paperID stringlengths 36 36 | pwc_id stringlengths 8 47 | arxiv_id stringlengths 6 16 ⌀ | nips_id float64 | url_abs stringlengths 18 329 | url_pdf stringlengths 18 742 | title stringlengths 8 325 | abstract stringlengths 1 7.27k ⌀ | authors stringlengths 2 7.06k | published stringlengths 10 10 ⌀ | conference stringlengths 12 47 ⌀ | conference_url_abs stringlengths 16 198 ⌀ | conference_url_pdf stringlengths 27 199 ⌀ | proceeding stringlengths 6 47 ⌀ | taskID stringlengths 7 1.44k | areaID stringclasses 688
values | embedding stringlengths 9.26k 12.5k | umap_embedding stringlengths 29 44 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
6babc05a-faab-46e7-838b-5fcfe251fa14 | what-makes-my-queries-slow-subgroup-discovery | 2108.03906 | null | https://arxiv.org/abs/2108.03906v1 | https://arxiv.org/pdf/2108.03906v1.pdf | "What makes my queries slow?": Subgroup Discovery for SQL Workload Analysis | Among daily tasks of database administrators (DBAs), the analysis of query workloads to identify schema issues and improving performances is crucial. Although DBAs can easily pinpoint queries repeatedly causing performance issues, it remains challenging to automatically identify subsets of queries that share some properties only (a pattern) and simultaneously foster some target measures, such as execution time. Patterns are defined on combinations of query clauses, environment variables, database alerts and metrics and help answer questions like what makes SQL queries slow? What makes I/O communications high? Automatically discovering these patterns in a huge search space and providing them as hypotheses for helping to localize issues and root-causes is important in the context of explainable AI. To tackle it, we introduce an original approach rooted on Subgroup Discovery. We show how to instantiate and develop this generic data-mining framework to identify potential causes of SQL workloads issues. We believe that such data-mining technique is not trivial to apply for DBAs. As such, we also provide a visualization tool for interactive knowledge discovery. We analyse a one week workload from hundreds of databases from our company, make both the dataset and source code available, and experimentally show that insightful hypotheses can be discovered. | ['Mehdi Kaytoue', 'Philippe Chaleat', 'Romain Mathonat', 'Anes Bendimerad', 'Youcef Remil'] | 2021-08-09 | null | null | null | null | ['subgroup-discovery'] | ['methodology'] | [-1.74817830e-01 4.40880880e-02 -4.32243496e-01 -4.82362807e-01
-4.90201771e-01 -6.58830881e-01 5.44828996e-02 9.00824904e-01
1.02674656e-01 3.20950657e-01 4.14579175e-02 -8.06525290e-01
-6.75155342e-01 -8.56781244e-01 -5.66627860e-01 -9.61397588e-03
-3.93901139e-01 6.78037047e-01 6.22238994e-01 -2.16882810e-01
6.81533217e-01 8.36143255e-01 -2.00619459e+00 6.38144970e-01
8.32031190e-01 8.78745198e-01 1.28687881e-02 5.69323123e-01
-5.42489231e-01 1.09171629e+00 -1.00883460e+00 1.44040430e-04
3.09470855e-02 -1.94833025e-01 -9.23286378e-01 -5.73169030e-02
2.59901762e-01 3.71711254e-02 3.63278389e-01 6.35322034e-01
1.77336425e-01 -3.60669822e-01 2.50862762e-02 -1.54575527e+00
2.18219552e-02 6.57500446e-01 -5.05970418e-01 5.93314290e-01
4.65331882e-01 5.45289703e-02 9.78513718e-01 -3.74336332e-01
7.24933922e-01 9.60903168e-01 3.25521141e-01 -1.24240965e-01
-1.50881135e+00 -3.45487982e-01 6.80935308e-02 4.01963919e-01
-1.36092901e+00 -3.53591412e-01 5.23646593e-01 -4.49254334e-01
1.33460963e+00 1.11629701e+00 3.32042724e-01 5.39009452e-01
-5.62523790e-02 5.07445455e-01 8.26421797e-01 -6.21249616e-01
3.80520672e-01 6.85316324e-01 6.04880154e-01 5.05085886e-01
4.52959567e-01 -2.15921581e-01 -7.32106149e-01 -7.00777769e-01
2.53774375e-01 1.61879435e-02 -8.77560154e-02 -4.97576773e-01
-1.12506425e+00 6.61015928e-01 -1.90545134e-02 5.31996369e-01
-4.18704152e-01 -1.28710389e-01 2.89670587e-01 6.82746828e-01
-7.44202211e-02 1.01443279e+00 -7.65798867e-01 -4.26964700e-01
-7.06099629e-01 4.48948711e-01 1.52803469e+00 1.07632291e+00
1.25604427e+00 -3.46439123e-01 -2.77355343e-01 5.02537668e-01
-2.27976561e-01 1.07989274e-01 2.56560594e-01 -8.69282782e-01
1.80000558e-01 1.41776073e+00 5.65932356e-02 -1.14293051e+00
-4.50518101e-01 -2.64429867e-01 -1.43195704e-01 -6.75448552e-02
3.18891793e-01 3.54907721e-01 -3.50869894e-01 1.32890940e+00
2.00312570e-01 -4.30240870e-01 -3.92578065e-01 5.80261409e-01
2.47491419e-01 2.16515437e-01 -1.95749909e-01 -5.32825291e-01
1.46167624e+00 -2.15022311e-01 -6.59560919e-01 1.22420024e-03
7.77256608e-01 -8.14916134e-01 1.44745743e+00 5.27010262e-01
-8.84770334e-01 -4.99735594e-01 -7.74630725e-01 3.95744979e-01
-5.50645649e-01 -1.09838113e-01 7.25350261e-01 6.34528041e-01
-6.09128892e-01 3.51731777e-01 -8.44202518e-01 -5.21179318e-01
-2.18313247e-01 2.97412932e-01 1.40685216e-01 2.37653524e-01
-6.71102107e-01 8.31106365e-01 3.12828958e-01 -6.12602174e-01
-3.24515343e-01 -1.02806365e+00 -8.96001607e-02 3.35322022e-01
1.21289182e+00 -5.03876567e-01 1.13054347e+00 -3.01580966e-01
-5.88180721e-01 7.32418001e-01 -3.75438452e-01 -2.89345533e-01
2.53220409e-01 -1.06412373e-01 -5.75941801e-01 -2.88201362e-01
9.38520059e-02 -3.87280226e-01 3.21302623e-01 -1.16780555e+00
-7.34527349e-01 -6.89821839e-01 5.31432219e-02 -5.01494586e-01
-3.28944296e-01 2.62390435e-01 -5.51502287e-01 -2.22274229e-01
7.56411627e-02 -5.46194255e-01 -9.36324745e-02 -7.29909420e-01
-8.69672239e-01 -2.30456471e-01 8.92429292e-01 -4.29452717e-01
2.06435609e+00 -1.85166585e+00 -2.29523912e-01 8.45133066e-01
3.35414767e-01 -1.31943181e-01 3.25122267e-01 7.03094840e-01
-1.77462310e-01 5.69575787e-01 1.58126995e-01 7.05694199e-01
6.85935766e-02 2.43739545e-01 -5.64789832e-01 -1.30131200e-01
3.79925221e-01 7.14528322e-01 -3.09058607e-01 -4.62353408e-01
1.83920041e-01 -2.89607972e-01 -6.51611507e-01 6.00746989e-01
-8.11313450e-01 -1.01547211e-01 -5.38289011e-01 9.50954080e-01
4.23016518e-01 -4.98803705e-01 4.35902506e-01 -2.94758290e-01
-5.30791521e-01 4.32446986e-01 -1.39114511e+00 8.71760190e-01
-2.71563381e-01 2.46680871e-01 -1.36045367e-01 -1.08297849e+00
1.01825511e+00 -1.53023303e-01 5.77748239e-01 -9.87661898e-01
-2.34078810e-01 1.76128566e-01 -3.35549861e-02 -9.59983885e-01
2.47815937e-01 4.85265583e-01 -1.21841051e-01 9.25161004e-01
-5.34998178e-01 2.05950275e-01 4.66582239e-01 -7.40633346e-03
1.81925082e+00 -5.47592998e-01 5.09268284e-01 -2.68985808e-01
4.25103277e-01 6.99720681e-01 4.26708013e-01 7.86658347e-01
1.25933245e-01 2.00037196e-01 1.18157375e+00 -7.17712581e-01
-9.82477725e-01 -9.68159318e-01 3.73953618e-02 1.35660541e+00
-3.48094590e-02 -9.02297020e-01 -3.28389138e-01 -5.52628756e-01
3.50303054e-01 8.67481232e-01 -4.63174403e-01 5.72825298e-02
-5.93477845e-01 -6.20964408e-01 1.98623970e-01 2.23775715e-01
5.84367290e-02 -1.04699659e+00 -1.17412400e+00 1.77528962e-01
-2.22913455e-03 -8.82536709e-01 6.46076053e-02 4.75163639e-01
-7.07602024e-01 -1.70282006e+00 4.57919359e-01 -1.30919218e-01
3.06074351e-01 2.02648565e-01 1.75193453e+00 5.88759720e-01
-1.00771320e+00 2.54486114e-01 -2.82426506e-01 -8.77095819e-01
-5.77498317e-01 1.45814881e-01 -3.47737104e-01 -3.46606344e-01
8.97741318e-01 -6.35411501e-01 -2.30136648e-01 7.57123649e-01
-8.66581440e-01 -1.73057497e-01 6.46814942e-01 2.38349319e-01
7.21018493e-01 4.61625099e-01 3.13590407e-01 -1.33262253e+00
1.05931818e+00 -6.91245198e-01 -8.30432832e-01 8.50761592e-01
-1.18281186e+00 2.93163449e-01 4.24473047e-01 -2.83200771e-01
-5.50158799e-01 -1.37705550e-01 3.75057459e-01 -3.68219525e-01
-4.60010856e-01 6.76972210e-01 -2.58940011e-01 4.36048895e-01
1.12599194e+00 1.10874332e-01 2.76072379e-02 -5.55558503e-01
1.34295508e-01 4.75311786e-01 3.51637453e-01 -7.27936447e-01
6.83816969e-01 3.84883523e-01 -2.65217215e-01 -7.65772402e-01
-4.04148489e-01 -7.70798147e-01 -4.08045292e-01 1.53463231e-02
3.43106985e-01 -2.88253337e-01 -1.19352341e+00 -5.36581814e-01
-9.16970670e-01 8.64597037e-02 -4.30182993e-01 1.33557711e-02
-5.50753891e-01 3.13695706e-02 3.92717309e-02 -8.53078365e-01
-3.75647135e-02 -9.63246942e-01 5.92886925e-01 3.25505026e-02
-7.86872208e-01 -5.24731576e-01 1.14259414e-01 4.71699625e-01
8.88094425e-01 3.17509681e-01 1.60571980e+00 -1.15837669e+00
-1.04784107e+00 -1.63114011e-01 -3.08484465e-01 -2.64572948e-01
1.47059113e-01 3.40191990e-01 -7.78462827e-01 -1.82278797e-01
2.44533885e-02 3.07378527e-02 1.36706829e-01 -7.19576329e-02
1.76341355e+00 -3.79894912e-01 -4.60103512e-01 3.14627200e-01
1.33722830e+00 3.44164610e-01 4.31295067e-01 5.91499984e-01
3.72568250e-01 9.26207721e-01 7.64873087e-01 7.26214647e-01
2.64656395e-01 9.79635835e-01 3.47559810e-01 -1.21598840e-01
4.44358170e-01 -6.22009784e-02 5.29818386e-02 3.82673591e-01
7.25856647e-02 6.12030104e-02 -1.34525919e+00 4.62812632e-01
-1.85704923e+00 -9.08773363e-01 -3.93877149e-01 2.24273682e+00
7.01693535e-01 3.14437300e-01 5.18883288e-01 2.94815600e-01
4.50674921e-01 -3.45597327e-01 -7.95045435e-01 -6.16891265e-01
2.37097234e-01 3.09429497e-01 3.92332524e-01 -8.80177505e-03
-5.51646113e-01 4.68663186e-01 6.53693724e+00 3.30515146e-01
-9.75411832e-01 -4.27878499e-01 4.37115610e-01 -5.15979603e-02
-5.77202380e-01 2.59831756e-01 -7.48123765e-01 2.09884048e-01
1.31764138e+00 -6.63981318e-01 4.15328741e-01 1.16929781e+00
5.06871760e-01 -2.95865446e-01 -1.42479992e+00 7.51975834e-01
-2.75248677e-01 -1.58581567e+00 -4.20889677e-03 3.52314860e-01
8.55464786e-02 -2.12915704e-01 -4.09048945e-01 2.29137763e-01
3.08143571e-02 -1.01667130e+00 2.70174652e-01 5.82820535e-01
2.98473895e-01 -8.06251347e-01 5.03413737e-01 5.47983110e-01
-8.10102463e-01 -3.13871324e-01 -3.59719247e-01 -8.58819485e-02
-4.23710287e-01 8.29291642e-01 -1.29399884e+00 4.17456031e-01
9.92648184e-01 2.98631564e-02 -9.70408082e-01 9.06489432e-01
3.19621176e-01 5.30431271e-01 -3.83219600e-01 -1.39873400e-01
-5.03283441e-01 9.71025527e-02 2.90906489e-01 1.13358700e+00
1.95121855e-01 -1.74609900e-01 6.57357946e-02 1.47615623e+00
2.61470199e-01 5.01043081e-01 -5.24748564e-01 -2.60986477e-01
7.26875424e-01 1.05935967e+00 -6.65414274e-01 -1.62526697e-01
-3.78538907e-01 3.09383720e-01 1.89178839e-01 2.72103459e-01
-5.54908693e-01 -4.38453019e-01 9.45717812e-01 8.45016420e-01
1.52984738e-01 -3.46224867e-02 -4.10901666e-01 -9.15995181e-01
5.01982868e-01 -1.31096601e+00 6.06466234e-01 -3.53677303e-01
-1.08311117e+00 3.74710530e-01 1.44410893e-01 -8.15992594e-01
-5.06885767e-01 -3.63648444e-01 -7.43649542e-01 8.87496889e-01
-1.12956655e+00 -4.70058262e-01 -6.58358872e-01 5.41385949e-01
1.66505665e-01 -2.06348374e-01 8.75996828e-01 1.33377269e-01
-6.42412364e-01 4.62486118e-01 -3.16245735e-01 -3.76193076e-01
5.68874359e-01 -1.42170668e+00 3.42964470e-01 6.23237371e-01
3.08708340e-01 1.06004238e+00 8.88675690e-01 -5.28213739e-01
-1.83428955e+00 -8.07451963e-01 9.43709612e-01 -7.86970437e-01
7.55477607e-01 -3.52434158e-01 -1.42703104e+00 4.08953667e-01
-1.03258967e-01 -2.10715294e-01 8.64942253e-01 4.93416697e-01
-4.65632260e-01 -3.49966019e-01 -7.69936323e-01 1.37288809e-01
7.81235993e-01 -4.44077253e-01 -5.57652414e-01 4.98677224e-01
6.46098018e-01 2.01736782e-02 -1.08819818e+00 4.85374600e-01
3.15750718e-01 -1.65518284e+00 8.25488985e-01 -1.00904047e+00
8.23360607e-02 -3.58296782e-01 -3.09846371e-01 -6.68726742e-01
9.25396159e-02 -7.60516465e-01 -2.39061221e-01 1.10837758e+00
6.52119637e-01 -3.67307097e-01 9.93626893e-01 9.61065233e-01
2.69496471e-01 -9.29688275e-01 -3.26206058e-01 -9.03044999e-01
-5.51792622e-01 -7.33208120e-01 8.75154078e-01 1.10543025e+00
7.26458430e-02 2.75249004e-01 7.82845914e-02 1.77471846e-01
4.14601922e-01 8.24001670e-01 1.38737750e+00 -1.38788426e+00
-5.52155495e-01 -5.24239302e-01 -2.26834357e-01 -4.52570677e-01
-4.07011569e-01 -5.19374967e-01 -5.60815692e-01 -1.13577533e+00
2.16967285e-01 -7.04220951e-01 -1.15630329e-01 4.76290196e-01
-5.63457683e-02 -6.78949654e-01 -6.02676421e-02 4.58036244e-01
-4.49031204e-01 -4.21357661e-01 4.55095112e-01 8.67128223e-02
-5.29075146e-01 3.67636144e-01 -9.12903428e-01 2.27843210e-01
6.05804563e-01 -4.31069911e-01 -4.77936298e-01 3.91524769e-02
6.14961386e-01 1.73302397e-01 2.57111192e-01 -9.83929336e-01
3.32017094e-01 -6.04815066e-01 -1.87366363e-02 -8.38250995e-01
-4.48642462e-01 -8.50214005e-01 5.49667001e-01 5.53174496e-01
-4.13545579e-01 4.58125681e-01 2.93192089e-01 3.30420613e-01
-4.37825531e-01 6.90301508e-03 2.30551764e-01 -1.60627261e-01
-7.74290204e-01 -1.34743834e-02 -1.78649589e-01 2.13591322e-01
1.33161688e+00 -1.63083434e-01 -6.12048328e-01 -1.52057797e-01
-5.24599135e-01 4.08897460e-01 6.94381356e-01 4.93036091e-01
3.65011871e-01 -7.79920280e-01 -4.03050095e-01 4.99473333e-01
7.84487247e-01 -2.82484621e-01 -1.99837014e-01 7.39402950e-01
-5.90597808e-01 6.29830599e-01 -8.33636671e-02 -7.32567012e-01
-1.48566937e+00 8.87494326e-01 1.48235634e-01 -2.84120321e-01
-3.35629135e-01 4.94589180e-01 -1.78282365e-01 -1.52404934e-01
4.82614100e-01 -5.23427546e-01 1.60990171e-02 1.25034884e-01
5.95641017e-01 3.84448856e-01 4.51616317e-01 2.54979491e-01
-6.12466872e-01 1.80111513e-01 -7.20417351e-02 4.93455023e-01
1.51410770e+00 -8.66071060e-02 -3.97603273e-01 6.66753888e-01
9.02592361e-01 2.73452193e-01 -2.69637734e-01 -2.28392586e-01
9.39019620e-01 -6.73014224e-01 -4.53240186e-01 -1.02609122e+00
-8.02579582e-01 6.82864130e-01 4.75837499e-01 1.09292924e+00
1.31005096e+00 3.17611843e-01 2.57538259e-01 6.07686698e-01
3.55116278e-01 -1.03621185e+00 1.13448009e-01 3.97901274e-02
8.28780055e-01 -1.21808553e+00 8.44446123e-02 -5.84190428e-01
-3.24003488e-01 1.35122871e+00 8.78082514e-01 4.45462883e-01
6.68755233e-01 5.99573314e-01 1.79748908e-01 -8.19947422e-01
-1.26543844e+00 -2.16553807e-01 -2.36943476e-02 2.19152302e-01
5.55242360e-01 1.16503976e-01 -3.43892306e-01 4.98091757e-01
-4.77227122e-01 -1.30942121e-01 4.40795541e-01 1.26751292e+00
-7.20193684e-01 -1.40645337e+00 -4.95537668e-01 8.86176646e-01
-3.10272276e-01 3.32361817e-01 -7.26511717e-01 1.11583304e+00
1.69762354e-02 9.15352643e-01 1.52397648e-01 -7.31675267e-01
1.05201375e+00 1.33187830e-01 -2.51307115e-02 -6.23123109e-01
-5.89961886e-01 -1.21445969e-01 2.51450241e-01 -1.05777979e+00
-1.13254758e-02 -4.43558306e-01 -1.01746714e+00 -5.88361442e-01
-2.31262937e-01 4.02127415e-01 7.46150494e-01 7.21928596e-01
7.10841417e-01 5.73376656e-01 6.99778557e-01 2.11845204e-01
-6.15534484e-01 -6.77834392e-01 -3.58520299e-01 6.68520093e-01
-1.77495945e-02 -4.25409555e-01 -1.32315084e-01 8.03221688e-02] | [8.965415954589844, 7.317539215087891] |
5da92c0a-90f3-4267-a39c-67b2991c90a2 | climategan-raising-climate-change-awareness | 2110.02871 | null | https://arxiv.org/abs/2110.02871v1 | https://arxiv.org/pdf/2110.02871v1.pdf | ClimateGAN: Raising Climate Change Awareness by Generating Images of Floods | Climate change is a major threat to humanity, and the actions required to prevent its catastrophic consequences include changes in both policy-making and individual behaviour. However, taking action requires understanding the effects of climate change, even though they may seem abstract and distant. Projecting the potential consequences of extreme climate events such as flooding in familiar places can help make the abstract impacts of climate change more concrete and encourage action. As part of a larger initiative to build a website that projects extreme climate events onto user-chosen photos, we present our solution to simulate photo-realistic floods on authentic images. To address this complex task in the absence of suitable training data, we propose ClimateGAN, a model that leverages both simulated and real data for unsupervised domain adaptation and conditional image generation. In this paper, we describe the details of our framework, thoroughly evaluate components of our architecture and demonstrate that our model is capable of robustly generating photo-realistic flooding. | ['Yoshua Bengio', 'Alex Hernandez-Garcia', 'Vahe Vardanyan', 'Adrien Juraver', 'Gautier Cosne', 'Sunand Raghupathi', 'Alexia Reynaud', 'Tianyu Zhang', 'Mélisande Teng', 'Alexandra Sasha Luccioni', 'Victor Schmidt'] | 2021-10-06 | climategan-raising-climate-change-awareness-1 | https://openreview.net/forum?id=EZNOb_uNpJk | https://openreview.net/pdf?id=EZNOb_uNpJk | iclr-2022-4 | ['conditional-image-generation'] | ['computer-vision'] | [ 6.11796439e-01 2.20330074e-01 4.34485048e-01 -5.07331073e-01
-6.74489260e-01 -7.82077789e-01 9.33929265e-01 -7.67741119e-03
-3.56949747e-01 8.75105679e-01 7.19175935e-01 -7.26277649e-01
4.02762234e-01 -1.01676202e+00 -6.94562137e-01 -5.82704842e-01
3.85489352e-02 1.07586794e-01 -3.87908071e-02 -5.03340840e-01
1.96825296e-01 7.43703365e-01 -1.33167994e+00 4.86976933e-03
8.57307553e-01 5.86640649e-03 3.23956162e-01 9.51503515e-01
2.53054351e-02 5.17607450e-01 -9.08437729e-01 1.52096348e-02
2.07137942e-01 -4.93419021e-01 -7.13190615e-01 2.89696231e-02
3.81247818e-01 -5.84599018e-01 -8.59445259e-02 6.72906101e-01
7.52677739e-01 -1.04899295e-02 7.89637804e-01 -9.59366918e-01
-6.44314826e-01 2.84945458e-01 -2.87793726e-01 2.15387255e-01
4.19191182e-01 7.49122202e-01 2.49526218e-01 -5.16709268e-01
4.68209445e-01 1.43259263e+00 6.87687278e-01 3.78082871e-01
-1.27583611e+00 -6.20683849e-01 2.29657397e-01 -1.56174466e-01
-1.17822182e+00 -5.70542336e-01 4.74874437e-01 -4.85527754e-01
8.85968387e-01 2.96410054e-01 6.41557753e-01 1.17234564e+00
1.01791121e-01 1.56682149e-01 1.29610682e+00 -4.52284634e-01
5.28068244e-01 5.37411608e-02 -6.31890178e-01 3.56554464e-02
1.24268569e-01 1.53261781e-01 -1.54972196e-01 -3.15889627e-01
5.31371653e-01 -1.06332786e-01 -2.66149640e-01 2.92349130e-01
-1.08433545e+00 5.23655891e-01 7.85358250e-01 2.99260374e-02
-4.41749901e-01 5.80805123e-01 -3.19679648e-01 -1.48418427e-01
5.59433699e-01 3.26008856e-01 -1.80310652e-01 -2.36234665e-02
-9.16354895e-01 4.36307728e-01 5.39549232e-01 4.90573615e-01
8.45693588e-01 8.53598937e-02 2.08900616e-01 4.02489156e-01
3.12504351e-01 1.36687946e+00 -2.00275928e-01 -1.13840580e+00
2.86324769e-01 3.19538474e-01 5.40433764e-01 -1.11882269e+00
-4.46874052e-01 3.25780399e-02 -5.84808290e-01 4.84518617e-01
4.07235861e-01 -6.80702388e-01 -1.13045228e+00 1.79829788e+00
4.94901061e-01 3.85098100e-01 1.03736214e-01 7.16449857e-01
3.97663534e-01 1.16249728e+00 6.54977262e-01 2.77521253e-01
1.12218881e+00 -4.44732308e-02 -5.06643713e-01 -6.29966497e-01
3.63831937e-01 -6.00806832e-01 1.19053316e+00 -2.82214791e-01
-7.73012340e-01 -1.55563235e-01 -7.59543121e-01 2.81390160e-01
-6.75912678e-01 -3.80195767e-01 2.91409284e-01 8.02098274e-01
-1.14871848e+00 3.85144591e-01 -6.10568821e-01 -8.91977847e-01
2.04120278e-01 -2.57565260e-01 -4.49073426e-02 -4.38751206e-02
-1.44802809e+00 1.20804870e+00 1.42078800e-02 2.67624170e-01
-9.12383378e-01 -9.10013855e-01 -8.28728259e-01 1.27194181e-01
-7.49646947e-02 -5.56958199e-01 1.02771199e+00 -7.09804177e-01
-8.79008472e-01 7.47244060e-01 -3.27951275e-02 -2.84084857e-01
4.71579015e-01 -2.12873537e-02 -3.64128381e-01 1.27263770e-01
3.38317007e-02 1.21458578e+00 4.53207076e-01 -1.46997941e+00
-4.52621460e-01 -1.91769272e-01 1.86585799e-01 5.61119139e-01
-2.68327624e-01 2.18782470e-01 1.43153638e-01 -4.92213905e-01
-5.20761490e-01 -1.11995924e+00 -8.02641869e-01 1.44363433e-01
-1.79290712e-01 4.38173413e-01 7.41928339e-01 -9.26696539e-01
7.88150907e-01 -2.14079237e+00 -4.11386311e-01 1.22574471e-01
-2.57691681e-01 1.58707052e-01 -4.44556437e-02 8.34548414e-01
1.91465899e-01 5.86658776e-01 -8.12748253e-01 -7.99363554e-02
5.08999601e-02 2.59629995e-01 -7.15660453e-01 3.91395777e-01
3.09529960e-01 6.63042188e-01 -9.63180065e-01 -3.37222159e-01
3.63340765e-01 6.71893179e-01 -1.84359670e-01 1.70039922e-01
-4.20148671e-01 9.03500438e-01 -4.09278184e-01 3.56771886e-01
8.97669852e-01 2.88071539e-02 3.26291651e-01 5.22128820e-01
-3.20241302e-01 1.43146142e-01 -9.06290829e-01 1.19383729e+00
-5.31817913e-01 6.32382989e-01 -9.28048743e-04 -2.55039513e-01
7.77324378e-01 3.11840713e-01 9.45098549e-02 -8.24255228e-01
-4.20245051e-01 4.07195762e-02 -4.67644185e-01 -5.68832457e-01
5.71505427e-01 -2.75830179e-01 -3.92183542e-01 8.57618928e-01
-9.15070415e-01 -6.48039699e-01 -2.16172844e-01 3.24499875e-01
8.73580039e-01 4.84809041e-01 1.56189099e-01 -2.69374877e-01
1.41340837e-01 4.20268625e-01 3.82336169e-01 7.25277424e-01
-3.15822124e-01 7.94432282e-01 1.51682287e-01 -7.76983619e-01
-1.25207889e+00 -1.18126559e+00 -1.34712785e-01 7.65457690e-01
1.97984688e-02 1.28219709e-01 -8.02785575e-01 -3.91942173e-01
7.75548117e-03 1.27174890e+00 -5.43817520e-01 -6.42465651e-02
-5.70676744e-01 -1.21407223e+00 8.14952552e-01 2.91951030e-01
7.00384259e-01 -1.47313988e+00 -1.19147253e+00 -4.25964780e-02
-4.35415655e-01 -8.79600883e-01 -1.82768315e-01 -2.69986182e-01
-5.54494798e-01 -7.41742194e-01 -8.82631004e-01 -3.61416757e-01
8.30617607e-01 4.24068391e-01 1.25903130e+00 8.73860568e-02
-4.74589407e-01 7.54332662e-01 -8.25137571e-02 -6.45653427e-01
-6.28307283e-01 -2.77461499e-01 -1.75322935e-01 -3.41007411e-01
2.74916887e-01 -7.60705352e-01 -1.00004244e+00 5.06647341e-02
-1.18606305e+00 7.81765673e-04 3.86514664e-01 7.26182833e-02
2.04572201e-01 -1.90228030e-01 6.28773928e-01 -6.71356380e-01
5.14087558e-01 -7.13465273e-01 -4.67588753e-01 3.14760834e-01
-2.80062526e-01 -1.91397294e-01 2.95148730e-01 -7.49324188e-02
-1.54166567e+00 3.91397774e-01 2.82765944e-02 4.11318660e-01
-7.10564375e-01 5.30605674e-01 -1.59322098e-02 1.68265849e-01
1.10728657e+00 1.76560491e-01 -4.73853648e-01 -3.14526439e-01
6.75795317e-01 8.42108667e-01 6.85386300e-01 -5.75374484e-01
1.05120075e+00 8.71511400e-01 -1.86253846e-01 -1.14844143e+00
-5.85519791e-01 -9.50396135e-02 -4.52219814e-01 -5.76106131e-01
8.85669291e-01 -1.23792517e+00 -1.02846578e-01 5.12845337e-01
-1.02227998e+00 -8.34330618e-01 -1.01278819e-01 6.60272911e-02
-2.49513015e-01 2.83409566e-01 1.29302114e-01 -9.18578029e-01
-1.46526501e-01 -6.77121222e-01 9.57422495e-01 4.89171624e-01
-3.28198701e-01 -1.08454549e+00 6.59531176e-01 6.78882748e-02
9.17840838e-01 8.23933423e-01 6.68837905e-01 1.58409290e-02
-6.27684116e-01 1.04958586e-01 -2.20463991e-01 -2.07734872e-02
4.43115145e-01 1.06758013e-01 -1.18523026e+00 -2.48783216e-01
-3.57173890e-01 -3.39837015e-01 6.88989520e-01 3.43011528e-01
4.88565207e-01 -3.93662035e-01 -3.20127606e-01 4.90162671e-01
1.46507192e+00 -6.25282004e-02 1.03675628e+00 5.30342281e-01
3.41283113e-01 9.06782210e-01 2.17227593e-01 4.97772455e-01
8.81798148e-01 3.68483722e-01 4.41080034e-01 -6.08595908e-01
-3.06578934e-01 -8.82294178e-02 4.30162370e-01 -1.93100899e-01
1.84887424e-01 -8.48713592e-02 -1.31548893e+00 1.03728485e+00
-1.40062702e+00 -1.38824832e+00 -1.73858941e-01 2.29028797e+00
7.76568711e-01 -1.92039415e-01 2.46101432e-02 -5.36971509e-01
6.22773290e-01 4.13313031e-01 -6.91289425e-01 -5.38727880e-01
-2.82069892e-01 3.66843566e-02 6.95293427e-01 6.28297567e-01
-1.00508785e+00 9.61083949e-01 7.58784294e+00 1.75911821e-02
-1.18437433e+00 -1.06748588e-01 1.01113486e+00 2.34369785e-02
-8.82516503e-01 3.47710490e-01 -5.14625669e-01 3.49807382e-01
1.37404931e+00 -3.10109705e-01 4.06039655e-01 3.75431418e-01
9.26176667e-01 -4.69308287e-01 -5.43234706e-01 1.38454422e-01
-1.56195551e-01 -1.21899772e+00 -2.31844321e-01 1.07332483e-01
9.12358224e-01 2.37877816e-01 1.81407541e-01 3.60385738e-02
1.09526992e+00 -1.18089545e+00 5.67570865e-01 5.79161882e-01
7.88223326e-01 -5.73649824e-01 1.19400643e-01 3.19502324e-01
-9.35115933e-01 1.10586159e-01 -3.42689782e-01 -3.74662727e-01
3.74006242e-01 5.05589664e-01 -9.54397857e-01 4.72233035e-02
8.33734751e-01 3.15466881e-01 -6.32164121e-01 9.05033708e-01
-5.23064315e-01 8.32512498e-01 -7.12408900e-01 2.75945723e-01
2.33387232e-01 -1.01855665e-01 3.34526300e-01 1.41779947e+00
4.38506067e-01 3.71378422e-01 2.05483273e-01 9.17256653e-01
2.14966625e-01 -1.66510969e-01 -1.04397190e+00 1.50724202e-01
6.13660753e-01 1.15521991e+00 -5.51127017e-01 -3.06371003e-01
-1.85520187e-01 8.07466209e-01 -1.86736416e-02 6.46477163e-01
-8.68663669e-01 -2.43182063e-01 8.45170200e-01 4.97678742e-02
8.08850378e-02 -3.68825853e-01 -4.94459003e-01 -8.69848847e-01
-3.54107134e-02 -7.57642984e-01 1.89932987e-01 -1.08267832e+00
-6.79274142e-01 3.68725598e-01 2.89560258e-01 -9.60832894e-01
-3.04065168e-01 -1.78831338e-03 -1.21326435e+00 1.28613675e+00
-1.98121572e+00 -1.44542134e+00 -5.06786048e-01 2.90518701e-01
6.61675856e-02 6.94153130e-01 9.50509191e-01 -2.30106562e-01
-2.30224609e-01 -4.20804136e-02 1.96080297e-01 -3.58480155e-01
9.06659782e-01 -1.29862511e+00 1.20302880e+00 1.20095038e+00
-2.84992903e-01 4.71183240e-01 9.89177704e-01 -1.00278628e+00
-8.55388105e-01 -1.33954716e+00 9.99475360e-01 -4.81714964e-01
4.13136721e-01 -2.41296470e-01 -7.70301044e-01 7.32674062e-01
4.15197819e-01 -4.17033166e-01 6.01086557e-01 -3.02877039e-01
-4.24401104e-01 4.95454259e-02 -1.46290612e+00 1.09996367e+00
7.31751800e-01 -4.68214154e-01 -3.91725719e-01 6.71815991e-01
6.29313827e-01 9.90201756e-02 -5.46691895e-01 6.71954900e-02
3.77287120e-01 -1.03759587e+00 1.02642727e+00 -2.86867797e-01
3.78250360e-01 -5.70354402e-01 -2.20417663e-01 -1.59031534e+00
-2.67000765e-01 -5.05921304e-01 6.61088169e-01 1.09973001e+00
5.76364160e-01 -5.97178638e-01 6.89021707e-01 1.10131443e+00
1.57629117e-01 2.45557025e-01 -7.33714759e-01 -4.94337618e-01
7.02165902e-01 -3.30277175e-01 8.74615312e-01 7.85099924e-01
-1.00178495e-01 -1.93774030e-01 -5.31172156e-01 8.71291220e-01
6.58962846e-01 5.97780533e-02 9.83355463e-01 -8.63393009e-01
4.22848836e-02 -1.13122061e-01 3.01362034e-02 -4.60122079e-01
-2.14887172e-01 -2.30351195e-01 2.59433687e-01 -1.86966705e+00
1.95338979e-01 -3.92911464e-01 1.67599514e-01 6.92231655e-01
-4.45230871e-01 3.72030497e-01 1.75393194e-01 1.35909021e-01
2.00268496e-02 5.32795489e-01 8.33207846e-01 -1.54093251e-01
-1.73068106e-01 -3.80094081e-01 -8.22322726e-01 6.02519989e-01
1.00352514e+00 -5.41488826e-01 -4.03001040e-01 -5.71654677e-01
3.95106047e-01 -2.34097555e-01 6.17840052e-01 -1.19145131e+00
4.42942977e-03 -9.29770470e-01 5.16355157e-01 -4.00484025e-01
2.37030149e-01 -6.73442602e-01 5.38771808e-01 6.10004425e-01
-3.42097849e-01 3.75673324e-02 4.70452845e-01 4.18263108e-01
3.81971121e-01 2.55166590e-02 7.29210079e-01 -4.79636371e-01
-6.39052391e-01 -3.52567025e-02 -9.09127355e-01 -3.35416384e-02
1.16501772e+00 1.34151638e-01 -6.82528794e-01 -9.25692737e-01
-5.29305220e-01 4.83725518e-01 1.05442750e+00 2.60435671e-01
4.50058073e-01 -1.00876880e+00 -1.19732344e+00 -1.53459460e-01
1.97297335e-01 -7.48017356e-02 4.67007518e-01 7.39734694e-02
-8.07312548e-01 7.18797976e-03 -8.71248990e-02 -2.42979303e-01
-9.96759653e-01 2.90779442e-01 4.62750316e-01 -1.32983038e-02
-7.76703119e-01 4.33066308e-01 3.86797667e-01 -9.07068014e-01
-1.36889219e-01 2.98541947e-03 -2.01761767e-01 -3.22611719e-01
9.94224846e-01 2.78156921e-02 -3.26788813e-01 -7.40515947e-01
-3.45033556e-01 4.05332297e-01 4.10241097e-01 -5.68067133e-01
1.54262447e+00 -5.46845376e-01 1.26380354e-01 1.37625769e-01
4.97092664e-01 -6.29281579e-03 -1.82770586e+00 4.91050584e-03
-1.19262405e-01 -5.06871402e-01 -6.11168779e-02 -1.32237971e+00
-7.14092195e-01 9.42039371e-01 7.49566436e-01 -6.87593445e-02
1.15368581e+00 -2.60800362e-01 8.51411939e-01 5.31411767e-01
1.61397904e-01 -9.10964549e-01 -2.49726489e-01 9.99277607e-02
9.79271293e-01 -1.01648939e+00 2.05816448e-01 -9.76349860e-02
-6.36231244e-01 7.69592762e-01 3.38576049e-01 1.47631064e-01
5.01063824e-01 2.63585597e-01 6.01463675e-01 -9.81968641e-02
-6.31645620e-01 -2.95944810e-01 -4.39059228e-01 1.12588918e+00
1.59647375e-01 2.86012322e-01 2.83513546e-01 -4.22793984e-01
4.91250493e-02 -3.91118303e-02 1.07432687e+00 1.10469079e+00
-7.23730981e-01 -8.60411286e-01 -1.05352151e+00 -2.28093997e-01
-1.34330690e-01 -2.90851384e-01 -5.94475865e-01 4.74403173e-01
-4.82162200e-02 1.14574337e+00 3.97612192e-02 5.33456244e-02
1.85266882e-01 6.53496161e-02 -1.74979054e-04 -3.60449493e-01
-5.14642715e-01 -3.14194769e-01 2.42587984e-01 -3.05535972e-01
-5.06387353e-01 -9.07674134e-01 -1.13878584e+00 -6.70288801e-01
3.73790205e-01 -1.72089383e-01 1.05601060e+00 8.22535932e-01
6.46791637e-01 9.21640918e-02 8.43165755e-01 -1.17956865e+00
-4.16399598e-01 -7.79336751e-01 -2.83876896e-01 5.85774064e-01
2.39558309e-01 -8.81097093e-02 -4.39721882e-01 4.27958488e-01] | [9.54433536529541, -1.6526341438293457] |
cba164e3-9495-414e-a023-80700e323060 | awq-activation-aware-weight-quantization-for | 2306.00978 | null | https://arxiv.org/abs/2306.00978v1 | https://arxiv.org/pdf/2306.00978v1.pdf | AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration | Large language models (LLMs) have shown excellent performance on various tasks, but the astronomical model size raises the hardware barrier for serving (memory size) and slows down token generation (memory bandwidth). In this paper, we propose Activation-aware Weight Quantization (AWQ), a hardware-friendly approach for LLM low-bit weight-only quantization. Our method is based on the observation that weights are not equally important: protecting only 1% of salient weights can greatly reduce quantization error. We then propose to search for the optimal per-channel scaling that protects the salient weights by observing the activation, not weights. AWQ does not rely on any backpropagation or reconstruction, so it can well preserve LLMs' generalization ability on different domains and modalities, without overfitting to the calibration set; it also does not rely on any data layout reordering, maintaining the hardware efficiency. AWQ outperforms existing work on various language modeling, common sense QA, and domain-specific benchmarks. Thanks to better generalization, it achieves excellent quantization performance for instruction-tuned LMs and, for the first time, multi-modal LMs. We also implement efficient tensor core kernels with reorder-free online dequantization to accelerate AWQ, achieving a 1.45x speedup over GPTQ and is 1.85x faster than the cuBLAS FP16 implementation. Our method provides a turn-key solution to compress LLMs to 3/4 bits for efficient deployment. | ['Song Han', 'Xingyu Dang', 'Shang Yang', 'Haotian Tang', 'Jiaming Tang', 'Ji Lin'] | 2023-06-01 | null | null | null | null | ['quantization', 'common-sense-reasoning'] | ['methodology', 'reasoning'] | [-7.78537393e-02 -3.81324857e-01 -6.15141988e-01 -1.09047167e-01
-1.14449179e+00 -3.52220505e-01 2.26646289e-01 3.83157194e-01
-8.45798969e-01 2.95431077e-01 2.46789262e-01 -7.80370951e-01
2.23942176e-01 -6.22192979e-01 -7.56642997e-01 -7.01942980e-01
-2.59397298e-01 1.78113565e-01 5.48040211e-01 -3.60476762e-01
5.01088381e-01 2.50403911e-01 -1.34171784e+00 5.59461951e-01
6.54001534e-01 1.32397783e+00 2.48879716e-01 8.96976948e-01
-3.03372562e-01 8.54279459e-01 -5.26920676e-01 -4.66854841e-01
2.20796883e-01 3.92202467e-01 -1.02893138e+00 -4.91871446e-01
6.59239590e-01 -5.42426229e-01 -4.48004305e-01 1.04729211e+00
6.00607276e-01 -5.26427291e-02 3.63243073e-01 -1.13763082e+00
-5.36966681e-01 1.00340867e+00 -7.86539137e-01 2.95597255e-01
-1.81568488e-01 1.07492052e-01 1.42910743e+00 -9.59151566e-01
1.63995847e-01 1.32101440e+00 6.10731423e-01 4.58154023e-01
-1.31791246e+00 -8.54605019e-01 -3.35192680e-02 3.76860529e-01
-1.82040453e+00 -6.03713930e-01 1.72906458e-01 -1.98981706e-02
1.40321183e+00 3.24666709e-01 2.54661322e-01 6.37916565e-01
3.02921504e-01 8.08149576e-01 1.01462793e+00 -5.47155678e-01
4.31075215e-01 -8.60050395e-02 3.33773702e-01 9.39356327e-01
2.29126960e-01 -2.45356247e-01 -8.14487219e-01 -6.42954588e-01
5.38412154e-01 -3.03150833e-01 -2.12233275e-01 6.36658594e-02
-1.36028838e+00 7.86033988e-01 2.89509356e-01 2.00004160e-01
-2.32029725e-02 1.02717137e+00 8.08585525e-01 4.16188329e-01
1.43035995e-02 2.27278844e-01 -6.04414582e-01 -4.43853468e-01
-1.27911937e+00 8.09953287e-02 6.07504666e-01 8.61333132e-01
7.21404314e-01 4.49202269e-01 -1.70043200e-01 6.67727411e-01
3.49239796e-01 8.36829662e-01 7.66655266e-01 -8.94100726e-01
6.66535735e-01 1.59419850e-01 -2.76862413e-01 -6.42201424e-01
-6.29671633e-01 -3.88431489e-01 -9.38708782e-01 -1.29063517e-01
4.42366421e-01 1.88968807e-01 -8.44657004e-01 1.73271990e+00
3.02939471e-02 7.63647407e-02 8.17072205e-03 7.70746589e-01
2.88196921e-01 7.93511808e-01 1.42267138e-01 1.72178492e-01
1.82673705e+00 -6.83401406e-01 -5.50880432e-01 -6.50999248e-02
1.26845932e+00 -9.10258174e-01 1.46290636e+00 7.80188084e-01
-1.11447942e+00 -3.72319788e-01 -1.46086538e+00 -5.16161859e-01
-1.97073206e-01 1.52082011e-01 7.91752279e-01 9.31781828e-01
-1.16741538e+00 5.25944591e-01 -9.73502100e-01 1.76578686e-01
1.59715667e-01 6.74299657e-01 -9.20181125e-02 1.22176059e-01
-1.43596590e+00 6.47860229e-01 5.46279907e-01 -3.67430061e-01
-5.96263051e-01 -9.01874602e-01 -5.91573894e-01 3.42235178e-01
2.00933009e-01 -3.40675056e-01 1.17711747e+00 -6.04933381e-01
-1.61705017e+00 4.71037477e-01 -2.80471683e-01 -9.57195222e-01
-1.19849928e-01 -1.40349455e-02 -5.53589404e-01 1.88885286e-01
-3.40351611e-01 8.17464411e-01 1.03475261e+00 -6.43620849e-01
-6.21422052e-01 -8.92582461e-02 1.78197864e-02 -1.70818597e-01
-1.01461458e+00 -1.51644558e-01 -4.27198380e-01 -7.35707998e-01
8.08358639e-02 -8.16920042e-01 -6.83092326e-02 -1.18015967e-01
-4.13784504e-01 -1.25377357e-01 3.42462867e-01 -4.44841027e-01
1.69683707e+00 -2.42944312e+00 -2.48188674e-01 4.25956011e-01
4.45035577e-01 3.78923863e-01 -2.28457049e-01 3.22659791e-01
2.11102799e-01 1.51696175e-01 1.08175159e-01 -3.98175389e-01
4.52160656e-01 2.47165725e-01 -8.01501215e-01 4.37367052e-01
-2.56404579e-02 8.40949357e-01 -6.59037292e-01 -5.43176234e-01
-3.16073820e-02 3.10085148e-01 -8.87297750e-01 -3.22867304e-01
-5.49946167e-02 -4.14589256e-01 -1.64604615e-02 5.00448823e-01
8.94788325e-01 -4.57938612e-01 1.74155310e-01 -7.61646807e-01
-1.00478686e-01 7.71244526e-01 -1.42471159e+00 1.77960682e+00
-7.16161788e-01 5.63121617e-01 1.23799115e-01 -9.25665557e-01
6.41636431e-01 1.61218330e-01 3.08430612e-01 -8.90141010e-01
3.50286113e-03 5.83626628e-01 2.63100546e-02 4.12146039e-02
1.08740723e+00 3.53431664e-02 -3.35428715e-01 5.37022829e-01
2.04968303e-01 -7.33782724e-03 1.64120033e-01 4.98120248e-01
9.15767193e-01 -5.59335470e-01 -6.36886731e-02 -5.75577855e-01
4.68044192e-01 -2.61781365e-01 3.38523954e-01 7.90507495e-01
-1.14448637e-01 1.45725593e-01 3.33890259e-01 -2.13846505e-01
-1.19668365e+00 -1.09941947e+00 -2.86845356e-01 1.58755565e+00
-2.79127266e-02 -1.09446239e+00 -6.60484254e-01 -7.75910839e-02
9.28238034e-03 6.46796942e-01 1.35788480e-02 -3.12060773e-01
-5.20830691e-01 -8.85853350e-01 1.38479567e+00 5.22852123e-01
6.36909485e-01 -1.12870947e-01 -5.18524468e-01 2.99710542e-01
-6.13366403e-02 -1.09698343e+00 -7.19395697e-01 4.75870162e-01
-8.67384672e-01 -4.88624454e-01 -4.03560072e-01 -4.27845418e-01
2.49571800e-01 2.27102190e-01 1.02025402e+00 2.03967229e-01
3.90952639e-02 8.32596198e-02 -2.54378524e-02 -1.77327469e-01
-4.19827431e-01 4.31189954e-01 4.28203136e-01 -1.73557565e-01
4.04345244e-01 -4.09826785e-01 -5.95746458e-01 2.30561532e-02
-9.61918175e-01 -6.96841702e-02 5.86002827e-01 1.06875932e+00
6.48585200e-01 -1.15228901e-02 1.89229280e-01 -5.91301203e-01
5.59157670e-01 -1.43357351e-01 -6.37300253e-01 1.70220122e-01
-7.77322471e-01 7.58230209e-01 7.64743328e-01 -5.48598051e-01
-5.24432063e-01 -2.20364600e-01 -4.55217779e-01 -3.60176414e-01
4.98728484e-01 3.12715113e-01 2.61937320e-01 -3.16953003e-01
7.71646798e-01 3.16719145e-01 -2.29235515e-01 -3.72204304e-01
7.12100565e-01 8.02136183e-01 4.64250773e-01 -1.08752894e+00
5.49734890e-01 5.70355952e-01 6.32822290e-02 -8.52558494e-01
-3.24564308e-01 -2.76217729e-01 -2.49358624e-01 4.26943749e-01
4.72873986e-01 -1.12183785e+00 -1.12762535e+00 2.18561426e-01
-9.74531054e-01 -4.39509034e-01 -2.16418818e-01 5.03657639e-01
-2.85059452e-01 7.05092192e-01 -1.16765141e+00 -6.01941526e-01
-7.57728338e-01 -1.42926288e+00 1.18666553e+00 -3.32335144e-01
-2.01219380e-01 -7.85375655e-01 -4.31829542e-01 8.46005604e-02
6.16081417e-01 -6.27903581e-01 1.04100931e+00 -3.73725593e-01
-6.59886003e-01 5.54787144e-02 -3.92388493e-01 3.64631832e-01
-3.92679751e-01 -1.39678746e-01 -1.01623929e+00 -5.76663435e-01
-2.67794639e-01 -5.49672902e-01 1.09908259e+00 6.13737702e-02
1.41985059e+00 -4.14341569e-01 -4.57053399e-03 8.35828185e-01
1.32768810e+00 -5.45388758e-01 5.41629791e-01 1.62465304e-01
7.04016149e-01 -9.27637145e-02 3.61832887e-01 6.25028789e-01
6.35565579e-01 6.34200513e-01 2.99337715e-01 -3.41104642e-02
-1.93369776e-01 -1.12295404e-01 8.09692919e-01 1.49021995e+00
2.13030681e-01 6.77473247e-02 -8.45053017e-01 2.89046049e-01
-1.45158970e+00 -6.00293398e-01 -3.21768314e-01 2.39766145e+00
1.21406627e+00 4.86424208e-01 -1.40837207e-01 4.67346609e-01
5.14373593e-02 5.31221814e-02 -3.96155715e-01 -9.94158328e-01
-1.05242230e-01 7.38332331e-01 1.44548488e+00 5.13217628e-01
-8.57392550e-01 9.33566213e-01 6.43177271e+00 1.74061787e+00
-1.43501663e+00 6.40026033e-01 4.76389885e-01 -2.90513247e-01
-3.30668628e-01 -5.52934520e-02 -1.20090985e+00 4.72507119e-01
1.51474011e+00 7.90254623e-02 5.76777756e-01 7.63738334e-01
-1.44919872e-01 2.43129134e-02 -9.27323699e-01 1.15405309e+00
-3.10887456e-01 -1.46576154e+00 1.09987289e-01 1.76087618e-01
4.51310784e-01 4.10585463e-01 3.91851306e-01 3.61850768e-01
2.53135413e-01 -8.27311695e-01 9.89131808e-01 1.88057929e-01
1.11393797e+00 -7.85652280e-01 5.30806959e-01 2.21804589e-01
-1.45385480e+00 -1.56036302e-01 -7.63266027e-01 -4.13999408e-02
-1.11233823e-01 6.99299991e-01 -6.76783979e-01 1.20999373e-01
5.87209404e-01 -9.04513896e-02 -4.25388038e-01 5.28655887e-01
1.07178733e-01 9.06172633e-01 -7.02763915e-01 -1.15041137e-01
2.90853322e-01 4.46775645e-01 7.59450197e-02 1.59040463e+00
3.73748153e-01 -1.90836713e-01 2.20730349e-01 5.30862212e-01
-2.14021593e-01 2.44422942e-01 9.92007926e-02 5.77510111e-02
9.80770826e-01 1.00339663e+00 -4.30647224e-01 -6.70824647e-01
-4.97184485e-01 9.70335066e-01 2.86778301e-01 1.51092216e-01
-8.45241249e-01 -5.14172196e-01 9.02063549e-01 -1.34140834e-01
2.27638572e-01 -6.08690441e-01 -3.90191883e-01 -1.31718814e+00
-1.99203417e-01 -9.33145523e-01 3.66369963e-01 -3.10435355e-01
-9.87375915e-01 2.03081593e-01 -2.00146332e-01 -1.03077686e+00
1.21957539e-02 -8.05887222e-01 -1.18778003e-02 9.16538000e-01
-1.68546641e+00 -9.59794044e-01 3.43516976e-01 7.28557110e-01
1.58586130e-01 -3.41037586e-02 9.47787702e-01 7.14736581e-01
-3.02485645e-01 1.45269394e+00 3.57451767e-01 2.71131918e-02
8.80473614e-01 -1.22022963e+00 5.22035122e-01 6.29615784e-01
1.68382555e-01 1.06986761e+00 4.44058865e-01 -2.15030879e-01
-2.13249302e+00 -9.77614820e-01 8.63139868e-01 -6.53540269e-02
1.10084093e+00 -5.95022857e-01 -9.93730545e-01 2.59086430e-01
-9.85191297e-03 1.32603511e-01 8.16870570e-01 1.26010090e-01
-9.31937397e-01 -4.00141358e-01 -8.58273983e-01 6.23632133e-01
6.24384701e-01 -9.29000318e-01 -1.52844444e-01 3.27553898e-01
9.28888261e-01 -6.55237556e-01 -1.28816295e+00 -4.01222408e-02
4.69732732e-01 -6.09939694e-01 1.19642293e+00 -3.15674514e-01
-3.39785516e-02 -3.41749310e-01 -6.92608297e-01 -8.80036533e-01
-4.12142038e-01 -6.61420882e-01 -5.49125433e-01 1.07947111e+00
3.72967184e-01 -6.55976057e-01 5.93764126e-01 4.57690656e-01
-1.23176470e-01 -6.84672177e-01 -1.20937669e+00 -8.44990253e-01
3.36881191e-01 -1.00339103e+00 8.66435528e-01 7.38457501e-01
2.25168601e-01 2.48936176e-01 -4.84324753e-01 2.66656458e-01
6.40882373e-01 -1.28176212e-01 5.89152634e-01 -6.19966328e-01
-5.49946785e-01 -6.42933011e-01 -4.06414837e-01 -1.49963355e+00
-4.51182500e-02 -1.28127897e+00 -3.60446364e-01 -7.42308736e-01
-9.24302563e-02 -7.82131433e-01 -4.93067533e-01 7.40126610e-01
5.88752478e-02 5.80678105e-01 2.61960685e-01 2.56782919e-01
-6.68358207e-01 2.76953936e-01 8.62579048e-01 -4.68312114e-01
1.42848834e-01 -5.35926282e-01 -5.39270461e-01 5.15538454e-01
7.93191850e-01 -2.15585887e-01 -3.27915728e-01 -6.26798868e-01
6.63048208e-01 -2.81395137e-01 1.44048825e-01 -1.07500303e+00
3.15358281e-01 1.57422721e-01 1.00452565e-01 -4.63845193e-01
4.81802076e-01 -5.66289604e-01 -2.85203815e-01 7.19412327e-01
-4.09604341e-01 1.15325220e-01 4.87732559e-01 1.57745719e-01
1.80621054e-02 -3.18972498e-01 7.60528445e-01 2.38239899e-01
-9.11944151e-01 1.74694195e-01 -4.72140163e-01 1.35303706e-01
2.08574593e-01 6.46033809e-02 -4.52538341e-01 -4.74347025e-02
-3.78446996e-01 1.63073093e-01 3.19537938e-01 -6.37280243e-03
5.24138451e-01 -1.44142818e+00 -5.35539865e-01 2.05251083e-01
4.63934503e-02 -4.61022079e-01 2.30580077e-01 8.88283372e-01
-5.71671426e-01 6.59125209e-01 2.24701777e-01 -7.16112018e-01
-1.04857922e+00 6.11043930e-01 7.23660588e-02 -4.19888467e-01
-4.23332304e-01 1.07852578e+00 -1.04049847e-01 -1.13167912e-01
3.49923193e-01 -9.44930494e-01 4.85221475e-01 -2.40969732e-01
9.56666887e-01 4.10526931e-01 4.60920244e-01 -4.10005897e-01
-2.92517722e-01 5.17134428e-01 -1.47590622e-01 -1.82641134e-01
7.42412865e-01 -4.03149650e-02 -2.40751892e-01 5.02591252e-01
1.41730392e+00 1.84081912e-01 -7.69315958e-01 -5.45406103e-01
1.23397440e-01 -2.38003969e-01 5.21967530e-01 -3.42934698e-01
-9.73738253e-01 1.38752449e+00 8.18014860e-01 1.16279423e-01
1.13868475e+00 -3.51240575e-01 1.43747377e+00 5.68182945e-01
6.86267436e-01 -1.45470858e+00 1.13562234e-02 7.18200564e-01
2.21923754e-01 -8.84788513e-01 1.13449790e-01 -1.70966417e-01
-3.80625188e-01 1.38844669e+00 2.28719592e-01 7.15957284e-02
6.41742527e-01 6.90461874e-01 -3.46493930e-01 1.29844964e-01
-1.03115523e+00 -7.66881704e-02 2.09927157e-01 2.82101095e-01
4.55292374e-01 4.41593468e-01 -3.28364521e-01 4.41241652e-01
-4.00907665e-01 -1.35070100e-01 3.44284147e-01 7.58309007e-01
-6.96779072e-01 -1.36972785e+00 -5.86689055e-01 6.13806307e-01
-5.82974851e-01 -6.59939051e-01 4.33983326e-01 4.09532726e-01
8.21774900e-02 7.45711982e-01 2.49521270e-01 -7.18928874e-01
6.27445355e-02 1.17194615e-01 5.22216320e-01 -1.79772824e-01
-5.71672976e-01 1.08001426e-01 -5.84898107e-02 -7.19910622e-01
1.15212955e-01 -2.84867197e-01 -1.74094379e+00 -1.04405951e+00
-5.07358670e-01 1.09388903e-02 7.64505208e-01 6.56382561e-01
7.27447510e-01 4.10470575e-01 3.90162051e-01 -4.95010018e-01
-1.16524112e+00 -6.75959170e-01 -5.86109459e-01 1.79731444e-01
3.45213950e-01 -4.14520532e-01 4.73526083e-02 -2.17122346e-01] | [8.672783851623535, 3.464522123336792] |
a571b5cb-051f-48c3-abb4-4561ac86549a | unsupervised-adaptation-with-domain | 1711.08010 | null | http://arxiv.org/abs/1711.08010v2 | http://arxiv.org/pdf/1711.08010v2.pdf | Unsupervised Adaptation with Domain Separation Networks for Robust Speech Recognition | Unsupervised domain adaptation of speech signal aims at adapting a
well-trained source-domain acoustic model to the unlabeled data from target
domain. This can be achieved by adversarial training of deep neural network
(DNN) acoustic models to learn an intermediate deep representation that is both
senone-discriminative and domain-invariant. Specifically, the DNN is trained to
jointly optimize the primary task of senone classification and the secondary
task of domain classification with adversarial objective functions. In this
work, instead of only focusing on learning a domain-invariant feature (i.e. the
shared component between domains), we also characterize the difference between
the source and target domain distributions by explicitly modeling the private
component of each domain through a private component extractor DNN. The private
component is trained to be orthogonal with the shared component and thus
implicitly increases the degree of domain-invariance of the shared component. A
reconstructor DNN is used to reconstruct the original speech feature from the
private and shared components as a regularization. This domain separation
framework is applied to the unsupervised environment adaptation task and
achieved 11.08% relative WER reduction from the gradient reversal layer
training, a representative adversarial training method, for automatic speech
recognition on CHiME-3 dataset. | ['Zhuo Chen', 'Yifan Gong', 'Zhong Meng', 'Vadim Mazalov', 'Jinyu Li'] | 2017-11-21 | null | null | null | null | ['robust-speech-recognition'] | ['speech'] | [ 6.21707022e-01 3.26437950e-01 1.03789657e-01 -5.42453349e-01
-1.05770802e+00 -7.35312343e-01 5.38668811e-01 -4.90864933e-01
-4.05137837e-01 5.97122133e-01 4.16853189e-01 -1.19441412e-01
1.70388773e-01 -4.92000252e-01 -8.23035896e-01 -1.05542600e+00
3.21337998e-01 4.97262120e-01 -5.67403249e-02 -3.31016570e-01
-4.73539919e-01 5.30548692e-01 -1.21036983e+00 1.80935770e-01
7.21909761e-01 9.69723165e-01 3.18712920e-01 6.27632439e-01
-6.94173649e-02 4.04925019e-01 -8.93056691e-01 1.73319522e-02
3.39417219e-01 -7.40776777e-01 -6.29552901e-01 1.94517523e-02
2.00298205e-01 -1.62338302e-01 -6.57314956e-01 1.25050962e+00
6.57877803e-01 3.80686164e-01 8.94937634e-01 -7.72166014e-01
-6.41605914e-01 5.60904801e-01 -2.13635620e-02 3.57188024e-02
-9.95309502e-02 -1.24108516e-01 8.27352047e-01 -8.96950603e-01
5.03869653e-01 1.15908921e+00 1.11875594e-01 9.77455795e-01
-1.41814291e+00 -7.99143553e-01 4.28397618e-02 1.13323092e-01
-1.15701354e+00 -8.04301262e-01 1.21272135e+00 -4.71576750e-01
7.92085111e-01 1.49379402e-01 -6.96782246e-02 1.66653478e+00
-2.80517697e-01 5.39006174e-01 9.40071404e-01 -4.66076136e-01
4.33505356e-01 3.06314617e-01 -1.80584714e-01 -1.55996718e-02
-4.96075034e-01 5.74327052e-01 -6.12261713e-01 1.16571158e-01
3.46573353e-01 -3.69607061e-01 -3.24613392e-01 -5.36435068e-01
-8.21816862e-01 6.97539568e-01 2.82406211e-01 2.71500796e-01
-2.99913108e-01 -2.15365082e-01 4.47797954e-01 5.37915468e-01
4.57270622e-01 1.84928834e-01 -6.12742484e-01 -2.48509590e-02
-6.75209582e-01 -8.13506767e-02 7.23927021e-01 8.38185847e-01
6.17557287e-01 6.87584817e-01 -1.17444761e-01 1.40697932e+00
3.79878372e-01 8.35348427e-01 8.09612751e-01 -6.42605901e-01
4.65493321e-01 6.66979551e-02 -2.49654293e-01 -2.14434430e-01
-8.97251591e-02 -6.50444150e-01 -8.10272455e-01 4.30050939e-01
5.46117485e-01 -3.26924115e-01 -1.22934246e+00 2.23353529e+00
1.73363999e-01 2.44350851e-01 6.11942887e-01 8.32736671e-01
6.78123355e-01 6.72954679e-01 5.87700680e-02 3.92044298e-02
1.01422107e+00 -7.72522211e-01 -5.59164464e-01 -7.25308955e-01
3.01362932e-01 -7.46462405e-01 1.02748311e+00 2.24166378e-01
-8.99468064e-01 -6.29355669e-01 -1.39404738e+00 6.06335625e-02
-4.06861842e-01 1.27088115e-01 -3.76073420e-01 6.70316637e-01
-6.64663911e-01 2.97607780e-01 -7.20152795e-01 -9.01825652e-02
1.78691238e-01 3.75292450e-01 -4.72386092e-01 -2.38845143e-02
-1.57914782e+00 7.21299946e-01 4.61269230e-01 -3.53312254e-01
-1.36698925e+00 -7.00070083e-01 -9.99497890e-01 8.41098130e-02
-2.55069267e-02 -2.79075086e-01 1.14020813e+00 -1.44114494e+00
-2.09285736e+00 8.11795771e-01 -1.29523546e-01 -5.94814897e-01
9.95786712e-02 1.47745917e-02 -7.99022079e-01 -1.10332994e-02
-5.56154512e-02 1.87527642e-01 1.41073358e+00 -1.19248164e+00
-3.35650116e-01 -5.62506497e-01 -4.29475367e-01 4.07577604e-01
-5.10511160e-01 3.16106738e-03 -5.29500321e-02 -1.04762745e+00
2.90989578e-01 -8.77463996e-01 1.08511887e-01 -3.98553818e-01
-2.60993809e-01 1.57553419e-01 1.10701907e+00 -1.18418562e+00
7.82730877e-01 -2.72644186e+00 5.71985066e-01 3.40637386e-01
-2.04916492e-01 4.32302952e-01 -3.69486839e-01 5.80628030e-02
-5.01829028e-01 -3.95510942e-01 -6.30000293e-01 -4.84534562e-01
1.71864092e-01 3.70511502e-01 -5.61506748e-01 5.69218338e-01
4.05025840e-01 3.87038261e-01 -7.71568179e-01 2.74685442e-01
1.99110895e-01 6.04174674e-01 -4.63625133e-01 4.12971377e-01
-1.25813842e-01 9.38686430e-01 -1.85459033e-02 3.47163558e-01
8.92821789e-01 5.81247926e-01 3.09752554e-01 7.28663057e-02
3.27946603e-01 7.66995430e-01 -1.10093975e+00 1.73054445e+00
-6.99924707e-01 5.91086864e-01 5.82251966e-01 -1.41288829e+00
1.30757022e+00 4.93091613e-01 3.26583624e-01 -8.44606817e-01
-1.57697886e-01 4.77370203e-01 1.87322736e-01 -4.85570822e-03
-9.73767601e-03 -6.05167270e-01 -2.90551722e-01 9.16597545e-02
6.82563424e-01 -2.69438148e-01 -6.59927666e-01 -1.65705547e-01
1.10077715e+00 -5.61186410e-02 2.86150724e-01 -2.33501792e-01
7.06598699e-01 -4.97383952e-01 6.06502473e-01 3.42310488e-01
-3.38231683e-01 6.47396922e-01 2.63741612e-01 2.44011939e-01
-1.06676090e+00 -1.72820485e+00 -9.60642248e-02 1.35061109e+00
-1.48480311e-01 2.36506924e-01 -7.63663828e-01 -8.69631648e-01
-8.17844719e-02 9.98388469e-01 -2.10975334e-01 -6.48980021e-01
-8.00593197e-01 -4.85523343e-02 8.87615740e-01 4.93759453e-01
3.31137955e-01 -9.28286672e-01 2.09162951e-01 2.98685730e-01
-1.69695050e-01 -1.24798334e+00 -6.61005437e-01 9.20191169e-01
-4.39843655e-01 -5.94822824e-01 -7.95980096e-01 -9.74837124e-01
3.55644464e-01 -2.12045357e-01 8.57934237e-01 -8.46565306e-01
3.61745596e-01 3.18858325e-01 -1.93224058e-01 -4.67036933e-01
-9.97650445e-01 -7.90169314e-02 4.22836423e-01 3.59314680e-01
4.31949049e-01 -9.92217541e-01 -1.04293264e-01 3.01957905e-01
-9.73750353e-01 -6.47398472e-01 4.10630077e-01 1.21154034e+00
6.43599689e-01 6.40101060e-02 8.50865901e-01 -4.65462327e-01
3.15379530e-01 -5.17465532e-01 -4.74060804e-01 -2.09023908e-01
-1.14599079e-01 2.99347460e-01 1.01212156e+00 -5.76287627e-01
-1.41953945e+00 2.65050501e-01 -5.76019645e-01 -5.41215181e-01
-4.92321312e-01 1.94696918e-01 -1.02882755e+00 2.84349710e-01
9.11674738e-01 5.66343129e-01 -3.78510132e-02 -6.64670527e-01
3.61493319e-01 9.86114085e-01 8.31204295e-01 -4.11264032e-01
1.08041143e+00 3.67622763e-01 -3.78893644e-01 -1.23020160e+00
-4.82681602e-01 -5.27497590e-01 -7.47881532e-01 2.42174461e-01
8.57100666e-01 -9.89187539e-01 1.98782891e-01 6.51358187e-01
-1.13882470e+00 -4.55004066e-01 -6.59155309e-01 6.66418612e-01
-6.54078245e-01 2.78662473e-01 -3.31684388e-02 -6.21626616e-01
1.80141460e-02 -1.08135974e+00 7.99798667e-01 -1.87247634e-01
-1.18287660e-01 -1.08599675e+00 2.56955653e-01 1.74004748e-01
3.30532193e-01 -1.21124275e-01 1.02588630e+00 -1.35317755e+00
5.45179173e-02 -6.28060102e-02 2.34479129e-01 1.42567277e+00
3.32180709e-01 -5.82652092e-01 -1.57564020e+00 -3.46366107e-01
5.27301490e-01 -1.60638750e-01 8.69966328e-01 3.59679252e-01
8.79784048e-01 -2.61172146e-01 -7.48883635e-02 9.66441572e-01
9.74909008e-01 4.88807082e-01 6.70969725e-01 8.20119157e-02
5.97383440e-01 5.61010540e-01 2.91186690e-01 1.02697439e-01
-1.30677685e-01 9.53482807e-01 2.42718339e-01 -1.23901013e-02
-6.82296336e-01 -3.61361533e-01 9.03805614e-01 8.23949575e-01
4.67070311e-01 -8.33042637e-02 -8.37462246e-01 6.48351133e-01
-1.28022289e+00 -7.41809666e-01 6.19585574e-01 2.31481791e+00
8.30811799e-01 1.88439459e-01 -4.05278765e-02 1.76461816e-01
8.21309328e-01 2.47043252e-01 -7.81192362e-01 -5.74964523e-01
-3.28504324e-01 6.05091393e-01 3.67911577e-01 5.60980439e-01
-1.12738955e+00 9.52087402e-01 5.19572449e+00 1.00023198e+00
-1.29694772e+00 2.94606388e-01 8.97912532e-02 -5.09070009e-02
-2.86124259e-01 -2.25970656e-01 -5.63691437e-01 4.59965497e-01
1.34267414e+00 -8.99209008e-02 6.95687950e-01 9.37474787e-01
5.00040129e-02 6.65795147e-01 -1.33333182e+00 7.08321929e-01
-8.89102146e-02 -6.97507143e-01 -1.18963316e-01 1.29020125e-01
6.33412838e-01 2.25789830e-01 4.44652289e-01 5.59103310e-01
3.35681260e-01 -1.04161060e+00 8.03872406e-01 1.83255777e-01
1.09207630e+00 -8.54914486e-01 3.99630487e-01 3.91066968e-01
-9.38472092e-01 -1.25539139e-01 -2.86724657e-01 1.46480083e-01
-5.59630319e-02 4.05092984e-01 -1.13096559e+00 3.35147083e-01
5.84762871e-01 5.55996537e-01 1.27569273e-01 1.93163604e-01
-3.56802434e-01 9.52764332e-01 -1.93959817e-01 6.16891742e-01
5.76837882e-02 -2.07923666e-01 1.23876143e+00 1.23173916e+00
1.51053637e-01 -2.81392097e-01 -1.84699148e-01 8.55024457e-01
-3.71551365e-01 -2.75221735e-01 -8.58208299e-01 -8.77983794e-02
6.70633554e-01 6.82208896e-01 1.05677836e-01 -1.01706758e-01
-3.39825839e-01 1.29850221e+00 3.46348822e-01 7.40827203e-01
-6.92349792e-01 -5.37126899e-01 1.34301436e+00 -1.80370092e-01
6.39571369e-01 -2.97150642e-01 -2.42115974e-01 -8.86656702e-01
1.40035778e-01 -1.09320068e+00 1.30389348e-01 -3.24060678e-01
-1.37703741e+00 6.85877383e-01 -2.39058599e-01 -1.27416396e+00
-6.10643506e-01 -6.13499045e-01 -6.82615578e-01 1.52328479e+00
-1.58441961e+00 -1.20636284e+00 1.78512663e-01 9.61762190e-01
6.15859866e-01 -7.35867620e-01 1.14260173e+00 3.35857272e-01
-4.04005468e-01 9.74635422e-01 8.03173959e-01 3.24446321e-01
1.08854842e+00 -1.38168573e+00 5.24775684e-01 1.04289758e+00
7.80258551e-02 3.30261469e-01 6.58826411e-01 -3.54158968e-01
-8.09288740e-01 -1.34014308e+00 6.01855099e-01 -2.85208762e-01
6.39152706e-01 -7.56009161e-01 -9.95578110e-01 6.11038923e-01
-1.39733195e-01 8.99121389e-02 7.10770607e-01 -2.42294148e-01
-6.51997209e-01 -3.44892353e-01 -1.17246521e+00 4.69932348e-01
9.21713412e-01 -1.19022381e+00 -7.57963359e-01 1.31351184e-02
1.02816737e+00 -3.32141340e-01 -7.35688150e-01 2.21886039e-01
2.25145414e-01 -6.18806541e-01 1.29678249e+00 -7.38629162e-01
2.14543313e-01 -1.34611368e-01 -7.11148679e-01 -1.74432194e+00
-7.37091228e-02 -4.15341109e-01 -1.71965390e-01 1.45831525e+00
3.95442158e-01 -7.89084792e-01 5.97041309e-01 1.73691049e-01
-5.65188289e-01 -1.17904998e-01 -1.41371453e+00 -1.16243732e+00
6.15661502e-01 -5.24423659e-01 5.39012372e-01 7.77044654e-01
-2.61798620e-01 4.71602678e-01 -2.54420012e-01 6.54510021e-01
4.94048625e-01 -4.95232195e-01 5.23731768e-01 -8.89031947e-01
-5.61517537e-01 -2.18720660e-01 -3.73021245e-01 -1.17141235e+00
7.29858398e-01 -1.15808904e+00 3.41404706e-01 -1.01041281e+00
-5.37790298e-01 -2.37165079e-01 -5.48659146e-01 1.39036968e-01
2.15886623e-01 -8.52233842e-02 -2.49917293e-03 7.00606406e-02
1.41467243e-01 8.16133201e-01 9.64948595e-01 -5.70214808e-01
-4.09492403e-01 3.37150574e-01 -5.74077547e-01 7.95015037e-01
8.32338572e-01 -6.54171050e-01 -5.25429666e-01 -3.60418141e-01
-5.79533637e-01 6.79604039e-02 4.06255931e-01 -1.04232407e+00
9.79151428e-02 -4.55668457e-02 2.37922728e-01 -1.74031526e-01
6.27621233e-01 -9.72113132e-01 -2.06081733e-01 2.34491974e-01
-4.01925355e-01 -9.61511254e-01 3.07011604e-01 7.56044447e-01
-5.32800376e-01 -2.47827590e-01 1.25325811e+00 2.96748549e-01
-7.98663795e-01 -6.07020268e-03 -3.18164676e-01 1.69912085e-01
7.30767846e-01 -5.82587793e-02 5.26709445e-02 -3.47851187e-01
-1.08445239e+00 -4.16661352e-01 1.42670378e-01 4.71309721e-01
5.78216970e-01 -1.51649034e+00 -8.89509797e-01 6.74645305e-01
8.53751004e-02 -1.60565991e-02 3.52015108e-01 2.65116274e-01
1.67588085e-01 2.71519393e-01 -1.61389127e-01 -5.60607016e-01
-1.11074519e+00 3.99707317e-01 7.47004747e-01 -2.66593218e-01
-3.20333898e-01 1.07042027e+00 7.39930809e-01 -9.04236138e-01
3.30302805e-01 6.99813850e-03 -7.67390355e-02 -1.95636988e-01
2.12678581e-01 3.05064142e-01 3.55930507e-01 -8.92346025e-01
-6.02482736e-01 1.86517447e-01 1.34168774e-01 -4.68932152e-01
1.25578797e+00 -2.80308217e-01 2.11958423e-01 2.46885568e-01
1.62900102e+00 3.04386884e-01 -1.48019469e+00 -5.83168566e-01
-7.75698423e-02 3.05334069e-02 1.40230373e-01 -1.02075016e+00
-7.76317298e-01 9.81239021e-01 8.01612914e-01 -6.37312159e-02
1.34292471e+00 1.80745438e-01 7.25159407e-01 1.44525498e-01
-1.14300668e-01 -1.30041313e+00 2.33067989e-01 9.98928010e-01
1.06798410e+00 -1.14393938e+00 -6.75793767e-01 -2.05428824e-02
-7.20697045e-01 9.77173328e-01 3.87587190e-01 -1.71142608e-01
8.61881495e-01 3.09874386e-01 1.86737955e-01 1.35215700e-01
-1.76461115e-01 -1.42478257e-01 6.53161108e-01 1.25511372e+00
1.63502723e-01 2.97832191e-01 3.71747851e-01 1.08924901e+00
-2.94388145e-01 -6.71685696e-01 6.20342232e-02 4.24831957e-01
-3.09862822e-01 -1.38200748e+00 -4.43408281e-01 2.28229836e-02
-2.75569737e-01 -1.46227583e-01 -5.48697293e-01 4.24952745e-01
3.09239775e-02 9.40531969e-01 5.49007133e-02 -5.13183236e-01
6.05379820e-01 4.31361407e-01 1.06971599e-01 -8.59683573e-01
-2.02008829e-01 1.18139833e-01 1.76993888e-02 -2.19261453e-01
9.15294886e-02 -6.05573416e-01 -1.26559782e+00 2.82597601e-01
5.08076847e-02 -9.54152942e-02 9.59262490e-01 1.07939363e+00
1.19801693e-01 8.30347478e-01 9.32045758e-01 -7.22887337e-01
-1.10923302e+00 -9.51298535e-01 -8.92844498e-01 5.59992790e-01
7.48506248e-01 -4.90783155e-01 -6.06315851e-01 1.78708777e-01] | [14.551300048828125, 6.386044979095459] |
95ca8973-5ae2-4284-8ff3-3cf839301b2f | behavior-cloned-transformers-are | 2210.07382 | null | https://arxiv.org/abs/2210.07382v2 | https://arxiv.org/pdf/2210.07382v2.pdf | Behavior Cloned Transformers are Neurosymbolic Reasoners | In this work, we explore techniques for augmenting interactive agents with information from symbolic modules, much like humans use tools like calculators and GPS systems to assist with arithmetic and navigation. We test our agent's abilities in text games -- challenging benchmarks for evaluating the multi-step reasoning abilities of game agents in grounded, language-based environments. Our experimental study indicates that injecting the actions from these symbolic modules into the action space of a behavior cloned transformer agent increases performance on four text game benchmarks that test arithmetic, navigation, sorting, and common sense reasoning by an average of 22%, allowing an agent to reach the highest possible performance on unseen games. This action injection technique is easily extended to new agents, environments, and symbolic modules. | ['Prithviraj Ammanabrolu', 'Marc-Alexandre Côté', 'Peter Jansen', 'Ruoyao Wang'] | 2022-10-13 | null | null | null | null | ['common-sense-reasoning'] | ['reasoning'] | [ 2.84540597e-02 4.73713517e-01 7.07402676e-02 8.57342184e-02
-3.95558357e-01 -8.37198377e-01 8.53327513e-01 6.48825690e-02
-5.60721159e-01 6.66621983e-01 8.45236853e-02 -6.34345114e-01
7.90640637e-02 -1.24676108e+00 -4.08065826e-01 -1.30484626e-01
-3.89540911e-01 9.45733666e-01 1.01308203e+00 -1.02039587e+00
3.33849430e-01 1.05074644e-01 -1.70538855e+00 3.86675000e-01
8.02585006e-01 2.13739321e-01 -2.84804702e-02 1.04127526e+00
-2.30730444e-01 1.57970047e+00 -8.75083745e-01 -3.27701718e-01
4.15979803e-01 -4.49004412e-01 -9.79584575e-01 -3.69070053e-01
-1.39264330e-01 -7.34239042e-01 -2.89905310e-01 9.71648633e-01
1.67708665e-01 3.93255055e-01 3.54206800e-01 -1.72471356e+00
-3.41197774e-02 1.14167869e+00 -4.56441194e-02 -1.48986906e-01
1.04965103e+00 6.56298041e-01 6.96086287e-01 -3.35747153e-02
6.84206784e-01 1.45605421e+00 7.05119908e-01 6.55731320e-01
-9.36268926e-01 -4.61552441e-01 1.98823199e-01 -3.19786333e-02
-1.24819207e+00 -3.63194376e-01 5.60725294e-02 -4.38419938e-01
1.70023167e+00 3.05713594e-01 7.68951178e-01 7.39290595e-01
1.45173669e-01 7.37250149e-01 7.34379709e-01 -3.04271400e-01
5.82889020e-01 -2.35778820e-02 -3.39504443e-02 9.77092922e-01
1.90245584e-01 3.44759107e-01 -5.13002515e-01 -4.03586686e-01
1.00628936e+00 -3.35816741e-01 3.14094782e-01 -4.89201754e-01
-1.17348719e+00 6.84838653e-01 1.33739367e-01 2.17152178e-01
-2.49213457e-01 5.17167091e-01 6.23564482e-01 5.57514310e-01
-1.87455058e-01 1.19818985e+00 -4.53259975e-01 -8.18703234e-01
-5.97754680e-02 7.95921743e-01 1.32482183e+00 1.09115016e+00
3.76708716e-01 2.12565526e-01 7.95597360e-02 1.84352726e-01
1.36182934e-01 4.69860792e-01 5.64624012e-01 -1.58089411e+00
3.03674608e-01 1.17406118e+00 3.99764717e-01 -6.22136950e-01
-7.57380068e-01 -8.83999094e-02 3.28461975e-02 8.91981900e-01
4.97757465e-01 -3.44646901e-01 -5.37954390e-01 1.64259195e+00
2.18431085e-01 1.95834041e-01 4.14747924e-01 5.78475237e-01
7.58315861e-01 3.53058010e-01 3.29393983e-01 1.96123406e-01
1.35920262e+00 -1.02858794e+00 -1.63299665e-01 -6.52115464e-01
1.37630653e+00 -2.00532433e-02 1.10101998e+00 3.63277644e-01
-1.27285981e+00 -1.74811170e-01 -1.08990073e+00 2.73741961e-01
-5.19747317e-01 -6.13254368e-01 1.05850816e+00 7.60362566e-01
-1.25643444e+00 3.52677703e-01 -1.14488006e+00 -3.55953515e-01
-1.51205540e-01 6.88980281e-01 -8.15412104e-02 1.90898672e-01
-1.00093925e+00 1.17513621e+00 7.54193485e-01 -6.48411155e-01
-9.65186536e-01 -3.21631461e-01 -1.20926845e+00 -6.70570508e-02
8.01912129e-01 -8.65890920e-01 1.75413752e+00 -6.08681917e-01
-1.86175644e+00 5.37629306e-01 2.65240431e-01 -6.72089517e-01
5.92592061e-01 -2.06663068e-02 2.65925638e-02 -1.89624146e-01
3.36113513e-01 7.73606479e-01 1.38949960e-01 -8.54008257e-01
-8.86600554e-01 -1.88581035e-01 1.24586797e+00 5.92938364e-01
2.74268031e-01 4.77770492e-02 -3.27213645e-01 -1.48979187e-01
4.66413982e-02 -1.10854781e+00 -4.73946035e-01 -3.35496992e-01
-9.25085545e-02 1.56299266e-05 3.80201668e-01 -3.86850178e-01
6.73220038e-01 -2.09363699e+00 4.05747056e-01 2.73492992e-01
3.00665885e-01 -1.10075019e-01 -2.94235498e-01 3.50721419e-01
3.24969471e-01 3.41499187e-02 -2.13912837e-02 1.48879260e-01
5.02386689e-01 3.42791736e-01 -6.72650263e-02 -1.22886546e-01
-2.64842957e-01 1.17801905e+00 -1.27984035e+00 -3.16612512e-01
3.73792946e-01 -1.23167202e-01 -1.05438614e+00 -4.16197218e-02
-7.87264109e-01 9.53190774e-02 -4.86986548e-01 3.89152437e-01
-4.23334120e-03 -2.96563748e-02 5.50363183e-01 7.28814006e-01
9.70890373e-02 6.72716856e-01 -1.25095582e+00 1.84184122e+00
-4.70385939e-01 4.00342256e-01 -2.41799161e-01 -1.02619402e-01
4.66803044e-01 1.87031835e-01 1.20097250e-01 -1.08258450e+00
2.67912485e-02 1.14497148e-01 5.42381763e-01 -1.81300014e-01
7.41714180e-01 1.23373047e-01 -5.72942197e-01 7.01223731e-01
-1.74990818e-01 -6.26550078e-01 6.29435539e-01 4.21566308e-01
1.80065691e+00 2.53429621e-01 4.52019691e-01 2.67764889e-02
3.03379923e-01 5.72517514e-01 1.10241637e-01 1.29954159e+00
3.63533050e-02 -2.96930730e-01 7.79516518e-01 -4.49675083e-01
-9.22698557e-01 -1.09149933e+00 6.43865585e-01 1.77902019e+00
2.70900995e-01 -8.47385585e-01 -7.91617632e-01 -4.51130837e-01
-1.45862103e-01 1.11886632e+00 -4.60864425e-01 -4.27413225e-01
-6.90139651e-01 -3.24829310e-01 1.21882427e+00 8.80061448e-01
8.01080644e-01 -1.20910907e+00 -1.12459064e+00 3.30416352e-01
-1.46088287e-01 -1.00571823e+00 1.22071825e-01 1.79600149e-01
-6.89967453e-01 -1.15326977e+00 6.77182227e-02 -6.91845298e-01
2.98520565e-01 -2.26174947e-02 1.22899222e+00 5.73256910e-01
1.13238946e-01 5.55430293e-01 -4.27650869e-01 -1.65824667e-01
-8.91195893e-01 3.05551976e-01 -8.61079544e-02 -1.24685693e+00
3.03150773e-01 -5.20232499e-01 3.24834324e-02 4.67273802e-01
-5.00354230e-01 4.14810002e-01 1.66655704e-01 5.78626931e-01
-3.68627816e-01 1.57179072e-01 2.75558028e-02 -6.57429397e-01
8.94989848e-01 -2.60200799e-01 -9.46880996e-01 -4.97965999e-02
-1.08939238e-01 2.32248232e-01 3.08196574e-01 -3.04678261e-01
-7.45462537e-01 -1.78092659e-01 1.86224245e-02 5.59980392e-01
-1.91571802e-01 4.88199979e-01 -1.63356543e-01 -1.78006873e-01
1.25371027e+00 3.38194668e-01 -2.07023527e-02 3.14154923e-01
4.98181254e-01 2.15631098e-01 7.49564409e-01 -1.16621339e+00
7.22151101e-01 8.92399251e-02 -1.17723316e-01 -3.78839910e-01
2.71618627e-02 -1.03815190e-01 -7.68157467e-02 6.24492913e-02
4.95535165e-01 -7.53624082e-01 -1.45372963e+00 5.60592771e-01
-1.00755632e+00 -1.18372536e+00 -4.32042986e-01 1.30270571e-01
-9.21215594e-01 4.64474298e-02 -8.65816236e-01 -7.88973570e-01
1.80580720e-01 -1.48733902e+00 9.55306709e-01 2.32028916e-01
-7.04117954e-01 -7.33208597e-01 2.71156311e-01 8.05436671e-02
3.84672254e-01 -1.56613290e-01 9.04754460e-01 -1.01313055e+00
-6.46633446e-01 -1.48925427e-02 1.17479235e-01 -4.40713972e-01
-1.22735493e-01 -4.54933077e-01 -6.09682918e-01 -2.51450930e-02
-3.53493959e-01 -3.67696583e-01 6.14958443e-02 -1.39727637e-01
2.40715235e-01 -2.66967177e-01 -4.31854159e-01 1.93307668e-01
9.17956531e-01 8.08126450e-01 9.38712537e-01 1.03216374e+00
1.93301037e-01 6.62717372e-02 6.26029909e-01 5.23010433e-01
5.83352685e-01 8.63019228e-01 4.48853046e-01 5.14391959e-01
2.92496085e-01 -1.36475652e-01 5.32007337e-01 -1.46584958e-01
-4.39990908e-01 -1.93178356e-01 -1.23395848e+00 3.54138076e-01
-2.11174273e+00 -1.10405278e+00 2.17111617e-01 1.76782489e+00
9.43695247e-01 5.62310159e-01 4.69878286e-01 -3.67396362e-02
1.40709803e-01 -2.57963359e-01 -4.91633773e-01 -5.04937053e-01
2.97758698e-01 3.44472796e-01 3.95885050e-01 8.54908466e-01
-8.22834134e-01 1.45330453e+00 6.99725580e+00 5.06226003e-01
-6.48671389e-01 -9.01463479e-02 -4.00025621e-02 3.47732641e-02
-4.12854441e-02 -9.90639850e-02 -4.89908665e-01 1.97483212e-01
8.99150789e-01 -3.81295383e-01 8.69975984e-01 1.09520411e+00
-1.54169977e-01 -5.73201120e-01 -1.26428032e+00 3.63728523e-01
-2.80956775e-01 -1.19650900e+00 -2.29952902e-01 -8.34363699e-02
4.26954895e-01 5.57962805e-02 -2.75276811e-03 9.93173122e-01
1.41615617e+00 -1.14311337e+00 8.48311186e-01 1.40903115e-01
2.29023337e-01 -6.75689161e-01 8.21198225e-01 4.93672073e-01
-1.00973916e+00 -1.38377354e-01 1.01923920e-01 -9.94908452e-01
-1.58710182e-01 -7.07697451e-01 -1.29085290e+00 1.49908319e-01
3.23925883e-01 9.92142931e-02 -7.26614773e-01 1.08045208e+00
-5.45840621e-01 1.72631204e-01 -4.49770361e-01 -2.16081440e-01
3.26548308e-01 4.93149646e-02 5.86439490e-01 7.82366157e-01
-1.13079054e-02 4.35579211e-01 3.54376197e-01 6.69910908e-01
4.33903813e-01 -3.91785353e-01 -7.96743035e-01 -4.52156179e-02
6.56810164e-01 6.64447010e-01 -9.49317217e-01 -6.22580290e-01
-1.96523964e-01 8.68581533e-01 -5.50347986e-03 2.23614454e-01
-1.04711950e+00 -3.78690958e-01 9.59241748e-01 1.31285191e-01
-2.00166315e-01 -5.98973334e-01 -4.58399355e-01 -7.93410361e-01
-2.32304379e-01 -1.46465230e+00 3.18008780e-01 -1.18541157e+00
-4.02931303e-01 5.41786313e-01 1.69311568e-01 -7.63142347e-01
-1.23830390e+00 -6.61419809e-01 -4.91528213e-01 5.08417726e-01
-4.18295383e-01 -6.96506679e-01 -5.45311272e-01 5.02046704e-01
4.26470548e-01 -5.84565520e-01 1.17746139e+00 -1.42239809e-01
1.42642176e-02 3.51442248e-01 -4.16882694e-01 3.58982116e-01
3.37026111e-04 -1.36591780e+00 1.19373775e+00 6.30332112e-01
-1.55674413e-01 7.04763293e-01 9.55847263e-01 -6.49901032e-01
-1.26111698e+00 -6.05099082e-01 -1.08628143e-02 -7.56787360e-01
9.25127566e-01 -2.08396375e-01 -5.17025888e-01 1.24169409e+00
1.69042110e-01 -6.88507080e-01 3.42551112e-01 1.95749402e-01
-4.47212994e-01 5.18026710e-01 -1.24053872e+00 1.25640261e+00
1.52530062e+00 -5.41143656e-01 -9.09330249e-01 1.13401271e-01
8.37098062e-01 -1.06882322e+00 -4.40081626e-01 2.84252670e-02
5.11707127e-01 -1.08469033e+00 1.11229455e+00 -8.51883650e-01
4.34422106e-01 -6.11190975e-01 -2.86085159e-01 -1.43487203e+00
-1.00582913e-01 -5.56778491e-01 1.23205781e-01 4.70549017e-01
3.56701046e-01 -9.20520604e-01 8.34406376e-01 9.58558321e-01
-2.64068246e-02 1.49094295e-02 -5.81185281e-01 -6.62009418e-01
-1.39693290e-01 -7.86961794e-01 9.26863432e-01 6.99138820e-01
7.52022088e-01 3.09510350e-01 2.27812648e-01 2.01813132e-01
2.23787382e-01 -2.33078480e-01 1.10238385e+00 -9.40405667e-01
-8.40773821e-01 -8.11116040e-01 -9.02230740e-01 -9.12285864e-01
2.69568622e-01 -7.50209332e-01 1.19413145e-01 -1.45879602e+00
7.89703876e-02 -3.71429980e-01 4.99124646e-01 7.89471507e-01
2.77303338e-01 1.37249961e-01 4.59345907e-01 -1.98810369e-01
-1.13918686e+00 4.81790751e-02 9.61813271e-01 -3.32978457e-01
-4.88116801e-01 -9.84529629e-02 -7.34062850e-01 1.11241066e+00
8.11580122e-01 -1.65177479e-01 -6.08462393e-01 -4.69686687e-01
8.44259620e-01 3.25519502e-01 6.48148477e-01 -1.39908421e+00
5.44060647e-01 -4.77596432e-01 -1.61309049e-01 6.54558763e-02
5.50454795e-01 -5.56025565e-01 1.86940238e-01 8.15309465e-01
-2.73642033e-01 1.72691569e-01 6.92945778e-01 -4.76422496e-02
1.12812117e-01 -3.45888764e-01 1.39039204e-01 -5.14210224e-01
-1.15404129e+00 -5.79118013e-01 -1.11232483e+00 2.21067041e-01
1.11986434e+00 -3.82463634e-01 -7.35934556e-01 -7.27816224e-01
-8.65772188e-01 3.56531590e-01 8.28449488e-01 1.64551511e-01
3.35682273e-01 -8.78382742e-01 -3.70909691e-01 1.11508586e-01
6.84952512e-02 -1.61050320e-01 -2.68009454e-01 5.12559593e-01
-1.05466163e+00 2.23336667e-01 -5.84498882e-01 -3.51552069e-01
-1.17921269e+00 4.33499664e-01 8.50782752e-01 -2.48968348e-01
-5.07801592e-01 6.78318977e-01 1.62873015e-01 -9.58653688e-01
1.18424915e-01 -6.29662275e-01 -3.29143256e-02 -6.63574278e-01
6.18046165e-01 4.17913496e-01 -4.35309485e-02 -1.26418173e-01
-5.98949254e-01 8.07574764e-02 1.00751154e-01 -6.57363176e-01
9.52676058e-01 4.50231224e-01 -1.31107226e-01 6.39860407e-02
9.64824781e-02 9.18684080e-02 -8.29257727e-01 1.67736504e-02
1.10408857e-01 -2.18390524e-01 -4.34341639e-01 -9.96063948e-01
-4.13757265e-01 2.07319394e-01 8.33850652e-02 4.32446480e-01
6.01639807e-01 4.93781865e-02 2.00267538e-01 1.08641839e+00
1.27469182e+00 -7.50493646e-01 1.56902671e-01 1.14294207e+00
5.49385667e-01 -8.21755290e-01 -1.91705063e-01 -3.44774365e-01
-6.82493389e-01 8.16834331e-01 1.07906401e+00 -1.24682873e-01
-1.39687851e-01 9.29302573e-01 -2.24951968e-01 -1.81588218e-01
-8.83156121e-01 -3.44726652e-01 -4.43797290e-01 1.03854764e+00
8.18587393e-02 2.31910452e-01 2.20704809e-01 6.79964721e-01
-6.91580534e-01 1.44629523e-01 9.21783626e-01 1.44331884e+00
-6.86860979e-01 -9.31951284e-01 -4.72921252e-01 1.81348875e-01
1.91514313e-01 -1.37930453e-01 -6.31787956e-01 1.16111386e+00
-6.55286759e-02 9.83385205e-01 2.05279082e-01 -5.98299921e-01
4.04145598e-01 1.77706257e-01 8.24444771e-01 -7.95649052e-01
-8.22209120e-01 -4.99335974e-01 7.01574981e-01 -1.00060642e+00
4.25253883e-02 -6.16788507e-01 -1.89034307e+00 -7.64583647e-01
9.44527909e-02 1.86072156e-01 1.70068711e-01 1.14107823e+00
1.08770087e-01 7.47453094e-01 -4.14859623e-01 -7.17823386e-01
-2.37387180e-01 -7.45861769e-01 -1.37136176e-01 -2.09701285e-02
-2.12610289e-01 -6.88440561e-01 3.21508870e-02 -2.83862501e-01] | [3.6539716720581055, 1.325090765953064] |
27249016-1426-40c0-86d8-9e89b24848a1 | using-knowledge-graphs-for-performance | 2301.09876 | null | https://arxiv.org/abs/2301.09876v1 | https://arxiv.org/pdf/2301.09876v1.pdf | Using Knowledge Graphs for Performance Prediction of Modular Optimization Algorithms | Empirical data plays an important role in evolutionary computation research. To make better use of the available data, ontologies have been proposed in the literature to organize their storage in a structured way. However, the full potential of these formal methods to capture our domain knowledge has yet to be demonstrated. In this work, we evaluate a performance prediction model built on top of the extension of the recently proposed OPTION ontology. More specifically, we first extend the OPTION ontology with the vocabulary needed to represent modular black-box optimization algorithms. Then, we use the extended OPTION ontology, to create knowledge graphs with fixed-budget performance data for two modular algorithm frameworks, modCMA, and modDE, for the 24 noiseless BBOB benchmark functions. We build the performance prediction model using a knowledge graph embedding-based methodology. Using a number of different evaluation scenarios, we show that a triple classification approach, a fairly standard predictive modeling task in the context of knowledge graphs, can correctly predict whether a given algorithm instance will be able to achieve a certain target precision for a given problem instance. This approach requires feature representation of algorithms and problems. While the latter is already well developed, we hope that our work will motivate the community to collaborate on appropriate algorithm representations. | ['Carola Doerr', 'Tome Eftimov', 'Panče Panov', 'Sašo Džeroski', 'Diederick Vermetten', 'Ana Kostovska'] | 2023-01-24 | null | null | null | null | ['triple-classification', 'knowledge-graph-embedding'] | ['graphs', 'graphs'] | [ 2.57888347e-01 1.89058900e-01 -2.73105264e-01 -3.45416695e-01
-8.54284912e-02 -4.18282479e-01 5.19275725e-01 6.85331285e-01
-2.10839882e-01 6.98363245e-01 -8.02602470e-02 -2.13903919e-01
-9.46200609e-01 -1.10135543e+00 -5.73916197e-01 -4.22056079e-01
-2.03367665e-01 5.92592180e-01 1.60016045e-01 -3.77657443e-01
3.73828411e-01 5.51325977e-01 -2.16668057e+00 3.53952765e-01
8.41748655e-01 1.12527931e+00 -1.87191144e-02 6.20173156e-01
-8.81460160e-02 7.17087746e-01 -5.02583563e-01 -5.70871174e-01
1.57158479e-01 -2.87414610e-01 -9.76951957e-01 -9.39313769e-02
2.76400834e-01 5.21461070e-01 -1.38382852e-01 8.00446212e-01
3.92963350e-01 9.70251933e-02 6.43885851e-01 -1.43268192e+00
-3.30106854e-01 7.61286080e-01 2.27565944e-01 8.81150831e-03
3.65278333e-01 -6.98229298e-02 1.35587573e+00 -3.07284951e-01
8.89540553e-01 8.96720469e-01 6.37105405e-01 3.83596152e-01
-1.49708998e+00 -2.32355148e-01 -1.48711473e-01 6.39260709e-01
-1.39156246e+00 -1.00126602e-01 6.72533035e-01 -5.09046376e-01
1.14492559e+00 5.00296474e-01 1.05828667e+00 7.73647547e-01
2.37452552e-01 5.64560711e-01 1.14722860e+00 -7.33172476e-01
5.42689264e-01 5.29375315e-01 5.14556825e-01 8.70032907e-01
5.70147574e-01 4.13390666e-01 -6.97022676e-01 -1.13241255e-01
-1.67443335e-01 -3.14526528e-01 -3.48198056e-01 -9.95630324e-01
-7.46585071e-01 8.05393755e-01 4.27189440e-01 7.24062562e-01
-1.26922131e-01 2.55628020e-01 3.81056070e-01 5.23514450e-01
3.95748556e-01 9.84312773e-01 -5.32537580e-01 -3.36411268e-01
-8.31822038e-01 3.87924880e-01 1.13142872e+00 7.44008958e-01
8.42059314e-01 -1.62378594e-01 -5.04300930e-02 4.95445430e-01
2.62235761e-01 -1.40836254e-01 5.36499918e-01 -6.57745361e-01
2.01310962e-01 1.24390435e+00 -2.12765142e-01 -1.13705361e+00
-5.38679481e-01 -6.82126582e-01 -3.49390417e-01 2.57781625e-01
2.20555142e-01 3.26524377e-01 -4.22987521e-01 1.56273353e+00
2.50634730e-01 2.35062048e-01 2.31129214e-01 5.93623877e-01
5.25134385e-01 2.92137951e-01 -2.44404934e-02 -1.08696654e-01
1.28174567e+00 -7.11448014e-01 -5.40194988e-01 2.45834574e-01
9.39560950e-01 -3.44604135e-01 8.44574749e-01 5.98887682e-01
-8.02088082e-01 -2.78076291e-01 -1.54453719e+00 2.45442137e-01
-9.27136242e-01 2.29314882e-02 8.00744295e-01 1.27880239e+00
-9.32256281e-01 9.39943075e-01 -7.23078966e-01 -4.91531432e-01
3.00347090e-01 4.64524806e-01 -3.00549120e-01 -6.90551177e-02
-1.27499664e+00 1.20064259e+00 9.98020053e-01 -2.18733653e-01
-5.46946049e-01 -8.88507903e-01 -7.37596869e-01 3.31826121e-01
5.07277727e-01 -1.00907552e+00 5.45710444e-01 -9.51423049e-01
-1.59226000e+00 8.26561451e-01 4.26939070e-01 -8.97191465e-01
3.64555210e-01 1.88486487e-01 -7.04866230e-01 -1.46914706e-01
-6.56280994e-01 2.80758768e-01 5.84169269e-01 -1.05983448e+00
-4.74235833e-01 -6.39693558e-01 3.40116948e-01 -9.02424529e-02
-7.41455555e-01 -1.72967851e-01 -2.23378390e-01 -4.97195154e-01
-3.24066550e-01 -1.03642094e+00 -7.91408941e-02 -3.26434612e-01
-1.45604312e-01 -1.30382434e-01 5.56699634e-01 -3.58439952e-01
1.82375669e+00 -1.86538458e+00 7.74389625e-01 5.07907331e-01
1.61938667e-01 2.38921314e-01 2.67508887e-02 5.20967245e-01
-3.32975537e-02 3.43577117e-01 -4.35174972e-01 4.62909341e-02
3.88557196e-01 3.41960400e-01 8.41875002e-02 3.15915197e-01
9.36670229e-02 8.06807578e-01 -5.33936262e-01 -3.06845218e-01
4.04988050e-01 3.82570535e-01 -9.28457558e-01 -1.11950867e-01
-5.02481103e-01 -2.16563046e-01 -4.56523091e-01 3.41680497e-01
2.81669229e-01 -2.84214735e-01 6.63120508e-01 -2.54998147e-01
-3.13626453e-02 -4.47804406e-02 -1.47024846e+00 1.82913792e+00
-5.56309104e-01 5.34991205e-01 -3.87421012e-01 -1.23846316e+00
8.77993464e-01 8.01969543e-02 6.19995654e-01 -5.60839772e-01
2.28614703e-01 2.59094298e-01 2.57190675e-01 -2.55763024e-01
7.35307932e-01 1.10893622e-01 8.57627578e-03 2.11426973e-01
3.03503633e-01 -8.31091329e-02 6.69880986e-01 -1.03826724e-01
1.20901859e+00 2.71018744e-01 6.03680074e-01 -3.81524771e-01
8.80085766e-01 1.99223340e-01 1.70599878e-01 4.89992201e-01
1.39415607e-01 1.47069335e-01 4.81704772e-01 -6.41762078e-01
-8.82835090e-01 -5.83483040e-01 -3.58081758e-01 9.33677971e-01
-1.36000499e-01 -1.01276088e+00 -8.45471859e-01 -6.40783787e-01
2.04390153e-01 8.52460921e-01 -7.48830080e-01 -5.28732121e-01
-2.80071169e-01 -1.06184101e+00 4.90948945e-01 5.47603816e-02
1.43189043e-01 -9.10053074e-01 -1.14973056e+00 2.98448950e-01
2.97340631e-01 -7.40357041e-01 2.44870812e-01 3.28476846e-01
-8.02401602e-01 -1.42763877e+00 -1.37664333e-01 -3.00601006e-01
1.66268110e-01 -4.24063116e-01 1.39053345e+00 3.94338906e-01
-7.65123606e-01 6.54562950e-01 -7.17129588e-01 -3.65031302e-01
-6.34140193e-01 2.73605704e-01 3.18425964e-03 6.01750873e-02
2.79361606e-01 -5.62109470e-01 -2.00954169e-01 2.30836689e-01
-1.16414332e+00 5.79069257e-02 4.18221265e-01 8.66536140e-01
5.18812835e-01 2.83301711e-01 1.88352019e-01 -7.51264811e-01
6.55262172e-01 -4.74018574e-01 -7.96389043e-01 6.72432005e-01
-1.26578152e+00 6.56361997e-01 4.94409591e-01 -1.64703310e-01
-5.49522698e-01 -6.86854720e-02 1.27668843e-01 -3.76839221e-01
-4.06102240e-02 8.67775977e-01 -2.40840584e-01 -3.95675510e-01
8.75492871e-01 9.23885256e-02 -1.72723606e-01 -4.83825624e-01
4.57285225e-01 4.51500148e-01 4.33745012e-02 -9.03062046e-01
5.46263635e-01 7.16414750e-02 4.77951556e-01 -6.61961973e-01
-3.56964856e-01 -2.25082338e-01 -4.16410744e-01 -1.40137583e-01
7.22348571e-01 -3.09422672e-01 -8.64957571e-01 -5.17713577e-02
-8.07204723e-01 -9.87038016e-02 -5.62590539e-01 3.38799089e-01
-7.16721594e-01 2.45102420e-01 1.57522142e-01 -7.29945779e-01
-1.60362870e-01 -1.20036602e+00 6.47287250e-01 3.81490663e-02
-1.25123948e-01 -1.09865582e+00 2.92429000e-01 3.07906657e-01
4.09672111e-01 2.70747751e-01 1.37183905e+00 -9.21803236e-01
-5.15688956e-01 -1.70246944e-01 2.28304550e-01 3.75465125e-01
-2.12266639e-01 1.40825421e-01 -7.52388716e-01 -2.47470632e-01
-2.90034950e-01 -1.34685606e-01 6.79900169e-01 -6.43005818e-02
1.27890766e+00 2.40204781e-02 -4.03240293e-01 5.46248317e-01
1.74107850e+00 2.29949672e-02 6.70561850e-01 6.07817233e-01
1.94023475e-01 5.95555961e-01 4.98016328e-01 4.93849605e-01
1.87014461e-01 1.26119363e+00 5.39292812e-01 5.63304007e-01
-1.13574274e-01 -1.29738182e-01 1.96150959e-01 6.18131816e-01
-2.68432170e-01 -4.49416190e-01 -1.20177209e+00 2.27208525e-01
-1.67190814e+00 -8.53819907e-01 -4.03059758e-02 2.25801420e+00
5.21380484e-01 7.27137700e-02 9.23969224e-03 5.64324379e-01
4.68554646e-01 -4.30390537e-02 -2.57795781e-01 -6.09415948e-01
-1.24250822e-01 5.66321611e-01 3.91112626e-01 2.76249856e-01
-7.77806640e-01 6.31846189e-01 5.66684198e+00 9.18929279e-01
-8.57966304e-01 7.92248524e-04 2.67114788e-01 -5.44634694e-03
-3.88495475e-01 2.16119215e-01 -6.17014050e-01 4.22578573e-01
1.44734037e+00 -4.78377670e-01 6.81869864e-01 8.18843484e-01
-2.79662699e-01 6.96328655e-02 -1.26263547e+00 9.75858629e-01
1.86733663e-01 -1.60088086e+00 4.58307415e-02 1.79668874e-01
4.74558532e-01 -3.29489648e-01 -1.09760724e-01 4.52764124e-01
1.47026762e-01 -1.15762484e+00 6.35837317e-01 6.88424587e-01
3.24019611e-01 -8.32160592e-01 6.55304432e-01 3.36619169e-01
-1.10497701e+00 -2.87082434e-01 -2.49258444e-01 -5.70947714e-02
-2.54372269e-01 5.16083181e-01 -8.70110512e-01 1.25496829e+00
6.00171566e-01 6.59174025e-01 -9.60737526e-01 1.17835534e+00
-6.14376329e-02 4.97141719e-01 -2.33094245e-01 -3.62808526e-01
-1.74594492e-01 -1.67600527e-01 6.74489379e-01 8.57367218e-01
5.00976503e-01 -3.07646543e-01 -2.53238380e-02 8.08874309e-01
-2.30635386e-02 6.59735203e-01 -5.03521919e-01 -3.71201307e-01
1.44782037e-01 1.14314246e+00 -4.48465526e-01 -1.23449944e-01
-1.38177440e-01 6.52290106e-01 4.89841372e-01 -1.08897343e-01
-7.73954809e-01 -3.03336620e-01 6.89970195e-01 -6.59557730e-02
1.78390786e-01 4.45105396e-02 -1.65209606e-01 -1.13295734e+00
-2.94160750e-02 -1.04511118e+00 7.03198254e-01 -6.41977191e-01
-1.08878791e+00 5.81180811e-01 -2.82967426e-02 -1.21593249e+00
-2.95944810e-01 -1.00406182e+00 -3.32015343e-02 6.15225255e-01
-1.26338470e+00 -8.09957623e-01 -3.45489532e-01 2.89910346e-01
8.78978819e-02 -4.77004737e-01 1.19462085e+00 2.93726236e-01
-5.54840088e-01 4.97551441e-01 1.87751025e-01 -4.47554797e-01
2.72343546e-01 -1.33068705e+00 -1.83487937e-01 4.36545879e-01
6.38421118e-01 4.37723428e-01 9.12355602e-01 -3.51282209e-01
-1.78964520e+00 -7.61057198e-01 6.22561514e-01 -5.87245464e-01
6.77589059e-01 -1.23743951e-01 -8.04380655e-01 3.59444648e-01
4.79325131e-02 4.42467891e-02 9.53504384e-01 3.60886753e-01
-3.39745700e-01 -1.48498386e-01 -1.06368113e+00 3.49854529e-01
1.08027279e+00 -1.95764408e-01 -8.15688133e-01 2.29165167e-01
5.49578905e-01 -2.11348593e-01 -1.33078587e+00 5.98472834e-01
5.81474245e-01 -1.04040086e+00 1.03573775e+00 -1.03371036e+00
4.09572750e-01 -3.38750690e-01 -5.16294539e-01 -1.43208635e+00
2.75798216e-02 -3.57039034e-01 -5.55149794e-01 9.99895453e-01
5.30003607e-01 -6.50121868e-01 9.22921598e-01 4.14704591e-01
-5.71677946e-02 -9.33907926e-01 -1.12034500e+00 -1.01116693e+00
3.85263450e-02 -7.28406131e-01 9.97858226e-01 8.07665765e-01
1.67271480e-01 9.69930515e-02 -2.53738940e-01 5.25652319e-02
2.88541108e-01 1.80883095e-01 7.52246916e-01 -1.60802591e+00
-5.84224701e-01 -8.46078098e-01 -1.10888994e+00 -9.87662002e-02
3.45584095e-01 -1.46137393e+00 -7.46952534e-01 -1.28083384e+00
3.20194364e-02 -5.52503943e-01 -6.35204911e-01 2.01535314e-01
1.99019209e-01 5.28768227e-02 2.64520526e-01 -2.39519179e-01
-4.91112828e-01 6.47105575e-01 7.62979448e-01 -3.08715582e-01
-1.40592828e-01 -2.81452000e-01 -4.51073617e-01 3.87271613e-01
7.36668110e-01 -6.09380305e-01 -4.73314106e-01 5.94496951e-02
7.47227967e-01 -1.02969952e-01 2.95989513e-01 -1.35212171e+00
1.53143317e-01 -9.06498060e-02 3.07315160e-02 -1.36464471e-02
2.54032373e-01 -1.25380778e+00 6.45524561e-01 6.78154826e-01
-3.17981243e-01 2.90314518e-02 2.93761849e-01 6.25422418e-01
-3.29468399e-01 -6.13364875e-01 4.37683553e-01 2.19022468e-01
-9.07584429e-01 1.12865016e-01 -3.18918414e-02 -2.44181659e-02
1.30438852e+00 -2.10463554e-01 -1.77419513e-01 2.41553098e-01
-8.35218549e-01 2.62363777e-02 6.70007944e-01 5.55061519e-01
3.51064920e-01 -1.13919842e+00 -5.00621498e-01 4.26950604e-02
6.04950488e-01 -8.24425220e-01 2.39979148e-01 7.35476136e-01
-6.76858246e-01 5.76291084e-01 -3.45444232e-01 -3.90580893e-01
-1.33489275e+00 9.13318336e-01 5.33058047e-01 -6.63788915e-01
-3.36895853e-01 3.93023700e-01 -6.84970260e-01 -5.64107776e-01
-7.45926332e-03 -3.10655951e-01 -2.71864027e-01 1.37748808e-01
2.27302581e-01 5.72125673e-01 5.11639774e-01 -4.12355512e-01
-4.31115448e-01 5.37915289e-01 4.11972731e-01 5.39733395e-02
1.33945179e+00 4.09872830e-01 -2.46264338e-01 3.24335128e-01
1.02000332e+00 2.47091260e-02 -4.00076270e-01 8.45471770e-02
4.29020017e-01 -4.72386211e-01 1.69216618e-01 -1.03357089e+00
-8.94218683e-01 8.14721704e-01 9.72790182e-01 3.90309244e-01
1.25931084e+00 -2.69569904e-01 -5.31954281e-02 5.77090263e-01
7.25293517e-01 -1.10049510e+00 -2.29609162e-01 3.14763874e-01
8.40187907e-01 -8.57743084e-01 2.67251700e-01 -4.60448056e-01
-2.31428578e-01 1.39415228e+00 3.65164846e-01 2.13376522e-01
6.12431824e-01 7.76814148e-02 -5.97808659e-01 -3.48382294e-01
-9.82619941e-01 -3.81175309e-01 6.93646371e-01 3.93005013e-01
3.99728447e-01 2.51699567e-01 -6.19620144e-01 8.35361421e-01
-2.90382951e-01 2.22155452e-01 3.43133658e-01 8.56580853e-01
-3.10627043e-01 -1.66604602e+00 -2.20857233e-01 5.79916298e-01
-1.50365457e-01 -7.12189302e-02 -4.29896921e-01 1.01444709e+00
3.73372138e-01 7.76295662e-01 -3.36911410e-01 -7.13956594e-01
5.70111394e-01 4.57787454e-01 7.91124821e-01 -5.12346148e-01
-8.00658941e-01 -9.17902887e-01 3.11064631e-01 -4.89143372e-01
-4.96960759e-01 -4.59069967e-01 -9.67417598e-01 -2.40328416e-01
-3.36529523e-01 2.84190685e-01 8.90975416e-01 8.21572065e-01
5.09585023e-01 8.81078780e-01 1.47344545e-01 -4.41547483e-01
-4.73722428e-01 -6.02749944e-01 -5.24324894e-01 4.41630810e-01
-4.18262720e-01 -1.02404118e+00 -1.78757861e-01 -3.07404429e-01] | [8.347735404968262, 4.599459171295166] |
52b56cd6-c224-4934-8bd1-e67655a43673 | gated-graph-sequence-neural-networks | 1511.05493 | null | http://arxiv.org/abs/1511.05493v4 | http://arxiv.org/pdf/1511.05493v4.pdf | Gated Graph Sequence Neural Networks | Graph-structured data appears frequently in domains including chemistry,
natural language semantics, social networks, and knowledge bases. In this work,
we study feature learning techniques for graph-structured inputs. Our starting
point is previous work on Graph Neural Networks (Scarselli et al., 2009), which
we modify to use gated recurrent units and modern optimization techniques and
then extend to output sequences. The result is a flexible and broadly useful
class of neural network models that has favorable inductive biases relative to
purely sequence-based models (e.g., LSTMs) when the problem is
graph-structured. We demonstrate the capabilities on some simple AI (bAbI) and
graph algorithm learning tasks. We then show it achieves state-of-the-art
performance on a problem from program verification, in which subgraphs need to
be matched to abstract data structures. | ['Daniel Tarlow', 'Yujia Li', 'Marc Brockschmidt', 'Richard Zemel'] | 2015-11-17 | null | null | null | null | ['sql-to-text'] | ['computer-code'] | [ 6.79083347e-01 6.19195461e-01 -4.20034379e-01 -4.58376497e-01
-1.45211294e-01 -5.95608711e-01 4.21911329e-01 4.07459259e-01
-1.71780005e-01 6.44562721e-01 2.95060873e-02 -1.07201374e+00
1.03813529e-01 -1.30261326e+00 -1.21164048e+00 -1.63004681e-01
-5.33183455e-01 2.85039783e-01 -7.89353549e-02 -3.03660899e-01
2.27411374e-01 4.98407662e-01 -1.31655717e+00 4.13099080e-01
4.16056603e-01 6.10823452e-01 1.43153034e-03 9.43863511e-01
-4.86628950e-01 1.39895678e+00 -4.79625344e-01 -3.12087566e-01
-4.66679148e-02 -3.27993006e-01 -1.23823261e+00 -4.15466040e-01
6.72327459e-01 1.28860269e-02 -6.26356781e-01 1.26107776e+00
2.66670316e-01 2.92187065e-01 5.29515684e-01 -1.23540139e+00
-1.26685178e+00 1.33868456e+00 -2.32088581e-01 1.34058714e-01
4.55159485e-01 1.45144269e-01 1.37847197e+00 -6.00829005e-01
1.00485289e+00 1.39231920e+00 8.43340874e-01 8.17113280e-01
-1.37082791e+00 -5.57961643e-01 6.07588410e-01 2.36839116e-01
-9.06753361e-01 -3.31111610e-01 7.07184553e-01 -4.87043351e-01
1.83090532e+00 -1.28345370e-01 6.94393694e-01 1.16919065e+00
3.37577194e-01 8.97467136e-01 5.40853441e-01 -5.39555311e-01
1.47981551e-02 -4.44076091e-01 6.47440374e-01 1.37575042e+00
4.45087433e-01 2.62661815e-01 -3.08637649e-01 -1.74420580e-01
3.83739293e-01 1.11170094e-02 -1.97481513e-01 -5.35780370e-01
-1.17200351e+00 1.00056207e+00 8.65105033e-01 3.32020074e-01
-5.73772453e-02 8.11009943e-01 5.19348383e-01 8.13905120e-01
2.12252602e-01 8.45193923e-01 -4.65062350e-01 2.79014528e-01
-6.91629827e-01 1.45213440e-01 1.20335650e+00 1.33654165e+00
9.76196527e-01 6.27754152e-01 -1.55363500e-01 1.78593814e-01
4.51504290e-01 1.65495738e-01 4.35077757e-01 -4.39510971e-01
5.28587580e-01 7.59025335e-01 -5.03795564e-01 -8.23264837e-01
-6.34712934e-01 -2.00288638e-01 -8.70507717e-01 1.07957728e-01
1.87747642e-01 -1.61716178e-01 -1.28011787e+00 1.90280616e+00
-1.98820516e-01 8.00180137e-02 2.51751959e-01 3.82607669e-01
1.40273690e+00 6.82453990e-01 1.66622028e-01 2.12612778e-01
8.22899759e-01 -9.44566429e-01 -4.70116049e-01 -3.70761991e-01
9.71362531e-01 -1.21942073e-01 9.66029525e-01 1.58088312e-01
-9.85447526e-01 -5.69253862e-01 -1.29949474e+00 -2.01803252e-01
-1.03962636e+00 -3.01429510e-01 1.11242127e+00 6.09706819e-01
-1.58544242e+00 9.69763756e-01 -6.64514482e-01 -4.73790646e-01
3.27764541e-01 6.11685574e-01 -2.18864173e-01 2.95701809e-02
-1.34835637e+00 7.23540127e-01 7.55354047e-01 5.63625954e-02
-9.91617262e-01 -6.09731197e-01 -1.55620098e+00 2.90860474e-01
5.87127447e-01 -7.42771089e-01 1.41575217e+00 -1.08437955e+00
-1.52212536e+00 8.18409681e-01 -1.58995435e-01 -9.13628876e-01
-1.41661108e-01 2.48782054e-01 -3.06614667e-01 -3.66358459e-01
-2.36042783e-01 6.15168393e-01 6.07793808e-01 -8.15621495e-01
-2.65431236e-02 -1.97990328e-01 2.20671281e-01 -3.57253373e-01
-1.78614244e-01 4.56265807e-02 9.48749706e-02 -4.57735240e-01
-3.54174763e-01 -9.09007967e-01 -5.71909249e-01 -3.82091433e-01
-6.94965184e-01 -4.96334285e-01 6.16782546e-01 -4.72916454e-01
1.27981544e+00 -1.65121794e+00 3.10099900e-01 5.11260033e-01
6.59192801e-01 2.59613782e-01 -3.18206131e-01 7.06273735e-01
-4.63270485e-01 5.39317012e-01 -3.14398289e-01 3.43152583e-01
2.17827782e-01 1.32818893e-01 -1.36223242e-01 3.22784126e-01
4.97440249e-01 1.66868198e+00 -9.70844269e-01 -1.69285327e-01
-1.26817256e-01 -4.20213118e-02 -5.84029019e-01 2.18702644e-01
-9.13061976e-01 -1.47625461e-01 -4.19106722e-01 7.04200149e-01
-1.65370461e-02 -8.91343772e-01 5.45139015e-01 1.10868607e-02
1.44690499e-01 5.12820661e-01 -7.66444921e-01 1.79266644e+00
-2.92017192e-01 7.72288918e-01 -1.23787992e-01 -1.24309182e+00
9.13778186e-01 1.92104161e-01 1.46685019e-01 -4.94423538e-01
2.51663979e-02 -8.49519297e-02 4.10286844e-01 -4.12125170e-01
5.29446721e-01 2.33872041e-01 -1.39523044e-01 6.95769846e-01
3.91710252e-01 -1.32499516e-01 3.72009575e-01 3.74885827e-01
1.31653559e+00 3.45632344e-01 5.93837023e-01 -3.48442376e-01
4.92408216e-01 2.41505764e-02 1.94331944e-01 1.07967997e+00
3.82071994e-02 3.95201035e-02 6.87709212e-01 -6.37968004e-01
-9.85204637e-01 -5.94673574e-01 6.12915277e-01 1.43264401e+00
-3.40737045e-01 -6.23463094e-01 -6.13660395e-01 -7.90861964e-01
8.66923556e-02 6.12782061e-01 -6.93976462e-01 -4.16427344e-01
-7.75897205e-01 -3.00411850e-01 6.92738593e-01 8.58586848e-01
-3.70952040e-02 -1.59719801e+00 -2.34093979e-01 4.31084067e-01
5.35386205e-01 -7.44821250e-01 -4.22796607e-01 6.71039701e-01
-8.07547092e-01 -1.21552193e+00 -2.79838443e-01 -1.24884975e+00
5.19073308e-01 -1.02553897e-01 1.85217464e+00 3.84200305e-01
-1.92279816e-01 3.96703839e-01 -5.08915670e-02 -6.15565121e-01
-8.11385393e-01 5.33480167e-01 -3.77428621e-01 -4.90599483e-01
2.91079491e-01 -4.98101920e-01 1.29741088e-01 -3.59809607e-01
-9.56505239e-01 -3.29970755e-03 3.22138816e-01 1.03042567e+00
5.15738189e-01 -4.63470995e-01 5.19542754e-01 -1.60428250e+00
8.07896078e-01 -4.97863024e-01 -8.84862661e-01 6.15482271e-01
-8.35948527e-01 5.63180864e-01 9.98546422e-01 -3.99975806e-01
-4.21256185e-01 1.60333544e-01 -2.55159494e-02 -3.34976107e-01
4.00487185e-02 1.11161542e+00 -1.22702509e-01 -4.58197743e-01
8.30444872e-01 1.46046832e-01 -4.97047082e-02 3.46226692e-02
6.03431463e-01 2.13144571e-01 2.63118863e-01 -8.10622096e-01
6.97945118e-01 -2.60877252e-01 2.37222254e-01 -6.84318542e-01
-4.92709041e-01 1.07291824e-04 -4.61828202e-01 8.57082829e-02
6.56536639e-01 -6.68141842e-01 -1.09281313e+00 1.59807876e-01
-1.23712623e+00 -9.15263712e-01 -2.58854806e-01 6.42921850e-02
-6.11781895e-01 3.15470904e-01 -8.82243872e-01 -5.80228806e-01
-6.46699309e-01 -1.15043485e+00 7.86998153e-01 5.21465279e-02
-2.82402486e-01 -1.23930240e+00 -9.71602742e-03 -5.57388961e-01
4.86713320e-01 5.46315372e-01 1.60340667e+00 -9.63205934e-01
-8.68693233e-01 1.72342196e-01 -2.32744262e-01 -1.30813569e-02
4.16798256e-02 8.75285640e-02 -8.19536269e-01 -4.20150638e-01
-6.52468979e-01 -6.43170476e-01 1.12587535e+00 4.11613494e-01
1.34630227e+00 -4.03033048e-01 -3.59003991e-01 7.01759279e-01
1.39410126e+00 2.74833649e-01 3.82074445e-01 1.47031639e-02
1.12152159e+00 3.55810672e-01 -1.90957308e-01 -2.30408162e-01
3.29516739e-01 4.74128500e-02 4.16741490e-01 -2.97269344e-01
-1.19681999e-01 -5.88652432e-01 4.28837657e-01 7.66988277e-01
2.11182266e-01 -6.73963845e-01 -1.03467572e+00 4.63354319e-01
-1.93959951e+00 -9.72835124e-01 7.84612149e-02 1.68389165e+00
6.97719157e-01 1.78997099e-01 1.76049158e-01 -1.15434140e-01
6.97533786e-01 3.91955167e-01 -8.43849897e-01 -1.00202787e+00
7.69643262e-02 7.00838804e-01 7.41225064e-01 5.27473807e-01
-1.06073368e+00 1.32589161e+00 6.99957895e+00 2.79134601e-01
-1.05797720e+00 -3.14499140e-01 4.94574606e-01 3.13584447e-01
-5.84764242e-01 -2.11361155e-01 -5.27483284e-01 -2.19631474e-02
1.49216855e+00 -3.70626450e-01 9.01494861e-01 9.72630501e-01
-4.20169830e-01 7.13753223e-01 -1.69362462e+00 6.35998547e-01
1.03553154e-01 -1.79398119e+00 1.69313192e-01 -1.13172263e-01
5.34757316e-01 4.99769688e-01 -1.66965470e-01 7.22675622e-01
1.17847407e+00 -1.56139290e+00 4.70952541e-01 2.88057357e-01
9.54424560e-01 -5.31660140e-01 3.61288458e-01 -1.60774827e-01
-1.36658728e+00 -3.30815054e-02 -2.76528120e-01 -8.14049169e-02
-2.71249563e-01 1.90159470e-01 -1.04278612e+00 6.75600648e-01
3.10873002e-01 1.22939765e+00 -6.20283782e-01 5.74290752e-01
-4.39456671e-01 6.64469898e-01 -2.06265803e-02 -7.52066851e-01
6.34147108e-01 -6.60844147e-02 1.95109382e-01 1.43879294e+00
9.76837333e-03 -2.98712432e-01 5.11834085e-01 1.27247429e+00
-6.13696337e-01 -1.37714654e-01 -1.59445262e+00 -6.53160632e-01
1.07218120e-02 9.44823682e-01 -5.94010711e-01 -5.51160395e-01
-6.95112526e-01 4.83937770e-01 7.55515516e-01 8.66287053e-01
-4.70279992e-01 -6.34157419e-01 3.96506071e-01 -2.13829532e-01
4.85965014e-01 -2.45948210e-01 -1.26494076e-02 -1.13290155e+00
-2.05462217e-01 -1.24095654e+00 7.43921220e-01 -5.86813927e-01
-1.17510915e+00 6.84830189e-01 -2.03399748e-01 -5.49440682e-01
-7.16112792e-01 -1.08874547e+00 -6.99861825e-01 9.24280763e-01
-1.53348684e+00 -1.20819592e+00 3.29780430e-02 5.84206522e-01
2.02848703e-01 -3.89632374e-01 9.50859964e-01 7.97774363e-03
-4.05905664e-01 6.32408202e-01 -1.81466773e-01 5.58345079e-01
8.04865509e-02 -1.52186573e+00 1.48731637e+00 9.47246313e-01
5.50784171e-01 9.23664153e-01 2.66640723e-01 -7.52746224e-01
-2.17414689e+00 -1.44864571e+00 7.99950957e-01 -2.16206953e-01
1.00394619e+00 -6.21877789e-01 -8.84431541e-01 1.41514194e+00
4.17360008e-01 7.23494142e-02 4.63595450e-01 4.83866215e-01
-5.81361711e-01 3.76940489e-01 -8.70134413e-01 6.62962556e-01
1.56479323e+00 -8.79961133e-01 -3.76446307e-01 4.05092984e-01
1.23279715e+00 -5.84472716e-01 -9.00311828e-01 4.74559307e-01
4.14815515e-01 -4.16930348e-01 7.95071006e-01 -1.68975651e+00
3.70024890e-01 -1.78099960e-01 -5.35315648e-02 -1.42874885e+00
-5.81675172e-01 -1.02065408e+00 -5.16498864e-01 6.93257451e-01
9.31005776e-01 -7.00422943e-01 1.00885844e+00 3.48706305e-01
-3.01749945e-01 -7.29134202e-01 -4.10904109e-01 -7.14951456e-01
1.48187369e-01 -2.57357687e-01 6.95120275e-01 9.69446659e-01
1.94969267e-01 6.57178581e-01 -7.32191429e-02 -1.80420522e-02
1.94191560e-01 6.94907427e-01 6.27692819e-01 -1.25081718e+00
-4.94466811e-01 -7.49986887e-01 -5.11334538e-01 -8.22324812e-01
8.99945438e-01 -1.71309507e+00 -1.22645468e-01 -1.64157021e+00
-6.01619333e-02 1.80289596e-01 -3.91458958e-01 8.82589817e-01
-4.10638973e-02 -3.81851166e-01 2.72680745e-02 -5.15924871e-01
-6.08744442e-01 2.65533715e-01 1.05829906e+00 -9.35909867e-01
-2.12582782e-01 -2.12445945e-01 -7.84395337e-01 2.13211477e-01
8.97491992e-01 -4.07337457e-01 -5.10826051e-01 -4.07498926e-01
5.99339485e-01 1.62454024e-02 1.99485689e-01 -5.98506987e-01
2.40674108e-01 -1.30295306e-01 2.66826630e-01 -2.87340730e-01
-2.11797625e-01 -4.83838469e-01 -2.28623264e-02 7.85448849e-01
-8.23195159e-01 4.73198473e-01 3.61513883e-01 6.58438206e-01
-1.62733406e-01 -3.11249912e-01 4.15443957e-01 -5.99556327e-01
-8.05146158e-01 7.23536253e-01 -5.29402614e-01 2.25995168e-01
4.97253418e-01 -9.28158686e-02 -4.30817276e-01 -4.99832958e-01
-6.64376795e-01 3.36475939e-01 3.19600403e-01 4.27224547e-01
6.76992536e-01 -1.12994373e+00 -5.20987988e-01 2.15569168e-01
2.77515054e-01 -1.35589987e-01 -4.84341472e-01 2.65881747e-01
-3.82761627e-01 5.74042857e-01 -4.88301888e-02 -2.92663366e-01
-1.04448652e+00 1.09182060e+00 4.28882241e-01 -5.24463654e-01
-5.96828997e-01 7.41696894e-01 4.47751135e-02 -7.99417615e-01
4.04199630e-01 -1.09180892e+00 -7.99913183e-02 -2.97751576e-01
1.65766940e-01 -1.35873333e-01 2.33222201e-01 -2.35295985e-02
-3.80995095e-01 2.25509554e-01 -6.77128136e-02 4.29029167e-01
1.39073062e+00 6.30347431e-01 -1.81104422e-01 4.52824175e-01
1.32944334e+00 -4.00391608e-01 -4.31099385e-01 -4.05692160e-01
5.38155317e-01 4.12135959e-01 -1.92337662e-01 -5.54631770e-01
-1.10776377e+00 9.65545475e-01 2.24559024e-01 4.53521371e-01
6.54288650e-01 -1.17698833e-01 6.91811681e-01 1.34088886e+00
2.79630542e-01 -6.16633832e-01 -1.48703516e-01 9.58695412e-01
5.73707581e-01 -1.14925289e+00 -4.08104286e-02 -1.22432679e-01
-1.94642350e-01 1.43556809e+00 6.10763669e-01 -1.95684344e-01
6.31354213e-01 4.01606441e-01 -4.89223689e-01 -6.45159602e-01
-9.68099833e-01 -4.49632734e-01 3.13539118e-01 7.79551923e-01
8.51851463e-01 8.91869813e-02 2.60384768e-01 1.94430172e-01
-1.35246679e-01 -2.43616654e-04 6.31350160e-01 1.09837508e+00
-2.58402169e-01 -1.09355891e+00 1.25572890e-01 8.29644501e-01
-3.77755374e-01 -6.45626962e-01 -8.26041639e-01 9.64641333e-01
-5.28242767e-01 7.54950583e-01 -2.42839947e-01 -4.12364870e-01
2.39358321e-01 4.15160239e-01 6.00587249e-01 -1.05589116e+00
-1.02400362e+00 -5.68798184e-01 5.25141180e-01 -7.18994379e-01
-3.47748876e-01 -1.90446660e-01 -1.32618952e+00 -2.40804762e-01
-1.55482918e-01 7.03512207e-02 3.40143204e-01 4.63497072e-01
4.42687690e-01 9.87394273e-01 2.47220188e-01 -6.57268465e-01
-2.89450437e-01 -9.02016759e-01 -2.11476177e-01 2.22304389e-01
5.69288015e-01 -2.41891637e-01 7.89328590e-02 -1.22235594e-02] | [7.018618583679199, 6.508023738861084] |
327fbac3-7f5f-46c9-b691-87f0b8f8c2e0 | swifttron-an-efficient-hardware-accelerator | 2304.03986 | null | https://arxiv.org/abs/2304.03986v2 | https://arxiv.org/pdf/2304.03986v2.pdf | SwiftTron: An Efficient Hardware Accelerator for Quantized Transformers | Transformers' compute-intensive operations pose enormous challenges for their deployment in resource-constrained EdgeAI / tinyML devices. As an established neural network compression technique, quantization reduces the hardware computational and memory resources. In particular, fixed-point quantization is desirable to ease the computations using lightweight blocks, like adders and multipliers, of the underlying hardware. However, deploying fully-quantized Transformers on existing general-purpose hardware, generic AI accelerators, or specialized architectures for Transformers with floating-point units might be infeasible and/or inefficient. Towards this, we propose SwiftTron, an efficient specialized hardware accelerator designed for Quantized Transformers. SwiftTron supports the execution of different types of Transformers' operations (like Attention, Softmax, GELU, and Layer Normalization) and accounts for diverse scaling factors to perform correct computations. We synthesize the complete SwiftTron architecture in a $65$ nm CMOS technology with the ASIC design flow. Our Accelerator executes the RoBERTa-base model in 1.83 ns, while consuming 33.64 mW power, and occupying an area of 273 mm^2. To ease the reproducibility, the RTL of our SwiftTron architecture is released at https://github.com/albertomarchisio/SwiftTron. | ['Muhammad Shafique', 'Guido Masera', 'Maurizio Martina', 'Maurizio Capra', 'Davide Dura', 'Alberto Marchisio'] | 2023-04-08 | null | null | null | null | ['neural-network-compression', 'neural-network-compression'] | ['methodology', 'miscellaneous'] | [ 6.41072839e-02 1.11062676e-01 -5.81701458e-01 -4.46627468e-01
-1.26396224e-01 -7.56454468e-02 1.70880049e-01 2.79848903e-01
-7.26713896e-01 2.16857374e-01 7.78799281e-02 -8.24026763e-01
2.86001503e-01 -9.46223319e-01 -6.41329706e-01 -5.11631310e-01
1.09014869e-01 1.76426545e-01 6.82289526e-02 -2.00356334e-01
3.74158546e-02 4.94041711e-01 -1.45227575e+00 3.21404427e-01
5.61923087e-01 1.64015710e+00 1.60943240e-01 6.80048227e-01
-1.09078120e-02 9.24509168e-01 -5.60961068e-01 -8.77428412e-01
4.58049804e-01 6.45696372e-02 -2.06739023e-01 -7.97826469e-01
5.17753243e-01 -6.79348171e-01 -9.59330559e-01 1.21563160e+00
5.68750381e-01 -2.55528539e-01 3.33118975e-01 -1.21872485e+00
-4.40547109e-01 1.23027182e+00 -3.84167224e-01 5.88803113e-01
-3.27132463e-01 2.73431033e-01 8.96116495e-01 -6.37314320e-01
1.41335800e-01 1.08350122e+00 7.33167827e-01 3.93691689e-01
-6.08642220e-01 -1.15084124e+00 -2.55463928e-01 2.93605566e-01
-1.59778368e+00 -9.38849747e-01 4.06551540e-01 3.74195457e-01
1.67007542e+00 1.85532913e-01 1.04054379e+00 9.78224814e-01
7.21086621e-01 6.07935727e-01 3.04801702e-01 -3.13070476e-01
7.13207126e-01 -3.71787161e-01 1.89659089e-01 7.41574526e-01
8.08445811e-01 -2.01210812e-01 -7.87378371e-01 2.13869754e-03
6.81863904e-01 1.64792165e-01 7.38420710e-02 3.09202284e-01
-8.64028811e-01 6.69215679e-01 7.77461171e-01 1.66733757e-01
-3.49945188e-01 9.73332345e-01 9.56576109e-01 5.87657243e-02
1.25116974e-01 1.20997056e-01 -4.83806819e-01 -5.86236894e-01
-1.06479561e+00 1.84900194e-01 5.46711206e-01 1.42682922e+00
2.60302842e-01 5.84091008e-01 6.57569841e-02 2.34645709e-01
2.47904688e-01 6.08858168e-01 8.24573874e-01 -7.87472665e-01
5.54838181e-01 7.01235890e-01 -7.24069476e-01 -5.87058127e-01
-6.51769698e-01 -5.97777247e-01 -1.42888927e+00 -6.35173544e-02
9.00777727e-02 3.41019742e-02 -9.54456747e-01 1.48299801e+00
5.90994991e-02 -2.55657099e-02 1.00689996e-02 8.12935591e-01
7.89890587e-01 6.93992496e-01 2.39903182e-01 2.76345909e-02
1.72831154e+00 -9.62405980e-01 -7.82799304e-01 -4.72135365e-01
7.71838725e-01 -4.69265789e-01 1.13642859e+00 3.48649263e-01
-1.42112279e+00 -4.29684550e-01 -1.64331639e+00 -7.89630949e-01
-3.73250991e-01 4.22678381e-01 1.13737667e+00 9.34695184e-01
-1.12367070e+00 5.77535689e-01 -1.44025862e+00 1.58974379e-01
7.34356523e-01 7.29292989e-01 2.97463715e-01 2.35806316e-01
-9.61178839e-01 5.87087750e-01 5.52267373e-01 7.81927928e-02
-3.38065833e-01 -1.09600878e+00 -7.38315046e-01 8.25379789e-01
-4.90327030e-02 -8.65320086e-01 1.48493803e+00 -3.93248171e-01
-1.54166842e+00 2.47309089e-01 3.53451297e-02 -1.13919532e+00
-8.37607980e-02 -1.62570644e-02 -4.48197693e-01 2.24722363e-02
-4.41950679e-01 7.52486229e-01 6.20197773e-01 2.35829532e-01
-6.41608477e-01 -3.13797921e-01 -1.40661731e-01 -1.02865212e-01
-9.43788528e-01 -1.87877849e-01 -3.25087041e-01 -6.44788682e-01
4.70664240e-02 -8.57919395e-01 -2.26446196e-01 2.49635085e-01
-8.70194733e-02 -2.61210166e-02 8.08357179e-01 -2.60678917e-01
1.50980759e+00 -2.25161290e+00 -3.52738321e-01 5.48264384e-02
4.54161555e-01 3.95496875e-01 4.72311497e-01 -1.93125159e-01
2.74702698e-01 -7.91503638e-02 3.35221104e-02 -3.21298331e-01
1.95985317e-01 1.06331743e-01 -3.81446838e-01 4.53675568e-01
-2.34521896e-01 1.10110486e+00 -3.93081158e-01 -2.64701277e-01
1.86014295e-01 7.37430692e-01 -8.79261494e-01 -6.07502818e-01
7.47600123e-02 -5.73080361e-01 -2.33653933e-01 1.00837672e+00
6.33212566e-01 -3.06009024e-01 2.85076231e-01 -7.68989563e-01
-1.80927590e-01 7.54528940e-01 -6.30413055e-01 1.98453283e+00
-4.19593126e-01 8.89957368e-01 1.85571790e-01 -6.83807254e-01
7.25019276e-01 1.48864627e-01 2.39393637e-01 -1.09980381e+00
8.63029778e-01 5.08055925e-01 1.22319549e-01 3.02946776e-01
9.13107634e-01 1.01481713e-01 -3.74569207e-01 3.57963234e-01
2.54855424e-01 -8.25480297e-02 3.03316087e-01 3.35895419e-01
1.32544792e+00 -4.35540915e-01 2.61473835e-01 -4.93203521e-01
-2.20137388e-02 -6.03107102e-02 5.70425034e-01 3.62156779e-01
-3.06585312e-01 -6.47495762e-02 1.74250692e-01 -6.19313538e-01
-1.37967181e+00 -9.14760530e-01 -3.10343266e-01 1.09968221e+00
-1.80533603e-01 -1.20627058e+00 -7.85474479e-01 1.53319463e-01
-2.86919683e-01 6.62088513e-01 1.42562851e-01 -4.88796026e-01
-6.64778411e-01 -9.49470162e-01 1.13523304e+00 8.83693397e-01
8.32407355e-01 -5.53722739e-01 -1.62899375e+00 2.10358486e-01
4.18788016e-01 -1.21455848e+00 -5.71531475e-01 7.62429714e-01
-1.21525013e+00 -2.94359416e-01 -5.69209009e-02 -5.82723022e-01
4.31041777e-01 -9.31205694e-03 1.03261507e+00 1.27763361e-01
-4.54309285e-01 -4.54582840e-01 -5.10951020e-02 -5.43597162e-01
-1.94627821e-01 3.14728200e-01 4.55353320e-01 -8.39319587e-01
5.18541992e-01 -9.67674673e-01 -8.15826893e-01 -2.69714773e-01
-6.86219215e-01 2.51454026e-01 8.65749300e-01 7.78991759e-01
5.92991650e-01 2.62491763e-01 1.48395136e-01 -4.31048363e-01
5.75100124e-01 5.10530584e-02 -8.33896101e-01 -1.45907462e-01
-8.09928834e-01 3.12739491e-01 1.03154421e+00 -5.11147082e-01
-5.73362648e-01 2.00858369e-01 -5.83405733e-01 -6.25018716e-01
6.69628859e-01 3.62765819e-01 -4.20430638e-02 -3.22172970e-01
8.88478637e-01 5.89255467e-02 -1.66958436e-01 -3.56992800e-03
3.07948887e-01 7.09922433e-01 9.05749798e-01 -3.59223217e-01
4.19780731e-01 2.10420474e-01 1.47087768e-01 -5.46911418e-01
-2.49044791e-01 1.56701848e-01 -1.25118092e-01 7.69944116e-02
4.33644474e-01 -1.31618834e+00 -1.27206886e+00 2.15736497e-02
-9.12891150e-01 -2.55569458e-01 -4.80848521e-01 5.07592380e-01
-1.86386466e-01 -1.95372850e-01 -1.14918029e+00 -5.14507532e-01
-1.44071150e+00 -1.41235554e+00 9.02449191e-01 6.30908132e-01
-2.81628251e-01 -5.19250333e-01 -6.80765450e-01 -3.84751968e-02
9.37139034e-01 -3.35393608e-01 8.24297011e-01 -2.19118759e-01
-6.90379202e-01 -3.09817880e-01 -1.92735121e-01 -3.92790185e-03
-3.01444352e-01 -7.88844451e-02 -1.01427186e+00 -4.00753826e-01
2.01266378e-01 -3.06911290e-01 6.48025095e-01 4.87872303e-01
1.46836305e+00 -3.76079053e-01 -3.23862016e-01 1.13709533e+00
1.11443102e+00 2.64742464e-01 5.39168179e-01 2.18824167e-02
4.37356770e-01 -6.40274465e-01 2.35288888e-01 7.50636876e-01
2.44410753e-01 4.81886923e-01 4.47207361e-01 2.45898992e-01
-7.85487294e-02 -2.66041458e-01 3.45023245e-01 1.42739904e+00
-1.79989189e-02 -1.53002724e-01 -7.87469447e-01 -8.05415958e-02
-1.34570432e+00 -8.34738731e-01 2.67920285e-01 1.98578465e+00
8.11572552e-01 6.91814780e-01 -2.17214242e-01 5.20974636e-01
9.80481431e-02 1.43018104e-02 -8.48001003e-01 -9.79504347e-01
1.41406819e-01 8.80187035e-01 1.21937740e+00 1.45974144e-01
-6.91661358e-01 8.26705873e-01 5.60717010e+00 1.20356882e+00
-1.41001081e+00 3.45494270e-01 7.99136937e-01 -5.54577529e-01
1.00865047e-02 -1.77122846e-01 -1.21511066e+00 4.54097331e-01
1.77717388e+00 -5.58521748e-01 3.95086825e-01 1.19166660e+00
-1.48135558e-01 5.44030331e-02 -1.10897183e+00 1.26527584e+00
-2.00744435e-01 -1.74392009e+00 -9.75730121e-02 1.57261595e-01
7.83388019e-02 2.07464829e-01 3.22797120e-01 5.38647652e-01
-2.52113715e-02 -8.60752404e-01 9.83124673e-01 -2.62905866e-01
1.19647014e+00 -1.18125820e+00 5.42488158e-01 2.40977943e-01
-1.46315169e+00 -3.34701240e-01 -8.50442886e-01 -3.29654932e-01
7.20125958e-02 8.00499499e-01 -7.78747559e-01 -1.51403964e-01
9.87703741e-01 2.41546586e-01 -4.72949028e-01 4.47450578e-01
1.64720267e-01 5.47090530e-01 -8.13954353e-01 -2.70241827e-01
4.96617518e-02 -9.22522694e-02 -9.60828643e-03 1.15205371e+00
6.23030365e-01 2.58189440e-01 -4.80260313e-01 5.55703521e-01
-4.92262959e-01 -2.82131582e-01 -1.71545133e-01 -1.37272984e-01
9.01225328e-01 1.44596255e+00 -9.26782787e-01 -5.59688747e-01
-1.90914094e-01 8.20649743e-01 1.50200218e-01 -4.38131958e-01
-1.21212864e+00 -6.33170307e-01 7.98938453e-01 -4.50974256e-02
-7.55172372e-02 -3.20618093e-01 -8.56713831e-01 -1.08818901e+00
6.74077645e-02 -8.16505253e-01 1.80432677e-01 -5.24515986e-01
-3.49403739e-01 8.07169080e-01 -3.39179367e-01 -8.51852238e-01
-5.42380698e-02 -7.29307055e-01 -4.06205326e-01 4.93388742e-01
-1.06719339e+00 -8.26491117e-01 -4.77418512e-01 4.94631797e-01
5.71986675e-01 -2.34095231e-01 8.36107969e-01 9.13422465e-01
-5.99670291e-01 1.31936538e+00 -1.74517184e-01 -1.74819648e-01
1.18389070e-01 -5.43926597e-01 1.10105968e+00 8.59084010e-01
-5.24888821e-02 8.13236654e-01 7.95225263e-01 -4.96818870e-01
-2.30169368e+00 -1.01186371e+00 5.09857118e-01 3.79626215e-01
7.45511889e-01 -7.10132480e-01 -6.15658700e-01 8.08539927e-01
4.28780951e-02 3.69620830e-01 6.29362702e-01 -3.06974977e-01
-1.80400312e-01 -3.37177634e-01 -1.19542420e+00 9.84323025e-01
1.22491252e+00 -4.19236809e-01 2.15327457e-01 5.33249736e-01
9.35671210e-01 -9.16945577e-01 -9.59520757e-01 2.50029355e-01
6.83622003e-01 -6.57747746e-01 1.03819692e+00 1.29475757e-01
3.41326088e-01 1.66570414e-02 -4.18500215e-01 -7.74173975e-01
-3.36011440e-01 -5.41707516e-01 -5.95680237e-01 6.36755526e-01
3.58763129e-01 -5.58281422e-01 1.18538332e+00 7.03073978e-01
-7.10742474e-01 -9.04661059e-01 -1.35989094e+00 -5.44318378e-01
-9.96420309e-02 -6.07342362e-01 8.74173939e-01 3.49899560e-01
5.88640392e-01 3.91947329e-01 -1.32603556e-01 -6.76588342e-02
5.12714326e-01 -3.08778435e-01 2.74955481e-01 -8.45012486e-01
-3.26531738e-01 -6.72527075e-01 -7.27843404e-01 -1.18472385e+00
-9.82198492e-02 -1.05797827e+00 -3.75926822e-01 -8.07550013e-01
-2.09673539e-01 -6.02989256e-01 8.18389431e-02 6.10242009e-01
4.72108096e-01 3.46301585e-01 2.00738057e-01 -8.92930180e-02
-4.20609921e-01 7.01995194e-01 7.61174321e-01 -2.52146900e-01
-5.64054810e-02 -3.09634835e-01 -6.95930243e-01 6.89866424e-01
1.02780080e+00 -2.50886351e-01 -6.10731304e-01 -6.73139989e-01
3.46022397e-01 2.93425238e-03 -4.17722389e-02 -1.63317299e+00
9.53843534e-01 3.72439831e-01 4.79647875e-01 -5.90781689e-01
7.62685597e-01 -6.93906844e-01 3.28835249e-01 1.03391719e+00
-4.27176878e-02 7.25645423e-01 4.82516736e-01 -8.82970989e-02
1.12853684e-01 -1.00780174e-01 7.22853601e-01 1.37692183e-01
-5.50317466e-01 4.03451920e-01 -3.87388259e-01 -2.36520827e-01
7.17540443e-01 -5.77242710e-02 -7.58351505e-01 1.98013391e-02
-1.80488124e-01 -1.26607448e-01 3.56902063e-01 -6.31362349e-02
8.32946181e-01 -1.19762361e+00 -1.47815064e-01 4.79829997e-01
-3.13499421e-01 1.69793233e-01 2.36461207e-01 7.09877670e-01
-1.13018417e+00 8.53941083e-01 -5.09326756e-01 -3.66176188e-01
-1.12991798e+00 4.27157193e-01 1.20472059e-01 2.02425700e-02
-1.01969302e+00 9.48114216e-01 -4.11936462e-01 2.33125135e-01
2.28329420e-01 -9.49802160e-01 4.28996086e-01 -3.00974607e-01
7.77181387e-01 2.47971028e-01 5.69954157e-01 -1.77184299e-01
-4.43319619e-01 5.98613769e-02 -1.61615327e-01 1.17069177e-01
1.02757466e+00 2.88381755e-01 4.07282561e-02 -8.99258777e-02
1.05311954e+00 -3.82468849e-01 -7.46247530e-01 -3.64627433e-03
-2.56315857e-01 -3.76876332e-02 6.97207868e-01 -1.90761805e-01
-1.50973475e+00 7.62733757e-01 7.40238845e-01 -3.15477639e-01
1.57035291e+00 -3.96770179e-01 1.14237988e+00 6.30582154e-01
6.83194399e-01 -1.03473103e+00 -4.73464668e-01 4.66664433e-01
2.00146675e-01 -6.36976600e-01 5.46408534e-01 -3.26770395e-01
-7.59023577e-02 1.25284910e+00 6.75747633e-01 3.61757278e-02
6.78171158e-01 1.36089087e+00 -5.27980030e-01 -1.59613043e-02
-1.13240528e+00 4.76545304e-01 -2.96769738e-01 9.49559882e-02
5.25151193e-01 2.63879538e-01 -4.73920971e-01 6.59370899e-01
-9.30603921e-01 2.78092504e-01 5.99931240e-01 1.19581091e+00
-3.91889483e-01 -7.23354042e-01 1.09266518e-02 8.40844631e-01
-4.87518787e-01 -4.96778458e-01 4.36506301e-01 3.38737458e-01
-2.90221255e-02 4.64540452e-01 5.95197022e-01 -7.85048842e-01
1.52948359e-02 -2.17059806e-01 3.28446835e-01 -7.28013441e-02
-8.39474738e-01 -5.65781668e-02 -9.62675512e-02 -5.43599069e-01
3.07012916e-01 -2.02358365e-01 -1.59803665e+00 -1.19161034e+00
-6.43046498e-02 -2.34079137e-01 1.19631219e+00 4.48354661e-01
5.97056329e-01 8.87168705e-01 -4.93902005e-02 -1.00022233e+00
-5.82632065e-01 -7.24673033e-01 -3.16905141e-01 -5.24914682e-01
-3.30079198e-02 -4.42771167e-01 -1.27632186e-01 -2.81281292e-01] | [8.432191848754883, 2.8759424686431885] |
d1003ea7-35d3-420b-9f03-cc58c665d4a1 | sep-stereo-visually-guided-stereophonic-audio | 2007.09902 | null | https://arxiv.org/abs/2007.09902v1 | https://arxiv.org/pdf/2007.09902v1.pdf | Sep-Stereo: Visually Guided Stereophonic Audio Generation by Associating Source Separation | Stereophonic audio is an indispensable ingredient to enhance human auditory experience. Recent research has explored the usage of visual information as guidance to generate binaural or ambisonic audio from mono ones with stereo supervision. However, this fully supervised paradigm suffers from an inherent drawback: the recording of stereophonic audio usually requires delicate devices that are expensive for wide accessibility. To overcome this challenge, we propose to leverage the vastly available mono data to facilitate the generation of stereophonic audio. Our key observation is that the task of visually indicated audio separation also maps independent audios to their corresponding visual positions, which shares a similar objective with stereophonic audio generation. We integrate both stereo generation and source separation into a unified framework, Sep-Stereo, by considering source separation as a particular type of audio spatialization. Specifically, a novel associative pyramid network architecture is carefully designed for audio-visual feature fusion. Extensive experiments demonstrate that our framework can improve the stereophonic audio generation results while performing accurate sound separation with a shared backbone. | ['Xudong Xu', 'Hang Zhou', 'Ziwei Liu', 'Xiaogang Wang', 'Dahua Lin'] | 2020-07-20 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/1483_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123570052.pdf | eccv-2020-8 | ['audio-generation'] | ['audio'] | [ 3.32953453e-01 -2.92933792e-01 1.75018609e-01 -1.69173896e-01
-1.06453454e+00 -5.63340008e-01 3.56248826e-01 3.53716165e-02
8.34314059e-03 6.04785860e-01 5.62703252e-01 1.37583420e-01
-1.67726696e-01 -4.71045643e-01 -5.61090767e-01 -9.47638273e-01
4.18555588e-01 -1.32435840e-02 2.29610041e-01 -2.82151222e-01
2.78825432e-01 3.04567367e-01 -2.41104221e+00 5.89730263e-01
8.87507677e-01 1.20016563e+00 4.84115154e-01 8.61381114e-01
1.04838528e-01 4.83771354e-01 -6.52656436e-01 -8.52136686e-02
2.19765738e-01 -3.95125717e-01 -5.23196578e-01 3.58146243e-02
6.07186258e-01 -1.26064047e-01 2.59151496e-02 1.05253327e+00
1.10713065e+00 9.25849229e-02 5.63663363e-01 -1.49726665e+00
-5.36239743e-01 8.34504247e-01 -4.57172573e-01 5.09040849e-03
6.15693152e-01 5.70745915e-02 1.16861200e+00 -8.80991101e-01
3.93426754e-02 1.10642540e+00 4.76737857e-01 2.69593894e-01
-1.23273933e+00 -8.64656746e-01 -2.26889085e-02 3.57321799e-01
-1.47903585e+00 -8.29904854e-01 9.34197783e-01 -5.57894111e-01
4.20437962e-01 4.25090849e-01 8.99333656e-01 8.65287483e-01
-1.57322437e-01 8.37859273e-01 1.00807893e+00 -3.55445713e-01
2.45794043e-01 2.24018544e-01 -2.10568190e-01 1.36680216e-01
-1.22214302e-01 1.99406773e-01 -1.09120357e+00 1.19088262e-01
5.83912432e-01 -1.79323971e-01 -6.94651544e-01 -3.64339024e-01
-9.80433762e-01 5.22785544e-01 3.45433593e-01 2.56242096e-01
-1.61682203e-01 -1.95655730e-02 2.38610595e-01 1.84039429e-01
1.87034056e-01 5.48783898e-01 1.44603783e-02 -1.46294847e-01
-9.83307779e-01 1.21006839e-01 4.55017149e-01 6.61539316e-01
6.11051440e-01 5.38966537e-01 -1.57602921e-01 9.38262284e-01
2.56210357e-01 6.39913499e-01 3.83874744e-01 -9.15477037e-01
3.24704170e-01 2.05742761e-01 1.04380921e-01 -9.30505455e-01
-2.10822836e-01 -6.48323298e-01 -8.23812544e-01 2.82315075e-01
1.54980838e-01 -1.03527412e-01 -7.13157117e-01 1.71898472e+00
3.05651903e-01 2.80724078e-01 -1.25534713e-01 1.28449035e+00
8.30632865e-01 5.67041874e-01 -1.26333490e-01 -1.01879157e-01
1.36276245e+00 -8.84969413e-01 -7.42541611e-01 1.04415014e-01
-8.41132775e-02 -1.05895531e+00 1.17124391e+00 6.31097376e-01
-1.19819999e+00 -8.48353505e-01 -1.12949705e+00 4.09037881e-02
-1.97407454e-01 1.67914718e-01 5.04831195e-01 7.32929230e-01
-1.18759036e+00 1.99882179e-01 -4.54536796e-01 5.95322922e-02
3.50186303e-02 2.23820984e-01 -2.85036623e-01 4.01421547e-01
-1.21125007e+00 3.10433447e-01 2.61651009e-01 9.07634869e-02
-1.10115576e+00 -9.56016302e-01 -9.20583427e-01 3.19602013e-01
2.36226082e-01 -7.64658928e-01 1.30244505e+00 -8.39673281e-01
-1.65239918e+00 4.94561344e-01 -1.70675777e-02 -5.18994093e-01
7.78390989e-02 -3.81779879e-01 -4.66129810e-01 4.75845993e-01
1.24869496e-01 7.55740821e-01 1.34895837e+00 -1.45472527e+00
-9.63874102e-01 -2.87451863e-01 1.44825764e-02 6.40970588e-01
-7.28136003e-01 -1.05158843e-01 -4.42480147e-01 -7.37541080e-01
8.31298307e-02 -6.33560598e-01 1.69773608e-01 -3.34274232e-01
-5.93457520e-01 2.05854863e-01 6.26776755e-01 -4.68213499e-01
1.22638297e+00 -2.19699621e+00 2.82208353e-01 3.40028703e-02
3.53877395e-01 1.23109706e-01 2.43799295e-02 3.34433377e-01
-2.00215787e-01 -3.57548594e-01 4.95513417e-02 -4.40370023e-01
1.00911513e-01 -5.29770136e-01 -7.34541059e-01 5.61904442e-03
-6.14612363e-02 4.47734356e-01 -9.11343992e-01 -6.36453331e-01
2.73095578e-01 7.44861901e-01 -8.84073675e-01 3.17969531e-01
2.61508524e-01 7.76767671e-01 -1.16488546e-01 7.61456907e-01
6.35556757e-01 5.66371158e-02 -5.86928315e-02 -4.02354062e-01
-3.31029683e-01 1.36650234e-01 -1.41138482e+00 1.83624899e+00
-5.44122040e-01 5.16445279e-01 4.77400720e-01 -6.14901841e-01
8.98290217e-01 5.85913897e-01 3.95303547e-01 -4.96211231e-01
-2.76443129e-03 2.93618858e-01 -1.58927664e-01 -2.76066542e-01
6.98207974e-01 -3.10038120e-01 -1.13589745e-02 3.70611727e-01
2.04263300e-01 -5.39805055e-01 -7.57822990e-02 3.08031626e-02
6.26544237e-01 -4.51604538e-02 8.98395479e-02 3.74380574e-02
5.61127007e-01 -5.22531092e-01 4.58129287e-01 6.40600204e-01
-1.66381463e-01 1.03023064e+00 2.49881912e-02 1.68518946e-01
-7.09580183e-01 -1.55735219e+00 5.38556240e-02 1.38985765e+00
2.07970321e-01 -5.74627042e-01 -8.57835054e-01 -2.35835388e-01
-2.31106430e-01 3.83877456e-01 -1.10223584e-01 -1.61096945e-01
-2.97437817e-01 -4.27841693e-01 6.59421504e-01 5.31428218e-01
3.20235103e-01 -9.25189853e-01 -4.01690662e-01 8.19153637e-02
-5.71641684e-01 -8.23075056e-01 -5.94896674e-01 1.75070137e-01
-2.84463376e-01 -7.21805692e-01 -1.04305983e+00 -8.16158175e-01
1.55941665e-01 6.76076591e-01 8.65283966e-01 -4.98209506e-01
-3.04194957e-01 7.29449570e-01 -2.02973947e-01 -5.06352186e-01
-1.93300471e-01 -6.22886382e-02 3.47481668e-01 5.47875285e-01
3.77414078e-02 -1.27925253e+00 -9.59904313e-01 3.50010037e-01
-9.40106452e-01 2.03412518e-01 5.75493872e-01 6.57103896e-01
6.41953886e-01 2.99664319e-01 7.88389802e-01 -1.52052194e-01
7.56540060e-01 -2.54467696e-01 -4.04604405e-01 7.86566641e-03
-2.84107745e-01 -2.99325973e-01 7.95118332e-01 -2.52088517e-01
-1.33943379e+00 2.53168613e-01 -5.34526370e-02 -6.70236468e-01
-4.32925195e-01 2.58916348e-01 -4.09463882e-01 2.19550636e-02
6.05271280e-01 4.79547441e-01 -2.42914289e-01 -7.30295897e-01
4.96717513e-01 1.02121377e+00 8.21421683e-01 -3.56975406e-01
7.03845978e-01 7.11825013e-01 -2.07795516e-01 -1.06242549e+00
-6.45207345e-01 -7.44936883e-01 -4.13979948e-01 -4.27219242e-01
8.30896497e-01 -1.20338559e+00 -8.62248182e-01 5.67966640e-01
-1.05400097e+00 7.78201148e-02 -2.62112021e-01 6.44798696e-01
-6.94894731e-01 4.46098000e-01 -3.67907822e-01 -9.91730750e-01
-1.73212454e-01 -1.24271727e+00 1.34596801e+00 2.18876764e-01
-1.75479159e-01 -4.74127233e-01 2.55165428e-01 6.26719832e-01
2.17881784e-01 -3.08760434e-01 6.50973260e-01 -3.59782428e-01
-5.90051472e-01 -9.71398409e-03 -1.48188367e-01 3.87861222e-01
3.00443858e-01 -3.63813907e-01 -1.61380708e+00 -1.45471767e-01
1.76223174e-01 -4.83488292e-01 8.43871653e-01 5.07802844e-01
1.12519658e+00 8.76081735e-02 7.06965476e-02 5.05780876e-01
9.46625173e-01 3.23627770e-01 3.93658638e-01 -1.88203737e-01
7.96750128e-01 8.07613075e-01 6.66254580e-01 6.03137612e-01
2.43107349e-01 8.93006682e-01 5.06886184e-01 -2.66741931e-01
-4.36802715e-01 -5.18879056e-01 3.40281069e-01 1.06929660e+00
-7.48452246e-02 -1.46057919e-01 -6.19146287e-01 6.76745534e-01
-1.63514853e+00 -9.71317887e-01 2.47148752e-01 2.33172083e+00
1.03486502e+00 -1.20709524e-01 2.41280630e-01 7.25023985e-01
1.09012544e+00 1.46508217e-01 -1.55839100e-01 -8.26726705e-02
-2.01751098e-01 2.70719528e-01 -2.23293714e-02 4.28495675e-01
-1.13424706e+00 7.12112844e-01 6.14832830e+00 1.27767003e+00
-1.40436995e+00 -2.64190435e-02 9.61557552e-02 -3.95403922e-01
-3.76501977e-01 -2.33223230e-01 -6.84202611e-01 4.30682242e-01
7.36981630e-01 -1.87911186e-02 4.83713835e-01 6.80406451e-01
3.86919022e-01 -2.26806495e-02 -1.08569169e+00 1.42295027e+00
1.63118541e-01 -1.10725510e+00 3.39014471e-01 -1.35556370e-01
5.49310088e-01 -4.99166638e-01 5.63951731e-01 -6.31848052e-02
1.18896142e-01 -8.92785490e-01 9.32438791e-01 4.18056428e-01
8.41246665e-01 -7.93648064e-01 1.22231014e-01 -8.53009969e-02
-1.59304225e+00 -8.42598602e-02 -3.55446972e-02 -1.12580948e-01
3.41019928e-01 5.07612646e-01 -9.06139672e-01 6.20394945e-01
1.00274897e+00 5.84701657e-01 -3.79577428e-01 1.41391563e+00
-2.58559644e-01 4.23176229e-01 -8.84806067e-02 5.33939004e-01
-9.54498500e-02 7.44057521e-02 9.29982126e-01 9.87220466e-01
5.15786290e-01 -5.02019465e-01 7.04620406e-02 6.26617312e-01
3.44133168e-01 2.98015565e-01 -7.40488410e-01 -2.80420277e-02
6.05590463e-01 1.25393999e+00 -6.80557251e-01 -1.06775880e-01
-1.35524437e-01 7.99875379e-01 -1.31558076e-01 3.15210819e-01
-8.97872269e-01 -6.74369097e-01 6.39833033e-01 -1.84622686e-02
4.69513118e-01 1.98737964e-01 -1.34271204e-01 -9.38534975e-01
6.72532842e-02 -9.48967636e-01 1.90030962e-01 -1.10250211e+00
-1.11072469e+00 8.51700187e-01 -2.28468493e-01 -1.83759665e+00
-2.84599721e-01 -2.80553401e-01 -4.12870765e-01 6.65518999e-01
-1.62638104e+00 -1.00967360e+00 -3.25238675e-01 1.00497031e+00
5.33774018e-01 -1.88770860e-01 8.01783204e-01 6.51874125e-01
-3.19875211e-01 7.30571747e-01 -2.84064189e-02 -2.56461412e-01
1.11967421e+00 -1.29900336e+00 -1.74187839e-01 6.35257959e-01
3.52859557e-01 5.10237277e-01 8.18182647e-01 -4.00108993e-01
-1.03140366e+00 -9.90089417e-01 7.29302824e-01 -1.11926079e-01
6.26731217e-01 -3.95856023e-01 -5.42240679e-01 1.89046472e-01
2.55535811e-01 -5.02621233e-01 1.10042822e+00 -1.53085245e-02
-3.68210793e-01 -4.33812201e-01 -5.03522813e-01 8.06569397e-01
8.60491455e-01 -8.94472539e-01 -6.60726130e-01 -3.61351818e-02
8.35323334e-01 -1.12664990e-01 -6.25336587e-01 1.50053367e-01
6.47776663e-01 -1.46592212e+00 1.21090758e+00 -1.78506523e-01
4.20690894e-01 -7.66015708e-01 -3.25251400e-01 -1.48448098e+00
-6.36228770e-02 -9.27516699e-01 1.28669083e-01 1.50765169e+00
1.95761204e-01 -4.38185036e-01 4.35829043e-01 -9.26082134e-02
-4.22888607e-01 -2.27705970e-01 -7.87604749e-01 -7.86611557e-01
-5.26708484e-01 -7.32514739e-01 7.05741167e-01 6.76628709e-01
1.99122727e-01 4.76077676e-01 -8.02875936e-01 3.14286888e-01
6.12644434e-01 4.45806205e-01 7.30139673e-01 -1.32086957e+00
-5.06664991e-01 -3.10790986e-01 -4.45121944e-01 -1.13873100e+00
-8.98158178e-03 -8.03017616e-01 3.44626874e-01 -1.22479320e+00
-7.30539635e-02 -1.57844514e-01 -6.20855451e-01 1.46867752e-01
1.26013368e-01 8.23754370e-01 2.84554064e-01 2.56528676e-01
-6.64326549e-01 9.33973968e-01 1.24519873e+00 -7.18509629e-02
-4.63818610e-01 2.65203655e-01 -1.02743030e+00 1.05056000e+00
5.11681676e-01 -1.90463990e-01 -7.18298912e-01 -4.30258542e-01
2.39962801e-01 4.43821102e-01 5.67702830e-01 -1.34472680e+00
4.81468469e-01 1.21023960e-01 4.31879722e-02 -7.72026181e-01
7.99409747e-01 -7.80683398e-01 3.12838219e-02 -5.31393215e-02
-4.72180516e-01 -3.63560230e-01 1.35947689e-01 8.90742958e-01
-8.05081606e-01 1.43422544e-01 6.00140691e-01 1.41660720e-01
-4.68746692e-01 5.44370823e-02 -5.19894481e-01 -4.66700606e-02
6.49992883e-01 -3.59185815e-01 -7.39323348e-02 -7.73646355e-01
-8.39126885e-01 -8.70067924e-02 6.70135915e-02 4.43381101e-01
7.07343161e-01 -1.53239906e+00 -3.83192241e-01 2.39328787e-01
1.16977692e-01 -2.58309543e-01 7.57116377e-01 9.76397753e-01
-7.32742548e-02 5.41589320e-01 -4.68266964e-01 -8.29055548e-01
-1.34993947e+00 4.67335254e-01 3.68937254e-02 1.54571444e-01
-4.01414573e-01 8.58208299e-01 9.01588738e-01 3.37114371e-02
5.90746522e-01 -4.97569563e-04 -6.07671976e-01 4.76848096e-01
6.46610141e-01 5.29382050e-01 5.55107147e-02 -5.95895410e-01
-1.55841440e-01 6.13775134e-01 1.06257290e-01 -5.50376117e-01
1.15228355e+00 -6.32188082e-01 4.85336855e-02 6.41921639e-01
9.60255682e-01 5.03963649e-01 -1.30535865e+00 -1.51495755e-01
-4.61049527e-01 -4.20257509e-01 1.67731553e-01 -6.20987058e-01
-1.14125705e+00 1.26099896e+00 5.30365884e-01 3.48808885e-01
1.55321097e+00 -2.42360741e-01 8.25811267e-01 1.66305318e-01
1.56887427e-01 -1.14053583e+00 3.79778355e-01 2.56417215e-01
9.33726549e-01 -9.66562390e-01 -4.83553946e-01 -5.91816485e-01
-7.40783632e-01 9.17798340e-01 5.48225164e-01 3.00243825e-01
5.91251612e-01 2.17049763e-01 2.55762726e-01 -1.35849103e-01
-4.10918534e-01 -5.04008174e-01 5.91767132e-01 8.04672003e-01
2.98383504e-01 -1.23915344e-01 2.76784211e-01 9.12331283e-01
-5.64583182e-01 -2.98876435e-01 2.92513520e-01 7.14970350e-01
-4.81101841e-01 -8.37511122e-01 -6.70152545e-01 -1.48047283e-01
-4.09092724e-01 -4.91745174e-01 -4.62724686e-01 2.05246672e-01
9.86484289e-02 1.20328975e+00 6.17286414e-02 -6.17214799e-01
8.57302248e-02 2.67505199e-01 2.82665879e-01 -5.95473886e-01
-5.48262596e-01 7.22956777e-01 -2.75379509e-01 -4.70472157e-01
-5.04649043e-01 -3.69230062e-01 -9.03851569e-01 1.12143554e-01
-2.46320695e-01 2.81454325e-01 4.69088763e-01 6.58430517e-01
5.04424155e-01 7.75539935e-01 8.32455575e-01 -1.20118904e+00
-3.20597142e-01 -6.06126726e-01 -9.27137017e-01 3.30348790e-01
5.10062218e-01 -7.58124411e-01 -6.41295671e-01 2.77487069e-01] | [14.95418643951416, 5.077826499938965] |
fcaf1250-b07b-4b0a-bd2c-d821b66de82c | high-fidelity-and-freely-controllable-talking | 2304.10168 | null | https://arxiv.org/abs/2304.10168v1 | https://arxiv.org/pdf/2304.10168v1.pdf | High-Fidelity and Freely Controllable Talking Head Video Generation | Talking head generation is to generate video based on a given source identity and target motion. However, current methods face several challenges that limit the quality and controllability of the generated videos. First, the generated face often has unexpected deformation and severe distortions. Second, the driving image does not explicitly disentangle movement-relevant information, such as poses and expressions, which restricts the manipulation of different attributes during generation. Third, the generated videos tend to have flickering artifacts due to the inconsistency of the extracted landmarks between adjacent frames. In this paper, we propose a novel model that produces high-fidelity talking head videos with free control over head pose and expression. Our method leverages both self-supervised learned landmarks and 3D face model-based landmarks to model the motion. We also introduce a novel motion-aware multi-scale feature alignment module to effectively transfer the motion without face distortion. Furthermore, we enhance the smoothness of the synthesized talking head videos with a feature context adaptation and propagation module. We evaluate our model on challenging datasets and demonstrate its state-of-the-art performance. More information is available at https://yuegao.me/PECHead. | ['Yan Lu', 'Xiang Ming', 'Xiao Li', 'Jinglu Wang', 'Yuan Zhou', 'Yue Gao'] | 2023-04-20 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Gao_High-Fidelity_and_Freely_Controllable_Talking_Head_Video_Generation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Gao_High-Fidelity_and_Freely_Controllable_Talking_Head_Video_Generation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['talking-head-generation', 'video-generation', 'face-model'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.09920710e-01 9.48593616e-02 -6.57158643e-02 -5.96523881e-01
-9.27097738e-01 -5.48096955e-01 5.96505225e-01 -7.93152750e-01
1.89931050e-01 5.04286528e-01 7.13927269e-01 5.88948548e-01
2.45501697e-01 -2.49959603e-01 -7.21072912e-01 -7.91142464e-01
1.36007398e-01 -5.99715076e-02 -1.86731443e-01 -1.20332345e-01
1.02462374e-01 4.18467641e-01 -1.72922754e+00 1.86586142e-01
6.76241159e-01 9.66093898e-01 8.65489319e-02 6.55477405e-01
4.48961288e-01 6.66040242e-01 -5.32953620e-01 -2.29237780e-01
2.52015203e-01 -5.11457086e-01 -3.91316533e-01 4.63683963e-01
9.39741552e-01 -6.48509026e-01 -6.40746057e-01 1.03414857e+00
7.92329252e-01 7.84877315e-02 4.06572223e-01 -1.47120881e+00
-4.77128506e-01 4.69229482e-02 -4.87896860e-01 -1.65085420e-01
8.29148889e-01 5.72935224e-01 4.64025855e-01 -1.06343293e+00
1.03455019e+00 1.37205243e+00 3.96136254e-01 9.74169612e-01
-1.14111924e+00 -1.03340983e+00 1.86489671e-01 2.69364536e-01
-1.52371728e+00 -1.29333246e+00 1.07367885e+00 -4.16546404e-01
3.85221720e-01 2.27182612e-01 6.76937938e-01 1.45228064e+00
-1.63005181e-02 5.66385269e-01 7.15477645e-01 -1.43640444e-01
1.02641344e-01 -1.83607310e-01 -5.90706468e-01 7.27431417e-01
-1.26001060e-01 1.66339070e-01 -8.96198153e-01 5.52291190e-03
7.03380227e-01 -2.20506757e-01 -6.56035364e-01 -4.54793096e-01
-1.16870379e+00 6.44616961e-01 1.61056712e-01 5.87535724e-02
-1.60619840e-01 2.44955029e-02 1.82442777e-02 -2.19055563e-01
3.68693024e-01 8.25013295e-02 -7.97222927e-02 -2.39209473e-01
-1.07695556e+00 3.10609639e-01 5.35600901e-01 1.19636321e+00
7.12663114e-01 2.97969967e-01 -2.29401425e-01 6.03075266e-01
3.85418802e-01 7.89558530e-01 4.03974146e-01 -1.34352672e+00
3.67074341e-01 1.70521706e-01 7.62550607e-02 -1.38617563e+00
-3.81683975e-01 -4.78084236e-02 -6.46207929e-01 2.93926839e-02
3.53636086e-01 -3.11291248e-01 -7.17837691e-01 2.15178370e+00
6.95493758e-01 4.95669991e-01 -2.77600408e-01 1.12419426e+00
8.61611784e-01 4.17010397e-01 -2.65738189e-01 -4.35432851e-01
1.00530756e+00 -8.72038722e-01 -1.14738286e+00 -2.54455060e-01
2.38633558e-01 -7.41348147e-01 7.48155117e-01 3.95549387e-02
-1.34276474e+00 -5.05533814e-01 -8.23652864e-01 -8.68411362e-02
2.21962526e-01 -2.36144401e-02 2.45135829e-01 3.80307883e-01
-1.08165967e+00 3.87160420e-01 -8.40461969e-01 -2.31026515e-01
4.13620830e-01 4.73772705e-01 -5.48012614e-01 -1.58413261e-01
-9.51406240e-01 6.21976435e-01 -1.35293886e-01 2.54829258e-01
-7.38616943e-01 -7.21679151e-01 -1.14617455e+00 -2.81165332e-01
2.35513613e-01 -7.86269784e-01 1.22842407e+00 -1.13024974e+00
-1.98254323e+00 8.06263328e-01 -6.16031349e-01 1.92230463e-01
7.54982591e-01 -1.40910983e-01 -5.13898730e-01 2.94854850e-01
1.00049493e-03 9.26090419e-01 1.15156925e+00 -1.30738211e+00
-2.80477345e-01 -3.59711021e-01 -2.42331684e-01 3.42788190e-01
-2.66748816e-01 6.56618103e-02 -8.03347111e-01 -8.08099687e-01
-3.11830156e-02 -1.16979182e+00 2.06532180e-01 3.14151943e-01
-3.53578240e-01 3.56466234e-01 9.79102790e-01 -7.04479039e-01
1.11683095e+00 -2.37870407e+00 3.23991060e-01 4.89624850e-02
2.28234917e-01 -7.41282776e-02 -2.87364483e-01 -4.02495125e-03
-5.18506393e-03 -2.12947994e-01 -2.83209328e-02 -6.34932876e-01
-8.99054706e-02 -4.78534261e-03 -2.09388405e-01 7.71698892e-01
1.67620897e-01 8.89460683e-01 -9.15493727e-01 -5.75757682e-01
1.69157326e-01 1.09795833e+00 -9.28609610e-01 3.60173494e-01
5.56007139e-02 1.08130145e+00 -4.02020752e-01 7.11174190e-01
8.87493491e-01 9.86327454e-02 1.74434230e-01 -6.13620102e-01
-1.84685718e-02 1.38854295e-01 -1.19552422e+00 1.94369566e+00
-2.91617155e-01 6.47596300e-01 3.07790428e-01 -4.41086203e-01
7.30918527e-01 4.76427138e-01 6.13544047e-01 -4.93626148e-01
2.44863793e-01 1.03900813e-01 -2.13672414e-01 -6.43199801e-01
3.85121435e-01 1.31621480e-01 1.41369984e-01 2.35482901e-01
8.60543549e-02 -3.58443797e-01 -1.90979764e-01 -7.15512931e-02
7.03507721e-01 4.27190244e-01 -9.86270830e-02 -1.14537962e-01
4.64651763e-01 -6.52264535e-01 9.72905815e-01 9.37985852e-02
-3.50990623e-01 1.12858665e+00 2.16988832e-01 -1.16742581e-01
-8.33582222e-01 -9.82330322e-01 7.07594166e-03 8.13354850e-01
1.07516550e-01 -3.32254082e-01 -1.09684515e+00 -3.22499335e-01
-2.45117754e-01 4.40908015e-01 -5.88489890e-01 -3.08956563e-01
-8.67822051e-01 -4.26531076e-01 3.81601393e-01 3.95275831e-01
3.32657903e-01 -8.45099449e-01 -3.24922919e-01 2.07541473e-02
-6.14892483e-01 -1.37206078e+00 -1.30415487e+00 -7.99334228e-01
-5.35172403e-01 -9.00180995e-01 -5.82033694e-01 -7.81076431e-01
1.03835189e+00 2.32932121e-01 7.23381341e-01 -9.25494805e-02
-2.24232048e-01 4.84318018e-01 -7.13556334e-02 -7.72497728e-02
-2.66128272e-01 -2.21258298e-01 4.58856821e-01 5.85310519e-01
3.41986828e-02 -6.63581014e-01 -8.50997448e-01 3.82187277e-01
-6.79437280e-01 2.21818313e-01 1.19839258e-01 7.57433414e-01
3.72619927e-01 -3.83729964e-01 4.08610255e-01 -2.19216809e-01
2.20833004e-01 -2.86692977e-01 -3.97630632e-01 -1.17315426e-01
-1.56387631e-02 -2.25409284e-01 4.06466007e-01 -7.36351252e-01
-1.12873328e+00 3.61070901e-01 8.55207741e-02 -7.29892492e-01
-8.35616589e-02 -1.80011407e-01 -8.04477990e-01 -1.80957645e-01
3.01543027e-01 1.90602332e-01 1.99039549e-01 -9.31607261e-02
4.27743375e-01 5.38677752e-01 8.77430260e-01 -4.59694058e-01
9.86121058e-01 6.83205247e-01 -8.90858844e-02 -9.95223284e-01
-5.91022372e-01 -1.99484285e-02 -7.18782246e-01 -5.93709052e-01
8.97520840e-01 -1.08452427e+00 -8.62876058e-01 7.84633875e-01
-1.17492855e+00 -4.21199679e-01 3.14313583e-02 5.73820233e-01
-7.54731596e-01 3.05913925e-01 -5.17144382e-01 -5.39432526e-01
-2.88717985e-01 -1.33383322e+00 1.34210777e+00 3.01575661e-01
-3.29760462e-01 -7.66554058e-01 -1.60338223e-01 5.09386003e-01
3.85144144e-01 4.64378744e-01 2.21638545e-01 1.21676110e-01
-7.52607107e-01 -1.12547345e-01 1.88917607e-01 8.17954168e-02
4.51988876e-01 2.14614034e-01 -1.14713323e+00 -5.89288354e-01
3.71497013e-02 -2.46063441e-01 2.94570863e-01 4.56886023e-01
9.37750757e-01 -6.97319984e-01 -2.09962770e-01 1.11024880e+00
7.29208946e-01 2.85319276e-02 4.94732410e-01 -1.88816455e-04
1.01255786e+00 9.73034203e-01 4.62146550e-01 5.06289899e-01
5.79330266e-01 1.17019045e+00 1.84545696e-01 1.72957763e-01
-3.41658652e-01 -5.34127235e-01 6.86866343e-01 9.27973390e-01
-6.87307045e-02 -7.03045428e-02 -5.43979585e-01 3.41521263e-01
-1.63901365e+00 -1.19708323e+00 3.60657901e-01 2.22111225e+00
8.41654837e-01 -3.17952096e-01 1.03312120e-01 -1.24584921e-01
8.18484306e-01 2.30397820e-01 -7.37356663e-01 1.21887267e-01
-1.39339134e-01 -7.67854601e-02 1.04397409e-01 7.41394579e-01
-9.03956652e-01 8.96632612e-01 5.68508291e+00 3.46938699e-01
-1.52517354e+00 1.36555301e-03 4.87544835e-01 -5.80105424e-01
-3.30361098e-01 -3.04129660e-01 -8.34347486e-01 6.14219964e-01
6.31206512e-01 -3.25618297e-01 4.77383912e-01 7.02154756e-01
6.72823548e-01 3.41572374e-01 -1.10417783e+00 1.17464852e+00
6.44510210e-01 -1.22146451e+00 -7.97927156e-02 2.03271374e-01
7.41744220e-01 -2.89264053e-01 3.29061300e-01 -1.54598221e-01
-1.39834672e-01 -9.26525950e-01 1.06939173e+00 6.08210802e-01
1.01821983e+00 -6.90353274e-01 1.32064924e-01 -6.46514399e-03
-1.18378687e+00 9.99146253e-02 1.03562437e-01 2.23001987e-01
4.49152589e-01 1.25687987e-01 -6.11321926e-01 2.99596548e-01
5.68980038e-01 7.24629939e-01 -4.13912863e-01 5.91327608e-01
-3.33103806e-01 2.35767275e-01 -3.18912089e-01 4.41870362e-01
-3.03815365e-01 -8.58326331e-02 7.50648201e-01 1.02062213e+00
5.61919928e-01 1.70010298e-01 1.28005549e-01 7.92936981e-01
-9.91150886e-02 -1.94637720e-02 -5.92798412e-01 3.59360635e-01
6.83658123e-01 1.23046482e+00 -1.31122246e-01 -1.03836447e-01
-4.24029708e-01 1.02742398e+00 3.42843682e-02 4.65086669e-01
-9.59857464e-01 1.79593568e-03 1.21816468e+00 3.81076038e-01
2.70028442e-01 -3.16433161e-01 1.21481627e-01 -1.55540669e+00
2.37622157e-01 -9.57004011e-01 -1.67566296e-02 -8.27322066e-01
-9.38039541e-01 6.74837649e-01 -4.30523269e-02 -1.43176973e+00
-6.66072190e-01 -1.70863658e-01 -6.55352056e-01 5.80357790e-01
-1.23391128e+00 -1.44148636e+00 -6.44044757e-01 8.51574898e-01
5.35508275e-01 7.11373016e-02 6.26479626e-01 3.96059573e-01
-7.41132379e-01 1.10468125e+00 -2.53978372e-01 1.65448576e-01
1.14893794e+00 -5.60294986e-01 1.90997198e-01 8.25397134e-01
-2.75665313e-01 5.58503926e-01 7.90430665e-01 -4.52343673e-01
-1.73555684e+00 -1.23178375e+00 6.65298641e-01 -5.51284075e-01
3.89286220e-01 -6.37106061e-01 -8.34083915e-01 8.73554111e-01
1.73893962e-02 4.34705496e-01 4.99502093e-01 -5.67834258e-01
-3.09544891e-01 -2.18571350e-01 -1.24059188e+00 7.81378031e-01
1.34746575e+00 -5.52749395e-01 -1.98784843e-01 9.08492357e-02
5.72495580e-01 -7.26215243e-01 -7.68720686e-01 3.71990293e-01
8.63080442e-01 -9.58216071e-01 8.53137076e-01 -1.97122082e-01
2.37605289e-01 -3.66158813e-01 -1.88602135e-01 -1.24708164e+00
-2.51180381e-01 -1.20024610e+00 -3.20960104e-01 1.53729725e+00
7.55442157e-02 -5.05940616e-01 7.32292712e-01 9.77178037e-01
-1.35978973e-02 -4.10038292e-01 -9.00268674e-01 -5.55560112e-01
-1.75380498e-01 -2.07357645e-01 8.00996721e-01 1.01861501e+00
4.35578488e-02 2.33107224e-01 -6.62438512e-01 2.87705809e-01
7.37866879e-01 2.55224369e-02 1.12114811e+00 -7.54387379e-01
-8.39211717e-02 -2.34882563e-01 -5.49998999e-01 -1.05440998e+00
5.80436170e-01 -5.51968336e-01 1.62148297e-01 -8.95712435e-01
2.12548655e-02 -9.32083875e-02 2.45467722e-01 2.18675986e-01
-1.74267307e-01 3.71576756e-01 4.00089532e-01 1.66935608e-01
-2.87599385e-01 8.19855213e-01 1.49577928e+00 9.83511209e-02
-2.79850066e-01 -1.75622866e-01 -5.13498902e-01 8.64925861e-01
7.31736362e-01 -2.08007589e-01 -4.56973761e-01 -5.64710259e-01
-3.12877327e-01 2.02666998e-01 3.40466917e-01 -8.71795177e-01
2.95594156e-01 -3.38948131e-01 4.80660617e-01 -1.34244785e-01
7.63957262e-01 -5.37815213e-01 3.59478056e-01 8.69135931e-02
-1.89494640e-01 9.70380604e-02 8.95799324e-02 4.07810271e-01
-1.69553921e-01 3.71357262e-01 9.44680154e-01 1.76863998e-01
-3.54220361e-01 6.83076441e-01 -1.03641607e-01 1.53097615e-01
9.13017035e-01 -2.79639721e-01 -6.77718073e-02 -9.93841767e-01
-4.61569637e-01 6.72432184e-02 8.90613258e-01 7.95654595e-01
5.95174551e-01 -1.48851204e+00 -7.05730557e-01 5.59432089e-01
8.55450332e-02 1.30817741e-01 5.41677594e-01 8.88796210e-01
-1.36723384e-01 2.29643229e-02 -1.17975019e-01 -6.90892458e-01
-1.40397930e+00 4.22893435e-01 3.26760411e-01 4.34728503e-01
-6.38952732e-01 5.87794781e-01 5.59987009e-01 -2.75920630e-01
1.41210794e-01 1.32749034e-02 -3.59236239e-03 -3.51248570e-02
7.52547503e-01 2.93296397e-01 -2.32734561e-01 -1.43965673e+00
-4.52752411e-01 9.57099617e-01 9.44860727e-02 -1.70924321e-01
1.14753759e+00 -4.47772652e-01 1.94219917e-01 4.78670001e-02
1.37210667e+00 3.08991641e-01 -1.85968459e+00 7.03960378e-03
-4.63073194e-01 -7.70458877e-01 -7.45897442e-02 -3.29063058e-01
-1.37859333e+00 6.23416185e-01 4.52856630e-01 -6.87658548e-01
1.24138498e+00 -1.52811557e-01 9.26954806e-01 -8.26386735e-02
3.37616891e-01 -1.01899993e+00 2.16441721e-01 2.45402709e-01
1.09295988e+00 -1.11361337e+00 -1.80129036e-01 -5.82771540e-01
-6.68546200e-01 1.05265355e+00 7.22665370e-01 2.72327662e-01
5.21760404e-01 4.03922081e-01 3.26930910e-01 1.39858574e-01
-6.51806712e-01 1.32656455e-01 2.94168204e-01 8.61784458e-01
3.91971081e-01 -7.48941302e-02 8.21209997e-02 3.27290565e-01
-5.66376805e-01 1.12666517e-01 4.84071106e-01 7.88894534e-01
3.48252133e-02 -8.47805917e-01 -5.79205990e-01 -1.38076589e-01
-3.16231132e-01 1.91279709e-01 -1.56978592e-01 4.73821402e-01
8.61153826e-02 1.06730354e+00 8.35813582e-02 -4.73349661e-01
3.68420154e-01 -1.61287896e-02 6.84242725e-01 -3.47426027e-01
-4.02063094e-02 3.11886579e-01 -7.44501427e-02 -9.42894518e-01
-4.92987216e-01 -8.46265435e-01 -1.14294577e+00 -5.95988452e-01
-8.14885646e-02 -1.95341021e-01 5.41873336e-01 7.02344537e-01
8.72664928e-01 1.52417660e-01 9.78596270e-01 -1.49860060e+00
-2.83456534e-01 -8.61458123e-01 -2.89676964e-01 6.90108597e-01
6.17085695e-01 -8.23980927e-01 -5.34820497e-01 5.35834789e-01] | [13.111870765686035, -0.3991010785102844] |
ee931418-5ea4-4dca-b983-a83185dd6b34 | learning-logic-specifications-for-soft-policy | 2303.09172 | null | https://arxiv.org/abs/2303.09172v1 | https://arxiv.org/pdf/2303.09172v1.pdf | Learning Logic Specifications for Soft Policy Guidance in POMCP | Partially Observable Monte Carlo Planning (POMCP) is an efficient solver for Partially Observable Markov Decision Processes (POMDPs). It allows scaling to large state spaces by computing an approximation of the optimal policy locally and online, using a Monte Carlo Tree Search based strategy. However, POMCP suffers from sparse reward function, namely, rewards achieved only when the final goal is reached, particularly in environments with large state spaces and long horizons. Recently, logic specifications have been integrated into POMCP to guide exploration and to satisfy safety requirements. However, such policy-related rules require manual definition by domain experts, especially in real-world scenarios. In this paper, we use inductive logic programming to learn logic specifications from traces of POMCP executions, i.e., sets of belief-action pairs generated by the planner. Specifically, we learn rules expressed in the paradigm of answer set programming. We then integrate them inside POMCP to provide soft policy bias toward promising actions. In the context of two benchmark scenarios, rocksample and battery, we show that the integration of learned rules from small task instances can improve performance with fewer Monte Carlo simulations and in larger task instances. We make our modified version of POMCP publicly available at https://github.com/GiuMaz/pomcp_clingo.git. | ['Alessandro Farinelli', 'Alberto Castellini', 'Daniele Meli', 'Giulio Mazzi'] | 2023-03-16 | null | null | null | null | ['inductive-logic-programming'] | ['methodology'] | [ 5.34023456e-02 3.08901548e-01 -4.91121441e-01 -1.24796920e-01
-9.87822473e-01 -8.27191234e-01 6.57009780e-01 2.24428028e-01
-4.51372176e-01 1.28332591e+00 3.41167331e-01 -5.63022912e-01
-3.40775847e-01 -8.58331323e-01 -8.80597591e-01 -5.90294182e-01
-4.73166049e-01 9.27069068e-01 3.55465323e-01 8.66314322e-02
5.24928309e-02 2.71078825e-01 -1.34038937e+00 3.88192803e-01
6.20880604e-01 8.55057478e-01 3.05392534e-01 6.90978408e-01
2.21740529e-01 1.17468059e+00 -2.24923864e-01 -1.31211989e-02
3.94530892e-01 -2.47147486e-01 -1.04584360e+00 -1.49625257e-01
-5.62153578e-01 -5.75200558e-01 -2.08296537e-01 1.02064788e+00
2.90155202e-01 3.59153599e-01 3.26641321e-01 -1.42690730e+00
1.07152671e-01 1.12866330e+00 6.32641315e-02 -3.54617275e-02
6.24921620e-01 9.18575883e-01 9.52339113e-01 -9.29595456e-02
6.84843302e-01 1.51404703e+00 7.61065483e-02 6.37621403e-01
-1.13272953e+00 -2.40610272e-01 5.75245678e-01 4.61884528e-01
-1.05631065e+00 -2.31016442e-01 3.05550635e-01 -1.48085266e-01
1.17360663e+00 3.46352577e-01 9.24116075e-01 1.42271173e+00
5.84267735e-01 1.10683656e+00 1.62375391e+00 -1.39774263e-01
1.05617571e+00 -2.76991427e-01 -9.81276184e-02 4.59630191e-01
1.01394817e-01 6.72998965e-01 -4.55192596e-01 -5.00489533e-01
5.42772710e-01 -4.32450362e-02 -3.47420238e-02 -5.41593373e-01
-1.19954693e+00 7.18210757e-01 1.23691745e-01 -1.96791157e-01
-7.31120586e-01 6.00761294e-01 3.46998334e-01 2.47535273e-01
-1.49961904e-01 4.95858639e-01 -3.29518318e-01 -5.97228527e-01
-4.14508224e-01 8.92441988e-01 1.16196632e+00 9.21831846e-01
4.22814786e-01 -2.63783544e-01 -6.57631457e-01 -1.81054041e-01
3.95324588e-01 6.01154327e-01 5.63317351e-02 -1.50232255e+00
5.58837771e-01 2.74312407e-01 7.89320111e-01 -1.90721855e-01
-4.36063737e-01 -8.10890943e-02 -9.41662863e-02 4.17836398e-01
3.89370531e-01 -3.92522633e-01 -8.26969206e-01 1.61146450e+00
4.87913162e-01 -9.56358202e-03 2.79121518e-01 9.14182782e-01
-1.86771750e-02 7.14275122e-01 9.64077786e-02 -4.12179768e-01
1.28354514e+00 -8.03754032e-01 -7.04037428e-01 -5.00064135e-01
3.12086403e-01 1.69938710e-02 9.91192102e-01 7.10669041e-01
-1.13217854e+00 1.32918939e-01 -9.28885043e-01 5.84648311e-01
1.37694821e-01 -5.52816272e-01 8.82047415e-01 3.52638543e-01
-7.38807678e-01 7.03759909e-01 -1.53238523e+00 4.10878174e-02
4.61880922e-01 2.79802680e-01 2.78388560e-01 -2.06939965e-01
-1.20351434e+00 1.00527990e+00 8.06436598e-01 -3.69085073e-02
-2.01008058e+00 -2.90794391e-02 -6.57058835e-01 1.56276464e-01
1.23235631e+00 -5.88113248e-01 1.81544864e+00 -2.81742215e-01
-1.96147382e+00 1.15475230e-01 1.03803925e-01 -9.24879074e-01
8.26660156e-01 -2.71162748e-01 -6.33519292e-02 4.29808907e-02
-8.84805471e-02 5.94945073e-01 6.18497312e-01 -1.00242710e+00
-6.60028160e-01 -1.61832809e-01 7.08696902e-01 4.59213972e-01
4.77544665e-01 9.38019063e-03 -1.51867300e-01 1.58684209e-01
-1.92574784e-01 -1.25497687e+00 -7.55383193e-01 -4.73784268e-01
-4.02222961e-01 -1.98102772e-01 1.48513056e-02 -2.03617990e-01
9.10909593e-01 -1.71950328e+00 3.02842259e-01 1.87899083e-01
-8.83218646e-02 -1.76559359e-01 1.25428876e-02 6.56222403e-01
4.86988991e-01 -1.96954116e-01 -2.51699448e-01 -3.14305956e-03
5.08585632e-01 5.50455987e-01 -5.16023576e-01 3.70178968e-01
-4.54041474e-02 9.41834867e-01 -1.23866439e+00 -4.00503218e-01
1.97377458e-01 -3.06359798e-01 -7.23444164e-01 7.41820782e-02
-1.15185797e+00 7.91767061e-01 -8.62977564e-01 5.69476247e-01
2.29260445e-01 -3.47374007e-02 6.79649889e-01 7.36352026e-01
-1.04264446e-01 6.30032361e-01 -1.44647968e+00 1.71632111e+00
-4.39742118e-01 -3.00877076e-02 3.15921158e-02 -5.19018710e-01
4.43914413e-01 4.34156507e-01 3.19567680e-01 -4.06156003e-01
3.31115574e-01 7.54198357e-02 -3.31169255e-02 -2.15560898e-01
3.89926225e-01 -2.96558112e-01 -5.45828760e-01 6.85577750e-01
-1.58308819e-01 -4.94875908e-01 2.43759036e-01 -3.85097824e-02
1.50011849e+00 4.48223144e-01 6.30244434e-01 -3.68553959e-02
2.25307390e-01 5.02975047e-01 7.44868338e-01 1.23835421e+00
-4.35327113e-01 -2.88820155e-02 9.09029186e-01 -4.47009325e-01
-5.75740159e-01 -1.18045771e+00 8.81842300e-02 9.45032895e-01
3.54342014e-01 -4.15916950e-01 -7.17956662e-01 -6.87090099e-01
-8.83326530e-02 1.23515344e+00 -2.97606498e-01 -1.03308663e-01
-6.42880082e-01 -5.54517090e-01 2.82376021e-01 4.03300613e-01
2.98894882e-01 -1.52016401e+00 -1.33693528e+00 4.47853506e-01
-2.18192637e-01 -1.07932460e+00 5.72058074e-02 4.92869169e-01
-9.29682612e-01 -1.06159174e+00 -1.48637518e-01 8.23893175e-02
3.20094019e-01 -3.36547643e-01 1.00289011e+00 -4.22698677e-01
2.35343128e-01 5.43818116e-01 -3.56851667e-01 -6.49368584e-01
-5.09584010e-01 -1.78117901e-01 2.45664641e-01 -5.52967250e-01
3.45438197e-02 -4.02219206e-01 -3.41009468e-01 2.81003356e-01
-6.39719188e-01 3.04811597e-01 4.33005244e-01 6.98360980e-01
7.58319199e-01 1.68340728e-01 -2.22390387e-02 -5.62694669e-01
8.26727569e-01 -4.12701309e-01 -1.10265040e+00 2.78497368e-01
-3.27433050e-01 4.77321088e-01 4.72045213e-01 -3.21977079e-01
-1.17035592e+00 1.49190292e-01 1.49713874e-01 -1.80609465e-01
-3.09833080e-01 5.93171120e-01 -2.22113699e-01 4.46622163e-01
6.59506142e-01 1.79064557e-01 -2.52496511e-01 -1.20341867e-01
3.64879817e-01 3.02357852e-01 3.28037679e-01 -1.17450833e+00
4.89818394e-01 6.35751843e-01 5.74836098e-02 8.96354169e-02
-8.21803272e-01 -8.56262352e-03 -7.81677887e-02 -2.62396514e-01
6.24835312e-01 -8.14800143e-01 -1.29851079e+00 1.42696902e-01
-8.51286650e-01 -1.01254237e+00 -7.28019953e-01 6.51152074e-01
-1.32855487e+00 -4.64439839e-02 -5.47640026e-01 -1.34204578e+00
-1.16490826e-01 -1.49509728e+00 8.29466641e-01 1.75852761e-01
-2.70126969e-01 -5.74347854e-01 2.36358479e-01 9.68991220e-02
-1.74669381e-02 3.67201269e-01 5.42425811e-01 -4.19540852e-01
-9.68211651e-01 6.45754952e-03 4.71989214e-01 -1.39072597e-01
-4.41538334e-01 -5.23477018e-01 -8.45919132e-01 -3.53607684e-01
1.86354980e-01 -4.78840202e-01 4.64073956e-01 5.04099786e-01
1.01732326e+00 -6.61541343e-01 -3.75424325e-01 1.92259312e-01
1.21071696e+00 3.54922354e-01 4.67976540e-01 6.16485834e-01
-5.90172485e-02 5.16366422e-01 1.23332405e+00 9.60903645e-01
4.02696669e-01 6.02284491e-01 9.16216373e-01 9.17249084e-01
4.70079243e-01 -4.95740801e-01 9.13991749e-01 6.36996403e-02
-2.62724072e-01 -2.58460343e-01 -1.13392413e+00 3.90573919e-01
-2.32334590e+00 -9.48737383e-01 3.13311696e-01 2.14192104e+00
9.69545305e-01 6.83609068e-01 1.66384786e-01 -1.90928176e-01
4.01381701e-01 1.06194671e-02 -8.91921401e-01 -2.84451634e-01
3.85228872e-01 1.51718676e-01 7.30242252e-01 7.43502557e-01
-9.00999010e-01 1.13384378e+00 5.58582306e+00 7.67696202e-01
-5.95895469e-01 3.91224384e-01 8.41244534e-02 -6.13394201e-01
-2.36548021e-01 5.29755533e-01 -9.00522411e-01 3.90369475e-01
1.09572160e+00 -3.98183525e-01 9.33263004e-01 9.20367301e-01
2.81499624e-01 -5.98874509e-01 -1.35049367e+00 5.70281625e-01
-7.95843661e-01 -1.20744157e+00 -3.93302411e-01 7.57154673e-02
5.92723489e-01 2.66136229e-01 -9.56879258e-02 6.31324291e-01
1.42143464e+00 -9.55357134e-01 1.38578260e+00 3.75990480e-01
3.41555864e-01 -8.19535434e-01 6.38009250e-01 9.59037662e-01
-8.77545297e-01 -6.02065086e-01 -2.96585828e-01 -6.02151573e-01
6.11067832e-01 3.17835152e-01 -1.15007544e+00 5.51754832e-01
5.58884203e-01 2.30536595e-01 2.49703191e-02 9.37540054e-01
-9.85880673e-01 5.11561334e-01 -5.28468430e-01 -3.89940321e-01
5.75060308e-01 -7.14051500e-02 8.24964225e-01 5.20640731e-01
1.02711655e-01 2.04835907e-01 5.52261412e-01 1.10721040e+00
5.58021426e-01 -6.26717925e-01 -4.51619178e-01 -6.24173470e-02
6.63902938e-01 9.45955098e-01 -6.88627064e-01 -3.93932432e-01
2.59908289e-01 4.89772499e-01 2.96889067e-01 3.16262305e-01
-1.13226056e+00 3.52823526e-01 7.07548201e-01 -2.51510531e-01
1.26144975e-01 -3.35259050e-01 -1.61847770e-01 -9.32711184e-01
-4.29596491e-02 -1.07581031e+00 5.48245668e-01 -6.62207127e-01
-7.75168300e-01 2.88580298e-01 5.17172396e-01 -1.08815944e+00
-7.32635498e-01 -3.64148945e-01 -5.33732355e-01 5.37623882e-01
-1.16693413e+00 -5.74569821e-01 6.02638386e-02 4.55423653e-01
3.86345893e-01 2.84795016e-01 8.13337147e-01 -5.20125926e-01
-4.04257506e-01 -1.35034218e-01 -8.72682407e-02 -4.41215754e-01
1.20169818e-01 -1.29548466e+00 5.09463906e-01 6.97514653e-01
-1.09191962e-01 3.18387389e-01 1.04041636e+00 -8.76381874e-01
-1.70580971e+00 -8.24343264e-01 1.75482452e-01 -5.56097925e-01
7.09475458e-01 -3.73983145e-01 -4.47894335e-01 9.49061930e-01
-1.22724930e-02 -2.18642876e-01 3.94353233e-02 4.37808074e-02
2.51035035e-01 3.37945163e-01 -1.16647995e+00 8.63500714e-01
1.17684925e+00 -2.08198115e-01 -6.86230838e-01 5.39848387e-01
7.39170909e-01 -8.79498899e-01 -6.98189557e-01 2.53086090e-01
4.37376291e-01 -8.45758915e-01 7.43257403e-01 -7.03387678e-01
1.63584426e-01 -4.63372141e-01 -1.27038687e-01 -1.35528088e+00
-1.56739146e-01 -9.08891082e-01 -5.61355889e-01 4.85432774e-01
3.01072627e-01 -6.49151862e-01 6.64203107e-01 6.95710361e-01
-1.02296151e-01 -7.55975544e-01 -1.19361246e+00 -1.10452747e+00
-8.47097859e-02 -8.35313737e-01 8.90542567e-01 2.69828767e-01
4.12421763e-01 -3.46824713e-02 -2.43442982e-01 2.90717721e-01
7.33956575e-01 3.16712677e-01 6.28271461e-01 -6.22323453e-01
-8.54490817e-01 -1.16192989e-01 1.54643506e-01 -9.33710158e-01
2.34304026e-01 -7.52686143e-01 5.47280610e-01 -1.61256397e+00
2.44177997e-01 -5.87116241e-01 -8.47666338e-02 7.40732551e-01
1.10676073e-01 -7.28365541e-01 6.31842375e-01 1.90817267e-01
-1.12856841e+00 6.55062199e-01 1.52155137e+00 -1.10433489e-01
-3.60984474e-01 3.12500030e-01 -3.34699571e-01 8.69333208e-01
1.10015786e+00 -6.67274296e-01 -3.32305729e-01 -2.11965784e-01
5.43490350e-01 8.57472956e-01 4.43352520e-01 -1.11047494e+00
1.40693009e-01 -8.83651435e-01 -1.76027879e-01 -4.99447018e-01
3.91665965e-01 -7.22347975e-01 4.83241737e-01 9.90493357e-01
-4.36491191e-01 -1.28815442e-01 5.83966151e-02 8.47889066e-01
1.26349837e-01 -4.61880624e-01 2.93551564e-01 -4.17198211e-01
-8.37844849e-01 1.27885535e-01 -8.40180159e-01 4.79312576e-02
1.41164398e+00 1.44905895e-01 -2.82985032e-01 -5.16040087e-01
-1.06452298e+00 7.84576595e-01 4.89505142e-01 9.11302269e-02
4.85590398e-01 -1.04007995e+00 -4.25483853e-01 -2.74126589e-01
-2.29708552e-02 2.76333690e-01 1.13253698e-01 9.65324044e-01
-2.76907057e-01 5.35240769e-01 -4.27811921e-01 -3.37879837e-01
-7.53496706e-01 6.09246731e-01 3.44202518e-01 -9.01228547e-01
-6.56976044e-01 4.11830842e-01 -3.64430957e-02 -5.22393167e-01
3.82248849e-01 -7.86706805e-01 -6.73209364e-03 -3.23568642e-01
5.35713017e-01 4.64321971e-01 -2.07080647e-01 2.74329841e-01
-5.26615024e-01 -3.05440187e-01 3.32277745e-01 -8.00289512e-01
1.28726280e+00 8.89798105e-02 1.89405605e-01 3.68715137e-01
1.26173794e-01 -3.88561070e-01 -1.87116873e+00 -1.54454663e-01
2.84265131e-01 -3.17645878e-01 -2.12614629e-02 -9.92372692e-01
-3.40616465e-01 5.59989810e-01 3.15870136e-01 6.32835506e-03
6.60886109e-01 6.74631074e-02 5.30007660e-01 7.55274057e-01
1.28593385e+00 -1.15358770e+00 -1.85235754e-01 7.92133331e-01
7.67209709e-01 -9.29574013e-01 5.90456836e-02 -2.14183014e-02
-1.03137624e+00 6.90663397e-01 3.87129486e-01 4.08307128e-02
1.07424736e-01 4.82034355e-01 -4.22312915e-01 -1.58102259e-01
-1.35250640e+00 -3.45634699e-01 -2.90895343e-01 4.82828379e-01
-4.84948158e-01 7.43282378e-01 -1.08707212e-01 6.84275746e-01
-1.91991210e-01 2.94403523e-01 5.84817588e-01 1.62598276e+00
-6.85983896e-01 -1.01971936e+00 -6.98525250e-01 3.95374596e-01
-2.74430484e-01 1.17673159e-01 -1.92217261e-01 4.30975407e-01
-2.22967878e-01 9.23305094e-01 -8.14808309e-02 -1.58570930e-01
2.29920059e-01 1.31832644e-01 8.55858564e-01 -7.21361995e-01
-4.85228509e-01 -1.42670304e-01 4.37059075e-01 -1.15026927e+00
-1.03533730e-01 -9.77741122e-01 -1.54372180e+00 -3.64039898e-01
1.27039989e-02 2.24687025e-01 4.45877314e-01 1.03369594e+00
1.39428690e-01 6.34435356e-01 2.64244974e-01 -8.92984331e-01
-1.33150363e+00 -7.32802331e-01 -5.38426459e-01 -1.48664355e-01
-6.86274609e-03 -8.24135900e-01 -1.37690350e-01 -3.47021371e-01] | [4.273430824279785, 2.1591553688049316] |
f4beca2a-b319-4a10-bade-46a8f52d5488 | sdfdiff-differentiable-rendering-of-signed | 1912.07109 | null | https://arxiv.org/abs/1912.07109v2 | https://arxiv.org/pdf/1912.07109v2.pdf | SDFDiff: Differentiable Rendering of Signed Distance Fields for 3D Shape Optimization | We propose SDFDiff, a novel approach for image-based shape optimization using differentiable rendering of 3D shapes represented by signed distance functions (SDFs). Compared to other representations, SDFs have the advantage that they can represent shapes with arbitrary topology, and that they guarantee watertight surfaces. We apply our approach to the problem of multi-view 3D reconstruction, where we achieve high reconstruction quality and can capture complex topology of 3D objects. In addition, we employ a multi-resolution strategy to obtain a robust optimization algorithm. We further demonstrate that our SDF-based differentiable renderer can be integrated with deep learning models, which opens up options for learning approaches on 3D objects without 3D supervision. In particular, we apply our method to single-view 3D reconstruction and achieve state-of-the-art results. | ['Zhizhong Han', 'Matthias Zwicker', 'Dantong Ji', 'Yue Jiang'] | 2019-12-15 | sdfdiff-differentiable-rendering-of-signed-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Jiang_SDFDiff_Differentiable_Rendering_of_Signed_Distance_Fields_for_3D_Shape_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Jiang_SDFDiff_Differentiable_Rendering_of_Signed_Distance_Fields_for_3D_Shape_CVPR_2020_paper.pdf | cvpr-2020-6 | ['single-view-3d-reconstruction'] | ['computer-vision'] | [-9.90900770e-02 -3.29329744e-02 3.31173778e-01 -3.37468565e-01
-8.67921293e-01 -5.38941324e-01 5.52909195e-01 -2.93615043e-01
3.56353000e-02 3.01822752e-01 7.88104236e-02 -1.69513896e-01
-1.05699701e-02 -9.97750938e-01 -8.92409265e-01 -3.94960761e-01
-7.92376846e-02 8.24922740e-01 8.62875357e-02 -1.80982381e-01
1.80604294e-01 1.19118798e+00 -1.45797813e+00 2.99811542e-01
7.69444108e-01 1.08949959e+00 1.26914635e-01 5.40945172e-01
-2.83415586e-01 2.88563460e-01 -1.29876956e-01 -2.41152406e-01
3.86558861e-01 3.20797950e-01 -7.01390803e-01 2.82630771e-01
9.37782049e-01 -5.49390316e-01 -1.27390862e-01 6.33297086e-01
5.56137741e-01 -1.18904471e-01 9.34674144e-01 -9.93117273e-01
-8.43004763e-01 -2.71376163e-01 -6.75393343e-01 -2.65144497e-01
4.55326587e-01 1.52456434e-02 8.72382820e-01 -1.47335947e+00
8.35086823e-01 1.74234641e+00 6.42511785e-01 5.30854940e-01
-1.52729642e+00 -2.54755557e-01 4.97816443e-01 -2.70355105e-01
-9.01233017e-01 -3.89236093e-01 1.05916488e+00 -5.59960246e-01
9.75258172e-01 2.33325571e-01 9.01889801e-01 8.44995618e-01
1.99842378e-01 1.02881527e+00 1.17334723e+00 -1.46806717e-01
7.37846456e-03 -3.63009721e-01 -3.03784668e-01 1.08276677e+00
-1.63588777e-01 2.42235675e-01 -3.83749664e-01 -3.65567118e-01
1.44104707e+00 4.45963517e-02 -2.40101904e-01 -8.91450763e-01
-1.26251197e+00 7.09885836e-01 5.23505449e-01 -1.41589999e-01
-3.53853643e-01 5.14334977e-01 9.04320851e-02 1.70363113e-01
1.09388602e+00 2.16011226e-01 -5.23100197e-01 1.13308214e-01
-4.53662217e-01 3.92387629e-01 6.58269525e-01 1.03978741e+00
6.95327103e-01 2.78872430e-01 1.29556209e-01 7.20392883e-01
5.77821910e-01 6.04620695e-01 -3.96415949e-01 -1.38220274e+00
2.62299448e-01 5.20405591e-01 1.68221086e-01 -8.88327360e-01
-4.14595306e-01 -2.73177236e-01 -8.96045387e-01 8.50962639e-01
9.37539041e-02 2.61826783e-01 -7.95820415e-01 1.24634969e+00
5.27747333e-01 3.64266574e-01 -2.84124315e-01 8.24045658e-01
1.00748003e+00 5.44726372e-01 -5.26329339e-01 1.02165021e-01
9.85850692e-01 -8.39174509e-01 -1.99251339e-01 2.16352612e-01
2.05714777e-01 -5.95145583e-01 1.31441903e+00 4.26680267e-01
-1.55277920e+00 -2.01813191e-01 -7.37822175e-01 -3.42925847e-01
3.57606672e-02 -2.90744394e-01 6.41053975e-01 2.57098645e-01
-1.26377773e+00 9.10533905e-01 -9.88122642e-01 2.09550798e-01
9.53626633e-01 2.18207225e-01 -2.72489369e-01 -4.51461561e-02
-3.19973677e-01 6.25603378e-01 -3.26946408e-01 -6.11279942e-02
-1.01036716e+00 -1.21101499e+00 -8.55590999e-01 -1.64177328e-01
8.61202478e-02 -1.28586912e+00 1.13680494e+00 -3.88785452e-01
-1.68034518e+00 1.44970763e+00 -2.44237691e-01 1.44639343e-01
7.79301047e-01 -6.91787750e-02 2.70388424e-01 1.69001207e-01
-2.27711737e-01 3.88246715e-01 9.50354040e-01 -1.74742997e+00
-1.30879864e-01 -5.77238798e-01 8.58033970e-02 2.08459929e-01
9.28685814e-02 -2.36993819e-01 -2.61565268e-01 -6.63965285e-01
4.59659278e-01 -5.36658883e-01 -3.45827579e-01 9.24991965e-01
-2.69187778e-01 -1.51604190e-01 9.07117248e-01 -4.31699097e-01
3.73157084e-01 -2.01435733e+00 5.47234774e-01 1.64874598e-01
5.56087673e-01 -6.82322681e-02 -3.68359089e-01 8.13046992e-02
3.45370620e-01 3.57823491e-01 -3.26659650e-01 -8.02854836e-01
1.07909344e-01 3.58176410e-01 -2.66175538e-01 4.81681049e-01
4.05383408e-01 1.20724392e+00 -8.78491759e-01 -3.16376895e-01
2.76879996e-01 9.34989452e-01 -7.39032030e-01 3.04146886e-01
-4.25199211e-01 9.49704945e-01 -6.90818012e-01 8.18528950e-01
9.30726707e-01 -4.98071969e-01 -2.76468158e-01 -2.73779184e-01
-1.26025364e-01 2.30106100e-01 -1.05949187e+00 1.87307847e+00
-7.37505734e-01 2.48183757e-01 6.23861849e-01 -7.98031807e-01
1.08890033e+00 1.92207083e-01 7.67003894e-01 -5.44803500e-01
-1.36826009e-01 2.12215453e-01 -7.12842047e-01 -1.34774268e-01
2.66677231e-01 -2.96604246e-01 3.40896189e-01 5.54274797e-01
-4.04334247e-01 -9.03752446e-01 -4.57166940e-01 1.15744434e-02
7.31752813e-01 6.05328918e-01 -7.00798556e-02 -2.45045111e-01
2.61832207e-01 -3.99181038e-01 2.85255671e-01 3.55623513e-01
3.61837029e-01 9.07540381e-01 3.57366741e-01 -7.42756784e-01
-1.19977462e+00 -1.59230900e+00 -4.27639663e-01 5.53813338e-01
1.70008391e-01 -3.73530909e-02 -1.83214098e-01 -5.84429026e-01
5.44699728e-01 3.54202718e-01 -3.67171437e-01 4.73875523e-01
-9.85938728e-01 -2.15617567e-01 3.63101214e-02 5.98018706e-01
1.64816216e-01 -7.99711943e-01 -6.75821543e-01 1.86037511e-01
2.01057822e-01 -1.09757936e+00 -4.55248773e-01 -2.62958974e-01
-1.45048606e+00 -1.12420177e+00 -8.48672569e-01 -7.39963174e-01
8.12409163e-01 4.20131296e-01 1.55940652e+00 2.78619796e-01
-1.46543041e-01 9.36851084e-01 1.45324349e-01 -2.57087529e-01
-4.87237930e-01 -2.62990505e-01 -2.53113598e-01 -1.44020796e-01
-5.84977567e-01 -1.20466399e+00 -6.46499395e-01 3.65823388e-01
-6.82459712e-01 2.93168336e-01 1.08475611e-01 5.53678095e-01
1.22208560e+00 -6.31704986e-01 4.11726773e-01 -7.33494341e-01
3.53203505e-01 -4.12432998e-02 -9.81963933e-01 2.04298213e-01
-2.90870726e-01 6.91180602e-02 5.53233266e-01 -3.36834788e-01
-9.10741687e-01 -7.27579445e-02 -2.96574563e-01 -8.54312599e-01
-1.26597539e-01 1.90165892e-01 -1.34259596e-01 -6.07352972e-01
3.09238732e-01 1.05324291e-01 3.80050302e-01 -9.05380607e-01
5.58331251e-01 9.49194878e-02 2.92732388e-01 -8.97283614e-01
7.99156785e-01 1.14090371e+00 4.32461530e-01 -8.25787544e-01
-7.25870907e-01 -5.69869950e-02 -8.56628537e-01 -3.24482083e-01
6.15036011e-01 -9.23378646e-01 -1.05268466e+00 5.25345325e-01
-1.47166169e+00 -5.29833794e-01 -3.18898499e-01 9.61393192e-02
-1.22057426e+00 3.95220071e-01 -6.61288321e-01 -7.42454708e-01
-4.04220045e-01 -1.15010726e+00 1.81273222e+00 -2.29667306e-01
1.92390531e-01 -1.20102775e+00 -1.81620985e-01 1.40329361e-01
3.83992434e-01 7.47395873e-01 1.25439811e+00 2.17632920e-01
-1.19888544e+00 3.51454347e-01 -3.02908242e-01 1.10827945e-01
2.27693632e-01 5.11817820e-02 -9.22294199e-01 -4.12630171e-01
1.84627175e-02 -3.21128041e-01 7.40245998e-01 5.34824491e-01
1.66155767e+00 -2.30421707e-01 -3.42023045e-01 1.35982811e+00
1.40472877e+00 -3.40107083e-01 4.51073825e-01 1.57735988e-01
9.54242408e-01 4.10566688e-01 3.77021760e-01 5.93460321e-01
5.42791545e-01 8.17434847e-01 9.04949307e-01 -2.80622333e-01
-2.36415729e-01 -3.06738257e-01 -2.51612868e-02 8.88617873e-01
-3.11604142e-01 1.43165635e-02 -7.43889987e-01 1.61079273e-01
-1.68626976e+00 -5.53600729e-01 -3.39280754e-01 2.07309365e+00
4.81396019e-01 -1.71476871e-01 1.85904920e-01 -2.15967119e-01
1.85471117e-01 3.85584086e-01 -7.46671319e-01 -2.65040755e-01
-1.22295469e-01 3.04411292e-01 9.45742801e-02 7.76744246e-01
-9.53107595e-01 7.68387496e-01 7.08360004e+00 5.42551100e-01
-1.23358297e+00 5.18261045e-02 5.34922481e-01 -2.43782178e-01
-1.13484514e+00 -3.33246231e-01 -4.94714141e-01 2.31705271e-02
1.60064045e-02 -6.61223382e-02 5.61306536e-01 5.95082223e-01
2.42145225e-01 3.99658740e-01 -1.21071899e+00 1.20356286e+00
-1.02953687e-01 -1.85085285e+00 1.91175044e-01 2.80455947e-01
8.51506710e-01 1.78560734e-01 2.41489545e-01 -2.53398210e-01
3.70496899e-01 -1.21463847e+00 8.52595031e-01 6.64659262e-01
9.81798589e-01 -6.59630597e-01 -1.14433728e-01 5.51009417e-01
-1.22860086e+00 3.71473402e-01 -4.58948761e-01 1.78720951e-01
6.02141678e-01 7.62500942e-01 -3.77903014e-01 7.46954262e-01
5.40811718e-01 1.13999069e+00 -8.43604747e-03 9.91827428e-01
-6.85340837e-02 5.11478400e-03 -4.39542502e-01 1.45144701e-01
4.57939133e-02 -6.06333554e-01 8.66880357e-01 7.37242639e-01
3.57115358e-01 1.51829451e-01 3.23984444e-01 1.28214622e+00
-1.97142765e-01 -9.67637345e-04 -9.75130677e-01 4.05862033e-01
2.26862535e-01 9.73151147e-01 -5.40781319e-01 -1.06457859e-01
-3.95474494e-01 8.21846306e-01 7.60757744e-01 4.73767340e-01
-4.83256042e-01 3.77430528e-01 8.40070546e-01 4.21753109e-01
4.80700552e-01 -7.56316900e-01 -5.88525712e-01 -1.30767059e+00
3.57621104e-01 -6.64119422e-01 -4.14877385e-02 -8.56139064e-01
-1.67341316e+00 4.01960373e-01 -8.77939984e-02 -1.20366693e+00
1.63972780e-01 -8.69893372e-01 -5.46699405e-01 8.52819324e-01
-2.06172967e+00 -1.37070358e+00 -2.46823236e-01 4.62729096e-01
4.24987108e-01 9.97278988e-02 8.86190534e-01 1.95406362e-01
1.74658895e-01 9.36020389e-02 -1.58959385e-02 -2.47180909e-01
2.10881904e-01 -1.25863051e+00 8.47009122e-01 2.16928884e-01
2.44720489e-01 1.86643824e-01 6.20003827e-02 -6.07964337e-01
-1.92834938e+00 -9.87542510e-01 3.27203095e-01 -6.73074782e-01
2.65353262e-01 -3.92173499e-01 -1.02923977e+00 6.97447538e-01
-2.31761456e-01 4.56487179e-01 3.06155741e-01 1.36916963e-02
-3.55977297e-01 1.71142057e-01 -1.30261362e+00 4.78832543e-01
1.71339464e+00 -4.54437971e-01 -3.61019552e-01 3.46162468e-01
6.73539996e-01 -9.25827146e-01 -1.12452745e+00 5.09506345e-01
6.80431724e-01 -1.11875558e+00 1.60988641e+00 -7.76233673e-01
4.95209903e-01 -7.88920522e-02 -2.38602355e-01 -1.48299599e+00
-3.19584578e-01 -7.30410278e-01 -5.08726299e-01 6.50379896e-01
1.40384153e-01 -7.61908352e-01 8.13800514e-01 4.54136580e-01
-5.02880812e-01 -1.22979450e+00 -1.12497807e+00 -9.29513037e-01
4.47361290e-01 -3.72270852e-01 8.63766432e-01 7.48238623e-01
-7.24829495e-01 -2.12980285e-01 -2.15202078e-01 1.82787403e-01
9.85758066e-01 7.47911453e-01 7.96285033e-01 -1.59082389e+00
-1.96586922e-01 -6.34127319e-01 -1.38490602e-01 -1.64052570e+00
4.82146055e-01 -1.30245626e+00 -2.78018057e-01 -1.77036953e+00
-9.48017016e-02 -9.47837770e-01 1.66741475e-01 1.66559011e-01
1.37741014e-01 1.48713261e-01 2.91278958e-01 1.48810849e-01
-2.47300163e-01 9.53180075e-01 2.15989399e+00 -9.25431252e-02
-2.43479200e-02 7.94312060e-02 -4.15269494e-01 9.82803822e-01
3.89719009e-01 -2.93438524e-01 -1.24131262e-01 -9.95126963e-01
2.06874147e-01 1.69084683e-01 7.69061744e-01 -9.74002779e-02
-1.71154737e-01 -3.56997490e-01 3.99118006e-01 -8.82797539e-01
7.51855552e-01 -8.40044022e-01 5.52480482e-03 2.25355476e-01
-3.29456455e-03 1.06557764e-01 1.06740266e-01 4.82893676e-01
1.36256695e-01 1.43005118e-01 9.28573728e-01 -4.64322597e-01
-2.90376484e-01 9.33873475e-01 9.83569473e-02 2.47103110e-01
6.85379982e-01 -2.84519076e-01 -5.41205034e-02 -2.90018737e-01
-7.13209152e-01 3.26513022e-01 9.81517911e-01 2.36703232e-01
1.18383694e+00 -1.71029735e+00 -9.73580778e-01 3.52680534e-01
-1.42177626e-01 6.78205132e-01 -5.81255406e-02 5.25557160e-01
-6.24977350e-01 -1.24594308e-01 -3.07304133e-02 -1.06730211e+00
-1.04528344e+00 2.45167077e-01 4.55120116e-01 -8.76224414e-02
-1.22136319e+00 6.68135941e-01 3.49854827e-01 -8.21224749e-01
2.28335708e-01 -4.68108416e-01 2.11984932e-01 -3.13468874e-01
2.55980462e-01 4.28217947e-01 6.65657446e-02 -3.73903930e-01
-3.10379833e-01 1.20238566e+00 4.07720119e-01 7.08973035e-02
1.94546640e+00 2.11110726e-01 -1.94928795e-01 4.50181037e-01
1.32188547e+00 5.81016317e-02 -1.77873993e+00 -1.07841358e-01
-3.25377077e-01 -9.98088956e-01 9.14108455e-02 -3.84289533e-01
-1.24995184e+00 1.05222285e+00 2.24352434e-01 2.26253569e-02
6.96888864e-01 2.85850853e-01 8.95771980e-01 2.17750058e-01
6.60901606e-01 -4.62389261e-01 5.39383233e-01 6.83818460e-01
1.46941948e+00 -1.14449811e+00 6.61950633e-02 -7.59239614e-01
-1.16861038e-01 1.25678003e+00 3.06478351e-01 -5.09083748e-01
8.77736092e-01 5.99458575e-01 -8.66115093e-02 -5.30667305e-01
-7.07769632e-01 8.47778469e-02 6.14596188e-01 9.16552961e-01
2.61913508e-01 -5.14415875e-02 2.57073492e-01 4.00556484e-03
-1.14053702e-02 -1.64258540e-01 2.17954472e-01 7.18161464e-01
-1.92090318e-01 -1.21983969e+00 -2.33953983e-01 3.88828605e-01
-1.14978313e-01 1.84871018e-01 -1.96305767e-01 6.24610722e-01
-4.74535704e-01 2.81442076e-01 1.69510394e-01 1.54781625e-01
7.15293586e-01 -1.65547907e-01 1.14333773e+00 -6.30306900e-01
-3.73801947e-01 2.36437008e-01 -1.78538263e-02 -8.73811722e-01
-4.62544501e-01 -6.37330472e-01 -1.22547579e+00 -3.81938785e-01
2.06972472e-02 -4.71917510e-01 6.72353506e-01 6.59403622e-01
7.28514910e-01 2.65997291e-01 9.97643828e-01 -1.51869035e+00
-6.50047600e-01 -1.99298695e-01 -5.22933662e-01 3.68250012e-01
5.59689522e-01 -7.49187648e-01 -2.57365674e-01 -1.87175110e-01] | [8.739187240600586, -3.5751750469207764] |
2d0f38cb-06c3-4ede-a9b6-db32a2ffe422 | discriminative-models-can-still-outperform-1 | 2206.02892 | null | https://arxiv.org/abs/2206.02892v1 | https://arxiv.org/pdf/2206.02892v1.pdf | Discriminative Models Can Still Outperform Generative Models in Aspect Based Sentiment Analysis | Aspect-based Sentiment Analysis (ABSA) helps to explain customers' opinions towards products and services. In the past, ABSA models were discriminative, but more recently generative models have been used to generate aspects and polarities directly from text. In contrast, discriminative models commonly first select aspects from the text, and then classify the aspect's polarity. Previous results showed that generative models outperform discriminative models on several English ABSA datasets. Here, we evaluate and contrast two state-of-the-art discriminative and generative models in several settings: cross-lingual, cross-domain, and cross-lingual and domain, to understand generalizability in settings other than English mono-lingual in-domain. Our more thorough evaluation shows that, contrary to previous studies, discriminative models can still outperform generative models in almost all settings. | ['Bilal Ghanem', 'Alona Fyshe', 'Dhruv Mullick'] | 2022-06-06 | null | null | null | null | ['aspect-based-sentiment-analysis'] | ['natural-language-processing'] | [-1.78237036e-01 6.75371066e-02 -3.78101587e-01 -9.06745255e-01
-1.06281352e+00 -1.02371955e+00 8.95292282e-01 5.14684990e-02
-1.51984468e-01 4.22722220e-01 5.70312142e-01 -3.85183156e-01
1.41801447e-01 -8.79508138e-01 -3.98845434e-01 -4.30337638e-01
5.56186497e-01 8.67332935e-01 -2.77342469e-01 -5.98619640e-01
1.19611718e-01 3.30916718e-02 -1.34160733e+00 6.99187994e-01
6.84005201e-01 8.65053296e-01 -2.03804359e-01 2.38025218e-01
-5.41631520e-01 3.69932324e-01 -6.11442566e-01 -1.24816000e+00
-5.01095969e-03 -5.54268539e-01 -6.11342371e-01 1.90787107e-01
2.13167161e-01 4.72917035e-02 1.30032361e-01 7.67260432e-01
5.14568508e-01 -2.22721547e-01 8.42194319e-01 -1.12461722e+00
-1.20169795e+00 7.68610835e-01 -4.45026040e-01 -7.81432763e-02
6.89657331e-01 -7.24559277e-02 1.64257228e+00 -8.64434838e-01
7.83716559e-01 1.42106771e+00 9.75589573e-01 2.98484445e-01
-1.53306901e+00 -4.34608102e-01 6.39098108e-01 -3.96147072e-02
-1.04205501e+00 -1.03179537e-01 9.10276115e-01 -2.69855797e-01
1.11376834e+00 2.76680380e-01 9.67446983e-01 1.57637346e+00
2.44250655e-01 1.25827181e+00 1.67136788e+00 -2.56770700e-01
1.42627448e-01 6.36022091e-01 1.06886692e-01 -8.49667639e-02
4.06171888e-01 -3.64299983e-01 -6.59184933e-01 -3.30769002e-01
1.43353254e-01 -1.71737164e-01 1.42127439e-01 -1.73129722e-01
-7.18591571e-01 1.44937229e+00 -2.07208127e-01 3.95067841e-01
-6.15276992e-01 -4.79302853e-01 2.68213093e-01 3.36821139e-01
9.46784675e-01 5.49952090e-01 -8.85972261e-01 -6.30310357e-01
-9.92592931e-01 5.61541319e-01 1.41930974e+00 1.13361740e+00
9.10765946e-01 1.57094792e-01 -1.55250579e-01 9.39881384e-01
5.59009850e-01 7.96646416e-01 6.50258362e-01 -2.40181372e-01
3.73578489e-01 6.42705142e-01 -1.99860662e-01 -1.05857360e+00
-2.12238759e-01 -4.92818385e-01 -1.96291611e-01 -2.94041306e-01
1.28136277e-01 -1.85132429e-01 -1.06844223e+00 1.43403411e+00
1.78448949e-02 -5.76248229e-01 4.56148200e-02 6.83749616e-01
9.03442264e-01 4.95058954e-01 1.36512727e-01 3.53809409e-02
1.67815042e+00 -5.30373991e-01 -7.08107293e-01 -7.16051817e-01
6.11579716e-01 -1.29614043e+00 1.31954026e+00 5.02430439e-01
-7.89856791e-01 -4.23877269e-01 -7.52499640e-01 -8.74200016e-02
-8.23721290e-01 1.03897505e-01 1.14912510e+00 1.11443233e+00
-8.97298992e-01 -9.15990025e-02 -5.52253604e-01 -5.08896232e-01
4.16566044e-01 1.53432757e-01 -2.09490180e-01 -2.50875831e-01
-1.17985785e+00 7.65167713e-01 -1.03953041e-01 -1.72936410e-01
-5.93793511e-01 -7.18685210e-01 -1.10795557e+00 -2.59614140e-01
6.26666322e-02 -6.86982810e-01 1.31971812e+00 -1.25685334e+00
-1.37451375e+00 1.01848710e+00 -5.52819371e-01 -3.22488576e-01
2.81504393e-02 -5.26912510e-01 -7.16825604e-01 -4.96473849e-01
4.02897328e-01 5.19276321e-01 5.47588289e-01 -1.30760658e+00
-6.14324689e-01 -5.03275216e-01 3.38315964e-01 1.95878625e-01
-1.15762644e-01 4.86192033e-02 -5.44090807e-01 -6.84988439e-01
-1.05266031e-02 -1.16750252e+00 -2.01996773e-01 -9.07657504e-01
-4.84694123e-01 -3.87747616e-01 8.30784857e-01 -4.50664371e-01
1.09704494e+00 -2.02751684e+00 -2.94759393e-01 1.86854035e-01
-2.12433740e-01 -3.59339677e-02 -3.20689641e-02 6.07883751e-01
-2.18855236e-02 9.50307027e-02 5.86370304e-02 -7.04233050e-01
5.08647084e-01 3.68999720e-01 -4.50673908e-01 7.67570585e-02
2.61382908e-01 1.23736358e+00 -5.05371094e-01 -2.94914007e-01
-1.02829197e-02 8.91147256e-01 -1.00560284e+00 -1.45843118e-01
-2.96459734e-01 7.94886425e-02 -5.66295981e-01 9.01384473e-01
6.61986709e-01 -1.66072488e-01 3.96417260e-01 -2.00463369e-01
1.53960511e-01 9.18635249e-01 -7.98246264e-01 1.42679393e+00
-8.44836473e-01 8.90259147e-01 -3.43310386e-01 -7.91605175e-01
9.07200575e-01 6.51105791e-02 4.04946178e-01 -8.89893293e-01
1.02144480e-01 1.84990734e-01 -2.25655828e-02 -2.17598304e-01
9.63111043e-01 -4.91490096e-01 -5.55619419e-01 4.69762266e-01
9.34648886e-02 -5.51171243e-01 3.55226636e-01 2.06528977e-01
3.43554199e-01 2.03007922e-01 3.18437964e-01 -2.58171141e-01
1.78813860e-01 1.67573377e-01 4.76940840e-01 4.55031991e-01
1.61792934e-01 8.81169617e-01 8.52408469e-01 -3.68097246e-01
-7.78568268e-01 -8.31700802e-01 -2.54476815e-01 1.28356993e+00
-1.63295344e-01 -9.14270282e-01 -4.60267484e-01 -9.88058925e-01
2.29243591e-01 1.28655589e+00 -9.56393659e-01 1.23049125e-01
-2.07223594e-01 -1.14370060e+00 8.05288255e-02 7.04522073e-01
6.47573024e-02 -8.81486773e-01 8.99786800e-02 2.00669691e-01
-8.71254951e-02 -1.22442961e+00 -2.30646968e-01 2.37789601e-01
-4.96362269e-01 -6.00208402e-01 -5.57625353e-01 -6.22434318e-01
3.18678319e-01 3.51002514e-02 1.87259734e+00 -7.53799140e-01
8.66676643e-02 5.96305430e-01 -5.01426220e-01 -6.80444837e-01
-2.74824560e-01 4.66425121e-01 -3.24246943e-01 2.14853846e-02
1.32061660e+00 -4.93670404e-01 -4.93728727e-01 2.62332171e-01
-7.97520041e-01 -3.01022470e-01 7.55878925e-01 5.13941109e-01
8.39322388e-01 -1.77709535e-01 5.01624286e-01 -1.57653356e+00
9.21558201e-01 -5.94971776e-01 -1.76292688e-01 -1.07356437e-01
-1.00518358e+00 -1.65374309e-01 3.26074094e-01 -3.05983633e-01
-1.03901422e+00 -3.55239093e-01 -4.89597261e-01 4.77927402e-02
-3.46638441e-01 8.23611736e-01 -2.78045565e-01 5.47310412e-01
3.15234035e-01 2.65707374e-01 -3.22221845e-01 -3.10368687e-01
4.89791811e-01 7.21568763e-01 -2.54753619e-01 -3.51120681e-01
3.90876234e-01 4.98822480e-01 -5.15081763e-01 -7.14713871e-01
-1.11553824e+00 -4.00251329e-01 -4.25096095e-01 1.26453817e-01
7.03236639e-01 -1.17635024e+00 -3.13716292e-01 2.40869388e-01
-7.79029667e-01 -3.33644338e-02 -5.22535861e-01 4.70145136e-01
-3.76260817e-01 -9.18527320e-02 -4.58136231e-01 -6.02314830e-01
-2.32677191e-01 -1.19338322e+00 1.41196632e+00 3.07368606e-01
-8.32398713e-01 -1.45397353e+00 2.46657237e-01 5.03057778e-01
6.60544693e-01 -7.53389820e-02 8.82912040e-01 -1.20255721e+00
-2.00386643e-02 -4.11577851e-01 1.62662640e-01 4.14332211e-01
4.17535067e-01 2.75241043e-02 -1.06073201e+00 -4.17684913e-02
1.00148745e-01 -1.73975170e-01 6.67572737e-01 5.49861848e-01
5.53932667e-01 -2.72679925e-01 -2.74650902e-01 3.98757666e-01
1.14802694e+00 2.48199686e-01 7.23103821e-01 5.01504183e-01
6.10802412e-01 6.14023685e-01 8.09683383e-01 3.78630161e-01
9.99183774e-01 6.23750985e-01 -5.99679612e-02 -1.67239219e-01
-1.22211091e-02 -3.67962688e-01 6.13366663e-01 9.62421060e-01
1.84300989e-01 -2.86000818e-01 -6.42562985e-01 8.10133815e-01
-1.37164748e+00 -7.54033327e-01 -2.19365820e-01 1.70728517e+00
9.24004436e-01 2.78627634e-01 4.40056741e-01 -2.62036413e-01
1.37082219e-01 4.36860412e-01 -3.03808123e-01 -9.15982366e-01
-5.18372893e-01 2.76984096e-01 2.15920314e-01 2.83336967e-01
-1.04256773e+00 1.01602280e+00 7.31154537e+00 8.73365879e-01
-1.27587712e+00 1.66736022e-01 9.93193507e-01 -1.23630092e-01
-1.30333304e+00 2.16806158e-01 -1.12520385e+00 4.43503141e-01
8.28151822e-01 -1.07772127e-01 1.95947215e-02 1.39865768e+00
-2.10647672e-01 -1.54597592e-02 -1.08718836e+00 7.88561583e-01
5.22220612e-01 -8.64765286e-01 2.21820876e-01 2.76184022e-01
1.08967614e+00 2.95666791e-02 5.61424434e-01 7.48122871e-01
5.66949189e-01 -1.00786769e+00 6.31057441e-01 -1.43799679e-02
4.69122380e-01 -9.69593763e-01 1.02267528e+00 -2.76812464e-01
-8.75232339e-01 4.27683681e-01 -1.77342325e-01 1.97982579e-01
4.24786806e-01 8.72130275e-01 -8.07655454e-01 5.97590387e-01
7.91811645e-01 9.44003880e-01 -7.83051968e-01 2.15958625e-01
-3.24059308e-01 9.56695378e-01 -6.89368844e-02 -1.14078745e-01
5.82729340e-01 -6.05297565e-01 4.09783959e-01 1.48017061e+00
3.05544257e-01 -3.67587239e-01 -7.48100877e-02 8.06735694e-01
2.23067418e-01 4.43801790e-01 -7.41160154e-01 -5.96847177e-01
-1.09904520e-02 1.32072318e+00 -6.98599696e-01 -3.59908521e-01
-8.98070097e-01 8.30702126e-01 -1.54570013e-01 4.29497719e-01
-8.04363430e-01 -1.08470611e-01 8.60523045e-01 3.26073110e-01
7.62375593e-01 -1.95889696e-01 -6.09478176e-01 -1.15177321e+00
-7.99416378e-02 -1.17592335e+00 4.07603115e-01 -7.58142948e-01
-1.58605134e+00 8.38790894e-01 -9.15087536e-02 -1.09522212e+00
-7.40914464e-01 -7.32998729e-01 -3.31846476e-01 9.37496364e-01
-1.46800435e+00 -1.59098399e+00 5.77197745e-02 4.41761732e-01
7.04134047e-01 -2.77159631e-01 8.66776168e-01 2.64013112e-01
-1.69273123e-01 5.30219793e-01 2.61818082e-03 -5.90053983e-02
9.09972429e-01 -1.43697786e+00 7.51036882e-01 5.78316391e-01
5.25791526e-01 1.06350493e+00 9.45474744e-01 -6.19810164e-01
-1.39141297e+00 -6.77945793e-01 1.28208065e+00 -7.59423256e-01
7.52466857e-01 -6.24273837e-01 -5.77643275e-01 1.02574193e+00
7.10174441e-01 -5.43182790e-01 1.26348352e+00 1.06474781e+00
-6.27452910e-01 -1.56597510e-01 -8.28189850e-01 6.07953072e-01
5.99905670e-01 -8.16543400e-01 -5.42656243e-01 1.02654502e-01
3.77227575e-01 -2.11896256e-01 -8.46523166e-01 2.80585796e-01
8.88553739e-01 -1.20324731e+00 5.98398864e-01 -5.78985751e-01
5.46962321e-01 -1.49394218e-02 -5.64266562e-01 -1.64833212e+00
-2.66273528e-01 -5.98865688e-01 3.71497571e-01 1.69559383e+00
9.66632843e-01 -9.09057617e-01 7.32569277e-01 6.48663819e-01
-1.25276312e-01 -9.02787685e-01 -4.97769505e-01 -4.75122809e-01
2.28002816e-01 -9.20122802e-01 8.15603554e-01 8.08287740e-01
-2.07120985e-01 6.24684632e-01 8.65726843e-02 -2.75526285e-01
-1.03753544e-01 7.05762208e-01 8.35952461e-01 -8.76426160e-01
-5.07618785e-01 -5.98862529e-01 -4.74919587e-01 -1.05579317e+00
3.53231579e-01 -6.46263540e-01 -2.07527414e-01 -1.65779150e+00
3.07306230e-01 -3.62192601e-01 -1.53907398e-02 4.37463999e-01
-2.41491869e-01 4.67241377e-01 -8.11418891e-02 -6.75880760e-02
-3.47330272e-01 5.63181520e-01 1.09392273e+00 -1.76543444e-01
-1.98903084e-01 4.01020855e-01 -1.55167902e+00 6.63075566e-01
6.30675018e-01 -4.07032281e-01 -6.03902280e-01 -2.18016028e-01
7.84373701e-01 -8.24047208e-01 -1.26604959e-01 -2.40117520e-01
-4.11542565e-01 -6.00397289e-02 3.80090386e-01 -7.72040784e-01
5.11303782e-01 -4.80883539e-01 -3.50227393e-03 -2.22277373e-01
-1.13278404e-01 4.04540509e-01 4.04958278e-01 3.07812005e-01
-6.30314887e-01 9.64247715e-03 1.84683487e-01 6.96388632e-02
-5.00154138e-01 5.58089698e-03 -5.81149876e-01 3.18595111e-01
6.64186478e-01 -3.19711357e-01 -2.78585762e-01 -7.43038058e-01
-5.62450707e-01 -3.18470001e-01 4.77890909e-01 7.12144554e-01
2.57040560e-01 -1.37598288e+00 -6.21635914e-01 4.25491095e-01
3.78024846e-01 -3.26890975e-01 2.41154835e-01 9.77931142e-01
-1.20440289e-01 8.50160420e-01 1.26637712e-01 -6.05296433e-01
-1.04498529e+00 3.72891426e-01 -4.77589704e-02 -4.07594770e-01
-8.32860321e-02 7.26052582e-01 4.66199160e-01 -6.04827404e-01
-5.54976940e-01 -4.12138492e-01 -2.52760977e-01 7.11200833e-01
5.80912083e-02 -6.86481893e-02 2.98289627e-01 -1.03481460e+00
-4.76976544e-01 5.02667725e-01 -4.59976882e-01 -4.87932563e-01
1.38490605e+00 -2.00109363e-01 1.19384095e-01 8.16542685e-01
1.22801340e+00 4.94322091e-01 -7.03122377e-01 -1.20786214e-02
-2.61862785e-01 -3.90451699e-01 -1.54331438e-02 -9.20109153e-01
-1.07700539e+00 7.13983178e-01 1.05781876e-01 6.75177276e-01
9.69178319e-01 3.83150190e-01 5.76571465e-01 -1.30042136e-01
2.02681988e-01 -1.14050508e+00 -1.78370878e-01 4.86375630e-01
6.79642022e-01 -1.23766434e+00 1.36290705e-02 -4.40325886e-01
-1.48724341e+00 6.29709721e-01 4.32123989e-01 5.22709750e-02
8.41257393e-01 2.95455247e-01 3.73389304e-01 -3.93424988e-01
-7.81357527e-01 -4.13134992e-01 5.91228485e-01 6.48085237e-01
1.07286763e+00 2.64950603e-01 -2.11446404e-01 1.08585370e+00
-9.21758115e-01 -4.73351032e-01 2.86453426e-01 8.61240566e-01
3.78727794e-01 -1.33752847e+00 -2.65984852e-02 5.82008004e-01
-7.90853620e-01 -4.23342705e-01 -6.54871643e-01 1.07230437e+00
-1.68827921e-02 1.07169127e+00 1.01888567e-01 -3.64829093e-01
5.03830075e-01 2.14658514e-01 2.50522465e-01 -7.12289155e-01
-7.17377782e-01 4.95644033e-01 4.99576747e-01 -5.23555279e-01
-4.49370444e-01 -1.27347529e+00 -5.71227491e-01 -1.84759870e-01
-3.17778051e-01 2.27791697e-01 8.72471511e-01 1.05363178e+00
6.47905886e-01 3.29413116e-01 5.23787200e-01 -4.12811607e-01
-2.23173052e-02 -1.19873428e+00 -7.60285079e-01 5.40195823e-01
-7.88832456e-02 -3.88937056e-01 -4.51497406e-01 7.61377066e-02] | [11.430286407470703, 6.716883182525635] |
40bdec32-a51b-4307-b947-7ec0ee2401a3 | simple-and-efficient-confidence-score-for | 2303.04604 | null | https://arxiv.org/abs/2303.04604v1 | https://arxiv.org/pdf/2303.04604v1.pdf | Simple and Efficient Confidence Score for Grading Whole Slide Images | Grading precancerous lesions on whole slide images is a challenging task: the continuous space of morphological phenotypes makes clear-cut decisions between different grades often difficult, leading to low inter- and intra-rater agreements. More and more Artificial Intelligence (AI) algorithms are developed to help pathologists perform and standardize their diagnosis. However, those models can render their prediction without consideration of the ambiguity of the classes and can fail without notice which prevent their wider acceptance in a clinical context. In this paper, we propose a new score to measure the confidence of AI models in grading tasks. Our confidence score is specifically adapted to ordinal output variables, is versatile and does not require extra training or additional inferences nor particular architecture changes. Comparison to other popular techniques such as Monte Carlo Dropout and deep ensembles shows that our method provides state-of-the art results, while being simpler, more versatile and less computationally intensive. The score is also easily interpretable and consistent with real life hesitations of pathologists. We show that the score is capable of accurately identifying mispredicted slides and that accuracy for high confidence decisions is significantly higher than for low-confidence decisions (gap in AUC of 17.1% on the test set). We believe that the proposed confidence score could be leveraged by pathologists directly in their workflow and assist them on difficult tasks such as grading precancerous lesions. | ['Thomas Walter', 'Cécile Badoual', 'Rutger Fick', 'Yaëlle Bellahsen-Harrar', 'Mélanie Lubrano'] | 2023-03-08 | null | null | null | null | ['whole-slide-images'] | ['computer-vision'] | [ 4.55387414e-01 2.76259392e-01 -1.93682373e-01 -3.66393238e-01
-1.04432511e+00 -7.25349247e-01 3.59596610e-01 6.61863804e-01
-5.77535808e-01 8.26808035e-01 -1.55480877e-01 -6.73063040e-01
-4.01704699e-01 -7.19091356e-01 -1.14980981e-01 -1.12261593e+00
2.10210189e-01 8.32941234e-01 3.08399826e-01 1.93640605e-01
2.86114782e-01 5.64949691e-01 -1.35630226e+00 5.37382185e-01
1.07615912e+00 8.34838867e-01 7.43320584e-02 9.57298458e-01
7.23856017e-02 5.26443064e-01 -6.46362185e-01 -7.16631174e-01
-1.21302396e-01 -3.09327096e-01 -6.84490919e-01 -2.29370341e-01
4.02242988e-01 8.48116279e-02 3.10221195e-01 1.00841105e+00
4.34775203e-01 -3.81420195e-01 1.13380075e+00 -7.71104038e-01
-3.37067574e-01 4.57127362e-01 -3.65084052e-01 1.12959035e-01
3.02278232e-02 4.58155036e-01 1.04053032e+00 -4.08361554e-01
5.92696488e-01 5.79154372e-01 8.41585934e-01 4.32483822e-01
-1.07716656e+00 -3.60536933e-01 -1.56787992e-01 3.40345502e-01
-1.32325959e+00 -7.13625699e-02 5.93024790e-02 -6.58951700e-01
6.36838973e-01 7.26364374e-01 6.21909797e-01 8.56012642e-01
5.69701850e-01 3.65603298e-01 1.41891754e+00 -3.45961273e-01
4.40457821e-01 3.75633299e-01 4.91128080e-02 8.06871355e-01
6.69988453e-01 -1.09405078e-01 -1.73737518e-02 -1.95757419e-01
3.42930287e-01 1.37679085e-01 -7.16885999e-02 -2.60025226e-02
-1.07455385e+00 7.70964146e-01 4.05365527e-01 6.09875262e-01
-1.58352509e-01 -9.13224295e-02 3.01764876e-01 1.13662839e-01
2.08397865e-01 6.29240751e-01 -1.42458618e-01 5.01463749e-03
-1.13406312e+00 -1.88577935e-01 6.55743480e-01 4.97135557e-02
1.53143406e-01 -4.47784543e-01 -2.52981454e-01 7.16495812e-01
1.51353955e-01 3.38175237e-01 7.77271152e-01 -7.51273513e-01
-3.49785715e-01 9.00497198e-01 -9.61180329e-02 -7.85506964e-01
-7.60446668e-01 -6.42298460e-01 -1.07123387e+00 5.23535788e-01
6.95454776e-01 4.44957837e-02 -1.18505144e+00 1.27097273e+00
1.01888984e-01 -6.52232394e-02 -1.08173467e-01 7.38739431e-01
6.04276955e-01 1.19857639e-01 2.22587436e-01 -1.23018451e-01
1.72041678e+00 -4.24459100e-01 -4.44217324e-01 -7.26411194e-02
8.76266897e-01 -5.93767881e-01 8.78605068e-01 6.62978768e-01
-8.33793998e-01 -8.69239643e-02 -1.10287809e+00 8.93651545e-02
-4.53773558e-01 2.40868375e-01 8.02011967e-01 1.03121150e+00
-1.17719567e+00 5.89051127e-01 -1.07908094e+00 -4.70867306e-01
6.24697387e-01 6.27794981e-01 -4.71312553e-01 1.35227174e-01
-8.18059862e-01 1.16317129e+00 3.30482095e-01 5.05236350e-02
-6.12521112e-01 -8.26205552e-01 -4.31826502e-01 2.77362652e-02
1.75102055e-01 -7.89121509e-01 1.06726253e+00 -6.71950758e-01
-1.46294022e+00 1.18502116e+00 -6.98259920e-02 -4.57017064e-01
7.60425031e-01 5.47149658e-01 -3.37357491e-01 1.20583937e-01
-1.90513059e-01 5.97586215e-01 3.36543441e-01 -7.85315156e-01
-8.16877782e-01 -5.24147093e-01 -1.95075437e-01 -8.85049626e-02
-2.27242306e-01 -3.32121521e-01 -6.64927950e-03 -3.61283153e-01
-1.36541739e-01 -9.99866128e-01 -4.85610962e-01 2.37416074e-01
-2.25028187e-01 -1.55139700e-01 2.89361358e-01 -4.71494526e-01
1.15950000e+00 -1.84528327e+00 -2.28414148e-01 4.68571126e-01
3.50458413e-01 5.29311240e-01 2.61827052e-01 4.57729436e-02
2.12396488e-01 4.38101828e-01 -2.74339825e-01 1.65661156e-01
-1.03527337e-01 1.22691497e-01 2.19774589e-01 5.44821560e-01
3.58238935e-01 8.31919432e-01 -9.93090391e-01 -6.70382440e-01
1.97113797e-01 4.05200660e-01 -4.11969602e-01 3.14453021e-02
1.16709368e-02 2.63833702e-01 -2.48840064e-01 7.22973764e-01
4.09782559e-01 -7.01335967e-01 4.35078412e-01 -1.58679694e-01
1.91210508e-01 1.51863554e-02 -9.18334424e-01 1.10599685e+00
-4.62091982e-01 6.38397336e-01 -2.87835598e-01 -6.61284029e-01
8.74251306e-01 2.80036002e-01 -8.86213034e-02 -3.03059906e-01
2.43489996e-01 5.42255998e-01 6.11901045e-01 -5.53095162e-01
8.03051293e-02 -4.56366181e-01 2.18616202e-01 4.16284859e-01
2.03863680e-02 -3.56622756e-01 3.11300457e-01 -7.04017803e-02
1.33900785e+00 -5.01086950e-01 8.18258822e-01 -3.25162739e-01
7.45585740e-01 -7.70170987e-02 3.60208511e-01 7.83787370e-01
-2.88381308e-01 6.23521388e-01 6.53818667e-01 -5.21784306e-01
-8.66058111e-01 -1.15517497e+00 -5.60666442e-01 6.45203531e-01
-1.82445079e-01 3.59507874e-02 -5.26078820e-01 -9.52607512e-01
5.85250705e-02 5.27866364e-01 -1.09116447e+00 3.37851420e-02
-1.12946168e-01 -1.19696021e+00 5.69468319e-01 5.05964518e-01
-5.38484938e-03 -7.37280667e-01 -5.93201160e-01 8.79516378e-02
5.61937355e-02 -6.54341221e-01 8.62114131e-02 2.97338277e-01
-8.20954382e-01 -1.31635344e+00 -8.04467738e-01 -5.17259538e-01
1.02857327e+00 -2.25167125e-01 9.95497763e-01 2.51705796e-01
-5.16217589e-01 -2.11413205e-01 -2.81010300e-01 -6.29525602e-01
-8.68165493e-01 1.88659072e-01 -3.10035944e-01 -2.43443158e-02
5.31762064e-01 -3.10407102e-01 -9.46519375e-01 4.95139599e-01
-9.53435183e-01 -3.53223607e-02 1.17663312e+00 1.12209582e+00
7.06388950e-01 -2.29285896e-01 5.63212991e-01 -1.30526805e+00
5.16138911e-01 -3.37820679e-01 -3.06263536e-01 3.32538843e-01
-8.65798056e-01 1.34699091e-01 7.72405624e-01 -2.89951414e-01
-8.12689066e-01 -1.94021866e-01 -3.10742259e-01 2.07204640e-01
-3.56840789e-01 4.16004002e-01 5.27037382e-01 -2.56978035e-01
9.81815934e-01 -2.10132971e-01 2.82083839e-01 -8.98532867e-02
-2.68937677e-01 7.33621776e-01 4.32282209e-01 -1.47628160e-02
4.60287213e-01 7.18880236e-01 4.19211000e-01 -3.41007262e-01
-7.67181158e-01 -4.70973343e-01 -3.60637903e-01 -1.97324246e-01
5.68935931e-01 -4.11080360e-01 -8.28590810e-01 2.42296070e-01
-5.48638940e-01 -2.73369938e-01 -1.27531275e-01 4.36955780e-01
-3.05875868e-01 3.60905021e-01 -6.67441130e-01 -6.77837193e-01
-5.03754079e-01 -1.26860511e+00 1.01004982e+00 5.18290401e-01
-7.51035750e-01 -1.33548498e+00 1.81304812e-01 4.77509946e-01
5.58057189e-01 4.49644834e-01 1.03296244e+00 -8.76226246e-01
-1.39889121e-01 -6.62408948e-01 -1.71110928e-01 1.06146380e-01
1.80952027e-01 5.89898884e-01 -1.22873640e+00 -2.18817100e-01
-3.30477625e-01 -2.08430022e-01 8.86758029e-01 3.81577790e-01
1.25861096e+00 -9.40707177e-02 -6.19293332e-01 3.57807308e-01
1.45764041e+00 1.69088289e-01 5.82260430e-01 2.00756907e-01
1.06617205e-01 6.96562707e-01 4.82878327e-01 9.43056643e-02
1.69762224e-01 4.55937922e-01 3.61707985e-01 -2.49217063e-01
-4.88071777e-02 7.65436217e-02 2.74761822e-02 5.70926428e-01
-1.99743152e-01 -4.06026065e-01 -1.12295270e+00 5.32810569e-01
-1.48060787e+00 -7.45181859e-01 -2.02013433e-01 2.19214249e+00
1.00031865e+00 4.04201090e-01 -1.57048747e-01 3.94894809e-01
5.72824895e-01 -2.80964404e-01 -4.89376962e-01 -7.57457256e-01
-8.78119394e-02 3.79028767e-01 4.34028000e-01 5.58224916e-01
-7.12471068e-01 3.83124650e-01 6.57352686e+00 1.12961674e+00
-1.35449374e+00 1.16800480e-02 1.31281030e+00 -7.02981139e-03
-1.38537079e-01 -2.76658058e-01 -5.98087192e-01 5.18716812e-01
9.98260915e-01 -2.75966376e-01 -2.54736304e-01 4.86972958e-01
1.23703629e-01 -5.33996046e-01 -1.27530253e+00 5.41020811e-01
-7.22616985e-02 -1.46112418e+00 -4.04226333e-02 1.28022298e-01
7.24339485e-01 -1.18444920e-01 2.50870198e-01 -3.83111439e-03
4.74086314e-01 -1.38365269e+00 4.99277115e-02 6.64321959e-01
1.02646410e+00 -6.26460433e-01 1.54205680e+00 1.54342711e-01
-4.89408940e-01 6.86143339e-02 -2.60677069e-01 2.43385490e-02
-2.08953202e-01 8.65149200e-01 -1.69993210e+00 3.48838031e-01
2.98240185e-01 2.09898815e-01 -9.45832729e-01 1.22698367e+00
-1.69822320e-01 7.15756059e-01 -3.64664674e-01 -5.44965506e-01
2.32646480e-01 2.36405820e-01 2.19807744e-01 1.40815532e+00
4.37789708e-01 -7.13001415e-02 -4.07644272e-01 5.88860035e-01
3.11251968e-01 2.42627516e-01 -1.71771437e-01 7.00485408e-02
3.01394105e-01 1.64435482e+00 -1.18108106e+00 -3.11715037e-01
3.42893004e-02 5.05057573e-01 1.87486142e-01 -1.13282099e-01
-6.02096438e-01 -3.30174893e-01 2.94043750e-01 9.15329605e-02
3.93444858e-02 6.31037593e-01 -6.89296722e-01 -7.41821170e-01
-2.28369936e-01 -7.38334179e-01 7.57605255e-01 -5.14371634e-01
-1.35397744e+00 7.02665150e-01 -4.49127525e-01 -1.34067488e+00
-4.96649891e-01 -1.11919594e+00 -7.29079843e-01 5.15885055e-01
-1.46264470e+00 -7.46522307e-01 -4.11554366e-01 1.60207972e-02
2.97839753e-02 1.44419000e-01 1.07384443e+00 -1.35354176e-01
-3.30770046e-01 9.83520269e-01 2.47282416e-01 1.70662329e-02
9.47015882e-01 -1.52128232e+00 -2.90974498e-01 5.15210569e-01
-1.95492133e-01 3.65808636e-01 7.88893163e-01 -2.17764288e-01
-8.11834216e-01 -8.07947040e-01 7.79201627e-01 -6.45140827e-01
7.89707184e-01 5.62529527e-02 -8.96943986e-01 7.59446323e-02
-1.22898668e-01 -1.70656908e-02 1.40898275e+00 3.78716365e-02
-2.54111528e-01 -2.17278928e-01 -1.40637338e+00 6.36662066e-01
4.58995312e-01 -1.06431261e-01 -2.89517969e-01 3.97579283e-01
-7.82724917e-02 -3.34192157e-01 -1.08055055e+00 6.41038001e-01
8.29812586e-01 -1.16164851e+00 5.73654830e-01 -3.76378566e-01
3.91875833e-01 -3.09951723e-01 3.13488513e-01 -1.27391815e+00
-4.69061673e-01 -8.89340565e-02 2.78395027e-01 7.43576527e-01
9.32050765e-01 -7.39276648e-01 1.04048526e+00 5.77557087e-01
3.70764849e-03 -1.19623423e+00 -1.03030467e+00 -3.33077908e-01
2.45276973e-01 -1.93132177e-01 3.12805831e-01 8.46066296e-01
3.32959652e-01 -2.28902370e-01 2.43189871e-01 1.95143491e-01
4.11555558e-01 7.14379475e-02 3.59593481e-01 -1.42822266e+00
-4.88699287e-01 -9.58154976e-01 -9.84084785e-01 -3.14408280e-02
-3.29502672e-01 -1.02024496e+00 -1.99644506e-01 -1.36800277e+00
4.76452768e-01 -5.21561801e-01 -5.58355212e-01 6.09919786e-01
-4.77840871e-01 7.61152565e-01 -1.83324888e-01 8.38123783e-02
-4.65682268e-01 -3.03754717e-01 1.17244828e+00 -3.00496787e-01
2.17934355e-01 5.87449521e-02 -9.46089387e-01 8.45215797e-01
7.69940197e-01 -5.18197119e-01 -5.21754753e-03 1.00183949e-01
4.60151404e-01 -6.45727888e-02 2.34859332e-01 -1.08180177e+00
2.72893220e-01 -2.72671163e-01 6.31766558e-01 -2.34118193e-01
-4.71440107e-02 -5.79800785e-01 4.29540604e-01 9.69138741e-01
-2.62411445e-01 -2.52063811e-01 3.55587341e-02 4.31292057e-01
-2.10163042e-01 -5.43810606e-01 9.00381744e-01 -7.15215579e-02
-3.06457579e-01 1.60336066e-02 -4.39014435e-01 -2.77371168e-01
1.27151513e+00 -5.68812728e-01 -6.10115707e-01 -1.67324692e-01
-8.13416481e-01 2.19559729e-01 7.00499058e-01 -1.17414974e-01
1.84214696e-01 -8.48384321e-01 -1.02184975e+00 -7.66823441e-02
3.52777630e-01 -1.05903614e-02 5.40844202e-01 1.22862923e+00
-9.71026242e-01 5.93461931e-01 -1.67824820e-01 -9.07553077e-01
-1.40443349e+00 1.79807201e-01 4.98002082e-01 -1.02616787e+00
-5.13990643e-03 9.55037236e-01 2.01914728e-01 -2.01400593e-01
-1.41129643e-01 -4.23228741e-01 -3.18186074e-01 2.08751157e-01
6.54747605e-01 3.38166595e-01 5.37040830e-01 -6.41031489e-02
-3.57573003e-01 4.61152196e-01 -6.08432651e-01 1.15928024e-01
1.10906029e+00 3.64218771e-01 -1.72551498e-02 4.98113543e-01
7.95246243e-01 7.75537193e-02 -9.54221606e-01 2.53815621e-01
-8.22550878e-02 -2.59698212e-01 -4.20949087e-02 -1.42432606e+00
-8.20695817e-01 9.32047307e-01 7.25127399e-01 2.89505184e-01
1.07766163e+00 -9.46770906e-02 1.15282990e-01 2.45578527e-01
1.98367506e-01 -8.85711908e-01 -1.61620155e-01 4.26634923e-02
4.59963977e-01 -1.31628311e+00 1.90282598e-01 -3.58558208e-01
-5.50270796e-01 1.27046096e+00 4.20082897e-01 4.69986647e-02
2.17704535e-01 4.91367847e-01 3.97168756e-01 1.86673203e-03
-1.11308348e+00 -2.53044964e-05 3.26304197e-01 7.21069515e-01
7.18298733e-01 3.97001535e-01 -6.90035582e-01 5.44851661e-01
-3.21991593e-01 1.35991529e-01 7.30248809e-01 6.11696601e-01
-3.93240422e-01 -1.12855005e+00 -3.22970510e-01 1.04191053e+00
-7.94813454e-01 1.22099765e-01 -3.88214529e-01 7.15313196e-01
9.22614858e-02 7.38431871e-01 1.70785606e-01 -1.80534169e-01
-1.39470816e-01 7.99810216e-02 4.02253121e-01 -4.69840616e-01
-7.99073339e-01 -2.50857115e-01 1.08221201e-02 -1.12415053e-01
-3.43860447e-01 -6.14919186e-01 -1.05984163e+00 -2.85871714e-01
-5.27465582e-01 2.94526666e-01 5.50457656e-01 8.01617980e-01
4.02141780e-01 4.44803089e-01 3.96169364e-01 -4.20888007e-01
-5.01046956e-01 -9.78535116e-01 -3.86779428e-01 2.92761296e-01
1.65897042e-01 -5.49763262e-01 -6.13014519e-01 -5.46997637e-02] | [15.120173454284668, -2.923309326171875] |
739f7fe1-16bc-4a07-a2ec-30f78c78a849 | minimizing-the-effect-of-noise-and-limited | 2208.10390 | null | https://arxiv.org/abs/2208.10390v1 | https://arxiv.org/pdf/2208.10390v1.pdf | Minimizing the Effect of Noise and Limited Dataset Size in Image Classification Using Depth Estimation as an Auxiliary Task with Deep Multitask Learning | Generalizability is the ultimate goal of Machine Learning (ML) image classifiers, for which noise and limited dataset size are among the major concerns. We tackle these challenges through utilizing the framework of deep Multitask Learning (dMTL) and incorporating image depth estimation as an auxiliary task. On a customized and depth-augmented derivation of the MNIST dataset, we show a) multitask loss functions are the most effective approach of implementing dMTL, b) limited dataset size primarily contributes to classification inaccuracy, and c) depth estimation is mostly impacted by noise. In order to further validate the results, we manually labeled the NYU Depth V2 dataset for scene classification tasks. As a contribution to the field, we have made the data in python native format publicly available as an open-source dataset and provided the scene labels. Our experiments on MNIST and NYU-Depth-V2 show dMTL improves generalizability of the classifiers when the dataset is noisy and the number of examples is limited. | ['Farzad Khalvati', 'Partoo Vafaeikia', 'Khashayar Namdar'] | 2022-08-22 | null | null | null | null | ['scene-classification'] | ['computer-vision'] | [ 1.57056689e-01 -7.29149356e-02 -1.18082501e-01 -5.27186215e-01
-1.16557276e+00 -5.51287174e-01 4.19556528e-01 -2.44792178e-02
-7.12462008e-01 6.55981779e-01 1.34247420e-02 -2.33676076e-01
1.51364073e-01 -6.28640652e-01 -9.95765209e-01 -6.17463052e-01
1.84189379e-01 2.89328098e-01 3.52268487e-01 2.42353976e-01
2.11360931e-01 1.83097079e-01 -1.45306349e+00 6.36483788e-01
7.09206104e-01 1.27206635e+00 4.70701605e-01 6.02016509e-01
-2.98562702e-02 9.92069304e-01 -4.61913884e-01 -1.59260452e-01
5.88842034e-01 7.50685781e-02 -9.55703974e-01 8.55228230e-02
1.05847538e+00 -7.24073648e-01 -2.28227556e-01 8.26560497e-01
7.67665803e-01 -1.46297095e-02 6.19672179e-01 -1.31731045e+00
-3.50881130e-01 3.25789034e-01 -4.55167234e-01 2.11388379e-01
-1.94362760e-01 3.87615204e-01 7.46070087e-01 -7.69000411e-01
5.27900636e-01 1.24969602e+00 7.88300693e-01 5.65038919e-01
-1.18127429e+00 -6.56993866e-01 3.82416755e-01 1.34870455e-01
-1.28086150e+00 -4.88320231e-01 5.12452364e-01 -6.94436133e-01
1.17528939e+00 -1.11312300e-01 4.26498353e-02 1.44430912e+00
9.57706645e-02 8.48817587e-01 1.35755253e+00 -3.08331102e-01
3.40655565e-01 1.63878262e-01 1.40566200e-01 7.32052326e-01
1.11655042e-01 9.19925272e-02 -3.46963286e-01 1.66234225e-02
7.56496727e-01 -1.75568581e-01 1.80871375e-02 -3.19935650e-01
-1.17765915e+00 8.71936977e-01 5.49612522e-01 -1.63818911e-01
-2.09957883e-02 3.73019338e-01 5.84783137e-01 1.86239287e-01
6.94785893e-01 1.26340032e-01 -6.53570414e-01 -1.77125365e-01
-6.72394097e-01 2.57093132e-01 3.85783166e-01 9.83509958e-01
8.93161297e-01 -1.31182969e-01 -1.39668593e-02 9.07023311e-01
4.18312661e-02 5.07748723e-01 4.70579654e-01 -1.03826272e+00
7.28207350e-01 4.41957653e-01 -2.10739091e-01 -5.13563097e-01
-6.13297582e-01 -2.85230160e-01 -5.37838340e-01 4.72291648e-01
7.64756024e-01 -2.47800648e-01 -9.66919482e-01 1.74670899e+00
1.03037551e-01 -3.14192325e-02 1.30893275e-01 8.69886100e-01
1.30139375e+00 2.74741292e-01 3.14878285e-01 2.89083481e-01
1.23569512e+00 -9.50477004e-01 -4.14286703e-01 -6.08404040e-01
1.04993320e+00 -6.78643644e-01 1.44234955e+00 5.30618608e-01
-7.57692695e-01 -6.72472477e-01 -1.08143687e+00 -4.36092019e-01
-3.70562732e-01 4.64928269e-01 8.10711801e-01 5.38278520e-01
-9.19189632e-01 3.63924265e-01 -8.53446782e-01 -4.22124565e-01
8.25314641e-01 7.73683637e-02 -5.12433946e-01 -4.26764369e-01
-7.49281764e-01 7.81087697e-01 3.94818187e-01 -9.83822271e-02
-1.18038797e+00 -6.89883590e-01 -9.88335133e-01 -2.35659689e-01
2.68819839e-01 -4.98994857e-01 1.21441031e+00 -7.54708409e-01
-1.09331357e+00 1.09354115e+00 1.09797820e-01 -2.61293828e-01
9.37265754e-01 -1.49278164e-01 -5.10116704e-02 1.39575973e-01
2.95957714e-01 1.08065343e+00 6.59369290e-01 -1.16402507e+00
-7.74599969e-01 -6.18304908e-01 2.72923768e-01 1.84240952e-01
-3.20465237e-01 -3.72657239e-01 -3.14497352e-01 -4.76408154e-01
1.45584941e-01 -9.01765645e-01 -9.83962268e-02 1.24584787e-01
-3.86728466e-01 -1.15501314e-01 9.47508693e-01 -4.10498112e-01
6.97207212e-01 -2.32146883e+00 -1.27008989e-01 -2.25358501e-01
1.34476408e-01 -9.19883251e-02 -2.93488890e-01 -5.59181869e-02
1.45428494e-01 1.62345394e-01 -6.86530843e-02 -6.04982555e-01
-4.12745833e-01 2.80785739e-01 -2.89956424e-02 4.82568145e-01
5.64627796e-02 8.07485521e-01 -5.75145960e-01 -4.38433200e-01
2.77477622e-01 1.40312955e-01 -4.68215853e-01 8.86818673e-03
-3.85298699e-01 6.39552653e-01 -2.60062397e-01 5.69440126e-01
7.96143889e-01 -3.67458850e-01 -2.48622894e-01 -5.87456584e-01
-1.08300075e-01 9.71792638e-02 -1.03110504e+00 1.93709111e+00
-7.44545162e-01 6.52738392e-01 -9.77415815e-02 -7.75538325e-01
7.27066219e-01 6.87485784e-02 2.44958267e-01 -8.20815325e-01
1.00779951e-01 1.13672726e-01 -1.52888477e-01 -7.99825668e-01
2.44192734e-01 8.48416910e-02 1.88255589e-02 3.05899978e-01
1.14043266e-01 -3.79999012e-01 2.23224256e-02 1.03080772e-01
1.20817697e+00 3.11224103e-01 1.53706297e-01 -2.87225753e-01
3.18480060e-02 2.85161555e-01 5.43641388e-01 8.12873304e-01
-4.82406169e-01 8.27546477e-01 5.30082107e-01 -5.31734228e-01
-1.09638643e+00 -8.31563056e-01 -5.45943737e-01 1.07324052e+00
-1.12362124e-01 -1.14202522e-01 -6.01386547e-01 -9.01616573e-01
1.70481026e-01 5.91983259e-01 -8.06652665e-01 2.33824342e-01
-2.33173102e-01 -7.55405366e-01 6.59527898e-01 7.43722200e-01
8.29646885e-01 -7.01447368e-01 -7.14475989e-01 -7.39424527e-02
-3.79113078e-01 -1.71918488e+00 -1.86932653e-01 4.99863893e-01
-7.81347990e-01 -1.16066313e+00 -5.83494544e-01 -5.27719975e-01
4.90499675e-01 3.07041943e-01 8.68722975e-01 -2.52038896e-01
-3.49943608e-01 2.79719234e-01 -2.35257968e-01 -6.26949728e-01
-1.27263606e-01 2.97739536e-01 -2.28159606e-01 -2.54029512e-01
2.20820621e-01 -5.12840509e-01 -5.86087465e-01 2.44217321e-01
-8.95125508e-01 2.00423285e-01 4.86504048e-01 6.67223454e-01
5.60417473e-01 -1.94734216e-01 5.23721993e-01 -1.25708807e+00
6.17721826e-02 -3.93489599e-01 -7.89674282e-01 -5.00424094e-02
-5.21868885e-01 -8.51773098e-02 3.48274946e-01 -4.20817733e-01
-1.03998792e+00 2.96243042e-01 -2.95178980e-01 -4.72198755e-01
-3.98898154e-01 3.31146389e-01 -1.98463902e-01 -2.64802098e-01
9.53283548e-01 -1.93264395e-01 -2.63596196e-02 -4.68728721e-01
1.84847444e-01 6.08365417e-01 4.14879918e-01 -6.10583484e-01
1.46282002e-01 7.75323808e-01 1.67239472e-01 -8.96641493e-01
-1.08900666e+00 -3.22833598e-01 -7.68512547e-01 -1.15521103e-01
9.34445620e-01 -1.40361059e+00 -5.87874770e-01 6.74739540e-01
-9.71737087e-01 -9.76196587e-01 1.27888069e-01 4.25305754e-01
-5.22410154e-01 3.36996526e-01 -5.45529783e-01 -6.47671819e-01
-1.22025125e-01 -1.42300832e+00 1.23548567e+00 -1.61348537e-01
1.72279198e-02 -1.04399180e+00 -2.78261065e-01 7.49093831e-01
3.12070310e-01 4.42293823e-01 1.07517862e+00 -6.45061553e-01
-5.93586683e-01 -6.96422085e-02 -5.50443828e-01 4.49821770e-01
6.87616691e-02 -1.05730094e-01 -1.50442028e+00 -3.21947396e-01
-1.59048721e-01 -8.98229778e-01 1.12565076e+00 7.64272332e-01
1.48409867e+00 6.32888004e-02 -2.00482428e-01 9.34756756e-01
1.69142056e+00 -1.74533859e-01 5.99027634e-01 6.29673421e-01
9.75777805e-01 7.14300513e-01 4.89405602e-01 4.96851891e-01
7.47897804e-01 5.83563089e-01 5.50122857e-01 -1.76788628e-01
-3.38265032e-01 -2.68716644e-02 8.77111405e-02 1.37653023e-01
3.21298987e-01 -3.81231382e-02 -1.04595423e+00 5.09421766e-01
-1.84785426e+00 -5.95052779e-01 -3.29393506e-01 2.07324743e+00
5.16518056e-01 2.88542032e-01 1.31368458e-01 -4.51376624e-02
3.31490964e-01 -7.93476403e-02 -6.80384457e-01 -2.01698858e-02
-3.70680243e-01 -2.36726686e-01 8.45762193e-01 4.84189838e-01
-1.37647426e+00 1.07278883e+00 6.19839621e+00 7.70547688e-01
-1.38055253e+00 3.76974612e-01 8.54943216e-01 -4.41726834e-01
3.98396030e-02 -1.08155690e-01 -1.03252149e+00 4.87064660e-01
7.15655565e-01 2.68826127e-01 5.76319136e-02 1.08585262e+00
2.59916484e-01 -4.32005584e-01 -1.33104777e+00 1.28318036e+00
2.40578931e-02 -1.15182209e+00 -1.67055324e-01 6.15072474e-02
6.76277339e-01 5.03374696e-01 1.75361335e-01 4.19874012e-01
1.89682126e-01 -1.09783697e+00 8.86654675e-01 1.36902519e-02
9.63786364e-01 -5.26734531e-01 7.58888245e-01 3.71746153e-01
-8.73281956e-01 -2.50566125e-01 -5.91993809e-01 -2.25564003e-01
-8.10069367e-02 8.10639143e-01 -5.89402258e-01 2.70089984e-01
1.19677043e+00 6.79107428e-01 -9.49139118e-01 7.98697829e-01
-1.74693570e-01 4.54107374e-01 -3.92866135e-01 3.47048193e-01
2.79760063e-01 1.15326919e-01 1.28637612e-01 1.09550929e+00
1.70648079e-02 -2.16584504e-01 4.19927418e-01 7.60007262e-01
-1.46753728e-01 -4.04921658e-02 -7.93090582e-01 3.91586125e-01
2.15309560e-01 1.48841143e+00 -7.00827777e-01 -1.77625775e-01
-6.08893991e-01 9.08156753e-01 6.02313280e-01 3.76355886e-01
-6.21695340e-01 -2.87037715e-02 6.74412847e-01 -1.31228641e-02
1.88247666e-01 -9.47660729e-02 -8.26494455e-01 -1.13193989e+00
5.29376939e-02 -7.52550304e-01 4.58188325e-01 -7.72140682e-01
-1.31408644e+00 6.53885961e-01 5.65137193e-02 -1.09199190e+00
-5.56357801e-02 -7.70007908e-01 -3.52273434e-01 7.76886165e-01
-1.69599271e+00 -1.42850399e+00 -8.71716321e-01 5.54800391e-01
5.94051600e-01 -1.62838936e-01 6.96245134e-01 4.65084225e-01
-7.45555103e-01 7.26593375e-01 -1.47529364e-01 1.11457705e-02
9.08970475e-01 -1.16350496e+00 4.12121266e-01 7.23375499e-01
-1.53015018e-01 6.69455603e-02 5.21920204e-01 -5.62876463e-01
-1.00792229e+00 -1.42718780e+00 3.79636526e-01 -6.95318282e-01
4.64413911e-01 -6.86549008e-01 -7.87306666e-01 8.87846291e-01
-3.26343685e-01 2.65432745e-01 5.71768522e-01 -1.50919305e-02
-3.61152500e-01 -7.35153109e-02 -1.27873051e+00 3.52958530e-01
1.09782135e+00 -6.03847623e-01 -1.01016864e-01 6.39576912e-01
6.05201185e-01 -6.57137156e-01 -7.11977303e-01 5.55721581e-01
4.91031140e-01 -1.02947903e+00 7.44510591e-01 -1.81869239e-01
5.58524370e-01 -2.59664189e-03 -4.71583396e-01 -1.26674771e+00
2.29115225e-02 1.71233714e-02 2.63195127e-01 1.15979540e+00
5.64312220e-01 -6.36746585e-01 9.61118639e-01 5.45062780e-01
-1.54487133e-01 -7.41573393e-01 -9.55244422e-01 -7.63638020e-01
3.89665365e-01 -8.28954995e-01 2.11665615e-01 7.59667218e-01
-4.85848814e-01 2.82991290e-01 -2.71566123e-01 1.21493682e-01
7.56859720e-01 -1.32619023e-01 1.09852064e+00 -1.08942735e+00
-2.50399888e-01 -1.90025374e-01 -3.31191629e-01 -1.07255828e+00
2.87162870e-01 -9.56229091e-01 -1.28897908e-03 -1.55260134e+00
2.34575823e-01 -7.00735927e-01 2.17857108e-01 7.67556429e-01
-6.36268035e-02 4.22468036e-01 2.01662444e-02 2.99460292e-01
-4.33073580e-01 3.72926831e-01 1.07252431e+00 -9.90515128e-02
1.88896917e-02 4.83059734e-02 -4.62006658e-01 7.06437528e-01
5.99222839e-01 -3.09608698e-01 -5.25746524e-01 -8.52824092e-01
-4.40954529e-02 -1.49127245e-01 4.58576858e-01 -1.04468083e+00
1.55409873e-01 8.44083801e-02 6.43834472e-01 -6.09227955e-01
5.12598753e-01 -8.07459712e-01 -3.31443489e-01 2.53745615e-01
-4.34003443e-01 4.16416116e-02 6.49041891e-01 3.30369681e-01
-5.34270369e-02 -2.71829125e-02 1.07059050e+00 -1.66997612e-01
-1.05094564e+00 4.26286697e-01 -1.42638937e-01 1.76088452e-01
1.01655102e+00 -1.87260419e-01 -7.08588660e-01 -2.75755793e-01
-4.54265088e-01 2.06009775e-01 5.24557173e-01 4.04350013e-01
5.06040812e-01 -9.63545561e-01 -7.99581230e-01 1.51346907e-01
5.90013325e-01 2.20826536e-01 3.74043882e-01 7.80809879e-01
-6.50600791e-01 4.51644123e-01 -3.27757001e-01 -9.69021440e-01
-1.21916676e+00 3.12158853e-01 3.54936481e-01 1.47209823e-01
-6.73479855e-01 1.03695083e+00 4.76723164e-01 -5.59021413e-01
4.55368429e-01 -5.72519243e-01 -1.93905775e-02 -8.30500424e-02
4.53797430e-01 3.79929036e-01 4.58921552e-01 -4.27947342e-01
-3.22060466e-01 4.35794592e-01 -1.97257191e-01 -6.96833581e-02
1.31579554e+00 -2.71127254e-01 1.54905379e-01 5.52709579e-01
1.39390600e+00 -3.76447260e-01 -1.62971437e+00 -2.65907377e-01
-1.41125336e-01 -4.95776296e-01 3.46705019e-01 -8.15916538e-01
-1.01552761e+00 1.04004061e+00 8.93527746e-01 -2.43212283e-01
9.66788948e-01 1.26804458e-02 5.90262413e-01 4.35464263e-01
3.63758922e-01 -9.70600426e-01 1.71610832e-01 6.28256083e-01
6.70348227e-01 -1.85915446e+00 -4.58620228e-02 -4.58269358e-01
-8.46203029e-01 1.03192699e+00 1.05071247e+00 6.63966238e-02
5.55392206e-01 4.26190794e-01 4.98588473e-01 -1.41507477e-01
-7.66511977e-01 -2.23277643e-01 1.15256622e-01 6.83154762e-01
4.88882750e-01 -2.82487303e-01 2.21467540e-01 3.90149921e-01
-2.60557085e-01 -5.95520772e-02 5.21187842e-01 8.15417886e-01
-1.21147111e-01 -8.45433056e-01 -1.77864477e-01 5.54717481e-01
-5.39708912e-01 -1.78202406e-01 -1.20617829e-01 7.43043065e-01
1.57968715e-01 1.07673693e+00 8.33891109e-02 -2.59451300e-01
3.27968776e-01 -5.01785576e-02 6.76960170e-01 -7.60812521e-01
-4.21234250e-01 -1.32302582e-01 7.23478943e-02 -6.82828128e-01
-7.76337460e-02 -4.79239762e-01 -1.04059124e+00 -2.68354237e-01
-1.30982220e-01 -5.08944869e-01 1.04435277e+00 1.14842236e+00
3.73668522e-01 6.63180828e-01 3.62788588e-01 -8.84382486e-01
-5.96552372e-01 -1.11163652e+00 -5.02280712e-01 5.81666410e-01
4.74499822e-01 -6.78618491e-01 -5.23571789e-01 -5.24289310e-02] | [9.530471801757812, 1.418951392173767] |
89801b7f-d173-4dc4-b838-60c9affd1706 | combining-vision-and-tactile-sensation-for | 2304.11193 | null | https://arxiv.org/abs/2304.11193v1 | https://arxiv.org/pdf/2304.11193v1.pdf | Combining Vision and Tactile Sensation for Video Prediction | In this paper, we explore the impact of adding tactile sensation to video prediction models for physical robot interactions. Predicting the impact of robotic actions on the environment is a fundamental challenge in robotics. Current methods leverage visual and robot action data to generate video predictions over a given time period, which can then be used to adjust robot actions. However, humans rely on both visual and tactile feedback to develop and maintain a mental model of their physical surroundings. In this paper, we investigate the impact of integrating tactile feedback into video prediction models for physical robot interactions. We propose three multi-modal integration approaches and compare the performance of these tactile-enhanced video prediction models. Additionally, we introduce two new datasets of robot pushing that use a magnetic-based tactile sensor for unsupervised learning. The first dataset contains visually identical objects with different physical properties, while the second dataset mimics existing robot-pushing datasets of household object clusters. Our results demonstrate that incorporating tactile feedback into video prediction models improves scene prediction accuracy and enhances the agent's perception of physical interactions and understanding of cause-effect relationships during physical robot interactions. | ['Amir Ghalamzan-E', 'Willow Mandil'] | 2023-04-21 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 4.08301353e-01 1.27952144e-01 -2.33073369e-01 -3.16701710e-01
1.20209932e-01 -2.21811533e-01 4.31464761e-01 1.12730235e-01
-2.10044429e-01 3.41571957e-01 1.47609383e-01 1.58595100e-01
-2.08039373e-01 -5.90800047e-01 -1.11411095e+00 -2.61314750e-01
-1.10866353e-01 1.62262276e-01 5.97829223e-01 -2.35662505e-01
6.10743642e-01 4.92120713e-01 -2.03397679e+00 6.87894702e-01
3.81795406e-01 1.14569378e+00 1.17395473e+00 9.53427196e-01
2.14690492e-01 1.29070830e+00 4.73608309e-03 3.23617429e-01
2.96785355e-01 7.38252327e-02 -5.94914675e-01 2.76292205e-01
8.64362419e-02 -7.73436189e-01 -5.92435062e-01 5.02515495e-01
1.21765681e-01 3.04953158e-01 8.35418701e-01 -1.64495373e+00
-7.68110633e-01 5.34229755e-01 -1.29926011e-01 -4.57938135e-01
8.66951108e-01 4.75393653e-01 4.58347857e-01 -5.69557726e-01
9.94041920e-01 1.51082456e+00 6.85205460e-01 6.07094944e-01
-1.16853774e+00 -3.22817683e-01 2.94900179e-01 7.69393921e-01
-7.78201044e-01 -3.04411352e-01 8.73940468e-01 -7.58209348e-01
1.20327890e+00 2.66101398e-02 8.22859824e-01 1.43765020e+00
4.84702677e-01 9.46565688e-01 7.48627961e-01 -3.79693598e-01
2.93537378e-01 1.28568903e-01 -3.08726162e-01 6.10031962e-01
-5.61749637e-02 3.97573084e-01 -8.76072764e-01 5.59934489e-02
9.66988266e-01 9.82932374e-02 -1.67847350e-01 -9.31202710e-01
-1.31384563e+00 2.00215042e-01 4.30947989e-01 -1.77116469e-01
-6.66852951e-01 5.35961747e-01 1.20825350e-01 7.38284588e-02
-5.68710677e-02 4.70506459e-01 -5.43340147e-01 -3.70064646e-01
7.41488114e-02 3.82402688e-01 7.47732818e-01 1.10181510e+00
6.51267648e-01 -3.72804701e-01 6.89719617e-02 8.64170194e-01
4.71073776e-01 7.59791791e-01 4.46446091e-01 -1.63539147e+00
2.19815254e-01 5.73305905e-01 5.86842120e-01 -1.30163014e+00
-5.89662910e-01 9.78129029e-01 1.11947536e-01 5.39460778e-01
1.33927286e-01 -2.30543874e-02 -8.90674710e-01 1.50659800e+00
3.96432549e-01 1.41972557e-01 1.13824226e-01 1.02973890e+00
7.02390373e-01 5.26950777e-01 4.26686525e-01 1.86828360e-01
8.18664968e-01 -9.16744113e-01 -6.43690705e-01 -5.18031359e-01
7.85092950e-01 -5.11927605e-01 9.06796217e-01 3.98662508e-01
-7.96415985e-01 -8.66002083e-01 -1.01828563e+00 -2.22826265e-02
-4.19248343e-01 2.68210143e-01 7.52394617e-01 -1.91985086e-01
-9.12865162e-01 1.02228522e+00 -1.09083724e+00 -8.26512218e-01
-9.33008790e-02 5.18246531e-01 -5.33795714e-01 -8.11727941e-02
-7.07933247e-01 1.27440846e+00 2.75877655e-01 -2.28280306e-01
-6.31699443e-01 -3.61814409e-01 -1.10247445e+00 -4.40686077e-01
3.34711790e-01 -4.06752318e-01 1.52945328e+00 -7.15328813e-01
-1.90929019e+00 4.07106876e-01 4.80220355e-02 -3.10744613e-01
4.27253604e-01 -4.54482526e-01 5.99952079e-02 2.98027843e-01
-1.06318250e-01 1.43461907e+00 7.61704147e-01 -1.79084980e+00
-4.85217780e-01 -1.91910744e-01 -3.65814194e-02 5.98778665e-01
-1.42255351e-01 -4.50567275e-01 -5.00439405e-01 -8.86518955e-02
1.86319724e-01 -1.25746191e+00 -3.34202498e-01 5.72416365e-01
-2.14965507e-01 -8.72545540e-02 1.08107066e+00 -5.44394195e-01
3.83503139e-01 -2.10428977e+00 3.67537618e-01 6.28864020e-02
-2.52663493e-01 -2.89455634e-02 -3.88698161e-01 6.71469033e-01
3.82145554e-01 -2.27473781e-01 4.31966305e-01 -4.06043440e-01
4.73876633e-02 3.64867955e-01 -3.26825887e-01 -8.10816810e-02
1.19770192e-01 7.45236635e-01 -9.93156910e-01 -6.00871861e-01
9.64620113e-01 2.98526049e-01 -5.63597798e-01 4.58681017e-01
-5.93346357e-01 5.20540535e-01 -6.08534873e-01 4.50905800e-01
4.28382128e-01 3.73944500e-03 5.08762658e-01 -2.88179964e-01
-1.59844816e-01 1.25445470e-01 -8.03633511e-01 1.69694269e+00
-3.49023730e-01 5.63844621e-01 1.05167493e-01 -6.79761887e-01
1.00407219e+00 9.64206755e-02 8.57371807e-01 -9.47817445e-01
2.13724732e-01 -1.42601609e-01 -2.53535599e-01 -1.29729295e+00
7.60038733e-01 3.29888344e-01 -7.09210038e-02 2.49131635e-01
-2.45275542e-01 -6.36009693e-01 -5.80667891e-02 7.79304430e-02
1.34704089e+00 8.95685315e-01 -8.82734265e-03 3.59437048e-01
-2.16547981e-01 4.64624286e-01 1.85733095e-01 8.67790282e-01
-4.36486155e-01 4.58787769e-01 6.66920394e-02 -2.95498043e-01
-1.14596987e+00 -1.19024456e+00 2.44976535e-01 1.29821527e+00
1.00874782e+00 -8.65424797e-02 -3.45537543e-01 -1.11884810e-01
4.53610003e-01 7.50554502e-01 -8.12359631e-01 -3.15866500e-01
-3.76102000e-01 -2.94251461e-02 -2.59021334e-02 9.62008655e-01
2.53812790e-01 -1.68414390e+00 -1.37130952e+00 1.23702563e-01
-2.26856887e-01 -1.33358538e+00 2.58301795e-01 2.97601223e-01
-6.43129408e-01 -1.03479016e+00 4.65966575e-03 -9.02943432e-01
5.36347687e-01 6.85487092e-01 4.66334611e-01 -9.20696706e-02
-3.45887452e-01 1.27181065e+00 -6.84114039e-01 -5.41730285e-01
-6.85208917e-01 -4.61957812e-01 3.74175668e-01 -5.24300158e-01
2.68127173e-01 -3.59554648e-01 -4.55286026e-01 6.05663300e-01
-5.06375194e-01 7.33078241e-01 6.50764406e-01 3.78849357e-01
2.86279559e-01 -5.27250767e-01 1.48238078e-01 -4.84132990e-02
5.00451148e-01 -4.43431497e-01 -1.70438886e-01 2.81535298e-01
2.17250306e-02 -3.18687931e-02 2.07540363e-01 -9.96026695e-01
-1.44507551e+00 7.44048297e-01 5.46636879e-01 -5.71915388e-01
-3.31174433e-01 3.81562829e-01 2.21515000e-01 -5.09647205e-02
5.78092098e-01 -2.24309012e-01 2.07798094e-01 -1.64461836e-01
3.10753047e-01 6.83723092e-01 7.02026010e-01 -4.30441648e-01
1.28830418e-01 4.76873934e-01 -1.54332638e-01 -9.13139105e-01
-1.25291571e-01 -4.89122570e-01 -9.11796451e-01 -8.33949983e-01
7.90387809e-01 -6.70825422e-01 -1.25186634e+00 6.69597983e-01
-1.20232308e+00 -1.13697934e+00 8.82043038e-03 7.13298023e-01
-1.20269883e+00 3.46805125e-01 -8.56748164e-01 -9.72927511e-01
3.39824498e-01 -9.98012841e-01 1.21247029e+00 4.92366403e-02
-6.53819442e-01 -5.95511734e-01 -7.53648132e-02 3.27839792e-01
3.45892012e-02 5.24585284e-02 5.91762602e-01 -7.90240616e-02
-5.61791897e-01 -1.06515311e-01 -2.53364801e-01 -1.27863869e-01
2.68007994e-01 -4.53881249e-02 -8.20752919e-01 2.40739867e-01
-2.54269838e-01 -7.06894159e-01 5.39273500e-01 5.33415854e-01
1.12928271e+00 5.23597077e-02 -8.15281034e-01 -1.57003850e-01
9.32083368e-01 5.75447679e-01 6.43738329e-01 3.20182800e-01
7.29926109e-01 9.89159524e-01 1.35474885e+00 6.19721711e-01
5.81725299e-01 8.73822033e-01 7.10924983e-01 4.95376408e-01
-5.19312620e-02 -4.71133173e-01 6.51612878e-01 5.23627102e-01
2.14186683e-03 -2.70590901e-01 -8.51003349e-01 3.73555362e-01
-2.36608839e+00 -9.05056775e-01 -2.86468603e-02 1.85231221e+00
4.43646759e-01 4.81470712e-02 -1.70667022e-01 -8.39370340e-02
5.97922981e-01 -4.05519754e-01 -9.70775127e-01 -5.12943387e-01
3.00327927e-01 -5.13326108e-01 6.04747593e-01 5.01641035e-01
-1.04311264e+00 1.02393293e+00 6.59736490e+00 1.91931635e-01
-1.18944764e+00 -4.44864303e-01 3.81205902e-02 -8.17577764e-02
2.24533170e-01 -3.83107692e-01 -2.35855132e-01 1.89858764e-01
6.61115050e-01 3.47540647e-01 4.92758751e-01 1.05629301e+00
5.19001722e-01 -8.75297010e-01 -1.46086133e+00 8.47915947e-01
1.25844926e-01 -9.48405445e-01 -2.87284493e-01 -9.66684222e-02
3.45239639e-01 1.18140914e-01 1.22384831e-01 2.26576999e-01
5.80496788e-01 -7.50637650e-01 9.67014492e-01 8.24660182e-01
3.47937763e-01 6.47034347e-02 3.69845688e-01 4.57082152e-01
-1.11510146e+00 -5.24257421e-01 -2.89836735e-01 -6.68248832e-01
2.75090456e-01 -2.80691147e-01 -1.35653532e+00 -4.67983298e-02
1.06375098e+00 8.23171735e-01 -3.22804719e-01 7.93656468e-01
1.61343828e-01 -7.64588937e-02 -4.99444991e-01 -5.23578823e-01
-1.28086478e-01 1.19813241e-01 3.76024365e-01 7.12919891e-01
3.90515596e-01 1.22774750e-01 4.80415940e-01 6.05480611e-01
4.22396123e-01 -1.67149842e-01 -1.03870618e+00 -5.22960722e-02
6.11881971e-01 8.11629176e-01 -6.68656886e-01 -1.79397658e-01
-2.22709164e-01 1.04804635e+00 4.10792351e-01 3.24797541e-01
-8.25325251e-01 1.61379725e-01 5.78545392e-01 -6.63319277e-03
3.21388215e-01 -5.11718750e-01 -3.42741579e-01 -5.46875000e-01
1.04263149e-01 -1.68109626e-01 -3.84008706e-01 -1.63874888e+00
-9.38089132e-01 -2.60754943e-01 1.74408033e-01 -1.46054173e+00
-3.94172132e-01 -1.14460039e+00 -1.47572353e-01 2.14671969e-01
-9.50894594e-01 -1.26953328e+00 -7.61915088e-01 1.18933558e-01
7.40556240e-01 3.68337810e-01 8.84510100e-01 -4.29537922e-01
7.82350302e-02 -2.02000558e-01 -1.53342500e-01 -3.80833000e-01
7.27931738e-01 -7.15081155e-01 6.89859539e-02 -4.29269858e-02
-4.81449187e-01 2.47191757e-01 9.77864146e-01 -1.25738227e+00
-1.79964602e+00 -7.87173212e-01 2.09165350e-01 -6.49868667e-01
5.95647335e-01 -1.87537938e-01 -7.76573777e-01 8.24588120e-01
-1.08831212e-01 -2.54966170e-01 3.59596819e-01 -2.97911376e-01
-6.40220894e-03 3.82083267e-01 -1.15325272e+00 9.04645979e-01
1.37241530e+00 -3.62810910e-01 -7.20889390e-01 1.29505619e-01
6.00724578e-01 -4.65317041e-01 -6.48616970e-01 7.26655126e-01
1.10712445e+00 -8.49343181e-01 1.09724903e+00 -4.12186474e-01
8.57470870e-01 -1.40654847e-01 -3.91836137e-01 -1.29113543e+00
-4.51703042e-01 2.38985550e-02 -1.07690848e-01 6.31827950e-01
1.60870299e-01 -6.66277334e-02 8.83089662e-01 1.09751236e+00
-2.12221384e-01 -4.74497288e-01 -4.32673305e-01 -6.48334742e-01
-4.52775955e-01 -8.95594835e-01 6.72894940e-02 5.56007504e-01
9.43983495e-01 -1.41496822e-01 -5.75750768e-01 2.02780321e-01
5.55292182e-02 -1.33940145e-01 1.18319404e+00 -1.08602178e+00
-1.83311597e-01 -1.10712782e-01 -6.56631410e-01 -1.14202070e+00
2.10821003e-01 -3.32165003e-01 8.07732463e-01 -1.84716821e+00
3.44994336e-01 -3.76046360e-01 -5.34479506e-04 5.10807097e-01
2.58604914e-01 -2.51769628e-02 5.32658815e-01 2.20799834e-01
-8.05470407e-01 6.48470819e-01 1.30567312e+00 -1.35338381e-01
-4.76127446e-01 -3.96745473e-01 1.59301594e-01 9.35106993e-01
8.35809588e-01 1.15726270e-01 -5.04608572e-01 -5.26064873e-01
1.18762739e-01 2.64754355e-01 7.56589711e-01 -1.31312203e+00
3.44158232e-01 -7.17532516e-01 5.21533847e-01 -5.95046997e-01
1.14183962e+00 -1.06511986e+00 9.46356058e-02 6.72468066e-01
-5.99458873e-01 -1.89560071e-01 5.14411449e-01 8.77220869e-01
3.23433459e-01 -1.52984872e-01 1.54763401e-01 -1.19574822e-01
-1.45384872e+00 -2.02476576e-01 -1.00902081e+00 -8.42767298e-01
1.37728477e+00 -6.09859049e-01 -3.64729285e-01 -6.58387244e-01
-7.83791840e-01 3.73948187e-01 7.52966404e-01 1.08074486e+00
9.85141039e-01 -1.27802753e+00 7.60636777e-02 -2.91696694e-02
2.74760455e-01 -3.19491327e-01 4.88341719e-01 6.15442038e-01
-5.56970417e-01 1.46086439e-01 -5.41889727e-01 -8.21012974e-01
-1.37819195e+00 5.95989406e-01 -5.35529889e-02 4.94483352e-01
-3.78144234e-01 7.38627493e-01 2.64532477e-01 -8.43984365e-01
3.09300572e-01 -5.23472250e-01 -2.05644846e-01 -5.86326241e-01
1.39641166e-01 4.73523885e-01 -5.83871663e-01 -5.96289515e-01
-1.50008246e-01 4.95424241e-01 2.62653470e-01 -2.21773461e-01
1.24243224e+00 -4.42057937e-01 3.48756790e-01 8.95773888e-01
8.51656437e-01 -6.14555538e-01 -1.97751319e+00 1.04398830e-02
-1.46870408e-02 -3.09635997e-01 -3.43525469e-01 -8.66422653e-01
-3.28431398e-01 6.92974865e-01 6.06414855e-01 -1.01676464e-01
6.18318856e-01 1.75833523e-01 5.72077513e-01 9.84531641e-01
7.69361675e-01 -1.77020204e+00 8.47786546e-01 5.90909362e-01
1.08922148e+00 -1.52770269e+00 -7.50342086e-02 -8.42135668e-01
-9.81723070e-01 9.83932555e-01 1.17810392e+00 2.43619941e-02
3.84944588e-01 4.68197316e-01 3.46183218e-02 7.06614461e-04
-9.03370738e-01 3.94117907e-02 -1.93142481e-02 1.03325915e+00
-1.37803316e-01 2.18794227e-01 4.51116532e-01 2.62212276e-01
-1.91630706e-01 3.41757566e-01 5.51016510e-01 1.43037522e+00
-7.12436080e-01 -6.13003492e-01 -3.75971138e-01 5.05186975e-01
4.66662705e-01 4.57809240e-01 -6.68032944e-01 5.59816062e-01
8.54398459e-02 1.21832442e+00 2.13147640e-01 -1.04974270e+00
2.52863973e-01 4.35809651e-03 8.58389556e-01 -5.16866505e-01
1.12916462e-01 -3.59359860e-01 2.34462634e-01 -8.95258605e-01
-4.97922421e-01 -9.86665666e-01 -1.67985761e+00 -6.08669929e-02
-2.02998906e-01 -5.70897043e-01 1.01145172e+00 9.84446287e-01
5.89331925e-01 3.68470430e-01 2.30116740e-01 -1.98318267e+00
-2.08082154e-01 -1.06530857e+00 -2.31737390e-01 7.72890627e-01
2.30176896e-02 -1.28057575e+00 -3.02346312e-02 5.05920112e-01] | [4.862826347351074, 0.5889959931373596] |
d2a07263-9589-40dc-be8c-5cb30f3545f3 | exploring-length-generalization-in-large | 2207.04901 | null | https://arxiv.org/abs/2207.04901v2 | https://arxiv.org/pdf/2207.04901v2.pdf | Exploring Length Generalization in Large Language Models | The ability to extrapolate from short problem instances to longer ones is an important form of out-of-distribution generalization in reasoning tasks, and is crucial when learning from datasets where longer problem instances are rare. These include theorem proving, solving quantitative mathematics problems, and reading/summarizing novels. In this paper, we run careful empirical studies exploring the length generalization capabilities of transformer-based language models. We first establish that naively finetuning transformers on length generalization tasks shows significant generalization deficiencies independent of model scale. We then show that combining pretrained large language models' in-context learning abilities with scratchpad prompting (asking the model to output solution steps before producing an answer) results in a dramatic improvement in length generalization. We run careful failure analyses on each of the learning modalities and identify common sources of mistakes that highlight opportunities in equipping language models with the ability to generalize to longer problems. | ['Behnam Neyshabur', 'Ethan Dyer', 'Guy Gur-Ari', 'Ambrose Slone', 'Vinay Ramasesh', 'Vedant Misra', 'Aitor Lewkowycz', 'Anders Andreassen', 'Yuhuai Wu', 'Cem Anil'] | 2022-07-11 | null | null | null | null | ['automated-theorem-proving', 'automated-theorem-proving'] | ['miscellaneous', 'reasoning'] | [ 4.84896749e-01 1.74118221e-01 -3.98594327e-02 -3.58728647e-01
-1.03790736e+00 -9.70353127e-01 3.96274537e-01 4.85437751e-01
-3.28366429e-01 6.14491642e-01 -8.25560745e-03 -1.15499473e+00
-6.03709698e-01 -1.03568494e+00 -9.23324585e-01 -1.27286136e-01
-1.26443177e-01 5.69816530e-01 2.41124406e-01 -3.11471939e-01
4.95751143e-01 5.05363703e-01 -1.31106377e+00 8.09733629e-01
1.03020096e+00 5.52478850e-01 1.80584311e-01 9.49525893e-01
-5.30774355e-01 1.39848483e+00 -6.68781936e-01 -4.01927590e-01
1.21817425e-01 -2.21609339e-01 -1.10649610e+00 -4.09030169e-01
8.53064120e-01 -3.22821021e-01 -1.88280895e-01 7.46943414e-01
1.71181932e-01 1.65916815e-01 5.92030823e-01 -1.26463223e+00
-1.05317259e+00 1.24547124e+00 -3.13879788e-01 8.69015038e-01
6.28761232e-01 4.42642391e-01 1.24172294e+00 -5.43326259e-01
4.58375335e-01 1.43868566e+00 8.57596695e-01 6.23651266e-01
-1.61468065e+00 -6.91214561e-01 2.78346270e-01 7.49826804e-02
-1.12207520e+00 -2.45830730e-01 4.62669581e-01 -5.30244410e-01
1.46869862e+00 2.34928146e-01 3.08083266e-01 1.09236252e+00
3.28619510e-01 6.03872836e-01 1.30450845e+00 -4.96291012e-01
1.00332595e-01 1.35828003e-01 4.42437410e-01 7.31031060e-01
2.36173719e-01 1.31462187e-01 -6.54943764e-01 -1.00901797e-01
5.98174572e-01 -1.03336558e-01 -6.40458763e-02 -9.49303433e-02
-1.17807937e+00 6.96647704e-01 2.67801970e-01 3.41454059e-01
2.35683680e-01 3.22169513e-01 4.61375803e-01 9.20608282e-01
1.12844855e-01 1.17073548e+00 -9.41712260e-01 -2.47037351e-01
-5.72293639e-01 5.88986874e-01 6.40463829e-01 1.01915681e+00
5.51220059e-01 -1.34500563e-01 -3.44637781e-01 5.00773370e-01
-2.29339257e-01 4.72967654e-01 4.85933542e-01 -1.06165898e+00
9.41913664e-01 7.81631410e-01 -1.81659147e-01 -6.04304075e-01
-5.72462738e-01 -7.93739974e-01 -2.90081203e-01 1.23900606e-03
8.09620380e-01 -1.53075159e-01 -4.29680109e-01 2.02695560e+00
-7.30035380e-02 -1.44923151e-01 2.14041993e-01 2.54843652e-01
5.11947036e-01 5.81086874e-01 3.69293213e-01 -1.08876258e-01
1.03050351e+00 -6.14655793e-01 -2.54210591e-01 -5.32296717e-01
1.55456817e+00 -3.79909575e-01 1.82182729e+00 6.41641080e-01
-1.23354661e+00 -6.08331144e-01 -8.56828451e-01 -5.62138498e-01
-4.73783195e-01 -3.21137607e-01 8.16790819e-01 4.98626024e-01
-1.01400101e+00 7.54776180e-01 -4.07296538e-01 -2.95976788e-01
4.18623120e-01 1.54860169e-01 -1.79486781e-01 -2.87178874e-01
-1.03680098e+00 1.15005124e+00 5.00520468e-01 -4.87605065e-01
-7.70192683e-01 -1.34339917e+00 -8.27932477e-01 4.23567206e-01
4.07980323e-01 -6.54982924e-01 1.60128093e+00 -5.62702775e-01
-1.01211393e+00 7.45622158e-01 -5.53148277e-02 -4.98655379e-01
4.52159911e-01 -2.47376725e-01 -1.64795473e-01 -2.61541009e-01
-1.14772217e-02 4.84965891e-01 3.89921159e-01 -7.75628805e-01
-5.17594159e-01 -4.10359502e-01 6.35679543e-01 1.92024410e-02
-3.61965716e-01 -2.20652923e-01 5.34388363e-01 -5.85172951e-01
9.21766609e-02 -5.31307578e-01 2.55047753e-02 -1.71023339e-01
-8.13151374e-02 -7.36959398e-01 3.58885109e-01 -4.66476262e-01
1.48685193e+00 -1.88527918e+00 1.84749484e-01 1.87521785e-01
3.57426524e-01 -3.04753214e-01 -3.75150651e-01 5.75897276e-01
-2.05961421e-01 3.36905688e-01 -3.69585672e-04 8.17742497e-02
3.94504994e-01 -3.16827260e-02 -7.15291739e-01 -5.71950562e-02
1.75843865e-01 1.25409472e+00 -9.98758197e-01 -3.22450995e-01
-1.57086655e-01 -3.06509316e-01 -9.53758299e-01 -4.07403335e-03
-6.97560310e-01 6.51004612e-02 -1.00520648e-01 5.07727802e-01
1.98459059e-01 -4.54739392e-01 1.21302105e-01 2.17719480e-01
2.53374428e-01 7.33242631e-01 -6.60623968e-01 1.72420859e+00
-7.76595116e-01 6.27596796e-01 -4.95024949e-01 -9.39235449e-01
5.89011848e-01 -1.62991900e-02 -3.47779304e-01 -9.86410499e-01
-1.67686149e-01 1.64606482e-01 5.48963785e-01 -7.22013891e-01
3.19024652e-01 -6.19531512e-01 -1.77360371e-01 7.59882152e-01
5.43243811e-02 -5.04217148e-01 4.84237850e-01 3.92790645e-01
1.13492167e+00 -1.38970956e-01 -1.78560633e-02 -4.46368486e-01
3.74922514e-01 -1.30938934e-02 8.58088732e-02 1.20264018e+00
1.69005141e-01 -1.32127507e-02 9.02478874e-01 -5.14402628e-01
-1.03672194e+00 -1.29247200e+00 8.16955492e-02 1.73794675e+00
-3.95834386e-01 -7.16673613e-01 -4.43222344e-01 -6.70168519e-01
1.37749687e-01 1.28037608e+00 -5.32285810e-01 -5.78406334e-01
-4.96167421e-01 -2.84662813e-01 7.62752473e-01 1.00765789e+00
6.90337867e-02 -9.09914732e-01 -5.50110400e-01 -1.21872658e-02
1.05638765e-01 -1.03200853e+00 -3.18319984e-02 4.27305281e-01
-1.27645957e+00 -1.04456902e+00 -1.35423928e-01 -7.10615575e-01
5.53715110e-01 -4.82319556e-02 1.37324739e+00 4.66002434e-01
-1.90313429e-01 4.99216229e-01 -1.61912143e-02 -5.57432711e-01
-6.87524617e-01 4.14349645e-01 1.50108799e-01 -1.04158127e+00
6.00021958e-01 -7.39414096e-01 1.66337222e-01 -2.55996622e-02
-8.99775743e-01 1.31710544e-01 5.69207013e-01 7.23055840e-01
-2.76332553e-02 1.66461825e-01 6.48531437e-01 -1.02937698e+00
1.16138363e+00 -4.20075804e-01 -6.72094762e-01 6.68063760e-01
-4.87400174e-01 5.42035580e-01 9.17090595e-01 -6.55912042e-01
-6.66403472e-01 -5.78712761e-01 7.88238049e-02 -4.18800376e-02
-2.76667215e-02 7.61671543e-01 2.03828037e-01 3.88547815e-02
1.24328375e+00 2.92533249e-01 -1.41053021e-01 -2.90304631e-01
2.64830649e-01 3.19182128e-02 2.59003431e-01 -1.52680790e+00
9.96700466e-01 -2.47206926e-01 6.77901600e-03 -5.63174188e-01
-1.24564087e+00 2.42487229e-02 -2.21705317e-01 2.76015997e-01
2.85614759e-01 -6.49096847e-01 -1.00302601e+00 1.39418572e-01
-8.92062008e-01 -1.07227051e+00 -4.89203990e-01 1.46619171e-01
-6.20228529e-01 3.07696983e-02 -7.18455970e-01 -6.52525187e-01
-5.30638546e-02 -1.05439818e+00 5.56073964e-01 2.44721636e-01
-5.11063516e-01 -1.20451164e+00 -8.35846663e-02 3.38103890e-01
5.53166568e-01 -2.94946313e-01 1.78075683e+00 -9.81484950e-01
-4.89133000e-01 1.84380170e-02 -1.12998486e-01 1.82955787e-01
-2.09153533e-01 -2.63946533e-01 -9.55059886e-01 -2.17176706e-01
-8.95314366e-02 -8.96856666e-01 7.18287289e-01 -6.27248585e-02
1.71904743e+00 -3.43520492e-01 -1.91495150e-01 4.82051015e-01
1.06345975e+00 -2.55882502e-01 3.08907837e-01 2.72288650e-01
3.33897680e-01 3.30887467e-01 9.76892784e-02 1.51339576e-01
3.26717257e-01 8.22030827e-02 -1.87155768e-01 4.30008024e-01
1.83768049e-01 -6.12939715e-01 4.02956516e-01 2.72659481e-01
2.12504894e-01 3.03658657e-02 -1.28203559e+00 4.92045730e-01
-1.33647263e+00 -1.01163936e+00 2.26573691e-01 2.06382370e+00
1.21069562e+00 7.42592990e-01 2.23356690e-02 2.56809682e-01
1.07040890e-02 -1.97857454e-01 -5.15007079e-01 -7.61702240e-01
9.65711996e-02 5.79761624e-01 2.01377869e-01 8.64943624e-01
-5.07324278e-01 1.01905143e+00 7.27009916e+00 8.19671810e-01
-8.47041070e-01 -8.47898498e-02 5.09078145e-01 -1.43817902e-01
-5.97466230e-01 -9.96860489e-02 -6.12902761e-01 -6.96881562e-02
1.22425377e+00 -3.46285611e-01 7.64135480e-01 8.37330997e-01
-3.31006944e-01 -7.53652155e-02 -1.99302888e+00 7.19574928e-01
-2.16820747e-01 -1.38194287e+00 3.69374216e-01 -1.39206529e-01
7.52056241e-01 -3.19340736e-01 3.25815916e-01 9.32965338e-01
3.76446009e-01 -1.54376507e+00 6.24512970e-01 3.83248866e-01
6.97603762e-01 -6.52738512e-01 4.43684101e-01 8.61519992e-01
-7.00794935e-01 -6.29991591e-01 -4.48839903e-01 -8.27965677e-01
-3.59567344e-01 1.96209654e-01 -9.83709037e-01 -6.61967918e-02
3.56069982e-01 2.96851724e-01 -9.96844232e-01 6.84871316e-01
-4.12130117e-01 5.46411991e-01 -3.20734739e-01 -1.21346444e-01
-6.18855171e-02 2.68328756e-01 1.64360747e-01 1.14036822e+00
2.17671007e-01 3.60997617e-01 2.06252426e-01 1.23265529e+00
-9.85806361e-02 -1.96333244e-01 -7.54887044e-01 -4.17429030e-01
5.49955785e-01 8.03107679e-01 -4.94495213e-01 -4.18462336e-01
-3.00749242e-01 4.49303895e-01 8.21993470e-01 5.06870031e-01
-5.44636071e-01 -3.19201499e-01 3.74726355e-01 2.82608449e-01
-3.71360928e-02 -3.93317491e-01 -7.29192317e-01 -9.91073847e-01
7.26215839e-02 -1.23024297e+00 5.67247272e-01 -1.04309487e+00
-1.35227919e+00 -2.76376605e-02 3.09541613e-01 -4.69033420e-01
-2.36845136e-01 -9.95650887e-01 -7.06312180e-01 9.36723232e-01
-1.12845206e+00 -9.04850066e-01 2.96147652e-02 4.50867712e-01
6.15760684e-01 -1.35892391e-01 8.28051686e-01 -1.70792431e-01
-1.29696503e-01 1.06898689e+00 -3.06022435e-01 1.04078427e-01
3.57769996e-01 -1.40045166e+00 4.70159113e-01 6.46228611e-01
9.27876011e-02 1.29757440e+00 7.84671783e-01 -3.59100163e-01
-1.46618855e+00 -6.70320809e-01 1.10315323e+00 -1.01967883e+00
1.00267291e+00 -5.02656519e-01 -9.88336802e-01 1.19585073e+00
-1.22237019e-01 -1.94322422e-01 7.04871058e-01 6.27952099e-01
-1.01561105e+00 -9.93386358e-02 -9.85674977e-01 6.45887494e-01
1.24893653e+00 -1.09263790e+00 -1.19829154e+00 4.00962353e-01
1.02770638e+00 -5.15248477e-01 -8.54422987e-01 2.01327503e-01
5.23331881e-01 -8.94519866e-01 9.27379966e-01 -1.36111748e+00
8.57309759e-01 1.05574720e-01 -2.63690561e-01 -1.11493695e+00
-4.24033076e-01 -4.08823878e-01 -1.86449721e-01 9.22256231e-01
6.68663502e-01 -6.59213901e-01 4.45373863e-01 7.35817909e-01
-2.29497347e-02 -8.73567522e-01 -4.61401492e-01 -8.17340136e-01
1.01605701e+00 -7.96631277e-01 6.04975283e-01 8.66460741e-01
5.84991932e-01 5.15550911e-01 5.40437758e-01 2.07724005e-01
3.11466634e-01 9.32080895e-02 6.90091908e-01 -1.31385827e+00
-4.87139314e-01 -7.07081139e-01 -1.23924360e-01 -1.16413629e+00
3.39655757e-01 -1.20711148e+00 -3.04056227e-01 -1.32741714e+00
2.75711536e-01 -5.59142709e-01 -2.40951657e-01 4.94681686e-01
-1.70148790e-01 -3.66017163e-01 1.67157114e-01 -3.17654520e-01
-6.44229293e-01 4.20934707e-02 1.18694878e+00 -2.13082626e-01
3.44836786e-02 -2.70421361e-03 -1.01518965e+00 6.36902273e-01
7.69057989e-01 -3.04753661e-01 -9.68621373e-01 -7.52351940e-01
9.84710336e-01 4.94532920e-02 4.91692334e-01 -1.15836716e+00
2.67930806e-01 -5.51944315e-01 4.58475620e-01 -2.41583765e-01
-1.40310705e-01 -5.75280070e-01 -5.32592595e-01 4.21383053e-01
-1.13227499e+00 6.03521168e-01 7.73857236e-01 1.96667209e-01
4.46605295e-01 -2.68375188e-01 5.28129935e-01 -3.87625158e-01
-5.48389614e-01 -1.75376058e-01 -2.06181601e-01 7.23133087e-01
6.72335505e-01 -1.67843904e-02 -6.51419282e-01 -1.12578854e-01
-7.28263199e-01 4.07506287e-01 4.13038641e-01 4.45411056e-01
4.41286623e-01 -8.42235923e-01 -6.92206681e-01 2.23071188e-01
2.39240766e-01 1.60588488e-01 2.80418783e-01 7.94388235e-01
-5.15727699e-01 6.87424362e-01 2.79964022e-02 -5.65614522e-01
-9.35757875e-01 1.01946652e+00 4.70387071e-01 -4.60159898e-01
-5.09065390e-01 1.29279041e+00 4.38105792e-01 -7.84953058e-01
4.79430318e-01 -1.29011488e+00 3.27548742e-01 -3.73748869e-01
7.24685729e-01 1.91455826e-01 8.19250494e-02 4.10702288e-01
-1.59221232e-01 3.74490857e-01 -3.38503152e-01 -6.90095350e-02
1.16649389e+00 2.27707535e-01 -1.28671691e-01 7.77636111e-01
8.81749630e-01 2.74610333e-02 -8.19172740e-01 -4.06246066e-01
2.92205304e-01 -8.19442347e-02 -4.08116758e-01 -1.23437226e+00
-3.51187855e-01 1.13680398e+00 8.00760314e-02 2.79852420e-01
7.90350139e-01 2.14693248e-01 3.95422250e-01 1.15590858e+00
4.94615883e-01 -9.18758512e-01 2.22936004e-01 9.14510489e-01
8.90985787e-01 -1.09120381e+00 2.74305046e-02 -2.88285706e-02
-1.35767788e-01 1.21723258e+00 1.04511976e+00 6.91763535e-02
2.42567614e-01 4.43698108e-01 -1.62410468e-01 -2.10312247e-01
-1.23485672e+00 3.90814096e-01 9.40881893e-02 5.44358552e-01
6.72997057e-01 -1.33786500e-01 3.53600681e-02 7.60712922e-01
-8.37770522e-01 5.19426763e-02 5.98917782e-01 9.66870964e-01
-5.38267791e-01 -7.73692846e-01 -3.81687224e-01 5.85331976e-01
-1.90145075e-01 -5.15770197e-01 -2.89582044e-01 8.24266315e-01
-2.42837099e-03 6.26068175e-01 -4.25967714e-03 -2.96944529e-01
2.10665435e-01 7.76386082e-01 9.73506331e-01 -9.09663916e-01
-6.07961476e-01 -7.88735330e-01 -8.48199353e-02 -4.43369925e-01
2.90362537e-01 -3.57875764e-01 -1.40429342e+00 -5.67179799e-01
-1.11565225e-01 1.28198639e-01 4.54831682e-02 1.36620080e+00
1.15242533e-01 6.16796315e-01 -1.50206909e-01 -2.30729252e-01
-1.42824411e+00 -1.00478303e+00 -2.79763103e-01 3.54610324e-01
5.93985796e-01 -3.64402652e-01 -5.27181029e-01 -1.00638546e-01] | [9.487804412841797, 7.268215656280518] |
43babe3e-8010-4416-af0f-ac9549c0fc16 | adaptive-as-natural-as-possible-image | null | null | http://openaccess.thecvf.com/content_cvpr_2015/html/Lin_Adaptive_As-Natural-As-Possible_Image_2015_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2015/papers/Lin_Adaptive_As-Natural-As-Possible_Image_2015_CVPR_paper.pdf | Adaptive As-Natural-As-Possible Image Stitching | The goal of image stitching is to create natural-looking mosaics free of artifacts that may occur due to relative camera motion, illumination changes, and optical aberrations. In this paper, we propose a novel stitching method, that uses a smooth stitching field over the entire target image, while accounting for all the local transformation variations. Computing the warp is fully automated and uses a combination of local homography and global similarity transformations, both of which are estimated with respect to the target. We mitigate the perspective distortion in the non-overlapping regions by linearizing the homography and slowly changing it to the global similarity. The proposed method is easily generalized to multiple images, and allows one to automatically obtain the best perspective in the panorama. It is also more robust to parameter selection, and hence more automated compared with state-of-the-art methods. The benefits of this method are demonstrated using a variety of challenging cases. | ['Karthikeyan Natesan Ramamurthy', 'Chung-Ching Lin', 'Aleksandr Y. Aravkin', 'Sharathchandra U. Pankanti'] | 2015-06-01 | null | null | null | cvpr-2015-6 | ['image-stitching'] | ['computer-vision'] | [ 8.29639256e-01 -4.03282255e-01 2.06934735e-01 -6.16970705e-03
-4.71252531e-01 -8.52030456e-01 6.68658674e-01 -3.66222352e-01
-1.12218827e-01 3.36590767e-01 1.07239008e-01 1.60419047e-01
-6.12664036e-02 -4.26195830e-01 -4.74824190e-01 -8.73761654e-01
2.39835799e-01 2.37802699e-01 5.42129934e-01 -1.10664509e-01
6.07312381e-01 5.83203912e-01 -1.33762360e+00 -2.57506937e-01
9.75951612e-01 6.74825251e-01 2.04733178e-01 6.64449275e-01
5.19389212e-01 3.87131691e-01 -5.23032188e-01 -1.93086743e-01
6.58137321e-01 -4.61413652e-01 -3.25951338e-01 7.90705264e-01
9.73727882e-01 -4.86447543e-01 -1.74616858e-01 1.14427483e+00
1.85006380e-01 5.29695265e-02 2.67923117e-01 -9.61457908e-01
-3.80220041e-02 -2.19412521e-01 -1.09457171e+00 -3.01172107e-01
5.06099522e-01 1.56997517e-01 6.12661898e-01 -6.08028233e-01
8.95459473e-01 1.04329598e+00 7.04309762e-01 -7.98183903e-02
-1.42713249e+00 -5.74013233e-01 -4.34980065e-01 2.15335861e-02
-1.23082197e+00 -5.27749538e-01 9.82384741e-01 -4.95727211e-01
3.25789213e-01 3.44679683e-01 6.42769277e-01 4.54957873e-01
6.76893234e-01 1.20019563e-01 1.27136850e+00 -8.77940178e-01
7.03311265e-02 -2.04026282e-01 -1.75780252e-01 5.73553264e-01
3.06954414e-01 3.77984285e-01 -3.98628801e-01 -3.47440451e-01
1.02437913e+00 1.16259396e-01 -7.27605760e-01 -8.46001446e-01
-1.63964939e+00 4.26900595e-01 2.23349929e-01 3.67626995e-01
-2.96095282e-01 1.16621792e-01 -1.60856366e-01 3.52638103e-02
2.81465054e-01 6.96159363e-01 2.11097803e-02 1.73893142e-02
-1.17952740e+00 2.52035439e-01 5.65217435e-01 9.04181004e-01
9.59422410e-01 -5.37937880e-03 2.46509299e-01 7.79388905e-01
4.41504456e-02 6.52829826e-01 3.33040148e-01 -1.21301258e+00
3.05841416e-01 3.59920055e-01 1.91035792e-01 -1.23947799e+00
-1.47566661e-01 -1.36245221e-01 -7.06697166e-01 6.56070530e-01
4.16611522e-01 -9.92131606e-02 -8.14145565e-01 1.34516859e+00
5.42975962e-01 2.38797709e-01 -9.83211026e-03 8.06078434e-01
5.85543700e-02 5.76153457e-01 -9.68599498e-01 -3.74490827e-01
1.25949490e+00 -1.04793441e+00 -8.86294663e-01 -5.49377620e-01
1.30674765e-01 -1.57651484e+00 5.46696603e-01 4.32445735e-01
-1.03316724e+00 -3.15130204e-01 -1.39226532e+00 -8.17032233e-02
2.10958645e-01 1.12324692e-01 1.62087694e-01 5.27174413e-01
-1.07924724e+00 6.65351748e-01 -6.95220232e-01 -3.96345466e-01
-2.77309984e-01 2.85239220e-01 -4.37264115e-01 -2.19864503e-01
-6.55530572e-01 1.08103096e+00 2.42584303e-01 -1.03105791e-01
-3.33925605e-01 -4.01820660e-01 -8.99788618e-01 -4.62305509e-02
6.27828062e-01 -7.29344189e-01 1.12167490e+00 -1.17011964e+00
-1.90782416e+00 7.02281773e-01 -3.40421796e-01 -1.59463093e-01
5.58631241e-01 -2.59283662e-01 -2.13451937e-01 2.50034153e-01
-4.78165634e-02 4.00033712e-01 1.39194238e+00 -1.39647925e+00
-3.85573387e-01 -3.11896712e-01 -4.14167166e-01 5.39221823e-01
-5.78182749e-02 -9.62434188e-02 -7.85938084e-01 -7.99070835e-01
4.27150249e-01 -1.33510077e+00 -3.60016257e-01 2.68515572e-02
-2.73851126e-01 8.33212435e-01 1.21166646e+00 -8.74992549e-01
1.13749480e+00 -2.21536350e+00 2.06158638e-01 3.20147842e-01
1.10534551e-02 1.22501753e-01 -5.89343756e-02 7.66987801e-01
4.85012233e-02 -6.20891035e-01 -5.29911935e-01 -3.43148559e-01
-6.29629254e-01 -3.00216153e-02 -2.90562123e-01 8.18378627e-01
-3.25365216e-01 4.75769192e-01 -7.49671221e-01 -3.41254860e-01
6.19586289e-01 4.81485128e-01 -2.19270274e-01 9.13057029e-02
8.48544016e-02 5.56242228e-01 -6.46182848e-03 5.14993608e-01
1.04994714e+00 2.04872087e-01 3.48345608e-01 -3.03754240e-01
-5.26157320e-01 -1.15900159e-01 -1.24115860e+00 1.61383641e+00
-3.03037494e-01 9.52436924e-01 1.46430284e-01 -2.72335529e-01
1.05083609e+00 1.77894667e-01 6.23838902e-01 -2.98069596e-01
2.68910527e-01 4.61447030e-01 -2.37717386e-02 -1.94945231e-01
6.77442491e-01 -1.69864535e-01 2.69508570e-01 4.73296255e-01
-1.41700357e-01 -7.29898274e-01 1.47018865e-01 -3.89654711e-02
7.49558508e-01 3.78740542e-02 7.93371499e-01 -3.36608261e-01
5.83006144e-01 -7.08606169e-02 4.50077266e-01 3.29702348e-01
1.26313224e-01 1.05307806e+00 2.00822636e-01 -3.69879842e-01
-1.30036736e+00 -7.18820989e-01 9.52592120e-02 -4.34171855e-02
7.53003120e-01 -3.68656158e-01 -9.64381039e-01 -1.79988191e-01
-1.64089084e-01 4.85030174e-01 -4.14201945e-01 7.47098401e-03
-7.95989811e-01 -2.98117220e-01 1.89552214e-02 2.15867963e-02
7.39735842e-01 -4.51375633e-01 -1.01470840e+00 1.33097574e-01
-2.91833371e-01 -1.36326516e+00 -1.12938726e+00 -4.71848816e-01
-1.16501176e+00 -1.26716065e+00 -9.40335095e-01 -6.23600006e-01
8.42375576e-01 1.06706464e+00 6.15775168e-01 -8.03512055e-03
-9.62731540e-02 2.85311937e-01 -6.87687397e-02 1.05677694e-02
-5.23503065e-01 -4.12404507e-01 -1.44430384e-01 4.57585096e-01
-2.73559600e-01 -5.34285605e-01 -5.40944576e-01 8.38252008e-01
-1.19055033e+00 4.00214106e-01 3.77075523e-01 8.97308588e-01
4.05522913e-01 2.29002014e-01 -4.75842148e-01 -6.46199167e-01
2.54178166e-01 3.03664416e-01 -1.06904137e+00 1.74560070e-01
-4.99687821e-01 -5.18336445e-02 4.26384330e-01 -5.62642217e-01
-1.32295799e+00 3.51984411e-01 5.33217728e-01 -5.68870723e-01
1.48443356e-01 1.45023242e-01 3.50774586e-04 -7.09271431e-01
5.41244030e-01 9.66466665e-02 2.75469989e-01 -3.93773228e-01
2.76974201e-01 3.07225376e-01 8.06130230e-01 -3.68319750e-02
1.30584705e+00 9.16661739e-01 3.17129105e-01 -1.24892437e+00
-2.76727706e-01 -5.90312839e-01 -8.58020246e-01 -4.26051855e-01
6.32263005e-01 -5.01558363e-01 -2.60118455e-01 9.88993108e-01
-1.20460284e+00 -9.38226953e-02 -4.78508063e-02 6.18093669e-01
-5.96588075e-01 9.80449736e-01 -2.11069241e-01 -4.45975065e-01
-3.14621925e-01 -1.27275538e+00 1.00235939e+00 3.66755128e-01
-1.79525658e-01 -9.08027828e-01 4.55636144e-01 1.79610536e-01
3.28613460e-01 3.91310334e-01 4.71008211e-01 2.77702421e-01
-8.83082390e-01 -2.99236745e-01 5.86991049e-02 2.48096094e-01
4.93032336e-01 4.07609195e-01 -7.70601094e-01 -5.49762726e-01
3.95760924e-01 3.04697126e-01 6.07974768e-01 5.98386168e-01
2.05963448e-01 -3.36551815e-01 -4.25660223e-01 8.76904190e-01
1.68342018e+00 4.67304558e-01 9.46433961e-01 6.24829590e-01
6.41277492e-01 6.53696358e-01 8.63681078e-01 9.31946412e-02
-3.05087063e-02 1.28641129e+00 3.22489858e-01 -3.46364647e-01
-1.22547410e-01 -2.34153178e-02 3.97135347e-01 4.28789467e-01
-7.12266639e-02 -1.82202030e-02 -6.56605959e-01 4.59399968e-01
-1.76695502e+00 -9.81028616e-01 -1.89635873e-01 2.68990970e+00
5.22350252e-01 -2.16540858e-01 -2.47520939e-01 5.27902730e-02
9.81100023e-01 5.57400525e-01 -2.11432323e-01 -2.41334200e-01
-1.06077552e-01 -6.66852742e-02 9.53223288e-01 9.20675278e-01
-9.07711923e-01 8.73508632e-01 6.60764170e+00 4.96680856e-01
-1.35582399e+00 -2.77005106e-01 3.91266495e-02 1.92340717e-01
-2.36256838e-01 6.15991533e-01 -2.58634001e-01 3.00843835e-01
2.30632469e-01 -1.33863553e-01 6.19764686e-01 4.49614853e-01
2.86442935e-02 -5.88659048e-01 -6.92581952e-01 1.01812661e+00
4.98269916e-01 -1.20370269e+00 -3.73367399e-01 2.03601882e-01
1.08680212e+00 -2.82544702e-01 -6.88051656e-02 -8.11965466e-01
-3.36801670e-02 -3.97591740e-01 6.85409844e-01 2.60160685e-01
7.50855744e-01 -5.19113064e-01 5.09825766e-01 1.71332419e-01
-1.05816042e+00 8.69268551e-02 5.12736253e-02 1.97485134e-01
5.39997339e-01 6.86925411e-01 -8.65284443e-01 6.28688872e-01
3.71152610e-01 6.59657955e-01 -4.26349491e-01 1.23463285e+00
-2.13822201e-01 -8.91523957e-02 -4.43134546e-01 5.29997408e-01
4.91628647e-02 -8.30920696e-01 9.89767253e-01 7.54908741e-01
8.06229293e-01 8.89520645e-02 -5.79749681e-02 4.50570941e-01
2.96873003e-01 -5.89710139e-02 -7.77330458e-01 2.43273422e-01
2.97727138e-01 1.30813968e+00 -7.96737015e-01 -2.70142823e-01
-2.18965694e-01 1.30344474e+00 -3.31071943e-01 3.94802451e-01
-4.19539660e-01 -4.01115447e-01 6.03948295e-01 8.34526792e-02
3.50976855e-01 -5.00104725e-01 -2.92005867e-01 -1.37974024e+00
2.30781093e-01 -1.09204888e+00 1.47248700e-01 -9.07601893e-01
-5.00768483e-01 5.58439553e-01 1.33033812e-01 -1.92946851e+00
-4.15563256e-01 -2.53412247e-01 -7.49866247e-01 6.75145149e-01
-1.37110472e+00 -1.10579908e+00 -4.71629232e-01 4.52552289e-01
5.41302383e-01 1.13547862e-01 4.62493181e-01 -4.91642244e-02
-3.19340415e-02 1.61882013e-01 4.19277579e-01 -5.01792252e-01
1.20954335e+00 -8.03490877e-01 4.30514604e-01 1.38679075e+00
2.00202037e-02 5.37651479e-01 1.00375509e+00 -6.22401536e-01
-1.28173113e+00 -4.91590142e-01 8.60378504e-01 -5.32377064e-02
4.87020046e-01 -9.36790928e-02 -7.51466691e-01 6.92590892e-01
5.59450626e-01 -3.51562172e-01 1.31954506e-01 -5.74007690e-01
-3.48502070e-01 -3.33382279e-01 -1.17043400e+00 7.16649830e-01
5.17786443e-01 -2.32840523e-01 -3.14147204e-01 -1.26926124e-01
3.49029213e-01 -9.12074625e-01 -5.54515362e-01 3.48937988e-01
9.05266285e-01 -1.38915634e+00 9.61703479e-01 6.14065826e-01
2.88754135e-01 -7.69255221e-01 1.77432373e-01 -1.49437892e+00
-2.72200257e-01 -1.25000274e+00 2.32560351e-01 1.07478952e+00
-1.55666679e-01 -9.25778925e-01 4.35777098e-01 2.24849582e-01
-1.56201525e-02 -3.74443345e-02 -7.99913168e-01 -6.56556308e-01
-7.52472401e-01 2.53883064e-01 4.42339063e-01 1.11135077e+00
-4.24025096e-02 1.60175070e-01 -8.91882837e-01 4.76410508e-01
9.75995362e-01 4.63904142e-01 1.13937938e+00 -9.92386043e-01
-3.46360177e-01 -2.00202391e-01 -5.01797259e-01 -1.06571567e+00
-4.14466530e-01 -8.41686502e-02 1.31902292e-01 -9.64780092e-01
-5.31669110e-02 -4.87449057e-02 4.56867456e-01 2.53889356e-02
-1.60523832e-01 4.09170717e-01 3.62608284e-01 5.68804681e-01
3.71284485e-01 2.99178749e-01 1.41851377e+00 1.17248811e-01
-4.05651897e-01 4.48035896e-02 -3.19850534e-01 7.45397627e-01
6.00371957e-01 -3.24535370e-01 -3.09469074e-01 -5.57245076e-01
-2.60027081e-01 2.98696339e-01 1.99832380e-01 -1.13660121e+00
3.33373696e-01 -3.70525330e-01 1.55425057e-01 -7.70362437e-01
6.05662346e-01 -1.17035580e+00 7.09506691e-01 4.79220152e-01
6.66275248e-02 1.57224581e-01 4.01460268e-02 3.31768364e-01
-3.57242584e-01 -4.00913000e-01 1.20189357e+00 8.64508227e-02
-3.36277723e-01 -3.21147591e-03 -3.22337225e-02 -5.68138659e-01
1.22642076e+00 -5.54811060e-01 -2.75016338e-01 -6.81438267e-01
8.28488842e-02 -1.45511717e-01 1.42693102e+00 2.41944239e-01
5.31663835e-01 -1.26716030e+00 -3.55610281e-01 4.99619812e-01
2.46031433e-02 -1.04981698e-01 4.04145837e-01 9.27623749e-01
-1.17583597e+00 2.02182174e-01 -3.23855430e-01 -7.86668301e-01
-1.77916610e+00 4.83938932e-01 3.50603908e-01 -2.36938238e-01
-7.06310928e-01 2.21945226e-01 2.34999880e-01 -4.46122847e-02
-1.32989600e-01 -1.83674932e-01 1.00400507e-01 -2.81595737e-01
6.49763167e-01 4.78754044e-01 7.65243769e-02 -8.88134778e-01
-2.17184469e-01 1.49371481e+00 2.30023172e-02 -6.12016141e-01
1.02276957e+00 -3.72801870e-01 -2.76885062e-01 1.27951056e-03
1.14408290e+00 5.93599021e-01 -1.46210229e+00 -2.80409604e-01
-2.73448139e-01 -1.25759947e+00 9.08838883e-02 -3.33030194e-01
-1.18009090e+00 6.99305058e-01 4.79104608e-01 5.30823208e-02
1.42571557e+00 -5.39795399e-01 8.54570091e-01 -6.84483349e-02
2.48962954e-01 -8.66938293e-01 4.85014021e-02 3.00126374e-01
7.97946751e-01 -7.67800987e-01 5.92636406e-01 -6.98588669e-01
-5.09782314e-01 1.42330313e+00 2.32423499e-01 -2.60340065e-01
1.47590354e-01 2.10217074e-01 3.95264894e-01 -2.70331353e-02
-3.41507345e-01 1.35059044e-01 3.45347315e-01 5.32149613e-01
1.13359265e-01 -2.87866026e-01 -4.06782180e-01 -8.12590539e-01
-4.90065701e-02 -1.33693919e-01 8.13461840e-01 1.06008899e+00
-3.49354357e-01 -1.36836088e+00 -1.15295458e+00 -1.60229325e-01
-4.58631665e-02 3.53108197e-02 -4.11278337e-01 7.52915382e-01
-2.14629129e-01 8.68260682e-01 -6.42865822e-02 -3.11256468e-01
3.32051247e-01 -3.08415502e-01 6.33325517e-01 -1.62015930e-01
-4.42577511e-01 7.08110511e-01 5.20417243e-02 -6.14310801e-01
-5.23247957e-01 -8.18432391e-01 -6.66092634e-01 -2.94868380e-01
-7.04237640e-01 -2.48661280e-01 6.64519012e-01 6.86786234e-01
2.65056461e-01 -7.02727586e-02 1.03167808e+00 -1.36444581e+00
-2.59780020e-01 -6.08173072e-01 -4.65798765e-01 4.79788214e-01
6.74692810e-01 -5.44339418e-01 -5.36134481e-01 3.35564375e-01] | [9.356209754943848, -2.3856749534606934] |
4362f88f-d613-4b98-ac6e-81a48f4b7cbe | langevin-thompson-sampling-with-logarithmic | 2306.08803 | null | https://arxiv.org/abs/2306.08803v1 | https://arxiv.org/pdf/2306.08803v1.pdf | Langevin Thompson Sampling with Logarithmic Communication: Bandits and Reinforcement Learning | Thompson sampling (TS) is widely used in sequential decision making due to its ease of use and appealing empirical performance. However, many existing analytical and empirical results for TS rely on restrictive assumptions on reward distributions, such as belonging to conjugate families, which limits their applicability in realistic scenarios. Moreover, sequential decision making problems are often carried out in a batched manner, either due to the inherent nature of the problem or to serve the purpose of reducing communication and computation costs. In this work, we jointly study these problems in two popular settings, namely, stochastic multi-armed bandits (MABs) and infinite-horizon reinforcement learning (RL), where TS is used to learn the unknown reward distributions and transition dynamics, respectively. We propose batched $\textit{Langevin Thompson Sampling}$ algorithms that leverage MCMC methods to sample from approximate posteriors with only logarithmic communication costs in terms of batches. Our algorithms are computationally efficient and maintain the same order-optimal regret guarantees of $\mathcal{O}(\log T)$ for stochastic MABs, and $\mathcal{O}(\sqrt{T})$ for RL. We complement our theoretical findings with experimental results. | ['Siddharth Mitra', 'Yi-An Ma', 'Nikki Lijing Kuang', 'Amin Karbasi'] | 2023-06-15 | null | null | null | null | ['thompson-sampling', 'multi-armed-bandits'] | ['methodology', 'miscellaneous'] | [ 1.54462561e-01 -1.25959694e-01 -4.50614274e-01 -3.32277566e-01
-1.01556814e+00 -5.68109870e-01 8.45655799e-02 3.01228702e-01
-5.89099646e-01 1.15514386e+00 -4.05883551e-01 -7.66203880e-01
-4.76956606e-01 -7.76310146e-01 -8.44517469e-01 -7.83540487e-01
-6.47913963e-02 7.67141640e-01 4.04327810e-02 7.28960186e-02
2.41915852e-01 3.06309253e-01 -1.20317030e+00 -2.62138665e-01
1.13559139e+00 1.36593294e+00 1.49369568e-01 4.86043423e-01
-7.56995901e-02 1.07063794e+00 -2.86589473e-01 -5.69346547e-01
3.09378564e-01 -7.08866835e-01 -5.42409480e-01 1.38941973e-01
-4.29463506e-01 -7.92754114e-01 -1.77552372e-01 1.18457067e+00
2.88619816e-01 4.01750296e-01 6.22042358e-01 -1.02099645e+00
-1.90510631e-01 7.42128253e-01 -1.10712552e+00 2.57282406e-01
-2.37825140e-02 8.24370161e-02 1.17836976e+00 -1.03942968e-01
5.14493212e-02 1.22448230e+00 2.93895602e-01 3.22783232e-01
-1.30295885e+00 -6.31751001e-01 5.43892026e-01 2.08280995e-01
-1.03587055e+00 -3.96019042e-01 4.30048734e-01 -1.61182597e-01
4.57956374e-01 3.44004959e-01 5.32059968e-01 7.27681816e-01
5.56833930e-02 1.24991810e+00 1.20305538e+00 -4.91937906e-01
7.69287884e-01 -8.92879367e-02 -7.00743347e-02 3.86887521e-01
3.85810882e-01 1.88933499e-02 -2.36841664e-01 -3.53192151e-01
9.16009724e-01 3.92089307e-01 5.65962680e-02 -2.37402648e-01
-8.00844252e-01 1.11950433e+00 -1.09687209e-01 -3.35547149e-01
-6.41945899e-01 4.54710990e-01 2.92170137e-01 4.76608783e-01
6.37363851e-01 -1.41286969e-01 -2.45790988e-01 -4.56414789e-01
-9.27859962e-01 4.61717486e-01 7.15720057e-01 1.08418334e+00
4.67492700e-01 6.80004656e-02 -4.64304954e-01 6.90172672e-01
2.04989076e-01 7.94671476e-01 1.68571591e-01 -1.19968522e+00
8.17137718e-01 -5.86187392e-02 9.67182994e-01 -4.50932384e-01
-1.68028712e-01 -4.75292265e-01 -7.06399798e-01 -5.00096418e-02
7.40282178e-01 -4.03696597e-01 -4.66804266e-01 1.58183706e+00
4.83513355e-01 -2.08521843e-01 -2.58541077e-01 8.77357185e-01
-3.95429760e-01 5.74678659e-01 -2.68763214e-01 -6.29252136e-01
1.06077051e+00 -8.28074694e-01 -4.76815015e-01 -2.27966771e-01
6.10747159e-01 -6.44890249e-01 8.63916874e-01 5.84862530e-01
-1.37107134e+00 6.54530972e-02 -6.93388820e-01 4.49032724e-01
4.80986863e-01 2.81728036e-03 7.93953776e-01 8.49618852e-01
-4.52988178e-01 7.88376391e-01 -1.12812853e+00 2.29953036e-01
5.22523761e-01 2.03646034e-01 6.82900727e-01 -3.05611670e-01
-6.98185325e-01 3.46592695e-01 5.20572513e-02 2.87023634e-01
-1.06219912e+00 -2.31602982e-01 -3.35803866e-01 1.47510082e-01
1.11945927e+00 -4.11188871e-01 1.73841953e+00 -7.50405729e-01
-1.90200245e+00 1.09050393e-01 -1.90491974e-01 -6.72164857e-01
1.06036830e+00 -1.72332451e-01 7.85279945e-02 2.06936330e-01
9.26146284e-02 -2.41058096e-01 9.08033431e-01 -6.14043713e-01
-8.51137757e-01 -5.47138274e-01 3.85234654e-01 1.53251305e-01
-1.65716425e-01 -9.58968922e-02 -1.27375469e-01 -5.67706943e-01
5.55751659e-03 -1.01463735e+00 -5.55791676e-01 -3.59442383e-01
-3.38887900e-01 -1.22261330e-01 -9.88914222e-02 -4.55211252e-01
1.15005445e+00 -1.81877875e+00 -1.05366237e-01 2.60418475e-01
-7.31011182e-02 -1.23545617e-01 2.45354295e-01 5.80120802e-01
6.93894804e-01 -1.22940473e-01 -1.24165304e-01 -3.51107359e-01
3.02939624e-01 1.87778845e-01 -5.23635387e-01 6.45784497e-01
-4.51700807e-01 6.20018840e-01 -9.90878046e-01 -3.20883989e-01
-8.39590933e-03 -3.43202829e-01 -4.99754220e-01 -3.98943434e-03
-6.74596369e-01 4.88410085e-01 -9.82699573e-01 4.81303930e-01
5.03291607e-01 -6.31843686e-01 5.58408797e-01 4.77234572e-01
-4.71597016e-02 1.41141355e-01 -1.36175823e+00 1.15305781e+00
-6.82245076e-01 -2.81632487e-02 2.54543930e-01 -1.25459456e+00
4.47773486e-01 8.72880742e-02 3.85316133e-01 -6.87229156e-01
2.70056248e-01 3.57959956e-01 -2.30449304e-01 -1.33318990e-01
3.89506936e-01 -3.48317444e-01 -7.49648511e-02 1.00780439e+00
-6.15122437e-01 5.46223186e-02 3.39245290e-01 -5.00320755e-02
9.81841505e-01 2.26981774e-01 3.22151989e-01 -9.48292390e-02
-7.93912038e-02 -1.31667897e-01 7.46294081e-01 1.19064701e+00
-1.14787623e-01 -1.64862871e-02 8.20408285e-01 -1.64134383e-01
-9.13228393e-01 -8.82145226e-01 1.25561655e-01 1.26372385e+00
2.84116298e-01 2.02653974e-01 -5.03709137e-01 -4.30103153e-01
2.22903296e-01 9.47515428e-01 -5.55796623e-01 2.42944181e-01
-1.69264838e-01 -8.71419549e-01 2.76142880e-02 4.20818418e-01
3.45420361e-01 -5.53369045e-01 -9.00411665e-01 5.35099506e-01
-1.51885282e-02 -8.41391146e-01 -4.14876133e-01 1.15946144e-01
-1.13257527e+00 -8.36132348e-01 -9.89536464e-01 1.00083984e-01
5.98456562e-01 5.32671094e-01 5.83828390e-01 -4.27704722e-01
8.44308585e-02 3.26778531e-01 -4.43804443e-01 -5.45915902e-01
-2.65477709e-02 -2.49160584e-02 -1.40053509e-02 2.44102031e-01
1.12167805e-01 -4.15506810e-01 -8.56061339e-01 4.12212878e-01
-1.01506221e+00 6.58442080e-02 6.50924563e-01 8.92084837e-01
7.39426613e-01 5.19007165e-03 4.95964825e-01 -9.09017980e-01
5.46237051e-01 -4.19481128e-01 -1.24956274e+00 4.19697493e-01
-5.30791879e-01 2.97362417e-01 7.64156461e-01 -4.83396560e-01
-1.05214596e+00 -3.48424733e-01 3.53210270e-01 -5.34211457e-01
3.46056134e-01 4.69466925e-01 4.33155239e-01 1.64161190e-01
1.57603502e-01 4.18812424e-01 7.30414391e-02 -4.00770724e-01
6.15378506e-02 5.50252438e-01 -6.03784844e-02 -9.94744837e-01
3.22566867e-01 5.97284019e-01 2.11393073e-01 -4.24834579e-01
-1.29649878e+00 -7.26171732e-02 1.52406752e-01 -2.56606974e-02
2.31554553e-01 -6.91695750e-01 -1.36302865e+00 2.26157501e-01
-5.82346857e-01 -5.57429552e-01 -3.08949947e-01 8.00172150e-01
-9.90307689e-01 4.83479023e-01 -6.18710995e-01 -1.74532676e+00
-2.68124521e-01 -9.76471841e-01 6.73091054e-01 1.90795779e-01
1.92469999e-01 -6.27312422e-01 -2.24639818e-01 6.58170462e-01
3.34247768e-01 1.06922857e-01 9.09311116e-01 -3.69591236e-01
-7.30513632e-01 -2.05527365e-01 -2.50972301e-01 1.89976245e-01
6.94446191e-02 -4.45051342e-01 -4.35517848e-01 -6.01418674e-01
2.73570456e-02 -5.41620076e-01 5.84355116e-01 6.62871242e-01
1.32830822e+00 -6.25637650e-01 -1.35878235e-01 9.82299224e-02
1.24164414e+00 5.18238246e-01 2.50570357e-01 1.90147087e-01
8.60004965e-03 4.53345776e-01 9.71044660e-01 1.36840343e+00
2.41855651e-01 5.09380341e-01 5.24685621e-01 5.82562804e-01
8.25271904e-01 -2.40964204e-01 3.06665480e-01 4.58530366e-01
-1.54784754e-01 -2.29441434e-01 -6.19477630e-01 5.58675885e-01
-2.20151758e+00 -9.12105978e-01 2.40514502e-01 2.79866505e+00
9.92703021e-01 3.18024099e-01 5.39768636e-01 -3.34136523e-02
7.39868164e-01 -9.61646885e-02 -1.14800417e+00 -3.57891798e-01
2.21673429e-01 2.48489529e-01 8.71261775e-01 3.16357493e-01
-6.37180030e-01 6.51667118e-01 4.97761202e+00 1.36221468e+00
-8.47351074e-01 1.94987610e-01 1.03117275e+00 -8.66974890e-01
-2.13051692e-01 1.28614888e-01 -7.46041298e-01 7.04914391e-01
1.03644145e+00 -1.78602904e-01 8.46989870e-01 7.80290842e-01
4.53665078e-01 -5.79363167e-01 -1.02914524e+00 1.02568257e+00
-6.80975974e-01 -1.17911470e+00 -3.33290845e-01 2.89402306e-01
8.30426574e-01 -1.21647678e-01 1.42842531e-01 3.39222729e-01
1.06229877e+00 -7.79313385e-01 9.41743076e-01 4.49167281e-01
6.67634189e-01 -1.27392066e+00 5.82513511e-01 7.14144349e-01
-9.10925090e-01 -4.37309086e-01 -4.57812816e-01 -3.08183074e-01
2.13422373e-01 9.01857495e-01 -3.89104933e-01 6.22585475e-01
6.77753389e-01 1.64082438e-01 3.38951647e-01 9.31183159e-01
-1.11050792e-01 7.79234886e-01 -7.09228277e-01 -6.11684203e-01
5.67039132e-01 -6.74019337e-01 2.74748564e-01 5.20653844e-01
3.66412103e-01 1.90251887e-01 4.22174245e-01 6.14361346e-01
5.62782772e-02 1.97371826e-01 4.55779508e-02 -2.79481262e-01
6.21607304e-01 7.73164988e-01 -9.83618081e-01 -2.67018408e-01
-1.76113069e-01 9.21426713e-01 2.67779708e-01 3.59270155e-01
-1.00593138e+00 -4.95311022e-01 4.41450536e-01 -1.38121443e-02
8.76642287e-01 -2.99625278e-01 -2.95926958e-01 -1.05806923e+00
3.81398141e-01 -5.82786679e-01 2.95449823e-01 -8.01558867e-02
-1.12929749e+00 -5.21402843e-02 3.03794555e-02 -1.11316085e+00
-4.77322370e-01 -1.08917736e-01 -2.06119642e-01 6.05404198e-01
-1.56568468e+00 -5.49996912e-01 3.03181857e-01 3.97055715e-01
3.68819058e-01 2.50312716e-01 3.25942367e-01 3.02667879e-02
-8.11792850e-01 5.65291941e-01 1.19677734e+00 -2.72569507e-01
1.51684433e-01 -9.84039307e-01 -9.69377160e-02 4.89316851e-01
-2.48233646e-01 2.31986374e-01 8.29535663e-01 -3.73034954e-01
-1.69772565e+00 -7.56877601e-01 1.82992876e-01 2.63237864e-01
8.95484090e-01 -1.79897904e-01 -2.50917584e-01 4.64583784e-01
-2.61121809e-01 -1.17124826e-01 6.31733239e-01 1.41894773e-01
-1.92624666e-02 -3.65761250e-01 -1.13906205e+00 7.36035764e-01
8.67994905e-01 -1.38648272e-01 6.25899732e-02 4.94729906e-01
4.84902322e-01 -2.47927293e-01 -8.96480680e-01 -4.37071845e-02
6.10972643e-01 -9.76010859e-01 4.69082385e-01 -5.58606505e-01
2.16992393e-01 1.32028356e-01 -3.11462671e-01 -1.07365823e+00
4.98119406e-02 -1.18966877e+00 -3.66582155e-01 8.66356909e-01
3.44569027e-01 -8.70558262e-01 7.71220088e-01 7.00746596e-01
4.41542566e-01 -9.16445434e-01 -1.31198025e+00 -1.07589900e+00
7.04472363e-02 -5.21573842e-01 6.55963659e-01 4.04693395e-01
-2.76307642e-01 2.55514327e-02 -6.44410670e-01 -8.07207748e-02
9.47595298e-01 7.96471357e-01 5.93564928e-01 -9.28962767e-01
-9.89385664e-01 -3.76300603e-01 3.00978988e-01 -1.55010712e+00
-2.13630319e-01 -2.88257152e-01 3.84362079e-02 -1.22493935e+00
2.48633295e-01 -9.66204524e-01 -4.26672816e-01 2.22163245e-01
3.48260924e-02 -2.96106040e-01 2.03903973e-01 3.56213123e-01
-1.05759645e+00 1.04433274e+00 1.08614671e+00 9.76838917e-02
-1.72897935e-01 6.55733049e-01 -6.66888833e-01 5.30145466e-01
7.17886090e-01 -5.67024708e-01 -5.26381373e-01 -5.64252257e-01
5.92157841e-01 8.79545510e-01 2.66540293e-02 -5.54052770e-01
-7.46752620e-02 -7.87229359e-01 -3.65882553e-02 -7.01982677e-01
3.02465171e-01 -5.15732348e-01 -1.08010247e-01 6.43632114e-01
-3.80341679e-01 -1.29219785e-01 -1.29074693e-01 1.17165709e+00
2.39274591e-01 -5.63005388e-01 7.03407526e-01 -1.65941149e-01
1.81566253e-01 4.16304737e-01 -5.48178673e-01 2.05106914e-01
1.18299723e+00 -6.64884271e-03 5.86392842e-02 -7.97629237e-01
-5.07524133e-01 5.67012072e-01 1.53547511e-01 -3.12072903e-01
2.88451076e-01 -9.69487667e-01 -5.46687901e-01 -4.83471513e-01
-2.85285085e-01 1.99762538e-01 5.21490812e-01 1.24373686e+00
-2.78702378e-01 5.92929959e-01 1.97678611e-01 -3.37853551e-01
-6.24180734e-01 5.25460780e-01 -2.72404235e-02 -6.51041269e-01
-2.93004721e-01 6.70920849e-01 -9.84361582e-03 2.82231700e-02
2.68589526e-01 -4.27283078e-01 3.91464114e-01 7.53343925e-02
5.46752155e-01 9.57504153e-01 -1.26427814e-01 3.35767478e-01
3.75428461e-02 -7.11816698e-02 -4.59270030e-01 -4.01169837e-01
1.36250746e+00 -3.87859792e-01 9.52121541e-02 5.34794569e-01
6.75930977e-01 -1.63985491e-01 -1.71193933e+00 -7.16301203e-01
6.76839575e-02 -5.46238422e-01 1.04917869e-01 -4.26118851e-01
-8.82976949e-01 7.10110843e-01 2.78130144e-01 4.50751752e-01
9.35985923e-01 -2.78097540e-01 8.77316475e-01 5.74396729e-01
9.65601087e-01 -1.38576674e+00 -2.09288582e-01 2.61244297e-01
2.53048271e-01 -9.39549029e-01 1.60046250e-01 1.66952163e-02
-6.82497919e-01 8.88944805e-01 6.33821962e-03 -9.72032771e-02
3.31014037e-01 -5.15727922e-02 -6.08466327e-01 3.47966164e-01
-7.93063045e-01 -1.77066579e-01 -3.70113373e-01 3.27696614e-02
6.21692929e-03 4.52523708e-01 -4.75406259e-01 6.70603395e-01
1.02820776e-01 1.64200872e-01 5.23007691e-01 1.35522127e+00
-4.67963785e-01 -1.05265343e+00 -3.75480175e-01 8.80042255e-01
-8.34498942e-01 3.42219211e-02 2.21403390e-01 3.27988327e-01
-7.12000787e-01 1.15836704e+00 -5.00526279e-02 2.70948142e-01
-6.14020713e-02 -2.73797005e-01 8.20338666e-01 -2.49761239e-01
-1.46613449e-01 3.62501055e-01 7.33130947e-02 -4.47870523e-01
-1.75575793e-01 -9.14403558e-01 -1.02651489e+00 -6.49381697e-01
-3.27224135e-01 4.39301193e-01 5.18621981e-01 1.31435072e+00
2.98066407e-01 1.78958073e-01 1.08168268e+00 -6.35532618e-01
-1.54345131e+00 -8.48486483e-01 -9.56742823e-01 -1.19476862e-01
8.72099027e-02 -7.88068354e-01 -1.57699035e-03 -5.49749136e-01] | [4.48383903503418, 3.1819210052490234] |
58dae5fd-0e80-4328-9a2b-2370ae08a788 | dp-util-comprehensive-utility-analysis-of | 2112.12998 | null | https://arxiv.org/abs/2112.12998v1 | https://arxiv.org/pdf/2112.12998v1.pdf | DP-UTIL: Comprehensive Utility Analysis of Differential Privacy in Machine Learning | Differential Privacy (DP) has emerged as a rigorous formalism to reason about quantifiable privacy leakage. In machine learning (ML), DP has been employed to limit inference/disclosure of training examples. Prior work leveraged DP across the ML pipeline, albeit in isolation, often focusing on mechanisms such as gradient perturbation. In this paper, we present, DP-UTIL, a holistic utility analysis framework of DP across the ML pipeline with focus on input perturbation, objective perturbation, gradient perturbation, output perturbation, and prediction perturbation. Given an ML task on privacy-sensitive data, DP-UTIL enables a ML privacy practitioner perform holistic comparative analysis on the impact of DP in these five perturbation spots, measured in terms of model utility loss, privacy leakage, and the number of truly revealed training samples. We evaluate DP-UTIL over classification tasks on vision, medical, and financial datasets, using two representative learning algorithms (logistic regression and deep neural network) against membership inference attack as a case study attack. One of the highlights of our results is that prediction perturbation consistently achieves the lowest utility loss on all models across all datasets. In logistic regression models, objective perturbation results in lowest privacy leakage compared to other perturbation techniques. For deep neural networks, gradient perturbation results in lowest privacy leakage. Moreover, our results on true revealed records suggest that as privacy leakage increases a differentially private model reveals more number of member samples. Overall, our findings suggest that to make informed decisions as to which perturbation mechanism to use, a ML privacy practitioner needs to examine the dynamics between optimization techniques (convex vs. non-convex), perturbation mechanisms, number of classes, and privacy budget. | ['Birhanu Eshete', 'Ismat Jarin'] | 2021-12-24 | null | null | null | null | ['membership-inference-attack'] | ['computer-vision'] | [ 4.70717162e-01 2.29700267e-01 -5.68071231e-02 -2.66070664e-01
-7.66849279e-01 -1.03730083e+00 4.38647389e-01 3.38110059e-01
-4.46857810e-01 6.23409033e-01 2.06858173e-01 -7.42352188e-01
-2.18684971e-01 -6.53506994e-01 -7.50729084e-01 -8.01057398e-01
-1.05016299e-01 -1.55603677e-01 -4.75859672e-01 2.61908203e-01
1.53165191e-01 7.29182303e-01 -9.50164437e-01 2.80069917e-01
4.75154042e-01 1.09573734e+00 -8.75475228e-01 4.29495364e-01
3.64609659e-01 7.82884181e-01 -6.81230962e-01 -1.01887548e+00
8.73813152e-01 -8.13951120e-02 -4.69362020e-01 -4.16595787e-01
5.59352040e-01 -6.56502306e-01 -4.54962015e-01 1.09290445e+00
8.73543143e-01 -7.09121600e-02 6.12563848e-01 -1.55696440e+00
-5.08001029e-01 7.13919461e-01 -5.32478988e-01 1.39716163e-01
1.43916786e-01 6.32467210e-01 7.54132807e-01 -2.63305426e-01
5.09919167e-01 1.08933151e+00 1.00687015e+00 6.60093844e-01
-1.81512392e+00 -8.75099063e-01 -1.00192443e-01 -3.23212653e-01
-1.21513665e+00 -5.84865153e-01 4.54885125e-01 -5.78005195e-01
7.51021266e-01 6.87268078e-01 1.70951605e-01 1.46699238e+00
2.54159898e-01 6.77111924e-01 1.13825548e+00 -1.40654251e-01
4.15388465e-01 7.31773436e-01 3.19293886e-01 1.26589596e-01
6.87557697e-01 3.81359845e-01 -4.53113973e-01 -9.82996106e-01
6.33521974e-02 1.84823405e-02 -5.41814327e-01 -4.54136878e-01
-3.03462625e-01 7.17183173e-01 2.07604002e-02 -3.48206908e-01
-2.53644675e-01 4.15583923e-02 7.98047125e-01 4.63803738e-01
3.95669878e-01 7.71367073e-01 -7.23292232e-01 4.74270731e-02
-5.96734047e-01 6.56462014e-01 1.10457408e+00 7.82645881e-01
4.89007801e-01 -3.70318383e-01 -8.39120150e-01 4.70310301e-01
-2.63550654e-02 1.20147355e-02 2.98658133e-01 -9.26853299e-01
7.34967649e-01 4.56250399e-01 6.83271214e-02 -1.29298973e+00
4.40153247e-03 -4.87927228e-01 -8.52445841e-01 3.43530506e-01
3.93580645e-01 -6.74934506e-01 -2.94671029e-01 2.01061416e+00
2.57410198e-01 -2.53640592e-01 3.66089970e-01 4.48215842e-01
3.05472851e-01 2.29144007e-01 2.10263491e-01 -1.69046685e-01
1.25271606e+00 -2.88605630e-01 -3.37832481e-01 2.11577443e-03
7.60955989e-01 -1.47310510e-01 1.11337745e+00 4.11767274e-01
-8.63503158e-01 1.63685635e-01 -1.02079439e+00 -1.23540379e-01
-4.60236073e-01 -3.07855695e-01 4.58150357e-01 1.26378596e+00
-8.18749011e-01 8.79867077e-01 -6.74790740e-01 -1.98994428e-01
1.02731371e+00 4.65211630e-01 -4.59332317e-01 2.95191677e-03
-1.07571781e+00 6.11512899e-01 1.85711071e-01 -2.43810728e-01
-5.30125439e-01 -1.44173563e+00 -5.60641885e-01 3.45724523e-01
5.91184571e-02 -8.81008148e-01 9.99935508e-01 -7.36504197e-01
-1.07364023e+00 9.61418808e-01 2.80317575e-01 -1.18191302e+00
1.34438086e+00 -2.00451136e-01 -5.39698154e-02 -2.85195649e-01
-3.71840209e-01 3.67046535e-01 5.47201037e-01 -1.11313426e+00
-5.02405465e-01 -7.24985242e-01 -5.40246740e-02 9.29548368e-02
-4.97809857e-01 -6.30302122e-03 9.74755436e-02 -4.44255978e-01
-4.64053124e-01 -7.58745313e-01 -2.54474699e-01 2.48799771e-01
-7.51941621e-01 2.03711346e-01 1.01495552e+00 -6.23467982e-01
1.36331356e+00 -2.48026395e+00 -3.51440847e-01 3.43681425e-01
3.46472412e-01 1.29855201e-01 2.62359023e-01 3.06440175e-01
2.94118356e-02 4.40727085e-01 -5.37473381e-01 -6.59790397e-01
3.41292202e-01 3.01800128e-02 -6.59810543e-01 7.62112200e-01
-3.79877426e-02 7.32444286e-01 -3.24938923e-01 -4.13374603e-02
-1.00494720e-01 5.07269621e-01 -8.70507419e-01 1.09795965e-01
-1.19012691e-01 1.79281384e-01 -2.50769317e-01 6.74278378e-01
1.08108234e+00 4.32037190e-02 1.08752437e-01 -1.50354812e-02
1.58694148e-01 2.87083015e-02 -7.15779066e-01 1.06149399e+00
-5.78097142e-02 7.03050017e-01 6.40939176e-02 -5.25937438e-01
6.64620101e-01 2.28319541e-01 3.20215613e-01 -3.59000146e-01
2.53007144e-01 -1.60485074e-01 -1.51877120e-01 -2.55837470e-01
3.20729107e-01 1.53555498e-01 -1.82832897e-01 4.72853690e-01
-3.75745088e-01 3.75330240e-01 -4.13865119e-01 -2.01917868e-02
1.44060421e+00 -3.82303208e-01 4.19188350e-01 -2.83666104e-01
2.66004741e-01 -2.08446547e-01 6.14580452e-01 1.01293170e+00
-5.96982181e-01 5.86774707e-01 1.03908813e+00 -3.07342470e-01
-9.22674954e-01 -8.37343633e-01 -3.92030865e-01 8.15910161e-01
-1.56834424e-01 -2.10049212e-01 -9.86304283e-01 -8.06844413e-01
7.93440461e-01 9.07548130e-01 -7.45251894e-01 -4.53624129e-01
-2.15344593e-01 -9.73691046e-01 1.21125364e+00 -9.43237618e-02
4.71680909e-01 -7.95999646e-01 -8.87107372e-01 -2.57968634e-01
3.89445782e-01 -9.66701806e-01 -5.16477108e-01 1.51576251e-01
-7.40563095e-01 -1.04870880e+00 -2.56876141e-01 1.14832528e-01
6.12490952e-01 -3.95296752e-01 7.12180257e-01 -4.64381814e-01
-3.85277152e-01 6.12669706e-01 1.21848643e-01 -7.35766232e-01
-4.90059376e-01 2.16553058e-03 -2.68256143e-02 1.53539076e-01
5.87202668e-01 -7.53178120e-01 -7.86137819e-01 -9.18641239e-02
-8.68914902e-01 -5.95249057e-01 4.10635322e-01 4.71905857e-01
4.98184383e-01 1.24944011e-02 -2.32872590e-02 -1.32573652e+00
1.21668947e+00 -8.13762009e-01 -4.98140484e-01 2.67483413e-01
-1.04067886e+00 2.31541134e-02 6.78442955e-01 -5.91898263e-01
-7.46914446e-01 -1.71598122e-01 1.90506727e-01 -7.26270497e-01
-1.55438453e-01 1.05252869e-01 -4.34451133e-01 -2.03773052e-01
9.57892239e-01 2.40120932e-01 2.35702187e-01 -4.81101513e-01
2.29362160e-01 6.28124774e-01 4.14842188e-01 -6.73634589e-01
3.91388953e-01 3.63474786e-01 3.26589614e-01 -4.88400608e-01
-2.88065434e-01 6.61482438e-02 -6.36690035e-02 3.76524299e-01
3.43112290e-01 -6.90740645e-01 -1.20564663e+00 4.93548691e-01
-7.80813277e-01 -1.22308053e-01 -5.70938230e-01 4.55057286e-02
-4.61155355e-01 5.12131274e-01 -3.66902500e-01 -1.01478803e+00
-7.79183149e-01 -1.05468404e+00 6.02341175e-01 -2.80969553e-02
-4.94571835e-01 -8.92824948e-01 -1.61502510e-01 2.14298829e-01
5.15250862e-01 1.02384877e+00 9.03155267e-01 -1.13411570e+00
-3.68346334e-01 -5.52448809e-01 -1.21491991e-01 5.79253793e-01
3.98267023e-02 -8.59225094e-02 -1.28920603e+00 -5.36668539e-01
3.42345864e-01 -1.11684941e-01 5.52770317e-01 3.09428155e-01
1.62748682e+00 -1.32342887e+00 -1.78537697e-01 1.23609161e+00
1.42212749e+00 7.63397068e-02 5.28740644e-01 3.89667779e-01
3.98187369e-01 7.14148402e-01 9.03141350e-02 7.51774669e-01
6.67427853e-02 2.75286227e-01 4.45945650e-01 1.53807953e-01
5.42968631e-01 -4.76985365e-01 5.03457725e-01 -5.05808473e-01
6.53402328e-01 -2.80066460e-01 -7.34897256e-01 1.66721627e-01
-1.54817200e+00 -9.02345240e-01 3.40262383e-01 2.41728210e+00
9.09429848e-01 3.74036878e-02 3.64511222e-01 8.79811589e-03
4.39276338e-01 1.64825018e-04 -1.03880632e+00 -8.51817787e-01
-2.27587909e-01 -1.22464985e-01 1.12207985e+00 1.15312621e-01
-1.04084706e+00 4.09122825e-01 6.14281893e+00 6.22295797e-01
-1.17470968e+00 -1.55580062e-02 1.27503538e+00 -6.87613487e-01
-2.96688378e-01 -2.26354420e-01 -7.36844957e-01 7.65698195e-01
9.29220736e-01 -6.33637369e-01 4.42332000e-01 1.09942961e+00
-3.23117152e-02 2.64701575e-01 -1.62734246e+00 9.78716791e-01
-3.34180593e-01 -1.26688850e+00 1.22704051e-01 5.64781666e-01
3.65528911e-01 -4.01008725e-02 7.76137590e-01 1.88663483e-01
4.89757866e-01 -1.10637093e+00 4.74192917e-01 4.59730566e-01
8.50264370e-01 -9.97203529e-01 5.40533006e-01 4.66623813e-01
-1.78454474e-01 -4.45596367e-01 -2.67076433e-01 1.25479400e-01
-3.40090901e-01 5.60938418e-01 -9.85797524e-01 3.13515455e-01
8.10803473e-01 2.10976601e-01 -2.20606774e-01 5.11419833e-01
2.50778019e-01 6.52653396e-01 -4.91440296e-01 2.63365716e-01
-5.18269762e-02 -8.98471624e-02 7.29654193e-01 1.45045733e+00
1.07151844e-01 1.53385609e-01 -2.99577653e-01 1.16156042e+00
-3.16218376e-01 -5.38846515e-02 -8.80730331e-01 -2.63597202e-02
7.84668505e-01 9.04670238e-01 1.46413684e-01 2.50089318e-01
1.57209709e-02 9.18229938e-01 2.01240748e-01 3.18268478e-01
-4.41259623e-01 -1.70186013e-01 1.45645249e+00 2.19564527e-01
-1.46391049e-01 3.82954746e-01 -8.14441204e-01 -7.49276698e-01
3.04386735e-01 -1.04705548e+00 7.26029694e-01 -8.37615598e-03
-1.57790565e+00 1.00935772e-01 9.70205851e-03 -8.54636431e-01
-1.10759549e-01 -4.14802223e-01 -5.37014723e-01 1.20875144e+00
-1.09370482e+00 -7.87239432e-01 2.03116074e-01 4.72121954e-01
-3.29192668e-01 -3.03337872e-01 7.42946506e-01 7.89226219e-02
-8.29055429e-01 1.67573285e+00 3.31224233e-01 2.52293140e-01
4.16487068e-01 -1.03331232e+00 4.83225465e-01 9.23094749e-01
-3.03468138e-01 1.10842884e+00 7.76716232e-01 -4.00218695e-01
-1.48833966e+00 -1.18828261e+00 4.36852723e-01 -8.09623003e-01
2.70809025e-01 -3.87929171e-01 -9.18623924e-01 7.81440377e-01
-1.42147034e-01 1.08377874e-01 1.01288736e+00 -1.30176961e-01
-5.34963548e-01 -3.17039043e-01 -2.21945214e+00 6.69330060e-01
8.14617038e-01 -7.18171477e-01 6.48047030e-02 6.85219988e-02
1.03347743e+00 -5.56359172e-01 -1.03235698e+00 4.49343711e-01
7.77965486e-01 -1.25709724e+00 8.95345151e-01 -6.82014406e-01
2.38294140e-01 1.69415727e-01 -4.43460047e-01 -7.16856539e-01
-1.38068572e-01 -1.14428341e+00 -4.68489498e-01 1.46237111e+00
4.59755123e-01 -1.09236276e+00 1.16370022e+00 1.50724161e+00
5.13537407e-01 -1.10748363e+00 -1.00323820e+00 -5.06985903e-01
2.83356875e-01 -3.40776950e-01 9.88129795e-01 1.18014669e+00
-2.53359914e-01 -4.42728549e-01 -3.46660018e-01 3.74363214e-01
8.37013066e-01 -4.92373317e-01 9.65661228e-01 -7.97140241e-01
-5.63451231e-01 -5.11891544e-01 -4.52285349e-01 -3.60384166e-01
9.90694761e-02 -8.71708214e-01 -6.58590019e-01 -6.54188693e-01
9.14463103e-02 -3.09172213e-01 -4.06326652e-01 6.13755643e-01
-1.08704634e-01 -2.79968560e-01 2.71719813e-01 8.86433274e-02
2.63914019e-01 1.49958208e-01 5.52260697e-01 -9.54763442e-02
-4.31444675e-01 4.79180694e-01 -1.33767819e+00 2.90048957e-01
8.30508351e-01 -6.10417604e-01 -4.12431240e-01 -4.25857194e-02
1.18686013e-01 -1.73174515e-01 4.56146061e-01 -7.24159896e-01
3.72589141e-01 -7.50364363e-02 3.34336698e-01 -2.92390019e-01
1.01754637e-02 -1.07337713e+00 3.89199972e-01 5.11283159e-01
-6.27389967e-01 -3.88443545e-02 4.48955446e-01 5.84149957e-01
2.31988207e-01 3.09080482e-01 8.75954390e-01 3.33190300e-02
-2.52182838e-02 5.02764702e-01 1.62371829e-01 7.36232474e-02
1.18910491e+00 -5.96331537e-01 -3.75276923e-01 -1.26278969e-02
-6.46609604e-01 3.14439386e-01 5.92864931e-01 1.34127945e-01
3.07442009e-01 -8.16412508e-01 -6.60410047e-01 3.73513460e-01
1.01868942e-01 1.64031109e-03 2.60862678e-01 5.77164054e-01
-3.59003425e-01 2.42374301e-01 -4.16910537e-02 -3.35996777e-01
-1.51405466e+00 6.06832802e-01 6.63006186e-01 -2.62744248e-01
-5.88619530e-01 1.00322199e+00 1.72701567e-01 -5.41434765e-01
7.64987111e-01 -3.02322924e-01 4.34430748e-01 -3.64545435e-02
4.30696249e-01 7.60369599e-01 2.58643217e-02 -6.08080961e-02
-3.28421742e-01 -5.23621030e-02 -3.43234450e-01 5.25056086e-02
1.01264000e+00 8.86898562e-02 -1.29613504e-01 3.51776406e-02
1.58815575e+00 -2.94962302e-02 -1.53434229e+00 -1.07658297e-01
-2.34537534e-02 -5.58002293e-01 -1.93548173e-01 -9.06127214e-01
-8.11055601e-01 5.19758821e-01 7.92620122e-01 1.06135182e-01
1.26698661e+00 -5.56480527e-01 5.19338012e-01 3.08262765e-01
1.18568875e-01 -6.33898318e-01 -8.62246335e-01 1.53210647e-02
7.30255604e-01 -9.37378109e-01 2.79667735e-01 -2.98927296e-02
-7.13754117e-01 5.02883852e-01 5.04258215e-01 1.26108289e-01
7.60453045e-01 3.99183035e-01 -1.51628926e-01 1.27664357e-01
-8.36246371e-01 9.04670179e-01 2.34589074e-02 6.74258173e-01
-4.87402305e-02 9.90596265e-02 -8.83771181e-02 8.69164169e-01
-5.16890943e-01 -1.19124576e-01 3.42571735e-01 8.02844882e-01
2.74300128e-01 -9.50209737e-01 -2.34039307e-01 7.05624580e-01
-1.09643149e+00 -3.11929304e-02 -8.14200997e-01 6.93241954e-01
1.29921883e-01 6.28984213e-01 2.91259084e-02 -3.82326186e-01
5.37978828e-01 -4.65581194e-03 1.40447691e-01 -2.08605483e-01
-1.25281084e+00 -8.60069275e-01 -1.08459890e-01 -6.48336291e-01
3.83614540e-01 -8.94615173e-01 -6.65554523e-01 -8.23118627e-01
9.62193906e-02 -3.10303178e-02 5.98007917e-01 4.63166207e-01
9.46124554e-01 -1.21709984e-02 7.15194821e-01 -2.39464462e-01
-1.34541190e+00 -5.75045824e-01 -6.25154734e-01 4.41156387e-01
7.72966504e-01 8.65229070e-02 -7.84348249e-01 -4.85978663e-01] | [5.965097427368164, 6.969370365142822] |
543934bd-d0c9-44e0-8f50-8cb5450ff9ad | local-global-temporal-difference-learning-for | 2304.04421 | null | https://arxiv.org/abs/2304.04421v1 | https://arxiv.org/pdf/2304.04421v1.pdf | Local-Global Temporal Difference Learning for Satellite Video Super-Resolution | Optical-flow-based and kernel-based approaches have been widely explored for temporal compensation in satellite video super-resolution (VSR). However, these techniques involve high computational consumption and are prone to fail under complex motions. In this paper, we proposed to exploit the well-defined temporal difference for efficient and robust temporal compensation. To fully utilize the temporal information within frames, we separately modeled the short-term and long-term temporal discrepancy since they provide distinctive complementary properties. Specifically, a short-term temporal difference module is designed to extract local motion representations from residual maps between adjacent frames, which provides more clues for accurate texture representation. Meanwhile, the global dependency in the entire frame sequence is explored via long-term difference learning. The differences between forward and backward segments are incorporated and activated to modulate the temporal feature, resulting in holistic global compensation. Besides, we further proposed a difference compensation unit to enrich the interaction between the spatial distribution of the target frame and compensated results, which helps maintain spatial consistency while refining the features to avoid misalignment. Extensive objective and subjective evaluation of five mainstream satellite videos demonstrates that the proposed method performs favorably for satellite VSR. Code will be available at \url{https://github.com/XY-boy/TDMVSR} | ['Chia-Wen Lin', 'Liangpei Zhang', 'Jiang He', 'Xianyu Jin', 'Kui Jiang', 'Qiangqiang Yuan', 'Yi Xiao'] | 2023-04-10 | null | null | null | null | ['video-super-resolution'] | ['computer-vision'] | [ 5.55281714e-02 -5.69822073e-01 -3.52755934e-01 -3.10342669e-01
-5.85585296e-01 -3.03452700e-01 3.71009797e-01 -3.53449225e-01
-2.04511687e-01 6.38100445e-01 5.39690316e-01 3.10955554e-01
-1.65827021e-01 -4.67939496e-01 -3.80189985e-01 -9.38865721e-01
-1.65103450e-01 -6.46863997e-01 5.43568850e-01 -4.34076130e-01
4.26777214e-01 3.89770806e-01 -1.45961070e+00 1.89437315e-01
1.15518129e+00 1.00654995e+00 5.22185206e-01 3.98702115e-01
1.12593159e-01 8.52395892e-01 -1.53210506e-01 3.30015302e-01
2.25146055e-01 -4.03668582e-01 -5.92227757e-01 2.47558415e-01
3.95361662e-01 -6.26850069e-01 -8.39784920e-01 1.02968848e+00
5.24190068e-01 5.08638561e-01 1.91615745e-01 -6.83867693e-01
-5.29098511e-01 9.39779133e-02 -8.92263889e-01 9.01005447e-01
3.51189464e-01 5.33272982e-01 7.81000912e-01 -9.61169899e-01
5.32919347e-01 1.04753494e+00 3.77408147e-01 4.14874345e-01
-1.04735434e+00 -6.62485600e-01 2.09561884e-01 5.88241875e-01
-1.29741085e+00 -5.98403394e-01 8.92233551e-01 -4.39799041e-01
5.04235804e-01 2.23047167e-01 6.28862321e-01 6.55390561e-01
2.63314903e-01 6.28637135e-01 1.08065522e+00 -9.37742069e-02
-1.97060168e-01 -3.33373189e-01 -9.02382806e-02 5.65744162e-01
-3.33997309e-01 3.83539438e-01 -7.07554162e-01 2.41738811e-01
1.19922531e+00 -1.06013343e-02 -8.64702165e-01 9.76793002e-03
-1.32523751e+00 5.06080329e-01 5.14896333e-01 5.60232759e-01
-3.37079883e-01 -1.00357480e-01 2.25678876e-01 -7.00689554e-02
6.29272223e-01 4.94885333e-02 -2.65546978e-01 -2.14663614e-03
-1.04130065e+00 4.74300422e-02 -4.40455787e-02 4.62798834e-01
9.48658705e-01 5.08430600e-01 -4.65029210e-01 8.86575460e-01
2.06224769e-01 2.50384301e-01 6.43522680e-01 -1.32258940e+00
4.85536695e-01 2.11055115e-01 3.32805455e-01 -1.37020934e+00
-2.70991474e-01 -4.81249183e-01 -1.03598630e+00 1.18604288e-01
3.73678625e-01 1.54558480e-01 -5.52936435e-01 1.65268350e+00
4.95641291e-01 6.93443298e-01 -1.37031332e-01 1.48168504e+00
6.43273473e-01 7.17406213e-01 -1.94063887e-03 -4.77576405e-01
1.14639902e+00 -9.33017671e-01 -8.61485541e-01 -2.36739665e-01
3.45386088e-01 -8.45354497e-01 9.02466059e-01 -7.75417872e-03
-1.10919333e+00 -8.12081337e-01 -1.03299081e+00 -1.42588168e-01
2.70481229e-01 1.75462052e-01 3.70328754e-01 8.62060860e-02
-9.50852096e-01 7.46096790e-01 -8.88520539e-01 -6.79592369e-03
2.88967639e-01 1.11372024e-01 -2.65480489e-01 -1.61097050e-01
-1.41683888e+00 6.66639149e-01 2.20646724e-01 6.02125347e-01
-7.42967010e-01 -5.96669018e-01 -7.82171965e-01 -1.12853855e-01
2.36416906e-01 -3.24185908e-01 9.26238179e-01 -1.01601911e+00
-1.48076332e+00 4.54991728e-01 -5.54113507e-01 -8.47015977e-02
4.10850167e-01 -4.69720438e-02 -4.97012526e-01 4.05284047e-01
6.42097816e-02 4.25706536e-01 9.62400079e-01 -1.03105772e+00
-7.12571561e-01 -2.94750869e-01 -8.99041891e-02 6.14893973e-01
-3.41972619e-01 -2.11388431e-02 -6.27579749e-01 -1.20638561e+00
3.72747719e-01 -6.17990434e-01 -1.11166373e-01 9.67001319e-02
1.67428821e-01 2.31538370e-01 7.88983583e-01 -1.22090137e+00
1.61810684e+00 -2.32656479e+00 2.84645587e-01 -1.54069826e-01
1.45676717e-01 4.42506582e-01 -1.76971585e-01 -4.92687114e-02
-9.46157724e-02 -1.14290446e-01 -2.38107145e-01 1.33124486e-01
-5.93294144e-01 3.87047343e-02 -2.41583273e-01 6.74957991e-01
1.61541119e-01 5.98801255e-01 -9.66734648e-01 -5.87336302e-01
5.31159639e-01 7.39147425e-01 -4.67939466e-01 1.65836796e-01
1.18625484e-01 1.07327354e+00 -5.57936609e-01 5.81830800e-01
9.38114285e-01 -4.03959267e-02 3.55554093e-03 -6.12021565e-01
-4.20222700e-01 1.04829464e-02 -1.26173198e+00 1.69933164e+00
-3.66235465e-01 7.33659029e-01 1.95333123e-01 -8.31487358e-01
8.60114634e-01 1.48649260e-01 6.94835305e-01 -1.07466471e+00
-1.63437314e-02 2.57785946e-01 -2.53714710e-01 -6.81298912e-01
6.31447017e-01 4.67686839e-02 4.31579173e-01 1.74504835e-02
-2.92666972e-01 2.60057062e-01 4.80122007e-02 -1.18289381e-01
5.84205151e-01 4.53717500e-01 1.81156307e-01 -1.74314544e-01
1.03706360e+00 -2.96769530e-01 1.00466347e+00 1.36345908e-01
-4.93009269e-01 7.85363972e-01 1.73870996e-02 -4.56787229e-01
-8.52119923e-01 -7.62139022e-01 -3.19093019e-02 9.48391140e-01
7.60252178e-01 -2.15102270e-01 -4.36892003e-01 -1.33488417e-01
-3.70455742e-01 2.69771725e-01 -4.74608451e-01 -1.30823284e-01
-9.78242874e-01 -6.64778113e-01 2.25147679e-01 4.76035744e-01
8.17096174e-01 -7.40605772e-01 -4.22544867e-01 3.11791956e-01
-7.83371985e-01 -1.02627385e+00 -1.05210161e+00 -4.07419562e-01
-1.15047133e+00 -7.94967592e-01 -1.01717055e+00 -6.45006657e-01
3.67207289e-01 7.80664861e-01 6.45360053e-01 1.53159618e-01
-2.09670350e-01 -6.92325607e-02 -3.11751723e-01 6.27152264e-01
1.60287857e-01 -3.69692951e-01 -1.04304187e-01 2.99143761e-01
-6.75198287e-02 -6.76374257e-01 -1.09174073e+00 7.17523575e-01
-1.01730597e+00 1.10096909e-01 3.95522475e-01 1.06685889e+00
5.63181102e-01 1.62770483e-03 1.28547162e-01 -1.88935846e-01
1.08296961e-01 -3.70633096e-01 -4.93223220e-01 8.97808298e-02
-2.38714471e-01 -1.22321891e-02 3.74266595e-01 -4.56239939e-01
-1.43113589e+00 -1.58027872e-01 3.24970335e-02 -6.29680693e-01
1.19318359e-01 3.82261902e-01 -1.85746059e-01 -2.01525599e-01
3.41960132e-01 6.75223351e-01 4.93580066e-02 -3.75597209e-01
7.57775307e-02 4.50628638e-01 6.20140076e-01 -4.30252105e-01
6.71604812e-01 7.87963510e-01 -1.28103867e-01 -8.43919277e-01
-7.01892614e-01 -6.09661996e-01 -5.79990327e-01 -4.84706908e-01
9.12217081e-01 -1.23351955e+00 -5.09253025e-01 7.02789128e-01
-9.47554171e-01 -3.95934522e-01 3.58577035e-02 6.82180583e-01
-4.52603996e-01 7.75304437e-01 -8.07412386e-01 -5.64017773e-01
-2.03734964e-01 -1.11129212e+00 8.72226059e-01 6.85947597e-01
2.02893555e-01 -9.96396363e-01 -9.98513773e-03 3.59356254e-01
5.73906243e-01 3.57439555e-02 3.70140761e-01 1.80655509e-01
-8.72845232e-01 2.73552388e-01 -4.08155560e-01 3.62047404e-01
3.45430821e-01 6.84861317e-02 -6.50876522e-01 -3.80409002e-01
8.62938464e-02 1.15221828e-01 9.58693206e-01 6.28822923e-01
1.06730878e+00 -2.60375321e-01 -5.99613376e-02 1.00478220e+00
1.39131522e+00 1.49851635e-01 9.32421446e-01 6.02433324e-01
8.21699321e-01 7.31756032e-01 1.02088547e+00 6.23188794e-01
2.61070848e-01 1.00574374e+00 1.67092443e-01 -5.79294264e-02
-3.17074805e-01 9.99108888e-03 5.06638110e-01 7.22434759e-01
-5.75610399e-01 4.15557139e-02 -4.93134469e-01 4.63934004e-01
-1.89231539e+00 -1.29646730e+00 -1.14287995e-01 2.20845938e+00
9.32121158e-01 -1.63212165e-01 -1.86293215e-01 -4.24330775e-03
9.14074361e-01 7.07709014e-01 -4.15472656e-01 2.01226786e-01
-5.39556205e-01 -8.70873705e-02 4.62190479e-01 6.74267352e-01
-1.06807745e+00 8.59767735e-01 5.46339893e+00 1.09949386e+00
-1.37246752e+00 1.77311853e-01 6.46007776e-01 -7.55912438e-02
-3.39797080e-01 6.29466176e-02 -5.09268522e-01 6.49008989e-01
4.17729139e-01 -4.63898405e-02 4.02168304e-01 2.84773499e-01
8.72750103e-01 -2.72543073e-01 -2.82901168e-01 9.01348829e-01
-1.03432678e-01 -1.29532290e+00 -6.56821281e-02 -1.28742218e-01
7.63786197e-01 -1.92164376e-01 1.84431836e-01 -1.58208191e-01
-2.99672544e-01 -7.83731580e-01 6.93699062e-01 8.35487723e-01
8.08803856e-01 -6.29661500e-01 5.34904897e-01 6.77078888e-02
-1.71667361e+00 -1.28970608e-01 -2.37246573e-01 5.52335531e-02
2.59958863e-01 4.62702960e-01 4.04750556e-02 8.56478751e-01
8.96377981e-01 1.29394233e+00 -5.79495788e-01 9.70587313e-01
-1.19744837e-01 3.15325975e-01 5.21491133e-02 6.70186520e-01
1.29467711e-01 -4.53107387e-01 7.38261580e-01 1.04815125e+00
4.13843781e-01 4.99210715e-01 3.57288718e-02 4.45968211e-01
5.67954242e-01 -4.09307033e-02 -4.35657874e-02 1.71391755e-01
2.83509493e-01 1.31756079e+00 -4.90933120e-01 -2.24791273e-01
-5.10149658e-01 1.12539065e+00 6.40037805e-02 6.10109210e-01
-1.01713109e+00 -8.78185108e-02 7.90827930e-01 1.75645828e-01
3.92941445e-01 -4.33668911e-01 -1.22785181e-01 -1.58558238e+00
1.22790188e-01 -8.13900173e-01 4.22953606e-01 -8.81850541e-01
-8.95930111e-01 5.51562309e-01 -1.82550430e-01 -1.89365935e+00
-5.17784199e-03 -1.12539925e-01 -5.42906225e-01 1.08710229e+00
-1.72382104e+00 -1.03573763e+00 -6.78201795e-01 8.16505909e-01
7.49734759e-01 1.01885028e-01 1.32325783e-01 6.38087571e-01
-6.63364112e-01 4.18130010e-01 1.64088085e-01 4.97955047e-02
9.68908548e-01 -6.73220754e-01 -1.18725777e-01 1.18821859e+00
-3.60185027e-01 3.92666817e-01 7.31151760e-01 -7.00281322e-01
-1.15130758e+00 -9.74550962e-01 5.47106087e-01 1.36007532e-01
5.48772395e-01 3.04113269e-01 -1.36365521e+00 3.74642223e-01
1.76434312e-02 3.43595296e-01 1.63716823e-01 -6.65828884e-01
-2.84285486e-01 -3.20576847e-01 -8.06304157e-01 5.37453413e-01
8.83866966e-01 -5.91015458e-01 -2.29074121e-01 -1.57651544e-01
5.36835015e-01 -4.49753672e-01 -8.12050462e-01 5.61276138e-01
6.42167866e-01 -1.28585827e+00 1.04694045e+00 -6.30341917e-02
6.32853866e-01 -8.98527026e-01 -2.28455111e-01 -1.00348997e+00
-5.27655482e-01 -5.78333437e-01 -1.93450451e-01 1.31840086e+00
-1.61491841e-01 -5.77432811e-01 4.92356807e-01 3.96337926e-01
-1.69162005e-01 -6.69043899e-01 -8.38849485e-01 -5.27007818e-01
-3.50715399e-01 -9.84052420e-02 2.09348440e-01 1.02015877e+00
-2.58206218e-01 -1.76215544e-01 -8.41603935e-01 3.71878028e-01
7.17889488e-01 5.67060485e-02 3.66268188e-01 -5.66114247e-01
-3.31968158e-01 -4.36860621e-01 -3.48298401e-01 -1.28542721e+00
-4.26864922e-02 -4.47957635e-01 3.01332586e-02 -1.31859684e+00
1.13505103e-01 -2.33432218e-01 -4.48427111e-01 -6.17743880e-02
-5.57732522e-01 5.36722898e-01 8.26576427e-02 6.51923120e-01
-3.51890951e-01 8.52871716e-01 1.71944678e+00 2.67699808e-02
-3.24458420e-01 -1.36540219e-01 -2.89297521e-01 5.54366529e-01
7.54343867e-01 1.03623997e-02 -2.10724950e-01 -5.77243209e-01
-2.86203355e-01 5.57898343e-01 4.23712522e-01 -8.69577825e-01
2.76732236e-01 -5.01442432e-01 5.67510188e-01 -4.79332656e-01
3.49296600e-01 -5.06809533e-01 2.37515584e-01 3.49460542e-01
-2.05810651e-01 -6.34436356e-03 -3.65335145e-03 7.63520002e-01
-7.66359329e-01 1.76908359e-01 1.10069335e+00 -4.37721573e-02
-1.24865460e+00 5.31752348e-01 -2.88685292e-01 -1.18578233e-01
9.04021680e-01 -4.69614208e-01 -2.50483871e-01 -3.93574238e-01
-7.72193313e-01 2.75180012e-01 5.58782220e-01 3.28857869e-01
7.74672925e-01 -1.22543240e+00 -7.65607834e-01 1.99423194e-01
-7.84472004e-02 -1.98821768e-01 1.26413870e+00 1.33539033e+00
-5.59380233e-01 1.37998372e-01 -5.29841959e-01 -7.07481205e-01
-1.17914140e+00 4.46881324e-01 5.57423770e-01 -5.67631535e-02
-6.65732741e-01 7.26356745e-01 4.38426614e-01 2.73775935e-01
3.96705121e-02 -8.42383727e-02 -5.22978187e-01 2.20727265e-01
8.41229916e-01 6.20496631e-01 -3.29125196e-01 -9.98858571e-01
-3.01826298e-01 9.92953479e-01 8.53231177e-02 -9.12610516e-02
1.19524252e+00 -9.28364277e-01 -1.03584051e-01 1.61945596e-01
1.10510623e+00 7.99409896e-02 -1.84807456e+00 -5.05302370e-01
-1.47022590e-01 -1.04480553e+00 2.32431129e-01 -2.20653787e-01
-1.44807351e+00 7.97932863e-01 7.94396818e-01 -2.71808684e-01
1.66113162e+00 -3.32359225e-01 8.98566961e-01 -5.23091555e-01
1.97763935e-01 -8.91214371e-01 3.22766513e-01 4.41663027e-01
7.60819256e-01 -1.28614342e+00 1.42146215e-01 -5.86824358e-01
-7.45473504e-01 1.31662202e+00 6.98340952e-01 -1.35673076e-01
3.22555542e-01 -4.43783142e-02 6.59185275e-02 2.62599558e-01
-6.20463192e-01 -2.86451012e-01 4.51126873e-01 4.91135836e-01
4.41201657e-01 -2.62699664e-01 -4.88675952e-01 2.55193532e-01
3.87269348e-01 -1.31544955e-02 4.25612003e-01 7.95614600e-01
-3.62707525e-01 -8.05613697e-01 -3.90594482e-01 -6.04271479e-02
-4.06465679e-01 -1.05466373e-01 2.96492577e-01 4.54603195e-01
9.80020240e-02 8.46531093e-01 -3.83692347e-02 -4.46347862e-01
1.38853148e-01 -4.12076503e-01 3.95869285e-01 -3.99289615e-02
-2.74213105e-01 5.56733429e-01 -9.55434069e-02 -9.51536000e-01
-9.08287525e-01 -5.98119318e-01 -1.13805187e+00 -2.82299906e-01
-1.10716030e-01 -9.46626812e-02 7.39231780e-02 7.47609973e-01
3.36995274e-01 4.89294827e-01 9.07945454e-01 -1.17767572e+00
-2.08173856e-01 -7.97384262e-01 -7.00327754e-01 3.14765692e-01
6.51915073e-01 -6.86334193e-01 -6.76832736e-01 3.25925350e-01] | [11.090127944946289, -1.8587006330490112] |
45a178a4-9156-47ed-8009-77e7b8cf0ffb | generalized-universal-domain-adaptation-with | 2305.04466 | null | https://arxiv.org/abs/2305.04466v1 | https://arxiv.org/pdf/2305.04466v1.pdf | Generalized Universal Domain Adaptation with Generative Flow Networks | We introduce a new problem in unsupervised domain adaptation, termed as Generalized Universal Domain Adaptation (GUDA), which aims to achieve precise prediction of all target labels including unknown categories. GUDA bridges the gap between label distribution shift-based and label space mismatch-based variants, essentially categorizing them as a unified problem, guiding to a comprehensive framework for thoroughly solving all the variants. The key challenge of GUDA is developing and identifying novel target categories while estimating the target label distribution. To address this problem, we take advantage of the powerful exploration capability of generative flow networks and propose an active domain adaptation algorithm named GFlowDA, which selects diverse samples with probabilities proportional to a reward function. To enhance the exploration capability and effectively perceive the target label distribution, we tailor the states and rewards, and introduce an efficient solution for parent exploration and state transition. We also propose a training paradigm for GUDA called Generalized Universal Adversarial Network (GUAN), which involves collaborative optimization between GUAN and GFlowNet. Theoretical analysis highlights the importance of exploration, and extensive experiments on benchmark datasets demonstrate the superiority of GFlowDA. | ['Chao Wu', 'Jun Xiao', 'Kun Kuang', 'Fei Wu', 'Jianye Hao', 'Yunfeng Shao', 'Yinchuan Li', 'Didi Zhu'] | 2023-05-08 | null | null | null | null | ['universal-domain-adaptation', 'unsupervised-domain-adaptation'] | ['computer-vision', 'methodology'] | [ 2.26715922e-01 7.30127916e-02 -4.84472513e-01 -2.51841694e-01
-7.70109892e-01 -6.85078561e-01 4.39459562e-01 -3.40214998e-01
-1.92708075e-01 8.76122236e-01 9.08770561e-02 -1.97898611e-01
-9.67422947e-02 -8.28273058e-01 -3.70672733e-01 -1.04742122e+00
2.07622677e-01 6.09157383e-01 -6.48761168e-02 -4.84547131e-02
-9.76841226e-02 4.49120164e-01 -1.06343472e+00 -3.77131820e-01
1.29318202e+00 1.02211332e+00 7.98388198e-02 3.43635798e-01
-2.17629522e-01 7.02868223e-01 -7.14671016e-01 -2.00312898e-01
3.16190839e-01 -7.75239289e-01 -7.39298880e-01 2.14979425e-01
1.65146142e-01 -3.81806582e-01 -3.76061499e-01 1.03717077e+00
6.75827384e-01 5.80271602e-01 8.64067972e-01 -1.59376705e+00
-9.48731840e-01 5.25394797e-01 -3.20452094e-01 1.31127164e-01
-9.08325464e-02 3.67819041e-01 9.61444438e-01 -7.44399011e-01
6.02777064e-01 1.19567680e+00 5.87446630e-01 1.04733074e+00
-1.37846851e+00 -9.95755374e-01 5.29929996e-01 2.30101682e-02
-1.13733542e+00 -3.20386559e-01 1.06684947e+00 -5.84381521e-01
3.60789388e-01 -4.67910878e-02 2.63185024e-01 1.39355946e+00
-3.22445095e-01 1.07755065e+00 9.28168535e-01 -3.69492739e-01
5.96645296e-01 9.99122560e-02 1.06861852e-01 5.73518276e-01
7.83200637e-02 4.05186743e-01 -3.33109230e-01 -2.45275632e-01
9.96733546e-01 9.95571688e-02 -9.45892483e-02 -1.09441280e+00
-1.18720114e+00 1.08277392e+00 5.42167068e-01 -1.69124067e-01
-2.16674536e-01 -1.43265501e-01 4.90676850e-01 2.37995476e-01
4.98863250e-01 6.11107111e-01 -3.24601591e-01 1.48578271e-01
-5.20670772e-01 2.62550503e-01 5.62824547e-01 1.04290855e+00
8.66695464e-01 3.77390534e-01 -4.82448548e-01 9.51969504e-01
3.63270104e-01 5.73166251e-01 7.12148786e-01 -9.03171480e-01
3.52018416e-01 8.07151020e-01 1.46521941e-01 -5.78370810e-01
-1.93642959e-01 -8.53592992e-01 -9.23044384e-01 3.86585742e-01
4.64260459e-01 -4.30131018e-01 -1.16768777e+00 2.15602016e+00
5.27073920e-01 4.12387401e-01 2.90634215e-01 7.65492797e-01
6.21040285e-01 6.26548886e-01 5.80467463e-01 -3.24800014e-01
7.57051468e-01 -1.11613059e+00 -6.46090567e-01 -3.27631533e-01
4.12138730e-01 -7.69021735e-02 1.24157190e+00 1.44320011e-01
-8.13442469e-01 -7.56622016e-01 -1.06499565e+00 1.53225541e-01
-3.73323113e-01 1.70454534e-03 6.67174459e-01 5.10921597e-01
-6.65225804e-01 4.23681498e-01 -7.98048675e-01 -1.09626472e-01
7.51154482e-01 4.26472932e-01 -5.81127650e-04 -3.09151001e-02
-1.34952724e+00 5.69512069e-01 7.35982180e-01 -9.86979380e-02
-1.20739830e+00 -7.29272664e-01 -8.88934314e-01 4.99718748e-02
4.25659508e-01 -6.88599885e-01 1.33334315e+00 -1.03740096e+00
-1.81569159e+00 6.86585486e-01 3.71744968e-02 -4.84751403e-01
4.83246952e-01 2.52662338e-02 -5.88278174e-01 -9.92332771e-02
2.20865414e-01 8.42314303e-01 7.74634659e-01 -1.29498291e+00
-6.91918612e-01 -2.53046870e-01 -4.13210988e-02 4.33069229e-01
-6.46595478e-01 -3.96774709e-01 -1.71753272e-01 -8.89416337e-01
-3.27838734e-02 -7.43196487e-01 -4.42508042e-01 -1.49862364e-01
-4.44118351e-01 -5.52654028e-01 8.85601759e-01 -3.86155546e-01
1.37656093e+00 -2.07705927e+00 3.85321707e-01 2.00989649e-01
5.37145853e-01 4.18909788e-01 -3.08413923e-01 1.03628442e-01
4.05156277e-02 -1.65245712e-01 -5.55807233e-01 -3.65310073e-01
1.48902193e-01 3.72925550e-01 -8.00393701e-01 4.03292142e-02
1.85693517e-01 1.13749778e+00 -1.34946215e+00 -2.57433742e-01
4.91465554e-02 2.16323167e-01 -3.80530119e-01 5.05664229e-01
-4.78088439e-01 7.14196146e-01 -7.89866209e-01 7.95982301e-01
7.31332958e-01 -4.07529056e-01 1.52689159e-01 1.16651237e-01
3.12928408e-01 -4.87488657e-02 -1.01564574e+00 1.73148870e+00
-4.56919044e-01 2.49617845e-01 -2.77135283e-01 -1.12213242e+00
1.51840913e+00 -2.40972415e-02 4.15558040e-01 -5.89476705e-01
8.69737491e-02 1.99867174e-01 -2.10180923e-01 -1.10316545e-01
2.96460003e-01 -2.18355283e-01 -2.26800382e-01 4.10153985e-01
4.14422512e-01 1.96159199e-01 -2.27227751e-02 1.79729715e-01
7.19987154e-01 4.65615451e-01 4.22496915e-01 -1.38757214e-01
6.00917757e-01 6.41867220e-02 9.87639725e-01 8.92345607e-01
-7.26584315e-01 3.33016098e-01 4.93368804e-01 -3.82352471e-01
-9.00098026e-01 -1.49160790e+00 -6.61214348e-03 1.36958838e+00
4.59495038e-01 -2.50194105e-03 -5.58461010e-01 -1.39654517e+00
-1.24927042e-02 7.17887282e-01 -9.40684199e-01 -5.30766249e-01
-6.02114081e-01 -8.17415297e-01 3.44792455e-01 8.15767407e-01
7.24912405e-01 -1.22887111e+00 -1.69412181e-01 3.09749037e-01
-7.01235980e-02 -6.43338442e-01 -5.26854038e-01 3.14311087e-01
-8.56714845e-01 -9.12443578e-01 -9.50604200e-01 -9.52368021e-01
6.38983488e-01 -5.96210510e-02 1.14589262e+00 -4.83218551e-01
1.04498319e-01 2.71900088e-01 -3.56926978e-01 -3.15705597e-01
-6.70469940e-01 3.97816330e-01 5.48951142e-02 1.08730271e-02
4.67133492e-01 -7.19750941e-01 -5.48606455e-01 5.07321835e-01
-7.49401391e-01 -8.39637369e-02 6.75175965e-01 1.15455413e+00
7.66340196e-01 -1.97481543e-01 1.23251820e+00 -1.20160651e+00
8.68714750e-01 -8.69284570e-01 -5.77287495e-01 3.66232961e-01
-8.92675817e-01 1.08156145e-01 9.26923394e-01 -7.45598733e-01
-1.16096890e+00 1.79937869e-01 -1.21690072e-01 -8.12540591e-01
-3.04714054e-01 2.02787086e-01 -5.77148974e-01 1.25389054e-01
8.28473449e-01 4.73595738e-01 8.48721564e-02 -4.30314064e-01
6.34435117e-01 4.65330482e-01 8.91149163e-01 -5.66332638e-01
9.62462425e-01 3.45081419e-01 -1.58524260e-01 -1.36340186e-01
-1.06956613e+00 -4.24872041e-01 -7.31550574e-01 -6.28772366e-04
6.62367225e-01 -8.70063663e-01 -4.15562123e-01 5.17190218e-01
-6.37412727e-01 -6.61647081e-01 -7.75972724e-01 2.24226192e-01
-7.19738305e-01 2.29521126e-01 -2.20519632e-01 -5.75195312e-01
-3.05210948e-01 -1.06045604e+00 5.10041237e-01 5.61507821e-01
8.71947035e-03 -1.39157772e+00 4.22379732e-01 -6.01494964e-03
4.23256934e-01 4.33182597e-01 8.13777626e-01 -9.63357091e-01
-4.70931858e-01 4.64624129e-02 -1.57808378e-01 5.11114836e-01
2.86142498e-01 -4.78643775e-01 -9.01141703e-01 -4.79338020e-01
-2.37793922e-01 -6.09806359e-01 8.46237540e-01 2.92140663e-01
1.24431419e+00 -5.18713966e-02 -5.24196148e-01 7.75900781e-01
1.26282179e+00 5.96652746e-01 3.88349324e-01 4.24577028e-01
7.83874035e-01 2.94085652e-01 8.27228487e-01 3.95368218e-01
2.34967276e-01 6.78420544e-01 4.04790193e-01 -9.82687771e-02
-2.61107624e-01 -6.39843762e-01 3.01686108e-01 6.81792915e-01
3.79746646e-01 -3.88984293e-01 -7.98932552e-01 4.59375411e-01
-1.82028353e+00 -6.08077228e-01 5.14531016e-01 2.03167367e+00
8.21710110e-01 5.93485460e-02 3.49852741e-01 -1.88469619e-01
9.02277529e-01 1.23603135e-01 -1.35049081e+00 -1.30537823e-01
-9.71158296e-02 1.73912987e-01 1.77746505e-01 3.93401712e-01
-1.36897171e+00 9.79314327e-01 6.62282372e+00 1.18197596e+00
-1.00149870e+00 8.67940336e-02 7.48451650e-01 3.44087124e-01
-2.61771977e-01 -7.26249442e-02 -8.83391917e-01 5.60947597e-01
6.19346857e-01 -2.99014390e-01 2.88959205e-01 1.22592974e+00
-2.91749150e-01 4.81376469e-01 -9.71069396e-01 8.06644797e-01
-4.41547446e-02 -1.19627166e+00 1.12266190e-01 -7.69226179e-02
1.06298399e+00 -2.19279304e-01 2.31589302e-01 7.62821078e-01
9.09542561e-01 -7.34527290e-01 3.08899850e-01 3.48207325e-01
1.08347642e+00 -9.19052303e-01 3.41375709e-01 2.61917531e-01
-1.03160059e+00 -4.16871756e-01 -5.56405544e-01 2.97958761e-01
6.23270012e-02 2.18544543e-01 -8.14244866e-01 4.38870341e-01
2.16219887e-01 9.82580900e-01 -4.88007516e-01 1.08933115e+00
-4.84546334e-01 8.52777839e-01 3.98384668e-02 1.17369689e-01
2.38637388e-01 -2.30395004e-01 6.54480815e-01 8.97806764e-01
1.69626966e-01 -1.29757538e-01 4.13440526e-01 1.07379282e+00
-3.17027628e-01 -1.43718123e-02 -2.55294859e-01 -6.05898760e-02
9.26757455e-01 1.12280893e+00 -5.77072978e-01 -3.41355890e-01
-5.40751480e-02 1.04660487e+00 5.08716464e-01 5.48307955e-01
-7.43485391e-01 -4.55930829e-01 6.32272482e-01 -4.25524205e-01
2.07083270e-01 2.11252924e-02 -3.21089268e-01 -1.25426519e+00
-3.16502661e-01 -7.84198165e-01 6.96839333e-01 -4.34441030e-01
-1.62331259e+00 5.98230898e-01 -1.95385918e-01 -1.65805888e+00
-3.54599684e-01 -3.81947607e-01 -7.77890623e-01 9.86562192e-01
-1.52722061e+00 -1.15366793e+00 -4.54973370e-01 7.01623619e-01
5.63241303e-01 -6.44541383e-01 8.35104585e-01 1.33291930e-01
-8.58642995e-01 1.04586339e+00 6.09337568e-01 2.17912614e-01
7.68990695e-01 -1.43886590e+00 4.81570601e-01 9.35331345e-01
-4.28576916e-02 4.90404546e-01 3.00009966e-01 -5.93194783e-01
-8.37541878e-01 -1.42923129e+00 4.83374149e-01 -2.98997402e-01
6.15590334e-01 -5.14860511e-01 -1.02764857e+00 8.12783420e-01
-8.87022987e-02 5.00981472e-02 7.19098508e-01 6.24866076e-02
-3.77521724e-01 -7.74286687e-02 -1.22068286e+00 4.27109778e-01
1.26027739e+00 -2.91293830e-01 -6.08558297e-01 3.17522317e-01
1.03090155e+00 -4.25766259e-01 -7.80277789e-01 4.58022207e-01
1.54060528e-01 -6.82558596e-01 1.10933232e+00 -8.13240111e-01
3.62269133e-01 -2.57218838e-01 1.58805639e-01 -1.63407588e+00
-7.17777491e-01 -5.51703632e-01 -5.76042056e-01 1.46996272e+00
1.36313930e-01 -8.72147381e-01 1.10154021e+00 2.77126878e-01
-3.06989729e-01 -7.77010024e-01 -7.10005879e-01 -9.29716945e-01
4.67077702e-01 4.96561676e-02 8.10966849e-01 1.27720332e+00
-3.08777511e-01 3.21587205e-01 -5.04758000e-01 3.94025482e-02
7.35932946e-01 2.26631239e-01 6.70049667e-01 -1.28879249e+00
-3.70062917e-01 -4.12698179e-01 -2.07291692e-01 -1.44151485e+00
3.65802854e-01 -1.03687286e+00 3.12193278e-02 -1.28457952e+00
4.89100292e-02 -9.04211044e-01 -7.75027692e-01 6.11642361e-01
-4.23995048e-01 1.37134105e-01 1.00798413e-01 5.07563949e-01
-6.78841472e-01 8.99121225e-01 1.44245803e+00 -9.25201103e-02
-4.80306178e-01 2.29647905e-01 -1.05875230e+00 3.83355856e-01
7.63368130e-01 -2.90275991e-01 -9.09890890e-01 -1.81006655e-01
-2.41046235e-01 -7.21591897e-03 2.89278150e-01 -1.05020702e+00
4.74788547e-02 -3.91572386e-01 4.70880836e-01 -5.35472393e-01
9.41275582e-02 -5.38447380e-01 -3.01996201e-01 3.33259881e-01
-7.75717497e-01 -4.23487604e-01 -5.47089428e-02 8.36345196e-01
-2.60602593e-01 2.42802575e-02 1.07097495e+00 1.41580537e-01
-1.26665533e+00 6.83307052e-01 -6.69567138e-02 3.61795336e-01
1.14084339e+00 -1.76311299e-01 -3.40500563e-01 -1.52380362e-01
-1.07429814e+00 5.48087060e-01 2.09003597e-01 4.77377295e-01
4.95416522e-01 -1.66752267e+00 -6.11706436e-01 4.52954710e-01
2.22050250e-01 2.42802560e-01 4.36379343e-01 2.42411599e-01
-1.06965363e-01 2.12929234e-01 -3.56431901e-01 -5.10766208e-01
-5.87171912e-01 7.26470828e-01 5.05304635e-01 -5.57603240e-01
-4.02431458e-01 1.11658251e+00 6.98289156e-01 -8.84844601e-01
2.51869470e-01 1.80995986e-01 -3.68537843e-01 -1.11339211e-01
4.06253517e-01 5.12685716e-01 -3.32842708e-01 -2.82109410e-01
-5.26683144e-02 2.27779061e-01 -2.26199448e-01 3.63531947e-01
1.08457470e+00 -3.32370937e-01 3.85804832e-01 2.27557093e-01
1.11036003e+00 -3.11806023e-01 -1.93009913e+00 -6.24012887e-01
-1.68616876e-01 -3.68008047e-01 -7.45291635e-02 -1.22391760e+00
-1.09285080e+00 8.28450203e-01 9.14038181e-01 6.18125834e-02
1.35200465e+00 -1.10421985e-01 7.71039665e-01 1.97410956e-01
3.99323367e-02 -1.02566969e+00 3.48505497e-01 4.98984277e-01
6.06721878e-01 -1.19000542e+00 -3.12024593e-01 -3.00826341e-01
-8.24242413e-01 9.30721760e-01 1.08526623e+00 -1.64002731e-01
3.13469231e-01 -1.12345614e-01 2.82455713e-01 2.41564795e-01
-4.18959260e-01 -1.50564268e-01 2.34057263e-01 1.07840502e+00
-1.07898332e-01 1.31856143e-01 7.11353775e-03 8.09413373e-01
1.55591428e-01 -5.78946285e-02 3.78417373e-02 6.31784320e-01
-4.02333200e-01 -1.46876609e+00 -1.40891299e-01 3.15814674e-01
7.94419795e-02 1.06650449e-01 -1.02248207e-01 7.77490735e-01
2.99838454e-01 4.73990619e-01 1.19898535e-01 -4.65331852e-01
2.18649104e-01 2.73790896e-01 6.43881559e-02 -4.86796528e-01
-2.53426313e-01 -9.66886356e-02 -3.74954820e-01 -1.72663629e-01
-1.52521536e-01 -5.32414615e-01 -1.07729590e+00 1.49958268e-01
-2.39016920e-01 2.95116365e-01 1.99159905e-01 8.08818758e-01
5.07462323e-01 5.95039010e-01 1.08153307e+00 -5.35759032e-01
-9.85200226e-01 -8.49378765e-01 -7.59719014e-01 4.60348368e-01
3.20664078e-01 -1.03000426e+00 -3.19611371e-01 -6.10333271e-02] | [10.32065486907959, 3.0520823001861572] |
41bfb318-1a1a-4e43-b1f4-a0b008164ce5 | ma2cl-masked-attentive-contrastive-learning | 2306.02006 | null | https://arxiv.org/abs/2306.02006v1 | https://arxiv.org/pdf/2306.02006v1.pdf | MA2CL:Masked Attentive Contrastive Learning for Multi-Agent Reinforcement Learning | Recent approaches have utilized self-supervised auxiliary tasks as representation learning to improve the performance and sample efficiency of vision-based reinforcement learning algorithms in single-agent settings. However, in multi-agent reinforcement learning (MARL), these techniques face challenges because each agent only receives partial observation from an environment influenced by others, resulting in correlated observations in the agent dimension. So it is necessary to consider agent-level information in representation learning for MARL. In this paper, we propose an effective framework called \textbf{M}ulti-\textbf{A}gent \textbf{M}asked \textbf{A}ttentive \textbf{C}ontrastive \textbf{L}earning (MA2CL), which encourages learning representation to be both temporal and agent-level predictive by reconstructing the masked agent observation in latent space. Specifically, we use an attention reconstruction model for recovering and the model is trained via contrastive learning. MA2CL allows better utilization of contextual information at the agent level, facilitating the training of MARL agents for cooperation tasks. Extensive experiments demonstrate that our method significantly improves the performance and sample efficiency of different MARL algorithms and outperforms other methods in various vision-based and state-based scenarios. Our code can be found in \url{https://github.com/ustchlsong/MA2CL} | ['Houqiang Li', 'Wengang Zhou', 'Mingxiao Feng', 'Haolin Song'] | 2023-06-03 | null | null | null | null | ['multi-agent-reinforcement-learning'] | ['methodology'] | [-2.12873250e-01 -4.81954068e-02 -3.90397519e-01 2.56648138e-02
-7.57538259e-01 -3.01984251e-01 8.69863331e-01 1.12989899e-02
-5.30491889e-01 9.20858204e-01 1.02904990e-01 1.51183635e-01
-2.05861017e-01 -4.57422912e-01 -7.50122488e-01 -1.06132829e+00
-1.16057232e-01 7.25944340e-01 -9.26586986e-02 -1.82877511e-01
1.06154367e-01 1.94978163e-01 -1.31291318e+00 3.83517109e-02
6.91746294e-01 6.59739852e-01 7.94195056e-01 6.72968388e-01
2.37616375e-01 1.36146259e+00 -7.09114313e-01 4.20869142e-02
2.40648583e-01 -6.31540179e-01 -4.70602810e-01 3.36727381e-01
-2.02006951e-01 -6.93797410e-01 -6.47421479e-01 8.90613377e-01
4.08215761e-01 5.62892735e-01 6.82493389e-01 -1.65125775e+00
-7.66947508e-01 4.72895205e-01 -9.19772685e-01 3.98322821e-01
3.40453982e-01 6.44652247e-01 7.96602070e-01 -8.64103854e-01
5.45184135e-01 1.55877125e+00 5.41330501e-02 8.02433014e-01
-1.08033442e+00 -7.33895242e-01 6.64371490e-01 6.29855096e-01
-8.04995000e-01 -4.52460855e-01 9.69146490e-01 -3.33933592e-01
9.24429536e-01 -1.06394894e-01 6.90668523e-01 1.53950703e+00
2.03038424e-01 1.39386928e+00 1.45625949e+00 -2.97485739e-01
3.69859338e-01 8.42297217e-04 8.34527686e-02 1.04520440e+00
2.31392533e-02 6.05208099e-01 -8.82610619e-01 -1.84056744e-01
1.12612796e+00 1.92406669e-01 -7.12496266e-02 -3.51116836e-01
-1.27880859e+00 1.07264256e+00 3.52513254e-01 -1.00595333e-01
-7.36221611e-01 3.96586925e-01 2.25756645e-01 4.61573988e-01
3.47274125e-01 1.86352000e-01 -2.23315820e-01 -1.05729163e-01
-2.46057659e-01 2.64814556e-01 2.99393207e-01 8.50112140e-01
8.18710208e-01 5.80929101e-01 -1.34431198e-01 8.31268668e-01
6.58869624e-01 7.09268987e-01 6.98547900e-01 -1.37261820e+00
3.65100980e-01 3.48647892e-01 1.97891340e-01 -4.06849027e-01
-4.46734160e-01 -2.51521021e-01 -7.78357744e-01 7.52933264e-01
1.36471316e-01 -4.15430993e-01 -7.68353999e-01 1.92778003e+00
3.72233838e-01 4.11969483e-01 5.53123653e-01 9.61683691e-01
7.81723797e-01 7.09825873e-01 7.03464597e-02 -5.47491133e-01
1.06931520e+00 -1.51386523e+00 -6.07851386e-01 -5.19900262e-01
2.77414292e-01 -5.05017340e-01 9.72570896e-01 2.01925695e-01
-9.58942115e-01 -5.26317358e-01 -8.50713372e-01 3.06525558e-01
7.08032474e-02 1.66945845e-01 7.25212514e-01 3.30783390e-02
-1.06289995e+00 5.12737259e-02 -1.03849757e+00 -2.36128315e-01
3.64151686e-01 1.71708733e-01 -1.68371767e-01 -4.36062776e-02
-8.31659496e-01 8.49248946e-01 2.93600649e-01 -3.08530107e-02
-1.76335585e+00 -9.27056372e-02 -9.91636813e-01 -2.86627561e-01
6.05000556e-01 -5.53019345e-01 1.34006894e+00 -8.70211661e-01
-1.78709900e+00 3.64893228e-01 -1.08756453e-01 -5.80033600e-01
4.31387663e-01 -7.64625818e-02 -6.93533868e-02 3.48163754e-01
3.77579808e-01 6.76215708e-01 1.26405323e+00 -1.58805990e+00
-6.61861181e-01 -4.02354747e-01 4.36371803e-01 6.37905240e-01
7.41040930e-02 -2.80617386e-01 -1.81582406e-01 -7.97322869e-01
-2.66878217e-01 -9.44267869e-01 -3.11595768e-01 -1.84101269e-01
9.73478183e-02 -6.18573189e-01 9.49918330e-01 -4.84738529e-01
5.67892551e-01 -1.96646118e+00 4.05810267e-01 -2.80114502e-01
2.99203813e-01 2.71887183e-01 -6.44373953e-01 6.81069314e-01
2.89368629e-01 -5.88716865e-01 9.82228890e-02 -6.24814332e-01
-3.87994736e-03 5.34414470e-01 -4.25530910e-01 4.96279091e-01
-2.09667790e-03 1.02676046e+00 -1.08050597e+00 -4.05508935e-01
5.39979041e-01 1.76287830e-01 -3.52815628e-01 5.40625453e-01
-5.06199419e-01 8.77784252e-01 -8.30603659e-01 6.59803689e-01
2.29685664e-01 -3.04346621e-01 1.37967877e-02 3.24784428e-01
7.84538686e-02 -1.83424726e-01 -8.26295972e-01 1.84551787e+00
-3.79016101e-01 4.15540725e-01 2.69217759e-01 -1.20790875e+00
8.38422120e-01 4.13781941e-01 7.37560570e-01 -8.18132877e-01
1.10407926e-01 -3.02139103e-01 -3.66125628e-02 -5.88845551e-01
2.80978322e-01 1.45204857e-01 6.82949945e-02 8.05049658e-01
1.40685499e-01 -2.23461390e-02 2.24615142e-01 3.23765308e-01
9.27006662e-01 4.53768432e-01 2.50932574e-01 2.26771593e-01
2.75737733e-01 1.26968045e-02 6.01289868e-01 9.92035866e-01
-5.41963041e-01 -1.54627666e-01 1.50644451e-01 -3.32814395e-01
-7.17741072e-01 -9.08805132e-01 4.31799918e-01 1.41473806e+00
3.69860888e-01 -2.14935809e-01 -5.13520479e-01 -7.44104147e-01
-1.58327803e-01 8.26907158e-01 -6.22666597e-01 -2.94278979e-01
-5.97018480e-01 -6.94807470e-01 2.30583087e-01 4.12568271e-01
6.61113918e-01 -1.68255568e+00 -9.98994648e-01 2.95885354e-01
-2.40785420e-01 -1.00850785e+00 -3.71448874e-01 1.78894475e-01
-6.43971503e-01 -1.14304173e+00 -6.20408773e-01 -6.72038019e-01
6.51391387e-01 5.00372946e-01 8.33710074e-01 1.36874160e-02
-1.28322557e-01 9.21363890e-01 -5.58723092e-01 -5.66972435e-01
-3.27709734e-01 -3.30385149e-01 4.39107120e-01 -5.00005148e-02
2.46599764e-01 -4.95529175e-01 -4.23032284e-01 1.61908910e-01
-6.64220095e-01 1.74914077e-01 7.24314630e-01 1.07750273e+00
4.40253794e-01 -2.11456999e-01 8.70569170e-01 -3.01993936e-01
7.64821529e-01 -4.63001966e-01 -6.82322562e-01 1.72348291e-01
-5.67470729e-01 1.02363594e-01 6.15736723e-01 -7.86261261e-01
-9.80074167e-01 -2.04907447e-01 2.54459560e-01 -8.92888844e-01
-2.50955492e-01 5.11574805e-01 3.63082290e-01 2.79402018e-01
4.62617189e-01 6.65745318e-01 3.43485802e-01 -2.62982339e-01
1.42114252e-01 4.45324302e-01 1.90992922e-01 -7.06559837e-01
7.63225317e-01 3.90897483e-01 -1.15546383e-01 -7.06630230e-01
-7.96822727e-01 -1.61068022e-01 -3.91751565e-02 -5.71331203e-01
6.85937226e-01 -1.15623629e+00 -1.15685523e+00 5.41011274e-01
-8.40542853e-01 -9.40946162e-01 -5.08684456e-01 7.92604923e-01
-9.93120730e-01 3.50739926e-01 -5.85844457e-01 -1.04287076e+00
-1.52166486e-01 -1.40990722e+00 7.88767338e-01 3.34560633e-01
2.97346860e-01 -9.43505108e-01 2.93258965e-01 6.35047674e-01
2.73356698e-02 -4.24983576e-02 6.32504821e-01 -4.94052976e-01
-7.58284330e-01 1.78119361e-01 5.61497547e-02 1.60503954e-01
2.04540879e-01 -5.90774655e-01 -8.37650895e-01 -9.03240919e-01
1.80569410e-01 -9.03329492e-01 8.82327914e-01 5.74624956e-01
7.78998613e-01 -4.81907994e-01 -7.92900398e-02 1.01126067e-01
1.07720077e+00 4.49592531e-01 1.40709296e-01 4.42576408e-01
5.26492536e-01 3.60178649e-01 9.17773247e-01 8.85009944e-01
8.32831442e-01 5.33368945e-01 7.91348100e-01 1.38586357e-01
-2.51943380e-01 -1.90103352e-01 9.52411354e-01 6.65410280e-01
-3.72506589e-01 -2.94991285e-01 -4.79522169e-01 4.03138340e-01
-2.47261000e+00 -1.23867095e+00 4.70263183e-01 1.77060831e+00
6.09760940e-01 -2.74396211e-01 1.37133002e-01 -4.95807767e-01
6.01299107e-01 4.80170071e-01 -1.19503272e+00 -4.79782857e-02
2.65785703e-03 -2.98309445e-01 2.94852499e-02 6.28132284e-01
-8.99351776e-01 1.11316073e+00 5.12572813e+00 8.01842451e-01
-7.95678377e-01 4.63151932e-01 3.20014268e-01 -3.04843396e-01
3.33725512e-02 -1.54579222e-01 -6.42949343e-01 4.16019350e-01
5.06685257e-01 -1.77177787e-01 8.85081828e-01 8.67436707e-01
4.67065394e-01 -4.72904652e-01 -1.05776823e+00 1.14778197e+00
2.68866062e-01 -1.17379880e+00 4.98528630e-02 2.12614298e-01
6.48711860e-01 2.87029654e-01 2.82698780e-01 8.05133283e-01
8.91469896e-01 -7.51942575e-01 7.08786130e-01 4.53410447e-01
3.54826450e-01 -4.69308376e-01 3.43204141e-01 6.85422897e-01
-1.17758656e+00 -3.25173438e-01 -2.71168202e-01 -2.06678241e-01
-1.21891178e-01 -1.90950781e-01 -8.00425291e-01 5.52357018e-01
6.46010041e-01 8.96863401e-01 -2.64091402e-01 5.40845811e-01
-3.87867838e-01 3.93483341e-01 4.64917682e-02 -1.56882897e-01
5.79425514e-01 -4.77217495e-01 9.06967282e-01 5.71603239e-01
2.62823198e-02 2.35901043e-01 9.49985445e-01 5.12550354e-01
8.80705193e-02 -3.19996029e-01 -5.77060163e-01 3.69497314e-02
4.26928192e-01 1.06574035e+00 -3.85529786e-01 -2.72037864e-01
-3.72557402e-01 9.32822943e-01 7.70360351e-01 7.14117110e-01
-9.43536937e-01 3.32354307e-01 6.56870306e-01 -4.87474412e-01
2.83552080e-01 -4.42929953e-01 4.34077710e-01 -1.30789173e+00
-8.84661153e-02 -9.93391335e-01 4.28963333e-01 -8.06831479e-01
-1.21368873e+00 5.13066769e-01 1.96602084e-02 -1.22144818e+00
-7.72596538e-01 -2.96634227e-01 -3.70793283e-01 3.83205026e-01
-1.70901740e+00 -1.37595367e+00 -1.47517309e-01 1.09391332e+00
1.29631805e+00 -9.50178444e-01 8.46680462e-01 -2.41339996e-01
-7.72042572e-01 2.38145381e-01 2.72299677e-01 -4.36363705e-02
5.57711840e-01 -1.17423832e+00 -3.25140804e-01 7.15995014e-01
2.84417242e-01 1.46569014e-01 6.33043289e-01 -5.99608898e-01
-1.72890377e+00 -1.12167895e+00 -9.16717872e-02 -2.30580151e-01
6.77171409e-01 4.53601703e-02 -4.89090830e-01 1.02310431e+00
5.37297606e-01 -7.35112652e-02 4.87122864e-01 -1.87719628e-01
-2.04398498e-01 9.55563560e-02 -1.07737339e+00 7.33337998e-01
6.72161639e-01 -4.16564703e-01 -6.72110260e-01 5.31390190e-01
5.63842893e-01 -2.57393122e-01 -5.19510686e-01 5.43448590e-02
1.63314819e-01 -8.05532515e-01 8.89450490e-01 -5.84345877e-01
2.80881256e-01 -3.48586828e-01 -8.55122656e-02 -1.50408912e+00
-2.89579660e-01 -5.96388280e-01 -6.46535277e-01 7.80661941e-01
-3.25767696e-02 -7.75327206e-01 6.07664466e-01 2.95963317e-01
-1.17129117e-01 -6.32920563e-01 -1.06331146e+00 -8.23910296e-01
-1.28090620e-01 -1.28383234e-01 1.50976956e-01 9.00695562e-01
-1.00305490e-01 3.55307221e-01 -8.16643536e-01 3.27479035e-01
9.70934868e-01 3.81396115e-01 8.93188119e-01 -8.84692192e-01
-5.43338656e-01 -2.92364776e-01 1.26667261e-01 -1.16391861e+00
5.06608546e-01 -7.16023803e-01 7.44379759e-02 -1.64954638e+00
3.68976951e-01 -4.25209403e-01 -4.16114956e-01 7.82924533e-01
-5.74801713e-02 5.09544089e-02 5.94534934e-01 6.03454590e-01
-9.52967823e-01 1.17091620e+00 1.41550112e+00 -3.09789807e-01
-1.91980824e-01 5.55761456e-02 -3.52285206e-01 8.40518951e-01
8.54828119e-01 -4.18163955e-01 -6.94274068e-01 -6.04416728e-01
-3.35113108e-01 5.27473688e-01 6.66745365e-01 -7.75136530e-01
2.91118056e-01 -5.78075290e-01 3.43404710e-01 -4.68675613e-01
9.79710937e-01 -7.35646307e-01 -2.79301792e-01 6.33663714e-01
-4.61706370e-01 1.08957835e-01 -2.59814002e-02 8.49765778e-01
-3.95986848e-02 -3.84822518e-01 6.14265621e-01 -5.04383564e-01
-8.89365971e-01 5.46127558e-01 -7.68927217e-01 -9.71202180e-02
1.25274444e+00 1.26823828e-01 -5.25474906e-01 -6.36285245e-01
-7.11382031e-01 6.48195565e-01 1.54134974e-01 5.11718333e-01
1.00301147e+00 -1.24467218e+00 -7.84945846e-01 9.38562080e-02
3.92152481e-02 -1.24785259e-01 3.38183105e-01 7.52242267e-01
1.08726464e-01 2.37786040e-01 -3.43001574e-01 -5.42360961e-01
-1.05456412e+00 7.05158949e-01 1.96982473e-01 -4.31309730e-01
-6.40145242e-01 5.77171743e-01 4.10269707e-01 -3.46220911e-01
3.88194412e-01 2.18414605e-01 -3.27855080e-01 -9.60822217e-03
4.77210104e-01 4.01980639e-01 -6.35990620e-01 -6.23795569e-01
-1.32671967e-01 2.13438824e-01 -4.81467396e-01 -3.58489931e-01
1.57537055e+00 -3.77904892e-01 1.35645941e-01 5.67279518e-01
5.51278889e-01 -4.01899159e-01 -1.93642974e+00 -5.30625582e-01
-4.32236046e-01 -2.70168841e-01 1.34465605e-01 -8.06453705e-01
-1.09648466e+00 5.23742318e-01 6.36786282e-01 1.62904948e-01
9.35895681e-01 2.64077336e-01 4.13256019e-01 6.64928496e-01
7.66429424e-01 -1.15907001e+00 8.28788400e-01 5.64926326e-01
1.26024425e+00 -1.57982504e+00 5.73269688e-02 2.83734083e-01
-1.16237688e+00 7.62106895e-01 9.62557375e-01 -3.57134461e-01
3.32852870e-01 2.56114081e-02 1.17463134e-01 -2.19910383e-01
-1.11031199e+00 -4.36149150e-01 -2.83927143e-01 8.40502083e-01
-3.05318445e-01 1.21046185e-01 1.51773810e-01 2.88627237e-01
1.74736798e-01 -2.24162981e-01 6.39197528e-01 1.15523374e+00
-4.25738096e-01 -1.02693641e+00 -3.59993994e-01 3.09286952e-01
5.99117130e-02 1.74012303e-01 1.80733427e-02 6.47810221e-01
-2.23218247e-01 1.27532530e+00 -1.47049174e-01 -6.27924204e-02
-7.84388483e-02 -2.48078093e-01 6.51494980e-01 -4.19012249e-01
-3.87412459e-01 4.19194728e-01 -1.96850196e-01 -6.24883771e-01
-8.10969412e-01 -7.83443868e-01 -1.37919056e+00 -7.36063495e-02
-7.43673295e-02 1.45555183e-01 2.28860602e-01 9.87417161e-01
2.52862632e-01 6.02695286e-01 8.89530182e-01 -1.08358061e+00
-5.84583282e-01 -9.81477022e-01 -5.46265602e-01 1.59967601e-01
6.58575773e-01 -1.05100203e+00 -1.37784094e-01 -1.30280629e-01] | [4.2074174880981445, 1.4280364513397217] |
3fd441c8-5e7b-4bcb-b237-f67e04206053 | towards-large-scale-single-shot-millimeter | 2305.15750 | null | https://arxiv.org/abs/2305.15750v2 | https://arxiv.org/pdf/2305.15750v2.pdf | Towards Large-scale Single-shot Millimeter-wave Imaging for Low-cost Security Inspection | Millimeter-wave (MMW) imaging is emerging as a promising technique for safe security inspection. It achieves a delicate balance between imaging resolution, penetrability and human safety, resulting in higher resolution compared to low-frequency microwave, stronger penetrability compared to visible light, and stronger safety compared to X ray. Despite of recent advance in the last decades, the high cost of requisite large-scale antenna array hinders widespread adoption of MMW imaging in practice. To tackle this challenge, we report a large-scale single-shot MMW imaging framework using sparse antenna array, achieving low-cost but high-fidelity security inspection under an interpretable learning scheme. We first collected extensive full-sampled MMW echoes to study the statistical ranking of each element in the large-scale array. These elements are then sampled based on the ranking, building the experimentally optimal sparse sampling strategy that reduces the cost of antenna array by up to one order of magnitude. Additionally, we derived an untrained interpretable learning scheme, which realizes robust and accurate image reconstruction from sparsely sampled echoes. Last, we developed a neural network for automatic object detection, and experimentally demonstrated successful detection of concealed centimeter-sized targets using 10% sparse array, whereas all the other contemporary approaches failed at the same sample sampling ratio. The performance of the reported technique presents higher than 50% superiority over the existing MMW imaging schemes on various metrics including precision, recall, and mAP50. With such strong detection ability and order-of-magnitude cost reduction, we anticipate that this technique provides a practical way for large-scale single-shot MMW imaging, and could advocate its further practical applications. | ['Huteng Liu', 'Jun Zhang', 'Shiyong Li', 'Guoqiang Zhao', 'Xuyang Chang', 'Hanwen Xu', 'Chunyang Teng', 'Shuoguang Wang', 'Daoyu Li', 'Liheng Bian'] | 2023-05-25 | null | null | null | null | ['image-reconstruction'] | ['computer-vision'] | [ 4.28213656e-01 1.82079688e-01 2.84566998e-01 -3.37013483e-01
-1.07778943e+00 -4.82461751e-01 3.65333319e-01 -1.62832975e-01
-2.54207373e-01 3.35762501e-01 1.05188869e-01 -5.11997044e-01
-7.86487162e-01 -7.97475636e-01 -6.08036339e-01 -1.23954654e+00
-3.41767043e-01 6.31164312e-02 3.00677240e-01 1.23877108e-01
1.36560962e-01 4.51281250e-01 -1.20903599e+00 1.00168549e-01
7.05524921e-01 1.40924382e+00 2.98985928e-01 4.30890203e-01
5.13807356e-01 6.28135502e-01 -5.36206543e-01 -2.91519314e-01
3.56779635e-01 -3.14853668e-01 -8.15217122e-02 -1.15680974e-03
4.91425544e-01 -4.81888950e-01 -1.99466363e-01 9.54608381e-01
6.92023337e-01 -2.17130482e-01 6.54694140e-01 -6.21839702e-01
-7.41300642e-01 2.52257437e-01 -6.71969891e-01 1.06356271e-01
3.70812476e-01 3.11375022e-01 6.15150630e-01 -7.68186331e-01
2.15711951e-01 5.05547464e-01 9.07214522e-01 2.33373657e-01
-7.92380571e-01 -7.17522085e-01 -2.98033625e-01 1.83104843e-01
-1.15547204e+00 -4.00006413e-01 9.80244577e-01 -3.68754357e-01
6.90096140e-01 4.05512989e-01 7.10201859e-01 1.15236747e+00
5.24544239e-01 1.46830246e-01 1.42007804e+00 -5.55800617e-01
3.75110239e-01 -1.91651378e-02 1.89799577e-01 1.15208459e+00
7.08877146e-01 4.09857929e-01 -6.01793230e-01 -2.96610564e-01
5.47435760e-01 9.36983526e-02 -6.31318510e-01 -2.18404308e-01
-9.88355041e-01 7.25669682e-01 4.35646266e-01 4.47457552e-01
-4.78977531e-01 2.47569736e-02 -1.12451367e-01 4.49863262e-02
4.68025595e-01 8.65197718e-01 7.23548010e-02 1.99644282e-01
-9.41656411e-01 -2.83075839e-01 5.97812414e-01 4.47227716e-01
4.85009491e-01 3.55227679e-01 -1.40660569e-01 5.36854267e-01
4.60368842e-01 1.09795415e+00 3.45125981e-02 -8.48361731e-01
2.67506003e-01 1.76087007e-01 1.25758797e-01 -1.30598998e+00
-4.53977287e-01 -9.99794006e-01 -1.15942705e+00 3.71772081e-01
6.25246838e-02 -1.34451553e-01 -8.24941456e-01 1.41241682e+00
3.07921857e-01 4.00651246e-01 1.84892476e-01 9.66792107e-01
8.21873009e-01 8.32168818e-01 -1.66855037e-01 -4.15539682e-01
1.61813235e+00 -4.98788565e-01 -5.30507028e-01 -4.41924095e-01
8.38313624e-02 -5.95102489e-01 6.56604826e-01 6.81285501e-01
-8.48677397e-01 -2.41821557e-01 -1.33474219e+00 8.06750536e-01
3.25487375e-01 9.52882245e-02 8.99886668e-01 1.01049972e+00
-5.85844398e-01 1.51680991e-01 -7.66448796e-01 -1.11976728e-01
4.20942158e-01 7.73409456e-02 -3.17161977e-01 -3.54641616e-01
-7.87025452e-01 6.45957351e-01 -2.15954170e-01 3.51450324e-01
-9.05414760e-01 -9.84454751e-01 -8.07614207e-01 -3.03633641e-02
3.69820416e-01 -5.88260949e-01 7.65122116e-01 1.71919525e-01
-1.40601945e+00 6.13304853e-01 4.98394817e-02 -5.44689775e-01
-4.86058444e-02 -1.39366180e-01 -5.30923605e-01 6.22965932e-01
-3.09276097e-02 -2.55675614e-01 1.13737690e+00 -1.43261147e+00
-2.16788948e-01 -4.80981201e-01 -1.94257215e-01 -3.25375915e-01
-4.82783496e-01 -1.93045348e-01 1.29739404e-01 -4.28681940e-01
7.68712938e-01 -5.98846912e-01 -2.54215389e-01 2.02633515e-02
-1.52783439e-01 2.07600817e-01 7.05454290e-01 -7.02611268e-01
9.35986280e-01 -2.11457586e+00 -2.62131423e-01 1.33274123e-01
5.71610093e-01 2.81248748e-01 -2.43519038e-01 6.79185271e-01
1.61483720e-01 -4.95869875e-01 -4.15353060e-01 -1.60084385e-02
-2.71376133e-01 -2.51358479e-01 -2.26371288e-01 8.70705307e-01
2.60912348e-02 7.67551899e-01 -7.72316158e-01 1.22091128e-03
1.79046601e-01 3.67702872e-01 -4.30402368e-01 3.57234538e-01
4.26641941e-01 5.41127205e-01 -6.52000606e-01 9.48849201e-01
9.53948498e-01 -2.27810755e-01 -1.19912565e-01 -7.45494366e-01
-1.34589538e-01 -4.21020061e-01 -1.00809669e+00 1.32986760e+00
-7.15143144e-01 4.40274298e-01 3.48540723e-01 -1.51384521e+00
1.33266032e+00 3.53171378e-01 4.92741674e-01 -1.14413297e+00
-1.20184183e-01 3.56916040e-01 -2.62224823e-01 -9.90958452e-01
-1.82531416e-01 -4.11198080e-01 -1.64830253e-01 3.56643111e-01
7.57904053e-02 -3.52248013e-01 -4.24537688e-01 2.33562361e-03
1.56535006e+00 -3.89512509e-01 4.03223187e-01 -3.37739617e-01
2.92539448e-01 -1.97474852e-01 1.90312952e-01 1.10475957e+00
1.05032027e-01 5.15092969e-01 -2.43583158e-01 -6.98869288e-01
-6.74800932e-01 -1.29130375e+00 -1.56796336e-01 3.97376776e-01
4.25026953e-01 1.70111656e-01 -3.95892471e-01 -2.86712110e-01
-1.86141402e-01 5.87537348e-01 -2.48938411e-01 -3.22561979e-01
-6.19073987e-01 -9.80980277e-01 4.89472508e-01 1.23650588e-01
7.45920718e-01 -6.31635070e-01 -9.65244174e-01 9.69651043e-02
-1.40245110e-01 -1.31452715e+00 2.29277760e-01 2.61346072e-01
-6.48061752e-01 -1.05396330e+00 -5.74977756e-01 -6.47489727e-01
8.62286687e-01 5.61416984e-01 9.31392372e-01 -1.47001401e-01
-8.22077096e-01 7.07660854e-01 -4.15872902e-01 -4.79971230e-01
-1.61023393e-01 -5.51649511e-01 1.65336818e-01 1.44557700e-01
-1.66045561e-01 -6.54212236e-01 -9.74972367e-01 2.48584807e-01
-7.40728974e-01 -2.49391362e-01 1.00767899e+00 1.02109385e+00
2.04378873e-01 2.27625981e-01 5.96325457e-01 -5.08378446e-01
3.58206064e-01 -3.50193322e-01 -8.19012940e-01 2.73255348e-01
-3.24352086e-01 -7.66570494e-02 5.12680292e-01 -2.98271656e-01
-1.03297424e+00 -2.86768287e-01 -5.04574597e-01 2.10225329e-01
-5.90267917e-03 5.37983358e-01 3.62898707e-02 -7.68103600e-01
7.97470391e-01 4.70952809e-01 6.21346943e-02 -1.04475744e-01
-6.87714741e-02 5.89905143e-01 6.37635350e-01 -4.22179639e-01
1.31014526e+00 6.73633397e-01 2.43700996e-01 -1.13223147e+00
-1.05202818e+00 -4.06823844e-01 -6.65166155e-02 -2.93833375e-01
7.08258986e-01 -8.32924783e-01 -9.27214205e-01 3.62582475e-01
-1.06087458e+00 1.14219509e-01 -1.48976713e-01 8.48788023e-01
-1.53728360e-02 5.69698036e-01 -3.76377016e-01 -1.24596858e+00
-6.24359965e-01 -7.86530733e-01 1.10646605e+00 -6.83045462e-02
2.80609019e-02 -7.40722358e-01 -7.44604915e-02 6.42940998e-01
8.02442908e-01 3.63215119e-01 9.38872635e-01 -5.98554015e-02
-9.52386260e-01 -6.91717207e-01 -1.53555810e-01 1.22956537e-01
-5.40983193e-02 -7.65426576e-01 -8.79351676e-01 -5.53039968e-01
9.14443493e-01 -2.77299464e-01 8.45573962e-01 6.73754215e-01
1.08723819e+00 -3.71231049e-01 -2.81248063e-01 9.64673877e-01
1.77908230e+00 3.36559564e-01 4.66413379e-01 1.00962957e-03
1.12490550e-01 5.24248302e-01 6.39079034e-01 5.31153023e-01
-2.22617611e-01 4.26113069e-01 8.10873270e-01 -2.72144675e-01
-1.25448136e-02 1.08758144e-01 1.26566449e-02 9.69746888e-01
-2.71360040e-01 -3.41762900e-01 -7.94037163e-01 3.51184756e-01
-1.18492901e+00 -1.15012062e+00 4.38516177e-02 2.30088902e+00
1.29009873e-01 2.36379579e-02 -5.55948794e-01 1.43700205e-02
1.26165226e-01 3.51810217e-01 -2.38850817e-01 1.30652115e-01
7.87394121e-02 4.00141895e-01 4.65642482e-01 5.19163549e-01
-9.50293601e-01 4.70624082e-02 6.42228985e+00 9.76114750e-01
-1.11404502e+00 2.13547006e-01 5.25919020e-01 5.07754833e-02
-6.06664538e-01 -3.75480533e-01 -6.90755010e-01 3.70665997e-01
5.15381753e-01 2.88983107e-01 4.32226323e-02 5.68389893e-01
4.86796200e-02 -3.15712392e-01 -6.02019608e-01 1.01028562e+00
2.65872419e-01 -1.36752820e+00 -1.30576909e-01 6.24339916e-02
5.04245341e-01 -1.42971158e-01 2.18400776e-01 -1.30219042e-01
-1.85227260e-01 -9.20363605e-01 4.20417398e-01 2.22789004e-01
7.82447994e-01 -5.25099158e-01 8.37937355e-01 6.92248881e-01
-1.05990016e+00 -3.15240115e-01 -5.59101760e-01 7.49421120e-02
1.93757385e-01 1.21059239e+00 -5.03237307e-01 7.95875907e-01
6.60793722e-01 1.19023368e-01 -1.69299975e-01 9.52828288e-01
-2.14730501e-01 1.01929283e+00 -3.45508009e-01 -2.46387452e-01
3.81746829e-01 -3.61545950e-01 8.13934624e-01 1.05530059e+00
9.89988089e-01 5.28340876e-01 2.01575220e-01 6.43139064e-01
3.40820551e-01 -4.65806991e-01 -9.30355847e-01 3.56398135e-01
5.03901064e-01 1.44159031e+00 -5.66335917e-01 -2.56675705e-02
-4.48876768e-01 6.62803650e-01 -1.51117191e-01 5.05667329e-01
-9.81099606e-01 -3.88722748e-01 2.40500614e-01 3.31944078e-01
2.55380005e-01 -5.39355278e-01 -1.81206420e-01 -8.51595640e-01
2.54826963e-01 -5.58678687e-01 1.29925489e-01 -5.38463950e-01
-1.37210631e+00 9.22993422e-01 4.14227434e-02 -1.46654558e+00
-1.97843257e-02 -6.80712521e-01 -6.45412028e-01 7.02371776e-01
-1.64986169e+00 -1.25875878e+00 -7.29988635e-01 1.07631773e-01
4.42428648e-01 -3.41141343e-01 1.09905910e+00 1.01310708e-01
-1.04671352e-01 3.56180638e-01 7.84802511e-02 -5.53238504e-02
3.94252449e-01 -5.78434587e-01 -3.41145433e-02 9.55031574e-01
1.96006104e-01 6.28692269e-01 7.54937530e-01 -5.66210747e-01
-2.02145648e+00 -9.72802162e-01 2.61365652e-01 -1.72211722e-01
6.44939959e-01 -4.08877045e-01 -5.92080593e-01 1.31937027e-01
2.44234338e-01 3.28166336e-01 7.84570932e-01 -1.02234066e-01
-3.64133358e-01 -4.36938614e-01 -1.39080274e+00 1.95736408e-01
1.08520877e+00 -2.21719116e-01 -5.94767272e-01 4.12471086e-01
5.86096704e-01 -1.55660897e-01 -7.51230717e-01 8.90576541e-01
6.18030429e-01 -1.07956815e+00 1.40353930e+00 1.34428993e-01
4.41565603e-01 -8.06636885e-02 -2.73714542e-01 -1.21483171e+00
-7.84979165e-01 -6.11194253e-01 -1.83333278e-01 8.23050261e-01
3.50569606e-01 -1.06850588e+00 9.40247238e-01 -1.13081492e-01
-3.90596420e-01 -1.04300833e+00 -1.09384525e+00 -1.02922487e+00
-4.70425487e-01 -3.91198814e-01 8.76706243e-02 6.61956966e-01
-3.02831858e-01 1.90753043e-02 -6.52894080e-01 9.81417954e-01
1.49532342e+00 4.53134000e-01 2.11663216e-01 -1.14028418e+00
-8.04163039e-01 1.85777336e-01 -5.36539018e-01 -1.05637538e+00
-2.78413355e-01 -6.69965386e-01 1.32227153e-01 -1.67012858e+00
2.98128664e-01 -3.03487867e-01 -1.39351368e-01 9.37755778e-02
-6.63208440e-02 8.25543344e-01 -1.77204505e-01 -5.64477816e-02
-5.02683759e-01 4.12333876e-01 1.11153269e+00 -4.55086410e-01
3.52739960e-01 2.87280511e-02 -6.45495117e-01 8.70985210e-01
5.99146485e-01 -4.05213207e-01 -3.63469452e-01 -7.53140509e-01
1.05717611e-02 2.55825669e-01 6.63591802e-01 -1.53589654e+00
5.06073594e-01 -5.35287336e-02 4.61402595e-01 -2.99124181e-01
5.05414844e-01 -9.84158576e-01 3.93275887e-01 8.72799158e-01
3.06318223e-01 -7.77017534e-01 -9.96396542e-02 6.73413754e-01
-3.07724297e-01 -5.25950909e-01 9.99584436e-01 -1.59553140e-01
-6.05723679e-01 1.84465662e-01 -4.66979444e-01 -2.53882527e-01
1.14933991e+00 -3.99326026e-01 -3.86086017e-01 -3.09680611e-01
-2.37854853e-01 -3.22936237e-01 -1.13411285e-01 -1.26098841e-01
9.99099255e-01 -1.03251445e+00 -8.43215466e-01 5.40480196e-01
2.42090300e-01 -4.10496354e-01 5.55365801e-01 8.89213383e-01
-4.74712819e-01 4.10907477e-01 -1.19714821e-02 -8.41439426e-01
-1.24287879e+00 2.43009984e-01 2.20379382e-02 -3.03213686e-01
-8.96425307e-01 1.06588268e+00 1.94327280e-01 -1.29329428e-01
1.87102798e-02 -1.21187247e-01 -2.21434105e-02 -4.00605142e-01
6.79815471e-01 1.74988091e-01 1.53564975e-01 6.28432631e-02
-3.78605068e-01 1.27182722e+00 4.88759637e-01 2.95571536e-01
1.79946733e+00 4.40507121e-02 -2.06094123e-02 -2.13781353e-02
9.99998033e-01 3.37091833e-01 -1.10745263e+00 3.27573791e-02
-2.67135352e-01 -6.24916732e-01 5.33211976e-02 -7.04206824e-01
-9.33613718e-01 7.29492128e-01 6.57326639e-01 5.00577390e-01
1.31706345e+00 3.47185820e-01 7.65742540e-01 6.02211177e-01
1.09361887e+00 -2.41178334e-01 5.31856239e-01 3.06856073e-02
8.19108009e-01 -1.28483129e+00 3.62678587e-01 -7.80634880e-01
-9.42941085e-02 9.12276626e-01 3.23001623e-01 -2.45414034e-01
4.32792693e-01 7.81898856e-01 -6.33794069e-02 -6.87188447e-01
-3.36413234e-01 3.53808790e-01 3.53554338e-01 9.08828616e-01
1.77422576e-02 -1.60440594e-01 -4.50110994e-02 6.79985285e-01
1.49816558e-01 -3.10223609e-01 1.56887695e-01 7.26587117e-01
-9.03001606e-01 -5.78809679e-01 -6.33581221e-01 5.80846488e-01
-4.06676263e-01 7.20842108e-02 3.86514485e-01 4.41820800e-01
-5.44348732e-02 9.92464781e-01 -3.88081431e-01 -2.12352917e-01
4.40999985e-01 -3.73825610e-01 6.35446012e-01 -3.61137837e-01
-1.91915870e-01 -6.66890442e-02 2.11348031e-02 -6.28801942e-01
-3.08490396e-01 -1.76092058e-01 -7.72172153e-01 2.37316713e-01
-4.29778963e-01 4.01754200e-01 7.95918524e-01 7.48316944e-01
2.81080097e-01 4.90612507e-01 8.79647672e-01 -7.03359604e-01
-8.79912078e-01 -7.73569405e-01 -7.49934435e-01 1.01698257e-01
3.83271962e-01 -5.62067926e-01 -8.32070112e-01 -2.72467524e-01] | [6.77614164352417, 0.8785808682441711] |
8f2be92b-40e8-48dc-ab9c-96ebb8ff6f05 | model-aware-contrastive-learning-towards | 2207.07874 | null | https://arxiv.org/abs/2207.07874v4 | https://arxiv.org/pdf/2207.07874v4.pdf | Model-Aware Contrastive Learning: Towards Escaping the Dilemmas | Contrastive learning (CL) continuously achieves significant breakthroughs across multiple domains. However, the most common InfoNCE-based methods suffer from some dilemmas, such as \textit{uniformity-tolerance dilemma} (UTD) and \textit{gradient reduction}, both of which are related to a $\mathcal{P}_{ij}$ term. It has been identified that UTD can lead to unexpected performance degradation. We argue that the fixity of temperature is to blame for UTD. To tackle this challenge, we enrich the CL loss family by presenting a Model-Aware Contrastive Learning (MACL) strategy, whose temperature is adaptive to the magnitude of alignment that reflects the basic confidence of the instance discrimination task, then enables CL loss to adjust the penalty strength for hard negatives adaptively. Regarding another dilemma, the gradient reduction issue, we derive the limits of an involved gradient scaling factor, which allows us to explain from a unified perspective why some recent approaches are effective with fewer negative samples, and summarily present a gradient reweighting to escape this dilemma. Extensive remarkable empirical results in vision, sentence, and graph modality validate our approach's general improvement for representation learning and downstream tasks. | ['Chunlin Chen', 'Huaxiong Li', 'Ziqi Wen', 'Haoxing Chen', 'Bo wang', 'Chao Zhang', 'Zizheng Huang'] | 2022-07-16 | null | null | null | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 5.51609635e-01 3.42553742e-02 -1.21911153e-01 -4.76701051e-01
-6.34234190e-01 -3.20910066e-01 4.21312213e-01 4.01236057e-01
-7.04115272e-01 6.54264867e-01 -1.49114519e-01 -4.00132090e-01
-3.72115076e-01 -5.21740437e-01 -6.31300867e-01 -9.18879330e-01
-1.37435002e-02 -1.24807134e-01 2.94503272e-01 -3.93919140e-01
6.00967944e-01 2.80058116e-01 -1.29518485e+00 1.47953287e-01
1.10124683e+00 1.23244274e+00 2.61500686e-01 2.49109089e-01
-2.31944606e-01 6.78944886e-01 -5.66550732e-01 -5.77719033e-01
2.47399628e-01 -5.53216279e-01 -6.78703070e-01 -3.29385370e-01
3.98532242e-01 1.77257042e-02 1.79172046e-02 1.37327409e+00
7.07136810e-01 -2.73702666e-02 8.24779630e-01 -1.30754173e+00
-6.89908922e-01 5.90252519e-01 -9.16718662e-01 6.40991688e-01
-1.46714687e-01 1.41600117e-01 1.11740065e+00 -7.64475942e-01
3.04143161e-01 1.17389500e+00 6.52582884e-01 5.51161826e-01
-1.03055263e+00 -3.70921195e-01 6.38949990e-01 3.72334778e-01
-1.22457576e+00 -2.00508624e-01 8.41769397e-01 -3.13779652e-01
6.49063528e-01 3.99068624e-01 3.57365429e-01 1.03740752e+00
3.12701702e-01 8.77364933e-01 1.24496639e+00 -6.24292791e-01
3.80825609e-01 1.77255064e-01 2.22904533e-01 8.08308899e-01
3.08255792e-01 -3.88530828e-02 -5.15336752e-01 1.62180979e-02
6.26999021e-01 -3.28940451e-01 -4.89905059e-01 -2.63765872e-01
-7.27690279e-01 7.34834790e-01 4.37945485e-01 1.35062501e-01
1.45405354e-02 4.00756262e-02 4.05015707e-01 5.81839681e-01
5.42830408e-01 5.30672252e-01 -4.71608788e-01 -5.25468215e-03
-6.02415800e-01 2.35852841e-02 4.74590868e-01 5.90766311e-01
6.40379608e-01 8.26627463e-02 -4.86075640e-01 1.01010025e+00
2.61243373e-01 1.83059260e-01 5.68841338e-01 -6.61829650e-01
6.63342834e-01 6.78117990e-01 -2.42784217e-01 -1.07886231e+00
-5.39146185e-01 -9.70826089e-01 -9.56980407e-01 2.88875639e-01
4.46504056e-01 -7.24823698e-02 -6.68836832e-01 2.05711126e+00
6.31456152e-02 -1.05885871e-01 -3.89888138e-01 1.00066936e+00
5.05158901e-01 1.89894468e-01 2.05724075e-01 -4.82877016e-01
1.13779306e+00 -7.99720347e-01 -5.28877378e-01 -5.26703775e-01
6.08717442e-01 -3.62779737e-01 1.56644714e+00 4.33953464e-01
-9.17065561e-01 -3.29528898e-01 -1.30964839e+00 4.45791930e-02
-2.43089154e-01 -8.69078562e-02 6.16626561e-01 7.53645539e-01
-9.07523096e-01 8.60422015e-01 -4.62446243e-01 -2.73751859e-02
4.79036808e-01 3.75932813e-01 -5.70322275e-02 7.00713769e-02
-1.10363722e+00 8.69310975e-01 2.24047914e-01 4.17666256e-01
-3.05816919e-01 -5.34055531e-01 -5.90716898e-01 1.23944536e-01
5.05228043e-01 -7.00920761e-01 6.95864975e-01 -1.07699847e+00
-1.43565381e+00 9.19294119e-01 8.18177089e-02 -5.33679128e-01
8.38697255e-01 -2.08335921e-01 -2.42896363e-01 -1.10927895e-01
-2.75195748e-01 4.92964536e-01 1.19569206e+00 -1.16643131e+00
-2.03438506e-01 -5.84179640e-01 1.78963263e-02 3.83964151e-01
-6.33031964e-01 -1.87194243e-01 -2.85496891e-01 -6.97159350e-01
3.00646067e-01 -7.11932003e-01 -2.48272538e-01 9.03094038e-02
-2.69020736e-01 -3.52344930e-01 2.20546886e-01 -2.70872295e-01
1.55856240e+00 -2.23881507e+00 1.85077712e-01 1.58964172e-01
3.15733910e-01 3.07067275e-01 -1.76264569e-01 1.69165581e-01
-1.22459091e-01 1.52440667e-01 -4.71828103e-01 -2.66813040e-01
1.14191300e-03 1.21135311e-03 -3.36123973e-01 4.85951006e-01
3.42259675e-01 7.93602288e-01 -8.53472710e-01 -5.30030727e-01
-1.40011594e-01 3.59982520e-01 -6.49713933e-01 -7.66754523e-02
-1.92921400e-01 2.20879287e-01 -3.71726543e-01 3.52045596e-01
7.41541505e-01 -4.02718663e-01 2.21671417e-01 -4.40976441e-01
-3.79881784e-02 1.18419439e-01 -1.24888587e+00 1.27522492e+00
2.51811873e-02 3.71596277e-01 -4.11159534e-04 -1.31165969e+00
8.96774590e-01 -3.34752828e-01 8.37282743e-03 -8.87282431e-01
2.06190243e-01 2.04617605e-01 1.07364029e-01 -3.76373261e-01
3.30880225e-01 -1.67235404e-01 1.20596014e-01 9.29763988e-02
-2.09834322e-01 -6.40981272e-02 -6.62976280e-02 2.24780124e-02
1.08785212e+00 7.50535354e-02 2.81861722e-01 -4.73760337e-01
5.94376504e-01 -4.51758564e-01 6.54403031e-01 9.80294466e-01
-5.50744832e-01 6.78469956e-01 9.50051546e-01 -2.74071574e-01
-6.82836056e-01 -1.00764382e+00 -1.10564008e-01 1.27719426e+00
1.91144213e-01 -3.35088402e-01 -7.33270347e-01 -7.35715926e-01
-1.48905903e-01 5.75093985e-01 -6.62183702e-01 -5.32013357e-01
-5.69697142e-01 -1.39673018e+00 3.86749744e-01 4.35715199e-01
6.04150712e-01 -7.89304078e-01 -5.56082129e-01 -1.47769731e-02
7.32208490e-02 -7.38358319e-01 -3.32821161e-01 4.53343391e-01
-8.36984813e-01 -8.56680036e-01 -6.27163112e-01 -6.38682067e-01
7.00537980e-01 1.30241469e-01 1.02784097e+00 1.49663478e-01
-3.11463028e-01 2.25701362e-01 -2.23678440e-01 -3.48318696e-01
-2.47691572e-01 1.31506324e-01 3.50875705e-02 -7.66669586e-02
1.80903643e-01 -6.27900720e-01 -8.31741869e-01 1.13646083e-01
-8.63216639e-01 -1.00161642e-01 6.44617617e-01 9.86447990e-01
5.45118988e-01 -1.52893767e-01 8.35671544e-01 -9.01238561e-01
9.86231387e-01 -3.49509090e-01 -4.77314323e-01 4.70164597e-01
-1.12437987e+00 3.62787694e-01 5.63288629e-01 -4.12118524e-01
-8.22845221e-01 -3.76807690e-01 -1.28672659e-01 -2.59380877e-01
3.10514480e-01 4.77338165e-01 -6.33945838e-02 -4.80206534e-02
6.38444543e-01 3.15656424e-01 -4.83291931e-02 -3.74029577e-01
2.72511542e-01 4.15516853e-01 3.00633699e-01 -6.70944750e-01
4.89423454e-01 3.21682245e-01 1.42382041e-01 -6.94292724e-01
-1.08502150e+00 7.05820993e-02 -3.30168337e-01 -1.53940275e-01
4.89721894e-01 -6.81963861e-01 -8.71037006e-01 5.17627001e-01
-9.17607784e-01 -2.99760282e-01 -2.84670055e-01 3.35130930e-01
-3.67997050e-01 6.17517352e-01 -6.30402565e-01 -9.18486714e-01
-3.32820743e-01 -9.28557158e-01 6.01059973e-01 2.56088734e-01
1.31879106e-01 -8.90547514e-01 -1.67329967e-01 2.25763008e-01
6.18599474e-01 1.17683254e-01 1.39799893e+00 -4.57614094e-01
-2.02993765e-01 1.09957270e-01 -4.99430448e-01 7.70396173e-01
-1.02805883e-01 -2.17216730e-01 -9.78095353e-01 -3.87375623e-01
2.86304802e-01 -2.94203401e-01 1.17746329e+00 3.43356371e-01
1.46312177e+00 -2.58721769e-01 -2.79890709e-02 5.39025426e-01
1.37445366e+00 -8.90465826e-03 6.80612922e-01 3.05767804e-01
5.18002033e-01 5.56175292e-01 2.45918930e-01 4.72168714e-01
2.32466027e-01 6.43069208e-01 6.69062138e-01 2.50424985e-02
-1.20140307e-01 -1.61719054e-01 2.44460881e-01 8.26927960e-01
3.44972834e-02 -3.12892526e-01 -6.74427330e-01 1.23968296e-01
-1.73164606e+00 -7.63579071e-01 -7.45490368e-04 2.32313204e+00
9.20939505e-01 6.45574749e-01 -7.00347200e-02 1.98761851e-01
6.82830930e-01 3.91474038e-01 -7.58800507e-01 -3.98562759e-01
-3.60688984e-01 1.04735522e-02 3.78793746e-01 4.87389356e-01
-9.33308899e-01 7.21098959e-01 6.18191671e+00 1.13108361e+00
-1.36802208e+00 9.45985131e-03 1.01051807e+00 -9.92023479e-03
-5.06580234e-01 -1.31420046e-01 -6.70601726e-01 6.31947219e-01
5.87394238e-01 1.72190424e-02 3.34549308e-01 6.97981298e-01
-1.70532614e-02 -8.31489787e-02 -9.96175826e-01 1.05465126e+00
2.00188637e-01 -1.15160096e+00 1.24097660e-01 -9.27462578e-02
3.65529746e-01 -1.11402199e-02 4.56879705e-01 5.49415827e-01
-1.71415433e-01 -9.42957759e-01 5.89139342e-01 3.69491458e-01
6.63781404e-01 -6.43845499e-01 5.39746523e-01 4.12987769e-01
-8.98302555e-01 -3.47653329e-01 -4.32620138e-01 -2.43993416e-01
-1.79331988e-01 8.36255789e-01 -5.45908093e-01 4.04701620e-01
6.31751359e-01 3.35529625e-01 -8.70762944e-01 8.53316247e-01
-9.69174206e-02 5.72490990e-01 -2.05299959e-01 -2.23832831e-01
9.46035981e-02 -3.40268433e-01 5.43984473e-01 1.16950643e+00
1.20845415e-01 -7.60402754e-02 -7.89746642e-02 8.69019747e-01
-1.04553118e-01 2.37935066e-01 -2.64850110e-01 1.91648126e-01
4.83918369e-01 1.03056490e+00 -8.22310627e-01 -1.82426721e-02
-9.12690163e-02 1.03684294e+00 6.91333950e-01 2.10785985e-01
-8.20872009e-01 -3.08963448e-01 3.25704783e-01 1.20234430e-01
1.89645812e-01 -9.76224393e-02 -5.04388630e-01 -1.09057200e+00
3.99344742e-01 -6.68022931e-01 3.98246258e-01 -3.97416383e-01
-1.43408823e+00 5.36723793e-01 -4.30846296e-04 -1.19324315e+00
1.05000578e-01 -6.76264286e-01 -4.77400512e-01 4.84085470e-01
-1.73069012e+00 -5.89611232e-01 -1.00460500e-01 4.02258575e-01
3.06446880e-01 -4.84409891e-02 5.05982280e-01 4.33589071e-01
-8.25396001e-01 1.02991986e+00 -1.63485557e-02 -1.32481098e-01
6.56209946e-01 -1.23187506e+00 1.27038881e-01 6.50832534e-01
-9.13608521e-02 5.08280933e-01 8.10770571e-01 -2.43488953e-01
-1.35436296e+00 -7.75542021e-01 6.85237646e-01 -3.32065642e-01
5.76551557e-01 -3.52168202e-01 -9.83084142e-01 2.08660841e-01
-3.06859165e-02 1.56504154e-01 4.45895195e-01 -4.51836996e-02
-6.18668795e-01 -3.54061812e-01 -1.14478838e+00 6.96706235e-01
1.36642587e+00 -3.19647759e-01 -4.44399655e-01 3.06073159e-01
7.19338596e-01 -5.58972880e-02 -4.92921025e-01 5.72646797e-01
3.36504519e-01 -1.31429350e+00 8.05277109e-01 -5.45024335e-01
3.53027582e-01 -1.09170705e-01 -2.13715062e-01 -1.18646812e+00
-2.31311992e-01 -7.14359760e-01 -1.07279003e-01 1.20750296e+00
3.66303056e-01 -8.32526267e-01 4.97080475e-01 4.15343493e-01
-1.26618803e-01 -9.76976156e-01 -1.26290381e+00 -8.28360617e-01
3.26327831e-01 -3.81749094e-01 2.21902952e-01 9.14073169e-01
-9.69915837e-02 5.51818848e-01 -3.19858342e-01 -9.90203246e-02
5.93364537e-01 -1.82608917e-01 2.34764650e-01 -1.24462771e+00
-4.99166846e-01 -8.05518091e-01 -2.64742881e-01 -1.18145955e+00
-1.27273006e-03 -7.35451818e-01 -1.22170918e-01 -1.18055153e+00
2.01318830e-01 -4.51940149e-01 -7.18037307e-01 3.55370283e-01
-4.21477795e-01 5.75268045e-02 2.48787224e-01 1.23610198e-01
-7.01612234e-01 6.44061446e-01 1.09101343e+00 -1.19673684e-01
-1.74663931e-01 -1.52719423e-01 -9.26174283e-01 6.69620752e-01
6.56614006e-01 -3.75241667e-01 -4.84091878e-01 -4.99998182e-01
8.99330854e-01 -2.48603746e-01 3.71356517e-01 -9.26341236e-01
1.88284695e-01 -9.06203538e-02 2.05697820e-01 -2.43423238e-01
2.43275166e-01 -4.63094741e-01 -4.61707681e-01 6.16136432e-01
-5.42651415e-01 6.41207695e-02 9.11501646e-02 7.35822141e-01
2.40450585e-03 -3.16940129e-01 9.05883074e-01 -1.10494941e-01
-5.40890098e-01 1.16087556e-01 -3.89007628e-02 4.15446758e-01
6.15294576e-01 -1.10618599e-01 -5.40163398e-01 -1.42224431e-01
-4.49657708e-01 1.98656827e-01 1.35114461e-01 3.83062243e-01
5.32242239e-01 -1.05517197e+00 -8.07109296e-01 3.87166709e-01
8.59289989e-02 -2.22726658e-01 4.25636530e-01 1.06653190e+00
-1.31804809e-01 7.92672932e-02 4.77387458e-02 -5.71833968e-01
-9.57732856e-01 4.54062700e-01 4.50251162e-01 -2.50542551e-01
-5.52363634e-01 1.18849719e+00 1.81925699e-01 -1.34668186e-01
4.11119521e-01 -3.11932802e-01 -1.70179546e-01 2.08556920e-01
4.67558652e-01 3.27879816e-01 2.62142539e-01 -7.72676915e-02
-5.70369661e-01 5.26002824e-01 -2.41776064e-01 1.45233780e-01
1.29677331e+00 -2.58961618e-01 -1.00576028e-01 4.80906725e-01
1.12726820e+00 -1.54870614e-01 -1.56224048e+00 -2.34989211e-01
1.21548399e-01 -2.04983771e-01 -4.59438674e-02 -8.19293261e-01
-9.10488784e-01 9.02637541e-01 9.09357667e-01 4.61868525e-01
1.11992407e+00 -1.39627382e-01 4.42896456e-01 4.16213900e-01
3.20198655e-01 -1.27813029e+00 1.93681255e-01 5.44449747e-01
8.59974563e-01 -1.38184667e+00 1.43289775e-01 -2.85988510e-01
-5.59218585e-01 9.51725960e-01 6.38499498e-01 -4.11623158e-02
6.09670758e-01 1.98842004e-01 -1.64583012e-01 -4.04559858e-02
-7.09725618e-01 7.35133508e-05 2.00023785e-01 3.66865367e-01
3.51486474e-01 -5.97190484e-02 -7.57023573e-01 6.01255715e-01
-3.68868113e-02 -2.37694964e-01 2.30625093e-01 7.72924066e-01
-5.94929993e-01 -1.01043594e+00 -2.73717660e-02 4.52001452e-01
-3.97443533e-01 -3.16620320e-01 -2.89958656e-01 6.83522940e-01
1.20158918e-01 7.46679127e-01 1.97402388e-02 -2.42822394e-01
2.40376905e-01 1.02542169e-01 7.01955914e-01 -2.72158146e-01
-5.52130282e-01 -1.55293941e-01 -1.77561343e-01 -3.23064655e-01
-2.90993512e-01 -2.94555902e-01 -1.02069664e+00 -2.89866831e-02
-3.26564103e-01 5.24148494e-02 4.51320857e-01 1.06803918e+00
4.04907733e-01 4.52165902e-01 6.82716310e-01 -4.50137973e-01
-1.16389096e+00 -8.09019208e-01 -5.42865038e-01 3.69014591e-01
3.15640748e-01 -6.15563810e-01 -8.24018240e-01 -4.05624866e-01] | [9.177568435668945, 3.2610669136047363] |
fc2f1c73-61ee-4055-ade1-f17356d15d51 | nuclick-from-clicks-in-the-nuclei-to-nuclear | 1909.03253 | null | https://arxiv.org/abs/1909.03253v1 | https://arxiv.org/pdf/1909.03253v1.pdf | NuClick: From Clicks in the Nuclei to Nuclear Boundaries | Best performing nuclear segmentation methods are based on deep learning algorithms that require a large amount of annotated data. However, collecting annotations for nuclear segmentation is a very labor-intensive and time-consuming task. Thereby, providing a tool that can facilitate and speed up this procedure is very demanding. Here we propose a simple yet efficient framework based on convolutional neural networks, named NuClick, which can precisely segment nuclei boundaries by accepting a single point position (or click) inside each nucleus. Based on the clicked positions, inclusion and exclusion maps are generated which comprise 2D Gaussian distributions centered on those positions. These maps serve as guiding signals for the network as they are concatenated to the input image. The inclusion map focuses on the desired nucleus while the exclusion map indicates neighboring nuclei and improve the results of segmentation in scenes with nuclei clutter. The NuClick not only facilitates collecting more annotation from unseen data but also leads to superior segmentation output for deep models. It is also worth mentioning that an instance segmentation model trained on NuClick generated labels was able to rank first in LYON19 challenge. | ['Navid Alemi Koohbanani', 'Mostafa Jahanifar', 'Nasir Rajpoot'] | 2019-09-07 | null | null | null | null | ['nuclear-segmentation'] | ['medical'] | [ 3.57302241e-02 3.30697298e-01 6.80850595e-02 -6.19956136e-01
-8.74103427e-01 -7.01315641e-01 4.73435551e-01 4.51184720e-01
-8.04029167e-01 7.32571542e-01 -1.90722778e-01 -1.51775882e-01
2.60478109e-01 -8.53516877e-01 -8.20298851e-01 -9.48298395e-01
1.71993539e-01 9.77683961e-01 6.65451169e-01 1.36772692e-01
2.22807322e-02 6.59847498e-01 -1.33171070e+00 2.53805548e-01
8.80176067e-01 7.71954060e-01 6.19611740e-01 5.02260029e-01
-1.81157827e-01 6.25585951e-03 -6.49866581e-01 -2.47265697e-01
2.16460787e-02 -2.10990325e-01 -6.95107102e-01 -1.75490439e-01
1.18134096e-01 -1.81664854e-01 1.62318915e-01 9.90126669e-01
5.29979646e-01 1.05177216e-01 8.73223007e-01 -8.57240915e-01
-2.68240958e-01 8.61009061e-01 -2.96294451e-01 1.55327275e-01
-2.22471803e-01 1.50742859e-01 1.00814283e+00 -8.58929455e-01
7.69255042e-01 6.62961602e-01 5.02618253e-01 5.98809659e-01
-1.15765131e+00 -4.46580440e-01 -2.28459150e-01 -7.97534138e-02
-1.32846057e+00 -4.13879119e-02 3.54606241e-01 -5.32570124e-01
3.12960148e-01 5.31881869e-01 7.20808446e-01 8.05260956e-01
-9.10491049e-02 1.00174642e+00 7.25830972e-01 -1.59184515e-01
3.55659515e-01 1.14100166e-01 -2.74045877e-02 4.78452712e-01
6.01420514e-02 -3.97181660e-01 1.54401302e-01 1.85553446e-01
6.67938650e-01 -2.27033906e-02 -1.41364858e-01 -5.62541366e-01
-1.27641952e+00 8.86148691e-01 9.48890448e-01 6.59897506e-01
-2.69442320e-01 4.16527577e-02 5.53924084e-01 -5.30801952e-01
2.69685209e-01 4.44808513e-01 -3.60022306e-01 1.25428855e-01
-1.23177218e+00 1.82497278e-01 3.33459646e-01 5.60744226e-01
7.62367129e-01 -5.13865292e-01 -5.82686365e-01 8.32028925e-01
1.92725390e-01 2.10905477e-01 2.24434644e-01 -5.49420118e-01
1.70060173e-01 8.67859304e-01 2.56671488e-01 -6.88997447e-01
-1.07640839e+00 -9.75393653e-01 -9.15570617e-01 1.74109951e-01
9.37826335e-01 -2.19172031e-01 -1.27181876e+00 1.45628715e+00
5.67832947e-01 -7.63710365e-02 -2.47914091e-01 1.23507762e+00
1.10558271e+00 5.15239000e-01 2.81294346e-01 1.01894766e-01
1.45694113e+00 -9.65277076e-01 -6.50414824e-01 1.55541673e-01
1.01413715e+00 -6.82857096e-01 1.02648866e+00 2.42288485e-01
-9.50531781e-01 -5.44279575e-01 -8.65282297e-01 -2.65639156e-01
-6.06817901e-01 8.10999572e-01 5.65549076e-01 5.63008130e-01
-1.19709444e+00 5.90192735e-01 -1.02687550e+00 -2.60192335e-01
5.75674176e-01 5.57032883e-01 -3.00851822e-01 3.19751769e-01
-1.13146102e+00 6.99214041e-01 7.56256521e-01 4.89177763e-01
-8.21027160e-01 -5.44530630e-01 -6.60664856e-01 1.87020943e-01
5.52376032e-01 -2.99558133e-01 1.19062090e+00 -5.76266170e-01
-1.18135023e+00 9.81140435e-01 1.72783360e-01 -5.81017435e-01
8.55658531e-01 3.70198712e-02 1.54757977e-01 1.72742054e-01
3.46745819e-01 1.25796366e+00 4.26990032e-01 -1.15021658e+00
-6.52089119e-01 -2.47348115e-01 -3.13209504e-01 7.53653795e-02
-7.91315511e-02 -2.39462286e-01 -1.05867660e+00 -3.82562548e-01
2.59405315e-01 -9.51353014e-01 -3.94315630e-01 -4.65046205e-02
-9.50678885e-01 -5.36016703e-01 7.55006313e-01 -7.95033038e-01
1.05094278e+00 -2.01919913e+00 4.07883301e-02 4.71795499e-01
2.79026330e-01 3.94286543e-01 3.14563632e-01 2.04664115e-02
-1.16572887e-01 2.05420777e-01 -1.50719628e-01 -2.42624179e-01
1.06714360e-01 1.68929808e-02 3.63570005e-01 3.96843493e-01
6.13447279e-02 8.70517969e-01 -8.30018222e-01 -7.19715357e-01
3.75772357e-01 3.93553764e-01 -5.23029089e-01 2.19482239e-02
-5.65960467e-01 7.35920966e-01 -3.25198472e-01 5.82015157e-01
7.46953905e-01 -4.23708051e-01 -4.52375188e-02 -4.59904730e-01
-1.66943818e-01 -2.81638473e-01 -9.43139791e-01 1.65580559e+00
-2.31067520e-02 4.13005233e-01 9.45058689e-02 -9.52465177e-01
1.09663975e+00 2.97718883e-01 4.71983999e-01 -4.22812462e-01
3.93643290e-01 3.30128580e-01 -2.91987620e-02 -4.79478836e-01
3.82692456e-01 -5.47649898e-03 -1.03393413e-01 2.20196500e-01
-9.08175204e-03 -1.77566409e-01 7.54696250e-01 1.91451207e-01
9.40829992e-01 1.23027943e-01 -7.89253041e-02 -1.56684622e-01
6.80214226e-01 -1.06173046e-02 5.61962485e-01 6.95187747e-01
-1.23469986e-01 9.26692069e-01 7.79750764e-01 -4.85908329e-01
-1.23408937e+00 -8.75933886e-01 -3.29665065e-01 8.94976020e-01
-9.44482163e-02 -2.72635490e-01 -1.06127882e+00 -1.07770967e+00
-3.10664743e-01 6.19797885e-01 -9.17288184e-01 4.38531339e-01
-3.13973576e-01 -7.68199325e-01 4.45740551e-01 6.46166027e-01
5.31701565e-01 -1.22025824e+00 -4.68591720e-01 7.40909874e-02
-3.74540120e-01 -8.68311584e-01 -2.38987163e-01 5.39827585e-01
-5.59056401e-01 -9.70676303e-01 -1.26887965e+00 -7.71955073e-01
1.03156018e+00 -4.71767932e-01 8.29770982e-01 1.38858557e-01
-4.75201666e-01 -4.23642069e-01 -3.95276815e-01 -3.66933882e-01
-3.44626784e-01 5.96856356e-01 -5.20148456e-01 3.81157398e-02
1.65989235e-01 -8.76253322e-02 -7.37957478e-01 4.70649183e-01
-9.82147276e-01 3.65625322e-01 6.89860225e-01 6.68464780e-01
7.74324358e-01 -7.53775164e-02 4.38312918e-01 -1.07273757e+00
1.23484746e-01 -3.32870275e-01 -9.38695371e-01 3.36946845e-02
1.62368774e-01 1.45900855e-02 5.71960270e-01 4.89198267e-02
-7.58723319e-01 5.91629028e-01 -6.20075226e-01 -6.35558665e-02
-4.34697568e-01 4.09055173e-01 -1.45917907e-01 2.49072015e-01
6.49275184e-01 8.77588242e-02 -1.06884040e-01 -4.86777663e-01
3.35488677e-01 5.10256410e-01 6.37542367e-01 -2.25160211e-01
5.57303011e-01 5.56085110e-01 6.99475110e-02 -6.27202094e-01
-7.72974551e-01 -5.08699954e-01 -1.04443073e+00 -3.47053081e-01
1.28385353e+00 -3.62137258e-01 -7.90337265e-01 2.10005194e-01
-1.01857197e+00 -3.11643809e-01 -1.25794590e-01 4.20102566e-01
-3.88860881e-01 1.55256584e-01 -4.74489719e-01 -4.73316610e-01
-2.30072975e-01 -1.43747628e+00 1.08562505e+00 6.57214761e-01
-1.96518630e-01 -8.07388544e-01 -1.98395699e-01 4.67495561e-01
2.51594961e-01 2.83384919e-01 9.76426065e-01 -1.05768812e+00
-6.77711546e-01 -5.38921535e-01 -4.72988099e-01 7.83773735e-02
-1.50479421e-01 2.04457268e-01 -9.37665880e-01 -5.04172035e-02
-6.64808333e-01 -3.44470561e-01 8.42563868e-01 7.06114352e-01
1.46598423e+00 3.31311315e-01 -6.41339242e-01 5.00819147e-01
1.14953315e+00 1.59392029e-01 5.08316100e-01 3.49164516e-01
4.91665661e-01 5.28949976e-01 6.24341249e-01 2.44726330e-01
-2.49563344e-02 7.40021229e-01 6.80212259e-01 -7.64011621e-01
6.27004802e-02 7.04698935e-02 -3.87879401e-01 1.31276891e-01
-1.61639631e-01 -3.15082490e-01 -9.38886464e-01 5.68340063e-01
-1.69452155e+00 -4.62995142e-01 -3.75427961e-01 2.01343393e+00
9.40130830e-01 3.53046864e-01 1.38398230e-01 1.54365003e-01
8.27522337e-01 -2.82912672e-01 -5.31358004e-01 -6.43544272e-02
9.10793394e-02 1.62748918e-01 3.12630296e-01 2.03608140e-01
-1.46197343e+00 8.55109334e-01 5.35415173e+00 1.16192174e+00
-1.10597610e+00 6.61075786e-02 1.00051129e+00 9.97248217e-02
9.49878711e-03 -2.87847608e-01 -9.82048512e-01 6.09717786e-01
5.15557706e-01 5.94235420e-01 -2.56359160e-01 8.86091769e-01
4.45367575e-01 -5.62421620e-01 -1.00894225e+00 7.56406367e-01
-3.34760547e-01 -1.56419468e+00 -1.89480379e-01 7.58684725e-02
5.72827101e-01 1.31963911e-02 -1.65063754e-01 1.93906650e-01
2.24669173e-01 -1.00860083e+00 5.13420880e-01 5.74453712e-01
6.90390527e-01 -8.29248846e-01 1.21886671e+00 5.11287034e-01
-7.73013592e-01 2.47074962e-01 -4.64684665e-01 6.69250607e-01
2.02075139e-01 9.29015279e-01 -1.38526106e+00 3.42840165e-01
5.88539362e-01 9.54136774e-02 -6.75039232e-01 1.60672760e+00
-3.89069468e-01 6.01975977e-01 -5.21663249e-01 -1.80363178e-01
4.50615168e-01 -2.25627795e-01 2.69932419e-01 1.34217966e+00
4.03860539e-01 -2.80943453e-01 2.58003831e-01 1.17077434e+00
-1.78579673e-01 4.07573551e-01 -1.58532709e-01 -3.20014283e-02
3.19666974e-02 1.86023188e+00 -1.75966573e+00 -3.15189391e-01
-4.89834324e-02 8.15166235e-01 2.55352527e-01 2.32227892e-01
-1.13427484e+00 -4.89082187e-01 -1.85465291e-01 1.74219951e-01
3.82348567e-01 -3.58571522e-02 -3.99864137e-01 -4.84962136e-01
-3.29522669e-01 -3.25535327e-01 4.61350471e-01 -8.50612223e-01
-7.38599181e-01 5.81901014e-01 -1.75164193e-01 -8.94355536e-01
1.03529721e-01 -4.25441861e-01 -4.93292302e-01 7.71041155e-01
-9.71377611e-01 -1.13108361e+00 -5.41108072e-01 2.19630852e-01
4.72088873e-01 2.61042625e-01 6.99863851e-01 4.42568719e-01
-6.47685647e-01 4.19872671e-01 6.85211569e-02 4.83537257e-01
5.50642848e-01 -1.63133025e+00 -3.96403559e-02 5.07699788e-01
1.29212007e-01 3.98326784e-01 7.09675848e-01 -7.14498401e-01
-4.51308012e-01 -1.08284283e+00 8.71789753e-01 -1.75725430e-01
2.03763455e-01 -5.32169700e-01 -7.58434355e-01 4.41650689e-01
7.50801787e-02 1.47296995e-01 6.44125164e-01 9.25632268e-02
4.67125535e-01 5.87970130e-02 -1.07933056e+00 6.34967625e-01
3.95987153e-01 1.26108617e-01 -1.37864336e-01 8.11922312e-01
3.66920710e-01 -7.64127493e-01 -5.64265549e-01 3.34150791e-01
1.99008267e-02 -1.05321312e+00 7.77543008e-01 -1.70384929e-01
2.62026697e-01 -6.32403553e-01 4.79751527e-01 -1.19670761e+00
-1.14675134e-01 -2.94346631e-01 5.87480187e-01 1.07863355e+00
9.16082680e-01 -1.60262004e-01 1.22674775e+00 5.68705976e-01
-4.07394201e-01 -8.48271847e-01 -8.22569311e-01 -3.47463071e-01
-8.08913410e-02 -1.79436192e-01 4.59301651e-01 5.36516666e-01
-2.89340794e-01 1.24346986e-01 6.35659471e-02 1.97597489e-01
4.12471503e-01 -1.71551585e-01 6.11363113e-01 -1.15655577e+00
9.48850438e-02 -3.09592485e-01 -2.85636127e-01 -8.53840113e-01
-1.83301762e-01 -1.07724392e+00 2.32421681e-01 -1.80832231e+00
2.05458954e-01 -7.61332810e-01 -6.81666806e-02 6.96183860e-01
-1.12032779e-01 7.46608973e-01 9.61219668e-02 5.59913404e-02
-8.31598520e-01 2.56936431e-01 1.37588573e+00 -8.76068100e-02
-1.14495501e-01 5.21876395e-01 -2.35739484e-01 8.37652922e-01
7.99606025e-01 -2.90612310e-01 -8.62521753e-02 7.58788735e-03
2.40088701e-01 -2.74334736e-02 3.29349399e-01 -1.20660579e+00
3.36248010e-01 2.68132299e-01 9.46347356e-01 -1.32723606e+00
3.55672002e-01 -7.77613580e-01 8.28723833e-02 3.55954856e-01
-3.85172516e-01 -4.65180635e-01 5.13329133e-02 2.17324793e-01
-6.41610026e-02 -6.68711066e-01 8.65227401e-01 -3.34833115e-01
-3.50394458e-01 1.35808170e-01 -4.07124162e-01 -2.47095227e-01
1.25710154e+00 -2.55518705e-01 1.22929536e-01 -2.18759850e-01
-1.62992597e+00 5.40221393e-01 8.85013044e-02 -1.31289795e-01
1.33896321e-01 -1.09854424e+00 -4.08590347e-01 2.42970794e-01
-9.75439548e-02 5.01801968e-01 4.07746226e-01 1.01941955e+00
-8.15014780e-01 6.43531501e-01 -5.02384678e-02 -9.69903946e-01
-1.37018347e+00 3.49474102e-01 4.11175787e-01 -5.06590724e-01
-6.65607274e-01 1.24084723e+00 2.76555359e-01 -6.16128564e-01
3.79290879e-01 -7.34067857e-01 -6.30206704e-01 3.45869124e-01
3.08985025e-01 1.75153002e-01 3.00332993e-01 -3.91175658e-01
-2.10846663e-01 3.16583157e-01 -2.24414647e-01 -3.41814645e-02
1.22714281e+00 1.50021255e-01 -7.89762065e-02 1.36467502e-01
8.36414933e-01 -1.53690606e-01 -1.47588420e+00 8.22153836e-02
1.48574874e-01 -9.71480086e-02 -9.37379226e-02 -1.10784936e+00
-1.21037507e+00 9.13616657e-01 5.43765843e-01 2.96527475e-01
7.94704854e-01 2.13664532e-01 6.06924534e-01 1.08908668e-01
-4.66674194e-02 -1.29073560e+00 -1.05775788e-01 3.27651620e-01
6.63577259e-01 -1.10227621e+00 -2.61635631e-01 -5.36396742e-01
-5.13741314e-01 1.21933877e+00 6.67367637e-01 1.52009532e-01
2.84741193e-01 2.22517416e-01 1.67779714e-01 -3.46144348e-01
-1.18542634e-01 -3.34834397e-01 2.00089991e-01 6.16543055e-01
5.08987248e-01 1.05831623e-01 -6.28750980e-01 6.99634612e-01
-1.77527457e-01 7.76578998e-03 2.70581722e-01 5.15327394e-01
-6.46302283e-01 -1.06528890e+00 -5.37821829e-01 5.58341682e-01
-5.61360240e-01 1.52880251e-01 -1.86672047e-01 8.64243388e-01
5.15045106e-01 4.15101618e-01 9.21338648e-02 1.57619759e-01
2.84909527e-03 6.65644705e-02 -7.07581341e-02 -7.50589967e-01
-5.45317054e-01 5.27903259e-01 -3.53957377e-02 -2.94023603e-01
4.50152485e-03 -4.64502364e-01 -1.85049987e+00 3.07832122e-01
-4.73710239e-01 5.24380565e-01 8.79158437e-01 1.03929377e+00
-9.08492208e-02 6.94399118e-01 1.98059306e-01 -1.09822893e+00
-2.64109433e-01 -1.05281234e+00 -6.09683871e-01 4.43490505e-01
-7.90930092e-02 -7.09931552e-01 -1.29583985e-01 8.32016915e-02] | [14.733368873596191, -2.732520818710327] |
58742a1d-1837-44d7-9f57-403af8a2e280 | diffswap-high-fidelity-and-controllable-face | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Zhao_DiffSwap_High-Fidelity_and_Controllable_Face_Swapping_via_3D-Aware_Masked_Diffusion_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Zhao_DiffSwap_High-Fidelity_and_Controllable_Face_Swapping_via_3D-Aware_Masked_Diffusion_CVPR_2023_paper.pdf | DiffSwap: High-Fidelity and Controllable Face Swapping via 3D-Aware Masked Diffusion | In this paper, we propose DiffSwap, a diffusion model based framework for high-fidelity and controllable face swapping. Unlike previous work that relies on carefully designed network architectures and loss functions to fuse the information from the source and target faces, we reformulate the face swapping as a conditional inpainting task, performed by a powerful diffusion model guided by the desired face attributes (e.g., identity and landmarks). An important issue that makes it nontrivial to apply diffusion models to face swapping is that we cannot perform the time-consuming multi-step sampling to obtain the generated image during training. To overcome this, we propose a midpoint estimation method to efficiently recover a reasonable diffusion result of the swapped face with only 2 steps, which enables us to introduce identity constraints to improve the face swapping quality. Our framework enjoys several favorable properties more appealing than prior arts: 1) Controllable. Our method is based on conditional masked diffusion on the latent space, where the mask and the conditions can be fully controlled and customized. 2) High-fidelity. The formulation of conditional inpainting can fully exploit the generative ability of diffusion models and can preserve the background of target images with minimal artifacts. 3) Shape-preserving. The controllability of our method enables us to use 3D-aware landmarks as the condition during generation to preserve the shape of the source face. Extensive experiments on both FF++ and FFHQ demonstrate that our method can achieve state-of-the-art face swapping results both qualitatively and quantitatively. | ['Jiwen Lu', 'Jie zhou', 'Zuyan Liu', 'Weikang Shi', 'Yongming Rao', 'Wenliang Zhao'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['face-swapping'] | ['computer-vision'] | [ 2.49671742e-01 2.96271265e-01 -5.19122481e-02 -2.98822105e-01
-4.82497275e-01 -5.46573341e-01 5.49986303e-01 -6.88226640e-01
7.62917995e-02 7.32536852e-01 1.73086375e-01 2.38460779e-01
-1.26523346e-01 -8.45547616e-01 -6.94055021e-01 -9.39731002e-01
3.63077104e-01 2.08998948e-01 -1.55973285e-01 -2.17822924e-01
-6.90025389e-02 8.37930441e-01 -1.46502209e+00 5.18706581e-03
1.15215111e+00 1.06411326e+00 2.37886295e-01 8.17680880e-02
6.12026826e-02 4.43793893e-01 -3.75821620e-01 -5.06707191e-01
6.26740277e-01 -8.40413570e-01 -3.09658676e-01 3.45003426e-01
6.61428392e-01 -4.89517152e-01 -4.07340497e-01 1.06234121e+00
5.65521777e-01 -4.67871688e-02 5.91586471e-01 -1.41165233e+00
-1.10680938e+00 2.11888447e-01 -8.77707124e-01 -3.94411385e-01
2.26149336e-01 3.06777507e-01 6.63831413e-01 -1.01193476e+00
8.02088916e-01 1.29553759e+00 5.63810706e-01 8.72913241e-01
-1.44434834e+00 -1.01959467e+00 4.72249053e-02 -1.06772862e-01
-1.51679730e+00 -7.57680714e-01 1.11944509e+00 -4.36474264e-01
9.06169713e-02 2.66685963e-01 6.66306674e-01 1.02979577e+00
-1.66743677e-02 4.58747894e-01 1.03981948e+00 -3.45397353e-01
4.78591546e-02 -1.92853007e-02 -6.39354944e-01 1.00519872e+00
-1.16735576e-02 1.86719343e-01 -5.44121265e-01 -5.21729775e-02
1.32167828e+00 5.95106259e-02 -6.93299890e-01 -4.95564908e-01
-1.12630475e+00 8.07596028e-01 4.79077756e-01 2.83700138e-01
-4.54371989e-01 9.33507308e-02 -1.34129345e-01 2.47604862e-01
6.23877347e-01 2.47130707e-01 1.36749474e-02 3.40975225e-01
-1.25335503e+00 1.41237319e-01 2.83215195e-01 9.44000125e-01
8.79610896e-01 2.80941695e-01 -4.21623588e-01 9.17262137e-01
3.33625317e-01 6.48343086e-01 1.81670040e-01 -1.31013632e+00
2.64113247e-01 3.01917672e-01 1.51323035e-01 -1.16453147e+00
2.38875479e-01 -2.27389812e-01 -1.08938324e+00 4.22286987e-01
1.43943965e-01 -1.12899683e-01 -1.10507214e+00 2.16335607e+00
4.52806711e-01 2.75092959e-01 -3.07553083e-01 6.81079209e-01
6.01566732e-01 5.25437176e-01 -3.13843846e-01 -5.32674074e-01
1.06822455e+00 -9.61030722e-01 -1.00051308e+00 -6.99661821e-02
1.02304533e-01 -8.44584107e-01 1.03708744e+00 -5.88479564e-02
-1.40128648e+00 -5.33465862e-01 -9.72076237e-01 -1.65924177e-01
2.46466830e-01 1.96374968e-01 4.57029283e-01 6.02306247e-01
-1.26192248e+00 6.56887412e-01 -7.59021044e-01 6.51623532e-02
7.18166411e-01 5.31015992e-01 -7.59924114e-01 -2.12300614e-01
-9.43958521e-01 5.20510674e-01 -1.70394033e-01 2.38053247e-01
-9.57471967e-01 -9.18148339e-01 -8.41706514e-01 1.23854809e-01
3.17700535e-01 -8.89465511e-01 7.92691767e-01 -1.19730091e+00
-1.80337906e+00 8.15429270e-01 -4.08068419e-01 7.99691901e-02
9.19025958e-01 -1.82039570e-02 -7.79847056e-02 2.37794876e-01
3.21123272e-01 7.45856881e-01 1.34384072e+00 -1.44503844e+00
-2.09659323e-01 -2.68462956e-01 -2.20450938e-01 -3.89252491e-02
-6.15296185e-01 -2.41484523e-01 -7.58211315e-01 -9.98519361e-01
1.58803016e-01 -9.79832590e-01 -2.17630211e-02 7.18943894e-01
-3.55753154e-01 2.59858698e-01 1.06972063e+00 -7.85630047e-01
9.11665440e-01 -2.39171219e+00 3.31921846e-01 2.31065631e-01
2.89812595e-01 2.08347648e-01 -4.49622065e-01 3.11515033e-01
-8.90395939e-02 4.34204713e-02 -5.19903064e-01 -8.87988806e-01
-1.23402633e-01 2.43745372e-01 -5.65329552e-01 5.09355068e-01
2.31003836e-01 9.49630082e-01 -6.96722209e-01 -4.84232515e-01
8.53798166e-02 9.89826858e-01 -7.02677071e-01 3.90042841e-01
-7.05964565e-02 9.17755246e-01 -3.49193513e-01 5.22978246e-01
1.06505477e+00 -2.82382313e-02 1.16483547e-01 -4.22769815e-01
8.03164095e-02 -2.43841678e-01 -1.06450021e+00 1.84850013e+00
-5.25275886e-01 2.80224562e-01 5.09295642e-01 -4.15616393e-01
9.10647869e-01 3.80613267e-01 5.01341581e-01 -5.00634313e-01
-3.90330143e-02 2.90558845e-01 -3.79227996e-01 -1.94338858e-01
8.94634947e-02 -2.90770054e-01 4.15659219e-01 4.42084014e-01
-2.18656231e-02 -2.39462391e-01 -6.55478984e-02 1.47043139e-01
5.97912192e-01 2.88851023e-01 -2.84015507e-01 -3.33831996e-01
5.01120806e-01 -6.54284298e-01 9.79564786e-01 1.85610220e-01
1.02895461e-01 1.05177701e+00 5.05886376e-01 -9.80521366e-02
-8.92512381e-01 -1.03146136e+00 3.63834649e-02 5.41651368e-01
9.67880562e-02 -1.61495835e-01 -1.02460718e+00 -6.40997708e-01
-9.75013003e-02 4.53170627e-01 -8.47838879e-01 -2.18558401e-01
-7.33115852e-01 -4.42943156e-01 3.57614309e-01 5.60237408e-01
7.78800964e-01 -8.02715242e-01 -1.68671887e-02 1.16857868e-02
-2.82682836e-01 -8.88418496e-01 -1.14272928e+00 -4.37013179e-01
-7.39777386e-01 -8.24877679e-01 -1.08220017e+00 -1.06403041e+00
1.20709896e+00 3.12607944e-01 6.40289485e-01 2.44606629e-01
1.74596766e-03 1.22750185e-01 4.51747626e-02 1.86164975e-01
-3.15549314e-01 -2.86233693e-01 5.17682508e-02 5.61279297e-01
-5.07579744e-01 -1.03934300e+00 -8.41524482e-01 4.61531967e-01
-1.09557199e+00 2.69437790e-01 4.78800982e-01 8.90126288e-01
6.24214292e-01 1.59422413e-01 4.49412644e-01 -7.10321724e-01
4.24331814e-01 -8.00295919e-02 -6.22464120e-01 3.50295961e-01
-4.81880367e-01 1.57421306e-02 5.68221509e-01 -6.55692101e-01
-1.28202403e+00 2.44253784e-01 -1.72955394e-01 -8.46167088e-01
3.79916906e-01 -1.06206372e-01 -5.86653173e-01 -5.20424783e-01
2.76048064e-01 2.58817524e-01 3.34553421e-01 -4.37620848e-01
6.38443947e-01 2.17537686e-01 5.35692990e-01 -6.32770479e-01
1.14531171e+00 8.64185810e-01 7.58716390e-02 -6.52829409e-01
-4.46174204e-01 3.11040193e-01 -6.11328542e-01 -1.08638212e-01
7.30891168e-01 -8.97689402e-01 -7.76084721e-01 6.75551474e-01
-1.06329870e+00 -3.40111971e-01 -3.94589990e-01 1.00823715e-01
-5.13200521e-01 4.22716349e-01 -6.45666003e-01 -6.81046844e-01
-2.47221231e-01 -1.17840767e+00 1.14014471e+00 1.91839740e-01
9.68242355e-04 -8.86873484e-01 -1.26586810e-01 3.33620220e-01
6.19769156e-01 4.54190910e-01 8.24710071e-01 3.13426405e-01
-8.46988201e-01 9.52495933e-02 -1.48052737e-01 4.37085181e-01
6.25488400e-01 1.08440362e-01 -8.81122470e-01 -5.33983529e-01
2.84320235e-01 -1.85588703e-01 8.86206806e-01 2.81044960e-01
1.03568733e+00 -4.34373260e-01 -4.71282363e-01 9.38015342e-01
1.14142668e+00 7.21531287e-02 7.55505204e-01 -2.37257734e-01
9.28086638e-01 8.30910742e-01 3.79120886e-01 2.78795153e-01
1.59418568e-01 7.06246257e-01 2.96962470e-01 -2.92580724e-01
-5.21752238e-01 -6.70320153e-01 3.22624683e-01 5.57879388e-01
-5.96000999e-02 -1.65377155e-01 -5.04956126e-01 5.54263294e-01
-1.72230458e+00 -8.63707900e-01 3.25655282e-01 2.23641443e+00
1.01749241e+00 -3.30508500e-01 -1.96672440e-01 -4.78411801e-02
8.81848097e-01 2.75008142e-01 -4.70179617e-01 2.62145311e-01
-9.31850299e-02 3.31720740e-01 6.00585341e-02 7.47659624e-01
-6.56705379e-01 9.36612844e-01 5.82261324e+00 9.86416638e-01
-1.24659479e+00 2.05988765e-01 8.24028671e-01 -3.40248168e-01
-7.05049396e-01 -3.95547450e-02 -5.38047194e-01 5.73188007e-01
1.94144547e-01 -1.02733765e-02 5.16457379e-01 5.91337919e-01
2.36279160e-01 3.29256147e-01 -9.06188369e-01 1.01573467e+00
1.72609285e-01 -1.50360346e+00 2.09155753e-01 2.02447489e-01
1.07490373e+00 -6.90711200e-01 5.60794652e-01 -1.94992945e-01
1.65466830e-01 -1.01669908e+00 8.34905326e-01 5.70961535e-01
1.22262442e+00 -8.52569997e-01 7.73433894e-02 6.07028604e-02
-1.14438760e+00 8.54050964e-02 -1.63507387e-01 4.27728772e-01
4.23863262e-01 8.26407850e-01 -2.25450471e-01 5.40261447e-01
4.81484026e-01 5.97472429e-01 -1.41878083e-01 5.51815391e-01
-5.92198014e-01 1.60341337e-01 -2.58677095e-01 6.59657359e-01
-2.04330072e-01 -5.02146900e-01 4.92125720e-01 4.53568727e-01
6.35486662e-01 4.06902060e-02 -3.03378459e-02 1.27813208e+00
-4.13029224e-01 -5.24326228e-02 -5.74579954e-01 -2.04754416e-02
5.26327848e-01 1.28563201e+00 -5.67203224e-01 -2.56104916e-02
-1.47161260e-01 1.42329013e+00 1.94649562e-01 4.47171718e-01
-8.81657124e-01 -2.18944535e-01 7.30056882e-01 4.18517441e-01
6.43648624e-01 -2.40224317e-01 -1.08448565e-01 -1.20755649e+00
3.44227791e-01 -8.29176128e-01 -1.88166082e-01 -6.11554086e-01
-1.32267594e+00 8.80380332e-01 -2.82960027e-01 -1.03123879e+00
-3.46497037e-02 -8.20270628e-02 -7.36130595e-01 1.04077530e+00
-1.59005952e+00 -1.69461262e+00 -3.77134442e-01 8.03117871e-01
1.54206052e-01 -7.81308785e-02 6.00184917e-01 4.35499370e-01
-5.16068637e-01 7.61476517e-01 -6.31340221e-02 1.05364267e-02
9.05511975e-01 -7.27089226e-01 3.22528452e-01 8.90688002e-01
-1.59635022e-02 8.36689532e-01 4.11926270e-01 -7.90045679e-01
-1.26912832e+00 -1.12732363e+00 7.90214837e-01 -2.36388028e-01
2.02880472e-01 -4.72804368e-01 -9.09552395e-01 6.96633637e-01
1.99813604e-01 2.39703625e-01 4.77426916e-01 -4.24778283e-01
-4.08093572e-01 -3.38719875e-01 -1.39556777e+00 7.49421239e-01
1.36399114e+00 -6.41707420e-01 -1.35466561e-01 3.44900578e-01
6.47684455e-01 -3.22581768e-01 -6.24720335e-01 4.59633887e-01
4.78233010e-01 -1.07566392e+00 8.64797294e-01 -6.90827817e-02
4.12480980e-01 -4.68401462e-01 2.83931382e-02 -1.16968191e+00
-3.65413934e-01 -1.26568949e+00 -7.31556639e-02 1.66894054e+00
1.19162850e-01 -8.34063888e-01 7.58739233e-01 7.62454271e-01
8.26180056e-02 -7.26082981e-01 -9.55114841e-01 -7.58650959e-01
7.12920278e-02 1.87683180e-01 8.92728269e-01 1.02459502e+00
-5.98051488e-01 -2.70687491e-02 -6.63141310e-01 2.12585300e-01
6.41462505e-01 3.64045411e-01 6.74034894e-01 -1.00887692e+00
-3.80592555e-01 -3.04676026e-01 1.15522798e-02 -1.19965124e+00
3.44330251e-01 -5.91238558e-01 -3.84561270e-02 -1.19637811e+00
3.26085798e-02 -5.98326564e-01 8.47171023e-02 7.14145184e-01
-3.21910754e-02 6.07850254e-01 2.63198465e-01 4.02210712e-01
1.93710476e-01 1.15256667e+00 1.75748205e+00 -2.79022660e-02
-1.87788755e-01 -1.52631477e-01 -8.68652523e-01 5.88978827e-01
4.95265514e-01 -4.23089027e-01 -6.91442192e-01 -5.45221388e-01
-3.03939842e-02 2.27226436e-01 2.70856053e-01 -7.10080922e-01
1.37789890e-01 -2.85154104e-01 3.79477262e-01 -2.63041002e-03
6.30164981e-01 -8.51315320e-01 4.07740414e-01 3.50214779e-01
-9.18983370e-02 -9.01042074e-02 3.39984447e-02 5.13251960e-01
-2.69795775e-01 7.08796456e-02 1.07582211e+00 1.28153861e-01
-1.43298596e-01 7.93447196e-01 2.54154205e-01 -2.11572230e-01
1.22449636e+00 -1.43787220e-01 -7.97192231e-02 -6.11072600e-01
-5.57606459e-01 4.13853899e-02 9.16198432e-01 4.32502776e-01
6.84478819e-01 -1.71331489e+00 -6.79779470e-01 7.71165729e-01
-3.37116748e-01 3.64444293e-02 4.05063629e-01 8.49013329e-01
-4.49041545e-01 -8.60440657e-02 -3.65410298e-01 -2.77766824e-01
-1.21439099e+00 5.65228701e-01 2.77297407e-01 1.39490189e-02
-5.47921062e-01 9.26810801e-01 5.99553764e-01 -2.01565325e-01
-5.75145595e-02 1.82097659e-01 1.79896921e-01 -6.81393817e-02
6.25350714e-01 6.19612820e-02 -2.43187949e-01 -7.31167316e-01
-2.43120819e-01 8.67408097e-01 -1.58830717e-01 -3.30088109e-01
1.46095574e+00 -1.53193980e-01 -3.87315661e-01 -2.20773757e-01
1.17192471e+00 5.16019762e-01 -1.84483957e+00 -9.54108760e-02
-7.19810367e-01 -9.10092175e-01 1.47465289e-01 -5.57489097e-01
-1.73482871e+00 7.73583651e-01 4.75087464e-01 -2.24020660e-01
1.34231400e+00 -2.20027715e-01 1.10938191e+00 -3.79700691e-01
2.59513617e-01 -7.17335701e-01 3.08040857e-01 2.72507686e-02
1.26444185e+00 -8.21142673e-01 -1.78322449e-01 -8.15483391e-01
-5.18628895e-01 9.85300541e-01 5.56147218e-01 -1.25708301e-02
6.29634917e-01 2.07087994e-01 7.25490600e-02 -4.32930440e-02
-2.88919657e-01 1.63668722e-01 3.34291250e-01 6.32127643e-01
1.57311857e-01 -3.24153095e-01 -8.16725194e-02 2.17098266e-01
-5.06477579e-02 5.39297126e-02 1.12718366e-01 7.86243916e-01
6.53963909e-03 -1.52161825e+00 -2.75565863e-01 -2.43867338e-02
-2.31251270e-01 -8.81131664e-02 -3.88651103e-01 5.52981317e-01
3.26766580e-01 7.33543575e-01 -1.40536338e-01 -2.03573689e-01
1.16848335e-01 -1.37634963e-01 7.12268472e-01 -5.26552379e-01
-2.09471971e-01 1.55326128e-01 -3.53298128e-01 -6.63406909e-01
-4.54338551e-01 -4.21805918e-01 -1.06920421e+00 -5.48448324e-01
-3.81808668e-01 -5.64255789e-02 3.43269378e-01 6.95500255e-01
8.19745719e-01 2.63970613e-01 8.42394471e-01 -9.16738749e-01
-2.66949534e-01 -5.61340928e-01 -7.39293158e-01 4.85206783e-01
4.96130705e-01 -8.08210969e-01 -3.62641066e-01 1.81358680e-01] | [12.628185272216797, -0.20709143579006195] |
6117dffd-b90c-4777-a207-4f157676b740 | is-chatgpt-a-good-keyphrase-generator-a | 2303.13001 | null | https://arxiv.org/abs/2303.13001v1 | https://arxiv.org/pdf/2303.13001v1.pdf | Is ChatGPT A Good Keyphrase Generator? A Preliminary Study | The emergence of ChatGPT has recently garnered significant attention from the computational linguistics community. To demonstrate its capabilities as a keyphrase generator, we conduct a preliminary evaluation of ChatGPT for the keyphrase generation task. We evaluate its performance in various aspects, including keyphrase generation prompts, keyphrase generation diversity, multi-domain keyphrase generation, and long document understanding. Our evaluation is based on six benchmark datasets, and we adopt the prompt suggested by OpenAI while extending it to six candidate prompts. We find that ChatGPT performs exceptionally well on all six candidate prompts, with minor performance differences observed across the datasets. Based on our findings, we conclude that ChatGPT has great potential for keyphrase generation. Moreover, we discover that ChatGPT still faces challenges when it comes to generating absent keyphrases. Meanwhile, in the final section, we also present some limitations and future expansions of this report. | ['Liping Jing', 'Huafeng Liu', 'Yi Feng', 'Shilong Lu', 'Songfang Yao', 'Shuming Shi', 'Haiyun Jiang', 'Mingyang Song'] | 2023-03-23 | null | null | null | null | ['keyphrase-generation'] | ['natural-language-processing'] | [ 1.84505686e-01 2.10280702e-01 -1.68425515e-01 3.12309951e-01
-1.24589634e+00 -1.23502433e+00 1.28877485e+00 5.74513435e-01
-3.96873027e-01 9.63863134e-01 8.75130594e-01 -4.23168927e-01
-2.63123453e-01 -5.81117392e-01 -3.15195054e-01 -2.90541559e-01
7.90046677e-02 4.28257376e-01 2.69243062e-01 -5.39767444e-01
8.05549920e-01 4.45435196e-02 -1.32479918e+00 3.98664057e-01
1.08020782e+00 3.80546480e-01 1.29772782e-01 7.04920173e-01
-3.63226831e-01 8.48410666e-01 -1.06295109e+00 -6.12024844e-01
1.50089175e-01 -2.78188109e-01 -1.25534630e+00 -2.91492313e-01
5.48406661e-01 -3.55885148e-01 -3.71621609e-01 6.24529839e-01
4.89201009e-01 2.85180926e-01 7.25859225e-01 -1.39800274e+00
-6.86773241e-01 1.29289949e+00 -9.54871103e-02 3.97319168e-01
8.66329670e-01 3.23467851e-02 1.68276143e+00 -8.86682391e-01
9.05778468e-01 1.23580277e+00 3.75887483e-01 3.82546991e-01
-1.03031552e+00 -5.15441239e-01 8.67547616e-02 1.91980347e-01
-1.27485323e+00 -2.57524222e-01 5.75533748e-01 -2.40158767e-01
1.02138817e+00 5.09713411e-01 4.93454516e-01 1.24885690e+00
1.13245845e-01 1.16484320e+00 1.30939257e+00 -5.33517182e-01
-2.68952176e-03 -6.10741600e-02 1.16133146e-01 2.89591998e-01
8.54922161e-02 -2.68714935e-01 -7.75900960e-01 -6.71073079e-01
5.40939867e-01 -6.01290941e-01 -2.03541711e-01 5.76810777e-01
-1.63270915e+00 8.73741210e-01 -1.56438842e-01 4.16368753e-01
-3.32031876e-01 -1.76251829e-01 4.08463091e-01 4.45671231e-01
5.67138612e-01 1.11696088e+00 -4.43394542e-01 -7.04413831e-01
-8.21915567e-01 1.09772074e+00 1.37150276e+00 1.08531463e+00
4.18333888e-01 -3.00018758e-01 -7.50973880e-01 8.99983287e-01
-2.94647247e-01 4.15855229e-01 3.59648973e-01 -9.95730758e-01
7.84558892e-01 5.70801318e-01 2.28776246e-01 -1.16284692e+00
-2.14514658e-01 -4.24644351e-02 -4.60878193e-01 -8.19460154e-01
3.65904689e-01 -5.09507477e-01 -5.17630041e-01 1.39709771e+00
9.14850980e-02 -2.32869431e-01 1.32244900e-01 3.84338200e-01
1.20307517e+00 1.00058448e+00 1.93904117e-02 -3.09094906e-01
1.54256511e+00 -8.35649967e-01 -7.10681975e-01 -5.16994037e-02
6.75520420e-01 -1.16355097e+00 1.21018279e+00 2.86730289e-01
-8.70117664e-01 -7.17741475e-02 -4.77288127e-01 -1.66963652e-01
-3.19228500e-01 2.37687349e-01 7.34215140e-01 3.02140683e-01
-9.43854988e-01 2.93612540e-01 -3.66677374e-01 -6.10583723e-01
-2.04195142e-01 -4.08763766e-01 -6.52287602e-02 1.72942523e-02
-1.45667696e+00 7.02532291e-01 6.30113721e-01 -6.28256202e-01
-4.98066038e-01 -1.05734825e+00 -4.31698889e-01 -1.05689235e-01
8.31817329e-01 -5.20130634e-01 1.77764881e+00 4.42995280e-02
-1.37261796e+00 5.92931867e-01 -1.15991570e-01 -4.06543106e-01
3.18974972e-01 -2.69157708e-01 -3.56418014e-01 3.80259544e-01
5.40592134e-01 7.80357540e-01 5.52787840e-01 -7.61170506e-01
-9.67994392e-01 2.64956206e-01 3.45321745e-01 3.79206032e-01
-6.10708952e-01 5.65876782e-01 -2.02916428e-01 -1.16428781e+00
-1.58720851e-01 -1.10809803e+00 -1.63318470e-01 -7.15485096e-01
-6.96996212e-01 -8.86995554e-01 6.60458744e-01 -6.10543728e-01
1.74762535e+00 -1.70449471e+00 -1.23868547e-01 -2.74375137e-02
2.39343181e-01 -3.32861207e-02 -2.11259395e-01 1.37154281e+00
1.52569070e-01 4.89035875e-01 -2.74689663e-02 6.05710261e-02
1.89684600e-01 5.70086017e-02 -8.00493121e-01 -3.33065778e-01
8.99247825e-02 1.24725056e+00 -1.02437401e+00 -6.23383105e-01
-3.41431201e-01 -4.58917618e-02 -3.36588532e-01 1.31720871e-01
-5.62710881e-01 1.01227991e-01 -7.68144965e-01 6.05860591e-01
1.33154452e-01 -3.98688674e-01 1.30343869e-01 8.80540982e-02
-2.87296623e-01 8.46717834e-01 -9.85590518e-01 1.40135944e+00
-2.46566564e-01 5.71859598e-01 -3.61264259e-01 -3.02353173e-01
8.34587276e-01 5.05684495e-01 4.90039080e-01 -3.68166089e-01
-1.24351837e-01 3.13376695e-01 2.85989628e-03 -9.84180644e-02
1.44181430e+00 2.01361656e-01 -5.11570275e-01 1.31969202e+00
6.60491288e-02 -6.12456203e-01 8.05095732e-01 1.20301199e+00
1.20381880e+00 -3.80833507e-01 4.25398052e-01 -3.32819700e-01
1.40491977e-01 2.50413805e-01 -1.72428280e-01 1.11005771e+00
2.52865791e-01 4.39834565e-01 8.07803035e-01 -1.55114099e-01
-8.38084877e-01 -7.40061820e-01 1.79995298e-01 1.24571073e+00
-7.73394406e-02 -1.38826203e+00 -7.00832009e-01 -9.23979938e-01
-1.18777946e-01 9.68604147e-01 -4.35983539e-01 1.34688109e-01
-6.44063771e-01 -6.62033260e-01 8.80576611e-01 3.58818173e-01
2.83767879e-01 -1.32977414e+00 -4.78227168e-01 3.37533802e-01
-8.36520493e-01 -1.41755199e+00 -7.57156193e-01 -6.71310425e-02
-4.98925000e-01 -9.70780790e-01 -1.06001949e+00 -6.61400139e-01
3.91993284e-01 6.19368374e-01 1.35715854e+00 -1.68657247e-02
7.73379207e-02 8.35836589e-01 -1.02275848e+00 -5.44749975e-01
-7.97011852e-01 8.32177162e-01 -1.18988596e-01 -5.00022054e-01
1.67777628e-01 -5.08246005e-01 -3.27604860e-01 8.27799812e-02
-1.00674617e+00 1.74785763e-01 4.07638133e-01 5.26826084e-01
-8.11108667e-03 -4.31006914e-03 5.70504665e-01 -8.55736077e-01
1.61426783e+00 -5.71048796e-01 -2.13972688e-01 1.84782118e-01
-6.40230536e-01 4.29840460e-02 5.56163788e-01 -4.01121706e-01
-1.08367610e+00 -6.20526195e-01 4.77574095e-02 4.03582186e-01
4.25962470e-02 7.75649130e-01 3.77945185e-01 1.93043038e-01
8.87584984e-01 5.11027336e-01 -3.41597587e-01 -6.23828948e-01
7.76470840e-01 5.39677799e-01 3.90827924e-01 -1.13867188e+00
9.86455739e-01 -1.56541765e-02 -4.43446815e-01 -1.06248069e+00
-8.33240747e-01 -5.81799090e-01 -1.54109806e-01 -1.50835648e-01
3.19107473e-01 -8.79360437e-01 -5.00758469e-01 3.86030436e-01
-1.44548428e+00 -2.07447007e-01 -2.40688130e-01 1.24661110e-01
-2.86034882e-01 6.21692359e-01 -8.07417452e-01 -4.45061415e-01
-8.15983057e-01 -7.67726064e-01 1.02371955e+00 3.25169593e-01
-9.31733668e-01 -9.56389546e-01 1.66810393e-01 2.92245954e-01
2.90765196e-01 5.96434176e-02 9.94603992e-01 -1.07385719e+00
-5.49954593e-01 -1.94833741e-01 -1.85229644e-01 -9.86847058e-02
2.48427048e-01 1.54626921e-01 -4.79306996e-01 -1.39832869e-01
-5.42551637e-01 -5.97195208e-01 8.86952043e-01 6.05133548e-03
7.09770739e-01 -8.37039769e-01 -2.29898036e-01 -1.11248074e-02
6.52768731e-01 1.31211542e-02 2.91103631e-01 5.55952549e-01
5.14095366e-01 6.25784159e-01 8.09379101e-01 6.43664241e-01
1.00025451e+00 4.42534506e-01 -8.58359262e-02 3.73100609e-01
4.90857065e-02 -6.94766104e-01 3.34349990e-01 1.04207551e+00
-9.83829424e-02 -5.11897683e-01 -8.63288105e-01 7.28112102e-01
-1.64583993e+00 -9.45219576e-01 -9.63153094e-02 1.58229780e+00
1.23454916e+00 1.02453098e-01 3.90124142e-01 1.03112228e-01
3.49169970e-01 5.51269531e-01 1.34635389e-01 -4.62625623e-01
-3.00716221e-01 4.08750117e-01 1.46173507e-01 2.48857617e-01
-7.96697557e-01 1.46755922e+00 6.92851686e+00 9.34678674e-01
-8.32805097e-01 -2.03992844e-01 3.33216935e-01 3.58781219e-01
-6.81462109e-01 2.53185540e-01 -1.06484950e+00 2.80452192e-01
5.97495794e-01 -8.97407770e-01 2.94544667e-01 7.71618724e-01
-4.44475673e-02 -2.29594171e-01 -9.23005342e-01 7.50682533e-01
2.45469455e-02 -1.55513513e+00 3.94639432e-01 -8.73967931e-02
9.59372401e-01 -1.45266756e-01 -9.27043185e-02 4.72971559e-01
7.32248783e-01 -6.67815685e-01 5.73043168e-01 -8.78743902e-02
5.53630769e-01 -6.42819405e-01 2.22475618e-01 5.08295119e-01
-1.03723872e+00 1.19205914e-01 -1.17013372e-01 -1.98035777e-01
2.55176961e-01 4.64091212e-01 -1.46026921e+00 6.51951194e-01
3.53606164e-01 4.48233545e-01 -7.20512211e-01 8.76808763e-01
-6.06360316e-01 7.69402325e-01 -3.67442757e-01 -4.88030523e-01
4.41537946e-01 6.30992502e-02 7.75938153e-01 1.53002107e+00
4.31330293e-01 3.44673872e-01 6.17098510e-01 8.03512216e-01
-1.08321398e-01 3.73577356e-01 -4.77095932e-01 -6.93207383e-01
9.41006362e-01 1.48623478e+00 -8.81402194e-01 -6.32891953e-01
-1.25690728e-01 7.38870442e-01 -5.08816987e-02 4.06189203e-01
-4.32106614e-01 -6.48495018e-01 4.67094570e-01 -9.52883735e-02
-2.82279197e-02 -4.55694020e-01 -1.68264117e-02 -1.26686907e+00
1.06777884e-01 -1.36389041e+00 6.15718305e-01 -7.64584720e-01
-1.25820744e+00 6.79861307e-01 5.07547438e-01 -8.93857539e-01
-7.22118676e-01 -1.12600043e-01 -7.49659956e-01 7.26114333e-01
-1.11968482e+00 -1.15949595e+00 4.59555909e-02 3.35825711e-01
7.48151541e-01 -7.03155994e-02 9.30532873e-01 -1.57905161e-01
-1.19440988e-01 6.69925094e-01 -3.25224251e-01 2.41009295e-01
9.15486813e-01 -1.36223876e+00 1.19975626e+00 8.67496729e-01
5.39606690e-01 9.30054903e-01 9.20406163e-01 -9.22239184e-01
-1.30308473e+00 -7.70248115e-01 1.41258395e+00 -6.60767734e-01
1.10615385e+00 -3.07382762e-01 -7.32535422e-01 5.94100118e-01
5.16431987e-01 -5.59159219e-01 4.82493162e-01 2.12475732e-01
-4.12678212e-01 4.38880831e-01 -5.61169147e-01 1.13980877e+00
7.81135499e-01 -5.42078674e-01 -9.27381098e-01 5.07042646e-01
1.19067228e+00 -6.82658494e-01 -8.33168805e-01 1.29803672e-01
4.73002255e-01 -5.46761096e-01 7.24511147e-01 -5.27550399e-01
6.32321954e-01 4.30080146e-02 8.68404135e-02 -1.66245282e+00
-1.86016202e-01 -1.38748825e+00 -1.59824640e-01 1.50958848e+00
6.67334437e-01 -5.20635128e-01 6.55385196e-01 5.58350861e-01
-1.21164978e-01 -5.02375305e-01 -6.11912370e-01 -7.03709841e-01
1.55724660e-01 -5.55236816e-01 6.07070327e-01 1.05115736e+00
4.15878594e-01 8.24487448e-01 -4.57747519e-01 -3.57921600e-01
2.43409991e-01 2.32376695e-01 1.13672101e+00 -1.11067712e+00
-2.68231273e-01 -4.72636759e-01 3.86560827e-01 -1.36417699e+00
4.28188406e-02 -9.31252182e-01 -1.87812433e-01 -1.65592504e+00
1.88995644e-01 -2.17685178e-01 2.34417424e-01 6.04293346e-01
-5.85035801e-01 -6.79356903e-02 5.44941306e-01 1.96318537e-01
-6.41593397e-01 4.07666326e-01 1.54103625e+00 -5.01079895e-02
-5.42783976e-01 1.79064706e-01 -1.24200833e+00 4.05126005e-01
7.03751206e-01 -4.96979952e-01 -4.19092715e-01 -1.73122138e-01
3.86322618e-01 6.81698024e-02 1.47211581e-01 -5.90502381e-01
2.22625151e-01 -5.17662644e-01 -2.90053487e-01 -5.93127131e-01
4.90797088e-02 2.26033460e-02 -3.32666576e-01 -1.30829532e-02
-6.24136388e-01 4.25738543e-01 2.48524666e-01 2.61226803e-01
-1.25542447e-01 -8.33713114e-02 9.89603698e-02 -3.55360359e-01
-7.15001225e-01 1.93712473e-01 -8.70427728e-01 7.53604650e-01
4.65209067e-01 -9.60628688e-02 -5.14984667e-01 -7.54253745e-01
-1.80446282e-01 2.62185663e-01 1.43559411e-01 8.35283816e-01
5.42576194e-01 -1.17499149e+00 -1.05332124e+00 -2.60689110e-01
5.41644752e-01 -2.20685437e-01 -8.26436281e-02 5.30899704e-01
-2.17325330e-01 8.49834800e-01 2.13000059e-01 -2.03528926e-01
-1.21439707e+00 7.88010806e-02 -4.62679237e-01 -5.92099249e-01
-7.68841326e-01 7.51789749e-01 -5.72927594e-02 -4.21514899e-01
3.87737267e-02 -5.98919511e-01 -3.17187876e-01 3.82641912e-01
8.00944328e-01 3.77820283e-01 5.45971356e-02 -3.48857433e-01
-1.03200367e-02 3.77724580e-02 -5.27567625e-01 -5.51302493e-01
1.08935070e+00 -9.79032442e-02 -1.93251014e-01 2.91586488e-01
8.20439816e-01 3.42569619e-01 -5.19623518e-01 -5.06875694e-01
4.41327572e-01 -1.33361578e-01 -3.28682840e-01 -1.02925420e+00
-3.20007503e-01 2.29361370e-01 -4.26105887e-01 6.44404113e-01
8.12497318e-01 1.57626525e-01 1.20230949e+00 6.60280108e-01
4.88480657e-01 -1.04360485e+00 4.11800981e-01 1.20135307e+00
1.06905282e+00 -9.45843816e-01 1.19113959e-01 -2.27837473e-01
-9.30498779e-01 1.11684930e+00 5.51666737e-01 2.34629795e-01
3.38749677e-01 2.08296012e-02 8.63782167e-02 -2.46880800e-01
-1.19751525e+00 -3.12738381e-02 4.41367656e-01 4.01813060e-01
5.70536017e-01 4.87185381e-02 -7.37053812e-01 6.05270803e-01
-8.60036492e-01 -2.30960324e-01 9.16940808e-01 9.13349986e-01
-4.29074645e-01 -1.26388788e+00 -3.88584673e-01 4.21482086e-01
-6.18151784e-01 -6.02465868e-01 -9.50707972e-01 8.02549005e-01
-5.46410799e-01 1.03911304e+00 -2.89888561e-01 -4.09353137e-01
1.78721294e-01 1.05509073e-01 2.74469197e-01 -9.71412182e-01
-8.54446411e-01 -1.13640856e-02 5.30453324e-01 -1.58644140e-01
-6.75304607e-02 -5.23673058e-01 -8.33533168e-01 -5.76776445e-01
-2.25452572e-01 7.25710690e-01 3.12193304e-01 9.53992188e-01
4.15911883e-01 1.37254268e-01 5.30661225e-01 -4.46323842e-01
-7.13844597e-01 -1.31468260e+00 -3.21836650e-01 7.59496763e-02
-7.93558285e-02 -2.64299333e-01 -1.21047154e-01 7.98720717e-02] | [12.268296241760254, 8.898992538452148] |
a669e057-0608-43bf-8704-e4d431ec34d4 | the-exponentiated-gumbel-type-2-distribution | null | null | https://doi.org/10.1155/2016/5898356 | https://doi.org/10.1155/2016/5898356 | The Exponentiated Gumbel Type-2 Distribution: Properties and Application | We introduce a generalized version of the standard Gumble type-2 distribution. The new lifetime distribution is called the ExponentiatedGumbel (EG) type-2 distribution. The EG type-2 distribution has three nested submodels, namely, theGumbel type2 distribution, the Exponentiated Fr´echet (EF) distribution, and the Fr´echet distribution. Some statistical and reliability properties of the new distribution were given and the method of maximum likelihood estimates was proposed for estimating the model parameters. The usefulness and flexibility ofthe ExponentiatedGumbel (EG) type-2 distributionwere illustrated with a real lifetime data set. Results based on the log-likelihood and information statistics values showed that the EG type-2 distribution provides a better fit to the data than the other competing distributions. Also, the consistency of the parameters of the new distribution was demonstrated through a simulation study. The EG type-2 distribution is therefore recommended for effective modelling of lifetime data. | ['J.', 'A. C. Ohakwe', 'I. E. Akpanta', 'Okorie'] | 2016-07-10 | null | null | null | international-journal-of-mathematics-and | ['type'] | ['speech'] | [-7.00802207e-01 6.92892745e-02 -6.38717175e-01 -2.99801141e-01
-3.58797818e-01 -3.61455292e-01 5.35437226e-01 -2.82044802e-02
-3.30018550e-01 1.35800779e+00 -2.45307252e-01 -7.26358593e-01
-6.40125453e-01 -7.06455588e-01 -2.53686339e-01 -1.09167612e+00
-5.56557119e-01 5.93478143e-01 2.63475150e-01 2.20882297e-01
2.17391297e-01 6.34106219e-01 -1.42065680e+00 -5.18657267e-01
8.73797894e-01 1.13133752e+00 4.52145599e-02 3.56417388e-01
8.11309833e-03 1.49100214e-01 -7.61068404e-01 -4.27624315e-01
-1.35701299e-01 -1.11361206e-01 -3.54038738e-02 -1.34212017e-01
-6.32482767e-01 -6.05234623e-01 4.02282439e-02 6.22925997e-01
2.64779359e-01 7.77776912e-02 1.15951955e+00 -1.76193476e+00
-3.03153336e-01 4.27135527e-01 -7.79112339e-01 2.49976292e-01
9.95494053e-02 -4.73539025e-01 2.50737131e-01 -4.06651199e-01
2.44793892e-01 1.65064812e+00 7.28007138e-01 8.91829655e-02
-1.13603175e+00 -7.43467987e-01 -2.14685678e-01 4.52712737e-02
-1.65903342e+00 1.53117284e-01 1.39312550e-01 -2.43836969e-01
7.02724934e-01 1.09159693e-01 7.03020513e-01 9.95768011e-01
1.19396794e+00 4.85003591e-01 1.44097316e+00 -4.24558759e-01
5.21297574e-01 3.45170826e-01 4.12390202e-01 4.40975353e-02
1.01588929e+00 7.41402805e-01 2.43812986e-02 -7.30400741e-01
7.66394973e-01 -1.21191129e-01 4.11469400e-01 1.57054156e-01
-1.58893391e-01 9.16874349e-01 -4.56246704e-01 3.05248529e-01
-3.40133548e-01 3.93087566e-01 -9.03767813e-03 2.22642988e-01
7.05879271e-01 -3.93100679e-01 -3.94108474e-01 -2.50203878e-01
-7.96709895e-01 3.33268076e-01 7.55096495e-01 9.09527302e-01
5.05752444e-01 1.99050397e-01 -8.47520977e-02 6.04543746e-01
9.59669828e-01 1.00090706e+00 1.78728715e-01 -7.61863291e-01
-9.30208340e-02 1.73693717e-01 5.72165549e-01 -4.85950112e-01
-6.88047826e-01 -2.67027050e-01 -5.95449150e-01 1.07663259e-01
5.87733865e-01 -1.67108893e-01 -8.79071891e-01 1.57772636e+00
1.27016515e-01 -2.57000268e-01 7.86710754e-02 -6.22604564e-02
2.15893671e-01 8.87703121e-01 2.35108241e-01 -6.22342825e-01
9.91321325e-01 -1.48360252e-01 -1.06491125e+00 1.38242260e-01
2.63750404e-01 -4.76591378e-01 3.17546904e-01 5.38692713e-01
-1.06913769e+00 -2.04419911e-01 -9.84916329e-01 7.50435233e-01
-1.16950748e-02 -3.21869791e-01 5.07377267e-01 1.11451852e+00
-1.04523158e+00 5.00441670e-01 -9.05235410e-01 -4.05696213e-01
-4.06715363e-01 2.07416281e-01 -2.38305524e-01 1.13619193e-01
-1.17770290e+00 8.74024868e-01 5.32256603e-01 -2.06654295e-01
-8.06907952e-01 -2.13435441e-01 -5.36964118e-01 2.44559478e-02
-7.55827129e-02 -1.68110192e-01 1.10083389e+00 -1.52176738e-01
-1.18056786e+00 3.21567565e-01 5.15137203e-02 -2.28950545e-01
3.90390396e-01 1.27535209e-01 -7.46022642e-01 1.63436055e-01
-2.66100559e-02 -2.54671201e-02 5.07233143e-01 -1.44858336e+00
-3.94799918e-01 -2.61866301e-01 -4.50954258e-01 -4.01197076e-01
1.70616910e-01 1.97444037e-01 7.94676170e-02 -7.38283932e-01
-1.60860308e-02 -8.29929471e-01 1.53845310e-01 -1.63942814e-01
-2.32257664e-01 -6.07414842e-01 3.74402553e-01 -6.99165940e-01
1.83463478e+00 -2.24191117e+00 -6.79202378e-01 5.80390394e-01
-2.79109329e-01 -3.14808369e-01 1.13164127e-01 1.19164431e+00
1.07303401e-03 1.72167540e-01 -2.69016594e-01 -1.18417770e-01
1.12002835e-01 4.29693371e-01 2.24140167e-01 6.14609182e-01
-4.80689883e-01 2.69880980e-01 -8.97553921e-01 -4.64835167e-01
-1.03189625e-01 2.06787214e-01 1.22168317e-01 5.04279137e-01
2.29069948e-01 -2.03299411e-02 -2.54033178e-01 7.36645758e-01
1.02913320e+00 2.45776772e-02 7.89830685e-02 1.67198077e-01
-3.38802069e-01 -5.02985060e-01 -7.90119529e-01 3.32033038e-01
-5.57504408e-02 7.58513361e-02 6.48495778e-02 -5.54814637e-01
1.27774036e+00 4.89048183e-01 5.70896327e-01 -3.65958273e-01
4.07612741e-01 5.39761841e-01 1.29374012e-01 -3.11043710e-01
3.67518574e-01 -6.50604010e-01 -2.93423027e-01 8.02079737e-01
-7.89191201e-02 -1.57928374e-02 3.12297016e-01 4.31561284e-02
7.99907744e-01 2.80048302e-03 6.25716090e-01 -8.14703822e-01
-8.13834295e-02 -6.22099817e-01 5.57793856e-01 6.93567693e-01
-2.94008881e-01 8.07639509e-02 7.68309593e-01 1.07152946e-01
-7.36725926e-01 -1.55251884e+00 -5.52541554e-01 3.04795593e-01
8.88909921e-02 -2.88568169e-01 -3.21163654e-01 -3.58692199e-01
2.68126696e-01 1.28092694e+00 -5.55163264e-01 -3.40491593e-01
1.83729768e-01 -1.09401119e+00 5.19360065e-01 5.13723493e-01
4.67312545e-01 -4.80087072e-01 -4.21036005e-01 9.78038684e-02
1.84773486e-02 -5.86568296e-01 -2.35840261e-01 -6.22071326e-03
-1.00466430e+00 -1.04274964e+00 -6.79584384e-01 -3.74690704e-02
4.52221185e-01 -6.46183193e-02 7.26744175e-01 4.86701019e-02
2.79999822e-01 7.58636236e-01 -4.37453568e-01 -4.37772185e-01
-7.54233479e-01 -5.75814903e-01 3.67106736e-01 -2.32599899e-01
3.94102722e-01 -4.02959824e-01 -5.33517361e-01 6.22690201e-01
-9.14045930e-01 -7.08327830e-01 7.43492842e-01 6.56388342e-01
3.33571523e-01 5.90018094e-01 1.09234977e+00 -3.84927869e-01
8.48672748e-01 -9.72434640e-01 -5.62566340e-01 3.21193427e-01
-1.43084109e+00 1.85506549e-02 2.81799883e-02 -5.99853694e-01
-1.26349759e+00 -5.76718271e-01 -1.41559234e-02 -7.95921311e-02
2.30074063e-01 6.81327462e-01 -2.53167331e-01 3.61593306e-01
1.20118432e-01 -9.07250419e-02 3.69406819e-01 -5.49350441e-01
-5.70045374e-02 1.13018811e+00 3.84034067e-01 -8.37277353e-01
3.74120712e-01 2.28425592e-01 2.92695761e-01 -9.12379563e-01
-1.85838789e-01 -7.45254681e-02 -1.42973930e-01 -3.16245556e-01
5.18959761e-01 -5.37583947e-01 -5.57997108e-01 1.04270303e+00
-8.33280921e-01 -2.81208098e-01 -3.55465785e-02 8.44150901e-01
-7.58361042e-01 5.86780965e-01 -6.55690730e-01 -1.78984010e+00
-5.92052005e-02 -7.75822937e-01 5.68756819e-01 3.75021636e-01
1.04835525e-03 -1.44419527e+00 3.50441575e-01 -2.03941613e-01
2.35927761e-01 5.04864514e-01 1.28401351e+00 -5.45219123e-01
2.37280786e-01 -6.71728075e-01 -1.38654947e-01 2.43101522e-01
1.95860684e-01 3.35410565e-01 -3.38254929e-01 -7.03381956e-01
6.91851005e-02 1.84652716e-01 9.71936733e-02 6.77501023e-01
7.58614898e-01 -2.81862557e-01 -4.08367634e-01 1.31974801e-01
1.31847978e+00 8.30669463e-01 7.41933525e-01 3.81079793e-01
-4.67640400e-01 6.07274055e-01 8.98226082e-01 8.17531824e-01
5.21039307e-01 4.95479226e-01 3.71630520e-01 3.35095137e-01
3.34613830e-01 -3.00047666e-01 5.78078508e-01 4.50003535e-01
3.06369774e-02 -6.66755855e-01 -7.00414419e-01 4.82279390e-01
-1.72560108e+00 -8.70545626e-01 -5.17845809e-01 2.72536469e+00
6.90610468e-01 3.03380668e-01 6.13897264e-01 6.12080581e-02
9.42770481e-01 -3.60399991e-01 -4.69654024e-01 -5.92686653e-01
1.40951395e-01 -9.60463285e-03 6.05133295e-01 3.97553861e-01
-5.08155048e-01 1.37798786e-01 8.13579082e+00 1.14638710e+00
-3.21971625e-01 -4.29213531e-02 5.67484081e-01 3.60347956e-01
-4.59411085e-01 3.86505246e-01 -1.08084941e+00 1.10522842e+00
1.67979443e+00 -1.04533899e+00 -1.87186450e-01 2.54509836e-01
4.66993600e-01 -8.86149108e-01 -6.55312419e-01 7.43299425e-01
-2.53096342e-01 -2.19157353e-01 -3.16034615e-01 5.09049535e-01
3.41308475e-01 -6.65955782e-01 -9.62127466e-04 3.38846654e-01
4.73107725e-01 -6.54248297e-01 7.76594400e-01 8.23607564e-01
9.83089328e-01 -9.80765760e-01 1.19275093e+00 3.51450771e-01
-9.75179195e-01 -2.41161600e-01 -4.31851268e-01 1.50784180e-01
7.18112111e-01 8.04425657e-01 -4.37432200e-01 6.69391632e-01
8.31397712e-01 1.32998303e-01 -3.52037668e-01 1.46857858e+00
-2.49654800e-02 7.52806365e-01 -5.86244464e-01 1.54913634e-01
1.67356636e-02 -3.23372215e-01 4.60995495e-01 5.36131024e-01
1.06245160e+00 7.58421747e-03 -1.34742364e-01 5.42860746e-01
7.50958085e-01 -4.79497731e-01 -1.29661173e-01 -3.44668984e-01
1.21525645e+00 9.87879813e-01 -8.55457485e-01 -2.05809083e-02
-3.21916580e-01 2.16872469e-01 -4.08084095e-01 3.55402470e-01
-6.71505868e-01 -5.33220172e-01 2.00789161e-02 5.12405872e-01
-1.52552485e-01 -3.42329800e-01 5.06449789e-02 -4.71829742e-01
-6.13871038e-01 -1.93964839e-01 6.85714304e-01 -8.74851823e-01
-1.64446962e+00 5.05593777e-01 1.38049722e+00 -1.01547468e+00
-7.07721889e-01 -6.52231812e-01 -6.22228384e-01 6.56132519e-01
-9.20242965e-01 -8.78058970e-01 9.06283632e-02 3.16032320e-01
-1.13057725e-01 -8.32773820e-02 4.90061373e-01 5.76808527e-02
-4.67633009e-01 6.98924541e-01 8.30948949e-01 -8.31531167e-01
5.58294535e-01 -1.25655770e+00 -8.85827318e-02 2.95858383e-01
-1.17491889e+00 5.07749736e-01 1.13926041e+00 -1.19444990e+00
-7.75155485e-01 -5.22444725e-01 5.02323806e-01 -1.64636537e-01
8.58059287e-01 -1.06245307e-02 -6.13212943e-01 6.24653876e-01
1.30151331e-01 -8.02920818e-01 1.00347865e+00 -2.86201090e-01
1.06000967e-01 -1.04899697e-01 -1.65655434e+00 6.55499026e-02
4.05687839e-01 1.29777297e-01 -4.75951105e-01 -1.80705898e-02
2.44360238e-01 4.14275229e-01 -1.29597783e+00 3.05137426e-01
1.07767761e+00 -1.03973150e+00 5.61703742e-01 -1.50666675e-02
-3.77480268e-01 2.19500855e-01 -3.02588016e-01 -1.24944282e+00
-2.67685771e-01 -6.16605401e-01 -4.72251028e-01 1.17619801e+00
2.40866229e-01 -1.24746597e+00 2.53153235e-01 7.47920752e-01
8.20192322e-02 -7.85061419e-01 -1.41375577e+00 -1.68147004e+00
5.34639657e-01 -6.30350038e-02 7.96684384e-01 4.45349328e-02
2.27775052e-02 -2.04190671e-01 -5.17130077e-01 -1.76495418e-01
1.04564369e+00 -4.89800543e-01 4.51761186e-01 -1.40211093e+00
-3.77990812e-01 -3.31848189e-02 -3.11257541e-01 -7.94939697e-01
-2.40959283e-02 -2.37298504e-01 -6.65921392e-03 -1.48816204e+00
4.92711544e-01 -5.53484142e-01 -3.67861539e-01 2.67371863e-01
-8.57467949e-02 -2.15828836e-01 -1.03381045e-01 2.95170069e-01
1.51680604e-01 7.54501820e-01 6.99155033e-01 4.10028517e-01
-3.21118012e-02 6.20859146e-01 -4.45913762e-01 3.55843693e-01
8.02943170e-01 -6.14544988e-01 -8.14839542e-01 3.36742580e-01
4.17279303e-02 5.46279073e-01 5.72565421e-02 -5.06056130e-01
-3.36060941e-01 -4.05963540e-01 1.20518290e-01 -1.20684791e+00
2.88292468e-01 -7.73167431e-01 9.09374058e-01 5.63570142e-01
4.22985435e-01 3.69257182e-01 2.67672062e-01 7.30515242e-01
2.67929494e-01 -6.33070707e-01 7.34589338e-01 2.56618649e-01
-7.63494223e-02 4.86788638e-02 -9.37085748e-01 -4.80816841e-01
1.39766920e+00 -4.26192194e-01 -6.68967307e-01 -8.94850850e-01
-9.22956884e-01 1.89827949e-01 5.00991821e-01 1.16342925e-01
3.83839846e-01 -1.46868360e+00 -4.32032019e-01 1.68345943e-02
-2.40856677e-01 -7.85579264e-01 2.33216912e-01 1.08101499e+00
-3.28376740e-01 3.81644815e-01 -1.72037140e-01 -2.90798455e-01
-8.68054509e-01 4.09713119e-01 2.91030049e-01 -3.94162118e-01
-1.33092597e-01 1.53669626e-01 2.34147221e-01 -1.13247007e-01
-1.46920100e-01 1.17452569e-01 -1.86786488e-01 3.38828042e-02
5.92505455e-01 1.18307602e+00 -5.31493962e-01 -7.15867877e-01
-4.79573399e-01 3.24726522e-01 2.52796756e-03 -4.11356360e-01
1.10243213e+00 -6.67641282e-01 -5.09746194e-01 8.09432805e-01
8.02292109e-01 -2.00189099e-01 -1.24837542e+00 5.35808742e-01
3.08912098e-01 -4.64637786e-01 -7.11833537e-02 -1.00709748e+00
-5.52971065e-01 4.02071357e-01 7.44854569e-01 7.12515116e-01
1.03376067e+00 -1.18400063e-02 4.25175816e-01 -4.06317532e-01
8.91565084e-01 -8.01057100e-01 -1.65645838e-01 6.20393790e-02
8.01038086e-01 -5.34913301e-01 2.98346519e-01 -2.14536458e-01
-2.50589430e-01 1.11963654e+00 5.07264495e-01 1.29777387e-01
1.50901186e+00 3.06186199e-01 -2.68222123e-01 1.12788804e-01
-7.99155235e-01 6.38633370e-02 -6.55980706e-02 8.83980334e-01
2.80898571e-01 1.82201594e-01 -1.34101427e+00 1.29622746e+00
-6.22919686e-02 -5.25673367e-02 8.28696311e-01 1.13826656e+00
-5.02446651e-01 -1.44795907e+00 -5.59143126e-01 5.32399476e-01
-4.50756967e-01 4.89188880e-01 9.18077305e-03 1.15883470e+00
-2.58811980e-01 1.62158418e+00 1.57130182e-01 -4.21233475e-01
1.05203815e-01 -1.67567477e-01 4.97664005e-01 -1.15086600e-01
5.49242087e-03 3.70664388e-01 1.49934947e-01 -2.11127177e-01
-2.34522730e-01 -6.10791802e-01 -1.12062538e+00 -7.33581066e-01
-6.33603573e-01 7.94319034e-01 7.48062372e-01 9.04580832e-01
8.54484141e-02 -8.11698381e-03 8.46908271e-01 -5.16795993e-01
-1.02434814e+00 -1.19519937e+00 -1.44607139e+00 -4.79248434e-01
-4.61375006e-02 -1.24885929e+00 -8.80931199e-01 -4.49593991e-01] | [6.980194568634033, 4.287303447723389] |
7ddf1810-8850-4ba8-bc38-e16019c1154f | picture-that-sketch-photorealistic-image | 2303.11162 | null | https://arxiv.org/abs/2303.11162v2 | https://arxiv.org/pdf/2303.11162v2.pdf | Picture that Sketch: Photorealistic Image Generation from Abstract Sketches | Given an abstract, deformed, ordinary sketch from untrained amateurs like you and me, this paper turns it into a photorealistic image - just like those shown in Fig. 1(a), all non-cherry-picked. We differ significantly from prior art in that we do not dictate an edgemap-like sketch to start with, but aim to work with abstract free-hand human sketches. In doing so, we essentially democratise the sketch-to-photo pipeline, "picturing" a sketch regardless of how good you sketch. Our contribution at the outset is a decoupled encoder-decoder training paradigm, where the decoder is a StyleGAN trained on photos only. This importantly ensures that generated results are always photorealistic. The rest is then all centred around how best to deal with the abstraction gap between sketch and photo. For that, we propose an autoregressive sketch mapper trained on sketch-photo pairs that maps a sketch to the StyleGAN latent space. We further introduce specific designs to tackle the abstract nature of human sketches, including a fine-grained discriminative loss on the back of a trained sketch-photo retrieval model, and a partial-aware sketch augmentation strategy. Finally, we showcase a few downstream tasks our generation model enables, amongst them is showing how fine-grained sketch-based image retrieval, a well-studied problem in the sketch community, can be reduced to an image (generated) to image retrieval task, surpassing state-of-the-arts. We put forward generated results in the supplementary for everyone to scrutinise. | ['Yi-Zhe Song', 'Tao Xiang', 'Pinaki Nath Chowdhury', 'Aneeshan Sain', 'Ayan Kumar Bhunia', 'Subhadeep Koley'] | 2023-03-20 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Koley_Picture_That_Sketch_Photorealistic_Image_Generation_From_Abstract_Sketches_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Koley_Picture_That_Sketch_Photorealistic_Image_Generation_From_Abstract_Sketches_CVPR_2023_paper.pdf | cvpr-2023-1 | ['sketch-based-image-retrieval'] | ['computer-vision'] | [ 5.29757857e-01 3.34950179e-01 1.03035577e-01 -2.25487545e-01
-8.13881278e-01 -9.04209316e-01 1.15568256e+00 -6.66698217e-01
-5.08045740e-02 4.71554101e-01 3.68858576e-01 -4.56534699e-02
1.77615881e-01 -7.73602366e-01 -9.04418766e-01 -6.17274046e-01
5.22662938e-01 4.15098071e-01 -4.15942818e-01 -2.52479047e-01
3.39616686e-01 6.77587986e-01 -1.34737408e+00 4.79878485e-01
1.05599798e-01 8.96087170e-01 1.12185448e-01 8.71232331e-01
-6.15218505e-02 5.09705245e-01 -4.66997147e-01 -1.08655703e+00
5.06843805e-01 -6.17190540e-01 -6.20447040e-01 3.31969649e-01
1.05103767e+00 -6.99133754e-01 -5.56422651e-01 7.06803262e-01
3.71782988e-01 -2.50750542e-01 9.07535553e-01 -1.23901320e+00
-1.27796078e+00 2.87102193e-01 -2.54813969e-01 -5.84097087e-01
3.54463011e-01 4.17869300e-01 1.18919504e+00 -1.20794392e+00
1.05180478e+00 1.35305572e+00 5.03423750e-01 8.71523976e-01
-1.51171947e+00 -6.06400311e-01 6.24998007e-03 -4.35621828e-01
-1.46837699e+00 -6.47199750e-01 8.89789402e-01 -4.28541929e-01
6.44191980e-01 4.23103601e-01 8.04018795e-01 1.43709934e+00
-1.24965750e-01 9.34062302e-01 8.63097191e-01 -4.24724877e-01
4.85643074e-02 7.14579672e-02 -6.85262918e-01 7.25360990e-01
-1.97809935e-01 1.24957949e-01 -5.38063407e-01 3.35194990e-02
1.26861346e+00 2.67970353e-01 -2.10565090e-01 -6.54200852e-01
-1.21698308e+00 6.47112131e-01 3.79703403e-01 2.32918799e-01
-3.95052195e-01 5.05107224e-01 1.75856680e-01 5.49926937e-01
3.26729715e-01 4.26084578e-01 8.78520384e-02 1.28244599e-02
-1.28907657e+00 4.02775913e-01 7.42806971e-01 1.35634756e+00
6.94146693e-01 1.40716895e-01 -2.23291114e-01 7.48808503e-01
1.24274567e-01 5.62694788e-01 -1.31453589e-01 -1.06497788e+00
1.88493505e-01 3.74601662e-01 8.41389224e-02 -8.16077530e-01
5.33931494e-01 -9.96889398e-02 -9.33098793e-01 4.93895292e-01
1.92724675e-01 2.92361349e-01 -9.45402205e-01 1.68905199e+00
-2.08339244e-01 -2.06376612e-01 -1.27980351e-01 1.07569194e+00
7.20796227e-01 6.50260210e-01 1.62402675e-01 4.14074779e-01
1.50769436e+00 -9.18127060e-01 -4.11597729e-01 -1.37029484e-01
-2.88258935e-03 -8.03459287e-01 1.38450718e+00 3.58357996e-01
-1.58281398e+00 -5.54850996e-01 -1.08046794e+00 -7.38054216e-01
-4.58081007e-01 3.36289525e-01 3.90228093e-01 4.28197622e-01
-1.29695964e+00 7.52706349e-01 -2.53557771e-01 -6.18293047e-01
6.39131308e-01 6.40900433e-02 -7.85620689e-01 -1.29064783e-01
-8.60328078e-01 8.69595706e-01 -1.22914419e-01 6.74331635e-02
-7.89049447e-01 -8.19630563e-01 -7.63843536e-01 2.88519919e-01
1.36752531e-01 -1.27173460e+00 1.06649268e+00 -1.29656339e+00
-1.74500477e+00 1.42498875e+00 -1.02835698e-02 -1.75407469e-01
8.48265469e-01 -2.56955288e-02 6.05587102e-03 2.32980520e-01
-9.45895687e-02 1.22908866e+00 1.29324150e+00 -1.54751301e+00
-1.85364768e-01 -8.66151899e-02 1.44441068e-01 1.16482489e-01
-1.23145893e-01 -1.00515954e-01 -7.40018904e-01 -9.57901537e-01
-2.77931511e-01 -9.71629620e-01 1.57763883e-01 7.66166508e-01
-4.01231349e-01 6.50039641e-03 6.54698908e-01 -5.29009640e-01
9.15878356e-01 -2.21701312e+00 4.83144432e-01 8.08039233e-02
1.33167490e-01 2.59168506e-01 -5.45490086e-01 1.13006794e+00
-1.73655063e-01 1.87977701e-01 -4.70640689e-01 -8.57305348e-01
5.31215966e-01 2.18450293e-01 -9.36852753e-01 3.01366031e-01
6.40046179e-01 1.43788099e+00 -8.74056101e-01 -3.29577267e-01
2.41456211e-01 7.30909705e-01 -4.78867561e-01 4.35108602e-01
-2.31952563e-01 3.69643986e-01 -2.67017692e-01 5.37308693e-01
7.46487141e-01 -1.34980991e-01 1.03107832e-01 -3.02669644e-01
-1.05619527e-01 -7.10560381e-02 -8.01777124e-01 2.24853063e+00
-6.67342603e-01 7.03189492e-01 2.93045521e-01 -6.12351775e-01
9.03014004e-01 2.07978502e-01 1.07577015e-02 -6.63529754e-01
-2.99824953e-01 2.24639088e-01 -6.01351857e-01 -1.92359805e-01
5.84651291e-01 -5.60805023e-01 -2.37837076e-01 5.76277077e-01
8.85630548e-02 -7.25412965e-01 -1.68995187e-01 4.19592500e-01
9.22786474e-01 5.68836689e-01 5.74745461e-02 -1.48630574e-01
4.07921970e-01 -3.35143864e-01 -1.89138725e-01 7.56743193e-01
3.90560538e-01 1.29576409e+00 5.59442699e-01 -5.28305113e-01
-1.59234929e+00 -1.34392285e+00 1.93090498e-01 9.28816140e-01
-1.27665147e-01 -4.37107116e-01 -7.00338721e-01 -5.87127507e-01
1.47404328e-01 5.62430918e-01 -7.26583600e-01 1.87875517e-02
-6.51726782e-01 4.34592932e-01 6.97312474e-01 2.76932031e-01
3.13572705e-01 -1.26531029e+00 -4.66301262e-01 -1.89327762e-01
2.18726411e-01 -8.59987557e-01 -9.03766453e-01 -3.64482671e-01
-4.54708844e-01 -6.12579048e-01 -1.49185133e+00 -7.62826681e-01
7.86656499e-01 1.81018785e-01 1.44435966e+00 3.86975080e-01
-3.56642663e-01 8.96563053e-01 -7.38959201e-03 -2.42699891e-01
-5.13266265e-01 -3.39214839e-02 -5.15103519e-01 3.24936956e-01
-1.18915707e-01 -8.92960489e-01 -9.36857939e-01 -1.93799958e-02
-1.35930204e+00 3.88532162e-01 9.26195145e-01 9.73870099e-01
4.03473258e-01 -7.12952852e-01 2.45710164e-01 -8.46032679e-01
5.58384418e-01 -1.53723851e-01 -3.36515456e-01 5.33816695e-01
-3.79382044e-01 2.55110145e-01 7.34416008e-01 -3.63438427e-01
-8.42280686e-01 2.06431106e-01 -1.99472606e-01 -7.74177432e-01
-1.44349352e-01 -2.38652423e-01 -2.57895619e-01 -2.63216775e-02
3.82547617e-01 4.40038145e-01 9.50978547e-02 -4.85241503e-01
1.03148389e+00 6.30654275e-01 6.81803048e-01 -6.53612673e-01
1.09583139e+00 6.73505604e-01 2.31087446e-01 -8.12942624e-01
-2.93225378e-01 -2.94152722e-02 -6.29506886e-01 -5.54925762e-02
7.75116444e-01 -8.66548181e-01 -8.94488573e-01 2.13526025e-01
-1.44092548e+00 -5.07272542e-01 -7.10356116e-01 -2.75324941e-01
-9.46501672e-01 4.13279295e-01 -5.93621254e-01 -8.50889564e-01
-5.72065532e-01 -9.07691121e-01 1.74745619e+00 -1.00065470e-01
-3.10696602e-01 -8.12402248e-01 -7.07092090e-03 -2.16052718e-02
6.16198957e-01 1.58236101e-01 9.62925136e-01 5.38108032e-03
-1.04245114e+00 -1.43414304e-01 -6.14310563e-01 3.70904535e-01
-2.73132682e-01 -6.05954137e-03 -1.13104308e+00 -1.81917131e-01
-3.35269511e-01 -4.53073472e-01 9.93536770e-01 -1.87772170e-01
1.02665210e+00 -5.71643174e-01 -8.65694657e-02 7.24728823e-01
1.52288675e+00 -3.58472794e-01 9.36985731e-01 -1.71444982e-01
8.18210006e-01 6.11750662e-01 3.50087285e-02 1.10135876e-01
3.64788115e-01 7.88747013e-01 1.92141920e-01 -3.44129145e-01
-6.43705070e-01 -1.00258291e+00 2.47578233e-01 4.81629759e-01
-1.58021841e-02 -3.44755948e-01 -4.05490965e-01 5.86626709e-01
-1.64946485e+00 -1.14403439e+00 2.60681421e-01 2.14380646e+00
7.38113523e-01 -3.55615854e-01 3.65961753e-02 -1.73516706e-01
4.23791856e-01 4.10814404e-01 -2.68611550e-01 -6.67928219e-01
-1.55461326e-01 6.03338420e-01 2.24405289e-01 6.32772207e-01
-7.81970203e-01 1.19362354e+00 5.83341217e+00 8.66937995e-01
-1.21925807e+00 -1.56865150e-01 4.61051404e-01 -8.98614526e-02
-8.18838000e-01 3.28514636e-01 -1.54817209e-01 3.60418618e-01
4.42253172e-01 1.92115635e-01 8.13373506e-01 6.66392088e-01
-2.22723827e-01 1.59439504e-01 -1.56478465e+00 1.25295043e+00
2.97027677e-01 -1.56099641e+00 6.23154402e-01 6.66895211e-02
5.19451916e-01 -4.26535368e-01 3.55945885e-01 2.62085885e-01
1.25728399e-01 -1.26203120e+00 1.05223691e+00 9.59650218e-01
1.42216313e+00 -3.01802099e-01 -1.64425328e-01 2.90510189e-02
-9.68425453e-01 1.43709615e-01 -2.66581148e-01 4.33046035e-02
3.28627676e-01 3.25606942e-01 -4.30751532e-01 6.31846309e-01
1.23281702e-01 6.26544118e-01 -2.50757873e-01 5.57135403e-01
-4.00695950e-01 -8.40246603e-02 -5.59195876e-02 9.34166238e-02
2.79848963e-01 -3.41875851e-01 4.25077856e-01 1.50554609e+00
4.77667898e-01 2.76272029e-01 -3.09902281e-01 1.50782931e+00
-3.59122425e-01 -1.96809992e-01 -1.07234371e+00 -2.86279261e-01
2.70820886e-01 1.35460281e+00 -3.99908036e-01 -4.43001986e-01
-2.71580189e-01 1.69087756e+00 2.43302479e-01 5.72396517e-01
-4.57005292e-01 -4.43760842e-01 5.32415152e-01 2.05365926e-01
5.06217539e-01 -2.20895976e-01 -2.95702070e-01 -1.28170907e+00
1.86789632e-01 -7.66290426e-01 -4.71107475e-02 -1.32487869e+00
-1.49747515e+00 5.33151031e-01 -2.37550825e-01 -1.15917003e+00
-3.02687854e-01 -5.11319757e-01 -7.82235086e-01 1.16734779e+00
-1.53555298e+00 -1.88649094e+00 -1.29602864e-01 3.52895677e-01
6.12629354e-01 8.16401914e-02 9.42229390e-01 2.48271137e-01
-4.81308922e-02 7.59125471e-01 -2.61339813e-01 2.42971212e-01
8.92584801e-01 -1.20176911e+00 9.71286476e-01 7.06504405e-01
4.73428547e-01 8.06791663e-01 5.47108591e-01 -4.67605650e-01
-1.92904937e+00 -8.06724310e-01 1.04032028e+00 -7.36852586e-01
5.10774434e-01 -9.18153048e-01 -5.88865221e-01 6.16459966e-01
4.71340477e-01 6.26215637e-02 1.16258733e-01 -3.50069165e-01
-6.57786846e-01 7.65340179e-02 -9.71218109e-01 9.63489294e-01
1.40779662e+00 -1.15194190e+00 -4.26988214e-01 3.92068028e-02
3.67610842e-01 -1.84722424e-01 -8.02812755e-01 -1.54523641e-01
1.23865926e+00 -9.87063766e-01 1.27723396e+00 -6.09810829e-01
9.39436972e-01 -1.59994856e-01 -1.36821404e-01 -1.11739206e+00
-3.15457195e-01 -1.01495886e+00 1.68881431e-01 1.26132381e+00
1.25564367e-01 -2.49422863e-01 8.85097861e-01 7.72940338e-01
-6.18773997e-02 -7.12720156e-01 -6.35075033e-01 -6.20283544e-01
2.36133888e-01 -4.38404754e-02 6.72665775e-01 7.44000077e-01
-1.92401528e-01 5.20205855e-01 -6.56415641e-01 -3.13956022e-01
5.20076513e-01 3.59793246e-01 1.25075448e+00 -9.68315959e-01
-4.11904931e-01 -7.83775806e-01 -2.37300023e-01 -1.39591014e+00
1.91744059e-01 -9.54051197e-01 -1.32842436e-01 -1.43815601e+00
1.60175353e-01 -3.43988895e-01 2.02871844e-01 3.56471330e-01
6.71294481e-02 6.94537759e-01 7.05005050e-01 3.26923460e-01
-4.00868326e-01 6.05419874e-01 1.54343629e+00 -2.43021488e-01
2.17960760e-01 -3.45056295e-01 -7.55712748e-01 1.82334274e-01
1.80994973e-01 -2.92473793e-01 -4.67184991e-01 -6.00909829e-01
3.92175496e-01 2.62131959e-01 9.20588136e-01 -5.98236620e-01
2.49860883e-01 5.02251387e-02 4.28030103e-01 -1.77542150e-01
7.76707530e-01 -8.12478125e-01 5.11354327e-01 1.18695490e-01
-6.10817850e-01 1.67821534e-03 -4.44026701e-02 4.32051718e-01
-1.90221295e-02 -7.94161335e-02 5.81908882e-01 -4.41775441e-01
-3.86158109e-01 4.80969638e-01 1.27833933e-01 -1.60702616e-01
5.96781611e-01 -2.90088028e-01 -1.38555974e-01 -7.34252214e-01
-5.43352067e-01 -2.64337987e-01 1.01618981e+00 3.12849790e-01
6.65782094e-01 -1.57406664e+00 -7.91046262e-01 4.51923490e-01
2.84241974e-01 -1.32203534e-01 4.41209286e-01 4.64587629e-01
-4.02282357e-01 4.86609221e-01 -3.82741541e-01 -2.87516207e-01
-9.55881238e-01 7.52608657e-01 6.54144362e-02 -2.98004085e-03
-8.30384612e-01 8.02268147e-01 6.23745918e-01 -3.37653011e-01
2.28933945e-01 -1.74341306e-01 5.81294060e-01 -1.44722328e-01
3.78715158e-01 -1.34024188e-01 -1.57667637e-01 -4.43651021e-01
-3.10927164e-02 8.12680125e-01 1.33838594e-01 -5.16484320e-01
1.26862109e+00 -2.13282425e-02 -4.77862433e-02 3.16667706e-01
1.38643670e+00 8.97559449e-02 -1.56782842e+00 -2.18844023e-02
-3.95424426e-01 -5.92900217e-01 -2.81334043e-01 -9.17586982e-01
-8.45038056e-01 1.31474006e+00 2.10504293e-01 1.35379136e-01
9.92567241e-01 1.32985422e-02 9.55540776e-01 2.26598978e-01
2.39767522e-01 -6.73267007e-01 1.95812747e-01 1.57020062e-01
1.49427199e+00 -1.03740966e+00 4.66975048e-02 6.78753778e-02
-7.18865812e-01 1.23462713e+00 -3.35874110e-02 -5.98650634e-01
2.57641852e-01 5.80797382e-02 -1.64131552e-01 -3.25083107e-01
-6.97505474e-01 -1.08506270e-01 5.19315302e-01 5.13694465e-01
2.65029907e-01 -4.59146053e-02 5.77747859e-02 2.79663175e-01
-1.91836834e-01 2.37230152e-01 2.86435694e-01 6.55112803e-01
4.00469564e-02 -1.50943398e+00 -1.80736363e-01 1.00399293e-02
-5.54697514e-02 -3.03835839e-01 -8.39655161e-01 8.24963808e-01
-6.42905682e-02 4.16889220e-01 1.40746415e-01 -1.98348835e-01
4.70096439e-01 2.56466120e-01 8.46636474e-01 -3.97956073e-01
-5.96625805e-01 -2.81779945e-01 -8.63712355e-02 -5.57281852e-01
-9.87479389e-02 -2.42537484e-01 -5.18088937e-01 -6.13828421e-01
2.80719966e-01 -1.88881487e-01 8.19190323e-01 5.65253794e-01
6.65880263e-01 7.31323473e-03 4.63909119e-01 -1.32240522e+00
-4.39998448e-01 -5.57181418e-01 -4.68159139e-01 7.11822152e-01
5.13679028e-01 -2.62060940e-01 -2.22625270e-01 2.11737409e-01] | [11.80747127532959, 0.20058897137641907] |
7d020165-9236-4f17-b032-bf28dd7f6b50 | grig-few-shot-generative-residual-image | 2304.12035 | null | https://arxiv.org/abs/2304.12035v1 | https://arxiv.org/pdf/2304.12035v1.pdf | GRIG: Few-Shot Generative Residual Image Inpainting | Image inpainting is the task of filling in missing or masked region of an image with semantically meaningful contents. Recent methods have shown significant improvement in dealing with large-scale missing regions. However, these methods usually require large training datasets to achieve satisfactory results and there has been limited research into training these models on a small number of samples. To address this, we present a novel few-shot generative residual image inpainting method that produces high-quality inpainting results. The core idea is to propose an iterative residual reasoning method that incorporates Convolutional Neural Networks (CNNs) for feature extraction and Transformers for global reasoning within generative adversarial networks, along with image-level and patch-level discriminators. We also propose a novel forgery-patch adversarial training strategy to create faithful textures and detailed appearances. Extensive evaluations show that our method outperforms previous methods on the few-shot image inpainting task, both quantitatively and qualitatively. | ['Hanli Zhao', 'Kaijie Shi', 'Tao Wang', 'Minglun Gong', 'Yong-Liang Yang', 'Xiaogang Jin', 'Xianta Jiang', 'Wanglong Lu'] | 2023-04-24 | null | null | null | null | ['image-inpainting'] | ['computer-vision'] | [ 5.71816146e-01 6.75275549e-02 -6.12109005e-02 -2.40094766e-01
-1.15038490e+00 -2.09353462e-01 2.71469355e-01 -6.38836622e-01
2.61100739e-01 9.11091030e-01 2.08553135e-01 1.84211686e-01
3.53412092e-01 -9.86812472e-01 -1.10492146e+00 -6.45965695e-01
4.13522929e-01 -4.57213074e-02 1.24816947e-01 -3.08477461e-01
6.14946224e-02 5.96657634e-01 -1.34952724e+00 6.42455876e-01
9.78132784e-01 1.04309750e+00 1.21624574e-01 7.10210681e-01
4.25684974e-02 1.35392630e+00 -7.50745356e-01 -4.98575658e-01
6.28971457e-01 -9.26094234e-01 -6.08069003e-01 6.06486559e-01
7.05622673e-01 -7.39987910e-01 -5.43739736e-01 9.25601423e-01
5.16227007e-01 1.15127601e-01 4.98951167e-01 -1.15171993e+00
-1.10184574e+00 2.28837773e-01 -7.07043469e-01 -1.29792869e-01
2.73049444e-01 2.72076339e-01 4.60842520e-01 -1.00634730e+00
8.03121328e-01 1.32431436e+00 8.78567338e-01 6.91787302e-01
-1.44408190e+00 -5.69743037e-01 -3.19149494e-01 -1.28748447e-01
-1.19258487e+00 -4.85703945e-01 1.30014801e+00 -2.94933226e-02
6.58590972e-01 3.15797091e-01 3.84069026e-01 1.15141082e+00
4.40770328e-01 6.81828678e-01 1.28953004e+00 -5.47095597e-01
2.23951846e-01 -7.95715228e-02 -7.95692027e-01 6.57686889e-01
-3.89123522e-02 3.28250021e-01 -4.65989441e-01 -1.64961051e-02
1.44119120e+00 4.00663197e-01 -1.25682876e-01 -2.89412141e-01
-8.08610439e-01 9.21747565e-01 6.33258283e-01 8.90349522e-02
-5.20951807e-01 4.47636217e-01 2.08713934e-01 6.79910600e-01
9.22685206e-01 4.87032741e-01 -6.25351593e-02 3.14759493e-01
-1.38020527e+00 4.63000327e-01 4.94248837e-01 8.71317267e-01
8.62050295e-01 4.80896860e-01 -4.33767080e-01 1.19452024e+00
-2.41647795e-01 3.56833845e-01 1.90703601e-01 -1.47094762e+00
2.00863540e-01 4.09630269e-01 3.04611206e-01 -1.21697676e+00
3.47434372e-01 -2.35865772e-01 -1.01459455e+00 6.95465624e-01
-1.27821565e-02 -6.18936270e-02 -1.14327002e+00 1.60313630e+00
2.42948160e-01 1.75591081e-01 -1.69604532e-02 8.41724098e-01
7.34894753e-01 7.23283291e-01 -6.24534860e-02 -7.58524090e-02
1.11818624e+00 -1.47889304e+00 -9.21224535e-01 -4.15713131e-01
-7.36363232e-02 -1.06967664e+00 1.03669739e+00 1.80179492e-01
-1.55100179e+00 -8.33804011e-01 -9.71841633e-01 -4.01339471e-01
-1.66995347e-01 -8.56902078e-03 6.88467383e-01 5.44417918e-01
-1.12177181e+00 7.79088855e-01 -4.84059870e-01 -6.32046983e-02
7.73798466e-01 -6.34701923e-02 -4.12642390e-01 -5.02467394e-01
-1.08648789e+00 7.46342301e-01 1.66242436e-01 -1.75324485e-01
-1.16696274e+00 -8.02138150e-01 -9.93407071e-01 -7.84931891e-03
2.47462004e-01 -6.87934935e-01 1.09126091e+00 -1.44858015e+00
-1.49714029e+00 7.80256569e-01 4.07543704e-02 -5.13148069e-01
6.42837465e-01 -1.64610520e-01 -3.73299688e-01 3.60872090e-01
3.06547701e-01 1.10812712e+00 1.47537279e+00 -1.60627389e+00
-2.16907278e-01 7.08647743e-02 -1.69012100e-01 4.04883623e-02
-7.98109174e-03 2.01037694e-02 -4.68730360e-01 -1.52575040e+00
-4.48369943e-02 -4.28462446e-01 -3.93534809e-01 5.22501230e-01
-1.89534590e-01 4.60208923e-01 1.20554864e+00 -1.16469085e+00
7.43270338e-01 -2.15193248e+00 -6.57531712e-03 -3.61653745e-01
1.24983825e-01 2.42018372e-01 -4.06263977e-01 6.29057109e-01
-1.36124298e-01 -1.46883607e-01 -6.27912760e-01 -6.97077632e-01
-3.39535415e-01 4.50862139e-01 -7.98384845e-01 2.56911516e-01
5.99746764e-01 1.21697772e+00 -8.08739126e-01 -5.50523937e-01
4.78292346e-01 6.98881328e-01 -6.17768109e-01 4.98577267e-01
-3.91522378e-01 5.26404500e-01 -6.54886067e-02 1.03532577e+00
1.08621418e+00 -7.10479692e-02 -3.48062456e-01 -1.31636381e-01
2.73765624e-01 -3.73034984e-01 -8.30898285e-01 2.05196261e+00
-5.07414937e-01 6.93747699e-01 1.83415085e-01 -8.65481734e-01
8.62319052e-01 2.20363036e-01 4.20341223e-01 -8.01386654e-01
1.31251335e-01 2.19309807e-01 -6.02973759e-01 -4.16757584e-01
6.47467792e-01 -4.74329770e-01 -1.59105770e-02 3.10408264e-01
1.02547415e-01 -5.20762920e-01 -1.53740689e-01 1.15397610e-01
1.13251197e+00 2.94491947e-01 2.04004198e-02 2.33636826e-01
1.79645881e-01 -8.40765759e-02 4.70431507e-01 6.91822469e-01
1.43021243e-02 1.43850505e+00 2.60769427e-01 -5.93307912e-01
-1.46936214e+00 -9.90443587e-01 2.23654896e-01 7.09595203e-01
1.55994490e-01 2.81027555e-02 -1.04839373e+00 -4.76273775e-01
-6.13478906e-02 7.24143803e-01 -8.94427538e-01 -3.42117876e-01
-5.23308694e-01 -3.29612702e-01 4.50131476e-01 5.96208394e-01
8.74699473e-01 -1.57224822e+00 -3.59419405e-01 3.67316157e-01
-3.25399995e-01 -1.18294191e+00 -6.81273639e-01 3.10368109e-02
-1.05323970e+00 -9.13040578e-01 -1.13743651e+00 -1.09159720e+00
9.37892914e-01 4.60844338e-01 1.17967963e+00 9.64019969e-02
-7.11713970e-01 4.21583615e-02 -4.95180607e-01 -5.27981808e-03
-7.67318547e-01 -4.18647319e-01 -5.81143200e-01 1.00971378e-01
-4.03319985e-01 -7.64351904e-01 -8.48058462e-01 2.33252719e-01
-1.39317346e+00 2.62653172e-01 8.58406126e-01 1.20982659e+00
7.08477378e-01 1.60629585e-01 6.01505756e-01 -8.94961238e-01
6.55414522e-01 -3.09468776e-01 -1.96889982e-01 1.54515401e-01
-3.52593958e-01 -7.80967847e-02 9.12035823e-01 -5.25080204e-01
-1.20765257e+00 -3.69774848e-02 -3.00529152e-01 -9.64397967e-01
-1.19624853e-01 -4.56425585e-02 -1.30435854e-01 -5.39300919e-01
6.73381627e-01 4.98887390e-01 1.92252189e-01 -4.01028097e-01
5.88219106e-01 3.31328630e-01 7.73982644e-01 -5.06580353e-01
8.55933309e-01 7.86089361e-01 -1.80276737e-01 -4.02296066e-01
-8.59644890e-01 4.63746823e-02 -3.05794209e-01 -8.34177509e-02
6.37838125e-01 -1.08127904e+00 7.21650124e-02 6.39645398e-01
-1.14072275e+00 -4.96724963e-01 -8.77985120e-01 -1.35308772e-01
-1.00035167e+00 3.32511723e-01 -9.10599411e-01 -5.04332483e-01
-5.24397433e-01 -1.07773530e+00 1.40107059e+00 7.56780133e-02
4.23762836e-02 -7.78252661e-01 -1.45767322e-02 4.34019804e-01
6.89094722e-01 8.45642626e-01 5.90794861e-01 2.59471357e-01
-8.27070892e-01 -2.06960186e-01 -3.66919875e-01 8.25213909e-01
3.24881703e-01 -2.67361850e-01 -8.98937941e-01 -3.31847668e-01
2.89357603e-01 -5.69457233e-01 9.96798038e-01 5.48433900e-01
1.36259913e+00 -4.36696589e-01 -1.16931178e-01 8.72748196e-01
1.70379746e+00 9.49500799e-02 1.33241606e+00 -9.23215412e-03
6.72974288e-01 3.73159677e-01 6.24295413e-01 3.23435605e-01
-3.25837791e-01 5.13267934e-01 4.21436250e-01 -5.54503918e-01
-8.69834244e-01 -6.16124034e-01 2.47018620e-01 3.32698554e-01
7.83418119e-02 -2.58636117e-01 -9.76355076e-02 7.32823849e-01
-1.75966454e+00 -1.26721179e+00 3.63287717e-01 1.73278379e+00
1.05677474e+00 -1.54431894e-01 -2.46715561e-01 4.14150022e-02
7.60082722e-01 3.51227164e-01 -5.52881658e-01 -3.98768991e-01
-2.17715204e-01 6.30487621e-01 3.96781474e-01 4.88036305e-01
-9.85119820e-01 1.25570905e+00 6.83879566e+00 1.38258302e+00
-9.93500412e-01 3.68926793e-01 1.10605109e+00 1.17215924e-01
-2.73603737e-01 1.00090615e-01 -1.90768182e-01 5.16377449e-01
4.67224598e-01 4.13981318e-01 6.31514072e-01 9.37262237e-01
1.28588632e-01 -1.53907597e-01 -5.67574501e-01 1.10065877e+00
4.21968520e-01 -1.71095896e+00 2.59503633e-01 -1.90055966e-01
1.35235596e+00 -5.22576630e-01 1.79089427e-01 1.16838694e-01
2.21140996e-01 -1.23585200e+00 7.60074496e-01 6.42000020e-01
1.16515446e+00 -1.06456006e+00 4.06260937e-01 1.13324329e-01
-8.43641102e-01 3.22984084e-02 -5.66784739e-01 5.68113886e-02
1.83791772e-01 6.23090088e-01 -3.49238843e-01 4.79559928e-01
5.69846213e-01 6.37733877e-01 -4.66921210e-01 8.14618349e-01
-3.88637364e-01 3.35689634e-01 1.91396326e-01 6.57046199e-01
2.57868785e-02 -1.09867997e-01 3.10917199e-01 7.96759903e-01
3.80336016e-01 9.46063641e-03 -1.37841955e-01 1.42553532e+00
-4.30826545e-01 -1.01942718e-01 -7.70773768e-01 1.09548047e-02
1.42261758e-01 1.25012839e+00 -8.45497310e-01 -3.78664255e-01
-3.42354625e-01 1.70770502e+00 2.43362501e-01 3.63016427e-01
-1.03606129e+00 -5.51270962e-01 5.84959269e-01 3.03312957e-01
5.59714675e-01 1.95133244e-03 -2.40727216e-01 -1.13961124e+00
-2.08336562e-02 -9.89991128e-01 -6.74810186e-02 -1.15340161e+00
-1.43174398e+00 6.24678135e-01 -3.41675729e-01 -1.34165454e+00
-2.06500426e-01 -1.61724448e-01 -8.17846179e-01 8.33382010e-01
-1.45285606e+00 -1.54902434e+00 -5.28043747e-01 6.89749539e-01
9.62398350e-01 -1.66199312e-01 8.13828528e-01 3.55997145e-01
-1.32409379e-01 6.18677199e-01 1.69759974e-01 4.45009209e-02
9.26991880e-01 -7.65050530e-01 4.68308330e-01 1.05295467e+00
-1.30856484e-01 3.03264409e-01 7.80326188e-01 -7.88304687e-01
-1.26080167e+00 -1.54162931e+00 4.13754612e-01 -1.37217402e-01
-2.47914391e-03 -2.32633621e-01 -7.48375535e-01 6.17038727e-01
5.84636271e-01 4.55377162e-01 1.87075391e-01 -7.19520450e-01
-3.81511003e-01 -5.13571687e-02 -1.70498908e+00 6.62392795e-01
8.83712530e-01 -4.92888957e-01 -3.60897332e-01 3.40612233e-01
8.79809141e-01 -5.24206698e-01 -7.29298711e-01 2.68134713e-01
3.63765538e-01 -1.11217237e+00 1.29420364e+00 -1.95720807e-01
1.06282783e+00 -2.78148681e-01 -1.61740512e-01 -1.24156022e+00
-2.09472761e-01 -8.48946929e-01 -2.17891634e-01 1.06736839e+00
-1.58758715e-01 -3.86981398e-01 8.12878668e-01 2.90791214e-01
-1.77016035e-01 -7.33166814e-01 -5.74561775e-01 -8.21691394e-01
-9.57177673e-03 -1.04297198e-01 5.51127434e-01 7.95748889e-01
-7.51456678e-01 -9.76680219e-02 -1.09942889e+00 -1.97538510e-01
7.75838137e-01 4.24835712e-01 8.27130556e-01 -5.18707395e-01
-2.90031850e-01 -8.19084048e-02 -2.77532339e-01 -7.02853084e-01
1.42235458e-02 -5.01398802e-01 1.27189010e-01 -1.57031500e+00
2.49173433e-01 -8.86314139e-02 7.40667284e-02 5.18131912e-01
-1.73120543e-01 8.30759764e-01 -4.65291291e-02 2.92091727e-01
-3.34669381e-01 7.84244001e-01 1.81665993e+00 -1.54142037e-01
6.53608218e-02 -2.82429129e-01 -8.40326130e-01 3.91812056e-01
7.37442434e-01 -5.73742151e-01 -4.35793132e-01 -2.37508804e-01
-1.21612988e-01 3.00893545e-01 8.03233743e-01 -1.18701184e+00
-1.52395681e-01 -2.40368262e-01 9.39421833e-01 -4.71880943e-01
5.38092673e-01 -7.28584468e-01 4.89670604e-01 4.42759156e-01
-2.63665080e-01 -2.78852105e-01 2.24622115e-01 6.18790865e-01
-5.91724157e-01 -3.96622941e-02 1.28495967e+00 -4.39357013e-01
-7.01843917e-01 4.22370911e-01 4.33798879e-02 9.78456140e-02
1.11679661e+00 -4.33360070e-01 -5.12171723e-02 -6.67686164e-01
-6.33668542e-01 -3.67511898e-01 8.96131694e-01 4.91434693e-01
9.63623047e-01 -1.64101291e+00 -6.85880184e-01 4.94645655e-01
-1.18046768e-01 3.45235057e-02 5.56559145e-01 4.25331920e-01
-9.87060487e-01 -1.03381649e-01 -6.46284819e-01 -1.39174074e-01
-9.37281489e-01 7.75016189e-01 2.41731927e-01 -1.95194751e-01
-7.48780012e-01 8.74156058e-01 2.36625552e-01 -2.61125088e-01
-1.03296535e-02 -1.89975783e-01 5.00772834e-01 -3.63878191e-01
5.83027840e-01 9.01319087e-02 -1.21509910e-01 -3.70672971e-01
1.84321135e-01 4.48373139e-01 -9.83122587e-02 -3.95455882e-02
1.50211787e+00 -3.33914906e-02 -2.06110373e-01 -1.25099301e-01
1.23201191e+00 8.74013919e-03 -1.83827460e+00 -1.08720846e-01
-7.56211460e-01 -9.66131866e-01 -4.17749807e-02 -7.67804623e-01
-1.54706454e+00 7.79215693e-01 5.42784512e-01 -3.65095772e-02
1.39587629e+00 -1.44578427e-01 1.27165234e+00 -2.94954181e-01
3.45624030e-01 -1.04385245e+00 5.74870229e-01 5.16172573e-02
1.39401019e+00 -1.26884258e+00 5.79547994e-02 -5.51075816e-01
-4.91220802e-01 9.92469668e-01 5.83958983e-01 -7.40947783e-01
3.78386706e-01 3.37315261e-01 2.38151755e-02 -4.35522571e-02
-4.56742942e-01 1.44065082e-01 1.02993183e-01 6.82308912e-01
1.31919816e-01 -1.04517654e-01 -1.69365644e-01 3.03852975e-01
1.91080987e-01 2.20156506e-01 2.69241184e-01 1.13244283e+00
-3.80790740e-01 -1.25395679e+00 -5.89610815e-01 1.42502725e-01
-4.49650198e-01 -2.88752794e-01 -3.13269854e-01 6.68734074e-01
2.96954542e-01 8.37995946e-01 -1.42016336e-01 -2.36142218e-01
2.25590039e-02 -1.64924338e-01 7.46374786e-01 -4.39289749e-01
-4.53449667e-01 1.83831781e-01 -3.14274698e-01 -7.96177268e-01
-3.60167950e-01 -1.76911220e-01 -7.94481277e-01 -4.12721515e-01
-9.46386755e-02 -3.21108580e-01 3.33093137e-01 5.61225235e-01
5.70538580e-01 7.78434217e-01 7.47602999e-01 -1.09011710e+00
-2.96184540e-01 -9.10081327e-01 -7.21771181e-01 6.69032097e-01
3.18378478e-01 -4.40474182e-01 -3.60768735e-01 3.45337540e-01] | [11.519621849060059, -0.9140453338623047] |
5cc5a8db-216d-43f1-9eed-9f40ca59a1c5 | a-framework-for-provably-stable-and | 2305.12125 | null | https://arxiv.org/abs/2305.12125v1 | https://arxiv.org/pdf/2305.12125v1.pdf | A Framework for Provably Stable and Consistent Training of Deep Feedforward Networks | We present a novel algorithm for training deep neural networks in supervised (classification and regression) and unsupervised (reinforcement learning) scenarios. This algorithm combines the standard stochastic gradient descent and the gradient clipping method. The output layer is updated using clipped gradients, the rest of the neural network is updated using standard gradients. Updating the output layer using clipped gradient stabilizes it. We show that the remaining layers are automatically stabilized provided the neural network is only composed of squashing (compact range) activations. We also present a novel squashing activation function - it is obtained by modifying a Gaussian Error Linear Unit (GELU) to have compact range - we call it Truncated GELU (tGELU). Unlike other squashing activations, such as sigmoid, the range of tGELU can be explicitly specified. As a consequence, the problem of vanishing gradients that arise due to a small range, e.g., in the case of a sigmoid activation, is eliminated. We prove that a NN composed of squashing activations (tGELU, sigmoid, etc.), when updated using the algorithm presented herein, is numerically stable and has consistent performance (low variance). The theory is supported by extensive experiments. Within reinforcement learning, as a consequence of our study, we show that target networks in Deep Q-Learning can be omitted, greatly speeding up learning and alleviating memory requirements. Cross-entropy based classification algorithms that suffer from high variance issues are more consistent when trained using our framework. One symptom of numerical instability in training is the high variance of the neural network update values. We show, in theory and through experiments, that our algorithm updates have low variance, and the training loss reduces in a smooth manner. | ['Naman Saxena', 'Shalabh Bhatnagar', 'Arunselvan Ramaswamy'] | 2023-05-20 | null | null | null | null | ['q-learning'] | ['methodology'] | [ 1.74902022e-01 3.01395118e-01 -8.51282030e-02 -1.68727368e-01
-3.70119989e-01 -4.72567499e-01 2.41814747e-01 2.37955451e-02
-8.87216449e-01 1.24881911e+00 -3.53756487e-01 -3.60593051e-01
-3.36455703e-01 -6.22164249e-01 -1.27249634e+00 -1.20415390e+00
-1.46646038e-01 4.06941250e-02 2.28860602e-01 -2.92895675e-01
2.63793916e-01 4.29113328e-01 -1.36335266e+00 -3.47205728e-01
9.65279043e-01 1.11871326e+00 -5.05561866e-02 7.02299714e-01
1.32844463e-01 9.08460319e-01 -5.32844424e-01 -1.88017666e-01
4.53158170e-01 -4.65590030e-01 -6.50235772e-01 -9.21431929e-02
2.00271845e-01 -2.79819548e-01 -1.43287286e-01 1.10675430e+00
5.61806500e-01 4.90683407e-01 5.99638343e-01 -1.09667695e+00
-5.31355619e-01 7.50123441e-01 -3.02457362e-01 -1.06812537e-01
-3.76803458e-01 -8.62499475e-02 8.17080796e-01 -7.50289321e-01
5.43493629e-01 8.79482269e-01 1.08231425e+00 8.35882902e-01
-1.40569627e+00 -5.36281466e-01 3.33265543e-01 -3.27662408e-01
-1.09104776e+00 -1.50462165e-01 7.33294666e-01 -3.10920268e-01
9.37531710e-01 2.03790307e-01 5.80843568e-01 8.76581848e-01
4.05947536e-01 7.42145896e-01 9.83108342e-01 -5.72867334e-01
6.32956982e-01 4.80817258e-01 1.72053367e-01 8.86412978e-01
2.24757969e-01 3.34403574e-01 -1.43257141e-01 -1.08340867e-01
7.71231711e-01 -2.13166978e-02 -3.10775131e-01 -8.30173969e-01
-7.70871997e-01 1.09994888e+00 5.91925383e-01 2.47323036e-01
-3.04062843e-01 5.36033392e-01 4.23270792e-01 6.09358430e-01
5.78739166e-01 6.01003945e-01 -6.07361376e-01 5.75756170e-02
-9.44037557e-01 2.74779588e-01 9.16895390e-01 5.62051892e-01
8.07867408e-01 5.15653491e-01 -7.32365400e-02 9.68178391e-01
-4.67553400e-02 3.65705013e-01 8.16289425e-01 -1.09840524e+00
2.58971423e-01 1.43079504e-01 2.56199092e-01 -5.38995564e-01
-6.59106851e-01 -7.53848732e-01 -9.99542296e-01 7.41804123e-01
5.16090631e-01 -8.41735423e-01 -7.72202313e-01 2.10756302e+00
1.44166231e-01 2.53502131e-02 1.52950630e-01 6.94208086e-01
1.70025378e-01 5.95962524e-01 -1.75691262e-01 -5.33906519e-01
7.06969738e-01 -7.77806044e-01 -6.03745103e-01 1.29833957e-02
8.07828546e-01 -1.90404147e-01 9.93509352e-01 3.99130076e-01
-1.27368867e+00 -2.99390227e-01 -1.20442247e+00 2.82399476e-01
-5.48139215e-01 4.81123142e-02 5.00646472e-01 7.63427556e-01
-1.36912048e+00 1.30262089e+00 -8.62201810e-01 7.31568933e-02
2.31651917e-01 7.32199252e-01 -1.60499021e-01 5.40043175e-01
-1.29710662e+00 1.06826627e+00 5.54922223e-01 2.24091485e-01
-4.10554737e-01 -7.85348773e-01 -8.27820539e-01 1.33565113e-01
3.12353253e-01 -4.82041359e-01 1.46967983e+00 -1.50103462e+00
-2.02005863e+00 2.94373572e-01 4.28622253e-02 -9.16971445e-01
8.47128451e-01 -1.18834421e-01 1.33791432e-01 -1.66929215e-02
-3.66985023e-01 6.65710032e-01 1.07492137e+00 -1.15614605e+00
-4.89680469e-01 -1.01981051e-01 -1.80645481e-01 1.62190944e-01
-5.75112939e-01 -2.14815319e-01 3.08881372e-01 -6.65181637e-01
-2.44323276e-02 -9.77839351e-01 -4.81910318e-01 -9.77074578e-02
-1.39532641e-01 -1.59679055e-02 6.94853961e-01 -3.78209084e-01
1.30452371e+00 -1.99077082e+00 7.57643357e-02 6.08276188e-01
1.22427218e-01 2.02443674e-01 1.52383655e-01 1.17677696e-01
-3.53762418e-01 2.36600086e-01 -7.73995340e-01 -3.46099198e-01
-5.59540987e-02 3.25283676e-01 -2.57748157e-01 5.32308638e-01
3.17937970e-01 8.51136029e-01 -9.78294313e-01 -2.37643957e-01
-2.19460074e-02 5.26911795e-01 -6.96649432e-01 -1.11941986e-01
1.79879535e-02 3.97405960e-02 -2.56111830e-01 1.02210954e-01
4.92022455e-01 -1.16160020e-01 2.61230581e-02 2.13743031e-01
-2.96176046e-01 6.23114556e-02 -1.42245626e+00 1.05450141e+00
-5.92624784e-01 6.86435223e-01 9.26086828e-02 -1.36218274e+00
1.01970530e+00 2.68827170e-01 4.59402889e-01 -2.45589986e-01
5.94827868e-02 5.01505733e-01 -7.45903999e-02 -1.16281480e-01
3.25617552e-01 -2.28043884e-01 2.48293981e-01 3.84649277e-01
2.34546646e-01 -3.15980464e-02 1.84202686e-01 -1.08447634e-02
8.94926786e-01 1.64990976e-01 6.24937341e-02 -4.61043775e-01
3.96269083e-01 -3.31084371e-01 4.62146908e-01 9.42342639e-01
-2.03083709e-01 3.32061350e-01 7.74219215e-01 -1.36177436e-01
-1.15544975e+00 -9.42471743e-01 -4.02768850e-01 1.07219219e+00
-3.00006658e-01 6.59448840e-03 -9.74866569e-01 -5.28760314e-01
2.38388449e-01 5.41993082e-01 -8.31675351e-01 -4.75277722e-01
-6.21791363e-01 -1.05743778e+00 5.11542201e-01 4.88000482e-01
5.83449125e-01 -1.13248742e+00 -7.32920885e-01 5.54083407e-01
5.13812304e-01 -5.54979622e-01 -4.73690182e-01 9.28239763e-01
-1.31239831e+00 -6.53048337e-01 -1.06746733e+00 -9.48992729e-01
7.45974839e-01 -4.13633555e-01 9.02102768e-01 -1.36824995e-01
2.01773897e-01 2.86574572e-01 2.94576418e-02 -2.79417872e-01
-4.16376084e-01 8.18421170e-02 2.79744238e-01 -2.98496690e-02
-2.03401029e-01 -6.16888881e-01 -5.11279821e-01 7.52149969e-02
-7.85392880e-01 -2.30699643e-01 3.54195416e-01 1.29583478e+00
4.69705999e-01 4.51305648e-03 8.69027615e-01 -1.02205598e+00
8.84249806e-01 -2.40426749e-01 -9.93952453e-01 7.06495643e-02
-1.06642497e+00 5.44334173e-01 9.00577486e-01 -8.66953850e-01
-8.61591041e-01 1.57161862e-01 -6.75246790e-02 -4.09895182e-01
2.83067703e-01 3.41302752e-01 4.92369026e-01 -4.85092819e-01
8.40079010e-01 2.96971183e-02 3.09376061e-01 -3.19585800e-01
2.53514767e-01 3.87199193e-01 3.19035560e-01 -4.83304828e-01
5.88638902e-01 7.13571087e-02 1.13355912e-01 -6.42380476e-01
-6.73943520e-01 7.86377490e-02 -2.16101199e-01 -3.04060787e-01
5.03979802e-01 -4.23237860e-01 -9.77407515e-01 4.60806131e-01
-8.46200049e-01 -6.17683709e-01 -7.42525101e-01 5.34886062e-01
-6.71648741e-01 1.05450884e-01 -8.63666832e-01 -1.25913048e+00
-4.04611468e-01 -1.05368483e+00 3.26746166e-01 3.39078754e-01
1.42750321e-02 -1.30140710e+00 5.03168302e-03 -4.67319638e-01
4.61996824e-01 3.96749824e-01 5.93286276e-01 -6.33087099e-01
8.19994211e-02 -2.73384694e-02 2.33548328e-01 9.11275268e-01
-1.08994126e-01 2.14166760e-01 -8.91702890e-01 -3.67218137e-01
2.88601905e-01 -4.01004463e-01 1.17978954e+00 8.33366811e-01
1.13905573e+00 -5.50314844e-01 8.59598219e-02 6.93276644e-01
1.42550087e+00 2.88325191e-01 3.07775646e-01 5.33076763e-01
4.26976055e-01 5.48450828e-01 2.05566794e-01 3.55695397e-01
-2.57338226e-01 2.02582434e-01 3.15888226e-01 -1.70586824e-01
4.04051065e-01 -2.67142672e-02 7.34165907e-01 6.78929448e-01
-9.08538401e-02 2.61664867e-01 -6.00057781e-01 2.53945261e-01
-2.02594924e+00 -1.00227451e+00 8.40723328e-03 2.41370368e+00
1.32520628e+00 5.31829774e-01 2.00079411e-01 3.91806304e-01
7.03994036e-01 -2.47710198e-01 -8.82635832e-01 -1.04941261e+00
-2.55904526e-01 5.59298098e-01 9.82371330e-01 9.49430466e-01
-1.08806932e+00 6.29031539e-01 6.68006706e+00 7.51202941e-01
-1.20592499e+00 -2.01553050e-02 8.81526232e-01 -1.18784465e-01
-6.63805604e-02 -3.24774832e-01 -8.22339416e-01 4.07536358e-01
1.01952934e+00 -9.73326247e-03 5.05639136e-01 1.16422522e+00
3.44341725e-01 -1.57061502e-01 -9.28136051e-01 6.90398455e-01
-3.53505760e-01 -1.34222102e+00 -3.16363424e-01 -1.57132179e-01
1.06071639e+00 -2.85392739e-02 2.34570920e-01 5.38440764e-01
2.87440151e-01 -8.93911541e-01 7.51013815e-01 3.85047644e-01
4.98514146e-01 -1.07411420e+00 7.55317926e-01 2.49822885e-01
-6.82744026e-01 -3.84028941e-01 -4.89273518e-01 -6.36753663e-02
-3.45458210e-01 7.90498197e-01 -5.66324115e-01 1.71808586e-01
4.44116026e-01 4.09222931e-01 -2.79334784e-01 1.00591207e+00
1.30140660e-02 7.39707053e-01 -5.28623581e-01 -5.18004835e-01
5.47841549e-01 -5.54471970e-01 4.47288990e-01 1.25825322e+00
1.45878449e-01 -2.72976458e-01 -1.54900998e-01 8.72145832e-01
-1.91061318e-01 1.49283007e-01 -4.68529969e-01 3.58445138e-01
2.14262694e-01 1.07570553e+00 -5.65016508e-01 -3.75358194e-01
-9.46694314e-02 7.46287227e-01 4.03171837e-01 6.93105102e-01
-7.93727934e-01 -7.58553803e-01 3.70318502e-01 -1.48937941e-01
3.78784835e-01 8.32470879e-02 -4.88956571e-01 -9.29231286e-01
1.29242823e-01 -5.32172203e-01 2.24071115e-01 -4.39151317e-01
-1.03384185e+00 4.57823485e-01 -6.29963651e-02 -9.49908614e-01
-6.02743983e-01 -8.08276176e-01 -5.05304277e-01 8.14212978e-01
-1.44348061e+00 -1.63167074e-01 3.79993647e-01 6.83669508e-01
1.64176062e-01 -3.95227745e-02 6.01415753e-01 5.70042655e-02
-7.24116087e-01 8.58888626e-01 7.77382314e-01 9.04103965e-02
3.27482134e-01 -1.56753695e+00 -1.47586927e-01 6.15753353e-01
-2.14572057e-01 6.15859866e-01 7.47902632e-01 -3.23030382e-01
-9.25537109e-01 -1.09693694e+00 6.77217841e-01 2.98973378e-02
7.81940579e-01 -3.74826223e-01 -1.08610070e+00 5.35115778e-01
-8.58613402e-02 4.73764129e-02 1.82188705e-01 -7.35725760e-02
2.41579022e-02 -3.37868750e-01 -1.20935404e+00 7.00063109e-01
4.68088120e-01 -2.38395393e-01 -4.84767616e-01 3.31329823e-01
6.42276585e-01 -3.83657217e-01 -7.94684649e-01 4.48471248e-01
5.64173698e-01 -8.45525503e-01 7.03882873e-01 -7.32807159e-01
1.90127611e-01 7.10908175e-02 3.62037301e-01 -1.57592511e+00
-7.86500499e-02 -1.03318477e+00 -3.82299989e-01 7.54185379e-01
6.48027420e-01 -1.07073295e+00 8.84643018e-01 6.82459950e-01
-9.27124918e-02 -1.22544873e+00 -1.02173293e+00 -1.04384065e+00
5.76892436e-01 -1.30388960e-01 -3.05950902e-02 7.42066324e-01
2.01084808e-01 1.88051432e-01 -2.52052009e-01 -2.15437442e-01
4.45900440e-01 -2.99786091e-01 1.63111001e-01 -1.03234327e+00
-3.99982989e-01 -8.34379554e-01 -6.50035292e-02 -9.45710778e-01
8.88418928e-02 -7.30309427e-01 1.34587333e-01 -9.98145223e-01
-2.49891669e-01 -5.05121648e-01 -5.81772089e-01 4.28271353e-01
-1.13941126e-01 1.00128569e-01 3.03676967e-02 -1.49262235e-01
-2.01734491e-02 5.79260886e-01 1.08558083e+00 1.48977995e-01
-6.89313948e-01 3.00493777e-01 -2.85543799e-01 8.02703440e-01
1.26067269e+00 -6.08014822e-01 -3.40119600e-01 1.48488328e-01
2.89391160e-01 -8.92006233e-02 2.53153861e-01 -1.00174832e+00
1.73613593e-01 7.69114792e-02 4.72755432e-01 -5.18197529e-02
1.58192590e-01 -5.97871423e-01 -3.20149362e-01 9.28244829e-01
-8.01059067e-01 2.33690478e-02 1.77170962e-01 2.47284979e-01
4.01561633e-02 -8.18769455e-01 1.04529524e+00 6.76459149e-02
-7.38867968e-02 1.30432874e-01 -5.96520185e-01 4.35829014e-02
6.86216533e-01 -1.80077061e-01 1.47542328e-01 -5.07470250e-01
-8.80837381e-01 3.08900923e-01 3.28834131e-02 -2.06604481e-01
4.93669301e-01 -1.30740380e+00 -5.03428340e-01 1.85846359e-01
-5.42300701e-01 -1.22561365e-01 -1.89149961e-01 9.01537359e-01
-3.92553240e-01 1.01875044e-01 7.22289532e-02 -2.93836385e-01
-7.40437210e-01 4.29270536e-01 9.58413243e-01 -3.50007147e-01
-4.91628766e-01 8.95693719e-01 -6.23429120e-02 -4.66140747e-01
6.09266639e-01 -7.11800873e-01 1.01288799e-02 1.00117512e-01
1.85239613e-01 2.66920507e-01 2.03019872e-01 8.69600326e-02
-4.86287586e-02 4.92401719e-01 1.18249906e-02 -3.17493081e-01
1.40851319e+00 1.47927627e-01 8.92008245e-02 6.82864189e-01
1.35126305e+00 -1.49032727e-01 -1.63535714e+00 -3.31709348e-02
2.05522515e-02 2.81038344e-01 2.25061342e-01 -4.95980322e-01
-1.12500954e+00 7.64610946e-01 7.18404472e-01 3.93566191e-01
1.10421336e+00 -6.49512649e-01 5.10179460e-01 9.29122150e-01
4.18208875e-02 -1.56307578e+00 -1.23455919e-01 7.32818067e-01
6.99413180e-01 -1.04348874e+00 -1.64840236e-01 1.46972388e-01
-4.54897255e-01 1.42983067e+00 4.34091896e-01 -6.52965963e-01
7.92671859e-01 3.95751983e-01 -2.16141745e-01 4.31584984e-01
-7.73540556e-01 -6.18224069e-02 1.35685375e-04 3.40647757e-01
2.61158913e-01 -1.72736838e-01 -5.57447135e-01 3.98555309e-01
-2.63640553e-01 -7.72227719e-02 6.44785285e-01 9.84838188e-01
-6.67066395e-01 -8.85734141e-01 -2.61839688e-01 4.78588700e-01
-6.02231085e-01 -2.46141225e-01 -3.77224199e-02 9.79821801e-01
-1.41016558e-01 6.05051696e-01 1.03293009e-01 -1.35680571e-01
1.43290401e-01 2.75085568e-01 4.80226427e-01 -1.93623245e-01
-9.37794030e-01 4.10002880e-02 -1.40946776e-01 -3.40193212e-01
-2.90737122e-01 -4.13502336e-01 -1.46358323e+00 -2.84086555e-01
-5.57999372e-01 3.97290468e-01 7.21865416e-01 7.81677246e-01
-1.26787528e-01 5.72193146e-01 9.01376009e-01 -7.92263269e-01
-1.25691664e+00 -7.89611459e-01 -7.32029319e-01 9.20706689e-02
7.12793469e-01 -6.27891600e-01 -8.06909800e-01 -1.13754466e-01] | [7.6360979080200195, 3.618036985397339] |
f6663462-75d6-4df2-8d9f-889114785a7e | spiral-contrastive-learning-an-efficient-3d | 2208.10694 | null | https://arxiv.org/abs/2208.10694v1 | https://arxiv.org/pdf/2208.10694v1.pdf | Spiral Contrastive Learning: An Efficient 3D Representation Learning Method for Unannotated CT Lesions | Computed tomography (CT) samples with pathological annotations are difficult to obtain. As a result, the computer-aided diagnosis (CAD) algorithms are trained on small datasets (e.g., LIDC-IDRI with 1,018 samples), limiting their accuracies and reliability. In the past five years, several works have tailored for unsupervised representations of CT lesions via two-dimensional (2D) and three-dimensional (3D) self-supervised learning (SSL) algorithms. The 2D algorithms have difficulty capturing 3D information, and existing 3D algorithms are computationally heavy. Light-weight 3D SSL remains the boundary to explore. In this paper, we propose the spiral contrastive learning (SCL), which yields 3D representations in a computationally efficient manner. SCL first transforms 3D lesions to the 2D plane using an information-preserving spiral transformation, and then learn transformation-invariant features using 2D contrastive learning. For the augmentation, we consider natural image augmentations and medical image augmentations. We evaluate SCL by training a classification head upon the embedding layer. Experimental results show that SCL achieves state-of-the-art accuracy on LIDC-IDRI (89.72%), LNDb (82.09%) and TianChi (90.16%) for unsupervised representation learning. With 10% annotated data for fine-tune, the performance of SCL is comparable to that of supervised learning algorithms (85.75% vs. 85.03% on LIDC-IDRI, 78.20% vs. 73.44% on LNDb and 87.85% vs. 83.34% on TianChi, respectively). Meanwhile, SCL reduces the computational effort by 66.98% compared to other 3D SSL algorithms, demonstrating the efficiency of the proposed method in unsupervised pre-training. | ['Jinpeng Li', 'Xin Wei', 'Baolian Qi', 'Enwei Zhu', 'Penghua Zhai'] | 2022-08-23 | null | null | null | null | ['unsupervised-pre-training'] | ['methodology'] | [ 3.26633990e-01 4.04321641e-01 -5.03119051e-01 -4.74021323e-02
-1.14182258e+00 1.90367736e-02 3.75024527e-01 -6.69047236e-02
-5.67763865e-01 4.69342291e-01 4.17376399e-01 -2.69339263e-01
-8.22786242e-02 -6.15939796e-01 -3.28585654e-01 -9.91987646e-01
-4.08657193e-02 6.45328820e-01 1.23622872e-01 2.84353137e-01
-2.32093275e-01 6.18594587e-01 -1.23839641e+00 3.05849612e-01
7.63471365e-01 1.24197006e+00 1.81418002e-01 4.74095821e-01
-1.00472786e-01 9.43417072e-01 -3.17583859e-01 -1.59986652e-02
1.70244023e-01 -3.71645093e-01 -9.36777830e-01 3.15191776e-01
1.84056491e-01 -3.80173117e-01 -6.75146937e-01 7.79958785e-01
6.01479232e-01 -3.83001029e-01 1.31390452e+00 -1.03853536e+00
-5.90970516e-01 2.16959491e-01 -8.87823999e-01 1.29244968e-01
5.59397191e-02 7.98133463e-02 5.69639266e-01 -1.02869499e+00
6.82646096e-01 8.96616697e-01 7.46116042e-01 6.64713740e-01
-1.03592610e+00 -6.96151793e-01 -3.19419473e-01 1.71091571e-01
-1.45233989e+00 -4.01902907e-02 7.77295113e-01 -5.80332756e-01
6.68268263e-01 2.63883501e-01 5.26952505e-01 1.01481152e+00
2.42509872e-01 9.71005917e-01 1.35009539e+00 -4.90511924e-01
4.37867045e-02 2.12375119e-01 1.47745967e-01 9.81669724e-01
4.44194466e-01 1.73000515e-01 -1.67218119e-01 -1.88989043e-01
1.01602578e+00 1.70447737e-01 -2.71483719e-01 -3.36047500e-01
-1.08903801e+00 8.22887659e-01 8.30791295e-01 2.71239042e-01
-3.18124831e-01 -2.47039989e-01 6.80882633e-01 3.03915441e-02
5.81701696e-01 1.74920529e-01 -1.16701514e-01 1.80576429e-01
-6.84510887e-01 -3.73872817e-01 4.15020645e-01 9.56170499e-01
4.38482881e-01 -7.23601738e-03 -3.17513868e-02 1.03221273e+00
3.13009262e-01 4.55277115e-01 1.15997422e+00 -4.56328988e-01
2.95597941e-01 8.45159173e-01 -5.21108627e-01 -6.96602046e-01
-7.40727067e-01 -6.43997252e-01 -1.63673472e+00 1.48060277e-01
7.03601912e-02 2.42357060e-01 -1.27582169e+00 1.26328135e+00
2.07501680e-01 7.47385025e-02 2.15052098e-01 8.12273622e-01
1.04500937e+00 4.68049288e-01 5.21272048e-02 -4.60356534e-01
1.40358794e+00 -9.73883808e-01 -7.73912847e-01 -1.02055736e-01
1.20636630e+00 -5.89549899e-01 1.19828963e+00 3.36382151e-01
-8.17936897e-01 -4.79581267e-01 -9.23198283e-01 5.35913333e-02
-1.87759236e-01 5.59673786e-01 5.47383785e-01 7.05400229e-01
-8.93964648e-01 3.62558588e-02 -1.10707915e+00 -3.30392420e-01
8.07216287e-01 3.82377982e-01 -6.18773639e-01 -5.19691885e-01
-8.92653644e-01 8.50360096e-01 2.17924386e-01 -1.46251842e-01
-8.49578381e-01 -7.34616876e-01 -8.20032418e-01 -3.34693223e-01
1.12782963e-01 -4.82319683e-01 8.69533181e-01 -4.95518297e-01
-1.38399613e+00 1.34518468e+00 9.11308825e-02 -3.93849075e-01
5.48637450e-01 6.10272773e-02 -3.26697141e-01 4.80644196e-01
2.58832276e-01 7.24137247e-01 4.51973319e-01 -1.13294983e+00
-3.48042518e-01 -6.33297682e-01 -3.52344483e-01 2.56821692e-01
-7.59431183e-01 -3.32102060e-01 -5.57091177e-01 -8.61611009e-01
7.32801437e-01 -1.06292760e+00 -3.24970871e-01 5.20289302e-01
-4.34707671e-01 -1.47432193e-01 9.84705567e-01 -6.62018239e-01
1.10394180e+00 -2.17190146e+00 -1.80079788e-01 1.36615247e-01
3.26997787e-01 3.33073765e-01 1.23386301e-01 -1.08285896e-01
-3.16487253e-01 1.36976957e-01 -4.95041639e-01 -4.35471773e-01
-4.50613439e-01 2.84607887e-01 3.66436511e-01 7.24875867e-01
2.19774410e-01 9.14363027e-01 -6.29907310e-01 -9.82609689e-01
2.70553142e-01 4.72526819e-01 -7.27991581e-01 2.44435057e-01
4.77706641e-01 5.39926589e-01 -6.12499237e-01 9.32421803e-01
6.77353442e-01 -5.80039442e-01 1.20070815e-01 -6.47286713e-01
1.48619279e-01 -1.37372874e-02 -8.78248870e-01 1.75265515e+00
-7.28639543e-01 4.18601036e-01 -3.96652788e-01 -1.26806939e+00
9.48860705e-01 5.17028272e-01 8.74093711e-01 -7.77834296e-01
2.78517306e-01 3.95132780e-01 -6.42071366e-02 -7.73663759e-01
-1.30929619e-01 -2.61766970e-01 -1.07773244e-01 5.40081501e-01
2.28980064e-01 -5.54603823e-02 -2.81016558e-01 8.64562467e-02
1.15292645e+00 -4.03424978e-01 6.36594713e-01 -1.50475651e-01
6.10806763e-01 -1.41527101e-01 4.07818258e-01 4.12426054e-01
-1.76152527e-01 8.06937397e-01 1.92855716e-01 -3.85018349e-01
-7.46073067e-01 -1.12328482e+00 -6.16758168e-01 2.99084455e-01
-1.82737578e-02 -1.98923647e-01 -5.91716051e-01 -9.30152118e-01
-2.02829853e-01 2.67993897e-01 -7.68923283e-01 -5.42452931e-01
-6.17697895e-01 -9.61119473e-01 5.80956578e-01 8.23781908e-01
7.58931577e-01 -8.28180790e-01 -3.92983228e-01 -2.78556664e-02
-1.03911713e-01 -1.14227915e+00 -3.43222141e-01 1.82685629e-01
-1.35240102e+00 -1.15032172e+00 -9.98482406e-01 -1.20952368e+00
1.14801323e+00 1.47861347e-01 7.17339694e-01 9.24991220e-02
-6.78727210e-01 4.23893154e-01 -4.19328868e-01 -2.27502495e-01
-4.21322376e-01 1.06737986e-01 1.78276658e-01 -4.50825617e-02
5.69188371e-02 -4.98443753e-01 -5.63757539e-01 4.84622449e-01
-1.03265679e+00 1.99305832e-01 1.22974789e+00 1.25537848e+00
1.14845645e+00 -9.85959619e-02 4.89528239e-01 -1.05716801e+00
1.10322721e-01 -3.59261870e-01 -1.71206117e-01 7.37373680e-02
-7.56948888e-01 -8.99767429e-02 5.73992610e-01 -5.95032990e-01
-8.80375862e-01 3.48549724e-01 -2.42524609e-01 -5.81120968e-01
-1.92991123e-01 5.08298159e-01 -1.95425786e-02 -1.16915658e-01
8.13200176e-01 4.50548112e-01 3.52897674e-01 -5.22096753e-01
7.08519667e-02 1.02082765e+00 4.76621687e-01 -4.10712183e-01
8.14074397e-01 6.00409567e-01 2.74608824e-02 -7.24452496e-01
-9.30863202e-01 -6.33294523e-01 -7.99952149e-01 -1.84846222e-01
7.74798334e-01 -9.25211012e-01 -3.74637932e-01 4.26304668e-01
-6.31355405e-01 -1.00531660e-01 -5.38211226e-01 9.42183673e-01
-7.29786992e-01 2.40217403e-01 -6.37159824e-01 -5.26613116e-01
-6.67206824e-01 -1.37149000e+00 1.06524909e+00 -4.04657284e-03
-1.46292597e-01 -7.87635505e-01 -8.68554413e-02 4.54707295e-01
2.10102662e-01 3.34014267e-01 1.06704211e+00 -5.22702813e-01
-3.37898612e-01 -5.21373630e-01 -4.70516980e-01 5.81148982e-01
4.54378068e-01 -6.14603281e-01 -1.05965185e+00 -3.74295980e-01
5.76117449e-02 -6.77078009e-01 7.81081021e-01 4.71810937e-01
1.53657258e+00 -2.79903144e-01 -6.00069880e-01 8.33622694e-01
1.27000928e+00 2.27086693e-01 7.68036544e-01 3.99257451e-01
5.71943462e-01 1.45467386e-01 5.92833459e-01 3.64378452e-01
2.04501867e-01 5.03623784e-01 3.71912122e-01 -4.41911906e-01
-6.33487701e-01 -1.79141700e-01 3.45513746e-02 1.14323533e+00
-1.44177347e-01 1.54902041e-01 -1.12255073e+00 6.01559043e-01
-1.35548663e+00 -4.72204506e-01 1.90032343e-03 1.88708460e+00
9.84249592e-01 2.55388975e-01 -1.45809084e-01 6.32358432e-01
4.09777015e-01 -5.38199358e-02 -7.13516057e-01 2.40994722e-01
8.35357308e-02 4.42915469e-01 4.87637639e-01 5.63031174e-02
-1.26957071e+00 5.10024965e-01 5.16165638e+00 9.33827817e-01
-1.13379836e+00 2.60015547e-01 8.77950430e-01 1.33009762e-01
2.33011812e-01 -5.95677316e-01 -6.57121480e-01 2.80751914e-01
6.14370763e-01 6.22724853e-02 -2.42640749e-01 9.41941023e-01
9.75339785e-02 1.30765317e-02 -8.45806837e-01 1.37011290e+00
3.47516477e-01 -1.35096109e+00 1.51121527e-01 1.96582690e-01
7.74554133e-01 -9.16489214e-02 2.65527666e-01 3.56506497e-01
-1.86446086e-01 -1.14914095e+00 1.16354302e-01 3.38771969e-01
1.46881390e+00 -5.96897066e-01 1.14871490e+00 3.44597131e-01
-1.11522722e+00 4.67159078e-02 -3.37870657e-01 4.95446205e-01
1.55420639e-02 5.54579675e-01 -9.97703433e-01 6.77519619e-01
7.96289325e-01 8.43876779e-01 -5.88455439e-01 1.05409157e+00
-1.00133330e-01 8.40848267e-01 -4.28599656e-01 1.95108309e-01
2.64840752e-01 2.01819614e-01 1.83192581e-01 1.17064273e+00
2.77140081e-01 3.27287465e-01 1.82433024e-01 3.53320658e-01
-2.54427314e-01 2.18187615e-01 -4.53413993e-01 2.34369680e-01
2.08014086e-01 1.16827190e+00 -6.12037659e-01 -3.11603278e-01
-2.77654767e-01 9.06780005e-01 1.21706560e-01 -6.73751812e-03
-8.22526515e-01 -2.18646303e-01 1.49851069e-01 2.62062997e-01
-6.99033961e-02 9.30879936e-02 -3.83437246e-01 -9.20153379e-01
-1.00742929e-01 -7.67423213e-01 7.43949354e-01 -6.98890567e-01
-1.44781971e+00 9.25662875e-01 -2.11944640e-01 -1.79114771e+00
-6.13822602e-02 -8.52185786e-01 -3.05925369e-01 4.76541460e-01
-1.57677805e+00 -1.25970626e+00 -6.08013511e-01 7.05173135e-01
4.64970440e-01 -4.37845767e-01 1.12989366e+00 3.96183044e-01
-4.41445261e-01 1.06503630e+00 2.24662706e-01 3.87606025e-01
6.82634592e-01 -9.51540232e-01 -3.04108679e-01 1.80878744e-01
-5.77161014e-02 6.92435727e-02 -1.21628225e-01 -2.40583435e-01
-1.31613719e+00 -1.45976317e+00 5.45310318e-01 -8.51506218e-02
3.86611849e-01 -2.49167290e-02 -9.85162377e-01 5.48756540e-01
-2.75210112e-01 6.14780545e-01 9.29709256e-01 -3.37244511e-01
-3.42470646e-01 -1.60339162e-01 -1.47597003e+00 4.16680872e-01
1.08021331e+00 -5.83454847e-01 -3.58530939e-01 5.06372750e-01
5.38802862e-01 -4.37198550e-01 -1.26245856e+00 6.71101987e-01
4.84023809e-01 -6.24227881e-01 1.12950170e+00 -4.23701674e-01
3.21662217e-01 -1.10201709e-01 -2.39213392e-01 -9.09217536e-01
-2.98517078e-01 3.68250161e-02 -4.59660888e-02 7.59720743e-01
2.67109275e-01 -6.59514487e-01 9.45486367e-01 7.00583905e-02
-4.42659259e-01 -1.30074656e+00 -9.98257875e-01 -8.19556773e-01
2.06594974e-01 -3.93409848e-01 3.63196991e-02 1.04129815e+00
-1.50541231e-01 8.51284266e-02 -8.26920271e-02 2.02354714e-01
6.79216325e-01 -1.81842908e-01 4.70304072e-01 -1.18033707e+00
-7.58599117e-02 -2.28278548e-01 -7.08021700e-01 -1.00055766e+00
-8.78184661e-02 -1.35873044e+00 -3.02431434e-01 -1.60293424e+00
4.90048826e-01 -7.07820058e-01 -4.95753974e-01 6.03614688e-01
5.31371497e-02 5.32955527e-01 -1.02382712e-01 5.69985926e-01
-2.17712685e-01 6.08035624e-01 1.61033523e+00 -4.13401902e-01
6.43607974e-03 -1.54912516e-01 -5.43967068e-01 8.61749113e-01
8.91718447e-01 -5.81598103e-01 -4.75060523e-01 -2.44408578e-01
-5.25896013e-01 -2.20185276e-02 3.26489985e-01 -1.15276778e+00
2.80456573e-01 2.62966931e-01 6.26153767e-01 -8.47567677e-01
3.54023486e-01 -7.56051600e-01 -2.30177253e-01 7.17175245e-01
-2.28075311e-01 -3.52925211e-01 2.27682784e-01 4.66965109e-01
-4.42691714e-01 -1.26405224e-01 1.17977071e+00 -9.30997953e-02
-4.06856924e-01 6.01365030e-01 -4.31375384e-01 9.65036899e-02
1.04105330e+00 -3.40800256e-01 2.31635217e-02 -1.56370357e-01
-7.65116096e-01 -9.40552577e-02 -4.60726733e-04 -1.74812917e-02
9.34250891e-01 -1.57903063e+00 -6.87646508e-01 4.16574836e-01
3.69956732e-01 3.61686140e-01 3.65529090e-01 1.24835432e+00
-4.71656054e-01 4.89362389e-01 -1.49440587e-01 -8.77048016e-01
-1.17619848e+00 3.07796061e-01 2.88803726e-01 -8.28718066e-01
-9.30789113e-01 8.00201178e-01 5.08165956e-01 -5.49641788e-01
3.06990832e-01 -3.01473260e-01 -2.09255874e-01 2.18097586e-02
4.98442322e-01 1.46016985e-01 3.78190607e-01 -5.37261009e-01
-3.52920532e-01 8.64389658e-01 -5.56259394e-01 3.24104548e-01
1.42496562e+00 3.20336908e-01 2.14953035e-01 1.41350195e-01
1.83168352e+00 -5.19862533e-01 -7.05003977e-01 -5.44136703e-01
-1.18790381e-01 -2.01270059e-01 3.99625063e-01 -8.78291667e-01
-1.30979085e+00 9.54413593e-01 1.23344326e+00 -3.34398836e-01
1.44198334e+00 3.30640823e-01 8.47517550e-01 3.09732795e-01
3.83343846e-01 -7.10751295e-01 4.67539757e-01 2.32670292e-01
1.05182838e+00 -1.25161195e+00 2.36700222e-01 -5.65112352e-01
-6.27677619e-01 1.12124026e+00 6.42624140e-01 -1.51382804e-01
9.48199213e-01 3.01154584e-01 1.05729185e-01 -1.65165722e-01
-4.81859207e-01 -1.01721421e-01 3.03837806e-01 6.41665816e-01
4.07914519e-01 8.82275552e-02 -2.13573158e-01 7.50956774e-01
-8.21187720e-02 6.25719503e-02 3.19487453e-01 9.21145618e-01
-1.30336955e-01 -8.07596982e-01 -3.72372866e-01 8.02859545e-01
-1.61806390e-01 5.64302988e-02 -1.15994878e-01 1.18082285e+00
1.02853306e-01 4.36727822e-01 1.01857528e-01 -4.83986497e-01
4.43279296e-01 -5.20412587e-02 4.12891120e-01 -6.48508549e-01
-3.56660821e-02 2.93838739e-01 -1.38405278e-01 -2.75018811e-01
-3.99370849e-01 -4.81121242e-01 -1.47983444e+00 2.02065989e-01
-3.88830066e-01 -1.04634069e-01 3.39371681e-01 5.34577370e-01
1.90460891e-01 6.79915667e-01 9.01673555e-01 -5.73534489e-01
-5.44524372e-01 -1.08009648e+00 -5.92541516e-01 3.42649937e-01
1.60897642e-01 -1.00405872e+00 -3.71370643e-01 1.80220380e-01] | [14.814016342163086, -2.155107021331787] |
a514713d-fc2b-48c0-b5f6-2d0b369be077 | sign-coded-exposure-sensing-for-noise-robust | 2305.03226 | null | https://arxiv.org/abs/2305.03226v1 | https://arxiv.org/pdf/2305.03226v1.pdf | Sign-Coded Exposure Sensing for Noise-Robust High-Speed Imaging | We present a novel Fourier camera, an in-hardware optical compression of high-speed frames employing pixel-level sign-coded exposure where pixel intensities temporally modulated as positive and negative exposure are combined to yield Hadamard coefficients. The orthogonality of Walsh functions ensures that the noise is not amplified during high-speed frame reconstruction, making it a much more attractive option for coded exposure systems aimed at very high frame rate operation. Frame reconstruction is carried out by a single-pass demosaicking of the spatially multiplexed Walsh functions in a lattice arrangement, significantly reducing the computational complexity. The simulation prototype confirms the improved robustness to noise compared to the binary-coded exposure patterns, such as one-hot encoding and pseudo-random encoding. Our hardware prototype demonstrated the reconstruction of 4kHz frames of a moving scene lit by ambient light only. | ['Keigo Hirakawa', 'Vijayan Asari', 'R. Wes Baldwin'] | 2023-05-05 | null | null | null | null | ['demosaicking'] | ['computer-vision'] | [ 1.15328586e+00 -1.99584618e-01 3.29796672e-01 -7.73235336e-02
-2.56105632e-01 -2.67987370e-01 5.31134963e-01 -6.38429821e-01
-6.99645281e-01 7.78618932e-01 1.64591685e-01 -3.68492663e-01
-1.47835705e-02 -6.58774555e-01 -5.94146252e-01 -9.38417733e-01
-3.44398439e-01 -4.89610583e-01 2.45204791e-01 1.16696626e-01
4.90004301e-01 2.51183331e-01 -1.85279107e+00 4.30013090e-01
5.26019782e-02 1.15646231e+00 4.96085793e-01 1.21777952e+00
1.67161494e-01 8.79133582e-01 -5.02863348e-01 -3.85809422e-01
6.85714006e-01 -7.33159721e-01 -2.28318080e-01 2.77783185e-01
4.04400796e-01 -7.54916072e-01 -6.64451182e-01 9.19541657e-01
8.40349570e-02 -2.07053959e-01 3.54527175e-01 -6.13061845e-01
-2.86461055e-01 1.56693980e-01 -4.41320956e-01 9.53666419e-02
6.41632915e-01 5.29891968e-01 3.29427987e-01 -6.93470657e-01
5.95598161e-01 7.47467518e-01 6.53004110e-01 2.42973700e-01
-1.27569139e+00 -4.92235571e-01 -9.81552243e-01 1.57919183e-01
-1.38251221e+00 -6.26611769e-01 6.48256779e-01 -3.11898962e-02
1.07351494e+00 5.35377502e-01 9.42557871e-01 4.59622234e-01
4.57415700e-01 -2.67458856e-01 1.43638420e+00 -1.02547801e+00
1.88910633e-01 -4.23882715e-02 -2.56518096e-01 5.57031155e-01
2.82260358e-01 3.07244182e-01 -5.78751266e-01 -1.71567604e-01
1.08983302e+00 -7.25463331e-02 -9.32157755e-01 2.49403492e-01
-1.44028604e+00 2.15989053e-01 -1.41674906e-01 3.59788805e-01
-3.34505558e-01 6.88875198e-01 -7.38456026e-02 4.97512370e-01
6.57837167e-02 4.83286053e-01 1.20305464e-01 -2.85826176e-01
-1.11630154e+00 -2.81894207e-01 4.87116933e-01 8.73524904e-01
5.93374074e-01 2.25196972e-01 6.04713969e-02 3.05100292e-01
2.58950174e-01 7.69489050e-01 4.38101500e-01 -1.13512182e+00
2.90572673e-01 -3.21133643e-01 1.92389145e-01 -8.06317329e-01
-1.12200819e-01 -1.92844719e-02 -8.92264724e-01 6.54317498e-01
5.11904180e-01 -2.19151706e-01 -6.26111865e-01 1.21876276e+00
-1.73514992e-01 4.59956139e-01 2.17146918e-01 1.03414237e+00
1.12493470e-01 9.29320633e-01 -5.58961451e-01 -6.79463089e-01
1.69397998e+00 -3.95081192e-01 -1.22280431e+00 2.43272707e-01
5.76748885e-02 -1.25739717e+00 7.20803380e-01 8.37526441e-01
-1.41873133e+00 -6.89336836e-01 -1.41731906e+00 -4.10677642e-02
2.94782758e-01 1.60660207e-01 4.19852525e-01 1.25175560e+00
-1.18737507e+00 5.07301271e-01 -4.88912016e-01 3.02836359e-01
1.99678794e-01 4.05321121e-01 -1.53236941e-01 -4.86656912e-02
-8.60736609e-01 6.18282378e-01 1.73207998e-01 -1.05849989e-01
1.60735995e-02 -5.79783678e-01 -4.87555236e-01 1.26447171e-01
-1.22147009e-01 -3.94157410e-01 9.83773232e-01 -9.29642022e-01
-1.94731236e+00 6.60051286e-01 -3.63517165e-01 -9.04304802e-01
2.62309283e-01 1.24188989e-01 -8.54389131e-01 7.91522205e-01
-5.73662102e-01 3.40168715e-01 1.48766780e+00 -6.44586980e-01
-5.48323750e-01 4.39168423e-01 -3.19479018e-01 -2.14207172e-01
-2.60364026e-01 -2.77655735e-03 -9.10105482e-02 -4.49929416e-01
2.43707985e-01 -5.66261947e-01 -9.40932184e-02 2.35163271e-01
1.50567964e-01 6.46590352e-01 1.21644163e+00 -3.60454440e-01
1.53893125e+00 -2.41915703e+00 -3.70872647e-01 8.15954208e-02
-1.70145273e-01 3.09371859e-01 1.61362410e-01 6.33795500e-01
-1.95985869e-01 -1.86003312e-01 -3.96413147e-01 -2.56793529e-01
-5.23237646e-01 -8.52998495e-02 -4.98577267e-01 8.03169131e-01
3.88656348e-01 2.76348174e-01 -7.13907361e-01 -2.62933135e-01
4.63679433e-01 5.20030916e-01 -5.76586008e-01 -6.57507256e-02
6.16540499e-02 2.62007654e-01 2.04244554e-01 3.78068238e-01
9.39613223e-01 -1.04036229e-02 3.30475777e-01 -3.63591552e-01
-5.99450469e-01 1.96333006e-01 -1.21298575e+00 1.44202757e+00
-4.46711183e-01 1.27692020e+00 -6.59331158e-02 -1.85892045e-01
1.03967261e+00 6.12464666e-01 2.76184767e-01 -8.27720225e-01
4.77155391e-03 3.30594838e-01 -2.27630928e-01 -6.06829047e-01
1.01678991e+00 -4.15535718e-01 1.94413558e-01 5.08562744e-01
-3.10962439e-01 -4.68406081e-01 1.42306730e-01 6.38075024e-02
1.20231211e+00 2.29990613e-02 2.75737226e-01 -3.97101581e-01
5.31319857e-01 -3.16820264e-01 9.36430544e-02 5.14874697e-01
1.62614495e-01 1.07587159e+00 1.36041522e-01 -4.11085844e-01
-1.57799160e+00 -6.74027503e-01 -3.95911068e-01 -3.86594906e-02
3.09334517e-01 -5.18356323e-01 -7.38084912e-01 4.89961863e-01
-4.93585885e-01 4.58836168e-01 1.87651902e-01 1.76683709e-01
-8.91982377e-01 -4.81284946e-01 5.17268836e-01 -2.75866836e-01
8.37349713e-01 -5.81827343e-01 -1.55396128e+00 4.05959845e-01
1.62225351e-01 -1.28277934e+00 -3.11512887e-01 4.69630957e-02
-7.17343569e-01 -1.04880214e+00 -6.63364232e-01 -7.73178577e-01
6.55814946e-01 5.36054850e-01 6.62591279e-01 -2.04279572e-02
-6.04170084e-01 3.92434388e-01 -3.51028174e-01 1.41699120e-01
-3.68158609e-01 -1.05698240e+00 1.35664210e-01 4.06846374e-01
1.61102504e-01 -6.29120350e-01 -9.26767766e-01 2.45716065e-01
-1.06239367e+00 2.65394956e-01 2.47828037e-01 7.87535727e-01
2.69535691e-01 1.77594244e-01 -2.23680392e-01 -3.24678808e-01
3.05604577e-01 1.78524673e-01 -9.07559395e-01 -4.06106040e-02
-1.86584428e-01 -1.05401911e-01 8.17079365e-01 -4.13731337e-01
-1.17007077e+00 2.14671195e-01 1.59797207e-01 -2.21620068e-01
1.66694164e-01 -1.25264704e-01 5.90764225e-01 -5.42668760e-01
7.26808310e-01 3.70695591e-01 1.87931404e-01 3.81539650e-02
1.64889097e-01 7.85359800e-01 1.05510986e+00 -3.71337757e-02
9.01629984e-01 8.44560444e-01 5.17990172e-01 -1.37333024e+00
5.13199449e-01 -2.22535774e-01 -1.25759244e-01 -5.09460747e-01
8.93599749e-01 -9.16102231e-01 -8.97538304e-01 7.79912710e-01
-1.29986596e+00 -1.33773506e-01 -5.35723925e-01 8.30930531e-01
-5.33660591e-01 7.78506875e-01 -5.87291241e-01 -1.08516538e+00
-8.54726285e-02 -9.71481800e-01 7.33308971e-01 2.95741886e-01
-1.63768113e-01 -5.61902881e-01 -2.28759453e-01 -2.61030078e-01
7.11543083e-01 4.78393696e-02 5.77007890e-01 7.56327987e-01
-1.16067708e+00 -2.28805318e-01 -5.32528423e-02 8.21250901e-02
2.37793475e-01 2.97445059e-01 -1.17558098e+00 -1.70306742e-01
6.09042346e-01 1.03869820e-02 7.44894743e-01 3.35175306e-01
1.02166641e+00 -2.01034665e-01 1.07648119e-01 9.01366472e-01
1.81544578e+00 4.62574124e-01 1.48322535e+00 2.27996096e-01
-1.53990835e-01 8.41277167e-02 4.68295068e-01 8.15824032e-01
-5.30269980e-01 6.88558638e-01 1.81739673e-01 -9.37315598e-02
-3.24594826e-01 1.48219377e-01 2.10164070e-01 4.38526630e-01
-1.96605429e-01 -5.87134004e-01 -3.62376153e-01 5.81739917e-02
-1.01792753e+00 -1.48959255e+00 -7.03207076e-01 2.44152164e+00
7.73581147e-01 5.87391667e-02 -3.77592772e-01 6.71938479e-01
8.19465876e-01 3.77441645e-01 4.39445317e-01 -5.83006561e-01
-4.56173480e-01 8.87365282e-01 1.21180606e+00 7.02036083e-01
-8.62165093e-01 2.56142169e-01 6.76064444e+00 8.71396184e-01
-1.26368034e+00 3.26588750e-02 3.73580158e-01 -3.26382697e-01
-4.22604322e-01 1.23717979e-01 -5.81169367e-01 1.05127549e+00
1.02091610e+00 2.31770366e-01 5.10669887e-01 1.01640418e-01
4.92893010e-01 -7.84982741e-01 -5.60610116e-01 1.26451623e+00
-5.74121550e-02 -1.74414718e+00 -3.30941081e-01 1.32249355e-01
6.37962103e-01 -5.82551718e-01 3.37106019e-01 -5.06532013e-01
-6.02228761e-01 -5.43487847e-01 8.41145217e-01 6.68255806e-01
1.49774706e+00 -4.16880816e-01 2.48217121e-01 9.43237171e-02
-1.14258015e+00 -1.64115831e-01 -3.93440753e-01 -5.46087921e-01
7.36067474e-01 8.80720198e-01 -6.85190916e-01 4.37051989e-02
3.35803330e-01 2.52376832e-02 1.17681675e-01 1.00857413e+00
2.97104567e-01 7.43628442e-01 -6.59604371e-01 -2.42095217e-02
1.27915710e-01 -5.92347264e-01 6.26835048e-01 1.25993800e+00
9.63360012e-01 7.65215456e-01 -6.27152741e-01 5.51907301e-01
1.79243743e-01 -5.39190710e-01 -6.20411456e-01 1.59914345e-01
1.49598807e-01 8.65437388e-01 -8.12466204e-01 -2.68264920e-01
-6.42948806e-01 1.21221626e+00 -6.59016073e-01 2.54243225e-01
-7.90378511e-01 -8.88124585e-01 5.89110672e-01 2.21711874e-01
6.19281530e-01 -6.76953614e-01 -3.85829538e-01 -8.34441245e-01
1.73315443e-02 -5.26009500e-01 -3.21051031e-01 -8.89953136e-01
-4.75454360e-01 4.05515671e-01 -3.02745879e-01 -1.83868778e+00
-1.31909370e-01 -5.43023586e-01 -3.58836204e-01 1.06219947e+00
-1.44208074e+00 -2.46521339e-01 -1.30421862e-01 5.16741753e-01
4.11502689e-01 1.20770093e-02 7.78576970e-01 2.90033132e-01
1.46007493e-01 1.23153232e-01 3.82256329e-01 -3.98016930e-01
5.01336396e-01 -6.05841994e-01 3.40109468e-01 1.23325932e+00
-8.81087705e-02 6.01520181e-01 7.36157119e-01 -6.46368802e-01
-1.87899709e+00 -5.50540388e-01 8.68405998e-01 3.78651410e-01
9.71456394e-02 -2.55056173e-01 -6.02441013e-01 -9.00103226e-02
6.33550346e-01 3.08625363e-02 5.69482028e-01 -1.21506643e+00
-3.44814472e-02 -1.65537037e-02 -1.49644661e+00 6.64124608e-01
8.20609748e-01 -6.58689737e-01 -1.31641820e-01 3.00797433e-01
5.50589561e-01 -5.53009927e-01 -3.86652768e-01 1.14242211e-01
7.02356279e-01 -1.49273276e+00 9.16081250e-01 5.57580650e-01
6.46530330e-01 -6.77177787e-01 -3.80055249e-01 -5.33492208e-01
-4.07867253e-01 -1.41197407e+00 1.71902820e-01 7.96829879e-01
5.12904190e-02 -9.15538669e-01 5.82910895e-01 4.09035832e-01
-9.00574103e-02 -1.39463797e-01 -1.12954211e+00 -5.42423368e-01
-1.04768181e+00 -3.82167071e-01 4.46598619e-01 5.66795290e-01
2.01984063e-01 -2.23480597e-01 -9.05519307e-01 2.88767606e-01
7.30247736e-01 -3.13758671e-01 4.12555814e-01 -5.59345186e-01
-7.82506406e-01 -1.44863129e-01 -8.21615994e-01 -1.45456135e+00
-5.50909460e-01 -2.75571585e-01 -7.71955103e-02 -6.63124561e-01
-4.29107934e-01 -1.73942685e-01 3.58599126e-01 -3.94614488e-01
3.17632586e-01 1.01778448e+00 1.69552103e-01 8.91044736e-02
1.98134422e-01 8.77058730e-02 1.01221657e+00 3.10046047e-01
-9.29318741e-03 -1.90716714e-01 1.53422907e-01 3.69868308e-01
4.50138122e-01 -2.29258195e-01 -4.42586660e-01 -4.34865355e-01
2.77839243e-01 5.18919230e-01 4.44700241e-01 -1.64461362e+00
4.45701957e-01 2.16717482e-01 4.35304940e-01 -4.87969577e-01
7.83454239e-01 -1.25025165e+00 9.34463501e-01 8.09440017e-01
-1.10958420e-01 1.29499301e-01 -2.95933392e-02 5.35802126e-01
-1.69706419e-01 -5.59286535e-01 8.13023388e-01 -1.55990154e-01
-8.27116013e-01 -3.26084375e-01 -8.36459696e-01 -5.43658257e-01
1.22976267e+00 -9.71949220e-01 -4.02255714e-01 -5.00983894e-01
-2.09430307e-01 -6.29257500e-01 7.63609767e-01 -5.25450230e-01
8.95743370e-01 -1.07724404e+00 -2.06710175e-01 8.87010932e-01
-6.39011502e-01 -4.36739028e-01 3.72528404e-01 7.40135252e-01
-1.62626493e+00 4.72381532e-01 -2.45751977e-01 -7.55166590e-01
-1.51746416e+00 2.06307277e-01 8.70278552e-02 2.79894620e-01
-8.52932751e-01 9.22580123e-01 -4.26216543e-01 9.72881675e-01
-2.31628448e-01 -3.92837763e-01 2.47231439e-01 -4.14363563e-01
1.02443361e+00 4.62374121e-01 5.13059311e-02 -1.63641915e-01
-4.89689820e-02 8.92176449e-01 5.74038744e-01 -5.99191427e-01
1.04224980e+00 -3.82095844e-01 -1.38822407e-01 -5.33051230e-02
1.28147769e+00 2.85598844e-01 -1.35499120e+00 6.25119805e-02
-4.75452572e-01 -1.14874911e+00 1.12446174e-01 -2.07568944e-01
-6.74241960e-01 8.55676413e-01 6.66097522e-01 3.75946701e-01
1.63381410e+00 -7.56815076e-01 7.87967682e-01 2.93346345e-01
8.38824809e-01 -8.59959781e-01 -1.98032603e-01 1.91768095e-01
2.37025455e-01 -4.38997656e-01 3.47827524e-01 -5.68934798e-01
-1.73102692e-01 1.79636467e+00 -2.85717338e-01 -2.09499523e-01
4.20692086e-01 7.72372663e-01 -2.13957593e-01 2.14976102e-01
-6.28989518e-01 3.72141190e-02 -3.00498337e-01 5.76932788e-01
3.73068154e-01 -2.30247527e-01 -6.22833550e-01 -3.75358433e-01
-9.42925066e-02 2.99524367e-01 1.11818182e+00 8.60838592e-01
-5.11942506e-01 -1.09453762e+00 -8.36884260e-01 1.83389440e-01
-5.34855664e-01 -2.98633128e-01 3.47249627e-01 5.12310266e-01
1.64619088e-01 9.33325768e-01 4.17936414e-01 -3.31601948e-01
6.17050454e-02 9.87198651e-02 7.92985380e-01 -3.46611515e-02
-5.64984739e-01 2.33133197e-01 9.13328081e-02 -5.80178440e-01
-6.36560917e-01 -5.68876445e-01 -1.10233724e+00 -2.58984149e-01
-1.33567736e-01 -1.61003381e-01 8.67461443e-01 3.42189193e-01
9.08328667e-02 2.42109284e-01 9.43238378e-01 -9.93744969e-01
-2.90357977e-01 -2.88587242e-01 -6.32979274e-01 1.18964329e-01
8.69054437e-01 9.60289389e-02 -5.35869777e-01 5.18770337e-01] | [11.24655818939209, -2.2504427433013916] |
08475e7e-9838-4abd-9468-c3addd66ae67 | performance-of-gan-based-augmentation-for | 2304.09067 | null | https://arxiv.org/abs/2304.09067v1 | https://arxiv.org/pdf/2304.09067v1.pdf | Performance of GAN-based augmentation for deep learning COVID-19 image classification | The biggest challenge in the application of deep learning to the medical domain is the availability of training data. Data augmentation is a typical methodology used in machine learning when confronted with a limited data set. In a classical approach image transformations i.e. rotations, cropping and brightness changes are used. In this work, a StyleGAN2-ADA model of Generative Adversarial Networks is trained on the limited COVID-19 chest X-ray image set. After assessing the quality of generated images they are used to increase the training data set improving its balance between classes. We consider the multi-class classification problem of chest X-ray images including the COVID-19 positive class that hasn't been yet thoroughly explored in the literature. Results of transfer learning-based classification of COVID-19 chest X-ray images are presented. The performance of several deep convolutional neural network models is compared. The impact on the detection performance of classical image augmentations i.e. rotations, cropping, and brightness changes are studied. Furthermore, classical image augmentation is compared with GAN-based augmentation. The most accurate model is an EfficientNet-B0 with an accuracy of 90.2 percent, trained on a dataset with a simple class balancing. The GAN augmentation approach is found to be subpar to classical methods for the considered dataset. | ['Rafał Możdżonek', 'Aleksander Ogonowski', 'Konrad Klimaszewski', 'Oleksandr Fedoruk'] | 2023-04-18 | null | null | null | null | ['image-augmentation'] | ['computer-vision'] | [ 6.88715279e-01 3.93038154e-01 4.62450869e-02 -2.98878282e-01
-9.07705843e-01 -2.70858645e-01 6.58316433e-01 2.02536970e-01
-7.86877930e-01 8.28460813e-01 -1.87888563e-01 -4.00060534e-01
5.00521399e-02 -9.59284306e-01 -8.45087588e-01 -8.76873136e-01
1.88954815e-01 7.74784148e-01 -1.66858763e-01 -4.55485523e-01
-2.75277913e-01 6.70103550e-01 -1.22448468e+00 4.35329586e-01
4.60613787e-01 8.80077720e-01 2.01127473e-02 9.03377175e-01
2.47233897e-01 6.23262107e-01 -9.31780457e-01 -2.23495662e-01
3.03744227e-01 -7.90508509e-01 -6.29521072e-01 1.43808067e-01
3.89215678e-01 -3.17379177e-01 -5.84050156e-02 7.87294507e-01
6.87938511e-01 -5.32580577e-02 9.01799142e-01 -1.04581141e+00
-2.26963967e-01 3.59827161e-01 -3.97251397e-01 4.84402895e-01
9.70647857e-02 1.51394442e-01 3.13332886e-01 -6.64195299e-01
7.28179276e-01 8.01352143e-01 5.89757085e-01 6.70172155e-01
-1.16051471e+00 -5.84406257e-01 -5.29286444e-01 5.30572571e-02
-1.18449640e+00 3.02218199e-01 6.96017444e-01 -5.52984059e-01
4.73800659e-01 5.43731213e-01 6.18850529e-01 1.32161140e+00
4.00151879e-01 2.71525800e-01 1.58664060e+00 -6.76748455e-01
2.14477554e-01 2.91192204e-01 -1.65172219e-01 4.87254202e-01
2.43860006e-01 1.82376057e-01 5.60760424e-02 8.95959735e-02
6.80859745e-01 4.18801978e-03 -9.06958878e-02 -1.07662953e-01
-9.25173163e-01 1.06597090e+00 8.25626552e-01 8.14770401e-01
-6.05530798e-01 1.63753495e-01 4.44334447e-01 2.63871104e-01
5.34341693e-01 5.68254411e-01 -1.80116445e-01 2.94265360e-01
-9.19572294e-01 2.53421426e-01 3.26377064e-01 5.14711082e-01
4.02855635e-01 3.01812887e-01 -1.88738614e-01 6.38493359e-01
-7.75404423e-02 4.75895256e-01 7.27062881e-01 -1.31421313e-01
2.78560311e-01 6.25563025e-01 -3.85160267e-01 -7.73968816e-01
-6.94409549e-01 -6.65438771e-01 -1.19501460e+00 7.04215705e-01
4.98030543e-01 -2.59267062e-01 -1.33840346e+00 1.54833627e+00
2.49997512e-01 -1.62281811e-01 2.40811691e-01 6.13255262e-01
8.57628584e-01 4.48868454e-01 8.93714800e-02 -2.60873530e-02
1.55587542e+00 -6.03136480e-01 -6.14754736e-01 -6.38633445e-02
6.04172468e-01 -9.36319888e-01 8.45574558e-01 5.94904542e-01
-1.00398648e+00 -8.08008552e-01 -1.24337435e+00 1.91247612e-01
-5.98646879e-01 4.91399020e-01 2.17235237e-01 1.10059452e+00
-9.01955843e-01 5.64413548e-01 -6.93172038e-01 -3.70282978e-01
4.82519329e-01 5.27853012e-01 -4.60066229e-01 -1.33054659e-01
-1.08699119e+00 1.10912383e+00 6.64371848e-01 -8.63405839e-02
-9.07422841e-01 -6.60860300e-01 -5.92895269e-01 -1.48270622e-01
2.71100134e-01 -6.16903961e-01 1.03359008e+00 -1.28782380e+00
-1.27394319e+00 1.17383993e+00 6.58418894e-01 -8.35461795e-01
1.02823865e+00 5.99395856e-02 -3.97921920e-01 2.81593621e-01
-1.52935386e-01 7.16622412e-01 1.05736768e+00 -1.21541286e+00
-2.38674611e-01 -2.41327956e-01 -1.37511849e-01 -1.81886218e-02
-6.87966645e-02 -1.52185977e-01 -1.21821079e-03 -1.06162465e+00
-2.47657299e-01 -1.15175188e+00 -2.79457480e-01 -3.74968112e-01
-3.84077400e-01 2.83164084e-01 8.80065560e-01 -8.91966164e-01
6.51151061e-01 -1.95694208e+00 -6.23343810e-02 4.86536175e-01
2.77436376e-02 4.63752985e-01 -2.04877947e-02 2.00907961e-01
-7.46127784e-01 9.47763696e-02 -3.71961087e-01 -8.75779912e-02
-5.45331359e-01 3.33280534e-01 1.35010593e-02 5.03155887e-01
4.17677462e-01 8.88398349e-01 -5.38314402e-01 -4.37797606e-01
5.82310736e-01 7.51317143e-01 -4.29258794e-01 4.45216835e-01
-1.84086591e-01 7.37526536e-01 -8.51410925e-02 3.60754222e-01
6.70638084e-01 -2.60402970e-02 -5.23834005e-02 -2.53074795e-01
3.33539426e-01 -3.14829081e-01 -7.86423564e-01 1.43374991e+00
-6.41172707e-01 5.21161616e-01 -4.52403903e-01 -1.10372460e+00
1.09205556e+00 5.39301872e-01 5.23350060e-01 -6.94994211e-01
5.00993729e-01 2.11284742e-01 3.98942888e-01 -3.45245510e-01
3.06730419e-01 -4.05385822e-01 3.68308579e-03 3.30184370e-01
1.98818654e-01 -3.92114729e-01 2.06833616e-01 -1.93809360e-01
1.06242788e+00 1.02083795e-02 4.69206244e-01 -1.71727523e-01
5.86858690e-01 3.81427780e-02 -1.46224052e-01 6.18067741e-01
2.81720042e-01 1.01230562e+00 5.09407163e-01 -4.75431204e-01
-1.28496742e+00 -7.39207923e-01 -2.04649270e-01 5.91469526e-01
-5.38294494e-01 1.23335660e-01 -1.16275203e+00 -8.72470558e-01
-4.45790172e-01 6.79828882e-01 -1.15684700e+00 -2.98799217e-01
-7.46204019e-01 -1.26837063e+00 6.22532070e-01 4.37671661e-01
5.02876103e-01 -1.25924945e+00 -9.23524141e-01 1.00651808e-01
3.28488387e-02 -1.04148805e+00 1.66801661e-01 3.51560682e-01
-9.59680855e-01 -1.26043963e+00 -1.04884696e+00 -4.79834586e-01
9.19642329e-01 -3.92891258e-01 1.18695414e+00 2.68943936e-01
-8.03427637e-01 3.78050864e-01 -6.02766156e-01 -7.05929339e-01
-1.22788167e+00 2.16038704e-01 -3.52762133e-01 3.05343401e-02
1.88374072e-02 -2.63792992e-01 -5.79107940e-01 2.82556862e-01
-1.29506910e+00 2.50360537e-02 9.92928207e-01 1.08578527e+00
6.16082847e-01 -2.11151510e-01 4.04256701e-01 -1.25083745e+00
4.45895523e-01 -3.62648219e-01 -4.08409387e-01 -3.76160145e-02
-5.36260307e-01 -1.90801620e-01 6.33396447e-01 -4.41300124e-01
-8.19701195e-01 2.06575155e-01 -5.95959067e-01 -3.57920319e-01
-5.19812047e-01 2.19778582e-01 1.88436881e-01 -3.15743387e-01
1.18448114e+00 -2.68826075e-02 3.25379409e-02 -2.53269911e-01
7.46098310e-02 2.75025606e-01 6.39948010e-01 -1.21647604e-01
8.06227088e-01 3.37539822e-01 3.90114337e-01 -1.02591026e+00
-4.08200622e-01 -1.68033779e-01 -6.74853027e-01 -3.19349438e-01
1.27985477e+00 -6.00801528e-01 -7.35722482e-02 4.99266893e-01
-9.37281370e-01 -4.53912377e-01 -7.46994734e-01 5.49933434e-01
-6.14194930e-01 3.23740020e-02 -4.07571942e-01 -4.75994796e-01
-5.59002340e-01 -1.21399009e+00 8.55214536e-01 3.66676413e-02
-1.36388808e-01 -9.00638044e-01 1.39363915e-01 4.10202265e-01
4.03075874e-01 9.42310870e-01 1.02608919e+00 -9.82273459e-01
-1.69591635e-01 -6.22935653e-01 9.43869278e-02 7.33026206e-01
1.83083475e-01 -1.67295635e-01 -1.17446053e+00 -3.72106612e-01
2.02901930e-01 -3.32355499e-01 7.01125264e-01 3.77304256e-01
1.34020722e+00 -2.21450552e-02 -2.85320822e-02 4.55096632e-01
1.51892781e+00 4.71410811e-01 8.86957586e-01 3.61436248e-01
5.03823757e-01 2.81137228e-01 7.56582677e-01 8.58768150e-02
-4.96024311e-01 6.51327431e-01 8.39361966e-01 -7.68721461e-01
-3.76922011e-01 1.27359748e-01 -2.79082954e-01 1.54880077e-01
-3.47959429e-01 -4.30981666e-01 -9.67702627e-01 2.50003278e-01
-1.00519156e+00 -6.02359176e-01 -1.80957645e-01 2.09311700e+00
4.89604443e-01 3.40468168e-01 1.42662629e-01 7.22814083e-01
6.38409734e-01 -1.20698340e-01 -1.53040722e-01 -3.98670405e-01
-7.30253160e-02 9.20159578e-01 5.32294035e-01 1.97441608e-01
-1.24313056e+00 2.43098319e-01 5.72510147e+00 6.66055501e-01
-1.33391893e+00 3.81001651e-01 1.04760766e+00 1.96286559e-01
1.96365267e-01 -5.27444541e-01 -3.58944684e-01 4.97841775e-01
9.08048809e-01 5.26469648e-01 -5.38732596e-02 8.96142006e-01
-1.71501040e-01 -2.64701486e-01 -9.46420431e-01 9.55748737e-01
4.45215940e-01 -1.14126754e+00 1.15134427e-02 1.15864463e-01
7.41582513e-01 -1.13361597e-01 2.87879407e-01 2.32753515e-01
-1.81083903e-01 -1.29727137e+00 2.73116887e-01 3.86401176e-01
1.13288665e+00 -8.22759211e-01 1.16396105e+00 6.81470931e-02
-5.53359807e-01 9.27997157e-02 1.01192575e-02 4.82940853e-01
3.06981169e-02 2.52383590e-01 -1.30657947e+00 6.02337003e-01
5.21177292e-01 4.74894755e-02 -9.18139398e-01 7.21975327e-01
-2.38591582e-01 6.43270075e-01 -3.30777466e-01 1.33724451e-01
3.37687939e-01 -6.62306398e-02 3.78447801e-01 1.25809777e+00
4.90159959e-01 -1.00650983e-02 -2.90408105e-01 6.85935080e-01
-1.53679758e-01 3.50691378e-01 -6.49318457e-01 2.34960958e-01
-2.98012793e-01 1.38200164e+00 -9.73544300e-01 -5.62749088e-01
-4.92846221e-02 1.02676308e+00 -3.82048547e-01 -5.45221716e-02
-8.78738940e-01 -8.61844495e-02 -1.36035845e-01 4.86886531e-01
1.82520926e-01 4.28990424e-01 -1.19931787e-01 -5.99084318e-01
-1.88641310e-01 -1.10739732e+00 5.41576326e-01 -9.59006727e-01
-8.17073882e-01 1.11690927e+00 2.34439746e-01 -1.23541236e+00
-5.69402516e-01 -8.22444737e-01 -6.40421271e-01 8.98173749e-01
-1.21940732e+00 -1.37012041e+00 -7.56255388e-01 5.91552079e-01
5.61918795e-01 -5.61439276e-01 1.07063448e+00 4.23790336e-01
-1.86353773e-01 6.25986755e-01 -6.87398240e-02 1.48027018e-01
4.89913255e-01 -1.44292641e+00 2.16932073e-01 8.79559577e-01
3.83726619e-02 -9.92992967e-02 7.97372341e-01 -5.18159568e-01
-7.13595569e-01 -1.21726704e+00 2.57046789e-01 -1.39315784e-01
1.66089222e-01 -2.25172624e-01 -9.75307286e-01 5.27434587e-01
4.27993387e-01 1.68390706e-01 5.84292591e-01 -6.47748649e-01
5.74227571e-02 -6.02674931e-02 -1.41902769e+00 1.82881430e-01
1.54347017e-01 -1.85418487e-01 -4.24278289e-01 4.84900028e-01
1.79223865e-01 -7.54628122e-01 -9.76824820e-01 5.11277497e-01
1.31546646e-01 -9.10876334e-01 1.11805117e+00 -5.56579709e-01
6.18840754e-01 -3.45595181e-03 6.36608303e-02 -1.46931267e+00
2.55623579e-01 -2.48314902e-01 3.90904695e-01 8.04634571e-01
2.85498798e-01 -4.57387984e-01 9.22873735e-01 1.69630367e-02
1.30827263e-01 -6.43253863e-01 -8.98157895e-01 -4.06588286e-01
2.00989440e-01 -1.71861351e-01 3.08605790e-01 8.87013555e-01
-8.01242352e-01 2.37264588e-01 -3.52124006e-01 -1.51876435e-01
2.37265512e-01 -3.57140809e-01 9.98154223e-01 -9.81575549e-01
-5.39108813e-01 -1.74314246e-01 -8.02709401e-01 9.56896916e-02
-3.36128235e-01 -8.29011261e-01 -3.39709699e-01 -1.17209530e+00
-1.34531900e-01 -5.68293333e-01 -2.03778893e-01 4.47719246e-01
-5.61976656e-02 9.31861520e-01 3.21444780e-01 -8.82819518e-02
3.48574907e-01 -2.22940072e-02 1.14632130e+00 -1.93511635e-01
-9.60620046e-02 2.96377987e-01 -1.26137495e-01 6.63764954e-01
1.16700649e+00 -5.77975452e-01 -4.09036458e-01 1.02012686e-01
1.85640439e-01 1.49570495e-01 6.09119892e-01 -1.36462402e+00
-3.76866460e-01 4.04080808e-01 7.14658916e-01 -6.43029630e-01
4.49276596e-01 -1.12006915e+00 4.07089621e-01 1.12768626e+00
-3.02788377e-01 4.54635054e-01 3.29612374e-01 2.83261120e-01
-2.76209384e-01 -5.83234072e-01 1.18703699e+00 -3.06504667e-01
-3.70654732e-01 6.78801313e-02 -3.33911031e-01 -8.61847475e-02
1.33293295e+00 -1.36215657e-01 4.25062999e-02 -3.26866895e-01
-1.02531695e+00 -5.97719848e-01 2.50822723e-01 2.80319721e-01
5.34656584e-01 -1.15255189e+00 -9.59703207e-01 3.73269588e-01
1.76050887e-01 -1.62542202e-02 3.52684259e-01 8.92339885e-01
-9.44654047e-01 2.84432829e-03 -7.80417740e-01 -7.97462881e-01
-1.71020639e+00 8.30488563e-01 6.41920209e-01 -6.51879072e-01
-4.60014164e-01 4.43500400e-01 1.73172608e-01 -1.14064008e-01
-1.87881321e-01 -4.78979111e-01 -4.60875511e-01 1.14346035e-01
2.08016321e-01 2.57220864e-01 6.89732790e-01 -5.25212824e-01
2.24823784e-02 4.92764443e-01 -1.65963396e-02 -6.06337488e-02
1.30969918e+00 5.58775723e-01 1.41433522e-01 2.11935744e-01
1.21024716e+00 -2.13130087e-01 -5.47948062e-01 7.97498822e-02
-3.63429993e-01 -2.02421769e-01 -2.13914011e-02 -9.64669287e-01
-1.22668993e+00 1.04889297e+00 1.42071414e+00 1.70193836e-01
1.43943071e+00 -2.73573697e-01 1.84964970e-01 1.27479956e-01
-1.82088450e-01 -8.74540448e-01 3.52121830e-01 -1.35799214e-01
1.13176429e+00 -1.28290939e+00 1.98503792e-01 -3.39782149e-01
-6.63496137e-01 1.22279167e+00 5.06009638e-01 -3.13272029e-01
3.01274836e-01 2.98791707e-01 2.82751292e-01 -2.30412796e-01
-1.08983994e-01 -8.32236186e-02 2.27623954e-01 6.20562673e-01
4.64964271e-01 9.78400707e-02 -5.20534575e-01 5.73098548e-02
-2.01490268e-01 6.98878765e-02 5.87531447e-01 8.18960190e-01
1.62983760e-01 -1.26411617e+00 -8.25705171e-01 5.85888147e-01
-9.38559830e-01 6.49739578e-02 -2.55984664e-01 1.46863544e+00
4.04154658e-01 4.75831151e-01 -1.87546555e-02 -2.61211962e-01
4.76811290e-01 2.26708408e-03 5.37740529e-01 -5.04590511e-01
-1.08501017e+00 2.64590811e-02 -1.05162315e-01 -1.49418861e-01
-5.46435177e-01 -4.59183902e-01 -8.03392291e-01 9.95996222e-02
-2.22484767e-01 -4.24189568e-02 1.06857991e+00 7.36142159e-01
-3.02374631e-01 1.17074597e+00 4.73943383e-01 -7.34295905e-01
-5.58061361e-01 -1.19844210e+00 -3.01106632e-01 7.54307330e-01
2.63002902e-01 -5.68674326e-01 -9.00091976e-02 1.96693838e-01] | [14.22390365600586, -1.983370065689087] |
b7eca03c-bdda-4ef8-8256-433fff86e321 | end-to-end-recovery-of-human-shape-and-pose | 1712.06584 | null | http://arxiv.org/abs/1712.06584v2 | http://arxiv.org/pdf/1712.06584v2.pdf | End-to-end Recovery of Human Shape and Pose | We describe Human Mesh Recovery (HMR), an end-to-end framework for
reconstructing a full 3D mesh of a human body from a single RGB image. In
contrast to most current methods that compute 2D or 3D joint locations, we
produce a richer and more useful mesh representation that is parameterized by
shape and 3D joint angles. The main objective is to minimize the reprojection
loss of keypoints, which allow our model to be trained using images in-the-wild
that only have ground truth 2D annotations. However, the reprojection loss
alone leaves the model highly under constrained. In this work we address this
problem by introducing an adversary trained to tell whether a human body
parameter is real or not using a large database of 3D human meshes. We show
that HMR can be trained with and without using any paired 2D-to-3D supervision.
We do not rely on intermediate 2D keypoint detections and infer 3D pose and
shape parameters directly from image pixels. Our model runs in real-time given
a bounding box containing the person. We demonstrate our approach on various
images in-the-wild and out-perform previous optimization based methods that
output 3D meshes and show competitive results on tasks such as 3D joint
location estimation and part segmentation. | ['Michael J. Black', 'David W. Jacobs', 'Angjoo Kanazawa', 'Jitendra Malik'] | 2017-12-18 | end-to-end-recovery-of-human-shape-and-pose-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Kanazawa_End-to-End_Recovery_of_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Kanazawa_End-to-End_Recovery_of_CVPR_2018_paper.pdf | cvpr-2018-6 | ['monocular-3d-human-pose-estimation', '3d-multi-person-pose-estimation', 'weakly-supervised-3d-human-pose-estimation', 'human-mesh-recovery'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 8.48307088e-02 5.07770300e-01 1.78600401e-01 -2.69166619e-01
-1.14604783e+00 -6.51629448e-01 1.69108883e-01 -1.08365744e-01
-6.70530796e-01 2.13229850e-01 -2.76790351e-01 1.93167061e-01
6.44776225e-01 -4.79807973e-01 -1.25325692e+00 -2.14591488e-01
8.42429772e-02 1.14722764e+00 3.78150791e-01 -2.03163281e-01
-2.04019889e-01 6.30852520e-01 -1.41617680e+00 -1.46686539e-01
2.45365486e-01 8.18655074e-01 -3.51036757e-01 1.04309571e+00
5.66613138e-01 8.12410936e-03 -4.42678273e-01 -6.06818259e-01
8.82834971e-01 -3.08166385e-01 -8.70034635e-01 6.15070879e-01
1.03864241e+00 -5.31797230e-01 -1.90249056e-01 9.72840548e-01
6.78511262e-01 -5.53048868e-03 5.09500563e-01 -1.21764493e+00
5.47265727e-03 -1.44585639e-01 -8.23634863e-01 -5.04501045e-01
8.69995236e-01 3.96335006e-01 6.19351268e-01 -8.90019000e-01
1.09419084e+00 1.46620750e+00 1.11512518e+00 8.21635187e-01
-1.57188797e+00 -4.00908947e-01 -2.14538127e-02 -6.38595521e-01
-1.53087103e+00 -2.83519983e-01 8.65755141e-01 -6.46645665e-01
5.84403753e-01 2.34304681e-01 1.07606554e+00 1.05486131e+00
-1.75277308e-01 7.24179506e-01 1.09256577e+00 -4.57139999e-01
-2.73935823e-03 -2.34524816e-01 -2.73658037e-01 1.23964095e+00
-2.34776605e-02 1.57181904e-01 -4.65173364e-01 -3.69428992e-01
1.15896916e+00 -5.25806621e-02 -1.56224981e-01 -9.16969717e-01
-1.33437097e+00 5.30907571e-01 5.19499183e-01 -6.16023600e-01
-2.61275023e-01 5.80952942e-01 1.28528401e-01 6.92035910e-03
3.69971693e-01 8.40169117e-02 -4.90110844e-01 1.57095920e-02
-8.55937123e-01 6.75971985e-01 7.42081344e-01 9.70070660e-01
8.54872882e-01 -4.08268243e-01 3.88285309e-01 4.18850362e-01
3.92580390e-01 7.90081024e-01 -3.03267866e-01 -1.60726190e+00
3.10386956e-01 3.73869151e-01 3.24202269e-01 -7.75973856e-01
-2.99093485e-01 1.93226233e-01 -4.95661169e-01 7.14228868e-01
7.01842427e-01 -1.63273230e-01 -1.43293405e+00 1.56068182e+00
1.08782136e+00 7.58559480e-02 -3.21221560e-01 1.38273692e+00
4.71396506e-01 1.37319759e-01 -2.84624428e-01 4.35540736e-01
1.44681549e+00 -8.97369027e-01 -2.00893298e-01 -3.70731801e-01
2.96588838e-01 -6.71989679e-01 8.89950335e-01 5.34404159e-01
-1.38779604e+00 -4.34833854e-01 -8.45284939e-01 -5.58268368e-01
-1.36974201e-01 -1.66004568e-01 3.17248821e-01 5.29548407e-01
-9.12080407e-01 7.42091954e-01 -1.31238008e+00 -3.67366761e-01
5.70822716e-01 6.44028366e-01 -8.25500131e-01 2.45205741e-02
-6.47127688e-01 7.10954189e-01 3.68949547e-02 1.56134188e-01
-8.74560952e-01 -6.07947409e-01 -1.18725872e+00 -7.22237468e-01
5.10193288e-01 -1.19905245e+00 1.28154790e+00 -6.74670339e-01
-1.43728340e+00 1.52637088e+00 -6.46108165e-02 -2.76573956e-01
1.25112164e+00 -6.14086866e-01 4.40625131e-01 3.79289061e-01
1.28898561e-01 9.61486399e-01 1.00300455e+00 -1.53554463e+00
-2.65112460e-01 -6.94487214e-01 -2.59504281e-03 2.41807014e-01
6.22569561e-01 -1.96017712e-01 -1.08319461e+00 -6.84628010e-01
3.85386169e-01 -1.39728296e+00 -4.76756781e-01 9.28705931e-01
-8.74277651e-01 1.83820248e-01 6.25464916e-01 -8.37787390e-01
2.16538072e-01 -1.85122502e+00 4.06272948e-01 3.94299328e-01
1.11364998e-01 -1.48856238e-01 2.01672569e-01 -8.38017538e-02
2.70338923e-01 4.85188775e-02 -5.50231457e-01 -1.02202749e+00
1.41386136e-01 4.32955265e-01 -4.92147096e-02 1.10308337e+00
2.59627700e-01 9.12369907e-01 -7.67549455e-01 -7.92739689e-01
3.67738396e-01 7.55084097e-01 -7.37484276e-01 3.87844563e-01
-3.64201516e-01 7.36520886e-01 -2.25325659e-01 8.69203866e-01
7.24116087e-01 -1.71390548e-01 -6.93106651e-02 -3.29241276e-01
3.17131072e-01 -8.32588896e-02 -1.44451725e+00 2.35896444e+00
-1.03025869e-01 1.43673373e-02 4.21988189e-01 -4.50653613e-01
3.62070233e-01 3.03115427e-01 4.49799716e-01 -8.22072700e-02
2.68511087e-01 1.50252104e-01 -4.97852594e-01 -2.62190491e-01
2.32114792e-01 -2.63653159e-01 -3.20713520e-01 3.99813175e-01
-5.26469760e-02 -5.97234845e-01 -4.34450120e-01 6.84533790e-02
9.85752761e-01 7.79233634e-01 2.36987006e-02 1.85639873e-01
-1.62015468e-01 9.81639400e-02 4.72871065e-01 4.49284256e-01
-1.28903970e-01 1.24505341e+00 2.60302603e-01 -4.30932850e-01
-1.25397706e+00 -1.29413605e+00 2.36039236e-02 6.41357541e-01
4.99621749e-01 -2.64388412e-01 -8.25408340e-01 -9.09145772e-01
3.51354390e-01 8.26769397e-02 -8.07158887e-01 3.61692548e-01
-8.37288082e-01 6.26732688e-03 7.72751033e-01 4.89487678e-01
2.82804102e-01 -6.32289469e-01 -1.08728969e+00 -1.37604877e-01
-1.66141719e-01 -1.30458260e+00 -8.61339450e-01 6.80041313e-02
-8.90406907e-01 -1.32821643e+00 -8.68929088e-01 -6.33663893e-01
1.05985355e+00 -2.73307711e-01 1.17645705e+00 1.72141597e-01
-7.58496940e-01 7.12016046e-01 -6.79328293e-02 -2.04472199e-01
-2.61446714e-01 -3.15854460e-01 6.80911541e-02 -2.30624139e-01
-2.03718200e-01 -5.61409891e-01 -7.19500422e-01 4.78867888e-01
-5.23913443e-01 8.25958550e-02 1.73195019e-01 6.16249979e-01
1.25678444e+00 -2.68279254e-01 -3.40409100e-01 -9.10450339e-01
-2.69289404e-01 8.27203542e-02 -6.10499263e-01 -7.11251330e-03
-9.89099517e-02 7.52879456e-02 -1.56399254e-02 -5.70477426e-01
-5.05847633e-01 8.67714763e-01 -2.58423567e-01 -1.02571845e+00
-3.51964295e-01 -2.86742032e-01 -3.07808310e-01 -2.34635860e-01
5.64375937e-01 -2.52133399e-01 2.91008204e-01 -7.14183927e-01
5.44352829e-01 1.48789346e-01 9.59987700e-01 -8.01829576e-01
1.09095502e+00 9.00679171e-01 2.45348528e-01 -5.24373591e-01
-6.70185864e-01 -4.81902361e-01 -1.06417382e+00 -2.15462893e-01
1.08828628e+00 -1.15112281e+00 -8.65178287e-01 5.40267587e-01
-1.24361849e+00 -5.43573022e-01 -4.83110160e-01 2.26065710e-01
-8.19593549e-01 5.16474366e-01 -6.18122399e-01 -8.50728869e-01
-2.43156955e-01 -1.25123656e+00 1.97900808e+00 -3.43896806e-01
-3.61840665e-01 -5.83446264e-01 9.18325502e-03 5.34357548e-01
-3.57126147e-01 1.06762838e+00 2.84272879e-01 -1.09010562e-01
-5.41252077e-01 -3.33754957e-01 3.18986237e-01 2.30467260e-01
-1.92761853e-01 -2.56305903e-01 -9.61833477e-01 -2.95768946e-01
-2.65041530e-01 -6.55219316e-01 5.92632771e-01 1.11142769e-01
9.44382489e-01 -3.16597283e-01 -3.55586350e-01 8.86905730e-01
1.20832717e+00 -7.61621416e-01 4.33140367e-01 3.88206542e-02
1.12798607e+00 6.39893651e-01 7.46794820e-01 3.35192800e-01
5.22300959e-01 8.14182103e-01 6.63399756e-01 -3.63771588e-01
-2.95085460e-01 -6.61251545e-01 2.02952534e-01 2.32477516e-01
-2.02652529e-01 1.70758009e-01 -8.79748702e-01 5.01199663e-01
-1.74925995e+00 -4.53605324e-01 -4.29225191e-02 2.35137606e+00
1.06732488e+00 1.34844586e-01 4.95848894e-01 -8.66262689e-02
6.11003757e-01 -1.53792903e-01 -7.87975311e-01 1.82828695e-01
2.38734975e-01 3.83812755e-01 7.57845700e-01 8.15706372e-01
-1.24664831e+00 1.00894153e+00 5.88192272e+00 2.81220794e-01
-7.62005687e-01 8.61062557e-02 3.28736275e-01 -3.12037319e-01
-8.46724957e-02 4.41152491e-02 -4.96445298e-01 1.19818926e-01
1.59968436e-01 7.24461257e-01 3.30050230e-01 8.89665842e-01
-1.97590798e-01 -1.05107665e-01 -1.42168343e+00 1.36977553e+00
1.06389530e-01 -8.82755756e-01 -2.21048236e-01 4.11735177e-01
6.25471890e-01 2.17735413e-02 -1.43385068e-01 -2.63521582e-01
4.24786389e-01 -9.82214451e-01 1.30342603e+00 5.18482208e-01
1.09736693e+00 -6.70857251e-01 3.24360818e-01 4.75521505e-01
-1.04919648e+00 6.61817431e-01 -6.97397217e-02 1.47751033e-01
3.48139346e-01 4.05105561e-01 -8.50862145e-01 3.53132397e-01
9.17225659e-01 3.61692697e-01 -4.04426694e-01 8.74286532e-01
-4.22949135e-01 1.84067845e-01 -9.74262595e-01 5.88770807e-01
-3.82419050e-01 -1.93654280e-02 7.37598181e-01 9.14646804e-01
1.37736842e-01 1.75563723e-01 5.76328218e-01 8.41470301e-01
-3.74988645e-01 -2.71162212e-01 -5.22303820e-01 3.54221672e-01
2.98192590e-01 9.54640031e-01 -9.55518723e-01 -1.87424913e-01
1.54782385e-01 1.68833113e+00 1.94160610e-01 1.27594128e-01
-8.10494840e-01 -1.03363633e-01 6.82957351e-01 5.83378136e-01
5.04080236e-01 -4.16101575e-01 -2.55014688e-01 -1.24133658e+00
3.48598629e-01 -8.24307621e-01 2.06592783e-01 -8.01102996e-01
-1.21737421e+00 2.91294128e-01 8.97671431e-02 -9.31447566e-01
-3.30271751e-01 -5.31268239e-01 -5.22290319e-02 7.27791548e-01
-9.86434281e-01 -1.56621659e+00 -2.19313964e-01 7.19943345e-01
2.61122137e-01 6.07232511e-01 7.70377934e-01 8.68965462e-02
-1.36331409e-01 7.12520599e-01 -6.41462326e-01 5.45418441e-01
7.94149160e-01 -1.54768324e+00 7.40733325e-01 7.59112954e-01
1.31727874e-01 4.18347746e-01 8.26316535e-01 -8.79691660e-01
-1.62299538e+00 -9.86125171e-01 4.09603804e-01 -1.05529761e+00
-2.77784709e-02 -8.89798641e-01 -4.68535066e-01 9.90814686e-01
-4.74362493e-01 5.05124867e-01 2.27598175e-01 -5.93087189e-02
-5.68899930e-01 2.86570877e-01 -1.60129893e+00 5.70831656e-01
1.41644728e+00 -5.76379597e-01 -5.26653528e-01 3.66932362e-01
7.50623822e-01 -1.12763739e+00 -1.02073240e+00 4.48795408e-01
7.20577836e-01 -7.26294339e-01 1.35154796e+00 -4.67976242e-01
5.17667346e-02 -6.50382936e-01 -2.86325067e-01 -8.35978329e-01
4.41654742e-01 -9.37890470e-01 -1.90817297e-01 7.56901026e-01
8.43851082e-03 -1.86628044e-01 1.13956618e+00 1.11001968e+00
2.09411040e-01 -6.80396736e-01 -1.18589735e+00 -5.90298831e-01
-1.30116567e-01 -6.39258265e-01 3.61672640e-01 6.18677318e-01
-5.41128695e-01 9.21867862e-02 -4.64526355e-01 6.24078155e-01
1.09290397e+00 1.24604210e-01 1.39444029e+00 -1.11340058e+00
-4.23594713e-01 1.52560562e-01 -7.97457755e-01 -1.22386920e+00
1.02572940e-01 -6.97515428e-01 4.57437515e-01 -1.17415559e+00
-3.73867899e-02 -3.95183623e-01 4.35249329e-01 7.97992945e-01
-8.86949524e-02 8.66560280e-01 2.51551032e-01 1.63320810e-01
-5.91824412e-01 2.72484869e-01 1.13766599e+00 -5.27599901e-02
-5.14849015e-02 -7.87240639e-03 -1.01809070e-01 1.15023756e+00
2.63962239e-01 -6.95519209e-01 -4.76147719e-02 -4.69622135e-01
6.37762249e-02 5.94005287e-02 1.15725219e+00 -8.75574052e-01
7.15197325e-02 6.99055865e-02 6.72627091e-01 -6.60104036e-01
8.62943709e-01 -9.38184202e-01 4.19876516e-01 3.94645184e-01
-1.07268989e-01 -6.63897917e-02 -1.98907275e-02 7.08868861e-01
4.88770664e-01 1.05223686e-01 8.49085629e-01 -4.35578316e-01
-2.75735974e-01 5.79410791e-01 3.00974399e-01 2.45780066e-01
9.46822226e-01 -3.89818460e-01 4.41039264e-01 -1.20027423e-01
-1.16572094e+00 1.26034170e-01 1.29834056e+00 2.13883698e-01
8.80813181e-01 -1.19951785e+00 -5.86839020e-01 2.33737230e-01
-6.45177066e-02 8.77921045e-01 6.64621443e-02 5.96008301e-01
-9.29854453e-01 -4.27377701e-01 1.83490992e-01 -9.60006595e-01
-1.51085246e+00 4.48876381e-01 7.14739859e-01 5.92762567e-02
-1.05980396e+00 9.72942710e-01 8.91191438e-02 -9.04107153e-01
4.72927153e-01 -2.72008300e-01 8.47920716e-01 -3.75175983e-01
1.94588557e-01 2.93521285e-01 -5.24424948e-04 -9.14750993e-01
-4.92375523e-01 1.06550610e+00 2.54693508e-01 -6.15557194e-01
1.10549819e+00 -4.59914729e-02 -7.85300136e-02 3.86958212e-01
1.35429549e+00 2.84067094e-01 -1.60839045e+00 -3.51383612e-02
-5.72220206e-01 -7.19017148e-01 -2.79088050e-01 -8.06584597e-01
-9.95784581e-01 8.63442957e-01 7.60465801e-01 -4.01966095e-01
6.15356565e-01 3.81596535e-01 1.13979483e+00 2.18154490e-01
7.62660742e-01 -1.04230583e+00 3.38911712e-02 1.14340372e-01
7.92051792e-01 -1.22543991e+00 2.79127568e-01 -5.77099025e-01
-3.76526684e-01 8.75791371e-01 4.81419683e-01 -4.66783047e-01
7.46022880e-01 3.33467185e-01 2.26526424e-01 -4.15604413e-01
-1.77874435e-02 -2.58262545e-01 3.12068194e-01 7.48202622e-01
-2.01531589e-01 3.13869193e-02 3.79041910e-01 1.27798453e-01
-4.28686738e-01 -9.15599093e-02 1.30861431e-01 1.19408834e+00
-7.93660209e-02 -1.22674310e+00 -8.28988016e-01 2.72608139e-02
-5.75686634e-01 3.67843091e-01 -5.26980817e-01 9.11876440e-01
4.27746534e-01 5.62554717e-01 -4.53229211e-02 -4.03360754e-01
5.87697744e-01 3.71377654e-02 9.62975442e-01 -6.16462827e-01
-4.54764634e-01 2.46996760e-01 4.37039398e-02 -9.41954970e-01
-4.45769459e-01 -6.22151613e-01 -1.60909104e+00 -2.04257816e-01
-2.70325810e-01 -2.29712620e-01 6.83573067e-01 7.82107294e-01
3.65382999e-01 5.37820086e-02 2.06565976e-01 -1.52449977e+00
-4.79245901e-01 -4.77901042e-01 -4.77238357e-01 7.25649059e-01
6.86745107e-01 -6.71388686e-01 -2.31678963e-01 4.36575055e-01] | [7.082042217254639, -1.1803479194641113] |
9fb4dc83-7c34-43aa-8d90-2210c58127c6 | temporal-recurrent-networks-for-online-action | 1811.07391 | null | http://arxiv.org/abs/1811.07391v2 | http://arxiv.org/pdf/1811.07391v2.pdf | Temporal Recurrent Networks for Online Action Detection | Most work on temporal action detection is formulated as an offline problem,
in which the start and end times of actions are determined after the entire
video is fully observed. However, important real-time applications including
surveillance and driver assistance systems require identifying actions as soon
as each video frame arrives, based only on current and historical observations.
In this paper, we propose a novel framework, Temporal Recurrent Network (TRN),
to model greater temporal context of a video frame by simultaneously performing
online action detection and anticipation of the immediate future. At each
moment in time, our approach makes use of both accumulated historical evidence
and predicted future information to better recognize the action that is
currently occurring, and integrates both of these into a unified end-to-end
architecture. We evaluate our approach on two popular online action detection
datasets, HDD and TVSeries, as well as another widely used dataset, THUMOS'14.
The results show that TRN significantly outperforms the state-of-the-art. | ['Yi-Ting Chen', 'Mingfei Gao', 'Mingze Xu', 'Larry S. Davis', 'David J. Crandall'] | 2018-11-18 | temporal-recurrent-networks-for-online-action-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Xu_Temporal_Recurrent_Networks_for_Online_Action_Detection_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Xu_Temporal_Recurrent_Networks_for_Online_Action_Detection_ICCV_2019_paper.pdf | iccv-2019-10 | ['online-action-detection'] | ['computer-vision'] | [ 4.42808092e-01 -2.40635604e-01 -5.28687298e-01 -3.14758241e-01
-4.07276094e-01 -2.92855740e-01 6.58538043e-01 -8.28115121e-02
-6.75686121e-01 4.22940999e-01 5.58279514e-01 -2.33625516e-01
7.47325346e-02 -4.20377851e-01 -2.93086231e-01 -4.45220023e-01
-1.53447419e-01 -8.70436728e-02 7.84425914e-01 2.84010824e-02
1.06217511e-01 5.85588634e-01 -1.70148504e+00 4.64970857e-01
2.34281614e-01 1.12071562e+00 2.37625882e-01 8.59897614e-01
2.08613873e-01 1.70552409e+00 -2.95321167e-01 -1.76896807e-02
3.12207162e-01 -5.21029651e-01 -5.97400308e-01 4.73946571e-01
8.44617113e-02 -8.77529144e-01 -1.06172919e+00 6.60968125e-01
2.94897765e-01 3.88865948e-01 -4.35160147e-03 -1.24099052e+00
-1.30037189e-01 2.98234135e-01 -4.17969823e-01 6.83497548e-01
5.13268411e-01 4.38715786e-01 7.60873973e-01 -5.11693597e-01
8.54922831e-01 9.41204309e-01 3.18047702e-01 5.59794962e-01
-5.76080859e-01 -3.47880155e-01 5.76971531e-01 6.81851029e-01
-1.05954587e+00 -7.40390718e-01 7.71258891e-01 -3.29569489e-01
1.22624278e+00 -6.01209328e-02 7.02596128e-01 8.36966276e-01
2.60440886e-01 1.28660059e+00 6.35721862e-01 -3.92455876e-01
1.04655951e-01 -4.52520490e-01 -1.90277191e-04 3.77512902e-01
-5.48346281e-01 2.87184596e-01 -6.39144063e-01 4.22082841e-02
5.67212045e-01 4.16645885e-01 -1.41706616e-01 3.86658162e-02
-1.32791150e+00 2.73154974e-01 -9.78441313e-02 3.21022898e-01
-9.23064053e-01 2.39332125e-01 6.10522330e-01 3.06343228e-01
5.71378112e-01 -1.85901120e-01 -2.63125271e-01 -9.17435467e-01
-9.02791917e-01 5.12118787e-02 5.05428612e-01 6.30694687e-01
4.97295707e-01 -1.40602574e-01 -4.40299869e-01 4.11686033e-01
4.03734706e-02 1.26561895e-01 5.12034178e-01 -1.23277545e+00
4.49550539e-01 6.05771780e-01 5.02598464e-01 -8.97251785e-01
-3.62927467e-01 -3.98477614e-02 -3.80539030e-01 1.24160290e-01
2.87086964e-01 -3.01200151e-01 -8.29917014e-01 1.61144340e+00
4.52248693e-01 9.09284592e-01 1.98245108e-01 9.16816890e-01
3.85381103e-01 8.25522244e-01 9.45256278e-02 -6.24132693e-01
1.00894904e+00 -9.25266385e-01 -8.89458597e-01 -4.41068321e-01
7.51570523e-01 -5.98947763e-01 3.36093068e-01 3.66461813e-01
-9.92342532e-01 -6.07269168e-01 -8.77698421e-01 9.54421759e-02
8.23274255e-03 4.74543124e-01 3.20075065e-01 8.18852261e-02
-1.01305866e+00 5.64642072e-01 -1.20041740e+00 -4.13247645e-01
1.30831733e-01 1.13000840e-01 -2.57193297e-01 -1.87776133e-01
-1.11960900e+00 7.06100225e-01 5.78331411e-01 3.24785709e-01
-1.07872474e+00 -1.94579393e-01 -6.75882757e-01 -2.54988521e-01
1.00354433e+00 -3.25596303e-01 1.85591960e+00 -1.08132768e+00
-1.67710650e+00 5.95268190e-01 -6.12737417e-01 -1.03252625e+00
6.77997530e-01 -4.64135587e-01 -7.90176928e-01 3.65405589e-01
-1.46681562e-01 4.06530321e-01 7.40214527e-01 -4.18908805e-01
-1.24485600e+00 -2.66890526e-01 3.75212193e-01 3.44299078e-01
-2.55900472e-01 5.40150642e-01 -9.07084048e-01 -4.06025559e-01
-4.34227996e-02 -1.01841462e+00 -2.66861051e-01 -3.13547403e-02
-8.76888856e-02 -4.54208285e-01 1.11032498e+00 -6.88548565e-01
1.57648063e+00 -2.26543140e+00 -7.43852034e-02 -2.08574876e-01
1.40265658e-01 4.83590245e-01 -3.69613543e-02 5.86264968e-01
-1.52255157e-02 -4.60360467e-01 3.10722530e-01 -2.95092970e-01
-1.80000097e-01 3.23747098e-01 -4.24071014e-01 4.81110960e-01
7.45509714e-02 6.94610298e-01 -1.05992723e+00 -4.77749616e-01
6.63983703e-01 3.06225777e-01 -1.05272919e-01 3.73368084e-01
-4.09496337e-01 4.60057944e-01 -5.99944234e-01 4.02358860e-01
5.31246327e-02 -1.69524431e-01 2.76888818e-01 1.27369851e-01
-5.31719804e-01 3.84397000e-01 -1.08454776e+00 1.64488149e+00
-2.76649117e-01 9.39774871e-01 -3.78575057e-01 -8.62623811e-01
5.09031892e-01 7.15069473e-01 8.89059007e-01 -9.86628413e-01
6.86906874e-02 -1.68985799e-01 -1.69622466e-01 -7.91539729e-01
6.93445563e-01 3.12031746e-01 2.40812987e-01 5.11253476e-01
-2.82926410e-01 8.80716324e-01 5.80192864e-01 2.43670136e-01
1.54931700e+00 3.23839575e-01 4.98949379e-01 6.69570327e-01
5.23577988e-01 -1.48751527e-01 9.79828775e-01 7.66778827e-01
-6.22386754e-01 2.79213071e-01 6.26987398e-01 -6.42407775e-01
-7.56858408e-01 -5.15251994e-01 4.98417914e-01 1.16742659e+00
7.91897401e-02 -5.82213104e-01 -4.41191018e-01 -7.13573515e-01
-3.70186806e-01 6.92101657e-01 -6.33969545e-01 -7.14301364e-03
-8.23581576e-01 -2.71495432e-01 1.21941037e-01 7.53904104e-01
7.03082204e-01 -1.24234664e+00 -1.23191094e+00 5.66622317e-01
-3.05185735e-01 -1.59365582e+00 -5.53065300e-01 -4.94678259e-01
-7.47038782e-01 -1.00974274e+00 -5.24650335e-01 -2.98237622e-01
1.37925476e-01 5.29878914e-01 9.31122601e-01 -2.05639023e-02
-3.48992944e-02 5.39051473e-01 -6.13234818e-01 -8.60112607e-02
-4.45744723e-01 -4.33539718e-01 -3.00937258e-02 4.80668455e-01
4.22928274e-01 -4.54580575e-01 -7.68275738e-01 5.73639452e-01
-7.65019834e-01 2.62615055e-01 6.53538704e-01 3.01037073e-01
5.80615461e-01 3.03890407e-01 2.93085724e-01 -4.47604686e-01
1.61032215e-01 -4.06060845e-01 -5.80379784e-01 4.28491175e-01
-2.07893774e-01 -2.90198714e-01 4.49482560e-01 -4.23910707e-01
-1.30085981e+00 1.79559603e-01 -1.57039747e-01 -7.18507230e-01
-1.69732019e-01 6.32611632e-01 4.31632511e-02 3.39060962e-01
1.95667669e-01 4.31674927e-01 -1.49951339e-01 -3.67899358e-01
9.96149853e-02 5.33620834e-01 8.17271531e-01 6.86995238e-02
2.15701804e-01 6.20935857e-01 -1.52462125e-01 -7.82827735e-01
-8.74349713e-01 -8.95579994e-01 -6.60419583e-01 -7.93161571e-01
8.44841123e-01 -1.08940899e+00 -9.22564149e-01 7.79610634e-01
-1.23002028e+00 -3.62319559e-01 -2.56372273e-01 7.59362757e-01
-5.66643357e-01 4.38981801e-01 -5.61875105e-01 -1.07580793e+00
7.41642714e-02 -9.39236999e-01 9.44682062e-01 1.47780567e-01
-1.34351730e-01 -8.16299677e-01 5.85029349e-02 1.81430161e-01
1.10737041e-01 3.24645281e-01 1.39482677e-01 -3.86953920e-01
-7.37575948e-01 -6.24258399e-01 3.89843471e-02 3.36218089e-01
1.79519922e-01 1.33277804e-01 -7.33863235e-01 -1.01358324e-01
1.01954684e-01 -6.68359026e-02 9.13239598e-01 5.07429421e-01
1.07357538e+00 -3.00043970e-01 -3.61445218e-01 9.52877030e-02
1.16932428e+00 7.26983130e-01 8.89098227e-01 2.00255126e-01
3.93828034e-01 4.90729988e-01 1.12822318e+00 8.63408387e-01
4.04230326e-01 8.70624125e-01 5.15477538e-01 1.33028999e-01
4.76689152e-02 -2.82933414e-01 8.22512507e-01 4.13849264e-01
-3.39072913e-01 -6.02886975e-01 -7.82087445e-01 6.58723474e-01
-2.41742349e+00 -1.51530790e+00 2.24507123e-01 2.29700708e+00
2.57592589e-01 4.18652862e-01 4.76568431e-01 1.17977850e-01
6.42810285e-01 5.52147627e-01 -8.14701736e-01 7.68528432e-02
3.89215797e-02 -3.82886946e-01 5.07838368e-01 2.02777997e-01
-1.41155934e+00 8.88118863e-01 6.20818329e+00 6.32189333e-01
-1.15420687e+00 4.73454148e-02 6.66954935e-01 -4.28377122e-01
4.52923805e-01 1.02797084e-01 -7.15515912e-01 4.12247688e-01
1.16922021e+00 -1.53562903e-01 1.64758042e-01 7.58300543e-01
9.36907291e-01 -5.03501058e-01 -1.04439569e+00 9.73703802e-01
3.31528597e-02 -1.35252655e+00 -4.44612443e-01 -8.93322080e-02
4.05205578e-01 2.49155939e-01 -1.25891224e-01 1.82051986e-01
1.87272072e-01 -4.29228276e-01 6.93184376e-01 7.75383294e-01
5.15182853e-01 -6.37987971e-01 5.84213853e-01 5.32185256e-01
-1.39611530e+00 -4.60331053e-01 8.52001384e-02 -4.46774036e-01
6.11184239e-01 4.99527425e-01 -7.06001401e-01 4.59705293e-01
5.03102124e-01 1.47828591e+00 -3.42060953e-01 8.79131913e-01
-4.24064219e-01 7.53986537e-01 -8.43673572e-02 1.11182876e-01
4.04277891e-01 6.90646768e-02 6.39908016e-01 8.43419373e-01
3.50846112e-01 5.80531299e-01 5.63796818e-01 2.27577031e-01
1.31214917e-01 -2.67606884e-01 -5.61630309e-01 -1.39211610e-01
2.02252865e-01 9.38377500e-01 -6.36608124e-01 -6.63214803e-01
-7.62477398e-01 9.96360719e-01 3.90704721e-02 3.67663920e-01
-1.06161988e+00 4.23001759e-02 7.58118272e-01 -1.49099246e-01
5.33113360e-01 -4.19018358e-01 4.84317958e-01 -1.23622453e+00
4.42722887e-01 -6.56404018e-01 6.19175136e-01 -9.24479485e-01
-6.57007754e-01 4.70924675e-01 -3.37905437e-02 -1.80725455e+00
-5.65998375e-01 -3.02523047e-01 -7.22532511e-01 3.64435613e-01
-1.49729419e+00 -8.79148066e-01 -1.50182307e-01 5.30494630e-01
8.47300708e-01 1.10485218e-01 3.30163479e-01 3.49645525e-01
-8.53617549e-01 1.41386181e-01 -2.63819955e-02 2.09186003e-01
5.04913390e-01 -7.38029301e-01 6.11509562e-01 1.27363169e+00
1.35785118e-01 -2.95469239e-02 7.86818147e-01 -7.00367093e-01
-1.36943460e+00 -1.14889920e+00 1.04929459e+00 -2.57621497e-01
7.82079637e-01 7.30588809e-02 -8.31828356e-01 8.10121596e-01
2.05036119e-01 1.72703788e-01 3.03017884e-01 -2.24150479e-01
1.49540067e-01 -1.31022528e-01 -6.78151965e-01 5.83759129e-01
1.16907704e+00 -6.19785726e-01 -4.68875378e-01 4.68008220e-01
7.23214447e-01 -5.48948109e-01 -6.15345299e-01 4.06075716e-01
6.51195347e-01 -1.33282137e+00 6.28760517e-01 -6.46427631e-01
3.80079687e-01 -3.53131086e-01 2.14700885e-02 -8.08860898e-01
-1.52231470e-01 -1.10969448e+00 -4.71223176e-01 9.02719617e-01
1.46730766e-01 -4.32504237e-01 8.50635767e-01 8.76137257e-01
-1.86249509e-01 -7.31677771e-01 -1.05448854e+00 -7.26609290e-01
-1.02487612e+00 -9.57237422e-01 4.54770982e-01 4.68890011e-01
-1.71582680e-02 5.59021495e-02 -9.35110509e-01 1.38034537e-01
5.45456894e-02 1.59088224e-01 8.56557310e-01 -6.58469319e-01
-2.82341123e-01 -1.58313550e-02 -9.00428832e-01 -1.60634744e+00
1.41127512e-01 -1.47406444e-01 2.20041841e-01 -1.32405651e+00
9.35010463e-02 2.60992855e-01 -6.38658822e-01 5.71770966e-01
-5.29526323e-02 -5.12604192e-02 2.75787383e-01 2.86985070e-01
-1.15049326e+00 6.16016805e-01 9.16420400e-01 8.33924040e-02
-3.53711277e-01 2.29641214e-01 3.91271338e-02 9.49253082e-01
5.68647504e-01 -2.42794126e-01 -5.36616087e-01 -3.84732813e-01
-1.31300222e-02 7.65758514e-01 4.97766167e-01 -1.34548664e+00
5.67661583e-01 -4.14927572e-01 4.89648525e-03 -9.56487358e-01
5.86377025e-01 -8.31510901e-01 4.77440618e-02 3.14982682e-01
-4.68854815e-01 2.20690653e-01 9.67504233e-02 1.00024104e+00
-4.20239896e-01 2.44590640e-01 4.42878306e-01 1.08157232e-01
-1.61093128e+00 5.70412517e-01 -6.63871646e-01 -1.90366089e-01
1.39065325e+00 -2.75924325e-01 -2.80717313e-01 -7.28199720e-01
-8.42220008e-01 3.99289817e-01 2.34118432e-01 5.76010823e-01
8.10023963e-01 -1.13036084e+00 -6.61700368e-01 -6.09977171e-02
8.91676694e-02 -6.28259838e-01 7.39851058e-01 1.12326205e+00
-1.97867990e-01 4.85136896e-01 8.52310937e-03 -5.59882462e-01
-1.58869696e+00 7.27305830e-01 2.16163650e-01 -3.60354245e-01
-9.05133605e-01 5.47221720e-01 7.10640997e-02 4.62046266e-01
3.88356030e-01 -3.17459673e-01 -5.28493881e-01 7.10335514e-03
9.90063488e-01 4.73776817e-01 -2.64312267e-01 -8.57536435e-01
-2.52670437e-01 2.03409478e-01 -1.36032447e-01 -1.89908549e-01
1.19027424e+00 -3.42145711e-01 3.48216575e-03 7.24411726e-01
1.08456755e+00 -5.40967464e-01 -1.80488873e+00 -6.93266451e-01
1.49362594e-01 -7.45713592e-01 8.78351480e-02 -5.24565399e-01
-1.12270796e+00 5.33615410e-01 6.14534855e-01 7.64221847e-02
1.52984369e+00 -1.29852176e-01 1.16129732e+00 4.56317872e-01
3.19346666e-01 -1.36397743e+00 6.73753098e-02 5.66010773e-01
5.88370621e-01 -1.23409677e+00 -4.46718698e-03 -2.21361816e-01
-7.80730903e-01 1.03470111e+00 4.85106200e-01 1.62445113e-01
5.65668523e-01 -1.29239649e-01 -8.40916559e-02 -5.01586124e-02
-1.39391863e+00 -4.41598237e-01 2.64854759e-01 2.26529706e-02
1.46011665e-01 -1.87284574e-01 -6.75443262e-02 -5.27217276e-02
4.90355670e-01 5.61003625e-01 4.89898682e-01 1.43675053e+00
-3.73402238e-01 -8.71392787e-01 -9.90649313e-02 4.05820698e-01
-5.29029012e-01 1.87776148e-01 -3.87803435e-01 5.75765610e-01
5.50508834e-02 1.21906900e+00 3.11733723e-01 -5.52980959e-01
3.15661073e-01 4.65440229e-02 1.93319544e-01 -3.43540430e-01
-1.17150098e-01 8.88445452e-02 4.86595780e-01 -1.09482872e+00
-9.94870067e-01 -1.02171171e+00 -1.06649303e+00 -1.29825130e-01
-1.85260754e-02 -1.96100250e-01 3.15927833e-01 1.23451126e+00
6.29363596e-01 5.61831594e-01 8.77116919e-01 -8.39935243e-01
-2.08121493e-01 -7.82117248e-01 -3.71172607e-01 3.39025617e-01
5.78429818e-01 -5.50688863e-01 -1.39326349e-01 2.29496017e-01] | [8.250199317932129, 0.45918816328048706] |
77e6edd2-f4fe-42e2-aa8d-eea5e2573fe0 | recommendations-for-datasets-for-source-code | 1904.02660 | null | http://arxiv.org/abs/1904.02660v1 | http://arxiv.org/pdf/1904.02660v1.pdf | Recommendations for Datasets for Source Code Summarization | Source Code Summarization is the task of writing short, natural language
descriptions of source code. The main use for these descriptions is in software
documentation e.g. the one-sentence Java method descriptions in JavaDocs. Code
summarization is rapidly becoming a popular research problem, but progress is
restrained due to a lack of suitable datasets. In addition, a lack of community
standards for creating datasets leads to confusing and unreproducible research
results -- we observe swings in performance of more than 33% due only to
changes in dataset design. In this paper, we make recommendations for these
standards from experimental results. We release a dataset based on prior work
of over 2.1m pairs of Java methods and one sentence method descriptions from
over 28k Java projects. We describe the dataset and point out key differences
from natural language data, to guide and support future researchers. | ['Alexander LeClair', 'Collin McMillan'] | 2019-04-04 | recommendations-for-datasets-for-source-code-1 | https://aclanthology.org/N19-1394 | https://aclanthology.org/N19-1394.pdf | naacl-2019-6 | ['code-summarization'] | ['computer-code'] | [ 2.70946950e-01 1.75207734e-01 -5.44661641e-01 -5.86485624e-01
-1.03209996e+00 -8.36206138e-01 4.62903351e-01 5.84383786e-01
-6.19782284e-02 4.98754203e-01 8.00928056e-01 -3.86147022e-01
5.47750341e-03 -7.57345790e-03 -4.72647190e-01 1.20283157e-01
7.70410001e-02 -2.00924844e-01 2.93535233e-01 -1.86529338e-01
1.09250760e+00 -6.03353791e-02 -1.59233451e+00 6.88423097e-01
9.11353171e-01 -7.31186271e-02 2.37438783e-01 7.35048473e-01
-6.78743243e-01 1.21561575e+00 -1.03420901e+00 -5.61248600e-01
-6.14680871e-02 -6.71247005e-01 -1.24064875e+00 -4.19533364e-02
6.96034908e-01 -2.47074232e-01 1.52903154e-01 1.03437090e+00
4.42073166e-01 -3.57879937e-01 2.21625805e-01 -1.60397482e+00
-4.36149567e-01 1.11098218e+00 -9.11925733e-01 1.42982528e-01
8.34615767e-01 -6.35982379e-02 1.00721610e+00 -5.25026679e-01
1.02454710e+00 8.36107671e-01 9.49561954e-01 7.45273590e-01
-1.18337476e+00 -3.96194339e-01 -2.61790454e-01 -2.67716646e-01
-1.07584691e+00 -9.55750704e-01 4.43701237e-01 -8.48472655e-01
1.65147579e+00 5.49579620e-01 3.24993163e-01 1.04469717e+00
5.16469717e-01 4.01680589e-01 6.01839662e-01 -6.35810316e-01
2.36742478e-02 3.76347095e-01 5.22680521e-01 5.74886322e-01
8.17768335e-01 -7.40116954e-01 -3.66498232e-01 -9.72575307e-01
4.00314182e-02 -1.18090138e-01 -1.40391707e-01 -2.86417037e-01
-1.20456815e+00 6.11643195e-01 -4.48561281e-01 4.83973742e-01
1.08701348e-01 2.65434265e-01 9.02197897e-01 4.04323339e-01
2.52239197e-01 8.33502829e-01 -6.15531504e-01 -1.14244163e+00
-1.09336257e+00 6.38434529e-01 1.37227035e+00 1.54913747e+00
8.05951416e-01 -9.95566174e-02 1.74054340e-01 9.84460711e-01
3.10391873e-01 1.84987083e-01 7.48455942e-01 -1.40414572e+00
7.24534214e-01 1.00048935e+00 1.64001852e-01 -1.17935491e+00
-2.77545333e-01 4.16908599e-02 -9.78136659e-02 4.18667123e-02
1.24162234e-01 -1.06288761e-01 -1.41247988e-01 1.19923580e+00
-2.12298125e-01 -8.69524956e-01 9.79242027e-02 1.70663282e-01
9.69898820e-01 3.06937754e-01 -2.82123208e-01 -3.53488863e-01
1.08897555e+00 -8.39050114e-01 -6.54175818e-01 -4.18170184e-01
1.05430603e+00 -1.31002045e+00 1.11933148e+00 2.64904827e-01
-1.01751518e+00 -6.90713525e-02 -8.46771240e-01 -1.83360040e-01
-3.45383808e-02 1.54362172e-02 6.57833636e-01 7.40173876e-01
-1.14391661e+00 4.85828906e-01 -8.75202000e-01 -9.19200480e-01
1.03486940e-01 -3.87684554e-02 -3.80123407e-01 1.16208382e-01
-8.66161585e-02 6.14742517e-01 2.30598122e-01 -4.10510033e-01
-5.04625201e-01 -8.09255838e-01 -8.25402379e-01 -1.98976144e-01
4.86690849e-01 -3.33518803e-01 1.91783428e+00 -8.15765202e-01
-9.36536908e-01 7.98587441e-01 -1.45356864e-01 -1.74526125e-02
2.60965288e-01 -2.30290040e-01 -1.79533854e-01 -2.00705037e-01
6.00588799e-01 2.45710641e-01 2.96371162e-01 -1.34093332e+00
-8.52366745e-01 -8.00476968e-02 2.65280157e-01 -1.78648844e-01
-4.33682114e-01 7.72630751e-01 -4.32615548e-01 -5.83959997e-01
-2.73776770e-01 -8.35605621e-01 -2.02706873e-01 -6.17241204e-01
-4.14877594e-01 -2.46560737e-01 4.95277286e-01 -9.14058208e-01
1.72796392e+00 -2.03366899e+00 -1.49243206e-01 -3.85926574e-01
4.63872194e-01 -2.69504428e-01 -2.81796604e-01 1.11300266e+00
4.21540737e-02 7.92512357e-01 -3.64669502e-01 -1.08216606e-01
1.02059975e-01 -2.21195757e-01 -3.86398673e-01 3.82418752e-01
-5.47697730e-02 4.11545992e-01 -1.09321749e+00 -6.62310779e-01
-4.48577791e-01 3.92897474e-03 -7.54792213e-01 -3.29456851e-03
-2.70186424e-01 -2.15101168e-01 -4.86973047e-01 7.66498268e-01
4.41747546e-01 -9.64597389e-02 2.74349004e-01 2.53027260e-01
-6.07336938e-01 7.83853471e-01 -8.46469700e-01 2.15191054e+00
-4.00517166e-01 9.77075934e-01 -7.83100575e-02 -4.08863515e-01
7.85379171e-01 2.20239118e-01 4.28927928e-01 -3.80589277e-01
-3.78864348e-01 4.46992338e-01 8.19595009e-02 -1.11442876e+00
8.36648703e-01 3.88800502e-01 -5.04374206e-01 9.18552458e-01
-2.46970311e-01 -4.53387231e-01 8.24024200e-01 5.44879735e-01
1.64161897e+00 2.28835240e-01 5.87291479e-01 -4.33941036e-01
5.83424158e-02 6.41410589e-01 6.46545291e-01 9.05614734e-01
1.45858169e-01 6.45655870e-01 1.05638695e+00 -3.44408840e-01
-1.10002387e+00 -3.57003242e-01 -6.67454749e-02 1.03851438e+00
-2.84939468e-01 -1.36215448e+00 -8.80711198e-01 -8.81377161e-01
-1.30018130e-01 8.82560492e-01 -3.48529309e-01 -7.56039619e-02
-7.18888283e-01 -5.10700703e-01 9.16725218e-01 3.99226695e-01
5.25290556e-02 -8.88463616e-01 -1.07169282e+00 2.33383372e-01
-3.33526224e-01 -7.20212102e-01 -6.46240652e-01 -1.89450577e-01
-9.94043410e-01 -1.20145488e+00 -3.43312353e-01 -6.36032879e-01
5.69062769e-01 4.92170602e-01 1.52229595e+00 5.07034123e-01
-3.85339558e-01 6.18382275e-01 -6.94893599e-01 -4.14928406e-01
-1.21470404e+00 2.07056984e-01 -3.11928570e-01 -1.06449771e+00
5.75831890e-01 -3.92482609e-01 -7.75600374e-02 2.92525366e-02
-9.92902100e-01 4.05516960e-02 6.74951732e-01 4.26664323e-01
-1.04320012e-01 -1.96348011e-01 3.83072257e-01 -1.25205016e+00
8.98954749e-01 -7.09561527e-01 -4.62977946e-01 3.23533922e-01
-9.01265144e-01 8.16982836e-02 4.50935572e-01 -9.25871506e-02
-1.17541134e+00 -1.01904809e-01 1.58332616e-01 5.61721385e-01
-2.24556357e-01 8.79395068e-01 4.15283114e-01 6.90063462e-02
1.06078792e+00 5.51517792e-02 -1.84229016e-02 -7.34243989e-01
9.87307429e-02 1.18588173e+00 1.95039868e-01 -7.39812374e-01
4.35813934e-01 4.92890477e-02 -8.41273308e-01 -1.01621139e+00
-2.07259357e-01 -4.63295549e-01 -1.97417185e-01 2.26207748e-01
1.86249688e-01 -6.45127177e-01 -1.59631491e-01 1.93883866e-01
-1.39208031e+00 -3.32486570e-01 -1.30167961e-01 1.81060195e-01
-5.11425376e-01 7.97771811e-01 -6.10212684e-01 -5.31989813e-01
-2.93808252e-01 -1.30447042e+00 9.98896718e-01 1.13163143e-01
-1.00175989e+00 -6.83382928e-01 4.31739420e-01 4.69691217e-01
6.45913482e-01 2.69100666e-01 1.00310647e+00 -5.75012505e-01
-2.89430767e-01 -1.93804383e-01 -1.19055785e-01 2.07630455e-01
2.89863080e-01 7.71240592e-01 -4.62314069e-01 -3.63166690e-01
1.63816176e-02 -2.66091079e-01 3.02471668e-01 -6.12512156e-02
9.21479821e-01 -6.31200492e-01 -4.58265215e-01 9.42121744e-02
1.55662024e+00 2.24763691e-01 4.36795503e-01 5.79378903e-01
5.66474855e-01 7.26699054e-01 4.54145014e-01 7.66428292e-01
6.62548304e-01 5.14016032e-01 1.11949988e-01 6.21666849e-01
-2.70602912e-01 -2.20329538e-02 6.21090293e-01 1.26623178e+00
1.46775708e-01 -3.26099619e-02 -1.60131383e+00 8.95621538e-01
-1.63298452e+00 -9.45225060e-01 -7.46419847e-01 2.09244251e+00
1.27178752e+00 4.65984941e-02 1.54187575e-01 -2.61305243e-01
3.84119332e-01 -1.51289869e-02 -1.91688925e-01 -7.16155410e-01
2.94834584e-01 -3.09114456e-01 3.30218643e-01 2.21186966e-01
-4.96550918e-01 3.23569506e-01 6.56936836e+00 4.26086843e-01
-9.62408721e-01 1.59419863e-03 -1.51808396e-01 -8.54125321e-02
-7.38768578e-01 6.23108983e-01 -7.90232658e-01 5.99259615e-01
1.12093771e+00 -1.00053608e+00 3.38742793e-01 1.03045988e+00
1.55610785e-01 -4.84475613e-01 -1.37502778e+00 7.53111064e-01
3.44151318e-01 -1.42934048e+00 -1.51114836e-01 -3.82561311e-02
9.29482043e-01 4.33932275e-01 -6.35376990e-01 4.08334821e-01
3.42332751e-01 -5.83874643e-01 9.41863000e-01 2.57858932e-01
6.15068793e-01 -2.28896618e-01 7.31103957e-01 4.79199499e-01
-7.41163552e-01 -4.77974042e-02 -3.00942391e-01 -3.27501386e-01
-4.19774890e-01 4.82931107e-01 -9.20870185e-01 4.18709219e-01
9.73728776e-01 8.05475771e-01 -9.62454855e-01 1.20951951e+00
1.89905658e-01 6.26317263e-01 1.00472830e-01 -3.28471541e-01
-2.28178352e-01 2.43878677e-01 5.56560993e-01 1.80321336e+00
1.94227085e-01 -3.35021019e-01 5.93063869e-02 8.44562769e-01
-6.20805845e-02 2.82312274e-01 -8.59708786e-01 -5.32609820e-01
5.42114496e-01 1.21876812e+00 -8.54143202e-01 -2.48702914e-01
-8.49753678e-01 6.72786713e-01 1.29082203e-01 2.43233025e-01
-5.25332451e-01 -1.18066168e+00 5.90769708e-01 1.24369077e-01
-3.77713367e-02 -1.72262877e-01 -3.48715842e-01 -1.31300843e+00
6.00578189e-01 -1.45638835e+00 6.97861090e-02 -6.84606075e-01
-8.61999691e-01 5.05350828e-01 3.15362185e-01 -9.91324544e-01
-4.82992887e-01 7.52675682e-02 -4.66716141e-01 6.27790213e-01
-1.01772928e+00 -4.69735712e-01 -4.25724566e-01 -3.96840364e-01
7.74037182e-01 -1.59841925e-01 7.84616590e-01 3.10944080e-01
-4.38723892e-01 4.03087676e-01 3.02283973e-01 1.64451543e-02
1.00188959e+00 -1.38879251e+00 8.95083904e-01 9.82848346e-01
-3.62120979e-02 1.43547380e+00 1.06473625e+00 -8.66373777e-01
-1.58646297e+00 -6.92101181e-01 1.37374520e+00 -9.05609965e-01
7.91135311e-01 -3.98114920e-01 -8.59704554e-01 8.22219789e-01
7.66524851e-01 -4.17271316e-01 7.76004195e-01 7.59629766e-03
-4.13371235e-01 -1.25170154e-02 -9.17552471e-01 6.58874273e-01
9.14749980e-01 -4.33258563e-01 -7.65570343e-01 4.27187830e-01
6.08501256e-01 -4.97025579e-01 -8.83437395e-01 -1.33178383e-02
6.70728743e-01 -1.14141178e+00 2.12109387e-01 -5.13511121e-01
9.57091093e-01 -2.15230152e-01 -1.39167365e-02 -1.09788609e+00
1.00308932e-01 -1.00434232e+00 1.94665775e-01 1.77521729e+00
4.73072290e-01 -4.52802539e-01 5.34801424e-01 8.67672145e-01
-2.97137499e-01 -4.03848588e-01 -2.92228699e-01 -6.67765319e-01
3.00856046e-02 -3.50346059e-01 4.04984385e-01 1.20465100e+00
6.40129507e-01 2.38614127e-01 1.11087389e-01 -4.12969828e-01
4.63217258e-01 2.31573626e-01 1.19008124e+00 -1.00711119e+00
-3.29971611e-01 -6.16717935e-01 -2.96494663e-01 -7.65302956e-01
8.11503753e-02 -9.18423772e-01 2.03465953e-01 -1.86548150e+00
1.02974880e+00 -1.98155448e-01 5.47306061e-01 7.35251844e-01
8.90888050e-02 -3.63036990e-02 4.48941551e-02 4.86547261e-01
-7.78984547e-01 -1.36412546e-01 3.08678478e-01 -9.08168331e-02
-2.46083364e-01 -1.11343116e-01 -1.19252264e+00 6.53136313e-01
8.54739189e-01 -1.01599109e+00 -4.91360307e-01 -8.45615864e-01
7.39609361e-01 1.91342741e-01 6.13668794e-03 -1.04361963e+00
1.37110949e-01 -2.44998693e-01 -3.74719650e-01 -2.86889374e-01
-5.65790534e-01 -5.16773283e-01 2.57779449e-01 3.69292051e-01
-6.27382636e-01 4.20282394e-01 4.09915894e-01 1.33499473e-01
-7.20385984e-02 -9.61586416e-01 2.88106531e-01 -3.71738315e-01
-6.67909503e-01 -4.21406209e-01 -6.60053670e-01 6.07167125e-01
8.61174703e-01 -3.64090055e-01 -8.88640940e-01 -6.00432158e-02
2.16398448e-01 4.14528102e-02 1.20896077e+00 8.09600115e-01
2.50676572e-01 -8.65777552e-01 -7.41188109e-01 -1.60122782e-01
6.33610427e-01 -2.46376127e-01 -1.07254416e-01 8.01897109e-01
-8.46479774e-01 3.80893499e-01 -2.24427044e-01 -3.89056027e-01
-1.36330700e+00 1.84097633e-01 -1.15278922e-01 2.36818179e-01
-7.31955290e-01 4.08149332e-01 -3.91677350e-01 -4.14137125e-01
1.48449168e-01 -4.65880007e-01 6.70882538e-02 -1.33515283e-01
8.14272285e-01 6.51806355e-01 3.32859844e-01 -3.40742975e-01
-6.50601685e-01 2.75075227e-01 -4.31143641e-01 -2.25887783e-02
1.56232297e+00 -1.13139473e-01 -7.55116940e-01 8.63798678e-01
1.42289186e+00 7.25391448e-01 -7.90474355e-01 1.55792609e-01
7.60856509e-01 -5.20615637e-01 -3.53043199e-01 -7.98387706e-01
-5.76805174e-01 3.37349236e-01 -5.71428947e-02 6.38760567e-01
6.57666147e-01 2.18587667e-01 2.77612209e-01 6.14689112e-01
5.63607454e-01 -9.76428032e-01 -1.40714481e-01 3.41367573e-01
1.05878317e+00 -1.14787281e+00 4.30034190e-01 -1.79591566e-01
-4.26737815e-01 1.45320678e+00 6.40228868e-01 3.47334951e-01
1.07994176e-01 6.61832035e-01 2.13288993e-01 -4.18019444e-01
-1.01304102e+00 6.14266694e-01 -2.91770130e-01 5.28959036e-01
1.23511147e+00 -3.43046129e-01 -8.06658208e-01 6.48818612e-01
-1.34931773e-01 1.06967755e-01 1.63271630e+00 1.67334378e+00
-3.11142445e-01 -1.36652589e+00 -3.65036875e-01 8.70264947e-01
-7.64844656e-01 -2.32125103e-01 -6.75544441e-01 9.20729756e-01
-2.96328694e-01 1.17764115e+00 -2.11432680e-01 -2.19091088e-01
4.91358191e-01 -1.10543162e-01 3.36124301e-01 -1.27531457e+00
-9.46800113e-01 -3.96496534e-01 3.35020632e-01 -7.10265100e-01
-4.47399378e-01 -1.09251034e+00 -1.29930723e+00 -4.91620541e-01
-2.70961791e-01 4.53685403e-01 8.43486905e-01 4.90137219e-01
7.87410915e-01 2.96273708e-01 2.20760882e-01 -3.88019145e-01
-8.11889470e-01 -9.12620306e-01 -2.33829364e-01 2.61331022e-01
3.56449634e-01 -2.07981050e-01 -4.39756185e-01 6.57703221e-01] | [7.667816638946533, 7.910769462585449] |
9d9d28e7-df9b-418e-a0f4-fea2f582c38d | effectively-using-long-and-short-sessions-for | 2205.04366 | null | https://arxiv.org/abs/2205.04366v1 | https://arxiv.org/pdf/2205.04366v1.pdf | Effectively Using Long and Short Sessions for Multi-Session-based Recommendations | It is not accurate to make recommendations only based one single current session. Therefore, multi-session-based recommendation(MSBR) is a solution for the problem. Compared with the previous MSBR models, we have made three improvements in this paper. First, the previous work choose to use all the history sessions of the user and/or of his similar users. When the user's current interest changes greatly from the past, most of these sessions can only have negative impacts. Therefore, we select a large number of randomly chosen sessions from the dataset as candidate sessions to avoid over depending on history data. Then we only choose to use the most similar sessions to get the most useful information while reduce the noise caused by dissimilar sessions. Second, in real-world datasets, short sessions account for a large proportion. The RNN often used in previous work is not suitable to process short sessions, because RNN only focuses on the sequential relationship, which we find is not the only relationship between items in short sessions. So, we designed a more suitable method named GAFE based on attention to process short sessions. Third, Although there are few long sessions, they can not be ignored. Not like previous models, which simply process long sessions in the same way as short sessions, we propose LSIS, which can split the interest of long sessions, to make better use of long sessions. Finally, to help recommendations, we also have considered users' long-term interests captured by a multi-layer GRU. Considering the four points above, we built the model ENIREC. Experiments on two real-world datasets show that the comprehensive performance of ENIREC is better than other existing models. | ['Yan Wang', 'Gang Wu', 'Zihan Wang'] | 2022-05-09 | null | null | null | null | ['session-based-recommendations'] | ['miscellaneous'] | [-1.42110988e-01 -2.94983715e-01 -4.03337985e-01 -4.25656676e-01
-5.40281534e-02 -2.00376123e-01 1.64589107e-01 -2.40327179e-01
-4.22480762e-01 6.63574934e-01 5.30116200e-01 -3.81282456e-02
-3.96686226e-01 -1.06035638e+00 -4.73931789e-01 -7.36086607e-01
4.86170389e-02 1.99027389e-01 4.37359720e-01 -5.29040098e-01
3.86763901e-01 5.87600581e-02 -1.63023484e+00 4.91105020e-01
1.08060646e+00 6.30058408e-01 6.78609073e-01 -9.92501527e-03
-5.26432514e-01 5.26115477e-01 -4.41911131e-01 -8.89283046e-02
6.00965396e-02 -6.59687638e-01 -6.69038773e-01 -1.00357771e-01
-3.65316451e-01 -4.42667156e-01 -3.36143494e-01 7.07970619e-01
5.61340213e-01 7.17707574e-01 3.18232059e-01 -9.47556615e-01
-7.44426668e-01 1.15640795e+00 -5.33186495e-01 3.74649465e-01
3.64461809e-01 -3.70739490e-01 1.14857757e+00 -7.52035797e-01
2.58022040e-01 1.26861370e+00 4.47902441e-01 8.44706833e-01
-6.64222300e-01 -8.10051084e-01 9.55996871e-01 3.84018838e-01
-1.13196588e+00 -1.36845514e-01 7.61565983e-01 1.45698622e-01
6.68255031e-01 4.97451425e-01 6.64633930e-01 1.17313373e+00
-3.03279161e-01 1.04539204e+00 6.84065402e-01 -1.34502396e-01
-5.50204106e-02 2.87114143e-01 5.49520135e-01 -8.57389122e-02
1.51826413e-02 -1.56057268e-01 -1.09386377e-01 -5.31992912e-02
7.34293938e-01 7.15334654e-01 -5.81725955e-01 2.03863189e-01
-1.17896938e+00 7.40146041e-01 5.19875050e-01 8.50163162e-01
-5.08517444e-01 -3.22392672e-01 1.28321365e-01 3.40912044e-01
3.74217600e-01 2.17706636e-01 -3.27780098e-01 -1.61468729e-01
-5.93846202e-01 4.77876440e-02 6.34497762e-01 7.89486408e-01
7.18176305e-01 -1.85286418e-01 -2.68159658e-01 1.21037388e+00
4.47412282e-01 1.35280207e-01 8.38817358e-01 -4.95857686e-01
5.03904343e-01 6.11388445e-01 2.12823704e-01 -1.11840153e+00
-2.80678540e-01 -7.78452158e-01 -1.05802572e+00 -5.47497034e-01
1.60564721e-01 -1.57072604e-01 -6.58750892e-01 1.68831253e+00
-1.26496613e-01 3.58613461e-01 -5.23903184e-02 1.13479650e+00
1.05296957e+00 1.03808069e+00 -7.10264221e-02 -8.74593139e-01
8.94533217e-01 -1.18545318e+00 -9.43781495e-01 -1.19200215e-01
2.76646972e-01 -7.09304452e-01 1.08769751e+00 5.63954532e-01
-9.85876977e-01 -6.96492136e-01 -7.66323030e-01 3.97593290e-01
-2.63817608e-01 3.25670578e-02 6.96665049e-01 3.44293386e-01
-9.15546894e-01 8.20269763e-01 -3.27667058e-01 -5.71495831e-01
4.45205569e-02 1.38664544e-01 2.95019299e-01 -3.99017669e-02
-1.54482400e+00 5.62336385e-01 2.89001018e-01 4.97167706e-01
-3.90962124e-01 -3.33341002e-01 -1.90006301e-01 2.52742827e-01
6.20330989e-01 -3.40633750e-01 1.12389505e+00 -1.15360248e+00
-1.28378272e+00 -2.02967357e-02 -4.63886589e-01 -1.27988353e-01
3.20525706e-01 -2.02384457e-01 -8.29943419e-01 -3.85680705e-01
-2.75613993e-01 1.57196000e-01 3.82172257e-01 -1.27924252e+00
-9.06247854e-01 -1.89038351e-01 2.94443637e-01 6.07347846e-01
-7.57596910e-01 -1.68486591e-02 -9.83763576e-01 -5.51292241e-01
3.37997764e-01 -9.04118955e-01 -3.81824583e-01 -5.74085534e-01
-5.49957901e-03 -5.24925292e-01 7.61480510e-01 -6.20954275e-01
1.83370650e+00 -2.18467212e+00 1.33098453e-01 3.29796761e-01
-2.18252316e-02 4.16775882e-01 -2.52755910e-01 5.60016513e-01
6.07246906e-02 3.72685462e-01 2.53552139e-01 -1.83935374e-01
-2.21064687e-01 3.75091940e-01 -2.90740699e-01 -1.72242239e-01
-6.28842533e-01 6.21467710e-01 -8.54753196e-01 -1.54094025e-01
-2.60978509e-02 3.93180430e-01 -4.26113367e-01 4.04443443e-01
-1.54474825e-01 4.20043498e-01 -8.59305322e-01 2.55693644e-01
6.46215737e-01 -3.37173373e-01 5.27783558e-02 -1.57711431e-01
-1.17870115e-01 4.16129351e-01 -1.35727990e+00 1.46583533e+00
-6.22960985e-01 -2.73617320e-02 -2.26209030e-01 -9.71062899e-01
9.79262948e-01 4.45243239e-01 6.12109661e-01 -8.01367164e-01
3.36743593e-02 1.13400243e-01 1.02448180e-01 -5.84334314e-01
4.23418701e-01 9.93328243e-02 3.09378445e-01 6.31794393e-01
-3.97802681e-01 8.35710406e-01 3.57770264e-01 3.60055715e-01
7.90261149e-01 8.63112807e-02 1.55121461e-01 1.13877311e-01
6.97452843e-01 -6.45871460e-01 1.07010162e+00 8.58885825e-01
1.51061034e-02 7.83767581e-01 1.81019187e-01 -1.91138417e-01
-5.42929530e-01 -6.11732543e-01 2.95127779e-01 1.28431201e+00
5.76870978e-01 -4.84760016e-01 -2.75944233e-01 -9.04321313e-01
-2.97911823e-01 6.47786140e-01 -5.14538288e-01 -1.46468267e-01
-5.62287569e-01 -1.02733445e+00 5.38709350e-02 3.99655968e-01
5.74323714e-01 -1.54472983e+00 1.82657659e-01 5.39554000e-01
-6.90936089e-01 -6.53323293e-01 -6.74275935e-01 -2.91735500e-01
-8.82221460e-01 -8.49194348e-01 -9.08261478e-01 -6.24427497e-01
4.45064604e-01 1.07474387e+00 8.99236321e-01 4.83539343e-01
5.45527101e-01 9.75922719e-02 -1.00461709e+00 -2.15870038e-01
-4.68348786e-02 1.31498396e-01 1.28289312e-01 3.42536569e-01
5.32623529e-01 -8.50255430e-01 -6.92450702e-01 6.15622163e-01
-7.04514921e-01 -7.71905109e-03 6.21975303e-01 6.12059891e-01
3.62472355e-01 4.71144676e-01 1.16839576e+00 -1.09190691e+00
7.62891531e-01 -7.51908362e-01 9.56346318e-02 4.12752926e-01
-8.94947529e-01 -5.80179572e-01 9.62832987e-01 -6.43182099e-01
-1.39040077e+00 -7.22572982e-01 -3.49568278e-01 -3.17152709e-01
-1.84890330e-01 6.59911811e-01 -3.50848228e-01 4.36429739e-01
2.77766705e-01 2.81657040e-01 -2.06699684e-01 -9.06474590e-01
-6.68817386e-02 1.02977979e+00 -2.20471233e-01 -2.59819835e-01
3.93958151e-01 4.98960838e-02 -6.90955818e-01 -5.18725216e-01
-9.77907956e-01 -8.38275313e-01 -1.02457605e-01 -2.23847672e-01
4.33862090e-01 -7.39644408e-01 -7.44994402e-01 3.03146362e-01
-1.01050293e+00 -2.30114404e-02 4.11972217e-02 8.39483917e-01
1.50269549e-02 6.16215706e-01 -7.83361018e-01 -1.08465886e+00
-5.39024532e-01 -8.47110271e-01 2.72832215e-01 7.10809648e-01
4.36223745e-02 -8.94733846e-01 -2.68345475e-01 3.04569721e-01
6.45776212e-01 -5.22555947e-01 4.84489948e-01 -8.29813898e-01
-5.31096578e-01 5.94361685e-02 -1.91842422e-01 3.04253072e-01
4.80082154e-01 -2.10932612e-01 -7.49392629e-01 -3.83566469e-01
5.34942858e-02 2.16093495e-01 9.22099590e-01 2.26990134e-01
1.45243633e+00 -4.02445585e-01 -3.77285331e-01 2.38835275e-01
1.34375155e+00 6.46607459e-01 9.61868763e-01 3.26580644e-01
7.82241523e-01 6.51781082e-01 1.00233150e+00 3.77341628e-01
3.60031724e-01 5.98150790e-01 3.29807132e-01 -6.37595030e-03
2.10059494e-01 -3.19649875e-01 5.42904794e-01 1.22212541e+00
-6.49456441e-01 -6.54036462e-01 -2.17946038e-01 5.15778005e-01
-2.34563804e+00 -1.30733871e+00 -3.38205755e-01 2.40100455e+00
6.15078807e-01 1.73626125e-01 3.61739933e-01 4.19371463e-02
9.05999780e-01 1.73951030e-01 -4.34836805e-01 -1.52630106e-01
-2.51305141e-02 -3.60186309e-01 1.13111369e-01 1.56265214e-01
-6.75105214e-01 4.89392459e-01 5.13952398e+00 1.01941872e+00
-8.98641586e-01 -1.64602083e-04 4.42419618e-01 -3.54051709e-01
-6.24666274e-01 -6.40902594e-02 -1.07624376e+00 1.01190126e+00
7.43169963e-01 -1.58592254e-01 7.78087318e-01 4.94716972e-01
7.56832719e-01 1.79892063e-01 -8.01941633e-01 9.39919770e-01
2.44347245e-01 -8.05825531e-01 8.51334706e-02 5.04954495e-02
5.82123339e-01 -1.39957741e-01 -2.29854196e-01 6.83024526e-01
1.69269562e-01 -6.38548613e-01 9.72777009e-02 7.56366968e-01
6.05400987e-02 -8.40563059e-01 9.62264180e-01 8.45872402e-01
-1.22622013e+00 -4.11937565e-01 -7.40806282e-01 -2.31292238e-03
3.17224473e-01 8.55061054e-01 -2.24113464e-01 9.86693501e-01
9.59740937e-01 1.25980759e+00 -2.99445987e-01 1.32020473e+00
-1.59055218e-01 9.16136503e-01 -2.44711891e-01 -8.42263177e-02
1.67395964e-01 -6.29873335e-01 4.61001724e-01 1.07654870e+00
7.77620852e-01 3.85579646e-01 3.18454683e-01 4.79790539e-01
1.26767874e-01 5.50858438e-01 -4.58113194e-01 3.09561521e-01
4.12271529e-01 1.38809037e+00 -4.43510622e-01 -2.57217050e-01
-6.44100308e-01 8.34348142e-01 7.09462836e-02 6.35464191e-01
-8.21749866e-01 -2.51430452e-01 3.21857691e-01 1.34764552e-01
3.59603852e-01 2.55526781e-01 1.01339497e-01 -1.27967811e+00
4.36665453e-02 -7.53173172e-01 6.94881737e-01 -7.51070738e-01
-1.54622197e+00 6.22994363e-01 -1.23820953e-01 -1.57564962e+00
-5.27892895e-02 1.87511772e-01 -9.40035105e-01 9.95230436e-01
-1.56747532e+00 -9.66108620e-01 -2.14378536e-01 5.20928621e-01
8.04728150e-01 -1.18176587e-01 5.58835804e-01 7.47609735e-01
-7.46971548e-01 5.84203124e-01 2.55016953e-01 -2.19456375e-01
8.21587801e-01 -8.85852218e-01 -3.49266902e-02 7.13647664e-01
-1.77910011e-02 1.01650071e+00 5.55596471e-01 -6.87136352e-01
-8.26456964e-01 -9.57997918e-01 9.67984915e-01 8.25903267e-02
2.45319664e-01 -4.58961874e-02 -1.42763269e+00 7.30854690e-01
9.04195309e-02 -5.08495629e-01 7.74414718e-01 5.42372108e-01
1.16179757e-01 -4.54297513e-01 -8.09853792e-01 6.90866649e-01
1.34722650e+00 -3.08041833e-02 -6.22093678e-01 9.08187628e-02
7.72691250e-01 4.43786494e-02 -6.92942441e-01 4.13572341e-01
4.76908416e-01 -9.15359139e-01 8.19904983e-01 -4.06231791e-01
2.20660329e-01 -3.85422528e-01 1.89376399e-01 -1.50845623e+00
-7.67739058e-01 -4.31563616e-01 -2.38848075e-01 1.69943380e+00
3.42258275e-01 -6.76149309e-01 7.38334835e-01 4.90529001e-01
-7.52258152e-02 -6.45960987e-01 -2.62207359e-01 -5.54880083e-01
-3.65382612e-01 -1.24117233e-01 8.72610033e-01 9.85112369e-01
2.63273008e-02 4.36361343e-01 -8.60298157e-01 -1.70129389e-02
2.88472205e-01 4.31762695e-01 4.52651918e-01 -1.45731878e+00
-4.90031570e-01 -3.09227675e-01 3.83139312e-01 -1.43924916e+00
-1.69281259e-01 -5.91573536e-01 2.58873049e-02 -1.89705956e+00
4.81058538e-01 -8.10640037e-01 -1.00272655e+00 4.33846414e-01
-3.44440371e-01 7.96293467e-02 7.00721368e-02 5.29664755e-01
-7.62163401e-01 5.34363389e-01 1.58392191e+00 2.23610282e-01
-6.68888032e-01 7.65378952e-01 -1.14672315e+00 5.59176922e-01
8.60192895e-01 -2.19845265e-01 -7.56699324e-01 -2.34802082e-01
2.89929092e-01 1.12617873e-01 -2.24167496e-01 -6.22615933e-01
2.98539549e-01 -5.63417077e-01 1.84499323e-01 -7.06837535e-01
2.06039131e-01 -9.54709589e-01 2.17758879e-01 1.04583688e-02
-4.72807705e-01 -3.29454422e-01 -2.18755990e-01 8.03186834e-01
-5.33585325e-02 -5.14227450e-01 3.50719690e-01 -4.09437507e-01
-8.34608912e-01 5.13117313e-01 -4.47339714e-01 -4.50386584e-01
7.29034007e-01 -2.22969636e-01 -1.54125363e-01 -7.54582644e-01
-9.49142575e-01 5.84957898e-01 1.24646030e-01 8.61880481e-01
5.52216947e-01 -1.38516998e+00 -5.60266674e-01 -4.14741077e-02
-3.45963687e-02 -3.46949175e-02 7.79452503e-01 7.25111127e-01
3.35252881e-01 3.20378691e-01 -1.18252756e-02 -9.20371488e-02
-1.35141993e+00 7.06355512e-01 1.05043903e-01 -3.62709403e-01
-5.46137571e-01 4.62620020e-01 3.65424246e-01 -4.01364326e-01
3.76707464e-01 -7.27767795e-02 -1.32803595e+00 3.53670269e-01
8.05103064e-01 3.33142489e-01 -1.56927288e-01 -4.65275079e-01
-1.92086156e-02 4.54538673e-01 -4.29596841e-01 1.80258140e-01
1.49242282e+00 -7.03120410e-01 -1.88882068e-01 7.07746327e-01
9.34507191e-01 1.24662444e-01 -7.45283842e-01 -5.69281161e-01
-3.90318841e-01 -5.10938048e-01 4.90367115e-02 -7.55616307e-01
-1.45640409e+00 6.42949581e-01 2.62758464e-01 6.24003887e-01
1.29440367e+00 -3.51401031e-01 1.12442899e+00 2.96596587e-01
4.21465486e-01 -1.17901993e+00 -1.58366814e-01 4.72055584e-01
6.95175648e-01 -1.04112351e+00 -3.47733907e-02 -4.91342336e-01
-7.56233513e-01 8.72855127e-01 1.08586872e+00 1.71122357e-01
7.23803520e-01 -4.36241090e-01 -1.03169926e-01 2.58171141e-01
-8.83043349e-01 -3.41446161e-01 2.89444655e-01 2.25875139e-01
5.98323584e-01 -1.73391879e-01 -9.99880433e-01 1.07091570e+00
2.27310270e-01 1.58497483e-01 7.73036182e-01 6.45811677e-01
-5.76632440e-01 -1.44776356e+00 -1.30311042e-01 7.92901278e-01
-5.43042183e-01 3.42498794e-02 1.19879685e-01 2.64859110e-01
1.07840344e-01 1.29545462e+00 -1.06764566e-02 -6.15481853e-01
4.14479345e-01 -1.33431882e-01 -1.14352822e-01 -5.98287225e-01
-7.03929484e-01 4.80900049e-01 7.14394301e-02 -4.68327612e-01
-6.66551113e-01 -4.32435691e-01 -1.16861713e+00 -4.88696724e-01
-6.07184052e-01 5.52503824e-01 2.96837449e-01 1.09822297e+00
2.94156700e-01 8.14791799e-01 1.03743410e+00 -5.92399657e-01
-4.19587612e-01 -1.20802307e+00 -8.60158741e-01 5.15966117e-01
-2.84596384e-02 -5.10742486e-01 -6.01703882e-01 -4.03904200e-01] | [10.143501281738281, 5.6048688888549805] |
f79bc62f-8924-4e60-adfc-45edfa36191e | bridging-the-gap-between-language-models-and | 2204.05210 | null | https://arxiv.org/abs/2204.05210v1 | https://arxiv.org/pdf/2204.05210v1.pdf | Bridging the Gap between Language Models and Cross-Lingual Sequence Labeling | Large-scale cross-lingual pre-trained language models (xPLMs) have shown effectiveness in cross-lingual sequence labeling tasks (xSL), such as cross-lingual machine reading comprehension (xMRC) by transferring knowledge from a high-resource language to low-resource languages. Despite the great success, we draw an empirical observation that there is a training objective gap between pre-training and fine-tuning stages: e.g., mask language modeling objective requires local understanding of the masked token and the span-extraction objective requires global understanding and reasoning of the input passage/paragraph and question, leading to the discrepancy between pre-training and xMRC. In this paper, we first design a pre-training task tailored for xSL named Cross-lingual Language Informative Span Masking (CLISM) to eliminate the objective gap in a self-supervised manner. Second, we present ContrAstive-Consistency Regularization (CACR), which utilizes contrastive learning to encourage the consistency between representations of input parallel sequences via unsupervised cross-lingual instance-wise training signals during pre-training. By these means, our methods not only bridge the gap between pretrain-finetune, but also enhance PLMs to better capture the alignment between different languages. Extensive experiments prove that our method achieves clearly superior results on multiple xSL benchmarks with limited pre-training data. Our methods also surpass the previous state-of-the-art methods by a large margin in few-shot data settings, where only a few hundred training examples are available. | ['Daxin Jiang', 'Jian Pei', 'Ming Gong', 'Linjun Shou', 'Nuo Chen'] | 2022-04-11 | null | https://aclanthology.org/2022.naacl-main.139 | https://aclanthology.org/2022.naacl-main.139.pdf | naacl-2022-7 | ['machine-reading-comprehension'] | ['natural-language-processing'] | [ 5.56648791e-01 4.76903245e-02 -5.67252815e-01 -6.12377465e-01
-1.26521385e+00 -6.33056760e-01 5.17391026e-01 8.18079486e-02
-7.07230926e-01 6.20607913e-01 4.49745685e-01 -7.03685224e-01
3.08134884e-01 -4.12422776e-01 -1.17992365e+00 -2.40064442e-01
4.29157168e-01 4.12432522e-01 1.50556847e-01 -3.81388515e-01
-8.50282162e-02 -1.07577972e-01 -1.34750974e+00 7.58424163e-01
1.36829412e+00 7.14631855e-01 5.74015796e-01 2.85992563e-01
-5.75527370e-01 9.40448761e-01 -2.83925772e-01 -2.94704854e-01
-8.56388882e-02 -6.83317840e-01 -1.12319398e+00 2.58475170e-02
6.48925662e-01 1.17784977e-01 6.31383210e-02 8.26775432e-01
4.14470464e-01 2.26756886e-01 4.11803335e-01 -8.60252380e-01
-9.02779639e-01 1.22886443e+00 -7.26513267e-01 2.30501518e-01
2.60196805e-01 3.65632892e-01 1.31782305e+00 -9.48349714e-01
5.72407722e-01 1.43025470e+00 6.58242881e-01 7.08429635e-01
-1.28162742e+00 -5.99722683e-01 5.08511782e-01 4.84198004e-01
-1.19714653e+00 -6.53386772e-01 6.94371998e-01 -2.46218368e-01
1.25976479e+00 2.37198576e-01 5.38720526e-02 1.41944230e+00
-7.31621236e-02 1.24896276e+00 1.48656881e+00 -1.02535999e+00
-1.57725047e-02 5.08324802e-02 3.50333780e-01 7.54886568e-01
-1.57262474e-01 -1.73711926e-01 -7.70595133e-01 3.22920233e-01
2.38987193e-01 -5.09630620e-01 -4.42912281e-01 4.48470265e-02
-1.35823548e+00 7.58340299e-01 -8.96026343e-02 5.51013172e-01
8.91776290e-05 -1.81472093e-01 9.40347910e-01 5.86161494e-01
4.29212302e-01 4.71220821e-01 -9.72750247e-01 -4.95436788e-02
-9.19978857e-01 -3.13083082e-01 4.36797231e-01 1.07535422e+00
8.50354075e-01 1.14830874e-01 -5.31456232e-01 1.11107075e+00
2.60246247e-02 4.11124915e-01 8.83144617e-01 -5.67284763e-01
1.05277324e+00 4.73245323e-01 -4.44397658e-01 -2.28562221e-01
-2.40943521e-01 -5.37210822e-01 -8.09261978e-01 -3.60754728e-01
4.98336494e-01 6.19989298e-02 -7.50678003e-01 2.24162531e+00
1.15654416e-01 9.19557139e-02 2.24485055e-01 5.50447285e-01
6.40471816e-01 8.01976025e-01 2.84354299e-01 -3.02373946e-01
1.64682233e+00 -1.31600666e+00 -8.15652490e-01 -6.71844602e-01
1.13552141e+00 -7.17420816e-01 2.23454738e+00 2.03269839e-01
-1.14879405e+00 -9.23268318e-01 -1.07954931e+00 -5.41895211e-01
-3.36929351e-01 3.13055903e-01 2.05791846e-01 3.15905303e-01
-5.45122027e-01 3.03938359e-01 -7.35524237e-01 -2.40166217e-01
1.35278448e-01 -2.56619602e-01 -1.50728285e-01 -4.16106552e-01
-1.60849452e+00 1.11822045e+00 7.39555895e-01 -1.99732855e-01
-7.09951162e-01 -1.15728354e+00 -1.03254008e+00 4.19812687e-02
7.34297991e-01 -3.94153804e-01 1.15367520e+00 -1.01278520e+00
-1.62600219e+00 1.29488635e+00 -2.81865209e-01 -4.48762983e-01
5.94765961e-01 -6.61672831e-01 -4.22027349e-01 -1.78708643e-01
2.34038189e-01 6.42465353e-01 6.11198187e-01 -1.07452726e+00
-5.08385181e-01 -2.84377635e-02 -1.90898374e-01 2.17949808e-01
-2.03186184e-01 1.79363072e-01 -5.41474640e-01 -7.70357728e-01
-2.37849787e-01 -6.67775035e-01 1.28151551e-01 -4.86856163e-01
-4.50963765e-01 -7.06640363e-01 3.98225427e-01 -8.76032472e-01
1.31764281e+00 -1.88918066e+00 1.12727284e-01 -2.13428706e-01
-2.42542401e-01 3.94598335e-01 -6.27291262e-01 5.13500214e-01
-7.65808970e-02 3.51056643e-02 -4.35089886e-01 -6.94902182e-01
1.85633540e-01 4.33149219e-01 -6.30114317e-01 3.24963868e-01
3.50722939e-01 1.10412478e+00 -1.05948114e+00 -6.44650519e-01
1.58306910e-03 9.01671350e-02 -3.85027528e-01 4.14126396e-01
-6.68158293e-01 5.35713494e-01 -6.87778071e-02 3.02381039e-01
3.94606739e-01 -4.78702307e-01 3.23612124e-01 -1.55080244e-01
-1.59378663e-01 1.04706156e+00 -7.10185230e-01 2.32968235e+00
-8.81505549e-01 4.07405704e-01 -3.38526249e-01 -1.18954182e+00
7.38539577e-01 2.51772553e-01 6.15119077e-02 -9.87761199e-01
4.25950959e-02 2.81245410e-01 2.01987363e-02 -6.59109950e-01
3.68620247e-01 -1.75565884e-01 -2.17147291e-01 7.19294310e-01
2.56338239e-01 1.41227931e-01 4.61496949e-01 1.05209500e-01
8.38971257e-01 5.42117059e-01 4.44019228e-01 -4.01972592e-01
8.44046950e-01 -6.43278733e-02 6.06051683e-01 7.75153100e-01
7.44091868e-02 3.63687187e-01 3.81969184e-01 2.75627915e-02
-1.00255787e+00 -8.61993670e-01 -3.72811295e-02 1.78728139e+00
-1.34119615e-01 -4.65890884e-01 -8.85112286e-01 -9.47138131e-01
-1.83827132e-01 1.22738743e+00 -5.39986432e-01 -1.74140111e-01
-9.54934716e-01 -4.68485802e-01 8.80098581e-01 5.86498320e-01
4.32893395e-01 -1.21308863e+00 -3.08724344e-01 2.58093625e-01
-4.91776109e-01 -1.63458896e+00 -8.05285275e-01 2.65917957e-01
-4.66170937e-01 -7.84677267e-01 -3.66303325e-01 -9.88781929e-01
5.55507302e-01 6.00522347e-02 1.61866677e+00 8.67337286e-02
-1.51242748e-01 1.32432252e-01 -5.38601339e-01 -5.42522185e-02
-7.58405328e-01 5.04291475e-01 1.14304423e-01 -3.71095492e-03
5.70805550e-01 -3.60564351e-01 -1.26584634e-01 2.71324754e-01
-8.18749249e-01 5.23780644e-01 5.60048878e-01 8.70208859e-01
8.12657833e-01 -3.79426569e-01 7.71647036e-01 -1.14537871e+00
5.18349469e-01 -4.30553228e-01 -5.76914489e-01 9.07048583e-01
-5.39815605e-01 4.83114958e-01 1.04633439e+00 -6.19013727e-01
-1.20537233e+00 -3.70614767e-01 -1.66721269e-01 -2.66540587e-01
-1.35704011e-01 8.41955185e-01 -4.77671146e-01 5.10015965e-01
6.62500501e-01 5.10156274e-01 -1.41306192e-01 -7.44297564e-01
7.13670790e-01 4.55850750e-01 6.58545315e-01 -1.06983376e+00
5.82312763e-01 2.07715966e-02 -5.54367363e-01 -6.27767801e-01
-1.44053507e+00 -3.13675374e-01 -8.65316331e-01 1.94990218e-01
8.40681791e-01 -1.08900785e+00 -4.07776326e-01 3.47936720e-01
-1.21928144e+00 -7.01673150e-01 -3.45626444e-01 3.88692468e-01
-5.29050887e-01 4.54788744e-01 -7.37213492e-01 -3.89673024e-01
-3.53032917e-01 -1.03284991e+00 8.80910695e-01 -2.04250738e-01
-3.80561143e-01 -1.16430116e+00 9.30047557e-02 5.76329529e-01
3.42853457e-01 -2.21219629e-01 1.34657371e+00 -7.50356555e-01
-4.46201235e-01 4.13535953e-01 -2.20484361e-01 6.61764383e-01
2.57995188e-01 -4.20221239e-01 -9.31923032e-01 -4.01526749e-01
-1.18162706e-02 -9.61385012e-01 8.76905859e-01 6.79431036e-02
1.35927498e+00 -3.68394285e-01 1.06972866e-02 5.86197436e-01
1.41776264e+00 -3.54885340e-01 4.90436673e-01 3.22285473e-01
9.07686353e-01 8.17341149e-01 5.98345101e-01 -1.02942429e-01
6.96685731e-01 6.37527168e-01 -2.26334199e-01 -1.92334399e-01
-4.97407407e-01 -6.32732093e-01 7.08305597e-01 1.60070157e+00
3.38238180e-01 -1.93021551e-01 -9.93173361e-01 6.13719821e-01
-1.74697542e+00 -7.08930075e-01 1.57634035e-01 2.04266214e+00
1.70106936e+00 1.44449234e-01 -2.21169129e-01 -1.55058816e-01
4.62247819e-01 2.89828062e-01 -6.57825232e-01 -3.12502146e-01
-3.80580723e-01 2.73519725e-01 4.06441927e-01 6.55238569e-01
-7.80494153e-01 1.53788149e+00 5.11382484e+00 1.37368023e+00
-1.18279970e+00 4.96965677e-01 5.44294417e-01 9.25274193e-02
-5.98623216e-01 -3.42337787e-02 -1.15059745e+00 5.04106224e-01
9.94608879e-01 -5.55524901e-02 4.63235438e-01 6.88591242e-01
2.00727686e-01 1.57294556e-01 -1.33148623e+00 7.27817297e-01
2.80961514e-01 -1.31005824e+00 1.14116389e-02 -4.39399779e-01
8.02056015e-01 3.40306878e-01 -1.49160057e-01 6.69010818e-01
3.79994273e-01 -9.46996689e-01 8.03760171e-01 2.65969992e-01
1.13606060e+00 -5.88319778e-01 3.84479523e-01 6.87854946e-01
-1.13002062e+00 3.97455722e-01 -3.66943181e-01 1.44495443e-01
4.38070446e-01 4.41629469e-01 -6.38331354e-01 6.20250106e-01
3.86843324e-01 6.46263361e-01 -6.00592554e-01 2.71594882e-01
-7.42262125e-01 1.12458229e+00 1.13815255e-02 1.29909456e-01
4.32125509e-01 -1.56549305e-01 2.83531159e-01 1.63830328e+00
-8.88963640e-02 -2.21385121e-01 3.78362209e-01 1.01350045e+00
-2.10410848e-01 4.32549655e-01 -2.23522738e-01 -6.31329715e-02
4.26746517e-01 7.94803560e-01 -2.96857327e-01 -5.65438807e-01
-6.62730694e-01 9.33044434e-01 9.22920942e-01 3.25690746e-01
-8.62840533e-01 -9.24301744e-02 4.65396643e-01 -6.07410446e-03
3.07984650e-01 -3.67650747e-01 -4.96224433e-01 -1.51451826e+00
1.64207816e-01 -1.42755735e+00 3.44258785e-01 -4.73436296e-01
-1.53893507e+00 6.92728877e-01 1.60757452e-02 -1.04883301e+00
-2.60034800e-01 -6.03653669e-01 -3.37549865e-01 9.64340925e-01
-2.03434634e+00 -1.51594281e+00 1.48777008e-01 4.46933627e-01
1.11962140e+00 -2.91126609e-01 6.78277016e-01 2.66159654e-01
-6.83195770e-01 1.01966166e+00 8.35948065e-02 2.25135058e-01
8.10589492e-01 -1.14071655e+00 6.15210891e-01 1.20796728e+00
3.02661270e-01 7.87906110e-01 4.03682888e-01 -5.60127676e-01
-1.32492769e+00 -1.28143096e+00 1.26837397e+00 -4.38422024e-01
8.56232941e-01 -7.10011423e-01 -1.51686299e+00 9.94042754e-01
5.12369871e-01 -9.89423618e-02 7.81803191e-01 4.27890450e-01
-7.85927832e-01 -7.04995415e-04 -3.92235011e-01 7.30880678e-01
1.19176340e+00 -1.02722919e+00 -9.74924445e-01 3.96451980e-01
1.13699353e+00 -4.18656230e-01 -6.76864207e-01 4.52642620e-01
1.79676652e-01 -5.96095264e-01 7.63008356e-01 -9.01705325e-01
6.32079303e-01 -2.10600689e-01 -1.99117482e-01 -1.38732207e+00
-1.34105593e-01 -6.84746146e-01 4.17038351e-02 1.60472667e+00
6.61974013e-01 -4.88397181e-01 1.75025225e-01 7.87423253e-02
-5.48194230e-01 -7.75548935e-01 -8.14446628e-01 -1.11152577e+00
4.75783676e-01 -6.87262774e-01 4.95541722e-01 1.18041611e+00
8.55288282e-02 8.45720947e-01 -4.28373605e-01 -3.29038016e-02
5.34835458e-01 1.55466929e-01 5.78442574e-01 -5.87109506e-01
-5.50661922e-01 -4.64766085e-01 3.98417026e-01 -1.50029814e+00
8.33749533e-01 -1.29658043e+00 3.02094817e-01 -1.12110233e+00
9.96584594e-02 -6.67809665e-01 -4.57794100e-01 7.58948624e-01
-6.16987169e-01 -2.37636700e-01 2.27188114e-02 4.86565381e-02
-7.40281641e-01 6.71999037e-01 1.20788157e+00 -7.70337805e-02
-8.55627283e-02 -3.45752686e-01 -5.39879918e-01 4.72789466e-01
7.03103483e-01 -4.04994547e-01 -5.26802182e-01 -9.01057184e-01
2.14139372e-01 -2.07148567e-01 -1.18984334e-01 -7.10417032e-01
6.53232262e-02 -2.40778074e-01 -1.05488114e-01 -5.27404904e-01
-3.81080508e-02 -2.97706872e-01 -4.56532121e-01 2.75293440e-01
-9.20499861e-01 -3.89172249e-02 2.95519501e-01 1.84275508e-01
-1.97781295e-01 -2.97071993e-01 8.83837700e-01 -2.06330776e-01
-8.99392962e-01 2.82874368e-02 -4.53469902e-02 8.34371150e-01
4.40048903e-01 1.99683294e-01 -5.19735157e-01 -1.45489136e-02
-1.79189295e-01 4.39123720e-01 3.27587366e-01 7.86971092e-01
2.17413723e-01 -1.15876508e+00 -1.04343820e+00 3.25812399e-01
4.08948541e-01 2.37270854e-02 2.11678088e-01 9.04620647e-01
-4.11265232e-02 5.38443327e-01 -6.73965141e-02 -6.59205914e-01
-1.06833673e+00 6.54621243e-01 6.02660887e-02 -7.81203806e-01
-6.11813843e-01 9.80687499e-01 3.89320374e-01 -8.11350286e-01
3.50359499e-01 -5.52380860e-01 -4.38841945e-03 -1.73441228e-02
6.30621076e-01 9.25420821e-02 1.76102221e-01 -5.70592463e-01
-2.89643168e-01 6.63847685e-01 -4.86980140e-01 -2.56575998e-02
1.02629399e+00 -4.29235786e-01 -9.97672454e-02 7.56245732e-01
1.33518183e+00 1.49795115e-01 -1.33758104e+00 -8.80408764e-01
4.10064399e-01 -2.55975053e-02 -2.15829998e-01 -9.23637450e-01
-5.93079448e-01 1.08081603e+00 -1.32213868e-02 -2.14847669e-01
8.73766780e-01 2.01972738e-01 1.13405120e+00 1.60710961e-01
1.50703624e-01 -1.36328316e+00 1.77312478e-01 9.29794729e-01
9.41540897e-01 -1.41818964e+00 -1.71809360e-01 -3.53051901e-01
-9.98431027e-01 7.90204942e-01 8.95823538e-01 1.30435035e-01
5.58665991e-02 1.16686113e-01 2.05651730e-01 1.79300249e-01
-1.09334040e+00 -3.11845481e-01 5.53018868e-01 2.79793531e-01
6.83270574e-01 2.61409767e-02 -2.74714261e-01 6.85217559e-01
-3.92992169e-01 -1.26212135e-01 3.51178236e-02 7.54961014e-01
-3.74171704e-01 -1.20174360e+00 -9.53876600e-02 1.59674585e-02
-3.10773522e-01 -5.91682792e-01 6.92062899e-02 7.45550632e-01
1.36216730e-01 6.82639658e-01 -1.26294531e-02 -2.79682782e-03
2.38765150e-01 4.82408643e-01 5.64080358e-01 -6.57552540e-01
-4.76372331e-01 3.16889696e-02 1.57165483e-01 -4.41875249e-01
-2.39398479e-01 -4.79673266e-01 -1.39966416e+00 1.17925420e-01
-2.08411247e-01 7.80900344e-02 3.31514060e-01 1.51578712e+00
2.62369692e-01 6.54025435e-01 4.90550697e-01 -1.98436216e-01
-8.62369478e-01 -1.14776480e+00 -1.30811989e-01 4.76661503e-01
2.95162231e-01 -3.00421715e-01 -4.05443817e-01 3.77431899e-01] | [11.012413024902344, 9.383408546447754] |
e5cf9241-1d63-476c-8b71-251c2027ffca | a-3d-shape-similarity-based-contrastive | 2211.02130 | null | https://arxiv.org/abs/2211.02130v1 | https://arxiv.org/pdf/2211.02130v1.pdf | A 3D-Shape Similarity-based Contrastive Approach to Molecular Representation Learning | Molecular shape and geometry dictate key biophysical recognition processes, yet many graph neural networks disregard 3D information for molecular property prediction. Here, we propose a new contrastive-learning procedure for graph neural networks, Molecular Contrastive Learning from Shape Similarity (MolCLaSS), that implicitly learns a three-dimensional representation. Rather than directly encoding or targeting three-dimensional poses, MolCLaSS matches a similarity objective based on Gaussian overlays to learn a meaningful representation of molecular shape. We demonstrate how this framework naturally captures key aspects of three-dimensionality that two-dimensional representations cannot and provides an inductive framework for scaffold hopping. | ['Kangway V. Chuang', 'Gabriele Scalia', 'Tommaso Biancalani', 'Ziqing Lu', 'Nathaniel L. Diamant', 'Austin Atsango'] | 2022-11-03 | null | null | null | null | ['molecular-property-prediction'] | ['miscellaneous'] | [ 6.31539285e-01 1.10433549e-01 -6.64256454e-01 -4.04319972e-01
-6.13567472e-01 -7.15868175e-01 6.09739840e-01 7.09197760e-01
-5.50697707e-02 9.87928450e-01 4.29931432e-02 -8.45224082e-01
-3.57492954e-01 -7.38664806e-01 -9.95455563e-01 -7.28172719e-01
-5.42696059e-01 5.63098311e-01 -3.27389017e-02 -6.06038570e-02
4.94776636e-01 1.16106296e+00 -9.51407731e-01 2.37584084e-01
6.76234365e-01 6.80853307e-01 -1.35992438e-01 8.21416914e-01
-1.82992935e-01 5.28606474e-01 -1.72583744e-01 -1.97970122e-01
3.99388894e-02 -4.98992801e-01 -7.19373882e-01 -4.60444838e-01
1.00010884e+00 2.52596587e-01 -5.81887782e-01 8.66554379e-01
6.35141909e-01 3.49613905e-01 1.54013085e+00 -8.61313939e-01
-9.38883185e-01 4.45934152e-03 -9.45363287e-03 1.72349170e-01
6.52571499e-01 2.79283851e-01 1.31635809e+00 -9.24929976e-01
9.08485889e-01 1.29511023e+00 8.97287369e-01 4.01256502e-01
-1.88473463e+00 -3.44895422e-01 5.14691360e-02 5.97088709e-02
-1.22819567e+00 -2.24169880e-01 1.03820741e+00 -6.75618052e-01
1.12957335e+00 3.89234304e-01 6.75617456e-01 1.10317969e+00
6.95146024e-01 4.31656003e-01 9.15515780e-01 -1.62622333e-02
4.56492215e-01 -4.75859106e-01 2.41958320e-01 9.70017493e-01
3.87678832e-01 3.91436309e-01 -4.41039115e-01 -7.96546876e-01
7.85341382e-01 2.31668890e-01 -2.37650514e-01 -1.16006303e+00
-7.48304486e-01 9.64881301e-01 7.41787970e-01 -1.98857244e-02
-2.38106236e-01 3.51475805e-01 1.26625627e-01 2.39487857e-01
1.97066858e-01 1.10070789e+00 -4.48930740e-01 1.24308184e-01
-5.24743497e-01 2.42175743e-01 9.24689651e-01 7.84042001e-01
9.26514149e-01 2.79883713e-01 1.35493651e-01 2.62618035e-01
3.09647620e-01 3.07220519e-01 3.11264414e-02 -5.90716064e-01
-1.19454034e-01 6.14126861e-01 -2.61007756e-01 -1.14362562e+00
-6.56934619e-01 -1.63586721e-01 -8.65900815e-01 2.93754607e-01
2.48589084e-01 3.87109071e-01 -1.23787177e+00 1.68868017e+00
1.48224533e-01 3.92509513e-02 1.58260196e-01 5.24838150e-01
1.22346473e+00 4.78154391e-01 4.97271508e-01 -2.27325916e-01
5.81955671e-01 -4.26028430e-01 -2.11319953e-01 2.99704552e-01
9.13484156e-01 -3.84078801e-01 8.14810276e-01 3.38497728e-01
-1.03577518e+00 -3.69917333e-01 -1.38385868e+00 -9.42622032e-03
-8.17712605e-01 -5.70077300e-01 1.21045458e+00 6.78403318e-01
-8.03971052e-01 1.47104752e+00 -5.55407286e-01 -1.97955407e-02
6.15232527e-01 7.16693699e-01 -6.51090860e-01 -4.04004417e-02
-9.82318401e-01 8.71433735e-01 6.81110382e-01 -4.39981222e-01
-9.89324749e-01 -8.68285120e-01 -9.53798473e-01 -3.09948385e-01
3.66631262e-02 -7.86855519e-01 6.82036638e-01 -4.90745813e-01
-1.51937103e+00 7.69773185e-01 -2.85955042e-01 -3.20586354e-01
-1.01477066e-02 3.72609317e-01 -2.68015832e-01 1.07934140e-01
-3.08422387e-01 6.60346806e-01 7.04094350e-01 -1.29771793e+00
3.75065804e-01 -4.90877390e-01 -4.29574139e-02 9.10034031e-02
1.19787261e-01 -6.01071537e-01 1.63285777e-01 -7.32161999e-01
5.03939092e-01 -8.21288586e-01 -6.55195773e-01 1.92747384e-01
-6.13075495e-01 -4.65151221e-02 5.99233389e-01 -1.76987335e-01
8.17451417e-01 -1.49423242e+00 6.23743117e-01 7.53315389e-01
8.06784809e-01 1.31991565e-01 -4.94252980e-01 6.39932692e-01
-5.62611878e-01 4.66067851e-01 -3.64838988e-01 4.85451102e-01
-1.05517179e-01 2.29279086e-01 -1.89742565e-01 5.93744338e-01
4.37161326e-01 1.19052565e+00 -1.09681964e+00 -6.97611496e-02
3.10705174e-02 6.90566897e-01 -6.82505429e-01 1.78304240e-01
-5.96978843e-01 4.26531911e-01 -5.82589447e-01 9.07301128e-01
6.55323029e-01 -5.30444086e-01 6.36568129e-01 -4.21062112e-01
2.14501813e-01 3.87469143e-01 -7.12689161e-01 1.66089165e+00
2.82287300e-01 2.21275851e-01 -5.09429634e-01 -1.15675449e+00
1.22236168e+00 -1.21572837e-02 6.24223650e-01 -6.94659770e-01
-1.13016471e-01 1.97654173e-01 3.99787501e-02 -8.49363729e-02
4.02342007e-02 -2.01987073e-01 1.09973222e-01 2.53615677e-01
2.03232914e-01 -6.14671052e-01 -1.67808846e-01 1.39988452e-01
1.02641916e+00 3.16951215e-01 5.46867073e-01 -2.74566829e-01
3.38263839e-01 -8.69111195e-02 3.89278352e-01 7.78744340e-01
-3.62060964e-02 4.50217813e-01 6.13748491e-01 -1.04356921e+00
-1.05940187e+00 -1.54748666e+00 -1.05182998e-01 1.05208635e+00
3.90716121e-02 -6.30691111e-01 -3.04111987e-01 -6.97347760e-01
5.31362712e-01 2.59738952e-01 -7.38822341e-01 -5.64489186e-01
-4.99634087e-01 -5.94938159e-01 5.71580708e-01 3.08581769e-01
-4.90506411e-01 -6.80066884e-01 2.07376078e-01 2.44725898e-01
8.18085134e-01 -4.41865176e-01 -5.31978369e-01 9.16263580e-01
-1.05430961e+00 -1.56305742e+00 -5.46022713e-01 -8.04408550e-01
5.96756220e-01 1.91327542e-01 1.24822664e+00 -1.13673899e-02
-6.93644643e-01 2.05635488e-01 1.46419227e-01 -2.92632610e-01
-5.78886330e-01 1.77798495e-02 2.68196821e-01 -3.78514260e-01
3.37906748e-01 -1.03782940e+00 -5.34934461e-01 8.66494477e-02
-5.99839509e-01 -2.22487479e-01 6.17375612e-01 8.29667568e-01
1.16180825e+00 -5.94819844e-01 5.76404095e-01 -9.22079384e-01
8.68142545e-01 -2.64128655e-01 -3.48002225e-01 3.11897635e-01
-5.80460131e-01 4.85518545e-01 7.22808361e-01 -5.56586325e-01
-1.74903780e-01 3.26990694e-01 -3.30803469e-02 -5.21821499e-01
-2.81704903e-01 5.71668446e-01 -1.69537753e-01 -7.96537578e-01
9.56120908e-01 4.32457715e-01 9.41302702e-02 -3.71600211e-01
6.36975825e-01 8.43900666e-02 2.88233370e-01 -9.26498771e-01
7.17719257e-01 1.85914468e-02 8.65239918e-01 -1.04133010e+00
-3.45433295e-01 -1.29793748e-01 -9.63441849e-01 2.97545969e-01
8.76843989e-01 -5.44220686e-01 -1.35229099e+00 -1.41050026e-01
-1.11646378e+00 -1.72879532e-01 -1.86289296e-01 4.46836770e-01
-9.15999949e-01 5.64102411e-01 -3.94849211e-01 -4.85075146e-01
-2.46830970e-01 -1.16247857e+00 8.24825287e-01 -1.25830218e-01
-4.57202941e-01 -1.20713317e+00 4.51083422e-01 -2.04932883e-01
1.55401811e-01 7.08595634e-01 1.58365965e+00 -8.71381938e-01
-6.52375102e-01 -8.30256790e-02 -5.45878932e-02 -2.86652833e-01
2.98495233e-01 2.66564079e-03 -7.68830717e-01 -3.91413152e-01
-7.74730921e-01 -6.03928387e-01 9.51190174e-01 5.40695429e-01
1.34435415e+00 -3.38024050e-01 -5.88097870e-01 9.64452744e-01
1.19302189e+00 4.65766728e-01 3.77011180e-01 -1.79500863e-01
1.04686940e+00 3.73520315e-01 -7.80400559e-02 7.58299306e-02
3.91694121e-02 5.08470416e-01 4.15400177e-01 -4.04783547e-01
7.83773512e-02 -7.57562220e-01 -2.88310070e-02 5.90697885e-01
-3.83882791e-01 -9.96049792e-02 -8.81723344e-01 -2.73155302e-01
-1.54563880e+00 -1.00085843e+00 2.03053191e-01 2.11984277e+00
1.05320120e+00 1.84514627e-01 2.56540984e-01 -2.19287679e-01
4.68146175e-01 3.95265281e-01 -1.31848609e+00 -6.84512496e-01
-4.62694347e-01 4.88095999e-01 4.95003670e-01 6.52200878e-01
-1.05086863e+00 1.03079569e+00 8.54713249e+00 7.62569010e-01
-1.07566988e+00 -6.53918624e-01 6.27222180e-01 3.58518481e-01
-8.92343521e-01 -1.04916366e-02 -4.22858208e-01 -4.36292849e-02
9.41608071e-01 -1.43544719e-01 3.78738701e-01 7.81310678e-01
-1.25915930e-02 6.87520683e-01 -1.58337271e+00 1.15071547e+00
-1.91021010e-01 -2.31161094e+00 9.38653827e-01 1.77938566e-01
5.82126558e-01 -9.38031971e-02 1.71378270e-01 -8.33221525e-02
4.64903295e-01 -1.70526934e+00 1.17772825e-01 7.74441063e-01
9.90364671e-01 -7.71018386e-01 -1.72047466e-01 -1.43623576e-01
-1.31468570e+00 3.80235434e-01 -5.16685426e-01 5.59547264e-03
-1.79966524e-01 8.89791623e-02 -9.42799449e-01 3.68246019e-01
-8.40772968e-03 1.07499599e+00 -4.15980011e-01 9.33161318e-01
2.83259191e-02 1.87487811e-01 -4.56662895e-03 -2.43291050e-01
3.81892920e-01 -5.39431989e-01 5.54811776e-01 1.20046854e+00
-1.92233756e-01 2.03348249e-01 5.12382627e-01 1.10685039e+00
-3.75786066e-01 1.49371147e-01 -1.27074432e+00 -6.76036596e-01
4.18931991e-01 5.67698240e-01 -5.11092842e-01 -9.10125896e-02
-2.12317988e-01 6.90446019e-01 4.59096611e-01 6.39349580e-01
-4.32456613e-01 -5.63386142e-01 1.03267050e+00 -6.19284324e-02
1.45561397e-01 -5.67014992e-01 -3.65180038e-02 -8.25687885e-01
-4.61885303e-01 -9.68773544e-01 2.63592124e-01 -5.95287800e-01
-1.43356323e+00 2.75206715e-01 -3.49913895e-01 -9.11104381e-01
-2.09669527e-02 -1.24376476e+00 -7.00485587e-01 8.45222950e-01
-1.37300551e+00 -1.02626121e+00 2.68108517e-01 3.49689096e-01
-4.79648635e-02 -3.07368785e-01 1.33103979e+00 -2.48532861e-01
-3.31810504e-01 7.41359949e-01 4.29218948e-01 -2.70939231e-01
5.59509039e-01 -1.48164928e+00 8.16715360e-01 -2.47132614e-01
2.92027265e-01 9.37338471e-01 6.81639493e-01 -8.09744298e-01
-1.99866414e+00 -9.96767938e-01 3.91377240e-01 -5.71094275e-01
5.43570340e-01 -2.36396536e-01 -1.08779442e+00 3.29735488e-01
-4.89809036e-01 1.35818005e-01 1.30915093e+00 1.93232194e-01
-9.85369503e-01 3.55665594e-01 -1.09240472e+00 6.84299588e-01
1.52246344e+00 -1.10238874e+00 -3.95214856e-01 5.49015403e-01
9.29106295e-01 -6.43386766e-02 -1.46351528e+00 5.08647740e-01
6.72220409e-01 -4.31748927e-01 1.45308149e+00 -1.82196510e+00
-2.87804659e-02 -1.81384191e-01 -4.77278471e-01 -1.19646180e+00
-5.69088221e-01 -1.00320804e+00 -4.43701416e-01 -2.97244526e-02
7.37158000e-01 -6.59633696e-01 1.24631119e+00 4.27933544e-01
-2.43719116e-01 -1.06429362e+00 -8.11813891e-01 -9.26252246e-01
5.59806585e-01 -6.67111352e-02 5.06659925e-01 1.18604720e+00
8.90124068e-02 5.78124285e-01 -1.18600719e-01 -9.08891708e-02
6.80591822e-01 2.19880968e-01 6.36232316e-01 -1.64546728e+00
-3.47708434e-01 -7.04059362e-01 -9.50944722e-01 -1.24430680e+00
3.12947124e-01 -1.42791784e+00 -5.68922222e-01 -1.15886176e+00
3.71272296e-01 -1.85472235e-01 -5.30232728e-01 3.50700438e-01
2.00199097e-01 9.06987190e-02 -2.63382941e-01 7.05955699e-02
-6.06148779e-01 8.24343383e-01 1.49583232e+00 -6.64541602e-01
-4.62372333e-01 -3.60692352e-01 -6.93243444e-01 6.00348294e-01
5.85055053e-01 -3.28488171e-01 -4.10868853e-01 1.32246971e-01
2.15810195e-01 6.48508742e-02 1.83244243e-01 -6.43726707e-01
6.35984764e-02 -5.97723186e-01 8.20763767e-01 -6.43409491e-01
6.21545613e-01 -3.65258664e-01 7.98148066e-02 6.01079524e-01
-6.23918295e-01 -6.68975711e-02 2.01990634e-01 1.04313946e+00
1.04710959e-01 2.22554296e-01 6.67729616e-01 -1.60448536e-01
-2.77492583e-01 9.49742079e-01 -3.94217581e-01 1.14750005e-01
6.46881342e-01 -8.23998749e-01 -2.59224683e-01 -1.72155097e-01
-1.12807739e+00 -2.44102091e-01 6.99872255e-01 1.38109952e-01
1.10081601e+00 -1.50707519e+00 -2.86561549e-01 2.77871668e-01
2.79692888e-01 -4.43795502e-01 -2.58948475e-01 2.33258292e-01
-5.07497370e-01 6.19862437e-01 -4.15486246e-01 -5.33392370e-01
-1.27079296e+00 7.03402281e-01 6.37105584e-01 1.54567748e-01
-1.56578436e-01 7.33682871e-01 3.59609783e-01 -6.64027274e-01
2.82489419e-01 -1.23772129e-01 -8.58791396e-02 -1.41217589e-01
2.21789151e-01 -3.00657600e-02 -9.59710032e-02 -5.76035023e-01
-4.31467712e-01 9.12673533e-01 -2.75974363e-01 5.34226954e-01
1.51438212e+00 5.78520238e-01 2.98296399e-02 2.46389911e-01
1.55827057e+00 -4.23915297e-01 -1.30718625e+00 -2.03154713e-01
2.04495892e-01 -1.82872862e-01 -6.54355958e-02 -5.36511660e-01
-3.65932584e-01 9.69905317e-01 5.39719880e-01 -1.40803114e-01
3.44360530e-01 1.05042672e-02 3.92976403e-01 1.40086269e+00
1.81874320e-01 -7.46808767e-01 5.19556344e-01 7.03419983e-01
8.88581634e-01 -1.09438848e+00 4.04150397e-01 -3.98376971e-01
-9.79264602e-02 1.48709035e+00 3.34210247e-01 -4.19053063e-02
7.55706012e-01 -1.42695740e-01 -3.77502680e-01 -7.58995354e-01
-5.22956729e-01 -1.64937362e-01 8.36480379e-01 1.14829814e+00
8.49182487e-01 9.45563316e-02 9.81757417e-02 2.02667728e-01
1.85066789e-01 -4.92576569e-01 3.34765688e-02 1.13428986e+00
-7.94230700e-01 -1.31859910e+00 7.29018003e-02 4.00630832e-01
-2.01105867e-02 -2.50174582e-01 -1.27844465e+00 6.13374233e-01
-2.83641040e-01 3.52374762e-01 -1.16782621e-01 -7.09175825e-01
3.29994142e-01 1.12253360e-01 8.99952590e-01 -6.95164979e-01
-9.67428759e-02 -1.92092031e-01 4.68481109e-02 -6.63157642e-01
-2.06698909e-01 -7.57066235e-02 -1.25821257e+00 -5.23537338e-01
-1.45381153e-01 3.83197397e-01 4.10856366e-01 6.00577593e-01
6.69036984e-01 2.36702487e-01 7.15417922e-01 -1.24166453e+00
-3.25635880e-01 -3.44464511e-01 -4.41210479e-01 3.70732844e-01
3.90911281e-01 -7.91082799e-01 -1.37776196e-01 -1.08027786e-01] | [5.139481067657471, 5.771726608276367] |
41fbcb82-a133-4cbd-8ab0-ae5cd70153e9 | bapose-bottom-up-pose-estimation-with | 2112.10716 | null | https://arxiv.org/abs/2112.10716v1 | https://arxiv.org/pdf/2112.10716v1.pdf | BAPose: Bottom-Up Pose Estimation with Disentangled Waterfall Representations | We propose BAPose, a novel bottom-up approach that achieves state-of-the-art results for multi-person pose estimation. Our end-to-end trainable framework leverages a disentangled multi-scale waterfall architecture and incorporates adaptive convolutions to infer keypoints more precisely in crowded scenes with occlusions. The multi-scale representations, obtained by the disentangled waterfall module in BAPose, leverage the efficiency of progressive filtering in the cascade architecture, while maintaining multi-scale fields-of-view comparable to spatial pyramid configurations. Our results on the challenging COCO and CrowdPose datasets demonstrate that BAPose is an efficient and robust framework for multi-person pose estimation, achieving significant improvements on state-of-the-art accuracy. | ['Andreas Savakis', 'Bruno Artacho'] | 2021-12-20 | null | null | null | null | ['multi-person-pose-estimation'] | ['computer-vision'] | [-3.90331388e-01 -2.46135935e-01 2.56856740e-01 -3.81891310e-01
-8.66412699e-01 -5.16937673e-01 4.39002514e-01 -3.08326602e-01
-7.05635965e-01 4.11373079e-01 6.59952998e-01 6.22740686e-01
-3.75764035e-02 -5.53323925e-01 -7.72234023e-01 -1.10808708e-01
-1.38652653e-01 7.43580997e-01 2.96782792e-01 -2.55712628e-01
-3.41253430e-01 3.99787903e-01 -1.63787496e+00 6.32657170e-01
5.86961567e-01 6.65842175e-01 -3.77604276e-01 1.11457300e+00
6.26205146e-01 4.91660535e-01 -3.04559767e-01 -1.24412668e+00
5.95952272e-01 3.38325828e-01 -5.02891541e-01 2.25534812e-01
1.56499612e+00 -8.17477465e-01 -9.79234457e-01 4.83151048e-01
8.41349781e-01 6.71367794e-02 3.88672978e-01 -1.11477637e+00
-7.07478106e-01 2.90848557e-02 -7.96703398e-01 3.74940008e-01
8.11878324e-01 5.74459076e-01 1.02585948e+00 -1.31567693e+00
4.28601027e-01 1.68410277e+00 1.17430079e+00 3.10374141e-01
-1.35104799e+00 -5.91880023e-01 4.76421028e-01 -4.41815853e-02
-1.54844379e+00 -4.66047555e-01 8.71552974e-02 -5.23590207e-01
1.07023084e+00 5.30597195e-02 8.86667252e-01 1.33989596e+00
1.12427194e-02 1.02613139e+00 9.14069116e-01 -8.62615928e-03
-4.12184834e-01 -3.82088721e-01 -3.42285074e-02 1.16200233e+00
7.18454480e-01 1.15242396e-02 -1.20185030e+00 -2.06479341e-01
9.71207559e-01 2.32868314e-01 2.38297856e-03 -6.12602592e-01
-1.23325515e+00 6.11152053e-01 6.22601926e-01 -4.30049390e-01
-3.88769299e-01 5.63501835e-01 1.72952875e-01 -2.56117761e-01
5.58885217e-01 2.57328600e-01 -5.61403811e-01 -2.64625609e-01
-1.01344156e+00 1.06920362e+00 5.47957063e-01 1.02341104e+00
3.88819367e-01 -4.79672581e-01 -5.21465123e-01 4.67602193e-01
3.11451018e-01 7.52690673e-01 -3.10038954e-01 -1.08880162e+00
9.54809606e-01 7.00686395e-01 5.18860042e-01 -8.29256117e-01
-5.15736878e-01 -7.86632657e-01 -3.57237697e-01 2.36535579e-01
8.34571004e-01 -2.48514190e-01 -9.83573020e-01 1.65082431e+00
4.86089617e-01 2.61749327e-01 -3.38351488e-01 1.08856702e+00
9.05237973e-01 4.91752438e-02 1.37525544e-01 6.90052271e-01
1.75273395e+00 -1.39975739e+00 -3.29490215e-01 -6.27587557e-01
1.87528245e-02 -4.91510034e-01 9.49514806e-01 3.34255427e-01
-1.22098958e+00 -5.33698857e-01 -8.23881865e-01 -6.86704576e-01
-1.69198513e-01 4.17745978e-01 8.22961330e-01 6.56202972e-01
-9.07798648e-01 5.26215374e-01 -9.42991018e-01 -2.68524766e-01
8.76671076e-01 3.46530050e-01 -9.26688790e-01 -3.70253801e-01
-7.71397769e-01 7.48431861e-01 -1.73398018e-01 3.59998524e-01
-7.45234191e-01 -8.97095144e-01 -9.41712201e-01 1.60065889e-01
5.11443794e-01 -1.50855482e+00 1.26985931e+00 -1.35267064e-01
-1.14502299e+00 9.21770692e-01 -3.14537317e-01 -3.57464015e-01
1.09950757e+00 -1.19363844e+00 1.29585778e-02 5.33148706e-01
3.92103612e-01 6.93575859e-01 8.03709686e-01 -8.93470347e-01
-7.02451646e-01 -7.73847818e-01 -1.20682688e-02 4.43484455e-01
-2.20215619e-01 7.92602450e-02 -9.61746395e-01 -7.02475309e-01
-1.16824955e-01 -1.01073956e+00 -3.46489429e-01 6.27042055e-01
-4.39037144e-01 -6.63290694e-02 4.04684126e-01 -9.88180637e-01
7.53456175e-01 -1.69514263e+00 3.97823215e-01 -1.25655994e-01
7.65360355e-01 9.17723253e-02 -1.07312560e-01 2.93091953e-01
1.48152947e-01 -2.53136545e-01 2.13166863e-01 -1.20590365e+00
2.67542809e-01 -2.51044303e-01 4.09006514e-02 5.69654524e-01
3.61801356e-01 1.29197836e+00 -7.55437374e-01 -4.86815155e-01
2.11344287e-01 8.77155304e-01 -8.60842347e-01 3.13828766e-01
1.08315632e-01 5.50047398e-01 -3.99812698e-01 6.12256110e-01
6.87758148e-01 -4.04536217e-01 -1.84886500e-01 -2.75470227e-01
2.04223655e-02 -1.09631337e-01 -1.21496403e+00 2.14213252e+00
-1.26436532e-01 4.90318298e-01 -1.07558571e-01 2.08146378e-01
2.79835790e-01 1.22254148e-01 3.01841497e-01 -9.70702544e-02
6.24962011e-03 -7.49694407e-02 -4.57545519e-01 -4.21557575e-01
6.78924739e-01 4.06424135e-01 -2.15515569e-01 -8.17414187e-03
3.85060728e-01 2.44604021e-01 1.90801382e-01 4.95887995e-01
9.99592364e-01 6.80704653e-01 -1.51589904e-02 -7.09669739e-02
1.28656551e-01 -4.20407295e-01 3.85570198e-01 8.30171049e-01
-2.07700744e-01 1.08104241e+00 3.56833428e-01 -9.27477777e-01
-1.05348873e+00 -1.58993137e+00 2.41681889e-01 1.44084358e+00
-9.03612226e-02 -6.79297268e-01 -6.86143875e-01 -6.24238551e-01
4.94325876e-01 -3.55564728e-02 -9.49945450e-01 4.11592215e-01
-7.07073927e-01 -5.35161376e-01 7.52107322e-01 9.23502028e-01
6.72526240e-01 -2.74973691e-01 -4.90306526e-01 6.95346668e-02
-3.61521304e-01 -1.67513788e+00 -7.94163287e-01 -5.43859959e-01
-4.39172626e-01 -1.34035909e+00 -1.16109943e+00 -2.22970441e-01
5.02732575e-01 3.67384255e-01 1.50334680e+00 -2.54908234e-01
-4.84414279e-01 5.31744480e-01 -8.10993239e-02 -3.74979198e-01
6.37580335e-01 1.20496929e-01 1.63157701e-01 6.48585036e-02
4.79737788e-01 -5.69124937e-01 -1.20646560e+00 2.79079288e-01
-1.89720616e-01 1.69575703e-03 5.47661185e-01 4.48390394e-01
4.30393964e-01 -6.57959640e-01 3.35617699e-02 -4.91617769e-01
3.10533941e-01 1.39713913e-01 -5.80569327e-01 1.71353310e-01
-2.85523254e-02 -1.26775019e-02 8.42951983e-02 -1.85430154e-01
-1.02234209e+00 3.25932533e-01 -1.27074748e-01 -4.63098228e-01
-1.75038338e-01 -2.61366010e-01 -1.92978114e-01 -3.24947208e-01
8.23286891e-01 -1.90458838e-02 -1.27022937e-01 -6.38762057e-01
7.28288949e-01 3.39200974e-01 1.02751672e+00 -6.62957966e-01
9.45878506e-01 9.14916694e-01 -4.80993390e-02 -4.63122606e-01
-1.30487859e+00 -6.68465018e-01 -1.05157435e+00 -2.09797904e-01
1.25395060e+00 -1.85897803e+00 -1.09118962e+00 7.11135447e-01
-1.19030488e+00 -2.81471550e-03 -1.47224531e-01 3.15132469e-01
-3.98740292e-01 2.29370505e-01 -8.19942772e-01 -6.40202463e-01
-5.22743285e-01 -9.87823665e-01 1.80891490e+00 2.35677361e-01
-2.95427322e-01 -4.71663445e-01 1.43502308e-02 9.09075320e-01
7.91581422e-02 5.43943644e-01 5.57956565e-03 -2.00291306e-01
-9.13498819e-01 -5.32320917e-01 -3.05155873e-01 -1.89757347e-01
-5.61286926e-01 -3.84469837e-01 -1.19008052e+00 -5.91657043e-01
-8.54639888e-01 -5.37137926e-01 1.14318478e+00 2.80203402e-01
8.45774293e-01 -1.98372945e-01 -3.50955755e-01 1.00120783e+00
9.45260823e-01 -1.20432413e+00 4.67725486e-01 2.15608299e-01
1.03709948e+00 2.86088824e-01 1.66821420e-01 5.29363275e-01
1.05695570e+00 9.42384005e-01 3.06527495e-01 -2.32721359e-01
-3.55508298e-01 -5.92055798e-01 1.16842560e-01 5.42649068e-02
-4.83602673e-01 5.61801950e-03 -7.24013388e-01 4.56260294e-01
-2.03362131e+00 -1.10288358e+00 -1.95767760e-01 2.01070333e+00
3.34782183e-01 7.78321773e-02 7.17344463e-01 -2.85554051e-01
6.24209940e-01 3.10870767e-01 -5.08832216e-01 3.47052008e-01
-2.28242248e-01 1.65065974e-01 5.14202356e-01 4.25630569e-01
-1.50538373e+00 9.80512083e-01 6.92200851e+00 4.35859233e-01
-4.35827941e-01 2.57444531e-01 3.37073207e-01 -7.27123797e-01
1.48188919e-01 -4.48951185e-01 -1.10122466e+00 2.23914176e-01
3.01682383e-01 3.60545695e-01 2.71549821e-01 7.76125252e-01
3.05535141e-02 -5.52075431e-02 -1.28709650e+00 1.41215515e+00
9.14220884e-02 -1.24638236e+00 -1.03596598e-01 1.37023374e-01
7.20827341e-01 2.12708279e-01 3.66556048e-01 2.22218215e-01
5.01192808e-01 -1.22741306e+00 1.05064929e+00 5.73252380e-01
7.87475884e-01 -8.19521546e-01 6.01171196e-01 1.40518993e-01
-1.47718048e+00 -1.87463269e-01 -1.71121627e-01 -2.85631210e-01
3.32322270e-01 3.41241717e-01 -4.47045773e-01 5.38383067e-01
1.03147852e+00 6.21836305e-01 -1.09240937e+00 1.18685663e+00
-4.73941863e-01 5.30410884e-03 -4.28873330e-01 1.77079067e-01
-2.07564197e-02 3.62027347e-01 6.03082180e-01 1.44604504e+00
-2.74549872e-02 1.76249325e-01 2.40205199e-01 7.06642032e-01
-1.33906990e-01 -4.39176291e-01 -1.84580863e-01 4.47935879e-01
4.16800141e-01 1.27193522e+00 -3.31916600e-01 -2.05938205e-01
-5.23815691e-01 1.33432901e+00 8.52645218e-01 4.46601540e-01
-8.42032552e-01 -1.26878768e-01 1.11645806e+00 4.53982800e-01
6.98325515e-01 -5.03825068e-01 -2.23919377e-01 -1.57862568e+00
2.41830081e-01 -8.81347060e-01 5.13451278e-01 -7.13264585e-01
-1.61845994e+00 5.23525298e-01 7.30725750e-02 -8.90658021e-01
-1.81810021e-01 -7.54628837e-01 -2.79817998e-01 1.07442594e+00
-1.12732279e+00 -1.99364102e+00 -6.56155765e-01 7.77624249e-01
3.13950509e-01 -1.08242169e-01 8.12166035e-01 2.38727495e-01
-5.20012021e-01 9.33608890e-01 -3.77550215e-01 6.17369652e-01
8.19061875e-01 -1.36360037e+00 9.95972872e-01 1.13452291e+00
1.67234093e-01 9.21516061e-01 7.13670552e-01 -6.06743217e-01
-1.34634006e+00 -1.01237190e+00 6.04564905e-01 -1.26327288e+00
2.81774521e-01 -8.91287446e-01 -2.15187922e-01 9.59630668e-01
1.35899395e-01 5.90780199e-01 7.68620551e-01 7.48538435e-01
-9.56329525e-01 7.07891434e-02 -1.09149039e+00 7.14184523e-01
1.54839647e+00 -5.23995101e-01 -7.37751365e-01 3.29459399e-01
7.54097879e-01 -8.87674391e-01 -7.66681790e-01 1.17040135e-01
1.10689855e+00 -1.23681819e+00 1.50119102e+00 -8.36919427e-01
3.75706136e-01 -2.69608289e-01 1.28084406e-01 -1.09412134e+00
-8.74289751e-01 -6.95871711e-01 -6.87543631e-01 6.12601578e-01
1.05964310e-01 -4.01443154e-01 1.24776065e+00 9.09055293e-01
3.43088567e-01 -9.32496667e-01 -8.84891689e-01 -4.06623393e-01
-1.98956951e-01 -2.31025770e-01 5.38376570e-01 2.65293986e-01
-4.64809448e-01 5.26352406e-01 -6.12019658e-01 4.85967755e-01
1.13726294e+00 -1.79163530e-01 1.41547930e+00 -1.19924676e+00
-6.45012796e-01 7.45547283e-03 -6.79319859e-01 -1.41758370e+00
-6.92228228e-02 -2.95734435e-01 -3.65226030e-01 -1.62744021e+00
5.80382526e-01 1.40757516e-01 1.56934559e-01 3.86877567e-01
-6.87682450e-01 8.46146703e-01 7.46555865e-01 9.65856314e-02
-1.13504910e+00 5.32079697e-01 1.22094357e+00 1.19627751e-01
5.57699092e-02 -1.25735238e-01 -8.83937776e-01 9.99500275e-01
6.37610853e-02 -3.63541573e-01 -5.37570044e-02 -1.09089303e+00
2.32394949e-01 -3.06637347e-01 1.00015807e+00 -1.32510293e+00
3.05064440e-01 3.97694200e-01 1.32218838e+00 -8.02675843e-01
1.10324645e+00 -3.71560395e-01 -1.11864880e-01 3.48224580e-01
-7.58686513e-02 4.16842520e-01 1.20812140e-01 9.33701038e-01
2.07205564e-01 6.80203378e-01 5.80691099e-01 -3.06961775e-01
-5.55908144e-01 8.29929411e-01 1.93014517e-01 3.33016455e-01
7.93804944e-01 -2.38518253e-01 -3.86990875e-01 -5.03937840e-01
-8.57616961e-01 5.02064884e-01 3.36285323e-01 5.41614592e-01
7.13270366e-01 -1.30987942e+00 -1.13786471e+00 3.41142416e-02
1.17094018e-01 2.68781215e-01 5.21275818e-01 5.91006219e-01
-7.08760858e-01 1.54813796e-01 -1.18073218e-01 -6.94353342e-01
-1.37610722e+00 1.73626870e-01 4.08326060e-01 -5.48556447e-01
-8.55138898e-01 1.42049909e+00 1.04015693e-01 -5.35978377e-01
9.65365544e-02 3.52889560e-02 9.71922576e-02 -1.44468606e-01
9.75384176e-01 5.45198977e-01 -1.07178800e-01 -8.67669344e-01
-6.02790117e-01 7.68123984e-01 -1.64733082e-01 -3.71268898e-01
1.55368555e+00 -6.16220720e-02 2.25683764e-01 -1.56109482e-01
9.98558104e-01 4.09716189e-01 -1.98301184e+00 -2.32688084e-01
-5.25382698e-01 -9.57939863e-01 -3.66826773e-01 -8.84042263e-01
-8.02580833e-01 7.17650712e-01 5.62636435e-01 -6.33301139e-01
6.06267154e-01 -2.50830352e-02 8.27062309e-01 4.56102788e-01
6.74632847e-01 -8.65547717e-01 4.91490513e-01 4.43314373e-01
9.17998374e-01 -1.22062802e+00 4.25978750e-01 -8.24787796e-01
-6.46896541e-01 9.24548805e-01 8.49016726e-01 -5.28322935e-01
3.04656357e-01 3.66933644e-01 -1.00667000e-01 -3.32355380e-01
-4.96363312e-01 -4.13982511e-01 6.35835707e-01 6.57902420e-01
2.90492982e-01 9.01622921e-02 3.27696681e-01 8.31543744e-01
-3.62580061e-01 8.53834301e-02 -3.94467870e-03 6.97321236e-01
-1.92608148e-01 -6.69555187e-01 -6.98382795e-01 1.97934195e-01
-5.59081078e-01 -5.78997247e-02 -4.38417464e-01 8.00849378e-01
3.63789856e-01 9.74246085e-01 3.57352756e-02 -3.18163544e-01
8.19959283e-01 -1.86891302e-01 1.02779257e+00 -4.44015115e-01
-7.64777243e-01 -4.14289236e-02 2.43786931e-01 -1.20173120e+00
-1.75989747e-01 -7.44819939e-01 -6.84267402e-01 -5.23347378e-01
8.15441981e-02 -2.81300187e-01 9.87914577e-02 9.27789986e-01
8.14986050e-01 4.90858227e-01 -1.79585591e-01 -1.36459756e+00
-5.35122991e-01 -1.00566125e+00 -2.05668509e-01 6.89582467e-01
4.21418458e-01 -8.70442986e-01 2.83700109e-01 -1.47267342e-01] | [7.104533672332764, -0.8044403791427612] |
73282413-a963-4de7-809b-c8d8d561954f | feedback-linearization-of-car-dynamics-for | 2110.10441 | null | https://arxiv.org/abs/2110.10441v1 | https://arxiv.org/pdf/2110.10441v1.pdf | Feedback Linearization of Car Dynamics for Racing via Reinforcement Learning | Through the method of Learning Feedback Linearization, we seek to learn a linearizing controller to simplify the process of controlling a car to race autonomously. A soft actor-critic approach is used to learn a decoupling matrix and drift vector that effectively correct for errors in a hand-designed linearizing controller. The result is an exactly linearizing controller that can be used to enable the well-developed theory of linear systems to design path planning and tracking schemes that are easy to implement and significantly less computationally demanding. To demonstrate the method of feedback linearization, it is first used to learn a simulated model whose exact structure is known, but varied from the initial controller, so as to introduce error. We further seek to apply this method to a system that introduces even more error in the form of a gym environment specifically designed for modeling the dynamics of car racing. To do so, we posit an extension to the method of learning feedback linearization; a neural network that is trained using supervised learning to convert the output of our linearizing controller to the required input for the racing environment. Our progress towards these goals is reported and the next steps in their accomplishment are discussed. | ['Xiangyu Cai', 'Sida Li', 'Michael Estrada'] | 2021-10-20 | null | null | null | null | ['carracing-v0'] | ['playing-games'] | [ 9.00405943e-02 7.99988031e-01 -4.24342364e-01 -1.17309652e-02
-4.90498126e-01 -5.47697365e-01 2.98640877e-01 -3.32285196e-01
-2.69691288e-01 6.83774352e-01 -9.10746902e-02 -6.02195561e-01
-2.74880920e-02 -3.77574831e-01 -1.05947340e+00 -6.96490049e-01
6.35996014e-02 3.41355562e-01 6.99524954e-02 -8.71087849e-01
1.47496074e-01 6.77858949e-01 -1.15423489e+00 -1.98556378e-01
6.71978652e-01 4.71662432e-01 1.18086353e-01 9.43797469e-01
5.95080793e-01 1.15565121e+00 -1.98853880e-01 5.34200370e-01
5.79370320e-01 -6.00717902e-01 -6.62784696e-01 2.56551147e-01
1.36145011e-01 -1.27990738e-01 -3.66167516e-01 7.40313292e-01
4.12505329e-01 4.36089575e-01 5.52115977e-01 -1.14036715e+00
-4.22942191e-01 3.35746318e-01 2.15915322e-01 -8.49277452e-02
2.02821046e-01 5.56079745e-01 5.66332519e-01 -3.87134671e-01
4.50982004e-01 1.28804421e+00 7.95061350e-01 7.75595546e-01
-1.20728648e+00 -2.93862104e-01 3.20998967e-01 -1.13834769e-01
-1.30746627e+00 -4.93645787e-01 5.77114344e-01 -4.86878484e-01
7.89415300e-01 2.17091724e-01 9.22518790e-01 6.27378106e-01
6.21070504e-01 5.73191106e-01 7.22094953e-01 -5.98826766e-01
4.49149251e-01 4.11978662e-01 -2.35544845e-01 8.29064190e-01
-1.98978595e-02 4.90449935e-01 4.51705754e-02 1.59936920e-01
9.72488821e-01 -2.65011698e-01 -2.01394737e-01 -8.68472934e-01
-7.53188312e-01 6.88414156e-01 7.73701847e-01 1.01866856e-01
-3.87567312e-01 3.84012997e-01 1.27046421e-01 5.73843539e-01
6.77117258e-02 1.10762930e+00 -3.66401285e-01 4.43797857e-02
-3.74343187e-01 7.41434813e-01 1.01019597e+00 9.34313178e-01
4.66089189e-01 6.45972848e-01 1.70210510e-01 2.80341655e-01
2.43037999e-01 2.45802298e-01 4.32273567e-01 -1.51399851e+00
2.39610896e-01 5.25287569e-01 4.98582482e-01 -6.47643447e-01
-5.55452704e-01 -1.71739906e-01 -2.24729344e-01 8.98168325e-01
4.20575380e-01 -6.86535001e-01 -6.42336369e-01 1.63619411e+00
3.62375051e-01 -1.18727520e-01 6.93574995e-02 9.55941021e-01
-1.59789667e-01 7.55557775e-01 -4.07972157e-01 -2.29817495e-01
5.90617359e-01 -8.56390417e-01 -6.72666311e-01 -3.60929757e-01
7.51167297e-01 -1.56752884e-01 1.39144504e+00 4.73558366e-01
-1.19594681e+00 -4.96283621e-01 -1.29706013e+00 1.88296601e-01
-3.66734087e-01 1.29282519e-01 1.77857667e-01 2.51825005e-01
-9.62069035e-01 8.99328768e-01 -1.09375608e+00 -2.33041003e-01
-3.13471287e-01 6.01123273e-01 -5.11175171e-02 4.31016564e-01
-1.09191799e+00 1.57790411e+00 5.49206376e-01 2.85676748e-01
-1.09335113e+00 -7.76247501e-01 -7.50447989e-01 -5.40193319e-02
7.21261382e-01 -6.78691149e-01 1.69277298e+00 -1.29805934e+00
-2.27914429e+00 2.29999542e-01 1.36761516e-01 -3.40882003e-01
4.33948338e-01 -1.79830462e-01 6.19262904e-02 -1.63689718e-01
-8.37311670e-02 3.94172072e-01 8.32848668e-01 -1.24154067e+00
-4.00972307e-01 4.51624300e-03 8.64743739e-02 3.59548658e-01
2.93795727e-02 -2.95087159e-01 9.94069315e-03 -1.96838439e-01
-2.20069095e-01 -1.38399780e+00 -6.00339830e-01 2.39893883e-01
-2.23592564e-01 5.95809370e-02 7.26526618e-01 -3.18796217e-01
9.95593607e-01 -1.91630054e+00 6.25072062e-01 2.52160132e-01
-4.48547751e-02 4.40000921e-01 7.20584244e-02 6.45858049e-01
-3.88380796e-01 -2.36640975e-01 -4.53538746e-02 7.04234391e-02
-1.69602484e-01 3.27589154e-01 -3.82191956e-01 5.57225406e-01
4.02253479e-01 8.42345059e-01 -9.98849630e-01 1.70342941e-02
2.95116395e-01 3.53425592e-01 -7.29443789e-01 4.00699824e-01
-4.81179237e-01 6.42625213e-01 -6.00363672e-01 1.25769764e-01
-2.38337949e-01 1.98859572e-01 1.88890323e-01 1.97715297e-01
-5.59803903e-01 5.83857261e-02 -1.51529539e+00 1.09967971e+00
-4.65221643e-01 6.02831423e-01 5.37943006e-01 -1.22397006e+00
8.76145303e-01 3.70870858e-01 5.15251219e-01 -3.66214037e-01
5.68858504e-01 2.28249878e-01 -3.27286720e-02 -7.04236507e-01
2.10279241e-01 -2.98339456e-01 -3.27683822e-03 2.12191939e-01
-8.31314698e-02 -7.68867910e-01 -9.90309119e-02 7.19586164e-02
1.00914645e+00 2.69602597e-01 2.29895696e-01 -4.44598168e-01
7.53063083e-01 4.97964740e-01 5.13951123e-01 4.01382774e-01
-2.68041581e-01 4.12888601e-02 5.41162252e-01 -5.01490295e-01
-1.25768387e+00 -5.89595139e-01 3.64551812e-01 9.59624052e-01
-1.91722900e-01 -1.13421157e-01 -9.20659244e-01 -2.41726786e-02
2.69347817e-01 8.40288758e-01 -5.36785007e-01 -6.81966722e-01
-9.24082816e-01 -3.66802700e-02 2.11052701e-01 5.13440073e-01
-4.26928542e-04 -7.08766282e-01 -6.66415155e-01 3.62807661e-01
3.09325725e-01 -5.70170105e-01 -6.26045346e-01 4.49004799e-01
-7.00553179e-01 -1.09851050e+00 -4.83224690e-01 -8.47063959e-01
8.54524016e-01 -1.58149153e-01 5.07269442e-01 4.05964628e-02
-2.15689391e-01 7.05425322e-01 9.05621499e-02 -7.28467345e-01
-7.13434100e-01 2.98761278e-02 4.72078443e-01 -2.96524405e-01
-3.55705351e-01 -1.23214953e-01 -2.14078933e-01 3.02489370e-01
-7.31296539e-01 -2.65794341e-02 2.57907361e-01 1.00568974e+00
3.84563655e-01 7.60335755e-03 5.01472354e-01 -6.07355177e-01
1.05943060e+00 -1.92364588e-01 -1.06658328e+00 2.62985639e-02
-8.09198022e-01 3.43976974e-01 1.19599533e+00 -7.49430537e-01
-9.11249936e-01 7.72550225e-01 -3.22843902e-02 -4.57835764e-01
2.88522303e-01 2.53942639e-01 2.10081972e-02 -6.07794881e-01
8.85425031e-01 -6.15583807e-02 5.63833952e-01 -1.69954836e-01
8.21643174e-01 5.69031894e-01 6.74939275e-01 -4.80914742e-01
8.32851648e-01 1.11023195e-01 1.10777162e-01 -7.45707154e-01
-6.33619905e-01 -1.30775094e-01 -6.81552112e-01 -3.86585534e-01
4.93826270e-01 -7.86115587e-01 -1.18399405e+00 2.21867025e-01
-7.40686834e-01 -1.11427379e+00 -6.70574069e-01 4.41486359e-01
-1.28649056e+00 -3.60444754e-01 -5.62503099e-01 -9.93795574e-01
1.58309519e-01 -1.09049165e+00 6.60022199e-01 2.09805548e-01
-1.79299921e-01 -1.08046544e+00 4.13102090e-01 -1.11088842e-01
4.59215999e-01 2.43536159e-01 6.95540488e-01 -2.00016703e-02
-4.40961301e-01 -3.71184886e-01 5.70229769e-01 5.23745596e-01
-4.19685096e-02 6.07763752e-02 -6.06514871e-01 -6.93755329e-01
1.52655959e-01 -3.55110884e-01 1.94557413e-01 3.66394907e-01
6.77029371e-01 -4.03669417e-01 -4.27844614e-01 2.68696547e-01
1.28969204e+00 4.76618141e-01 2.38240376e-01 4.32195544e-01
7.65204847e-01 6.33684754e-01 6.18690312e-01 8.66719112e-02
3.50374311e-01 6.70140147e-01 3.53435904e-01 -9.84865576e-02
2.22133338e-01 -5.17275691e-01 4.93671864e-01 6.49859369e-01
1.26132280e-01 2.68564403e-01 -8.92952025e-01 5.12252867e-01
-2.02647543e+00 -7.59639740e-01 2.19322056e-01 2.16423893e+00
7.71516025e-01 1.72060952e-01 3.88766885e-01 -1.90672167e-02
4.35818583e-01 -5.38246930e-01 -9.93562937e-01 -8.09852719e-01
5.61512113e-01 -1.75265118e-01 9.26433980e-01 1.02346909e+00
-8.26630056e-01 1.01497829e+00 7.77791834e+00 9.71830115e-02
-1.33471179e+00 -4.63724881e-01 2.11083636e-01 -2.92748418e-02
3.27248462e-02 1.96714625e-01 -6.72389269e-01 2.82878548e-01
1.34435916e+00 -3.44035715e-01 1.09243107e+00 1.04809570e+00
9.46506739e-01 -9.33548063e-03 -1.22278559e+00 3.22300076e-01
-1.44466519e-01 -1.20103967e+00 -6.28962219e-01 -2.88689375e-01
7.86902964e-01 -3.79131973e-01 6.87500136e-03 6.66275442e-01
7.67038524e-01 -7.17659831e-01 8.22853506e-01 6.89893007e-01
3.10280442e-01 -7.22252429e-01 2.06544980e-01 9.73898947e-01
-9.87516582e-01 -7.64849842e-01 -1.17682785e-01 -6.10333502e-01
2.24552497e-01 -1.73521370e-01 -1.09027684e+00 -5.90865240e-02
8.03422108e-02 3.92508239e-01 -1.24325700e-01 8.82477462e-01
-2.77251869e-01 4.43943411e-01 -3.93220305e-01 -4.10451770e-01
3.72115433e-01 -1.05599672e-01 5.74062765e-01 6.49599552e-01
-1.02913953e-01 2.79331654e-01 3.70711774e-01 8.98999572e-01
4.07358080e-01 -2.58066021e-02 -9.67366636e-01 1.43947646e-01
1.28770173e-01 8.69721889e-01 -2.26721182e-01 -3.26505959e-01
1.21837616e-01 4.98352081e-01 5.45078039e-01 3.54224443e-01
-6.98374450e-01 -6.37934089e-01 7.06772983e-01 3.29264939e-01
1.64043993e-01 -5.51893175e-01 -1.08376242e-01 -9.21650648e-01
-2.58869141e-01 -9.20898855e-01 -2.88743973e-01 -8.69506299e-01
-7.04293728e-01 1.09429844e-01 2.09512576e-01 -1.17487812e+00
-1.02254963e+00 -4.46017563e-01 -6.63356543e-01 9.07522202e-01
-1.14980972e+00 -5.53728342e-01 4.86369543e-02 4.59309936e-01
4.59598809e-01 -4.15742062e-02 5.60762584e-01 1.44040123e-01
-8.54413450e-01 2.76887834e-01 4.04470891e-01 -2.12699503e-01
6.78522706e-01 -1.34926999e+00 1.86629981e-01 8.13363552e-01
-6.23673022e-01 7.08643198e-01 1.03992057e+00 -6.34733379e-01
-1.86445510e+00 -1.08184183e+00 6.11960173e-01 -4.21828032e-01
8.26159120e-01 -2.69585103e-01 -8.14584196e-01 1.14681137e+00
-1.00841850e-01 -1.07657336e-01 -2.59178370e-01 -4.08130467e-01
3.84284228e-01 -3.22597504e-01 -9.23857272e-01 9.13764060e-01
3.98799360e-01 -4.30175453e-01 -6.22866213e-01 4.20515299e-01
8.15324128e-01 -9.36935782e-01 -6.13079429e-01 -1.37971088e-01
4.97893125e-01 -1.49253637e-01 8.21770370e-01 -9.05419707e-01
-9.62044522e-02 -5.49655795e-01 4.15264159e-01 -1.84349716e+00
-4.67847884e-01 -1.10095489e+00 -5.85861243e-02 5.72560430e-01
3.81513655e-01 -7.75802851e-01 7.51267314e-01 8.73926401e-01
-3.60692024e-01 -7.89077342e-01 -6.60103261e-01 -8.04416537e-01
5.31216860e-01 -1.93285756e-02 -2.27403175e-03 5.74885726e-01
4.32909548e-01 3.88970464e-01 -5.01756430e-01 2.74163872e-01
2.62058020e-01 -4.99281198e-01 8.68589640e-01 -8.28582168e-01
-3.36593419e-01 -9.77306664e-02 -2.03440696e-01 -1.01830590e+00
3.21035266e-01 -4.42410856e-01 6.94009721e-01 -1.20489812e+00
-6.79349065e-01 -4.27640706e-01 1.38156891e-01 6.03183985e-01
6.49902076e-02 -3.92421484e-01 4.66759294e-01 8.59983936e-02
-2.53674924e-01 4.26001191e-01 1.40095043e+00 -7.81182945e-02
-8.54151785e-01 2.80939519e-01 -7.25866735e-01 6.90618992e-01
8.92586470e-01 -3.23109001e-01 -7.40059078e-01 -2.83534259e-01
1.20616540e-01 4.46759462e-01 1.36898980e-01 -1.17088580e+00
4.67187434e-01 -3.64557266e-01 1.43804923e-01 5.26629984e-02
1.89575791e-01 -9.73629892e-01 7.58581907e-02 8.53276193e-01
-9.17155147e-01 1.30180091e-01 3.18801552e-01 3.76021087e-01
-2.07081083e-02 -2.43235886e-01 9.59547043e-01 -1.67599186e-01
-5.35122752e-01 -1.62420005e-01 -9.53508973e-01 -1.54167667e-01
1.34737992e+00 4.53199558e-02 1.63338229e-01 -5.97052634e-01
-9.26376402e-01 5.87038636e-01 3.35295916e-01 3.28051597e-01
2.09951967e-01 -1.15528190e+00 -2.62863278e-01 4.19413269e-01
-2.69383281e-01 -1.57733455e-01 -2.57832468e-01 5.57456434e-01
-5.29664695e-01 6.84462488e-01 -4.96539950e-01 -3.90434712e-01
-9.76043046e-01 7.09934294e-01 8.82885337e-01 7.81028122e-02
-5.60607195e-01 3.77317011e-01 -3.83929610e-01 -6.94801569e-01
2.90314376e-01 -7.15130389e-01 -2.60501243e-02 -6.02232814e-01
4.82626706e-01 3.47477496e-01 -1.77148711e-02 -2.57990658e-01
8.94337445e-02 5.09310305e-01 2.58236676e-01 -1.60742864e-01
1.21319056e+00 -9.13455635e-02 2.66209900e-01 6.33137167e-01
8.55224371e-01 -4.93142307e-01 -1.73332250e+00 6.04219548e-02
-7.78384209e-02 -1.32499151e-02 1.80276409e-01 -4.36767548e-01
-6.35429800e-01 5.62828720e-01 7.19853044e-01 1.83335796e-01
7.73789346e-01 -4.53018129e-01 3.15710813e-01 8.85612369e-01
3.07508796e-01 -1.35277510e+00 9.43775568e-03 6.89193666e-01
9.04298723e-01 -1.02995014e+00 -3.62841003e-02 -7.12530538e-02
-7.13669598e-01 1.30521679e+00 7.27011204e-01 -7.49557734e-01
5.61043680e-01 5.59208691e-01 2.09633391e-02 9.11381543e-02
-8.86842787e-01 -2.59560030e-02 3.69689725e-02 5.31746745e-01
1.47448376e-01 5.59082837e-04 -2.02466488e-01 -9.90956128e-02
9.91769806e-02 4.62701738e-01 7.95326114e-01 1.15839791e+00
-7.48005152e-01 -1.06954765e+00 -6.00299299e-01 -2.17322656e-03
1.76458418e-01 6.27957225e-01 -3.75035197e-01 1.00357962e+00
-1.24901861e-01 6.44918501e-01 -5.40792197e-02 -2.43124589e-01
9.60666656e-01 -7.73827210e-02 2.90536761e-01 -5.85444212e-01
-6.13915503e-01 -1.59460589e-01 -6.64137527e-02 -8.78405750e-01
1.83381379e-01 -4.68944758e-01 -1.36189544e+00 -2.48388082e-01
-2.63492167e-01 1.55291617e-01 6.86646640e-01 7.25527287e-01
9.56409872e-02 8.48773837e-01 7.21788764e-01 -1.36270547e+00
-1.13640881e+00 -4.61809665e-01 -3.74891341e-01 -9.66099054e-02
8.61729860e-01 -7.90270329e-01 -3.46507579e-01 1.69827163e-01] | [4.855642318725586, 2.027982234954834] |
fecb8036-5b7d-4179-956a-7ab760bc4c1c | loglg-weakly-supervised-log-anomaly-detection | 2208.10833 | null | https://arxiv.org/abs/2208.10833v5 | https://arxiv.org/pdf/2208.10833v5.pdf | LogLG: Weakly Supervised Log Anomaly Detection via Log-Event Graph Construction | Fully supervised log anomaly detection methods suffer the heavy burden of annotating massive unlabeled log data. Recently, many semi-supervised methods have been proposed to reduce annotation costs with the help of parsed templates. However, these methods consider each keyword independently, which disregards the correlation between keywords and the contextual relationships among log sequences. In this paper, we propose a novel weakly supervised log anomaly detection framework, named LogLG, to explore the semantic connections among keywords from sequences. Specifically, we design an end-to-end iterative process, where the keywords of unlabeled logs are first extracted to construct a log-event graph. Then, we build a subgraph annotator to generate pseudo labels for unlabeled log sequences. To ameliorate the annotation quality, we adopt a self-supervised task to pre-train a subgraph annotator. After that, a detection model is trained with the generated pseudo labels. Conditioned on the classification results, we re-extract the keywords from the log sequences and update the log-event graph for the next iteration. Experiments on five benchmarks validate the effectiveness of LogLG for detecting anomalies on unlabeled log data and demonstrate that LogLG, as the state-of-the-art weakly supervised method, achieves significant performance improvements compared to existing methods. | ['Bo Zhang', 'Liangfan Zheng', 'Weichao Hou', 'Tieqiao Zheng', 'Renjie Chen', 'Zhoujun Li', 'Jiaheng Liu', 'Jian Yang', 'Yuhui Guo', 'Hongcheng Guo'] | 2022-08-23 | null | null | null | null | ['scene-recognition'] | ['computer-vision'] | [ 2.03227967e-01 -3.20312046e-02 -1.73443839e-01 -4.86265212e-01
-5.63081205e-01 -4.78067994e-01 2.61220723e-01 4.88534659e-01
-1.88646510e-01 3.32474858e-01 -2.58111730e-02 -3.35890710e-01
2.42130250e-01 -6.36734605e-01 -6.59983873e-01 -5.56069493e-01
-1.21276081e-01 4.10476267e-01 6.46494687e-01 2.76633799e-01
1.61550403e-01 3.62260006e-02 -1.22154832e+00 5.07899895e-02
8.41939509e-01 1.07540369e+00 -1.32841066e-01 2.13859528e-01
-3.94095629e-01 9.23568130e-01 -4.20819223e-01 -1.00447580e-01
4.12684400e-03 -8.53095412e-01 -5.98736107e-01 5.91805220e-01
1.42336756e-01 -2.68097490e-01 -4.15325254e-01 1.12229061e+00
3.63213830e-02 1.01376414e-01 3.98885578e-01 -1.24394953e+00
-2.64303416e-01 7.70231962e-01 -7.33588576e-01 4.57017809e-01
4.27757442e-01 6.37456495e-03 1.27471614e+00 -1.02394354e+00
2.30922386e-01 9.13527429e-01 5.30104518e-01 2.25837886e-01
-9.72057998e-01 -6.12749398e-01 6.31975770e-01 4.99052793e-01
-1.39318800e+00 -1.44994110e-01 9.31776106e-01 -2.33339667e-01
7.78908849e-01 9.79853868e-02 3.08811247e-01 9.12098467e-01
-1.59201831e-01 8.50549459e-01 4.91721243e-01 -4.67443973e-01
3.20019484e-01 -1.84569672e-01 5.53975880e-01 1.05683053e+00
4.19806957e-01 -3.34863037e-01 -6.67107701e-01 -7.26368070e-01
5.35569079e-02 2.53871411e-01 -1.23431556e-01 -2.62523204e-01
-8.06151748e-01 6.29693151e-01 -1.83313561e-03 2.27664366e-01
-3.49919349e-01 -1.99114174e-01 7.52359688e-01 2.42284119e-01
6.78794742e-01 -1.64063141e-01 -3.97015214e-01 6.56402484e-02
-6.90915525e-01 -1.61762476e-01 9.83544767e-01 8.95527780e-01
9.83030140e-01 -1.10301346e-01 -7.06415027e-02 6.57524407e-01
6.85187519e-01 1.50805071e-01 5.53407729e-01 -1.86144054e-01
5.73468208e-01 9.87833917e-01 -2.22332329e-01 -1.00771379e+00
-2.18660384e-01 -4.99206692e-01 -5.77294350e-01 -4.63876516e-01
1.70763567e-01 1.17817082e-01 -9.02277589e-01 1.36873090e+00
4.78652209e-01 6.30496502e-01 -1.91846818e-01 5.17928302e-01
2.55798101e-01 6.91392541e-01 1.91402987e-01 -6.51313961e-01
1.13350761e+00 -1.04444933e+00 -7.48347402e-01 -5.07064283e-01
9.59348202e-01 -5.17273664e-01 1.19598520e+00 3.22380245e-01
-4.05142844e-01 -1.94171056e-01 -1.09626961e+00 5.04589140e-01
-1.95143633e-02 1.09176964e-01 1.71560526e-01 3.80795062e-01
-3.63099664e-01 4.55739528e-01 -1.36307514e+00 -3.53240669e-01
3.33179057e-01 3.18952240e-02 4.93957959e-02 -7.61334077e-02
-9.82694447e-01 2.19474465e-01 1.14346731e+00 1.21181682e-01
-1.24268210e+00 -1.17244318e-01 -1.11532843e+00 1.31763130e-01
1.11895168e+00 1.18611336e-01 1.31784272e+00 -7.81186521e-01
-9.80488658e-01 6.35360360e-01 -3.88166875e-01 -3.83196026e-01
2.88993537e-01 -4.28593159e-01 -6.01394117e-01 -4.09563398e-03
2.89154381e-01 -4.09183979e-01 8.06967497e-01 -1.23186696e+00
-9.42758083e-01 -4.96738434e-01 -3.73377293e-01 -7.68663138e-02
-6.83971047e-01 7.64552550e-03 -8.82284403e-01 -6.81541324e-01
3.07426214e-01 -8.60746920e-01 -1.26055196e-01 -5.86824477e-01
-6.55355930e-01 -6.03037298e-01 1.41903007e+00 -7.25104153e-01
1.82167757e+00 -2.41505265e+00 -3.94562453e-01 5.64155042e-01
2.71416426e-01 2.95190662e-01 5.85069060e-02 3.94362301e-01
-8.95982012e-02 -7.05281794e-02 -7.98110127e-01 -5.74197888e-01
-2.20304295e-01 4.87262011e-01 -4.37334448e-01 5.15403867e-01
1.48353621e-01 6.28713906e-01 -1.11121380e+00 -8.41914475e-01
-1.70920953e-01 -3.34692985e-01 -4.00953025e-01 5.83575189e-01
-6.99807525e-01 4.77446705e-01 -8.59296501e-01 7.91955411e-01
4.31859761e-01 -6.11442089e-01 3.38508964e-01 1.33768857e-01
1.85966477e-01 3.81946892e-01 -1.04055607e+00 1.59341872e+00
-1.21673919e-01 1.59480155e-01 -3.57660621e-01 -1.37984920e+00
9.21511948e-01 2.62616932e-01 5.27341425e-01 -2.84731716e-01
-1.41104281e-01 4.36839640e-01 -1.65802717e-01 -6.35356367e-01
-2.65846532e-02 3.07737738e-01 -1.50861472e-01 8.31224918e-01
-5.16965352e-02 5.66204250e-01 4.87145364e-01 5.00327229e-01
1.44116938e+00 2.26712540e-01 3.57005358e-01 1.39610633e-01
1.06010044e+00 1.16437292e-02 8.58931303e-01 6.76907599e-01
-6.17051423e-02 3.32569271e-01 6.02911592e-01 -3.98028225e-01
-6.32967532e-01 -1.00271976e+00 2.85461545e-01 1.21191370e+00
2.66041517e-01 -8.44372332e-01 -7.79250503e-01 -1.53641927e+00
-1.73544452e-01 5.85364521e-01 -2.76460320e-01 -4.44563538e-01
-8.59691978e-01 -9.12738144e-01 5.82907379e-01 5.14061749e-01
5.39390087e-01 -1.18497920e+00 -1.19678907e-01 5.54154754e-01
-3.68992418e-01 -1.25141394e+00 -7.56521344e-01 2.33597040e-01
-8.37683678e-01 -1.23845077e+00 1.71282008e-01 -9.74900723e-01
7.85722494e-01 8.53077099e-02 9.20053363e-01 4.42387104e-01
-1.78870201e-01 8.41494799e-02 -5.82108617e-01 -8.90911818e-02
-4.44517821e-01 1.83238193e-01 1.42499238e-01 4.15503860e-01
7.11312532e-01 -7.91636765e-01 -3.96358013e-01 4.23375338e-01
-1.03464985e+00 -4.68022436e-01 6.28732383e-01 6.98717535e-01
9.12515759e-01 2.15953082e-01 7.49969602e-01 -1.32765210e+00
5.14659047e-01 -7.03498006e-01 -6.07721865e-01 3.24803859e-01
-1.07928205e+00 2.70767510e-01 1.01634860e+00 -4.23502415e-01
-1.23069394e+00 3.09700847e-01 -1.82035882e-02 -3.96248221e-01
-1.85610667e-01 8.29171836e-01 -5.35222173e-01 3.67569208e-01
5.88680446e-01 5.47515988e-01 -3.95672441e-01 -7.37433434e-01
1.15582310e-01 8.94703686e-01 7.67851591e-01 -4.60303873e-01
1.08941555e+00 4.36485469e-01 -3.55773985e-01 -6.12776518e-01
-1.31400776e+00 -8.41957569e-01 -5.55793226e-01 1.20058991e-01
6.62780523e-01 -6.83599830e-01 -2.86343843e-01 5.84631562e-01
-1.01040971e+00 -1.16155427e-02 -1.25781000e-01 4.36979353e-01
-2.16207117e-01 9.36525047e-01 -6.73130155e-01 -6.91688478e-01
-3.54457796e-01 -5.73891521e-01 9.38502312e-01 2.07257822e-01
-1.15624912e-01 -8.29827726e-01 2.44210750e-01 1.60561621e-01
-1.44822702e-01 1.25860155e-01 1.02157569e+00 -1.58591938e+00
-5.26471317e-01 -5.27538717e-01 -1.97681099e-01 3.92937094e-01
4.33851033e-01 -3.53948861e-01 -9.07462716e-01 -2.34938949e-01
1.77066773e-01 -2.55704284e-01 8.19024086e-01 -4.28263456e-01
1.30809164e+00 -5.67279100e-01 -5.89150786e-01 3.90366375e-01
9.39382255e-01 4.35753465e-01 9.70769301e-02 2.51351893e-01
1.15969515e+00 3.38163853e-01 9.47497964e-01 7.20202684e-01
2.06343666e-01 2.35051543e-01 3.84538740e-01 2.55456239e-01
5.17619789e-01 -7.54784346e-01 5.58181465e-01 9.96226192e-01
3.31607640e-01 -5.74985921e-01 -8.59835982e-01 4.09133822e-01
-2.05797625e+00 -4.60526347e-01 -1.26943439e-01 2.15540767e+00
8.43017340e-01 6.32177234e-01 4.59329337e-02 3.84212762e-01
8.35092187e-01 3.30874056e-01 -7.21260428e-01 3.24607551e-01
3.01960409e-01 -8.84363055e-02 2.95818746e-01 2.04441160e-01
-1.16995990e+00 1.05139017e+00 4.45583534e+00 9.19775605e-01
-6.88082039e-01 1.45899147e-01 4.59962547e-01 2.69784570e-01
-7.70324543e-02 5.08979499e-01 -9.19502795e-01 6.37156546e-01
1.08364189e+00 -2.04066291e-01 2.06870139e-01 1.08091509e+00
2.14303941e-01 1.26446769e-01 -1.29036379e+00 5.80581069e-01
3.25438291e-01 -6.69046581e-01 5.19134812e-02 2.13286988e-02
5.18316329e-01 -1.14388160e-01 -6.96557283e-01 3.96784961e-01
1.34074345e-01 -4.81255978e-01 3.94165635e-01 1.56252563e-01
3.49406719e-01 -8.17139685e-01 7.19209015e-01 6.15631700e-01
-1.72009373e+00 5.32976165e-02 -1.47315621e-01 1.65168405e-01
2.08872661e-01 7.42120922e-01 -7.61510611e-01 6.75419986e-01
7.16658473e-01 9.81449544e-01 -7.72287786e-01 8.24640095e-01
-5.87301731e-01 1.21021914e+00 -4.16018695e-01 8.78879949e-02
2.26569250e-01 -1.91186786e-01 5.10387063e-01 1.20352101e+00
1.21866062e-01 9.66028944e-02 9.67230797e-01 5.58443248e-01
-2.35359743e-01 4.39622670e-01 -4.58386511e-01 -4.03266549e-01
5.01735032e-01 1.17086256e+00 -9.51891363e-01 -4.58964139e-01
-7.60634243e-01 1.18407035e+00 3.55893075e-01 2.30319798e-01
-8.18626940e-01 -5.93957663e-01 9.89859328e-02 8.00073296e-02
1.76511690e-01 -1.61232099e-01 1.42217934e-01 -1.19098592e+00
5.52141488e-01 -8.06761384e-01 1.06750274e+00 -1.26702726e-01
-1.32840574e+00 5.23968637e-01 -1.22902608e-02 -1.34992981e+00
-3.71139973e-01 -1.30346358e-01 -1.00308013e+00 3.58977556e-01
-1.19898069e+00 -8.29889178e-01 -5.82639635e-01 6.34503484e-01
6.49630547e-01 2.18656752e-02 4.26706225e-01 3.93968165e-01
-9.61577892e-01 6.39899492e-01 -1.44258469e-01 4.04728681e-01
6.17466033e-01 -1.16495502e+00 6.46024168e-01 1.34877253e+00
4.09093112e-01 4.45299268e-01 3.81375700e-01 -1.06709361e+00
-1.09954405e+00 -1.34856415e+00 9.28665876e-01 -1.80175096e-01
8.10996473e-01 -3.34705830e-01 -1.67659330e+00 1.02264047e+00
-2.92541683e-01 3.51981908e-01 7.02244878e-01 -2.03138947e-01
-4.03641582e-01 2.35731807e-02 -6.72549784e-01 2.89379478e-01
1.43923676e+00 -5.48711240e-01 -7.00100958e-01 5.25594532e-01
9.54753458e-01 -1.80279374e-01 -3.41728002e-01 6.00845277e-01
7.47477859e-02 -6.68324947e-01 5.64429879e-01 -5.01463652e-01
2.05777418e-02 -5.70463777e-01 2.45584607e-01 -1.00646818e+00
1.32184133e-01 -7.29437590e-01 -6.77815735e-01 1.56459141e+00
3.42288226e-01 -7.92772472e-01 9.65426385e-01 6.13157488e-02
-3.13389719e-01 -5.06985307e-01 -5.85516691e-01 -9.91125941e-01
-9.15182948e-01 -5.12197137e-01 3.41103882e-01 8.36454630e-01
1.92128599e-01 5.39960682e-01 -3.72751653e-01 4.93641734e-01
7.03015625e-01 1.99537069e-01 6.34170651e-01 -1.31594872e+00
-4.98611927e-01 7.59745240e-02 -2.48523310e-01 -1.14951134e+00
4.24582303e-01 -9.94372427e-01 4.49327648e-01 -1.06889307e+00
1.99712873e-01 -3.21302384e-01 -5.07753849e-01 7.34419107e-01
-7.00696409e-01 3.78161669e-03 -4.49027479e-01 6.43158376e-01
-1.08644629e+00 6.20221555e-01 5.66004097e-01 1.33814126e-01
-5.28587222e-01 2.14970440e-01 -3.77984613e-01 1.08962250e+00
7.51285255e-01 -9.10331786e-01 -5.09049535e-01 8.01433846e-02
-9.62872431e-02 -1.83106184e-01 8.69311392e-03 -8.21814418e-01
2.47981936e-01 -8.54270086e-02 1.25076681e-01 -7.62225688e-01
-3.87909651e-01 -7.93263495e-01 -2.98211724e-01 5.56944549e-01
-1.99444935e-01 8.14847350e-02 -1.71401650e-01 1.25165474e+00
-4.06024754e-01 -4.20303166e-01 5.03106713e-01 -3.51859460e-04
-6.64850533e-01 6.18248463e-01 -3.33109319e-01 4.53202814e-01
1.10977340e+00 1.64252058e-01 1.89539548e-02 -2.10854754e-01
-6.95403099e-01 5.92694581e-01 3.09478730e-01 2.81135231e-01
4.24120545e-01 -1.23169863e+00 -5.08501530e-01 2.36559331e-01
6.23614252e-01 3.30478191e-01 -2.06132531e-01 6.43864214e-01
-4.51316893e-01 1.37784705e-01 5.19180298e-01 -6.52020454e-01
-1.17193818e+00 7.99116850e-01 -3.18442285e-02 -7.19722569e-01
-8.46836984e-01 4.70443249e-01 2.00978771e-01 -7.85607174e-02
5.21697164e-01 2.41037109e-03 -1.36814892e-01 -4.80652899e-02
3.85867178e-01 2.74611205e-01 2.85707682e-01 -2.99555749e-01
-5.75252831e-01 3.60064387e-01 -4.79382455e-01 9.58462209e-02
1.00615585e+00 -1.93084389e-01 -1.42811060e-01 5.04069865e-01
1.12548828e+00 1.22501500e-01 -1.09872115e+00 -8.48677397e-01
9.96514618e-01 -3.57983172e-01 -4.48826134e-01 -1.61702320e-01
-1.09388649e+00 5.31892717e-01 3.50746870e-01 2.87589997e-01
1.26370406e+00 2.77841151e-01 1.13701618e+00 6.99726522e-01
1.51439667e-01 -7.83282876e-01 5.24393320e-01 5.35036027e-01
1.54085591e-01 -1.22617769e+00 -1.42901018e-01 -6.45431161e-01
-5.02074897e-01 7.93602705e-01 1.01498926e+00 8.72206129e-03
5.28604150e-01 -6.08895812e-03 2.28770869e-03 -2.98823327e-01
-4.96064693e-01 -2.14924395e-01 1.71580553e-01 2.11801469e-01
2.95135200e-01 -3.49718571e-01 -5.71775615e-01 5.71747839e-01
2.37550572e-01 -1.76265284e-01 2.23281085e-01 1.35516095e+00
-6.56280696e-01 -1.34525383e+00 -1.08791173e-01 5.32132447e-01
-6.80722177e-01 -2.91906223e-02 -5.56621253e-01 3.29231083e-01
1.54626379e-02 1.03298855e+00 -2.63309125e-02 -5.91665089e-01
4.29238528e-01 6.07024431e-01 -2.07399577e-01 -1.06370807e+00
-3.60494882e-01 2.67979681e-01 1.06281385e-01 -5.12981772e-01
5.63980453e-02 -5.05605280e-01 -1.69825876e+00 3.25053513e-01
-5.04261911e-01 6.33835554e-01 4.10931744e-02 1.28067482e+00
2.50159442e-01 4.20100689e-01 1.01541984e+00 -3.06311160e-01
-5.93985498e-01 -1.09488690e+00 -4.94980335e-01 6.77093148e-01
1.03102215e-01 -3.79699141e-01 -7.67512500e-01 3.70723039e-01] | [7.47288703918457, 2.4986965656280518] |
ebf7f436-bc94-47ab-a6a4-9d18e5097347 | hyperbolic-audio-source-separation | 2212.05008 | null | https://arxiv.org/abs/2212.05008v1 | https://arxiv.org/pdf/2212.05008v1.pdf | Hyperbolic Audio Source Separation | We introduce a framework for audio source separation using embeddings on a hyperbolic manifold that compactly represent the hierarchical relationship between sound sources and time-frequency features. Inspired by recent successes modeling hierarchical relationships in text and images with hyperbolic embeddings, our algorithm obtains a hyperbolic embedding for each time-frequency bin of a mixture signal and estimates masks using hyperbolic softmax layers. On a synthetic dataset containing mixtures of multiple people talking and musical instruments playing, our hyperbolic model performed comparably to a Euclidean baseline in terms of source to distortion ratio, with stronger performance at low embedding dimensions. Furthermore, we find that time-frequency regions containing multiple overlapping sources are embedded towards the center (i.e., the most uncertain region) of the hyperbolic space, and we can use this certainty estimate to efficiently trade-off between artifact introduction and interference reduction when isolating individual sounds. | ['Jonathan Le Roux', 'Aswin Subramanian', 'Gordon Wichern', 'Darius Petermann'] | 2022-12-09 | null | null | null | null | ['audio-source-separation'] | ['audio'] | [ 2.76689921e-02 1.16848715e-01 3.73248875e-01 -9.87030938e-02
-1.16511536e+00 -9.23768163e-01 2.57963568e-01 2.38147914e-01
-2.24028364e-01 2.42303498e-02 8.08013141e-01 1.37218237e-01
-6.52442157e-01 -4.07839090e-01 -3.64753336e-01 -6.05509639e-01
-6.07780218e-01 -1.17687389e-01 1.09338149e-01 2.87277699e-01
5.61217442e-02 2.54254550e-01 -1.37353277e+00 3.71570766e-01
3.98330122e-01 1.11832273e+00 -2.54241049e-01 1.19245172e+00
4.95285749e-01 6.04931414e-01 -9.41019654e-01 -2.19124377e-01
2.48235077e-01 -5.68950593e-01 -1.92170456e-01 1.03449663e-02
5.41858435e-01 -1.22929625e-01 -7.86954284e-01 7.68922865e-01
8.72877002e-01 2.39520803e-01 8.18619370e-01 -1.42488551e+00
-7.93472111e-01 1.05030215e+00 -3.65009665e-01 2.31268689e-01
1.58073410e-01 -1.05430156e-01 1.26915848e+00 -1.22646546e+00
1.58881634e-01 1.04773974e+00 9.58892643e-01 -8.62715989e-02
-1.56682599e+00 -6.55388951e-01 -3.66901726e-01 2.04917148e-01
-1.69059646e+00 -6.62070990e-01 1.05825794e+00 -5.76987445e-01
6.72475159e-01 4.01662856e-01 4.88183796e-01 1.00515819e+00
9.06614289e-02 5.46633780e-01 6.03527308e-01 -2.84690112e-01
2.33158424e-01 3.54571909e-01 -7.61566758e-02 2.42893353e-01
-2.66891092e-01 1.26327336e-01 -9.51488256e-01 -5.51694334e-01
6.88173473e-01 -2.94939607e-01 -5.00896156e-01 -1.67624980e-01
-1.26568925e+00 5.59839010e-01 4.10718888e-01 3.50182384e-01
-1.67239398e-01 2.64451951e-01 1.23947166e-01 9.50211808e-02
2.97845960e-01 8.38587582e-01 -1.75852761e-01 -2.67135091e-02
-1.35857201e+00 9.69094932e-02 6.33417785e-01 6.81889594e-01
1.71060666e-01 3.42895120e-01 -1.65325537e-01 1.02328444e+00
3.23855966e-01 2.42728174e-01 3.41704369e-01 -1.40966713e+00
1.68482587e-01 2.22560018e-02 1.01823017e-01 -1.16819429e+00
-6.09415293e-01 -7.34356523e-01 -5.86424828e-01 3.21288198e-01
7.10296988e-01 -3.76919985e-01 -4.62542653e-01 1.83841693e+00
1.18305251e-01 4.09109145e-01 5.08503467e-02 9.14528608e-01
5.50074279e-01 8.61893594e-01 -4.99577999e-01 -1.03607811e-01
1.53655016e+00 -5.98371029e-01 -8.12074184e-01 4.22805957e-02
-1.28835440e-01 -8.43472958e-01 8.87940168e-01 6.30020916e-01
-1.37498248e+00 -6.51813328e-01 -1.45947313e+00 -2.79846042e-01
-2.59971708e-01 7.00947642e-02 -7.82312676e-02 5.22824764e-01
-8.67478907e-01 8.09425116e-01 -4.66331214e-01 3.56030464e-01
6.50085285e-02 5.96449338e-02 -8.87839720e-02 4.56024140e-01
-1.18625212e+00 2.56859273e-01 7.11201355e-02 -9.97740701e-02
-8.13644409e-01 -1.22074771e+00 -8.79465580e-01 3.59634608e-01
-7.24414662e-02 -1.27261624e-01 1.13810992e+00 -5.37312806e-01
-1.29915106e+00 1.54873952e-01 2.30371147e-01 -5.56576073e-01
3.70257586e-01 -1.47496387e-01 -6.36902809e-01 2.71051049e-01
-1.61645040e-01 6.48118913e-01 1.36343539e+00 -1.17975616e+00
-5.22372425e-01 -2.92224675e-01 -1.78707808e-01 1.41651645e-01
-6.60722613e-01 3.74487713e-02 -6.92160474e-03 -1.14774585e+00
4.40182805e-01 -8.30573499e-01 2.24899977e-01 2.68610299e-01
-4.43262100e-01 -1.04287617e-01 7.55492032e-01 -8.50370347e-01
1.23827863e+00 -2.72135639e+00 2.75284886e-01 1.78896189e-01
5.57510436e-01 -3.64622951e-01 -8.36812705e-02 3.86652380e-01
-1.36977181e-01 7.47851431e-02 -2.77783215e-01 -5.87345898e-01
4.17255133e-01 -5.12913287e-01 -5.82513452e-01 7.99188733e-01
2.04918489e-01 2.42775321e-01 -7.38714099e-01 -3.66748750e-01
1.61205325e-02 1.01740336e+00 -7.90338218e-01 2.48947680e-01
2.26750359e-01 -1.06775895e-01 4.98968959e-01 4.84843582e-01
6.19560480e-01 9.50570181e-02 -2.71526724e-01 -2.42299095e-01
-9.42526385e-02 3.07166308e-01 -1.65916526e+00 1.53415561e+00
-4.04533118e-01 1.50035203e+00 3.96793544e-01 -3.12873453e-01
7.82193542e-01 7.83332467e-01 7.07956731e-01 -8.91113579e-02
1.80974677e-01 7.89788179e-03 1.61680385e-01 -2.92904109e-01
4.88218278e-01 -1.13829277e-01 -5.98468520e-02 5.73263168e-01
5.21993041e-01 -3.66925448e-01 -3.67158830e-01 6.34777173e-02
1.11558104e+00 -4.49695408e-01 -1.45668641e-01 -1.76486239e-01
-9.36730430e-02 -6.75834060e-01 2.55823165e-01 5.50387025e-01
-4.19440329e-01 1.17503870e+00 6.66755974e-01 4.62689772e-02
-9.96354938e-01 -1.75678337e+00 -3.89349282e-01 1.38285911e+00
-1.29930198e-01 -6.01496100e-01 -7.64030635e-01 -9.33523402e-02
6.76407889e-02 8.80345881e-01 -5.84582806e-01 -2.93961406e-01
-4.00832206e-01 -4.63812232e-01 9.84957576e-01 6.28009081e-01
-1.67952448e-01 -7.17670560e-01 -6.16813540e-01 2.93699950e-01
-1.57599390e-01 -1.06701624e+00 -9.61717486e-01 7.09714651e-01
-2.27502018e-01 -7.34422028e-01 -7.00369596e-01 -6.34973943e-01
-4.77531776e-02 1.46714836e-01 9.00338829e-01 -6.99920893e-01
-3.79194528e-01 5.11256576e-01 -1.71449244e-01 -6.90730095e-01
-2.37916723e-01 -3.78234059e-01 3.19606036e-01 2.38995150e-01
-1.38499275e-01 -8.81341279e-01 -7.43168890e-01 2.28600174e-01
-7.81490386e-01 -4.78002340e-01 1.49002835e-01 5.11180162e-01
1.35754079e-01 6.42160118e-01 5.94394207e-01 1.71788454e-01
8.69903207e-01 -6.02878690e-01 -3.35357785e-01 -2.31862977e-01
-1.66335732e-01 -3.97809334e-02 6.10791862e-01 -8.81653547e-01
-6.17240965e-01 -1.21746831e-01 1.74832672e-01 -6.05860949e-01
-3.98305105e-03 -1.41890556e-01 -5.11294864e-02 2.76368082e-01
9.61885333e-01 -2.88679097e-02 -3.85089785e-01 -3.68700981e-01
7.00734675e-01 8.91735077e-01 7.99429595e-01 -1.40546545e-01
6.09970272e-01 4.68711227e-01 -4.73772436e-01 -1.23128474e+00
-3.35937381e-01 -3.16859841e-01 -3.96156669e-01 -5.89208640e-02
9.31111872e-01 -8.11981916e-01 -6.60483301e-01 3.18365470e-02
-1.24098408e+00 -4.32726443e-02 -7.30214059e-01 8.44230056e-01
-6.16696358e-01 -1.43174827e-01 -6.65766537e-01 -1.19631910e+00
7.63491839e-02 -8.93088043e-01 1.20039117e+00 -6.99409423e-03
-7.23214507e-01 -7.26790309e-01 2.87136555e-01 -2.17447449e-02
8.20290893e-02 8.50579143e-02 9.57697213e-01 -4.35746700e-01
-2.30669111e-01 -2.11089715e-01 8.60853568e-02 5.45992136e-01
8.87304768e-02 1.75872296e-01 -1.68297243e+00 -7.08721429e-02
3.56945544e-01 -2.55276747e-02 8.18616450e-01 7.31642962e-01
9.50592220e-01 -4.95527297e-01 2.40351539e-02 7.13992774e-01
1.02314329e+00 1.83610231e-01 3.38216096e-01 -2.19578594e-01
4.42161262e-01 7.53996015e-01 -2.65300512e-01 6.19098842e-01
1.23216033e-01 5.96617937e-01 1.28569379e-01 -1.41462475e-01
-3.14238787e-01 -2.95374542e-01 4.95751768e-01 7.43314862e-01
5.30531585e-01 -3.48654240e-01 -6.02158010e-01 7.84798741e-01
-1.35115457e+00 -9.86632884e-01 3.11661333e-01 2.28780127e+00
9.24557328e-01 2.80760884e-01 5.66509426e-01 8.82306814e-01
5.29683053e-01 3.02734911e-01 -5.12183547e-01 -2.10569739e-01
-3.23396325e-01 4.86796945e-02 2.64712125e-01 7.66187549e-01
-1.26550806e+00 2.45140925e-01 7.35660648e+00 5.97761989e-01
-8.32033932e-01 4.28174734e-02 1.47356763e-01 -7.71548450e-01
-2.24932134e-01 -4.95058864e-01 -2.99484372e-01 4.84224439e-01
1.09010363e+00 -3.87044072e-01 6.26213133e-01 3.97829145e-01
1.16171002e-01 4.40882176e-01 -1.47908103e+00 1.14520550e+00
1.61710069e-01 -1.13065124e+00 -4.39891011e-01 1.32207070e-02
4.27222669e-01 -2.65516043e-01 7.71458983e-01 2.45607235e-02
1.69560105e-01 -1.25264919e+00 1.23115659e+00 1.85286865e-01
4.85300720e-01 -7.53183663e-01 2.36563444e-01 2.21770838e-01
-1.27469993e+00 -4.96605366e-01 -2.72401154e-01 2.02654526e-01
2.28983685e-02 5.36514223e-01 -1.08817828e+00 -6.18159622e-02
9.29877341e-01 3.01801562e-01 -4.26455438e-01 1.04747808e+00
7.61869028e-02 5.83978415e-01 -4.18928534e-01 3.78007412e-01
1.90135427e-02 1.84192825e-02 8.55740309e-01 1.50176287e+00
4.46838289e-01 -3.64508480e-02 -2.46356651e-01 1.05301154e+00
-1.02627225e-01 -1.33443698e-01 -5.93584538e-01 -4.98734452e-02
7.81915188e-01 1.11316049e+00 -6.40629351e-01 -4.27268893e-02
-2.46526673e-01 7.90462255e-01 -3.74511853e-02 4.40697610e-01
-1.07943821e+00 -8.22775126e-01 9.37856197e-01 3.28116983e-01
2.48532712e-01 -3.95283103e-01 -3.63632232e-01 -5.61274171e-01
9.91275162e-02 -7.40355849e-01 4.26716328e-01 -9.03139710e-01
-1.03989112e+00 8.09797704e-01 -8.78863782e-02 -1.35455930e+00
-1.64892256e-01 -3.86426717e-01 -4.09211308e-01 5.36010683e-01
-5.93941152e-01 -5.09140790e-01 1.64048478e-01 4.70428824e-01
4.71132100e-01 1.47513717e-01 6.92570686e-01 2.94670969e-01
-1.92938477e-01 8.90670002e-01 2.88950741e-01 1.62928358e-01
8.16764414e-01 -1.61343968e+00 1.37452170e-01 6.49323106e-01
9.68274117e-01 3.35632861e-01 9.55495179e-01 1.51199521e-03
-9.56565619e-01 -8.74461710e-01 5.37670910e-01 -6.69487536e-01
7.62725294e-01 -9.36243176e-01 -8.75517488e-01 4.64304447e-01
4.01342422e-01 -2.57533733e-02 1.15582860e+00 8.79177824e-02
-7.30575144e-01 -1.73260331e-01 -9.90327001e-01 7.47803092e-01
7.70915568e-01 -1.06305635e+00 -6.35896921e-01 3.00946116e-01
1.07282376e+00 -1.15897648e-01 -9.51255977e-01 9.42708403e-02
8.74654531e-01 -1.05336344e+00 1.18963003e+00 -3.81111979e-01
1.95191786e-01 -4.38050210e-01 -5.83212554e-01 -1.54646957e+00
-5.92585027e-01 -1.06291425e+00 -3.09432864e-01 1.28114200e+00
4.78073239e-01 -3.93513031e-02 3.51448417e-01 1.40117943e-01
-2.17248444e-02 -4.79914546e-01 -1.24305320e+00 -9.56162393e-01
1.49115309e-01 -7.51905918e-01 3.52071673e-01 6.55371428e-01
3.71546358e-01 3.10072184e-01 -3.42520028e-01 7.98556209e-01
8.27734649e-01 -1.35842308e-01 1.79230720e-01 -1.11339891e+00
-4.13474053e-01 -7.26665676e-01 -6.21874571e-01 -7.33499467e-01
-2.83385701e-02 -6.81729555e-01 3.64631176e-01 -1.07286596e+00
-3.71826708e-01 -8.48041624e-02 -6.35344326e-01 1.24704577e-02
1.11063488e-01 4.51533198e-01 2.49606267e-01 -1.99959517e-01
-1.72187477e-01 6.03119969e-01 5.52627385e-01 -4.38475192e-01
-5.31569183e-01 -1.28325179e-01 -7.84526289e-01 9.70119119e-01
4.56245959e-01 -7.09099591e-01 -5.48677146e-01 -6.13932312e-01
-2.72748303e-02 2.43205890e-01 3.56773168e-01 -1.40781331e+00
3.92648131e-01 3.30021352e-01 6.30768597e-01 -3.61155480e-01
9.44451332e-01 -9.13999498e-01 1.39934599e-01 -4.57403325e-02
-7.67807364e-01 -1.76226765e-01 2.78379947e-01 5.51174939e-01
-1.63251847e-01 1.52437672e-01 9.25818801e-01 4.33121473e-01
1.98348865e-01 -1.30478948e-01 -5.55805027e-01 2.46640846e-01
8.53156984e-01 -7.76110217e-02 -7.36853182e-02 -8.14720690e-01
-8.80848527e-01 -1.87650770e-01 -4.80672754e-02 7.29808271e-01
6.14030361e-01 -1.58115709e+00 -8.46808255e-01 3.84578824e-01
-8.29315558e-02 -4.53259677e-01 3.02574366e-01 5.08462846e-01
-2.26441696e-02 -7.86652490e-02 1.04597457e-01 -6.61526680e-01
-1.23275793e+00 5.71066141e-01 5.82688510e-01 5.07368863e-01
-5.94895422e-01 1.25944376e+00 2.79523462e-01 3.07760071e-02
8.46490562e-01 -6.84902608e-01 3.54404859e-02 4.19804364e-01
7.84666657e-01 8.65140855e-01 -1.52634963e-01 -6.23606622e-01
-5.35293162e-01 4.94515657e-01 6.25836492e-01 -9.29726899e-01
1.03402853e+00 2.63105966e-02 1.30662516e-01 9.51886714e-01
1.63802218e+00 5.83602965e-01 -1.50904346e+00 -2.83575863e-01
-8.97146314e-02 -4.20439333e-01 1.35367990e-01 -4.74178225e-01
-7.83914983e-01 9.68712807e-01 9.07221973e-01 8.56907606e-01
1.01863742e+00 2.45106280e-01 3.94962311e-01 -4.02589440e-02
-2.98839152e-01 -1.07792187e+00 4.21236873e-01 1.19213551e-01
1.18807268e+00 -7.00233281e-01 -8.39689374e-02 -1.00536302e-01
-4.44294870e-01 1.07739699e+00 -6.90095918e-03 -2.92467564e-01
1.06235147e+00 9.00007010e-01 2.69066960e-01 1.10040186e-02
-6.67240381e-01 2.27776587e-01 5.58042467e-01 7.12253749e-01
3.28716576e-01 2.17394426e-01 6.93031609e-01 1.03578901e+00
-8.54801595e-01 -7.44592369e-01 4.36710268e-01 3.62041473e-01
-5.07613361e-01 -3.06435257e-01 -8.61025810e-01 2.41375025e-02
-4.06308144e-01 -2.29091376e-01 -4.68483448e-01 3.92557949e-01
4.33097839e-01 1.49398577e+00 4.33376700e-01 -6.96104467e-01
5.80243409e-01 3.00950527e-01 4.09395188e-01 -2.18246505e-01
-4.89879310e-01 7.87695944e-01 -2.89738715e-01 -4.18327123e-01
3.84031415e-01 -7.06033468e-01 -1.08747756e+00 3.08989342e-02
-2.19200894e-01 2.53590256e-01 5.24525642e-01 3.10577959e-01
2.46657655e-01 8.75765264e-01 8.73721838e-01 -9.30544496e-01
-7.34015882e-01 -1.01058602e+00 -1.03833914e+00 1.35574967e-01
9.83714521e-01 -3.89708668e-01 -8.34302843e-01 2.38453150e-01] | [15.377578735351562, 5.6021223068237305] |
71a3be29-a34a-4067-999b-55d5360c45fe | neural-data-to-text-generation-with-lm-based | 2102.03556 | null | https://arxiv.org/abs/2102.03556v1 | https://arxiv.org/pdf/2102.03556v1.pdf | Neural Data-to-Text Generation with LM-based Text Augmentation | For many new application domains for data-to-text generation, the main obstacle in training neural models consists of a lack of training data. While usually large numbers of instances are available on the data side, often only very few text samples are available. To address this problem, we here propose a novel few-shot approach for this setting. Our approach automatically augments the data available for training by (i) generating new text samples based on replacing specific values by alternative ones from the same category, (ii) generating new text samples based on GPT-2, and (iii) proposing an automatic method for pairing the new text samples with data samples. As the text augmentation can introduce noise to the training data, we use cycle consistency as an objective, in order to make sure that a given data sample can be correctly reconstructed after having been formulated as text (and that text samples can be reconstructed from data). On both the E2E and WebNLG benchmarks, we show that this weakly supervised training paradigm is able to outperform fully supervised seq2seq models with less than 10% annotations. By utilizing all annotated data, our model can boost the performance of a standard seq2seq model by over 5 BLEU points, establishing a new state-of-the-art on both datasets. | ['Hui Su', 'Vera Demberg', 'Dawei Zhu', 'Xiaoyu Shen', 'Ernie Chang'] | 2021-02-06 | null | https://aclanthology.org/2021.eacl-main.64 | https://aclanthology.org/2021.eacl-main.64.pdf | eacl-2021-2 | ['text-augmentation', 'data-to-text-generation'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.88757718e-01 3.75710875e-01 -1.34798065e-01 -4.13366705e-01
-1.12210548e+00 -6.03468060e-01 7.92241991e-01 4.10796255e-01
-6.02997780e-01 1.15874636e+00 3.16698849e-01 -1.39157280e-01
2.81152546e-01 -8.53632390e-01 -1.07539892e+00 -5.24293184e-01
5.56847572e-01 9.21865344e-01 4.04373296e-02 -2.94854283e-01
-5.03503680e-02 -5.46181574e-02 -1.78936934e+00 4.78319496e-01
1.08975470e+00 8.38722289e-01 4.56885755e-01 6.48828685e-01
-6.86358631e-01 4.56890911e-01 -7.54678071e-01 -5.75804830e-01
2.71900088e-01 -7.51759768e-01 -7.98432112e-01 -8.98295641e-03
3.50762069e-01 -2.45150566e-01 7.08544478e-02 8.65243673e-01
7.45309472e-01 2.51939893e-01 5.63480973e-01 -1.04195952e+00
-4.06316400e-01 1.21937406e+00 -2.14451998e-01 -3.82057935e-01
1.12665348e-01 5.75811453e-02 1.18493533e+00 -9.76971686e-01
8.10019016e-01 8.24790955e-01 4.97855276e-01 1.07797372e+00
-1.32848334e+00 -4.57164884e-01 -2.43500099e-02 -9.29637477e-02
-9.85100687e-01 -8.14638197e-01 4.55624253e-01 -5.36033958e-02
8.21210325e-01 4.39144164e-01 4.68570024e-01 1.42409980e+00
-2.28723422e-01 1.05954373e+00 7.44608045e-01 -7.09932506e-01
4.65089321e-01 1.17739901e-01 1.23764269e-01 1.54435679e-01
1.58005327e-01 -6.52282685e-02 -7.52181888e-01 8.41838196e-02
3.89948264e-02 -1.99792907e-01 -5.74944913e-02 1.13666348e-01
-1.26573265e+00 6.70870364e-01 2.53450330e-02 3.89330447e-01
-3.79410416e-01 1.11668132e-01 6.46772563e-01 9.81199667e-02
6.90134823e-01 4.82580394e-01 -6.89313948e-01 -4.99531299e-01
-1.15095544e+00 3.96510273e-01 8.71871769e-01 1.16754413e+00
6.70530498e-01 -1.69695038e-02 -7.36770272e-01 1.04409277e+00
-1.42187089e-01 5.84936321e-01 8.23641360e-01 -4.38874334e-01
9.02565300e-01 4.67595279e-01 -2.14549247e-02 -1.86770812e-01
-3.99277732e-02 -3.17016065e-01 -1.03883815e+00 -2.13187158e-01
4.18993205e-01 -3.41911614e-01 -9.85628307e-01 1.78226626e+00
2.44989499e-01 4.86320034e-02 3.95428419e-01 4.20398146e-01
8.39184105e-01 9.54672337e-01 -1.12431400e-01 -3.36620420e-01
1.12476838e+00 -1.00997710e+00 -8.21507096e-01 -4.78224605e-01
1.08153319e+00 -6.28238022e-01 1.18375123e+00 4.06505585e-01
-9.95557845e-01 -5.95845103e-01 -7.67244041e-01 -9.57059190e-02
-3.68514657e-01 3.52584541e-01 2.65825987e-01 6.79867864e-01
-8.44699562e-01 8.03517342e-01 -5.87972760e-01 -4.08075213e-01
3.60832810e-01 -1.03789847e-02 -3.98151189e-01 -2.66548693e-01
-1.14501131e+00 5.07393420e-01 8.57151270e-01 3.07125580e-02
-6.80430293e-01 -8.59957099e-01 -8.06240439e-01 1.20400213e-01
4.85085875e-01 -5.68879128e-01 1.36141956e+00 -9.84627783e-01
-1.57658076e+00 5.47081292e-01 -3.41154158e-01 -6.97199643e-01
6.94490075e-01 -2.04456300e-01 -3.03684324e-01 -3.10462177e-01
8.12894180e-02 9.80631351e-01 6.91877782e-01 -1.12125885e+00
-7.99467921e-01 -1.49895668e-01 -3.43343973e-01 2.08701029e-01
-5.16147554e-01 -2.33064964e-01 -5.34524500e-01 -7.69191086e-01
-2.74550259e-01 -8.28313112e-01 -9.46978629e-02 -1.89432234e-01
-7.49030292e-01 -3.54002148e-01 6.28232896e-01 -7.42373884e-01
1.12077022e+00 -1.91567504e+00 3.44722979e-02 -1.00499414e-01
-4.44862060e-02 5.42693734e-01 -4.92413998e-01 4.74145949e-01
-4.91760112e-02 3.19780767e-01 -5.36198497e-01 -7.85783112e-01
4.18727547e-02 4.24390197e-01 -5.89949608e-01 -2.77275264e-01
4.02649969e-01 9.73640263e-01 -1.08599150e+00 -2.84759700e-01
2.70897765e-02 3.45966145e-02 -6.36949003e-01 4.05747950e-01
-6.82418823e-01 3.22247297e-01 -4.60438840e-02 9.85516831e-02
5.77029884e-01 -2.91698873e-02 1.78608924e-01 3.86638232e-02
1.48880750e-01 6.38726711e-01 -1.20059800e+00 1.81703222e+00
-6.38136983e-01 5.75003684e-01 -5.10893464e-01 -1.02721310e+00
9.17079329e-01 4.57107604e-01 2.36969218e-01 -7.37986207e-01
-7.27803707e-02 3.15303564e-01 -1.08397357e-01 -3.43509287e-01
8.96007061e-01 -2.41073936e-01 -8.20544809e-02 5.89419365e-01
3.48568797e-01 -9.98920649e-02 8.11699212e-01 1.22841291e-01
1.06511176e+00 1.96287781e-01 7.86164925e-02 1.37789935e-01
2.75765389e-01 -1.30337566e-01 5.23281276e-01 8.47592056e-01
5.04282534e-01 6.50009215e-01 5.86722672e-01 -1.28765658e-01
-1.40527868e+00 -6.43381655e-01 3.25865224e-02 1.11998594e+00
-4.48912233e-01 -5.43685138e-01 -9.11384284e-01 -1.12843311e+00
-2.48741582e-01 1.31776321e+00 -6.75186872e-01 -2.42469057e-01
-4.29731578e-01 -6.58113718e-01 8.04389596e-01 5.12335062e-01
3.11149836e-01 -1.11877644e+00 -2.18512788e-01 4.00653213e-01
-6.00696683e-01 -1.05818844e+00 -6.59525573e-01 5.23156106e-01
-9.33213115e-01 -6.70883417e-01 -7.12869167e-01 -5.33419549e-01
7.85605490e-01 -5.88952601e-02 1.18946540e+00 -1.23458356e-01
-2.75060553e-02 -3.01773548e-01 -7.61300802e-01 -4.87175614e-01
-1.03308642e+00 4.49520737e-01 -1.64446041e-01 6.65342361e-02
1.07644826e-01 -2.12712392e-01 -7.47951642e-02 2.48612799e-02
-1.23341560e+00 4.60586846e-01 6.89558804e-01 1.15553904e+00
6.12629890e-01 -5.62729277e-02 8.48395526e-01 -1.17772460e+00
3.98081064e-01 -3.91747624e-01 -4.12569761e-01 1.25574961e-01
-5.87374449e-01 4.41865414e-01 1.04476714e+00 -4.48072135e-01
-1.05437064e+00 7.31977075e-02 -5.43974459e-01 -1.35775238e-01
-2.03495428e-01 7.29880154e-01 -3.15625817e-01 6.69427991e-01
7.77175486e-01 5.37543118e-01 -3.31383683e-02 -5.25963008e-01
5.97632468e-01 8.49342108e-01 4.22514856e-01 -4.62645590e-01
6.40195906e-01 5.14016300e-02 -1.93592250e-01 -8.02858055e-01
-1.03467643e+00 -3.17240655e-01 -6.55010700e-01 -6.43662140e-02
5.98622739e-01 -7.31408715e-01 -1.51394472e-01 5.02731383e-01
-1.22816217e+00 -4.93050605e-01 -7.88810074e-01 2.31370822e-01
-4.21609491e-01 3.99870992e-01 -2.15678349e-01 -8.30629766e-01
-6.29212916e-01 -8.37558448e-01 1.14866722e+00 -1.30193263e-01
-2.41200477e-01 -8.46941769e-01 1.62189081e-01 2.31771871e-01
1.90243065e-01 -1.23938262e-01 8.70792627e-01 -1.29916596e+00
-1.46953702e-01 -3.71166885e-01 1.07359514e-01 6.50929451e-01
2.12796658e-01 -1.01554945e-01 -1.00212085e+00 -1.88200399e-01
-1.57466576e-01 -4.75086182e-01 8.89126241e-01 5.67803858e-03
1.33283687e+00 -5.55480838e-01 -2.62981236e-01 3.01316738e-01
1.16078293e+00 -5.17239906e-02 7.74147570e-01 -1.13182418e-01
6.12566054e-01 6.86160624e-01 6.27513468e-01 4.98807013e-01
1.75361961e-01 6.99431300e-01 2.81352460e-01 1.44590437e-01
-1.39440030e-01 -6.04683459e-01 4.43385452e-01 9.03102517e-01
3.69130135e-01 -8.31293344e-01 -8.41854572e-01 6.44722283e-01
-1.80983543e+00 -9.88862038e-01 -2.81901062e-01 2.50601363e+00
1.36097193e+00 2.55916715e-01 -6.14850828e-03 3.59269112e-01
6.89024568e-01 -8.23348686e-02 -4.87366647e-01 -1.06700957e-01
-1.27171874e-01 4.51406270e-01 4.31491375e-01 2.00711414e-01
-8.39149892e-01 8.76981378e-01 5.52956438e+00 1.19960380e+00
-1.10140872e+00 4.36886474e-02 6.90997779e-01 -2.16878965e-01
-5.14704168e-01 -7.85443857e-02 -1.20100677e+00 8.47366512e-01
1.22199023e+00 -3.10130298e-01 3.30328673e-01 7.19637692e-01
2.55811483e-01 -1.64346751e-02 -1.40790951e+00 8.05232584e-01
1.49351954e-01 -1.34203136e+00 4.70858574e-01 -1.16659604e-01
7.87083685e-01 -1.74692124e-01 -3.06311101e-01 6.20251358e-01
4.04821664e-01 -1.04003811e+00 8.59256029e-01 4.22154963e-01
1.00902724e+00 -5.91991305e-01 9.22772646e-01 7.42849231e-01
-8.00798476e-01 1.85163096e-01 -4.59720880e-01 2.28206635e-01
1.22618668e-01 1.08848035e+00 -1.20295382e+00 8.38274658e-01
2.43359432e-01 7.59225488e-01 -4.78851646e-01 9.87908065e-01
-2.54980713e-01 7.99160659e-01 -3.82685065e-01 -2.95366496e-01
1.39938846e-01 8.41956660e-02 4.13380533e-01 1.21151948e+00
6.43844128e-01 -2.87574768e-01 -4.22207266e-02 9.94765401e-01
-5.43574989e-01 2.81348526e-01 -5.17866135e-01 -2.83011287e-01
5.14597058e-01 1.26981056e+00 -4.32175696e-01 -5.75033963e-01
-1.88794896e-01 1.20266235e+00 3.48596424e-01 6.84276149e-02
-5.77760279e-01 -5.15319288e-01 2.42632821e-01 -1.25079066e-01
3.81981134e-01 2.09764645e-01 -3.32633018e-01 -1.13038695e+00
3.32495600e-01 -9.39991355e-01 2.10604802e-01 -7.24688470e-01
-1.32457817e+00 5.20699799e-01 -3.84151518e-01 -1.38296700e+00
-6.90610230e-01 -2.98158705e-01 -4.54396129e-01 1.03813446e+00
-1.26308477e+00 -9.70162928e-01 -2.58257151e-01 1.59649938e-01
8.71879041e-01 -1.43903881e-01 8.98649633e-01 3.02736193e-01
-4.16618764e-01 8.67222428e-01 5.15309572e-01 1.57153785e-01
8.91231954e-01 -1.41625392e+00 9.10491586e-01 9.31134224e-01
2.49662593e-01 1.95036471e-01 6.95414543e-01 -7.85123765e-01
-1.31875706e+00 -1.53348660e+00 1.23669815e+00 -3.68320256e-01
3.74186695e-01 -7.06261098e-01 -1.03768218e+00 5.49126029e-01
-1.58814881e-02 -1.00400135e-01 5.27373135e-01 -3.99280898e-02
-1.23109616e-01 -1.53789401e-01 -9.40502822e-01 6.64997458e-01
1.00811863e+00 -2.17525944e-01 -5.18217146e-01 3.46043140e-01
9.38238621e-01 -5.46709001e-01 -5.28701305e-01 1.82941556e-01
2.36028627e-01 -5.69477499e-01 3.64671052e-01 -7.42936075e-01
8.55922580e-01 -6.27128705e-02 -8.54815990e-02 -1.74395001e+00
2.12758973e-01 -5.85841775e-01 -1.57940611e-01 1.60004520e+00
7.11650014e-01 -3.81945789e-01 8.35087299e-01 3.30063641e-01
-3.97392958e-01 -6.16648674e-01 -1.06013262e+00 -1.00555515e+00
2.23196417e-01 -6.57774508e-01 6.32679701e-01 8.35579336e-01
-3.83218601e-02 5.56337655e-01 -5.36517262e-01 -4.16492909e-01
2.89437979e-01 -3.46939385e-01 1.02209198e+00 -1.12756002e+00
-3.57594579e-01 -3.30605477e-01 1.09237574e-01 -1.23691237e+00
1.44051701e-01 -1.28733790e+00 5.65542161e-01 -1.64306736e+00
2.75070280e-01 -6.12892032e-01 1.89800814e-01 7.24604964e-01
-5.03011048e-01 5.23879491e-02 1.69957310e-01 -8.68891925e-02
-3.41664851e-01 7.89477050e-01 9.52386737e-01 -9.25930869e-03
-1.60108998e-01 3.90378274e-02 -7.55361915e-01 2.10345820e-01
6.76237762e-01 -5.70230842e-01 -2.87864864e-01 -4.59353268e-01
4.77002442e-01 1.02269702e-01 7.95694664e-02 -9.78588641e-01
8.74352157e-02 1.76520236e-02 2.06021935e-01 -7.74064779e-01
4.92874123e-02 -7.76805818e-01 5.04279323e-02 3.32541078e-01
-8.00099254e-01 -3.20516169e-01 3.85921896e-01 5.63867092e-01
-6.12085797e-02 -7.35374629e-01 5.91808081e-01 1.06068570e-02
-3.56638759e-01 2.30668277e-01 -3.52555603e-01 4.04237062e-01
7.22610116e-01 1.20121110e-02 -3.26856345e-01 -2.38530919e-01
-4.42200571e-01 2.31430098e-01 2.71110922e-01 4.97703016e-01
3.00757736e-01 -1.37855077e+00 -9.92676973e-01 3.05724055e-01
4.66147393e-01 3.08224201e-01 2.31131554e-01 4.96578068e-01
3.78807448e-03 4.45106983e-01 1.00814074e-01 -4.86209631e-01
-1.17725754e+00 3.97367299e-01 1.31844327e-01 -6.73856795e-01
-4.00728941e-01 6.14684820e-01 -4.11434844e-02 -8.82630467e-01
2.02721506e-01 -3.80295664e-01 -5.74774444e-02 1.92701682e-01
6.26351953e-01 2.20059574e-01 5.41172266e-01 -8.28061402e-02
2.30660483e-01 -7.45222420e-02 -2.69766897e-01 -3.78724635e-01
1.37224507e+00 9.81839895e-02 4.17269282e-02 7.02012599e-01
1.05958521e+00 2.70233117e-02 -1.06397092e+00 -3.91538620e-01
-2.98558380e-02 -3.01034510e-01 -2.09934205e-01 -1.11272824e+00
-7.24123836e-01 9.44500446e-01 2.72242665e-01 1.89285308e-01
1.05990195e+00 -9.43478420e-02 9.87877786e-01 6.58329368e-01
7.70568028e-02 -1.21214557e+00 1.21784888e-01 8.97531927e-01
7.87161648e-01 -1.14425194e+00 -4.74431843e-01 -1.80166543e-01
-7.04116344e-01 1.11658454e+00 5.19222558e-01 3.78390431e-01
4.53236252e-02 1.93293244e-01 -3.06970954e-01 3.74039114e-01
-1.09646416e+00 -1.02861509e-01 3.37988853e-01 6.02548718e-01
5.82847476e-01 -1.22227937e-01 -3.45110506e-01 7.52942860e-01
-4.71636236e-01 6.13438934e-02 6.44106984e-01 7.19289780e-01
-4.20833051e-01 -1.49316680e+00 -6.73035011e-02 9.87414539e-01
-4.28566456e-01 -5.29660285e-01 -3.27153265e-01 4.54942018e-01
1.11847576e-02 7.83263445e-01 2.52034068e-01 -2.55350202e-01
3.48418891e-01 5.38411736e-01 2.15258792e-01 -9.86196637e-01
-4.81159568e-01 -2.14971915e-01 4.30610031e-01 -8.73306915e-02
-7.50509091e-03 -7.09268451e-01 -1.16477835e+00 -1.67862996e-01
-3.51466656e-01 2.39913374e-01 8.32575440e-01 1.11137843e+00
4.14359421e-01 6.79755449e-01 6.57236278e-01 -6.59771323e-01
-6.59252107e-01 -1.32042551e+00 -3.74763668e-01 4.81778979e-01
8.89190212e-02 -6.26751110e-02 -2.21324399e-01 4.93067384e-01] | [11.659449577331543, 8.964717864990234] |
16beae60-2438-4dcc-8e94-835f51a942ab | seismic-data-interpolation-based-on-denoising | 2307.04226 | null | https://arxiv.org/abs/2307.04226v1 | https://arxiv.org/pdf/2307.04226v1.pdf | Seismic Data Interpolation based on Denoising Diffusion Implicit Models with Resampling | The incompleteness of the seismic data caused by missing traces along the spatial extension is a common issue in seismic acquisition due to the existence of obstacles and economic constraints, which severely impairs the imaging quality of subsurface geological structures. Recently, deep learning-based seismic interpolation methods have attained promising progress, while achieving stable training of generative adversarial networks is not easy, and performance degradation is usually notable if the missing patterns in the testing and training do not match. In this paper, we propose a novel seismic denoising diffusion implicit model with resampling. The model training is established on the denoising diffusion probabilistic model, where U-Net is equipped with the multi-head self-attention to match the noise in each step. The cosine noise schedule, serving as the global noise configuration, promotes the high utilization of known trace information by accelerating the passage of the excessive noise stages. The model inference utilizes the denoising diffusion implicit model, conditioning on the known traces, to enable high-quality interpolation with fewer diffusion steps. To enhance the coherency between the known traces and the missing traces within each reverse step, the inference process integrates a resampling strategy to achieve an information recap on the former interpolated traces. Extensive experiments conducted on synthetic and field seismic data validate the superiority of our model and its robustness on various missing patterns. In addition, uncertainty quantification and ablation studies are also investigated. | ['Sang-Woon Kim', 'Jiangshe Zhang', 'Baisong Jiang', 'Deng Xiong', 'Chengli Tan', 'Hongtao Wang', 'Chunxia Zhang', 'Xiaoli Wei'] | 2023-07-09 | null | null | null | null | ['denoising'] | ['computer-vision'] | [ 6.48498833e-02 -8.77432004e-02 3.08882505e-01 -9.62992013e-02
-9.49477315e-01 -2.87119355e-02 4.53286767e-01 -1.63622200e-01
-3.67858261e-01 8.50700438e-01 3.66117626e-01 3.40754874e-02
-4.06972677e-01 -1.07929075e+00 -8.32618654e-01 -1.23451221e+00
-6.77817017e-02 2.55290747e-01 3.80473077e-01 -1.80564240e-01
1.75443664e-01 3.15848202e-01 -1.07646894e+00 2.32570261e-01
1.27024877e+00 9.23596025e-01 4.23010975e-01 1.09419085e-01
-4.26305644e-02 7.60368764e-01 -4.70417649e-01 -2.15417668e-01
1.10328995e-01 -2.06563577e-01 -2.10547104e-01 -1.77999854e-01
-1.22185290e-01 -6.51230693e-01 -6.61750019e-01 1.28315055e+00
6.49981558e-01 1.24466911e-01 5.91469526e-01 -6.77447259e-01
-5.39896846e-01 7.68588245e-01 -7.40012646e-01 9.38297883e-02
-7.89724663e-02 2.51269698e-01 4.62427199e-01 -1.11877239e+00
3.25242072e-01 9.19271171e-01 1.10078216e+00 2.01152563e-01
-1.20387161e+00 -6.18951857e-01 -2.31242925e-01 1.57414690e-01
-1.44414222e+00 -3.67148787e-01 1.28345144e+00 -4.76649880e-01
3.06746542e-01 -1.35857631e-02 4.44632769e-01 1.18341398e+00
1.96266040e-01 5.76876283e-01 9.25666809e-01 -5.17180376e-02
3.28501225e-01 -2.21440092e-01 -1.41002640e-01 3.78827006e-01
1.70292825e-01 3.42260897e-01 -5.31343520e-01 -1.25052750e-01
9.51875687e-01 -1.32114831e-02 -4.37197328e-01 1.39222801e-01
-7.83426106e-01 5.79937637e-01 6.55627608e-01 4.90798622e-01
-6.08314097e-01 6.93491921e-02 2.99083829e-01 -6.32107332e-02
5.97583830e-01 1.56722575e-01 1.43667772e-01 1.51778176e-01
-1.46953022e+00 1.80305615e-01 4.39200521e-01 7.26414084e-01
8.04908514e-01 7.74863720e-01 -7.72227794e-02 7.09975004e-01
3.49704862e-01 7.13434339e-01 2.28051633e-01 -9.01980400e-01
5.60135305e-01 1.06794037e-01 8.98348466e-02 -1.23949862e+00
-2.40256414e-01 -8.37279320e-01 -1.55914104e+00 2.34886497e-01
5.88898778e-01 -2.11675733e-01 -8.34006310e-01 1.67883039e+00
1.27257541e-01 5.53591430e-01 -4.43613864e-02 7.32785583e-01
3.39590758e-01 8.39085758e-01 7.15189204e-02 -2.02929094e-01
8.89378130e-01 -2.63475865e-01 -8.67120504e-01 -3.26790303e-01
2.56629914e-01 -5.69077551e-01 6.67866826e-01 3.48201483e-01
-1.19337285e+00 -6.62694752e-01 -1.19383323e+00 9.11953375e-02
2.11435780e-01 -1.05905272e-01 1.91718772e-01 4.66711223e-01
-7.88061261e-01 8.78867626e-01 -1.11121964e+00 7.25410819e-01
4.37926978e-01 -5.03332540e-02 -1.11695103e-01 -3.73188823e-01
-1.56457317e+00 6.21560633e-01 1.59953505e-01 8.35017383e-01
-1.13670373e+00 -9.31886017e-01 -6.63648963e-01 8.91346186e-02
1.55312002e-01 -4.28385109e-01 5.33642948e-01 -6.01343334e-01
-1.18050849e+00 7.82387778e-02 7.50485361e-02 -4.71061975e-01
9.56404924e-01 -1.33297071e-01 -4.90056396e-01 8.38835686e-02
2.57208973e-01 2.81653106e-02 1.04276848e+00 -1.47291613e+00
-1.90190986e-01 -3.26787978e-01 -5.12919426e-01 -1.33699536e-01
-6.52983710e-02 -5.72386563e-01 -2.03527704e-01 -1.05991018e+00
5.54902673e-01 -4.70133007e-01 -4.05259192e-01 -2.39500981e-02
-3.86526197e-01 4.87775952e-01 5.55056572e-01 -1.19264555e+00
1.23343813e+00 -2.49809265e+00 9.95588526e-02 7.55539954e-01
6.81177601e-02 -7.18269274e-02 1.40317351e-01 3.97307962e-01
7.28345960e-02 -1.35622159e-01 -1.00995588e+00 -2.81618983e-01
-2.81974971e-01 2.70085514e-01 -4.51280445e-01 4.95945960e-01
1.60888359e-01 3.96473378e-01 -8.57329965e-01 -3.37835610e-01
-5.46498261e-02 6.64684534e-01 -4.14860338e-01 1.37141630e-01
-1.04665205e-01 8.02633166e-01 -5.85012317e-01 5.39629877e-01
1.16034961e+00 8.77604038e-02 -1.42818436e-01 -4.49983507e-01
-2.67684549e-01 -1.88528642e-01 -1.47602355e+00 1.76537609e+00
-3.24941427e-01 1.59198210e-01 3.19098920e-01 -9.18765485e-01
1.11005604e+00 2.84526318e-01 3.20447087e-01 -8.77138495e-01
-2.32086420e-01 6.90949976e-01 -1.18442632e-01 -6.98667407e-01
5.33420682e-01 -3.30334991e-01 1.52552471e-01 1.26993671e-01
-1.65458396e-01 -2.51454026e-01 -1.27795264e-01 2.12565795e-01
8.16831648e-01 1.73717409e-01 -4.41020548e-01 -3.64839107e-01
3.72253239e-01 -2.83961236e-01 7.47937858e-01 8.37838590e-01
2.74049371e-01 8.90364766e-01 3.02484453e-01 -1.34856790e-01
-1.24091113e+00 -1.12170923e+00 -1.99203596e-01 3.08630347e-01
1.68964148e-01 3.33302051e-01 -7.10482299e-01 -1.52875081e-01
-1.59373343e-01 8.08657706e-01 -7.38083482e-01 -2.49192491e-01
-7.81034946e-01 -1.08504200e+00 8.41698349e-01 6.16716802e-01
8.23640049e-01 -6.79579198e-01 -1.01602495e-01 6.83181345e-01
-4.41051066e-01 -6.25000775e-01 -2.12829784e-01 1.84327826e-01
-9.73394513e-01 -5.92454851e-01 -9.68210816e-01 -5.76236486e-01
6.52144969e-01 -2.96867061e-02 6.12145126e-01 1.05798997e-01
1.65272221e-01 -2.34488800e-01 -8.44777375e-02 3.97593305e-02
-4.68024969e-01 -2.48469055e-01 -2.76738465e-01 4.04221982e-01
-2.89480180e-01 -9.01146412e-01 -7.70315349e-01 2.73283839e-01
-1.18866158e+00 -1.00756362e-01 5.13413250e-01 1.35393119e+00
5.15450358e-01 3.27709109e-01 7.31169999e-01 -6.45203590e-01
4.13263708e-01 -8.95905852e-01 -5.08751214e-01 -8.42240974e-02
-2.72012562e-01 2.46700615e-01 5.14552474e-01 -4.40614551e-01
-1.62255681e+00 -2.68772423e-01 -6.74180806e-01 -2.86374897e-01
2.33537927e-01 8.68552089e-01 -2.83465326e-01 -7.65901282e-02
6.93136752e-01 5.33544064e-01 -1.58452839e-02 -4.72163856e-01
1.97709668e-02 3.13454986e-01 7.13740706e-01 -6.68530762e-01
8.90502572e-01 7.67312169e-01 5.80297671e-02 -1.03229117e+00
-5.64889014e-01 -6.13009185e-02 -2.74403661e-01 -4.18703079e-01
6.74593210e-01 -9.29025769e-01 -2.04299420e-01 9.33862984e-01
-1.12006807e+00 -2.50686467e-01 -3.63114476e-01 5.87639987e-01
-2.34353125e-01 6.94304109e-01 -8.42888176e-01 -9.27431405e-01
-1.10790804e-01 -1.28543448e+00 6.44369423e-01 1.41916379e-01
1.51058272e-01 -9.17229712e-01 -1.59322277e-01 5.53682894e-02
6.00775063e-01 2.67519802e-01 9.06730473e-01 -3.47736150e-01
-8.49958777e-01 -1.82606518e-01 -8.41111690e-03 4.66881335e-01
-1.05543450e-01 -1.39660120e-01 -9.90840673e-01 -7.82826319e-02
6.03349626e-01 -2.94823460e-02 1.06054962e+00 6.58107698e-01
9.95019078e-01 -5.14507331e-02 -1.32483348e-01 7.31841266e-01
1.45909834e+00 2.46500611e-01 1.18545866e+00 3.37462187e-01
5.62249362e-01 5.55714846e-01 3.12426746e-01 6.56105280e-01
-2.45378111e-02 8.45399871e-02 5.03744245e-01 -1.48342431e-01
-2.00395167e-01 -2.98808038e-01 -8.02592188e-02 9.72324848e-01
-3.72875988e-01 -1.71886057e-01 -8.96928906e-01 7.29027867e-01
-1.47003508e+00 -1.26794803e+00 -2.38873422e-01 2.21512485e+00
9.65692401e-01 3.92309427e-01 -5.74847817e-01 4.56985205e-01
7.56837547e-01 3.43413830e-01 -4.36685950e-01 5.18511117e-01
-4.96690422e-01 1.51927769e-01 4.44518149e-01 6.46713316e-01
-6.94511294e-01 2.58966833e-01 5.49648857e+00 1.25659287e+00
-9.18530881e-01 1.63642034e-01 7.34325767e-01 4.71699178e-01
-7.67101467e-01 -1.33644328e-01 -5.09908319e-01 8.97601664e-01
4.80415702e-01 2.19057515e-01 1.86947986e-01 3.68167311e-01
4.22895819e-01 -2.60441244e-01 -5.75576842e-01 5.07666171e-01
-2.74701685e-01 -1.44122052e+00 -6.86349869e-02 -1.88474998e-01
8.34007025e-01 -1.75007954e-01 2.38705292e-01 -6.18121848e-02
2.45827958e-01 -7.38894105e-01 9.03486967e-01 1.16158223e+00
7.20554113e-01 -8.69122863e-01 8.37032616e-01 5.47336578e-01
-9.26719844e-01 -1.25332192e-01 -2.21353680e-01 1.64359093e-01
5.50034344e-01 1.15558243e+00 -2.50409693e-01 9.47826445e-01
6.32373154e-01 5.03201365e-01 3.79988626e-02 1.04509747e+00
-4.05211657e-01 8.36969197e-01 -4.49549764e-01 4.67107356e-01
2.28235066e-01 -6.08169496e-01 7.54119277e-01 1.05927980e+00
6.95341051e-01 1.48651823e-01 -1.89582795e-01 1.12094724e+00
-4.09753658e-02 -3.93849313e-01 -4.73181129e-01 5.49134433e-01
5.14108479e-01 6.15735710e-01 -3.66288930e-01 -1.41826451e-01
-3.75106692e-01 4.70292598e-01 -6.95107356e-02 7.10183322e-01
-8.10864747e-01 -3.04601640e-01 3.74565758e-02 2.98213214e-01
1.58705771e-01 -3.66099238e-01 -7.47711301e-01 -8.14535439e-01
1.41405076e-01 -5.91081619e-01 1.15640916e-01 -5.92070222e-01
-1.42173982e+00 4.95218426e-01 -1.43155798e-01 -1.32678294e+00
7.16223707e-03 2.00743988e-01 -7.42246270e-01 1.22102344e+00
-1.48873949e+00 -1.06023669e+00 -2.79591978e-01 4.37977850e-01
4.06283081e-01 2.87818387e-02 4.35370475e-01 8.33914936e-01
-3.64921659e-01 3.36676836e-01 5.70873499e-01 1.30102232e-01
4.74004775e-01 -7.01462924e-01 2.07916141e-01 1.14703643e+00
-4.04641777e-01 3.65248680e-01 9.31725979e-01 -1.14871621e+00
-9.40738440e-01 -9.79883969e-01 3.83191824e-01 3.16366583e-01
8.11782956e-01 2.56186426e-02 -1.69321859e+00 3.38756561e-01
-4.60876077e-02 -5.72625138e-02 1.76076889e-01 -4.46119636e-01
-1.19878426e-01 -3.46983910e-01 -1.09972703e+00 5.22956669e-01
5.60623646e-01 -5.14193952e-01 -5.39227664e-01 -6.66948780e-02
4.96262610e-01 -3.85970831e-01 -8.83264422e-01 5.47060549e-01
2.66186655e-01 -8.82267356e-01 1.13685763e+00 1.87924460e-01
7.35162377e-01 -4.39423740e-01 -1.38774291e-01 -1.29453790e+00
-3.23120922e-01 -4.68017578e-01 6.54645488e-02 1.36257362e+00
4.03397948e-01 -5.23515701e-01 6.97361469e-01 6.31126046e-01
-5.55469453e-01 -3.53329092e-01 -1.20115423e+00 -6.95885122e-01
1.29323289e-01 -3.89232963e-01 5.21309853e-01 9.25201893e-01
-4.44979221e-01 -3.25865179e-01 -6.73141479e-01 5.78491569e-01
1.19964707e+00 -3.40923071e-01 1.35488108e-01 -8.51106346e-01
-4.93906170e-01 -2.26392686e-01 9.10415500e-02 -1.12888551e+00
-2.71410406e-01 -4.84799206e-01 3.22758347e-01 -1.27146804e+00
-2.20152229e-01 -8.10724556e-01 -1.70802578e-01 6.80112792e-03
-2.92036891e-01 1.13163836e-01 -2.48373538e-01 3.75821888e-01
3.89293164e-01 9.22358453e-01 1.40901053e+00 -4.47651535e-01
-1.30569078e-02 2.38020331e-01 -1.02755949e-01 7.96060920e-01
6.45452142e-01 -7.51689136e-01 -2.90403157e-01 -8.28394055e-01
3.10699731e-01 6.67337537e-01 4.54497308e-01 -1.37599218e+00
6.30881250e-01 1.05378047e-01 4.56002146e-01 -7.08566785e-01
4.41487044e-01 -9.25398707e-01 6.35043085e-01 5.61281681e-01
-1.41387984e-01 -3.27328563e-01 1.15175657e-01 8.63631725e-01
-5.60738206e-01 -5.39751768e-01 9.64681149e-01 -8.46977904e-02
-5.06748915e-01 2.97852129e-01 -2.44093105e-01 7.66643649e-03
5.65153301e-01 -2.31775418e-01 4.58611622e-02 -2.16146931e-01
-8.34576309e-01 6.51340559e-02 8.87865648e-02 -3.04527074e-01
7.51889765e-01 -1.06960940e+00 -9.17598248e-01 4.18278486e-01
-3.99850547e-01 3.38551372e-01 1.05512381e+00 8.04476261e-01
-5.95869660e-01 -5.04113138e-01 -4.41932119e-02 -5.20945966e-01
-2.98933744e-01 3.03791851e-01 3.95460099e-01 -3.83712143e-01
-6.74663901e-01 8.31408739e-01 5.79500534e-02 1.05296202e-01
1.21644415e-01 -1.46691784e-01 -3.94168347e-02 5.57997413e-02
4.96416658e-01 7.01680064e-01 1.12972401e-01 -3.07284385e-01
1.19793035e-01 3.75018150e-01 8.79346058e-02 -3.44167829e-01
1.45799124e+00 -2.32424244e-01 -1.11662624e-02 3.05152178e-01
7.69629657e-01 2.21119106e-01 -1.70542133e+00 -4.19591814e-01
-7.56501481e-02 -3.75064760e-01 1.71829954e-01 -4.55700725e-01
-1.26002300e+00 9.81118739e-01 4.16762143e-01 8.82459655e-02
1.02119648e+00 -5.46657860e-01 8.21428120e-01 -6.09441660e-02
3.01495343e-01 -1.09634292e+00 -4.63884277e-03 4.51082587e-01
8.63640606e-01 -8.76893640e-01 3.83229069e-02 -3.20227742e-01
-4.54651415e-01 1.03270316e+00 2.06304774e-01 -2.37965882e-01
7.41173685e-01 6.82576299e-01 -1.69669569e-01 -1.04407389e-02
-8.51765350e-02 3.79901439e-01 -2.60708123e-01 4.50932324e-01
-4.41153683e-02 -1.90061674e-01 -4.66000102e-02 6.85456872e-01
3.82030487e-01 -4.17202041e-02 4.82296318e-01 9.31920707e-01
-4.33230698e-01 -7.41503835e-01 -6.69864893e-01 1.60351783e-01
-4.30478483e-01 -3.19314182e-01 6.45914018e-01 5.62339783e-01
1.68092623e-02 6.66372061e-01 -4.29431610e-02 -1.05226964e-01
3.55454177e-01 -2.37725258e-01 -1.12517243e-02 2.39872327e-03
-3.99025410e-01 4.79501992e-01 -2.82986611e-02 -2.53699452e-01
-3.02250892e-01 -6.65923178e-01 -1.24567425e+00 -3.37625176e-01
-3.46637726e-01 3.27384919e-01 5.29443145e-01 9.34590220e-01
2.76180748e-02 7.26091683e-01 6.25119388e-01 -9.62712705e-01
-8.95223916e-01 -1.01221538e+00 -9.42217588e-01 2.62385249e-01
4.27599072e-01 -6.56775534e-01 -6.22700214e-01 1.83441773e-01] | [6.833965301513672, 2.646696090698242] |
d856d620-505f-442e-b0ef-f94cbd684852 | misrobaerta-transformers-versus | 2304.07759 | null | https://arxiv.org/abs/2304.07759v1 | https://arxiv.org/pdf/2304.07759v1.pdf | MisRoBÆRTa: Transformers versus Misinformation | Misinformation is considered a threat to our democratic values and principles. The spread of such content on social media polarizes society and undermines public discourse by distorting public perceptions and generating social unrest while lacking the rigor of traditional journalism. Transformers and transfer learning proved to be state-of-the-art methods for multiple well-known natural language processing tasks. In this paper, we propose MisRoB{\AE}RTa, a novel transformer-based deep neural ensemble architecture for misinformation detection. MisRoB{\AE}RTa takes advantage of two transformers (BART \& RoBERTa) to improve the classification performance. We also benchmarked and evaluated the performances of multiple transformers on the task of misinformation detection. For training and testing, we used a large real-world news articles dataset labeled with 10 classes, addressing two shortcomings in the current research: increasing the size of the dataset from small to large, and moving the focus of fake news detection from binary classification to multi-class classification. For this dataset, we manually verified the content of the news articles to ensure that they were correctly labeled. The experimental results show that the accuracy of transformers on the misinformation detection problem was significantly influenced by the method employed to learn the context, dataset size, and vocabulary dimension. We observe empirically that the best accuracy performance among the classification models that use only one transformer is obtained by BART, while DistilRoBERTa obtains the best accuracy in the least amount of time required for fine-tuning and training. The proposed MisRoB{\AE}RTa outperforms the other transformer models in the task of misinformation detection. To arrive at this conclusion, we performed ample ablation and sensitivity testing with MisRoB{\AE}RTa on two datasets. | ['Elena-Simona Apostol', 'Ciprian-Octavian Truică'] | 2023-04-16 | null | null | null | null | ['misinformation', 'fake-news-detection'] | ['miscellaneous', 'natural-language-processing'] | [ 1.23421058e-01 -3.98193076e-02 -1.78467900e-01 -1.32020041e-01
-6.73810720e-01 -6.60899043e-01 1.06543612e+00 2.10256800e-01
-4.07141060e-01 6.89882398e-01 3.38631600e-01 -7.75018096e-01
6.20100237e-02 -9.30667281e-01 -7.93918371e-01 -5.50629377e-01
1.83709174e-01 4.03730512e-01 -8.53006393e-02 -5.66613078e-01
5.46896577e-01 -1.80441868e-02 -1.32433760e+00 5.97896636e-01
1.18587887e+00 1.01109219e+00 -4.27535832e-01 2.46396795e-01
1.61096659e-02 1.13090587e+00 -1.01463068e+00 -7.67173827e-01
1.33036092e-01 -4.40897673e-01 -8.26040089e-01 -1.75798327e-01
6.04142010e-01 -3.90456349e-01 -2.49216557e-01 1.22490942e+00
3.28670442e-01 -2.26642713e-01 8.17744493e-01 -1.05064166e+00
-9.11788523e-01 9.73730505e-01 -3.42050105e-01 3.98951650e-01
1.47484541e-01 3.49650644e-02 8.34901750e-01 -9.20977354e-01
5.42214990e-01 1.27609098e+00 7.38163531e-01 2.02458769e-01
-9.40315962e-01 -1.06463206e+00 6.50265142e-02 2.98155516e-01
-1.21052551e+00 -4.15651858e-01 7.78265953e-01 -7.83783555e-01
6.99851930e-01 3.84778380e-01 6.61056578e-01 1.57009888e+00
4.73797500e-01 5.02440751e-01 1.71559572e+00 -3.58695269e-01
1.29920870e-01 4.15804565e-01 2.13884279e-01 5.99769056e-01
4.65716958e-01 1.81873679e-01 -3.21707249e-01 -3.16957176e-01
2.07876667e-01 -2.81746775e-01 -1.86485723e-01 4.46702838e-01
-1.28328192e+00 1.07907486e+00 5.73436320e-01 7.52613962e-01
-1.10177174e-01 -9.48219597e-02 5.38429081e-01 7.66705215e-01
1.00780606e+00 9.29802656e-01 -2.78295130e-01 -3.54592912e-02
-7.25827754e-01 1.80105582e-01 7.65909553e-01 3.51878077e-01
3.88833255e-01 4.39853892e-02 -2.77328789e-01 7.79637873e-01
-1.81037560e-02 7.84152687e-01 6.60053432e-01 -5.68039775e-01
7.42464006e-01 8.09550524e-01 3.09556425e-01 -1.79269731e+00
-3.40350479e-01 -8.26340914e-01 -9.25857306e-01 -3.95763367e-02
7.21492290e-01 -1.67707860e-01 -7.97494888e-01 1.50007915e+00
6.86252117e-02 -2.24826276e-01 8.24390650e-02 9.72730100e-01
8.41246903e-01 8.21104586e-01 -1.05241723e-01 7.54150003e-02
1.34325564e+00 -7.07185209e-01 -7.42403090e-01 -3.88325870e-01
8.00503552e-01 -7.63918757e-01 1.16631079e+00 5.69745541e-01
-3.69564384e-01 -1.52441591e-01 -1.20901835e+00 -8.55804086e-02
-7.05051899e-01 2.57834196e-01 4.30994987e-01 7.28901088e-01
-5.96603930e-01 5.28666794e-01 -3.21821660e-01 -9.59278718e-02
5.11843264e-01 -1.00094281e-01 -1.91601947e-01 8.32067281e-02
-1.61332881e+00 1.18121600e+00 3.42207432e-01 1.43469140e-01
-7.96449184e-01 -1.77925929e-01 -5.92346191e-01 5.47445603e-02
3.44073921e-01 -4.61198866e-01 1.06479084e+00 -1.38051808e+00
-1.37986577e+00 1.14610505e+00 4.19121176e-01 -5.93722463e-01
8.76103580e-01 -2.99063146e-01 -6.91346049e-01 -2.07936555e-01
1.84498742e-01 -2.42414307e-02 1.06790090e+00 -1.22866464e+00
-2.55414754e-01 -4.05440778e-01 -2.09548399e-02 -1.10463470e-01
-5.69811404e-01 3.53091322e-02 5.02608716e-01 -9.15626884e-01
1.23045564e-01 -8.75572681e-01 3.36338013e-01 -4.53062445e-01
-5.78437448e-01 -3.67107540e-02 7.83505917e-01 -7.74641931e-01
1.30635893e+00 -2.05493712e+00 -1.46034375e-01 2.30343059e-01
4.90296006e-01 6.30168021e-01 1.00020908e-01 4.53126699e-01
3.89657728e-02 3.84214848e-01 -4.70169000e-02 -3.12652513e-02
-4.77036275e-02 -1.17557012e-02 -5.85166633e-01 7.21233368e-01
-2.03306615e-01 7.90841043e-01 -9.09557819e-01 -1.84280455e-01
-1.02247037e-01 2.36700147e-01 -4.39197004e-01 -1.59730971e-01
-3.02363217e-01 4.95161861e-01 -4.65652347e-01 5.25389194e-01
5.46247602e-01 -4.19734776e-01 1.28964946e-01 -1.65038079e-01
-1.08655319e-01 7.07325876e-01 -5.43565452e-01 9.23805296e-01
-4.14800853e-01 9.41889942e-01 -1.93078116e-01 -1.10520327e+00
9.56368387e-01 2.22479150e-01 2.25915283e-01 -1.00740850e+00
6.48784041e-01 4.48527575e-01 1.24666825e-01 -5.74144661e-01
5.82545519e-01 -1.71821877e-01 -3.99296373e-01 5.97002327e-01
-4.27534491e-01 9.24938470e-02 -1.58673555e-01 8.57235342e-02
8.05639863e-01 -4.43184793e-01 3.10297906e-01 -3.66502613e-01
4.07316148e-01 1.49961382e-01 4.14527684e-01 9.03159857e-01
-1.15249515e-01 2.53002811e-02 5.73580623e-01 -6.66298509e-01
-9.87728715e-01 -6.65893376e-01 -1.81962743e-01 1.29319596e+00
1.03497393e-01 -2.87224710e-01 -6.76001847e-01 -8.51351619e-01
4.84244600e-02 1.15547788e+00 -6.99729204e-01 -2.26293981e-01
-3.77759874e-01 -1.15101421e+00 8.33554566e-01 -1.50948077e-01
8.71990383e-01 -8.70470762e-01 -4.88026112e-01 1.07334711e-01
-7.05980837e-01 -1.11693394e+00 -2.38052309e-01 -2.76881874e-01
-4.84112710e-01 -1.32253385e+00 -3.05957168e-01 -4.86009479e-01
5.37348330e-01 1.48248687e-01 9.69585836e-01 8.93125385e-02
2.95060068e-01 -1.70884252e-01 -4.67128307e-01 -5.44523120e-01
-1.05942893e+00 3.98158506e-02 -4.80409451e-02 1.73097700e-01
2.32183591e-01 -4.30920213e-01 -3.79801691e-01 3.26745182e-01
-9.34964120e-01 1.21247187e-01 5.58607996e-01 1.05064392e+00
-2.39397734e-01 8.83879513e-02 6.83731496e-01 -1.11285388e+00
9.30909932e-01 -5.21013379e-01 -5.40730596e-01 -1.01420805e-01
-8.35003734e-01 -7.26929978e-02 6.76765978e-01 -4.46213365e-01
-6.67332232e-01 -7.79769659e-01 5.75179514e-03 -4.16587330e-02
1.67908162e-01 8.56655180e-01 3.23645800e-01 9.15163755e-02
1.15854347e+00 2.47428194e-01 -2.41299421e-02 -4.25745100e-01
1.63267255e-01 1.05259240e+00 1.66507497e-01 -2.24024877e-01
5.15810668e-01 3.29945058e-01 -5.16122282e-01 -6.50461793e-01
-1.28325593e+00 -6.34191632e-02 -1.36766449e-01 -1.43424526e-01
4.83643591e-01 -8.66517484e-01 -5.99406719e-01 8.14173877e-01
-1.11776388e+00 -6.77450523e-02 1.80628777e-01 3.05783391e-01
-5.36504500e-02 2.67055601e-01 -6.23549759e-01 -6.48416102e-01
-4.99009639e-01 -9.40377831e-01 7.73977101e-01 -5.55852473e-01
-1.10967807e-01 -8.72972727e-01 -8.54856297e-02 7.88963199e-01
6.41358137e-01 4.25022393e-01 9.96075690e-01 -1.06375778e+00
-1.88147545e-01 -2.32144773e-01 -2.81221241e-01 4.77550447e-01
6.41759485e-02 -2.18704507e-01 -1.06522596e+00 -3.50157589e-01
3.07422370e-01 -3.48438203e-01 9.75652754e-01 -3.61342356e-02
1.04952073e+00 -1.22222710e+00 -2.23116055e-01 2.03198850e-01
1.08891547e+00 -3.47276367e-02 5.51723301e-01 7.66562164e-01
5.36816478e-01 3.81659240e-01 3.79004717e-01 2.54571319e-01
4.04640079e-01 5.30221939e-01 4.73737568e-01 -2.84656659e-02
1.69488132e-01 -3.15982461e-01 6.26517236e-01 7.84757972e-01
8.50926265e-02 -4.91848707e-01 -1.13382900e+00 5.03988922e-01
-1.76455057e+00 -1.16766286e+00 -7.78694451e-02 2.14995527e+00
9.43032622e-01 3.22293550e-01 5.97019419e-02 3.44199777e-01
6.70626581e-01 2.50093311e-01 -1.28011867e-01 -5.09603202e-01
-1.77880660e-01 -3.35299909e-01 5.24618387e-01 5.25687754e-01
-1.32144213e+00 9.71956372e-01 5.84176540e+00 6.92235589e-01
-1.61906302e+00 3.66939723e-01 7.65649199e-01 1.66723639e-01
-4.38592911e-01 -2.29381412e-01 -4.38043773e-01 8.24541986e-01
8.31735253e-01 -6.68305904e-02 4.75070626e-01 7.10089922e-01
2.97158092e-01 -8.18904396e-03 -7.87454784e-01 7.35628366e-01
2.76857942e-01 -1.35994184e+00 1.52430773e-01 1.87352467e-02
9.28163826e-01 3.16554308e-01 2.37880558e-01 4.72822279e-01
2.81599939e-01 -1.20430303e+00 8.87117684e-01 1.32039398e-01
6.46047652e-01 -3.35065007e-01 9.30437386e-01 7.02439010e-01
-1.56329229e-01 -3.75794321e-01 4.43963222e-02 -3.74347210e-01
-3.72994542e-01 8.63414943e-01 -1.12232733e+00 2.37503901e-01
5.33565819e-01 6.68287218e-01 -7.01522768e-01 5.45869470e-01
-3.11491251e-01 9.80115354e-01 -2.53214180e-01 -2.52954572e-01
5.07571399e-01 9.44994092e-02 8.05046976e-01 1.31226170e+00
1.98790401e-01 -2.42474213e-01 2.51460165e-01 9.25942540e-01
-3.11664551e-01 9.42537189e-02 -7.56864309e-01 -2.35450402e-01
4.88318294e-01 7.66936719e-01 -3.61932129e-01 -5.06004393e-01
2.27399226e-02 5.66245198e-01 2.81998098e-01 2.03426465e-01
-9.42474425e-01 -8.33569393e-02 2.07492128e-01 4.62728322e-01
1.22316331e-01 -9.25931260e-02 -4.31227267e-01 -1.44480348e+00
1.01634912e-04 -1.23224747e+00 4.13526058e-01 -3.99963558e-01
-1.46448743e+00 5.89346290e-01 -4.43637557e-03 -1.19419134e+00
-8.17938671e-02 -6.34330690e-01 -2.82308221e-01 5.57600081e-01
-1.44011891e+00 -1.22594118e+00 -2.82274306e-01 3.84674251e-01
2.48287439e-01 -2.46252179e-01 7.56878495e-01 2.86850333e-01
-5.11061728e-01 6.73685312e-01 4.13871050e-01 4.53729898e-01
7.83472896e-01 -8.36575568e-01 2.27125332e-01 7.23430932e-01
-1.35148525e-01 4.84740525e-01 8.57522905e-01 -5.91421008e-01
-1.00981367e+00 -1.04572117e+00 1.11339784e+00 -4.09095138e-01
9.07206297e-01 -4.30778712e-01 -7.24137902e-01 8.33155990e-01
1.67726323e-01 -3.68970633e-01 5.40716648e-01 1.74525231e-01
-8.18106353e-01 -7.46487230e-02 -1.35074139e+00 3.56128067e-01
6.32678688e-01 -5.53548396e-01 -7.77877212e-01 6.11040890e-01
3.66097957e-01 -3.33066791e-01 -6.45325661e-01 1.63805038e-01
5.73892593e-01 -1.01654541e+00 5.46251595e-01 -6.04132116e-01
8.12962532e-01 4.18407004e-03 -4.97813001e-02 -1.54278326e+00
-2.97240496e-01 -4.12725151e-01 8.44633430e-02 9.30233777e-01
5.90523303e-01 -1.12649107e+00 3.14239532e-01 3.73311751e-02
6.22062571e-02 -5.08679092e-01 -1.00549328e+00 -6.30875647e-01
3.85510862e-01 -3.18710923e-01 4.75682348e-01 1.60373926e+00
2.64623053e-02 6.49184048e-01 -4.95642453e-01 4.62533906e-03
3.97084504e-01 3.72434646e-01 7.51848757e-01 -1.12563419e+00
-5.56688905e-02 -5.17266750e-01 -2.34624594e-01 -7.35174775e-01
-1.24321639e-04 -1.04438901e+00 -2.02736855e-01 -1.22878540e+00
1.33258805e-01 -5.15654147e-01 -1.48296788e-01 5.71768045e-01
4.41540964e-03 3.29535812e-01 5.06709665e-02 4.32171434e-01
-2.80514330e-01 5.90871215e-01 1.34746587e+00 -5.91352165e-01
-2.83657908e-02 4.45263088e-02 -9.71576810e-01 8.01070929e-01
7.81488061e-01 -6.12803340e-01 -1.57728776e-01 -5.92557669e-01
6.64452136e-01 -1.37789547e-01 6.67504489e-01 -7.73351073e-01
-1.02452412e-01 -6.39166087e-02 1.66757524e-01 -2.58238494e-01
2.34683886e-01 -6.25234842e-01 -1.96964502e-01 5.89103580e-01
-4.64783281e-01 6.04459345e-02 1.03003621e-01 4.29331690e-01
-9.34834257e-02 1.83786228e-01 8.66198659e-01 -1.64027333e-01
-3.61583084e-01 -2.70304590e-01 -6.40689015e-01 2.50888288e-01
7.27214277e-01 8.72694552e-02 -9.59894419e-01 -5.12101352e-01
-5.42087734e-01 -1.14143349e-01 2.60844737e-01 6.05187476e-01
4.17252988e-01 -1.04725933e+00 -1.17275715e+00 2.46140227e-01
2.22934425e-01 -4.69427407e-01 -2.39346385e-01 1.04091573e+00
-5.24983287e-01 4.99954522e-01 -1.49940252e-01 -4.88426030e-01
-1.02874327e+00 4.86035317e-01 4.85833794e-01 -4.09329027e-01
-4.39279228e-01 4.93278205e-01 -1.91611007e-01 -5.84576249e-01
-5.08457385e-02 -4.38830078e-01 -3.26188654e-01 1.50154799e-01
4.16918337e-01 4.89788920e-01 2.15686306e-01 -7.84380198e-01
-1.92559436e-01 9.59011912e-02 -3.55484217e-01 1.27896801e-01
1.20310521e+00 -7.04523455e-03 -2.52385348e-01 3.89651000e-01
1.07060623e+00 2.17834875e-01 -3.32774848e-01 -2.47883633e-01
-1.83303341e-01 -6.24599934e-01 4.05838639e-01 -1.42056155e+00
-7.30254292e-01 5.17734289e-01 4.34706062e-01 4.49961483e-01
8.25759828e-01 -2.94835836e-01 7.56371617e-01 4.16100830e-01
3.34650725e-01 -9.40649807e-01 9.59312096e-02 7.05262065e-01
1.16700089e+00 -1.36587322e+00 6.69087917e-02 -3.21566075e-01
-5.99269271e-01 7.83421338e-01 4.36090052e-01 -8.70124698e-02
5.47594011e-01 -1.67249545e-01 2.75702268e-01 -4.87013906e-01
-5.09793997e-01 3.90394956e-01 4.06799883e-01 3.81595525e-03
3.57126713e-01 2.36905441e-01 -7.48826206e-01 4.08407211e-01
-6.34698629e-01 -4.91287149e-02 5.85002303e-01 6.12893820e-01
-5.26553214e-01 -6.06611431e-01 -5.59767127e-01 5.00150383e-01
-6.86518312e-01 -1.80080503e-01 -8.58455122e-01 7.76555598e-01
1.92621320e-01 1.17259002e+00 -1.05186045e-01 -6.54534280e-01
1.00241952e-01 -1.36304870e-01 1.62953749e-01 -2.85273314e-01
-9.92726743e-01 -3.27552468e-01 5.74548900e-01 -3.13109368e-01
-3.55458736e-01 -3.75231177e-01 -6.83090210e-01 -6.95901215e-01
-4.74778682e-01 2.88359851e-01 6.15612328e-01 1.25593126e+00
4.46788251e-01 2.90699780e-01 7.70275295e-01 -3.28465283e-01
-1.11828446e+00 -1.37385428e+00 -2.09524617e-01 6.21847749e-01
5.90878308e-01 -6.63581133e-01 -5.68149388e-01 -3.34529102e-01] | [8.218731880187988, 10.262718200683594] |
760c7068-10ec-42d8-a250-98f8ac95933a | reqa-coarse-to-fine-assessment-of-image | 2209.01760 | null | https://arxiv.org/abs/2209.01760v4 | https://arxiv.org/pdf/2209.01760v4.pdf | REQA: Coarse-to-fine Assessment of Image Quality to Alleviate the Range Effect | Blind image quality assessment (BIQA) of user generated content (UGC) suffers from the range effect which indicates that on the overall quality range, mean opinion score (MOS) and predicted MOS (pMOS) are well correlated; focusing on a particular range, the correlation is lower. The reason for the range effect is that the predicted deviations both in a wide range and in a narrow range destroy the uniformity between MOS and pMOS. To tackle this problem, a novel method is proposed from coarse-grained metric to fine-grained prediction. Firstly, we design a rank-and-gradient loss for coarse-grained metric. The loss keeps the order and grad consistency between pMOS and MOS, thereby reducing the predicted deviation in a wide range. Secondly, we propose multi-level tolerance loss to make fine-grained prediction. The loss is constrained by a decreasing threshold to limite the predicted deviation in narrower and narrower ranges. Finally, we design a feedback network to conduct the coarse-to-fine assessment. On the one hand, the network adopts feedback blocks to process multi-scale distortion features iteratively and on the other hand, it fuses non-local context feature to the output of each iteration to acquire more quality-aware feature representation. Experimental results demonstrate that the proposed method can alleviate the range effect compared to the state-of-the-art methods effectively. | ['Fushuo Huo', 'Bingheng Li'] | 2022-09-05 | null | null | null | null | ['blind-image-quality-assessment'] | ['computer-vision'] | [ 1.07121408e-01 -7.03578413e-01 -7.26671293e-02 -4.93371725e-01
-5.81189036e-01 -2.87188768e-01 1.63217843e-01 4.50492688e-02
-2.30817124e-01 5.53046346e-01 3.86011690e-01 1.11908689e-01
-4.60762650e-01 -8.43880415e-01 -2.28057206e-01 -8.24379981e-01
2.96102405e-01 -4.80336785e-01 6.47560298e-01 -2.24042460e-01
4.73157197e-01 2.13718072e-01 -1.64086914e+00 3.90568912e-01
1.32467222e+00 1.33466554e+00 2.93533444e-01 1.75249442e-01
-8.46117809e-02 5.18671632e-01 -8.43224406e-01 -2.24498451e-01
2.86517620e-01 -6.16132915e-01 -1.75839901e-01 4.29589339e-02
2.11482644e-01 -3.60193282e-01 -1.94440693e-01 1.44163752e+00
7.31338024e-01 2.94145942e-02 6.13237798e-01 -1.12517881e+00
-6.82460725e-01 1.42758086e-01 -7.72752881e-01 4.33958411e-01
3.75738710e-01 2.12391183e-01 8.41944158e-01 -8.53354573e-01
1.38683304e-01 1.24439347e+00 4.68140572e-01 1.39413223e-01
-7.91403592e-01 -7.38690615e-01 3.76262993e-01 4.79847431e-01
-1.58030677e+00 -3.33795428e-01 8.77945840e-01 -1.72365770e-01
3.81294578e-01 2.95582235e-01 4.83718604e-01 4.30300921e-01
3.84505242e-01 3.87719065e-01 1.27443707e+00 -2.68799365e-01
2.33950555e-01 6.88062608e-02 -2.25792855e-01 5.49866796e-01
1.71705723e-01 -3.10484110e-03 -4.56142128e-01 -4.66948282e-03
7.67857850e-01 -5.03531247e-02 -6.33649111e-01 -2.03315049e-01
-9.83179152e-01 5.01325727e-01 4.47187215e-01 3.38152856e-01
-2.80694783e-01 -1.78821027e-01 3.71151656e-01 5.14310300e-01
6.85609430e-02 -3.36727798e-02 -3.42183769e-01 -2.45186612e-01
-1.03011024e+00 -6.78019971e-02 5.35809755e-01 5.68089843e-01
9.23581719e-01 -1.76042303e-01 -5.44432819e-01 1.05012643e+00
3.51435572e-01 3.90394479e-01 7.78212607e-01 -8.80975366e-01
6.14071429e-01 6.28364623e-01 1.85561642e-01 -1.42481887e+00
-1.74573436e-01 -5.92756629e-01 -9.59615469e-01 4.40319985e-01
3.56447101e-01 1.83235303e-01 -8.09630156e-01 1.62691355e+00
2.56964743e-01 -1.18731298e-01 -3.80322725e-01 1.34719777e+00
6.50804281e-01 8.69457483e-01 1.75374257e-03 -6.83323741e-01
1.36679113e+00 -8.21837187e-01 -8.80361915e-01 2.74469052e-02
-4.27747108e-02 -9.17721033e-01 1.33187938e+00 5.86677074e-01
-1.09410536e+00 -9.16681588e-01 -1.43079329e+00 1.05774209e-01
-8.10972229e-02 6.14538714e-02 -7.27995485e-02 7.69235432e-01
-1.02642405e+00 7.30185390e-01 -3.15514058e-01 -1.64383009e-01
2.11509019e-01 1.10607646e-01 1.54339159e-02 -8.80397931e-02
-1.41309154e+00 7.02521265e-01 6.62363619e-02 1.59072384e-01
-4.92528021e-01 -4.90285486e-01 -3.96964073e-01 1.79885626e-01
2.20719665e-01 -4.31795210e-01 7.92496204e-01 -1.07138205e+00
-1.59561789e+00 3.52535367e-01 -1.64247826e-01 -7.83488899e-02
5.61630189e-01 3.40171494e-02 -9.55966949e-01 1.95671678e-01
1.74933039e-02 2.97608256e-01 1.04705310e+00 -1.31351447e+00
-8.23947966e-01 -2.80791432e-01 1.02741746e-02 5.37627876e-01
-7.92301536e-01 -7.19994828e-02 -6.79332614e-01 -7.66622543e-01
4.55295742e-01 -3.01729232e-01 5.34698293e-02 1.71603546e-01
-1.05214849e-01 -1.61730275e-01 7.81371772e-01 -6.18319571e-01
1.99480438e+00 -2.25017214e+00 -2.32521623e-01 4.06572074e-01
1.81409955e-01 3.40956897e-01 -1.84916615e-01 2.60540903e-01
4.46900241e-02 -1.48685398e-02 -1.80779591e-01 2.83590019e-01
-1.65379629e-01 5.27276024e-02 8.59631673e-02 4.34168905e-01
5.05419411e-02 3.28807205e-01 -8.83924723e-01 -6.36001825e-01
2.34937653e-01 3.17076653e-01 -4.30034429e-01 1.50756434e-01
2.22414479e-01 3.99520546e-01 -6.04424655e-01 6.41408682e-01
1.11901379e+00 -1.37691334e-01 -1.90323591e-01 -6.38957739e-01
-3.39233518e-01 7.33681992e-02 -1.63163733e+00 1.41921842e+00
-4.33256596e-01 1.84075192e-01 -1.12439189e-02 -6.81229472e-01
1.25833774e+00 2.81793296e-01 2.32948065e-01 -1.08953691e+00
5.50787076e-02 3.27541232e-01 -3.44036333e-02 -4.88148957e-01
3.54338527e-01 -1.61810383e-01 1.52091280e-01 1.75427496e-01
-3.95688534e-01 2.06157327e-01 6.62957132e-02 -1.16982892e-01
8.61445725e-01 -6.89242184e-02 4.31730658e-01 -9.70666707e-02
1.07723868e+00 -5.54987907e-01 9.52949524e-01 4.97064352e-01
-6.37536108e-01 7.49876559e-01 3.96838278e-01 -2.25252509e-01
-8.61293912e-01 -1.12492609e+00 -1.82799295e-01 9.53761220e-01
6.79300249e-01 -3.12428206e-01 -7.28207946e-01 -6.29314184e-01
-1.87774196e-01 2.82407552e-01 -2.94251353e-01 -3.96623939e-01
-3.82522523e-01 -7.50304282e-01 2.12619960e-01 3.35477233e-01
1.01967347e+00 -1.09131801e+00 -2.21803874e-01 2.80954063e-01
-2.73442656e-01 -7.29920447e-01 -7.96111107e-01 -2.89768815e-01
-7.44887590e-01 -7.25774825e-01 -8.45865011e-01 -8.15718055e-01
5.32151163e-01 3.81622851e-01 1.04778981e+00 1.82152554e-01
2.08130121e-01 -1.11101419e-01 -5.10672569e-01 4.82686795e-02
-4.08368222e-02 -2.87406176e-01 6.95281997e-02 2.94854134e-01
3.30625594e-01 -7.74604440e-01 -1.13411760e+00 7.03857005e-01
-1.02329433e+00 -3.38303000e-01 6.99669421e-01 8.72794807e-01
6.47436500e-01 5.72503626e-01 8.19560528e-01 -5.81680499e-02
9.24130142e-01 -2.60688007e-01 -5.52535772e-01 2.50568837e-01
-1.00739050e+00 -2.44670331e-01 9.48898375e-01 -3.88239354e-01
-1.13230038e+00 -3.84159207e-01 -2.71997079e-02 -1.55609056e-01
3.21810767e-02 3.06480885e-01 -6.16925776e-01 -1.13781422e-01
4.60136324e-01 2.96880752e-01 -2.45752037e-01 -4.03483748e-01
1.94015428e-01 9.78125930e-01 5.56474566e-01 -2.76505291e-01
8.62121701e-01 2.28246748e-01 -1.01604663e-01 -2.65698969e-01
-5.52490771e-01 -2.94507593e-01 -1.41973957e-01 -3.57890278e-01
6.94291711e-01 -9.32076752e-01 -5.28282046e-01 5.12065828e-01
-8.25429261e-01 1.44751504e-01 -5.74692488e-02 4.63514447e-01
-2.10113496e-01 6.10865116e-01 -6.32871389e-01 -7.72603095e-01
-4.15735632e-01 -1.16956544e+00 6.81496382e-01 7.15989053e-01
8.35031793e-02 -4.79303539e-01 -1.76142767e-01 1.58490330e-01
4.60724860e-01 -4.92572933e-02 7.66287029e-01 -9.02058780e-02
-5.02937138e-01 1.62805021e-02 -7.88861156e-01 7.03794301e-01
3.61427158e-01 -1.27342656e-01 -7.69739985e-01 -3.77433300e-01
2.66539663e-01 4.03402597e-02 7.52674997e-01 3.38178486e-01
1.42874515e+00 -2.87482262e-01 4.93426360e-02 5.78881979e-01
1.63721251e+00 4.36486602e-01 9.28378701e-01 5.56759596e-01
4.71442610e-01 3.06038052e-01 8.55377376e-01 5.58049798e-01
3.14685851e-01 6.77457571e-01 4.27007914e-01 -4.80533540e-02
-2.43527040e-01 -2.10817635e-01 3.59703511e-01 9.07777190e-01
5.11255711e-02 -1.93412855e-01 -3.57664287e-01 4.61313933e-01
-1.60692465e+00 -8.04100275e-01 3.53427418e-02 2.51555896e+00
1.01152027e+00 4.26910728e-01 -3.84911411e-02 4.10381973e-01
1.03500903e+00 3.87784570e-01 -5.14634550e-01 -3.77432227e-01
-1.73243076e-01 -1.53175846e-01 3.86331260e-01 2.53571928e-01
-7.84863889e-01 4.38235670e-01 5.10922575e+00 1.46608281e+00
-1.17088032e+00 2.27007400e-02 6.67442560e-01 -7.79203922e-02
-2.92704940e-01 -8.07683691e-02 -4.11425620e-01 1.00567222e+00
5.35178661e-01 -3.25041339e-02 5.74824333e-01 5.00608504e-01
5.89712381e-01 -1.86040685e-01 -5.98705053e-01 1.17377341e+00
-1.47206128e-01 -7.50552833e-01 6.81216717e-02 -9.41319168e-02
8.03804755e-01 -3.92311394e-01 9.60216373e-02 4.46369797e-02
-1.91294611e-01 -7.98113227e-01 7.47056305e-01 6.88592851e-01
1.00153351e+00 -9.47054386e-01 8.00527275e-01 1.87168166e-01
-1.44531357e+00 -2.51784474e-01 -5.60773432e-01 1.63276568e-02
2.17255130e-01 1.06835306e+00 5.47625422e-02 7.00183928e-01
9.04693723e-01 5.20883918e-01 -6.30451739e-01 1.24445438e+00
-2.52737254e-01 4.57444489e-01 -1.64924078e-02 5.56335188e-02
7.77790546e-02 -4.46973234e-01 6.37549639e-01 9.56016123e-01
5.73478878e-01 9.25339535e-02 3.81323323e-02 5.71053803e-01
6.07085135e-03 2.07439199e-01 1.05819009e-01 5.23272157e-01
7.87752151e-01 1.19179308e+00 -4.13059503e-01 -3.17194521e-01
-4.85560179e-01 1.12974775e+00 -1.35586992e-01 4.02850688e-01
-8.38809431e-01 -8.14923823e-01 3.67680013e-01 1.00130998e-01
4.69684303e-01 2.12733001e-01 -4.52250242e-01 -9.36595559e-01
4.93011922e-01 -8.20243299e-01 3.07545155e-01 -7.80151248e-01
-1.48630595e+00 6.07646585e-01 -4.01130050e-01 -1.84767580e+00
2.53971785e-01 -1.15101770e-01 -8.16177845e-01 1.17424130e+00
-1.75077820e+00 -6.03415906e-01 -5.85922062e-01 7.15919137e-01
4.20917720e-01 2.85407100e-02 5.03432691e-01 6.02826238e-01
-5.14888227e-01 9.08389509e-01 2.42390990e-01 -1.41804725e-01
1.03443122e+00 -9.83016908e-01 -2.44198620e-01 8.62033010e-01
-4.78716552e-01 4.76101816e-01 6.89520895e-01 -5.01362264e-01
-8.11716735e-01 -9.65807915e-01 7.29822338e-01 1.77565441e-01
3.41339141e-01 1.72840342e-01 -1.04108906e+00 -2.42448807e-01
-3.87397371e-02 2.34000593e-01 3.18870097e-01 -2.77783602e-01
-4.29612130e-01 -8.21481645e-01 -1.48467290e+00 4.67746079e-01
8.58151257e-01 -5.58136284e-01 -3.83213907e-01 -1.72907636e-01
6.85559928e-01 -1.96600407e-01 -1.08248448e+00 5.59338689e-01
8.40193212e-01 -1.42995679e+00 7.66160607e-01 3.04786325e-01
3.80992740e-01 -1.05389488e+00 -1.32701650e-01 -1.32793880e+00
-6.50950789e-01 -4.31557000e-01 -4.76473756e-02 1.66749382e+00
1.47165611e-01 -6.05149150e-01 3.78855795e-01 1.16362736e-01
-5.71242906e-02 -1.05727100e+00 -8.88445199e-01 -7.58103669e-01
-1.91485152e-01 -3.46993953e-02 9.41767871e-01 5.68819523e-01
-1.69479147e-01 5.06472066e-02 -3.71227831e-01 2.50766516e-01
5.06606162e-01 9.35178995e-02 2.74817526e-01 -1.00734115e+00
-1.19665071e-01 -4.84970361e-01 -5.02967894e-01 -1.30231845e+00
-6.52676523e-01 -1.72121555e-01 6.95835650e-02 -1.67499053e+00
2.22870529e-01 -5.15127897e-01 -9.44954634e-01 -8.71297121e-02
-4.92841721e-01 5.09814978e-01 2.17719063e-01 4.83134627e-01
-7.35344529e-01 5.94748855e-01 1.59447289e+00 -2.71280296e-03
-2.96188205e-01 -8.69342238e-02 -7.96200156e-01 6.29744351e-01
7.36809969e-01 -3.73581380e-01 -3.35267246e-01 -2.80079365e-01
2.03338221e-01 8.12442154e-02 2.03425080e-01 -1.37805009e+00
3.53145957e-01 -1.50219560e-01 7.47286677e-01 -7.26732969e-01
1.30171636e-02 -7.80606329e-01 -2.32335314e-01 2.99157023e-01
-5.89966960e-02 -1.40993357e-01 -1.98899999e-01 5.90190113e-01
-4.74227428e-01 -3.28459628e-02 9.81720090e-01 5.93730472e-02
-8.31828892e-01 3.35200518e-01 -8.13254938e-02 -3.72975618e-02
6.33525252e-01 -5.03165483e-01 -2.74822146e-01 -4.80758518e-01
-3.61936629e-01 2.54101694e-01 5.87380171e-01 4.20911700e-01
6.55863822e-01 -1.58015525e+00 -6.24282777e-01 2.40999699e-01
2.15557322e-01 -3.38389546e-01 6.05789721e-01 7.78762400e-01
-2.03274459e-01 1.46272155e-02 -3.35900337e-01 -4.48508203e-01
-1.00224566e+00 4.03088987e-01 4.10522580e-01 -4.31206524e-01
-1.79049313e-01 7.66761363e-01 2.27547526e-01 -1.72492731e-02
1.56822637e-01 -1.77220330e-01 -6.14244103e-01 3.48959602e-02
7.95716882e-01 5.65390944e-01 1.62854761e-01 -6.53150260e-01
-3.34731251e-01 8.71243596e-01 7.64444768e-02 5.87076619e-02
1.00834358e+00 -8.93098950e-01 -2.76631802e-01 1.63859636e-01
1.18506193e+00 2.75954813e-01 -1.42225790e+00 -1.96133867e-01
-4.84913468e-01 -7.51937568e-01 2.83245265e-01 -1.03138649e+00
-1.14158249e+00 5.81083000e-01 1.14257753e+00 4.01755214e-01
1.87693310e+00 -3.80559266e-01 1.05327284e+00 -2.86703348e-01
1.93310112e-01 -1.39931047e+00 1.63285762e-01 2.52203375e-01
7.78878510e-01 -1.00766039e+00 1.19944572e-01 -3.99578601e-01
-6.19113564e-01 1.10804152e+00 8.01315367e-01 -5.88106588e-02
5.41176498e-01 -9.12769213e-02 1.20313339e-01 5.16668186e-02
-3.73105854e-01 -4.48902324e-03 5.05870283e-01 5.75547218e-01
3.10339093e-01 -9.34334099e-02 -7.52345145e-01 7.69145787e-01
-1.12023026e-01 1.45649612e-01 3.25205624e-01 5.23610532e-01
-7.33635306e-01 -1.03694224e+00 -6.09417558e-01 5.50026357e-01
-5.37140489e-01 -2.05922782e-01 3.01139772e-01 2.19521075e-01
6.43546820e-01 1.42776680e+00 5.56942187e-02 -6.31973565e-01
5.14343143e-01 -4.74708468e-01 1.08210519e-01 -2.69487686e-02
-3.28873068e-01 2.14121476e-01 -9.94452387e-02 -8.01528871e-01
-4.43261236e-01 -4.11666214e-01 -1.14694798e+00 -1.83555365e-01
-4.59625453e-01 9.55047011e-02 3.39149088e-01 8.23105216e-01
1.72172353e-01 7.51662791e-01 1.18101847e+00 -4.29220915e-01
-6.18883908e-01 -1.01035011e+00 -8.59808743e-01 4.00806189e-01
4.65730399e-01 -5.49505353e-01 -6.27303779e-01 -2.21029699e-01] | [11.735082626342773, -1.9158419370651245] |
fe128a6c-3566-4cf3-82f3-0e34259dd267 | 190600551 | 1906.00551 | null | https://arxiv.org/abs/1906.00551v1 | https://arxiv.org/pdf/1906.00551v1.pdf | HERA: Partial Label Learning by Combining Heterogeneous Loss with Sparse and Low-Rank Regularization | Partial Label Learning (PLL) aims to learn from the data where each training instance is associated with a set of candidate labels, among which only one is correct. Most existing methods deal with such problem by either treating each candidate label equally or identifying the ground-truth label iteratively. In this paper, we propose a novel PLL approach called HERA, which simultaneously incorporates the HeterogEneous Loss and the SpaRse and Low-rAnk procedure to estimate the labeling confidence for each instance while training the model. Specifically, the heterogeneous loss integrates the strengths of both the pairwise ranking loss and the pointwise reconstruction loss to provide informative label ranking and reconstruction information for label identification, while the embedded sparse and low-rank scheme constrains the sparsity of ground-truth label matrix and the low rank of noise label matrix to explore the global label relevance among the whole training data for improving the learning model. Extensive experiments on both artificial and real-world data sets demonstrate that our method can achieve superior or comparable performance against the state-of-the-art methods. | ['Yidong Li', 'Congyan Lang', 'Yi Jin', 'Songhe Feng', 'Gengyu Lyu', 'Guojun Dai'] | 2019-06-03 | null | null | null | null | ['partial-label-learning'] | ['methodology'] | [ 4.44089055e-01 1.18242562e-01 -5.56241572e-01 -6.32307351e-01
-1.34047306e+00 -3.82981598e-01 2.28280082e-01 3.36280882e-01
-1.78289995e-01 5.96534908e-01 -7.90788140e-03 4.31563765e-01
-3.29834551e-01 -5.25220752e-01 -5.05331635e-01 -9.72636998e-01
2.62037247e-01 7.11462379e-01 -1.37729406e-01 5.94125807e-01
3.82294245e-02 1.34870768e-01 -1.45635259e+00 2.37982556e-01
7.87312567e-01 1.31343865e+00 1.07424952e-01 -1.95513994e-01
-1.37320250e-01 1.15871871e+00 -8.15393925e-02 -3.09467524e-01
2.48255938e-01 -3.17070812e-01 -7.16395378e-01 3.13736022e-01
6.50732815e-01 4.81092595e-02 -8.98510814e-02 1.38320673e+00
5.03746986e-01 -2.60361005e-02 7.79455781e-01 -1.14390934e+00
-2.03014895e-01 4.40589964e-01 -8.90739679e-01 -3.38140517e-01
1.25160053e-01 -1.73385113e-01 1.31909788e+00 -1.16655350e+00
5.30943215e-01 1.13041639e+00 8.39135349e-01 2.58432060e-01
-1.32989907e+00 -8.43983769e-01 3.44000369e-01 9.50189829e-02
-1.58461976e+00 -2.27788165e-01 9.56669986e-01 -5.74046016e-01
5.31847216e-02 8.60209167e-02 1.09882161e-01 5.99265158e-01
-3.49667698e-01 7.84157038e-01 1.45940304e+00 -2.52267420e-01
1.97538838e-01 9.48300809e-02 5.07665813e-01 1.21213520e+00
1.91220760e-01 3.86371762e-02 -5.56876779e-01 -4.50017095e-01
4.30563331e-01 1.76265955e-01 -2.87445396e-01 -7.24086106e-01
-1.22053993e+00 7.45610952e-01 4.48414445e-01 6.89591914e-02
-4.31608766e-01 -2.68643927e-02 3.76600683e-01 -9.15678442e-02
5.56825459e-01 2.15407431e-01 -4.28595930e-01 5.77485383e-01
-1.06322360e+00 -5.46254776e-02 5.07684529e-01 7.54802227e-01
1.27413166e+00 -4.47551697e-01 -5.84172547e-01 1.13756263e+00
7.60552585e-01 3.76847476e-01 2.93837845e-01 -1.20299399e+00
5.52220821e-01 8.52476776e-01 9.28834528e-02 -1.22410595e+00
-3.65992010e-01 -9.53938544e-01 -8.79979849e-01 5.67336893e-03
4.40957278e-01 8.92121419e-02 -6.78915381e-01 2.15243196e+00
6.80115759e-01 7.41170347e-01 -2.32054040e-01 8.89059365e-01
8.45557213e-01 6.22329593e-01 2.73840308e-01 -5.52248955e-01
1.27257204e+00 -1.04583180e+00 -7.87451267e-01 -2.91709840e-01
6.89102232e-01 -6.78650737e-01 8.89263988e-01 2.62884647e-01
-8.61764848e-01 -6.89838767e-01 -9.33464587e-01 -1.46483444e-03
2.27798089e-01 7.40492582e-01 5.46625555e-01 2.50476956e-01
-4.90192056e-01 5.39013028e-01 -5.23162842e-01 1.54893622e-01
4.41651791e-01 3.70451659e-01 -4.31089520e-01 -6.86380386e-01
-1.07630455e+00 4.72451419e-01 3.90364528e-01 1.28192037e-01
-9.27976429e-01 -6.55088603e-01 -7.80036271e-01 4.00305539e-02
6.71887815e-01 -5.88482857e-01 9.33289587e-01 -6.72450304e-01
-1.06783581e+00 1.05918837e+00 -1.96774527e-01 6.52954578e-02
3.60617489e-01 4.04764116e-02 -8.75187293e-02 -1.80152357e-02
5.62261462e-01 6.53449774e-01 6.24601245e-01 -1.62648451e+00
-7.66622066e-01 -5.28470874e-01 -2.25731522e-01 2.95649469e-01
-1.40325710e-01 -4.40475345e-01 -4.68211383e-01 -6.91160381e-01
7.00690925e-01 -9.16261852e-01 -2.15258554e-01 6.94323033e-02
-4.18844283e-01 -3.99368495e-01 6.66641891e-01 -5.39434731e-01
1.22627497e+00 -2.25265670e+00 1.55100882e-01 4.83795524e-01
2.42193460e-01 -1.18613251e-01 -2.86744177e-01 2.45646983e-01
-1.50558621e-01 -2.15463996e-01 -3.33807945e-01 -5.78394115e-01
-6.64614886e-02 2.78543264e-01 -1.32230699e-01 6.36647701e-01
-3.39804068e-02 6.56662881e-01 -1.22352350e+00 -7.65098631e-01
-4.69598956e-02 4.83184397e-01 -3.52662414e-01 2.80337065e-01
-1.52843520e-01 6.44605219e-01 -7.65380621e-01 6.38260841e-01
7.37046480e-01 -7.15616047e-01 3.34472358e-01 -5.77243984e-01
2.66632825e-01 4.35710289e-02 -1.61162579e+00 1.71975100e+00
-2.55358994e-01 -2.10831612e-01 2.09741727e-01 -8.08198035e-01
7.82142103e-01 2.40643471e-01 6.87751532e-01 -4.57788795e-01
6.50693029e-02 4.28013265e-01 -6.83066189e-01 -3.89846146e-01
-1.14508398e-01 -3.11781436e-01 6.79116696e-02 5.22351325e-01
1.41928032e-01 3.25335413e-01 1.05355032e-01 2.72399038e-01
9.28426802e-01 6.01774417e-02 2.49329910e-01 -2.38984451e-01
7.61129975e-01 -3.15436810e-01 1.04337752e+00 5.73411345e-01
-8.49739686e-02 4.47518975e-01 2.94717729e-01 -2.18117312e-01
-5.77769160e-01 -8.76504421e-01 -2.12336376e-01 1.08090806e+00
4.02407378e-01 -2.71392703e-01 -6.24338448e-01 -1.26536238e+00
3.14834937e-02 4.03502136e-01 -6.19821548e-01 -1.86648503e-01
-2.66556531e-01 -6.74266398e-01 1.01425022e-01 2.09157571e-01
5.63666403e-01 -7.03247309e-01 3.56931299e-01 1.47199467e-01
-5.39786160e-01 -6.72038615e-01 -8.23350430e-01 3.01562071e-01
-7.93605149e-01 -1.36106038e+00 -3.54606926e-01 -1.12769759e+00
1.08136785e+00 3.09430212e-01 1.05916810e+00 7.99493417e-02
5.27028320e-03 1.25286222e-01 -1.98529735e-01 3.41715515e-01
-1.15810007e-01 -9.90879163e-02 -1.65200338e-01 5.13317168e-01
1.71439990e-01 -3.77179772e-01 -4.05442536e-01 5.34059703e-01
-6.94794893e-01 -7.51784397e-03 6.62801564e-01 1.19728196e+00
1.26040709e+00 4.18246955e-01 7.47134149e-01 -1.48392630e+00
1.90611273e-01 -6.48994207e-01 -4.77788091e-01 5.07082403e-01
-1.04662907e+00 1.98227003e-01 2.66758323e-01 -2.85605282e-01
-9.51807439e-01 6.05758965e-01 1.39903529e-02 -3.34954500e-01
1.52433082e-01 7.61123955e-01 -4.72425222e-01 -1.65206060e-01
2.92198002e-01 -1.91481151e-02 -2.07882777e-01 -7.25362360e-01
2.59563178e-01 4.03283209e-01 4.77100849e-01 -6.82664990e-01
8.01443696e-01 4.54546720e-01 4.37226534e-01 2.83375476e-02
-1.90853143e+00 -9.82401550e-01 -5.46895742e-01 -2.18365550e-01
5.35449028e-01 -1.23805058e+00 -4.22972739e-01 4.70627636e-01
-7.25197077e-01 1.12150535e-01 -3.79047662e-01 4.66487944e-01
-3.18093151e-01 5.50549746e-01 -6.72648370e-01 -5.70736766e-01
-5.32146394e-02 -1.27010012e+00 1.31420696e+00 2.61821002e-02
1.95154101e-01 -8.79033923e-01 2.99519092e-01 7.08843350e-01
-1.34912342e-01 1.16239198e-01 9.97995019e-01 -5.56762636e-01
-6.16445124e-01 -3.72814149e-01 -5.28696418e-01 4.04278278e-01
1.63124233e-01 -6.19854093e-01 -1.05440044e+00 -6.03681982e-01
1.96742907e-01 -6.86110318e-01 1.05701792e+00 2.66146690e-01
1.11138988e+00 -3.62105042e-01 -3.06527138e-01 2.99482912e-01
1.51933360e+00 -2.22908244e-01 1.64850250e-01 -8.03767443e-02
1.06593812e+00 7.88926065e-01 1.14255667e+00 4.07963365e-01
5.30285120e-01 6.52710259e-01 3.95774394e-01 -1.02097496e-01
-1.20282628e-01 -5.31808555e-01 1.99489500e-02 1.12303460e+00
5.03771484e-01 1.35696558e-02 -5.70353925e-01 1.85821816e-01
-2.07122064e+00 -7.34966099e-01 -6.14744239e-02 2.29710913e+00
9.80596900e-01 -8.73404816e-02 -1.66030735e-01 2.11061418e-01
1.02922702e+00 2.02217728e-01 -6.36018276e-01 6.14077926e-01
-5.27975820e-02 -3.21499348e-01 4.39599931e-01 6.15671277e-01
-1.31413555e+00 5.37461817e-01 5.80555105e+00 1.18991566e+00
-7.35534549e-01 3.68874073e-01 9.04902756e-01 2.22618505e-01
-3.85363162e-01 8.25666785e-02 -8.80585074e-01 4.72287625e-01
3.57229054e-01 3.15368861e-01 2.77752787e-01 8.70627999e-01
-2.61134957e-03 -5.36555797e-03 -1.13748372e+00 1.07333434e+00
1.20775960e-01 -9.38862622e-01 -9.58195478e-02 5.15943840e-02
1.11944985e+00 -2.00867370e-01 1.37208953e-01 3.31491858e-01
3.76389056e-01 -7.95426428e-01 7.06370771e-01 6.56951845e-01
1.00745237e+00 -7.03794777e-01 8.53625536e-01 5.27499795e-01
-1.36440015e+00 -1.59762427e-01 -2.61379957e-01 2.63633221e-01
-6.24815784e-02 1.06589854e+00 -4.94583637e-01 5.76933622e-01
1.92745984e-01 1.14607692e+00 -6.07209146e-01 9.03758049e-01
-3.43332112e-01 7.54230261e-01 -2.40440190e-01 5.66609561e-01
2.84801066e-01 -3.00711423e-01 1.85405448e-01 9.16579187e-01
5.78671992e-02 5.33963926e-02 9.59685862e-01 5.69476724e-01
-4.09054220e-01 3.62297058e-01 -2.11887330e-01 1.98840126e-01
6.83367908e-01 1.47487104e+00 -5.94961286e-01 -1.28451020e-01
-4.52100128e-01 6.35360599e-01 6.15532279e-01 3.13364357e-01
-5.55437386e-01 1.86459064e-01 2.44270504e-01 8.68634600e-03
-3.51334624e-02 2.87276417e-01 -3.23567599e-01 -1.05825257e+00
-6.22316785e-02 -7.34260321e-01 6.92889035e-01 -5.27602136e-01
-1.76554692e+00 1.98397592e-01 -3.70793551e-01 -1.32167041e+00
7.06643537e-02 -3.03141385e-01 -2.38577649e-01 8.45307827e-01
-1.65636349e+00 -1.31007206e+00 -3.05082589e-01 3.69534194e-01
2.56118834e-01 -1.30903184e-01 6.26515210e-01 6.57360852e-01
-7.06094980e-01 4.65362102e-01 2.56079108e-01 -5.69535680e-02
9.30578291e-01 -1.15173340e+00 -3.61199617e-01 4.52218026e-01
2.44419724e-01 4.44034606e-01 4.11510795e-01 -6.27202988e-01
-8.88891816e-01 -1.41806531e+00 9.63274240e-01 -1.70614302e-01
2.87099212e-01 -3.28856036e-02 -9.92702723e-01 5.33796012e-01
-4.56619322e-01 3.37463528e-01 8.47049475e-01 2.57993907e-01
-7.65319407e-01 -3.47270638e-01 -1.35941148e+00 1.11974761e-01
7.43014872e-01 -6.50806189e-01 -1.70780808e-01 5.38976967e-01
5.43434501e-01 -1.59930706e-01 -8.48828733e-01 7.08301902e-01
3.76190960e-01 -6.31058455e-01 9.17546570e-01 -2.07230523e-01
1.76417783e-01 -8.22331131e-01 -3.10217053e-01 -9.72960472e-01
-8.27617288e-01 1.06488906e-01 -2.61762049e-02 1.61213243e+00
3.42510581e-01 -3.80244702e-01 9.94560838e-01 4.94765043e-01
5.47557026e-02 -1.00317299e+00 -8.82170081e-01 -3.17111552e-01
-4.47171539e-01 -1.35312319e-01 3.31242889e-01 1.28122270e+00
-3.51271659e-01 6.07291758e-01 -6.27408683e-01 3.58147770e-01
1.07802474e+00 2.86142707e-01 1.75136119e-01 -1.56159294e+00
-4.33917046e-01 7.47052878e-02 -2.64023393e-01 -1.06525004e+00
5.22555053e-01 -1.31955814e+00 2.47473598e-01 -1.50330889e+00
8.22231770e-01 -9.46208894e-01 -8.64348769e-01 6.75702393e-01
-4.88312721e-01 3.62268984e-01 4.26246859e-02 5.39948940e-01
-1.10444403e+00 5.00642598e-01 1.23012686e+00 -3.13401431e-01
2.68505901e-01 4.64264154e-02 -7.81242967e-01 7.53220499e-01
3.52376461e-01 -1.00756657e+00 -6.04703724e-01 -2.28037849e-01
3.10236484e-01 1.14008665e-01 2.49368384e-01 -9.04788256e-01
3.41891527e-01 -7.99081698e-02 2.39700869e-01 -6.08373165e-01
1.68901518e-01 -1.03916526e+00 2.87555039e-01 2.65882850e-01
-8.30902219e-01 -5.54436743e-01 -4.50142413e-01 1.02489758e+00
-3.59056354e-01 -4.33553487e-01 1.14436507e+00 -9.77078751e-02
-4.83996183e-01 6.19970739e-01 3.23286593e-01 2.34728232e-01
8.25260222e-01 2.64902920e-01 -1.11061908e-01 -1.09894589e-01
-7.42953122e-01 5.75717986e-01 4.74221021e-01 -4.45097871e-03
3.46242338e-01 -1.73351300e+00 -7.88222969e-01 1.11653097e-01
2.69616187e-01 1.22292973e-01 4.08449858e-01 6.20516062e-01
-2.33560335e-03 1.41755462e-01 2.04341680e-01 -6.27604008e-01
-1.18859410e+00 7.37050772e-01 2.56181002e-01 -8.72733474e-01
-2.18358517e-01 1.05171120e+00 5.97991765e-01 -7.71691799e-01
5.40445566e-01 2.90487647e-01 -3.63818079e-01 2.28954449e-01
4.51532632e-01 4.24941123e-01 -6.94093406e-02 -9.16483164e-01
-2.92865276e-01 9.39241588e-01 -1.19148649e-01 2.23753110e-01
1.03066909e+00 -3.66630048e-01 -3.86769563e-01 6.30596280e-01
1.36427426e+00 2.24438161e-02 -1.28920829e+00 -8.35342109e-01
1.04834750e-01 -5.05848706e-01 3.36412758e-01 -1.03289950e+00
-1.34181261e+00 5.84092975e-01 6.70726299e-01 -3.01512688e-01
1.15487742e+00 -4.03798521e-02 5.57542980e-01 2.67048150e-01
5.73707163e-01 -9.32161927e-01 3.14179480e-01 3.09387147e-01
4.91300970e-01 -1.42913496e+00 3.96875553e-02 -8.14944088e-01
-5.83897948e-01 5.32775342e-01 4.52745348e-01 -4.83841002e-02
6.70135796e-01 -1.19030185e-01 1.58853397e-01 -2.81248063e-01
-6.71377838e-01 -3.79651487e-02 4.56368238e-01 1.57849342e-01
3.18188578e-01 2.01895311e-02 -3.54422867e-01 7.50669301e-01
5.02947867e-01 -1.74876466e-01 -2.58623689e-01 5.49291313e-01
-5.00905216e-01 -1.27560484e+00 -3.68272513e-01 6.20301604e-01
-4.23786521e-01 6.59907386e-02 -2.14755997e-01 1.53240263e-01
5.12950063e-01 9.49235320e-01 -5.15673459e-01 -3.31464827e-01
2.20398501e-01 2.31686234e-01 3.14412057e-01 -8.88246715e-01
-4.29385960e-01 4.06158566e-01 -5.27314506e-02 -4.99840081e-01
-5.68557024e-01 -8.33302557e-01 -1.21116495e+00 1.74767777e-01
-5.11953831e-01 4.18708056e-01 3.63016158e-01 8.99240315e-01
1.50430322e-01 3.82523775e-01 9.48331237e-01 -6.21009171e-01
-7.27933288e-01 -8.27493906e-01 -1.11400831e+00 6.14951372e-01
2.70245492e-01 -9.46391344e-01 -4.85183686e-01 5.36420830e-02] | [9.528142929077148, 4.041176795959473] |
67e87a5a-c7cb-42f8-9883-85e49dd34118 | heterogeneously-distributed-joint-radar | 2107.13838 | null | https://arxiv.org/abs/2107.13838v2 | https://arxiv.org/pdf/2107.13838v2.pdf | Heterogeneously-Distributed Joint Radar Communications: Bayesian Resource Allocation | Due to spectrum scarcity, the coexistence of radar and wireless communication has gained substantial research interest recently. Among many scenarios, the heterogeneouslydistributed joint radar-communication system is promising due to its flexibility and compatibility of existing architectures. In this paper, we focus on a heterogeneous radar and communication network (HRCN), which consists of various generic radars for multiple target tracking (MTT) and wireless communications for multiple users. We aim to improve the MTT performance and maintain good throughput levels for communication users by a well-designed resource allocation. The problem is formulated as a Bayesian Cram\'er-Rao bound (CRB) based minimization subjecting to resource budgets and throughput constraints. The formulated nonconvex problem is solved based on an alternating descent-ascent approach. Numerical results demonstrate the efficacy of the proposed allocation scheme for this heterogeneous network. | ['Björn Ottersten', 'Bhavani Shankar M. R.', 'Kumar Vijay Mishra', 'Linlong Wu'] | 2021-07-29 | null | null | null | null | ['joint-radar-communication'] | ['robots'] | [ 3.22750241e-01 -1.93998083e-01 -1.61179423e-01 -1.44415468e-01
-8.70287538e-01 -2.61012852e-01 1.76040336e-01 -4.04620975e-01
-4.18318301e-01 1.17727602e+00 -1.33553669e-01 -4.15811330e-01
-7.62914419e-01 -6.87282562e-01 1.55673876e-01 -1.12394249e+00
-2.36538157e-01 3.68629903e-01 -3.90834004e-01 1.01992987e-01
-9.61269513e-02 7.76695251e-01 -7.91592956e-01 -9.21543598e-01
1.01018071e+00 1.25334954e+00 2.43473768e-01 5.54190874e-01
3.16440284e-01 3.21512908e-01 -6.87982738e-01 -2.66349733e-01
5.46188533e-01 -3.97257268e-01 2.67145783e-01 2.11919576e-01
-8.28831494e-02 -2.87062563e-02 -2.67421961e-01 1.19366586e+00
8.08621407e-01 7.40607977e-02 6.58314586e-01 -1.32967579e+00
-3.00816387e-01 3.29924017e-01 -1.16736007e+00 2.47018501e-01
-2.37230703e-01 -4.58442956e-01 9.23482835e-01 -5.69093823e-01
7.23082945e-02 1.07850182e+00 3.85245383e-01 3.17066669e-01
-8.94795418e-01 -1.07716870e+00 4.77988347e-02 8.96420628e-02
-1.82826173e+00 -4.49737817e-01 3.90490085e-01 -9.96268317e-02
7.37639964e-02 4.53101069e-01 5.36675215e-01 3.50993246e-01
3.86244565e-01 3.36071312e-01 1.02229369e+00 -3.14989954e-01
2.03647479e-01 -1.92222908e-01 1.35545895e-01 5.20321667e-01
1.10561633e+00 3.67058724e-01 -1.89615458e-01 -3.33016515e-01
5.73079288e-01 -1.95685670e-01 -5.31884730e-01 -1.75727561e-01
-8.59823346e-01 9.78835940e-01 2.95547932e-01 1.32887140e-01
-8.07567656e-01 2.17543229e-01 -2.82657623e-01 1.79723442e-01
5.74052274e-01 8.70587602e-02 8.50728527e-02 3.72625738e-01
-1.02541506e+00 1.55620158e-01 8.85635555e-01 1.18915570e+00
1.07156202e-01 6.48608744e-01 -2.88347065e-01 6.36087000e-01
7.69284606e-01 1.42395437e+00 -3.67983371e-01 -5.67617357e-01
4.43830699e-01 -1.65233657e-01 4.86603290e-01 -1.09894288e+00
-6.28583014e-01 -1.55571735e+00 -1.27754891e+00 1.59771591e-01
1.27962142e-01 -1.03748345e+00 -6.78474307e-01 1.66642785e+00
3.13147098e-01 2.31023952e-01 5.05105972e-01 1.13667774e+00
2.99529403e-01 7.09508359e-01 -1.48148209e-01 -9.25624371e-01
1.15592194e+00 -2.77355969e-01 -1.08968520e+00 -3.16350579e-01
-9.24868695e-03 -8.59116971e-01 -4.31917787e-01 2.83253491e-01
-1.07397771e+00 -6.51787296e-02 -1.34776819e+00 1.06552899e+00
3.11319679e-01 1.69942215e-01 4.28192496e-01 1.29903352e+00
-7.85713673e-01 -3.43294472e-01 -3.42260033e-01 -1.79443926e-01
3.41558963e-01 2.87091255e-01 4.71060365e-01 -4.86953914e-01
-1.04911232e+00 9.15911078e-01 1.23199239e-01 7.48112261e-01
-8.77721965e-01 -5.51233709e-01 -6.72079980e-01 -1.18027024e-01
6.86331153e-01 -8.74847770e-01 9.31178093e-01 -2.12100282e-01
-1.37838876e+00 3.94528322e-02 -2.58897077e-02 -5.33547640e-01
4.00155991e-01 4.74280901e-02 -5.56168079e-01 -5.83627038e-02
-1.40679991e-02 -1.21114932e-01 6.84942961e-01 -1.28442693e+00
-9.20966089e-01 -4.20857102e-01 -2.11216688e-01 4.00854707e-01
-7.32438127e-03 1.74948856e-01 -6.30005077e-02 -7.78905511e-01
2.13446096e-01 -1.10954571e+00 -6.78405881e-01 -2.27653712e-01
-6.13780558e-01 2.98325270e-01 8.28633547e-01 -3.07156295e-01
1.10814178e+00 -1.90554702e+00 2.99639672e-01 6.49972498e-01
1.49854451e-01 6.01646081e-02 -1.01839110e-01 1.79810658e-01
4.25613314e-01 -4.11861360e-01 -1.43453270e-01 -1.47940204e-01
-5.46230376e-02 -7.57046193e-02 3.73664796e-02 1.17658567e+00
-4.43005204e-01 3.34368497e-01 -5.11664867e-01 -3.62000704e-01
-5.13483733e-02 3.33541721e-01 -6.11228589e-03 2.31635526e-01
7.90406689e-02 4.75099444e-01 -1.00092280e+00 1.13652933e+00
1.10013294e+00 1.79539829e-01 2.28896335e-01 -3.03795189e-01
-3.94148886e-01 -8.38453591e-01 -1.19266450e+00 9.90630388e-01
-6.39450073e-01 5.14448822e-01 1.01116943e+00 -1.07566309e+00
1.01710892e+00 3.81769031e-01 7.33115017e-01 -4.80492592e-01
5.73115706e-01 3.22454840e-01 3.64434421e-01 1.20176397e-01
2.84745961e-01 -6.57254994e-01 -2.59541601e-01 3.23240310e-01
-5.86979091e-01 5.31275012e-02 -4.87572290e-02 1.05341382e-01
1.04068553e+00 -5.99035800e-01 5.36785901e-01 -4.96354669e-01
5.77668488e-01 -1.63169160e-01 6.64299309e-01 1.08994198e+00
-2.17134610e-01 -2.72582561e-01 -3.39516997e-01 3.57820332e-01
-3.98952156e-01 -1.12644601e+00 -1.94698915e-01 5.82553566e-01
5.74401796e-01 4.44091380e-01 1.53860275e-03 8.38203430e-02
1.43447235e-01 6.36765301e-01 -1.71654269e-01 6.98148645e-03
-1.50536805e-01 -1.14791691e+00 3.60855103e-01 -1.38022706e-01
7.08716452e-01 -2.28848025e-01 -6.46098614e-01 3.81506175e-01
1.44554630e-01 -1.35456002e+00 -1.34035870e-01 2.12100133e-01
-5.01793742e-01 -7.64859378e-01 -1.01837444e+00 -3.78450513e-01
4.77380812e-01 8.82930875e-01 4.94262964e-01 -3.09308976e-01
-3.90226871e-01 7.19979882e-01 -3.39747608e-01 -6.67756677e-01
2.38827869e-01 -1.22190341e-01 3.45518857e-01 2.78191656e-01
-2.27862597e-01 -4.66273129e-01 -2.33757690e-01 3.91326368e-01
-4.03384030e-01 -2.69777685e-01 9.47824836e-01 5.77359617e-01
2.54340857e-01 5.34779966e-01 9.30437148e-01 -4.00148362e-01
8.99252057e-01 -5.88136017e-01 -1.04662740e+00 4.24827397e-01
-2.89400071e-01 -3.01815987e-01 2.82431245e-01 -6.54087514e-02
-1.34192216e+00 -8.59349668e-02 4.14402962e-01 -1.48881793e-01
3.70904356e-01 7.90799022e-01 -4.57248360e-01 -8.15907359e-01
1.31367221e-01 1.90398186e-01 -1.12628385e-01 9.04410779e-02
4.37993050e-01 7.93584228e-01 3.94614190e-01 -5.06660938e-01
1.40704036e+00 4.00872529e-01 5.92535555e-01 -1.23909533e+00
-1.22698629e+00 -4.37065035e-01 -3.04496735e-02 -5.73488951e-01
6.24453247e-01 -1.09370208e+00 -9.72658932e-01 -6.11856207e-03
-9.86148775e-01 3.20255399e-01 2.55475312e-01 1.22050560e+00
-3.14375222e-01 2.69392163e-01 2.06139795e-02 -1.59085119e+00
-5.36881745e-01 -7.05873251e-01 4.25995141e-01 4.11795706e-01
3.02569121e-01 -8.34166050e-01 -1.65579896e-02 1.77149668e-01
7.99123585e-01 5.13402402e-01 2.76486129e-01 -1.33604348e-01
-9.65277970e-01 -4.41488832e-01 -5.67834556e-01 -1.95267066e-01
2.02501267e-01 -7.49233544e-01 -2.97491640e-01 -7.80635715e-01
2.20369816e-01 2.43926374e-03 5.52235305e-01 1.03657913e+00
4.73970205e-01 -1.57360971e-01 -7.38022387e-01 6.06789649e-01
1.70140600e+00 6.84269011e-01 1.99099019e-01 7.66240135e-02
1.26375005e-01 2.71311581e-01 9.81238127e-01 9.85087335e-01
1.34319952e-02 6.24904335e-01 7.42887259e-01 3.44794840e-02
4.65782404e-01 7.80079544e-01 8.05714130e-02 3.76702428e-01
9.01575945e-03 -7.19173133e-01 -4.76589054e-01 3.11708480e-01
-1.80840039e+00 -9.11232829e-01 -1.62773296e-01 2.15776873e+00
3.02552968e-01 -8.65448192e-02 -1.89313412e-01 -2.94597387e-01
1.01638269e+00 2.29822546e-01 -5.61514914e-01 3.68300565e-02
-2.38145620e-01 -1.29923984e-01 1.14622319e+00 4.46558565e-01
-9.28560138e-01 2.52833456e-01 5.80446768e+00 1.20311940e+00
-5.30267894e-01 2.61697918e-01 1.07243076e-01 -1.82506576e-01
-9.52573866e-02 -2.42670313e-01 -1.01591253e+00 1.76312417e-01
9.20378923e-01 -8.45256746e-01 1.15172446e-01 3.20679575e-01
5.65330684e-01 -4.77754474e-01 -2.53564686e-01 1.02727318e+00
1.84555843e-01 -1.13769245e+00 -4.45065677e-01 4.46961403e-01
5.75699329e-01 -7.95365274e-02 -2.59296000e-02 2.02146560e-01
7.13510156e-01 -6.91926777e-01 4.90580380e-01 5.54991066e-01
5.84812403e-01 -1.08881474e+00 8.67330730e-01 2.69473404e-01
-1.51021445e+00 -3.53992879e-01 -4.41536307e-01 -2.91006290e-03
8.10434043e-01 1.10715568e+00 -4.26376015e-01 1.19113767e+00
1.08187981e-01 3.40616524e-01 3.51494879e-01 1.68552160e+00
-5.51479124e-02 2.20090374e-01 -4.12694514e-01 -4.14179802e-01
4.11755562e-01 -5.60704470e-01 1.04723620e+00 1.09861410e+00
8.68454695e-01 6.68026805e-01 6.68458760e-01 3.85438234e-01
1.56641409e-01 -6.16168566e-02 -5.97391367e-01 2.02886492e-01
8.57996404e-01 1.60878766e+00 -5.17407775e-01 1.88933853e-02
-4.09981847e-01 1.68058619e-01 -2.58992136e-01 4.31367815e-01
-1.11325741e+00 -5.58786929e-01 6.96611941e-01 -5.70127666e-01
3.66373837e-01 -5.40388465e-01 -3.23742419e-01 -6.81894362e-01
-2.85594076e-01 -2.09568858e-01 3.62273127e-01 -2.91567892e-01
-1.31124783e+00 7.04368114e-01 6.98190590e-04 -1.51067400e+00
2.77835160e-01 -1.00304432e-01 -4.58314240e-01 1.09894228e+00
-1.89228404e+00 -1.11213660e+00 -3.33230764e-01 5.61658800e-01
2.75636554e-01 -5.99588037e-01 4.10864085e-01 4.14683700e-01
-7.71993279e-01 3.32020968e-01 4.28174108e-01 -3.16556364e-01
1.38784304e-01 -5.85612953e-01 -8.16837251e-01 1.09896803e+00
-4.32613224e-01 1.84523851e-01 1.18120861e+00 -7.31529951e-01
-1.51777542e+00 -1.25862408e+00 2.74783760e-01 6.04710937e-01
8.19162011e-01 -9.51774120e-02 1.91229448e-01 2.14728653e-01
5.04814208e-01 -2.37613067e-01 1.02138793e+00 -5.68448789e-02
1.33198485e-01 -5.67045435e-02 -1.14032125e+00 5.39211392e-01
7.25195229e-01 3.15408677e-01 -4.71672937e-02 5.61210513e-01
3.07330459e-01 -2.22799048e-01 -7.69610584e-01 4.34893757e-01
5.58432579e-01 -3.57325703e-01 9.02690947e-01 -7.10993260e-02
-4.81713384e-01 -4.62306648e-01 -6.42963469e-01 -1.52031696e+00
-5.33105850e-01 -8.08497012e-01 7.74577036e-02 1.11387396e+00
4.09508377e-01 -6.38524294e-01 7.75151670e-01 6.37491643e-02
-1.92738231e-02 -4.28763270e-01 -1.28231466e+00 -1.16563439e+00
-4.83897299e-01 -3.01604509e-01 4.37215865e-02 3.69888246e-01
-8.20105821e-02 5.01217365e-01 -8.99081767e-01 8.66169930e-01
1.50919819e+00 3.50089431e-01 4.26172584e-01 -1.51518142e+00
-2.38616198e-01 -7.42609501e-02 -1.93709627e-01 -1.16457415e+00
3.31797630e-01 -6.59515500e-01 3.05427194e-01 -1.62094939e+00
-1.42364223e-02 -8.08043063e-01 -7.96672627e-02 -6.35932609e-02
1.78524643e-01 1.19512156e-01 3.53982687e-01 1.09960176e-01
-7.50389636e-01 9.85918641e-01 1.22028291e+00 -1.97216883e-01
7.03799427e-02 7.74988055e-01 -8.62132370e-01 3.46524775e-01
9.15786862e-01 -4.74560231e-01 -3.21026623e-01 -3.34134132e-01
-2.00055718e-01 9.77743506e-01 -2.11276151e-02 -1.12524903e+00
5.13907373e-01 -4.98094082e-01 1.89360857e-01 -8.37450922e-01
7.79375136e-01 -1.38791585e+00 3.22934657e-01 8.57053161e-01
1.64659426e-01 -3.33159953e-01 -1.08956672e-01 1.23407805e+00
-1.23911731e-01 -2.60158688e-01 1.06607544e+00 1.73597053e-01
-2.68775433e-01 4.58293974e-01 -7.95247853e-01 -2.42763296e-01
1.49915171e+00 -1.20604530e-01 -4.63081777e-01 -8.13534260e-01
-6.67550385e-01 7.00135529e-01 -5.06742835e-01 -9.77472514e-02
6.79555535e-01 -1.18877923e+00 -1.01163352e+00 -3.29806626e-01
-2.59477884e-01 -6.41725063e-01 3.79939348e-01 1.03555942e+00
1.74956024e-02 8.28936934e-01 -7.92709813e-02 -2.76383758e-01
-1.31086528e+00 -8.28751996e-02 3.54272962e-01 -1.88020676e-01
-1.65987194e-01 6.90903187e-01 -2.93389320e-01 1.66867197e-01
2.53209651e-01 4.48973805e-01 -3.03356975e-01 1.41910002e-01
3.93935442e-01 5.46368778e-01 -2.39835158e-01 -5.31689167e-01
-4.23185557e-01 6.15714431e-01 8.46169367e-02 -2.84107208e-01
1.20633614e+00 -5.63439846e-01 -1.12634614e-01 -3.32593441e-01
5.00818551e-01 2.48309255e-01 -7.75523841e-01 -4.64516848e-01
-9.33259204e-02 -7.56303489e-01 5.38356006e-01 -5.02024531e-01
-1.27163196e+00 2.68146507e-02 6.28444374e-01 2.91789860e-01
1.00643826e+00 -4.46862131e-01 4.14058417e-01 5.06673634e-01
8.01303446e-01 -6.50457799e-01 -2.59547174e-01 4.77637529e-01
5.25712669e-01 -7.89795101e-01 4.17636335e-01 -5.95179737e-01
-2.93536007e-01 9.96814132e-01 5.10586977e-01 2.81081516e-02
1.03299510e+00 5.93373120e-01 1.93268619e-02 -1.39795601e-01
-5.36302030e-01 -7.88541555e-01 -3.40108834e-02 7.45161772e-01
2.27904692e-01 3.83441150e-01 -7.31934845e-01 4.56496119e-01
3.88684869e-02 -2.40266666e-01 7.20611095e-01 1.01481318e+00
-1.01026487e+00 -9.08936262e-01 -9.07611549e-01 8.08248520e-01
-2.65271574e-01 9.22526494e-02 1.11441173e-01 7.05156267e-01
-2.78724849e-01 1.51423681e+00 -3.50895792e-01 -8.57879668e-02
3.45174998e-01 -8.90362561e-01 3.94476950e-01 -4.73748326e-01
1.46939978e-01 5.35686314e-01 3.99530917e-01 1.16333812e-01
-6.68903112e-01 -6.05015576e-01 -6.86610341e-01 -6.21840879e-02
-8.62202108e-01 7.54502416e-01 6.47755086e-01 1.06877303e+00
-3.72748598e-02 8.73280704e-01 8.88060689e-01 -6.58839881e-01
-7.47842789e-01 -6.98977411e-01 -1.17959929e+00 -8.57464850e-01
1.06447265e-01 -9.16100085e-01 -3.21385413e-01 -7.45973587e-01] | [6.2727460861206055, 1.354749083518982] |
3fea2165-32a7-4727-80ff-0b9f016b1022 | multimodal-age-and-gender-classification | 1907.10081 | null | https://arxiv.org/abs/1907.10081v1 | https://arxiv.org/pdf/1907.10081v1.pdf | Multimodal Age and Gender Classification Using Ear and Profile Face Images | In this paper, we present multimodal deep neural network frameworks for age and gender classification, which take input a profile face image as well as an ear image. Our main objective is to enhance the accuracy of soft biometric trait extraction from profile face images by additionally utilizing a promising biometric modality: ear appearance. For this purpose, we provided end-to-end multimodal deep learning frameworks. We explored different multimodal strategies by employing data, feature, and score level fusion. To increase representation and discrimination capability of the deep neural networks, we benefited from domain adaptation and employed center loss besides softmax loss. We conducted extensive experiments on the UND-F, UND-J2, and FERET datasets. Experimental results indicated that profile face images contain a rich source of information for age and gender classification. We found that the presented multimodal system achieves very high age and gender classification accuracies. Moreover, we attained superior results compared to the state-of-the-art profile face image or ear image-based age and gender classification methods. | ['Hazim Kemal Ekenel', 'Fevziye Irem Eyiokur', 'Dogucan Yaman'] | 2019-07-23 | null | null | null | null | ['age-and-gender-classification'] | ['computer-vision'] | [-1.76033694e-02 5.77815622e-02 3.72619624e-03 -9.04189229e-01
-8.26783121e-01 -1.91861331e-01 4.65284109e-01 -1.12791471e-01
-5.84233999e-01 7.53274679e-01 -1.07383601e-01 1.29713327e-01
-2.61799544e-01 -4.31508064e-01 -2.74610162e-01 -1.04417098e+00
-4.80378978e-02 3.72803688e-01 -7.70516455e-01 4.57881428e-02
1.76140815e-01 3.52852583e-01 -2.00003886e+00 1.40129626e-01
8.78510118e-01 1.77605498e+00 -4.67421025e-01 4.14810061e-01
1.48614615e-01 2.79649079e-01 -4.83440131e-01 -1.20659196e+00
-1.94867030e-01 3.24914493e-02 -5.03995776e-01 -1.24488793e-01
7.69110084e-01 -5.86178005e-01 -1.13505349e-01 9.32430863e-01
1.28352559e+00 -2.37828478e-01 1.05007803e+00 -1.52980971e+00
-7.97128201e-01 6.03082418e-01 -9.03557301e-01 -4.52485979e-01
4.24382567e-01 -1.73398554e-01 6.62687182e-01 -1.11914492e+00
8.99933577e-02 1.40612876e+00 7.27537453e-01 9.33744550e-01
-7.84398675e-01 -1.05350053e+00 -3.81057680e-01 2.47372255e-01
-1.46743107e+00 -7.98140883e-01 8.34250689e-01 -3.46032351e-01
3.53827477e-01 8.67916420e-02 2.58389145e-01 1.46749699e+00
-4.67810407e-02 9.51688945e-01 1.32387185e+00 -5.31575441e-01
-6.16643764e-02 2.44577825e-01 -1.31449010e-02 9.81810033e-01
-2.13568076e-03 2.04115525e-01 -9.50483203e-01 -3.94173384e-01
4.11671340e-01 -3.20132375e-01 2.43096069e-01 -6.05744421e-02
-6.90744996e-01 7.02210963e-01 8.51860046e-02 7.73694441e-02
-3.53864342e-01 6.98805675e-02 4.08090651e-01 2.45650023e-01
4.99783993e-01 -4.92324010e-02 -4.77199286e-01 -1.12453200e-01
-9.66138899e-01 -3.45352888e-02 5.68952739e-01 3.45945209e-01
3.16124380e-01 5.28420508e-02 -1.93358853e-01 1.28849292e+00
5.50082564e-01 9.09053624e-01 4.25014406e-01 -8.23192000e-01
2.40899786e-01 5.34816444e-01 -2.77776837e-01 -6.68237090e-01
-4.98529226e-01 -3.07871401e-01 -9.10537302e-01 2.37068355e-01
5.44741511e-01 -1.39595985e-01 -1.20290554e+00 2.08080292e+00
5.59596457e-02 -2.58341759e-01 -6.52215537e-03 5.90693772e-01
1.31928861e+00 -9.83714908e-02 5.52627265e-01 4.22961041e-02
1.65216744e+00 -2.08276257e-01 -6.74841404e-01 8.29440728e-02
2.63469368e-01 -8.34321797e-01 6.81277514e-01 4.41750795e-01
-1.54640353e+00 -7.00522661e-01 -7.57409513e-01 1.35062784e-01
-5.14327288e-01 8.74292612e-01 7.78911889e-01 1.39323914e+00
-1.13603354e+00 3.29281092e-01 -5.75105846e-01 -3.42759341e-01
6.71511292e-01 1.02891624e+00 -7.62387693e-01 2.21426278e-01
-1.14154911e+00 5.92111707e-01 8.70135948e-02 2.44743377e-01
-5.99821746e-01 -6.54166937e-01 -1.13948047e+00 4.33391146e-02
-3.38565439e-01 -8.75496089e-01 1.15718710e+00 -8.66987944e-01
-1.40502608e+00 1.48422623e+00 -3.20992768e-01 -1.41461536e-01
2.86483824e-01 -1.01563051e-01 -5.01795113e-01 2.77483582e-01
-3.35518837e-01 9.74334121e-01 1.01247072e+00 -1.13757038e+00
-5.29327571e-01 -1.11849666e+00 -5.24262547e-01 6.41567782e-02
-8.76026571e-01 3.04602265e-01 -1.11578219e-01 -1.10764787e-01
9.52699557e-02 -7.38288820e-01 5.82216680e-01 -3.22612487e-02
-2.20511615e-01 -4.53991175e-01 5.72424591e-01 -1.10962164e+00
1.10123920e+00 -2.17194390e+00 8.19444880e-02 4.42931682e-01
3.06569576e-01 1.72796220e-01 -2.18111753e-01 1.59069803e-02
-1.63396135e-01 -1.68330278e-02 3.82234678e-02 -9.35133457e-01
3.31303716e-01 -1.17252566e-01 3.27917188e-01 4.69550878e-01
1.99856311e-01 8.75269771e-01 -1.47576049e-01 -8.01694095e-01
3.03160250e-02 9.31983531e-01 -3.08471322e-01 2.77548790e-01
4.41183627e-01 3.78457308e-01 -1.81691483e-01 1.53450108e+00
1.10710704e+00 2.36161649e-01 -2.97760218e-01 -4.19741809e-01
4.26032394e-01 -5.84610701e-01 -7.84537554e-01 1.45194745e+00
-4.68463123e-01 3.59227926e-01 3.02970171e-01 -7.16838121e-01
1.29615641e+00 3.45098436e-01 6.79491162e-01 -7.69173205e-01
7.25062668e-01 3.56751978e-01 -1.20650195e-01 -6.69803977e-01
5.79231739e-01 -3.48944776e-02 -1.91478893e-01 -4.95648086e-02
5.48262119e-01 3.51980180e-01 -3.69347930e-02 -2.88526356e-01
2.59053230e-01 4.28293198e-02 -1.56269684e-01 1.39295861e-01
1.00263727e+00 -1.03856480e+00 1.89866155e-01 3.87150019e-01
-4.62052912e-01 5.27134717e-01 5.09201109e-01 -1.77960768e-01
-7.58247256e-01 -1.28457236e+00 -5.47354400e-01 1.47301602e+00
-3.92677963e-01 1.62703600e-02 -8.85630846e-01 -4.74089652e-01
2.38460943e-01 1.06681935e-01 -8.84030998e-01 -2.52277315e-01
-1.60903752e-01 -9.86130774e-01 1.12249327e+00 7.09365427e-01
7.88287640e-01 -6.99782014e-01 -5.46944700e-02 -2.54114002e-01
-1.90884024e-01 -9.73826587e-01 -2.77438313e-01 -4.87424135e-02
-5.67387223e-01 -1.01995182e+00 -1.46176100e+00 -9.85446095e-01
6.73940539e-01 -9.14527118e-01 1.01324391e+00 -3.44580352e-01
-2.98261434e-01 6.43543661e-01 9.24399048e-02 -5.05609035e-01
-1.12857066e-01 3.76219511e-01 4.03353691e-01 4.34410661e-01
7.54543543e-01 -4.58856165e-01 -8.37690234e-01 2.37940118e-01
-5.04047275e-01 -6.32217050e-01 7.32045472e-01 1.05075240e+00
-1.05880694e-02 -4.07836735e-01 9.78743970e-01 -2.79037446e-01
7.24517524e-01 -1.73471391e-01 -2.94267386e-01 4.04749095e-01
-6.50278151e-01 -1.73683196e-01 -1.15391217e-01 -4.25024837e-01
-1.35722125e+00 4.68102433e-02 -6.46976829e-01 -1.36020958e-01
-3.04220200e-01 5.58342934e-01 -3.61432642e-01 -3.76641721e-01
4.14556950e-01 1.81810632e-01 5.60050726e-01 -5.22141814e-01
1.11616947e-01 1.29561794e+00 7.49919772e-01 -9.33623850e-01
3.63042831e-01 1.37174278e-01 2.13459149e-01 -7.04668760e-01
-2.80710429e-01 -1.94489032e-01 -6.38977587e-01 -3.45752180e-01
7.71259785e-01 -9.96337175e-01 -1.55410278e+00 1.30429685e+00
-6.91591501e-01 3.83295417e-01 5.05397320e-01 3.72956872e-01
-4.51306075e-01 3.66828740e-01 -7.01825857e-01 -1.30627203e+00
-8.28219533e-01 -9.89907742e-01 1.45469582e+00 6.09860539e-01
-7.12086409e-02 -9.89581466e-01 -3.05741906e-01 7.16198087e-01
5.94169438e-01 3.26476723e-01 6.89961374e-01 -6.17963612e-01
2.05270663e-01 -5.32668591e-01 -3.94945472e-01 3.57097089e-01
-2.60198396e-03 1.13923274e-01 -1.54483938e+00 -3.44232023e-01
-6.33875549e-01 -8.29601765e-01 9.63215888e-01 4.99858797e-01
1.52416825e+00 3.62594947e-02 -1.04985423e-02 6.50156915e-01
1.01960981e+00 1.95900515e-01 7.30453849e-01 -5.99119291e-02
4.40139949e-01 8.87048125e-01 2.92268634e-01 5.86966038e-01
5.88878453e-01 4.94332820e-01 2.98904359e-01 -4.23747152e-01
1.42101184e-01 2.30188277e-02 1.94266617e-01 2.22346455e-01
-4.89052445e-01 4.40850370e-02 -8.69433820e-01 5.15422642e-01
-1.49010694e+00 -7.50959814e-01 4.09999490e-01 2.09587693e+00
9.41654861e-01 -4.27119136e-01 5.01045763e-01 3.22939098e-01
6.69755161e-01 -1.94149703e-01 -3.61891448e-01 -5.24694443e-01
-2.70466417e-01 4.65655446e-01 2.52494752e-01 -1.23265296e-01
-1.28793383e+00 5.71728468e-01 5.69177437e+00 8.36210966e-01
-1.28705370e+00 -1.96994916e-01 1.10511017e+00 -6.28652275e-02
2.73768961e-01 -1.04623890e+00 -9.57516730e-01 4.51344311e-01
1.11932051e+00 8.34769011e-02 1.89711973e-01 6.80639863e-01
-3.11170161e-01 -6.97442563e-03 -1.11257112e+00 1.48276556e+00
2.01599881e-01 -4.76481825e-01 -3.58820647e-01 2.17910513e-01
4.42911327e-01 -5.22122920e-01 9.68814254e-01 3.62207025e-01
-3.44830424e-01 -1.39747488e+00 2.08015651e-01 7.85027206e-01
1.22695434e+00 -9.88588035e-01 1.12882209e+00 -3.59424114e-01
-1.06071758e+00 -2.45964587e-01 2.05943175e-02 2.81254232e-01
-2.08925068e-01 1.63272321e-01 -4.67655003e-01 6.78108454e-01
8.33751380e-01 3.26887399e-01 -9.07110274e-01 1.06641567e+00
2.02956170e-01 3.36550802e-01 -2.39667416e-01 6.49457872e-02
-2.27733940e-01 1.68443263e-01 -2.33935088e-01 9.24325347e-01
7.12500453e-01 -2.93051153e-01 -3.30316156e-01 5.44718325e-01
-2.48525634e-01 1.55993119e-01 -4.17313725e-01 -1.04726821e-01
3.92920524e-01 1.42174530e+00 4.94370349e-02 -7.54276440e-02
-2.55634040e-01 5.06704092e-01 -1.18436970e-01 3.45438182e-01
-7.21014678e-01 -3.71223271e-01 7.51387298e-01 -2.01142922e-01
-5.84642179e-02 1.19236961e-01 -5.83132744e-01 -8.63243878e-01
-2.20390424e-01 -7.52936482e-01 6.52441919e-01 -7.14429259e-01
-1.49946880e+00 5.41078687e-01 -2.11990569e-02 -8.79385889e-01
-4.59135741e-01 -9.24008727e-01 -3.33262742e-01 1.18924379e+00
-1.36819100e+00 -1.98748720e+00 -4.44812864e-01 7.67186761e-01
-1.66466117e-01 -8.70763540e-01 9.56672609e-01 6.62446976e-01
-4.42038417e-01 1.57952821e+00 -9.39781368e-02 3.25857460e-01
1.11197817e+00 -1.15404212e+00 -3.29604805e-01 1.98460221e-01
-5.50726831e-01 7.36444235e-01 2.82359332e-01 -2.80621052e-01
-1.23526263e+00 -5.77365696e-01 9.65960920e-01 -2.95145214e-01
3.56579781e-01 -8.14189985e-02 -4.23104852e-01 1.75496191e-01
4.21659172e-01 -4.06872034e-01 1.09370172e+00 4.65384632e-01
-3.54882360e-01 -3.61387521e-01 -1.81249225e+00 2.99058110e-01
6.63144052e-01 -6.90920174e-01 -2.52299458e-01 -3.85271609e-01
-4.04088944e-02 -3.52134526e-01 -1.33804870e+00 9.89897430e-01
1.27049112e+00 -7.01043904e-01 1.22592044e+00 -4.44417447e-01
7.35249281e-01 2.55007029e-01 -2.84919858e-01 -1.08472347e+00
6.06040023e-02 -5.84514327e-02 -3.44692051e-01 1.71514153e+00
4.38672483e-01 -5.89486361e-01 9.28942561e-01 9.60942090e-01
2.44485646e-01 -8.60778570e-01 -8.64012182e-01 -4.00907904e-01
3.51789862e-01 -5.83636500e-02 7.46232986e-01 5.58209896e-01
-2.59169519e-01 -7.79707059e-02 -5.71922362e-01 8.82098973e-02
1.18972695e+00 -3.27819765e-01 5.17430663e-01 -1.37718928e+00
2.75702924e-01 -6.53548539e-01 -6.37824774e-01 -6.83488324e-02
4.84416842e-01 -6.14470720e-01 -4.38453645e-01 -9.86670911e-01
4.34869081e-01 9.07390341e-02 -7.15200424e-01 6.63782716e-01
-1.46141022e-01 8.20674598e-01 -7.22635612e-02 -4.48277920e-01
-1.83247060e-01 7.10041523e-01 1.00717723e+00 -4.07300860e-01
8.28808770e-02 3.68331492e-01 -8.95002663e-01 4.80752856e-01
9.06786501e-01 2.60638207e-01 -1.56803876e-01 -1.54859036e-01
9.59912464e-02 1.31840289e-01 4.28669900e-01 -7.92754114e-01
2.79465675e-01 3.76679420e-01 1.04623842e+00 -8.48016262e-01
8.90377760e-01 -5.30336678e-01 -3.78480822e-01 2.37445161e-01
-3.42214197e-01 -1.65288672e-01 3.15138578e-01 -1.21960580e-01
-4.11218941e-01 7.77660757e-02 8.98926437e-01 3.32619041e-01
-6.06316686e-01 3.85069996e-01 6.57878257e-03 -5.67958474e-01
6.37996912e-01 -3.25702876e-01 -5.24851501e-01 -4.98274863e-01
-1.24111235e+00 2.43341953e-01 1.50458694e-01 4.89702851e-01
8.31560075e-01 -1.48919618e+00 -9.17725325e-01 5.45871913e-01
3.21828157e-01 -8.33167970e-01 4.24198061e-01 9.56776619e-01
5.16563132e-02 2.27656335e-01 -7.62312412e-01 -7.23213732e-01
-2.07624483e+00 8.08326229e-02 4.97904330e-01 2.44324580e-01
5.52616715e-01 1.14040864e+00 -2.09164787e-02 -8.06889355e-01
9.50793922e-01 1.96241334e-01 -6.43495142e-01 5.69720268e-01
4.80223119e-01 2.82317013e-01 9.86079499e-02 -7.42495716e-01
-5.34888387e-01 8.37251246e-01 -8.51820111e-02 -1.42223567e-01
1.18846166e+00 -1.44498363e-01 -2.03272060e-01 -6.55217543e-02
1.23879719e+00 -1.83511451e-01 -6.49339795e-01 -9.89133343e-02
-2.15952799e-01 -4.49347973e-01 -3.54367821e-03 -1.32453346e+00
-1.22523665e+00 1.04659593e+00 1.47798240e+00 -1.11088797e-01
1.55306935e+00 1.15583256e-01 7.12744236e-01 2.27623463e-01
1.96087644e-01 -1.16592491e+00 2.94897612e-02 2.51951426e-01
6.47211790e-01 -1.66267240e+00 -2.44222015e-01 -1.51997820e-01
-5.12845516e-01 1.14708996e+00 7.70058453e-01 6.89718127e-01
6.36892498e-01 1.16320789e-01 2.16387063e-01 -8.17933679e-03
-3.72449994e-01 -3.88156414e-01 8.90058994e-01 8.82406235e-01
8.13440025e-01 -4.04651044e-03 -4.91293967e-01 1.08311927e+00
-4.12100136e-01 2.26161256e-01 -1.15753695e-01 5.85440278e-01
-1.41809702e-01 -1.31616592e+00 -6.44047379e-01 5.27809799e-01
-9.44467545e-01 6.26962781e-02 -5.47238469e-01 5.76539636e-01
2.43107975e-01 9.24077094e-01 -1.72429696e-01 -6.50490046e-01
-2.29974855e-02 5.25851667e-01 9.35971022e-01 1.72411472e-01
-5.14841974e-01 -1.40617058e-01 2.40636989e-01 -2.27550328e-01
-4.80334014e-01 -6.70507431e-01 -6.58107877e-01 -4.90588576e-01
-2.32084896e-02 -1.91781729e-01 1.15167856e+00 7.25388765e-01
3.72493975e-02 2.81521648e-01 8.54079306e-01 -7.47095525e-01
-6.14049435e-01 -1.25032735e+00 -7.47149050e-01 3.67013037e-01
2.95589328e-01 -7.68403351e-01 -1.54890269e-01 6.32642806e-02] | [13.560420036315918, 0.950089693069458] |
8311bc2b-e01d-45e5-ad3c-1e625cf57bcb | r-mbo-a-multi-surrogate-approach-for | 2204.13166 | null | https://arxiv.org/abs/2204.13166v1 | https://arxiv.org/pdf/2204.13166v1.pdf | R-MBO: A Multi-surrogate Approach for Preference Incorporation in Multi-objective Bayesian Optimisation | Many real-world multi-objective optimisation problems rely on computationally expensive function evaluations. Multi-objective Bayesian optimisation (BO) can be used to alleviate the computation time to find an approximated set of Pareto optimal solutions. In many real-world problems, a decision-maker has some preferences on the objective functions. One approach to incorporate the preferences in multi-objective BO is to use a scalarising function and build a single surrogate model (mono-surrogate approach) on it. This approach has two major limitations. Firstly, the fitness landscape of the scalarising function and the objective functions may not be similar. Secondly, the approach assumes that the scalarising function distribution is Gaussian, and thus a closed-form expression of an acquisition function e.g., expected improvement can be used. We overcome these limitations by building independent surrogate models (multi-surrogate approach) on each objective function and show that the distribution of the scalarising function is not Gaussian. We approximate the distribution using Generalised value distribution. We present an a-priori multi-surrogate approach to incorporate the desirable objective function values (or reference point) as the preferences of a decision-maker in multi-objective BO. The results and comparison with the existing mono-surrogate approach on benchmark and real-world optimisation problems show the potential of the proposed approach. | ['Tinkle Chugh'] | 2022-04-27 | null | null | null | null | ['bayesian-optimisation'] | ['methodology'] | [ 1.63177744e-01 -2.23625824e-01 4.90429401e-02 -3.97582054e-01
-9.77602303e-01 -5.96189857e-01 2.95403272e-01 3.42133552e-01
-5.37503421e-01 1.15083539e+00 -6.52504265e-02 -2.28005182e-02
-1.06093299e+00 -8.02786946e-01 -4.15274352e-01 -1.11015689e+00
6.29805550e-02 8.11935723e-01 2.64508605e-01 -1.24248870e-01
5.61750889e-01 2.66823500e-01 -1.56105483e+00 1.10323038e-02
1.13956678e+00 1.01613116e+00 4.06134337e-01 9.12148237e-01
1.25397712e-01 -9.38710272e-02 -1.08605158e+00 -3.59023511e-01
3.64149660e-01 -3.22956532e-01 -5.55906832e-01 -3.76835227e-01
-6.00814044e-01 2.19195440e-01 6.54012501e-01 1.10447633e+00
7.37347186e-01 5.73018730e-01 1.07823944e+00 -1.51576090e+00
-3.02016884e-01 8.13752562e-02 -4.86841410e-01 -1.74636424e-01
2.73114860e-01 1.39269203e-01 8.90495002e-01 -2.94523686e-01
-5.78294657e-02 1.50401545e+00 7.01678813e-01 1.07198782e-01
-1.34232414e+00 -1.37838915e-01 -3.97830904e-02 4.92711067e-02
-1.63486028e+00 -7.88956359e-02 6.00198150e-01 -1.85011715e-01
6.08103752e-01 7.83703089e-01 6.05598927e-01 4.22884583e-01
7.53268063e-01 5.16343892e-01 1.44879496e+00 -4.95691270e-01
6.71193838e-01 2.17662737e-01 -3.62157345e-01 5.08318692e-02
2.67940909e-01 5.07921219e-01 -7.59273395e-02 -5.56144714e-01
2.72022545e-01 -3.06696236e-01 -3.48277271e-01 -4.46507514e-01
-8.31395447e-01 1.05693281e+00 2.37080678e-01 5.85751906e-02
-6.72414601e-01 3.12675864e-01 2.24057019e-01 7.75620062e-03
4.18473572e-01 6.55782938e-01 -4.32559282e-01 -2.39425778e-01
-1.15871716e+00 4.84103918e-01 7.88653016e-01 4.61814672e-01
5.28751969e-01 -1.20250292e-01 -4.15186137e-01 8.33658099e-01
1.09958065e+00 4.93539095e-01 4.11401778e-01 -8.51001382e-01
2.42289618e-01 3.38604510e-01 6.72260940e-01 -8.80737841e-01
-2.90407419e-01 -6.43389940e-01 -4.70076352e-01 8.38493526e-01
6.41182780e-01 -2.02272609e-01 -8.36801350e-01 1.32806516e+00
4.07517016e-01 -4.87006605e-01 1.39315769e-01 9.15821731e-01
4.24135417e-01 9.49030876e-01 -1.11764885e-01 -4.84447300e-01
9.74565744e-01 -6.21322632e-01 -7.56673217e-01 1.04484465e-02
1.48025796e-01 -9.44738925e-01 7.93419778e-01 7.55738676e-01
-1.10530007e+00 -3.17907989e-01 -1.10841346e+00 7.28979230e-01
-2.26373911e-01 -1.28178820e-01 3.01162571e-01 1.10099745e+00
-7.27422953e-01 7.49790788e-01 -5.21755576e-01 1.37787968e-01
-3.48689705e-02 7.55684137e-01 2.99627602e-01 6.85496330e-02
-1.09874761e+00 1.31289661e+00 9.02857602e-01 3.59000713e-01
-8.09096217e-01 -6.05966985e-01 -5.61921299e-01 1.06028683e-01
2.90871680e-01 -7.02572048e-01 1.02373445e+00 -9.60440040e-01
-1.98309076e+00 6.79612309e-02 -7.11808121e-03 1.35837734e-01
5.37086189e-01 -1.06985960e-02 -3.72933984e-01 -3.41383964e-01
-1.47660553e-01 1.61709756e-01 6.81531906e-01 -1.53503859e+00
-7.14139640e-01 -1.20595619e-01 1.21611290e-01 4.35913712e-01
2.18986854e-01 3.31111670e-01 3.07908300e-02 -4.42222804e-01
-2.48736486e-01 -4.98253644e-01 -4.42437172e-01 -3.53354305e-01
-1.14384927e-01 -2.53079623e-01 2.71987468e-01 -5.43714881e-01
1.63014710e+00 -1.78327060e+00 2.84786850e-01 6.24516129e-01
-3.22252423e-01 1.04842268e-01 5.24752811e-02 4.16080743e-01
2.29079738e-01 1.44184589e-01 -6.13455713e-01 8.69180635e-02
3.68294895e-01 3.86946261e-01 2.45992735e-01 7.52021730e-01
1.87448651e-01 5.83095372e-01 -1.11708772e+00 -3.89866143e-01
2.62631059e-01 4.28153336e-01 -3.99573773e-01 1.17516495e-01
-2.22001255e-01 2.50855416e-01 -4.28858519e-01 5.29774487e-01
7.57882714e-01 2.95221955e-01 -3.69985104e-02 -1.12598114e-01
-2.43528038e-01 -4.03229445e-02 -1.86284292e+00 1.14120138e+00
-6.05290711e-01 2.42203161e-01 2.19042212e-01 -1.27082551e+00
1.12763035e+00 3.60437721e-01 4.28143948e-01 -2.16172740e-01
3.70034516e-01 5.45527458e-01 3.49085599e-01 -5.03325090e-02
2.73835748e-01 -6.37722552e-01 -1.30035862e-01 2.75590897e-01
-2.78666858e-02 -5.93637764e-01 2.81442702e-01 -7.95854151e-01
5.37811875e-01 5.37203610e-01 6.14698350e-01 -6.95216417e-01
7.46686935e-01 -3.37372236e-02 7.44806647e-01 5.69282770e-01
3.68047915e-02 6.47675097e-01 3.06669116e-01 -1.67268649e-01
-7.48118699e-01 -9.04753149e-01 -3.38183045e-01 8.61612678e-01
1.26283422e-01 -7.82959387e-02 -2.57286638e-01 -3.03744763e-01
6.20065257e-02 1.22133636e+00 -4.03187424e-01 -1.72501892e-01
-3.80600363e-01 -1.55166090e+00 1.64870039e-01 2.30477810e-01
8.61141160e-02 -4.45821136e-01 -9.34623122e-01 6.52127683e-01
-3.38436626e-02 -9.97182429e-02 -1.34936631e-01 4.58833665e-01
-7.81460285e-01 -6.64926648e-01 -1.07249463e+00 -1.31808072e-01
5.38751185e-01 -4.69655991e-01 1.13738537e+00 -2.19956785e-01
2.03659292e-02 1.58210322e-01 -1.06441915e-01 -7.87461102e-01
-4.05250013e-01 -5.72982371e-01 -7.95287862e-02 1.19149517e-02
1.40491262e-01 -2.20178321e-01 -4.37102526e-01 8.08540940e-01
-7.71650553e-01 -5.95899284e-01 4.95540023e-01 1.00291002e+00
7.76359439e-01 8.15568089e-01 6.36822045e-01 1.65909883e-02
1.19594836e+00 -5.33953846e-01 -1.05847704e+00 5.93892276e-01
-7.75987804e-01 3.30727160e-01 5.11915207e-01 -5.55665135e-01
-1.12311733e+00 -2.40043849e-01 2.66716052e-05 -1.05829269e-01
1.76932588e-02 7.92029679e-01 -4.51865494e-01 -3.13492082e-02
4.50806916e-01 -2.20630497e-01 -2.40051690e-02 -5.88439286e-01
1.09924868e-01 8.41457307e-01 3.83283436e-01 -1.03283465e+00
4.36750382e-01 -2.98385229e-02 5.22728682e-01 -3.04324597e-01
-4.16229784e-01 -5.89645207e-01 -3.20794731e-01 -4.71899629e-01
5.34137905e-01 -2.04552010e-01 -1.04306042e+00 1.13789558e-01
-1.06446505e+00 -2.22646389e-02 -4.29575503e-01 7.90900409e-01
-8.21151853e-01 2.19820529e-01 5.55997491e-01 -1.61624503e+00
-1.58416957e-01 -1.48858118e+00 7.42551208e-01 4.49605048e-01
-4.16840285e-01 -1.31219971e+00 2.72550344e-01 7.03519657e-02
4.81820673e-01 6.80669427e-01 7.54374623e-01 -5.16923070e-01
6.28484413e-03 -3.75172466e-01 1.32917598e-01 3.32811385e-01
2.61874106e-02 1.70269296e-01 -5.40321350e-01 -4.82350796e-01
4.27989930e-01 -4.40671891e-02 3.42606127e-01 1.00409544e+00
6.83223248e-01 -2.98653781e-01 -3.07326525e-01 4.50286627e-01
1.67774057e+00 6.27046287e-01 5.67835450e-01 7.34005392e-01
-1.17722526e-01 8.67088199e-01 1.12220240e+00 5.80671906e-01
-2.51504015e-02 9.10227418e-01 5.95053315e-01 1.68261945e-01
3.36122215e-01 3.11656594e-01 5.08195698e-01 3.33759785e-01
-4.57710057e-01 -7.24327743e-01 -1.02552664e+00 5.73091805e-01
-2.06398010e+00 -6.91150844e-01 -2.80458063e-01 2.73215342e+00
7.20641434e-01 1.67835355e-01 3.21158528e-01 4.05041963e-01
9.75620985e-01 -3.67144674e-01 -4.43985224e-01 -1.02803504e+00
-5.79435490e-02 1.32098645e-01 7.11722136e-01 7.55537391e-01
-8.64328265e-01 -9.92998332e-02 6.92006207e+00 1.19651854e+00
-7.69509792e-01 5.47183044e-02 4.20082927e-01 -2.14362025e-01
-3.55946302e-01 1.20550469e-01 -7.08146036e-01 8.42459321e-01
1.15471804e+00 -6.31554306e-01 5.40790141e-01 4.36623126e-01
3.96426052e-01 -6.43871903e-01 -8.65666509e-01 5.94098210e-01
-3.29577208e-01 -7.37932444e-01 -4.10822839e-01 3.15464795e-01
7.65782535e-01 -4.54277754e-01 -9.51937884e-02 4.53063473e-03
6.47878408e-01 -1.37235808e+00 7.65794992e-01 8.14142525e-01
5.01474142e-01 -1.29955006e+00 1.47993946e+00 4.17651594e-01
-1.04443109e+00 -2.53564268e-01 -3.50990266e-01 -4.13492434e-02
5.28070509e-01 7.09430814e-01 -5.51266968e-01 1.18541169e+00
7.45042086e-01 -1.64620951e-01 -4.96680662e-02 1.72169602e+00
-1.47058040e-01 4.00844455e-01 -9.11415756e-01 -4.79255289e-01
4.16595519e-01 -5.44653952e-01 9.53797877e-01 8.81118119e-01
6.73638344e-01 -3.34336013e-01 -8.04128423e-02 1.07147336e+00
8.73471320e-01 1.58997834e-01 5.66281080e-02 -7.64369443e-02
2.58805037e-01 8.38542759e-01 -6.03569508e-01 1.31175861e-01
-1.44180864e-01 4.08890784e-01 -4.19616878e-01 3.14105064e-01
-8.20094585e-01 -6.79267883e-01 4.29630935e-01 -2.68220782e-01
2.44057074e-01 2.60166284e-02 -5.79716682e-01 -2.81930089e-01
2.95145754e-02 -7.51875877e-01 4.57414359e-01 -6.53078377e-01
-1.14896250e+00 2.99494594e-01 8.00390542e-01 -1.21149874e+00
-4.76174533e-01 -6.81098878e-01 -8.26155603e-01 1.52221584e+00
-1.24484468e+00 -7.56766200e-01 2.15632468e-01 7.13285580e-02
1.80324361e-01 -1.87775895e-01 7.26401567e-01 2.69683963e-03
-2.93714792e-01 3.20515722e-01 8.54536474e-01 -6.88357949e-01
3.76306981e-01 -1.42258489e+00 -4.75923598e-01 5.74287295e-01
-6.04950309e-01 5.13578475e-01 1.33621812e+00 -6.45283520e-01
-9.79766130e-01 -4.88817096e-01 4.85605866e-01 -3.09204966e-01
5.33558786e-01 1.26928955e-01 -5.59750736e-01 -7.16086999e-02
1.96102947e-01 -3.74184310e-01 8.61702561e-01 -2.89608017e-02
7.81558812e-01 -5.35356998e-02 -1.60811234e+00 1.74534142e-01
7.35284463e-02 2.60037005e-01 -8.71810675e-01 2.40399502e-02
-4.19535637e-02 -8.68066102e-02 -1.31528091e+00 7.65943050e-01
6.95254445e-01 -7.51951277e-01 1.09135330e+00 -3.69248599e-01
4.87732291e-02 -8.19497228e-01 -3.41797620e-01 -1.88730657e+00
-2.80853570e-01 -7.95919895e-01 -1.64358988e-01 1.30475330e+00
2.31409490e-01 -9.06791925e-01 2.48924524e-01 7.83746660e-01
3.16114612e-02 -1.22077763e+00 -1.42864585e+00 -1.22081888e+00
2.69586325e-01 -3.09020698e-01 9.58459198e-01 2.34363809e-01
-3.48042369e-01 -5.32700308e-02 -1.58983842e-01 1.88782722e-01
8.58677268e-01 -1.22074760e-03 3.98801655e-01 -1.36975312e+00
-5.60644567e-01 -7.38328040e-01 -2.89787203e-01 -7.10279346e-01
-2.92116106e-01 -5.01225352e-01 2.25051358e-01 -1.71950352e+00
-2.63240617e-02 -4.83618379e-01 -5.74475706e-01 4.90157977e-02
-2.44615853e-01 -2.11346418e-01 3.51621062e-02 -9.03141871e-02
-1.49128780e-01 9.01059806e-01 1.19505441e+00 -2.86552906e-02
-3.65978062e-01 4.81836766e-01 -6.06714606e-01 8.63926768e-01
7.75231421e-01 -9.36297417e-01 -3.76120597e-01 9.72992927e-02
5.11885405e-01 -1.35533391e-02 1.72695726e-01 -6.84379637e-01
2.14708969e-03 -7.37458885e-01 2.85944432e-01 -6.83921456e-01
3.52860779e-01 -1.05315232e+00 7.34161377e-01 3.93968344e-01
1.17183976e-01 3.11878473e-02 3.23652416e-01 4.70647186e-01
-2.67544299e-01 -1.17175567e+00 7.92263269e-01 -9.92159545e-02
-1.04418255e-01 -5.30723631e-01 -3.29973161e-01 -2.52208978e-01
1.09454727e+00 -6.76145732e-01 8.28398857e-03 -3.12740475e-01
-5.90331912e-01 6.34860933e-01 4.31407273e-01 -3.00680231e-02
4.38560694e-01 -1.42023635e+00 -9.45556462e-01 -3.85468006e-01
-2.42474824e-01 8.97910967e-02 -2.85706699e-01 9.13502395e-01
-5.19965112e-01 4.64892089e-01 -2.78178871e-01 -5.62542140e-01
-1.03041387e+00 3.11900645e-01 5.92317343e-01 -5.55763960e-01
2.89366990e-01 7.82983243e-01 -7.98833445e-02 -4.90520358e-01
-6.89430386e-02 -2.98764445e-02 -1.92395866e-01 2.28691280e-01
2.67589092e-01 1.03228462e+00 -1.07084215e-01 -6.18766427e-01
-6.12518072e-01 7.62243807e-01 7.47300029e-01 -5.24245024e-01
1.60050011e+00 -2.61384964e-01 -2.53799796e-01 4.80650634e-01
8.98568630e-01 -1.06035300e-01 -1.07735693e+00 3.42534631e-01
6.82440493e-03 -7.11398959e-01 5.71677089e-01 -1.17740321e+00
-5.62659800e-01 5.99501789e-01 6.30476892e-01 2.47942969e-01
1.39153910e+00 -4.86931622e-01 7.72828385e-02 7.44261220e-02
2.79770285e-01 -1.41375077e+00 -5.22869051e-01 9.72677693e-02
1.32693100e+00 -1.00245249e+00 3.96255374e-01 -9.02276859e-02
-5.06704092e-01 1.29810393e+00 1.22387528e-01 6.13876767e-02
6.23318970e-01 9.32934582e-02 -1.51814133e-01 3.01054958e-02
-3.87861580e-01 -2.71389857e-02 9.50217903e-01 6.54724300e-01
3.06474268e-01 1.83881804e-01 -8.75169158e-01 7.19084859e-01
-1.10598020e-01 -1.18908271e-01 1.64011389e-01 9.94180560e-01
-1.99630409e-01 -1.31866539e+00 -1.20764625e+00 2.37601131e-01
-5.65277874e-01 7.78434277e-02 -6.35969788e-02 6.12955153e-01
1.23235896e-01 1.33003592e+00 -1.71923757e-01 -5.24025559e-02
3.32217693e-01 2.85479743e-02 4.98470873e-01 -3.05337697e-01
-6.80631578e-01 3.75590682e-01 2.46554881e-01 -2.57387042e-01
-5.55481076e-01 -6.19677782e-01 -8.87181401e-01 -2.17507705e-01
-8.93355608e-01 5.98542511e-01 1.01867211e+00 9.14700687e-01
-1.20916925e-01 7.12846994e-01 5.98834038e-01 -1.07960975e+00
-8.52919996e-01 -8.81164789e-01 -4.54641700e-01 -1.72218636e-01
1.43845364e-01 -1.11851633e+00 -6.67166233e-01 -5.60189247e-01] | [6.087121486663818, 3.6086814403533936] |
1f3b5779-7ae1-4a16-82b9-5f54ffe8ab01 | motrv3-release-fetch-supervision-for-end-to | 2305.14298 | null | https://arxiv.org/abs/2305.14298v1 | https://arxiv.org/pdf/2305.14298v1.pdf | MOTRv3: Release-Fetch Supervision for End-to-End Multi-Object Tracking | Although end-to-end multi-object trackers like MOTR enjoy the merits of simplicity, they suffer from the conflict between detection and association seriously, resulting in unsatisfactory convergence dynamics. While MOTRv2 partly addresses this problem, it demands an additional detection network for assistance. In this work, we serve as the first to reveal that this conflict arises from the unfair label assignment between detect queries and track queries during training, where these detect queries recognize targets and track queries associate them. Based on this observation, we propose MOTRv3, which balances the label assignment process using the developed release-fetch supervision strategy. In this strategy, labels are first released for detection and gradually fetched back for association. Besides, another two strategies named pseudo label distillation and track group denoising are designed to further improve the supervision for detection and association. Without the assistance of an extra detection network during inference, MOTRv3 achieves impressive performance across diverse benchmarks, e.g., MOT17, DanceTrack. | ['Wenbing Tao', 'Xiangyu Zhang', 'Yuang Zhang', 'Zhuoling Li', 'Tiancai Wang', 'En Yu'] | 2023-05-23 | null | null | null | null | ['multi-object-tracking', 'pseudo-label'] | ['computer-vision', 'miscellaneous'] | [ 1.46111652e-01 -8.64466205e-02 -4.20113206e-01 -2.96001047e-01
-7.49970675e-01 -4.82321471e-01 4.97023195e-01 1.37620606e-02
-5.24552226e-01 6.33028984e-01 -1.33575365e-01 -8.36255401e-02
7.05835521e-02 -3.37649673e-01 -7.31293380e-01 -9.31068480e-01
1.54036745e-01 5.45202136e-01 6.92014456e-01 8.73529762e-02
-6.94666207e-02 1.26686588e-01 -1.55425596e+00 -1.14619285e-01
8.45340133e-01 9.90219057e-01 3.44438314e-01 3.67570758e-01
2.84278467e-02 9.32672858e-01 -5.58646679e-01 -3.40251327e-01
1.78411648e-01 -2.89637089e-01 -2.29663134e-01 -6.41560331e-02
6.89128876e-01 -3.43901604e-01 -3.39141935e-01 1.18474400e+00
8.07942927e-01 1.46374643e-01 3.51133049e-01 -1.48710990e+00
-3.27560186e-01 7.37550855e-01 -1.04206359e+00 3.40972006e-01
1.21149877e-02 3.53575170e-01 1.20427299e+00 -6.94013953e-01
4.40983683e-01 1.06800389e+00 7.46435285e-01 6.35578454e-01
-1.16784263e+00 -1.00708997e+00 5.28214037e-01 5.19376397e-02
-1.32934391e+00 -6.83395922e-01 5.28150797e-01 -3.95475835e-01
3.23144883e-01 4.69387919e-01 4.43523228e-01 1.01455069e+00
-1.01399682e-01 1.07830906e+00 7.82085717e-01 7.44124427e-02
5.02095781e-02 5.92170283e-02 3.03818822e-01 6.87057793e-01
2.49163240e-01 1.96489856e-01 -6.11139834e-01 -1.29642025e-01
6.39583886e-01 2.15964153e-01 -2.85776228e-01 -2.87725180e-01
-1.25679255e+00 6.12349689e-01 5.04518867e-01 9.00058448e-02
-2.90704399e-01 3.83746505e-01 3.96791488e-01 1.54543161e-01
3.53518993e-01 2.91989651e-02 -3.40970099e-01 -1.30099319e-02
-1.04218662e+00 1.88992638e-02 2.95103073e-01 1.13697338e+00
8.83817434e-01 -2.57412754e-02 -6.79723084e-01 6.73502684e-01
6.56024098e-01 6.66644692e-01 2.29486585e-01 -9.38146055e-01
3.78747910e-01 4.49836582e-01 2.43359581e-01 -8.75046849e-01
-4.11335677e-01 -8.84573400e-01 -7.02753782e-01 7.03925043e-02
4.96417284e-01 -1.53656840e-01 -9.62266088e-01 1.96928680e+00
7.69518554e-01 6.25379145e-01 -1.35513082e-01 1.18257821e+00
8.03448558e-01 4.00578588e-01 2.05229238e-01 -2.17097968e-01
1.56920409e+00 -1.32302845e+00 -8.23780835e-01 -2.22719386e-01
8.15095186e-01 -6.06016994e-01 7.74606764e-01 2.80879408e-01
-1.03565562e+00 -6.48898721e-01 -9.24193144e-01 3.00424211e-02
1.82149097e-01 3.98345232e-01 6.18721545e-01 3.75375330e-01
-8.19021940e-01 2.00518072e-01 -1.03302836e+00 -3.45840275e-01
3.74128968e-01 4.54689413e-01 -1.40236880e-04 5.59860878e-02
-8.90875876e-01 4.39302772e-01 1.42125294e-01 3.87486875e-01
-1.05075490e+00 -5.52179873e-01 -5.09969950e-01 -2.44349316e-02
7.72131205e-01 -5.81013083e-01 1.38731515e+00 -6.82309031e-01
-1.20527375e+00 6.99036181e-01 -1.81236878e-01 -4.90542084e-01
7.16341555e-01 -2.72408247e-01 -3.98783654e-01 -1.73487723e-01
3.77410650e-01 6.81324661e-01 8.14489901e-01 -1.21308196e+00
-1.25211549e+00 -3.91368479e-01 3.47948894e-02 1.81587413e-01
-2.83407897e-01 -9.97947231e-02 -9.46804285e-01 -6.17330670e-01
1.38430461e-01 -1.06211138e+00 -6.35965466e-02 6.11289917e-03
-5.02352595e-01 -5.63953340e-01 9.84162450e-01 -8.26035514e-02
1.30456543e+00 -2.34698033e+00 -4.01225388e-02 -6.36094213e-02
5.37598372e-01 2.44022563e-01 -6.20360970e-02 9.96829011e-03
2.56625086e-01 -2.86674261e-01 2.05272183e-01 -8.38004887e-01
4.89480533e-02 3.22880834e-01 -3.18889916e-01 7.02904522e-01
1.01138070e-01 8.38248551e-01 -1.14006937e+00 -7.57278264e-01
7.99721852e-03 2.39468813e-01 -5.86245954e-01 1.58765078e-01
-5.09158969e-01 7.25872993e-01 -7.30977535e-01 7.98823178e-01
6.81435227e-01 -6.36715889e-01 9.98782367e-02 -3.61569971e-01
-2.99528807e-01 2.79800028e-01 -1.36387157e+00 1.63520920e+00
-8.61255229e-02 4.62601393e-01 3.66128802e-01 -7.77812421e-01
7.89954960e-01 2.60651559e-01 6.55830026e-01 -7.20338941e-01
1.48754209e-01 1.09796353e-01 -1.09923996e-01 -2.72369474e-01
7.09131300e-01 1.17134094e-01 4.01339047e-02 3.72668505e-01
-2.70410568e-01 7.66168118e-01 1.82159275e-01 3.38144183e-01
1.15652883e+00 2.47487053e-01 -2.95386285e-01 5.33151589e-02
5.33122361e-01 1.11175127e-01 1.09528840e+00 9.51246679e-01
-4.13387299e-01 4.55752045e-01 6.58991486e-02 -2.65679181e-01
-5.05893648e-01 -1.02242935e+00 -1.20758098e-02 1.53741264e+00
6.50171041e-01 -4.24484342e-01 -3.57918531e-01 -7.96163678e-01
-3.58949304e-02 3.87647212e-01 -4.71033275e-01 -2.23746672e-01
-6.80756092e-01 -8.77784550e-01 6.81735754e-01 4.72769380e-01
3.53056878e-01 -9.65924144e-01 -6.37287974e-01 3.93718451e-01
-3.61261487e-01 -1.01562929e+00 -7.85388589e-01 3.10865939e-01
-6.37870491e-01 -1.00084746e+00 -6.17674291e-01 -6.34173155e-01
5.33318698e-01 7.22842038e-01 1.03345609e+00 3.49231988e-01
1.58256683e-02 1.18083782e-01 -2.83345938e-01 -2.42066175e-01
-1.60380125e-01 2.92648941e-01 7.66048282e-02 1.67265475e-01
3.25204134e-01 -4.03819352e-01 -7.45062590e-01 8.04221213e-01
-6.39172435e-01 3.93543579e-02 6.76476479e-01 7.57149816e-01
8.01972389e-01 -2.49157399e-01 5.44129312e-01 -7.13150859e-01
-8.16019028e-02 -6.07472479e-01 -8.54759216e-01 3.28736484e-01
-5.59510112e-01 -6.20962605e-02 3.90212417e-01 -7.11770833e-01
-8.54286671e-01 1.59038424e-01 -2.88851582e-03 -5.69874644e-01
4.92656119e-02 3.32974680e-02 -9.49367434e-02 1.34132668e-01
5.23288250e-01 1.61026567e-01 -1.55660689e-01 -6.78928435e-01
2.54498869e-01 5.95507205e-01 7.64864028e-01 -3.68409842e-01
8.83025050e-01 6.14528656e-01 -2.79099792e-01 -4.10803139e-01
-1.26056898e+00 -7.68716395e-01 2.27597281e-02 -3.99361432e-01
9.70160067e-01 -1.33195651e+00 -1.10689795e+00 3.59335095e-01
-1.00192869e+00 -2.11500689e-01 -1.28539175e-01 5.54487884e-01
-2.55207539e-01 2.44322583e-01 -5.58013678e-01 -7.86860585e-01
-3.26101631e-01 -1.15073669e+00 1.37180424e+00 4.74093199e-01
1.95356593e-01 -6.14872873e-01 -4.02564816e-02 4.78309780e-01
2.82674015e-01 -1.09599736e-02 1.85618684e-01 -6.43798351e-01
-1.04702580e+00 3.81815657e-02 -4.01449323e-01 -2.69633234e-01
2.06431329e-01 -1.40155137e-01 -1.04909182e+00 -7.01127768e-01
-1.76446497e-01 -3.90495896e-01 9.94029641e-01 2.80558407e-01
7.92047679e-01 -5.43275289e-03 -7.38219559e-01 6.78202331e-01
1.14635360e+00 1.96132399e-02 1.92493752e-01 3.36315274e-01
9.30253565e-01 1.94924340e-01 9.53401625e-01 4.59366530e-01
6.44740582e-01 1.07720697e+00 7.10650742e-01 -1.16709791e-01
-2.56910533e-01 -3.02079588e-01 6.17199838e-01 7.78425336e-01
1.86007589e-01 -4.82697815e-01 -5.28518796e-01 2.95209885e-01
-2.24693203e+00 -7.03285992e-01 -4.89161402e-01 2.12463808e+00
8.58367980e-01 3.26524466e-01 3.26312840e-01 -2.72731900e-01
8.83763909e-01 2.74168372e-01 -7.36280680e-01 4.95280623e-01
-3.71886268e-02 -5.23577571e-01 6.79795980e-01 2.18934596e-01
-1.05183637e+00 1.05796826e+00 5.18861961e+00 1.05386889e+00
-1.10355484e+00 3.54079068e-01 2.38025442e-01 -1.79430798e-01
9.17568207e-02 9.03627276e-02 -1.43189025e+00 6.72196925e-01
7.28941619e-01 3.69320005e-01 2.34138444e-01 6.97085202e-01
1.60282522e-01 -1.32458985e-01 -1.00199461e+00 8.38098407e-01
-2.72255868e-01 -1.04447079e+00 -4.22434419e-01 6.20861053e-02
4.32410359e-01 4.06948328e-01 -4.35296707e-02 4.17723924e-01
6.05853140e-01 -4.77292448e-01 9.50592875e-01 3.57144177e-01
5.06692171e-01 -4.50439602e-01 4.96103764e-01 5.59696972e-01
-1.44963944e+00 -5.07796230e-03 -4.51058328e-01 2.50374109e-01
3.43407750e-01 6.44138515e-01 -4.97996211e-01 7.35058486e-01
6.94085658e-01 8.55403721e-01 -5.26323676e-01 1.17702937e+00
-1.73484445e-01 6.66150987e-01 -4.95578080e-01 2.29105338e-01
2.52965122e-01 -9.09648761e-02 6.93244636e-01 8.60703588e-01
2.11013451e-01 -1.83966339e-01 6.56663775e-01 7.40216434e-01
-2.84723341e-02 -3.46076518e-01 -2.63655335e-01 2.84263730e-01
8.20000589e-01 1.51152432e+00 -7.55976200e-01 -5.02021968e-01
-4.88751203e-01 6.87851548e-01 3.56367677e-01 2.10019395e-01
-1.39124286e+00 1.37309164e-01 6.47335589e-01 5.76007850e-02
5.93696833e-01 8.15904420e-03 1.17099527e-02 -1.01619196e+00
2.63675973e-02 -6.58062518e-01 4.62508559e-01 -5.25305152e-01
-1.13771927e+00 4.28508997e-01 -2.53025442e-01 -1.37181079e+00
1.44055277e-01 -5.19375615e-02 -6.30720496e-01 3.86230618e-01
-1.76648748e+00 -1.04391634e+00 -4.45788920e-01 5.17330348e-01
3.66174579e-01 2.24591389e-01 1.52848229e-01 9.54001486e-01
-1.01653659e+00 9.20573235e-01 3.82668115e-02 4.63308804e-02
1.00029063e+00 -1.16268039e+00 1.93999961e-01 6.42296731e-01
2.98629075e-01 4.20871347e-01 7.05151558e-01 -6.83656931e-01
-1.34236538e+00 -1.43224061e+00 5.72726369e-01 -4.08821791e-01
6.21634245e-01 -4.01648253e-01 -9.14410233e-01 7.14839995e-01
-9.25450027e-03 3.32138807e-01 4.43053752e-01 5.87512106e-02
-1.40254855e-01 -2.59858519e-01 -7.39826143e-01 4.54580784e-01
1.23822653e+00 -4.44260128e-02 -3.09102952e-01 4.68154907e-01
8.07478189e-01 -5.05805612e-01 -7.17359185e-01 3.99170160e-01
3.31887782e-01 -8.63832414e-01 8.90593350e-01 -8.48063976e-02
-2.34164596e-01 -8.80345047e-01 1.92906912e-02 -8.19414020e-01
-3.42264205e-01 -8.40519786e-01 -3.86788726e-01 1.63803637e+00
2.57436186e-01 -4.94466186e-01 9.90827203e-01 -4.73050028e-02
-3.29619825e-01 -5.74992776e-01 -9.20014083e-01 -9.66832697e-01
-5.82352698e-01 -1.03397183e-01 4.95874375e-01 8.82480621e-01
-6.56299949e-01 6.66317225e-01 -7.43612409e-01 3.71356279e-01
9.35617805e-01 2.88936526e-01 1.07951975e+00 -1.40109861e+00
-5.09719491e-01 -1.50584832e-01 4.28709835e-02 -1.73812962e+00
-2.11860880e-01 -6.96211994e-01 4.48093891e-01 -1.17191124e+00
1.79459646e-01 -1.03064322e+00 -4.91702557e-01 6.73161805e-01
-3.93422008e-01 4.25573677e-01 3.82661641e-01 6.48598492e-01
-1.35845900e+00 6.60674274e-01 1.30497575e+00 -6.67769983e-02
-2.51834601e-01 2.35518917e-01 -6.88603580e-01 5.40886760e-01
5.33216715e-01 -9.59261000e-01 -3.02187741e-01 -5.92146158e-01
2.26500347e-01 1.01652481e-01 3.84570152e-01 -1.05611348e+00
6.06616497e-01 9.64448974e-02 2.73045190e-02 -1.01091301e+00
2.94331461e-01 -8.52481902e-01 2.91132331e-01 4.66194630e-01
-1.54911593e-01 1.33988902e-01 -9.38853528e-03 9.70589042e-01
-2.12201830e-02 1.98161211e-02 7.36189961e-01 2.34995082e-01
-6.23976827e-01 4.89079505e-01 -1.99058533e-01 1.92319989e-01
9.37034547e-01 -2.08646357e-01 -4.88060385e-01 -5.94161591e-03
-6.61267757e-01 1.03705359e+00 4.59343404e-01 3.82633388e-01
1.09127071e-02 -1.43661165e+00 -5.81963956e-01 -5.45148142e-02
3.19290012e-02 4.91057128e-01 2.16361180e-01 1.46419990e+00
1.12978399e-01 1.51087016e-01 2.72765696e-01 -9.91321385e-01
-1.31228220e+00 5.03524482e-01 1.05193540e-01 -5.85079610e-01
-8.03586721e-01 8.56762588e-01 4.21195954e-01 -3.04199159e-01
6.98047280e-01 -2.42867842e-01 -7.80603215e-02 1.07910082e-01
4.73142892e-01 3.37790310e-01 -1.92533270e-01 -4.82860833e-01
-4.16895747e-01 5.20534217e-01 -3.20072860e-01 1.58292338e-01
9.85097170e-01 -4.30066198e-01 2.24834144e-01 3.46811712e-01
6.11225605e-01 5.95301278e-02 -1.60054374e+00 -4.55249965e-01
1.59708261e-01 -3.10839415e-01 2.21653432e-02 -6.21032476e-01
-1.43347073e+00 5.46194792e-01 6.61679685e-01 2.49722466e-01
1.05932438e+00 1.23983778e-01 1.09413338e+00 3.30328554e-01
2.53998190e-01 -9.37575221e-01 1.68348983e-01 4.31845009e-01
1.37230888e-01 -1.33354890e+00 5.73393889e-03 -3.43726575e-01
-3.65473479e-01 4.28396344e-01 9.40455437e-01 1.17370047e-01
2.15472281e-01 2.58560449e-01 1.12103902e-01 -3.17860961e-01
-9.20169175e-01 -3.66636455e-01 -1.49953952e-02 4.08186734e-01
1.25277519e-01 -2.19636247e-01 -2.22429484e-01 3.78121108e-01
2.28833959e-01 -9.22979712e-02 5.62504679e-02 1.01375759e+00
-6.08643651e-01 -9.90929425e-01 -5.64841986e-01 3.20633978e-01
-5.22258103e-01 1.14043064e-01 -1.08295143e-01 7.07036138e-01
3.77409756e-01 9.51317787e-01 6.70481846e-02 -3.98539007e-01
2.52263248e-01 -3.75625849e-01 1.16676599e-01 -5.00059664e-01
-7.90820777e-01 4.69548911e-01 1.18045090e-02 -4.81204957e-01
-6.41444206e-01 -6.33135915e-01 -1.41873682e+00 -7.57934377e-02
-9.69368458e-01 3.85024041e-01 3.44887972e-01 8.25221956e-01
4.39530015e-01 8.21705103e-01 5.36378443e-01 -6.17372692e-01
-8.01104367e-01 -1.06147015e+00 -4.55143422e-01 1.55661941e-01
6.43973231e-01 -9.63281214e-01 -2.98232436e-01 -1.12403527e-01] | [6.2501726150512695, -2.0748770236968994] |
e42f8b02-1bf1-4401-b490-087fb27d8e0b | saf-bage-salient-approach-for-facial-soft | 1803.05719 | null | http://arxiv.org/abs/1803.05719v2 | http://arxiv.org/pdf/1803.05719v2.pdf | SAF- BAGE: Salient Approach for Facial Soft-Biometric Classification - Age, Gender, and Facial Expression | How can we improve the facial soft-biometric classification with help of the
human visual system? This paper explores the use of saliency which is
equivalent to the human visual system to classify Age, Gender and Facial
Expression soft-biometric for facial images. Using the Deep Multi-level Network
(ML-Net) [1] and off-the-shelf face detector [2], we propose our approach -
SAF-BAGE, which first detects the face in the test image, increases the
Bounding Box (B-Box) margin by 30%, finds the saliency map using ML-Net, with
30% reweighted ratio of saliency map, it multiplies with the input cropped face
and extracts the Convolutional Neural Networks (CNN) predictions on the
multiplied reweighted salient face. Our CNN uses the model AlexNet [3], which
is pre-trained on ImageNet. The proposed approach surpasses the performance of
other approaches, increasing the state-of-the-art by approximately 0.8% on the
widely-used Adience [28] dataset for Age and Gender classification and by
nearly 3% on the recent AffectNet [36] dataset for Facial Expression
classification. We hope our simple, reproducible and effective approach will
help ease future research in facial soft-biometric classification using
saliency. | ['Viraj Mavani', 'Kenil Shah', 'Ayesha Gurnani', 'Yash Khandhediya', 'Vandit Gajjar'] | 2018-03-13 | null | null | null | null | ['age-and-gender-classification'] | ['computer-vision'] | [ 8.50889757e-02 4.23771828e-01 -7.61952326e-02 -6.73800707e-01
1.04660289e-02 6.56733215e-02 2.82171249e-01 -3.91342014e-01
-3.08652669e-01 3.97441536e-01 4.32273485e-02 2.73747027e-01
3.08208406e-01 -4.83787954e-01 -4.59856629e-01 -6.79569364e-01
-5.86877577e-02 -3.48891728e-02 1.43576056e-01 -4.05880511e-01
2.60545045e-01 6.77204311e-01 -2.18829679e+00 2.98154861e-01
4.85180438e-01 1.77528572e+00 -3.75035584e-01 4.57284391e-01
7.02243075e-02 7.59381711e-01 -3.92849714e-01 -9.21218455e-01
1.83147848e-01 -1.58541650e-01 -7.07666993e-01 -1.85200304e-01
9.71059561e-01 -4.53367740e-01 -1.90517336e-01 1.43949115e+00
7.52352715e-01 -3.97627681e-01 6.23205483e-01 -1.67440796e+00
-7.37927139e-01 1.58481181e-01 -1.02182233e+00 1.81371376e-01
3.43289167e-01 -2.78446283e-02 3.95345032e-01 -1.35486174e+00
4.11592275e-01 1.74923503e+00 7.36681104e-01 1.07720077e+00
-4.62538987e-01 -1.22596586e+00 -2.48867154e-01 3.72116089e-01
-1.48789275e+00 -7.43729115e-01 7.36998737e-01 -2.36288875e-01
7.68369138e-01 1.99980468e-01 7.44658887e-01 8.57617497e-01
1.24602400e-01 8.20322037e-01 1.31427681e+00 -3.18637460e-01
-3.69899087e-02 2.46532649e-01 1.05116181e-02 1.24842560e+00
-2.16877505e-01 9.81855169e-02 -7.54129231e-01 -1.77889153e-01
6.95916951e-01 8.77596885e-02 1.88434094e-01 1.33746490e-01
-4.46693420e-01 5.58013916e-01 6.59027576e-01 2.97491044e-01
-5.09720385e-01 3.48266698e-02 3.97688776e-01 1.18664004e-01
7.35281169e-01 4.52155136e-02 -2.45265841e-01 1.66486025e-01
-1.47854614e+00 2.63656676e-01 4.21077579e-01 5.12409329e-01
7.39042699e-01 3.56511533e-01 -2.56320685e-01 8.69573653e-01
5.51080227e-01 6.70541227e-01 4.52576488e-01 -6.41057611e-01
-3.81024092e-01 9.15960670e-01 -3.59376878e-01 -1.13706994e+00
-4.49522376e-01 -9.18928236e-02 -8.02076399e-01 7.79573023e-01
3.26224715e-01 -1.42807186e-01 -1.20637798e+00 1.72956812e+00
3.60630482e-01 4.71905410e-01 -3.64923179e-01 9.94773328e-01
1.26945162e+00 1.93043023e-01 4.42856312e-01 -5.06986352e-03
1.55865526e+00 -8.05969894e-01 -6.74495757e-01 -5.12070730e-02
1.73782706e-02 -7.15274870e-01 7.65003324e-01 1.53691620e-01
-1.38261938e+00 -6.25154078e-01 -1.08311963e+00 9.06732902e-02
-5.05850315e-01 4.45179582e-01 6.04094207e-01 9.96125638e-01
-1.54573262e+00 6.24881566e-01 -4.38892514e-01 -5.54649413e-01
1.01322174e+00 9.18195248e-01 -5.49326301e-01 4.54814523e-01
-1.09116864e+00 1.13382339e+00 -2.26219207e-01 6.25389740e-02
-8.97231817e-01 -7.78997064e-01 -9.82590556e-01 2.03106925e-02
-1.67513281e-01 -4.46751714e-01 9.44116712e-01 -2.02374721e+00
-1.62108421e+00 1.60298371e+00 -4.70585763e-01 -2.16603398e-01
1.52431875e-01 -7.06275925e-02 -5.09875357e-01 3.09124827e-01
-1.28398791e-01 1.21876514e+00 1.45950580e+00 -7.29352534e-01
-3.16880137e-01 -8.25105011e-01 -2.58497864e-01 -6.07284866e-02
-4.38878655e-01 8.38605702e-01 -3.61476764e-02 -4.41830724e-01
-1.97363794e-01 -7.98789799e-01 1.85570478e-01 4.98335630e-01
-6.31741527e-03 -4.68176723e-01 1.22305059e+00 -9.67282116e-01
1.12318683e+00 -2.26120067e+00 -6.63341135e-02 3.01474720e-01
3.92846823e-01 6.30181074e-01 -1.42640039e-01 -4.95612979e-01
-4.69644338e-01 -9.14046615e-02 -6.29248321e-02 -5.45612931e-01
-2.20039174e-01 -4.17267740e-01 1.22095747e-02 7.48283327e-01
5.04270375e-01 1.00055122e+00 -5.51408112e-01 -7.30359375e-01
2.29397595e-01 5.69661379e-01 -4.12685901e-01 1.61715522e-01
3.71175855e-01 -1.91689208e-01 2.44567301e-02 1.35264778e+00
1.21815503e+00 1.41777351e-01 -3.94660205e-01 -3.10573637e-01
8.27921256e-02 -3.94967526e-01 -1.00629497e+00 1.29682052e+00
-6.19030297e-02 6.68695629e-01 4.16505605e-01 -6.48017585e-01
1.21211886e+00 2.63621032e-01 3.20408136e-01 -5.86680412e-01
5.26909113e-01 2.69192100e-01 -6.00032955e-02 -5.39675176e-01
1.85041755e-01 -2.19119444e-01 2.67820865e-01 -2.15873532e-02
6.16802752e-01 3.26877981e-01 -3.70746642e-01 -3.33222412e-02
6.16571367e-01 1.16705552e-01 3.11836600e-01 -6.19877517e-01
9.13052559e-01 -8.06345224e-01 3.88120651e-01 1.94659978e-02
-9.32309449e-01 5.11545420e-01 5.53425133e-01 -5.96826196e-01
-9.09838617e-01 -8.62457871e-01 -1.53328925e-01 1.55147648e+00
-2.90879235e-02 -1.21558651e-01 -1.45265329e+00 -8.69041264e-01
3.08311939e-01 2.04368513e-02 -1.14326811e+00 -1.78613961e-01
-2.28746921e-01 -6.25453472e-01 7.46980727e-01 3.94307852e-01
8.88032794e-01 -1.46173799e+00 -5.34388959e-01 -3.32593560e-01
2.87866563e-01 -7.97229588e-01 -2.79214680e-01 -3.89805049e-01
-3.92803818e-01 -9.84960675e-01 -1.08453178e+00 -1.03254759e+00
7.77043521e-01 -2.58554876e-01 9.93735492e-01 4.59013671e-01
-4.58755434e-01 1.16760731e-01 1.77447125e-01 -7.76515424e-01
-1.84858423e-02 -7.21528903e-02 2.53857285e-01 6.78132176e-01
9.15037990e-01 -4.12941664e-01 -8.33624542e-01 1.11194760e-01
-3.67664546e-01 -3.40650454e-02 4.43139374e-01 6.18788242e-01
1.23843434e-03 -7.48345077e-01 8.87318254e-01 -3.47864777e-01
3.16498488e-01 -2.59089828e-01 -2.93997288e-01 1.67725474e-01
-4.52198714e-01 -2.69203663e-01 1.04781069e-01 -3.16549957e-01
-7.87948132e-01 2.48105645e-01 -4.44987357e-01 -6.95745707e-01
-1.38868153e-01 -2.30947778e-01 -9.15594678e-03 -7.18828499e-01
5.42368829e-01 2.09287927e-02 4.22269315e-01 -5.37864789e-02
-5.11067137e-02 1.12260938e+00 4.88799870e-01 2.90193092e-02
5.48924029e-01 5.12917161e-01 2.94109553e-01 -5.64490020e-01
-8.14067841e-01 -1.70607880e-01 -6.87768400e-01 -7.03123987e-01
9.70692158e-01 -9.11833227e-01 -1.31934547e+00 1.02388048e+00
-1.03630924e+00 2.26439029e-01 2.52389401e-01 -1.66360214e-01
-2.59275407e-01 1.69798791e-01 -6.02246165e-01 -1.19681954e+00
-9.01594818e-01 -9.93241191e-01 1.49564302e+00 7.91717947e-01
-3.37851256e-01 -6.09070480e-01 -4.49644595e-01 8.66650641e-02
7.10756183e-01 3.86414617e-01 4.08427447e-01 -3.49177420e-01
6.35318160e-02 -5.28104156e-02 -5.67521393e-01 5.37727296e-01
-1.06456585e-01 4.86665726e-01 -1.62400603e+00 -1.01138882e-01
-1.05161302e-01 -5.90970337e-01 9.46612537e-01 5.48161983e-01
1.61851025e+00 -2.82843918e-01 -1.39467970e-01 6.21179342e-01
1.24740577e+00 -1.00743070e-01 9.13011074e-01 1.92825627e-02
4.39294338e-01 7.73212433e-01 4.01683241e-01 5.02337754e-01
4.75839436e-01 5.30379832e-01 7.04542220e-01 -8.02483261e-01
-2.62987286e-01 3.65582034e-02 3.85833114e-01 1.51975080e-01
-3.99737269e-01 7.72044361e-01 -6.71115875e-01 4.70273614e-01
-1.65337467e+00 -1.08981228e+00 3.50083500e-01 1.71979046e+00
9.50181246e-01 -1.82663053e-01 5.45032084e-01 1.76382124e-01
8.62398922e-01 1.81975126e-01 -4.12988573e-01 -7.28545606e-01
-1.91394687e-01 8.90695095e-01 3.57148498e-01 4.02842909e-01
-1.26691806e+00 1.26858664e+00 6.47619915e+00 9.01305020e-01
-1.83845019e+00 2.40963981e-01 1.13595581e+00 -2.78383851e-01
2.93471217e-01 -6.32860899e-01 -9.53528464e-01 6.82662249e-01
1.02236509e+00 -2.72131041e-02 4.99876022e-01 1.21598995e+00
-6.69421181e-02 -1.62316009e-01 -8.80261660e-01 1.47415066e+00
7.35942125e-01 -1.10618186e+00 -1.31733567e-01 -2.58483648e-01
5.00059962e-01 -3.54570568e-01 5.27658582e-01 3.25435847e-01
-3.67074579e-01 -1.47315729e+00 7.30907977e-01 6.53170764e-01
1.07793593e+00 -8.63377571e-01 9.52451169e-01 -3.04507881e-01
-1.02875197e+00 -2.97093838e-01 -3.48272115e-01 -2.06865385e-01
-3.94745201e-01 3.18488657e-01 -6.75697386e-01 -1.52742431e-01
1.10131907e+00 6.33032084e-01 -9.93862212e-01 7.27024019e-01
5.99073097e-02 4.87391800e-01 -1.93450779e-01 -3.06914419e-01
9.80878323e-02 1.64149359e-01 2.55082697e-01 1.22384894e+00
2.39789039e-01 -6.49011061e-02 -3.20252150e-01 6.58362925e-01
-2.03426540e-01 3.79052728e-01 -4.52532351e-01 3.99490923e-01
1.65996701e-01 1.71396053e+00 -4.54327196e-01 -5.75502574e-01
-2.88222969e-01 9.75889266e-01 1.55795559e-01 2.19487883e-02
-8.39679956e-01 -4.98935044e-01 8.91860366e-01 2.08957851e-01
-1.22075276e-02 4.37348515e-01 -6.31544054e-01 -7.16801882e-01
-1.38342351e-01 -9.31141734e-01 2.01103613e-01 -9.84520495e-01
-9.90650654e-01 8.04074585e-01 -3.90204817e-01 -7.49531806e-01
-1.80152223e-01 -8.51910472e-01 -6.03131831e-01 1.14246440e+00
-1.50952566e+00 -1.50343120e+00 -5.37692964e-01 8.50660741e-01
3.29802841e-01 -5.82914591e-01 8.53568614e-01 3.53815317e-01
-5.05244970e-01 1.03696275e+00 -7.38044798e-01 3.16205025e-01
8.54386747e-01 -9.37547028e-01 1.56661555e-01 5.21584570e-01
-4.27662998e-01 5.09757817e-01 5.09117842e-01 -4.16902572e-01
-9.55188751e-01 -9.06392038e-01 1.12333572e+00 -4.05352652e-01
3.72650087e-01 -5.33514321e-01 -7.50194192e-01 2.58838564e-01
4.10693049e-01 2.45429814e-01 6.49993002e-01 -1.97961792e-01
-3.90444249e-01 -5.10477901e-01 -1.92023981e+00 6.87093794e-01
9.36053872e-01 -5.67609251e-01 -4.88340765e-01 -3.00817803e-04
2.57215500e-01 -2.63838261e-01 -6.50816798e-01 8.58501434e-01
9.54637706e-01 -1.03329730e+00 8.07141066e-01 -5.74514151e-01
6.40856087e-01 -3.30986716e-02 1.24115050e-01 -8.68243992e-01
-3.79741311e-01 -5.37699878e-01 -1.58290550e-01 1.08440709e+00
2.01939330e-01 -3.83814871e-01 9.71082211e-01 4.54818308e-01
2.74610996e-01 -1.13067138e+00 -1.15150774e+00 -1.56264946e-01
-2.57834882e-01 9.33476835e-02 8.83559346e-01 7.93527365e-01
2.79569793e-02 -1.20224319e-01 -4.10172790e-01 -6.19122759e-02
5.87399542e-01 -3.66658121e-01 3.90404105e-01 -1.30145121e+00
4.77853298e-01 -6.55967414e-01 -1.04151881e+00 -1.43690243e-01
5.82324743e-01 -8.14746082e-01 -4.12157893e-01 -7.35212743e-01
3.21889400e-01 2.52260059e-01 -5.48004866e-01 9.14343119e-01
-2.12108672e-01 9.75279093e-01 2.68216074e-01 -4.40102249e-01
-4.95015204e-01 5.49978673e-01 1.07195139e+00 -1.70210063e-01
2.06801474e-01 -2.20196545e-01 -8.83123040e-01 9.53727067e-01
5.03158092e-01 -7.21054226e-02 1.87018931e-01 3.50675792e-01
-1.40065432e-01 -3.20698231e-01 4.96573985e-01 -1.03678679e+00
1.42275795e-01 7.00068474e-02 1.08064258e+00 -4.28109080e-01
4.16823238e-01 -6.28664851e-01 -3.93368840e-01 5.32943249e-01
-1.46658629e-01 1.30308464e-01 4.60876405e-01 -4.07062143e-01
-2.28000790e-01 -1.40728429e-01 1.40422142e+00 2.85362778e-03
-7.89199471e-01 5.53965986e-01 -1.58738002e-01 -5.59049070e-01
1.18397164e+00 -4.23386276e-01 -2.50004262e-01 -3.53789061e-01
-8.07329416e-01 -1.45312458e-01 2.61858642e-01 5.91218054e-01
7.74528444e-01 -1.52550793e+00 -7.49433339e-01 6.47367537e-01
-1.31201759e-01 -8.14203024e-01 2.42032900e-01 8.24894786e-01
-3.04687530e-01 9.05303285e-03 -9.29052353e-01 -6.87426209e-01
-1.97931564e+00 3.30035299e-01 7.25593448e-01 3.77898932e-01
1.83944434e-01 1.17356181e+00 9.56307501e-02 -1.91970363e-01
2.81170964e-01 8.65240470e-02 -5.81118882e-01 1.05572075e-01
8.61070454e-01 3.68790865e-01 1.19411275e-02 -1.08188438e+00
-7.47315168e-01 7.65784502e-01 -3.04943463e-03 4.30766903e-02
1.23286867e+00 1.89488769e-01 -7.16164768e-01 -3.42282318e-02
1.30073082e+00 -2.10520208e-01 -9.89435315e-01 4.86734621e-02
-1.54379934e-01 -5.55226207e-01 1.05736189e-01 -8.27185452e-01
-1.41480660e+00 8.27848971e-01 1.41565561e+00 -1.54776737e-01
1.50644267e+00 -1.07425116e-02 4.08720136e-01 -2.57857502e-01
1.23434752e-01 -1.25805473e+00 1.90059751e-01 4.10499841e-01
1.16176128e+00 -1.35602272e+00 -1.19294815e-01 -3.75201553e-01
-6.01457179e-01 1.05446029e+00 1.04812896e+00 -3.45717341e-01
1.08384907e+00 3.57856929e-01 2.10461915e-01 -3.33237976e-01
-4.74468350e-01 -2.87527591e-01 5.80230236e-01 5.50922334e-01
4.83514339e-01 8.25678334e-02 -1.68360904e-01 6.94557369e-01
-5.19311130e-01 3.70409787e-01 1.17457274e-03 4.67760712e-01
-5.37342072e-01 -5.88875711e-01 -4.67528760e-01 5.44509709e-01
-9.78644431e-01 -1.18724093e-01 -6.66802943e-01 2.74128109e-01
6.25482798e-01 6.24030232e-01 3.31496596e-01 -5.03907204e-01
7.44001046e-02 3.51161629e-01 6.08836770e-01 -3.25024843e-01
-9.64661419e-01 -1.93567798e-01 -2.16141179e-01 -7.17270613e-01
-5.05901396e-01 -7.05650389e-01 -1.00890815e+00 -4.49097306e-01
-1.51043599e-02 -4.86997634e-01 8.34306121e-01 8.12672019e-01
4.28239435e-01 4.16645221e-02 7.96776354e-01 -1.15842152e+00
-2.13071987e-01 -1.35370624e+00 -5.78570306e-01 4.96915907e-01
4.96084780e-01 -9.28592622e-01 -2.52835006e-01 -1.26812607e-02] | [13.48362922668457, 1.2658003568649292] |
0b88eb9f-cf0b-478b-a8c3-1bf5002090e2 | dense-label-encoding-for-boundary | 2011.09670 | null | https://arxiv.org/abs/2011.09670v4 | https://arxiv.org/pdf/2011.09670v4.pdf | Dense Label Encoding for Boundary Discontinuity Free Rotation Detection | Rotation detection serves as a fundamental building block in many visual applications involving aerial image, scene text, and face etc. Differing from the dominant regression-based approaches for orientation estimation, this paper explores a relatively less-studied methodology based on classification. The hope is to inherently dismiss the boundary discontinuity issue as encountered by the regression-based detectors. We propose new techniques to push its frontier in two aspects: i) new encoding mechanism: the design of two Densely Coded Labels (DCL) for angle classification, to replace the Sparsely Coded Label (SCL) in existing classification-based detectors, leading to three times training speed increase as empirically observed across benchmarks, further with notable improvement in detection accuracy; ii) loss re-weighting: we propose Angle Distance and Aspect Ratio Sensitive Weighting (ADARSW), which improves the detection accuracy especially for square-like objects, by making DCL-based detectors sensitive to angular distance and object's aspect ratio. Extensive experiments and visual analysis on large-scale public datasets for aerial images i.e. DOTA, UCAS-AOD, HRSC2016, as well as scene text dataset ICDAR2015 and MLT, show the effectiveness of our approach. The source code is available at https://github.com/Thinklab-SJTU/DCL_RetinaNet_Tensorflow and is also integrated in our open source rotation detection benchmark: https://github.com/yangxue0827/RotationDetection. | ['Junchi Yan', 'Wentao Wang', 'Yue Zhou', 'Liping Hou', 'Xue Yang'] | 2020-11-19 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Yang_Dense_Label_Encoding_for_Boundary_Discontinuity_Free_Rotation_Detection_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Yang_Dense_Label_Encoding_for_Boundary_Discontinuity_Free_Rotation_Detection_CVPR_2021_paper.pdf | cvpr-2021-1 | ['object-detection-in-aerial-images'] | ['computer-vision'] | [-1.85628496e-02 -2.95418024e-01 -2.91191816e-01 -1.12950943e-01
-4.78821307e-01 -5.27845502e-01 6.48692369e-01 -9.57217999e-03
-2.35278383e-01 2.75937557e-01 2.59053886e-01 -3.68804723e-01
-1.79268450e-01 -6.12130642e-01 -4.16472703e-01 -8.44198585e-01
-1.18366413e-01 -5.18590920e-02 2.70002514e-01 -3.45886528e-01
4.85510945e-01 8.99498940e-01 -1.60110927e+00 5.60777746e-02
5.34037769e-01 1.17188692e+00 -1.44472316e-01 6.33596659e-01
4.16518569e-01 7.70894170e-01 -2.09652945e-01 -2.08192781e-01
7.16465831e-01 -4.28665765e-02 -6.84761822e-01 1.42613336e-01
7.24163890e-01 -3.72201383e-01 -3.60214919e-01 9.28847194e-01
6.62397623e-01 -6.80376589e-02 9.07592058e-01 -1.11453056e+00
-6.40522897e-01 2.43984967e-01 -1.42585719e+00 4.32773679e-01
3.17832559e-01 1.85056642e-01 9.18016434e-01 -1.06482589e+00
5.31359315e-01 1.17732513e+00 7.69965053e-01 2.42741287e-01
-7.77088344e-01 -5.91293633e-01 1.51178226e-01 2.37471238e-01
-1.65608895e+00 -4.91822094e-01 6.63967431e-01 -4.62266952e-01
8.42548966e-01 4.02113408e-01 3.20859581e-01 8.88708353e-01
5.42146079e-02 6.61238432e-01 1.13658369e+00 -4.64481920e-01
-6.74535930e-02 1.06609546e-01 -4.02966812e-02 1.02987325e+00
4.10283715e-01 3.11677635e-01 -1.87685072e-01 2.05163315e-01
8.33310187e-01 -3.13621163e-02 -4.27413344e-01 -4.30368900e-01
-1.16543615e+00 7.07749605e-01 7.99366772e-01 -3.34205963e-02
-1.44658104e-01 6.36800230e-02 3.83368641e-01 3.70411992e-01
6.21146381e-01 3.09160531e-01 -4.27788585e-01 3.45148563e-01
-7.99075246e-01 3.61698493e-02 2.38376364e-01 9.37483966e-01
5.45642018e-01 1.44895375e-01 -2.39194289e-01 9.77954566e-01
4.36505973e-01 6.30574048e-01 4.39668059e-01 -4.85382378e-01
3.90289873e-01 6.07614279e-01 -7.59120332e-03 -1.16526663e+00
-6.59255803e-01 -5.85624933e-01 -9.35918987e-01 3.19530368e-01
3.11685562e-01 -1.24478772e-01 -9.92938995e-01 1.31033826e+00
5.75948298e-01 -6.40752241e-02 -1.88357353e-01 1.16328192e+00
9.17997301e-01 4.79824901e-01 -2.15576664e-01 1.30704679e-02
1.55516255e+00 -8.77629519e-01 -1.61197990e-01 -1.04107305e-01
8.45195711e-01 -1.13053751e+00 7.33113229e-01 5.52368164e-01
-5.86154640e-01 -6.66539133e-01 -1.20103061e+00 -4.24699485e-02
-4.80804831e-01 7.87971318e-01 8.79882395e-01 6.99101150e-01
-8.94730508e-01 2.27350742e-01 -4.49889779e-01 -5.83578348e-01
6.35174811e-01 2.57626444e-01 -3.44158083e-01 -1.08213812e-01
-6.60787642e-01 8.07990968e-01 1.21905021e-02 1.56316951e-01
-6.98899567e-01 -4.40172583e-01 -8.00229907e-01 -3.17659289e-01
4.03458446e-01 -4.36870486e-01 9.60470438e-01 -8.57466877e-01
-1.22146869e+00 1.10481226e+00 1.50171191e-01 -4.69035327e-01
5.47993898e-01 -1.34392396e-01 -4.49136436e-01 1.13821015e-01
1.01758547e-01 1.01284373e+00 1.15068161e+00 -1.11612272e+00
-7.65572369e-01 -5.93691468e-01 1.57315254e-01 3.42227221e-01
-2.76508212e-01 1.30791366e-01 -2.94802457e-01 -7.73392975e-01
2.14375392e-01 -9.88640070e-01 -3.97719219e-02 2.54006177e-01
-4.90083396e-01 -1.69178665e-01 8.89610767e-01 -4.54419404e-01
1.17269051e+00 -2.17633224e+00 -8.78778547e-02 5.08962199e-02
2.27931097e-01 4.35406655e-01 -2.21290290e-01 2.64928043e-01
-3.78212780e-01 -1.28198013e-01 -1.32224068e-01 -5.71121760e-02
-3.19905847e-01 -4.61237997e-01 -4.26298350e-01 9.62055802e-01
4.48498487e-01 5.99883676e-01 -6.61983907e-01 -3.78206074e-01
5.40999711e-01 5.86014569e-01 -3.59435588e-01 -2.77224183e-01
3.13108206e-01 1.72263652e-01 -3.95617485e-01 1.23943520e+00
9.34797525e-01 -1.35425523e-01 -2.24333152e-01 -5.61107099e-01
-4.10133898e-01 -9.52381343e-02 -1.45687652e+00 1.36095107e+00
-3.64314795e-01 7.59329736e-01 -3.25133055e-01 -1.04827607e+00
1.15758395e+00 5.12934066e-02 3.42467427e-01 -6.63353443e-01
4.22708094e-01 7.45156547e-03 -8.44923556e-02 -3.80838513e-01
5.02416849e-01 3.87477338e-01 1.07439101e-01 -1.66102260e-01
5.73527953e-03 -2.26802770e-02 9.23713222e-02 -8.08666199e-02
9.41102684e-01 2.61726469e-01 6.31491601e-01 -3.14447731e-01
6.63576484e-01 2.49515567e-02 3.39168310e-01 3.12350303e-01
-3.27091604e-01 8.38680863e-01 3.60148489e-01 -5.75350404e-01
-7.80284882e-01 -7.91382253e-01 -6.68274343e-01 9.67616737e-01
2.28541553e-01 -3.23269278e-01 -4.19101179e-01 -5.58651388e-01
4.32270095e-02 1.57100722e-01 -6.88157022e-01 3.75312865e-02
-4.05275404e-01 -1.23212969e+00 6.21212959e-01 6.36420250e-01
6.71580970e-01 -6.59965456e-01 -6.92642987e-01 -2.90527165e-01
1.64372072e-01 -1.23878324e+00 -3.04959029e-01 1.96440920e-01
-8.27262104e-01 -1.17250776e+00 -6.85070395e-01 -6.07246518e-01
6.87641501e-01 7.48736978e-01 8.23384881e-01 2.28576530e-02
-1.00725520e+00 3.49171072e-01 -5.64817190e-01 -4.15711612e-01
3.63448054e-01 -2.70424504e-02 8.08271319e-02 7.27557205e-03
3.45556319e-01 -2.71813750e-01 -9.37680721e-01 5.78144848e-01
-5.62386930e-01 6.15138896e-02 7.51894355e-01 6.27228200e-01
3.63316715e-01 -2.47647434e-01 3.43940824e-01 -5.85745692e-01
1.32616445e-01 -4.66862798e-01 -9.95623708e-01 5.03664720e-04
-6.99353456e-01 -8.66836756e-02 4.02353525e-01 -2.66839206e-01
-6.83189929e-01 2.47213230e-01 4.18971926e-02 -2.66187429e-01
-2.82424033e-01 -6.77229930e-03 2.26567209e-01 -4.25008833e-01
8.56939375e-01 6.61734566e-02 -3.07908028e-01 -2.59772748e-01
3.01620573e-01 7.47885168e-01 3.62467110e-01 -1.54456943e-01
9.93105292e-01 7.37135947e-01 3.60800564e-01 -1.14307618e+00
-6.74393296e-01 -8.18157554e-01 -6.24958932e-01 -1.79240301e-01
8.01189721e-01 -1.31857085e+00 -6.31646752e-01 4.10226017e-01
-9.03078556e-01 -1.51483547e-02 -2.70109698e-02 5.57369351e-01
-2.05877870e-01 3.34349900e-01 -3.45129073e-01 -8.05315077e-01
-3.25066566e-01 -1.23184121e+00 1.13256180e+00 4.39509720e-01
1.93383873e-01 -7.48204529e-01 6.02434315e-02 3.37816268e-01
3.09718430e-01 4.53613192e-01 4.06817615e-01 -2.95561254e-01
-6.59440339e-01 -3.18275899e-01 -7.18479753e-01 5.01727760e-01
-7.78689161e-02 3.09339792e-01 -1.20240116e+00 -3.88237834e-01
-4.14644241e-01 -4.15421367e-01 1.13296759e+00 5.00418782e-01
1.09493744e+00 2.83083729e-02 -3.37154210e-01 1.01206756e+00
1.72350776e+00 -6.71068728e-02 6.83652878e-01 3.24922711e-01
8.18544090e-01 4.02520835e-01 9.87758279e-01 5.81672490e-01
3.32719237e-01 9.10566151e-01 6.32433653e-01 -4.02738065e-01
-4.91049081e-01 5.26940227e-02 4.02132660e-01 5.02850473e-01
-6.51357710e-01 -1.78162798e-01 -8.35022807e-01 2.90921718e-01
-1.56789756e+00 -8.12953115e-01 -3.73656094e-01 2.34227800e+00
4.35131907e-01 -4.82347459e-02 1.27779350e-01 3.91009301e-01
5.47224939e-01 4.12790000e-01 -2.94181317e-01 -2.92242199e-01
-1.54047191e-01 1.70817226e-01 1.18118477e+00 3.53591084e-01
-1.60993278e+00 9.67647791e-01 4.87927771e+00 1.02435946e+00
-1.38096869e+00 -2.54186778e-03 7.60039032e-01 6.23194128e-02
4.59892273e-01 -7.05698058e-02 -1.21699488e+00 1.38657957e-01
3.32860261e-01 3.46202016e-01 2.99361736e-01 1.03302944e+00
2.18870729e-01 -2.27358252e-01 -7.10973382e-01 1.18937361e+00
3.31417024e-01 -1.08047688e+00 -1.90392315e-01 1.21093102e-01
6.49326384e-01 3.40155900e-01 2.59873062e-01 1.52334526e-01
-7.84969553e-02 -9.74438846e-01 6.36753976e-01 7.03936219e-02
1.16357446e+00 -7.07633197e-01 6.28218532e-01 -1.64908439e-01
-1.49933064e+00 -2.75423497e-01 -7.05832243e-01 2.16595773e-02
-4.52707738e-01 5.61376154e-01 -7.85165548e-01 6.82675362e-01
8.74871969e-01 9.86735642e-01 -1.06459808e+00 1.17960215e+00
-3.21057677e-01 4.83410895e-01 -1.82578593e-01 9.27372128e-02
2.99293458e-01 -9.41050798e-02 6.79781258e-01 1.31948519e+00
2.14398608e-01 -6.65466115e-02 -7.82959722e-03 3.06617409e-01
-2.98484825e-02 1.74114451e-01 -7.66080260e-01 4.62746471e-01
4.45220262e-01 1.76481473e+00 -1.10053086e+00 4.46377732e-02
-3.53814930e-01 7.79478014e-01 6.51036426e-02 1.99534148e-01
-1.18333125e+00 -4.52239096e-01 4.82866079e-01 3.69988739e-01
5.16621172e-01 -2.09463537e-01 -1.26389742e-01 -1.12678695e+00
2.88361218e-02 -8.69710863e-01 1.62177339e-01 -6.85633719e-01
-9.24705327e-01 6.56776071e-01 1.42218754e-01 -1.77632725e+00
2.38724098e-01 -1.07949960e+00 -5.91655672e-01 4.00935560e-01
-1.73839235e+00 -1.16357100e+00 -7.38078475e-01 4.73679423e-01
6.65134728e-01 -3.65229845e-01 4.83019769e-01 5.89484096e-01
-7.85962462e-01 6.24696791e-01 2.87367385e-02 3.02961200e-01
9.90695119e-01 -1.08330357e+00 1.96588248e-01 8.76439393e-01
3.28323811e-01 3.04174423e-01 5.61787486e-01 -1.34105101e-01
-1.40142655e+00 -1.22703540e+00 4.33460355e-01 -5.36695719e-01
4.23568040e-01 -2.38756448e-01 -5.10410547e-01 3.86711478e-01
2.76893917e-02 3.65950018e-01 4.34582978e-01 -3.38571556e-02
-6.21898770e-01 -3.62274855e-01 -1.13239527e+00 3.82609546e-01
1.18122530e+00 -1.87381282e-01 -2.38742381e-01 5.86152792e-01
2.63453305e-01 -3.41217339e-01 -6.49459243e-01 7.42063761e-01
6.25633419e-01 -1.17527318e+00 1.29626918e+00 -1.78734675e-01
4.60053325e-01 -7.13315248e-01 -2.34377787e-01 -9.71855402e-01
-2.80840367e-01 -2.60054946e-01 2.89860424e-02 1.09101593e+00
2.16011614e-01 -6.82167172e-01 4.30852115e-01 -4.12214845e-01
-7.46698976e-02 -9.46030140e-01 -8.04177403e-01 -8.09725106e-01
-3.31553102e-01 -3.33026797e-01 2.67504692e-01 1.04272532e+00
-5.62044859e-01 3.76435310e-01 -3.82105947e-01 3.45181495e-01
6.11701965e-01 -8.49880800e-02 7.64257967e-01 -1.05474150e+00
-1.31837681e-01 -4.88163024e-01 -7.58465946e-01 -1.06113029e+00
-4.17854875e-01 -8.14597309e-01 -2.56485194e-01 -1.17063570e+00
-1.32597378e-02 -5.14650881e-01 -7.84280151e-02 5.36996603e-01
2.76216209e-01 8.67623746e-01 1.36881039e-01 1.70057029e-01
-6.25233889e-01 3.82437408e-01 1.00727940e+00 2.20428985e-02
8.33233993e-04 -4.54103537e-02 -6.72246337e-01 9.49408948e-01
8.67133856e-01 -4.74653840e-01 -3.45414519e-01 -4.41626012e-01
2.66314745e-01 -1.61205322e-01 7.23109126e-01 -1.35614800e+00
1.08718745e-01 2.32589357e-02 7.81305969e-01 -7.29206502e-01
1.77467704e-01 -5.82258701e-01 -2.97410756e-01 5.52816749e-01
7.96447974e-03 1.77920431e-01 1.57348663e-01 4.31733221e-01
-6.69092909e-02 -2.09831093e-02 1.16038597e+00 1.91658512e-01
-8.01940143e-01 3.26783568e-01 -7.82544632e-03 -5.92771173e-02
1.12981641e+00 -3.69713515e-01 -6.09477758e-01 1.05348252e-01
-3.26186895e-01 6.29692525e-02 3.56843442e-01 5.48233926e-01
6.35392547e-01 -1.15678501e+00 -8.22753429e-01 1.79084659e-01
4.03974265e-01 -4.29815054e-02 1.38882488e-01 1.19256186e+00
-9.08476412e-01 5.76322019e-01 -3.14252406e-01 -7.67954767e-01
-1.41734111e+00 4.22227055e-01 2.16849878e-01 -8.38258862e-03
-5.35032690e-01 1.01320374e+00 1.74173221e-01 -1.56084716e-01
2.63534039e-01 -5.40794909e-01 -4.25883561e-01 1.91675872e-01
4.95226383e-01 5.65338910e-01 1.71301469e-01 -6.80657208e-01
-6.53901160e-01 1.01146340e+00 -5.02542593e-02 2.88403034e-01
1.21817255e+00 -4.21977639e-02 1.27199113e-01 3.30796763e-02
1.12435937e+00 1.23438582e-01 -1.07243562e+00 -1.11650005e-01
-1.80944055e-01 -6.73689008e-01 3.79357874e-01 -6.72868788e-01
-1.13260055e+00 8.04181457e-01 1.14391756e+00 7.00908247e-03
1.37790883e+00 -1.69386178e-01 2.51731962e-01 3.93129170e-01
2.00887993e-01 -1.00765753e+00 6.62409961e-02 3.33396494e-01
1.07089531e+00 -1.67158329e+00 6.21748507e-01 -5.08869410e-01
-7.05830872e-01 1.15938509e+00 6.14285648e-01 -5.08417189e-01
5.48468113e-01 1.32217750e-01 -4.67968062e-02 -1.55476317e-01
-5.13888478e-01 -4.96191591e-01 5.07212996e-01 6.19331896e-01
6.17236316e-01 1.56952471e-01 -3.43143374e-01 -2.53927559e-02
5.84892370e-03 -2.97801942e-01 4.29361343e-01 6.93270981e-01
-4.24109876e-01 -7.79516697e-01 -4.89402831e-01 3.85027617e-01
-3.08919400e-01 -1.84832796e-01 -2.66604751e-01 9.26700354e-01
3.83964300e-01 6.79850936e-01 5.74150831e-02 -4.30153877e-01
2.14212775e-01 -5.28495312e-01 6.15296483e-01 -4.96040583e-01
-2.95713425e-01 -1.58529468e-02 -7.28392228e-02 -5.71521282e-01
-4.88548785e-01 -5.92598617e-01 -8.92567277e-01 -1.43125892e-01
-5.46800137e-01 -4.69608992e-01 7.41835952e-01 4.75024253e-01
1.57658800e-01 4.31454331e-01 1.04199553e+00 -8.29020262e-01
-5.45189083e-01 -9.21680927e-01 -4.33358997e-01 4.41575497e-02
3.64984393e-01 -9.33056414e-01 -4.95189399e-01 -7.81961009e-02] | [8.701553344726562, -0.8252307772636414] |
f336f3a3-8a41-497c-b780-a4b570a225d0 | shallow-bayesian-meta-learning-for-real-world | 2101.02833 | null | https://arxiv.org/abs/2101.02833v2 | https://arxiv.org/pdf/2101.02833v2.pdf | Shallow Bayesian Meta Learning for Real-World Few-Shot Recognition | Current state-of-the-art few-shot learners focus on developing effective training procedures for feature representations, before using simple, e.g. nearest centroid, classifiers. In this paper, we take an orthogonal approach that is agnostic to the features used and focus exclusively on meta-learning the actual classifier layer. Specifically, we introduce MetaQDA, a Bayesian meta-learning generalization of the classic quadratic discriminant analysis. This setup has several benefits of interest to practitioners: meta-learning is fast and memory-efficient, without the need to fine-tune features. It is agnostic to the off-the-shelf features chosen and thus will continue to benefit from advances in feature representations. Empirically, it leads to robust performance in cross-domain few-shot learning and, crucially for real-world applications, it leads to better uncertainty calibration in predictions. | ['Timothy Hospedales', 'Henry Gouk', 'Debin Meng', 'Xueting Zhang'] | 2021-01-08 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Zhang_Shallow_Bayesian_Meta_Learning_for_Real-World_Few-Shot_Recognition_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Zhang_Shallow_Bayesian_Meta_Learning_for_Real-World_Few-Shot_Recognition_ICCV_2021_paper.pdf | iccv-2021-1 | ['cross-domain-few-shot', 'cross-domain-few-shot-learning'] | ['computer-vision', 'computer-vision'] | [-7.82809034e-02 -2.08618075e-01 -3.65402192e-01 -4.42395180e-01
-1.34700751e+00 -3.06834638e-01 7.41649687e-01 1.85116991e-01
-2.35193968e-01 7.82355547e-01 -3.55852470e-02 -1.09557115e-01
-6.54545486e-01 -7.84402728e-01 -4.73780811e-01 -7.64219284e-01
-1.46956399e-01 4.70396549e-01 2.07665965e-01 -4.02119845e-01
3.88442546e-01 3.00807595e-01 -2.00127292e+00 9.72569082e-03
8.00324798e-01 1.18852234e+00 -5.94067052e-02 6.17817581e-01
-5.60497381e-02 4.85787541e-01 -4.19947714e-01 -4.71050858e-01
7.55185932e-02 -3.44853610e-01 -6.37559593e-01 -8.13128576e-02
2.41519496e-01 3.21565717e-02 -3.32081877e-02 8.59187186e-01
4.19926673e-01 5.75028360e-01 1.12611747e+00 -1.17456472e+00
-5.36663711e-01 4.70216244e-01 -4.13564205e-01 3.04235309e-01
2.19914567e-04 2.13287786e-01 1.07278621e+00 -1.02308309e+00
3.82880300e-01 1.03085625e+00 8.65310907e-01 5.80431342e-01
-1.29404247e+00 -5.67276716e-01 6.30323635e-03 3.16918552e-01
-1.24305689e+00 -5.93232751e-01 5.72602689e-01 -6.25078142e-01
1.04671216e+00 5.68750203e-02 5.05386889e-01 1.19878280e+00
2.89869636e-01 5.85539818e-01 9.43354607e-01 -6.85940981e-01
7.73709297e-01 3.54880035e-01 4.73814130e-01 5.41101336e-01
1.62863567e-01 4.06134725e-01 -5.47177851e-01 -2.32266501e-01
1.46098822e-01 2.32989304e-02 8.06864202e-02 -8.86512458e-01
-7.20511377e-01 1.34550917e+00 8.44307169e-02 2.88832575e-01
-2.18431458e-01 2.48867184e-01 5.73725998e-01 3.28673273e-01
7.85654128e-01 6.04983568e-01 -6.17379248e-01 -5.25565982e-01
-1.14610803e+00 3.44373316e-01 7.28026509e-01 8.21981907e-01
9.61741745e-01 3.17023918e-02 -5.72633259e-02 9.58711803e-01
1.93462312e-01 8.65878463e-02 8.28770936e-01 -9.78772521e-01
2.55320907e-01 2.97218919e-01 1.90757886e-01 -4.23350751e-01
-3.71717215e-01 -4.61076081e-01 -4.98752356e-01 3.32885087e-01
1.53917998e-01 -3.20087731e-01 -9.08646524e-01 1.54281044e+00
2.76183933e-01 2.71842599e-01 5.96908256e-02 5.34092307e-01
5.11121094e-01 4.38877642e-01 1.84919208e-01 -2.44807199e-01
1.18769920e+00 -8.01515579e-01 -3.19608361e-01 -2.57849067e-01
7.91757703e-01 -4.91209894e-01 9.48579431e-01 4.44524229e-01
-7.41785884e-01 -7.16559827e-01 -1.39381921e+00 2.88962781e-01
-8.53750527e-01 -2.73185641e-01 9.49879169e-01 9.92519557e-01
-5.58695912e-01 1.07465947e+00 -8.22680295e-01 -2.66425073e-01
4.82157171e-01 3.58778119e-01 -2.78255850e-01 -2.02838048e-01
-1.24520016e+00 1.28653753e+00 6.37851238e-01 -5.44071376e-01
-7.58812249e-01 -1.22255993e+00 -9.85396206e-01 2.17390001e-01
4.63849217e-01 -7.46857285e-01 1.60178483e+00 -9.42331195e-01
-1.64322889e+00 6.06840789e-01 -3.55500802e-02 -5.00010908e-01
3.97680700e-01 -2.96720237e-01 -4.45576966e-01 -2.36570567e-01
-3.85413244e-02 4.04128283e-01 1.16919839e+00 -8.26826930e-01
-7.96594441e-01 -3.48497778e-01 -2.56508410e-01 -5.59932701e-02
-3.26293021e-01 -1.50217935e-01 3.62150297e-02 -5.24796247e-01
-3.64758521e-01 -7.92775154e-01 -2.52995163e-01 -1.48974344e-01
1.27774239e-01 -3.65699172e-01 8.31907094e-01 -7.97899812e-03
1.14616370e+00 -2.09971619e+00 7.44507089e-02 2.23622322e-01
-3.58930379e-02 2.67547697e-01 4.84227799e-02 6.39387012e-01
-1.30678311e-01 -1.90893576e-01 -2.75792450e-01 -2.43272841e-01
1.05544120e-01 2.84165405e-02 -4.03375059e-01 6.26186192e-01
3.45110357e-01 8.66127253e-01 -9.38370347e-01 -3.98939729e-01
4.85540509e-01 3.85712206e-01 -4.49639708e-01 1.14429690e-01
-1.42126098e-01 -5.92937581e-02 -4.42284793e-01 6.53443515e-01
4.39907938e-01 -1.97944909e-01 -1.36816740e-01 2.32988715e-01
-2.39535525e-01 7.77618065e-02 -1.21378720e+00 1.84396350e+00
-7.08097339e-01 4.50845987e-01 -3.85792464e-01 -1.16043258e+00
8.53070617e-01 2.65936665e-02 5.97233355e-01 -3.42924595e-01
2.82422215e-01 3.89502913e-01 -5.98676614e-02 -1.83295816e-01
4.89092857e-01 -3.88943911e-01 -2.03557819e-01 5.10741949e-01
7.34336615e-01 -2.09275693e-01 1.96437746e-01 -2.19298705e-01
8.08643460e-01 2.53283203e-01 8.79317462e-01 -3.91937613e-01
2.59092569e-01 5.43237366e-02 2.97699898e-01 7.44618773e-01
-1.40065953e-01 4.35119897e-01 3.28176431e-02 -2.72039235e-01
-9.65487421e-01 -1.09640920e+00 -5.16622901e-01 1.53928506e+00
-2.38868773e-01 -5.13020337e-01 -5.53277016e-01 -6.17308021e-01
4.28242713e-01 1.06306326e+00 -8.67923856e-01 -5.84149897e-01
-8.17828402e-02 -8.49170089e-01 1.47381335e-01 6.58160150e-01
-3.40052322e-02 -8.12747478e-01 -9.11499441e-01 3.40446353e-01
5.13084352e-01 -4.85784560e-01 -1.93764493e-02 8.59391928e-01
-1.08829951e+00 -1.09349406e+00 -8.94460261e-01 -2.43056983e-01
-1.53107764e-02 1.80506185e-01 1.15194714e+00 -4.73421454e-01
-5.48683465e-01 6.01376593e-01 -4.98574138e-01 -8.11367631e-01
-2.20207602e-01 3.14592689e-01 8.27116221e-02 -1.65096372e-01
7.38016725e-01 -5.13237059e-01 -2.23665491e-01 2.57805996e-02
-5.21141768e-01 -5.97502530e-01 5.24640918e-01 1.24468994e+00
4.08370197e-01 5.12297861e-02 7.90148556e-01 -1.13234544e+00
7.63557673e-01 -6.97659731e-01 -5.71512580e-01 4.32045460e-01
-9.92241800e-01 2.36452162e-01 6.34318352e-01 -2.57454008e-01
-9.94277179e-01 1.36855841e-01 6.08223565e-02 -4.33932930e-01
-2.10074335e-01 5.17378092e-01 1.38446569e-01 -1.71967089e-01
1.00932813e+00 3.22943181e-02 -4.46157679e-02 -3.48702639e-01
6.96613789e-01 8.07946861e-01 1.93065792e-01 -6.67967856e-01
5.91938019e-01 2.09416062e-01 1.11702904e-01 -9.50259030e-01
-1.02014792e+00 -7.39756286e-01 -8.91594350e-01 -1.45179937e-02
4.29792136e-01 -7.55553603e-01 -3.34956318e-01 2.47250363e-01
-6.03850424e-01 -2.48201683e-01 -6.30268574e-01 4.87400770e-01
-1.08948946e+00 1.24430060e-01 -9.36990157e-02 -9.38853502e-01
-3.07540685e-01 -9.88139451e-01 8.68253350e-01 2.78969824e-01
-3.14886391e-01 -1.12882364e+00 3.21439117e-01 -1.24802291e-01
4.49693352e-01 -5.52598946e-02 1.03424895e+00 -8.49788845e-01
-7.65510127e-02 -3.46956730e-01 4.58971038e-02 3.50967824e-01
1.65491328e-01 1.61613271e-01 -1.56614816e+00 -2.87501454e-01
-2.28810593e-01 -7.10217357e-01 1.15111446e+00 5.61733246e-01
1.22384715e+00 2.06621662e-01 -2.71873593e-01 6.06871963e-01
1.27996397e+00 9.01497900e-02 2.47277662e-01 3.24406952e-01
1.27596810e-01 6.79767311e-01 9.96207476e-01 7.31928110e-01
2.07185969e-01 7.43056357e-01 7.36152455e-02 3.38296592e-01
5.10817319e-02 -4.93189357e-02 1.40865907e-01 5.98291814e-01
2.75072455e-02 1.85295850e-01 -8.95173311e-01 4.30521578e-01
-2.09479165e+00 -1.07471371e+00 3.64431530e-01 2.23800302e+00
6.74281538e-01 2.06511751e-01 3.45334768e-01 2.43202507e-01
5.40946901e-01 1.96483508e-01 -7.30179548e-01 -5.12209892e-01
1.93532914e-01 6.08220994e-01 3.06684852e-01 1.39502004e-01
-1.41328561e+00 8.50489914e-01 6.56697559e+00 1.10305786e+00
-1.05532372e+00 2.71616757e-01 5.03742695e-01 -2.01885596e-01
2.56307032e-02 -1.22862197e-01 -8.84624004e-01 3.50294501e-01
1.34143412e+00 -3.50055277e-01 4.32190895e-01 1.37966025e+00
-2.75231272e-01 -2.41604656e-01 -1.31641579e+00 1.04898643e+00
4.72620502e-02 -1.43849325e+00 -2.70351410e-01 -5.33785783e-02
7.83585370e-01 5.10622300e-02 2.60460198e-01 9.25829232e-01
2.79829502e-01 -1.03858638e+00 5.59415400e-01 5.85301280e-01
7.78063238e-01 -1.29989612e+00 8.01307201e-01 3.19500417e-01
-1.01826012e+00 -4.97225612e-01 -8.17544460e-01 6.81008771e-02
-3.30054536e-02 6.67343378e-01 -5.17129958e-01 4.99033034e-01
5.71480036e-01 5.75053334e-01 -6.38617814e-01 1.21907735e+00
2.01993197e-01 3.74950737e-01 -4.85663787e-02 -1.71956897e-01
3.27685952e-01 1.00417197e-01 1.56193912e-01 1.33597374e+00
4.16270405e-01 -8.91052112e-02 1.53527915e-01 5.45227766e-01
1.85897306e-01 8.79656896e-02 -7.64147103e-01 -8.18125159e-02
5.43510199e-01 1.23228240e+00 -4.01141614e-01 -3.00581634e-01
-5.10062754e-01 5.73597908e-01 5.57122707e-01 2.32412741e-02
-6.25213385e-01 -7.96603560e-01 7.32494235e-01 -1.04413867e-01
6.74794734e-01 -2.36334875e-02 -2.37083137e-01 -1.10764706e+00
-4.12762195e-01 -7.86235690e-01 5.65940619e-01 -2.81163782e-01
-1.71132171e+00 2.01148331e-01 2.34948844e-01 -1.25028980e+00
-7.40245640e-01 -9.38423872e-01 -7.65706062e-01 7.94839382e-01
-1.62761056e+00 -1.07105279e+00 -1.25394419e-01 5.81004798e-01
6.22293472e-01 -3.87971014e-01 1.26413274e+00 -2.99414899e-03
-3.09518993e-01 7.28009403e-01 5.75184047e-01 -2.51936078e-01
8.64952564e-01 -1.17906988e+00 3.42617929e-01 3.36400747e-01
2.98154742e-01 7.47875392e-01 7.03721464e-01 -3.03064317e-01
-1.46302068e+00 -9.82962906e-01 4.24517721e-01 -4.02046472e-01
8.50658119e-01 -1.80778980e-01 -7.91038275e-01 3.82528394e-01
-1.96995959e-01 3.16161186e-01 1.10639894e+00 8.52585435e-01
-5.02121031e-01 -2.54681885e-01 -1.16251016e+00 1.70547456e-01
5.89839935e-01 -6.64384663e-01 -8.87256444e-01 1.27740607e-01
1.97754383e-01 -1.32588372e-01 -8.29748631e-01 2.70883828e-01
7.97846913e-01 -1.03855884e+00 9.54105079e-01 -1.14362371e+00
2.25024372e-01 2.05904737e-01 -3.45275968e-01 -1.75020421e+00
-5.88002205e-01 -3.77814680e-01 -6.58241689e-01 1.14712787e+00
4.71068561e-01 -5.11727035e-01 8.55802000e-01 5.68988502e-01
-2.12438479e-01 -8.94216120e-01 -1.06614411e+00 -1.14943755e+00
3.08364838e-01 -7.17749715e-01 4.91056174e-01 8.94836485e-01
2.62383372e-01 3.00484300e-01 -3.39145690e-01 -2.11002216e-01
7.52877533e-01 3.36439371e-01 7.22284198e-01 -1.77344763e+00
-5.82674265e-01 -6.24168575e-01 -4.87227827e-01 -5.12805641e-01
2.31833860e-01 -7.52498627e-01 8.27150047e-02 -9.73193169e-01
1.55641854e-01 -4.31189656e-01 -7.09297419e-01 2.45640799e-01
-1.78659692e-01 1.01049453e-01 7.41991401e-02 -8.64307731e-02
-6.13256395e-01 7.51572609e-01 6.17085636e-01 -1.01534165e-01
-2.68506110e-01 4.24166173e-01 -7.47739792e-01 6.11436903e-01
8.24109674e-01 -6.08174860e-01 -5.25301456e-01 2.79264480e-01
-1.05918489e-01 -1.06369749e-01 1.98951632e-01 -1.14471781e+00
9.53478217e-02 -2.13906780e-01 5.92356980e-01 -2.72648424e-01
6.42139077e-01 -5.79361498e-01 -2.35966429e-01 3.68865520e-01
-2.06773579e-01 -2.01207623e-01 1.73783004e-01 7.07774162e-01
-6.61608800e-02 -8.98166835e-01 1.08329821e+00 -3.20510656e-01
-9.49529171e-01 1.82014883e-01 -3.05343360e-01 2.65389889e-01
1.34323418e+00 -3.17014843e-01 -2.07935348e-01 -2.23396882e-01
-5.46989501e-01 -5.83432540e-02 2.96023220e-01 4.21171784e-01
3.91566247e-01 -1.18741834e+00 -4.33768570e-01 1.52403757e-01
6.59218431e-01 -4.92513329e-01 2.17035323e-01 5.92697382e-01
1.36618003e-01 6.87928200e-01 -2.82242984e-01 -5.06052136e-01
-9.02565658e-01 6.96385503e-01 2.35598192e-01 -7.42464960e-02
-5.23678660e-01 8.98764133e-01 -1.52008906e-01 -3.39603186e-01
2.62879461e-01 -1.21890686e-01 1.87273789e-02 5.90925932e-01
6.70127153e-01 7.00408995e-01 2.45950013e-01 -2.32321188e-01
-3.46786678e-01 5.42631984e-01 -1.93606704e-01 -9.00707096e-02
1.50181568e+00 1.89904988e-01 5.87916434e-01 1.07177019e+00
1.05461323e+00 -3.28286618e-01 -1.31848598e+00 -5.56538701e-02
3.97362143e-01 -5.68225265e-01 3.67730200e-01 -7.31992960e-01
-5.87346137e-01 1.05299139e+00 7.15556324e-01 1.36304349e-01
7.24628508e-01 -1.08549096e-01 3.65111381e-01 7.37744689e-01
5.46518385e-01 -1.40482914e+00 3.33541594e-02 5.07919550e-01
4.98580277e-01 -1.55304170e+00 3.69002342e-01 4.37044688e-02
-7.11602628e-01 1.50841904e+00 4.43789929e-01 -1.95632935e-01
1.00572276e+00 6.91459030e-02 -8.33201632e-02 -2.05013141e-01
-9.56119299e-01 -2.65050232e-01 5.96507549e-01 8.68806601e-01
5.03734112e-01 9.99234989e-03 9.04864073e-02 6.87994242e-01
-2.93943733e-02 3.22214738e-02 2.01521277e-01 8.84555578e-01
-7.57451773e-01 -1.13868475e+00 -5.58384657e-02 7.44356930e-01
-2.44963035e-01 1.70118231e-02 -1.54932171e-01 1.02528977e+00
3.33382785e-02 7.40329564e-01 3.24619748e-02 -2.67860264e-01
2.43264079e-01 5.97753704e-01 6.58712804e-01 -9.13796902e-01
-5.11534572e-01 -1.26920432e-01 -2.89602415e-03 -3.06391597e-01
-3.19811583e-01 -8.47816408e-01 -8.07263911e-01 -2.01803476e-01
-6.11522555e-01 2.85981238e-01 7.05844343e-01 1.05161321e+00
3.75826508e-01 4.53010082e-01 6.38301015e-01 -1.11306000e+00
-1.24825037e+00 -1.17232561e+00 -8.05762768e-01 -6.59375712e-02
3.25597197e-01 -1.25340712e+00 -2.64795959e-01 -5.71270764e-01] | [9.88387393951416, 3.088408946990967] |
9ad1036e-308c-4504-8134-0c8a4ee6a74a | source-free-adaptive-gaze-estimation-by | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Cai_Source-Free_Adaptive_Gaze_Estimation_by_Uncertainty_Reduction_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Cai_Source-Free_Adaptive_Gaze_Estimation_by_Uncertainty_Reduction_CVPR_2023_paper.pdf | Source-Free Adaptive Gaze Estimation by Uncertainty Reduction | Gaze estimation across domains has been explored recently because the training data are usually collected under controlled conditions while the trained gaze estimators are used in real and diverse environments. However, due to privacy and efficiency concerns, simultaneous access to annotated source data and to-be-predicted target data can be challenging. In light of this, we present an unsupervised source-free domain adaptation approach for gaze estimation, which adapts a source-trained gaze estimator to unlabeled target domains without source data. We propose the Uncertainty Reduction Gaze Adaptation (UnReGA) framework, which achieves adaptation by reducing both sample and model uncertainty. Sample uncertainty is mitigated by enhancing image quality and making them gaze-estimation-friendly, whereas model uncertainty is reduced by minimizing prediction variance on the same inputs. Extensive experiments are conducted on six cross-domain tasks, demonstrating the effectiveness of UnReGA and its components. Results show that UnReGA outperforms other state-of-the-art cross-domain gaze estimation methods under both protocols, with and without source data | ['Xilin Chen', 'Shiguang Shan', 'Jiabei Zeng', 'Xin Cai'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['gaze-estimation', 'source-free-domain-adaptation'] | ['computer-vision', 'computer-vision'] | [ 3.53443861e-01 1.95050001e-01 -2.65256673e-01 -7.60042369e-01
-5.68633974e-01 -3.92647833e-01 1.31464109e-01 -2.80676752e-01
-5.16222119e-01 9.23870325e-01 2.59358762e-03 3.06031942e-01
-4.94351014e-02 -3.76818590e-02 -7.50956416e-01 -6.28541827e-01
3.98266673e-01 3.80857550e-02 1.60769016e-01 2.59816259e-01
4.57803488e-01 -1.29006132e-01 -2.22914314e+00 -2.89637566e-01
1.39247298e+00 1.18247306e+00 2.18697116e-01 3.11756432e-01
1.39278352e-01 3.62237453e-01 -5.00151992e-01 -5.55600524e-01
1.65638492e-01 -4.57841575e-01 -4.95169133e-01 -9.00568534e-03
7.94413328e-01 -4.33001459e-01 4.34898674e-01 1.20129216e+00
5.48450828e-01 2.53410637e-01 6.63928628e-01 -1.76889038e+00
-9.82101738e-01 -3.31761390e-02 -1.04349864e+00 9.37065631e-02
5.91088235e-01 1.51894405e-01 6.47468925e-01 -8.04428995e-01
5.76721013e-01 1.08448315e+00 2.73372918e-01 1.13629627e+00
-1.43888438e+00 -1.26006019e+00 4.42099005e-01 2.33405128e-01
-1.39897895e+00 -8.78239989e-01 5.75050771e-01 -3.89408529e-01
4.25866276e-01 -1.02817133e-01 2.12114781e-01 1.35631859e+00
-1.22091368e-01 8.74916017e-01 1.37637401e+00 -5.91142833e-01
2.75667876e-01 5.97248495e-01 1.29967570e-01 3.94714952e-01
4.04981494e-01 2.10359812e-01 -1.24359322e+00 4.05374430e-02
1.82853818e-01 -7.33518302e-02 -6.70397699e-01 -7.29446888e-01
-8.92512560e-01 5.36904454e-01 4.73069698e-01 -4.10294592e-01
-2.42940739e-01 -2.66346782e-01 1.41871208e-03 1.01504065e-01
8.17321301e-01 1.93299636e-01 -3.30009937e-01 -2.75383830e-01
-8.51953328e-01 1.32815331e-01 5.31000733e-01 1.41895843e+00
8.96708965e-01 -2.76632398e-01 -1.43820181e-01 7.32393622e-01
8.88431370e-01 9.41460490e-01 4.87535983e-01 -9.54840004e-01
3.30728352e-01 6.64411843e-01 3.26488495e-01 -5.11894166e-01
-1.63077697e-01 9.90906134e-02 -2.05919206e-01 4.91968125e-01
6.42977297e-01 -3.10240239e-01 -9.45931792e-01 2.25500941e+00
6.72715306e-01 2.00599551e-01 -6.05694950e-02 1.14600241e+00
6.65242314e-01 1.61917657e-01 3.81619960e-01 -4.92442340e-01
1.27186024e+00 -7.55676866e-01 -1.05662763e+00 -4.85059410e-01
3.99908453e-01 -4.33198810e-01 1.44721150e+00 3.47278029e-01
-8.34163845e-01 -3.38492870e-01 -9.46493566e-01 -2.68810183e-01
-1.32741377e-01 9.86939445e-02 2.08708391e-01 9.08114135e-01
-9.83085096e-01 -1.72106531e-02 -6.52755201e-01 -6.33725762e-01
8.99588764e-01 5.30212939e-01 -4.15273458e-01 -1.19816817e-01
-7.83767223e-01 8.33490252e-01 3.70679647e-02 -1.48699597e-01
-6.19524598e-01 -8.80466402e-01 -9.96698678e-01 3.04852389e-02
5.35489738e-01 -5.52075624e-01 1.43184340e+00 -1.31715894e+00
-1.72251308e+00 9.71811414e-01 -9.56896782e-01 -7.17918947e-02
2.56313205e-01 -5.40006042e-01 -3.67560804e-01 -1.45610899e-01
2.09461451e-01 1.02163494e+00 1.31478775e+00 -1.30904186e+00
-7.95662701e-01 -7.88472950e-01 -1.12203673e-01 5.57042420e-01
-4.54179555e-01 1.65441241e-02 -3.45584452e-01 1.52604580e-01
-2.58864194e-01 -9.58949268e-01 4.31942552e-01 3.21676433e-01
-2.92598635e-01 -2.58202046e-01 9.37891066e-01 -3.08701903e-01
1.31688964e+00 -2.36816382e+00 7.44801313e-02 7.87078440e-02
5.36205828e-01 8.05864781e-02 3.17742042e-02 -2.72788197e-01
1.05497383e-01 -7.99189806e-02 -2.20300391e-01 -8.10244441e-01
7.28861839e-02 -1.95733681e-01 -1.54270485e-01 3.35278720e-01
1.52287856e-01 6.23022914e-01 -7.92250216e-01 -6.44823968e-01
-2.37488039e-02 3.39334011e-01 -3.29101980e-01 4.84814554e-01
-1.88958570e-01 6.42609298e-01 -3.47832769e-01 6.73340917e-01
9.71132040e-01 -3.85392457e-01 -7.92028904e-02 1.22403152e-01
-5.03985351e-03 7.10976571e-02 -8.47920775e-01 1.67855263e+00
-2.80176461e-01 8.27797234e-01 -2.94738431e-02 -8.94392133e-02
7.23588705e-01 4.52283621e-02 4.91705276e-02 -8.18597376e-01
3.62089247e-01 -3.11578270e-02 -9.59580466e-02 -5.25180638e-01
5.76811910e-01 2.07904458e-01 1.12154648e-01 7.68325984e-01
1.41639903e-01 2.07231387e-01 6.42823800e-02 1.02130629e-01
5.18592179e-01 5.72117209e-01 4.56628352e-01 -1.88357346e-02
3.79418582e-01 -2.50184089e-01 6.09902740e-01 2.99692035e-01
-8.81090701e-01 5.48140645e-01 4.16666418e-01 2.43761569e-01
-4.56516862e-01 -9.52829361e-01 -1.72992513e-01 1.55529094e+00
4.76192385e-01 -1.58891410e-01 -1.29608154e+00 -1.09543872e+00
-8.32146928e-02 8.91881347e-01 -1.01171720e+00 -1.59914717e-01
9.53736529e-02 -2.04340696e-01 5.74380085e-02 3.10903877e-01
4.09181356e-01 -8.94896567e-01 -8.31832826e-01 -5.18172860e-01
-1.97063074e-01 -8.66132379e-01 -8.16552401e-01 -2.50529110e-01
-6.11203551e-01 -1.27284634e+00 -8.13426256e-01 -2.50255823e-01
8.65551889e-01 5.81417859e-01 9.66270387e-01 -3.83325636e-01
2.10555047e-01 5.49033523e-01 -2.54532963e-01 -1.08160388e+00
1.11372866e-01 1.81949809e-01 2.78301209e-01 2.88373768e-01
1.32838082e+00 -1.89102933e-01 -5.77545285e-01 5.37638009e-01
-5.14033914e-01 -2.40473941e-01 4.10369605e-01 7.39596725e-01
5.82917333e-01 -8.02039266e-01 5.38012862e-01 -9.90900695e-01
7.64311373e-01 -6.48789823e-01 -9.47523117e-01 4.57895935e-01
-1.12414432e+00 4.87604365e-02 -1.89711690e-01 -5.88800192e-01
-1.58991766e+00 3.37988473e-02 6.90095067e-01 -7.19611228e-01
-4.22388136e-01 7.45521188e-02 -3.12498689e-01 -2.53498346e-01
1.03668106e+00 -1.61324799e-01 2.91205883e-01 -2.12157935e-01
1.30332515e-01 1.05521941e+00 1.24867015e-01 -2.36950144e-01
5.03455997e-01 3.34408075e-01 -2.17797190e-01 -5.16657591e-01
-1.09299111e+00 -3.22389781e-01 -5.46004415e-01 -4.68737125e-01
7.71318018e-01 -9.56469119e-01 -1.03730321e+00 5.83913267e-01
-8.66566956e-01 -1.84085518e-01 -1.26361892e-01 5.87522507e-01
-4.33893174e-01 -3.59837373e-04 4.62154806e-01 -1.05136406e+00
-2.03599408e-01 -1.14470804e+00 1.15255892e+00 9.32713449e-01
-3.20709705e-01 -7.13500440e-01 1.10684454e-01 3.14984590e-01
3.72776359e-01 -1.04298726e-01 1.27570182e-01 -6.33050919e-01
-7.04946160e-01 1.34118751e-01 -5.61177313e-01 1.02142312e-01
2.27640167e-01 -7.15804622e-02 -1.44958591e+00 -1.79475501e-01
-1.17570944e-01 -6.90233767e-01 4.28006530e-01 6.78171277e-01
1.04347551e+00 4.74080956e-03 -6.24470770e-01 6.96714222e-01
1.10249722e+00 -1.27775609e-01 3.60106379e-01 2.82279521e-01
5.74701428e-01 9.25250828e-01 1.10769331e+00 4.14500177e-01
7.83086896e-01 4.96110588e-01 5.31885087e-01 3.65569025e-01
3.25918831e-02 -2.68143445e-01 2.83352911e-01 -2.23996460e-01
-1.08491920e-01 -4.53064859e-01 -8.43670964e-01 5.23891211e-01
-1.80407822e+00 -7.23660171e-01 -4.15249020e-02 2.44269443e+00
9.53140497e-01 -2.16866314e-01 2.93174267e-01 -3.65633190e-01
7.31900632e-01 -1.02326795e-01 -1.16469669e+00 -5.86697049e-02
2.94724822e-01 -1.22975886e-01 3.61258864e-01 1.42617077e-01
-8.86178017e-01 8.54645967e-01 6.39208746e+00 3.85709375e-01
-1.22276664e+00 3.26429397e-01 3.83430213e-01 -6.35071754e-01
-8.79257992e-02 -2.61565804e-01 -1.04207826e+00 7.19471991e-01
1.00940251e+00 -2.01476336e-01 4.55470860e-01 1.02976584e+00
4.09804247e-02 -7.57493734e-01 -1.26232505e+00 1.11770201e+00
3.77253681e-01 -5.12372613e-01 -6.70674026e-01 1.93533733e-01
5.20517528e-01 4.83516976e-02 4.74327564e-01 1.69319645e-01
1.71515524e-01 -8.18359673e-01 7.17810750e-01 7.21147299e-01
1.27289212e+00 -4.96510208e-01 4.86026108e-01 1.94287807e-01
-7.00595975e-01 -9.93535444e-02 -1.93683058e-01 1.76954314e-01
7.68203735e-02 -2.49530748e-02 -7.49792635e-01 1.04479477e-01
1.11385036e+00 7.82478034e-01 -6.19563997e-01 9.91487801e-01
-4.52174455e-01 5.36059439e-01 -3.39398742e-01 -1.55868202e-01
-4.09361213e-01 -1.12102538e-01 5.17310679e-01 5.42403996e-01
3.40995044e-01 5.88760935e-02 -5.54175019e-01 1.15860069e+00
-2.16171399e-01 -9.94022116e-02 -5.56866229e-01 3.15306038e-01
9.28803682e-01 9.70720470e-01 4.11724225e-02 1.75460234e-01
-3.73501837e-01 8.00369561e-01 5.26817560e-01 7.03801394e-01
-7.67873883e-01 -3.87668282e-01 1.02496219e+00 1.39195710e-01
1.68437794e-01 3.34984779e-01 -4.40818071e-01 -1.12297964e+00
1.22761019e-01 -6.32565320e-01 2.90974796e-01 -1.15518034e+00
-1.23780394e+00 5.16828239e-01 3.03860426e-01 -1.49314475e+00
-4.35978025e-01 -3.62734884e-01 -2.63122976e-01 1.17164636e+00
-2.03092241e+00 -1.10840225e+00 -7.41641641e-01 8.71219456e-01
2.70156175e-01 -1.75652444e-01 6.94026291e-01 -1.13424100e-01
-7.10518241e-01 1.15490317e+00 7.87355378e-02 -4.74093020e-01
1.51623440e+00 -1.11387658e+00 -1.67796686e-01 8.97021472e-01
-2.83320695e-01 8.08968782e-01 5.59752882e-01 -4.49440598e-01
-9.08168077e-01 -8.03255916e-01 7.70679593e-01 -7.89357126e-01
1.92256317e-01 -4.45418268e-01 -1.07606578e+00 8.02647948e-01
4.97009873e-01 1.40117809e-01 1.01506138e+00 3.80001813e-01
-5.19530535e-01 -2.87782371e-01 -1.46524727e+00 4.35214847e-01
1.10844886e+00 -4.60168928e-01 -4.37697232e-01 -2.63238996e-01
5.94280064e-01 -6.95708036e-01 -4.44266737e-01 3.19709666e-02
7.59197414e-01 -1.20296466e+00 3.59199345e-01 -2.92128503e-01
1.59332007e-01 -3.05134267e-01 2.23206848e-01 -1.46447396e+00
3.84082645e-02 -6.31862283e-01 -4.77138549e-01 1.40541887e+00
3.76574010e-01 -7.86330700e-01 5.69070637e-01 1.32294703e+00
2.50557125e-01 -1.64072528e-01 -7.36112833e-01 -4.45951790e-01
-3.72317165e-01 -2.62295067e-01 8.82673383e-01 6.04602396e-01
5.38425297e-02 3.66842359e-01 -3.95625114e-01 3.17925513e-01
9.53338683e-01 -1.24244541e-01 1.10601485e+00 -1.46927989e+00
4.12050217e-01 -9.53125730e-02 -1.96192697e-01 -1.01446342e+00
4.57022399e-01 5.90223121e-03 4.07775193e-01 -8.26583803e-01
5.02273366e-02 -2.65352428e-01 -3.25605571e-01 3.12188596e-01
-6.13886237e-01 2.59592474e-01 1.59787655e-01 4.80617851e-01
-7.72535443e-01 6.06886625e-01 1.14471149e+00 2.65849322e-01
-4.40324724e-01 2.96127915e-01 -1.09055769e+00 4.94129211e-01
4.33656752e-01 -6.47199333e-01 -9.19237196e-01 -5.12885809e-01
2.38400996e-01 -3.87940824e-01 9.76445228e-02 -9.09813881e-01
4.06924665e-01 -3.12867254e-01 2.26227254e-01 -3.75056684e-01
4.30896759e-01 -1.01452935e+00 -2.63660401e-01 -3.34680796e-01
-4.11227584e-01 -5.55128634e-01 2.61932522e-01 1.02636898e+00
-8.84109512e-02 -1.75740004e-01 8.83362591e-01 5.20402014e-01
-7.45023072e-01 3.52243215e-01 2.18449861e-01 2.14038551e-01
1.37155700e+00 -6.26746595e-01 -5.13476431e-01 -4.74721014e-01
-4.69225436e-01 3.85797173e-01 9.33004856e-01 5.05811512e-01
5.23815274e-01 -1.04840004e+00 -3.54147047e-01 5.72784424e-01
8.68055761e-01 1.64778724e-01 3.24995339e-01 9.46131468e-01
3.18373680e-01 2.10906833e-01 -3.40882182e-01 -9.44601476e-01
-1.55555320e+00 6.26200259e-01 1.77489519e-01 3.84025544e-01
2.92953819e-01 1.13364625e+00 5.57397127e-01 -2.09051147e-01
3.33320707e-01 -1.85613424e-01 -4.74998891e-01 1.95214242e-01
8.22959065e-01 3.63041490e-01 -2.07120031e-01 -7.10025012e-01
-3.75271320e-01 5.62071145e-01 -2.59406328e-01 -7.84288198e-02
1.06251538e+00 -1.00104249e+00 2.05852956e-01 4.95509416e-01
6.69454277e-01 -3.01094097e-03 -1.81359112e+00 -5.53195536e-01
-1.61092758e-01 -9.95214283e-01 1.05528712e-01 -1.01014555e+00
-7.94362187e-01 1.00313449e+00 9.72014725e-01 -1.08411267e-01
1.56027734e+00 -5.54863624e-02 3.77653152e-01 7.28256926e-02
3.44813287e-01 -1.04428172e+00 -1.42451406e-01 1.54531315e-01
6.00135148e-01 -1.97767293e+00 -1.99341953e-01 -3.21761996e-01
-1.23363960e+00 5.58988869e-01 1.27081788e+00 3.73301089e-01
7.34076500e-01 -2.45730579e-01 2.49656156e-01 -8.04588571e-02
-7.67818809e-01 -3.13633591e-01 6.65880859e-01 1.21276760e+00
1.87845439e-01 -3.10893804e-01 1.29550353e-01 5.43872893e-01
1.41722634e-01 4.89318967e-01 3.91543925e-01 8.47126424e-01
-2.47509494e-01 -9.18096006e-01 -5.68113923e-01 4.03769553e-01
-2.79469579e-01 -1.99630782e-02 -3.89497399e-01 7.91065812e-01
5.81328720e-02 1.08749485e+00 2.00014398e-01 -1.93902701e-01
2.31993347e-01 2.23563939e-01 4.62158114e-01 -6.66223824e-01
-1.25324339e-01 -8.93203914e-02 -2.61377335e-01 -8.39604020e-01
-9.51143682e-01 -9.46487427e-01 -7.82521307e-01 -2.40409166e-01
-7.77822673e-01 -8.74337405e-02 7.46454656e-01 9.15335476e-01
1.00203872e+00 1.89282075e-01 5.63241124e-01 -9.30831730e-01
-4.40292746e-01 -1.29009163e+00 -6.73564970e-01 2.50439286e-01
7.33484685e-01 -1.28697598e+00 -4.09603775e-01 3.82303149e-01] | [14.122090339660645, 0.03078402206301689] |
ce12f6c7-bd2b-488b-8935-5e066acc8976 | triplet-based-embedding-distance-and | 1908.02283 | null | https://arxiv.org/abs/1908.02283v1 | https://arxiv.org/pdf/1908.02283v1.pdf | Triplet Based Embedding Distance and Similarity Learning for Text-independent Speaker Verification | Speaker embeddings become growing popular in the text-independent speaker verification task. In this paper, we propose two improvements during the training stage. The improvements are both based on triplet cause the training stage and the evaluation stage of the baseline x-vector system focus on different aims. Firstly, we introduce triplet loss for optimizing the Euclidean distances between embeddings while minimizing the multi-class cross entropy loss. Secondly, we design an embedding similarity measurement network for controlling the similarity between the two selected embeddings. We further jointly train the two new methods with the original network and achieve state-of-the-art. The multi-task training synergies are shown with a 9% reduction equal error rate (EER) and detected cost function (DCF) on the 2016 NIST Speaker Recognition Evaluation (SRE) Test Set. | ['Shugong Xu', 'Zongze Ren', 'Zhiyong Chen'] | 2019-08-06 | null | null | null | null | ['text-independent-speaker-verification'] | ['speech'] | [ 4.27009016e-02 -1.72855392e-01 -8.53441563e-03 -8.04308176e-01
-1.11912608e+00 -3.76782894e-01 6.14580750e-01 -2.99060702e-01
-5.77738583e-01 1.09021120e-01 4.21632797e-01 -3.78288090e-01
-3.53801325e-02 1.42857134e-01 -2.48757109e-01 -9.51643288e-01
-8.19987655e-02 -6.06851056e-02 -3.97642910e-01 7.20816925e-02
1.40988484e-01 3.29669058e-01 -1.31427813e+00 -1.18805729e-01
8.18528533e-01 1.25888848e+00 -1.44240856e-01 5.17244875e-01
2.73886025e-01 2.28062898e-01 -5.87307572e-01 -6.22475445e-01
1.35742575e-01 -3.50020468e-01 -5.77551126e-01 -5.00658900e-02
4.86910373e-01 -1.01514228e-01 -4.70539719e-01 1.07639313e+00
1.04726243e+00 3.70436907e-01 6.15182281e-01 -1.28624249e+00
-6.91813231e-01 6.89602792e-01 -3.26976746e-01 2.65515506e-01
2.56543197e-02 -5.74516952e-02 1.09045827e+00 -1.34686720e+00
3.33073400e-02 1.20499933e+00 6.35647178e-01 9.49480116e-01
-1.07998323e+00 -7.74034262e-01 2.12664574e-01 7.09080219e-01
-1.65075588e+00 -1.05866325e+00 9.99867380e-01 -2.29957253e-01
9.68535841e-01 4.05125231e-01 -2.91545186e-02 1.21834707e+00
-1.70395598e-01 8.66886258e-01 8.15004826e-01 -6.50735319e-01
1.94618031e-01 5.78397095e-01 4.37100261e-01 4.29459274e-01
-3.71395350e-01 4.44870174e-01 -8.19078624e-01 8.80689640e-03
9.81693044e-02 -2.03545883e-01 -5.00797808e-01 -2.55523741e-01
-1.05210936e+00 9.78513479e-01 3.17405641e-01 4.79407132e-01
-6.33666515e-02 -4.02924120e-02 6.07286334e-01 4.46313024e-01
4.94081706e-01 6.37097508e-02 -4.57011193e-01 -1.90836459e-01
-9.93037820e-01 -2.34698728e-01 5.12823105e-01 6.27285421e-01
3.45193118e-01 3.34195852e-01 -2.76424348e-01 1.20505416e+00
9.22030807e-01 3.77901852e-01 8.29292953e-01 -3.28805000e-01
6.18836284e-01 1.55507550e-01 -3.89489740e-01 -5.64686179e-01
-5.60752302e-02 -6.69599950e-01 -7.66995430e-01 -6.52067289e-02
-7.59680867e-02 -2.58692652e-01 -7.65288830e-01 1.86747801e+00
3.05823565e-01 3.14434618e-01 9.30730626e-02 7.45573997e-01
7.94742703e-01 6.64954662e-01 2.84820907e-02 -1.54864058e-01
1.48044968e+00 -1.27716351e+00 -1.14409029e+00 -1.98514685e-01
5.28359711e-01 -7.41675556e-01 1.05783486e+00 5.73853739e-02
-9.39557672e-01 -7.86313772e-01 -1.48159790e+00 1.08613499e-01
-2.53340989e-01 5.26177645e-01 -2.21967623e-01 1.11136031e+00
-1.10515213e+00 4.00861472e-01 -7.65006006e-01 8.33574235e-02
2.62935966e-01 5.38385153e-01 -3.12889218e-01 2.30878338e-01
-1.34105396e+00 1.11660385e+00 -8.94513875e-02 2.73670614e-01
-8.15613568e-01 -8.89703512e-01 -9.73393559e-01 3.18783045e-01
-5.22180535e-02 -2.10328355e-01 1.09159076e+00 -4.90933836e-01
-2.11721444e+00 7.75272906e-01 -2.83284605e-01 -3.59690726e-01
3.62136394e-01 -2.31410131e-01 -7.44828463e-01 -4.02592361e-01
-2.88589001e-01 5.15517294e-01 9.58785236e-01 -9.92755473e-01
-3.04975569e-01 -6.03187680e-01 -6.30670428e-01 1.56695843e-01
-8.69124055e-01 4.91904706e-01 -1.98305771e-01 -6.12089097e-01
-8.08179155e-02 -8.80484879e-01 9.97906774e-02 -1.72522962e-01
-4.67688948e-01 -5.63396811e-01 1.21085727e+00 -9.67780054e-01
1.36324441e+00 -2.72806191e+00 2.70250708e-01 7.44131431e-02
-8.46615657e-02 5.09241283e-01 -3.25120211e-01 5.41074947e-03
-2.58969575e-01 1.35617390e-01 -2.36979261e-01 -1.12270916e+00
4.48720008e-01 -3.30690026e-01 -3.70357573e-01 6.81212246e-01
2.58024514e-01 5.89040518e-01 -2.64386177e-01 -2.77188927e-01
2.32150987e-01 1.02367628e+00 -3.37218732e-01 3.57842118e-01
4.89122003e-01 3.01480770e-01 -1.44510195e-02 3.98652315e-01
8.74480724e-01 3.23985636e-01 -4.19081151e-02 -2.67086774e-01
-1.08153764e-02 7.71009266e-01 -1.07373881e+00 1.89817452e+00
-5.32068014e-01 7.30202913e-01 2.88879246e-01 -9.08553064e-01
8.97076666e-01 7.23569155e-01 1.35616988e-01 -3.09777290e-01
3.06257308e-01 2.25535586e-01 5.71472570e-02 -1.71301737e-01
2.06573486e-01 -3.34581017e-01 2.17525527e-01 4.25657868e-01
3.15242171e-01 1.90595329e-01 -4.58623856e-01 -2.32800096e-01
5.60606480e-01 -2.08551660e-01 1.72048565e-02 -2.61076987e-01
9.70186591e-01 -8.97155404e-01 5.19387424e-01 1.23463146e-01
-6.57227099e-01 5.00605166e-01 1.61379009e-01 9.68915783e-03
-7.31584072e-01 -9.80481207e-01 -6.02944911e-01 9.98723328e-01
-9.63811576e-02 -3.15222144e-01 -7.02919543e-01 -1.03135419e+00
-3.86472382e-02 1.03537667e+00 -6.22069180e-01 -3.23235810e-01
-5.30478299e-01 -5.69460690e-01 7.43339002e-01 5.61380863e-01
4.24448222e-01 -6.61594510e-01 1.51079819e-02 -6.09161183e-02
-1.48830757e-01 -1.21396172e+00 -1.13758600e+00 2.55376637e-01
-6.77154183e-01 -4.24564511e-01 -6.82780683e-01 -9.47700620e-01
3.81954461e-01 4.21859100e-02 5.14866173e-01 -2.66458869e-01
8.19359422e-02 1.68306202e-01 -7.18570501e-02 -4.95880038e-01
-4.51017886e-01 1.95403352e-01 5.57992697e-01 3.95339876e-01
6.78736806e-01 -3.03280205e-01 -4.91031319e-01 5.94574749e-01
-4.23898488e-01 -6.32473528e-01 2.53443033e-01 1.08298826e+00
2.62728184e-01 -3.62858206e-01 7.36335456e-01 7.04331622e-02
5.84084809e-01 -3.64656635e-02 -4.88939077e-01 4.29187924e-01
-8.30325842e-01 2.01768950e-01 3.05860579e-01 -4.79397565e-01
-9.86475527e-01 -6.55088946e-02 -4.13209885e-01 -5.98570347e-01
1.49655668e-02 1.85749739e-01 -5.76860189e-01 -1.22739568e-01
3.55141312e-01 3.90313476e-01 1.28542796e-01 -5.80234170e-01
3.97826761e-01 1.32412589e+00 2.17450693e-01 -2.29491871e-02
8.76305282e-01 -9.08963680e-02 -5.43565273e-01 -8.55511248e-01
-5.08826554e-01 -6.00502133e-01 -5.83066046e-01 4.94221933e-02
7.59364545e-01 -1.01949537e+00 -9.20894504e-01 4.08519894e-01
-1.15713763e+00 7.13533238e-02 -1.87865853e-01 9.76189613e-01
-1.88410878e-01 4.99594688e-01 -3.02485496e-01 -9.30910110e-01
-7.10595608e-01 -1.54827881e+00 1.11326134e+00 6.08973205e-02
-1.54648619e-02 -1.09356439e+00 2.06937477e-01 4.35813993e-01
9.05247808e-01 -4.89850372e-01 6.89090014e-01 -1.05556297e+00
-1.01250283e-01 -3.48266661e-01 -1.09351180e-01 9.73907113e-01
2.18191460e-01 -1.07295118e-01 -1.55164480e+00 -5.96806407e-01
6.74808383e-01 1.20188752e-02 8.63109708e-01 4.13288623e-01
1.01707613e+00 -3.54152441e-01 -2.30534658e-01 7.02215493e-01
1.10471451e+00 2.16953799e-01 5.64907432e-01 8.51444900e-02
5.86275697e-01 5.69309473e-01 1.00261547e-01 1.06798172e-01
2.05629662e-01 1.29533434e+00 -2.74736788e-02 2.28768244e-01
-3.71992797e-01 -1.75180092e-01 8.97338748e-01 1.43701351e+00
3.13952267e-01 -3.67128626e-02 -7.12807178e-01 4.98032987e-01
-1.40788805e+00 -1.05271220e+00 3.87533009e-01 2.47746682e+00
8.27974319e-01 -2.42224187e-01 2.07070298e-02 3.18136394e-01
8.55324090e-01 3.02439332e-01 -4.88751858e-01 -4.44403201e-01
2.14728229e-02 1.50272595e-02 1.43755689e-01 9.14643288e-01
-1.23270845e+00 5.57536066e-01 6.13286495e+00 8.60995471e-01
-1.44634306e+00 4.19134736e-01 5.11594415e-01 -3.10972482e-01
-1.56237051e-01 -3.26627374e-01 -1.18400109e+00 4.40190494e-01
1.43842566e+00 -2.58378148e-01 2.86213756e-01 8.53781223e-01
4.06102948e-02 7.09424317e-01 -1.29233801e+00 1.33381474e+00
5.22618473e-01 -9.69229221e-01 -3.53765875e-01 2.00000152e-01
4.88882035e-01 1.80036455e-01 4.64334935e-01 4.76167291e-01
-2.20123246e-01 -9.32748675e-01 7.40105510e-01 4.02237289e-02
8.94012749e-01 -6.64085388e-01 8.04233909e-01 -1.32535011e-01
-1.10316396e+00 -1.04447566e-01 -8.61642361e-02 5.09640872e-01
3.34050417e-01 4.95245665e-01 -7.02589929e-01 4.41041827e-01
4.62940633e-01 5.35253763e-01 -3.34384382e-01 9.44059968e-01
-3.32229614e-01 7.88155735e-01 -1.12107016e-01 -8.04208815e-02
-2.69029438e-02 -2.30495036e-01 8.50897074e-01 1.19442320e+00
2.31350362e-01 -4.84141409e-01 -4.38566267e-01 7.72691071e-01
-2.85956204e-01 1.42507344e-01 -2.24718451e-01 1.66876569e-01
5.77212036e-01 1.04428744e+00 1.79964930e-01 -2.01457277e-01
-2.53568441e-01 9.67132390e-01 2.28179559e-01 4.06413198e-01
-1.07137704e+00 -6.34142816e-01 1.19436812e+00 -3.37050021e-01
5.35646319e-01 -1.80747584e-01 -4.54456180e-01 -1.25791574e+00
2.93552905e-01 -8.83975923e-01 -1.15442887e-01 -7.08967298e-02
-1.28512073e+00 8.81768584e-01 -3.02081674e-01 -1.14407837e+00
-1.80744588e-01 -5.73251307e-01 -8.31434250e-01 1.09034348e+00
-1.82294095e+00 -8.91740024e-01 1.12542987e-01 3.61722112e-01
6.14871860e-01 -5.78125119e-01 1.05666971e+00 7.40822971e-01
-1.10405123e+00 1.52228725e+00 3.62532735e-01 2.04818353e-01
8.19051862e-01 -1.04454195e+00 7.10553765e-01 8.68604302e-01
3.20391297e-01 7.10433066e-01 3.59265864e-01 1.98498532e-01
-1.31624746e+00 -1.01034439e+00 1.26151228e+00 -4.03037071e-01
4.64317828e-01 -7.21292496e-01 -8.09128106e-01 6.06306732e-01
3.51335734e-01 -5.82989454e-02 1.07792139e+00 2.61634707e-01
-8.04617047e-01 -3.59631926e-01 -1.17912376e+00 2.86636710e-01
6.45613492e-01 -9.63243544e-01 -4.88884628e-01 1.79723144e-01
6.54490709e-01 -1.34129867e-01 -9.01773155e-01 4.52285916e-01
5.21770835e-01 -6.42101288e-01 1.03520095e+00 -6.03823364e-01
-1.01018034e-01 -2.70790994e-01 -3.94216925e-01 -1.56164229e+00
-1.74005792e-01 -6.71286464e-01 -1.41614348e-01 1.49064934e+00
7.51663685e-01 -8.12552869e-01 4.89182264e-01 5.23793638e-01
-3.38724375e-01 -6.84962571e-01 -1.63242769e+00 -9.94711399e-01
-1.99794360e-02 -3.90816510e-01 6.60369813e-01 9.24872100e-01
2.24445343e-01 7.39283442e-01 -3.14968914e-01 2.73632526e-01
7.31734157e-01 -4.26359057e-01 3.94311816e-01 -9.87386703e-01
-9.14470851e-02 -6.51150763e-01 -6.14365160e-01 -1.14511275e+00
3.26806784e-01 -9.85219419e-01 1.44341484e-01 -1.01112473e+00
1.04482226e-01 -2.55681723e-01 -6.30103052e-01 3.89591873e-01
-3.92858446e-01 -1.54937223e-01 2.93867379e-01 6.73419610e-02
-2.00883463e-01 1.24825573e+00 5.16943276e-01 -4.14485216e-01
-2.44215995e-01 1.87603667e-01 -5.79504669e-01 1.90141261e-01
7.63644397e-01 -4.92331982e-01 -1.99204102e-01 -5.91962397e-01
-6.13477767e-01 -3.03152263e-01 6.37779906e-02 -9.41748440e-01
2.89375484e-01 5.19976974e-01 -7.61306435e-02 -3.87912065e-01
5.49163997e-01 -7.23113298e-01 -3.19050670e-01 4.09706205e-01
-5.44156730e-01 -6.78669810e-02 2.93265969e-01 3.92754227e-01
-5.14183164e-01 -1.77764878e-01 1.17127013e+00 8.48288357e-01
-2.87282504e-02 1.80858612e-01 3.95646207e-02 -3.21922481e-01
8.50016832e-01 -6.51519522e-02 -7.73950368e-02 -1.91072017e-01
-5.56419671e-01 1.47390952e-02 -9.09425244e-02 7.77991474e-01
6.59028053e-01 -1.60500264e+00 -1.17492783e+00 5.70032954e-01
1.94298103e-01 -5.56276321e-01 3.19849253e-01 9.76856172e-01
1.61911666e-01 6.95896447e-01 2.18498513e-01 -5.77200830e-01
-1.58214450e+00 3.29117537e-01 6.27993345e-01 -1.79626718e-01
-1.09819837e-01 1.31086302e+00 1.25315517e-01 -8.66036773e-01
7.76351213e-01 -2.42813513e-01 -8.22617039e-02 1.15033977e-01
7.87900686e-01 4.48729485e-01 4.02404308e-01 -8.81458044e-01
-8.27026188e-01 5.75200737e-01 -3.87647688e-01 -3.07651788e-01
1.33455324e+00 -2.64954239e-01 1.83547810e-01 2.90340960e-01
1.84828734e+00 2.17837077e-02 -1.06794405e+00 -4.24380034e-01
-1.00871615e-01 -3.22531015e-01 4.66114700e-01 -7.40628302e-01
-1.14334309e+00 1.15660477e+00 1.25504994e+00 1.11463711e-01
8.11624229e-01 -1.38828844e-01 7.80047357e-01 -1.57870557e-02
-3.56756330e-01 -9.45630491e-01 -1.31936669e-01 2.83735096e-01
1.06663036e+00 -1.38601220e+00 -1.11140534e-01 -1.10432848e-01
-7.85505712e-01 9.43611145e-01 3.06769639e-01 3.98624927e-01
1.09298408e+00 2.80873943e-03 3.84248137e-01 2.58115958e-02
-6.59985363e-01 2.86642551e-01 7.00506866e-01 4.50681716e-01
6.72558427e-01 1.36479348e-01 -2.59742558e-01 5.83372414e-01
-2.83406645e-01 -4.20947880e-01 -5.49814068e-02 2.31691852e-01
-9.91176590e-02 -1.20334840e+00 -3.26891571e-01 -4.24070433e-02
-3.58997077e-01 -1.32281676e-01 -1.80986181e-01 2.17479467e-01
-3.23989540e-01 1.27176416e+00 -7.01059774e-02 -7.92119503e-01
4.57744777e-01 4.27751809e-01 1.54203698e-01 -1.76859111e-01
-6.93030715e-01 -8.98751244e-02 -7.02006668e-02 -2.09741145e-01
-1.64987415e-01 -8.11007082e-01 -8.43860209e-01 -9.41757411e-02
-1.01150882e+00 3.51523727e-01 1.55365992e+00 8.02131653e-01
6.47196352e-01 4.43933785e-01 1.31478953e+00 -5.86173832e-01
-1.32033634e+00 -1.43118644e+00 -4.69257116e-01 1.66706011e-01
7.23600984e-01 -4.88886327e-01 -9.10867870e-01 -1.68710485e-01] | [14.314634323120117, 6.110069751739502] |
d9644260-1b18-4fd5-baad-9838a3d08341 | realtime-global-attention-network-for | 2112.12939 | null | https://arxiv.org/abs/2112.12939v1 | https://arxiv.org/pdf/2112.12939v1.pdf | Realtime Global Attention Network for Semantic Segmentation | In this paper, we proposed an end-to-end realtime global attention neural network (RGANet) for the challenging task of semantic segmentation. Different from the encoding strategy deployed by self-attention paradigms, the proposed global attention module encodes global attention via depth-wise convolution and affine transformations. The integration of these global attention modules into a hierarchy architecture maintains high inferential performance. In addition, an improved evaluation metric, namely MGRID, is proposed to alleviate the negative effect of non-convex, widely scattered ground-truth areas. Results from extensive experiments on state-of-the-art architectures for semantic segmentation manifest the leading performance of proposed approaches for robotic monocular visual perception. | ['Xiangyu Chen', 'Xi Mo'] | 2021-12-24 | null | null | null | null | ['2d-semantic-segmentation'] | ['computer-vision'] | [ 1.64080322e-01 4.50991511e-01 7.40665123e-02 -5.78979135e-01
-6.02723181e-01 -1.55377790e-01 4.15465862e-01 -1.35946929e-01
-4.94471043e-01 4.95562315e-01 4.75169159e-02 6.53834920e-03
-2.52140611e-02 -6.59965038e-01 -1.01551044e+00 -5.90639889e-01
1.32491946e-01 4.16016340e-01 3.15733314e-01 -1.31680384e-01
3.74835789e-01 5.00910401e-01 -1.45990038e+00 3.64994034e-02
1.14805436e+00 1.31656766e+00 5.11011481e-01 5.53682923e-01
1.76281817e-02 9.07531619e-01 -6.66092455e-01 -1.61273956e-01
2.59369373e-01 -2.62144715e-01 -1.05569458e+00 1.15011998e-01
8.37432086e-01 -3.18825871e-01 -3.26890111e-01 1.31990874e+00
3.35382968e-01 3.91789883e-01 3.58229011e-01 -1.18423223e+00
-1.14357054e+00 3.83828253e-01 -5.04031599e-01 3.33039314e-01
7.54649863e-02 2.17474729e-01 1.20950532e+00 -9.05105948e-01
5.27334034e-01 1.63129234e+00 5.29270947e-01 7.35669062e-02
-9.40292120e-01 -3.81526768e-01 6.47654116e-01 3.55516434e-01
-1.21330023e+00 -9.46885571e-02 6.32877827e-01 -4.13023889e-01
1.21221113e+00 -5.35112508e-02 5.25436699e-01 7.19619572e-01
3.07559669e-01 1.02256167e+00 8.00704420e-01 -3.66613455e-02
1.35476723e-01 -3.50321591e-01 1.81442827e-01 7.08908141e-01
2.75266945e-01 -9.68666226e-02 -2.68574387e-01 2.73664981e-01
9.18992937e-01 -1.05483897e-01 -2.12042123e-01 -5.30863106e-01
-9.89036739e-01 7.71518707e-01 1.28769290e+00 2.79133350e-01
-5.71633101e-01 5.14385939e-01 2.52282888e-01 -1.96113005e-01
6.25911593e-01 5.38167298e-01 -6.83055401e-01 3.76960009e-01
-6.93775773e-01 1.46962419e-01 3.21573645e-01 1.20223153e+00
9.27658081e-01 2.73483902e-01 -5.47273517e-01 7.96084166e-01
2.20399141e-01 1.45756990e-01 5.29231787e-01 -1.05099404e+00
3.69568110e-01 9.25079763e-01 4.66219075e-02 -1.12394309e+00
-6.13154113e-01 -8.88184190e-01 -7.83104539e-01 2.05272764e-01
6.02207445e-02 1.45046368e-01 -1.29785502e+00 1.78806019e+00
1.27185509e-01 1.74306571e-01 3.85581553e-02 1.44999862e+00
9.15952265e-01 2.55662233e-01 3.79650563e-01 3.81478399e-01
1.23813701e+00 -1.68520939e+00 -6.12710953e-01 -7.09796071e-01
4.99788788e-04 -4.06028152e-01 9.08394158e-01 9.93873551e-02
-1.28446484e+00 -9.15656328e-01 -1.10489464e+00 -7.14706957e-01
-3.87279093e-01 2.58712322e-01 7.41557479e-01 5.53645082e-02
-1.28715420e+00 3.72642338e-01 -8.07693958e-01 -4.99716341e-01
8.35372210e-01 3.30276817e-01 -1.09327838e-01 -5.30237965e-02
-7.00422406e-01 7.05377758e-01 6.43982470e-01 4.18122917e-01
-1.09500539e+00 -5.66420972e-01 -9.06940401e-01 2.56683856e-01
4.34994698e-01 -9.85355377e-01 1.28039896e+00 -1.04675698e+00
-1.54336679e+00 8.51185083e-01 -5.25642000e-02 -7.85071313e-01
3.76003772e-01 -7.67448723e-01 2.49533623e-01 2.78701991e-01
2.78146178e-01 1.09444082e+00 7.95929730e-01 -1.29500067e+00
-6.25409007e-01 -6.00712061e-01 2.03757688e-01 3.67287695e-01
-3.69109735e-02 -2.06630439e-01 -6.10592604e-01 -6.99651182e-01
4.93102044e-01 -7.84715414e-01 -5.62248409e-01 -1.68773070e-01
-3.94124717e-01 -2.18254030e-01 8.09286892e-01 -8.13881636e-01
6.93364084e-01 -2.03418183e+00 5.80755532e-01 -2.62653112e-01
1.54413477e-01 5.62942885e-02 -1.35289639e-01 -1.29871443e-01
1.40615776e-02 -4.84508127e-02 -6.26104474e-01 -4.64632988e-01
4.34466489e-02 2.33800322e-01 -2.46897191e-01 4.76538956e-01
5.17332315e-01 1.35170758e+00 -9.91807699e-01 -2.76391298e-01
3.08411628e-01 2.76238322e-01 -7.57503927e-01 4.43497032e-01
-5.11129022e-01 5.67485094e-01 -3.28326523e-01 8.71951699e-01
6.61548495e-01 -4.04194832e-01 -1.73542455e-01 -1.35232583e-01
-3.96337286e-02 8.08030367e-02 -6.88427389e-01 2.44617724e+00
-2.88826406e-01 5.57958007e-01 3.21770698e-01 -1.02934182e+00
9.95669901e-01 -6.73756227e-02 3.02947611e-01 -7.82621086e-01
5.15401185e-01 7.51951803e-03 -1.26982242e-01 -2.78861731e-01
8.12742233e-01 4.35387760e-01 -1.18907362e-01 -1.60919607e-01
5.19601643e-01 -3.24439317e-01 -1.05290830e-01 1.00846007e-01
7.86127627e-01 5.27347922e-01 3.55297565e-01 -6.21542752e-01
5.62221289e-01 1.94789335e-01 6.35184765e-01 6.81232393e-01
-2.50154614e-01 8.92411053e-01 3.11416447e-01 -4.58698303e-01
-8.24609935e-01 -8.56384635e-01 2.33005062e-02 1.45105529e+00
6.09950006e-01 -1.12750664e-01 -9.37851965e-01 -4.29573894e-01
3.88073027e-02 5.78000486e-01 -7.58657694e-01 -4.73844334e-02
-4.57983971e-01 -6.37545586e-01 4.22488600e-01 9.91762877e-01
1.03406799e+00 -1.31886089e+00 -1.14652658e+00 3.79811347e-01
-2.23426357e-01 -1.50997043e+00 -3.01175684e-01 3.48834127e-01
-8.21424246e-01 -9.69726205e-01 -6.46329045e-01 -8.38233888e-01
6.05620325e-01 3.38807374e-01 1.17032516e+00 -1.14099212e-01
-3.65286916e-01 3.25660706e-01 -1.78470910e-01 -4.05330092e-01
2.22626880e-01 2.72887468e-01 -5.44574618e-01 3.58609892e-02
1.63790509e-01 -3.49886119e-01 -8.73843074e-01 2.84619063e-01
-7.11304724e-01 4.63989712e-02 6.06659055e-01 7.84884691e-01
7.58476436e-01 -4.20653105e-01 5.03156841e-01 -5.41863978e-01
3.72943461e-01 -4.68727052e-01 -8.20470989e-01 -1.58878472e-02
-3.05601478e-01 9.37022641e-02 4.34419632e-01 -2.00502966e-02
-1.16012931e+00 -4.43093963e-02 -7.58870095e-02 -5.81941366e-01
-2.38709509e-01 3.23456168e-01 -1.21305250e-01 -1.92236036e-01
3.85979861e-01 -7.24117877e-03 -1.63010880e-01 -4.20648575e-01
5.97241819e-01 2.82011479e-01 1.05307150e+00 -4.84854490e-01
2.93618351e-01 4.77143109e-01 -1.46506846e-01 -6.95809782e-01
-1.02656627e+00 -5.02257466e-01 -7.53330052e-01 1.33603960e-01
1.51003897e+00 -1.23180830e+00 -6.76394701e-01 8.04709375e-01
-1.28025806e+00 -5.51537812e-01 -2.35674053e-01 2.85617393e-02
-8.12724710e-01 2.01508582e-01 -5.17376602e-01 -6.68990672e-01
-4.70963180e-01 -1.24273336e+00 1.44967341e+00 4.13595885e-01
-6.90451190e-02 -8.52827430e-01 -3.01837772e-01 5.49599886e-01
5.64054489e-01 3.29351515e-01 6.59187257e-01 -4.43194270e-01
-1.19301450e+00 1.79546088e-01 -7.54804075e-01 4.22866404e-01
-1.59287766e-01 -3.82904708e-01 -1.17338264e+00 -3.26953501e-01
-3.88450623e-02 -4.30286288e-01 1.09851837e+00 6.71955526e-01
1.53152776e+00 -4.83408943e-02 -1.43603474e-01 1.16188252e+00
1.55523038e+00 2.08375037e-01 7.93316960e-01 4.51065242e-01
1.15680218e+00 3.62709910e-01 5.93381464e-01 2.27588311e-01
6.40502334e-01 5.80585957e-01 9.25840080e-01 -4.19188291e-01
-4.10151213e-01 -6.11110181e-02 -9.09833238e-02 2.83695638e-01
-1.20797111e-02 -3.00397217e-01 -9.22894478e-01 8.23681474e-01
-2.17383909e+00 -5.08505821e-01 -5.03478497e-02 1.68626916e+00
2.34700263e-01 8.66843760e-02 -2.16957450e-01 -3.89088869e-01
4.85555410e-01 2.06703231e-01 -7.92248428e-01 -5.03696442e-01
-3.72339338e-01 2.07537338e-01 7.37479627e-01 5.10080993e-01
-1.37262380e+00 1.56046104e+00 6.50468302e+00 5.63571393e-01
-1.07235229e+00 2.99658805e-01 6.50903702e-01 2.29860127e-01
8.62498656e-02 -2.24076271e-01 -5.73226929e-01 8.50661322e-02
4.39823627e-01 1.09364927e-01 3.88856351e-01 1.01710784e+00
-2.30523095e-01 -1.90293223e-01 -7.75925219e-01 8.33493233e-01
2.74341971e-01 -1.02432597e+00 3.69646922e-02 -1.35567978e-01
1.04761088e+00 4.20790464e-01 1.17505506e-01 2.54267156e-01
6.05882883e-01 -1.13402224e+00 9.68863070e-01 4.10224408e-01
5.90830803e-01 -5.60503781e-01 8.40705335e-01 5.16832247e-02
-1.25875092e+00 -5.74582577e-01 -5.08421004e-01 -9.24905464e-02
5.37138246e-02 2.67352819e-01 -5.92658639e-01 7.68846273e-01
1.00652444e+00 7.56544769e-01 -8.38776231e-01 9.43805933e-01
-4.98195171e-01 2.15405926e-01 -1.84245467e-01 2.79853314e-01
7.39787281e-01 -1.65195301e-01 5.75193107e-01 1.16731906e+00
4.02886681e-02 2.27132365e-01 2.00686529e-01 1.15392470e+00
-1.22474425e-01 -1.38965189e-01 -3.65356177e-01 2.03762114e-01
4.02459763e-02 1.28835857e+00 -7.45212972e-01 -4.54790235e-01
-3.53448421e-01 1.17644715e+00 7.67787397e-01 6.10483527e-01
-8.52983117e-01 -2.26106361e-01 9.25874472e-01 -2.95702726e-01
6.29753053e-01 -4.16675419e-01 -8.15492332e-01 -8.95188272e-01
-9.58038867e-02 -5.81662059e-01 4.05043781e-01 -1.10230827e+00
-1.10080516e+00 8.70687366e-01 -8.15289319e-02 -6.55684769e-01
6.94119260e-02 -6.26638889e-01 -5.09065449e-01 8.65810394e-01
-1.60651124e+00 -1.30402493e+00 -9.49670792e-01 4.57199425e-01
7.87445486e-01 -1.57970622e-01 5.80248415e-01 -4.03732881e-02
-4.71609592e-01 4.67024177e-01 -4.53205377e-01 -3.67434770e-02
2.70360321e-01 -1.46400523e+00 6.77310586e-01 1.05177474e+00
-1.19650364e-01 4.57118481e-01 5.90763807e-01 -5.53002357e-01
-1.03800249e+00 -1.36091554e+00 3.48899722e-01 -3.35787714e-01
5.01617849e-01 -1.49170846e-01 -1.07457113e+00 8.52690101e-01
5.70992231e-01 6.76028281e-02 -1.38175666e-01 -4.12760349e-03
-2.62354672e-01 1.01719722e-02 -1.12930095e+00 4.29634392e-01
1.41752601e+00 -2.92874277e-01 -7.56657004e-01 1.58541560e-01
1.14059865e+00 -6.42392039e-01 -6.37173772e-01 6.49975777e-01
1.85420901e-01 -1.00625873e+00 1.09963858e+00 -5.26778579e-01
5.13376713e-01 -3.03253621e-01 -3.91359001e-01 -1.21299708e+00
-6.70592606e-01 -3.37178826e-01 2.23038569e-02 9.37066853e-01
7.01653808e-02 -6.55169904e-01 6.69507742e-01 2.45592356e-01
-7.95145690e-01 -6.54644668e-01 -9.14202511e-01 -5.74075282e-01
-4.41478826e-02 -5.04907221e-02 5.36911666e-01 6.12184346e-01
-5.25661349e-01 3.17556381e-01 1.70182493e-02 3.97846133e-01
7.01272070e-01 2.67882377e-01 7.50536263e-01 -1.12034249e+00
-1.25708953e-01 -5.29329062e-01 -6.67905390e-01 -1.41746581e+00
3.37982297e-01 -8.25619400e-01 5.09241998e-01 -1.87548125e+00
1.60818875e-01 -9.37723219e-02 -5.97902417e-01 4.64619249e-01
-3.90094608e-01 4.68054235e-01 2.33621612e-01 -9.48652700e-02
-8.69081259e-01 9.89450157e-01 1.31428826e+00 -1.84243441e-01
-1.84650972e-01 -5.28274119e-01 -7.75338888e-01 8.37855816e-01
7.90416360e-01 -1.93992034e-01 -3.66926283e-01 -9.00537372e-01
-1.41053647e-01 -2.22610295e-01 7.09146023e-01 -1.16364300e+00
3.74218225e-01 -5.19449310e-03 2.56487429e-01 -8.11312854e-01
4.14235890e-01 -8.08518708e-01 -2.69765943e-01 2.32846290e-01
-1.73394695e-01 1.46522671e-01 4.00391966e-01 6.08685791e-01
-4.99854118e-01 2.82692790e-01 8.64474893e-01 -2.01783270e-01
-1.13575792e+00 3.25826883e-01 -3.39794904e-02 2.47862428e-01
9.52208161e-01 -1.25297934e-01 -4.09514517e-01 4.43958007e-02
-6.81351244e-01 3.52761000e-01 3.19490463e-01 6.70457006e-01
6.31087899e-01 -1.08151007e+00 -6.07625961e-01 2.28024989e-01
1.43040612e-01 5.45664191e-01 2.79255927e-01 6.49149776e-01
-7.31916606e-01 6.87052190e-01 -5.85931838e-01 -9.58117247e-01
-7.90957451e-01 4.92314607e-01 4.16349739e-01 -3.17346156e-02
-6.42219901e-01 1.20679510e+00 6.92728281e-01 -4.33027178e-01
2.19525710e-01 -7.40720153e-01 -4.86778766e-02 -1.00347243e-01
3.58195752e-02 2.92135596e-01 9.57161188e-02 -5.66458642e-01
-3.89843166e-01 6.32339239e-01 1.29248068e-01 1.00516275e-01
1.34401357e+00 -3.66902053e-01 -2.05228403e-01 1.90042123e-01
8.93223524e-01 -5.53436756e-01 -1.76182473e+00 -9.94063318e-02
-8.07030946e-02 -3.52847099e-01 2.64951885e-01 -9.34012532e-01
-1.20581400e+00 8.10603738e-01 4.33101624e-01 -1.01834081e-01
1.26536000e+00 5.47578856e-02 6.76075757e-01 1.96393207e-01
4.91350472e-01 -1.01777864e+00 1.50575593e-01 8.39021325e-01
1.32377756e+00 -1.27320838e+00 -9.98968184e-02 -5.30340612e-01
-6.64797008e-01 7.06203461e-01 1.11159551e+00 -4.93847847e-01
2.48081923e-01 -5.17373607e-02 -2.58124620e-02 -2.51465797e-01
-5.35653114e-01 -5.21583915e-01 4.67530459e-01 4.71346796e-01
1.29319012e-01 5.60967959e-02 -6.68094009e-02 6.64822280e-01
-1.76798016e-01 -1.25183031e-01 1.69215932e-01 5.52552938e-01
-4.20833886e-01 -2.73214281e-01 -2.37019375e-01 -4.14150581e-02
-3.25715035e-01 -2.95692384e-02 -2.80287147e-01 8.49027097e-01
2.88764328e-01 8.52712929e-01 3.48179519e-01 -8.86414871e-02
3.26340884e-01 -7.03852922e-02 4.42975551e-01 -5.82782745e-01
-6.97139204e-01 3.91777717e-02 -1.50164157e-01 -7.99885273e-01
-3.61995876e-01 -3.50709379e-01 -1.38839698e+00 2.34368742e-01
-1.81933790e-02 -4.22628880e-01 6.11925185e-01 8.99810851e-01
5.30649364e-01 1.15385890e+00 4.23898362e-02 -1.15599298e+00
-2.24299043e-01 -1.22210407e+00 -2.39815697e-01 5.13251662e-01
3.20676357e-01 -6.83741629e-01 -3.16629075e-02 -3.71336117e-02] | [9.525145530700684, 0.18217137455940247] |
8c57d806-dd47-444a-afcc-7aab5b178870 | geometric-disentanglement-for-generative | 1908.06386 | null | https://arxiv.org/abs/1908.06386v1 | https://arxiv.org/pdf/1908.06386v1.pdf | Geometric Disentanglement for Generative Latent Shape Models | Representing 3D shape is a fundamental problem in artificial intelligence, which has numerous applications within computer vision and graphics. One avenue that has recently begun to be explored is the use of latent representations of generative models. However, it remains an open problem to learn a generative model of shape that is interpretable and easily manipulated, particularly in the absence of supervised labels. In this paper, we propose an unsupervised approach to partitioning the latent space of a variational autoencoder for 3D point clouds in a natural way, using only geometric information. Our method makes use of tools from spectral differential geometry to separate intrinsic and extrinsic shape information, and then considers several hierarchical disentanglement penalties for dividing the latent space in this manner, including a novel one that penalizes the Jacobian of the latent representation of the decoded output with respect to the latent encoding. We show that the resulting representation exhibits intuitive and interpretable behavior, enabling tasks such as pose transfer and pose-aware shape retrieval that cannot easily be performed by models with an entangled representation. | ['Tristan Aumentado-Armstrong', 'Stavros Tsogkas', 'Sven Dickinson', 'Allan Jepson'] | 2019-08-18 | geometric-disentanglement-for-generative-1 | http://openaccess.thecvf.com/content_ICCV_2019/html/Aumentado-Armstrong_Geometric_Disentanglement_for_Generative_Latent_Shape_Models_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Aumentado-Armstrong_Geometric_Disentanglement_for_Generative_Latent_Shape_Models_ICCV_2019_paper.pdf | iccv-2019-10 | ['3d-shape-generation', 'pose-transfer', '3d-shape-representation', '3d-object-retrieval'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 3.53239059e-01 3.64139497e-01 2.14282826e-01 -3.20636332e-01
-3.54348749e-01 -9.32628393e-01 9.02718425e-01 -6.70373440e-02
-9.66182277e-02 2.70010084e-01 1.84209004e-01 -1.32644489e-01
-4.23212826e-01 -9.51273680e-01 -6.81526661e-01 -1.04275990e+00
2.25260824e-01 8.39942813e-01 -2.86614299e-01 -4.05643955e-02
2.90216450e-02 9.46717858e-01 -1.55051494e+00 -1.20927371e-01
6.44831717e-01 7.23829031e-01 7.40868226e-02 6.09989166e-01
2.01928560e-02 1.34048149e-01 -3.33693206e-01 -2.05030188e-01
3.18381488e-01 -3.02445352e-01 -5.28002381e-01 5.02555370e-01
2.67142415e-01 -8.93726051e-02 -1.72876522e-01 9.16899681e-01
3.25416982e-01 1.37407601e-01 1.19961941e+00 -1.10929847e+00
-6.93829060e-01 7.67408386e-02 -1.49147362e-01 -1.92999318e-01
9.65358093e-02 -4.23334613e-02 1.09515631e+00 -7.13341177e-01
7.46932447e-01 1.37328517e+00 8.92821029e-02 4.00245488e-01
-1.74314582e+00 -2.00249791e-01 -1.79234728e-01 -4.54925597e-01
-1.44099498e+00 -2.88304538e-01 1.13319528e+00 -8.45377386e-01
7.45478809e-01 2.96367377e-01 7.24462092e-01 9.19398665e-01
-3.87660116e-02 6.69659138e-01 9.64137137e-01 -4.94632065e-01
3.96274984e-01 2.67876118e-01 -2.31695578e-01 7.48372853e-01
2.28029370e-01 7.91928098e-02 -4.06841725e-01 -2.92295367e-01
1.21675158e+00 -5.31119760e-03 -2.09691167e-01 -1.40058577e+00
-1.10543239e+00 1.15400922e+00 4.44953084e-01 2.44587809e-01
-2.36987129e-01 2.13849917e-01 -2.69561201e-01 1.61204576e-01
6.43402755e-01 4.84177977e-01 -7.87850171e-02 3.49473432e-02
-8.39881778e-01 1.55914634e-01 9.48615670e-01 7.94065475e-01
8.71032596e-01 6.65980205e-02 2.37575918e-01 5.08880615e-01
7.11778879e-01 6.45892143e-01 2.42897030e-02 -1.03884315e+00
6.15332611e-02 6.21905267e-01 3.41125624e-03 -1.07075298e+00
-1.54064186e-02 -3.90580416e-01 -7.82555044e-01 4.97780204e-01
2.12387636e-01 9.03896540e-02 -8.01365972e-01 1.74375021e+00
2.84761250e-01 -1.65530771e-01 -1.27135217e-02 9.50470686e-01
4.91925925e-01 4.99122143e-01 -1.72952235e-01 -5.51001355e-02
9.29946482e-01 -2.97704518e-01 -2.47904718e-01 -2.93928683e-02
2.37946287e-01 -6.67681336e-01 6.29116952e-01 5.91810383e-02
-1.06530702e+00 -1.86622798e-01 -1.07069778e+00 -3.11237514e-01
-1.92086220e-01 1.17332108e-01 6.65392756e-01 5.64952672e-01
-1.11463106e+00 6.80611312e-01 -1.10799122e+00 -3.47621590e-01
1.88008547e-01 4.97422874e-01 -4.93209690e-01 2.66946852e-01
-6.59844577e-01 7.23295808e-01 2.88747489e-01 2.35500205e-02
-5.82002938e-01 -2.99296916e-01 -9.34914112e-01 9.80304033e-02
2.48402625e-01 -1.09712541e+00 8.24481905e-01 -7.37603843e-01
-1.74653423e+00 9.82016683e-01 -7.14865997e-02 -6.20147027e-02
4.83183235e-01 5.60017638e-02 3.08591276e-01 8.50600675e-02
-9.68540013e-02 8.55078518e-01 1.26999331e+00 -1.41709888e+00
3.03964436e-01 -6.80057168e-01 2.84966499e-01 3.23546112e-01
-1.75903648e-01 -6.26144528e-01 -2.99870670e-01 -5.40960252e-01
5.65440953e-01 -1.32818162e+00 -2.56031275e-01 3.54861170e-01
-4.18193638e-01 -5.22364378e-02 6.37245834e-01 -2.39682078e-01
5.38604617e-01 -2.25696707e+00 1.05744886e+00 4.37029421e-01
4.59241569e-01 -1.84781432e-01 4.35748622e-02 4.16721135e-01
-3.02505270e-02 9.93432924e-02 -4.95062053e-01 -4.15226042e-01
2.86824256e-01 4.88932073e-01 -5.24262488e-01 5.84705651e-01
3.90899211e-01 9.51254308e-01 -7.82344520e-01 -1.83367744e-01
3.42636704e-01 8.77149999e-01 -9.34817374e-01 1.32841378e-01
-3.69688898e-01 7.86017239e-01 -6.67771637e-01 1.77502796e-01
4.78925020e-01 -3.31567854e-01 9.93596315e-02 -2.52438813e-01
-1.31658986e-01 2.57822573e-01 -1.21706009e+00 1.88934600e+00
-4.30548221e-01 5.28612971e-01 7.17497393e-02 -9.20515120e-01
9.30509210e-01 2.84136206e-01 6.13578796e-01 -1.83981750e-02
2.63617516e-01 2.44395673e-01 -1.64648533e-01 -1.87402859e-01
5.06895840e-01 -3.58379811e-01 -1.13743827e-01 5.72036326e-01
2.42649302e-01 -4.42291528e-01 -2.60722041e-01 2.42211930e-02
6.75634027e-01 3.25146705e-01 1.55419856e-01 -1.31721556e-01
2.61666179e-01 -3.26832265e-01 9.81171895e-03 4.73585129e-01
3.73849690e-01 7.85002470e-01 3.92124891e-01 -1.33452877e-01
-1.17126322e+00 -1.45923913e+00 -1.28005341e-01 5.33451200e-01
-2.47847270e-02 -1.21587977e-01 -6.75301313e-01 -3.29487681e-01
1.57745212e-01 6.28875077e-01 -4.96774793e-01 -3.04619342e-01
-3.57135803e-01 -4.58792716e-01 3.27936321e-01 2.77015060e-01
1.26585527e-03 -8.01021636e-01 -7.77625620e-01 2.83356868e-02
1.07385822e-01 -9.22026873e-01 -1.31234214e-01 2.38342211e-01
-1.16605675e+00 -6.67818666e-01 -7.75629878e-01 -3.07571650e-01
8.99149895e-01 1.50495827e-01 9.39720333e-01 -3.23211670e-01
-3.69334280e-01 8.92171085e-01 -3.69350314e-02 -1.10893913e-01
-4.69288945e-01 1.18101589e-01 -1.21511500e-02 3.25537950e-01
1.10529453e-01 -9.36564505e-01 -3.49509001e-01 4.13833745e-02
-1.19862390e+00 2.27354273e-01 6.14308655e-01 7.21760690e-01
7.12109566e-01 -4.34732974e-01 -8.98407493e-03 -5.78742564e-01
4.61364925e-01 -3.74636590e-01 -6.65569961e-01 4.91892397e-02
-2.81050503e-01 7.78115988e-01 2.17936128e-01 -2.56174475e-01
-8.62593055e-01 3.45324904e-01 -6.37269393e-02 -7.05081224e-01
-2.27679864e-01 3.06101441e-01 -4.21348155e-01 -2.17743248e-01
4.46699053e-01 2.98224419e-01 2.27510497e-01 -6.44171238e-01
7.65264034e-01 3.22235614e-01 3.58367920e-01 -5.90219855e-01
1.02768612e+00 6.49490416e-01 4.63751048e-01 -1.18208754e+00
-5.11555791e-01 -3.57834518e-01 -8.84585917e-01 -5.44157960e-02
1.07161856e+00 -6.11929417e-01 -6.75711215e-01 9.66802761e-02
-1.16389608e+00 1.39082178e-01 -5.57717741e-01 3.77011001e-01
-9.69827116e-01 3.09728503e-01 -2.75721759e-01 -7.86081076e-01
1.29422285e-02 -1.25776470e+00 1.47101009e+00 -1.04594670e-01
-3.17162782e-01 -1.03449905e+00 1.65589139e-01 1.67325333e-01
2.57405430e-01 4.24497455e-01 1.27788985e+00 -3.55471283e-01
-1.04217041e+00 -1.37022808e-02 1.80188864e-02 2.18247458e-01
9.02104676e-02 -6.55372888e-02 -1.01635253e+00 -3.54704529e-01
2.71026164e-01 -1.16281562e-01 8.71049583e-01 3.27050477e-01
8.06578100e-01 -2.26659581e-01 -1.79495648e-01 6.95815384e-01
1.38780200e+00 -1.22559503e-01 4.08478528e-01 -1.34477958e-01
8.53207886e-01 5.67256331e-01 -2.29780361e-01 2.31527165e-01
1.92403898e-01 8.34600747e-01 5.55132806e-01 9.60197076e-02
2.29522184e-01 -3.06524813e-01 8.77088681e-02 9.77090299e-01
-3.77572954e-01 -1.72656000e-01 -6.22303903e-01 2.57910311e-01
-1.63221192e+00 -7.24504828e-01 1.83535621e-01 2.25980449e+00
4.33468521e-01 -2.64276061e-02 -2.26318538e-01 2.34598756e-01
3.25089306e-01 2.05616385e-01 -5.84119618e-01 -3.77562553e-01
-8.43540207e-03 3.42924893e-01 2.28195682e-01 5.41522264e-01
-1.03400600e+00 6.23996973e-01 5.79801178e+00 3.94914091e-01
-1.10826385e+00 -8.76399800e-02 1.77586526e-01 2.17320859e-01
-8.44546258e-01 2.12802857e-01 -3.65289450e-01 9.12524983e-02
5.70957363e-01 -7.53807696e-03 4.53348279e-01 7.30829358e-01
-9.08094794e-02 1.27135843e-01 -1.40523958e+00 8.72809529e-01
1.12870321e-01 -1.16003382e+00 4.48314905e-01 5.11526167e-01
5.99848926e-01 -1.87422216e-01 4.46534246e-01 -1.76675737e-01
3.38374078e-02 -9.37933147e-01 7.72566855e-01 6.94460511e-01
6.53483272e-01 -4.97458726e-01 1.64862815e-02 5.37160218e-01
-8.21774781e-01 3.00687373e-01 -2.17465878e-01 -4.79277549e-03
4.05186042e-02 4.11810309e-01 -8.48545372e-01 4.46555465e-01
9.56109837e-02 5.45517504e-01 -3.59193772e-01 9.52150941e-01
-2.07665414e-01 3.27472180e-01 -5.53038299e-01 2.70820428e-02
4.60266136e-02 -6.02001369e-01 1.09838068e+00 6.53436780e-01
5.48638761e-01 -1.95928589e-02 1.22872323e-01 1.37398744e+00
-1.14189484e-03 -7.76879489e-02 -9.98849332e-01 -2.91769594e-01
6.40011281e-02 1.05663323e+00 -9.09015000e-01 6.21926673e-02
-4.75093024e-03 1.04844832e+00 1.94529399e-01 3.94995272e-01
-3.82899553e-01 2.43130192e-01 5.33074081e-01 1.37015000e-01
4.04611558e-01 -7.87527919e-01 -2.26011962e-01 -1.23773193e+00
3.34651768e-02 -2.56504744e-01 -1.02560788e-01 -8.49677563e-01
-1.04562712e+00 4.64978695e-01 3.40250582e-01 -1.30621958e+00
-6.65891647e-01 -6.35924280e-01 -1.56977832e-01 9.15230334e-01
-1.09363663e+00 -1.14487362e+00 -8.84582773e-02 3.36593151e-01
1.30725980e-01 1.01825178e-01 1.06188381e+00 1.29429894e-02
-9.45463106e-02 2.10208252e-01 1.73626348e-01 -1.09093875e-01
2.23503679e-01 -1.30659270e+00 2.24140942e-01 6.16274059e-01
6.92496896e-01 8.24321806e-01 8.73636425e-01 -2.95718640e-01
-1.73399067e+00 -8.74394536e-01 6.50629461e-01 -6.48214281e-01
4.08923030e-01 -4.46172148e-01 -8.03394854e-01 6.68273628e-01
-1.67686611e-01 -1.66467369e-01 7.00150788e-01 -5.82132377e-02
-5.06866455e-01 1.87343732e-01 -8.11124325e-01 5.27227938e-01
9.23443735e-01 -7.94691861e-01 -5.30445278e-01 7.18711242e-02
5.90766788e-01 -4.07559305e-01 -6.73371017e-01 1.93473056e-01
5.04717231e-01 -7.26820588e-01 1.14128268e+00 -4.51992452e-01
3.73448402e-01 -3.22134048e-01 -2.68414676e-01 -1.23516250e+00
-2.71939158e-01 -5.30300558e-01 -1.25538513e-01 8.89699578e-01
2.30172634e-01 -5.02391875e-01 8.36067975e-01 5.13637602e-01
3.31287570e-02 -6.64816082e-01 -9.63644147e-01 -6.27392173e-01
1.70201585e-01 -2.93248177e-01 3.46183360e-01 7.09223807e-01
-5.20688772e-01 5.34082651e-01 -1.88522965e-01 1.99826598e-01
7.61193156e-01 5.34309983e-01 6.79890871e-01 -1.58158672e+00
-4.68891233e-01 -4.88599360e-01 -7.97711670e-01 -1.34981692e+00
2.48039141e-01 -1.30299973e+00 -6.86463341e-02 -1.35593522e+00
8.33460838e-02 -4.07290757e-01 9.28197205e-02 2.86688775e-01
4.87258911e-01 3.45532626e-01 3.04854482e-01 3.51546645e-01
-1.35102421e-01 9.47897077e-01 1.23647106e+00 -2.17001289e-01
-1.86478913e-01 -3.29343490e-02 -5.00512779e-01 8.75494003e-01
3.79175514e-01 -3.91839623e-01 -4.36492413e-01 -6.52991116e-01
4.22062814e-01 9.22127664e-02 8.07153940e-01 -6.76639378e-01
1.58832565e-01 -3.56118493e-02 2.89602369e-01 -4.62953001e-01
7.74840117e-01 -9.15119827e-01 5.56811392e-01 5.98583855e-02
-2.72408605e-01 -1.51748627e-01 -1.28582299e-01 7.43844569e-01
-1.07026912e-01 -3.03562075e-01 5.49541295e-01 -1.30511597e-01
-2.65012413e-01 3.16668034e-01 -2.40204260e-01 -1.90151080e-01
6.97159648e-01 -2.98070431e-01 1.70478225e-01 -4.27368760e-01
-1.06363022e+00 -2.49840632e-01 7.52279222e-01 3.59654844e-01
5.65448403e-01 -1.53226078e+00 -4.43336546e-01 4.59223121e-01
1.39463514e-01 1.01398230e-01 1.40523463e-01 5.90252578e-01
-4.46641356e-01 5.21443546e-01 -1.46616355e-01 -1.01383781e+00
-1.02097762e+00 3.29090476e-01 2.82357275e-01 -1.26047030e-01
-6.85790896e-01 5.70302904e-01 3.95441920e-01 -5.05351305e-01
-7.02987462e-02 -3.58439654e-01 -5.57011627e-02 4.28359322e-02
-7.93274790e-02 1.03225559e-01 4.41037752e-02 -1.15787411e+00
-1.83347359e-01 8.23521614e-01 4.34496164e-01 -3.85945559e-01
1.46949565e+00 -1.21615725e-02 -2.19860986e-01 6.29596055e-01
1.42535782e+00 -1.07985940e-02 -1.26934457e+00 -9.43168402e-02
-2.05257192e-01 -4.15020078e-01 3.23184431e-02 -2.69960403e-01
-8.57735813e-01 1.09061337e+00 4.14682776e-01 4.10984546e-01
8.22994232e-01 2.98503041e-01 3.81703407e-01 3.82553697e-01
4.73146379e-01 -5.55612028e-01 7.88387805e-02 5.35441399e-01
1.09347522e+00 -9.60822523e-01 -1.15771741e-01 -5.25704920e-01
-1.64436668e-01 1.05702484e+00 -6.94382861e-02 -3.24035525e-01
8.52847934e-01 -1.73594337e-02 -3.47383320e-01 -3.80916655e-01
-5.06483316e-01 -2.29913622e-01 6.94587648e-01 3.67721498e-01
2.58309454e-01 3.50968659e-01 1.07999472e-02 2.08109856e-01
-4.18228149e-01 -2.62535989e-01 3.10343057e-01 8.12965155e-01
-1.53732851e-01 -1.32301044e+00 -2.17710212e-01 2.08713114e-01
-1.28450632e-01 1.69370160e-01 -5.51734269e-01 4.61733460e-01
6.29613474e-02 3.43320608e-01 1.11104473e-01 -1.38665721e-01
1.47319719e-01 1.93052769e-01 9.04771805e-01 -7.18123257e-01
1.53868273e-01 3.08691055e-01 -3.57882142e-01 -1.69194683e-01
-5.86442411e-01 -7.63227224e-01 -9.83939767e-01 2.28125930e-01
-3.19976836e-01 5.56981079e-02 8.68468821e-01 9.86712515e-01
3.69262218e-01 3.25289220e-01 5.49114525e-01 -1.27676523e+00
-5.70216537e-01 -4.98916000e-01 -5.65195084e-01 4.28733557e-01
4.82433289e-01 -1.03522980e+00 -4.49608773e-01 7.22705275e-02] | [8.821550369262695, -3.3199331760406494] |
f4587095-a1cc-4e36-85b2-3124aa807b94 | posematcher-one-shot-6d-object-pose | 2304.01382 | null | https://arxiv.org/abs/2304.01382v1 | https://arxiv.org/pdf/2304.01382v1.pdf | PoseMatcher: One-shot 6D Object Pose Estimation by Deep Feature Matching | Estimating the pose of an unseen object is the goal of the challenging one-shot pose estimation task. Previous methods have heavily relied on feature matching with great success. However, these methods are often inefficient and limited by their reliance on pre-trained models that have not be designed specifically for pose estimation. In this paper we propose PoseMatcher, an accurate model free one-shot object pose estimator that overcomes these limitations. We create a new training pipeline for object to image matching based on a three-view system: a query with a positive and negative templates. This simple yet effective approach emulates test time scenarios by cheaply constructing an approximation of the full object point cloud during training. To enable PoseMatcher to attend to distinct input modalities, an image and a pointcloud, we introduce IO-Layer, a new attention layer that efficiently accommodates self and cross attention between the inputs. Moreover, we propose a pruning strategy where we iteratively remove redundant regions of the target object to further reduce the complexity and noise of the network while maintaining accuracy. Finally we redesign commonly used pose refinement strategies, zoom and 2D offset refinements, and adapt them to the one-shot paradigm. We outperform all prior one-shot pose estimation methods on the Linemod and YCB-V datasets as well achieve results rivaling recent instance-level methods. The source code and models are available at https://github.com/PedroCastro/PoseMatcher. | ['Tae-Kyun Kim', 'Pedro Castro'] | 2023-04-03 | null | null | null | null | ['6d-pose-estimation'] | ['computer-vision'] | [ 8.23244825e-03 -1.86768517e-01 8.44454840e-02 -3.08130354e-01
-1.04297757e+00 -6.29348636e-01 6.51380658e-01 -2.37844616e-01
-4.23368782e-01 9.29920450e-02 -2.35907003e-01 2.05775425e-01
-6.71262443e-02 -5.39324999e-01 -1.07262015e+00 -3.47404301e-01
4.97954965e-01 9.40152466e-01 6.10418439e-01 -1.02300383e-01
3.76130670e-01 8.79505873e-01 -1.88847983e+00 6.05160184e-02
3.65434706e-01 1.19028211e+00 3.70561212e-01 8.12395096e-01
1.20737977e-01 2.80049592e-01 -3.57371867e-01 -4.76938009e-01
5.49800634e-01 6.11280948e-02 -6.75504923e-01 3.07468444e-01
1.04251945e+00 -5.76675236e-01 -3.08665693e-01 6.80836678e-01
7.60910869e-01 1.69962376e-01 5.43937743e-01 -1.40552247e+00
-2.14125305e-01 -1.86300069e-01 -7.47816145e-01 4.00779061e-02
3.88245642e-01 3.50931376e-01 8.64581466e-01 -1.50679922e+00
8.92723858e-01 1.07423508e+00 9.84418273e-01 5.88460565e-01
-1.28032386e+00 -4.67321038e-01 2.01164752e-01 2.27088958e-01
-1.55221832e+00 -7.38031924e-01 8.38307083e-01 -4.73611683e-01
1.28699172e+00 2.46439606e-01 6.50220871e-01 1.08132565e+00
-1.12741783e-01 6.89768910e-01 4.85515773e-01 -2.42220849e-01
4.40406911e-02 -5.23404218e-03 -2.08617777e-01 5.29014468e-01
-9.16114193e-04 8.49354416e-02 -5.28858304e-01 -2.06396610e-01
6.69897616e-01 3.09204817e-01 -2.85465568e-01 -1.30593038e+00
-1.12532365e+00 5.38986742e-01 6.14100635e-01 -1.01762548e-01
-3.18521678e-01 3.55509013e-01 8.21659267e-02 -8.80681258e-03
4.25716579e-01 4.42486614e-01 -6.59244835e-01 -8.20968002e-02
-1.04660511e+00 3.78257513e-01 6.70174778e-01 1.26013553e+00
8.74943435e-01 -3.06675315e-01 -3.21066119e-02 6.92084849e-01
2.28276029e-01 5.48420250e-01 -3.82253192e-02 -1.07599080e+00
4.01753902e-01 6.17919922e-01 2.49071479e-01 -7.99279213e-01
-3.39438975e-01 -4.69826847e-01 -1.63044572e-01 3.20547968e-01
2.09833562e-01 2.87447631e-01 -1.15452707e+00 1.43152690e+00
7.56128192e-01 2.30011135e-01 -3.87133121e-01 1.06186378e+00
8.20393920e-01 2.45490745e-01 -2.08307937e-01 2.57459968e-01
1.22502005e+00 -1.17997515e+00 -1.68448180e-01 -4.80304003e-01
4.77899343e-01 -8.13050091e-01 1.09597707e+00 3.19300920e-01
-1.32344961e+00 -5.42438745e-01 -1.01704431e+00 -4.19292629e-01
-5.14847100e-01 8.01144466e-02 4.73932564e-01 2.58557677e-01
-8.87050033e-01 7.57505894e-01 -9.83818114e-01 -4.57605273e-01
6.47471070e-01 5.77491701e-01 -5.65419614e-01 -2.72394717e-01
-4.84712571e-01 9.57898498e-01 1.78990319e-01 4.94468352e-03
-9.18703198e-01 -9.87347364e-01 -9.31296647e-01 1.41442660e-02
7.13459134e-01 -1.03477907e+00 1.52752745e+00 -6.71033680e-01
-1.20465350e+00 1.03502655e+00 -7.15314820e-02 -1.56667113e-01
5.78535616e-01 -5.39268851e-01 1.06322415e-01 2.14409560e-01
2.02090904e-01 9.15489793e-01 8.96561861e-01 -1.32156265e+00
-4.04399335e-01 -5.17541528e-01 1.33531622e-03 3.16753566e-01
-3.33712343e-03 -3.81555706e-02 -1.04889643e+00 -3.69696766e-01
3.55889976e-01 -8.75355780e-01 -1.83980733e-01 5.31959116e-01
-9.64813605e-02 -9.19032097e-03 1.14732552e+00 -2.85046220e-01
6.84124172e-01 -2.20051646e+00 1.93958491e-01 -1.09878937e-02
1.97309449e-01 1.81377932e-01 -3.36736143e-01 4.88983512e-01
-3.54447551e-02 -3.18095833e-01 -1.31519228e-01 -7.94114411e-01
8.30861330e-02 1.34263560e-01 -2.97070503e-01 7.54115880e-01
4.42078561e-01 1.08017123e+00 -7.45358944e-01 -3.51025462e-01
4.38459933e-01 6.16227269e-01 -8.52813005e-01 4.50823009e-01
-4.32489514e-01 2.67165720e-01 -2.47888323e-02 8.27775538e-01
8.87830853e-01 -4.84521359e-01 -2.48393893e-01 -4.86422896e-01
-1.32180557e-01 1.80316120e-01 -1.31910551e+00 2.17007613e+00
-4.95733976e-01 3.58074725e-01 1.58233166e-01 -6.41996682e-01
6.41016841e-01 1.89300388e-01 6.67421401e-01 -4.21295077e-01
2.82440484e-01 1.60755813e-01 -2.99398303e-01 -4.60798383e-01
4.26484317e-01 -8.38107765e-02 1.26166865e-01 2.72194862e-01
4.86809403e-01 -5.09501576e-01 6.51033595e-02 1.99394569e-01
9.87505019e-01 7.59654701e-01 3.00621659e-01 2.69820303e-01
2.93111295e-01 8.56063515e-02 4.14220512e-01 6.30759776e-01
-6.91877380e-02 1.28475392e+00 8.83826837e-02 -3.31012666e-01
-1.15857220e+00 -1.03434622e+00 -1.73993185e-01 1.21320462e+00
3.26923758e-01 -4.54482883e-01 -4.83207881e-01 -5.94116747e-01
3.10264200e-01 4.68793482e-01 -5.13606489e-01 -5.26973791e-02
-5.88777065e-01 -1.98346302e-01 1.21952504e-01 7.79947162e-01
1.61758289e-01 -7.92729378e-01 -8.96724045e-01 1.66864544e-01
3.84223275e-02 -1.17964387e+00 -5.26495457e-01 2.94266433e-01
-8.16046476e-01 -1.23305070e+00 -7.01052547e-01 -5.07837474e-01
6.12937748e-01 4.10942465e-01 1.34197402e+00 1.00844115e-01
-4.92597282e-01 8.27402949e-01 -1.80323243e-01 -5.14831007e-01
9.48072672e-02 2.19086096e-01 6.27940595e-02 -6.37209862e-02
4.19349253e-01 -6.35384023e-01 -7.99285591e-01 4.44197446e-01
-5.80627799e-01 8.16089511e-02 5.88842392e-01 5.75668275e-01
7.88794756e-01 -8.13557684e-01 9.38476026e-02 -6.08114064e-01
-1.42893121e-01 -3.45898330e-01 -7.56628990e-01 8.20424631e-02
-4.16048735e-01 -1.57618061e-01 2.78573602e-01 -5.62035203e-01
-8.10373545e-01 5.74759960e-01 -1.46152586e-01 -1.21488249e+00
-1.63060531e-01 -4.48490120e-02 -1.91574231e-01 -4.18271065e-01
5.90583324e-01 -3.72945964e-02 -5.57935834e-02 -7.11336672e-01
4.16185826e-01 4.24175441e-01 7.22391009e-01 -4.16279614e-01
1.02887189e+00 7.82553792e-01 -5.99922016e-02 -5.51240981e-01
-8.71570587e-01 -8.91666591e-01 -9.48292315e-01 -2.50111967e-01
6.37629867e-01 -9.70218897e-01 -8.36229503e-01 3.20073783e-01
-1.32287741e+00 -1.13802217e-01 -3.79757524e-01 4.42759126e-01
-9.26025391e-01 1.70756623e-01 -3.14746052e-01 -6.02908313e-01
-3.06009680e-01 -1.00970232e+00 1.76585364e+00 -1.75096989e-02
-6.93659857e-02 -5.35464287e-01 1.26659051e-01 2.80100733e-01
2.97565103e-01 2.17682734e-01 2.91856438e-01 -6.18116856e-01
-9.95917499e-01 -3.71018678e-01 -2.09679097e-01 -1.49610313e-03
-2.97568440e-01 4.84323082e-03 -1.18025291e+00 -5.19488573e-01
3.28325704e-02 -5.28264701e-01 7.13579953e-01 2.17290998e-01
1.06974292e+00 6.71219528e-02 -6.03721976e-01 1.07040203e+00
1.54243720e+00 -1.75157145e-01 3.68993372e-01 3.59113574e-01
8.71799529e-01 4.56171006e-01 7.28468359e-01 5.12315333e-01
2.76707530e-01 9.37227607e-01 8.60405982e-01 -6.02332391e-02
-1.60966143e-01 -3.37462753e-01 -7.59372255e-03 4.89505261e-01
2.46305075e-02 -1.49491683e-01 -1.08284986e+00 6.44249141e-01
-1.76722801e+00 -8.74155521e-01 2.49104425e-02 2.18587828e+00
4.32123810e-01 3.64201330e-02 1.98955089e-02 -1.05168521e-01
3.59848648e-01 1.31277218e-01 -7.85486519e-01 7.24069700e-02
1.99355900e-01 2.34185338e-01 4.22399908e-01 3.83812129e-01
-1.17718470e+00 9.25340116e-01 5.65125704e+00 4.04851645e-01
-1.09923565e+00 9.22649503e-02 1.34950191e-01 -5.70969880e-01
-2.04884950e-02 7.01590031e-02 -9.49438989e-01 6.46911636e-02
6.07712030e-01 9.68135968e-02 3.49042952e-01 1.20540774e+00
-1.80448622e-01 -6.19679950e-02 -1.57324660e+00 1.24794424e+00
3.56246531e-01 -1.25599301e+00 -2.23665938e-01 -1.15920968e-01
4.96074826e-01 4.84645277e-01 -5.81398532e-02 4.14906472e-01
-1.57492667e-01 -9.17218864e-01 9.20635402e-01 5.82016706e-01
8.27974975e-01 -4.84490842e-01 4.13937509e-01 5.03043592e-01
-1.30734241e+00 -5.86681888e-02 -3.88824552e-01 3.47695239e-02
2.14673862e-01 7.70525858e-02 -7.57558942e-01 5.20606577e-01
8.96013677e-01 6.47343397e-01 -7.78632402e-01 1.35306859e+00
1.09502584e-01 -3.48925143e-02 -6.15664005e-01 3.46432894e-01
7.03690797e-02 2.04124838e-01 6.81722403e-01 8.79062593e-01
3.47987801e-01 -1.16589349e-02 1.81356430e-01 8.71262968e-01
-1.84879396e-02 -1.51873142e-01 -7.17002630e-01 3.71017724e-01
4.23891902e-01 1.34627211e+00 -6.29695833e-01 -2.05186233e-01
-5.52821159e-01 1.02171230e+00 5.72310150e-01 1.78459749e-01
-9.55392063e-01 -3.38187277e-01 6.38291895e-01 3.59234363e-01
8.23843658e-01 -1.57620877e-01 -6.39180616e-02 -1.17012167e+00
2.89794475e-01 -7.02205300e-01 1.36579871e-01 -1.10227144e+00
-1.14818418e+00 5.19297957e-01 1.44803703e-01 -1.49239933e+00
-2.95794070e-01 -6.75620019e-01 -4.35856700e-01 7.58572876e-01
-1.25297916e+00 -1.36369324e+00 -6.23344302e-01 4.36052144e-01
6.15764797e-01 2.16793150e-01 6.71126842e-01 4.20643032e-01
-3.17157567e-01 4.61431891e-01 -3.06561530e-01 -1.12670392e-01
7.15398073e-01 -9.30033386e-01 6.43472195e-01 7.03210056e-01
2.77360529e-01 6.02748930e-01 5.50275087e-01 -5.21350801e-01
-1.63872147e+00 -9.48952556e-01 5.68842769e-01 -9.31534886e-01
3.73141706e-01 -8.01469505e-01 -9.94099498e-01 8.07813942e-01
-1.03717990e-01 5.03833711e-01 1.90738708e-01 -7.66543671e-02
-3.90936911e-01 1.77450031e-02 -9.91176844e-01 4.55197036e-01
1.32651401e+00 -4.95419413e-01 -5.61899185e-01 3.45193833e-01
7.52528965e-01 -8.75778437e-01 -7.07054555e-01 6.31802142e-01
6.33798540e-01 -1.02107251e+00 1.16708052e+00 -6.12034917e-01
1.20493896e-01 -4.15886015e-01 -2.03571141e-01 -8.90759945e-01
-3.06798428e-01 -4.69055027e-01 -4.43254918e-01 9.64148700e-01
1.24200828e-01 -3.00869673e-01 1.11118281e+00 6.15859210e-01
-2.83230692e-01 -1.07590914e+00 -8.88902128e-01 -8.29816639e-01
-3.59661818e-01 -5.89585006e-01 6.08874381e-01 5.72360337e-01
-5.85529566e-01 3.79423648e-01 -1.36916012e-01 3.46420705e-01
6.70619369e-01 3.26787680e-01 1.28382838e+00 -1.26205778e+00
-3.88707846e-01 -2.73903221e-01 -4.97117877e-01 -1.17878366e+00
-1.24072924e-01 -7.53940701e-01 1.52712658e-01 -1.46097159e+00
2.42533967e-01 -1.75822809e-01 1.11760207e-01 4.32562977e-01
-4.00441373e-03 5.71410358e-01 3.03048164e-01 2.30595008e-01
-6.89809322e-01 6.24953091e-01 1.00891769e+00 1.69939429e-01
-1.99912153e-02 -9.56507865e-03 -2.36813754e-01 8.56396019e-01
5.25303602e-01 -5.66769421e-01 -2.93402493e-01 -5.78546405e-01
1.85521677e-01 -6.09739944e-02 7.97837794e-01 -1.16586983e+00
4.29753751e-01 -2.68276650e-02 5.22068441e-01 -1.13877368e+00
9.03185308e-01 -1.09496331e+00 3.66958916e-01 2.61454254e-01
5.55494539e-02 2.24021018e-01 2.25674883e-01 5.56536496e-01
7.08042607e-02 -2.17340291e-01 8.01658690e-01 -2.20209479e-01
-7.92762578e-01 7.50070810e-01 4.09012854e-01 -1.68731324e-02
1.19095874e+00 -5.80267668e-01 -1.50284171e-01 -1.31769881e-01
-7.01105237e-01 3.60223800e-01 9.18973267e-01 5.61661661e-01
7.34340012e-01 -1.31729507e+00 -4.76966619e-01 4.59929079e-01
4.49127197e-01 3.75915587e-01 2.15533510e-01 1.06703925e+00
-5.51979065e-01 2.78926581e-01 -8.09992626e-02 -9.32919323e-01
-1.25404751e+00 8.46783638e-01 4.92763877e-01 1.28945917e-01
-7.16414511e-01 9.99687374e-01 3.05949956e-01 -6.52914524e-01
4.48002905e-01 -8.20160583e-02 1.71423569e-01 -6.42874837e-02
3.64337116e-01 4.23672438e-01 3.64438802e-01 -7.07931757e-01
-6.27032578e-01 9.55383122e-01 -8.77760276e-02 -8.59817490e-02
1.53093970e+00 -2.98948828e-02 1.73672751e-01 3.67743134e-01
1.37114739e+00 -1.04020298e-01 -1.68891561e+00 -2.28912458e-01
-9.75984931e-02 -7.63720572e-01 -1.26490161e-01 -4.83106494e-01
-9.80113149e-01 8.88535082e-01 5.43614030e-01 -2.50586778e-01
8.45709741e-01 3.03133667e-01 6.35609627e-01 3.89874130e-01
3.55122685e-01 -9.82055068e-01 2.98566848e-01 4.48916435e-01
1.09414101e+00 -1.28681755e+00 8.91582370e-02 -3.54106724e-01
-3.40435386e-01 9.48323727e-01 1.06870794e+00 -2.22324848e-01
4.81328756e-01 4.01501387e-01 -6.72059283e-02 -4.86795783e-01
-8.96991670e-01 -3.45402837e-01 5.17688036e-01 5.77424705e-01
1.27831802e-01 -5.42189777e-01 2.59858340e-01 3.16210747e-01
-5.04402779e-02 -5.60468212e-02 5.80122694e-02 1.11116397e+00
-2.89522111e-01 -8.08068156e-01 -5.18044591e-01 3.18576068e-01
-1.37801096e-01 3.94240320e-02 -4.61070806e-01 9.45574284e-01
1.69283688e-01 3.93383026e-01 3.08324099e-01 -3.46417099e-01
7.01227605e-01 5.11411056e-02 7.09326863e-01 -8.51456761e-01
-5.62275589e-01 7.06289187e-02 -2.06588790e-01 -9.81154859e-01
-3.19967836e-01 -6.52772963e-01 -9.61735010e-01 -3.89904492e-02
-5.61641097e-01 -3.48072827e-01 6.66491985e-01 6.36546314e-01
7.01416135e-01 3.00536186e-01 4.54731911e-01 -1.66988122e+00
-7.28861392e-01 -8.19545984e-01 -1.29020855e-01 4.89378035e-01
3.44460845e-01 -9.39033866e-01 -2.18085662e-01 -1.84926629e-01] | [7.609992504119873, -2.601003408432007] |
d5435477-a811-4a50-9862-37207ab23966 | deep-svbrdf-estimation-on-real-materials | 2010.04143 | null | https://arxiv.org/abs/2010.04143v1 | https://arxiv.org/pdf/2010.04143v1.pdf | Deep SVBRDF Estimation on Real Materials | Recent work has demonstrated that deep learning approaches can successfully be used to recover accurate estimates of the spatially-varying BRDF (SVBRDF) of a surface from as little as a single image. Closer inspection reveals, however, that most approaches in the literature are trained purely on synthetic data, which, while diverse and realistic, is often not representative of the richness of the real world. In this paper, we show that training such networks exclusively on synthetic data is insufficient to achieve adequate results when tested on real data. Our analysis leverages a new dataset of real materials obtained with a novel portable multi-light capture apparatus. Through an extensive series of experiments and with the use of a novel deep learning architecture, we explore two strategies for improving results on real data: finetuning, and a per-material optimization procedure. We show that adapting network weights to real data is of critical importance, resulting in an approach which significantly outperforms previous methods for SVBRDF estimation on real materials. Dataset and code are available at https://lvsn.github.io/real-svbrdf | ['Jean-François Lalonde', 'Denis Laurendeau', 'Louis-Philippe Asselin'] | 2020-10-08 | null | null | null | null | ['svbrdf-estimation'] | ['computer-vision'] | [ 4.96587932e-01 -2.00229421e-01 9.89834219e-02 -3.23568612e-01
-8.57711613e-01 -5.55316210e-01 4.22218263e-01 -3.46293956e-01
-9.04614702e-02 8.72969747e-01 -1.21033743e-01 -3.10734600e-01
-1.50922194e-01 -7.29466081e-01 -9.09703135e-01 -7.04521060e-01
1.44428357e-01 2.64471382e-01 2.07198575e-01 -3.49933244e-02
1.82502508e-01 9.80328560e-01 -1.86041117e+00 1.65764183e-01
4.81787533e-01 1.15998542e+00 3.70235205e-01 7.92525411e-01
1.49760321e-01 5.02440035e-01 -6.06250644e-01 -4.66360599e-02
5.55935323e-01 -1.32598123e-02 -6.96038961e-01 -4.26260047e-02
9.03444588e-01 -7.33423948e-01 -3.24248105e-01 6.52926326e-01
7.32454896e-01 2.24217117e-01 5.82421780e-01 -8.89445841e-01
-6.57045424e-01 -2.88922954e-02 -5.45901418e-01 1.13055751e-01
2.16395110e-01 3.57277870e-01 9.59190011e-01 -7.93041170e-01
5.44892073e-01 9.14398074e-01 8.20983052e-01 4.45426792e-01
-1.48994076e+00 -5.38466573e-01 -2.69005209e-01 -1.15584873e-01
-1.29396975e+00 -7.84472227e-01 9.79841053e-01 -4.47564691e-01
8.41683626e-01 1.25526443e-01 6.22671545e-01 1.24299800e+00
1.30058110e-01 4.94465232e-01 1.25342596e+00 -5.73963165e-01
4.92521375e-02 -9.66222063e-02 -2.89543182e-01 4.43379164e-01
3.99180770e-01 2.86513209e-01 -3.04695159e-01 -2.68719047e-02
1.05448472e+00 -2.46763423e-01 -4.63767976e-01 -6.50388718e-01
-1.15796053e+00 6.36663139e-01 3.62368464e-01 1.63418263e-01
-3.06148291e-01 4.36163098e-01 1.48263782e-01 3.03464115e-01
6.89953864e-01 5.33120811e-01 -5.57285607e-01 -8.23888406e-02
-8.47704351e-01 2.49386013e-01 6.94883883e-01 6.93274856e-01
8.38101506e-01 1.08082227e-01 2.21668825e-01 1.08214915e+00
2.14200646e-01 8.18903625e-01 4.81884405e-02 -1.53309011e+00
1.77712247e-01 1.54500082e-01 4.26709056e-01 -1.00382245e+00
-3.73409569e-01 -3.51149082e-01 -5.04067838e-01 6.83477998e-01
7.20041335e-01 -2.34259754e-01 -9.89319682e-01 1.62919903e+00
3.10852647e-01 1.51665835e-02 -2.26774871e-01 8.56674731e-01
8.29877436e-01 4.19620484e-01 -4.35987979e-01 2.24357732e-02
7.90528595e-01 -6.29825950e-01 -4.39536631e-01 -2.39873871e-01
3.07697117e-01 -9.14824605e-01 1.37536979e+00 6.04719818e-01
-1.18851531e+00 -3.95699799e-01 -1.13018799e+00 -1.16685569e-01
-4.21436846e-01 2.25118339e-01 7.93058097e-01 4.71200198e-01
-1.14062226e+00 7.43570030e-01 -7.53631532e-01 -3.27127486e-01
6.12853587e-01 3.63983363e-01 -3.44779342e-01 -3.56122464e-01
-7.74245024e-01 6.95776582e-01 -4.57562767e-02 4.56040531e-01
-8.52521181e-01 -8.56874228e-01 -7.05038190e-01 -2.12747931e-01
4.30720568e-01 -5.86609960e-01 1.41486692e+00 -1.04299462e+00
-1.56504703e+00 8.73516381e-01 8.40670522e-03 1.89215448e-02
4.68406945e-01 -1.90118492e-01 -2.36608729e-01 2.99213141e-01
-1.82378203e-01 5.31116724e-01 9.81783986e-01 -1.79128301e+00
-3.46149385e-01 -1.65295094e-01 3.34939152e-01 -1.79527283e-01
-1.30262464e-01 -1.40115410e-01 -3.86630110e-02 -4.84600574e-01
-1.02121957e-01 -8.17026198e-01 4.31926660e-02 4.66382831e-01
-2.90180862e-01 1.69118792e-01 6.11003995e-01 -6.05151772e-01
6.88335001e-01 -1.91954505e+00 -2.33235836e-01 1.26317799e-01
2.80502617e-01 4.74889934e-01 -2.59741932e-01 3.94357294e-01
-1.34495437e-01 -5.83989471e-02 -3.53623599e-01 -2.52510786e-01
-1.06913179e-01 1.29519925e-01 -2.45015830e-01 7.35977590e-01
2.60211557e-01 8.37158084e-01 -7.07760572e-01 -2.12551340e-01
3.82031471e-01 7.88840473e-01 -2.67354906e-01 1.92234635e-01
-3.25070739e-01 4.53694016e-01 -1.59919098e-01 8.44729066e-01
8.80874991e-01 -3.80577356e-01 -5.23528550e-04 -4.39745665e-01
-1.36311516e-01 -5.23804836e-02 -1.06295419e+00 1.60763192e+00
-7.77681947e-01 1.00034583e+00 3.33680183e-01 -9.29502308e-01
7.40621150e-01 1.13916881e-01 6.57597184e-01 -8.19578409e-01
1.98091105e-01 3.42110425e-01 -3.43892761e-02 -6.72884166e-01
3.37486625e-01 -1.33742407e-01 4.45358187e-01 6.62201345e-01
2.61601079e-02 -3.54194731e-01 5.34610748e-02 -2.33792260e-01
1.07835031e+00 3.42675775e-01 -8.28299001e-02 -2.11904600e-01
2.93910503e-01 -1.16861120e-01 1.46223143e-01 8.16305459e-01
-4.26886976e-02 9.84759927e-01 2.55547673e-01 -5.61486304e-01
-1.18097258e+00 -1.15692711e+00 -3.57233614e-01 8.50514710e-01
1.31026685e-01 1.72561496e-01 -6.13336265e-01 -3.34078163e-01
2.15355664e-01 5.41718066e-01 -7.16479421e-01 9.74292830e-02
-5.39230287e-01 -9.02661026e-01 2.83837646e-01 4.15801316e-01
1.81895912e-01 -1.19949174e+00 -7.73046672e-01 -1.96847022e-02
1.95650503e-01 -1.26248968e+00 1.38461411e-01 8.30054954e-02
-6.58993006e-01 -1.18738985e+00 -8.61959517e-01 -4.37683851e-01
4.98099566e-01 5.94654381e-01 1.32086551e+00 2.41921887e-01
-5.41868687e-01 6.35201275e-01 -5.42576760e-02 -3.63492459e-01
-4.15460020e-01 9.91248935e-02 -3.14937353e-01 -5.06883897e-02
1.03735030e-01 -8.38626325e-01 -8.69052529e-01 2.07621604e-01
-1.00173831e+00 7.62940496e-02 5.13857782e-01 7.22867131e-01
4.91411746e-01 -1.58752844e-01 4.58541811e-01 -7.42832601e-01
5.76999605e-01 -3.67981374e-01 -7.34867156e-01 5.04171960e-02
-6.38042390e-01 -7.09910244e-02 5.27079821e-01 -5.07166326e-01
-1.19436395e+00 -1.22684613e-01 -3.16506326e-01 -3.90820622e-01
-4.89683867e-01 2.60161430e-01 1.77241072e-01 -5.72757661e-01
7.80481279e-01 -1.74155474e-01 1.25540048e-01 -4.32581365e-01
1.54053167e-01 5.95934331e-01 3.94318342e-01 -7.05901086e-01
6.56249702e-01 8.49234223e-01 2.32874632e-01 -9.85902727e-01
-8.66355836e-01 -2.51400918e-01 -6.52804911e-01 -4.33003157e-01
4.51520205e-01 -7.33245134e-01 -6.76005602e-01 7.27828145e-01
-1.01980293e+00 -1.00874877e+00 -4.02026832e-01 3.41763169e-01
-7.52326846e-01 4.34426442e-02 -3.59851241e-01 -6.48160934e-01
-1.10782981e-01 -1.03008401e+00 1.36950839e+00 1.22356549e-01
1.20606601e-01 -1.15654886e+00 1.84003308e-01 3.80599737e-01
7.54981935e-01 5.59718788e-01 6.88406050e-01 -3.46254334e-02
-6.45746291e-01 -1.50530890e-01 -5.97402573e-01 5.73135376e-01
3.75333905e-01 3.88421237e-01 -1.35090613e+00 -3.16731691e-01
-9.25159305e-02 -5.90454400e-01 7.58419871e-01 5.91242731e-01
1.35918891e+00 1.02098331e-01 -9.39789489e-02 7.62281954e-01
1.71734905e+00 -1.32917896e-01 5.96698105e-01 4.13738489e-01
9.42206502e-01 4.97078091e-01 3.93640101e-01 2.07065612e-01
2.84294069e-01 6.89840555e-01 5.77680290e-01 -3.69941950e-01
-6.39184475e-01 2.07477495e-01 1.01903126e-01 4.73553628e-01
-3.74031126e-01 -4.59477723e-01 -9.81592357e-01 6.48267508e-01
-1.44675660e+00 -7.14034200e-01 -3.25327665e-02 2.20691204e+00
5.92843294e-01 9.93317273e-03 -3.34829278e-03 3.36210281e-02
5.11625230e-01 2.94699341e-01 -7.94195712e-01 -2.87710935e-01
-1.76597074e-01 4.05693650e-01 6.68904066e-01 4.48500603e-01
-7.83098221e-01 6.36772752e-01 7.26442814e+00 6.42945290e-01
-1.46400487e+00 1.28472537e-01 4.62564081e-01 -2.72810370e-01
-5.05841136e-01 -3.51109207e-01 -3.84627670e-01 3.58964652e-01
9.22310233e-01 2.54477501e-01 8.00402284e-01 4.68372524e-01
2.39913315e-01 -2.02355877e-01 -8.15203547e-01 9.67761755e-01
8.56062621e-02 -1.21146131e+00 -1.75603941e-01 6.55021816e-02
7.38992691e-01 4.21490639e-01 5.60908616e-02 -2.51841187e-01
3.74192238e-01 -1.06273937e+00 5.26768923e-01 6.83055460e-01
9.81615901e-01 -5.00052214e-01 5.75003564e-01 5.24340197e-02
-7.27566600e-01 -2.31957110e-03 -2.68131942e-01 1.16610982e-01
2.16822833e-01 6.67174578e-01 -5.79136908e-01 3.77678007e-01
7.98743427e-01 6.71475768e-01 -6.87747717e-01 9.21458185e-01
-1.25565708e-01 6.09235346e-01 -3.92047673e-01 1.58522323e-01
-1.47962019e-01 -6.93777204e-02 3.79788518e-01 1.11795557e+00
3.55908632e-01 -2.65942246e-01 -3.46069694e-01 9.18613136e-01
-6.88189492e-02 -1.80305377e-01 -8.95566761e-01 1.06715709e-02
1.53603181e-01 1.36481524e+00 -6.82862818e-01 2.98168529e-02
-4.96076971e-01 5.10035634e-01 4.74462032e-01 6.56417012e-01
-6.39870286e-01 -2.91695535e-01 6.26372337e-01 3.92637730e-01
5.22835732e-01 -4.92240369e-01 -2.81285882e-01 -1.00580001e+00
1.61019161e-01 -7.83513188e-01 -1.41026080e-01 -1.36665916e+00
-1.32257521e+00 4.33721840e-01 2.32660398e-01 -1.01687241e+00
-1.52967989e-01 -9.52821612e-01 -3.45406622e-01 7.27615356e-01
-1.74166656e+00 -1.17835736e+00 -6.80644274e-01 3.25873584e-01
2.92791784e-01 1.67974811e-02 6.90377057e-01 4.04333055e-01
-4.15600061e-01 3.18618506e-01 4.31806892e-01 -3.31155509e-01
6.95165217e-01 -1.17005670e+00 4.95948851e-01 7.41140127e-01
-2.26524496e-03 4.14730519e-01 8.42227280e-01 -2.19749302e-01
-1.50026178e+00 -6.13828659e-01 1.06808379e-01 -4.12379414e-01
7.73098290e-01 -4.95942503e-01 -1.02543902e+00 5.13103247e-01
1.68796167e-01 3.44159871e-01 4.33332771e-01 -1.09608918e-01
-8.35447237e-02 -2.68373430e-01 -1.26548481e+00 4.76065069e-01
1.24070454e+00 -6.06176972e-01 -8.00153390e-02 4.15236473e-01
6.05464876e-01 -5.72727084e-01 -1.07442236e+00 6.44497335e-01
9.75274146e-01 -1.22595489e+00 1.09871054e+00 -3.58046770e-01
6.34414136e-01 -1.59161702e-01 -3.26988578e-01 -1.36940813e+00
-6.75739050e-02 -4.62583870e-01 -3.76658261e-01 9.74945962e-01
2.65566707e-01 -8.20749283e-01 7.32708216e-01 5.69436431e-01
-6.49375767e-02 -9.03833628e-01 -6.70529723e-01 -6.88972950e-01
2.62462683e-02 -5.70146859e-01 6.61240458e-01 6.87473238e-01
-1.00742316e+00 -3.79313044e-02 -1.91539720e-01 7.90201426e-02
7.08324969e-01 2.88388908e-01 9.56393540e-01 -1.39822853e+00
-2.01802015e-01 -3.15863073e-01 -2.36264303e-01 -8.42373312e-01
3.77224356e-01 -5.13861835e-01 2.87317902e-01 -1.58812225e+00
-5.95696904e-02 -6.24357224e-01 -1.63636148e-01 3.42154622e-01
2.48801365e-01 6.17682576e-01 -1.62118420e-01 1.50507957e-01
-1.09956257e-01 3.77228618e-01 1.59105086e+00 -7.37969801e-02
8.59016031e-02 -1.89933218e-02 -5.87948382e-01 6.86419487e-01
9.36939299e-01 -3.68515760e-01 -4.56782788e-01 -7.84405112e-01
4.27576005e-01 -2.40250304e-01 6.72425389e-01 -9.60140884e-01
-1.32647157e-01 -1.54753581e-01 4.39750940e-01 -2.38915548e-01
5.46276867e-01 -8.42010915e-01 1.81631342e-01 -1.91006940e-02
-1.85841158e-01 -1.88615784e-01 4.85012531e-01 3.07019979e-01
1.16373830e-01 -3.17166865e-01 7.69138992e-01 -1.40472546e-01
-5.68375945e-01 2.17620283e-01 -1.54532716e-01 -2.43730769e-02
5.77757537e-01 -3.89725357e-01 -5.17901242e-01 -3.60071808e-01
-4.31295037e-01 -2.74313897e-01 8.38497400e-01 2.55750328e-01
5.03633380e-01 -1.28149164e+00 -4.65398014e-01 4.25359219e-01
-3.00399549e-02 2.85118550e-01 9.72689912e-02 5.81456661e-01
-9.04544592e-01 1.14932589e-01 -3.00392628e-01 -6.33867145e-01
-1.06758022e+00 3.34437370e-01 5.34032345e-01 1.07671008e-01
-6.20468915e-01 5.11160851e-01 9.49089602e-02 -5.53707898e-01
-2.68930867e-02 -4.52042162e-01 1.60673261e-01 -2.61366963e-01
2.27627769e-01 3.42033595e-01 3.18890244e-01 -5.40881991e-01
-1.18527254e-02 7.51209140e-01 1.41856745e-01 -7.90062100e-02
1.76187479e+00 -1.65571928e-01 -5.98958619e-02 6.85285747e-01
1.26218891e+00 9.21987519e-02 -1.71179974e+00 3.45335668e-03
-4.46381062e-01 -7.25751460e-01 3.57876241e-01 -7.03023255e-01
-1.37748289e+00 8.69655430e-01 6.91185057e-01 5.52442789e-01
1.18119061e+00 -8.64264891e-02 7.68067479e-01 5.71252406e-01
3.77054065e-01 -9.99384224e-01 1.07073039e-01 2.33743668e-01
1.01437795e+00 -1.48860407e+00 2.08833560e-01 -3.37656647e-01
-1.28073782e-01 1.20152915e+00 4.30485874e-01 -2.20781520e-01
8.13531816e-01 4.33378756e-01 2.40229651e-01 -4.57890928e-01
-5.05291402e-01 5.67619540e-02 1.72917545e-01 6.76241338e-01
3.48184735e-01 -1.74935699e-01 3.11550498e-01 -1.76483616e-01
-4.33023237e-02 2.02073500e-01 6.64170265e-01 9.88285482e-01
-2.89541066e-01 -8.74180615e-01 -3.46678764e-01 6.16951048e-01
-5.61569989e-01 1.19199939e-01 -2.21254811e-01 1.00275946e+00
-1.36862714e-02 7.19626009e-01 9.68315378e-02 -4.62043993e-02
3.98968011e-01 -4.34631169e-01 9.76773798e-01 -4.57425773e-01
-7.17059299e-02 -2.39910811e-01 9.85666662e-02 -7.52462626e-01
-8.46643269e-01 -6.99687243e-01 -7.59349644e-01 -3.01838517e-01
-2.86643475e-01 -5.81814945e-01 8.32902908e-01 9.10360992e-01
2.61933327e-01 5.64511597e-01 5.56714058e-01 -1.43146038e+00
-1.46749169e-01 -8.35151315e-01 -5.89581609e-01 3.03050816e-01
8.28648627e-01 -9.52393293e-01 -6.03085995e-01 3.12324464e-02] | [9.725398063659668, -2.7667195796966553] |
32229abf-a30e-4bba-80e7-f3e7b314f9dc | a-benchmark-for-toxic-comment-classification | 2301.11125 | null | https://arxiv.org/abs/2301.11125v1 | https://arxiv.org/pdf/2301.11125v1.pdf | A benchmark for toxic comment classification on Civil Comments dataset | Toxic comment detection on social media has proven to be essential for content moderation. This paper compares a wide set of different models on a highly skewed multi-label hate speech dataset. We consider inference time and several metrics to measure performance and bias in our comparison. We show that all BERTs have similar performance regardless of the size, optimizations or language used to pre-train the models. RNNs are much faster at inference than any of the BERT. BiLSTM remains a good compromise between performance and inference time. RoBERTa with Focal Loss offers the best performance on biases and AUROC. However, DistilBERT combines both good AUROC and a low inference time. All models are affected by the bias of associating identities. BERT, RNN, and XLNet are less sensitive than the CNN and Compact Convolutional Transformers. | ['Reda Dehak', 'Pierre Guillaume', 'Henri Jamet', 'Corentin Duchene'] | 2023-01-26 | null | null | null | null | ['toxic-comment-classification'] | ['natural-language-processing'] | [-3.00196886e-01 -5.31098843e-02 -2.18006186e-02 -4.50561702e-01
-9.03702617e-01 -6.22676790e-01 9.01162505e-01 2.17951924e-01
-6.65117681e-01 8.52241933e-01 4.00214702e-01 -4.19573903e-01
5.30232526e-02 -5.73006988e-01 -4.20696974e-01 -6.27803206e-01
1.54624000e-01 4.20949966e-01 2.31770769e-01 -2.87167519e-01
1.02824181e-01 2.08696291e-01 -9.91213500e-01 2.91728169e-01
4.11237240e-01 8.95959914e-01 -5.76120555e-01 7.98567533e-01
1.40622437e-01 1.27454531e+00 -1.02487016e+00 -9.80218053e-01
-1.88423812e-01 -3.30713272e-01 -6.59036398e-01 -6.84200108e-01
8.49307597e-01 -3.18251014e-01 -2.42516115e-01 9.16565597e-01
1.09212685e+00 -7.92550892e-02 8.69075000e-01 -1.18713188e+00
-5.87960184e-01 1.36866641e+00 -4.10040379e-01 6.81217372e-01
-1.01473063e-01 2.08833382e-01 1.36660480e+00 -8.39537859e-01
4.06743765e-01 1.53530335e+00 1.18134367e+00 4.75248128e-01
-1.17870033e+00 -1.02557635e+00 8.81581232e-02 1.05397940e-01
-1.08788300e+00 -4.99768287e-01 4.34387624e-01 -4.78874117e-01
8.26013982e-01 3.02780926e-01 1.69659719e-01 1.69554710e+00
-4.80063222e-02 6.05444074e-01 1.16976166e+00 -1.73685122e-02
-1.70625091e-01 5.48814416e-01 4.08924490e-01 8.00773025e-01
6.25166744e-02 -4.08587083e-02 -7.08669722e-01 -4.75169837e-01
8.63208547e-02 -2.18590543e-01 -1.48031890e-01 4.64871079e-01
-9.24792826e-01 1.15755749e+00 5.24862945e-01 3.15310806e-01
2.33271588e-02 6.59548402e-01 7.66442299e-01 4.32179332e-01
1.02059758e+00 8.02865922e-01 -3.17097604e-01 -2.43213221e-01
-1.02875412e+00 4.34949130e-01 8.55522394e-01 4.58684236e-01
3.82564753e-01 2.51295298e-01 -6.06947780e-01 9.71772313e-01
4.44918573e-02 5.87385893e-01 2.47333273e-01 -9.50061202e-01
4.13162827e-01 2.56379973e-02 -2.13691935e-01 -1.16790235e+00
-6.62117541e-01 -9.07079160e-01 -5.96437156e-01 8.58990848e-02
7.12547719e-01 -6.55363262e-01 -6.02077544e-01 1.70647311e+00
-1.40258476e-01 -5.78231700e-02 -5.42903900e-01 5.11666238e-01
9.08002138e-01 5.61405361e-01 2.72167951e-01 2.50941098e-01
1.07600105e+00 -9.18099046e-01 -6.13136053e-01 -1.59217164e-01
8.41958344e-01 -7.79323161e-01 9.27455902e-01 3.04482281e-01
-1.18616438e+00 -1.20824926e-01 -1.01225877e+00 -3.15537512e-01
-6.20628178e-01 -1.76223531e-01 3.07053745e-01 9.47950542e-01
-1.13881767e+00 8.43843043e-01 -2.18708858e-01 -5.19436486e-02
6.23065054e-01 2.23064125e-01 9.87610519e-02 2.37752110e-01
-1.51422453e+00 1.33395326e+00 -9.97710899e-02 -5.57640344e-02
-1.05636823e+00 -8.91522884e-01 -4.73461449e-01 1.30850121e-01
-1.91807486e-02 -3.02404106e-01 1.43683481e+00 -7.85918474e-01
-1.20264030e+00 9.80356932e-01 1.69056699e-01 -8.46604228e-01
7.92346537e-01 -2.47369960e-01 -3.75397027e-01 -2.53570825e-01
-1.78824160e-02 6.18747354e-01 7.83500433e-01 -9.66441989e-01
-5.98791897e-01 -5.75883836e-02 1.20547451e-01 -1.84852295e-02
-5.89808404e-01 5.06855369e-01 2.78155476e-01 -3.54281276e-01
-7.57653296e-01 -8.75360370e-01 3.99363637e-01 -3.73181552e-01
-5.90426445e-01 -4.17636633e-01 8.74435902e-01 -8.26713324e-01
1.36763251e+00 -1.99276638e+00 -1.73144788e-01 1.63337439e-01
5.13408363e-01 4.12791669e-01 -9.11425203e-02 2.23671913e-01
1.17706895e-01 6.46740735e-01 2.01925859e-01 -6.79663301e-01
1.72687411e-01 -1.53750688e-01 -2.77518541e-01 6.48626745e-01
3.39510031e-02 9.06248808e-01 -7.95128167e-01 -4.44801837e-01
-6.76152781e-02 6.95869744e-01 -5.74901640e-01 -3.84789556e-02
-1.72075972e-01 3.32714617e-02 1.56014547e-01 4.00502324e-01
2.65611559e-01 -3.15916300e-01 -1.49619237e-01 2.16876134e-01
-1.07981794e-01 8.14245641e-01 -3.32806855e-01 8.33289862e-01
-5.63303590e-01 1.24951959e+00 2.48590093e-02 -4.99364644e-01
7.95512140e-01 3.41776907e-01 -1.64441645e-01 -6.39763653e-01
3.81245643e-01 2.93846339e-01 3.01355183e-01 -3.06757450e-01
3.36822838e-01 -2.30603412e-01 4.39355001e-02 5.68903327e-01
-2.30939388e-02 1.45020589e-01 -3.25458422e-02 3.49078804e-01
1.16805685e+00 -3.33641887e-01 -1.67669266e-01 -3.67313117e-01
8.59233811e-02 -4.10217375e-01 3.26784104e-01 9.60518479e-01
-1.33616418e-01 6.40773773e-01 1.01992118e+00 -2.80599445e-01
-1.16317213e+00 -9.49559212e-01 -1.32150069e-01 1.66265869e+00
-5.31962872e-01 -3.52310717e-01 -5.86461604e-01 -7.98651993e-01
-5.74051738e-02 1.07978034e+00 -9.04351950e-01 -6.19416051e-02
-5.01264632e-01 -1.06997979e+00 1.26076198e+00 3.03726256e-01
3.83539617e-01 -9.46506917e-01 -1.68547645e-01 -7.80259222e-02
-2.40871266e-01 -8.39650035e-01 -3.43922675e-01 2.55384356e-01
-2.83468962e-01 -7.29814470e-01 -6.06910646e-01 -4.45757806e-01
1.12575665e-01 -3.92889194e-02 1.42878652e+00 2.09406644e-01
5.89363202e-02 -1.13410473e-01 -1.89041331e-01 -1.01746547e+00
-6.26611292e-01 6.87433779e-01 -2.77939260e-01 -1.74045101e-01
3.31318587e-01 -6.29972160e-01 -3.75781566e-01 1.93040058e-01
-4.41476613e-01 -3.48190606e-01 3.15796822e-01 8.91665339e-01
-3.52382183e-01 -2.18132332e-01 5.91251969e-01 -1.37239087e+00
8.52160454e-01 -8.44978750e-01 -1.92359284e-01 -3.95826586e-02
-6.93162143e-01 -1.59831613e-01 5.72894812e-01 -3.30075592e-01
-7.76010633e-01 -7.43831635e-01 -2.81069040e-01 -1.88419655e-01
7.06382021e-02 3.70828837e-01 3.36538374e-01 1.44020021e-01
1.09306610e+00 -5.83446383e-01 -5.94507158e-02 -4.42035109e-01
7.63704926e-02 6.54478610e-01 9.78494957e-02 -1.66130871e-01
6.81816220e-01 3.33307207e-01 -3.17729414e-01 -7.94990778e-01
-1.46078968e+00 -2.41475344e-01 -3.15237641e-01 -3.14757913e-01
8.46562207e-01 -9.82712865e-01 -8.04546237e-01 6.51708364e-01
-1.29713190e+00 -5.43317854e-01 7.17690140e-02 3.11664283e-01
9.32320114e-03 -1.75597981e-01 -1.04170656e+00 -8.57956469e-01
-5.39704323e-01 -7.67803073e-01 6.91649318e-01 -1.48532018e-01
-5.05496383e-01 -1.28194201e+00 5.04049473e-02 3.11859429e-01
7.45819569e-01 3.85775805e-01 9.40276742e-01 -1.12398660e+00
-1.63849339e-01 -2.05544069e-01 -4.67044353e-01 5.44797719e-01
-2.26908341e-01 2.55686074e-01 -1.56758749e+00 -2.39194542e-01
-2.29427636e-01 -6.05008423e-01 1.22356999e+00 2.97871709e-01
1.08645976e+00 -6.37305915e-01 -5.70793673e-02 3.95414621e-01
1.08673346e+00 -2.67228127e-01 4.72875357e-01 3.14145267e-01
8.70308340e-01 6.27328753e-01 2.59589404e-02 1.74997017e-01
3.94880831e-01 3.26315284e-01 4.23712820e-01 -2.73703516e-01
-1.27762735e-01 -2.08309978e-01 7.76848733e-01 7.88682222e-01
9.28097218e-02 -5.23674428e-01 -1.19064009e+00 5.31656981e-01
-1.67233479e+00 -1.36500680e+00 -3.76771867e-01 1.70021784e+00
8.31772506e-01 4.29990292e-01 5.65876663e-01 7.75470510e-02
7.07757056e-01 5.26731551e-01 -1.97141632e-01 -7.12763309e-01
-3.47467810e-01 -5.11958525e-02 7.62361526e-01 8.17743003e-01
-1.00455034e+00 7.09620774e-01 7.47680140e+00 8.15602899e-01
-1.12348580e+00 5.97880602e-01 1.01509082e+00 -7.00424910e-01
-2.56688714e-01 -4.36372489e-01 -1.02838254e+00 6.66762948e-01
1.45214188e+00 1.58638343e-01 3.29634398e-01 7.10163772e-01
7.36309066e-02 1.07636437e-01 -1.11745524e+00 6.03875935e-01
3.53392512e-01 -1.31367385e+00 -1.99413180e-01 1.25190988e-01
7.99916148e-01 6.55021966e-01 4.93932843e-01 4.79326040e-01
7.56570816e-01 -1.34986234e+00 9.34235156e-01 3.41040641e-01
5.52262843e-01 -9.17385519e-01 1.05046237e+00 1.57552823e-01
-2.73788441e-02 -1.89978868e-01 -2.15216085e-01 -2.29859456e-01
-7.86400959e-02 8.34860146e-01 -1.05533683e+00 -2.03343555e-01
7.23411918e-01 6.06378913e-01 -8.28501642e-01 8.00848067e-01
-2.61986136e-01 1.13293684e+00 -2.88702726e-01 -5.87591231e-01
6.28893018e-01 3.02978575e-01 3.73126239e-01 1.77757323e+00
-1.85662545e-02 -5.49956799e-01 -2.39200041e-01 7.70976186e-01
-3.02493244e-01 -5.02817519e-02 -6.18937492e-01 -8.03085417e-03
3.82602125e-01 1.24086607e+00 -2.38764778e-01 -3.15332979e-01
-4.77775514e-01 4.58673000e-01 6.72726512e-01 1.90981477e-01
-1.13629401e+00 -2.48488382e-01 6.67350411e-01 3.58229458e-01
1.74391001e-01 1.30847588e-01 -3.40944171e-01 -6.83223307e-01
-4.31656152e-01 -9.68990147e-01 4.62799907e-01 -6.32758975e-01
-1.40937006e+00 4.70675975e-01 -1.34725630e-01 -5.36325812e-01
-1.11722387e-01 -6.02464855e-01 -5.95881939e-01 8.98854494e-01
-1.58954740e+00 -1.04853415e+00 -1.05366759e-01 3.19990844e-01
2.11977080e-01 -5.79619035e-02 7.14814782e-01 5.98332584e-01
-7.55764782e-01 8.45380068e-01 1.36395350e-01 5.43265104e-01
1.02296412e+00 -1.50584412e+00 4.73306656e-01 4.66844857e-01
-1.76629554e-02 4.28647786e-01 8.84702682e-01 -3.28631014e-01
-4.20546293e-01 -1.15696943e+00 1.06187832e+00 -1.00179410e+00
8.72935593e-01 -5.53147554e-01 -5.83577693e-01 9.13062274e-01
4.08834308e-01 -4.89380658e-01 7.38143444e-01 8.00473452e-01
-9.66599584e-01 -1.32457256e-01 -8.33104849e-01 6.82629704e-01
7.67604172e-01 -9.69393790e-01 -2.61845708e-01 5.45263410e-01
6.94198191e-01 -1.56861976e-01 -7.73857713e-01 2.73348019e-02
6.01812124e-01 -1.21040118e+00 6.45519733e-01 -6.88123941e-01
7.11645603e-01 2.69516498e-01 4.85360511e-02 -1.47427917e+00
-6.90488636e-01 -5.49175501e-01 1.88175127e-01 1.55082178e+00
1.19413161e+00 -6.98492825e-01 5.66600621e-01 4.38141972e-01
1.81022003e-01 -6.08090580e-01 -5.39115369e-01 -5.14826953e-01
4.84934628e-01 -4.13051665e-01 3.19871813e-01 1.06421506e+00
-1.63022250e-01 9.80083644e-01 -6.03451967e-01 -2.07245722e-01
4.68747616e-01 -4.46995527e-01 5.51755786e-01 -1.25413048e+00
-6.51725605e-02 -7.69550085e-01 -1.03208669e-01 -3.31297815e-01
3.84234309e-01 -1.07181096e+00 -1.98807657e-01 -1.16944218e+00
2.91970074e-01 -4.41599876e-01 -3.95464033e-01 4.58800077e-01
-1.13485783e-01 6.69637322e-01 2.40160525e-01 6.38038944e-03
-5.99515915e-01 8.95806178e-02 8.69061232e-01 -3.43986094e-01
1.37257680e-01 6.10398389e-02 -8.79577816e-01 8.40245962e-01
1.14355528e+00 -7.52582490e-01 -2.06683591e-01 -5.90145111e-01
7.77637959e-01 -4.24520075e-01 4.33868676e-01 -9.08469021e-01
1.56478792e-01 4.60297137e-01 3.89598042e-01 -4.48199689e-01
5.03251910e-01 -2.91529059e-01 -2.58442163e-01 3.51627499e-01
-1.01882863e+00 2.85150945e-01 -2.04943959e-02 2.24651992e-01
9.51352865e-02 -1.86458752e-01 1.12637925e+00 -2.41883487e-01
1.78759396e-01 1.91860080e-01 -4.19734895e-01 5.05064547e-01
2.96918511e-01 2.81605691e-01 -8.10609579e-01 -7.84051776e-01
-5.54887950e-01 1.50479078e-01 6.28889054e-02 3.94739121e-01
1.04362303e-02 -9.96977568e-01 -1.11453736e+00 -3.36285532e-01
-2.10771590e-01 -5.02514601e-01 -1.54390991e-01 1.12627745e+00
-4.59765047e-01 2.85498649e-01 1.73105882e-03 -3.46854210e-01
-1.45073950e+00 1.78807214e-01 6.39181554e-01 -3.32161158e-01
-1.40755206e-01 1.18140852e+00 -3.48493382e-02 -4.88230646e-01
5.16152799e-01 -1.37068793e-01 -7.64008909e-02 5.68100929e-01
6.71458721e-01 8.07682812e-01 1.04831114e-01 -4.86048639e-01
-1.58434868e-01 -6.64173812e-02 -3.51858258e-01 -2.77967989e-01
1.25779259e+00 8.19418430e-02 -2.03076020e-01 9.91235435e-01
1.51978421e+00 9.08252150e-02 -9.17195916e-01 -1.33763000e-01
-3.79969776e-02 -9.48471576e-02 2.70104587e-01 -1.06288934e+00
-8.61238241e-01 1.14320660e+00 2.15266332e-01 5.48647463e-01
2.15015441e-01 -2.49271557e-01 5.89225829e-01 2.34304115e-01
-2.59586215e-01 -1.34844089e+00 1.75298274e-01 7.82830060e-01
7.68477201e-01 -1.05738103e+00 1.06230773e-01 7.18128979e-02
-6.11659288e-01 6.86945617e-01 6.17318094e-01 -2.59578675e-02
8.13621283e-01 3.70756000e-01 3.26887816e-01 -4.19946194e-01
-9.63445067e-01 4.70357984e-02 -2.66635176e-02 1.93971232e-01
6.41089916e-01 -6.36265054e-02 -1.31541654e-01 -2.56979372e-02
-7.87439466e-01 -7.51385510e-01 4.28528458e-01 6.68241680e-02
-3.64947557e-01 -4.24567848e-01 -1.28947943e-01 7.09574461e-01
-1.16545630e+00 -4.02647138e-01 -7.79566050e-01 8.36125433e-01
1.04513772e-01 1.12020898e+00 1.47544980e-01 -4.70935434e-01
1.16606683e-01 2.72174597e-01 1.22706525e-01 -4.88633484e-01
-1.44457185e+00 -2.10672945e-01 6.75339282e-01 -2.37391993e-01
-3.38724822e-01 -6.98547840e-01 -6.29951179e-01 -9.97466922e-01
-6.73749387e-01 6.13078848e-02 6.92930222e-01 7.52689123e-01
1.79284308e-02 4.89891559e-01 7.05357015e-01 -4.33994770e-01
-6.27501905e-01 -1.38348198e+00 -5.72963595e-01 3.15574169e-01
7.54909337e-01 -3.09424758e-01 -8.14780891e-01 -4.46096539e-01] | [8.65045166015625, 10.367128372192383] |
843d9f3f-a1bc-469a-9fd5-0a29a3639077 | optimal-image-smoothing-and-its-applications | 2003.08210 | null | https://arxiv.org/abs/2003.08210v1 | https://arxiv.org/pdf/2003.08210v1.pdf | Optimal Image Smoothing and Its Applications in Anomaly Detection in Remote Sensing | This paper is focused on deriving an optimal image smoother. The optimization is done through the minimization of the norm of the Laplace operator in the image coordinate system. Discretizing the Laplace operator and using the method of Euler-Lagrange result in a weighted average scheme for the optimal smoother. Satellite imagery can be smoothed by this optimal smoother. It is also very fast and can be used for detecting the anomalies in the image. A real anomaly detecting problem is considered for the Qom region in Iran. Satellite image in different bands are smoothed. Comparing the smoothed and original images in different bands, the maps of anomalies are presented. Comparison between the derived method and the existing methods reveals that it is more efficient in detecting anomalies in the region. | ['M. Kiani'] | 2020-03-17 | null | null | null | null | ['image-smoothing'] | ['computer-vision'] | [ 6.33297265e-02 -1.41916856e-01 4.00217503e-01 -1.58508852e-01
-5.10707796e-01 -3.00054193e-01 2.08766609e-01 1.39833242e-01
-5.65967321e-01 5.51827729e-01 -7.07829371e-02 -3.93455625e-01
-3.02549630e-01 -7.39635110e-01 -7.11739138e-02 -1.05454993e+00
-4.95638967e-01 -2.34662876e-01 2.63335884e-01 -3.96968305e-01
6.36484563e-01 9.04649794e-01 -1.38844943e+00 -1.93047360e-01
1.23850799e+00 9.75793302e-01 3.42623919e-01 8.52943182e-01
-2.22854137e-01 1.94192052e-01 -6.57861292e-01 3.39472115e-01
5.88974357e-01 -4.42626894e-01 -7.47715116e-01 7.11526513e-01
2.63429701e-01 -2.36226544e-01 1.20740421e-01 1.50780475e+00
2.29709789e-01 5.86912990e-01 7.92006075e-01 -9.25289214e-01
-6.10065341e-01 -9.61726755e-02 -1.33843958e+00 7.86566198e-01
1.05549134e-01 -3.87013406e-01 4.24320400e-01 -7.65210688e-01
3.35688323e-01 1.40378666e+00 8.42797458e-01 -2.86281258e-01
-1.22480905e+00 -2.97084272e-01 -1.64160773e-01 3.62328768e-01
-1.51925492e+00 1.83793470e-01 5.59755385e-01 -3.83470297e-01
5.52546263e-01 5.94076455e-01 5.78325987e-01 -2.92145669e-01
6.82231963e-01 2.48906910e-01 1.17646646e+00 -6.06648147e-01
-2.16555402e-01 -2.38223444e-03 2.97761112e-01 6.13554001e-01
2.84937263e-01 -2.05350220e-01 1.47146121e-01 -2.62580454e-01
5.70599377e-01 7.79059678e-02 -2.77655095e-01 5.27137995e-01
-7.40769088e-01 1.07085025e+00 4.23779607e-01 5.33401847e-01
-7.12257206e-01 -4.11338896e-01 3.61411095e-01 4.46119189e-01
9.85718608e-01 3.24037045e-01 -1.10566750e-01 3.06246191e-01
-1.31105387e+00 3.80700141e-01 2.23951131e-01 2.25835279e-01
5.87265432e-01 4.98326331e-01 1.69044167e-01 6.51895106e-01
5.20607531e-01 8.00838113e-01 2.50844628e-01 -1.05710900e+00
-2.82504857e-02 4.46938932e-01 1.37786835e-01 -1.19078910e+00
-2.67362714e-01 -1.37600541e-01 -7.14780509e-01 8.54035378e-01
4.10990983e-01 -4.35850382e-01 -9.56191599e-01 8.94221365e-01
5.06975412e-01 3.35411429e-01 1.17003582e-01 6.82360590e-01
3.04668874e-01 1.40846825e+00 -5.42531647e-02 -3.55748355e-01
1.38302636e+00 -4.57819998e-01 -1.17272747e+00 2.03582197e-01
4.35205996e-01 -1.04094207e+00 7.02830195e-01 5.65727949e-01
-8.20451617e-01 -3.30407351e-01 -8.31324160e-01 3.74805123e-01
-4.51062232e-01 9.84849632e-02 2.43657261e-01 3.26277286e-01
-1.06218588e+00 7.05587447e-01 -6.07275963e-01 -5.83040178e-01
-1.67884082e-02 5.46496287e-02 -4.32959676e-01 6.80903852e-01
-8.71811807e-01 1.09142077e+00 6.19533062e-01 2.84870595e-01
1.19787022e-01 -4.18844134e-01 -8.24290991e-01 -1.50930867e-01
-2.01516807e-01 2.38206163e-01 8.23236644e-01 -1.23007393e+00
-1.04400027e+00 8.80092025e-01 -3.48264396e-01 -3.00365925e-01
3.76546949e-01 -6.83232695e-02 -8.74179482e-01 1.69138506e-01
3.41922879e-01 -1.00669838e-01 9.44730043e-01 -9.74504530e-01
-1.23612535e+00 -4.51097876e-01 -6.74116910e-01 -4.19369973e-02
1.92247838e-01 2.84630507e-01 1.45515546e-01 -7.84224629e-01
5.95111668e-01 -6.47529840e-01 -3.99811119e-01 -2.73486763e-01
-1.58221751e-01 -2.71251231e-01 1.42632580e+00 -1.57701397e+00
1.23214126e+00 -2.57441831e+00 -4.45710689e-01 7.82097101e-01
-3.97978425e-01 1.09972000e-01 1.34984791e-01 4.35382694e-01
-2.38296881e-01 6.50659427e-02 -5.11410832e-01 2.16206715e-01
-4.94956851e-01 2.75056154e-01 -5.10550201e-01 1.03283107e+00
8.23767036e-02 -2.19561726e-01 -6.85334623e-01 -6.62285924e-01
3.00543636e-01 3.28887641e-01 -1.54149741e-01 2.44296081e-02
5.00027478e-01 7.25524366e-01 -4.76092935e-01 4.91155237e-01
1.49450016e+00 3.56833696e-01 -4.42288131e-01 -7.79597834e-02
-7.86990643e-01 -6.10811830e-01 -1.60035741e+00 8.75652969e-01
4.33835573e-03 1.04715192e+00 4.62074161e-01 -1.32069445e+00
1.25566328e+00 4.39910769e-01 5.71256757e-01 -5.40197492e-01
-7.38924146e-02 2.42218852e-01 -3.53737622e-01 -9.57243621e-01
6.63246870e-01 -1.07081622e-01 3.24327826e-01 1.10516183e-01
-4.58995253e-01 -2.25341618e-01 3.97579610e-01 1.00962222e-01
2.42903486e-01 -1.99077129e-01 4.50319171e-01 -9.32725966e-01
1.01909435e+00 2.54358143e-01 6.20693445e-01 6.15910769e-01
-2.68733352e-01 1.20867994e-02 4.76839513e-01 -3.50165814e-01
-1.13846159e+00 -1.00665915e+00 -7.53992140e-01 7.63664961e-01
1.16374865e-01 2.29034424e-01 -8.27001989e-01 -1.91131532e-01
1.85402751e-01 8.35478187e-01 -4.43641663e-01 4.17645484e-01
-3.54030252e-01 -1.06023896e+00 3.52210879e-01 -9.89131778e-02
1.02690649e+00 -7.89393723e-01 -5.41440010e-01 1.00419290e-01
-7.30856135e-02 -6.55875146e-01 -2.43376985e-01 -4.30542231e-01
-1.46823168e+00 -7.94121981e-01 -7.74904549e-01 -6.29934669e-01
9.91452575e-01 3.37665081e-01 5.08879125e-01 1.27951041e-01
-3.68078262e-01 1.81939065e-01 -1.81223184e-01 -8.00233305e-01
-3.44857633e-01 -6.05959415e-01 -1.86970443e-01 2.91467547e-01
3.20434600e-01 -2.20055252e-01 -4.24300581e-01 -4.77242982e-03
-9.80541050e-01 -5.40598929e-01 2.66851366e-01 7.35088468e-01
3.99427950e-01 8.28178167e-01 3.14044833e-01 -7.97249675e-01
8.62888932e-01 -6.22905374e-01 -1.17607391e+00 3.85333300e-02
-6.41939223e-01 4.78236116e-02 1.77957296e-01 1.36370927e-01
-1.66627944e+00 -2.13377690e-03 -8.40558186e-02 4.25503880e-01
-3.81821513e-01 5.55806994e-01 5.29512823e-01 -3.28218997e-01
6.96320057e-01 2.02641204e-01 2.49421105e-01 -8.65108669e-01
1.56698316e-01 9.07952428e-01 7.88582504e-01 -7.47905597e-02
8.96369219e-01 8.40611517e-01 3.31296831e-01 -1.62265956e+00
-6.72718883e-01 -8.65155339e-01 -5.54400384e-01 -4.88265723e-01
1.15879750e+00 -4.26313251e-01 -2.79071659e-01 6.55170202e-01
-1.06976700e+00 3.86913925e-01 -7.02508092e-02 7.01665878e-01
-5.77180684e-02 8.02417755e-01 -4.18637246e-01 -1.31612384e+00
-3.38126272e-01 -9.99969900e-01 8.27708721e-01 6.22931480e-01
3.30684818e-02 -1.40013421e+00 1.99977875e-01 -1.64293632e-01
4.76905704e-01 7.16880381e-01 5.86027920e-01 -3.66447628e-01
-1.74115241e-01 -4.72763807e-01 -3.62853318e-01 4.44102138e-01
3.39678884e-01 4.42800820e-01 -8.27616870e-01 -2.67366797e-01
3.72869194e-01 5.14154255e-01 6.96200132e-01 7.74581671e-01
9.48669434e-01 -3.01296026e-01 -4.95949909e-02 7.74997830e-01
1.82074201e+00 4.59179074e-01 7.05630064e-01 8.17477942e-01
4.18752246e-02 5.47819793e-01 1.02364397e+00 5.29517293e-01
-3.18061978e-01 2.75050163e-01 2.91958511e-01 -5.52331448e-01
6.86175287e-01 4.16372418e-01 3.03595930e-01 3.04125696e-01
-6.47794008e-01 3.86253029e-01 -8.89543533e-01 6.43905401e-01
-1.77490759e+00 -1.32525659e+00 -1.01855576e+00 2.07820368e+00
3.57561231e-01 -2.18359709e-01 -5.37610389e-02 4.37386960e-01
8.88522208e-01 2.32605398e-01 1.26652315e-01 -9.89312351e-01
-1.17922299e-01 2.77639776e-01 1.20919287e+00 1.01083243e+00
-1.30559719e+00 5.23032129e-01 7.31849051e+00 6.42094553e-01
-1.47007179e+00 5.00710160e-02 1.89866111e-01 5.90227187e-01
1.18291631e-01 2.25846276e-01 -2.46786192e-01 5.62199235e-01
6.49170399e-01 -5.53597450e-01 9.21007022e-02 6.95626497e-01
1.02366030e+00 -1.01857483e+00 2.16211200e-01 8.70718002e-01
-2.40329012e-01 -9.59718287e-01 -3.03738937e-02 2.05004752e-01
8.48498821e-01 -2.68412810e-02 -2.03859910e-01 -3.14191252e-01
-1.28135830e-01 -6.68402612e-01 4.71938401e-01 1.05941606e+00
1.49718553e-01 -1.16106558e+00 1.10762835e+00 1.68042138e-01
-1.28493202e+00 -6.74316660e-03 -5.72557449e-01 -2.05906704e-01
3.29273760e-01 8.60847414e-01 -6.13336205e-01 5.79744101e-01
9.61092412e-01 5.34140050e-01 -5.95755994e-01 1.39255583e+00
1.31748676e-01 7.17130899e-01 -3.69384855e-01 5.06338537e-01
6.40202224e-01 -1.22170687e+00 8.88823211e-01 1.61615622e+00
8.47270787e-01 1.98608220e-01 6.61033485e-03 5.75034022e-01
8.37767720e-01 4.61670965e-01 -8.23989749e-01 1.22489460e-01
8.04801732e-02 1.06519854e+00 -7.82346427e-01 -5.01340687e-01
-4.56231773e-01 9.49512243e-01 -5.30811965e-01 7.19903529e-01
-3.78826499e-01 -9.56198692e-01 6.37427330e-01 -1.72897980e-01
2.12318078e-02 -2.37286896e-01 -5.05644202e-01 -6.65503621e-01
-2.34066084e-01 -5.79942286e-01 7.42705464e-01 -5.66015899e-01
-8.33865523e-01 4.96248275e-01 3.33832473e-01 -1.03667879e+00
-6.16052235e-03 -5.99037468e-01 -1.21091712e+00 1.52227056e+00
-1.21759248e+00 -5.69997728e-01 -3.10372233e-01 6.18794382e-01
5.23007989e-01 -1.71200559e-01 5.46572089e-01 2.20158070e-01
-4.66951251e-01 -2.00459808e-01 6.74651682e-01 -3.87936048e-02
3.76728296e-01 -1.40746248e+00 2.15113372e-01 1.40068066e+00
-5.08392334e-01 1.29481584e-01 1.42035627e+00 -7.27875412e-01
-4.70213264e-01 -6.10021114e-01 9.89075124e-01 1.61763966e-01
6.78389013e-01 4.40470338e-01 -1.25657761e+00 6.49837196e-01
4.72783029e-01 -2.27421120e-01 2.68974066e-01 -5.57028770e-01
4.78848696e-01 -2.03510493e-01 -1.65563488e+00 3.57457101e-01
-1.21401072e-01 -2.44944319e-01 -6.67011499e-01 3.96189302e-01
-1.77427437e-02 -1.11965507e-01 -8.78624916e-01 7.85286501e-02
1.14874512e-01 -1.05122387e+00 7.38428891e-01 -4.37524855e-01
-2.92512298e-01 -5.76022804e-01 1.97154522e-01 -1.48048353e+00
-4.26502019e-01 -4.99563813e-01 7.51910925e-01 1.03852129e+00
2.09768921e-01 -1.11116385e+00 2.37376541e-01 1.62452877e-01
1.15825683e-01 -2.03161642e-01 -8.34153295e-01 -7.43349493e-01
-4.27133441e-01 -1.40667096e-01 2.92362422e-01 7.91421890e-01
-7.55258724e-02 -4.61317807e-01 -2.80959457e-01 7.66106725e-01
1.15960181e+00 -3.69386971e-01 4.63136792e-01 -1.63470578e+00
3.62971306e-01 -2.17364177e-01 -7.47794151e-01 -3.27834308e-01
-1.05608307e-01 -3.42969328e-01 -5.76884262e-02 -1.54222190e+00
-4.33786690e-01 1.35398405e-02 -2.03980915e-02 5.35157025e-02
-2.38640726e-01 5.44642061e-02 2.26856265e-02 2.94794202e-01
4.14204925e-01 -5.49008697e-02 1.12023234e+00 1.62061051e-01
-1.55058071e-01 1.56775430e-01 -2.33219936e-01 1.06596339e+00
1.06550050e+00 -3.98632973e-01 -1.99502215e-01 -3.19785886e-02
-2.27711305e-01 -5.42531163e-02 1.34996727e-01 -8.66668046e-01
-2.00969666e-01 -5.04561722e-01 1.21678233e-01 -8.87827814e-01
-1.32092252e-01 -8.94535005e-01 2.09662810e-01 4.69279677e-01
1.80855334e-01 4.52289790e-01 2.91725636e-01 3.97050261e-01
-5.64632952e-01 -7.51268446e-01 1.37737226e+00 -1.91531464e-01
-1.20438004e+00 -1.29479080e-01 -7.47321486e-01 -3.26594919e-01
1.26432669e+00 -4.07981753e-01 4.90851142e-02 -5.26852548e-01
-8.53595495e-01 7.16711059e-02 2.51821011e-01 -2.46415392e-01
4.46008861e-01 -1.17262435e+00 -9.43754137e-01 2.42410451e-01
-4.22927529e-01 -2.75992036e-01 3.70510668e-01 1.36038291e+00
-1.48857665e+00 1.14253096e-01 -4.41291958e-01 -6.44815326e-01
-1.62740088e+00 1.07487716e-01 5.39058328e-01 -2.09824438e-03
-7.46163905e-01 4.32688117e-01 -2.11946413e-01 1.61808223e-01
-1.04431354e-01 -2.44195879e-01 -6.61651254e-01 1.72739401e-01
7.53776014e-01 9.05657709e-01 -2.73188800e-01 -9.53200996e-01
-1.43791676e-01 9.27858770e-01 4.35932785e-01 -3.24079692e-01
1.28146482e+00 -4.79742795e-01 -7.35431612e-01 4.06667143e-01
1.05840504e+00 3.60974729e-01 -9.95715201e-01 -1.75885558e-02
5.04700780e-01 -9.81749833e-01 5.02492964e-01 3.99068836e-03
-9.96492982e-01 4.56553251e-01 1.10021079e+00 8.19463849e-01
1.48782623e+00 -5.14371693e-01 5.49106538e-01 2.99606562e-01
-3.16826999e-01 -1.47055435e+00 -7.04102218e-01 5.68101048e-01
9.89809632e-01 -1.15825653e+00 2.57942647e-01 -3.04480672e-01
-6.47193313e-01 1.45368934e+00 3.27439383e-02 -5.01165926e-01
9.77570415e-01 1.19081736e-01 5.21407783e-01 -2.06057891e-01
1.74146175e-01 -3.13823432e-01 2.98069984e-01 4.37344253e-01
6.15934610e-01 6.23967499e-03 -9.70988333e-01 -3.41406286e-01
4.50140005e-03 -1.96571037e-01 7.88217962e-01 9.48260367e-01
-1.11007833e+00 -6.67760491e-01 -1.44267774e+00 4.34360027e-01
-9.00819540e-01 1.89412698e-01 6.88982457e-02 7.91355312e-01
1.64937451e-01 8.28902483e-01 3.71930003e-01 3.84708345e-01
5.05206943e-01 1.77632764e-01 -1.82327643e-01 3.39489393e-02
-9.90537331e-02 7.98269361e-02 -7.69308582e-02 -2.76279330e-01
-4.90987688e-01 -8.05722177e-01 -1.54980993e+00 -4.54018325e-01
-3.92860144e-01 7.87762284e-01 9.42421615e-01 9.55052435e-01
-3.09129477e-01 2.22238481e-01 8.52766395e-01 -9.28186059e-01
-5.17140269e-01 -9.54564869e-01 -1.49853754e+00 5.39606690e-01
7.17456818e-01 -3.90650809e-01 -8.05147767e-01 1.67293370e-01] | [10.314562797546387, -2.224564790725708] |
dca28ae2-5edd-4704-b568-ea83e842fb10 | comparing-of-term-clustering-frameworks-for | 1901.09037 | null | http://arxiv.org/abs/1901.09037v1 | http://arxiv.org/pdf/1901.09037v1.pdf | Comparing of Term Clustering Frameworks for Modular Ontology Learning | This paper aims to use term clustering to build a modular ontology according
to core ontology from domain-specific text. The acquisition of semantic
knowledge focuses on noun phrase appearing with the same syntactic roles in
relation to a verb or its preposition combination in a sentence. The
construction of this co-occurrence matrix from context helps to build feature
space of noun phrases, which is then transformed to several encoding
representations including feature selection and dimensionality reduction. In
addition, the content has also been presented with the construction of word
vectors. These representations are clustered respectively with K-Means and
Affinity Propagation (AP) methods, which differentiate into the term clustering
frameworks. Due to the randomness of K-Means, iteration efforts are adopted to
find the optimal parameter. The frameworks are evaluated extensively where AP
shows dominant effectiveness for co-occurred terms and NMF encoding technique
is salient by its promising facilities in feature compression. | ['Ziwei Xu', 'Fabrice Guillet', 'Mounira Harzallah'] | 2019-01-25 | null | null | null | null | ['feature-compression'] | ['computer-vision'] | [ 1.95591733e-01 -1.28510654e-01 7.95229673e-02 -4.02712584e-01
-3.47555995e-01 -3.03159952e-01 5.00040591e-01 4.73742455e-01
-5.58159232e-01 4.21699643e-01 5.46590984e-01 1.14090264e-01
-8.94076407e-01 -8.02218676e-01 1.15875565e-01 -9.32625294e-01
-2.10160762e-01 4.92909849e-01 -2.05853313e-01 -3.32630277e-01
5.98493755e-01 4.05527413e-01 -2.11764741e+00 2.06474021e-01
7.39852667e-01 6.89261734e-01 6.41266286e-01 1.87499180e-01
-8.77001643e-01 2.60630757e-01 -4.03278232e-01 -2.10000828e-01
3.46287014e-03 -1.98790774e-01 -9.72983778e-01 2.70821184e-01
-5.07416368e-01 4.23443109e-01 1.82620168e-01 1.06793535e+00
1.76644266e-01 5.85815787e-01 1.00815296e+00 -8.40567052e-01
-5.67869365e-01 8.14037323e-01 -3.08924973e-01 2.27336600e-01
4.07702178e-01 -6.52139068e-01 1.17437482e+00 -1.00082910e+00
7.55663574e-01 1.18678474e+00 3.06129694e-01 2.40707770e-01
-8.88130844e-01 4.92844172e-02 -1.53811440e-01 5.40704370e-01
-1.74891829e+00 -2.22033449e-02 9.58039045e-01 -3.80648583e-01
1.33433878e+00 3.14984739e-01 6.33377612e-01 5.56975067e-01
5.82942218e-02 3.46217990e-01 5.89403510e-01 -9.20709133e-01
4.12506759e-01 4.40278143e-01 8.64459515e-01 1.57590926e-01
2.37508059e-01 -7.22454429e-01 -3.20899934e-01 -2.62774289e-01
-1.09149069e-01 6.18641637e-03 -2.99211927e-02 -1.19227886e-01
-7.23000586e-01 8.61063480e-01 1.70227736e-01 9.72101092e-01
-7.81110525e-01 -3.62455040e-01 4.23611581e-01 -1.97242876e-03
5.73171265e-02 5.13398707e-01 -5.03562212e-01 -8.46671388e-02
-6.04379296e-01 2.56489456e-01 5.86871088e-01 1.01496422e+00
8.23107958e-01 -2.99374282e-01 2.45616063e-02 1.14721167e+00
6.62662804e-01 8.08835849e-02 1.08930194e+00 -4.83869225e-01
3.14118236e-01 1.21052420e+00 -2.32720494e-01 -1.24514341e+00
-6.16499662e-01 -3.51353526e-01 -4.75854903e-01 -3.75969052e-01
-2.78946728e-01 1.16347924e-01 -7.19220638e-01 1.29463184e+00
5.34056425e-01 -3.04057181e-01 5.72474360e-01 5.39026856e-01
8.14280868e-01 8.98339033e-01 1.86195418e-01 -6.89283013e-01
1.95022631e+00 -4.78902310e-01 -9.28257287e-01 2.67893344e-01
5.78468382e-01 -7.26188123e-01 6.58892334e-01 3.68192673e-01
-6.18082225e-01 -5.03446341e-01 -8.97010326e-01 -4.00864193e-03
-8.97946835e-01 -1.51296005e-01 6.98442340e-01 7.20034599e-01
-8.62490892e-01 3.51480126e-01 -3.74527633e-01 -5.55076361e-01
6.16551898e-02 4.94540066e-01 -4.88701284e-01 3.34879830e-02
-1.28107643e+00 8.64905119e-01 1.05573630e+00 -1.46437481e-01
-1.29823372e-01 -3.63603234e-01 -8.07001173e-01 1.88589543e-01
1.92420650e-02 -6.72996581e-01 3.80216360e-01 -7.39755809e-01
-9.23736513e-01 6.53847039e-01 -4.04860258e-01 -4.82755959e-01
-4.44162220e-01 1.65546358e-01 -7.18934059e-01 2.85274118e-01
1.87679619e-01 3.90285522e-01 7.97628462e-01 -1.25691521e+00
-9.09012675e-01 -6.26996100e-01 -1.73504174e-01 7.73543537e-01
-9.95965481e-01 2.82956064e-01 -5.52445948e-01 -4.39628422e-01
7.30753601e-01 -5.71006238e-01 -1.48608878e-01 -9.33104753e-01
-1.01571269e-01 -7.94160247e-01 8.16915154e-01 -5.02981484e-01
1.84060991e+00 -2.07152605e+00 5.19511104e-01 8.52213442e-01
1.67576328e-01 -1.73871428e-01 4.18655962e-01 7.51032770e-01
-1.80286273e-01 5.42393364e-02 -2.98647225e-01 8.19077939e-02
1.06031679e-01 4.73574758e-01 -1.36428550e-02 -2.23880466e-02
-1.79808378e-01 2.70263255e-01 -5.93774021e-01 -9.35603619e-01
1.49298370e-01 4.67116028e-01 -4.34626937e-01 -2.44214073e-01
1.10547364e-01 -1.43214449e-01 -7.18162596e-01 6.01637423e-01
5.73021472e-01 8.45464319e-02 6.09073341e-01 -4.46057081e-01
-2.28418216e-01 2.78107356e-02 -1.33988833e+00 1.58470953e+00
-1.25683755e-01 -1.09081969e-01 -2.79462665e-01 -1.26374865e+00
1.07475221e+00 3.39159191e-01 8.06492805e-01 -4.53657001e-01
5.16736686e-01 1.62460580e-01 1.02327354e-02 -9.14592147e-01
6.72264397e-01 1.91724468e-02 -8.02244022e-02 1.44793028e-02
3.64413172e-01 9.31143910e-02 7.64103889e-01 1.65033907e-01
7.66952753e-01 -1.08598948e-01 5.17752349e-01 -5.48402965e-01
1.03824413e+00 1.54977232e-01 4.28101927e-01 1.52584717e-01
2.97416210e-01 2.34680936e-01 1.16037853e-01 -2.28515968e-01
-8.56505990e-01 -7.72439718e-01 -7.43386805e-01 1.05913830e+00
1.19601134e-02 -8.22872877e-01 -8.15742552e-01 -3.20032716e-01
-1.02417819e-01 7.81711400e-01 -6.71507001e-01 -6.70341551e-02
-2.28839308e-01 -1.04704380e+00 2.13937894e-01 -1.87629368e-02
2.71238089e-01 -1.24705780e+00 -5.09192228e-01 4.50417817e-01
-4.72404202e-03 -4.33835745e-01 3.32842261e-01 6.28673673e-01
-7.81300187e-01 -8.99311244e-01 -3.02494168e-01 -1.00607157e+00
7.51907527e-01 1.38499811e-01 7.91068375e-01 1.30188555e-01
-3.65262866e-01 3.21399450e-01 -1.08347702e+00 -3.59557062e-01
-1.31652191e-01 1.23790957e-01 2.27247342e-01 8.15691054e-02
1.16468263e+00 -7.81140804e-01 -2.71951765e-01 -1.87319934e-01
-1.09732759e+00 -4.85243708e-01 3.10486972e-01 5.86889386e-01
5.52061737e-01 7.18789041e-01 2.85680294e-01 -6.69649422e-01
9.02854621e-01 -8.13510954e-01 -1.45088568e-01 3.11439663e-01
-8.56509805e-01 2.21372440e-01 3.07439983e-01 8.48920494e-02
-1.05720675e+00 4.70135212e-02 -2.11884757e-03 -5.59087321e-02
-2.57038087e-01 8.27830195e-01 -2.05378294e-01 2.91140705e-01
5.08835852e-01 5.27784109e-01 -1.33074075e-01 -4.43507731e-01
3.66646796e-01 9.04329658e-01 7.92681873e-02 -7.86584198e-01
5.10584593e-01 2.23558068e-01 -5.63572608e-02 -1.17921591e+00
-2.11536989e-01 -8.05418730e-01 -8.16243052e-01 -1.29480481e-01
1.09955490e+00 -4.88167554e-01 -4.14694577e-01 -1.07913628e-01
-9.21503425e-01 1.09896481e+00 -2.35807240e-01 5.27359426e-01
-2.13889778e-01 6.27699256e-01 -2.57333905e-01 -1.05605233e+00
-5.43265224e-01 -1.00427973e+00 5.72367430e-01 2.86506206e-01
-4.00579125e-01 -8.23508978e-01 2.08528899e-02 3.40218842e-01
2.02244129e-02 -1.64057046e-01 1.36950076e+00 -1.00122690e+00
2.98678372e-02 -1.74929544e-01 2.38424122e-01 3.09076130e-01
3.34492445e-01 2.10042913e-02 -7.43951738e-01 3.77641395e-02
2.91175961e-01 3.13457489e-01 6.06067419e-01 2.83086568e-01
7.99829006e-01 -1.46918520e-01 -4.13524479e-01 4.18046296e-01
1.64090478e+00 8.55361223e-01 6.66728616e-01 6.36879563e-01
5.45626521e-01 9.14163172e-01 6.48477435e-01 6.31188512e-01
2.85769522e-01 6.04002595e-01 1.74736589e-01 6.00205839e-01
3.66570711e-01 2.69432068e-01 -1.71024531e-01 1.29834926e+00
-1.46955624e-01 -1.40409395e-01 -8.77748191e-01 6.15837753e-01
-1.62827516e+00 -1.07933950e+00 -2.78904289e-01 1.74392152e+00
5.43217719e-01 -7.11493641e-02 -4.26076427e-02 7.54716575e-01
7.96718299e-01 -1.57383218e-01 7.91610852e-02 -5.50569236e-01
-2.60052890e-01 1.78468391e-01 2.13544384e-01 3.06691498e-01
-9.83394325e-01 9.35937643e-01 5.19285393e+00 1.12604237e+00
-5.04958332e-01 8.97150412e-02 7.71852136e-02 1.50494218e-01
-5.70786655e-01 1.64418053e-02 -8.88686895e-01 7.12265849e-01
7.80113757e-01 -3.22541952e-01 3.19395632e-01 7.83031404e-01
1.49799526e-01 -1.99165925e-01 -4.98293132e-01 1.03853905e+00
3.94462228e-01 -9.65193331e-01 5.06457150e-01 1.00246042e-01
5.05755842e-01 -4.68542367e-01 -1.72788262e-01 2.56363302e-01
-3.57731953e-02 -5.98561764e-01 3.35942447e-01 4.39983398e-01
5.42678684e-02 -1.26872766e+00 7.50678599e-01 1.47299021e-01
-1.44370663e+00 -3.89597058e-01 -7.43989944e-01 1.52062356e-01
1.78553089e-01 4.33928847e-01 -7.16613770e-01 1.04543900e+00
6.61777020e-01 5.35215974e-01 -5.62610805e-01 7.28724599e-01
1.61702752e-01 3.67178410e-01 -5.04134417e-01 -3.76462340e-01
4.66233075e-01 -6.30254924e-01 5.32386780e-01 1.24303353e+00
4.40228432e-01 2.81432837e-01 -5.25401533e-03 5.03601134e-01
2.24491164e-01 1.02400780e+00 -5.77241123e-01 5.63685745e-02
7.63728619e-01 1.41379738e+00 -1.04774928e+00 -2.76617348e-01
-3.54800284e-01 7.67790556e-01 1.38240010e-01 4.81859967e-02
-5.24305522e-01 -5.49025178e-01 6.10288143e-01 -1.54018462e-01
3.19449246e-01 -5.24079055e-03 -3.04628760e-01 -7.02630579e-01
1.64868891e-01 -5.34605324e-01 6.04477465e-01 -5.81181765e-01
-1.08097470e+00 8.92714620e-01 3.36171597e-01 -1.24055290e+00
-1.82556629e-01 -4.98255581e-01 -1.99370176e-01 8.21799397e-01
-1.06518233e+00 -9.08193469e-01 -1.26539217e-02 8.29207599e-01
7.52811909e-01 -5.47186613e-01 1.14467585e+00 3.96212220e-01
-4.10285681e-01 9.37253311e-02 2.38430411e-01 -1.72659263e-01
1.79594025e-01 -1.12933350e+00 -3.54874074e-01 7.09064364e-01
2.52261668e-01 1.11829722e+00 7.35045373e-01 -7.23092437e-01
-1.31677604e+00 -6.17561340e-01 1.33374286e+00 -3.10766876e-01
5.89850366e-01 1.72683671e-01 -8.19854558e-01 1.56831190e-01
1.84640139e-01 -4.87138569e-01 1.07959569e+00 2.62028217e-01
9.16671194e-03 -2.12457613e-03 -1.21495700e+00 5.09244502e-01
8.59711111e-01 -3.00887376e-01 -1.25402999e+00 3.33553404e-01
6.20827734e-01 3.34975302e-01 -1.03285110e+00 5.78165762e-02
2.74229407e-01 -5.99255204e-01 9.45334852e-01 -6.65773094e-01
1.59193173e-01 -5.50592065e-01 -4.92151797e-01 -9.07470167e-01
-6.87205732e-01 -9.77072641e-02 1.33768409e-01 1.67404640e+00
5.29399276e-01 -2.58590490e-01 5.25158107e-01 4.18023765e-01
-8.48977491e-02 -4.60226893e-01 -9.68727052e-01 -4.93780017e-01
-3.24461222e-01 -4.79223549e-01 6.32364571e-01 1.24512339e+00
4.96881008e-01 4.83159602e-01 -4.37543839e-02 1.78431153e-01
5.27821660e-01 -1.12877488e-01 8.74651968e-02 -1.54705310e+00
4.64272164e-02 -5.07212043e-01 -7.73023009e-01 -3.90233606e-01
8.63374099e-02 -1.26043534e+00 -2.06470430e-01 -1.46466231e+00
1.66216537e-01 -3.48258585e-01 -5.15076697e-01 1.75151289e-01
-1.04806490e-01 -2.04311281e-01 5.00890985e-02 4.04273093e-01
-4.44419503e-01 5.24682045e-01 4.49661672e-01 -1.70141794e-02
-3.58710647e-01 -3.16501200e-01 -6.18637323e-01 5.29748857e-01
6.85139477e-01 -5.65391958e-01 -6.51212275e-01 -1.43341154e-01
3.93676579e-01 -1.64047554e-01 -3.92104596e-01 -1.04742658e+00
2.63682961e-01 1.51144583e-02 2.48605460e-01 -6.45844638e-01
3.42307687e-01 -1.44818306e+00 3.64471078e-01 1.21126443e-01
-1.82139546e-01 4.67206627e-01 -1.48222804e-01 5.09325266e-01
-3.66596639e-01 -9.83998597e-01 3.65120351e-01 -2.04791501e-01
-1.08546054e+00 -3.11669171e-01 -6.05128825e-01 -2.95987606e-01
1.08842885e+00 -7.19947815e-01 4.12098378e-01 1.94680855e-01
-9.39579487e-01 1.53048024e-01 3.38050723e-02 4.91204083e-01
6.26431823e-01 -1.37952793e+00 -4.60203171e-01 9.15615708e-02
2.57003307e-01 -2.80632138e-01 5.15968382e-01 4.25940126e-01
-5.87753356e-01 4.02528018e-01 -1.51476711e-01 -3.63056719e-01
-1.54907572e+00 8.46694529e-01 -1.64893001e-01 -1.27038747e-01
-6.17875159e-01 7.47801900e-01 -3.29910278e-01 -1.29328415e-01
1.77664369e-01 -8.83261785e-02 -1.33065939e+00 6.69881344e-01
4.79333580e-01 3.25460672e-01 2.48842180e-01 -1.08626342e+00
-5.73585749e-01 8.68151486e-01 -1.04538068e-01 -9.67965648e-02
1.31758654e+00 -3.47017407e-01 -5.67278028e-01 2.73936957e-01
1.25120687e+00 -1.69102445e-01 -2.21941844e-01 -1.13469891e-01
5.92936873e-01 -1.94339737e-01 -1.86200012e-02 -3.00772756e-01
-8.01228940e-01 1.42334357e-01 6.12801671e-01 4.76498961e-01
1.12938523e+00 -7.12705031e-02 3.90589565e-01 5.53465784e-01
4.81184572e-01 -1.46233678e+00 -5.03351688e-01 6.35034502e-01
4.75166678e-01 -6.71595633e-01 4.82889265e-02 -4.68347311e-01
-7.64168203e-01 1.33971596e+00 2.05720335e-01 -3.72756086e-02
9.13896441e-01 1.27580747e-01 -1.45224884e-01 -5.63511193e-01
-4.56532359e-01 -5.07231653e-01 3.37082326e-01 5.19923210e-01
4.19375837e-01 3.54681760e-02 -1.23664808e+00 7.39567995e-01
-4.24133778e-01 -4.42429572e-01 7.90385976e-02 1.06101906e+00
-9.34401870e-01 -1.31462681e+00 -4.19314027e-01 4.32962358e-01
-3.92410010e-01 -3.16139102e-01 -1.52985886e-01 5.33292890e-01
6.33585453e-01 1.17473888e+00 1.97016612e-01 -4.04503077e-01
1.88080177e-01 6.72253549e-01 1.41166791e-01 -6.19956136e-01
-8.49236190e-01 1.73257977e-01 6.62377104e-02 -8.82078707e-02
-5.79388678e-01 -8.64812672e-01 -1.59016931e+00 9.01677758e-02
-2.13704079e-01 8.97045672e-01 9.89210725e-01 1.15734828e+00
2.15853810e-01 6.06915474e-01 6.88300848e-01 -4.81296778e-01
-9.22085717e-02 -1.05574977e+00 -6.71713948e-01 7.36762106e-01
-4.38702375e-01 -7.01286376e-01 -4.80308563e-01 1.31938294e-01] | [10.25118350982666, 8.496562004089355] |
6f4bc9f6-3a8f-42a0-8db9-a38481e089e0 | un-nouvel-algorithme-pour-la-detection-des | 2112.13280 | null | https://arxiv.org/abs/2112.13280v1 | https://arxiv.org/pdf/2112.13280v1.pdf | Un nouvel algorithme pour la detection des transferts horizontaux de genes partiels entre les especes et pour la classification des transferts inferes | In this article we are describing a new algorithm for detecting and validating partial horizontal gene transfers (HGT). The presented algorithm is based on a sliding window procedure which analyzes fragments of the given multiple sequence alignment. A bootstrap procedure incorporated in our method can be used to estimate the support of each inferred partial HGT. The new algorithm can be also applied to confirm or discard complete (i.e., traditional) horizontal gene transfers detected by any HGT inferring algorithm. While working on a full-genome scale, the introduced algorithm can be used to assess the level of mosaicism of the whole species genomes as well as the rates of complete and partial HGT underlying the evolution of the considered set of species. | ['Makarenkov Vladimir', 'Diallo Alpha Boubacar', 'Boc Alix'] | 2021-12-25 | null | null | null | null | ['multiple-sequence-alignment'] | ['medical'] | [ 6.66018724e-01 1.15385853e-01 -1.30615145e-01 -3.93368751e-02
-1.11303635e-01 -6.08981073e-01 6.27109110e-01 6.35028899e-01
-4.27973390e-01 9.14170980e-01 -2.78301865e-01 -6.42104745e-01
-3.24123681e-01 -9.41043139e-01 -5.53326368e-01 -9.60132599e-01
-6.70640841e-02 8.61370802e-01 7.14895964e-01 -1.44230679e-01
2.92568207e-01 5.42251706e-01 -1.61312222e+00 2.62965590e-01
1.02975976e+00 1.01400971e-01 4.96878505e-01 7.26328969e-01
-4.08364624e-01 -1.88118428e-01 -5.33424377e-01 -3.54976058e-01
2.66925320e-02 -6.84353530e-01 -7.00262606e-01 2.53221169e-02
-3.51894721e-02 -1.60557553e-01 3.55279177e-01 6.31946683e-01
1.86409846e-01 -5.00973761e-01 6.54322445e-01 -7.90889025e-01
3.55763853e-01 8.15338850e-01 -3.88164133e-01 8.91276002e-02
4.37869906e-01 9.03220326e-02 6.43077970e-01 -5.32644153e-01
1.21073282e+00 1.04474497e+00 5.84730029e-01 -1.42320260e-01
-1.28806138e+00 -1.93690836e-01 -6.48142755e-01 2.79771507e-01
-1.34271109e+00 1.40812863e-02 1.20622486e-01 -8.38634014e-01
9.29567754e-01 3.78893644e-01 1.01525128e+00 4.89894986e-01
2.90695786e-01 3.65719736e-01 1.13071954e+00 -2.35331059e-01
3.96286964e-01 -9.96625945e-02 1.69599876e-01 5.08891106e-01
1.02697897e+00 -2.69640088e-01 -2.10196838e-01 -5.21791518e-01
4.75036025e-01 -5.15158892e-01 -3.48304272e-01 -2.47326538e-01
-1.05702889e+00 7.39669502e-01 9.98254344e-02 5.80598533e-01
-6.05638921e-01 -3.33234221e-01 6.58746898e-01 4.18143198e-02
3.27742249e-01 -3.11482288e-02 -2.29522496e-01 -2.33750418e-01
-9.63512838e-01 -5.20717055e-02 7.21831083e-01 4.12257165e-01
9.17702436e-01 -2.01202691e-01 3.25183213e-01 2.18932524e-01
-1.25961572e-01 1.74048528e-01 2.09225863e-01 -4.86898363e-01
-4.82469909e-02 6.02633238e-01 2.15064019e-01 -7.16396451e-01
-2.78238773e-01 -2.66635239e-01 -6.04761124e-01 1.82610061e-02
6.48418725e-01 1.43402845e-01 -3.47846985e-01 1.41346276e+00
7.47036397e-01 3.99345934e-01 1.14943638e-01 8.38297755e-02
1.06476344e-01 6.40918136e-01 -2.17758983e-01 -7.05587924e-01
1.12163103e+00 -4.43405248e-02 -5.23242950e-01 4.35028166e-01
4.90159988e-01 -8.21414232e-01 2.74098933e-01 4.43783283e-01
-6.98817968e-01 -3.34129393e-01 -9.73804891e-01 3.48921329e-01
-1.28724635e-01 -5.37743717e-02 1.55960634e-01 8.28546107e-01
-8.21375608e-01 8.39500248e-01 -8.03786159e-01 -8.55218768e-01
-2.52780497e-01 7.21639395e-02 -4.29351687e-01 -1.60110891e-02
-6.74224854e-01 8.09124827e-01 9.78306472e-01 3.52898121e-01
-4.53294128e-01 -1.82133570e-01 -1.24956742e-01 2.38581344e-01
2.63159275e-01 -6.54410779e-01 3.13580155e-01 -5.28012693e-01
-1.03058791e+00 1.20754862e+00 -2.56439656e-01 -4.96546954e-01
8.50611269e-01 9.77211669e-02 5.05647734e-02 3.24424118e-01
1.90616399e-02 8.63485709e-02 3.08707923e-01 -8.16796958e-01
-4.22499508e-01 -5.04608989e-01 -6.60625339e-01 -2.57528454e-01
2.37489998e-01 1.43789481e-02 -9.27626938e-02 -3.15167576e-01
1.75091922e-01 -1.10583961e+00 8.77993107e-02 -2.86998928e-01
-3.21448177e-01 -2.54518926e-01 3.29274297e-01 -9.47746813e-01
8.22535634e-01 -1.92815185e+00 6.37364268e-01 2.95389116e-01
-3.44848484e-01 4.95069087e-01 -1.64430961e-03 1.01584077e+00
-1.59420236e-03 -6.81236610e-02 -6.16334081e-01 3.19899827e-01
-2.38270968e-01 4.25315976e-01 -8.94491822e-02 7.32019126e-01
-7.96868876e-02 3.08177650e-01 -5.64464867e-01 -1.97010815e-01
2.67236650e-01 5.96046925e-01 -2.10835755e-01 1.72132507e-01
-1.69745788e-01 4.51647282e-01 -1.28350899e-01 1.03633255e-01
1.00335264e+00 8.15473646e-02 7.80245483e-01 2.15503708e-01
-7.40619779e-01 1.33432224e-01 -1.02319574e+00 1.35741723e+00
2.79670507e-01 4.07107621e-01 -2.64809757e-01 -1.01639926e+00
1.19448125e+00 3.12810212e-01 3.08926016e-01 1.92638949e-01
8.16140324e-02 6.57375097e-01 2.34113932e-01 -3.39080274e-01
3.76435667e-01 -1.24827094e-01 3.98172975e-01 3.23652864e-01
1.11275405e-01 2.92414844e-01 3.99476588e-01 -1.21349335e-01
7.71735430e-01 3.09281707e-01 8.52711260e-01 -5.34413576e-01
8.93992424e-01 1.00967139e-01 5.28231859e-01 5.57883024e-01
1.51414454e-01 2.50008591e-02 6.76226974e-01 -3.58079642e-01
-1.43609190e+00 -8.15494001e-01 -5.40074646e-01 6.56983614e-01
-1.98925734e-01 -5.19355126e-02 -8.30252111e-01 3.94492527e-04
2.29502786e-02 5.79538345e-01 -3.38286281e-01 8.20861664e-03
-5.60507178e-01 -1.16932654e+00 7.66192138e-01 -2.19917729e-01
4.53778684e-01 -7.46879160e-01 -7.97179103e-01 3.99233341e-01
-4.86787796e-01 -7.74776042e-01 4.08366263e-01 1.60066068e-01
-9.74891901e-01 -1.32608724e+00 -6.37513697e-01 -3.61357689e-01
3.28815043e-01 4.84228805e-02 4.23470080e-01 2.66410440e-01
-5.62420070e-01 -1.45730674e-01 -3.41484547e-01 -9.40248929e-03
-1.02958250e+00 1.05258211e-01 -1.59278885e-01 -1.24300987e-01
2.17574567e-01 -3.85962784e-01 -2.67968327e-01 6.31287992e-01
-8.63329411e-01 -1.37873918e-01 2.63599247e-01 4.75121379e-01
2.52996117e-01 2.06459627e-01 2.64267474e-01 -7.07089007e-01
3.44801307e-01 -5.47494292e-01 -8.76259804e-01 7.92090416e-01
-4.50494021e-01 -1.43293180e-02 1.42422259e-01 -6.60145804e-02
-1.11942744e+00 -1.76495254e-01 -6.74203813e-01 4.47377741e-01
-3.88873935e-01 4.65444893e-01 1.05546266e-02 3.88965011e-04
3.26441795e-01 5.96478045e-01 2.86188759e-02 -5.78183055e-01
7.67070428e-02 5.69926918e-01 4.72932458e-01 -4.39597964e-01
5.04344642e-01 3.48133922e-01 4.02922183e-01 -1.21711016e+00
1.20036617e-01 -7.12489545e-01 -8.91627192e-01 -2.24127531e-01
8.92307281e-01 -5.77465475e-01 -9.04963493e-01 7.50720561e-01
-1.09131575e+00 -1.70305327e-01 1.45391911e-01 4.91882682e-01
-4.65418696e-01 9.84662175e-01 -6.72583818e-01 -9.27343369e-01
-1.78017706e-01 -7.36379027e-01 7.75052905e-01 -2.50746161e-01
-2.41351068e-01 -8.63946378e-01 6.01964772e-01 1.60155639e-01
1.55451521e-01 4.77865368e-01 1.07386732e+00 -7.08206892e-01
-4.19727534e-01 -2.63619214e-01 1.69160157e-01 -1.03362270e-01
5.71060702e-02 6.68891072e-01 -5.35039246e-01 -3.71488541e-01
6.25267103e-02 2.42960155e-01 7.94202268e-01 2.57305264e-01
-4.36506756e-02 3.49278450e-02 -4.12460953e-01 3.28690678e-01
1.60500193e+00 2.55934983e-01 6.61903679e-01 7.38293767e-01
-1.80392694e-02 8.34176302e-01 9.62297976e-01 5.61502814e-01
-3.18147331e-01 8.84261906e-01 2.45169431e-01 4.10742760e-01
2.88449049e-01 2.04496123e-02 2.11693555e-01 3.89635563e-01
-4.48905140e-01 -4.75931764e-01 -1.24443305e+00 2.71207124e-01
-1.75520921e+00 -1.28958654e+00 -8.08089077e-01 2.49490452e+00
4.28560287e-01 1.08243125e-02 3.29040200e-01 3.35906386e-01
1.25313604e+00 -3.46413016e-01 -3.91893506e-01 -5.73361635e-01
-4.63099629e-01 2.98161712e-02 3.49208713e-01 6.41799808e-01
-4.21204418e-01 4.27744716e-01 7.00261545e+00 5.60462296e-01
-9.47945356e-01 -3.37475181e-01 8.23175237e-02 4.39630747e-01
-3.27996731e-01 3.36557925e-01 -7.06676066e-01 2.46549308e-01
9.24300909e-01 -4.80764389e-01 2.03324139e-01 5.58590770e-01
5.68826124e-02 -3.11596870e-01 -4.70162749e-01 3.63582939e-01
-1.02281332e-01 -1.21884072e+00 -2.20256466e-02 3.45711738e-01
1.52827919e-01 1.53325006e-01 -4.49495733e-01 -2.88620085e-01
-5.89764528e-02 -1.76171750e-01 2.45531499e-01 1.16809875e-01
3.75425339e-01 -7.28010297e-01 1.00083852e+00 5.96157908e-01
-1.06675315e+00 2.02443004e-01 -5.55018663e-01 -2.20838264e-02
5.64137340e-01 7.80126870e-01 -1.23186636e+00 9.44590688e-01
2.13598490e-01 3.58702302e-01 -4.53189850e-01 1.22585249e+00
-3.61297190e-01 5.07750630e-01 -5.62742829e-01 2.90522445e-02
-2.82092928e-03 -6.89096451e-01 8.94865274e-01 1.23062539e+00
9.17701960e-01 -2.22105443e-01 -2.91934997e-01 3.96277130e-01
6.82458937e-01 3.18378657e-01 -4.22554165e-01 -6.08949251e-02
3.09251815e-01 7.57481039e-01 -1.25647974e+00 -6.07625306e-01
8.22191760e-02 1.11662543e+00 7.07762614e-02 -2.03363709e-02
-5.57711959e-01 -1.63154289e-01 6.70813978e-01 1.61776677e-01
6.92573786e-01 -1.23063751e-01 3.73591065e-01 -1.10539818e+00
-3.43664527e-01 -3.55344892e-01 7.41730571e-01 -5.08248031e-01
-6.03487134e-01 6.33368254e-01 1.88840911e-01 -6.94772184e-01
-4.55528945e-01 -4.97611254e-01 -5.44856787e-01 7.02370226e-01
-9.29189444e-01 -6.01788223e-01 -3.97844948e-02 3.43528658e-01
1.66279748e-02 1.82554424e-01 6.33623540e-01 -8.20502639e-03
-5.53247333e-01 -2.91431159e-01 6.14625216e-01 -3.69220376e-01
4.22420293e-01 -9.75744545e-01 3.01602900e-01 9.43211138e-01
-4.22212988e-01 8.26999485e-01 1.12334251e+00 -1.21642542e+00
-8.75224948e-01 -5.92804074e-01 9.26415741e-01 3.10326308e-01
6.33753061e-01 -3.08637410e-01 -1.06202745e+00 9.36647415e-01
3.70562553e-01 -7.35082507e-01 9.52162027e-01 -3.40213031e-02
-2.40575626e-01 1.98050469e-01 -1.23975170e+00 4.87512834e-02
5.80398083e-01 -2.84811556e-01 -6.85584605e-01 4.84926142e-02
3.01795006e-01 1.27272308e-01 -1.14641929e+00 4.00239795e-01
8.54535878e-01 -1.26703513e+00 8.72166634e-01 -2.78744370e-01
-1.42458037e-01 -5.81762016e-01 1.26464069e-01 -7.54302740e-01
-2.43206441e-01 -5.14520347e-01 6.63187265e-01 1.24151742e+00
-1.02787457e-01 -8.64178956e-01 3.42797697e-01 -2.41284877e-01
1.37875378e-02 2.05725476e-01 -1.34454298e+00 -6.70704484e-01
-1.71248376e-01 2.92910933e-01 5.43255091e-01 9.37208593e-01
6.52456433e-02 1.48824349e-01 -5.37931800e-01 5.92980906e-02
8.59962642e-01 3.66704673e-01 7.65386999e-01 -1.30715895e+00
-4.76195753e-01 -3.72606277e-01 -7.91701078e-01 -5.70140220e-02
-3.57824206e-01 -7.70968199e-01 3.64492834e-02 -1.21475959e+00
2.75285929e-01 -6.73324838e-02 -8.94736201e-02 3.19525413e-02
-1.63108319e-01 6.67158067e-02 -1.88239748e-04 2.42759258e-01
2.41451800e-01 6.08418845e-02 3.89457822e-01 4.88723695e-01
1.16442509e-01 2.51589552e-03 2.48305112e-01 5.50614476e-01
9.07660604e-01 -5.77002645e-01 -7.80601287e-03 8.51050317e-02
4.31973785e-01 1.46544933e-01 1.28106728e-01 -8.72771502e-01
-1.60167843e-01 -4.41695638e-02 1.58592351e-02 -6.87459171e-01
-6.85769767e-02 -5.50805748e-01 1.15714526e+00 1.48870897e+00
-5.44219688e-02 7.43061304e-04 2.42178679e-01 5.82510591e-01
-5.73253967e-02 -6.28628969e-01 7.90574253e-01 -1.59124330e-01
-3.63914132e-01 -2.98932165e-01 -7.27720499e-01 -7.59194613e-01
1.17037153e+00 -3.14299077e-01 -4.86795962e-01 1.97521985e-01
-7.39403307e-01 -2.86499023e-01 8.82286668e-01 2.75098234e-02
3.04867744e-01 -6.85620308e-01 -9.67886508e-01 -7.88397640e-02
2.67193675e-01 -8.16827953e-01 5.20836532e-01 1.06010413e+00
-1.32140243e+00 3.56696934e-01 -8.78926277e-01 -9.12240326e-01
-1.94668067e+00 1.10386682e+00 -9.87813771e-02 -1.99213579e-01
-5.19691527e-01 3.70458141e-02 -1.04484461e-01 -2.41957009e-01
-2.59669155e-01 2.03165874e-01 -2.17652380e-01 3.07927966e-01
3.02700162e-01 5.95931411e-01 -1.53999388e-01 -8.43753278e-01
-4.02552068e-01 8.42469692e-01 2.24052832e-01 8.07235762e-03
1.39075041e+00 -3.11288953e-01 -8.69974136e-01 6.80749357e-01
8.89991462e-01 4.39149514e-02 -5.61991394e-01 1.73987851e-01
4.12300378e-01 -5.38781166e-01 -5.37550211e-01 -2.39513665e-01
-1.10288322e-01 7.06475258e-01 4.37771916e-01 2.35515758e-01
8.80306482e-01 -1.07143953e-01 1.46373719e-01 2.39149034e-01
5.12130618e-01 -6.19447112e-01 -7.00699687e-01 -1.65494427e-01
4.28161681e-01 -4.42156345e-01 5.48115745e-02 -7.29659319e-01
-4.28904705e-02 1.47548044e+00 -1.66476458e-01 1.35586575e-01
-1.07955690e-02 1.44447714e-01 -4.56663251e-01 1.74250394e-01
-6.54745400e-01 -2.05992445e-01 -3.04279268e-01 3.24816346e-01
5.00493765e-01 9.82254595e-02 -1.27405298e+00 -6.88426733e-01
1.44308329e-01 2.53445208e-01 8.07348013e-01 7.84794509e-01
-8.16894948e-01 -1.35939634e+00 -5.33749163e-01 -1.43273249e-01
-1.76341698e-01 2.15420187e-01 -5.98324060e-01 9.22798693e-01
2.28047103e-01 5.53170204e-01 7.41101801e-02 -6.25247955e-02
-5.34903370e-02 4.21034366e-01 5.90277195e-01 -1.26592703e-02
-6.40415907e-01 5.15433013e-01 2.63770431e-01 -4.69495170e-02
-6.39510274e-01 -9.43512261e-01 -8.16098571e-01 -6.10665977e-01
-3.52121294e-01 6.02783620e-01 8.46819341e-01 5.66200912e-01
-8.53927657e-02 1.74895659e-01 2.42734849e-01 -3.34278196e-01
-3.32671590e-02 -9.77675855e-01 -6.77812815e-01 3.28245044e-01
-3.74828763e-02 -2.85812736e-01 -3.86118054e-01 -2.57457793e-01] | [4.858279705047607, 5.186275005340576] |
1a6c3b96-c843-4878-967b-376042751c02 | renderers-are-good-zero-shot-representation | 2306.10721 | null | https://arxiv.org/abs/2306.10721v1 | https://arxiv.org/pdf/2306.10721v1.pdf | Renderers are Good Zero-Shot Representation Learners: Exploring Diffusion Latents for Metric Learning | Can the latent spaces of modern generative neural rendering models serve as representations for 3D-aware discriminative visual understanding tasks? We use retrieval as a proxy for measuring the metric learning properties of the latent spaces of Shap-E, including capturing view-independence and enabling the aggregation of scene representations from the representations of individual image views, and find that Shap-E representations outperform those of the classical EfficientNet baseline representations zero-shot, and is still competitive when both methods are trained using a contrative loss. These findings give preliminary indication that 3D-based rendering and generative models can yield useful representations for discriminative tasks in our innately 3D-native world. Our code is available at \url{https://github.com/michaelwilliamtang/golden-retriever}. | ['David Shustin', 'Michael Tang'] | 2023-06-19 | null | null | null | null | ['neural-rendering', 'metric-learning', 'metric-learning'] | ['computer-vision', 'computer-vision', 'methodology'] | [-1.66085824e-01 1.42793491e-01 -1.43129721e-01 -5.65742671e-01
-7.42211699e-01 -7.71186590e-01 1.08860409e+00 -5.23282647e-01
-6.39026389e-02 2.26663336e-01 7.21772194e-01 -1.09759659e-01
-1.02123708e-01 -6.49571121e-01 -6.82234764e-01 -5.12874365e-01
1.19367786e-01 5.67909360e-01 -1.05670810e-01 -1.07379153e-01
2.00290322e-01 7.00313210e-01 -1.63698900e+00 3.02362740e-01
5.00022411e-01 7.51251578e-01 3.84498924e-01 6.83676124e-01
1.19815087e-02 6.96143985e-01 -3.27575624e-01 -3.79319608e-01
5.26751578e-01 -1.93217397e-01 -5.37121415e-01 2.17840914e-02
1.10338962e+00 -9.50125754e-01 -9.32655871e-01 7.24049628e-01
5.66350937e-01 3.70550692e-01 8.60642731e-01 -1.18587399e+00
-1.15326643e+00 -8.81854910e-03 -2.47765824e-01 5.32837212e-01
3.40404332e-01 2.78259903e-01 1.26740432e+00 -9.84111905e-01
9.16188657e-01 1.52150226e+00 2.55051017e-01 6.80325687e-01
-1.44896472e+00 -6.93788052e-01 2.38988131e-01 9.70258042e-02
-1.15580416e+00 -8.03450704e-01 7.31910288e-01 -6.98996484e-01
1.24444938e+00 1.67088225e-01 7.93306172e-01 1.64575350e+00
1.03360303e-01 8.92686844e-01 1.18252540e+00 5.80854937e-02
6.08780310e-02 -9.98082384e-02 1.52219072e-01 8.87398005e-01
2.78052807e-01 4.96460676e-01 -9.17005360e-01 -2.27052346e-01
1.29719722e+00 2.21783653e-01 -2.45860934e-01 -7.63320446e-01
-9.25462425e-01 9.65610504e-01 6.66694939e-01 7.16442466e-02
-3.41762394e-01 6.43367112e-01 1.66803181e-01 2.32382074e-01
8.13075483e-01 4.73712564e-01 -2.10203663e-01 -1.34276569e-01
-9.35663283e-01 3.20673436e-01 5.11778951e-01 1.22484553e+00
6.94455206e-01 4.65808183e-01 -1.27433762e-01 7.28701949e-01
2.48165771e-01 4.08554703e-01 2.22464919e-01 -1.35619378e+00
1.63441136e-01 2.26931751e-01 -1.87606409e-01 -7.97021449e-01
-6.10261038e-02 -3.15688312e-01 -4.75007534e-01 2.10031494e-01
1.10700727e-01 3.08144927e-01 -1.15710151e+00 2.07878113e+00
-3.33224237e-02 1.25973865e-01 -2.46992737e-01 8.00005019e-01
9.59449172e-01 6.28395736e-01 3.74802887e-01 3.02593648e-01
9.85733807e-01 -6.16974235e-01 -1.22406855e-01 -3.26074332e-01
2.98506081e-01 -6.14856184e-01 1.22541118e+00 -7.46279536e-03
-1.20557678e+00 -7.81477630e-01 -8.73658061e-01 -7.04110086e-01
-3.45148891e-01 -1.62530750e-01 9.95083988e-01 4.89168972e-01
-1.56239271e+00 5.05798638e-01 -8.12795401e-01 -5.26611447e-01
8.63899589e-01 -1.04899496e-01 -5.60041487e-01 -2.98385441e-01
-5.81921577e-01 8.83066893e-01 9.36187208e-02 -4.33071256e-01
-1.80640221e+00 -7.98258543e-01 -8.88347387e-01 -1.57948613e-01
-9.42476615e-02 -1.16110969e+00 9.10322309e-01 -6.59115613e-01
-1.05261683e+00 1.34985733e+00 -1.60231456e-01 -1.38566360e-01
2.80044943e-01 -5.71343303e-01 1.42744213e-01 2.69320697e-01
1.61293283e-01 9.05593395e-01 9.66108799e-01 -1.49600601e+00
2.95709707e-02 -5.65869749e-01 4.81082797e-01 4.04592335e-01
1.33947194e-01 -2.71455646e-01 -1.88606650e-01 -6.02772295e-01
2.69307047e-01 -7.88071394e-01 -3.86390872e-02 2.26992235e-01
-1.85855195e-01 -1.95535481e-01 5.76161683e-01 -9.12819266e-01
3.36091518e-01 -2.26180482e+00 3.32493454e-01 -2.32243776e-01
6.35168970e-01 2.55357195e-02 -5.29274881e-01 4.71732378e-01
-2.37829596e-01 3.00801307e-01 1.72074050e-01 -4.63565469e-01
2.44972274e-01 3.21363091e-01 -7.27638543e-01 5.17836928e-01
1.24425262e-01 1.24417686e+00 -8.59341264e-01 -2.02381343e-01
5.41349828e-01 8.47947001e-01 -6.90090418e-01 4.19004679e-01
-3.17291826e-01 4.99401867e-01 -3.41309100e-01 5.45603991e-01
5.50726354e-01 -4.36811656e-01 -8.17091689e-02 -2.64085531e-01
2.18810186e-01 4.98055875e-01 -4.98830616e-01 2.16962266e+00
-4.65881258e-01 7.99491525e-01 -2.86708772e-01 -6.48895264e-01
6.78998172e-01 1.15124956e-01 2.31488347e-01 -8.21291924e-01
3.45193110e-02 -1.17594063e-01 -3.29452097e-01 -1.69698209e-01
5.02904654e-01 -2.37872109e-01 9.74786505e-02 4.76573110e-01
6.98952556e-01 -4.53697622e-01 -2.66504467e-01 6.81135952e-01
9.95474517e-01 7.04009175e-01 6.99392557e-02 -3.05811286e-01
-3.90407473e-01 -3.45446736e-01 1.60812020e-01 7.63707221e-01
-4.51383926e-02 8.11757743e-01 2.77498901e-01 -4.65627581e-01
-1.28922248e+00 -1.81253207e+00 -7.97321275e-02 1.26625478e+00
2.19462235e-02 -4.79830831e-01 -2.33122453e-01 -5.48867106e-01
1.82452518e-02 1.13678741e+00 -6.83499873e-01 -3.77980232e-01
-1.29237592e-01 -6.16711192e-02 5.09767413e-01 6.74907923e-01
2.48967141e-01 -6.78954899e-01 -5.41716754e-01 -2.00322390e-01
-9.62255076e-02 -9.16608632e-01 -1.79801583e-01 2.04369307e-01
-9.74406362e-01 -8.25140357e-01 -7.76310921e-01 -3.46264482e-01
5.34210265e-01 6.19292557e-01 1.63860369e+00 8.37002546e-02
-4.85843062e-01 9.95543778e-01 -1.31274894e-01 -1.87311560e-01
-1.48827150e-01 -5.15490659e-02 -2.86796857e-02 -5.47113776e-01
2.86095619e-01 -1.14942217e+00 -8.04287553e-01 4.55090404e-02
-7.59713054e-01 2.28191465e-01 4.04032558e-01 4.94259864e-01
6.14000499e-01 -7.97961712e-01 8.28732997e-02 -6.91521704e-01
4.09647822e-01 -6.37635708e-01 -4.45040017e-01 3.93479876e-02
-1.74697757e-01 1.15234315e-01 1.67622238e-01 -2.45895222e-01
-1.08886242e+00 -2.34138116e-01 -1.23922244e-01 -9.47381616e-01
-5.20618618e-01 -1.80344239e-01 -1.86371967e-01 -1.35749057e-02
7.04129636e-01 3.07654619e-01 -2.27187514e-01 -6.36803031e-01
9.82445776e-01 -1.05408812e-02 2.81169027e-01 -8.60459864e-01
9.43624437e-01 6.81347728e-01 -7.17593059e-02 -9.35183227e-01
-9.44075108e-01 -5.27787983e-01 -7.08019376e-01 -1.06536299e-01
1.11522448e+00 -1.22195423e+00 -4.10625726e-01 6.66315481e-02
-1.22621405e+00 -3.01890969e-01 -6.33908272e-01 3.21945161e-01
-1.09133291e+00 1.54984057e-01 -6.08016372e-01 -6.11169934e-01
3.31624411e-02 -9.44220483e-01 1.30423272e+00 6.50651976e-02
-4.61656928e-01 -1.01521027e+00 2.45345622e-01 4.34695214e-01
4.25928801e-01 2.83600479e-01 1.20624173e+00 -5.54599047e-01
-1.07696819e+00 -4.13759537e-02 -5.10235608e-01 3.45359623e-01
-8.16787705e-02 -2.47543767e-01 -1.49938750e+00 -4.11929399e-01
3.34297568e-02 -6.74787283e-01 1.17524886e+00 3.91493350e-01
1.22225058e+00 -2.32209936e-01 -2.09128894e-02 1.23138654e+00
1.57577801e+00 -2.72285827e-02 7.78813481e-01 -1.51416898e-01
8.61987770e-01 3.93478125e-01 -1.57306358e-01 3.67582470e-01
4.14672852e-01 3.79281372e-01 4.97975379e-01 -2.87625361e-02
-5.49799562e-01 -6.99805021e-01 2.61109501e-01 7.27032661e-01
-5.10451615e-01 -3.50434721e-01 -6.22232616e-01 5.17903984e-01
-1.48109043e+00 -1.17589855e+00 4.23244745e-01 2.01582146e+00
3.81976038e-01 -1.73742265e-01 -1.14173748e-01 -7.33309925e-01
2.40374789e-01 8.90524745e-01 -6.94143534e-01 -3.44337225e-01
-2.86410034e-01 5.55326462e-01 3.15284610e-01 3.74341726e-01
-8.97035599e-01 1.09846413e+00 7.19224262e+00 6.94783628e-01
-7.11273670e-01 3.99800181e-01 5.97873688e-01 -5.33502162e-01
-8.39619696e-01 2.15271950e-01 -5.99686146e-01 -1.29033970e-02
7.94046581e-01 -1.05901860e-01 4.74494636e-01 9.49743807e-01
-2.17343345e-01 1.92767531e-01 -1.41348946e+00 9.99245763e-01
2.44631112e-01 -1.40580022e+00 6.46070659e-01 4.64940131e-01
7.36645997e-01 3.52836549e-01 6.76751018e-01 2.47108549e-01
7.45182276e-01 -1.25129712e+00 7.21990526e-01 8.55316043e-01
8.77832055e-01 -2.74781495e-01 -1.69855449e-02 1.74656644e-01
-8.91599000e-01 2.86176115e-01 -6.89407229e-01 2.70518940e-02
9.02330354e-02 1.68733940e-01 -5.32907426e-01 3.27315062e-01
5.92168450e-01 7.97680378e-01 -7.29916871e-01 7.04270244e-01
-2.69347459e-01 4.57167447e-01 -1.87464207e-01 3.60542029e-01
3.10533047e-01 -7.04904646e-02 8.35621715e-01 8.67431641e-01
3.42472762e-01 1.05323590e-01 -3.06406114e-02 1.38190138e+00
-6.41961768e-02 -1.63374454e-01 -1.24436092e+00 -2.47190252e-01
2.68631220e-01 8.51322532e-01 -4.05899614e-01 -2.69069582e-01
-2.86837697e-01 1.12951028e+00 6.40662253e-01 6.80559754e-01
-7.49711037e-01 2.80074060e-01 1.07297325e+00 2.22222328e-01
2.81925887e-01 -7.76632726e-01 -2.76194569e-02 -1.35961103e+00
-3.00326526e-01 -5.73448718e-01 1.56368002e-01 -1.31536198e+00
-1.52577007e+00 4.93299961e-01 3.01971436e-01 -8.28844547e-01
-3.94592434e-01 -7.95869052e-01 -4.11828637e-01 9.20522153e-01
-1.15559125e+00 -1.49154663e+00 -3.85655165e-01 7.10343659e-01
6.59615695e-01 -2.35202104e-01 1.05226314e+00 -8.70526880e-02
-4.31018211e-02 4.26630616e-01 -6.57379553e-02 -1.28970131e-01
4.28903043e-01 -1.22553408e+00 8.24783206e-01 8.12779009e-01
9.01384175e-01 9.22338188e-01 4.93226200e-01 -5.19555390e-01
-1.36144674e+00 -6.64736629e-01 3.25672507e-01 -1.00182152e+00
3.31000775e-01 -5.35439610e-01 -7.38567352e-01 1.06781387e+00
3.02747905e-01 -2.57120460e-01 9.50595140e-01 2.53097296e-01
-1.11147177e+00 3.01582634e-01 -8.30700040e-01 6.10064864e-01
1.76460505e+00 -1.16359341e+00 -7.01061666e-01 4.87300605e-01
8.40997517e-01 -9.84807909e-02 -7.93861270e-01 1.04891971e-01
5.98552942e-01 -1.30842829e+00 1.45524025e+00 -8.77948642e-01
5.16310811e-01 1.20139934e-01 -8.11249077e-01 -1.17598116e+00
-7.61359870e-01 -3.87493253e-01 -5.71320467e-02 8.63874376e-01
4.64169271e-02 -5.97399473e-01 5.86858213e-01 5.71365535e-01
-1.73316523e-01 -4.28924561e-01 -9.39867377e-01 -7.47175813e-01
2.53638327e-01 -4.90960300e-01 4.40488160e-01 7.76537359e-01
-8.23825061e-01 3.65202844e-01 -2.31331557e-01 3.30759250e-02
1.06492531e+00 1.41167581e-01 7.53385961e-01 -1.21638381e+00
-4.48484063e-01 -4.61755991e-01 -6.73322380e-01 -1.27670038e+00
4.21649754e-01 -1.28063297e+00 -4.13894564e-01 -1.64350355e+00
2.79083759e-01 -2.56760061e-01 -2.98384428e-01 3.50116104e-01
2.80288368e-01 3.41883063e-01 4.59178150e-01 4.17223394e-01
-5.66955924e-01 7.65958786e-01 1.40496492e+00 -1.42510710e-02
4.12115067e-01 -3.02348971e-01 -8.14294934e-01 8.80989313e-01
6.42147481e-01 -1.39259741e-01 -7.90724754e-01 -8.19881260e-01
1.41502708e-01 -1.63864136e-01 9.86797333e-01 -7.34625757e-01
-1.33965135e-01 -1.54122457e-01 8.37587059e-01 -5.23672163e-01
9.46078122e-01 -4.44307119e-01 3.37273747e-01 2.53680684e-02
-3.90035272e-01 1.52172716e-02 2.50680715e-01 6.42359197e-01
1.18718475e-01 1.54783905e-01 6.76138997e-01 -6.50712013e-01
-1.10336065e+00 7.16787755e-01 -4.04266082e-02 3.13322842e-01
5.77671409e-01 -3.83298934e-01 -7.88601637e-01 -6.80226564e-01
-7.86951005e-01 -3.82164359e-01 8.40752721e-01 5.42481065e-01
9.28809762e-01 -1.40739644e+00 -5.96025050e-01 3.52153152e-01
2.84692496e-01 -3.29738468e-01 3.79157484e-01 2.22392663e-01
-4.40957963e-01 2.96606034e-01 -6.27399445e-01 -5.87814510e-01
-8.67093563e-01 4.29776311e-01 3.15942436e-01 2.57245153e-01
-6.90627217e-01 1.38204479e+00 9.94107664e-01 -2.63822645e-01
-1.56874976e-05 -3.22510488e-02 3.25848252e-01 -7.53029361e-02
2.31703758e-01 2.65143365e-01 -3.70670378e-01 -8.01555395e-01
-2.49910370e-01 4.49825913e-01 -3.73198055e-02 -2.73624241e-01
1.46895647e+00 -4.39722091e-02 3.75774167e-02 6.78786278e-01
1.46948409e+00 -3.20983261e-01 -1.46371174e+00 -1.63199063e-02
-5.72428942e-01 -8.72871339e-01 6.33277074e-02 -5.66643298e-01
-9.73570824e-01 1.30904567e+00 5.80316365e-01 -1.20678395e-01
7.65536129e-01 5.26552498e-01 3.75573963e-01 2.30381846e-01
6.02022171e-01 -4.23634976e-01 4.10947233e-01 4.31575418e-01
1.32851732e+00 -8.65923941e-01 6.87102675e-02 -1.15501240e-01
-5.49641848e-01 6.56977415e-01 5.93010068e-01 -5.93906164e-01
6.87746108e-01 -9.56705958e-02 -2.11235568e-01 -5.65501451e-01
-9.22547221e-01 -4.46101874e-01 5.88665128e-01 9.55560029e-01
4.87960815e-01 2.08212331e-01 3.62103671e-01 1.83518767e-01
-5.75916767e-01 -4.61827248e-01 5.77368475e-02 5.58132291e-01
-3.28950167e-01 -8.13033402e-01 3.03454876e-01 4.98165727e-01
-3.81037183e-02 -3.18272710e-01 -4.67271894e-01 8.48014116e-01
-2.05486361e-02 3.23754907e-01 2.31808901e-01 -4.58127707e-01
2.01067865e-01 2.98819214e-01 1.07279944e+00 -8.56294036e-01
-4.60565388e-02 -5.13457367e-03 6.63174391e-02 -1.03191137e+00
-3.17468554e-01 -4.81860489e-01 -5.73374569e-01 -3.52727562e-01
1.75105408e-02 -4.29226428e-01 4.26647633e-01 5.05450904e-01
4.13663357e-01 2.87032127e-01 3.42459947e-01 -1.19976258e+00
-4.55068827e-01 -7.07591832e-01 -6.25641704e-01 6.29247129e-01
2.46164337e-01 -9.99140322e-01 -4.98806208e-01 5.37804961e-02] | [8.675751686096191, -3.0969018936157227] |
c467205f-f9ef-49c5-9648-246dfc307d4a | topic-guided-sampling-for-data-efficient | 2306.00765 | null | https://arxiv.org/abs/2306.00765v1 | https://arxiv.org/pdf/2306.00765v1.pdf | Topic-Guided Sampling For Data-Efficient Multi-Domain Stance Detection | Stance Detection is concerned with identifying the attitudes expressed by an author towards a target of interest. This task spans a variety of domains ranging from social media opinion identification to detecting the stance for a legal claim. However, the framing of the task varies within these domains, in terms of the data collection protocol, the label dictionary and the number of available annotations. Furthermore, these stance annotations are significantly imbalanced on a per-topic and inter-topic basis. These make multi-domain stance detection a challenging task, requiring standardization and domain adaptation. To overcome this challenge, we propose $\textbf{T}$opic $\textbf{E}$fficient $\textbf{St}$anc$\textbf{E}$ $\textbf{D}$etection (TESTED), consisting of a topic-guided diversity sampling technique and a contrastive objective that is used for fine-tuning a stance classifier. We evaluate the method on an existing benchmark of $16$ datasets with in-domain, i.e. all topics seen and out-of-domain, i.e. unseen topics, experiments. The results show that our method outperforms the state-of-the-art with an average of $3.5$ F1 points increase in-domain, and is more generalizable with an averaged increase of $10.2$ F1 on out-of-domain evaluation while using $\leq10\%$ of the training data. We show that our sampling technique mitigates both inter- and per-topic class imbalances. Finally, our analysis demonstrates that the contrastive learning objective allows the model a more pronounced segmentation of samples with varying labels. | ['Isabelle Augenstein', 'Arnav Arora', 'Erik Arakelyan'] | 2023-06-01 | null | null | null | null | ['stance-detection'] | ['natural-language-processing'] | [ 1.76969096e-01 1.08058169e-01 -4.12126750e-01 -6.50762260e-01
-1.47037745e+00 -8.91245008e-01 6.26775980e-01 3.34475696e-01
-4.61885035e-01 9.25338745e-01 6.33319467e-02 -2.13010699e-01
9.16889030e-03 -6.78573251e-01 -6.64113283e-01 -7.66163766e-01
2.39291623e-01 7.79707789e-01 3.60947311e-01 -3.67934257e-01
3.48790258e-01 -2.32925594e-01 -1.50788772e+00 5.13504505e-01
1.02911270e+00 1.18395913e+00 -2.85875142e-01 1.26494065e-01
-1.64882451e-01 4.09787059e-01 -1.18449211e+00 -8.05113494e-01
1.50433734e-01 -2.95612633e-01 -7.28697240e-01 2.03079835e-01
6.93612158e-01 1.00217722e-01 2.48990461e-01 1.09204483e+00
7.03831136e-01 -1.04340762e-01 8.78278553e-01 -1.09546208e+00
-2.51213074e-01 5.02495408e-01 -7.83190131e-01 3.34867805e-01
2.02852055e-01 1.59315264e-03 1.31219471e+00 -8.66127014e-01
9.00285304e-01 1.20169640e+00 5.76355636e-01 2.54917979e-01
-1.28836143e+00 -1.00320315e+00 4.67221230e-01 -6.52257055e-02
-1.01290274e+00 -2.64929444e-01 1.08855689e+00 -8.38074625e-01
6.02736652e-01 1.73928738e-01 3.35157692e-01 1.27103341e+00
3.72672975e-02 7.22617149e-01 1.63379288e+00 -3.45594466e-01
3.67478848e-01 5.43490708e-01 6.02742493e-01 3.03629607e-01
2.60925114e-01 -2.82114178e-01 -6.57459855e-01 -5.66600740e-01
4.30361889e-02 -4.24439013e-01 -2.27927212e-02 -2.43772402e-01
-7.73523986e-01 1.23527336e+00 1.01681647e-03 3.00971687e-01
-1.66287154e-01 -3.63407761e-01 6.76313162e-01 3.63534480e-01
1.03925741e+00 5.89693248e-01 -7.04016030e-01 -2.09138170e-01
-1.13705826e+00 6.33362174e-01 8.78007531e-01 7.02447474e-01
7.11942673e-01 -1.26259446e-01 -2.25103930e-01 1.34589922e+00
4.05255798e-03 7.77534902e-01 3.88514280e-01 -9.04254019e-01
7.66053319e-01 8.59132946e-01 6.52081966e-02 -8.73086452e-01
-2.08985761e-01 -5.68131506e-01 -2.99466103e-01 2.88196951e-01
7.43938208e-01 -3.34674299e-01 -8.46841753e-01 1.92419517e+00
4.68170524e-01 -6.57920241e-01 -2.44451717e-01 5.35632670e-01
5.67493320e-01 3.50209415e-01 2.38341972e-01 -3.71181786e-01
1.78266537e+00 -5.37425637e-01 -5.51047206e-01 -5.06061912e-01
5.57370484e-01 -1.02573144e+00 1.20657420e+00 5.50338507e-01
-1.04135740e+00 -3.13083231e-01 -9.73603725e-01 1.48292378e-01
-4.53115344e-01 -4.49170172e-02 2.16265008e-01 1.02483416e+00
-6.17416382e-01 2.01847732e-01 -2.14723736e-01 -1.09791502e-01
7.00345993e-01 3.52850378e-01 7.02418387e-02 -4.26463261e-02
-1.40297914e+00 5.48355341e-01 1.27827868e-01 -5.27096391e-01
-6.83122218e-01 -9.03250694e-01 -5.71748853e-01 -4.16458160e-01
4.95321691e-01 -2.33296931e-01 1.16131580e+00 -9.72032487e-01
-1.13357246e+00 1.44074011e+00 -3.02218080e-01 -3.77810806e-01
6.65474176e-01 -1.13547072e-01 -5.65427065e-01 6.69792295e-02
6.21406257e-01 4.60211307e-01 9.93545532e-01 -1.11812806e+00
-7.22686827e-01 -4.87425089e-01 1.89670339e-01 9.04125869e-02
-2.89829880e-01 2.61336714e-01 1.51742192e-03 -8.07802558e-01
2.78743468e-02 -1.05680192e+00 1.97749257e-01 -4.09545690e-01
-3.20921153e-01 -4.18749571e-01 1.10512543e+00 -5.19465625e-01
1.33629513e+00 -2.00306034e+00 -1.66089416e-01 2.87823915e-01
4.31602210e-01 1.21540435e-01 3.71630967e-01 1.15924492e-01
-1.45775810e-01 2.23292604e-01 -2.36405686e-01 -2.90305823e-01
1.51517302e-01 -1.82250321e-01 -3.45800549e-01 4.82940942e-01
-1.65771134e-03 3.89574200e-01 -5.28734028e-01 -5.58144689e-01
-3.16508174e-01 4.72978875e-02 -6.14164948e-01 -7.79145733e-02
-6.11839712e-01 3.56911570e-01 -5.46550095e-01 7.20289469e-01
6.01954401e-01 -3.43251884e-01 3.66150469e-01 -2.79931482e-02
-9.40661132e-02 6.12834096e-01 -1.15248561e+00 1.42321193e+00
-1.72014907e-01 5.27363598e-01 2.37798125e-01 -1.18701637e+00
1.16225851e+00 3.21151197e-01 6.63053989e-01 -8.71934712e-01
3.13001901e-01 6.91338718e-01 1.08155176e-01 -1.03874296e-01
5.21265388e-01 -3.95831227e-01 -6.06051803e-01 7.41297305e-01
-9.24276561e-02 -8.33599120e-02 5.11706352e-01 1.00045629e-01
9.58387375e-01 -2.15904653e-01 1.41987041e-01 -8.23761582e-01
4.33672160e-01 1.44031465e-01 7.22995341e-01 5.41046977e-01
-4.36197221e-01 3.44874054e-01 8.90554249e-01 -1.66721106e-01
-9.81386185e-01 -7.96721995e-01 -4.70565826e-01 1.53839362e+00
-1.24529414e-01 -1.67976335e-01 -8.84411156e-01 -1.06369007e+00
2.89283879e-03 5.93866348e-01 -7.04498529e-01 2.17179239e-01
-6.98527813e-01 -1.11722183e+00 4.23444629e-01 1.31123468e-01
5.87637305e-01 -9.66466606e-01 -6.23533249e-01 1.09953955e-01
-4.60156232e-01 -1.09556055e+00 -4.02276844e-01 2.69071132e-01
-5.67349613e-01 -1.12108827e+00 -7.36645222e-01 -6.49578571e-01
3.28894198e-01 -1.88444048e-01 1.46855390e+00 -4.44498271e-01
3.79871167e-02 2.01880652e-02 -3.61819923e-01 -6.14977896e-01
-3.71721715e-01 3.92654151e-01 -2.10875452e-01 -7.37978518e-02
7.56762803e-01 -4.61588204e-01 -7.15391517e-01 5.29525995e-01
-7.51387656e-01 -4.62277949e-01 3.32907647e-01 8.91074777e-01
5.88083684e-01 -4.10939865e-02 7.82930195e-01 -1.53774774e+00
8.61165702e-01 -4.47608829e-01 -4.01142478e-01 4.46986258e-02
-7.92460740e-01 -1.78686827e-01 3.99236053e-01 -3.89361501e-01
-9.38965380e-01 -5.65792263e-01 -1.23433016e-01 3.25045250e-02
-1.71527907e-01 3.43998015e-01 -2.58400351e-01 4.52911496e-01
1.14948857e+00 -8.79525393e-02 -1.54784486e-01 -2.82022446e-01
2.25954596e-02 9.01036084e-01 -9.60085839e-02 -9.54473913e-01
4.31513071e-01 4.92969304e-01 -4.82176781e-01 -6.58007801e-01
-1.19118714e+00 -5.39830506e-01 -3.51879418e-01 -1.05647221e-01
6.99403524e-01 -1.03415394e+00 -4.53993082e-01 4.72106904e-01
-8.34936798e-01 -3.76610577e-01 -1.94442049e-01 7.37485364e-02
-4.14311111e-01 2.34767601e-01 -4.20058995e-01 -6.57748103e-01
-5.18085241e-01 -1.33560288e+00 1.15313578e+00 -2.32875913e-01
-7.73057878e-01 -9.62907791e-01 8.04255232e-02 9.04782712e-01
9.52849314e-02 4.54996318e-01 1.12939823e+00 -9.68792975e-01
-8.47315937e-02 -1.66960180e-01 -7.19480813e-02 5.04436016e-01
-1.44272121e-02 -4.15883303e-01 -1.17130733e+00 -4.56828892e-01
2.00140059e-01 -5.60159504e-01 7.25792825e-01 4.86020923e-01
8.33779097e-01 -3.26747596e-01 -2.92933464e-01 -2.08093114e-02
1.13662302e+00 2.52092928e-01 4.51003492e-01 4.95240301e-01
1.78390637e-01 7.17628837e-01 9.14289832e-01 4.77105319e-01
2.69313455e-01 7.60551035e-01 4.52291518e-02 9.22913942e-03
-5.47540598e-02 2.44982690e-01 3.91527057e-01 3.23686838e-01
1.73805967e-01 -2.07213059e-01 -1.12125742e+00 6.60931289e-01
-1.57115555e+00 -9.39569235e-01 4.34572473e-02 2.09953141e+00
1.29883754e+00 7.03150809e-01 5.48509955e-01 2.82702804e-01
7.35136867e-01 4.09584135e-01 -5.96911609e-01 -4.91513491e-01
-2.53764391e-01 5.73704243e-01 2.43440613e-01 5.53943694e-01
-1.27384603e+00 8.06191921e-01 5.30822182e+00 1.13805473e+00
-1.31077147e+00 3.16644818e-01 9.43489850e-01 -2.98978508e-01
-3.44084799e-01 -2.41138935e-01 -1.11893070e+00 8.38985741e-01
9.27861869e-01 -5.07274605e-02 -2.31493428e-01 9.85095143e-01
8.63334313e-02 -2.24751577e-01 -9.48933065e-01 7.34980464e-01
2.76179522e-01 -1.10221744e+00 -2.50071704e-01 3.36461574e-01
7.99117982e-01 -2.52419058e-02 3.26678663e-01 5.45967460e-01
3.65047574e-01 -6.69441640e-01 1.02347767e+00 -2.35550806e-01
9.83438790e-01 -6.26521230e-01 6.36497319e-01 3.25854540e-01
-8.82883191e-01 -4.80473004e-02 1.67295143e-01 6.98835626e-02
1.23654027e-02 1.13652813e+00 -7.53750086e-01 3.30485761e-01
8.48277986e-01 4.70569223e-01 -3.37046534e-01 2.49865308e-01
-1.41628817e-01 7.27106750e-01 -2.64055133e-01 -1.29935771e-01
1.60240829e-01 9.40489173e-02 5.75692177e-01 1.27431130e+00
1.33825436e-01 1.62765142e-02 4.36744988e-01 6.44763827e-01
-2.38117725e-01 1.91193059e-01 -1.50779814e-01 1.17975168e-01
5.72271347e-01 9.58445251e-01 -6.64378583e-01 -4.12565559e-01
-8.17814395e-02 4.62046176e-01 2.10437313e-01 2.11675406e-01
-9.30269718e-01 -2.58397281e-01 6.93519056e-01 6.35314167e-01
4.32409078e-01 9.75256190e-02 -5.64739883e-01 -1.04021120e+00
2.13940129e-01 -1.38377130e+00 5.71143210e-01 -1.23691842e-01
-1.45647860e+00 5.60722768e-01 1.03496954e-01 -1.17690110e+00
-2.84353971e-01 -6.66233838e-01 -8.21932778e-02 9.35258806e-01
-1.48761439e+00 -1.05704522e+00 -3.62741575e-02 4.83154178e-01
5.94653785e-01 -2.07780197e-01 6.64839625e-01 5.43854117e-01
-4.11061168e-01 7.35001028e-01 -6.63593039e-02 2.33144075e-01
1.18443978e+00 -1.27060151e+00 -1.13166094e-01 5.84404171e-01
-3.14669341e-01 6.69330478e-01 9.03902054e-01 -6.19837821e-01
-5.38928866e-01 -9.92559612e-01 1.09657204e+00 -5.74833035e-01
7.06666350e-01 -6.11064851e-01 -7.33593106e-01 5.30239046e-01
1.44811779e-01 -3.95907611e-01 1.18763018e+00 5.97913027e-01
-7.64791548e-01 -1.49439827e-01 -1.34181249e+00 3.81216764e-01
9.01665330e-01 -5.07397950e-01 -5.81315994e-01 4.60876912e-01
3.70540321e-01 -5.38967371e-01 -1.04556012e+00 3.35507929e-01
4.99857724e-01 -1.10554421e+00 7.80726373e-01 -1.65797874e-01
6.31836355e-01 1.09573826e-01 -3.42773080e-01 -1.05019927e+00
-1.34934902e-01 -3.14131737e-01 2.03859940e-01 1.28858912e+00
8.87306392e-01 -6.46820426e-01 9.97149050e-01 6.07097387e-01
-1.45658344e-01 -9.02631044e-01 -9.83907759e-01 -5.31554461e-01
6.30945683e-01 -3.77495438e-01 2.94674367e-01 1.12574553e+00
8.44933838e-03 6.07199311e-01 -6.15893342e-02 -2.83921778e-01
6.02273166e-01 4.94417131e-01 6.64072156e-01 -1.51825726e+00
-3.69252533e-01 -5.10896266e-01 1.28200725e-01 -9.88545120e-01
1.36432633e-01 -6.04235530e-01 -1.33462206e-01 -9.57395613e-01
2.57973284e-01 -8.47850204e-01 -9.90225375e-02 2.56000578e-01
-2.04207644e-01 1.66365325e-01 3.10297720e-02 3.27072799e-01
-6.77969575e-01 1.59712955e-01 1.03333068e+00 -3.33907455e-01
-9.50316712e-02 3.36729810e-02 -1.05205441e+00 7.78601646e-01
7.48712301e-01 -7.03098595e-01 -3.65016609e-01 -2.97505438e-01
2.88133860e-01 -1.99576303e-01 2.11326271e-01 -7.89646029e-01
-9.47058797e-02 -6.23822249e-02 1.42469639e-02 -7.01795578e-01
5.25851786e-01 -4.63455468e-01 -3.91057640e-01 3.01671207e-01
-3.21032047e-01 -1.40361190e-01 1.86896205e-01 5.22208095e-01
-3.52116585e-01 -6.48223013e-02 9.60169852e-01 -2.57260054e-01
-3.56586158e-01 6.71405345e-02 -5.64132109e-02 8.13211560e-01
1.02543914e+00 -2.83162385e-01 -7.21077740e-01 -2.18041778e-01
-6.82157576e-01 6.92199031e-03 3.91163528e-01 1.75690427e-01
-3.38287205e-02 -8.62879813e-01 -7.78164029e-01 1.04862889e-02
2.57789791e-01 -8.36802274e-02 3.27719629e-01 8.20950449e-01
3.78690241e-03 4.23009247e-01 1.65829241e-01 -8.44377697e-01
-1.41930366e+00 4.73101102e-02 2.26168469e-01 -9.00495708e-01
8.38006586e-02 9.72638190e-01 1.90946296e-01 -4.44138974e-01
2.83330195e-02 -1.09324858e-01 -1.19893394e-01 6.96314871e-01
1.87060922e-01 2.54110873e-01 4.13710773e-01 -6.28288507e-01
-3.34727317e-01 4.52822894e-01 -4.72438604e-01 -2.51155019e-01
1.10997939e+00 9.24614444e-02 -1.37972713e-01 5.55909455e-01
1.19902301e+00 2.12547570e-01 -9.87190008e-01 -3.51925999e-01
6.02577906e-03 -2.88218915e-01 -3.15328449e-01 -1.08338058e+00
-7.18186915e-01 7.86773860e-01 4.39867228e-01 4.40837562e-01
8.47634792e-01 1.71845838e-01 6.96597040e-01 -2.13126782e-02
3.51967722e-01 -1.56909025e+00 4.30806994e-01 5.11624753e-01
6.99424505e-01 -1.39272106e+00 1.75500587e-01 -6.25796854e-01
-7.21114278e-01 5.21891892e-01 4.08292830e-01 -2.89016645e-02
8.83299291e-01 6.80385455e-02 2.94429451e-01 -5.55884063e-01
-5.24501860e-01 1.77507222e-01 1.41546637e-01 2.28480369e-01
7.31276512e-01 1.16452754e-01 -7.40361869e-01 5.93013585e-01
-5.24361312e-01 -2.68307716e-01 1.33837789e-01 9.99243021e-01
-6.03577018e-01 -1.44198763e+00 -3.94614130e-01 5.93098879e-01
-8.63018513e-01 -6.07578270e-02 -4.79896009e-01 8.21782589e-01
5.25471389e-01 1.12849462e+00 -1.12783402e-01 -4.31996062e-02
4.92814332e-01 3.81364942e-01 9.94000509e-02 -8.25860560e-01
-9.44606900e-01 3.75427246e-01 4.54085827e-01 -2.04264894e-01
-5.76106131e-01 -1.15241849e+00 -8.75872731e-01 -3.40301484e-01
-1.65884346e-01 3.87949020e-01 3.37871999e-01 9.33949292e-01
4.06342268e-01 4.72857505e-01 5.70602775e-01 -2.55286455e-01
-5.88310778e-01 -1.08505976e+00 -6.49862111e-01 8.12680721e-01
7.62559474e-02 -8.90604377e-01 -3.83860081e-01 1.14111580e-01] | [8.926483154296875, 9.935812950134277] |
8d7252f0-a763-4f23-a743-fa416442b118 | sheffield-submissions-for-wmt18-multimodal | null | null | https://aclanthology.org/W18-6442 | https://aclanthology.org/W18-6442.pdf | Sheffield Submissions for WMT18 Multimodal Translation Shared Task | This paper describes the University of Sheffield{'}s submissions to the WMT18 Multimodal Machine Translation shared task. We participated in both tasks 1 and 1b. For task 1, we build on a standard sequence to sequence attention-based neural machine translation system (NMT) and investigate the utility of multimodal re-ranking approaches. More specifically, n-best translation candidates from this system are re-ranked using novel multimodal cross-lingual word sense disambiguation models. For task 1b, we explore three approaches: (i) re-ranking based on cross-lingual word sense disambiguation (as for task 1), (ii) re-ranking based on consensus of NMT n-best lists from German-Czech, French-Czech and English-Czech systems, and (iii) data augmentation by generating English source data through machine translation from French to English and from German to English followed by hypothesis selection using a multimodal-reranker. | ['Pranava Swaroop Madhyastha', 'Lucia Specia', 'Chiraag Lala', 'Carolina Scarton'] | 2018-10-01 | null | null | null | ws-2018-10 | ['multimodal-machine-translation'] | ['natural-language-processing'] | [ 8.18187654e-01 1.02090776e-01 -2.16250300e-01 -3.35003257e-01
-1.74317551e+00 -9.93333280e-01 9.48093235e-01 3.58707011e-01
-9.28976476e-01 1.13885224e+00 6.39653087e-01 -7.23404109e-01
-3.08936127e-02 -5.30482642e-02 -6.03545666e-01 -2.62953937e-01
5.13696313e-01 1.25054145e+00 -2.56445467e-01 -6.51769638e-01
2.16333374e-01 -8.49545375e-03 -1.04514170e+00 9.95958805e-01
1.08704710e+00 2.99188793e-01 3.94341350e-01 7.57715881e-01
-2.08113462e-01 -8.62934738e-02 -3.56974274e-01 -7.44570255e-01
3.29089053e-02 -8.48078072e-01 -1.36936474e+00 -4.49271083e-01
5.95070899e-01 3.09270501e-01 2.99974412e-01 1.09311068e+00
8.49717855e-01 2.98643172e-01 6.80963993e-01 -8.26876402e-01
-7.79173672e-01 9.62289035e-01 -3.39382738e-01 6.23964518e-03
7.56207168e-01 -1.66022275e-02 1.31245267e+00 -1.48008168e+00
1.10187721e+00 1.68700588e+00 2.80889750e-01 8.25307429e-01
-1.55604601e+00 -3.41092885e-01 -1.99147388e-01 1.61649823e-01
-1.13754880e+00 -6.18054569e-01 1.63236558e-01 -1.89127207e-01
1.55525398e+00 4.16425914e-01 8.89523402e-02 1.42985642e+00
3.72337773e-02 8.31007659e-01 1.24499238e+00 -9.44901526e-01
-9.62608233e-02 2.07804114e-01 -1.73787430e-01 3.28023583e-01
-1.54838651e-01 -1.72623202e-01 -7.53833175e-01 -4.67637509e-01
1.86447024e-01 -8.40568841e-01 -1.25723928e-01 4.08001281e-02
-2.13244438e+00 8.53432238e-01 -3.60842310e-02 5.27643323e-01
-3.36207300e-01 -1.69545382e-01 5.53197801e-01 6.33812308e-01
5.61380804e-01 9.06493664e-01 -9.24439013e-01 5.51000461e-02
-7.89439797e-01 1.52065223e-02 6.26730382e-01 7.70786703e-01
8.16212177e-01 -3.96947920e-01 -5.05415797e-01 1.45936835e+00
5.07132530e-01 8.79038632e-01 9.53823686e-01 -5.08156419e-01
1.02261186e+00 2.88938493e-01 2.33727694e-01 -4.12161559e-01
-3.97540629e-01 7.22872391e-02 -4.59661394e-01 -2.96130061e-01
1.04703896e-01 -4.94080812e-01 -1.03105903e+00 1.90005612e+00
-9.66957584e-02 -4.72937852e-01 6.66966915e-01 1.04452026e+00
8.65020752e-01 7.39731908e-01 3.34078461e-01 -4.26768601e-01
1.45848930e+00 -8.85607302e-01 -6.27891600e-01 -3.54456693e-01
8.50229740e-01 -1.35579360e+00 1.05309355e+00 -3.67451906e-02
-1.15735531e+00 -6.43484414e-01 -7.31525004e-01 -7.41046220e-02
-6.69967890e-01 2.97008276e-01 8.53349864e-02 3.48462224e-01
-1.45104361e+00 3.47884089e-01 -3.47262114e-01 -1.02159810e+00
-3.76494974e-01 6.27455056e-01 -6.80378139e-01 -1.10324070e-01
-1.76949430e+00 1.44240797e+00 6.62773371e-01 -4.97563416e-03
-4.75129336e-01 -1.51289567e-01 -1.02331269e+00 -4.06094015e-01
-1.66601464e-01 -1.00385046e+00 1.22582173e+00 -1.53871727e+00
-1.56446612e+00 1.32387948e+00 -6.30772471e-01 -2.16652691e-01
2.01921791e-01 -2.23268881e-01 -6.33373737e-01 -3.92444730e-02
5.67007840e-01 1.08876801e+00 4.25160408e-01 -1.32291400e+00
-7.66160309e-01 -2.40262598e-01 -4.47764516e-01 7.16591418e-01
2.03915492e-01 6.53621614e-01 -1.95921063e-01 -4.73981559e-01
-1.13251887e-01 -1.16033494e+00 -3.79210025e-01 -1.08025897e+00
-6.28829777e-01 -5.03609419e-01 1.95936158e-01 -8.31543744e-01
9.50305462e-01 -1.65679634e+00 5.94340146e-01 2.13139102e-01
-3.15479130e-01 2.24815324e-01 -1.01951492e+00 7.22158134e-01
-3.37855101e-01 4.32511657e-01 -2.31882378e-01 -5.15941560e-01
2.24284381e-01 1.26114696e-01 -1.96182206e-01 -5.54461330e-02
6.05357051e-01 1.17580426e+00 -1.22079849e+00 -4.61314887e-01
-3.85830499e-04 2.59069592e-01 8.26230869e-02 -7.32184350e-02
-2.25373775e-01 6.03473008e-01 -1.36576772e-01 4.67682987e-01
2.54095644e-01 9.60285515e-02 5.20847499e-01 -1.85391828e-01
-2.11941034e-01 7.51218617e-01 -5.45647919e-01 1.82659292e+00
-6.01596355e-01 5.68125546e-01 -8.52831230e-02 -5.48564255e-01
7.66586483e-01 6.52854562e-01 1.65706351e-01 -8.50469887e-01
-4.15558927e-02 7.79414535e-01 8.80628675e-02 -4.61145341e-01
7.97538698e-01 -3.45265180e-01 -3.52332592e-01 4.67120171e-01
3.93799335e-01 -2.66541336e-02 3.63788843e-01 8.19722489e-02
5.70840061e-01 5.15068948e-01 2.84528971e-01 -3.48378152e-01
5.00259221e-01 3.33926082e-01 2.28990510e-01 7.01673567e-01
1.24813482e-01 7.23318338e-01 8.64802822e-02 -1.91953793e-01
-1.04581261e+00 -8.94495964e-01 1.58561364e-01 1.57659078e+00
-1.49383262e-01 -3.54537010e-01 -6.13190711e-01 -6.72907352e-01
-3.22639704e-01 9.37545478e-01 -4.92301196e-01 6.63404018e-02
-6.84486330e-01 -8.45398545e-01 8.59156728e-01 8.97882804e-02
-1.76961608e-02 -1.20957279e+00 -1.05445608e-01 2.84995139e-01
-9.41360533e-01 -1.01867032e+00 -7.37581491e-01 3.37897509e-01
-6.84549868e-01 -4.79277790e-01 -1.15453148e+00 -1.09659505e+00
2.20606312e-01 -1.60667703e-01 1.39214969e+00 -2.67322421e-01
2.61129081e-01 5.02674341e-01 -5.04204452e-01 -4.22589242e-01
-8.17749321e-01 5.08039474e-01 3.52853835e-01 -5.36970869e-02
5.52932441e-01 1.03259444e-01 -1.18040644e-01 2.32046440e-01
-6.76137447e-01 9.27570239e-02 9.88657653e-01 1.02655673e+00
6.41130149e-01 -9.54505742e-01 7.31046796e-01 -6.21932149e-01
1.08937371e+00 -3.09806585e-01 -2.37123743e-01 6.47097170e-01
-5.04044354e-01 1.46479294e-01 9.29770693e-02 -6.04832828e-01
-9.59267139e-01 3.03499192e-01 9.18241404e-03 -8.58893842e-02
-2.38262981e-01 8.22281659e-01 -2.04643250e-01 1.80532038e-01
7.01357424e-01 2.66484916e-01 -2.31874183e-01 -3.89021516e-01
8.00566852e-01 7.18415439e-01 4.05870914e-01 -6.55557513e-01
5.88729501e-01 -3.85143936e-01 -2.95302600e-01 -6.55488193e-01
-3.46250176e-01 -4.85259444e-01 -1.08569753e+00 -2.04509776e-02
1.30890203e+00 -7.04601884e-01 -3.49555969e-01 8.67134407e-02
-1.71932352e+00 -2.45498464e-01 1.12304263e-01 8.34699869e-01
-4.74964172e-01 3.70130897e-01 -6.16168916e-01 -6.92886949e-01
-6.43291295e-01 -1.24183595e+00 1.55026925e+00 -1.19023494e-01
-8.00642848e-01 -1.05028200e+00 6.26865029e-01 4.96873796e-01
2.22787857e-01 -1.47016928e-01 1.22691870e+00 -1.24403620e+00
-7.29722679e-02 6.48777336e-02 -2.34688580e-01 2.14520730e-02
-5.36018088e-02 -3.66232365e-01 -6.46032870e-01 -2.76227623e-01
-7.20113873e-01 -5.70758164e-01 8.69918108e-01 6.38294518e-02
-2.10768029e-01 -2.84105480e-01 -3.32899153e-01 -1.41263008e-01
1.30107117e+00 1.08362392e-01 4.63424444e-01 4.35926944e-01
5.81416786e-01 8.51348042e-01 5.99587381e-01 -3.56929809e-01
5.36684871e-01 8.27170610e-01 -9.31898430e-02 -2.77434736e-01
-1.84991527e-02 -8.94187167e-02 7.45544612e-01 1.08111608e+00
-1.73394054e-01 -4.30641860e-01 -1.16498852e+00 9.32616174e-01
-1.98178828e+00 -6.13924384e-01 -1.32544920e-01 2.39427137e+00
9.66607630e-01 -2.24856138e-01 1.38931453e-01 -6.10021114e-01
9.09761667e-01 -2.72968352e-01 1.00470737e-01 -8.35386813e-01
-5.14223337e-01 2.67674506e-01 3.92986536e-01 9.76022720e-01
-9.91867065e-01 1.47308969e+00 6.05532074e+00 7.51890063e-01
-9.18338358e-01 3.12361896e-01 4.17652369e-01 2.45158896e-01
-5.74648023e-01 1.16830125e-01 -6.41854525e-01 2.01062888e-01
1.14707851e+00 -1.03078544e-01 5.59346735e-01 1.78102404e-01
-1.17383465e-01 3.38780768e-02 -1.17027116e+00 8.70871902e-01
3.75987202e-01 -9.40236866e-01 5.41117549e-01 -8.95930603e-02
8.06365788e-01 6.51273608e-01 -1.33440033e-01 4.20517892e-01
4.74862635e-01 -1.14937508e+00 5.06356835e-01 5.47251701e-01
1.03565967e+00 -8.02188694e-01 1.16685784e+00 5.11233918e-02
-9.66358840e-01 4.45288211e-01 -3.22583526e-01 5.05758286e-01
2.76280940e-01 -3.84492590e-03 -1.16147780e+00 1.01386786e+00
3.95580411e-01 2.96824336e-01 -4.39177275e-01 6.10951483e-01
-2.24171266e-01 5.52535176e-01 -5.36336042e-02 -2.36349851e-01
4.00879741e-01 -1.11315399e-01 9.67351615e-01 1.86080289e+00
3.53092432e-01 -2.91128725e-01 3.05096179e-01 3.68718922e-01
-2.05117777e-01 6.49219930e-01 -5.93463600e-01 -2.85434872e-01
2.26545334e-01 1.12436342e+00 -4.27800655e-01 -4.49118555e-01
-1.78654373e-01 1.37391150e+00 1.85043573e-01 5.89390278e-01
-1.52298346e-01 -4.74402428e-01 5.40434122e-01 -4.86177772e-01
-1.74545497e-01 -8.80119428e-02 -5.89541048e-02 -1.16790020e+00
-3.33366305e-01 -1.06306922e+00 3.78344446e-01 -1.01367652e+00
-1.53658175e+00 1.32436752e+00 -7.76343346e-02 -1.09827566e+00
-6.60611629e-01 -5.93968570e-01 -2.85059541e-01 1.46281385e+00
-1.24082851e+00 -1.43702841e+00 8.06551635e-01 3.97973508e-01
6.77211821e-01 -5.41354179e-01 1.28308320e+00 2.58161962e-01
-1.13833919e-01 7.33594954e-01 7.02889115e-02 8.23114961e-02
1.33129668e+00 -1.28398514e+00 5.66389740e-01 5.70966661e-01
2.73073018e-01 8.43935966e-01 6.43548548e-01 -8.54617715e-01
-1.18782508e+00 -8.42120707e-01 2.02925086e+00 -8.55682075e-01
7.78552175e-01 -2.84277529e-01 -6.96839094e-01 5.22391379e-01
9.82258797e-01 -8.74637663e-01 7.70858943e-01 2.06861377e-01
-2.22698361e-01 4.16595727e-01 -7.74838686e-01 8.17314923e-01
6.51944876e-01 -8.07955980e-01 -8.62500429e-01 5.54060102e-01
8.16199422e-01 -2.66293645e-01 -7.39812136e-01 4.46153164e-01
5.23358643e-01 -1.17006591e-02 8.98496211e-01 -1.21209967e+00
4.59429741e-01 -2.58388221e-01 -4.40030098e-01 -1.78184235e+00
-2.87979096e-01 -9.00952876e-01 8.83188248e-01 1.10818553e+00
1.36336172e+00 -6.26123071e-01 -3.02982312e-02 1.19165428e-01
-2.56471969e-02 -3.37026119e-01 -1.42389107e+00 -3.63609850e-01
4.55642074e-01 -2.37700537e-01 3.92686665e-01 1.28113842e+00
3.86867255e-01 1.21449649e+00 -4.04469937e-01 1.68657042e-02
1.97795942e-01 9.73680988e-04 3.73261571e-01 -1.04538643e+00
-7.00530633e-02 -7.91596234e-01 -4.67148162e-02 -7.89563239e-01
4.28193957e-01 -1.39416873e+00 3.80647689e-01 -1.78360915e+00
3.69799912e-01 2.07854494e-01 -4.33125108e-01 7.61901259e-01
-4.38589543e-01 4.60142046e-01 2.27034688e-01 2.80712634e-01
-8.64545584e-01 4.06940073e-01 9.46393073e-01 -2.51577526e-01
-1.44587576e-01 -1.79727867e-01 -4.41564888e-01 2.46788293e-01
6.54967010e-01 -5.25707722e-01 -1.90645270e-03 -7.79783487e-01
6.02332830e-01 3.83752674e-01 1.01642996e-01 -1.50443152e-01
3.29708010e-02 -1.66027710e-01 3.35838407e-01 -4.86724705e-01
2.98291475e-01 -4.46693033e-01 -3.75730306e-01 1.69950724e-01
-7.22088158e-01 7.51994789e-01 4.86182779e-01 1.96513683e-01
-2.04641595e-01 -2.29213476e-01 5.35185933e-01 -3.33755970e-01
-5.05330861e-01 -2.99186945e-01 -9.59088743e-01 -9.00060758e-02
2.22932830e-01 1.07569031e-01 -3.70339066e-01 -4.16573882e-01
-9.10905480e-01 4.12480235e-01 3.31244797e-01 8.60750973e-01
4.53943342e-01 -1.60943067e+00 -1.41064262e+00 -1.78905904e-01
4.81001079e-01 -9.06268954e-01 -1.05953045e-01 9.79857385e-01
1.75456271e-01 8.04630578e-01 -1.41592622e-01 -3.90568793e-01
-1.51083601e+00 2.85893589e-01 2.67548442e-01 -4.55208898e-01
2.42774755e-01 7.58125246e-01 -8.39718580e-02 -1.25209606e+00
-2.26979092e-01 1.54395938e-01 -4.32169110e-01 1.64766684e-01
3.00911188e-01 5.42712547e-02 1.76414564e-01 -1.31093836e+00
-5.79572260e-01 6.15944803e-01 8.98099318e-02 -1.11584306e+00
6.91249549e-01 -4.35956240e-01 -5.55542290e-01 5.74963212e-01
1.09435892e+00 1.31289214e-01 -3.19667570e-02 -4.06226635e-01
5.80508411e-01 1.63213640e-01 -4.12345886e-01 -1.53813374e+00
-1.59814417e-01 7.46325791e-01 7.33666718e-01 -2.41798773e-01
8.34863663e-01 1.71060041e-01 7.86472201e-01 7.38050222e-01
2.10862249e-01 -1.30903351e+00 -2.40198553e-01 1.20867443e+00
1.08957601e+00 -1.53411138e+00 -6.44917369e-01 2.15591505e-01
-1.02894127e+00 1.21021366e+00 2.96362817e-01 6.30288303e-01
-1.65903077e-01 -4.58365828e-01 6.14960015e-01 4.16387655e-02
-8.87128115e-01 -5.38428426e-01 1.09862924e+00 5.73012948e-01
8.88591886e-01 1.89575732e-01 -7.14068413e-01 5.62167585e-01
-1.23842299e-01 -3.94271612e-01 8.28173831e-02 5.95799863e-01
-2.51616418e-01 -1.57446301e+00 -4.32942927e-01 9.63414311e-02
-3.84505302e-01 -7.69666016e-01 -1.17663109e+00 5.20959556e-01
9.65928957e-02 1.34580672e+00 -2.89744437e-01 -6.21804953e-01
2.52781153e-01 6.03708506e-01 4.33638722e-01 -8.69065702e-01
-8.20133924e-01 4.97361392e-01 6.57999575e-01 -2.25910798e-01
-6.23958290e-01 -9.07874525e-01 -8.56975377e-01 3.78587216e-01
-1.42877117e-01 4.12644267e-01 8.35715413e-01 1.18045974e+00
3.94064903e-01 2.92439200e-02 2.45053396e-01 -1.01474702e+00
-2.72161186e-01 -1.48962402e+00 2.28205442e-01 2.82767624e-01
1.34638235e-01 -1.32153794e-01 -1.58526264e-02 1.64342448e-02] | [11.50938606262207, 1.5520899295806885] |
2b54100b-80ac-4a3f-8de0-2f1b33ff626a | population-based-evolutionary-gaming-for | 2306.05236 | null | https://arxiv.org/abs/2306.05236v1 | https://arxiv.org/pdf/2306.05236v1.pdf | Population-Based Evolutionary Gaming for Unsupervised Person Re-identification | Unsupervised person re-identification has achieved great success through the self-improvement of individual neural networks. However, limited by the lack of diversity of discriminant information, a single network has difficulty learning sufficient discrimination ability by itself under unsupervised conditions. To address this limit, we develop a population-based evolutionary gaming (PEG) framework in which a population of diverse neural networks is trained concurrently through selection, reproduction, mutation, and population mutual learning iteratively. Specifically, the selection of networks to preserve is modeled as a cooperative game and solved by the best-response dynamics, then the reproduction and mutation are implemented by cloning and fluctuating hyper-parameters of networks to learn more diversity, and population mutual learning improves the discrimination of networks by knowledge distillation from each other within the population. In addition, we propose a cross-reference scatter (CRS) to approximately evaluate re-ID models without labeled samples and adopt it as the criterion of network selection in PEG. CRS measures a model's performance by indirectly estimating the accuracy of its predicted pseudo-labels according to the cohesion and separation of the feature space. Extensive experiments demonstrate that (1) CRS approximately measures the performance of models without labeled samples; (2) and PEG produces new state-of-the-art accuracy for person re-identification, indicating the great potential of population-based network cooperative training for unsupervised learning. | ['Yonghong Tian', 'Xuesong Gao', 'Weiqiang Chen', 'Shiyong Li', 'Mengxi Jia', 'Peixi Peng', 'Yunpeng Zhai'] | 2023-06-08 | null | null | null | null | ['person-re-identification', 'unsupervised-person-re-identification'] | ['computer-vision', 'computer-vision'] | [ 2.55190611e-01 -1.07267536e-01 -1.43014237e-01 -1.83891252e-01
9.16278362e-02 -2.18612865e-01 3.95590365e-01 -1.82206810e-01
-6.76291943e-01 7.73933113e-01 -2.02040404e-01 3.11882168e-01
-6.13331318e-01 -9.29386377e-01 -2.78681964e-01 -9.26383436e-01
4.78479303e-02 6.86461627e-01 1.69553198e-02 -1.88143864e-01
1.08576277e-02 3.24649423e-01 -1.62255728e+00 -2.84736723e-01
1.53963864e+00 7.57196724e-01 -7.81835765e-02 3.55239511e-01
1.18418232e-01 4.69919533e-01 -6.19906902e-01 -4.78103012e-01
1.94298908e-01 -8.00827265e-01 -4.05344754e-01 6.41177595e-02
-9.72033069e-02 7.68737495e-02 -3.09680611e-01 1.23095715e+00
8.03018332e-01 3.20218831e-01 7.41923392e-01 -1.34147620e+00
-8.11706722e-01 8.44053507e-01 -4.04478490e-01 -6.58844337e-02
7.37696365e-02 2.81221718e-01 7.12894976e-01 -5.09825230e-01
2.68818319e-01 1.15614188e+00 7.14746833e-01 9.58471656e-01
-1.47922969e+00 -1.03399038e+00 1.86412930e-01 2.00502183e-02
-1.64819682e+00 -2.64259249e-01 8.94470215e-01 -3.82610947e-01
5.66971362e-01 2.83740103e-01 1.05706501e+00 8.74603748e-01
-2.93619990e-01 6.11007273e-01 7.65223742e-01 -3.95770729e-01
2.64132291e-01 3.98052186e-01 1.71689749e-01 7.44080424e-01
3.94183159e-01 5.04424155e-01 -5.65639138e-01 -4.84129451e-02
9.03605759e-01 -1.13880351e-01 -4.58824076e-02 -3.96821737e-01
-8.60682428e-01 6.82429910e-01 4.47835505e-01 2.38888457e-01
-2.33921438e-01 -2.82491773e-01 1.27774552e-01 3.14757347e-01
2.84494489e-01 7.83110678e-01 -1.22655675e-01 -2.21245289e-02
-7.66205668e-01 1.30163595e-01 6.22608960e-01 3.69414538e-01
6.94637418e-01 2.73727924e-01 -9.50051025e-02 1.23389339e+00
1.56529710e-01 4.18416649e-01 9.48401928e-01 -8.71919096e-01
1.67851448e-01 1.11715066e+00 -2.48508707e-01 -1.11895609e+00
-3.54242653e-01 -9.53234494e-01 -1.23524833e+00 1.21847801e-01
2.82433361e-01 -5.09711683e-01 -5.96346855e-01 2.08172989e+00
1.66154996e-01 3.95061940e-01 1.09221093e-01 6.59572959e-01
6.87425971e-01 2.33265206e-01 6.36976883e-02 -4.43462610e-01
7.61172473e-01 -7.64577508e-01 -3.08298200e-01 -1.37651294e-01
4.23667997e-01 2.34069347e-01 5.84476709e-01 2.37343118e-01
-9.93112683e-01 -8.79146457e-01 -9.78831351e-01 9.58683372e-01
-2.65717000e-01 2.22047329e-01 3.94466072e-01 9.87882495e-01
-9.00338590e-01 6.47346020e-01 -4.85875517e-01 -2.01633021e-01
5.76527178e-01 8.09113145e-01 -1.95086822e-01 2.61682272e-01
-1.31001568e+00 4.03705716e-01 6.88283086e-01 3.20286632e-01
-4.67823088e-01 -5.18121362e-01 -4.37054306e-01 1.49787873e-01
2.95055032e-01 -8.95615935e-01 4.48238164e-01 -1.45760524e+00
-1.87097132e+00 5.93695462e-01 2.89769799e-01 -4.36013043e-01
6.46592081e-01 1.22400790e-01 -6.24769688e-01 -1.77748322e-01
-1.21161781e-01 8.37977231e-01 7.66397834e-01 -1.29583716e+00
-6.55436337e-01 -4.82434541e-01 -1.43573597e-01 4.64351863e-01
-8.37316215e-01 -1.36345550e-01 -2.75127769e-01 -5.25070012e-01
1.55556425e-01 -9.93034601e-01 -2.89781541e-01 -2.57851124e-01
-3.76605004e-01 -2.80294359e-01 3.63437921e-01 -4.35544670e-01
1.36781096e+00 -2.09455156e+00 5.01104772e-01 7.82137275e-01
3.57708573e-01 4.78252858e-01 -2.29365826e-01 -5.25066331e-02
-5.44652715e-02 1.97425008e-01 -2.33595371e-01 -1.04733326e-01
-1.98279575e-01 -7.60034546e-02 2.94547498e-01 5.38893044e-02
1.06376350e-01 9.35262024e-01 -8.32602441e-01 -4.50440973e-01
-2.11416883e-03 2.35659003e-01 -5.88757694e-01 1.48741886e-01
2.27287412e-01 5.17641127e-01 -5.09270728e-01 5.03609061e-01
2.75578260e-01 -2.74767041e-01 5.36078155e-01 -1.32089391e-01
1.77977338e-01 -5.37999988e-01 -1.29911232e+00 9.80301559e-01
9.90623459e-02 1.96257770e-01 -3.65114778e-01 -1.10879982e+00
1.30833399e+00 -1.05197713e-01 4.51099724e-01 -6.35403693e-01
2.81753540e-01 1.07381642e-01 4.25044298e-01 -2.44140506e-01
1.67644247e-01 2.81718373e-01 1.44777551e-01 5.51786959e-01
-1.13342647e-02 4.22105521e-01 1.71331823e-01 -1.25564784e-01
6.53467834e-01 -1.51701316e-01 1.78758174e-01 -4.06776145e-02
6.95971668e-01 -3.17653030e-01 7.87717819e-01 1.05939376e+00
-3.78572613e-01 2.27352008e-01 2.34527841e-01 -2.29013249e-01
-9.05777097e-01 -1.03104496e+00 6.60990104e-02 1.08271456e+00
2.77510464e-01 1.45731568e-01 -7.83998549e-01 -7.80748725e-01
2.07807273e-01 4.75613773e-01 -6.97159290e-01 -8.39726627e-01
-4.29165125e-01 -1.14278483e+00 7.38499165e-01 5.27460456e-01
9.84479010e-01 -1.08016515e+00 -1.48801580e-02 2.93397270e-02
3.26699317e-02 -4.28287804e-01 -3.41621637e-01 -3.80168259e-02
-8.63578498e-01 -1.05105138e+00 -9.34693992e-01 -9.16577518e-01
9.54417527e-01 7.75007382e-02 6.46818757e-01 3.86925012e-01
-3.13205235e-02 2.40938067e-01 -8.48126486e-02 1.41671412e-02
-3.49442840e-01 2.64057934e-01 5.81230462e-01 3.37773651e-01
2.57430494e-01 -8.43488395e-01 -4.35035080e-01 6.77562654e-01
-4.01757926e-01 -5.84976599e-02 4.83902216e-01 1.08010232e+00
3.43540817e-01 5.21226406e-01 8.87743950e-01 -6.31018221e-01
1.11854959e+00 -2.94119209e-01 -3.69557649e-01 5.74591517e-01
-9.22709405e-01 1.49829602e-02 4.99252319e-01 -9.90704060e-01
-1.03520751e+00 -2.00411323e-02 3.23133796e-01 -1.61644161e-01
-3.98284802e-03 4.53872770e-01 -2.27427706e-01 -3.22632343e-01
6.81129932e-01 4.65997428e-01 4.34076250e-01 -1.89249694e-01
8.91792104e-02 6.26573443e-01 5.46281815e-01 -5.48641443e-01
8.04865956e-01 -5.16525432e-02 -2.09747076e-01 -4.68549162e-01
-4.09192473e-01 -9.21965539e-02 -7.88594842e-01 -6.32615447e-01
6.03249967e-01 -7.81196117e-01 -1.19959080e+00 8.70421648e-01
-4.96668756e-01 -2.79358834e-01 -4.67502624e-01 5.58361709e-01
-1.05481103e-01 2.67120272e-01 -2.37211645e-01 -1.11613536e+00
-3.85206461e-01 -8.42223465e-01 1.60138667e-01 8.08474362e-01
-2.14179620e-01 -8.19046676e-01 8.65323935e-03 3.35008979e-01
3.35518420e-01 -8.28273073e-02 8.10136676e-01 -8.75102520e-01
-1.04810156e-01 -2.73779005e-01 -2.23954380e-01 4.72216338e-01
8.94039944e-02 5.06437048e-02 -6.44655108e-01 -4.08610135e-01
-4.47388917e-01 -3.54570568e-01 7.48596430e-01 4.08726126e-01
1.07646048e+00 -1.14468910e-01 -5.63920319e-01 5.49222708e-01
1.02003515e+00 6.84519589e-01 4.52942014e-01 4.29883957e-01
7.22086430e-01 6.75855994e-01 2.01890748e-02 3.68023753e-01
2.90089995e-01 4.61462200e-01 3.59296687e-02 1.15888566e-01
1.74528480e-01 -3.37796897e-01 1.03972264e-01 8.25465500e-01
-6.99616313e-01 -1.90255761e-01 -9.14740145e-01 7.01851696e-02
-1.99044621e+00 -1.06652844e+00 4.38429803e-01 2.44699287e+00
6.99028015e-01 1.59542844e-01 6.62503958e-01 8.23022351e-02
1.22451293e+00 -3.43545318e-01 -1.02901101e+00 2.88586497e-01
-4.39984113e-01 -2.70001411e-01 3.02642822e-01 2.14795828e-01
-7.86150217e-01 7.29238987e-01 6.32008314e+00 8.96935642e-01
-9.36069548e-01 -2.04248846e-01 9.81762946e-01 -1.06539585e-01
1.57677699e-02 -2.62026012e-01 -8.54328156e-01 7.56152868e-01
5.80708623e-01 -2.63746262e-01 7.77784050e-01 6.70284331e-01
-5.22899330e-02 3.65162119e-02 -7.27623522e-01 1.19412649e+00
1.20524600e-01 -1.06073606e+00 1.88153952e-01 8.50354061e-02
9.93488073e-01 -4.71372485e-01 2.21370131e-01 3.44276756e-01
5.69240451e-01 -9.44998562e-01 4.76632893e-01 8.99134576e-01
6.09170496e-01 -1.02405441e+00 7.29062140e-01 5.04008472e-01
-1.04884052e+00 -6.83249950e-01 -2.84637332e-01 1.36609837e-01
-2.06233621e-01 1.31465629e-01 -4.93975967e-01 4.70339090e-01
4.64043289e-01 6.19265974e-01 -8.95233512e-01 1.09804428e+00
5.87877929e-02 4.94265735e-01 -2.57763624e-01 -3.87120128e-01
-2.57794768e-01 -5.00406504e-01 5.06047785e-01 6.26424730e-01
2.12518439e-01 2.92765559e-03 2.32219964e-01 9.21135724e-01
-4.03373502e-02 -1.47259217e-02 -8.20322633e-02 -1.49279982e-01
8.57317984e-01 8.31175923e-01 -6.59937024e-01 -1.85007185e-01
2.22468123e-01 8.74996066e-01 4.90309209e-01 5.05932271e-01
-6.20822906e-01 -2.86085635e-01 3.37007016e-01 -1.84612125e-01
-9.42418873e-02 8.47639814e-02 -2.47517169e-01 -9.95640278e-01
-2.42262021e-01 -8.12800765e-01 4.57445920e-01 -4.14491445e-01
-1.58619523e+00 6.54250622e-01 -2.14900792e-01 -1.30608940e+00
-2.76407450e-01 -2.36910701e-01 -5.32774448e-01 6.93116665e-01
-8.72162700e-01 -7.51660585e-01 -4.99704361e-01 5.62875271e-01
3.95091251e-03 -1.02786279e+00 6.09666348e-01 1.99243426e-01
-1.21880853e+00 1.14767838e+00 4.09728765e-01 4.55009699e-01
4.91482794e-01 -8.25311005e-01 -1.15037769e-01 5.39957047e-01
-7.53837749e-02 6.04457974e-01 2.72226810e-01 -8.62684369e-01
-9.53502655e-01 -9.48748171e-01 4.76000398e-01 5.07010818e-02
3.54428679e-01 -1.02795526e-01 -7.42737114e-01 2.05773383e-01
-4.40887570e-01 -2.93337822e-01 7.43069053e-01 2.65807986e-01
-1.46421790e-01 -5.03353655e-01 -1.11724448e+00 6.28499687e-01
1.25143504e+00 -3.83307844e-01 -2.05996647e-01 7.87548050e-02
4.32250082e-01 4.72057126e-02 -7.73366988e-01 4.09624666e-01
9.31536257e-01 -8.16385746e-01 1.09882092e+00 -6.25335217e-01
9.27497521e-02 -1.13057010e-01 2.45043531e-01 -1.39055336e+00
-7.69894123e-01 -3.63858730e-01 8.56534988e-02 1.44800520e+00
4.13214356e-01 -1.04092586e+00 1.04445207e+00 7.57061541e-01
4.10475552e-01 -6.29950702e-01 -7.70647824e-01 -9.17066395e-01
-1.51916236e-01 3.78624499e-02 8.38504314e-01 1.13063300e+00
6.31848862e-03 2.85207480e-01 -4.46769595e-01 -5.21285944e-02
7.40121305e-01 -2.40420401e-01 5.63121915e-01 -1.67863846e+00
-4.46434468e-01 -8.71683776e-01 -4.46500808e-01 -7.90590167e-01
2.36633748e-01 -9.12036777e-01 -1.25029281e-01 -8.79023194e-01
4.81209189e-01 -7.21843839e-01 -4.93425131e-01 2.10006535e-01
-3.53593141e-01 1.32866919e-01 2.19938755e-01 5.27200401e-01
-7.16095448e-01 7.74972975e-01 1.13621032e+00 -2.25746229e-01
-7.76341558e-01 2.61200756e-01 -8.25604618e-01 5.72783649e-01
8.28127325e-01 -3.43558997e-01 -4.91502196e-01 7.07481895e-03
1.12219743e-01 5.80374040e-02 4.01013613e-01 -1.39750183e+00
5.66522837e-01 -2.47115432e-03 8.52247119e-01 -1.24459438e-01
1.89803004e-01 -5.35440564e-01 4.01133537e-01 7.62117922e-01
-5.78731894e-01 -9.90926772e-02 -1.99690148e-01 7.07743406e-01
5.62701225e-02 -3.00471187e-01 7.21803844e-01 -4.61306497e-02
-4.94484633e-01 4.41233009e-01 -2.20909134e-01 4.71545830e-02
7.91585386e-01 -7.58494973e-01 -1.26429006e-01 -1.83650598e-01
-9.21159387e-01 3.39984983e-01 3.66001517e-01 2.82801569e-01
5.88599384e-01 -1.35536897e+00 -7.81279743e-01 4.98806775e-01
-1.44304946e-01 -3.43188614e-01 3.48650694e-01 5.28206348e-01
-5.98839521e-02 -1.09826021e-01 -4.37087536e-01 -6.47347093e-01
-1.28969789e+00 1.20986603e-01 7.99421787e-01 -3.00899237e-01
1.68893137e-03 9.20486391e-01 4.52225190e-03 -7.68642545e-01
2.67018497e-01 5.62296391e-01 -6.86968446e-01 1.14735886e-01
4.14616138e-01 6.04420960e-01 -5.14498711e-01 -5.15525103e-01
-1.88356370e-01 4.59957898e-01 -1.12751789e-01 -1.17049711e-02
1.21418357e+00 -7.52197355e-02 -1.26490280e-01 2.62534320e-01
7.04654634e-01 -3.86497319e-01 -1.14464974e+00 -3.41286838e-01
-2.36828774e-01 -2.13178247e-01 -2.43925184e-01 -1.01395798e+00
-1.11167252e+00 2.09158808e-01 9.90727127e-01 1.69748902e-01
1.20288062e+00 -3.83406967e-01 2.71699071e-01 5.31178951e-01
2.35211655e-01 -1.31928372e+00 3.64085317e-01 4.55286801e-01
4.70913202e-01 -1.15808427e+00 -1.23517133e-01 -2.02004045e-01
-6.47689641e-01 1.08040726e+00 1.02389300e+00 -6.78645447e-02
6.75010681e-01 -1.91300556e-01 -3.17255586e-01 1.81958750e-01
-3.47324103e-01 -4.59787175e-02 4.66469795e-01 9.23288405e-01
-8.83708447e-02 2.09053949e-01 -1.56964973e-01 8.97001505e-01
-1.68451011e-01 5.22130392e-02 -1.73763141e-01 4.84661609e-01
-4.82832968e-01 -9.24296916e-01 -1.95203930e-01 6.43011868e-01
2.39260420e-01 3.56616452e-02 -5.63753903e-01 6.17354870e-01
3.19866806e-01 8.67299914e-01 1.31951556e-01 -9.55007076e-01
1.87630892e-01 8.86379182e-03 4.33921099e-01 -2.42476076e-01
-8.15611362e-01 -1.90037534e-01 -1.12861253e-01 -7.28616118e-02
-4.22827214e-01 -5.76330602e-01 -9.76150990e-01 -3.06209236e-01
-5.26759982e-01 3.68174404e-01 2.14788437e-01 9.51848626e-01
4.02666450e-01 5.74613512e-01 8.51472080e-01 -5.89900434e-01
-6.78016841e-01 -8.15464437e-01 -6.84718549e-01 2.61633188e-01
-3.14730912e-01 -7.94335663e-01 -4.22865182e-01 -1.89022481e-01] | [14.818310737609863, 1.1178867816925049] |
df8b007b-5806-462e-a639-95b770692d20 | continuous-conditional-random-field | 2110.06085 | null | https://arxiv.org/abs/2110.06085v1 | https://arxiv.org/pdf/2110.06085v1.pdf | Continuous Conditional Random Field Convolution for Point Cloud Segmentation | Point cloud segmentation is the foundation of 3D environmental perception for modern intelligent systems. To solve this problem and image segmentation, conditional random fields (CRFs) are usually formulated as discrete models in label space to encourage label consistency, which is actually a kind of postprocessing. In this paper, we reconsider the CRF in feature space for point cloud segmentation because it can capture the structure of features well to improve the representation ability of features rather than simply smoothing. Therefore, we first model the point cloud features with a continuous quadratic energy model and formulate its solution process as a message-passing graph convolution, by which it can be easily integrated into a deep network. We theoretically demonstrate that the message passing in the graph convolution is equivalent to the mean-field approximation of a continuous CRF model. Furthermore, we build an encoder-decoder network based on the proposed continuous CRF graph convolution (CRFConv), in which the CRFConv embedded in the decoding layers can restore the details of high-level features that were lost in the encoding stage to enhance the location ability of the network, thereby benefiting segmentation. Analogous to the CRFConv, we show that the classical discrete CRF can also work collaboratively with the proposed network via another graph convolution to further improve the segmentation results. Experiments on various point cloud benchmarks demonstrate the effectiveness and robustness of the proposed method. Compared with the state-of-the-art methods, the proposed method can also achieve competitive segmentation performance. | ['Zhong Jin', 'Huan Wang', 'Franck Davoine', 'Fei Yang'] | 2021-10-12 | null | null | null | null | ['point-cloud-segmentation'] | ['computer-vision'] | [ 1.36280864e-01 1.28386825e-01 2.19641730e-01 -6.19704247e-01
-3.08878779e-01 -3.32533032e-01 3.91006261e-01 1.77873760e-01
-3.72045338e-01 1.83685988e-01 -3.06450099e-01 -2.68856853e-01
4.52450328e-02 -1.13550854e+00 -1.01315594e+00 -7.14145362e-01
1.25117227e-01 3.13620389e-01 4.13105547e-01 7.52261430e-02
2.58731931e-01 7.00529397e-01 -1.31862688e+00 1.45840188e-02
1.27795851e+00 1.08424759e+00 6.66831136e-01 1.58526391e-01
-7.27894902e-01 4.95269924e-01 -3.14701378e-01 -1.53174505e-01
2.33463258e-01 -9.31007043e-02 -8.22765291e-01 4.39277589e-01
1.57968447e-01 -2.34916762e-01 -3.04248482e-01 1.45304513e+00
5.49208336e-02 2.51366317e-01 6.34521604e-01 -1.10851431e+00
-6.91685915e-01 3.25598568e-01 -7.43595839e-01 -4.90576386e-01
7.40339467e-03 -1.70816302e-01 6.35417819e-01 -8.19991589e-01
2.84160703e-01 1.39923167e+00 5.98597527e-01 1.49070695e-01
-9.51098204e-01 -6.80378497e-01 5.38758337e-01 -6.42232448e-02
-1.59417725e+00 9.04422328e-02 8.71348977e-01 -4.22027290e-01
7.13486314e-01 1.07433043e-01 7.61243641e-01 1.98917940e-01
2.08165124e-01 8.16282213e-01 8.17685664e-01 -1.28343940e-01
1.89025417e-01 -7.40607008e-02 2.61085331e-01 7.99609065e-01
-5.32636307e-02 -1.21954188e-01 1.55964956e-01 6.30648509e-02
1.06334686e+00 4.58307803e-01 -2.23592609e-01 -1.55075535e-01
-9.98540878e-01 8.34150434e-01 1.13098669e+00 1.73956767e-01
-2.89936781e-01 3.34535003e-01 3.51687847e-03 -1.03610568e-01
6.27767026e-01 -1.92760974e-01 -2.29034632e-01 5.47940910e-01
-9.46345448e-01 1.03657618e-02 5.36114335e-01 1.32288599e+00
1.26566195e+00 -2.08339289e-01 -2.44019330e-01 7.06984878e-01
9.79922354e-01 4.91361648e-01 -1.40243992e-02 -8.70093405e-01
2.70317495e-01 8.15151274e-01 -1.60000250e-01 -1.02887678e+00
-3.60254884e-01 -5.49626529e-01 -1.18386483e+00 2.12868392e-01
7.47213811e-02 9.27831456e-02 -1.38074565e+00 1.41380608e+00
4.29184228e-01 6.70325637e-01 -1.74009934e-01 9.79577839e-01
8.99343669e-01 1.00887775e+00 1.56740159e-01 -6.44784495e-02
1.17692924e+00 -8.99104714e-01 -5.82963884e-01 -1.19750328e-01
3.59940559e-01 -6.91078067e-01 6.58636391e-01 1.22406535e-01
-6.99405611e-01 -7.84919679e-01 -9.25694287e-01 -2.63495266e-01
-3.15907419e-01 -6.64921384e-03 8.18875551e-01 3.46219599e-01
-1.01685035e+00 6.71262801e-01 -1.04773235e+00 -1.36376858e-01
5.66080272e-01 5.04906356e-01 -3.54367718e-02 -1.90239713e-01
-9.64264929e-01 4.40020502e-01 4.68216866e-01 6.62092566e-01
-5.36985755e-01 -3.46810490e-01 -9.20672238e-01 1.66062891e-01
2.22457185e-01 -6.87400997e-01 8.66685331e-01 -6.88548446e-01
-1.45814705e+00 6.21441543e-01 -2.49454543e-01 -1.33316949e-01
2.56570607e-01 -5.31479716e-02 -9.42874178e-02 6.66430369e-02
1.35201424e-01 9.76765752e-01 7.17735410e-01 -1.51808393e+00
-6.82331622e-01 -2.46336028e-01 8.78967643e-02 2.31675550e-01
2.17208266e-01 -2.30016202e-01 -8.67578328e-01 -3.75343621e-01
7.78164327e-01 -8.64452183e-01 -6.24436975e-01 1.30282845e-02
-6.11126542e-01 -4.19807136e-01 8.28331232e-01 -4.47092384e-01
7.81118095e-01 -2.44091105e+00 -7.94815421e-02 5.42713761e-01
3.46669197e-01 -5.90253435e-02 1.14821725e-01 1.26323074e-01
3.21376650e-03 2.46109575e-01 -7.75467813e-01 -5.11989594e-01
4.79953922e-02 4.92058784e-01 -1.52048275e-01 7.51553595e-01
3.47566724e-01 1.01726770e+00 -7.43111551e-01 -6.79172873e-01
4.71310407e-01 5.86110353e-01 -5.19176662e-01 2.19147608e-01
-4.12356526e-01 6.56254232e-01 -8.05137634e-01 4.66125786e-01
1.42214906e+00 -3.29787970e-01 -2.04341114e-01 -2.14176640e-01
-3.45372021e-01 -1.08923621e-01 -1.21268368e+00 1.90919375e+00
-1.58974186e-01 1.79682553e-01 1.68446496e-01 -1.18756902e+00
1.18606007e+00 -9.08710063e-02 5.63350976e-01 -4.32827890e-01
2.74555087e-01 4.17690277e-02 -3.93189907e-01 -1.40559316e-01
4.47678804e-01 -1.38765052e-01 -3.08358911e-02 -1.78623274e-01
-8.43709409e-02 -4.41117167e-01 -2.80654132e-01 1.03641286e-01
6.36983335e-01 1.56261653e-01 -3.11155945e-01 -2.62322932e-01
6.68188035e-01 7.63455182e-02 7.05082297e-01 6.68733776e-01
1.90008000e-01 7.04018652e-01 9.04990584e-02 -7.16122091e-02
-6.42244458e-01 -1.10559809e+00 -3.71656448e-01 4.56442803e-01
8.54193568e-01 -1.87733367e-01 -8.23253632e-01 -6.96559131e-01
-9.77673009e-02 5.15760779e-01 -1.84336960e-01 -3.55491671e-03
-4.50851440e-01 -6.06502831e-01 1.57942846e-01 4.58322138e-01
1.07767022e+00 -9.92979527e-01 -1.00044847e-01 3.11866790e-01
-9.31789503e-02 -1.20703483e+00 -3.64848435e-01 3.07036549e-01
-8.28423142e-01 -7.07434893e-01 -6.01505637e-01 -1.19288111e+00
9.63299990e-01 4.82238829e-01 8.09020996e-01 5.19141078e-01
2.24942520e-01 1.15818895e-01 -5.26403308e-01 -2.54862398e-01
-1.06574595e-01 -1.76514573e-02 -4.29786831e-01 1.17127441e-01
3.11861694e-01 -6.61936224e-01 -6.11794829e-01 9.99601930e-02
-1.12540114e+00 3.92824590e-01 6.72214448e-01 6.29895747e-01
1.04218054e+00 4.13572192e-01 1.14246212e-01 -8.95007908e-01
2.67401040e-01 -4.23687279e-01 -9.48291004e-01 9.90607589e-02
-4.57696676e-01 -1.59322675e-02 6.13705397e-01 6.40368313e-02
-1.15259051e+00 6.37813747e-01 -6.12281561e-01 -5.37329912e-01
-3.89840811e-01 6.35447502e-01 -3.83458287e-01 -4.52767760e-01
-6.81979284e-02 3.41063410e-01 -8.20440799e-02 -6.45068407e-01
6.38684273e-01 7.10929513e-01 5.64331710e-01 -5.19010127e-01
9.87546146e-01 6.55882359e-01 1.96818456e-01 -7.18586743e-01
-7.23004222e-01 -6.22506797e-01 -7.99522042e-01 -1.85418814e-01
1.36228251e+00 -9.54736590e-01 -7.49747217e-01 7.26489305e-01
-1.53030026e+00 -1.65784299e-01 -1.66416481e-01 5.29858053e-01
-5.09638488e-01 6.43963337e-01 -5.76083004e-01 -6.67919874e-01
-4.28955257e-02 -1.34943175e+00 1.42442584e+00 6.38593674e-01
7.37184703e-01 -9.50314999e-01 -2.42151424e-01 -8.33787471e-02
8.91012624e-02 2.46890739e-01 8.93514812e-01 -3.14898074e-01
-1.03403437e+00 3.62771265e-02 -6.47500396e-01 4.49216992e-01
-1.26658633e-01 -4.10398170e-02 -9.75313127e-01 -1.34013951e-01
2.24449828e-01 1.23011194e-01 1.04631329e+00 5.34613431e-01
1.45037401e+00 2.30325945e-02 -5.31261444e-01 1.07507634e+00
1.72693741e+00 1.63584024e-01 1.00680542e+00 7.09840050e-03
1.15881741e+00 4.17914689e-01 5.51273346e-01 1.70028612e-01
6.33463204e-01 3.18957776e-01 7.74925172e-01 -6.07416749e-01
3.09368968e-02 -3.32527667e-01 4.94187251e-02 1.09530413e+00
7.30553716e-02 -2.27349520e-01 -7.67638862e-01 2.65385628e-01
-1.99372375e+00 -5.12242556e-01 -7.60932446e-01 1.82447827e+00
4.04044807e-01 -3.20360176e-02 -4.64052171e-01 -7.30766356e-02
9.37068999e-01 8.01253691e-02 -4.27905321e-01 -2.38580585e-01
2.57629491e-02 3.12993795e-01 7.76490808e-01 4.89680499e-01
-1.27420473e+00 1.06992269e+00 5.50467777e+00 1.04255605e+00
-1.12507427e+00 7.59347305e-02 4.62333053e-01 7.89636791e-01
-3.82528186e-01 3.37436974e-01 -6.87719047e-01 5.09088814e-01
2.88256437e-01 1.98804170e-01 4.28699046e-01 8.27245653e-01
5.46591245e-02 -1.00181445e-01 -8.28994036e-01 9.38627899e-01
-3.98670286e-01 -1.14863348e+00 -3.37846652e-02 2.55477637e-01
6.22306168e-01 2.29678586e-01 -1.81561634e-01 2.93370575e-01
3.06313336e-01 -1.06725049e+00 6.86556041e-01 6.83207631e-01
7.15508342e-01 -7.28945255e-01 7.82289803e-01 6.45668209e-01
-1.53162539e+00 2.13413626e-01 -8.87261212e-01 -5.79741225e-03
3.97955626e-01 9.09103096e-01 -5.03955662e-01 9.75984216e-01
6.66408956e-01 9.18156207e-01 -4.23479944e-01 1.34152055e+00
-2.30599672e-01 4.69788522e-01 -5.83916724e-01 1.47807330e-01
4.70404387e-01 -6.34094656e-01 3.98225784e-01 1.26017463e+00
3.22651237e-01 3.25726151e-01 5.89019060e-01 1.26250339e+00
-4.85635623e-02 1.47769883e-01 -3.35768849e-01 2.17300758e-01
2.72166461e-01 1.28677177e+00 -1.07987523e+00 -3.03006113e-01
-5.13099194e-01 1.00032437e+00 3.51269931e-01 4.18043911e-01
-8.67327571e-01 -4.91626024e-01 3.51174086e-01 -1.83230698e-01
4.56316203e-01 -5.22582710e-01 -4.01544154e-01 -1.06625700e+00
-1.02475069e-01 -1.76818535e-01 -5.86583652e-02 -7.01572001e-01
-1.45561373e+00 6.45776272e-01 -5.64699098e-02 -1.29382706e+00
4.05702233e-01 -4.02089268e-01 -6.87285841e-01 1.19443989e+00
-1.90386128e+00 -1.30445373e+00 -6.66229784e-01 7.35617697e-01
2.14738294e-01 5.29064238e-01 5.42172372e-01 3.61943662e-01
-4.04788226e-01 9.19607505e-02 1.40483946e-01 1.37563139e-01
2.00549468e-01 -1.29229712e+00 4.82314855e-01 9.10374045e-01
-8.13861936e-02 4.88872916e-01 1.23706728e-01 -9.57092285e-01
-1.08912861e+00 -1.58165681e+00 4.75944817e-01 1.28812596e-01
1.75068706e-01 -4.75020736e-01 -1.25302994e+00 5.19470930e-01
7.37985894e-02 1.11023232e-01 1.59361973e-01 -3.45426381e-01
2.38100842e-01 8.95031542e-02 -1.13393402e+00 1.38108179e-01
1.14484489e+00 -3.53919476e-01 -4.94595736e-01 4.54026192e-01
1.21217036e+00 -6.57479525e-01 -7.40345001e-01 6.53067946e-01
6.97774952e-03 -8.17316771e-01 9.54346538e-01 -8.09348077e-02
1.81675613e-01 -7.65911520e-01 -1.88123062e-01 -1.11616302e+00
-6.42600358e-01 -1.98258981e-01 4.14485574e-01 1.42765796e+00
6.58530220e-02 -6.33868694e-01 7.06848741e-01 4.75904495e-01
-6.03646457e-01 -6.70283318e-01 -9.28415298e-01 -6.19928837e-01
1.25611097e-01 -7.02668786e-01 9.38773632e-01 7.89287567e-01
-5.84832251e-01 1.08782738e-01 8.23448151e-02 6.74178123e-01
8.32932115e-01 2.69326597e-01 5.50333321e-01 -1.39803100e+00
2.43538115e-02 -3.48563015e-01 -4.13818866e-01 -1.79743993e+00
4.03119981e-01 -1.32502401e+00 4.27254111e-01 -2.05259466e+00
-3.01724300e-03 -9.51740086e-01 -1.61483347e-01 3.55195999e-01
-2.26793379e-01 5.37339300e-02 2.26712421e-01 2.77938336e-01
-5.23884475e-01 8.44248891e-01 1.59480810e+00 -3.34969342e-01
-2.24553347e-01 9.29002836e-02 -4.74466443e-01 7.71376491e-01
5.44521391e-01 -5.67028701e-01 -3.03481340e-01 -5.81522286e-01
-7.55396262e-02 -4.32360768e-02 5.81778586e-01 -9.93020713e-01
5.75703382e-01 -2.39528477e-01 4.47724313e-01 -9.99272227e-01
2.18044207e-01 -1.12839866e+00 1.97157308e-01 8.78986791e-02
1.02912329e-01 -3.83739978e-01 4.56072465e-02 6.67222440e-01
-3.37062180e-01 -2.87232310e-01 6.48694038e-01 -2.22142741e-01
-7.08884001e-01 7.17424095e-01 -1.73511863e-01 -3.51995587e-01
9.33153629e-01 -2.45660305e-01 -1.32131025e-01 -8.12104866e-02
-6.06320918e-01 5.88636279e-01 4.76928443e-01 8.99121836e-02
8.84762526e-01 -1.32062948e+00 -4.34873372e-01 3.03671449e-01
-2.43378833e-01 9.16530073e-01 3.43574792e-01 7.41788387e-01
-6.87015116e-01 1.31402597e-01 2.47485824e-02 -1.05499089e+00
-5.79003036e-01 5.45552433e-01 3.74307901e-01 -8.69995281e-02
-8.27636540e-01 8.23747575e-01 6.46682441e-01 -5.43601692e-01
7.21976310e-02 -6.41696036e-01 -3.21914941e-01 -3.66738677e-01
-4.26749028e-02 -4.27025631e-02 -9.61579531e-02 -7.42085040e-01
-4.36512023e-01 9.63477492e-01 1.41721219e-01 1.34433508e-01
1.27958512e+00 -2.64377564e-01 -5.52346230e-01 1.31041378e-01
1.32621098e+00 -9.51279625e-02 -1.31450009e+00 -1.91294298e-01
-3.30865055e-01 -4.27856654e-01 4.38705027e-01 -3.24261099e-01
-1.28960228e+00 1.02428997e+00 6.07648075e-01 3.31987143e-01
1.18054175e+00 7.87131339e-02 8.99519026e-01 1.03198528e-01
4.59825218e-01 -7.37100005e-01 -5.46984494e-01 6.14953995e-01
5.97854078e-01 -1.08774543e+00 -9.32019800e-02 -1.02453768e+00
-3.27244520e-01 1.15521169e+00 5.10536849e-01 -4.40994889e-01
1.10314131e+00 1.51405275e-01 -1.14977561e-01 -4.13942754e-01
-3.16338576e-02 -5.17633677e-01 3.60125601e-01 5.99480689e-01
4.68121096e-02 1.56194404e-01 -2.26573139e-01 7.77329862e-01
-5.55050634e-02 -2.18551815e-03 2.39435956e-01 7.05984414e-01
-5.92229903e-01 -1.00769353e+00 -2.33218789e-01 4.23245043e-01
-8.86086673e-02 -1.27134904e-01 -7.87486881e-02 4.37409788e-01
5.13814688e-01 1.05860138e+00 3.14640582e-01 -5.82886279e-01
3.67093354e-01 -2.76438236e-01 1.98724061e-01 -7.57905066e-01
-3.78815413e-01 3.73540074e-01 -5.42927086e-01 -5.12443483e-01
-5.69035411e-01 -4.32825893e-01 -1.90180683e+00 -9.21778679e-02
-6.46592736e-01 2.37814426e-01 8.58877897e-01 1.12800813e+00
3.04048151e-01 7.29564905e-01 8.54212284e-01 -9.46181297e-01
-1.25003859e-01 -7.19207227e-01 -8.17439079e-01 1.83195710e-01
1.96755990e-01 -7.65730321e-01 -2.82874346e-01 -1.65931452e-02] | [8.122750282287598, -3.1921639442443848] |
3ee90f04-e3e1-499b-a60a-210ed9f292dd | synthesizing-mixed-type-electronic-health | 2302.14679 | null | https://arxiv.org/abs/2302.14679v1 | https://arxiv.org/pdf/2302.14679v1.pdf | Synthesizing Mixed-type Electronic Health Records using Diffusion Models | Electronic Health Records (EHRs) contain sensitive patient information, which presents privacy concerns when sharing such data. Synthetic data generation is a promising solution to mitigate these risks, often relying on deep generative models such as Generative Adversarial Networks (GANs). However, recent studies have shown that diffusion models offer several advantages over GANs, such as generation of more realistic synthetic data and stable training in generating data modalities, including image, text, and sound. In this work, we investigate the potential of diffusion models for generating realistic mixed-type tabular EHRs, comparing TabDDPM model with existing methods on four datasets in terms of data quality, utility, privacy, and augmentation. Our experiments demonstrate that TabDDPM outperforms the state-of-the-art models across all evaluation metrics, except for privacy, which confirms the trade-off between privacy and utility. | ['David A. Clifton', 'Andrew P. Creagh', 'Tingting Zhu', 'Vinod Kumar Chauhan', 'Ghadeer O. Ghosheh', 'Taha Ceritli'] | 2023-02-28 | null | null | null | null | ['synthetic-data-generation', 'synthetic-data-generation', 'type'] | ['medical', 'miscellaneous', 'speech'] | [ 1.87156916e-01 6.75645292e-01 1.57361284e-01 -2.73993224e-01
-1.09768665e+00 -4.76299256e-01 4.98879880e-01 1.74810439e-01
-1.50173202e-01 9.58284676e-01 6.63362145e-01 -2.66873270e-01
2.97922373e-01 -1.10338473e+00 -6.79362118e-01 -5.70818543e-01
1.14564866e-01 3.05790484e-01 -5.95014811e-01 6.55257106e-02
-4.55060512e-01 5.63936494e-02 -7.20641553e-01 3.91488791e-01
9.56529200e-01 9.78160381e-01 -5.57354033e-01 3.46894890e-01
5.66002429e-02 7.43148565e-01 -7.74751604e-01 -1.21270704e+00
4.54640329e-01 -6.94277346e-01 -4.95568335e-01 -2.47262880e-01
-3.76844145e-02 -7.12228835e-01 -5.69112718e-01 1.07081032e+00
1.05562198e+00 -4.30755526e-01 3.51917088e-01 -1.44063306e+00
-1.18950272e+00 6.90195501e-01 -3.12084556e-01 -5.15167058e-01
5.02350986e-01 6.61260903e-01 6.03177428e-01 -2.76981562e-01
8.45606148e-01 1.18671775e+00 9.60225105e-01 1.00999498e+00
-1.38239241e+00 -7.73395658e-01 -4.07598406e-01 -4.59633052e-01
-1.29825366e+00 -4.84067738e-01 5.50063729e-01 -3.42406511e-01
4.02573079e-01 5.04128397e-01 6.36149168e-01 1.65010083e+00
3.51962894e-01 8.95576715e-01 1.21532738e+00 5.24269752e-02
3.13416570e-01 4.38654095e-01 -4.15175825e-01 2.42467239e-01
4.79474962e-01 1.70909539e-01 -3.01718414e-01 -6.58511281e-01
6.89224422e-01 1.99973315e-01 -3.86595458e-01 -1.16057567e-01
-1.09806061e+00 8.46178174e-01 2.41739586e-01 -8.98259282e-02
-7.45911002e-01 7.10024685e-02 4.60453391e-01 1.98443606e-01
5.59842110e-01 4.83265072e-01 -3.48836072e-02 -9.01462361e-02
-5.70701361e-01 4.18336868e-01 7.03198314e-01 1.31949294e+00
1.06538944e-01 -1.93212964e-02 -9.04343605e-01 5.01148283e-01
2.65287817e-01 6.58498347e-01 3.20434868e-01 -6.99687779e-01
7.15224326e-01 5.24385571e-01 1.09967671e-01 -7.78372347e-01
-1.15911542e-02 -7.43192285e-02 -1.43866670e+00 -7.35876411e-02
1.48648918e-01 -6.55179203e-01 -1.05866730e+00 1.91537941e+00
2.01148942e-01 -2.04893470e-01 4.61998105e-01 6.45937681e-01
1.13450682e+00 5.66591024e-01 4.98264432e-01 -1.98780864e-01
1.29005373e+00 -5.14021218e-01 -1.22929251e+00 1.20398775e-01
4.39556628e-01 -4.49910790e-01 9.51898396e-01 -3.94184096e-03
-1.42657208e+00 -1.91903189e-01 -6.35649204e-01 -1.71763115e-02
-1.81536034e-01 -4.30154324e-01 5.38127065e-01 1.19188142e+00
-1.09597266e+00 2.92723417e-01 -7.56715238e-01 -1.86303124e-01
9.93720889e-01 7.10921660e-02 -3.60039681e-01 -2.37334833e-01
-1.36211991e+00 3.74069899e-01 9.69790891e-02 -2.31008425e-01
-9.68460023e-01 -9.20105875e-01 -9.70355570e-01 1.97602838e-01
1.19048217e-02 -1.24142575e+00 9.64145839e-01 -5.95571458e-01
-1.33011544e+00 7.37856209e-01 3.13602567e-01 -7.07125962e-01
1.25890779e+00 -9.11131501e-02 -4.51419234e-01 -1.98840797e-01
-1.87689275e-01 8.36949825e-01 4.64659214e-01 -1.32572579e+00
-1.40779745e-02 -4.61830825e-01 -1.15685232e-01 -4.54524681e-02
-3.76124620e-01 -2.39651769e-01 -1.92102253e-01 -8.83722603e-01
-4.18900013e-01 -9.32835340e-01 -5.15989244e-01 1.92740541e-02
-9.10324156e-01 2.93955982e-01 6.74361289e-01 -9.51985717e-01
1.16984153e+00 -2.31837916e+00 -2.66902804e-01 2.59900908e-03
3.01482707e-01 2.98096359e-01 1.83544923e-02 4.97761220e-01
1.73736200e-01 7.05978036e-01 -3.80305797e-01 -6.62569344e-01
2.97823846e-02 1.73268057e-02 -2.12956980e-01 1.49289444e-01
2.86936939e-01 1.52202737e+00 -6.39550269e-01 -4.20017600e-01
-1.10552713e-01 9.13620353e-01 -5.75924397e-01 5.22853315e-01
-2.77126938e-01 8.74518454e-01 -4.48224038e-01 6.65079176e-01
1.08138764e+00 -2.76655138e-01 1.90766469e-01 7.41042644e-02
5.25900841e-01 9.97816548e-02 -7.34711111e-01 1.53261375e+00
-8.95202979e-02 8.29197392e-02 -1.89413100e-01 -1.25139862e-01
6.77873731e-01 7.07139850e-01 5.46216667e-01 -7.99656093e-01
1.47436291e-01 -2.30691135e-01 -2.12906405e-01 -4.26651686e-01
3.85862261e-01 -1.63977295e-01 -2.80550092e-01 5.52536249e-01
-3.32524121e-01 6.68419078e-02 -4.04759824e-01 2.86933511e-01
1.09425497e+00 -1.76216707e-01 4.26949933e-02 8.84156227e-02
-2.04401180e-01 -2.96642363e-01 5.25624573e-01 8.18041801e-01
-8.61766115e-02 1.02027047e+00 6.52982891e-01 -1.88922033e-01
-1.13280404e+00 -9.41145122e-01 -9.28451791e-02 7.71425292e-02
-4.17771712e-02 -5.22889137e-01 -1.12507272e+00 -8.72459531e-01
-3.87868960e-03 8.62379253e-01 -7.39829302e-01 -2.45316118e-01
-1.66014388e-01 -1.00848889e+00 9.76899385e-01 6.08405411e-01
6.55065775e-01 -1.01282263e+00 -5.76312304e-01 2.09280282e-01
-3.66015792e-01 -1.06731737e+00 -5.49039125e-01 -4.00436044e-01
-8.25321138e-01 -7.30511904e-01 -1.01839948e+00 -9.21333507e-02
7.85341024e-01 -1.98040202e-01 1.11653590e+00 -2.00639576e-01
-1.59127265e-01 3.49777997e-01 -2.68632323e-01 -7.68068314e-01
-8.97029519e-01 -4.33386862e-02 -3.96334141e-01 -6.30148873e-03
-1.78192109e-02 -4.17125463e-01 -1.10758877e+00 2.83314325e-02
-1.34905076e+00 1.58388451e-01 5.99349976e-01 9.52964544e-01
7.15636492e-01 -1.35632649e-01 6.80366158e-01 -1.49243844e+00
1.07929718e+00 -7.23925948e-01 -2.81048924e-01 2.96124309e-01
-1.05136001e+00 -1.21018820e-01 6.80582702e-01 -2.14066386e-01
-1.08517420e+00 -1.17958263e-01 -3.84224474e-01 -3.74048114e-01
-1.49637759e-01 4.63665813e-01 -5.48150003e-01 2.11234495e-01
5.69664001e-01 1.47450477e-01 2.32683852e-01 -3.44797671e-01
3.73545498e-01 6.96770191e-01 2.70273238e-01 -2.11506262e-01
5.16961217e-01 4.72534269e-01 -2.22962990e-01 -1.53868794e-01
-3.55124831e-01 2.43526354e-01 6.26870915e-02 2.01371506e-01
9.78383064e-01 -1.04727662e+00 -6.78422630e-01 7.11284816e-01
-8.97888541e-01 -9.89415124e-02 -6.69128895e-01 7.85661638e-02
-4.26055938e-01 2.30941340e-01 -9.43849742e-01 -9.74589229e-01
-9.30168629e-01 -1.02973795e+00 1.12757170e+00 1.67495832e-01
-2.59896010e-01 -7.62760162e-01 -3.43030388e-03 5.16728401e-01
8.03665400e-01 9.60977435e-01 9.06096220e-01 -5.83906889e-01
-7.05780268e-01 -4.34277087e-01 -6.61900118e-02 3.05781811e-01
1.88869625e-01 -4.19072092e-01 -1.12148416e+00 -5.44310927e-01
9.52002034e-02 -2.72295743e-01 5.16617775e-01 3.38404983e-01
1.41816521e+00 -9.78734136e-01 -2.98547089e-01 9.31950867e-01
1.32763183e+00 3.52244020e-01 1.33991635e+00 1.55830385e-05
6.05272830e-01 3.21054548e-01 3.59420031e-01 6.94537342e-01
6.12000883e-01 2.54272163e-01 4.93751436e-01 -6.16744995e-01
1.27815008e-01 -6.62571132e-01 4.80676256e-03 3.14430594e-01
2.43348196e-01 -6.89043641e-01 -6.92330599e-01 5.37152767e-01
-1.79612303e+00 -7.82601297e-01 9.73838493e-02 2.25725031e+00
9.58205044e-01 -1.62911206e-01 2.44151190e-01 -1.60282508e-01
5.15571713e-01 4.76129316e-02 -7.65035152e-01 -1.78866670e-01
-3.09816658e-01 1.52315557e-01 5.61923206e-01 -2.19119236e-01
-8.81390452e-01 4.56748843e-01 6.91446590e+00 2.86653996e-01
-9.58814681e-01 2.29118392e-01 1.50559008e+00 -1.48615941e-01
-9.34406102e-01 -4.53215241e-01 -3.48518193e-01 8.41948152e-01
1.04032266e+00 -2.93392211e-01 1.62926897e-01 7.86928535e-01
2.57121120e-02 4.32880431e-01 -1.15648961e+00 8.96987677e-01
-7.45714381e-02 -1.36991107e+00 2.50445336e-01 5.50530672e-01
9.39493775e-01 -2.95912027e-01 5.35551190e-01 1.18738458e-01
5.37514865e-01 -1.42840111e+00 4.47975278e-01 5.55993378e-01
1.18050647e+00 -7.37652004e-01 1.09531176e+00 7.15480149e-02
-3.62751424e-01 2.63119727e-01 -1.18522150e-02 6.31216347e-01
4.18851942e-01 5.34495771e-01 -8.56633365e-01 7.24790215e-01
7.70151436e-01 2.25800753e-01 -5.10092437e-01 9.00172830e-01
8.07436407e-02 5.17286897e-01 -8.88626352e-02 3.39482516e-01
-1.08520150e-01 -1.74718082e-01 2.74547100e-01 1.04883993e+00
6.07837915e-01 2.22463012e-01 -2.21662581e-01 1.26810491e+00
-5.09129167e-01 2.74110381e-02 -8.63268435e-01 -2.79338479e-01
3.89655977e-01 9.45112467e-01 -1.31741226e-01 -1.95693403e-01
-2.54413307e-01 1.09177005e+00 -1.84742555e-01 2.32778847e-01
-9.20703650e-01 -3.83006223e-02 7.29628980e-01 3.83961797e-01
-9.22863260e-02 3.51378113e-01 -5.91187418e-01 -9.82010841e-01
2.76732624e-01 -1.25347793e+00 6.34393334e-01 -6.89584196e-01
-1.44516003e+00 8.21579754e-01 -2.37541571e-01 -1.29018044e+00
-4.26775992e-01 1.05635770e-01 -2.63348877e-01 1.02555335e+00
-1.22908366e+00 -1.52957976e+00 -4.73340094e-01 6.68804884e-01
-9.92981493e-02 -4.75840420e-02 1.23508012e+00 2.77625442e-01
-4.17978764e-01 1.26425076e+00 1.47860661e-01 2.89343208e-01
5.08791745e-01 -1.10296512e+00 9.03263032e-01 6.83916867e-01
-2.39233032e-01 5.75327158e-01 4.93065089e-01 -7.48956203e-01
-1.54829502e+00 -1.38818145e+00 4.33144659e-01 -5.06201863e-01
-1.20117940e-01 -5.34912646e-01 -9.71489787e-01 7.57379174e-01
5.19933820e-01 -1.12684958e-01 1.13525355e+00 -3.72878850e-01
-1.84558854e-01 8.73966590e-02 -1.92491138e+00 6.82850301e-01
9.38975632e-01 -3.81642282e-01 1.41069531e-01 1.92830771e-01
1.00020909e+00 -6.22781456e-01 -1.40148616e+00 5.21435916e-01
4.59473789e-01 -1.00123417e+00 8.42939556e-01 -5.91609597e-01
6.94943488e-01 1.17155127e-01 1.02880597e-01 -1.38504255e+00
-1.91932335e-01 -1.07371986e+00 -1.66955113e-01 1.62332797e+00
5.04176795e-01 -8.50099564e-01 9.64297473e-01 1.34819436e+00
2.95655042e-01 -6.66121423e-01 -7.20699728e-01 -5.14914870e-01
-1.69446263e-02 -8.85200426e-02 1.31441772e+00 1.23611128e+00
-2.53968567e-01 2.80454308e-02 -7.82263994e-01 -9.23713483e-03
6.04590893e-01 -2.64956981e-01 9.83252227e-01 -7.71492422e-01
-2.52068758e-01 -9.82985571e-02 -3.64790916e-01 -5.54201245e-01
-1.93400621e-01 -5.62654972e-01 -3.84036243e-01 -1.66386557e+00
3.34680378e-01 -5.82654655e-01 -2.30881691e-01 5.74838281e-01
-4.43929493e-01 1.13632835e-01 3.03484589e-01 7.72598684e-02
-1.87679797e-01 6.76809251e-01 1.51550329e+00 -2.53912538e-01
-7.86298588e-02 5.40839285e-02 -1.29145610e+00 1.33770227e-01
6.97095752e-01 -5.26065350e-01 -4.31768060e-01 -4.76890445e-01
9.70711708e-02 5.81508875e-01 2.77164310e-01 -7.17080057e-01
-6.31293282e-02 5.96972648e-03 4.13471520e-01 -3.16823602e-01
2.61988193e-01 -9.68290031e-01 9.59125996e-01 4.67907459e-01
-4.75275099e-01 1.61745518e-01 3.35830033e-01 6.58505321e-01
-2.03197613e-01 3.56616408e-01 6.46028519e-01 -1.55620530e-01
1.24122530e-01 5.85984707e-01 1.06681492e-02 2.43497178e-01
1.01554918e+00 -8.24961290e-02 -5.07260799e-01 -9.23883498e-01
-5.65069199e-01 2.81456769e-01 6.42206490e-01 4.15566444e-01
7.29261220e-01 -1.49849117e+00 -1.04124534e+00 3.63797396e-01
3.14889163e-01 2.85330236e-01 5.52223206e-01 5.27906418e-01
-3.79086077e-01 9.56363454e-02 -2.48738527e-01 -3.48193109e-01
-9.86622155e-01 7.79267251e-01 1.60148337e-01 -4.65080738e-01
-7.45239317e-01 5.02275050e-01 4.64258164e-01 -2.97414869e-01
1.58973515e-01 -1.55010685e-01 3.39213818e-01 -3.41312438e-01
4.68977779e-01 1.36364028e-01 -1.52204363e-02 -2.71307498e-01
-1.53815284e-01 -2.01416805e-01 -1.22701026e-01 -1.49529511e-02
1.29131711e+00 -1.02039333e-03 1.90145206e-02 -1.83570907e-01
1.00943887e+00 -2.24371124e-02 -1.30581701e+00 7.12672621e-03
-4.44715410e-01 -6.82994843e-01 -3.61720711e-01 -1.22837424e+00
-1.42152190e+00 5.00987351e-01 8.60763609e-01 4.07108605e-01
1.28342378e+00 -1.22525752e-01 1.27145350e+00 -3.17247957e-01
3.95290703e-01 -4.81856257e-01 -2.55024016e-01 -2.27924570e-01
8.20423782e-01 -1.36935437e+00 -3.24414730e-01 -4.07776475e-01
-1.18253350e+00 2.06668958e-01 4.55430210e-01 2.38193288e-01
5.63536108e-01 4.84521300e-01 2.96949238e-01 -6.60991371e-02
-7.19823539e-01 3.88935387e-01 5.56259006e-02 8.65877271e-01
4.34328169e-01 3.54219854e-01 -2.38098800e-01 8.67854178e-01
-3.57632220e-01 1.08996995e-01 4.23641622e-01 8.79412472e-01
8.06244254e-01 -1.29273164e+00 -2.26770014e-01 5.33317804e-01
-1.03868461e+00 -3.63496035e-01 -5.26582897e-01 6.17510140e-01
-1.28198490e-01 8.19392323e-01 -1.81313649e-01 -3.41002822e-01
5.26974559e-01 -7.64409155e-02 1.30587757e-01 -2.25071937e-01
-1.15562904e+00 -4.50993329e-02 8.22141692e-02 -4.33648407e-01
-7.59915933e-02 -6.24397337e-01 -7.29861021e-01 -5.42788863e-01
-1.12791128e-01 2.44795494e-02 6.27640128e-01 2.90292442e-01
1.03293777e+00 6.54844105e-01 5.29519975e-01 8.54024589e-02
-6.58264995e-01 -8.87925684e-01 -3.45437467e-01 1.04069746e+00
2.46433035e-01 1.73410580e-01 5.70734888e-02 -1.22838132e-02] | [6.245790481567383, 6.7914252281188965] |
3e3ba2a5-45a2-41d4-a50f-6e90b5bda38a | generative-adversarial-network-driven | 2202.07802 | null | https://arxiv.org/abs/2202.07802v1 | https://arxiv.org/pdf/2202.07802v1.pdf | Generative Adversarial Network-Driven Detection of Adversarial Tasks in Mobile Crowdsensing | Mobile Crowdsensing systems are vulnerable to various attacks as they build on non-dedicated and ubiquitous properties. Machine learning (ML)-based approaches are widely investigated to build attack detection systems and ensure MCS systems security. However, adversaries that aim to clog the sensing front-end and MCS back-end leverage intelligent techniques, which are challenging for MCS platform and service providers to develop appropriate detection frameworks against these attacks. Generative Adversarial Networks (GANs) have been applied to generate synthetic samples, that are extremely similar to the real ones, deceiving classifiers such that the synthetic samples are indistinguishable from the originals. Previous works suggest that GAN-based attacks exhibit more crucial devastation than empirically designed attack samples, and result in low detection rate at the MCS platform. With this in mind, this paper aims to detect intelligently designed illegitimate sensing service requests by integrating a GAN-based model. To this end, we propose a two-level cascading classifier that combines the GAN discriminator with a binary classifier to prevent adversarial fake tasks. Through simulations, we compare our results to a single-level binary classifier, and the numeric results show that proposed approach raises Adversarial Attack Detection Rate (AADR), from $0\%$ to $97.5\%$ by KNN/NB, from $45.9\%$ to $100\%$ by Decision Tree. Meanwhile, with two-levels classifiers, Original Attack Detection Rate (OADR) improves for the three binary classifiers, with comparison, such as NB from $26.1\%$ to $61.5\%$. | ['Burak Kantarci', 'Zhiyan Chen'] | 2022-02-16 | null | null | null | null | ['adversarial-attack-detection', 'adversarial-attack-detection'] | ['computer-vision', 'knowledge-base'] | [ 5.19737005e-01 1.87358916e-01 8.00553858e-02 1.65024787e-01
-8.03849816e-01 -1.00108361e+00 5.64041615e-01 -1.54236063e-01
-1.69828951e-01 9.19470131e-01 -2.40291849e-01 -5.79875350e-01
3.83673668e-01 -1.12566340e+00 -9.24574077e-01 -6.58083081e-01
-5.92946038e-02 -9.64521989e-02 2.92179942e-01 -4.85763013e-01
-7.92216435e-02 2.90323466e-01 -1.28459561e+00 3.45372170e-01
9.56258833e-01 1.30900419e+00 -8.38113308e-01 9.78219151e-01
2.27027133e-01 7.97926962e-01 -1.24442959e+00 -1.03085530e+00
5.30043066e-01 -4.33498174e-01 -7.39931986e-02 -5.77069700e-01
-1.98653005e-02 -4.04311866e-01 -2.09122479e-01 1.37868500e+00
6.64465606e-01 -4.55160588e-01 5.93238533e-01 -1.93006504e+00
-6.98829234e-01 5.88102460e-01 -5.10241091e-01 -8.34144503e-02
5.63535094e-01 5.67473531e-01 3.97854596e-01 -1.53805390e-01
-3.50499675e-02 1.09126091e+00 9.23496425e-01 8.88388932e-01
-9.11286116e-01 -1.26256955e+00 -1.32600158e-01 -3.00061971e-01
-1.44284737e+00 -2.84683764e-01 8.70891094e-01 -2.96617419e-01
3.99800360e-01 7.10763991e-01 3.79974723e-01 1.57180095e+00
4.21885610e-01 3.26150894e-01 1.45177889e+00 -2.16347769e-01
5.24265289e-01 4.84365553e-01 -2.36337930e-01 4.55812782e-01
6.48285985e-01 3.03706169e-01 -5.95474169e-02 -8.14224601e-01
3.48455280e-01 1.83482990e-01 -1.11935712e-01 5.12133181e-01
-3.10786366e-01 8.84257257e-01 4.23232943e-01 1.31151944e-01
-4.29979414e-01 2.79807061e-01 3.10634822e-01 1.15972027e-01
6.31143004e-02 1.44980043e-01 -1.77719384e-01 -1.69397891e-01
-5.33141077e-01 2.29599163e-01 8.14809084e-01 9.84335899e-01
3.72653514e-01 3.53423238e-01 -1.93584830e-01 2.00305507e-01
1.98067054e-01 1.10080659e+00 4.71140802e-01 -5.86280763e-01
5.52344978e-01 6.11547887e-01 3.84360075e-01 -1.42053974e+00
-3.67613770e-02 -3.10248405e-01 -9.80741858e-01 2.45198295e-01
1.27005830e-01 -6.35709167e-01 -6.01705670e-01 1.75398827e+00
3.21203977e-01 3.48455459e-01 3.44678253e-01 5.19658327e-01
3.77846807e-01 3.92236680e-01 2.10115328e-01 -1.72342986e-01
1.32485235e+00 -4.05700594e-01 -5.82106471e-01 -1.46198481e-01
4.16807532e-01 -5.90619504e-01 1.37525821e+00 4.06799406e-01
-9.32462215e-01 -4.61964309e-01 -1.45460355e+00 7.73198843e-01
-5.01698136e-01 -3.56289536e-01 3.50330591e-01 1.85715306e+00
-6.87550783e-01 1.56900883e-02 -4.96812016e-01 1.03427529e-01
7.87386239e-01 3.63178581e-01 2.20425308e-01 1.68383941e-02
-1.47094011e+00 3.86325300e-01 -1.68743446e-01 -1.06891960e-01
-1.16965795e+00 -3.58222067e-01 -4.22623366e-01 -3.03444892e-01
2.93031447e-02 -4.49070722e-01 9.76097584e-01 -1.02859628e+00
-1.43353105e+00 5.07901311e-01 1.71859011e-01 -8.48399818e-01
8.24335873e-01 9.56804007e-02 -9.99234617e-01 -2.38937084e-02
-4.25048880e-02 -1.68392938e-02 1.09065187e+00 -1.54627013e+00
-5.52439272e-01 -3.56683671e-01 4.95295346e-01 -3.00572157e-01
-4.96747643e-01 5.89751415e-02 5.45066953e-01 -7.17768848e-01
-3.02868754e-01 -1.13210261e+00 -2.27156907e-01 -3.70512456e-01
-6.63069010e-01 4.80504930e-01 1.07782674e+00 -5.02837300e-01
1.13105452e+00 -1.97860515e+00 -7.31986582e-01 3.32862347e-01
3.43535364e-01 7.19927013e-01 2.92521805e-01 3.12309772e-01
3.80307674e-01 5.72311699e-01 -4.12099063e-01 -1.55222058e-01
8.34877640e-02 8.35349485e-02 -5.05790055e-01 4.29855257e-01
1.87068552e-01 9.58480179e-01 -7.09297299e-01 3.59586626e-02
-8.00264031e-02 3.06978703e-01 -5.02350926e-01 3.28068495e-01
-7.21967891e-02 3.82458955e-01 -6.18561327e-01 9.54900324e-01
1.07438350e+00 1.15582839e-01 5.03584705e-02 1.51792541e-01
4.68142122e-01 -1.34331405e-01 -1.13726664e+00 5.91965139e-01
-4.21408832e-01 1.37317821e-01 2.02642113e-01 -7.94556379e-01
1.02419293e+00 3.19729477e-01 1.13392457e-01 -7.02655435e-01
5.21382272e-01 2.17584342e-01 -1.52549427e-02 -5.04619360e-01
1.00739293e-01 -1.69734225e-01 -6.29020214e-01 5.14431059e-01
-6.82716906e-01 -1.40861119e-03 -7.72688091e-01 8.50665942e-02
1.60779560e+00 -4.24996644e-01 1.48530930e-01 1.45017609e-01
7.76679099e-01 -1.69509321e-01 3.69308025e-01 1.03275692e+00
-4.65833187e-01 1.43186212e-01 4.10736918e-01 -2.19157144e-01
-6.82422161e-01 -1.11809194e+00 1.61369160e-01 7.71762252e-01
3.85093540e-01 4.95971851e-02 -1.23536074e+00 -9.00039554e-01
6.89985082e-02 5.87329865e-01 -5.10691464e-01 -6.43210173e-01
-4.50440228e-01 -7.97337711e-01 1.69181418e+00 3.62898648e-01
1.03028846e+00 -8.77970994e-01 -4.25874919e-01 8.33566710e-02
-5.44190444e-02 -1.30045033e+00 4.32776241e-03 -2.94904649e-01
-2.30338410e-01 -1.09439874e+00 -3.40573788e-01 -2.39166811e-01
5.38507044e-01 1.23507269e-01 6.04859352e-01 2.80878067e-01
-1.39113948e-01 1.44477233e-01 -5.83897173e-01 -8.58333170e-01
-9.34409797e-01 -1.75476730e-01 3.38839829e-01 4.85995412e-01
3.95757288e-01 -7.41282463e-01 -7.11326122e-01 5.49513042e-01
-1.07903481e+00 -6.89812958e-01 4.62039918e-01 4.94336516e-01
1.53423041e-01 1.83016181e-01 9.83850360e-01 -9.25259352e-01
9.17866349e-01 -7.95918345e-01 -4.82611060e-01 -2.55701393e-02
-6.07255459e-01 -4.09519941e-01 1.24073863e+00 -9.67858315e-01
-6.09019935e-01 -3.10140550e-01 -3.62678379e-01 -3.57871741e-01
-2.95505852e-01 -1.91288859e-01 -6.37051344e-01 -5.27385771e-01
1.16548383e+00 3.92263353e-01 -2.21770048e-01 1.65583715e-01
3.63733292e-01 1.17729223e+00 3.50007892e-01 -3.55191618e-01
1.27392673e+00 6.98886871e-01 -1.05012678e-01 -4.88754094e-01
-3.43964636e-01 3.62546593e-01 3.17575932e-01 -1.29744992e-01
5.15250921e-01 -7.78578877e-01 -1.12714207e+00 7.10164249e-01
-9.04404044e-01 -2.00115412e-01 -8.32865685e-02 -7.72998184e-02
-2.31046796e-01 3.21435362e-01 -5.51349580e-01 -1.37535179e+00
-5.87645590e-01 -1.25015807e+00 1.08241439e+00 2.73521841e-01
-1.06465526e-01 -5.67342877e-01 -2.90129006e-01 6.76324546e-01
7.76603281e-01 9.37659323e-01 5.65546989e-01 -8.37213397e-01
-3.41980159e-01 -7.89998412e-01 -1.25838444e-02 5.29792309e-01
1.68857411e-01 -2.18936890e-01 -1.28845537e+00 -2.66906828e-01
6.09702647e-01 -7.39112794e-02 8.73934925e-02 -6.73432350e-02
9.82888937e-01 -1.01411748e+00 -2.93397218e-01 4.22630250e-01
1.28831530e+00 6.06834710e-01 1.02584624e+00 3.27886268e-02
5.61716676e-01 1.62155062e-01 4.42524105e-01 5.26836395e-01
2.26599276e-01 3.84157836e-01 8.28396022e-01 7.62295052e-02
2.57791877e-01 -4.97084528e-01 6.35775089e-01 1.35937735e-01
2.05353856e-01 -4.65459228e-01 -5.74063718e-01 -1.86930895e-02
-1.35953188e+00 -9.13794219e-01 -2.75867045e-01 2.17478466e+00
6.47706509e-01 7.40139604e-01 4.97264385e-01 7.08889246e-01
8.97444487e-01 6.65178848e-03 -5.59375763e-01 -4.96793509e-01
-1.87615231e-01 4.08813983e-01 8.48972797e-01 4.33615685e-01
-9.32726085e-01 9.18700993e-01 5.28243113e+00 9.56131458e-01
-1.26309288e+00 5.90332031e-01 6.96400583e-01 5.42825498e-02
-3.80016476e-01 -1.21929280e-01 -6.30965471e-01 1.11721444e+00
1.11447012e+00 6.03410527e-02 4.95393306e-01 8.74778450e-01
2.16145024e-01 1.43497393e-01 -5.00202894e-01 9.79030967e-01
1.66664988e-01 -1.09313047e+00 -1.00754194e-01 2.01056540e-01
7.12893188e-01 -3.63733470e-01 3.17363948e-01 3.92467707e-01
6.06867850e-01 -1.13074899e+00 7.46320546e-01 2.63575077e-01
8.70737612e-01 -8.98768067e-01 1.01821220e+00 6.28585339e-01
-1.06638110e+00 -4.31354195e-01 -1.04022399e-01 -4.01361883e-01
8.71635787e-03 5.14084399e-01 -7.46757746e-01 3.12621802e-01
5.46562076e-01 -3.43479306e-01 -5.93908727e-01 3.10926914e-01
-2.63387382e-01 8.65772188e-01 -2.60186553e-01 -5.98132372e-01
1.10479549e-01 2.05220267e-01 5.47942817e-01 8.90819967e-01
1.59169525e-01 2.13143855e-01 1.14722483e-01 8.95450711e-01
-2.38411635e-01 -1.16256386e-01 -8.90590191e-01 1.70531675e-01
8.52484047e-01 1.04570711e+00 -3.30208153e-01 -1.90508902e-01
9.64076892e-02 1.05487180e+00 -2.72818983e-01 2.51653641e-01
-1.51996696e+00 -4.97762978e-01 6.12908721e-01 2.87604451e-01
-1.22517250e-01 1.00415155e-01 -5.06551623e-01 -7.31768072e-01
3.43530685e-01 -1.25915980e+00 1.08490057e-01 -4.97359782e-01
-1.30621481e+00 8.02052200e-01 -4.32058632e-01 -1.33624375e+00
-1.08154684e-01 -2.14926302e-01 -6.93495333e-01 7.72704005e-01
-1.06946528e+00 -1.41766763e+00 -4.76520389e-01 8.42335403e-01
-2.46606078e-02 -2.53648996e-01 7.64761090e-01 3.55325848e-01
-5.00132322e-01 1.47506368e+00 -1.63823113e-01 3.20749193e-01
1.02923296e-01 -6.50933146e-01 4.59144324e-01 1.05294669e+00
-1.92623317e-01 5.59967697e-01 6.60195589e-01 -7.42787182e-01
-1.44135249e+00 -1.29663634e+00 1.28601894e-01 -6.52089894e-01
4.59934801e-01 -7.05591321e-01 -5.29192448e-01 3.63789409e-01
-1.24942452e-01 1.96508259e-01 6.31696880e-01 -8.25235248e-01
-5.51918149e-01 -2.96247333e-01 -2.15695643e+00 8.75754535e-01
9.17546809e-01 -6.15863442e-01 2.80124079e-02 2.33103544e-01
1.05404019e+00 -1.17179126e-01 -7.20413506e-01 3.14299315e-01
6.79063737e-01 -1.04224825e+00 7.58073390e-01 -5.02832353e-01
-7.28749111e-02 -3.16651016e-01 -4.90437955e-01 -7.35446513e-01
2.40025431e-01 -1.07874382e+00 -3.65889728e-01 1.43805611e+00
4.68053579e-01 -1.27898324e+00 9.12335038e-01 4.41167355e-01
1.96692601e-01 -6.26957536e-01 -1.03514421e+00 -7.96094060e-01
1.22173585e-01 -7.11509943e-01 1.00802791e+00 9.04156506e-01
-2.80889302e-01 7.32487291e-02 -3.69725704e-01 6.08292818e-01
5.85076213e-01 -6.68976843e-01 1.05542505e+00 -8.27805758e-01
-4.28640455e-01 -3.88154089e-01 -5.18078983e-01 -4.52230036e-01
-1.88919753e-01 -3.64693075e-01 -1.82944432e-01 -6.35358632e-01
-4.24391061e-01 -8.15363050e-01 -1.36955634e-01 2.69127518e-01
-2.13050321e-01 6.79165304e-01 1.67960729e-02 3.20697799e-02
-1.13235041e-01 1.81832999e-01 6.22735798e-01 -3.69824231e-01
-1.97381616e-01 4.48917687e-01 -1.04779303e+00 7.23435104e-01
1.05596721e+00 -7.26411164e-01 -3.87664884e-01 1.52081594e-01
2.72617280e-01 1.34561595e-03 5.84273934e-01 -1.29806125e+00
-9.71455351e-02 -2.15215787e-01 1.21380039e-01 8.50883722e-02
3.24575454e-01 -9.29689348e-01 3.95737916e-01 8.67996514e-01
-3.12116779e-02 7.13791326e-02 1.12382889e-01 6.97396457e-01
3.57206434e-01 -2.72325724e-02 7.69941151e-01 -6.23487718e-02
2.39855424e-01 1.05186671e-01 -2.82061189e-01 1.02157868e-01
1.35053170e+00 -3.98243785e-01 -6.71598315e-01 -6.98822439e-01
-3.94559950e-01 -2.35350266e-01 5.01410127e-01 1.86392576e-01
4.51929867e-01 -1.31755662e+00 -4.93004113e-01 2.71307915e-01
2.92455368e-02 -3.36740136e-01 1.17642179e-01 5.80533147e-01
-4.97437358e-01 -1.13537967e-01 -5.20682596e-02 -3.19838047e-01
-1.07709193e+00 6.78348839e-01 4.28919256e-01 -1.73648059e-01
-3.73372324e-02 7.79441237e-01 -3.28231573e-01 -2.72901297e-01
1.70868725e-01 -1.58833802e-01 1.70281351e-01 -3.82051110e-01
6.67325437e-01 6.45604849e-01 1.13575578e-01 -6.30735874e-01
-4.44006592e-01 2.12456614e-01 4.35906380e-01 -2.90998369e-01
5.06524146e-01 6.44284785e-02 1.14517979e-01 -9.25134644e-02
1.04391932e+00 5.57405829e-01 -8.55473518e-01 1.28868595e-01
-3.31193805e-01 -3.74612689e-01 -6.00369453e-01 -8.23855639e-01
-9.26723361e-01 6.31368518e-01 9.37724590e-01 9.98786509e-01
1.08668613e+00 -3.09735060e-01 1.19917738e+00 -1.32410958e-01
7.43382871e-01 -5.66867411e-01 1.81583613e-01 -9.38126147e-02
4.92235094e-01 -1.03798378e+00 -4.54041272e-01 -4.98031199e-01
-5.02032816e-01 3.34724367e-01 6.53558612e-01 -4.44195330e-01
5.21354795e-01 5.39885640e-01 1.16487779e-01 2.12509651e-02
-2.04704970e-01 1.45958319e-01 -2.81713933e-01 1.02904475e+00
-3.70539486e-01 4.27681208e-01 -1.91570953e-01 1.33045006e+00
-5.18725574e-01 -1.48637205e-01 6.95597410e-01 1.01469028e+00
-2.56554574e-01 -1.06969023e+00 -7.87088156e-01 4.74774629e-01
-8.46172333e-01 -3.90655128e-03 -6.15034699e-01 4.89088327e-01
7.48209357e-01 1.61645067e+00 -3.01755250e-01 -1.13640285e+00
3.94530892e-01 -1.12634614e-01 -1.14201810e-02 -1.91085219e-01
-1.03566384e+00 -3.25645000e-01 8.27357322e-02 -3.76465738e-01
-2.61160284e-02 -2.08328158e-01 -1.03090024e+00 -6.19789183e-01
-3.29522878e-01 7.89695606e-02 5.83251417e-01 9.47950602e-01
4.85513091e-01 3.20284635e-01 1.14228618e+00 -4.88625079e-01
-7.79614866e-01 -8.53888571e-01 -3.61420751e-01 4.32864517e-01
2.44478807e-01 -5.11267185e-01 -7.19860256e-01 -2.79982507e-01] | [13.62140941619873, 5.757315635681152] |
fd1da713-6f12-404e-bb26-5aa1a57bbc1a | sleep-quality-chronotype-and-social-jet-lag | 2103.10795 | null | https://arxiv.org/abs/2103.10795v1 | https://arxiv.org/pdf/2103.10795v1.pdf | Sleep quality, chronotype and social jet lag of adolescents from a population with a very late chronotype | Sleep disorders can be a negative factor both for learning as for the mental and physical development of adolescents. It has been shown that, in many populations, adolescents tend to have a poor sleep quality, and a very late chronotype. Furthermore, these features peak at adolescence, in the sense that adults tend to sleep better and have an earlier chronotype. But what happens when we consider adolescents in a population where already adults have poor sleep quality and a very late chronotype? We have conducted two non-clinical studies in the city of Bariloche, Argentina aimed at measuring sleep quality, chronotype, and social jet lag, using the Pittsburgh and Munich questionnaires. These were administered individually to groups of high school students, as well as to smaller samples of adults and preadolescents, in order to study differences between adolescents and these groups. The results show that in this population sleep quality is much poorer than in most other healthy populations recorded elsewhere. Furthermore, sleep quality is consistently worse for adolescents than for the other groups. The difference with adults seems to be due mainly to increased daytime sleepiness and sleep latency, whereas the difference with preadolescents seems to be due mainly to shorter sleep duration. We also found that the chronotypes of all the groups are very late, with a peak at an age between 18 and 24 ys. Social jet lag and sleep onset latency are also large, and they peak at adolescence, which suggests that they might be closely related to the large prevalence of poor sleep quality that we find in adolescents. | ['Sebastián Risau-Gusman', 'Fernanda R. Román', 'Sabrina C. Riva', 'Mara López-Wortzman', 'Sergio Lindenbaum', 'Pablo M. Gleiser', 'D. Lorena Franco', 'Damián Dellavale', 'Romina A. Capellino', 'Andrés H. Calderón'] | 2021-03-19 | null | null | null | null | ['sleep-quality-prediction'] | ['medical'] | [-5.58476031e-01 8.29493031e-02 -3.91074389e-01 -1.18296526e-01
1.96996368e-02 -1.47378162e-01 2.20394775e-01 6.73218131e-01
-8.54223251e-01 7.87088096e-01 2.77035952e-01 -4.38220054e-02
-2.57306784e-01 -7.35231996e-01 -8.21609721e-02 -5.73870778e-01
1.13120921e-01 6.15509033e-01 3.39214593e-01 -2.30136618e-01
8.53573456e-02 9.95992720e-02 -1.70461357e+00 -7.73456395e-01
1.25918126e+00 1.00718930e-01 3.06433201e-01 3.13479632e-01
-7.22190663e-02 -1.87911749e-01 -4.87325788e-01 -4.00947660e-01
-3.72227788e-01 -5.13917804e-01 -3.00389588e-01 -6.95223883e-02
3.80014598e-01 2.63638288e-01 8.13633129e-02 8.80273998e-01
5.50875902e-01 5.76716661e-01 4.65998203e-01 -1.20326531e+00
-1.84038535e-01 3.48037630e-01 -4.35002327e-01 7.20753729e-01
4.22523707e-01 5.44582717e-02 6.29050195e-01 -4.68064249e-01
2.58148551e-01 1.00755715e+00 3.95235181e-01 8.76232564e-01
-1.34268129e+00 -1.28047478e+00 -2.63512909e-01 4.15468365e-01
-1.36860955e+00 -7.73773313e-01 3.27123433e-01 -3.76547635e-01
6.81788743e-01 -3.41145769e-02 1.70740223e+00 7.19288647e-01
5.16047359e-01 -1.91441104e-02 1.04944956e+00 -1.32836238e-01
5.34167707e-01 1.18347913e-01 8.99289846e-02 2.99296081e-01
8.72297883e-01 -1.36302695e-01 -4.26556975e-01 4.78437155e-01
1.04793623e-01 5.57571556e-03 -1.30161777e-01 1.69914261e-01
-7.43403494e-01 6.76198661e-01 -6.73586503e-02 8.63813102e-01
-3.85820754e-02 -2.61801649e-02 3.00228029e-01 2.64908433e-01
4.66651678e-01 1.67311341e-01 -2.30611376e-02 -8.62468660e-01
-1.33208728e+00 -3.87051031e-02 4.95871902e-01 1.22650124e-01
5.79893410e-01 -1.19736038e-01 5.90540826e-01 9.95551348e-01
4.58866864e-01 6.95459127e-01 7.77693212e-01 -9.74186301e-01
4.37323183e-01 5.49363196e-01 -2.16022715e-01 -9.46611285e-01
-8.05575311e-01 -3.78420055e-01 -4.41159874e-01 2.94295013e-01
7.12504029e-01 1.76054418e-01 -1.25827760e-01 1.99783206e+00
1.82213277e-01 -5.12928143e-02 6.09687194e-02 6.34623170e-01
7.35124528e-01 3.25087368e-01 1.78692386e-01 -5.69611549e-01
1.34605014e+00 -5.99641263e-01 -5.91951311e-01 -7.59938538e-01
2.01081187e-01 -7.70578980e-01 1.16459203e+00 3.87901306e-01
-1.47019517e+00 -6.27983272e-01 -1.01297784e+00 7.57325068e-02
-2.78664023e-01 -4.00275886e-01 3.29705387e-01 1.34141743e+00
-1.23679543e+00 9.01221931e-01 -1.22530043e+00 -1.04863942e+00
9.19982269e-02 4.02687371e-01 -4.40255791e-01 4.15733248e-01
-8.29613805e-01 1.01282930e+00 1.44061506e-01 -4.11641687e-01
-4.88694124e-02 -6.19708955e-01 -8.09235036e-01 3.04437518e-01
-9.14581411e-04 -5.11710525e-01 8.41676295e-01 -7.51887083e-01
-1.31039774e+00 9.03182566e-01 -3.44730675e-01 -6.40537590e-02
-3.25317949e-01 -5.78546785e-02 -8.46133947e-01 3.83474380e-01
4.80184883e-01 4.23504710e-01 4.80048388e-01 -3.91607106e-01
-3.79634708e-01 -8.03501487e-01 -3.69058132e-01 1.18153043e-01
-5.22912554e-02 2.33893946e-01 1.07156828e-01 -9.79907960e-02
4.10265326e-01 -8.98736358e-01 -2.04326406e-01 -2.62808502e-01
3.16207893e-02 -5.98808825e-01 8.71357247e-02 -2.46103913e-01
1.22633517e+00 -2.30088949e+00 -5.02485573e-01 1.49811015e-01
5.74332297e-01 -8.36111698e-03 1.67640343e-01 4.99824524e-01
-2.04028219e-01 3.11569333e-01 3.34034622e-01 -5.71596146e-01
1.82932273e-01 5.33938706e-01 4.81596947e-01 8.97684455e-01
4.96225990e-03 2.08510712e-01 -1.18369901e+00 -7.40329623e-01
1.69895828e-01 1.06921814e-01 -3.87025416e-01 -2.70425290e-01
4.31645989e-01 2.50129998e-01 -7.32851177e-02 1.07786871e-01
4.39441323e-01 1.65090695e-01 -1.51429608e-01 8.48587036e-01
-8.77912462e-01 7.29527235e-01 -7.12310910e-01 1.32538950e+00
-2.85416067e-01 1.02967060e+00 1.59955900e-02 -5.56103170e-01
6.94868088e-01 3.85856360e-01 1.92460462e-01 -1.14329040e+00
9.47596803e-02 3.70695144e-01 8.78231466e-01 -4.15291280e-01
7.34945714e-01 -6.99023664e-01 3.78721625e-01 6.02275372e-01
-2.16755256e-01 -2.07114011e-01 8.36097002e-01 7.80802891e-02
8.15605581e-01 -5.30797660e-01 2.50959188e-01 -4.59090233e-01
4.49477941e-01 -7.23771214e-01 9.02157664e-01 2.88551718e-01
-2.96718746e-01 2.00330734e-01 8.42794657e-01 3.22560191e-01
-7.29662240e-01 -1.40211093e+00 -3.20934802e-01 6.54405594e-01
5.21187596e-02 -6.96419477e-01 -4.81723934e-01 -3.28226797e-02
-3.38402182e-01 1.00028336e+00 -3.10850501e-01 -4.76129204e-01
-3.08618695e-01 -3.88657570e-01 1.09719314e-01 2.87996709e-01
3.99920464e-01 -8.14513326e-01 -4.70851362e-01 1.33889958e-01
1.83397964e-01 -7.85336256e-01 -4.37753946e-01 -1.99519411e-01
-1.06612074e+00 -9.79547501e-01 -6.50903404e-01 -4.35716808e-01
5.55812955e-01 1.51580691e-01 1.10543680e+00 4.82262462e-01
1.65004611e-01 3.26922417e-01 -2.57121623e-01 -4.16316241e-01
-3.18155140e-01 8.29272419e-02 4.51051354e-01 -6.12144768e-01
5.06605923e-01 -1.22287011e+00 -1.10918403e+00 3.67048472e-01
-4.37946230e-01 -2.66261071e-01 3.82128417e-01 -3.91344773e-03
4.57009971e-02 2.11269781e-01 4.21909869e-01 -2.64328808e-01
3.74700576e-02 -7.30025411e-01 -3.40951174e-01 -4.59295601e-01
-1.33767009e+00 -4.70498681e-01 4.40854162e-01 -2.86674678e-01
-5.22346020e-01 -8.45793605e-01 -5.11568189e-01 4.21890259e-01
-4.61339086e-01 -9.59085971e-02 5.11556789e-02 3.15193176e-01
1.25816494e-01 8.35113749e-02 2.66299725e-01 -6.16229773e-01
-4.59105939e-01 5.39675593e-01 1.56724125e-01 -3.41922522e-01
9.19054806e-01 2.20922247e-01 3.14780951e-01 -1.49075937e+00
-6.44499242e-01 -5.98098636e-01 -3.48870486e-01 -2.62351334e-01
1.18798971e+00 -6.67397261e-01 -9.75872993e-01 -5.41450717e-02
-1.55022368e-01 -6.08350575e-01 -3.71752203e-01 1.13057029e+00
-4.02042001e-01 8.79559368e-02 -5.17132543e-02 -7.34947443e-01
-2.65204579e-01 -1.05455089e+00 5.97802877e-01 8.35065365e-01
-6.28317177e-01 -1.50905979e+00 6.55067444e-01 2.94279903e-01
1.85899183e-01 2.86460146e-02 9.59109843e-01 -6.78683758e-01
-1.40015781e-01 3.20634067e-01 1.23109028e-01 1.44553274e-01
1.01957977e-01 8.19834322e-02 -3.19807261e-01 -3.11615139e-01
-4.40381914e-02 1.26510143e-01 3.98900032e-01 3.20595443e-01
1.28146380e-01 1.59718052e-01 -3.60404909e-01 1.06859230e-01
1.38094413e+00 6.80630982e-01 5.07178426e-01 6.17086053e-01
-4.44872826e-02 5.06123781e-01 4.79161143e-01 -4.49201316e-02
5.39844632e-01 6.13758206e-01 3.40476967e-02 1.22370258e-01
-3.48810494e-01 6.91414401e-02 4.98527378e-01 9.35832024e-01
-1.42296538e-01 1.33153662e-01 -8.70457232e-01 7.15291500e-01
-1.05919766e+00 -1.18634713e+00 -5.75708687e-01 2.41964912e+00
4.91261482e-01 3.95340949e-01 8.28971863e-01 4.15185124e-01
5.31037748e-01 1.46979153e-01 -3.67686637e-02 -8.82658839e-01
3.84214848e-01 8.48669469e-01 5.68756983e-02 1.80428252e-01
-7.60539696e-02 4.65045989e-01 6.59196377e+00 4.20146704e-01
-1.17303801e+00 2.19898269e-01 2.58329451e-01 -5.41737437e-01
-1.82320088e-01 3.45298767e-01 -8.77129018e-01 9.14557517e-01
1.70927250e+00 -5.39846241e-01 1.17298551e-01 3.70019257e-01
8.21917534e-01 -9.12611842e-01 -8.42046380e-01 7.08279788e-01
-1.26139149e-01 -5.39184690e-01 -1.13979983e+00 3.47645700e-01
3.01832974e-01 1.09149925e-02 4.90681157e-02 3.02531093e-01
-4.32845831e-01 -9.10996079e-01 6.00905001e-01 5.72489977e-01
6.83964789e-01 -7.65451431e-01 4.24504966e-01 1.84799075e-01
-1.03787148e+00 8.64048526e-02 -1.95501834e-01 -6.52576029e-01
-1.09302372e-01 6.80662751e-01 -7.94561386e-01 -7.96911344e-02
7.25243986e-01 6.50741339e-01 -7.67217577e-01 1.50393736e+00
-3.85831118e-01 1.00247490e+00 -6.33310199e-01 -5.77729702e-01
-8.92763287e-02 -6.76243842e-01 5.91593444e-01 5.32524705e-01
5.03935874e-01 -2.75923237e-02 -6.58910871e-01 8.41274500e-01
4.15849715e-01 2.58951396e-01 -2.62992829e-01 -2.94778496e-01
3.34163457e-01 1.30945754e+00 -1.23056448e+00 3.05326357e-02
-8.48485351e-01 4.22567308e-01 -1.19225167e-01 6.96457475e-02
-5.66227198e-01 -4.25905108e-01 8.79685760e-01 5.34361482e-01
-8.97515118e-02 -3.48269612e-01 -2.02676430e-01 -8.26791465e-01
-3.36633503e-01 -2.11256847e-01 2.52383817e-02 -4.49713767e-01
-5.84009469e-01 -3.64445820e-02 -1.45420907e-02 -8.05528820e-01
-1.41404003e-01 3.60549055e-02 -1.35680556e+00 5.59483647e-01
-9.00889516e-01 -1.47126630e-01 -1.86167452e-02 2.73015261e-01
3.95030439e-01 1.90622449e-01 3.94010931e-01 4.51033413e-01
-1.05204153e+00 4.38859224e-01 1.28195331e-01 -1.71987280e-01
5.74590981e-01 -1.63842010e+00 -1.31549239e-01 6.76396370e-01
-1.17887102e-01 9.21853781e-01 8.67145479e-01 -5.54704547e-01
-6.66117728e-01 -5.20583332e-01 1.31514239e+00 -1.29277125e-01
6.64892435e-01 -2.93398529e-01 -4.01000828e-01 3.08672518e-01
4.82163370e-01 -9.23069417e-01 1.33415520e+00 7.28788912e-01
5.69923878e-01 -4.58427578e-01 -8.85383487e-01 6.31224573e-01
9.43286479e-01 -1.01059951e-01 -9.02601659e-01 7.08272085e-02
3.33379716e-01 1.61255583e-01 -9.51503396e-01 -3.06088030e-01
7.30334520e-01 -1.54577827e+00 6.01357222e-01 3.02438110e-01
3.93663198e-01 7.86655471e-02 5.47396898e-01 -1.31245983e+00
-1.82200983e-01 -5.98843157e-01 6.39384508e-01 1.74688923e+00
2.65773237e-01 -7.87865937e-01 1.02965713e+00 7.06423998e-01
-3.33047122e-01 -5.77771187e-01 -1.13060403e+00 -8.42094898e-01
3.50739688e-01 -3.40688437e-01 5.95638573e-01 3.97931904e-01
9.68725458e-02 4.74937826e-01 6.75830305e-01 -2.67570943e-01
2.33671263e-01 1.44857824e-01 5.00987887e-01 -1.48583710e+00
9.90489274e-02 -6.73686147e-01 -6.65588200e-01 -1.72421858e-01
1.70358181e-01 -7.13348627e-01 -3.08004737e-01 -1.68149185e+00
9.05235186e-02 -5.15776217e-01 6.70707524e-02 1.67515367e-01
2.74158851e-03 3.51287484e-01 -1.75744705e-02 -3.28532964e-01
-3.13543826e-01 3.92439842e-01 1.31093037e+00 4.43090469e-01
-7.55598962e-01 6.20391607e-01 -7.39351869e-01 9.78792489e-01
9.28299129e-01 -5.98814607e-01 -7.20294237e-01 3.40079159e-01
6.84466302e-01 2.31005982e-01 -2.04617195e-02 -1.45067918e+00
2.09093690e-02 -1.90353364e-01 2.56602019e-01 -5.98328054e-01
5.90836763e-01 -7.03540325e-01 1.86660022e-01 7.43748248e-01
3.27657908e-01 2.15144962e-01 -2.85823531e-02 -4.79050130e-02
2.20667660e-01 -5.55273712e-01 8.12044144e-01 1.99924842e-01
-2.08882764e-01 -4.44727056e-02 -1.10909295e+00 3.97697598e-01
7.86189973e-01 -7.16719568e-01 -2.69224972e-01 -4.89942729e-01
-1.06937850e+00 2.35560924e-01 9.49602783e-01 7.68927634e-02
3.46692383e-01 -1.01754415e+00 3.46982293e-02 3.05560917e-01
-1.99760184e-01 -4.28647727e-01 2.13506132e-01 1.67968988e+00
-3.85324210e-01 3.78184170e-01 -3.25449526e-01 -2.05057025e-01
-1.29771388e+00 2.61819929e-01 5.38707227e-02 1.44010991e-01
-7.13333905e-01 6.16646528e-01 -1.06772892e-01 4.68472064e-01
2.84098275e-02 -3.72209072e-01 -4.95016724e-01 4.26833451e-01
3.55902612e-01 9.03131187e-01 -3.33250552e-01 -7.77494133e-01
-4.34918344e-01 8.26872587e-01 2.61121541e-01 -1.28108621e-01
1.15909719e+00 -4.46982145e-01 -4.29042369e-01 7.03356981e-01
1.04602742e+00 8.89806986e-01 -4.54273432e-01 5.92544973e-01
6.12425879e-02 -4.81707841e-01 -2.22980320e-01 -3.88066433e-02
-9.54575896e-01 7.38454461e-01 8.00698102e-01 5.20532966e-01
1.16633904e+00 3.57646376e-01 1.09687364e+00 -1.92848027e-01
2.75300723e-02 -1.17211008e+00 -6.62840828e-02 2.25859240e-01
3.26940715e-01 -5.68886697e-01 1.66697949e-02 -6.40472677e-03
5.29465415e-02 8.83953214e-01 6.62814081e-01 -2.34361872e-01
5.56122839e-01 -2.69238204e-01 -3.24948400e-01 -1.84885845e-01
-6.51297808e-01 -7.13099241e-01 2.13104963e-01 5.08063018e-01
5.57119966e-01 -1.10509872e-01 -1.10438335e+00 4.01146591e-01
-1.13756418e+00 -2.28032246e-01 9.02509689e-01 3.67790997e-01
-6.98985338e-01 -1.65549731e+00 -4.67212081e-01 5.33933222e-01
-7.62641311e-01 8.45068544e-02 -1.05516419e-01 1.14312565e+00
7.25202560e-01 1.29305589e+00 8.37463677e-01 -7.89388716e-02
2.40398630e-01 2.82974452e-01 4.63473767e-01 -1.06099260e+00
-2.88809776e-01 1.89214535e-02 3.05298984e-01 -5.08383095e-01
-4.04593289e-01 -1.12105441e+00 -1.39743495e+00 -5.72276652e-01
-7.22900555e-02 8.74943316e-01 6.64063871e-01 1.08277392e+00
-2.85150737e-01 3.14232767e-01 1.89541027e-01 -2.89113045e-01
4.86524612e-01 -7.92563379e-01 -1.28712082e+00 -1.26265556e-01
5.41153848e-01 -7.12238193e-01 -5.52639723e-01 -6.77328348e-01] | [13.50318431854248, 3.477952718734741] |
1a4e5072-d94c-4fc8-b3ec-968908cd2cdc | even-the-simplest-baseline-needs-careful-re | null | null | https://aclanthology.org/2022.naacl-main.145 | https://aclanthology.org/2022.naacl-main.145.pdf | Even the Simplest Baseline Needs Careful Re-investigation: A Case Study on XML-CNN | The power and the potential of deep learning models attract many researchers to design advanced and sophisticated architectures. Nevertheless, the progress is sometimes unreal due to various possible reasons. In this work, through an astonishing example we argue that more efforts should be paid to ensure the progress in developing a new deep learning method. For a highly influential multi-label text classification method XML-CNN, we show that the superior performance claimed in the original paper was mainly due to some unbelievable coincidences. We re-examine XML-CNN and make a re-implementation which reveals some contradictory findings to the claims in the original paper. Our study suggests suitable baselines for multi-label text classification tasks and confirms that the progress on a new architecture cannot be confidently justified without a cautious investigation. | ['Chih-Jen Lin', 'Hsuan-Tien Lin', 'Tsung-Han Yang', 'Jie-Jyun Liu', 'Si-An Chen'] | null | null | null | null | naacl-2022-7 | ['multi-label-text-classification', 'multi-label-text-classification'] | ['methodology', 'natural-language-processing'] | [ 1.65414304e-01 1.20179012e-01 -1.37917757e-01 -7.54043102e-01
-6.54479861e-01 -5.69791257e-01 7.40216911e-01 3.69506210e-01
-7.60017276e-01 6.45386875e-01 2.46555775e-01 -7.22531497e-01
-2.97882795e-01 -4.91384715e-01 -3.88684958e-01 -6.53362393e-01
4.45838273e-01 5.65732300e-01 -3.78489941e-02 -2.47771740e-01
4.38813657e-01 3.92208993e-01 -1.21982598e+00 4.44571465e-01
2.53097057e-01 6.94743812e-01 -9.91471857e-02 5.95457613e-01
-3.14449489e-01 1.14738631e+00 -6.33982539e-01 -1.02628684e+00
2.02493928e-02 -2.59826839e-01 -1.25123262e+00 -2.14051709e-01
6.22136593e-01 -3.13576370e-01 9.84616056e-02 9.80281115e-01
5.74062586e-01 -1.73216954e-01 6.29614651e-01 -1.23565018e+00
-6.53496265e-01 7.83110738e-01 -5.21741390e-01 1.74469620e-01
-1.54774502e-01 -2.52615273e-01 1.31313574e+00 -4.52247947e-01
3.96178931e-01 1.04491353e+00 1.21745384e+00 5.93144655e-01
-8.33656788e-01 -8.31073165e-01 1.96671680e-01 1.85889721e-01
-1.12957144e+00 -3.08428645e-01 6.38291061e-01 -3.62531483e-01
1.03449655e+00 3.61124545e-01 2.90490717e-01 1.55197632e+00
3.76954228e-01 8.65201712e-01 1.22945571e+00 -6.95975482e-01
-1.76309958e-01 2.69366711e-01 1.58544853e-01 7.86749363e-01
4.84035403e-01 -6.98214844e-02 -4.43683043e-02 -2.24681035e-01
4.01828140e-01 -3.02297384e-01 1.22186532e-02 5.15593179e-02
-1.03336072e+00 1.08242106e+00 2.48633206e-01 1.12474191e+00
-2.35782731e-02 2.17913926e-01 8.42458546e-01 4.50418413e-01
7.21254110e-01 5.91838062e-01 -7.60804236e-01 -3.83014023e-01
-1.12400126e+00 1.75754681e-01 5.78992963e-01 5.12347639e-01
1.99912176e-01 -1.56872720e-01 3.09576213e-01 9.38602507e-01
2.54479647e-01 -1.91212445e-01 5.81581950e-01 -8.87518585e-01
3.17323864e-01 4.76571530e-01 -2.21302062e-01 -1.06207168e+00
-7.68892765e-01 -6.83824062e-01 -9.32374954e-01 2.84096599e-01
6.16538584e-01 -1.25159279e-01 -3.56694013e-01 1.43873596e+00
-2.39412710e-01 -3.75417292e-01 -1.91768780e-01 5.08242190e-01
7.52316236e-01 2.16644332e-01 4.22193736e-01 1.16977222e-01
1.38005745e+00 -9.41307545e-01 -6.56933606e-01 -1.17750004e-01
1.12321723e+00 -9.33409154e-01 9.30108547e-01 5.95858753e-01
-7.29640245e-01 -5.46600044e-01 -1.07167220e+00 -1.11631557e-01
-5.22275031e-01 2.28272572e-01 8.67531836e-01 8.76078665e-01
-7.31898367e-01 7.53669143e-01 -7.15744734e-01 -5.07799983e-01
4.19180691e-01 4.05842066e-01 -3.25649291e-01 1.85664341e-01
-1.12233579e+00 9.29699361e-01 4.14957762e-01 1.65733889e-01
-3.27790946e-01 -2.36732081e-01 -2.91921526e-01 -7.88118467e-02
4.25956994e-01 -6.16789639e-01 1.38998711e+00 -1.14662170e+00
-1.15510392e+00 1.31667173e+00 1.91081151e-01 -3.94556433e-01
7.14058697e-01 9.16352645e-02 -4.66266394e-01 -9.81288627e-02
-6.31014854e-02 5.69686711e-01 4.76824999e-01 -1.25879109e+00
-9.94474649e-01 -3.89393032e-01 2.75531054e-01 -2.69722819e-01
-5.57462394e-01 5.10011673e-01 7.86064491e-02 -9.70638156e-01
-7.45272860e-02 -1.11085117e+00 1.99776404e-02 -3.67532998e-01
-3.01062047e-01 -7.50918090e-01 5.28036416e-01 -1.81060046e-01
1.12263322e+00 -1.93890607e+00 -3.83127302e-01 -4.42721993e-01
5.34931958e-01 3.11228335e-01 -1.45813793e-01 4.50062782e-01
-3.60154599e-01 6.66315734e-01 7.88627639e-02 -5.83127081e-01
2.53375798e-01 5.12353657e-03 -2.38857597e-01 5.42499483e-01
-1.75455138e-02 1.11728489e+00 -7.95144081e-01 -5.43126643e-01
-4.52459231e-02 4.60213900e-01 -2.54587773e-02 -2.59843051e-01
-3.66713777e-02 -6.72574388e-04 -5.62863827e-01 4.82339829e-01
5.17099500e-01 -5.10165274e-01 2.29230955e-01 -3.01361233e-01
-2.36449748e-01 4.02406991e-01 -7.20370829e-01 1.60285568e+00
-2.29926452e-01 9.95749593e-01 -1.12942882e-01 -1.27912521e+00
6.39672637e-01 6.24015510e-01 6.32150829e-01 -6.96186483e-01
5.83390296e-01 5.02994895e-01 1.35260433e-01 -6.49364591e-01
6.21354103e-01 -5.18958509e-01 3.15524228e-02 8.13874602e-01
-9.40245204e-03 2.72642970e-01 -2.32890233e-01 -8.87475014e-02
8.96496236e-01 -1.47368321e-02 5.63232526e-02 -5.50114930e-01
3.42379481e-01 -8.52959603e-02 9.69695896e-02 9.24822688e-01
-4.03998762e-01 4.61729825e-01 5.89127123e-01 -9.37052250e-01
-1.14786279e+00 -3.27006489e-01 -3.21548790e-01 1.47914839e+00
-5.34535706e-01 -4.31540459e-01 -6.76920474e-01 -1.30296397e+00
-2.08130211e-01 5.73915720e-01 -8.42296362e-01 1.07095920e-01
-5.03579795e-01 -1.13873696e+00 1.18460369e+00 6.47677124e-01
5.38666070e-01 -8.97694528e-01 -7.57504702e-01 2.47256577e-01
-1.56790972e-01 -1.03534031e+00 1.32474497e-01 5.83781183e-01
-9.57302451e-01 -8.68511558e-01 -7.84320295e-01 -8.32985401e-01
2.36543447e-01 2.96080798e-01 1.03463411e+00 4.38423634e-01
2.57582277e-01 -3.84891704e-02 -5.44065952e-01 -4.87775356e-01
-7.59849429e-01 6.66302145e-01 -2.86523849e-01 -5.00953197e-01
9.19745862e-01 -3.04449260e-01 -3.10549587e-01 9.87594500e-02
-9.01745737e-01 -1.42472144e-02 5.63478649e-01 8.65323842e-01
-2.08107561e-01 1.70412600e-01 7.96530366e-01 -1.08293641e+00
8.22457492e-01 -3.69459927e-01 -2.45245382e-01 3.18937033e-01
-1.16012776e+00 1.90939873e-01 4.29614961e-01 -2.20949993e-01
-9.14539754e-01 -3.27931911e-01 -6.36520922e-01 2.20487803e-01
-5.42390168e-01 5.66661417e-01 4.36341524e-01 -1.56199252e-02
6.87016964e-01 -1.59723535e-01 -3.49559784e-01 -6.28346741e-01
1.52674749e-01 9.59259033e-01 9.45095345e-02 -6.20193362e-01
2.76831746e-01 5.11167347e-01 -1.60839766e-01 -6.49340808e-01
-1.14363170e+00 -2.29834080e-01 -6.17511809e-01 -1.09380484e-01
8.26108575e-01 -6.62574947e-01 -1.01718879e+00 5.00939906e-01
-1.28669703e+00 -3.63860607e-01 2.17072293e-01 2.48513311e-01
-3.13139141e-01 6.74485207e-01 -8.60131323e-01 -5.52250326e-01
-2.63513118e-01 -1.19063318e+00 9.57036912e-01 -1.67789504e-01
-4.50154960e-01 -1.44277370e+00 6.18115477e-02 7.25240469e-01
6.58062756e-01 1.62239261e-02 1.01702905e+00 -1.03888988e+00
-4.66953665e-02 -3.44936639e-01 -4.85389978e-01 2.74906009e-01
1.25190809e-01 2.11009890e-01 -1.37813163e+00 -3.83209527e-01
2.58111328e-01 -7.71895528e-01 8.31723452e-01 1.43881381e-01
1.21898675e+00 -2.74667498e-02 -2.50287116e-01 1.53201088e-01
1.56303942e+00 3.91454518e-01 4.01852071e-01 9.44788218e-01
5.41721940e-01 6.80787265e-01 1.18504567e-02 2.26074591e-01
4.93968159e-01 6.68808103e-01 3.54077190e-01 -1.23021536e-01
-3.54176462e-02 4.27617840e-02 -1.31950766e-01 6.80392742e-01
-1.24015868e-01 -5.19441724e-01 -1.05042708e+00 2.11707979e-01
-1.62513506e+00 -9.90866661e-01 -4.87393498e-01 1.78264058e+00
6.90295577e-01 4.40563530e-01 1.62553266e-01 3.64436477e-01
4.37381327e-01 3.06791574e-01 -1.83611602e-01 -5.73779345e-01
-2.13088229e-01 4.79348414e-02 2.04917997e-01 3.13284069e-01
-1.08597350e+00 7.29507983e-01 7.04668951e+00 9.01648998e-01
-1.29536867e+00 3.47313792e-01 8.91708732e-01 1.74645320e-01
-2.28886098e-01 -2.61848509e-01 -7.64090538e-01 2.20548376e-01
1.19245422e+00 1.15995020e-01 -1.07072569e-01 5.26374578e-01
-1.23585723e-01 4.99101467e-02 -9.44613218e-01 8.54921877e-01
5.40994294e-02 -1.33487201e+00 -3.17798644e-01 1.42542377e-01
3.66673231e-01 2.81137973e-01 2.03342319e-01 4.52521145e-01
3.37735921e-01 -9.75193143e-01 7.10487843e-01 1.39030620e-01
4.51863319e-01 -6.93049729e-01 9.62379456e-01 2.68183380e-01
-6.89950764e-01 -1.49267405e-01 -3.45871925e-01 -2.66257674e-01
-1.18616551e-01 4.70994830e-01 -7.01229990e-01 4.69004601e-01
4.82932150e-01 4.29269850e-01 -8.41642797e-01 6.59580767e-01
9.35003608e-02 9.59558189e-01 3.90849747e-02 -2.02907756e-01
7.64697433e-01 1.35536820e-01 7.68281054e-03 1.52050126e+00
7.53204152e-03 -1.70062408e-01 -3.66972923e-01 6.28370583e-01
-1.64317906e-01 3.30295563e-01 -5.46366870e-01 -1.72666132e-01
-7.39357769e-02 1.45258224e+00 -1.13457096e+00 -3.24504942e-01
-9.89852369e-01 9.81568873e-01 4.51933295e-01 -3.19020241e-03
-7.89306879e-01 -1.49565548e-01 1.11614496e-01 -1.98621348e-01
9.39402729e-03 -1.05746001e-01 -7.29674637e-01 -1.18994570e+00
-8.45252648e-02 -1.04900610e+00 4.41819072e-01 -6.68667257e-01
-1.30468416e+00 9.20806885e-01 -1.30812243e-01 -7.65519798e-01
-4.83341403e-02 -7.16302216e-01 -1.21730074e-01 6.76530361e-01
-1.45556450e+00 -1.10766101e+00 4.85675149e-02 1.28433019e-01
5.29986262e-01 -1.40824005e-01 1.04297411e+00 5.50855875e-01
-5.68186998e-01 8.72363508e-01 2.76927888e-01 3.04959565e-01
8.86293948e-01 -1.21213377e+00 4.57436919e-01 2.71877348e-01
3.98204803e-01 5.15126526e-01 8.25575292e-01 -2.23945573e-01
-6.03062451e-01 -6.30040228e-01 1.47167540e+00 -6.81773245e-01
8.37963343e-01 -2.97689617e-01 -8.32012773e-01 8.48059118e-01
7.70475686e-01 -5.48718810e-01 9.88909662e-01 3.75347376e-01
-4.63884652e-01 2.24783540e-01 -9.48460877e-01 3.81584674e-01
5.16213775e-01 -5.03845036e-01 -7.24789917e-01 3.98250908e-01
4.07586694e-01 6.55611008e-02 -7.89850771e-01 3.37662965e-01
8.28224480e-01 -1.15534592e+00 5.49181819e-01 -6.90530837e-01
6.05810761e-01 4.14093405e-01 -5.74276410e-02 -9.84563768e-01
-3.62881094e-01 -9.39616282e-03 5.70179522e-01 1.23898220e+00
4.28450704e-01 -5.86342812e-01 9.74031031e-01 4.90688950e-01
-1.66832998e-01 -8.32459450e-01 -7.97210693e-01 -6.77851260e-01
8.25874627e-01 -8.25185537e-01 5.18926382e-01 1.60013533e+00
-3.88301499e-02 5.72976053e-01 -3.96618128e-01 -3.69332701e-01
4.27082121e-01 -5.68638602e-03 3.51670563e-01 -1.58600819e+00
-7.38623831e-03 -9.15651560e-01 8.49481486e-03 -8.42342436e-01
4.42869633e-01 -8.97013605e-01 -3.89098376e-01 -1.48962247e+00
4.74218875e-01 -6.31900012e-01 -5.87786615e-01 7.58109868e-01
-2.89135743e-02 3.56939703e-01 2.08795190e-01 2.15641782e-01
-6.62553132e-01 1.16211645e-01 1.13777316e+00 -2.71677613e-01
3.95757496e-01 1.59966648e-01 -1.16362071e+00 7.99379528e-01
9.83084619e-01 -8.92462075e-01 -1.00653023e-01 -6.70070767e-01
8.70069981e-01 -1.35524049e-01 2.05098182e-01 -8.26635122e-01
8.17403719e-02 1.44751728e-01 1.94400653e-01 -3.67624700e-01
1.25797442e-03 -9.48342025e-01 7.42173800e-03 5.30140519e-01
-9.31831539e-01 2.80327111e-01 4.13624704e-01 1.43601194e-01
-1.28305018e-01 -7.00788379e-01 6.99626505e-01 -3.17147583e-01
-4.22777146e-01 4.38732058e-02 -5.68274677e-01 -2.43112683e-01
5.69609165e-01 -7.13011548e-02 -4.57030207e-01 -3.42574328e-01
-6.62444413e-01 -1.81866720e-01 4.35656190e-01 6.46469891e-01
7.56504834e-02 -1.16210938e+00 -7.53873348e-01 -1.91430733e-01
-9.16417222e-03 -5.77679217e-01 1.49815083e-01 8.06806207e-01
-4.13757235e-01 8.78597498e-01 -1.08388521e-01 -1.34654865e-01
-1.24919164e+00 6.56162560e-01 5.12982547e-01 -4.79379922e-01
-6.69194460e-01 7.99801230e-01 -1.96149811e-01 -3.41571987e-01
4.53660101e-01 -3.06907594e-01 -2.03113362e-01 3.00407261e-01
4.50114936e-01 4.80512828e-01 4.09861386e-01 -5.04445493e-01
-2.40983039e-01 3.90344501e-01 -2.89822966e-01 -5.11809513e-02
1.34457695e+00 -1.00410223e-01 -8.18849653e-02 6.79024160e-01
1.34634829e+00 -5.55497147e-02 -8.24164331e-01 6.17082464e-03
1.68626249e-01 -1.42975375e-01 2.59550422e-01 -1.14986348e+00
-1.16807652e+00 1.15047204e+00 4.68593776e-01 8.44993591e-01
9.09675419e-01 -1.64876476e-01 6.68279111e-01 4.63317037e-01
2.13839844e-01 -1.09756756e+00 -4.83883079e-03 5.28640509e-01
6.21460199e-01 -1.37133431e+00 7.53871948e-02 2.16739789e-01
-3.90370160e-01 1.52486205e+00 3.72238964e-01 1.83817133e-01
6.08789086e-01 2.33443365e-01 2.98528790e-01 -5.38875937e-01
-7.67133355e-01 -7.93691073e-03 2.13680300e-03 2.58253604e-01
1.23184741e+00 -1.49112850e-01 -6.63823724e-01 3.97721052e-01
-7.02267811e-02 1.48243219e-01 4.89761740e-01 7.39086270e-01
-3.70995790e-01 -1.45381069e+00 -1.98743299e-01 4.25397933e-01
-1.07721508e+00 -1.35443926e-01 -4.56661731e-01 1.14925015e+00
3.74609947e-01 1.05539489e+00 -9.04132351e-02 -4.22852397e-01
1.93495989e-01 3.35486859e-01 5.15794277e-01 -3.24452221e-01
-8.89834762e-01 1.55168921e-02 1.56202435e-01 -1.16372287e-01
-8.32413852e-01 -5.06049573e-01 -9.03396428e-01 -5.25471270e-01
-4.51535225e-01 2.68848538e-01 9.68082786e-01 1.11390877e+00
1.44531578e-01 5.11751175e-01 1.64105430e-01 -4.45415109e-01
-6.69059277e-01 -1.18805480e+00 -4.64146465e-01 3.19079965e-01
1.43527955e-01 -4.70373452e-01 -4.31570828e-01 -2.02498868e-01] | [9.533815383911133, 4.634410858154297] |
3ea86dbc-0526-4689-acb5-74a902b96550 | romo-her-robust-model-based-hindsight | 2306.16061 | null | https://arxiv.org/abs/2306.16061v1 | https://arxiv.org/pdf/2306.16061v1.pdf | RoMo-HER: Robust Model-based Hindsight Experience Replay | Sparse rewards are one of the factors leading to low sample efficiency in multi-goal reinforcement learning (RL). Based on Hindsight Experience Replay (HER), model-based relabeling methods have been proposed to relabel goals using virtual trajectories obtained by interacting with the trained model, which can effectively enhance the sample efficiency in accurately modelable sparse-reward environments. However, they are ineffective in robot manipulation environment. In our paper, we design a robust framework called Robust Model-based Hindsight Experience Replay (RoMo-HER) which can effectively utilize the dynamical model in robot manipulation environments to enhance the sample efficiency. RoMo-HER is built upon a dynamics model and a novel goal relabeling technique called Foresight relabeling (FR), which selects the prediction starting state with a specific strategy, predicts the future trajectory of the starting state, and then relabels the goal using the dynamics model and the latest policy to train the agent. Experimental results show that RoMo-HER has higher sample efficiency than HER and Model-based Hindsight Experience Replay in several simulated robot manipulation environments. Furthermore, we integrate RoMo-HER and Relay Hindsight Experience Replay (RHER), which currently exhibits the highest sampling efficiency in most benchmark environments, resulting in a novel approach called Robust Model-based Relay Hindsight Experience Replay (RoMo-RHER). Our experimental results demonstrate that RoMo-RHER achieves higher sample efficiency over RHER, outperforming RHER by 25% and 26% in FetchPush-v1 and FetchPickandPlace-v1, respectively. | ['Bin Ren', 'Yuming Huang'] | 2023-06-28 | null | null | null | null | ['multi-goal-reinforcement-learning', 'robot-manipulation'] | ['methodology', 'robots'] | [-3.77290964e-01 9.97395590e-02 -3.37426096e-01 3.24639492e-02
-5.81918299e-01 -2.40755469e-01 5.63543320e-01 -1.00207016e-01
-7.11718261e-01 8.87548983e-01 1.70240983e-01 -7.25895539e-02
-4.41474468e-01 -5.38127780e-01 -6.66942835e-01 -7.32162178e-01
-3.98114949e-01 5.65313876e-01 1.47361577e-01 -6.83570743e-01
3.00563544e-01 3.36850286e-01 -1.33858585e+00 -2.88440645e-01
8.08406949e-01 5.39119720e-01 9.40101504e-01 5.55835724e-01
5.23351252e-01 1.30441320e+00 -1.95446298e-01 5.10611832e-01
4.26465601e-01 -4.36047465e-01 -5.29891908e-01 -1.75273582e-01
-7.09747553e-01 -8.75748992e-01 -8.02720010e-01 7.26194620e-01
6.71201706e-01 8.63859892e-01 5.22728622e-01 -1.44146454e+00
-1.95547238e-01 8.19865942e-01 -6.35066569e-01 -2.55901694e-01
3.50767583e-01 3.46798867e-01 5.72892547e-01 -8.56140316e-01
6.81099534e-01 1.52641332e+00 3.33482355e-01 6.97989881e-01
-8.80279183e-01 -5.13738751e-01 3.83330882e-01 2.34342322e-01
-1.11851263e+00 -7.93740153e-02 5.12965858e-01 -6.48808554e-02
9.84887481e-01 -2.23826647e-01 7.52885044e-01 1.08787811e+00
5.28358877e-01 1.23500872e+00 1.16173398e+00 -1.99754834e-01
6.38260007e-01 -2.68944055e-01 1.02022262e-02 9.54481363e-01
-1.39988318e-01 6.26424670e-01 -4.67792660e-01 -2.88303405e-01
1.17236638e+00 4.63877618e-01 -2.49865204e-01 -6.20180428e-01
-1.46505356e+00 8.64312530e-01 6.33991539e-01 -2.27259099e-01
-8.19587708e-01 4.19774771e-01 4.74373966e-01 7.62115300e-01
-3.34424019e-01 5.64149618e-01 -3.56773615e-01 -4.74402457e-01
-2.35367969e-01 7.12254822e-01 6.57418489e-01 1.34334147e+00
7.35549033e-01 3.11710358e-01 -1.90999508e-01 7.99358308e-01
2.07958952e-01 6.44491196e-01 8.42836022e-01 -1.26108193e+00
4.61290479e-01 3.39861572e-01 5.41325688e-01 -5.85455418e-01
-7.89953411e-01 -1.51079699e-01 -5.77423215e-01 5.12581527e-01
-1.31254913e-02 -3.80310684e-01 -8.76406729e-01 1.75795662e+00
4.13262337e-01 5.28700091e-02 5.84125161e-01 1.18259811e+00
3.30887377e-01 8.44556510e-01 -7.59288743e-02 -3.03149313e-01
8.43479931e-01 -1.41839504e+00 -3.84916425e-01 -3.38436991e-01
8.31316113e-01 -1.01093337e-01 1.20097625e+00 7.29978263e-01
-8.32210124e-01 -5.08610845e-01 -1.13459933e+00 5.17506897e-01
2.82655448e-01 2.45155960e-01 6.60955608e-01 -1.67289004e-01
-6.95141256e-01 8.80513847e-01 -1.11776423e+00 -2.95482337e-01
2.33849455e-02 4.24393266e-01 -3.52429420e-01 -3.60840857e-01
-9.56207812e-01 1.10533643e+00 5.92392623e-01 -9.23593491e-02
-1.90942800e+00 -2.60842979e-01 -9.63439465e-01 -1.50212556e-01
7.72850752e-01 -3.14843595e-01 1.60378993e+00 -2.62242138e-01
-2.11488509e+00 -4.88842875e-02 2.76721537e-01 -6.51714504e-01
4.75165397e-01 -4.28953558e-01 -8.07007253e-02 1.29803643e-01
1.28582329e-01 6.28993869e-01 8.14311206e-01 -1.43002069e+00
-6.92872047e-01 -1.44303277e-01 7.54021853e-02 7.04000890e-01
3.79515835e-03 -3.56367975e-01 -4.57055233e-02 -3.34895045e-01
1.57962009e-01 -1.23515952e+00 -6.87574387e-01 -4.48433578e-01
5.44505939e-02 -3.50484401e-01 8.33643377e-01 -2.54558891e-01
9.13997054e-01 -1.97153151e+00 4.41272646e-01 -1.06765162e-02
4.28555347e-02 2.70637780e-01 -3.90528977e-01 9.22276974e-01
4.52416837e-01 -6.94469392e-01 7.19526112e-02 -2.70651668e-01
9.08340588e-02 6.62032068e-01 -6.14317715e-01 5.61658144e-01
-3.40867609e-01 8.03544581e-01 -1.43109357e+00 -1.96989998e-01
3.55826676e-01 -3.76794636e-02 -6.40251994e-01 5.14542043e-01
-3.84027600e-01 6.03232205e-01 -8.10914278e-01 3.72317314e-01
3.33638787e-01 -8.12423900e-02 3.19923401e-01 4.71095353e-01
-2.07472630e-02 -5.16701788e-02 -1.22040963e+00 1.88235772e+00
-6.58818126e-01 -4.39455658e-02 1.13106668e-01 -8.80107284e-01
1.25764000e+00 9.16346386e-02 5.57640493e-01 -6.51061356e-01
1.53371081e-01 2.78002679e-01 -1.58054546e-01 -5.50183475e-01
9.86261308e-01 2.01560944e-01 -4.26180243e-01 5.35818875e-01
1.13020703e-01 -1.04495578e-01 -4.60197421e-04 3.20857167e-01
1.29220116e+00 8.36162150e-01 5.85238695e-01 7.06746802e-02
3.01904410e-01 2.04248041e-01 8.59147787e-01 1.09463727e+00
-4.85225081e-01 9.80926901e-02 1.69040844e-01 -2.80811578e-01
-8.62220883e-01 -1.04066956e+00 4.85324174e-01 1.27505147e+00
7.54646659e-01 -3.36250752e-01 -2.01038539e-01 -5.81604600e-01
-6.92660920e-03 8.94516170e-01 -4.08711940e-01 -4.29483801e-01
-9.08897281e-01 -3.56083572e-01 2.05108523e-01 3.86670411e-01
6.64768636e-01 -1.57880425e+00 -1.15172601e+00 6.15500629e-01
-1.84109196e-01 -7.20326066e-01 -3.95151794e-01 5.23717582e-01
-7.46347845e-01 -8.37875664e-01 -8.44410419e-01 -7.76374936e-01
5.35571814e-01 5.04502118e-01 4.70019609e-01 -1.44250140e-01
6.75987601e-02 5.62555611e-01 -7.53830731e-01 -1.81104928e-01
-4.60680157e-01 -1.60270214e-01 5.11234701e-01 -4.44462687e-01
-1.62119791e-01 -4.69864458e-01 -4.91461009e-01 4.82827932e-01
-5.39355338e-01 1.13917373e-01 8.11514676e-01 1.23537326e+00
6.19092762e-01 -1.00158878e-01 1.11836088e+00 -4.59852844e-01
8.59088182e-01 -7.35327840e-01 -6.14210665e-01 1.01596810e-01
-7.60441840e-01 2.95321137e-01 9.23826814e-01 -9.27736044e-01
-1.02989256e+00 9.61147025e-02 1.35582224e-01 -6.87352777e-01
1.39528066e-01 6.12415195e-01 1.66528627e-01 1.33514419e-01
7.35634565e-01 5.62397599e-01 3.02098840e-01 -2.82084852e-01
4.41032410e-01 6.79504097e-01 5.18413961e-01 -8.01117599e-01
3.83422405e-01 2.20099136e-01 1.04488946e-01 -5.72821081e-01
-3.91516358e-01 -4.74384457e-01 -3.73969926e-03 -4.16959554e-01
3.71365786e-01 -1.08972704e+00 -1.08114004e+00 6.13068104e-01
-6.81286991e-01 -1.00441480e+00 -3.77334535e-01 9.58915234e-01
-1.41829503e+00 3.33007485e-01 -7.40049660e-01 -1.24566483e+00
-2.68798977e-01 -1.29128683e+00 6.96243286e-01 3.70792300e-01
1.20275646e-01 -4.39765662e-01 -1.96780991e-02 -2.76380867e-01
2.46090278e-01 3.19049329e-01 6.91388071e-01 -5.59630334e-01
-4.40706372e-01 1.73824243e-02 2.71542370e-01 -1.47855133e-01
-2.04070240e-01 -7.10883558e-01 -3.45778674e-01 -7.71539450e-01
3.29862870e-02 -9.88824248e-01 5.75862110e-01 4.37117517e-01
6.22949779e-01 -3.33393663e-01 -5.20453513e-01 1.66173115e-01
1.33111835e+00 4.87197518e-01 5.17526209e-01 6.01236463e-01
4.05165523e-01 2.95174897e-01 1.64893627e+00 1.03077841e+00
6.79673076e-01 5.74829817e-01 9.25388217e-01 3.98164958e-01
3.41126829e-01 -7.81922936e-01 9.19585884e-01 8.91407967e-01
7.06732944e-02 -1.09383620e-01 -3.71442318e-01 5.42759180e-01
-2.70061183e+00 -8.01660955e-01 2.27171570e-01 2.19915557e+00
5.57017982e-01 7.49589829e-03 3.51645589e-01 -2.27851570e-01
5.41687310e-01 -4.17403318e-02 -1.04557490e+00 -2.51576126e-01
3.71993482e-01 -2.57139385e-01 6.50482833e-01 4.59482163e-01
-6.41434252e-01 1.20827079e+00 4.98869085e+00 9.41754878e-01
-7.72023380e-01 1.91104505e-02 -1.95879743e-01 6.62671402e-02
2.02618748e-01 2.24802092e-01 -7.84433007e-01 2.19541073e-01
7.28594780e-01 -2.55407184e-01 1.04829144e+00 1.44757211e+00
3.90017599e-01 -4.02298212e-01 -9.70676899e-01 9.08789456e-01
-2.11395308e-01 -8.31933081e-01 -2.91330963e-01 -3.79494466e-02
6.24328673e-01 1.96548045e-01 -1.64862558e-01 1.15785182e+00
1.16060424e+00 -8.05001855e-01 8.46015811e-01 5.76262236e-01
3.60706925e-01 -9.00008619e-01 5.16477585e-01 1.03428304e+00
-1.22800910e+00 -7.70883381e-01 -7.58401990e-01 -3.33287895e-01
3.36264104e-01 -1.86377078e-01 -1.15217304e+00 8.83972526e-01
4.86190856e-01 8.85808945e-01 1.94552794e-01 8.13858032e-01
-4.20023441e-01 2.94317901e-01 -2.10274935e-01 -5.23795247e-01
5.39331615e-01 -2.18361422e-01 7.37356603e-01 3.41506034e-01
3.75231236e-01 -1.82165077e-03 8.65026474e-01 4.17435050e-01
4.02167350e-01 -1.47561550e-01 -5.72262108e-01 1.53813958e-01
7.96499968e-01 1.15504801e+00 -3.42231065e-01 -2.71941185e-01
2.35101022e-02 8.25016260e-01 6.00601971e-01 3.54708880e-01
-8.95619333e-01 -4.24576670e-01 4.04703856e-01 -3.20179820e-01
2.52690941e-01 -4.43325698e-01 5.00936925e-01 -9.01338696e-01
-3.18113536e-01 -9.47821021e-01 2.87337899e-01 -1.01791179e+00
-8.46901774e-01 5.22149980e-01 1.51705414e-01 -1.53899443e+00
-6.90884054e-01 -3.12086314e-01 -3.11435640e-01 5.07276177e-01
-1.40869999e+00 -1.00411403e+00 -2.07398057e-01 6.86095357e-01
8.06366324e-01 -4.90226477e-01 8.47709537e-01 -3.01479846e-01
-2.76796401e-01 2.56278843e-01 2.77592361e-01 -3.98048669e-01
5.11576831e-01 -1.15170264e+00 6.73302934e-02 3.80582452e-01
-3.07746559e-01 4.70389545e-01 7.97339022e-01 -6.04675651e-01
-1.86091948e+00 -1.14340520e+00 -3.80564220e-02 1.38602436e-01
5.58187306e-01 1.86829809e-02 -6.27443135e-01 8.16737890e-01
-1.57193407e-01 -8.61312672e-02 -7.86973685e-02 -3.56926769e-01
1.36829674e-01 1.50031686e-01 -1.23526990e+00 1.02038932e+00
1.00533068e+00 3.86347771e-02 -7.06042528e-01 2.51637340e-01
9.96566057e-01 -6.77165091e-01 -8.10244262e-01 3.74118567e-01
5.16440690e-01 -6.68651044e-01 7.11257458e-01 -4.29824650e-01
2.38386244e-01 -4.16974723e-01 -1.44272372e-01 -1.71951699e+00
-2.68662691e-01 -9.44017351e-01 -3.17853183e-01 8.28281283e-01
-4.62725945e-02 -8.80516887e-01 5.41360915e-01 1.41092524e-01
-4.33314502e-01 -1.03490186e+00 -7.42675364e-01 -1.22258222e+00
-2.67447140e-02 -7.40886852e-02 6.76690042e-01 4.50626373e-01
3.79344672e-01 1.53577089e-01 -9.42279398e-01 2.28008926e-02
4.79081362e-01 2.07219914e-01 1.27909243e+00 -8.31088126e-01
-5.16288400e-01 1.69651061e-01 1.67554766e-02 -1.61919999e+00
2.49103069e-01 -8.80637467e-01 5.42999744e-01 -1.69149160e+00
-1.22440318e-02 -7.02958345e-01 -2.31189683e-01 5.49701929e-01
-3.73186544e-02 -7.12760627e-01 3.80673170e-01 4.99264121e-01
-8.95811856e-01 1.25104833e+00 1.69421327e+00 1.60698712e-01
-8.22828531e-01 6.40556514e-02 -3.55581671e-01 6.77393556e-01
9.12989438e-01 -5.84172189e-01 -7.12440610e-01 -7.19651803e-02
-1.07070185e-01 9.60093975e-01 1.62788600e-01 -1.09660244e+00
2.02242196e-01 -5.98904312e-01 -4.39668037e-02 -6.00804329e-01
5.66219807e-01 -6.95524275e-01 -6.29505292e-02 1.00547445e+00
-4.55490023e-01 1.97987199e-01 -7.02265278e-02 1.11311793e+00
2.00767472e-01 -4.48154807e-01 5.82274675e-01 -3.55452001e-01
-1.23613787e+00 3.42425346e-01 -7.23979056e-01 -1.53674945e-01
1.22674179e+00 2.02327892e-01 -3.87080640e-01 -6.06998980e-01
-6.71886146e-01 8.93813610e-01 3.27935100e-01 5.26034117e-01
9.73267913e-01 -1.44454598e+00 -4.26743299e-01 8.38735029e-02
1.42110407e-01 -3.08350883e-02 2.86434770e-01 6.50063038e-01
-2.47427315e-01 1.74849838e-01 -3.76358181e-01 -4.47742730e-01
-7.98015714e-01 7.68034697e-01 6.46784678e-02 -6.65769756e-01
-1.01300669e+00 4.30971682e-01 -5.18040210e-02 -9.13704395e-01
2.11472794e-01 -2.47345477e-01 -2.63621718e-01 -6.63079619e-01
3.80920559e-01 7.06898153e-01 -5.29191196e-01 -3.47935408e-01
-8.98850113e-02 7.09912926e-02 -1.53499693e-01 -3.60989273e-01
1.44616139e+00 -3.10847998e-01 2.86985785e-01 3.52558374e-01
6.35321796e-01 -4.78722215e-01 -1.86667788e+00 -2.84352809e-01
-1.56797633e-01 -3.34599882e-01 -9.15707499e-02 -6.48743033e-01
-5.36967099e-01 3.71289730e-01 4.29531604e-01 -2.75740057e-01
7.74799049e-01 -3.81960213e-01 8.53416443e-01 1.06089520e+00
1.33979595e+00 -1.24260914e+00 5.48116744e-01 8.18134010e-01
1.07912457e+00 -1.11275983e+00 -2.19186004e-02 2.22117230e-01
-1.24439657e+00 7.78009892e-01 9.81077969e-01 -5.66256642e-01
4.43314433e-01 -5.46419658e-02 -2.33384609e-01 3.56769748e-02
-9.35310543e-01 -2.28893742e-01 -6.21926069e-01 7.57551014e-01
-5.03380179e-01 1.63918793e-01 -2.56732285e-01 7.98089445e-01
-6.66752607e-02 2.63410002e-01 7.01003671e-01 1.37803388e+00
-8.80272627e-01 -8.65098357e-01 -2.47248277e-01 2.69394964e-01
1.71124309e-01 4.06814218e-01 3.37991238e-01 8.13228548e-01
-5.82816243e-01 1.23723662e+00 -3.65110725e-01 -6.77374065e-01
3.04532737e-01 -4.28292066e-01 5.69534004e-01 -5.44577241e-01
-5.35894930e-01 -1.56268161e-02 -1.93546414e-02 -7.25022197e-01
3.86743248e-02 -4.48477328e-01 -2.00358868e+00 -2.74875194e-01
-3.83021384e-01 2.68350542e-01 4.29709643e-01 8.90099406e-01
3.90943050e-01 4.55193579e-01 9.94381249e-01 -1.16423547e+00
-1.38325977e+00 -1.08187830e+00 -8.59462976e-01 5.68377264e-02
3.49776298e-01 -1.11159909e+00 -1.51475385e-01 -5.22314072e-01] | [4.216062545776367, 1.6197943687438965] |
180645c8-536a-4bc8-bcb6-2ad48e341022 | an-automatic-image-content-retrieval-method | 2108.12068 | null | https://arxiv.org/abs/2108.12068v1 | https://arxiv.org/pdf/2108.12068v1.pdf | An Automatic Image Content Retrieval Method for better Mobile Device Display User Experiences | A growing number of commercially available mobile phones come with integrated high-resolution digital cameras. That enables a new class of dedicated applications to image analysis such as mobile visual search, image cropping, object detection, content-based image retrieval, image classification. In this paper, a new mobile application for image content retrieval and classification for mobile device display is proposed to enrich the visual experience of users. The mobile application can extract a certain number of images based on the content of an image with visual saliency methods aiming at detecting the most critical regions in a given image from a perceptual viewpoint. First, the most critical areas from a perceptual perspective are extracted using the local maxima of a 2D saliency function. Next, a salient region is cropped using the bounding box centred on the local maxima of the thresholded Saliency Map of the image. Then, each image crop feds into an Image Classification system based on SVM and SIFT descriptors to detect the class of object present in the image. ImageNet repository was used as the reference for semantic category classification. Android platform was used to implement the mobile application on a client-server architecture. A mobile client sends the photo taken by the camera to the server, which processes the image and returns the results (image contents such as image crops and related target classes) to the mobile client. The application was run on thousands of pictures and showed encouraging results towards a better user visual experience with mobile displays. | ['Alessandro Bruno'] | 2021-08-26 | null | null | null | null | ['content-based-image-retrieval', 'image-cropping'] | ['computer-vision', 'computer-vision'] | [ 5.39834619e-01 -1.97346732e-01 -3.56899738e-01 -4.99624908e-02
-4.14293945e-01 -3.84567618e-01 4.04603928e-01 4.28882897e-01
-3.67531180e-01 1.60250083e-01 -2.04354078e-01 -6.12987876e-02
-2.48989016e-02 -7.51805186e-01 -5.20575643e-01 -6.47316456e-01
1.87600970e-01 -1.93833977e-01 7.82486796e-01 -2.78993994e-01
8.15099597e-01 4.96625632e-01 -2.41464901e+00 4.50855106e-01
5.79738617e-01 1.46089566e+00 1.14013910e+00 5.82330167e-01
-1.76626146e-01 4.33034539e-01 -4.64970618e-01 1.42480358e-01
4.96989600e-02 -1.61252916e-01 -4.77943391e-01 3.72806758e-01
2.49755010e-01 -1.02925092e-01 1.86705261e-01 1.29212856e+00
3.04167539e-01 9.97025073e-02 3.50236714e-01 -1.30446255e+00
-2.07740426e-01 -1.48865923e-01 -4.85185832e-01 4.49714899e-01
4.96923029e-01 -2.29746908e-01 5.86133242e-01 -1.23116374e+00
7.21832812e-01 8.82192194e-01 1.55582473e-01 4.06254679e-02
-6.27956271e-01 -1.83295533e-01 -2.05632418e-01 5.91863811e-01
-1.20108318e+00 -9.44928676e-02 8.81074727e-01 -2.43721321e-01
7.10307240e-01 5.89237869e-01 8.42633188e-01 3.26464623e-01
2.83851266e-01 6.41621470e-01 1.04975104e+00 -4.64830339e-01
4.45999503e-01 6.95989728e-01 1.04455091e-01 5.13334036e-01
-4.47368287e-02 -2.47854725e-01 -5.06992459e-01 1.02102242e-01
5.49815416e-01 5.23588121e-01 -4.56096120e-02 -6.33958161e-01
-9.73422825e-01 6.18812084e-01 8.74388456e-01 5.96693695e-01
-7.86331594e-01 -2.75644869e-01 2.06558779e-01 -4.07618284e-02
4.22456026e-01 1.35173336e-01 -2.79489428e-01 7.73464888e-02
-1.05139649e+00 1.60748251e-02 1.72709867e-01 5.79733014e-01
1.08009732e+00 -3.09872389e-01 1.47368520e-01 6.24133885e-01
3.26860279e-01 7.82920837e-01 7.95246899e-01 -6.96470916e-01
1.33321628e-01 1.18174052e+00 -2.22550277e-02 -1.47567606e+00
-1.32663876e-01 -2.02252537e-01 -4.06334966e-01 5.13089538e-01
7.20776916e-02 3.49033713e-01 -6.14211142e-01 7.54015803e-01
6.56156600e-01 -3.55203077e-02 1.06974011e-02 1.16395414e+00
7.33682334e-01 7.52915740e-01 5.03783077e-02 -5.82153238e-02
1.71206820e+00 -6.76411867e-01 -3.87805313e-01 -3.32103342e-01
2.51049042e-01 -9.68428791e-01 1.19234753e+00 2.61319488e-01
-9.20210063e-01 -7.16446936e-01 -9.98431802e-01 5.68077937e-02
-8.60786498e-01 4.26078290e-01 2.59904325e-01 4.43012506e-01
-1.09814239e+00 1.61585659e-01 -2.91289806e-01 -6.64418280e-01
3.05909753e-01 3.77066731e-01 -2.41248161e-01 7.71147832e-02
-8.19244146e-01 8.15639615e-01 5.15116930e-01 -2.83869714e-01
-4.77062166e-01 -3.55568469e-01 -7.67309964e-01 1.92129090e-01
1.67675301e-01 -1.85033098e-01 6.44710124e-01 -1.82153451e+00
-1.20956779e+00 1.14239049e+00 -2.40664646e-01 -4.54507321e-01
-3.17476168e-02 1.12954102e-01 -3.54638726e-01 6.56004727e-01
3.43242586e-01 8.16935658e-01 1.33889627e+00 -1.00389957e+00
-1.15301538e+00 -5.11112690e-01 -2.37916797e-01 6.89801812e-01
-3.66572767e-01 2.02185258e-01 -5.86523354e-01 -3.42840970e-01
1.40773252e-01 -6.73306763e-01 2.19149753e-01 -4.69561756e-01
7.06862435e-02 -1.42137662e-01 1.62538707e+00 -6.83465958e-01
9.07641470e-01 -2.37785196e+00 -6.53607473e-02 4.75845754e-01
-9.66252312e-02 4.08311784e-01 1.75758839e-01 5.21320663e-02
-4.23203185e-02 -2.26591974e-01 1.11253135e-01 1.54191226e-01
-6.61670148e-01 -3.73053879e-01 -2.45697767e-01 8.38843733e-02
-2.33052224e-02 7.93715179e-01 -6.75189972e-01 -6.82443976e-01
6.94642365e-01 3.22352648e-01 -2.29431421e-01 -4.92956070e-03
-3.79824564e-02 1.10707432e-01 -3.73696238e-01 8.31068099e-01
6.11162543e-01 -9.87060592e-02 -1.89981043e-01 -1.06002800e-01
-3.65490317e-01 -1.47991925e-01 -9.90854621e-01 1.28005075e+00
-3.99938703e-01 8.30241680e-01 -8.50641280e-02 -9.42725658e-01
1.08733118e+00 -5.27491942e-02 4.47410822e-01 -8.79376113e-01
1.70089304e-01 3.95079255e-01 -5.24640024e-01 -6.23190761e-01
8.76178682e-01 5.35290360e-01 5.42092994e-02 1.31665722e-01
-2.19299763e-01 -4.23993096e-02 8.79207533e-03 1.09029137e-01
4.60649341e-01 3.43638845e-02 3.21797401e-01 -2.60014355e-01
7.81286657e-01 3.13440084e-01 -3.19436431e-01 2.48593315e-01
-4.87187654e-02 5.44007957e-01 -4.17501070e-02 -5.10530233e-01
-1.07398498e+00 -7.60129392e-01 -1.09203337e-02 1.31223309e+00
8.12984765e-01 -2.58761980e-02 -1.10950816e+00 -4.97302651e-01
-2.05433458e-01 3.21003139e-01 -2.93798983e-01 1.01399552e-02
-2.52671838e-01 -2.88437575e-01 -3.68320882e-01 -1.33857489e-01
8.10147583e-01 -1.55751193e+00 -1.48004603e+00 -1.14486456e-01
-9.96407419e-02 -7.30796993e-01 -3.76280636e-01 -4.91428263e-02
-9.48669195e-01 -1.01873219e+00 -9.63234425e-01 -1.36486483e+00
7.39957988e-01 9.37530279e-01 5.00156879e-01 7.22870156e-02
-4.42640960e-01 4.27316278e-01 -4.03125107e-01 -3.82969618e-01
-1.66509137e-01 8.84961113e-02 -1.63612172e-01 3.14623058e-01
3.23299825e-01 -3.16985026e-02 -9.05841708e-01 3.62519920e-01
-8.63799036e-01 1.29543245e-01 5.23307264e-01 3.98610622e-01
7.44975746e-01 1.89763904e-01 3.91709000e-01 -1.21988408e-01
5.08309424e-01 -4.74429190e-01 -7.27228761e-01 1.27905354e-01
-4.18438703e-01 -2.90097922e-01 2.60539263e-01 -3.12068820e-01
-7.32415318e-01 1.93485007e-01 3.01132202e-01 -2.19071612e-01
-2.13837042e-01 3.57463717e-01 -1.92067608e-01 -2.09462680e-02
5.96954942e-01 6.48368835e-01 2.81134367e-01 -1.83990747e-01
2.23535627e-01 1.23285842e+00 4.97630268e-01 5.48188448e-01
2.97005475e-01 3.59570861e-01 -9.27333608e-02 -1.15546191e+00
-2.24478826e-01 -9.03800964e-01 -3.67182016e-01 -6.56098247e-01
8.45983446e-01 -7.61215389e-01 -5.29338479e-01 4.21830118e-01
-8.08751404e-01 1.29924893e-01 -1.10409841e-01 3.46275359e-01
-4.83458400e-01 1.45838082e-01 3.72170322e-02 -7.05703318e-01
-6.92773104e-01 -1.21587026e+00 1.24548054e+00 7.58475721e-01
-6.29556328e-02 -5.33957243e-01 -4.06568706e-01 2.77181447e-01
4.13741887e-01 -1.09799892e-01 8.09176683e-01 -3.85373056e-01
-6.65591598e-01 -4.63705987e-01 -3.95287126e-01 2.64962137e-01
1.97871983e-01 -9.49870273e-02 -8.04173350e-01 9.32686031e-02
1.24540469e-02 2.48080269e-01 4.48403805e-01 6.51419163e-01
1.05063128e+00 -2.71311492e-01 -7.25201666e-01 1.69589326e-01
1.59429872e+00 5.34498632e-01 8.28638136e-01 6.35764658e-01
5.07395327e-01 5.58636487e-01 1.17547405e+00 5.29383533e-02
9.21623036e-02 9.89650309e-01 7.30663478e-01 -2.19365701e-01
-5.75257465e-02 -1.70515031e-02 4.28288311e-01 2.42608413e-01
2.30852991e-01 2.59387970e-01 -6.64487779e-01 6.11349583e-01
-1.60520697e+00 -8.77688169e-01 -1.40045062e-01 2.51313734e+00
3.49770099e-01 2.18746047e-02 3.87818068e-01 4.09146965e-01
1.05804014e+00 -2.65122652e-01 -3.61622840e-01 -4.51036513e-01
5.07483073e-02 5.78270890e-02 5.28067827e-01 2.38146842e-01
-1.14556861e+00 9.48558450e-01 5.10854197e+00 1.00874579e+00
-1.64407122e+00 2.22726688e-02 7.27355957e-01 1.72030866e-01
1.42963827e-01 -1.93352669e-01 -7.57202685e-01 8.80995989e-01
7.01962650e-01 5.25990538e-02 3.26128989e-01 1.30187988e+00
3.21393490e-01 -9.07933235e-01 -3.36031139e-01 1.20151448e+00
2.87716895e-01 -1.32478249e+00 9.47572440e-02 1.10211186e-01
5.62154531e-01 -4.01809104e-02 3.98630857e-01 -2.99852848e-01
-8.46245706e-01 -5.71793675e-01 7.25542367e-01 5.23443580e-01
6.33707821e-01 -9.22960222e-01 6.79656446e-01 2.59810805e-01
-1.04599750e+00 -3.48553389e-01 -2.58935481e-01 -7.52951624e-03
-2.79461801e-01 3.09690267e-01 -1.06026483e+00 6.81099743e-02
9.86572921e-01 5.13024271e-01 -9.38808978e-01 1.17502880e+00
3.38793278e-01 9.49672312e-02 -1.92863926e-01 -4.53091413e-01
1.65174440e-01 -2.88786709e-01 6.36070669e-01 8.59083831e-01
6.87256157e-01 -3.10936630e-01 -2.07031444e-01 3.90190244e-01
2.61365235e-01 5.44708729e-01 -8.78828704e-01 3.63891721e-01
3.24866891e-01 1.57277751e+00 -1.40225232e+00 -6.07290208e-01
-1.30090371e-01 1.12626946e+00 -4.14217055e-01 7.07190335e-02
-4.91950870e-01 -6.73493087e-01 2.15882838e-01 5.10403991e-01
6.21295571e-01 -2.32813768e-02 -2.41491660e-01 -6.20368123e-01
6.54880106e-02 -8.63680720e-01 9.11704004e-02 -1.33779860e+00
-3.37327242e-01 5.01248837e-01 -1.16912439e-01 -1.40694344e+00
-2.68915772e-01 -5.69529414e-01 -5.11748016e-01 7.54240513e-01
-1.00701725e+00 -8.51059437e-01 -6.54934168e-01 7.53090978e-01
8.32103133e-01 -2.77647883e-01 6.21441782e-01 9.27724317e-02
-4.39436398e-02 -1.48007557e-01 1.50671929e-01 -2.71221787e-01
2.70297468e-01 -1.09702659e+00 -1.33423908e-02 8.61772597e-01
2.50792831e-01 2.43106514e-01 5.19717872e-01 -7.67930448e-01
-1.20904660e+00 -8.73173714e-01 9.36371922e-01 -3.37128192e-01
2.50974387e-01 -1.46595642e-01 -5.66627026e-01 -1.11793198e-01
1.85679197e-01 -1.80304661e-01 1.33606583e-01 -7.33410239e-01
3.95117968e-01 -3.26460689e-01 -1.28367591e+00 6.37605667e-01
2.40877405e-01 -4.27683383e-01 -3.23793650e-01 2.69741893e-01
3.81983638e-01 -2.27524236e-01 -3.50605339e-01 -1.29642515e-02
3.78671080e-01 -8.73828232e-01 1.07586801e+00 6.62248358e-02
2.94841349e-01 -5.30083001e-01 -1.22167222e-01 -8.70900214e-01
1.63026884e-01 -1.70031190e-01 2.46176004e-01 8.71067166e-01
1.34134367e-01 -2.59738237e-01 7.10580766e-01 1.33455858e-01
2.09146902e-01 -5.41033566e-01 -7.57370353e-01 -1.32756963e-01
-9.96735156e-01 -1.44275114e-01 3.80533665e-01 3.12586457e-01
1.12013824e-01 2.61608124e-01 -2.56112218e-02 1.15918443e-01
4.55542803e-01 2.85106868e-01 5.88019311e-01 -1.10118353e+00
2.71749407e-01 -4.60942388e-01 -8.86204660e-01 -5.13707101e-01
-3.69369626e-01 -9.01668131e-01 -2.68474340e-01 -1.46109152e+00
1.77526802e-01 -2.05826879e-01 -1.54798418e-01 1.67763770e-01
-6.86539561e-02 8.12063992e-01 3.15392256e-01 4.73639101e-01
-6.99420273e-01 -1.53625786e-01 8.43759477e-01 -1.97641060e-01
-5.30991912e-01 4.01361614e-01 -5.52550733e-01 6.51271105e-01
8.98684621e-01 -2.07319915e-01 -3.12837034e-01 2.37613589e-01
1.76933810e-01 -1.48638785e-01 6.93122566e-01 -1.25789320e+00
2.07113773e-01 7.52290636e-02 5.29030085e-01 -8.40216815e-01
3.24866414e-01 -1.04600179e+00 2.51956154e-02 7.13853896e-01
-7.99415186e-02 3.07775699e-02 -2.34240144e-02 2.56352335e-01
-3.06573808e-01 -4.19250160e-01 6.52873635e-01 -9.05683823e-03
-1.28123081e+00 -2.16392651e-01 -6.27177298e-01 -5.63164771e-01
1.44706643e+00 -6.67576611e-01 5.70289381e-02 -3.49541336e-01
-5.46084762e-01 -3.27498823e-01 4.97047126e-01 6.06944382e-01
8.80852461e-01 -1.21939909e+00 -4.35402207e-02 4.43170369e-01
1.58562601e-01 -4.03514475e-01 2.86534518e-01 8.25834811e-01
-6.32801354e-01 3.83271873e-01 -4.66245800e-01 -9.67896163e-01
-1.87314665e+00 8.81287158e-01 1.22413054e-01 3.07019800e-01
-4.43526119e-01 1.80196151e-01 1.21619757e-02 3.40002179e-01
2.70698546e-03 -4.56037402e-01 -6.77142501e-01 3.82849872e-01
7.11060286e-01 4.57149267e-01 2.77454227e-01 -1.08111250e+00
-5.51068544e-01 8.57586026e-01 2.86076069e-01 8.46424513e-03
9.82309699e-01 -4.50540423e-01 -1.41590238e-01 3.05828571e-01
1.43308210e+00 -1.99858136e-02 -9.93257821e-01 2.58581452e-02
-3.61957774e-03 -4.76396084e-01 1.97594702e-01 -6.99143708e-01
-9.02536929e-01 6.76787257e-01 1.36614466e+00 5.43651700e-01
1.50413251e+00 9.09079611e-02 3.97564024e-01 5.20680472e-02
4.69490618e-01 -1.25516379e+00 2.10671991e-01 7.07602054e-02
7.28014827e-01 -1.24310410e+00 -7.01763406e-02 -1.70919016e-01
-1.05012989e+00 9.71982658e-01 1.10619493e-01 -2.07245886e-01
5.80972731e-01 -1.21134818e-01 1.09600708e-01 -2.09766567e-01
6.32599294e-02 -4.90191638e-01 5.44714749e-01 7.22509086e-01
-1.68851256e-01 3.18397284e-02 -2.22002029e-01 1.92809612e-01
-4.61601540e-02 1.43706407e-02 2.90684998e-01 7.85035431e-01
-1.04918540e+00 -7.54125953e-01 -6.64381385e-01 3.04138511e-01
-5.51236808e-01 -1.28769353e-01 -2.57795274e-01 4.78264511e-01
2.62898028e-01 9.61403131e-01 3.72446090e-01 -2.76771605e-01
4.26714425e-04 -1.63915232e-01 1.00010350e-01 -3.75847280e-01
-5.17806828e-01 1.04440689e-01 -6.01655185e-01 -4.64484781e-01
-3.22884709e-01 -5.18641114e-01 -1.22866893e+00 3.32241088e-01
-1.87955692e-01 2.08686814e-01 1.42789543e+00 5.76473355e-01
5.27713180e-01 1.22544721e-01 8.48451614e-01 -1.19716990e+00
1.69894710e-01 -7.67571628e-01 -4.63390708e-01 4.13821757e-01
1.04158141e-01 -3.74013007e-01 -8.92340466e-02 2.45410666e-01] | [9.947887420654297, -0.4740324914455414] |
a2710ae6-ed26-4b72-8730-f46874a84f9c | energy-based-processes-for-exchangeable-data | 2003.07521 | null | https://arxiv.org/abs/2003.07521v2 | https://arxiv.org/pdf/2003.07521v2.pdf | Energy-Based Processes for Exchangeable Data | Recently there has been growing interest in modeling sets with exchangeability such as point clouds. A shortcoming of current approaches is that they restrict the cardinality of the sets considered or can only express limited forms of distribution over unobserved data. To overcome these limitations, we introduce Energy-Based Processes (EBPs), which extend energy based models to exchangeable data while allowing neural network parameterizations of the energy function. A key advantage of these models is the ability to express more flexible distributions over sets without restricting their cardinality. We develop an efficient training procedure for EBPs that demonstrates state-of-the-art performance on a variety of tasks such as point cloud generation, classification, denoising, and image completion. | ['Dale Schuurmans', 'Hanjun Dai', 'Bo Dai', 'Mengjiao Yang'] | 2020-03-17 | null | https://proceedings.icml.cc/static/paper_files/icml/2020/2926-Paper.pdf | https://proceedings.icml.cc/static/paper_files/icml/2020/2926-Paper.pdf | icml-2020-1 | ['point-cloud-generation'] | ['computer-vision'] | [ 1.73504367e-01 -2.67511487e-01 1.33424059e-01 -3.10608864e-01
-8.38686049e-01 -5.11516929e-01 7.35105753e-01 8.71200040e-02
-1.95946902e-01 7.41818011e-01 1.92238782e-02 -1.87920574e-02
-3.34569097e-01 -1.20907998e+00 -9.54771221e-01 -7.22354293e-01
2.87485886e-02 7.17134237e-01 1.34896589e-02 2.25943141e-02
4.72028442e-02 9.75252688e-01 -1.69560158e+00 7.16197416e-02
9.32262361e-01 9.11279142e-01 2.91122288e-01 4.53746855e-01
-1.88696608e-01 4.27812427e-01 -2.58045137e-01 -1.42102852e-01
2.68666863e-01 -2.54419893e-02 -5.87509990e-01 2.05992877e-01
2.25396082e-01 -1.63632572e-01 -4.97784674e-01 8.63066137e-01
3.53823155e-01 4.83216524e-01 8.27491462e-01 -1.35085130e+00
-9.25875068e-01 6.85050711e-02 -2.83440173e-01 -4.69188355e-02
-2.99626738e-02 9.41515639e-02 8.12580764e-01 -1.08655727e+00
5.77380002e-01 1.17784905e+00 8.36264491e-01 6.57255352e-01
-1.66061568e+00 -4.46370542e-01 1.94663122e-01 -1.95847210e-02
-1.47102451e+00 -4.07862157e-01 8.30902636e-01 -5.10327220e-01
1.20826888e+00 3.80762994e-01 7.99691498e-01 8.34657252e-01
2.33047813e-01 6.60291970e-01 9.74997163e-01 -3.94254982e-01
5.77732146e-01 -2.60497570e-01 8.54720362e-03 4.04032856e-01
2.33061075e-01 7.15375841e-02 -5.60033441e-01 -5.61703146e-01
9.08013344e-01 1.81833684e-01 -2.28886887e-01 -6.03631675e-01
-9.85701025e-01 8.86390984e-01 2.50664026e-01 -4.59870622e-02
-5.77911377e-01 6.57199800e-01 -6.11538142e-02 -2.51693632e-02
1.00546050e+00 4.23863292e-01 -3.59178305e-01 -1.18156262e-02
-9.11465824e-01 4.67916518e-01 6.52340412e-01 9.98519242e-01
9.68662977e-01 3.94207090e-02 -1.65971100e-01 7.90304482e-01
2.57500917e-01 5.48575222e-01 -8.12791735e-02 -1.42455757e+00
3.50412786e-01 3.56136858e-01 3.26437712e-01 -6.36630654e-01
-1.12962894e-01 -2.36294195e-01 -9.82150197e-01 5.71782827e-01
-1.11382216e-01 3.37434337e-02 -1.27484834e+00 1.96683884e+00
2.62587339e-01 2.17624709e-01 -5.49466014e-02 5.23270726e-01
4.90959018e-01 8.95534694e-01 2.15508625e-01 -8.86288881e-02
9.82446134e-01 -7.16108978e-01 -5.49430549e-01 -1.00314558e-01
8.01605284e-02 -4.86688107e-01 9.14081872e-01 3.34443629e-01
-1.44442475e+00 -3.06724161e-01 -6.55784667e-01 -1.35926425e-01
-2.87870318e-01 -3.36181164e-01 8.94040585e-01 5.32423377e-01
-1.34498370e+00 9.73584473e-01 -1.32400966e+00 -1.74841926e-01
6.04492605e-01 6.56574309e-01 -1.35648280e-01 -8.02179053e-02
-6.38630629e-01 7.20042169e-01 3.91748324e-02 -4.09275964e-02
-7.73699462e-01 -1.01617646e+00 -6.74114048e-01 4.26497966e-01
-8.55694350e-04 -1.22781968e+00 1.05949295e+00 -7.35469222e-01
-1.29974794e+00 6.20951176e-01 -2.69527555e-01 -3.80322456e-01
4.83548135e-01 -4.82069068e-02 -1.73738554e-01 4.65927310e-02
-1.15608349e-01 1.04708791e+00 8.38377535e-01 -1.39568746e+00
-2.53773957e-01 -1.99255690e-01 -2.91508548e-02 4.19243217e-01
-2.97662288e-01 -1.81695864e-01 -4.06429976e-01 -6.88239038e-01
2.23314852e-01 -1.25662553e+00 -4.15955573e-01 3.78308713e-01
-3.49509358e-01 -1.82067975e-01 8.24086607e-01 -2.68175304e-01
5.64317524e-01 -1.97158277e+00 3.92272264e-01 6.40160367e-02
5.29114529e-02 -1.91153750e-01 -1.72627926e-01 6.72888875e-01
-3.05653904e-02 4.58899796e-01 -5.61717212e-01 -9.34899867e-01
2.10024446e-01 7.72526443e-01 -5.04747391e-01 3.14292341e-01
4.01117623e-01 6.24837339e-01 -6.45265877e-01 -2.69564569e-01
4.89279360e-01 8.76701593e-01 -6.75845444e-01 -5.22809960e-02
-5.19678593e-01 6.99022353e-01 -4.56668466e-01 3.95215303e-01
9.64818597e-01 -4.03185785e-01 -3.98286611e-01 -5.95527217e-02
-8.15876201e-02 1.78353205e-01 -1.29172766e+00 1.82345665e+00
-4.84025270e-01 4.45358872e-01 2.58068025e-01 -5.11390984e-01
3.36884677e-01 4.69135433e-01 1.08266425e+00 1.12332165e-01
-2.84814686e-01 -1.02042496e-01 -3.79925728e-01 8.28026037e-04
6.96282089e-01 -4.75380480e-01 2.14776456e-01 3.17239255e-01
-7.16887116e-02 -6.36711895e-01 -5.22820577e-02 -2.73695029e-02
1.22751021e+00 2.11271286e-01 -1.99176550e-01 -3.11895221e-01
5.64740002e-02 -3.96731310e-02 5.78428805e-01 8.24032307e-01
3.45331997e-01 1.14129913e+00 -8.81535336e-02 -2.55679160e-01
-1.14610183e+00 -1.31911194e+00 -3.29498291e-01 7.15181708e-01
1.23098187e-01 -1.08348466e-01 -4.92071003e-01 -2.79689938e-01
1.40994027e-01 8.45661104e-01 -4.54744697e-01 1.88534502e-02
-4.54159707e-01 -1.12097406e+00 2.07806364e-01 5.98342001e-01
4.14234489e-01 -9.79411185e-01 -4.36540991e-01 1.89424977e-01
-2.69139677e-01 -1.10614610e+00 -3.05427700e-01 1.28108442e-01
-1.32438076e+00 -6.95116341e-01 -3.71768951e-01 -4.06859398e-01
8.57869208e-01 2.12308004e-01 1.45539749e+00 -3.58830243e-02
-9.03007984e-02 7.73845732e-01 8.45536664e-02 -5.17813146e-01
-3.14720780e-01 -5.72886951e-02 -1.13243144e-02 -1.98334426e-01
-1.15318231e-01 -7.74589419e-01 -7.19891012e-01 3.19469124e-02
-1.28185570e+00 3.01643424e-02 2.82751709e-01 6.18855834e-01
1.12518084e+00 2.49478087e-01 2.78103530e-01 -6.36359334e-01
4.56243813e-01 -4.93673831e-01 -4.92039531e-01 7.91314915e-02
-5.14458895e-01 3.68644595e-01 3.10472041e-01 -4.56759274e-01
-1.26082075e+00 1.54713705e-01 -3.26248705e-01 -6.79705203e-01
-4.19966057e-02 3.01940590e-01 -2.01738223e-01 -3.16102535e-01
2.50459164e-01 -5.29316114e-03 -1.76411077e-01 -4.77301270e-01
3.26471746e-01 9.05280188e-02 4.25858587e-01 -9.93538857e-01
6.53153837e-01 9.77353811e-01 4.58626479e-01 -7.35640287e-01
-5.52647173e-01 -3.95610839e-01 -2.91602552e-01 2.61128861e-02
8.40797842e-01 -9.22058105e-01 -4.80299145e-01 4.43693995e-01
-1.28697801e+00 -2.97135979e-01 -8.18667650e-01 3.61943036e-01
-8.50592196e-01 3.55583996e-01 -6.52925313e-01 -9.05118167e-01
-4.99974966e-01 -1.04258728e+00 1.36345077e+00 -7.89975077e-02
-5.89680187e-02 -1.09181583e+00 3.46888714e-02 9.57621541e-03
3.57482255e-01 4.06604558e-01 1.00983703e+00 -2.32894465e-01
-1.12818027e+00 -2.62244865e-02 1.37605160e-01 4.51741815e-01
3.75660509e-02 1.93446577e-01 -9.00506139e-01 -2.56384015e-01
2.21297994e-01 -1.17679529e-01 8.74251544e-01 8.04089308e-01
1.73181188e+00 -1.05701089e-01 -5.84838033e-01 7.71803796e-01
1.65604901e+00 -7.32714906e-02 7.12850213e-01 -2.19471082e-02
5.37320852e-01 3.13806564e-01 7.75807202e-02 4.62854207e-01
2.50582099e-01 6.15874708e-01 6.68906629e-01 -1.09739155e-01
-6.95537627e-02 -1.79943025e-01 -8.85502174e-02 7.24836409e-01
-3.13114166e-01 -7.68075109e-01 -7.47971952e-01 7.08050370e-01
-1.80817723e+00 -9.56980526e-01 -1.68154910e-01 2.24926114e+00
5.43876529e-01 -1.00724123e-01 -3.67296308e-01 -1.76577643e-01
6.32252514e-01 1.08792596e-01 -6.47692204e-01 -8.74987319e-02
-2.06946328e-01 4.32403803e-01 6.47294402e-01 5.33386827e-01
-1.01068103e+00 5.17870843e-01 7.13539219e+00 7.14366198e-01
-7.05700636e-01 2.86429852e-01 5.62677741e-01 -2.83929110e-01
-6.85227931e-01 4.52910215e-02 -5.39499819e-01 5.01233578e-01
7.27880478e-01 9.73883048e-02 5.97954035e-01 6.33635283e-01
2.21820995e-01 7.02420250e-02 -1.28928101e+00 9.25472438e-01
-2.30906472e-01 -1.63179135e+00 1.43196389e-01 3.94687176e-01
1.01909959e+00 5.28757393e-01 2.31939554e-01 6.25990033e-02
6.40148878e-01 -1.11460257e+00 7.26856649e-01 8.41791928e-01
7.21692920e-01 -6.47447050e-01 2.41716012e-01 4.21055108e-01
-8.56377482e-01 1.68130457e-01 -5.44516683e-01 -2.82987133e-02
3.72966945e-01 7.98788249e-01 -2.67231882e-01 5.06575465e-01
7.53242493e-01 3.62135619e-01 1.01030745e-01 1.10511136e+00
7.86818415e-02 5.15187562e-01 -9.72739637e-01 4.68001723e-01
-5.74044771e-02 -4.30862218e-01 7.93612421e-01 8.80287349e-01
8.08750987e-01 3.03989351e-01 1.91934943e-01 1.17445862e+00
-3.27642679e-01 -5.02461612e-01 -5.36737263e-01 1.86061993e-01
7.21503556e-01 1.11636913e+00 -7.71156669e-01 -9.20343250e-02
-4.07779843e-01 9.00384963e-01 2.44058877e-01 5.08583605e-01
-7.07019210e-01 9.32863057e-02 1.05654967e+00 3.17997307e-01
3.24587822e-01 -6.04024351e-01 -5.23885250e-01 -1.17466056e+00
7.33520538e-02 -3.11307162e-01 9.14538950e-02 -1.08536029e+00
-1.59383678e+00 3.98927987e-01 3.97170454e-01 -9.86806512e-01
2.39221146e-04 -5.33650577e-01 -7.00036228e-01 1.01368260e+00
-1.33987319e+00 -1.30886805e+00 -4.05525535e-01 4.05385643e-01
4.64094073e-01 2.54184663e-01 9.18941677e-01 2.00775504e-01
-2.09710911e-01 -1.40514985e-01 5.31531751e-01 -4.80255336e-01
1.79052263e-01 -1.33386064e+00 6.91491246e-01 6.60205066e-01
2.04465643e-01 5.13982415e-01 9.93070126e-01 -6.47313952e-01
-1.45693910e+00 -1.11997485e+00 5.05032122e-01 -7.43463635e-01
3.25165391e-01 -4.28315938e-01 -9.85536277e-01 7.16244638e-01
7.57818669e-02 1.99548721e-01 4.94589150e-01 1.51154622e-02
7.23911077e-02 2.15937104e-02 -1.29608595e+00 4.27395195e-01
1.27941072e+00 -3.88056397e-01 -2.16640979e-01 6.28552318e-01
5.80380380e-01 -5.41735351e-01 -9.81476665e-01 6.34830832e-01
-6.77073328e-03 -7.82527626e-01 1.30558753e+00 -3.42832953e-01
3.23151857e-01 -2.11326763e-01 -2.63183266e-01 -1.44348729e+00
-4.58390743e-01 -5.53077817e-01 -2.70750672e-01 1.20391285e+00
7.28191286e-02 -5.71837544e-01 1.01687813e+00 1.24223220e+00
-4.11409199e-01 -6.97631538e-01 -1.18942797e+00 -8.84698927e-01
3.27172071e-01 -6.72232270e-01 9.17766929e-01 6.94059253e-01
-5.69759607e-01 -7.94296935e-02 -1.89205498e-01 3.22521627e-01
7.53804684e-01 1.19872674e-01 3.28948349e-01 -1.50971758e+00
-3.46724927e-01 -1.21124163e-01 -1.57008141e-01 -1.05452049e+00
1.84747413e-01 -7.91065037e-01 1.21026188e-01 -1.96847272e+00
2.82173008e-01 -9.45595741e-01 -2.10442320e-01 3.83647531e-01
9.15341154e-02 1.91886470e-01 1.33869186e-01 5.45061111e-01
-3.38611364e-01 1.02293241e+00 1.04355562e+00 3.38908434e-02
-1.00036979e-01 8.16696435e-02 -3.87474447e-01 8.18228364e-01
6.94892108e-01 -7.49088585e-01 -5.78412771e-01 -9.10175800e-01
3.30915391e-01 7.62746930e-02 7.37075567e-01 -1.08749866e+00
8.24249908e-02 -3.82699549e-01 4.61466998e-01 -8.14664841e-01
9.73005891e-01 -9.19660211e-01 8.97768438e-01 -9.86087695e-03
-1.07446074e-01 2.38345131e-01 2.85260439e-01 8.71082366e-01
-8.50763023e-02 -3.07398856e-01 6.65085137e-01 -2.43490860e-01
-2.93426394e-01 7.38800049e-01 -3.39945793e-01 -1.21487699e-01
9.08780932e-01 -2.09223881e-01 -1.34075031e-01 -4.11693901e-01
-7.75680542e-01 3.17268185e-02 9.74272192e-01 1.08084716e-01
5.95685303e-01 -1.38881969e+00 -6.36272609e-01 5.50615229e-02
-1.59569517e-01 4.69814450e-01 2.93349862e-01 4.34273779e-01
-4.57068801e-01 1.02779016e-01 7.26126581e-02 -7.06667662e-01
-1.02601111e+00 4.52646405e-01 4.30777699e-01 -3.32692176e-01
-8.24746549e-01 5.89481533e-01 3.27973008e-01 -5.68311810e-01
-2.63763905e-01 -4.63876814e-01 4.33922380e-01 -3.75345379e-01
-1.10159621e-01 4.48490381e-01 2.63476759e-01 -5.56710839e-01
-2.65511245e-01 4.05422360e-01 3.37223202e-01 -3.56499135e-01
1.59369457e+00 -2.60651354e-02 -1.39448017e-01 2.76371866e-01
8.28140080e-01 -2.39990223e-02 -1.81083703e+00 2.61057336e-02
-4.43653166e-01 -4.44822043e-01 3.09226543e-01 -6.40934944e-01
-9.80458140e-01 8.73171389e-01 5.53539574e-01 1.79850891e-01
1.11737120e+00 1.06017694e-01 8.01178277e-01 2.81885862e-01
4.84223485e-01 -1.03269231e+00 -2.26652920e-01 3.85754585e-01
9.27189708e-01 -9.50541735e-01 1.55887067e-01 -5.79698622e-01
-3.39248955e-01 8.43807578e-01 2.19815403e-01 -2.48819113e-01
8.41679513e-01 3.43189895e-01 -5.28912723e-01 -2.75367528e-01
-7.92966306e-01 -2.43853852e-02 3.61637324e-01 6.83459878e-01
1.89703509e-01 -9.31396801e-03 1.93456635e-02 1.57960191e-01
5.33954762e-02 6.88085929e-02 4.97774363e-01 1.12656343e+00
-2.97209144e-01 -1.14260042e+00 -3.83085489e-01 5.08743525e-01
-5.50778091e-01 -2.04224080e-01 -2.75509320e-02 5.93142509e-01
1.48302540e-01 9.39179778e-01 3.47138911e-01 1.33569315e-01
2.49850422e-01 5.62911592e-02 7.72074699e-01 -7.60973454e-01
-3.80329072e-01 -8.45344067e-02 -8.74768049e-02 -2.15236157e-01
-6.47905588e-01 -8.76323104e-01 -1.20298016e+00 -4.29945558e-01
-4.86786306e-01 9.55859050e-02 7.61911392e-01 7.57009268e-01
7.41031349e-01 6.08478546e-01 1.89731225e-01 -1.18195438e+00
-5.65875590e-01 -7.14610159e-01 -3.89724791e-01 5.54043949e-01
2.79014468e-01 -7.48651445e-01 -1.67914480e-01 4.96652983e-02] | [8.866793632507324, -3.6494011878967285] |
544f4a6c-9ef6-4f76-a967-a3dedd88e028 | radial-distortion-homography | null | null | http://openaccess.thecvf.com/content_cvpr_2015/html/Kukelova_Radial_Distortion_Homography_2015_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2015/papers/Kukelova_Radial_Distortion_Homography_2015_CVPR_paper.pdf | Radial Distortion Homography | The importance of precise homography estimation is often underestimated even though it plays a crucial role in various vision applications such as plane or planarity detection, scene degeneracy tests, camera motion classification, image stitching, and many more. Ignoring the radial distortion component in homography estimation---even for classical perspective cameras---may lead to significant errors or totally wrong estimates. In this paper, we fill the gap among the homography estimation methods by presenting two algorithms for estimating homography between two cameras with different radial distortions. Both algorithms can handle planar scenes as well as scenes where the relative motion between the cameras is a pure rotation. The first algorithm uses the minimal number of five image point correspondences and solves a nonlinear system of polynomial equations using the Groebner basis method. The second algorithm uses a non-minimal number of six image point correspondences and leads to a simple system of two quadratic equations in two unknowns and one system of six linear equations. The proposed algorithms are fast, stable, and can be efficiently used inside a RANSAC loop. | ['Jan Heller', 'Zuzana Kukelova', 'Tomas Pajdla', 'Martin Bujnak'] | 2015-06-01 | null | null | null | cvpr-2015-6 | ['image-stitching', 'homography-estimation'] | ['computer-vision', 'computer-vision'] | [ 3.09763759e-01 -3.69818002e-01 6.74799457e-02 -8.03495422e-02
-2.54786491e-01 -7.23702610e-01 5.12636781e-01 -2.92373091e-01
-2.98868895e-01 5.75124204e-01 -3.08686346e-01 -3.25489610e-01
-2.17586786e-01 -3.47684324e-01 -5.50315559e-01 -7.19755054e-01
5.26560485e-01 7.35635161e-01 2.97956556e-01 -1.07405275e-01
7.72896111e-01 6.50596142e-01 -1.40525770e+00 -4.45585072e-01
6.83674753e-01 7.48055637e-01 1.16921186e-01 9.25332069e-01
1.99233904e-01 5.46866775e-01 -2.62262672e-01 -3.99631113e-01
6.30672336e-01 -4.30635780e-01 -6.28181279e-01 6.34341955e-01
7.46926010e-01 -5.65328121e-01 -2.62963802e-01 1.20709538e+00
2.72852480e-01 6.08257502e-02 5.75688124e-01 -1.14105880e+00
-1.30442038e-01 -1.39811173e-01 -8.85518372e-01 -2.46252984e-01
7.51538634e-01 -1.42276853e-01 4.95007575e-01 -8.33169281e-01
6.84162855e-01 1.07696986e+00 7.70857334e-01 5.24183773e-02
-1.19972563e+00 -3.11514497e-01 -5.90434909e-01 2.23533466e-01
-1.42505085e+00 -4.64151025e-01 7.71458447e-01 -6.20168328e-01
7.09714174e-01 4.99679357e-01 6.68967247e-01 4.08546269e-01
2.46679276e-01 9.96674746e-02 1.08841562e+00 -5.79414129e-01
-2.15554804e-01 1.34756312e-01 1.80837855e-01 7.29293883e-01
6.33393466e-01 1.22632375e-02 -1.76648125e-02 -2.45894670e-01
1.27079570e+00 9.14927721e-02 -5.48067629e-01 -7.98492312e-01
-1.58652711e+00 7.39662826e-01 -9.93492920e-03 2.25833878e-01
1.33192390e-02 -7.35100135e-02 4.88430262e-02 1.54816896e-01
-1.04335181e-01 6.54803157e-01 8.86093304e-02 -2.02553794e-02
-8.25696766e-01 1.95296317e-01 9.83064592e-01 1.24203861e+00
8.34477365e-01 4.96734381e-02 8.15619111e-01 7.06721783e-01
1.37728080e-01 6.61019981e-01 2.94113755e-01 -1.23653626e+00
5.69779754e-01 4.07484412e-01 2.06121206e-01 -1.53442240e+00
-2.99436629e-01 -7.77026936e-02 -9.33797479e-01 2.27591991e-01
5.74404299e-01 7.11109117e-02 -3.77521366e-01 1.00578380e+00
3.71013582e-01 -2.01001376e-01 -8.17442909e-02 9.86126542e-01
4.07624245e-01 5.61216533e-01 -1.11643243e+00 -5.81119359e-01
1.29885852e+00 -9.32766557e-01 -7.36191273e-01 -7.82563463e-02
3.90730947e-01 -1.50334322e+00 4.69907373e-01 5.64243376e-01
-1.16202080e+00 -3.15942764e-01 -1.16925359e+00 -3.30209881e-01
-8.57333690e-02 1.88590586e-01 3.71966302e-01 4.15527046e-01
-1.04395580e+00 5.17720640e-01 -3.78157020e-01 -4.28517222e-01
-5.41236460e-01 6.42436922e-01 -7.00290442e-01 -5.52407978e-03
-5.04694104e-01 1.10602927e+00 2.65292078e-01 2.31147379e-01
-6.94671944e-02 -1.07896060e-01 -8.46230924e-01 -1.44427121e-01
4.71372038e-01 -9.32285428e-01 9.21434402e-01 -6.71653211e-01
-1.70731080e+00 9.90802169e-01 -3.28021854e-01 8.28012265e-03
9.25142944e-01 -1.61244661e-01 -1.30727395e-01 1.57625660e-01
3.01741157e-02 2.81268001e-01 9.06373203e-01 -1.39477909e+00
-3.04214329e-01 -4.12378818e-01 -6.59164116e-02 5.11106491e-01
3.98910642e-01 -8.04095715e-02 -6.38074577e-01 -2.91449249e-01
8.31766903e-01 -1.32352066e+00 -2.10837603e-01 -6.59068525e-02
-5.02885461e-01 4.25152928e-01 8.71954143e-01 -6.95803523e-01
9.41264033e-01 -2.09852815e+00 2.61989355e-01 3.36309731e-01
7.89585412e-02 8.04818794e-02 2.63456255e-01 5.30801237e-01
-2.19052002e-01 -3.67563516e-01 -2.19042063e-01 -3.71729843e-02
-4.30611670e-01 1.65076613e-01 -3.55520509e-02 1.04964650e+00
-2.90557742e-01 2.49247879e-01 -6.98304057e-01 -4.19188589e-01
6.37060225e-01 4.79256868e-01 -2.60262668e-01 4.00933549e-02
4.71081734e-01 5.50961971e-01 -1.46085426e-01 3.30045491e-01
1.00052822e+00 1.12638213e-01 2.73062643e-02 -4.92659748e-01
-6.22469902e-01 -5.30107096e-02 -1.75606596e+00 1.18165576e+00
-1.21226773e-01 9.33699965e-01 2.17255101e-01 -7.67654002e-01
9.95250046e-01 3.18823397e-01 6.65656269e-01 -1.91290244e-01
3.94852817e-01 3.34563196e-01 -1.56953419e-03 -4.01204646e-01
7.45037556e-01 -8.51015598e-02 3.33520055e-01 3.93834673e-02
-2.10825279e-01 -7.32760310e-01 3.16888124e-01 -8.27706456e-02
5.82620621e-01 -1.98559254e-01 8.14139009e-01 -2.62674153e-01
8.87996495e-01 6.04778267e-02 4.79071319e-01 3.21591467e-01
6.62577823e-02 9.15218949e-01 5.50853908e-01 -5.78497112e-01
-1.41127288e+00 -6.20942593e-01 -2.99991339e-01 -2.17399985e-01
6.97466373e-01 -2.29725689e-01 -6.22339785e-01 1.20914102e-01
-1.30982399e-01 2.47128263e-01 -7.14693964e-02 1.05491281e-01
-7.23973036e-01 -3.38714749e-01 1.46671170e-02 5.79601480e-03
6.81499004e-01 -2.99644291e-01 -7.06595123e-01 -1.37410253e-01
-2.25802690e-01 -1.37425470e+00 -6.05372488e-01 -1.85515061e-01
-1.13883007e+00 -1.41386187e+00 -8.13510299e-01 -7.12737381e-01
1.05130589e+00 9.80551720e-01 7.88928866e-01 -1.33471176e-01
-1.87708572e-01 3.84314895e-01 -1.39388386e-02 1.17636256e-01
-2.87417769e-01 -4.39641416e-01 1.82594866e-01 -2.70925220e-02
1.38133079e-01 -3.10937434e-01 -4.44716126e-01 1.05726182e+00
-6.30370080e-01 2.98410072e-03 3.07582170e-01 7.88144708e-01
5.42031169e-01 1.58454254e-01 -4.66590196e-01 -7.54983842e-01
3.12307864e-01 1.63105801e-01 -1.24172640e+00 9.63416249e-02
-4.59803969e-01 3.97334285e-02 6.29835367e-01 -2.83065528e-01
-8.71082306e-01 4.46529269e-01 2.40919545e-01 -5.73814809e-01
2.12214142e-03 1.42083436e-01 -1.35088459e-01 -6.79459989e-01
5.06724715e-01 1.61480591e-01 8.90258253e-02 -3.60804886e-01
2.65615433e-01 4.09539491e-01 9.90459383e-01 -8.05580616e-02
1.16962826e+00 6.26803041e-01 6.27179503e-01 -1.34573889e+00
-9.26915705e-02 -1.16735554e+00 -9.79827523e-01 -2.19532833e-01
8.83478582e-01 -6.36486709e-01 -8.14454556e-01 4.93310571e-01
-1.49731791e+00 4.35117841e-01 6.46305457e-02 8.36110711e-01
-6.55352235e-01 1.07107282e+00 -4.12175000e-01 -6.82973564e-01
-1.27171213e-02 -1.60425258e+00 8.81593943e-01 1.57274023e-01
-9.56928805e-02 -9.22327757e-01 2.17216030e-01 4.87438649e-01
-3.92513461e-02 2.18515068e-01 4.44051236e-01 5.96694574e-02
-8.02350521e-01 -5.14564097e-01 -2.01271623e-01 4.75115851e-02
7.67676607e-02 3.51633221e-01 -6.06164932e-01 -3.23773593e-01
5.77564597e-01 3.26438785e-01 3.90733033e-01 4.42457318e-01
5.77868581e-01 -3.15034211e-01 -1.71749353e-01 1.01003015e+00
1.60995400e+00 5.64129591e-01 7.39028335e-01 4.48076695e-01
9.85829115e-01 7.17693150e-01 5.01887023e-01 2.41775408e-01
8.38683099e-02 9.78166521e-01 4.96509463e-01 -1.67580750e-02
2.59790003e-01 1.52783301e-02 1.72276855e-01 1.22654092e+00
-4.40696537e-01 1.69684827e-01 -6.98005795e-01 3.18320304e-01
-1.80704868e+00 -1.03999496e+00 -9.35172260e-01 2.82698512e+00
2.59689897e-01 -2.25614786e-01 -3.27955522e-02 2.96891868e-01
8.22001696e-01 -6.04018010e-02 -2.42242888e-01 -6.46300912e-01
-9.05071795e-02 -5.16330898e-01 9.87809896e-01 9.50408518e-01
-9.86937165e-01 6.47233427e-01 6.54170942e+00 3.14880252e-01
-1.10867167e+00 -2.52168864e-01 -8.51876289e-02 3.87499541e-01
-3.35145704e-02 6.26002252e-01 -8.57478976e-01 3.14342350e-01
1.90008804e-01 -1.04651526e-02 3.91428739e-01 7.87040532e-01
-4.51334491e-02 -6.26261115e-01 -1.00778818e+00 1.67235339e+00
4.91775215e-01 -1.00260425e+00 -7.14614615e-02 3.67696255e-01
9.03915405e-01 -3.39387715e-01 -8.21685493e-02 -6.08384550e-01
-2.81337172e-01 -5.02685487e-01 4.52429622e-01 2.82436609e-01
6.20288670e-01 -4.79739308e-01 7.73021638e-01 4.64906454e-01
-1.06248486e+00 2.96953708e-01 -7.45898545e-01 -1.54601872e-01
4.89932507e-01 6.28190219e-01 -7.86284387e-01 6.70436978e-01
2.27460727e-01 5.58285773e-01 -2.11021572e-01 1.39534760e+00
1.16543300e-01 -1.72017589e-01 -4.49110210e-01 3.00106466e-01
9.83781219e-02 -1.16274369e+00 9.11986172e-01 8.65879774e-01
6.16284430e-01 3.74342769e-01 -1.20242618e-01 4.59060639e-01
3.72170717e-01 1.76823363e-01 -1.10289109e+00 3.83561432e-01
2.48064306e-02 1.19943619e+00 -8.75130296e-01 -1.49029478e-01
-4.90279645e-01 1.19414437e+00 -3.33820432e-01 3.23121041e-01
-4.77083713e-01 -4.83258963e-01 2.47077152e-01 1.70763582e-01
-9.88506377e-02 -7.13035345e-01 -2.39632651e-01 -1.60557592e+00
3.71503644e-02 -9.19382453e-01 -3.28930430e-02 -8.85969698e-01
-6.63915873e-01 4.96732086e-01 3.00477445e-01 -1.66315675e+00
-4.60745007e-01 -9.13359582e-01 -5.92801034e-01 7.08882213e-01
-1.03887951e+00 -7.73296356e-01 -4.60929215e-01 6.99822068e-01
5.82927465e-01 -3.66162620e-02 4.92997289e-01 2.96250522e-01
-4.23272759e-01 1.15056142e-01 2.85087258e-01 -9.47172046e-02
8.01578522e-01 -1.14556408e+00 4.80933376e-02 9.85456884e-01
-1.71335518e-01 7.63729274e-01 1.01659858e+00 -4.77068096e-01
-1.61411285e+00 -1.78396627e-01 1.14309740e+00 -3.56332451e-01
5.19071698e-01 5.90772461e-03 -7.00093687e-01 6.49792194e-01
1.97068036e-01 -2.42413819e-01 2.68199772e-01 -3.74939531e-01
-1.28760427e-01 -8.00032839e-02 -9.55087841e-01 4.68964696e-01
6.48214042e-01 -4.52930063e-01 -4.34083819e-01 3.51879418e-01
1.41488865e-01 -7.84685731e-01 -7.30960190e-01 2.39924356e-01
7.46474862e-01 -1.30331659e+00 1.02733552e+00 1.41632721e-01
1.46655604e-01 -6.44905984e-01 -1.17492735e-01 -9.91546512e-01
-2.80214071e-01 -9.52376902e-01 2.85750806e-01 8.16988170e-01
-8.98339897e-02 -7.65584528e-01 6.23843491e-01 3.98754567e-01
3.06924190e-02 -2.51825005e-01 -7.64812171e-01 -7.43117213e-01
-5.66956699e-01 1.35494873e-01 1.38583243e-01 1.27775097e+00
5.31730652e-02 4.84267056e-01 -7.24925816e-01 2.96148837e-01
9.58827496e-01 1.47273049e-01 1.24935842e+00 -1.29334199e+00
-3.83081645e-01 -4.61952150e-01 -8.39297652e-01 -1.38598192e+00
-1.95927978e-01 -2.83545345e-01 -1.06344640e-01 -1.15781212e+00
7.63423890e-02 -1.08873412e-01 7.95796633e-01 -2.23456368e-01
3.42395693e-01 7.89850280e-02 1.36273772e-01 4.90296185e-01
-2.56070681e-02 9.62343886e-02 1.20025742e+00 1.54638305e-01
-1.40430212e-01 2.64057934e-01 -7.29470775e-02 1.16362977e+00
4.15967047e-01 -1.97046921e-01 -4.49402183e-01 -5.43365359e-01
3.66202772e-01 4.75868404e-01 2.56363690e-01 -9.26035941e-01
5.62651038e-01 -1.40963525e-01 2.01643392e-01 -8.14772129e-01
5.07848799e-01 -1.17820179e+00 7.07366467e-01 3.49824786e-01
3.29518855e-01 6.55368090e-01 -2.30961576e-01 2.07315788e-01
-4.84907329e-01 -7.60015488e-01 8.74600470e-01 -3.07808787e-01
-4.76155251e-01 -3.46610956e-02 -1.54065490e-01 -5.19190192e-01
1.11615729e+00 -8.60095859e-01 -2.79416889e-01 -6.13418102e-01
-4.56203133e-01 -1.40078187e-01 9.74322677e-01 3.00133266e-02
6.96748137e-01 -1.25780773e+00 -2.59636581e-01 4.80816394e-01
-6.73865750e-02 1.39527753e-01 9.45860818e-02 1.25646996e+00
-1.47111034e+00 6.76273048e-01 -2.67995238e-01 -9.42497492e-01
-1.56802607e+00 5.66124320e-01 3.75012547e-01 4.73374836e-02
-3.73716176e-01 3.94573450e-01 3.13977391e-01 -2.35347494e-01
-2.35410202e-02 -2.23109454e-01 -1.31054267e-01 -2.29749560e-01
1.77916065e-01 9.42057014e-01 -1.24380320e-01 -1.32512689e+00
-1.75307944e-01 1.54097259e+00 6.31796122e-02 -1.66560650e-01
8.75244200e-01 -3.13649803e-01 -4.95755970e-01 3.78861487e-01
1.35046446e+00 2.52595454e-01 -9.13320363e-01 1.31276473e-01
-2.59245098e-01 -9.95945394e-01 -1.39275163e-01 9.43824351e-02
-7.31002808e-01 9.52007174e-01 3.16070974e-01 2.31784746e-01
9.28302646e-01 -3.58108848e-01 4.75831151e-01 4.93780285e-01
4.13999170e-01 -7.75732815e-01 -2.12943032e-01 5.81115127e-01
8.29271317e-01 -1.15005338e+00 4.98896152e-01 -8.56512964e-01
-4.49491739e-01 1.48191071e+00 4.56039280e-01 -3.13924521e-01
4.25474256e-01 -5.53825721e-02 -7.37906843e-02 1.30888879e-01
-2.25281909e-01 5.44074662e-02 4.16155815e-01 3.84383917e-01
4.36526835e-01 -2.17490435e-01 -6.68833554e-01 -5.30744612e-01
-2.82777518e-01 -3.37122917e-01 9.86656249e-01 6.49336100e-01
-4.96191174e-01 -1.26720822e+00 -9.98460889e-01 -6.16491353e-03
-6.00122474e-02 2.59000868e-01 -5.42516887e-01 9.37146962e-01
-4.76383381e-02 7.68915713e-01 1.31629622e-02 -2.57603168e-01
4.45288181e-01 -4.26517010e-01 6.81414962e-01 -9.47603732e-02
-2.01726034e-01 6.03300929e-01 -8.33492279e-02 -5.82683086e-01
-5.57997763e-01 -7.39381909e-01 -7.47537673e-01 -6.92035794e-01
-6.68495357e-01 1.33887410e-01 8.45235050e-01 8.06077480e-01
-1.63227171e-01 -2.91014701e-01 6.82250857e-01 -1.11153436e+00
-5.41696370e-01 -5.39671898e-01 -7.00819731e-01 3.05024654e-01
4.92684960e-01 -5.73532164e-01 -6.29171789e-01 2.13959619e-01] | [8.018314361572266, -2.3238368034362793] |
f810dd61-228f-448b-8c74-df1a9c2e668b | forecasting-human-dynamics-from-static-images | 1704.03432 | null | http://arxiv.org/abs/1704.03432v1 | http://arxiv.org/pdf/1704.03432v1.pdf | Forecasting Human Dynamics from Static Images | This paper presents the first study on forecasting human dynamics from static
images. The problem is to input a single RGB image and generate a sequence of
upcoming human body poses in 3D. To address the problem, we propose the 3D Pose
Forecasting Network (3D-PFNet). Our 3D-PFNet integrates recent advances on
single-image human pose estimation and sequence prediction, and converts the 2D
predictions into 3D space. We train our 3D-PFNet using a three-step training
strategy to leverage a diverse source of training data, including image and
video based human pose datasets and 3D motion capture (MoCap) data. We
demonstrate competitive performance of our 3D-PFNet on 2D pose forecasting and
3D pose recovery through quantitative and qualitative results. | ['Yu-Wei Chao', 'Brian Price', 'Jia Deng', 'Scott Cohen', 'Jimei Yang'] | 2017-04-11 | forecasting-human-dynamics-from-static-images-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Chao_Forecasting_Human_Dynamics_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Chao_Forecasting_Human_Dynamics_CVPR_2017_paper.pdf | cvpr-2017-7 | ['human-dynamics'] | ['computer-vision'] | [ 1.50415853e-01 8.76078382e-02 -1.01360507e-01 -2.49019668e-01
-5.20880640e-01 -3.21865708e-01 4.10906643e-01 -1.01216269e+00
-5.53059757e-01 5.60891688e-01 5.92748165e-01 1.97100103e-01
4.87940639e-01 -2.68347144e-01 -9.00095046e-01 -1.37230933e-01
-3.28465551e-01 8.50311875e-01 1.57179549e-01 -4.68995571e-01
-2.71406859e-01 6.33573949e-01 -1.36867750e+00 1.56867504e-01
3.40853259e-02 1.00578880e+00 -7.58914575e-02 1.12999916e+00
6.64687276e-01 5.45025349e-01 -3.79438162e-01 -1.42146796e-01
7.93787837e-01 -5.04093826e-01 -7.51333892e-01 2.66169965e-01
5.71227908e-01 -7.96880305e-01 -7.40190089e-01 3.35923374e-01
9.89004076e-01 2.33080566e-01 5.53178728e-01 -1.41759026e+00
-4.14007198e-04 -7.35351816e-02 -3.11208904e-01 3.13418508e-02
1.05501604e+00 7.05752015e-01 4.61323887e-01 -9.62300718e-01
1.02423489e+00 1.51400924e+00 1.13061547e+00 1.16605902e+00
-1.01178873e+00 -4.49297279e-01 2.89064422e-02 -6.85848817e-02
-1.17153156e+00 -1.67102590e-01 6.50757134e-01 -7.17926800e-01
1.17329597e+00 -1.12819768e-01 1.49530828e+00 1.73178136e+00
5.14562786e-01 9.68040407e-01 7.94148386e-01 -2.17209011e-01
-1.80457458e-01 -8.79591644e-01 -3.71388018e-01 8.80003393e-01
-3.71378772e-02 5.52265882e-01 -9.55232441e-01 1.14888921e-01
1.26990032e+00 -5.70803210e-02 -2.87032574e-01 -6.30957842e-01
-1.41111803e+00 2.99860656e-01 4.06653851e-01 -3.29448998e-01
-4.71958101e-01 6.74978673e-01 2.48421267e-01 -1.21423781e-01
3.65430504e-01 1.62375599e-01 -6.37508392e-01 -3.56593519e-01
-8.28081191e-01 1.07253313e+00 7.39009857e-01 8.84774327e-01
1.26718372e-01 3.25614251e-02 -4.88657504e-02 2.35282049e-01
5.39016128e-01 9.11724508e-01 5.74320376e-01 -1.26016903e+00
5.64209878e-01 2.57065117e-01 3.17103475e-01 -1.09242761e+00
-9.13930953e-01 -1.41614303e-01 -4.89972889e-01 1.58001348e-01
4.57023114e-01 -3.00784379e-01 -1.37854004e+00 1.87016892e+00
4.70966786e-01 -7.55987763e-02 -3.42115283e-01 1.55782127e+00
8.31449687e-01 4.30166811e-01 1.79869980e-02 2.57167339e-01
8.92642677e-01 -1.06928456e+00 -2.67856479e-01 -5.27862430e-01
5.06004751e-01 -1.72325432e-01 7.95351684e-01 4.00044411e-01
-1.24275696e+00 -8.38008046e-01 -8.49543810e-01 -2.06101060e-01
1.24436274e-01 9.37656462e-02 4.10108447e-01 3.91428560e-01
-9.77808774e-01 8.17634463e-01 -1.24014080e+00 -4.93040860e-01
1.54481858e-01 5.28720856e-01 -7.50904620e-01 2.40943953e-01
-1.22030938e+00 9.88820612e-01 2.80805111e-01 4.79035050e-01
-1.04525912e+00 -3.56902868e-01 -1.05498695e+00 -5.87271154e-01
2.83042848e-01 -1.45557821e+00 1.42539632e+00 -4.97059703e-01
-1.62129867e+00 1.13618290e+00 -2.33487152e-02 -5.59428155e-01
9.94636476e-01 -1.00103509e+00 1.29149361e-02 1.42645732e-01
1.03058061e-02 1.10453200e+00 7.45009542e-01 -1.07107127e+00
-2.26046965e-01 -5.99851668e-01 -3.98294955e-01 5.89252949e-01
3.35863531e-01 -4.42969620e-01 -7.60479569e-01 -7.90367246e-01
4.15489852e-01 -1.45741844e+00 -4.13285673e-01 1.75403073e-01
-5.84104776e-01 1.75621524e-01 2.83805788e-01 -9.09773827e-01
8.44056904e-01 -1.35177875e+00 8.35386276e-01 6.28894866e-02
3.32506061e-01 1.92233026e-01 8.63283686e-03 1.09701678e-01
2.09082998e-02 -2.07578152e-01 6.53789267e-02 -6.30915463e-01
5.93610853e-02 2.69064158e-01 -1.23243578e-01 4.72196877e-01
3.02418862e-02 1.34913909e+00 -7.94194221e-01 -4.95062411e-01
3.73527348e-01 5.49264967e-01 -8.83947253e-01 4.13097978e-01
-3.09830040e-01 8.94106269e-01 -3.09117585e-01 6.91011488e-01
1.12689987e-01 -3.07653993e-01 1.24834143e-01 -4.29060936e-01
2.12834820e-01 8.89524817e-02 -9.08946037e-01 2.33737206e+00
7.25148767e-02 3.05005997e-01 -5.07353961e-01 -4.47382957e-01
6.54606104e-01 2.79742897e-01 9.96132016e-01 -5.11944294e-01
4.72316116e-01 1.17874302e-01 -1.73238680e-01 -6.30128741e-01
4.62537110e-01 -3.37239295e-01 -4.25012678e-01 3.70672911e-01
3.71844143e-01 5.13904868e-03 -2.60486394e-01 -1.80178031e-01
1.03202152e+00 1.30613685e+00 2.65490413e-01 2.16303304e-01
1.64192274e-01 1.89924359e-01 5.69048822e-01 4.59986299e-01
-6.78827822e-01 1.11855423e+00 1.09343842e-01 -9.91566896e-01
-1.29941225e+00 -1.38904738e+00 6.96953416e-01 8.80612493e-01
-1.28624486e-02 -3.84321600e-01 -7.65550673e-01 -5.58934748e-01
2.60154694e-01 -1.98924467e-01 -8.04277956e-01 -1.90387398e-01
-1.18456435e+00 -2.78913558e-01 7.06905127e-01 9.42141116e-01
5.04993856e-01 -1.02992141e+00 -1.19670868e+00 1.19453952e-01
-5.81477582e-01 -1.17205679e+00 -6.39718294e-01 -8.53541419e-02
-8.76156271e-01 -1.16067350e+00 -9.48735297e-01 -7.16156900e-01
4.78523016e-01 -1.97680935e-01 1.23795009e+00 -1.78907350e-01
-3.66333127e-01 6.13858759e-01 -1.73319712e-01 -1.89379528e-01
-4.40623946e-02 5.31704538e-03 7.84312963e-01 -4.62153465e-01
1.28164545e-01 -5.06065726e-01 -7.62312353e-01 3.94784302e-01
-1.46886975e-01 4.27231878e-01 4.38934028e-01 7.69092739e-01
8.80721748e-01 -8.16154540e-01 5.39193004e-02 -2.59782314e-01
2.14391097e-01 -4.42665592e-02 -1.58616394e-01 -2.15095337e-02
-1.22741781e-01 -6.96877670e-03 1.75824910e-01 -5.19582033e-01
-8.77155006e-01 9.40806866e-01 -5.07929385e-01 -8.20113301e-01
-2.15102822e-01 1.61441162e-01 -3.06116641e-02 2.51205433e-02
8.90294254e-01 1.85819551e-01 3.19299340e-01 -4.30497646e-01
4.16241556e-01 9.33211446e-02 1.27828848e+00 -4.09667492e-01
8.38340044e-01 4.77516413e-01 1.84841871e-01 -2.85008878e-01
-8.54331076e-01 -2.61781871e-01 -1.40328681e+00 -7.42633641e-01
1.20258737e+00 -1.30897582e+00 -9.69307482e-01 8.12463403e-01
-1.12471223e+00 -7.59532213e-01 -1.04156494e-01 6.04710460e-01
-1.37792802e+00 1.98525593e-01 -7.30612397e-01 -8.31610918e-01
-5.12151957e-01 -9.02722120e-01 1.52737880e+00 -1.89294115e-01
-7.41964102e-01 -6.79607511e-01 4.57137406e-01 2.88679689e-01
-1.90552711e-01 1.03278482e+00 1.29555598e-01 -1.35144114e-01
-3.11401606e-01 -3.36044252e-01 4.72245693e-01 2.13575773e-02
-2.23740697e-01 -5.23970544e-01 -5.60707629e-01 -3.20364684e-01
-2.49180242e-01 -7.74476349e-01 6.48889422e-01 7.34057307e-01
7.79374361e-01 -7.98895583e-02 -3.14109504e-01 1.01203835e+00
8.35447013e-01 -2.75754660e-01 5.40263414e-01 3.98640096e-01
1.11902690e+00 4.93188977e-01 7.26311982e-01 5.86660624e-01
6.99173272e-01 9.79765534e-01 3.00376892e-01 9.05220881e-02
-4.32714224e-01 -9.76710320e-01 4.56685781e-01 6.97878361e-01
-6.26802981e-01 -6.37486717e-03 -1.15138125e+00 2.11075783e-01
-1.98869669e+00 -8.60454023e-01 3.11528802e-01 1.92151034e+00
6.82331264e-01 2.65896112e-01 8.36015582e-01 -3.13760787e-02
4.87646252e-01 -5.92891276e-02 -8.31371069e-01 2.70825446e-01
1.50312021e-01 -2.15294547e-02 3.81736040e-01 2.77848661e-01
-1.10455346e+00 9.95034337e-01 7.10108805e+00 3.05025689e-02
-1.05583870e+00 -2.17947438e-01 3.37813586e-01 -6.73290133e-01
1.48661166e-01 -3.93927306e-01 -9.18638349e-01 1.67400390e-01
7.51121104e-01 -4.21820171e-02 4.36107703e-02 7.49415934e-01
2.84898400e-01 7.64595866e-02 -1.30124807e+00 1.36461818e+00
2.77422518e-01 -9.78784621e-01 1.89859550e-02 1.42190561e-01
6.30930722e-01 8.10376406e-02 -4.02749591e-02 1.84070751e-01
2.62822837e-01 -1.22746038e+00 1.16407275e+00 8.48632455e-01
8.88088942e-01 -5.05194008e-01 4.82916117e-01 5.88149428e-01
-1.19758010e+00 -7.65764937e-02 1.42797336e-01 -3.96334857e-01
7.91461766e-01 -5.09757362e-03 -6.68045104e-01 3.10218334e-01
1.05203974e+00 9.96060967e-01 -4.25714999e-01 8.51814389e-01
-2.41102010e-01 1.21143177e-01 -4.32599366e-01 2.09121943e-01
-1.19527839e-01 2.78616309e-01 5.56044042e-01 7.56858945e-01
3.87345880e-01 2.79337913e-01 4.48121071e-01 6.27614439e-01
1.10781260e-01 -4.87434238e-01 -6.67801738e-01 1.81919053e-01
1.29036158e-01 8.44498932e-01 -5.31654954e-01 -2.42449999e-01
1.84651300e-01 1.26665294e+00 1.41559005e-01 1.34700462e-01
-8.59003365e-01 1.88760817e-01 6.83633745e-01 2.67471969e-01
1.85366943e-01 -7.31184065e-01 -1.71205446e-01 -1.32476187e+00
-1.10385325e-02 -8.52306187e-01 2.64834195e-01 -1.06177580e+00
-1.07648969e+00 4.01568770e-01 3.15649539e-01 -1.48602021e+00
-1.03320742e+00 -8.30458760e-01 3.20872776e-02 6.26531243e-01
-7.40108430e-01 -1.25421798e+00 -4.44639117e-01 6.80756569e-01
4.33394045e-01 1.00645669e-01 6.87471509e-01 -1.12843039e-02
-1.74217671e-01 4.19717491e-01 -7.78587222e-01 4.17982101e-01
6.40891433e-01 -1.10467279e+00 1.19681919e+00 5.77667713e-01
1.06276564e-01 4.30252135e-01 8.04321289e-01 -9.66838658e-01
-1.67872095e+00 -8.34267020e-01 7.25260675e-01 -1.28465378e+00
1.15169451e-01 -4.99228060e-01 -1.19248882e-01 1.00033295e+00
-5.47509968e-01 2.24230558e-01 3.41598988e-01 -2.32252553e-01
-1.83078393e-01 4.02038574e-01 -9.73648727e-01 6.65045798e-01
1.97285950e+00 -3.32866132e-01 -7.52054155e-01 1.94235623e-01
6.38517737e-01 -1.20516467e+00 -9.96338546e-01 6.99162006e-01
1.28283632e+00 -7.96352148e-01 1.37001061e+00 -7.43315339e-01
6.61455035e-01 -1.99381888e-01 -1.83386162e-01 -1.27849889e+00
-2.05769017e-01 -6.44099057e-01 -5.88886023e-01 9.37452074e-03
-8.16401653e-03 1.42502636e-01 1.43719482e+00 5.70589364e-01
-1.04650520e-01 -7.61612952e-01 -9.00128007e-01 -7.82139719e-01
1.53224215e-01 -5.51925480e-01 3.13843340e-01 8.38686228e-02
-1.23854429e-01 9.32526439e-02 -1.04884219e+00 -1.37043506e-01
7.22564340e-01 2.32198909e-02 1.40020919e+00 -1.06683576e+00
-4.73367333e-01 9.97915715e-02 -7.65518486e-01 -1.62209415e+00
3.99795920e-02 -5.39758325e-01 6.18671298e-01 -1.38450325e+00
6.22389428e-02 3.32397670e-01 2.57870585e-01 4.72582608e-01
-1.06616691e-01 5.52675307e-01 4.96902883e-01 3.70556265e-01
-7.36285925e-01 6.40274286e-01 1.48927176e+00 2.58195639e-01
-2.35508487e-01 3.16174440e-02 7.95343593e-02 8.41496050e-01
4.06653464e-01 -2.24770442e-01 -2.72093713e-01 -3.95936906e-01
1.31639272e-01 4.37147677e-01 9.08976316e-01 -1.43159795e+00
1.08514644e-01 -4.79683317e-02 1.17053795e+00 -1.13339102e+00
6.04634464e-01 -4.73023683e-01 5.12471914e-01 9.71202195e-01
-3.17793936e-01 3.28926027e-01 4.50475477e-02 6.44032717e-01
2.35816151e-01 7.03826666e-01 2.72176415e-01 -6.37923181e-01
-1.09782326e+00 7.15457797e-01 -1.17893696e-01 1.95550710e-01
7.24292755e-01 -5.87151945e-01 1.35362953e-01 -5.74065566e-01
-1.32085562e+00 2.26521090e-01 5.79405308e-01 6.50467694e-01
8.87295902e-01 -1.77179658e+00 -3.58086437e-01 2.63080537e-01
-2.01668794e-04 1.59922913e-01 2.23670676e-01 6.61273479e-01
-6.95747435e-01 5.29580355e-01 -7.05418348e-01 -9.22811091e-01
-1.10183275e+00 1.77809775e-01 7.00043976e-01 -2.44337857e-01
-9.37150300e-01 9.70863402e-01 -2.28747968e-02 -9.29692805e-01
1.86460927e-01 -2.24393234e-01 1.55123949e-01 -5.18799782e-01
1.81831479e-01 2.81625986e-01 -2.97263831e-01 -1.22143006e+00
-4.76240516e-01 7.97650993e-01 4.27578181e-01 -5.92684805e-01
1.33076310e+00 -1.99495301e-01 4.68008041e-01 5.27408183e-01
1.20727694e+00 -6.94295168e-01 -2.03587198e+00 -1.53170051e-02
-2.50705123e-01 -4.74588990e-01 -4.85520124e-01 -8.64056766e-01
-8.82547915e-01 7.47136116e-01 7.39157796e-01 -7.38212824e-01
8.94127786e-01 -3.41760218e-02 1.22441208e+00 3.77807617e-01
8.36666584e-01 -1.22041881e+00 6.50179625e-01 7.14789629e-01
9.45125222e-01 -1.09288001e+00 2.21935272e-01 -2.20647633e-01
-6.16921365e-01 1.07672179e+00 1.02135265e+00 -3.76019508e-01
6.10968411e-01 6.44912496e-02 2.14424342e-01 -2.68568635e-01
-6.10549092e-01 -2.61355489e-01 7.29782879e-01 7.51352489e-01
3.57730180e-01 1.38582122e-02 2.02197637e-02 8.14329326e-01
-6.69466674e-01 5.67228496e-01 1.82415053e-01 1.13283873e+00
-2.51988083e-01 -8.51288021e-01 -4.46873248e-01 8.73274356e-03
-2.81030923e-01 4.24953401e-01 -5.90660632e-01 9.57585514e-01
1.89370960e-01 3.46505493e-01 -5.07507436e-02 -1.11963975e+00
6.42155707e-01 1.46570161e-01 9.97088969e-01 -5.38157701e-01
-4.08442289e-01 1.24151267e-01 1.30416714e-02 -1.19714010e+00
-6.50447309e-01 -6.62929952e-01 -1.39481020e+00 -3.56183171e-01
1.32807240e-01 -6.19771063e-01 5.22056878e-01 1.01877129e+00
2.86177814e-01 3.79568815e-01 9.37236995e-02 -1.93123770e+00
-5.46133399e-01 -8.14777553e-01 -3.00657809e-01 7.31394053e-01
4.56239581e-01 -9.68148530e-01 7.25376159e-02 3.39973867e-01] | [7.12231969833374, -0.6259563565254211] |
b75f5f43-1a8b-4767-ad73-e854f2b647a1 | towards-a-privacy-preserving-deep-learning | 2106.06765 | null | https://arxiv.org/abs/2106.06765v1 | https://arxiv.org/pdf/2106.06765v1.pdf | Towards a Privacy-preserving Deep Learning-based Network Intrusion Detection in Data Distribution Services | Data Distribution Service (DDS) is an innovative approach towards communication in ICS/IoT infrastructure and robotics. Being based on the cross-platform and cross-language API to be applicable in any computerised device, it offers the benefits of modern programming languages and the opportunities to develop more complex and advanced systems. However, the DDS complexity equally increases its vulnerability, while the existing security measures are limited to plug-ins and static rules, with the rest of the security provided by third-party applications and operating system. Specifically, traditional intrusion detection systems (IDS) do not detect any anomalies in the publish/subscribe method. With the exponentially growing global communication exchange, securing DDS is of the utmost importance to futureproofing industrial, public, and even personal devices and systems. This report presents an experimental work on the simulation of several specific attacks against DDS, and the application of Deep Learning for their detection. The findings show that even though Deep Learning allows to detect all simulated attacks using only metadata analysis, their detection level varies, with some of the advanced attacks being harder to detect. The limitations imposed by the attempts to preserve privacy significantly decrease the detection rate. The report also reviews the drawbacks and limitations of the Deep Learning approach and proposes a set of selected solutions and configurations, that can further improve the DDS security. | ['Stanislav Abaimov'] | 2021-06-12 | null | null | null | null | ['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning'] | ['methodology', 'natural-language-processing'] | [-2.45997801e-01 1.65451039e-02 -1.23670176e-01 -1.25636384e-01
5.23097329e-02 -6.87214136e-01 9.00586188e-01 3.97257149e-01
-3.31463039e-01 5.39024293e-01 -4.40548152e-01 -7.62092590e-01
-4.59199101e-01 -1.02412188e+00 -4.19308335e-01 -7.05595553e-01
-5.30919552e-01 5.18066645e-01 6.29887879e-01 -2.58966804e-01
1.64214939e-01 1.18087649e+00 -1.56246936e+00 9.35578197e-02
-2.36950349e-02 1.40622735e+00 -2.70852417e-01 5.35295665e-01
-2.67752320e-01 3.18700463e-01 -1.11929965e+00 -1.54601768e-01
3.66423100e-01 5.08029163e-01 -5.92323184e-01 -5.83541572e-01
4.70504053e-02 -4.64311272e-01 -2.35534519e-01 9.50624526e-01
5.76683164e-01 -5.73120475e-01 6.96904361e-02 -1.93875432e+00
-7.34244660e-02 4.92975444e-01 -1.19462050e-01 1.16505157e-02
6.26928031e-01 1.69813797e-01 2.00667545e-01 -7.58750737e-03
5.92790961e-01 1.11957693e+00 3.93234044e-01 4.61915702e-01
-8.58064711e-01 -1.16112125e+00 -1.37984663e-01 1.24349177e-01
-1.04903913e+00 -7.84660280e-02 6.35845959e-01 -2.08828256e-01
1.07469487e+00 4.87244546e-01 1.73271745e-01 1.40697360e+00
5.21743715e-01 2.38548219e-01 9.20075238e-01 -3.32305729e-01
6.02276206e-01 4.73402441e-01 3.62600952e-01 1.81797758e-01
8.33135128e-01 3.11720550e-01 7.07397461e-02 -5.37475467e-01
1.61636814e-01 2.18191594e-01 2.53787428e-01 -5.17798305e-01
-7.75966942e-01 6.80556536e-01 5.76781668e-02 8.93423200e-01
-3.20081711e-01 -2.52691489e-02 1.08248246e+00 6.01473212e-01
-8.43056962e-02 4.02602404e-01 -9.92572069e-01 -1.26316145e-01
-4.31683183e-01 -2.10713670e-02 1.44336104e+00 8.96976948e-01
3.61793369e-01 2.90572405e-01 4.99344885e-01 -2.43459828e-02
4.42182660e-01 4.01539922e-01 1.39553651e-01 -5.37504196e-01
5.31712510e-02 5.42645097e-01 -1.95929155e-01 -1.13081169e+00
-7.14588225e-01 -3.05546224e-01 -9.41400945e-01 7.20147789e-01
2.29966015e-01 -2.60643899e-01 -5.93418896e-01 1.30998421e+00
5.91946065e-01 -4.26388159e-02 9.48915035e-02 9.26162023e-03
3.21462244e-01 5.27311623e-01 9.58062932e-02 -1.46113962e-01
1.11825776e+00 1.26959896e-02 -1.04410851e+00 2.27468014e-01
4.00110006e-01 -7.12785065e-01 5.01418769e-01 7.03188181e-01
-6.03193998e-01 -3.03399175e-01 -1.50619268e+00 6.30469084e-01
-1.28592944e+00 -6.03288174e-01 6.61887586e-01 1.25159037e+00
-1.02864575e+00 5.72970450e-01 -8.29091787e-01 -6.09134436e-01
4.00550008e-01 1.21376526e+00 -2.08757460e-01 3.76463145e-01
-1.29385710e+00 1.00684571e+00 6.12625897e-01 -2.83525079e-01
-7.49884248e-01 -3.76053095e-01 -5.51894426e-01 3.66915017e-02
2.66052425e-01 -1.09400682e-01 8.94407451e-01 -3.03059191e-01
-1.49781394e+00 5.56448996e-01 6.66824281e-01 -9.80409622e-01
3.52682918e-01 4.82383296e-02 -1.04724789e+00 -2.38755252e-02
-3.27936292e-01 7.58062005e-02 6.31986380e-01 -1.16260612e+00
-8.29135835e-01 -2.49829888e-01 2.67028868e-01 -9.96263027e-01
-6.24758363e-01 3.34672540e-01 1.96177512e-01 3.16847712e-02
-2.72360831e-01 -7.47692406e-01 -1.30494684e-01 -2.70050671e-02
-4.29912865e-01 -2.32494235e-01 1.80923438e+00 -2.46520236e-01
9.71475959e-01 -2.24383616e+00 -4.49109465e-01 7.01327026e-01
7.82673359e-02 8.56531620e-01 2.49560952e-01 8.16609383e-01
-1.18878372e-01 1.33705422e-01 1.25043079e-01 -3.28270867e-02
2.72051692e-01 3.84895176e-01 -4.39095587e-01 5.96425712e-01
-6.36081547e-02 2.02525795e-01 -4.83859539e-01 -7.15174451e-02
7.47326553e-01 5.20134211e-01 -1.17814369e-01 3.72702815e-02
-1.84279606e-01 2.79353499e-01 -6.57852769e-01 7.65826881e-01
8.79056931e-01 1.62661284e-01 3.76554698e-01 -5.34755662e-02
-3.13849241e-01 4.08524573e-01 -1.40810668e+00 9.44049656e-01
-5.23126602e-01 4.60307747e-01 3.59171242e-01 -1.17406142e+00
1.09349954e+00 3.85073900e-01 7.04760373e-01 -6.60475612e-01
4.65170801e-01 4.47334051e-01 1.65236201e-02 -5.30909538e-01
-1.06803127e-01 4.67648417e-01 -3.20129812e-01 4.30660754e-01
-3.45069207e-02 4.45123240e-02 -2.13424210e-02 1.84317663e-01
1.39278638e+00 -2.52373248e-01 3.50239456e-01 -1.99896976e-01
9.17898059e-01 -4.75887775e-01 3.32941502e-01 6.64579690e-01
-1.63023978e-01 -2.60206938e-01 5.98678589e-01 -7.96354771e-01
-7.16280580e-01 -1.22552216e+00 -2.78011054e-01 5.04566729e-01
1.93536244e-02 -3.11407775e-01 -5.94978333e-01 -8.97402823e-01
5.08756973e-02 9.20664489e-01 -1.54258117e-01 -3.65739226e-01
-4.53008026e-01 -5.63162684e-01 8.68652821e-01 1.63336664e-01
5.64404786e-01 -1.20731163e+00 -7.37820089e-01 2.33122274e-01
6.09982312e-01 -1.28911579e+00 4.46286976e-01 5.17155468e-01
-6.28748238e-01 -1.34008634e+00 2.91662306e-01 -6.15425646e-01
1.89769775e-01 -2.20044583e-01 6.67490363e-01 6.46197051e-02
-4.72864509e-01 5.49545705e-01 -2.47071251e-01 -8.04724693e-01
-9.13481593e-01 9.29691270e-02 3.35009933e-01 -3.47253174e-01
7.18747914e-01 -7.74267137e-01 -1.06918819e-01 1.99476078e-01
-1.05763197e+00 -1.00880969e+00 4.40539777e-01 2.82985091e-01
-3.64358686e-02 6.62587643e-01 5.97902775e-01 -7.91927755e-01
4.28211242e-01 -6.16127670e-01 -9.77826536e-01 -1.03087537e-01
-8.87371123e-01 -1.97960317e-01 8.01019430e-01 -5.02966225e-01
-5.34261584e-01 -2.52572373e-02 -4.70079154e-01 9.11812298e-03
-6.84802890e-01 -2.22005114e-01 -6.34424031e-01 -5.68425179e-01
3.84560108e-01 4.49952707e-02 4.76188362e-02 -3.89819354e-01
-2.51537889e-01 9.13278580e-01 1.19325861e-01 -2.70625472e-01
9.16658044e-01 7.55820453e-01 2.06510291e-01 -1.12521303e+00
1.35709224e-02 -1.38408214e-01 -2.24794269e-01 -1.26085773e-01
6.36784673e-01 -2.44815335e-01 -1.47354484e+00 5.59375882e-01
-1.29602945e+00 2.57368088e-01 -2.60920316e-01 3.17765206e-01
-1.64277405e-01 4.00262654e-01 -5.50692320e-01 -7.84864068e-01
-4.08348531e-01 -1.20925307e+00 6.22418106e-01 -3.82569642e-03
-3.55206847e-01 -7.77360499e-01 -8.27444345e-02 -7.89823309e-02
8.06221902e-01 5.75418353e-01 1.18034828e+00 -1.42098463e+00
-6.39544427e-01 -8.10011268e-01 4.46653403e-02 6.95638120e-01
2.98625380e-01 1.60237834e-01 -9.26224470e-01 -4.25030947e-01
4.30407256e-01 1.73086032e-01 -1.50268957e-01 -1.23716377e-01
1.09776604e+00 -3.69480163e-01 -3.88172299e-01 3.86226088e-01
1.37307572e+00 9.49594140e-01 8.06796730e-01 9.04489875e-01
3.60649139e-01 6.24677002e-01 4.86507148e-01 4.78950024e-01
-1.40073195e-01 4.34978157e-01 1.16962135e+00 1.20785594e-01
3.51626724e-01 3.53609979e-01 2.83497542e-01 3.56169343e-01
4.15264159e-01 -2.27002785e-01 -8.63065481e-01 1.16413809e-01
-1.37965703e+00 -7.59074092e-01 -2.96623558e-01 2.25768328e+00
2.23991767e-01 7.35992074e-01 9.52217206e-02 6.02778494e-01
7.78456330e-01 9.44696292e-02 -4.47749048e-01 -1.02692580e+00
1.58771470e-01 5.87352991e-01 7.89175093e-01 6.70941696e-02
-1.13725090e+00 4.96021837e-01 6.20225573e+00 5.94139457e-01
-1.27581131e+00 2.11356394e-02 -8.20023194e-02 1.62857786e-01
2.06939802e-01 -2.50397563e-01 -9.37308252e-01 6.47612154e-01
1.40663111e+00 -1.51047260e-02 1.92436412e-01 1.12027788e+00
-1.35107100e-01 1.87489286e-01 -1.02118218e+00 6.31478786e-01
-2.70919651e-01 -1.28398287e+00 -1.19915120e-01 4.55790788e-01
1.32567659e-01 2.08271444e-01 8.30166638e-02 3.33634838e-02
2.35473379e-01 -5.14010608e-01 3.88202369e-01 3.77992876e-02
2.43996039e-01 -1.13294983e+00 1.23550296e+00 -3.77810560e-02
-7.88475156e-01 -4.82563674e-01 -1.84228588e-02 3.06776515e-03
-6.60359710e-02 4.29968864e-01 -8.38485181e-01 6.40952229e-01
9.70188081e-01 1.55594185e-01 -2.58258611e-01 7.47833610e-01
1.43319815e-01 3.17822725e-01 -8.09713006e-01 -4.43824857e-01
2.16159090e-01 2.50758648e-01 5.84281921e-01 1.13703191e+00
1.60716102e-01 -5.89557827e-01 -8.14890396e-03 3.23568732e-01
3.65564466e-01 -3.57310176e-01 -9.69772637e-01 -8.91855285e-02
6.71543002e-01 1.07963705e+00 -7.13729143e-01 -1.97336059e-02
-5.32842159e-01 4.09950972e-01 -5.60403764e-01 -3.54241766e-02
-5.55229068e-01 -7.20471323e-01 1.03034627e+00 3.72788399e-01
3.25772703e-01 -3.99123073e-01 -3.30825150e-01 -3.19653094e-01
-1.33796316e-02 -1.15463150e+00 5.87329745e-01 -3.04216780e-02
-1.17288446e+00 5.72241724e-01 8.95165056e-02 -1.22760415e+00
-3.21148098e-01 -8.82654965e-01 -5.48008680e-01 7.83500746e-02
-1.15871656e+00 -1.22638893e+00 2.72688031e-01 7.66742468e-01
4.21129242e-02 -5.46905875e-01 1.14515746e+00 6.66345656e-01
-2.88049251e-01 4.27402884e-01 2.22481638e-01 -7.15764910e-02
4.03142303e-01 -8.99043858e-01 3.35540146e-01 6.69206321e-01
-1.64408833e-01 7.35038996e-01 9.90869701e-01 -5.66031694e-01
-1.57795548e+00 -7.04882443e-01 5.58884740e-01 -1.84321497e-02
9.97978330e-01 -7.26434946e-01 -7.38440692e-01 6.78244829e-01
3.62410456e-01 -2.42045492e-01 4.94175464e-01 -1.28437877e-02
-3.30172390e-01 -7.05845296e-01 -1.73869920e+00 3.71021867e-01
2.99657792e-01 -5.48587143e-01 -3.59979331e-01 3.95340204e-01
8.49894345e-01 5.25388196e-02 -6.94261611e-01 4.04479057e-01
5.13961971e-01 -1.15140414e+00 1.10261536e+00 -3.35853696e-01
-4.48453039e-01 -2.79046208e-01 -2.25963384e-01 -4.68478441e-01
9.78265479e-02 -7.58338690e-01 -3.49140227e-01 1.73949194e+00
8.87878165e-02 -1.17921853e+00 7.50289738e-01 3.98829937e-01
1.42652348e-01 -2.87190050e-01 -1.32445204e+00 -1.09648764e+00
-2.37007722e-01 -6.52322829e-01 1.07755446e+00 8.78587127e-01
-2.52777964e-01 -2.60740191e-01 -6.77000061e-02 7.62206554e-01
5.69052994e-01 -4.00415748e-01 7.27133691e-01 -1.47210765e+00
-1.61309883e-01 -9.75611806e-02 -1.07367313e+00 1.58921238e-02
-4.45941649e-03 -3.37974191e-01 -4.81203794e-01 -1.04920709e+00
-8.71822178e-01 -4.42953795e-01 -5.44122040e-01 4.71024513e-01
1.17699039e+00 -1.39833748e-01 2.26423591e-02 -2.58180588e-01
-4.14608240e-01 2.15056967e-02 3.32197547e-01 -3.59000415e-01
-1.39904708e-01 4.15839761e-01 -2.06138074e-01 1.05064392e+00
1.08373201e+00 -7.27077305e-01 -3.03751945e-01 2.15936266e-02
1.51145667e-01 -4.47778344e-01 4.38347965e-01 -1.35221863e+00
4.86415029e-01 1.85756415e-01 1.76369995e-01 -7.02163875e-01
2.29430616e-01 -1.76015830e+00 1.80866718e-01 1.21587229e+00
2.30218902e-01 1.52034406e-02 3.80170286e-01 3.97576988e-01
7.77426362e-02 -1.13106228e-01 8.55640054e-01 1.64043978e-01
-6.47193670e-01 1.21866249e-01 -7.19655156e-01 -7.36273527e-01
1.63216376e+00 -2.13755727e-01 -3.04561615e-01 -1.57634035e-01
-5.74614584e-01 1.41788684e-02 3.53631705e-01 7.73393512e-01
3.54668587e-01 -7.93223143e-01 1.47211835e-01 5.76161087e-01
-1.04230449e-01 -2.92980880e-01 -1.03689127e-01 2.83903122e-01
-5.85178673e-01 4.50628132e-01 -4.40034717e-01 -3.34704161e-01
-1.43288648e+00 1.03427541e+00 4.90569696e-02 -2.07606345e-01
-4.51638162e-01 2.39902362e-01 -4.31703478e-01 -4.43709821e-01
8.02317321e-01 -1.41377553e-01 -3.34372491e-01 -3.07788383e-02
7.39799976e-01 6.06438994e-01 3.00645262e-01 -1.21337496e-01
-7.58435547e-01 1.88441202e-01 -3.55364025e-01 3.02521497e-01
1.44058144e+00 -3.48400231e-03 -4.81254786e-01 2.54792631e-01
1.20119381e+00 5.50652593e-02 -4.94999439e-01 3.13639700e-01
5.98311901e-01 -2.02289999e-01 -1.06607459e-01 -8.76576722e-01
-1.09168983e+00 6.06742978e-01 1.08801663e+00 1.05313528e+00
9.79020715e-01 -1.20435646e-02 1.04881680e+00 7.02443182e-01
8.62589240e-01 -9.92238998e-01 -2.20365226e-01 3.83183032e-01
3.19739610e-01 -8.10887635e-01 -9.71223339e-02 -4.98026222e-01
9.48221162e-02 1.37048173e+00 4.00283337e-01 -2.65467525e-01
1.06278706e+00 1.13578916e+00 -9.81264561e-02 -3.85115743e-01
-2.80207366e-01 3.64635140e-01 -5.86518705e-01 1.18176842e+00
-1.99546188e-01 -2.25219294e-01 -3.71119022e-01 2.81206280e-01
6.55429065e-02 1.03534467e-03 5.40906787e-01 1.22968638e+00
-2.58728921e-01 -1.72154021e+00 -4.29268450e-01 2.80404299e-01
-7.72570431e-01 5.35229504e-01 -5.06177843e-01 1.14756036e+00
5.44549108e-01 1.17424488e+00 6.34933403e-03 -7.23277748e-01
3.70882690e-01 -1.41813695e-01 -8.69955271e-02 -1.70346767e-01
-7.68532097e-01 -5.22575378e-01 7.92956278e-02 -9.61431026e-01
-9.97608230e-02 -5.47922790e-01 -1.28498065e+00 -6.07820213e-01
-1.22857459e-01 9.59409252e-02 1.47667575e+00 8.48726094e-01
4.32530850e-01 7.39460766e-01 8.31846774e-01 -3.58122259e-01
-6.32826090e-01 -3.33945602e-01 -6.93000495e-01 -8.77986923e-02
4.98421490e-01 -5.06004870e-01 -6.39450490e-01 -6.28512681e-01] | [5.252415180206299, 7.157679080963135] |
d25be932-963b-46dd-a593-34107a6b4266 | fully-unsupervised-feature-alignment-for | 1907.09204 | null | https://arxiv.org/abs/1907.09204v2 | https://arxiv.org/pdf/1907.09204v2.pdf | Domain Adaptation for One-Class Classification: Monitoring the Health of Critical Systems Under Limited Information | The failure of a complex and safety critical industrial asset can have extremely high consequences. Close monitoring for early detection of abnormal system conditions is therefore required. Data-driven solutions to this problem have been limited for two reasons: First, safety critical assets are designed and maintained to be highly reliable and faults are rare. Fault detection can thus not be solved with supervised learning. Second, complex industrial systems usually have long lifetime during which they face very different operating conditions. In the early life of the system, the collected data is probably not representative of future operating conditions, making it challenging to train a robust model. In this paper, we propose a methodology to monitor the systems in their early life. To do so, we enhance the training dataset with other units from a fleet, for which longer observations are available. Since each unit has its own specificity, we propose to extract features made independent of their origin by three unsupervised feature alignment techniques. First, using a variational encoder, we impose a shared probabilistic encoder/decoder for both units. Second, we introduce a new loss designed to conserve inter-point spacial relationships between the input and the learned features. Last, we propose to train in an adversarial manner a discriminator on the origin of the features. Once aligned, the features are fed to a one-class classifier to monitor the health of the system. By exploring the different combinations of the proposed alignment strategies, and by testing them on a real case study, a fleet composed of 112 power plants operated in different geographical locations and under very different operating regimes, we demonstrate that this alignment is necessary and beneficial. | ['Gabriel Michau', 'Olga Fink'] | 2019-07-22 | null | null | null | null | ['one-class-classifier'] | ['methodology'] | [ 4.36571389e-01 4.51197699e-02 1.46180704e-01 -3.01794130e-02
-3.23676974e-01 -6.48496509e-01 5.90657473e-01 4.43577707e-01
-3.91710438e-02 9.12308753e-01 -7.56722033e-01 -2.48128399e-01
-6.35238826e-01 -8.63708615e-01 -7.88675785e-01 -1.07206285e+00
-3.58105987e-01 3.84501994e-01 1.97135761e-01 4.15490562e-04
2.34803334e-01 8.79712582e-01 -1.63874722e+00 -1.13885336e-01
6.80613875e-01 1.20254123e+00 2.99738586e-01 5.61578453e-01
4.02116954e-01 3.50152016e-01 -9.73321617e-01 1.91224739e-02
5.33054888e-01 -4.49582130e-01 -5.23983896e-01 4.08773929e-01
-2.29697004e-01 -7.05997348e-02 4.89836335e-02 9.21507299e-01
1.46075368e-01 2.10779592e-01 8.30650330e-01 -1.43556154e+00
1.96910411e-01 1.48884401e-01 -1.77232012e-01 -4.61032130e-02
1.66730285e-01 2.95660853e-01 7.63538957e-01 -3.76778454e-01
2.16918707e-01 5.50369382e-01 4.32444006e-01 2.25598812e-01
-1.28066003e+00 -2.68894196e-01 3.19170393e-02 -2.85363756e-02
-1.29523480e+00 -2.92900801e-01 8.90365064e-01 -6.19144857e-01
6.46046102e-01 9.21134204e-02 5.02919257e-01 1.07960999e+00
6.92516685e-01 1.52515635e-01 9.38951969e-01 -3.82534117e-01
5.41897178e-01 4.99375045e-01 -2.28470534e-01 4.62219328e-01
3.56364936e-01 3.36761802e-01 -1.27342530e-02 -1.21495798e-01
4.94337589e-01 2.63556212e-01 -4.09487963e-01 -6.38954163e-01
-9.54497755e-01 5.98990083e-01 2.59709686e-01 6.92425847e-01
-5.05895913e-01 -1.23862281e-01 3.21038127e-01 7.44314969e-01
2.18280956e-01 6.38143778e-01 -4.61627483e-01 -1.73335604e-03
-6.82816863e-01 -1.15388326e-01 9.04482067e-01 6.58966601e-01
5.88981330e-01 -3.57787348e-02 1.51043683e-01 3.14723909e-01
1.40953869e-01 2.42272124e-01 2.63666570e-01 -4.31723028e-01
4.05566096e-01 4.91213948e-01 1.99865937e-01 -9.30519581e-01
-3.71686846e-01 -5.28309703e-01 -9.35475290e-01 6.56308413e-01
3.26935053e-01 -2.86912411e-01 -6.14875913e-01 1.45413852e+00
2.00372741e-01 2.65245605e-02 2.86573589e-01 5.21699250e-01
-4.65258479e-01 7.92693138e-01 -2.07727194e-01 -4.54860926e-01
8.41925442e-01 -3.82325917e-01 -5.51012695e-01 9.59688947e-02
4.79493916e-01 -4.77655947e-01 6.19216084e-01 8.94834757e-01
-8.05350244e-01 -6.09385669e-01 -1.36017072e+00 8.71953607e-01
-4.97392029e-01 1.52768835e-01 3.80486548e-02 5.94440877e-01
-4.86147135e-01 1.14554608e+00 -9.61399019e-01 -2.95963675e-01
1.63336918e-01 3.84920418e-01 -3.95769805e-01 1.16948478e-01
-1.17932498e+00 1.13954031e+00 5.25417209e-01 4.01614636e-01
-1.27269959e+00 -4.03390676e-01 -7.37343848e-01 1.63250983e-01
4.71078932e-01 -2.04675063e-01 8.23559165e-01 -7.76684463e-01
-1.40802324e+00 9.07418504e-02 3.62521261e-01 -4.51414824e-01
7.27880895e-01 -1.05491757e-01 -4.87834603e-01 8.63543227e-02
-5.85930608e-02 -2.83929646e-01 1.18854785e+00 -1.36061609e+00
-6.69433177e-01 -1.63238883e-01 8.12573079e-03 -1.67464733e-01
-5.11516809e-01 -3.80472869e-01 3.28597367e-01 -3.96248370e-01
-1.21679120e-01 -9.24205065e-01 -2.09594414e-01 -1.38062418e-01
-4.85874027e-01 1.11699469e-01 9.89469528e-01 -5.17149091e-01
8.30529690e-01 -2.37699032e+00 3.02542984e-01 6.17899299e-01
-2.65249342e-01 1.52880490e-01 3.59207630e-01 6.56122446e-01
-2.83226669e-01 -1.65162191e-01 -6.70895219e-01 -1.90319568e-01
-1.54720262e-01 3.12900871e-01 -2.89799392e-01 6.13097072e-01
6.31930172e-01 1.76092118e-01 -7.21332908e-01 -1.65412858e-01
5.12602329e-01 3.27276915e-01 -2.22618207e-02 4.52502310e-01
-1.63769528e-01 6.46534324e-01 -7.45663285e-01 3.70374739e-01
2.70863861e-01 2.14074567e-01 2.99502853e-02 8.84154346e-04
-3.41764390e-02 -2.78605849e-01 -1.26299405e+00 1.35457623e+00
-8.77843022e-01 3.23685527e-01 -3.08092590e-03 -1.47985160e+00
1.15149534e+00 7.32097089e-01 7.63043284e-01 -3.82980585e-01
2.02821121e-01 2.82969475e-01 -1.99010640e-01 -4.86865014e-01
4.62663248e-02 -3.17577869e-01 -2.94602573e-01 2.18794674e-01
1.57544315e-01 -3.46053571e-01 1.31061524e-01 -3.46703291e-01
1.43059051e+00 -6.74076751e-02 2.80410588e-01 -1.90066934e-01
8.19100440e-01 -3.05304259e-01 6.16657376e-01 2.80679047e-01
6.04275949e-02 3.47544998e-01 7.60910809e-01 -2.29560971e-01
-8.43101621e-01 -1.11758804e+00 -3.04048002e-01 1.64313421e-01
1.15840323e-01 9.53153074e-02 -4.47327465e-01 -9.51533854e-01
1.29702151e-01 8.73721302e-01 -4.78822887e-01 -6.72745943e-01
-3.50579530e-01 -5.60408175e-01 2.29288921e-01 3.43868971e-01
2.32867837e-01 -9.91869330e-01 -1.05917716e+00 4.64689553e-01
4.56919104e-01 -8.17084134e-01 2.19534487e-01 8.51753235e-01
-8.16256464e-01 -1.36031365e+00 -5.08944809e-01 -4.40189362e-01
8.71025443e-01 -2.14391679e-01 8.93045664e-01 -4.86782044e-02
-4.09375608e-01 1.62432700e-01 -4.37335223e-01 -2.72272915e-01
-8.38646293e-01 4.16999348e-02 3.28531384e-01 2.50898868e-01
-2.42551610e-01 -6.08995676e-01 -1.38401046e-01 5.19308984e-01
-1.06189346e+00 -6.08787775e-01 7.91887999e-01 9.44902301e-01
2.54152566e-01 1.08203232e+00 7.30088830e-01 -6.82403564e-01
4.45372134e-01 -5.25493026e-01 -8.92875552e-01 3.57891470e-01
-7.58173227e-01 1.70326591e-01 1.11250663e+00 -3.30396473e-01
-9.46943283e-01 2.80937642e-01 7.68967122e-02 -4.58638310e-01
-5.56515455e-01 3.26694310e-01 -5.78344226e-01 7.72310272e-02
3.06814373e-01 3.48127745e-02 7.67784789e-02 -4.38003689e-01
-4.70742919e-02 5.16543567e-01 4.02001143e-01 -5.27451217e-01
1.15402448e+00 4.78734002e-02 4.45890635e-01 -8.76402140e-01
-3.52835476e-01 -2.24645212e-01 -7.70184040e-01 -3.51568609e-01
5.83861947e-01 -3.07729423e-01 -6.21442318e-01 4.41886634e-01
-9.15419400e-01 -1.42803326e-01 -6.36965275e-01 5.74650526e-01
-6.31900012e-01 2.58114576e-01 -2.10787177e-01 -9.39045191e-01
-4.82870340e-02 -1.20822346e+00 8.16720963e-01 1.90653149e-02
6.53199255e-02 -1.04600930e+00 6.80332184e-02 -1.69117987e-01
2.01634273e-01 7.39142895e-01 8.52838039e-01 -7.47882724e-01
-3.03765088e-01 -6.76076412e-01 4.02004093e-01 1.00528276e+00
6.43865407e-01 1.22369155e-01 -8.21586430e-01 -7.10558653e-01
4.93439674e-01 -7.08080307e-02 4.35162306e-01 -4.19200445e-03
1.02685523e+00 -1.41787440e-01 -4.41764832e-01 1.44301891e-01
1.35860491e+00 5.48163712e-01 3.63605231e-01 1.68760374e-01
3.00364405e-01 8.83875430e-01 1.02212524e+00 5.42324483e-01
-4.67230558e-01 5.77377737e-01 7.82746196e-01 1.44724876e-01
6.60934925e-01 1.62801817e-01 6.12692952e-01 4.49956626e-01
1.52303688e-02 -3.33270282e-01 -5.71768939e-01 6.17560744e-01
-1.52830172e+00 -8.29171658e-01 1.00085445e-01 2.66404033e+00
3.56565922e-01 6.50571406e-01 -4.35080491e-02 7.96540439e-01
7.07041502e-01 -8.85284543e-02 -4.51388121e-01 -2.84506023e-01
1.61182418e-01 5.37019297e-02 5.27592778e-01 2.04452097e-01
-1.17888165e+00 -2.56373677e-02 4.69326448e+00 5.07637799e-01
-1.11391068e+00 -1.11356370e-01 3.33298117e-01 1.99980307e-02
1.39190763e-01 1.36074215e-01 -4.08700734e-01 6.50948644e-01
1.16004813e+00 -9.10934247e-03 1.38769105e-01 7.71081686e-01
1.39855638e-01 -2.29872063e-01 -1.43425214e+00 4.13682818e-01
-3.58145609e-02 -6.52667284e-01 -4.83673573e-01 2.13783830e-01
4.23584163e-01 -4.76430774e-01 3.07672704e-03 1.41477408e-02
-2.15100095e-01 -8.21488082e-01 6.43492222e-01 7.83388555e-01
4.07192498e-01 -1.09284234e+00 9.88935292e-01 5.15083849e-01
-1.02344632e+00 -4.31542397e-01 -1.49757877e-01 1.81249633e-01
5.14228284e-01 9.04353976e-01 -7.01862216e-01 1.02180183e+00
3.68922293e-01 5.15727878e-01 -4.11119580e-01 9.52706635e-01
-2.23716125e-01 4.41914499e-01 -3.51836592e-01 1.41277924e-01
2.86579076e-02 -1.14391752e-01 7.30921447e-01 6.06527805e-01
7.21108615e-01 -5.65005064e-01 1.60967559e-01 9.00906622e-01
3.76069993e-01 -4.29360092e-01 -1.05252826e+00 -2.06143651e-02
1.99284092e-01 1.18043911e+00 -6.76792502e-01 1.58807501e-01
-2.68822581e-01 9.43796515e-01 -1.72084317e-01 1.11199811e-01
-8.43229115e-01 -5.65840781e-01 4.99883831e-01 4.19936292e-02
3.05618137e-01 -2.36694098e-01 1.05346277e-01 -7.15990067e-01
3.53619009e-01 -4.50415522e-01 1.86670110e-01 -2.32415304e-01
-1.35876846e+00 6.42032683e-01 2.07682624e-01 -1.63307703e+00
-6.74825907e-01 -7.35846400e-01 -7.51044869e-01 7.66555607e-01
-1.32056952e+00 -7.46043444e-01 -1.00085102e-01 5.30726135e-01
3.80134314e-01 -3.16788346e-01 8.09351206e-01 3.69304657e-01
-7.75288999e-01 1.70644999e-01 2.78247356e-01 -2.04924077e-01
4.68380630e-01 -1.32451379e+00 1.86105520e-02 9.03504729e-01
-3.81407328e-02 2.41499439e-01 8.97938311e-01 -6.32408500e-01
-1.22912490e+00 -1.14773428e+00 4.50926214e-01 -1.78784743e-01
7.39778996e-01 -3.87919188e-01 -1.00118411e+00 4.96085316e-01
1.83392856e-02 7.25742429e-02 2.76674896e-01 -2.84605533e-01
4.38488066e-01 -3.56127709e-01 -1.36871147e+00 -3.11298121e-04
3.08422565e-01 -3.48967940e-01 -5.80342114e-01 4.00844902e-01
3.21694046e-01 1.58566609e-01 -1.18999195e+00 5.09657025e-01
2.11384133e-01 -8.98182273e-01 5.67714274e-01 -2.32557222e-01
5.78620881e-02 -3.63772780e-01 5.41695021e-02 -1.61691642e+00
2.96184514e-02 -5.92379510e-01 -1.26868725e-01 1.51635385e+00
4.30182666e-01 -9.29514587e-01 5.59926748e-01 3.24511319e-01
-2.29819641e-01 -7.09614396e-01 -1.12086546e+00 -1.14761126e+00
-2.70916615e-02 -1.64750874e-01 5.97670913e-01 6.75164402e-01
-6.83416352e-02 1.29623357e-02 -2.31122851e-01 5.21552622e-01
5.21118283e-01 5.70764765e-02 4.36570585e-01 -1.39736712e+00
-4.16320175e-01 -1.63580805e-01 -6.49745643e-01 -2.19060495e-01
2.33773932e-01 -3.46385390e-01 4.35241967e-01 -9.61801231e-01
-3.93724829e-01 -7.18290448e-01 -5.22697270e-01 2.71231145e-01
2.23090202e-01 -2.60407209e-01 -2.49173231e-02 2.93192603e-02
1.12333193e-01 6.57308877e-01 7.23988175e-01 -2.45034203e-01
-1.90189444e-02 6.36883557e-01 1.36999622e-01 4.83893722e-01
8.87261093e-01 -4.26636249e-01 -5.39605856e-01 1.45813152e-01
-9.15916229e-04 3.09820265e-01 5.67593336e-01 -1.50872350e+00
4.65742461e-02 -9.19177830e-02 2.66960859e-01 -3.83007407e-01
3.63592923e-01 -1.63244104e+00 4.72336620e-01 6.77704394e-01
-6.33820444e-02 6.60953745e-02 8.00680276e-03 7.26764917e-01
-3.28323692e-01 -7.15849400e-01 7.03088999e-01 1.74362496e-01
-4.41366374e-01 5.81044555e-02 -4.70772266e-01 -5.73449850e-01
1.64898837e+00 -7.22246990e-02 1.62404642e-01 -1.76100750e-02
-7.47779965e-01 2.86259681e-01 5.11294305e-01 2.82336861e-01
4.88606155e-01 -1.01039517e+00 -4.18421239e-01 4.77606684e-01
2.04432219e-01 9.98666659e-02 1.49013162e-01 8.31275165e-01
-3.25190425e-01 2.63192773e-01 -3.16136748e-01 -4.87306535e-01
-9.66487765e-01 9.46658671e-01 3.23485851e-01 -2.79269099e-01
-5.52975714e-01 3.25287819e-01 -1.15057625e-01 -1.60420924e-01
7.56300539e-02 -4.60031122e-01 -1.40903667e-01 1.32472858e-01
1.12474844e-01 2.87655950e-01 4.56542552e-01 -4.39571798e-01
-2.37007901e-01 4.05026108e-01 2.22459257e-01 9.09335613e-02
1.42645013e+00 -1.24948993e-02 1.20923646e-01 6.82355881e-01
1.06795573e+00 -1.82698771e-01 -1.47302079e+00 2.38896683e-01
1.43958569e-01 -4.20244038e-01 -9.96061321e-03 -4.54724818e-01
-1.32915425e+00 8.07002902e-01 6.67444766e-01 8.92317235e-01
1.42034781e+00 -2.20420465e-01 4.11174864e-01 3.80608499e-01
6.02023184e-01 -9.89444494e-01 -1.16761252e-01 7.35834092e-02
7.30609238e-01 -1.01072526e+00 -1.81957170e-01 -1.92948207e-01
-3.75371486e-01 1.23876762e+00 2.75842875e-01 -1.40416056e-01
7.26008058e-01 3.31769556e-01 -2.70087421e-01 -1.15412429e-01
-7.02096701e-01 8.41179267e-02 2.31745187e-02 4.89982039e-01
-9.37761459e-03 -8.46146420e-02 -8.18189606e-02 2.81576693e-01
1.56820506e-01 -2.36426219e-01 4.93546426e-01 1.29820383e+00
-1.79830566e-01 -1.23518932e+00 -4.83544439e-01 4.06484812e-01
-2.91260749e-01 6.18428826e-01 -5.02001569e-02 8.71424556e-01
2.38918915e-01 9.73394334e-01 -7.04960972e-02 -4.03267562e-01
6.48532510e-01 1.38026297e-01 3.09358388e-01 -5.33613503e-01
-3.95334780e-01 -2.82682151e-01 -1.29902020e-01 -4.66922611e-01
-2.23014027e-01 -9.10267413e-01 -1.04981863e+00 1.56351477e-01
-5.35131752e-01 4.32643384e-01 8.54609668e-01 1.08263958e+00
-1.46662779e-02 1.00388515e+00 1.37277484e+00 -8.83815646e-01
-1.11362112e+00 -8.55218112e-01 -1.14136255e+00 2.62723893e-01
5.15027404e-01 -1.06439006e+00 -6.76879287e-01 -1.12810219e-02] | [6.735410690307617, 2.396639347076416] |
790f178b-9481-45f1-9f81-b56e08696ed9 | multilingual-representation-distillation-with | 2210.05033 | null | https://arxiv.org/abs/2210.05033v2 | https://arxiv.org/pdf/2210.05033v2.pdf | Multilingual Representation Distillation with Contrastive Learning | Multilingual sentence representations from large models encode semantic information from two or more languages and can be used for different cross-lingual information retrieval and matching tasks. In this paper, we integrate contrastive learning into multilingual representation distillation and use it for quality estimation of parallel sentences (i.e., find semantically similar sentences that can be used as translations of each other). We validate our approach with multilingual similarity search and corpus filtering tasks. Experiments across different low-resource languages show that our method greatly outperforms previous sentence encoders such as LASER, LASER3, and LaBSE. | ['Philipp Koehn', 'Holger Schwenk', 'Kevin Heffernan', 'Weiting Tan'] | 2022-10-10 | null | null | null | null | ['cross-lingual-information-retrieval'] | ['natural-language-processing'] | [-2.49987409e-01 -3.35979164e-01 -4.14837658e-01 -4.62229401e-01
-1.53806448e+00 -6.98901296e-01 6.41760945e-01 7.08212495e-01
-8.76949012e-01 8.31525624e-01 7.00609744e-01 -2.70335734e-01
1.16664872e-01 -6.61262453e-01 -9.53079224e-01 1.45125717e-01
4.64384109e-01 7.41595864e-01 4.78763655e-02 -6.43518269e-01
1.36159107e-01 1.63367465e-02 -1.06145740e+00 5.70928574e-01
1.00836694e+00 4.66599822e-01 5.59569180e-01 2.64290422e-01
-5.10184884e-01 6.63639009e-01 -5.66317797e-01 -8.89637232e-01
-8.07919800e-02 -3.67740154e-01 -1.06037915e+00 -5.63378215e-01
7.78076172e-01 2.16574341e-01 -1.86826989e-01 1.27024877e+00
6.44367576e-01 -7.12241754e-02 4.38117534e-01 -7.62195647e-01
-1.17301202e+00 1.03439081e+00 -2.69053340e-01 3.89745086e-01
8.18807364e-01 -3.58963996e-01 1.36289346e+00 -1.22492540e+00
1.06686068e+00 1.72841573e+00 6.63385928e-01 3.29479873e-01
-1.10326970e+00 -6.66867554e-01 -1.86608806e-01 4.65443581e-01
-1.55623031e+00 -7.66868174e-01 3.51935327e-01 -1.37528747e-01
1.49980938e+00 1.06910832e-01 -4.07358743e-02 1.14120555e+00
3.54036272e-01 8.08238268e-01 8.39456975e-01 -8.62420082e-01
-3.06643963e-01 3.59037280e-01 3.10657937e-02 8.07231069e-01
2.28003919e-01 -2.07346782e-01 -5.23507237e-01 -2.17613038e-02
1.40405938e-01 -2.82332748e-01 -1.74864814e-01 1.74394906e-01
-1.34775519e+00 1.01109719e+00 2.09789455e-01 7.28848696e-01
-2.07543626e-01 -1.28275290e-01 8.48862052e-01 8.77193511e-01
5.51214278e-01 7.64423668e-01 -5.74631214e-01 8.19315612e-02
-7.72431612e-01 1.19363487e-01 6.60684347e-01 1.28237760e+00
8.30901742e-01 -4.03911114e-01 -2.29289562e-01 1.37396371e+00
2.87782282e-01 1.00991213e+00 9.39272821e-01 -6.70561373e-01
1.00573671e+00 3.81912410e-01 -3.17357272e-01 -6.49631321e-01
-1.88961536e-01 -4.31490764e-02 -5.92714369e-01 -6.87670887e-01
-2.30774492e-01 7.57344514e-02 -3.34133178e-01 1.61591232e+00
-1.42102301e-01 -1.28370777e-01 6.27077460e-01 5.77971756e-01
1.25102520e+00 6.92250013e-01 7.55967293e-03 -1.11344025e-01
1.45908177e+00 -1.00892377e+00 -6.79974735e-01 -3.88387203e-01
1.08147085e+00 -1.44855094e+00 1.18761706e+00 -1.37051478e-01
-1.14936566e+00 -8.32863450e-01 -1.03516328e+00 -4.36448812e-01
-5.24018347e-01 8.23528767e-02 3.16493332e-01 3.61169875e-01
-1.08071816e+00 5.79917431e-01 -4.62363034e-01 -6.34259462e-01
1.42326849e-02 -2.85678655e-01 -5.11640728e-01 -5.68269610e-01
-1.67929602e+00 1.40316093e+00 5.99573076e-01 -4.24409479e-01
-4.38382775e-01 -6.46269858e-01 -1.32927012e+00 -1.81486920e-01
-2.24194735e-01 -6.91789687e-01 1.23255801e+00 -6.05225980e-01
-1.08716273e+00 1.17105508e+00 -4.24671888e-01 -5.79334378e-01
-1.33459702e-01 -3.46280158e-01 -8.80009055e-01 7.70925879e-02
7.37216830e-01 7.43318915e-01 1.23478457e-01 -5.77576458e-01
-4.25165802e-01 -1.75650388e-01 8.90930966e-02 4.51583713e-01
-2.76926935e-01 7.09788144e-01 -5.37263274e-01 -5.68962336e-01
6.35812953e-02 -6.32945240e-01 -1.10413924e-01 -4.41266328e-01
-2.45317981e-01 -5.06799936e-01 2.93776780e-01 -1.00195885e+00
1.09008884e+00 -1.88692153e+00 2.07718819e-01 -2.07709134e-01
-4.26518470e-01 8.83238465e-02 -6.54248059e-01 7.88720906e-01
3.59881483e-02 2.91060328e-01 4.97269705e-02 -4.69868839e-01
7.79155418e-02 2.87777781e-01 -2.09625125e-01 2.04266727e-01
3.35707992e-01 1.09319234e+00 -1.08609498e+00 -8.96893382e-01
-2.74722185e-02 3.56042415e-01 -3.63871545e-01 1.98185906e-01
1.21363334e-03 2.87577808e-01 -1.11890048e-01 5.59328318e-01
2.62136459e-01 2.73039639e-02 3.10128897e-01 -2.03728929e-01
1.17754899e-01 1.10993791e+00 -7.55315721e-01 2.40477157e+00
-1.27270079e+00 6.30927086e-01 -3.67993802e-01 -8.48677099e-01
1.04819810e+00 4.77704376e-01 9.80692804e-02 -1.47445905e+00
-1.65559709e-01 6.74521983e-01 -3.43169391e-01 -5.13627052e-01
9.31738436e-01 -1.93971694e-01 -5.63980758e-01 5.51923037e-01
5.28004110e-01 -3.48737299e-01 7.03119874e-01 2.82279342e-01
6.94762707e-01 -2.10834648e-02 5.04521847e-01 -4.40798640e-01
8.14505339e-01 -7.10531324e-02 4.41696614e-01 6.30642653e-01
1.61857292e-01 3.05823445e-01 -1.58129379e-01 -1.50410369e-01
-1.18746269e+00 -1.19791472e+00 -3.86091202e-01 1.10055733e+00
6.81203902e-02 -7.01173127e-01 -3.27847868e-01 -7.62985229e-01
-5.98225892e-02 8.44164908e-01 -8.37029051e-03 -1.38187721e-01
-8.09235096e-01 -2.87375391e-01 8.14206481e-01 4.38847452e-01
3.03132199e-02 -1.12703943e+00 9.38072503e-02 3.33176523e-01
-7.52026379e-01 -1.34712267e+00 -6.73841119e-01 -2.54698079e-02
-6.93697214e-01 -9.51478124e-01 -4.88604367e-01 -1.30278075e+00
1.80246130e-01 1.84957013e-01 1.83778048e+00 -2.25259528e-01
-2.03262225e-01 2.90155888e-01 -4.45332497e-01 -1.53750271e-01
-9.56677675e-01 1.55257300e-01 2.71194339e-01 -6.71275198e-01
6.89707339e-01 -2.79644787e-01 -2.06109062e-02 -5.59972413e-02
-7.42527843e-01 -3.91817272e-01 5.07393122e-01 9.35872614e-01
8.17363024e-01 -3.79596561e-01 7.75098920e-01 -8.10849965e-01
1.12284374e+00 -5.36421001e-01 -5.27261794e-01 7.99398839e-01
-5.64437270e-01 5.75545907e-01 6.83023095e-01 -8.92476812e-02
-8.72546732e-01 -4.64522123e-01 -5.05170643e-01 -2.68690377e-01
2.11080126e-02 9.22491550e-01 -1.41967326e-01 3.06860566e-01
6.09248638e-01 2.85367846e-01 -3.90659541e-01 -7.39652455e-01
7.68712819e-01 8.42473269e-01 4.95957255e-01 -6.80355728e-01
4.28205073e-01 -1.55582935e-01 -4.45121735e-01 -5.71099162e-01
-8.60864937e-01 -5.62130392e-01 -8.02142680e-01 2.11658850e-01
7.02950656e-01 -1.34887993e+00 -9.10370573e-02 -1.92147210e-01
-1.62983334e+00 4.47008878e-01 -1.72153696e-01 6.88389957e-01
-2.95354217e-01 5.17359912e-01 -8.65495801e-01 -1.35557249e-01
-7.61070549e-01 -1.17710698e+00 1.38444686e+00 -1.28734514e-01
-4.22229230e-01 -1.35696208e+00 6.40513480e-01 2.38964781e-01
2.96065599e-01 -4.75774139e-01 9.45000947e-01 -7.49769211e-01
-1.11796901e-01 -2.06512943e-01 -1.40701324e-01 4.91364777e-01
4.56605926e-02 -2.94882774e-01 -5.29435337e-01 -5.68658769e-01
-2.89590091e-01 -8.64493906e-01 7.47847497e-01 1.73114929e-02
5.78928530e-01 -3.57051194e-01 -1.23244047e-01 3.29549611e-01
1.62290692e+00 -1.95769325e-01 4.65900570e-01 3.63872498e-01
4.93551046e-01 6.34565115e-01 6.20685697e-01 -4.38094363e-02
6.11443520e-01 8.51418257e-01 -2.95704544e-01 1.20871879e-01
-3.84396434e-01 -4.94606256e-01 6.30664289e-01 1.96008039e+00
6.47659659e-01 -5.80887273e-02 -7.94718206e-01 6.66423380e-01
-1.65306985e+00 -9.93622303e-01 9.43573564e-02 2.08070040e+00
1.21490240e+00 -6.37244657e-02 -4.25801009e-01 -6.01962864e-01
6.78861380e-01 3.67251001e-02 -6.48893788e-02 -6.66188657e-01
-5.34634113e-01 6.75924480e-01 4.08970445e-01 7.27051556e-01
-9.99312460e-01 1.27494514e+00 6.65626431e+00 8.76868904e-01
-8.23695183e-01 6.09047353e-01 2.28936031e-01 1.41637444e-01
-7.00015843e-01 1.28313944e-01 -9.83363748e-01 3.13370168e-01
1.18038023e+00 -6.70522392e-01 1.76095724e-01 6.61040604e-01
-1.56563535e-01 2.07740307e-01 -1.34935534e+00 1.12938297e+00
4.93866175e-01 -1.40976405e+00 2.62851924e-01 -6.70869470e-01
7.15848565e-01 6.69233441e-01 -2.61956096e-01 7.67632723e-01
5.79704583e-01 -9.80307221e-01 5.35745800e-01 4.65396285e-01
8.30630124e-01 -7.18049467e-01 9.51085925e-01 1.76330730e-01
-1.33416700e+00 5.16804516e-01 -7.63168573e-01 3.02286088e-01
3.54827374e-01 4.03248757e-01 -8.15934479e-01 1.06704021e+00
5.29375434e-01 1.05477226e+00 -8.25507700e-01 7.38068640e-01
-3.66723120e-01 2.27516517e-01 -1.07226841e-01 -1.11869134e-01
1.60705879e-01 -1.04413532e-01 3.63578409e-01 1.57902110e+00
4.15556222e-01 -7.61242568e-01 4.11019564e-01 7.90416479e-01
-4.37363237e-01 8.00434709e-01 -8.48084927e-01 -4.30845842e-02
8.44149649e-01 8.43523502e-01 3.35130058e-02 -5.38197994e-01
-7.90481925e-01 1.18494940e+00 7.78326035e-01 7.54964948e-02
-4.70635355e-01 -5.95374048e-01 6.62655473e-01 -4.99331892e-01
-3.21643949e-02 -1.52384102e-01 1.49157584e-01 -1.65245199e+00
3.07810515e-01 -9.38145459e-01 5.54140151e-01 -6.80508852e-01
-1.84055221e+00 8.50748301e-01 -1.84503108e-01 -1.23192036e+00
-6.34668350e-01 -3.83843094e-01 -3.79173100e-01 1.25927198e+00
-1.74116445e+00 -1.17700171e+00 4.30686831e-01 6.98625445e-01
8.17075372e-01 -5.46363235e-01 1.19712460e+00 7.34379113e-01
-2.34202400e-01 7.67793119e-01 1.95986181e-01 3.87513399e-01
1.02624130e+00 -1.03826559e+00 6.62592471e-01 9.48650837e-01
8.59780788e-01 6.83998287e-01 3.23336452e-01 -6.15064025e-01
-1.25044191e+00 -1.24334335e+00 1.92464471e+00 -4.18792367e-01
1.04393411e+00 -2.58674145e-01 -9.02297974e-01 5.34208953e-01
7.90495038e-01 -3.97895128e-01 7.87500441e-01 4.50068980e-01
-8.41579020e-01 -6.11167848e-02 -7.68251777e-01 5.27178049e-01
8.63419652e-01 -1.23747742e+00 -9.96982098e-01 7.03102946e-01
1.13830090e+00 -1.78531930e-01 -1.11230183e+00 2.90021390e-01
2.24820003e-01 -3.62878203e-01 1.07309699e+00 -8.88510704e-01
4.38622117e-01 7.55050927e-02 -5.79664886e-01 -1.62367046e+00
-2.18670011e-01 -1.39659401e-02 3.48182499e-01 1.33082330e+00
7.70622730e-01 -4.17610168e-01 -4.02332507e-02 -2.08934844e-01
-3.03252429e-01 -2.69492775e-01 -1.17814577e+00 -1.22164905e+00
6.59777343e-01 -3.69249016e-01 6.00645483e-01 1.18278813e+00
3.93365115e-01 1.08615768e+00 -9.37399939e-02 7.26022944e-02
4.96946752e-01 3.24831307e-01 3.94819587e-01 -1.00309312e+00
-3.25851321e-01 -5.65280616e-01 -3.37150633e-01 -9.94922340e-01
8.39374363e-01 -1.81001806e+00 6.09335452e-02 -1.57117188e+00
4.57316071e-01 -3.14644903e-01 -2.68012315e-01 3.42968702e-01
-3.42988700e-01 5.35629392e-02 3.93328704e-02 2.91602850e-01
-8.11817408e-01 7.05562770e-01 8.43712568e-01 -5.16735852e-01
2.41888925e-01 -6.03104770e-01 -4.65247333e-01 3.17841113e-01
7.31878281e-01 -6.38325989e-01 -1.08336054e-01 -1.00559664e+00
4.42387938e-01 1.15727641e-01 -1.10790268e-01 -8.81759286e-01
8.30797106e-02 1.31003186e-01 -1.19368667e-02 -4.67986286e-01
2.79067494e-02 -5.15491784e-01 -2.97072113e-01 3.58595401e-01
-6.95198238e-01 7.82392740e-01 2.98416883e-01 2.84897089e-01
-8.26048434e-01 -6.26788735e-01 5.36329269e-01 -3.03365231e-01
-7.15025067e-01 -5.86093264e-03 -2.14563072e-01 6.06461942e-01
4.65349257e-01 6.25902176e-01 -4.82705563e-01 -3.74941379e-01
-3.80272090e-01 3.19630325e-01 3.25912654e-01 1.10274851e+00
6.68916285e-01 -1.75765240e+00 -1.41022539e+00 8.30262229e-02
7.63505280e-01 -6.15293324e-01 -2.27584422e-01 6.02417111e-01
-3.72302234e-01 8.76988590e-01 -8.40008333e-02 -4.75061595e-01
-1.28029108e+00 5.28842330e-01 1.92332566e-01 -5.18161297e-01
-2.75683343e-01 7.07243085e-01 -3.45176786e-01 -1.09839702e+00
-1.23117073e-02 -4.63170633e-02 -2.15408996e-01 -1.26239331e-02
5.53769171e-01 -9.16991830e-02 4.01784331e-01 -1.01937962e+00
-6.34808123e-01 4.07787621e-01 -3.09645087e-01 -2.23734543e-01
1.13974094e+00 -3.61697078e-01 -5.34258187e-01 6.34688020e-01
1.79788637e+00 9.75723863e-02 -2.67494116e-02 -8.04205775e-01
5.50128102e-01 -3.08667719e-01 -1.47612184e-01 -5.78522325e-01
-6.74920440e-01 8.74317408e-01 4.99224901e-01 -3.97411585e-01
7.39992261e-01 4.72888023e-01 1.04739511e+00 8.18764389e-01
6.25840604e-01 -1.24124432e+00 1.17613794e-02 8.71316731e-01
9.14430141e-01 -1.50471973e+00 -1.70534715e-01 1.61111850e-04
-5.70705175e-01 1.11313796e+00 4.15391088e-01 4.60168999e-03
4.33963120e-01 1.14059404e-01 1.84905589e-01 -1.83479816e-01
-8.59101713e-01 -3.37563187e-01 6.44058585e-01 3.74693871e-01
1.12612236e+00 9.17751491e-02 -7.46527910e-01 3.73850286e-01
-3.66462886e-01 -4.28487152e-01 2.21372604e-01 8.09962392e-01
-3.30515772e-01 -1.63893497e+00 -1.27382772e-02 1.86418802e-01
-4.28400248e-01 -9.04229939e-01 -3.27620029e-01 4.15493876e-01
-1.94120705e-02 1.00775170e+00 9.44609717e-02 -1.87227517e-01
3.97086263e-01 2.92791069e-01 6.70356274e-01 -9.46242392e-01
-6.03963315e-01 -1.88809782e-01 5.00346005e-01 -5.67196846e-01
-5.38802624e-01 -7.90788293e-01 -8.40680659e-01 -5.74992821e-02
-2.28010476e-01 4.22865540e-01 7.85624504e-01 9.82074678e-01
3.84566516e-01 3.19882214e-01 5.64449489e-01 -6.42940179e-02
-4.79211748e-01 -1.27939510e+00 -7.94115290e-02 6.60772264e-01
-6.07172847e-02 -2.23738074e-01 4.36262898e-02 -6.63201809e-02] | [11.138365745544434, 9.83722972869873] |
7a05a2d5-99dc-4466-8627-7cb02d6bee4e | patch-wise-spatial-temporal-quality | null | null | https://www.researchgate.net/publication/353113483_Patch-Wise_Spatial-Temporal_Quality_Enhancement_for_HEVC_Compressed_Video | https://www.researchgate.net/publication/353113483_Patch-Wise_Spatial-Temporal_Quality_Enhancement_for_HEVC_Compressed_Video | Patch-Wise Spatial-Temporal Quality Enhancement for HEVC Compressed Video | Recently, many deep learning based researches are conducted to explore the potential quality improvement of compressed videos. These methods mostly utilize either the spatial or temporal information to perform frame-level video enhancement. However, they fail in combining different spatial-temporal information to adaptively utilize adjacent patches to enhance the current patch and achieve limited enhancement performance especially on scene-changing and strong-motion videos. To overcome these limitations, we propose a patch-wise spatial-temporal quality enhancement network which firstly extracts spatial and temporal features, then recalibrates and fuses the obtained spatial and temporal features. Specifically, we design a temporal and spatial-wise attention-based feature distillation structure to adaptively utilize the adjacent patches for distilling patch-wise temporal features. For adaptively enhancing different patch with spatial and temporal information, a channel and spatial-wise attention fusion block is proposed to achieve patch-wise recalibration and fusion of spatial and temporal features. Experimental results demonstrate our network achieves peak signal-to-noise ratio improvement, 0.55 – 0.69 dB compared with the compressed videos at different quantization parameters, outperforming state- of-the-art approaches. | ['Mai Xu', 'Hao Yang', 'Liangwei Yu', 'Liquan Shen', 'Qing Ding'] | 2021-07-08 | null | null | null | journal-2021-7 | ['video-enhancement'] | ['computer-vision'] | [ 1.89789101e-01 -8.26075613e-01 -1.98757201e-01 -3.64559174e-01
-7.11911440e-01 -2.45101061e-02 1.82333142e-01 9.68602747e-02
-5.53562164e-01 5.33947468e-01 5.20549059e-01 -9.09396857e-02
-3.65181297e-01 -7.27059603e-01 -5.72451651e-01 -8.54462683e-01
-4.40575093e-01 -7.46497691e-01 3.64118963e-01 -9.75972563e-02
2.07561210e-01 1.52847469e-01 -1.57665336e+00 4.34149206e-01
9.76301372e-01 1.46263802e+00 5.95431864e-01 6.72505319e-01
1.01224929e-01 7.08955705e-01 -2.07084551e-01 1.03608444e-01
2.67033637e-01 -5.98406017e-01 -2.59050965e-01 2.27852792e-01
3.93687338e-01 -7.50057578e-01 -7.76216567e-01 1.16317952e+00
7.36855805e-01 2.89246321e-01 8.86638649e-03 -8.86351943e-01
-9.76953566e-01 4.05391246e-01 -1.03262556e+00 8.09720814e-01
2.26083383e-01 4.15934145e-01 6.48557723e-01 -9.36215580e-01
2.09083498e-01 1.01820743e+00 5.16266763e-01 2.18293160e-01
-7.81456470e-01 -7.25617528e-01 2.90667504e-01 8.83676589e-01
-1.52539086e+00 -4.12461489e-01 1.09169376e+00 -5.88181168e-02
8.56931508e-01 -1.50673222e-02 6.25777602e-01 4.02205169e-01
2.12496132e-01 9.08316016e-01 8.65061641e-01 -1.37148887e-01
-1.39561482e-02 -3.52399588e-01 -2.75365889e-01 6.02460861e-01
-2.42907524e-01 2.64475852e-01 -6.47462785e-01 2.58873403e-01
1.03547049e+00 1.81856886e-01 -7.40301907e-01 -8.07868764e-02
-1.44180405e+00 4.98486757e-01 8.24697137e-01 8.58875871e-01
-8.63906324e-01 1.60795674e-01 4.19765532e-01 7.61862472e-02
1.42819151e-01 1.58157811e-01 -4.37771499e-01 -2.50597745e-01
-1.33399129e+00 -1.47469088e-01 -6.66902289e-02 7.79871881e-01
7.77981877e-01 2.77370512e-01 -5.92418313e-01 8.65176260e-01
1.27383411e-01 3.07850063e-01 6.43000126e-01 -1.13594437e+00
5.23125827e-01 2.75288641e-01 1.06496029e-01 -1.29346323e+00
-2.17445254e-01 -6.06072366e-01 -1.22770202e+00 -1.25251174e-01
-2.35208452e-01 -1.91047356e-01 -8.79895985e-01 1.60312021e+00
1.68615371e-01 6.36202216e-01 -9.93015319e-02 1.30665004e+00
8.32472980e-01 1.00679171e+00 2.47481212e-01 -6.88884318e-01
1.20723057e+00 -1.10220635e+00 -1.06023407e+00 4.93889824e-02
1.98335052e-01 -6.89505935e-01 8.58297706e-01 3.00785393e-01
-1.44549739e+00 -1.15760696e+00 -1.24320173e+00 7.12093860e-02
1.40801564e-01 1.84657454e-01 2.10518882e-01 2.45735422e-01
-1.07765841e+00 6.34353518e-01 -7.11746454e-01 2.95468301e-01
5.05736053e-01 3.49643439e-01 -2.03581214e-01 -3.99443328e-01
-1.47947216e+00 3.78547758e-01 4.52001035e-01 1.35567859e-01
-8.40075016e-01 -8.11338067e-01 -9.06849265e-01 2.95689762e-01
3.83408904e-01 -5.24934590e-01 9.72148240e-01 -9.98200655e-01
-1.12098062e+00 7.58869648e-02 -3.59064698e-01 -3.28517526e-01
1.18925937e-01 -1.10982955e-01 -7.92311847e-01 5.89212596e-01
1.32629769e-02 7.37797022e-01 9.97908175e-01 -1.08598351e+00
-1.00996327e+00 -2.39240870e-01 -5.17741172e-03 3.97107571e-01
-7.78660595e-01 -3.93905155e-02 -9.12029028e-01 -1.08136010e+00
2.86621273e-01 -1.21898524e-01 -2.78835028e-01 1.81395695e-01
1.18908554e-01 1.74369350e-01 1.24694419e+00 -9.98125494e-01
1.74453425e+00 -2.31876874e+00 5.47676459e-02 -2.46237561e-01
3.59507799e-01 7.39565790e-01 -4.44341451e-01 2.13600136e-02
1.16484277e-01 4.44865078e-02 -2.43102893e-01 -4.19584066e-02
-2.81089067e-01 -2.03752249e-01 2.06615672e-01 3.76867563e-01
3.41691375e-01 8.68183792e-01 -1.03314030e+00 -4.98799920e-01
6.91300750e-01 1.01640236e+00 -7.19864011e-01 1.93225831e-01
1.57806426e-01 5.61873317e-01 -5.29908776e-01 5.85330486e-01
9.70892012e-01 -2.29009047e-01 -8.87831971e-02 -7.02218771e-01
-2.13925809e-01 5.15829884e-02 -1.03871453e+00 1.74154270e+00
-5.27748227e-01 7.01715231e-01 2.74149925e-01 -1.11094666e+00
7.86860585e-01 3.37099850e-01 7.57833958e-01 -1.25086737e+00
2.43483081e-01 1.02327034e-01 -5.97270317e-02 -6.67977035e-01
5.66540480e-01 -1.34324014e-01 2.35746667e-01 -1.03414953e-01
5.61652556e-02 5.26051998e-01 8.10029451e-03 1.14274789e-02
1.03116858e+00 5.13251573e-02 9.27926749e-02 8.79108682e-02
9.48001921e-01 -6.00406826e-01 8.99518490e-01 3.38079691e-01
-6.49337411e-01 6.34876013e-01 -9.55846384e-02 -2.98010081e-01
-1.08450222e+00 -7.87062526e-01 -1.12890042e-02 9.87196863e-01
7.30973125e-01 -4.75746214e-01 -4.91602600e-01 -4.22612816e-01
-2.67619044e-01 3.98781419e-01 -4.80947912e-01 -4.26529765e-01
-6.40428483e-01 -4.94473547e-01 7.27133825e-02 6.99912369e-01
1.29609025e+00 -8.44610572e-01 -6.49566531e-01 6.43481672e-01
-5.39085746e-01 -9.72732782e-01 -7.94730246e-01 -2.92670965e-01
-6.93632305e-01 -6.47586882e-01 -1.10544538e+00 -8.02475870e-01
2.46241659e-01 6.36762440e-01 8.02193522e-01 2.37807840e-01
-1.36267751e-01 1.28080055e-01 -6.36974812e-01 2.84969211e-01
1.98971778e-01 -3.44924808e-01 -2.12576345e-01 2.82176763e-01
1.61260188e-01 -7.97141492e-01 -1.21751297e+00 4.17699367e-01
-1.12436712e+00 2.70106364e-02 7.74373293e-01 1.03690493e+00
6.06755137e-01 5.24716556e-01 6.54422879e-01 1.13729713e-02
4.93515462e-01 -4.87901002e-01 -1.74236685e-01 9.59343761e-02
-3.36978793e-01 -1.69490919e-01 6.50544643e-01 -5.77638865e-01
-9.98663962e-01 -3.65995675e-01 -1.96385428e-01 -8.11662853e-01
-2.25871950e-02 6.22317731e-01 -4.68839377e-01 -7.26263076e-02
9.62922722e-02 7.33629882e-01 -2.58884996e-01 -4.16367203e-01
1.99983463e-01 6.41019464e-01 6.81744814e-01 -2.53679097e-01
5.13014019e-01 3.43675494e-01 -3.47325742e-01 -4.72791940e-01
-3.37814599e-01 -3.87868583e-01 -4.03373390e-01 -4.01104838e-01
1.05648291e+00 -1.11708331e+00 -4.51718688e-01 6.18341446e-01
-9.17644560e-01 -1.62395954e-01 -1.48759380e-01 7.90479779e-01
-4.34863865e-01 6.79593325e-01 -8.79173040e-01 -4.17841017e-01
-3.68599594e-01 -1.38168049e+00 1.02458680e+00 5.13004422e-01
4.58133370e-01 -6.58931077e-01 -4.55963254e-01 1.62771717e-01
8.88658047e-01 -5.84198684e-02 6.86764657e-01 1.75302863e-01
-6.53961658e-01 9.43632200e-02 -6.84231579e-01 4.28194970e-01
3.76346678e-01 -2.86191493e-01 -5.84568560e-01 -4.96981323e-01
4.33468334e-02 3.14777754e-02 1.05357563e+00 7.61144042e-01
1.56627107e+00 -4.01027769e-01 -1.15836270e-01 8.80126715e-01
1.39918673e+00 5.38922012e-01 8.34274411e-01 3.34919035e-01
7.04131365e-01 1.42636240e-01 8.48251462e-01 7.53397584e-01
2.94724047e-01 7.08730221e-01 4.89036351e-01 -3.21391195e-01
-3.57515216e-01 -1.39115140e-01 2.39450321e-01 6.86398685e-01
-7.87261799e-02 -1.30122438e-01 -3.58240753e-01 7.64658391e-01
-1.69428515e+00 -1.06644487e+00 2.64078945e-01 1.95920169e+00
9.71305907e-01 1.08623914e-02 -7.88755193e-02 2.96731532e-01
9.75438535e-01 4.01637942e-01 -6.88489556e-01 1.67211965e-01
-2.89503962e-01 4.00353707e-02 4.00911093e-01 3.26087356e-01
-1.28255856e+00 5.37609339e-01 5.07697773e+00 1.20712686e+00
-1.38090599e+00 1.85565770e-01 8.32186580e-01 -2.58788347e-01
-2.54730880e-01 -2.51172125e-01 -2.11192653e-01 7.49977112e-01
7.69715428e-01 -1.48063794e-01 4.40357476e-01 5.82714200e-01
5.62548757e-01 1.22213196e-02 -6.15674376e-01 1.19849086e+00
7.50177950e-02 -1.31519628e+00 5.84411882e-02 -1.10025786e-01
9.12932754e-01 -4.00936097e-01 2.66717255e-01 3.12448382e-01
-2.16881633e-01 -8.06712687e-01 7.31562018e-01 6.07550800e-01
8.76094759e-01 -9.70605075e-01 7.74992347e-01 1.04151264e-01
-1.65840936e+00 -4.11663234e-01 -2.29286849e-01 6.34353906e-02
3.49172056e-01 6.44480765e-01 5.26204444e-02 6.48566008e-01
9.45285916e-01 1.10935354e+00 -2.87560433e-01 1.30066133e+00
1.47965446e-03 5.94746709e-01 1.52115580e-02 3.92528892e-01
4.69630420e-01 3.04867402e-02 4.62909192e-01 1.12115610e+00
6.91021442e-01 6.35148942e-01 1.50214583e-01 5.74971795e-01
5.51924715e-03 -8.44896361e-02 4.86132614e-02 1.04026988e-01
5.41536510e-01 1.09115601e+00 -2.49273747e-01 -7.84505486e-01
-6.25252008e-01 1.07100832e+00 -2.30311200e-01 4.81861949e-01
-1.06498766e+00 -6.84496045e-01 5.36598086e-01 1.99999809e-01
8.93094420e-01 -3.15379232e-01 6.43161312e-02 -1.05347300e+00
1.24094501e-01 -8.45563412e-01 2.77052790e-01 -1.01657975e+00
-1.02075458e+00 6.12587392e-01 -1.72909901e-01 -1.63613808e+00
2.04327926e-01 -1.42346799e-01 -5.41729867e-01 8.73681009e-01
-1.74993908e+00 -9.87899125e-01 -5.64245343e-01 8.30592871e-01
7.57740021e-01 -6.81600422e-02 1.07539505e-01 8.57274950e-01
-5.14750004e-01 7.17056394e-01 -5.83615899e-03 7.30771348e-02
7.15505779e-01 -7.14821339e-01 -1.45292744e-01 1.10760713e+00
-3.12797427e-01 3.64554584e-01 3.82695228e-01 -4.62215751e-01
-1.47449362e+00 -1.27854645e+00 5.22272587e-01 5.96602201e-01
3.06916267e-01 3.06617141e-01 -1.25836897e+00 1.01423919e-01
3.82398933e-01 4.38259065e-01 3.70846957e-01 -4.59508359e-01
-1.54559731e-01 -4.14348394e-01 -1.18203378e+00 3.89376462e-01
9.46998715e-01 -4.96013403e-01 -2.30513796e-01 -2.13760167e-01
9.27619755e-01 -1.34100690e-01 -9.28072989e-01 9.10038888e-01
3.72796416e-01 -1.17665732e+00 8.96686971e-01 -9.46048796e-02
6.51206613e-01 -7.76836395e-01 -4.42281514e-01 -1.12789822e+00
-8.35831881e-01 -5.12235999e-01 -2.26224676e-01 1.18624640e+00
-3.91299743e-03 -1.91703930e-01 4.55700219e-01 1.82134360e-01
-4.22797889e-01 -9.66467917e-01 -9.50911105e-01 -5.44135511e-01
-2.49365643e-01 -4.89857554e-01 7.09366500e-01 9.75860059e-01
-5.45527674e-02 3.04392949e-02 -6.66852951e-01 2.39201441e-01
3.95200640e-01 1.18853226e-01 2.32708842e-01 -3.58316779e-01
-3.43209982e-01 -5.82428873e-01 -5.39231718e-01 -1.60236096e+00
-2.57301539e-01 -5.14945269e-01 1.10517107e-01 -1.54253864e+00
3.84903669e-01 -3.19856852e-01 -9.67305541e-01 3.75684321e-01
-5.81806898e-01 5.17305851e-01 3.19674760e-01 8.00262615e-02
-8.95675302e-01 9.78128791e-01 1.59518266e+00 -4.03309524e-01
-1.96157262e-01 -4.96217668e-01 -7.15713978e-01 3.28429341e-01
5.68877816e-01 5.01554832e-02 -3.44012439e-01 -5.96478045e-01
-2.75935352e-01 5.04470110e-01 4.76414919e-01 -1.39272022e+00
2.37278879e-01 -1.90956116e-01 8.57279301e-01 -8.70270550e-01
3.14912409e-01 -7.18351722e-01 -6.04666881e-02 3.62090945e-01
-3.16218317e-01 -1.06114700e-01 3.46927762e-01 6.97257996e-01
-7.55553544e-01 2.50541598e-01 8.47246408e-01 -6.49719406e-03
-1.20981371e+00 6.91614032e-01 -3.88157278e-01 -3.17788303e-01
1.01569164e+00 -2.76448995e-01 2.46469327e-03 -6.03137136e-01
-6.06674075e-01 2.64762759e-01 3.41367930e-01 4.51299012e-01
1.06757998e+00 -1.61060309e+00 -8.02605450e-01 3.78936321e-01
-1.48515239e-01 -4.74519193e-01 1.11516738e+00 8.97620678e-01
-2.85605490e-01 3.02927554e-01 -5.59084713e-01 -7.16097713e-01
-1.14335930e+00 8.71228099e-01 4.04058009e-01 -1.90075874e-01
-3.47404361e-01 9.54737246e-01 3.00392449e-01 4.97519165e-01
5.78640550e-02 -2.36937091e-01 -2.10503459e-01 -3.37809116e-01
8.78792703e-01 3.51011217e-01 -1.11743264e-01 -8.40276599e-01
-2.97910362e-01 7.72749126e-01 6.70068413e-02 -2.12860536e-02
1.19926369e+00 -4.94714797e-01 1.28865749e-01 -9.64916721e-02
1.50779700e+00 -6.79332167e-02 -1.68457890e+00 -6.51828349e-01
-6.49023116e-01 -1.07815015e+00 5.96177876e-01 -7.35914588e-01
-1.74857378e+00 1.13305569e+00 9.81759548e-01 -4.65867631e-02
2.13453627e+00 -3.35426897e-01 1.27254200e+00 -4.28328902e-01
2.09070593e-01 -8.33302319e-01 3.34109873e-01 2.48851478e-01
8.44004810e-01 -1.16461027e+00 5.88492379e-02 -1.74101934e-01
-4.06497896e-01 9.03695166e-01 7.13307500e-01 7.19280541e-02
7.18441844e-01 2.26455942e-01 -3.33108455e-01 4.12089936e-02
-6.75145030e-01 -2.47372925e-01 6.04367614e-01 5.80080092e-01
4.71695006e-01 -7.43176267e-02 -5.38227141e-01 6.60768092e-01
3.77400726e-01 1.65285498e-01 4.83358242e-02 9.26722825e-01
-7.44512081e-01 -8.87278974e-01 -1.96968138e-01 4.61521894e-01
-4.45999444e-01 -3.09586078e-01 3.57781768e-01 4.49302375e-01
6.86261892e-01 1.17616618e+00 2.08391190e-01 -8.88095617e-01
1.78183079e-01 -6.11822963e-01 2.99607873e-01 3.36733833e-02
-3.35319459e-01 6.11318946e-01 -2.74603724e-01 -6.92809641e-01
-6.60263181e-01 -4.19744909e-01 -1.28050530e+00 -2.78395742e-01
-2.81209499e-01 1.29743725e-01 3.71470064e-01 7.31881261e-01
6.11885488e-01 8.69031966e-01 9.67831075e-01 -1.13853991e+00
-5.78893647e-02 -9.31479812e-01 -5.11394322e-01 4.21571553e-01
6.35603130e-01 -5.82530558e-01 -2.14421019e-01 1.89863205e-01] | [11.236333847045898, -1.7677173614501953] |
424619fd-6c3a-4c86-a3a8-2a2f82e7ea17 | vi-net-view-invariant-quality-of-human | 2008.04999 | null | https://arxiv.org/abs/2008.04999v1 | https://arxiv.org/pdf/2008.04999v1.pdf | VI-Net: View-Invariant Quality of Human Movement Assessment | We propose a view-invariant method towards the assessment of the quality of human movements which does not rely on skeleton data. Our end-to-end convolutional neural network consists of two stages, where at first a view-invariant trajectory descriptor for each body joint is generated from RGB images, and then the collection of trajectories for all joints are processed by an adapted, pre-trained 2D CNN (e.g. VGG-19 or ResNeXt-50) to learn the relationship amongst the different body parts and deliver a score for the movement quality. We release the only publicly-available, multi-view, non-skeleton, non-mocap, rehabilitation movement dataset (QMAR), and provide results for both cross-subject and cross-view scenarios on this dataset. We show that VI-Net achieves average rank correlation of 0.66 on cross-subject and 0.65 on unseen views when trained on only two views. We also evaluate the proposed method on the single-view rehabilitation dataset KIMORE and obtain 0.66 rank correlation against a baseline of 0.62. | ['Adeline Paiement', 'Sion Hannuna', 'Majid Mirmehdi', 'Faegheh Sardari'] | 2020-08-11 | null | null | null | null | ['action-analysis', 'action-assessment', '3d-human-action-recognition'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-1.91922799e-01 -7.51868337e-02 -2.19714746e-01 -2.37400964e-01
-9.37517881e-01 -4.20650512e-01 3.62238765e-01 -4.17958647e-01
-7.38533080e-01 3.48288298e-01 6.83942437e-01 6.55475974e-01
-1.77536413e-01 -3.86635870e-01 -6.41674876e-01 -3.60532880e-01
-2.36404061e-01 3.96955401e-01 2.97590733e-01 -3.90856534e-01
-1.69283211e-01 3.46292108e-01 -1.44921839e+00 3.55501950e-01
9.73448381e-02 9.81130838e-01 -2.97252059e-01 9.11679327e-01
9.26488936e-01 3.46611738e-01 -3.75881791e-01 -6.40686452e-01
4.09035414e-01 -3.40928376e-01 -8.35793853e-01 3.28531176e-01
7.02841043e-01 -5.64841509e-01 -7.20802486e-01 5.45489490e-01
1.08897161e+00 3.32299858e-01 5.75139344e-01 -1.01022553e+00
-3.58440459e-01 2.94894874e-01 -6.16733968e-01 3.29170525e-01
6.17990673e-01 4.71665353e-01 1.02806067e+00 -8.46686721e-01
1.06636572e+00 1.03918874e+00 6.38642311e-01 8.14001799e-01
-9.13145304e-01 -2.98049778e-01 -1.26892462e-01 2.08414942e-01
-8.85597408e-01 -3.15682948e-01 6.12503529e-01 -5.38489401e-01
8.29151988e-01 6.39820769e-02 1.12560713e+00 1.61707318e+00
5.26693046e-01 9.45230067e-01 9.40147698e-01 -2.92235287e-03
5.84375039e-02 -9.36141670e-01 -1.95059314e-01 7.52983987e-01
7.73069113e-02 1.84064567e-01 -9.08649206e-01 2.45999545e-01
8.45484734e-01 6.32781908e-02 -1.89053908e-01 -8.60540330e-01
-1.52602279e+00 4.36267614e-01 7.52146363e-01 1.41362876e-01
-7.68408537e-01 4.69328940e-01 5.85026622e-01 1.30081579e-01
3.15932035e-01 2.65773013e-02 -5.59894264e-01 -5.31991899e-01
-9.83977497e-01 4.96557891e-01 3.13644886e-01 5.38987756e-01
2.77946834e-02 -1.21403821e-01 -3.97195041e-01 6.87507510e-01
3.24749351e-01 4.70203817e-01 5.88291466e-01 -1.14331698e+00
6.72648549e-01 5.17319858e-01 -7.68066421e-02 -7.59166539e-01
-9.10656631e-01 -5.57030082e-01 -5.42131603e-01 5.02967358e-01
6.53542876e-01 -1.01664774e-01 -1.13370717e+00 1.75702178e+00
5.18546641e-01 -3.84933323e-01 -2.18093157e-01 1.43766356e+00
8.89500082e-01 -2.27782682e-01 -2.01422330e-02 3.28828484e-01
1.46769643e+00 -1.25451005e+00 -3.37784439e-01 -1.33777559e-01
3.95159602e-01 -3.38168293e-01 1.18845117e+00 6.72621965e-01
-1.38226712e+00 -7.72751689e-01 -1.00714743e+00 -2.14722902e-01
-1.08665237e-02 5.52228749e-01 3.27298135e-01 2.64019608e-01
-8.94761086e-01 9.84313607e-01 -1.41131699e+00 -4.43809777e-01
4.63939935e-01 5.28624296e-01 -9.71106708e-01 5.36254756e-02
-8.84364665e-01 8.41789722e-01 -7.81395733e-02 3.71456534e-01
-1.22150862e+00 -2.85631150e-01 -7.77416050e-01 -3.94493699e-01
1.01414584e-01 -1.20559704e+00 9.29070532e-01 -7.79648125e-01
-1.56011498e+00 1.15808082e+00 2.97216803e-01 -3.06517184e-01
1.01511931e+00 -7.86596298e-01 -2.82575965e-01 5.21607339e-01
2.30319887e-01 6.03101969e-01 6.72427475e-01 -1.02585423e+00
-4.09462243e-01 -9.11767960e-01 1.12385526e-01 4.04750317e-01
1.91299769e-03 -1.38142005e-01 -9.42245424e-01 -6.11624897e-01
2.47380435e-01 -1.35782993e+00 -2.04224978e-02 3.58354181e-01
-5.40788651e-01 6.43607751e-02 1.72280222e-01 -1.10181129e+00
9.49893117e-01 -1.83278620e+00 9.31293726e-01 4.10406440e-02
2.97530234e-01 8.91690608e-03 -1.27225667e-01 3.80427331e-01
-1.40103996e-01 -3.91244292e-01 -1.88989997e-01 -4.00296420e-01
-6.67525753e-02 5.83650321e-02 5.95225036e-01 9.64015305e-01
-8.81056711e-02 1.09395742e+00 -8.85236979e-01 -3.58346015e-01
2.33131617e-01 6.26157582e-01 -4.92938280e-01 1.36130929e-01
1.81966767e-01 7.71509111e-01 -2.23794922e-01 6.94348454e-01
1.89037368e-01 -1.88771307e-01 4.15559188e-02 -4.28571403e-01
3.66249770e-01 -2.26389408e-01 -9.44817603e-01 2.56599450e+00
-3.20176810e-01 4.79386210e-01 -1.86447322e-01 -4.43354428e-01
4.74120677e-01 2.37458155e-01 9.30189550e-01 -6.07291281e-01
2.36791939e-01 -1.24912569e-02 9.53011364e-02 -7.32134700e-01
2.92600483e-01 6.45859763e-02 -2.23282069e-01 3.59369427e-01
4.32960778e-01 4.57157135e-01 2.13936642e-01 1.01764668e-02
1.36608958e+00 8.47548962e-01 7.81742483e-02 2.17600957e-01
2.77368039e-01 -1.68946952e-01 2.95526087e-01 1.16790876e-01
-5.00900745e-01 1.06355131e+00 3.34195167e-01 -5.19641519e-01
-1.10474741e+00 -1.46812558e+00 3.74900639e-01 9.87683952e-01
-1.18464632e-02 -4.88682091e-01 -8.03647995e-01 -8.65345716e-01
2.10248142e-01 -3.63780484e-02 -1.00272477e+00 -3.17533523e-01
-7.06054270e-01 -4.02844727e-01 6.05812252e-01 9.86023247e-01
5.20756185e-01 -1.08470297e+00 -9.83145237e-01 3.10235526e-02
-3.47233295e-01 -1.02494240e+00 -4.38272566e-01 -3.16252410e-01
-1.04939032e+00 -1.31012988e+00 -1.12758219e+00 -3.84468377e-01
3.64974350e-01 -1.45509496e-01 1.11473393e+00 -7.93521628e-02
-2.60618716e-01 6.74704194e-01 -4.03811336e-01 3.56588483e-01
9.34089795e-02 6.41484335e-02 1.67794779e-01 -9.27955434e-02
3.93800214e-02 -6.07380509e-01 -1.17457187e+00 3.94352198e-01
-5.12409806e-01 -2.07940117e-01 6.63972318e-01 8.10802400e-01
7.90599525e-01 -6.01268828e-01 1.79978281e-01 -7.45165348e-02
3.27017307e-01 -8.05986077e-02 2.29230188e-02 1.60245985e-01
-2.72189558e-01 -1.76398128e-01 1.35064140e-01 -4.27106738e-01
-5.41427970e-01 3.73411834e-01 -2.89086223e-01 -5.88706076e-01
-7.44942874e-02 2.87819475e-01 -1.27109215e-01 6.58404008e-02
7.50040710e-01 -3.76766431e-03 7.03633726e-02 -5.14578879e-01
6.57432377e-01 2.99271822e-01 9.82612431e-01 -3.49257171e-01
6.96959972e-01 8.30934882e-01 1.85163051e-01 -4.83898222e-01
-7.37790108e-01 -6.89373612e-01 -1.23591697e+00 -7.61722028e-01
1.46571863e+00 -1.19805300e+00 -8.10115457e-01 5.91633558e-01
-8.64190996e-01 -5.39665580e-01 -2.94531614e-01 9.92037237e-01
-1.17856252e+00 3.93400729e-01 -7.58833826e-01 -3.05428237e-01
-6.58803165e-01 -1.02405608e+00 1.56888986e+00 -9.67045799e-02
-5.25990069e-01 -6.89547062e-01 4.47583944e-01 8.42639148e-01
-6.81322813e-02 9.74738061e-01 3.53291541e-01 -3.45060706e-01
-1.16220027e-01 -3.71172816e-01 1.63984045e-01 3.17866445e-01
-5.74535653e-02 -2.19748586e-01 -7.29423821e-01 -6.20296717e-01
-4.68745530e-01 -7.33302951e-01 8.62285078e-01 4.71967369e-01
6.76260531e-01 1.11639142e-01 -5.50249871e-03 6.32197142e-01
1.17314792e+00 -4.63079661e-01 7.56834507e-01 6.12358034e-01
1.03623438e+00 5.25056064e-01 6.84795618e-01 3.98140311e-01
4.90381867e-01 1.04924166e+00 6.15256488e-01 -1.59158587e-01
-4.18996096e-01 -2.51593649e-01 5.18412709e-01 7.99048424e-01
-8.05742621e-01 -2.65720021e-02 -7.33276367e-01 5.74640036e-01
-1.77951527e+00 -1.02739632e+00 -1.55509874e-01 2.09064317e+00
4.80219692e-01 1.65954664e-01 7.92755306e-01 1.60260096e-01
2.96813518e-01 1.84909478e-02 -7.53037751e-01 6.60073683e-02
2.01363146e-01 3.55790347e-01 3.84617448e-01 6.57338724e-02
-1.18081379e+00 7.99176216e-01 5.80671644e+00 3.53481233e-01
-1.02986288e+00 1.55684561e-01 4.47906815e-02 -6.74302936e-01
2.97678024e-01 -5.01196086e-01 -2.95515448e-01 1.97847113e-01
7.71410048e-01 3.44407767e-01 1.88586026e-01 7.42908120e-01
2.62557328e-01 -8.48713592e-02 -1.16947126e+00 1.08010316e+00
2.32973352e-01 -6.52428150e-01 -2.74832636e-01 2.14688197e-01
6.19576514e-01 3.96203846e-01 -6.45875186e-02 1.53599009e-01
2.40653548e-02 -8.92630219e-01 8.79963338e-01 8.56527805e-01
6.75127745e-01 -8.57741594e-01 5.89319170e-01 9.74643454e-02
-1.25242579e+00 3.36219892e-02 -1.33857459e-01 2.57330298e-01
3.51191133e-01 6.34650066e-02 -7.87930191e-02 8.17898035e-01
1.00161052e+00 9.31480467e-01 -7.91355073e-01 9.84974325e-01
-3.87115270e-01 2.33940452e-01 -7.34739052e-03 5.26535034e-01
6.39742911e-02 1.29033104e-01 5.51932514e-01 7.82189608e-01
2.31052667e-01 -1.95965916e-01 -3.17789130e-02 2.93458879e-01
1.02789387e-01 -9.10121128e-02 -4.14983690e-01 3.01765949e-01
-4.54253405e-01 1.38412547e+00 -6.79198146e-01 -1.74039021e-01
-2.42021412e-01 1.55111396e+00 3.60765159e-01 2.02806115e-01
-8.93691063e-01 -5.33869155e-02 6.39993906e-01 1.36154726e-01
4.09334809e-01 -4.37832385e-01 1.08897664e-01 -1.28805661e+00
3.15675616e-01 -1.04094458e+00 5.36207020e-01 -9.51786637e-01
-1.25014544e+00 5.03984571e-01 -3.55368890e-02 -1.63290918e+00
-5.43070674e-01 -6.36435986e-01 -3.23206574e-01 7.18478501e-01
-8.12205672e-01 -1.30414474e+00 -5.43417633e-01 7.81729877e-01
4.55910891e-01 4.23715897e-02 6.69888616e-01 3.06946069e-01
-4.55830216e-01 5.69465697e-01 -2.71880150e-01 5.24190068e-01
7.15910852e-01 -1.27504289e+00 6.30608499e-01 6.61349416e-01
1.63837060e-01 6.53103888e-01 5.25379479e-01 -6.92611814e-01
-1.42014647e+00 -8.88249159e-01 5.10808706e-01 -9.74157393e-01
3.85447621e-01 -1.70137942e-01 -4.34307963e-01 5.56858480e-01
2.81013008e-02 2.23543569e-01 6.65373802e-01 -2.31444798e-02
-2.01226100e-01 4.58801985e-02 -1.00979233e+00 5.62309325e-01
1.65791357e+00 -3.91439736e-01 -7.43116617e-01 2.54419059e-01
4.09826398e-01 -8.10065269e-01 -1.41000605e+00 3.43359172e-01
1.33122814e+00 -1.11526370e+00 1.24614429e+00 -9.85512614e-01
8.32474113e-01 -1.98351905e-01 -1.72994047e-01 -1.23366129e+00
-2.26562485e-01 -2.29415223e-01 -1.64258078e-01 5.91011822e-01
2.05007225e-01 1.33489683e-01 1.06089854e+00 2.51348108e-01
-2.29034439e-01 -1.03344738e+00 -1.05602276e+00 -9.49528456e-01
-8.24458152e-02 -4.46251869e-01 1.09500229e-01 4.67697769e-01
-1.78133503e-01 2.10806862e-01 -7.19607234e-01 -4.30427380e-02
7.63894439e-01 1.11659780e-01 1.14272058e+00 -9.81080174e-01
-5.65801859e-01 -1.67815506e-01 -9.28372264e-01 -7.84120858e-01
-1.56344131e-01 -9.35072005e-01 -2.29218788e-02 -1.87035930e+00
2.31629118e-01 4.08096731e-01 -2.50976205e-01 3.33387345e-01
6.45330176e-02 5.47571242e-01 1.72487184e-01 3.05745244e-01
-8.30600381e-01 6.95735216e-01 1.68586850e+00 9.76556316e-02
-7.38746449e-02 1.33050472e-01 -1.39373183e-01 7.03242004e-01
6.21811330e-01 -4.04278457e-01 -2.60784477e-01 -3.40389490e-01
1.82600111e-01 2.52152085e-01 8.04672122e-01 -1.46666551e+00
-1.76035225e-01 2.99028009e-01 9.30264652e-01 -8.36114287e-01
6.25403345e-01 -5.95301151e-01 2.64408857e-01 8.37309122e-01
-3.95591080e-01 4.52697843e-01 -1.73618644e-01 7.37602413e-01
8.09595287e-02 4.37002569e-01 6.65164173e-01 -3.06266785e-01
-5.46361744e-01 6.05268180e-01 1.29287720e-01 3.15354317e-01
8.51497054e-01 -3.89977366e-01 2.15020142e-02 -4.32206154e-01
-1.45620835e+00 1.70982838e-01 6.11653924e-01 6.04957521e-01
6.81960523e-01 -1.72986257e+00 -6.26354575e-01 -6.08606823e-02
3.34826827e-01 -3.50128919e-01 5.32832563e-01 1.08645642e+00
-5.47022939e-01 1.25958681e-01 -8.36351752e-01 -8.43446612e-01
-1.33801520e+00 2.65614629e-01 3.99992257e-01 -3.95666599e-01
-8.84282589e-01 6.37579620e-01 -2.47922808e-01 -4.30321097e-01
1.75808966e-01 -2.80480087e-01 -2.23748729e-01 6.56365603e-02
2.95896173e-01 7.25314319e-01 5.17967753e-02 -1.18942451e+00
-6.03521109e-01 1.08170986e+00 3.50161344e-01 -6.30897641e-01
1.47512674e+00 -1.95442187e-03 3.50757062e-01 4.76518601e-01
1.39014041e+00 -1.65616885e-01 -1.53132272e+00 8.04383773e-03
-2.67243236e-01 -3.29320878e-01 -2.24967584e-01 -7.86988080e-01
-1.42836893e+00 9.37502801e-01 1.07469285e+00 -5.46941519e-01
1.07905197e+00 5.81031963e-02 9.57837760e-01 8.77484530e-02
5.17603815e-01 -1.10920441e+00 6.37974381e-01 1.63370505e-01
1.26578760e+00 -1.17687953e+00 2.77294815e-01 4.74854708e-02
-9.29878354e-01 1.08387697e+00 7.18746305e-01 -3.90483826e-01
4.44104642e-01 -4.17612851e-01 1.36060417e-01 -6.53119445e-01
-4.89439487e-01 -4.73557562e-01 7.40570903e-01 6.75931871e-01
4.84066278e-01 1.29118627e-02 -4.46720570e-01 3.91490936e-01
-3.76058728e-01 3.13482225e-01 1.57013014e-01 9.63779747e-01
-3.33369859e-02 -8.96812081e-01 -6.02336274e-03 2.74619371e-01
-6.74377859e-01 5.75000823e-01 -5.63903987e-01 1.12277877e+00
1.24271289e-01 7.44647622e-01 -2.57539988e-01 -8.51977766e-01
1.01201427e+00 -1.29025191e-01 9.77458358e-01 -4.05785233e-01
-8.92870247e-01 1.25646636e-01 2.56831020e-01 -1.38243675e+00
-7.52991498e-01 -8.61356139e-01 -1.20432782e+00 -1.49661243e-01
9.40646455e-02 -5.33181727e-01 5.26146650e-01 7.35501349e-01
1.97749883e-01 6.14478469e-01 2.90817916e-01 -1.22340322e+00
-6.26769066e-01 -9.69106436e-01 -4.86932427e-01 9.04191434e-01
2.27889359e-01 -9.86474276e-01 -6.42168894e-02 -4.59676571e-02] | [7.118964195251465, -0.677645742893219] |
2d5ef8c0-8c0e-4621-8404-087f89874738 | a-koopman-approach-to-understanding-sequence | 2102.07824 | null | https://arxiv.org/abs/2102.07824v4 | https://arxiv.org/pdf/2102.07824v4.pdf | An Operator Theoretic Approach for Analyzing Sequence Neural Networks | Analyzing the inner mechanisms of deep neural networks is a fundamental task in machine learning. Existing work provides limited analysis or it depends on local theories, such as fixed-point analysis. In contrast, we propose to analyze trained neural networks using an operator theoretic approach which is rooted in Koopman theory, the Koopman Analysis of Neural Networks (KANN). Key to our method is the Koopman operator, which is a linear object that globally represents the dominant behavior of the network dynamics. The linearity of the Koopman operator facilitates analysis via its eigenvectors and eigenvalues. Our method reveals that the latter eigendecomposition holds semantic information related to the neural network inner workings. For instance, the eigenvectors highlight positive and negative n-grams in the sentiments analysis task; similarly, the eigenvectors capture the salient features of healthy heart beat signals in the ECG classification problem. | ['Omri Azencot', 'Ilan Naiman'] | 2021-02-15 | a-koopman-approach-to-understanding-sequence-1 | https://openreview.net/forum?id=4j4qVy8OQA1 | https://openreview.net/pdf?id=4j4qVy8OQA1 | null | ['ecg-classification'] | ['medical'] | [-4.66012321e-02 3.87957543e-01 -2.64298648e-01 -4.59760427e-02
3.27570230e-01 -5.57266831e-01 2.84596324e-01 -1.62097048e-02
-1.99886277e-01 1.07032649e-01 1.46925390e-01 -3.40335965e-01
-7.71245539e-01 -4.00996715e-01 -6.46154761e-01 -9.20778275e-01
-5.97109556e-01 -2.89470434e-01 -2.00869143e-01 -4.74237889e-01
2.05259383e-01 4.68750119e-01 -9.89257455e-01 1.66229084e-01
4.44847077e-01 1.37392759e+00 -3.01623553e-01 6.06544912e-01
4.12008986e-02 1.02850604e+00 -2.23298326e-01 -4.50153440e-01
3.20334941e-01 -5.48862219e-01 -7.39768565e-01 -3.71239215e-01
1.13478966e-01 1.55930340e-01 -4.28373694e-01 1.47015417e+00
1.54901370e-01 -8.97431597e-02 6.46209121e-01 -1.46012008e+00
-5.79791665e-01 6.97227418e-01 -3.22643667e-01 6.12524927e-01
-3.86179477e-01 -3.28560174e-01 1.65829289e+00 -8.86710584e-01
5.08239806e-01 1.00160122e+00 1.03973889e+00 3.10944766e-01
-1.37535799e+00 -3.99132907e-01 -7.93453678e-02 2.40096658e-01
-1.19947290e+00 -2.75876164e-01 1.05306065e+00 -7.40532339e-01
4.94288802e-01 3.01660508e-01 9.58140969e-01 8.10626924e-01
6.18960381e-01 8.07079136e-01 5.77656925e-01 -3.19550395e-01
2.65702695e-01 1.13808163e-01 5.62542856e-01 7.32263565e-01
1.62863716e-01 3.16986889e-02 -6.23614490e-01 -9.35889035e-02
9.51136172e-01 2.38132253e-01 -4.08410251e-01 -6.33349836e-01
-1.07982075e+00 8.63113225e-01 4.55592036e-01 4.75718498e-01
-5.80335677e-01 2.88463235e-01 5.64037740e-01 5.35750210e-01
2.71459401e-01 4.73849386e-01 -6.07730448e-01 -4.49933670e-02
-6.80020392e-01 -5.50840516e-03 8.78209293e-01 4.30458516e-01
7.61928380e-01 -1.73921399e-02 2.96648771e-01 4.73207474e-01
4.38502610e-01 1.22556619e-01 3.89608949e-01 -1.06231606e+00
1.29060969e-01 7.50496089e-01 -4.69891667e-01 -1.40674841e+00
-7.00763047e-01 -8.65975797e-01 -1.11946726e+00 -1.99957248e-02
5.18241882e-01 -3.56753856e-01 -1.15465052e-01 1.88510656e+00
4.16640332e-03 1.09143024e-02 1.25188269e-02 8.95802081e-01
4.37387973e-01 2.48554796e-01 -2.57507622e-01 -2.21963450e-01
1.27266669e+00 -5.23004234e-01 -9.04074252e-01 -1.22881345e-02
5.88375509e-01 -1.78264037e-01 7.61629462e-01 2.62788266e-01
-1.10308659e+00 -1.85406789e-01 -9.41794991e-01 2.07421347e-01
-2.90231794e-01 -8.08131602e-03 5.15859902e-01 3.59698504e-01
-1.04812014e+00 8.94719362e-01 -8.80162477e-01 -1.74907446e-01
4.80541110e-01 3.45416456e-01 -2.10906655e-01 6.37577653e-01
-1.25760067e+00 5.68059742e-01 2.80958623e-01 3.31753165e-01
-4.39049453e-01 -1.08352590e+00 -5.45283139e-01 4.57689881e-01
3.44562650e-01 -5.31813562e-01 9.84842479e-01 -1.30267870e+00
-1.45226634e+00 5.41957200e-01 2.79368088e-02 -5.71969271e-01
1.57363445e-01 -1.04397602e-01 8.78996775e-03 3.95964712e-01
-4.68035698e-01 2.24194959e-01 9.23985720e-01 -8.96480739e-01
-7.95204192e-02 -6.07769787e-01 -2.45466456e-02 -2.04208463e-01
-9.09081638e-01 -1.68342233e-01 1.41609788e-01 -7.74261475e-01
5.11541128e-01 -8.86775434e-01 -5.86171262e-02 -6.78158477e-02
-4.48557854e-01 -2.73963302e-01 5.83944917e-01 -5.58280885e-01
1.48145020e+00 -2.47855043e+00 3.71686965e-01 7.01016486e-01
7.81770349e-01 -1.64546102e-01 5.85897937e-02 5.82468092e-01
-6.75978482e-01 2.52862513e-01 9.71592367e-02 2.12197646e-01
1.24084882e-01 -1.64860249e-01 -4.77194875e-01 7.14934886e-01
3.15854669e-01 1.14314604e+00 -6.75034761e-01 -1.74558029e-01
-2.01914728e-01 5.11415243e-01 -6.37432516e-01 -3.33446652e-01
1.53629631e-01 -7.24706128e-02 -3.94330502e-01 1.75564855e-01
1.17807239e-01 -5.04055977e-01 3.94525886e-01 -7.02399433e-01
-1.27006978e-01 2.27900550e-01 -1.05781353e+00 1.46794689e+00
2.06954721e-02 1.12584758e+00 1.96491480e-01 -1.55509984e+00
6.55899286e-01 4.47804153e-01 1.03548157e+00 -4.36961085e-01
4.79476452e-01 -4.95224632e-02 5.19280434e-01 -5.05911052e-01
4.11540829e-02 -6.81308508e-02 3.36598665e-01 6.23427331e-01
2.70354569e-01 7.87973225e-01 1.46854773e-01 4.63095874e-01
1.10496128e+00 -4.37435836e-01 4.90040660e-01 -1.00710213e+00
2.53775209e-01 -2.62000382e-01 3.13490391e-01 5.83463669e-01
-4.81337726e-01 1.92534402e-01 1.26872241e+00 -4.77400512e-01
-8.99220228e-01 -9.85714316e-01 -3.99953455e-01 1.02785540e+00
-2.95176506e-01 -6.18623018e-01 -1.09647000e+00 -2.69570798e-01
-1.58527657e-01 -7.96099827e-02 -9.63713288e-01 -6.09237611e-01
-3.11958879e-01 -6.40031993e-01 5.01750231e-01 5.31597853e-01
3.29765350e-01 -9.21562374e-01 -9.92674589e-01 -2.79999543e-02
-4.47523855e-02 -8.66436958e-01 -4.33952510e-01 2.81880170e-01
-1.20974612e+00 -1.05942798e+00 -5.10459900e-01 -5.86063564e-01
5.82828641e-01 -8.70243683e-02 9.48179066e-01 -9.56708863e-02
-4.33634579e-01 5.74981630e-01 2.29534935e-02 -6.43473864e-01
-9.55901593e-02 7.30191395e-02 2.93110341e-01 5.11617184e-01
2.38356680e-01 -9.46724474e-01 -6.52963936e-01 1.65014222e-01
-8.14159572e-01 -2.26225078e-01 6.02477133e-01 8.07443619e-01
2.68133074e-01 4.27376837e-01 5.29277563e-01 -5.41522145e-01
9.12503660e-01 -3.74220341e-01 -3.90113622e-01 1.33068383e-01
-5.68953156e-01 3.25226992e-01 6.24597728e-01 -5.57944834e-01
-4.55400467e-01 -1.02528803e-01 3.48098129e-01 -5.16108871e-01
3.08707356e-01 7.26519644e-01 9.58314463e-02 -1.59876540e-01
7.36620665e-01 1.51676223e-01 2.80286700e-01 -3.58184844e-01
1.44436315e-01 1.11338526e-01 3.70478928e-01 -3.39048833e-01
6.95072293e-01 7.10360348e-01 2.86470801e-01 -1.11802971e+00
-8.65434289e-01 -4.09380704e-01 -6.94402337e-01 -5.18722773e-01
7.26158917e-01 -5.31989336e-01 -1.67449415e+00 2.61361986e-01
-1.03364503e+00 -9.81523767e-02 -3.81418824e-01 4.52469379e-01
-4.53185439e-01 1.66275457e-01 -7.85357237e-01 -1.09584939e+00
-3.35376769e-01 -5.60251653e-01 6.30325556e-01 1.06594890e-01
-4.78092104e-01 -1.43386161e+00 1.43310726e-01 -1.49187952e-01
3.45682561e-01 1.53554082e-01 1.19192910e+00 -5.79198420e-01
-2.69758344e-01 -2.61759102e-01 -7.23345205e-02 4.58429813e-01
-4.22366828e-01 5.67834750e-02 -9.37697172e-01 -5.01980409e-02
4.25611794e-01 8.01047385e-02 7.55734861e-01 8.22538733e-01
1.11193085e+00 -5.48985660e-01 1.06086142e-01 4.92908955e-01
1.30656755e+00 3.29577662e-02 3.69782269e-01 1.77337229e-01
6.45516992e-01 1.11556864e+00 -2.27909654e-01 5.15663743e-01
1.25310436e-01 1.90940946e-01 4.11720246e-01 -5.85973747e-02
4.51074690e-01 -2.28529111e-01 4.94850755e-01 1.00042641e+00
-3.20651025e-01 4.12346154e-01 -1.10653841e+00 9.34964120e-02
-2.11725831e+00 -1.10333538e+00 -1.88319340e-01 1.80583799e+00
4.85037237e-01 1.67204142e-01 1.70055330e-01 4.63448346e-01
4.73069698e-01 5.40028922e-02 -6.50212824e-01 -1.26520917e-01
-8.01534429e-02 -6.03618426e-03 3.43006611e-01 1.88644901e-01
-8.93931150e-01 4.37993318e-01 6.90694094e+00 4.92585689e-01
-1.18129420e+00 1.14406049e-02 3.73480797e-01 1.78855390e-03
-9.97870192e-02 -1.14763208e-01 -2.00866446e-01 1.23898119e-01
8.44190240e-01 -1.57312855e-01 5.79750061e-01 9.44778681e-01
2.73141921e-01 2.12650046e-01 -1.28929412e+00 1.10965645e+00
-1.26958668e-01 -1.50423932e+00 -3.43280822e-01 6.09705985e-01
4.62122411e-01 1.14501707e-01 4.66581523e-01 -9.14921239e-02
-1.13226846e-01 -8.35966468e-01 7.66674042e-01 7.51030505e-01
1.23021834e-01 -6.59789860e-01 4.44455653e-01 4.47629899e-01
-1.03568041e+00 -3.97421390e-01 -2.70398885e-01 -2.83894032e-01
-2.58331269e-01 8.96998167e-01 -5.02303481e-01 6.72693551e-02
7.30020165e-01 1.15943670e+00 -3.62348050e-01 5.89550376e-01
1.86031535e-01 7.19923139e-01 -1.53891191e-01 -3.71311679e-02
5.66142909e-02 -5.59813738e-01 6.27229095e-01 9.75528359e-01
-6.85618967e-02 1.66576430e-02 -3.32933217e-01 1.30142403e+00
-1.84646219e-01 1.50096938e-01 -6.09451234e-01 -6.01127028e-01
-6.26816824e-02 1.46676981e+00 -8.64656985e-01 7.28902640e-03
-2.50116736e-01 4.46313769e-01 2.18092144e-01 6.81641519e-01
-4.41798776e-01 -4.26790029e-01 9.61201489e-01 -4.48943973e-02
-4.00567390e-02 -2.76173621e-01 -6.41130626e-01 -1.12545359e+00
2.24376962e-01 -7.49126792e-01 2.03717276e-01 -3.95374298e-01
-1.28489268e+00 2.31614977e-01 -2.59936303e-01 -9.54465330e-01
-8.59952047e-02 -1.02946353e+00 -8.46343100e-01 5.54598689e-01
-1.01446581e+00 -4.47243869e-01 9.96805727e-02 7.81733930e-01
-2.27141809e-02 -1.36624157e-01 4.33213890e-01 3.29376966e-01
-1.10173106e+00 4.55875754e-01 9.42803547e-02 5.25254905e-01
6.58604652e-02 -1.19371748e+00 -5.19329868e-02 7.55356967e-01
3.67238820e-01 1.31146801e+00 6.86464548e-01 -1.19597979e-01
-1.55957067e+00 -5.89977086e-01 8.24246407e-01 -4.12815660e-01
1.25369382e+00 -4.12930965e-01 -8.95631552e-01 5.87772131e-01
2.20176168e-02 2.35835537e-02 7.63578773e-01 3.96889180e-01
-3.54279637e-01 -2.38578022e-01 -3.18030357e-01 6.67067409e-01
9.38854992e-01 -7.95026422e-01 -2.79564738e-01 1.77504838e-01
4.06953990e-01 1.39421761e-01 -9.64994252e-01 3.63223016e-01
9.67732191e-01 -1.07047045e+00 9.61051941e-01 -1.13592649e+00
6.21814549e-01 2.40903467e-01 -2.18264386e-02 -1.22264755e+00
-4.44516748e-01 -7.80620754e-01 -2.71584600e-01 7.40107894e-01
4.10210222e-01 -7.43670881e-01 6.80889726e-01 5.13638079e-01
1.24990456e-02 -1.03784978e+00 -7.29872346e-01 -7.18934000e-01
-3.50577175e-03 -5.37856460e-01 -1.03818104e-01 1.19229734e+00
6.20362818e-01 4.18382645e-01 -3.83243114e-02 -3.48022580e-02
5.35612285e-01 -2.11367145e-01 1.37804046e-01 -1.59805369e+00
-2.78824568e-01 -9.94988859e-01 -5.68514109e-01 -6.82400346e-01
3.72840196e-01 -1.12793970e+00 -2.66547978e-01 -9.00951326e-01
3.06922793e-01 1.51420454e-03 -5.98716497e-01 2.75522053e-01
4.24220860e-01 6.01380318e-02 2.45041192e-01 3.46722573e-01
-3.57088298e-01 3.30380559e-01 1.06859410e+00 1.09911658e-01
-2.89223224e-01 -4.79065143e-02 -7.60707498e-01 1.03489447e+00
7.76857376e-01 -3.80787283e-01 -4.97700781e-01 -1.80430755e-01
1.04166746e+00 -2.82212734e-01 6.79940403e-01 -7.62871981e-01
3.28514457e-01 1.25279322e-01 1.36627287e-01 3.41307670e-02
8.22662860e-02 -8.76708210e-01 -1.68118894e-01 7.74342954e-01
-6.32818103e-01 2.32234955e-01 -1.69680178e-01 5.38128614e-01
-3.39365393e-01 -6.06792942e-02 6.79620087e-01 -1.12326732e-02
-1.72046900e-01 3.32754731e-01 -4.32722747e-01 2.68234581e-01
6.48271799e-01 1.87302232e-02 -3.47450793e-01 -4.13325131e-01
-7.64596224e-01 -1.22128099e-01 -2.21799120e-01 1.51016533e-01
5.05256236e-01 -1.30114603e+00 -1.66434273e-01 2.10609734e-01
-2.34796271e-01 -4.84118760e-01 2.34109536e-01 1.64458644e+00
-2.32614025e-01 4.79516804e-01 -2.26155505e-01 -7.60027468e-01
-1.10346484e+00 3.34590793e-01 7.10637867e-01 -4.07854430e-02
-5.44317067e-01 6.98928297e-01 4.49590504e-01 -3.09023540e-02
3.33844662e-01 -3.94774705e-01 -4.82656747e-01 5.06272197e-01
3.90122920e-01 5.53973913e-01 1.11063092e-03 -5.13368368e-01
-3.82582068e-01 5.40817082e-01 2.76903361e-01 -2.07419008e-01
1.33727849e+00 3.56708206e-02 -5.69818318e-01 8.36292326e-01
1.53387213e+00 -2.99402535e-01 -9.36377764e-01 -3.44590366e-01
3.32954675e-01 1.88416302e-01 2.11558864e-01 -1.56045139e-01
-1.18379653e+00 1.26598883e+00 5.31471014e-01 8.29207957e-01
1.18029857e+00 1.00078389e-01 4.25019473e-01 8.78228724e-01
-1.81648314e-01 -1.22936332e+00 3.56327623e-01 5.92815340e-01
7.57210672e-01 -8.39621603e-01 -3.97474021e-01 -2.37585112e-01
-4.18713480e-01 1.44012368e+00 1.84040725e-01 -2.33115152e-01
1.21116054e+00 2.28562161e-01 3.71530391e-02 -4.88832682e-01
-7.81810582e-01 9.89844054e-02 4.31945354e-01 -2.29956377e-02
3.79404932e-01 8.61137956e-02 -1.47233829e-01 8.11572134e-01
-4.02923018e-01 -3.60161334e-01 4.19793397e-01 6.09363496e-01
-2.97951818e-01 -4.73803252e-01 -1.90321445e-01 3.64247769e-01
-5.30400276e-01 -2.06793204e-01 -5.57075143e-01 5.30138791e-01
-5.91158010e-02 5.52841604e-01 1.72923371e-01 -5.37468731e-01
2.96507269e-01 5.02064884e-01 2.63550580e-01 -2.48726115e-01
-4.55848902e-01 4.81162220e-02 -5.43905795e-01 -7.30733871e-01
-5.40319979e-01 -7.25083470e-01 -1.26501393e+00 -2.33306348e-01
-1.34963512e-01 6.30499348e-02 8.99003804e-01 1.12412024e+00
3.22330594e-01 7.08498716e-01 5.44712126e-01 -4.80511755e-01
-6.02554917e-01 -7.12007761e-01 -8.75035107e-01 3.50169539e-01
6.24471664e-01 -4.74733174e-01 -4.36650783e-01 2.39249840e-01] | [7.928850173950195, 3.588822603225708] |
49af5643-010a-4701-8938-11b1448a7fda | prediction-of-reynolds-stresses-in-high-mach | 1808.07752 | null | http://arxiv.org/abs/1808.07752v1 | http://arxiv.org/pdf/1808.07752v1.pdf | Prediction of Reynolds Stresses in High-Mach-Number Turbulent Boundary Layers using Physics-Informed Machine Learning | Modeled Reynolds stress is a major source of model-form uncertainties in
Reynolds-averaged Navier-Stokes (RANS) simulations. Recently, a
physics-informed machine-learning (PIML) approach has been proposed for
reconstructing the discrepancies in RANS-modeled Reynolds stresses. The merits
of the PIML framework has been demonstrated in several canonical incompressible
flows. However, its performance on high-Mach-number flows is still not clear.
In this work we use the PIML approach to predict the discrepancies in RANS
modeled Reynolds stresses in high-Mach-number flat-plate turbulent boundary
layers by using an existing DNS database. Specifically, the discrepancy
function is first constructed using a DNS training flow and then used to
correct RANS-predicted Reynolds stresses under flow conditions different from
the DNS. The machine-learning technique is shown to significantly improve
RANS-modeled turbulent normal stresses, the turbulent kinetic energy, and the
Reynolds-stress anisotropy. Improvements are consistently observed when
different training datasets are used. Moreover, a high-dimensional
visualization technique and distance metrics are used to provide a priori
assessment of prediction confidence based only on RANS simulations. This study
demonstrates that the PIML approach is a computationally affordable technique
for improving the accuracy of RANS-modeled Reynolds stresses for
high-Mach-number turbulent flows when there is a lack of experiments and
high-fidelity simulations. | ['Jian-Xun Wang', 'Lian Duan', 'Heng Xiao', 'Junji Huang'] | 2018-08-19 | null | null | null | null | ['physics-informed-machine-learning'] | ['graphs'] | [-3.91563594e-01 -9.29646671e-01 1.54639423e-01 -8.10249709e-03
-5.23812532e-01 -4.97787356e-01 6.20830774e-01 1.92777708e-01
-1.56800434e-01 1.00810087e+00 -2.18250901e-01 -7.90644169e-01
-3.19545716e-01 -4.86016572e-01 -4.56948839e-02 -7.57066429e-01
-3.14098746e-01 3.92888755e-01 -5.43519016e-03 -3.03211540e-01
6.25165522e-01 1.21043444e+00 -1.63048315e+00 -6.10712543e-02
1.03238666e+00 1.00560629e+00 -3.25170398e-01 9.37999010e-01
-2.11297095e-01 6.57697678e-01 -1.17328368e-01 1.30934104e-01
4.42291111e-01 -6.03567183e-01 -5.18954694e-01 2.76551638e-02
2.91020602e-01 -3.88714850e-01 2.49792725e-01 6.60964727e-01
3.60882938e-01 8.08087885e-01 8.63885403e-01 -4.60259318e-01
3.13466221e-01 -2.20016003e-01 -3.82235289e-01 3.51362377e-01
4.61492352e-02 4.43539947e-01 6.37883604e-01 -1.16791141e+00
3.65132302e-01 1.01575649e+00 1.09840536e+00 2.73481339e-01
-1.23600364e+00 -2.27702051e-01 -7.60919392e-01 -3.50519389e-01
-1.16554582e+00 -1.03177823e-01 9.09983456e-01 -1.22853315e+00
8.37945223e-01 6.16356909e-01 5.58823168e-01 1.99631140e-01
5.12259066e-01 -5.42069077e-01 1.36375129e+00 -4.22264546e-01
5.99298000e-01 2.28310943e-01 -6.33150041e-02 3.38108003e-01
7.44935155e-01 8.20588946e-01 -1.40559167e-01 -4.18699324e-01
1.06313229e+00 -3.21283400e-01 -9.56357569e-02 -2.84981340e-01
-6.34179056e-01 8.33636820e-01 7.52191246e-02 5.58207393e-01
-1.73899725e-01 -2.66413301e-01 6.53875351e-01 -5.12291938e-02
1.02928007e+00 1.12150574e+00 -7.88777471e-01 -6.74514890e-01
-1.28628135e+00 4.31793630e-01 9.64745581e-01 9.44142863e-02
5.76133013e-01 6.66073382e-01 5.51410377e-01 6.60035491e-01
3.71247321e-01 6.49719059e-01 3.40842813e-01 -1.26238453e+00
7.62642622e-02 2.83043593e-01 6.94167018e-01 -1.11693215e+00
-5.20711653e-02 -4.27199870e-01 -8.57424319e-01 9.86060321e-01
5.26883304e-01 -4.86877263e-01 -4.86917764e-01 6.48117304e-01
4.80504632e-01 5.42607486e-01 5.02964079e-01 1.41865015e+00
4.84306097e-01 6.89321101e-01 1.85290202e-01 -3.71271968e-01
6.84708893e-01 -8.60719919e-01 -4.94841397e-01 1.15374520e-01
8.63242030e-01 -1.20885336e+00 6.35040998e-01 9.13601890e-02
-1.05270863e+00 -8.78736675e-01 -8.53911459e-01 2.74188757e-01
-3.93363625e-01 1.59185037e-01 4.15135503e-01 6.81877792e-01
-6.92337751e-01 1.78133011e+00 -9.74985123e-01 -1.08612059e-02
-2.40495875e-01 -3.04421604e-01 -2.47645557e-01 5.11170566e-01
-6.64573193e-01 9.83635068e-01 1.00540616e-01 2.93963164e-01
-4.67406720e-01 -1.16716540e+00 -8.16668689e-01 1.28454924e-01
-2.74831057e-01 -5.26566684e-01 1.19604790e+00 -3.99402082e-01
-1.91227651e+00 2.91587383e-01 -1.10851742e-01 -2.95399986e-02
1.13402867e+00 -2.35171035e-01 -2.83487290e-01 2.61647820e-01
-1.28875732e-01 -4.30189162e-01 3.93289477e-01 -1.85152650e+00
-2.27352858e-01 2.36808971e-01 -4.77671266e-01 1.97365597e-01
3.93161327e-01 -1.10341072e-01 4.50908035e-01 -5.47513306e-01
3.28116059e-01 -7.98584223e-01 -5.95742643e-01 -1.39442533e-01
1.49557122e-03 4.23921108e-01 8.08603466e-01 -6.50771320e-01
1.00299597e+00 -1.88204181e+00 -2.73846596e-01 2.18419656e-01
-1.60966814e-01 7.92321146e-01 5.86613536e-01 6.30080044e-01
-2.80174613e-01 4.13697928e-01 -6.59378052e-01 -3.73748481e-01
-5.40862560e-01 2.36905262e-01 -7.07988918e-01 4.37621802e-01
2.85711735e-01 1.38864219e-01 -8.86328757e-01 -2.73126096e-01
1.00461292e+00 4.14567739e-01 -4.80816752e-01 5.77560782e-01
3.53514642e-01 1.22814465e+00 -5.79402335e-02 1.24895096e-01
1.31190419e+00 1.05931573e-01 2.13875026e-02 -5.05011342e-02
-8.72807384e-01 6.28221333e-02 -1.14823186e+00 9.21422243e-01
-8.75012636e-01 4.19577360e-01 1.85371414e-01 -6.67444885e-01
1.40273631e+00 1.58422664e-01 5.55868566e-01 -1.44403309e-01
2.17835978e-01 6.97301984e-01 1.91823188e-02 -6.68899298e-01
5.43592453e-01 -9.42236483e-01 6.88317657e-01 9.42044780e-02
-2.99695015e-01 -8.95250201e-01 1.54339671e-01 -1.81261063e-01
3.80040079e-01 1.39305428e-01 1.79927498e-01 -8.05345774e-01
9.88659739e-01 1.54726177e-01 3.23431581e-01 4.51477110e-01
4.61844960e-03 7.37033010e-01 3.49692047e-01 -7.60448694e-01
-1.40091741e+00 -8.10480535e-01 -7.50394166e-01 3.60325694e-01
2.61215150e-01 -5.20618141e-01 -3.00219178e-01 -1.59010619e-01
2.21356079e-01 8.81966889e-01 -2.95021057e-01 1.25276372e-01
-6.15670979e-01 -5.24719834e-01 2.11964771e-01 5.27676404e-01
4.99304026e-01 -6.69706702e-01 -6.29957139e-01 1.72140062e-01
5.80293596e-01 -1.22216487e+00 3.14768136e-01 -8.99818912e-02
-1.55896890e+00 -1.31096017e+00 -3.83332342e-01 -5.55604100e-02
4.75481868e-01 -9.69289988e-02 1.02973390e+00 4.37911659e-01
-2.09459409e-01 2.36381739e-01 -2.07631245e-01 -4.21773493e-02
-9.37147439e-01 -5.52553356e-01 1.52154446e-01 -2.52661228e-01
-2.19458491e-01 -3.74965161e-01 -5.54098785e-01 5.78998089e-01
-6.83536947e-01 -2.20046178e-01 -4.40954268e-02 6.70849800e-01
3.27973038e-01 1.49508134e-01 2.76555747e-01 -7.40031421e-01
5.01597404e-01 -3.45590055e-01 -8.59846175e-01 -3.25167537e-01
-8.31605911e-01 -2.71953166e-01 1.15539169e+00 -1.64176479e-01
-1.45081544e+00 -5.37460387e-01 -2.82476962e-01 -8.51181626e-01
-5.18893778e-01 6.73866808e-01 6.24877810e-01 -2.93942928e-01
6.38314068e-01 -1.78613305e-01 4.13373560e-02 -1.02589214e+00
-3.61022979e-01 3.75352651e-01 2.02416733e-01 -1.11144567e+00
7.30606437e-01 2.82783508e-01 7.41892040e-01 -1.38781929e+00
-3.63813251e-01 -7.25255251e-01 -8.76673937e-01 -3.33298683e-01
5.62759876e-01 -4.57459122e-01 -7.10574806e-01 5.70182145e-01
-1.03111792e+00 -4.34632540e-01 -6.45847738e-01 1.17319179e+00
-3.22360486e-01 6.41198933e-01 -9.04383957e-01 -1.23156869e+00
-1.39734745e-01 -1.32982397e+00 7.25500822e-01 2.51134902e-01
-1.40126422e-01 -1.74967062e+00 4.01117593e-01 1.30157858e-01
8.51939201e-01 6.54643476e-01 8.36608112e-01 -3.62165123e-01
1.17432095e-01 -4.36741635e-02 -2.80023366e-01 6.77198470e-01
1.75032437e-01 6.93059087e-01 -1.03198016e+00 -1.94207549e-01
3.39134723e-01 7.03306198e-02 5.95088303e-01 4.05504823e-01
9.18621004e-01 -1.99260935e-01 -1.35115013e-02 5.48069417e-01
1.59878302e+00 1.63061842e-02 8.58347863e-02 -8.53265449e-03
5.87987185e-01 4.40952688e-01 4.35007453e-01 6.52599990e-01
-5.31324029e-01 2.93054402e-01 7.92617798e-02 -2.95352161e-01
2.41141524e-02 -6.84693232e-02 2.87844446e-02 1.00824153e+00
-8.09797227e-01 3.11337262e-01 -1.20162213e+00 1.19494222e-01
-1.37529564e+00 -7.71549165e-01 -6.46080494e-01 2.27090454e+00
4.58652407e-01 9.32415277e-02 -4.92574900e-01 2.86476821e-01
4.89138931e-01 1.03846855e-01 -3.68944138e-01 -1.18780720e+00
3.19194525e-01 4.53392655e-01 5.01320362e-01 9.83628213e-01
-1.11166322e+00 4.06450361e-01 6.19590616e+00 3.86685997e-01
-1.76615775e+00 -1.22376829e-01 6.51522458e-01 5.25048733e-01
1.14501186e-01 1.84408024e-01 -5.06670356e-01 3.98408681e-01
1.22809994e+00 -1.18995853e-01 1.37226686e-01 1.01840639e+00
8.48635554e-01 -4.04568017e-01 -5.09621561e-01 8.09721351e-01
-3.90206695e-01 -1.60704768e+00 -1.31143659e-01 -7.62196928e-02
9.08406019e-01 -1.95808247e-01 -2.98186928e-01 8.32876489e-02
-1.67494118e-01 -8.79971683e-01 2.64727533e-01 8.49589884e-01
1.14194822e+00 -6.80709183e-01 1.16389763e+00 2.94078588e-01
-1.33088434e+00 3.96409422e-01 -3.69303107e-01 -5.43995917e-01
4.01492268e-01 9.19182777e-01 -9.09874976e-01 9.77183759e-01
5.88570952e-01 9.02635813e-01 -2.08807588e-01 1.15024924e+00
1.23718418e-01 9.21177089e-01 -2.41961405e-01 3.83528352e-01
2.68326551e-01 -8.06621850e-01 7.16578901e-01 1.25260699e+00
6.89245284e-01 4.59797770e-01 2.01102644e-01 7.81505644e-01
4.44129229e-01 2.87880063e-01 -6.77808225e-01 1.77009746e-01
-5.69694936e-02 1.31755042e+00 -7.58521199e-01 -3.55226725e-01
1.30089030e-01 2.18197018e-01 -3.88443649e-01 4.40413892e-01
-5.83968759e-01 -1.04488529e-01 1.21449327e+00 4.85440493e-01
2.48350669e-02 -8.10596645e-01 -6.53670013e-01 -1.01899922e+00
-5.76106191e-01 -1.61467850e-01 1.18989885e-01 -8.91559184e-01
-1.57787955e+00 5.34417570e-01 2.91203141e-01 -1.42279828e+00
-2.71380633e-01 -1.19507480e+00 -9.89751995e-01 1.48923731e+00
-1.66152740e+00 -3.22558612e-01 -1.49820045e-01 -9.00328457e-02
5.36351085e-01 -2.94237342e-02 8.30625594e-01 2.24902146e-02
-3.25717688e-01 -2.07478732e-01 6.67275846e-01 -1.54064167e-02
4.86950099e-01 -1.34821510e+00 1.22723453e-01 6.76566422e-01
-7.31172919e-01 6.94201767e-01 1.11687243e+00 -9.37137187e-01
-9.94246960e-01 -9.53796506e-01 7.83796668e-01 6.92083500e-03
6.94234490e-01 4.33246434e-01 -1.37036133e+00 1.33258849e-01
1.42427795e-02 6.92664623e-01 6.99280977e-01 -1.96728900e-01
2.25401238e-01 6.00760952e-02 -1.42409265e+00 2.83067167e-01
2.43774354e-01 -3.61521751e-01 -6.37466311e-01 -1.22815430e-01
-4.27403934e-02 -7.88527310e-01 -1.12136722e+00 9.20247793e-01
2.72869796e-01 -1.11548388e+00 7.39319861e-01 -1.02877831e+00
6.84462130e-01 -4.05558646e-01 1.22969441e-01 -1.42238581e+00
2.79888719e-01 -4.46693748e-01 2.14579642e-01 1.04405332e+00
1.49361446e-01 -7.56044626e-01 5.06111264e-01 9.72224355e-01
-1.68018818e-01 -8.83770168e-01 -1.13521135e+00 -7.24125206e-01
6.05369925e-01 -6.95119023e-01 3.49812582e-02 1.21041894e+00
-8.27325433e-02 -3.29392910e-01 -2.38360226e-01 3.44797164e-01
4.79045808e-01 1.47470281e-01 6.95554793e-01 -1.52626777e+00
8.33669025e-03 -2.26136044e-01 -2.64061779e-01 -5.45308292e-01
2.65185267e-01 -6.44042611e-01 9.52995792e-02 -1.00036478e+00
-5.98843098e-01 -7.78644085e-01 6.64148480e-02 -2.31075987e-01
3.99189629e-02 2.56281756e-02 1.63832456e-01 4.65549082e-01
3.20556074e-01 5.58890164e-01 1.36982524e+00 6.47949874e-01
-2.96080798e-01 1.52200788e-01 4.01417941e-01 8.18226457e-01
1.04842091e+00 -3.07509005e-01 9.15365368e-02 1.82053387e-01
-3.04227322e-01 2.58556485e-01 3.40277940e-01 -1.50420797e+00
-5.77225387e-02 -4.25931334e-01 6.52762532e-01 -3.36577922e-01
4.56873514e-02 -8.93984079e-01 -3.87450419e-02 4.53351319e-01
-2.28449494e-01 1.60437420e-01 9.63461816e-01 1.43645585e-01
-2.56053388e-01 -2.92586595e-01 1.24995399e+00 -3.65970343e-01
-3.53632271e-01 -2.01243207e-01 -6.17612839e-01 6.73314556e-02
9.24216688e-01 -4.22370248e-02 -4.06459495e-02 1.16961123e-02
-4.74196702e-01 -4.79719192e-01 6.14955127e-01 -2.73725063e-01
5.45970619e-01 -7.76146948e-01 -4.66545552e-01 5.88067830e-01
-5.55639029e-01 9.47714671e-02 4.43802953e-01 8.48758638e-01
-1.60154068e+00 2.34755024e-01 -1.73028544e-01 -6.16252959e-01
-1.00713563e+00 3.02935034e-01 9.15087461e-01 -3.95349532e-01
-3.68178874e-01 5.34457088e-01 -3.04492503e-01 -6.44979656e-01
-7.02323794e-01 -5.08710146e-01 -6.10483363e-02 -1.67746842e-01
2.50902265e-01 1.05428183e+00 3.49021405e-01 -1.17726433e+00
-2.38832403e-02 1.11458397e+00 8.09644341e-01 -4.99630384e-02
9.99879301e-01 -1.27131805e-01 -4.84453887e-02 7.27475286e-01
1.12452066e+00 3.41090471e-01 -1.56177127e+00 4.48459566e-01
-1.21335387e-01 -1.03313029e+00 3.70257586e-01 -7.05530107e-01
-1.00704384e+00 1.12751472e+00 3.32959086e-01 2.36674726e-01
6.55754864e-01 -5.75561285e-01 6.43560708e-01 -3.25678103e-02
-9.96509492e-02 -7.42783070e-01 -1.32224262e-01 8.51450205e-01
8.56480598e-01 -1.25500059e+00 2.12908223e-01 -5.67647278e-01
-7.40327656e-01 1.79748464e+00 6.32340848e-01 -6.88872710e-02
9.88974929e-01 5.32428920e-01 6.99344158e-01 2.19855547e-01
-1.84871584e-01 2.45420054e-01 3.01321507e-01 -3.86218019e-02
7.39633501e-01 -2.05201805e-01 -5.12427032e-01 -6.53909519e-02
-1.41366079e-01 -7.39306360e-02 4.32747394e-01 9.08318043e-01
-2.37582192e-01 -9.92621362e-01 -7.52547860e-01 2.09958866e-01
-3.77723068e-01 4.82891351e-02 1.60655200e-01 7.56970823e-01
1.30167022e-01 9.36510801e-01 2.27180630e-01 -8.75083134e-02
2.36060679e-01 2.92114377e-01 -2.34808758e-01 -3.92670959e-01
-5.77305853e-01 2.20997911e-02 1.35972500e-01 -2.01024443e-01
-4.98468369e-01 -4.08150733e-01 -1.45577729e+00 -7.15469599e-01
-1.55469850e-01 9.23597991e-01 7.83152521e-01 8.64310741e-01
2.11063534e-01 3.17835689e-01 8.67691934e-01 -1.63823140e+00
-2.65928090e-01 -9.51560259e-01 -9.77192342e-01 6.10099912e-01
3.98530275e-01 -1.07539368e+00 -1.26198351e+00 -4.55140769e-02] | [6.375761032104492, 3.3202733993530273] |
ef3358a7-63b9-4b8f-8da0-c90f89809b93 | ma-net-a-multi-scale-attention-network-for | null | null | https://ieeexplore.ieee.org/document/9201310 | https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9201310 | MA-Net: A Multi-Scale Attention Network for Liver and Tumor Segmentation | Automatic assessing the location and extent of liver and liver tumor is critical for radiologists, diagnosis and the clinical process. In recent years, a large number of variants of U-Net based on Multi-scale feature fusion are proposed to improve the segmentation performance for medical image segmentation. Unlike the previous works which extract the context information of medical image via applying the multi-scale feature fusion, we propose a novel network named Multi-scale Attention Net (MA-Net) by introducing self-attention mechanism into our method to adaptively integrate local features with their global dependencies. The MA-Net can capture rich contextual dependencies based on the attention mechanism. We design two blocks: Position-wise Attention Block (PAB) and Multi-scale Fusion Attention Block (MFAB). The PAB is used to model the feature interdependencies in spatial dimensions, which capture the spatial dependencies between pixels in a global view. In addition, the MFAB is to capture the channel dependencies between any feature map by multi-scale semantic feature fusion. We evaluate our method on the dataset of MICCAI 2017 LiTS Challenge. The proposed method achieves better performance than other state-of-the-art methods. The Dice values of liver and tumors segmentation are 0.960 ± 0.03 and 0.749 ± 0.08 respectively. | ['Hongrui Wang', 'Yan Li', 'Guanglei Wang', 'Tongle Fan'] | 2020-09-21 | null | null | null | ieee-access-2020-9 | ['tumor-segmentation'] | ['computer-vision'] | [-1.91804707e-01 -2.38089979e-01 8.00380111e-03 -4.94165510e-01
-6.86836541e-01 -1.69483572e-01 3.63628209e-01 4.13383275e-01
-5.17303348e-01 4.22620505e-01 6.08144403e-01 -1.06156923e-01
-1.71039626e-01 -7.18851745e-01 -5.97231984e-01 -8.49494219e-01
-1.31235868e-01 -9.20406953e-02 5.55836678e-01 -2.61176121e-03
1.22049667e-01 3.90023619e-01 -8.71604919e-01 2.79185772e-01
1.11967719e+00 1.02235997e+00 4.11889821e-01 7.13106096e-01
-1.39018685e-01 5.79104185e-01 -2.86580145e-01 1.20312370e-01
1.95926487e-01 -6.45272672e-01 -8.83614242e-01 1.31801993e-01
4.67677005e-02 -2.39801809e-01 -2.11628348e-01 1.35017455e+00
7.03197896e-01 7.70688755e-03 5.78865051e-01 -8.51168156e-01
-7.78538764e-01 6.18120432e-01 -8.87028873e-01 8.05207014e-01
-2.51849350e-02 2.09289715e-01 5.76763451e-01 -6.09061062e-01
2.23065689e-01 1.16338003e+00 5.13597608e-01 1.19363084e-01
-6.35409117e-01 -5.14456093e-01 1.67847186e-01 4.52656239e-01
-1.36863029e+00 1.35854840e-01 6.05565190e-01 -4.45824623e-01
6.30465388e-01 2.80159384e-01 7.94245064e-01 3.28895152e-01
6.67570174e-01 9.01328981e-01 1.05658221e+00 -3.04413676e-01
-2.41577730e-01 -9.77725536e-02 2.47795716e-01 9.68715906e-01
1.14501052e-01 -2.08598631e-03 2.52952939e-03 8.90471712e-02
1.13423038e+00 4.35408980e-01 -3.77909243e-01 -1.10540502e-01
-1.66923833e+00 9.84030783e-01 1.21921849e+00 6.20664299e-01
-6.24353588e-01 8.64723623e-02 3.25757235e-01 -9.64852050e-02
2.17862338e-01 1.19607411e-01 -5.62970042e-01 4.97068763e-01
-5.73035359e-01 -3.67771924e-01 4.27642524e-01 7.12821186e-01
6.26155496e-01 -2.88182199e-01 -6.44217908e-01 5.61640024e-01
6.27043903e-01 1.92435816e-01 9.93434012e-01 -3.74039710e-01
-3.84891629e-02 6.53966844e-01 -2.85561502e-01 -8.65669489e-01
-7.58931756e-01 -7.66048789e-01 -1.15479803e+00 -1.46875322e-01
1.13386780e-01 -3.08323085e-01 -1.30665982e+00 1.47077906e+00
5.95135391e-01 6.31575167e-01 -3.40455547e-02 1.10919237e+00
1.30321395e+00 4.42440093e-01 3.95040661e-01 -1.64358363e-01
1.76356339e+00 -1.36424255e+00 -8.59662116e-01 2.24142417e-01
6.78682029e-01 -8.89869213e-01 7.27558076e-01 -2.14010939e-01
-1.00111079e+00 -5.67443907e-01 -9.40158308e-01 1.47576630e-01
-3.45192045e-01 2.14511417e-02 5.47949433e-01 3.57826799e-01
-1.13058352e+00 1.93718910e-01 -9.17141736e-01 -3.38020772e-01
5.57306111e-01 4.70572948e-01 -3.21499050e-01 -6.12080134e-02
-1.28870988e+00 8.36101234e-01 3.11632961e-01 3.16917509e-01
-8.32405210e-01 -6.30609035e-01 -9.37141538e-01 2.14753807e-01
2.35550582e-01 -8.49098444e-01 9.21716213e-01 -8.44429851e-01
-1.22866857e+00 4.24259096e-01 1.21635534e-02 -4.11882401e-01
3.62467319e-01 1.75339580e-01 -1.93973750e-01 3.71389061e-01
1.75789952e-01 8.22897851e-01 4.25109535e-01 -8.87239754e-01
-7.32554317e-01 -3.59710872e-01 -1.56838670e-01 5.24818301e-01
-1.29028082e-01 9.53670070e-02 -6.24717534e-01 -8.45644057e-01
2.14018479e-01 -6.21311605e-01 -5.83519995e-01 -1.46392763e-01
-2.81635851e-01 -1.61586165e-01 9.07670617e-01 -8.66587043e-01
1.14114881e+00 -2.06805778e+00 1.87898248e-01 1.42444611e-01
2.28511930e-01 2.52502978e-01 -1.06458671e-01 -3.46121341e-01
6.38777241e-02 2.28834376e-01 -2.87334174e-01 1.82650030e-01
-3.90271932e-01 1.83751181e-01 5.05438924e-01 5.66356897e-01
1.35652155e-01 1.36682522e+00 -9.56523478e-01 -9.01212990e-01
5.66905797e-01 6.13599420e-01 -5.35500705e-01 2.42817551e-01
2.98862785e-01 8.23574305e-01 -6.50083065e-01 6.55535638e-01
7.11028934e-01 -4.18846160e-01 -3.28371167e-01 -6.17702723e-01
-2.50953361e-02 -3.78997266e-01 -9.14885163e-01 1.91000998e+00
-3.57825607e-01 1.42830268e-01 2.08439201e-01 -8.34777355e-01
6.21009231e-01 4.28894341e-01 7.97433436e-01 -5.87218642e-01
4.40197110e-01 -1.60088241e-02 4.75706607e-01 -7.05242157e-01
-1.87943690e-02 -1.38930991e-01 -2.87801269e-02 1.44214645e-01
2.22286999e-01 1.38161615e-01 7.50753507e-02 1.33409292e-01
9.21786785e-01 -1.80126280e-01 6.44881666e-01 -7.25929260e-01
1.04229045e+00 -2.92518824e-01 7.24383652e-01 4.89535362e-01
-7.30652034e-01 5.72382510e-01 1.38359159e-01 -5.48370659e-01
-5.21654785e-01 -8.01714718e-01 -2.85798639e-01 8.71131897e-01
4.78280783e-01 -1.30579680e-01 -8.12492251e-01 -1.15827394e+00
-4.38499227e-02 2.88509969e-02 -9.82531071e-01 -8.99117216e-02
-5.25274336e-01 -1.03968072e+00 6.52736723e-02 8.76542807e-01
9.02626336e-01 -1.07940006e+00 -7.30346322e-01 2.63155490e-01
-1.83079928e-01 -8.27136278e-01 -9.45547163e-01 1.41058087e-01
-6.88320458e-01 -1.20697916e+00 -1.01743996e+00 -8.69251788e-01
8.65505040e-01 2.33549014e-01 9.34521616e-01 3.31035167e-01
-5.62471390e-01 2.25847065e-01 -3.99280131e-01 -2.37186551e-01
3.28750089e-02 1.26840830e-01 -5.28825641e-01 -3.00417058e-02
2.17699736e-01 -2.34983280e-01 -1.06469929e+00 4.24808025e-01
-9.65699732e-01 1.54113978e-01 1.08337915e+00 1.05907214e+00
7.24621475e-01 -9.68487561e-02 5.80121994e-01 -7.95350730e-01
4.20111269e-01 -6.46837354e-01 -3.37698072e-01 3.64694208e-01
-2.35164106e-01 -1.45860547e-02 3.84185106e-01 -2.31929630e-01
-8.56108308e-01 1.90817446e-01 -1.93064362e-01 -3.92844111e-01
-2.03318477e-01 6.08299315e-01 -1.15500838e-01 -3.69432896e-01
2.34694690e-01 2.68029273e-01 -1.94588333e-01 -2.53020823e-01
2.49526605e-01 5.91154277e-01 5.14276922e-01 -1.68355003e-01
2.21380264e-01 2.56354958e-01 6.64928509e-03 -4.77232933e-01
-7.26043105e-01 -6.55577123e-01 -8.53711426e-01 -9.19732749e-02
1.36551273e+00 -8.49216878e-01 -5.58976173e-01 4.31817293e-01
-8.50193679e-01 -1.54566124e-01 -1.40591204e-01 7.84115911e-01
-3.07268202e-01 3.53710413e-01 -8.15350533e-01 -1.92501768e-01
-6.49320662e-01 -1.77191496e+00 9.08658266e-01 6.95493400e-01
3.02824378e-01 -1.24545085e+00 -2.95816302e-01 1.52864650e-01
6.89410567e-01 3.62519592e-01 6.93400741e-01 -6.79516256e-01
-5.74603796e-01 9.32314023e-02 -7.82687962e-01 1.32232189e-01
4.44703013e-01 -3.34002256e-01 -5.01043916e-01 -3.13381553e-01
2.28867698e-02 1.59607261e-01 9.87005055e-01 1.04723084e+00
1.32816553e+00 -1.92752197e-01 -3.75382841e-01 9.54341412e-01
1.53533268e+00 3.50175321e-01 2.57012695e-01 2.25884020e-02
9.76551652e-01 8.88349861e-02 3.79409432e-01 2.19525144e-01
6.35191798e-01 3.06762338e-01 6.79607809e-01 -5.75941801e-01
-2.54844636e-01 2.96492666e-01 -1.53022304e-01 8.35110962e-01
1.61136672e-01 1.02835275e-01 -1.00489604e+00 7.31476247e-01
-1.88980436e+00 -4.72181469e-01 -8.75512958e-02 1.78926408e+00
6.47971928e-01 -2.07719162e-01 -2.15547100e-01 -3.80911499e-01
9.30437922e-01 -6.69741705e-02 -3.64085108e-01 -7.84254000e-02
6.37471899e-02 6.10936135e-02 6.83329046e-01 6.64475203e-01
-1.51619232e+00 5.39162457e-01 5.61527777e+00 8.13241422e-01
-1.19659722e+00 4.61259484e-01 9.08625126e-01 3.29806864e-01
-1.37410164e-01 -3.11877728e-01 -5.45418441e-01 6.08607888e-01
5.57341874e-01 -5.26382029e-02 -3.23517285e-02 4.47346538e-01
-8.14622082e-03 -3.51959467e-01 -6.22929990e-01 7.80099630e-01
3.13863724e-01 -1.13504803e+00 -2.56058834e-02 -3.19005586e-02
9.09322083e-01 3.10069472e-01 -9.38584805e-02 8.78191963e-02
3.13934624e-01 -1.10283220e+00 1.85793385e-01 6.10859573e-01
5.97517967e-01 -6.79077506e-01 1.31152117e+00 1.50386021e-01
-1.50982153e+00 -6.89486489e-02 -8.61322507e-02 4.21734899e-01
6.87682852e-02 5.04463315e-01 -5.69438279e-01 8.85431588e-01
6.81792617e-01 9.17679250e-01 -7.28909910e-01 1.35908508e+00
-6.14267178e-02 3.05509627e-01 -3.86506587e-01 2.00295448e-01
6.72150254e-01 -7.41605461e-02 2.60415256e-01 1.48751307e+00
3.85244131e-01 2.61162698e-01 4.91415679e-01 4.16519344e-01
2.13781688e-02 3.30766559e-01 5.40405959e-02 4.12464827e-01
-1.12853935e-02 1.69361901e+00 -1.15071452e+00 -6.74749374e-01
-6.10265195e-01 9.84326065e-01 -1.06276467e-01 1.04136892e-01
-9.12227213e-01 -2.83084929e-01 1.23703077e-01 -1.72967330e-01
4.14663166e-01 1.26123697e-01 -2.37752244e-01 -1.16304636e+00
-4.55216885e-01 -4.26365912e-01 7.71447122e-01 -5.38733006e-01
-1.31374717e+00 8.20372164e-01 -2.12436199e-01 -1.00537598e+00
2.68380612e-01 -3.25141728e-01 -8.93997133e-01 1.14322400e+00
-1.93976545e+00 -1.44204283e+00 -7.60293961e-01 7.61264086e-01
6.39227152e-01 8.58000815e-02 6.30240142e-01 2.94839382e-01
-5.22370517e-01 2.54307628e-01 -2.00422242e-01 3.86851192e-01
7.29913473e-01 -1.38119864e+00 -1.28068328e-01 8.50161076e-01
-2.08130404e-01 4.14332598e-01 2.46626288e-01 -5.81614971e-01
-9.77572680e-01 -1.26529145e+00 3.73637944e-01 -8.62217322e-02
3.05788040e-01 2.36413941e-01 -8.71945620e-01 6.74572289e-01
5.74957669e-01 8.89648318e-01 6.94305658e-01 -3.83702517e-01
1.16882168e-01 5.54331876e-02 -1.41640246e+00 7.99807906e-02
5.17107129e-01 -4.13807556e-02 -4.90792394e-01 2.25851223e-01
7.68069506e-01 -6.73250616e-01 -1.24012339e+00 7.53886759e-01
3.58487964e-01 -7.82877147e-01 1.04132521e+00 -2.53957927e-01
3.70226592e-01 -5.98351657e-01 -3.54908705e-02 -1.40738416e+00
-7.61772096e-01 -1.10581785e-01 1.01833992e-01 7.03829229e-01
1.30734250e-01 -5.14196634e-01 3.42195094e-01 1.56366631e-01
-4.37346339e-01 -1.13114429e+00 -7.46901453e-01 -1.65625170e-01
2.66447067e-01 2.45763674e-01 7.44465590e-01 9.93461311e-01
-1.45615578e-01 2.39619091e-01 8.34083557e-02 1.94521964e-01
5.14054894e-01 7.38526136e-02 1.52777478e-01 -1.01010585e+00
-1.52045771e-01 -7.19045103e-01 -5.47612965e-01 -8.37709367e-01
-1.40302300e-01 -1.02184701e+00 -1.11239307e-01 -1.93621063e+00
6.70963645e-01 -3.71780366e-01 -9.90275025e-01 4.69874740e-01
-6.76031530e-01 4.15708959e-01 2.08639055e-01 9.14443880e-02
-7.12113738e-01 4.69167233e-01 1.68939948e+00 -2.12895066e-01
-1.14825211e-01 -2.08963931e-01 -7.11125433e-01 7.92922139e-01
7.97076285e-01 -7.10445791e-02 -1.64638206e-01 -4.82234955e-01
-4.34601754e-01 2.84242749e-01 3.28150451e-01 -9.97633934e-01
4.85150307e-01 -3.43122669e-02 9.24152792e-01 -8.04112971e-01
-2.55645394e-01 -1.00430965e+00 -1.46482214e-01 8.21188986e-01
-2.17156693e-01 1.07177287e-01 1.26598954e-01 5.25405824e-01
-4.90077078e-01 6.31642863e-02 1.00476313e+00 -5.33956528e-01
-6.63552821e-01 7.62405515e-01 -1.39793381e-01 -1.98540121e-01
1.38619518e+00 2.30977163e-01 -3.73695306e-02 8.92433897e-03
-9.51991498e-01 6.20230377e-01 2.68162694e-02 1.99061558e-01
5.84644616e-01 -1.45750070e+00 -7.64358938e-01 4.56255585e-01
-6.31471053e-02 3.29268605e-01 6.90556467e-01 1.47474873e+00
-6.21670008e-01 4.84933645e-01 -3.51280689e-01 -7.66467869e-01
-1.19201827e+00 5.23255348e-01 6.07631326e-01 -5.91820419e-01
-5.66841722e-01 1.00757623e+00 7.12937176e-01 -1.38091251e-01
-1.30343765e-01 -6.66702986e-01 -4.51761454e-01 -1.44409508e-01
5.46509385e-01 6.46994710e-02 -8.32507387e-03 -8.73801231e-01
-4.33289737e-01 6.85328066e-01 -1.76598296e-01 2.34752089e-01
1.24011433e+00 -3.64165217e-01 -3.30870479e-01 -8.60418305e-02
1.20873260e+00 -2.08435982e-01 -1.28786755e+00 -4.68229532e-01
-3.09489369e-01 -4.20847356e-01 5.17149508e-01 -1.14586937e+00
-1.54670548e+00 9.56072450e-01 8.57573211e-01 -3.80486920e-02
1.37501752e+00 -6.71546161e-03 8.70033562e-01 -4.46730822e-01
-1.02236204e-01 -5.83366871e-01 -1.06566502e-02 4.37060267e-01
7.69665599e-01 -1.53839636e+00 -6.35056570e-02 -3.82124245e-01
-6.49438798e-01 1.00295687e+00 7.51151025e-01 -2.86775470e-01
1.03349030e+00 4.23925787e-01 1.08820587e-01 -9.62781385e-02
-3.78323495e-01 -4.00576949e-01 5.33070445e-01 3.52042675e-01
4.90287751e-01 1.96438745e-01 -5.09597719e-01 8.81900132e-01
3.26952815e-01 5.96172251e-02 2.34884486e-01 7.51902103e-01
-6.09343827e-01 -7.73197055e-01 -4.33945298e-01 5.69620609e-01
-7.01801121e-01 -2.28348196e-01 2.69603789e-01 5.55863082e-01
4.23324287e-01 5.99060178e-01 1.66867137e-01 -1.68326154e-01
9.14257485e-03 -7.97116235e-02 4.00662750e-01 -4.30418670e-01
-9.87061024e-01 6.04469180e-01 -7.02375948e-01 -5.03104806e-01
-5.77453494e-01 -4.84030783e-01 -1.48778379e+00 1.58063009e-01
-4.33319390e-01 2.32410163e-01 5.87182045e-01 8.94273639e-01
2.50059634e-01 1.09626579e+00 6.20682716e-01 -6.24696314e-01
-2.19510034e-01 -1.18411291e+00 -4.00703549e-01 5.10000348e-01
4.62990701e-01 -5.22104383e-01 -9.94126201e-02 1.05627086e-02] | [14.59321117401123, -2.6279382705688477] |
41f0719a-ffe7-4d66-9397-6c479d84b73e | handwriting-recognition-for-scottish-gaelic | null | null | https://aclanthology.org/2022.cltw-1.9 | https://aclanthology.org/2022.cltw-1.9.pdf | Handwriting recognition for Scottish Gaelic | Like most other minority languages, Scottish Gaelic has limited tools and resources available for Natural Language Processing research and applications. These limitations restrict the potential of the language to participate in modern speech technology, while also restricting research in fields such as corpus linguistics and the Digital Humanities. At the same time, Gaelic has a long written history, is well-described linguistically, and is unusually well-supported in terms of potential NLP training data. For instance, archives such as the School of Scottish Studies hold thousands of digitised recordings of vernacular speech, many of which have been transcribed as paper-based, handwritten manuscripts. In this paper, we describe a project to digitise and recognise a corpus of handwritten narrative transcriptions, with the intention of re-purposing it to develop a Gaelic speech recognition system. | ['Mark Sinclair', 'Beatrice Alex', 'William Lamb'] | null | null | null | null | cltw-lrec-2022-6 | ['handwriting-recognition'] | ['computer-vision'] | [ 2.61855423e-01 2.37547472e-01 -2.15682343e-01 -3.05227607e-01
-9.20208275e-01 -7.99998164e-01 1.11770451e+00 2.04164878e-01
-4.57011729e-01 6.19016171e-01 8.82325947e-01 -6.86405957e-01
-5.95327094e-02 -5.21096826e-01 -1.84079885e-01 -4.41778481e-01
4.06361341e-01 5.00206649e-01 5.73452301e-02 -2.66580135e-01
8.85818481e-01 6.70230627e-01 -1.23367512e+00 2.96117097e-01
4.05082434e-01 2.02813089e-01 3.44498068e-01 6.36789978e-01
-3.08263391e-01 1.11151493e+00 -7.77894795e-01 -4.49940383e-01
-3.29090089e-01 -6.10535681e-01 -9.99108851e-01 3.40078294e-01
1.97688788e-01 -1.02565631e-01 -6.66225851e-01 5.33522785e-01
3.03055286e-01 1.28900796e-01 4.94247258e-01 -3.99161369e-01
-7.75330842e-01 6.29221976e-01 -3.30212270e-03 1.12560205e-01
6.72140479e-01 -1.52198985e-01 8.24986160e-01 -7.79274046e-01
1.01278126e+00 1.24394822e+00 3.13617706e-01 4.27475274e-01
-8.14057469e-01 -3.52960557e-01 -3.14094692e-01 -1.34347484e-01
-1.23594975e+00 -1.18005812e+00 8.05033743e-01 -4.45915043e-01
1.03832352e+00 2.53341049e-01 6.02422893e-01 1.08959293e+00
-2.87339296e-02 6.06180549e-01 9.42746937e-01 -8.71621072e-01
2.82949150e-01 1.29692599e-01 -2.44308323e-01 6.09283447e-01
-1.61326498e-01 -4.02728379e-01 -7.46237576e-01 -2.76538342e-01
7.06610858e-01 -5.20533919e-01 -2.78647453e-01 4.51287836e-01
-1.23739243e+00 7.99099803e-01 -3.67457509e-01 9.30804014e-01
3.76315154e-02 -2.01043472e-01 8.53718042e-01 3.42019022e-01
3.46913517e-01 3.81029844e-01 -3.21517624e-02 -8.40719819e-01
-9.53591168e-01 1.34893432e-01 1.08031166e+00 8.48627627e-01
-1.69654023e-02 1.79609686e-01 6.76314533e-01 1.44432032e+00
4.46455538e-01 5.42987406e-01 7.13376880e-01 -8.09263051e-01
7.48608708e-01 4.52887475e-01 -3.62607986e-01 -1.15123773e+00
1.06259026e-01 3.95569503e-01 -4.08902615e-01 -1.24049217e-01
5.50371110e-01 2.75314122e-01 -6.63785398e-01 9.81831670e-01
-1.13253757e-01 -7.42723584e-01 2.91764587e-01 6.93479598e-01
6.57815456e-01 9.92858529e-01 -3.24324295e-02 -5.26580036e-01
1.29640317e+00 -3.86834383e-01 -6.23180628e-01 -5.58839560e-01
3.51227731e-01 -1.18119287e+00 1.32987869e+00 5.99204719e-01
-1.35498393e+00 4.19443846e-02 -1.02400458e+00 -2.66054451e-01
-1.81798011e-01 -2.50421837e-02 5.42789400e-01 9.10608470e-01
-6.39992177e-01 3.63945127e-01 -9.35878158e-01 -7.66751766e-01
2.94506520e-01 -2.22541586e-01 -4.73520219e-01 -9.75229144e-02
-7.78319597e-01 9.95924354e-01 3.30241084e-01 1.19140364e-01
-3.23508769e-01 -1.93730183e-02 -8.56229067e-01 -3.11190039e-01
2.28232116e-01 3.31947029e-01 1.36576021e+00 -8.98057520e-01
-1.41551721e+00 1.40222454e+00 -1.81415781e-01 -7.68847242e-02
4.40653652e-01 2.43988633e-01 -9.46500957e-01 3.77487302e-01
1.05118059e-01 4.80109416e-02 4.19763178e-01 -7.04022527e-01
-5.97955287e-01 -4.65614945e-01 -4.10830140e-01 -1.00601964e-01
-2.82911658e-01 7.54731774e-01 -3.52026016e-01 -9.76385534e-01
3.69498163e-01 -5.25113940e-01 2.29754150e-01 -2.74647802e-01
-3.34069468e-02 -1.18923053e-01 6.49735391e-01 -1.20902538e+00
1.37388051e+00 -2.34199429e+00 -2.47169808e-01 2.37838969e-01
-1.17922418e-01 1.09974444e-01 2.79828787e-01 1.05718946e+00
4.19991255e-01 3.38248938e-01 -2.67299294e-01 1.01577207e-01
6.58019409e-02 6.61340833e-01 -4.24089104e-01 6.39110804e-01
1.34716466e-01 4.94298309e-01 -9.53311861e-01 -5.11395931e-01
-2.91929021e-02 4.09794629e-01 -4.16635871e-02 -5.31271040e-01
1.75692469e-01 -7.62660280e-02 -4.88886416e-01 8.32031310e-01
5.04703671e-02 2.16411218e-01 5.72555602e-01 6.84595525e-01
-7.25015223e-01 1.00384271e+00 -8.33023608e-01 1.35099864e+00
-5.17651320e-01 1.40672922e+00 1.55961543e-01 -7.56866395e-01
1.10662723e+00 6.08151257e-01 -1.26365677e-01 -6.58818364e-01
1.38700604e-01 8.06524515e-01 6.09415658e-02 -6.06579602e-01
1.02428043e+00 -4.55591589e-01 -2.64429122e-01 5.64514577e-01
-1.48607954e-01 -5.19672692e-01 1.30456209e-01 1.33350655e-01
1.02387691e+00 -1.54483646e-01 4.99079466e-01 -2.87090451e-01
5.39636612e-01 3.37290764e-01 3.16272497e-01 2.99341172e-01
-1.07309788e-01 6.13566279e-01 3.86264026e-01 -2.12677479e-01
-1.53453362e+00 -7.52421260e-01 -5.05897105e-01 9.80816185e-01
-4.79635835e-01 -4.08260912e-01 -7.39001751e-01 1.28785133e-01
-4.47222739e-01 7.79856741e-01 -3.83129274e-03 2.03415841e-01
-8.52753580e-01 -1.46106496e-01 1.12612545e+00 4.26234454e-01
3.02552015e-01 -1.60307157e+00 -5.06831348e-01 4.65657592e-01
7.85033926e-02 -8.87616575e-01 -2.29968607e-01 -4.06020358e-02
-5.00802338e-01 -8.70928526e-01 -6.67809606e-01 -1.28956938e+00
1.81323275e-01 -1.60605848e-01 5.85548162e-01 -1.03298761e-01
-1.54175892e-01 3.80712003e-01 -3.86783332e-01 -4.66164589e-01
-1.17464030e+00 1.83054283e-01 -1.09285213e-01 -4.77700889e-01
4.48463619e-01 -3.30239862e-01 4.83378708e-01 -3.70637327e-02
-9.45837796e-01 -4.22110677e-01 2.44330004e-01 7.36596823e-01
1.28262192e-01 -1.32366687e-01 1.01195407e+00 -8.91713560e-01
6.75466180e-01 -3.13701332e-01 -2.29538932e-01 3.08712214e-01
-1.64417639e-01 -4.39810127e-01 4.75141585e-01 -2.86343902e-01
-1.13816249e+00 -3.07855844e-01 -2.55049825e-01 4.04627562e-01
-2.85441786e-01 7.85307944e-01 -1.38371944e-01 1.61905780e-01
6.82093143e-01 3.84635001e-01 1.29501194e-01 -3.39233279e-01
-6.11229278e-02 1.58002019e+00 9.40800011e-01 -5.56158960e-01
4.59016651e-01 7.86655545e-02 -5.17907977e-01 -1.93322444e+00
-1.47713542e-01 -4.74093735e-01 -4.73548442e-01 -3.92694682e-01
5.86947560e-01 -7.00689495e-01 -5.36028385e-01 4.74441975e-01
-1.07308221e+00 -3.35947692e-01 -1.72418371e-01 5.71880758e-01
-5.41054189e-01 2.81210870e-01 -8.36741567e-01 -1.17756999e+00
1.52232379e-01 -5.98198473e-01 7.08181381e-01 5.55883273e-02
-7.25256979e-01 -1.09058774e+00 2.00277388e-01 6.98214471e-01
3.16083501e-03 1.44773424e-01 1.12507153e+00 -6.84384167e-01
-1.97610855e-01 -4.48761612e-01 9.55699608e-02 3.26055586e-01
2.24804014e-01 2.48992607e-01 -9.43471968e-01 -6.41864119e-03
1.91572294e-01 -4.22819674e-01 3.36714238e-01 -1.80988312e-01
2.55880803e-01 -4.37402934e-01 1.23356916e-01 -1.85894623e-01
1.05741704e+00 7.87922978e-01 9.65703309e-01 6.33133888e-01
2.51331121e-01 6.30736232e-01 1.59774423e-01 3.94887298e-01
2.37285614e-01 1.46672398e-01 -6.38927400e-01 6.38767362e-01
7.46887457e-03 -3.68368387e-01 6.07580543e-01 1.60565174e+00
-1.46789283e-01 -3.72083127e-01 -1.57517147e+00 1.11125767e+00
-1.54689467e+00 -1.29681277e+00 -2.49417379e-01 1.84284818e+00
9.11776662e-01 3.71512413e-01 -3.53038125e-02 3.88255030e-01
6.93820119e-01 4.07803893e-01 2.90371850e-02 -9.30934846e-01
-3.23089838e-01 3.25800538e-01 3.32248390e-01 6.03392839e-01
-6.75540924e-01 1.01684773e+00 7.05031872e+00 7.61537313e-01
-8.79101038e-01 -2.90400326e-01 3.10635149e-01 -4.39777412e-02
-4.22509104e-01 3.51374183e-04 -3.79915178e-01 2.96304405e-01
1.13748550e+00 -3.77537429e-01 5.68328857e-01 5.44720709e-01
3.41813654e-01 -2.00560033e-01 -6.83272302e-01 6.72126830e-01
4.76952463e-01 -1.59327745e+00 -2.18247473e-01 1.76385537e-01
3.91669393e-01 3.77486385e-02 -3.48229080e-01 -5.03980741e-02
4.82399166e-02 -1.20454133e+00 1.23745346e+00 3.18564564e-01
8.17878962e-01 -8.18594038e-01 4.13548827e-01 4.72947776e-01
-7.22013235e-01 -2.71199755e-02 -3.82470965e-01 -4.39917982e-01
2.78534293e-01 6.80567464e-03 -1.03021777e+00 1.09004691e-01
2.67506927e-01 2.85536051e-01 -4.17227983e-01 6.55755103e-01
-2.01913446e-01 1.30885434e+00 -3.10845524e-01 -4.92389888e-01
6.14662528e-01 -3.24601501e-01 5.33623457e-01 1.35131085e+00
8.03401023e-02 5.53987324e-01 3.01747601e-02 4.57706660e-01
8.93540457e-02 3.98583949e-01 -6.61431134e-01 -1.04780889e+00
8.18441451e-01 6.81808352e-01 -9.98760700e-01 -7.86092281e-02
-5.78623176e-01 6.36404276e-01 -9.51364115e-02 8.24780166e-02
-2.13708445e-01 -7.40792811e-01 1.30996883e-01 5.18115103e-01
8.89015049e-02 -6.78782642e-01 -3.64602327e-01 -8.37309480e-01
2.14825824e-01 -1.11865079e+00 2.82934774e-02 -6.77719831e-01
-1.23556399e+00 6.43284678e-01 -2.80853391e-01 -5.58374882e-01
-5.00513852e-01 -6.06844127e-01 -5.65885603e-01 9.14790809e-01
-5.51120520e-01 -1.21593308e+00 3.20968509e-01 1.72445625e-01
6.81481123e-01 -8.30704987e-01 8.91353905e-01 2.59566545e-01
-3.46524060e-01 2.90090770e-01 3.33988577e-01 4.94048625e-01
4.54045951e-01 -7.78695226e-01 5.00847161e-01 6.25269592e-01
3.44780803e-01 6.20598257e-01 6.29922509e-01 -7.42989540e-01
-1.33999574e+00 -4.64898020e-01 1.63987911e+00 -2.76464820e-01
1.28302777e+00 -4.45620775e-01 -9.18393910e-01 6.54155314e-01
1.83667630e-01 -7.30138361e-01 6.69995487e-01 1.01094125e-02
-1.63358390e-01 3.78944516e-01 -9.49975610e-01 6.05074227e-01
7.66393185e-01 -1.29968667e+00 -1.12152028e+00 3.43828589e-01
2.17954949e-01 -6.26306310e-02 -8.14903617e-01 -5.71221173e-01
7.28938818e-01 -4.12363440e-01 4.53022033e-01 -3.74529511e-01
6.15069330e-01 -2.51770834e-03 -4.15134519e-01 -7.78516293e-01
-5.51051497e-02 -9.90585387e-01 6.71743095e-01 1.89401424e+00
4.29255307e-01 -6.82901859e-01 8.33520412e-01 6.44697785e-01
-5.18450260e-01 -2.77719915e-01 -9.59962428e-01 -7.49322653e-01
3.27463448e-01 -6.06174350e-01 2.41269335e-01 1.27489555e+00
6.01607561e-01 3.33684713e-01 7.52929151e-02 -3.47690165e-01
2.41396308e-01 -3.29118520e-01 4.11564082e-01 -1.02866209e+00
-1.14975549e-01 -5.84703922e-01 -7.60218203e-01 -4.36109781e-01
2.55787015e-01 -9.43122268e-01 3.14522356e-01 -1.39543152e+00
-3.35061103e-01 -3.46310616e-01 7.16355801e-01 4.87597942e-01
6.41131520e-01 3.66167687e-02 1.88082770e-01 3.22379768e-01
2.87886292e-01 1.37593657e-01 8.98039103e-01 -8.74646157e-02
-4.53272492e-01 -2.49199659e-01 -6.79015398e-01 6.60242736e-01
8.71770382e-01 -3.66664886e-01 -3.04277152e-01 -1.56633273e-01
2.38493577e-01 2.29594171e-01 2.00884566e-01 -6.46637857e-01
3.24006915e-01 -5.27041316e-01 2.38488033e-01 -3.89120132e-01
6.12762980e-02 -3.41912746e-01 2.48647705e-01 -7.42659764e-03
-4.10502672e-01 -4.64665107e-02 1.59229368e-01 9.80622619e-02
-5.57221532e-01 -6.39475167e-01 6.93440259e-01 -2.65003592e-01
-6.92321062e-01 -4.04720575e-01 -1.26016116e+00 3.45385224e-01
8.64287257e-01 -5.79239964e-01 -4.36074406e-01 -2.68610686e-01
-3.76692623e-01 -1.91735715e-01 8.17404211e-01 2.75348872e-01
5.77834308e-01 -1.15524137e+00 -6.47092700e-01 3.48173440e-01
1.66097119e-01 -2.12134942e-01 -8.57439172e-03 4.06595469e-01
-1.21032619e+00 4.20799464e-01 -2.29047015e-01 1.00340120e-01
-8.91989350e-01 1.56125620e-01 -1.80164069e-01 3.20637465e-01
-1.01987541e+00 5.08918703e-01 -3.57046038e-01 -3.49569917e-01
1.45740896e-01 -1.16489865e-02 4.83242124e-02 2.32315376e-01
6.52537465e-01 3.95842105e-01 3.21564190e-02 -1.04823720e+00
-4.70467150e-01 -1.31826818e-01 7.31738284e-02 -8.43527257e-01
1.59054601e+00 -2.24473011e-02 -2.71797866e-01 1.06729853e+00
1.11221099e+00 3.89443308e-01 -5.58700502e-01 1.78992540e-01
5.34059405e-01 -4.57478881e-01 1.05648696e-01 -7.49921620e-01
-3.73770624e-01 7.30956137e-01 -3.36204410e-01 2.94884354e-01
7.87965953e-01 5.22927083e-02 6.97854936e-01 7.33669639e-01
2.17600986e-01 -1.67247617e+00 -4.27259415e-01 7.94968188e-01
1.13911808e+00 -4.96345431e-01 5.38343228e-02 -3.71411890e-02
-8.53905976e-01 1.34864092e+00 -2.80540168e-01 -1.25315040e-01
4.29989606e-01 4.33982849e-01 1.32765502e-01 -3.76924455e-01
-5.92682958e-01 2.08235353e-01 3.02269543e-03 5.97013712e-01
7.28181124e-01 1.67662296e-02 -6.54426515e-01 4.30259347e-01
-7.75007904e-01 -6.12182394e-02 1.00736487e+00 1.21266150e+00
-6.96191549e-01 -9.95740235e-01 -6.35664105e-01 2.64958888e-01
-5.66239297e-01 -3.66562158e-02 -8.53601933e-01 1.07076907e+00
-3.64434421e-01 8.39002788e-01 4.10860837e-01 2.00886447e-02
2.84223020e-01 3.76414776e-01 4.92432952e-01 -5.41580260e-01
-6.53204262e-01 4.42030668e-01 8.36820960e-01 1.40145048e-01
-3.81772697e-01 -1.21959531e+00 -1.27731252e+00 -7.52880752e-01
-5.59312180e-02 2.91183293e-01 7.66102076e-01 1.02695215e+00
-3.76246959e-01 1.32682752e-02 2.10490584e-01 -5.85395932e-01
-2.28711396e-01 -9.34934437e-01 -1.11061394e+00 -2.76930898e-01
1.13748955e-02 -5.07246964e-02 -8.55840370e-02 4.17692065e-01] | [10.415834426879883, 10.138031959533691] |
c4a7d0f6-5d4c-466a-9a31-51e94d39dd77 | robust-outlier-rejection-for-3d-registration | 2304.01514 | null | https://arxiv.org/abs/2304.01514v1 | https://arxiv.org/pdf/2304.01514v1.pdf | Robust Outlier Rejection for 3D Registration with Variational Bayes | Learning-based outlier (mismatched correspondence) rejection for robust 3D registration generally formulates the outlier removal as an inlier/outlier classification problem. The core for this to be successful is to learn the discriminative inlier/outlier feature representations. In this paper, we develop a novel variational non-local network-based outlier rejection framework for robust alignment. By reformulating the non-local feature learning with variational Bayesian inference, the Bayesian-driven long-range dependencies can be modeled to aggregate discriminative geometric context information for inlier/outlier distinction. Specifically, to achieve such Bayesian-driven contextual dependencies, each query/key/value component in our non-local network predicts a prior feature distribution and a posterior one. Embedded with the inlier/outlier label, the posterior feature distribution is label-dependent and discriminative. Thus, pushing the prior to be close to the discriminative posterior in the training step enables the features sampled from this prior at test time to model high-quality long-range dependencies. Notably, to achieve effective posterior feature guidance, a specific probabilistic graphical model is designed over our non-local model, which lets us derive a variational low bound as our optimization objective for model training. Finally, we propose a voting-based inlier searching strategy to cluster the high-quality hypothetical inliers for transformation estimation. Extensive experiments on 3DMatch, 3DLoMatch, and KITTI datasets verify the effectiveness of our method. | ['Mathieu Salzmann', 'Jian Yang', 'Jin Xie', 'Zhen Wei', 'Zheng Dang', 'Haobo Jiang'] | 2023-04-04 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Jiang_Robust_Outlier_Rejection_for_3D_Registration_With_Variational_Bayes_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Jiang_Robust_Outlier_Rejection_for_3D_Registration_With_Variational_Bayes_CVPR_2023_paper.pdf | cvpr-2023-1 | ['bayesian-inference'] | ['methodology'] | [-1.94139645e-01 4.06306703e-03 -2.63782233e-01 -4.24412727e-01
-1.35034871e+00 -2.51884729e-01 4.82313693e-01 3.19458134e-02
2.39174180e-02 3.56780261e-01 2.73325950e-01 2.17217699e-01
-5.97115159e-01 -4.58952636e-01 -9.31523263e-01 -9.23678041e-01
1.58929393e-01 8.33782434e-01 -4.28316519e-02 1.86040416e-01
2.37048030e-01 9.04922783e-01 -1.21442580e+00 -2.03774229e-01
9.95862961e-01 9.04472411e-01 -1.14048056e-01 1.83972865e-01
4.18646723e-01 1.02777325e-01 -3.68826330e-01 9.26908944e-03
5.15696466e-01 -3.10282648e-01 -8.37852359e-02 2.89200455e-01
8.01101208e-01 -4.03929889e-01 -3.44843119e-01 1.07303524e+00
6.45607233e-01 2.76908934e-01 1.10049748e+00 -1.24516284e+00
-4.70683463e-02 4.69686277e-02 -6.85794294e-01 -1.79395825e-01
4.57023352e-01 1.20608918e-01 1.16054630e+00 -1.45058036e+00
5.77658653e-01 1.09313202e+00 6.10429108e-01 -3.26897055e-02
-1.17331171e+00 -5.85341454e-01 2.10207358e-01 -5.06701171e-02
-1.60599554e+00 -4.13255841e-01 1.16415215e+00 -6.56215191e-01
3.96779180e-01 3.12697738e-01 9.75009441e-01 1.08287001e+00
2.49591693e-01 8.08262944e-01 5.85760534e-01 -1.85772911e-01
2.28942141e-01 -4.38130885e-01 6.06852286e-02 7.18903720e-01
5.22891045e-01 2.67927349e-01 -8.71662021e-01 -5.71064532e-01
9.92201507e-01 2.96961963e-01 -2.71863908e-01 -8.73573959e-01
-1.08884251e+00 7.66129076e-01 4.21210319e-01 6.37594089e-02
-3.18466753e-01 2.96535075e-01 4.07508723e-02 1.07243918e-02
5.34068704e-01 1.61130354e-01 -2.37461805e-01 1.07752301e-01
-1.15975821e+00 2.39966661e-01 5.99480391e-01 1.03508043e+00
1.01313865e+00 1.30816447e-02 -4.01480675e-01 7.77264178e-01
9.94675815e-01 5.10036588e-01 -9.51942578e-02 -9.85790610e-01
3.91034663e-01 5.09776771e-01 3.13997269e-03 -1.20247376e+00
-1.91221371e-01 -8.53589773e-01 -9.59256291e-01 5.50834686e-02
5.28884768e-01 4.12499219e-01 -9.66709316e-01 1.52456450e+00
6.36496186e-01 7.42800474e-01 -4.42076743e-01 9.09102023e-01
3.36258531e-01 3.36747020e-01 -5.48653126e-01 -4.39607829e-01
9.35285270e-01 -3.56803864e-01 -3.88713330e-01 7.13146627e-02
5.83832026e-01 -8.94975543e-01 8.46050560e-01 3.31788003e-01
-7.38417327e-01 -2.36801520e-01 -1.01515961e+00 -1.86414085e-02
4.54301775e-01 5.33916131e-02 -9.83218476e-03 1.20307505e-01
-5.35314918e-01 5.57789385e-01 -1.03797817e+00 -2.01769724e-01
3.22333008e-01 3.29620630e-01 -2.80668288e-01 -2.80516714e-01
-6.10599816e-01 5.80389798e-01 1.48038402e-01 4.21185434e-01
-9.41117585e-01 -6.46920502e-01 -9.03192341e-01 -2.53282547e-01
5.11709571e-01 -9.91762996e-01 6.42073095e-01 -2.99394101e-01
-1.29063284e+00 7.22903848e-01 -2.54225910e-01 -5.46300150e-02
7.08829582e-01 -2.87064165e-01 1.57215744e-01 2.81809010e-02
2.56703705e-01 2.00047478e-01 1.19604802e+00 -1.59055793e+00
-2.06504628e-01 -5.24491906e-01 -4.50668216e-01 2.14880288e-01
7.03491196e-02 -2.55512148e-01 -7.69956052e-01 -9.77558970e-01
1.05110288e+00 -1.01933408e+00 -2.65272945e-01 1.04753666e-01
-6.60340846e-01 -2.84724325e-01 5.98961294e-01 -5.63663006e-01
1.00754797e+00 -2.45779705e+00 1.36476502e-01 1.23550081e+00
4.56446052e-01 -6.00093782e-01 -3.55999544e-02 1.59236178e-01
4.41329814e-02 -2.50505209e-01 -2.99364448e-01 -6.15630150e-01
1.48468748e-01 3.61292720e-01 -2.64165252e-01 1.15856266e+00
-4.52685915e-02 7.03090906e-01 -9.38550234e-01 -5.50383151e-01
4.34324265e-01 3.83572459e-01 -9.70391095e-01 1.68421865e-01
-1.46885328e-02 8.55521977e-01 -8.38422239e-01 1.01684880e+00
6.29537225e-01 1.40333772e-01 -2.85849899e-01 -4.86076832e-01
1.34318471e-01 -1.23396948e-01 -1.49724209e+00 1.89450264e+00
-2.80873664e-02 1.63789958e-01 -2.57916525e-02 -7.72853255e-01
1.09988976e+00 4.19304669e-02 7.89506495e-01 -1.61103368e-01
9.73958671e-02 6.52482569e-01 -2.39799768e-01 -3.09952378e-01
3.22725892e-01 4.13715616e-02 -1.74138978e-01 -8.37388039e-02
2.02150829e-02 -4.32021737e-01 -2.10717648e-01 1.26995653e-01
8.10145438e-01 4.78049070e-01 1.48062944e-01 -2.82603055e-01
3.97700757e-01 -3.97126466e-01 1.15832567e+00 8.12708557e-01
-1.88225985e-01 1.04586351e+00 4.12587821e-01 -8.28794837e-02
-9.54807043e-01 -1.26892543e+00 -3.79386902e-01 3.33869964e-01
3.22234362e-01 -4.52181578e-01 -4.09286588e-01 -6.85364485e-01
2.21163020e-01 5.88966608e-01 -2.73371786e-01 -2.27940843e-01
-7.30389297e-01 -2.89863437e-01 2.85822481e-01 2.66915351e-01
3.48726735e-02 -3.36106271e-01 7.29416981e-02 1.83557793e-01
-3.41057740e-02 -8.05834353e-01 -6.68026328e-01 6.26397282e-02
-1.02588892e+00 -1.17906499e+00 -4.88138914e-01 -4.92451638e-01
1.04047465e+00 -1.38625488e-01 8.10023785e-01 1.75739184e-01
8.79852567e-03 5.63905835e-01 -1.02925330e-01 1.23392910e-01
-4.03299741e-02 -2.68545061e-01 4.50375110e-01 2.55981982e-01
1.08667627e-01 -9.08163965e-01 -6.33750677e-01 4.68519449e-01
-6.96365774e-01 -3.39904636e-01 5.34466803e-01 1.16071653e+00
1.11307311e+00 -2.81605482e-01 7.91175589e-02 -3.29460114e-01
1.19697057e-01 -3.63156438e-01 -6.79391980e-01 3.12495884e-03
-2.13189200e-01 1.16723944e-02 2.51527548e-01 -2.15238109e-01
-6.68893933e-01 2.67535031e-01 -1.59085706e-01 -9.52823937e-01
-9.17499885e-02 7.16761827e-01 -5.64521015e-01 -1.43438175e-01
4.66186315e-01 2.69760434e-02 -1.43397912e-01 -4.73610491e-01
2.21862212e-01 2.60027617e-01 6.07262015e-01 -1.14552474e+00
1.33159959e+00 5.17733335e-01 3.26757491e-01 -8.42837214e-01
-8.50740433e-01 -6.56649709e-01 -8.60827863e-01 -4.25311863e-01
6.57867730e-01 -9.90104079e-01 -4.66985971e-01 4.38418210e-01
-9.33695853e-01 7.45057538e-02 -6.09125614e-01 8.47824574e-01
-7.49150038e-01 5.96755922e-01 -3.52964342e-01 -7.67778218e-01
2.26468444e-01 -1.38998747e+00 1.35639143e+00 -1.24342769e-01
-2.79501289e-01 -7.88816810e-01 2.92365521e-01 2.21092150e-01
-2.55701303e-01 5.47585487e-01 6.74316645e-01 -7.66942918e-01
-1.11210835e+00 -4.67733711e-01 -3.58823612e-02 3.29314381e-01
-1.64832041e-01 3.19846690e-01 -6.80143714e-01 -5.23153484e-01
2.09773272e-01 1.69262681e-02 8.67612958e-01 6.25048816e-01
1.09705496e+00 -6.78212345e-02 -4.27036047e-01 9.97232974e-01
1.23305702e+00 -3.22037339e-01 4.42847788e-01 8.80365968e-02
1.00877988e+00 2.87843704e-01 9.47272122e-01 6.73833549e-01
2.37135738e-01 7.56060243e-01 6.01450741e-01 2.86599219e-01
2.09535718e-01 -5.08652449e-01 4.38536078e-01 1.03445184e+00
-3.64979282e-02 1.64440293e-02 -8.93206775e-01 4.15726066e-01
-2.12228799e+00 -6.44896328e-01 -2.19816059e-01 2.53828263e+00
5.96737564e-01 1.21353798e-01 -1.53022811e-01 -1.67417094e-01
6.22564971e-01 1.96377844e-01 -5.40682316e-01 4.65495974e-01
-1.35314628e-01 -1.36383221e-01 2.24588066e-01 8.02494407e-01
-9.17519093e-01 7.04198897e-01 4.69700575e+00 1.16210806e+00
-5.63303471e-01 -9.83459726e-02 2.58381277e-01 -2.66368706e-02
-7.01729059e-01 4.20677423e-01 -9.47704256e-01 3.92139196e-01
4.18415844e-01 1.03215225e-01 -6.42422959e-02 6.71338558e-01
4.86767590e-01 -1.52591527e-01 -1.31728280e+00 1.05880368e+00
1.74465448e-01 -1.04604888e+00 1.76102921e-01 5.02174795e-01
6.72910869e-01 3.68237384e-02 -3.26136649e-02 2.80879512e-02
1.70659319e-01 -6.25249207e-01 8.03506136e-01 1.00931823e+00
6.22760594e-01 -8.63532603e-01 5.15003204e-01 4.00852114e-01
-1.09439337e+00 5.71849160e-02 -3.88168216e-01 3.10729533e-01
3.48793387e-01 1.20133245e+00 -5.68501472e-01 8.94286036e-01
5.88608980e-01 1.17758942e+00 -2.65816987e-01 1.42607427e+00
-3.78902525e-01 4.44599718e-01 -8.22843552e-01 5.05495965e-01
-2.64324658e-02 -7.96477735e-01 1.31604505e+00 6.42789423e-01
6.95416927e-01 2.72813835e-03 6.21878445e-01 8.70264530e-01
-8.16464517e-03 4.52469252e-02 -5.98422289e-01 5.32044768e-01
6.17568851e-01 1.01686251e+00 -5.64932823e-01 -3.76257747e-02
-1.52751416e-01 7.67331898e-01 2.02644870e-01 5.07145345e-01
-7.83131063e-01 4.36821058e-02 7.17718005e-01 1.31875962e-01
-4.28194627e-02 -5.21035790e-01 -3.59659821e-01 -1.49858832e+00
3.29081088e-01 -7.48780191e-01 3.32351536e-01 -5.62414706e-01
-1.51720893e+00 1.76520310e-02 2.03128204e-01 -1.76927805e+00
-3.04827154e-01 -4.16215777e-01 -6.68940306e-01 6.33442760e-01
-1.27205408e+00 -1.36970866e+00 -3.49359959e-01 7.44786739e-01
1.72199622e-01 3.63626108e-02 3.53747666e-01 3.49283010e-01
-7.00564086e-01 5.78998625e-01 2.21693188e-01 1.64460629e-01
1.07468617e+00 -1.02495587e+00 -1.87872320e-01 1.13765657e+00
3.08431864e-01 7.03098595e-01 6.93223894e-01 -9.99598026e-01
-1.23903728e+00 -1.04093111e+00 6.09878242e-01 -5.95645368e-01
6.12398267e-01 -2.73866445e-01 -7.99952686e-01 8.44835401e-01
-4.56488252e-01 2.12373123e-01 4.91836011e-01 7.01230392e-02
-2.77040988e-01 -2.00850099e-01 -1.03684354e+00 5.57906270e-01
1.16069782e+00 -5.58947861e-01 -6.98960483e-01 4.57560688e-01
3.92912418e-01 -6.69030786e-01 -9.61945415e-01 8.14248860e-01
3.75216156e-01 -8.37452710e-01 1.07653272e+00 -1.36457533e-01
-4.71306266e-04 -7.51309156e-01 -3.98236096e-01 -1.08863032e+00
-1.19412810e-01 -8.20630193e-01 -4.34460282e-01 1.27977002e+00
-4.28592190e-02 -6.73639119e-01 8.08471918e-01 4.19478625e-01
-7.82900810e-01 -8.20975363e-01 -1.43719697e+00 -8.61581326e-01
-8.01919177e-02 -7.48699009e-01 2.65260279e-01 9.72632706e-01
-4.77852225e-01 -1.87102586e-01 -4.12373602e-01 6.37604535e-01
1.12802351e+00 -6.16485737e-02 1.09304869e+00 -1.47686231e+00
-2.08135650e-01 -2.43919849e-01 -5.87595105e-01 -1.47706282e+00
2.84647167e-01 -1.01089239e+00 1.66521713e-01 -1.22936428e+00
1.16079241e-01 -5.45312405e-01 -2.48905793e-01 2.43598521e-01
-9.12985727e-02 4.12593894e-02 -1.08301654e-01 7.01477587e-01
-3.73582214e-01 1.02577519e+00 1.05423939e+00 -3.40546966e-02
-2.11276621e-01 1.92092836e-01 -1.46835014e-01 1.00915480e+00
1.86074704e-01 -6.94248676e-01 -5.77683793e-03 1.63301229e-02
3.77848357e-01 -1.81953404e-02 7.09141612e-01 -8.69722903e-01
4.16050613e-01 -1.76465645e-01 4.80496466e-01 -1.16252720e+00
4.09743011e-01 -1.00960541e+00 9.83786732e-02 8.88554901e-02
1.11660898e-01 -2.63183028e-01 -5.04773974e-01 8.71012628e-01
-2.72924423e-01 -1.18890025e-01 7.17687011e-01 3.20586681e-01
-2.57491916e-01 8.43446195e-01 9.80758071e-02 1.06938273e-01
8.28812063e-01 -6.58451796e-01 2.41657495e-01 -3.23319018e-01
-9.89991307e-01 4.88303870e-01 7.76380897e-01 6.42582849e-02
9.62849140e-01 -1.64318466e+00 -8.32080901e-01 6.18729174e-01
4.55881625e-01 5.54103732e-01 2.34114200e-01 1.29317057e+00
-2.97968358e-01 -1.89493477e-01 3.57458562e-01 -1.35709691e+00
-9.06281888e-01 -3.58051546e-02 4.11760718e-01 -1.11686707e-01
-6.02988958e-01 8.39966059e-01 2.97650900e-02 -5.46016157e-01
2.57101715e-01 -4.23965335e-01 2.38319770e-01 1.42562926e-01
-1.90103039e-01 4.45405960e-01 -7.14641064e-02 -7.33207345e-01
-3.24960649e-01 9.78353143e-01 1.62326381e-01 -1.08601786e-01
1.21763122e+00 -1.31022066e-01 -3.09201002e-01 5.07365406e-01
1.22617042e+00 5.80510199e-01 -1.43004823e+00 -3.29307079e-01
1.22248262e-01 -7.88475394e-01 -2.51020510e-02 5.02953455e-02
-8.81964922e-01 4.05077755e-01 2.83262283e-01 -3.42949033e-01
7.53325582e-01 2.02133089e-01 5.54247022e-01 3.82841378e-01
4.75645304e-01 -1.06802726e+00 2.30714917e-01 6.84883773e-01
1.02557397e+00 -1.08568811e+00 2.33489409e-01 -4.74202901e-01
-3.51217464e-02 1.10490787e+00 5.37732661e-01 -6.14672065e-01
8.16272318e-01 -4.61999737e-02 -1.43078551e-01 -3.92033607e-01
-6.99838176e-02 -3.74752991e-02 8.18040669e-01 3.37566465e-01
-2.47961041e-02 -2.17362940e-01 6.74811304e-02 2.81790197e-01
-1.52020425e-01 -4.73966181e-01 1.99139610e-01 7.02943742e-01
-3.90900433e-01 -1.10126472e+00 -6.23538852e-01 5.00373483e-01
-8.21132436e-02 5.75221889e-02 -5.67215215e-03 8.05670142e-01
4.62456048e-02 4.80868638e-01 1.03180863e-01 -1.57183632e-01
4.56838399e-01 -6.06469251e-02 4.10847306e-01 -6.52266920e-01
-7.09282979e-02 8.64622176e-01 -1.38555169e-01 -9.38848257e-01
-3.28460008e-01 -9.80869055e-01 -1.24202073e+00 9.66897309e-02
-6.71719432e-01 1.37629196e-01 3.63089204e-01 1.03673851e+00
2.61713535e-01 1.42110780e-01 8.68193507e-01 -1.12276554e+00
-6.38698816e-01 -6.66637897e-01 -7.66538382e-01 3.37063432e-01
4.35519338e-01 -9.72068191e-01 -8.82517338e-01 -5.16030133e-01] | [7.820902347564697, -2.9573278427124023] |
6e55e27f-320d-40dd-8c5c-340d43bca6ec | improving-shadow-suppression-for-illumination | 1710.05073 | null | http://arxiv.org/abs/1710.05073v1 | http://arxiv.org/pdf/1710.05073v1.pdf | Improving Shadow Suppression for Illumination Robust Face Recognition | 2D face analysis techniques, such as face landmarking, face recognition and
face verification, are reasonably dependent on illumination conditions which
are usually uncontrolled and unpredictable in the real world. An illumination
robust preprocessing method thus remains a significant challenge in reliable
face analysis. In this paper we propose a novel approach for improving lighting
normalization through building the underlying reflectance model which
characterizes interactions between skin surface, lighting source and camera
sensor, and elaborates the formation of face color appearance. Specifically,
the proposed illumination processing pipeline enables the generation of
Chromaticity Intrinsic Image (CII) in a log chromaticity space which is robust
to illumination variations. Moreover, as an advantage over most prevailing
methods, a photo-realistic color face image is subsequently reconstructed which
eliminates a wide variety of shadows whilst retaining the color information and
identity details. Experimental results under different scenarios and using
various face databases show the effectiveness of the proposed approach to deal
with lighting variations, including both soft and hard shadows, in face
recognition. | ['Liming Chen', 'Jean-Marie Morvan', 'Xi Zhao', 'Wuming Zhang'] | 2017-10-13 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 5.23132801e-01 -5.50830007e-01 5.31484783e-01 -6.02964580e-01
-1.68380737e-01 -7.71452904e-01 6.34495378e-01 -2.63038993e-01
-1.50435328e-01 5.68862975e-01 -2.24248618e-01 1.00310527e-01
-1.11894019e-01 -5.24719954e-01 -4.36393201e-01 -1.04131782e+00
3.22017372e-01 1.58500317e-02 -2.77384728e-01 8.26572701e-02
3.87475580e-01 1.00810897e+00 -1.94287765e+00 -2.63375700e-01
6.57322466e-01 9.84150708e-01 -1.61991373e-01 6.40084028e-01
-1.26307547e-01 1.76512420e-01 -6.66136086e-01 -2.27502197e-01
4.91373450e-01 -3.10598195e-01 -7.71665350e-02 5.29418886e-01
4.78862673e-01 -2.18540370e-01 3.77469122e-01 1.10966837e+00
5.04758120e-01 2.15731218e-01 6.22101188e-01 -1.16675830e+00
-2.97450781e-01 -3.59745771e-01 -8.03148210e-01 -8.64536762e-02
4.37788725e-01 1.43723726e-01 1.38194859e-01 -1.01951694e+00
4.39803183e-01 1.23868167e+00 4.65775400e-01 4.37651575e-01
-1.23889399e+00 -7.05837488e-01 -1.93618704e-02 2.26801991e-01
-1.45573831e+00 -8.49519670e-01 1.13640487e+00 -1.22127250e-01
3.46150517e-01 4.29644853e-01 4.44623172e-01 8.59908998e-01
-1.51070282e-02 -1.60105318e-01 1.97962153e+00 -8.31417263e-01
1.31969884e-01 5.45020700e-01 -1.44481376e-01 6.12513840e-01
3.14156085e-01 8.23221058e-02 -5.25823295e-01 -2.32217293e-02
6.05108976e-01 -1.49370730e-01 -3.10312986e-01 -2.23444074e-01
-6.38875067e-01 3.45630944e-02 -4.36219871e-02 8.57130066e-02
-3.62311453e-01 -3.11313331e-01 3.06052491e-02 5.50523363e-02
3.34472507e-01 -2.27877468e-01 -2.03980327e-01 1.92058235e-01
-8.14152062e-01 -1.87356964e-01 6.19977832e-01 6.96806610e-01
9.70924795e-01 2.19955966e-01 1.39699191e-01 9.00140464e-01
7.64071763e-01 9.26585853e-01 -1.91462394e-02 -6.44563556e-01
-7.33008096e-03 7.42483199e-01 2.26219684e-01 -1.07653773e+00
-1.91136822e-01 -1.81364343e-01 -6.67629480e-01 6.76563084e-01
4.06453788e-01 -3.10129654e-02 -1.10529089e+00 1.51331222e+00
9.73218143e-01 4.33638513e-01 1.21831015e-01 7.30159760e-01
6.28920019e-01 2.16248319e-01 -7.02763274e-02 -5.84901273e-01
1.42324972e+00 -3.70683044e-01 -8.85999978e-01 1.21508062e-01
-3.23429078e-01 -1.37278807e+00 8.20935428e-01 5.14072776e-01
-9.44740415e-01 -7.08318651e-01 -1.07957268e+00 8.66413862e-02
-4.32192266e-01 2.28666514e-01 4.09802586e-01 1.32764173e+00
-1.17629278e+00 2.06291631e-01 -6.57462656e-01 -4.37267005e-01
1.76711217e-01 5.23462534e-01 -5.85913360e-01 -2.80169129e-01
-6.49990857e-01 8.06012690e-01 1.27774805e-01 7.32782841e-01
-8.33632588e-01 -3.39949876e-01 -6.80447340e-01 -2.65328020e-01
8.41306522e-02 -9.32244658e-02 4.88829166e-01 -1.22095883e+00
-1.95314634e+00 9.38496709e-01 -5.61023235e-01 1.89899653e-01
4.47610348e-01 9.61727574e-02 -7.80738950e-01 8.70144553e-03
-4.84928966e-01 -1.17690429e-01 1.43894351e+00 -1.52350068e+00
-2.03202888e-01 -8.06788385e-01 -3.54790092e-01 3.92571151e-01
-1.72877148e-01 3.51386219e-01 -7.02883661e-01 -1.19423613e-01
4.84235249e-02 -8.23794901e-01 2.57657945e-01 -1.55131817e-02
-1.68195620e-01 2.10235357e-01 1.09639204e+00 -7.74871290e-01
6.95744395e-01 -2.17949152e+00 -2.47485444e-01 8.05758119e-01
-3.02537054e-01 3.75913709e-01 2.18321290e-02 1.56596661e-01
-2.34410256e-01 -3.89903545e-01 -2.16513366e-01 -4.05236185e-01
-5.33611067e-02 1.56865895e-01 2.69357085e-01 9.23404694e-01
2.58379906e-01 2.31586009e-01 -3.10155749e-01 -4.76925135e-01
5.36436856e-01 1.25772941e+00 -5.87798357e-02 3.06730032e-01
1.28592089e-01 8.52864206e-01 -4.39528413e-02 9.92636442e-01
1.38239610e+00 6.61223114e-01 2.74565637e-01 -3.43028665e-01
-2.75884181e-01 -4.77719754e-01 -1.65713775e+00 1.29591739e+00
-5.38542449e-01 4.46025014e-01 7.07771182e-01 -5.92346489e-01
1.30233574e+00 3.53542477e-01 3.24840993e-01 -6.94332123e-01
5.46305597e-01 3.14470846e-03 -2.20255941e-01 -3.10432643e-01
2.03911215e-01 -1.18643217e-01 6.56719983e-01 3.27062935e-01
-1.26835614e-01 6.21729344e-02 -6.49068430e-02 -3.88493568e-01
3.42454731e-01 5.27691066e-01 1.05947725e-01 -3.37155491e-01
1.20098686e+00 -5.72768092e-01 5.31559169e-01 3.91392820e-02
-4.28078443e-01 4.56998020e-01 6.63354695e-02 -1.42482385e-01
-8.37017596e-01 -7.51805544e-01 -3.89604479e-01 6.81796610e-01
1.67656571e-01 1.30319774e-01 -1.11378562e+00 -1.78411886e-01
-3.07261825e-01 3.88478816e-01 -5.92942059e-01 2.01275736e-01
-5.78802288e-01 -9.72461581e-01 1.83748081e-01 -7.32499883e-02
4.43827838e-01 -7.56013513e-01 -4.52916384e-01 -2.41050020e-01
2.18186051e-01 -1.01710892e+00 -1.78856179e-01 -2.16141373e-01
-5.56110322e-01 -1.29570222e+00 -2.70697027e-01 -4.70129222e-01
1.08319545e+00 3.83142143e-01 8.80552351e-01 3.80806088e-01
-9.49080110e-01 5.27826071e-01 -1.53253630e-01 -6.12658203e-01
-3.13473314e-01 -6.87863350e-01 6.54387176e-02 8.24756086e-01
4.24354941e-01 -3.83771628e-01 -8.32915723e-01 3.19406748e-01
-8.83097231e-01 -3.15162331e-01 3.85105491e-01 4.76056159e-01
5.33554256e-01 5.03139913e-01 2.66498536e-01 -1.01281059e+00
3.33708405e-01 -1.09374635e-01 -1.00016975e+00 4.86268878e-01
-6.21542633e-01 -2.83521146e-01 5.27737021e-01 -2.59353593e-02
-1.85393822e+00 3.29456925e-01 5.08072153e-02 -9.08710286e-02
-7.20286012e-01 -1.76965103e-01 -6.64794683e-01 -6.53148830e-01
4.44818795e-01 2.15131342e-01 7.84116909e-02 -4.87109184e-01
2.76113421e-01 5.94739974e-01 5.19453287e-01 -6.13936841e-01
1.33542836e+00 7.81158745e-01 4.87728745e-01 -1.22037590e+00
-1.51294053e-01 -4.75930899e-01 -9.42830265e-01 -5.95392466e-01
6.95379555e-01 -6.89141691e-01 -9.87123430e-01 7.76325524e-01
-8.22058380e-01 2.63685137e-01 3.04001480e-01 2.57256895e-01
6.32754266e-02 5.08257866e-01 1.66533571e-02 -1.40614128e+00
-3.50303411e-01 -1.13884163e+00 8.69004011e-01 7.67997146e-01
3.10999095e-01 -1.10564184e+00 -1.12384677e-01 4.58276540e-01
4.96141851e-01 5.77634037e-01 8.28783214e-01 1.49798319e-01
-5.49332023e-01 -1.64867342e-01 -2.87092596e-01 4.11842376e-01
5.79043329e-01 6.38216138e-01 -1.51408708e+00 -3.22483629e-01
1.48760185e-01 8.56975019e-02 3.26163352e-01 1.36217520e-01
8.75015497e-01 9.19925198e-02 1.34043634e-01 7.40784168e-01
1.85698652e+00 2.75676668e-01 6.59839630e-01 1.23896137e-01
5.88348210e-01 8.95576656e-01 6.35130584e-01 5.94981909e-01
1.02885589e-01 5.24186790e-01 5.42118609e-01 -5.33931136e-01
-3.52207005e-01 2.48398021e-01 2.29984030e-01 3.24837327e-01
-3.69641274e-01 1.43815652e-01 -4.54328209e-01 1.27598077e-01
-1.13717401e+00 -7.50239491e-01 -2.40312919e-01 2.50853276e+00
6.67056561e-01 -4.74618971e-01 -1.00401804e-01 4.40322608e-01
7.70830512e-01 -2.81784207e-01 -4.10105735e-01 -4.76666987e-01
-4.09716070e-01 6.15487218e-01 4.97550279e-01 5.88188350e-01
-9.18019414e-01 7.07666636e-01 5.57090378e+00 2.56197870e-01
-1.36358857e+00 -1.09997019e-01 3.90746385e-01 1.86962306e-01
-1.92129061e-01 -9.92746130e-02 -6.81456506e-01 2.83576965e-01
7.03289032e-01 1.72150478e-01 7.58326173e-01 4.45641994e-01
4.19541091e-01 -4.20299768e-01 -7.16521621e-01 1.14287055e+00
4.71401840e-01 -3.50764304e-01 -2.29214624e-01 -2.83981077e-02
6.96170092e-01 -5.95705748e-01 4.01903361e-01 -5.05772054e-01
-1.32491216e-01 -1.03767705e+00 4.11548942e-01 6.09073102e-01
8.80980372e-01 -9.12534475e-01 4.69839036e-01 -1.29279271e-01
-1.08814192e+00 1.83912903e-01 -2.18193397e-01 2.08500043e-01
-1.33319989e-01 3.60817343e-01 -1.05304933e+00 5.96552253e-01
5.63052416e-01 3.44937861e-01 -8.02512765e-01 8.27213585e-01
-3.12543929e-01 3.86373758e-01 -4.64192510e-01 4.12447929e-01
-3.13894838e-01 -6.19422197e-01 4.13454741e-01 1.11782873e+00
1.85866490e-01 1.65174797e-01 -1.96103871e-01 5.78605175e-01
4.00391631e-02 3.29366267e-01 -3.92622858e-01 2.04777911e-01
3.75536293e-01 1.67669249e+00 -9.88361955e-01 1.50526285e-01
-4.52480644e-01 9.16626155e-01 -1.96859568e-01 6.99752867e-01
-7.11032093e-01 -3.94775458e-02 7.04637468e-01 -1.74798332e-02
-6.86079934e-02 -2.02395245e-01 -2.39152804e-01 -8.89257729e-01
2.54064858e-01 -8.57090235e-01 1.17483944e-01 -4.79085594e-01
-7.64463246e-01 6.48016751e-01 -1.06557108e-01 -9.32580113e-01
2.94924229e-01 -8.78892481e-01 -5.78361571e-01 1.15109265e+00
-2.14753723e+00 -1.48898351e+00 -6.72947466e-01 9.80883002e-01
3.59956861e-01 -9.33123007e-02 1.03824115e+00 5.05507827e-01
-9.62195158e-01 6.20416582e-01 2.08387390e-01 -2.95667917e-01
9.08943117e-01 -9.42359746e-01 -2.14078218e-01 1.18502271e+00
-9.59177837e-02 8.20932984e-01 7.13122189e-01 -4.15099174e-01
-1.87425792e+00 -9.13310707e-01 2.84162790e-01 -2.13463143e-01
1.11025661e-01 -4.03196394e-01 -7.29412198e-01 1.92450508e-01
3.24527889e-01 -5.68012111e-02 7.87694633e-01 -9.66429263e-02
-3.04257005e-01 -5.84436119e-01 -1.50099552e+00 2.87504166e-01
5.33055782e-01 -4.89261448e-01 -3.19315903e-02 2.27753639e-01
-4.40543592e-01 -2.10740238e-01 -8.13530862e-01 1.66671634e-01
8.64535332e-01 -1.15101373e+00 9.71343994e-01 1.65191889e-02
-3.66635293e-01 -7.87056386e-01 -9.97223854e-02 -8.55646849e-01
1.36275426e-01 -7.49374032e-01 4.07646328e-01 1.77559197e+00
9.38150063e-02 -7.98993289e-01 6.25408828e-01 9.27402735e-01
3.02626580e-01 -9.75041911e-02 -7.60908246e-01 -3.32495183e-01
-6.02610111e-01 -5.13706990e-02 6.04302227e-01 7.15474844e-01
-7.12816477e-01 -2.00046942e-01 -3.94351095e-01 6.63692892e-01
1.30905938e+00 1.02722012e-01 9.09133554e-01 -1.38039792e+00
1.41694486e-01 -1.04069427e-01 -3.35881799e-01 1.60296798e-01
2.52082199e-01 -3.82148743e-01 7.87682459e-02 -1.00949252e+00
1.64049968e-01 -5.71190834e-01 -4.00020540e-01 8.43497887e-02
-1.73857346e-01 7.72724569e-01 -7.11282268e-02 1.73783340e-02
-5.44025600e-02 1.47851259e-01 9.01449442e-01 1.68312341e-01
-1.45103857e-01 -9.47831795e-02 -3.54662061e-01 7.94538021e-01
7.01926112e-01 -6.27182201e-02 -5.13293326e-01 -4.01474416e-01
-1.48377940e-01 -3.99445832e-01 3.42574328e-01 -9.42817807e-01
1.36208788e-01 -3.11482221e-01 8.65566492e-01 -4.08086061e-01
5.76735735e-01 -1.44749355e+00 6.27881765e-01 7.55837187e-02
1.79816246e-01 -4.82292473e-02 2.57478923e-01 4.68091458e-01
-2.49789841e-02 2.43829954e-02 1.26327848e+00 -8.26850384e-02
-6.33223534e-01 7.52863064e-02 1.46748096e-01 -4.96799439e-01
1.26374328e+00 -6.77335382e-01 -1.90155417e-01 4.33109626e-02
-3.82805914e-01 -2.51885653e-01 8.23211491e-01 2.09290668e-01
5.64867437e-01 -1.00777912e+00 -6.88367188e-01 8.36549342e-01
-5.01434393e-02 -3.00483167e-01 5.51150024e-01 7.87903309e-01
-7.19175637e-01 1.02387607e-01 -4.84037966e-01 -4.53817457e-01
-1.92734706e+00 2.77592957e-01 1.77141145e-01 5.02777934e-01
-1.74562737e-01 7.98041701e-01 7.27277771e-02 -6.91909418e-02
2.83781290e-01 9.90048721e-02 -4.98748869e-01 -7.36000240e-02
6.82230234e-01 5.95947444e-01 3.20850909e-01 -1.22591662e+00
-5.12901187e-01 1.12392974e+00 1.90336391e-01 5.33885323e-03
1.19303024e+00 -6.40066147e-01 -5.13584912e-01 1.32853284e-01
9.89312470e-01 2.37023294e-01 -1.17671633e+00 -2.61754040e-02
-1.15665436e-01 -1.00006580e+00 4.91110235e-02 -8.94880712e-01
-1.15658677e+00 8.82184207e-01 1.08703184e+00 -1.66417331e-01
1.63359141e+00 -7.55115032e-01 2.23489717e-01 -7.37190694e-02
1.98440701e-01 -1.33047831e+00 -3.47267777e-01 -2.03301627e-02
6.89845681e-01 -1.23980439e+00 1.47310421e-01 -6.80110276e-01
-2.44827896e-01 1.29432249e+00 4.73976433e-01 2.37145171e-01
6.63242102e-01 3.35956246e-01 5.66978395e-01 -7.20729753e-02
-2.71856397e-01 -2.44337440e-01 3.13936919e-01 7.71577656e-01
5.22037089e-01 -3.73811573e-02 -4.67880517e-02 -5.07547259e-02
1.86792135e-01 -3.51642281e-01 3.03543866e-01 8.20189834e-01
-1.30207077e-01 -1.26949263e+00 -9.88835037e-01 -1.77644327e-01
-6.20360613e-01 2.10442215e-01 -4.13696855e-01 5.23938000e-01
5.01510262e-01 1.14213085e+00 -3.05675894e-01 9.21027884e-02
3.35327059e-01 2.52999604e-01 7.31022894e-01 -3.53567779e-01
-3.98004919e-01 4.25218731e-01 -2.17971206e-01 -4.26486641e-01
-9.19792831e-01 -6.77971065e-01 -9.73368704e-01 -1.92519933e-01
-5.10140955e-01 -2.13558868e-01 1.52041304e+00 7.18875349e-01
1.24192543e-01 1.78500623e-01 1.05402529e+00 -8.60384881e-01
-3.77757400e-02 -6.98404074e-01 -7.52088368e-01 5.88839710e-01
3.26445848e-01 -8.79848897e-01 -2.89233059e-01 3.90974432e-01] | [13.127579689025879, 0.6567527055740356] |
8e298381-af84-4e6f-9a7e-93e94ec35948 | jointly-complementary-competitive-influence | 2302.09620 | null | https://arxiv.org/abs/2302.09620v1 | https://arxiv.org/pdf/2302.09620v1.pdf | Jointly Complementary&Competitive Influence Maximization with Concurrent Ally-Boosting and Rival-Preventing | In this paper, we propose a new influence spread model, namely, Complementary\&Competitive Independent Cascade (C$^2$IC) model. C$^2$IC model generalizes three well known influence model, i.e., influence boosting (IB) model, campaign oblivious (CO)IC model and the IC-N (IC model with negative opinions) model. This is the first model that considers both complementary and competitive influence spread comprehensively under multi-agent environment. Correspondingly, we propose the Complementary\&Competitive influence maximization (C$^2$IM) problem. Given an ally seed set and a rival seed set, the C$^2$IM problem aims to select a set of assistant nodes that can boost the ally spread and prevent the rival spread concurrently. We show the problem is NP-hard and can generalize the influence boosting problem and the influence blocking problem. With classifying the different cascade priorities into 4 cases by the monotonicity and submodularity (M\&S) holding conditions, we design 4 algorithms respectively, with theoretical approximation bounds provided. We conduct extensive experiments on real social networks and the experimental results demonstrate the effectiveness of the proposed algorithms. We hope this work can inspire abundant future exploration for constructing more generalized influence models that help streamline the works of this area. | ['Minghui Wu', 'Can Wang', 'Mengqi Xue', 'Wujian Yang', 'Wenjie Tian', 'Qihao Shi'] | 2023-02-19 | null | null | null | null | ['blocking'] | ['natural-language-processing'] | [ 3.92041802e-01 4.44614559e-01 -6.11055791e-01 -7.83965960e-02
6.29236773e-02 -7.48518348e-01 7.61843741e-01 -1.40180290e-01
-2.31272936e-01 1.15048611e+00 1.69208080e-01 -3.98938149e-01
-6.26303494e-01 -1.03402197e+00 -6.17310166e-01 -8.96386564e-01
-5.46554506e-01 7.65902102e-01 4.14160192e-01 -6.87635124e-01
5.43687582e-01 -3.62667814e-02 -1.16096687e+00 -2.01599374e-01
8.62970054e-01 5.29755592e-01 4.02507745e-02 7.04684317e-01
-1.17366135e-01 7.41956055e-01 -6.52813196e-01 -3.85878623e-01
4.96305943e-01 -6.38188183e-01 -6.81655288e-01 -6.23518936e-02
-4.65660572e-01 -1.41435126e-02 -4.68592867e-02 9.68874335e-01
3.61920983e-01 -1.19492583e-01 6.90051913e-01 -1.97788894e+00
-6.14142895e-01 1.14588273e+00 -1.16240799e+00 6.12233162e-01
2.17747822e-01 -3.71191293e-01 9.92137134e-01 -2.42356941e-01
8.25422764e-01 1.35870242e+00 1.81614280e-01 1.69140562e-01
-7.45059788e-01 -7.51075387e-01 1.01595640e+00 2.20473677e-01
-1.00612664e+00 1.23958945e-01 8.47352445e-01 -3.28767635e-02
5.05183041e-01 4.15848672e-01 1.13580680e+00 2.89081395e-01
8.20221566e-03 1.09194684e+00 1.66403913e+00 -3.07345182e-01
2.37751499e-01 1.88399002e-01 2.65962511e-01 4.65862662e-01
5.99961579e-01 1.28316730e-01 -4.83084530e-01 -4.40610975e-01
6.67892754e-01 -5.02224378e-02 -3.36558402e-01 1.36861697e-01
-9.53178346e-01 1.06421137e+00 7.17433631e-01 4.65006143e-01
-3.09282005e-01 2.94736773e-01 -1.55387998e-01 8.52042437e-01
7.85715878e-01 2.66879916e-01 -5.40366352e-01 2.26142108e-01
-6.70901060e-01 1.37701496e-01 1.14623451e+00 1.18568254e+00
7.37064242e-01 -2.56516427e-01 3.27493735e-02 3.06074589e-01
6.39288545e-01 8.10733497e-01 -4.37998176e-01 -6.34714246e-01
3.24903697e-01 7.76044428e-01 1.50079861e-01 -1.07329118e+00
-3.28619897e-01 -7.10207224e-01 -8.94446671e-01 -1.46212414e-01
-5.05473912e-02 -5.76749742e-01 -5.75351179e-01 1.72085810e+00
6.89695954e-01 3.30434203e-01 -2.40774274e-01 8.17172706e-01
6.51929200e-01 6.30190134e-01 -1.54559076e-01 -9.67365265e-01
9.92169559e-01 -1.03161979e+00 -7.40247846e-01 -2.28119656e-01
3.21101248e-01 -8.36656272e-01 3.18059415e-01 2.07554400e-01
-1.09384966e+00 1.20187409e-01 -1.32035422e+00 8.63865614e-01
-2.09606588e-01 -7.57336259e-01 1.19966078e+00 7.10679352e-01
-1.47637820e+00 -9.27020703e-03 -1.53227150e-01 -1.28213376e-01
3.74694496e-01 7.94532120e-01 5.97351231e-02 -4.02474552e-01
-1.49103725e+00 5.23965418e-01 -2.55384624e-01 6.94784522e-02
-1.30406928e+00 -4.97738302e-01 -1.63113207e-01 -3.49121660e-01
9.03430760e-01 -9.10301208e-01 7.92060733e-01 -1.18863654e+00
-1.03465414e+00 4.63110894e-01 -1.27330169e-01 -7.45863691e-02
2.99881220e-01 2.65125513e-01 -3.07502836e-01 8.63151401e-02
3.09882432e-01 2.26365477e-01 6.52577460e-01 -1.60914540e+00
-1.05224168e+00 -5.65198839e-01 7.99552917e-01 6.09709561e-01
-1.22519128e-01 3.79032522e-01 -2.94595301e-01 -1.45672649e-01
1.60573814e-02 -9.84303296e-01 -9.12993789e-01 -8.24302793e-01
-5.37685692e-01 -2.65678942e-01 5.26079774e-01 4.93508801e-02
1.32874405e+00 -1.40894747e+00 4.49988335e-01 7.55006373e-01
4.52870041e-01 -1.46410942e-01 -2.54177392e-01 5.49132824e-01
3.36339891e-01 6.72887385e-01 -6.87420815e-02 3.34729254e-01
-3.13784182e-01 2.78341901e-02 1.64006263e-01 5.81902266e-01
-2.15419009e-01 8.91900778e-01 -8.53876948e-01 -4.01151031e-01
-4.98482525e-01 1.26287028e-01 -4.31310803e-01 -1.80248216e-01
-3.07534724e-01 3.69733036e-01 -1.00923383e+00 8.58022273e-01
9.37802970e-01 -3.02832276e-01 2.96553761e-01 4.30947512e-01
-9.57988873e-02 -3.15287352e-01 -1.30088949e+00 6.30886495e-01
-1.19710900e-01 7.83739164e-02 8.94938409e-01 -7.40229249e-01
6.10812485e-01 1.02143243e-01 4.96741325e-01 -2.68025279e-01
5.19193709e-01 1.98084921e-01 2.70248026e-01 -8.30666870e-02
3.08527678e-01 -4.29991961e-01 -1.99168637e-01 1.06496620e+00
-4.39304113e-01 -1.12741277e-01 4.93514150e-01 9.70060885e-01
1.28064382e+00 -7.31198370e-01 1.27008319e-01 -6.98489308e-01
7.42762685e-01 2.42474928e-01 5.26195288e-01 1.08720160e+00
-4.87704605e-01 -2.64660448e-01 9.38759267e-01 7.57179260e-02
-6.29504383e-01 -5.98682165e-01 3.54886115e-01 1.39601600e+00
8.71148348e-01 -1.21256746e-01 -6.30335093e-01 -8.57899487e-01
1.02416798e-01 2.90937334e-01 -8.73567581e-01 2.81687468e-01
-6.97376668e-01 -1.30122817e+00 8.09726268e-02 2.71992922e-01
6.41429901e-01 -8.30728829e-01 4.17206526e-01 2.48239174e-01
-1.07974537e-01 -3.40748996e-01 -5.42579234e-01 5.21889590e-02
-8.37114632e-01 -1.44232380e+00 -5.84726095e-01 -7.12469280e-01
1.11669838e+00 7.12857842e-01 1.18536198e+00 6.70085013e-01
1.62473135e-03 5.47393024e-01 -7.25156784e-01 -6.38708413e-01
-9.06686261e-02 7.57902442e-03 -1.45999482e-02 -9.82529297e-02
3.54260772e-01 -6.45362437e-01 -1.03717399e+00 8.43067229e-01
-9.52264607e-01 -2.47860789e-01 8.84069026e-01 3.66148144e-01
5.10508239e-01 3.70237470e-01 1.04765117e+00 -1.14601672e+00
1.23833191e+00 -8.89187217e-01 -4.74442214e-01 1.52434841e-01
-1.02020240e+00 -2.44827643e-01 3.28076690e-01 -2.28506774e-01
-1.13340712e+00 -4.96704042e-01 2.51028687e-01 4.69981104e-01
6.59925938e-01 7.60924160e-01 -1.99139923e-01 -2.22239360e-01
3.62934440e-01 -4.80980724e-02 -2.70511899e-02 1.83778186e-03
5.61543465e-01 6.82790399e-01 -4.10504013e-01 -1.91945478e-01
7.65279412e-01 8.53278816e-01 2.61521161e-01 -3.97787631e-01
-6.10927045e-01 -6.23174369e-01 -2.21481755e-01 -6.93822742e-01
3.16852272e-01 -7.38202453e-01 -7.85087705e-01 2.56594539e-01
-9.04017627e-01 1.37773767e-01 4.15516049e-01 1.79256573e-01
-2.65777260e-01 1.05970293e-01 -4.84778434e-01 -1.23719263e+00
-4.40788209e-01 -8.42910528e-01 2.77205825e-01 3.84110391e-01
3.03097785e-01 -8.23466063e-01 1.85652539e-01 5.33125401e-01
6.45336926e-01 2.69596606e-01 6.26503289e-01 -5.50899029e-01
-1.13756967e+00 -2.99687654e-01 -6.37362748e-02 -2.18888566e-01
-2.36923784e-01 -6.01439439e-02 -1.56048194e-01 -4.24386472e-01
1.38425073e-02 7.53703639e-02 1.05206823e+00 7.56720483e-01
1.03429534e-01 -4.03304398e-01 -9.58374500e-01 -8.71707797e-02
1.34540975e+00 5.75646758e-01 5.78112543e-01 4.26709592e-01
2.31394932e-01 6.34375691e-01 1.03108108e+00 3.68631184e-01
3.63909185e-01 2.81423032e-02 8.43634069e-01 -2.16262251e-01
2.98752636e-01 -1.08846597e-01 1.80312917e-01 1.09950042e+00
-7.02067077e-01 -5.08242249e-01 -3.38321090e-01 5.75309336e-01
-1.87492168e+00 -8.88068020e-01 -4.74334776e-01 1.78346884e+00
7.00132132e-01 2.54231572e-01 3.05912584e-01 1.90723583e-01
1.20469606e+00 1.45095319e-01 -3.59440863e-01 -3.72868180e-01
-3.02006930e-01 3.71115445e-03 7.96370566e-01 9.62685287e-01
-8.62454772e-01 8.73258054e-01 6.29170084e+00 9.83587921e-01
-3.83681804e-01 5.77693641e-01 9.20332670e-01 -4.39054728e-01
-7.21221685e-01 4.10910845e-01 -1.00110555e+00 2.28493243e-01
3.57732505e-01 -6.64446294e-01 6.49534881e-01 7.13683486e-01
1.77986622e-01 -2.81553894e-01 -4.43184197e-01 1.50918245e-01
1.46148831e-01 -1.14323139e+00 -2.50029385e-01 1.14356063e-01
1.47079694e+00 7.87956193e-02 -2.19902322e-01 2.10368499e-01
8.91575098e-01 -7.78857529e-01 2.33427882e-01 8.26763585e-02
4.40681189e-01 -1.05704427e+00 8.10904622e-01 5.04766047e-01
-1.19617355e+00 -2.85922617e-01 -3.92264694e-01 -4.65703964e-01
5.80106139e-01 7.52568245e-01 -4.69131082e-01 7.66076863e-01
5.00658810e-01 6.48860574e-01 -8.90143290e-02 8.66127253e-01
-7.22267210e-01 5.01985669e-01 -4.20642942e-01 -9.02424097e-01
4.55787361e-01 -3.82571697e-01 1.01951146e+00 6.80197835e-01
-1.16812043e-01 4.17960644e-01 1.34045973e-01 3.30759883e-01
-1.35946572e-01 3.54941398e-01 -5.20987630e-01 4.31683920e-02
4.86300737e-01 1.41835046e+00 -1.16614997e+00 -3.11933845e-01
-2.57155653e-02 4.43331182e-01 -1.14295311e-01 4.19592202e-01
-8.53339434e-01 -1.98871285e-01 2.45464504e-01 4.37270075e-01
3.61245543e-01 2.33421475e-01 -2.35643372e-01 -8.19671571e-01
-2.95495778e-01 -8.35625827e-01 5.14549255e-01 -1.88139707e-01
-1.21394014e+00 7.58645535e-01 2.21297309e-01 -7.05886245e-01
2.23522291e-01 -1.51741147e-01 -8.14269304e-01 5.06266952e-01
-1.74083352e+00 -1.04759645e+00 9.51625314e-03 6.96318746e-01
1.78177878e-01 5.17208390e-02 -1.78311728e-02 7.87757784e-02
-2.62542218e-01 -4.26293649e-02 8.47310796e-02 -5.60165584e-01
3.13947231e-01 -1.01411200e+00 -2.96828330e-01 6.59049988e-01
-4.81390297e-01 7.12783098e-01 6.78250015e-01 -1.00746417e+00
-1.41770792e+00 -8.07757437e-01 8.80730867e-01 -1.94549724e-01
6.86432958e-01 -1.27700120e-01 6.27314150e-02 4.97675329e-01
6.51575804e-01 -6.57793880e-01 8.01308870e-01 3.83185863e-01
1.76605001e-01 -1.35295659e-01 -1.34085035e+00 6.68537140e-01
1.48606741e+00 3.38741779e-01 -8.97008032e-02 4.87147719e-01
9.21281695e-01 3.75308007e-01 -6.88774943e-01 6.98649466e-01
2.77101040e-01 -8.76011431e-01 6.72269881e-01 -5.84794641e-01
5.94513714e-01 -4.30068582e-01 5.91871329e-02 -1.32559347e+00
-5.80271780e-01 -7.05032110e-01 1.34163409e-01 1.00666106e+00
8.92395854e-01 -8.45135689e-01 8.54189038e-01 2.62178212e-01
2.98556715e-01 -1.12810361e+00 -8.74527812e-01 -3.87553126e-01
2.00815741e-02 -1.00562468e-01 8.61475706e-01 9.52883363e-01
2.62255102e-01 7.03145742e-01 -3.48926038e-01 -3.78121212e-02
7.01464534e-01 3.53703707e-01 6.81238592e-01 -1.24377406e+00
-6.42595470e-01 -5.54651856e-01 7.60368034e-02 -1.28483558e+00
-1.49804920e-01 -7.45573819e-01 -2.00393617e-01 -1.63885963e+00
8.02852690e-01 -9.44877505e-01 -3.47236335e-01 1.56659380e-01
-3.96037608e-01 1.72037095e-01 1.59146592e-01 4.36212987e-01
-1.01404810e+00 3.75777304e-01 1.85176623e+00 -4.32248741e-01
-4.03075635e-01 3.92824143e-01 -1.57608199e+00 6.61722064e-01
8.59711349e-01 -5.68884432e-01 -7.51801670e-01 -9.63595212e-02
9.52528059e-01 2.51262277e-01 -2.44657069e-01 5.02181910e-02
4.51990634e-01 -5.23452342e-01 -9.85018387e-02 -7.87123680e-01
-2.37637721e-02 -5.22149861e-01 1.57144830e-01 6.09725893e-01
-2.26972386e-01 -1.06337771e-01 -3.02851558e-01 1.12595499e+00
-1.41672865e-01 -2.24547088e-01 2.34990865e-01 -4.44953710e-01
-2.50407726e-01 3.92574072e-01 -3.24094355e-01 -4.21718918e-02
1.36125398e+00 -1.24407960e-02 -8.97179842e-01 -9.27654147e-01
-6.99110150e-01 8.61585498e-01 1.37270510e-01 2.33294815e-01
4.45787400e-01 -1.13819957e+00 -1.18311322e+00 -1.75837383e-01
-2.25494474e-01 -2.80949801e-01 2.62898117e-01 1.35801494e+00
-1.61910638e-01 3.40321809e-01 1.64896920e-01 -2.59574711e-01
-1.40818381e+00 6.93387568e-01 -8.21823180e-02 -7.24848390e-01
4.28830415e-01 1.14829755e+00 1.94637015e-01 -3.26684177e-01
-1.08658373e-01 6.56965435e-01 -5.47292173e-01 7.44984858e-03
4.76323813e-01 7.69774795e-01 -5.61331689e-01 -4.46829736e-01
-4.95875150e-01 2.35995039e-01 -3.38258535e-01 -4.01850015e-01
1.29401720e+00 -3.23041946e-01 -9.74835575e-01 -2.23383173e-01
5.51145732e-01 1.42814711e-01 -5.49529672e-01 -2.11844873e-02
-3.62389773e-01 -5.11576116e-01 1.63020343e-01 -1.00947452e+00
-1.28739369e+00 -8.61072615e-02 -8.25950429e-02 8.50155890e-01
1.32846439e+00 4.28844422e-01 1.57464281e-01 4.41359915e-02
6.76781416e-01 -1.16639090e+00 -1.21633597e-01 3.77111822e-01
9.15542006e-01 -7.42823362e-01 4.11372900e-01 -1.03915644e+00
-5.72317064e-01 2.75145561e-01 6.23351812e-01 -2.65828401e-01
1.24959099e+00 4.11102116e-01 -2.68233597e-01 -5.34842551e-01
-1.05570519e+00 -5.05230367e-01 -3.16246480e-01 4.59123701e-01
2.10743204e-01 3.63074303e-01 -1.33505988e+00 7.44381189e-01
1.59182236e-01 -9.59960073e-02 5.98258674e-01 1.13237500e+00
-9.05133009e-01 -1.36358762e+00 -6.23665571e-01 8.09126318e-01
-4.06238496e-01 -3.50101084e-01 -1.13537323e+00 8.57353985e-01
2.65033841e-01 1.70948052e+00 -3.64903182e-01 -5.00641048e-01
1.26405656e-01 -6.61695898e-01 3.89640003e-01 -1.28531232e-01
-8.70117605e-01 5.70927918e-01 3.80357355e-01 3.16678658e-02
-9.43082988e-01 -3.31025839e-01 -8.52302551e-01 -8.64559770e-01
-1.07024288e+00 4.94732022e-01 5.80869675e-01 7.60879636e-01
2.46048376e-01 3.47542644e-01 1.15753973e+00 -3.08848739e-01
-3.42839271e-01 -1.31497204e+00 -8.90242457e-01 -1.50391385e-01
-3.29759359e-01 -7.69751012e-01 -9.36787307e-01 -4.46396887e-01] | [6.859138011932373, 5.365538597106934] |
bc8be382-01fe-4f91-b996-8ce9d38d96ce | class-anchor-margin-loss-for-content-based | 2306.00630 | null | https://arxiv.org/abs/2306.00630v2 | https://arxiv.org/pdf/2306.00630v2.pdf | Class Anchor Margin Loss for Content-Based Image Retrieval | The performance of neural networks in content-based image retrieval (CBIR) is highly influenced by the chosen loss (objective) function. The majority of objective functions for neural models can be divided into metric learning and statistical learning. Metric learning approaches require a pair mining strategy that often lacks efficiency, while statistical learning approaches are not generating highly compact features due to their indirect feature optimization. To this end, we propose a novel repeller-attractor loss that falls in the metric learning paradigm, yet directly optimizes for the L2 metric without the need of generating pairs. Our loss is formed of three components. One leading objective ensures that the learned features are attracted to each designated learnable class anchor. The second loss component regulates the anchors and forces them to be separable by a margin, while the third objective ensures that the anchors do not collapse to zero. Furthermore, we develop a more efficient two-stage retrieval system by harnessing the learned class anchors during the first stage of the retrieval process, eliminating the need of comparing the query with every image in the database. We establish a set of four datasets (CIFAR-100, Food-101, SVHN, and Tiny ImageNet) and evaluate the proposed objective in the context of few-shot and full-set training on the CBIR task, by using both convolutional and transformer architectures. Compared to existing objective functions, our empirical evidence shows that the proposed objective is generating superior and more consistent results. | ['Radu Tudor Ionescu', 'Alexandru Ghita'] | 2023-06-01 | null | null | null | null | ['metric-learning', 'content-based-image-retrieval', 'metric-learning'] | ['computer-vision', 'computer-vision', 'methodology'] | [ 5.67858443e-02 -2.92046756e-01 -2.33272761e-01 -5.61665356e-01
-1.03785944e+00 -3.34221721e-01 6.70004487e-01 3.43993515e-01
-7.80275643e-01 5.53086579e-01 2.13825256e-02 8.73492807e-02
-7.65477896e-01 -8.64338100e-01 -5.41671693e-01 -8.06806028e-01
-1.52410060e-01 3.71707261e-01 1.76656559e-01 -3.25139850e-01
2.29446441e-01 4.35485899e-01 -1.73248267e+00 1.40718371e-01
6.20054662e-01 1.71261787e+00 1.70865119e-01 2.82052785e-01
3.72351445e-02 7.90438414e-01 -4.81825441e-01 -3.73231322e-01
4.66705412e-01 -4.58980858e-01 -6.46906793e-01 -5.36745429e-01
3.50933164e-01 -3.18567395e-01 -4.47324961e-01 9.17226493e-01
6.96466565e-01 3.25989455e-01 9.33818579e-01 -1.27921927e+00
-8.90299380e-01 4.59546775e-01 -3.13310266e-01 1.28896460e-01
1.77504607e-02 -3.06283087e-01 1.53660107e+00 -9.50269759e-01
5.09328008e-01 9.81530547e-01 4.64942247e-01 5.18119216e-01
-9.10490692e-01 -6.20740354e-01 -1.30122542e-01 3.57322842e-01
-1.66376925e+00 -2.53662050e-01 8.38625729e-01 -2.83046752e-01
5.87372065e-01 1.56107530e-01 3.27605993e-01 8.95793676e-01
-2.03522444e-02 8.34436119e-01 6.79330826e-01 -4.75147575e-01
3.79722685e-01 3.50500286e-01 3.23955983e-01 6.84720099e-01
-2.55336985e-02 2.79376388e-01 -6.03058457e-01 -2.28973955e-01
2.33121470e-01 2.11854175e-01 -3.42426181e-01 -6.51622117e-01
-9.37907636e-01 1.15427780e+00 7.03413904e-01 6.02364361e-01
-1.57883838e-01 2.90151030e-01 3.63216132e-01 5.78265429e-01
3.65093112e-01 6.11884773e-01 -1.37308151e-01 2.08193064e-01
-1.05683100e+00 2.51002342e-01 6.22300327e-01 7.25541353e-01
8.92841339e-01 -3.09281737e-01 -6.08596742e-01 9.10005331e-01
2.78132498e-01 4.25278902e-01 8.40405822e-01 -5.68913519e-01
1.04367964e-01 6.72285616e-01 -1.09339125e-01 -1.24174750e+00
-1.21326119e-01 -4.64199156e-01 -7.42442429e-01 9.23826322e-02
1.32819578e-01 2.56360829e-01 -7.28043258e-01 1.89450514e+00
-2.04949528e-02 -5.82669526e-02 -1.17286555e-01 1.02131081e+00
9.05606627e-01 4.50897723e-01 -1.72738135e-01 8.29205960e-02
8.95347953e-01 -8.45662355e-01 -4.03438658e-01 1.29525840e-01
6.06304944e-01 -4.31075543e-01 1.16857600e+00 2.56410569e-01
-8.69294047e-01 -3.49425435e-01 -1.50639784e+00 -1.71139985e-01
-6.10765338e-01 8.17156676e-03 4.13686544e-01 4.08454180e-01
-1.02748215e+00 9.31754410e-01 -4.47379082e-01 5.61612137e-02
4.21815813e-01 3.11192721e-01 -3.36033970e-01 -3.94780152e-02
-1.38721836e+00 8.36705744e-01 2.75004059e-01 -1.08235016e-01
-8.78867030e-01 -7.39164591e-01 -7.62642682e-01 4.59670097e-01
-2.39566918e-02 -4.80664819e-01 8.59718323e-01 -1.06532776e+00
-1.17645741e+00 8.78337741e-01 2.68700361e-01 -6.71379268e-01
5.22464335e-01 -1.05611131e-01 -4.89337929e-02 1.99619502e-01
-4.41363864e-02 8.67237210e-01 8.84276330e-01 -1.09550083e+00
-4.77924258e-01 -3.70446086e-01 1.15926675e-01 1.91461429e-01
-1.04467583e+00 -4.35650162e-02 -3.46746325e-01 -5.75063407e-01
-1.87932476e-01 -6.58960879e-01 1.17089711e-02 3.48906815e-01
-2.57626772e-01 -3.65470588e-01 7.06748068e-01 -1.92867830e-01
1.22517264e+00 -2.23806214e+00 1.26046658e-01 4.45796072e-01
2.56606072e-01 2.44601652e-01 -4.08125460e-01 1.86912954e-01
7.40748793e-02 1.22426286e-01 -1.86719432e-01 -3.91596615e-01
1.39837980e-01 -6.06075600e-02 -4.07560050e-01 4.31698143e-01
1.38175502e-01 9.54600215e-01 -8.33340645e-01 -4.22464728e-01
1.17402911e-01 6.89302802e-01 -5.79427481e-01 1.66533664e-01
2.13884003e-02 -1.95070952e-01 -4.38762695e-01 5.31948864e-01
3.30157220e-01 -2.50388592e-01 -1.99239105e-01 -2.27484807e-01
2.01896071e-01 7.76259229e-02 -1.02687645e+00 1.81644154e+00
-3.42419028e-01 6.45901680e-01 -3.20623308e-01 -1.18841219e+00
1.08332145e+00 6.94295019e-02 8.20015788e-01 -1.21975827e+00
1.88936442e-01 2.15828642e-01 -1.53865770e-01 -1.02777392e-01
3.84026110e-01 6.79199323e-02 -2.33241245e-02 5.53126872e-01
2.65916318e-01 5.98767288e-02 2.51899809e-01 2.12188780e-01
1.10896850e+00 -1.08192727e-01 -6.56101853e-02 -4.17072833e-01
4.66260314e-01 -2.05692813e-01 4.39850688e-01 7.47079790e-01
-2.41959915e-01 7.08027124e-01 2.16135532e-01 -4.74393219e-01
-9.11095321e-01 -1.06362915e+00 -2.60923564e-01 1.21149099e+00
3.91936570e-01 -2.71249563e-01 -5.10473907e-01 -6.02155328e-01
1.37259275e-01 4.64775026e-01 -8.93803596e-01 -7.62658775e-01
-2.73167044e-01 -5.59385717e-01 4.89612490e-01 3.25345844e-01
4.79095161e-01 -1.06854391e+00 -6.92095280e-01 3.64246927e-02
-1.28647134e-01 -6.77784204e-01 -6.01313949e-01 2.30850637e-01
-4.70408440e-01 -1.19399738e+00 -6.91657424e-01 -7.52013445e-01
4.91431057e-01 3.94163251e-01 1.14180005e+00 2.00973004e-01
-5.91488540e-01 4.32445586e-01 -4.64286238e-01 -4.56086904e-01
3.62177454e-02 1.34588957e-01 -1.37249921e-02 1.50844440e-01
5.77953219e-01 -4.47300613e-01 -9.59257901e-01 4.76134419e-01
-1.00358737e+00 -4.29947346e-01 4.06203449e-01 1.03780675e+00
7.21334457e-01 -1.86250523e-01 7.05941260e-01 -3.37090135e-01
5.97303450e-01 -4.85649198e-01 -5.02657413e-01 3.99496377e-01
-7.94392705e-01 2.15016231e-01 4.25442249e-01 -5.72857201e-01
-5.49376130e-01 -1.60573199e-01 2.05158129e-01 -6.67062938e-01
4.17868674e-01 5.68975747e-01 4.40294072e-02 -1.07428379e-01
8.23292375e-01 2.22672388e-01 1.60209894e-01 -3.34512711e-01
2.73488522e-01 6.79256082e-01 2.14691162e-01 -4.29771781e-01
7.51547277e-01 4.76192027e-01 -4.55935374e-02 -5.17311692e-01
-9.44039285e-01 -5.51628590e-01 -3.47929716e-01 -1.08983837e-01
6.32289827e-01 -6.25567913e-01 -6.60644710e-01 2.04061702e-01
-7.37586558e-01 -9.64560211e-02 -5.72827101e-01 4.81587321e-01
-5.24248719e-01 6.67067021e-02 -3.03341210e-01 -6.39746189e-01
-7.28616416e-01 -1.02498150e+00 9.38266873e-01 4.87310998e-02
4.74096611e-02 -7.68461168e-01 1.92505762e-01 -3.97659875e-02
7.64906824e-01 1.71363816e-01 1.10107136e+00 -9.98478174e-01
-3.28352541e-01 -4.49738264e-01 -3.81279290e-01 6.25815690e-01
4.03843224e-02 -6.85575679e-02 -9.54594016e-01 -3.84746641e-01
8.99552107e-02 -8.39856386e-01 1.18411672e+00 2.62142718e-01
1.27598095e+00 -2.90688008e-01 -1.26524940e-01 6.11551106e-01
1.46710241e+00 9.02290568e-02 5.32827079e-01 5.14470100e-01
3.45265925e-01 6.41253829e-01 6.34329736e-01 3.68192106e-01
1.78925917e-01 7.56574750e-01 5.39603114e-01 -7.14549348e-02
-1.21033192e-01 -7.21344203e-02 1.94830030e-01 4.78083611e-01
3.33409101e-01 -1.64754510e-01 -7.01925039e-01 6.13087475e-01
-1.86088383e+00 -1.01444399e+00 5.42959750e-01 2.42937732e+00
9.94506717e-01 -4.90867496e-02 -8.57235938e-02 2.44895980e-01
4.12755638e-01 1.31527126e-01 -5.93413770e-01 -1.11480013e-01
-1.22104466e-01 2.42680371e-01 1.68068007e-01 2.76079953e-01
-1.18381941e+00 6.08255923e-01 5.71792221e+00 1.05403316e+00
-1.21573317e+00 -3.62881459e-03 6.49684310e-01 -5.65239549e-01
-3.74402523e-01 -3.16973209e-01 -6.24262154e-01 3.75590503e-01
7.60353684e-01 -2.90865779e-01 3.18036765e-01 8.97621810e-01
-1.47058889e-01 2.35245794e-01 -1.40993571e+00 1.03812850e+00
1.81548968e-01 -1.31704938e+00 3.15650314e-01 -2.83825118e-02
5.22275746e-01 2.91236669e-01 3.08392465e-01 3.83704960e-01
6.76584020e-02 -1.17242432e+00 8.05834949e-01 6.48583353e-01
7.82474697e-01 -8.88884902e-01 5.80666065e-01 2.37844557e-01
-8.57809365e-01 -3.21979016e-01 -5.60046971e-01 3.56153131e-01
-2.20057487e-01 5.79543114e-01 -3.85393888e-01 3.50394100e-01
7.25705862e-01 7.03669846e-01 -5.90071797e-01 1.13088226e+00
2.14200467e-01 2.69266546e-01 -2.96734393e-01 -1.28984988e-01
4.22522485e-01 -2.66800761e-01 3.49412113e-01 9.42768335e-01
2.90551215e-01 -3.13136101e-01 1.61246059e-03 8.72991741e-01
-4.17860270e-01 3.24405909e-01 -7.17152834e-01 2.30216654e-04
4.60597456e-01 1.21311772e+00 -2.89234728e-01 -1.69400945e-02
-2.19247282e-01 7.12180734e-01 4.86322463e-01 2.22637981e-01
-7.97695041e-01 -7.21612275e-01 6.52964532e-01 -3.99710238e-02
3.52359474e-01 9.87453386e-02 -6.54588193e-02 -8.84461582e-01
2.18094230e-01 -7.39820004e-01 4.61525887e-01 -4.05191600e-01
-1.31222844e+00 7.57584870e-01 -1.10952705e-01 -1.26493800e+00
-2.48973668e-01 -5.31528175e-01 -3.38086039e-01 5.51817417e-01
-1.91610050e+00 -9.98676121e-01 -3.50309819e-01 8.21743429e-01
1.99433953e-01 -3.56086254e-01 9.17795837e-01 6.05141759e-01
-2.90286511e-01 1.01708400e+00 3.33478123e-01 2.47371539e-01
7.81448901e-01 -1.12598026e+00 -2.36723036e-01 2.96724230e-01
4.85271156e-01 4.93403375e-01 3.15970838e-01 -8.74626487e-02
-1.12273920e+00 -9.81342971e-01 8.29655349e-01 -2.33234897e-01
5.62781394e-01 -4.65946913e-01 -7.91480303e-01 1.44006878e-01
-1.54350117e-01 2.51626730e-01 7.44953394e-01 -7.81247243e-02
-7.86462605e-01 -5.38253129e-01 -1.20342493e+00 3.00745696e-01
7.89147973e-01 -5.71872532e-01 -4.19608742e-01 4.11933392e-01
6.52528822e-01 1.72611654e-01 -8.49167705e-01 6.23832583e-01
8.54564190e-01 -9.27913308e-01 1.11926639e+00 -6.90417826e-01
5.33322930e-01 2.48076860e-02 -5.61185181e-01 -1.05274415e+00
-1.76301301e-01 -3.06513995e-01 -1.63475946e-01 1.10955918e+00
4.33237284e-01 -3.47182393e-01 6.83645964e-01 5.11615992e-01
7.69845769e-02 -1.17528737e+00 -1.00836706e+00 -9.13397551e-01
2.70744234e-01 -1.40745670e-01 4.80636597e-01 8.22691917e-01
-1.87746391e-01 2.70536453e-01 -2.80543298e-01 -4.29170042e-01
5.61650276e-01 1.29891768e-01 3.60212266e-01 -1.38513064e+00
-2.62410790e-01 -7.92558253e-01 -5.83439827e-01 -6.65085495e-01
8.29166472e-02 -1.04256082e+00 8.16495046e-02 -1.21118605e+00
4.41184223e-01 -7.08719909e-01 -9.51540709e-01 5.07316828e-01
3.22022587e-02 4.50466543e-01 1.15531392e-01 3.68537992e-01
-6.89262629e-01 9.32584584e-01 8.83859038e-01 -3.97420049e-01
-6.39660805e-02 -1.26763582e-01 -6.60688043e-01 5.39893091e-01
6.10885143e-01 -5.48348367e-01 -6.15904987e-01 -4.09023941e-01
4.01226252e-01 -3.91120613e-01 3.26354355e-01 -8.63584757e-01
3.80130678e-01 1.14258334e-01 2.39388347e-01 -3.02113950e-01
3.63686949e-01 -8.34090710e-01 -3.71161431e-01 4.07685399e-01
-6.98133290e-01 -7.61942342e-02 -2.74442852e-01 5.86922646e-01
-3.98589998e-01 -3.81666124e-01 9.20378923e-01 1.21961258e-01
-4.05686945e-01 5.95874608e-01 2.37016797e-01 1.32225513e-01
9.98227894e-01 -1.45185918e-01 -1.27128735e-01 -3.02571416e-01
-2.86507845e-01 2.80810416e-01 1.50586963e-01 5.68854690e-01
7.83215165e-01 -1.51172292e+00 -7.28856385e-01 8.60938653e-02
3.99620563e-01 -1.86964780e-01 -1.20152988e-01 6.13124311e-01
-2.01692939e-01 4.20886278e-01 -8.26468468e-02 -6.24147534e-01
-9.78583694e-01 5.33787549e-01 4.91514415e-01 -4.80369717e-01
-3.28469217e-01 8.53928983e-01 1.64614677e-01 -3.06170702e-01
6.79800570e-01 1.06720448e-01 -2.87882179e-01 3.63684922e-01
7.17218578e-01 3.38188410e-01 2.95543879e-01 -3.75412971e-01
-3.20618868e-01 4.28511053e-01 -3.60919803e-01 2.23170910e-02
1.44944179e+00 7.26002306e-02 -6.50500953e-02 2.78712928e-01
1.73846233e+00 -4.99931067e-01 -9.27489817e-01 -5.73738098e-01
2.18907952e-01 -3.39812100e-01 2.92617649e-01 -6.71749175e-01
-1.13751578e+00 7.51145124e-01 8.96264613e-01 7.82893151e-02
1.10501575e+00 -8.49301964e-02 7.88613617e-01 6.85909569e-01
3.44310939e-01 -1.21250296e+00 5.19531846e-01 5.13192594e-01
8.72150004e-01 -1.34602785e+00 -1.65254980e-01 2.34992206e-01
-3.53040278e-01 7.95662642e-01 4.80886340e-01 -2.96845883e-01
8.24555457e-01 4.38755602e-02 -1.57970011e-01 -3.61565024e-01
-6.88820541e-01 -1.23977683e-01 6.73228800e-01 3.21198851e-01
2.23239020e-01 -3.44905108e-01 -4.45986331e-01 6.10493183e-01
-2.40409784e-02 -1.70757726e-01 -7.35216811e-02 7.68625796e-01
-5.77304661e-01 -9.82252598e-01 1.93407327e-01 5.49452007e-01
-5.26569247e-01 -2.93984145e-01 -4.79298443e-01 6.97351158e-01
-2.01292232e-01 8.25085282e-01 1.90174133e-01 -3.55499774e-01
1.63007259e-01 -7.62589276e-02 2.96426833e-01 -4.57880646e-01
-5.09055197e-01 -3.44237357e-01 -3.88485223e-01 -6.56548440e-01
-3.54386091e-01 -1.82639927e-01 -9.72488701e-01 -1.00065377e-02
-5.97997844e-01 3.23245078e-01 7.63282180e-01 8.43853593e-01
3.93946379e-01 1.76830590e-01 1.05344617e+00 -5.49921989e-01
-1.04403996e+00 -8.58942151e-01 -5.69760978e-01 5.88655412e-01
3.66264433e-01 -7.30205059e-01 -4.86380130e-01 -3.33933353e-01] | [9.522726058959961, 3.031306505203247] |
5224c0e0-6f62-44d3-b1ff-9115aca7163b | synchronous-speech-recognition-and-speech-to | 1912.07240 | null | https://arxiv.org/abs/1912.07240v1 | https://arxiv.org/pdf/1912.07240v1.pdf | Synchronous Speech Recognition and Speech-to-Text Translation with Interactive Decoding | Speech-to-text translation (ST), which translates source language speech into target language text, has attracted intensive attention in recent years. Compared to the traditional pipeline system, the end-to-end ST model has potential benefits of lower latency, smaller model size, and less error propagation. However, it is notoriously difficult to implement such a model without transcriptions as intermediate. Existing works generally apply multi-task learning to improve translation quality by jointly training end-to-end ST along with automatic speech recognition (ASR). However, different tasks in this method cannot utilize information from each other, which limits the improvement. Other works propose a two-stage model where the second model can use the hidden state from the first one, but its cascade manner greatly affects the efficiency of training and inference process. In this paper, we propose a novel interactive attention mechanism which enables ASR and ST to perform synchronously and interactively in a single model. Specifically, the generation of transcriptions and translations not only relies on its previous outputs but also the outputs predicted in the other task. Experiments on TED speech translation corpora have shown that our proposed model can outperform strong baselines on the quality of speech translation and achieve better speech recognition performances as well. | ['Cheng-qing Zong', 'Zhongjun He', 'Jiajun Zhang', 'Hua Wu', 'Yuchen Liu', 'Long Zhou', 'Hao Xiong', 'Haifeng Wang'] | 2019-12-16 | null | null | null | null | ['speech-to-text-translation'] | ['natural-language-processing'] | [ 2.15075105e-01 3.62930931e-02 -1.29184201e-01 -4.36032057e-01
-1.33381569e+00 -4.19751704e-01 5.38755000e-01 -4.33490604e-01
-2.45855063e-01 5.42441487e-01 4.17921633e-01 -6.79912329e-01
7.92063236e-01 -3.74261647e-01 -7.61565208e-01 -5.80099583e-01
6.28123164e-01 5.10154366e-01 2.67196685e-01 -1.92712590e-01
-2.35367373e-01 6.52209520e-02 -7.44625628e-01 6.96962476e-01
9.85305607e-01 7.28955150e-01 5.79000354e-01 6.39740348e-01
-2.93256521e-01 7.18965173e-01 -8.06804061e-01 -5.15981019e-01
-6.90501854e-02 -6.69169009e-01 -6.97422683e-01 8.82262737e-03
-6.82409704e-02 -4.59396034e-01 -4.66765195e-01 8.49014521e-01
9.15951908e-01 -2.53516492e-02 2.63903409e-01 -9.36425269e-01
-8.36030185e-01 1.00828242e+00 -3.79725426e-01 1.84166789e-01
2.28911445e-01 3.08317512e-01 9.06247199e-01 -1.18805802e+00
1.47148311e-01 1.47157586e+00 2.09631324e-01 6.97763801e-01
-9.94916797e-01 -8.65594089e-01 2.47764319e-01 7.50818029e-02
-1.29326737e+00 -1.06227779e+00 4.82751429e-01 -7.28711933e-02
1.37415004e+00 1.25655979e-01 2.21116424e-01 1.34550512e+00
1.00838982e-01 1.17094409e+00 8.62203121e-01 -3.74369800e-01
-1.59798294e-01 1.28304869e-01 -3.62282664e-01 4.83118206e-01
-4.07113433e-01 6.51349276e-02 -7.21304417e-01 1.35271266e-01
6.62919581e-01 -9.54336151e-02 -2.76040614e-01 5.46593845e-01
-1.58003378e+00 4.95298058e-01 1.96820289e-01 3.49553496e-01
-3.85450393e-01 5.79050928e-02 4.45574194e-01 5.65279365e-01
7.45602190e-01 -6.23877011e-02 -3.92247140e-01 -3.43168765e-01
-9.68348682e-01 -3.58400375e-01 6.28813267e-01 1.23768139e+00
4.69140559e-01 3.53539884e-01 -4.99881893e-01 8.77710879e-01
4.78808999e-01 7.78329313e-01 7.79557407e-01 -2.99609512e-01
1.23354781e+00 2.59435356e-01 -4.52135131e-02 -4.03771251e-01
1.38672635e-01 -6.16367280e-01 -9.15680766e-01 -4.20511663e-01
-4.87925299e-02 -4.88980234e-01 -9.98391628e-01 1.56648946e+00
1.39854223e-01 2.74161041e-01 3.75030428e-01 9.77283955e-01
7.99785256e-01 1.25897801e+00 -6.97067901e-02 -4.13259774e-01
8.94049942e-01 -1.61682773e+00 -1.00623894e+00 -5.03602982e-01
5.87780356e-01 -1.16879022e+00 1.12711906e+00 3.80263105e-02
-1.38208759e+00 -7.85095453e-01 -8.85748148e-01 -2.33966336e-01
1.65154010e-01 5.96078992e-01 1.16401456e-01 3.36067915e-01
-1.18231368e+00 2.49815926e-01 -1.14651883e+00 -1.98750630e-01
1.42512724e-01 4.37724233e-01 6.86542019e-02 1.83735296e-01
-1.35524929e+00 9.36939418e-01 5.25158904e-02 3.70646954e-01
-1.09204149e+00 -6.01423830e-02 -6.40034378e-01 2.73330927e-01
2.32178777e-01 -6.64960682e-01 1.83882034e+00 -1.14137256e+00
-2.28497148e+00 2.73488164e-01 -8.01372528e-01 -3.12655091e-01
5.24005413e-01 -2.98531294e-01 -4.96206135e-01 -1.34772822e-01
-8.58773105e-03 4.68284398e-01 8.23555589e-01 -7.91258335e-01
-6.66923642e-01 -1.67037591e-01 -2.07867220e-01 5.30598760e-01
-3.55395675e-01 4.25244689e-01 -7.15255618e-01 -5.57271957e-01
1.27916008e-01 -8.10470819e-01 -1.98024705e-01 -4.58087713e-01
-4.46407884e-01 -3.89148504e-01 6.62215948e-01 -9.12234366e-01
1.40569437e+00 -2.19741893e+00 3.02946746e-01 -2.42693871e-01
-2.01439574e-01 5.52627265e-01 -2.24096075e-01 6.67924821e-01
1.37663811e-01 1.64823413e-01 -3.23836543e-02 -7.39010990e-01
-3.16574946e-02 7.54550844e-02 -5.88595331e-01 5.45371957e-02
4.14848477e-01 1.11330891e+00 -7.26107121e-01 -4.45882887e-01
7.37206861e-02 5.31354785e-01 -2.17154786e-01 6.56015873e-01
-1.67090341e-01 6.82192922e-01 -4.58424091e-01 3.88343096e-01
2.86549360e-01 -2.53263742e-01 6.61460236e-02 1.03154838e-01
-1.90674931e-01 1.07033324e+00 -6.44272983e-01 1.68431664e+00
-8.77633452e-01 4.93555844e-01 1.85465198e-02 -9.32378113e-01
1.07300889e+00 9.82824445e-01 -7.73037970e-02 -8.30893040e-01
-1.19632296e-02 3.94488543e-01 1.51499823e-01 -4.78695363e-01
1.51474446e-01 -1.43015116e-01 5.94358072e-02 5.84707677e-01
-1.28269857e-02 4.51690480e-02 -2.10545808e-01 -1.80730615e-02
7.82125354e-01 2.19935775e-01 3.89675647e-02 3.93822551e-01
5.46700656e-01 -7.62660354e-02 7.03258455e-01 3.54995847e-01
-1.25049129e-02 6.50848031e-01 1.97521210e-01 -4.86124828e-02
-1.04221630e+00 -8.94201934e-01 5.34519136e-01 1.23017681e+00
4.58642803e-02 -3.20946544e-01 -8.61365914e-01 -7.21082926e-01
-5.10136425e-01 7.39218175e-01 3.83681841e-02 -3.06821674e-01
-9.15262818e-01 -5.40959537e-01 8.03111196e-01 5.23831248e-01
5.48160374e-01 -1.14187741e+00 9.58498344e-02 4.32823151e-01
-7.64957130e-01 -1.34465468e+00 -1.01572847e+00 -5.06702960e-02
-8.96154583e-01 -1.15954705e-01 -8.94251943e-01 -1.08838463e+00
5.65228224e-01 4.96387810e-01 8.73172581e-01 4.86267023e-02
6.48173928e-01 -4.07967478e-01 -4.47323442e-01 -2.46943846e-01
-8.64427149e-01 4.66367513e-01 3.75776622e-03 3.21198076e-01
2.83421516e-01 -4.45618987e-01 -3.43932807e-01 5.37549257e-01
-5.67830682e-01 6.48244858e-01 1.17575598e+00 8.72199476e-01
4.60903615e-01 -4.49704289e-01 8.78784597e-01 -5.06237090e-01
6.40164852e-01 -3.56425583e-01 -3.82344604e-01 4.91595358e-01
-4.69746947e-01 7.91018009e-02 9.78147149e-01 -4.28089321e-01
-1.30997443e+00 1.17381878e-01 -5.22124410e-01 -5.05287766e-01
1.66050028e-02 6.60656989e-01 -4.54087377e-01 4.70138818e-01
2.51702249e-01 7.11008012e-01 -7.14861155e-02 -5.19440532e-01
1.24575824e-01 1.37207162e+00 2.95293212e-01 -2.02470765e-01
7.58243501e-01 -1.34973660e-01 -7.37187862e-01 -5.96135795e-01
-7.10586250e-01 -3.39576155e-01 -5.69026887e-01 7.23891407e-02
7.72458553e-01 -1.25774860e+00 -4.80411649e-01 3.94579738e-01
-1.73569906e+00 -3.96194756e-01 3.58628154e-01 7.67986298e-01
-4.17359829e-01 1.57794297e-01 -8.42838526e-01 -8.18107009e-01
-7.71899879e-01 -1.51737165e+00 1.27279353e+00 -2.35072784e-02
6.37526587e-02 -7.22117901e-01 -2.36426383e-01 5.94332933e-01
5.72275758e-01 -6.81714058e-01 6.22237504e-01 -6.93423629e-01
-7.45026827e-01 2.66630333e-02 -1.10657789e-01 5.06238043e-01
1.45722911e-01 -6.87567890e-02 -1.00906491e+00 -3.34447175e-01
9.48897377e-02 -2.25530028e-01 6.60538852e-01 6.96518198e-02
8.15099120e-01 -6.14681125e-01 -2.24967584e-01 4.90410298e-01
9.28007841e-01 4.93071616e-01 5.64173400e-01 -9.87322405e-02
7.74368584e-01 4.61291045e-01 4.89231259e-01 -8.28721970e-02
5.08180618e-01 9.49454308e-01 -2.84087770e-02 -3.56024384e-01
-3.08956832e-01 -4.77596045e-01 1.25467706e+00 1.79631400e+00
2.42898479e-01 -5.07336795e-01 -8.33554924e-01 4.19132471e-01
-1.98895645e+00 -8.30094516e-01 5.94394095e-02 2.11192203e+00
1.09893239e+00 2.01209545e-01 -5.71101122e-02 -2.72343338e-01
9.49550569e-01 1.00885548e-01 -4.88969028e-01 -3.76210988e-01
6.89246729e-02 1.00121692e-01 1.53034031e-01 7.40446091e-01
-6.49331808e-01 1.41020632e+00 5.97152662e+00 9.08722758e-01
-1.51754487e+00 5.79085469e-01 6.30621314e-01 -2.11546794e-01
-2.76374996e-01 7.01208711e-02 -9.95434523e-01 5.64455390e-01
1.33354568e+00 -2.72211730e-01 5.70209503e-01 5.85747004e-01
6.66733801e-01 4.40202326e-01 -1.08526373e+00 8.86770964e-01
2.92235217e-03 -1.01515365e+00 3.19043994e-01 -2.18068004e-01
5.22126079e-01 3.15598935e-01 -4.35374565e-02 5.01119435e-01
3.16918969e-01 -8.76635134e-01 8.22372139e-01 9.91915092e-02
9.68181133e-01 -7.60234833e-01 7.41468012e-01 7.41073549e-01
-1.29908812e+00 1.34994686e-01 -2.57629156e-01 -1.25360012e-01
5.78699410e-01 3.48049551e-01 -1.20281911e+00 6.51587188e-01
1.94156259e-01 5.76499403e-01 -1.08688265e-01 6.95737600e-01
-6.55757129e-01 1.14625812e+00 -1.02326125e-01 -2.84706503e-01
3.90746385e-01 -1.55779347e-01 3.56832623e-01 1.43497956e+00
5.64642251e-01 -3.02163009e-02 3.82640064e-01 6.81906819e-01
-2.29521856e-01 2.75259912e-01 -4.40394729e-01 -2.50167489e-01
5.99277198e-01 1.01407158e+00 -2.89137661e-01 -6.66457474e-01
-7.29265630e-01 1.31265557e+00 3.54962111e-01 6.17120326e-01
-9.41934824e-01 -3.85572851e-01 5.37451386e-01 -4.39629368e-02
2.33092755e-01 -4.40976411e-01 -2.98357457e-01 -1.30773902e+00
2.87413150e-01 -1.09100223e+00 -1.68688834e-01 -7.61741996e-01
-9.08548594e-01 1.04049051e+00 -5.89894533e-01 -1.31232071e+00
-4.08577353e-01 -5.39834388e-02 -7.02884138e-01 1.42084229e+00
-1.51123071e+00 -1.29486954e+00 1.55008703e-01 4.67385948e-01
1.20065296e+00 -1.82114050e-01 7.60418773e-01 4.76066262e-01
-1.02319384e+00 8.50602329e-01 6.68063238e-02 3.39349985e-01
8.71104240e-01 -8.96532893e-01 1.07264626e+00 1.21912658e+00
2.32721925e-01 5.42307496e-01 2.45304406e-01 -7.41356075e-01
-1.25650334e+00 -1.26415420e+00 1.47513342e+00 -1.72090486e-01
4.60095644e-01 -5.83541691e-01 -8.56355906e-01 8.25776756e-01
5.58445513e-01 -3.52532297e-01 4.78640407e-01 -1.75345913e-01
-1.03702605e-01 -2.17795417e-01 -3.88289332e-01 7.56541610e-01
1.05622530e+00 -8.40390861e-01 -5.40197253e-01 3.47465813e-01
1.23603761e+00 -5.58113396e-01 -5.04765213e-01 1.91571757e-01
2.85777360e-01 -5.18382609e-01 5.58847427e-01 -5.04099548e-01
4.52650726e-01 -3.64882469e-01 1.38331100e-03 -1.59702432e+00
-2.19435021e-01 -1.05419707e+00 -4.92973290e-02 1.27800488e+00
9.65884626e-01 -6.04609668e-01 2.94901013e-01 2.28557244e-01
-7.58827567e-01 -8.13847601e-01 -8.68761778e-01 -7.79356122e-01
-2.95283031e-02 -2.30593979e-01 7.90869892e-01 6.54133558e-01
4.56624329e-02 1.12828410e+00 -5.99755526e-01 4.24738586e-01
8.90158489e-03 1.52121440e-01 7.85264313e-01 -6.37381554e-01
-4.87345815e-01 -3.58053952e-01 2.71062016e-01 -1.84796035e+00
1.43221319e-01 -1.06914473e+00 4.59459871e-01 -1.71418631e+00
5.68063781e-02 -3.08077931e-01 -4.28736135e-02 5.74629128e-01
-4.61649716e-01 -1.80371970e-01 3.39321382e-02 5.02912164e-01
-4.83458310e-01 8.52329433e-01 1.47096002e+00 -4.57123928e-02
-2.82635719e-01 2.28904694e-01 -5.30301929e-01 3.15989524e-01
7.43374467e-01 -5.33542097e-01 -4.62165922e-01 -1.24221814e+00
-1.58299148e-01 5.43629646e-01 -2.32853428e-01 -7.17204332e-01
5.49061179e-01 -7.08203912e-02 1.26702279e-01 -5.18913925e-01
3.66013527e-01 -4.56814975e-01 -1.32432170e-02 3.42264831e-01
-4.66055751e-01 2.96338052e-01 -7.76200593e-02 2.88099021e-01
-5.50620675e-01 4.80715856e-02 4.93291676e-01 -8.45064595e-02
-1.79267883e-01 4.74202633e-01 -4.85418797e-01 -2.58330703e-01
6.08270705e-01 1.21348679e-01 -1.12160213e-01 -6.08592272e-01
-6.24700367e-01 2.30409756e-01 1.93981044e-02 7.60125816e-01
6.36213422e-01 -1.37928295e+00 -1.08181238e+00 3.08040142e-01
-1.50012225e-01 2.12538481e-01 -2.92728208e-02 9.92231965e-01
-9.34298187e-02 5.48292339e-01 2.86535263e-01 -6.03989124e-01
-1.13716412e+00 2.93381661e-01 3.41853648e-01 -3.09565187e-01
-3.32442820e-01 8.38377655e-01 1.90371931e-01 -4.95014787e-01
3.55314523e-01 -2.19759524e-01 1.33743346e-01 -3.19677383e-01
6.35325074e-01 6.21953122e-02 2.41139233e-01 -7.25659192e-01
-1.71568409e-01 2.17451155e-01 -3.68390113e-01 -5.92150688e-01
1.09312177e+00 -4.74253833e-01 3.65595073e-02 5.52437305e-01
1.04107189e+00 -1.25829518e-01 -1.12526631e+00 -5.78956068e-01
-5.85448481e-02 -2.56870508e-01 1.02452226e-01 -9.80117679e-01
-1.03617430e+00 1.38779163e+00 1.42622754e-01 -1.45637691e-01
1.07804251e+00 -8.71391073e-02 1.40757728e+00 3.40995878e-01
2.98305780e-01 -1.02451885e+00 1.72811732e-01 7.72919893e-01
9.41414475e-01 -1.19995093e+00 -7.28282273e-01 -4.71488655e-01
-8.36341798e-01 9.92376208e-01 7.18576491e-01 3.17476869e-01
1.67704016e-01 3.13899219e-01 4.35372502e-01 5.02899051e-01
-1.23445320e+00 -1.96215332e-01 1.80988863e-01 2.69994408e-01
7.86883354e-01 9.42541882e-02 -1.67368144e-01 7.18483210e-01
-1.65876597e-01 -3.65635045e-02 1.71021014e-01 6.12016261e-01
-3.97607386e-01 -1.34933317e+00 -1.69318095e-01 1.07324980e-01
-6.22171879e-01 -5.20279169e-01 -5.42399645e-01 7.73009211e-02
-2.53241360e-01 1.35765588e+00 -1.15573674e-01 -6.91976726e-01
4.93072212e-01 1.70707554e-01 1.97618511e-02 -9.40306544e-01
-8.50560427e-01 5.23556173e-01 2.17690125e-01 -4.47854489e-01
-2.64744516e-02 -4.97618020e-01 -1.26982367e+00 -2.43275627e-01
-5.87650061e-01 3.41066927e-01 7.25819170e-01 1.17728984e+00
6.69974625e-01 5.70342779e-01 1.07184470e+00 -5.94027281e-01
-8.32075059e-01 -1.40752816e+00 1.38657168e-01 -5.71365356e-02
4.79641110e-01 -8.35372135e-02 -1.37526527e-01 1.50112137e-01] | [14.473341941833496, 7.097713947296143] |
9b27faee-3213-4ca8-adeb-90c08465ebeb | ease-entity-aware-contrastive-learning-of-1 | 2205.04260 | null | https://arxiv.org/abs/2205.04260v1 | https://arxiv.org/pdf/2205.04260v1.pdf | EASE: Entity-Aware Contrastive Learning of Sentence Embedding | We present EASE, a novel method for learning sentence embeddings via contrastive learning between sentences and their related entities. The advantage of using entity supervision is twofold: (1) entities have been shown to be a strong indicator of text semantics and thus should provide rich training signals for sentence embeddings; (2) entities are defined independently of languages and thus offer useful cross-lingual alignment supervision. We evaluate EASE against other unsupervised models both in monolingual and multilingual settings. We show that EASE exhibits competitive or better performance in English semantic textual similarity (STS) and short text clustering (STC) tasks and it significantly outperforms baseline methods in multilingual settings on a variety of tasks. Our source code, pre-trained models, and newly constructed multilingual STC dataset are available at https://github.com/studio-ousia/ease. | ['Isao Echizen', 'Yoshimasa Tsuruoka', 'Ikuya Yamada', 'Ryokan Ri', 'Sosuke Nishikawa'] | 2022-05-09 | null | https://aclanthology.org/2022.naacl-main.284 | https://aclanthology.org/2022.naacl-main.284.pdf | naacl-2022-7 | ['text-clustering', 'short-text-clustering'] | ['natural-language-processing', 'natural-language-processing'] | [-2.75445551e-01 -4.63050976e-02 -3.69408756e-01 -6.60797238e-01
-9.57317948e-01 -6.64338708e-01 9.26050127e-01 8.73555601e-01
-9.82421935e-01 5.63650191e-01 8.39599609e-01 -2.09390596e-01
4.74408157e-02 -3.36120576e-01 -6.27401769e-01 -2.02411562e-01
7.75897084e-03 6.35506988e-01 -2.06280667e-02 -3.16726536e-01
4.51795422e-02 -4.89230156e-02 -1.05582571e+00 1.19841360e-01
1.29737306e+00 4.18668211e-01 3.85144472e-01 3.95415306e-01
-2.38018394e-01 5.72189271e-01 -3.31620038e-01 -8.55638027e-01
-8.86712745e-02 -1.01076126e-01 -1.06807554e+00 -3.90713662e-01
5.09864867e-01 2.70783633e-01 -3.42606962e-01 9.61016715e-01
6.71225369e-01 7.35585839e-02 7.63142049e-01 -1.02686822e+00
-1.03045464e+00 1.10235631e+00 -2.33010560e-01 3.88608605e-01
4.13218826e-01 -3.43887895e-01 1.59444630e+00 -1.25544083e+00
1.05939937e+00 9.22878206e-01 8.29956472e-01 3.50270122e-01
-1.25251758e+00 -3.89580429e-01 -1.40511766e-01 2.38019347e-01
-1.36376095e+00 -7.87782669e-01 4.48787034e-01 -3.78012031e-01
1.32942843e+00 1.23505011e-01 2.42557704e-01 1.34420705e+00
-8.89604259e-03 9.84867990e-01 9.31702971e-01 -6.50837064e-01
-1.31911486e-01 6.51329935e-01 2.81508446e-01 4.93352473e-01
2.75148183e-01 -4.75651264e-01 -5.21205962e-01 -2.28888616e-02
1.21822909e-01 -2.62605816e-01 -3.45678300e-01 -3.07636470e-01
-1.43184566e+00 9.40074682e-01 4.45285231e-01 7.99508631e-01
-1.76149592e-01 4.03858162e-02 8.58259261e-01 3.66526574e-01
7.09403396e-01 7.53995538e-01 -7.03154743e-01 -1.91595346e-01
-5.39262712e-01 -1.00867137e-01 7.82205939e-01 9.13376927e-01
5.89028537e-01 -8.36619884e-02 8.84054154e-02 1.42331791e+00
1.50219351e-01 6.53904021e-01 7.94883490e-01 -6.26565516e-01
8.21688533e-01 4.22477901e-01 -1.10661320e-01 -9.61614490e-01
-3.48340571e-01 -4.07722116e-01 -4.75342065e-01 -5.92521071e-01
1.49467841e-01 -6.95369840e-02 -2.34691814e-01 1.84550333e+00
1.36324435e-01 -8.92466158e-02 3.53251994e-01 5.45100391e-01
1.13444316e+00 4.95332122e-01 7.48644471e-02 5.31722829e-02
1.51591575e+00 -9.67438459e-01 -7.56270468e-01 -4.14019436e-01
1.12631929e+00 -1.05550539e+00 1.20501268e+00 -4.17136133e-01
-7.57898629e-01 -4.69202846e-01 -7.73070037e-01 -4.34407890e-01
-7.68265843e-01 3.98355067e-01 5.30916691e-01 4.14428413e-01
-9.95629609e-01 4.15389746e-01 -7.33621955e-01 -8.63955557e-01
2.03079090e-01 9.17292759e-02 -7.09991217e-01 3.95415239e-02
-1.51319897e+00 1.19625378e+00 3.97988409e-01 -3.39281708e-01
-1.75444007e-01 -6.69973791e-01 -1.38763142e+00 8.84845704e-02
3.38759460e-02 -4.75132376e-01 1.04011047e+00 -7.88049877e-01
-9.90931392e-01 1.36742115e+00 -3.11886609e-01 -5.94534159e-01
7.21046478e-02 -4.09139782e-01 -6.02061987e-01 1.78979650e-01
6.60973072e-01 6.24330878e-01 2.06710726e-01 -9.25995767e-01
-3.69488537e-01 -2.56223470e-01 -1.41813174e-01 4.99056846e-01
-8.01660478e-01 2.54366308e-01 -3.84807557e-01 -7.86358714e-01
-3.04853559e-01 -8.20886374e-01 -1.83311030e-02 -3.73622358e-01
-3.24546784e-01 -5.49103498e-01 4.66063350e-01 -8.97397339e-01
1.32655036e+00 -2.04317069e+00 -2.94688698e-02 -2.98962295e-01
2.00153627e-02 3.14595550e-01 -1.07435904e-01 9.46030378e-01
-2.18503192e-01 2.93650120e-01 -2.52938509e-01 -6.63232207e-01
2.64189750e-01 1.65824339e-01 -1.13392822e-01 3.58280331e-01
3.36274683e-01 1.22154713e+00 -1.17780995e+00 -7.21011698e-01
1.59360498e-01 2.90633678e-01 -2.61871964e-01 -1.45369962e-01
1.61214039e-01 2.07701340e-01 -1.96777478e-01 2.70816356e-01
3.23579639e-01 -3.67120892e-01 5.38467646e-01 -1.35244727e-01
-4.29399824e-03 8.31574380e-01 -8.98378015e-01 1.93266213e+00
-7.24477112e-01 9.71938312e-01 -2.94151574e-01 -9.93863583e-01
7.05806613e-01 5.18161714e-01 3.19391787e-01 -6.55015945e-01
1.20386839e-01 4.04747725e-01 -1.38647497e-01 -6.13611996e-01
7.34272540e-01 9.43369493e-02 -4.75360304e-01 6.19914532e-01
4.44907099e-01 -1.55455302e-02 4.63119686e-01 5.36413848e-01
9.75022614e-01 -7.11413100e-02 5.57320714e-01 -6.34753287e-01
4.88588363e-01 2.90821046e-02 6.12117350e-01 2.66570747e-01
-1.31124496e-01 2.28211939e-01 3.89073253e-01 -1.23013332e-01
-1.18711567e+00 -1.26328599e+00 -6.78768754e-01 1.05946004e+00
4.29906286e-02 -8.86157870e-01 -3.74871433e-01 -8.04803014e-01
1.54679939e-01 8.76716077e-01 -5.00985861e-01 1.77268967e-01
-5.58740139e-01 -5.04372597e-01 6.09157801e-01 7.33238637e-01
1.26279652e-01 -8.32374990e-01 7.59106278e-02 5.09175658e-02
-5.67243099e-01 -1.46617210e+00 -8.03513169e-01 1.31510511e-01
-5.68316638e-01 -1.07347882e+00 -4.45842713e-01 -1.25868428e+00
5.56721330e-01 1.82401747e-01 1.37946486e+00 -2.08762288e-01
-1.07231140e-01 5.68397582e-01 -4.87767547e-01 -1.62476242e-01
-4.75769997e-01 4.14634526e-01 4.45130795e-01 -2.98805445e-01
6.48949027e-01 -4.02362436e-01 -3.08684081e-01 -4.27600369e-02
-7.67197907e-01 -1.94046989e-01 3.05709064e-01 1.05939305e+00
3.43312711e-01 -2.95381755e-01 6.84683859e-01 -1.17281187e+00
7.90385783e-01 -5.77104509e-01 -3.30521971e-01 4.69888628e-01
-5.70986450e-01 2.27893308e-01 6.45446599e-01 6.38571288e-03
-9.24396753e-01 -3.00336599e-01 -1.85104325e-01 -6.75230324e-02
-1.83739498e-01 7.85666823e-01 6.51433095e-02 5.15524745e-01
7.04242945e-01 1.05638346e-02 -1.96257144e-01 -4.56112742e-01
5.27935445e-01 1.01966512e+00 3.21722448e-01 -7.61918187e-01
7.50831425e-01 2.22853392e-01 -5.35620630e-01 -1.00898063e+00
-7.68502951e-01 -8.30142856e-01 -9.40468788e-01 8.44745114e-02
1.11273062e+00 -1.14501750e+00 -2.38797709e-01 4.05726470e-02
-1.10230374e+00 -1.87214047e-01 -9.28939432e-02 7.86475599e-01
-1.20332785e-01 5.22420168e-01 -8.46450806e-01 -2.91991144e-01
-4.05986398e-01 -8.41331661e-01 1.08088565e+00 1.87976286e-02
-5.49823880e-01 -1.71897972e+00 2.95759916e-01 4.13325965e-01
1.90628693e-01 7.76499584e-02 8.68317425e-01 -1.15713990e+00
-6.10122532e-02 -1.89058289e-01 5.60821332e-02 4.47815299e-01
3.36156070e-01 1.13763884e-01 -8.21600974e-01 -4.06664431e-01
-3.57400149e-01 -5.42627811e-01 7.06131876e-01 7.89482370e-02
4.81944203e-01 -2.07663938e-01 -3.29268754e-01 4.46333736e-01
1.41030812e+00 -4.07364905e-01 1.79156110e-01 5.40131152e-01
5.67471564e-01 6.16652071e-01 4.49674815e-01 1.33162826e-01
8.11493516e-01 8.35796952e-01 -2.53750056e-01 -8.77948180e-02
-5.48549742e-02 -3.25910270e-01 5.01978695e-01 1.73371363e+00
3.74092132e-01 -1.25670046e-01 -1.12948287e+00 9.39937472e-01
-1.85975337e+00 -1.08724630e+00 -3.30431372e-01 2.02392983e+00
1.10641670e+00 -5.48507757e-02 3.24663632e-02 -2.96971709e-01
6.56813145e-01 1.92772374e-01 -5.85831180e-02 -4.27020311e-01
-5.17549396e-01 4.43456262e-01 5.41060328e-01 5.59408367e-01
-1.30730784e+00 1.19846189e+00 5.70774698e+00 7.83281147e-01
-8.16471517e-01 4.78373766e-01 2.55506307e-01 1.53375447e-01
-5.65732300e-01 3.23024429e-02 -8.28656971e-01 4.63086784e-01
9.79660273e-01 -6.04790568e-01 -1.07392848e-01 6.98096812e-01
-7.29924366e-02 1.24270625e-01 -1.15921426e+00 7.20529377e-01
1.60708532e-01 -1.34870279e+00 -2.81046212e-01 -2.17975438e-01
8.64182413e-01 6.40739858e-01 -7.49602821e-03 4.14655358e-01
5.83270013e-01 -7.59683788e-01 5.00028551e-01 -5.43956719e-02
8.95731986e-01 -5.30501306e-01 9.58729446e-01 2.09991038e-02
-1.26967049e+00 3.19021165e-01 -5.27294397e-01 3.63608599e-01
2.47978479e-01 5.18777966e-01 -8.27678442e-01 8.57087314e-01
7.16895640e-01 1.33147883e+00 -9.70151424e-01 7.14271069e-01
-5.99031031e-01 5.73523283e-01 -2.59420604e-01 -7.07262084e-02
3.10064524e-01 -1.98575333e-01 5.89699149e-01 1.68623459e+00
1.60444640e-02 -5.67911267e-01 -1.87522657e-02 6.63618684e-01
-3.90988857e-01 6.69062376e-01 -7.24005103e-01 -3.91902536e-01
8.61206770e-01 1.28385043e+00 -6.29403412e-01 -3.63166153e-01
-6.85346246e-01 1.13884389e+00 7.38940537e-01 2.42224723e-01
-7.69848526e-01 -7.61421144e-01 7.42733061e-01 -3.11077386e-01
3.65455329e-01 -2.91165531e-01 -1.65218577e-01 -1.68263948e+00
2.11581945e-01 -6.40302658e-01 5.77596903e-01 -5.95426440e-01
-1.59649026e+00 6.60251737e-01 -2.75377512e-01 -1.19285250e+00
-3.05094421e-01 -7.40271032e-01 -5.82558692e-01 7.79676318e-01
-1.45590043e+00 -1.16081882e+00 1.29579231e-01 5.06755650e-01
6.03972256e-01 -3.44458252e-01 1.12775373e+00 4.16096300e-01
-6.30766869e-01 7.99302876e-01 7.12773919e-01 5.90001941e-01
1.15991688e+00 -1.53442001e+00 5.85610926e-01 8.59185040e-01
7.68164814e-01 9.35047030e-01 4.49880391e-01 -5.40849030e-01
-9.55727398e-01 -1.05799317e+00 1.72350860e+00 -8.73043537e-01
1.18165600e+00 -7.36577511e-01 -8.31581652e-01 9.51523244e-01
7.79355645e-01 -3.22191775e-01 1.22904217e+00 7.76487768e-01
-6.32109046e-01 -3.19663584e-02 -6.97546005e-01 7.69875288e-01
8.26110303e-01 -9.93382394e-01 -9.11042333e-01 6.73452377e-01
8.43311548e-01 -1.86411664e-01 -1.34776652e+00 7.74061754e-02
2.35056996e-01 -6.65387988e-01 8.72331023e-01 -4.98067737e-01
5.88938296e-01 -1.75537933e-02 -3.10563058e-01 -1.44247282e+00
-1.86472252e-01 -2.39673257e-01 2.82809317e-01 1.58176219e+00
8.34464431e-01 -8.75596642e-01 2.25151867e-01 1.96435690e-01
-3.29875857e-01 -6.47107422e-01 -7.91908860e-01 -1.05245125e+00
3.41089129e-01 -3.74400973e-01 3.40701878e-01 1.69874585e+00
5.48911572e-01 8.31153929e-01 4.53624986e-02 6.11578524e-02
4.64629591e-01 -7.48374388e-02 5.18432319e-01 -1.00126231e+00
-2.69486487e-01 -4.84191120e-01 -4.29543823e-01 -8.39755595e-01
7.14817166e-01 -1.66854525e+00 -2.32233554e-01 -1.75337040e+00
1.85493290e-01 -4.67578679e-01 -2.61958361e-01 4.26449120e-01
-4.06678706e-01 -8.10634438e-03 4.76037338e-02 3.27813864e-01
-7.66149879e-01 6.62392795e-01 4.89265978e-01 -7.93438926e-02
2.54674405e-01 -2.92902023e-01 -6.43345654e-01 4.04777348e-01
8.84850621e-01 -5.35785556e-01 -1.37214303e-01 -7.41088808e-01
2.27793068e-01 -4.01911288e-01 -3.61669101e-02 -5.80591917e-01
1.24681279e-01 2.04080671e-01 1.14461869e-01 -2.30190456e-01
1.24295205e-01 -5.82390845e-01 -2.88294822e-01 2.62964547e-01
-4.96848792e-01 3.77672046e-01 2.07190782e-01 3.20468456e-01
-4.81708497e-01 -5.88386416e-01 4.36641872e-01 9.91904289e-02
-8.10960352e-01 2.17244262e-03 -3.50326598e-01 6.04649484e-01
6.73709631e-01 1.18905470e-01 -4.64248687e-01 -2.63816744e-01
-5.33781230e-01 5.71735144e-01 7.43180275e-01 7.45325208e-01
2.69596249e-01 -1.56322300e+00 -8.03802490e-01 -1.00238383e-01
5.60925484e-01 -4.45963889e-01 -1.09984584e-01 1.02116048e+00
-4.40731347e-01 6.51167154e-01 1.93714008e-01 -5.57960033e-01
-1.30098331e+00 3.65582883e-01 -1.77515764e-02 -2.24218965e-01
-4.63240176e-01 8.46004963e-01 2.16843504e-02 -1.12970829e+00
-2.12313933e-03 1.68155245e-02 -1.02136828e-01 8.93764049e-02
1.48073629e-01 1.68479860e-01 2.76339389e-02 -9.16409612e-01
-6.00653827e-01 4.25262004e-01 -2.33406380e-01 -6.32047653e-02
1.48642969e+00 -2.96375096e-01 -3.15213859e-01 7.99290359e-01
1.53327787e+00 3.28569412e-01 -3.24063838e-01 -5.99792659e-01
5.86467147e-01 -2.31327489e-01 -2.39919558e-01 -3.99054766e-01
-6.01191521e-01 7.76655018e-01 1.34132385e-01 1.59146860e-01
6.48427069e-01 4.05500114e-01 8.41549993e-01 4.47268605e-01
6.04731366e-02 -1.43287253e+00 4.56367806e-02 8.19328129e-01
5.56251526e-01 -1.45607603e+00 -1.57663435e-01 -2.65994102e-01
-1.01218927e+00 9.81739938e-01 4.29644883e-01 8.92268643e-02
6.52728796e-01 1.84601903e-01 2.54102618e-01 -3.76857877e-01
-7.25354195e-01 -4.23330277e-01 4.03132439e-01 5.28725028e-01
1.12363970e+00 -1.96268223e-03 -5.27771175e-01 4.92339075e-01
-4.89570498e-01 -6.08587325e-01 3.16219866e-01 7.36732483e-01
-9.99147892e-02 -1.38959467e+00 1.99464560e-01 3.70864004e-01
-5.22297323e-01 -5.28511107e-01 -3.12111706e-01 9.11886811e-01
-1.28596649e-01 7.58765042e-01 7.02203363e-02 -1.14604250e-01
1.13242358e-01 3.38892192e-01 2.94529080e-01 -8.93044293e-01
-5.60646236e-01 -1.72319591e-01 5.88436007e-01 -1.78264186e-01
-5.05120397e-01 -9.89134848e-01 -1.09482574e+00 -2.42905185e-01
-4.30395365e-01 4.98556823e-01 4.80050743e-01 7.99949408e-01
5.30892193e-01 2.26045415e-01 5.85982382e-01 -3.49738032e-01
-2.95423627e-01 -1.10770631e+00 -4.95684654e-01 8.87818813e-01
-2.04241872e-01 -4.05749559e-01 -4.49713647e-01 1.10164620e-01] | [10.903192520141602, 9.68709659576416] |
d47bb9f2-99c3-478d-bbf5-3b8df2473636 | improving-multi-label-emotion-classification | null | null | https://aclanthology.org/D18-1137 | https://aclanthology.org/D18-1137.pdf | Improving Multi-label Emotion Classification via Sentiment Classification with Dual Attention Transfer Network | In this paper, we target at improving the performance of multi-label emotion classification with the help of sentiment classification. Specifically, we propose a new transfer learning architecture to divide the sentence representation into two different feature spaces, which are expected to respectively capture the general sentiment words and the other important emotion-specific words via a dual attention mechanism. Experimental results on two benchmark datasets demonstrate the effectiveness of our proposed method. | ["Lu{\\'\\i}s Marujo", 'Jing Jiang', 'William Brendel', 'Jianfei Yu', 'Pradeep Karuturi'] | 2018-10-01 | null | null | null | emnlp-2018-10 | ['stock-market-prediction'] | ['time-series'] | [-4.09888662e-02 -2.22706214e-01 -1.43447876e-01 -6.68046474e-01
-6.25730693e-01 -7.83323199e-02 1.90763578e-01 7.55484104e-02
-4.86087531e-01 6.76837325e-01 3.05187851e-01 1.37575790e-01
1.68176532e-01 -4.13527489e-01 -2.44181260e-01 -7.57233560e-01
4.20662284e-01 -1.03443764e-01 -3.67547274e-01 -3.62876832e-01
3.46129328e-01 -5.88027313e-02 -1.28196847e+00 3.63172382e-01
8.50977242e-01 1.51678526e+00 -1.65873468e-01 2.84049481e-01
-2.23470762e-01 9.73530293e-01 -6.98333025e-01 -4.96862382e-01
-2.47661769e-01 -3.30476910e-01 -7.03249574e-01 1.80772230e-01
-1.45046785e-01 4.05371748e-03 2.12509885e-01 1.12389123e+00
5.36479652e-01 2.54517764e-01 5.04736960e-01 -1.20098305e+00
-9.80618954e-01 4.00285095e-01 -8.23680997e-01 2.78272808e-01
1.20668180e-01 -2.76402891e-01 1.30144525e+00 -1.26067877e+00
1.07979923e-01 1.28602862e+00 5.76952636e-01 5.73217034e-01
-4.86061245e-01 -8.75818729e-01 6.80365503e-01 2.66680747e-01
-1.13617218e+00 -1.89924613e-01 1.14172947e+00 -3.24590027e-01
8.14472258e-01 -1.24062691e-03 3.41329545e-01 1.17511702e+00
4.30771470e-01 9.29973006e-01 1.12506008e+00 -3.51303697e-01
-2.90993098e-02 5.56351125e-01 5.38211048e-01 6.51773393e-01
-6.82020485e-02 -3.57647091e-01 -5.37349463e-01 -1.94288716e-01
1.62216976e-01 1.54594019e-01 -3.65038693e-01 -2.90565938e-01
-7.19598591e-01 1.26431072e+00 3.40722233e-01 4.72748309e-01
-4.95162696e-01 2.23741066e-02 7.78387427e-01 3.41751993e-01
1.34301507e+00 2.50675052e-01 -9.89469409e-01 -1.69835128e-02
-2.30790928e-01 -2.86538154e-01 5.34448802e-01 6.51429951e-01
6.80954099e-01 -1.63469225e-01 -8.62914026e-02 1.09813738e+00
6.74820602e-01 1.77709088e-01 9.27441359e-01 -8.76650661e-02
5.24968445e-01 8.44675243e-01 -5.03305309e-02 -1.12282467e+00
-6.03036940e-01 -5.74734747e-01 -7.27733135e-01 -2.23251045e-01
-4.52880204e-01 -6.34663522e-01 -4.13690358e-01 1.59819973e+00
2.68027663e-01 1.22917518e-01 4.47367877e-01 9.24484432e-01
1.05371678e+00 7.38200247e-01 5.07300496e-01 -1.93928957e-01
1.50073349e+00 -1.56468618e+00 -9.37739611e-01 -4.91018444e-01
7.07067966e-01 -4.55593348e-01 9.81892049e-01 1.20435797e-01
-7.77966142e-01 -5.48488319e-01 -1.08214831e+00 3.29264440e-02
-5.77613354e-01 4.49297518e-01 8.30093384e-01 6.10820174e-01
-6.21189892e-01 1.10186383e-01 -4.60680574e-01 6.99217170e-02
5.02100348e-01 2.20539510e-01 -3.78127217e-01 2.38544330e-01
-1.57257283e+00 8.74840260e-01 2.96176761e-01 5.48334241e-01
-1.17620766e-01 -3.92439872e-01 -1.21393847e+00 3.62132102e-01
3.84960435e-02 -4.97992784e-01 1.05537426e+00 -1.51447511e+00
-1.59973991e+00 6.82442188e-01 -2.31547132e-01 2.53895402e-01
-4.30656187e-02 -4.93860036e-01 -7.24812627e-01 2.24889725e-01
1.63903430e-01 3.42514783e-01 6.98805928e-01 -1.21328211e+00
-5.93418837e-01 -5.70089757e-01 -1.30072758e-01 4.80398208e-01
-1.18970919e+00 3.63690257e-01 -2.18783811e-01 -6.01022899e-01
-2.19313517e-01 -5.77762723e-01 -3.59353125e-01 -4.90002871e-01
-2.21692607e-01 -6.43079937e-01 7.82132626e-01 -3.01369667e-01
1.10416603e+00 -2.32860637e+00 2.95327485e-01 3.96270445e-03
1.31370351e-01 7.91595876e-02 -3.29561949e-01 1.40142575e-01
-3.19981098e-01 1.23336032e-01 -1.02756903e-01 -6.65628791e-01
-1.73253175e-02 -9.30250883e-02 -4.48280096e-01 3.39448929e-01
6.11821115e-01 1.04884601e+00 -7.40892768e-01 -4.30950105e-01
-1.40097648e-01 5.62420845e-01 -2.92070299e-01 2.86080629e-01
2.00662330e-01 3.25816900e-01 -9.10954118e-01 6.22269928e-01
5.97369075e-01 -4.04922217e-01 1.20260753e-01 -8.30379426e-02
1.33218333e-01 1.29746258e-01 -8.35975826e-01 1.49526513e+00
-7.51137257e-01 2.13416874e-01 -5.75930811e-03 -1.05962992e+00
1.11480069e+00 3.24544221e-01 4.56214219e-01 -5.60229123e-01
7.42704511e-01 -2.23278403e-02 -3.36915672e-01 -4.92887139e-01
5.27156770e-01 -6.00148380e-01 -5.34090459e-01 5.82398593e-01
2.64533430e-01 2.27554068e-01 -1.45472556e-01 -1.64272428e-01
4.48834509e-01 -1.78804100e-01 3.15423310e-01 -3.28912675e-01
8.57103646e-01 -4.50153202e-01 6.16427124e-01 8.16035792e-02
-4.85004991e-01 2.28739679e-01 5.39315581e-01 -5.27137935e-01
-1.35754481e-01 -3.14105511e-01 -1.47116840e-01 1.62788904e+00
4.29656714e-01 -3.27812642e-01 -4.95073944e-01 -1.26386631e+00
-2.25741193e-01 4.15326506e-01 -9.11264181e-01 -6.29922330e-01
-1.42576009e-01 -1.09724534e+00 3.36684614e-01 8.22041869e-01
3.84108067e-01 -1.46004939e+00 -6.68666959e-01 4.77409810e-02
-3.13481927e-01 -1.06294525e+00 -5.26668906e-01 4.90101904e-01
-5.19819140e-01 -7.70975709e-01 -6.55174196e-01 -1.01207113e+00
5.73591232e-01 3.04485142e-01 9.90882337e-01 2.55678952e-01
1.25352740e-01 3.99042934e-01 -6.88615024e-01 -6.26640975e-01
2.17172876e-01 2.94688076e-01 -8.56427178e-02 6.57630205e-01
6.57435536e-01 -1.33651689e-01 -3.70075166e-01 -1.46033661e-02
-7.64616430e-01 -2.15937734e-01 6.35921299e-01 9.85466361e-01
2.74694890e-01 -1.01408064e-01 1.03709185e+00 -8.28241110e-01
1.14406157e+00 -8.01808357e-01 1.29803017e-01 2.76180923e-01
-6.80662870e-01 -1.21519506e-01 7.71330476e-01 -3.54238898e-01
-1.05763829e+00 -1.24555573e-01 -3.36427540e-01 -2.47249559e-01
-3.23967002e-02 8.63668621e-01 -3.82844478e-01 -1.10064089e-01
-1.60709828e-01 3.13295484e-01 -2.30359182e-01 -3.13926250e-01
2.86164194e-01 1.00007391e+00 -1.34660631e-01 -4.47659284e-01
-2.96657626e-03 2.38666117e-01 -4.04178023e-01 -2.79760599e-01
-1.30917668e+00 -6.27197981e-01 -3.12274158e-01 -2.59362608e-01
8.29006672e-01 -9.57621813e-01 -5.82876086e-01 4.91522372e-01
-1.08645332e+00 1.92672700e-01 -6.10483112e-03 4.19331342e-01
-4.48430985e-01 2.58766115e-01 -8.10734749e-01 -7.10271478e-01
-7.71155834e-01 -1.35866892e+00 1.44022501e+00 4.78379816e-01
1.12039810e-02 -1.33409059e+00 2.88449019e-01 2.28904083e-01
3.31419230e-01 -1.31330445e-01 8.10849965e-01 -9.31362867e-01
3.79986405e-01 -3.08806419e-01 -3.82694811e-01 5.13561666e-01
9.48239639e-02 -3.63192171e-01 -1.01837480e+00 -1.00475542e-01
2.65857548e-01 -8.33333373e-01 1.13464403e+00 -1.83209814e-02
1.26290321e+00 9.62863211e-03 -2.32527956e-01 3.60060602e-01
1.35666597e+00 7.79116899e-02 4.49937195e-01 3.17773551e-01
6.19977593e-01 7.06280112e-01 8.51721168e-01 6.09908581e-01
6.31234944e-01 4.00352746e-01 5.49654722e-01 -2.67839193e-01
4.60685968e-01 1.28292754e-01 2.98239589e-01 1.14314246e+00
3.05348903e-01 -3.05332214e-01 -5.82952917e-01 4.79510754e-01
-1.94641614e+00 -5.94579995e-01 -9.92924646e-02 1.21142673e+00
6.04343295e-01 -7.72731975e-02 -3.00097853e-01 -1.12023644e-01
6.69835746e-01 4.46374506e-01 -5.06083548e-01 -7.31837511e-01
-1.89751357e-01 3.93677920e-01 -5.03547490e-02 2.91850775e-01
-1.51263869e+00 9.12736833e-01 6.10123539e+00 5.73863447e-01
-1.22460961e+00 1.85711727e-01 7.91998446e-01 -1.64325330e-02
-1.88530058e-01 -5.66476643e-01 -7.81882524e-01 4.59874123e-01
8.70807528e-01 -9.85315740e-02 -2.34098226e-01 1.01739681e+00
-1.49086386e-01 3.29081833e-01 -4.58579510e-01 9.78019595e-01
5.23167789e-01 -7.43078768e-01 8.70874301e-02 -3.12383592e-01
5.91284513e-01 -2.89742351e-01 2.16115430e-01 8.10147762e-01
-8.01790655e-02 -7.06835806e-01 5.89709401e-01 6.06181800e-01
4.11549062e-01 -1.03497863e+00 1.03076148e+00 1.33852884e-01
-1.17307568e+00 -3.36605519e-01 -3.38947535e-01 -2.93618411e-01
3.98152545e-02 4.21541661e-01 -2.10233852e-02 7.81342924e-01
6.68435454e-01 1.04862154e+00 -6.55098975e-01 4.51075643e-01
-3.18959057e-01 5.42218685e-01 4.45701867e-01 -4.15639222e-01
4.64045525e-01 -1.49245620e-01 -4.25401935e-03 1.30645490e+00
7.06279278e-02 2.68291593e-01 1.96846426e-01 3.91262382e-01
-2.76018202e-01 7.61458814e-01 -3.25723082e-01 -2.92229708e-02
-1.16803057e-01 1.90925932e+00 -5.47244132e-01 -3.52278054e-01
-6.74813449e-01 1.22332227e+00 7.53159404e-01 2.69169688e-01
-9.23600912e-01 -7.15611160e-01 7.08171427e-01 -9.66092944e-01
4.81946290e-01 2.52796113e-01 -3.48712564e-01 -1.46053910e+00
-1.45554775e-02 -6.57694817e-01 4.16318417e-01 -7.70853877e-01
-1.61138737e+00 8.86086643e-01 -6.52020395e-01 -1.01411605e+00
1.29604757e-01 -9.34335172e-01 -7.06568182e-01 9.70743954e-01
-1.85856807e+00 -1.25696027e+00 -1.94834724e-01 5.44547379e-01
5.74417770e-01 -5.94462715e-02 1.13693202e+00 4.89848405e-01
-8.68563712e-01 7.58487821e-01 1.03172414e-01 1.20959520e-01
8.18671703e-01 -1.15141070e+00 -1.42990455e-01 3.41812402e-01
-5.73336929e-02 4.24583435e-01 2.54095376e-01 -2.36688778e-01
-9.60437834e-01 -1.04918969e+00 9.53055322e-01 -1.75504684e-01
6.99510872e-01 -2.79322356e-01 -7.99469292e-01 5.71574330e-01
5.55566549e-01 -4.62615825e-02 1.38228428e+00 4.64082479e-01
-4.34759021e-01 3.33953090e-02 -9.65930820e-01 1.32248104e-01
3.56065184e-01 -5.10235667e-01 -5.90976477e-01 2.00609833e-01
8.66294742e-01 -7.16524571e-02 -8.00099552e-01 5.51704764e-01
5.70580661e-01 -5.24054229e-01 5.66364288e-01 -1.08779526e+00
9.58342791e-01 1.95522130e-01 -1.97704107e-01 -1.80153751e+00
-2.95650154e-01 2.72233725e-01 -4.26982753e-02 1.20451093e+00
3.01049709e-01 -7.14042187e-01 4.07682627e-01 2.74243265e-01
-2.26835415e-01 -1.29444122e+00 -7.18547881e-01 -2.30573028e-01
1.35478497e-01 -3.44736695e-01 5.92745543e-01 1.20575762e+00
4.21327055e-01 1.01491666e+00 -5.34402370e-01 -2.87772287e-02
-1.30521245e-02 5.36863983e-01 1.91622883e-01 -1.15987527e+00
-5.04569486e-02 -7.11630404e-01 -1.79198205e-01 -1.03119636e+00
9.34679329e-01 -7.12099075e-01 1.24588981e-01 -1.30940509e+00
5.24552405e-01 3.01811527e-02 -9.55111980e-01 4.64170516e-01
-8.36467147e-01 2.95428425e-01 -1.01823092e-01 -1.74321651e-01
-1.06854975e+00 1.17919743e+00 1.11274409e+00 -2.22410962e-01
1.41813144e-01 -1.12046733e-01 -1.39060402e+00 7.66503274e-01
8.12071085e-01 -3.95483732e-01 -1.31534994e-01 -6.06603444e-01
2.77361125e-01 -3.86056080e-02 -4.68145721e-02 -4.41481441e-01
-1.21467955e-01 -6.89941347e-02 2.72909492e-01 -3.61724108e-01
3.48844051e-01 -9.42113876e-01 -7.30882525e-01 3.24815691e-01
-5.66787481e-01 6.22666813e-02 2.67106205e-01 6.30925298e-01
-5.50946176e-01 -3.36798966e-01 5.38263619e-01 1.83222920e-01
-8.11887920e-01 3.45124125e-01 -2.02547118e-01 -1.63645387e-01
1.01726103e+00 5.59548378e-01 -2.07564935e-01 -2.49272883e-01
-6.15479231e-01 4.52958614e-01 -1.40868261e-01 9.50175822e-01
7.85381079e-01 -1.64238691e+00 -5.59590459e-01 1.68020099e-01
5.23173571e-01 -5.90514660e-01 4.02836502e-01 6.40251517e-01
1.68646455e-01 4.59196895e-01 -1.32051945e-01 -5.63478880e-02
-1.17171550e+00 6.86787844e-01 4.44753408e-01 -5.46342611e-01
-4.56905887e-02 1.01227498e+00 3.63395274e-01 -6.56757474e-01
-3.61544378e-02 9.21310931e-02 -9.59144294e-01 3.93490255e-01
6.56749308e-01 4.48902398e-02 4.15428579e-02 -9.02421355e-01
-5.95548332e-01 6.04665577e-01 -3.22778046e-01 2.09019735e-01
1.38759172e+00 -3.71313572e-01 -3.29747528e-01 7.75588632e-01
1.52336943e+00 2.69541591e-02 -8.84025872e-01 -1.12535536e-01
1.10969014e-01 -1.42015502e-01 2.08134025e-01 -7.50900745e-01
-1.34899020e+00 1.04660559e+00 5.41806579e-01 3.95649821e-01
1.38972139e+00 6.60507753e-02 8.05007577e-01 1.69236496e-01
6.09319247e-02 -9.54603791e-01 2.48747304e-01 7.19851732e-01
7.22597539e-01 -1.41411495e+00 -3.81873578e-01 -2.84949601e-01
-1.00132608e+00 1.17277253e+00 9.65763211e-01 -1.62024543e-01
7.75831044e-01 -1.45598933e-01 3.88982147e-01 -5.27343035e-01
-8.98571908e-01 -2.05689877e-01 4.57312703e-01 -2.30210871e-02
6.77038014e-01 -1.71634611e-02 -7.04869449e-01 1.52765405e+00
1.95976943e-01 -1.99030548e-01 2.81392038e-01 7.23564148e-01
-4.80791122e-01 -1.00369835e+00 7.23315449e-03 3.10549051e-01
-7.95929790e-01 -1.51173159e-01 -5.22861004e-01 3.55196148e-01
3.84464338e-02 1.23147988e+00 2.28576479e-03 -4.88810718e-01
4.64295268e-01 3.45927805e-01 9.52173173e-02 -7.05468595e-01
-8.70857894e-01 -5.66811971e-02 -1.69624716e-01 -3.00876170e-01
-6.47594929e-01 -2.82415926e-01 -1.10155261e+00 4.24531460e-01
-6.83880448e-01 5.65503597e-01 6.65556073e-01 1.12824774e+00
6.04291320e-01 9.05830026e-01 1.13776147e+00 -5.97130120e-01
-6.33942783e-01 -1.26902080e+00 -6.07427418e-01 4.60438937e-01
2.57271647e-01 -7.78770268e-01 -4.04375613e-01 -3.39273691e-01] | [11.561187744140625, 6.660256385803223] |
263fedf7-3e81-44ac-b873-0e525bdb1b72 | late-reverberation-suppression-using-u-nets | 2110.02144 | null | https://arxiv.org/abs/2110.02144v1 | https://arxiv.org/pdf/2110.02144v1.pdf | Late reverberation suppression using U-nets | In real-world settings, speech signals are almost always affected by reverberation produced by the working environment; these corrupted signals need to be \emph{dereverberated} prior to performing, e.g., speech recognition, speech-to-text conversion, compression, or general audio enhancement. In this paper, we propose a supervised dereverberation technique using \emph{U-nets with skip connections}, which are fully-convolutional encoder-decoder networks with layers arranged in the form of an "U" and connections that "skip" some layers. Building on this architecture, we address speech dereverberation through the lens of Late Reverberation Suppression (LS). Via experiments on synthetic and real-world data with different noise levels and reverberation settings, we show that our proposed method termed "LS U-net" improves quality, intelligibility and other performance metrics compared to the original U-net method and it is on par with the state-of-the-art GAN-based approaches. | ['Felipe Tobar', 'Diego León'] | 2021-10-05 | null | null | null | null | ['speech-dereverberation'] | ['speech'] | [ 3.93541425e-01 -1.45424716e-02 5.08192182e-01 -1.93902329e-01
-6.27646923e-01 -3.31801146e-01 2.87801981e-01 -3.26920033e-01
-3.63577567e-02 7.06074953e-01 6.91215038e-01 -4.28747386e-01
4.96926568e-02 -5.11130333e-01 -8.97901773e-01 -7.05596805e-01
7.84347877e-02 -3.75638545e-01 -1.39058009e-01 -3.82309288e-01
-1.79144934e-01 2.18259349e-01 -1.56945515e+00 3.40424657e-01
6.71357572e-01 1.12340081e+00 2.74284601e-01 1.02492940e+00
2.42408171e-01 1.07586730e+00 -1.20574522e+00 -1.24738954e-01
8.38335603e-02 -7.64829099e-01 -3.02327931e-01 6.46778941e-02
3.14772762e-02 -5.02892613e-01 -6.93968594e-01 9.68616962e-01
1.03830719e+00 3.64590466e-01 1.82614729e-01 -6.13670588e-01
-7.90774763e-01 7.92867362e-01 -2.08038583e-01 3.21489424e-01
4.69543815e-01 7.37581104e-02 7.34051645e-01 -8.69653642e-01
1.90118343e-01 1.12002826e+00 8.43695700e-01 5.32175422e-01
-1.08070660e+00 -7.83175051e-01 -1.61069289e-01 4.10154551e-01
-1.06591988e+00 -8.50719094e-01 1.05768478e+00 -5.55819459e-03
1.10143387e+00 5.76034904e-01 2.74737269e-01 1.32349384e+00
6.46947324e-02 6.12304091e-01 8.93750608e-01 -5.56047499e-01
1.87689617e-01 -3.08498628e-02 -1.63044170e-01 9.56869647e-02
-3.41008395e-01 6.14895344e-01 -7.50721157e-01 3.68830740e-01
4.26046222e-01 -4.23810035e-01 -8.43898654e-01 5.45980752e-01
-9.18693423e-01 2.12179288e-01 5.41772485e-01 4.39631313e-01
-3.47349048e-01 2.89025247e-01 4.00253087e-01 7.11692095e-01
7.74207771e-01 2.90394813e-01 -3.34503919e-01 -2.26075336e-01
-9.74225938e-01 -1.57696694e-01 5.94065070e-01 8.43558490e-01
2.62812674e-01 7.90398359e-01 -2.26263568e-01 1.23774469e+00
8.54279548e-02 4.98899400e-01 5.51118255e-01 -5.99274993e-01
4.66057360e-01 -3.01365882e-01 1.67077586e-01 -5.29056728e-01
-1.75204828e-01 -1.05901539e+00 -1.22390008e+00 3.00923854e-01
-1.94805384e-01 -5.76502204e-01 -9.62311387e-01 1.80604517e+00
-5.45824692e-02 5.62155545e-01 3.04725200e-01 9.14263606e-01
8.53345692e-01 9.15829897e-01 -3.69996369e-01 -4.39201862e-01
9.12337005e-01 -1.12137973e+00 -1.30545580e+00 -1.03361882e-01
-4.38924432e-02 -1.14839709e+00 1.15886390e+00 6.63983583e-01
-1.19178140e+00 -9.69227672e-01 -1.22808039e+00 1.64562151e-01
-1.79150030e-01 1.43249407e-01 -5.92918210e-02 1.05148077e+00
-1.34432840e+00 7.36655593e-01 -4.53082919e-01 2.20480397e-01
-4.18834761e-02 1.29238263e-01 -1.37154773e-01 2.82158554e-01
-1.33556545e+00 8.48880589e-01 -1.17848543e-02 4.07377630e-01
-1.13149738e+00 -6.06708348e-01 -7.09614277e-01 3.58408749e-01
1.55529797e-01 -4.27742064e-01 1.43203056e+00 -1.11113346e+00
-1.95316899e+00 5.15017658e-02 -8.27924162e-02 -6.74616456e-01
4.80448723e-01 -6.55717313e-01 -1.04969943e+00 -5.29342405e-02
-5.78923106e-01 -1.22877862e-02 1.07437897e+00 -1.33921993e+00
-2.99420118e-01 1.93051454e-02 -7.29791969e-02 1.79119706e-01
-2.77678460e-01 2.53034681e-01 2.98463064e-03 -1.15788639e+00
5.50422408e-02 -5.16754329e-01 9.57302451e-02 -4.35604304e-01
-5.00153363e-01 2.16445670e-01 1.18872261e+00 -1.25510228e+00
1.29661345e+00 -2.42477274e+00 -1.25881255e-01 -5.25288470e-02
-3.75306085e-02 8.07660520e-01 -2.01476693e-01 5.27246356e-01
-4.70063120e-01 -4.25562542e-03 -2.14325935e-01 -7.16338992e-01
3.17005627e-02 4.80609909e-02 -3.93936843e-01 4.02103871e-01
5.21229394e-02 2.84253865e-01 -6.85374737e-01 1.58417106e-01
5.84334552e-01 9.78613973e-01 -4.25707161e-01 5.72352648e-01
7.61884898e-02 7.32388198e-01 2.70724505e-01 2.88365960e-01
7.49636829e-01 5.99057078e-01 -2.40017131e-01 -9.04267803e-02
-2.28775308e-01 5.86917877e-01 -1.03160787e+00 1.49587798e+00
-1.08855045e+00 1.16951072e+00 5.00490725e-01 -7.69030511e-01
9.97426927e-01 8.66319418e-01 -5.25353774e-02 -8.34783494e-01
2.58066654e-01 4.24793869e-01 1.25627249e-01 -4.72016573e-01
3.08494598e-01 -1.31745204e-01 5.28937519e-01 1.93444878e-01
2.17434242e-01 -1.85089678e-01 -1.94286719e-01 -1.71510234e-01
1.17803669e+00 -1.13423400e-01 -3.15138958e-02 6.83507696e-02
6.56101823e-01 -1.11333740e+00 3.28702807e-01 5.88742673e-01
2.13019317e-03 9.15989518e-01 1.76629961e-01 6.12408072e-02
-1.12911606e+00 -1.21877444e+00 -1.05031632e-01 8.14449668e-01
-2.42459938e-01 -2.94005305e-01 -1.16326296e+00 -8.10853615e-02
-6.00533485e-01 1.13638592e+00 -3.41084838e-01 -4.09218490e-01
-5.56256235e-01 -4.09323663e-01 7.30739415e-01 4.58284706e-01
7.93671966e-01 -1.27408135e+00 -2.88640201e-01 5.58736801e-01
-3.14712346e-01 -9.28268075e-01 -5.85559070e-01 6.06983066e-01
-2.90945381e-01 -5.07421970e-01 -8.21318269e-01 -6.48423314e-01
3.82549077e-01 1.86374590e-01 9.11173046e-01 -1.44026682e-01
-3.70146483e-02 3.63780931e-02 -7.92000413e-01 -4.75187212e-01
-6.81811750e-01 -3.26742411e-01 1.06657214e-01 2.67944127e-01
-2.11007416e-01 -1.21090484e+00 -8.24858546e-01 2.22462937e-01
-1.04996228e+00 -9.58337858e-02 4.92623031e-01 8.77149045e-01
2.87596852e-01 5.88843644e-01 9.32014406e-01 -5.32693803e-01
9.43958819e-01 -3.03677857e-01 -2.32373565e-01 -5.68024181e-02
-1.88134864e-01 -3.11484307e-01 1.18574107e+00 -4.21258420e-01
-1.38061345e+00 -4.87465233e-01 -7.40469992e-01 -4.73212481e-01
-2.76580960e-01 2.94293493e-01 -5.09096324e-01 1.47528186e-01
6.23614907e-01 3.85862023e-01 -4.63746995e-01 -8.61605704e-01
2.29640424e-01 1.32148576e+00 8.27482522e-01 7.41814673e-02
6.88569069e-01 -2.48606969e-02 -3.90705645e-01 -1.07134616e+00
-5.24523914e-01 -1.71091601e-01 -1.23497024e-01 -3.13646615e-01
6.17651761e-01 -8.19517255e-01 -3.62407923e-01 8.52865040e-01
-1.38203549e+00 -5.04489005e-01 -3.75717223e-01 6.21050239e-01
-5.28780222e-01 2.13472739e-01 -6.83214247e-01 -1.11694300e+00
-4.91665423e-01 -1.05371416e+00 7.14768887e-01 1.50232598e-01
-2.20235274e-03 -6.13674223e-01 -2.14347709e-02 3.67951632e-01
8.55484486e-01 -3.73710431e-02 7.21539974e-01 -2.59618044e-01
-4.35207598e-02 -1.08111903e-01 1.26925439e-01 1.24755299e+00
3.97159785e-01 -3.50659758e-01 -1.56904662e+00 -3.27425629e-01
5.58388770e-01 4.70266538e-03 7.73312747e-01 5.32903254e-01
1.45889378e+00 -5.21951735e-01 3.75298291e-01 7.45648742e-01
1.12876093e+00 5.20156860e-01 1.02829266e+00 -1.68521255e-01
3.75518739e-01 2.82567620e-01 -3.68146300e-02 5.34245372e-01
-2.73457438e-01 4.55100387e-01 4.59885985e-01 -5.36519766e-01
-9.15629327e-01 -2.78326303e-01 5.50635815e-01 1.24764049e+00
1.98304895e-02 -8.46522450e-01 -3.56107205e-01 5.30020118e-01
-1.24174893e+00 -8.93486798e-01 2.84207799e-03 2.19470716e+00
8.08883011e-01 7.22839460e-02 -3.17724884e-01 7.96244919e-01
7.94436574e-01 3.43309581e-01 -4.37609285e-01 -7.05686688e-01
-2.78493136e-01 6.33423269e-01 3.08319747e-01 6.27848744e-01
-7.85104334e-01 6.33027852e-01 6.03188848e+00 8.64918351e-01
-1.27466369e+00 4.53103691e-01 7.74448931e-01 -1.45478085e-01
-3.47506344e-01 -4.75202650e-01 -1.77059919e-02 4.73519325e-01
1.14906728e+00 1.27688706e-01 8.95366669e-01 4.95963514e-01
6.99975789e-01 1.15860544e-01 -9.22664702e-01 9.60714757e-01
1.55108377e-01 -8.24960589e-01 -3.26197833e-01 -3.41284841e-01
6.64401948e-01 -9.59279463e-02 3.21946770e-01 2.10595861e-01
1.78043246e-01 -1.21495986e+00 9.66517031e-01 3.28090519e-01
1.08157647e+00 -9.20687556e-01 7.14893043e-01 1.78244874e-01
-1.04421031e+00 -1.80451453e-01 -2.14093760e-01 -4.02342193e-02
2.79217273e-01 9.95747983e-01 -7.54196763e-01 6.84723616e-01
7.80448139e-01 1.68410748e-01 8.91275108e-02 8.80636930e-01
-5.91803253e-01 1.15220463e+00 -8.24643373e-02 1.95515245e-01
5.63285649e-02 7.08772168e-02 7.69215107e-01 1.32929540e+00
6.38761103e-01 1.28662467e-01 -5.28014481e-01 6.67856395e-01
-2.84691304e-01 -1.44365177e-01 -4.50135529e-01 1.85366258e-01
4.47681457e-01 6.87268615e-01 -2.58836985e-01 -1.26305833e-01
-2.83586740e-01 1.07534289e+00 -3.25829804e-01 7.40128040e-01
-9.12330210e-01 -8.52515399e-01 6.70833349e-01 -2.06320975e-02
4.55070525e-01 -1.86809525e-01 -2.35069245e-01 -7.80567646e-01
2.56895334e-01 -9.67870712e-01 -3.34217459e-01 -1.23940241e+00
-9.57757175e-01 1.12928951e+00 -4.10865784e-01 -1.33326197e+00
-1.62460640e-01 -4.45441723e-01 -7.72350311e-01 1.07732069e+00
-1.55325246e+00 -8.09428573e-01 -1.71614513e-01 6.86493337e-01
8.08564126e-01 -2.12274820e-01 7.66462862e-01 6.90417707e-01
-4.21313524e-01 6.78581476e-01 4.54665959e-01 -9.46688205e-02
6.03486538e-01 -1.07276201e+00 5.84365129e-01 1.30760777e+00
1.17634557e-01 3.14228654e-01 1.06522071e+00 -3.62999946e-01
-1.00634110e+00 -1.25122476e+00 8.30277741e-01 3.10763031e-01
4.01482135e-01 -6.67970359e-01 -6.02503121e-01 4.03295279e-01
8.13788831e-01 -8.77573937e-02 6.57881021e-01 -2.27056995e-01
-1.97057933e-01 -5.23455739e-01 -1.23935068e+00 6.00810349e-01
9.54615593e-01 -7.47015357e-01 -4.56291735e-01 2.19633490e-01
1.12994909e+00 -3.90714526e-01 -4.06576931e-01 2.57928848e-01
2.26055592e-01 -1.19366026e+00 8.59556496e-01 -9.47371647e-02
4.23161954e-01 -3.39948893e-01 -4.03437763e-01 -2.04725862e+00
-3.10200565e-02 -1.26196611e+00 -1.14190720e-01 1.51446652e+00
4.25263762e-01 -5.95813394e-01 1.31721467e-01 -1.70214579e-01
-7.42584884e-01 -2.81003118e-01 -1.19842708e+00 -6.93154454e-01
-1.83575243e-01 -7.83526599e-01 7.64305711e-01 4.57056910e-01
-1.56441078e-01 1.88375965e-01 -9.66335177e-01 2.82309949e-01
3.95659447e-01 -4.64386493e-01 4.77121800e-01 -4.48991776e-01
-5.49095392e-01 -1.97598830e-01 -1.29048005e-01 -1.09546351e+00
2.20357161e-02 -3.69954497e-01 4.46753770e-01 -1.48055816e+00
-5.25108993e-01 -8.34516883e-02 -5.28293133e-01 1.08260050e-01
3.91219370e-02 1.93431258e-01 7.58626685e-02 -3.53080630e-01
2.93976218e-02 9.36231196e-01 1.15795577e+00 -9.58318114e-02
-1.68438405e-01 3.31205785e-01 -4.63657230e-01 4.95306522e-01
8.66853774e-01 -1.79040730e-01 -6.84971333e-01 -6.99460447e-01
-1.39446393e-01 4.70945239e-01 2.85855591e-01 -1.57494509e+00
1.42385036e-01 4.82765317e-01 2.16700122e-01 -4.48073894e-01
5.65013945e-01 -9.38778818e-01 2.59380728e-01 3.21930945e-01
-4.91958678e-01 -1.32589117e-01 1.39465645e-01 4.25079376e-01
-5.38425803e-01 6.52209222e-02 8.65920544e-01 1.09088659e-01
-4.84337285e-02 -1.41229272e-01 -6.54712319e-01 -3.79717529e-01
4.65188652e-01 -4.09350209e-02 -3.52579415e-01 -9.21734154e-01
-7.58624971e-01 -4.28129464e-01 -3.64576429e-01 2.84444958e-01
7.61997163e-01 -1.13591063e+00 -8.44961345e-01 4.15828437e-01
-4.78913426e-01 -2.06877977e-01 5.49094915e-01 6.52917862e-01
-4.55191523e-01 3.31001520e-01 1.52062404e-03 -1.06681086e-01
-1.19095957e+00 4.45320815e-01 5.80787659e-01 3.96762341e-02
-6.40302181e-01 1.02234423e+00 2.05694422e-01 -1.95908248e-01
6.35155261e-01 -5.17299950e-01 -9.56894606e-02 -3.47653419e-01
6.39361322e-01 5.01200557e-01 5.89638591e-01 -4.19418007e-01
3.14005907e-03 -1.88711584e-02 2.64901310e-01 -4.08302188e-01
1.31300211e+00 -3.58436763e-01 3.78152393e-02 4.20538992e-01
1.04983962e+00 1.93640411e-01 -1.15767014e+00 -2.02236235e-01
-6.11851096e-01 -4.51518267e-01 4.02199805e-01 -1.24687099e+00
-1.41512120e+00 1.07228875e+00 8.59369516e-01 4.05035794e-01
1.75323439e+00 -3.94991934e-01 1.03735340e+00 1.95487380e-01
9.90256369e-02 -1.21874070e+00 1.84340239e-01 5.31142175e-01
1.28770769e+00 -5.64269722e-01 -6.76350772e-01 -1.74555019e-01
-2.27758259e-01 8.87532830e-01 3.00816447e-01 3.84074147e-03
7.90609658e-01 7.34264553e-01 4.17897314e-01 4.46439475e-01
-6.35524571e-01 -6.14893138e-02 -2.74798321e-03 7.84302473e-01
6.31080151e-01 7.40707368e-02 -9.10120755e-02 6.96485758e-01
-8.42767894e-01 -2.39281639e-01 4.55292046e-01 5.80376625e-01
-4.05843735e-01 -9.77442741e-01 -6.86481655e-01 1.84524044e-01
-6.43511057e-01 -4.44333702e-01 -3.91373247e-01 1.40881449e-01
3.44207972e-01 1.65116799e+00 -1.84260383e-02 -7.76272297e-01
5.24824500e-01 -4.59825806e-02 2.41424963e-01 -2.78481215e-01
-1.01100385e+00 4.72023070e-01 2.88503021e-01 -2.29712933e-01
-2.55918354e-01 -3.81818444e-01 -8.45070362e-01 -3.34044755e-01
-5.60896099e-01 -3.07364129e-02 8.20289493e-01 9.20487106e-01
1.54315710e-01 1.44162524e+00 9.97507811e-01 -7.18478084e-01
-4.41596806e-01 -1.54430139e+00 -7.88012624e-01 2.88963616e-01
9.49569046e-01 -1.78773955e-01 -6.02993429e-01 1.29935041e-01] | [15.046734809875488, 5.935602188110352] |
75d6394a-169e-4d6b-b2ba-65979d235ca9 | predicting-eye-gaze-location-on-websites | 2211.08074 | null | https://arxiv.org/abs/2211.08074v3 | https://arxiv.org/pdf/2211.08074v3.pdf | Predicting Eye Gaze Location on Websites | World-wide-web, with the website and webpage as the main interface, facilitates the dissemination of important information. Hence it is crucial to optimize them for better user interaction, which is primarily done by analyzing users' behavior, especially users' eye-gaze locations. However, gathering these data is still considered to be labor and time intensive. In this work, we enable the development of automatic eye-gaze estimations given a website screenshots as the input. This is done by the curation of a unified dataset that consists of website screenshots, eye-gaze heatmap and website's layout information in the form of image and text masks. Our pre-processed dataset allows us to propose an effective deep learning-based model that leverages both image and text spatial location, which is combined through attention mechanism for effective eye-gaze prediction. In our experiment, we show the benefit of careful fine-tuning using our unified dataset to improve the accuracy of eye-gaze predictions. We further observe the capability of our model to focus on the targeted areas (images and text) to achieve high accuracy. Finally, the comparison with other alternatives shows the state-of-the-art result of our model establishing the benchmark for the eye-gaze prediction task. | ['Steffen Staab', 'Decky Aspandi', 'Ciheng Zhang'] | 2022-11-15 | null | null | null | null | ['eye-tracking'] | ['computer-vision'] | [ 1.21982500e-01 -7.82064423e-02 -1.49775282e-01 -2.29952753e-01
-3.74884427e-01 -4.22451973e-01 4.24646229e-01 7.14035705e-02
-2.54083276e-01 1.73420697e-01 1.70267895e-01 -5.45464039e-01
-1.76293701e-01 -4.82361078e-01 -6.50351346e-01 -5.61458528e-01
2.35154912e-01 -2.43239671e-01 1.41045049e-01 -2.16455311e-01
8.58284771e-01 1.33470014e-01 -2.00426984e+00 2.28338704e-01
1.12637734e+00 1.22152293e+00 3.74249279e-01 7.38478243e-01
-1.66990116e-01 5.50750971e-01 -3.12386185e-01 -3.22920620e-01
1.26446083e-01 -1.60622224e-01 -6.23596370e-01 1.78350747e-01
6.51796281e-01 -3.55712116e-01 1.22494489e-01 9.09518957e-01
4.32255894e-01 2.02204213e-01 4.44700122e-01 -1.25081730e+00
-9.77063239e-01 -1.34653971e-01 -9.07321453e-01 5.16865194e-01
4.71589684e-01 4.97283161e-01 1.21584928e+00 -7.09763229e-01
2.89031804e-01 8.95294547e-01 4.14125770e-01 3.12206000e-01
-9.65876043e-01 -4.64509428e-01 1.89725801e-01 5.42155087e-01
-1.26245451e+00 -5.77674747e-01 8.82107854e-01 -4.72505331e-01
7.72713840e-01 6.05289757e-01 4.53582525e-01 1.02502847e+00
2.59840846e-01 8.05275142e-01 9.37722266e-01 -1.00323987e+00
-1.43036157e-01 5.52516341e-01 4.34476912e-01 6.10831559e-01
8.55209753e-02 -7.30127320e-02 -7.73766756e-01 4.19346094e-01
3.78975779e-01 2.35859081e-01 -5.62819183e-01 -3.18612576e-01
-9.02400494e-01 5.50947368e-01 7.50317276e-01 1.40509412e-01
-4.43762451e-01 -3.22484463e-01 -1.29245833e-01 3.91793810e-02
6.18527770e-01 2.86308348e-01 -2.93212712e-01 -2.31364548e-01
-8.84759545e-01 4.75973263e-02 5.86870849e-01 7.89489388e-01
8.60678732e-01 -6.41281545e-01 -3.49442303e-01 7.56356239e-01
7.22593188e-01 4.10273015e-01 3.83850783e-01 -4.81324971e-01
6.17484272e-01 1.05169892e+00 4.05108154e-01 -1.10025370e+00
-4.26923335e-01 -2.49927282e-01 -3.46882135e-01 3.28484982e-01
6.03948951e-01 2.12543149e-04 -7.19982445e-01 1.50635970e+00
3.17201912e-01 -1.44817963e-01 -5.70869207e-01 9.28935826e-01
4.34133053e-01 3.00250709e-01 3.14646363e-01 -1.93053514e-01
1.64274764e+00 -1.01616192e+00 -8.71609926e-01 -1.93208843e-01
7.93945074e-01 -8.52710426e-01 1.82750058e+00 1.15190327e-01
-8.34073782e-01 -5.84460676e-01 -8.56340170e-01 -4.08215582e-01
-6.45839214e-01 3.82618427e-01 3.75039726e-01 9.89518881e-01
-1.10196435e+00 2.93506831e-01 -6.03124201e-01 -7.71972656e-01
6.05141461e-01 4.78390753e-01 -1.71911359e-01 2.99672455e-01
-7.33600795e-01 8.65246534e-01 -5.87659217e-02 1.98997587e-01
-1.83473304e-01 -8.75184655e-01 -6.76116526e-01 5.08487046e-01
3.51344645e-01 -5.14007330e-01 1.09965360e+00 -1.15346587e+00
-1.44781470e+00 7.94648528e-01 -5.94965696e-01 5.96726127e-02
3.16934139e-01 -4.35618877e-01 -3.48652035e-01 -1.28191516e-01
-2.29243502e-01 4.95792091e-01 8.60484660e-01 -1.14266324e+00
-1.10977912e+00 -7.30920017e-01 1.62241980e-01 1.33312762e-01
-9.38055992e-01 1.61841556e-01 -7.07489133e-01 -1.71636552e-01
-3.40072274e-01 -8.97580981e-01 2.95021147e-01 -1.63711961e-02
-6.23482525e-01 -4.83926922e-01 8.85083318e-01 -9.18739140e-01
1.98826063e+00 -2.11402082e+00 -8.59807655e-02 2.06878290e-01
5.52733898e-01 1.81499779e-01 8.71903449e-02 2.42505655e-01
-2.49434978e-01 4.23541427e-01 4.92639631e-01 -5.21733284e-01
8.46694335e-02 -5.51145434e-01 -8.04365426e-02 4.15343165e-01
2.26197198e-01 1.04201806e+00 -4.77529705e-01 -7.24566996e-01
2.54281431e-01 4.23279405e-01 -6.46770120e-01 2.69711733e-01
-9.88616347e-02 3.05846274e-01 -4.55897719e-01 6.04077995e-01
5.94564259e-01 -7.44055271e-01 -1.83862820e-01 -3.00506353e-01
-4.94681507e-01 2.87820399e-01 -7.51438498e-01 1.45221531e+00
-5.39647102e-01 9.99949694e-01 -3.95509839e-01 -2.25299507e-01
4.80130017e-01 -1.35396630e-01 3.71525437e-01 -1.15737712e+00
3.81548554e-01 -3.01816314e-01 5.79856932e-02 -9.13531721e-01
5.43699980e-01 5.71115434e-01 3.15679818e-01 1.01244760e+00
-2.25734204e-01 9.06548798e-01 2.30804861e-01 -4.97502349e-02
7.20452428e-01 3.86794806e-01 2.26284638e-01 -3.62949401e-01
7.04983830e-01 -1.85421184e-01 -2.56604433e-01 5.35736024e-01
-4.73524809e-01 4.72358137e-01 4.59053546e-01 -4.26474661e-01
-1.00109673e+00 -3.49091530e-01 -2.62011196e-02 1.76378930e+00
1.26819193e-01 -4.76667583e-01 -1.12638438e+00 -9.36756134e-01
-2.68333286e-01 8.57141137e-01 -1.01455557e+00 -1.40408933e-01
-4.36273307e-01 -7.36029983e-01 -1.68828174e-01 2.74171352e-01
2.67071337e-01 -1.04818785e+00 -8.24786127e-01 -4.63946670e-01
-1.25095472e-01 -6.93370998e-01 -7.62387991e-01 -1.96579784e-01
-4.74495739e-01 -1.33250201e+00 -6.81387365e-01 -5.51586986e-01
6.61234081e-01 7.05539227e-01 9.10703659e-01 3.01343918e-01
-1.03766300e-01 2.63645798e-01 -3.79976451e-01 -5.25812328e-01
1.40496477e-01 5.69565594e-01 -3.71446133e-01 4.50951904e-01
9.58087325e-01 -2.49057427e-01 -9.76828456e-01 3.38146806e-01
-6.74111187e-01 1.85563922e-01 6.32995725e-01 4.15046960e-01
-5.51879453e-03 -9.83897075e-02 -4.23936620e-02 -8.62915158e-01
8.29587698e-01 -5.53139925e-01 -5.86335659e-01 5.79967260e-01
-7.81589270e-01 8.69713575e-02 2.97295958e-01 -2.07494885e-01
-1.17882192e+00 -1.70046344e-01 8.80431682e-02 -1.93050236e-01
-2.63401777e-01 3.51564944e-01 -1.55031607e-01 -2.51392007e-01
9.07233119e-01 1.13483816e-01 6.95126429e-02 -6.84647262e-01
1.53056011e-01 1.25237000e+00 -1.18678100e-01 -1.58994142e-02
6.05941117e-01 2.74715334e-01 -2.06405997e-01 -7.46173024e-01
-1.00604892e+00 -6.27641261e-01 -8.72867584e-01 -4.93236452e-01
7.07904160e-01 -6.35215938e-01 -1.33931065e+00 3.70431513e-01
-9.02527034e-01 -4.62202579e-02 4.49489504e-01 9.48089287e-02
-2.04570293e-01 2.52032131e-01 -7.32938424e-02 -1.02362645e+00
-4.04543787e-01 -1.13387430e+00 1.23482907e+00 5.01545668e-01
-9.75962654e-02 -1.11036360e+00 -7.10203946e-02 5.26797414e-01
4.72604692e-01 -2.18681574e-01 8.15239906e-01 -7.39263952e-01
-9.34820533e-01 -2.35476062e-01 -6.09605968e-01 -6.03202060e-02
3.23346734e-01 2.73384005e-01 -1.18149412e+00 -1.04148194e-01
-1.93544120e-01 1.03592739e-01 5.46004593e-01 3.79014015e-01
1.30171812e+00 -1.66133538e-01 -5.96746445e-01 6.18937671e-01
1.38505185e+00 -1.30573928e-01 8.22025418e-01 8.02794695e-01
1.01751423e+00 8.31651688e-01 7.46705174e-01 5.42159975e-01
7.28407443e-01 8.59204590e-01 6.77074611e-01 -1.08321592e-01
3.59629840e-02 1.00637101e-01 2.12827343e-02 2.26854965e-01
-2.74121106e-01 -4.28351283e-01 -1.12798274e+00 3.71738315e-01
-1.71114695e+00 -9.78710949e-01 -3.56992722e-01 2.35422015e+00
6.47156715e-01 -1.38088837e-02 4.01666373e-01 -6.25694096e-02
7.66375005e-01 8.58458132e-02 -4.22251225e-01 -1.44200390e-02
5.16440451e-01 -1.88763410e-01 3.97368342e-01 3.01925004e-01
-1.26775134e+00 6.99834585e-01 6.04207516e+00 4.45863366e-01
-1.31882560e+00 4.58888821e-02 6.81312263e-01 -3.94327164e-01
-2.19174936e-01 -4.01862353e-01 -1.18642163e+00 9.82901394e-01
1.18904173e+00 3.05167306e-02 7.35263109e-01 7.52380788e-01
6.27738833e-01 -1.98111802e-01 -1.10856891e+00 1.12930954e+00
3.05293262e-01 -1.13699281e+00 -2.29104415e-01 4.31984872e-01
2.35630527e-01 -2.76223153e-01 3.62072051e-01 1.72729090e-01
-2.67507523e-01 -8.57572794e-01 6.12007558e-01 6.25403166e-01
6.78090870e-01 -4.52483624e-01 4.51723993e-01 3.15901101e-01
-7.18468308e-01 -4.68912482e-01 4.88766804e-02 2.79357694e-02
-7.81869590e-02 1.05321281e-01 -8.88917327e-01 9.88110676e-02
1.09309411e+00 7.78819740e-01 -1.20130646e+00 1.23605299e+00
6.82274625e-02 3.19249421e-01 -1.50519684e-01 -2.72234023e-01
-3.31712402e-02 -9.39105004e-02 1.37897208e-01 8.67417991e-01
2.96103567e-01 -3.04287821e-01 -2.84619480e-01 9.18467939e-01
-8.30803439e-03 2.89126337e-01 -2.28953287e-01 3.84422462e-03
2.66896427e-01 1.45490289e+00 -3.91705990e-01 1.49604470e-01
-6.85316741e-01 6.33481622e-01 6.26789868e-01 4.88796473e-01
-1.05237198e+00 -4.32975620e-01 7.83545732e-01 6.05824888e-01
3.13318282e-01 1.34283686e-02 -6.09332561e-01 -1.02461171e+00
7.22540468e-02 -6.92378640e-01 1.61785975e-01 -9.80440259e-01
-7.76780009e-01 5.22154748e-01 -4.34052885e-01 -1.13904238e+00
-1.12329416e-01 -8.56270492e-01 -5.41318178e-01 1.24101818e+00
-1.79749894e+00 -1.41361129e+00 -9.52828348e-01 7.08978236e-01
5.29341936e-01 4.57834303e-02 4.01700914e-01 3.80924582e-01
-1.21512043e+00 7.70932972e-01 -1.74980350e-02 -3.61812152e-02
9.08186615e-01 -1.34155846e+00 3.05440813e-01 9.41348791e-01
-6.73577487e-02 8.97213519e-01 7.67260015e-01 -3.94226640e-01
-1.35296106e+00 -8.18495750e-01 9.84820187e-01 -9.77388501e-01
6.47901475e-01 -3.00880522e-01 -8.14666867e-01 8.80278945e-01
5.96173584e-01 -4.29434389e-01 9.18975294e-01 5.75195551e-01
5.60761206e-02 -1.45874977e-01 -5.90949774e-01 1.04758489e+00
8.17330420e-01 -5.05473852e-01 -3.90734613e-01 2.61748701e-01
4.08034503e-01 -1.87994465e-01 -5.70646882e-01 -1.32412344e-01
6.75479054e-01 -1.13678586e+00 8.51468682e-01 -6.28924906e-01
4.59212959e-01 -1.58409029e-01 3.61618638e-01 -1.04302692e+00
-5.19075334e-01 -6.26111925e-01 -2.50120372e-01 1.12844968e+00
3.06396186e-01 -3.80118519e-01 7.64791906e-01 1.01621330e+00
1.30755141e-01 -4.42687750e-01 -2.28788480e-01 1.93483204e-01
-6.32165551e-01 -2.51960576e-01 6.86020613e-01 4.97443259e-01
7.46898353e-02 5.63274205e-01 -3.23350638e-01 1.81026444e-01
4.68778670e-01 1.76103357e-02 1.06675303e+00 -1.44894123e+00
2.83635944e-01 -6.43752515e-01 -1.37405582e-02 -1.06900465e+00
-3.66125032e-02 -1.97460756e-01 -1.68536827e-01 -1.11549306e+00
3.14921886e-01 -2.13556811e-01 -5.62110424e-01 4.06823546e-01
-6.99633956e-01 2.34467760e-01 2.64299899e-01 5.20624995e-01
-9.93093431e-01 1.88895643e-01 1.28052139e+00 2.21994445e-01
-4.40869391e-01 2.88186133e-01 -1.20219350e+00 3.39999795e-01
4.79107469e-01 -9.27816778e-02 -3.11151803e-01 -3.01003307e-01
5.26470482e-01 -4.80770677e-01 3.39342922e-01 -7.13824809e-01
3.96446556e-01 -2.12812752e-01 5.97376764e-01 -6.46648586e-01
8.28340724e-02 -9.92274463e-01 -4.44318861e-01 2.77635437e-02
-4.41025734e-01 -4.32638265e-03 1.42342582e-01 6.36303842e-01
4.70485650e-02 -2.41796836e-01 4.97267932e-01 1.52671278e-01
-7.87561417e-01 1.23642974e-01 1.11316442e-01 -3.75988722e-01
1.10175657e+00 -5.23967326e-01 -5.68607032e-01 -2.76820332e-01
-5.61109781e-01 1.17610864e-01 6.41415536e-01 6.87081993e-01
9.13721174e-02 -9.32427168e-01 -1.69513464e-01 5.40015519e-01
3.50404799e-01 -4.22024667e-01 3.83325338e-01 1.18497121e+00
-5.25833130e-01 8.18303883e-01 -4.10168082e-01 -8.47311795e-01
-1.42105711e+00 9.25427973e-01 2.49086052e-01 -8.28203931e-02
-1.25535876e-01 5.54321110e-01 4.63607907e-01 -4.39939322e-04
2.97529429e-01 -2.85714358e-01 -8.95003736e-01 1.34929433e-01
9.65269387e-01 3.50129426e-01 2.74684399e-01 -6.58470690e-01
-1.83691666e-01 4.84165668e-01 -2.63372868e-01 4.25678283e-01
1.27861738e+00 -8.82026970e-01 5.08587388e-03 1.44346833e-01
1.03540146e+00 2.17316225e-01 -1.47200906e+00 -8.89715403e-02
2.47859955e-01 -9.72025931e-01 3.44187349e-01 -8.86642396e-01
-8.67605209e-01 1.02801943e+00 8.14874887e-01 6.15738094e-01
1.30579400e+00 -1.85374677e-01 5.46780586e-01 1.17586233e-01
1.17260776e-02 -8.60246778e-01 7.11846054e-02 2.42291912e-01
6.79041684e-01 -1.74562991e+00 -1.21623233e-01 -3.01752239e-01
-6.15895212e-01 1.08732009e+00 8.29105735e-01 3.15956026e-01
7.80358315e-01 -3.35613728e-01 1.69275418e-01 -2.33664513e-01
-7.95976579e-01 -5.32071173e-01 8.35424662e-01 4.15491909e-01
6.14427745e-01 -3.70046318e-01 5.67701012e-02 5.83957076e-01
-1.09166823e-01 5.59670851e-03 2.27490112e-01 7.38164783e-01
-5.37837744e-01 -9.06676292e-01 -5.21950364e-01 5.41311800e-01
-6.06110215e-01 -2.57174730e-01 -1.61743045e-01 8.19167495e-01
-1.95885718e-01 8.61856401e-01 2.56276876e-01 -2.87539423e-01
2.65686184e-01 2.02080101e-01 1.67187825e-01 -5.09377837e-01
-6.57697499e-01 9.80649292e-02 -2.61651546e-01 -6.61273897e-01
-2.94903457e-01 -6.62876844e-01 -4.46279496e-01 -5.43449223e-01
-5.48828483e-01 -4.44990277e-01 8.96919668e-01 9.91638839e-01
8.30878258e-01 6.22869968e-01 6.02101743e-01 -1.05051959e+00
-3.31579953e-01 -1.15017688e+00 -2.97519177e-01 7.71638751e-01
4.72423226e-01 -8.28725100e-01 -3.22118044e-01 3.37949336e-01] | [14.094232559204102, 0.08388006687164307] |
bba49847-220c-44de-af0c-e1906ad2298a | lets-take-this-online-adapting-scene | 1906.08744 | null | https://arxiv.org/abs/1906.08744v1 | https://arxiv.org/pdf/1906.08744v1.pdf | Let's Take This Online: Adapting Scene Coordinate Regression Network Predictions for Online RGB-D Camera Relocalisation | Many applications require a camera to be relocalised online, without expensive offline training on the target scene. Whilst both keyframe and sparse keypoint matching methods can be used online, the former often fail away from the training trajectory, and the latter can struggle in textureless regions. By contrast, scene coordinate regression (SCoRe) methods generalise to novel poses and can leverage dense correspondences to improve robustness, and recent work has shown how to adapt SCoRe forests between scenes, allowing their state-of-the-art performance to be leveraged online. However, because they use features hand-crafted for indoor use, they do not generalise well to harder outdoor scenes. Whilst replacing the forest with a neural network and learning suitable features for outdoor use is possible, the techniques used to adapt forests between scenes are unfortunately harder to transfer to a network context. In this paper, we address this by proposing a novel way of leveraging a network trained on one scene to predict points in another scene. Our approach replaces the appearance clustering performed by the branching structure of a regression forest with a two-step process that first uses the network to predict points in the original scene, and then uses these predicted points to look up clusters of points from the new scene. We show experimentally that our online approach achieves state-of-the-art performance on both the 7-Scenes and Cambridge Landmarks datasets, whilst running in under 300ms, making it highly effective in live scenarios. | ['Stuart Golodetz', 'Philip Torr', 'Tommaso Cavallari', 'Jishnu Mukhoti', 'Luca Bertinetto'] | 2019-06-20 | null | null | null | null | ['camera-relocalization'] | ['computer-vision'] | [ 4.17373687e-01 -1.88932851e-01 -1.28138736e-01 -3.76194715e-01
-7.94680357e-01 -6.92267179e-01 6.43832147e-01 -4.53055371e-03
-5.74238896e-01 3.55879486e-01 -7.71306902e-02 -3.16075295e-01
-2.08777152e-02 -7.07270741e-01 -8.34408998e-01 -5.24545670e-01
-9.08648297e-02 5.07247686e-01 7.46460557e-01 -7.83135965e-02
2.84707755e-01 9.35498297e-01 -1.62529230e+00 5.51954433e-02
2.90024906e-01 9.30330873e-01 2.42758512e-01 9.07414973e-01
3.75847444e-02 6.73758268e-01 -1.85461074e-01 -2.45394677e-01
7.87389040e-01 -3.02613080e-01 -7.76819289e-01 3.05230230e-01
9.34639215e-01 -1.46017060e-01 -4.60543275e-01 5.54191172e-01
3.23625892e-01 4.58641112e-01 3.36113036e-01 -1.07272303e+00
9.89391655e-03 -1.59647614e-01 -5.77024043e-01 8.95061865e-02
5.76982975e-01 1.89374968e-01 7.87802100e-01 -8.55441153e-01
7.72849739e-01 8.53283107e-01 1.03725910e+00 2.72031367e-01
-1.38246930e+00 -7.42109358e-01 2.85146922e-01 2.28564307e-01
-1.36700833e+00 -8.25842142e-01 5.95057905e-01 -2.68335611e-01
1.04354608e+00 2.12535247e-01 9.90872979e-01 6.60814166e-01
-4.65727411e-02 5.09563625e-01 1.07775640e+00 -4.51119363e-01
2.68798053e-01 -1.32531852e-01 -5.14214456e-01 7.41046369e-01
-1.90694794e-01 3.29506427e-01 -5.90035975e-01 9.41180158e-03
9.03462112e-01 2.21407831e-01 -1.04677543e-01 -1.05186427e+00
-1.26346517e+00 7.49277830e-01 9.79025066e-01 9.36321542e-03
-2.92559236e-01 3.59259367e-01 1.92393199e-01 3.33121687e-01
3.60842496e-01 3.96331728e-01 -5.13043523e-01 -1.80512726e-01
-1.37302840e+00 1.29511699e-01 7.96849132e-01 9.40293193e-01
1.21312261e+00 -3.53037775e-01 4.40578133e-01 5.42775989e-01
1.33942604e-01 3.73874843e-01 2.05195889e-01 -1.24571824e+00
2.87593126e-01 3.91922474e-01 -1.67828664e-01 -1.12739384e+00
-4.13869590e-01 -1.68669641e-01 -5.42549729e-01 5.40691912e-01
5.84384859e-01 8.47740322e-02 -1.11608696e+00 1.29768586e+00
7.06857920e-01 4.18558180e-01 -2.23177612e-01 6.66152656e-01
1.86915234e-01 4.99591500e-01 -2.23648563e-01 7.44636804e-02
7.41034567e-01 -1.21498156e+00 2.19779119e-01 -6.08802021e-01
4.75590616e-01 -9.50825989e-01 8.00703168e-01 4.32247877e-01
-9.35320616e-01 -4.74403024e-01 -8.82746279e-01 -1.47048697e-01
-3.76481473e-01 -4.93777007e-01 7.54246116e-01 3.48090112e-01
-1.26204777e+00 8.31124604e-01 -1.04493392e+00 -9.47752655e-01
4.30256188e-01 6.47698283e-01 -6.48033023e-01 -3.88134718e-01
-4.43217516e-01 1.02705312e+00 2.30832756e-01 1.58997942e-02
-5.52081048e-01 -6.08082414e-01 -8.32197964e-01 -3.48677129e-01
4.81866837e-01 -8.87901068e-01 1.24080181e+00 -1.27898645e+00
-1.68104064e+00 7.20024049e-01 -2.77349830e-01 -3.79149407e-01
6.20602667e-01 -2.97845632e-01 -5.60981408e-03 1.97318256e-01
1.51504204e-01 9.10852134e-01 8.86162162e-01 -1.17033815e+00
-9.79651511e-01 -2.43620068e-01 2.51524568e-01 4.50447857e-01
1.41393512e-01 -7.12609738e-02 -7.55033910e-01 -3.16798806e-01
4.41302657e-01 -1.14739931e+00 -5.68175673e-01 4.25137430e-01
-1.24308104e-02 2.01155394e-01 8.71307790e-01 -4.77428317e-01
6.44526899e-01 -2.06134605e+00 5.64078130e-02 6.63964391e-01
-1.34360701e-01 4.02186401e-02 -3.95699143e-02 4.29502279e-01
-1.14817743e-03 -2.91234434e-01 -1.70103341e-01 -5.05169511e-01
-3.18108857e-01 3.21688175e-01 -1.55665666e-01 7.12357521e-01
-7.20989471e-03 6.06898665e-01 -9.93344665e-01 -3.53939921e-01
5.83748758e-01 4.74986851e-01 -7.42177248e-01 8.01519677e-02
-3.69711639e-03 5.74513972e-01 -1.21239178e-01 5.80446005e-01
3.80486131e-01 -7.42845312e-02 4.61883657e-02 8.95970035e-03
-1.51667789e-01 1.56665117e-01 -1.51074958e+00 2.12038279e+00
-7.22103715e-01 5.83920479e-01 1.66923687e-01 -6.93215251e-01
7.62850285e-01 -2.05239818e-01 4.92883474e-01 -4.19700384e-01
-2.37794727e-01 2.50904769e-01 -2.80986160e-01 -1.52109578e-01
4.97916132e-01 -2.42564067e-01 1.01393498e-01 2.88334399e-01
7.76073383e-03 -5.10154486e-01 -6.62464350e-02 1.22883223e-01
1.43366086e+00 7.16263652e-01 4.26437944e-01 1.73291102e-01
2.21867904e-01 3.23740900e-01 2.91898370e-01 7.23738492e-01
9.47805471e-04 9.05494869e-01 -1.57127514e-01 -8.35514486e-01
-1.18384826e+00 -1.03563714e+00 5.69908619e-02 1.16034937e+00
2.93975174e-01 -5.09963334e-01 -3.63513350e-01 -8.72590363e-01
-4.41400968e-02 4.04223114e-01 -4.21699345e-01 6.57713264e-02
-6.72907889e-01 -2.65858412e-01 1.27433121e-01 6.68991089e-01
5.58568835e-01 -8.28823686e-01 -8.84399831e-01 3.66565764e-01
6.66307211e-02 -1.03087986e+00 -3.78862381e-01 5.55057824e-01
-7.79039323e-01 -1.11804998e+00 -4.67095733e-01 -5.60776711e-01
8.13600898e-01 6.17238879e-01 1.06161666e+00 2.19607815e-01
-1.96960881e-01 7.05725729e-01 -4.01735514e-01 -1.56620786e-01
-1.14954539e-01 2.56277651e-01 7.89301395e-02 -9.79154482e-02
1.83872044e-01 -7.97502875e-01 -7.02898204e-01 4.13081795e-01
-6.63745463e-01 1.72574773e-01 6.85369730e-01 7.83605158e-01
7.36025333e-01 -1.45933375e-01 -1.32111773e-01 -7.98609436e-01
-2.19638065e-01 -2.83279955e-01 -6.66329741e-01 4.45597712e-03
-4.68824953e-01 -4.16628681e-02 6.50222421e-01 -2.97136486e-01
-5.65446615e-01 8.84242237e-01 -1.24514259e-01 -5.89716256e-01
-3.31074297e-01 2.42453694e-01 9.02226716e-02 -6.27063096e-01
7.64398575e-01 -5.40346699e-03 -5.13822287e-02 -3.18672657e-01
6.26032889e-01 3.80567640e-01 7.32199967e-01 -5.38316250e-01
1.27341473e+00 7.53951609e-01 2.23709524e-01 -5.99423409e-01
-9.33259428e-01 -8.88412893e-01 -1.32548451e+00 -2.44592577e-01
6.81326747e-01 -8.76309335e-01 -2.59908468e-01 9.37050134e-02
-8.20571780e-01 -5.93216598e-01 -5.57588100e-01 2.83514887e-01
-8.51493895e-01 3.33797902e-01 -1.06639132e-01 -5.63736975e-01
8.12622607e-02 -9.28094983e-01 1.20032835e+00 1.37680903e-01
-2.76595980e-01 -8.65750909e-01 1.66634589e-01 7.03688413e-02
4.09588575e-01 4.00945693e-01 3.09106022e-01 -3.76759410e-01
-8.91922712e-01 -5.59309781e-01 -8.56665671e-02 -2.17683874e-02
-4.91835400e-02 -5.30861877e-02 -1.05627239e+00 -4.49920148e-01
-2.53121346e-01 -2.12983787e-01 8.49760175e-01 9.60344225e-02
1.01392353e+00 -1.20571084e-01 -4.94043469e-01 1.06753087e+00
1.48492503e+00 -1.45293951e-01 5.82066476e-01 9.69794571e-01
7.74514556e-01 5.39131761e-01 6.05334342e-01 2.19633263e-02
5.23490131e-01 8.80000234e-01 5.16397715e-01 -2.97142565e-01
-5.60886115e-02 -4.91723508e-01 3.59537631e-01 5.83197951e-01
-2.53250748e-01 4.02920812e-01 -1.03141832e+00 4.56019282e-01
-1.99650145e+00 -9.51646686e-01 1.45322502e-01 2.52761650e+00
5.58586359e-01 1.95958063e-01 2.49672949e-01 -9.75853875e-02
3.58606011e-01 1.04100555e-01 -6.02501154e-01 -2.33826354e-01
2.02578723e-01 4.39908504e-01 9.14441288e-01 4.88720238e-01
-1.25584161e+00 1.02451169e+00 6.15325642e+00 5.74598014e-01
-1.17884672e+00 6.10819198e-02 5.20106733e-01 -2.25978956e-01
2.03121260e-01 6.85223699e-01 -5.60568690e-01 1.42667098e-02
9.80927110e-01 2.70815581e-01 7.41323471e-01 1.13155985e+00
-1.17430231e-02 -3.93638104e-01 -1.26439643e+00 1.04444194e+00
6.44969419e-02 -1.25142586e+00 -3.47153336e-01 1.05372004e-01
6.37124896e-01 4.22822505e-01 -2.71434098e-01 1.53396711e-01
5.11894882e-01 -1.10048687e+00 7.75513828e-01 5.07318199e-01
8.38781059e-01 -7.97183812e-01 4.71831560e-01 4.85245377e-01
-1.32396388e+00 -8.56447816e-02 -4.76262152e-01 -2.14497223e-01
-1.82120353e-02 2.78373569e-01 -1.01506913e+00 5.11722684e-01
9.58157063e-01 7.94884026e-01 -8.67945850e-01 1.50173199e+00
-2.29853183e-01 2.45574489e-01 -7.89801836e-01 2.01996997e-01
2.42813185e-01 -3.17514800e-02 3.51950675e-01 1.14620328e+00
3.97437423e-01 -2.25780487e-01 5.55791557e-01 2.42903113e-01
1.68879628e-01 2.65560318e-02 -9.26582992e-01 5.38325846e-01
3.41836005e-01 1.41725564e+00 -1.03986883e+00 -2.46831000e-01
-4.67181563e-01 1.16718841e+00 4.83200371e-01 2.69287586e-01
-5.14067650e-01 -3.58041167e-01 4.50222313e-01 5.01625896e-01
6.61841691e-01 -5.14411032e-01 -5.27535118e-02 -1.16613829e+00
-8.96403939e-02 -8.15766752e-01 3.00172508e-01 -8.47516000e-01
-1.19388294e+00 4.56172615e-01 9.99955982e-02 -1.36853945e+00
-3.87577564e-01 -4.08355802e-01 -5.43131948e-01 6.55245841e-01
-1.54979908e+00 -1.50923789e+00 -4.64735061e-01 7.98029244e-01
6.70300007e-01 1.88609138e-01 8.67331147e-01 -1.94998346e-02
-8.49438906e-02 2.96665996e-01 2.34626979e-01 -5.74494898e-02
7.87729084e-01 -1.33135796e+00 6.01815283e-01 1.04908800e+00
6.34025633e-01 5.66059053e-01 5.39021611e-01 -4.82757777e-01
-1.43704104e+00 -1.06637788e+00 6.24525964e-01 -6.90901458e-01
6.06941879e-01 -5.42675018e-01 -5.78957558e-01 7.85287380e-01
5.50691895e-02 4.20006663e-01 4.26041961e-01 3.41289848e-01
-5.02021432e-01 -1.54463097e-01 -9.38002408e-01 6.91729128e-01
1.23362792e+00 -5.60651183e-01 -3.40309739e-01 4.54283267e-01
3.68996650e-01 -7.24561632e-01 -7.18657076e-01 1.60742357e-01
6.45861447e-01 -1.11970639e+00 1.16727352e+00 -3.19984347e-01
1.07127108e-01 -6.25717700e-01 -3.16316873e-01 -1.33024752e+00
-4.26421255e-01 -7.09562838e-01 -3.02578341e-02 9.32531118e-01
2.67796308e-01 -5.33612370e-01 1.01182604e+00 6.94248974e-01
-5.21184728e-02 -6.64763272e-01 -1.06643879e+00 -6.83583379e-01
-1.77387536e-01 -6.60670817e-01 4.98613328e-01 7.11710751e-01
-3.28478545e-01 3.27094883e-01 -2.40594044e-01 1.34899110e-01
4.27370667e-01 1.49008766e-01 1.48396015e+00 -1.11372507e+00
-5.23678243e-01 -2.34176219e-01 -7.99247503e-01 -1.25061560e+00
2.96235122e-02 -8.53282332e-01 2.16508344e-01 -1.44707012e+00
1.75330658e-02 -7.81832874e-01 1.54476753e-02 6.77829683e-01
-1.92504190e-02 5.40540159e-01 4.99365062e-01 4.97886956e-01
-7.28529453e-01 1.54731289e-01 7.74545610e-01 6.29326627e-02
-2.16207549e-01 1.68635324e-01 -4.54175472e-01 9.88453031e-01
5.24819314e-01 -5.73406041e-01 -2.72857368e-01 -3.40662211e-01
2.51420975e-01 -2.25257829e-01 6.01428807e-01 -1.56219113e+00
6.59993291e-01 -1.23940721e-01 8.76920521e-01 -4.91795748e-01
5.65269232e-01 -1.21003830e+00 3.56973350e-01 1.56594843e-01
8.84842593e-03 3.12999219e-01 2.28614137e-01 8.79895389e-01
1.11470699e-01 -1.20311856e-01 6.72120571e-01 -4.15603280e-01
-9.15126264e-01 3.56758624e-01 -7.92602450e-02 -2.57616609e-01
1.05089092e+00 -8.90413225e-01 2.35073447e-01 -5.39357662e-01
-7.05047488e-01 2.03193873e-02 1.19620836e+00 2.05075830e-01
5.47395170e-01 -1.14826453e+00 -2.63215303e-01 3.53259534e-01
-5.02056023e-03 5.37666142e-01 -1.80413455e-01 9.07141984e-01
-7.75435686e-01 -5.91033623e-02 -6.99355304e-02 -9.97652531e-01
-1.21354234e+00 5.60642898e-01 4.11899060e-01 -1.64020956e-01
-9.44054246e-01 8.30591142e-01 1.14809126e-01 -6.87093377e-01
1.89109936e-01 -6.44051209e-02 2.84947008e-01 -1.50759622e-01
2.67118692e-01 8.51387680e-02 2.71900743e-01 -7.77974248e-01
-3.43077093e-01 8.90306413e-01 -2.45971251e-02 -1.46984488e-01
1.57098150e+00 -2.50121981e-01 -1.07984571e-02 4.99360174e-01
1.17896628e+00 2.05633283e-01 -1.62944412e+00 -3.41734707e-01
9.79015678e-02 -9.12757099e-01 -4.57655033e-03 -6.19902909e-01
-7.88689494e-01 6.17466033e-01 6.25945687e-01 1.93057097e-02
1.17221463e+00 -3.43469530e-02 7.80626714e-01 4.97520387e-01
7.53075600e-01 -1.06273603e+00 -3.03398576e-02 6.18715048e-01
5.25368333e-01 -1.20361543e+00 3.24632406e-01 -2.84570634e-01
-3.99064243e-01 1.51185262e+00 3.84180963e-01 -3.10311556e-01
6.12032115e-01 1.81193650e-01 1.44856021e-01 -1.78608954e-01
-3.38617533e-01 -2.50788271e-01 1.07174993e-01 8.45965087e-01
2.04923330e-03 -2.53939986e-01 6.62971258e-01 -3.01140040e-01
-4.51907098e-01 5.38098393e-03 2.70934492e-01 1.27585006e+00
-4.02887523e-01 -1.21786082e+00 -5.51790953e-01 5.80127716e-01
-1.58140689e-01 -4.66628671e-02 -3.92220646e-01 9.35675919e-01
1.46759287e-01 7.64624000e-01 4.90914099e-02 -4.15934324e-01
4.83430207e-01 8.45240727e-02 5.64225018e-01 -8.07288885e-01
-6.97387159e-01 9.02894288e-02 -4.80472073e-02 -1.08107853e+00
-8.43805611e-01 -1.00847101e+00 -8.20123792e-01 -3.68652433e-01
-4.70587641e-01 -1.49498969e-01 7.93386519e-01 9.54001486e-01
3.91901195e-01 8.74577388e-02 6.51672184e-01 -1.58296001e+00
-1.81832165e-01 -5.02298594e-01 -1.10757947e-01 2.18040392e-01
4.27294165e-01 -6.47588849e-01 -2.81567633e-01 2.03493476e-01] | [7.7704901695251465, -2.216183662414551] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.