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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
34408229-60a2-4b63-9658-eaf4c6628b0d | new-designs-on-mvdr-robust-adaptive | 1810.11360 | null | http://arxiv.org/abs/1810.11360v1 | http://arxiv.org/pdf/1810.11360v1.pdf | New Designs on MVDR Robust Adaptive Beamforming Based on Optimal Steering Vector Estimation | The robust adaptive beamforming design problem based on estimation of the
signal of interest steering vector is considered in the paper. In this case,
the optimal beamformer is obtained by computing the sample matrix inverse and
an optimal estimate of the signal of interest steering vector. The common
criteria to find the best estimate of the steering vector are the beamformer
output SINR and output power, while the constraints assume as little as
possible prior inaccurate knowledge about the signal of interest, the
propagation media, and the antenna array. Herein, a new beamformer output power
maximization problem is formulated and solved subject to a double-sided norm
perturbation constraint, a similarity constraint, and a quadratic constraint
that guarantees that the direction-of-arrival (DOA) of the signal of interest
is away from the DOA region of all linear combinations of the interference
steering vectors. In the new robust design, the prior information required
consists of some allowable error norm bounds, the approximate knowledge of the
antenna array geometry, and the angular sector of the signal of interest. It
turns out that the array output power maximization problem is a non-convex QCQP
problem with inhomogeneous constraints. However, we show that the problem is
still solvable, and develop efficient algorithms for finding globally optimal
estimate of the signal of interest steering vector. The results are generalized
to the case where an ellipsoidal constraint is considered, and sufficient
conditions for the global optimality are derived. In addition, a new quadratic
constraint on the actual signal steering vector is proposed in order to improve
the array performance. To validate our results, simulation examples are
presented, and they demonstrate the improved performance of the new robust
beamformers in terms of the output SINR as well as the output power. | [] | 2018-10-26 | null | null | null | null | ['robust-design'] | ['miscellaneous'] | [ 2.60763168e-01 1.08328521e-01 3.12574774e-01 7.55736604e-02
-5.51719189e-01 -7.89746284e-01 -2.29871318e-01 -4.22943741e-01
-1.59317940e-01 7.35038579e-01 4.26251531e-01 -1.63190439e-01
-6.89160824e-01 -5.49032032e-01 -4.90781724e-01 -1.39680827e+00
-2.00612515e-01 -2.12412700e-01 -3.78072321e-01 -2.38241464e-01
5.86463325e-02 7.09413290e-01 -8.76617074e-01 -4.22401309e-01
6.63891077e-01 1.50871944e+00 1.77041635e-01 5.65702438e-01
6.96364880e-01 -9.08231139e-02 -6.87735617e-01 3.70358974e-02
5.60293078e-01 -4.79133368e-01 1.97149277e-01 6.04434550e-01
2.04702571e-01 -7.51370341e-02 -2.54069299e-01 1.22979462e+00
7.95909762e-01 7.74567574e-02 4.40010726e-01 -1.10756767e+00
-1.90111846e-02 2.09661841e-01 -4.70377594e-01 -2.73162108e-02
2.06190303e-01 -3.20391685e-01 8.42517853e-01 -1.21356475e+00
2.26857185e-01 7.06736445e-01 5.16509831e-01 -2.36226134e-02
-9.02065694e-01 -7.53108382e-01 -1.00830965e-01 -3.80095318e-02
-1.96818614e+00 -6.52522564e-01 6.23152733e-01 -2.10606247e-01
4.59975339e-02 7.43693590e-01 5.00446260e-01 2.91544110e-01
1.31978124e-01 2.58869529e-01 6.01397514e-01 -5.61099589e-01
4.81609821e-01 2.41856147e-02 -1.79975435e-01 7.84230471e-01
6.51017249e-01 -2.62989588e-02 -3.95730525e-01 -4.19990599e-01
7.34905660e-01 -2.69630045e-01 -1.10094333e+00 -6.57911122e-01
-1.22732174e+00 6.56828225e-01 3.30142379e-01 4.77893531e-01
-4.75580722e-01 -1.03323743e-01 -3.92963409e-01 1.08514272e-01
2.81881183e-01 5.70514023e-01 -1.59452751e-01 4.65080112e-01
-8.94430757e-01 1.24245994e-01 9.05890703e-01 1.13116133e+00
4.07639951e-01 4.77097958e-01 -2.60134637e-01 7.20193744e-01
6.79972172e-01 1.24017227e+00 -2.77956039e-01 -1.00464463e+00
7.58488297e-01 1.68102346e-02 6.73418462e-01 -1.37448204e+00
-3.89545590e-01 -1.38548505e+00 -9.15397704e-01 -1.68167558e-02
5.97084939e-01 -9.22044754e-01 -4.43934172e-01 1.69449317e+00
1.92243159e-01 1.08849756e-01 6.55355975e-02 1.26035047e+00
1.86450675e-01 9.24137890e-01 -8.94840539e-01 -1.24734378e+00
9.45219040e-01 -2.87089348e-01 -1.04520833e+00 -3.52078885e-01
2.17465803e-01 -7.85127282e-01 8.22296888e-02 3.99680078e-01
-1.09530723e+00 3.45106646e-02 -1.25751853e+00 8.38062584e-01
4.20666993e-01 3.73261273e-01 3.83960381e-02 9.11790967e-01
-7.82987416e-01 -4.51222032e-01 -3.74834865e-01 1.18292637e-01
-2.58912712e-01 2.95014948e-01 -1.62559301e-01 -1.91836879e-01
-7.34380603e-01 5.11121333e-01 1.45983443e-01 7.28156149e-01
-1.91289946e-01 -7.59088576e-01 -6.73980713e-01 3.54741544e-01
3.91582847e-01 -6.09121919e-01 7.49611080e-01 -1.03628814e+00
-1.33257484e+00 -2.40994338e-02 -4.02461141e-01 7.12438524e-02
6.37054592e-02 -3.23035419e-02 -2.66686529e-01 5.70195280e-02
2.57314332e-02 -5.89034081e-01 8.08455050e-01 -1.39120555e+00
-7.25170910e-01 -3.83608699e-01 -2.37738147e-01 4.56491917e-01
-3.36267829e-01 -2.14001387e-01 -6.38784111e-01 -7.27944553e-01
5.93898356e-01 -9.65229928e-01 -5.66277623e-01 -1.89069331e-01
-4.16733116e-01 3.71787131e-01 2.13838562e-01 -4.99197900e-01
1.49748528e+00 -2.57320380e+00 4.35683131e-01 1.01966465e+00
-2.43976876e-01 -3.48059565e-01 -1.07019372e-01 5.69735408e-01
-8.90298039e-02 -3.56132090e-01 -2.12971017e-01 1.20208204e-01
-3.47168386e-01 -5.28616905e-02 -2.83736914e-01 1.11958086e+00
-4.85464394e-01 3.93399261e-02 -7.96424687e-01 8.83424059e-02
-1.05549276e-01 2.05742896e-01 -6.07802033e-01 4.31850731e-01
4.69468206e-01 3.73166025e-01 -5.99829197e-01 4.72817928e-01
8.78472328e-01 2.93153860e-02 3.13183933e-01 -6.73534989e-01
-5.49194992e-01 -5.68832517e-01 -1.97801113e+00 1.09212041e+00
-5.62285542e-01 4.68475252e-01 1.00451767e+00 -8.76442313e-01
6.75700545e-01 5.72964907e-01 7.21680462e-01 -4.39885944e-01
6.13451600e-02 2.98373461e-01 3.38459581e-01 -3.42302293e-01
-1.12439552e-02 -2.01721266e-01 2.21812632e-02 2.34894112e-01
-2.03833699e-01 -1.49010465e-01 2.39507303e-01 -7.88676068e-02
8.32997918e-01 -7.81550229e-01 6.21293485e-01 -6.04043245e-01
6.95934534e-01 -5.35054862e-01 8.16490233e-01 6.39847398e-01
4.37265754e-01 5.29679716e-01 2.18203187e-01 2.71445841e-01
-7.84928679e-01 -7.35126495e-01 -2.51730084e-01 7.35636294e-01
3.41224879e-01 8.83529112e-02 -3.89265180e-01 -4.40755002e-02
-6.12718351e-02 7.84887612e-01 -2.18165979e-01 3.15711498e-01
-5.24755001e-01 -9.74385381e-01 -2.45984375e-01 -8.63518417e-02
1.17198169e-01 1.93671599e-01 -3.49182844e-01 2.01417923e-01
-2.97567457e-01 -9.79818165e-01 -7.09589660e-01 9.04811174e-03
-4.98786062e-01 -1.01283610e+00 -8.95838201e-01 -6.81771636e-01
1.25670660e+00 5.21525860e-01 3.60502809e-01 -2.80559719e-01
2.84686655e-01 7.72106051e-01 -2.25391135e-01 -3.53968769e-01
1.50115013e-01 -5.64403832e-01 2.50254095e-01 7.71767557e-01
-9.67010617e-01 -4.59204584e-01 -6.44534111e-01 9.00771618e-01
-5.01978993e-01 -1.62633434e-01 5.82786560e-01 7.30668485e-01
4.05618042e-01 3.78066957e-01 5.27132928e-01 -2.07879916e-01
5.71671009e-01 -5.18781841e-01 -9.25048470e-01 8.58178213e-02
-4.06283259e-01 -9.70062017e-02 6.24677598e-01 -1.00490190e-01
-1.15726650e+00 1.74263030e-01 2.38583684e-02 -7.13234991e-02
3.28558534e-01 7.05446124e-01 -6.73352599e-01 -4.42059934e-01
3.89209390e-01 2.59783924e-01 -3.25751573e-01 -1.47919267e-01
3.02252144e-01 6.40834093e-01 3.32479030e-01 -7.65023604e-02
1.18981552e+00 5.33833981e-01 5.48512936e-01 -1.26673925e+00
-9.19072986e-01 -8.21304739e-01 -1.56477183e-01 -4.77189332e-01
2.27803051e-01 -7.83431053e-01 -5.74553549e-01 9.97509137e-02
-9.54288423e-01 2.86549032e-01 7.21955895e-02 8.80518377e-01
-4.74996030e-01 3.75539273e-01 1.82389066e-01 -1.16896319e+00
-2.22383618e-01 -9.63134110e-01 5.28294027e-01 -1.39437184e-01
8.93450007e-02 -7.90926158e-01 -1.26146749e-01 -1.83383524e-02
4.66992408e-01 2.93558657e-01 6.90549254e-01 -2.10836604e-01
-7.06154525e-01 -4.08656925e-01 -8.15869793e-02 1.86469913e-01
2.13409260e-01 -6.01249814e-01 -2.12034315e-01 -6.83641970e-01
3.92410785e-01 5.58361828e-01 5.29759564e-02 8.23226511e-01
7.20073223e-01 -8.01974595e-01 -4.87920582e-01 8.50546479e-01
1.75873435e+00 3.58530551e-01 2.08481058e-01 -2.09548194e-02
1.98020101e-01 2.02035144e-01 6.66691720e-01 9.21209395e-01
-1.27199188e-01 9.20158446e-01 7.13238537e-01 1.44384712e-01
3.08498234e-01 5.71141362e-01 9.16591287e-02 9.91029084e-01
-1.00332730e-01 -7.73150861e-01 -4.05686498e-01 4.91059661e-01
-1.81168616e+00 -7.64961362e-01 -2.78742433e-01 2.64836574e+00
3.89914632e-01 -4.98123854e-01 -3.09468806e-01 3.18335891e-01
6.54913366e-01 3.58345173e-02 -3.59227240e-01 -4.53913864e-03
-1.87452316e-01 -1.83799535e-01 1.03190911e+00 8.54067981e-01
-9.02394950e-01 -2.70971507e-01 6.28831339e+00 5.59401810e-01
-1.02150404e+00 -1.28798470e-01 -4.59746346e-02 -2.16211259e-01
-3.33745331e-01 -2.94334382e-01 -6.91034973e-01 4.07081276e-01
3.43427211e-01 -5.18481076e-01 5.06254494e-01 4.65339333e-01
6.34093583e-01 -1.98351249e-01 -8.08216751e-01 1.14068675e+00
3.25112671e-01 -9.66153562e-01 -4.89829451e-01 2.92722344e-01
7.21773803e-01 -5.32073915e-01 -2.40793615e-03 -3.79844189e-01
-2.81183392e-01 -4.86374944e-01 6.94011807e-01 7.50888944e-01
4.97276366e-01 -8.90463948e-01 8.96091402e-01 4.56849277e-01
-1.05779207e+00 -5.31298339e-01 -6.89311549e-02 -4.59203757e-02
4.22220916e-01 1.19509351e+00 -6.79004312e-01 7.00156331e-01
2.49848917e-01 2.88180232e-01 1.72513500e-01 1.74901223e+00
-3.39686424e-01 6.86804950e-01 -6.41612649e-01 -1.35058329e-01
2.09165681e-02 -4.78949726e-01 1.15792644e+00 1.11742079e+00
9.65435028e-01 7.13933468e-01 2.46253759e-01 2.79114872e-01
7.98123553e-02 4.55221176e-01 -2.81050742e-01 4.59752530e-01
6.46911144e-01 1.08369792e+00 -2.58816391e-01 1.98403910e-01
-5.13783455e-01 5.64739287e-01 -3.74425352e-01 1.01172578e+00
-3.89940292e-01 -5.72526753e-01 6.61625922e-01 7.62555301e-02
3.33467335e-01 -5.88305295e-01 -4.59813744e-01 -7.15426147e-01
2.29756132e-01 -7.15775430e-01 9.37843993e-02 -5.11201024e-01
-8.62078846e-01 1.61114022e-01 -1.05174996e-01 -1.50847328e+00
-3.67991269e-01 -3.76300931e-01 -5.24001956e-01 1.02663541e+00
-1.00009823e+00 -6.05236113e-01 5.72219901e-02 4.31814194e-01
1.72325611e-01 -3.10284674e-01 5.36171138e-01 2.79457897e-01
-4.78512377e-01 3.70279104e-01 6.89614713e-01 1.31408302e-02
2.10305974e-01 -7.66333282e-01 -9.64379966e-01 1.22374892e+00
-2.98528045e-01 6.15428984e-01 1.24648249e+00 -8.78033042e-02
-1.93776000e+00 -9.10830617e-01 6.10764205e-01 1.65017948e-01
6.73066735e-01 -2.57987052e-01 -1.42064586e-01 3.23811144e-01
2.89819807e-01 -6.39914796e-02 8.92794371e-01 -1.41168267e-01
3.80365968e-01 -3.80213857e-01 -6.97886348e-01 6.27819180e-01
7.10475862e-01 2.17066064e-01 -6.12877607e-02 5.72506368e-01
1.26188126e-04 -4.67405260e-01 -8.36326778e-01 6.30398393e-01
6.55183673e-01 -6.34825408e-01 9.46331263e-01 8.50606151e-03
-4.11320597e-01 -6.40569687e-01 -6.93102837e-01 -1.52868259e+00
-7.67137706e-01 -7.78895199e-01 3.16745192e-02 8.78009081e-01
5.83261788e-01 -3.97415102e-01 6.49023831e-01 2.66497165e-01
-2.24845052e-01 -8.31283987e-01 -1.31741023e+00 -7.32000053e-01
-6.73073530e-01 -2.79847383e-01 -1.23587839e-01 6.12056792e-01
2.06007883e-01 3.67327631e-01 -4.90768433e-01 1.21004844e+00
8.96647036e-01 2.31250033e-01 5.14675736e-01 -8.13697398e-01
-3.42047602e-01 -1.71826348e-01 -1.95136681e-01 -1.38460600e+00
-1.18885189e-01 -5.07918239e-01 3.64460051e-01 -1.62647462e+00
-2.71941900e-01 -4.89665329e-01 -9.78319049e-02 -1.12384751e-01
1.17734782e-01 9.13746506e-02 2.54681110e-01 -1.43642843e-01
-9.57541913e-02 4.03399885e-01 1.28459549e+00 -2.17784956e-01
-3.31732720e-01 1.00569320e+00 -6.27685666e-01 7.39620209e-01
4.46540266e-01 -2.39582792e-01 -4.76468742e-01 -5.57371497e-01
6.00193262e-01 6.08278573e-01 -1.49231091e-01 -8.97396147e-01
3.98869246e-01 -2.95871764e-01 2.00699255e-01 -5.31443298e-01
5.64132452e-01 -1.35358012e+00 3.37728232e-01 3.09954435e-01
2.67494991e-02 -3.94677460e-01 -1.11209340e-01 6.92800939e-01
-2.93039437e-02 -6.29501402e-01 8.32318604e-01 3.63161594e-01
-1.62430838e-01 1.11565061e-01 -7.61911392e-01 -3.25806737e-01
9.49282825e-01 -1.63466930e-01 2.71611035e-01 -1.01155233e+00
-7.92488039e-01 3.14491600e-01 -1.84378833e-01 -3.99687663e-02
6.05013907e-01 -1.29690754e+00 -9.15237308e-01 2.60246694e-01
-1.69554681e-01 -4.82495904e-01 -3.68032767e-03 1.00571489e+00
-3.76568623e-02 8.53159130e-01 3.04958999e-01 -4.48989451e-01
-1.30864131e+00 2.08365917e-01 5.37482917e-01 2.53885895e-01
2.11828589e-01 8.20340574e-01 2.34931871e-01 1.54080048e-01
2.90171832e-01 -1.54145792e-01 -1.28676832e-01 -1.16539046e-01
5.10886788e-01 5.73682964e-01 1.26905411e-01 -7.44692028e-01
-3.31090271e-01 7.74827421e-01 8.14277589e-01 -3.09018016e-01
1.02802849e+00 -6.34870708e-01 -3.70130420e-01 -1.36969775e-01
1.07688808e+00 1.08181775e+00 -7.60511994e-01 -3.11048359e-01
-4.13249642e-01 -7.69366324e-01 2.73110896e-01 -6.38205171e-01
-1.15648115e+00 2.67262876e-01 5.95454156e-01 1.23407997e-01
1.50156403e+00 -1.56058878e-01 3.22176740e-02 4.92581159e-01
6.57212138e-01 -7.23601401e-01 -2.33902767e-01 5.26325405e-01
1.43299961e+00 -4.55464453e-01 3.65936011e-01 -8.18988085e-01
4.45943139e-02 1.12422097e+00 2.14916438e-01 1.82346124e-02
9.82925475e-01 2.39616737e-01 -1.29492581e-01 2.82531410e-01
-2.02691630e-01 -2.17011854e-01 6.26683354e-01 4.01252806e-01
4.06286627e-01 1.00717552e-01 -6.49380386e-01 6.06911242e-01
-5.12453988e-02 -4.28142726e-01 5.90611637e-01 4.71223533e-01
-8.83249462e-01 -6.16082251e-01 -1.03972077e+00 2.77103603e-01
-3.86040688e-01 -4.39359061e-02 1.86389580e-01 3.08140397e-01
2.26860017e-01 1.33343601e+00 -1.53051302e-01 1.76282719e-01
7.59531140e-01 -3.98854673e-01 3.83404583e-01 -4.95189846e-01
2.20501751e-01 5.51733255e-01 2.86044151e-01 -3.37767512e-01
-1.55536011e-01 -5.10775685e-01 -8.37212324e-01 2.90987551e-01
-7.05405295e-01 6.38237000e-01 7.04221964e-01 9.44719136e-01
2.23859981e-01 2.33690992e-01 1.23092735e+00 -6.37619615e-01
-4.34860229e-01 -6.58643723e-01 -9.17809069e-01 -3.05398285e-01
4.59193945e-01 -2.82772899e-01 -3.52652043e-01 -5.45966614e-04] | [6.420071125030518, 1.3550186157226562] |
deb392c9-a682-4204-a0c5-68f5936c45a6 | full-transformer-framework-for-robust-point | 2112.09385 | null | https://arxiv.org/abs/2112.09385v1 | https://arxiv.org/pdf/2112.09385v1.pdf | Full Transformer Framework for Robust Point Cloud Registration with Deep Information Interaction | Recent Transformer-based methods have achieved advanced performance in point cloud registration by utilizing advantages of the Transformer in order-invariance and modeling dependency to aggregate information. However, they still suffer from indistinct feature extraction, sensitivity to noise, and outliers. The reasons are: (1) the adoption of CNNs fails to model global relations due to their local receptive fields, resulting in extracted features susceptible to noise; (2) the shallow-wide architecture of Transformers and lack of positional encoding lead to indistinct feature extraction due to inefficient information interaction; (3) the omission of geometrical compatibility leads to inaccurate classification between inliers and outliers. To address above limitations, a novel full Transformer network for point cloud registration is proposed, named the Deep Interaction Transformer (DIT), which incorporates: (1) a Point Cloud Structure Extractor (PSE) to model global relations and retrieve structural information with Transformer encoders; (2) a deep-narrow Point Feature Transformer (PFT) to facilitate deep information interaction across two point clouds with positional encoding, such that Transformers can establish comprehensive associations and directly learn relative position between points; (3) a Geometric Matching-based Correspondence Confidence Evaluation (GMCCE) method to measure spatial consistency and estimate inlier confidence by designing the triangulated descriptor. Extensive experiments on clean, noisy, partially overlapping point cloud registration demonstrate that our method outperforms state-of-the-art methods. | ['Li Yuan', 'Qingxiang Zhang', 'Yufeng Yue', 'Meiling Wang', 'Guangyan Chen'] | 2021-12-17 | null | null | null | null | ['geometric-matching'] | ['computer-vision'] | [-3.44515264e-01 -1.81607947e-01 1.29681513e-01 -2.42288873e-01
-9.70781863e-01 -2.61127174e-01 5.18177271e-01 1.62292242e-01
-3.69896591e-02 1.60328358e-01 -6.76606670e-02 5.03518991e-02
-3.96319956e-01 -9.66791272e-01 -1.00611806e+00 -5.09909451e-01
-2.03919351e-01 7.51211345e-01 2.82948494e-01 -3.84125084e-01
3.67117047e-01 1.04147112e+00 -1.60863245e+00 1.26874156e-03
1.12784481e+00 1.52541041e+00 4.50353511e-03 -2.49937773e-01
-4.41757776e-03 1.90079361e-01 -3.37972999e-01 -3.28106523e-01
5.81604004e-01 1.25940114e-01 -3.89130056e-01 -2.13414058e-01
7.19587743e-01 -4.06327188e-01 -3.73533279e-01 1.09408343e+00
3.79762620e-01 2.11787019e-02 6.96693897e-01 -1.32493711e+00
-7.23192811e-01 -8.48347470e-02 -6.44913733e-01 -2.44438097e-01
3.62209827e-01 3.38051766e-02 9.04255688e-01 -1.41262031e+00
3.40132326e-01 1.25336766e+00 1.32791662e+00 -1.34223104e-01
-1.02539062e+00 -9.32406068e-01 -1.40775740e-01 1.13072187e-01
-1.88764071e+00 -3.45033944e-01 9.65053380e-01 -3.15008312e-01
1.19398868e+00 2.98851192e-01 9.85222459e-01 6.99378252e-01
4.08177793e-01 2.31399804e-01 6.95405781e-01 6.91728527e-03
6.71872422e-02 -3.07507902e-01 -2.13629439e-01 7.19798625e-01
2.19683349e-01 5.75241804e-01 -5.09003401e-01 -3.59684616e-01
1.27990675e+00 3.06183904e-01 -1.79647595e-01 -5.36467016e-01
-1.34491026e+00 7.57968187e-01 1.12243009e+00 2.84710348e-01
-5.16461670e-01 1.29898399e-01 1.75563887e-01 3.78062099e-01
4.11848694e-01 4.36150223e-01 -4.22433823e-01 -1.17516536e-02
-9.21553016e-01 3.19135964e-01 4.75897312e-01 1.37391078e+00
1.17937374e+00 -2.60670260e-02 5.94879054e-02 8.10394526e-01
4.86690253e-01 6.95523679e-01 2.79760718e-01 -7.50980675e-01
6.87907219e-01 9.59486008e-01 -1.80671975e-01 -1.58537471e+00
-4.83090460e-01 -6.25401139e-01 -1.14782119e+00 3.13429147e-01
-3.43780108e-02 6.30998909e-01 -8.18207085e-01 1.26931942e+00
1.14693500e-01 4.72838074e-01 -3.03681821e-01 9.82177079e-01
9.72792685e-01 2.18212828e-01 -3.38915080e-01 1.32270887e-01
1.20202494e+00 -4.57777798e-01 -3.99144441e-01 9.34275761e-02
4.61115420e-01 -9.56231415e-01 6.96535230e-01 2.48594195e-01
-1.07001805e+00 -7.61050522e-01 -1.14630234e+00 -3.68505806e-01
-4.99326944e-01 6.66557625e-02 5.43500662e-01 1.04494058e-01
-1.06675625e+00 7.43224323e-01 -9.37688291e-01 -2.65224814e-01
5.46398997e-01 8.22041094e-01 -5.83348036e-01 -3.25815454e-02
-8.78653169e-01 8.14576626e-01 7.28140920e-02 5.58146656e-01
-2.67723113e-01 -9.65315938e-01 -1.17388999e+00 1.89421862e-01
-7.81361088e-02 -1.07038426e+00 6.39499485e-01 -5.63711226e-01
-1.38616431e+00 6.40802324e-01 -1.13975592e-01 -1.27293065e-01
5.81887782e-01 -2.89455384e-01 -1.44363895e-01 5.52981719e-03
4.25536126e-01 6.85808480e-01 7.13103533e-01 -1.34508884e+00
-4.45999265e-01 -7.26875305e-01 -3.09858203e-01 2.47455552e-01
2.13107956e-03 -2.52881318e-01 -4.86275434e-01 -8.37226331e-01
1.07652259e+00 -7.90997803e-01 -1.27813920e-01 1.29656360e-01
-3.55193704e-01 -2.45083645e-01 1.13117659e+00 -4.87263232e-01
8.66694272e-01 -2.28966355e+00 -1.66707709e-01 6.39107704e-01
3.49793375e-01 7.52342418e-02 -1.27440423e-01 4.79394108e-01
-2.03037977e-01 -1.17532209e-01 -8.95880833e-02 -4.61844087e-01
8.93902034e-04 2.93457359e-01 -5.19284904e-01 7.23127067e-01
3.31651211e-01 1.08005607e+00 -8.43951881e-01 -5.33287525e-01
7.45270073e-01 7.04688311e-01 -5.68142653e-01 3.36394273e-02
3.62276524e-01 5.04417837e-01 -7.06822336e-01 1.04209960e+00
1.24626756e+00 -1.02122635e-01 -6.81540906e-01 -9.13567960e-01
-2.06510887e-01 2.46298119e-01 -1.38185191e+00 1.82132661e+00
-3.43610257e-01 3.11861783e-01 -1.90096393e-01 -7.00954020e-01
1.26656711e+00 2.95932233e-01 8.55188191e-01 -7.49315143e-01
2.01128304e-01 5.26779473e-01 -3.25308293e-01 -1.10802045e-02
4.67489928e-01 1.17043354e-01 9.53484923e-02 -2.13839278e-01
1.08435005e-01 -4.41962391e-01 -5.46452761e-01 -1.16504520e-01
1.03362644e+00 2.02695251e-01 1.97644774e-02 -1.61155999e-01
5.19566655e-01 -1.58085153e-01 7.39207149e-01 5.31242430e-01
4.81497794e-02 1.06612194e+00 -1.71470523e-01 -7.46778727e-01
-8.90373230e-01 -1.18918931e+00 -4.89621639e-01 2.45577186e-01
5.42942822e-01 -5.09438455e-01 -1.31791830e-01 -2.72191942e-01
2.50549197e-01 2.17573002e-01 -3.70858997e-01 -2.57316262e-01
-7.39388168e-01 -2.82847375e-01 4.73222107e-01 7.62269139e-01
8.71471882e-01 -5.52802742e-01 -2.16078326e-01 1.78882211e-01
-2.53680348e-01 -1.16246331e+00 -3.42545837e-01 1.21504337e-01
-1.31717491e+00 -1.15551257e+00 -2.63322473e-01 -6.47667646e-01
8.60701680e-01 4.03013468e-01 1.06569219e+00 3.29940140e-01
2.00646862e-01 1.63799018e-01 -1.49906933e-01 -2.02909335e-01
1.77874461e-01 -3.74154001e-02 1.69779241e-01 -2.43216231e-01
4.39996421e-01 -9.10578549e-01 -7.30106950e-01 6.95799470e-01
-6.62068009e-01 -1.62290752e-01 7.04716682e-01 1.05702567e+00
8.26253951e-01 -8.18688795e-02 -2.22536357e-04 -3.24471831e-01
2.35484689e-01 -1.63047165e-01 -6.76932931e-01 1.40610086e-02
-5.74929833e-01 -2.18197107e-01 4.37605023e-01 -6.40645549e-02
-6.53617024e-01 4.38263230e-02 -2.79308736e-01 -9.38027322e-01
-2.53722095e-03 3.86495471e-01 -1.39944687e-01 -7.64982760e-01
3.44993770e-01 1.89020023e-01 4.42017950e-02 -4.59532708e-01
1.18598156e-01 3.12201947e-01 6.65553808e-01 -7.41533160e-01
1.21835423e+00 6.12724304e-01 3.52221161e-01 -6.55690253e-01
-4.22855884e-01 -4.61674303e-01 -1.07793880e+00 5.99818826e-02
5.85398078e-01 -1.19493747e+00 -9.57078993e-01 4.13614869e-01
-1.37841368e+00 4.18928832e-01 -2.61767119e-01 5.87575555e-01
-6.75511003e-01 4.65488076e-01 -6.86303973e-01 -4.26561087e-01
-4.75216538e-01 -1.42170846e+00 1.67438483e+00 1.13129858e-02
-1.89821012e-02 -6.66108370e-01 -1.77103534e-01 2.62241364e-01
3.61574948e-01 5.51166415e-01 5.54498613e-01 -3.86405557e-01
-1.16161489e+00 -4.69838023e-01 -4.38190907e-01 3.02623808e-01
1.98411494e-01 1.25990957e-01 -8.93797874e-01 -4.28879172e-01
1.01279043e-01 4.60959449e-02 3.98853213e-01 3.74410629e-01
1.14479113e+00 -1.45283133e-01 -6.28991723e-01 1.21518612e+00
1.37797153e+00 -1.55945778e-01 8.66862118e-01 4.48039681e-01
9.90112960e-01 2.51848519e-01 7.11941540e-01 1.48092479e-01
4.30291533e-01 8.59966457e-01 7.68968344e-01 -4.21763599e-01
6.22331537e-02 -3.83488804e-01 3.69595252e-02 1.20599341e+00
-4.53097939e-01 4.25819784e-01 -8.31951737e-01 3.56172204e-01
-1.93820739e+00 -6.62618160e-01 -3.42083007e-01 2.29724431e+00
2.96977341e-01 3.63411903e-02 -3.70911449e-01 -1.41081102e-02
3.81067485e-01 -7.73622245e-02 -2.71668017e-01 1.83601782e-01
-1.71177313e-01 3.66702944e-01 6.46023393e-01 1.60807475e-01
-9.80862975e-01 9.09272075e-01 5.33182716e+00 9.60075378e-01
-1.19608521e+00 4.22528200e-02 2.37918273e-01 4.24429446e-01
-1.01429075e-01 6.84528276e-02 -5.01942575e-01 3.59002113e-01
2.02756882e-01 1.99242979e-01 2.10309830e-02 8.70899439e-01
-1.34654492e-01 3.10744435e-01 -1.11862266e+00 1.38234949e+00
-1.51001275e-01 -1.28738976e+00 2.77244508e-01 3.50109369e-01
4.83328789e-01 4.67735142e-01 4.37302217e-02 9.96975303e-02
-2.14427095e-02 -1.01131833e+00 7.51798511e-01 6.49890900e-01
9.43259716e-01 -8.65772128e-01 9.89315271e-01 1.94464907e-01
-1.68543100e+00 1.08823463e-01 -7.28508830e-01 1.97397515e-01
-1.11161448e-01 5.66091478e-01 -5.90865254e-01 1.14300764e+00
9.56806719e-01 1.03074706e+00 -5.63381255e-01 1.16369510e+00
-4.89198463e-03 -2.88335569e-02 -8.53029191e-01 5.64199626e-01
2.26196229e-01 -5.09955645e-01 6.77184224e-01 7.92839587e-01
7.51097202e-01 6.41174614e-02 6.41588122e-02 1.21640670e+00
2.39425123e-01 -1.14598416e-01 -7.95741916e-01 6.38076007e-01
8.36268663e-01 1.19601214e+00 -6.77537680e-01 1.49239684e-02
-4.51254815e-01 8.33277643e-01 2.18571678e-01 3.55501920e-01
-5.77768505e-01 -2.51723111e-01 8.83784533e-01 1.77088410e-01
3.05168182e-01 -5.09358764e-01 -5.82290232e-01 -1.11540568e+00
3.28136891e-01 -5.89420259e-01 7.11695105e-02 -8.63210082e-01
-1.53249729e+00 9.27699089e-01 -5.10999113e-02 -1.98719406e+00
7.19112828e-02 -4.92877692e-01 -5.71328521e-01 9.97482896e-01
-1.46788001e+00 -1.55247903e+00 -6.09456956e-01 8.90497386e-01
1.75870895e-01 -8.25039670e-02 7.08230615e-01 5.95324099e-01
-1.46430194e-01 6.52131975e-01 -1.20292179e-01 3.37792248e-01
6.03737652e-01 -9.54339147e-01 4.97129202e-01 5.53492486e-01
8.44643936e-02 1.01714325e+00 3.05440933e-01 -6.89091027e-01
-1.66088068e+00 -1.09664774e+00 7.22732186e-01 -7.14936793e-01
3.28086138e-01 -3.22954595e-01 -1.06750882e+00 7.27436602e-01
-3.82530004e-01 3.98035049e-01 2.31351003e-01 4.75866832e-02
-4.08313364e-01 -4.06731516e-01 -1.22485220e+00 4.22831833e-01
1.15292943e+00 -5.28448462e-01 -6.88816786e-01 1.57378510e-01
6.42872870e-01 -9.29932714e-01 -1.22944939e+00 9.17707145e-01
4.88019109e-01 -1.23537672e+00 1.21066272e+00 1.91284671e-01
1.03943825e-01 -6.10979557e-01 -1.46130815e-01 -1.10967624e+00
-6.27129853e-01 -3.68340909e-01 5.98901622e-02 1.13682926e+00
-9.87278745e-02 -9.84444857e-01 6.38090909e-01 5.23671150e-01
-6.35265768e-01 -8.94130111e-01 -1.41943944e+00 -7.43320346e-01
-1.57083333e-01 -4.91290420e-01 1.20524347e+00 1.19111037e+00
-3.64201367e-01 6.39369041e-02 4.86581624e-02 6.45401835e-01
5.31109333e-01 6.51542842e-02 8.55311096e-01 -1.66137087e+00
2.82852769e-01 -4.24160570e-01 -9.38028812e-01 -1.38850152e+00
-5.21935299e-02 -6.65747106e-01 -4.26689088e-02 -1.24752176e+00
-3.26519996e-01 -1.02194512e+00 -2.52351314e-01 2.98595101e-01
2.53521055e-01 4.68405396e-01 5.18008135e-02 7.13446617e-01
-3.09373021e-01 9.43638265e-01 1.22994149e+00 -1.75821893e-02
6.27526492e-02 -6.82374043e-03 -3.77079666e-01 7.68839240e-01
1.80495292e-01 -4.20320243e-01 -7.10955635e-02 -5.64702094e-01
3.16653043e-01 6.32444723e-03 7.18648612e-01 -1.42560983e+00
5.89032769e-01 2.06153944e-01 7.87914693e-01 -1.11574578e+00
5.77153385e-01 -1.46772170e+00 4.17287320e-01 1.41920790e-01
4.73282725e-01 5.54191351e-01 1.69891864e-01 4.85130280e-01
-6.06450617e-01 2.52510071e-01 4.39966917e-01 4.57249954e-02
-4.36595589e-01 8.63157153e-01 3.54093522e-01 -5.25466383e-01
6.22704923e-01 -8.82117629e-01 -1.15985170e-01 -1.62447542e-01
-2.28770599e-01 2.20778346e-01 8.61290812e-01 4.93426323e-01
9.51181352e-01 -1.85922444e+00 -5.10609150e-01 7.58027971e-01
3.88418883e-01 7.11459816e-01 1.15266591e-01 1.05932426e+00
-6.66531503e-01 2.63153851e-01 -1.59953579e-01 -1.20033920e+00
-9.15093601e-01 2.64789075e-01 4.82091129e-01 -7.10653663e-02
-9.35231805e-01 7.49594688e-01 2.72633612e-01 -6.30176961e-01
2.19618883e-02 -7.11393237e-01 1.33955419e-01 -2.84357667e-01
2.32175905e-02 3.30277085e-01 6.77816033e-01 -1.09799957e+00
-5.24478316e-01 1.26833272e+00 3.29800658e-02 3.55649531e-01
1.28407896e+00 5.15246540e-02 -3.88018012e-01 -1.62495580e-02
1.41273844e+00 -1.21932447e-01 -9.58280444e-01 -3.78332436e-01
-1.77755237e-01 -9.32464063e-01 1.08761698e-01 -2.87184060e-01
-1.23091245e+00 6.89891279e-01 6.92783237e-01 -2.61620693e-02
1.08056474e+00 -1.13130957e-01 9.97860312e-01 3.62702966e-01
8.21468592e-01 -7.98973858e-01 -2.89412022e-01 7.44421065e-01
1.17267036e+00 -1.26778090e+00 1.28626436e-01 -7.19853461e-01
-1.00707524e-01 1.47354209e+00 6.44158006e-01 -4.90532875e-01
9.17575896e-01 1.70262396e-01 -1.17654437e-02 -8.03491056e-01
-2.15440214e-01 1.92324258e-02 7.40459621e-01 7.25473166e-01
1.89253554e-01 -2.78927922e-01 1.79849342e-01 3.56933177e-01
-6.14100397e-01 -1.13269471e-01 -3.08755249e-01 7.51788259e-01
-8.54367507e-04 -9.13758278e-01 -5.62720537e-01 4.74453598e-01
4.14235108e-02 -6.68055564e-02 -1.88544303e-01 9.90680039e-01
4.23872769e-01 6.19057536e-01 4.29112315e-01 -6.08700275e-01
7.89950848e-01 -4.96609867e-01 3.45730186e-01 -3.99483234e-01
-7.36184299e-01 1.49593338e-01 -4.61486548e-01 -1.04815376e+00
-3.13718647e-01 -4.46938604e-01 -1.05035210e+00 -3.99251908e-01
-5.41879714e-01 1.12622224e-01 4.84094828e-01 8.85069430e-01
7.18640149e-01 3.59891683e-01 6.18867934e-01 -1.23268378e+00
-5.13337135e-01 -9.74484205e-01 -4.92442757e-01 4.71019894e-01
4.42668140e-01 -1.11848557e+00 -2.45458215e-01 -4.37756151e-01] | [7.688384532928467, -3.0790059566497803] |
c64bd691-d81c-4336-a422-46abd03ddd66 | effective-matching-of-patients-to-clinical | 2307.00381 | null | https://arxiv.org/abs/2307.00381v1 | https://arxiv.org/pdf/2307.00381v1.pdf | Effective Matching of Patients to Clinical Trials using Entity Extraction and Neural Re-ranking | Clinical trials (CTs) often fail due to inadequate patient recruitment. This paper tackles the challenges of CT retrieval by presenting an approach that addresses the patient-to-trials paradigm. Our approach involves two key components in a pipeline-based model: (i) a data enrichment technique for enhancing both queries and documents during the first retrieval stage, and (ii) a novel re-ranking schema that uses a Transformer network in a setup adapted to this task by leveraging the structure of the CT documents. We use named entity recognition and negation detection in both patient description and the eligibility section of CTs. We further classify patient descriptions and CT eligibility criteria into current, past, and family medical conditions. This extracted information is used to boost the importance of disease and drug mentions in both query and index for lexical retrieval. Furthermore, we propose a two-step training schema for the Transformer network used to re-rank the results from the lexical retrieval. The first step focuses on matching patient information with the descriptive sections of trials, while the second step aims to determine eligibility by matching patient information with the criteria section. Our findings indicate that the inclusion criteria section of the CT has a great influence on the relevance score in lexical models, and that the enrichment techniques for queries and documents improve the retrieval of relevant trials. The re-ranking strategy, based on our training schema, consistently enhances CT retrieval and shows improved performance by 15\% in terms of precision at retrieving eligible trials. The results of our experiments suggest the benefit of making use of extracted entities. Moreover, our proposed re-ranking schema shows promising effectiveness compared to larger neural models, even with limited training data. | ['Allan Hanbury', 'Gabriella Pasi', 'Petr Knoth', 'Óscar E. Mendoza', 'Wojciech Kusa'] | 2023-07-01 | null | null | null | null | ['retrieval', 'negation-detection'] | ['methodology', 'natural-language-processing'] | [ 2.15892196e-01 1.19371735e-01 -6.95593059e-01 -2.23808035e-01
-1.13295960e+00 -3.90432507e-01 5.54323852e-01 9.50871825e-01
-1.00602186e+00 7.18308628e-01 6.77248180e-01 -4.18304890e-01
-6.73561454e-01 -8.27931046e-01 -4.99759823e-01 -2.99571931e-01
1.43028542e-01 7.85151660e-01 3.09616238e-01 -1.84533730e-01
1.63491324e-01 6.30169451e-01 -1.44086409e+00 6.03205860e-01
7.45212853e-01 9.40970004e-01 2.52233654e-01 2.50093490e-01
-1.62742063e-01 6.96514904e-01 -5.11516511e-01 -2.10102513e-01
4.45720442e-02 -2.32864812e-01 -8.69881332e-01 -6.11866593e-01
3.75232324e-02 -3.10356885e-01 -2.32748613e-01 6.93371773e-01
1.10056591e+00 -9.93508622e-02 6.64201677e-01 -5.66349149e-01
-9.78653431e-02 9.90727246e-01 5.52390702e-03 1.78691655e-01
6.19753480e-01 -1.36338755e-01 9.23843861e-01 -9.25447047e-01
1.17012608e+00 7.65789747e-01 6.33106530e-01 5.70757329e-01
-7.01314628e-01 -5.92651904e-01 -2.12659538e-01 1.79724425e-01
-1.36415243e+00 -3.01903814e-01 1.72650963e-01 -3.50463092e-01
1.21121514e+00 4.88974929e-01 6.65836811e-01 9.09506559e-01
1.34497076e-01 5.57630122e-01 6.74512386e-01 -7.32611120e-01
3.55407417e-01 3.21243703e-01 2.86075950e-01 2.77192533e-01
2.69108474e-01 2.04397827e-01 -3.17558646e-01 -4.19216394e-01
3.86589050e-01 -1.76371576e-03 -2.81250477e-01 -6.79089725e-02
-9.00382161e-01 8.47180724e-01 4.87130672e-01 7.48206377e-01
-7.26832569e-01 -1.55959606e-01 7.75044620e-01 2.32907739e-02
2.15931028e-01 7.41151512e-01 -5.33537090e-01 2.10121840e-01
-1.15579903e+00 1.67432621e-01 8.36308956e-01 8.92523825e-01
1.21872030e-01 -6.06649220e-01 -8.28630745e-01 9.12380159e-01
2.50277221e-01 3.70507360e-01 8.51059496e-01 -3.24641943e-01
6.26886189e-01 9.92198169e-01 -1.10069424e-01 -6.47692382e-01
-7.18609989e-01 -5.08116841e-01 -7.00396597e-01 -4.78446603e-01
-1.48560300e-01 -1.44041821e-01 -1.28066313e+00 1.73342419e+00
2.36677200e-01 -2.84293979e-01 1.72339693e-01 5.74240506e-01
1.47334456e+00 2.37502918e-01 7.35751987e-01 -4.45453703e-01
1.65888226e+00 -4.78487909e-01 -9.57140088e-01 2.02043712e-01
1.05035579e+00 -7.89671957e-01 6.46050453e-01 3.06594595e-02
-1.20020270e+00 -3.98265779e-01 -9.91051614e-01 1.39466003e-01
-7.07800210e-01 5.05909026e-01 3.21744859e-01 4.99222398e-01
-1.05950141e+00 6.74444139e-01 -4.36286420e-01 -5.74069262e-01
3.77149493e-01 8.55417669e-01 -2.51784235e-01 -2.55198363e-04
-1.63607287e+00 1.13913906e+00 7.61064768e-01 -4.69502360e-02
-7.26571560e-01 -9.09738481e-01 -7.38759398e-01 3.37063491e-01
4.67164338e-01 -1.01407099e+00 1.04918694e+00 -3.44195157e-01
-9.71032560e-01 6.70329571e-01 -2.98222527e-02 -3.77202600e-01
3.53722006e-01 -1.29329056e-01 -5.14815390e-01 4.00340170e-01
2.18098834e-01 5.94913721e-01 3.23868304e-01 -7.06342101e-01
-7.49809086e-01 -5.66526413e-01 -2.47695193e-01 4.43228573e-01
-5.69378138e-01 2.34830096e-01 -9.49661613e-01 -6.44082606e-01
-2.12796271e-01 -7.45992839e-01 -4.45819199e-01 -4.97300237e-01
-2.66826928e-01 -4.04102653e-01 1.98754594e-01 -6.00757420e-01
1.89846361e+00 -1.77291489e+00 4.17149402e-02 6.03198111e-01
3.13931108e-01 5.16841173e-01 -1.32691160e-01 6.07231855e-01
-4.58532095e-01 4.85676497e-01 1.78232491e-01 5.02076047e-03
-2.84021497e-01 -4.87512089e-02 -1.58817694e-02 -1.11062154e-01
2.55589485e-01 1.04902005e+00 -7.19630301e-01 -7.73292542e-01
-1.01941466e-01 4.35811341e-01 -4.01812911e-01 1.31860510e-01
-9.35058743e-02 1.23734795e-01 -9.29401517e-01 4.62119102e-01
3.49415153e-01 -3.65315259e-01 5.33327639e-01 -5.24900138e-01
-7.56004378e-02 6.29443228e-01 -9.57306564e-01 1.49576139e+00
-2.67104059e-01 -5.92355020e-02 -3.93608004e-01 -6.79711699e-01
4.08682913e-01 7.87151039e-01 7.64227331e-01 -1.17688882e+00
3.46753865e-01 5.45798421e-01 -6.02939203e-02 -8.51698160e-01
4.92907226e-01 -1.30558804e-01 4.01307922e-03 4.19762015e-01
-7.63539821e-02 9.04171914e-02 4.30171907e-01 3.94369751e-01
1.20917630e+00 -5.11633269e-02 5.43016434e-01 9.49342176e-02
7.56121159e-01 1.97993934e-01 3.84157717e-01 8.53086710e-01
2.82068402e-01 2.89883345e-01 2.62044787e-01 -2.78018385e-01
-8.71268034e-01 -4.43069100e-01 -3.32899243e-01 7.83402205e-01
-3.55961889e-01 -5.78595996e-01 -4.89495248e-01 -8.22819471e-01
-2.45491732e-02 5.40452600e-01 -8.26400936e-01 -3.47595364e-01
-7.37656653e-01 -1.06795239e+00 7.45224297e-01 7.15295553e-01
1.62114829e-01 -1.31728852e+00 -7.21632063e-01 2.27733567e-01
-1.50610685e-01 -9.15052116e-01 -2.50801027e-01 5.57121515e-01
-1.17931819e+00 -1.21791542e+00 -9.76028562e-01 -7.33575821e-01
4.72236782e-01 -4.66077596e-01 1.05603826e+00 3.27301800e-01
-3.42282295e-01 4.97797489e-01 -6.11239791e-01 -3.88917774e-01
-5.58403730e-01 3.59750390e-01 -2.10946620e-01 -6.75214350e-01
4.36655521e-01 -2.36029774e-02 -7.35472798e-01 9.13272500e-02
-1.42623723e+00 -4.65950102e-01 1.01712978e+00 8.06078434e-01
6.63050175e-01 -1.26112267e-01 5.76237798e-01 -9.31997359e-01
9.47177827e-01 -5.25469780e-01 -4.08405364e-01 6.62066877e-01
-1.08727610e+00 3.55894327e-01 9.96490121e-02 -4.14757341e-01
-8.40965748e-01 1.24231026e-01 -4.47831064e-01 -1.07707702e-01
-2.73860395e-02 1.12986577e+00 -2.21636239e-02 1.12564325e-01
7.64798403e-01 -1.27873242e-01 -1.87507018e-01 -6.16555214e-01
1.75065726e-01 6.44753873e-01 -6.17679246e-02 -3.70505273e-01
2.23973125e-01 6.63878545e-02 1.63172036e-01 -4.31022078e-01
-3.95421237e-01 -8.02580595e-01 -5.75239778e-01 1.72417536e-01
8.86852980e-01 -7.08443165e-01 -7.49073684e-01 -5.22214137e-02
-9.81353581e-01 -2.80519258e-02 -5.81462801e-01 8.08479786e-01
-2.36901324e-02 4.22195882e-01 -6.43629968e-01 -5.21003723e-01
-7.86672652e-01 -1.29322612e+00 1.07575858e+00 1.52190760e-01
-2.16133848e-01 -7.48868048e-01 3.03662390e-01 8.45806953e-03
2.87531555e-01 -9.22385380e-02 1.52490783e+00 -1.27544940e+00
-2.53967553e-01 -4.93775457e-01 -2.52245247e-01 -5.91283664e-02
3.18860859e-02 -3.68208289e-01 -6.33444011e-01 -3.81729119e-02
-3.66218910e-02 -2.09329829e-01 8.88814449e-01 5.46073556e-01
8.40173006e-01 -7.08695278e-02 -7.90946066e-01 1.60622999e-01
1.51928365e+00 5.64382076e-01 6.30465746e-01 4.68241781e-01
3.69207591e-01 7.24219501e-01 6.58170521e-01 3.45979691e-01
4.60350551e-02 7.79389620e-01 2.52732217e-01 -3.13941300e-01
-2.51942188e-01 -5.38752563e-02 -4.68644649e-02 6.20652139e-01
5.55378944e-02 -3.10359776e-01 -1.05892122e+00 5.99145293e-01
-1.66457510e+00 -5.69225490e-01 4.07297164e-02 2.47987461e+00
1.09136355e+00 8.71484578e-02 4.21018526e-02 5.28558623e-03
3.97780925e-01 -4.72243339e-01 -3.39862019e-01 -2.50935048e-01
-6.62036017e-02 8.19343090e-01 5.71975887e-01 2.48053491e-01
-9.07878935e-01 6.75015211e-01 6.55932713e+00 1.05179179e+00
-1.02076149e+00 5.57735004e-02 4.24872100e-01 -6.26808852e-02
-3.02164823e-01 -2.33952329e-01 -1.09913397e+00 6.47548735e-02
1.09871733e+00 -9.73923206e-02 -3.09686214e-01 3.76467764e-01
2.15181604e-01 -1.99347526e-01 -1.28073537e+00 4.41918373e-01
1.32209316e-01 -1.42129576e+00 4.63615865e-01 6.49835542e-02
2.59177893e-01 9.69305634e-02 -4.26820032e-02 4.04802263e-01
5.81866056e-02 -1.02256000e+00 3.22170079e-01 5.24153531e-01
1.00440145e+00 -3.30584466e-01 1.20145059e+00 1.25187054e-01
-1.14623404e+00 -1.54307187e-01 -9.61934179e-02 7.23428488e-01
-9.67952609e-03 4.11936760e-01 -1.37802589e+00 1.16633570e+00
6.41153753e-01 5.02595305e-01 -6.78385377e-01 1.38983333e+00
-9.93351042e-02 3.43215078e-01 -3.37699771e-01 -1.79369599e-01
2.25659817e-01 3.17978710e-01 4.29689795e-01 1.42045987e+00
3.61460686e-01 2.29998484e-01 -3.20553891e-02 5.69551110e-01
-2.50861257e-01 7.78075635e-01 -4.51874465e-01 -1.70880184e-01
3.60640377e-01 9.91254151e-01 -7.47945189e-01 -6.46783471e-01
-1.22568935e-01 3.34766150e-01 -1.37662530e-01 2.90566832e-01
-5.57615101e-01 -2.98077345e-01 -3.61390442e-01 1.02289662e-01
4.50809717e-01 6.54987454e-01 -1.87738556e-02 -6.96350813e-01
-4.80479635e-02 -1.08394563e+00 1.01205981e+00 -6.82231903e-01
-9.43440437e-01 1.05480421e+00 3.40935230e-01 -1.30264223e+00
-4.76128966e-01 -4.90962386e-01 1.79733142e-05 9.82678890e-01
-1.45096910e+00 -1.05382597e+00 -6.83351234e-02 6.52265131e-01
2.73775250e-01 -7.19195455e-02 1.02885377e+00 8.84867847e-01
-5.26364207e-01 6.00329041e-01 -1.13227479e-02 1.07990578e-01
8.93771648e-01 -9.35374618e-01 -3.36205870e-01 1.94408461e-01
-1.96456105e-01 1.00361526e+00 1.73354477e-01 -1.23640954e+00
-9.76519585e-01 -8.08685303e-01 1.35364377e+00 -3.25132430e-01
3.82146478e-01 2.20531747e-01 -8.01927686e-01 2.18054801e-01
5.93468547e-02 -5.54739654e-01 8.90320182e-01 1.08544037e-01
-1.15802281e-01 6.87285736e-02 -1.09120274e+00 4.45315212e-01
5.72195947e-01 -4.06760335e-01 -8.77530575e-01 4.12167639e-01
7.19597220e-01 -3.72247100e-01 -1.12322462e+00 9.35507357e-01
6.55510783e-01 -9.08636302e-02 1.12048066e+00 -1.04050589e+00
3.43527138e-01 5.87929133e-03 -5.82501898e-03 -9.03403878e-01
-1.82860658e-01 -3.45828868e-02 2.16896400e-01 9.26162064e-01
9.27242994e-01 -2.47663021e-01 7.22637236e-01 5.80705822e-01
-1.38445143e-02 -8.68203759e-01 -8.74605656e-01 -3.37041765e-01
3.40515785e-02 -2.84954518e-01 5.20889938e-01 8.35756779e-01
1.22045130e-01 4.52176601e-01 1.63983971e-01 -1.14128754e-01
1.42773941e-01 -1.53816521e-01 -9.23450571e-03 -1.40811062e+00
-1.03500985e-01 -5.64328492e-01 9.42833163e-03 -4.65890646e-01
-3.42366636e-01 -1.08188915e+00 -9.77127720e-03 -1.74250185e+00
6.15513504e-01 -6.63244724e-01 -5.57463646e-01 5.67703843e-01
-1.79339170e-01 1.58567838e-02 3.88620514e-03 3.79872322e-01
-5.01424611e-01 3.42181414e-01 9.02563989e-01 -1.85892537e-01
-5.11900842e-01 2.08837446e-02 -5.86151838e-01 4.56715971e-01
4.36879992e-01 -9.36064780e-01 -3.33609402e-01 -2.92999029e-01
6.09252453e-01 2.51116902e-01 1.67105183e-01 -7.02175200e-01
5.54713666e-01 2.95304209e-01 3.52549344e-01 -9.52104747e-01
8.14915523e-02 -1.02264941e+00 1.32140517e-01 8.69153559e-01
-6.19864941e-01 2.75381029e-01 6.16131961e-01 4.14908350e-01
-2.67358035e-01 -7.54886508e-01 2.23273560e-01 -2.26642936e-01
-3.81923646e-01 8.28529447e-02 -4.08231765e-01 -3.41576687e-03
5.29761553e-01 -6.67720009e-03 -1.77241400e-01 -5.22461981e-02
-1.20007670e+00 3.02346706e-01 -9.93561000e-02 1.99653909e-01
6.37265921e-01 -1.15970051e+00 -6.48628056e-01 -1.77685112e-01
2.65501410e-01 -3.29407603e-01 1.53367877e-01 1.12787187e+00
-2.88090199e-01 1.02191365e+00 -3.01186438e-03 -4.32075500e-01
-1.55999589e+00 6.45370781e-01 1.26951396e-01 -1.23739052e+00
-2.60004789e-01 5.98774076e-01 -7.92634934e-02 -2.56965876e-01
6.52568102e-01 -4.21444774e-01 -9.87131059e-01 5.47436655e-01
4.24995184e-01 1.43049791e-01 6.76156938e-01 -5.04635453e-01
-5.39783537e-01 6.50905132e-01 -4.81064796e-01 -3.53880852e-01
1.56996596e+00 2.59339422e-01 -2.63914555e-01 -1.66203976e-01
1.05611849e+00 3.80787924e-02 2.27985561e-01 -3.06924045e-01
3.65828961e-01 2.40185514e-01 2.87685812e-01 -1.38401473e+00
-9.36739087e-01 5.74847937e-01 9.32042897e-01 -1.42426610e-01
1.43769670e+00 -3.86719555e-02 4.34371293e-01 5.03933907e-01
6.61326498e-02 -1.07546401e+00 -1.27327472e-01 3.61944169e-01
7.44373381e-01 -9.35280204e-01 4.05224264e-01 -4.16321248e-01
-4.84259784e-01 9.79788721e-01 1.57630503e-01 4.59923983e-01
6.81354642e-01 9.98351052e-02 -1.59710422e-01 -7.65675366e-01
-5.96124649e-01 -3.49263698e-01 7.57697761e-01 2.61546582e-01
5.30699849e-01 -2.69533694e-01 -1.04022658e+00 7.88636148e-01
4.65002447e-01 5.20387769e-01 3.47224772e-02 1.04468369e+00
-8.81856680e-02 -1.62972784e+00 -3.15393060e-01 6.18543983e-01
-7.73393631e-01 -6.19658351e-01 -5.83486140e-01 1.07479322e+00
2.31306672e-01 7.18760192e-01 -4.77088928e-01 -3.78336787e-01
7.30367899e-01 2.59118080e-01 2.98831254e-01 -8.54585052e-01
-1.45735586e+00 5.66515863e-01 2.86864281e-01 -5.70918381e-01
-6.83087230e-01 -4.25934672e-01 -1.06678772e+00 4.05195862e-01
-6.18928075e-01 6.87639058e-01 7.55701363e-01 8.11121762e-01
5.31003714e-01 8.80488038e-01 2.48267248e-01 -1.36176094e-01
-5.29291809e-01 -1.01913273e+00 -7.10636675e-02 2.72425175e-01
-4.18567508e-02 -5.91681421e-01 -4.09530662e-02 -2.46774510e-01] | [8.710406303405762, 8.643304824829102] |
2a4726e2-4116-47f2-b518-18df189bd18e | person-retrieval-in-surveillance-video-using-1 | 1810.05080 | null | https://arxiv.org/abs/1810.05080v1 | https://arxiv.org/pdf/1810.05080v1.pdf | Person Retrieval in Surveillance Video using Height, Color and Gender | A person is commonly described by attributes like height, build, cloth color, cloth type, and gender. Such attributes are known as soft biometrics. They bridge the semantic gap between human description and person retrieval in surveillance video. The paper proposes a deep learning-based linear filtering approach for person retrieval using height, cloth color, and gender. The proposed approach uses Mask R-CNN for pixel-wise person segmentation. It removes background clutter and provides precise boundary around the person. Color and gender models are fine-tuned using AlexNet and the algorithm is tested on SoftBioSearch dataset. It achieves good accuracy for person retrieval using the semantic query in challenging conditions. | ['Mehul S. Raval', 'Vandit Gajjar', 'Kenil Shah', 'Hiren Galiyawala'] | 2018-09-24 | person-retrieval-in-surveillance-video-using | https://ieeexplore.ieee.org/document/8639145 | https://ieeexplore.ieee.org/document/8639145 | 2018-15th-ieee-international-conference-on | ['person-retrieval'] | ['computer-vision'] | [-2.90446937e-01 -5.01201928e-01 4.93347682e-02 -6.69010103e-01
-4.15215880e-01 -5.36251485e-01 4.67669278e-01 1.94187269e-01
-7.61835039e-01 5.19344866e-01 -6.27975306e-03 4.98502821e-01
1.19202308e-01 -1.02534175e+00 -2.51558691e-01 -5.64130962e-01
2.63453811e-01 5.93766391e-01 -4.75935042e-02 -6.00130670e-02
-1.52037010e-01 5.00061393e-01 -1.43016279e+00 2.08576202e-01
3.77495676e-01 1.32288003e+00 -4.95238334e-01 8.51446152e-01
6.57003224e-02 3.63450795e-02 -9.57908273e-01 -9.14572597e-01
5.57515621e-01 9.39484760e-02 -5.70434511e-01 1.88687369e-01
1.02868986e+00 -6.63459539e-01 -7.33404577e-01 1.07424295e+00
9.13618386e-01 2.40722243e-02 5.28578699e-01 -1.06153846e+00
-1.07486987e+00 1.13011420e-01 -8.48965645e-01 1.32154524e-01
7.18596399e-01 2.44248033e-01 4.25100207e-01 -4.64923233e-01
3.03127974e-01 1.80580056e+00 1.04445505e+00 8.13704193e-01
-6.97918594e-01 -5.57258189e-01 7.70753324e-02 7.48136714e-02
-1.87052763e+00 5.09158969e-02 4.01966512e-01 -4.43610370e-01
6.25028789e-01 4.92309123e-01 1.24351776e+00 1.07475209e+00
1.97702184e-01 8.50825608e-01 1.19325531e+00 -9.98207405e-02
-1.52700111e-01 2.51581967e-01 4.38205421e-01 1.35073876e+00
5.14235675e-01 1.99460134e-01 -4.04145598e-01 -2.19562829e-01
8.34814250e-01 4.63977486e-01 8.84295627e-03 2.04234108e-01
-1.04590738e+00 4.76127774e-01 5.93714297e-01 -1.20906577e-01
-2.75319189e-01 2.67321885e-01 4.48915809e-01 -4.70939539e-02
3.31533521e-01 1.08610742e-01 -2.03662246e-01 3.99747252e-01
-1.13302493e+00 4.25412029e-01 7.41543770e-01 1.19216132e+00
3.17651093e-01 5.85827716e-02 -8.34447622e-01 8.11581671e-01
4.22828913e-01 1.29842317e+00 7.09229484e-02 -7.57901669e-01
-3.75159264e-01 7.66007125e-01 1.57665908e-01 -1.24482381e+00
-4.86686379e-01 -3.49517643e-01 -6.98074996e-01 -1.84970751e-01
5.63930273e-01 -1.07756302e-01 -1.50661027e+00 1.38513732e+00
3.95696521e-01 -1.77606121e-01 -3.05857658e-01 1.41595364e+00
1.71192801e+00 5.26516140e-01 3.63468677e-01 2.69037217e-01
1.84609413e+00 -7.43297815e-01 -8.31203163e-01 -1.76076934e-01
-5.53861141e-01 -6.31851614e-01 7.20244288e-01 2.60318309e-01
-1.19290769e+00 -9.21337664e-01 -8.15903783e-01 -2.10310698e-01
-9.16986287e-01 5.25987804e-01 7.19017386e-01 1.19096649e+00
-9.44676220e-01 2.21203417e-01 -4.28706825e-01 -8.83273065e-01
5.36918879e-01 6.41934872e-01 -9.89238545e-02 2.09604055e-02
-1.28239095e+00 5.79871655e-01 7.75512531e-02 3.75186831e-01
-7.51841486e-01 -3.25392097e-01 -8.29109251e-01 -1.27890348e-01
9.99492477e-04 -1.34895527e+00 5.69394290e-01 -8.40301394e-01
-1.26978016e+00 1.79496837e+00 -7.09351674e-02 -2.13912994e-01
7.57676899e-01 -2.09054127e-01 -4.99249518e-01 4.31597710e-01
6.33347631e-02 9.48441327e-01 1.22923410e+00 -1.12680221e+00
-5.52353382e-01 -8.33444953e-01 -3.67239267e-02 6.74937889e-02
-3.04393142e-01 1.94160730e-01 -1.13637280e+00 -5.39139807e-01
-2.91055385e-02 -6.97894275e-01 8.11557285e-03 2.71948606e-01
-5.20026028e-01 -3.08369994e-01 6.41578555e-01 -1.12971735e+00
7.56621659e-01 -1.78684998e+00 -1.08693957e-01 6.11527801e-01
2.45356351e-01 2.61082947e-01 1.13182485e-01 -2.33175039e-01
5.38501561e-01 -2.01509699e-01 2.68570989e-01 -3.45201463e-01
2.20905423e-01 -1.30538613e-01 2.34831110e-01 8.56131732e-01
-1.91803798e-01 1.20080173e+00 -4.71936047e-01 -9.05182481e-01
3.52409184e-01 9.31589842e-01 -5.06153740e-02 2.69547701e-01
2.03023106e-01 1.02500081e-01 -3.13861787e-01 1.44334269e+00
8.83539557e-01 8.02738145e-02 -3.27736199e-01 -6.03558004e-01
1.99107647e-01 -8.83296728e-01 -1.11403728e+00 1.34135604e+00
2.64835149e-01 6.07979953e-01 3.54120672e-01 -5.04344106e-01
1.19646168e+00 2.37137377e-02 4.69495207e-01 -5.41798353e-01
6.61342084e-01 -1.36652768e-01 -5.16244829e-01 -7.24765599e-01
4.72744375e-01 2.14005649e-01 -2.93457896e-01 -2.58065611e-01
-8.15029144e-02 1.26488522e-01 1.06583305e-01 -6.90331981e-02
4.72150326e-01 1.76848590e-01 4.46013995e-02 -3.55728984e-01
7.17955351e-01 7.38432780e-02 4.82171118e-01 1.04952824e+00
-6.85383558e-01 7.32798338e-01 -2.41442636e-01 -1.04914880e+00
-1.06797361e+00 -1.22769296e+00 -8.09532404e-02 1.16770637e+00
6.23473644e-01 -5.94121031e-02 -1.20772719e+00 -3.76772672e-01
4.96167153e-01 -3.31751019e-01 -8.75677526e-01 -1.93801716e-01
-4.97601569e-01 -6.53050542e-01 8.80213261e-01 6.07341826e-01
1.11833203e+00 -8.67112577e-01 -5.58180869e-01 -2.12710828e-01
-5.86360134e-02 -1.12175143e+00 -7.20204651e-01 -5.54443836e-01
-2.14457184e-01 -1.24771690e+00 -1.23141539e+00 -8.58603835e-01
7.79825747e-01 -6.72543049e-02 1.14440417e+00 3.73619676e-01
-9.92997110e-01 1.07301950e+00 1.35628209e-01 -5.89143991e-01
3.69978964e-01 -1.75694257e-01 1.97283700e-01 2.19314307e-01
1.05162668e+00 4.97700423e-01 -1.19767487e+00 2.12934449e-01
-3.96605760e-01 -4.26179618e-01 3.43014449e-01 6.33394241e-01
6.87226832e-01 2.77112797e-02 -2.11798260e-03 -4.72201020e-01
5.78805387e-01 2.53488980e-02 -5.32979429e-01 6.48536146e-01
-2.90082216e-01 -6.01545751e-01 2.52209101e-02 -5.48006654e-01
-8.81683707e-01 5.55556715e-02 3.54949608e-02 -3.37178856e-01
-4.94603336e-01 -1.76282063e-01 -1.86470851e-01 -3.33914727e-01
3.31349701e-01 1.86266199e-01 -7.98802227e-02 -2.06383467e-01
7.86534771e-02 6.88324213e-01 7.50329971e-01 -5.51477253e-01
7.48733401e-01 8.82475555e-01 -6.06012419e-02 -8.28314066e-01
-8.06240261e-01 -6.61162674e-01 -6.50665879e-01 -6.43865645e-01
1.34359145e+00 -1.08319414e+00 -1.50063968e+00 7.79707313e-01
-1.00905669e+00 1.12550311e-01 6.46783784e-02 2.33173966e-01
-4.19030599e-02 2.03456342e-01 -8.49246025e-01 -1.05125391e+00
-1.02273715e+00 -7.27980316e-01 1.48019612e+00 8.63373041e-01
-6.46952838e-02 -7.14870691e-01 -5.13506770e-01 6.49372280e-01
4.39028442e-01 5.56353331e-01 1.49263471e-01 -4.79808629e-01
-5.48106611e-01 -7.41966903e-01 -5.72579682e-01 -1.78180095e-02
-1.42120168e-01 -1.19977761e-02 -1.06304765e+00 -5.86985469e-01
-5.35948217e-01 -1.13787919e-01 1.19630873e+00 8.54819775e-01
1.39035487e+00 -1.54242232e-01 -4.67954874e-01 9.50279117e-01
1.46210682e+00 1.58925787e-01 3.84619862e-01 1.63631082e-01
9.55470502e-01 5.44253647e-01 4.00473803e-01 4.66942489e-01
4.75986421e-01 4.26062405e-01 -1.24849603e-01 -6.27960801e-01
-3.66594762e-01 7.98498094e-02 -1.87316045e-01 -1.09850720e-01
-4.48225647e-01 -1.03252351e-01 -8.69970322e-01 2.30650738e-01
-1.60098135e+00 -1.02554178e+00 1.05582237e-01 1.84788644e+00
5.92015982e-01 -2.19594315e-02 4.83781040e-01 -3.03167969e-01
9.76457059e-01 -1.80818498e-01 -5.40891409e-01 -3.52360681e-02
-4.98115093e-01 1.59743831e-01 8.71397913e-01 3.54719371e-01
-1.60856950e+00 1.08732319e+00 6.78950119e+00 5.66049755e-01
-9.50060189e-01 -3.91596481e-02 7.93156147e-01 -1.05523221e-01
4.14499998e-01 -8.47925782e-01 -1.16155577e+00 2.39039972e-01
1.41557068e-01 2.46479258e-01 5.30875921e-01 6.68026924e-01
-7.04561323e-02 -8.21402669e-02 -1.07317650e+00 1.66386306e+00
5.86982608e-01 -9.76154268e-01 -9.19887889e-03 -2.90098667e-01
5.99733174e-01 -6.86363697e-01 4.00829434e-01 2.17997640e-01
-5.56175821e-02 -1.14603412e+00 7.64665604e-01 1.14898086e+00
8.53783906e-01 -6.60059929e-01 9.41589653e-01 -3.55323792e-01
-1.27175951e+00 -2.10897654e-01 -4.98187512e-01 4.04270500e-01
-1.75774649e-01 5.53845502e-02 -5.17753124e-01 -3.37856752e-03
1.39132845e+00 3.66270602e-01 -7.72341728e-01 1.19373202e+00
3.07204187e-01 1.80423334e-01 -3.47774327e-01 -1.77418873e-01
-2.15258133e-02 -1.40609309e-01 5.75402081e-01 1.79311359e+00
6.43496141e-02 3.28632981e-01 7.31112480e-01 1.04522288e+00
3.59006338e-02 3.48576419e-02 -3.03178459e-01 9.75776315e-02
2.19364539e-01 1.30044687e+00 -9.41053271e-01 -5.78453720e-01
-1.79948211e-01 1.08456075e+00 -4.74680871e-01 7.25586772e-01
-8.66750777e-01 -2.23606750e-01 5.84868133e-01 2.70564973e-01
-3.68155017e-02 -9.13598482e-03 -2.97583520e-01 -7.85441637e-01
-2.46759295e-01 -8.28394830e-01 7.44967401e-01 -6.73135459e-01
-1.54444003e+00 4.87778276e-01 -1.01685831e-02 -6.04121327e-01
5.50754309e-01 -9.16681767e-01 -1.90587863e-01 9.16965485e-01
-1.06646574e+00 -1.79230082e+00 -9.55038846e-01 8.63896847e-01
4.36128318e-01 -7.19048202e-01 7.40989804e-01 5.41776717e-01
-6.43734276e-01 8.40783656e-01 -3.65700215e-01 7.72737384e-01
8.91647756e-01 -1.40678275e+00 1.26776859e-01 5.00957906e-01
-4.52757776e-01 1.11576045e+00 6.78364754e-01 -1.02487850e+00
-1.49581373e+00 -1.01759338e+00 3.79509509e-01 -4.68514860e-01
-2.06844136e-01 -4.22568083e-01 -4.18648452e-01 3.18437904e-01
2.96876013e-01 1.46809034e-02 7.25583553e-01 -8.32159445e-02
-3.30917388e-01 -4.97911483e-01 -1.71303642e+00 3.17203790e-01
6.88005388e-01 -4.56877291e-01 -2.13495702e-01 5.16641736e-01
3.61485392e-01 -6.74110889e-01 -8.33358765e-01 3.09249401e-01
9.13993955e-01 -5.45018077e-01 1.61711597e+00 -5.45201182e-01
-2.62948960e-01 -2.87915856e-01 -8.34728777e-02 -5.74890494e-01
-4.81956482e-01 -3.33153069e-01 -7.15142787e-02 1.01660633e+00
-9.38652232e-02 -2.37159103e-01 8.89196396e-01 9.37378287e-01
3.43779564e-01 -4.41170037e-01 -4.65710998e-01 -6.05929852e-01
-2.69290537e-01 3.23366135e-01 8.34462523e-01 5.95548809e-01
-6.68595016e-01 -2.35136524e-01 -5.25392473e-01 2.52821535e-01
1.24296701e+00 7.49610588e-02 6.49735510e-01 -1.47388816e+00
4.01598215e-02 -4.97552156e-01 -5.14912188e-01 -5.94370127e-01
2.61711975e-04 -5.66984475e-01 -1.55531406e-01 -1.69561517e+00
7.15738475e-01 -7.22719878e-02 -3.96478295e-01 2.77778119e-01
-2.30509296e-01 8.34736288e-01 9.54669267e-02 -1.47390634e-01
-8.63032818e-01 -8.28323662e-02 1.12508309e+00 -7.88309336e-01
-4.72554527e-02 -1.19751357e-02 -3.70646864e-01 7.94725716e-01
8.25004220e-01 8.40052664e-02 2.90212423e-01 -3.91799003e-01
-1.86814126e-02 -1.43056393e-01 9.39918876e-01 -1.07749653e+00
4.65177834e-01 -1.63407475e-02 1.57673621e+00 -8.46360624e-01
7.91801691e-01 -8.38039398e-01 -5.01580387e-02 7.20304549e-01
-3.95193756e-01 4.38347412e-03 1.40441194e-01 4.25697714e-01
1.14918597e-01 2.27744296e-01 9.04482961e-01 -5.66781342e-01
-9.22264695e-01 7.59152949e-01 -3.82313207e-02 -3.45054507e-01
8.58732283e-01 -5.55253685e-01 -3.30690175e-01 -2.62049228e-01
-1.10740054e+00 3.17526817e-01 2.78088868e-01 4.81732845e-01
9.70615685e-01 -1.32601726e+00 -7.40020633e-01 3.07722718e-01
-5.77825271e-02 -6.76810086e-01 1.22162409e-01 4.34850872e-01
-7.10888445e-01 2.95678973e-01 -3.90697420e-01 -6.52888834e-01
-1.86074901e+00 5.17662644e-01 8.17888081e-01 4.99384463e-01
-4.47814763e-01 1.21851254e+00 -5.48435077e-02 -1.99205771e-01
7.84930825e-01 2.00687736e-01 -3.56052399e-01 -4.01820391e-02
6.27943397e-01 5.29248118e-01 -5.69458663e-01 -9.53905880e-01
-6.46451652e-01 9.85708892e-01 1.46753475e-01 8.45067874e-02
9.07685757e-01 -2.77349532e-01 -3.92905802e-01 -2.19426565e-02
1.01494658e+00 -2.61163205e-01 -9.36754882e-01 -3.11708242e-01
-3.86059314e-01 -7.47414768e-01 -3.80023532e-02 -1.07508123e+00
-1.32136393e+00 6.51252806e-01 1.45650482e+00 9.14436132e-02
9.30398762e-01 -2.20750403e-02 1.11256909e+00 3.13324302e-01
2.11947739e-01 -1.63887477e+00 1.21806048e-01 1.96270436e-01
6.66034460e-01 -1.57374954e+00 2.61523902e-01 -3.64475369e-01
-3.76208931e-01 1.05711460e+00 9.39869761e-01 -1.85723662e-01
5.24359226e-01 1.45704299e-01 1.20536551e-01 -3.73694450e-01
2.18804181e-01 -5.82940757e-01 8.96607339e-01 8.47818196e-01
3.21240455e-01 4.34056759e-01 -1.62619799e-01 6.84804082e-01
-2.32620403e-01 -2.03512594e-01 -3.68606925e-01 5.71367443e-01
-5.16989529e-01 -4.06502694e-01 -1.00517571e+00 3.84141028e-01
-8.25829208e-01 1.65022254e-01 -7.14043260e-01 4.94931012e-01
7.11721063e-01 1.01127875e+00 1.24775544e-01 -1.97236463e-01
3.34474891e-01 -6.22610264e-02 8.10988486e-01 -6.03088699e-02
-9.72194731e-01 2.41577893e-01 -1.61799476e-01 -5.58194339e-01
-4.32509750e-01 -4.37930346e-01 -9.69436049e-01 -6.18180931e-01
1.95597634e-01 -2.41809279e-01 5.74211955e-01 7.10938573e-01
-2.21626714e-01 3.50521415e-01 3.26244161e-02 -4.52891469e-01
-1.85917526e-01 -7.31386185e-01 -5.44724166e-01 7.73227513e-01
2.60727525e-01 -5.23573399e-01 2.76927561e-01 3.49910319e-01] | [14.58775520324707, 0.882739782333374] |
e7f1c82b-43be-47d7-aa2b-9d775edfd62e | causalr-causal-reasoning-over-natural | null | null | https://openreview.net/forum?id=vuAX_4bv8A | https://openreview.net/pdf?id=vuAX_4bv8A | CausalR: Causal Reasoning over Natural Language Rulebases | Transformers have been shown to perform deductive reasoning on a logical rulebase containing rules and statements written in natural language. Recent works show that such models can also produce the reasoning steps (i.e., the proof graph) that emulate the model’s logical reasoning process. But these models behave as a black-box unit that emulates the reasoning process without any causal constraints in the reasoning steps, thus questioning the faithfulness. In this work, we frame the deductive logical reasoning task as a causal process by defining three modular components: rule selection, fact selection, and knowledge composition. The rule and fact selection steps select the candidate rule and facts to be used and then the knowledge composition combines them to generate new inferences. This ensures model faithfulness by assured causal relation from the proof step to the inference reasoning. To test our causal reasoning framework, we propose CausalR, where the above three components are independently modeled by transformers. We observe that CausalR is robust to novel language perturbations, and performs on par with previous works on existing reasoning datasets. Furthermore, the errors made by CausalR are more interpretable due to our multi-modular approach compared to black-box generative models. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['fact-selection'] | ['natural-language-processing'] | [ 2.94606864e-01 1.03716707e+00 -1.49196684e-01 -1.55518770e-01
-5.86912706e-02 -7.40644217e-01 1.32654488e+00 1.38944983e-01
3.82091820e-01 8.75142276e-01 3.68078947e-01 -8.63133967e-01
-4.41088200e-01 -1.44098318e+00 -1.16712260e+00 -2.71625161e-01
-3.13485712e-02 6.36805236e-01 6.92548752e-01 -2.49299169e-01
6.81161061e-02 2.36940533e-01 -1.57650483e+00 5.31150997e-01
9.34319317e-01 5.12331724e-01 -3.51934582e-01 6.65258467e-01
-8.58080238e-02 2.12575984e+00 -3.89100134e-01 -8.34980488e-01
-2.88405828e-02 -8.08924317e-01 -1.22367120e+00 -3.63378167e-01
1.58462510e-01 -2.83011794e-01 -1.31947443e-01 1.10818839e+00
-1.67244211e-01 -2.08246447e-02 4.83375371e-01 -1.76898849e+00
-8.23666394e-01 1.51327395e+00 1.40284956e-01 -3.24487984e-01
8.72917473e-01 1.28243104e-01 1.15757585e+00 -4.43571746e-01
8.92886996e-01 1.58999503e+00 5.45794129e-01 7.01757133e-01
-1.35422504e+00 -4.42669690e-01 2.03156233e-01 7.89414942e-01
-1.03710818e+00 -4.51481164e-01 7.30103970e-01 -3.52428854e-01
9.95081127e-01 5.52239060e-01 7.48058200e-01 1.28718185e+00
3.73451352e-01 5.23490071e-01 1.09387255e+00 -6.29669070e-01
6.49827242e-01 2.73886383e-01 1.69985473e-01 8.32507312e-01
5.92494249e-01 3.23804498e-01 -7.56463051e-01 -3.75797689e-01
4.46909755e-01 -3.65115970e-01 -3.31744224e-01 -4.04537916e-01
-1.00590587e+00 6.83900356e-01 9.57756042e-02 1.99685380e-01
-3.58549505e-01 4.40578103e-01 2.16799989e-01 4.51638192e-01
-3.12476665e-01 4.90638793e-01 -4.98280019e-01 7.80056417e-02
-5.69756269e-01 6.07429922e-01 1.20998645e+00 9.62731838e-01
2.97529399e-01 -2.84933418e-01 -3.23568285e-01 1.55637279e-01
6.51758730e-01 6.62030280e-01 1.35786787e-01 -1.21658385e+00
1.01328999e-01 9.66052771e-01 1.24463804e-01 -9.53982592e-01
-1.38807774e-01 -3.61750647e-02 -4.69681770e-01 2.50440121e-01
3.74871552e-01 1.15733914e-01 -5.72745085e-01 2.01048112e+00
3.42907608e-01 2.52773345e-01 3.05035502e-01 4.58754361e-01
6.85795844e-01 4.58639234e-01 4.23247926e-02 -3.75268698e-01
1.33707726e+00 -5.04547238e-01 -8.39125454e-01 4.88472730e-02
3.45686287e-01 -2.59117812e-01 8.90083075e-01 4.31392074e-01
-1.44253206e+00 -3.16116475e-02 -1.09504890e+00 6.11931272e-02
-1.47770643e-01 -5.94361603e-01 7.68733859e-01 2.40133494e-01
-9.57336307e-01 4.20293480e-01 -6.49937153e-01 -1.57729194e-01
2.10664362e-01 -2.26407230e-01 -2.72873282e-01 -9.45445970e-02
-1.73527563e+00 1.20730150e+00 4.70376581e-01 -1.12764567e-01
-1.37940025e+00 -7.75694311e-01 -8.62680376e-01 1.78128555e-01
9.16294932e-01 -1.17399764e+00 1.38167477e+00 -4.50410008e-01
-1.56729376e+00 5.94224811e-01 -2.76655763e-01 -8.98442030e-01
8.43109787e-01 -1.09041996e-01 -4.61938143e-01 1.13057569e-01
2.75572855e-02 1.18900992e-01 5.47026932e-01 -1.56131291e+00
-4.82987285e-01 -1.54294744e-01 6.37709856e-01 -3.82416219e-01
4.30866629e-01 -2.96122767e-03 -1.01037197e-01 -1.03132278e-01
9.01464224e-02 -8.50743771e-01 5.28030805e-02 -2.73920208e-01
-8.89611125e-01 -2.47236922e-01 4.86993432e-01 -2.68569440e-01
1.31454337e+00 -1.56267297e+00 2.22049057e-01 3.87707353e-01
2.53809243e-01 -1.93451926e-01 3.81354153e-01 6.49535000e-01
-3.73128653e-01 4.75946337e-01 -1.27062544e-01 1.75581053e-01
4.85717207e-01 3.37224066e-01 -9.41681027e-01 6.32349625e-02
1.16250895e-01 1.06750309e+00 -9.07036483e-01 -6.35947108e-01
1.74693897e-01 2.70592887e-02 -6.67999208e-01 1.43792272e-01
-8.51432741e-01 1.86992567e-02 -2.53635973e-01 3.96833539e-01
3.96423191e-01 -2.37722471e-01 6.88648701e-01 -3.83687317e-02
9.12997052e-02 9.51460361e-01 -1.30556142e+00 1.17440236e+00
-3.92571449e-01 2.09993720e-01 -5.34455121e-01 -6.54233932e-01
6.65526867e-01 5.16113579e-01 -9.34415981e-02 -4.24486339e-01
-4.34085131e-02 -9.21971798e-02 2.27151573e-01 -7.31044471e-01
-1.05262868e-01 -6.27749205e-01 -1.85421237e-03 8.10375273e-01
-7.58561725e-03 -4.63091135e-01 5.23070812e-01 7.47542679e-01
1.25763845e+00 5.81829190e-01 6.37354672e-01 -1.81850970e-01
5.61960578e-01 2.62266845e-01 7.46025980e-01 9.33011591e-01
2.46037126e-01 -6.28873408e-02 1.16833210e+00 -5.11984944e-01
-6.81361735e-01 -1.30770957e+00 9.49737802e-02 5.52593887e-01
-1.16171679e-02 -8.68891418e-01 -5.77755690e-01 -7.71952689e-01
-2.00876407e-02 1.74702859e+00 -8.87578785e-01 -6.87549472e-01
-3.22752714e-01 -2.39586785e-01 9.22840953e-01 3.31300110e-01
3.27463359e-01 -1.21522486e+00 -8.49622011e-01 6.55646846e-02
-3.93455714e-01 -9.80541050e-01 4.35851157e-01 8.10317621e-02
-5.04528522e-01 -1.74719143e+00 7.35714614e-01 -1.02693224e-02
6.59793496e-01 -3.39303195e-01 1.33553004e+00 2.71514714e-01
1.72951043e-01 2.64104515e-01 -3.99570227e-01 -5.60788631e-01
-1.06469560e+00 -5.91045856e-01 -3.15384306e-02 -2.39794880e-01
1.00564390e-01 -9.29058015e-01 8.77527818e-02 9.36417207e-02
-1.05116606e+00 4.79237258e-01 1.28053412e-01 6.09900713e-01
3.13811749e-01 7.03026831e-01 1.26674339e-01 -1.10641241e+00
6.46873772e-01 -4.05557990e-01 -6.69780791e-01 8.15973818e-01
-7.42512703e-01 5.72678447e-01 9.64215279e-01 -8.09319317e-03
-1.43923688e+00 -4.33409572e-01 3.37415010e-01 1.20937508e-02
-1.18833937e-01 6.55896246e-01 -4.59038794e-01 6.76704228e-01
8.99610877e-01 2.16698304e-01 -3.18607360e-01 4.43779007e-02
5.78062594e-01 -1.23925515e-01 5.62346816e-01 -1.19252813e+00
1.36055720e+00 5.31351030e-01 4.24283177e-01 1.76108778e-01
-8.71818304e-01 6.40736461e-01 -3.08824837e-01 -2.26245686e-01
5.04912198e-01 -4.27866787e-01 -1.29310358e+00 -6.44413605e-02
-1.26241136e+00 -5.48575878e-01 -6.46671474e-01 1.98788822e-01
-8.38636458e-01 7.27473795e-02 -3.96272987e-01 -1.13828588e+00
-2.25581117e-02 -9.05596018e-01 5.81748128e-01 -1.41200379e-01
-6.94484949e-01 -9.77529705e-01 2.39743337e-01 1.66161343e-01
-3.83761637e-02 2.85553664e-01 1.45848036e+00 -6.19311452e-01
-6.72327578e-01 7.88083002e-02 1.13913924e-01 8.18641409e-02
-5.72880544e-02 6.43630981e-01 -9.01668847e-01 5.47618210e-01
2.03913022e-02 -1.58043608e-01 4.71012414e-01 1.54969553e-02
7.22069919e-01 -6.91896796e-01 -2.41174236e-01 1.81121871e-01
1.23805678e+00 2.58396894e-01 9.05651331e-01 3.03569406e-01
2.94067591e-01 5.41414022e-01 2.65992969e-01 8.48378614e-02
6.04972959e-01 5.76639593e-01 4.19918001e-01 4.22796905e-01
-1.34710521e-01 -8.87629628e-01 5.99123180e-01 1.69351056e-01
-3.90144676e-01 -1.03578791e-01 -1.19748449e+00 2.19237030e-01
-2.25101304e+00 -1.80445945e+00 -4.96550441e-01 1.90292132e+00
1.34745145e+00 3.29682887e-01 -2.93199092e-01 5.14257669e-01
2.39315376e-01 -2.22687125e-01 -2.00947255e-01 -5.14066458e-01
-2.18376666e-01 1.73954606e-01 -3.03489696e-02 9.92617130e-01
-3.31187606e-01 8.14125061e-01 6.49939489e+00 2.91958153e-01
-5.02825499e-01 -3.07944249e-02 -8.80068019e-02 -1.50427356e-01
-1.07212901e+00 7.26694167e-01 -3.90159220e-01 1.93791479e-01
8.31368983e-01 -4.89669502e-01 7.91835189e-01 6.49727762e-01
3.18099290e-01 -3.45742971e-01 -1.55452180e+00 1.51276439e-01
-8.21519792e-02 -1.56742728e+00 5.43821216e-01 -3.28125507e-01
5.62795579e-01 -6.14981174e-01 -4.35703397e-01 1.99379116e-01
1.37952697e+00 -9.81283665e-01 1.37607193e+00 8.29926848e-01
2.27841660e-01 -6.47748947e-01 6.16717875e-01 3.75466079e-01
-7.14215577e-01 -3.16097289e-01 1.39431536e-01 -3.79505634e-01
1.87211469e-01 1.08301187e+00 -9.63620901e-01 9.08863664e-01
6.01977110e-01 3.97699475e-01 -5.04557312e-01 4.51536357e-01
-1.38899398e+00 7.76382625e-01 -2.11917702e-02 1.34969532e-01
-3.73961657e-01 -3.69096473e-02 6.32039070e-01 8.71126652e-01
-3.35299820e-02 -3.47356945e-02 -2.85704613e-01 1.48730206e+00
8.20682123e-02 -4.98567760e-01 -6.54564023e-01 1.35214657e-01
5.72434783e-01 7.28281081e-01 -4.26842362e-01 -7.48825729e-01
-4.97505553e-02 5.12622476e-01 3.34798515e-01 4.57124978e-01
-1.25597441e+00 3.32266353e-02 5.13669968e-01 7.81509057e-02
-1.02063581e-01 1.39182284e-01 -4.98643309e-01 -1.24562263e+00
1.63415357e-01 -1.08876085e+00 3.91927093e-01 -1.30418324e+00
-9.85358298e-01 1.94054574e-01 3.86330038e-01 -5.99770308e-01
-5.22904217e-01 -3.60131830e-01 -8.11720908e-01 6.79121137e-01
-1.27691555e+00 -1.13212979e+00 -8.16939473e-02 8.05616498e-01
-1.66391373e-01 3.36023152e-01 1.00186253e+00 -3.06120247e-01
-4.88785684e-01 1.13800898e-01 -9.27385449e-01 -2.63562668e-02
4.47081655e-01 -1.42097950e+00 1.22373670e-01 1.44548404e+00
1.25982851e-01 1.35358238e+00 1.15894687e+00 -8.51989985e-01
-1.53025520e+00 -1.00667727e+00 1.30546486e+00 -7.90619731e-01
9.61640835e-01 -3.64074945e-01 -7.00964749e-01 1.19211006e+00
2.04975501e-01 -3.33079636e-01 4.72517848e-01 3.81624192e-01
-9.85270977e-01 -9.39309299e-02 -1.12331247e+00 1.07027018e+00
1.43132937e+00 -5.12334645e-01 -1.36497247e+00 2.06209168e-01
8.19617093e-01 -1.41607627e-01 -5.91328621e-01 3.94016176e-01
5.67756236e-01 -1.19954610e+00 7.66797304e-01 -9.54326808e-01
1.07265854e+00 -9.83768165e-01 -1.79933175e-01 -1.24763000e+00
-4.33120966e-01 -6.99868679e-01 -5.59383333e-01 1.29696202e+00
7.27496564e-01 -8.21147084e-01 -2.61135418e-02 8.95321727e-01
7.05788359e-02 -3.27644348e-01 -7.36133695e-01 -7.06284583e-01
-1.85712650e-01 -8.87626886e-01 9.41889584e-01 9.77145016e-01
7.69625068e-01 4.76955891e-01 -1.03302471e-01 2.64258564e-01
5.53063035e-01 5.83179891e-01 7.89794803e-01 -1.30132782e+00
-6.67348385e-01 -4.93060082e-01 -3.76657210e-02 -1.12231024e-01
4.02409613e-01 -1.04922020e+00 4.80618738e-02 -1.90107882e+00
3.91295940e-01 -1.28655612e-01 1.89918354e-01 1.11995423e+00
-1.80002034e-01 -4.70735669e-01 1.99927047e-01 6.33037090e-02
-4.04429853e-01 3.18857014e-01 1.25681829e+00 -2.22043917e-01
1.51279429e-02 -3.03831846e-01 -9.83647108e-01 9.36296940e-01
6.27495408e-01 -4.79658008e-01 -9.00804520e-01 4.22942229e-02
1.09316790e+00 2.37294380e-02 9.94891107e-01 -7.47401476e-01
3.84535611e-01 -6.48419321e-01 -5.95231578e-02 7.27866441e-02
-2.19744250e-01 -8.06943297e-01 7.40457952e-01 6.89474046e-01
-6.63852870e-01 -2.03407526e-01 1.22245951e-02 1.71657071e-01
-9.01328772e-02 -1.07854337e-01 4.42002922e-01 -8.28628689e-02
-2.81105042e-01 -2.69045502e-01 -4.82138962e-01 3.19895446e-02
1.02554059e+00 2.57124305e-01 -7.18409717e-01 -3.19979161e-01
-7.91728020e-01 9.52694565e-02 4.36191440e-01 2.55576700e-01
4.21526074e-01 -1.19599879e+00 -7.30192840e-01 -8.13201293e-02
1.54175058e-01 5.47270961e-02 -7.52591481e-03 1.07196414e+00
-1.89068288e-01 3.15038234e-01 -4.97821793e-02 -1.32771298e-01
-9.12615776e-01 8.61249030e-01 5.52412629e-01 -5.01113653e-01
-4.85881716e-01 5.51433027e-01 -1.34395927e-01 -4.57301944e-01
-1.28648967e-01 -6.60587132e-01 -5.83338179e-03 -1.67741731e-01
6.75500035e-01 2.24607259e-01 -3.92838642e-02 -9.47446853e-04
-6.28294647e-01 8.45945850e-02 3.99016649e-01 -3.27254415e-01
1.17768359e+00 1.77759439e-01 -7.33501971e-01 6.43224239e-01
1.77184865e-01 4.61388201e-01 -7.12475121e-01 -6.92269504e-02
-8.09128955e-02 -5.26552983e-02 -3.78444672e-01 -1.35548091e+00
-5.28667867e-01 4.51413244e-01 -5.34859776e-01 6.65898204e-01
8.94183993e-01 2.51237422e-01 6.82098744e-03 3.30284238e-01
6.11950099e-01 -8.37746263e-01 -4.06756252e-01 5.44315219e-01
1.23298490e+00 -5.86793840e-01 4.65387851e-02 -7.47146487e-01
-4.59231585e-01 9.58221674e-01 3.88444424e-01 1.45416260e-01
2.28785232e-01 7.24975288e-01 -1.99849799e-01 -3.27683657e-01
-1.48740208e+00 3.07568274e-02 -2.49077030e-03 6.28060162e-01
2.09451497e-01 3.25542837e-01 -2.55552471e-01 8.76061201e-01
-6.27410591e-01 4.88642991e-01 6.21289968e-01 7.34796286e-01
-1.19494803e-01 -1.09237337e+00 -6.14807129e-01 2.89036576e-02
-4.19373065e-02 -2.35395387e-01 -8.15543175e-01 7.88776100e-01
2.77156234e-01 1.39197457e+00 -4.28383559e-01 -4.43838388e-01
4.07403886e-01 5.24378717e-01 7.92782784e-01 -4.67677027e-01
-3.92747790e-01 -8.31374407e-01 5.68212986e-01 -8.94324422e-01
-3.94027114e-01 -6.69743419e-01 -1.67151463e+00 -6.75637126e-01
-5.85249178e-02 3.64166558e-01 9.51248333e-02 1.24811745e+00
2.53012896e-01 9.02059138e-01 2.05930784e-01 1.41835615e-01
-4.67468679e-01 -6.83516622e-01 -1.69118360e-01 5.88992894e-01
-1.16656264e-02 -6.63983583e-01 -3.92593592e-01 5.43880582e-01] | [9.207515716552734, 7.145236015319824] |
b31fd768-8264-49c7-b9dd-7a7e912ae04a | msr-net-multi-scale-relighting-network-for | 2107.06125 | null | https://arxiv.org/abs/2107.06125v1 | https://arxiv.org/pdf/2107.06125v1.pdf | MSR-Net: Multi-Scale Relighting Network for One-to-One Relighting | Deep image relighting allows photo enhancement by illumination-specific retouching without human effort and so it is getting much interest lately. Most of the existing popular methods available for relighting are run-time intensive and memory inefficient. Keeping these issues in mind, we propose the use of Stacked Deep Multi-Scale Hierarchical Network, which aggregates features from each image at different scales. Our solution is differentiable and robust for translating image illumination setting from input image to target image. Additionally, we have also shown that using a multi-step training approach to this problem with two different loss functions can significantly boost performance and can achieve a high quality reconstruction of a relighted image. | ['Saikat Dutta', 'Nisarg A. Shah', 'Sourya Dipta Das'] | 2021-07-13 | null | null | null | null | ['image-relighting'] | ['computer-vision'] | [ 3.38623077e-01 -2.59655654e-01 1.90453678e-01 -1.98946044e-01
-9.98296797e-01 -3.50034714e-01 1.46768689e-01 -2.30536520e-01
-3.70806277e-01 7.44926929e-01 2.18727991e-01 -4.13392074e-02
3.18616778e-01 -8.50576043e-01 -9.76342440e-01 -7.74773896e-01
3.04642111e-01 -1.87538460e-01 4.67347413e-01 -3.94949257e-01
2.61053890e-01 6.73658729e-01 -1.37828660e+00 2.21505389e-01
7.63479173e-01 9.55741465e-01 2.13984936e-01 8.56840968e-01
2.59261221e-01 7.79291570e-01 -5.59390783e-01 -6.61887050e-01
5.55045605e-01 -3.54051352e-01 -8.29126298e-01 2.74913758e-01
9.90138173e-01 -7.62397945e-01 -5.77177703e-01 8.84078860e-01
6.80752695e-01 3.36785614e-01 2.27075666e-01 -9.47073042e-01
-7.65726209e-01 9.12248641e-02 -9.14070606e-01 8.17888230e-02
1.23740144e-01 9.23373848e-02 7.20897257e-01 -9.28695381e-01
6.67988360e-01 1.26474941e+00 8.43165517e-01 3.43561113e-01
-1.48290837e+00 -4.52406555e-01 -2.98470885e-01 2.27070659e-01
-1.21874154e+00 -2.95522690e-01 8.17223966e-01 -5.40661812e-02
8.31493437e-01 2.70339757e-01 5.09376466e-01 9.45883453e-01
4.13548827e-01 6.91794336e-01 1.40212214e+00 -3.73189151e-01
-1.16244540e-01 -2.86391322e-02 -2.20362440e-01 7.81453669e-01
-2.46549491e-02 3.16245973e-01 -3.18959266e-01 1.39634430e-01
1.27774715e+00 5.42235076e-02 -4.71171081e-01 -2.14038610e-01
-9.55744624e-01 7.78408527e-01 8.67443383e-01 2.05323324e-01
-2.45418578e-01 5.82973480e-01 4.65474755e-01 2.83222616e-01
5.14784336e-01 3.95075560e-01 -3.01637828e-01 1.88617587e-01
-1.02048326e+00 -1.15713356e-02 2.54461795e-01 5.52435338e-01
1.17563534e+00 2.26640120e-01 -6.23988882e-02 7.36851990e-01
-1.37618780e-01 2.66494572e-01 3.18387657e-01 -1.33342028e+00
3.65389854e-01 3.36245358e-01 2.44945973e-01 -1.20259106e+00
-3.08833718e-01 -4.09009904e-01 -1.08759975e+00 6.76204741e-01
2.58273482e-01 -1.40375018e-01 -1.00597227e+00 1.58988500e+00
5.05888641e-01 2.10003525e-01 8.78404975e-02 9.56069589e-01
6.85726643e-01 8.82135212e-01 -1.07842185e-01 -8.94868188e-03
1.19690323e+00 -1.25556242e+00 -7.29479551e-01 -1.30700111e-01
3.38472962e-01 -9.81834114e-01 1.26551008e+00 4.82077658e-01
-1.33265078e+00 -5.99642098e-01 -1.21301317e+00 -8.47588062e-01
-2.32712269e-01 1.75832268e-02 6.17497206e-01 5.36878824e-01
-1.38162434e+00 8.27916920e-01 -7.32250929e-01 -2.46145993e-01
4.07875180e-01 3.71885866e-01 -5.67143023e-01 -2.19894379e-01
-9.23188627e-01 1.07067573e+00 2.94859350e-01 1.57073662e-01
-8.67487848e-01 -5.21840870e-01 -7.81922281e-01 1.31508544e-01
2.49168068e-01 -7.74891615e-01 9.94085729e-01 -1.21705246e+00
-1.75165868e+00 7.03339636e-01 -8.99777487e-02 -3.05102676e-01
6.78523183e-01 -1.83037490e-01 -1.14077806e-01 3.91879052e-01
-2.53230512e-01 7.06978142e-01 1.15563428e+00 -1.49655616e+00
-3.96920264e-01 -2.49645203e-01 3.88485938e-02 3.52394164e-01
-4.26663667e-01 2.67334711e-02 -4.92675573e-01 -8.46119106e-01
-1.16763236e-02 -7.09010661e-01 -2.73846745e-01 2.67533511e-01
-2.43944436e-01 3.80982816e-01 1.04637909e+00 -1.05258846e+00
8.64986360e-01 -2.06003881e+00 2.66912639e-01 -1.41317055e-01
2.00671837e-01 3.33517760e-01 -3.80523205e-01 4.43292707e-01
-8.91131684e-02 1.51496353e-02 -2.34761760e-01 -5.47251582e-01
-3.23042870e-01 2.83769935e-01 -4.31962013e-01 5.34985363e-01
6.91630021e-02 1.10661566e+00 -7.25508332e-01 -5.53426921e-01
5.47816575e-01 9.99550343e-01 -6.16384149e-01 3.46332759e-01
4.41199429e-02 5.41455925e-01 8.36992860e-02 4.76997107e-01
8.79807174e-01 -2.61037618e-01 -1.00349091e-01 -6.80280924e-01
-1.25950605e-01 -2.07633659e-01 -9.72765982e-01 1.63574302e+00
-8.59394491e-01 9.39460814e-01 6.94169477e-02 -7.04031408e-01
6.23713613e-01 1.81704134e-01 2.38920555e-01 -8.24706376e-01
2.17301279e-01 1.69502616e-01 -5.72258472e-01 -3.36724937e-01
7.73586273e-01 -2.82290041e-01 2.06487834e-01 2.42218018e-01
-1.58663634e-02 -4.12611663e-01 1.54028879e-02 6.50159828e-03
8.96501184e-01 2.09974244e-01 1.24968924e-01 1.15481146e-01
2.60725498e-01 -3.70350122e-01 4.51930642e-01 3.97858739e-01
-9.69997495e-02 1.09792590e+00 2.02209875e-01 -6.46955192e-01
-1.41568232e+00 -9.03290689e-01 1.51566556e-02 1.03140724e+00
3.63039613e-01 -6.43968806e-02 -8.26151192e-01 -5.15398979e-01
-2.98782676e-01 4.84694421e-01 -6.70298159e-01 -3.38412449e-02
-8.37841690e-01 -6.11380041e-01 2.48814583e-01 4.44243520e-01
1.03909874e+00 -9.78934050e-01 -6.01796687e-01 1.55740559e-01
-3.33462059e-01 -1.23115790e+00 -7.05662191e-01 1.30924180e-01
-9.46587801e-01 -8.26896489e-01 -1.12082410e+00 -9.26994324e-01
6.67148829e-01 6.11405134e-01 1.17836702e+00 2.77645409e-01
-3.95469606e-01 2.38662273e-01 -2.79299110e-01 1.99477375e-01
-3.36478263e-01 -6.51764274e-02 -4.30321157e-01 5.86325377e-02
-6.25374258e-01 -6.75807893e-01 -8.92858267e-01 3.34323674e-01
-1.30588984e+00 3.22563648e-01 5.66766620e-01 9.42393482e-01
5.72038889e-01 2.32380852e-01 5.63209951e-02 -5.39902747e-01
2.80532002e-01 1.26270875e-01 -6.20737493e-01 2.44742602e-01
-3.98919344e-01 6.51304564e-03 8.28594923e-01 -2.96841413e-01
-1.22494483e+00 -4.18835506e-02 -2.62731731e-01 -4.96038944e-01
1.86529562e-01 2.55453177e-02 -3.00060436e-02 -7.49180496e-01
6.64985955e-01 2.72024244e-01 -1.67537048e-01 -3.00047815e-01
4.80145544e-01 1.75152838e-01 7.00084984e-01 -1.92755595e-01
9.46781337e-01 7.67128468e-01 2.78915346e-01 -7.40400791e-01
-9.48793054e-01 -4.03015949e-02 -4.58461225e-01 -1.68327883e-01
8.53845656e-01 -8.27522576e-01 -6.75488889e-01 7.53516555e-01
-1.18972290e+00 -7.21690178e-01 -3.42900753e-01 5.32623678e-02
-5.46545923e-01 7.23991454e-01 -9.10962164e-01 -3.56118232e-01
-5.30534983e-01 -1.20334983e+00 1.28560019e+00 3.84209067e-01
4.46954608e-01 -1.06292295e+00 -4.25725281e-02 3.78675491e-01
5.53557754e-01 4.77552772e-01 6.65385842e-01 4.76755530e-01
-8.73667300e-01 -5.73551767e-02 -7.27614462e-01 5.97145975e-01
1.43950179e-01 -4.41744514e-02 -9.27882969e-01 -5.78572154e-01
3.33153047e-02 -6.21994317e-01 8.99950743e-01 4.84731525e-01
1.31903660e+00 -4.72031295e-01 -1.71570051e-02 1.04101992e+00
1.87529933e+00 -2.24810973e-01 1.04732430e+00 4.39261764e-01
9.04225051e-01 2.68146217e-01 4.93640304e-01 2.19284117e-01
3.05993289e-01 8.52038443e-01 6.43985271e-01 -9.05557811e-01
-5.47012627e-01 -1.87471628e-01 3.80588084e-01 3.47862512e-01
-1.47662889e-02 -3.00782859e-01 -5.25882483e-01 4.92624819e-01
-1.65204501e+00 -9.68821287e-01 5.57552353e-02 2.21581268e+00
9.40282464e-01 -3.01262587e-01 -1.06461048e-01 -5.37814051e-02
6.39535367e-01 2.03468263e-01 -5.87037742e-01 -4.33853358e-01
-3.75672787e-01 2.20466852e-01 7.41036952e-01 7.95581996e-01
-9.87485111e-01 1.15332878e+00 7.10086679e+00 9.00993645e-01
-1.32708418e+00 1.24836273e-01 9.70787346e-01 7.15468526e-02
-2.92308211e-01 -5.11463210e-02 -3.77081603e-01 2.36745521e-01
6.34070396e-01 -3.25636044e-02 6.61930919e-01 5.13693929e-01
2.77124435e-01 -3.11391890e-01 -6.82647586e-01 1.19106627e+00
2.05667600e-01 -1.34102643e+00 8.26082975e-02 -1.50842950e-01
1.06776094e+00 -5.95773757e-03 3.20131719e-01 -1.41702304e-02
3.53371978e-01 -9.00725007e-01 5.34521341e-01 2.41079405e-01
8.33049238e-01 -8.54529738e-01 5.40830195e-01 -3.01353224e-02
-1.13556170e+00 -4.16312180e-02 -4.59393263e-01 2.55752116e-01
2.99260050e-01 5.53130567e-01 -4.39863622e-01 5.51811337e-01
7.77083158e-01 6.08914435e-01 -6.17692113e-01 1.01740599e+00
-3.52961063e-01 3.51529866e-01 -1.65061280e-01 4.77363080e-01
1.51562572e-01 -1.75283045e-01 3.38555574e-01 1.06428003e+00
5.42889595e-01 -3.56239732e-03 -1.81654412e-02 5.77609301e-01
-1.80684820e-01 -4.21922142e-03 -5.41646540e-01 3.64106834e-01
-6.44680783e-02 1.41446686e+00 -5.92756391e-01 -3.48373622e-01
-2.77334988e-01 1.67491031e+00 3.93325150e-01 5.64096808e-01
-9.48646963e-01 -4.22619402e-01 4.51658309e-01 5.75920567e-02
4.65727478e-01 -3.40823710e-01 -2.52399653e-01 -1.29412413e+00
-1.44928336e-01 -6.36202335e-01 1.54833406e-01 -1.22497392e+00
-1.04395866e+00 6.74482048e-01 -3.12084168e-01 -1.14524353e+00
1.55957043e-01 -3.98574382e-01 -4.88697678e-01 7.85153270e-01
-1.95422852e+00 -1.34842372e+00 -5.53435743e-01 7.25796998e-01
4.95316118e-01 3.56068879e-01 5.96711993e-01 4.37723339e-01
-5.79122603e-01 6.83313549e-01 2.52746850e-01 -3.99087649e-03
8.89820874e-01 -1.20163441e+00 4.17336226e-01 1.08380055e+00
5.20105241e-03 3.88616294e-01 9.81271565e-01 -2.97692865e-01
-1.39846659e+00 -1.19378209e+00 5.60253620e-01 -1.40257943e-02
3.52416426e-01 -1.05751768e-01 -9.58811522e-01 6.48946822e-01
7.68270493e-01 1.74108848e-01 3.20691854e-01 -3.61582011e-01
-4.00925636e-01 -3.48268092e-01 -1.40167952e+00 8.35201859e-01
7.58768320e-01 -4.12708700e-01 -7.03167021e-02 3.37302893e-01
8.45204771e-01 -6.39576733e-01 -8.57478023e-01 1.74052075e-01
4.22563702e-01 -1.34981275e+00 1.11675000e+00 -5.37814610e-02
6.07253432e-01 -2.71930665e-01 -1.24635905e-01 -1.30714059e+00
-3.29751283e-01 -8.59260499e-01 9.88019705e-02 9.61181998e-01
1.39429020e-02 -6.37713373e-01 6.75654948e-01 5.43837368e-01
-1.36515066e-01 -5.84747672e-01 -8.79615486e-01 -6.38540149e-01
6.23509288e-02 2.47100461e-02 3.09355080e-01 6.70771658e-01
-4.86108780e-01 2.56502777e-01 -9.37526703e-01 2.40899131e-01
7.96512008e-01 2.20055014e-01 9.51833367e-01 -5.40995479e-01
-1.98977008e-01 -2.41358116e-01 -2.66651034e-01 -9.91013706e-01
-7.78564364e-02 -5.73668778e-01 3.81569155e-02 -1.64037323e+00
3.24406862e-01 -6.13844097e-02 -3.50516587e-02 6.96228921e-01
-2.84061968e-01 9.81697917e-01 1.42567828e-01 5.81920259e-02
-3.37059885e-01 7.42558539e-01 1.65689433e+00 -9.59231891e-03
1.23528671e-02 -4.71057534e-01 -5.61344802e-01 4.99393314e-01
7.42141068e-01 -2.39877746e-01 -2.88826823e-01 -7.23555624e-01
2.07950592e-01 1.11136623e-01 5.74019253e-01 -9.13365662e-01
-1.25263527e-01 -1.97635457e-01 4.82452720e-01 -4.07676041e-01
5.87316513e-01 -9.01314080e-01 1.44036070e-01 3.80589515e-01
-1.21494822e-01 8.24656487e-02 3.31671417e-01 4.13496166e-01
-2.70120651e-01 -2.95183081e-02 1.40262771e+00 -4.71235476e-02
-7.24049151e-01 4.77965742e-01 -1.71461105e-02 -2.12088943e-01
9.36059773e-01 -3.06192607e-01 -2.75787473e-01 -5.19998014e-01
-3.44471574e-01 -1.33667797e-01 8.91635656e-01 -1.05280746e-02
7.67185986e-01 -1.36926484e+00 -6.25599325e-01 6.39166161e-02
-3.25153112e-01 -2.09023625e-01 5.68790436e-01 6.58256650e-01
-8.93713951e-01 -7.03113154e-02 -4.21310902e-01 -4.31295753e-01
-1.43725872e+00 5.97997248e-01 4.31688726e-01 -4.09825176e-01
-8.48251700e-01 8.19483578e-01 2.23665550e-01 -1.19798467e-01
5.83610460e-02 -2.21965775e-01 1.46409392e-01 -1.67455465e-01
5.53645432e-01 3.99698257e-01 3.80058005e-03 -6.14528537e-01
4.92314100e-02 9.53620195e-01 -1.49211749e-01 -3.15051883e-01
1.39749110e+00 -4.70678627e-01 -8.90334323e-02 1.98024604e-02
1.59800065e+00 -1.37479007e-01 -1.76023591e+00 -1.15207329e-01
-6.13199472e-01 -9.19227302e-01 4.65502799e-01 -7.85923302e-01
-1.51616228e+00 9.51418936e-01 7.63267815e-01 1.09076202e-01
1.53863835e+00 -4.45119381e-01 1.20247245e+00 1.60300523e-01
2.84922600e-01 -1.11888921e+00 5.51522374e-01 2.37818331e-01
1.17069459e+00 -1.45190763e+00 1.76343411e-01 -2.26354361e-01
-5.83130538e-01 1.07604206e+00 4.51649964e-01 -3.34494144e-01
1.63522288e-01 2.73186177e-01 2.49809071e-01 8.82590935e-02
-3.75494629e-01 -2.84494292e-02 1.40931040e-01 5.34518361e-01
2.61015326e-01 -2.67466694e-01 -1.79973871e-01 -3.27563375e-01
7.93325827e-02 -6.82240352e-02 6.43152118e-01 6.55752242e-01
-4.01440859e-01 -1.22211623e+00 -4.70818192e-01 -5.32250432e-03
-4.88155603e-01 -1.91462696e-01 4.70168237e-03 7.16622710e-01
-1.36565566e-01 7.34296203e-01 -2.78135866e-01 -1.47405729e-01
1.61540762e-01 -5.06076813e-01 9.15678799e-01 -2.73050308e-01
-6.58755660e-01 6.75433949e-02 -2.97375202e-01 -8.51654828e-01
-4.59990948e-01 -3.12888592e-01 -1.06846488e+00 -4.63255227e-01
-3.35316122e-01 -3.97922695e-01 5.95158279e-01 5.45635819e-01
3.48569900e-01 4.07619566e-01 1.02568376e+00 -1.38175082e+00
-3.37782770e-01 -5.52530468e-01 -4.88537520e-01 4.49487031e-01
6.40343428e-01 -4.14917827e-01 -4.07290787e-01 1.60359994e-01] | [10.867100715637207, -2.178257942199707] |
1f8f697f-16bc-445f-b61d-80096dcc2c83 | few-shot-learning-for-video-object-detection | 2103.14724 | null | https://arxiv.org/abs/2103.14724v3 | https://arxiv.org/pdf/2103.14724v3.pdf | When Few-Shot Learning Meets Video Object Detection | Different from static images, videos contain additional temporal and spatial information for better object detection. However, it is costly to obtain a large number of videos with bounding box annotations that are required for supervised deep learning. Although humans can easily learn to recognize new objects by watching only a few video clips, deep learning usually suffers from overfitting. This leads to an important question: how to effectively learn a video object detector from only a few labeled video clips? In this paper, we study the new problem of few-shot learning for video object detection. We first define the few-shot setting and create a new benchmark dataset for few-shot video object detection derived from the widely used ImageNet VID dataset. We employ a transfer-learning framework to effectively train the video object detector on a large number of base-class objects and a few video clips of novel-class objects. By analyzing the results of two methods under this framework (Joint and Freeze) on our designed weak and strong base datasets, we reveal insufficiency and overfitting problems. A simple but effective method, called Thaw, is naturally developed to trade off the two problems and validate our analysis. Extensive experiments on our proposed benchmark datasets with different scenarios demonstrate the effectiveness of our novel analysis in this new few-shot video object detection problem. | ['Jiebo Luo', 'Sebastian Raschka', 'Lin Chen', 'Gaoang Wang', 'Zhongjie Yu'] | 2021-03-26 | null | null | null | null | ['few-shot-video-object-detection'] | ['computer-vision'] | [ 1.16081543e-01 -4.54163194e-01 -1.99338064e-01 -3.77859473e-01
-8.34072709e-01 -3.81783009e-01 3.16570789e-01 -1.64095417e-01
-5.38456261e-01 4.30995136e-01 -2.99802154e-01 1.96735173e-01
-2.42740437e-02 -5.16427040e-01 -1.18196809e+00 -6.16242051e-01
-2.86859035e-01 2.64557358e-02 9.10598814e-01 -8.25127494e-03
1.91335864e-02 1.24378517e-01 -1.69143069e+00 4.13910210e-01
6.41910315e-01 1.22394943e+00 5.52848041e-01 5.85228860e-01
1.64770529e-01 1.22175622e+00 -4.90690351e-01 -1.99598834e-01
5.82384944e-01 -4.69594419e-01 -4.57974404e-01 3.94563258e-01
7.24431336e-01 -9.50094700e-01 -4.44636256e-01 1.25700593e+00
3.56351852e-01 4.19562340e-01 4.18149620e-01 -1.45297170e+00
-3.79120409e-01 4.81342852e-01 -6.55493557e-01 8.77168775e-01
3.26474279e-01 4.21445340e-01 6.38090730e-01 -1.02993071e+00
6.21824145e-01 9.22912776e-01 7.46352792e-01 6.77511036e-01
-5.64750314e-01 -6.94135249e-01 1.84776455e-01 4.50626671e-01
-1.47174525e+00 -4.82731491e-01 7.40160942e-01 -5.57797849e-01
6.48673534e-01 4.67359647e-02 6.57235444e-01 1.22608447e+00
-1.34904191e-01 9.72322047e-01 6.42124116e-01 -1.70198426e-01
3.63183677e-01 2.85284400e-01 2.31068283e-01 7.11397886e-01
1.28465131e-01 2.23782703e-01 -3.54532987e-01 7.02129900e-02
6.04262888e-01 5.87343574e-01 -3.00919920e-01 -4.67142612e-01
-1.05166507e+00 6.62542045e-01 3.34445953e-01 3.52147073e-01
-1.70820206e-01 1.61965773e-01 6.02143586e-01 3.42207611e-01
2.68804789e-01 -6.26855791e-02 -3.76107574e-01 -1.32884607e-01
-9.18744087e-01 1.79623485e-01 5.37590921e-01 1.23024511e+00
8.23977232e-01 1.85494542e-01 -3.31399828e-01 7.51761615e-01
-8.86657462e-02 4.27586347e-01 4.72976655e-01 -7.31571555e-01
3.62675071e-01 3.52602065e-01 1.94709942e-01 -1.04746389e+00
-1.39235049e-01 -3.06664407e-01 -5.98876417e-01 -5.44685535e-02
5.01043022e-01 -1.16634808e-01 -8.66394579e-01 1.53992176e+00
3.11425745e-01 8.12928796e-01 -1.86510459e-01 1.18121195e+00
8.24519932e-01 8.86051357e-01 1.98553130e-01 -4.81567681e-01
1.24287391e+00 -1.16194379e+00 -5.77596962e-01 -2.37241179e-01
7.21272826e-01 -2.88335323e-01 1.00926983e+00 2.72360891e-01
-8.96799743e-01 -8.50754678e-01 -1.13506830e+00 2.90681392e-01
-3.93200070e-01 3.29709388e-02 5.21470964e-01 5.32318115e-01
-7.25472152e-01 4.63159144e-01 -6.37597978e-01 -5.15976369e-01
6.16579294e-01 2.34602913e-01 -3.55356961e-01 -2.98831373e-01
-1.17468297e+00 5.12973964e-01 6.82583153e-01 3.39050628e-02
-1.36031985e+00 -6.85886145e-01 -8.16140592e-01 1.11399040e-01
9.09289539e-01 -4.86756474e-01 1.25108993e+00 -1.48474658e+00
-1.03828895e+00 7.55585849e-01 2.59671450e-01 -5.50297081e-01
5.74501574e-01 -3.89494658e-01 -4.10907090e-01 4.11258668e-01
1.27442434e-01 6.20294750e-01 1.17213750e+00 -9.82972443e-01
-9.76354659e-01 -2.32571766e-01 2.98765600e-01 -4.96625677e-02
-6.10121608e-01 3.56039315e-01 -6.58443391e-01 -6.33891165e-01
-4.31465030e-01 -6.63952947e-01 6.77503943e-02 8.21808353e-02
2.26231311e-02 -1.76703513e-01 1.12126505e+00 -3.68110865e-01
1.12735140e+00 -2.28118873e+00 -1.21354751e-01 -2.63503402e-01
1.63545147e-01 6.01334691e-01 -2.15926021e-01 -9.57520232e-02
-5.67618385e-02 -1.51521638e-01 -1.59088876e-02 2.24733278e-02
-3.31779748e-01 5.59186600e-02 -3.64352763e-01 4.31886256e-01
2.91223168e-01 8.05293500e-01 -1.09955335e+00 -7.15322256e-01
3.13690901e-01 2.07908332e-01 -5.51065147e-01 4.47466135e-01
-1.64438352e-01 5.00502847e-02 -3.90057564e-01 1.04842544e+00
6.61643088e-01 -2.86101282e-01 -1.86845154e-01 -1.81361720e-01
1.19488277e-01 -4.69115436e-01 -1.30011702e+00 1.66216874e+00
1.34781823e-01 5.96671760e-01 -2.85024136e-01 -1.28797686e+00
6.67926729e-01 2.36071125e-01 6.03286505e-01 -6.08374298e-01
3.68310750e-01 1.09272346e-01 -1.01373538e-01 -1.09105861e+00
1.83549389e-01 -3.46656561e-01 5.28857708e-02 2.08226994e-01
5.23848116e-01 5.53826332e-01 5.11729300e-01 1.84863999e-01
1.31078279e+00 -1.88030005e-02 3.96394670e-01 3.07335015e-02
3.57513428e-01 -3.10583673e-02 9.82125342e-01 9.60885465e-01
-7.99205244e-01 7.02365816e-01 3.54712516e-01 -8.92726123e-01
-1.05935454e+00 -8.50374818e-01 4.94729169e-02 1.53391778e+00
4.18574125e-01 -4.10449773e-01 -8.27120066e-01 -9.35418129e-01
-1.02293104e-01 2.98441350e-01 -7.28120804e-01 -3.37720841e-01
-6.28050566e-01 -8.25401247e-01 3.13688606e-01 8.15210164e-01
4.87650454e-01 -1.07678103e+00 -8.67133796e-01 1.71567261e-01
3.68672535e-02 -1.41519380e+00 -4.69378799e-01 -3.12293079e-02
-7.22196817e-01 -1.28063810e+00 -9.50251818e-01 -9.84773695e-01
5.15137434e-01 8.13446581e-01 7.88925409e-01 1.35266155e-01
-4.68461245e-01 4.76418734e-01 -6.61510408e-01 -4.35942888e-01
-2.41597176e-01 -2.11073011e-01 3.22363108e-01 3.15412819e-01
6.08908474e-01 -3.05436969e-01 -6.73674285e-01 4.52199072e-01
-1.10220599e+00 -1.18961349e-01 6.98912978e-01 6.76869571e-01
3.20196599e-01 1.90803297e-02 6.60098732e-01 -5.93919992e-01
9.33770314e-02 -6.63102567e-01 -6.26471937e-01 4.17842478e-01
9.17758867e-02 -3.22421312e-01 6.69262409e-01 -8.05252492e-01
-7.99367607e-01 2.82059133e-01 3.23070288e-01 -1.08692014e+00
-1.36762470e-01 2.24356279e-01 -9.88824442e-02 -8.85601789e-02
6.39385283e-01 3.58124822e-01 -2.00779766e-01 -3.35199505e-01
2.37921014e-01 6.40751541e-01 6.04555845e-01 -2.80350298e-01
8.49638462e-01 5.68748653e-01 -4.88020271e-01 -8.92559171e-01
-9.58615720e-01 -7.56063879e-01 -7.88575470e-01 -6.29935086e-01
8.46200764e-01 -1.13653636e+00 -4.38950658e-01 3.95843595e-01
-1.07876587e+00 -1.71789527e-01 -2.52708554e-01 4.90487456e-01
-6.20528102e-01 3.81406784e-01 -3.25579226e-01 -6.65966988e-01
-2.36982584e-01 -9.89410281e-01 1.02518415e+00 3.20853710e-01
2.57959455e-01 -6.04070723e-01 3.07353698e-02 1.86809704e-01
9.05521140e-02 2.27646768e-01 4.46358800e-01 -7.00638115e-01
-8.48824084e-01 -4.77789581e-01 -3.04998577e-01 5.01150131e-01
-2.33600363e-02 5.19606136e-02 -9.18331027e-01 -5.22869110e-01
2.71720618e-01 -4.87959564e-01 1.15354395e+00 3.21595252e-01
1.28031158e+00 -1.40050963e-01 -3.89197022e-01 7.36225426e-01
1.64338708e+00 3.74032974e-01 5.03280163e-01 2.15502739e-01
7.45387614e-01 2.17809424e-01 9.23001409e-01 4.69037026e-01
-3.55605781e-02 6.21174693e-01 3.00399572e-01 1.50753334e-01
1.08756900e-01 -1.12354904e-01 6.19146526e-01 7.16568291e-01
-3.55099857e-01 -9.44722742e-02 -9.20749426e-01 3.93451661e-01
-2.07208824e+00 -1.35841215e+00 4.65072036e-01 2.04605675e+00
4.87960875e-01 2.80863106e-01 4.25131530e-01 1.72862262e-02
9.26465929e-01 1.15393929e-01 -5.58060050e-01 3.11813056e-01
4.88965623e-02 -3.13895106e-01 3.35153013e-01 -1.77861601e-01
-1.57401001e+00 8.79779994e-01 5.95433140e+00 9.07430291e-01
-1.15484786e+00 3.24210554e-01 5.64009786e-01 -3.40499669e-01
4.64481413e-01 -6.83265328e-02 -1.00949931e+00 6.23719633e-01
7.51736939e-01 -2.44496167e-01 2.25950122e-01 1.32568622e+00
5.89627028e-02 9.38344933e-03 -1.36285067e+00 1.33766937e+00
5.27046740e-01 -1.34068239e+00 2.23292798e-01 -4.72173363e-01
7.86036909e-01 -2.40581986e-02 -1.20165415e-01 7.39166915e-01
-3.37837934e-01 -6.18966043e-01 7.13815808e-01 4.90561515e-01
6.52320743e-01 -4.78211045e-01 6.04063928e-01 4.57872778e-01
-1.30157828e+00 -7.48457313e-01 -7.20904112e-01 -2.48694628e-01
1.69235170e-02 1.43642664e-01 -4.57813889e-01 2.11725906e-01
9.95675981e-01 1.06221330e+00 -6.86045051e-01 1.53752768e+00
2.78190494e-01 4.56380844e-01 -5.72935678e-02 7.31692612e-02
3.70393336e-01 7.66268447e-02 6.30357981e-01 9.40569401e-01
3.47033262e-01 4.24707174e-01 3.09141964e-01 7.39878356e-01
-1.97106794e-01 4.95595904e-03 -7.98713624e-01 1.36080816e-01
2.94446796e-01 1.38001788e+00 -9.51335609e-01 -6.62386417e-01
-8.72461975e-01 7.51867592e-01 4.21940476e-01 4.19541270e-01
-1.25736070e+00 -3.46384317e-01 1.72391370e-01 3.23740423e-01
6.16947055e-01 2.94658840e-02 4.90661472e-01 -1.48351157e+00
1.29471004e-01 -9.81663048e-01 6.33364677e-01 -7.87482738e-01
-1.25471807e+00 4.14964080e-01 1.45587564e-01 -1.70441365e+00
-5.76213188e-02 -6.99664712e-01 -8.37963760e-01 -1.43686190e-01
-1.30816317e+00 -9.55269396e-01 -6.25057936e-01 7.52186179e-01
8.71350765e-01 -3.62626970e-01 2.02752680e-01 7.33193636e-01
-8.26944053e-01 6.06046796e-01 1.92199752e-01 4.22122687e-01
7.52615869e-01 -8.89096022e-01 3.57909575e-02 8.53144109e-01
2.75748670e-01 3.22820038e-01 5.46074808e-01 -4.69445229e-01
-1.62653720e+00 -1.29172289e+00 9.18927118e-02 -4.94678110e-01
5.48636854e-01 -4.36205357e-01 -1.08935559e+00 9.26325142e-01
-9.87912714e-02 5.51991224e-01 4.58529741e-01 -3.40692937e-01
-1.89697295e-01 -4.01441246e-01 -8.09447289e-01 2.12385044e-01
1.14042664e+00 -2.49266684e-01 -8.89860034e-01 5.09157360e-01
8.84209096e-01 -9.16944221e-02 -4.06019479e-01 6.96436226e-01
5.07795691e-01 -9.63646531e-01 1.07987487e+00 -8.93832028e-01
2.86330223e-01 -2.54973769e-01 -2.51696080e-01 -9.40043032e-01
-1.82312161e-01 -5.49308956e-01 -4.00728464e-01 1.05688977e+00
-2.12728634e-01 -7.72717893e-02 7.21746743e-01 5.07414758e-01
-7.57601671e-03 -6.58323407e-01 -6.93429530e-01 -1.24226344e+00
-2.80786604e-01 -3.86988193e-01 2.69853681e-01 9.56688344e-01
-3.14195128e-03 2.17479214e-01 -7.24246740e-01 1.09901868e-01
7.68583536e-01 1.01639800e-01 9.28575516e-01 -1.03075266e+00
-1.98664561e-01 -2.17325673e-01 -8.22362900e-01 -8.58648717e-01
-1.26548052e-01 -5.90783417e-01 1.42893746e-01 -8.78693223e-01
9.05317843e-01 1.20528433e-02 -6.72215700e-01 2.36628130e-01
-2.93231338e-01 3.82297903e-01 3.19152236e-01 3.02659452e-01
-1.43903029e+00 5.86822450e-01 9.13143277e-01 -2.42521644e-01
-1.03677005e-01 -6.13170564e-02 -1.92164764e-01 1.07098198e+00
4.44902301e-01 -6.57498598e-01 -3.66994530e-01 -3.10396940e-01
8.10825750e-02 7.70852864e-02 6.63193643e-01 -1.40504110e+00
2.84306437e-01 -1.05093770e-01 4.12771225e-01 -6.44397140e-01
1.21353664e-01 -8.03880095e-01 -2.30022982e-01 4.89636749e-01
-1.59905538e-01 -3.49041492e-01 1.51347846e-01 9.19266105e-01
-3.04431766e-01 -2.74713904e-01 8.74092758e-01 -3.20360869e-01
-1.56166518e+00 5.43151915e-01 -1.67943090e-01 1.68666855e-01
1.60974479e+00 -3.78496110e-01 -2.44180366e-01 -1.06440067e-01
-7.57608235e-01 2.79624999e-01 2.47953743e-01 6.64974689e-01
8.07348549e-01 -1.43697655e+00 -5.31813502e-01 1.32572323e-01
3.83898139e-01 -2.87362516e-01 5.40214181e-01 8.81517053e-01
-4.38810438e-01 1.43144235e-01 -5.45177817e-01 -8.37953448e-01
-1.20294452e+00 1.18610144e+00 3.78479838e-01 3.88403952e-01
-8.04371774e-01 8.34298790e-01 5.37860155e-01 7.91439265e-02
4.69315171e-01 -4.36776936e-01 -1.73405439e-01 1.46672249e-01
9.78846431e-01 4.76368904e-01 -3.59371573e-01 -5.25614023e-01
-3.13445449e-01 5.21618366e-01 -2.58066833e-01 5.22537112e-01
1.34185994e+00 -1.37864828e-01 3.10897052e-01 6.02234900e-01
1.31543863e+00 -6.82454765e-01 -1.64797580e+00 -3.22455347e-01
-6.25872388e-02 -7.12440431e-01 -1.46648020e-01 -1.80898860e-01
-1.10543919e+00 9.80023026e-01 8.32285941e-01 1.69215947e-01
1.05688810e+00 4.28381264e-02 1.01045990e+00 6.89200997e-01
3.90374750e-01 -1.29068995e+00 6.36332333e-01 2.86995828e-01
5.52992702e-01 -1.78831542e+00 -1.73271716e-01 -1.69504955e-01
-6.17171884e-01 1.06467342e+00 9.18862522e-01 -2.65283465e-01
6.78102732e-01 1.66723102e-01 -1.62067503e-01 -2.75034666e-01
-7.15385377e-01 -2.40565687e-01 1.48328930e-01 4.95759517e-01
-2.31522113e-01 -3.11922371e-01 5.88285401e-02 8.86882603e-01
4.93965119e-01 4.01207060e-01 5.45021296e-01 9.66413915e-01
-8.97902846e-01 -3.64012241e-01 -3.34268451e-01 4.85733688e-01
-3.79432052e-01 2.26146787e-01 -3.52132544e-02 9.15480733e-01
3.05923402e-01 5.85455716e-01 2.29535788e-01 -4.83824193e-01
8.50046575e-02 -6.24196045e-02 3.69132996e-01 -6.61181688e-01
-1.14824496e-01 -2.49764249e-02 -4.78697479e-01 -6.25371635e-01
-7.20433414e-01 -4.95218277e-01 -9.75369751e-01 4.59620357e-02
-4.93694812e-01 -6.56980798e-02 1.84170797e-01 1.01647782e+00
1.82586834e-02 3.66999894e-01 6.29782319e-01 -9.18275833e-01
-7.11175621e-01 -8.75454903e-01 -7.73137152e-01 6.83054030e-01
3.64327997e-01 -8.87535393e-01 -3.52964163e-01 4.41491127e-01] | [8.730405807495117, 0.8950499296188354] |
a4dba4f4-61d0-499f-88a1-6fb18d94f38a | unsupervised-text-style-transfer-using | 1805.11749 | null | http://arxiv.org/abs/1805.11749v3 | http://arxiv.org/pdf/1805.11749v3.pdf | Unsupervised Text Style Transfer using Language Models as Discriminators | Binary classifiers are often employed as discriminators in GAN-based
unsupervised style transfer systems to ensure that transferred sentences are
similar to sentences in the target domain. One difficulty with this approach is
that the error signal provided by the discriminator can be unstable and is
sometimes insufficient to train the generator to produce fluent language. In
this paper, we propose a new technique that uses a target domain language model
as the discriminator, providing richer and more stable token-level feedback
during the learning process. We train the generator to minimize the negative
log likelihood (NLL) of generated sentences, evaluated by the language model.
By using a continuous approximation of discrete sampling under the generator,
our model can be trained using back-propagation in an end- to-end fashion.
Moreover, our empirical results show that when using a language model as a
structured discriminator, it is possible to forgo adversarial steps during
training, making the process more stable. We compare our model with previous
work using convolutional neural networks (CNNs) as discriminators and show that
our approach leads to improved performance on three tasks: word substitution
decipherment, sentiment modification, and related language translation. | ['Taylor Berg-Kirkpatrick', 'Zhiting Hu', 'Chris Dyer', 'Zichao Yang', 'Eric P. Xing'] | 2018-05-30 | unsupervised-text-style-transfer-using-1 | http://papers.nips.cc/paper/7959-unsupervised-text-style-transfer-using-language-models-as-discriminators | http://papers.nips.cc/paper/7959-unsupervised-text-style-transfer-using-language-models-as-discriminators.pdf | neurips-2018-12 | ['decipherment'] | ['natural-language-processing'] | [ 6.48176193e-01 4.66112286e-01 -3.69906314e-02 -3.69546592e-01
-7.91447997e-01 -8.68640721e-01 7.06363559e-01 -1.11245766e-01
-4.93721128e-01 1.10051191e+00 1.19817473e-01 -3.26630771e-01
8.39942217e-01 -1.00304651e+00 -1.00027132e+00 -5.20835400e-01
5.17907798e-01 4.04079676e-01 1.47861661e-02 -4.83128756e-01
-5.64284995e-02 4.03487802e-01 -9.70408380e-01 4.40607101e-01
1.09319019e+00 6.54797435e-01 1.23312831e-01 8.59704971e-01
-8.24689493e-02 8.31757188e-01 -1.17136633e+00 -7.78421402e-01
3.01321656e-01 -1.16823506e+00 -8.08107018e-01 -1.50640592e-01
2.02218011e-01 -1.15331016e-01 1.19490437e-02 1.18716586e+00
6.40253961e-01 -1.21857226e-01 8.33544075e-01 -8.97834539e-01
-7.27793097e-01 7.64918685e-01 -1.14244774e-01 -2.16794208e-01
3.85414422e-01 1.86313301e-01 7.89209545e-01 -6.79648161e-01
7.01711655e-01 1.33413804e+00 6.73353136e-01 1.05536866e+00
-1.59636724e+00 -7.09134758e-01 -1.36903286e-01 -4.48259026e-01
-1.01460922e+00 -4.85774517e-01 8.03316712e-01 -2.17752278e-01
8.30721796e-01 3.04235071e-01 4.13627625e-01 1.54648411e+00
5.16299486e-01 6.17458403e-01 1.12636125e+00 -7.96452701e-01
3.31491888e-01 6.16939306e-01 -6.36667430e-01 5.40187001e-01
-4.35360707e-02 2.37764239e-01 -2.37532929e-01 -5.78848571e-02
6.09585404e-01 -4.58919883e-01 -1.63381502e-01 -7.66920894e-02
-7.77902067e-01 1.05374384e+00 4.75835741e-01 4.35925990e-01
-2.35309392e-01 2.50315577e-01 4.43162531e-01 7.92191327e-01
8.29084218e-01 8.02897155e-01 -1.22764610e-01 -9.07172542e-03
-1.08939409e+00 2.36385018e-01 9.81355071e-01 8.16618323e-01
6.65058553e-01 3.75839949e-01 -4.73941714e-01 8.03616226e-01
-2.85619013e-02 5.36127388e-01 8.12212825e-01 -4.59430009e-01
6.29596174e-01 3.27660531e-01 -6.55790865e-02 -5.77980638e-01
1.74342692e-01 -5.79345167e-01 -1.03973353e+00 5.98877847e-01
3.55723679e-01 -5.62621832e-01 -9.83784080e-01 2.00857568e+00
-3.86878923e-02 -1.75547287e-01 3.97236288e-01 4.62287575e-01
2.77724296e-01 7.42536545e-01 -3.85556668e-02 -5.17019443e-02
8.74813199e-01 -1.02507770e+00 -6.11267865e-01 -3.74940336e-01
8.01719964e-01 -8.74276638e-01 1.26974642e+00 2.13789210e-01
-1.20213675e+00 -6.23763800e-01 -1.14339089e+00 9.91435423e-02
-2.58650392e-01 2.37877015e-02 5.44343181e-02 7.83930004e-01
-1.15849209e+00 8.54557931e-01 -6.17356241e-01 -2.28577226e-01
3.45230222e-01 3.26297343e-01 -1.97785407e-01 2.26790830e-01
-1.49646544e+00 1.16607606e+00 3.86877865e-01 -2.20709503e-01
-8.15525591e-01 -4.68549162e-01 -8.83452058e-01 -4.47997078e-02
-1.88470662e-01 -8.15880954e-01 1.37728286e+00 -1.83806491e+00
-2.09045839e+00 7.76091397e-01 -3.76153588e-02 -6.99909031e-01
1.09011424e+00 7.36778155e-02 -1.95835590e-01 -1.04969710e-01
-4.22125077e-03 7.68269718e-01 1.13177395e+00 -1.10162687e+00
-2.94518113e-01 2.46822909e-01 -6.09937916e-03 1.00136392e-01
-3.19565028e-01 -2.74801813e-03 1.44675851e-01 -1.09682024e+00
-5.98350286e-01 -1.01505792e+00 -3.54549050e-01 -1.76048204e-01
-3.54244500e-01 -4.52056341e-02 6.27690494e-01 -8.70712221e-01
1.02404332e+00 -2.02984881e+00 2.28384852e-01 1.88502789e-01
-2.65989631e-01 6.00130975e-01 -3.18164021e-01 5.01504302e-01
-1.16760939e-01 2.71925598e-01 -6.87604725e-01 -6.98365033e-01
-1.26676187e-01 2.03193203e-01 -5.57388306e-01 1.71419904e-01
7.27711201e-01 1.10728276e+00 -9.26457644e-01 -2.13599503e-01
-1.76824674e-01 4.94257540e-01 -5.59294879e-01 5.03724575e-01
-3.57977301e-01 6.92195415e-01 -8.19325224e-02 1.85603872e-02
5.08083940e-01 2.49052227e-01 5.52575700e-02 2.85398245e-01
2.62778342e-01 6.41247451e-01 -7.94380307e-01 1.59915137e+00
-1.05934846e+00 7.52062142e-01 -2.23526105e-01 -1.01186574e+00
1.19607723e+00 4.65576947e-01 -3.04980814e-01 -6.36740983e-01
1.14421010e-01 3.84026170e-01 -3.38723287e-02 -5.04729114e-02
4.24828678e-01 -5.19574404e-01 -3.02124351e-01 5.84744811e-01
1.21898301e-01 -5.74768603e-01 9.05747190e-02 1.79525074e-02
9.95230436e-01 5.30691892e-02 2.38246769e-02 -8.89286473e-02
5.80571413e-01 -2.03114733e-01 2.89518535e-01 7.36996174e-01
3.86474162e-01 9.07554030e-01 5.84279120e-01 2.67808046e-02
-1.37314081e+00 -8.51037145e-01 2.74881095e-01 6.96194112e-01
-3.78743291e-01 -1.09950989e-01 -1.13240027e+00 -1.00937486e+00
-3.34548682e-01 1.05039787e+00 -5.83456337e-01 -6.01592779e-01
-7.61400878e-01 -5.00200331e-01 1.05265725e+00 5.73944569e-01
3.78164679e-01 -1.31561112e+00 -3.33288282e-01 4.03098226e-01
-7.73601532e-02 -7.98012733e-01 -6.70250654e-01 3.80167574e-01
-8.88856769e-01 -3.70420277e-01 -1.02815771e+00 -9.75942850e-01
9.07938242e-01 -4.35981154e-01 1.16485369e+00 -8.28502104e-02
1.40022695e-01 -2.29054615e-01 -3.64789724e-01 -3.63199949e-01
-1.52286971e+00 3.75351220e-01 -1.83043808e-01 8.03542733e-02
-1.15672119e-01 -5.67589283e-01 -1.71481177e-01 -4.25457582e-02
-1.19429600e+00 1.32076696e-01 5.83974719e-01 1.16749871e+00
2.70081401e-01 -2.10453585e-01 7.82192767e-01 -1.24772394e+00
1.00256026e+00 -2.04234004e-01 -5.68802118e-01 1.79565728e-01
-7.12072134e-01 4.88060951e-01 1.22822762e+00 -7.63902426e-01
-1.05806088e+00 7.88760781e-02 -5.26016474e-01 -3.03660870e-01
-7.37913400e-02 2.83483654e-01 -2.67821997e-01 1.72147918e-02
1.05733120e+00 4.28202212e-01 1.17145963e-01 -4.01212454e-01
4.36121821e-01 8.09533954e-01 3.63800645e-01 -3.16791445e-01
9.88744318e-01 5.49456030e-02 -2.07413927e-01 -4.41936910e-01
-6.17973983e-01 2.89043576e-01 -4.42998290e-01 6.62616640e-02
6.76897645e-01 -8.29385877e-01 -4.93586138e-02 6.09643400e-01
-1.44028819e+00 -6.39061451e-01 -5.63142836e-01 2.03081682e-01
-5.74730635e-01 1.05932705e-01 -6.63456202e-01 -6.86980426e-01
-6.39185429e-01 -1.03591859e+00 9.34990406e-01 7.28878528e-02
-4.69090462e-01 -1.29477370e+00 2.74243414e-01 -8.85717943e-02
7.13147819e-01 4.26179409e-01 9.18393493e-01 -6.78921580e-01
-9.31775942e-02 -3.87441039e-01 3.12198430e-01 1.03906643e+00
3.32809120e-01 -1.62869319e-01 -1.00900507e+00 -4.36997384e-01
1.15029037e-01 -4.56488460e-01 8.04958403e-01 -1.95481509e-01
8.90409529e-01 -7.45451808e-01 6.74888119e-02 5.29704571e-01
1.27526915e+00 3.13002467e-02 8.56262624e-01 1.54339358e-01
5.10287404e-01 5.22528708e-01 2.82516807e-01 -5.22958413e-02
-1.92224056e-01 5.75930595e-01 6.57522678e-02 -4.65456456e-01
-3.77460182e-01 -6.31815195e-01 9.00558174e-01 7.32338667e-01
3.28950226e-01 -5.28433084e-01 -5.06936193e-01 4.33785856e-01
-1.44132912e+00 -9.55738306e-01 1.39306441e-01 2.14435339e+00
1.41837144e+00 3.60362768e-01 1.04830042e-01 1.46625206e-01
7.95590878e-01 5.87322153e-02 -3.40179145e-01 -9.54746842e-01
-2.74352044e-01 7.46651232e-01 4.95228827e-01 8.04526567e-01
-8.79828453e-01 1.09447932e+00 6.18930292e+00 8.22152436e-01
-1.64530075e+00 2.29362160e-01 7.34550297e-01 9.16654542e-02
-6.04886174e-01 -1.07223444e-01 -4.54202771e-01 6.83840632e-01
1.04875124e+00 -1.22018494e-01 4.46136415e-01 7.85764396e-01
1.63624004e-01 1.72510654e-01 -1.20743263e+00 5.48799813e-01
1.18732400e-01 -1.14572442e+00 5.46339825e-02 -2.16282189e-01
9.52832401e-01 -2.39154130e-01 1.81074262e-01 2.58844405e-01
4.78479862e-01 -1.16882920e+00 1.04994738e+00 2.64772594e-01
1.06701279e+00 -9.01627123e-01 8.20121169e-01 4.82927412e-01
-5.76749325e-01 3.91534954e-01 -3.37870806e-01 -1.20910794e-01
1.71797827e-01 5.71163416e-01 -1.33937156e+00 4.30385441e-01
5.82711305e-03 3.73228610e-01 -4.84468848e-01 5.39592981e-01
-6.49334311e-01 8.69987726e-01 -1.05319470e-01 -2.45669529e-01
1.31847426e-01 -1.72211900e-01 5.99146903e-01 1.45296919e+00
2.64028460e-01 -5.57942688e-01 -2.00585127e-01 1.20021665e+00
-4.78602380e-01 4.69691791e-02 -8.01126897e-01 -2.50450164e-01
9.31683555e-02 9.56006765e-01 -5.13302684e-01 -3.33494931e-01
1.52088758e-02 1.65462506e+00 3.84259462e-01 3.09160143e-01
-7.77307987e-01 -7.14338183e-01 3.51432204e-01 1.02375615e-02
3.18935543e-01 -5.63450530e-02 -5.09770751e-01 -1.17641103e+00
1.72912017e-01 -1.20883930e+00 -9.60209593e-02 -4.99723345e-01
-1.35856509e+00 1.09156048e+00 -3.68032724e-01 -1.24666107e+00
-7.15622604e-01 -3.49377364e-01 -8.26169789e-01 1.36173820e+00
-1.40163136e+00 -1.08336735e+00 2.15226069e-01 4.62490439e-01
4.76498663e-01 -2.26393119e-01 1.07961714e+00 2.23681808e-01
-1.83009714e-01 1.14376175e+00 1.08624555e-01 3.43394339e-01
8.10483098e-01 -1.25072014e+00 7.69185901e-01 1.02723444e+00
1.15877360e-01 3.65514696e-01 8.53971303e-01 -6.72931254e-01
-8.28189671e-01 -1.40102232e+00 1.20859468e+00 -3.31726521e-01
4.05721754e-01 -7.99322784e-01 -7.92173505e-01 5.10393202e-01
4.59153682e-01 -4.40386057e-01 4.37616259e-01 -3.58497888e-01
-3.23515415e-01 -5.72365895e-03 -1.34577513e+00 5.95709145e-01
6.72891855e-01 -6.21604741e-01 -5.21898329e-01 2.95136660e-01
7.96083808e-01 -4.40229118e-01 -3.30684900e-01 1.06688209e-01
2.68959790e-01 -5.87093651e-01 4.32704359e-01 -6.93540931e-01
6.93260074e-01 -1.42572060e-01 2.37496957e-01 -1.86069512e+00
6.45543588e-03 -7.77452528e-01 2.25554824e-01 1.49661100e+00
9.18631017e-01 -7.79663324e-01 6.93036795e-01 3.15586895e-01
8.44034031e-02 -5.37541270e-01 -9.14452791e-01 -9.99345183e-01
5.98758161e-01 -1.15588047e-01 5.58524370e-01 7.31495261e-01
-2.99328417e-01 6.09142601e-01 -5.42587996e-01 -2.75676429e-01
2.08121628e-01 -9.34501514e-02 7.38352180e-01 -7.94216037e-01
-6.54521763e-01 -3.82105649e-01 -2.13815182e-01 -1.05160272e+00
4.91009802e-01 -1.27232277e+00 2.41757631e-01 -1.19253564e+00
-2.50374973e-01 -4.39209431e-01 -7.89554883e-03 4.40552682e-01
-2.18434691e-01 3.59393597e-01 2.60815084e-01 -7.46577829e-02
8.43443349e-02 6.42368436e-01 1.24201775e+00 -3.00388694e-01
-2.24837527e-01 2.98925102e-01 -6.67172492e-01 3.94204497e-01
9.56397057e-01 -8.77870500e-01 -2.01414555e-01 -5.03018022e-01
1.79827437e-01 -2.82257050e-01 2.10880443e-01 -8.07080507e-01
-1.64049521e-01 1.16806142e-01 4.14555788e-01 1.08076759e-01
1.53592393e-01 -6.25410676e-01 1.10903643e-01 7.80089676e-01
-7.15135574e-01 6.99817166e-02 8.25364143e-02 2.42436171e-01
-2.89560050e-01 -5.46032369e-01 1.07599366e+00 -1.06767282e-01
1.62761956e-01 -1.16527610e-01 -5.39495111e-01 1.16169050e-01
6.92502081e-01 5.37179001e-02 1.74630225e-01 -6.71011746e-01
-3.51558715e-01 -1.85820818e-01 7.37501621e-01 4.22982574e-01
3.32797617e-01 -1.44250119e+00 -9.87631679e-01 3.55262548e-01
-1.98578089e-01 -2.28751376e-02 -3.08752030e-01 3.59782368e-01
-6.38594389e-01 9.84169170e-02 -1.10350549e-01 -3.03850561e-01
-1.05056703e+00 4.18151975e-01 5.77593446e-01 -5.32853961e-01
-3.07424247e-01 9.83797789e-01 -1.69783160e-02 -5.90429246e-01
8.12517479e-02 -2.38902360e-01 1.94473431e-01 -8.59533474e-02
3.78411323e-01 -9.26198885e-02 2.40438223e-01 -3.33373487e-01
-3.63727622e-02 1.77082449e-01 -3.10795935e-04 -5.28786600e-01
1.25337160e+00 1.49242416e-01 -1.51248053e-01 3.71864110e-01
1.32865727e+00 2.95394152e-01 -1.14584947e+00 -2.50239730e-01
-1.83682501e-01 -1.75702840e-01 -3.13169420e-01 -8.95718932e-01
-9.33706701e-01 9.58439350e-01 4.95086998e-01 3.20527315e-01
1.18813300e+00 -2.26340830e-01 9.41470444e-01 7.17478096e-02
1.96191549e-01 -1.01036465e+00 6.96694106e-02 5.91517329e-01
1.01513016e+00 -9.86145437e-01 -6.21254027e-01 -2.16861144e-01
-7.66251445e-01 1.12428296e+00 4.26624864e-01 -3.73881489e-01
2.05841705e-01 4.16013092e-01 3.01788718e-01 6.02486908e-01
-5.54598451e-01 1.67951524e-01 1.87703162e-01 5.52241325e-01
4.64080602e-01 -1.40747372e-02 -4.23248500e-01 3.94774646e-01
-6.51964426e-01 -8.51777047e-02 5.62984169e-01 9.73764896e-01
-1.13617040e-01 -1.84621239e+00 -2.64998853e-01 8.49189833e-02
-6.61651015e-01 -3.82414073e-01 -7.00354874e-01 2.90528268e-01
4.20756750e-02 9.33553576e-01 -2.08097175e-02 -2.88983166e-01
2.77734995e-01 2.92577088e-01 5.06828308e-01 -8.36768508e-01
-9.96380210e-01 -2.57624686e-01 1.70844808e-01 -1.66925818e-01
-1.59737825e-01 -3.21557015e-01 -9.63648081e-01 -1.81560159e-01
-3.98820400e-01 3.46586764e-01 6.89107418e-01 9.22112703e-01
2.51097351e-01 5.71550310e-01 1.05514765e+00 -6.15941882e-01
-9.10075784e-01 -1.12503231e+00 -2.45478719e-01 5.67264080e-01
4.16895688e-01 4.42079641e-02 -2.88627237e-01 3.70596677e-01] | [11.82553482055664, 9.525010108947754] |
f091e59b-0bdf-4467-80ce-40eefcc12913 | psd-principled-synthetic-to-real-dehazing | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Chen_PSD_Principled_Synthetic-to-Real_Dehazing_Guided_by_Physical_Priors_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Chen_PSD_Principled_Synthetic-to-Real_Dehazing_Guided_by_Physical_Priors_CVPR_2021_paper.pdf | PSD: Principled Synthetic-to-Real Dehazing Guided by Physical Priors | Deep learning-based methods have achieved remarkable performance for image dehazing. However, previous studies are mostly focused on training models with synthetic hazy images, which incurs performance drop when the models are used for real-world hazy images. We propose a Principled Synthetic-to-real Dehazing (PSD) framework to improve the generalization performance of dehazing. Starting from a dehazing model backbone that is pre-trained on synthetic data, PSD exploits real hazy images to fine-tune the model in an unsupervised fashion. For the fine-tuning, we leverage several well-grounded physical priors and combine them into a prior loss committee. PSD allows for most of the existing dehazing models as its backbone, and the combination of multiple physical priors boosts dehazing significantly. Through extensive experiments, we demonstrate that our PSD framework establishes the new state-of-the-art performance for real-world dehazing, in terms of visual quality assessed by no-reference quality metrics as well as subjective evaluation and downstream task performance indicator. | ['Dong Liu', 'Yang Yang', 'Yangchao Wang', 'Zeyuan Chen'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['image-dehazing'] | ['computer-vision'] | [ 1.90359682e-01 -1.38454869e-01 4.22825009e-01 -2.02808127e-01
-6.81739926e-01 2.51832530e-02 6.53183281e-01 -1.80851430e-01
-2.02779070e-01 4.85704482e-01 2.09716111e-01 -3.38081084e-02
-4.38467562e-02 -1.01872790e+00 -8.64257038e-01 -1.14162874e+00
9.00446251e-02 -1.72029376e-01 6.09833658e-01 -4.94446218e-01
6.04691803e-02 1.70541346e-01 -1.52620161e+00 1.94741592e-01
1.39851463e+00 9.74188924e-01 2.57224202e-01 6.86683536e-01
5.44341028e-01 8.84378374e-01 -9.85491157e-01 -5.20242572e-01
6.95440769e-01 -2.21670806e-01 -2.61728853e-01 1.45075679e-01
6.86777353e-01 -7.53819346e-01 -6.34446442e-01 1.20621669e+00
5.61568081e-01 2.50887185e-01 7.15626478e-01 -1.23892951e+00
-1.24983239e+00 -6.63412502e-03 -5.95638156e-01 1.73662454e-01
-3.55601996e-01 5.19128740e-01 7.39336073e-01 -8.76323760e-01
-2.46037589e-03 1.13511801e+00 6.82629466e-01 4.42682534e-01
-1.07520139e+00 -8.88304293e-01 -4.19702865e-02 2.86275178e-01
-1.49558258e+00 -4.71324980e-01 7.89923906e-01 -4.12376523e-01
5.78170836e-01 -1.28631130e-01 4.14808899e-01 1.02653635e+00
4.34253633e-01 6.43489122e-01 1.05738616e+00 -3.69012892e-01
2.49251619e-01 2.58667730e-02 -2.41456583e-01 4.67615396e-01
5.08250296e-01 3.79582942e-01 -5.71056426e-01 3.81120533e-01
6.86465502e-01 -2.35863887e-02 -3.96172494e-01 -2.48047844e-01
-9.17812645e-01 8.02188993e-01 6.71999097e-01 -2.02765495e-01
-2.92693555e-01 2.46146694e-01 -1.61454782e-01 2.92542875e-01
6.91985428e-01 7.68080235e-01 -2.03925706e-02 4.44461882e-01
-1.22712529e+00 3.97374302e-01 1.76631942e-01 8.31041098e-01
8.72043848e-01 3.35931063e-01 -4.12752092e-01 1.00479102e+00
3.42822969e-01 6.26873314e-01 2.95583129e-01 -1.04812539e+00
2.40781248e-01 2.25596935e-01 3.61031294e-01 -9.68582392e-01
1.60505712e-01 -5.35580277e-01 -8.90161276e-01 7.74141252e-01
7.75664449e-02 3.70491706e-02 -1.40489805e+00 1.55038655e+00
1.93196893e-01 4.51306581e-01 1.81979313e-01 1.01202643e+00
7.02408731e-01 1.01727819e+00 -3.70405465e-02 1.92329600e-01
1.03644156e+00 -1.44296658e+00 -6.69855356e-01 -2.27784663e-01
1.22146621e-01 -6.34910941e-01 1.29818046e+00 7.05875278e-01
-1.02997589e+00 -5.85958183e-01 -1.55933690e+00 -2.09928259e-01
-3.41542214e-01 -1.15963846e-01 1.37740821e-01 6.36756778e-01
-1.33567107e+00 5.17448127e-01 -8.89654994e-01 -5.49546145e-02
5.72136641e-01 6.20564388e-04 -4.13180664e-02 -3.91568631e-01
-1.46070302e+00 9.03634250e-01 3.54300469e-01 8.60329941e-02
-1.61314285e+00 -1.12083423e+00 -8.23014677e-01 1.13709830e-02
2.40861118e-01 -8.73452246e-01 9.35857236e-01 -6.59094632e-01
-1.79159927e+00 6.63984239e-01 4.47831392e-01 -6.88836277e-01
4.78829145e-01 -5.67421913e-01 -6.51973307e-01 4.47183549e-01
-1.28656760e-01 8.62554610e-01 1.57658184e+00 -1.69313252e+00
-4.41957265e-01 1.31037056e-01 3.68683964e-01 1.70323119e-01
-5.74280262e-01 -2.08715186e-01 -4.47734833e-01 -1.17257261e+00
-4.48994994e-01 -6.11228645e-01 -9.16252136e-02 3.36243242e-01
-3.54975104e-01 3.38553786e-01 8.95448923e-01 -7.80291796e-01
1.24507666e+00 -2.34295106e+00 -6.03165962e-02 -1.80404589e-01
6.12002909e-01 7.19382823e-01 -1.61585718e-01 3.60861599e-01
1.22454889e-01 7.50172213e-02 -5.07070422e-01 -4.99050856e-01
1.83225516e-02 1.18192494e-01 -4.00265455e-01 4.47475225e-01
5.40806949e-01 7.46519566e-01 -9.02430236e-01 -3.95312428e-01
4.49792027e-01 8.32721829e-01 -6.54192328e-01 7.49749482e-01
-3.03982198e-01 3.23334813e-01 -9.90319923e-02 3.68248612e-01
1.03852439e+00 -2.48263374e-01 -6.64154410e-01 -1.78781539e-01
1.54813267e-02 8.42337087e-02 -7.05352426e-01 1.56888866e+00
-5.95955014e-01 5.93816400e-01 7.68943354e-02 -6.73169315e-01
6.19152188e-01 1.23996653e-01 1.93077996e-02 -6.85326040e-01
-1.33676320e-01 2.68868823e-02 -2.53542066e-01 -5.93511641e-01
5.55959821e-01 -3.79935145e-01 3.69052052e-01 1.02274276e-01
1.21859640e-01 -6.22322500e-01 -1.47305027e-01 2.08202481e-01
9.60939705e-01 6.21859692e-02 -2.62531433e-02 -3.21750641e-01
4.96088386e-01 -2.13541538e-01 3.07816893e-01 6.80090845e-01
-2.48511240e-01 1.22159910e+00 4.57722731e-02 -9.12039876e-02
-1.17546105e+00 -1.39096999e+00 -1.78453147e-01 8.69529605e-01
5.38274348e-01 -2.38176376e-01 -1.06691611e+00 -3.40974599e-01
-1.83824241e-01 9.29918349e-01 -8.36658239e-01 -7.59150326e-01
-2.16470689e-01 -9.68881726e-01 4.94986415e-01 -4.12051156e-02
1.10344255e+00 -6.95235491e-01 -3.94629449e-01 8.29848985e-04
-1.42755024e-02 -1.23490572e+00 -4.59431678e-01 -1.92429006e-01
-4.19448406e-01 -6.01060808e-01 -1.02722764e+00 -4.96898383e-01
3.26886922e-01 8.43348622e-01 9.55179691e-01 2.70720989e-01
-1.06078446e-01 1.65616810e-01 -4.76958841e-01 -7.48048723e-01
-5.26575148e-01 -2.48208150e-01 2.50698607e-02 4.63052899e-01
-1.47245154e-01 -9.24614370e-01 -1.29607749e+00 3.92679483e-01
-1.55671322e+00 1.33059993e-01 8.32306743e-01 7.51907408e-01
1.97027773e-01 6.21272385e-01 3.36532593e-01 -5.72551668e-01
2.53542781e-01 -5.74455500e-01 -5.95807552e-01 1.40262201e-01
-8.96455407e-01 5.99037074e-02 5.50722361e-01 -4.66400564e-01
-1.33949816e+00 -6.92438424e-01 -1.09373324e-01 -8.46226096e-01
-5.93807064e-02 3.34024400e-01 -3.44262362e-01 -4.42835748e-01
9.03568685e-01 3.50080639e-01 -9.06831920e-02 -4.13076371e-01
4.03273553e-01 5.63107729e-01 8.55747938e-01 -5.39512575e-01
1.53126359e+00 7.61949480e-01 -2.42742926e-01 -8.64244163e-01
-1.18896031e+00 -3.37402850e-01 -2.74828404e-01 -2.26540398e-02
1.07191455e+00 -1.35127616e+00 -1.31630734e-01 9.81152117e-01
-9.01749074e-01 -7.47451365e-01 -5.09724095e-02 3.68786573e-01
-2.45311737e-01 4.78582114e-01 -5.57423413e-01 -6.58872128e-01
-3.16346407e-01 -1.05440426e+00 1.29181993e+00 8.33011940e-02
5.64009607e-01 -9.76419091e-01 6.00077696e-02 3.58610302e-01
8.02436948e-01 2.45994031e-01 8.56990874e-01 1.23377247e-02
-8.57231259e-01 6.21943437e-02 -4.65806544e-01 9.34680939e-01
4.09920551e-02 -9.63670239e-02 -1.20241773e+00 -4.77317482e-01
2.12274194e-01 -3.12101394e-01 1.19788527e+00 3.55871528e-01
1.19021738e+00 -2.29821712e-01 1.50442705e-01 1.06997669e+00
1.47995043e+00 -1.37509301e-01 1.12276554e+00 5.20801246e-01
8.51641238e-01 5.06797373e-01 4.52344924e-01 1.77154720e-01
3.21846992e-01 7.09243774e-01 8.54993582e-01 -4.10008103e-01
-5.79869807e-01 -1.85812473e-01 4.43834424e-01 5.61214983e-01
3.16886306e-02 -6.42469347e-01 -7.37732649e-01 6.14329159e-01
-1.48506260e+00 -7.10623026e-01 3.28955978e-01 2.16082549e+00
9.08398330e-01 1.90059170e-01 -9.73822176e-02 4.25029993e-02
5.36122084e-01 5.31296492e-01 -3.87218118e-01 1.68515131e-01
-1.60582677e-01 2.24824160e-01 5.12438118e-01 4.67046499e-01
-1.28340447e+00 1.12186098e+00 6.33801699e+00 1.06451178e+00
-1.13153672e+00 2.87081957e-01 5.51453650e-01 -1.24146782e-01
-2.75493264e-01 -1.96952283e-01 -5.10877609e-01 6.82790875e-01
1.01812172e+00 -6.50073308e-03 5.88311613e-01 4.69284803e-01
4.00817931e-01 2.19721757e-02 -7.63799727e-01 1.11113524e+00
2.53012121e-01 -1.31994355e+00 4.91383821e-01 6.58819526e-02
1.09525657e+00 7.33198896e-02 5.65823972e-01 3.09187531e-01
4.06286299e-01 -1.14902484e+00 1.03790116e+00 3.21290851e-01
7.93515682e-01 -4.83537525e-01 6.13221586e-01 1.13666110e-01
-8.14044476e-01 1.30459309e-01 -5.43317676e-01 2.14925632e-01
-9.64613538e-03 9.33347583e-01 -5.00584304e-01 7.74261177e-01
9.09863532e-01 7.89422274e-01 -6.70012891e-01 1.11636555e+00
-5.07018209e-01 9.57280219e-01 3.79342400e-02 7.24301100e-01
3.57134283e-01 -2.02329174e-01 4.74203855e-01 1.16755509e+00
2.66626686e-01 2.00539291e-01 -6.27317950e-02 9.94097233e-01
-1.42074183e-01 -3.57546002e-01 -3.15530390e-01 1.91102773e-01
3.20550293e-01 9.46449399e-01 -1.87701508e-01 -2.45068282e-01
-4.25293475e-01 1.02331722e+00 8.32062438e-02 6.51536942e-01
-1.17893040e+00 -6.42547250e-01 9.76069570e-01 2.01543301e-01
4.42951769e-01 -3.54851961e-01 -1.64542021e-03 -1.24224770e+00
-3.84850167e-02 -9.37431216e-01 -1.41033540e-02 -1.15641451e+00
-1.54163849e+00 4.92647678e-01 1.88222945e-01 -1.35238707e+00
3.63342494e-01 -7.18812048e-01 -7.89192200e-01 8.93484116e-01
-2.29202843e+00 -1.31749117e+00 -8.20959032e-01 5.73693871e-01
5.52163661e-01 1.36918938e-02 3.29729706e-01 3.58024657e-01
-6.76871002e-01 7.79401004e-01 1.35788545e-01 -1.87804736e-02
1.03453636e+00 -1.17189193e+00 3.76030028e-01 1.43403709e+00
-1.72466204e-01 4.30139482e-01 9.60008919e-01 -3.04193348e-01
-1.04001558e+00 -1.54375815e+00 -2.87208445e-02 -6.13387525e-01
7.40851641e-01 -3.08179498e-01 -1.28833210e+00 3.01762640e-01
3.33419174e-01 1.83266997e-01 4.33035314e-01 -6.42931461e-01
-6.61254764e-01 -3.48664314e-01 -9.45608079e-01 7.17805445e-01
8.54007006e-01 -5.16725540e-01 -5.02907276e-01 2.78019935e-01
1.31055725e+00 -3.89006376e-01 -7.65417576e-01 3.98940682e-01
2.98293442e-01 -1.21207380e+00 1.28026438e+00 -2.27105871e-01
7.21996427e-01 -6.84386373e-01 -2.02025145e-01 -1.58737898e+00
-3.65554303e-01 -1.01148248e+00 -2.23080799e-01 9.80543494e-01
2.74486780e-01 -5.75579822e-01 5.86948752e-01 1.18636183e-01
-5.87882876e-01 -5.61473608e-01 -4.60703254e-01 -1.02021158e+00
2.29928046e-01 -3.26128930e-01 7.55959332e-01 8.51968408e-01
-7.84413815e-01 4.26488705e-02 -8.15904140e-01 8.21194232e-01
9.74336267e-01 -3.79153073e-01 8.45285177e-01 -8.24142754e-01
-4.57550466e-01 -2.29405120e-01 -5.43455243e-01 -1.12031651e+00
-2.83681080e-02 -4.01211143e-01 4.00524408e-01 -1.42046142e+00
-3.87011073e-03 -2.33517930e-01 -5.80902576e-01 1.87386289e-01
-6.15378797e-01 5.20738721e-01 3.76623571e-02 4.56113100e-01
-5.16118407e-01 1.07952631e+00 1.37888920e+00 -4.28756386e-01
-7.04886541e-02 -1.90929517e-01 -8.47303033e-01 3.90913546e-01
6.45818651e-01 -4.68443751e-01 -7.53291249e-01 -8.07309330e-01
-1.77217252e-03 -4.80329156e-01 5.66538334e-01 -1.34896541e+00
5.24621345e-02 -3.14464211e-01 2.12738782e-01 -7.18540773e-02
5.09172678e-01 -4.78757709e-01 3.04342676e-02 1.85771286e-01
-1.58134595e-01 -4.07411844e-01 8.12247917e-02 9.37275052e-01
-4.26554292e-01 1.01212874e-01 1.32006943e+00 1.40289351e-01
-6.80216551e-01 6.17855072e-01 -1.26672179e-01 1.44891903e-01
8.39843214e-01 -3.21517169e-01 -6.91266239e-01 -6.16820157e-01
-2.85564184e-01 1.16387777e-01 7.54566073e-01 2.43426770e-01
8.12165082e-01 -1.02430749e+00 -9.02001858e-01 1.68076679e-01
4.07383740e-01 2.95869201e-01 2.77869344e-01 7.70258844e-01
-8.87515664e-01 -4.46218476e-02 -2.52546102e-01 -4.26723689e-01
-7.51507282e-01 9.24729109e-01 5.19437253e-01 -7.16412589e-02
-9.02977765e-01 9.51007962e-01 9.89466190e-01 7.84481391e-02
1.30673766e-01 -2.65620679e-01 6.65742531e-02 -5.92976630e-01
7.42830336e-01 1.61704525e-01 2.45380655e-01 -3.40492308e-01
6.35207668e-02 4.07370508e-01 -7.98604786e-02 -1.63322315e-01
1.36511898e+00 -3.46072137e-01 -3.53774950e-02 1.84842572e-01
1.04428697e+00 6.20910861e-02 -1.83633900e+00 -3.29649568e-01
-4.26456034e-01 -8.67480814e-01 4.60605770e-01 -6.96167290e-01
-1.11294341e+00 1.16970456e+00 6.08253360e-01 5.03521860e-02
1.41376579e+00 -3.46984416e-01 1.04292572e+00 3.11993986e-01
3.08720559e-01 -6.13604069e-01 6.99605942e-01 2.61594802e-01
9.18379903e-01 -1.21045792e+00 2.84539536e-02 -5.32280803e-01
-5.92071593e-01 6.03082240e-01 6.27405524e-01 -3.73899490e-01
8.37186277e-01 4.24835016e-04 2.35938638e-01 -1.80664629e-01
-5.89915872e-01 -4.55340696e-03 4.56598222e-01 8.05676937e-01
-1.66367352e-01 -2.42457733e-01 4.15892571e-01 2.60454893e-01
-2.03243271e-01 -2.51744062e-01 6.94872141e-01 6.49150014e-01
-5.63908100e-01 -5.93786359e-01 -3.93947810e-01 1.37420192e-01
-2.90290296e-01 -4.57919300e-01 -5.16171865e-02 7.23862708e-01
2.21059635e-01 1.11552751e+00 -1.66025236e-01 -5.08763075e-01
2.54937530e-01 -3.89523476e-01 4.79827046e-01 -6.78337216e-01
-2.63432890e-01 -1.48172244e-01 -2.90630162e-01 -4.42947179e-01
-3.88609052e-01 3.62896584e-02 -7.34739780e-01 -4.08579469e-01
-3.04426759e-01 1.14028297e-01 3.81552249e-01 8.49230886e-01
4.40602630e-01 7.13358760e-01 7.11780548e-01 -1.08399343e+00
-5.43438017e-01 -9.91616905e-01 -7.61870623e-01 4.93282676e-01
8.24460924e-01 -1.03123581e+00 -6.32864654e-01 2.86829710e-01] | [10.94159984588623, -3.137730598449707] |
730386b7-d2fe-4001-bfa1-0d97eb1e04e5 | mesa-offline-meta-rl-for-safe-adaptation-and | 2112.03575 | null | https://arxiv.org/abs/2112.03575v1 | https://arxiv.org/pdf/2112.03575v1.pdf | MESA: Offline Meta-RL for Safe Adaptation and Fault Tolerance | Safe exploration is critical for using reinforcement learning (RL) in risk-sensitive environments. Recent work learns risk measures which measure the probability of violating constraints, which can then be used to enable safety. However, learning such risk measures requires significant interaction with the environment, resulting in excessive constraint violations during learning. Furthermore, these measures are not easily transferable to new environments. We cast safe exploration as an offline meta-RL problem, where the objective is to leverage examples of safe and unsafe behavior across a range of environments to quickly adapt learned risk measures to a new environment with previously unseen dynamics. We then propose MEta-learning for Safe Adaptation (MESA), an approach for meta-learning a risk measure for safe RL. Simulation experiments across 5 continuous control domains suggest that MESA can leverage offline data from a range of different environments to reduce constraint violations in unseen environments by up to a factor of 2 while maintaining task performance. See https://tinyurl.com/safe-meta-rl for code and supplementary material. | ['Ken Goldberg', 'Ion Stoica', 'Chelsea Finn', 'Jie Tan', 'Julian Ibarz', 'Suraj Nair', 'Brijen Thananjeyan', 'Ashwin Balakrishna', 'Michael Luo'] | 2021-12-07 | null | null | null | null | ['safe-exploration'] | ['robots'] | [ 1.58320963e-02 7.36477003e-02 -1.31096020e-01 -3.15367758e-01
-1.02731764e+00 -7.46796966e-01 3.14789981e-01 3.97035599e-01
-7.52638936e-01 9.66152549e-01 1.43270746e-01 -3.68428469e-01
-3.47994834e-01 -5.25335073e-01 -9.42297101e-01 -4.32019114e-01
-7.28353620e-01 -3.35602090e-02 1.64826602e-01 -3.00590813e-01
3.34173143e-01 2.74702251e-01 -1.53351092e+00 1.10268593e-04
9.02806282e-01 6.03130460e-01 8.61719474e-02 8.06119502e-01
4.87856746e-01 5.23457646e-01 -6.32297873e-01 2.47170612e-01
5.50360918e-01 -2.16915593e-01 -7.78805017e-01 -6.71642780e-01
3.07698399e-01 -5.40520489e-01 -1.20705895e-01 9.92410123e-01
4.11927283e-01 8.74853313e-01 4.41823721e-01 -1.37853909e+00
-2.61068702e-01 6.73587799e-01 -1.36205778e-01 3.34792346e-01
5.14077067e-01 7.62725949e-01 6.47426903e-01 -2.77483072e-02
3.80639285e-01 1.27174497e+00 4.36456800e-01 1.01364875e+00
-1.21379781e+00 -7.96577811e-01 6.46311343e-01 -1.00920327e-01
-1.00088978e+00 -2.55161315e-01 3.95878136e-01 -4.39335942e-01
1.16843772e+00 1.95324525e-01 7.66357005e-01 1.57230377e+00
6.71932995e-01 4.96280521e-01 1.28351235e+00 -6.49084002e-02
9.06874001e-01 -8.60769004e-02 -1.36643305e-01 4.79253143e-01
3.47634643e-01 9.28324223e-01 -6.81376398e-01 -2.72815764e-01
3.30540866e-01 -1.80982858e-01 -6.38518929e-02 -4.85048324e-01
-9.45705295e-01 6.20154202e-01 4.15072680e-01 -2.31118515e-01
-3.38881947e-02 4.98549998e-01 6.48133934e-01 5.69772065e-01
1.40740657e-02 1.25930321e+00 -5.99492133e-01 -5.78012705e-01
-2.49901354e-01 6.52013361e-01 7.51609385e-01 7.99409270e-01
4.58612412e-01 2.62020081e-01 -2.18913347e-01 3.55511516e-01
1.16553053e-01 5.24893105e-01 2.69323379e-01 -1.20040452e+00
5.08395314e-01 1.59843624e-01 5.24570346e-01 -5.29087901e-01
-5.14278769e-01 -2.51314908e-01 8.17789957e-02 9.02843535e-01
3.46378684e-01 -5.99540591e-01 -6.78569138e-01 2.13196230e+00
5.86827874e-01 1.24729857e-01 6.34843782e-02 6.43828630e-01
-1.74710020e-01 5.10407746e-01 3.07195425e-01 -2.47102380e-01
7.39420533e-01 -7.03670859e-01 -5.37698507e-01 -5.25084138e-01
6.12158716e-01 -1.60960898e-01 1.57299066e+00 8.39012027e-01
-9.78082180e-01 -3.38145047e-01 -1.28959739e+00 4.23316032e-01
-4.04983550e-01 -1.09290898e+00 4.94216412e-01 5.12429118e-01
-8.22928369e-01 1.03166974e+00 -1.16607344e+00 -1.70842543e-01
5.24414837e-01 4.41367447e-01 -9.26114544e-02 2.11638346e-01
-1.17245495e+00 1.09714937e+00 6.50352359e-01 -1.74245939e-01
-1.77562308e+00 -8.33852351e-01 -9.91099834e-01 -2.24284202e-01
9.00713921e-01 -2.58963257e-01 1.50223804e+00 -5.67394555e-01
-1.56421852e+00 4.06668223e-02 4.53390300e-01 -5.56936026e-01
6.03277981e-01 -9.93835568e-01 -2.58572876e-01 -2.45130092e-01
-1.32997185e-01 4.06675488e-01 1.01483142e+00 -1.29479611e+00
-5.50848484e-01 -9.74929035e-02 3.40521723e-01 3.58913600e-01
-2.36478254e-01 -1.99735224e-01 2.79742777e-01 -5.01074970e-01
-6.17850840e-01 -1.11223900e+00 -5.29473960e-01 -1.96608558e-01
-2.00053394e-01 1.73040796e-02 5.42475045e-01 -3.84371370e-01
1.20661354e+00 -2.08585525e+00 2.31522009e-01 3.32399428e-01
-6.16276711e-02 1.54687732e-01 -3.55144739e-01 2.73822248e-01
1.59614906e-01 2.73021847e-01 -2.36349091e-01 1.22986278e-02
1.28412142e-01 1.49303600e-01 -6.40550435e-01 2.87604243e-01
2.40602903e-02 6.34761095e-01 -1.40913045e+00 -1.79869205e-01
1.90640509e-01 1.61774442e-01 -8.04680169e-01 5.84787130e-01
-6.25072956e-01 6.99184537e-01 -5.24017811e-01 5.52005708e-01
1.27662018e-01 5.44515312e-01 1.91697732e-01 4.42574710e-01
-1.17777504e-01 1.71973825e-01 -1.05870247e+00 1.50397301e+00
-7.40694642e-01 2.07528397e-01 -9.84674245e-02 -4.17330265e-01
8.17966819e-01 -1.57738864e-01 4.13127869e-01 -7.63296545e-01
2.59990633e-01 5.07161580e-02 -3.08762938e-02 -5.14294088e-01
4.42656606e-01 -1.02500305e-01 -5.45450807e-01 6.69636428e-01
-1.60546705e-01 -4.27219510e-01 -1.17993318e-01 -1.39538765e-01
1.42900980e+00 5.74387968e-01 2.49373496e-01 -2.75872648e-01
3.77124990e-03 -1.49228871e-01 9.12478149e-01 1.05065906e+00
-7.90636361e-01 -2.05954909e-01 4.51287866e-01 -5.41508436e-01
-7.20227778e-01 -1.42067289e+00 4.95565720e-02 1.29043806e+00
1.62538961e-01 -5.38790345e-01 -8.24414790e-01 -8.55508327e-01
3.20927292e-01 1.21117282e+00 -6.79426372e-01 -8.73630226e-01
-6.75074399e-01 -2.81411767e-01 3.75528395e-01 6.26835704e-01
2.06233963e-01 -1.21962857e+00 -1.41704440e+00 1.48789808e-01
1.76723048e-01 -5.24205327e-01 -5.92690170e-01 6.99729323e-01
-8.72965693e-01 -9.96509492e-01 7.95712546e-02 -2.11042121e-01
4.16006118e-01 -1.38587847e-01 7.65732765e-01 2.59543750e-02
-4.74261940e-01 6.56644702e-01 -1.85107902e-01 -5.26299477e-01
-5.20328104e-01 -1.90264061e-01 8.39372575e-01 -6.92610264e-01
9.43085924e-02 -5.88657498e-01 -5.61536908e-01 2.58460760e-01
-6.72362924e-01 -4.85374242e-01 9.32611823e-02 7.95796871e-01
6.08167112e-01 1.28305733e-01 6.58381760e-01 -6.63429260e-01
9.07425582e-01 -6.37775302e-01 -1.03126693e+00 2.64192909e-01
-9.03609931e-01 4.08055633e-01 7.10200787e-01 -8.73992205e-01
-1.03682458e+00 4.80642468e-02 2.59038508e-01 -3.13710332e-01
-6.20722696e-02 1.23808481e-01 -8.50771964e-02 -1.44106150e-01
1.13677144e+00 -1.77855477e-01 -2.99052782e-02 -1.90713629e-01
3.61950487e-01 1.68848902e-01 3.34727019e-01 -1.22069049e+00
9.76497412e-01 3.97423767e-02 -1.19784269e-02 -1.97039679e-01
-1.06181276e+00 4.19272929e-02 -5.31070650e-01 -4.24459249e-01
7.31922030e-01 -6.48298860e-01 -1.12071145e+00 1.68695778e-01
-3.71708065e-01 -1.38336933e+00 -5.16526520e-01 3.29477429e-01
-1.11611593e+00 1.22432709e-02 -3.02523285e-01 -1.05439019e+00
-1.74174219e-01 -1.02165592e+00 6.39135957e-01 4.48751777e-01
-5.33619463e-01 -9.51587379e-01 4.64866132e-01 9.57675055e-02
4.92960602e-01 6.75219715e-01 7.73541868e-01 -4.96240228e-01
-4.80912268e-01 1.95109963e-01 6.77981257e-01 3.40031743e-01
1.01370737e-01 -2.29278684e-01 -9.18990970e-01 -7.77948797e-01
3.12316846e-02 -9.41933274e-01 6.29610598e-01 7.01109245e-02
1.30959964e+00 -7.00527251e-01 -1.42805338e-01 7.02996731e-01
1.18100786e+00 4.89778757e-01 2.51540065e-01 8.13687623e-01
3.85608792e-01 4.23415333e-01 1.26586282e+00 7.09540665e-01
9.86814424e-02 5.15332580e-01 7.09658086e-01 5.55750191e-01
4.23469931e-01 -5.29260874e-01 9.79870319e-01 1.28311560e-01
3.79591167e-01 1.16922796e-01 -1.04818952e+00 4.74784344e-01
-1.90132308e+00 -1.02866924e+00 7.96573877e-01 2.69090939e+00
1.11761189e+00 6.15525901e-01 4.57999676e-01 -3.08992356e-01
1.21074855e-01 4.46452722e-02 -1.35935903e+00 -9.38775837e-01
5.39818287e-01 6.77306131e-02 4.42354739e-01 8.38144124e-01
-1.14563215e+00 9.59424734e-01 6.80755329e+00 3.25713456e-01
-9.87152874e-01 -8.54755268e-02 4.36071515e-01 -7.28657424e-01
-3.40467274e-01 7.26725012e-02 -6.95958912e-01 3.45784664e-01
1.30290341e+00 -5.06934166e-01 9.50964808e-01 1.17944348e+00
7.39683509e-02 -3.43030453e-01 -1.34076905e+00 4.39213157e-01
-2.20159963e-01 -7.14354515e-01 -3.97900999e-01 -1.71091139e-01
7.37781644e-01 6.54202253e-02 3.73112977e-01 9.12657678e-01
8.70662391e-01 -1.24464500e+00 8.48325551e-01 5.01071036e-01
6.23380482e-01 -1.12323463e+00 2.02076465e-01 4.18851286e-01
-8.92385960e-01 -6.86687827e-01 -2.67933786e-01 -4.48803715e-02
-6.80068061e-02 1.39391748e-02 -8.73733640e-01 -1.12758480e-01
9.51535761e-01 3.51215005e-01 -5.91962218e-01 9.34263527e-01
-3.99048090e-01 4.92856979e-01 -2.92924255e-01 -1.19581580e-01
2.27778539e-01 1.90641195e-01 8.07059467e-01 9.00256336e-01
1.84384614e-01 -2.23421361e-02 3.90884697e-01 9.76716101e-01
3.41387898e-01 -3.95249337e-01 -1.00406659e+00 2.08819639e-02
7.63470709e-01 9.65065241e-01 -2.80655205e-01 1.22033983e-01
2.64181495e-01 7.11784303e-01 6.38658762e-01 3.31056476e-01
-9.06298637e-01 -4.30443436e-01 1.08394015e+00 -2.53510773e-01
-1.67963788e-01 -5.06799996e-01 -1.79138526e-01 -8.52297664e-01
-1.34912372e-01 -1.12507534e+00 6.84652090e-01 -4.01619613e-01
-1.07391548e+00 3.18021268e-01 4.49142039e-01 -1.19340265e+00
-3.28343242e-01 -5.03535151e-01 -6.31166160e-01 5.56955755e-01
-1.35393953e+00 -5.25656998e-01 -2.24561438e-01 5.59526503e-01
5.90249240e-01 -1.20802626e-01 8.10612202e-01 -4.37327057e-01
-7.29076147e-01 8.18324089e-01 1.91422030e-02 -5.15120029e-01
9.55868244e-01 -1.43642771e+00 5.29435277e-01 8.10651362e-01
-4.04428601e-01 8.29235792e-01 9.73123133e-01 -1.02841449e+00
-1.46848285e+00 -1.27180135e+00 -7.60645047e-02 -8.70994210e-01
6.81778729e-01 -6.11398518e-01 -9.90376174e-01 7.29937196e-01
-7.37238675e-02 2.41900841e-03 6.21846378e-01 3.67376298e-01
-6.84395552e-01 -1.50578827e-01 -1.35267544e+00 1.10002422e+00
1.33044755e+00 -4.19852972e-01 -5.47742963e-01 1.87875107e-01
1.08248055e+00 -6.25005901e-01 -1.01815474e+00 2.97802597e-01
5.51364124e-01 -8.76700759e-01 9.09035027e-01 -9.82654095e-01
-2.85687204e-02 -2.38989204e-01 -1.21566430e-01 -1.75128520e+00
-2.76020259e-01 -1.21346080e+00 -5.45804024e-01 6.69150174e-01
2.32088417e-01 -6.16440058e-01 3.44328910e-01 9.82517123e-01
-1.92459613e-01 -7.02705979e-01 -9.77476239e-01 -1.40626872e+00
4.31463897e-01 -5.46076417e-01 6.31820977e-01 6.93132818e-01
4.44649398e-01 -2.11719111e-01 -3.43466640e-01 3.09738457e-01
9.63660061e-01 -2.69841611e-01 6.85117900e-01 -7.19041526e-01
-4.38292623e-01 -2.78995156e-01 4.74695750e-02 -2.05853775e-01
4.97544557e-01 -6.62916422e-01 5.86707234e-01 -9.81396556e-01
-4.04193485e-03 -6.65453553e-01 -7.00883090e-01 7.94658601e-01
-3.29453379e-01 -5.87080777e-01 2.98607707e-01 -1.68651327e-01
-9.18429017e-01 7.71465242e-01 9.89957392e-01 -1.33368159e-02
-6.40569568e-01 -1.01150207e-01 -7.66437292e-01 6.43341184e-01
1.31614602e+00 -6.40413046e-01 -8.65627646e-01 -2.23299235e-01
4.19491827e-01 -1.21309966e-01 1.71934545e-01 -1.34397805e+00
-8.79932418e-02 -9.47653413e-01 3.00227284e-01 4.59973849e-02
1.47343442e-01 -7.88641870e-01 -1.18774250e-01 8.13754976e-01
-7.59379923e-01 1.62505671e-01 8.83938849e-01 8.94838512e-01
5.25386751e-01 -1.63171977e-01 8.23339224e-01 -3.95149365e-02
-6.83762431e-01 1.44695252e-01 -4.23961639e-01 5.35076916e-01
1.43115866e+00 -5.98362796e-02 -3.80154759e-01 -1.73956320e-01
-6.93648279e-01 7.45243728e-01 5.71027875e-01 7.49583542e-01
7.73453176e-01 -1.08385003e+00 -1.91340595e-01 2.68472493e-01
3.20054322e-01 9.76333469e-02 2.29007795e-01 2.84954041e-01
-6.47407323e-02 -9.81044918e-02 -6.67864978e-01 -2.32732445e-01
-1.02428734e+00 7.85688698e-01 6.33569837e-01 -1.01364367e-01
-6.08267605e-01 7.59724438e-01 -9.08982456e-02 -5.04292965e-01
6.77386999e-01 -4.16227847e-01 1.41957432e-01 -3.58933568e-01
7.13650882e-01 4.38678801e-01 -3.23475033e-01 2.36020103e-01
-3.84431213e-01 3.06841075e-01 -2.64178872e-01 -3.29594731e-01
1.31145394e+00 6.57996023e-03 4.44143802e-01 7.20607698e-01
7.56105542e-01 -2.29890302e-01 -2.11712551e+00 1.43823683e-01
2.70110846e-01 -6.52138889e-01 -4.31528352e-02 -1.11825466e+00
-3.73095602e-01 5.74944973e-01 8.10870886e-01 -3.73681903e-01
1.02048767e+00 -4.73440140e-01 6.31183028e-01 8.78309131e-01
8.97972703e-01 -1.69172192e+00 6.95371926e-01 5.78832984e-01
1.02446389e+00 -1.28071952e+00 -2.45742593e-03 3.80563676e-01
-9.18723464e-01 8.57848108e-01 1.29926229e+00 -4.09314856e-02
5.49942613e-01 5.65614820e-01 -1.04154751e-01 5.18471710e-02
-1.06641746e+00 1.47267804e-01 -7.45219886e-02 1.01536107e+00
-1.88541412e-01 2.04602852e-01 4.28762376e-01 3.43511999e-01
-1.78787991e-01 -1.78282753e-01 5.80714524e-01 1.33921528e+00
-6.91491425e-01 -9.84422922e-01 -2.83862203e-01 4.04920101e-01
-1.49560869e-01 3.22544217e-01 -1.68619946e-01 6.93524063e-01
-1.26959056e-01 8.84565413e-01 -2.04372287e-01 -6.35048926e-01
4.86526132e-01 1.58680230e-01 5.71031570e-01 -8.34146440e-01
-7.76350617e-01 -3.28138173e-01 1.93255708e-01 -1.28583539e+00
2.85118103e-01 -8.59774470e-01 -1.51234293e+00 -9.98490229e-02
1.25391379e-01 -8.78554136e-02 3.74185085e-01 6.12592995e-01
2.92594492e-01 5.55145860e-01 7.16101527e-01 -6.82958186e-01
-1.26747286e+00 -4.42322075e-01 -3.12192023e-01 3.47766310e-01
8.14937353e-01 -9.06005621e-01 -4.85799581e-01 -2.77234375e-01] | [4.457327842712402, 2.096083879470825] |
1cc8c7e9-997f-4741-8942-22a47e36e15e | describing-videos-by-exploiting-temporal | 1502.08029 | null | http://arxiv.org/abs/1502.08029v5 | http://arxiv.org/pdf/1502.08029v5.pdf | Describing Videos by Exploiting Temporal Structure | Recent progress in using recurrent neural networks (RNNs) for image
description has motivated the exploration of their application for video
description. However, while images are static, working with videos requires
modeling their dynamic temporal structure and then properly integrating that
information into a natural language description. In this context, we propose an
approach that successfully takes into account both the local and global
temporal structure of videos to produce descriptions. First, our approach
incorporates a spatial temporal 3-D convolutional neural network (3-D CNN)
representation of the short temporal dynamics. The 3-D CNN representation is
trained on video action recognition tasks, so as to produce a representation
that is tuned to human motion and behavior. Second we propose a temporal
attention mechanism that allows to go beyond local temporal modeling and learns
to automatically select the most relevant temporal segments given the
text-generating RNN. Our approach exceeds the current state-of-art for both
BLEU and METEOR metrics on the Youtube2Text dataset. We also present results on
a new, larger and more challenging dataset of paired video and natural language
descriptions. | ['Hugo Larochelle', 'Kyunghyun Cho', 'Christopher Pal', 'Nicolas Ballas', 'Li Yao', 'Atousa Torabi', 'Aaron Courville'] | 2015-02-27 | describing-videos-by-exploiting-temporal-1 | http://openaccess.thecvf.com/content_iccv_2015/html/Yao_Describing_Videos_by_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Yao_Describing_Videos_by_ICCV_2015_paper.pdf | iccv-2015-12 | ['video-description'] | ['computer-vision'] | [ 1.00864798e-01 -2.91667700e-01 -6.74722731e-01 -4.07484114e-01
-6.31507874e-01 -4.39906955e-01 1.07499051e+00 -1.33979887e-01
-3.80698174e-01 3.03711772e-01 8.60115170e-01 1.18843809e-01
1.20950714e-01 -3.50823134e-01 -6.64694071e-01 -3.76667589e-01
-2.88333923e-01 2.61247814e-01 2.07240403e-01 -7.54140541e-02
2.14209870e-01 4.72055644e-01 -1.53035450e+00 7.03530192e-01
-9.56266373e-02 9.19911265e-01 3.58093679e-01 1.08296061e+00
-5.84548041e-02 1.60514843e+00 -3.40666175e-01 5.65118156e-03
-2.41482388e-02 -7.55238831e-01 -1.03238058e+00 3.83091658e-01
4.76862878e-01 -6.53785706e-01 -8.62015784e-01 4.96073693e-01
1.83852375e-01 5.58234692e-01 6.73036098e-01 -1.05639279e+00
-7.66679108e-01 5.56170523e-01 -1.71141595e-01 3.86818856e-01
6.15678906e-01 3.93671304e-01 1.09802723e+00 -5.20035863e-01
1.18841982e+00 1.13805938e+00 4.89922643e-01 1.06919456e+00
-9.95257497e-01 -2.71315962e-01 3.86950612e-01 3.64664793e-01
-1.24486184e+00 -4.60617006e-01 5.74824989e-01 -6.38263404e-01
1.44449770e+00 -3.05841547e-02 8.09659719e-01 1.67966902e+00
1.96135804e-01 9.79308724e-01 3.36879343e-01 -1.60797030e-01
-8.73720050e-02 -9.26478952e-02 -1.22247875e-01 4.78483260e-01
-5.95993042e-01 6.89750612e-02 -6.94183648e-01 2.83536434e-01
9.20115709e-01 1.23824917e-01 -1.85081869e-01 -3.73529166e-01
-1.52125323e+00 6.59411311e-01 4.26704139e-01 6.93360209e-01
-5.40853798e-01 8.83124053e-01 9.98253703e-01 8.62719491e-02
5.59959352e-01 3.44854563e-01 -2.98886180e-01 -6.34244800e-01
-1.30787265e+00 4.77726191e-01 6.45466924e-01 1.09029794e+00
3.60734731e-01 1.34338334e-01 -6.45270169e-01 5.92860043e-01
1.48782849e-01 5.18650003e-02 7.36342669e-01 -1.18687260e+00
3.45420808e-01 3.59268695e-01 1.48497447e-01 -8.88292730e-01
-2.45795816e-01 -1.02670928e-02 -6.84920430e-01 -2.59546608e-01
2.35688686e-01 4.26409990e-01 -1.03367102e+00 1.76875067e+00
-2.96191454e-01 2.73529083e-01 1.36972591e-01 1.03255224e+00
9.24412489e-01 1.08051229e+00 2.99255192e-01 -2.05178514e-01
1.13307345e+00 -1.26908481e+00 -9.17213023e-01 7.58392215e-02
7.22990811e-01 -3.15845162e-01 7.55731940e-01 -1.76370069e-02
-1.40131664e+00 -6.98775768e-01 -8.62774909e-01 -2.89502174e-01
-3.32463443e-01 2.32453361e-01 2.57309884e-01 -2.31282368e-01
-1.43432164e+00 8.10361147e-01 -9.91533041e-01 -7.13808298e-01
2.55751193e-01 2.55019486e-01 -4.45805788e-01 5.08164475e-03
-1.15322077e+00 8.42665076e-01 5.41334689e-01 -8.40986893e-02
-1.36773658e+00 -4.28072035e-01 -1.00502622e+00 3.78238820e-02
2.13375553e-01 -6.98790669e-01 1.60834956e+00 -1.36501801e+00
-1.46485615e+00 8.00657749e-01 -3.78450215e-01 -8.75608802e-01
4.99110371e-01 -1.77788273e-01 -9.46120918e-02 4.70119417e-01
-1.71933125e-03 1.25460386e+00 6.93985105e-01 -1.00338387e+00
-3.47012162e-01 1.57039568e-01 3.53979290e-01 1.61910594e-01
-2.01100856e-01 1.44401029e-01 -7.94798732e-01 -8.03199232e-01
-3.11212808e-01 -9.92310822e-01 -2.54667401e-01 1.47635579e-01
-1.62289619e-01 -3.10561508e-01 8.58054817e-01 -6.45968616e-01
1.45208633e+00 -2.12828565e+00 5.02793789e-01 -3.42073798e-01
6.36214837e-02 1.78398490e-01 -3.01920921e-01 6.42702997e-01
-1.89998433e-01 3.83848816e-01 6.63104877e-02 -7.71721780e-01
-4.04885709e-02 2.52030164e-01 -3.89464408e-01 2.99325168e-01
3.31946403e-01 1.05737150e+00 -9.82937276e-01 -4.20528203e-01
3.39798540e-01 8.32732677e-01 -5.44097185e-01 3.34007263e-01
-6.10623956e-01 4.55776662e-01 -4.62316126e-01 2.95505166e-01
-3.57528068e-02 -3.76913697e-01 2.73245256e-02 -3.17359149e-01
-2.47586474e-01 4.19260174e-01 -5.08716643e-01 1.99026620e+00
-5.95431268e-01 1.25906372e+00 -5.30452073e-01 -9.56238687e-01
7.01079130e-01 8.12988997e-01 8.19502711e-01 -7.99384296e-01
1.07265480e-01 2.50045136e-02 -4.48880553e-01 -8.56604815e-01
7.19626725e-01 6.85408711e-02 -9.22316015e-02 6.17118537e-01
4.34814274e-01 1.09798715e-01 3.78058314e-01 3.83886307e-01
9.93019044e-01 8.35484803e-01 3.39715213e-01 1.54438883e-01
4.90152717e-01 1.80831552e-01 2.45946780e-01 5.95859230e-01
-3.86659473e-01 1.04212749e+00 6.05979621e-01 -1.01356459e+00
-1.53595257e+00 -5.58085203e-01 4.22845066e-01 9.69591022e-01
-5.66710196e-02 -6.45739496e-01 -5.88970721e-01 -5.96517801e-01
-3.64257306e-01 5.49380124e-01 -1.00896311e+00 -1.66991338e-01
-6.89006090e-01 4.51225489e-02 3.88130814e-01 6.69011950e-01
4.25284058e-01 -1.46586025e+00 -9.12603736e-01 4.00501609e-01
-4.02529836e-01 -1.48914135e+00 -8.11491668e-01 -1.25383511e-01
-7.81455159e-01 -7.96519816e-01 -9.01005208e-01 -6.04622602e-01
2.25095555e-01 1.05261989e-01 1.24688244e+00 -5.03588244e-02
-2.09976479e-01 7.55278826e-01 -7.28052914e-01 9.97030288e-02
-5.81644177e-01 -1.87905822e-02 1.13738352e-03 4.39749695e-02
3.54927570e-01 -3.42978269e-01 -3.77022296e-01 1.63893729e-01
-1.11415005e+00 3.59777629e-01 2.95938522e-01 6.70813024e-01
3.78385931e-01 -4.66616601e-01 1.73189849e-01 -3.41354877e-01
2.73764461e-01 -5.44493556e-01 -1.73790917e-01 2.29104221e-01
3.85812744e-02 2.04778314e-01 4.12202090e-01 -7.72083700e-01
-8.93831730e-01 4.13697213e-01 1.41374379e-01 -1.01163018e+00
-1.48020878e-01 4.79788333e-01 5.11136651e-01 4.32872653e-01
4.38922554e-01 5.18457949e-01 -4.68753912e-02 -1.17171235e-01
3.24956119e-01 4.00533617e-01 5.26537955e-01 -1.02884330e-01
4.79960263e-01 5.46064138e-01 -1.90756828e-01 -6.95627272e-01
-9.65931296e-01 -6.51657164e-01 -1.06172156e+00 -6.04510844e-01
1.30750334e+00 -1.14361131e+00 -6.89347744e-01 1.91777796e-01
-1.69028163e+00 -5.48495471e-01 -2.63941228e-01 5.66478431e-01
-1.19001281e+00 6.93772957e-02 -6.72704935e-01 -7.08373189e-01
-1.00686692e-01 -1.17021728e+00 1.19134152e+00 -1.02304690e-01
-5.31336844e-01 -1.22807789e+00 1.94650829e-01 1.26570240e-01
4.96315926e-01 4.93747145e-01 3.87558132e-01 -5.55021882e-01
-8.09932709e-01 -1.77822322e-01 -2.48159975e-01 2.12636396e-01
-5.13236001e-02 2.76484042e-01 -8.19130242e-01 -8.09508711e-02
-2.21999824e-01 -4.74277854e-01 1.00680757e+00 6.61167026e-01
1.16601431e+00 -5.34573317e-01 -1.56045169e-01 5.17822683e-01
1.42182708e+00 2.47589976e-01 6.63244903e-01 4.36211318e-01
6.92758501e-01 6.81605935e-01 5.71923912e-01 6.62915945e-01
4.26097512e-01 1.03931499e+00 5.64706922e-01 7.40129920e-03
-3.04834604e-01 -3.56014013e-01 6.53721333e-01 6.67341650e-01
-2.97119349e-01 -5.60670614e-01 -8.74546051e-01 8.62021387e-01
-2.16481519e+00 -1.51064682e+00 2.97370315e-01 1.82351935e+00
5.75914264e-01 -6.45613857e-03 3.21431220e-01 -3.06295335e-01
5.32812178e-01 5.48483372e-01 -4.99727011e-01 -6.29661024e-01
-6.03417167e-03 -3.14723462e-01 3.22214454e-01 2.56104261e-01
-1.16605830e+00 9.81031179e-01 6.33390999e+00 4.72685993e-01
-1.20962632e+00 4.09922637e-02 5.62097728e-01 -3.37661922e-01
-1.41268253e-01 -2.32714847e-01 -5.18925130e-01 1.50775835e-01
1.40742099e+00 -2.40592986e-01 3.40045959e-01 6.06932521e-01
8.12487364e-01 1.98773116e-01 -1.42879009e+00 1.01497638e+00
3.23161811e-01 -1.77787042e+00 5.11556983e-01 -1.02625310e-01
7.66798317e-01 1.01462498e-01 -2.44127400e-02 2.82021284e-01
-1.70026738e-02 -1.21663129e+00 1.07742035e+00 8.68899226e-01
8.47556710e-01 -5.32433927e-01 6.45486593e-01 2.18829691e-01
-1.51009202e+00 -7.85440356e-02 -2.89115068e-02 -9.34857992e-04
4.15460557e-01 -2.09498286e-01 -9.06276822e-01 3.40129048e-01
7.31068850e-01 1.66516566e+00 -3.09683651e-01 8.00112188e-01
1.29645139e-01 3.22025865e-01 2.11979896e-01 -1.59733668e-01
9.40270722e-01 1.99576929e-01 6.11081541e-01 1.59152830e+00
3.23419958e-01 4.72885780e-02 1.49975911e-01 8.86702359e-01
-9.13757011e-02 -6.27798811e-02 -9.98531282e-01 -6.84360445e-01
2.16298960e-02 8.07661951e-01 -6.02854848e-01 -5.13896108e-01
-3.96813720e-01 1.09986365e+00 2.40030676e-01 5.05487561e-01
-1.00427020e+00 2.00112894e-01 6.05165422e-01 1.12696737e-01
4.01697516e-01 -5.24201334e-01 4.42159265e-01 -1.27240419e+00
-3.49591300e-02 -5.96411645e-01 2.73709625e-01 -1.21499944e+00
-8.83064151e-01 9.54785228e-01 2.39710674e-01 -1.63163221e+00
-8.82837236e-01 -3.93142492e-01 -4.49082345e-01 5.31091154e-01
-1.34567237e+00 -1.28687978e+00 -2.97740519e-01 4.62938130e-01
1.34808803e+00 -1.25521362e-01 6.98051095e-01 1.61477819e-01
-2.92452037e-01 2.23438628e-02 -1.86782286e-01 1.58792242e-01
6.64527595e-01 -9.12294924e-01 5.90822220e-01 5.82852483e-01
2.08781853e-01 4.72137123e-01 7.60978997e-01 -4.22452241e-01
-1.20365727e+00 -1.28430462e+00 1.19728541e+00 -5.22063494e-01
7.39238441e-01 -2.58746862e-01 -7.41262913e-01 1.00088143e+00
4.06517208e-01 1.22580804e-01 3.02515417e-01 -5.20494759e-01
-4.34764057e-01 2.55700171e-01 -6.63584948e-01 6.57716513e-01
1.04879558e+00 -9.19170082e-01 -3.53426814e-01 5.37815928e-01
9.32836890e-01 -3.76228154e-01 -8.02934527e-01 1.41367301e-01
7.02544928e-01 -8.95540655e-01 8.75156760e-01 -8.38032007e-01
9.89998162e-01 -1.12637959e-01 -1.85797483e-01 -8.69694591e-01
-1.62686139e-01 -7.38838077e-01 -1.51947305e-01 9.80966806e-01
3.88878495e-01 3.80218834e-01 6.60164297e-01 4.82737005e-01
-1.67345330e-01 -6.70204639e-01 -7.43664920e-01 -6.65366232e-01
-2.42181748e-01 -5.57527304e-01 1.36781000e-02 8.45801115e-01
-5.32180779e-02 1.35919049e-01 -9.70964789e-01 -3.24769825e-01
6.08632602e-02 -2.44793504e-01 5.65047026e-01 -7.58926272e-01
5.53036779e-02 -4.77356404e-01 -7.46175110e-01 -1.31103373e+00
5.47538221e-01 -7.52951741e-01 2.29526594e-01 -1.66392386e+00
4.32516724e-01 3.30446482e-01 -4.96524423e-02 4.33299661e-01
4.24462199e-01 3.44321579e-01 4.75946873e-01 3.53488684e-01
-1.08678269e+00 6.38051212e-01 1.23970389e+00 -2.50940174e-01
-2.49628887e-01 -2.65278876e-01 8.40839520e-02 3.37089032e-01
4.22004640e-01 -3.63273352e-01 -3.70163977e-01 -5.72492659e-01
7.76134953e-02 4.09507394e-01 5.01431346e-01 -1.05841398e+00
3.24024022e-01 -2.18503699e-01 2.73335636e-01 -7.06128061e-01
6.70631111e-01 -7.11054206e-01 2.20181763e-01 4.45988983e-01
-1.05584729e+00 3.66878241e-01 3.39561790e-01 5.92531502e-01
-5.79696774e-01 -5.20892218e-02 6.32656932e-01 -5.16386449e-01
-1.09763110e+00 6.38394058e-01 -6.91141129e-01 -2.48577595e-01
1.05861390e+00 -3.97909999e-01 3.29395458e-02 -9.88935351e-01
-1.11172187e+00 1.56610623e-01 4.93722469e-01 8.66604984e-01
8.03353608e-01 -1.51345110e+00 -7.16802299e-01 -1.63021162e-01
2.62459755e-01 -5.12441874e-01 3.44201118e-01 5.85847497e-01
-5.62515020e-01 8.12532127e-01 -4.05984879e-01 -8.60915840e-01
-1.20570087e+00 6.01868093e-01 5.63226283e-01 -2.33563885e-01
-7.79704750e-01 5.85607469e-01 1.62038803e-01 5.38002402e-02
4.08999383e-01 -4.64956909e-01 -6.13658667e-01 1.52382970e-01
6.78302109e-01 -2.86116470e-02 -5.07809103e-01 -1.10755384e+00
-2.11380407e-01 7.68791139e-01 -6.63482351e-03 -3.22707862e-01
1.49987400e+00 -3.78081053e-01 1.61283001e-01 8.56592178e-01
1.66489804e+00 -7.39414215e-01 -1.52935636e+00 -2.43179783e-01
1.41762748e-01 -1.91884249e-01 -9.01628062e-02 -3.82412016e-01
-9.69248295e-01 9.25688088e-01 2.74739653e-01 2.00185493e-01
1.00834870e+00 2.02767011e-02 7.20945895e-01 3.87678087e-01
8.84250328e-02 -8.64087582e-01 6.45009518e-01 8.76980901e-01
1.07674551e+00 -1.25107968e+00 -1.38763845e-01 2.26169452e-01
-1.02561712e+00 1.68822610e+00 4.40923482e-01 -2.32412234e-01
3.31972629e-01 -1.78171545e-01 -3.64913903e-02 -2.45774776e-01
-1.43555617e+00 -1.89775467e-01 5.42614281e-01 2.92715669e-01
6.87460184e-01 -3.57428491e-01 8.35916027e-02 1.00907043e-01
3.44007909e-01 2.73025572e-01 8.21648061e-01 7.63746679e-01
-1.03290893e-01 -8.09468448e-01 1.78692743e-01 6.00698516e-02
-5.18448770e-01 1.05781011e-01 -3.83881032e-01 9.56420064e-01
-1.12811498e-01 5.49096942e-01 3.61069083e-01 -4.09849733e-01
1.61715761e-01 -3.67250852e-02 2.85986960e-01 -6.81300402e-01
-6.68968558e-01 1.08356066e-01 9.17438492e-02 -1.05984747e+00
-1.12375748e+00 -7.84905612e-01 -1.18868995e+00 -4.82161120e-02
2.61331111e-01 -3.91804986e-03 8.20942938e-01 1.01748228e+00
1.37467280e-01 7.34438717e-01 5.55378079e-01 -1.35283351e+00
-1.87190086e-01 -9.90537941e-01 -2.67472118e-01 5.92131674e-01
9.13374841e-01 -4.38789636e-01 -3.77197504e-01 6.38410926e-01] | [10.441507339477539, 0.6188522577285767] |
cbe4d7a1-e85e-40cb-8d3f-7aaddf169758 | pathgan-visual-scanpath-prediction-with | 1809.00567 | null | http://arxiv.org/abs/1809.00567v1 | http://arxiv.org/pdf/1809.00567v1.pdf | PathGAN: Visual Scanpath Prediction with Generative Adversarial Networks | We introduce PathGAN, a deep neural network for visual scanpath prediction
trained on adversarial examples. A visual scanpath is defined as the sequence
of fixation points over an image defined by a human observer with its gaze.
PathGAN is composed of two parts, the generator and the discriminator. Both
parts extract features from images using off-the-shelf networks, and train
recurrent layers to generate or discriminate scanpaths accordingly. In scanpath
prediction, the stochastic nature of the data makes it very difficult to
generate realistic predictions using supervised learning strategies, but we
adopt adversarial training as a suitable alternative. Our experiments prove how
PathGAN improves the state of the art of visual scanpath prediction on the iSUN
and Salient360! datasets. Source code and models are available at
https://imatge-upc.github.io/pathgan/ | ["Noel E. O'Connor", 'Kevin McGuinness', 'Xavier Giro-i-Nieto', 'Marc Assens'] | 2018-09-03 | null | null | null | null | ['scanpath-prediction'] | ['computer-vision'] | [ 4.02583688e-01 3.49140316e-01 -4.10929829e-01 -3.32159698e-01
-3.72553527e-01 -7.45349348e-01 6.57233059e-01 -6.68156803e-01
-1.08296119e-01 4.10356492e-01 7.08530471e-03 -6.63182616e-01
5.06441593e-01 -8.04257095e-01 -1.14395237e+00 -5.24133325e-01
2.05009639e-01 3.29274386e-01 1.83062911e-01 -1.34992540e-01
2.50593752e-01 2.36460701e-01 -1.29938805e+00 3.88301134e-01
5.32528341e-01 8.31637681e-01 9.83384177e-02 8.52983892e-01
1.73994079e-01 1.01767492e+00 -4.76146132e-01 -6.69960558e-01
6.16041481e-01 -9.57940578e-01 -7.68484652e-01 -2.58205771e-01
5.27939379e-01 -3.27756017e-01 -6.71751022e-01 1.22825575e+00
3.94541919e-01 -5.39394133e-02 5.90271533e-01 -1.40240037e+00
-1.08941412e+00 5.62489510e-01 -7.29252696e-01 4.04226333e-01
2.74742246e-01 8.97058904e-01 8.79223108e-01 -6.94245160e-01
9.25693393e-01 1.12266219e+00 4.06046718e-01 1.00181341e+00
-1.26223922e+00 -1.02658629e+00 -9.20462906e-02 3.15343708e-01
-1.09274173e+00 -4.35525447e-01 6.59885824e-01 -4.20403808e-01
5.63160121e-01 2.32473925e-01 8.57478440e-01 1.78066981e+00
4.87652540e-01 9.41453576e-01 1.01898754e+00 -2.26695970e-01
-5.32376133e-02 -3.61484811e-02 -3.52882534e-01 7.58036911e-01
-3.74531075e-02 6.77547097e-01 -2.81961679e-01 1.61780819e-01
1.01554084e+00 -1.44675970e-01 -4.44001287e-01 -2.86082596e-01
-1.11035681e+00 9.25823689e-01 9.07372653e-01 -2.89962739e-01
-2.66698778e-01 4.35674161e-01 1.09934896e-01 5.83502829e-01
2.91892201e-01 6.93399549e-01 1.65672805e-02 3.38511318e-02
-6.89789951e-01 3.54548663e-01 5.82093477e-01 9.80834186e-01
6.32877648e-01 3.15711737e-01 -3.17562968e-01 3.23266208e-01
7.04376996e-02 8.66716564e-01 5.41548431e-01 -1.17982244e+00
3.46963167e-01 3.14563572e-01 8.98495987e-02 -9.01354492e-01
-9.95717645e-02 -3.23908657e-01 -6.69445276e-01 6.37590110e-01
5.12812734e-01 -3.30194354e-01 -1.32844961e+00 1.72436666e+00
6.94617108e-02 7.95544311e-02 -2.44868338e-01 1.03443325e+00
8.79964590e-01 6.81491017e-01 -3.29914130e-03 2.19534695e-01
7.77634501e-01 -1.35281515e+00 -3.33260030e-01 -3.53544205e-01
2.74502277e-01 -7.30932057e-01 1.31204057e+00 1.41646013e-01
-1.26733959e+00 -4.86034036e-01 -8.30848932e-01 1.94985922e-02
-2.44076952e-01 -2.10043788e-02 3.85839641e-01 2.45443150e-01
-1.22370994e+00 5.14720321e-01 -7.64819562e-01 -1.96129773e-02
7.06489384e-01 9.12720785e-02 -3.81496817e-01 -6.31595179e-02
-1.16511464e+00 7.20367968e-01 1.96440771e-01 -1.64491672e-03
-1.41608334e+00 -6.07831955e-01 -9.80248213e-01 -1.06037371e-02
1.98318392e-01 -7.59699762e-01 1.47069895e+00 -1.78558385e+00
-1.31432354e+00 1.09638309e+00 -9.72365886e-02 -8.07866573e-01
9.34253335e-01 -4.68229726e-02 -3.10453057e-01 2.25924507e-01
1.11916512e-01 1.13465834e+00 1.56626141e+00 -1.05602610e+00
-2.95549273e-01 1.40859649e-01 2.24898085e-01 1.19696967e-01
4.48804528e-01 -2.35125795e-03 -5.18856585e-01 -6.50673687e-01
-5.49379349e-01 -1.17125356e+00 -2.13317737e-01 2.13391513e-01
-9.04187322e-01 1.90255344e-02 6.71568215e-01 -5.55571079e-01
8.53582323e-01 -2.12440419e+00 5.86465076e-02 1.79821655e-01
3.01866829e-01 5.77201247e-01 -5.94991207e-01 2.45336831e-01
-5.92266381e-01 3.64516556e-01 -1.51586831e-02 -1.21324688e-01
-2.41066530e-01 -1.41735315e-01 -5.93521357e-01 4.21386719e-01
2.70706952e-01 1.40500498e+00 -8.83215070e-01 -8.54401737e-02
7.39634708e-02 3.76885504e-01 -5.15667319e-01 5.49873531e-01
-6.34361506e-01 7.09515512e-01 -1.51990369e-01 6.66682303e-01
6.22866094e-01 -5.62963486e-01 -2.57237732e-01 3.39576542e-01
1.01584509e-01 1.85175270e-01 -4.41867858e-01 1.31956387e+00
-1.64887100e-01 1.04700518e+00 -4.89384174e-01 -5.87707460e-01
6.17471039e-01 -8.51288810e-03 -1.08590372e-01 -9.34657335e-01
1.92862526e-01 -1.52860746e-01 2.49712437e-01 -4.62464154e-01
3.46887559e-01 1.33132726e-01 -7.49146119e-02 7.34645188e-01
9.52026248e-02 5.32409996e-02 6.80318773e-02 1.60647750e-01
1.15641642e+00 2.15319648e-01 2.53309369e-01 1.19529039e-01
1.81477115e-01 9.28008482e-02 3.66800159e-01 8.03966165e-01
-2.92788804e-01 9.40944374e-01 8.87436628e-01 -7.16816783e-01
-1.26409352e+00 -1.19861352e+00 3.14031005e-01 9.06745851e-01
2.34467328e-01 -1.99108303e-01 -9.46128905e-01 -9.16573763e-01
-1.65299550e-01 7.95988083e-01 -1.27615821e+00 -4.09393221e-01
-5.06179571e-01 -9.47131440e-02 5.81566930e-01 3.91138226e-01
5.80478728e-01 -1.64403009e+00 -7.31765389e-01 -3.96410137e-01
-4.77680340e-02 -7.89766610e-01 -7.05340266e-01 -1.54190913e-01
-3.17327410e-01 -1.35637200e+00 -7.47798622e-01 -5.36936224e-01
8.03229332e-01 3.25790584e-01 1.42312360e+00 2.48846591e-01
-2.22805887e-01 1.19930200e-01 -3.66586655e-01 -6.54987633e-01
-6.46669209e-01 2.17258751e-01 -3.47382128e-01 -6.07364848e-02
2.23172858e-01 -5.04673481e-01 -1.06283295e+00 3.65508348e-01
-6.96016610e-01 3.50076914e-01 6.84500694e-01 9.23174202e-01
8.50913584e-01 -4.09614295e-01 2.23699049e-03 -1.23939526e+00
2.94840276e-01 -6.90281093e-01 -8.39764535e-01 -2.50150487e-02
-4.58215594e-01 -4.15165089e-02 8.20399880e-01 -4.28467363e-01
-7.61337638e-01 -6.09823242e-02 -1.14323393e-01 -1.00386548e+00
-2.55016685e-01 1.47759780e-01 7.55138919e-02 -5.11309169e-02
9.57628131e-01 4.04421389e-01 5.11177517e-02 -6.14476725e-02
3.88309687e-01 1.48492515e-01 6.07931912e-01 1.19467556e-01
1.17221689e+00 4.67101693e-01 -9.29255337e-02 -4.52668518e-01
-9.65978920e-01 8.21175650e-02 -3.76188129e-01 -2.14265332e-01
9.26714420e-01 -7.17325687e-01 -6.00817561e-01 4.41207230e-01
-1.04180908e+00 -9.89994526e-01 -4.40922946e-01 -4.54749316e-02
-7.40036726e-01 -2.73571402e-01 -1.87118009e-01 -2.56540418e-01
-5.44389069e-01 -9.98707652e-01 7.24306703e-01 5.81851006e-01
-2.03247249e-01 -9.38650250e-01 1.18988894e-01 2.52394259e-01
4.26046520e-01 6.12198293e-01 6.68048918e-01 -5.79984665e-01
-9.26856279e-01 -2.93909490e-01 -2.05061838e-01 3.32270920e-01
-1.83450654e-01 1.78650409e-01 -9.83662546e-01 -3.23706418e-01
-2.86607891e-01 -5.38669467e-01 9.68268216e-01 4.08889502e-01
1.58563650e+00 -4.68760192e-01 -3.65920812e-01 1.18526173e+00
1.42995739e+00 1.71896458e-01 9.00093615e-01 2.99646616e-01
9.54486728e-01 3.43353540e-01 5.89216709e-01 3.89277562e-02
1.12296306e-01 4.05612171e-01 1.10247147e+00 -2.51768231e-01
-3.58606368e-01 -7.48658419e-01 3.56800348e-01 -7.04060346e-02
-3.21326293e-02 -6.93291724e-01 -1.06335151e+00 5.72590411e-01
-1.60988629e+00 -1.31596804e+00 -3.58877070e-02 2.01594329e+00
5.87878406e-01 1.15562826e-01 2.59453803e-01 -2.01411277e-01
7.54279494e-01 5.08462131e-01 -9.64414299e-01 -5.48912644e-01
4.92565706e-03 2.84227639e-01 6.12423420e-01 3.92743498e-01
-1.03999996e+00 1.12037504e+00 6.02820444e+00 4.94934350e-01
-1.51141429e+00 8.23379382e-02 9.86312330e-01 -4.08353567e-01
-3.22350591e-01 -1.91591203e-01 -4.16850835e-01 7.54470825e-01
1.09741414e+00 -1.46793425e-01 5.81254959e-01 7.73690641e-01
1.85386568e-01 1.74990118e-01 -1.00320101e+00 7.53070474e-01
5.12434989e-02 -1.43407738e+00 2.21260250e-01 3.44576016e-02
8.52100313e-01 5.31273246e-01 7.76130915e-01 1.96091980e-01
6.36784494e-01 -1.56340182e+00 7.88154662e-01 5.95212758e-01
1.22587514e+00 -7.44461596e-01 3.87976021e-01 2.23105997e-01
-6.41213179e-01 2.84267813e-02 -4.62426960e-01 1.01381466e-01
-6.80168718e-02 6.00986332e-02 -8.04056048e-01 5.95600568e-02
7.13898957e-01 8.08653593e-01 -7.50234187e-01 9.95142579e-01
-8.70183527e-01 7.99347103e-01 1.99002787e-01 2.45789990e-01
3.69068861e-01 9.41853449e-02 6.63689494e-01 8.42802584e-01
1.77558437e-01 -2.24495977e-01 -2.42173567e-01 1.33460724e+00
-4.74641353e-01 -2.66431868e-01 -9.87654269e-01 -7.89317787e-02
3.36119562e-01 1.06848860e+00 -3.59853804e-01 -3.71816382e-02
-2.49034107e-01 1.22197032e+00 3.81916821e-01 5.27920842e-01
-1.16062582e+00 -4.26175237e-01 7.63361871e-01 4.51111108e-01
5.06741643e-01 3.07019383e-01 -2.12590635e-01 -1.18075097e+00
-1.47050083e-01 -1.08793592e+00 2.72425115e-01 -1.10412490e+00
-1.00084639e+00 9.64003682e-01 -2.68748790e-01 -1.59453845e+00
-6.84057176e-01 -5.95006943e-01 -1.17925477e+00 9.44877803e-01
-1.46789038e+00 -1.36509776e+00 -5.22252738e-01 8.03743660e-01
6.03125930e-01 -2.42184162e-01 7.14261532e-01 -3.91656727e-01
-4.88157451e-01 8.70829642e-01 -2.04679161e-01 4.18368429e-01
5.60898423e-01 -1.27966082e+00 1.16519821e+00 1.10164666e+00
1.73588812e-01 3.19183230e-01 7.78670549e-01 -4.95277494e-01
-9.51827168e-01 -1.42158151e+00 6.32065415e-01 -7.38334596e-01
6.09454274e-01 -3.25099230e-01 -8.87867630e-01 1.16215050e+00
5.52592039e-01 4.52843815e-01 5.17438352e-01 -3.80686760e-01
-5.91928840e-01 2.49331877e-01 -9.78143692e-01 1.00898027e+00
1.26252365e+00 -4.95003700e-01 -2.87187696e-01 2.00096190e-01
8.10757756e-01 -7.82364249e-01 -3.10148299e-01 4.73623015e-02
4.99240875e-01 -1.29733860e+00 8.03591073e-01 -7.17261255e-01
1.05697823e+00 -2.81863380e-02 3.51451755e-01 -1.62901390e+00
-3.32639873e-01 -9.26989436e-01 -1.06324032e-01 7.16196418e-01
6.08472347e-01 -6.93611681e-01 9.27130044e-01 2.20865697e-01
9.92008820e-02 -8.37234318e-01 -5.17049491e-01 -5.30049443e-01
2.77854025e-01 -2.53133059e-01 6.78108454e-01 6.60565317e-01
-4.47274536e-01 2.41056725e-01 -5.46745479e-01 -8.67391452e-02
5.41023970e-01 3.12486500e-01 1.02415073e+00 -7.99212873e-01
-4.61435735e-01 -3.68080705e-01 -3.50700021e-01 -8.76374662e-01
5.67229152e-01 -8.31034184e-01 -7.52871484e-02 -1.02988601e+00
5.90696000e-02 -1.95948094e-01 -2.55002946e-01 6.66306257e-01
-2.18842685e-01 6.99553311e-01 4.16468084e-01 4.33984995e-01
-5.12300849e-01 4.02196735e-01 1.65136647e+00 -2.87115514e-01
7.15174573e-03 3.70394677e-01 -8.17682445e-01 7.18385994e-01
1.20459509e+00 -4.90539908e-01 -7.79785693e-01 -6.95008457e-01
3.06415975e-01 1.29361778e-01 6.32692218e-01 -8.42160583e-01
-1.46186635e-01 -1.78912133e-01 6.22001588e-01 -2.09273994e-01
1.69514477e-01 -5.30004084e-01 1.38178021e-01 5.48820257e-01
-5.04113019e-01 2.99654901e-01 2.19757080e-01 5.07977784e-01
-1.44821510e-01 -2.19619013e-02 1.03217041e+00 -3.45753670e-01
-7.42314517e-01 6.40591204e-01 -2.20681220e-01 1.95493683e-01
1.06732047e+00 -7.49863610e-02 -8.00357759e-01 -7.87106574e-01
-5.56532741e-01 1.87423855e-01 9.34799731e-01 6.20997190e-01
8.38865280e-01 -1.23963940e+00 -6.53544128e-01 4.49815989e-01
1.19792618e-01 9.68980417e-03 1.90943509e-01 6.26604915e-01
-7.28030086e-01 3.10708046e-01 -6.09063625e-01 -5.15891075e-01
-1.04203296e+00 7.70528555e-01 5.16651213e-01 -2.11547673e-01
-4.38275099e-01 1.13370311e+00 4.54109550e-01 -1.51813701e-01
-2.17016280e-01 -3.93755920e-03 -1.63731202e-01 -4.11593050e-01
7.03775704e-01 8.16626940e-03 -4.58795100e-01 -7.38120675e-01
3.71264876e-03 2.92840302e-01 -1.53269514e-01 -8.15337226e-02
1.11678541e+00 2.15959363e-02 2.13843316e-01 2.77149200e-01
1.13884187e+00 -1.47058889e-01 -1.64768469e+00 -5.91707863e-02
-5.83336711e-01 -5.99284589e-01 -2.67321169e-01 -9.37050700e-01
-1.34423220e+00 1.01217699e+00 4.84047115e-01 1.69203848e-01
1.18226767e+00 -4.43383753e-02 8.41341674e-01 9.06459913e-02
1.37293354e-01 -3.34913462e-01 2.29254693e-01 4.21409965e-01
1.09289563e+00 -1.45356011e+00 -5.89261532e-01 -1.31365269e-01
-9.33424354e-01 8.11169386e-01 9.02732015e-01 -4.59041208e-01
4.27362591e-01 -1.19852707e-01 4.61571991e-01 -2.45564535e-01
-9.73927081e-01 -9.79823768e-02 3.05906087e-01 8.71839881e-01
6.89270794e-02 4.51835878e-02 2.29640290e-01 1.33608729e-01
-5.17781973e-01 -1.36784717e-01 6.69990957e-01 4.97473717e-01
-3.24132852e-02 -9.31056857e-01 -5.02162129e-02 4.84228998e-01
-6.86343253e-01 -3.15647930e-01 -4.80352759e-01 9.39146161e-01
7.12068304e-02 5.44112802e-01 3.48418504e-01 -6.86464727e-01
7.63200596e-02 -2.74843097e-01 3.11794162e-01 -6.40645504e-01
-4.53275800e-01 -5.41668653e-01 -2.00093731e-01 -1.02219558e+00
-1.19636811e-01 -4.77835059e-01 -8.65920782e-01 -5.33636808e-01
1.45762011e-01 -1.78025842e-01 3.03842723e-01 5.22781193e-01
4.18199718e-01 4.24293816e-01 8.60697567e-01 -8.66523802e-01
-3.90261799e-01 -9.05015469e-01 -1.83974743e-01 4.96240199e-01
5.12010574e-01 -3.62997890e-01 -4.38058227e-01 2.02818468e-01] | [11.588163375854492, -0.19343160092830658] |
7f0e3498-39f4-47ca-ba08-924482257d77 | share-price-prediction-of-aerospace-relevant | 2008.11788 | null | https://arxiv.org/abs/2008.11788v1 | https://arxiv.org/pdf/2008.11788v1.pdf | Share Price Prediction of Aerospace Relevant Companies with Recurrent Neural Networks based on PCA | The capital market plays a vital role in marketing operations for aerospace industry. However, due to the uncertainty and complexity of the stock market and many cyclical factors, the stock prices of listed aerospace companies fluctuate significantly. This makes the share price prediction challengeable. To improve the prediction of share price for aerospace industry sector and well understand the impact of various indicators on stock prices, we provided a hybrid prediction model by the combination of Principal Component Analysis (PCA) and Recurrent Neural Networks. We investigated two types of aerospace industries (manufacturer and operator). The experimental results show that PCA could improve both accuracy and efficiency of prediction. Various factors could influence the performance of prediction models, such as finance data, extracted features, optimisation algorithms, and parameters of the prediction model. The selection of features may depend on the stability of historical data: technical features could be the first option when the share price is stable, whereas fundamental features could be better when the share price has high fluctuation. The delays of RNN also depend on the stability of historical data for different types of companies. It would be more accurate through using short-term historical data for aerospace manufacturers, whereas using long-term historical data for aerospace operating airlines. The developed model could be an intelligent agent in an automatic stock prediction system, with which, the financial industry could make a prompt decision for their economic strategies and business activities in terms of predicted future share price, thus improving the return on investment. Currently, COVID-19 severely influences aerospace industries. The developed approach can be used to predict the share price of aerospace industries at post COVID-19 time. | ['Linyu Zheng', 'Hongmei He'] | 2020-08-26 | null | null | null | null | ['stock-prediction'] | ['time-series'] | [-4.45394903e-01 -3.02654058e-01 -1.47751585e-01 1.02046505e-01
1.34157300e-01 -6.11884713e-01 4.63196546e-01 -1.90583989e-01
-3.89779508e-01 7.65506208e-01 7.82709345e-02 -5.92536211e-01
-5.63268006e-01 -9.72015202e-01 -1.95775628e-01 -8.18499148e-01
-6.40924722e-02 4.60778236e-01 1.20056607e-02 -5.18558741e-01
4.95229274e-01 9.65366304e-01 -1.39499331e+00 -1.07092187e-01
1.99314848e-01 1.14240086e+00 4.23868775e-01 2.55013973e-01
-2.86545187e-01 7.75977194e-01 -6.06919289e-01 4.18620296e-02
6.88775718e-01 -2.76043504e-01 -2.59474292e-02 -1.29456982e-01
-1.03829122e+00 -2.37721875e-02 1.53809071e-01 9.26588833e-01
2.46076226e-01 -1.70172438e-01 4.39846367e-01 -1.09909356e+00
8.57098773e-02 8.22989464e-01 -2.28951633e-01 4.04890418e-01
-3.13575193e-02 4.81916368e-02 8.53596330e-01 -7.87703097e-01
3.97095472e-01 7.23490417e-01 3.06906015e-01 -2.07507044e-01
-7.58679688e-01 -7.13611960e-01 1.49937332e-01 2.73665577e-01
-8.97369921e-01 1.22799911e-01 1.02190721e+00 -6.17907047e-01
6.34299695e-01 3.67904454e-01 1.08527231e+00 6.52546942e-01
6.77700162e-01 2.29466423e-01 1.17738056e+00 -2.40417063e-01
4.95319143e-02 3.91426474e-01 -1.71239480e-01 -1.30620942e-01
4.82874095e-01 3.72758508e-01 -1.83878183e-01 6.79268083e-03
6.76977932e-01 2.61249363e-01 -3.07854801e-01 2.89954115e-02
-9.78179812e-01 8.37191105e-01 7.37020597e-02 7.20974982e-01
-9.51659799e-01 -3.25494558e-01 1.72496498e-01 6.33641660e-01
1.34610400e-01 6.36841476e-01 -8.84162068e-01 -5.00323594e-01
-8.08643878e-01 1.80270761e-01 9.73669708e-01 3.41143101e-01
2.33583808e-01 5.29347181e-01 3.44735771e-01 1.73600882e-01
3.41248512e-01 6.46710753e-01 1.03386176e+00 -4.80788469e-01
3.17673832e-01 4.75530744e-01 1.77802905e-01 -1.22637033e+00
-6.58168554e-01 -1.14000690e+00 -4.72736418e-01 3.19448352e-01
3.11803818e-01 -4.32387888e-01 -4.29063976e-01 1.02685750e+00
-9.61808488e-02 -3.42418742e-03 2.78363258e-01 8.20643663e-01
-1.32669732e-01 9.27621543e-01 -5.40436745e-01 -7.36139059e-01
1.05034876e+00 -5.22056341e-01 -8.04044247e-01 8.54426995e-02
2.55092472e-01 -1.08783543e+00 2.46549726e-01 6.56221092e-01
-8.34080935e-01 -4.87917215e-01 -9.74974155e-01 1.07877386e+00
-4.13258582e-01 1.78243458e-01 3.42703968e-01 4.60521430e-01
-5.24746299e-01 8.87670994e-01 -7.14196146e-01 1.54937744e-01
-5.69454551e-01 4.90920782e-01 7.80012012e-02 5.73575616e-01
-1.36154449e+00 1.16493785e+00 6.98275149e-01 6.90127373e-01
-2.56120145e-01 -3.28017473e-01 -2.64945596e-01 2.86527604e-01
2.32712209e-01 -2.79866874e-01 8.18522692e-01 -9.54941273e-01
-1.40122736e+00 -2.40999028e-01 5.35843372e-01 -9.22597885e-01
6.19451821e-01 -7.17985481e-02 -8.71446788e-01 -2.07087174e-01
-3.08005452e-01 -3.64379168e-01 6.25455141e-01 -7.17423081e-01
-9.14355993e-01 -2.48137817e-01 -3.71219248e-01 2.19779268e-01
-8.48316103e-02 -1.22372666e-02 2.53044158e-01 -6.57985091e-01
1.48533404e-01 -1.16326451e+00 -3.51916164e-01 -1.01831996e+00
1.70199051e-01 3.77115458e-01 8.15050304e-01 -7.95794189e-01
1.34044099e+00 -1.94730711e+00 -4.12753187e-02 7.03517854e-01
-4.78691161e-01 6.01252690e-02 4.58006918e-01 7.41775393e-01
-5.67936301e-01 1.49560079e-01 2.47912988e-01 6.34425223e-01
-3.08248788e-01 2.29947791e-01 -4.66490477e-01 1.60355717e-01
3.18936139e-01 4.57448691e-01 -2.50633955e-01 2.06736177e-01
3.34551036e-01 1.32702500e-01 -4.41043600e-02 1.03087492e-01
-1.29296616e-01 3.49016696e-01 -5.44818699e-01 5.10021567e-01
3.47597122e-01 1.18607849e-01 6.93398044e-02 -6.16760887e-02
-6.85399234e-01 1.70712158e-01 -1.45382035e+00 6.48582458e-01
-5.13791144e-01 4.47734594e-01 -6.70845732e-02 -9.31071341e-01
1.33735144e+00 4.17530715e-01 7.99309433e-01 -5.11045694e-01
2.33122185e-01 4.41075087e-01 5.37729263e-01 -2.54379034e-01
4.84512091e-01 -2.83090085e-01 2.20603779e-01 2.87710786e-01
-4.66384053e-01 1.05084464e-01 2.37322062e-01 -4.37118262e-01
6.16652191e-01 -6.14360534e-02 3.02018523e-01 -8.71673673e-02
5.77908337e-01 -2.75577128e-01 8.75036895e-01 -3.18304040e-02
1.02340490e-01 1.86195537e-01 8.73726249e-01 -4.52140719e-01
-8.47589970e-01 -4.48092133e-01 -1.53234005e-01 1.82918817e-01
-2.70030320e-01 2.65132397e-01 1.04923315e-01 -9.73797664e-02
1.50742278e-01 7.58187830e-01 -8.69288146e-02 -1.01647720e-01
-2.57297784e-01 -1.09225237e+00 -2.33901247e-01 5.76141663e-02
2.26489171e-01 -1.02357006e+00 -8.44739556e-01 7.17943847e-01
3.51496249e-01 -8.32961857e-01 2.81578094e-01 5.58507025e-01
-1.00919139e+00 -9.15643752e-01 -6.35425866e-01 -3.34985219e-02
3.55137527e-01 -7.99409151e-02 8.68255317e-01 -1.79010496e-01
3.36706698e-01 -2.17212979e-02 -3.10062259e-01 -9.29510057e-01
-3.75872731e-01 1.66401148e-01 4.01567668e-02 1.11077324e-01
-5.88381141e-02 -4.32527810e-01 -4.06210482e-01 4.55010772e-01
-6.87401175e-01 -2.14764327e-01 7.36272812e-01 7.29145110e-01
3.68438393e-01 1.00208867e+00 7.60922074e-01 -6.96706414e-01
7.59487033e-01 -5.12713194e-01 -1.16585803e+00 6.17853925e-02
-1.20245039e+00 2.25344390e-01 4.19359654e-01 -6.00335523e-02
-1.00724983e+00 -7.69868046e-02 4.14869860e-02 -3.66069317e-01
3.15784961e-01 1.21274233e+00 2.65751593e-02 2.45041132e-01
-4.85547706e-02 2.69928545e-01 3.25709462e-01 -4.84296620e-01
-4.55316365e-01 4.12360400e-01 1.07397929e-01 -2.08489224e-02
9.67831075e-01 -1.12171046e-01 2.34087393e-01 -4.85334843e-01
-7.52045363e-02 -3.37051481e-01 -3.37905318e-01 -5.52625954e-01
4.87415463e-01 -9.01107788e-01 -5.93416929e-01 4.89883810e-01
-7.82528698e-01 3.33156765e-01 -1.39611676e-01 1.02917683e+00
-2.20342427e-01 -1.61722675e-01 -5.24604917e-01 -1.20245111e+00
-4.01657522e-01 -1.17448533e+00 8.67389366e-02 2.79995859e-01
-1.32070124e-01 -8.67823005e-01 -4.50006463e-02 1.63635954e-01
5.62447131e-01 4.74107802e-01 8.26912761e-01 -9.70253170e-01
-7.55849302e-01 -5.04940093e-01 2.73232341e-01 6.71020627e-01
1.68460235e-01 4.86952126e-01 -4.27142709e-01 -1.75118387e-01
5.05822897e-01 5.98554134e-01 3.70290041e-01 4.75088418e-01
4.71384227e-01 -2.71845311e-01 9.79337543e-02 1.69923469e-01
1.50637841e+00 1.08272016e+00 5.78523517e-01 8.93941402e-01
1.18316583e-01 8.15623164e-01 1.19283497e+00 7.35662997e-01
-2.53029376e-01 2.95045733e-01 6.35189474e-01 2.27558419e-01
6.52552485e-01 1.21458173e-01 5.44569790e-01 1.24621689e+00
-3.75788927e-01 1.93713114e-01 -9.80983973e-01 2.75581330e-01
-1.69343865e+00 -1.14299595e+00 -1.05056524e-01 2.09994388e+00
2.60649413e-01 8.93502295e-01 1.99652448e-01 5.89280188e-01
6.07354462e-01 8.17254633e-02 -2.66602069e-01 -4.86725807e-01
-1.37960434e-01 -2.41559014e-01 9.74339783e-01 1.13272265e-01
-5.03370106e-01 2.10257635e-01 5.30875540e+00 4.83765304e-01
-1.51691806e+00 -4.70366150e-01 6.52964234e-01 -2.14022979e-01
-4.35828209e-01 1.48197994e-01 -9.06960368e-01 8.98611367e-01
1.33831728e+00 -5.24138808e-01 3.10389131e-01 9.90877867e-01
6.34847462e-01 -6.17020801e-02 -3.45262676e-01 6.21712387e-01
-4.06390816e-01 -1.29129016e+00 -2.99480617e-01 4.31785107e-01
4.97166574e-01 1.04164407e-01 1.65319875e-01 7.94685781e-02
-2.30085254e-01 -4.96765971e-01 6.41572535e-01 1.07117081e+00
-2.04207525e-01 -1.28411341e+00 1.52911377e+00 4.28053856e-01
-1.15535653e+00 -5.16674399e-01 -2.54324913e-01 -4.38437998e-01
3.11078966e-01 7.61832416e-01 -8.47866833e-01 8.08231235e-01
4.57405895e-01 5.12190938e-01 -2.02553734e-01 7.66301811e-01
1.61690682e-01 4.80793834e-01 -4.02429074e-01 -2.45997205e-01
2.22187206e-01 -9.06557977e-01 4.07734990e-01 4.34110850e-01
7.79613078e-01 -8.11707377e-02 -2.56224036e-01 4.85409766e-01
5.37116230e-01 2.58404195e-01 -7.58146107e-01 -4.56613034e-01
4.86529559e-01 9.41531599e-01 -6.34217143e-01 1.41104475e-01
-4.78305310e-01 1.61857069e-01 -7.17652142e-01 2.61156201e-01
-5.90310395e-01 -3.07051033e-01 6.11202836e-01 3.62795889e-01
3.61614525e-01 -3.06454659e-01 -4.15261835e-01 -7.59171546e-01
5.96237071e-02 -9.09569740e-01 1.44764287e-02 -5.25407910e-01
-8.36622238e-01 6.02174580e-01 -9.42673385e-02 -1.63699567e+00
-8.29427004e-01 -7.90807664e-01 -6.17716730e-01 1.01640606e+00
-1.45633769e+00 -5.43873608e-01 4.58615005e-01 2.97481030e-01
3.27653080e-01 -9.61060405e-01 5.18417180e-01 7.92854000e-03
-7.21608996e-01 -3.87139499e-01 5.91806114e-01 6.24340065e-02
2.15939060e-01 -1.05549169e+00 6.40062848e-03 9.54102099e-01
1.12317048e-01 3.29774559e-01 8.87117088e-01 -9.31505263e-01
-1.22310066e+00 -4.95908231e-01 8.04143727e-01 6.51964992e-02
1.16258264e+00 3.55702400e-01 -6.62711740e-01 3.75603914e-01
2.05548301e-01 -6.51249707e-01 6.15074515e-01 1.09291665e-01
4.70349312e-01 -5.34234583e-01 -7.29550004e-01 3.90634477e-01
-1.04067646e-01 -2.31495798e-01 -5.73577762e-01 -2.65894383e-02
5.45539021e-01 -1.24120146e-01 -1.18290460e+00 4.69555885e-01
7.45679379e-01 -1.09911597e+00 6.25585496e-01 -2.92235047e-01
7.08650146e-03 -1.64451391e-01 1.61917061e-01 -1.34566772e+00
-4.22846705e-01 -3.86559814e-01 2.13456094e-01 1.27802968e+00
7.75068045e-01 -1.13797069e+00 6.46924734e-01 1.01935041e+00
2.03139037e-01 -7.68281043e-01 -9.02405798e-01 -7.58618653e-01
-3.32921654e-01 -4.11057174e-01 8.70714366e-01 7.31718659e-01
-2.31711581e-01 -7.67786987e-03 -4.67828721e-01 1.65175065e-01
1.05719484e-01 4.37521279e-01 5.66359818e-01 -1.42223668e+00
-4.23577428e-01 -5.19517243e-01 -4.46605057e-01 -1.28005326e-01
-1.64273426e-01 -3.29144895e-01 -7.19110966e-01 -1.01252258e+00
-6.40322089e-01 -3.35032791e-01 -8.79656732e-01 3.71626392e-02
3.46980661e-01 -5.77687740e-01 6.54903352e-01 4.88687694e-01
6.98859632e-01 3.98158044e-01 1.01928973e+00 1.97196275e-01
-3.91511858e-01 8.25832427e-01 -2.19269320e-01 6.20657444e-01
9.63435113e-01 -3.56367230e-01 -4.04502779e-01 2.49741822e-01
7.50213146e-01 4.77679133e-01 -2.63982147e-01 -1.11380517e+00
1.40727192e-01 -3.53844434e-01 4.56013948e-01 -9.69870090e-01
1.07142448e-01 -1.58820808e+00 1.11041522e+00 9.80919182e-01
1.36526152e-01 7.87246823e-01 1.28073350e-01 3.99815589e-01
-7.97046244e-01 -5.67981243e-01 1.98043585e-01 -1.08717740e-01
-5.31745613e-01 -7.16826692e-02 -5.93163311e-01 -6.83651507e-01
1.14508677e+00 -3.93389821e-01 -3.73075008e-02 -4.96947289e-01
-6.95862114e-01 1.52409524e-01 1.93024352e-01 4.62268502e-01
3.46309274e-01 -1.10698318e+00 -5.28182626e-01 2.54247606e-01
-4.74331379e-01 -3.06857824e-01 -5.23218289e-02 8.01283181e-01
-8.23106289e-01 7.82285213e-01 -5.25025845e-01 -1.31521240e-01
-8.39317620e-01 5.25834680e-01 3.27889383e-01 -4.97453094e-01
-2.80704210e-03 3.14534694e-01 -4.65620637e-01 2.18994617e-01
-1.64054200e-01 -4.42419171e-01 -6.85533345e-01 8.73611271e-01
2.81347573e-01 4.27928209e-01 2.61161625e-01 -5.09841979e-01
-1.52006567e-01 5.65284848e-01 1.54796410e-02 -2.29371011e-01
1.63002932e+00 -9.47124138e-02 -1.84164524e-01 9.83173907e-01
8.62977982e-01 1.63679078e-01 -8.88882756e-01 1.77146494e-01
5.64552665e-01 -2.96585262e-01 2.32323259e-01 -7.21656203e-01
-1.41235960e+00 4.09430653e-01 5.22160470e-01 5.43235600e-01
1.21929026e+00 -7.24599659e-01 4.96581048e-01 2.18516082e-01
4.47171539e-01 -1.26074374e+00 -5.08002996e-01 3.56637746e-01
1.01784503e+00 -8.90216112e-01 -7.12656230e-02 -1.12948008e-01
-8.65336001e-01 1.45326924e+00 -1.85792763e-02 -3.18651609e-02
1.25029886e+00 3.34523976e-01 2.53701031e-01 -3.27408686e-03
-9.10104632e-01 1.13257185e-01 2.07021013e-01 -6.02233075e-02
4.22354519e-01 1.85729086e-01 -7.45047450e-01 8.64078939e-01
-5.85597515e-01 -1.73540358e-02 6.63713634e-01 8.40490639e-01
-3.83325785e-01 -1.38164008e+00 -6.62772536e-01 5.60885906e-01
-6.00552022e-01 6.75785020e-02 -1.24307327e-01 1.00209558e+00
9.42018330e-02 5.80025494e-01 1.30279779e-01 -4.80926245e-01
5.26092350e-01 1.94228828e-01 -3.37959290e-01 -2.81910747e-01
-7.23237097e-01 5.36472619e-01 1.93356648e-01 -2.09048182e-01
-2.71011174e-01 -1.09604168e+00 -1.20168281e+00 -2.85923630e-01
-5.69564044e-01 4.80707616e-01 1.15092146e+00 9.36604738e-01
6.94463253e-02 5.70440590e-01 1.25880325e+00 -4.44831520e-01
-9.51095760e-01 -8.56295168e-01 -1.09152067e+00 -2.48095006e-01
-1.26863346e-01 -6.17331088e-01 -6.33919418e-01 -3.18840086e-01] | [4.563955783843994, 4.172496318817139] |
4b138b99-0fd8-4b55-8215-52860e6df258 | exploring-feature-representation-learning-for | 2111.10989 | null | https://arxiv.org/abs/2111.10989v1 | https://arxiv.org/pdf/2111.10989v1.pdf | Exploring Feature Representation Learning for Semi-supervised Medical Image Segmentation | This paper presents a simple yet effective two-stage framework for semi-supervised medical image segmentation. Our key insight is to explore the feature representation learning with labeled and unlabeled (i.e., pseudo labeled) images to enhance the segmentation performance. In the first stage, we present an aleatoric uncertainty-aware method, namely AUA, to improve the segmentation performance for generating high-quality pseudo labels. Considering the inherent ambiguity of medical images, AUA adaptively regularizes the consistency on images with low ambiguity. To enhance the representation learning, we propose a stage-adaptive contrastive learning method, including a boundary-aware contrastive loss to regularize the labeled images in the first stage and a prototype-aware contrastive loss to optimize both labeled and pseudo labeled images in the second stage. The boundary-aware contrastive loss only optimizes pixels around the segmentation boundaries to reduce the computational cost. The prototype-aware contrastive loss fully leverages both labeled images and pseudo labeled images by building a centroid for each class to reduce computational cost for pair-wise comparison. Our method achieves the best results on two public medical image segmentation benchmarks. Notably, our method outperforms the prior state-of-the-art by 5.7% on Dice for colon tumor segmentation relying on just 5% labeled images. | ['Kwang-Ting Cheng', 'Xiaomeng Li', 'Huimin Wu'] | 2021-11-22 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 5.35404921e-01 4.34388608e-01 -4.27973032e-01 -7.32411623e-01
-1.39581323e+00 -6.07133627e-01 3.27420443e-01 2.67170608e-01
-7.19791114e-01 3.98745060e-01 4.01300238e-03 -3.43861789e-01
1.29121104e-02 -5.33379734e-01 -6.73361719e-01 -8.93532872e-01
2.59505302e-01 5.33640981e-01 2.28220478e-01 4.41923469e-01
-1.42953349e-02 1.92854181e-01 -9.93491888e-01 2.36970231e-01
1.34828377e+00 1.18201888e+00 3.05802286e-01 4.09380615e-01
-1.96700007e-01 4.46053147e-01 -2.25113690e-01 -2.66807199e-01
5.88828802e-01 -5.33402443e-01 -1.01429641e+00 5.52406311e-01
4.34922487e-01 -3.08170319e-01 4.15679328e-02 1.38885403e+00
3.94243121e-01 -1.78110332e-03 9.02700424e-01 -8.92535210e-01
-5.05350292e-01 7.93266058e-01 -1.01319695e+00 1.22697540e-01
-2.65896708e-01 3.61153066e-01 8.98122728e-01 -8.08430493e-01
6.41310573e-01 9.24898207e-01 5.07786751e-01 4.79332656e-01
-1.27929425e+00 -5.50929248e-01 1.80312365e-01 -2.35739067e-01
-1.39177012e+00 -6.20745681e-02 6.77691281e-01 -4.92678672e-01
1.02583311e-01 1.28971711e-01 4.32449818e-01 4.69091833e-01
7.79189989e-02 1.12877500e+00 1.35613453e+00 -3.13321829e-01
2.96802640e-01 -5.97957559e-02 3.03534567e-01 1.11702657e+00
3.31489801e-01 7.12176934e-02 9.97433886e-02 -4.57215533e-02
8.64630342e-01 4.36104462e-02 -2.02334836e-01 -4.52779353e-01
-1.18699527e+00 7.27472961e-01 8.17026675e-01 4.38315654e-03
-4.39677328e-01 8.98136348e-02 3.35250556e-01 -2.76466608e-01
4.83301491e-01 5.55207789e-01 -3.70964825e-01 2.97893763e-01
-1.28619158e+00 -2.33016744e-01 3.57757509e-01 7.76167452e-01
8.67640495e-01 -3.70394498e-01 -7.07655191e-01 7.99253404e-01
5.81017017e-01 3.71790051e-01 3.76071930e-01 -1.13960218e+00
2.42930666e-01 7.85553455e-01 -1.18249215e-01 -4.70583469e-01
-4.34349716e-01 -6.53261185e-01 -7.94596672e-01 9.61253569e-02
6.45890296e-01 -2.07800373e-01 -1.72380209e+00 1.60603082e+00
5.97112954e-01 3.99547577e-01 -1.15760960e-01 9.97786343e-01
7.98122644e-01 3.91251147e-01 1.91175193e-01 -2.51348406e-01
1.48027921e+00 -1.47293437e+00 -6.05025411e-01 -1.71307683e-01
6.93576276e-01 -7.41254508e-01 1.05775774e+00 2.80600898e-02
-1.21816385e+00 -1.89496949e-01 -1.08881772e+00 3.25314216e-02
1.11331835e-01 3.31049949e-01 5.21796823e-01 7.26098895e-01
-7.70670891e-01 5.16686857e-01 -1.23323441e+00 1.93331972e-01
9.73691046e-01 3.45485121e-01 -4.89971600e-02 -3.16864014e-01
-8.84283602e-01 3.76326084e-01 3.49948168e-01 -1.42978027e-01
-8.90362501e-01 -1.11257398e+00 -1.07214737e+00 -2.06946254e-01
5.54653883e-01 -6.21423662e-01 1.09327340e+00 -8.71013045e-01
-1.36796057e+00 1.15252542e+00 -1.22049592e-01 -5.40637910e-01
8.64571929e-01 1.37389198e-01 9.51901674e-02 5.15027523e-01
3.93262476e-01 1.17955446e+00 7.46655405e-01 -1.59250951e+00
-7.00036108e-01 -2.67886668e-01 -2.09423512e-01 3.34866852e-01
1.14883585e-02 -5.56135774e-01 -1.01052034e+00 -9.54095542e-01
4.23419148e-01 -1.23265719e+00 -6.72027290e-01 3.31751227e-01
-6.83045805e-01 1.43931821e-01 4.49052632e-01 -5.50417602e-01
1.07298422e+00 -2.04491830e+00 -1.12892322e-01 4.48383570e-01
3.83612424e-01 1.40518814e-01 -9.12208855e-02 -6.99507594e-01
1.97175026e-01 2.72366881e-01 -1.05868292e+00 -6.17304981e-01
-2.25460663e-01 3.06581914e-01 1.29471332e-01 5.27622461e-01
2.35014364e-01 1.10332394e+00 -1.17708242e+00 -9.23331797e-01
3.32289666e-01 4.16365892e-01 -6.82946265e-01 5.13201840e-02
-1.23587608e-01 8.43126893e-01 -5.57873666e-01 7.19648898e-01
9.77466345e-01 -6.57044411e-01 1.42472163e-02 -4.00951326e-01
2.04956442e-01 -6.47074357e-02 -1.04301322e+00 1.96677649e+00
-4.24789548e-01 1.77866325e-01 2.78315872e-01 -6.89250946e-01
4.86487061e-01 1.25993267e-01 8.16432476e-01 -3.63961071e-01
3.10609221e-01 3.93620819e-01 -6.39869571e-02 -1.44501701e-01
1.37690201e-01 2.60097496e-02 3.15287150e-02 5.46766579e-01
-1.28559932e-01 -4.96981084e-01 1.84433758e-01 2.84442097e-01
8.31712902e-01 1.41728759e-01 1.65755883e-01 -5.58377266e-01
5.22774816e-01 1.11384108e-03 8.44164550e-01 7.18766868e-01
-6.00507557e-01 1.18602872e+00 2.80592531e-01 -2.10383818e-01
-6.77626550e-01 -1.17246032e+00 -4.52916980e-01 6.35507047e-01
5.51807523e-01 1.19959039e-03 -1.13137400e+00 -1.37755954e+00
-1.22213505e-01 4.76702988e-01 -8.26088667e-01 -1.27442284e-02
-4.97505307e-01 -1.05750132e+00 3.73668492e-01 5.18056273e-01
7.46908367e-01 -6.35120869e-01 -5.78338742e-01 5.92246391e-02
-2.90015370e-01 -1.15116394e+00 -1.09352970e+00 3.34633857e-01
-9.11875427e-01 -1.08364856e+00 -1.06770563e+00 -7.39473522e-01
1.31151307e+00 2.26935558e-02 1.04637206e+00 1.07010350e-01
-4.32491601e-01 1.73127785e-01 -2.26101309e-01 -1.10718437e-01
-3.21174979e-01 5.13379872e-02 -4.66810495e-01 4.74086553e-02
-1.64047271e-01 -1.13062933e-01 -1.08060622e+00 4.44796860e-01
-1.08801317e+00 9.07905921e-02 7.84818828e-01 1.17108226e+00
1.30501509e+00 -9.73784998e-02 1.07972033e-01 -1.36806417e+00
4.68758531e-02 -3.23056430e-01 -5.66919506e-01 5.43037236e-01
-9.49572504e-01 3.88112396e-01 3.17434609e-01 -3.41807395e-01
-1.14973545e+00 5.10693848e-01 -1.16861761e-01 -3.44013363e-01
5.26102223e-02 3.36595476e-01 4.88130860e-02 -2.52569944e-01
5.09477675e-01 2.28859186e-02 1.57574527e-02 -1.47381604e-01
5.95474660e-01 3.99724364e-01 6.38402641e-01 -5.51027119e-01
6.32492721e-01 7.51267612e-01 1.16549253e-01 -2.08845139e-01
-1.22683299e+00 -6.57299757e-01 -5.19754112e-01 6.16151989e-02
1.09548819e+00 -8.63522410e-01 -1.61767781e-01 5.65288544e-01
-6.73876226e-01 -5.23037314e-01 -6.46732926e-01 5.68694949e-01
-4.67851162e-01 5.66063762e-01 -8.12241256e-01 -1.91608578e-01
-6.73234761e-01 -1.86691058e+00 1.34932363e+00 4.79532868e-01
6.68324009e-02 -1.10049438e+00 -1.98907256e-01 6.47828758e-01
2.36598641e-01 2.84029067e-01 7.33554423e-01 -5.76247573e-01
-5.41870534e-01 8.44414607e-02 -5.91817677e-01 4.90299910e-01
3.88960034e-01 -1.66166231e-01 -8.88636827e-01 -2.34796658e-01
-1.33087784e-01 -3.00077468e-01 1.25081754e+00 9.90322351e-01
1.43523812e+00 -1.56053193e-02 -3.70122612e-01 1.02750361e+00
1.17117381e+00 1.96962636e-02 2.68759996e-01 -5.24826050e-02
9.63595390e-01 4.08330739e-01 8.19230020e-01 2.60081679e-01
4.05049860e-01 4.35494125e-01 3.16034615e-01 -5.47385335e-01
-5.49827397e-01 -1.39480442e-01 -2.62107372e-01 5.60741901e-01
4.88862872e-01 -1.02973290e-01 -9.30217385e-01 5.34685969e-01
-1.68210042e+00 -2.35922232e-01 1.13299765e-01 2.11185288e+00
1.17926323e+00 1.16937578e-01 -1.91790491e-01 -1.89188957e-01
8.28715920e-01 1.03854626e-01 -8.99894714e-01 2.46581912e-01
1.21488310e-01 2.18459114e-01 7.88316846e-01 7.60309339e-01
-1.50766802e+00 8.56204033e-01 5.70564890e+00 1.12541711e+00
-1.14726329e+00 1.57728598e-01 1.40302157e+00 1.68448180e-01
-4.65707958e-01 -8.38843584e-02 -5.85078239e-01 5.31981826e-01
4.58641857e-01 1.29996136e-01 -1.60825974e-03 7.99265683e-01
6.97594583e-02 -3.12561959e-01 -9.69165564e-01 8.27707112e-01
-5.21804355e-02 -1.51352131e+00 1.33100376e-01 4.68512438e-02
1.24771047e+00 -9.81173478e-03 3.47082525e-01 -1.84099097e-02
3.55464667e-01 -8.55318129e-01 5.77624917e-01 2.93461829e-01
1.00488305e+00 -7.74942517e-01 8.67775500e-01 -4.65376154e-02
-1.13213181e+00 2.63656139e-01 1.60397440e-02 5.96073329e-01
3.30771834e-01 8.83070707e-01 -7.32062340e-01 5.43188453e-01
4.23723280e-01 6.60099685e-01 -6.13626838e-01 1.17553794e+00
-3.45091671e-01 6.85406744e-01 -3.91832590e-01 5.08270442e-01
6.31460071e-01 -3.38027835e-01 3.78369242e-01 1.18184280e+00
2.23470703e-02 2.02971801e-01 6.22707725e-01 8.98187935e-01
-3.45288903e-01 -3.37092578e-02 2.47746781e-01 2.98781604e-01
4.67153221e-01 1.49154019e+00 -1.33461869e+00 -4.43015754e-01
-4.90642600e-02 9.98535752e-01 -5.69403358e-02 2.13570684e-01
-7.94278562e-01 -9.73902494e-02 2.40351558e-02 5.88088259e-02
7.86996409e-02 1.87278286e-01 -7.96025157e-01 -9.59039211e-01
-1.90184638e-01 -5.47863543e-01 6.92500114e-01 -2.06211805e-01
-1.28948128e+00 5.83065987e-01 1.44147381e-01 -1.14984548e+00
-1.15644652e-02 -3.17321092e-01 -5.25266945e-01 7.85284221e-01
-1.71875119e+00 -1.21012414e+00 -3.71395856e-01 2.57565618e-01
5.65621018e-01 3.10018718e-01 3.95059705e-01 3.30399662e-01
-6.61355197e-01 8.13233376e-01 6.29183725e-02 2.58958012e-01
8.63197923e-01 -1.55839813e+00 1.64723769e-01 8.65108013e-01
3.31252185e-03 4.36568081e-01 2.70659894e-01 -5.60238719e-01
-7.32281506e-01 -1.32539022e+00 2.51230955e-01 -1.94395155e-01
1.59267098e-01 9.02552754e-02 -8.25418830e-01 3.99958581e-01
-6.22569881e-02 7.76207983e-01 7.50267446e-01 -4.93481487e-01
-1.24864668e-01 9.56953503e-04 -1.64331567e+00 5.93415141e-01
6.87664449e-01 -1.88255370e-01 -2.77341813e-01 4.15833265e-01
8.74286592e-01 -8.25429678e-01 -8.83378565e-01 7.84762502e-01
2.78948873e-01 -5.69161594e-01 9.46397066e-01 3.39842774e-02
3.55100721e-01 -4.30500537e-01 1.60023525e-01 -1.06096852e+00
-7.46408626e-02 -6.13545299e-01 2.73964971e-01 1.00061572e+00
7.68467546e-01 -4.92226005e-01 1.12532735e+00 9.20057714e-01
-3.84239018e-01 -1.21768785e+00 -9.25493777e-01 -5.17442346e-01
2.81997532e-01 -2.11775854e-01 2.60668814e-01 8.33064139e-01
-4.29173082e-01 -1.13647804e-02 6.67293444e-02 6.32926077e-02
9.61200714e-01 2.58381039e-01 1.61440089e-01 -7.98788786e-01
-3.51964056e-01 -5.06067693e-01 -4.07358371e-02 -1.13112438e+00
9.78790596e-02 -1.17032993e+00 2.51440197e-01 -1.54853427e+00
5.50977886e-01 -8.32087278e-01 -4.67947066e-01 4.84944671e-01
-7.24730611e-01 4.32725161e-01 1.94751248e-02 2.21298933e-01
-6.87778592e-01 3.66516352e-01 1.70570934e+00 -4.03111070e-01
-3.38632405e-01 1.27031460e-01 -6.51139259e-01 9.41411138e-01
6.29998863e-01 -6.68288887e-01 -5.37601471e-01 -3.85298193e-01
-3.38066101e-01 -1.87757015e-01 1.50990129e-01 -7.62814045e-01
1.88756704e-01 -4.32729088e-02 4.05022949e-01 -5.79062641e-01
-5.82416691e-02 -6.90154850e-01 -3.82827550e-01 7.74052799e-01
-6.36907279e-01 -3.41136605e-01 3.05012502e-02 6.08102381e-01
-2.11252674e-01 -3.54735404e-01 1.45655620e+00 -1.28048658e-01
-4.60583061e-01 6.50907874e-01 1.07948199e-01 4.61291164e-01
1.07727611e+00 4.22760472e-02 -6.46002814e-02 -1.01144865e-01
-7.29119718e-01 6.51253104e-01 5.40380955e-01 -1.29973933e-01
5.07713795e-01 -1.07465172e+00 -6.47939265e-01 2.58659068e-02
-1.02945780e-02 6.23408079e-01 4.16724920e-01 1.00741720e+00
-7.61570394e-01 4.85915765e-02 3.41153517e-02 -8.94450843e-01
-1.04299629e+00 3.47801626e-01 4.50353235e-01 -7.96453893e-01
-5.19968688e-01 1.28062463e+00 6.00609720e-01 -4.65639323e-01
1.62867770e-01 -4.91097301e-01 8.19519162e-03 -2.07577690e-01
3.04122120e-01 3.27162027e-01 -8.63459632e-02 -6.00979328e-01
-4.90261585e-01 6.85389996e-01 -5.15929759e-01 -3.40588778e-01
8.91548574e-01 -1.28443792e-01 1.09180860e-01 9.84707549e-02
1.33467567e+00 -7.90318996e-02 -1.79261136e+00 -4.37293261e-01
-1.15603171e-01 -3.92172039e-01 5.45353591e-01 -1.08937204e+00
-1.54109335e+00 6.78190410e-01 8.34009707e-01 -4.96903390e-01
1.26806438e+00 -3.06483470e-02 9.86153364e-01 -5.58794700e-02
9.03826728e-02 -1.04798639e+00 1.89125180e-01 1.31316900e-01
3.05779576e-01 -1.76372385e+00 3.50240350e-01 -8.11103821e-01
-9.94606495e-01 7.56891072e-01 4.12179440e-01 -1.09586835e-01
7.51164913e-01 4.12796676e-01 2.84035593e-01 -9.64818299e-02
-1.24288358e-01 -1.71585083e-01 7.60937870e-01 2.81794310e-01
3.23018819e-01 2.55817354e-01 -4.60910976e-01 6.30989969e-01
3.27622950e-01 -1.84755698e-01 2.03931898e-01 8.80684614e-01
-2.41423428e-01 -1.12455177e+00 -2.14321211e-01 6.66197598e-01
-6.83956742e-01 -1.36777297e-01 -2.77663972e-02 5.56962788e-01
1.63446113e-01 8.28757882e-01 6.52962178e-02 -3.39743122e-03
3.20460461e-02 -3.45796078e-01 4.73693579e-01 -7.85878122e-01
-5.73569894e-01 3.64174813e-01 -4.20874119e-01 -5.91964126e-01
-4.64571595e-01 -6.63365901e-01 -1.84595764e+00 3.58204395e-01
-4.40989166e-01 4.65491526e-02 5.00179887e-01 8.49848747e-01
3.18569869e-01 5.51084280e-01 7.86669195e-01 -6.68655753e-01
-5.71681321e-01 -5.49709380e-01 -4.92349118e-01 5.26171684e-01
2.86826819e-01 -5.23677707e-01 -3.59265238e-01 1.61591768e-02] | [14.654770851135254, -2.0935847759246826] |
c739c719-a541-41e0-9fdb-354e640477eb | modeling-semantic-compositionality-with | 1907.04744 | null | https://arxiv.org/abs/1907.04744v1 | https://arxiv.org/pdf/1907.04744v1.pdf | Modeling Semantic Compositionality with Sememe Knowledge | Semantic compositionality (SC) refers to the phenomenon that the meaning of a complex linguistic unit can be composed of the meanings of its constituents. Most related works focus on using complicated compositionality functions to model SC while few works consider external knowledge in models. In this paper, we verify the effectiveness of sememes, the minimum semantic units of human languages, in modeling SC by a confirmatory experiment. Furthermore, we make the first attempt to incorporate sememe knowledge into SC models, and employ the sememeincorporated models in learning representations of multiword expressions, a typical task of SC. In experiments, we implement our models by incorporating knowledge from a famous sememe knowledge base HowNet and perform both intrinsic and extrinsic evaluations. Experimental results show that our models achieve significant performance boost as compared to the baseline methods without considering sememe knowledge. We further conduct quantitative analysis and case studies to demonstrate the effectiveness of applying sememe knowledge in modeling SC. All the code and data of this paper can be obtained on https://github.com/thunlp/Sememe-SC. | ['Jun-Jie Huang', 'Chenghao Yang', 'Qun Liu', 'Fanchao Qi', 'Maosong Sun', 'Zhiyuan Liu', 'Xiao Chen'] | 2019-07-10 | modeling-semantic-compositionality-with-1 | https://aclanthology.org/P19-1571 | https://aclanthology.org/P19-1571.pdf | acl-2019-7 | ['multi-word-expression-embedding', 'multi-word-expression-sememe-prediction'] | ['natural-language-processing', 'natural-language-processing'] | [ 8.00992623e-02 1.20924719e-01 -3.74442458e-01 -4.28613931e-01
-3.13589126e-01 -5.40342212e-01 7.69028604e-01 -4.84792516e-02
-5.60162485e-01 4.40281093e-01 6.61062539e-01 -3.16759109e-01
2.13392705e-01 -8.25664520e-01 -8.71461093e-01 -1.64658248e-01
4.31340456e-01 2.02349499e-01 2.45580822e-01 -4.53590304e-01
2.14426130e-01 -7.54362270e-02 -1.37149119e+00 6.14715219e-01
1.15066350e+00 8.58418882e-01 3.36077094e-01 -1.09457232e-01
-5.68762362e-01 1.17977262e+00 -4.72163588e-01 -8.73774469e-01
-3.49017493e-02 -5.00576854e-01 -8.50132883e-01 -2.64557481e-01
1.64093360e-01 2.38501742e-01 1.06411777e-01 1.21438944e+00
1.40283883e-01 7.12070167e-02 7.13669419e-01 -9.66505706e-01
-1.35240376e+00 1.44100249e+00 -2.35460047e-02 3.75433341e-02
4.76818502e-01 -1.01296291e-01 1.50490606e+00 -1.04053640e+00
4.94035393e-01 1.50612080e+00 5.32019556e-01 5.45990407e-01
-1.11223269e+00 -8.63425672e-01 4.02928233e-01 5.25851071e-01
-1.30041504e+00 -4.89846617e-01 9.88124013e-01 -5.99932522e-02
1.10854459e+00 1.02391466e-01 4.86482888e-01 1.19968796e+00
-9.28462446e-02 1.03842723e+00 1.18405223e+00 -8.04401398e-01
-3.78184617e-02 2.79528230e-01 3.82610202e-01 5.27719378e-01
3.58334929e-01 -6.40787557e-02 -6.44661605e-01 2.70833462e-01
5.61818779e-01 -3.79084617e-01 -2.15723440e-01 4.14607152e-02
-1.08191550e+00 7.78055847e-01 6.30795658e-01 6.48103535e-01
-1.96944490e-01 2.95161396e-01 3.53940099e-01 2.70452499e-01
1.17206298e-01 5.09352267e-01 -3.90399158e-01 1.66812122e-01
-3.86266083e-01 -1.48194015e-01 6.97118580e-01 1.14614713e+00
6.47897780e-01 1.03308067e-01 -3.58470455e-02 1.03054452e+00
4.23089415e-01 6.14872575e-01 9.31498110e-01 -9.42708731e-01
2.69810975e-01 1.02696228e+00 -2.40693688e-01 -7.83732831e-01
-2.27251217e-01 -2.23462060e-01 -2.43980333e-01 -5.91862559e-01
-6.54464588e-02 -9.32298526e-02 -6.24602973e-01 1.97873938e+00
4.29049656e-02 2.73370236e-01 2.60938674e-01 7.24245191e-01
1.26306498e+00 5.88412166e-01 5.12345374e-01 -5.25918044e-02
1.49563098e+00 -1.09344590e+00 -9.25159037e-01 -5.27105391e-01
7.89032876e-01 -6.59213364e-01 1.69621968e+00 7.62891173e-02
-1.01100922e+00 -4.71372724e-01 -9.59023476e-01 -3.17261249e-01
-7.12435901e-01 1.62603870e-01 7.79892981e-01 5.54607034e-01
-8.62897336e-01 2.13087752e-01 -7.20321357e-01 -2.66538322e-01
1.34213239e-01 1.32902339e-01 -1.64604187e-01 9.26949754e-02
-1.71625721e+00 1.25701988e+00 6.71596169e-01 -1.00535743e-01
-6.45685375e-01 -6.12501860e-01 -1.10120034e+00 -7.81526137e-03
4.98962194e-01 -7.09141791e-01 1.24905550e+00 -1.44797313e+00
-1.32946503e+00 8.61775756e-01 -4.34695005e-01 -3.09742719e-01
-1.04114518e-01 -2.64341921e-01 -7.78020024e-01 -2.11768657e-01
1.47107139e-01 5.85467994e-01 3.93766791e-01 -1.32499337e+00
-5.92404187e-01 -1.84027821e-01 4.66534197e-01 2.36651972e-01
-7.18972385e-01 3.06643814e-01 -4.58374470e-01 -7.48254120e-01
-1.23969451e-01 -7.74442673e-01 2.17112601e-01 -6.20586395e-01
-1.65045515e-01 -4.66323853e-01 1.38598949e-01 -5.48806727e-01
1.59545541e+00 -2.16597390e+00 2.27774173e-01 1.16828486e-01
-2.29450107e-01 1.63386494e-01 -1.56282350e-01 4.29228693e-01
8.62820148e-02 3.92941028e-01 -3.41349959e-01 -2.44618535e-01
3.89369220e-01 3.24761927e-01 -2.71940261e-01 -2.39873827e-01
1.70592159e-01 1.45140195e+00 -7.97861099e-01 -4.26331818e-01
-7.62867182e-03 4.26447123e-01 -4.67625946e-01 -1.52767792e-01
-2.89904207e-01 1.26498770e-02 -2.23704591e-01 5.01180649e-01
2.33806491e-01 -4.17181104e-01 6.59713268e-01 -4.31121618e-01
1.32690057e-01 9.66457248e-01 -9.88611400e-01 1.52900743e+00
-7.74428904e-01 9.33148041e-02 -4.12543237e-01 -7.46454477e-01
6.38080418e-01 2.65663713e-01 -2.38084674e-01 -9.36257899e-01
3.74005765e-01 4.06338096e-01 2.51014143e-01 -2.42348194e-01
4.49444145e-01 -7.23342538e-01 -2.02546030e-01 3.51351202e-01
8.11035857e-02 -2.41600610e-02 4.33198631e-01 7.74724483e-02
7.35358477e-01 2.52208352e-01 7.18406796e-01 -4.23537582e-01
8.91150594e-01 -6.14978038e-02 5.92848003e-01 2.28455439e-01
1.08946338e-01 -1.44568607e-02 2.29541957e-01 3.49135771e-02
-4.76728678e-01 -1.08286154e+00 -2.11657181e-01 1.23715055e+00
3.16058725e-01 -7.93121636e-01 -6.76416039e-01 -7.15091348e-01
-5.75567037e-02 1.45362353e+00 -6.47304773e-01 -2.48140603e-01
-6.03723764e-01 -5.29484212e-01 7.73776352e-01 8.84835064e-01
4.63501394e-01 -1.51282239e+00 -3.80999953e-01 6.68153241e-02
-4.08243209e-01 -1.23018181e+00 -3.17480296e-01 -9.54923704e-02
-7.01653481e-01 -9.44863975e-01 1.10060953e-01 -1.06006551e+00
4.17168289e-01 1.17690079e-01 1.16903794e+00 2.80052453e-01
1.98430479e-01 2.39035666e-01 -6.32237613e-01 -5.92798293e-01
-4.24876183e-01 -1.74385700e-02 -1.34346206e-02 -1.12736009e-01
1.04477417e+00 -4.66795772e-01 -1.92068517e-01 2.00501591e-01
-1.05571175e+00 1.27497256e-01 4.41422611e-01 4.22852576e-01
4.98703212e-01 -1.15574904e-01 5.75444281e-01 -1.21411681e+00
8.13669205e-01 -4.92717236e-01 -2.74350405e-01 3.84958535e-01
-6.49395466e-01 2.82508910e-01 6.95537567e-01 -3.27542067e-01
-1.29368210e+00 -3.56660753e-01 -2.35938549e-01 -2.58781249e-03
-4.97625172e-02 9.24607694e-01 -5.00203073e-01 2.72512823e-01
5.49037874e-01 3.50136250e-01 -9.90805477e-02 -4.30173367e-01
6.60625041e-01 5.20412266e-01 2.60146171e-01 -1.00103402e+00
4.70988512e-01 4.01123494e-01 -4.02770311e-01 -5.90923190e-01
-1.20804513e+00 -2.86289036e-01 -4.81414944e-01 -3.08025070e-02
7.20046043e-01 -1.10501564e+00 -4.02942568e-01 8.53789300e-02
-9.90522981e-01 -2.89712727e-01 -1.33338198e-01 3.76275331e-01
-2.97318369e-01 4.58617061e-01 -7.63689756e-01 -4.01782632e-01
-3.24308723e-01 -7.70841777e-01 7.89155424e-01 -1.53438998e-02
-4.75037247e-01 -1.42052472e+00 -7.78829455e-02 3.42053801e-01
5.23410559e-01 -1.53216705e-01 1.26784027e+00 -6.54582977e-01
-2.93963552e-01 1.44477159e-01 -2.46267971e-02 5.77904999e-01
4.07684237e-01 -2.43405655e-01 -8.95661533e-01 1.64395019e-01
2.52221879e-02 -4.01904702e-01 8.86763275e-01 1.36704057e-01
8.50969136e-01 -3.57461512e-01 -4.55776565e-02 4.45126772e-01
1.54514337e+00 1.23336844e-01 5.41688561e-01 4.24280643e-01
7.56315053e-01 6.40175223e-01 5.45543611e-01 1.10454336e-01
1.06223595e+00 5.63423574e-01 1.18146114e-01 3.62276107e-01
-3.63855690e-01 -6.02388382e-01 7.24376559e-01 1.54037130e+00
-1.62597790e-01 -1.04382202e-01 -9.26786602e-01 7.65543997e-01
-1.75696158e+00 -7.36862242e-01 -2.67040670e-01 1.84202433e+00
1.22381866e+00 9.41364616e-02 -2.01892808e-01 -1.08336486e-01
5.32011509e-01 2.60947999e-02 -1.29670739e-01 -2.85769820e-01
-4.01678085e-01 2.99306005e-01 1.42807364e-01 6.27525866e-01
-7.49755383e-01 1.65336061e+00 5.64882421e+00 8.73052835e-01
-9.81675923e-01 4.30977941e-01 -1.18623756e-01 3.50971706e-02
-8.36418033e-01 2.17093110e-01 -7.70105958e-01 4.93134767e-01
1.01853704e+00 -4.13022339e-01 5.29443860e-01 5.96618772e-01
-1.03490785e-01 2.45227903e-01 -1.21882164e+00 7.95948029e-01
3.93172473e-01 -1.12900507e+00 6.15045726e-01 -3.59183043e-01
6.78517759e-01 -5.32626770e-02 -1.22280620e-01 5.48816621e-01
5.19142687e-01 -9.30949330e-01 9.07965422e-01 3.05864006e-01
4.46077704e-01 -6.49585724e-01 6.75836146e-01 2.40003183e-01
-1.33679581e+00 -8.62433985e-02 -4.33146894e-01 -1.95592716e-01
2.94431031e-01 1.46478802e-01 -4.29693192e-01 6.76188827e-01
4.32831079e-01 8.87486815e-01 -7.91619062e-01 5.24871051e-01
-1.02066767e+00 8.58343482e-01 -1.30191818e-01 -2.57643968e-01
8.50746930e-02 -1.16012722e-01 2.38506421e-01 1.46284974e+00
1.88377857e-01 1.35898620e-01 -8.40203464e-02 1.12712348e+00
-3.50770414e-01 6.38678968e-01 -4.76202041e-01 -2.88520962e-01
7.10567355e-01 9.22550142e-01 -4.14624810e-01 -4.88787919e-01
-7.61235714e-01 8.39947522e-01 5.37579656e-01 2.41813451e-01
-8.14951181e-01 -1.45916179e-01 5.26477098e-01 -1.21072726e-02
2.37496495e-01 -9.68423113e-02 -4.78729010e-01 -1.31956792e+00
1.74759284e-01 -9.03873801e-01 5.29666543e-01 -7.85932004e-01
-1.55636442e+00 5.61968863e-01 1.61866218e-01 -8.92407358e-01
-1.29505783e-01 -9.47966754e-01 -4.80862558e-01 5.54176569e-01
-1.70920503e+00 -1.50872743e+00 -8.14128518e-02 5.52898884e-01
5.35711110e-01 -1.84952945e-01 9.07641232e-01 1.52194396e-01
-3.36940169e-01 6.61762893e-01 -4.29920226e-01 2.14718968e-01
6.04973674e-01 -1.16480207e+00 4.34707254e-01 1.01625645e+00
4.39865261e-01 1.20065188e+00 3.77990037e-01 -8.52048457e-01
-1.21319878e+00 -9.65718806e-01 1.36541033e+00 -8.23919654e-01
1.01952767e+00 -3.00722480e-01 -1.00084639e+00 1.20488119e+00
4.43960726e-01 -3.55455846e-01 1.10228646e+00 3.00051630e-01
-6.53799355e-01 9.31439102e-02 -6.27207637e-01 7.90075064e-01
1.46080709e+00 -5.47862768e-01 -1.32919419e+00 -1.90993130e-01
1.11064947e+00 7.16976896e-02 -8.62799346e-01 5.60870171e-01
4.33684349e-01 -6.36423528e-01 7.63440907e-01 -6.88268602e-01
6.85549080e-01 -5.44152021e-01 -5.08883834e-01 -1.46054614e+00
-4.90398318e-01 1.48854077e-01 -1.97304577e-01 1.31435907e+00
9.40326095e-01 -6.26017451e-01 1.67633027e-01 4.58845496e-01
-2.49782816e-01 -2.95327574e-01 -5.54373205e-01 -9.26376283e-01
4.00837779e-01 -7.10325897e-01 9.30268049e-01 1.20791888e+00
4.12112564e-01 7.76477993e-01 3.76974419e-02 7.39360973e-02
3.42416316e-01 1.60429806e-01 3.56061071e-01 -1.04509997e+00
-3.51722896e-01 -5.52175462e-01 -1.45343423e-01 -8.40156496e-01
8.79901052e-01 -1.54717577e+00 -1.54733136e-01 -1.73826444e+00
3.34850371e-01 -2.24708214e-01 -6.43622518e-01 7.06534088e-01
-4.14201349e-01 5.58251590e-02 4.64281499e-01 8.71478394e-02
-6.20570481e-01 7.56150424e-01 1.15462005e+00 -5.99538558e-04
8.67110491e-02 -5.76250017e-01 -1.13754082e+00 8.77661645e-01
9.28882420e-01 -2.74498999e-01 -6.70137823e-01 -7.65477955e-01
4.64596480e-01 -6.12908185e-01 1.45856500e-01 -7.04976141e-01
1.57297671e-01 -2.59830892e-01 3.94754522e-02 4.23528776e-02
2.53928632e-01 -9.40883517e-01 4.91961800e-02 1.47661150e-01
-4.23161924e-01 1.86492994e-01 3.90196472e-01 1.56040952e-01
-4.39043552e-01 -4.51478928e-01 4.27882880e-01 -3.31840754e-01
-1.19214404e+00 -2.66878694e-01 -2.36820102e-01 3.86751473e-01
7.08782613e-01 -5.77545092e-02 -4.04899567e-01 -1.87933713e-01
-4.91503328e-01 2.24273488e-01 6.04306579e-01 9.06698942e-01
6.59830868e-01 -1.51668382e+00 -5.72595239e-01 8.76644477e-02
4.93903071e-01 -4.86746460e-01 -1.93421677e-01 6.99191689e-01
-3.00811857e-01 4.71071661e-01 -2.12644696e-01 -3.15173492e-02
-9.94601846e-01 6.79492652e-01 2.28056312e-01 1.26404598e-01
-4.06682342e-01 7.82078683e-01 4.53602344e-01 -8.42024148e-01
6.10137135e-02 -4.29369330e-01 -4.33629036e-01 -1.28937185e-01
5.18340945e-01 1.42726406e-01 -3.70872729e-02 -8.47002387e-01
-5.09602487e-01 3.80327284e-01 3.94250490e-02 -1.76294297e-01
1.10038686e+00 -1.74651623e-01 -5.24228811e-01 7.84601271e-01
8.71635139e-01 3.27482879e-01 -4.69200879e-01 -4.76955891e-01
4.46767509e-01 -1.59270510e-01 -2.00231627e-01 -1.04954517e+00
-7.88288176e-01 7.51034379e-01 5.45019768e-02 -2.76658863e-01
9.79159534e-01 1.99462891e-01 6.77806616e-01 3.79855871e-01
5.34494221e-01 -1.21406698e+00 -9.99567732e-02 9.25609171e-01
9.32103515e-01 -9.74494338e-01 -4.09249008e-01 -6.51673079e-01
-1.05778861e+00 6.82758868e-01 9.30311263e-01 -1.33632913e-01
7.53481090e-01 9.45351273e-02 3.37293655e-01 -2.37354249e-01
-9.59628284e-01 -3.78403574e-01 3.95246148e-01 2.73211092e-01
1.00498188e+00 4.61118996e-01 -8.15587878e-01 1.15657640e+00
-5.97344697e-01 -1.17661521e-01 2.94134974e-01 7.49359965e-01
-4.09259915e-01 -1.47907114e+00 1.02878466e-01 3.31258953e-01
-4.42470372e-01 -7.39171624e-01 -8.18204761e-01 7.84734666e-01
1.55915692e-01 8.78969073e-01 -2.02756748e-01 -2.97765255e-01
4.41836953e-01 5.46395600e-01 4.49915886e-01 -8.57613266e-01
-6.00302100e-01 -1.79786444e-01 4.77560878e-01 -5.41366458e-01
-7.55931437e-01 -4.99058574e-01 -1.88412118e+00 -1.77782089e-01
-1.41742498e-01 8.14477429e-02 4.62456495e-01 1.08743119e+00
2.40296930e-01 5.01682103e-01 1.31941527e-01 -1.33059844e-01
-3.81819576e-01 -1.13639450e+00 -3.98927301e-01 7.51755357e-01
-2.20928863e-01 -7.28259742e-01 -4.19283271e-01 1.41370520e-01] | [10.537849426269531, 9.090995788574219] |
5d2b1139-7564-4de6-9426-5415e5bdbb3c | stillfast-an-end-to-end-approach-for-short | 2304.03959 | null | https://arxiv.org/abs/2304.03959v1 | https://arxiv.org/pdf/2304.03959v1.pdf | StillFast: An End-to-End Approach for Short-Term Object Interaction Anticipation | Anticipation problem has been studied considering different aspects such as predicting humans' locations, predicting hands and objects trajectories, and forecasting actions and human-object interactions. In this paper, we studied the short-term object interaction anticipation problem from the egocentric point of view, proposing a new end-to-end architecture named StillFast. Our approach simultaneously processes a still image and a video detecting and localizing next-active objects, predicting the verb which describes the future interaction and determining when the interaction will start. Experiments on the large-scale egocentric dataset EGO4D show that our method outperformed state-of-the-art approaches on the considered task. Our method is ranked first in the public leaderboard of the EGO4D short term object interaction anticipation challenge 2022. Please see the project web page for code and additional details: https://iplab.dmi.unict.it/stillfast/. | ['Antonino Furnari', 'Giovanni Maria Farinella', 'Francesco Ragusa'] | 2023-04-08 | null | null | null | null | ['short-term-object-interaction-anticipation', 'human-object-interaction-detection'] | ['computer-vision', 'computer-vision'] | [-2.98154622e-01 3.62203158e-02 -1.18186690e-01 -5.24967790e-01
-1.24904230e-01 -5.58612525e-01 8.21383715e-01 -1.29387215e-01
-4.05482888e-01 2.18021855e-01 8.62151444e-01 3.34790260e-01
-2.16126710e-01 -2.20676914e-01 -6.20394707e-01 -3.70202988e-01
-4.56783175e-01 7.99015939e-01 1.41910836e-01 1.35735825e-01
4.60262060e-01 4.61824328e-01 -1.73421824e+00 5.85673571e-01
3.40849966e-01 8.95176828e-01 5.57484627e-01 9.60365534e-01
4.51383382e-01 9.96473610e-01 1.99532509e-01 -6.66714758e-02
2.72529215e-01 5.57410978e-02 -1.11150038e+00 -1.25328973e-01
2.84271121e-01 -4.69664305e-01 -7.54551530e-01 6.14598811e-01
7.39902079e-01 5.91994226e-01 3.58261347e-01 -1.67004573e+00
-2.18420908e-01 5.75243115e-01 -2.30935156e-01 3.06494534e-01
9.61655498e-01 6.37514532e-01 7.09356129e-01 -1.05999124e+00
1.19619787e+00 1.32718408e+00 3.05552423e-01 5.90763330e-01
-8.20645452e-01 -2.42515936e-01 5.34479558e-01 8.64053309e-01
-1.32134521e+00 -8.68706882e-01 7.22871482e-01 -7.77822554e-01
1.43188512e+00 5.13160676e-02 7.39223421e-01 1.34985864e+00
1.98620647e-01 1.46923733e+00 5.62058449e-01 -1.05092451e-01
2.44920954e-01 -2.49384880e-01 3.40758145e-01 3.43612880e-01
-3.32346261e-01 5.82015961e-02 -8.02162170e-01 1.82269156e-01
2.38595054e-01 2.64261328e-02 -2.80611794e-02 -7.95425355e-01
-1.84175575e+00 1.47887290e-01 4.04372752e-01 2.75317550e-01
-1.04516840e+00 5.65366447e-01 5.97860396e-01 -8.94302726e-02
6.17093861e-01 5.08498400e-02 -5.56520522e-01 -7.74146378e-01
-3.98074001e-01 7.22063601e-01 6.11224055e-01 1.25243580e+00
3.76006335e-01 -6.01036370e-01 -5.45875967e-01 1.17766708e-01
2.47423694e-01 1.54669657e-01 9.45531949e-02 -1.19467044e+00
7.13923037e-01 3.94500971e-01 6.90643311e-01 -7.67427742e-01
-8.74476969e-01 -6.88449442e-02 -2.77240217e-01 8.45822599e-03
4.07533348e-01 -2.25406155e-01 -5.22172093e-01 1.64444685e+00
6.78787410e-01 3.24536175e-01 2.59626135e-02 1.21133792e+00
7.57168949e-01 4.10191566e-01 3.12499255e-01 6.87095225e-02
1.41882384e+00 -1.35315859e+00 -6.89630806e-01 -4.15529400e-01
9.88092959e-01 -4.97599125e-01 7.53610253e-01 3.43873292e-01
-1.34067535e+00 -8.45596969e-01 -4.58118439e-01 -1.36656523e-01
-2.99582511e-01 3.46361130e-01 8.94186318e-01 -3.18925828e-02
-1.05092096e+00 6.07661784e-01 -1.07469058e+00 -8.05845082e-01
3.30802709e-01 3.26958120e-01 -6.20566368e-01 -1.34602021e-02
-9.21575665e-01 8.96462858e-01 5.06772876e-01 3.65183860e-01
-1.18470407e+00 -5.17556965e-01 -6.71452463e-01 7.96347484e-02
4.72661853e-01 -9.43983376e-01 1.50536919e+00 -6.83438838e-01
-1.33966196e+00 9.76551473e-01 -3.50307614e-01 -6.44834697e-01
8.86767447e-01 -8.59937072e-01 -1.90935090e-01 -3.84037457e-02
3.67454253e-02 9.64107513e-01 4.41235840e-01 -5.51455140e-01
-6.69381618e-01 -6.82471991e-01 3.05082858e-01 5.93855858e-01
4.72464934e-02 1.74425974e-01 -4.92950857e-01 -5.55339158e-02
2.05142628e-02 -1.53374243e+00 -2.23733053e-01 -2.83666044e-01
-2.71081597e-01 -9.30879653e-01 7.63185680e-01 -3.48226428e-01
9.00666833e-01 -2.22483134e+00 6.06180429e-01 -4.00127977e-01
1.23168752e-01 5.83791547e-03 -2.57104516e-01 6.35580778e-01
-4.65246320e-01 -4.94172037e-01 4.95780021e-01 -7.43196666e-01
4.35948342e-01 -3.49103481e-01 -1.93516910e-01 7.11298823e-01
-3.71385604e-01 1.19988775e+00 -1.31720853e+00 -3.32349747e-01
7.23670065e-01 4.46088910e-01 -4.53072339e-01 4.14630085e-01
-3.43411922e-01 7.84036696e-01 -5.88489652e-01 4.72626865e-01
5.61601460e-01 -6.46008924e-02 -4.82208431e-02 -1.62616968e-01
-4.22280014e-01 1.95932791e-01 -1.05131805e+00 2.35535955e+00
-3.53832126e-01 7.84978330e-01 -2.21100569e-01 -5.30144155e-01
4.51929003e-01 4.49250638e-01 8.41931462e-01 -4.82980072e-01
1.08414613e-01 -3.10254216e-01 -1.94662422e-01 -1.06576729e+00
6.44239843e-01 5.25268972e-01 -2.35486299e-01 3.05290818e-01
-3.57308425e-02 2.78785020e-01 2.62702256e-01 3.17768633e-01
1.20390654e+00 8.24078321e-01 2.11618766e-01 -1.94266468e-01
4.44895804e-01 1.06403515e-01 4.18334633e-01 6.65062129e-01
-7.87159383e-01 4.40967649e-01 3.95153850e-01 -7.80238628e-01
-6.18590891e-01 -1.02201760e+00 4.36013490e-01 1.31484151e+00
2.93327987e-01 -6.49545789e-01 -8.19268882e-01 -8.15291464e-01
-1.15289964e-01 1.18387866e+00 -7.55317867e-01 1.14614032e-01
-4.99375701e-01 -9.09765139e-02 -3.63099389e-02 6.76748514e-01
4.08730119e-01 -1.59555805e+00 -1.03597593e+00 3.08624078e-02
-7.35547185e-01 -1.23196411e+00 -6.08696461e-01 -2.71326095e-01
-5.75746834e-01 -1.04702342e+00 -6.82446241e-01 -4.01182145e-01
3.76575887e-01 3.76926273e-01 1.17575026e+00 -5.69038451e-01
-2.91672915e-01 8.81701529e-01 -3.24527681e-01 -5.73350072e-01
2.91155994e-01 1.39099553e-01 4.64419335e-01 8.24768245e-02
6.72158182e-01 -5.20885944e-01 -1.13994169e+00 2.85662919e-01
-7.57830068e-02 2.44650826e-01 2.64588416e-01 5.15347794e-02
2.30190068e-01 -4.62487966e-01 3.05878431e-01 -4.68679398e-01
3.02847505e-01 -5.13289988e-01 -5.09520113e-01 6.72482252e-02
-2.49474674e-01 -1.98405847e-01 -5.92519380e-02 -1.98007852e-01
-1.20612705e+00 7.17239559e-01 8.88333544e-02 -7.54967570e-01
-6.28515542e-01 7.18418956e-02 -1.83025599e-01 4.79127020e-01
4.14512962e-01 -4.13386122e-04 -4.57306921e-01 -3.71321321e-01
5.36673307e-01 3.71637344e-01 4.47306275e-01 -1.78274110e-01
2.34922975e-01 6.64238870e-01 -7.29924664e-02 -5.13006330e-01
-6.55895948e-01 -9.00293410e-01 -1.13339245e+00 -8.19794893e-01
1.21281981e+00 -1.19495130e+00 -1.51931047e+00 5.48103154e-01
-1.57961595e+00 -5.85920751e-01 -2.79713809e-01 8.27861071e-01
-1.17728961e+00 8.67189467e-02 -1.41960308e-01 -9.57575262e-01
-2.13068575e-01 -9.85885143e-01 1.25548673e+00 7.65398517e-02
-6.87530339e-01 -7.97472775e-01 4.23736334e-01 4.33488816e-01
1.62158564e-01 2.43988723e-01 -2.15619113e-02 -4.85361338e-01
-9.57088232e-01 -3.78432810e-01 4.67242263e-02 -2.62423545e-01
-2.97136724e-01 -3.38276416e-01 -8.82886171e-01 -1.69870108e-01
-2.26812735e-01 -1.31577417e-01 7.69898593e-01 6.86829567e-01
1.10435140e+00 -3.23561281e-02 -7.70990908e-01 3.78688246e-01
9.68660533e-01 1.12221269e-02 8.64980400e-01 2.30645239e-01
6.69585645e-01 1.10123575e+00 1.41508842e+00 8.96428287e-01
6.45195484e-01 1.03774774e+00 7.25298107e-01 5.41332304e-01
2.48255376e-02 -1.72691062e-01 6.12845182e-01 1.25169978e-01
-4.50207084e-01 -6.94164872e-01 -1.03422487e+00 9.40305889e-01
-2.54976726e+00 -1.55270517e+00 -2.95396596e-01 1.86828578e+00
-1.84394673e-01 -1.79024041e-02 4.80872869e-01 -4.51714605e-01
6.02956057e-01 3.62805575e-01 -6.47250593e-01 -2.21889317e-01
4.26709354e-01 -5.51556110e-01 1.99578434e-01 4.62154299e-01
-1.27728164e+00 1.25631511e+00 5.48105335e+00 3.73002917e-01
-6.58893526e-01 2.95475423e-01 2.24279419e-01 -5.65831065e-01
4.77562308e-01 1.35555327e-01 -8.15578163e-01 3.98060501e-01
8.84632111e-01 -2.31321707e-01 4.05056328e-01 1.15332913e+00
6.85015082e-01 -4.14889306e-01 -1.52444637e+00 1.24391651e+00
-7.55324811e-02 -1.07448304e+00 -4.98241901e-01 -2.80477703e-02
4.22264904e-01 5.91891527e-01 -3.01932637e-02 2.68066615e-01
6.31768852e-02 -6.35630667e-01 8.56530845e-01 1.13456368e+00
1.51749060e-01 -5.21631777e-01 4.98335570e-01 4.96054709e-01
-1.38149047e+00 -1.83280438e-01 7.44706020e-02 -4.40241635e-01
6.16657853e-01 3.09025586e-01 -1.02732205e+00 3.77918154e-01
8.60180974e-01 1.08588076e+00 -4.53935772e-01 9.71068025e-01
-2.56727308e-01 3.27712707e-02 -1.27751321e-01 -8.45462084e-02
2.61863112e-01 -1.29457906e-01 1.08118260e+00 1.27609932e+00
2.42476687e-01 3.38069052e-01 1.26386061e-01 5.34736454e-01
2.37253845e-01 -3.34978312e-01 -8.94807518e-01 3.41166295e-02
-7.42154866e-02 1.24682248e+00 -6.38165414e-01 -4.79540288e-01
3.06516048e-02 1.27919447e+00 2.39296526e-01 3.52170914e-01
-1.05912876e+00 -3.16566020e-01 9.10432756e-01 7.43457749e-02
3.04741442e-01 -3.54386359e-01 1.39961913e-01 -1.12466311e+00
1.60993665e-01 -1.17581077e-01 3.91183257e-01 -1.37905753e+00
-5.98962903e-01 5.41415751e-01 3.11864525e-01 -1.45635343e+00
-6.62081301e-01 -4.45153534e-01 -6.61165357e-01 5.74994624e-01
-8.23493004e-01 -1.26159704e+00 -8.41925740e-01 6.46092772e-01
8.93333733e-01 6.34028018e-02 5.63597143e-01 1.36160851e-01
-4.12648708e-01 4.72332872e-02 -1.67858288e-01 -2.61631995e-01
7.80750573e-01 -1.12569261e+00 7.92978704e-01 6.77216887e-01
-5.04848994e-02 2.90992439e-01 1.13387108e+00 -7.47112751e-01
-1.56134248e+00 -9.28390503e-01 1.39983153e+00 -1.07573056e+00
4.50699598e-01 -6.71361864e-01 -1.12422191e-01 1.30231965e+00
4.24432993e-01 1.00152358e-01 1.65115505e-01 3.15460563e-01
1.88064232e-01 1.23473652e-01 -9.63748515e-01 9.05055285e-01
1.92969012e+00 -3.82694811e-01 -4.40076470e-01 1.01575959e+00
4.95735019e-01 -5.37170529e-01 -4.43730205e-01 2.57590383e-01
8.45763505e-01 -1.17639041e+00 1.08423042e+00 -6.68045521e-01
3.45191717e-01 -2.30763614e-01 -8.33844095e-02 -1.06098115e+00
-6.63935959e-01 -8.60765040e-01 -4.64728445e-01 1.02619791e+00
-1.33988217e-01 -3.57690573e-01 1.05911720e+00 7.93495357e-01
-3.10838938e-01 -4.12738323e-01 -7.67561138e-01 -6.95669949e-01
-8.47743273e-01 -5.63491344e-01 3.51289630e-01 3.75569284e-01
3.50459903e-01 2.64963597e-01 -5.80837846e-01 9.44940075e-02
6.47125065e-01 2.04554215e-01 1.32998383e+00 -9.88846958e-01
1.15831010e-01 -8.18765536e-02 -7.15417266e-01 -1.21072018e+00
4.79255885e-01 -8.71758878e-01 1.24420062e-01 -1.60749280e+00
1.86004028e-01 -8.24253168e-03 -2.30976209e-01 2.07594290e-01
2.02366546e-01 -1.81481928e-01 5.22026300e-01 1.41819164e-01
-1.43574643e+00 6.35538042e-01 9.79188800e-01 1.08462926e-02
-4.02649164e-01 3.52999181e-01 -1.54518858e-01 8.36788893e-01
7.78820455e-01 -4.02881712e-01 -4.43767995e-01 -5.77412486e-01
-3.80813740e-02 1.73633531e-01 6.59981430e-01 -1.36813700e+00
6.33455873e-01 -1.06236354e-01 2.47984827e-01 -1.24281788e+00
9.37890828e-01 -8.10003221e-01 3.86335015e-01 6.22099996e-01
-4.62798417e-01 2.50983328e-01 1.86034426e-01 6.06578767e-01
1.62068367e-01 1.10741928e-02 1.26820132e-01 -2.27751613e-01
-1.32864368e+00 5.17246246e-01 -4.50804204e-01 -1.90690875e-01
1.57686496e+00 -1.17459968e-01 -5.19623645e-02 -4.39546883e-01
-1.12431204e+00 6.33179784e-01 2.86131322e-01 9.15312409e-01
5.56540370e-01 -1.13854992e+00 -7.59175718e-01 -2.83893049e-01
4.03168052e-01 -2.72594541e-01 9.05106902e-01 1.14818156e+00
-2.45331183e-01 9.40614879e-01 -3.83312792e-01 -8.92852485e-01
-1.60167396e+00 9.29923058e-01 2.32185349e-02 -2.82607198e-01
-6.62943304e-01 8.39381099e-01 6.41790807e-01 -4.01163787e-01
3.36479336e-01 -1.17230862e-01 -2.53596306e-01 2.19747461e-02
5.69468498e-01 8.41458917e-01 -2.69916117e-01 -7.73960054e-01
-8.96456480e-01 4.72467065e-01 2.46857833e-02 -6.95995912e-02
1.48800790e+00 -5.10499120e-01 9.33329985e-02 5.58174670e-01
1.10960984e+00 -4.93113577e-01 -1.48879147e+00 2.15311814e-02
1.22050568e-01 -6.59896910e-01 -6.85818121e-02 -6.81595147e-01
-6.11619413e-01 9.56200123e-01 8.96115541e-01 -1.67076483e-01
9.43226099e-01 2.50542372e-01 5.03663719e-01 6.42076612e-01
6.35343254e-01 -1.14689291e+00 1.29110217e-01 6.21776164e-01
1.27923119e+00 -1.27374256e+00 -4.48021479e-02 -1.72680631e-01
-1.07929862e+00 7.53355205e-01 8.32185805e-01 -2.64706910e-01
5.67842662e-01 -2.84157306e-01 -3.40122253e-01 -3.95222276e-01
-1.13851738e+00 -3.20481718e-01 1.82186067e-01 5.51225364e-01
3.02423030e-01 1.93601936e-01 -1.19794317e-01 3.02316904e-01
8.29183012e-02 4.90106225e-01 1.61501154e-01 8.30624700e-01
2.78826747e-02 -5.67224681e-01 -8.21962655e-02 2.95399502e-02
-2.35284269e-02 2.81993449e-01 -3.86997968e-01 4.83559549e-01
1.20690972e-01 8.69948685e-01 3.51777822e-01 -4.21488822e-01
6.56600237e-01 2.12391466e-01 6.28375769e-01 -4.20973986e-01
-6.66599512e-01 -3.51172149e-01 2.50152022e-01 -1.37778056e+00
-6.06347740e-01 -1.22926950e+00 -1.06202328e+00 -4.22269076e-01
9.97354612e-02 -5.10309562e-02 8.17764044e-01 7.13183284e-01
1.00283086e+00 5.16226709e-01 3.70397687e-01 -1.86570418e+00
-1.77403182e-01 -1.00190055e+00 -1.62803978e-01 4.99398082e-01
2.60806501e-01 -7.09292054e-01 -1.19674355e-01 -2.90963911e-02] | [8.146052360534668, 0.48572850227355957] |
be5e172e-5aca-41ab-bc8f-1e22d99c658a | multi-body-se-3-equivariance-for-unsupervised | 2306.05584 | null | https://arxiv.org/abs/2306.05584v1 | https://arxiv.org/pdf/2306.05584v1.pdf | Multi-body SE(3) Equivariance for Unsupervised Rigid Segmentation and Motion Estimation | A truly generalizable approach to rigid segmentation and motion estimation is fundamental to 3D understanding of articulated objects and moving scenes. In view of the tightly coupled relationship between segmentation and motion estimates, we present an SE(3) equivariant architecture and a training strategy to tackle this task in an unsupervised manner. Our architecture comprises two lightweight and inter-connected heads that predict segmentation masks using point-level invariant features and motion estimates from SE(3) equivariant features without the prerequisites of category information. Our unified training strategy can be performed online while jointly optimizing the two predictions by exploiting the interrelations among scene flow, segmentation mask, and rigid transformations. We show experiments on four datasets as evidence of the superiority of our method both in terms of model performance and computational efficiency with only 0.25M parameters and 0.92G FLOPs. To the best of our knowledge, this is the first work designed for category-agnostic part-level SE(3) equivariance in dynamic point clouds. | ['Niki Trigoni', 'Andrew Markham', 'Kaichen Zhou', 'Kai Lu', 'Yuhang He', 'Ta-Ying Cheng', 'Jia-Xing Zhong'] | 2023-06-08 | null | null | null | null | ['motion-estimation'] | ['computer-vision'] | [ 7.19882771e-02 -1.19566303e-02 -5.46694286e-02 -3.71142805e-01
-7.03535318e-01 -8.16173434e-01 6.90881491e-01 -3.44315618e-01
-4.38176155e-01 2.42071211e-01 -9.79898125e-02 -1.70532987e-01
-6.59397021e-02 -4.05872762e-01 -8.62079680e-01 -4.88409430e-01
-6.10668026e-02 9.61024344e-01 7.23643184e-01 -1.23855494e-01
2.46571884e-01 1.04079962e+00 -1.42078030e+00 -1.80800617e-01
6.75317585e-01 6.86941564e-01 1.88488439e-01 1.06056738e+00
-9.13130865e-02 3.32566887e-01 -9.01709776e-03 -1.42501116e-01
5.19271076e-01 -1.76826254e-01 -1.14063227e+00 6.13887429e-01
9.06947196e-01 -1.04367621e-01 -2.90774494e-01 6.57567084e-01
1.73572913e-01 4.17972982e-01 6.12567663e-01 -8.87622774e-01
1.61606878e-01 1.28023595e-01 -5.39986253e-01 3.52479309e-01
1.30371124e-01 4.06274766e-01 9.22685683e-01 -1.04347420e+00
9.15273666e-01 1.17920589e+00 7.01092541e-01 4.56575990e-01
-1.43295586e+00 -1.06447019e-01 3.90883267e-01 -1.73047036e-01
-1.23084760e+00 -5.79065561e-01 7.69646466e-01 -8.39851975e-01
9.33863640e-01 2.22442627e-01 7.35145032e-01 5.75062871e-01
1.48485303e-01 6.16165400e-01 6.01723313e-01 -9.18770805e-02
2.23312601e-01 -1.38440296e-01 2.16701359e-01 9.60866213e-01
1.66331470e-01 -1.11690126e-01 -3.73415500e-01 1.30287439e-01
1.08184123e+00 -2.85909384e-01 -4.21115058e-03 -9.13130879e-01
-1.41786063e+00 6.27861083e-01 2.49333620e-01 5.32142296e-02
-1.32721245e-01 5.58129191e-01 1.79350957e-01 -7.80179501e-02
5.06425142e-01 1.59771979e-01 -7.19399333e-01 -1.13489829e-01
-1.08153260e+00 2.39794463e-01 8.50605607e-01 1.00597262e+00
9.36362088e-01 2.95334160e-01 1.89606085e-01 3.86092156e-01
3.78458202e-01 6.05824471e-01 2.39061907e-01 -1.49275815e+00
2.75519639e-01 4.35744107e-01 1.07956916e-01 -8.67934227e-01
-6.69379234e-01 -4.49232876e-01 -3.41688126e-01 1.10698730e-01
6.56737685e-01 -1.74473431e-02 -1.12366617e+00 1.57235873e+00
7.75271356e-01 4.03279394e-01 -2.72224247e-01 1.03405023e+00
5.10275424e-01 3.71448785e-01 -1.42906234e-02 1.11490957e-01
1.05561817e+00 -9.30414617e-01 -8.27134550e-02 -2.47416764e-01
6.14455044e-01 -8.11619163e-01 9.17915702e-01 -7.53990412e-02
-1.35457182e+00 -5.09923398e-01 -6.04088068e-01 -2.32362017e-01
1.25715453e-02 -1.24675706e-02 8.30098212e-01 4.72724676e-01
-1.05391717e+00 7.14633286e-01 -1.43783414e+00 -3.72337043e-01
3.82894635e-01 6.42175734e-01 -3.71951193e-01 3.59730899e-01
-4.01408434e-01 6.67158961e-01 2.10932747e-01 1.34426683e-01
-6.86508954e-01 -8.50355268e-01 -8.97291005e-01 -1.59886107e-01
3.48945141e-01 -1.03967559e+00 1.23478436e+00 -6.68094873e-01
-1.66602516e+00 9.73533154e-01 -3.98130745e-01 -4.25908029e-01
9.35817897e-01 -2.48461276e-01 3.09701085e-01 2.67975897e-01
1.27262741e-01 8.70582938e-01 7.69721150e-01 -1.12794757e+00
-5.96105516e-01 -4.53128815e-01 -1.18972428e-01 3.49779338e-01
1.96480945e-01 -2.91766405e-01 -7.40252376e-01 -5.18297315e-01
6.57530248e-01 -1.30223978e+00 -4.89780754e-01 2.62135655e-01
-1.54063269e-01 1.85656212e-02 7.42022395e-01 -5.29319584e-01
5.93170762e-01 -1.81327105e+00 5.50427556e-01 2.66896993e-01
1.26114115e-01 5.63559383e-02 7.24860933e-04 -1.40459046e-01
2.23331541e-01 4.14233170e-02 -5.78427255e-01 -3.78961772e-01
-3.47832367e-02 2.80091345e-01 -1.13407202e-01 9.64319646e-01
2.93085694e-01 1.00637305e+00 -8.31555367e-01 -6.66159272e-01
6.36565924e-01 3.75597745e-01 -8.61172378e-01 9.71811190e-02
-3.64803761e-01 8.81278992e-01 -5.37419736e-01 4.77525741e-01
5.29690444e-01 -2.51981974e-01 9.38224420e-02 -2.86493182e-01
-3.34394395e-01 2.30157271e-01 -1.36631608e+00 2.05634594e+00
-3.35333645e-01 6.30649447e-01 2.29542643e-01 -1.01601541e+00
5.58601797e-01 1.84266120e-01 1.00575805e+00 -8.79017487e-02
1.40759706e-01 1.75116271e-01 -1.09670743e-01 -3.02203774e-01
4.39944744e-01 -6.26754090e-02 -3.19182761e-02 1.99676424e-01
2.61905611e-01 -6.46930754e-01 2.09282860e-01 1.24022834e-01
8.71319294e-01 6.26399100e-01 1.00058511e-01 -4.74319160e-01
6.19872570e-01 2.33957767e-01 5.50811172e-01 5.59410751e-01
-1.11366950e-01 7.16451883e-01 1.26999393e-01 -6.20650351e-01
-1.15794861e+00 -1.16468418e+00 -2.29618117e-01 8.57535183e-01
3.24460626e-01 -2.13731900e-01 -6.33958519e-01 -5.12993515e-01
7.99348652e-02 3.30455750e-01 -3.63958806e-01 2.77821660e-01
-1.10381293e+00 -4.81088012e-01 2.04153612e-01 6.36393845e-01
2.51196086e-01 -6.83063745e-01 -8.79945219e-01 2.62337834e-01
-5.77226281e-02 -1.43328273e+00 -6.81259632e-01 7.98525196e-03
-1.27618015e+00 -1.09097421e+00 -4.62287575e-01 -6.00654960e-01
6.38820052e-01 3.26077729e-01 1.12917924e+00 -6.55512437e-02
-3.04678351e-01 8.13248992e-01 9.73203406e-02 -8.10152367e-02
-1.64559245e-01 2.94205099e-01 -1.33488365e-02 -7.24853128e-02
-1.76561981e-01 -6.66486919e-01 -6.68879271e-01 4.76341933e-01
-6.03375137e-01 1.51799530e-01 2.55432934e-01 4.21241194e-01
9.38850999e-01 -5.35713494e-01 -2.71107584e-01 -8.50705981e-01
-4.01525557e-01 -2.59061426e-01 -9.72696722e-01 -3.62196490e-02
-2.62737393e-01 3.02299887e-01 2.70382047e-01 -2.32047826e-01
-1.07410026e+00 7.87473857e-01 -5.56082390e-02 -5.71333230e-01
-3.31738681e-01 -1.27297230e-02 -8.47813040e-02 -4.74295735e-01
3.68341148e-01 6.41847104e-02 -1.84007883e-01 -5.16175508e-01
7.42793620e-01 -3.54937539e-02 8.74154210e-01 -7.03405201e-01
1.06385446e+00 1.02225602e+00 3.87885451e-01 -1.04016292e+00
-6.89096153e-01 -7.98053980e-01 -1.27138388e+00 -3.35655898e-01
1.10744095e+00 -8.75459790e-01 -6.92974985e-01 3.19809496e-01
-1.16013038e+00 -4.42628324e-01 -4.91168708e-01 6.20135605e-01
-9.52723145e-01 5.31850696e-01 -5.66679120e-01 -6.41258717e-01
-7.76679069e-02 -1.16938674e+00 1.24949515e+00 -1.33574232e-02
-3.12831581e-01 -1.13775814e+00 2.94759162e-02 5.43527424e-01
9.72167477e-02 4.45813924e-01 3.67731333e-01 -4.52723324e-01
-1.06895292e+00 -1.98472179e-02 4.83905859e-02 -7.47698992e-02
-3.27348411e-02 2.69097388e-01 -8.39941323e-01 -2.32768029e-01
-2.62083579e-02 8.25926363e-02 9.07090008e-01 5.96349180e-01
1.06936443e+00 -1.01671293e-01 -2.45914817e-01 1.07252693e+00
1.28847313e+00 -1.84491560e-01 2.25345820e-01 3.00470948e-01
1.18366945e+00 5.42118132e-01 5.34128487e-01 4.13118124e-01
4.64713007e-01 9.05586183e-01 4.38935727e-01 -1.46898091e-01
-4.16015804e-01 -9.60918143e-03 9.95733067e-02 9.42435265e-01
-3.12509328e-01 -4.49977927e-02 -1.13161492e+00 6.94886446e-01
-1.82693696e+00 -8.88008177e-01 -4.84970778e-01 2.06477237e+00
4.97759521e-01 1.48823991e-01 1.12533316e-01 -2.96270311e-01
5.70432365e-01 1.81397095e-01 -5.85288346e-01 -2.19928905e-01
1.18179709e-01 2.25440115e-01 6.80721343e-01 7.96608508e-01
-1.38687086e+00 1.32521772e+00 5.97761059e+00 4.29773301e-01
-9.72841799e-01 -4.20695394e-02 3.03600460e-01 -1.26509756e-01
-3.27827632e-01 3.14543396e-01 -8.37924480e-01 2.51538813e-01
8.32199097e-01 4.74596322e-02 2.72439897e-01 7.76105344e-01
2.58968532e-01 2.39083633e-01 -1.13278377e+00 8.10708404e-01
-1.23325838e-02 -1.40797770e+00 -4.92769759e-03 1.89984798e-01
7.52318978e-01 3.75043988e-01 -1.12062760e-01 -2.16512188e-01
4.20616269e-01 -6.05152428e-01 1.06570029e+00 6.34462237e-01
4.00438964e-01 -4.58329767e-01 3.86469126e-01 2.85314590e-01
-1.48267162e+00 3.25135320e-01 -1.33305326e-01 1.16277777e-01
4.76168454e-01 2.25762948e-01 -6.48077786e-01 5.40140450e-01
5.75560629e-01 8.86454046e-01 -3.47204030e-01 9.52962458e-01
1.08683802e-01 5.38145185e-01 -6.68575048e-01 6.54614985e-01
3.49626213e-01 -4.15817171e-01 9.61641431e-01 1.31286919e+00
1.86121181e-01 2.14378998e-01 3.89385790e-01 7.05383003e-01
6.17068633e-02 -1.00739755e-01 -4.21344668e-01 3.91470164e-01
1.85106263e-01 1.19353569e+00 -1.15393841e+00 -3.19019467e-01
-2.84352988e-01 8.79909992e-01 2.63229936e-01 1.34177953e-01
-8.71353745e-01 3.36282611e-01 8.18162382e-01 3.06483895e-01
7.10494995e-01 -8.57171595e-01 -4.93406236e-01 -1.40578830e+00
5.23278071e-03 -2.83379406e-01 2.74894744e-01 -3.99400800e-01
-9.10891891e-01 2.65519321e-01 1.65730283e-01 -1.21019924e+00
-2.56921470e-01 -5.76149166e-01 -4.19302464e-01 4.03020889e-01
-1.42431426e+00 -1.15434361e+00 -2.55754292e-01 5.60040295e-01
8.97687733e-01 2.91961819e-01 4.58442718e-01 -4.32822779e-02
-5.65568984e-01 1.08213894e-01 7.76806846e-02 2.53226664e-02
3.21788192e-01 -1.39719892e+00 6.54162288e-01 1.02880216e+00
4.29910004e-01 4.09557790e-01 6.68533087e-01 -4.88648742e-01
-1.47274995e+00 -1.13251972e+00 6.26593292e-01 -9.68907952e-01
6.60277247e-01 -2.69625455e-01 -9.37744915e-01 8.27254593e-01
-3.97807509e-01 3.54419470e-01 3.79779994e-01 -1.76722199e-01
-9.94435474e-02 -1.19415820e-02 -8.30654323e-01 5.46891987e-01
1.45466232e+00 -1.91582128e-01 -5.43462753e-01 3.16107005e-01
6.60665512e-01 -9.25708413e-01 -1.07617986e+00 5.04369140e-01
5.47691405e-01 -7.50390232e-01 1.22472143e+00 -5.14202535e-01
9.38966274e-02 -5.22627234e-01 -2.58978754e-01 -7.14351237e-01
-2.98534274e-01 -7.01606989e-01 -1.86850935e-01 8.82434666e-01
2.63380587e-01 -2.81604886e-01 1.05040073e+00 8.57527554e-01
-5.07727206e-01 -4.08545554e-01 -1.00501513e+00 -6.60771906e-01
8.55541155e-02 -5.06708741e-01 2.55976200e-01 7.81392634e-01
-6.78661704e-01 2.62517482e-01 -1.43843606e-01 3.31424028e-01
6.65190041e-01 4.60611165e-01 1.14472330e+00 -1.33402956e+00
-1.81872681e-01 -4.32521135e-01 -6.67637467e-01 -1.33176517e+00
3.49249810e-01 -9.13200438e-01 1.75073668e-01 -1.26425707e+00
-5.90797290e-02 -4.19807911e-01 1.88822150e-02 3.77135247e-01
-6.52222484e-02 2.69039571e-01 2.73945540e-01 4.58186835e-01
-7.43891716e-01 4.45537001e-01 1.17161775e+00 2.91305501e-02
-4.22684312e-01 8.90845433e-03 -6.56121969e-02 1.16196656e+00
7.56695807e-01 -3.45282972e-01 -4.45716858e-01 -7.33209908e-01
-3.26312304e-01 2.21593268e-02 5.60335279e-01 -1.01256347e+00
2.65412182e-01 -4.43522722e-01 2.66785681e-01 -7.08733499e-01
5.37561297e-01 -8.35095823e-01 3.06248844e-01 3.80908817e-01
-9.26045403e-02 8.36191550e-02 2.17997074e-01 7.29126632e-01
1.14813514e-01 1.89557806e-01 8.54217350e-01 -3.33170742e-01
-1.04875493e+00 6.09659016e-01 -1.85881093e-01 1.68888316e-01
1.04196608e+00 -5.45077145e-01 1.30948395e-01 -3.78908291e-02
-8.98016214e-01 1.21700265e-01 7.64174759e-01 1.34862348e-01
4.85459477e-01 -8.18233371e-01 -5.96614540e-01 1.52180195e-01
-2.01617479e-01 4.98894423e-01 1.94878981e-01 8.88322711e-01
-8.77956033e-01 5.35375774e-01 -7.40158185e-02 -1.18760777e+00
-1.06486475e+00 1.27310425e-01 4.87430751e-01 -1.13458812e-01
-8.54715228e-01 8.22162390e-01 3.57990086e-01 -5.79494238e-01
-1.25229120e-01 -3.31282169e-01 1.21714622e-01 -2.87083775e-01
-1.15912989e-01 6.25421882e-01 3.96019816e-02 -1.14912879e+00
-6.34531617e-01 1.21352518e+00 1.09300047e-01 -3.06626767e-01
1.30886042e+00 -2.68009156e-01 1.22505659e-03 4.20705765e-01
1.10082626e+00 -7.18510821e-02 -1.74631345e+00 -4.28701844e-03
6.76438361e-02 -5.19661605e-01 -4.29574959e-02 -1.30038336e-01
-1.18803644e+00 7.97303736e-01 3.81504327e-01 -2.16920629e-01
6.43843830e-01 2.98660159e-01 7.51367688e-01 2.09172100e-01
2.91220576e-01 -8.68765175e-01 -1.64908413e-02 4.86945301e-01
5.58927000e-01 -1.07269406e+00 1.36583656e-01 -7.04870403e-01
-3.53201509e-01 1.06333852e+00 6.16770804e-01 -4.25599396e-01
5.59500039e-01 4.85281879e-03 -2.50105392e-02 -3.20284575e-01
-5.93248963e-01 -2.95711190e-01 5.31997263e-01 4.54279244e-01
3.88023525e-01 -5.91518208e-02 -2.18860924e-01 -2.12920487e-01
-3.34959269e-01 -3.63164455e-01 2.76869595e-01 8.05327773e-01
-4.05244201e-01 -1.02506447e+00 -2.00144023e-01 1.20309509e-01
-3.23305875e-01 2.33657122e-01 -1.55456319e-01 9.77451146e-01
-3.27653848e-02 6.02268159e-01 3.29106063e-01 -7.39577636e-02
4.21038449e-01 9.53290705e-03 5.74107468e-01 -5.62696457e-01
-3.21710646e-01 2.21009657e-01 -1.12488724e-01 -9.06651735e-01
-7.88756728e-01 -8.92376721e-01 -1.49361503e+00 -2.92647630e-01
-1.84343472e-01 -1.32512540e-01 6.51699185e-01 9.24437821e-01
4.46130276e-01 4.66940403e-01 5.37767887e-01 -1.34110081e+00
-3.37873757e-01 -3.03105861e-01 -2.88044840e-01 5.33383906e-01
2.71896392e-01 -6.29969716e-01 -3.23769063e-01 4.49052840e-01] | [8.427116394042969, -2.0043795108795166] |
17879dab-85db-4dc0-af4f-dd1cc63b4b99 | quality-matters-embracing-quality-clues-for | 2208.10976 | null | https://arxiv.org/abs/2208.10976v1 | https://arxiv.org/pdf/2208.10976v1.pdf | Quality Matters: Embracing Quality Clues for Robust 3D Multi-Object Tracking | 3D Multi-Object Tracking (MOT) has achieved tremendous achievement thanks to the rapid development of 3D object detection and 2D MOT. Recent advanced works generally employ a series of object attributes, e.g., position, size, velocity, and appearance, to provide the clues for the association in 3D MOT. However, these cues may not be reliable due to some visual noise, such as occlusion and blur, leading to tracking performance bottleneck. To reveal the dilemma, we conduct extensive empirical analysis to expose the key bottleneck of each clue and how they correlate with each other. The analysis results motivate us to efficiently absorb the merits among all cues, and adaptively produce an optimal tacking manner. Specifically, we present Location and Velocity Quality Learning, which efficiently guides the network to estimate the quality of predicted object attributes. Based on these quality estimations, we propose a quality-aware object association (QOA) strategy to leverage the quality score as an important reference factor for achieving robust association. Despite its simplicity, extensive experiments indicate that the proposed strategy significantly boosts tracking performance by 2.2% AMOTA and our method outperforms all existing state-of-the-art works on nuScenes by a large margin. Moreover, QTrack achieves 48.0% and 51.1% AMOTA tracking performance on the nuScenes validation and test sets, which significantly reduces the performance gap between pure camera and LiDAR based trackers. | ['Wenbing Tao', 'Xiaoping Li', 'Zeming Li', 'En Yu', 'Jinrong Yang'] | 2022-08-23 | null | null | null | null | ['3d-multi-object-tracking'] | ['computer-vision'] | [-2.44550049e-01 -7.82136083e-01 -5.45360565e-01 1.00662090e-01
-6.08953118e-01 -6.53695047e-01 3.71357769e-01 -1.37481792e-02
-2.82194048e-01 4.48336840e-01 -1.72649160e-01 9.93766170e-03
-2.72062838e-01 -5.24692178e-01 -5.56411982e-01 -8.14388156e-01
1.31827770e-02 2.78838336e-01 6.99674189e-01 1.31629556e-01
1.53533354e-01 5.46560705e-01 -1.75642419e+00 -4.61450070e-01
8.92196000e-01 1.42800224e+00 3.34677815e-01 5.53674459e-01
-1.62276477e-02 3.81552994e-01 -6.44833267e-01 -1.60630897e-01
3.35487157e-01 2.87879109e-02 1.62543535e-01 1.60936862e-02
6.65962100e-01 -5.58784187e-01 -2.90664732e-01 1.06774426e+00
6.25845909e-01 -1.23285353e-01 6.48221135e-01 -1.55566549e+00
-5.37380755e-01 1.19392477e-01 -9.54066336e-01 4.30279881e-01
1.63940370e-01 4.92649257e-01 9.77067888e-01 -7.92249203e-01
1.60272241e-01 1.33926797e+00 7.19360948e-01 3.89305979e-01
-9.71120119e-01 -1.11162329e+00 3.81963819e-01 1.54621318e-01
-1.39440966e+00 -2.61121750e-01 7.15859652e-01 -4.91996527e-01
3.73839110e-01 2.22349346e-01 8.39342475e-01 8.26604068e-01
2.25452691e-01 9.54109550e-01 1.02489626e+00 -1.30701974e-01
-1.75820157e-01 2.03658089e-01 3.74326035e-02 6.85126245e-01
7.16208518e-01 5.73510885e-01 -5.42104483e-01 -4.36171219e-02
7.82579362e-01 6.55019209e-02 -3.34503464e-02 -6.62087619e-01
-1.19501448e+00 3.71350586e-01 5.73966205e-01 -1.62643567e-01
-1.68799624e-01 2.60836989e-01 1.72490522e-01 -5.20694107e-02
2.11354136e-01 4.13765572e-02 -2.05468461e-01 -9.08167288e-02
-4.36666846e-01 2.36285135e-01 1.62733391e-01 1.20873392e+00
5.68878353e-01 1.39088243e-01 -4.56145406e-01 5.47117651e-01
7.84302413e-01 1.14546359e+00 -1.26341134e-01 -7.59363174e-01
4.52307522e-01 7.69226611e-01 4.67771649e-01 -1.11153042e+00
-3.19209188e-01 -5.09026170e-01 -5.40436924e-01 3.18246782e-01
5.33552229e-01 -5.00856377e-02 -7.30475962e-01 1.62376320e+00
7.98436880e-01 3.67193520e-01 -3.63353729e-01 1.09426701e+00
7.60895431e-01 3.26453269e-01 2.62441099e-01 -1.67216137e-01
1.38344252e+00 -6.82114542e-01 -6.52625442e-01 -1.91910461e-01
3.91030788e-01 -8.98229003e-01 8.03393483e-01 1.44637510e-01
-7.99562454e-01 -8.86629105e-01 -1.14106500e+00 5.04066885e-01
-1.09230503e-01 3.42825025e-01 6.88196123e-01 8.69681656e-01
-5.80913663e-01 3.10800165e-01 -8.67794991e-01 -1.96927890e-01
4.93485749e-01 5.49825132e-01 1.94687605e-01 4.68634479e-02
-8.92273188e-01 7.36356437e-01 3.47132981e-01 1.11368157e-01
-8.77584100e-01 -7.92296708e-01 -5.08377254e-01 -2.70676523e-01
7.17307091e-01 -6.79519594e-01 9.98187602e-01 -2.01921761e-01
-1.21878016e+00 5.39246678e-01 -4.42235097e-02 -1.84915692e-01
5.51039696e-01 -4.08617556e-01 -5.16418219e-01 -2.81710550e-02
1.37711406e-01 6.78137660e-01 8.40438664e-01 -1.28015912e+00
-1.26201475e+00 -4.06849802e-01 6.55879900e-02 1.64438441e-01
-3.98405641e-01 -9.60705429e-02 -7.65917778e-01 -4.09313589e-01
1.73875317e-01 -9.42750096e-01 -2.04430278e-02 5.77936888e-01
-1.91862643e-01 -4.69096929e-01 1.05901504e+00 -8.40718299e-02
1.21841717e+00 -2.07449698e+00 -1.21262856e-01 -4.28769253e-02
4.57485288e-01 2.80852526e-01 8.89144316e-02 -6.97665811e-02
5.40503383e-01 -1.10281833e-01 4.65314627e-01 -1.43710032e-01
3.81460823e-02 -2.34129988e-02 -1.04826227e-01 5.74750721e-01
1.94890156e-01 1.04886103e+00 -1.11474717e+00 -8.21285307e-01
5.66366017e-01 5.05308807e-01 -2.60203242e-01 2.38557950e-01
-2.94154584e-01 4.49984521e-01 -9.50070381e-01 1.16473997e+00
8.08151841e-01 -3.92901123e-01 -1.36828110e-01 -5.49749553e-01
-3.40480477e-01 -1.34537131e-01 -1.34190214e+00 1.30677187e+00
-1.22265205e-01 4.71258223e-01 -1.39076978e-01 -3.87542397e-01
1.12881517e+00 2.91841738e-02 7.89090812e-01 -6.63030863e-01
3.17760497e-01 1.27014652e-01 1.42491842e-02 -3.52535874e-01
5.51063776e-01 2.21749485e-01 4.51244712e-02 1.03398554e-01
-3.37002546e-01 2.60468453e-01 -1.11028533e-02 2.54939403e-02
8.39978933e-01 3.37130159e-01 2.38071978e-01 3.21761928e-02
5.12145162e-01 -2.04785615e-02 7.95102358e-01 7.93133914e-01
-7.68218279e-01 2.26838738e-01 2.39663832e-02 -2.89468616e-01
-8.84822667e-01 -1.07881868e+00 -8.39562789e-02 8.72878253e-01
8.94184113e-01 -2.52590328e-01 -1.44694641e-01 -6.44473970e-01
4.56650138e-01 1.56922624e-01 -5.31068861e-01 -2.36700103e-01
-5.20038068e-01 -7.15652704e-01 4.05754477e-01 7.79293597e-01
4.44234580e-01 -6.02536142e-01 -6.65500522e-01 1.18878603e-01
6.73788637e-02 -1.27386868e+00 -5.04049182e-01 -2.40483001e-01
-8.76869202e-01 -1.14407241e+00 -4.33406115e-01 -3.01359832e-01
4.01726842e-01 9.12133515e-01 7.80766726e-01 3.27930838e-01
-1.71622291e-01 3.54076743e-01 -2.35518515e-01 -4.11250174e-01
-6.46620169e-02 -7.71461576e-02 4.45434242e-01 -6.49914294e-02
5.03743470e-01 -3.75044525e-01 -8.09906065e-01 8.91565263e-01
-2.74010688e-01 -1.09786816e-01 8.67544532e-01 5.06717682e-01
8.04586589e-01 6.14603311e-02 3.37675691e-01 -2.39008486e-01
1.09889030e-01 -4.66803968e-01 -1.11989212e+00 2.52889693e-01
-8.53521943e-01 -1.41167596e-01 3.01378280e-01 -7.79082119e-01
-7.30524361e-01 1.28423452e-01 3.19028080e-01 -9.65355098e-01
-6.58418909e-02 -4.24878206e-03 -3.19886655e-01 -4.52537626e-01
4.47650701e-01 7.56105930e-02 -2.00711802e-01 -4.03626591e-01
1.76343277e-01 5.02608895e-01 4.63323474e-01 -6.53348327e-01
1.27444148e+00 5.64935565e-01 1.96210921e-01 -4.11297500e-01
-1.11942959e+00 -6.66243255e-01 -5.86531401e-01 -7.80393898e-01
7.34340489e-01 -1.11840737e+00 -1.28358090e+00 3.23766083e-01
-8.70861530e-01 1.72399044e-01 -2.37956159e-02 5.98241627e-01
-2.24736661e-01 3.21391463e-01 -2.75158525e-01 -1.24023294e+00
-1.56282648e-01 -1.07332194e+00 1.16811478e+00 6.61337912e-01
3.04752946e-01 -5.92583537e-01 -2.09048390e-01 2.95884818e-01
2.80556709e-01 2.56497681e-01 3.88287306e-01 -1.75811604e-01
-1.30830586e+00 -2.28978261e-01 -6.14143312e-01 -1.97903082e-01
1.53447151e-01 1.42157733e-01 -8.67896974e-01 -5.25946259e-01
-4.76968199e-01 -1.07942335e-01 5.67588449e-01 5.29440522e-01
9.46423411e-01 2.69452840e-01 -7.74943233e-01 6.71389878e-01
1.32132888e+00 3.34582090e-01 1.02382302e-01 4.20216382e-01
9.14603293e-01 1.85226575e-01 1.31608057e+00 5.02548337e-01
4.84203786e-01 9.97177303e-01 8.11986327e-01 2.44268149e-01
-3.89677972e-01 -3.97788823e-01 3.25058043e-01 6.07548952e-01
-2.32701361e-01 -2.45061025e-01 -6.99924350e-01 2.71775275e-01
-1.97352207e+00 -7.52414703e-01 -1.70657739e-01 2.14433503e+00
4.13107544e-01 5.64286351e-01 4.32851851e-01 -1.68442979e-01
8.60007167e-01 1.60203919e-01 -8.10032368e-01 7.06131816e-01
-5.71145713e-02 -4.89006549e-01 8.42034817e-01 1.49056464e-01
-1.20328856e+00 8.61259460e-01 6.05395222e+00 9.80241179e-01
-9.21949029e-01 -1.39363363e-01 3.25797759e-02 -1.47430152e-01
-5.26046986e-03 -1.41166523e-01 -1.63691735e+00 6.38899207e-01
6.16628408e-01 -1.88805491e-01 1.67528257e-01 9.23791528e-01
2.38869429e-01 1.79515593e-02 -8.74481440e-01 1.02470922e+00
-5.03663346e-02 -1.09453869e+00 -4.66742888e-02 2.27650717e-01
4.76829708e-01 2.26514246e-02 2.62085676e-01 1.87209561e-01
3.15126449e-01 -6.08087063e-01 7.44190156e-01 3.39675069e-01
7.33135819e-01 -5.53086340e-01 5.62070668e-01 3.03851575e-01
-1.96834075e+00 -1.76514819e-01 -4.79996622e-01 1.28264293e-01
1.24121584e-01 2.47936234e-01 -7.10111797e-01 7.74838150e-01
7.57604480e-01 9.78449345e-01 -6.23430729e-01 1.47577608e+00
-9.56749544e-02 4.47696805e-01 -5.14532566e-01 -3.11715037e-01
1.69178516e-01 3.04372236e-02 8.72860968e-01 7.57893443e-01
3.33605379e-01 7.17897564e-02 6.28058493e-01 7.80633569e-01
1.24972470e-01 -1.41309470e-01 -3.36711764e-01 1.38882563e-01
1.00122046e+00 1.27977908e+00 -6.82587922e-01 -1.27076015e-01
-5.22057831e-01 2.11946964e-01 1.14259094e-01 1.03887171e-01
-1.16960502e+00 -1.07774734e-02 7.93592811e-01 1.40090317e-01
6.51984274e-01 -2.95234978e-01 -3.34396586e-02 -8.54330778e-01
8.20373222e-02 -5.09862423e-01 4.27819401e-01 -7.04058051e-01
-1.26817393e+00 3.49097252e-01 2.40209382e-02 -1.92974377e+00
1.42618507e-01 -6.06072366e-01 -3.23096156e-01 6.27420247e-01
-1.77802980e+00 -1.29844093e+00 -6.71435356e-01 3.32853377e-01
4.39013451e-01 -1.86497331e-01 1.32742554e-01 6.15924120e-01
-6.94249630e-01 7.39936352e-01 -6.83590621e-02 5.11002131e-02
7.15015292e-01 -9.63692605e-01 1.85812324e-01 8.42476547e-01
3.27090435e-02 3.67012471e-01 6.76499784e-01 -8.05198133e-01
-1.80653536e+00 -1.07657242e+00 2.32606769e-01 -9.10904944e-01
7.90449500e-01 -1.39575899e-01 -7.02661574e-01 3.44469696e-01
-4.06850249e-01 3.06819052e-01 2.58920401e-01 1.11749455e-01
-2.51810551e-01 -3.77606034e-01 -7.50039876e-01 4.93458122e-01
1.24686921e+00 -1.18174113e-01 -3.58945608e-01 3.50601040e-02
9.18695748e-01 -6.67203903e-01 -1.01724041e+00 6.42993629e-01
8.87204289e-01 -8.53838027e-01 1.33069527e+00 -1.75396129e-01
-1.78768784e-01 -8.89517844e-01 -1.41243413e-01 -7.56964028e-01
-4.50167567e-01 -4.07100946e-01 -5.95890641e-01 1.28609133e+00
-3.60678434e-02 -4.08092380e-01 9.34275627e-01 1.90062746e-01
-1.12756528e-01 -7.42807508e-01 -9.30219948e-01 -1.02924252e+00
-5.59151530e-01 -4.74326015e-01 7.50934899e-01 6.31626904e-01
-6.52594924e-01 3.15473706e-01 -6.50999546e-01 5.74813783e-01
1.15601301e+00 4.42216694e-01 1.09736550e+00 -1.57196963e+00
-2.62928680e-02 -4.79040086e-01 -4.74730492e-01 -1.65028489e+00
-2.92600602e-01 -4.97023791e-01 6.18601516e-02 -1.07880998e+00
2.86069781e-01 -1.00122797e+00 -6.62508190e-01 1.80849031e-01
-6.36173606e-01 2.29659215e-01 3.42375547e-01 4.43425715e-01
-1.06931186e+00 6.45069420e-01 1.40894473e+00 -1.75620526e-01
-2.20534340e-01 3.79215896e-01 -6.03411615e-01 5.46835840e-01
4.33107287e-01 -5.61796188e-01 -1.67910680e-01 -3.64750445e-01
8.38440210e-02 5.34519777e-02 6.40547097e-01 -1.13476086e+00
3.79786372e-01 -3.08472842e-01 6.29478872e-01 -1.26484454e+00
4.72756833e-01 -1.09579158e+00 1.45981759e-01 4.97199148e-01
1.43303409e-01 -1.55217769e-02 3.32334340e-01 9.04109180e-01
1.10461533e-01 1.88973635e-01 6.69476807e-01 2.28961751e-01
-1.00065875e+00 8.14035237e-01 3.84281874e-01 -5.57738021e-02
1.11955440e+00 -6.27100170e-01 -4.49957401e-01 1.71293784e-02
-8.72195512e-02 7.07122266e-01 4.76897418e-01 7.17476845e-01
3.88409317e-01 -1.63264775e+00 -4.90149021e-01 1.44847929e-01
3.07180107e-01 -1.78639337e-01 2.68119037e-01 1.02247345e+00
1.21417210e-01 4.85497326e-01 -1.94130972e-01 -1.26937461e+00
-1.42089593e+00 5.29443800e-01 1.82627812e-01 -2.10346021e-02
-5.59149086e-01 6.33234799e-01 9.73020196e-02 1.49260655e-01
4.74974543e-01 -2.41794840e-01 -2.46102095e-01 -7.80192316e-02
5.00828803e-01 4.82160121e-01 -3.77850294e-01 -6.03875518e-01
-5.39287627e-01 1.06769741e+00 -4.89882864e-02 4.41505104e-01
9.61216450e-01 -4.17665124e-01 5.21288216e-01 3.21232736e-01
4.85716790e-01 2.41705142e-02 -1.79685235e+00 -4.29191917e-01
-1.38640970e-01 -9.67845798e-01 1.13846704e-01 -5.74005008e-01
-1.08625531e+00 6.95041418e-01 8.96207213e-01 2.80428022e-01
9.06169891e-01 1.43690318e-01 7.54168034e-01 1.72026008e-01
4.75747436e-01 -7.95744956e-01 2.73629308e-01 2.12005958e-01
2.21346036e-01 -1.58176005e+00 3.73528153e-01 -4.58302706e-01
-3.33543450e-01 8.44194591e-01 1.06691766e+00 1.70847893e-01
3.82228464e-01 2.54797876e-01 1.97153017e-02 -2.03449592e-01
-6.16747856e-01 -5.27112365e-01 5.68130970e-01 7.29986072e-01
3.04396059e-02 -8.20214599e-02 1.65661395e-01 3.35092872e-01
1.83288649e-01 -2.50781953e-01 -2.23871231e-01 7.66499937e-01
-7.52027273e-01 -9.08581316e-01 -6.64107144e-01 4.75916803e-01
-1.65520892e-01 3.79228622e-01 -3.90314963e-04 1.05239868e+00
2.15699762e-01 7.66835093e-01 -7.03132227e-02 -6.28760993e-01
5.94232619e-01 -6.11182570e-01 5.18717766e-01 -1.77683935e-01
-3.05726707e-01 3.04050714e-01 -2.00451761e-01 -5.60855389e-01
-6.18076563e-01 -6.40624166e-01 -1.04436648e+00 -4.45246637e-01
-9.36841607e-01 -1.57025293e-01 3.71923566e-01 7.83368826e-01
2.73520231e-01 6.06660306e-01 6.69613838e-01 -8.24690163e-01
-5.87522924e-01 -6.46004617e-01 -2.68196702e-01 1.22214355e-01
4.23621595e-01 -1.45513141e+00 -2.56102175e-01 -3.63884628e-01] | [6.43826961517334, -2.184223175048828] |
948c51ee-b7f5-4b92-869f-7826baf4756e | neural-program-synthesis-with-query-1 | 2205.07857 | null | https://arxiv.org/abs/2205.07857v1 | https://arxiv.org/pdf/2205.07857v1.pdf | Neural Program Synthesis with Query | Aiming to find a program satisfying the user intent given input-output examples, program synthesis has attracted increasing interest in the area of machine learning. Despite the promising performance of existing methods, most of their success comes from the privileged information of well-designed input-output examples. However, providing such input-output examples is unrealistic because it requires the users to have the ability to describe the underlying program with a few input-output examples under the training distribution. In this work, we propose a query-based framework that trains a query neural network to generate informative input-output examples automatically and interactively from a large query space. The quality of the query depends on the amount of the mutual information between the query and the corresponding program, which can guide the optimization of the query framework. To estimate the mutual information more accurately, we introduce the functional space (F-space) which models the relevance between the input-output examples and the programs in a differentiable way. We evaluate the effectiveness and generalization of the proposed query-based framework on the Karel task and the list processing task. Experimental results show that the query-based framework can generate informative input-output examples which achieve and even outperform well-designed input-output examples. | ['Yunji Chen', 'Qi Guo', 'Zidong Du', 'Nan Li', 'Pengwei Jin', 'Xishan Zhang', 'Xing Hu', 'Rui Zhang', 'Di Huang'] | 2022-05-08 | neural-program-synthesis-with-query | https://openreview.net/forum?id=NyJ2KIN8P17 | https://openreview.net/pdf?id=NyJ2KIN8P17 | iclr-2022-4 | ['program-synthesis'] | ['computer-code'] | [ 3.50317508e-01 -7.67515078e-02 -3.92297179e-01 -6.33643985e-01
-8.09683383e-01 -5.85846305e-01 3.41479748e-01 4.10008430e-01
-2.84220070e-01 3.22726399e-01 -1.44898906e-01 -3.42885137e-01
-2.58592844e-01 -1.07603276e+00 -1.02356231e+00 -2.99330533e-01
1.36399895e-01 4.73470002e-01 1.52478144e-01 -1.43429533e-01
4.85565245e-01 1.50560126e-01 -2.03554463e+00 3.09180200e-01
1.33235705e+00 7.24519193e-01 5.82965910e-01 6.73404455e-01
-3.71214777e-01 5.15490413e-01 -6.47866309e-01 -3.54078352e-01
2.53097098e-02 -5.68006039e-01 -5.52243114e-01 -2.04688758e-01
2.48704389e-01 -2.39937268e-02 -1.25927106e-01 1.34041798e+00
2.65035272e-01 2.21556962e-01 6.26103044e-01 -1.24893582e+00
-7.44771123e-01 6.52321160e-01 1.40831605e-01 -1.14241406e-01
5.75970709e-01 3.16821158e-01 1.40328717e+00 -9.93619561e-01
5.40390670e-01 1.14058995e+00 1.58702970e-01 4.44818765e-01
-1.39012861e+00 -5.45982480e-01 -5.66560812e-02 1.37715966e-01
-1.36181116e+00 2.63881870e-03 7.07577944e-01 -6.12137616e-01
6.88762307e-01 5.55267930e-01 7.06605494e-01 3.63781154e-01
-1.09748587e-01 1.09107149e+00 5.07721364e-01 -3.72576118e-01
1.50606290e-01 7.30771542e-01 4.36037928e-01 7.24467039e-01
1.56636406e-02 1.88906193e-01 -4.64565188e-01 -3.81746113e-01
5.13345063e-01 1.76075026e-01 -5.62319100e-01 -5.76338947e-01
-1.10535383e+00 8.66201520e-01 6.31260574e-01 3.72657061e-01
-1.47946730e-01 -5.87602444e-02 2.74221987e-01 4.34783429e-01
2.05537483e-01 8.70397031e-01 -3.81691694e-01 -1.41168207e-01
-8.80036891e-01 4.37982798e-01 9.25796568e-01 1.28677142e+00
1.07720959e+00 -2.22826034e-01 -5.01744747e-01 5.49751163e-01
1.44117296e-01 5.16368270e-01 4.52582300e-01 -4.15018827e-01
6.29946947e-01 1.27729821e+00 9.30529982e-02 -1.05816770e+00
1.33661121e-01 -3.18310022e-01 -6.10114515e-01 -7.90532902e-02
4.13755655e-01 1.43638596e-01 -3.34375829e-01 1.89123487e+00
9.42016616e-02 -3.56821746e-01 1.91496983e-01 8.50310922e-01
6.56595170e-01 9.50710118e-01 -2.78744727e-01 -2.65675038e-01
9.42684412e-01 -8.79617155e-01 -5.19109666e-01 -2.39636302e-01
9.22077358e-01 -5.09475172e-01 1.86514282e+00 9.99229550e-02
-7.57324100e-01 -6.86377645e-01 -9.31420982e-01 1.58793643e-01
-2.38275781e-01 5.95108092e-01 4.09971774e-01 4.44446146e-01
-6.99985147e-01 5.33411562e-01 -3.07621151e-01 -6.28932565e-02
2.87660003e-01 4.16126937e-01 -1.42741308e-01 -2.92228431e-01
-1.04596674e+00 4.24060881e-01 6.90777302e-01 -9.81161743e-02
-7.10241616e-01 -8.66251767e-01 -9.03017879e-01 4.89305645e-01
5.78157246e-01 -5.95403612e-01 1.13055563e+00 -1.17217910e+00
-1.14607644e+00 4.24102694e-01 -3.48713607e-01 -1.13138027e-01
4.05221671e-01 -4.94671799e-02 -4.68217917e-02 -1.45849586e-01
-9.27225202e-02 4.19053286e-01 5.70175648e-01 -9.40803051e-01
-5.40984929e-01 -3.59714985e-01 1.64284304e-01 2.74603784e-01
-6.09490991e-01 -2.11887121e-01 -5.25560975e-01 -4.17965233e-01
-6.43911436e-02 -7.60753572e-01 -1.98205993e-01 -5.53677380e-02
-6.19319141e-01 -5.09413242e-01 5.70421278e-01 -2.42723867e-01
1.66864109e+00 -2.40460491e+00 2.60430753e-01 2.21905932e-01
-7.01445416e-02 1.14034027e-01 2.06093807e-02 4.78374869e-01
1.59655865e-02 1.96067318e-01 -4.47799683e-01 1.47044972e-01
2.11984023e-01 -8.45904201e-02 -4.26874399e-01 1.18615001e-01
3.21856856e-01 8.69721472e-01 -9.07383442e-01 -4.15686846e-01
3.26448344e-02 9.40934941e-02 -8.99776816e-01 8.23343635e-01
-8.88736069e-01 2.57970124e-01 -6.65145099e-01 4.07028168e-01
1.74538672e-01 -4.15864050e-01 -1.17194906e-01 -2.16079265e-01
1.34187862e-01 1.60747394e-01 -1.10688901e+00 1.38745368e+00
-5.72278619e-01 5.54267943e-01 -5.13368666e-01 -7.74892092e-01
1.13997114e+00 9.29076821e-02 -5.08178547e-02 -3.54829431e-01
-2.57569522e-01 3.68798763e-01 7.11488277e-02 -7.80637741e-01
5.29631972e-01 -5.64674437e-02 -3.65407795e-01 5.99273324e-01
-2.21629307e-01 -3.56119812e-01 2.55726039e-01 2.80494899e-01
8.46556306e-01 -5.62895946e-02 3.86597604e-01 -1.48214936e-01
8.55057716e-01 8.45629796e-02 2.52283812e-01 6.99755490e-01
3.81225973e-01 3.35561067e-01 7.58083463e-01 -2.93301910e-01
-8.31809461e-01 -5.37119865e-01 2.08436981e-01 1.28223646e+00
1.15543850e-01 -4.69391733e-01 -5.79419672e-01 -7.24709630e-01
3.01906839e-03 9.88987207e-01 -4.55144167e-01 -3.51596475e-01
-4.45180058e-01 -1.66098967e-01 3.24781001e-01 3.03556085e-01
2.13698208e-01 -1.28404319e+00 -6.66657090e-01 9.31711122e-03
-9.52685401e-02 -4.21884149e-01 -8.90643299e-01 1.73879519e-01
-8.04321408e-01 -1.09265387e+00 -6.19028449e-01 -8.17186117e-01
1.09307075e+00 -5.19523732e-02 1.09399629e+00 2.10069269e-01
-1.70748562e-01 1.97704032e-01 -6.04439974e-02 -1.05672345e-01
-6.64877415e-01 1.72275379e-01 -3.39223832e-01 2.71973014e-02
3.79179537e-01 -1.76303267e-01 -2.33976826e-01 2.95503199e-01
-1.19242132e+00 2.45286778e-01 7.79686928e-01 9.68203902e-01
6.27358198e-01 2.54224181e-01 6.77144706e-01 -1.19330490e+00
8.75750422e-01 -4.18735147e-01 -9.76497948e-01 5.50222397e-01
-7.37193108e-01 5.92708230e-01 1.00147140e+00 -6.36298180e-01
-7.41035283e-01 2.76253879e-01 8.14874321e-02 -3.50162446e-01
1.29150460e-02 9.13399875e-01 -6.79192126e-01 2.61689723e-01
7.91303992e-01 7.37386703e-01 -8.73081088e-02 -2.22323611e-01
4.65609282e-01 5.58693588e-01 4.03548211e-01 -7.87937164e-01
7.24797130e-01 -3.68844032e-01 -2.74173141e-01 -6.76752269e-01
-4.65233207e-01 -4.33806390e-01 -4.53465521e-01 -6.17236942e-02
4.23162669e-01 -3.97281289e-01 -6.80614829e-01 9.58931968e-02
-1.25382066e+00 -1.21405609e-01 -1.30414829e-01 2.08059177e-01
-5.95079541e-01 1.18466057e-01 -5.31773828e-02 -9.41365600e-01
-3.08924198e-01 -1.72491455e+00 9.22708631e-01 1.88610360e-01
-1.98195145e-01 -7.97069430e-01 -2.13736340e-01 -1.87418386e-01
3.46398056e-01 -4.25007157e-02 1.63246012e+00 -1.11473835e+00
-1.28349388e+00 -6.51808143e-01 -1.79176643e-01 3.29593539e-01
1.06068909e-01 -1.07278675e-01 -6.26179338e-01 -2.88375374e-02
-4.91128005e-02 -3.24506164e-01 4.58797514e-01 8.65622908e-02
1.26348567e+00 -7.17189014e-01 -2.86894828e-01 4.73715156e-01
1.49259162e+00 1.80394232e-01 3.39237779e-01 -7.44678900e-02
5.40393054e-01 6.73871160e-01 8.45460474e-01 3.39435071e-01
1.40092060e-01 5.51693261e-01 3.04091185e-01 2.88851291e-01
5.40390670e-01 -6.30814672e-01 2.70487309e-01 5.74755013e-01
6.38009667e-01 -1.17698178e-01 -8.10144246e-01 5.48809469e-01
-1.84571195e+00 -7.50167668e-01 1.22611523e-01 2.54849792e+00
1.15292239e+00 2.07203627e-01 3.07664778e-02 -8.13832954e-02
7.01546907e-01 -1.50555596e-01 -9.06358778e-01 -1.29488900e-01
2.33652249e-01 -1.64835632e-01 -2.74984669e-02 4.43384320e-01
-7.41891205e-01 6.21489048e-01 5.28208780e+00 9.36142504e-01
-1.04035020e+00 -5.95320463e-01 5.82930088e-01 3.18488538e-01
-7.67067373e-01 -3.42652798e-02 -8.58647883e-01 3.83642077e-01
8.23839188e-01 -9.43781495e-01 6.08002961e-01 1.35783911e+00
4.75501455e-03 -5.08077815e-02 -1.85494566e+00 8.52675736e-01
-1.16279446e-01 -1.14552259e+00 2.81435519e-01 -9.12198052e-02
3.66982460e-01 -6.12350285e-01 2.41702154e-01 6.41519010e-01
-1.30991206e-01 -1.00957274e+00 5.55780292e-01 6.49440169e-01
7.45213270e-01 -8.85642231e-01 6.29647911e-01 1.04043019e+00
-1.07955110e+00 -8.22728276e-02 -3.18540424e-01 5.22099547e-02
-4.58537549e-01 5.64426541e-01 -1.12214339e+00 3.39167982e-01
2.28095561e-01 5.19892991e-01 -6.51934147e-01 1.09730899e+00
-2.17418179e-01 2.76547194e-01 -2.20290005e-01 -8.11616719e-01
-7.30414689e-02 -2.88550407e-01 4.55969006e-01 1.07045901e+00
5.04477859e-01 8.88206735e-02 3.33437890e-01 1.54447401e+00
-1.69490367e-01 4.52341318e-01 -1.06272590e+00 -4.18836355e-01
5.81780016e-01 9.18549478e-01 -2.55553097e-01 -4.49755192e-01
-1.58934236e-01 7.30260789e-01 3.50605667e-01 3.32391769e-01
-5.03880322e-01 -6.57715559e-01 3.49545121e-01 -6.68931454e-02
9.22016427e-02 1.59125090e-01 -2.59190112e-01 -9.48784590e-01
3.63315254e-01 -1.07671225e+00 3.03149164e-01 -5.54791689e-01
-1.08676410e+00 6.24855459e-01 1.61325067e-01 -1.15819442e+00
-5.86430609e-01 -2.38030225e-01 -5.64351916e-01 1.30685735e+00
-9.77896512e-01 -6.34136319e-01 -2.51079202e-01 2.96429604e-01
5.35512924e-01 -2.38859057e-01 8.53266418e-01 1.39983758e-01
-4.16729093e-01 7.90070057e-01 -1.66632712e-01 4.94354479e-02
4.42017406e-01 -1.40561223e+00 1.57850802e-01 6.58693254e-01
-1.90236140e-02 1.07940125e+00 8.42520416e-01 -5.71625233e-01
-1.84622967e+00 -1.21251476e+00 8.87865841e-01 -3.98404390e-01
4.05241311e-01 -1.90562040e-01 -1.18189168e+00 5.84211707e-01
-9.03879404e-02 -1.54542267e-01 6.21425629e-01 -7.61067495e-02
-2.45946199e-01 -7.21691102e-02 -9.54286993e-01 6.46449804e-01
5.40970623e-01 -6.70092523e-01 -4.87259865e-01 3.09721828e-01
1.06059313e+00 -2.19378725e-01 -6.24413252e-01 3.04612905e-01
3.45349580e-01 -8.24575007e-01 6.94696844e-01 -6.01925015e-01
6.23466253e-01 -3.94794852e-01 -2.92027593e-01 -1.35048759e+00
2.21184805e-01 -3.55712771e-01 6.07915781e-02 1.10398710e+00
6.71439588e-01 -3.33043158e-01 7.33398378e-01 9.27361488e-01
-1.14640750e-01 -9.64973330e-01 -4.65212703e-01 -5.37213266e-01
-2.39368990e-01 -5.36888242e-01 8.58778238e-01 5.41167736e-01
2.40095973e-01 6.20931566e-01 -9.16106701e-02 1.36976048e-01
3.52032483e-01 7.14151978e-01 1.07019222e+00 -1.02162075e+00
-6.69286311e-01 -6.27363086e-01 -1.53481051e-01 -1.36167777e+00
3.08922261e-01 -1.17454219e+00 2.32317403e-01 -1.23919284e+00
2.17914149e-01 -5.36229670e-01 1.82159737e-01 2.56589323e-01
-6.92906678e-01 -6.88864410e-01 6.18849024e-02 2.85628170e-01
-4.30731654e-01 7.43521750e-01 1.20655262e+00 -2.92428643e-01
-6.18557334e-01 6.08467340e-01 -7.04179883e-01 3.94099444e-01
4.49725062e-01 -4.60667282e-01 -7.08398938e-01 -1.87377125e-01
5.10320187e-01 6.10359490e-01 1.79759309e-01 -7.10877478e-01
4.52949435e-01 -3.06299537e-01 -8.41270611e-02 -6.92622066e-01
3.02941706e-02 -9.66828883e-01 1.43814579e-01 4.91419375e-01
-1.02925229e+00 1.91974670e-01 4.65339376e-03 7.79550672e-01
-4.10776615e-01 -6.92484975e-01 5.23539245e-01 -2.27496862e-01
-5.25426030e-01 2.55708575e-01 2.45212000e-02 -1.64993890e-02
7.63234556e-01 1.35786250e-01 -2.65340954e-01 -4.15193439e-01
-3.63810301e-01 2.84851253e-01 5.15334547e-01 1.82949647e-01
9.26771462e-01 -1.33995080e+00 -4.34152186e-01 5.39243162e-01
6.52617216e-01 1.50261000e-01 -1.21089578e-01 4.59313542e-01
-2.52816260e-01 6.30035043e-01 2.71029323e-01 -7.29392827e-01
-1.20790124e+00 8.40597332e-01 2.30161726e-01 -2.59947121e-01
-3.09266686e-01 5.61218202e-01 3.86256844e-01 -7.59745002e-01
5.59297621e-01 -5.28005898e-01 -2.51105577e-01 -3.44075501e-01
5.33105373e-01 -4.79692779e-02 -2.00351626e-01 -8.40917453e-02
-1.54861398e-02 4.11806852e-02 2.86428574e-02 -3.21880654e-02
1.12502277e+00 4.52165365e-01 -2.49209434e-01 6.15071118e-01
1.46465814e+00 -2.60328978e-01 -7.72278547e-01 -5.28729141e-01
3.48214388e-01 -6.91414416e-01 -2.95598984e-01 -5.50509572e-01
-7.17215300e-01 1.05873561e+00 3.78473341e-01 2.71005422e-01
1.11165011e+00 3.53752598e-02 3.85446340e-01 9.15936828e-01
3.51228803e-01 -5.71613908e-01 3.59861016e-01 4.29588854e-01
8.89032781e-01 -1.10745358e+00 -3.26817989e-01 -4.45978880e-01
-5.05741060e-01 1.24831963e+00 9.35362816e-01 1.35683328e-01
4.07322735e-01 -1.33443460e-01 -4.50927794e-01 -1.95864633e-01
-8.22516084e-01 -4.95991744e-02 6.17611408e-01 3.30910802e-01
5.81106067e-01 3.57083976e-02 -3.63077372e-02 8.79581571e-01
-2.58768350e-01 9.33658257e-02 2.40929455e-01 8.07310879e-01
-7.41141737e-01 -1.08866322e+00 -2.37965405e-01 7.56450891e-01
4.32009846e-02 -2.04096630e-01 -2.86167294e-01 7.00121641e-01
-4.36717451e-01 5.30565500e-01 -3.66246551e-01 -4.61249858e-01
6.15839422e-01 3.14773530e-01 1.17181070e-01 -1.04840112e+00
-5.50491512e-01 -1.22970216e-01 -1.55036524e-01 -2.19123185e-01
3.33519615e-02 -2.88885564e-01 -1.23769987e+00 6.30461872e-02
-5.82294285e-01 4.39271450e-01 5.45423985e-01 7.83534288e-01
3.09126645e-01 4.58298862e-01 7.82463670e-01 -3.98891568e-01
-1.03263068e+00 -8.57035756e-01 -3.68837476e-01 4.79301333e-01
3.26631337e-01 -2.36582339e-01 -2.86333770e-01 1.62840709e-02] | [7.907402038574219, 7.702089786529541] |
25f96eeb-4353-4121-bfcb-ec0917b0219d | learning-object-language-alignments-for-open | 2211.14843 | null | https://arxiv.org/abs/2211.14843v1 | https://arxiv.org/pdf/2211.14843v1.pdf | Learning Object-Language Alignments for Open-Vocabulary Object Detection | Existing object detection methods are bounded in a fixed-set vocabulary by costly labeled data. When dealing with novel categories, the model has to be retrained with more bounding box annotations. Natural language supervision is an attractive alternative for its annotation-free attributes and broader object concepts. However, learning open-vocabulary object detection from language is challenging since image-text pairs do not contain fine-grained object-language alignments. Previous solutions rely on either expensive grounding annotations or distilling classification-oriented vision models. In this paper, we propose a novel open-vocabulary object detection framework directly learning from image-text pair data. We formulate object-language alignment as a set matching problem between a set of image region features and a set of word embeddings. It enables us to train an open-vocabulary object detector on image-text pairs in a much simple and effective way. Extensive experiments on two benchmark datasets, COCO and LVIS, demonstrate our superior performance over the competing approaches on novel categories, e.g. achieving 32.0% mAP on COCO and 21.7% mask mAP on LVIS. Code is available at: https://github.com/clin1223/VLDet. | ['Jianfei Cai', 'Zehuan Yuan', 'Gholamreza Haffari', 'Lizhen Qu', 'Ping Luo', 'Yi Jiang', 'Peize Sun', 'Chuang Lin'] | 2022-11-27 | null | null | null | null | ['set-matching', 'open-vocabulary-object-detection'] | ['computer-vision', 'computer-vision'] | [ 1.23588070e-01 3.14167365e-02 -3.59643430e-01 -6.19212329e-01
-1.04874706e+00 -6.02388203e-01 6.38768494e-01 3.36458892e-01
-6.37800336e-01 2.89975435e-01 -2.12958649e-01 -4.39289063e-02
3.51839572e-01 -4.61313456e-01 -7.62750030e-01 -4.97007102e-01
2.27137700e-01 7.27556527e-01 5.42935371e-01 9.99843515e-03
5.84416203e-02 2.20189884e-01 -1.56838167e+00 3.11595082e-01
5.86481035e-01 1.16032994e+00 4.51059103e-01 4.87698197e-01
-3.69311571e-01 4.80284333e-01 -2.22727075e-01 -4.71683294e-01
4.90834296e-01 5.27374819e-02 -7.63687670e-01 3.45173895e-01
1.02733421e+00 -1.77606285e-01 -1.74667642e-01 1.30964100e+00
2.82136261e-01 -1.85453191e-01 7.74258733e-01 -1.23992324e+00
-9.45227265e-01 4.18543428e-01 -7.93209970e-01 3.52531463e-01
1.68429703e-01 2.74743378e-01 1.20667624e+00 -1.48836315e+00
5.57795882e-01 1.23717749e+00 5.47423899e-01 5.75229704e-01
-1.33671141e+00 -8.23375940e-01 3.11736137e-01 1.35995105e-01
-1.84822345e+00 -2.79738069e-01 4.06358242e-01 -6.86949134e-01
8.46100986e-01 2.14979991e-01 4.18112099e-01 8.91099155e-01
-1.42914683e-01 9.23124313e-01 8.86638224e-01 -4.77789581e-01
-6.77488968e-02 6.19949520e-01 3.14178735e-01 8.85751128e-01
5.89580894e-01 -5.47272414e-02 -3.30420643e-01 -3.56503623e-03
4.94435608e-01 2.29875728e-01 -1.56469256e-01 -8.01523983e-01
-1.30670643e+00 9.23091590e-01 7.68782437e-01 1.96252093e-01
5.03694378e-02 1.09516203e-01 4.56919938e-01 9.16577652e-02
7.02495515e-01 2.78648078e-01 -2.87057579e-01 5.08626282e-01
-7.30324268e-01 1.29432827e-01 4.39201623e-01 1.35502613e+00
8.55169177e-01 -3.12778324e-01 -1.42724782e-01 8.97658944e-01
5.67339122e-01 7.64677227e-01 3.53066593e-01 -2.24761650e-01
4.31636333e-01 6.52237773e-01 -7.49494210e-02 -7.63497412e-01
-2.46586651e-01 -2.96292990e-01 -6.25123560e-01 -1.25432670e-01
3.62789989e-01 2.73373723e-01 -1.08428478e+00 1.34727645e+00
5.83386302e-01 3.68825123e-02 -1.18303277e-01 9.22110081e-01
1.06566477e+00 4.87548798e-01 9.83844101e-02 1.14950523e-01
1.75480390e+00 -1.12450147e+00 -5.95249295e-01 -5.11195540e-01
7.89429307e-01 -8.79303634e-01 1.26427960e+00 3.96828055e-02
-6.62270308e-01 -5.55231631e-01 -9.28692460e-01 -3.96324396e-01
-5.96504748e-01 1.70137167e-01 2.59041667e-01 3.38599890e-01
-6.77094758e-01 -1.70474425e-01 -6.02981806e-01 -5.53003967e-01
7.88605332e-01 3.30926239e-01 -3.62067640e-01 -2.30188385e-01
-8.11847448e-01 8.58944535e-01 6.72101259e-01 -2.00530499e-01
-9.28758800e-01 -5.77990472e-01 -1.08399332e+00 -6.43999800e-02
7.80054152e-01 -4.56624091e-01 1.40052605e+00 -9.61227119e-01
-7.15272486e-01 1.52746153e+00 1.46950021e-01 -4.65533197e-01
4.84405220e-01 -2.51631796e-01 -1.98263675e-01 1.11553676e-01
4.98266846e-01 1.05646443e+00 1.13027704e+00 -1.19717360e+00
-9.52013969e-01 -5.13445973e-01 -2.36569010e-02 -1.72779895e-02
-5.40181220e-01 2.21017703e-01 -6.28632605e-01 -7.65043855e-01
-8.31872504e-03 -1.00795269e+00 -1.19347259e-01 5.73401213e-01
-3.88198882e-01 -7.04128683e-01 9.22784090e-01 -3.70244771e-01
8.94356072e-01 -2.12572408e+00 -2.05755860e-01 -1.59771353e-01
4.89508778e-01 3.44487727e-01 -2.67683655e-01 -6.08463995e-02
8.28639269e-02 4.26484607e-02 -1.82315946e-01 -3.34094524e-01
-8.22029933e-02 4.94619533e-02 -4.69035476e-01 6.00653529e-01
5.94177246e-01 1.07673430e+00 -8.91808212e-01 -9.49020147e-01
1.86634988e-01 1.72423005e-01 -4.88240689e-01 1.81547090e-01
-4.65380222e-01 1.98648036e-01 -4.56276238e-01 8.90978634e-01
5.97067893e-01 -5.28033972e-01 -1.54315889e-01 -1.44042864e-01
-3.36889774e-02 -9.49664712e-02 -1.01112890e+00 1.68574607e+00
-3.84520382e-01 7.78234065e-01 -1.48659408e-01 -1.03189385e+00
1.02490056e+00 -4.55674194e-02 1.48893610e-01 -5.29854953e-01
3.64433229e-01 1.91438287e-01 -1.07888833e-01 -6.25919998e-01
2.93605298e-01 1.09253572e-02 -1.72313541e-01 2.01623291e-01
2.65873909e-01 -3.76808733e-01 2.57720679e-01 2.94690579e-01
7.25623131e-01 -1.57032296e-01 5.39650917e-01 -4.69445467e-01
6.34040177e-01 1.61554471e-01 3.46549481e-01 8.89852524e-01
-2.56459981e-01 6.27394021e-01 3.45615521e-02 -6.10980749e-01
-1.11553705e+00 -1.08545363e+00 -6.75813735e-01 1.24148619e+00
3.72350931e-01 -4.29283619e-01 -5.80562532e-01 -8.88636112e-01
1.88019663e-01 4.03447628e-01 -5.92669368e-01 -3.31931002e-02
-4.64051723e-01 -3.54359627e-01 4.12333965e-01 6.33302689e-01
3.53339851e-01 -8.66126776e-01 -3.92282426e-01 -6.21185228e-02
-1.27522379e-01 -1.45157957e+00 -9.04657304e-01 2.23098814e-01
-6.19887292e-01 -1.13531721e+00 -6.93411946e-01 -1.41246295e+00
9.20054376e-01 6.76688433e-01 1.08868694e+00 1.73342362e-01
-9.55683827e-01 5.15420139e-01 -2.68434733e-01 -8.91523242e-01
-2.89986253e-01 -1.42434668e-02 2.01885954e-01 2.35123113e-01
8.87936294e-01 1.79907858e-01 -4.80035812e-01 3.89753997e-01
-7.82956898e-01 5.17482832e-02 5.58345616e-01 1.03482187e+00
9.32718456e-01 -5.36151350e-01 3.68283659e-01 -5.83266318e-01
1.49660900e-01 -3.36805254e-01 -9.98800039e-01 3.76314610e-01
-4.95005727e-01 -1.43607315e-02 2.42449626e-01 -6.84684634e-01
-7.65225768e-01 4.26194191e-01 2.38221273e-01 -6.44188285e-01
-1.56838194e-01 1.07434787e-01 -4.43210006e-02 -7.80004635e-02
7.61480153e-01 2.37418517e-01 -1.50803356e-02 -4.50247288e-01
6.27540052e-01 9.85743940e-01 5.35229445e-01 -3.65873992e-01
1.03016329e+00 7.65371442e-01 -6.16517007e-01 -9.25747871e-01
-1.24168932e+00 -1.09721649e+00 -9.18127775e-01 -1.24743860e-02
1.05160844e+00 -1.21811330e+00 -2.69650042e-01 2.21608430e-01
-1.12998903e+00 -5.77262007e-02 -4.15378451e-01 4.55951124e-01
-4.33455527e-01 3.62767547e-01 -1.60353869e-01 -4.86908764e-01
-4.58444268e-01 -1.08694494e+00 1.54831374e+00 4.15560156e-02
-6.76479563e-03 -6.84435904e-01 -1.27192900e-01 6.29209876e-01
-4.82971184e-02 -2.33767673e-01 3.87352526e-01 -9.29087222e-01
-7.82340527e-01 -5.74881315e-01 -6.60243452e-01 5.18618286e-01
1.08625136e-01 -2.22637206e-01 -9.83579934e-01 -4.60112691e-01
-6.70976564e-02 -6.85941577e-01 1.02143943e+00 3.83482836e-02
1.31147707e+00 -1.98737115e-01 -6.33681834e-01 5.95638335e-01
1.31075478e+00 -1.88405827e-01 9.78620350e-02 3.35328966e-01
8.61083984e-01 5.00786722e-01 9.88777816e-01 2.43177399e-01
2.63390601e-01 7.51528084e-01 3.80873650e-01 -1.42019480e-01
-1.94997519e-01 -1.30715504e-01 1.72411352e-01 4.43248719e-01
4.92755234e-01 -1.57363400e-01 -1.35871565e+00 8.38376403e-01
-1.74853420e+00 -5.88145554e-01 -2.08336160e-01 1.96781266e+00
9.51627851e-01 2.32183158e-01 -4.46669422e-02 -3.67187470e-01
9.02131021e-01 -1.27464071e-01 -6.19808018e-01 8.77890810e-02
-2.63611935e-02 -6.61967741e-03 6.11562669e-01 3.22358727e-01
-1.52338660e+00 1.16943038e+00 4.86830616e+00 8.19580138e-01
-9.51384842e-01 4.91228372e-01 4.92630482e-01 -3.51616561e-01
1.52264178e-01 -1.37288451e-01 -1.34885561e+00 3.08652073e-01
4.29576159e-01 -1.17187157e-01 -1.69679597e-01 1.14012611e+00
-1.86706230e-01 2.74679326e-02 -1.44804144e+00 1.38713551e+00
4.78965402e-01 -1.42248678e+00 1.79860622e-01 -1.06145732e-01
6.81908607e-01 3.93450737e-01 1.54129798e-02 4.21055645e-01
1.22117050e-01 -9.05166566e-01 8.13799977e-01 5.79923913e-02
1.02049637e+00 -2.22286373e-01 6.11916542e-01 3.31105649e-01
-1.32435894e+00 -4.96834964e-02 -4.85582292e-01 1.22655913e-01
-7.90649801e-02 3.23940009e-01 -1.04701626e+00 4.50939462e-02
1.05213463e+00 7.27510929e-01 -8.54070544e-01 1.00166702e+00
-1.67089790e-01 4.57678318e-01 -4.49217409e-01 -1.73209563e-01
3.66013527e-01 6.73127621e-02 4.60435808e-01 1.25460505e+00
-1.95300877e-01 -2.21591666e-02 7.14899302e-01 9.40424621e-01
-2.91454673e-01 3.53569537e-01 -8.59275281e-01 1.26515403e-01
3.13893437e-01 1.39171040e+00 -8.11256289e-01 -5.51422238e-01
-9.41528440e-01 7.92550147e-01 5.56159139e-01 -7.14316666e-02
-9.46360290e-01 -2.73093402e-01 6.48715436e-01 2.58471191e-01
5.28804243e-01 -4.84104045e-02 -1.50944754e-01 -1.42194247e+00
2.38289312e-01 -7.06920326e-01 5.05054414e-01 -6.09494209e-01
-1.48866463e+00 4.94900167e-01 7.27860853e-02 -1.29539597e+00
2.65670449e-01 -9.22711134e-01 -1.96102887e-01 5.41762412e-01
-1.64759088e+00 -1.46492708e+00 -5.34051299e-01 6.53845727e-01
1.00965202e+00 -2.38578275e-01 5.53454876e-01 4.14449543e-01
-4.67726618e-01 7.56810248e-01 2.26143226e-01 5.14833272e-01
8.58607352e-01 -1.13083279e+00 4.27200556e-01 7.57936358e-01
5.27212620e-01 3.34804803e-01 4.51084971e-01 -5.34880400e-01
-1.19583857e+00 -1.57068253e+00 7.58669376e-01 -9.15402293e-01
9.98931408e-01 -9.43692029e-01 -1.15314341e+00 8.89564276e-01
6.07775599e-02 6.58055246e-01 4.79982883e-01 -7.51431733e-02
-6.36228144e-01 -6.51987121e-02 -8.07443142e-01 4.92351443e-01
1.09199119e+00 -5.78499675e-01 -8.67423356e-01 8.98717344e-01
9.32255924e-01 -3.69780391e-01 -6.28338039e-01 3.98359060e-01
3.01814765e-01 -2.81072944e-01 1.08740628e+00 -7.45056152e-01
9.13236663e-03 -3.69666636e-01 -2.26034611e-01 -6.85941458e-01
-2.54231215e-01 -1.57716885e-01 8.92540067e-02 1.16479993e+00
2.73146778e-01 -6.26029670e-01 4.74492192e-01 2.66594917e-01
1.10837355e-01 -6.61746085e-01 -8.57391834e-01 -1.08289611e+00
6.78073242e-02 -4.75426257e-01 1.76818818e-01 8.58360589e-01
-3.14982563e-01 6.88116193e-01 -7.47547001e-02 3.36201489e-01
7.80190051e-01 1.97911233e-01 9.28592563e-01 -1.27829456e+00
1.18509345e-01 -3.70833486e-01 -7.03839362e-01 -1.11936283e+00
2.51556993e-01 -1.25862300e+00 2.97395647e-01 -1.23388922e+00
6.32437050e-01 -6.28463745e-01 -2.10262239e-01 6.23615801e-01
-2.42800578e-01 5.87836683e-01 2.47690618e-01 4.50999588e-01
-9.99433756e-01 6.30394399e-01 9.20360386e-01 -6.84314489e-01
1.34551570e-01 -2.37769336e-01 -4.66874868e-01 1.01677573e+00
8.62090528e-01 -8.22029531e-01 -2.37601042e-01 -5.27275264e-01
-7.27334321e-02 -5.50115407e-01 7.00972974e-01 -8.14847231e-01
3.02916855e-01 -3.96222286e-02 1.31215349e-01 -7.94785380e-01
4.84161936e-02 -8.36910367e-01 -4.72939193e-01 5.15170097e-01
-3.03400546e-01 -2.88928211e-01 3.12039018e-01 8.85925531e-01
-1.93262637e-01 -3.45721275e-01 1.05962145e+00 2.60229539e-02
-1.12286103e+00 4.68391359e-01 -1.16776936e-01 3.25645000e-01
1.38679922e+00 -2.34213382e-01 -1.05188645e-01 2.63279885e-01
-6.89340651e-01 5.96570671e-01 3.63436013e-01 1.10223925e+00
6.26544595e-01 -1.28013933e+00 -8.74508619e-01 2.32995406e-01
1.00972176e+00 4.31270748e-01 -1.01644851e-01 6.90713108e-01
-4.08593386e-01 7.03806400e-01 1.78631008e-01 -1.23112142e+00
-1.70124853e+00 9.68464196e-01 3.85360241e-01 -5.39158620e-02
-7.72135437e-01 1.22949767e+00 9.31231022e-01 -5.61181307e-01
4.72985357e-01 -4.86923903e-01 4.70572785e-02 1.05173327e-01
5.98618507e-01 -2.47687951e-01 -4.11975831e-02 -7.36144423e-01
-5.52470624e-01 6.83855534e-01 -4.94511157e-01 4.51781124e-01
9.84993041e-01 -2.74978489e-01 -2.86606466e-03 5.02961695e-01
1.27896392e+00 -3.98813605e-01 -1.10046983e+00 -9.64726150e-01
3.32969755e-01 -5.40758193e-01 2.84908842e-02 -4.57765520e-01
-7.04125285e-01 8.62619281e-01 1.01932192e+00 -1.00016184e-01
7.07590342e-01 7.21773744e-01 5.54491341e-01 6.79048657e-01
3.26653898e-01 -1.02768672e+00 4.90886360e-01 4.35554296e-01
1.06773257e+00 -2.04371762e+00 -5.15012443e-02 -5.53682446e-01
-3.63757014e-01 9.20324028e-01 9.27581966e-01 -2.82167643e-02
6.93111241e-01 1.80262282e-01 1.55959651e-01 -4.23699975e-01
-6.23740494e-01 -5.93333364e-01 5.62357128e-01 4.68883991e-01
2.06821516e-01 8.38601068e-02 6.81869388e-02 3.95764738e-01
3.65638584e-02 -1.91803411e-01 2.38614306e-01 8.53497028e-01
-7.21080720e-01 -7.27241099e-01 -5.36831498e-01 5.68124294e-01
-1.84405386e-01 -2.61170924e-01 -2.47313693e-01 9.45543587e-01
1.34815559e-01 6.69492781e-01 4.56442744e-01 1.85404584e-01
2.23708615e-01 1.44953318e-02 3.08616012e-01 -1.27199101e+00
-1.40434399e-01 -7.95986429e-02 -2.19641671e-01 -3.49649489e-01
-3.27885091e-01 -7.92551041e-01 -1.18537664e+00 2.84228772e-01
-8.87536168e-01 -1.42270043e-01 5.14496505e-01 8.03495467e-01
2.14545518e-01 3.68563086e-01 3.68508905e-01 -6.18872225e-01
-8.04988086e-01 -8.96214128e-01 -3.91506135e-01 5.66956818e-01
4.11868900e-01 -7.92671740e-01 -7.95994550e-02 3.24003339e-01] | [9.653886795043945, 1.4865347146987915] |
5d77bd09-3bf5-4e57-9a1d-2dc339503d66 | scalable-gaussian-process-regression-enables | 2302.03294 | null | https://arxiv.org/abs/2302.03294v2 | https://arxiv.org/pdf/2302.03294v2.pdf | Linear-scaling kernels for protein sequences and small molecules outperform deep learning while providing uncertainty quantitation and improved interpretability | Gaussian process (GP) is a Bayesian model which provides several advantages for regression tasks in machine learning such as reliable quantitation of uncertainty and improved interpretability. Their adoption has been precluded by their excessive computational cost and by the difficulty in adapting them for analyzing sequences (e.g. amino acid and nucleotide sequences) and graphs (e.g. ones representing small molecules). In this study, we develop efficient and scalable approaches for fitting GP models as well as fast convolution kernels which scale linearly with graph or sequence size. We implement these improvements by building an open-source Python library called xGPR. We compare the performance of xGPR with the reported performance of various deep learning models on 20 benchmarks, including small molecule, protein sequence and tabular data. We show that xGRP achieves highly competitive performance with much shorter training time. Furthermore, we also develop new kernels for sequence and graph data and show that xGPR generally outperforms convolutional neural networks on predicting key properties of proteins and small molecules. Importantly, xGPR provides uncertainty information not available from typical deep learning models. Additionally, xGPR provides a representation of the input data that can be used for clustering and data visualization. These results demonstrate that xGPR provides a powerful and generic tool that can be broadly useful in protein engineering and drug discovery. | ['Wei Wang', 'Jonathan Parkinson'] | 2023-02-07 | null | null | null | null | ['protein-function-prediction', 'formation-energy'] | ['medical', 'miscellaneous'] | [-2.77024046e-05 -8.67805332e-02 7.20168557e-03 -3.01619053e-01
-6.83172047e-01 -6.72816217e-01 4.45847780e-01 6.98812187e-01
-3.56499553e-01 9.94738519e-01 -1.20012403e-01 -6.43773139e-01
-1.62077069e-01 -7.97743201e-01 -1.04648304e+00 -9.97361660e-01
-3.22527856e-01 8.09206009e-01 2.21062452e-01 2.11759150e-01
2.16348350e-01 6.75970018e-01 -9.92756546e-01 6.35555163e-02
9.87079799e-01 9.65278149e-01 2.31495515e-01 6.91177726e-01
-1.93011358e-01 3.33320647e-01 -4.36496347e-01 -5.09238958e-01
-1.79742083e-01 5.19730002e-02 -5.13619602e-01 -4.50273216e-01
1.59529716e-01 1.74929127e-01 -2.71128565e-01 1.01736760e+00
7.49151051e-01 1.36403486e-01 9.18619573e-01 -1.06100464e+00
-9.21389461e-01 5.83242357e-01 -6.47775710e-01 1.77034605e-02
2.26313964e-01 3.70615840e-01 9.93819118e-01 -7.90829957e-01
5.73437631e-01 1.43115306e+00 9.26607549e-01 2.93316483e-01
-1.77448034e+00 -4.50526059e-01 7.64736161e-02 2.86871761e-01
-1.32772124e+00 -8.17869976e-03 3.30428749e-01 -5.96270382e-01
1.37364590e+00 1.56618848e-01 3.09946120e-01 1.27139235e+00
6.83272839e-01 6.31376982e-01 9.34305072e-01 -3.66385244e-02
6.79829419e-01 -2.56681532e-01 4.34992522e-01 7.74978876e-01
2.01551169e-01 -5.87497512e-03 -5.31737685e-01 -7.29561985e-01
5.44044256e-01 2.16419250e-03 -3.95283759e-01 -5.08133888e-01
-1.06332421e+00 9.32332158e-01 3.35590392e-01 -2.39675060e-01
-4.12467688e-01 5.32395840e-01 3.17645431e-01 6.93766177e-02
6.95782840e-01 4.79197890e-01 -8.16613555e-01 -2.85726219e-01
-5.43765128e-01 2.51021028e-01 1.06820512e+00 8.96050215e-01
6.89207315e-01 -2.80314926e-02 -6.37849197e-02 6.93911850e-01
4.46813136e-01 2.80946642e-01 2.51166105e-01 -5.77185571e-01
6.45381585e-02 3.10143679e-01 1.09619856e-01 -9.55604911e-01
-7.39704788e-01 -2.86201477e-01 -7.16762900e-01 2.23016113e-01
5.11074007e-01 -2.00366955e-02 -1.20933175e+00 1.62051010e+00
2.59607852e-01 2.09650651e-01 7.00525614e-03 4.95396882e-01
1.15258276e+00 8.11252892e-01 4.02763486e-01 -1.33833975e-01
1.36312211e+00 -6.79016292e-01 -4.64653134e-01 1.56856533e-02
5.15579164e-01 -5.77317238e-01 9.41641808e-01 5.39849937e-01
-6.38118207e-01 -2.32838809e-01 -9.61889744e-01 -8.64618942e-02
-6.25315070e-01 3.32650021e-02 1.01959121e+00 6.83935404e-01
-9.96320069e-01 1.20945883e+00 -1.17534816e+00 -2.16683343e-01
7.52500653e-01 6.46338224e-01 -4.83662784e-01 -1.09717520e-02
-1.05753922e+00 7.97338903e-01 6.67441368e-01 6.37706229e-03
-7.66409636e-01 -9.52465415e-01 -8.07212114e-01 2.90715188e-01
1.96688578e-01 -6.21211648e-01 9.79673862e-01 -3.24743003e-01
-1.66549933e+00 5.24318278e-01 -5.18632792e-02 -7.71671057e-01
3.75200629e-01 -1.80635244e-01 -1.13382638e-01 -8.65508467e-02
-4.09969836e-01 5.42901993e-01 7.63499796e-01 -7.12265968e-01
-1.42118573e-01 -4.27034080e-01 -3.25355887e-01 -1.28303528e-01
1.46457866e-01 1.89203769e-01 -4.08968091e-01 -3.66484791e-01
-9.22609419e-02 -9.81084466e-01 -5.67421079e-01 -1.10660486e-01
-5.68278372e-01 -3.65924478e-01 5.82214475e-01 -7.91122794e-01
9.00794446e-01 -1.78824723e+00 3.09804946e-01 2.81094015e-01
6.06355965e-01 2.14363649e-01 -6.81373477e-02 5.53159654e-01
-2.98825622e-01 2.15294406e-01 -4.79795486e-01 -2.01386958e-03
2.29645744e-01 2.07076594e-01 -9.18833539e-02 5.89886606e-01
4.62953985e-01 1.29219139e+00 -8.78962815e-01 -1.19674489e-01
2.05860004e-01 9.15045261e-01 -3.84880245e-01 -9.88435838e-03
-7.08320856e-01 4.86936122e-01 -3.23805034e-01 6.15002751e-01
8.32833171e-01 -8.22153389e-01 2.02239677e-01 -2.53101379e-01
2.47004941e-01 2.90464699e-01 -8.63444746e-01 1.48308945e+00
2.18274109e-02 5.46194673e-01 -3.03085625e-01 -8.56336236e-01
9.52347517e-01 7.97937661e-02 3.71947944e-01 1.41589820e-01
4.68951650e-02 -3.80753949e-02 1.65547177e-01 -6.73918054e-02
1.05763949e-01 5.26279099e-02 2.52592355e-01 3.21049571e-01
3.20093334e-01 -2.42074043e-01 1.66003361e-01 1.31669626e-01
1.20363915e+00 4.11819994e-01 5.09627223e-01 -4.07517523e-01
1.58292741e-01 -9.84909236e-02 3.70395124e-01 6.24988735e-01
-4.16958481e-02 4.44601268e-01 8.91219616e-01 -5.42937875e-01
-8.54489982e-01 -1.16280389e+00 -2.73387492e-01 1.25920081e+00
-2.99562126e-01 -6.97036266e-01 -4.77380335e-01 -5.94488442e-01
3.68202120e-01 5.51793218e-01 -6.27654195e-01 -2.00449288e-01
-1.36061922e-01 -1.44751251e+00 3.62083644e-01 6.70358002e-01
-7.92518482e-02 -8.84155095e-01 7.73744807e-02 4.37514484e-01
2.49318406e-01 -8.88186991e-01 -2.63063759e-01 6.68524206e-01
-7.84268796e-01 -1.13011420e+00 -3.76560599e-01 -4.11230385e-01
4.11096901e-01 -2.53373653e-01 1.16340089e+00 -4.93413836e-01
-4.33671266e-01 1.81339532e-01 1.59854237e-02 -7.30619907e-01
-3.82134765e-01 -7.36238295e-03 9.90788564e-02 -4.15832043e-01
7.21705258e-01 -7.65103161e-01 -5.57513475e-01 5.17748557e-02
-7.99358785e-01 -1.31385595e-01 4.51009423e-01 9.73714709e-01
8.05364430e-01 -1.36566222e-01 5.43020844e-01 -1.01718652e+00
1.05236065e+00 -5.43171942e-01 -9.72732484e-01 3.01259130e-01
-7.47001469e-01 4.35776472e-01 4.83842194e-01 -4.36376274e-01
-7.58944690e-01 2.34570116e-01 -4.37611789e-01 -2.58045852e-01
-1.49917081e-01 8.22860122e-01 -4.18107621e-02 -2.47648329e-01
9.19153452e-01 1.17703848e-01 1.48046732e-01 -5.31061351e-01
5.54337144e-01 4.90649283e-01 2.98115373e-01 -7.39577651e-01
2.32461259e-01 2.39229932e-01 3.49092633e-01 -8.29479635e-01
-3.32250148e-01 -3.41484129e-01 -5.63943923e-01 2.77366042e-01
7.76194632e-01 -7.28901327e-01 -1.14093184e+00 4.02298868e-01
-1.09559298e+00 -2.70567864e-01 1.16307534e-01 4.50088024e-01
-6.68628812e-01 7.57064342e-01 -9.09237206e-01 -5.22416592e-01
-5.34443259e-01 -1.29105949e+00 1.09686375e+00 1.36900857e-01
-1.74486712e-01 -1.26282477e+00 1.34664848e-01 -5.52242063e-02
3.04955304e-01 5.75125277e-01 1.18372989e+00 -1.10903180e+00
-6.14493489e-01 -1.21875606e-01 -3.45602483e-01 2.29137480e-01
9.26038250e-02 3.47543925e-01 -9.87327099e-01 -3.03364158e-01
-3.41679156e-01 -3.77116442e-01 1.17411685e+00 7.75525987e-01
1.32645655e+00 -5.11985309e-02 -7.29498148e-01 8.80960345e-01
1.26527667e+00 4.86879349e-02 5.61411560e-01 4.36140299e-02
8.34325731e-01 4.08620805e-01 2.97462523e-01 3.24254781e-01
-2.86178607e-02 4.70462322e-01 4.64287966e-01 6.81779087e-02
3.68770063e-01 -1.07193217e-01 1.31684318e-01 3.50183040e-01
-1.85219809e-01 -3.30487162e-01 -1.18016005e+00 -1.07133284e-01
-2.27018857e+00 -6.40159726e-01 -4.47887391e-01 2.12174654e+00
1.05597103e+00 -1.19381443e-01 -1.14789069e-01 -3.95606726e-01
5.52165151e-01 -1.46768004e-01 -8.94508362e-01 -4.73370105e-01
-2.70914882e-01 6.66651309e-01 6.84781849e-01 4.09306794e-01
-1.38801026e+00 9.39380705e-01 7.14212465e+00 1.08529222e+00
-1.03369677e+00 -1.62072435e-01 7.65132010e-01 1.25970766e-01
-2.48479452e-02 -2.84721017e-01 -8.50501418e-01 5.22594273e-01
1.31959963e+00 -3.15339208e-01 3.89524192e-01 1.03029919e+00
1.12525016e-01 1.45162761e-01 -1.36937833e+00 1.00758600e+00
-4.11112309e-01 -1.69508457e+00 -1.59780458e-01 1.28027663e-01
5.88768780e-01 5.01612306e-01 1.42031014e-01 1.08771570e-01
1.01750588e+00 -1.48908615e+00 -2.89289579e-02 4.45296347e-01
5.60277402e-01 -9.97347295e-01 6.87274098e-01 2.20336877e-02
-7.16281831e-01 3.21092099e-01 -8.54352415e-01 1.99494809e-01
-7.36695379e-02 7.13112414e-01 -1.18760848e+00 5.55437863e-01
6.38815880e-01 7.83447981e-01 -4.20436561e-01 1.08167899e+00
-2.94179350e-01 6.43665373e-01 -4.36907709e-01 -1.70584634e-01
2.07388476e-01 -5.08799493e-01 3.83330703e-01 1.51183641e+00
1.60356700e-01 -3.18971008e-01 7.22625777e-02 1.14224660e+00
-1.45585984e-01 1.70454308e-01 -5.69678843e-01 -5.37258446e-01
2.78093964e-01 1.24165154e+00 -6.69234157e-01 -1.63306803e-01
-4.18670595e-01 7.00899661e-01 5.93762577e-01 6.24793291e-01
-9.74708676e-01 -4.02822822e-01 9.99014258e-01 -2.43270487e-01
5.51747501e-01 -4.39394385e-01 -6.07997887e-02 -9.33428109e-01
-4.75107849e-01 -8.47689569e-01 3.44799995e-01 -6.86999500e-01
-1.54887152e+00 5.41874349e-01 -2.01835245e-01 -5.22122383e-01
-2.79889107e-01 -1.30465794e+00 -2.38234863e-01 1.20176208e+00
-1.32693172e+00 -1.00876892e+00 2.59155706e-02 2.90853709e-01
3.38335298e-02 -8.04508105e-02 1.08992147e+00 -1.76975742e-01
-6.12137318e-01 3.17263514e-01 8.43246222e-01 -2.68653095e-01
7.56067991e-01 -1.67705953e+00 9.57616985e-01 4.49946761e-01
1.14701971e-01 9.80681896e-01 9.00583386e-01 -7.81539917e-01
-1.45135117e+00 -1.21408248e+00 2.64926791e-01 -6.29965425e-01
9.60573733e-01 -5.12417972e-01 -1.25178242e+00 7.00368702e-01
-9.23771337e-02 1.55259408e-02 1.16113794e+00 2.99878687e-01
-5.06841958e-01 3.45203817e-01 -1.06402135e+00 5.59353769e-01
7.75654137e-01 -4.06773776e-01 -2.99201667e-01 7.10152984e-01
9.10985053e-01 -5.00968575e-01 -1.18628824e+00 3.73491794e-01
5.28040469e-01 -7.01096058e-01 1.12677038e+00 -1.10387516e+00
3.07328552e-01 -3.75439346e-01 -2.67877504e-02 -1.62300074e+00
-5.97755551e-01 -7.03778923e-01 -5.68337619e-01 6.83749855e-01
6.55941069e-01 -8.10557783e-01 9.42855775e-01 8.23989272e-01
-1.68035328e-01 -9.25761700e-01 -7.08540797e-01 -7.91864812e-01
1.97925687e-01 -3.82434964e-01 5.87014496e-01 6.48497701e-01
5.58567829e-02 3.50445777e-01 -2.05156058e-01 2.16972336e-01
5.74064255e-01 1.21423736e-01 6.57287121e-01 -1.46962154e+00
-6.18631601e-01 -4.32746172e-01 -6.83002651e-01 -8.45240474e-01
2.46296018e-01 -9.13803697e-01 -4.13756967e-02 -1.59931076e+00
2.73109049e-01 -1.97956175e-01 -4.38456982e-01 4.67208922e-01
-4.67789233e-01 -4.74525243e-02 -4.28627282e-01 8.08971971e-02
-4.22376901e-01 4.88508493e-01 8.57894540e-01 -1.70645446e-01
-2.17829198e-01 5.88438176e-02 -6.89976573e-01 5.50187290e-01
9.49641943e-01 -2.78728962e-01 -3.02055001e-01 -6.04881113e-03
3.13955188e-01 -4.38920647e-01 1.88346788e-01 -5.53993762e-01
-1.80833824e-02 -1.05488263e-01 6.11759901e-01 -5.51624835e-01
3.74077916e-01 -4.65119094e-01 4.17937666e-01 4.01960850e-01
-4.39123660e-02 -1.39011471e-02 5.36882162e-01 1.18420386e+00
1.21700704e-01 -4.78079692e-02 7.41633713e-01 -9.82513279e-02
-4.18982446e-01 6.04906082e-01 -3.41406524e-01 -3.39496225e-01
9.37966287e-01 -7.33602196e-02 -5.36165595e-01 -2.62662977e-01
-9.49243069e-01 2.63434857e-01 5.68979204e-01 1.86521471e-01
6.35864675e-01 -9.57597792e-01 -5.93772352e-01 1.79545227e-02
2.23735049e-01 -7.07764104e-02 4.38193195e-02 7.83750892e-01
-9.53378677e-01 6.26377523e-01 -2.21378878e-01 -9.54284430e-01
-1.42837501e+00 8.93204391e-01 2.58754194e-01 -3.32819343e-01
-4.93057966e-01 1.04478502e+00 3.74677569e-01 -5.11759877e-01
2.75460035e-01 -5.80986857e-01 -3.06831077e-02 -2.99587101e-01
5.61933219e-01 2.02474624e-01 1.16564751e-01 -1.40176281e-01
-4.50342685e-01 1.74450278e-01 -2.77197391e-01 4.46983129e-01
1.73484290e+00 2.74880111e-01 -4.10767138e-01 3.64836663e-01
9.46543634e-01 -2.64576554e-01 -1.45526659e+00 -1.52577028e-01
3.44807446e-01 -1.46440864e-01 -1.29440397e-01 -8.01096618e-01
-5.07179856e-01 9.70423877e-01 5.22889137e-01 -3.10216262e-03
6.03959978e-01 1.41802147e-01 3.00855160e-01 6.44574940e-01
2.46599779e-01 -7.96623349e-01 -1.59954235e-01 5.33149600e-01
8.10169399e-01 -1.25132561e+00 1.20765969e-01 -4.93312061e-01
-4.62134778e-01 1.25787985e+00 2.47136250e-01 1.14373133e-01
7.11639702e-01 2.75263757e-01 -3.23818296e-01 -3.92071784e-01
-9.73653555e-01 -8.32023919e-02 4.30161208e-01 1.10132146e+00
7.82826960e-01 2.59588212e-01 -2.71166772e-01 6.11112595e-01
-1.10588502e-02 -1.06918380e-01 2.55694479e-01 7.05779970e-01
-5.72421014e-01 -1.25016773e+00 -1.17708221e-01 7.68503129e-01
-6.43729508e-01 -5.64329028e-01 -4.43486154e-01 4.88728911e-01
-3.42684180e-01 5.73531330e-01 -2.73344845e-01 -1.99443638e-01
-1.45089194e-01 3.61691862e-01 6.24698758e-01 -6.61715567e-01
-2.86682457e-01 7.06127211e-02 2.04303086e-01 -5.74068964e-01
-5.24248406e-02 -2.84768909e-01 -1.20477808e+00 -4.66253698e-01
-3.68222594e-01 2.59378791e-01 8.64640951e-01 5.64855635e-01
8.00574720e-01 5.78589320e-01 -1.88261151e-01 -6.68045819e-01
-6.38857782e-01 -1.04685843e+00 -6.48049295e-01 5.06662354e-02
2.37839874e-02 -6.54843569e-01 -7.48367757e-02 -6.16603121e-02] | [5.054610252380371, 5.607425212860107] |
f31adb28-b73e-4422-b635-8219bacca040 | conv-nilm-net-a-causal-and-multi-appliance | 2208.02173 | null | https://arxiv.org/abs/2208.02173v2 | https://arxiv.org/pdf/2208.02173v2.pdf | Conv-NILM-Net, a causal and multi-appliance model for energy source separation | Non-Intrusive Load Monitoring (NILM) seeks to save energy by estimating individual appliance power usage from a single aggregate measurement. Deep neural networks have become increasingly popular in attempting to solve NILM problems. However most used models are used for Load Identification rather than online Source Separation. Among source separation models, most use a single-task learning approach in which a neural network is trained exclusively for each appliance. This strategy is computationally expensive and ignores the fact that multiple appliances can be active simultaneously and dependencies between them. The rest of models are not causal, which is important for real-time application. Inspired by Convtas-Net, a model for speech separation, we propose Conv-NILM-net, a fully convolutional framework for end-to-end NILM. Conv-NILM-net is a causal model for multi appliance source separation. Our model is tested on two real datasets REDD and UK-DALE and clearly outperforms the state of the art while keeping a significantly smaller size than the competing models. | ['Simo Alami C.', 'Jesse Read', 'Rim Kaddah', 'Jérémie Decock'] | 2022-08-03 | null | null | null | null | ['non-intrusive-load-monitoring', 'non-intrusive-load-monitoring', 'speech-separation', 'non-intrusive-load-monitoring'] | ['knowledge-base', 'miscellaneous', 'speech', 'time-series'] | [ 2.58776903e-01 5.63242398e-02 -6.13791645e-01 -3.58652472e-01
-8.60924184e-01 -2.15752244e-01 5.21641076e-01 3.69839519e-02
1.90349475e-01 6.58107877e-01 3.99036556e-01 -3.16608436e-02
-1.22213811e-01 -5.24994791e-01 -5.81486523e-01 -6.65109694e-01
5.30686742e-03 5.43218791e-01 -2.46771157e-01 2.14935914e-01
-4.07183319e-01 1.22429766e-01 -1.28884947e+00 2.09221184e-01
4.73502100e-01 1.08811641e+00 1.82614341e-01 7.75362074e-01
-7.59800002e-02 1.60211658e+00 -9.40440238e-01 3.89391258e-02
-2.22280309e-01 -6.97181046e-01 -8.67877126e-01 -3.54610085e-02
5.98846599e-02 -5.27237177e-01 -4.42963660e-01 8.85573626e-01
7.04179645e-01 -3.29472601e-01 6.16645873e-01 -1.82390487e+00
-5.74732065e-01 1.43068135e+00 -8.45130205e-01 4.17841733e-01
4.30681497e-01 -2.75391471e-02 9.14560080e-01 1.54764310e-01
-3.03692669e-01 1.07458985e+00 7.35393107e-01 2.74008602e-01
-1.35305524e+00 -8.68311584e-01 1.25521138e-01 4.17587131e-01
-1.24136090e+00 -6.78858280e-01 1.10669076e+00 1.25597930e-02
1.47657156e+00 5.07818580e-01 2.63763815e-02 1.51807547e+00
1.75538778e-01 1.20450175e+00 8.23190153e-01 -1.59456819e-01
2.08901212e-01 -7.31592253e-02 3.21910501e-01 4.01185960e-01
-1.02340970e-02 -2.78606921e-01 -7.59515285e-01 -4.40222651e-01
1.53238222e-01 1.86928481e-01 -1.47927180e-01 1.96438104e-01
-9.58755076e-01 6.50924981e-01 2.01542750e-01 4.19695616e-01
-6.03431404e-01 5.73540151e-01 5.68174899e-01 -1.16072856e-02
6.52454793e-01 -2.16917276e-01 -6.16737902e-01 -3.69246989e-01
-1.24212766e+00 -1.37032107e-01 1.02791309e+00 6.92376435e-01
2.66388297e-01 3.85120094e-01 -2.06594139e-01 8.23117137e-01
4.31309044e-01 5.24433076e-01 7.36797631e-01 -5.25640309e-01
2.98753858e-01 6.72815382e-01 -3.39796782e-01 -3.80220562e-01
-8.13720405e-01 -2.10421458e-01 -1.18262339e+00 1.90704800e-02
-1.56066209e-01 -1.65373117e-01 -7.43040025e-01 1.81548727e+00
-4.51090634e-02 7.85881281e-01 -2.19205339e-02 4.03033078e-01
7.35182285e-01 6.20513499e-01 3.99647534e-01 -5.60421526e-01
1.28048885e+00 -9.82203186e-01 -9.50471401e-01 -1.92300573e-01
3.14478517e-01 -6.48063004e-01 6.29452348e-01 5.17047465e-01
-1.01134479e+00 -2.37105772e-01 -1.11251414e+00 -5.07467724e-02
-4.18271899e-01 2.55199730e-01 6.66938126e-01 5.32148004e-01
-8.36611748e-01 6.90415025e-01 -9.49335754e-01 -3.47902596e-01
8.00872743e-01 5.24860442e-01 1.16706602e-01 5.23273706e-01
-9.08571541e-01 5.11161327e-01 3.78817707e-01 1.26032680e-01
-1.19398880e+00 -7.19020069e-01 -7.01953828e-01 2.62539566e-01
4.03003216e-01 -4.85087872e-01 1.42978311e+00 -9.47771490e-01
-1.71714509e+00 4.73380566e-01 -1.01700589e-01 -8.23062420e-01
3.93005520e-01 -3.73227298e-01 -8.01193774e-01 -3.55728835e-01
2.31761664e-01 1.86714500e-01 1.00077701e+00 -1.04723978e+00
-7.24353194e-01 -3.57924819e-01 -1.77558348e-01 -2.31720939e-01
-4.20418650e-01 1.99888572e-02 -1.05473399e-01 -4.30359036e-01
-6.10012233e-01 -8.20582390e-01 3.30095232e-01 -8.04153979e-01
-9.86815393e-01 -7.88619101e-01 1.42770791e+00 -6.23281956e-01
1.23459303e+00 -2.00786996e+00 -1.93715110e-01 -2.67469466e-01
3.69993001e-01 3.26475918e-01 -1.72958404e-01 3.10775965e-01
-5.30811489e-01 -1.87808767e-01 -1.94474414e-01 -1.02552974e+00
3.99995565e-01 1.91633791e-01 -3.61121297e-01 6.51008248e-01
-6.39442429e-02 1.30813479e+00 -8.39761376e-01 -3.62676561e-01
7.52595186e-01 7.07076609e-01 1.47852182e-01 4.37933385e-01
-4.45681334e-01 1.18882604e-01 -7.05671310e-03 6.81124747e-01
6.80950105e-01 -3.83901149e-01 4.00534272e-01 -3.67011040e-01
1.46132022e-01 5.75947940e-01 -1.00750589e+00 1.89812505e+00
-9.20482755e-01 7.78685927e-01 -4.72910106e-02 -1.07187879e+00
3.30964029e-01 6.28801703e-01 1.10075796e+00 -7.63204396e-01
4.30181444e-01 -5.21876365e-02 -2.84096122e-01 -3.02258998e-01
1.60634890e-01 2.66163319e-01 -1.48255631e-01 7.27037132e-01
1.94677114e-01 4.59443569e-01 2.95608249e-02 8.59832615e-02
1.39813721e+00 1.11188479e-02 3.80017817e-01 -4.01324093e-01
2.14085445e-01 -5.55075407e-01 6.20390832e-01 4.29583818e-01
-1.62178189e-01 2.41242617e-01 3.96876961e-01 -2.62274891e-01
-4.00102973e-01 -1.06898451e+00 6.08225130e-02 1.19679523e+00
1.97235327e-02 -4.93789613e-01 -7.93136358e-01 -9.31818724e-01
5.31715937e-02 1.14502800e+00 -6.71659410e-01 -1.97446048e-01
-5.57164013e-01 -1.41010571e+00 7.04526782e-01 6.91944599e-01
4.71199483e-01 -1.26168418e+00 -7.35761225e-01 5.45651674e-01
-3.93279552e-01 -1.16379726e+00 -5.53993702e-01 8.07408094e-01
-4.53522205e-01 -1.33425283e+00 -1.68735653e-01 -3.61859471e-01
-4.88753729e-02 2.25004911e-01 1.73588026e+00 -3.55656147e-01
-4.40321505e-01 2.07051650e-01 6.17949516e-02 -9.19176638e-01
-5.13719201e-01 4.93555129e-01 -5.59267327e-02 1.22724988e-01
1.00406039e+00 -1.27192247e+00 -6.04778767e-01 1.60701901e-01
-6.87195361e-01 -2.38479629e-01 4.53507066e-01 6.33286536e-02
3.13840657e-01 5.60074627e-01 9.46077049e-01 -8.87381494e-01
7.41629899e-01 -9.64373171e-01 -4.30365384e-01 2.06775174e-01
-7.57781982e-01 -1.11031413e-01 7.92306364e-01 -4.97731805e-01
-8.71952593e-01 1.04709096e-01 -1.49104401e-01 -6.03728235e-01
-4.08989668e-01 -2.45812282e-01 -4.20589924e-01 8.22459459e-01
1.63632080e-01 1.58851013e-01 -6.18063450e-01 -7.24310279e-01
2.54295379e-01 8.20380449e-01 7.37599850e-01 -5.23250438e-02
2.83196002e-01 4.94340658e-01 -3.26934420e-02 -8.89757633e-01
-8.47986400e-01 -4.30141628e-01 -4.62715298e-01 1.35969132e-01
1.11093152e+00 -9.29092824e-01 -1.45488572e+00 8.48066628e-01
-1.23616552e+00 -6.51466012e-01 -4.23643827e-01 3.38218808e-02
-4.03639317e-01 4.22100797e-02 -5.74077845e-01 -1.12065446e+00
-6.48097456e-01 -9.65465009e-01 1.56432188e+00 1.80703551e-01
-4.96892869e-01 -1.07230389e+00 1.46130249e-01 2.84777999e-01
3.86974543e-01 3.60199332e-01 1.05958939e+00 -6.57162189e-01
-9.98079553e-02 -9.65824649e-02 -2.41130233e-01 2.89618790e-01
7.48444617e-01 -2.57421613e-01 -1.56386149e+00 -5.88375181e-02
3.05541873e-01 -3.00701648e-01 9.25813675e-01 6.17483318e-01
1.53686082e+00 -2.71627486e-01 -6.29279315e-01 3.65583122e-01
1.33205426e+00 2.16583803e-01 5.23559451e-01 -2.77516395e-01
1.07725370e+00 2.60523576e-02 -4.97664183e-01 3.78971219e-01
7.41293490e-01 4.49220419e-01 7.36463249e-01 -3.90736818e-01
-3.55107039e-01 -1.61466077e-01 5.36527097e-01 8.53718758e-01
4.43354279e-01 -8.55807126e-01 -3.79591852e-01 6.31919205e-01
-2.27415323e+00 -8.63169253e-01 -2.66845167e-01 1.81502831e+00
6.63553357e-01 1.56707257e-01 5.46456933e-01 5.87811410e-01
3.61174703e-01 5.38359463e-01 -9.02210951e-01 -2.66349643e-01
-6.17810041e-02 1.19207129e-01 6.34526432e-01 3.30271184e-01
-1.25927711e+00 1.22584626e-01 6.11503792e+00 9.51770186e-01
-8.89322460e-01 6.64101541e-01 6.72239125e-01 -4.46421444e-01
9.86552313e-02 -6.33163691e-01 -5.84688365e-01 9.11950469e-01
1.46862090e+00 2.69335181e-01 5.48656404e-01 8.46700907e-01
3.06082517e-01 -4.06527132e-01 -1.58184195e+00 1.35346794e+00
1.94848269e-01 -8.69239867e-01 -5.51155627e-01 5.70964888e-02
5.24686754e-01 4.34833854e-01 -3.30282897e-01 3.09977591e-01
7.76700854e-01 -1.06831360e+00 4.63314742e-01 3.98296088e-01
3.69160235e-01 -8.71241271e-01 6.64342403e-01 3.33787799e-01
-1.44987881e+00 -2.33658150e-01 4.22505513e-02 1.65417984e-01
3.44730765e-01 1.01887012e+00 -5.39696932e-01 5.77510953e-01
7.56651938e-01 7.93509901e-01 -4.05612826e-01 2.23319605e-01
-3.36998925e-02 9.56908107e-01 -5.43094695e-01 2.99619317e-01
-1.28586292e-01 2.49124691e-01 2.50455827e-01 1.31757343e+00
-1.40891403e-01 -4.39007849e-01 2.60898322e-01 1.01936615e+00
-4.06291813e-01 -4.48002309e-01 -4.82330918e-01 -2.30724993e-03
1.03765830e-01 1.45309591e+00 -7.73789346e-01 -3.34143400e-01
-4.04612660e-01 1.49625444e+00 5.07719591e-02 4.75660712e-03
-1.26976216e+00 -4.73010242e-02 8.44842374e-01 -1.20421834e-01
9.91384089e-02 1.53417736e-01 -2.17166021e-01 -9.52227175e-01
-1.51861578e-01 -6.45852685e-01 4.03764039e-01 -6.19757831e-01
-1.52924597e+00 5.53358555e-01 8.38896409e-02 -6.65869296e-01
-5.45856178e-01 -1.23953305e-01 -9.59681213e-01 8.00385237e-01
-1.65557051e+00 -1.37639678e+00 -3.61036539e-01 7.74482727e-01
1.00250328e+00 -4.33909819e-02 9.80698228e-01 7.47236848e-01
-1.03029716e+00 4.95848656e-01 -1.69562902e-02 1.59515604e-01
3.93544525e-01 -1.62195706e+00 5.34968615e-01 6.70120239e-01
4.48316604e-01 -9.32102054e-02 5.01853704e-01 -5.69911540e-01
-1.33904016e+00 -1.49561894e+00 8.35060358e-01 -7.56700516e-01
5.29958010e-01 -7.24449813e-01 -6.28125012e-01 6.12040043e-01
9.27582383e-01 -2.34412640e-01 8.73654604e-01 1.24998003e-01
-1.67383432e-01 -4.44946498e-01 -1.07586598e+00 2.55163051e-02
8.25598776e-01 -5.77727675e-01 -1.88865840e-01 5.73065639e-01
6.72995746e-01 7.69472420e-02 -7.06383705e-01 1.46065131e-01
2.92116374e-01 -1.01740980e+00 7.76637137e-01 -1.21197440e-01
9.33647975e-02 -3.32779765e-01 -2.92601019e-01 -1.27554703e+00
-1.64051041e-01 -9.32362795e-01 -9.72553492e-01 1.84334946e+00
1.63857594e-01 -4.49008405e-01 5.30663490e-01 3.73011559e-01
1.41010076e-01 -3.32534730e-01 -7.49217033e-01 -6.67185187e-01
-4.27836686e-01 -8.68986070e-01 1.14260590e+00 7.68469214e-01
-2.43259281e-01 9.01580215e-01 -6.07229531e-01 3.31880718e-01
9.00231600e-01 1.74735770e-01 4.12736446e-01 -1.23784816e+00
-4.55445558e-01 -4.58545208e-01 -2.23794170e-02 -6.47376001e-01
5.95032811e-01 -8.13970506e-01 8.44179615e-02 -1.45465147e+00
3.47572982e-01 -3.22331153e-02 -6.99815750e-01 8.39077175e-01
4.25840020e-02 5.53880870e-01 1.10258035e-01 2.01020837e-01
-6.23142481e-01 4.24296647e-01 8.71149972e-02 -5.58640122e-01
-4.57831323e-01 9.20554474e-02 -7.39848852e-01 9.73201334e-01
1.06514764e+00 -6.83341324e-01 -8.62748265e-01 -5.15122175e-01
4.59988043e-02 -6.48035407e-02 4.94442016e-01 -9.90138054e-01
9.69459191e-02 1.37879774e-01 4.27462995e-01 -9.50004935e-01
2.21728727e-01 -1.08208406e+00 3.29083383e-01 1.81518525e-01
-1.37777835e-01 1.11638568e-02 2.12693915e-01 5.90603173e-01
2.41983756e-01 -5.35321934e-03 4.92477626e-01 3.99485379e-02
-4.35255468e-01 2.65588045e-01 -1.55345887e-01 -2.47066990e-01
8.89318824e-01 4.05712903e-01 -3.60922247e-01 -2.39673540e-01
-3.99535179e-01 2.22113803e-01 -3.16685051e-01 8.05592060e-01
-1.11271188e-01 -1.48033583e+00 -6.13694847e-01 5.03886072e-03
-1.27406642e-01 -4.23253030e-02 1.68603644e-01 8.54565740e-01
2.29831785e-01 4.57414001e-01 4.61380124e-01 -6.78200483e-01
-1.15881157e+00 8.96389782e-01 4.06952351e-01 -5.67495644e-01
-5.72183728e-01 3.54809195e-01 2.19013304e-01 2.97150705e-02
4.54691947e-01 -6.46551967e-01 -6.39418587e-02 2.11950138e-01
6.37121320e-01 8.49727750e-01 4.12066370e-01 -5.13327599e-01
-5.96695900e-01 1.56634063e-01 1.11190945e-01 3.73192817e-01
1.56601501e+00 -3.11333358e-01 -2.97927320e-01 8.51781666e-01
1.45287848e+00 -2.24455044e-01 -9.95541513e-01 -3.79427262e-02
3.72594334e-02 1.99328229e-01 5.34254909e-01 -1.07404149e+00
-1.48753691e+00 6.67738080e-01 8.86074960e-01 8.75824630e-01
1.53180635e+00 7.90747404e-02 1.20931482e+00 -1.41972244e-01
4.95804921e-02 -1.09424806e+00 -1.83535948e-01 -1.41157374e-01
6.61248624e-01 -1.47249174e+00 -1.08778909e-01 2.82335598e-02
-1.41331807e-01 8.42601359e-01 4.12427753e-01 1.62082780e-02
9.84974682e-01 8.41682196e-01 -1.67321816e-01 -3.15785885e-01
-7.43668199e-01 -2.14730904e-01 8.62053502e-03 6.08608127e-01
4.65083241e-01 3.92260790e-01 2.77178109e-01 7.85064697e-01
2.33289897e-02 3.47630441e-01 1.80165559e-01 6.93722844e-01
3.44891727e-01 -1.02971947e+00 -1.87284917e-01 6.74526751e-01
-7.30122447e-01 -5.38387522e-02 -2.82659888e-01 3.72782826e-01
3.37024063e-01 1.57631898e+00 1.95522383e-01 -3.56923908e-01
2.85912752e-01 1.81620911e-01 2.94253290e-01 -2.31565610e-01
-6.28796816e-01 3.36720556e-01 -1.12694517e-01 -7.03215361e-01
-8.12196732e-01 -4.99051899e-01 -9.41647053e-01 -6.23201370e-01
-2.50674576e-01 -1.97362304e-01 6.07915342e-01 1.05497837e+00
3.01832289e-01 1.39956689e+00 9.16492283e-01 -8.43991220e-01
-1.57304347e-01 -1.17662060e+00 -7.88055122e-01 1.79184645e-01
6.67981446e-01 -3.36219907e-01 -7.26263896e-02 1.40439227e-01] | [16.07076644897461, 7.582086086273193] |
748731c1-fa88-4eb7-8ace-6936bc35f452 | spectral-goodness-of-fit-tests-for-complete | 2106.09702 | null | https://arxiv.org/abs/2106.09702v1 | https://arxiv.org/pdf/2106.09702v1.pdf | Spectral goodness-of-fit tests for complete and partial network data | Networks describe the, often complex, relationships between individual actors. In this work, we address the question of how to determine whether a parametric model, such as a stochastic block model or latent space model, fits a dataset well and will extrapolate to similar data. We use recent results in random matrix theory to derive a general goodness-of-fit test for dyadic data. We show that our method, when applied to a specific model of interest, provides an straightforward, computationally fast way of selecting parameters in a number of commonly used network models. For example, we show how to select the dimension of the latent space in latent space models. Unlike other network goodness-of-fit methods, our general approach does not require simulating from a candidate parametric model, which can be cumbersome with large graphs, and eliminates the need to choose a particular set of statistics on the graph for comparison. It also allows us to perform goodness-of-fit tests on partial network data, such as Aggregated Relational Data. We show with simulations that our method performs well in many situations of interest. We analyze several empirically relevant networks and show that our method leads to improved community detection algorithms. R code to implement our method is available on Github. | ['Tyler H. McCormick', 'Bolun Liu', 'Shane Lubold'] | 2021-06-17 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 1.51748091e-01 -4.26213443e-02 -1.37464345e-01 -2.60590583e-01
-2.61879086e-01 -7.02174485e-01 6.72052145e-01 2.30841309e-01
-9.81957316e-02 5.87763667e-01 1.21065147e-01 -6.25720561e-01
-8.34376872e-01 -1.03386283e+00 -5.22546232e-01 -6.43105567e-01
-5.80102921e-01 7.87367344e-01 4.22204882e-01 6.00745492e-02
1.63277745e-01 5.70885360e-01 -9.33805704e-01 -4.29884866e-02
7.72393703e-01 3.50645959e-01 -1.21069603e-01 6.60236418e-01
2.30187699e-01 5.22866726e-01 -3.37671220e-01 -5.40735841e-01
2.83412457e-01 -6.53208196e-01 -8.89608264e-01 2.69260019e-01
3.28624211e-02 -8.48880559e-02 -3.28691125e-01 8.46154273e-01
3.46022666e-01 -2.34344620e-02 8.98387015e-01 -1.44734049e+00
-1.64348215e-01 8.50517809e-01 -5.48246682e-01 3.41096133e-01
5.04669905e-01 -7.81985819e-02 1.24787354e+00 -5.33193886e-01
7.34391987e-01 1.39050853e+00 8.32868397e-01 -1.89211622e-01
-1.93569064e+00 -5.86867750e-01 -7.58785009e-02 -1.49327263e-01
-1.53898394e+00 -4.28693473e-01 7.28612065e-01 -5.74375689e-01
6.00437403e-01 2.08408281e-01 8.08336258e-01 8.08475256e-01
-1.63791422e-02 4.20379162e-01 1.14493585e+00 -4.28727955e-01
1.00286633e-01 -1.33506253e-01 2.43566617e-01 5.90520322e-01
6.45719409e-01 -1.24626800e-01 -1.30332321e-01 -8.92561495e-01
1.02545094e+00 9.65181366e-02 -6.39446005e-02 -8.17935884e-01
-1.32471824e+00 1.08484793e+00 1.74667269e-01 4.84370112e-01
-3.84928018e-01 2.97568053e-01 1.35885566e-01 4.49644744e-01
4.58217442e-01 1.98096678e-01 2.53928341e-02 -7.79949203e-02
-1.00618887e+00 4.80128750e-02 1.30207181e+00 6.81993663e-01
8.26837659e-01 -3.19862276e-01 2.14840800e-01 7.41943717e-01
2.84086049e-01 2.07236484e-01 -2.39640653e-01 -1.12076974e+00
3.62362951e-01 4.50962484e-01 1.74057391e-02 -1.28671539e+00
-3.93759459e-01 -4.84770060e-01 -9.01514530e-01 -1.27753541e-01
8.86813521e-01 -1.16866596e-01 -3.67558002e-01 1.71542299e+00
1.40766442e-01 1.90357193e-01 -2.49961615e-01 3.41711611e-01
3.30461740e-01 5.44180214e-01 -2.52061397e-01 -6.21614218e-01
9.21026170e-01 -5.18713713e-01 -4.28902328e-01 1.97870046e-01
6.39331043e-01 -7.67411709e-01 7.54437447e-01 2.65498459e-01
-1.17362416e+00 -7.96324983e-02 -6.04613006e-01 4.99998778e-01
3.24898474e-02 -4.12762076e-01 8.80920708e-01 6.93071008e-01
-1.38528430e+00 8.81474793e-01 -8.66217792e-01 -9.71700549e-01
4.20863647e-03 5.49433827e-01 -4.26576793e-01 -1.53159305e-01
-9.94974136e-01 7.02346087e-01 3.47525738e-02 -2.08643973e-02
-6.70149446e-01 -1.27025798e-01 -5.26569903e-01 2.58148491e-01
6.67841077e-01 -8.06181729e-01 7.60007799e-01 -9.53769386e-01
-1.12831736e+00 5.09779453e-01 -1.58159584e-01 -2.35850170e-01
5.28486788e-01 4.19674665e-01 -2.00706989e-01 2.47844502e-01
1.13841400e-01 1.56328589e-01 5.24210632e-01 -1.12098169e+00
2.46905625e-01 -5.55643700e-02 1.66295379e-01 9.36642811e-02
-1.87833145e-01 3.95157069e-01 -3.82975012e-01 -3.76177430e-01
4.21479315e-01 -1.17239261e+00 -5.69466472e-01 -1.39762267e-01
-4.58386064e-01 -1.59617811e-01 3.33143115e-01 -1.52915701e-01
1.31167567e+00 -1.89731753e+00 2.72676915e-01 1.05790925e+00
4.90959495e-01 -2.89724469e-01 -1.08938254e-01 1.25221956e+00
-2.50684202e-01 4.50313628e-01 -2.92551428e-01 -1.01831034e-01
-7.89098218e-02 1.91335008e-01 1.97793946e-01 8.08335125e-01
-2.20747620e-01 7.12804735e-01 -8.07013690e-01 -6.31626368e-01
5.63809648e-03 2.61641830e-01 -5.63750505e-01 -2.42726505e-01
3.39531094e-01 1.82368129e-01 -2.94960290e-01 2.96015650e-01
4.17395502e-01 -8.18067908e-01 7.55587399e-01 2.22850218e-01
1.24241926e-01 2.93810308e-01 -1.53660226e+00 9.90305185e-01
-1.73256230e-02 5.73172331e-01 1.41280234e-01 -1.13037109e+00
8.07771146e-01 2.82644629e-01 6.13308668e-01 6.76312447e-02
4.86523919e-02 1.56612620e-01 4.87437874e-01 -1.63824990e-01
-6.23412319e-02 -1.93747029e-01 -3.32333818e-02 1.04311395e+00
-1.67477608e-01 4.97264080e-02 4.37667489e-01 5.82678020e-01
1.57222331e+00 -3.91755879e-01 6.31346345e-01 -5.42810380e-01
1.54485375e-01 -2.01239586e-01 3.65714461e-01 8.72879326e-01
-6.08039573e-02 1.77043125e-01 1.08892453e+00 -1.26827106e-01
-1.20276833e+00 -9.81727839e-01 -8.73689130e-02 6.88314021e-01
-7.89202526e-02 -5.62732756e-01 -5.65211594e-01 -3.13930273e-01
9.64865312e-02 1.84451967e-01 -5.03609121e-01 2.28916377e-01
-5.12084126e-01 -9.27236021e-01 3.68708909e-01 1.34963095e-01
-5.11000231e-02 -7.89094985e-01 1.77609757e-01 1.66848913e-01
-1.23844832e-01 -8.37896407e-01 -3.17510635e-01 1.00341767e-01
-1.17041159e+00 -1.17611098e+00 -4.47687477e-01 -4.70261306e-01
8.29728842e-01 5.32778740e-01 1.11024880e+00 3.70693058e-01
5.55409603e-02 4.62703586e-01 -1.78550556e-01 4.78578895e-01
-5.68298101e-01 1.61264583e-01 1.25339791e-01 2.39786878e-02
2.93429106e-01 -9.62116301e-01 -3.97945195e-01 7.22663641e-01
-8.50872219e-01 -8.21135044e-02 4.33068871e-01 7.21857429e-01
2.06455752e-01 3.58089268e-01 4.04251218e-01 -1.11077154e+00
1.03895390e+00 -8.38298738e-01 -6.43392384e-01 2.66786635e-01
-6.61666870e-01 -5.10053560e-02 4.27999973e-01 -4.47006881e-01
-3.14234972e-01 3.57516520e-02 3.78081650e-01 -7.50795081e-02
1.69215709e-01 9.73026216e-01 -4.69874553e-02 -1.64746776e-01
5.84580183e-01 -2.34678879e-01 2.77839690e-01 -3.34670335e-01
2.56693065e-01 4.22135741e-01 -2.02822894e-01 -5.70242345e-01
9.52859163e-01 5.05782068e-01 4.41698402e-01 -8.10098648e-01
-1.22670338e-01 -6.21798038e-01 -8.68483186e-01 -2.26424202e-01
2.93500602e-01 -6.32905424e-01 -7.88610399e-01 9.48056579e-02
-8.52699637e-01 -3.88558835e-01 9.76123586e-02 5.97183049e-01
-6.60654187e-01 6.48504853e-01 -7.19898522e-01 -9.43490326e-01
2.65249938e-01 -9.57228184e-01 4.44779277e-01 -2.84365714e-01
-5.07445574e-01 -1.47054565e+00 4.07155156e-01 7.15401396e-02
1.55379638e-01 2.74706900e-01 7.78589845e-01 -6.93722129e-01
-7.61575222e-01 -2.64295340e-01 -1.96758464e-01 -1.02581389e-01
1.18263669e-01 5.85581779e-01 -3.30818266e-01 -5.23139417e-01
-2.71947473e-01 1.07006304e-01 7.85607517e-01 7.22258985e-01
6.85583413e-01 -3.94521892e-01 -6.98647380e-01 3.69984895e-01
1.31728947e+00 -1.96290426e-02 5.49430490e-01 -1.11704454e-01
5.52807271e-01 9.40946043e-01 2.46927679e-01 2.70527840e-01
3.98138523e-01 7.36420512e-01 1.28406689e-01 -1.74255073e-01
3.89207065e-01 -2.57616341e-01 1.19608827e-01 1.00381637e+00
-2.06535459e-01 -4.04517621e-01 -1.21537566e+00 5.52167594e-01
-2.13914204e+00 -1.26095057e+00 -5.38949788e-01 2.23902130e+00
5.51400781e-01 1.83922589e-01 7.83888996e-01 1.95539087e-01
1.10589063e+00 -3.31981368e-02 -9.49924514e-02 -1.22457132e-01
-1.16106197e-01 -4.77474444e-02 6.42012537e-01 8.54735911e-01
-7.36315370e-01 7.83802867e-01 7.53855181e+00 7.75997937e-01
-5.33434510e-01 9.21322480e-02 4.85328704e-01 4.42519747e-02
-5.23013175e-01 6.88470006e-01 -3.55989337e-01 2.56804764e-01
1.15155208e+00 -3.80289227e-01 5.87285042e-01 4.76259112e-01
5.40790558e-01 -2.50964105e-01 -1.12691987e+00 6.78114176e-01
-2.60682583e-01 -1.05710113e+00 -3.51602077e-01 6.70463979e-01
7.04424798e-01 -1.09762162e-01 -2.90669441e-01 -2.24841923e-01
8.26211929e-01 -1.16278183e+00 1.74785182e-01 6.08280122e-01
4.91572171e-01 -6.30005240e-01 4.18789715e-01 3.59313577e-01
-1.10586655e+00 -9.15550813e-03 -4.56824809e-01 -3.56060296e-01
2.01507196e-01 8.34800780e-01 -9.46509302e-01 3.59722167e-01
3.71405125e-01 7.25247264e-01 -5.42485654e-01 1.36524856e+00
-2.18895078e-02 8.21769595e-01 -6.74689114e-01 -1.46540970e-01
7.72770541e-03 -3.90618503e-01 6.72536373e-01 9.90966260e-01
3.51230413e-01 -1.02111958e-01 1.12831317e-01 7.35069513e-01
2.11232901e-01 2.56821394e-01 -1.01920223e+00 -3.36590618e-01
7.03761578e-01 1.17286384e+00 -1.29623175e+00 -2.00978652e-01
-3.03381324e-01 5.24771333e-01 3.91185284e-01 5.16329765e-01
-4.03524935e-01 -2.30679616e-01 1.22955553e-01 6.26817226e-01
2.27599427e-01 -6.35615587e-01 1.25228288e-02 -1.20564890e+00
-2.17885643e-01 -8.71872723e-01 5.15175223e-01 -5.69097936e-01
-1.50905502e+00 3.86332929e-01 5.00712097e-01 -1.10846806e+00
-5.82195401e-01 -2.37595335e-01 -6.70527577e-01 7.32462764e-01
-7.84571767e-01 -8.02027822e-01 -1.18086256e-01 6.03199005e-01
-1.76498950e-01 7.26993084e-02 5.19415557e-01 1.27291396e-01
-4.79233563e-01 1.11190371e-01 3.86071116e-01 8.54217634e-02
6.06996000e-01 -1.17567968e+00 3.77949327e-01 9.37586010e-01
2.34387442e-01 9.98401701e-01 9.57170546e-01 -8.71753812e-01
-9.64806855e-01 -4.28714603e-01 9.58419323e-01 -3.28160346e-01
1.11203957e+00 -5.70684850e-01 -7.57219732e-01 8.91776919e-01
-1.42929924e-03 -4.60086837e-02 9.16032791e-01 6.37408972e-01
-1.50969297e-01 -3.36872377e-02 -7.90408611e-01 6.42075360e-01
1.29871213e+00 -4.11254674e-01 -1.34677976e-01 4.92994875e-01
3.29969555e-01 3.50024372e-01 -1.11186671e+00 1.95122987e-01
5.99997640e-01 -1.13654602e+00 9.07307506e-01 -4.89975631e-01
1.25636578e-01 -1.85614929e-01 -1.06748112e-01 -1.14054251e+00
-6.88616693e-01 -8.93774927e-01 4.79127690e-02 1.18182111e+00
4.61725503e-01 -8.70656490e-01 6.88746333e-01 4.94908869e-01
5.54851949e-01 -5.43311775e-01 -9.01311815e-01 -8.89143765e-01
-2.59924438e-02 -1.64231032e-01 5.33797562e-01 1.21320832e+00
1.99804187e-01 4.79314983e-01 -2.97543854e-01 4.02650004e-03
8.43447030e-01 2.39377469e-01 1.09078884e+00 -1.70543206e+00
-6.42741024e-01 -4.85268652e-01 -4.22312975e-01 -9.97835577e-01
1.27837017e-01 -8.26823056e-01 -5.37870288e-01 -1.77429318e+00
5.18717349e-01 -8.19830179e-01 -4.22869362e-02 2.96648145e-01
5.02918735e-02 1.84525266e-01 9.57156047e-02 6.92313552e-01
-5.55366635e-01 1.60051644e-01 1.02016425e+00 2.80156821e-01
-3.24622273e-01 3.47083241e-01 -7.24959671e-01 5.57780683e-01
6.38154209e-01 -8.10457170e-01 -5.15120625e-01 2.35840544e-01
6.95193648e-01 6.14915788e-01 5.60296297e-01 -5.19335985e-01
3.07885498e-01 -2.51733035e-01 8.32246616e-02 -4.18861777e-01
2.31557623e-01 -8.27390850e-01 8.31910312e-01 4.69740301e-01
-1.71117753e-01 2.70810485e-01 -3.16137314e-01 7.06754982e-01
2.29891054e-02 -5.38752556e-01 4.52838451e-01 -9.74135175e-02
2.06130221e-01 1.72448680e-01 -4.17553574e-01 -2.12958485e-01
9.58464980e-01 -4.46972817e-01 -3.55827332e-01 -1.04585195e+00
-9.09498215e-01 1.49030492e-01 8.61593962e-01 -1.91537514e-01
2.02800274e-01 -1.24846363e+00 -7.29879141e-01 -1.50140941e-01
-1.93070158e-01 -5.99639475e-01 -1.55813992e-01 1.19187105e+00
-5.85557044e-01 1.97520927e-01 -6.81551024e-02 -6.83876753e-01
-1.42211616e+00 7.19088435e-01 2.58830547e-01 -4.26894814e-01
-3.90907198e-01 3.56089205e-01 1.09522395e-01 -2.59388626e-01
-2.55952120e-01 -3.84493060e-02 -7.70757422e-02 2.33441487e-01
1.40233627e-02 4.93697315e-01 -4.22507107e-01 -7.10996091e-01
-4.49137866e-01 5.47878325e-01 2.18915477e-01 -3.14158529e-01
1.45349932e+00 -3.67298663e-01 -6.13115549e-01 6.69366539e-01
1.07677472e+00 3.04348737e-01 -1.01278365e+00 -2.89126426e-01
1.39499858e-01 -5.59967995e-01 -2.46184036e-01 -1.02835402e-01
-8.16402912e-01 5.60158491e-01 -1.40725777e-01 1.10201991e+00
8.38145196e-01 2.20785633e-01 1.83474794e-02 3.08694959e-01
3.63531947e-01 -5.61567664e-01 -6.55326918e-02 1.90965146e-01
5.77969015e-01 -9.47527289e-01 4.77495521e-01 -7.33980715e-01
-2.75603026e-01 1.03014505e+00 2.39188932e-02 -3.55513543e-01
9.49581802e-01 1.33557795e-02 -4.20883179e-01 -4.90188003e-01
-8.93849552e-01 -1.83798119e-01 2.02260222e-02 4.21099693e-01
4.44104016e-01 1.44076869e-01 -5.32996893e-01 -5.19327149e-02
-3.02043945e-01 -3.46440852e-01 9.78262424e-01 7.23887980e-01
-2.99160838e-01 -1.48200715e+00 -3.26047301e-01 6.92163825e-01
-3.47537190e-01 -1.82514966e-01 -5.94347358e-01 8.76257002e-01
-3.93876106e-01 1.03237414e+00 6.50211573e-02 -1.37502789e-01
-9.07869488e-02 7.92875141e-03 5.15728951e-01 -6.56220198e-01
-4.24592972e-01 5.16721070e-01 4.69287723e-01 -2.83989727e-01
-8.23266327e-01 -9.76038814e-01 -5.59107244e-01 -1.01992762e+00
-7.01340735e-01 3.86055589e-01 3.39995533e-01 7.10379362e-01
1.62014738e-01 1.23877920e-01 5.72690845e-01 -5.79501390e-01
-2.28877485e-01 -9.73757565e-01 -8.44344378e-01 1.96518153e-01
7.40928501e-02 -8.60077739e-01 -5.60166061e-01 -2.21363157e-01] | [6.966357231140137, 5.259241580963135] |
554a6faf-74d1-416a-a7e1-ef7511578202 | autonomous-extraction-of-gleason-patterns-for | 2011.00527 | null | https://arxiv.org/abs/2011.00527v5 | https://arxiv.org/pdf/2011.00527v5.pdf | A Dilated Residual Hierarchically Fashioned Segmentation Framework for Extracting Gleason Tissues and Grading Prostate Cancer from Whole Slide Images | Prostate cancer (PCa) is the second deadliest form of cancer in males, and it can be clinically graded by examining the structural representations of Gleason tissues. This paper proposes \RV{a new method} for segmenting the Gleason tissues \RV{(patch-wise) in order to grade PCa from the whole slide images (WSI).} Also, the proposed approach encompasses two main contributions: 1) A synergy of hybrid dilation factors and hierarchical decomposition of latent space representation for effective Gleason tissues extraction, and 2) A three-tiered loss function which can penalize different semantic segmentation models for accurately extracting the highly correlated patterns. In addition to this, the proposed framework has been extensively evaluated on a large-scale PCa dataset containing 10,516 whole slide scans (with around 71.7M patches), where it outperforms state-of-the-art schemes by 3.22% (in terms of mean intersection-over-union) for extracting the Gleason tissues and 6.91% (in terms of F1 score) for grading the progression of PCa. | ['Ayman El-Baz', 'Bilal Hassan', 'Naoufel Werghi', 'Taimur Hassan'] | 2020-11-01 | null | null | null | null | ['scene-parsing'] | ['computer-vision'] | [ 5.82928717e-01 4.96960551e-01 -3.01168770e-01 -5.59316397e-01
-1.45500588e+00 -7.19278216e-01 4.37876612e-01 2.13658199e-01
-1.28634170e-01 3.75436634e-01 6.42005429e-02 5.33190668e-02
-6.35819554e-01 -8.10228229e-01 -2.36628398e-01 -1.31636894e+00
-3.43995720e-01 8.93204749e-01 2.22134382e-01 4.09778655e-01
4.82432455e-01 5.05014002e-01 -1.06482637e+00 2.40454063e-01
1.00318170e+00 9.85605896e-01 9.59061924e-03 7.40966201e-01
-2.29535177e-01 2.04930812e-01 -3.36977422e-01 -7.29223371e-01
1.04884230e-01 -2.34892637e-01 -5.21033108e-01 9.57214385e-02
7.86709726e-01 1.83768243e-01 3.75782885e-02 1.33193052e+00
2.43995592e-01 -4.90196735e-01 1.07271862e+00 -6.28064156e-01
4.08744402e-02 3.85895103e-01 -9.28413510e-01 2.70214111e-01
-2.60070652e-01 9.89766885e-03 1.17558229e+00 -5.26179731e-01
6.84530795e-01 1.02380502e+00 8.16598058e-01 4.91143495e-01
-1.15855873e+00 -6.18425250e-01 -3.25493306e-01 3.63195315e-02
-9.70646203e-01 -1.70559853e-01 8.29205215e-01 -3.58934015e-01
3.72475624e-01 9.01068151e-01 7.09549606e-01 9.34712052e-01
8.33016396e-01 7.57310212e-01 1.40254653e+00 2.72865575e-02
3.58595431e-01 -4.05977249e-01 5.37419140e-01 6.42302871e-01
5.65214217e-01 -3.77518803e-01 -2.16261595e-01 -5.22665381e-01
8.57512057e-01 1.37745768e-01 3.08401644e-01 -3.59241754e-01
-7.79105186e-01 6.78999066e-01 6.22128665e-01 2.80872494e-01
-2.22626522e-01 1.54131770e-01 3.27278316e-01 -5.52324653e-01
1.69508904e-01 4.17217046e-01 1.29343480e-01 3.24901380e-02
-1.38358414e+00 1.96619436e-01 4.85937864e-01 6.09023273e-01
3.45577925e-01 -4.57867116e-01 -3.42816174e-01 9.02957261e-01
6.35313869e-01 7.95776010e-01 6.43196285e-01 -8.95574152e-01
4.56976414e-01 9.88560319e-01 -5.41004121e-01 -1.10066819e+00
-5.54979384e-01 -7.71929741e-01 -1.04506052e+00 -1.22848220e-01
4.40029383e-01 3.93972874e-01 -1.47066116e+00 1.18193698e+00
-4.56701890e-02 7.81964734e-02 -1.50461838e-01 7.01574624e-01
8.18023562e-01 2.82951981e-01 2.98335701e-01 -1.20306075e-01
1.82898605e+00 -8.05974662e-01 -5.83374858e-01 -5.65165639e-01
1.03276610e-01 -5.91638267e-01 5.72745204e-01 4.15638119e-01
-1.10051715e+00 -2.61569649e-01 -1.10532629e+00 4.10200981e-03
-1.99059799e-01 4.91138190e-01 1.15841901e+00 8.33547235e-01
-9.19132352e-01 3.68106425e-01 -1.38864005e+00 -3.97022873e-01
8.44763339e-01 5.86447299e-01 -3.31498682e-01 -2.20481530e-01
-2.77734190e-01 2.13954836e-01 -6.09013364e-02 3.73254955e-01
-9.08619225e-01 -8.29250872e-01 -6.10982835e-01 2.33909085e-01
2.07531787e-02 -6.47913992e-01 4.97488171e-01 -4.54736799e-01
-1.15674722e+00 9.84297216e-01 -4.52390999e-01 -1.71973705e-01
4.45779115e-01 1.64126649e-01 -6.36447519e-02 5.34644186e-01
1.16164461e-01 4.45598662e-01 5.13545513e-01 -1.35369241e+00
-5.71860194e-01 -1.22260392e+00 -5.64853609e-01 1.55758306e-01
-4.67412025e-01 -4.01622683e-01 -8.74128342e-01 -5.14164329e-01
7.95106411e-01 -1.16285896e+00 -6.67951286e-01 -3.17366123e-01
-5.79186141e-01 -9.94422361e-02 4.29348052e-01 -1.03406787e+00
1.00193870e+00 -2.21018100e+00 3.68344307e-01 7.47078359e-01
4.68326122e-01 -7.45818466e-02 1.57070711e-01 -1.48748398e-01
1.75584346e-01 3.27594817e-01 -4.60969955e-01 -5.51539004e-01
5.95942372e-03 3.65942538e-01 6.08781166e-02 3.81611347e-01
1.00790262e-01 9.49592650e-01 -7.74880052e-01 -8.79399955e-01
-9.61906984e-02 3.70473862e-01 -1.81610107e-01 -7.06532523e-02
2.54690021e-01 1.18378535e-01 -7.38762558e-01 1.18216205e+00
8.51740479e-01 -2.88481325e-01 5.29065311e-01 -9.16938111e-02
4.92074996e-01 -3.47341001e-01 -7.89331436e-01 1.69549429e+00
7.44919330e-02 1.97891995e-01 1.95761368e-01 -5.27474701e-01
1.09182775e+00 -5.75012304e-02 7.27977693e-01 -3.92844111e-01
-7.92872459e-02 4.11725909e-01 -6.65293187e-02 -3.31348956e-01
1.29904494e-01 9.15333852e-02 -5.44719756e-01 -1.49545699e-01
1.11479573e-01 -3.85958925e-02 3.99538934e-01 1.02054372e-01
1.69446397e+00 -4.00714874e-01 2.96261963e-02 -5.58955550e-01
5.16303122e-01 7.76799098e-02 8.62434506e-01 5.99376559e-01
-4.90198076e-01 4.60216582e-01 8.96607041e-01 -1.33155435e-01
-6.64286613e-01 -1.58190286e+00 -3.49379867e-01 4.51964438e-01
2.73037881e-01 -1.76329121e-01 -5.83252370e-01 -8.66574585e-01
2.44816899e-01 3.47399712e-01 -7.89365113e-01 5.76268137e-02
-6.60925090e-01 -1.55823791e+00 6.34329796e-01 8.27329636e-01
4.61207449e-01 -4.82238442e-01 -1.25842407e-01 -3.71320933e-01
-6.39135484e-03 -6.04588211e-01 -1.29770011e-01 1.78759143e-01
-1.54992867e+00 -1.22267902e+00 -7.70230651e-01 -6.90790534e-01
1.23800743e+00 -7.59850070e-02 8.25359464e-01 -4.13829118e-01
-7.00332224e-01 5.86391352e-02 2.41585344e-01 8.55518728e-02
-2.96249185e-02 7.89898112e-02 -8.18699241e-01 8.75857994e-02
5.44530869e-01 -5.98451734e-01 -8.72565866e-01 1.35166436e-01
-6.87578261e-01 -1.53336570e-01 1.52350485e+00 7.22910464e-01
1.24623597e+00 1.42485395e-01 5.51647954e-02 -1.19160211e+00
4.11049366e-01 -3.38193685e-01 -3.86300117e-01 3.37172806e-01
-6.53442740e-01 -2.59666741e-01 2.28166968e-01 1.11493610e-01
-1.10056388e+00 1.77207708e-01 1.28259107e-01 -1.02249227e-01
-1.76720724e-01 3.80697310e-01 -1.48286194e-01 -7.43611380e-02
1.62944689e-01 3.66918713e-01 -7.23650530e-02 -4.58378583e-01
1.86426137e-02 2.45728657e-01 8.22528243e-01 -3.18838388e-01
5.79529881e-01 7.21811354e-01 7.60163844e-01 -4.96317923e-01
-5.48867345e-01 -8.41203272e-01 -6.04447961e-01 -1.87876031e-01
1.05072820e+00 -6.10068500e-01 -6.42634273e-01 4.19403762e-01
-3.18614900e-01 1.00530021e-01 -2.64660385e-03 3.35122973e-01
-3.55269283e-01 6.45548820e-01 -1.01597929e+00 -6.62047327e-01
-5.80463350e-01 -1.26789820e+00 1.47846317e+00 5.83130360e-01
-1.56880453e-01 -8.78936231e-01 1.75279409e-01 1.03316236e+00
3.57071787e-01 5.34808517e-01 1.22218442e+00 -6.68468952e-01
-4.96116906e-01 -6.45224035e-01 -3.72448772e-01 -6.92862272e-02
-6.22383282e-02 3.77917551e-02 -9.98701513e-01 -3.04543316e-01
-4.30043750e-02 2.22271279e-01 1.22682297e+00 7.09401488e-01
1.26417410e+00 -7.48714581e-02 -8.14666450e-01 7.77887225e-01
1.84733689e+00 3.17563534e-01 8.16394210e-01 2.09947512e-01
5.67047834e-01 4.56555188e-01 5.38712680e-01 1.21712752e-01
1.73186824e-01 4.09670413e-01 6.45092189e-01 -1.02872901e-01
-2.58666843e-01 2.12528840e-01 -3.43131199e-02 8.88418615e-01
-4.32170808e-01 2.39555836e-01 -1.11996734e+00 6.53368294e-01
-1.56931984e+00 -6.66692436e-01 -3.65444392e-01 1.98175883e+00
6.64204478e-01 2.90654242e-01 -2.56277144e-01 -1.49310738e-01
3.98043007e-01 1.33259907e-01 -5.10827243e-01 -1.74191564e-01
1.44632608e-01 3.69685650e-01 8.18009734e-01 3.70770991e-01
-1.45213819e+00 3.10513824e-01 6.18470669e+00 1.17689586e+00
-9.17576492e-01 -3.80789936e-02 1.12297142e+00 2.58588437e-02
-2.53950268e-01 -6.71604127e-02 -1.05361092e+00 6.27004445e-01
7.00650990e-01 2.86379345e-02 -1.07298672e-01 8.32464814e-01
-1.55567080e-01 -2.39087835e-01 -9.30034220e-01 8.06421280e-01
2.35186815e-01 -1.05137861e+00 -4.08436023e-02 5.23389995e-01
6.95163727e-01 -3.99407864e-01 1.45353764e-01 3.44045945e-02
4.01769839e-02 -1.16559184e+00 -1.82735324e-01 1.01155758e+00
4.79024470e-01 -6.25197232e-01 1.37856460e+00 -1.74653009e-01
-1.15582299e+00 -3.44161317e-02 -1.31317139e-01 8.27377737e-01
-3.59123677e-01 8.71222198e-01 -9.80211079e-01 8.07533801e-01
6.30184472e-01 5.63463509e-01 -1.28551781e+00 1.28540027e+00
5.61096370e-02 8.41789901e-01 -2.94198722e-01 -1.69703253e-02
3.02253128e-03 -1.89599201e-01 4.94044602e-01 1.45242608e+00
4.68332261e-01 -2.60162562e-01 3.91650535e-02 7.43654847e-01
3.17217201e-01 -9.27720219e-02 1.42203689e-01 -1.62522674e-01
2.49477118e-01 1.73222673e+00 -1.28397202e+00 -5.40266111e-02
1.29315451e-01 5.97711980e-01 -1.56815186e-01 1.74338698e-01
-6.00836456e-01 -2.87165612e-01 1.72262296e-01 -6.10651001e-02
2.56438732e-01 6.19744062e-02 -8.41020584e-01 -5.97389579e-01
1.27633587e-01 -3.12526435e-01 7.08320916e-01 -1.89246312e-01
-1.54758871e+00 4.06296045e-01 -1.92587748e-01 -1.00075889e+00
4.56480794e-02 -7.52837300e-01 -6.71619713e-01 6.77519441e-01
-1.39538252e+00 -1.47075522e+00 -7.65004694e-01 7.82883242e-02
4.20920342e-01 1.38903493e-02 1.01555991e+00 2.92905957e-01
-6.44561231e-01 5.66152990e-01 1.82555541e-01 1.00651406e-01
6.72545314e-01 -1.91481459e+00 -2.85881579e-01 6.68690026e-01
-5.78640401e-01 7.25828171e-01 4.13439900e-01 -8.12151253e-01
-1.69968033e+00 -1.14410400e+00 7.62826920e-01 -3.41673553e-01
5.80442011e-01 -2.64488637e-01 -4.31786209e-01 5.87838471e-01
-1.80711195e-01 -5.78830466e-02 1.11723483e+00 5.71220107e-02
3.93314958e-02 -5.51840365e-01 -1.39735305e+00 3.89772356e-01
5.36974967e-01 1.01759054e-01 -3.35738212e-01 2.25196362e-01
-7.53418058e-02 -3.21147650e-01 -1.38471305e+00 6.90311313e-01
6.37391448e-01 -7.60308087e-01 1.08969223e+00 8.62155929e-02
6.80220723e-01 -2.04392567e-01 -3.57999168e-02 -6.92746460e-01
-6.27999842e-01 -1.38193175e-01 -7.19093606e-02 1.37054765e+00
2.52521336e-01 -2.94686764e-01 1.55110705e+00 6.24349296e-01
-3.81112546e-01 -1.21566772e+00 -1.28986716e+00 -4.01957601e-01
-1.07550450e-01 7.38941878e-02 2.26518765e-01 4.88835394e-01
-2.03128412e-01 -1.57055840e-01 3.58877182e-01 2.51158625e-01
1.24802697e+00 2.35831186e-01 5.80904007e-01 -1.31259978e+00
-3.52708489e-01 -8.58537793e-01 -1.00871778e+00 -5.31956732e-01
-6.31205857e-01 -1.08081460e+00 8.47628526e-03 -1.60231745e+00
1.05744112e+00 -4.39063460e-01 -6.13895059e-01 3.56889576e-01
-2.92380065e-01 3.94538671e-01 -9.63431746e-02 5.85742295e-01
-6.27600074e-01 1.04977064e-01 1.33962381e+00 -4.44060415e-01
7.25518316e-02 2.66628444e-01 -8.54677618e-01 8.24741840e-01
4.34427947e-01 -3.74301583e-01 -1.66341692e-01 1.43583581e-01
-2.37056576e-02 1.08018890e-01 4.81639177e-01 -1.06731641e+00
2.33610779e-01 -9.81046781e-02 9.01046395e-01 -9.63201582e-01
5.93527317e-01 -6.49336934e-01 3.09468925e-01 6.17753804e-01
-2.59667546e-01 -4.34643120e-01 8.20736215e-02 8.70762169e-01
-2.08041981e-01 -3.88553202e-01 5.23798108e-01 -2.10785970e-01
-4.72838134e-01 1.01572953e-01 -3.20206508e-02 -4.76566821e-01
1.22118962e+00 -3.93005908e-01 -7.01767027e-01 3.94459724e-01
-9.49216843e-01 3.30780506e-01 2.60874897e-01 -7.35485032e-02
4.69377786e-01 -1.19805074e+00 -6.83838964e-01 9.20896605e-02
1.19004495e-01 1.62633300e-01 5.22138774e-01 1.16248298e+00
-7.53231585e-01 5.83065212e-01 -2.30067253e-01 -1.00547206e+00
-1.70502269e+00 -1.92746166e-02 1.45250872e-01 -1.27219594e+00
-4.05106813e-01 1.13400257e+00 2.43241653e-01 -1.86734334e-01
2.96127617e-01 -3.06248993e-01 -4.32959288e-01 5.53037189e-02
1.79623768e-01 5.08175790e-01 1.54190168e-01 -3.17155540e-01
-5.98715246e-01 6.68345749e-01 -3.77403408e-01 2.25925907e-01
1.43386400e+00 3.39262933e-01 -5.85179985e-01 1.23564646e-01
9.76070821e-01 3.83468002e-01 -1.12216902e+00 1.63139105e-01
1.81879848e-01 -5.23630142e-01 -4.75598574e-02 -9.65937316e-01
-1.19652760e+00 5.75969398e-01 1.02886260e+00 1.05135761e-01
1.26268756e+00 1.62288815e-01 7.82267988e-01 -1.67204112e-01
1.31720409e-01 -8.05439770e-01 8.88334885e-02 -3.00711304e-01
5.76952100e-01 -8.80962908e-01 4.17896688e-01 -8.90448511e-01
-4.29371625e-01 1.29779875e+00 2.25078076e-01 -5.80770493e-01
5.18317282e-01 2.60634065e-01 -9.70460325e-02 -7.20494151e-01
-3.96361142e-01 3.98676723e-01 4.72300529e-01 3.33701402e-01
3.27167362e-01 4.88788456e-01 -7.30505586e-01 1.28638589e+00
-2.52639711e-01 -1.81553498e-01 2.97660436e-02 6.95563376e-01
-3.46787989e-01 -9.03354526e-01 -3.31365615e-01 9.95992541e-01
-6.49085462e-01 3.49382460e-01 -6.09928548e-01 5.51029325e-01
1.41151756e-01 4.84238684e-01 2.90791132e-02 -1.22075438e-01
-3.88570763e-02 -1.38149321e-01 3.61384660e-01 -3.33203703e-01
-4.94986475e-01 5.61589420e-01 -8.06591436e-02 -6.24455750e-01
-2.62599111e-01 -1.05186689e+00 -1.13717198e+00 2.58645922e-01
4.28742124e-03 -2.81947255e-02 7.80446172e-01 6.43144488e-01
2.25705698e-01 8.89403403e-01 4.49497849e-01 -6.32629097e-01
-5.74014008e-01 -9.08876836e-01 -7.60207176e-01 4.74043638e-01
-2.03150615e-01 -4.74544019e-01 -5.44985175e-01 1.18743725e-01] | [14.921553611755371, -2.8670806884765625] |
f0d020b7-ca0e-4070-af55-842ad73b085e | hokem-human-and-object-keypoint-based | 2306.14260 | null | https://arxiv.org/abs/2306.14260v1 | https://arxiv.org/pdf/2306.14260v1.pdf | HOKEM: Human and Object Keypoint-based Extension Module for Human-Object Interaction Detection | Human-object interaction (HOI) detection for capturing relationships between humans and objects is an important task in the semantic understanding of images. When processing human and object keypoints extracted from an image using a graph convolutional network (GCN) to detect HOI, it is crucial to extract appropriate object keypoints regardless of the object type and to design a GCN that accurately captures the spatial relationships between keypoints. This paper presents the human and object keypoint-based extension module (HOKEM) as an easy-to-use extension module to improve the accuracy of the conventional detection models. The proposed object keypoint extraction method is simple yet accurately represents the shapes of various objects. Moreover, the proposed human-object adaptive GCN (HO-AGCN), which introduces adaptive graph optimization and attention mechanism, accurately captures the spatial relationships between keypoints. Experiments using the HOI dataset, V-COCO, showed that HOKEM boosted the accuracy of an appearance-based model by a large margin. | ['Yoshiki Ito'] | 2023-06-25 | null | null | null | null | ['human-object-interaction-detection'] | ['computer-vision'] | [-3.40665244e-02 7.88253266e-03 2.17341363e-01 -9.01814401e-02
-5.69724478e-03 -1.56745687e-01 5.37532568e-01 3.29759389e-01
-3.12836468e-01 4.80692126e-02 -3.01426083e-01 4.35783379e-02
-2.61507124e-01 -5.92607081e-01 -7.31471241e-01 -2.87519217e-01
-5.67206601e-03 4.37374741e-01 7.05544531e-01 -1.23010173e-01
4.06274527e-01 9.71527934e-01 -1.70577502e+00 -6.82774037e-02
5.89902580e-01 1.17741954e+00 4.21370745e-01 7.40649641e-01
-3.91299985e-02 6.14447594e-01 -5.08865416e-01 -2.08460569e-01
4.78702277e-01 -1.62228912e-01 -5.31882405e-01 5.87643497e-02
7.17148960e-01 -2.62966573e-01 -3.49583089e-01 1.06533992e+00
3.61730933e-01 2.03677922e-01 5.71018636e-01 -2.14594316e+00
-7.89191067e-01 2.20039904e-01 -7.63474643e-01 2.35521108e-01
2.19733119e-01 1.12120375e-01 7.52106428e-01 -1.07669878e+00
8.11093330e-01 1.48579574e+00 4.99724418e-01 2.54984498e-01
-5.62237561e-01 -7.68228531e-01 7.27262944e-02 5.94851315e-01
-1.65546679e+00 -7.50003848e-03 8.68538857e-01 -2.81863987e-01
9.31778848e-01 3.42459083e-01 1.05183530e+00 5.05553544e-01
3.85362446e-01 1.16322827e+00 4.74150807e-01 -6.45465791e-01
-1.62141040e-01 2.03214735e-01 5.39617538e-01 8.62300217e-01
6.19925499e-01 9.45761204e-02 -3.34534794e-01 -3.29947621e-02
1.01917779e+00 3.39704126e-01 -2.76578844e-01 -6.89514756e-01
-1.28147542e+00 6.27371132e-01 9.12550867e-01 1.67420506e-01
-4.30870831e-01 3.49369317e-01 8.62052515e-02 -1.72007218e-01
-2.48129129e-01 3.39953274e-01 -8.18388462e-02 4.82229441e-01
-3.91827643e-01 4.71527457e-01 3.35773319e-01 1.50276315e+00
8.12045872e-01 -1.95878446e-01 -3.40134889e-01 4.01139945e-01
4.42061722e-01 6.23084664e-01 1.94787145e-01 -6.53223574e-01
1.77335367e-01 1.38291562e+00 2.20713422e-01 -1.60829794e+00
-7.41931856e-01 -4.07061011e-01 -8.42930794e-01 4.27693009e-01
2.72707731e-01 4.34271961e-01 -9.93330359e-01 1.23159397e+00
7.75175095e-01 -1.57471985e-01 -3.69515359e-01 1.16864884e+00
1.10949695e+00 4.44896221e-01 2.49281466e-01 4.33420211e-01
1.59945476e+00 -1.10549235e+00 -6.37188315e-01 -4.34795052e-01
4.63691562e-01 -7.65062988e-01 9.03480887e-01 6.56544790e-02
-7.91857958e-01 -9.18507993e-01 -1.14451456e+00 -2.20519021e-01
-7.47201145e-01 5.08677483e-01 7.65460551e-01 9.67115909e-02
-8.53826642e-01 2.05846995e-01 -2.90923476e-01 -7.21516788e-01
3.22971761e-01 5.00567079e-01 -7.09774256e-01 -5.22849290e-03
-8.71936381e-01 8.50157797e-01 8.49078715e-01 3.16785783e-01
-6.81620896e-01 -5.33470869e-01 -9.86742496e-01 1.64777219e-01
5.08690715e-01 -7.15184748e-01 8.03716004e-01 -1.03921878e+00
-8.48392069e-01 8.42246354e-01 1.07312508e-01 -4.15814787e-01
5.92666030e-01 -2.35400558e-01 -4.04945165e-01 3.56726676e-01
7.27532208e-02 1.02782404e+00 1.17925465e+00 -1.23734665e+00
-9.59235251e-01 -5.40500283e-01 2.08373722e-02 1.79652229e-01
6.87392876e-02 1.87788323e-01 -8.08788776e-01 -5.27000070e-01
2.87478298e-01 -1.24158192e+00 9.62985530e-02 4.55652267e-01
-4.67162043e-01 -6.57822251e-01 1.43806171e+00 -5.71089447e-01
1.00962079e+00 -2.11440468e+00 -1.08648159e-01 2.86860973e-01
6.41094625e-01 4.83113259e-01 -2.92754352e-01 2.81969965e-01
-6.86082095e-02 -2.90507525e-01 2.06539795e-01 6.12594932e-02
-1.14693753e-01 1.34223641e-03 4.02103543e-01 4.60351259e-01
2.77331650e-01 1.56796467e+00 -7.89154470e-01 -7.11286068e-01
6.10217035e-01 4.24016476e-01 -1.95206687e-01 3.37969065e-01
2.80565619e-02 -4.36181463e-02 -2.34678581e-01 8.32496345e-01
7.03722179e-01 -3.52607548e-01 -3.57778460e-01 -8.32970321e-01
1.44753054e-01 -7.78249204e-01 -1.45379579e+00 1.20605123e+00
-3.94422747e-02 7.59853423e-01 1.33723356e-02 -5.96819997e-01
9.52940941e-01 1.07807249e-01 4.09827709e-01 -6.67524636e-01
3.74426425e-01 -1.82953719e-02 1.74037933e-01 -5.27035117e-01
5.56729853e-01 7.27937877e-01 1.92426249e-01 9.24731120e-02
1.36485532e-01 3.98802496e-02 2.08948597e-01 4.16544050e-01
5.59577882e-01 -8.57762620e-02 8.27256978e-01 -2.40564436e-01
6.66412771e-01 1.42458249e-02 5.26703298e-01 7.91173398e-01
-4.62372690e-01 6.02246881e-01 2.14204788e-02 -7.12880552e-01
-1.07239020e+00 -5.40048122e-01 1.29828945e-01 9.29960728e-01
7.16883779e-01 -3.34558517e-01 -5.81317604e-01 -7.94497669e-01
2.58589149e-01 4.92162615e-01 -8.88315082e-01 -1.66753486e-01
-5.76232195e-01 -2.64815956e-01 -1.01354994e-01 8.49239528e-01
7.61994421e-01 -1.37169373e+00 -9.81665075e-01 2.80969925e-02
1.73295904e-02 -1.19508481e+00 -7.48474836e-01 -6.76202253e-02
-2.18739331e-01 -1.51664710e+00 -6.73587322e-01 -8.60068619e-01
8.80824685e-01 6.99810565e-01 7.83225775e-01 5.39679170e-01
-9.42465007e-01 5.02870500e-01 -4.23874944e-01 -9.89803672e-01
-8.94751698e-02 -4.14622501e-02 -1.37983605e-01 2.14090884e-01
4.86801773e-01 3.11925020e-02 -7.39008605e-01 5.47022760e-01
-7.39053130e-01 2.28352711e-01 6.70931697e-01 4.80763048e-01
3.41794401e-01 2.33664811e-02 1.79103300e-01 -5.09883225e-01
2.85409272e-01 1.12203304e-02 -7.18990505e-01 6.85069025e-01
-4.20011163e-01 -1.15561575e-01 1.99618176e-01 -5.18898904e-01
-6.54763460e-01 2.62210488e-01 3.77580076e-01 -5.55168271e-01
-1.89715102e-01 -5.93556911e-02 -3.96632791e-01 -6.15079403e-01
3.50236684e-01 3.03398892e-02 -1.43287197e-01 -1.89355239e-01
5.32866478e-01 6.72032475e-01 5.62024057e-01 -1.66132614e-01
7.74508119e-01 3.65588695e-01 2.04665080e-01 -1.03249991e+00
-6.39313042e-01 -9.56896901e-01 -1.07138813e+00 -4.62496668e-01
1.09686792e+00 -5.77414632e-01 -9.20353591e-01 6.33501232e-01
-1.40913820e+00 -6.89541176e-02 -4.09106612e-02 4.42680061e-01
-2.56759703e-01 3.41154456e-01 -4.46558967e-02 -9.57956135e-01
-7.10195363e-01 -1.01629376e+00 1.40382552e+00 4.93234098e-01
-7.87133723e-02 -7.12078035e-01 -5.61195016e-01 2.30497897e-01
2.81418830e-01 3.03180993e-01 9.08821940e-01 -6.09443426e-01
-8.75990808e-01 -5.43715417e-01 -8.55298460e-01 -8.00146013e-02
-3.91465016e-02 1.50150776e-01 -6.13816917e-01 -3.09873998e-01
-3.84430707e-01 1.62417844e-01 5.01157582e-01 3.09028804e-01
9.05976832e-01 -2.17836976e-01 -6.42906487e-01 6.12751126e-01
1.48506856e+00 4.33305502e-01 5.88813543e-01 5.02745211e-01
1.17440605e+00 4.66785610e-01 7.89041162e-01 2.71584928e-01
2.13019431e-01 8.41164589e-01 5.14064610e-01 -3.58761400e-01
-4.56150174e-01 -2.33551130e-01 -2.39672869e-01 2.88526356e-01
-2.96997391e-02 -1.27352193e-01 -9.74685252e-01 6.11664951e-01
-2.10892224e+00 -6.49515450e-01 -3.86476010e-01 1.85283983e+00
1.69567694e-03 9.62608457e-02 2.02368155e-01 1.00328170e-01
9.57044899e-01 -4.44776654e-01 -5.70735276e-01 1.15985386e-01
8.92361626e-03 -2.07778171e-01 7.53387868e-01 2.48893723e-01
-1.12672281e+00 1.06162286e+00 5.57271004e+00 7.22694278e-01
-9.63360965e-01 2.11049290e-03 1.36858709e-02 4.72344846e-01
3.55915666e-01 -1.44403607e-01 -1.18663538e+00 1.15783758e-01
1.21742100e-01 -1.46834627e-01 2.25462541e-01 1.21508992e+00
4.15227860e-02 -1.84691980e-01 -1.00419331e+00 1.33541524e+00
2.44369179e-01 -1.04794300e+00 3.75542283e-01 -2.09653005e-01
4.08353418e-01 -4.60908711e-01 -2.72068977e-01 1.68991480e-02
-2.50061769e-02 -7.66021192e-01 9.00176108e-01 5.39388657e-01
3.95286560e-01 -6.93221986e-01 8.26939881e-01 1.56206906e-01
-1.67592704e+00 -3.73346955e-01 -6.46118581e-01 2.41649613e-01
3.31727490e-02 6.77696317e-02 -1.07534289e+00 4.49568808e-01
8.81870031e-01 5.47062576e-01 -1.22041476e+00 1.40039289e+00
-2.63976574e-01 2.46066949e-03 -4.13198233e-01 -2.30237380e-01
2.49245584e-01 2.58901790e-02 5.89217126e-01 1.06090677e+00
8.49732161e-02 2.48599559e-01 3.35084200e-01 8.96173716e-01
1.67642683e-01 2.33108684e-01 -5.76226115e-01 1.26018198e-02
4.10165966e-01 1.61790550e+00 -1.21565723e+00 -3.59665692e-01
-3.38130474e-01 1.06845403e+00 3.47668558e-01 2.93118566e-01
-8.41782928e-01 -8.13085914e-01 3.65110368e-01 4.43111449e-01
3.52982968e-01 -3.50260347e-01 1.79594398e-01 -6.61123872e-01
-9.73587185e-02 -6.93680346e-01 4.78064090e-01 -1.32385027e+00
-9.13325250e-01 5.82439482e-01 2.69001752e-01 -1.10781240e+00
1.23509973e-01 -8.54883015e-01 -5.74420214e-01 5.60835838e-01
-1.10344517e+00 -1.91947556e+00 -9.84369814e-01 7.51651645e-01
5.39147377e-01 3.79314879e-03 2.62857735e-01 1.16917007e-01
-5.52244127e-01 6.11009955e-01 -4.07393664e-01 3.38966519e-01
2.86495715e-01 -9.94022727e-01 5.33603549e-01 8.41006398e-01
3.26396972e-01 7.48041451e-01 4.05908316e-01 -7.75861859e-01
-1.39445364e+00 -1.24876475e+00 5.55681050e-01 -4.94270533e-01
2.08206147e-01 -4.17016745e-01 -9.12685573e-01 7.11624861e-01
-2.71907505e-02 1.88545614e-01 -4.85936962e-02 -5.93405187e-01
-4.94041108e-02 -7.89330527e-02 -9.13656294e-01 5.78759968e-01
1.19593024e+00 -2.60374188e-01 -4.41580474e-01 3.60491127e-01
7.85235822e-01 -2.69705176e-01 -4.86142427e-01 5.07898510e-01
6.63601339e-01 -6.53410077e-01 1.25895500e+00 -4.38589454e-01
-3.36313039e-01 -6.91843569e-01 -1.06950384e-02 -8.46122742e-01
-6.70275211e-01 -2.71716923e-01 -3.09789181e-01 9.20553982e-01
-1.38190180e-01 -4.14230615e-01 5.70696652e-01 7.45361328e-01
2.35376909e-01 -5.38647711e-01 -4.81015116e-01 -6.91705644e-01
-8.10390532e-01 -2.63002038e-01 7.83187568e-01 6.26047134e-01
-3.44805181e-01 3.10073107e-01 -3.16386968e-01 4.43914920e-01
8.38318169e-01 9.25717130e-02 1.20820332e+00 -1.43624449e+00
1.67189583e-01 -3.43926281e-01 -1.16575491e+00 -6.51115000e-01
-1.38731480e-01 -7.74233043e-01 -1.88718900e-01 -1.70237315e+00
2.50015676e-01 -2.97708571e-01 -2.81528383e-01 5.61804712e-01
-5.10436296e-01 3.70819747e-01 5.87270439e-01 1.96760044e-01
-6.86822116e-01 4.06630576e-01 1.29239011e+00 -4.01621222e-01
-2.66479522e-01 -1.37268037e-01 -3.53777975e-01 7.06776023e-01
4.76036072e-01 -2.88439959e-01 -3.68634284e-01 -1.24424256e-01
5.03293090e-02 -3.62254143e-01 9.41477776e-01 -1.24342656e+00
5.87454915e-01 -3.17280143e-02 7.06466079e-01 -1.09865367e+00
7.16976598e-02 -1.12721300e+00 3.45554233e-01 5.14688015e-01
-9.83538926e-02 3.63216430e-01 2.50621587e-01 5.83223283e-01
8.05254355e-02 -2.25734532e-01 7.92427123e-01 -2.49359757e-01
-1.48507810e+00 4.90850240e-01 1.38732120e-01 -2.14472190e-01
1.48814678e+00 -4.38656121e-01 -2.80888006e-02 -2.13814333e-01
-4.24540132e-01 2.02313766e-01 3.10453176e-01 8.09551179e-01
1.07112563e+00 -1.36265337e+00 -4.86228913e-01 5.18221140e-01
6.05804563e-01 -5.07872701e-02 1.35991767e-01 6.98495746e-01
-6.93476319e-01 5.04399359e-01 -3.55793029e-01 -7.59431243e-01
-1.69713819e+00 9.50282693e-01 3.48512411e-01 2.29244679e-01
-9.73989069e-01 8.33031595e-01 4.24136102e-01 -2.61327744e-01
3.99157852e-01 -4.67256457e-01 -2.74925590e-01 -9.65668634e-02
6.20405376e-01 5.39281726e-01 -2.05982793e-02 -9.01194274e-01
-4.64537233e-01 9.18642223e-01 -1.38666397e-02 3.64096582e-01
9.63550746e-01 -1.54303834e-01 -9.57461521e-02 7.02448413e-02
1.33833480e+00 -5.17599344e-01 -9.97763753e-01 -3.71405274e-01
-1.03625387e-01 -6.59490168e-01 8.45941901e-02 -6.18241191e-01
-1.03711069e+00 7.96288252e-01 9.79834855e-01 -1.44313738e-01
7.06843257e-01 -1.00563047e-03 7.53750086e-01 3.97538871e-01
5.52357078e-01 -9.30097342e-01 2.67578691e-01 6.45396411e-02
1.20589435e+00 -1.33430076e+00 1.21702962e-01 -6.02195323e-01
-6.21095479e-01 1.27519393e+00 1.06774914e+00 -1.61417015e-02
6.95272207e-01 -1.59077376e-01 1.28630539e-02 -6.95892036e-01
-1.64519865e-02 -3.24148595e-01 8.85968328e-01 7.76644349e-01
-1.37516454e-01 -5.93138523e-02 -1.80105910e-01 2.25935742e-01
4.25086208e-02 -1.54262707e-01 9.85161290e-02 8.61289620e-01
-6.20223045e-01 -5.48295915e-01 -5.43157816e-01 2.89209872e-01
1.20217428e-01 2.23818615e-01 -5.23803592e-01 1.25120378e+00
2.87455946e-01 6.83530509e-01 1.54927634e-02 -1.89590856e-01
4.88862723e-01 -2.07870558e-01 4.37144607e-01 -2.65892446e-01
-3.57896566e-01 -3.51278335e-01 -5.34741580e-01 -8.41082513e-01
-2.89721727e-01 -1.28845662e-01 -1.33075261e+00 2.91948412e-02
-7.56567895e-01 -1.61883831e-01 7.74039090e-01 7.22488284e-01
3.55068564e-01 5.12302518e-01 8.87597650e-02 -1.03118229e+00
-4.75510992e-02 -8.48334968e-01 -6.74790025e-01 8.42804611e-01
1.03274547e-01 -9.53396380e-01 -6.35258853e-02 -1.73268870e-01] | [9.589496612548828, 1.3682410717010498] |
cdeeb020-23c1-4011-abf9-117f143567e6 | personapkt-building-personalized-dialogue | 2306.08126 | null | https://arxiv.org/abs/2306.08126v1 | https://arxiv.org/pdf/2306.08126v1.pdf | PersonaPKT: Building Personalized Dialogue Agents via Parameter-efficient Knowledge Transfer | Personalized dialogue agents (DAs) powered by large pre-trained language models (PLMs) often rely on explicit persona descriptions to maintain personality consistency. However, such descriptions may not always be available or may pose privacy concerns. To tackle this bottleneck, we introduce PersonaPKT, a lightweight transfer learning approach that can build persona-consistent dialogue models without explicit persona descriptions. By representing each persona as a continuous vector, PersonaPKT learns implicit persona-specific features directly from a small number of dialogue samples produced by the same persona, adding less than 0.1% trainable parameters for each persona on top of the PLM backbone. Empirical results demonstrate that PersonaPKT effectively builds personalized DAs with high storage efficiency, outperforming various baselines in terms of persona consistency while maintaining good response generation quality. In addition, it enhances privacy protection by avoiding explicit persona descriptions. Overall, PersonaPKT is an effective solution for creating personalized DAs that respect user privacy. | ['Chenlei Guo', 'Xiaohu Liu', 'Yu Zhang', 'Benjamin Yao', 'Yoon Jung', 'Bin Guo', 'Xu Han'] | 2023-06-13 | null | null | null | null | ['response-generation'] | ['natural-language-processing'] | [-6.16217554e-01 5.19716203e-01 -1.75521806e-01 -7.49996960e-01
-8.39964628e-01 -7.11788952e-01 8.99306476e-01 2.56232083e-01
-5.99940062e-01 1.11551404e+00 4.57827657e-01 2.58993983e-01
2.45754614e-01 -6.96193337e-01 -3.69348466e-01 -3.62138242e-01
1.92470834e-01 9.65713739e-01 -4.28857833e-01 -1.84880003e-01
-2.25814775e-01 6.86891377e-02 -6.97652161e-01 3.73778701e-01
9.41787183e-01 6.98289573e-01 -3.84952903e-01 4.78774935e-01
-2.51432452e-02 5.61076224e-01 -9.22216773e-01 -1.33313560e+00
2.03829288e-01 -2.15187222e-01 -1.11344779e+00 3.36429961e-02
5.17738521e-01 -1.08164740e+00 -4.92947906e-01 6.65380538e-01
6.59410894e-01 2.40073249e-01 4.54692751e-01 -1.34313583e+00
-7.57967651e-01 8.08818221e-01 -1.73092172e-01 -7.45516062e-01
5.49175024e-01 2.61684597e-01 1.02154899e+00 -5.74993312e-01
4.24906969e-01 1.43330622e+00 8.70016694e-01 1.10643744e+00
-1.73100042e+00 -5.32293141e-01 -7.00261146e-02 -3.40780795e-01
-1.22048521e+00 -7.48826206e-01 3.90105367e-01 -1.81864679e-01
7.13887155e-01 4.93961930e-01 4.24408972e-01 1.74539077e+00
-1.98214397e-01 1.00028276e+00 8.73817265e-01 -9.70209017e-02
1.37487039e-01 9.03446317e-01 5.07148087e-01 6.29141510e-01
-1.02630444e-02 -5.32741129e-01 -7.75137961e-01 -8.73811483e-01
4.94405001e-01 -1.28466100e-01 -2.31870636e-01 -5.98340571e-01
-1.09232950e+00 1.07147634e+00 -7.39634084e-03 -2.44589359e-01
-3.63758802e-01 -2.12659657e-01 9.75449622e-01 4.01504129e-01
2.78156340e-01 8.00442219e-01 -4.57411230e-01 -3.42651367e-01
-3.80370289e-01 7.51935959e-01 1.32674634e+00 1.30909717e+00
7.86775351e-01 -3.02783936e-01 -6.69620574e-01 1.13928676e+00
-4.11709547e-02 4.91925567e-01 7.19570458e-01 -1.15683174e+00
5.05197883e-01 7.06410408e-01 6.77815199e-01 -1.01185596e+00
-1.64524823e-01 -5.73276402e-03 -8.28762352e-01 -3.45995992e-01
6.71803951e-01 -4.91399407e-01 -1.19293764e-01 2.06526136e+00
3.32682252e-01 -5.59506297e-01 3.72850448e-01 5.95511198e-01
8.63195598e-01 7.57619500e-01 2.48046577e-01 -3.40404660e-02
1.52790856e+00 -1.08860958e+00 -7.55469561e-01 -9.62713510e-02
7.08629191e-01 -2.60136873e-01 1.29098225e+00 1.60365775e-01
-1.12802637e+00 -2.32702628e-01 -5.76420367e-01 -2.95586377e-01
-2.92163312e-01 4.76001263e-01 6.81988180e-01 1.10747576e+00
-9.94640827e-01 5.70175111e-01 -3.97141039e-01 -5.52216053e-01
2.80758977e-01 4.86464381e-01 -7.66956210e-01 1.82442814e-02
-1.33249807e+00 7.36034274e-01 1.61637381e-01 -1.81447580e-01
-7.34265983e-01 -6.10385478e-01 -9.86978233e-01 1.79513663e-01
3.57859612e-01 -9.83809888e-01 1.56796634e+00 -5.81371188e-01
-1.91439164e+00 8.49539220e-01 -3.88341746e-03 -8.51946473e-01
1.01150644e+00 -5.19704401e-01 -1.27162591e-01 -1.49664953e-01
-2.39345487e-02 6.42503500e-01 8.22125793e-01 -1.07068384e+00
-2.03664064e-01 -2.41532892e-01 7.38048926e-02 4.18062687e-01
-8.50259662e-01 1.20097756e-01 -5.23072183e-01 -2.06528187e-01
-7.04967916e-01 -9.75737751e-01 -1.09797150e-01 1.11545645e-01
-7.43688762e-01 -5.57467103e-01 6.33262813e-01 -9.24690187e-01
8.39844823e-01 -1.92830634e+00 -5.16383089e-02 5.97874150e-02
4.31825817e-01 6.79638386e-01 -1.05400249e-01 7.79813349e-01
6.08456194e-01 -6.16569370e-02 1.79271340e-01 -1.15661287e+00
5.29200077e-01 3.11008953e-02 -2.84917951e-01 1.15600109e-01
-2.38694370e-01 8.57010961e-01 -7.52916992e-01 -2.42685705e-01
3.80969644e-02 3.43431413e-01 -4.87547278e-01 6.34066045e-01
-3.32760185e-01 2.57639557e-01 -3.61106038e-01 6.93971589e-02
5.40185034e-01 -3.85897309e-02 5.12005746e-01 9.02815629e-03
3.09519112e-01 5.13760865e-01 -6.68010294e-01 1.50544751e+00
-2.01191872e-01 2.00838909e-01 3.50360334e-01 -4.06229645e-01
1.24639773e+00 5.11118948e-01 3.91987801e-01 -3.86872739e-01
-9.82235838e-03 -1.43005297e-01 -3.85107607e-01 -1.80596992e-01
9.31327879e-01 2.23993912e-01 -6.10937178e-01 7.88461864e-01
3.78994271e-02 3.32538933e-01 -2.15085655e-01 4.51142192e-01
8.27276886e-01 -3.55822802e-01 3.06487203e-01 -5.06481342e-02
5.61977565e-01 -3.98568362e-01 7.89846599e-01 1.05239773e+00
-5.32740474e-01 1.99235946e-01 4.69866395e-01 -4.97563303e-01
-1.16537583e+00 -8.50539625e-01 3.24095726e-01 1.18624163e+00
-3.10926050e-01 -9.49815452e-01 -1.02884507e+00 -8.60945225e-01
3.36230308e-01 8.70349705e-01 -4.83599663e-01 -1.79249972e-01
-4.88745242e-01 -4.55333829e-01 9.87067580e-01 2.11037472e-01
9.30341721e-01 -7.55625486e-01 -3.40834670e-02 3.66110355e-01
-3.97627980e-01 -1.14108348e+00 -8.84665728e-01 -4.17581707e-01
-2.48261794e-01 -4.97960746e-01 -9.59760606e-01 -4.80620652e-01
4.71651912e-01 1.05914041e-01 1.06240296e+00 -4.61438447e-01
-1.45402309e-02 6.07954443e-01 -6.61068112e-02 -2.09825203e-01
-6.32322490e-01 3.19178641e-01 4.22534198e-01 1.23417914e-01
7.08896935e-01 -4.82322127e-01 -5.50167382e-01 3.03416699e-01
-4.79211323e-02 2.56487966e-01 2.17674255e-01 1.03311086e+00
-1.86444998e-01 -2.13641018e-01 6.44947946e-01 -1.09404075e+00
1.15490425e+00 -2.89628446e-01 -1.95528299e-01 2.91555107e-01
-7.28385568e-01 -2.32162431e-01 1.02377224e+00 -4.38016355e-01
-1.28391612e+00 -7.24777952e-02 1.67010978e-01 -1.42974496e-01
-1.89745322e-01 -1.06757134e-03 -5.71889162e-01 2.29797632e-01
7.06905067e-01 5.19947708e-01 5.33373535e-01 -6.02998614e-01
5.46964467e-01 9.94703591e-01 4.72946703e-01 -9.84903276e-01
6.67799592e-01 -3.38331461e-02 -7.44785726e-01 -6.64656281e-01
-3.68159622e-01 -3.58300894e-01 -3.26642990e-01 -1.22244912e-03
6.46564007e-01 -1.10466003e+00 -1.32632840e+00 7.06782877e-01
-1.02704394e+00 -4.16377902e-01 -1.90896597e-02 1.17381841e-01
-5.94267130e-01 5.24950564e-01 -1.10088408e+00 -9.71271873e-01
-9.53750789e-01 -7.76473820e-01 7.66515493e-01 1.55689895e-01
-9.28555071e-01 -8.09859157e-01 -2.63418816e-02 8.60652328e-01
2.82142967e-01 -2.00766116e-01 7.52205491e-01 -1.06602299e+00
-3.44833642e-01 -5.92582047e-01 -8.88565406e-02 2.87414968e-01
3.74894351e-01 -5.40435553e-01 -1.02567255e+00 -6.54810488e-01
-2.85587072e-01 -8.75772595e-01 2.01577783e-01 -2.49085307e-01
9.91853952e-01 -1.28639400e+00 -2.34086901e-01 2.23864719e-01
7.38125622e-01 -1.44489244e-01 3.27316821e-01 1.81960166e-01
6.23449504e-01 8.53439867e-01 4.40480858e-01 9.41717625e-01
8.82108331e-01 9.33003724e-01 -2.23270848e-01 1.12702742e-01
3.64056855e-01 -6.81217611e-01 5.10444462e-01 3.61979157e-01
3.25188756e-01 -2.02980369e-01 -5.59491873e-01 4.84483778e-01
-2.02132583e+00 -9.29315686e-01 2.53540486e-01 2.42499423e+00
1.42021036e+00 -5.03927410e-01 8.58771324e-01 -4.97472137e-01
5.22097468e-01 5.67053035e-02 -7.41612554e-01 -6.09227300e-01
-8.95147473e-02 -4.92331058e-01 4.92407352e-01 4.32447135e-01
-1.06458449e+00 1.01865947e+00 5.98122787e+00 3.40592146e-01
-6.81818426e-01 -1.85573585e-02 5.94538569e-01 -2.05436662e-01
-3.22391748e-01 -3.12520862e-01 -1.00610566e+00 6.68471634e-01
1.08612442e+00 -6.63684845e-01 4.21571583e-01 1.26121676e+00
1.40489534e-01 4.22093302e-01 -1.39137530e+00 1.05992055e+00
-5.69020445e-03 -1.05965006e+00 1.13231942e-01 3.96893173e-01
4.43362117e-01 -4.69055772e-01 2.17249140e-01 5.90535045e-01
9.27533865e-01 -1.01576245e+00 2.61629969e-01 3.34502935e-01
4.96398240e-01 -8.09865832e-01 5.59863746e-01 5.34147322e-01
-3.91081393e-01 -4.91382927e-02 -6.85097635e-01 1.53364450e-01
-2.27934383e-02 8.97478685e-02 -1.51465452e+00 3.77219200e-01
5.98387182e-01 3.87128532e-01 -3.43091637e-01 6.06289029e-01
1.96756691e-01 3.75911742e-01 -1.54064968e-01 -1.50566190e-01
1.15453057e-01 -3.15618783e-01 4.15117145e-01 1.31031549e+00
7.37530887e-02 1.00844428e-01 3.12244296e-01 8.72119427e-01
-4.74977344e-01 4.47016776e-01 -5.95163822e-01 -2.80094385e-01
9.19336557e-01 1.23288238e+00 5.08048236e-01 -3.76812160e-01
-3.12128246e-01 1.60757411e+00 7.81847298e-01 1.74607575e-01
-4.48884726e-01 -3.57189253e-02 1.22997904e+00 -1.49802864e-01
-8.01267475e-02 -1.90586299e-01 -4.12002616e-02 -1.21645296e+00
4.38829288e-02 -1.36835289e+00 5.60694158e-01 -5.18646598e-01
-1.64554870e+00 4.52101678e-01 -3.66140783e-01 -7.51153469e-01
-6.47375464e-01 -2.30835527e-01 -4.72167522e-01 1.19470525e+00
-6.85072958e-01 -1.52343428e+00 -1.45077080e-01 7.64255404e-01
2.34283715e-01 -4.59317446e-01 1.61935425e+00 4.56530713e-02
-8.76739800e-01 1.57221580e+00 2.65104145e-01 4.53468591e-01
1.34164441e+00 -1.42188346e+00 6.28381729e-01 2.22710356e-01
-5.35907865e-01 1.29392314e+00 7.87015319e-01 -6.12443507e-01
-1.32271409e+00 -1.16021705e+00 1.20544374e+00 -6.96618140e-01
2.43826434e-01 -8.52859557e-01 -1.06184554e+00 1.04849148e+00
4.03116405e-01 -6.07670665e-01 1.11256254e+00 7.46197701e-01
-7.34045684e-01 -1.84834421e-01 -1.61582708e+00 9.16076839e-01
6.57140315e-01 -7.71395385e-01 -3.95856023e-01 4.62770730e-01
5.54750919e-01 -2.73662567e-01 -1.30558932e+00 -4.25644517e-01
4.88491535e-01 -7.00054586e-01 1.05995941e+00 -8.77922475e-01
1.27293944e-01 1.07394941e-01 1.59460008e-01 -1.25689840e+00
-4.44620848e-01 -1.25982344e+00 -2.92288363e-01 1.79384911e+00
3.03650856e-01 -7.41586268e-01 1.07547879e+00 2.00295615e+00
3.59761417e-01 -2.90582031e-01 -6.61383808e-01 -9.48803484e-01
2.77153045e-01 3.56100470e-01 9.52797055e-01 1.26491988e+00
5.08663177e-01 6.23964012e-01 -1.42510831e+00 1.37520663e-03
7.92153895e-01 -1.59568518e-01 1.54317176e+00 -1.02643347e+00
-4.11338478e-01 -1.23012304e-01 -8.95692334e-02 -1.03604400e+00
3.85879397e-01 -6.27990007e-01 -1.72972884e-02 -1.23076642e+00
4.05634880e-01 -5.93138814e-01 2.31803536e-01 7.98239470e-01
-2.75362194e-01 -3.47440720e-01 -2.14161035e-02 2.17029512e-01
-6.50867701e-01 9.16842282e-01 8.09020698e-01 -1.22243077e-01
-3.65263313e-01 1.20062917e-01 -9.57862198e-01 3.37014943e-01
9.98316348e-01 -2.76027452e-02 -5.87355554e-01 -2.93729782e-01
-4.01541591e-01 1.51719391e-01 2.33457670e-01 -6.41029596e-01
3.07577759e-01 -4.14514206e-02 4.54407722e-01 -2.32408062e-01
8.98285508e-01 -4.55465972e-01 1.94253065e-02 4.99449670e-02
-7.19732404e-01 -1.76671088e-01 2.22680699e-02 5.61759889e-01
2.03837648e-01 3.69659923e-02 5.70116043e-01 -3.46889794e-01
-3.99273932e-01 5.67394972e-01 -4.31561232e-01 -2.57490337e-01
7.91542530e-01 2.97571383e-02 -5.32303751e-01 -9.30172741e-01
-4.45897579e-01 5.86264253e-01 7.43911684e-01 2.97175020e-01
2.55389333e-01 -1.20194352e+00 -7.15910673e-01 1.63886528e-02
3.32909167e-01 -2.83820987e-01 4.72177237e-01 1.55973285e-01
1.17170021e-01 5.53975821e-01 -2.94525653e-01 -4.69667614e-02
-1.78510404e+00 5.29317200e-01 3.77335578e-01 -9.16023031e-02
-8.57953787e-01 1.04540145e+00 3.60613644e-01 -1.09072864e+00
5.04512668e-01 4.98438001e-01 -9.99573544e-02 -1.18729398e-01
9.66065466e-01 1.82919517e-01 -4.37244862e-01 -5.99217176e-01
-1.08818583e-01 -3.94991815e-01 -8.64501774e-01 -3.71907860e-01
1.18025887e+00 -4.06715959e-01 -1.31173879e-01 1.67524591e-01
1.11617565e+00 1.72331512e-01 -1.35928786e+00 -7.17330158e-01
-2.60955423e-01 -6.66231215e-01 -5.98202527e-01 -1.10821187e+00
-4.58451748e-01 4.19314176e-01 1.47767738e-01 8.96811858e-02
4.25876021e-01 -2.75663346e-01 1.21110332e+00 7.97968268e-01
5.21365881e-01 -1.18459713e+00 1.12135056e-02 4.12568718e-01
9.81645703e-01 -1.31315744e+00 -1.83515102e-01 -3.85393202e-01
-1.22636008e+00 8.49557698e-01 9.92461979e-01 3.91197026e-01
3.77712920e-02 -4.56981063e-01 2.23373413e-01 3.50728631e-01
-8.38139176e-01 5.20873129e-01 6.32741973e-02 1.02584517e+00
2.71685094e-01 3.91345739e-01 -2.06880137e-01 1.19923651e+00
-6.48380399e-01 -3.79037797e-01 3.99716049e-01 5.88454843e-01
5.24598323e-02 -1.52045858e+00 -7.25627085e-03 3.55185062e-01
-1.92934960e-01 3.77708988e-04 -9.43265498e-01 4.70231950e-01
-5.23804605e-01 9.43917811e-01 -1.22684993e-01 -5.69629550e-01
2.66301781e-01 4.48736191e-01 1.38749748e-01 -4.41523910e-01
-9.87944901e-01 -4.84930813e-01 8.17449331e-01 -5.82490504e-01
5.11148512e-01 -1.00810766e+00 -6.82354987e-01 -1.14059520e+00
1.60579532e-01 4.91079420e-01 2.98990011e-01 4.84647155e-01
6.84859574e-01 -3.84221673e-01 7.48904228e-01 -1.74809843e-01
-1.07834160e+00 -9.02069867e-01 -5.21901965e-01 6.11805856e-01
1.16847716e-01 -7.15760365e-02 4.13056575e-02 -2.31257007e-01] | [12.70572280883789, 8.184429168701172] |
b5e18b51-ff3b-4846-95c9-aa9d7946bd7f | sefnet-bridging-tabular-datasets-with | 2306.11636 | null | https://arxiv.org/abs/2306.11636v1 | https://arxiv.org/pdf/2306.11636v1.pdf | SeFNet: Bridging Tabular Datasets with Semantic Feature Nets | Machine learning applications cover a wide range of predictive tasks in which tabular datasets play a significant role. However, although they often address similar problems, tabular datasets are typically treated as standalone tasks. The possibilities of using previously solved problems are limited due to the lack of structured contextual information about their features and the lack of understanding of the relations between them. To overcome this limitation, we propose a new approach called Semantic Feature Net (SeFNet), capturing the semantic meaning of the analyzed tabular features. By leveraging existing ontologies and domain knowledge, SeFNet opens up new opportunities for sharing insights between diverse predictive tasks. One such opportunity is the Dataset Ontology-based Semantic Similarity (DOSS) measure, which quantifies the similarity between datasets using relations across their features. In this paper, we present an example of SeFNet prepared for a collection of predictive tasks in healthcare, with the features' relations derived from the SNOMED-CT ontology. The proposed SeFNet framework and the accompanying DOSS measure address the issue of limited contextual information in tabular datasets. By incorporating domain knowledge and establishing semantic relations between features, we enhance the potential for meta-learning and enable valuable insights to be shared across different predictive tasks. | ['Przemysław Biecek', 'Piotr Wilczyński', 'Katarzyna Woźnica'] | 2023-06-20 | null | null | null | null | ['meta-learning', 'semantic-textual-similarity', 'semantic-similarity'] | ['methodology', 'natural-language-processing', 'natural-language-processing'] | [ 2.64593929e-01 2.18311876e-01 -5.91191709e-01 -4.50469345e-01
-5.37529647e-01 -2.95969009e-01 6.85832679e-01 1.01716936e+00
-7.68241733e-02 8.22947860e-01 4.99085397e-01 1.14884719e-01
-1.21318257e+00 -1.04884255e+00 -3.82222384e-01 -5.45946479e-01
-6.00076132e-02 5.35801172e-01 1.72629029e-01 -4.57948774e-01
2.93818355e-01 1.92220733e-01 -1.96829605e+00 8.75637829e-01
1.04104304e+00 1.32644916e+00 2.27505684e-01 -2.28815109e-01
-7.21968949e-01 9.22679484e-01 -3.93436193e-01 -4.93220776e-01
1.36885457e-02 -2.63482094e-01 -1.06079292e+00 -2.22925454e-01
2.53175765e-01 4.74173039e-01 7.20862746e-02 8.74053776e-01
1.92240119e-01 6.63281381e-02 5.61563492e-01 -1.55383503e+00
-2.62903750e-01 5.23299992e-01 8.95829871e-02 2.67825983e-02
6.16003096e-01 -4.42018867e-01 1.27220547e+00 -3.98401827e-01
9.16994274e-01 1.20295000e+00 8.28061640e-01 2.65499800e-01
-9.55413461e-01 -3.23268235e-01 4.01307978e-02 6.37945831e-01
-1.00101638e+00 -4.20829803e-02 5.98736644e-01 -5.52879930e-01
9.17167544e-01 4.93201703e-01 6.52944386e-01 8.01688790e-01
1.82758197e-01 4.87262458e-01 1.17221534e+00 -5.65601289e-01
2.77633190e-01 4.74631757e-01 1.78820416e-01 5.24727166e-01
2.97383875e-01 -2.39014313e-01 -7.57479489e-01 -2.42079794e-01
1.69842318e-01 4.50637579e-01 -1.30322233e-01 -6.47902429e-01
-1.24704766e+00 8.38468850e-01 5.47737300e-01 7.08213687e-01
-1.97000414e-01 -4.15913224e-01 8.42519522e-01 2.64214784e-01
3.96569997e-01 8.09783340e-01 -7.47657061e-01 7.43691698e-02
-6.04224980e-01 2.69887596e-01 6.99985027e-01 9.92739916e-01
8.68855715e-01 -6.09828413e-01 1.18151633e-03 9.72338319e-01
9.21581089e-02 -2.79522948e-02 8.89886200e-01 -7.34485328e-01
4.58563149e-01 1.27142060e+00 -1.48700580e-01 -1.15559554e+00
-7.37275183e-01 -4.21449333e-01 -4.48338956e-01 -2.40317553e-01
2.22217709e-01 3.54816288e-01 -6.51317477e-01 1.49331748e+00
4.16888863e-01 -2.16795802e-01 7.42684454e-02 4.60223585e-01
1.11966503e+00 2.22223133e-01 3.88691574e-01 -2.06743568e-01
1.53915370e+00 -7.07178473e-01 -8.50366652e-01 -1.41074196e-01
9.01628613e-01 -4.72558051e-01 8.53846788e-01 3.00204962e-01
-3.96380186e-01 -3.72935653e-01 -8.58190536e-01 -1.77955199e-02
-1.11579645e+00 -3.78954977e-01 8.16130698e-01 3.76872331e-01
-6.40786171e-01 7.84077406e-01 -4.00182903e-01 -7.91068614e-01
4.39244151e-01 2.84119636e-01 -7.01791286e-01 -3.03662837e-01
-1.43763494e+00 1.23240674e+00 9.85033751e-01 -3.26161444e-01
-1.08969525e-01 -9.35991287e-01 -8.22486222e-01 1.40830368e-01
7.84489632e-01 -8.10961485e-01 7.87496746e-01 -8.73718560e-01
-7.99453437e-01 7.09047496e-01 9.33770239e-02 -4.45104808e-01
2.43988469e-01 -2.71408237e-03 -8.01379859e-01 5.88095449e-02
2.98002660e-01 4.00972486e-01 2.97795415e-01 -1.08003044e+00
-8.18706274e-01 -5.74996471e-01 2.08689436e-01 1.58527136e-01
-5.70831299e-01 -2.55287975e-01 -1.73917532e-01 -4.23900694e-01
1.73560739e-01 -6.20428026e-01 -2.06755817e-01 -1.44929662e-01
-1.92447945e-01 -2.00920254e-01 5.86196601e-01 -2.60856360e-01
1.45491672e+00 -1.93196166e+00 1.72869384e-01 3.52010399e-01
2.54997671e-01 6.88275397e-02 1.91804916e-01 7.28737056e-01
-7.55412653e-02 1.74878351e-02 -3.29064399e-01 2.00918972e-01
-1.30948201e-01 5.68570614e-01 1.21618891e-02 -1.67723387e-01
-6.32247403e-02 7.38672614e-01 -1.10241139e+00 -6.28737152e-01
3.71663153e-01 2.98928738e-01 -3.60287398e-01 -1.63564265e-01
-4.04191822e-01 3.22941989e-01 -8.26085210e-01 6.22915983e-01
3.15448642e-01 -2.54062414e-01 4.61118013e-01 -4.20084596e-01
1.91040132e-02 4.57937956e-01 -9.63182449e-01 1.82300448e+00
-6.59514606e-01 1.91564873e-01 -6.69760942e-01 -1.25584376e+00
1.06682408e+00 2.52146065e-01 9.98541176e-01 -9.59685326e-01
9.55451131e-02 4.87885326e-01 -1.40405923e-01 -7.53402650e-01
4.26465988e-01 -6.45505309e-01 -1.27157301e-01 6.49266988e-02
2.93144822e-01 7.77617320e-02 3.34958881e-01 -3.55422236e-02
1.05378079e+00 1.74192712e-01 1.06416929e+00 -3.30482721e-01
7.27379858e-01 2.86102355e-01 4.44809049e-01 4.77491677e-01
1.38935715e-01 3.34697604e-01 5.53569376e-01 -8.91706169e-01
-8.44940186e-01 -7.69558847e-01 -6.70697927e-01 8.89152706e-01
-5.97902294e-03 -1.02764833e+00 -4.17116523e-01 -8.82279813e-01
3.57762098e-01 5.00984132e-01 -8.71614277e-01 -2.45267570e-01
-1.77653104e-01 -7.55027831e-01 2.10467204e-01 4.04985607e-01
3.19697708e-01 -1.12709212e+00 -8.37910354e-01 2.51650959e-01
-2.75263935e-01 -1.10713553e+00 3.45140249e-01 4.12436724e-01
-1.23074377e+00 -1.27337921e+00 -3.63709219e-02 -3.91684264e-01
3.13119113e-01 8.54246914e-02 1.35111058e+00 -5.18969037e-02
-3.62984776e-01 2.49396607e-01 -7.49339581e-01 -5.43220639e-01
-4.25485492e-01 2.53843993e-01 -2.72853732e-01 -1.36596218e-01
5.37684560e-01 -4.62016016e-01 -4.25182819e-01 3.31937432e-01
-1.07343876e+00 7.24309459e-02 3.82509500e-01 8.24063838e-01
6.81284845e-01 1.59662202e-01 9.48409617e-01 -1.35678995e+00
3.31048667e-01 -8.98067951e-01 -2.17742488e-01 5.70984840e-01
-9.09268200e-01 4.24507231e-01 7.12062895e-01 1.51373342e-01
-9.86553967e-01 -2.11817399e-01 -1.49959968e-02 -2.02476997e-02
-1.21921942e-01 1.01192617e+00 -4.08642501e-01 1.73968390e-01
6.69386148e-01 -1.08145187e-02 -4.08316478e-02 -6.74162567e-01
3.17906827e-01 4.98678863e-01 3.05214971e-01 -6.66465402e-01
3.28358859e-01 5.42255461e-01 4.27282423e-01 -3.16747576e-01
-1.34094405e+00 -7.48821199e-01 -7.33528852e-01 4.23796438e-02
7.17652202e-01 -5.88385761e-01 -5.82768083e-01 -2.60251582e-01
-6.06923938e-01 4.07258987e-01 -5.81909955e-01 3.91911775e-01
-6.52055621e-01 1.10724911e-01 1.46784440e-01 -4.39544380e-01
-2.56657839e-01 -9.07480299e-01 8.77306283e-01 -3.16270664e-02
-4.13052291e-01 -1.25229073e+00 3.46400067e-02 5.87746918e-01
4.26072448e-01 5.84806502e-01 1.48483431e+00 -1.09032440e+00
-1.93397522e-01 -2.73001552e-01 -1.20315880e-01 7.76731893e-02
7.01087058e-01 -5.43995917e-01 -9.05814767e-01 5.24219535e-02
-3.83939706e-02 -1.15943313e-01 6.49648845e-01 9.16432515e-02
1.35634947e+00 -3.61974686e-01 -6.06530130e-01 3.56834322e-01
1.65707850e+00 3.81971031e-01 4.75911796e-01 9.17937398e-01
6.97187304e-01 9.93502796e-01 1.00940299e+00 6.44715071e-01
3.96293491e-01 9.02861595e-01 4.36924905e-01 1.76864967e-01
1.48841575e-01 -1.94624230e-01 -2.59405941e-01 6.50990188e-01
-1.81972116e-01 6.35276586e-02 -1.00114751e+00 4.36126590e-01
-2.01865625e+00 -9.68380034e-01 -1.03868186e-01 2.21343946e+00
6.87792301e-01 1.18347980e-01 1.04797684e-01 3.99889559e-01
4.56825644e-01 -6.47974387e-02 -2.85533220e-01 -4.48000699e-01
-2.00060457e-02 2.68499732e-01 2.21819833e-01 1.27708226e-01
-9.95215893e-01 5.90177178e-01 5.41681385e+00 9.44823027e-01
-8.01084161e-01 1.83518603e-01 3.02382231e-01 1.02031626e-01
-4.60664332e-01 1.02913693e-01 -5.47630370e-01 3.53081584e-01
9.81969118e-01 -4.27622348e-01 9.98398885e-02 8.50516319e-01
-7.07039563e-03 -2.01250941e-01 -1.32420969e+00 8.78534138e-01
2.28201568e-01 -1.48114204e+00 4.02200341e-01 -1.82173122e-02
4.84601676e-01 -2.75338590e-01 -1.36025950e-01 1.31595835e-01
-1.52055487e-01 -1.05164647e+00 4.72363353e-01 7.28603721e-01
6.55451894e-01 -7.07879245e-01 8.89637113e-01 4.09456007e-02
-1.19731009e+00 -4.04544681e-01 -2.94354856e-01 1.50122076e-01
-2.03608498e-01 6.86256588e-01 -9.80940223e-01 1.36952937e+00
7.69619107e-01 1.05611217e+00 -6.53950512e-01 1.16216004e+00
1.90358818e-01 1.12759419e-01 1.23652676e-02 2.40218505e-01
1.59684896e-01 -5.86068407e-02 2.35232592e-01 1.10101628e+00
4.83003110e-01 -1.05476834e-01 -1.75792724e-03 5.06649196e-01
7.90983886e-02 5.74722946e-01 -7.87082195e-01 1.51979372e-01
3.75927210e-01 1.23510695e+00 -7.06536591e-01 -2.53788233e-01
-6.08283699e-01 4.03635919e-01 1.89267963e-01 -2.64592141e-01
-5.12176633e-01 -1.91439405e-01 7.05518663e-01 3.19853634e-01
3.34104225e-02 4.86444980e-01 -4.24692988e-01 -9.82430875e-01
3.61793637e-02 -9.21725869e-01 9.35844958e-01 -6.37306035e-01
-1.53931689e+00 6.89360142e-01 3.49339128e-01 -1.55649948e+00
-2.49366850e-01 -6.37714803e-01 1.00371176e-02 5.22289515e-01
-1.51036322e+00 -1.19703770e+00 -5.47864676e-01 6.51417196e-01
3.81744623e-01 -3.09983194e-01 1.07357073e+00 4.37202781e-01
-4.82707433e-02 1.87203243e-01 3.18101197e-01 -4.36147779e-01
7.89928675e-01 -1.23705983e+00 -2.36813620e-01 -4.07425538e-02
5.94396330e-02 6.16987586e-01 6.61404133e-01 -6.52924418e-01
-1.03940380e+00 -1.00363338e+00 1.02940738e+00 -5.76942325e-01
6.36913002e-01 -3.96077596e-02 -1.04664648e+00 4.47615355e-01
-2.13234335e-01 2.20084190e-01 1.11654747e+00 3.94798785e-01
-5.71647227e-01 -4.25173968e-01 -1.32683146e+00 1.49897888e-01
1.19574559e+00 -3.79757017e-01 -8.51364374e-01 3.49878937e-01
5.81407905e-01 -7.94839188e-02 -1.34210682e+00 5.06192625e-01
6.94563508e-01 -1.01081753e+00 9.95907843e-01 -8.68069410e-01
6.43201351e-01 -2.59192914e-01 -4.91602272e-01 -1.30921566e+00
-7.66171962e-02 5.51124923e-02 1.26655117e-01 1.15632236e+00
4.12209004e-01 -5.84262192e-01 4.78575855e-01 4.84227568e-01
-2.35498518e-01 -8.68104994e-01 -1.02115119e+00 -8.42872202e-01
3.88869680e-02 -4.26854163e-01 9.46588337e-01 1.36152852e+00
5.49459040e-01 9.87979323e-02 -2.21342310e-01 -2.66068727e-01
3.33855003e-01 2.79283345e-01 2.61307836e-01 -1.85450566e+00
-1.25366852e-01 -1.54969394e-01 -8.85186374e-01 1.19466543e-01
1.09227665e-01 -1.40570700e+00 -4.73354876e-01 -1.80562246e+00
4.34536219e-01 -7.80689061e-01 -7.29240000e-01 6.49718523e-01
6.82549402e-02 3.84379202e-03 2.88587153e-01 1.14229903e-01
-7.63872981e-01 3.42614025e-01 1.09755361e+00 -2.00549334e-01
-7.82402381e-02 -2.51158237e-01 -7.47845471e-01 7.85445154e-01
6.90711439e-01 -7.47088134e-01 -7.25684464e-01 -8.68132710e-02
4.87408370e-01 1.87946498e-01 1.52006745e-01 -1.07642043e+00
1.45304263e-01 -2.76831865e-01 1.40630633e-01 -3.07890922e-01
1.80383459e-01 -1.26670146e+00 6.34822667e-01 4.56189632e-01
-3.89215738e-01 -8.12968537e-02 1.73561290e-01 4.56529438e-01
-5.69867015e-01 -3.72352868e-01 3.55861872e-01 -2.83494562e-01
-8.74758303e-01 5.45396917e-02 6.28379136e-02 9.93914753e-02
1.00241351e+00 -3.06754351e-01 -4.43720132e-01 2.31421575e-01
-1.02343023e+00 -3.88849080e-02 3.13822836e-01 7.33361304e-01
4.76483017e-01 -1.26693654e+00 -2.75151372e-01 -1.46671399e-01
8.34439397e-01 -1.51236236e-01 3.87167573e-01 7.92833209e-01
-2.55869001e-01 7.39107728e-01 -6.53297782e-01 -3.64419907e-01
-1.25590360e+00 8.42742801e-01 3.32298487e-01 -5.33120215e-01
-7.52801776e-01 1.60837442e-01 1.11168645e-01 -4.57134575e-01
-5.57384714e-02 -1.98711291e-01 -5.66300869e-01 7.03537643e-01
3.24331820e-01 3.88755143e-01 6.50715232e-01 -5.38031757e-01
-5.68284810e-01 5.20940125e-01 -1.70422010e-02 3.98011059e-01
1.70276952e+00 -3.60387228e-02 -3.69725317e-01 5.45387745e-01
1.05712950e+00 -1.90636590e-01 -5.17507672e-01 -4.14030582e-01
5.71621060e-01 -6.85040236e-01 -2.07632691e-01 -1.14967620e+00
-9.28954840e-01 5.28720260e-01 5.19412160e-01 1.60450771e-01
1.30122197e+00 2.76305854e-01 5.42428076e-01 4.66155142e-01
6.21673584e-01 -9.15498435e-01 -1.14485070e-01 2.74376422e-01
7.73524046e-01 -1.26135480e+00 2.37083301e-01 -7.94034302e-01
-6.32179201e-01 1.32042527e+00 3.07202429e-01 3.81162524e-01
5.81023991e-01 -9.41734947e-03 -2.41175026e-01 -5.17028153e-01
-7.19705343e-01 -3.39818388e-01 5.23714483e-01 6.94121122e-01
7.24091172e-01 3.56467925e-02 -6.83227420e-01 7.93983042e-01
-2.17395842e-01 1.08467415e-01 1.95770636e-01 9.41935003e-01
-4.56808239e-01 -1.51984859e+00 -1.93056375e-01 8.22736382e-01
-4.17825937e-01 -1.71311647e-01 -2.40322486e-01 8.85317147e-01
5.75221777e-01 6.67757154e-01 -1.79781243e-01 -4.48079854e-01
4.86341417e-01 2.78242141e-01 4.87572849e-01 -6.18564248e-01
-7.66303360e-01 -4.04955149e-01 2.52783567e-01 -5.30261517e-01
-7.56754100e-01 -6.92846358e-01 -1.19769752e+00 -1.59941334e-02
-1.72517672e-02 2.64899999e-01 6.14602447e-01 1.36397457e+00
2.91608095e-01 7.43372560e-01 3.74758214e-01 -2.30672926e-01
-2.36029372e-01 -5.90763450e-01 -5.96900642e-01 7.65382648e-01
4.03971374e-02 -1.15951276e+00 6.29398003e-02 -1.02520637e-01] | [9.023551940917969, 7.924521446228027] |
8f5651d1-54ac-455e-9060-bc4e050562c6 | taskmix-data-augmentation-for-meta-learning | 2210.06341 | null | https://arxiv.org/abs/2210.06341v1 | https://arxiv.org/pdf/2210.06341v1.pdf | TaskMix: Data Augmentation for Meta-Learning of Spoken Intent Understanding | Meta-Learning has emerged as a research direction to better transfer knowledge from related tasks to unseen but related tasks. However, Meta-Learning requires many training tasks to learn representations that transfer well to unseen tasks; otherwise, it leads to overfitting, and the performance degenerates to worse than Multi-task Learning. We show that a state-of-the-art data augmentation method worsens this problem of overfitting when the task diversity is low. We propose a simple method, TaskMix, which synthesizes new tasks by linearly interpolating existing tasks. We compare TaskMix against many baselines on an in-house multilingual intent classification dataset of N-Best ASR hypotheses derived from real-life human-machine telephony utterances and two datasets derived from MTOP. We show that TaskMix outperforms baselines, alleviates overfitting when task diversity is low, and does not degrade performance even when it is high. | ['Surya Kant Sahu'] | 2022-09-26 | null | null | null | null | ['intent-classification'] | ['natural-language-processing'] | [ 3.75008821e-01 8.64349753e-02 -2.73632407e-01 -5.20745516e-01
-1.35858393e+00 -6.72755301e-01 6.91621482e-01 -3.36892694e-01
-5.69711685e-01 9.23634827e-01 3.63990396e-01 -3.09338927e-01
3.23611140e-01 9.93715483e-04 -7.26144016e-01 -4.44712669e-01
5.28579235e-01 8.60916793e-01 -1.35336360e-02 -2.97528833e-01
-9.94102061e-02 -2.58454412e-01 -1.31309557e+00 7.89905250e-01
1.11577916e+00 4.99056607e-01 4.34058547e-01 5.34555197e-01
2.70044431e-03 6.99104309e-01 -9.35556352e-01 -4.47227567e-01
6.80776611e-02 -3.40757102e-01 -1.17870176e+00 -3.59217785e-02
6.90631926e-01 1.74536500e-02 -2.27581263e-01 6.81798756e-01
5.32885849e-01 1.75269723e-01 7.42136419e-01 -1.45123112e+00
-8.91508400e-01 4.85717744e-01 -2.55746752e-01 2.88232207e-01
-1.28247306e-01 -1.14268646e-01 1.02162039e+00 -9.72872913e-01
4.70568299e-01 1.37613499e+00 9.01581109e-01 8.18830013e-01
-1.52800739e+00 -7.37670481e-01 7.77243227e-02 -2.83449516e-03
-9.33851004e-01 -7.52102494e-01 6.22226596e-01 -3.43363196e-01
1.11410534e+00 1.08313099e-01 -5.52565306e-02 1.84080458e+00
3.89157087e-02 1.03884983e+00 1.24561691e+00 -6.05154276e-01
-4.20161057e-03 3.09378505e-01 2.44851410e-01 3.60598117e-01
3.27655524e-02 9.01490264e-03 -6.07091248e-01 -4.37413514e-01
4.62752491e-01 -1.93030030e-01 -2.41286144e-01 -3.42074409e-02
-1.49023664e+00 7.37961352e-01 1.07068874e-01 5.48696876e-01
-2.59421110e-01 5.45233376e-02 4.97122198e-01 7.10254788e-01
8.72682035e-01 8.27338994e-01 -9.93412375e-01 -1.42249674e-01
-7.48162687e-01 1.25588730e-01 7.22810209e-01 1.04351354e+00
9.25897717e-01 2.07111031e-01 -2.21621737e-01 1.38792336e+00
-6.25195652e-02 4.11884964e-01 8.76177490e-01 -9.29085553e-01
9.65037704e-01 3.57048035e-01 1.69252574e-01 -1.23406872e-01
-4.00412947e-01 -5.69410861e-01 -6.27594829e-01 -4.33200479e-01
2.99133658e-01 -5.37315190e-01 -1.05420136e+00 2.04574609e+00
-1.94711462e-01 9.07623172e-02 4.01932478e-01 6.49211407e-01
8.56507659e-01 8.98335159e-01 2.46723250e-01 -2.51308203e-01
1.32970750e+00 -1.39828706e+00 -8.67511928e-01 -8.77711952e-01
9.00723100e-01 -8.33989978e-01 1.38294482e+00 3.47379535e-01
-9.44337130e-01 -8.76871526e-01 -8.23719680e-01 -1.48917288e-01
-3.00307035e-01 1.22878663e-01 5.67678690e-01 5.81776142e-01
-9.65915620e-01 2.85012335e-01 -4.77466553e-01 -4.71711606e-01
1.18091665e-01 1.52152136e-01 -4.51806635e-01 -1.45505846e-01
-1.21923697e+00 1.29536033e+00 7.88802147e-01 -1.95269957e-01
-1.17633998e+00 -8.58235955e-01 -8.58965576e-01 -7.13429004e-02
3.54509652e-01 -7.27519929e-01 1.68016422e+00 -1.12638032e+00
-1.14381468e+00 9.40854669e-01 -3.07973772e-01 -2.84441888e-01
1.93630174e-01 -6.10082150e-01 -6.57220602e-01 -3.75787467e-01
1.82346910e-01 9.47204411e-01 9.62373495e-01 -1.45716679e+00
-7.22842157e-01 -2.83290535e-01 -1.21246405e-01 4.16095614e-01
-6.42236233e-01 -3.73287089e-02 -2.08129287e-01 -8.71196091e-01
-4.95587811e-02 -1.13622832e+00 -1.91118836e-01 -8.36416960e-01
-3.46710235e-01 -3.47907662e-01 8.57730091e-01 -5.70978999e-01
9.54797506e-01 -2.12483287e+00 2.93735534e-01 -4.93803233e-01
-1.15942888e-01 4.31054175e-01 -5.12054384e-01 3.82873058e-01
-1.34172425e-01 1.62860170e-01 -1.12782285e-01 -7.57098615e-01
-1.23596512e-01 5.01724124e-01 -3.82750213e-01 -1.44172516e-02
1.58381432e-01 9.89268661e-01 -1.01964855e+00 -3.44821841e-01
4.40637805e-02 1.86160117e-01 -3.99100095e-01 3.13096672e-01
-4.50408340e-01 6.30365968e-01 -2.35616401e-01 5.68397224e-01
3.53009850e-01 -3.82618755e-01 2.52729833e-01 -1.61254197e-01
1.90676093e-01 6.19287133e-01 -4.76007760e-01 2.11698747e+00
-9.71791506e-01 7.06761599e-01 -1.53461382e-01 -9.27643120e-01
8.71694624e-01 7.32834160e-01 2.07968935e-01 -6.22097313e-01
-1.87831178e-01 3.75267953e-01 -2.31306050e-02 -4.90073115e-01
5.70665479e-01 -3.98837954e-01 -3.27695727e-01 4.82356429e-01
4.59969968e-01 -1.10215321e-01 -1.56027228e-01 -1.66444361e-01
1.05123794e+00 2.27686837e-01 2.99376190e-01 -2.49812558e-01
2.79806823e-01 2.00231209e-01 7.11795032e-01 8.97764564e-01
-7.95405954e-02 5.89904368e-01 3.75906713e-02 -6.03763521e-01
-1.07264543e+00 -1.00959277e+00 -8.26483890e-02 1.92392027e+00
-4.36793596e-01 -2.65458792e-01 -6.38165295e-01 -1.10826588e+00
-8.54346231e-02 9.34033096e-01 -4.90335166e-01 -1.51706561e-01
-8.43299150e-01 -9.25659597e-01 7.39218950e-01 5.29821157e-01
4.48708534e-01 -1.32002866e+00 3.86548862e-02 4.44710463e-01
-8.02733898e-01 -1.35829306e+00 -4.92137700e-01 6.14266038e-01
-1.12063241e+00 -6.55549705e-01 -9.43792105e-01 -1.09153688e+00
6.11782968e-01 3.22519839e-01 1.47998571e+00 -2.42656827e-01
2.14408860e-01 -1.53419152e-02 -3.84104103e-01 -5.21353900e-01
-8.87234628e-01 6.25153244e-01 1.93684310e-01 -2.09885523e-01
4.88657534e-01 -4.51407492e-01 -2.29097884e-02 5.43724000e-01
-7.07557142e-01 -3.92385349e-02 7.61931062e-01 1.21250165e+00
2.72283584e-01 -2.68301934e-01 1.08905768e+00 -9.89419103e-01
6.39229774e-01 -5.99220932e-01 -2.47392952e-01 5.13031840e-01
-4.02081668e-01 3.02312911e-01 6.36724710e-01 -6.22878373e-01
-1.29790604e+00 -5.32440543e-02 7.59751648e-02 -5.22316813e-01
-3.22418272e-01 2.32071623e-01 -8.28442052e-02 3.80138189e-01
8.10636044e-01 6.02135658e-02 -2.58166313e-01 -7.83122480e-01
5.56653619e-01 1.07536316e+00 5.78672290e-01 -6.86934769e-01
5.28089941e-01 1.49759740e-01 -5.81409097e-01 -7.74667621e-01
-1.40979242e+00 -5.73778987e-01 -5.47770143e-01 1.75056055e-01
7.28518903e-01 -1.08909595e+00 -2.17286482e-01 2.24129692e-01
-1.43332434e+00 -5.44882059e-01 -9.85917747e-02 5.03364325e-01
-7.02230990e-01 -3.98438796e-02 -6.47087097e-01 -6.05656862e-01
-3.79388899e-01 -1.13692260e+00 1.07895136e+00 -2.53839284e-01
-4.42388266e-01 -1.11576986e+00 2.89680451e-01 7.18185425e-01
3.30293119e-01 -2.59905457e-01 1.14063692e+00 -1.13056099e+00
-5.98466173e-02 3.20967019e-01 9.62322056e-02 7.31126964e-01
3.92413825e-01 -6.19831145e-01 -1.40563464e+00 -4.91733521e-01
1.99343145e-01 -8.67698431e-01 1.05611289e+00 1.69833675e-01
8.79118919e-01 -5.67817926e-01 -4.58561778e-01 4.34892595e-01
1.01884735e+00 2.44252518e-01 4.73345399e-01 4.71103132e-01
6.23523772e-01 7.77958095e-01 6.77435696e-01 -1.97313264e-01
4.36984688e-01 7.99059927e-01 -2.47795098e-02 -2.02654302e-01
-2.35591605e-01 -3.55157793e-01 5.90505779e-01 1.09053242e+00
1.19569451e-01 -4.20322567e-01 -1.09712219e+00 6.70396984e-01
-1.91491437e+00 -7.78626382e-01 7.57472962e-02 2.12828398e+00
1.09521258e+00 9.12822336e-02 1.21369876e-01 -1.80095047e-01
8.49277854e-01 4.00339998e-02 -4.08210754e-01 -4.78376776e-01
-1.33337706e-01 9.71063320e-03 8.28484818e-02 4.42727834e-01
-1.12992477e+00 1.33621800e+00 7.08172941e+00 7.15252995e-01
-8.68165910e-01 8.02046776e-01 6.17372692e-01 -3.99064906e-02
-2.21615523e-01 -7.83266034e-03 -9.48036253e-01 3.51236612e-01
1.19380844e+00 1.84715599e-01 1.83066234e-01 1.03601897e+00
-2.49535248e-01 3.77826482e-01 -1.41006994e+00 9.04075563e-01
3.61035556e-01 -1.08713031e+00 5.42825200e-02 1.49183776e-02
9.09768105e-01 4.54581946e-01 1.03099443e-01 9.01607692e-01
5.41771114e-01 -9.73487616e-01 5.05840957e-01 -5.22281788e-02
6.32803082e-01 -5.90549290e-01 7.33031094e-01 7.38538742e-01
-6.91704571e-01 -1.86707601e-01 -3.36952806e-01 1.33192793e-01
2.96135664e-01 1.91826150e-01 -1.36927748e+00 3.85629594e-01
4.94558036e-01 3.94412637e-01 -5.83411574e-01 7.38450885e-01
-1.53027460e-01 6.84188545e-01 2.20551342e-02 2.74081141e-01
2.88351715e-01 2.58794069e-01 5.54188788e-01 1.39562845e+00
9.42310020e-02 -2.69082367e-01 3.72830033e-01 6.08859599e-01
-3.03074777e-01 -2.25858632e-04 -9.95549679e-01 6.55708238e-02
5.83761334e-01 1.11079538e+00 -6.47648573e-02 -4.08314586e-01
-5.40473223e-01 1.18073452e+00 6.57362282e-01 5.72637916e-01
-5.92638135e-01 -1.45808071e-01 4.95194882e-01 -1.36574924e-01
9.53184441e-03 -1.15086488e-01 -9.12748743e-03 -1.21859145e+00
-5.71915647e-03 -1.27742457e+00 5.43670714e-01 -7.28441894e-01
-1.61457932e+00 9.17352915e-01 8.67154971e-02 -1.18712652e+00
-7.79126525e-01 -4.15441096e-01 -3.45629513e-01 7.50252843e-01
-1.55621791e+00 -1.36816359e+00 -3.11328620e-02 4.19264615e-01
1.11979365e+00 -4.93367642e-01 9.39280033e-01 2.66836256e-01
-3.99732679e-01 7.51320958e-01 5.86995892e-02 1.38492465e-01
1.23851442e+00 -1.09469402e+00 8.23692858e-01 5.07814348e-01
4.67510819e-01 5.29123187e-01 4.97390032e-01 -6.15913391e-01
-8.46812665e-01 -1.19555259e+00 1.17563140e+00 -9.76987004e-01
5.10933697e-01 -7.52185166e-01 -1.13055658e+00 1.21936214e+00
4.69320893e-01 -1.53965682e-01 8.55800748e-01 6.05717242e-01
-5.93419373e-01 -1.59695745e-02 -9.36255336e-01 4.37857658e-01
1.00659955e+00 -6.10074878e-01 -1.25215423e+00 5.89178562e-01
1.02383733e+00 -2.16067106e-01 -8.93184066e-01 5.99411607e-01
3.02967429e-01 -5.93187332e-01 8.87608886e-01 -1.11783123e+00
3.82331498e-02 1.12061560e-01 -3.04888219e-01 -1.82748508e+00
-3.21341425e-01 -8.78931642e-01 -8.42181668e-02 1.34369183e+00
9.09180164e-01 -5.74387789e-01 5.28668523e-01 1.12338394e-01
-6.41936839e-01 -2.72766083e-01 -9.56299782e-01 -1.25733078e+00
3.81429762e-01 -3.05764318e-01 3.84500831e-01 1.38958895e+00
-3.27041857e-02 9.67723310e-01 -5.24999022e-01 -2.27436684e-02
5.95886528e-01 -2.33931214e-01 7.86248147e-01 -1.37043083e+00
-2.44553134e-01 -3.90819907e-02 1.44950435e-01 -1.02401006e+00
7.30071306e-01 -1.01861072e+00 2.94404477e-01 -1.26872611e+00
1.68778390e-01 -7.47484863e-01 -4.28607821e-01 9.26127076e-01
-4.14505810e-01 1.88528448e-01 1.25279397e-01 5.45463741e-01
-6.46424949e-01 4.97494102e-01 1.13180113e+00 -8.96416903e-02
-4.23799217e-01 4.83371839e-02 -6.57320142e-01 8.70312393e-01
9.45493221e-01 -8.02031517e-01 -4.72314388e-01 -8.59638512e-01
1.42491087e-02 1.68940797e-01 6.26935586e-02 -1.03996634e+00
-2.04594675e-02 3.91216464e-02 3.05750102e-01 -5.74618399e-01
5.82963943e-01 -5.27547836e-01 -8.47794190e-02 1.98591053e-01
-6.24880195e-01 2.69608021e-01 3.97856623e-01 3.07308406e-01
-1.97599143e-01 -4.64553267e-01 6.28909469e-01 -2.75372237e-01
-6.36471689e-01 -2.74050571e-02 -2.05246896e-01 6.09375477e-01
6.29245460e-01 8.67976546e-02 -6.95163369e-01 -2.41570786e-01
-7.16532111e-01 2.10713819e-01 2.96490192e-01 1.00025845e+00
3.20564359e-01 -1.52715552e+00 -1.08717251e+00 1.29312381e-01
4.08680797e-01 -2.63641953e-01 1.09141376e-02 5.11085451e-01
1.80267528e-01 7.36252606e-01 -1.22698136e-02 -7.67423451e-01
-1.08912933e+00 4.14734572e-01 1.27343595e-01 -3.52957159e-01
-2.91260928e-01 7.45676875e-01 3.50589842e-01 -8.89701903e-01
2.11414933e-01 -2.11561799e-01 -3.70020270e-02 7.16821849e-02
4.20363486e-01 2.85957813e-01 2.87666172e-01 -5.84242225e-01
-2.55305678e-01 2.90407926e-01 -5.86030483e-01 -4.56916332e-01
1.21396184e+00 -1.02626421e-01 3.09284151e-01 7.63107240e-01
1.09601247e+00 -3.26894432e-01 -1.12088227e+00 -5.81558049e-01
2.22238615e-01 -2.04165488e-01 -1.43191405e-02 -9.39636648e-01
-6.05126083e-01 8.78196180e-01 2.93927103e-01 1.26663536e-01
8.47647309e-01 2.46079400e-01 8.54822278e-01 9.31213379e-01
5.30225158e-01 -1.16678703e+00 6.15214944e-01 9.00499046e-01
9.71720397e-01 -1.59008944e+00 -3.63225669e-01 -9.74098369e-02
-1.12865269e+00 6.95293009e-01 9.12487924e-01 3.16306412e-01
2.41612524e-01 1.94634408e-01 2.53245533e-01 1.09344259e-01
-1.07279027e+00 -1.83866858e-01 3.86168629e-01 8.35859895e-01
5.77657282e-01 -1.58881783e-01 2.76124895e-01 5.16303539e-01
-3.07533354e-01 -5.32807484e-02 1.54746875e-01 9.54022765e-01
-4.40787882e-01 -1.31930482e+00 -3.47494543e-01 2.58467555e-01
-6.26021862e-01 -4.32832330e-01 -3.10438216e-01 9.70985591e-01
1.48869067e-01 1.08646941e+00 1.59070060e-01 -2.56383330e-01
3.65923822e-01 6.92038357e-01 4.43351358e-01 -1.06579173e+00
-8.96997333e-01 4.71993908e-02 4.94461507e-01 -2.02064112e-01
-3.38015705e-01 -4.54039872e-01 -8.65830183e-01 2.35023886e-01
-4.89836693e-01 3.23095024e-01 6.47978544e-01 1.02569449e+00
4.57168907e-01 4.16214019e-01 5.52188575e-01 -5.79325974e-01
-7.56362498e-01 -1.47369051e+00 -7.46926293e-02 4.57707822e-01
5.73729515e-01 -6.22767746e-01 -3.35726410e-01 6.09544069e-02] | [10.877861022949219, 8.249903678894043] |
02401c3d-b1d3-49c4-a0e7-c5836075ed8a | sheffield-s-submission-to-the-americasnlp | 2306.09830 | null | https://arxiv.org/abs/2306.09830v1 | https://arxiv.org/pdf/2306.09830v1.pdf | Sheffield's Submission to the AmericasNLP Shared Task on Machine Translation into Indigenous Languages | In this paper we describe the University of Sheffield's submission to the AmericasNLP 2023 Shared Task on Machine Translation into Indigenous Languages which comprises the translation from Spanish to eleven indigenous languages. Our approach consists of extending, training, and ensembling different variations of NLLB-200. We use data provided by the organizers and data from various other sources such as constitutions, handbooks, news articles, and backtranslations generated from monolingual data. On the dev set, our best submission outperforms the baseline by 11% average chrF across all languages, with substantial improvements particularly for Aymara, Guarani and Quechua. On the test set, we achieve the highest average chrF of all the submissions, we rank first in four of the eleven languages, and at least one of our submissions ranks in the top 3 for all languages. | ['Danae Sánchez Villegas', 'Edward Gow-Smith'] | 2023-06-16 | null | null | null | null | ['machine-translation'] | ['natural-language-processing'] | [-1.97792143e-01 9.49636474e-02 -5.94878793e-01 -1.92609802e-01
-1.47343135e+00 -1.12662971e+00 1.00221276e+00 7.75633263e-04
-6.26079440e-01 1.42525935e+00 5.66838384e-01 -8.37604225e-01
3.65780443e-01 -3.87412935e-01 -9.49836850e-01 -1.42577171e-01
3.85523021e-01 1.00034809e+00 -1.87651902e-01 -7.12605894e-01
2.02076867e-01 1.71667784e-01 -6.16062462e-01 3.36209953e-01
1.42712379e+00 2.49547377e-01 -1.63396224e-02 3.92316163e-01
-1.30000293e-01 4.00967717e-01 -3.78208399e-01 -7.93022454e-01
2.40809411e-01 -5.81692040e-01 -1.21159029e+00 -6.98901713e-01
6.20401800e-01 -8.33677873e-02 -3.22414786e-02 8.46951365e-01
5.56524098e-01 -3.51690739e-01 6.27935588e-01 -6.10754251e-01
-1.06851888e+00 1.21907294e+00 -6.27700090e-01 3.14778239e-01
6.79218411e-01 -4.60328683e-02 9.08598661e-01 -1.35654974e+00
1.09035492e+00 1.24772012e+00 7.53141701e-01 4.28427577e-01
-1.12491596e+00 -6.01118147e-01 -9.95624810e-02 -5.54960780e-02
-1.43045795e+00 -8.81461859e-01 3.14772725e-02 -4.34606075e-01
1.26085842e+00 2.83753306e-01 2.54895419e-01 1.34052789e+00
-6.91915629e-03 6.40898943e-01 1.51624453e+00 -7.89249539e-01
-1.49416640e-01 1.90859988e-01 -3.35519612e-01 4.93321836e-01
6.24431856e-02 -1.68787628e-01 -5.32201231e-01 -2.42755398e-01
3.82679731e-01 -7.29673445e-01 -4.68468256e-02 3.88519317e-01
-1.73855817e+00 6.75084651e-01 1.76095422e-02 3.39252114e-01
-3.66617382e-01 -3.63632470e-01 2.94830143e-01 6.12199724e-01
9.35752213e-01 4.27182674e-01 -8.65249097e-01 -3.12811792e-01
-9.28312004e-01 4.50941205e-01 1.05514956e+00 1.14108634e+00
5.47522068e-01 -1.44654050e-01 4.43635024e-02 1.21097207e+00
-3.03801801e-02 9.95965481e-01 4.31721449e-01 -9.15111423e-01
1.03863752e+00 1.38718575e-01 4.16752487e-01 -3.81795704e-01
-2.97489285e-01 -1.79558337e-01 -5.16939223e-01 -4.81290430e-01
4.40980792e-01 -5.83828747e-01 -9.15750504e-01 1.75800633e+00
1.36220619e-01 -7.39266515e-01 4.85598594e-01 8.24625731e-01
8.03122759e-01 9.83708203e-01 -1.02627464e-01 -3.14497828e-01
1.18933702e+00 -1.12383580e+00 -4.19959813e-01 -2.38217771e-01
6.14512324e-01 -1.29812896e+00 1.13927662e+00 2.42007911e-01
-1.15926516e+00 -4.23937410e-01 -8.05592120e-01 -2.53825366e-01
-3.87639225e-01 4.11366284e-01 4.27395105e-01 1.21531330e-01
-1.42305338e+00 3.00061911e-01 -6.48601532e-01 -1.05286252e+00
-2.07441568e-01 1.54838845e-01 -6.80584133e-01 -2.25482181e-01
-1.49488163e+00 1.32922029e+00 4.87039447e-01 -2.17352673e-01
-7.70578742e-01 -4.82570827e-01 -5.55803537e-01 -5.28242469e-01
-5.10759391e-02 -3.34803730e-01 1.12919152e+00 -1.00693572e+00
-1.62596226e+00 1.16579080e+00 -1.25829428e-01 -3.89325708e-01
8.56183290e-01 -3.48313689e-01 -7.23883450e-01 -2.29595616e-01
7.50510812e-01 7.86630154e-01 4.12256047e-02 -6.38805628e-01
-6.55950010e-01 -2.45511442e-01 -7.18708336e-02 3.23135048e-01
-1.04349278e-01 7.84235716e-01 -5.15220761e-01 -5.76855421e-01
-1.64296299e-01 -1.18197620e+00 -1.62087396e-01 -7.40007639e-01
-4.19288337e-01 -2.44453490e-01 1.51073813e-01 -1.53396761e+00
9.19770539e-01 -1.85249400e+00 4.30913061e-01 -8.05091932e-02
-4.40702766e-01 -3.07771005e-02 -3.93096626e-01 8.69819164e-01
1.52115956e-01 2.98052102e-01 -2.62492806e-01 8.94585438e-03
-1.81322079e-02 1.16465777e-01 -4.56250489e-01 4.25310582e-01
2.50686437e-01 8.02291691e-01 -9.28845882e-01 -2.45774359e-01
-3.49226981e-01 2.11291656e-01 -2.81036049e-01 -8.69804919e-02
-1.69422612e-01 8.28819990e-01 -1.38507053e-01 4.70058918e-01
5.36812723e-01 2.75228918e-01 3.36622328e-01 3.17433715e-01
-6.07395053e-01 8.16331506e-01 -3.91810328e-01 2.01748395e+00
-6.61573470e-01 5.62648892e-01 -3.35797183e-02 -3.31009090e-01
9.49622273e-01 3.87145489e-01 7.78917596e-02 -8.70761812e-01
-3.94112617e-01 1.10294187e+00 1.38191909e-01 -2.09427819e-01
6.75862312e-01 5.78981042e-02 -3.75792384e-01 5.26703119e-01
1.11010388e-01 -1.10897973e-01 5.08700073e-01 2.34286129e-01
6.32456660e-01 5.37488103e-01 3.46718311e-01 -9.01673377e-01
3.68221909e-01 3.84419620e-01 5.21124661e-01 4.20195550e-01
2.08209291e-01 6.17133677e-01 3.70678365e-01 -4.27466482e-01
-1.48813772e+00 -1.06844735e+00 -2.08798394e-01 1.31424522e+00
-4.22849208e-01 -3.48906398e-01 -8.69629264e-01 -4.78369087e-01
-3.46229553e-01 8.11926961e-01 -4.38386381e-01 3.99351418e-01
-7.73525894e-01 -7.69843578e-01 1.04300797e+00 8.70572552e-02
4.71421301e-01 -9.70818520e-01 2.23076101e-02 1.27242804e-01
-8.11583638e-01 -1.25978768e+00 -5.36776602e-01 -6.22403175e-02
-6.76456034e-01 -4.68427092e-01 -1.37675810e+00 -9.28490520e-01
2.69545913e-01 -4.21675622e-01 1.15582633e+00 -6.14129841e-01
2.62736887e-01 -2.90277869e-01 -3.35255444e-01 -4.22085047e-01
-8.29158962e-01 8.84417117e-01 3.10721934e-01 -4.81957972e-01
2.85749942e-01 -1.80213168e-01 -1.13095948e-02 -8.38892385e-02
-2.14125037e-01 3.15345973e-01 3.58129412e-01 5.46902120e-01
3.82784456e-01 -7.66573429e-01 7.14570642e-01 -9.64228988e-01
5.91331244e-01 -7.53418505e-01 -5.69198251e-01 4.90849257e-01
-3.58895928e-01 -7.88970813e-02 8.72091353e-01 -1.24150790e-01
-1.07622826e+00 -2.19178006e-01 -2.08761841e-01 3.95047307e-01
-1.20118916e-01 7.10141599e-01 6.42537847e-02 2.84399122e-01
8.13324153e-01 1.62874416e-01 -4.26308542e-01 -7.24893570e-01
5.78555524e-01 1.08742583e+00 6.69118404e-01 -1.06388330e+00
4.56044614e-01 -6.22908100e-02 -5.87684274e-01 -7.93816864e-01
-7.04901099e-01 -9.81866866e-02 -7.05559373e-01 1.88982964e-01
7.15783119e-01 -1.28468835e+00 1.02144955e-02 4.62096304e-01
-1.46233904e+00 -4.53711092e-01 8.17514881e-02 6.20047867e-01
-3.66092801e-01 -5.42685762e-02 -1.09235013e+00 -3.28942716e-01
-7.55749762e-01 -1.19473743e+00 1.09659183e+00 4.68464009e-03
-5.31115830e-01 -9.80935574e-01 6.41378939e-01 2.60422915e-01
5.41402042e-01 2.59430081e-01 1.24670732e+00 -6.90841854e-01
1.02756694e-01 1.36358395e-01 -1.50860444e-01 2.01691866e-01
1.99480623e-01 1.31636724e-01 -4.99960035e-01 -4.53962743e-01
-5.18770754e-01 -6.17954433e-01 6.31663501e-01 -3.88611518e-02
1.41492531e-01 -4.99389023e-01 -9.65716317e-02 4.10022378e-01
1.34558833e+00 9.40305218e-02 5.06215096e-01 6.24142170e-01
5.45489013e-01 6.23787463e-01 2.91048497e-01 -1.20220743e-02
8.00048649e-01 7.05854893e-01 -3.93167943e-01 -1.66619018e-01
-3.52217034e-02 -3.73738199e-01 8.82948101e-01 1.39508331e+00
-3.82466912e-01 -1.56516492e-01 -1.39137065e+00 9.41991627e-01
-1.70833993e+00 -5.33346176e-01 -2.12151214e-01 2.17030644e+00
1.46802676e+00 -3.56121689e-01 2.61516988e-01 -8.00263643e-01
5.51721931e-01 -8.76808837e-02 -2.39633694e-01 -8.93079937e-01
-4.32376683e-01 5.08306205e-01 5.98114252e-01 9.51856971e-01
-9.42521274e-01 1.75274611e+00 7.14623880e+00 6.52885139e-01
-1.23275995e+00 2.76585788e-01 5.32028615e-01 -2.78609935e-02
-5.30486822e-01 2.48023435e-01 -8.47238481e-01 3.98132652e-01
1.45473480e+00 -4.10805732e-01 8.80799174e-01 3.28572839e-01
1.02492079e-01 1.54679447e-01 -9.06246364e-01 3.28935742e-01
7.25509897e-02 -1.05735850e+00 4.68625277e-02 3.44734564e-02
1.32150435e+00 1.13277435e+00 -1.05011731e-01 3.23139280e-01
8.53375852e-01 -1.09243298e+00 1.01828718e+00 4.02751297e-01
1.19784832e+00 -7.59496868e-01 6.41781747e-01 4.77363110e-01
-4.14095163e-01 5.16750038e-01 -4.64833915e-01 2.88896542e-02
5.93115613e-02 2.72437721e-01 -7.92380214e-01 9.47174430e-01
6.27216399e-01 6.67167962e-01 -5.19189775e-01 4.99948978e-01
-4.44075346e-01 9.95354235e-01 -5.00092447e-01 1.46481022e-01
5.28268218e-01 -2.52142102e-01 5.25860667e-01 1.54975760e+00
5.53565741e-01 -8.83821696e-02 3.18318069e-01 4.44159836e-01
-3.81400198e-01 7.96838999e-01 -7.12002814e-01 -1.63782462e-01
4.16853160e-01 9.00040090e-01 -2.22942814e-01 -3.82424384e-01
-3.70100915e-01 1.22523880e+00 5.58728456e-01 7.02277482e-01
-7.04070032e-01 -4.19655472e-01 3.50673914e-01 -1.67120010e-01
-7.97272548e-02 -2.71316946e-01 -1.24889647e-03 -1.32440341e+00
5.35576455e-02 -1.35769606e+00 1.66175708e-01 -6.25340700e-01
-1.17894781e+00 1.04408252e+00 3.63266356e-02 -8.06767106e-01
-5.98595619e-01 -4.75610614e-01 -2.55303741e-01 1.44402993e+00
-1.20875180e+00 -1.35524905e+00 5.36059856e-01 2.40763724e-01
4.93347675e-01 -5.83263695e-01 1.06766868e+00 5.36749721e-01
-4.43022370e-01 7.28471041e-01 5.03425062e-01 1.62745073e-01
1.09210455e+00 -9.78057921e-01 1.05815077e+00 9.01744843e-01
1.67028725e-01 8.17945838e-01 5.84202647e-01 -7.31803000e-01
-9.22836900e-01 -9.44886625e-01 1.67926431e+00 -5.12693167e-01
1.03275728e+00 -3.59763384e-01 -5.72322190e-01 1.12929869e+00
8.83644819e-01 -6.38150275e-01 3.51273984e-01 3.07511896e-01
-4.79878187e-01 2.36766875e-01 -8.74772906e-01 6.44179285e-01
8.98544371e-01 -6.31859004e-01 -5.32711089e-01 7.90422499e-01
7.73370564e-01 -6.28472865e-01 -1.07303226e+00 2.20283613e-01
7.10223615e-01 -2.33142123e-01 4.83251601e-01 -1.04773116e+00
8.22155297e-01 -7.72399455e-02 -4.59952831e-01 -1.68648529e+00
-2.25104690e-01 -6.66407168e-01 6.25805080e-01 1.52084839e+00
1.11953700e+00 -7.72498012e-01 6.62468001e-02 -3.42843160e-02
-1.07326530e-01 -4.38131243e-01 -9.75009859e-01 -8.57527554e-01
9.34208274e-01 -2.45846584e-02 5.59002697e-01 1.24352109e+00
-1.33932874e-01 7.79652715e-01 -5.06348968e-01 -1.85758710e-01
3.51411134e-01 8.02665800e-02 8.54691386e-01 -6.91327572e-01
-2.65691251e-01 -3.32547843e-01 2.17053831e-01 -7.28449821e-01
2.79295117e-01 -1.46145678e+00 1.29926845e-01 -1.78391325e+00
3.39305997e-01 -3.50266606e-01 1.17224216e-01 7.56809592e-01
-1.59289062e-01 4.42452371e-01 1.27931580e-01 5.30053318e-01
-2.99381644e-01 1.65254071e-01 9.77744699e-01 -1.02602430e-01
3.49697261e-03 -4.23759073e-01 -8.60856533e-01 5.14249384e-01
7.73081183e-01 -4.62021202e-01 1.02007158e-01 -1.21351516e+00
3.27485740e-01 -1.23437725e-01 -2.04044625e-01 -5.59075356e-01
-1.78956404e-01 -3.38196099e-01 2.49957278e-01 -3.33073378e-01
-1.70438394e-01 -2.43807450e-01 3.34645927e-01 3.82966459e-01
-3.79792541e-01 5.55822849e-01 3.40457857e-01 -2.42515579e-01
-1.15473658e-01 2.04686806e-01 6.41070604e-01 -2.15642959e-01
-4.33157951e-01 4.51646186e-02 -3.70092541e-01 5.82247436e-01
5.41239083e-01 4.46233451e-01 -6.17672920e-01 -2.49642417e-01
-4.42122906e-01 3.47929746e-01 6.58306718e-01 6.77770972e-01
-9.23365280e-02 -1.34487808e+00 -1.65996313e+00 -1.98174417e-02
2.59771168e-01 -5.27926326e-01 -3.59192669e-01 9.07754302e-01
-9.49605405e-01 6.64690256e-01 -3.55927467e-01 -3.43554258e-01
-5.69884062e-01 -1.35589577e-02 7.97669291e-02 -4.15980816e-01
-2.44097367e-01 5.80081284e-01 -1.26211971e-01 -1.37736428e+00
-3.21167409e-01 1.49289697e-01 6.66022077e-02 -1.00897208e-01
3.34398776e-01 3.88437271e-01 1.41658872e-01 -1.02849996e+00
-6.36436820e-01 5.16613185e-01 -3.16976637e-01 -8.07768047e-01
1.25763774e+00 -4.84350957e-02 -6.67388320e-01 5.80372453e-01
1.17792547e+00 5.03334999e-01 -4.96340454e-01 -1.76418766e-01
2.31063008e-01 7.91996494e-02 -4.46697295e-01 -1.40396607e+00
-5.25618255e-01 6.84096456e-01 1.79353282e-01 -3.88656288e-01
7.69328356e-01 1.15698189e-01 7.95378268e-01 3.75127941e-01
8.04228723e-01 -1.25794160e+00 -8.11060190e-01 1.03280735e+00
1.05236495e+00 -1.13903582e+00 -2.13249341e-01 1.02452792e-01
-7.83834457e-01 8.46651733e-01 2.41066575e-01 5.85824139e-02
9.24090147e-02 6.32299632e-02 5.03570318e-01 2.66243756e-01
-7.46129036e-01 1.39941588e-01 2.08163634e-01 3.90054286e-01
9.70546603e-01 4.73107159e-01 -9.90422368e-01 3.86645913e-01
-6.21453762e-01 -4.45641391e-02 4.07471299e-01 4.47466373e-01
-1.17701374e-01 -1.31610680e+00 -1.46631420e-01 1.69799462e-01
-9.34744835e-01 -7.26569891e-01 -8.62020135e-01 9.42730248e-01
-3.25431265e-02 7.34296143e-01 -5.31261340e-02 -2.86309749e-01
2.52716124e-01 1.01103351e-01 5.06188691e-01 -6.40913665e-01
-7.93145537e-01 1.47055805e-01 6.09862626e-01 -1.28493950e-01
-1.85334131e-01 -8.18210542e-01 -1.08585000e+00 -6.53387070e-01
1.38426885e-01 6.22481108e-01 7.93690264e-01 8.89421165e-01
2.30959177e-01 -1.64850205e-01 5.43942094e-01 -4.40029889e-01
-3.83274734e-01 -1.37872791e+00 -1.05282061e-01 -6.98151952e-03
9.79739130e-02 9.75128710e-02 2.99208798e-02 -9.02070329e-02] | [11.529510498046875, 10.305675506591797] |
fb9c6f6d-318a-40da-b2e6-9deae8f807e5 | style-interleaved-learning-for-generalizable | 2207.03132 | null | https://arxiv.org/abs/2207.03132v3 | https://arxiv.org/pdf/2207.03132v3.pdf | Style Interleaved Learning for Generalizable Person Re-identification | Domain generalization (DG) for person re-identification (ReID) is a challenging problem, as access to target domain data is not permitted during the training process. Most existing DG ReID methods update the feature extractor and classifier parameters based on the same features. This common practice causes the model to overfit to existing feature styles in the source domain, resulting in sub-optimal generalization ability on target domains. To solve this problem, we propose a novel style interleaved learning (IL) framework. Unlike conventional learning strategies, IL incorporates two forward propagations and one backward propagation for each iteration. We employ the features of interleaved styles to update the feature extractor and classifiers using different forward propagations, which helps to prevent the model from overfitting to certain domain styles. To generate interleaved feature styles, we further propose a new feature stylization approach. It produces a wide range of meaningful styles that are both different and independent from the original styles in the source domain, which caters to the IL methodology. Extensive experimental results show that our model not only consistently outperforms state-of-the-art methods on large-scale benchmarks for DG ReID, but also has clear advantages in computational efficiency. The code is available at https://github.com/WentaoTan/Interleaved-Learning. | ['Changxing Ding', 'Kui Jia', 'Mingming Gong', 'Pengfei Wang', 'Wentao Tan'] | 2022-07-07 | null | null | null | null | ['generalizable-person-re-identification'] | ['computer-vision'] | [ 7.61943832e-02 -3.04730147e-01 -1.71421707e-01 -5.03384590e-01
-3.14111590e-01 -6.25238419e-01 6.32766187e-01 2.97565684e-02
-5.10075808e-01 7.96941042e-01 5.32817990e-02 1.07980691e-01
-1.51094124e-01 -6.97829485e-01 -5.38028479e-01 -5.88936567e-01
1.10262036e-01 5.76596797e-01 1.80094838e-01 -1.61765561e-01
1.32134125e-01 3.37228596e-01 -1.57815599e+00 4.61903624e-02
1.14835548e+00 6.92844808e-01 -7.36423954e-02 3.90055746e-01
-3.89357507e-01 2.96466082e-01 -5.80368817e-01 -8.19998324e-01
3.87644500e-01 -3.19887549e-01 -7.26557255e-01 5.44216335e-02
3.96818668e-01 -3.89520586e-01 -2.56223261e-01 1.04135132e+00
4.13081348e-01 3.22111577e-01 7.17477798e-01 -1.33806920e+00
-9.60824966e-01 6.04279160e-01 -6.89199686e-01 7.08720007e-04
2.87980914e-01 -4.92118159e-03 7.80262947e-01 -8.31672907e-01
3.12903434e-01 1.40721321e+00 7.88640618e-01 9.62010264e-01
-1.41000843e+00 -1.12400174e+00 6.40156507e-01 1.67442515e-01
-1.45212877e+00 -2.75643975e-01 1.03251731e+00 -2.22928151e-01
4.60206628e-01 1.52818248e-01 5.46092868e-01 1.28620148e+00
-1.40857726e-01 9.79745328e-01 1.09034169e+00 -4.18535203e-01
7.57866129e-02 5.33070207e-01 5.83571255e-01 4.06137496e-01
3.91667038e-01 2.71268338e-01 -5.53346038e-01 -3.19345534e-01
7.26717949e-01 1.95140883e-01 -9.73113552e-02 -4.22171593e-01
-8.88091445e-01 7.22163618e-01 3.75314444e-01 1.33149311e-01
-3.83416899e-02 -2.54316896e-01 3.99103969e-01 4.94562328e-01
4.83003259e-01 3.05359125e-01 -5.79498351e-01 -1.42786026e-01
-6.17491841e-01 6.32347345e-01 5.99377513e-01 1.02227557e+00
8.87300551e-01 -2.16304600e-01 -1.66457757e-01 1.24712825e+00
1.29606709e-01 4.26258504e-01 7.05115855e-01 -4.53500688e-01
4.36959207e-01 8.75000298e-01 1.56999275e-01 -6.91671908e-01
-3.21644008e-01 -6.65126085e-01 -9.08318043e-01 -7.41977757e-03
5.79682529e-01 -2.95220762e-01 -9.53358054e-01 1.96093583e+00
4.02899325e-01 4.40198034e-01 -1.45257292e-02 4.75643635e-01
6.56989455e-01 4.37243611e-01 1.97653741e-01 1.41065300e-01
1.26869202e+00 -8.22551370e-01 -4.50133651e-01 -3.76290649e-01
6.09820366e-01 -5.60811222e-01 1.21942532e+00 3.92918676e-01
-6.96633875e-01 -9.12136555e-01 -1.00162077e+00 1.51617274e-01
-4.93499994e-01 2.02407494e-01 5.16308248e-01 9.30969357e-01
-6.22702181e-01 5.29779255e-01 -4.77936864e-01 -2.64850616e-01
4.69426602e-01 5.05922735e-01 -3.96080673e-01 -5.18763587e-02
-1.18763101e+00 4.82668698e-01 5.07320404e-01 -7.36600682e-02
-2.12128699e-01 -9.86100614e-01 -7.96244860e-01 -8.53540897e-02
1.87817976e-01 -7.03251600e-01 1.34306192e+00 -1.22274947e+00
-1.58377922e+00 6.24693215e-01 -3.47749114e-01 -2.80434400e-01
7.08619118e-01 -5.27426541e-01 -5.86127281e-01 -4.09455448e-01
8.33400637e-02 5.58377445e-01 1.03415763e+00 -1.41384637e+00
-9.21043158e-01 -3.58658046e-01 -1.77675948e-01 2.64522552e-01
-7.63718903e-01 -1.21973440e-01 -5.91574907e-01 -9.57351685e-01
-1.89999163e-01 -1.02395904e+00 4.93147224e-02 -2.80434340e-01
-3.41418475e-01 -4.09076154e-01 7.75891185e-01 -5.35051703e-01
1.31264293e+00 -2.23414469e+00 1.33970052e-01 4.02485132e-01
1.38028979e-01 4.00374264e-01 -1.06229044e-01 1.29452989e-01
-1.32974274e-02 -6.71545835e-03 -1.72025964e-01 -6.70936346e-01
-7.03659356e-02 2.06595227e-01 -2.61435419e-01 1.49432465e-01
6.52732551e-02 6.33829176e-01 -9.02069807e-01 -2.90491045e-01
3.37254629e-02 3.65356445e-01 -6.44448817e-01 3.70243818e-01
5.93703873e-02 6.68275654e-01 -3.37657064e-01 4.65782851e-01
9.37487245e-01 -2.42259949e-02 7.23352581e-02 -1.13375541e-02
5.55885993e-02 7.72074088e-02 -1.44879317e+00 1.34501553e+00
-4.07389611e-01 9.02681872e-02 -5.57105482e-01 -6.94230258e-01
1.27129877e+00 -1.00022376e-01 2.82575130e-01 -6.14730477e-01
4.54858243e-02 2.80254781e-01 -1.44432455e-01 -1.83659829e-02
4.38144058e-01 -4.89222072e-02 -3.23170602e-01 3.71671081e-01
1.05812073e-01 4.37806636e-01 2.72306561e-01 -3.11271846e-02
5.70854306e-01 1.62334293e-01 1.33838400e-01 -1.23734446e-02
8.48950684e-01 -3.04327667e-01 1.02262414e+00 8.71070385e-01
-1.12430543e-01 3.88256937e-01 2.77381033e-01 -5.03793538e-01
-8.71236742e-01 -1.19062972e+00 -1.13522947e-01 1.24866760e+00
3.32352132e-01 -3.88139248e-01 -7.02342629e-01 -1.11078000e+00
3.13931346e-01 6.66022837e-01 -7.41520882e-01 -4.50053543e-01
-6.95784986e-01 -8.23624134e-01 6.15311265e-01 6.85891807e-01
8.76010835e-01 -8.01304281e-01 8.34121462e-03 2.33623579e-01
-4.18996438e-02 -9.09319341e-01 -7.13758469e-01 -4.93380316e-02
-7.61520922e-01 -9.16572094e-01 -1.08237982e+00 -9.54072177e-01
8.36060047e-01 2.04815120e-01 8.63597035e-01 1.28550753e-01
2.12814659e-01 1.98778301e-01 -5.16451538e-01 -4.53398615e-01
-4.79759812e-01 4.36171263e-01 3.35652292e-01 3.88615340e-01
7.81684935e-01 -5.30171931e-01 -4.51857358e-01 4.24037606e-01
-7.27885246e-01 1.18683346e-01 5.39581060e-01 1.08869636e+00
4.42473352e-01 2.42850214e-01 8.34737122e-01 -1.19072497e+00
6.85095131e-01 -4.59813267e-01 -5.21358550e-01 2.87211627e-01
-7.35552728e-01 2.02142775e-01 9.40656364e-01 -8.59114707e-01
-1.38401949e+00 1.17535718e-01 -1.76908374e-01 -2.27701902e-01
-4.44700718e-01 9.19975042e-02 -4.61270392e-01 -1.28367171e-01
5.57893813e-01 4.12599325e-01 -8.18081945e-02 -9.22988892e-01
2.16795638e-01 7.58256376e-01 6.04884028e-01 -7.67611027e-01
1.18906569e+00 2.18023539e-01 -4.58791614e-01 -4.96588677e-01
-6.97465718e-01 -5.49796581e-01 -8.98366392e-01 3.39234062e-02
3.25581551e-01 -7.65170097e-01 -4.26451415e-01 1.02056384e+00
-8.78470659e-01 -2.76442498e-01 -1.96143329e-01 2.58937597e-01
-2.11837038e-01 5.57679415e-01 -3.92852426e-01 -5.88767290e-01
-4.24474537e-01 -9.08054709e-01 7.72413492e-01 6.72737956e-01
-3.45864564e-01 -1.00854647e+00 8.49967301e-02 7.58851832e-03
2.23939061e-01 -1.22287393e-01 9.89307284e-01 -9.85998213e-01
2.50786468e-02 -2.27771848e-01 -1.38828382e-01 4.45214033e-01
5.52319288e-01 -3.23075354e-01 -9.68034565e-01 -5.16633034e-01
-3.31977844e-01 -5.02113700e-02 7.40449548e-01 -1.82455312e-02
1.32816374e+00 -2.77008235e-01 -4.61625636e-01 8.22956562e-01
9.77945268e-01 2.00727150e-01 3.86186868e-01 5.90871155e-01
8.42845082e-01 5.10349870e-01 5.11115670e-01 4.59935158e-01
4.98792350e-01 8.40282440e-01 -2.86779255e-01 -1.27011299e-01
-1.36515722e-01 -5.64558804e-01 3.77754331e-01 4.10203487e-01
-7.06872940e-02 6.32248595e-02 -7.53920317e-01 5.18233538e-01
-1.80876744e+00 -8.36857438e-01 1.90035105e-01 2.49617648e+00
1.09271979e+00 3.07757497e-01 6.65352404e-01 2.48839960e-01
7.42010891e-01 -2.21962124e-01 -7.86629736e-01 -3.26986253e-01
2.91067865e-02 1.60631731e-01 5.09199560e-01 3.57216835e-01
-1.24452007e+00 9.85032618e-01 5.41684866e+00 8.09725821e-01
-1.02649891e+00 -8.60154107e-02 4.68667060e-01 7.37694278e-03
-1.82036355e-01 -1.89790085e-01 -1.34240127e+00 8.44662368e-01
5.16179264e-01 -3.52065444e-01 3.44440877e-01 8.64236951e-01
-1.32506549e-01 2.14724600e-01 -1.16433835e+00 1.19401336e+00
-7.73012415e-02 -1.01799536e+00 3.31524551e-01 -3.01086660e-02
6.91662908e-01 -4.52561200e-01 1.96153924e-01 5.55048883e-01
3.80991161e-01 -5.80119193e-01 6.31224453e-01 3.15397680e-01
7.06187189e-01 -1.14783645e+00 7.02607095e-01 2.83242583e-01
-1.31049776e+00 -4.59970623e-01 -3.65531862e-01 2.46167302e-01
-1.15610942e-01 4.19136226e-01 -6.44072890e-01 6.63887143e-01
8.91206920e-01 7.41747856e-01 -9.53941464e-01 1.16380799e+00
-1.17971048e-01 4.76913691e-01 -2.01672390e-01 1.78050533e-01
-9.47008282e-02 -7.64348032e-03 5.75187266e-01 1.22073662e+00
2.07224160e-01 -2.85326451e-01 2.51794040e-01 7.51885533e-01
-1.50285259e-01 1.28690764e-01 -3.78169686e-01 2.72475958e-01
7.61012673e-01 9.01303053e-01 -4.15734202e-01 -3.66853148e-01
-5.07186413e-01 1.26726735e+00 4.24498677e-01 4.91463691e-01
-7.16250777e-01 -5.28816223e-01 8.95716667e-01 1.80835009e-01
2.53476769e-01 -1.46198183e-01 -3.56556237e-01 -1.18364298e+00
1.20875716e-01 -1.03986049e+00 7.21686840e-01 1.16266854e-01
-1.84081244e+00 5.61215103e-01 2.84692287e-01 -1.45355797e+00
-2.37914339e-01 -4.84169573e-01 -6.70589685e-01 9.72378612e-01
-1.66105628e+00 -1.33521903e+00 -4.49343681e-01 7.54691422e-01
6.50607049e-01 -4.45205927e-01 7.21532404e-01 3.67752939e-01
-8.06209385e-01 1.41046441e+00 1.86052188e-01 2.84690559e-01
1.07451963e+00 -1.26487064e+00 7.50686526e-01 8.21051002e-01
-6.72080889e-02 8.56293142e-01 5.22227764e-01 -8.11080933e-01
-1.06501603e+00 -1.22475624e+00 7.31653690e-01 -3.02951038e-01
3.32723618e-01 -4.60884690e-01 -1.22229981e+00 7.23210633e-01
-2.32818916e-01 -2.68275112e-01 9.64744210e-01 3.82993609e-01
-5.64064980e-01 -3.02325606e-01 -1.18067563e+00 6.67539001e-01
1.13883626e+00 -2.31095627e-01 -7.04296887e-01 -5.63709438e-02
3.51204157e-01 -3.65720928e-01 -7.13807166e-01 3.73973191e-01
7.04680383e-01 -7.42122769e-01 1.05127537e+00 -4.63058978e-01
-1.91364199e-01 -2.57903576e-01 2.67990053e-01 -1.56775975e+00
-6.37837708e-01 -6.17755771e-01 -2.74145782e-01 1.63373411e+00
2.87827879e-01 -1.09263909e+00 7.49459565e-01 6.92742884e-01
1.44132942e-01 -5.77366173e-01 -6.80652380e-01 -1.25287902e+00
2.82482058e-01 -1.62415564e-01 1.16688299e+00 8.24081004e-01
-2.07601622e-01 2.25022569e-01 -4.10832286e-01 2.31556132e-01
8.94667685e-01 6.41154051e-02 1.09786272e+00 -1.60150647e+00
-4.41019386e-01 -5.23521304e-01 -2.27848768e-01 -1.23186374e+00
3.00358355e-01 -8.82589817e-01 -2.90953606e-01 -9.52351093e-01
2.30462909e-01 -8.70343268e-01 -3.82668704e-01 6.32854044e-01
-5.91994524e-01 1.54777333e-01 2.00923204e-01 3.07036906e-01
-2.33507648e-01 5.96622348e-01 9.25766587e-01 -1.35415554e-01
-7.37591684e-01 3.04854721e-01 -1.08732152e+00 6.99406922e-01
1.02362704e+00 -4.59168464e-01 -5.57732463e-01 -3.38389367e-01
-2.29530051e-01 -7.65137076e-01 4.09224480e-01 -1.13692629e+00
7.61432275e-02 -5.30468039e-02 8.17378581e-01 -3.81727159e-01
1.37440771e-01 -7.50007629e-01 5.04914299e-02 2.82362252e-01
-2.91640311e-01 -7.94311017e-02 3.65415394e-01 5.31971574e-01
4.77373227e-02 -1.90950304e-01 8.48455608e-01 1.56984612e-01
-9.55589831e-01 4.97847408e-01 -8.69873837e-02 -1.68819219e-01
8.88640463e-01 -4.85778809e-01 -1.96570739e-01 6.35530194e-03
-6.26749635e-01 4.01107550e-01 6.03491366e-01 7.86638141e-01
4.91982847e-01 -1.48409164e+00 -7.28145897e-01 6.53397858e-01
3.62800568e-01 -5.62917404e-02 2.82602131e-01 4.09652829e-01
3.38651286e-03 4.02105078e-02 -3.10093880e-01 -4.08623099e-01
-1.36253893e+00 5.04146278e-01 4.14016217e-01 -2.59713620e-01
-7.54402280e-01 8.51321340e-01 3.63922179e-01 -6.98570430e-01
3.49326462e-01 -4.17533033e-02 -4.31342721e-01 -1.08081833e-01
8.78530443e-01 4.13798422e-01 -1.19516790e-01 -5.86570978e-01
-3.03288281e-01 6.77897274e-01 -7.47210205e-01 1.96049407e-01
1.12903404e+00 -2.66384363e-01 3.14821303e-01 2.81810135e-01
1.04361701e+00 9.84703228e-02 -1.34437513e+00 -5.87372899e-01
2.08162978e-01 -6.47054613e-01 -2.95956463e-01 -7.76633263e-01
-8.15107644e-01 3.76553535e-01 8.21805954e-01 -3.59051079e-02
1.24755716e+00 -1.94649145e-01 1.11837518e+00 2.92892516e-01
3.56654435e-01 -1.26704931e+00 3.28944298e-03 4.98303801e-01
7.23974228e-01 -1.15532911e+00 -1.56574190e-01 -4.64671165e-01
-6.68809950e-01 1.06306148e+00 9.36467826e-01 -1.39077857e-01
5.71439683e-01 4.18614186e-02 6.18634224e-02 3.09513867e-01
-3.10132504e-01 -5.03202006e-02 4.04805571e-01 8.11195731e-01
2.91270405e-01 9.40704197e-02 -1.67851642e-01 1.07405162e+00
-2.88840294e-01 4.53205965e-02 1.23068899e-01 9.30163503e-01
-1.96715310e-01 -1.77841413e+00 -5.76858461e-01 5.00082672e-01
-1.55316308e-01 1.89189211e-01 -5.00415206e-01 6.44051850e-01
4.12506670e-01 8.18253517e-01 -2.47645415e-02 -6.36802077e-01
5.24915516e-01 2.67511934e-01 5.26116610e-01 -5.89471877e-01
-6.28372073e-01 -1.41177550e-01 -4.69247848e-02 -1.38793975e-01
-5.07799201e-02 -7.66511559e-01 -1.15163994e+00 -4.85035121e-01
-1.02773495e-01 1.14528731e-01 1.23184524e-01 9.00122225e-01
3.78359079e-01 2.93021351e-01 9.37886178e-01 -7.47625828e-01
-5.81973910e-01 -8.50851059e-01 -5.43119967e-01 6.82901144e-01
2.92901754e-01 -1.05592740e+00 -1.14093825e-01 5.38954884e-02] | [14.726839065551758, 1.100634217262268] |
13a11af7-e026-4327-bfe0-ab0661362b03 | evaluating-language-models-for-knowledge-base | 2303.11082 | null | https://arxiv.org/abs/2303.11082v1 | https://arxiv.org/pdf/2303.11082v1.pdf | Evaluating Language Models for Knowledge Base Completion | Structured knowledge bases (KBs) are a foundation of many intelligent applications, yet are notoriously incomplete. Language models (LMs) have recently been proposed for unsupervised knowledge base completion (KBC), yet, despite encouraging initial results, questions regarding their suitability remain open. Existing evaluations often fall short because they only evaluate on popular subjects, or sample already existing facts from KBs. In this work, we introduce a novel, more challenging benchmark dataset, and a methodology tailored for a realistic assessment of the KBC potential of LMs. For automated assessment, we curate a dataset called WD-KNOWN, which provides an unbiased random sample of Wikidata, containing over 3.9 million facts. In a second step, we perform a human evaluation on predictions that are not yet in the KB, as only this provides real insights into the added value over existing KBs. Our key finding is that biases in dataset conception of previous benchmarks lead to a systematic overestimate of LM performance for KBC. However, our results also reveal strong areas of LMs. We could, for example, perform a significant completion of Wikidata on the relations nativeLanguage, by a factor of ~21 (from 260k to 5.8M) at 82% precision, usedLanguage, by a factor of ~2.1 (from 2.1M to 6.6M) at 82% precision, and citizenOf by a factor of ~0.3 (from 4.2M to 5.3M) at 90% precision. Moreover, we find that LMs possess surprisingly strong generalization capabilities: even on relations where most facts were not directly observed in LM training, prediction quality can be high. | ['Gerhard Weikum', 'Simon Razniewski', 'Sneha Singhania', 'Blerta Veseli'] | 2023-03-20 | null | null | null | null | ['knowledge-base-completion', 'knowledge-base-completion'] | ['graphs', 'knowledge-base'] | [-1.83672607e-01 9.17944670e-01 -3.85038406e-01 -2.10640311e-01
-1.20767462e+00 -6.46609187e-01 7.73257434e-01 3.22362989e-01
-6.31106436e-01 1.41770482e+00 3.45745951e-01 -4.06770080e-01
-4.43346888e-01 -9.06068742e-01 -9.80010808e-01 -2.01434940e-01
-7.26606473e-02 1.02136528e+00 4.04155523e-01 -5.38728952e-01
-1.54745551e-02 4.83314395e-02 -1.29247510e+00 5.70404291e-01
1.08016288e+00 6.09088659e-01 -2.27621216e-02 4.21541363e-01
-2.87999928e-01 1.07522035e+00 -8.33948076e-01 -1.04100287e+00
-8.25236142e-02 1.70702130e-01 -1.40536284e+00 -5.43331921e-01
5.04739344e-01 -9.21845529e-03 -1.45193070e-01 8.64319384e-01
2.86299735e-01 -1.84644446e-01 6.87095821e-01 -1.01222169e+00
-5.25058746e-01 1.14120150e+00 9.59971473e-02 1.92889366e-02
7.07234919e-01 -7.65337050e-02 1.18299532e+00 -7.98119783e-01
1.05414402e+00 7.32942283e-01 9.30519044e-01 3.69051605e-01
-1.14191401e+00 -5.13339281e-01 -8.07385445e-02 1.95727080e-01
-1.57121122e+00 -6.22119725e-01 1.12226076e-01 -4.43066567e-01
1.33065081e+00 4.23348099e-01 5.60732543e-01 9.00510550e-01
-1.81814834e-01 5.25511920e-01 1.06799901e+00 -7.20270991e-01
3.14312652e-02 8.12359750e-01 2.16351867e-01 5.44635236e-01
8.18568349e-01 -3.93649280e-01 -8.08780849e-01 -3.38434905e-01
2.34515697e-01 -7.88719118e-01 -4.35694456e-01 6.79169968e-02
-1.09179246e+00 3.71025950e-01 8.42156857e-02 2.34854355e-01
-2.67654806e-01 -8.70228484e-02 2.44295999e-01 4.18022543e-01
4.28124309e-01 1.04120994e+00 -9.85930860e-01 -3.90266210e-01
-8.08027506e-01 6.25050783e-01 1.30950522e+00 1.22951019e+00
7.56837368e-01 -4.64472532e-01 1.59242198e-01 7.32972383e-01
-9.52150226e-02 5.69118857e-01 4.33446169e-01 -8.63072336e-01
8.81245017e-01 7.84401417e-01 5.04789829e-01 -1.09891534e+00
-5.16944230e-01 -4.19959813e-01 -5.86657584e-01 -2.97742277e-01
6.31708682e-01 -5.97555265e-02 -4.89061952e-01 1.56403160e+00
1.20799482e-01 -2.47886881e-01 4.67853218e-01 5.26101291e-01
1.04320824e+00 3.89337540e-01 1.67500764e-01 -2.35113353e-02
1.25321829e+00 -5.78162074e-01 -4.74299729e-01 -1.70858830e-01
1.18603313e+00 -7.03790128e-01 1.26653886e+00 4.74214345e-01
-1.06414461e+00 -3.56941313e-01 -9.50760961e-01 4.33388874e-02
-6.78195894e-01 9.20660868e-02 9.12649930e-01 7.42299914e-01
-1.11662102e+00 5.03295898e-01 -4.33366686e-01 -5.45544147e-01
1.41556635e-01 2.26655573e-01 -5.96862733e-01 -1.98739633e-01
-1.53985739e+00 1.36976707e+00 7.74017155e-01 -3.09616536e-01
-5.27955532e-01 -9.60276544e-01 -5.13001204e-01 -6.41223043e-02
7.81273186e-01 -5.12407899e-01 1.03849244e+00 -4.25744832e-01
-1.00947368e+00 7.22586572e-01 -4.31747437e-02 -7.01343119e-01
2.89505363e-01 -5.07083714e-01 -7.71588087e-01 -5.07023558e-02
2.83630103e-01 3.26788604e-01 -8.70917458e-03 -1.36300290e+00
-7.65659809e-01 -6.21520914e-02 4.35707539e-01 1.90352499e-01
-4.13564354e-01 -1.08369015e-01 -6.62689149e-01 -4.55927610e-01
-6.60989657e-02 -7.04489231e-01 -8.21269304e-02 -7.28911817e-01
-5.14965773e-01 -8.14516693e-02 -3.94239388e-02 -9.88782227e-01
1.58678794e+00 -1.58335400e+00 -4.06755805e-02 3.12624484e-01
2.42006093e-01 3.30683798e-01 5.65548316e-02 6.32662833e-01
2.68927962e-01 5.61826110e-01 -9.88386869e-02 -6.67340010e-02
-1.06102236e-01 1.79609209e-01 -4.14747417e-01 -6.47222102e-02
2.02727556e-01 1.04231703e+00 -9.53201711e-01 -5.85527539e-01
-3.04743260e-01 3.34017649e-02 -4.43904221e-01 -2.95537889e-01
-3.20884734e-01 -9.97425690e-02 -2.22068980e-01 7.54305303e-01
1.86248511e-01 -4.91524041e-01 5.27161300e-01 -1.46206394e-01
5.36741614e-02 7.41088569e-01 -1.10566795e+00 1.47624648e+00
-3.85497600e-01 5.58450818e-01 -4.73891318e-01 -8.10739279e-01
6.63206160e-01 4.76418257e-01 2.99209327e-01 -5.17980635e-01
-4.49142724e-01 4.21109200e-01 5.02132736e-02 -4.71764356e-01
9.93048251e-01 -1.12536237e-01 -9.45855305e-02 2.28948608e-01
2.53753513e-01 -1.58791810e-01 4.62140977e-01 5.85069478e-01
1.33244944e+00 -1.16313465e-01 3.04401278e-01 -3.77936482e-01
3.85152400e-01 5.41399956e-01 4.90889966e-01 8.69536757e-01
2.65053988e-01 2.53621459e-01 5.61481237e-01 -3.85843694e-01
-9.51568007e-01 -7.82046020e-01 -2.15365365e-01 5.48600972e-01
-9.68268067e-02 -1.07626188e+00 -4.88194972e-01 -8.96629274e-01
1.34453446e-01 8.72486055e-01 -2.50559598e-01 -6.95547462e-02
-3.56064141e-01 -7.94814348e-01 1.07308769e+00 4.40705806e-01
3.00014108e-01 -9.27541614e-01 -1.32477164e-01 2.43086830e-01
-4.88227278e-01 -1.39755392e+00 3.65550131e-01 -5.69449365e-03
-7.01801360e-01 -1.13927376e+00 -3.74321997e-01 -4.77266818e-01
5.23488283e-01 -2.23644897e-01 1.67173314e+00 8.31718892e-02
7.68064260e-02 4.19136792e-01 -5.63216388e-01 -4.94119465e-01
-5.54254532e-01 5.58664322e-01 3.83117378e-01 -6.77765310e-01
5.31083763e-01 -5.00903189e-01 -2.07556650e-01 2.35275403e-01
-5.84322810e-01 -4.29916605e-02 7.39652336e-01 7.02620566e-01
2.52860636e-01 3.03595573e-01 8.51270914e-01 -1.51360214e+00
6.03920937e-01 -5.83014667e-01 -4.19997364e-01 6.81008399e-01
-1.01863372e+00 7.82036558e-02 5.20626307e-01 -1.78204402e-01
-1.22377098e+00 -2.64053524e-01 5.83305582e-02 3.40348035e-01
-8.40979144e-02 1.23240697e+00 -7.89407194e-02 9.99468416e-02
1.25590599e+00 1.14584826e-01 -3.32520902e-01 -4.79549468e-01
6.23906314e-01 6.13675058e-01 5.80386937e-01 -1.02694952e+00
8.12922180e-01 2.69512892e-01 -3.44390273e-01 -7.21687973e-01
-1.06301665e+00 -4.87266153e-01 -5.39697945e-01 -6.84059337e-02
2.11819991e-01 -1.10579467e+00 -5.91610312e-01 2.90646423e-02
-9.08544958e-01 -6.85578704e-01 -2.59928733e-01 4.57444400e-01
-2.35900253e-01 3.35254967e-01 -6.23313129e-01 -7.12594986e-01
-4.32014257e-01 -5.74464440e-01 6.69513762e-01 -1.10161014e-01
-6.37914658e-01 -1.10670674e+00 1.21944714e-02 7.41739690e-01
2.53775120e-01 -2.17596702e-02 1.28714955e+00 -8.00661325e-01
-4.45913404e-01 -4.16878134e-01 -1.91056475e-01 3.37624162e-01
7.90100619e-02 -1.19864121e-01 -8.56723189e-01 -1.35062978e-01
-6.48693264e-01 -6.07798576e-01 5.68318486e-01 -1.85678691e-01
6.95648372e-01 -4.25565720e-01 -5.26047587e-01 6.91153929e-02
1.44166362e+00 4.83325608e-02 8.04315209e-01 5.72311819e-01
4.77744311e-01 7.66257465e-01 7.02593505e-01 2.76399553e-01
7.30983317e-01 7.41376638e-01 -2.90129445e-02 3.30025166e-01
-1.59149542e-01 -4.44460422e-01 2.70069569e-01 1.06735289e+00
-6.62517965e-01 -9.46598276e-02 -1.42736661e+00 7.07151651e-01
-1.61083615e+00 -7.13605046e-01 -2.92188674e-01 2.41183853e+00
1.47195864e+00 6.43390119e-01 -3.49396169e-02 1.42599791e-01
2.02357933e-01 -3.69485587e-01 -1.12705164e-01 2.00856373e-01
-3.11287075e-01 3.25449497e-01 6.12365484e-01 5.99463463e-01
-7.57638335e-01 1.21650040e+00 6.42910767e+00 9.52194035e-01
-6.30466461e-01 -2.12141559e-01 4.70529288e-01 -6.92519546e-02
-4.91986006e-01 3.80638480e-01 -1.21540022e+00 4.51667815e-01
1.09159803e+00 -3.60403091e-01 3.18105191e-01 8.35469365e-01
-4.27155703e-01 -4.27799344e-01 -1.00860345e+00 8.02318037e-01
2.37820242e-02 -1.61880231e+00 2.20427707e-01 2.45384369e-02
8.68348837e-01 -3.09887491e-02 -2.55315453e-01 8.76493573e-01
5.23899019e-01 -1.18240499e+00 6.56703949e-01 8.02413166e-01
9.97726679e-01 -5.98093152e-01 1.07046020e+00 5.49271762e-01
-7.85920978e-01 2.40828171e-01 -6.22793078e-01 -2.68581808e-01
8.85786787e-02 9.65357721e-01 -1.22577584e+00 8.90403688e-01
7.38057733e-01 3.69690239e-01 -8.52994621e-01 6.49448037e-01
-3.29745919e-01 8.76575589e-01 -3.96160454e-01 -1.30460590e-01
-1.49161965e-01 2.30259925e-01 2.37593666e-01 1.30101132e+00
3.81372571e-02 1.36075959e-01 -5.11725470e-02 5.58357179e-01
-2.52557844e-01 2.89420336e-01 -5.30418694e-01 -1.16390519e-01
8.47549200e-01 1.18083346e+00 -2.62231171e-01 -6.05328262e-01
-4.62877333e-01 5.77528417e-01 8.46771061e-01 3.37310135e-01
-5.89389682e-01 -4.79070395e-01 3.12446862e-01 7.39289746e-02
-1.13054596e-01 3.62158194e-02 -2.26122946e-01 -1.20857978e+00
4.80426461e-01 -9.95468378e-01 3.81312668e-01 -6.03218496e-01
-1.28230941e+00 4.93702948e-01 1.56098321e-01 -6.91578388e-01
-3.59278798e-01 -6.35298908e-01 -8.53656903e-02 9.31675613e-01
-1.26909578e+00 -1.01837921e+00 -1.86930239e-01 3.20798129e-01
1.46374926e-02 -3.32329452e-01 1.09121060e+00 5.11772811e-01
-2.79970884e-01 6.46258533e-01 7.27314427e-02 9.31259319e-02
8.24250519e-01 -1.41687644e+00 5.74133694e-01 5.09949863e-01
4.35101330e-01 1.05298042e+00 6.04136050e-01 -1.02074051e+00
-1.09159338e+00 -9.55792189e-01 1.44524741e+00 -1.26612341e+00
7.56916165e-01 -2.19921887e-01 -9.71059501e-01 8.07786167e-01
-1.96660802e-01 -2.88279533e-01 7.53391445e-01 9.22485888e-01
-4.37263519e-01 -1.50787115e-01 -8.91047239e-01 6.32185578e-01
1.15303051e+00 -6.36980712e-01 -6.49330914e-01 4.54527110e-01
7.11470068e-01 -3.90526295e-01 -1.33273077e+00 5.97825587e-01
6.86886370e-01 -7.33251393e-01 8.78992558e-01 -9.69690323e-01
5.19386232e-01 -2.09878415e-01 -3.11267138e-01 -1.18215024e+00
-2.29398787e-01 -4.51925367e-01 -2.95774996e-01 1.49535155e+00
1.16140747e+00 -5.05661726e-01 1.04603910e+00 9.46043313e-01
-5.04321158e-02 -9.29899275e-01 -6.44813836e-01 -1.01496387e+00
1.07241511e-01 -8.47074807e-01 4.55316275e-01 1.22753084e+00
4.96195525e-01 4.35297012e-01 -2.57518381e-01 2.41853341e-01
2.35345721e-01 -2.23411202e-01 1.03248465e+00 -1.29858947e+00
-4.08103406e-01 -1.35155365e-01 -3.53387356e-01 -7.50108242e-01
-5.31404279e-02 -1.00376666e+00 -2.74674118e-01 -1.58186245e+00
3.69524062e-01 -8.86128485e-01 -9.33035612e-02 6.66075945e-01
-2.71283597e-01 9.42499265e-02 -4.54037748e-02 3.16654533e-01
-5.18754661e-01 9.47334468e-02 7.21770823e-01 -3.20598967e-02
-1.24259077e-01 -2.52095461e-01 -9.01417673e-01 7.99163759e-01
8.67747247e-01 -3.08596373e-01 -5.46250582e-01 -3.54942530e-01
1.02396560e+00 -9.53122154e-02 2.75741041e-01 -9.88735020e-01
2.87720352e-01 3.84418517e-02 1.55484930e-01 -2.86953837e-01
3.17575783e-01 -4.17234629e-01 2.36065403e-01 1.75627753e-01
-3.50052446e-01 -3.24387968e-01 2.70325273e-01 3.62462670e-01
-2.25591823e-01 -4.81009424e-01 -1.23969711e-01 -3.16818357e-01
-7.07311749e-01 -1.63645566e-01 -1.59957260e-01 4.72312957e-01
4.47060019e-01 1.24938451e-01 -6.57719314e-01 -4.82062101e-01
-5.83021104e-01 6.73117489e-02 2.67612129e-01 8.36909637e-02
3.07533145e-01 -1.14836168e+00 -7.29322851e-01 -3.23203832e-01
5.14184713e-01 1.81522101e-01 -6.03721142e-02 7.73744524e-01
-5.68786621e-01 7.90662706e-01 2.35171914e-01 -7.52568692e-02
-9.67977881e-01 3.31949472e-01 -1.49384558e-01 -6.08918011e-01
-4.68493462e-01 7.54965782e-01 -3.93997461e-01 -5.42926908e-01
1.85098633e-01 -4.83289480e-01 -1.58894375e-01 2.73759011e-02
4.24663782e-01 4.30765033e-01 4.01723862e-01 -2.53178716e-01
-2.58916795e-01 2.29075745e-01 -3.73077244e-01 -1.93610340e-01
1.25103843e+00 2.16253847e-01 -1.66546345e-01 4.51599747e-01
6.28492415e-01 5.28595686e-01 -4.78806913e-01 -2.71578550e-01
5.63660443e-01 -1.62782013e-01 -4.73135382e-01 -1.36003768e+00
-4.59457844e-01 2.45558575e-01 -2.20270976e-01 3.24912846e-01
7.65104413e-01 2.68476844e-01 4.57533896e-01 9.63545799e-01
9.24407423e-01 -1.04817426e+00 -3.01211208e-01 6.10055029e-01
8.39028895e-01 -1.25463963e+00 3.80845904e-01 -6.36775613e-01
-6.40072763e-01 6.66047812e-01 6.10530853e-01 4.60693538e-01
3.40748072e-01 2.45219454e-01 -3.74246128e-02 -3.86734903e-01
-9.31602001e-01 -1.33632660e-01 2.86171079e-01 6.30740225e-01
5.81034184e-01 1.72545478e-01 -3.54052156e-01 1.01547849e+00
-6.95852995e-01 -3.79663073e-02 4.30960506e-01 6.86492801e-01
-3.23445439e-01 -1.12647569e+00 -7.37705231e-02 6.76943541e-01
-3.66667956e-01 -5.03297031e-01 -4.68225896e-01 1.08921897e+00
5.54962605e-02 8.27772796e-01 -4.30148125e-01 -5.91936171e-01
4.61559236e-01 4.07123446e-01 3.76477301e-01 -6.98972225e-01
-2.79368430e-01 -6.43706679e-01 9.02107298e-01 -2.90221214e-01
-5.01277447e-01 -4.80405837e-01 -1.02953768e+00 -2.77821064e-01
-4.92868572e-01 4.11788315e-01 3.71569365e-01 9.08740044e-01
3.23430061e-01 1.39770582e-01 -1.79211631e-01 -4.81301732e-02
-3.92823070e-01 -1.14772022e+00 -5.05153537e-01 4.26433593e-01
-3.57896030e-01 -5.54458380e-01 -1.68299377e-01 3.28485161e-01] | [9.508665084838867, 8.399956703186035] |
c6fb00a2-eabd-4df5-8457-63bf415298e4 | hynet-local-descriptor-with-hybrid-similarity | 2006.10202 | null | https://arxiv.org/abs/2006.10202v3 | https://arxiv.org/pdf/2006.10202v3.pdf | HyNet: Learning Local Descriptor with Hybrid Similarity Measure and Triplet Loss | Recent works show that local descriptor learning benefits from the use of L2 normalisation, however, an in-depth analysis of this effect lacks in the literature. In this paper, we investigate how L2 normalisation affects the back-propagated descriptor gradients during training. Based on our observations, we propose HyNet, a new local descriptor that leads to state-of-the-art results in matching. HyNet introduces a hybrid similarity measure for triplet margin loss, a regularisation term constraining the descriptor norm, and a new network architecture that performs L2 normalisation of all intermediate feature maps and the output descriptors. HyNet surpasses previous methods by a significant margin on standard benchmarks that include patch matching, verification, and retrieval, as well as outperforming full end-to-end methods on 3D reconstruction tasks. | ['Axel Barroso-Laguna', 'Yurun Tian', 'Krystian Mikolajczyk', 'Vassileios Balntas', 'Tony Ng'] | 2020-06-17 | null | http://proceedings.neurips.cc/paper/2020/hash/52d2752b150f9c35ccb6869cbf074e48-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/52d2752b150f9c35ccb6869cbf074e48-Paper.pdf | neurips-2020-12 | ['patch-matching'] | ['computer-vision'] | [ 2.07617983e-01 -1.94829985e-01 -2.78581619e-01 -5.46704650e-01
-9.68105614e-01 -2.90395021e-01 9.05393779e-01 3.06474745e-01
-4.36583519e-01 1.01441227e-01 3.31842840e-01 1.77073389e-01
-3.68927956e-01 -5.96510291e-01 -7.12786555e-01 -6.30080283e-01
-2.18515620e-01 1.30763367e-01 2.20056847e-01 -9.53220427e-02
3.38439614e-01 9.61913228e-01 -1.56431413e+00 1.88204885e-01
2.65533358e-01 1.46165872e+00 -3.22673351e-01 2.78710037e-01
3.53431851e-01 5.34754038e-01 -3.03505421e-01 -4.87285614e-01
7.67927051e-01 -2.06181169e-01 -5.24459720e-01 -1.86811835e-01
1.30016208e+00 -3.85460794e-01 -4.84164089e-01 8.59327614e-01
1.03757834e+00 2.94771016e-01 4.92142588e-01 -9.60876167e-01
-9.34349954e-01 8.23783055e-02 -2.16966510e-01 1.33652940e-01
6.15527272e-01 2.40765452e-01 1.22620726e+00 -1.27684045e+00
9.46404040e-01 1.36570990e+00 1.32134926e+00 2.96351492e-01
-1.50802064e+00 -6.25260115e-01 -2.67572775e-02 2.58055151e-01
-1.77583933e+00 -6.35328472e-01 7.59686887e-01 -1.76887915e-01
1.42071784e+00 -6.39688969e-02 5.45402348e-01 9.04135048e-01
3.32783550e-01 8.95869493e-01 8.97819042e-01 -5.08835495e-01
-3.42505649e-02 -1.33273080e-01 -1.33606166e-01 6.27349198e-01
-8.07070732e-02 7.37944484e-01 -7.60271132e-01 -1.93111807e-01
5.93315482e-01 -1.37415096e-01 -1.66917354e-01 -9.10645604e-01
-9.61725116e-01 8.64055097e-01 8.30351710e-01 3.53904888e-02
-4.97846961e-01 4.30863231e-01 5.89821458e-01 6.55345976e-01
7.08244443e-01 4.84555572e-01 -2.00022250e-01 -1.49877891e-01
-1.14487088e+00 5.28421938e-01 7.53825486e-01 8.13312709e-01
9.07196224e-01 -1.92357674e-01 -3.76646549e-01 1.02716660e+00
3.34902078e-01 5.95394790e-01 9.25591215e-02 -9.39016342e-01
2.34091401e-01 6.04385734e-01 -4.05499518e-01 -1.29833043e+00
-3.20079774e-01 -4.61857259e-01 -6.41657889e-01 4.86271948e-01
1.76855296e-01 4.20251846e-01 -8.80599082e-01 1.75728858e+00
2.22045809e-01 1.74785778e-01 -2.88789123e-01 1.09421897e+00
8.88610840e-01 2.13112950e-01 -1.52428290e-02 4.51172829e-01
7.41545379e-01 -1.04629767e+00 -3.96406531e-01 -2.69825896e-03
4.74206030e-01 -1.00224411e+00 7.20370173e-01 1.40090585e-01
-1.26275289e+00 -8.15651596e-01 -1.27028060e+00 -4.70047355e-01
-4.50296849e-01 -1.39794081e-01 7.02593684e-01 3.41415644e-01
-1.25226557e+00 1.16094637e+00 -8.91246140e-01 -6.42655790e-01
4.74879771e-01 4.91981417e-01 -7.37917840e-01 -4.05103147e-01
-1.08049989e+00 1.21062148e+00 -6.83973432e-02 1.16672605e-01
-6.88831508e-01 -1.08140695e+00 -1.18223083e+00 -4.47434522e-02
-9.56462696e-03 -8.51506472e-01 1.00334716e+00 -4.93557155e-01
-1.47158074e+00 1.19241428e+00 2.10146502e-01 -4.31147128e-01
6.29119039e-01 -2.48022631e-01 -1.33820370e-01 5.81020191e-02
-1.16365235e-02 1.09601450e+00 7.55552113e-01 -6.52688980e-01
-3.78904194e-01 -2.98497051e-01 -1.43946826e-01 9.84070748e-02
-9.76736769e-02 -1.18244775e-01 -5.16937733e-01 -7.77579725e-01
-2.08154004e-02 -9.80535150e-01 -7.23192915e-02 7.09489346e-01
1.23228831e-02 -3.89711022e-01 6.72847748e-01 -3.29703867e-01
1.06805086e+00 -2.48990679e+00 1.09527171e-01 5.11971176e-01
1.85671821e-01 3.58946413e-01 -5.62988997e-01 6.15102589e-01
-1.48712993e-01 -2.41976336e-01 -1.67669803e-01 -6.06200695e-01
5.86256504e-01 2.08935160e-02 -1.22625239e-01 9.23499882e-01
3.55292737e-01 1.02589381e+00 -6.26044333e-01 -2.25986481e-01
4.51795220e-01 9.58671927e-01 -6.73196971e-01 1.70542687e-01
3.32061797e-01 -2.56584227e-01 -1.10939499e-02 6.45621002e-01
9.43066001e-01 -1.03796043e-01 -3.52979988e-01 -3.25326830e-01
-2.64183991e-03 4.24650729e-01 -1.11489177e+00 2.22721601e+00
-4.62681949e-01 8.11673999e-01 4.97809472e-03 -8.25093389e-01
1.08547151e+00 -8.03297833e-02 3.94920498e-01 -1.01634634e+00
3.20518166e-02 4.26368207e-01 -2.61108398e-01 -2.61333934e-03
1.88614532e-01 -1.71615686e-02 1.32027790e-01 2.68054634e-01
2.51589686e-01 -2.28976935e-01 -8.24439712e-03 -9.66309160e-02
1.25249994e+00 3.16666603e-01 1.36279896e-01 -2.65924603e-01
5.36985099e-01 -3.42796504e-01 2.38305703e-01 9.16939437e-01
-1.58686787e-01 8.76515806e-01 2.20976785e-01 -6.11431837e-01
-1.05567002e+00 -1.05096221e+00 -4.94907051e-01 9.87266541e-01
1.44645423e-01 -5.71090102e-01 -4.23109919e-01 -6.06672943e-01
7.26428330e-01 1.56899616e-01 -8.56298745e-01 -5.49434841e-01
-5.38678825e-01 -1.46700114e-01 8.39210510e-01 5.75358629e-01
5.66105425e-01 -1.12132394e+00 -3.56446058e-01 2.46981919e-01
5.40657520e-01 -1.14350200e+00 -6.47420287e-01 3.25613916e-01
-8.10334682e-01 -9.03065026e-01 -7.96285987e-01 -8.45436811e-01
4.88163590e-01 7.50995055e-02 1.22469604e+00 2.43982092e-01
-5.57864547e-01 3.93958777e-01 -2.04182625e-01 -4.31717113e-02
-1.69579595e-01 4.27879691e-01 -3.41319323e-01 -3.27158570e-01
6.21715605e-01 -6.68785393e-01 -7.30480492e-01 2.82893896e-01
-8.53968322e-01 -4.38588649e-01 8.41605783e-01 9.94688511e-01
5.70043087e-01 -7.00986862e-01 1.77432179e-01 -3.53320390e-01
6.41130269e-01 -2.88504474e-02 -5.73272884e-01 1.30889356e-01
-1.03202736e+00 3.74237865e-01 3.60070646e-01 -3.50877613e-01
-4.78845865e-01 -1.22466702e-02 -4.29655254e-01 -5.54096103e-01
-1.30285770e-01 4.62751031e-01 2.26293772e-01 -8.47080588e-01
7.75593817e-01 -7.36514255e-02 3.29592645e-01 -6.17605269e-01
3.62450927e-01 2.99734950e-01 4.18691576e-01 -4.32023108e-01
9.07265246e-01 4.03609008e-01 3.31055880e-01 -6.36667550e-01
-7.29496837e-01 -6.76851392e-01 -6.61754906e-01 1.58045247e-01
4.67399329e-01 -8.73537362e-01 -5.84412873e-01 5.11003077e-01
-9.92908359e-01 -2.48407543e-01 -5.76160312e-01 4.42502618e-01
-6.15169525e-01 3.07115406e-01 -6.10072434e-01 -2.57871542e-02
-5.44446766e-01 -1.11228585e+00 1.44491541e+00 -1.78364739e-01
-9.12250876e-02 -1.10104775e+00 4.86779839e-01 -2.05330402e-01
9.00624573e-01 3.06598753e-01 6.31863475e-01 -4.55469817e-01
-5.82697093e-01 -6.31995082e-01 -4.93147999e-01 6.62473381e-01
-2.67983496e-01 -2.03690022e-01 -1.00283909e+00 -5.57074428e-01
-5.22583425e-01 -4.65641290e-01 1.22338736e+00 3.16256464e-01
7.65937805e-01 1.73908487e-01 -1.23403057e-01 1.25377405e+00
1.44834208e+00 -3.89560610e-01 7.40436256e-01 5.83460033e-01
5.12667239e-01 3.88795495e-01 3.69892061e-01 3.23610604e-01
2.48862356e-01 1.15636444e+00 2.64829963e-01 -4.77572680e-01
-7.71380484e-01 -3.01693410e-01 3.30976129e-01 3.85812759e-01
2.59176999e-01 1.23788878e-01 -6.70613050e-01 5.46813130e-01
-1.80762100e+00 -8.03182304e-01 3.95045340e-01 2.14564919e+00
5.80027163e-01 2.94926502e-02 -1.04613632e-01 -1.62021682e-01
3.91161740e-01 4.93070751e-01 -6.59063041e-01 -4.21420872e-01
-2.86876529e-01 7.64537454e-01 6.32020950e-01 5.83633304e-01
-1.24382019e+00 1.04741633e+00 6.62283754e+00 9.49533999e-01
-1.27584279e+00 -8.97730216e-02 7.73662031e-02 6.63932189e-02
-5.56295924e-02 1.01993978e-01 -7.62098491e-01 2.65755225e-02
5.92130780e-01 2.30841950e-01 3.46657544e-01 8.97350132e-01
-2.39901409e-01 2.22099856e-01 -1.48239005e+00 1.19808519e+00
4.02257055e-01 -1.26113725e+00 4.70957197e-02 1.31542519e-01
7.16443181e-01 3.98715526e-01 2.06858337e-01 3.44838589e-01
-1.46755859e-01 -9.65352297e-01 6.14734948e-01 5.18073857e-01
7.92396545e-01 -7.34474003e-01 9.47537541e-01 -2.15807989e-01
-1.23918056e+00 5.90263531e-02 -5.71655095e-01 5.39943092e-02
4.18943316e-02 4.53840762e-01 -5.68976641e-01 3.72828394e-01
5.97754896e-01 1.09126723e+00 -9.89739120e-01 1.38586366e+00
-3.18470329e-01 1.45607069e-01 -5.43778837e-01 2.48080030e-01
5.76769710e-01 1.62168667e-01 4.28239733e-01 1.48213851e+00
1.86063811e-01 -4.33727980e-01 2.60750592e-01 8.28812659e-01
-1.87987372e-01 7.81542808e-02 -6.86906993e-01 3.26074183e-01
3.10359150e-01 1.06720483e+00 -2.25868106e-01 -4.96721379e-02
-3.30969810e-01 1.08084416e+00 3.59266728e-01 1.43093511e-01
-4.58440185e-01 -8.64497423e-01 8.72778833e-01 1.22080125e-01
5.79280972e-01 -1.30837277e-01 2.05717571e-02 -8.33337128e-01
3.13913316e-01 -7.80308664e-01 2.23836392e-01 -4.95421410e-01
-1.59660959e+00 4.01164114e-01 -3.50383595e-02 -1.11818981e+00
-4.57344949e-01 -7.61877894e-01 -3.62762988e-01 8.58833849e-01
-1.90901542e+00 -1.32073903e+00 -5.56259215e-01 5.15010953e-01
9.62922163e-03 -2.04864949e-01 8.70812714e-01 7.15244770e-01
-1.08709335e-01 1.10470343e+00 -2.89289784e-02 1.00285754e-01
1.20643175e+00 -1.12676299e+00 9.60463226e-01 4.92281139e-01
8.94600078e-02 6.80567861e-01 3.83683056e-01 -1.76689595e-01
-1.28719997e+00 -5.86303711e-01 1.08097875e+00 -3.54374290e-01
6.01882458e-01 -4.43476081e-01 -8.01680505e-01 2.33994931e-01
1.92743257e-01 7.25207448e-01 4.47211802e-01 1.38370946e-01
-8.09498727e-01 -3.41245830e-01 -1.17428803e+00 1.64887205e-01
1.32657659e+00 -1.04767358e+00 -3.81828755e-01 2.69616395e-01
3.75724852e-01 -4.97733355e-01 -1.18666005e+00 5.67807436e-01
1.00579011e+00 -1.04245055e+00 1.38267708e+00 -3.93360734e-01
2.06361458e-01 -9.13926736e-02 -3.95461142e-01 -1.07393754e+00
-6.00703120e-01 -5.77870548e-01 6.26666918e-02 9.18587983e-01
3.25101316e-01 -4.06580657e-01 7.81811833e-01 4.15448010e-01
-3.45614612e-01 -9.98495996e-01 -1.03802729e+00 -9.92563903e-01
2.25411013e-01 -3.48652899e-01 5.41614830e-01 7.15551019e-01
-3.24941009e-01 -5.49395308e-02 -2.54877925e-01 -1.22462407e-01
8.00298214e-01 3.53760226e-03 9.26571250e-01 -1.04187500e+00
6.94590248e-03 -8.54325175e-01 -1.20569086e+00 -1.44115162e+00
4.60721701e-01 -1.22881174e+00 1.07625328e-01 -1.34869599e+00
5.78217804e-02 -4.31771696e-01 -4.10958707e-01 5.15419126e-01
4.86714281e-02 6.17373347e-01 3.70142668e-01 1.00806274e-01
-6.89781308e-01 7.43275583e-01 1.01226711e+00 -2.05773950e-01
-7.04461560e-02 -2.78536618e-01 -2.16505364e-01 2.51229197e-01
3.42619956e-01 -5.14966905e-01 -5.40500600e-03 -5.45361519e-01
6.65056109e-02 -4.86104161e-01 5.44745982e-01 -1.07210815e+00
4.23818499e-01 4.67263490e-01 4.81865019e-01 -4.30218756e-01
4.20230538e-01 -8.30908895e-01 -1.87907562e-01 2.81664610e-01
-6.26558185e-01 3.62114280e-01 2.27535427e-01 2.41593793e-01
-5.62377036e-01 1.90264266e-03 8.74172211e-01 1.55956477e-01
-7.79112875e-01 4.84570771e-01 2.39076272e-01 6.54579923e-02
5.82732499e-01 -5.34550190e-01 -2.17408106e-01 -1.25992477e-01
-4.63291258e-01 1.72719490e-02 7.28426158e-01 5.95763445e-01
7.33216882e-01 -1.80093133e+00 -8.65666628e-01 6.50853932e-01
3.21177661e-01 -3.58956158e-01 2.79329307e-02 9.71587062e-01
-6.71425879e-01 4.06463951e-01 -3.18641394e-01 -8.07200551e-01
-1.07065690e+00 3.16633999e-01 4.68186796e-01 -2.46603891e-01
-8.07808042e-01 9.08509135e-01 -1.30969957e-01 -8.02418113e-01
7.39488661e-01 -3.24616075e-01 3.43003452e-01 2.04631705e-02
3.51012558e-01 3.64031732e-01 4.40166563e-01 -5.80044448e-01
-6.59232080e-01 1.19093609e+00 -1.61498100e-01 1.79734960e-01
1.47852230e+00 1.22211456e-01 3.43879797e-02 -4.83364500e-02
2.00525594e+00 -3.29081297e-01 -1.28706276e+00 -4.21926558e-01
8.60181004e-02 -7.38844633e-01 3.61216933e-01 -7.80031681e-01
-1.07882094e+00 8.59303713e-01 1.06657386e+00 -1.35704577e-01
1.01123488e+00 -1.77681446e-02 1.03703094e+00 6.20077789e-01
7.36163110e-02 -9.86705899e-01 5.33102117e-02 6.23852134e-01
1.04171240e+00 -1.31415915e+00 3.89721930e-01 6.11502007e-02
-1.43596590e-01 9.89054739e-01 2.39930242e-01 -6.88099384e-01
9.15416718e-01 1.50754467e-01 2.44404346e-01 -4.16807324e-01
-6.72716022e-01 -3.47337604e-01 7.98448861e-01 6.07438147e-01
4.02945578e-01 -4.05256003e-01 -1.07146487e-01 -3.16547751e-01
3.94269824e-02 -5.30135967e-02 -3.72274041e-01 9.49158430e-01
-1.67195275e-01 -1.39950883e+00 -1.68571055e-01 6.18882068e-02
-4.05638337e-01 -1.86945274e-01 -5.70764780e-01 1.02584517e+00
1.78024516e-01 4.30551082e-01 8.95524863e-03 -3.63011897e-01
8.47455859e-01 -1.44246608e-01 7.67907798e-01 -3.09646517e-01
-8.92961204e-01 -9.18607116e-02 -1.67248473e-01 -1.00757766e+00
-5.38697124e-01 -6.57590449e-01 -7.66772509e-01 -5.29690027e-01
-5.02518952e-01 -1.78575233e-01 7.92809546e-01 6.03410244e-01
6.73103809e-01 3.21841277e-02 5.69036484e-01 -1.21182048e+00
-8.45643163e-01 -8.91582489e-01 -4.38564390e-01 7.94807613e-01
5.83354592e-01 -8.64974797e-01 -6.72176838e-01 -4.51581508e-01] | [8.264126777648926, -1.8211110830307007] |
9ed2570b-52c7-48cb-8909-f006f07b1a8e | on-the-computation-communication-trade-off | 2306.07159 | null | https://arxiv.org/abs/2306.07159v1 | https://arxiv.org/pdf/2306.07159v1.pdf | On the Computation-Communication Trade-Off with A Flexible Gradient Tracking Approach | We propose a flexible gradient tracking approach with adjustable computation and communication steps for solving distributed stochastic optimization problem over networks. The proposed method allows each node to perform multiple local gradient updates and multiple inter-node communications in each round, aiming to strike a balance between computation and communication costs according to the properties of objective functions and network topology in non-i.i.d. settings. Leveraging a properly designed Lyapunov function, we derive both the computation and communication complexities for achieving arbitrary accuracy on smooth and strongly convex objective functions. Our analysis demonstrates sharp dependence of the convergence performance on graph topology and properties of objective functions, highlighting the trade-off between computation and communication. Numerical experiments are conducted to validate our theoretical findings. | ['Jinming Xu', 'Yan Huang'] | 2023-06-12 | null | null | null | null | ['stochastic-optimization'] | ['methodology'] | [-3.59003365e-01 -9.08422023e-02 -3.89021903e-01 -1.73507109e-01
-6.25792682e-01 -6.99603438e-01 3.07580233e-02 3.13556850e-01
-5.08616447e-01 9.72347319e-01 -3.34120125e-01 -3.03759634e-01
-6.53089643e-01 -6.62777483e-01 -4.35782343e-01 -8.06542933e-01
-7.96263695e-01 1.81589097e-01 3.15873511e-02 -1.43654734e-01
8.96362364e-02 5.12104273e-01 -3.45419377e-01 -8.88099670e-01
9.77868557e-01 1.16612697e+00 5.32253124e-02 9.22625661e-01
2.22447559e-01 3.06265771e-01 -5.99375129e-01 -3.68598461e-01
4.46825147e-01 -3.04241717e-01 -3.20411056e-01 3.04426134e-01
-1.62659869e-01 -3.54783647e-02 -3.57915252e-01 1.24643815e+00
7.13267446e-01 -4.21964042e-02 3.97571534e-01 -1.42715132e+00
-4.69483644e-01 6.97180092e-01 -8.76222968e-01 3.98481101e-01
1.17775179e-01 1.51377544e-01 8.00347149e-01 -6.51302338e-01
3.92997384e-01 8.04343998e-01 8.56206596e-01 3.71419251e-01
-1.11383045e+00 -7.58470953e-01 4.06730294e-01 -2.33815551e-01
-1.61224258e+00 -3.03467065e-01 8.71435285e-01 -9.02341753e-02
2.50733554e-01 2.90788352e-01 6.10026658e-01 4.00303811e-01
5.10359287e-01 9.75746438e-02 7.51294315e-01 -6.46959320e-02
4.48404163e-01 2.13797733e-01 -1.17679745e-01 8.54734063e-01
5.63057423e-01 -1.85528964e-01 -4.68843699e-01 -4.92492467e-01
9.03703511e-01 -1.43714935e-01 -4.76851732e-01 -4.14517552e-01
-9.74403620e-01 6.89679205e-01 6.45069659e-01 -1.16843827e-01
-5.21734595e-01 5.19618392e-01 3.34339410e-01 6.68555081e-01
5.71440279e-01 -1.08518638e-01 -2.62447357e-01 -4.35102396e-02
-6.35033429e-01 -1.62873223e-01 1.09146774e+00 1.05558944e+00
5.30667186e-01 1.59129784e-01 -1.33538872e-01 3.58576536e-01
2.65946656e-01 7.71412194e-01 -1.56914368e-01 -8.22972953e-01
9.11218703e-01 3.28246444e-01 4.55037266e-01 -1.37535477e+00
-6.29495323e-01 -7.87531495e-01 -1.23278809e+00 -6.31438941e-02
3.40971768e-01 -1.09821594e+00 -1.98076561e-01 1.93446195e+00
7.20685303e-01 4.72375043e-02 -2.24033862e-01 1.08500624e+00
6.68509351e-03 5.22468328e-01 -2.98122644e-01 -6.99085772e-01
5.70705354e-01 -6.20506287e-01 -6.32511795e-01 8.09231922e-02
6.25049829e-01 -3.17286432e-01 8.60947490e-01 -1.67410985e-01
-1.38002849e+00 1.81042522e-01 -1.05282462e+00 5.37739456e-01
1.12428881e-01 2.05830768e-01 3.59974325e-01 9.02712226e-01
-1.24488926e+00 4.44815546e-01 -8.83569717e-01 -1.35064736e-01
2.05312923e-01 6.95147276e-01 2.49192029e-01 3.95666391e-01
-7.38029718e-01 2.32764393e-01 1.19850961e-02 6.45266950e-01
-8.54731083e-01 -8.26335251e-01 -2.66416699e-01 1.76728338e-01
3.91566336e-01 -7.25734055e-01 7.94081211e-01 -6.05208278e-01
-1.75954485e+00 2.59419531e-01 -7.06491694e-02 -4.65307266e-01
9.76708710e-01 8.21318626e-02 -1.63944453e-01 1.86865136e-01
-2.72811521e-02 -2.18708813e-01 6.24838233e-01 -1.02091646e+00
-5.40574193e-01 -4.79265183e-01 1.71100169e-01 3.89431298e-01
-8.92335951e-01 -9.05656517e-02 -3.11630577e-01 -3.14224362e-01
-1.76496387e-01 -1.02082515e+00 -6.58448040e-01 4.97635156e-01
-4.92088467e-01 9.32692811e-02 8.79420877e-01 -2.34073505e-01
1.21941495e+00 -1.69463551e+00 1.74095348e-01 8.20168555e-01
6.37149215e-01 -1.93416029e-01 -6.07282855e-02 5.26535153e-01
8.23340893e-01 1.59762383e-01 1.09669760e-01 -2.76657820e-01
-7.32752457e-02 -3.10323015e-02 3.38926502e-02 1.16529393e+00
-2.83823013e-01 6.30348444e-01 -9.65525687e-01 -3.16147298e-01
-7.02107996e-02 4.00938302e-01 -4.98038709e-01 -4.69859056e-02
1.87893048e-01 3.67291749e-01 -1.03693891e+00 3.48729849e-01
6.83523953e-01 -6.44254982e-01 6.23219430e-01 8.06443170e-02
-2.92358473e-02 -1.51346609e-01 -1.31185007e+00 1.13750112e+00
-8.52851927e-01 4.92842138e-01 1.24366748e+00 -1.09873664e+00
6.87500119e-01 2.25546613e-01 6.09582305e-01 -1.51470572e-01
2.16836467e-01 3.35654527e-01 -2.56502390e-01 1.13314025e-01
1.97922960e-01 1.31471758e-03 -9.82148293e-03 5.86134613e-01
-3.50438803e-01 1.01779237e-01 2.09944338e-01 3.89152020e-01
9.74243999e-01 -8.15187871e-01 3.15015353e-02 -9.08527195e-01
6.03459179e-01 -3.55034500e-01 5.64602971e-01 7.84453392e-01
-4.14941251e-01 -2.25900292e-01 5.77700734e-01 -7.48232529e-02
-7.95542121e-01 -1.04449761e+00 2.32113257e-01 1.13621366e+00
7.21157610e-01 -1.80372387e-01 -5.94720721e-01 -5.01868427e-01
1.82878762e-01 -8.88302326e-02 -5.12959719e-01 -1.28703313e-02
-4.43193674e-01 -9.23240602e-01 3.56200546e-01 8.07158798e-02
4.59622145e-01 -5.94347827e-02 -3.78636718e-01 2.74094254e-01
4.06956494e-01 -1.14233780e+00 -1.15911949e+00 -6.72359318e-02
-7.28103042e-01 -1.07930624e+00 -6.32112324e-01 -8.15411150e-01
1.03402483e+00 5.31723261e-01 7.85070837e-01 3.22999984e-01
-1.39054889e-02 8.66946042e-01 1.77481279e-01 -5.69545776e-02
-9.86725613e-02 4.28562075e-01 2.14331493e-01 2.05565393e-01
-7.61433840e-01 -9.38133955e-01 -9.59620774e-01 5.75725079e-01
-5.67562699e-01 -3.10514659e-01 2.98168212e-01 5.58531344e-01
4.39488292e-01 2.44678736e-01 6.42507195e-01 -3.83110106e-01
1.34468806e+00 -6.98341846e-01 -1.21107697e+00 4.09695923e-01
-6.93124115e-01 -4.55490174e-03 1.05120194e+00 -5.14499366e-01
-6.93387568e-01 -1.74292505e-01 5.63416302e-01 -3.03013716e-02
9.36535060e-01 4.73135024e-01 2.33701587e-01 -9.25837576e-01
3.49922776e-01 2.20697820e-01 1.96824856e-02 1.49327770e-01
4.28266019e-01 3.67476285e-01 1.36548683e-01 -7.06235886e-01
1.04897380e+00 7.01349735e-01 3.90342206e-01 -8.22858453e-01
-5.93372881e-01 -2.43441481e-02 -6.93842918e-02 -5.18147469e-01
4.97091673e-02 -7.27111280e-01 -1.46954131e+00 1.73405707e-01
-8.10099542e-01 -3.96677405e-01 -6.32033050e-02 4.24519300e-01
-2.38894641e-01 2.01275200e-01 -7.39085913e-01 -1.14149117e+00
-8.69385779e-01 -7.06191540e-01 4.25591648e-01 3.54943067e-01
4.03398782e-01 -1.53402507e+00 -9.16136652e-02 -3.33440661e-01
8.25504303e-01 5.33479869e-01 2.15665847e-01 -6.63178414e-02
-6.29768729e-01 -3.83068532e-01 -4.85025376e-01 -2.09974989e-01
1.22140020e-01 3.76037881e-02 1.94184870e-01 -8.94432306e-01
4.55405675e-02 -1.91996172e-01 2.20268697e-01 5.32524824e-01
1.01829553e+00 -9.04256523e-01 -6.10697746e-01 7.76407778e-01
1.71155214e+00 -2.34078303e-01 -2.77835697e-01 -1.48781344e-01
5.27968109e-01 1.03144467e-01 2.47799158e-01 1.02260721e+00
5.66265643e-01 5.03542840e-01 4.78120297e-01 1.08740918e-01
2.40133911e-01 -5.84640242e-02 3.94689053e-01 9.59759712e-01
1.51094452e-01 -4.16121215e-01 -8.64643335e-01 5.12218237e-01
-1.87054694e+00 -5.85265160e-01 8.46005827e-02 2.34985924e+00
9.17871356e-01 2.17112303e-01 3.61815959e-01 -1.81750610e-01
9.33827996e-01 7.56016672e-02 -8.19603205e-01 -1.70635596e-01
-3.12210526e-02 -3.26365709e-01 1.24693716e+00 8.17710102e-01
-6.13045633e-01 3.67888331e-01 6.56696796e+00 8.87169480e-01
-1.33358788e+00 2.02709854e-01 7.87019432e-01 -4.71775711e-01
-2.15756983e-01 -1.98510334e-01 -6.60973728e-01 6.39925480e-01
9.29109156e-01 -9.56580520e-01 7.87394702e-01 6.73968971e-01
6.70364141e-01 1.95493177e-01 -4.68686819e-01 7.40574062e-01
-5.12290239e-01 -1.52649522e+00 -3.69936705e-01 2.61401355e-01
9.55589533e-01 2.62158424e-01 5.98367862e-02 -3.59798312e-01
5.72660744e-01 -6.20597422e-01 3.87052149e-01 1.68595105e-01
4.86385971e-01 -9.18985009e-01 3.16679507e-01 1.86347097e-01
-1.51132762e+00 -3.26636106e-01 -2.24065870e-01 -1.31078243e-01
4.30953979e-01 9.63338375e-01 -4.53480810e-01 3.83841604e-01
4.28946495e-01 3.54401261e-01 2.46343259e-02 1.17783260e+00
5.35979718e-02 4.50901538e-01 -9.05992389e-01 -6.08427107e-01
4.36847985e-01 -6.49619222e-01 1.01804149e+00 1.02860868e+00
1.83312014e-01 -8.05394799e-02 7.48741746e-01 7.23818183e-01
-3.80686462e-01 2.73377001e-01 -1.72947496e-01 1.44728437e-01
9.98579562e-01 1.41968036e+00 -7.77813137e-01 1.70016158e-02
-1.51656538e-01 7.84933150e-01 5.43332696e-01 6.48262799e-01
-9.87229586e-01 -7.51432538e-01 8.97469878e-01 7.55408481e-02
2.48388812e-01 -8.77966523e-01 -5.42346895e-01 -8.66364896e-01
5.39616525e-01 1.15729291e-02 2.75828481e-01 1.54351607e-01
-1.11638927e+00 3.75581771e-01 -3.27266991e-01 -8.53017926e-01
2.18735754e-01 -1.26165941e-01 -9.69405591e-01 4.43521202e-01
-1.44644892e+00 -7.32344687e-01 -1.71256274e-01 6.81308270e-01
-8.60705152e-02 1.60979182e-01 8.72679800e-02 2.98716187e-01
-8.62825215e-01 9.51740801e-01 6.80622339e-01 2.40689050e-02
-3.97056863e-02 -1.02257514e+00 -1.31887019e-01 8.58157873e-01
-2.70015359e-01 3.89872879e-01 7.87727654e-01 -3.90880823e-01
-1.98642206e+00 -9.61728394e-01 3.50399405e-01 1.39664724e-01
1.23538017e+00 -5.04000783e-01 -2.64682531e-01 2.83450156e-01
-2.71899849e-02 3.67361158e-01 4.18918639e-01 -6.69996515e-02
1.50939792e-01 -5.40378511e-01 -1.48734498e+00 8.72455657e-01
1.13691974e+00 -2.72821069e-01 7.69413173e-01 7.89580226e-01
6.34445965e-01 -4.89654541e-01 -1.06474888e+00 7.27718621e-02
1.86439812e-01 -3.04921865e-01 8.22612762e-01 -3.42671007e-01
-2.94785917e-01 -1.63755074e-01 3.38339657e-02 -1.39462042e+00
2.14317385e-02 -1.61826718e+00 -3.62092465e-01 9.29496944e-01
6.70653105e-01 -1.10969687e+00 1.01088428e+00 7.28731394e-01
4.99349684e-01 -1.13182998e+00 -1.15736568e+00 -1.09951735e+00
1.86094902e-02 -4.31469604e-02 1.84110552e-01 4.60369945e-01
2.74157077e-01 3.39770496e-01 -4.44557279e-01 5.93007147e-01
9.59865212e-01 1.48982301e-01 6.52624726e-01 -7.39517450e-01
-1.12333700e-01 -4.57077891e-01 -1.75913587e-01 -1.28478909e+00
3.03337350e-02 -6.24753058e-01 -2.09222168e-01 -1.16108835e+00
-2.10447572e-02 -9.12120283e-01 -2.47541845e-01 6.49567991e-02
-1.30181357e-01 1.22126997e-01 2.69260794e-01 1.59144759e-01
-1.02540934e+00 8.45559895e-01 1.16291952e+00 2.61357486e-01
-6.14704967e-01 4.14219350e-01 -5.86258829e-01 3.65604013e-01
1.03912711e+00 -4.18320090e-01 -6.51105583e-01 -6.15311980e-01
3.01518828e-01 4.35586303e-01 2.34687135e-01 -8.17560256e-01
5.81032634e-01 -3.36796522e-01 -5.38617074e-02 -1.22261085e-01
2.53839880e-01 -1.02815068e+00 -1.51562765e-01 8.06674898e-01
-3.87349904e-01 3.30143213e-01 -2.01966554e-01 1.18172526e+00
2.83589214e-01 4.48113084e-01 8.49629164e-01 4.61020738e-01
1.42490074e-01 7.54127562e-01 -2.85010546e-01 3.04719895e-01
1.25943422e+00 2.61294469e-02 -6.57790005e-02 -9.74274099e-01
-6.37559891e-01 9.19375360e-01 1.69059083e-01 -1.13724232e-01
3.99391502e-01 -1.21423924e+00 -7.31621683e-01 -2.94319898e-01
-5.51114500e-01 -4.83468115e-01 -1.23785526e-01 1.32380033e+00
-4.58483189e-01 2.79935569e-01 2.86229223e-01 -4.25168753e-01
-9.94077682e-01 2.10296109e-01 5.63537598e-01 -4.76059645e-01
-2.74053037e-01 7.17635572e-01 -3.74782592e-01 -3.33467662e-01
4.07671571e-01 -1.54115632e-01 6.51560187e-01 -3.28278601e-01
2.88898587e-01 8.24890137e-01 -2.42415935e-01 -3.69815342e-02
-4.19723183e-01 4.15552884e-01 1.46618336e-01 -9.74525213e-02
1.24219131e+00 -1.00535309e+00 -1.57381848e-01 -5.34691475e-02
1.39011383e+00 2.08834186e-01 -1.45187938e+00 -3.55180174e-01
-1.12236634e-01 -4.94316965e-01 2.89518237e-01 -3.77888560e-01
-1.64491427e+00 1.07805505e-01 3.81504178e-01 5.04697263e-01
1.05096841e+00 -3.17042351e-01 7.90824294e-01 3.78587842e-01
7.74627745e-01 -1.16172361e+00 -7.93563500e-02 3.51885051e-01
5.57836652e-01 -9.65653002e-01 1.77572310e-01 -6.46743953e-01
-3.01892579e-01 1.03647447e+00 4.40384954e-01 -4.81181383e-01
1.05336475e+00 3.87830496e-01 -2.88138211e-01 -3.43659334e-02
-7.82413185e-01 3.08241189e-01 -4.66056541e-02 3.20341855e-01
2.42678642e-01 3.20754111e-01 -7.89083123e-01 2.56724536e-01
1.93700921e-02 -3.31028610e-01 5.39026320e-01 8.18106055e-01
-4.97507066e-01 -8.39944899e-01 -1.02793753e-01 4.09562975e-01
-5.16104996e-01 8.71786401e-02 -2.95717895e-01 6.27153158e-01
-8.02791715e-01 1.00817955e+00 -2.59841919e-01 -3.83896530e-02
-5.35893962e-02 -8.28379154e-01 2.18718618e-01 -5.41042723e-02
-5.74341178e-01 -3.65971357e-01 1.23520354e-02 -6.42659843e-01
-2.97729690e-02 -9.85929072e-02 -1.13726425e+00 -9.21635866e-01
-5.33131182e-01 5.30110180e-01 7.48199642e-01 7.01973677e-01
8.47795904e-01 1.48481622e-01 1.37212193e+00 -6.37866139e-01
-1.01755655e+00 -3.31321359e-01 -6.29043937e-01 -2.50759423e-01
6.38066947e-01 -2.48341650e-01 -5.94264984e-01 -7.38336563e-01] | [6.243598937988281, 4.928659915924072] |
ff342912-8db0-47e4-8083-97c5a3f39cb6 | all-about-knowledge-graphs-for-actions | 2008.12432 | null | https://arxiv.org/abs/2008.12432v1 | https://arxiv.org/pdf/2008.12432v1.pdf | All About Knowledge Graphs for Actions | Current action recognition systems require large amounts of training data for recognizing an action. Recent works have explored the paradigm of zero-shot and few-shot learning to learn classifiers for unseen categories or categories with few labels. Following similar paradigms in object recognition, these approaches utilize external sources of knowledge (eg. knowledge graphs from language domains). However, unlike objects, it is unclear what is the best knowledge representation for actions. In this paper, we intend to gain a better understanding of knowledge graphs (KGs) that can be utilized for zero-shot and few-shot action recognition. In particular, we study three different construction mechanisms for KGs: action embeddings, action-object embeddings, visual embeddings. We present extensive analysis of the impact of different KGs in different experimental setups. Finally, to enable a systematic study of zero-shot and few-shot approaches, we propose an improved evaluation paradigm based on UCF101, HMDB51, and Charades datasets for knowledge transfer from models trained on Kinetics. | ['Abhinav Shrivastava', 'Nirat Saini', 'Larry S. Davis', 'Pallabi Ghosh'] | 2020-08-28 | null | null | null | null | ['zero-shot-action-recognition', 'few-shot-action-recognition'] | ['computer-vision', 'computer-vision'] | [ 1.62173554e-01 7.17053469e-03 -5.19192576e-01 -3.85876417e-01
-3.56287032e-01 -2.56166577e-01 7.97724903e-01 1.47733063e-01
-4.43360865e-01 6.85701251e-01 3.53684425e-01 6.60714656e-02
-1.84513792e-01 -9.44213748e-01 -5.77256799e-01 -5.75302064e-01
-1.22323044e-01 1.95427492e-01 6.59377038e-01 1.85762420e-02
3.57728899e-01 4.81542915e-01 -1.91816366e+00 4.76419389e-01
3.78660232e-01 9.31136191e-01 -1.43983185e-01 5.01046956e-01
-3.17365170e-01 1.17554736e+00 -5.52349925e-01 -4.01433051e-01
1.36583656e-01 -4.98323023e-01 -1.04443061e+00 1.87591404e-01
4.35496271e-01 -3.85706753e-01 -3.46158594e-01 8.19989860e-01
4.22366142e-01 5.27855456e-01 1.07625246e+00 -1.59424484e+00
-1.13723779e+00 3.84435773e-01 -1.39908388e-01 4.06962782e-01
4.27481711e-01 3.85492742e-01 7.54101157e-01 -6.41609132e-01
9.30857539e-01 1.34127545e+00 3.09148520e-01 8.05233419e-01
-9.06112373e-01 -3.80459845e-01 1.68845549e-01 9.54797983e-01
-1.30778933e+00 -3.75916928e-01 5.88946760e-01 -6.31758630e-01
1.08150005e+00 -1.41984627e-01 4.95712906e-01 1.52315271e+00
1.75037347e-02 7.85386324e-01 9.89858449e-01 -7.33285129e-01
6.36960149e-01 3.16831410e-01 4.77020890e-01 4.59879011e-01
3.15338701e-01 2.91213952e-02 -8.67669165e-01 -6.47066981e-02
6.10913336e-01 1.20824710e-01 -2.29824319e-01 -8.90083551e-01
-9.62244213e-01 9.51078713e-01 4.08855617e-01 5.76343119e-01
-2.28386313e-01 2.34096125e-01 5.47681749e-01 2.65765399e-01
3.41481775e-01 2.31161952e-01 -2.56992012e-01 -4.46285516e-01
-2.29439318e-01 -2.10763708e-01 7.93632925e-01 1.07857120e+00
8.98080051e-01 6.07552752e-02 -4.72479284e-01 7.44706154e-01
9.86617729e-02 2.47335434e-01 8.05267453e-01 -5.24489224e-01
2.53283292e-01 6.98776245e-01 1.37849972e-01 -7.48672783e-01
-1.33715466e-01 1.51853189e-01 -3.47899020e-01 1.50686935e-01
3.76108795e-01 1.75799102e-01 -1.21168447e+00 1.52094805e+00
2.72006869e-01 4.36421841e-01 2.79811919e-01 8.05161834e-01
7.57558048e-01 1.76870659e-01 5.41774213e-01 1.53490916e-01
1.29271924e+00 -1.13956618e+00 -6.78799391e-01 -1.59562245e-01
1.15089035e+00 -1.48605630e-01 1.15527391e+00 5.32204024e-02
-4.73513514e-01 -5.96452177e-01 -8.50359619e-01 -4.90799136e-02
-1.16569412e+00 7.47647211e-02 6.00887060e-01 7.45216012e-01
-7.96601474e-01 8.52782607e-01 -8.86895776e-01 -9.61632013e-01
7.15719402e-01 -3.26525234e-02 -5.78435481e-01 -3.40390533e-01
-1.16368413e+00 1.24682295e+00 8.23801577e-01 -5.22882760e-01
-1.17223954e+00 -4.98167276e-01 -1.05441093e+00 -3.95023562e-02
4.93610352e-01 -3.80315989e-01 9.95415092e-01 -1.00023305e+00
-1.34911728e+00 9.43035245e-01 2.08713010e-01 -6.56835735e-01
6.67031631e-02 -3.47432375e-01 -5.73938847e-01 3.47624987e-01
-4.87068035e-02 5.35574377e-01 8.44901085e-01 -1.03379464e+00
-3.74769777e-01 -4.56495106e-01 3.43642294e-01 -4.56543230e-02
-6.63071692e-01 3.75176445e-02 2.44191084e-02 -4.06389624e-01
-3.78396660e-01 -6.99270964e-01 5.00786901e-02 1.22708514e-01
-3.96155491e-02 -4.40163910e-01 8.14792693e-01 -2.55757987e-01
9.85604882e-01 -2.36490750e+00 9.58486944e-02 -2.50699133e-01
-1.47303402e-01 5.94339252e-01 -3.88545394e-01 6.70255065e-01
-6.20997399e-02 2.36620996e-02 7.88051262e-02 2.72596143e-02
1.79443881e-01 6.17814183e-01 -2.08237737e-01 2.23406285e-01
3.48499626e-01 1.10715830e+00 -1.11865103e+00 -5.06379128e-01
4.88315105e-01 4.57979232e-01 -2.80191958e-01 1.75422192e-01
-9.99756306e-02 2.14122787e-01 -6.28986001e-01 7.70296931e-01
1.48050517e-01 -3.58117491e-01 1.36942893e-01 -2.19065025e-01
2.24276215e-01 1.57337263e-02 -1.01369357e+00 1.65454936e+00
-3.00779074e-01 3.32965553e-01 -6.52129829e-01 -1.23429084e+00
8.33020568e-01 3.16741735e-01 2.36102208e-01 -6.32236004e-01
2.71819413e-01 -1.25345007e-01 -2.16200501e-01 -6.73357606e-01
2.98655361e-01 -4.09088939e-01 1.28822336e-02 3.87132972e-01
7.23021865e-01 4.04265255e-01 5.11232555e-01 2.13913411e-01
1.33687615e+00 2.55416423e-01 6.32835329e-01 1.21375501e-01
2.87538916e-01 9.38860551e-02 2.84308285e-01 7.34784186e-01
-5.29051721e-01 3.79558444e-01 3.37493837e-01 -4.98516887e-01
-5.59757173e-01 -1.01814997e+00 -7.82798417e-03 1.39937520e+00
2.08569944e-01 -6.04801536e-01 -5.57888150e-01 -9.76871669e-01
1.76396355e-01 1.11413276e+00 -8.97095323e-01 -6.98128283e-01
-9.61562619e-02 -3.83860767e-01 5.50390601e-01 8.95829499e-01
2.65317231e-01 -1.42553711e+00 -9.01793838e-01 2.43120790e-02
2.15666071e-01 -1.05856192e+00 9.97214839e-02 2.31518492e-01
-8.57529581e-01 -1.41594124e+00 -6.50159538e-01 -4.62219745e-01
6.12299204e-01 3.48256260e-01 1.08026445e+00 -1.79812491e-01
-5.25159061e-01 1.23729026e+00 -1.07929409e+00 -5.33597827e-01
-2.56998390e-01 -2.20071927e-01 1.74081087e-01 2.67343760e-01
1.07515407e+00 -6.20962262e-01 -3.37803453e-01 3.64283651e-01
-9.31761384e-01 -4.16252285e-01 5.88912725e-01 4.77837116e-01
4.05728400e-01 -2.45295256e-01 5.29773533e-01 -7.62261629e-01
5.68227172e-01 -6.08004987e-01 -1.19367518e-01 6.65366769e-01
-3.45267504e-01 1.70131713e-01 4.56141353e-01 -6.47721529e-01
-1.09250867e+00 -7.15439916e-02 3.21997255e-01 -8.34461033e-01
-5.91041386e-01 3.04947376e-01 -8.34474117e-02 -1.74949497e-01
8.97145689e-01 2.76041240e-01 -1.76954255e-01 -6.57092988e-01
8.86538923e-01 6.85281098e-01 7.71459565e-02 -4.63711798e-01
4.51073945e-01 5.64740896e-01 -2.75900990e-01 -8.59930634e-01
-9.43391621e-01 -6.92913592e-01 -1.09497571e+00 -2.87848949e-01
1.18462241e+00 -7.11791098e-01 -3.04963380e-01 4.06888485e-01
-8.35648179e-01 -3.22724432e-01 -7.74661243e-01 6.54106498e-01
-7.27454841e-01 2.92422354e-01 -3.54308128e-01 -7.18963265e-01
6.20593280e-02 -6.13669097e-01 9.49621081e-01 2.95104682e-01
-2.18653798e-01 -9.82826591e-01 2.82472342e-01 3.27109694e-01
4.28274959e-01 1.14817470e-01 9.08458352e-01 -1.06467426e+00
-4.61596340e-01 -1.91039339e-01 -1.06884114e-01 5.91463387e-01
4.03879046e-01 -1.91558763e-01 -1.04015124e+00 -1.24056272e-01
-2.59356409e-01 -9.31147099e-01 9.07371640e-01 3.51496264e-02
1.03148460e+00 -6.69742450e-02 -4.77320105e-01 9.19653699e-02
1.42497325e+00 3.45226616e-01 1.04183114e+00 2.52950549e-01
6.37405515e-01 4.71618026e-01 7.98878074e-01 4.96840656e-01
1.84858158e-01 5.90868652e-01 2.01250538e-01 4.06536728e-01
-3.70400727e-01 -3.06651831e-01 3.94199193e-01 6.16186261e-01
-2.36329168e-01 -2.92884678e-01 -8.76971662e-01 7.83522367e-01
-1.87725651e+00 -1.07333350e+00 3.34347069e-01 2.13724136e+00
6.21727586e-01 -3.16146109e-03 -3.17749642e-02 -2.67694797e-02
6.19645596e-01 2.95718551e-01 -5.74645758e-01 -2.99858391e-01
1.53504804e-01 2.08980188e-01 2.84720659e-01 -5.90554103e-02
-1.20712900e+00 1.05197978e+00 6.41232872e+00 9.51694667e-01
-8.01692486e-01 3.31151158e-01 2.58556753e-02 -4.72273864e-02
3.31896096e-01 9.73107889e-02 -8.20456445e-01 2.98023939e-01
1.11543846e+00 -2.53386974e-01 7.73828924e-02 1.27119517e+00
-2.89850175e-01 -2.86800206e-01 -1.40709102e+00 9.73295927e-01
4.74516928e-01 -1.12520075e+00 3.05595189e-01 -2.16392592e-01
5.90997338e-01 3.26076150e-02 -4.38074380e-01 7.11858094e-01
4.91020024e-01 -8.36002171e-01 2.32442573e-01 7.58553088e-01
5.81469059e-01 -3.13429534e-01 5.09831250e-01 1.84465274e-01
-1.04182804e+00 -1.74282148e-01 -5.91955781e-01 -1.50242135e-01
1.37873292e-01 1.09063676e-02 -6.06138527e-01 7.54531264e-01
6.28462791e-01 1.06811166e+00 -8.30278277e-01 8.77188623e-01
-4.22129542e-01 5.58209300e-01 1.47151858e-01 -7.63460100e-02
1.69738084e-01 4.97371741e-02 5.25588989e-02 1.03467333e+00
8.69530588e-02 2.85554826e-01 3.05852145e-01 5.79821706e-01
1.27676100e-01 1.49374038e-01 -1.05805969e+00 -4.88392949e-01
1.87659368e-01 1.02994800e+00 -7.29022443e-01 -6.40704572e-01
-7.91474283e-01 9.43010509e-01 6.22363508e-01 3.23247790e-01
-7.00813532e-01 -3.91615182e-01 5.93517363e-01 1.49355099e-01
5.57267368e-01 -1.46866366e-01 4.03490186e-01 -1.20303571e+00
-2.94157237e-01 -6.06126308e-01 6.69763684e-01 -1.00046623e+00
-1.53881216e+00 1.58358306e-01 3.75794113e-01 -1.41078198e+00
-1.80680454e-01 -8.95995080e-01 -5.63950181e-01 3.07232916e-01
-1.35201430e+00 -1.03442121e+00 -3.63261819e-01 6.85548127e-01
4.73777890e-01 -3.15669000e-01 1.07661605e+00 2.11702704e-01
-3.04153472e-01 2.27718726e-01 1.09757818e-01 3.79775375e-01
7.88929224e-01 -9.12191987e-01 8.11387897e-02 4.52661604e-01
5.61524153e-01 4.00712430e-01 2.95580029e-01 -6.08530700e-01
-1.34090030e+00 -1.16694260e+00 4.94227618e-01 -7.54136145e-01
8.34119201e-01 -7.84889683e-02 -1.13747573e+00 9.31782544e-01
2.39543840e-02 3.26690078e-01 9.31210399e-01 2.44049370e-01
-6.29255474e-01 6.24138378e-02 -1.03714025e+00 4.12661076e-01
1.34323001e+00 -5.57687938e-01 -1.00723064e+00 4.37714368e-01
4.89094943e-01 3.13946486e-01 -9.58378494e-01 2.08491594e-01
3.54366183e-01 -8.78389060e-01 9.28577006e-01 -1.31405616e+00
3.38997096e-01 -2.01570272e-01 -2.31328905e-01 -1.37473595e+00
-3.68564785e-01 2.04425886e-01 -3.23038578e-01 9.90433753e-01
-4.02811356e-02 -5.70407033e-01 5.51145256e-01 5.26184142e-01
-9.23493430e-02 -4.61362600e-01 -9.28920984e-01 -1.38222551e+00
-6.70579150e-02 -2.15458825e-01 3.32349688e-01 1.10323191e+00
1.75866291e-01 3.72256905e-01 -4.45332527e-01 -2.25706264e-01
4.96882498e-01 2.24716030e-02 8.69570196e-01 -1.06701124e+00
-1.85912117e-01 2.39264835e-02 -1.22447979e+00 -5.01973152e-01
2.44415447e-01 -1.02450538e+00 -1.89361125e-01 -1.65159225e+00
3.53300959e-01 1.26151755e-01 -7.88852632e-01 9.41383719e-01
7.24022537e-02 1.46911860e-01 2.80516773e-01 -4.02609631e-03
-1.11589313e+00 8.77943158e-01 1.01129377e+00 -1.52911752e-01
9.02478471e-02 -5.01907587e-01 -2.89005697e-01 8.17485392e-01
8.51804614e-01 -5.05681098e-01 -7.31674135e-01 -2.38660634e-01
-2.29567662e-01 -3.99121612e-01 6.40486360e-01 -1.14050603e+00
1.21877097e-01 -4.23145711e-01 2.02790752e-01 -3.28121483e-01
4.45101589e-01 -6.43460751e-01 -1.31380567e-02 3.30160260e-01
-2.94055283e-01 -6.20728850e-01 2.01419666e-01 9.40833390e-01
-2.88136780e-01 -5.42240560e-01 7.48710871e-01 -3.86405349e-01
-1.43865824e+00 1.03915282e-01 -3.41877699e-01 2.20238715e-01
1.36886239e+00 -4.90987539e-01 -5.11938632e-01 -1.33921191e-01
-1.08449626e+00 1.54323047e-02 4.42580700e-01 8.48104537e-01
5.54226756e-01 -1.56387591e+00 -2.52523035e-01 -6.95154024e-03
8.00125837e-01 -7.60827661e-01 3.42467487e-01 7.04626501e-01
-9.86066833e-02 4.29398596e-01 -6.90244079e-01 -2.86835432e-01
-1.24039817e+00 1.03963172e+00 3.33747342e-02 -2.34407131e-02
-6.21270418e-01 6.24762237e-01 1.73996687e-01 -2.51603335e-01
3.01656753e-01 -7.69161135e-02 -3.52064043e-01 2.51344442e-01
7.72065163e-01 5.05862832e-01 -2.06666112e-01 -5.85966408e-01
-5.76409101e-01 4.55108494e-01 -1.09324150e-01 2.15841815e-01
1.22830975e+00 1.86054200e-01 2.62953669e-01 9.14367795e-01
1.05016851e+00 -7.05693960e-01 -1.08325970e+00 -2.88292944e-01
2.86408961e-01 -7.10258603e-01 -3.37244987e-01 -7.91918695e-01
-8.07854116e-01 1.08146405e+00 7.97729373e-01 9.46923196e-02
8.11979890e-01 2.81171083e-01 2.69869506e-01 5.77715158e-01
8.43112409e-01 -1.38780713e+00 5.02433419e-01 4.12764639e-01
8.89908791e-01 -1.25348580e+00 -8.75004679e-02 -3.38725626e-01
-8.79990935e-01 1.02087295e+00 7.29718268e-01 8.04749969e-03
7.11997330e-01 -2.89110363e-01 1.55559480e-02 -4.69028741e-01
-7.71723449e-01 -6.52860165e-01 2.37159491e-01 9.73106682e-01
1.88172907e-01 9.26403552e-02 -3.77912462e-01 5.74887753e-01
3.67595822e-01 4.61149752e-01 4.57906067e-01 1.49659061e+00
-6.32165134e-01 -1.14810658e+00 9.87644307e-03 5.21555364e-01
-3.07831969e-02 1.58696592e-01 -6.58605337e-01 8.27224731e-01
1.37244016e-01 8.83456349e-01 -5.10571450e-02 -3.20451975e-01
3.96993548e-01 6.63838506e-01 7.72951901e-01 -1.10362530e+00
-1.09485120e-01 -7.35557616e-01 2.02177957e-01 -6.88332200e-01
-8.34097981e-01 -4.93823647e-01 -1.01235259e+00 1.97447404e-01
-3.68428558e-01 -2.11149573e-01 3.38523090e-01 9.86578703e-01
5.43719769e-01 3.99216145e-01 1.03073135e-01 -5.01339018e-01
-6.75347626e-01 -1.18439734e+00 -9.99859095e-01 8.81639957e-01
-4.90031183e-01 -1.28757310e+00 -4.17612463e-01 3.73116285e-01] | [8.640262603759766, 1.0306315422058105] |
809f2032-80a9-48da-8cae-24e688638a15 | describe-me-if-you-can-characterized-instance | 2201.09594 | null | https://arxiv.org/abs/2201.09594v1 | https://arxiv.org/pdf/2201.09594v1.pdf | Describe me if you can! Characterized Instance-level Human Parsing | Several computer vision applications such as person search or online fashion rely on human description. The use of instance-level human parsing (HP) is therefore relevant since it localizes semantic attributes and body parts within a person. But how to characterize these attributes? To our knowledge, only some single-HP datasets describe attributes with some color, size and/or pattern characteristics. There is a lack of dataset for multi-HP in the wild with such characteristics. In this article, we propose the dataset CCIHP based on the multi-HP dataset CIHP, with 20 new labels covering these 3 kinds of characteristics. In addition, we propose HPTR, a new bottom-up multi-task method based on transformers as a fast and scalable baseline. It is the fastest method of multi-HP state of the art while having precision comparable to the most precise bottom-up method. We hope this will encourage research for fast and accurate methods of precise human descriptions. | ['Romaric Audigier', 'Angelique Loesch'] | 2022-01-24 | null | null | null | null | ['human-parsing', 'person-search'] | ['computer-vision', 'computer-vision'] | [-9.21226442e-02 1.04115754e-01 -3.06548804e-01 -6.72257602e-01
-9.07181501e-01 -5.72585285e-01 6.10312104e-01 4.22615319e-01
-4.88994390e-01 7.15709925e-01 1.97226733e-01 5.26497900e-01
-3.44266184e-02 -5.62416255e-01 -6.43267870e-01 -3.81253332e-01
2.12189272e-01 1.32852709e+00 4.22381818e-01 -1.63418688e-02
-1.08166084e-01 8.50709677e-02 -1.76949835e+00 5.51222146e-01
1.79371223e-01 1.09806418e+00 -1.14150316e-01 4.29167509e-01
5.35348654e-02 4.50714678e-01 -3.17552805e-01 -1.06478548e+00
1.29274756e-01 -1.33071423e-01 -1.05105555e+00 7.93071985e-02
1.05326271e+00 -1.17919378e-01 2.41962418e-01 8.28250885e-01
7.90483415e-01 1.49920331e-02 5.63081861e-01 -1.47967780e+00
-6.39352083e-01 6.14380896e-01 -6.85079336e-01 -1.01067223e-01
9.58940625e-01 -5.44609390e-02 1.19316745e+00 -7.60474741e-01
8.89011741e-01 1.64220369e+00 1.04854846e+00 9.73415613e-01
-1.14158309e+00 -4.78739053e-01 2.80568182e-01 1.59005761e-01
-1.46873546e+00 -1.87436536e-01 4.80279803e-01 -2.75666565e-01
9.21339989e-01 3.32396597e-01 7.15850353e-01 1.48504162e+00
-2.44165286e-01 1.07354879e+00 1.50173771e+00 -4.54155415e-01
-7.14772493e-02 2.82316864e-01 3.01848799e-01 1.10092902e+00
4.35118526e-01 -2.00764924e-01 -8.50508869e-01 -2.19766527e-01
5.99082530e-01 -3.19452256e-01 3.33485842e-01 -6.44749761e-01
-1.56119299e+00 5.98896086e-01 2.17898726e-01 1.45445973e-01
-2.51666218e-01 2.88332582e-01 6.11137748e-01 -1.60959527e-01
3.82112801e-01 5.47746599e-01 -8.38457346e-01 -7.97235891e-02
-8.44327867e-01 8.58292341e-01 9.59400535e-01 1.33521521e+00
5.81475377e-01 -5.91331482e-01 -3.09216589e-01 1.03603125e+00
8.47122967e-02 5.48908949e-01 1.74283504e-01 -1.20208824e+00
4.36594516e-01 6.14438772e-01 4.59145457e-02 -7.66269684e-01
-9.77744937e-01 -1.85914829e-01 -5.26388943e-01 -1.13605596e-01
8.38074684e-01 2.22406656e-01 -7.30804145e-01 1.79693985e+00
5.72122514e-01 -3.12209398e-01 -3.96018773e-01 1.11806476e+00
1.22200394e+00 1.45912647e-01 7.01461136e-01 2.48343050e-01
2.02457762e+00 -1.13818312e+00 -5.04160166e-01 -2.88213789e-01
3.94457996e-01 -3.50992858e-01 1.13479340e+00 4.11005288e-01
-1.07365596e+00 -6.76827848e-01 -7.71198571e-01 -4.89402801e-01
-8.54077518e-01 2.55385876e-01 1.26539731e+00 7.74421573e-01
-9.40775335e-01 5.60906708e-01 -5.83176613e-01 -1.03803372e+00
2.14358374e-01 4.34707165e-01 -7.31187582e-01 4.24086675e-02
-1.26162171e+00 9.95508194e-01 4.76624161e-01 -3.51225972e-01
-4.90074605e-01 -6.37285948e-01 -1.05900502e+00 -2.81091452e-01
6.75703585e-01 -1.38117683e+00 1.17531633e+00 -5.90902388e-01
-1.08491790e+00 1.66760623e+00 -6.50076717e-02 -3.76457095e-01
6.30109668e-01 -3.81314874e-01 -3.80397171e-01 1.00439981e-01
5.94749331e-01 1.15138400e+00 6.29763663e-01 -1.14120281e+00
-7.73107648e-01 -6.97615206e-01 5.74112721e-02 6.12646118e-02
3.52631025e-02 3.18878561e-01 -8.37368011e-01 -5.13164937e-01
1.64844207e-02 -1.11737299e+00 1.35959294e-02 1.34332981e-02
-7.34031141e-01 -6.22563064e-01 1.58605769e-01 -6.71487272e-01
9.09569561e-01 -1.80145061e+00 1.69223845e-01 -7.83212259e-02
2.64156997e-01 -1.09637603e-01 5.51658608e-02 1.99943408e-01
3.03667456e-01 2.29377404e-01 -1.29196957e-01 -7.29237139e-01
4.99489516e-01 1.70759588e-01 2.65300840e-01 4.62874502e-01
-3.42833400e-02 1.07435310e+00 -9.12252188e-01 -1.15724409e+00
1.34130523e-01 3.93820971e-01 -3.69057149e-01 5.67119569e-02
-7.26884007e-02 3.50307792e-01 -6.60635412e-01 1.27758503e+00
4.41768348e-01 -2.03250825e-01 1.63102783e-02 -3.97072524e-01
5.02338111e-02 -1.87336460e-01 -9.76692975e-01 1.90596318e+00
-2.20768392e-01 3.73150967e-02 -5.56074157e-02 -4.86821353e-01
7.03854442e-01 1.50774807e-01 5.05224705e-01 -5.49874187e-01
2.49455720e-02 2.61975497e-01 -5.15826762e-01 -3.43970686e-01
5.23088694e-01 -1.12566322e-01 -7.64832973e-01 -3.37107740e-02
2.70098329e-01 -1.83842052e-02 4.97893959e-01 8.45719203e-02
8.70725811e-01 7.32626975e-01 7.55297482e-01 -3.36556941e-01
5.83200634e-01 8.97664875e-02 7.54732251e-01 9.38021898e-01
-5.72153807e-01 9.24698651e-01 2.84159094e-01 -9.28385675e-01
-1.23590827e+00 -9.48291898e-01 -2.50788689e-01 1.67514873e+00
1.08429827e-01 -7.31900692e-01 -9.18657243e-01 -7.41992295e-01
2.90098011e-01 4.25157547e-01 -6.49246931e-01 4.67517585e-01
-5.30948937e-01 -7.27729142e-01 6.71790898e-01 7.53924847e-01
7.80827045e-01 -1.29305172e+00 -6.76621258e-01 2.20224962e-01
-4.18450952e-01 -1.85850394e+00 -2.99451113e-01 -6.44868016e-02
-4.51880187e-01 -1.15138590e+00 -8.79044294e-01 -6.38097584e-01
4.58800942e-01 -6.15966320e-01 1.62637198e+00 -3.24041620e-02
-3.50405365e-01 4.77919728e-01 -4.60464716e-01 -4.57759470e-01
-5.07671237e-02 3.57693464e-01 2.01775491e-01 -2.32977003e-01
6.33352339e-01 -1.52813792e-01 -4.07995194e-01 3.22596014e-01
-6.38776571e-02 3.01447809e-01 5.96239686e-01 5.23647547e-01
9.64890718e-01 -2.33692229e-01 1.38533071e-01 -1.40014100e+00
2.23928303e-01 -1.94220155e-01 -2.12384403e-01 4.56249923e-01
-4.15211529e-01 1.48859203e-01 3.99797350e-01 -2.99994260e-01
-1.04242218e+00 5.76113284e-01 -1.84735298e-01 7.64020607e-02
-6.55628920e-01 -1.46646976e-01 -4.01693791e-01 3.98203507e-02
5.53025305e-01 -6.30974993e-02 -5.32726467e-01 -7.47389019e-01
5.12533367e-01 2.90947646e-01 9.15818453e-01 -1.02451444e+00
5.00314713e-01 4.41439062e-01 3.28352928e-01 -5.01492381e-01
-1.36229491e+00 -8.41015279e-01 -1.10173535e+00 -6.58285096e-02
1.24651504e+00 -9.71562386e-01 -8.56010497e-01 5.32518327e-01
-1.09729850e+00 -1.04916841e-01 -5.35521619e-02 1.36372168e-02
-9.77435350e-01 4.02435511e-01 -9.02128875e-01 -7.49701321e-01
-4.31689799e-01 -6.93898439e-01 1.79262877e+00 9.54197571e-02
-6.11447692e-01 -6.97483182e-01 -1.65467083e-01 7.11137235e-01
2.07686156e-01 6.64428294e-01 6.03347480e-01 -7.53849804e-01
-1.75234601e-01 -2.77763098e-01 -2.82685488e-01 -2.11574614e-01
-3.71465743e-01 -3.22776824e-01 -9.36360717e-01 -3.24680284e-02
-6.72523618e-01 -6.29875124e-01 8.71594369e-01 3.55379492e-01
1.27299201e+00 -1.02570586e-01 -6.20145202e-01 6.94649816e-01
1.30189991e+00 -3.87225598e-01 2.47732177e-01 6.95430934e-01
9.35082018e-01 9.25556898e-01 8.35200012e-01 5.44008970e-01
1.01086688e+00 1.12392688e+00 -1.47365415e-02 -1.25144258e-01
-4.76019442e-01 -4.66653973e-01 -1.98608898e-02 2.72815794e-01
-5.16915619e-01 -1.60043370e-02 -9.02483046e-01 4.74635005e-01
-1.94807899e+00 -7.63178945e-01 -1.89688310e-01 2.08023763e+00
7.18951106e-01 6.55704364e-02 5.51046431e-01 -6.57210574e-02
8.85319710e-01 -3.31668071e-02 -3.05487871e-01 -2.59885728e-01
-2.49849632e-01 -9.40920133e-03 5.99206209e-01 8.62062275e-02
-1.81573820e+00 1.23426473e+00 6.35484362e+00 6.45073473e-01
-2.75503278e-01 3.87516797e-01 5.81981897e-01 1.08746603e-01
2.03071088e-01 -2.35915348e-01 -1.51453602e+00 3.12538415e-01
7.22372830e-01 3.29424500e-01 2.34477937e-01 1.13907039e+00
-5.73660076e-01 -1.00596316e-01 -1.46688318e+00 1.36425745e+00
3.86065304e-01 -6.48958266e-01 1.12809226e-01 -2.36767665e-01
2.89206475e-01 -5.63666940e-01 -2.27875978e-01 5.85572958e-01
1.26008853e-01 -7.73114681e-01 1.03881991e+00 5.31824589e-01
8.32038105e-01 -5.96361995e-01 7.41480708e-01 2.17772216e-01
-1.56000292e+00 1.81192420e-02 -2.24209875e-01 2.44051456e-01
3.35597456e-01 3.72832805e-01 -3.79333586e-01 3.30446005e-01
1.09908223e+00 5.07774889e-01 -1.04800820e+00 8.44648600e-01
-3.81083518e-01 6.84628263e-02 -1.90115854e-01 -6.89027607e-02
-1.06989965e-01 2.62979627e-01 3.94481033e-01 1.64290953e+00
1.95464984e-01 1.04671329e-01 5.52189410e-01 7.30708003e-01
6.95618019e-02 2.18045086e-01 -4.11978126e-01 1.23664670e-01
3.40961963e-01 1.46666658e+00 -8.63895714e-01 -5.40858746e-01
-4.62742805e-01 1.29913294e+00 4.05478925e-01 -1.35804623e-01
-9.13503468e-01 1.44576862e-01 3.79906625e-01 2.83710748e-01
3.13709863e-03 -7.13409632e-02 -2.22067088e-01 -1.03919733e+00
3.92873287e-02 -6.97934628e-01 1.01578772e+00 -6.94348276e-01
-1.69498837e+00 6.22500062e-01 3.53925556e-01 -6.89833105e-01
-3.87904108e-01 -7.97584414e-01 3.39959651e-01 5.15152037e-01
-1.30787194e+00 -2.10604739e+00 -4.51390624e-01 5.97823262e-01
7.02964842e-01 -1.66701563e-02 1.15675521e+00 2.29512364e-01
-3.49388033e-01 7.73873150e-01 -8.57464731e-01 1.10807665e-01
1.00533438e+00 -1.71123683e+00 6.31150842e-01 4.29780662e-01
1.28029570e-01 4.82789069e-01 9.25556958e-01 -7.47879267e-01
-1.12722647e+00 -8.81492078e-01 1.30177903e+00 -1.16415811e+00
2.68112540e-01 -6.92215264e-01 -6.46191597e-01 8.24784756e-01
-3.48542184e-02 2.09029004e-01 3.95816654e-01 5.58598638e-01
-6.38055742e-01 -1.14381453e-02 -1.34252012e+00 1.32327139e-01
1.43657470e+00 -3.30172807e-01 -6.47120655e-01 5.32593787e-01
6.24572098e-01 -6.60750389e-01 -9.86584425e-01 5.28922021e-01
7.37928569e-01 -8.69951189e-01 1.41808248e+00 -3.34030390e-01
2.54686266e-01 -1.53291389e-01 -2.14215294e-01 -1.01936913e+00
-5.70539355e-01 -9.84084830e-02 -7.19377324e-02 1.57552910e+00
5.06415725e-01 -3.88408571e-01 8.15977097e-01 1.02805841e+00
1.77435856e-02 -5.01701951e-01 -7.09238648e-01 -1.03342938e+00
-2.24884748e-01 -3.19002688e-01 6.13049865e-01 7.81471848e-01
-1.75252289e-01 1.93175077e-01 -6.46916449e-01 5.70611618e-02
1.10832036e+00 2.54872769e-01 8.00357819e-01 -1.42538440e+00
-3.28427613e-01 -2.11762950e-01 -5.34338593e-01 -4.17666972e-01
3.89802456e-01 -7.98165917e-01 -7.31557906e-02 -1.58806229e+00
8.80273223e-01 -4.11381692e-01 -6.54702038e-02 8.50513458e-01
-2.40598723e-01 4.48190629e-01 3.02501529e-01 7.64097571e-02
-1.26283932e+00 1.48555383e-01 1.02473700e+00 -8.71101916e-02
2.05094129e-01 -1.31283835e-01 -5.96790671e-01 9.20910597e-01
6.50890529e-01 -3.67071211e-01 1.36795953e-01 -2.33071446e-01
2.89259136e-01 -2.08694145e-01 5.67027569e-01 -1.30434942e+00
2.52943367e-01 5.76622188e-02 7.26895273e-01 -6.03543222e-01
7.44948626e-01 -6.64531589e-01 1.54194593e-01 1.96518615e-01
-2.35865101e-01 3.17749470e-01 -1.38667047e-01 3.19641680e-01
-1.40841812e-01 1.22018866e-01 6.27759874e-01 -8.27595115e-01
-1.48151255e+00 4.00788784e-01 1.88630894e-01 2.02797532e-01
8.77761602e-01 -1.18996181e-01 -3.41712832e-01 -5.72398603e-02
-9.40161347e-01 3.28638554e-01 4.07931954e-01 5.53649426e-01
3.33795100e-01 -1.41671109e+00 -7.73688734e-01 -2.32569754e-01
5.14676690e-01 -3.57437283e-01 4.69070934e-02 5.75991035e-01
-3.60329270e-01 5.57021320e-01 -5.26263058e-01 -4.23406482e-01
-1.48739970e+00 8.30191195e-01 7.04860315e-02 -3.31928223e-01
-7.59314120e-01 8.50867391e-01 1.06263354e-01 -7.13672757e-01
2.37979949e-01 9.98175815e-02 -3.70952874e-01 1.52167574e-01
5.53141356e-01 4.57352638e-01 -1.46770105e-01 -1.02593517e+00
-8.96239579e-01 9.22504485e-01 2.50609428e-01 -1.23573117e-01
1.04967248e+00 -2.33666450e-01 -2.49266773e-01 4.63463426e-01
8.87431562e-01 -1.94613248e-01 -8.68460238e-01 -9.95671600e-02
4.17013973e-01 -3.71689677e-01 -5.03097355e-01 -1.19776058e+00
-8.00932169e-01 5.79752803e-01 4.82297987e-01 -1.72664598e-01
9.97836232e-01 6.40915573e-01 9.83388007e-01 2.08972186e-01
1.12571847e+00 -1.38440347e+00 -1.82630643e-01 2.56353617e-01
9.21686769e-01 -1.41789842e+00 3.17419052e-01 -9.11591828e-01
-1.02984464e+00 8.32723796e-01 9.68668103e-01 1.53513387e-01
1.54410124e-01 1.73531592e-01 -1.93956837e-01 -4.51986790e-01
-5.10463655e-01 -7.57507741e-01 5.45373023e-01 7.76082277e-01
5.61196923e-01 5.21915853e-01 -4.46637064e-01 1.02662826e+00
-4.90547925e-01 -1.26761556e-01 -6.45597801e-02 6.02584124e-01
-2.92505920e-01 -1.32159519e+00 -3.60506564e-01 2.92214066e-01
-6.84813976e-01 3.94367240e-03 -7.11490631e-01 1.21625245e+00
6.07025087e-01 6.95985496e-01 -3.28220367e-01 -2.83032626e-01
6.52501106e-01 2.46441454e-01 8.35009694e-01 -6.84508860e-01
-6.77425861e-01 -3.05597067e-01 6.67020798e-01 -1.01112163e+00
-6.56862378e-01 -9.99681830e-01 -1.28908873e+00 -1.74441695e-01
1.98554978e-01 -3.21834832e-01 5.90785742e-01 8.48241925e-01
-2.92501296e-03 4.93129268e-02 -1.62523404e-01 -8.53147566e-01
-2.15931803e-01 -8.22552919e-01 -8.56433928e-01 1.05881619e+00
-1.56998932e-01 -1.03878021e+00 -6.85612783e-02 8.31524655e-02] | [8.180706977844238, -0.23299065232276917] |
422d10d5-68c1-432b-af37-9ff2e6ede317 | grayscale-data-construction-and-multi-level | 2004.02421 | null | https://arxiv.org/abs/2004.02421v4 | https://arxiv.org/pdf/2004.02421v4.pdf | The World is Not Binary: Learning to Rank with Grayscale Data for Dialogue Response Selection | Response selection plays a vital role in building retrieval-based conversation systems. Despite that response selection is naturally a learning-to-rank problem, most prior works take a point-wise view and train binary classifiers for this task: each response candidate is labeled either relevant (one) or irrelevant (zero). On the one hand, this formalization can be sub-optimal due to its ignorance of the diversity of response quality. On the other hand, annotating grayscale data for learning-to-rank can be prohibitively expensive and challenging. In this work, we show that grayscale data can be automatically constructed without human effort. Our method employs off-the-shelf response retrieval models and response generation models as automatic grayscale data generators. With the constructed grayscale data, we propose multi-level ranking objectives for training, which can (1) teach a matching model to capture more fine-grained context-response relevance difference and (2) reduce the train-test discrepancy in terms of distractor strength. Our method is simple, effective, and universal. Experiments on three benchmark datasets and four state-of-the-art matching models show that the proposed approach brings significant and consistent performance improvements. | ['Hai-Tao Zheng', 'Zibo Lin', 'Xiaojiang Liu', 'Shuming Shi', 'Deng Cai', 'Yan Wang'] | 2020-04-06 | null | https://aclanthology.org/2020.emnlp-main.741 | https://aclanthology.org/2020.emnlp-main.741.pdf | emnlp-2020-11 | ['conversational-response-selection'] | ['natural-language-processing'] | [ 4.79685485e-01 -2.46780410e-01 -3.31041992e-01 -7.74678111e-01
-1.45259953e+00 -4.55650866e-01 6.12127066e-01 2.49315873e-01
-2.11777046e-01 7.32000053e-01 1.83222651e-01 -3.30778450e-01
-2.11172178e-01 -8.85553837e-01 -1.42085955e-01 -5.85896254e-01
4.01536614e-01 7.03769088e-01 3.27271432e-01 -5.39308548e-01
6.49130583e-01 1.44790113e-01 -1.68377209e+00 7.02045858e-01
1.13040948e+00 1.30040205e+00 1.19015805e-01 7.21000612e-01
-1.28303647e-01 8.93145025e-01 -8.19447219e-01 -2.86921144e-01
1.10085919e-01 -6.87794089e-01 -9.33265865e-01 -3.04304659e-01
5.59138775e-01 -1.37514099e-01 -1.99096873e-01 8.61808717e-01
6.72859669e-01 2.19070986e-01 7.13427305e-01 -1.07327831e+00
-1.00331974e+00 3.31650853e-01 -4.68500316e-01 6.65974841e-02
6.20061040e-01 9.97848660e-02 1.23491883e+00 -9.23969150e-01
4.04699415e-01 1.35817945e+00 4.69167791e-02 8.18875790e-01
-1.26348126e+00 -5.59111774e-01 2.07501784e-01 3.00507873e-01
-1.10996938e+00 -2.33293772e-01 8.21030736e-01 -2.27952108e-01
4.69717830e-01 7.13511050e-01 2.07856819e-01 1.00690842e+00
1.59944132e-01 8.55083883e-01 1.74793684e+00 -5.85143149e-01
3.44012350e-01 3.48462135e-01 3.05930883e-01 4.48914170e-01
-1.90818399e-01 7.89643452e-02 -6.43171251e-01 -1.81455538e-01
3.85940611e-01 -5.88212686e-04 -2.29888633e-01 -4.96588871e-02
-8.50578845e-01 1.04527342e+00 4.77927297e-01 4.57793958e-02
-1.31318375e-01 -1.78548440e-01 7.20392317e-02 8.12993288e-01
6.03894770e-01 5.78649998e-01 -1.68790877e-01 -1.80132240e-02
-6.99642956e-01 3.21531206e-01 7.11755693e-01 7.33355820e-01
8.95988882e-01 -1.53834060e-01 -7.55962670e-01 1.32690859e+00
4.92854677e-02 4.43250060e-01 5.90736508e-01 -6.64559603e-01
5.02234280e-01 8.07745934e-01 2.03986838e-01 -1.05009818e+00
-2.54958689e-01 -3.05079073e-01 -7.09732950e-01 2.09228829e-01
3.66320997e-01 1.01949066e-01 -6.66829944e-01 1.70166576e+00
2.85793096e-01 -2.82560676e-01 -1.13838278e-01 1.30321324e+00
1.01093411e+00 4.47297335e-01 1.17917685e-02 -1.23997644e-01
1.46977985e+00 -1.06914067e+00 -5.17232180e-01 -3.74451965e-01
5.27987540e-01 -8.86721015e-01 1.70392859e+00 2.42868170e-01
-1.00957382e+00 -6.46220982e-01 -1.03416932e+00 -9.02972892e-02
-2.52933413e-01 3.92891876e-02 6.13067627e-01 8.79430711e-01
-1.04963005e+00 3.41297686e-01 4.94212285e-02 -1.00123174e-01
-7.72769377e-02 3.36399853e-01 -9.82597694e-02 -1.50374189e-01
-1.60392296e+00 9.68966603e-01 -2.53438145e-01 -1.48650780e-01
-6.55095816e-01 -4.49059725e-01 -4.45902854e-01 4.44065370e-02
4.50989395e-01 -6.03556454e-01 1.43758035e+00 -1.03810823e+00
-1.71390092e+00 9.67308700e-01 -4.92734537e-02 8.94366112e-03
3.94076407e-01 4.92116734e-02 -4.13016319e-01 2.39970803e-01
-1.77811399e-01 6.64001167e-01 7.00813472e-01 -1.28128910e+00
-7.89815366e-01 -3.64838570e-01 3.14081132e-01 3.97120714e-01
-4.91316229e-01 2.55462021e-01 -3.35014313e-01 -7.31359065e-01
1.15579501e-01 -9.84160006e-01 -4.64866944e-02 -1.51653379e-01
-2.44009253e-02 -6.52678490e-01 4.46468621e-01 -1.92536280e-01
1.46411002e+00 -1.75006807e+00 -2.57317662e-01 2.32230768e-01
7.29212314e-02 8.00041035e-02 -4.33105588e-01 3.68255854e-01
1.05804734e-01 1.79406181e-01 1.42649412e-01 -3.88444923e-02
1.95733950e-01 -1.52712330e-01 -4.59722161e-01 2.14501619e-01
2.97291070e-01 8.62568974e-01 -1.00793111e+00 -6.81732774e-01
4.73663658e-02 1.98858798e-01 -7.64375627e-01 5.69396496e-01
-1.34767815e-01 3.37782562e-01 -6.54622376e-01 6.47955835e-01
4.60016936e-01 -2.85895616e-01 2.53041953e-01 9.58922282e-02
1.27916440e-01 6.58052266e-01 -1.17541933e+00 1.21920013e+00
-8.71972024e-01 5.09311855e-01 -1.04236200e-01 -9.45895553e-01
1.18749273e+00 5.77061772e-02 2.07668781e-01 -1.29896784e+00
-2.78975279e-03 3.13695759e-01 3.25567313e-02 -2.25213155e-01
8.53517413e-01 -1.03137441e-01 -4.93799984e-01 8.29483092e-01
-4.31310862e-01 -3.90090406e-01 1.75583214e-01 1.46258950e-01
9.90635157e-01 -2.02493593e-01 1.03232920e-01 -1.74386263e-01
6.51297927e-01 -1.02407970e-01 3.63058776e-01 9.94808018e-01
1.38703557e-02 7.41903007e-01 5.50715506e-01 -3.70908946e-01
-5.31166732e-01 -7.90032923e-01 -1.34904953e-02 1.76332688e+00
2.51858443e-01 5.34365512e-03 -7.20916450e-01 -7.28932798e-01
-3.10981423e-01 4.46108460e-01 -6.54879868e-01 -2.52659559e-01
-6.85092986e-01 -5.60055375e-01 2.35038400e-01 3.70049268e-01
2.69251168e-01 -1.32661951e+00 -5.15313864e-01 -2.61561889e-02
-5.07418096e-01 -7.20243394e-01 -8.78017724e-01 1.44633129e-01
-5.82456231e-01 -8.05737913e-01 -6.65819049e-01 -9.33516204e-01
8.60487223e-01 7.35567153e-01 1.51703167e+00 5.77225208e-01
-1.21605173e-01 8.35103393e-02 -5.53531349e-01 -2.26153374e-01
-4.22677040e-01 1.40981581e-02 -1.79494649e-01 1.68346486e-03
5.21718442e-01 -1.26111388e-01 -1.00078058e+00 7.25883663e-01
-9.17372584e-01 -9.26517844e-02 6.23257935e-01 1.16660058e+00
6.98508680e-01 -4.02304024e-01 9.18499768e-01 -1.05004156e+00
1.26699936e+00 -2.32942477e-01 -4.08250749e-01 5.39350748e-01
-9.95705187e-01 2.17716262e-01 7.16520965e-01 -6.20525241e-01
-1.01025367e+00 -2.41457909e-01 6.19351352e-03 -1.04653239e-02
6.25852868e-02 4.03907150e-01 4.75092605e-02 -1.86017931e-01
8.46434474e-01 4.62354422e-02 -3.31065655e-01 -1.84054628e-01
3.73418272e-01 9.58274543e-01 3.15406889e-01 -9.41609263e-01
4.69355047e-01 1.59906968e-01 -2.54842013e-01 -3.16106737e-01
-8.12706947e-01 -4.79832977e-01 -2.37739190e-01 -3.40465128e-01
4.44809705e-01 -5.85305572e-01 -8.50594461e-01 9.45734158e-02
-8.20837677e-01 -3.20177972e-01 -4.40796390e-02 1.29160389e-01
-3.49049479e-01 1.28228649e-01 -6.12868369e-01 -8.08498442e-01
-6.15524769e-01 -1.16262329e+00 8.99628043e-01 3.39154124e-01
-2.81517029e-01 -7.34927416e-01 1.60181507e-01 5.15123963e-01
6.47236109e-01 -1.61172047e-01 1.09091806e+00 -7.37673521e-01
-4.80424613e-01 -4.63211924e-01 -3.95483166e-01 7.93475062e-02
-1.58153296e-01 -1.09305978e-01 -1.07904446e+00 -1.07027918e-01
4.93129976e-02 -8.03434134e-01 8.08082402e-01 4.51985560e-03
1.31772113e+00 -3.70183766e-01 9.55566689e-02 -1.23608513e-02
1.11173224e+00 1.94874048e-01 6.88296437e-01 2.60108918e-01
2.33123213e-01 1.05161083e+00 9.94725049e-01 4.18828070e-01
5.23806453e-01 1.08598483e+00 2.43963778e-01 -2.91296959e-01
-1.68146655e-01 -3.23076695e-01 1.63128749e-01 7.85576761e-01
3.04555744e-01 -3.31996113e-01 -6.47030592e-01 3.30979586e-01
-1.79169416e+00 -9.92609262e-01 -1.09785490e-01 2.43189096e+00
1.22899747e+00 1.39015257e-01 1.88154787e-01 5.66670559e-02
5.42207420e-01 1.79945398e-02 -2.44018957e-01 -6.27686143e-01
-1.29611298e-01 4.57153201e-01 3.68575798e-03 6.27665102e-01
-7.42199481e-01 8.56631100e-01 5.97695732e+00 8.90237689e-01
-1.26237166e+00 1.15100713e-02 9.63886023e-01 -1.42009839e-01
-6.34212077e-01 -2.05616131e-02 -5.13173819e-01 3.57677847e-01
7.20000863e-01 -4.43703309e-02 3.72150064e-01 8.39109361e-01
1.12217396e-01 -1.30927041e-01 -1.10543549e+00 8.58234227e-01
1.33053496e-01 -9.51404035e-01 1.00751311e-01 -2.74496019e-01
8.14020813e-01 -5.61180770e-01 1.75167158e-01 7.31322467e-01
4.93572354e-01 -1.10277164e+00 6.69869959e-01 4.89206254e-01
9.79764223e-01 -7.72852480e-01 7.53207207e-01 2.44560137e-01
-9.96786892e-01 -4.02425043e-02 -5.07303834e-01 -8.29526335e-02
-3.03930879e-01 3.73368502e-01 -5.55186331e-01 2.66317576e-01
4.70544606e-01 -6.40218928e-02 -7.68333077e-01 8.04306269e-01
-4.20985639e-01 4.66104895e-01 2.92692259e-02 -6.67899728e-01
1.35349855e-01 -1.52414843e-01 2.74353474e-02 1.18860614e+00
1.84451804e-01 3.59428704e-01 4.35910404e-01 5.75457811e-01
-2.67500013e-01 2.54827827e-01 -1.69726714e-01 3.76664132e-01
7.23992288e-01 1.42899990e+00 -5.87973118e-01 -3.05564165e-01
-3.52998763e-01 6.40876055e-01 4.31420147e-01 2.32173041e-01
-6.22301221e-01 -5.03599703e-01 4.25176054e-01 1.59373745e-01
-1.30619451e-01 4.63138551e-01 -6.44918382e-01 -8.67563665e-01
2.15677872e-01 -1.27283990e+00 5.89602113e-01 -5.06349385e-01
-1.40833020e+00 7.74955869e-01 -2.19735891e-01 -1.52855098e+00
-4.72188354e-01 -5.06313741e-01 -5.60759485e-01 1.00676250e+00
-1.61165893e+00 -8.87377203e-01 -5.41138053e-01 5.35654485e-01
6.88771129e-01 4.88638356e-02 9.88061845e-01 3.82533222e-01
-3.58357698e-01 1.04648602e+00 -1.57847479e-01 -1.91712573e-01
1.16709661e+00 -1.31242466e+00 2.38331985e-02 4.10228640e-01
-4.41423468e-02 7.97286034e-01 5.69849849e-01 -1.89588830e-01
-1.41304493e+00 -6.73774898e-01 1.11983395e+00 -3.89750004e-01
4.80702281e-01 -4.20396864e-01 -9.83391464e-01 -3.02915454e-01
1.07408144e-01 -1.01905793e-01 8.52088034e-01 3.42281312e-01
-6.32380009e-01 -4.14927214e-01 -9.29372489e-01 7.58668423e-01
6.41924024e-01 -7.00711250e-01 -4.16908681e-01 4.36672539e-01
4.21393871e-01 -4.42017287e-01 -5.36912680e-01 2.58421242e-01
7.68134773e-01 -1.01837885e+00 8.67205203e-01 -6.04271591e-01
6.90063596e-01 -1.11291647e-01 -4.13632281e-02 -1.34559596e+00
-3.23840797e-01 -6.00820243e-01 2.50100493e-01 1.10532200e+00
4.96030748e-01 -3.19712132e-01 6.47932768e-01 8.22347105e-01
-1.03665411e-01 -1.11311567e+00 -7.62918174e-01 -5.27913868e-01
2.40967393e-01 -2.57636935e-01 5.58850586e-01 7.65972316e-01
2.37569481e-01 6.92774415e-01 -6.08957171e-01 -3.29370052e-01
1.78061128e-01 5.48031569e-01 8.94564986e-01 -1.35533226e+00
-2.41973236e-01 -7.90941119e-01 2.05935284e-01 -1.47189951e+00
7.22957402e-02 -6.30067647e-01 3.62958819e-01 -1.59755552e+00
4.02597696e-01 -7.42191732e-01 -5.24686337e-01 3.70171428e-01
-6.30088985e-01 2.92401254e-01 6.71370747e-03 3.26513290e-01
-7.92194843e-01 2.60454535e-01 1.34988403e+00 -1.83957338e-01
-1.20477028e-01 1.66901723e-01 -1.10023379e+00 3.03341687e-01
7.56142914e-01 -6.54750884e-01 -5.31432331e-01 -2.81006128e-01
4.32610393e-01 4.60166901e-01 1.66637883e-01 -5.09177566e-01
2.74266124e-01 -6.07383370e-01 6.84857890e-02 -5.09142339e-01
2.74244785e-01 -4.93872464e-01 -4.29727703e-01 1.58908054e-01
-9.53227401e-01 2.43175551e-01 -1.88973725e-01 3.52984309e-01
-2.95648068e-01 -2.60818869e-01 8.62622440e-01 -4.72703204e-03
-4.70017552e-01 5.71642816e-02 -6.41519427e-02 3.91235113e-01
5.80916524e-01 -7.52820373e-02 -7.46183753e-01 -6.44204140e-01
1.70621604e-01 2.71990895e-01 3.38224530e-01 7.84179807e-01
7.84868240e-01 -1.46015954e+00 -9.73877966e-01 7.15337601e-03
5.15591741e-01 -5.01719713e-01 4.34215814e-01 5.42576611e-01
-9.53368396e-02 5.14474034e-01 -7.06394240e-02 -3.92184734e-01
-1.37338829e+00 4.40836638e-01 1.06076509e-01 -7.24416316e-01
-2.88951844e-02 8.48373532e-01 2.60314018e-01 -5.70875525e-01
2.83250391e-01 2.75854617e-02 -4.55500931e-01 1.14895463e-01
7.96381652e-01 1.88762963e-01 2.96360642e-01 -3.12665790e-01
-3.05055916e-01 4.07610804e-01 -2.67796308e-01 -2.27450132e-01
8.66901875e-01 -1.07276514e-01 3.85806221e-03 2.02921972e-01
1.04223049e+00 9.11284313e-02 -8.74245882e-01 -4.51967508e-01
4.54572476e-02 -7.15853214e-01 -6.76832795e-02 -9.41819429e-01
-7.99389780e-01 1.07130814e+00 6.15021586e-01 3.82742375e-01
1.24556613e+00 -2.59183407e-01 6.31812513e-01 3.82922858e-01
3.83775353e-01 -1.61195552e+00 6.62440300e-01 4.44151878e-01
1.02870643e+00 -1.46498728e+00 -4.77259681e-02 -2.61398137e-01
-9.65568721e-01 1.03616655e+00 8.64798427e-01 8.91174376e-02
7.27785677e-02 3.67442742e-02 4.21692282e-01 -2.63754390e-02
-1.21033013e+00 -1.89725891e-01 6.81999862e-01 4.32830572e-01
8.99610400e-01 1.84699416e-01 -5.57742894e-01 5.81132650e-01
-1.99660793e-01 -4.06377256e-01 2.38753200e-01 8.43980193e-01
-6.18373156e-01 -1.32683408e+00 -4.15785313e-01 5.21465361e-01
-4.77087677e-01 -2.75279015e-01 -7.02423453e-01 4.40856308e-01
-4.28483963e-01 1.60877085e+00 -2.07672909e-01 -6.96588755e-01
6.28978193e-01 -9.28529873e-02 2.27100298e-01 -6.45810306e-01
-9.09538746e-01 1.03087738e-01 2.07005665e-01 -2.60662019e-01
-1.46803498e-01 -2.56341964e-01 -1.00182962e+00 -3.47948968e-01
-4.63383198e-01 2.62005180e-01 3.98534447e-01 7.05311120e-01
2.63287425e-01 5.44272780e-01 1.12502468e+00 -2.29154095e-01
-1.08842766e+00 -1.14893508e+00 -1.47651061e-01 7.31769681e-01
-1.56712346e-02 -6.24116242e-01 -2.34747693e-01 -3.27880055e-01] | [12.379620552062988, 7.933199882507324] |
33777c0d-4ad6-4493-bed7-362473d9017f | bkt-lstm-efficient-student-modeling-for | 2012.12218 | null | https://arxiv.org/abs/2012.12218v3 | https://arxiv.org/pdf/2012.12218v3.pdf | BKT-LSTM: Efficient Student Modeling for knowledge tracing and student performance prediction | Recently, we have seen a rapid rise in usage of online educational platforms. The personalized education became crucially important in future learning environments. Knowledge tracing (KT) refers to the detection of students' knowledge states and predict future performance given their past outcomes for providing adaptive solution to Intelligent Tutoring Systems (ITS). Bayesian Knowledge Tracing (BKT) is a model to capture mastery level of each skill with psychologically meaningful parameters and widely used in successful tutoring systems. However, it is unable to detect learning transfer across skills because each skill model is learned independently and shows lower efficiency in student performance prediction. While recent KT models based on deep neural networks shows impressive predictive power but it came with a price. Ten of thousands of parameters in neural networks are unable to provide psychologically meaningful interpretation that reflect to cognitive theory. In this paper, we proposed an efficient student model called BKT-LSTM. It contains three meaningful components: individual \textit{skill mastery} assessed by BKT, \textit{ability profile} (learning transfer across skills) detected by k-means clustering and \textit{problem difficulty}. All these components are taken into account in student's future performance prediction by leveraging predictive power of LSTM. BKT-LSTM outperforms state-of-the-art student models in student's performance prediction by considering these meaningful features instead of using binary values of student's past interaction in DKT. We also conduct ablation studies on each of BKT-LSTM model components to examine their value and each component shows significant contribution in student's performance prediction. Thus, it has potential for providing adaptive and personalized instruction in real-world educational systems. | ['Sein Minn'] | 2020-12-22 | null | null | null | null | ['skill-mastery'] | ['robots'] | [-4.58569527e-02 2.24631980e-01 -2.11005062e-01 -3.47450912e-01
-1.86574072e-01 -5.25937080e-01 2.11986437e-01 5.77967167e-01
-3.52135390e-01 7.95612574e-01 3.41429450e-02 -6.18828356e-01
-1.13881683e+00 -9.93791461e-01 -6.50880337e-01 -4.11499113e-01
-4.28507701e-02 5.73951006e-01 5.21704733e-01 -4.06222045e-01
3.70747626e-01 4.98951882e-01 -1.95023608e+00 1.49506599e-01
1.60160053e+00 8.17876518e-01 4.51985747e-01 6.49029553e-01
-4.47642297e-01 1.08858800e+00 -8.53804469e-01 -3.13908935e-01
-3.11957151e-01 -4.12675828e-01 -1.02520144e+00 -6.86185777e-01
6.25437379e-01 -2.74783790e-01 -3.26383412e-01 1.03715253e+00
3.70325655e-01 5.45850515e-01 7.50603974e-01 -1.21775019e+00
-1.08178151e+00 7.46097565e-01 -1.97904900e-01 3.60486865e-01
3.17263752e-01 -1.15572155e-01 2.34463841e-01 -2.34698355e-01
-8.69364962e-02 1.09043312e+00 8.01497281e-01 6.68750584e-01
-8.90433729e-01 -5.91613173e-01 5.95507771e-02 8.02747130e-01
-6.90674365e-01 1.49044275e-01 4.95276511e-01 -6.54221356e-01
7.50690222e-01 1.04422592e-01 1.05927396e+00 1.01076460e+00
3.20893943e-01 9.95447218e-01 1.45699918e+00 -3.67673665e-01
1.80624783e-01 3.71545315e-01 7.94065177e-01 7.25158989e-01
8.29403251e-02 1.27714917e-01 -8.52662206e-01 1.94979817e-01
7.95052528e-01 1.04572937e-01 -1.27711385e-01 1.04417711e-01
-8.62290502e-01 4.66306061e-01 2.78027833e-01 4.52099115e-01
-2.95832604e-01 6.33677468e-02 1.90639585e-01 8.53881657e-01
1.22816592e-01 4.01653737e-01 -8.20078135e-01 -6.87071860e-01
-7.19088554e-01 8.79018232e-02 6.05092108e-01 7.94700623e-01
4.64278281e-01 4.61759686e-01 -6.23563647e-01 9.60829616e-01
-5.44724576e-02 3.29109877e-01 1.38459706e+00 -7.98577845e-01
9.06515867e-03 1.24807203e+00 -3.47868592e-01 -7.92398155e-01
-4.77437139e-01 -6.93622768e-01 -3.24119270e-01 3.01332682e-01
4.10164088e-01 -1.94999874e-01 -1.03652513e+00 1.82864523e+00
1.12453192e-01 7.12588191e-01 2.71487683e-01 2.71376133e-01
1.23697948e+00 4.57608253e-01 3.08098018e-01 -1.70987472e-01
1.28348470e+00 -6.48753524e-01 -7.73898661e-01 2.53541991e-02
8.70835066e-01 -4.92159814e-01 9.35695887e-01 7.67253637e-01
-1.33378005e+00 -9.65025067e-01 -8.31193507e-01 1.63553298e-01
-7.71727800e-01 -1.67949528e-01 5.31300306e-01 9.93688345e-01
-1.29985666e+00 8.94970894e-01 -4.96470928e-01 -2.26253331e-01
2.51422495e-01 6.51987314e-01 5.42567261e-02 3.08279723e-01
-1.33742130e+00 1.13622510e+00 4.58002537e-01 -2.16211826e-01
-9.64398742e-01 -9.56916809e-01 -4.70064551e-01 7.20224321e-01
2.77896494e-01 -6.52002513e-01 1.35976458e+00 -9.51170981e-01
-2.07901955e+00 4.69277769e-01 -6.04946464e-02 -3.51648957e-01
2.76845485e-01 -4.28218365e-01 -2.89224446e-01 -1.93381816e-01
-3.10414851e-01 2.43165478e-01 2.24646553e-01 -8.30606341e-01
-8.49662423e-01 -6.74436927e-01 -2.66413897e-01 5.50452054e-01
-9.71048772e-01 -3.51314545e-01 -1.75036341e-01 -2.55206347e-01
2.63488412e-01 -4.97631818e-01 1.35315761e-01 -7.81519413e-01
1.23585090e-01 -1.07405353e+00 8.83212149e-01 -7.19279706e-01
1.51041889e+00 -1.40492833e+00 -4.66168206e-03 2.57022291e-01
2.53780633e-01 8.59097421e-01 1.17369473e-01 1.26864240e-01
3.64065468e-02 6.44276738e-02 5.02707899e-01 1.54781446e-01
1.91032633e-01 2.72029519e-01 -9.61641967e-02 -2.71435678e-01
-1.50369138e-01 9.24195826e-01 -1.02753925e+00 -4.03504759e-01
3.07777971e-01 7.23669901e-02 -3.61395121e-01 4.84472454e-01
-1.31452933e-01 2.73350209e-01 -6.05551541e-01 4.33241636e-01
3.48841041e-01 1.82023346e-01 -1.04815528e-01 5.75123072e-01
-1.45835474e-01 2.22493231e-01 -8.79719138e-01 1.27502406e+00
-2.82270432e-01 5.74035645e-01 -4.15098220e-01 -1.21602201e+00
1.26417768e+00 4.51497197e-01 1.68366060e-01 -9.39553261e-01
1.42772987e-01 3.18814218e-02 4.42600429e-01 -8.43865573e-01
4.56208199e-01 3.08895111e-02 2.48400509e-01 3.47472310e-01
3.95457417e-01 8.52490067e-02 -2.80904859e-01 1.08990334e-01
1.08338368e+00 2.96304256e-01 -1.22131117e-01 -3.03815633e-01
6.65115416e-01 -2.44014069e-01 6.61131203e-01 9.14913118e-01
-5.03667951e-01 -2.64099479e-01 4.22744393e-01 -2.30674788e-01
-3.93523812e-01 -8.93427908e-01 2.62955930e-02 1.65291715e+00
-1.97678551e-01 1.05748884e-01 -8.81120026e-01 -5.18022776e-01
1.15879901e-01 1.00079262e+00 -7.15615273e-01 -8.02002132e-01
-5.13759553e-01 -4.48074788e-01 6.51521623e-01 5.98194480e-01
6.82559729e-01 -1.33198190e+00 -5.14016688e-01 2.54910678e-01
1.37874454e-01 -5.38942397e-01 2.61306018e-01 2.35270858e-01
-1.10742271e+00 -8.76574874e-01 -6.13889992e-01 -9.35233116e-01
2.96536565e-01 7.28156567e-02 1.06253695e+00 2.42967665e-01
-1.01669375e-02 8.23167920e-01 -1.77757621e-01 -8.29774618e-01
-2.10509792e-01 4.61575435e-03 5.86763799e-01 -4.93256688e-01
8.66932869e-01 -7.64920712e-01 -6.19702578e-01 2.78222561e-01
-4.12913114e-01 -8.82624164e-02 4.60957140e-01 6.58591449e-01
1.26232669e-01 4.05527592e-01 1.18618894e+00 -7.29940474e-01
1.16420925e+00 -6.92036390e-01 -2.85734683e-01 6.18288875e-01
-1.12187231e+00 -2.94912420e-02 5.51339149e-01 -6.99746251e-01
-1.53441048e+00 -7.09461808e-01 -1.14186496e-01 -5.28768599e-01
-5.22525370e-01 8.04747045e-01 3.30435961e-01 -5.23795858e-02
5.97020268e-01 5.94256163e-01 -2.07511723e-01 -6.32615268e-01
-2.43354842e-01 5.34446657e-01 6.82595730e-01 -1.18000007e+00
3.77595365e-01 -5.33706248e-01 3.33390087e-02 -4.71457452e-01
-1.06721580e+00 -1.24234490e-01 -5.65321803e-01 -7.41720319e-01
5.27560949e-01 -5.98280907e-01 -1.83689845e+00 7.46626496e-01
-5.09147882e-01 -5.82557142e-01 -1.68561384e-01 6.33320570e-01
-4.41687912e-01 1.06623387e-02 -7.08630919e-01 -1.26005149e+00
-2.89421827e-01 -1.10437310e+00 -4.89063235e-03 1.15129077e+00
-9.52838659e-02 -1.39730144e+00 7.48325586e-02 7.82152236e-01
5.02607405e-01 -6.40159100e-02 1.30626142e+00 -1.06719637e+00
-1.44538105e-01 1.15580987e-02 1.87704891e-01 3.51728588e-01
-1.95288375e-01 -1.51392639e-01 -1.09295428e+00 -1.09848291e-01
1.97120398e-01 -7.85147727e-01 6.68290794e-01 5.01540899e-01
1.46044552e+00 -9.49032456e-02 -2.16761101e-02 4.00375277e-01
1.07747543e+00 6.79191768e-01 3.83038640e-01 5.79907238e-01
7.31428981e-01 9.21648681e-01 4.38263327e-01 -8.77425820e-02
6.52428985e-01 3.99357468e-01 1.34288028e-01 4.73849088e-01
-1.30533576e-01 -4.12558258e-01 6.62063241e-01 1.36856782e+00
-3.47186387e-01 5.31612597e-02 -1.30524063e+00 5.56160033e-01
-1.90966392e+00 -9.80057478e-01 -4.31022763e-01 2.31620336e+00
9.91316497e-01 1.91074878e-01 -1.26074731e-01 -2.09539905e-02
3.96904796e-01 -6.15968823e-01 -1.03066027e+00 -1.01225221e+00
1.94532290e-01 4.48192239e-01 2.53087848e-01 4.21352863e-01
-2.31389597e-01 1.11150527e+00 6.08761644e+00 1.04486096e+00
-9.00461257e-01 8.47323090e-02 6.36760771e-01 8.90676603e-02
-1.35379940e-01 -3.68590266e-01 -1.05196476e+00 4.11948621e-01
1.57994115e+00 -4.64381635e-01 1.07411183e-01 8.02386105e-01
2.23290667e-01 -5.01104593e-01 -8.54080915e-01 5.63753128e-01
-1.63994342e-01 -8.03616464e-01 9.98332798e-02 -6.02026097e-02
9.19440985e-01 -4.60218549e-01 5.74360549e-01 1.12276268e+00
7.69445717e-01 -1.24169445e+00 1.79815814e-01 9.75976646e-01
2.47975886e-01 -9.78069067e-01 7.38826513e-01 7.26173818e-01
-6.01660848e-01 -4.43983704e-01 -4.45065975e-01 -4.41431254e-01
-5.79124570e-01 2.31446102e-02 -1.16310728e+00 3.06093752e-01
9.05658722e-01 3.35893989e-01 -7.28485346e-01 9.37270224e-01
-1.93970352e-01 1.17938280e+00 -5.69496378e-02 -2.58088559e-01
2.80453116e-01 -2.26652518e-01 1.02344334e-01 7.57782757e-01
8.44960928e-01 5.20585597e-01 8.57453793e-02 8.36902022e-01
4.52127941e-02 2.37733237e-02 -3.66805494e-01 1.62369773e-01
7.59857893e-01 9.40561414e-01 -5.00765979e-01 -5.98979056e-01
-1.59266874e-01 7.25982010e-01 6.23350978e-01 3.70828152e-01
-4.82226253e-01 -1.50648132e-01 6.05012357e-01 -9.44162607e-02
-1.22463159e-01 1.22430488e-01 -5.20959258e-01 -6.14326537e-01
-4.47353363e-01 -6.46808386e-01 5.09875834e-01 -9.02774811e-01
-1.09622037e+00 1.49288833e-01 1.50890211e-02 -7.54587471e-01
-1.64875075e-01 -9.27868307e-01 -1.19738603e+00 1.14639688e+00
-1.27289581e+00 -7.24457562e-01 -5.19454718e-01 9.35486555e-01
4.57705379e-01 -5.29165924e-01 8.11393559e-01 -1.90650091e-01
-1.05767465e+00 9.16390598e-01 6.35567844e-01 -8.03407282e-02
6.81365550e-01 -1.79447794e+00 -2.29579687e-01 3.16831350e-01
-5.23206294e-01 7.51231551e-01 6.48654401e-01 -9.02516007e-01
-1.24211442e+00 -6.23295963e-01 6.34503663e-01 -7.07237601e-01
5.70543051e-01 3.58121395e-01 -1.41323483e+00 7.44882464e-01
2.51409262e-01 -8.55814576e-01 1.06326246e+00 5.56655765e-01
2.11454257e-01 -1.06655069e-01 -1.12044013e+00 6.12785876e-01
7.21828699e-01 -7.45268837e-02 -1.11371505e+00 1.02606222e-01
6.27665401e-01 -3.99780005e-01 -1.53417277e+00 4.75529552e-01
3.74585539e-01 -1.27242815e+00 8.82836878e-01 -6.67885900e-01
2.80084640e-01 3.00591171e-01 6.68288589e-01 -1.53169763e+00
-4.72162783e-01 -3.32969606e-01 -5.65707862e-01 1.27663708e+00
4.62762266e-02 -5.41387737e-01 1.23137665e+00 1.09047079e+00
-5.68381488e-01 -1.05243957e+00 -7.49181330e-01 -6.14512801e-01
5.73580503e-01 -2.14845479e-01 6.77549422e-01 1.47759020e+00
4.21285510e-01 1.32863328e-01 -3.77884284e-02 1.23596285e-02
5.40823102e-01 -1.06103897e-01 5.74017465e-01 -1.79553938e+00
-5.18016145e-02 -9.65554893e-01 -3.37128133e-01 -8.30678642e-01
1.14368349e-01 -6.98184848e-01 -4.56605345e-01 -1.38070190e+00
2.36695379e-01 -3.17426205e-01 -8.02295923e-01 5.73640764e-01
-4.47992802e-01 -5.66337168e-01 -2.94010788e-01 -2.53006548e-01
-4.96800184e-01 6.98865891e-01 1.53430951e+00 3.69114786e-01
-5.06288707e-01 1.01701073e-01 -6.32034898e-01 1.02353406e+00
1.06753755e+00 -2.94823587e-01 -8.05834591e-01 -2.50950605e-01
1.76833048e-01 4.12256211e-01 9.13746282e-03 -1.35207701e+00
7.50995278e-01 -5.23139417e-01 9.71607208e-01 -4.44828033e-01
3.29494685e-01 -5.17457545e-01 -3.74838442e-01 5.09305358e-01
-3.10476899e-01 3.12499162e-02 5.52336395e-01 2.44195446e-01
-5.58035374e-02 -7.62989640e-01 5.50666153e-01 -2.50415921e-01
-8.38768244e-01 8.50547031e-02 -6.87338710e-01 -3.07288229e-01
1.10168386e+00 -6.05938852e-01 -5.28820932e-01 -4.00874823e-01
-8.62264574e-01 7.70877779e-01 -1.27703443e-01 6.37021184e-01
8.49469841e-01 -1.17739046e+00 -3.42589408e-01 5.53497821e-02
-3.68040502e-01 -2.36848406e-02 8.59223664e-01 7.90302873e-01
-3.74794185e-01 6.45418644e-01 -6.56678438e-01 -4.50586230e-01
-1.52019656e+00 3.87784123e-01 5.80583394e-01 -3.44084412e-01
-2.38021091e-01 1.02012610e+00 -2.67972145e-02 -6.83878720e-01
6.41215444e-01 -3.52165133e-01 -1.08809507e+00 1.18912168e-01
5.60395062e-01 7.39223897e-01 -1.73596814e-01 -1.68619737e-01
4.18554276e-01 3.23504716e-01 -3.29212666e-01 8.95515457e-02
1.34720838e+00 6.86477497e-03 7.12723434e-02 7.60285497e-01
4.48371142e-01 -5.49256384e-01 -9.25462961e-01 -3.80273461e-01
3.26319396e-01 -5.28448001e-02 2.71238118e-01 -1.44075501e+00
-7.65023351e-01 1.01004755e+00 8.98637712e-01 1.64834514e-01
9.84966040e-01 -4.05769080e-01 7.04661191e-01 7.40834236e-01
8.55399817e-02 -1.44382083e+00 4.21133846e-01 9.91011798e-01
3.74029815e-01 -1.26899064e+00 -3.42975765e-01 2.29878575e-01
-3.16447556e-01 1.16386521e+00 1.65092695e+00 2.39830151e-01
4.96991813e-01 -2.95170009e-01 1.00563407e-01 -3.76298875e-01
-9.07960236e-01 -4.68184687e-02 7.45134234e-01 3.83080155e-01
7.56676137e-01 2.64860213e-01 -4.43994477e-02 1.11670113e+00
-7.00426757e-01 -7.24142492e-02 5.48988104e-01 8.90079916e-01
-1.20475698e+00 -9.40407515e-01 -8.10214758e-01 7.39559770e-01
-2.47192293e-01 -3.96652371e-02 -3.76696378e-01 4.56525594e-01
3.17411512e-01 9.24714386e-01 -2.21898630e-01 -5.77982366e-01
4.30747360e-01 7.82242537e-01 5.42171419e-01 -4.29176211e-01
-1.22794080e+00 -4.21789378e-01 -3.44763905e-01 -2.40458027e-01
-1.31803319e-01 -4.63739485e-01 -1.23513389e+00 -6.53978050e-01
-2.07618609e-01 4.50151086e-01 4.79754150e-01 1.13382173e+00
5.82868271e-02 1.13293266e+00 -1.44551573e-02 -3.09194297e-01
-4.52633888e-01 -1.36640561e+00 -5.57748675e-01 2.38604531e-01
-2.59160642e-02 -8.12201023e-01 -2.57729366e-02 -1.46627024e-01] | [10.131118774414062, 7.173466682434082] |
6c277f33-aa0b-4341-8349-c4238ad0c7d1 | negation-handling-in-machine-learning-based | 2107.11597 | null | https://arxiv.org/abs/2107.11597v1 | https://arxiv.org/pdf/2107.11597v1.pdf | Negation Handling in Machine Learning-Based Sentiment Classification for Colloquial Arabic | One crucial aspect of sentiment analysis is negation handling, where the occurrence of negation can flip the sentiment of a sentence and negatively affects the machine learning-based sentiment classification. The role of negation in Arabic sentiment analysis has been explored only to a limited extent, especially for colloquial Arabic. In this paper, the author addresses the negation problem of machine learning-based sentiment classification for a colloquial Arabic language. To this end, we propose a simple rule-based algorithm for handling the problem; the rules were crafted based on observing many cases of negation. Additionally, simple linguistic knowledge and sentiment lexicon are used for this purpose. The author also examines the impact of the proposed algorithm on the performance of different machine learning algorithms. The results given by the proposed algorithm are compared with three baseline models. The experimental results show that there is a positive impact on the classifiers accuracy, precision and recall when the proposed algorithm is used compared to the baselines. | ['Omar Al-Harbi'] | 2021-07-24 | null | null | null | null | ['arabic-sentiment-analysis'] | ['natural-language-processing'] | [ 1.52400015e-02 -1.14715798e-02 -2.44185445e-03 -6.19559348e-01
-2.90444251e-02 -7.93317497e-01 6.57888055e-01 6.48156106e-01
-6.16593421e-01 9.40483510e-01 4.84752171e-02 -2.69405127e-01
1.28064275e-01 -8.42165649e-01 -2.71145910e-01 -6.17126465e-01
2.11947367e-01 3.73236835e-03 4.87169214e-02 -1.03534794e+00
7.91628182e-01 2.40228534e-01 -1.66058969e+00 5.62899411e-01
6.59466743e-01 8.55794251e-01 -2.10461244e-01 4.98492479e-01
-2.93481141e-01 1.06319654e+00 -8.65423858e-01 -6.55008972e-01
1.57554820e-01 -5.46844423e-01 -6.84693754e-01 -6.46734759e-02
2.06480455e-02 3.13454300e-01 6.04268610e-01 1.16523731e+00
2.28113070e-01 -1.39830187e-02 9.57383156e-01 -1.26492107e+00
-2.97354460e-01 4.99143034e-01 -6.47431076e-01 9.80501175e-02
4.51224327e-01 -5.31587243e-01 8.29435408e-01 -1.00178289e+00
3.90731931e-01 1.08476102e+00 7.78975904e-01 1.35661244e-01
-3.34611565e-01 -5.46139479e-01 1.26458764e-01 1.67952105e-01
-1.13470566e+00 -4.90830801e-02 7.84676075e-01 -1.91270158e-01
1.01307666e+00 4.79763746e-02 6.09967589e-01 3.00205141e-01
6.43751025e-01 4.06090587e-01 1.60092080e+00 -1.06072366e+00
1.87081262e-01 6.41946793e-01 5.46734393e-01 6.38148606e-01
4.91501004e-01 -4.27901477e-01 -5.52345216e-01 -5.24154305e-03
-3.09347272e-01 -2.60013998e-01 1.39647126e-01 2.26867959e-01
-5.13506353e-01 9.59129751e-01 8.82639065e-02 6.85779870e-01
-2.95575768e-01 -3.96263152e-01 8.34777355e-01 6.32395983e-01
4.72448677e-01 3.46962690e-01 -7.63831615e-01 -1.37062624e-01
-7.80033112e-01 3.03023577e-01 9.94081676e-01 6.10476017e-01
5.13837457e-01 -3.50997299e-02 3.67583960e-01 6.00611866e-01
4.79318380e-01 6.25199437e-01 6.45968437e-01 -1.14284001e-01
1.87823161e-01 1.12467206e+00 3.85573238e-01 -1.35742128e+00
-4.74297613e-01 -1.02237254e-01 -5.17017901e-01 3.64811420e-01
4.66591716e-01 -4.46207047e-01 -7.23986804e-01 1.28789461e+00
3.56829166e-01 -5.07243931e-01 7.09289312e-01 7.16516972e-01
7.37480223e-01 7.07967579e-01 1.44385979e-01 -3.61589402e-01
1.38887608e+00 -8.52099895e-01 -1.21429670e+00 -4.94296402e-02
8.98620784e-01 -1.34914839e+00 1.09093142e+00 9.41219628e-01
-6.80930614e-01 -2.64854550e-01 -1.34084427e+00 2.33231187e-01
-1.00610721e+00 2.98466116e-01 6.47776902e-01 1.22010934e+00
-6.34790599e-01 2.76820540e-01 -4.89089489e-01 -5.35627663e-01
-8.62666294e-02 6.02558136e-01 -3.46168846e-01 1.55411795e-01
-1.34319746e+00 1.44563973e+00 5.28274596e-01 3.21093142e-01
1.38391539e-01 1.44421935e-01 -6.87581360e-01 -3.26478511e-01
1.46120608e-01 1.60849661e-01 1.19000185e+00 -1.70309639e+00
-1.31941307e+00 8.11011374e-01 -3.27733368e-01 -3.19182396e-01
2.05282569e-01 -2.11113766e-01 -6.31686270e-01 -6.28052056e-02
1.46952113e-02 7.90233985e-02 4.85611588e-01 -1.17802763e+00
-8.87594998e-01 -3.99111360e-01 3.77382278e-01 3.79585654e-01
-4.66999203e-01 2.55122304e-01 4.46979469e-03 -6.71761274e-01
1.26879200e-01 -9.17281926e-01 4.73618805e-02 -6.38576806e-01
-3.52291726e-02 -4.27060992e-01 9.75215495e-01 -3.79834443e-01
1.32326269e+00 -1.84675539e+00 -3.42587799e-01 5.81494629e-01
-6.04259551e-01 4.36092526e-01 3.32707763e-01 7.33692646e-01
-4.96802181e-02 2.41784289e-01 -2.05496117e-01 7.13356361e-02
-3.00907820e-01 2.71968812e-01 -1.91786364e-01 3.55292171e-01
1.61187023e-01 2.03817174e-01 -6.65418863e-01 -4.52448368e-01
7.56216347e-02 3.97012860e-01 -2.99435705e-01 -3.95857953e-02
-1.81094840e-01 -1.28311798e-01 -3.99383217e-01 8.66719067e-01
5.94072104e-01 2.59665698e-01 5.81809819e-01 -4.69064564e-01
-6.41135946e-02 6.53025284e-02 -1.37053406e+00 9.03924167e-01
-3.75620127e-01 3.04640234e-01 -3.29203725e-01 -9.31263387e-01
1.10258615e+00 4.20082837e-01 -8.94510671e-02 -4.28947091e-01
7.68430531e-01 5.47587574e-01 1.14383280e-01 -5.82222939e-01
7.49139726e-01 -5.72265923e-01 -9.25720707e-02 5.90978503e-01
-1.33941904e-01 -2.48578787e-01 7.23100960e-01 -4.02270496e-04
4.15859550e-01 2.34429404e-01 8.73639584e-01 -5.06377995e-01
1.16035759e+00 6.23574913e-01 2.67563611e-01 3.94855261e-01
-1.63160786e-01 -4.04498093e-02 6.64718926e-01 -3.74836117e-01
-6.49384916e-01 -2.34437421e-01 -1.24011368e-01 1.08957660e+00
5.20612560e-02 -3.30691099e-01 -8.23753774e-01 -9.89708126e-01
-1.75576985e-01 5.19235909e-01 -6.93927586e-01 -4.08676118e-02
-4.25829083e-01 -1.24301541e+00 4.64162320e-01 2.07974926e-01
4.28501338e-01 -1.18329668e+00 -7.03923047e-01 1.06010465e-02
9.05585475e-03 -8.79937172e-01 3.29405159e-01 4.31449920e-01
-9.01179492e-01 -1.28128445e+00 -1.30163953e-01 -1.00119400e+00
8.30732346e-01 5.88381523e-03 9.17913020e-01 4.95240897e-01
4.10931468e-01 -5.36012948e-02 -1.13175297e+00 -1.22026718e+00
-5.24642467e-01 1.40706882e-01 7.99418613e-02 -1.57727286e-01
1.00999820e+00 5.15873432e-02 -1.56549782e-01 2.39324495e-02
-1.06771410e+00 -5.84494889e-01 5.87792695e-01 6.87849998e-01
1.31480709e-01 5.26420951e-01 1.06353891e+00 -1.53187501e+00
1.12410498e+00 -4.28940445e-01 -4.07689542e-01 2.42046475e-01
-8.40649903e-01 1.20540122e-02 8.72757614e-01 -1.17262118e-01
-1.03233135e+00 -2.19431128e-02 -1.79975912e-01 6.07223213e-01
-6.77288100e-02 9.88272250e-01 -1.53642995e-02 -1.55897871e-01
6.51373684e-01 -3.56850363e-02 1.06854081e-01 -7.63223991e-02
3.33002247e-02 1.02108252e+00 -9.39516351e-02 -3.15761834e-01
3.04439574e-01 3.44853729e-01 1.64512798e-01 -8.62987339e-01
-1.16121614e+00 -3.92448187e-01 -5.75496852e-01 -3.50441217e-01
5.29694974e-01 -6.44729435e-01 -5.13435423e-01 6.82747662e-01
-9.35522974e-01 3.24726492e-01 2.42353633e-01 4.43919986e-01
-1.23127788e-01 3.36984009e-01 -3.76097113e-01 -1.15424442e+00
-4.60664481e-01 -1.01303399e+00 3.46498340e-01 4.07105893e-01
-5.02993882e-01 -1.14146674e+00 2.46248324e-03 3.35893750e-01
1.23188175e-01 3.39953780e-01 1.10436928e+00 -1.06712854e+00
5.48976362e-01 -5.20289063e-01 1.61365241e-01 6.94393694e-01
4.20487165e-01 1.69346645e-01 -8.63972008e-01 -3.95597629e-02
3.18373322e-01 -4.82338309e-01 3.55586559e-01 -4.31380086e-02
2.97001362e-01 -2.11004317e-01 1.65276825e-01 -2.73301214e-01
1.81473029e+00 5.20141900e-01 3.83589268e-01 8.70532036e-01
1.65315405e-01 8.34033966e-01 1.40028524e+00 3.24461192e-01
2.38878906e-01 -4.10712101e-02 3.46752197e-01 1.14537515e-01
4.11207169e-01 2.37048909e-01 4.48695153e-01 1.11112964e+00
7.01946691e-02 -4.84387964e-01 -9.99915600e-01 5.70596099e-01
-1.67724323e+00 -5.33638597e-01 -5.28342783e-01 1.76400149e+00
9.42069054e-01 5.01924217e-01 -9.15712342e-02 1.03837430e+00
4.08213973e-01 -1.00851953e-01 9.26417112e-02 -1.25771236e+00
-3.47816557e-01 4.61827517e-01 2.95025706e-01 7.44009316e-01
-1.21973240e+00 1.09298193e+00 5.83074188e+00 4.61312622e-01
-1.49686801e+00 4.49221916e-02 2.96612054e-01 2.22913027e-01
3.97852808e-02 -1.11674748e-01 -7.51475513e-01 1.70044616e-01
6.78883791e-01 -1.33170960e-02 -2.21766874e-01 6.91261590e-01
4.11228031e-01 -7.78674841e-01 -4.95257944e-01 4.80437726e-01
5.54472983e-01 -7.23873436e-01 2.64014781e-01 -4.74182010e-01
7.70440996e-01 -4.55069780e-01 -1.17427789e-01 2.05918238e-01
-1.03194475e-01 -1.00947642e+00 6.81863129e-01 4.47613925e-01
6.89248741e-03 -1.41504967e+00 1.70687473e+00 3.16537946e-01
-5.92189968e-01 -2.84505840e-02 -1.83427230e-01 -7.69571245e-01
-3.97373550e-02 4.83160108e-01 -9.04280663e-01 5.08409321e-01
6.52472317e-01 4.84043032e-01 -6.27077401e-01 4.72444892e-01
-4.58815813e-01 7.72114515e-01 -2.86586016e-01 -8.11158121e-01
5.40514171e-01 -2.32040584e-01 1.14642590e-01 1.40032041e+00
8.06219056e-02 2.58902520e-01 -1.48254916e-01 -2.39870712e-01
1.96466491e-01 1.03925633e+00 -7.01628506e-01 -2.90803853e-02
1.69109106e-01 1.23816335e+00 -1.20091379e+00 -3.24383974e-01
-5.60048580e-01 6.09847009e-01 -3.68401669e-02 -1.62548870e-02
-3.44253778e-01 -7.42402554e-01 -4.24874313e-02 1.12445571e-01
2.40983784e-01 1.43216893e-01 -6.71041429e-01 -7.81185329e-01
1.84714273e-01 -1.20514882e+00 3.96116018e-01 -6.42709792e-01
-1.24181771e+00 7.56299019e-01 -1.33844510e-01 -1.03730214e+00
-1.78984478e-01 -1.10633433e+00 -4.65271473e-01 8.03598881e-01
-1.57821321e+00 -1.23792338e+00 3.33726592e-02 4.22601134e-01
3.08047086e-01 -5.04989326e-01 9.99682248e-01 1.36468068e-01
-1.92454055e-01 4.98128623e-01 -1.85555533e-01 1.42578349e-01
9.53553498e-01 -1.13655782e+00 -4.87439871e-01 1.07579327e+00
-1.77700326e-01 8.14884961e-01 1.23926198e+00 -7.77276993e-01
-1.12117720e+00 -7.36071587e-01 1.47128356e+00 -3.70119780e-01
6.04461193e-01 1.57080516e-02 -6.54162467e-01 4.60178524e-01
5.08986235e-01 -3.66561621e-01 1.05788517e+00 1.28174171e-01
-1.68259457e-01 -1.05619788e-01 -1.51240456e+00 4.33732897e-01
-1.43988073e-01 -2.96464771e-01 -9.55927193e-01 2.09183559e-01
-6.63395450e-02 -1.67413950e-01 -6.21082127e-01 3.15405399e-01
7.45444298e-01 -9.43014443e-01 1.39899984e-01 -6.36254370e-01
5.66573620e-01 -6.78293169e-01 -4.02960390e-01 -1.05192697e+00
4.67003256e-01 4.90713455e-02 2.76954263e-01 1.09379351e+00
8.53833854e-01 -4.99657393e-01 7.92104065e-01 1.97123960e-01
2.59195328e-01 -7.84620583e-01 -4.26981300e-01 -1.51459441e-01
3.13644856e-01 -3.51024538e-01 2.32915297e-01 1.15040350e+00
3.25268418e-01 5.07818162e-01 -2.21300408e-01 4.92159761e-02
8.64112675e-02 -8.91479030e-02 5.70978522e-01 -1.08726072e+00
2.75908083e-01 1.59037858e-02 -5.88501036e-01 -1.40577599e-01
-1.96256135e-02 -6.47807002e-01 -1.96447611e-01 -1.22915506e+00
-1.28670216e-01 -2.57798284e-01 -2.90341824e-01 5.62092602e-01
-2.12971285e-01 6.25758469e-01 3.07618469e-01 -1.98466018e-01
-3.20981979e-01 -1.02087669e-02 9.85868871e-01 1.21687792e-01
-1.21824086e-01 -1.13724075e-01 -9.75376606e-01 1.20420408e+00
1.06699419e+00 -8.56317997e-01 -4.18845743e-01 -1.07620955e-02
1.04871309e+00 -4.55312580e-01 -3.64692241e-01 -8.21706593e-01
1.41080379e-01 -1.66088551e-01 2.73154169e-01 -5.97113907e-01
4.19936478e-02 -1.05589283e+00 -4.64658976e-01 5.90445220e-01
-8.40023383e-02 6.82403564e-01 2.07944795e-01 1.54101834e-01
-6.46258891e-01 -8.69110584e-01 7.88467109e-01 -2.97583699e-01
-6.27248287e-01 -6.65883303e-01 -7.50779808e-01 -7.05685541e-02
1.25074756e+00 -2.99104303e-01 3.09401359e-02 -2.82251954e-01
-5.54429233e-01 1.31782144e-03 3.65179151e-01 4.31272477e-01
3.21841121e-01 -8.80168259e-01 -4.97010350e-01 -1.53302001e-02
1.86104253e-01 -4.54812527e-01 -3.45737606e-01 6.91898525e-01
-1.04629850e+00 3.46297413e-01 -3.67454201e-01 -1.09728955e-01
-1.60581744e+00 1.47425339e-01 2.51382142e-01 -2.04706475e-01
2.39107057e-01 4.93277073e-01 -7.14478076e-01 -5.99078238e-01
3.08660269e-02 -3.34897995e-01 -1.07605469e+00 5.43180645e-01
3.55259806e-01 1.67295322e-01 5.22602022e-01 -1.02528238e+00
-4.87929016e-01 6.27628267e-01 -3.39462757e-01 -2.65998632e-01
1.22301900e+00 -1.76405329e-02 -6.44428194e-01 9.10131633e-01
6.81952417e-01 5.00981510e-01 -8.85825083e-02 2.34140977e-01
3.08666080e-01 -1.67112902e-01 3.04956883e-02 -1.32358229e+00
-5.70966542e-01 5.17375290e-01 6.55658722e-01 3.06308359e-01
1.24054492e+00 -7.49280930e-01 3.89134318e-01 7.99162924e-01
1.87001169e-01 -1.47555602e+00 -2.45668128e-01 1.01950538e+00
4.47440177e-01 -1.51798785e+00 3.01426411e-01 -4.69054490e-01
-9.12461460e-01 1.35799682e+00 5.31206310e-01 -3.07296723e-01
9.97747600e-01 4.11571115e-01 8.77733886e-01 -2.51411587e-01
-5.61950207e-01 1.27276421e-01 4.03718688e-02 4.43712682e-01
1.15170097e+00 -1.33113965e-01 -1.37021351e+00 9.33642209e-01
-5.45558214e-01 2.15529814e-01 1.07346475e+00 1.64156210e+00
-6.44557834e-01 -1.46300709e+00 -6.00127280e-01 4.17847633e-01
-1.24962485e+00 -1.63899109e-01 -7.51180649e-01 1.01152408e+00
4.36779022e-01 1.41139781e+00 -4.14091706e-01 -2.02860162e-01
4.50210005e-01 1.77830502e-01 1.81048989e-01 -6.17492139e-01
-1.32650971e+00 -1.96069583e-01 3.55357170e-01 1.73308730e-01
-1.12143648e+00 -6.19715154e-01 -1.47303522e+00 -2.54579484e-01
-6.02822483e-01 7.20410764e-01 9.59002554e-01 1.24027812e+00
-2.10217714e-01 1.76859021e-01 4.86891180e-01 -1.78012893e-01
-4.95994419e-01 -1.12744856e+00 -5.09623230e-01 3.75505060e-01
2.41339147e-01 -5.98335564e-01 -5.69372475e-01 1.61004379e-01] | [11.051799774169922, 6.91135835647583] |
96858f20-b829-4a36-94c8-da8b53b7e2e2 | a-computational-model-of-infant-learning-and | 2106.16059 | null | https://arxiv.org/abs/2106.16059v1 | https://arxiv.org/pdf/2106.16059v1.pdf | A Computational Model of Infant Learning and Reasoning with Probabilities | Recent experiments reveal that 6- to 12-month-old infants can learn probabilities and reason with them. In this work, we present a novel computational system called Neural Probability Learner and Sampler (NPLS) that learns and reasons with probabilities, providing a computationally sufficient mechanism to explain infant probabilistic learning and inference. In 24 computer simulations, NPLS simulations show how probability distributions can emerge naturally from neural-network learning of event sequences, providing a novel explanation of infant probabilistic learning and reasoning. Three mathematical proofs show how and why NPLS simulates the infant results so accurately. The results are situated in relation to seven other active research lines. This work provides an effective way to integrate Bayesian and neural-network approaches to cognition. | ['Ardavan S Nobandegani', 'Thomas R Shultz'] | 2021-06-30 | null | null | null | null | ['mathematical-proofs'] | ['miscellaneous'] | [-3.81596549e-03 5.63914418e-01 -1.29142299e-01 -5.75409770e-01
3.75262275e-02 -2.29478702e-01 5.70336521e-01 3.79128635e-01
-3.91864896e-01 5.23404837e-01 4.03709292e-01 -5.65300107e-01
-5.44606864e-01 -9.34482992e-01 -1.08716726e+00 -4.54178810e-01
-3.35128248e-01 3.75289381e-01 1.90527365e-01 3.00455898e-01
2.99308360e-01 4.56215978e-01 -1.86123383e+00 3.31700981e-01
6.76803052e-01 2.04057515e-01 3.17517638e-01 9.97920990e-01
-2.98408568e-01 1.12964821e+00 -7.72798732e-02 -7.79605627e-01
-5.68847418e-01 -3.45688552e-01 -5.20212114e-01 -9.91030157e-01
1.87282130e-01 -8.85247111e-01 -4.04582769e-01 8.47346306e-01
2.86139846e-01 1.29138887e-01 1.09046090e+00 -1.12757134e+00
-1.00240016e+00 1.57343590e+00 -1.85599960e-02 5.52588344e-01
7.67066240e-01 -3.57880175e-01 6.54812157e-01 -7.18545258e-01
-9.77220237e-02 1.63503087e+00 8.63479495e-01 1.12872112e+00
-1.25818801e+00 -6.48169935e-01 1.73281208e-01 4.97572839e-01
-1.17008352e+00 -1.25713527e-01 5.45116961e-01 -2.89585710e-01
9.21298683e-01 -1.88196316e-01 1.22194016e+00 1.31180191e+00
2.25638360e-01 1.03833663e+00 9.95671391e-01 -6.38378859e-01
5.77931166e-01 5.49437925e-02 4.97694135e-01 7.24040091e-01
5.89669168e-01 8.22060168e-01 -1.08749330e+00 -1.95629850e-01
9.39252853e-01 9.73355304e-03 2.06687480e-01 -1.68730810e-01
-8.06531250e-01 6.68771565e-01 -1.55973583e-01 1.79558918e-01
-4.84842092e-01 7.78859794e-01 -1.38738528e-01 -2.74243683e-01
-7.32096806e-02 -3.10580656e-02 -6.72018349e-01 -4.48819309e-01
-8.10610473e-01 4.21701580e-01 1.18712139e+00 7.75199234e-01
2.40803510e-01 1.20150775e-01 -1.22247068e-02 5.80185294e-01
1.06651890e+00 8.63095522e-01 1.92571610e-01 -1.13403273e+00
-1.56630665e-01 -4.47638154e-01 -1.20036295e-02 -6.93613768e-01
-4.37529653e-01 -4.42043878e-03 -2.44362622e-01 1.48950815e-01
7.37336218e-01 -1.68525875e-01 -4.69969004e-01 2.26487374e+00
1.53461397e-01 4.88266945e-01 1.85064524e-01 -1.70268081e-02
1.03507745e+00 6.65204108e-01 6.59408152e-01 -2.40935534e-01
1.26839507e+00 3.14452350e-02 -7.88922012e-01 -7.61597529e-02
-7.58118415e-03 -2.46398032e-01 7.86364138e-01 7.94018626e-01
-1.50644743e+00 -2.54513413e-01 -8.37239146e-01 3.73213410e-01
-2.79399753e-01 -4.84520078e-01 1.36455238e+00 1.20232642e+00
-1.28986311e+00 9.42571402e-01 -1.22367775e+00 -4.50239807e-01
7.73921728e-01 2.13230610e-01 1.29524440e-01 -3.15108038e-02
-1.20727968e+00 1.29578722e+00 7.18536615e-01 -2.23379910e-01
-1.10080254e+00 -9.86597180e-01 -9.48362648e-01 2.68163055e-01
-4.57259506e-01 -6.17200255e-01 1.78021955e+00 -3.73142362e-01
-1.54128051e+00 6.92890704e-01 -4.07845855e-01 -6.20108128e-01
-2.22082213e-01 -5.16439140e-01 -3.87654573e-01 4.93256569e-01
-2.78258175e-01 9.22581434e-01 5.08177578e-01 -1.17915738e+00
-4.47985232e-01 -7.10683540e-02 -1.37509450e-01 -1.01513341e-01
-3.20791006e-02 3.47735077e-01 2.52782941e-01 -5.07784367e-01
3.52384627e-01 -1.92966387e-01 1.12618104e-01 -1.02511272e-01
1.60946786e-01 -6.93095267e-01 -3.09676051e-01 -4.69825745e-01
9.19707000e-01 -1.72277629e+00 -5.59072495e-01 1.25345439e-01
2.45062381e-01 -1.04501799e-01 -1.52352005e-02 3.37068796e-01
-4.73618627e-01 -3.49790789e-02 1.06419630e-01 -5.82006201e-03
4.60911065e-01 3.66292864e-01 -4.71327573e-01 3.12871248e-01
2.14177191e-01 9.27436173e-01 -1.29757285e+00 -5.49933195e-01
2.75230020e-01 8.38719904e-01 -8.10111821e-01 3.37745070e-01
-2.21011221e-01 -3.97809893e-02 -1.48200035e-01 3.12632561e-01
5.77275038e-01 1.31174073e-01 5.43648340e-02 3.99758458e-01
7.55588189e-05 7.66029596e-01 -1.02081740e+00 1.29945326e+00
-2.01535717e-01 6.75931811e-01 -6.75353169e-01 -7.48212755e-01
4.88348961e-01 6.07012033e-01 -4.24295485e-01 -2.47266829e-01
4.10163194e-01 -1.60888836e-01 2.28092253e-01 -6.52455807e-01
-2.13875964e-01 -3.38333249e-01 4.54927862e-01 9.48748052e-01
4.70859170e-01 -4.69640404e-01 6.81399852e-02 2.44673312e-01
8.69401693e-01 4.81313854e-01 4.08951789e-01 -4.37022001e-01
8.28799829e-02 -7.25382149e-01 2.34293088e-01 1.37953210e+00
-1.45183861e-01 2.56532848e-01 4.81667638e-01 -3.68621171e-01
-6.81511760e-01 -2.19906330e+00 -2.55706459e-01 1.41455841e+00
-4.56725955e-01 -1.61944830e-03 -1.24689841e+00 -2.32463442e-02
-2.60369629e-01 1.81217575e+00 -6.66301847e-01 -3.02073121e-01
-1.51285946e-01 -6.07864797e-01 6.76812410e-01 9.35904086e-01
-1.84454411e-01 -1.57005489e+00 -8.16913784e-01 3.45948577e-01
2.77118057e-01 -5.31007886e-01 1.24581575e-01 4.13997114e-01
-1.06374395e+00 -7.36836553e-01 -4.03596252e-01 -8.05765152e-01
6.47312582e-01 -2.23623440e-01 1.07892036e+00 8.73656273e-02
-9.05581266e-02 1.07417297e+00 -1.90342143e-02 -1.05312192e+00
-6.27034605e-01 -5.59973717e-01 3.55723768e-01 -8.84007692e-01
9.46455956e-01 -1.49663162e+00 -3.62432003e-01 -4.30632591e-01
-6.19194746e-01 1.78840563e-01 2.81452596e-01 5.02983809e-01
-2.55793557e-02 1.15287028e-01 6.41540766e-01 -5.83617926e-01
4.21150535e-01 -6.19161785e-01 -7.75770187e-01 3.02580088e-01
-6.12827182e-01 5.41634977e-01 2.12125033e-01 -8.64165902e-01
-1.53833604e+00 -1.07908636e-01 -2.16703147e-01 1.71654195e-01
-7.74369001e-01 1.73504606e-01 -2.23659556e-02 4.49595660e-01
6.20762408e-01 4.02207017e-01 -2.62905419e-01 -3.14687192e-01
7.45002806e-01 -6.98931217e-02 9.08896148e-01 -1.21985900e+00
7.89940596e-01 1.16780311e-01 -2.23740071e-01 -3.35836947e-01
-1.18834341e+00 3.45640540e-01 -5.07805109e-01 -3.36228728e-01
9.47896004e-01 -6.94918871e-01 -1.35677791e+00 3.29866946e-01
-1.62126458e+00 -4.44323003e-01 -3.59917581e-01 1.06161809e+00
-9.10295188e-01 -9.40444767e-02 -7.16377020e-01 -1.33633339e+00
-2.01086067e-02 -3.56115460e-01 2.97013879e-01 7.70240664e-01
-7.81107247e-01 -1.33316505e+00 2.37257212e-01 -1.35524377e-01
2.34023675e-01 -2.30481550e-01 1.17398071e+00 -9.79317546e-01
-4.00388241e-01 1.74343407e-01 -1.14572205e-01 1.33294731e-01
-3.69712412e-01 3.52160543e-01 -1.21940720e+00 6.42626107e-01
2.44373634e-01 -2.01739177e-01 7.30665445e-01 5.82520127e-01
1.27336323e+00 -2.30074823e-01 -1.55003354e-01 3.63057196e-01
1.14505076e+00 4.29237992e-01 4.84763891e-01 -3.40988785e-01
3.87285769e-01 8.74786258e-01 -1.66101396e-01 4.64150101e-01
9.50284541e-01 -5.26209414e-01 8.42808187e-02 7.80074239e-01
-9.59081128e-02 -9.41121221e-01 4.14164126e-01 8.73436987e-01
-2.57715583e-01 9.40126330e-02 -1.09347594e+00 7.89358616e-01
-1.51615417e+00 -1.66511595e+00 3.71561125e-02 1.98251808e+00
1.62373817e+00 3.28984916e-01 -2.05748826e-01 2.85348952e-01
7.77074337e-01 -2.70282090e-01 -2.88537085e-01 -8.25443089e-01
1.28770888e-01 9.43808973e-01 3.22247297e-02 5.65936744e-01
-4.30570334e-01 7.20828414e-01 8.70414925e+00 6.58580542e-01
-3.42492647e-02 7.04107359e-02 4.41312820e-01 1.02244556e-01
-8.58644605e-01 -2.43945763e-01 -9.39661443e-01 1.52295932e-01
1.67569053e+00 -1.58062324e-01 8.33738983e-01 8.39174032e-01
-2.04624869e-02 -5.12380719e-01 -1.86414087e+00 1.02539396e+00
9.98944938e-02 -1.25784826e+00 -9.42714512e-02 -6.44939303e-01
5.30118465e-01 -1.83110029e-01 6.48085997e-02 2.74278045e-01
9.93497014e-01 -1.21630800e+00 1.22669888e+00 9.91680384e-01
3.91717911e-01 -8.07623744e-01 2.18350619e-01 5.60340881e-01
-6.45206690e-01 1.77877158e-01 -3.41930985e-01 -6.88860536e-01
-1.35000139e-01 5.20283043e-01 -7.61715531e-01 -6.41616225e-01
7.80062973e-01 2.10923985e-01 -2.33912408e-01 1.14500856e+00
-1.09635389e+00 1.13640356e+00 -4.77783948e-01 -8.62789512e-01
-4.81442869e-01 3.22352171e-01 2.02978984e-01 1.41327727e+00
2.04120457e-01 7.51087725e-01 -9.23711777e-01 1.47793269e+00
4.44156796e-01 -3.12001020e-01 -5.83904862e-01 -8.40326920e-02
8.84129941e-01 6.29804254e-01 -8.64018977e-01 -3.77865791e-01
-2.63696551e-01 3.47068548e-01 3.22733402e-01 1.89884633e-01
-8.22311163e-01 -3.91941667e-02 4.55540866e-01 -3.62127274e-01
4.45112437e-01 -2.35339954e-01 -5.61153293e-01 -7.48407006e-01
-7.62306929e-01 -3.40983301e-01 -2.43935995e-02 -1.13591480e+00
-1.39664376e+00 -2.18869776e-01 8.27496409e-01 1.01906108e-03
-4.60734218e-01 -6.72231317e-01 -9.10219193e-01 6.52766645e-01
-9.75747406e-01 -5.65085948e-01 2.57829815e-01 3.06948394e-01
7.24807531e-02 1.50834829e-01 1.12099695e+00 -1.46098942e-01
-2.13162214e-01 6.23567820e-01 -2.61788487e-01 -2.59470530e-02
-2.64133930e-01 -1.58254409e+00 5.54341614e-01 5.80211401e-01
3.69477004e-01 1.14037383e+00 9.70575869e-01 -5.37827969e-01
-1.14440179e+00 -3.01585704e-01 1.22531605e+00 -7.85561800e-01
7.05859721e-01 -4.01068330e-01 -8.47620666e-01 5.48138320e-01
4.19612139e-01 -3.49465728e-01 9.54029500e-01 2.06626385e-01
-7.06656694e-01 7.63370097e-02 -1.43274260e+00 9.54414904e-01
1.28837991e+00 -7.35940933e-01 -1.67464447e+00 1.92048490e-01
1.05696929e+00 -2.79938638e-01 -6.28887653e-01 1.53958231e-01
1.11553180e+00 -1.21099436e+00 1.02328622e+00 -3.63941252e-01
4.50398564e-01 -5.29071093e-02 -3.78188342e-02 -9.25901651e-01
-1.52774438e-01 -7.35245168e-01 -5.85920632e-01 1.39664137e+00
4.00480479e-01 -8.26297998e-01 5.42480886e-01 1.00725317e+00
2.39772484e-01 -5.87493300e-01 -7.44153857e-01 -3.55070233e-01
4.22217965e-01 -1.41590357e+00 9.59196627e-01 3.16893697e-01
1.48050413e-01 -1.86483026e-01 4.76110786e-01 3.15134168e-01
1.11451554e+00 -6.55313551e-01 -1.93745986e-01 -1.35431814e+00
-2.42089719e-01 -5.45938790e-01 -4.83315066e-02 -7.38112211e-01
2.84124732e-01 -6.49026394e-01 4.26552594e-01 -1.50875795e+00
3.73974204e-01 1.52923232e-02 -4.91873533e-01 3.31506640e-01
-1.61015004e-01 -3.14461648e-01 -2.00523168e-01 -7.76413083e-01
-3.97187829e-01 2.62630403e-01 4.38478947e-01 4.89055008e-01
1.82262883e-01 2.48199761e-01 -9.38946187e-01 1.48537827e+00
9.65387642e-01 -8.37361634e-01 -6.59308672e-01 -2.21341014e-01
8.77293766e-01 1.50987133e-01 6.86578631e-01 -9.61865127e-01
3.16960514e-01 -3.29666078e-01 8.99036288e-01 -1.24665153e+00
-7.30284303e-02 -5.07034421e-01 -2.04363406e-01 5.44154704e-01
-5.44226944e-01 -7.73204342e-02 4.65951771e-01 2.94469297e-01
5.06776035e-01 -6.93274021e-01 6.49723530e-01 -2.96965763e-02
-3.69497299e-01 -5.34894317e-02 -1.12190008e+00 2.27137297e-01
4.19318587e-01 -3.12663093e-02 -1.85799062e-01 -3.00297320e-01
-9.10784066e-01 -2.52677072e-02 -1.62868440e-01 9.79655050e-03
9.95242238e-01 -1.17573273e+00 -4.66233253e-01 5.36953807e-02
-3.76936555e-01 -1.73828885e-01 6.65800795e-02 3.32471430e-01
-6.86664641e-01 4.30553108e-01 3.45370695e-02 -2.62474000e-01
-6.30113423e-01 5.22014916e-01 1.10918626e-01 4.90720421e-01
-3.68321151e-01 1.52386987e+00 1.86744273e-01 -3.21918011e-01
7.48227417e-01 -7.66682327e-01 -4.81310517e-01 -9.75480005e-02
1.11858284e+00 4.02024627e-01 -7.65730739e-01 9.50189754e-02
-2.44352549e-01 3.72336030e-01 2.30139568e-01 -4.84570086e-01
1.41972411e+00 -1.06662497e-01 -3.73445451e-01 1.07445824e+00
5.41793585e-01 -1.04596622e-01 -1.19055736e+00 -8.35568905e-02
2.09824830e-01 1.84441641e-01 -4.38553452e-01 -8.14071774e-01
-3.81238937e-01 1.06326723e+00 5.26651084e-01 9.13004801e-02
8.71056139e-01 4.01290536e-01 3.39102626e-01 6.58370852e-01
3.34527373e-01 -7.60074377e-01 -1.42061576e-01 4.73218173e-01
6.59496427e-01 -7.37217963e-01 7.34536424e-02 -3.94200087e-02
-6.47232533e-02 1.31263971e+00 4.65193868e-01 -2.99456120e-01
1.01297259e+00 3.42832923e-01 -4.79040295e-01 -1.45651530e-02
-1.13974714e+00 6.90609124e-03 3.59836221e-01 1.15398574e+00
5.17290652e-01 3.61153394e-01 -2.60168791e-01 1.33733702e+00
-8.26814294e-01 1.68113992e-01 5.30079126e-01 5.91159880e-01
-7.57146418e-01 -8.54667962e-01 -5.96844256e-01 2.85310060e-01
-5.71733713e-01 -7.77567565e-01 3.37892473e-01 4.82628018e-01
4.41038430e-01 9.58391428e-01 9.25914571e-03 5.68602867e-02
-2.54920840e-01 4.82589573e-01 1.23334777e+00 -8.28641534e-01
-2.18352005e-01 -5.51892519e-01 -6.74253330e-02 -4.92022634e-01
-4.37726825e-01 -1.27347815e+00 -1.83402300e+00 -3.56662899e-01
-1.70886219e-01 9.35961828e-02 7.23586440e-01 1.16123390e+00
-4.79119927e-01 4.08523917e-01 1.03123970e-01 -6.37472868e-01
-3.25693727e-01 -6.99826539e-01 -6.29349113e-01 -5.56180298e-01
2.80473173e-01 -5.17416656e-01 -7.24430799e-01 2.57320046e-01] | [8.81006908416748, 6.858198165893555] |
86eba407-ae32-4936-a4e6-7434684548b4 | nemto-neural-environment-matting-for-novel | 2303.11963 | null | https://arxiv.org/abs/2303.11963v1 | https://arxiv.org/pdf/2303.11963v1.pdf | NEMTO: Neural Environment Matting for Novel View and Relighting Synthesis of Transparent Objects | We propose NEMTO, the first end-to-end neural rendering pipeline to model 3D transparent objects with complex geometry and unknown indices of refraction. Commonly used appearance modeling such as the Disney BSDF model cannot accurately address this challenging problem due to the complex light paths bending through refractions and the strong dependency of surface appearance on illumination. With 2D images of the transparent object as input, our method is capable of high-quality novel view and relighting synthesis. We leverage implicit Signed Distance Functions (SDF) to model the object geometry and propose a refraction-aware ray bending network to model the effects of light refraction within the object. Our ray bending network is more tolerant to geometric inaccuracies than traditional physically-based methods for rendering transparent objects. We provide extensive evaluations on both synthetic and real-world datasets to demonstrate our high-quality synthesis and the applicability of our method. | ['Sabine Süsstrunk', 'Tong Zhang', 'Dongqing Wang'] | 2023-03-21 | null | null | null | null | ['transparent-objects', 'image-matting'] | ['computer-vision', 'computer-vision'] | [ 3.60649586e-01 7.23490119e-02 7.80161440e-01 -4.54114228e-01
-2.94472009e-01 -5.83201230e-01 6.78376377e-01 -4.76843417e-01
2.21265763e-01 4.71582323e-01 -1.59274098e-02 -3.40176404e-01
3.35727066e-01 -9.13943470e-01 -9.13050830e-01 -2.84224302e-01
-9.10311639e-02 5.39171934e-01 5.20634234e-01 -1.30615816e-01
2.62925830e-02 9.90795910e-01 -1.61534655e+00 4.31008279e-01
9.22088981e-01 9.62208509e-01 -1.89839542e-01 8.77531111e-01
-2.91148514e-01 4.96752530e-01 -5.34112751e-01 -3.99815738e-01
6.07727051e-01 -1.46396786e-01 -3.68787378e-01 1.07565232e-01
1.38769281e+00 -9.91997719e-01 -1.73531413e-01 5.25053322e-01
4.96264875e-01 7.81831890e-02 7.13833928e-01 -1.06869042e+00
-6.43206358e-01 -1.81686059e-01 -6.69856250e-01 -4.15134460e-01
3.18607956e-01 2.55882949e-01 6.56645179e-01 -1.12697053e+00
8.17584991e-01 1.55111778e+00 8.95218134e-01 8.40219021e-01
-1.47368801e+00 -6.86818659e-01 4.92165297e-01 -4.43436444e-01
-1.04994035e+00 -5.89992225e-01 8.68808329e-01 -4.44894463e-01
8.58448207e-01 6.31281853e-01 1.08172965e+00 9.16828752e-01
1.75013751e-01 3.49848241e-01 1.15251577e+00 -9.33468863e-02
3.08668107e-01 -1.87503010e-01 -1.36180162e-01 9.17487264e-01
1.26469240e-01 2.20729083e-01 -6.34654522e-01 -3.24200451e-01
1.50214207e+00 -3.06729704e-01 -3.64149123e-01 -7.26829886e-01
-1.04014277e+00 3.00513238e-01 6.96403444e-01 -5.84482908e-01
3.95111851e-02 7.67495334e-01 -1.22226663e-01 -1.40815089e-02
1.10951996e+00 1.08938225e-01 -5.63962996e-01 5.69105633e-02
-7.26905107e-01 5.12273431e-01 8.03425014e-01 1.01975787e+00
4.77948546e-01 3.75965953e-01 7.95040801e-02 7.37644970e-01
7.13961363e-01 6.25429094e-01 -5.73028326e-01 -1.53036630e+00
1.23368606e-01 4.59747046e-01 1.40992299e-01 -7.81830668e-01
-2.62056589e-01 -4.78341818e-01 -5.29767275e-01 1.39935064e+00
2.81777322e-01 1.13853574e-01 -1.33415174e+00 1.45926893e+00
8.40798974e-01 4.62427914e-01 -3.08271170e-01 1.00312138e+00
1.24964845e+00 5.46851933e-01 -4.61716413e-01 2.05361694e-01
9.69853938e-01 -8.52473676e-01 -2.99481332e-01 -1.87196378e-02
1.71994597e-01 -7.81582117e-01 1.18874407e+00 4.83683974e-01
-1.58969557e+00 5.54746762e-03 -9.00900245e-01 -5.70605814e-01
2.06542462e-01 -3.83386105e-01 8.08171809e-01 4.90155637e-01
-1.33734262e+00 6.48412883e-01 -1.01672924e+00 2.11141869e-01
7.50614762e-01 4.17701811e-01 -1.69020798e-02 -9.09749866e-02
-3.89952838e-01 4.22501624e-01 -6.04089677e-01 1.74128354e-01
-8.51046681e-01 -1.47341204e+00 -6.36603236e-01 -1.43615663e-01
-5.97195849e-02 -1.16932857e+00 1.23837376e+00 -1.07559156e+00
-1.85994136e+00 8.95437121e-01 -3.21141966e-02 1.60679847e-01
9.32353020e-01 -1.90198034e-01 1.21742666e-01 1.80372760e-01
-5.96963704e-01 4.74679530e-01 7.06384301e-01 -2.09991145e+00
1.37392297e-01 -3.70728046e-01 2.78556615e-01 4.66985345e-01
9.87251624e-02 -8.20343718e-02 -4.00872469e-01 -7.30413854e-01
4.32217062e-01 -7.44721234e-01 7.95664731e-04 1.39916587e+00
-4.80323672e-01 5.10865867e-01 1.04004014e+00 -7.15074897e-01
2.89045185e-01 -1.91535890e+00 -9.64992493e-02 3.40001673e-01
5.18010080e-01 8.52900669e-02 -2.50750065e-01 8.31900537e-02
8.37602690e-02 -3.28415744e-02 -4.15586591e-01 -9.44269121e-01
-2.67861597e-02 7.20569268e-02 -3.17110270e-01 4.07906771e-01
-1.26370415e-01 7.22556531e-01 -7.03900754e-01 -2.67095387e-01
1.73954383e-01 1.40256333e+00 -7.96723783e-01 2.03463823e-01
-7.19750106e-01 6.55669987e-01 -1.51524872e-01 6.49351895e-01
1.07977760e+00 -2.05317348e-01 -1.31107643e-01 -3.36247265e-01
-1.79499343e-01 4.28966641e-01 -9.88101065e-01 1.72468543e+00
-7.17174947e-01 6.98565125e-01 5.25119901e-01 3.77653122e-01
6.20171726e-01 1.83462054e-01 3.31911892e-01 -7.02152014e-01
7.80184846e-03 1.13884367e-01 -1.89194232e-01 3.19800600e-02
2.46237412e-01 -6.46303594e-02 8.32885742e-01 3.55600893e-01
-5.88938892e-01 -8.26456487e-01 -5.96227229e-01 2.23934576e-01
1.04099667e+00 8.00442994e-01 -6.07112706e-01 -2.13529855e-01
-2.63199747e-01 -3.80747736e-01 4.33954716e-01 6.65944889e-02
5.65185666e-01 1.28613186e+00 1.63554013e-01 -5.87242007e-01
-1.12000906e+00 -1.67962539e+00 -3.81649882e-01 6.95793033e-01
2.09683642e-01 -8.46213624e-02 -7.27138937e-01 -2.53619611e-01
2.29766116e-01 6.87884748e-01 -6.46795452e-01 1.50476620e-01
-5.45328617e-01 -4.78415132e-01 2.08310857e-01 3.30542624e-01
2.79045671e-01 -7.29252160e-01 -7.78174639e-01 2.31611077e-02
2.62459159e-01 -1.09693146e+00 -3.62840652e-01 -4.61651504e-01
-1.08439851e+00 -9.26817477e-01 -7.13532031e-01 -3.80959421e-01
9.31989193e-01 3.01385820e-01 1.58966136e+00 5.18161595e-01
-5.25010884e-01 4.91828501e-01 2.26470366e-01 -3.80128235e-01
-5.71407020e-01 -6.54091537e-01 -2.77214974e-01 -5.76254390e-02
-7.24794626e-01 -1.04936361e+00 -1.08200383e+00 4.52482164e-01
-8.95509362e-01 8.61557066e-01 -1.25945732e-01 1.31862685e-01
5.66362500e-01 -6.32446289e-01 -1.80209264e-01 -8.73490810e-01
1.83232263e-01 9.67987180e-02 -1.03079057e+00 1.19113646e-01
-2.30182901e-01 -2.83582240e-01 2.88094193e-01 -4.27930474e-01
-1.40454137e+00 -1.62716702e-01 -3.53302024e-02 -5.39052248e-01
1.46402985e-01 -1.28155127e-01 -1.47704482e-01 -6.18532360e-01
5.43776274e-01 -2.67967522e-01 -1.14317313e-01 -6.40956163e-01
1.61828190e-01 2.31377482e-02 3.42542082e-01 -7.31907189e-01
1.13369167e+00 1.00175905e+00 5.22613764e-01 -8.96087706e-01
-5.31222522e-01 3.29580009e-01 -5.22054493e-01 -3.65736246e-01
5.24487615e-01 -8.12294364e-01 -1.11013234e+00 8.64741802e-01
-1.51247370e+00 -1.04573715e+00 -2.90344119e-01 8.74592364e-02
-5.36143064e-01 1.97182462e-01 -7.14375854e-01 -9.22969818e-01
-4.84606385e-01 -9.99917805e-01 1.33944130e+00 -4.64823842e-02
6.46238700e-02 -1.06934977e+00 1.22648612e-01 2.84776092e-01
5.02641201e-01 7.80101061e-01 9.84743595e-01 6.88757300e-01
-1.20174110e+00 3.48297685e-01 -5.72371483e-01 1.02165908e-01
2.01839939e-01 7.99216688e-01 -1.39340758e+00 -1.70908779e-01
-3.34064722e-01 -1.97663426e-01 5.63635051e-01 3.99713427e-01
1.23884237e+00 -2.61278749e-01 -1.05751529e-01 1.39936316e+00
1.47585034e+00 -2.02349737e-01 6.80418670e-01 -2.48113245e-01
1.33643711e+00 5.80281496e-01 3.84840034e-02 4.45994556e-01
5.81165016e-01 8.13564360e-01 1.04913890e+00 -6.63663924e-01
-7.34381139e-01 -9.03468132e-02 8.61105770e-02 6.54526949e-01
-3.98621887e-01 -4.37556356e-01 -7.65030444e-01 -4.34370562e-02
-1.22914374e+00 -6.07585549e-01 -7.10016429e-01 2.43001080e+00
9.59649980e-01 9.64868441e-02 -1.00337461e-01 -2.98474759e-01
-5.72074056e-02 -7.92715326e-02 -6.91510201e-01 -5.64098060e-01
-1.42506018e-01 2.76619434e-01 3.40063632e-01 9.25998092e-01
-3.68102938e-01 7.39868224e-01 6.34826994e+00 2.23682940e-01
-1.20357859e+00 -9.07433927e-02 5.75034797e-01 -5.02059877e-01
-1.17993891e+00 -1.01745799e-01 -5.00234783e-01 1.50031209e-01
3.92362356e-01 3.60512793e-01 7.17357278e-01 1.59196809e-01
2.98559695e-01 -3.68410237e-02 -1.00965214e+00 7.31491029e-01
5.11330180e-02 -1.37578046e+00 3.23609203e-01 3.27213138e-01
9.04504180e-01 2.43185848e-01 2.38731518e-01 -5.60121000e-01
6.98002875e-01 -9.83488560e-01 1.15037429e+00 6.92432761e-01
1.15823257e+00 -4.73378450e-01 -2.66674757e-01 9.44439992e-02
-1.00928104e+00 5.88534415e-01 -1.97823271e-02 1.47883862e-01
3.04671347e-01 7.54114151e-01 -7.42831171e-01 9.74771902e-02
5.70664644e-01 4.75642920e-01 -4.28075254e-01 1.19677150e+00
3.37158889e-02 5.02722561e-01 -7.27125108e-01 4.42529380e-01
-1.42823040e-01 -3.17221969e-01 6.23285532e-01 7.59213746e-01
1.46700189e-01 1.36362806e-01 -1.99149504e-01 1.30814636e+00
-1.72532812e-01 -1.41007170e-01 -4.02714908e-01 3.98924261e-01
1.21253133e-01 9.39990580e-01 -8.25444460e-01 -6.04583584e-02
-3.32261890e-01 1.11315989e+00 3.93521518e-01 7.44911909e-01
-8.30964506e-01 2.15825796e-01 1.04386926e+00 6.04417145e-01
-3.07247862e-02 -5.92981100e-01 -5.47304690e-01 -1.12349308e+00
2.71433443e-01 -5.71099341e-01 -4.74460244e-01 -1.29003525e+00
-1.12698829e+00 5.83774090e-01 -8.60168040e-02 -1.15799880e+00
4.94371474e-01 -7.42844224e-01 -5.39665878e-01 9.88479972e-01
-1.70562267e+00 -1.61867332e+00 -5.07327795e-01 4.16705668e-01
3.57808083e-01 5.50421357e-01 8.85669410e-01 1.42646283e-01
-7.27912039e-02 2.83096135e-01 7.17009231e-02 -2.58269072e-01
4.75234002e-01 -1.21175516e+00 1.02286005e+00 5.58558285e-01
-1.04376234e-01 6.03389382e-01 7.30455279e-01 -7.63636589e-01
-1.59312177e+00 -8.06152582e-01 7.57259429e-02 -6.94406092e-01
6.07951507e-02 -8.73790920e-01 -1.01885438e+00 6.54240727e-01
-2.34305318e-02 4.93219525e-01 4.61363733e-01 -1.94912106e-01
-8.17074716e-01 -6.88641742e-02 -1.30662525e+00 9.74049985e-01
1.60576725e+00 -2.20277920e-01 1.56737536e-01 3.15159559e-01
8.49656940e-01 -1.03676343e+00 -6.54637575e-01 4.72787261e-01
1.11720371e+00 -1.28930545e+00 1.36949277e+00 -1.74824014e-01
4.18877035e-01 -2.58452743e-01 -1.50152426e-02 -1.37346613e+00
-8.72771442e-02 -8.08891952e-01 -2.84496725e-01 9.96321797e-01
1.98330224e-01 -6.09977782e-01 8.69528830e-01 1.02261305e+00
-4.85902578e-01 -8.13668191e-01 -8.00103545e-01 -4.75618154e-01
4.56916876e-02 -4.75774854e-01 7.28720367e-01 8.81497204e-01
-8.99142146e-01 -1.87644616e-01 -1.36779966e-02 3.80224973e-01
1.17363286e+00 3.04101259e-01 9.26118672e-01 -1.65171790e+00
-4.06395227e-01 -3.20287645e-01 -7.66859129e-02 -1.05692565e+00
-4.65147123e-02 -6.05427563e-01 6.97739571e-02 -1.85441089e+00
-1.86759338e-01 -1.01886070e+00 4.75725621e-01 2.23283440e-01
2.18635470e-01 6.95019066e-01 2.97913738e-02 -1.02920309e-01
5.86134903e-02 6.41762495e-01 1.87158275e+00 1.64775401e-01
-2.56946445e-01 1.64892413e-02 -2.18537286e-01 1.12731528e+00
3.81848156e-01 -1.66185781e-01 -6.57418668e-01 -1.32131135e+00
7.20073700e-01 -1.17285967e-01 8.74233663e-01 -7.52099335e-01
-2.89408594e-01 -2.97654957e-01 6.37758017e-01 -5.76077580e-01
1.10790491e+00 -1.14206183e+00 4.25115734e-01 -8.28880724e-03
-2.29116797e-01 -1.52380373e-02 3.93550217e-01 3.74078214e-01
7.08365917e-01 3.90779555e-01 9.27945793e-01 -1.55869722e-01
1.18370719e-01 7.46142685e-01 6.42990172e-02 1.44765794e-01
5.66191196e-01 -5.79568505e-01 -3.81052792e-01 -3.58160406e-01
-4.07446027e-01 -1.11094333e-01 1.31205964e+00 1.90850884e-01
1.04354000e+00 -1.23410761e+00 -7.27866232e-01 3.75709057e-01
-6.27937913e-02 6.28606260e-01 1.14167944e-01 2.54034191e-01
-1.29928553e+00 -6.36944771e-01 -1.16070500e-02 -5.85512996e-01
-1.59151876e+00 -1.57816991e-01 7.08238482e-01 4.08439428e-01
-9.95772541e-01 1.12031925e+00 7.69497335e-01 -5.97857833e-01
2.99792886e-01 -6.91885054e-01 5.45566857e-01 -6.61176503e-01
3.69953930e-01 4.30319667e-01 2.17752963e-01 -6.11613274e-01
-1.83661565e-01 8.96369219e-01 2.44262204e-01 -3.63431364e-01
1.38554752e+00 1.10061085e-02 -1.56154394e-01 4.89618152e-01
1.01309526e+00 2.98596084e-01 -1.88557541e+00 -3.35527286e-02
-9.82511640e-01 -7.70824909e-01 2.70780265e-01 -1.07602799e+00
-1.26619506e+00 1.04290771e+00 3.22674483e-01 -1.72749266e-01
6.80303156e-01 -4.30583596e-01 9.06738102e-01 4.04773280e-02
5.28544843e-01 -5.45785725e-01 -8.50663781e-02 3.75518084e-01
1.21465635e+00 -6.95075452e-01 2.77895927e-01 -1.07509089e+00
7.65381157e-02 1.01917970e+00 7.89991736e-01 -8.32525194e-02
8.08909476e-01 8.38004351e-01 3.63517553e-01 -3.55877250e-01
-6.91196859e-01 4.99147028e-01 6.54283345e-01 6.59529626e-01
4.97502327e-01 -1.83914766e-01 4.55787271e-01 -5.13754249e-01
-2.27897063e-01 -1.28279716e-01 4.61605489e-01 9.33332384e-01
-5.27330711e-02 -8.65657210e-01 -3.49516958e-01 2.96286672e-01
-1.57739550e-01 -1.73967272e-01 -5.81583261e-01 3.03837448e-01
-7.79188126e-02 4.10149515e-01 4.08645421e-01 2.09020153e-01
3.44014943e-01 -4.88919020e-01 1.07730138e+00 -7.70316422e-01
-6.31905258e-01 1.09620139e-01 9.79690850e-02 -8.73224497e-01
-3.50006402e-01 -4.61887002e-01 -1.59576380e+00 -3.65733832e-01
-1.53059319e-01 -6.59471333e-01 1.08444917e+00 4.24380302e-01
5.76122344e-01 4.89113629e-01 5.81203461e-01 -1.43280780e+00
-2.79121231e-02 -2.97396809e-01 -3.95883441e-01 4.62101251e-01
7.20381558e-01 -6.05099618e-01 -4.43607569e-01 2.85921358e-02] | [9.615654945373535, -3.1530494689941406] |
547602d2-f831-4024-88be-45e560aea83f | training-ensembles-with-inliers-and-outliers | 2307.03741 | null | https://arxiv.org/abs/2307.03741v1 | https://arxiv.org/pdf/2307.03741v1.pdf | Training Ensembles with Inliers and Outliers for Semi-supervised Active Learning | Deep active learning in the presence of outlier examples poses a realistic yet challenging scenario. Acquiring unlabeled data for annotation requires a delicate balance between avoiding outliers to conserve the annotation budget and prioritizing useful inlier examples for effective training. In this work, we present an approach that leverages three highly synergistic components, which are identified as key ingredients: joint classifier training with inliers and outliers, semi-supervised learning through pseudo-labeling, and model ensembling. Our work demonstrates that ensembling significantly enhances the accuracy of pseudo-labeling and improves the quality of data acquisition. By enabling semi-supervision through the joint training process, where outliers are properly handled, we observe a substantial boost in classifier accuracy through the use of all available unlabeled examples. Notably, we reveal that the integration of joint training renders explicit outlier detection unnecessary; a conventional component for acquisition in prior work. The three key components align seamlessly with numerous existing approaches. Through empirical evaluations, we showcase that their combined use leads to a performance increase. Remarkably, despite its simplicity, our proposed approach outperforms all other methods in terms of performance. Code: https://github.com/vladan-stojnic/active-outliers | ['Giorgos Tolias', 'Zakaria Laskar', 'Vladan Stojnić'] | 2023-07-07 | null | null | null | null | ['active-learning', 'outlier-detection', 'active-learning'] | ['methodology', 'methodology', 'natural-language-processing'] | [-4.12409678e-02 1.28852919e-01 -1.14914954e-01 -4.05707508e-01
-1.49126124e+00 -7.40912735e-01 4.49506372e-01 4.49504048e-01
-4.96618956e-01 6.12976074e-01 1.33437544e-01 -2.37395540e-02
-5.53395599e-02 -1.60470486e-01 -8.96507382e-01 -6.05809629e-01
1.90617517e-01 4.00708765e-01 -1.76872276e-02 1.24361791e-01
1.74830839e-01 5.01871467e-01 -1.43605232e+00 -1.31918639e-01
1.03875828e+00 9.84186351e-01 -4.51625466e-01 4.24304157e-01
1.71875641e-01 8.90701950e-01 -3.91733557e-01 -4.85916644e-01
6.87207282e-01 -1.81887224e-01 -5.74206233e-01 4.61249322e-01
6.44442141e-01 -5.01282394e-01 7.70646706e-02 7.69715250e-01
5.19504249e-01 1.51715711e-01 3.84289056e-01 -1.23293936e+00
-1.47609055e-01 3.90178472e-01 -6.25304401e-01 -4.48481366e-02
4.31752093e-02 3.34397316e-01 1.10707283e+00 -1.38560987e+00
2.64911622e-01 6.00927472e-01 1.16419160e+00 2.28666589e-01
-1.29064465e+00 -5.93792856e-01 6.56847134e-02 -1.27260789e-01
-1.39720047e+00 -9.45298731e-01 6.95334196e-01 -6.47968173e-01
6.22497618e-01 2.63539702e-01 5.26260316e-01 9.76749241e-01
-4.96235371e-01 9.00124311e-01 6.27879620e-01 -5.56566775e-01
3.78189236e-01 6.80163503e-02 4.37240273e-01 6.54807448e-01
5.52126288e-01 8.59317556e-02 -7.43703902e-01 -4.34339523e-01
4.27270532e-01 3.04638267e-01 -2.28512034e-01 -8.25777888e-01
-1.09039414e+00 4.35293317e-01 1.69992432e-01 -6.76410943e-02
-2.54519582e-01 2.33153805e-01 4.59077060e-01 9.88191068e-02
6.88972235e-01 5.51256061e-01 -4.41288233e-01 -1.92090377e-01
-1.42779052e+00 -1.02198504e-01 5.73466599e-01 1.00917923e+00
7.99455345e-01 5.37600927e-02 4.05802503e-02 7.22374499e-01
4.32634741e-01 1.83430016e-01 1.86139137e-01 -1.13930368e+00
3.56135547e-01 7.54510283e-01 4.23674554e-01 -6.19065523e-01
-1.41362667e-01 -7.36359119e-01 -4.77662891e-01 2.20463380e-01
7.01688826e-01 -1.17077753e-01 -7.30151653e-01 1.52598524e+00
6.57259941e-01 4.22416359e-01 -3.10024600e-02 6.79734826e-01
3.05159330e-01 -6.33912906e-02 5.97462058e-02 -2.70600915e-01
8.44558656e-01 -1.04008532e+00 -5.29487908e-01 -8.48036632e-02
7.48750269e-01 -7.52296805e-01 1.12219739e+00 5.47415257e-01
-1.08209348e+00 -3.55029583e-01 -1.24195313e+00 -2.17262685e-01
-3.15760635e-02 4.99064386e-01 6.43974066e-01 4.60852474e-01
-8.53136957e-01 7.30455458e-01 -1.26473236e+00 -2.56051898e-01
7.24876165e-01 4.65629101e-01 -4.85070169e-01 -1.29539119e-02
-4.95891780e-01 4.68508750e-01 2.74393529e-01 2.59538174e-01
-8.90230417e-01 -8.65187168e-01 -8.23809922e-01 -4.43996042e-02
6.91967428e-01 -3.88442069e-01 1.43325269e+00 -9.76552784e-01
-1.12860286e+00 7.72148907e-01 -1.07237488e-01 -4.70511049e-01
1.07290757e+00 -8.22790027e-01 1.00825399e-01 5.45878746e-02
1.73909720e-02 5.04150033e-01 7.54091978e-01 -1.33665121e+00
-4.45400506e-01 -2.95604318e-01 -2.82519639e-01 8.38287473e-02
-5.62023044e-01 -2.13736430e-01 -6.50124848e-01 -6.97743654e-01
3.27773541e-01 -9.81275320e-01 -3.47305954e-01 1.49872974e-01
-3.87760818e-01 8.62484053e-02 6.58818126e-01 -5.53022742e-01
1.07818043e+00 -2.50043488e+00 -2.10525304e-01 3.06810975e-01
5.86890399e-01 2.10428357e-01 2.33287483e-01 4.69569802e-01
-1.95314109e-01 5.55271655e-02 -4.08667356e-01 -8.90923679e-01
9.80214253e-02 7.88383782e-02 -3.98960263e-01 7.76022255e-01
3.68777066e-01 9.32037115e-01 -9.61454451e-01 -4.00185734e-01
2.27886096e-01 3.60040963e-01 -5.57975292e-01 1.97087556e-01
-1.33663014e-01 6.16165340e-01 -6.49637729e-02 8.97668481e-01
4.14501756e-01 -4.04200763e-01 5.92587851e-02 1.70618873e-02
3.80549231e-03 1.88851282e-01 -1.41282105e+00 1.75930822e+00
-5.20948730e-02 2.84012824e-01 7.87797868e-02 -7.75163710e-01
8.19918811e-01 3.98013800e-01 7.19778240e-01 -2.19078839e-01
5.87853827e-02 5.76206446e-01 -2.07745954e-01 -1.58889890e-01
5.02079904e-01 2.68476516e-01 1.84113011e-01 5.40745497e-01
2.20048770e-01 6.37480170e-02 2.96412498e-01 4.22649413e-01
1.17886579e+00 4.83255059e-01 5.25672138e-01 -1.57456324e-01
9.49631557e-02 1.60117410e-02 7.97925115e-01 8.47811222e-01
-4.75709319e-01 8.19259465e-01 4.20267224e-01 -2.97726840e-01
-1.14484394e+00 -9.07106340e-01 1.11578451e-02 9.46062505e-01
-8.91661048e-02 -4.81809348e-01 -6.53262258e-01 -8.93900216e-01
6.17616735e-02 5.37188590e-01 -5.11759043e-01 -2.31089424e-02
-5.79087317e-01 -7.04157293e-01 4.87042487e-01 5.54555833e-01
1.43613219e-01 -6.74131036e-01 -5.41205943e-01 -4.57079224e-02
1.14181221e-01 -9.61664557e-01 -4.35772508e-01 5.06580472e-01
-1.15597510e+00 -1.23854542e+00 -4.46263105e-01 -3.83729935e-01
9.46146548e-01 1.70951620e-01 1.10691261e+00 4.92467701e-01
-3.00996959e-01 4.88424569e-01 -4.00694489e-01 -3.55506569e-01
-3.29365015e-01 9.59233120e-02 1.30690768e-01 -1.81896165e-02
2.07150951e-01 -7.27184951e-01 -6.02815032e-01 1.26019299e-01
-8.07661057e-01 -1.20256133e-01 6.02558970e-01 9.04731452e-01
7.90432572e-01 -3.55290443e-01 3.79832953e-01 -1.14834559e+00
-3.19470204e-02 -5.10328889e-01 -7.01654434e-01 1.32610887e-01
-6.35831952e-01 2.14176551e-02 4.82945412e-01 -1.93267837e-01
-8.37487698e-01 5.53239107e-01 -1.70698419e-01 -7.07712591e-01
-1.80296615e-01 1.86644778e-01 -1.79537028e-01 -4.72295955e-02
8.39596868e-01 -1.46070078e-01 -7.56364912e-02 -5.62679529e-01
3.99591863e-01 4.68211651e-01 5.91242075e-01 -6.72159553e-01
1.04907179e+00 6.84439480e-01 -2.46057346e-01 -4.80046362e-01
-1.09634435e+00 -7.51242518e-01 -1.03468692e+00 -1.42984822e-01
3.00894350e-01 -1.07373095e+00 -4.31012839e-01 3.22616518e-01
-7.54064679e-01 -3.57419521e-01 -7.51409531e-01 4.54670489e-01
-3.30074757e-01 6.75040662e-01 -4.98404950e-01 -9.70060766e-01
-3.22396308e-01 -1.12311566e+00 1.20936096e+00 2.39155740e-02
-4.00921434e-01 -8.00442576e-01 8.18216354e-02 5.56197226e-01
1.48111209e-01 4.03644323e-01 2.83631742e-01 -1.24966073e+00
-7.53731847e-01 -4.78412747e-01 3.66324633e-02 3.73208344e-01
4.92130816e-02 1.91067249e-01 -1.37559986e+00 -4.51488107e-01
-5.43447137e-02 -5.41530430e-01 7.64799595e-01 4.50843424e-02
1.01597238e+00 -7.16238990e-02 -4.83798869e-02 6.71676815e-01
1.25373900e+00 -2.83737093e-01 4.32919651e-01 2.43675545e-01
8.69675517e-01 3.76077831e-01 9.61851120e-01 6.04373157e-01
1.45866543e-01 4.95718211e-01 6.00699663e-01 -4.51477975e-01
8.66620913e-02 -1.57795742e-01 1.21192276e-01 7.86509514e-01
9.53328460e-02 8.34165514e-02 -1.15273440e+00 5.27485728e-01
-2.16709590e+00 -7.96047390e-01 -2.25888953e-01 2.52392840e+00
9.20166135e-01 2.19425395e-01 1.43893123e-01 4.53804106e-01
5.03716588e-01 -2.54445076e-01 -7.08660603e-01 2.96515226e-01
1.06266782e-01 2.63892394e-02 5.38905263e-01 6.13256454e-01
-1.23411524e+00 7.41399169e-01 5.66222954e+00 6.64012015e-01
-8.36764395e-01 2.86160588e-01 7.49308348e-01 -4.50661182e-01
-1.28267407e-01 1.78949878e-01 -6.59384608e-01 3.21329355e-01
8.01944017e-01 2.56697088e-01 9.35445949e-02 1.03334558e+00
2.75232255e-01 -1.85864389e-01 -1.39880693e+00 8.50892365e-01
4.23536971e-02 -1.14444852e+00 -3.71206343e-01 7.44388849e-02
6.75839961e-01 1.75963268e-01 -9.04705599e-02 1.43628508e-01
5.06130219e-01 -6.23011351e-01 1.01769888e+00 4.80017394e-01
6.02086067e-01 -5.60262740e-01 7.44040430e-01 3.29247922e-01
-8.78211796e-01 -2.89371815e-02 6.66723102e-02 -3.71950343e-02
4.88570482e-02 1.19065511e+00 -9.09932077e-01 7.04835176e-01
5.48085392e-01 8.01884830e-01 -7.26019919e-01 1.15490174e+00
-2.10960239e-01 9.22850668e-01 -4.49250013e-01 6.35631800e-01
-1.16788141e-01 -2.00597540e-01 4.61354673e-01 1.03070104e+00
1.73348531e-01 -1.72368571e-01 4.33113694e-01 7.26716578e-01
-2.06338689e-01 1.99696034e-01 -4.99798775e-01 1.92522295e-02
7.08736241e-01 1.30534101e+00 -8.94925773e-01 -3.02798778e-01
-3.29488456e-01 8.25851738e-01 6.21133089e-01 1.84122279e-01
-8.96217585e-01 -8.99303108e-02 4.02844489e-01 -8.29438586e-03
1.84501395e-01 -3.42986435e-01 -6.36043131e-01 -1.25431097e+00
4.42380548e-01 -9.79877293e-01 4.26134229e-01 -4.11130995e-01
-1.23991334e+00 2.47181594e-01 -3.64327788e-01 -1.38108981e+00
-5.81425726e-02 -2.41689026e-01 -3.97191018e-01 5.15157163e-01
-1.35651028e+00 -1.24649608e+00 -5.90559423e-01 1.80599242e-01
4.28286493e-01 1.23127021e-01 5.78882933e-01 6.00341380e-01
-9.07280385e-01 8.01560819e-01 1.29330322e-01 1.40853256e-01
9.93308425e-01 -1.49951410e+00 3.95196706e-01 1.30255735e+00
4.25932050e-01 8.76866579e-01 5.33523440e-01 -6.92179739e-01
-1.16249692e+00 -1.02538037e+00 5.52505553e-01 -8.12142670e-01
6.03038132e-01 -5.94425499e-01 -1.01957917e+00 8.88123333e-01
-2.34765038e-01 1.83710366e-01 9.06847417e-01 1.98626414e-01
-3.65942836e-01 -1.32957906e-01 -9.79243338e-01 5.20119131e-01
9.75422442e-01 -3.78903329e-01 -2.74418056e-01 4.60382581e-01
7.25243032e-01 -5.78652024e-01 -8.00080955e-01 4.65932518e-01
4.00375247e-01 -1.20055091e+00 7.61286914e-01 -3.33650917e-01
2.25574926e-01 -5.12760758e-01 -1.19159453e-01 -8.50081742e-01
1.44789517e-01 -8.68491232e-01 -3.55923951e-01 1.58836448e+00
4.37269151e-01 -4.88064945e-01 1.01858175e+00 8.63729298e-01
-4.89206254e-01 -8.71326387e-01 -6.10080719e-01 -6.16910696e-01
-2.58302003e-01 -5.99091351e-01 4.28060353e-01 1.16946077e+00
-1.37990713e-01 9.74702165e-02 -4.14809436e-01 3.66446078e-01
7.30798602e-01 -2.79006332e-01 1.13996160e+00 -1.30051434e+00
-4.56319004e-01 -3.50820720e-02 -2.49412253e-01 -8.42392206e-01
-4.72963676e-02 -8.01269770e-01 2.12523058e-01 -1.20348895e+00
2.13522345e-01 -7.75931001e-01 -3.21257174e-01 8.20685804e-01
-4.47148144e-01 5.29321790e-01 -5.11100851e-02 6.45323157e-01
-9.47301388e-01 3.90092432e-01 5.34820497e-01 3.67058218e-01
-3.49111438e-01 -5.30209094e-02 -7.81299591e-01 8.73572648e-01
7.95987725e-01 -5.79295635e-01 -2.57723987e-01 -4.66029108e-01
2.94385254e-01 -3.87178332e-01 4.23953861e-01 -1.09575641e+00
3.00652236e-01 1.17232248e-01 3.50541145e-01 -5.33083320e-01
3.34248871e-01 -1.01039994e+00 1.74255863e-01 2.74703741e-01
-3.48000586e-01 5.26603404e-03 1.20080195e-01 7.03553677e-01
-1.14071891e-01 -1.78666860e-01 8.12875271e-01 2.21712776e-02
-5.06757140e-01 1.84356064e-01 2.18116224e-01 1.31585792e-01
1.14178538e+00 -3.60755652e-01 -5.59300371e-02 -2.14945391e-01
-7.24399447e-01 1.78158909e-01 7.87598789e-01 -4.52662483e-02
1.60044804e-01 -9.77722168e-01 -4.18582708e-01 1.96973711e-01
2.16431811e-01 3.80812645e-01 4.82730480e-04 1.36893654e+00
-5.67320824e-01 -2.55398601e-02 2.86413521e-01 -8.14595997e-01
-1.13995707e+00 1.83553085e-01 3.01967114e-01 -2.78803051e-01
-4.83790129e-01 8.36935878e-01 -2.10667700e-01 -4.75746363e-01
7.06179619e-01 -2.40652189e-01 3.30769002e-01 1.54446229e-01
2.85566777e-01 4.82379019e-01 5.74229717e-01 -2.72660464e-01
-1.76961467e-01 1.02644369e-01 -2.58558720e-01 7.37025402e-03
1.51463771e+00 -1.20232761e-01 3.51879969e-02 5.99234462e-01
8.26867938e-01 5.26345193e-01 -1.61341131e+00 -3.05894941e-01
4.76318866e-01 -5.09663403e-01 -1.83431029e-01 -7.40524948e-01
-9.06662941e-01 6.56829476e-01 4.13454920e-01 -9.11753532e-03
9.70863461e-01 -9.07067582e-02 6.61127806e-01 5.22945583e-01
2.84446388e-01 -1.07792819e+00 3.34110439e-01 3.16609293e-01
4.34265077e-01 -1.60788620e+00 2.92981148e-01 -5.39960921e-01
-6.14043772e-01 7.14771032e-01 6.26538098e-01 -5.98140396e-02
3.21366042e-01 3.78046423e-01 3.09573501e-01 -2.03478470e-01
-6.18701458e-01 -2.13355765e-01 1.29709810e-01 2.38076374e-01
4.09824938e-01 -3.59988242e-01 1.12318680e-01 6.24117732e-01
1.36589436e-02 -5.69079518e-02 4.01699841e-01 1.23735631e+00
-3.97925586e-01 -1.11013389e+00 -4.56238300e-01 4.74770069e-01
-5.16373098e-01 -1.77416101e-01 -4.93485063e-01 8.92972708e-01
1.21681213e-01 7.99570441e-01 -1.22821964e-01 1.74680427e-02
2.71763146e-01 3.34171653e-01 1.19469069e-01 -8.58663738e-01
-6.50147676e-01 2.87929118e-01 -1.43948281e-02 -6.65185034e-01
-3.18616480e-01 -6.93547845e-01 -1.19718421e+00 -7.78336823e-02
-7.30020225e-01 2.43269011e-01 5.32206178e-01 1.02561712e+00
6.03333056e-01 3.46158922e-01 4.43483949e-01 -9.01184976e-01
-8.33367288e-01 -7.78292179e-01 -4.42164034e-01 5.30311167e-01
5.05230725e-01 -7.15209842e-01 -6.98904693e-01 2.91142583e-01] | [9.105600357055664, 1.6285475492477417] |
e49dc172-142c-425f-b726-5171e405c753 | knowledge-graph-embedding-with-electronic | 2305.19997 | null | https://arxiv.org/abs/2305.19997v1 | https://arxiv.org/pdf/2305.19997v1.pdf | Knowledge Graph Embedding with Electronic Health Records Data via Latent Graphical Block Model | Due to the increasing adoption of electronic health records (EHR), large scale EHRs have become another rich data source for translational clinical research. Despite its potential, deriving generalizable knowledge from EHR data remains challenging. First, EHR data are generated as part of clinical care with data elements too detailed and fragmented for research. Despite recent progress in mapping EHR data to common ontology with hierarchical structures, much development is still needed to enable automatic grouping of local EHR codes to meaningful clinical concepts at a large scale. Second, the total number of unique EHR features is large, imposing methodological challenges to derive reproducible knowledge graph, especially when interest lies in conditional dependency structure. Third, the detailed EHR data on a very large patient cohort imposes additional computational challenge to deriving a knowledge network. To overcome these challenges, we propose to infer the conditional dependency structure among EHR features via a latent graphical block model (LGBM). The LGBM has a two layer structure with the first providing semantic embedding vector (SEV) representation for the EHR features and the second overlaying a graphical block model on the latent SEVs. The block structures on the graphical model also allows us to cluster synonymous features in EHR. We propose to learn the LGBM efficiently, in both statistical and computational sense, based on the empirical point mutual information matrix. We establish the statistical rates of the proposed estimators and show the perfect recovery of the block structure. Numerical results from simulation studies and real EHR data analyses suggest that the proposed LGBM estimator performs well in finite sample. | ['Tianxi Cai', 'Jin Yin', 'Junwei Lu'] | 2023-05-31 | null | null | null | null | ['graph-embedding', 'knowledge-graph-embedding'] | ['graphs', 'graphs'] | [ 4.44416776e-02 3.17050487e-01 -2.38996774e-01 -4.17199552e-01
-5.03802657e-01 -2.14635506e-01 4.94517460e-02 7.70729721e-01
-8.66738260e-02 6.02638543e-01 5.55400491e-01 -3.70154232e-01
-6.39418602e-01 -7.21552372e-01 -4.74086136e-01 -6.01364970e-01
-4.71108556e-01 1.78440183e-01 -3.00652474e-01 2.07121179e-01
-2.19740078e-01 2.02485040e-01 -9.41695988e-01 1.85183942e-01
1.01575863e+00 6.66888535e-01 2.23725170e-01 4.01938885e-01
-1.10468909e-01 6.97362125e-01 -2.19034135e-01 -3.98929358e-01
1.61645532e-01 -3.51358831e-01 -6.43147349e-01 -1.68286026e-01
-3.15126419e-01 -3.94375548e-02 -4.23028171e-01 1.06352913e+00
4.51294959e-01 -4.64387357e-01 5.87925196e-01 -1.35019422e+00
-6.81147516e-01 7.18500376e-01 -3.46515328e-01 -2.01736480e-01
5.01582146e-01 -2.46946320e-01 1.28967917e+00 -7.33429432e-01
8.71021926e-01 9.18437600e-01 9.51415539e-01 2.41121560e-01
-1.24689591e+00 -5.50270855e-01 -1.38605595e-01 1.49008498e-01
-1.81543529e+00 5.74470824e-03 5.76817870e-01 -7.50976264e-01
6.15195274e-01 3.34302425e-01 8.15574765e-01 5.91238081e-01
3.42890918e-01 4.25138652e-01 9.15523171e-01 -2.46158749e-01
2.41122648e-01 3.82283807e-01 4.35315520e-01 7.29249656e-01
8.65018308e-01 -3.41979295e-01 -3.67012501e-01 -7.93302298e-01
5.92334151e-01 5.24099410e-01 -3.87708187e-01 -5.04925430e-01
-1.18209398e+00 7.38403499e-01 3.91339093e-01 3.54131281e-01
-6.52791679e-01 -6.11724779e-02 2.95341551e-01 3.21626335e-01
1.50801927e-01 1.28852442e-01 -5.19193411e-01 1.05235301e-01
-6.98077142e-01 -1.21863633e-01 9.81553018e-01 1.17091799e+00
6.49861515e-01 -4.69460100e-01 2.15168204e-02 4.97086078e-01
6.40381038e-01 2.96268582e-01 2.48423919e-01 -3.74996424e-01
4.43060756e-01 9.88684118e-01 -1.05798859e-02 -1.27237678e+00
-5.12464583e-01 -3.73180479e-01 -1.29196537e+00 -7.58938432e-01
2.01939002e-01 -3.80205400e-02 -4.65064645e-01 1.78520036e+00
4.88357484e-01 2.44971216e-01 1.64239332e-01 3.50804657e-01
7.97484577e-01 2.87734956e-01 3.52288455e-01 -3.36349487e-01
1.68263745e+00 -9.89441052e-02 -1.01827884e+00 5.68415165e-01
1.22669399e+00 -3.96888524e-01 6.03120089e-01 -1.22305434e-02
-5.25027156e-01 -1.65210530e-01 -7.50238299e-01 1.43619284e-01
-2.51149356e-01 1.74949656e-03 7.41425514e-01 6.14499509e-01
-1.04044771e+00 2.77034998e-01 -9.12262976e-01 -4.68793303e-01
5.83370209e-01 4.05951202e-01 -6.30346060e-01 -3.77922446e-01
-1.37576711e+00 5.26470482e-01 4.35875386e-01 2.56612986e-01
-2.60027558e-01 -9.95878100e-01 -1.01502943e+00 3.12186778e-01
-2.24544927e-02 -1.03204346e+00 4.43733126e-01 -2.59963393e-01
-9.36080098e-01 4.66092139e-01 -1.19522981e-01 -2.36256197e-01
4.37292904e-01 3.39127854e-02 -5.97369254e-01 1.11431047e-01
1.48869723e-01 2.77237773e-01 4.04204130e-01 -1.07863188e+00
-3.89470130e-01 -5.36051035e-01 -3.99023026e-01 4.07968508e-03
-6.13665700e-01 -1.23603970e-01 -2.50566304e-01 -3.73804182e-01
3.13099712e-01 -9.13961828e-01 -5.29334486e-01 -1.74307257e-01
-5.12218416e-01 -1.69928372e-01 6.02321267e-01 -8.37961793e-01
1.69723284e+00 -2.30050206e+00 4.47190665e-02 5.05864203e-01
7.29033709e-01 -1.74935743e-01 1.73541874e-01 8.15929472e-01
-2.36656442e-01 1.63724408e-01 -2.79534429e-01 1.63206726e-01
-1.56049773e-01 1.81607053e-01 -1.04449622e-01 5.82718492e-01
8.78896285e-03 1.00549340e+00 -1.02539742e+00 -6.78756118e-01
1.48504646e-02 6.52646720e-01 -7.48753428e-01 2.52612144e-01
3.87701184e-01 4.90657479e-01 -8.37245882e-01 4.72596169e-01
5.76323032e-01 -9.51595664e-01 7.81936526e-01 -4.54667896e-01
1.97446167e-01 1.02013750e-02 -1.17860615e+00 1.62065136e+00
-7.10089207e-02 9.92217436e-02 -2.52966791e-01 -1.20392227e+00
7.26920366e-01 6.17828071e-01 1.11174524e+00 -1.40963625e-02
-8.15378502e-02 -2.32527144e-02 7.82925040e-02 -6.15727305e-01
5.23773767e-02 -1.19813226e-01 -2.41068587e-01 2.19940513e-01
-1.48685917e-01 5.55060387e-01 -1.67454273e-01 4.77680683e-01
1.53269053e+00 -2.43769407e-01 7.98047006e-01 -4.60131168e-01
4.08154935e-01 -2.02601831e-02 7.79283106e-01 4.56345528e-01
-5.97105846e-02 2.18529582e-01 6.77100956e-01 -4.19593155e-01
-8.06543589e-01 -1.02412641e+00 -6.54064298e-01 1.87192142e-01
8.94034002e-03 -8.38622153e-01 -4.32429522e-01 -5.29181778e-01
3.29977542e-01 2.91454196e-01 -5.73468328e-01 -1.59339577e-01
-1.98766559e-01 -9.89323914e-01 6.65223658e-01 3.10588777e-01
1.15677357e-01 -5.61976671e-01 -5.22750974e-01 3.78119171e-01
-2.73138702e-01 -1.28244579e+00 -1.73724040e-01 -3.25528830e-02
-1.05563235e+00 -1.35187602e+00 -4.31503713e-01 -5.62799931e-01
9.50011194e-01 7.25982115e-02 8.05501521e-01 1.08145706e-01
-6.31086886e-01 4.85180974e-01 -3.82436991e-01 -2.67810851e-01
-3.48436028e-01 -1.86605975e-01 1.48859188e-01 1.71619922e-01
5.84274352e-01 -4.62943465e-01 -8.69699836e-01 1.47756949e-01
-1.15658438e+00 1.93468705e-01 5.95960319e-01 9.39737558e-01
5.30098498e-01 6.91911206e-02 7.80968010e-01 -1.09957576e+00
5.16603708e-01 -9.20250893e-01 -4.18547183e-01 3.69507521e-01
-8.30269396e-01 8.01781714e-02 4.61028337e-01 1.10749125e-01
-7.08446503e-01 7.75228143e-02 3.04904673e-02 -6.93570375e-02
-4.65496350e-03 1.12204218e+00 -1.25570163e-01 4.36995953e-01
3.17857295e-01 7.61279240e-02 3.43012959e-02 -5.01855135e-01
3.85280758e-01 1.04500043e+00 -6.39336407e-02 -3.07270527e-01
6.46431148e-01 5.67082942e-01 1.58645198e-01 -9.73830044e-01
-5.67101061e-01 -7.94856489e-01 -7.16909826e-01 2.23562121e-01
1.09978378e+00 -1.19006121e+00 -9.17230546e-01 -8.87770429e-02
-7.93451130e-01 1.94190860e-01 -1.64935604e-01 8.29568446e-01
-2.57806569e-01 6.39996827e-01 -7.37419486e-01 -6.91761911e-01
-2.95450389e-01 -9.11360681e-01 9.32559311e-01 -3.08763325e-01
-5.34949422e-01 -1.23240650e+00 2.65879095e-01 2.59786874e-01
4.84274663e-02 4.46368545e-01 1.42325902e+00 -6.40936732e-01
-5.74595869e-01 -4.62563217e-01 -4.43316519e-01 -5.87696172e-02
6.72527611e-01 -2.36186773e-01 -6.32048130e-01 -4.65543538e-01
-1.15028217e-01 2.36404270e-01 3.92239213e-01 5.29840887e-01
1.01836193e+00 -2.88320988e-01 -7.47646809e-01 5.60063422e-01
1.62483644e+00 2.11969092e-02 4.48621333e-01 -7.67940655e-02
9.62959826e-01 5.98492026e-01 2.94530720e-01 7.70436645e-01
6.51579559e-01 5.56155205e-01 -4.04108241e-02 -2.03877181e-01
1.76704481e-01 -3.55514199e-01 3.75744924e-02 1.60171974e+00
-2.52608005e-02 1.11298166e-01 -1.09209299e+00 5.08473456e-01
-2.06336093e+00 -7.12916315e-01 -4.74253058e-01 2.34947562e+00
9.50578213e-01 -4.34804261e-01 -1.52349010e-01 -7.21555203e-02
6.84509337e-01 -3.85985315e-01 -3.04082066e-01 2.27246489e-02
8.46799389e-02 -3.48386243e-02 5.05188048e-01 1.67433023e-01
-6.27040863e-01 3.83978903e-01 6.23917627e+00 3.03319067e-01
-7.00494230e-01 9.42885131e-02 3.57272625e-01 4.14271235e-01
-4.99073565e-01 1.31180823e-01 -7.82828271e-01 4.28929806e-01
9.52489376e-01 -3.98701429e-01 -1.71359424e-02 6.63560808e-01
3.68426681e-01 2.03052580e-01 -1.16322839e+00 1.05714643e+00
-3.81359816e-01 -1.30560601e+00 2.03867987e-01 4.73272413e-01
7.35242069e-01 -2.81724986e-02 -2.79186279e-01 -2.81690881e-02
5.50256073e-01 -8.96510243e-01 -1.15766721e-02 4.89408672e-01
1.08105373e+00 -3.92745703e-01 7.92876065e-01 2.36142993e-01
-1.42266595e+00 -8.09001178e-02 -5.30299127e-01 1.63956136e-01
1.40437394e-01 9.73207176e-01 -1.03652167e+00 1.07988942e+00
6.42681420e-01 1.01617610e+00 -3.24387640e-01 9.37686443e-01
6.34010807e-02 6.21987045e-01 -8.72887075e-02 2.83987463e-01
-1.45639464e-01 -4.22770381e-01 2.72362769e-01 1.26286244e+00
5.57123840e-01 9.59746540e-02 1.28381133e-01 7.19796360e-01
-1.00189606e-02 5.15232205e-01 -9.76469934e-01 -3.17836344e-01
4.57173169e-01 1.23422801e+00 -6.68196857e-01 -3.07789892e-01
-7.11034596e-01 5.72432578e-01 3.27952504e-01 4.70532387e-01
-6.03044391e-01 -3.35707814e-01 7.82058954e-01 1.10103354e-01
1.08362652e-01 -9.28990096e-02 -2.34244496e-01 -1.32992029e+00
-7.34317079e-02 -7.12954819e-01 7.97226846e-01 -2.65779316e-01
-1.48355329e+00 4.04263526e-01 -1.61922965e-02 -1.37887609e+00
-1.00323446e-01 -2.89663970e-01 7.85833970e-02 8.68812203e-01
-1.35063863e+00 -8.40658009e-01 -3.45488608e-01 6.90374911e-01
-1.42845839e-01 5.97880408e-02 1.26424694e+00 4.40039277e-01
-5.97120345e-01 5.74547291e-01 3.97362113e-01 3.22111517e-01
5.17101645e-01 -1.09107339e+00 -6.30177408e-02 2.58428603e-01
8.15434232e-02 1.23825395e+00 4.93276924e-01 -9.48916197e-01
-1.63326156e+00 -1.18663824e+00 1.08418417e+00 -5.89741290e-01
7.65073836e-01 -3.75409216e-01 -1.02605927e+00 7.99112260e-01
-1.78124651e-01 1.12503484e-01 1.42781866e+00 3.01433802e-01
-4.04762477e-01 -6.17927276e-02 -1.03645551e+00 5.09877205e-01
9.68400061e-01 -6.09903216e-01 -5.96936941e-01 2.21596777e-01
6.60647333e-01 1.12720311e-01 -1.60889566e+00 5.27818620e-01
6.68982565e-01 -4.76300359e-01 8.42688084e-01 -7.98980892e-01
2.39271551e-01 -4.98349577e-01 -2.14291662e-01 -1.15210998e+00
-4.49475378e-01 -5.76767445e-01 -7.73532316e-02 7.93780148e-01
3.15161049e-01 -8.30784261e-01 5.43365479e-01 7.54692137e-01
1.55537248e-01 -7.00193524e-01 -9.20155287e-01 -6.57223105e-01
-3.10765445e-01 -1.85127154e-01 5.94056308e-01 1.42471611e+00
5.92981279e-01 3.34595025e-01 -3.81621778e-01 3.09103340e-01
7.09682643e-01 1.38361841e-01 7.00201929e-01 -1.61871636e+00
-3.93560827e-01 2.13914998e-02 -8.73651147e-01 -6.72407150e-01
-2.05146417e-01 -1.29838288e+00 -4.32508349e-01 -1.77865052e+00
7.87528753e-01 -7.99331367e-01 -8.25604439e-01 4.14884925e-01
-3.83769244e-01 -2.41098523e-01 -1.25039563e-01 4.28214222e-01
-5.01378834e-01 4.10085917e-01 9.83437538e-01 7.01007247e-02
-3.06566179e-01 -3.42497408e-01 -7.53338873e-01 5.41151464e-01
4.85915214e-01 -7.79718995e-01 -5.51672161e-01 8.80602375e-02
6.17452502e-01 4.38780636e-01 1.75503388e-01 -5.10574639e-01
2.06989795e-01 1.35789886e-01 1.87915862e-01 -3.74407321e-01
-1.15751065e-01 -1.11483097e+00 8.31123829e-01 7.49377012e-01
-3.03659111e-01 -1.16244564e-02 -6.71808869e-02 1.04093981e+00
-3.37930739e-01 1.42886043e-01 2.68004715e-01 2.07820490e-01
-6.68577030e-02 4.32866275e-01 -1.08148597e-01 -4.71052714e-02
9.56683815e-01 -2.19834745e-01 1.15904309e-01 -2.42822200e-01
-8.05608809e-01 1.64824530e-01 2.19683409e-01 2.56553590e-01
5.70202231e-01 -1.48408413e+00 -7.01203346e-01 3.02256465e-01
4.75481957e-01 -1.12712644e-01 4.09924805e-01 1.22695780e+00
-2.70475686e-01 6.48116469e-01 6.02626428e-02 -7.76684999e-01
-1.28743076e+00 5.81870556e-01 -3.15967798e-01 -5.91123104e-01
-1.04977107e+00 2.11199671e-01 5.40441453e-01 -3.20932180e-01
-3.03190183e-02 -3.41391653e-01 -2.45214343e-01 5.20636775e-02
3.31329256e-01 2.08021253e-01 -6.18105866e-02 -6.03114486e-01
-3.99658620e-01 5.47930479e-01 2.57916842e-02 3.70783091e-01
1.46951401e+00 -1.69292256e-01 -3.96336347e-01 3.87190729e-01
1.36063564e+00 -1.78921334e-02 -6.12239182e-01 -2.93305159e-01
3.87929320e-01 -3.08545828e-01 -1.80541784e-01 -1.91644147e-01
-6.61858559e-01 6.28604412e-01 3.69301796e-01 3.02659601e-01
9.03138816e-01 -6.72116410e-03 5.16241193e-01 4.43655044e-01
4.26218808e-01 -7.24395871e-01 -3.95014256e-01 5.76290563e-02
4.34061676e-01 -9.56404924e-01 1.19843133e-01 -8.01917195e-01
-4.88737285e-01 8.28900278e-01 -1.35371327e-01 2.42278844e-01
1.25527787e+00 9.55607966e-02 6.28448352e-02 -6.27407730e-01
-7.01508701e-01 7.99849555e-02 3.01092923e-01 4.65603769e-01
6.38762832e-01 3.96271348e-01 -6.26178145e-01 7.38547325e-01
5.28571196e-02 2.46044204e-01 3.15721542e-01 8.09464693e-01
-4.11076248e-02 -1.25775194e+00 -6.26675934e-02 7.81647086e-01
-6.21236205e-01 -3.11088592e-01 8.67845938e-02 7.34978676e-01
-1.80833116e-01 7.60097861e-01 -2.60072887e-01 -2.50395685e-01
2.01738298e-01 2.37417638e-01 1.44247040e-01 -6.26306713e-01
-2.24282831e-01 4.23358083e-02 -1.37522697e-01 -4.29582655e-01
-2.88661271e-01 -6.29943609e-01 -1.42236924e+00 7.08907694e-02
-4.27778393e-01 4.26117092e-01 6.70870423e-01 8.16489339e-01
8.71768951e-01 5.22517741e-01 6.42119110e-01 7.44494349e-02
-6.34649038e-01 -6.85862362e-01 -8.33171964e-01 7.72643864e-01
-2.19543949e-02 -4.73835528e-01 -9.39710662e-02 1.86897323e-01] | [7.461201190948486, 6.2496657371521] |
7c694f3a-1a42-4353-96d7-4f2bbdbbea16 | evconv-fast-cnn-inference-on-event-camera | 2303.04670 | null | https://arxiv.org/abs/2303.04670v1 | https://arxiv.org/pdf/2303.04670v1.pdf | EvConv: Fast CNN Inference on Event Camera Inputs For High-Speed Robot Perception | Event cameras capture visual information with a high temporal resolution and a wide dynamic range. This enables capturing visual information at fine time granularities (e.g., microseconds) in rapidly changing environments. This makes event cameras highly useful for high-speed robotics tasks involving rapid motion, such as high-speed perception, object tracking, and control. However, convolutional neural network inference on event camera streams cannot currently perform real-time inference at the high speeds at which event cameras operate - current CNN inference times are typically closer in order of magnitude to the frame rates of regular frame-based cameras. Real-time inference at event camera rates is necessary to fully leverage the high frequency and high temporal resolution that event cameras offer. This paper presents EvConv, a new approach to enable fast inference on CNNs for inputs from event cameras. We observe that consecutive inputs to the CNN from an event camera have only small differences between them. Thus, we propose to perform inference on the difference between consecutive input tensors, or the increment. This enables a significant reduction in the number of floating-point operations required (and thus the inference latency) because increments are very sparse. We design EvConv to leverage the irregular sparsity in increments from event cameras and to retain the sparsity of these increments across all layers of the network. We demonstrate a reduction in the number of floating operations required in the forward pass by up to 98%. We also demonstrate a speedup of up to 1.6X for inference using CNNs for tasks such as depth estimation, object recognition, and optical flow estimation, with almost no loss in accuracy. | ['Nandita Vijaykumar', 'Yushi Guan', 'Sankeerth Durvasula'] | 2023-03-08 | null | null | null | null | ['object-recognition'] | ['computer-vision'] | [-1.97691638e-02 -4.76330817e-01 -1.25690803e-01 -8.89897570e-02
-2.86710203e-01 -5.39183140e-01 4.95763808e-01 1.99590653e-01
-8.31296206e-01 3.76395434e-01 -8.47733170e-02 -2.57737935e-01
1.21491998e-01 -7.84378290e-01 -9.87788796e-01 -4.19041395e-01
-2.07734495e-01 3.91748808e-02 6.35729373e-01 3.06270838e-01
9.92167294e-02 8.61120343e-01 -1.88408720e+00 4.81865764e-01
1.51262313e-01 1.36562324e+00 1.19716465e-01 1.20353985e+00
8.61200616e-02 1.03400338e+00 -3.97812784e-01 1.33758515e-01
3.86918128e-01 7.96989799e-02 -4.81525391e-01 8.16467106e-02
9.66769099e-01 -1.08485544e+00 -7.11151719e-01 6.57188594e-01
1.25877902e-01 3.79048258e-01 -2.59015523e-02 -1.31078577e+00
3.77819352e-02 1.53115526e-01 -5.45506895e-01 5.26736438e-01
1.98860496e-01 5.02186596e-01 7.88489521e-01 -6.76122546e-01
7.54313469e-01 1.10404789e+00 5.13023555e-01 3.85214955e-01
-1.15041447e+00 -6.22486889e-01 3.70892614e-01 3.46308619e-01
-1.06879008e+00 -6.02577329e-01 2.69291371e-01 -5.50948918e-01
1.38151407e+00 -8.69430676e-02 1.01350915e+00 8.90377462e-01
2.69457787e-01 4.01935399e-01 4.51215714e-01 2.11509779e-01
4.28750664e-01 -6.53727412e-01 1.36466961e-05 8.17603409e-01
2.66435087e-01 1.24784946e-01 -1.06650317e+00 9.83275175e-02
1.37064636e+00 4.71506417e-01 -3.19376856e-01 9.71417353e-02
-1.72341156e+00 6.71217203e-01 5.71511388e-01 -3.18786114e-01
-5.16932309e-01 1.06350625e+00 6.69576287e-01 2.22467199e-01
9.47740972e-02 1.94319502e-01 -4.66306329e-01 -5.06117463e-01
-7.40471423e-01 1.21842496e-01 8.01307201e-01 9.60668981e-01
7.72911608e-01 2.58220881e-01 8.35654065e-02 6.30199462e-02
4.87851650e-02 6.24318957e-01 2.29245290e-01 -1.51247609e+00
4.43955451e-01 3.30936879e-01 1.03166819e-01 -9.67815876e-01
-3.29429686e-01 -3.67145315e-02 -9.37634170e-01 6.27663791e-01
6.59527123e-01 -1.96111664e-01 -7.88521409e-01 1.50554609e+00
5.20542681e-01 6.09876394e-01 -6.59139305e-02 1.00371635e+00
5.22649705e-01 8.67442608e-01 -2.22324520e-01 -1.81169972e-01
1.56456709e+00 -7.43983150e-01 -3.39724928e-01 -3.21085393e-01
4.15929973e-01 -7.07744062e-01 9.80381250e-01 5.38802207e-01
-1.02493429e+00 -4.90980119e-01 -1.00894666e+00 -4.44744796e-01
3.16144563e-02 4.70102252e-03 9.28941071e-01 -1.68370560e-01
-8.31248581e-01 7.28597105e-01 -1.72508860e+00 -1.42867997e-01
4.46688920e-01 6.50345445e-01 -4.56385434e-01 -9.11855325e-02
-3.67095411e-01 2.71737516e-01 2.27374330e-01 8.38809311e-02
-9.95741725e-01 -1.13275301e+00 -8.19303691e-01 1.79530963e-01
3.60834032e-01 -6.03967190e-01 1.43184519e+00 -7.83288658e-01
-1.46463537e+00 3.69143426e-01 -3.63560647e-01 -7.91555941e-01
4.08001304e-01 -4.04214621e-01 1.46249890e-01 4.61004525e-01
-1.24302350e-01 8.42358112e-01 8.24455619e-01 -4.30081815e-01
-8.52761209e-01 -2.54293174e-01 5.37852108e-01 -4.78836447e-02
-3.38943303e-01 -2.23726258e-01 -5.98657966e-01 -2.67341614e-01
7.63214082e-02 -1.21538210e+00 -1.68134630e-01 7.98851728e-01
6.52251318e-02 8.98196548e-02 1.10516047e+00 -1.78638294e-01
6.34479463e-01 -2.25213957e+00 -1.03656948e-01 -2.82920718e-01
5.54164052e-01 1.41501352e-01 1.81297451e-01 5.03258295e-02
2.30135709e-01 -3.99336874e-01 8.22703019e-02 -4.40904230e-01
-3.47014755e-01 4.54131335e-01 -5.22557616e-01 5.46246886e-01
2.58229673e-01 6.92826927e-01 -9.34720337e-01 -2.35306591e-01
5.68349063e-01 6.96499288e-01 -7.67462015e-01 1.67230442e-01
-4.82688874e-01 4.02217090e-01 -2.99387187e-01 2.94489115e-01
2.75213748e-01 -6.36763752e-01 -2.01533530e-02 -5.84918737e-01
-4.98525053e-01 4.56725329e-01 -1.27172267e+00 1.71528161e+00
-6.14863157e-01 1.23620784e+00 8.44322890e-02 -5.19250631e-01
4.22189921e-01 2.94005066e-01 5.50411403e-01 -5.58881462e-01
9.10977125e-02 -2.42002984e-03 -1.40196547e-01 -1.92496568e-01
6.78669930e-01 7.61156380e-02 1.32048115e-01 3.54142815e-01
-1.71733260e-01 9.51571837e-02 4.65602726e-01 1.60709038e-01
1.34602416e+00 4.20642421e-02 9.45454612e-02 1.43372923e-01
-2.43046377e-02 2.22501352e-01 6.43274426e-01 5.91431022e-01
-1.18672401e-01 3.57704341e-01 4.53049093e-01 -8.97368729e-01
-1.21679258e+00 -8.92102301e-01 6.44336939e-02 9.57014084e-01
1.36158183e-01 -4.87075150e-01 -3.11719120e-01 1.79589707e-02
1.13437548e-01 -2.73193326e-02 -3.83322001e-01 -2.63125598e-02
-8.97723138e-01 -4.27565545e-01 4.49950457e-01 1.06165528e+00
6.77965283e-01 -6.61645114e-01 -1.44506717e+00 4.09939080e-01
8.71310607e-02 -1.82422256e+00 -4.20190096e-01 1.28078192e-01
-1.16605568e+00 -1.11894476e+00 -7.32542500e-02 -4.67785120e-01
4.76417273e-01 4.61584240e-01 9.07606363e-01 -1.69169977e-02
-3.63527954e-01 2.80169189e-01 -3.16159688e-02 -2.23627493e-01
1.38443470e-01 -1.59839079e-01 -2.21565589e-02 6.78644404e-02
4.65591885e-02 -7.40564525e-01 -7.19085336e-01 -9.77119245e-03
-1.03469241e+00 3.39741081e-01 2.00668290e-01 6.61872983e-01
7.53112078e-01 -1.51725620e-01 -1.64704487e-01 -3.24213922e-01
-1.18325859e-01 -2.18171343e-01 -1.38339341e+00 -4.37923372e-01
-8.65929760e-03 1.41051710e-01 9.58305717e-01 -9.13370609e-01
-8.05570900e-01 3.20041895e-01 1.56162739e-01 -9.49529588e-01
-3.89881060e-02 3.08465004e-01 5.18719673e-01 -1.72405496e-01
5.19870222e-01 -1.50135234e-01 3.94633152e-02 -7.74871409e-02
1.17049672e-01 8.97271112e-02 8.10521185e-01 -4.81298596e-01
3.53364915e-01 1.09011304e+00 3.80153656e-01 -8.84239554e-01
-7.18902588e-01 -4.19614255e-01 -2.83757180e-01 -2.52216041e-01
9.07993555e-01 -1.24642491e+00 -1.48580885e+00 6.75133228e-01
-1.46351147e+00 -7.31554747e-01 -2.87869513e-01 9.01109993e-01
-3.78049910e-01 -1.48621444e-02 -1.12417817e+00 -4.71872985e-01
-2.54651070e-01 -1.21998191e+00 1.23312199e+00 3.01751822e-01
-4.53955028e-03 -7.32925832e-01 -3.10365885e-01 5.65768741e-02
3.44301671e-01 4.37943935e-01 3.41800004e-01 1.31836787e-01
-1.37326980e+00 -8.94981343e-03 -5.03277183e-01 -1.88839771e-02
-1.63353980e-01 3.37394208e-01 -8.34663987e-01 -3.78433168e-01
-8.63975063e-02 -1.34979755e-01 8.47368240e-01 4.26893622e-01
1.21835434e+00 -2.04461887e-01 -1.35483608e-01 1.02330828e+00
1.67934585e+00 1.09952413e-01 5.16129255e-01 2.36542709e-02
9.49920535e-01 1.01451248e-01 4.61230427e-01 7.97172427e-01
2.50335574e-01 6.89101815e-01 7.05344677e-01 1.97404772e-01
-2.80991793e-01 1.30699515e-01 5.53448319e-01 7.81246006e-01
-1.99688345e-01 -1.53071046e-01 -8.67967725e-01 6.71609461e-01
-1.87093735e+00 -1.03180408e+00 -2.86751002e-01 2.39384460e+00
6.13350391e-01 2.63402551e-01 -1.71044201e-01 -3.29600573e-02
5.11585891e-01 2.70394713e-01 -8.50775003e-01 -3.46993893e-01
2.65926778e-01 2.02243716e-01 7.93757737e-01 5.00296772e-01
-8.90896738e-01 9.01318252e-01 5.59102106e+00 2.93228716e-01
-1.66611540e+00 -8.38785470e-02 2.60795951e-01 -8.37216198e-01
2.44760618e-01 7.81860724e-02 -1.00962961e+00 4.25589263e-01
1.14710128e+00 4.20468971e-02 5.65666080e-01 8.38135719e-01
3.37408751e-01 -2.74872899e-01 -1.36813962e+00 1.25182545e+00
-2.69587666e-01 -1.75471091e+00 4.08147201e-02 2.87149996e-01
6.57851934e-01 4.92981911e-01 -2.77761638e-01 -1.85878217e-01
2.60146528e-01 -7.16191053e-01 6.98837638e-01 3.44280064e-01
8.41989994e-01 -6.36406958e-01 3.54654282e-01 2.02243149e-01
-1.52999568e+00 -6.70039579e-02 -4.80029374e-01 -5.15364289e-01
3.24334294e-01 9.99721169e-01 -7.12673426e-01 -9.57947746e-02
8.24772418e-01 8.68609190e-01 -1.61726270e-02 8.03712666e-01
1.19881079e-01 4.41430300e-01 -8.14858079e-01 4.15659100e-02
3.13762933e-01 8.18231925e-02 4.18942183e-01 1.08474267e+00
3.71423125e-01 2.01671094e-01 2.40712613e-01 6.45092189e-01
-3.07984382e-01 -6.95042789e-01 -5.55145860e-01 3.67939100e-02
6.21387661e-01 1.12248218e+00 -7.37048030e-01 -5.55023789e-01
-7.10481048e-01 1.10022402e+00 3.02838981e-01 1.88115939e-01
-1.04743469e+00 -3.40369254e-01 1.24503326e+00 6.67588934e-02
6.66863978e-01 -9.70418930e-01 -1.94118723e-01 -1.32652879e+00
3.41335237e-01 -5.70403576e-01 1.35012418e-01 -8.20169091e-01
-7.76845574e-01 2.51387298e-01 -1.28009439e-01 -1.12940037e+00
-3.25175285e-01 -8.25184047e-01 -3.72531831e-01 3.46103013e-01
-1.40989387e+00 -8.33289921e-01 -8.62427771e-01 8.87042165e-01
5.81799865e-01 3.63767624e-01 4.40945238e-01 4.18357551e-01
-5.08514583e-01 1.44091234e-01 -6.04924895e-02 4.16277707e-01
4.41077441e-01 -9.18245792e-01 6.18896961e-01 1.00757468e+00
7.65140578e-02 5.15586555e-01 4.01079953e-01 -4.15396392e-01
-2.09558034e+00 -1.27271533e+00 4.90271509e-01 -1.92299545e-01
7.35441446e-01 -2.54020959e-01 -6.97830975e-01 8.94449174e-01
-2.45250881e-01 6.82487249e-01 1.78937346e-01 -2.07882911e-01
-7.03638315e-01 -5.49393713e-01 -6.31669581e-01 6.19497776e-01
9.87186134e-01 -7.73385942e-01 -5.02116084e-02 2.30111569e-01
9.94713843e-01 -8.39369535e-01 -9.97273266e-01 9.37395692e-02
7.44216800e-01 -9.53440368e-01 1.05797040e+00 -3.85394543e-01
6.01082981e-01 -5.34963489e-01 -1.16224639e-01 -5.56583762e-01
-4.40794118e-02 -6.28056109e-01 -6.51476383e-01 7.22665250e-01
-7.96793029e-02 -5.97192883e-01 7.67967463e-01 5.46487510e-01
-9.34828296e-02 -5.99507034e-01 -8.49402249e-01 -6.68346941e-01
-6.11384571e-01 -6.95035815e-01 3.54039013e-01 7.35091090e-01
-3.90966445e-01 2.49549106e-01 -1.07310645e-01 4.11718518e-01
6.82297468e-01 2.78271705e-01 8.78770590e-01 -1.01816022e+00
-4.54626232e-01 -1.70099825e-01 -7.93869078e-01 -1.45556104e+00
-1.06895119e-01 -4.02550757e-01 7.73177668e-02 -1.17801607e+00
-5.73161095e-02 -1.45062268e-01 7.31359944e-02 4.85169411e-01
1.49085745e-01 4.79664087e-01 3.20205539e-01 2.80538142e-01
-5.33482015e-01 2.18275294e-01 1.11627591e+00 1.94844738e-01
-1.29466012e-01 -5.06698549e-01 -7.10187703e-02 9.08362329e-01
5.61704874e-01 -3.15796465e-01 -3.86558563e-01 -1.10198271e+00
3.65210056e-01 3.69231641e-01 1.02499962e+00 -1.39687514e+00
6.55093968e-01 -2.53816575e-01 4.05763566e-01 -4.88005936e-01
7.36349702e-01 -6.49791598e-01 2.85178572e-01 4.32171792e-01
-1.49465054e-01 3.92625332e-01 4.55064893e-01 6.42655134e-01
-2.30241433e-01 3.86075258e-01 8.10150743e-01 -1.97919652e-01
-9.05545354e-01 6.64406538e-01 -4.85920638e-01 2.51694024e-02
9.11993742e-01 -2.27659479e-01 -4.49506968e-01 -2.11452708e-01
-4.45586294e-01 -7.96133131e-02 5.35108209e-01 1.51412264e-01
6.07408464e-01 -1.22706509e+00 -3.58919948e-01 1.55672640e-01
-1.14836961e-01 5.91716826e-01 2.45295823e-01 9.13235784e-01
-1.00014532e+00 2.33631283e-01 -4.00556624e-01 -1.05670071e+00
-1.25428832e+00 2.55000204e-01 2.61229604e-01 1.04114739e-03
-1.03435636e+00 8.09230447e-01 2.44010329e-01 3.75740737e-01
2.12630779e-01 -8.59198511e-01 4.20188785e-01 -2.13085085e-01
8.82711411e-01 5.64552307e-01 1.41676739e-01 -1.69893429e-01
-2.92115450e-01 5.98324120e-01 6.06205966e-03 -1.79479942e-01
1.18378067e+00 1.03618562e-01 -1.69029832e-01 4.67819691e-01
1.41123843e+00 -2.40833834e-01 -2.07954574e+00 -1.46005824e-02
-4.43343312e-01 -5.66687167e-01 4.53903526e-01 -1.94087420e-02
-1.39826238e+00 1.07021224e+00 3.55736464e-01 -1.01189874e-01
1.11645746e+00 -2.41619572e-01 1.26178050e+00 7.32541800e-01
4.99831080e-01 -8.48092973e-01 1.62998900e-01 7.87901223e-01
4.80248988e-01 -1.08630049e+00 -6.86139381e-03 -3.80533040e-01
-4.78942208e-02 1.31080258e+00 6.49859786e-01 -4.67567652e-01
3.57141465e-01 7.42705345e-01 -2.89347768e-01 -8.62517133e-02
-1.19466758e+00 -4.01878357e-03 -1.10848255e-01 3.54599267e-01
1.64441153e-01 -9.24909860e-02 3.26765120e-01 -2.81713605e-01
1.55474218e-02 1.72260880e-01 7.92238474e-01 1.05980492e+00
-3.72024596e-01 -5.85688889e-01 -2.84601986e-01 4.35746998e-01
-3.10459882e-01 -9.23510715e-02 2.72717416e-01 5.80402315e-01
3.48927639e-02 8.36681068e-01 8.29387486e-01 -2.20970199e-01
1.44460410e-01 -3.62620294e-01 5.02426088e-01 -4.38773990e-01
-4.37843263e-01 -1.38525635e-01 1.65870655e-02 -1.29036486e+00
-3.80350083e-01 -5.52595675e-01 -1.64676702e+00 -9.53059614e-01
-4.71787825e-02 -4.67319965e-01 8.01508009e-01 8.04747045e-01
6.93993330e-01 5.64016461e-01 1.94261834e-01 -1.16553652e+00
-6.68502897e-02 -3.46733630e-01 -5.25872260e-02 2.42599726e-01
7.56875932e-01 -4.00566459e-01 -3.61567587e-01 5.41213632e-01] | [8.621220588684082, -0.940251350402832] |
32b3206c-32f8-4bfb-81d9-d5c302535234 | towards-efficient-and-accurate-approximation | 2211.14828 | null | https://arxiv.org/abs/2211.14828v1 | https://arxiv.org/pdf/2211.14828v1.pdf | Towards Efficient and Accurate Approximation: Tensor Decomposition Based on Randomized Block Krylov Iteration | Efficient and accurate low-rank approximation (LRA) methods are of great significance for large-scale data analysis. Randomized tensor decompositions have emerged as powerful tools to meet this need, but most existing methods perform poorly in the presence of noise interference. Inspired by the remarkable performance of randomized block Krylov iteration (rBKI) in reducing the effect of tail singular values, this work designs an rBKI-based Tucker decomposition (rBKI-TK) for accurate approximation, together with a hierarchical tensor ring decomposition based on rBKI-TK for efficient compression of large-scale data. Besides, the error bound between the deterministic LRA and the randomized LRA is studied. Numerical experiences demonstrate the efficiency, accuracy and scalability of the proposed methods in both data compression and denoising. | ['Qibin Zhao', 'Guoxu Zhou', 'Weijun Sun', 'Yichun Qiu'] | 2022-11-27 | null | null | null | null | ['data-compression'] | ['time-series'] | [-2.03619376e-01 -6.01223290e-01 1.25898898e-01 2.25780517e-01
-8.54294121e-01 -2.22758040e-01 5.91200218e-02 6.47363812e-02
-1.55817062e-01 4.86372083e-01 6.90618515e-01 -1.63668767e-01
-6.20150805e-01 -6.48959100e-01 -4.92517442e-01 -1.15299606e+00
-3.38222772e-01 1.41187375e-02 -9.77224782e-02 -2.25232348e-01
1.48578525e-01 2.54595131e-01 -1.04020691e+00 3.52613717e-01
7.58603156e-01 1.14776564e+00 -1.80409532e-02 4.60504115e-01
1.37084261e-01 7.63392985e-01 -1.26986131e-01 -2.62255251e-01
4.24211651e-01 -2.06051230e-01 -5.96738338e-01 1.61535472e-01
4.03407821e-03 -4.60984617e-01 -7.61004508e-01 9.18578506e-01
4.37191755e-01 2.92682499e-01 3.45444292e-01 -6.82644188e-01
-3.85101914e-01 6.25204146e-01 -1.11582887e+00 2.30759025e-01
2.89345741e-01 -3.61977160e-01 1.06497478e+00 -1.16760826e+00
2.30459347e-01 1.26831520e+00 8.63012552e-01 -1.38020948e-01
-1.46951103e+00 -5.31256616e-01 -2.04181984e-01 2.89229095e-01
-1.85764611e+00 -5.29454172e-01 8.40820909e-01 -2.97000915e-01
2.78877795e-01 5.57554543e-01 3.46679866e-01 5.38283527e-01
-1.56861003e-02 9.25878584e-01 1.30121195e+00 -3.35026473e-01
2.58081794e-01 -5.12019336e-01 3.64155978e-01 5.01377046e-01
6.21115029e-01 -7.70775154e-02 -6.14783406e-01 -7.49761164e-01
7.31340528e-01 3.33801448e-01 -3.75915706e-01 -6.22858740e-02
-1.45240748e+00 7.09633708e-01 1.73385933e-01 4.48857903e-01
-7.38196969e-01 2.91762918e-01 8.38108301e-01 8.76366422e-02
3.82904530e-01 -5.69903553e-02 -1.53150693e-01 -4.10615327e-03
-1.02312350e+00 3.75902832e-01 4.11690295e-01 7.52551556e-01
4.78467584e-01 2.41282642e-01 -5.17301440e-01 9.82254684e-01
1.82624653e-01 6.25437796e-01 4.26062524e-01 -1.00387418e+00
6.43040240e-01 1.76006705e-01 1.73827231e-01 -1.63834429e+00
-3.24760288e-01 -7.60481954e-01 -1.66126025e+00 -4.12328035e-01
2.97313005e-01 3.56742710e-01 -3.42144191e-01 1.13902771e+00
5.14988184e-01 4.29193854e-01 2.21555486e-01 1.08573389e+00
4.78824347e-01 9.62218463e-01 -4.69805300e-01 -7.55709529e-01
1.31549919e+00 -5.09007871e-01 -9.13546860e-01 4.96197581e-01
6.97946489e-01 -8.63607705e-01 8.14035118e-01 8.30439508e-01
-1.02843869e+00 -4.11286354e-01 -7.57522821e-01 -9.89926234e-03
4.85365480e-01 3.09529901e-01 8.03710580e-01 4.97011691e-01
-7.09740698e-01 5.62613428e-01 -7.75018454e-01 1.91975698e-01
2.93456614e-01 2.37623915e-01 -2.92500108e-01 -6.18520677e-01
-9.59895492e-01 3.95897105e-02 2.52872501e-02 7.26311743e-01
-7.58153617e-01 -5.65191984e-01 -2.12479845e-01 2.76141111e-02
4.49863642e-01 -4.63164747e-01 7.03692675e-01 -7.14976713e-02
-1.16376793e+00 2.63494521e-01 -3.86064678e-01 -5.51173329e-01
2.68350393e-01 -2.89923519e-01 -7.27898255e-02 3.72740537e-01
2.21599396e-02 -5.14771640e-01 1.06366825e+00 -1.10265601e+00
-1.76830873e-01 -6.65463567e-01 -2.56720692e-01 -3.30344811e-02
-5.64250946e-01 -1.64688081e-01 -3.11809897e-01 -1.10101199e+00
8.41875672e-01 -7.59460151e-01 -7.18156815e-01 -5.36880970e-01
-4.16057616e-01 -1.41151547e-01 6.61005139e-01 -9.93067145e-01
1.67150414e+00 -2.30756235e+00 3.78429055e-01 4.01685834e-01
4.78259772e-01 2.03235224e-01 -3.99692096e-02 8.12165558e-01
-3.32928710e-02 -1.53491020e-01 3.90852876e-02 -1.56563282e-01
-3.82408887e-01 2.52517134e-01 -6.82750702e-01 6.78952336e-01
-6.83090389e-01 2.56668001e-01 -8.74684393e-01 -2.69473404e-01
-8.41161534e-02 3.83297354e-01 -7.11434662e-01 1.05355605e-01
4.06854838e-01 4.77444589e-01 -6.99112535e-01 5.28126478e-01
8.96794736e-01 -1.67661816e-01 1.36872113e-01 -8.99672806e-01
-1.33692488e-01 -2.35160872e-01 -1.55734372e+00 1.26279104e+00
-3.78210247e-01 1.01935863e-01 5.59456289e-01 -1.07171154e+00
8.54829013e-01 4.72981721e-01 6.97964668e-01 -4.64647502e-01
-3.56596708e-02 3.41358364e-01 -2.92926043e-01 -2.68616676e-01
5.97470343e-01 -1.54487818e-01 8.56394619e-02 2.69397050e-01
-5.55160880e-01 2.39444330e-01 4.76663262e-01 5.73565960e-01
1.19405937e+00 -3.47149193e-01 1.48729175e-01 -3.10516268e-01
7.64309406e-01 -2.53524095e-01 8.88166964e-01 5.69389105e-01
1.56564191e-01 6.00399733e-01 2.94366658e-01 -6.18020415e-01
-1.11968565e+00 -5.16879857e-01 -1.11932091e-01 8.11069965e-01
5.14053144e-02 -8.71557832e-01 -4.33367074e-01 -2.39924882e-02
4.62731943e-02 2.13171691e-01 -2.77570903e-01 1.65694635e-02
-5.78865647e-01 -9.17231500e-01 3.85760367e-01 2.57267833e-01
7.20674336e-01 -7.19343871e-02 1.58878356e-01 3.30033362e-01
-5.85541248e-01 -1.23154843e+00 -4.55078512e-01 -3.33394259e-01
-1.27803111e+00 -8.91440988e-01 -7.26396978e-01 -1.65025949e-01
7.26632833e-01 1.08501709e+00 4.59240079e-01 7.05078319e-02
-1.61527634e-01 3.12382728e-01 -6.24945343e-01 4.65326428e-01
-1.93783641e-01 -4.25502092e-01 4.12904859e-01 5.89058697e-01
-2.86317170e-01 -8.50794733e-01 -7.66762495e-01 4.09003824e-01
-1.17983294e+00 7.83052370e-02 5.63794315e-01 9.83577788e-01
8.93990755e-01 6.55353427e-01 1.71380907e-01 -5.57350039e-01
8.44875813e-01 -4.20620382e-01 -7.59826899e-01 1.23399431e-02
-5.75082719e-01 5.42488322e-02 1.02475417e+00 -1.44762367e-01
-7.46872425e-01 -8.73617977e-02 1.18556149e-01 -7.45056748e-01
6.87277377e-01 9.01238620e-01 2.45306507e-01 -1.47592440e-01
4.25558954e-01 5.06587744e-01 -1.66849971e-01 -9.63001192e-01
3.75848264e-01 6.51275337e-01 4.06084567e-01 -8.50191116e-01
8.90409648e-01 8.16128910e-01 5.58453143e-01 -1.05175507e+00
-9.87744629e-01 -8.44550192e-01 -3.29160810e-01 1.21467352e-01
3.72882307e-01 -1.17858958e+00 -8.47942054e-01 3.24266821e-01
-8.65250468e-01 2.92725831e-01 -1.27040043e-01 7.38756597e-01
-2.79184967e-01 9.54228342e-01 -8.45399976e-01 -8.98370326e-01
-5.39861023e-01 -9.79503632e-01 6.91478610e-01 -3.86898726e-01
3.38362962e-01 -4.44816262e-01 3.69418338e-02 6.91558957e-01
2.58288443e-01 2.45308444e-01 6.75563276e-01 -2.95376573e-02
-5.88755965e-01 -5.77973545e-01 -5.16396940e-01 5.17406583e-01
-1.54703721e-01 -2.47360975e-01 -3.80100995e-01 -7.11912930e-01
1.30620584e-01 -1.16673224e-01 7.60359168e-01 4.72742856e-01
1.14837921e+00 -8.52671504e-01 -1.42433494e-01 7.08961427e-01
1.48530459e+00 -3.51634055e-01 6.20393991e-01 -1.91749039e-03
9.16792095e-01 7.53600076e-02 6.71276569e-01 1.05670142e+00
1.79818884e-01 6.34637892e-01 1.90052614e-01 3.21438491e-01
4.29470800e-02 -1.55420944e-01 2.73503482e-01 1.77520502e+00
-5.20157278e-01 2.86819607e-01 -7.05222189e-01 5.56947708e-01
-1.89375043e+00 -9.27960873e-01 -7.90079236e-01 2.61496162e+00
7.45891094e-01 -2.19179407e-01 1.18127905e-01 6.63066089e-01
4.31595951e-01 2.93300807e-01 -1.36318043e-01 -1.34230122e-01
-1.95784479e-01 -2.12336272e-01 7.94182658e-01 2.68924683e-01
-7.80818343e-01 4.65112835e-01 6.21127367e+00 1.42515707e+00
-5.48721552e-01 2.50065774e-01 4.32838053e-01 1.02485657e-01
-3.20422411e-01 2.41432533e-01 -5.91067553e-01 3.53859514e-01
8.18618655e-01 -1.13603510e-01 7.85733759e-01 7.93584764e-01
7.57622778e-01 -1.32590950e-01 -3.51083994e-01 1.32183647e+00
-3.06751668e-01 -1.41001344e+00 1.90533400e-01 2.38837123e-01
8.49969804e-01 -2.35459119e-01 -1.06551675e-02 -4.48062085e-02
1.62110358e-01 -6.48935139e-01 3.00926894e-01 4.22020823e-01
6.45894647e-01 -8.15462291e-01 6.74386322e-01 2.13513449e-01
-1.35327923e+00 -3.42479408e-01 -8.98060441e-01 -1.71777353e-01
-2.02893261e-02 1.45701432e+00 -3.73078942e-01 8.15788865e-01
7.29918301e-01 7.07892656e-01 -1.19377278e-01 9.79665041e-01
5.34065366e-02 1.04874218e+00 -5.03130972e-01 3.19971502e-01
2.33305499e-01 -8.13761175e-01 8.34455967e-01 8.34977269e-01
3.42941761e-01 6.04781330e-01 3.48599851e-01 2.11977020e-01
3.34995016e-02 4.65826929e-01 -3.45536858e-01 -2.03344241e-01
3.36518735e-01 1.16689873e+00 -4.90243584e-01 -2.41997004e-01
-3.14423412e-01 7.91683912e-01 -6.58705411e-03 5.45311451e-01
-3.31266433e-01 -1.69629529e-01 6.26650155e-01 2.23516285e-01
5.08157969e-01 -7.34962881e-01 -3.60262245e-01 -1.35104883e+00
1.63585708e-01 -1.07100976e+00 5.63706100e-01 -4.82670277e-01
-1.05943334e+00 3.57762873e-01 -9.95218977e-02 -1.55104339e+00
2.51916438e-01 -8.84782001e-02 1.84802786e-02 4.76607800e-01
-1.00324643e+00 -1.14301074e+00 -1.20180383e-01 7.05394328e-01
1.23777762e-01 -1.81623932e-03 5.66343129e-01 5.39056063e-01
-6.40760601e-01 4.41617757e-01 1.07636034e+00 2.65421141e-02
1.53924897e-01 -5.50718069e-01 -2.21351653e-01 9.26862597e-01
-2.11959966e-02 8.60545695e-01 9.71635520e-01 -5.50255895e-01
-2.19381452e+00 -8.01133394e-01 3.63079458e-01 2.98944622e-01
7.00158060e-01 -3.13719623e-02 -8.76887262e-01 3.09542179e-01
-4.01963145e-01 3.66374761e-01 5.75097263e-01 2.64220506e-01
-6.68828309e-01 -7.30579019e-01 -9.42554593e-01 5.41491985e-01
7.50482142e-01 -4.80732769e-01 -1.12302639e-01 6.89399660e-01
6.69015944e-01 -2.93086022e-01 -1.19305182e+00 3.49729002e-01
4.97773260e-01 -9.69645083e-01 1.27537894e+00 -3.70651871e-01
4.12834972e-01 -6.20281935e-01 -6.11664832e-01 -8.56873214e-01
-6.35123551e-01 -1.02853596e+00 -3.24413896e-01 1.04232025e+00
-3.37321579e-01 -2.99960434e-01 6.68025255e-01 3.49476367e-01
1.63979292e-01 -7.45391250e-01 -1.20031857e+00 -9.32787836e-01
-4.63554084e-01 -4.69805866e-01 2.79568285e-01 8.31889033e-01
-4.68546860e-02 2.15557575e-01 -9.37088132e-01 3.88781637e-01
1.12460423e+00 2.93657869e-01 8.88342142e-01 -8.87025952e-01
-5.34294486e-01 1.67402835e-03 -5.41538000e-01 -1.15631855e+00
-4.95481372e-01 -7.49118567e-01 -3.27991903e-01 -1.34840572e+00
3.88708770e-01 -5.75778961e-01 -3.21281880e-01 9.70099419e-02
-3.45700979e-02 4.22975749e-01 1.25040144e-01 8.59409928e-01
-5.83555937e-01 7.81981945e-01 1.15448511e+00 6.64249137e-02
-2.25607723e-01 1.06415130e-01 -4.49676901e-01 4.70291466e-01
1.46729112e-01 -5.03830135e-01 -2.56747931e-01 -3.48283619e-01
6.35272324e-01 4.74706113e-01 -2.94098668e-02 -9.60977793e-01
1.53620198e-01 -1.16879687e-01 -1.42349198e-01 -7.72155404e-01
2.76609927e-01 -7.78450370e-01 3.21471423e-01 3.99651557e-01
-2.35352039e-01 -6.45064563e-02 -2.16830269e-01 1.01998305e+00
-4.23073590e-01 4.65599995e-04 5.71850836e-01 -4.73836670e-03
-1.58504218e-01 4.57241982e-01 -2.30849549e-01 -1.57986388e-01
6.72584653e-01 -5.52070774e-02 -1.49625055e-02 -5.19723237e-01
-7.23034143e-01 -8.55006836e-03 3.24814916e-02 -1.98908702e-01
8.88359189e-01 -1.55384195e+00 -1.01005328e+00 4.20137942e-02
-8.26900974e-02 2.04095487e-02 6.52124822e-01 1.31856191e+00
-4.99328315e-01 4.12110478e-01 3.51866990e-01 -3.91757548e-01
-1.29028344e+00 6.39924467e-01 -4.04947668e-01 -6.49493456e-01
-7.28113711e-01 7.32494652e-01 1.66387230e-01 5.27575985e-02
3.47318426e-02 -1.65763706e-01 5.63482121e-02 -5.40185347e-02
9.11427259e-01 8.84515226e-01 1.25621915e-01 -7.30911434e-01
1.96344350e-02 5.55655122e-01 -1.37225702e-01 -5.81889711e-02
1.46687043e+00 -5.47682703e-01 -7.65195072e-01 3.15522164e-01
1.35327029e+00 4.53092813e-01 -7.53944039e-01 -5.52118063e-01
-1.75064519e-01 -9.38207924e-01 5.74108183e-01 1.69344665e-03
-1.03481328e+00 6.13978922e-01 2.23790079e-01 2.39932194e-01
1.24587500e+00 -4.76141930e-01 1.23067772e+00 4.54068929e-01
6.53054595e-01 -7.85608828e-01 5.10052443e-02 3.91335487e-01
1.03170776e+00 -6.85540080e-01 7.52549767e-01 -4.39265847e-01
-3.09658557e-01 1.14699936e+00 -1.79748133e-01 -2.55312353e-01
5.78308344e-01 -1.88160568e-01 -4.06383485e-01 -2.24332958e-02
-5.72714269e-01 3.53613906e-02 1.82662874e-01 1.91326916e-01
2.49316230e-01 4.40734327e-02 -9.05059338e-01 4.68478352e-01
1.52258441e-01 -1.03208767e-02 5.38902998e-01 6.47955239e-01
-1.68969274e-01 -9.01697218e-01 -9.75883842e-01 6.05346859e-01
-5.43006718e-01 -1.37703761e-01 3.46650362e-01 1.68634638e-01
-4.39259022e-01 9.51232612e-01 -5.85931718e-01 -7.44519770e-01
6.87101111e-02 -4.79366809e-01 2.09730715e-01 -4.06709760e-02
-2.64317721e-01 4.81499344e-01 -6.74597248e-02 -7.43152857e-01
-3.48830938e-01 -7.18807220e-01 -9.55700457e-01 -5.38073599e-01
-3.12512368e-01 4.98780966e-01 5.49392819e-01 7.55358696e-01
4.29597199e-01 2.82294191e-02 1.26512539e+00 -5.68074524e-01
-9.69511688e-01 -6.83309019e-01 -9.79198813e-01 4.19511676e-01
3.43562245e-01 -3.15568030e-01 -4.69188631e-01 -1.48432434e-01] | [7.397461891174316, 4.4796271324157715] |
95a3e563-002b-45ca-9568-6f7a2f05ffef | remasc-realistic-replay-attack-corpus-for | 1904.03365 | null | https://arxiv.org/abs/1904.03365v2 | https://arxiv.org/pdf/1904.03365v2.pdf | ReMASC: Realistic Replay Attack Corpus for Voice Controlled Systems | This paper introduces a new database of voice recordings with the goal of supporting research on vulnerabilities and protection of voice-controlled systems (VCSs). In contrast to prior efforts, the proposed database contains both genuine voice commands and replayed recordings of such commands, collected in realistic VCSs usage scenarios and using modern voice assistant development kits. Specifically, the database contains recordings from four systems (each with a different microphone array) in a variety of environmental conditions with different forms of background noise and relative positions between speaker and device. To the best of our knowledge, this is the first publicly available database that has been specifically designed for the protection of state-of-the-art voice-controlled systems against various replay attacks in various conditions and environments. | ['Christian Poellabauer', 'Jian Yang', 'Yuan Gong', 'Jacob Huber', 'Mitchell MacKnight'] | 2019-04-06 | null | null | null | null | ['voice-anti-spoofing'] | ['audio'] | [ 8.94492120e-02 -4.10884678e-01 2.06820428e-01 9.03936774e-02
-6.37877047e-01 -9.95325565e-01 4.31549430e-01 -8.96092430e-02
2.65144277e-03 4.32927728e-01 1.65159583e-01 -8.85724187e-01
1.81052536e-02 -1.42637268e-01 -2.62350887e-01 -5.79963028e-01
-1.18531108e-01 -2.04313412e-01 4.89484221e-01 -2.11216122e-01
2.23533273e-01 9.68860567e-01 -1.61808562e+00 -1.54716700e-01
5.39676324e-02 8.55726004e-01 6.95741400e-02 1.09093750e+00
1.48906246e-01 3.46549153e-01 -1.36796856e+00 -2.19008029e-01
1.74870461e-01 -2.09710568e-01 -4.80185300e-01 -4.16258313e-02
3.36101770e-01 -2.35836461e-01 -5.01828551e-01 8.98098767e-01
1.04132271e+00 6.32289201e-02 3.39176477e-04 -1.02538097e+00
-2.91405320e-01 5.61137497e-01 -2.09065713e-02 6.57933772e-01
7.15775907e-01 3.09572637e-01 6.34108782e-01 -5.67356348e-01
1.53942987e-01 9.49139535e-01 2.90502489e-01 6.93706989e-01
-1.15798187e+00 -8.94512236e-01 -4.61501986e-01 -1.41111538e-01
-1.31915736e+00 -8.71514320e-01 8.78720105e-01 -2.92743504e-01
1.17093194e+00 8.01163137e-01 1.42955020e-01 1.40748906e+00
-3.76238860e-02 8.07962492e-02 7.91381001e-01 -5.72138429e-01
3.18802685e-01 6.36834145e-01 2.39236951e-01 8.83476213e-02
8.52195024e-02 1.15255624e-01 -4.27571326e-01 -8.41514409e-01
1.16362385e-01 -5.54564357e-01 -8.94266307e-01 -1.06880784e-01
-9.38720822e-01 3.70677978e-01 -5.40794253e-01 6.20407701e-01
-3.12975287e-01 -2.56079912e-01 5.06352007e-01 4.35389638e-01
-1.25180021e-01 3.49569738e-01 -5.56136072e-01 -4.11205322e-01
-6.60861969e-01 5.77812567e-02 1.32491255e+00 1.18183768e+00
-1.40850142e-01 7.68464863e-01 2.44823053e-01 9.03053761e-01
5.11143863e-01 5.96749485e-01 3.99057060e-01 -3.26736689e-01
6.99398398e-01 -4.76450294e-01 2.49745145e-01 -8.49714875e-01
8.32431540e-02 -2.79605806e-01 -1.91361457e-01 2.11496651e-02
3.18789333e-01 -6.06354237e-01 -4.07919109e-01 1.35160577e+00
3.32910240e-01 5.17420053e-01 2.27492601e-01 4.53960896e-01
6.64819241e-01 5.93252957e-01 -4.32157844e-01 -5.71946681e-01
1.29915798e+00 -4.90496308e-01 -1.10008287e+00 3.11636773e-04
-7.59280846e-02 -1.34029186e+00 1.22243035e+00 7.47575760e-01
-7.13528931e-01 -3.47681731e-01 -1.32931781e+00 6.42931104e-01
-1.49952456e-01 -2.99383312e-01 -1.35085851e-01 1.85815954e+00
-8.98977637e-01 -6.55619130e-02 -3.23639840e-01 -4.38331664e-01
-2.02227265e-01 3.36582541e-01 -3.11113745e-01 5.58123350e-01
-1.07033396e+00 6.11681342e-01 -5.11808097e-01 -2.85582952e-02
-8.95626664e-01 -5.14241278e-01 -4.59405243e-01 -1.28965572e-01
4.72898722e-01 1.07548870e-01 1.72464824e+00 -6.82661161e-02
-2.01161623e+00 3.75454187e-01 -1.63019717e-01 -3.65794390e-01
1.68457314e-01 -4.35713798e-01 -1.36487627e+00 6.41603842e-02
-4.62175697e-01 -7.52174854e-01 1.20282495e+00 -1.26435351e+00
-1.54175028e-01 -2.56280661e-01 -3.13876212e-01 -3.85530949e-01
-3.30832034e-01 9.09430325e-01 -5.87190911e-02 -6.13902509e-01
-2.61304140e-01 -8.61173689e-01 2.61049241e-01 -7.29000807e-01
-9.80974019e-01 3.46790433e-01 1.19438982e+00 -5.79131246e-01
1.50167155e+00 -2.37759328e+00 -3.33286524e-01 3.10816914e-01
-3.79278690e-01 1.02147019e+00 2.59801537e-01 9.30501342e-01
-1.37312472e-01 4.56698805e-01 2.26486668e-01 -3.26806486e-01
-1.18646018e-01 7.20085129e-02 -9.23503518e-01 6.04511082e-01
-3.50588173e-01 -1.76938206e-01 -2.96022505e-01 5.14335092e-03
4.22217369e-01 2.92015433e-01 -2.91036129e-01 4.94067758e-01
3.12399507e-01 5.65521896e-01 -4.55373108e-01 6.55029476e-01
6.50000930e-01 9.60389078e-01 2.18753088e-02 1.84968308e-01
-4.69222069e-01 7.43030012e-01 -1.47142196e+00 7.30740249e-01
-5.00125825e-01 7.72701561e-01 7.28043437e-01 -4.21758771e-01
8.57218146e-01 1.17004311e+00 -1.20838620e-01 2.21558124e-01
3.60644460e-01 2.92633057e-01 1.03211828e-01 -7.59217322e-01
4.27752972e-01 1.64569053e-03 3.81663293e-02 3.60424221e-01
9.25626978e-02 -5.36241591e-01 -4.71910894e-01 -1.32043794e-01
1.07367527e+00 -8.25838685e-01 3.43012065e-01 -7.16795214e-04
9.87427354e-01 -6.04389846e-01 3.25430214e-01 7.87985265e-01
-6.13655865e-01 4.28965271e-01 1.51116192e-01 4.12281841e-01
-7.51504123e-01 -1.03492093e+00 -3.43912840e-01 7.97477067e-01
-3.64481986e-01 -4.77582216e-01 -8.44148219e-01 -1.35608032e-01
-1.46371543e-01 1.05195451e+00 2.28490204e-01 2.36482531e-01
-6.27167225e-01 -1.63371935e-02 1.53434074e+00 1.09473586e-01
2.55902052e-01 -1.01271093e+00 -1.51985496e-01 1.44827470e-01
1.69487551e-01 -1.36759460e+00 -7.29569435e-01 1.23551451e-01
-5.27900159e-01 -1.00719440e+00 -6.46072850e-02 -7.29794979e-01
-1.47725120e-01 4.36645746e-01 6.82502985e-01 -7.37065673e-02
-2.32293218e-01 7.28580892e-01 -3.21508855e-01 -7.51462519e-01
-9.15948808e-01 -2.52690911e-01 6.72010481e-01 1.33354485e-01
1.48301214e-01 -7.07327008e-01 -1.31337911e-01 6.25061750e-01
-6.80633068e-01 -1.22265482e+00 -2.55965650e-01 5.80793321e-01
-3.32593694e-02 5.25716662e-01 5.13530195e-01 -4.59754378e-01
1.12622166e+00 -6.48368001e-01 -6.42402589e-01 -3.00078988e-02
-1.44285396e-01 -5.67582726e-01 8.80287707e-01 -6.36216521e-01
-9.53259408e-01 -2.42753789e-01 -4.75673646e-01 -1.73367992e-01
-9.14364874e-01 1.68824457e-02 -8.20836425e-01 -1.74436495e-01
6.49043798e-01 3.22499067e-01 -8.18926468e-02 -9.84573185e-01
1.46377608e-01 1.60498571e+00 7.98590183e-01 -2.84336507e-01
1.03491592e+00 -7.93248415e-03 -4.95054841e-01 -1.65675855e+00
1.34689510e-01 -6.56299949e-01 -2.07595959e-01 -9.19800177e-02
4.07563925e-01 -4.75898445e-01 -6.83023453e-01 1.30378795e+00
-1.09768009e+00 1.88805521e-01 3.32245618e-01 6.63758099e-01
-1.77429855e-01 6.07371032e-01 -6.03155613e-01 -1.33562636e+00
-3.82771522e-01 -1.18614066e+00 5.94349146e-01 6.55180812e-02
-2.64767885e-01 -6.16006613e-01 7.87396729e-02 5.28957963e-01
5.69161415e-01 -9.92537588e-02 5.92436790e-01 -9.29040492e-01
-7.56217465e-02 -5.43726861e-01 8.33826303e-01 7.95200408e-01
5.63820124e-01 4.52614844e-01 -1.45502186e+00 -3.91784430e-01
7.94347942e-01 -5.22404537e-02 -1.05286784e-01 3.89204025e-02
6.14607632e-01 -4.33628201e-01 5.26458621e-02 2.44477689e-01
1.01386690e+00 7.90144205e-01 7.11621225e-01 9.40180048e-02
3.52320701e-01 4.14049655e-01 3.30022156e-01 5.72143435e-01
-4.53160673e-01 7.57797539e-01 4.04439956e-01 4.59952205e-01
1.93070874e-01 -3.23610045e-02 5.84751844e-01 7.37659037e-01
4.31747854e-01 -8.14943612e-01 -8.25921774e-01 4.38077927e-01
-7.88030446e-01 -9.65173066e-01 -2.63858050e-01 2.34080625e+00
5.18203557e-01 1.15663894e-01 2.86312431e-01 7.43231356e-01
1.17163002e+00 5.66687763e-01 -3.34530413e-01 -8.40931714e-01
1.07269853e-01 5.98107576e-01 4.77663815e-01 6.27935469e-01
-1.06607902e+00 6.60875678e-01 7.37117195e+00 4.28353310e-01
-1.28170431e+00 4.43653837e-02 -1.83209226e-01 -1.98280379e-01
2.67095387e-01 -1.19136699e-01 -7.13573992e-01 3.32044095e-01
1.49465454e+00 -3.21650039e-03 5.67516506e-01 8.51549149e-01
4.30129260e-01 2.69783765e-01 -7.29908466e-01 7.82860160e-01
6.21812344e-02 -7.56894827e-01 -4.20164526e-01 2.75619417e-01
-4.53866227e-03 6.46989346e-02 2.51591057e-01 -9.88866836e-02
-1.09407388e-01 -7.03899145e-01 8.62785518e-01 -2.21983716e-02
5.63246727e-01 -7.91853487e-01 7.38649786e-01 3.34206909e-01
-9.92340922e-01 -8.99711475e-02 2.20534690e-02 -7.74662048e-02
4.74661201e-01 2.27869451e-01 -1.05478597e+00 1.39389768e-01
6.43292964e-01 -2.75404871e-01 -1.91743135e-01 1.03600514e+00
-1.55836821e-01 1.53350925e+00 -3.87177974e-01 -1.92960411e-01
-1.24222256e-01 3.38847488e-01 1.10497975e+00 1.10554934e+00
5.44443369e-01 7.73587972e-02 -3.52371305e-01 3.50932717e-01
3.10639948e-01 1.55682247e-02 -9.88167763e-01 -1.88624129e-01
1.25675631e+00 8.95103693e-01 1.24875657e-01 2.38654181e-01
-4.39802825e-01 6.29704654e-01 -5.04849255e-01 5.03209114e-01
-6.06186926e-01 -8.61851156e-01 1.11581707e+00 1.92681253e-01
3.64670545e-01 -6.51184142e-01 -8.98356661e-02 -7.70659566e-01
1.06143557e-01 -1.11210859e+00 1.36461928e-01 -5.14546692e-01
-1.04080570e+00 7.98376679e-01 -2.30504666e-03 -1.20941472e+00
-3.29353273e-01 -5.51355302e-01 -1.00932086e+00 1.05826294e+00
-7.95741320e-01 -3.88272613e-01 1.74120799e-01 7.63696551e-01
4.00097460e-01 -8.11448991e-01 1.12214100e+00 2.80516922e-01
-7.93004930e-01 8.20948064e-01 2.75924027e-01 -3.35956104e-02
6.85142040e-01 -9.09163535e-01 6.75154030e-01 1.03916013e+00
7.26402849e-02 1.29794073e+00 1.14775848e+00 -2.78325170e-01
-1.74536896e+00 -5.50352693e-01 7.26487458e-01 -2.78388828e-01
8.85666013e-01 -6.65246844e-01 -8.90947759e-01 3.79113972e-01
4.82198954e-01 -1.62757084e-01 1.22713244e+00 -1.74868375e-01
-3.09352994e-01 -4.58824970e-02 -1.43828654e+00 4.21172678e-01
4.70297039e-01 -9.98067260e-01 -8.05318594e-01 2.05263253e-02
7.05557048e-01 -2.79520363e-01 -7.27876782e-01 -7.03986287e-02
4.16981369e-01 -1.00727916e+00 8.48532557e-01 -5.05421996e-01
-6.82161391e-01 -4.57930267e-01 -6.74105644e-01 -1.24284506e+00
2.19332814e-01 -1.47829270e+00 -4.66445200e-02 1.94766009e+00
3.24171156e-01 -9.96218562e-01 2.77561218e-01 4.85184222e-01
-1.60818860e-01 -1.74809456e-01 -1.21918082e+00 -8.86561632e-01
-2.45971322e-01 -9.24986482e-01 8.97426188e-01 6.12984180e-01
2.35390440e-01 2.03320682e-01 -7.52825379e-01 9.26796198e-01
2.31050447e-01 -6.69402242e-01 1.04676127e+00 -7.67756641e-01
-4.36152220e-01 -2.28442252e-03 -6.03212416e-01 -7.68973351e-01
-1.03528453e-02 -2.56049782e-01 1.02874488e-02 -5.75854659e-01
-7.76733220e-01 -8.58680159e-03 1.90470275e-02 -7.81719238e-02
2.41361424e-01 -1.53298154e-01 1.76025361e-01 6.13694498e-03
4.32033300e-01 2.05852047e-01 3.33242148e-01 9.09898505e-02
-3.26130688e-01 6.82533741e-01 -4.40952033e-01 7.71182358e-01
1.19548666e+00 -5.36151230e-01 -6.75235152e-01 1.57822464e-02
-6.78355336e-01 3.22331041e-01 1.60645649e-01 -1.01124036e+00
1.90910771e-01 -2.28052199e-01 -4.43049848e-01 -5.29253781e-01
2.77086049e-01 -9.80913281e-01 3.60642135e-01 1.72375545e-01
-8.60264897e-02 2.53931321e-02 4.85648394e-01 5.69271266e-01
-2.57233232e-01 -3.99746478e-01 7.09557950e-01 3.29161316e-01
-1.66253507e-01 -3.34221303e-01 -1.12273669e+00 -2.40307033e-01
7.99040616e-01 -2.70733058e-01 -3.90805960e-01 -5.93749821e-01
-6.26556396e-01 -4.50719357e-01 2.24044591e-01 5.59413373e-01
6.46503150e-01 -9.05121446e-01 -1.48635358e-01 6.82224870e-01
-1.35147572e-01 -7.12513149e-01 1.44341514e-01 4.53765571e-01
-4.96083438e-01 6.51399612e-01 6.78940415e-02 -2.51393974e-01
-1.96522331e+00 5.77056229e-01 4.08085287e-01 4.58492756e-01
-3.71138245e-01 7.12951362e-01 -5.09676576e-01 -4.20003921e-01
6.43453717e-01 -1.84513211e-01 -2.33945832e-01 -3.48288566e-01
7.76261210e-01 6.23039126e-01 4.21593636e-01 -8.89271319e-01
-6.83167756e-01 1.51297987e-01 9.31596681e-02 -6.33588791e-01
7.07406580e-01 -2.83421308e-01 1.39393518e-02 6.32883549e-01
1.13369191e+00 1.05445755e+00 -4.49144930e-01 -9.90614891e-02
3.57429758e-02 -8.52962673e-01 5.27381562e-02 -5.76739728e-01
-8.76721501e-01 6.46420419e-01 7.46008694e-01 5.32608032e-01
9.41233873e-01 -3.57207805e-01 9.45811093e-01 4.64211375e-01
7.11115003e-01 -7.77456462e-01 -1.41070858e-01 5.54066658e-01
8.47335994e-01 -5.37909627e-01 -4.54851270e-01 -6.21870518e-01
-6.71672523e-01 7.86605120e-01 2.18337402e-01 1.75910085e-01
1.05323958e+00 6.74447954e-01 5.44965267e-01 2.44759619e-01
-4.97812122e-01 2.45132238e-01 4.54065427e-02 1.07778394e+00
4.68765259e-01 -5.90694649e-03 -9.67680290e-02 7.10762978e-01
-6.03810012e-01 -4.17029083e-01 8.15220475e-01 1.06413674e+00
-3.89917910e-01 -1.29616129e+00 -9.61068690e-01 2.76313741e-02
-1.13359737e+00 -1.53436959e-01 -8.42554748e-01 5.40379107e-01
-2.62405157e-01 2.01915026e+00 -4.80484098e-01 -8.07474971e-01
7.78487325e-01 2.72946298e-01 -1.45373926e-01 -5.07897794e-01
-9.93988216e-01 1.93932146e-01 5.99013686e-01 -2.45721832e-01
9.09556299e-02 -8.44059944e-01 -9.99371469e-01 -4.88374472e-01
-7.75867820e-01 4.17086124e-01 9.15473700e-01 4.30359811e-01
3.44950438e-01 4.73018914e-01 1.00531507e+00 -6.71244740e-01
-9.98467267e-01 -1.11288834e+00 -1.16610110e+00 9.96518321e-03
8.16787124e-01 -5.42598486e-01 -9.09033597e-01 -3.87106448e-01] | [14.07616138458252, 5.871841907501221] |
eb932b03-f124-41c8-9bba-2f7b72b90c4e | small-footprint-keyword-spotting-using-deep | 1709.03665 | null | http://arxiv.org/abs/1709.03665v1 | http://arxiv.org/pdf/1709.03665v1.pdf | Small-footprint Keyword Spotting Using Deep Neural Network and Connectionist Temporal Classifier | Mainly for the sake of solving the lack of keyword-specific data, we propose
one Keyword Spotting (KWS) system using Deep Neural Network (DNN) and
Connectionist Temporal Classifier (CTC) on power-constrained small-footprint
mobile devices, taking full advantage of general corpus from continuous speech
recognition which is of great amount. DNN is to directly predict the posterior
of phoneme units of any personally customized key-phrase, and CTC to produce a
confidence score of the given phoneme sequence as responsive decision-making
mechanism. The CTC-KWS has competitive performance in comparison with purely
DNN based keyword specific KWS, but not increasing any computational
complexity. | ['Jun Zhou', 'Zhiming Wang', 'Xiaolong Li'] | 2017-09-12 | null | null | null | null | ['small-footprint-keyword-spotting'] | ['speech'] | [ 2.30083227e-01 -1.46286100e-01 -4.51516360e-01 -1.46425933e-01
-9.10715580e-01 -3.66769791e-01 4.58622247e-01 -4.49368745e-01
-6.72127843e-01 8.03422689e-01 3.45862031e-01 -8.16033602e-01
-1.59106866e-01 -3.18921804e-01 -3.86408031e-01 -5.98606288e-01
2.61227548e-01 3.29387069e-01 3.40355098e-01 1.10919811e-02
9.44378674e-02 3.22089374e-01 -1.12898004e+00 3.35045666e-01
4.34770286e-01 1.39890969e+00 7.03801930e-01 1.09353971e+00
8.06638822e-02 4.32718843e-01 -5.45799732e-01 -2.45616600e-01
-2.35294148e-01 -2.12648764e-01 -5.67054451e-01 -4.66749609e-01
-3.00370723e-01 -2.46485934e-01 -5.22791743e-01 7.55891323e-01
1.08242178e+00 2.76392072e-01 5.17445683e-01 -9.68537033e-01
-4.72834408e-01 8.00992846e-01 1.78891182e-01 5.78474104e-01
1.74838364e-01 1.18325815e-01 1.03287208e+00 -1.01206243e+00
1.71452880e-01 9.60315585e-01 5.87642133e-01 8.33339036e-01
-5.54652870e-01 -4.26134318e-01 2.16444358e-01 5.00365555e-01
-1.68516326e+00 -6.85934365e-01 5.55283010e-01 1.05299450e-01
1.73752213e+00 4.39652681e-01 7.05219030e-01 1.85407352e+00
2.27458850e-01 1.03268325e+00 4.81932938e-01 -4.98715371e-01
4.32930917e-01 1.80933326e-01 -1.28509209e-01 2.88208574e-01
-5.34589767e-01 5.13481125e-02 -9.82449055e-01 -1.26117036e-01
2.25219339e-01 5.36389835e-02 -3.67489576e-01 4.30087030e-01
-1.35605264e+00 4.45245951e-01 -4.04610753e-01 4.63819027e-01
-4.32478219e-01 2.74856895e-01 4.45129186e-01 1.90785408e-01
3.27840596e-01 1.50226906e-01 -1.14220667e+00 -1.14410567e+00
-1.12303758e+00 -2.20081195e-01 8.53415251e-01 1.17836642e+00
-4.26486274e-03 3.47459733e-01 -5.44541597e-01 8.49861443e-01
3.78424339e-02 5.21611333e-01 1.10612142e+00 -4.98770505e-01
5.44059038e-01 2.82607730e-02 -6.98827505e-02 -2.91806340e-01
-2.80471087e-01 -3.99060667e-01 -7.03993618e-01 -6.66083038e-01
-2.81665146e-01 -4.83937949e-01 -8.54865611e-01 1.41357672e+00
-1.12822250e-01 3.51864547e-01 1.60852939e-01 4.61296588e-01
3.49916905e-01 1.06791520e+00 -1.01220302e-01 -5.04123390e-01
1.22645319e+00 -1.00011325e+00 -8.65425527e-01 -1.36517882e-01
6.46130502e-01 -4.30423677e-01 1.23862851e+00 5.85862756e-01
-7.11986661e-01 -2.97507793e-01 -9.73163009e-01 1.24452047e-01
-5.14108658e-01 5.04889846e-01 5.20965576e-01 8.49255860e-01
-1.21940041e+00 4.84809369e-01 -7.01949239e-01 -4.16142106e-01
7.21966252e-02 5.16271651e-01 -3.50735127e-03 5.64831793e-01
-1.52121449e+00 6.65185988e-01 6.51187062e-01 2.87679851e-01
-9.51029181e-01 -5.43042243e-01 -3.49033117e-01 3.86276692e-01
6.41174257e-01 -3.90846193e-01 1.68157554e+00 -5.79590619e-01
-2.15474415e+00 2.55116880e-01 -4.41679627e-01 -6.93072438e-01
4.07552302e-01 -1.88974693e-01 -1.20636654e+00 -1.20273590e-01
-4.70079184e-01 6.22099340e-01 8.84306014e-01 -2.01944172e-01
-9.84388113e-01 3.74452062e-02 -5.64664721e-01 4.53100294e-01
-9.98510122e-01 -3.06059141e-02 -8.70373905e-01 -6.05502307e-01
-4.19010967e-01 -8.04411829e-01 2.88977623e-02 -4.59968120e-01
-7.53037572e-01 -8.45708609e-01 9.32214439e-01 -9.08972502e-01
1.98891926e+00 -2.30649590e+00 -2.80827045e-01 1.56323798e-02
-3.93046707e-01 6.15453482e-01 8.43933299e-02 4.63711023e-01
1.86813056e-01 5.34649678e-02 1.73626080e-01 -4.21468556e-01
7.08979517e-02 2.48897418e-01 -4.90651250e-01 5.45167103e-02
1.33014947e-01 1.17360127e+00 -5.34550965e-01 -3.63068640e-01
1.61635756e-01 2.58788407e-01 -2.30360344e-01 7.01730773e-02
-4.56465036e-01 -1.31188154e-01 -3.54932785e-01 5.88858366e-01
6.01541437e-02 6.74189478e-02 2.90254831e-01 1.35034963e-01
2.17960756e-02 8.49900186e-01 -9.46163416e-01 1.56880629e+00
-7.44986117e-01 7.31082022e-01 -4.38703328e-01 -1.13136244e+00
6.59622550e-01 8.64421308e-01 9.84099135e-02 -8.01857948e-01
8.24543089e-02 2.87848324e-01 -1.97422296e-01 -4.75962281e-01
4.73152429e-01 4.06238794e-01 -1.25250235e-01 3.62530679e-01
5.19700833e-02 4.83815104e-01 -4.72088009e-01 4.72596437e-02
1.22377288e+00 -9.34911892e-02 1.23298898e-01 7.19915479e-02
3.61462116e-01 -5.12561798e-01 5.87827146e-01 7.97028959e-01
-1.37726501e-01 4.32382345e-01 2.91538984e-01 -1.01419777e-01
-1.01849830e+00 -6.88343346e-01 9.65278894e-02 1.22439253e+00
-2.98680782e-01 -4.06602681e-01 -6.87812388e-01 -5.36839366e-01
-2.66111374e-01 8.87025237e-01 -1.60063580e-01 -2.06024066e-01
-4.05168653e-01 -4.42435473e-01 8.18271935e-01 6.65425301e-01
3.91606241e-01 -1.39222133e+00 -3.48095834e-01 3.85368347e-01
-1.42023981e-01 -1.42788899e+00 -1.10660386e+00 8.55525315e-01
-6.69930577e-01 -2.27861971e-01 -8.28845203e-01 -9.60299850e-01
-7.30236620e-02 1.06720231e-01 6.88529789e-01 -4.00726587e-01
4.47993577e-02 1.80355772e-01 -3.52321237e-01 -1.87737748e-01
-1.01261087e-01 5.13554513e-01 6.01468265e-01 1.34350300e-01
6.08227134e-01 -7.76807249e-01 -6.58924937e-01 1.69960916e-01
-4.11804974e-01 6.55960012e-03 6.58959091e-01 7.79315591e-01
3.19085956e-01 3.81423295e-01 6.60878360e-01 -3.70974898e-01
1.08092833e+00 -3.99334103e-01 -2.49740735e-01 4.91232544e-01
-9.95571375e-01 -1.01252951e-01 5.76912165e-01 -9.75160480e-01
-9.14261639e-01 -7.70214796e-02 -4.14200515e-01 -3.62496048e-01
8.87124315e-02 4.85628217e-01 -2.73105949e-01 5.27943313e-01
2.43626252e-01 8.83933187e-01 -6.12878323e-01 -7.05736160e-01
2.55484074e-01 1.49791920e+00 7.82138288e-01 -7.89798051e-02
1.38066724e-01 -1.53287500e-01 -6.90235674e-01 -8.27232182e-01
-3.41744542e-01 -4.87790525e-01 -2.73638695e-01 2.47766227e-02
7.21380591e-01 -1.02477527e+00 -1.13357365e+00 4.75868553e-01
-1.25450897e+00 -3.74637306e-01 -5.04405797e-02 5.88303089e-01
-4.40255940e-01 2.71293730e-01 -5.50391912e-01 -1.32285202e+00
-7.40880966e-01 -7.79163659e-01 1.07055175e+00 1.71086520e-01
-4.56628114e-01 -7.24832773e-01 -3.48023564e-01 1.33008108e-01
3.07049125e-01 -7.93763101e-01 7.93928087e-01 -1.05395710e+00
-2.89979339e-01 -6.17030263e-01 2.09170535e-01 6.36672199e-01
-1.00952275e-01 -2.57129788e-01 -1.16674554e+00 -4.66389284e-02
1.89981237e-02 4.41368259e-02 5.80372274e-01 5.11184931e-01
1.76980591e+00 -4.66395169e-01 -6.45768940e-01 3.41056019e-01
1.05575311e+00 9.04804528e-01 5.53051412e-01 -6.07262775e-02
4.64986324e-01 -5.05694970e-02 3.53685290e-01 4.89744604e-01
2.16073886e-01 1.04655361e+00 -2.43605837e-01 3.15127373e-01
-1.90858245e-02 -5.39274991e-01 8.14433813e-01 1.35402369e+00
3.97700548e-01 -1.03220379e+00 -6.56219900e-01 7.51532197e-01
-2.02552414e+00 -7.75267780e-01 4.66240555e-01 2.14089751e+00
9.77729738e-01 5.52441895e-01 2.41232999e-02 3.74162316e-01
7.78249502e-01 -5.54779842e-02 -7.29191124e-01 -6.84082508e-01
-1.33720532e-01 2.89647460e-01 5.98007619e-01 4.36453432e-01
-6.51025116e-01 1.03006053e+00 6.29310989e+00 1.75064445e+00
-1.18335640e+00 4.51328486e-01 6.61781907e-01 -3.76518458e-01
-2.26040170e-01 -3.59113634e-01 -1.16528118e+00 9.76618946e-01
1.78643632e+00 4.04280275e-02 6.44269288e-01 8.54890347e-01
5.49728930e-01 -2.32301667e-01 -1.22472274e+00 1.50421071e+00
-1.74882665e-01 -1.41571724e+00 -1.11363709e-01 7.48478621e-02
2.12510198e-01 1.02197215e-01 1.74364626e-01 4.82053101e-01
3.19638513e-02 -9.57631469e-01 7.01855838e-01 5.88657558e-01
1.24178481e+00 -9.51716840e-01 5.04464090e-01 6.90558434e-01
-1.01973152e+00 -2.45067015e-01 -4.74842712e-02 3.60179059e-02
1.79039419e-01 6.42406821e-01 -1.26910508e+00 2.33549997e-02
8.38244915e-01 3.33581567e-01 7.37269223e-02 7.15664744e-01
-1.06257699e-01 1.13579726e+00 -5.98180830e-01 -7.66845703e-01
3.30178857e-01 3.64616275e-01 4.27802205e-01 1.51274407e+00
6.68396771e-01 -7.89935365e-02 -1.72317699e-01 3.01224709e-01
-1.95030004e-01 -1.29475832e-01 -3.50347936e-01 -5.91632903e-01
9.97079015e-01 1.17935729e+00 -6.47473514e-01 -3.21508527e-01
-2.62657143e-02 1.68942642e+00 -1.05394367e-02 3.80084753e-01
-5.72039843e-01 -5.29596508e-01 5.15877426e-01 -2.57529557e-01
7.81532884e-01 -2.67550498e-01 -2.40198568e-01 -8.72940838e-01
3.23643297e-01 -5.54477215e-01 -3.99518460e-02 -9.13243651e-01
-8.46402526e-01 5.39249718e-01 -4.31734741e-01 -1.11090899e+00
-4.12011802e-01 -5.71858704e-01 -6.02299571e-01 1.06972468e+00
-1.10374153e+00 -7.29581714e-01 3.42806637e-01 5.37683725e-01
9.61881340e-01 -5.33857405e-01 9.40289021e-01 4.01711285e-01
-7.49295652e-01 9.91988361e-01 4.04410094e-01 -4.66640517e-02
5.27572155e-01 -1.04321039e+00 8.57463360e-01 8.06025505e-01
2.28481770e-01 4.77287292e-01 3.23659629e-01 -5.84326744e-01
-1.42837226e+00 -8.72436941e-01 1.47471499e+00 -5.60309768e-01
5.48938751e-01 -9.27209139e-01 -3.49007845e-01 1.77135646e-01
1.79663137e-01 -2.14000478e-01 5.11930943e-01 2.91946959e-02
1.15319453e-01 -2.65385270e-01 -6.44005537e-01 9.51905191e-01
1.23391318e+00 -1.06067002e+00 -1.81151584e-01 4.83746618e-01
1.33739591e+00 -1.74535215e-01 -5.54425776e-01 1.66695610e-01
5.53831458e-01 -4.41485703e-01 8.08685660e-01 -4.55771834e-01
-1.98283613e-01 -8.08272436e-02 -2.68244892e-01 -1.16597307e+00
-6.66213483e-02 -1.30947804e+00 -5.93877554e-01 1.16758704e+00
8.65286529e-01 -4.61880207e-01 7.73832440e-01 4.16733354e-01
-3.34269077e-01 -8.50431800e-01 -1.48079050e+00 -8.93011093e-01
-6.22222602e-01 -1.15853965e+00 5.61558425e-01 5.82720160e-01
3.36946070e-01 6.72744870e-01 -9.31914032e-01 2.36740801e-02
-2.30282053e-01 -5.84608436e-01 -1.43377800e-02 -7.66157627e-01
-5.04809141e-01 -4.02406275e-01 -3.64603885e-02 -1.51365042e+00
-6.20561279e-03 -7.16324508e-01 2.80780196e-01 -1.08137369e+00
-5.67223737e-03 -1.11001872e-01 -5.52821100e-01 5.45568287e-01
1.62421286e-01 -1.82636842e-01 -2.64949381e-01 -3.77184004e-02
-7.28353322e-01 6.83049679e-01 6.45846963e-01 -1.68734819e-01
-4.07242894e-01 4.29974437e-01 -5.83999217e-01 2.79595107e-01
7.04138815e-01 -4.02028263e-01 -6.99641109e-01 -5.78512475e-02
3.96738499e-01 4.19406742e-01 -1.49988890e-01 -1.08235419e+00
7.06916690e-01 1.13015873e-02 2.45022967e-01 -9.34029341e-01
5.75902164e-01 -8.02868247e-01 1.16419373e-02 4.11291152e-01
-5.63004017e-01 1.10104829e-01 2.61764936e-02 9.84597862e-01
7.81067088e-02 -2.23844647e-01 1.84056193e-01 1.65283650e-01
-9.94664550e-01 4.30105776e-01 -9.86930430e-01 -2.82869875e-01
7.10251987e-01 -4.81310546e-01 1.70843437e-01 -5.58098137e-01
-7.78477073e-01 1.01712808e-01 -5.54665208e-01 8.10045600e-01
6.93609059e-01 -1.40726483e+00 -1.01332270e-01 1.45497650e-01
5.54039627e-02 -2.85991699e-01 1.81127116e-01 5.51858962e-01
3.60109657e-02 1.22693670e+00 5.51580369e-01 -2.73948580e-01
-1.14179540e+00 3.83254170e-01 4.10805345e-01 -1.15685977e-01
-5.97024322e-01 1.11103570e+00 -3.79262000e-01 -2.57780105e-01
9.35356736e-01 -5.38649738e-01 -1.56522378e-01 -1.03827879e-01
4.91946757e-01 2.23997176e-01 4.93705869e-01 1.39019385e-01
-6.51116788e-01 -7.61028454e-02 -1.40365556e-01 -6.42769039e-01
9.07393754e-01 -2.97138900e-01 4.71320599e-01 6.15061641e-01
1.19434249e+00 -3.30371797e-01 -1.03167737e+00 -5.31760752e-01
2.27389976e-01 3.27239871e-01 4.89968449e-01 -1.09682941e+00
-7.06099331e-01 8.20757151e-01 8.61293614e-01 -1.63693488e-01
1.02317560e+00 -1.95635140e-01 1.43902338e+00 7.17759550e-01
2.28857577e-01 -1.82840633e+00 1.02118164e-01 7.63027668e-01
4.76650208e-01 -8.41197312e-01 -5.45728087e-01 2.37772062e-01
-8.60327780e-01 1.11431503e+00 4.15032983e-01 5.27534306e-01
9.53044236e-01 4.50306267e-01 -4.28116739e-01 2.31099755e-01
-1.26424944e+00 1.59195773e-02 1.28687724e-01 6.46985233e-01
-5.91694005e-02 2.74972796e-01 -2.29223222e-01 9.68968272e-01
-1.87733173e-01 2.08587453e-01 5.08970134e-02 6.68812633e-01
-3.11551780e-01 -9.10410106e-01 2.44131014e-01 5.44201672e-01
-4.50925678e-01 -7.14061916e-01 -1.36374339e-01 -1.52977660e-01
-4.11994047e-02 1.00645947e+00 1.80515379e-01 -1.03372931e+00
1.08667500e-01 4.11048114e-01 -1.10855475e-01 -4.19516236e-01
-5.31408250e-01 2.94152379e-01 3.42772543e-01 -3.37549716e-01
1.46193758e-01 -5.09142280e-01 -9.27529395e-01 -3.07842225e-01
-3.92585188e-01 1.47958592e-01 8.95006895e-01 1.26339054e+00
6.14938498e-01 4.99840796e-01 7.05695510e-01 -4.82039511e-01
-6.44810021e-01 -1.23441291e+00 -6.20722413e-01 -3.96733463e-01
2.73068756e-01 -9.29612368e-02 -1.20336577e-01 -1.85857847e-01] | [14.362648963928223, 6.3947224617004395] |
c5891514-9b34-4914-85d4-52e0a21b5c61 | towards-generating-adversarial-examples-on | 2210.09405 | null | https://arxiv.org/abs/2210.09405v1 | https://arxiv.org/pdf/2210.09405v1.pdf | Towards Generating Adversarial Examples on Mixed-type Data | The existence of adversarial attacks (or adversarial examples) brings huge concern about the machine learning (ML) model's safety issues. For many safety-critical ML tasks, such as financial forecasting, fraudulent detection, and anomaly detection, the data samples are usually mixed-type, which contain plenty of numerical and categorical features at the same time. However, how to generate adversarial examples with mixed-type data is still seldom studied. In this paper, we propose a novel attack algorithm M-Attack, which can effectively generate adversarial examples in mixed-type data. Based on M-Attack, attackers can attempt to mislead the targeted classification model's prediction, by only slightly perturbing both the numerical and categorical features in the given data samples. More importantly, by adding designed regularizations, our generated adversarial examples can evade potential detection models, which makes the attack indeed insidious. Through extensive empirical studies, we validate the effectiveness and efficiency of our attack method and evaluate the robustness of existing classification models against our proposed attack. The experimental results highlight the feasibility of generating adversarial examples toward machine learning models in real-world applications. | ['Hao Yang', 'Mahashweta Das', 'Xiaoting Li', 'Huiyuan Chen', 'Zhimeng Jiang', 'Menghai Pan', 'Han Xu'] | 2022-10-17 | null | null | null | null | ['type'] | ['speech'] | [ 1.82446763e-01 1.36042222e-01 3.89840342e-02 -1.00308463e-01
-5.20874858e-01 -9.36380923e-01 5.78133881e-01 2.09237963e-01
-9.96678174e-02 7.72778213e-01 -3.64246398e-01 -5.25747836e-01
1.11096300e-01 -1.21625781e+00 -7.98423350e-01 -8.45850289e-01
-1.77511007e-01 1.10327184e-01 -1.86295718e-01 -2.91166604e-01
2.76162624e-01 5.85776269e-01 -9.65765595e-01 2.35596150e-01
1.00772882e+00 9.34419811e-01 -7.75925279e-01 5.37776589e-01
1.23646490e-01 9.29831922e-01 -1.08532274e+00 -1.05990660e+00
5.74631333e-01 -3.19168091e-01 -3.65928322e-01 -1.42393261e-01
3.18122238e-01 -5.92197180e-01 -4.34115857e-01 1.41168106e+00
3.71897787e-01 -1.47950545e-01 6.77889526e-01 -1.81003952e+00
-6.87633157e-01 5.81165314e-01 -4.50706840e-01 -6.49303838e-04
1.98457211e-01 5.90310156e-01 6.07505500e-01 -6.35172665e-01
3.36344503e-02 1.37218547e+00 5.68792939e-01 8.81072998e-01
-9.54867661e-01 -1.17747509e+00 1.65459201e-01 -1.47811115e-01
-1.20068741e+00 -1.03344068e-01 1.04120553e+00 -3.77982169e-01
1.44397616e-01 6.79930866e-01 3.51683915e-01 1.34373784e+00
3.59461755e-01 8.66879046e-01 1.01775324e+00 -5.52577805e-03
4.30868983e-01 4.63549197e-01 -2.50952214e-01 4.81861383e-01
4.38992232e-01 5.55726230e-01 -3.17784608e-03 -9.05192435e-01
5.26448309e-01 3.55370194e-01 -1.72004029e-01 -4.78116870e-02
-9.21698093e-01 1.15740764e+00 4.30487841e-01 -1.79988369e-01
-2.48072997e-01 -3.09327412e-02 5.86339116e-01 3.84358495e-01
4.10819709e-01 5.17231166e-01 -3.71199429e-01 3.70284230e-01
-1.75194100e-01 4.65030581e-01 7.65055716e-01 7.95674384e-01
2.05423340e-01 5.09548008e-01 -6.33436367e-02 4.78484452e-01
1.82582200e-01 7.25132048e-01 4.88452971e-01 -4.56109226e-01
7.40428984e-01 5.43146551e-01 2.06526443e-01 -1.44793355e+00
2.63231248e-02 -5.05530238e-01 -1.19945002e+00 3.84687454e-01
5.11660039e-01 -6.10090554e-01 -4.64696378e-01 1.60174596e+00
5.73278129e-01 4.47168291e-01 4.44612920e-01 6.82791829e-01
4.42185670e-01 6.35718584e-01 1.05836712e-01 -2.93960035e-01
8.64495754e-01 -5.05771101e-01 -5.52100599e-01 -1.54168338e-01
7.15189040e-01 -6.71624780e-01 9.55699682e-01 3.60268414e-01
-7.66129017e-01 -2.59091109e-01 -1.05182743e+00 7.51959860e-01
-3.48125100e-01 -3.31234485e-01 6.72355711e-01 1.10009587e+00
-2.36633956e-03 5.65889239e-01 -4.50257003e-01 5.26595831e-01
6.97121263e-01 3.05224836e-01 -2.47387961e-01 -3.32747330e-03
-1.67798162e+00 4.44678009e-01 2.67735809e-01 8.68004933e-02
-9.48853254e-01 -7.70660400e-01 -7.30057299e-01 -1.47157729e-01
3.39751840e-01 -4.28585261e-01 9.48618293e-01 -9.70090985e-01
-9.85094309e-01 5.31791925e-01 3.04768831e-01 -7.22286284e-01
1.10027742e+00 -5.47243692e-02 -7.82382011e-01 -5.93641400e-02
-1.94022715e-01 -3.55831236e-02 1.20450568e+00 -1.28412175e+00
-5.35847485e-01 -3.68149966e-01 2.16082171e-01 -1.67796016e-01
-6.59591556e-01 -6.54244125e-02 4.61507171e-01 -1.25437367e+00
-4.08402443e-01 -7.79891133e-01 -5.65583825e-01 8.40495974e-02
-6.87085509e-01 8.34038779e-02 1.09710503e+00 -3.57554108e-01
1.18670940e+00 -2.41804028e+00 -3.74237716e-01 4.48933631e-01
3.21058512e-01 6.86613321e-01 3.96171212e-02 2.38354549e-01
-1.64029166e-01 4.74028438e-01 -5.14694691e-01 1.06480904e-01
-1.44086152e-01 1.12284340e-01 -1.01456070e+00 5.41383147e-01
4.00282741e-01 7.96252787e-01 -9.61539209e-01 -1.37304932e-01
7.69788399e-02 1.33925319e-01 -5.81801653e-01 4.46882904e-01
-1.12547897e-01 4.33416337e-01 -8.98921013e-01 7.26554155e-01
8.89034867e-01 1.35943726e-01 -1.18946113e-01 1.39826685e-01
6.94262385e-01 -1.66039228e-01 -1.13458681e+00 4.09990579e-01
-2.92773724e-01 2.10996196e-01 -2.16927931e-01 -1.01212811e+00
1.06641412e+00 2.48432755e-01 1.11643381e-01 -3.52319390e-01
1.15743980e-01 2.25673854e-01 3.03592652e-01 -3.65964681e-01
6.67163581e-02 -2.88931757e-01 -3.95897597e-01 5.30726850e-01
-7.34070122e-01 3.65956686e-02 -4.23470914e-01 1.57433584e-01
1.05579233e+00 -5.84365666e-01 2.26455003e-01 1.04858287e-01
7.46874690e-01 -8.91892985e-02 7.52996385e-01 8.70570302e-01
-3.38124663e-01 2.01106623e-01 6.54802024e-01 -5.76267958e-01
-8.56429338e-01 -1.07164836e+00 -5.38487136e-02 4.69742775e-01
1.19574904e-01 -5.65367676e-02 -6.69323623e-01 -1.40637648e+00
4.18439180e-01 7.81414032e-01 -6.34267807e-01 -7.31501460e-01
-6.11526966e-01 -9.90050972e-01 1.03585541e+00 3.22695255e-01
4.91472721e-01 -1.03082025e+00 -1.25568151e-01 2.48094723e-01
2.32352063e-01 -9.81398940e-01 -4.45963979e-01 -3.05133998e-01
-6.52181745e-01 -1.37708294e+00 -4.03129131e-01 -4.67270076e-01
1.04363036e+00 -4.25569266e-02 7.55302906e-01 3.73437554e-01
-3.56573045e-01 -6.39967769e-02 -3.42198938e-01 -6.78593636e-01
-9.90322948e-01 -4.28494871e-01 3.13082546e-01 4.29071605e-01
2.48520866e-01 -3.74163866e-01 -4.58323926e-01 3.43541086e-01
-1.16427660e+00 -4.69711423e-01 3.23316306e-01 8.98710907e-01
9.92435813e-02 4.14019287e-01 1.12265265e+00 -1.36464334e+00
8.74678195e-01 -7.94736087e-01 -5.72970808e-01 4.77664396e-02
-4.68120128e-01 -2.70782858e-01 1.48369360e+00 -1.03296983e+00
-6.05736971e-01 -2.74939239e-01 -1.63026869e-01 -7.74678528e-01
-1.53172016e-01 2.29788348e-01 -4.91350174e-01 -3.25235009e-01
6.87034369e-01 3.41667205e-01 7.53970221e-02 -1.64054587e-01
1.51866600e-01 5.79275310e-01 3.98923010e-01 -5.05162358e-01
1.49887371e+00 3.16717356e-01 1.64272875e-01 -4.90752727e-01
-6.33439541e-01 3.34569454e-01 -9.76735950e-02 -1.30469888e-03
2.40371153e-01 -6.44505501e-01 -7.76632726e-01 9.48786974e-01
-9.11777794e-01 8.68307892e-03 2.19916403e-02 2.61306286e-01
-1.52231142e-01 4.32749242e-01 -6.34821057e-01 -9.95191395e-01
-3.48027647e-01 -1.03277624e+00 5.46108007e-01 -6.71653301e-02
-5.09454198e-02 -1.12542069e+00 -3.27310741e-01 3.11004221e-01
2.70131797e-01 9.96719956e-01 1.03125095e+00 -1.27923048e+00
-4.11629885e-01 -8.45785677e-01 1.90848991e-01 6.45532548e-01
3.13558042e-01 1.28362551e-01 -8.22568297e-01 -4.07708108e-01
3.48443151e-01 -3.91780972e-01 3.16251576e-01 -2.07913846e-01
1.63584709e+00 -1.26211333e+00 -8.76367316e-02 5.97644508e-01
1.05040371e+00 4.26739842e-01 5.17239928e-01 1.68883890e-01
7.82028258e-01 5.87079406e-01 9.77997959e-01 7.02426195e-01
-2.44518936e-01 2.47635663e-01 8.50814223e-01 -1.22514568e-01
6.99957728e-01 -3.13740373e-01 4.17492241e-01 2.27507591e-01
3.70968401e-01 -1.57145023e-01 -6.53885365e-01 1.93288580e-01
-1.54270363e+00 -1.16110110e+00 -8.76810104e-02 2.33767676e+00
9.88610327e-01 5.38449824e-01 -1.52470153e-02 4.32660699e-01
8.84442985e-01 1.35018840e-01 -8.67717445e-01 -5.91590345e-01
3.57961543e-02 4.08455580e-02 4.85449463e-01 3.32774520e-01
-1.23086536e+00 5.91412306e-01 5.80716515e+00 1.08384442e+00
-1.16458178e+00 -2.28486046e-01 1.07245636e+00 -1.24034740e-01
-4.14200693e-01 -1.54963911e-01 -6.73189640e-01 9.11117017e-01
6.24578714e-01 -5.82612336e-01 2.62578994e-01 1.19774282e+00
-2.63402727e-03 7.23673642e-01 -1.15612125e+00 6.21913075e-01
-8.20845440e-02 -1.24819767e+00 3.35731268e-01 1.09874755e-01
5.45305133e-01 -6.88604295e-01 4.37797576e-01 5.68606853e-01
5.33151805e-01 -1.30167460e+00 4.26209778e-01 3.92508842e-02
5.31964600e-01 -1.34699094e+00 8.02663028e-01 7.51702309e-01
-8.06028962e-01 -3.13588917e-01 -2.78459758e-01 -1.25014121e-02
-2.06611782e-01 5.28114259e-01 -7.25785911e-01 5.11535645e-01
2.14540571e-01 4.64135408e-01 -4.91118312e-01 5.79609632e-01
-1.25611156e-01 7.63486445e-01 -1.60490096e-01 -1.02087751e-01
2.88440108e-01 -9.86943468e-02 6.35024369e-01 7.59639084e-01
7.98580348e-02 2.22820550e-01 3.13620657e-01 8.14011693e-01
-2.64334470e-01 9.79703888e-02 -9.56590235e-01 -1.22667793e-02
7.89575517e-01 9.75842595e-01 -9.31927711e-02 -3.06153178e-01
-1.07314982e-01 6.27613068e-01 4.36952971e-02 2.26314068e-01
-1.10679495e+00 -4.99954581e-01 7.96246111e-01 5.09755686e-02
-2.13056847e-01 1.62520036e-01 -3.86131972e-01 -1.15681791e+00
2.87537456e-01 -1.48066807e+00 5.19913673e-01 -1.73578575e-01
-1.78390753e+00 4.88639086e-01 -3.24761063e-01 -1.68954039e+00
-3.76495093e-01 -3.69430900e-01 -1.10557735e+00 8.45978439e-01
-9.61177886e-01 -1.06906664e+00 1.27063945e-01 8.59178066e-01
2.11421132e-01 -6.78121805e-01 7.60506749e-01 1.78396940e-01
-8.49508941e-01 1.29331577e+00 1.88636392e-01 6.55507088e-01
5.27118266e-01 -9.49696779e-01 6.97976410e-01 9.90485489e-01
-2.46936560e-01 7.51527190e-01 6.69615567e-01 -7.77565241e-01
-1.36371315e+00 -1.64183772e+00 2.34052435e-01 -4.66727227e-01
9.34098303e-01 -4.53126997e-01 -1.18992472e+00 6.46852136e-01
-4.39063102e-01 3.84130150e-01 7.39910781e-01 -4.30989385e-01
-5.87669134e-01 -1.09558605e-01 -1.84416795e+00 8.76966655e-01
5.35629928e-01 -4.10209984e-01 -4.70514029e-01 6.02863252e-01
8.12413514e-01 -3.44626129e-01 -8.70145023e-01 6.51583254e-01
2.47582421e-01 -7.69294679e-01 1.23955071e+00 -1.10328734e+00
6.06865466e-01 -4.28863987e-02 -1.35215484e-02 -1.20001221e+00
1.09379523e-01 -7.16860116e-01 -4.67455059e-01 1.34252238e+00
2.96051413e-01 -1.25264215e+00 6.69472277e-01 6.56549990e-01
3.79954338e-01 -9.12963092e-01 -8.77418816e-01 -9.24950898e-01
5.09432495e-01 -3.34375560e-01 9.64089572e-01 1.40259206e+00
-2.75830626e-01 -2.40478098e-01 -6.82323933e-01 6.65954947e-01
9.21612322e-01 2.27337645e-04 1.07626641e+00 -9.14741993e-01
-4.48345512e-01 -4.13023114e-01 -4.78970617e-01 -3.19116980e-01
3.05422038e-01 -5.95053077e-01 -4.31758285e-01 -4.36289102e-01
-2.80938298e-01 -8.16698849e-01 -1.67497426e-01 5.03503025e-01
-6.79807901e-01 1.72989026e-01 2.01532513e-01 2.53766745e-01
7.80836567e-02 5.14876604e-01 1.15574074e+00 -2.98867851e-01
6.33211434e-02 5.47542810e-01 -8.30213904e-01 8.53013694e-01
8.85994554e-01 -7.79602945e-01 -4.35292751e-01 2.00128481e-01
9.64434519e-02 2.16197342e-01 5.61468840e-01 -6.83803976e-01
-1.64193898e-01 -6.00776553e-01 4.45479065e-01 -1.26035362e-01
3.20351683e-03 -9.12355423e-01 -5.82678318e-02 1.03396332e+00
-4.43256348e-01 -1.18069075e-01 9.55689922e-02 7.40288854e-01
-1.27921715e-01 -2.51513809e-01 1.03524566e+00 5.41833006e-02
-8.42047855e-02 7.02925146e-01 -1.14522345e-01 2.56179482e-01
1.55847239e+00 2.32700497e-01 -4.60399061e-01 -3.13674420e-01
-5.28658211e-01 3.39378983e-01 4.05997038e-01 5.20278394e-01
8.36249232e-01 -1.52162600e+00 -9.69632030e-01 5.13481617e-01
-3.02817230e-03 7.22744614e-02 1.23899452e-01 2.62532413e-01
-2.85796404e-01 -1.67355672e-01 -5.93939498e-02 -2.02059075e-01
-1.34029317e+00 1.11888409e+00 4.04725075e-01 -2.22983658e-01
-4.54545170e-01 5.44045389e-01 3.90545607e-01 -5.25040746e-01
1.99714914e-01 2.12530762e-01 -1.00953482e-01 -2.28661537e-01
6.76755130e-01 3.41209322e-01 -2.79307812e-01 -2.44235158e-01
-2.57770956e-01 7.00520575e-02 -3.96145046e-01 4.40933853e-01
8.68749619e-01 3.79191607e-01 -4.26170789e-02 6.57471493e-02
1.10800910e+00 2.90007591e-01 -1.09884477e+00 -8.65379050e-02
-3.81765097e-01 -8.79962921e-01 -5.49782932e-01 -4.24655765e-01
-1.26900613e+00 7.63102710e-01 2.65940756e-01 7.35200107e-01
9.66980934e-01 -5.72887421e-01 9.03852880e-01 4.04870987e-01
3.48520011e-01 -4.73651707e-01 1.26647323e-01 1.33622676e-01
8.98983836e-01 -1.21544576e+00 -1.72487423e-01 -4.80201900e-01
-8.45477343e-01 8.46569121e-01 8.67925048e-01 -5.32233179e-01
5.34023821e-01 3.40835094e-01 1.06079988e-01 2.45322287e-01
-6.33799493e-01 8.03336859e-01 6.41768724e-02 6.52555168e-01
-1.42354280e-01 1.66094005e-01 -7.74980932e-02 8.52904379e-01
-3.01269889e-01 -5.39815187e-01 7.67739415e-01 9.31031227e-01
-1.14908732e-01 -1.13379240e+00 -8.25532138e-01 6.37396574e-01
-8.07693005e-01 -8.47323388e-02 -5.00141621e-01 7.38176703e-01
-1.45553991e-01 9.13384438e-01 -1.73493639e-01 -4.89162207e-01
2.36090198e-01 -1.47969410e-01 -2.03098536e-01 -3.07556480e-01
-8.99728775e-01 -4.16005969e-01 -9.70559195e-02 -3.28830630e-01
2.04743266e-01 -4.54237610e-01 -1.11727870e+00 -8.83783519e-01
-1.88422740e-01 2.33672515e-01 1.94531813e-01 9.25512254e-01
1.66578174e-01 3.91841650e-01 1.59081435e+00 -3.29786867e-01
-1.44732440e+00 -6.52957201e-01 -6.15866363e-01 7.93349326e-01
5.13795733e-01 -4.08982038e-01 -9.97930586e-01 -1.62094027e-01] | [5.789521217346191, 7.772523880004883] |
1f7c82d7-54fd-4fac-8dc9-430c3d149d25 | federated-multi-agent-deep-reinforcement | 2301.00641 | null | https://arxiv.org/abs/2301.00641v1 | https://arxiv.org/pdf/2301.00641v1.pdf | Federated Multi-Agent Deep Reinforcement Learning Approach via Physics-Informed Reward for Multi-Microgrid Energy Management | The utilization of large-scale distributed renewable energy promotes the development of the multi-microgrid (MMG), which raises the need of developing an effective energy management method to minimize economic costs and keep self energy-sufficiency. The multi-agent deep reinforcement learning (MADRL) has been widely used for the energy management problem because of its real-time scheduling ability. However, its training requires massive energy operation data of microgrids (MGs), while gathering these data from different MGs would threaten their privacy and data security. Therefore, this paper tackles this practical yet challenging issue by proposing a federated multi-agent deep reinforcement learning (F-MADRL) algorithm via the physics-informed reward. In this algorithm, the federated learning (FL) mechanism is introduced to train the F-MADRL algorithm thus ensures the privacy and the security of data. In addition, a decentralized MMG model is built, and the energy of each participated MG is managed by an agent, which aims to minimize economic costs and keep self energy-sufficiency according to the physics-informed reward. At first, MGs individually execute the self-training based on local energy operation data to train their local agent models. Then, these local models are periodically uploaded to a server and their parameters are aggregated to build a global agent, which will be broadcasted to MGs and replace their local agents. In this way, the experience of each MG agent can be shared and the energy operation data is not explicitly transmitted, thus protecting the privacy and ensuring data security. Finally, experiments are conducted on Oak Ridge national laboratory distributed energy control communication lab microgrid (ORNL-MG) test system, and the comparisons are carried out to verify the effectiveness of introducing the FL mechanism and the outperformance of our proposed F-MADRL. | ['Zhigang Zeng', 'Yang Shi', 'Yang Li', 'Shangyang He', 'Yuanzheng Li'] | 2022-12-29 | null | null | null | null | ['energy-management'] | ['time-series'] | [-1.08207321e+00 -1.35584071e-01 -3.69135197e-03 -1.13485239e-01
-3.84676546e-01 -5.27891278e-01 2.58889496e-01 2.18234405e-01
-4.51514363e-01 1.19140208e+00 -2.28688359e-01 -5.67946620e-02
-2.45075375e-02 -1.27172804e+00 -5.28672993e-01 -1.56377125e+00
-3.81179780e-01 2.58754939e-01 -2.10707173e-01 7.34405518e-02
-2.36268938e-01 2.79636890e-01 -1.19059885e+00 -6.97257221e-01
1.10496998e+00 1.22175717e+00 2.31453553e-01 6.11321367e-02
3.61964554e-01 4.68323827e-01 -8.87526035e-01 2.62601495e-01
4.87219632e-01 -2.68286854e-01 -3.48214567e-01 -1.38129428e-01
-8.23264837e-01 -8.76020789e-01 -2.16026738e-01 1.20038629e+00
6.85926259e-01 4.41010475e-01 2.56589614e-02 -2.03569603e+00
-5.46146631e-01 4.12631005e-01 -4.74790841e-01 -2.78268874e-01
-1.90830037e-01 3.67379248e-01 7.24993587e-01 7.77921453e-02
1.52492732e-01 7.37493694e-01 3.37271690e-02 5.90698838e-01
-7.78346479e-01 -1.07518983e+00 2.86716282e-01 4.33888108e-01
-1.31675255e+00 -2.19367705e-02 7.71492243e-01 2.42772564e-01
8.80398214e-01 2.28864402e-01 9.91212249e-01 5.41465521e-01
4.85563993e-01 6.50294721e-01 1.12508392e+00 -1.09946989e-01
9.79264557e-01 1.82311490e-01 -2.49667719e-01 4.71614301e-01
3.96274954e-01 2.68845379e-01 -1.07239909e-01 -3.57457995e-01
4.47981149e-01 1.86357737e-01 -3.62471312e-01 -4.99554217e-01
-7.02006280e-01 8.66548777e-01 6.76016152e-01 1.73302248e-01
-4.81759816e-01 4.43939976e-02 5.39920449e-01 2.86241651e-01
3.83382916e-01 -1.86666161e-01 -5.80246568e-01 4.47955012e-01
-3.69797945e-01 2.27413662e-02 8.68745148e-01 8.42593968e-01
9.89468396e-01 4.90969837e-01 3.07839930e-01 2.31043443e-01
5.49080193e-01 5.69398761e-01 8.13160956e-01 -7.67457366e-01
-6.62096739e-02 7.89679408e-01 2.69702762e-01 -8.28579783e-01
-5.31152546e-01 -3.19990844e-01 -1.31233788e+00 6.14190340e-01
-3.26624393e-01 -7.55553722e-01 -1.81703925e-01 1.85893214e+00
7.32393444e-01 -4.32441682e-02 4.32255656e-01 1.00401366e+00
3.11537951e-01 1.13747275e+00 1.02853388e-01 -9.14889276e-01
1.13736165e+00 -7.48896360e-01 -8.56762648e-01 3.26204956e-01
4.52489465e-01 -3.45006324e-02 5.06855249e-01 3.56766731e-02
-8.30587029e-01 -2.54683048e-01 -1.36693311e+00 4.25232679e-01
-7.53611624e-01 6.96193948e-02 4.66613352e-01 4.62119222e-01
-1.04476261e+00 2.68061787e-01 -9.33973432e-01 -1.60539821e-01
4.43162292e-01 8.69761407e-01 -1.07553162e-01 3.35086823e-01
-1.34139633e+00 6.44308090e-01 6.86303258e-01 -2.38068495e-02
-1.08984029e+00 -3.74881268e-01 -7.38082409e-01 1.08523406e-01
3.67108166e-01 -7.45782793e-01 1.03441262e+00 -6.84998930e-01
-1.63064289e+00 3.79711837e-02 4.50631022e-01 -5.79325378e-01
4.83855098e-01 3.98526013e-01 -6.59954786e-01 9.98050943e-02
3.71372513e-02 1.18110716e-01 5.12447774e-01 -1.34564912e+00
-8.99608135e-01 -5.09333789e-01 6.81072548e-02 4.29281563e-01
-6.09079182e-01 -3.35225612e-01 3.24319035e-01 -2.64693946e-01
-5.71689665e-01 -6.53623819e-01 -9.65382010e-02 -4.73929465e-01
-1.69943005e-01 -5.01145661e-01 1.43484402e+00 -5.90502799e-01
9.14487004e-01 -2.08090425e+00 2.46153884e-02 3.09206724e-01
1.64688542e-01 9.94049162e-02 1.62228078e-01 5.20952165e-01
2.99420089e-01 -4.31983508e-02 -3.01506966e-02 -4.07156438e-01
2.93393761e-01 5.85397899e-01 -5.18852053e-03 4.93272394e-01
-3.68172586e-01 6.69037879e-01 -7.74946094e-01 -2.68474072e-01
4.20240045e-01 4.64657471e-02 2.29148008e-02 6.02681220e-01
-4.82167155e-01 3.70258629e-01 -8.91439438e-01 4.76654232e-01
9.15395379e-01 -1.39588535e-01 3.37233007e-01 -2.86924124e-01
-3.86562884e-01 -4.47797447e-01 -1.17435336e+00 1.18970191e+00
-5.39012313e-01 -2.76119560e-01 7.83482969e-01 -1.02853489e+00
8.14065695e-01 3.60338897e-01 9.73919213e-01 -9.27549303e-01
3.83218676e-01 2.49566540e-01 -2.80409783e-01 -1.08763807e-01
1.54631853e-01 1.21457204e-01 -1.97541267e-01 7.98309386e-01
-7.82119036e-02 1.98891729e-01 -1.46402463e-01 1.65134415e-01
8.30245495e-01 -1.31553993e-01 3.20302457e-01 -4.90529895e-01
8.08421195e-01 -6.58818632e-02 8.09108377e-01 1.37362525e-01
-3.76993269e-01 -5.73294938e-01 -1.63330525e-01 -5.95243871e-01
-6.77046597e-01 -7.91595638e-01 3.22852582e-01 7.35991061e-01
5.37735045e-01 -1.23061374e-01 -7.93078482e-01 -9.96730328e-01
9.37676728e-02 8.45539093e-01 -2.58691311e-01 -4.54285741e-01
-2.81756699e-01 -1.13094902e+00 7.20772380e-03 1.70751750e-01
1.19148231e+00 -1.06299734e+00 -8.18525791e-01 4.25478160e-01
-1.38388807e-03 -7.08517253e-01 -4.63182151e-01 4.04445857e-01
-2.42491230e-01 -9.83122945e-01 -2.48442635e-01 -6.35629117e-01
7.65632212e-01 1.37118652e-01 5.79101682e-01 8.07980523e-02
-2.27070339e-02 2.91715682e-01 -2.00162306e-01 -4.81932223e-01
-4.85391468e-01 9.43823382e-02 5.81666470e-01 9.24028084e-02
3.00522685e-01 -7.80591369e-01 -7.58643985e-01 2.87142485e-01
-1.01489556e+00 -1.86906546e-01 4.71974224e-01 8.62414598e-01
6.50219262e-01 8.41398001e-01 1.30527961e+00 -1.06009282e-01
6.33645296e-01 -6.41224146e-01 -1.28119326e+00 2.81867385e-01
-1.04777610e+00 -2.60556132e-01 1.43706071e+00 -1.37723655e-01
-8.25867414e-01 -8.22303221e-02 3.64058614e-01 -3.25428009e-01
2.87034758e-03 -1.96448304e-02 -7.57222533e-01 -3.93776625e-01
-3.47362965e-01 6.21987402e-01 2.80238241e-01 -1.43396512e-01
8.80472139e-02 8.09281528e-01 2.86647737e-01 -3.97121698e-01
1.03812861e+00 1.47147149e-01 2.29355454e-01 -3.85754704e-01
-2.25633550e-02 2.32897937e-01 1.87853590e-01 -1.63531184e-01
8.62645745e-01 -1.17187381e+00 -1.50450933e+00 8.56377423e-01
-7.81571627e-01 -1.29254267e-01 -2.15357363e-01 4.09317702e-01
-4.77642626e-01 2.95347899e-01 -4.51361269e-01 -9.82676864e-01
-8.52841854e-01 -1.06128109e+00 4.54682499e-01 8.48827541e-01
5.45934439e-01 -1.00809371e+00 -3.10748033e-02 7.27364719e-02
5.32749176e-01 3.43419194e-01 1.03601766e+00 -7.24565804e-01
-9.15425837e-01 4.66150753e-02 9.99465510e-02 4.59797382e-01
5.43459833e-01 -3.41975868e-01 -7.22781599e-01 -1.08659637e+00
3.90732706e-01 -5.58199644e-01 -1.07453957e-01 2.15712517e-01
1.14164102e+00 -9.70602691e-01 -1.40807167e-01 5.40987134e-01
1.89429355e+00 6.00093126e-01 1.30852655e-01 4.79904383e-01
4.22068626e-01 1.74276710e-01 5.33922613e-01 7.13405907e-01
9.53660667e-01 2.05231324e-01 1.14149547e+00 -7.67352656e-02
6.96579099e-01 -7.52089769e-02 5.58573246e-01 9.57678139e-01
1.05584860e-01 -3.33076835e-01 -2.22681556e-02 2.59189546e-01
-2.15207601e+00 -6.36396289e-01 4.77268100e-01 2.19660854e+00
7.11511493e-01 -3.07170480e-01 2.51743406e-01 -1.66832715e-01
5.07544160e-01 1.85482100e-01 -1.08258677e+00 -3.34997833e-01
-1.74577758e-01 -3.15366507e-01 7.19807446e-01 1.40974879e-01
-8.19928706e-01 3.14127982e-01 4.21371841e+00 9.37617600e-01
-1.02610087e+00 2.06347108e-01 7.12609649e-01 -1.36493713e-01
-1.92538798e-02 -1.45264566e-01 -5.70321083e-01 9.45036113e-01
9.45224166e-01 -8.38232815e-01 1.08172512e+00 9.90909219e-01
6.49066389e-01 -3.80983204e-01 -7.01451361e-01 6.94263816e-01
-4.55146492e-01 -1.17238092e+00 -2.45869920e-01 4.32748228e-01
9.18137729e-01 1.38095379e-01 -3.59214395e-01 2.74027973e-01
8.19419444e-01 -5.37625968e-01 4.11279231e-01 2.76703060e-01
2.15300158e-01 -1.40580678e+00 8.82583141e-01 6.16894603e-01
-1.42246687e+00 -5.43501675e-01 -4.22165573e-01 1.66941091e-01
-1.33621231e-01 3.63442481e-01 -2.61761397e-01 1.39454472e+00
9.35675919e-01 3.34473729e-01 -1.95255488e-01 6.12242639e-01
1.15819916e-01 1.69572130e-01 -5.57398975e-01 -3.74167711e-01
1.53551400e-01 -6.47944748e-01 3.24843287e-01 3.01368475e-01
2.72193909e-01 2.48972997e-01 8.03946435e-01 7.71512926e-01
-3.03354383e-01 1.35484159e-01 -3.06894600e-01 1.42912537e-01
7.52711356e-01 1.74988604e+00 -2.51916856e-01 -8.29143673e-02
-3.34175885e-01 8.08680117e-01 1.23117432e-01 3.33603621e-01
-7.20622361e-01 -4.44707990e-01 8.31335545e-01 -5.31995416e-01
7.48217553e-02 -1.32673666e-01 9.62055773e-02 -9.83388245e-01
2.53316462e-02 -6.68203115e-01 6.52656555e-01 -6.07368231e-01
-1.56279361e+00 3.74066800e-01 -2.29298845e-01 -1.07920671e+00
-3.44309837e-01 -1.46637872e-01 -1.05664027e+00 8.42403948e-01
-1.81447136e+00 -1.22789478e+00 -4.13928866e-01 1.06945622e+00
-1.66939292e-02 -4.23245788e-01 1.20150983e+00 1.12935282e-01
-9.70827758e-01 5.38460135e-01 6.60999954e-01 4.64719795e-02
2.31257230e-01 -1.24289727e+00 -4.84789401e-01 7.49783516e-01
-5.70367455e-01 7.03014657e-02 4.07259315e-01 -5.69243848e-01
-1.89850283e+00 -1.38256633e+00 -3.54638584e-02 4.91132617e-01
7.14406133e-01 -3.32724661e-01 -7.85994947e-01 5.23991585e-01
7.83144712e-01 2.00591460e-01 4.44417536e-01 -7.58831978e-01
3.33692521e-01 -7.56503761e-01 -1.78448820e+00 3.36716533e-01
2.23010838e-01 -1.43220559e-01 -7.77052939e-02 3.97593737e-01
7.68337548e-01 4.85282764e-02 -1.05639470e+00 1.39535010e-01
-1.23955131e-01 -5.15049100e-01 4.42375004e-01 -2.31844425e-01
-3.50367665e-01 -7.13759124e-01 -2.36369744e-01 -1.95450759e+00
-6.41572326e-02 -8.46417844e-01 -4.74472016e-01 1.64094925e+00
-3.80170226e-01 -1.23427904e+00 5.61438441e-01 6.81488097e-01
5.29260971e-02 -6.44338131e-01 -1.25172102e+00 -8.64410579e-01
1.75257295e-01 4.41581130e-01 1.34672904e+00 9.79407966e-01
3.86368041e-03 -1.00518629e-01 -2.69542754e-01 7.14425087e-01
1.06759346e+00 3.18223864e-01 6.24656737e-01 -7.33842313e-01
-4.71103527e-02 -2.21760999e-02 -6.50548413e-02 -2.46765390e-01
3.43267202e-01 -6.17324650e-01 -1.56491071e-01 -1.50123358e+00
-2.06113189e-01 -4.50090915e-01 -6.47866547e-01 8.53311777e-01
2.41776213e-01 -3.59770745e-01 2.28090852e-01 3.94159146e-02
-7.01513648e-01 1.15564466e+00 9.04995918e-01 -2.32783601e-01
-7.73518607e-02 5.91788292e-02 -4.58083659e-01 4.24423009e-01
1.20425582e+00 -1.14156850e-01 -6.24735296e-01 -1.92637160e-01
-4.86725599e-01 2.79271156e-01 5.18941641e-01 -9.20694053e-01
5.45461416e-01 -4.40170944e-01 4.75590050e-01 -6.72434390e-01
1.06893592e-02 -1.46889281e+00 4.33990151e-01 8.12523365e-01
4.07279879e-01 2.07030565e-01 8.87271203e-03 5.69507003e-01
1.31273732e-01 2.28444412e-01 7.10940897e-01 -2.73056805e-01
-5.56238770e-01 6.21469855e-01 -3.65763128e-01 -3.55453521e-01
1.71246898e+00 3.08396876e-01 -5.68400681e-01 -3.36962551e-01
-3.32455128e-01 1.36941171e+00 5.45297563e-01 2.15318382e-01
2.51742750e-01 -1.44524014e+00 -5.03945947e-01 3.39076132e-01
-1.31067917e-01 1.85065836e-01 3.81163657e-01 3.81872743e-01
9.26589519e-02 -1.32231805e-02 -3.42923552e-01 -2.18352467e-01
-8.43075573e-01 6.50925100e-01 6.68048978e-01 -2.77744055e-01
-4.05737698e-01 6.21378496e-02 -6.44097254e-02 -3.22675228e-01
2.18441367e-01 -3.59874107e-02 7.08853230e-02 2.36838795e-02
3.32753569e-01 6.05718672e-01 5.46102226e-02 -2.91781664e-01
-3.69412571e-01 9.83638912e-02 8.27340409e-02 2.61504263e-01
1.55451822e+00 -5.15023947e-01 -5.11142015e-01 -2.56829318e-02
1.25861692e+00 -1.84373796e-01 -1.20765316e+00 9.21140909e-02
-6.15784526e-01 -1.36493161e-01 2.66016662e-01 -1.06750953e+00
-1.60923719e+00 1.36238441e-01 7.87072301e-01 5.88231266e-01
1.61422884e+00 -4.08540636e-01 9.35954690e-01 2.32429773e-01
1.13347602e+00 -1.37172365e+00 -3.80398989e-01 -5.57816587e-02
3.64433378e-01 -1.04874921e+00 -2.57118773e-02 3.54231268e-01
-4.44446087e-01 1.01490998e+00 1.07034981e+00 -5.72177358e-02
7.55070686e-01 3.43384832e-01 -1.67128772e-01 1.66897178e-01
-6.60792530e-01 3.05471838e-01 -6.09276295e-01 7.37517595e-01
-8.18066537e-01 3.90839189e-01 -2.20660046e-01 1.03658378e+00
5.04913330e-02 1.04301229e-01 5.76779723e-01 9.70226169e-01
-2.86233753e-01 -1.18154752e+00 -2.56877720e-01 2.35844165e-01
-1.84205011e-01 6.30716681e-01 8.53372738e-02 6.53133929e-01
4.47992325e-01 1.10215712e+00 -9.70716178e-02 -2.31308192e-01
-8.84175450e-02 -2.33649790e-01 -1.98720500e-01 1.99597217e-02
-6.99958086e-01 -5.03017381e-03 -3.78564298e-01 -3.59175354e-01
-1.85335845e-01 -4.26108897e-01 -1.77039754e+00 -6.38061702e-01
-2.81532019e-01 7.99085915e-01 7.55473018e-01 8.53442907e-01
4.61852521e-01 4.82525110e-01 1.70439243e+00 -6.39052391e-01
-9.85713661e-01 -5.15229404e-01 -9.40442860e-01 -4.49401736e-02
4.67632532e-01 -3.26987922e-01 -6.30705655e-01 -7.03524292e-01] | [5.592895030975342, 2.568253755569458] |
717bb6b9-b8bd-4a83-9f6d-3eb0a067844c | adversarial-attacks-and-defenses-on-3d-point | 2307.00309 | null | https://arxiv.org/abs/2307.00309v1 | https://arxiv.org/pdf/2307.00309v1.pdf | Adversarial Attacks and Defenses on 3D Point Cloud Classification: A Survey | Deep learning has successfully solved a wide range of tasks in 2D vision as a dominant AI technique. Recently, deep learning on 3D point clouds is becoming increasingly popular for addressing various tasks in this field. Despite remarkable achievements, deep learning algorithms are vulnerable to adversarial attacks. These attacks are imperceptible to the human eye but can easily fool deep neural networks in the testing and deployment stage. To encourage future research, this survey summarizes the current progress on adversarial attack and defense techniques on point cloud classification. This paper first introduces the principles and characteristics of adversarial attacks and summarizes and analyzes the adversarial example generation methods in recent years. Besides, it classifies defense strategies as input transformation, data optimization, and deep model modification. Finally, it presents several challenging issues and future research directions in this domain. | ['Ivan V. Bajić', 'Hanieh Naderi'] | 2023-07-01 | null | null | null | null | ['adversarial-attack', '3d-point-cloud-classification', 'point-cloud-classification'] | ['adversarial', 'computer-vision', 'computer-vision'] | [-9.31088999e-03 -1.52447507e-01 2.29238793e-01 -1.62997440e-01
-4.23361301e-01 -1.00696802e+00 7.60801613e-01 -2.85125166e-01
-2.61542290e-01 2.47957841e-01 -5.78856945e-01 -6.39564037e-01
1.72729075e-01 -8.14905882e-01 -8.61447394e-01 -7.48533249e-01
-1.09731458e-01 4.16129082e-01 1.13981001e-01 -2.12583005e-01
3.77497852e-01 1.56749511e+00 -1.16803741e+00 -3.96717079e-02
6.60591245e-01 1.00888586e+00 -5.06763697e-01 7.87656963e-01
-8.13935697e-02 6.38258338e-01 -1.35961950e+00 -8.16827297e-01
8.88036191e-01 2.62796849e-01 -5.20585954e-01 -1.43829182e-01
8.84877324e-01 -4.79681492e-01 -6.88906252e-01 1.49129772e+00
7.92132020e-01 -1.66745871e-01 5.32239676e-01 -1.83495951e+00
-9.35228407e-01 9.22969878e-02 -4.87317830e-01 3.07142466e-01
-4.37620915e-02 4.54733044e-01 3.56759638e-01 -7.08296418e-01
1.07701987e-01 1.28787148e+00 6.86051071e-01 9.92412210e-01
-7.10583150e-01 -1.14892483e+00 2.20010832e-01 2.47686148e-01
-1.10662043e+00 1.30934611e-01 1.16923940e+00 -4.70997125e-01
1.11058497e+00 4.68279749e-01 5.75182557e-01 1.46928489e+00
4.35165405e-01 8.93385708e-01 8.13255072e-01 -5.16564846e-02
2.55542368e-01 -1.49982706e-01 -1.00753665e-01 3.39509219e-01
3.10006648e-01 8.13565969e-01 2.61539929e-02 -5.71058631e-01
6.72970235e-01 -2.27826424e-02 5.31818457e-02 -4.14229661e-01
-6.43941224e-01 1.02884197e+00 9.59483027e-01 -8.41057599e-02
-3.57810467e-01 3.21713060e-01 5.33735514e-01 3.69763911e-01
4.71949965e-01 7.18281269e-01 -3.23724359e-01 4.32918847e-01
-4.38945621e-01 6.37828648e-01 3.34040403e-01 1.09547007e+00
1.22398213e-01 7.06964195e-01 3.27716261e-01 4.73433167e-01
2.71097898e-01 9.00718570e-01 -1.60019800e-01 -8.31294477e-01
2.74214029e-01 2.39422768e-01 -7.21130744e-02 -1.16498971e+00
-2.62636781e-01 -4.75841165e-01 -9.58321512e-01 1.18710089e+00
5.51848225e-02 -2.73701489e-01 -1.24133277e+00 1.18059528e+00
4.48656023e-01 6.39575481e-01 6.12605177e-03 9.39423084e-01
1.03612101e+00 5.71860492e-01 -2.73021962e-02 3.84198874e-01
1.00454950e+00 -6.54037654e-01 -2.67155319e-01 -4.40983951e-01
-2.19172016e-01 -8.75561178e-01 6.56399846e-01 4.53820139e-01
-9.49256301e-01 -5.66367090e-01 -1.17150283e+00 1.30243868e-01
-5.21992743e-01 -5.77134490e-01 7.45079935e-01 1.06759787e+00
-7.74156213e-01 6.19039476e-01 -1.07993197e+00 1.07578948e-01
1.08342350e+00 7.24083364e-01 -1.85970739e-01 1.22534633e-01
-1.21380639e+00 9.95738089e-01 -8.52945521e-02 1.30365357e-01
-1.21289408e+00 -7.96978951e-01 -6.27980292e-01 -3.72275740e-01
-1.04830340e-01 -9.07748222e-01 1.31841934e+00 -5.01867115e-01
-1.40635765e+00 1.22704732e+00 5.22266746e-01 -9.64480340e-01
5.92951477e-01 -5.06424010e-01 -4.14296418e-01 8.50878358e-02
-2.49648958e-01 6.10671937e-01 1.34529638e+00 -1.33851600e+00
-5.82178950e-01 -6.03964806e-01 3.03968757e-01 -4.25083525e-02
1.74106681e-03 3.09512436e-01 -2.51357734e-01 -8.45180452e-01
-8.52511898e-02 -1.29464912e+00 -4.55498815e-01 3.12635481e-01
-6.54734850e-01 -2.57090241e-01 1.11919272e+00 1.50697574e-01
2.70038366e-01 -2.21132541e+00 -5.36254607e-02 1.64825112e-01
4.79404718e-01 8.63058567e-01 -1.67866036e-01 1.66212931e-01
-4.14181113e-01 3.06735218e-01 -1.11722171e-01 -2.81934977e-01
2.68759355e-02 1.78750977e-01 -1.08905137e+00 8.48394215e-01
4.73676145e-01 1.09124899e+00 -7.61023104e-01 8.53520706e-02
7.38007426e-01 4.18153554e-01 -4.99914646e-01 3.78432572e-01
-1.22941211e-01 5.54793179e-01 -6.75585091e-01 8.99412930e-01
1.28110623e+00 1.97682843e-01 -8.03340375e-01 9.56479311e-02
3.30998331e-01 4.94404696e-02 -6.36311412e-01 1.14409399e+00
-2.00416803e-01 7.80408084e-01 -6.96886554e-02 -9.54296052e-01
1.16632247e+00 1.68634336e-02 4.23460633e-01 -7.12740719e-02
4.45232242e-01 2.08634675e-01 -1.02933114e-02 -2.06672177e-01
1.78975835e-01 -2.27291808e-01 -2.16235653e-01 3.80691588e-02
-2.86460221e-01 -8.20057631e-01 -7.34266698e-01 -4.09367308e-02
1.17192817e+00 -2.05461398e-01 -9.36831385e-02 3.05713862e-01
6.10851526e-01 2.91080445e-01 3.56562585e-01 8.46509516e-01
-5.96836686e-01 5.02072394e-01 1.69824496e-01 -9.66161728e-01
-9.63260353e-01 -1.37434173e+00 -1.44714206e-01 4.04044539e-01
3.83019000e-01 8.66622478e-02 -5.69186151e-01 -1.17438757e+00
5.36910713e-01 6.63744450e-01 -6.42099738e-01 -7.61483550e-01
-5.74316025e-01 -5.18511415e-01 1.24728680e+00 5.55894971e-01
5.64651787e-01 -1.06100774e+00 -3.15618396e-01 -2.39925966e-01
3.50928992e-01 -1.15300488e+00 2.45674089e-01 6.79470645e-03
-8.53114188e-01 -1.07518554e+00 -6.37031853e-01 -6.94212735e-01
4.43176836e-01 5.74282587e-01 1.31746805e+00 2.15224668e-01
-2.15255409e-01 3.47961992e-01 -3.09064001e-01 -1.42191815e+00
-5.80144405e-01 -1.55786991e-01 4.70424592e-01 -5.09940565e-01
1.01094937e+00 -5.28037667e-01 -3.29343617e-01 2.71601677e-01
-7.76090205e-01 -7.12535143e-01 2.44191721e-01 7.09089339e-01
5.36729872e-01 2.58350998e-01 8.47221687e-02 -7.32431293e-01
5.98812580e-01 -2.99049020e-01 -9.61729467e-01 -2.21618533e-01
-2.26594314e-01 -4.28850472e-01 6.43358767e-01 -4.78982955e-01
-4.03218448e-01 -5.50682545e-02 -8.31052005e-01 -1.38475311e+00
-5.02725124e-01 -2.06680045e-01 -1.66160449e-01 -9.63523746e-01
1.05415237e+00 -1.34471595e-01 -4.74891588e-02 -3.23424757e-01
2.56857395e-01 4.78858203e-01 5.97408116e-01 -3.35263073e-01
1.80703878e+00 6.68789387e-01 2.05100298e-01 -9.17877674e-01
-6.62681282e-01 1.74280077e-01 -3.80325049e-01 -1.31541207e-01
8.36580753e-01 -6.87243402e-01 -1.01043105e+00 1.03621995e+00
-1.44497812e+00 -8.59065056e-02 -3.10469389e-01 1.91723108e-01
-4.88932371e-01 4.33186650e-01 -4.94100124e-01 -5.43376207e-01
-3.74835253e-01 -1.38425958e+00 1.11082351e+00 1.69909164e-01
2.23028794e-01 -8.54452431e-01 -7.77623504e-02 2.74377584e-01
2.40079939e-01 8.28037858e-01 7.05139875e-01 -8.53664339e-01
-8.72565627e-01 -7.74976492e-01 6.03267364e-02 6.73277318e-01
-1.46173507e-01 3.24950069e-01 -1.22542608e+00 -3.94115657e-01
5.32432437e-01 -2.94880956e-01 4.82918888e-01 3.75508666e-01
1.52144718e+00 -9.51600596e-02 -2.50470191e-01 1.17517650e+00
1.13198984e+00 3.91341805e-01 7.06853867e-01 5.95469594e-01
9.25160527e-01 1.90090820e-01 4.86199439e-01 -6.64475858e-02
-3.11585575e-01 4.53073949e-01 1.26044631e+00 -2.39024416e-01
1.38726830e-01 4.57275435e-02 1.38022169e-01 5.88391498e-02
-1.03828110e-01 -4.52592611e-01 -1.21451962e+00 1.56499878e-01
-1.28128862e+00 -9.00085986e-01 -1.98360737e-02 1.73355532e+00
1.58570066e-01 6.49271846e-01 -8.88832286e-02 2.24741459e-01
7.81827986e-01 2.90207267e-01 -1.00708878e+00 -5.22660136e-01
-1.74919948e-01 4.71376032e-01 7.63945043e-01 2.79255390e-01
-1.51114821e+00 1.35563171e+00 6.87591743e+00 4.96365905e-01
-1.37685239e+00 -2.15654686e-01 1.79432005e-01 -1.18757464e-01
5.89004755e-02 -4.00940299e-01 -6.49638236e-01 3.16129506e-01
4.66296792e-01 -1.92864358e-01 2.80141592e-01 1.33887362e+00
-4.01438206e-01 6.66914701e-01 -9.79969025e-01 1.08089960e+00
-2.52220146e-02 -1.63925362e+00 3.13829422e-01 2.55527020e-01
6.45205617e-01 8.02311838e-01 6.01628900e-01 1.78946227e-01
5.90226591e-01 -1.07951367e+00 5.65952897e-01 2.44238470e-02
6.58276141e-01 -1.18330383e+00 8.19592297e-01 3.42395812e-01
-6.29184961e-01 -1.79102018e-01 -7.97820032e-01 2.39549987e-02
-8.25327933e-02 2.57156223e-01 -5.92208028e-01 4.65283483e-01
9.99607146e-01 6.83694839e-01 -4.75221395e-01 1.05760419e+00
-6.17662072e-01 3.44073176e-01 -4.25535262e-01 7.73321316e-02
6.20543182e-01 6.01691380e-02 1.26666832e+00 7.43990600e-01
-3.99196381e-03 7.30078220e-02 1.17514031e-02 7.23117232e-01
-4.91141640e-02 -5.33566296e-01 -1.21193552e+00 -3.09156929e-03
6.56680524e-01 7.28164911e-01 -3.34368110e-01 1.43834919e-01
-2.86928028e-01 8.30558002e-01 2.69327518e-02 2.85284400e-01
-1.03174150e+00 -5.04005194e-01 1.60539627e+00 -3.15745950e-01
4.96121198e-02 -5.60364902e-01 -6.11568213e-01 -9.41566885e-01
-2.63171017e-01 -1.16441691e+00 1.99126184e-01 -7.17608333e-01
-1.73405564e+00 9.18734729e-01 -1.44125968e-01 -1.59767199e+00
-8.81380364e-02 -1.17461860e+00 -8.42438459e-01 9.14781213e-01
-1.39657390e+00 -1.16591525e+00 -3.33988547e-01 8.49808037e-01
4.10978734e-01 -8.45980942e-01 8.05340588e-01 2.33463928e-01
-4.07330096e-01 9.10906434e-01 -1.58537656e-01 4.15399462e-01
3.67172539e-01 -1.02597356e+00 1.58160329e+00 1.13575852e+00
3.15655947e-01 5.40351152e-01 8.20961893e-01 -6.49779379e-01
-1.35809863e+00 -1.25594628e+00 2.00373501e-01 -8.87078583e-01
6.62279427e-01 -4.33064789e-01 -9.96811390e-01 6.03886783e-01
-1.30059058e-02 1.75433397e-01 6.31842792e-01 -2.22928673e-01
-4.27549183e-01 9.28575173e-02 -1.53036666e+00 7.99131572e-01
9.26433682e-01 -6.32974148e-01 -5.86405277e-01 5.23560762e-01
9.74434614e-01 -1.05464458e+00 -5.52547336e-01 5.18055677e-01
-2.18939688e-02 -9.67077255e-01 1.56700695e+00 -9.93536949e-01
1.95465595e-01 -3.46852005e-01 -5.64733744e-02 -1.18212187e+00
-5.08969843e-01 -9.17182505e-01 -3.32441509e-01 6.97171509e-01
-1.28166854e-01 -7.04864562e-01 1.23803520e+00 5.54682255e-01
-3.79958242e-01 -6.29758716e-01 -1.09570181e+00 -9.35898125e-01
6.23091459e-01 -7.85302281e-01 1.05949295e+00 9.20893192e-01
-9.55452561e-01 -9.33559425e-03 -3.41196984e-01 9.03833389e-01
1.07403171e+00 -8.19225088e-02 1.27076828e+00 -1.22330141e+00
1.82552874e-01 -4.44382727e-01 -1.02758563e+00 -8.23339939e-01
4.42791760e-01 -8.93652320e-01 -3.31650376e-01 -1.05199933e+00
-7.73109376e-01 -6.55004859e-01 -2.64948815e-01 3.81783843e-01
-1.42977133e-01 5.65481246e-01 5.25461018e-01 2.00347483e-01
3.06078374e-01 3.80501747e-01 1.11146092e+00 -6.38622701e-01
9.53139439e-02 7.99932420e-01 -6.35226727e-01 1.01074445e+00
1.09823585e+00 -7.26586223e-01 -3.16502690e-01 -8.36023510e-01
1.43200621e-01 -5.32256126e-01 7.69098938e-01 -1.09999061e+00
5.57948649e-02 -2.57161230e-01 5.19811094e-01 -9.58176851e-01
4.88630772e-01 -1.37594473e+00 -7.14401342e-03 6.82852149e-01
3.14970240e-02 1.04749635e-01 7.24420726e-01 5.76850295e-01
-4.36873883e-02 -1.15884997e-01 1.20298648e+00 7.29651097e-03
-7.38183320e-01 7.59992957e-01 -1.40822440e-01 -1.31629810e-01
1.40445304e+00 -2.33900592e-01 -4.42027777e-01 -3.93284783e-02
-6.81255460e-01 2.61352718e-01 6.53111696e-01 9.01893198e-01
9.97182846e-01 -1.43543661e+00 -7.76561499e-01 4.41744268e-01
-3.61157097e-02 3.10632825e-01 1.59055710e-01 -6.40400574e-02
-8.75927567e-01 5.50145693e-02 -5.63730419e-01 -8.41561794e-01
-1.33347547e+00 1.13226223e+00 7.34682679e-01 2.29158387e-01
-6.71613038e-01 1.36810362e+00 2.26205304e-01 -4.42010581e-01
4.79381680e-01 1.37594476e-01 7.29642734e-02 -6.31443560e-01
5.34397662e-01 2.32379153e-01 2.24237338e-01 -4.56478000e-01
-6.97578192e-01 5.16025722e-01 -2.60856062e-01 4.39521939e-01
1.32964242e+00 3.43355328e-01 7.34865740e-02 -1.30457073e-01
1.06224215e+00 -9.15637538e-02 -1.16047835e+00 -9.19647701e-03
-4.20344174e-01 -7.76203513e-01 1.90885514e-02 -5.57264507e-01
-1.27458012e+00 1.36613810e+00 7.16194212e-01 3.99295598e-01
1.10024631e+00 -1.46462664e-01 1.01210690e+00 5.68968415e-01
5.04905224e-01 -4.02381212e-01 3.70639078e-02 7.94491827e-01
1.13264155e+00 -1.14578247e+00 8.75862464e-02 -4.87704098e-01
-6.16235554e-01 8.19856048e-01 7.60149896e-01 -9.73185658e-01
9.47177231e-01 4.28676158e-01 5.64865232e-01 -2.83877134e-01
-3.44178319e-01 2.00034723e-01 2.98819393e-01 1.28006327e+00
-3.41419339e-01 -3.11761022e-01 6.03832304e-01 4.07256819e-02
-5.45366585e-01 -4.09857780e-01 2.29219034e-01 8.21568072e-01
-1.44796908e-01 -1.16998899e+00 -7.18542039e-01 9.99983326e-02
-6.54938340e-01 1.67796060e-01 -9.08842862e-01 7.62875855e-01
1.84905663e-01 7.67710209e-01 1.50462523e-01 -9.01779294e-01
6.22526705e-01 -4.16902393e-01 4.47785616e-01 -5.89977264e-01
-8.87738943e-01 -4.06249732e-01 -4.40649182e-01 -4.92203325e-01
-1.03443213e-01 -3.81568968e-01 -8.64920855e-01 -6.94995761e-01
-2.30145916e-01 -2.31197849e-02 8.30705225e-01 4.93589103e-01
4.05716538e-01 4.99083608e-01 1.07677829e+00 -1.27062476e+00
-9.44253445e-01 -3.47888589e-01 -2.42881149e-01 2.34397054e-01
7.28542745e-01 -8.02047849e-01 -5.78138292e-01 -4.14996296e-01] | [7.698335647583008, -4.480578899383545] |
86bf39a5-5df0-4fbb-8004-85375d8fa799 | kernel-learning-for-visual-perception | null | null | https://dr.ntu.edu.sg/handle/10220/47835 | https://dr.ntu.edu.sg/bitstream/10356/105527/1/thesis.pdf | Kernel learning for visual perception | The visual perceptual system in animals allows them to assimilate information from their surroundings. In artificial intelligence, the objective of visual perception is to enable the capability of a computer system to interpret the surrounding environment using data acquired from cameras and other aided sensors. Since the last century, researchers in visual perception have delivered many marvelous technologies and algorithms for various applications, such as object detection and image recognition, etc. Despite the technological progresses, human beings are still confronted by the unsatisfactory performance of artificial visual perceptual systems. One of the main reasons is that the traditional methods usually rely on large amount of training data, powerful processors, and require great efforts and time for process modeling. The research goal of this thesis is to develop visual perceptual systems that requires less computational resources but with higher performance. To this end, the novel kernel learning methods for several basic visual perceptual tasks, including object tracking, localization, mapping, and image recognition, are proposed and demonstrated both theoretically and practically. In visual object tracking, the state-of-the-art algorithms that leverage on kernelized correlation filters are limited by circulant training data and non-weighted kernel functions. This makes them only applicable for translation prediction and prevents their usage in other applications. To overcome the problems, a kernel cross-correlator (KCC) is introduced. First, by introducing the kernel trick, the KCC extends linear cross-correlation to non-linear space, which is more robust to signal noises and distortions. Second, connections to the existing works show that the KCC provides a unified solution for correlation filters. Third, the KCC is not only applicable to any training data and kernel functions, but also able to predict affine transforms with customized properties. Last, by leveraging the fast Fourier transform (FFT), the KCC eliminates direct calculation of kernel vectors, thus achieving better performance at a reasonable computational cost. Comprehensive experiments on visual tracking and human activity recognition using wearable devices have demonstrated its robustness, flexibility, and efficiency. Optical flow is the pattern of apparent motion of objects, surfaces, and edges in a visual scene caused by the relative motion between an observer and a scene. It is calculated from sequences of ordered images and allows the estimation of motion as instantaneous image velocities, which is crucial for autonomous robot navigation. This thesis proposes a KCC-based algorithm to determine optical flow using a monocular camera, which is named as correlation flow (CF). CF can provide reliable and accurate velocity estimation and is robust to motion blur. In addition, a joint kernel scale-rotation correlator is proposed to estimate the altitude velocity and yaw rate which are not available by traditional methods. Autonomous flight tests on a quadcopter show that correlation flow can provide robust trajectory estimation with very low processing power. In the problem of simultaneous localization and mapping (SLAM), traditional odometry methods resort to iterative algorithms which are usually computationally expensive or require well-designed initialization. To overcome this problem, a KCC-based non-iterative solution to RGB-D-inertial odometry system is proposed. To reduce the odometry and inertial drifts, two frameworks for non-iterative SLAM (NI-SLAM) are presented. One is to combine a visual loop closure detection, another one is to seek the aids from ultra wide-band (UWB) technology. Dominated by the FFT, the non-iterative front-end is only of $\mathcal{O}(n\log n)$ complexity, where $n$ is the number of pixels. Therefore, both frameworks can provide reliable performance and are of very low computational complexity. The map fusion is conducted by element-wise operation, so that both time and space complexity are further reduced. Extensive experiments show that, due to the lightweight of the proposed non-iterative front-end, both frameworks of NI-SLAM can run at a much faster speed and yet still with comparable accuracy with the state-of-the-arts. Convolutional neural network (CNN) is one of the most powerful tools in visual perception. It has enabled many state-of-the-art performances in image recognition, object detection, etc. However, little effort has been devoted to establishing convolution in non-linear space. In this thesis, a new operation, kervolution (kernel convolution), is introduced to approximate the non-linear behavior of the human perceptual system. It generalizes traditional convolution and increases the model capacity without introducing more parameters. Similarly, kervolution can also be calculated through element-wise multiplication via Fourier transform. The extensive experiments show that the kervolutional neural networks (KNN) achieve better performance and faster convergence than traditional CNN on the MNIST, CIFAR, and ImageNet datasets. In summary, the thesis demonstrates the superiority of the proposed kernel tools for visual perceptual problems, including KCC, CF, NI-SLAM and KNN. With the kernel tools, we may expect their usage in more applications, such as internet of things, robotics, transfer learning, reinforcement learning, etc. | ['Chen Wang'] | 2019-12-06 | null | null | null | null | ['loop-closure-detection'] | ['computer-vision'] | [ 1.24918327e-01 -8.56459498e-01 -1.43421069e-01 1.06819153e-01
3.88722122e-01 -4.00108337e-01 2.84057945e-01 -7.11184368e-02
-6.60346389e-01 4.63555634e-01 -4.05853242e-01 -8.42871610e-03
-2.65659630e-01 -4.37823653e-01 -3.28088135e-01 -8.67696166e-01
-1.74418211e-01 -4.67419565e-01 5.43455899e-01 1.42505974e-01
2.86016762e-01 6.07480764e-01 -1.92007577e+00 -1.25005618e-01
8.51352572e-01 1.09567106e+00 6.63255930e-01 7.78658092e-01
3.23887132e-02 6.24771059e-01 -4.20597255e-01 -2.73003299e-02
1.20409429e-02 -4.97268140e-01 -1.64973930e-01 5.54198213e-03
2.39687651e-01 -9.31612104e-02 -2.08994657e-01 1.17097878e+00
3.16649348e-01 1.63878173e-01 6.93366706e-01 -1.25152159e+00
-8.65819156e-01 -1.77887142e-01 -4.72000301e-01 2.68703967e-01
3.36274266e-01 1.37579769e-01 2.12149262e-01 -6.57689214e-01
2.05033094e-01 1.10477543e+00 6.62202954e-01 4.96151626e-01
-1.27949393e+00 -4.51816022e-01 -1.51350096e-01 5.55979311e-01
-1.12240636e+00 -1.02947213e-01 8.05256188e-01 -7.21041560e-01
7.20256984e-01 3.62070829e-01 9.56750929e-01 8.15016031e-01
5.99385023e-01 4.87728655e-01 1.23292899e+00 -3.33153218e-01
2.53428102e-01 3.19920599e-01 1.68196797e-01 6.98217630e-01
4.01786774e-01 3.22342217e-01 -3.62559438e-01 1.57096386e-01
1.06361008e+00 7.63387308e-02 -8.10453296e-01 -6.26461506e-01
-1.35306108e+00 4.60345566e-01 3.60328734e-01 5.05787432e-01
-2.00351030e-01 -2.87501384e-02 3.05050462e-01 2.30431303e-01
4.95787933e-02 3.64132166e-01 -3.21060419e-01 -3.02732229e-01
-6.06375694e-01 -1.99144572e-01 7.73453712e-01 5.80689967e-01
4.73198354e-01 2.50901163e-01 1.48230016e-01 7.06242740e-01
4.44876671e-01 1.03070390e+00 9.07955468e-01 -8.53987098e-01
-1.75952345e-01 6.01970196e-01 2.10526027e-02 -1.40046501e+00
-6.19779229e-01 -3.87238771e-01 -1.11117625e+00 8.03900599e-01
5.34915209e-01 2.06812561e-01 -5.78772187e-01 1.47983742e+00
3.59363616e-01 2.71595240e-01 1.19492188e-01 1.20784009e+00
5.75998783e-01 6.92617297e-01 -1.00671299e-01 -5.96253276e-01
1.43605042e+00 -6.21145487e-01 -8.96761835e-01 -1.01815298e-01
3.15214455e-01 -9.92946863e-01 9.90573704e-01 5.97642899e-01
-6.43527865e-01 -1.07113481e+00 -1.17902946e+00 2.24055886e-01
-5.70060313e-01 4.65489358e-01 7.56328344e-01 8.27322364e-01
-7.40616679e-01 4.37498033e-01 -9.27799523e-01 -4.64058995e-01
-6.16251081e-02 1.81717500e-01 -4.69895005e-01 1.32400125e-01
-8.37816715e-01 1.08158302e+00 3.64179313e-01 3.67270172e-01
-2.43296817e-01 -5.41101098e-01 -8.12825680e-01 -8.77673626e-02
2.68674754e-02 -5.95468879e-01 6.19938970e-01 -1.15816522e+00
-1.74657702e+00 3.93163145e-01 -1.97145134e-01 -3.09906960e-01
4.66968656e-01 -3.63375932e-01 -6.11943364e-01 2.63476133e-01
-8.78127217e-02 3.20726275e-01 1.26343417e+00 -9.46360648e-01
-5.49517334e-01 -2.27868989e-01 -3.60750914e-01 -8.83517601e-03
-3.93244386e-01 -2.23120719e-01 -2.53924340e-01 -4.62284982e-01
1.53015152e-01 -8.95812333e-01 1.29685074e-01 1.77360728e-01
1.18212715e-01 -8.65269303e-02 1.16532874e+00 -6.18156910e-01
1.10899687e+00 -2.35002017e+00 1.10367104e-01 5.11723720e-02
-9.75075588e-02 5.96077442e-01 2.83773914e-02 4.06335713e-03
-1.98121294e-02 -5.34437418e-01 -7.36507177e-02 1.49073079e-01
-4.09715474e-01 4.64514419e-02 -2.22525969e-01 7.27211654e-01
5.64857200e-02 6.10717416e-01 -1.00724840e+00 -4.06355917e-01
7.53056526e-01 7.52750754e-01 -1.68513685e-01 3.91246863e-02
2.48084843e-01 5.88976502e-01 -2.33701915e-01 4.67369914e-01
6.58853531e-01 -9.10551399e-02 -2.24312738e-01 -7.02518225e-01
-4.49030995e-01 -5.66744328e-01 -1.32320070e+00 1.43778110e+00
-4.03829038e-01 1.00109935e+00 -9.65457689e-03 -1.05561781e+00
1.06573725e+00 1.61105782e-01 5.58674872e-01 -7.93266296e-01
1.17312267e-01 2.94937730e-01 2.43126303e-02 -8.38322580e-01
2.71258086e-01 1.93541840e-01 4.52416897e-01 -3.50201651e-02
-1.09185740e-01 1.07532844e-01 1.55623868e-01 -2.17119932e-01
7.95043886e-01 3.60406786e-01 3.26852918e-01 -2.41337627e-01
1.04016030e+00 -1.88681439e-01 5.45831263e-01 3.74268532e-01
-2.48464853e-01 2.62045890e-01 -3.97638977e-03 -5.00116646e-01
-8.09661627e-01 -1.11636925e+00 -2.25315154e-01 4.70038980e-01
5.33024430e-01 1.16421767e-02 -5.26867092e-01 -1.44966498e-01
7.81561807e-02 1.48822159e-01 -3.08522314e-01 -2.74683505e-01
-4.57392752e-01 -5.48825026e-01 2.38194510e-01 4.78462756e-01
8.73418987e-01 -8.94893646e-01 -1.23795021e+00 2.04085469e-01
3.24804671e-02 -1.27188635e+00 -3.44211221e-01 -1.95080042e-01
-9.30263460e-01 -1.31927383e+00 -7.91855991e-01 -7.71315217e-01
6.24551177e-01 6.88168705e-01 3.94394219e-01 -1.54113621e-01
-7.35998273e-01 6.51648998e-01 -2.93708384e-01 -2.32823104e-01
-1.06456466e-01 -5.12933612e-01 3.48834097e-01 3.28738034e-01
2.99865127e-01 -4.46255028e-01 -6.83229089e-01 5.47986329e-01
-7.28068054e-01 -1.12415962e-01 7.22964466e-01 8.83936942e-01
2.98818529e-01 1.33534670e-01 2.50342190e-01 -9.86223742e-02
5.31554103e-01 1.00291327e-01 -1.00776196e+00 3.02885711e-01
-5.77417016e-01 6.82728291e-02 7.92250395e-01 -9.25072074e-01
-1.03231561e+00 1.58656135e-01 4.27398741e-01 -4.56438780e-01
-2.30255485e-01 2.34619558e-01 8.96576047e-02 -3.74293864e-01
7.48171926e-01 5.26169062e-01 4.18465078e-01 -1.72966748e-01
2.64947802e-01 5.69770455e-01 7.62854755e-01 -9.44481045e-02
7.65415311e-01 6.48751974e-01 4.01621938e-01 -1.55554056e+00
-2.11019397e-01 -7.32439280e-01 -6.76464319e-01 -6.89809799e-01
1.16806650e+00 -6.14038289e-01 -1.37140834e+00 7.57471681e-01
-1.18371677e+00 -3.87142901e-03 6.04247898e-02 1.35168958e+00
-5.16404033e-01 7.34293699e-01 -4.82259750e-01 -1.10628462e+00
-9.46220011e-02 -1.00271416e+00 5.31071901e-01 6.58560336e-01
-7.79036433e-02 -1.15395498e+00 8.32415186e-04 2.23796934e-01
5.02817452e-01 1.05644464e-01 6.43690526e-01 1.65013880e-01
-6.94879055e-01 -1.18918270e-01 -2.72661924e-01 7.51086354e-01
3.36730003e-01 9.07907784e-02 -8.83897781e-01 -3.77411723e-01
2.70091653e-01 1.21835731e-01 6.40026450e-01 6.40180588e-01
6.43698573e-01 8.92355293e-02 -3.17195296e-01 7.00559258e-01
1.35104966e+00 6.02198482e-01 6.13199592e-01 3.45221817e-01
6.51894450e-01 6.38146162e-01 6.02981031e-01 1.34789392e-01
-7.49147981e-02 7.29519248e-01 2.49164999e-01 -9.37288925e-02
-1.85420439e-01 1.77620709e-01 5.54358184e-01 8.48515809e-01
-5.32606483e-01 3.57650667e-01 -3.75205427e-01 2.15198696e-01
-1.86083817e+00 -1.10174823e+00 -4.40762192e-01 2.50195408e+00
3.10925663e-01 -5.32480963e-02 -6.89592808e-02 3.06940854e-01
6.37215972e-01 -2.39114434e-01 -5.36211848e-01 -2.14759111e-01
-1.48359463e-01 -2.20125471e-03 6.62490845e-01 3.71770799e-01
-1.19591343e+00 6.47906065e-01 5.38047075e+00 4.93366748e-01
-1.47589684e+00 -1.32674187e-01 -5.85499927e-02 3.44267070e-01
4.84085202e-01 -1.08442649e-01 -5.30029774e-01 6.18192017e-01
5.40739417e-01 3.64595801e-02 5.16219676e-01 8.60964537e-01
3.64558697e-01 -5.33071220e-01 -8.75691175e-01 1.53145099e+00
2.43330419e-01 -8.03899527e-01 -3.60727310e-01 -1.45069035e-02
3.66441399e-01 -4.38448429e-01 2.44256169e-01 -1.37984455e-01
-3.92827034e-01 -7.66660213e-01 4.55971152e-01 7.72940218e-01
3.12803209e-01 -4.15568531e-01 7.66867995e-01 3.08533311e-01
-1.30527496e+00 -2.10234717e-01 -6.97788596e-01 -2.73620486e-01
1.39208212e-02 6.24930978e-01 -4.54151154e-01 4.09007907e-01
7.32109308e-01 7.88711071e-01 -3.79717648e-01 1.46974063e+00
-1.80117577e-01 2.16805175e-01 -1.27934545e-01 -3.82361948e-01
-1.10710338e-01 -5.73840499e-01 6.11521125e-01 1.14232469e+00
4.93927866e-01 -1.58485532e-01 -3.67194265e-02 8.42582583e-01
7.64536262e-01 2.29047954e-01 -6.60393476e-01 1.48053229e-01
1.65271237e-01 1.36454105e+00 -9.18530703e-01 -1.41727269e-01
-5.89864969e-01 1.15557885e+00 -1.43215299e-01 5.65177798e-01
-7.99182773e-01 -4.94407356e-01 7.68152475e-01 -2.24872410e-01
1.31517708e-01 -6.99515224e-01 -7.87984505e-02 -1.14335656e+00
1.31847799e-01 -4.41617370e-01 -2.19771490e-02 -8.42325568e-01
-1.12868559e+00 3.89663130e-01 -1.19246721e-01 -1.62313747e+00
2.88654398e-03 -1.10416543e+00 -2.64815420e-01 9.06149745e-01
-1.43260610e+00 -7.97807336e-01 -5.97160101e-01 7.48859525e-01
4.18338656e-01 -2.27036208e-01 7.42394686e-01 2.93257117e-01
-3.63660276e-01 9.87217575e-02 7.54945353e-02 1.77921712e-01
8.41355979e-01 -1.02448499e+00 -1.96510196e-01 1.00062799e+00
4.37935963e-02 5.95793486e-01 6.81193531e-01 -5.06804109e-01
-1.58191228e+00 -6.25314891e-01 4.53838915e-01 -2.56593466e-01
5.96655309e-01 -8.77536535e-02 -9.11130965e-01 3.79201435e-02
-1.02427006e-01 1.35573879e-01 5.39703012e-01 -3.80866677e-01
-1.30622759e-01 -5.50218940e-01 -7.47952580e-01 4.59564954e-01
5.73685527e-01 -4.91325736e-01 -5.32268882e-01 -3.50943543e-02
2.35055894e-01 -1.63629681e-01 -7.88994551e-01 2.94327617e-01
1.03510308e+00 -1.16630256e+00 1.09187818e+00 -1.31774545e-01
-1.01342343e-01 -8.60467076e-01 1.06890753e-01 -1.18973947e+00
-5.05582571e-01 -4.05980140e-01 -7.90444240e-02 8.86164546e-01
-5.10826781e-02 -9.65191782e-01 5.65539479e-01 3.16933662e-01
2.52434295e-02 -2.11736664e-01 -9.20338452e-01 -1.13729870e+00
-5.68968534e-01 -4.30399656e-01 -1.10623583e-01 9.50906038e-01
9.25630778e-02 2.31160253e-01 -4.61000443e-01 4.12491202e-01
9.47193384e-01 6.49515688e-02 7.90582359e-01 -1.33291793e+00
-3.63097340e-01 -4.41481829e-01 -1.05049324e+00 -1.22834373e+00
-8.57383385e-02 -4.97392058e-01 -7.97357336e-02 -1.38217664e+00
-2.24646315e-01 -7.98730478e-02 -7.33140856e-02 -2.20318697e-02
-3.90334241e-02 2.00600058e-01 3.05021673e-01 4.05859858e-01
-7.57395923e-02 3.45618039e-01 1.42433333e+00 -1.15969338e-01
-4.44183320e-01 1.70527607e-01 1.19092457e-01 8.26249421e-01
6.71265244e-01 5.35108447e-02 -5.61309814e-01 -1.53233618e-01
4.91369516e-04 -1.59222782e-01 7.41059422e-01 -1.49014235e+00
5.72033882e-01 3.90233696e-02 7.00288177e-01 -3.66895854e-01
3.56401354e-01 -1.15516853e+00 2.51250505e-01 9.43432927e-01
2.08967999e-01 1.18002824e-01 9.47027057e-02 6.34609222e-01
-3.62183064e-01 -1.55537337e-01 8.98536325e-01 1.20301321e-02
-1.19606507e+00 -7.01835155e-02 -5.17308652e-01 -5.16016901e-01
1.32890403e+00 -6.41226411e-01 -5.02903104e-01 -1.41971588e-01
-5.29691815e-01 -1.73504055e-01 3.49543512e-01 4.07341540e-01
8.55792642e-01 -1.17039406e+00 -1.86337695e-01 5.69067717e-01
-5.36950752e-02 -4.10953492e-01 3.77533019e-01 1.25815177e+00
-7.50332117e-01 5.65954506e-01 -5.28501689e-01 -8.75355601e-01
-1.48623443e+00 8.79720569e-01 2.67622352e-01 3.16055596e-01
-5.38330138e-01 3.94001931e-01 1.92401662e-01 2.42637366e-01
1.61374360e-01 -7.12832928e-01 -4.73840237e-01 -2.42924951e-02
6.75128877e-01 6.57625437e-01 -2.78559625e-01 -6.86146736e-01
-3.45220447e-01 1.22078705e+00 4.79765654e-01 1.16463974e-01
7.69849539e-01 -1.39745578e-01 -7.21184313e-02 5.80880821e-01
1.11330736e+00 -5.10693453e-02 -1.29961848e+00 9.94003564e-02
-1.13251016e-01 -6.13847852e-01 -6.11760318e-02 -5.75897932e-01
-8.75601888e-01 9.22524333e-01 1.06887078e+00 4.73752737e-01
1.34146738e+00 -4.02575731e-01 3.89082134e-01 2.16638476e-01
4.24326211e-01 -1.01258755e+00 2.70856887e-01 3.18217278e-01
5.11761844e-01 -1.05742443e+00 1.86061487e-03 -6.59736991e-01
-5.26715994e-01 1.55108786e+00 5.30386388e-01 4.09435816e-02
6.12426281e-01 1.77340776e-01 2.66140521e-01 2.66297430e-01
-2.37928182e-01 -3.78084898e-01 4.08632457e-01 1.00338173e+00
3.75337005e-01 -1.73562113e-03 -4.03474718e-01 -3.26218195e-02
3.68767940e-02 1.17894262e-01 3.47887307e-01 8.05567741e-01
-6.25327528e-01 -9.05329704e-01 -7.07798481e-01 -2.09736228e-02
-1.80269465e-01 2.87634164e-01 1.27696723e-01 7.00297356e-01
2.63151675e-01 9.90357757e-01 -2.18276437e-02 -3.52413565e-01
4.17750359e-01 -1.58325791e-01 6.32527053e-01 5.53195812e-02
-3.61697190e-03 7.90731609e-02 -4.45597410e-01 -6.15403771e-01
-8.99486840e-01 -4.55848664e-01 -1.08518267e+00 5.07822856e-02
-3.69634718e-01 1.64617211e-01 1.06763518e+00 6.98916912e-01
1.16034038e-01 5.37641644e-01 4.07460570e-01 -6.98045611e-01
-2.60936856e-01 -7.96915531e-01 -6.55905843e-01 4.31224793e-01
3.01154673e-01 -9.30101275e-01 -4.45205569e-01 3.46543729e-01] | [8.271773338317871, -1.7379542589187622] |
c94b96ec-7cad-4ddc-bfd3-095cfb13d9c3 | towards-robust-semantic-segmentation-of | 2203.10395 | null | https://arxiv.org/abs/2203.10395v1 | https://arxiv.org/pdf/2203.10395v1.pdf | Towards Robust Semantic Segmentation of Accident Scenes via Multi-Source Mixed Sampling and Meta-Learning | Autonomous vehicles utilize urban scene segmentation to understand the real world like a human and react accordingly. Semantic segmentation of normal scenes has experienced a remarkable rise in accuracy on conventional benchmarks. However, a significant portion of real-life accidents features abnormal scenes, such as those with object deformations, overturns, and unexpected traffic behaviors. Since even small mis-segmentation of driving scenes can lead to serious threats to human lives, the robustness of such models in accident scenarios is an extremely important factor in ensuring safety of intelligent transportation systems. In this paper, we propose a Multi-source Meta-learning Unsupervised Domain Adaptation (MMUDA) framework, to improve the generalization of segmentation transformers to extreme accident scenes. In MMUDA, we make use of Multi-Domain Mixed Sampling to augment the images of multiple-source domains (normal scenes) with the target data appearances (abnormal scenes). To train our model, we intertwine and study a meta-learning strategy in the multi-source setting for robustifying the segmentation results. We further enhance the segmentation backbone (SegFormer) with a HybridASPP decoder design, featuring large window attention spatial pyramid pooling and strip pooling, to efficiently aggregate long-range contextual dependencies. Our approach achieves a mIoU score of 46.97% on the DADA-seg benchmark, surpassing the previous state-of-the-art model by more than 7.50%. Code will be made publicly available at https://github.com/xinyu-laura/MMUDA. | ['Rainer Stiefelhagen', 'Kunyu Peng', 'Alina Roitberg', 'Kailun Yang', 'Jiaming Zhang', 'Xinyu Luo'] | 2022-03-19 | null | null | null | null | ['scene-segmentation'] | ['computer-vision'] | [ 2.80899107e-01 2.57593781e-01 -1.39993176e-01 -5.97779036e-01
-9.51259017e-01 -3.68037999e-01 5.09846866e-01 -2.97429800e-01
-4.78083163e-01 4.97811377e-01 -2.94473600e-02 -3.01090509e-01
2.83054948e-01 -8.93417120e-01 -1.10484588e+00 -7.03643084e-01
5.54112554e-01 4.04485315e-01 7.91430533e-01 -4.15823430e-01
-8.58694315e-03 3.39975327e-01 -1.53782654e+00 4.74251211e-01
1.08668518e+00 9.23583984e-01 3.56285214e-01 5.95116735e-01
1.14613269e-02 6.32351696e-01 -4.91737843e-01 -6.20536208e-01
3.44861895e-01 -2.13267952e-01 -7.57370591e-01 2.95100659e-01
3.80208045e-01 -4.25924212e-01 -4.30401534e-01 1.15530992e+00
2.01565951e-01 1.85471222e-01 5.41586697e-01 -1.35383809e+00
-3.79542023e-01 2.50010520e-01 -6.97101772e-01 2.25543991e-01
-1.72606483e-01 5.99883497e-01 6.46724999e-01 -6.82538092e-01
5.05641937e-01 1.20574582e+00 5.28517723e-01 5.90251505e-01
-9.82909560e-01 -8.24367702e-01 3.19449127e-01 6.05694950e-01
-1.06349778e+00 -4.76395547e-01 7.51330018e-01 -3.40884298e-01
7.08489895e-01 2.76714526e-02 3.44847530e-01 1.35279953e+00
2.49717057e-01 1.03112042e+00 8.04793298e-01 1.17695443e-01
1.46423683e-01 3.14619602e-03 2.19264060e-01 4.30425227e-01
2.21161619e-01 4.31098044e-02 -3.90546918e-02 4.45030004e-01
3.90738577e-01 3.38762477e-02 2.16214091e-01 -2.07126379e-01
-1.01708627e+00 5.64573944e-01 5.78389645e-01 -1.00745440e-01
-5.32173991e-01 1.43320233e-01 3.48346710e-01 -3.68633145e-03
4.56803679e-01 1.09131202e-01 -4.01639581e-01 -7.45269060e-02
-6.70157075e-01 3.15572590e-01 1.93409353e-01 9.90990937e-01
8.27946603e-01 5.74463839e-03 -2.04964221e-01 9.40022826e-01
6.73961639e-02 7.59798884e-01 1.62441567e-01 -1.18303812e+00
7.80798495e-01 6.41775668e-01 -2.41228323e-02 -6.81382954e-01
-5.36017239e-01 -4.91171539e-01 -7.19415188e-01 2.07722515e-01
4.61192429e-01 -1.28994480e-01 -1.30575228e+00 1.73988152e+00
4.22943026e-01 3.87765855e-01 1.37705594e-01 9.93036330e-01
6.94383085e-01 7.79304385e-01 3.65079701e-01 3.96537989e-01
1.20478809e+00 -1.17597234e+00 -4.13616270e-01 -7.54222333e-01
5.26646674e-01 -4.58551049e-01 1.04840660e+00 2.94019371e-01
-1.05668986e+00 -7.64739335e-01 -9.32602465e-01 -2.42862046e-01
-4.28352296e-01 -1.12589844e-01 2.02773988e-01 5.61503410e-01
-8.27124596e-01 1.32618546e-01 -9.53670979e-01 -4.02694851e-01
8.71106982e-01 2.02009767e-01 -2.57115960e-01 -3.43673259e-01
-1.08121955e+00 8.90544355e-01 5.31774580e-01 4.18372564e-02
-9.30034578e-01 -7.19528198e-01 -8.74824524e-01 -3.29305202e-01
6.03467226e-01 -5.87798476e-01 1.35315323e+00 -9.88720775e-01
-1.12257326e+00 9.46450770e-01 -3.88003737e-01 -7.23660588e-01
7.35928118e-01 -3.05186898e-01 -4.51395601e-01 1.58475950e-01
2.78914720e-01 9.80665684e-01 7.26427078e-01 -1.35345733e+00
-9.69994843e-01 -3.36429209e-01 4.81539704e-02 1.27187207e-01
9.83941033e-02 -8.17025453e-02 -7.81286299e-01 -5.16456723e-01
-1.07218504e-01 -9.58592653e-01 -5.22207320e-01 -2.37226531e-01
-5.52439749e-01 2.56761722e-02 1.01287997e+00 -6.82999849e-01
8.69696200e-01 -2.36145353e+00 -4.91126999e-02 4.94427308e-02
2.19056591e-01 5.26606202e-01 -3.25307608e-01 1.12337926e-02
4.91742603e-02 -1.14199676e-01 -7.54154384e-01 -5.03549695e-01
-9.52270627e-02 4.15423244e-01 -2.24735230e-01 3.68876249e-01
4.51395899e-01 1.14077020e+00 -7.82064021e-01 -4.64452654e-01
4.29201961e-01 2.34135494e-01 -6.01714015e-01 2.54080176e-01
-2.61204660e-01 7.44184673e-01 -5.50099254e-01 6.43373787e-01
9.26662624e-01 -3.98514643e-02 -3.85932386e-01 -9.88757014e-02
-5.61484545e-02 -9.96629149e-02 -8.57665718e-01 1.50457668e+00
-3.50901991e-01 6.33561492e-01 1.40166014e-01 -1.18271816e+00
6.09291673e-01 -7.73429275e-02 4.40251589e-01 -1.16101766e+00
2.43136585e-01 7.00600445e-02 -1.77398711e-01 -7.31630504e-01
4.93438959e-01 4.18397933e-02 -2.86121041e-01 -1.11761063e-01
-1.69454217e-01 -5.91976792e-02 3.10286194e-01 8.01619515e-02
1.07953644e+00 -1.35410316e-02 -2.95740485e-01 9.18703973e-02
4.57456231e-01 2.98243344e-01 7.33825564e-01 6.48086429e-01
-5.53618073e-01 9.72847879e-01 5.55787027e-01 -2.48146251e-01
-1.16242433e+00 -1.27167416e+00 -8.03735331e-02 1.01976299e+00
4.98699069e-01 1.84492797e-01 -1.18550885e+00 -6.38582587e-01
-9.22942832e-02 1.08921492e+00 -4.62720901e-01 -4.93393153e-01
-6.79696441e-01 -8.38455319e-01 6.84672713e-01 7.42062032e-01
1.03004301e+00 -1.02460527e+00 -6.09170794e-01 2.45789066e-01
-4.09447491e-01 -1.67281234e+00 -3.87602001e-01 -7.15144724e-02
-4.69760060e-01 -1.09025788e+00 -6.26170337e-01 -6.49304330e-01
4.88070458e-01 3.06086212e-01 8.70855510e-01 -1.50495276e-01
-2.24958122e-01 1.17286801e-01 -2.95351267e-01 -4.63133544e-01
-6.02752805e-01 1.16016783e-01 -2.20999673e-01 3.54084462e-01
4.13240969e-01 -3.23602080e-01 -6.77703381e-01 5.29695392e-01
-1.03677058e+00 3.84541065e-01 7.72291064e-01 5.48427939e-01
6.30548418e-01 -5.20689376e-02 6.87638283e-01 -9.52397287e-01
1.01680078e-01 -7.68691778e-01 -5.31207860e-01 -1.37274593e-01
-2.44502783e-01 -2.51078159e-01 6.50999367e-01 -1.63472071e-01
-1.29244971e+00 9.54765156e-02 -5.33041954e-01 -5.51904321e-01
-7.38836050e-01 1.41935647e-01 -5.85856080e-01 2.46848255e-01
4.48543936e-01 1.27355322e-01 3.80797498e-02 -2.57579505e-01
4.80276972e-01 6.68863177e-01 9.46755052e-01 -4.69758064e-01
6.93296373e-01 6.56907380e-01 -1.86874166e-01 -9.35477376e-01
-8.24423373e-01 -6.48887038e-01 -6.00751400e-01 -3.57974738e-01
1.26880455e+00 -9.25215721e-01 -1.78482518e-01 8.81708741e-01
-1.05005801e+00 -7.84504592e-01 -1.62044480e-01 1.57593563e-01
-6.88549340e-01 2.05307782e-01 -3.75039786e-01 -4.92954254e-01
2.40251407e-01 -1.46338832e+00 1.13067651e+00 1.86884537e-01
1.36628181e-01 -6.92045689e-01 -3.16347659e-01 9.36257422e-01
3.39583069e-01 3.95047754e-01 8.02323341e-01 -6.88887000e-01
-7.35975444e-01 -6.84355795e-02 -5.03285885e-01 5.72121501e-01
-7.10547566e-02 -2.32667118e-01 -1.07027614e+00 -8.75115022e-03
-2.30809793e-01 -1.07804939e-01 1.15882552e+00 4.72746730e-01
1.40799022e+00 -4.51889560e-02 -3.53838176e-01 6.80943847e-01
1.08362734e+00 3.36402655e-01 9.14925814e-01 3.88775021e-01
8.42827201e-01 7.90835559e-01 7.57339716e-01 1.94583118e-01
7.78442323e-01 5.78345716e-01 6.82067096e-01 -3.62974703e-01
-4.84262943e-01 -1.03187390e-01 3.86687994e-01 1.22293197e-01
1.01416156e-01 -5.36269605e-01 -1.18057454e+00 9.06620145e-01
-1.97354841e+00 -9.60038781e-01 -4.64352816e-01 1.93131936e+00
4.42416728e-01 5.33842087e-01 2.33613342e-01 -1.10355735e-01
7.96098948e-01 2.43695572e-01 -8.91873479e-01 -3.73782396e-01
-1.93007722e-01 -6.58150613e-02 7.97433257e-01 4.13315743e-01
-1.38005877e+00 1.31573975e+00 4.69028282e+00 9.81721342e-01
-1.02631950e+00 4.01586592e-01 9.55245018e-01 6.55103475e-02
-6.47564083e-02 -2.90263027e-01 -8.13256383e-01 6.29991114e-01
1.13600516e+00 2.47179747e-01 2.49286983e-02 7.45342612e-01
4.62927133e-01 -8.75077173e-02 -7.65208781e-01 7.53635764e-01
-5.99373654e-02 -1.20843351e+00 -1.13421366e-01 -9.86544639e-02
8.73730123e-01 5.06068170e-01 3.30927074e-01 4.52935517e-01
1.24337986e-01 -7.48406231e-01 8.51496756e-01 3.78784835e-01
5.87811232e-01 -7.53896654e-01 7.33168006e-01 4.45144266e-01
-1.01546931e+00 -2.31995761e-01 -1.11006290e-01 3.24014157e-01
4.27951872e-01 4.49670523e-01 -5.73789120e-01 5.02558887e-01
7.83697367e-01 6.83564603e-01 -6.77818835e-01 1.10083234e+00
-8.95214379e-02 7.88213432e-01 -3.11283797e-01 4.18479800e-01
5.80881715e-01 -1.01004049e-01 7.10288465e-01 1.29752779e+00
4.21850197e-02 1.62298083e-01 1.49544641e-01 9.06143546e-01
-2.75396053e-02 -3.16500366e-01 -5.85250139e-01 3.10511738e-01
8.96538422e-02 9.65436637e-01 -6.12052500e-01 -4.11391348e-01
-6.41409516e-01 9.95269060e-01 1.76513474e-02 6.52824104e-01
-1.30363226e+00 -2.02668488e-01 9.73811388e-01 2.76557028e-01
3.48949909e-01 -1.95243135e-01 -6.58730686e-01 -9.28503573e-01
7.50596076e-02 -6.22565567e-01 2.09750950e-01 -6.64493203e-01
-9.70858097e-01 4.99553233e-01 3.04185539e-01 -1.22582948e+00
-4.77992408e-02 -5.65382123e-01 -7.10956633e-01 5.04101932e-01
-1.75696313e+00 -1.36370158e+00 -3.60160708e-01 5.76295614e-01
1.08900285e+00 5.07091582e-02 8.53714272e-02 4.96202290e-01
-8.27788234e-01 6.38604939e-01 -4.51087728e-02 1.65067375e-01
6.54700696e-01 -8.93821180e-01 7.80679882e-01 1.24440336e+00
-3.32950234e-01 -4.66746725e-02 6.58667624e-01 -6.65157497e-01
-9.98377383e-01 -1.85908532e+00 4.73414719e-01 -6.57882750e-01
4.38544661e-01 -3.09596032e-01 -1.07300651e+00 7.63943255e-01
1.24598220e-01 2.42171325e-02 8.65980014e-02 -3.92580241e-01
-1.61554888e-01 -2.77777493e-01 -1.24036944e+00 8.69205594e-01
1.09791219e+00 -2.03283966e-01 -5.39737105e-01 1.91228569e-01
8.15231562e-01 -5.88178337e-01 -3.47728074e-01 6.64763212e-01
1.47088647e-01 -1.16072857e+00 1.10099673e+00 -4.80330288e-01
5.63331902e-01 -2.95498729e-01 -1.30927995e-01 -1.12922513e+00
-1.63519695e-01 -2.54309416e-01 1.42125115e-01 1.01514637e+00
5.63498557e-01 -8.42080176e-01 9.02429879e-01 8.18535805e-01
-8.02340031e-01 -6.32260919e-01 -1.01185727e+00 -7.56819844e-01
2.67052472e-01 -1.05832684e+00 6.08321190e-01 4.12460953e-01
-5.70999861e-01 -9.13981348e-03 -1.69717018e-02 5.84293902e-01
7.39680290e-01 -1.57603830e-01 8.90293300e-01 -7.98886478e-01
2.12001558e-02 -6.63907707e-01 -5.28213978e-01 -1.06324399e+00
3.59969497e-01 -7.17439234e-01 2.42430508e-01 -1.58774829e+00
9.76546556e-02 -2.94214457e-01 -3.14444155e-01 5.06247282e-01
-1.87829211e-01 5.11457980e-01 1.03476204e-01 -1.65135376e-02
-8.60156476e-01 4.80678439e-01 1.23411500e+00 -3.12328756e-01
-9.65712965e-02 2.69720286e-01 -5.45622349e-01 8.82241189e-01
1.11410272e+00 -3.70123208e-01 -3.83480489e-01 -6.96358502e-01
-2.31895268e-01 -1.47734985e-01 6.80280805e-01 -1.23950326e+00
2.93871969e-01 -2.82338828e-01 7.57184252e-02 -7.51901031e-01
4.64237869e-01 -7.70316422e-01 -5.92576861e-02 2.71925956e-01
-1.76410496e-01 -1.77659035e-01 4.48903084e-01 6.26802087e-01
-2.89054364e-01 1.34428188e-01 1.02441931e+00 4.14831080e-02
-1.32500672e+00 4.55508769e-01 -5.09451032e-01 2.89684862e-01
1.28567314e+00 -5.00243127e-01 -3.39584440e-01 -1.52881652e-01
-5.26781976e-01 6.40281796e-01 5.52805245e-01 7.68599749e-01
6.75861299e-01 -8.96771669e-01 -8.09638977e-01 3.74285012e-01
2.90998489e-01 4.10177678e-01 7.14935303e-01 9.65817153e-01
-4.61353898e-01 2.70267099e-01 -1.60721496e-01 -7.02653825e-01
-8.99429381e-01 3.46744359e-01 3.48940700e-01 1.33094177e-01
-7.48228669e-01 7.98010409e-01 7.11442471e-01 -2.95816541e-01
1.60654224e-02 -5.02466321e-01 -2.83964127e-02 -2.42978945e-01
3.40955049e-01 3.95300537e-01 1.86753005e-01 -1.03385043e+00
-2.88528889e-01 5.86175561e-01 -1.22066326e-01 1.06960453e-01
1.02854645e+00 -3.22414935e-01 2.65329719e-01 1.29569128e-01
1.25663733e+00 -3.17469448e-01 -1.78003407e+00 -9.85797718e-02
-1.18363149e-01 -2.69888550e-01 2.92377807e-02 -7.89094210e-01
-1.22362411e+00 9.01830852e-01 5.75976372e-01 -1.44483089e-01
1.16275620e+00 2.61985421e-01 1.34031439e+00 2.65587419e-01
2.48124227e-01 -1.12556922e+00 -3.09076123e-02 6.08548164e-01
6.95245802e-01 -1.60841823e+00 -5.75435817e-01 -3.87389898e-01
-1.07049525e+00 7.24849582e-01 8.64861012e-01 -1.85800478e-01
5.25029480e-01 1.38522640e-01 2.14387015e-01 -9.76965204e-02
-5.19701004e-01 -4.59358394e-01 2.47418806e-01 7.15897501e-01
-2.99475312e-01 1.18305780e-01 1.81068420e-01 8.14484537e-01
-4.54441011e-02 -1.94782495e-01 4.50461686e-01 6.57475591e-01
-5.32021761e-01 -9.18704212e-01 -3.62486988e-01 3.23386908e-01
-2.77067900e-01 1.83477581e-01 -9.23675597e-02 6.93107665e-01
3.95335704e-01 1.11088061e+00 2.41190776e-01 -3.79642934e-01
7.12966800e-01 6.63625775e-03 1.21537279e-02 -4.88084853e-01
-2.81899989e-01 -2.03706756e-01 1.48531720e-01 -8.30432355e-01
-2.50128925e-01 -7.78243482e-01 -1.50659359e+00 -1.48663148e-01
2.02909186e-01 -3.94702762e-01 5.03556013e-01 1.10835588e+00
4.41739798e-01 6.70351744e-01 5.07601261e-01 -9.64926302e-01
-1.25588462e-01 -5.84370077e-01 -2.17375234e-01 5.65535367e-01
6.10319316e-01 -6.38918757e-01 -6.93753734e-02 1.54902756e-01] | [8.48824691772461, -1.021630048751831] |
a33763ad-2db9-4a0f-ade2-de1de2e4f457 | mgh-metadata-guided-hypergraph-modeling-for | 2110.05886 | null | https://arxiv.org/abs/2110.05886v1 | https://arxiv.org/pdf/2110.05886v1.pdf | MGH: Metadata Guided Hypergraph Modeling for Unsupervised Person Re-identification | As a challenging task, unsupervised person ReID aims to match the same identity with query images which does not require any labeled information. In general, most existing approaches focus on the visual cues only, leaving potentially valuable auxiliary metadata information (e.g., spatio-temporal context) unexplored. In the real world, such metadata is normally available alongside captured images, and thus plays an important role in separating several hard ReID matches. With this motivation in mind, we propose~\textbf{MGH}, a novel unsupervised person ReID approach that uses meta information to construct a hypergraph for feature learning and label refinement. In principle, the hypergraph is composed of camera-topology-aware hyperedges, which can model the heterogeneous data correlations across cameras. Taking advantage of label propagation on the hypergraph, the proposed approach is able to effectively refine the ReID results, such as correcting the wrong labels or smoothing the noisy labels. Given the refined results, We further present a memory-based listwise loss to directly optimize the average precision in an approximate manner. Extensive experiments on three benchmarks demonstrate the effectiveness of the proposed approach against the state-of-the-art. | ['Jian Tian', 'Xi Li', 'Xintian Wu', 'Yiming Wu'] | 2021-10-12 | null | null | null | null | ['unsupervised-person-re-identification'] | ['computer-vision'] | [-1.31449960e-02 -4.86076325e-02 -1.03491008e-01 -5.64559162e-01
-5.82343280e-01 -5.14648318e-01 5.16366661e-01 5.33593237e-01
-5.51780701e-01 5.68665147e-01 1.62725151e-01 1.67919993e-01
-4.57534581e-01 -7.77003169e-01 -7.69208848e-01 -9.04958904e-01
1.21977657e-01 4.49438125e-01 1.48864806e-01 2.04006225e-01
2.41594553e-01 3.56058478e-01 -1.71999836e+00 -2.06457213e-01
9.51577127e-01 8.21524918e-01 1.85330838e-01 1.50807172e-01
-1.33112222e-01 6.90221727e-01 -3.78339797e-01 -8.16751063e-01
3.64378631e-01 -6.82488456e-02 -7.97316194e-01 2.06567928e-01
7.42028952e-01 -7.99078047e-02 -5.84139526e-01 1.38649786e+00
4.96054560e-01 4.09391731e-01 4.22568113e-01 -1.11109436e+00
-3.70510966e-01 5.52901268e-01 -9.05217707e-01 1.66226014e-01
4.13512349e-01 2.58290321e-02 8.29866707e-01 -8.54094863e-01
7.71346629e-01 1.26938725e+00 5.87259591e-01 2.37656191e-01
-1.37307322e+00 -8.66570294e-01 6.06988847e-01 5.34500062e-01
-1.97003126e+00 -5.13943434e-01 8.00513506e-01 -3.30231190e-01
2.40254328e-01 3.64744633e-01 5.36006689e-01 7.75577605e-01
-1.63523331e-01 8.05323839e-01 1.25724041e+00 -1.73198149e-01
7.82106519e-02 2.76225120e-01 3.58771116e-01 8.34828675e-01
4.29252177e-01 7.23019764e-02 -7.80828774e-01 -2.72024810e-01
4.13523525e-01 2.37885684e-01 -4.87327129e-01 -5.59460223e-01
-1.32469797e+00 3.30339462e-01 8.25145900e-01 -4.27662320e-02
-2.75955170e-01 -3.20558250e-02 2.72478700e-01 -1.45088926e-01
3.80370647e-01 1.43719018e-01 5.52941151e-02 2.96287119e-01
-9.67269659e-01 2.33556166e-01 4.96091127e-01 1.30866694e+00
1.23888445e+00 -5.27181089e-01 -3.30442905e-01 6.67031348e-01
2.26180777e-01 5.28556347e-01 -1.14047907e-01 -7.39313245e-01
5.18294632e-01 6.78467572e-01 2.43735969e-01 -1.32878017e+00
-5.42465746e-01 -5.72542489e-01 -9.80318010e-01 -2.63083845e-01
5.42971075e-01 1.30698204e-01 -9.42334831e-01 1.78896201e+00
7.81829834e-01 6.24062359e-01 -3.50190699e-01 1.16835308e+00
7.64355302e-01 3.13352555e-01 1.79109946e-01 -1.63084686e-01
1.32872868e+00 -8.08148086e-01 -6.60023808e-01 -6.67373165e-02
4.40052152e-01 -6.11143708e-01 5.35704970e-01 9.44014862e-02
-8.83195758e-01 -5.17848730e-01 -8.81042242e-01 -7.21652061e-02
-3.34810883e-01 8.14083517e-02 5.12453914e-01 5.38397968e-01
-1.12372446e+00 4.16950434e-01 -6.97262287e-01 -5.39493263e-01
3.57521862e-01 5.78935862e-01 -5.83246887e-01 -5.32662034e-01
-9.52173293e-01 5.12513399e-01 5.66588283e-01 3.60630929e-01
-4.33587372e-01 -6.17415071e-01 -1.00487792e+00 7.53192753e-02
6.93999648e-01 -9.34196591e-01 8.20388138e-01 -4.65466589e-01
-1.02301323e+00 8.08250546e-01 -5.98866582e-01 -1.15442358e-01
7.38396347e-01 7.04017654e-02 -2.02211961e-01 2.29110032e-01
2.13474140e-01 7.43975997e-01 7.24162877e-01 -1.65374029e+00
-9.79676723e-01 -6.33349657e-01 1.81070670e-01 2.64878869e-01
-3.98144603e-01 -1.45549387e-01 -1.12650931e+00 -5.66989481e-01
2.95554250e-01 -1.11555219e+00 -1.40474707e-01 -2.69824993e-02
-6.75158918e-01 -2.81953186e-01 4.75988150e-01 -6.18634343e-01
1.07353508e+00 -2.00811028e+00 1.36027291e-01 7.65880644e-01
4.95756269e-01 -7.89273009e-02 1.29508926e-02 2.64388084e-01
2.65293062e-01 -8.89495686e-02 -1.13859676e-01 -6.80844724e-01
-6.92228146e-04 3.47182155e-02 1.20717965e-01 1.00305688e+00
-2.30102047e-01 7.67779827e-01 -1.18408668e+00 -9.24057543e-01
1.68724120e-01 5.15152931e-01 -2.62301862e-01 7.50960037e-02
1.42553523e-01 8.57370555e-01 -6.76684737e-01 7.45714009e-01
1.00091338e+00 -1.87092051e-01 8.06762278e-02 -2.75195003e-01
-1.21751457e-01 -7.47281238e-02 -1.47913551e+00 1.95846581e+00
-1.73921317e-01 3.41034263e-01 1.82318047e-01 -7.82054305e-01
7.83616304e-01 -5.83282858e-02 4.96947229e-01 -6.21557057e-01
7.66205713e-02 -8.99689049e-02 -5.65442741e-01 -2.47409716e-01
6.47534788e-01 3.47730875e-01 8.78323149e-03 2.30088264e-01
-2.73082286e-01 6.67523742e-01 2.37978801e-01 3.53248358e-01
9.22711432e-01 -2.03687558e-03 2.58712154e-02 -2.77834207e-01
8.08045447e-01 6.72050053e-03 8.88754249e-01 9.94778216e-01
-9.25810412e-02 4.35964435e-01 1.00482069e-01 -2.35291556e-01
-7.99115658e-01 -1.00505769e+00 -2.02231571e-01 8.67929757e-01
9.87178981e-01 -4.82777029e-01 -8.05586219e-01 -6.65337384e-01
1.77937061e-01 2.33946741e-01 -5.28060436e-01 -6.43516928e-02
-5.94413817e-01 -6.96615815e-01 2.90206403e-01 3.41599286e-01
5.30869424e-01 -5.55845857e-01 -6.57065809e-02 9.26402658e-02
-3.20702493e-01 -1.14333403e+00 -6.68659747e-01 -3.44664782e-01
-5.97990274e-01 -1.14620697e+00 -8.19136381e-01 -8.30297410e-01
1.14198196e+00 7.46961653e-01 8.88494790e-01 3.54962558e-01
-2.58090645e-01 5.18310964e-01 -2.42209896e-01 -6.12308793e-02
1.71188921e-01 1.10276975e-01 1.42566383e-01 4.82400417e-01
4.21142131e-01 -4.16748166e-01 -8.87929857e-01 4.53715205e-01
-6.42453194e-01 6.69467598e-02 6.91571534e-01 9.23239052e-01
8.05170298e-01 2.64010936e-01 4.07875508e-01 -1.19148242e+00
1.66577041e-01 -4.42591071e-01 -7.77731538e-01 4.93583620e-01
-6.82899058e-01 9.94233489e-02 3.77312720e-01 -2.98961550e-01
-1.37380028e+00 2.72509545e-01 3.83531272e-01 -5.45662940e-01
-1.66103363e-01 3.51682246e-01 -4.92261827e-01 -3.96862090e-01
1.24621034e-01 2.34105691e-01 -4.34203088e-01 -6.15615547e-01
4.78215247e-01 5.34919381e-01 9.41148400e-01 -7.61282504e-01
1.07237995e+00 8.41039479e-01 1.85752571e-01 -5.84851265e-01
-7.01491117e-01 -1.12570524e+00 -7.74601817e-01 -3.60277206e-01
7.22947598e-01 -1.04076886e+00 -1.03803539e+00 3.05537969e-01
-9.80872691e-01 2.50117868e-01 5.11668101e-02 4.96659607e-01
-2.16459960e-01 5.49425662e-01 -4.75053489e-01 -8.04621220e-01
-1.40208736e-01 -1.04864049e+00 1.25920832e+00 4.70184594e-01
1.79624334e-01 -7.14710057e-01 -8.24467614e-02 4.65014756e-01
-9.99943987e-02 1.53881952e-01 7.83794582e-01 -5.70003092e-01
-1.08639252e+00 -4.02879685e-01 -6.64451599e-01 -3.13002139e-01
-8.77019241e-02 -4.89095807e-01 -1.11078179e+00 -4.66333926e-01
-4.33337450e-01 1.09159112e-01 1.04767179e+00 4.50968407e-02
1.13781083e+00 -9.68173072e-02 -7.48780310e-01 8.03319871e-01
1.29584491e+00 -3.19067866e-01 3.28237265e-01 2.48685807e-01
1.14196110e+00 8.50252330e-01 7.70000756e-01 6.20463073e-01
8.45508277e-01 8.96203279e-01 2.84198254e-01 -3.93343009e-02
-9.53500867e-02 -3.72541457e-01 -8.08050185e-02 5.62163413e-01
-6.80203214e-02 -2.68645376e-01 -7.37091064e-01 6.34864926e-01
-2.08347344e+00 -7.85157681e-01 -2.44000643e-01 2.46035051e+00
5.48940778e-01 -1.87865078e-01 5.92423696e-03 8.14346131e-03
1.32828999e+00 1.82152569e-01 -6.35023952e-01 4.14928168e-01
-8.88072327e-02 -2.37178355e-01 8.24085414e-01 2.21150503e-01
-1.30430162e+00 7.48781919e-01 4.54232931e+00 7.55432785e-01
-6.18236601e-01 2.13863775e-02 2.43720368e-01 -1.55648135e-03
-2.05339581e-01 1.33664459e-01 -1.01522088e+00 5.02618968e-01
4.07517642e-01 -2.41678745e-01 4.44995552e-01 6.10616088e-01
2.41212785e-01 -2.78863668e-01 -1.23975456e+00 1.18354058e+00
1.89378172e-01 -1.02378023e+00 -1.08767986e-01 1.94503948e-01
9.10007775e-01 -4.54790533e-01 2.93764696e-02 8.28180015e-02
1.52544454e-01 -5.69632590e-01 8.01359773e-01 7.98174620e-01
6.39991760e-01 -9.84735668e-01 8.34954023e-01 3.25680941e-01
-1.62211764e+00 -2.00310722e-02 -5.53145587e-01 3.32247615e-01
1.66899040e-01 6.18793428e-01 -6.30203009e-01 8.53942335e-01
7.93678701e-01 7.85026312e-01 -7.32483029e-01 1.54418278e+00
-2.17421457e-01 2.39002645e-01 -3.30831438e-01 2.97764599e-01
-6.46349043e-02 -1.37853131e-01 6.58754110e-01 1.33359337e+00
1.11337386e-01 3.37117523e-01 5.74744225e-01 7.16086924e-01
-3.70073020e-01 2.76474357e-01 -3.13109577e-01 4.77398694e-01
6.90098524e-01 1.28116977e+00 -8.71647537e-01 -2.34164804e-01
-4.48496103e-01 1.02355480e+00 4.85582441e-01 5.45221686e-01
-5.51198900e-01 -1.72321439e-01 6.58997476e-01 1.64814159e-01
8.90466571e-02 -7.03432783e-02 -1.34029418e-01 -1.20931590e+00
3.08775812e-01 -4.13725138e-01 6.27758741e-01 -3.87552470e-01
-1.41826487e+00 2.20387965e-01 1.20511085e-01 -1.17178130e+00
8.46040398e-02 4.21495028e-02 -3.88807833e-01 8.83471847e-01
-1.83805525e+00 -1.29501748e+00 -7.78659999e-01 7.01596737e-01
2.13257715e-01 2.22409517e-01 3.29286933e-01 5.35433769e-01
-8.27042580e-01 9.34466243e-01 1.27503231e-01 1.92718312e-01
1.07372797e+00 -1.19351149e+00 1.89803660e-01 1.01604891e+00
1.48332361e-02 7.96521902e-01 5.04413605e-01 -8.31044137e-01
-1.53787053e+00 -1.44206977e+00 8.18506896e-01 -1.92320019e-01
3.85292262e-01 -3.67586255e-01 -9.94115412e-01 4.81516510e-01
-6.46111444e-02 1.34361640e-01 4.43153799e-01 1.08380280e-01
-3.64942282e-01 -3.64254832e-01 -1.07519996e+00 3.95951033e-01
1.40582180e+00 -4.84433115e-01 -2.83156604e-01 4.18686777e-01
4.76217061e-01 -5.04996538e-01 -7.76065707e-01 2.66198903e-01
4.45609272e-01 -8.01081896e-01 1.09429121e+00 -2.28201538e-01
-2.54290253e-01 -5.47689259e-01 2.93500155e-01 -1.03939950e+00
-3.31446946e-01 -5.40992558e-01 1.85462534e-02 1.65818429e+00
4.95192222e-02 -5.33178985e-01 9.69028294e-01 9.86407578e-01
4.66003269e-02 -2.70455211e-01 -8.64156425e-01 -8.61140907e-01
-5.87700486e-01 -8.48506317e-02 8.44271004e-01 8.75779390e-01
-1.48027092e-01 -1.64543733e-01 -5.87263465e-01 7.79625535e-01
1.19158030e+00 2.44824842e-01 9.21505034e-01 -1.49102378e+00
1.02801323e-02 -2.81561375e-01 -8.00788343e-01 -1.12393343e+00
5.95380604e-01 -9.93935645e-01 2.52432674e-01 -1.50703907e+00
6.27780914e-01 -9.06402588e-01 -5.39535999e-01 3.06292355e-01
-6.49110675e-01 2.13908792e-01 2.44459599e-01 4.43697393e-01
-9.94228542e-01 3.77967298e-01 9.33350623e-01 -4.69132900e-01
-1.86744899e-01 5.18865138e-02 -4.99254167e-01 6.84843957e-01
3.75549614e-01 -6.85441136e-01 -3.58973145e-01 -4.27677482e-01
9.91488993e-02 -1.20581321e-01 5.18996119e-01 -9.06392395e-01
9.16600347e-01 -3.51339243e-02 4.33713228e-01 -5.47079921e-01
2.86662340e-01 -9.58798110e-01 2.17885390e-01 -7.42199495e-02
-3.27901334e-01 -9.39408019e-02 -1.58790618e-01 1.12550688e+00
-1.93753704e-01 -1.84496343e-02 5.60727417e-01 6.78952932e-02
-9.00487542e-01 5.99936247e-01 3.00372630e-01 -1.04890853e-01
9.71067369e-01 -1.38509452e-01 -4.93734986e-01 -2.36189321e-01
-4.92463082e-01 7.56894529e-01 8.27830374e-01 3.30817342e-01
5.52405238e-01 -1.21749365e+00 -7.19348729e-01 5.28430156e-02
5.34370482e-01 1.64773896e-01 5.97047210e-01 9.05053616e-01
-6.84539229e-02 2.90084571e-01 1.49371162e-01 -7.27049589e-01
-1.32813811e+00 8.93917501e-01 5.21175191e-02 -9.36112180e-02
-8.33097994e-01 7.59047687e-01 5.42322755e-01 -1.20759122e-01
5.44689178e-01 1.48334503e-01 -3.80071133e-01 1.79057941e-01
7.28402972e-01 5.84202886e-01 9.70224291e-02 -9.67028737e-01
-4.57810104e-01 8.00368071e-01 -1.86335251e-01 1.09336779e-01
9.78349745e-01 -6.79542303e-01 -9.28067490e-02 7.28434995e-02
1.08519495e+00 9.53095034e-02 -1.11882353e+00 -6.98793054e-01
2.29872346e-01 -9.33905363e-01 9.75885093e-02 -3.35805029e-01
-1.21387482e+00 5.20256281e-01 5.63734174e-01 -2.64304072e-01
1.21307290e+00 3.41158323e-02 9.18395579e-01 4.54444468e-01
7.68578351e-01 -9.50672925e-01 -3.74962479e-01 -1.10573873e-01
3.25288743e-01 -1.36271644e+00 2.29718283e-01 -8.01995873e-01
-3.60151589e-01 7.49371409e-01 5.06406605e-01 2.61968255e-01
3.90948355e-01 -1.56496346e-01 -7.05558658e-02 -1.65834755e-01
-2.49356970e-01 -5.08321166e-01 4.17007178e-01 5.04123569e-01
-1.06255695e-01 4.25281823e-02 -2.54611999e-01 4.16857243e-01
6.04548529e-02 -2.47916624e-01 2.36017838e-01 8.13594520e-01
-2.22360745e-01 -1.07602608e+00 -6.43796861e-01 4.06883001e-01
-1.53035849e-01 1.03597350e-01 -2.94104159e-01 6.62192762e-01
3.52459848e-01 1.21824336e+00 -1.87344521e-01 -3.61260623e-01
4.76318955e-01 -1.97068244e-01 2.63312280e-01 -3.71602595e-01
-4.69469488e-01 4.60633971e-02 -1.37104362e-01 -6.82078838e-01
-5.50223529e-01 -9.38239396e-01 -1.05620408e+00 -5.42151868e-01
-4.20296848e-01 1.72223046e-01 5.60832143e-01 8.52066875e-01
3.60971868e-01 1.34280026e-01 7.07255840e-01 -8.47572565e-01
-2.21299872e-01 -4.99199390e-01 -6.91410542e-01 5.95014036e-01
3.93585265e-01 -9.39447165e-01 -2.17112094e-01 1.17803231e-01] | [14.838753700256348, 1.0297423601150513] |
f3765c95-7d1e-4810-ba3c-d68c706b5e9f | towards-good-practices-for-video-object | 1909.13583 | null | https://arxiv.org/abs/1909.13583v1 | https://arxiv.org/pdf/1909.13583v1.pdf | Towards Good Practices for Video Object Segmentation | Semi-supervised video object segmentation is an interesting yet challenging task in machine learning. In this work, we conduct a series of refinements with the propagation-based video object segmentation method and empirically evaluate their impact on the final model performance through ablation study. By taking all the refinements, we improve the space-time memory networks to achieve a Overall of 79.1 on the Youtube-VOS Challenge 2019. | ['Dongdong Yu', 'Jian Wang', 'Minghui Dong', 'Jie Shao', 'Kaihui Zhou', 'Hengkai Guo', 'Changhu Wang', 'Yuanyuan Huang', 'Kai Su'] | 2019-09-30 | null | null | null | null | ['one-shot-visual-object-segmentation'] | ['computer-vision'] | [ 1.84264839e-01 -4.48485278e-02 -4.41170990e-01 -4.53495115e-01
-7.40536690e-01 -4.38671082e-01 2.68841088e-01 -3.20968390e-01
-9.79903281e-01 5.60709000e-01 -1.68678612e-01 -4.15483296e-01
3.58803004e-01 -2.54495829e-01 -9.70880747e-01 -3.62440079e-01
-2.09313944e-01 2.47741029e-01 7.61138856e-01 2.24573001e-01
8.71341154e-02 1.89264625e-01 -1.03163528e+00 2.97570199e-01
7.85156429e-01 1.12350905e+00 -7.95613304e-02 8.92053306e-01
-2.25506015e-02 9.99675810e-01 -5.77244937e-01 -3.97754580e-01
5.96523285e-01 -2.67310202e-01 -1.26501238e+00 9.85601768e-02
6.62086308e-01 -6.91474259e-01 -4.45845932e-01 1.14021385e+00
3.48285794e-01 2.95419157e-01 5.85475385e-01 -1.41062045e+00
-2.77922779e-01 8.55501056e-01 -6.62903905e-01 6.32184505e-01
6.72523072e-03 2.33277336e-01 9.12641287e-01 -7.08380580e-01
6.93615735e-01 1.00945151e+00 8.03341627e-01 7.08744586e-01
-8.86510372e-01 -8.10391963e-01 5.37078321e-01 3.85352701e-01
-1.50776458e+00 -4.76808667e-01 6.39199257e-01 -2.53378898e-01
1.02754724e+00 8.01288486e-02 7.26414502e-01 9.84738708e-01
-2.56818384e-01 1.34557569e+00 7.36523211e-01 -3.78407873e-02
2.72051245e-01 -1.29872590e-01 4.26966608e-01 7.87224889e-01
2.32756063e-01 -2.15668917e-01 -3.54798108e-01 1.30607620e-01
7.14579642e-01 -7.30128735e-02 -1.56111404e-01 -7.93486089e-02
-8.96591604e-01 7.11873949e-01 5.73914170e-01 3.83880064e-02
-1.28897130e-01 5.10995388e-01 4.44324285e-01 1.47478253e-01
6.22071862e-01 4.70388323e-01 -5.92597425e-01 -4.15665120e-01
-1.46204603e+00 2.30825484e-01 7.05901742e-01 1.14938843e+00
4.55378294e-01 3.25397789e-01 -3.01946521e-01 5.90672851e-01
3.82536054e-01 2.31653303e-01 2.25533769e-01 -1.16106260e+00
5.63388824e-01 4.66009796e-01 -5.54557778e-02 -4.01420325e-01
-5.34530044e-01 -4.29187864e-01 -5.67823589e-01 -8.05654451e-02
4.77237076e-01 -7.46165156e-01 -1.65187216e+00 1.34101760e+00
2.22720146e-01 7.01757729e-01 -4.34600860e-02 1.09905088e+00
1.03835070e+00 6.27906382e-01 5.40905893e-01 -2.90935487e-02
9.27711248e-01 -1.83765709e+00 -4.40790445e-01 -4.06975359e-01
7.69511282e-01 -4.69260901e-01 8.60592663e-01 2.74682909e-01
-1.09313428e+00 -4.32431251e-01 -1.03799105e+00 2.07225941e-02
-7.01267347e-02 -4.87612076e-02 8.20904911e-01 6.04792118e-01
-1.00356364e+00 6.65101349e-01 -1.25419223e+00 -1.35667652e-01
1.07502830e+00 5.62474310e-01 1.85343381e-02 5.76822273e-02
-9.13568258e-01 4.46479827e-01 3.99499863e-01 1.19496748e-01
-1.02728891e+00 -8.16122711e-01 -6.61366999e-01 -3.37503217e-02
6.89145863e-01 -4.78138089e-01 1.36635160e+00 -9.94680882e-01
-1.31844521e+00 7.19981790e-01 -1.90299317e-01 -9.10130978e-01
7.79267251e-01 -2.92365491e-01 -1.63199231e-01 2.19810337e-01
-1.68722346e-01 1.44112694e+00 7.51870513e-01 -1.03623152e+00
-7.66291440e-01 -7.07500577e-02 2.20482498e-01 1.78983554e-01
-3.60995293e-01 6.09322041e-02 -1.08372903e+00 -6.90701365e-01
-1.22531734e-01 -1.32619727e+00 -5.82584739e-01 -1.25773281e-01
-5.66183329e-01 -1.03535794e-01 7.12123573e-01 -6.98573291e-01
1.14182007e+00 -2.08125496e+00 6.50033578e-02 9.85595807e-02
2.17814252e-01 5.36685705e-01 -3.55846405e-01 -3.51174563e-01
2.20443621e-01 6.03785813e-01 -3.57215971e-01 -6.65176094e-01
-2.11287692e-01 3.57627906e-02 1.19262099e-01 2.40140408e-01
2.08552048e-01 1.34495389e+00 -5.26219428e-01 -7.27757990e-01
-1.08906671e-01 4.40842211e-01 -8.19495976e-01 7.15773553e-02
-4.15515006e-01 3.48889321e-01 -3.77543151e-01 8.41853678e-01
5.31536162e-01 -3.57634425e-01 -1.85392588e-01 -2.00213835e-01
2.89177746e-01 7.09808022e-02 -8.00784528e-01 2.00700974e+00
1.40659958e-01 9.10184979e-01 -1.39109433e-01 -7.96992481e-01
3.14308941e-01 1.06884010e-01 6.39668643e-01 -6.21034086e-01
2.82083273e-01 -1.04427516e-01 1.28337234e-01 -4.92149174e-01
5.25606573e-01 2.38730207e-01 2.22986683e-01 2.39460528e-01
2.09253222e-01 2.17774376e-01 4.64767754e-01 2.11557016e-01
9.47224975e-01 2.90697217e-01 -3.66781533e-01 -3.71406764e-01
4.84212823e-02 3.51068914e-01 6.21301353e-01 7.99686372e-01
-5.31074286e-01 9.62125838e-01 5.16580522e-01 -3.81381720e-01
-9.41954494e-01 -8.49239528e-01 6.70244098e-02 1.02453268e+00
3.79729956e-01 -5.32767653e-01 -1.18269610e+00 -1.18951416e+00
-2.07584158e-01 5.73471427e-01 -6.35251164e-01 -1.11637168e-01
-6.49632812e-01 -1.03330994e+00 7.70410776e-01 1.00128603e+00
7.97418296e-01 -1.06797421e+00 -6.49040937e-01 -4.52105105e-02
-1.47179723e-01 -1.64395952e+00 -6.06803834e-01 5.35126552e-02
-1.02834713e+00 -1.17046452e+00 -9.75571275e-01 -9.15379465e-01
6.74278319e-01 3.36146116e-01 1.02871013e+00 2.46883303e-01
-1.99074805e-01 2.20184550e-01 -5.05186915e-01 -3.99216413e-01
-1.57713488e-01 7.23079503e-01 -2.54093677e-01 -1.89254776e-01
2.74805635e-01 -2.23947704e-01 -6.31132782e-01 2.17745870e-01
-8.73760462e-01 1.81199789e-01 2.85891563e-01 4.31830108e-01
7.18074024e-01 1.34742260e-01 2.49542922e-01 -1.21306121e+00
3.00966889e-01 -3.91391009e-01 -6.13740742e-01 1.01097137e-01
-7.20604777e-01 -1.99592352e-01 3.51291627e-01 -6.16511822e-01
-6.99933946e-01 2.70710111e-01 -2.86017805e-01 -6.57258391e-01
-2.99378708e-02 4.84062284e-01 1.18976414e-01 -3.89435351e-01
3.05672705e-01 1.23088591e-01 -2.84166187e-01 -5.20480096e-01
9.36334282e-02 4.23805445e-01 2.61614531e-01 -2.29582146e-01
7.61989653e-01 3.58256370e-01 -3.12678605e-01 -6.23720288e-01
-1.00446093e+00 -3.44425589e-01 -5.61269164e-01 -1.54899105e-01
1.07162917e+00 -1.14823699e+00 -3.84988874e-01 7.02157140e-01
-1.00682127e+00 -9.50186253e-01 -2.04409748e-01 4.27904874e-01
-2.78401881e-01 1.69171736e-01 -8.14768255e-01 -3.14733207e-01
-4.13898647e-01 -1.30687463e+00 8.01450431e-01 3.73972505e-01
-1.38006464e-01 -8.98127794e-01 -2.67209351e-01 5.80065906e-01
4.62099373e-01 4.69534053e-03 4.23415601e-01 -8.54264975e-01
-6.86930835e-01 -1.90786570e-02 -4.63724971e-01 5.19418418e-01
-1.69151291e-01 1.33922338e-01 -7.35397518e-01 -3.86982501e-01
-2.55073071e-01 -2.33567521e-01 1.38795102e+00 5.03286600e-01
1.41495335e+00 -1.97113931e-01 -3.26686651e-01 1.02166247e+00
1.30057406e+00 2.67546266e-01 6.37551069e-01 2.66422302e-01
1.15788448e+00 1.33820310e-01 5.35099030e-01 9.16118845e-02
3.28799486e-01 3.54652435e-01 4.14838910e-01 -3.06893378e-01
-2.71089762e-01 -1.41067030e-02 1.99196801e-01 5.64093411e-01
-1.27351001e-01 -4.00049955e-01 -1.02755463e+00 6.42121315e-01
-1.71943665e+00 -4.85116571e-01 3.60091403e-02 1.67599988e+00
7.41001964e-01 6.50937438e-01 2.45903596e-01 -1.95654631e-01
6.55597746e-01 3.46064806e-01 -7.97892749e-01 -8.77107009e-02
2.97435056e-02 4.35503609e-02 8.24203849e-01 4.40465271e-01
-1.58592546e+00 1.58385837e+00 7.60310030e+00 7.78074205e-01
-1.23083889e+00 1.77063525e-01 8.82769406e-01 -4.68922764e-01
4.92361747e-02 -2.10837916e-01 -9.83364880e-01 3.35644335e-01
1.03185713e+00 5.41307665e-02 4.46970016e-01 6.66192412e-01
-1.04258750e-02 -1.39010414e-01 -1.05774856e+00 8.47529590e-01
8.98506120e-02 -1.38900149e+00 1.43136129e-01 -1.80202752e-01
1.03337848e+00 6.82404995e-01 -6.94233403e-02 3.95265609e-01
1.48691803e-01 -1.01901710e+00 7.30607152e-01 2.75365449e-02
6.10358179e-01 -6.32573247e-01 7.10052252e-01 -3.28982323e-02
-9.82757390e-01 4.32182290e-02 -3.27413566e-02 1.24556087e-01
4.34622288e-01 1.31557420e-01 -5.19885778e-01 1.52460369e-03
7.69219697e-01 6.96670353e-01 -8.78064275e-01 1.49743187e+00
-8.16981867e-02 1.32685864e+00 -6.44834936e-01 1.01083465e-01
6.62841380e-01 1.45584166e-01 5.09919167e-01 1.44864917e+00
-1.81914508e-01 3.72957923e-02 3.29706579e-01 7.94698298e-01
-5.69361389e-01 -7.19870776e-02 -3.03510502e-02 -8.88036564e-02
5.52032478e-02 1.06581032e+00 -1.31955671e+00 -4.70281988e-01
-3.68390560e-01 9.84001517e-01 2.33061701e-01 8.11908305e-01
-1.41743517e+00 -1.38138518e-01 6.37812436e-01 1.22626964e-02
6.32774115e-01 -3.22207928e-01 -3.20537210e-01 -1.05259514e+00
5.81042208e-02 -6.24280691e-01 2.00128242e-01 -6.16358757e-01
-7.32488215e-01 7.98505187e-01 3.97980586e-02 -8.62565160e-01
1.04283132e-01 -3.12567860e-01 -5.36204338e-01 3.05250973e-01
-1.76672757e+00 -9.88389134e-01 -2.37931624e-01 3.79524112e-01
8.64861131e-01 -8.60745609e-02 3.40578496e-01 6.23753548e-01
-1.09160876e+00 9.04170454e-01 -2.04802275e-01 5.88907540e-01
3.92125607e-01 -1.04150343e+00 9.46955919e-01 1.03312731e+00
5.44857085e-01 2.17346415e-01 4.93018299e-01 -7.10469425e-01
-1.06988311e+00 -1.56147182e+00 2.09636644e-01 -2.19412521e-01
4.63879645e-01 -3.48874867e-01 -8.78441393e-01 9.78595972e-01
2.32146367e-01 2.58920372e-01 3.90170157e-01 -1.23260111e-01
-2.54360408e-01 1.39142781e-01 -9.50383723e-01 7.83217967e-01
1.27842247e+00 -2.43081868e-01 -1.62912682e-01 2.68094093e-01
1.15392148e+00 -5.08443177e-01 -7.71891594e-01 6.63660884e-01
2.94608861e-01 -6.01896405e-01 8.43712747e-01 -1.09294486e+00
4.75658864e-01 -2.55124811e-02 1.75083295e-01 -9.19072688e-01
-8.45342129e-02 -7.55572915e-01 -2.78131604e-01 1.09354734e+00
8.14527988e-01 -2.71147341e-01 1.54922271e+00 8.61808419e-01
1.06125539e-02 -1.05990958e+00 -7.83084810e-01 -8.43794346e-01
9.21102539e-02 -4.92428988e-01 1.31750733e-01 7.23689318e-01
-4.03926492e-01 2.99166441e-01 -3.65789741e-01 -1.89036459e-01
4.96291459e-01 -7.04801157e-02 5.58504462e-01 -8.21955740e-01
-1.71401471e-01 -5.75182915e-01 -2.91334957e-01 -1.57080579e+00
2.42593944e-01 -8.13182414e-01 5.57970218e-02 -1.62075758e+00
2.61229396e-01 -3.63259524e-01 -5.24715960e-01 5.99292696e-01
-2.42466196e-01 8.61862004e-01 4.99666482e-01 3.38745505e-01
-1.46092796e+00 4.91538465e-01 1.18528664e+00 -3.66533279e-01
-4.62625414e-01 1.41304642e-01 -5.85681856e-01 9.94427681e-01
9.18686271e-01 -8.06018054e-01 -3.32597554e-01 -8.95644426e-01
-1.49062529e-01 -2.21956715e-01 2.71915913e-01 -1.09932017e+00
2.64961600e-01 1.35805523e-02 1.50026068e-01 -4.32724118e-01
3.50985616e-01 -7.17052460e-01 -2.72491634e-01 4.53940362e-01
-3.73116255e-01 -1.19735099e-01 4.82109964e-01 4.42949176e-01
-2.26650104e-01 -1.44074589e-01 7.50384510e-01 6.02187738e-02
-1.09221280e+00 5.13679862e-01 -3.97016823e-01 4.56453681e-01
1.18990505e+00 -3.69028062e-01 -1.49499059e-01 -2.11270005e-01
-7.93002486e-01 5.51798284e-01 3.06127667e-01 5.21308899e-01
5.06213427e-01 -9.38799143e-01 -6.58596039e-01 -4.72185500e-02
-2.57168889e-01 3.15768808e-01 9.35687944e-02 9.13034320e-01
-9.37880158e-01 5.58122396e-01 -2.09577516e-01 -6.83650672e-01
-1.21192133e+00 3.03795964e-01 2.97196925e-01 -1.36040539e-01
-6.23315394e-01 1.24434590e+00 -1.12347290e-01 -1.18475094e-01
6.70150042e-01 -4.02899176e-01 -4.19656098e-01 -1.11673690e-01
3.15157890e-01 5.46578348e-01 -1.04152381e-01 -5.38088441e-01
-4.47777063e-01 4.57154751e-01 -2.15795472e-01 3.98660526e-02
1.26615095e+00 -2.02559941e-02 2.41197139e-01 2.59541720e-01
1.45659244e+00 -4.52876955e-01 -1.76285625e+00 -1.61364406e-01
1.36401996e-01 -1.04193613e-01 2.76257575e-01 -7.01628685e-01
-1.84147179e+00 5.92801630e-01 4.97011572e-01 -6.34808466e-02
1.05984151e+00 -1.42807374e-02 1.19440508e+00 5.61146915e-01
1.25953272e-01 -1.05243707e+00 -3.01321596e-01 4.34977144e-01
2.99753308e-01 -1.56845665e+00 1.76188663e-01 -6.76830888e-01
-7.19327509e-01 6.65525198e-01 9.39812660e-01 -3.16065013e-01
8.09099615e-01 4.74256128e-01 1.27148300e-01 -7.75423571e-02
-6.48563504e-01 -2.06972182e-01 2.93684989e-01 2.21917644e-01
2.52457500e-01 -3.55100751e-01 -2.55408525e-01 6.26397789e-01
9.03825983e-02 2.85408139e-01 4.10031110e-01 8.77131522e-01
-2.07245871e-01 -7.63362706e-01 2.67313719e-01 5.44409275e-01
-8.54010642e-01 -3.41141760e-01 -3.99762034e-01 7.28744209e-01
-2.09331512e-01 8.18114638e-01 2.56616563e-01 -4.17299539e-01
1.42715082e-01 -8.48129094e-02 4.16789025e-01 -3.89625192e-01
-6.49303138e-01 3.07386201e-02 2.07137212e-01 -6.89619303e-01
-6.65535927e-01 -5.12789607e-01 -1.60891187e+00 -1.85402468e-01
-5.54171562e-01 3.71026099e-02 7.95352936e-01 1.13670599e+00
3.79943490e-01 7.54580379e-01 1.18337832e-01 -9.07833099e-01
-2.72316158e-01 -7.90085554e-01 -1.22879006e-01 3.07038575e-01
2.22166732e-01 -3.63472790e-01 -2.00879335e-01 1.44521371e-01] | [9.24412727355957, -0.0812799409031868] |
8397195a-520a-47aa-b252-779158d94355 | morphology-is-not-just-a-naive-bayes-unimelb | null | null | https://aclanthology.org/2022.sigmorphon-1.25 | https://aclanthology.org/2022.sigmorphon-1.25.pdf | Morphology is not just a naive Bayes – UniMelb Submission to SIGMORPHON 2022 ST on Morphological Inflection | The paper describes the Flexica team’s submission to the SIGMORPHON 2022 Shared Task 1 Part 1: Typologically Diverse Morphological Inflection. Our team submitted a nonneural system that extracted transformation patterns from alignments between a lemma and inflected forms. For each inflection category, we chose a pattern based on its abstractness score. The system outperformed the non-neural baseline, the extracted patterns covered a substantial part of possible inflections. However, we discovered that such score that does not account for all possible combinations of string segments as well as morphosyntactic features is not sufficient for a certain proportion of inflection cases. | ['Ekaterina Vylomova', 'Andreas Sherbakov'] | null | null | null | null | naacl-sigmorphon-2022-7 | ['morphological-inflection'] | ['natural-language-processing'] | [ 1.36452824e-01 2.34935462e-01 -2.04325289e-01 -3.99249434e-01
-4.77892756e-01 -1.12477732e+00 5.89622617e-01 2.46398866e-01
-6.75171018e-01 7.96880424e-01 4.10970032e-01 -5.34296870e-01
-8.50379393e-02 -6.84543848e-01 -6.53250992e-01 -2.29685068e-01
6.78765634e-03 7.66111135e-01 1.00230068e-01 -4.51588452e-01
3.08528274e-01 4.99393731e-01 -1.14763582e+00 5.50794125e-01
8.87627184e-01 4.18064952e-01 -8.96722600e-02 2.18528315e-01
-2.95493454e-01 4.62902114e-02 -8.30269277e-01 -6.78426266e-01
2.96866804e-01 -3.90300363e-01 -8.97382796e-01 -5.33738792e-01
1.02324450e+00 3.07065964e-01 8.15262571e-02 1.27459252e+00
2.76903689e-01 9.94054601e-02 1.08206177e+00 -7.96007097e-01
-1.18255162e+00 1.60778761e+00 -4.39675421e-01 6.80851936e-01
1.03215016e-01 1.63367093e-01 1.38558388e+00 -1.06016624e+00
8.64350915e-01 1.23745608e+00 8.79728734e-01 3.63408357e-01
-1.42820919e+00 -6.44234717e-01 5.43432087e-02 1.24236465e-01
-1.07504785e+00 -2.79702932e-01 7.15164602e-01 -1.92174435e-01
1.75113750e+00 1.44714564e-01 9.96945679e-01 1.00023460e+00
3.63113523e-01 7.14352548e-01 1.08505976e+00 -5.18871844e-01
5.46743758e-02 -2.30433717e-01 6.95787370e-01 3.33319634e-01
4.43969786e-01 2.76415706e-01 -3.34905982e-01 -1.10086255e-01
6.95889771e-01 -8.49514544e-01 -1.51046142e-01 3.29523861e-01
-1.43510258e+00 5.53282440e-01 3.93446892e-01 7.47138321e-01
-4.19763833e-01 4.85041514e-02 5.24233639e-01 5.89384615e-01
1.04951367e-01 1.13087094e+00 -8.23709071e-01 -1.29310057e-01
-8.80190909e-01 3.00129712e-01 6.77982748e-01 7.79099226e-01
6.58358634e-01 4.66588497e-01 7.64782215e-03 1.23945010e+00
-3.06052059e-01 2.25704893e-01 8.01835299e-01 -8.67866397e-01
3.93145382e-01 6.51675582e-01 -3.99497241e-01 -7.38874078e-01
-6.18031144e-01 -3.55870903e-01 -6.88368917e-01 9.77290445e-04
6.75623596e-01 -2.04786435e-01 -1.00783980e+00 2.06988406e+00
-2.30804890e-01 -5.94902158e-01 -4.78623733e-02 5.45524895e-01
6.53905988e-01 4.97402042e-01 2.18495592e-01 -1.44484088e-01
1.27710402e+00 -5.21358609e-01 -7.23269761e-01 -4.51304376e-01
4.79159981e-01 -8.42710435e-01 1.32456303e+00 5.88295043e-01
-1.65287781e+00 -5.13705194e-01 -1.22353244e+00 -4.82889377e-02
-7.61134684e-01 -1.34695679e-01 7.89453983e-01 5.35296917e-01
-1.24497521e+00 7.11618602e-01 -4.45341110e-01 -3.85415912e-01
-4.76437658e-02 5.14827490e-01 -4.15880173e-01 5.82630515e-01
-1.42906129e+00 1.32564414e+00 9.23096418e-01 -3.04705389e-02
-1.38843432e-01 -7.99140394e-01 -7.56928444e-01 -1.13997750e-01
-2.09825099e-01 -6.55546725e-01 1.07758009e+00 -1.16444039e+00
-1.12039173e+00 1.08537817e+00 3.63112725e-02 -4.89705443e-01
1.24922045e-01 6.95305467e-02 -6.47922575e-01 -1.69061512e-01
-3.22947800e-01 9.37262237e-01 5.35941660e-01 -9.00347829e-01
-4.93474901e-01 -2.63436288e-01 -9.55281258e-02 1.03423275e-01
-1.26800850e-01 3.61568719e-01 2.74979502e-01 -1.10545051e+00
1.54559404e-01 -1.02266848e+00 2.89438982e-02 -6.44875228e-01
-5.23560047e-01 -6.80210948e-01 9.71880630e-02 -6.38109267e-01
1.07373846e+00 -1.96157253e+00 9.94443074e-02 2.43659198e-01
-8.58377144e-02 1.01729266e-01 -2.51983374e-01 3.40907514e-01
-4.80488092e-01 4.68766749e-01 -4.41827208e-01 -3.89613323e-02
2.66681999e-01 -2.22795252e-02 -4.54679787e-01 1.43116534e-01
2.87036121e-01 1.14294243e+00 -7.34321058e-01 -1.59601495e-01
-1.89725801e-01 -7.44932657e-03 -5.62369108e-01 -4.20857579e-01
-3.26633751e-01 -2.01065451e-01 3.66843551e-01 5.64563215e-01
6.59776986e-01 4.38713461e-01 1.55784547e-01 -1.41775236e-02
-1.85353234e-01 8.09038043e-01 -8.11261654e-01 1.52996922e+00
-4.34288949e-01 5.04934728e-01 -3.66819113e-01 -3.91423941e-01
9.79750991e-01 1.61016107e-01 1.95501037e-02 -3.14707994e-01
1.47073820e-01 6.69676900e-01 7.03877926e-01 -1.08888568e-02
9.06689703e-01 -4.11480963e-01 -4.59019601e-01 5.01281321e-01
2.72112250e-01 -3.58254969e-01 4.52180594e-01 -1.09155275e-01
1.11515474e+00 8.69210064e-02 6.94723666e-01 -7.87058890e-01
1.06036663e-01 2.56532848e-01 7.13988185e-01 5.51443219e-01
-5.40781878e-02 7.06519663e-01 2.16068447e-01 -5.60400486e-01
-9.35688376e-01 -1.50488091e+00 -3.56016994e-01 1.09120941e+00
-2.35165134e-01 -3.49039793e-01 -8.92426252e-01 -6.59685791e-01
-1.32327214e-01 1.37355125e+00 -5.83321273e-01 -4.02913749e-01
-1.05142069e+00 -8.55697870e-01 1.13372755e+00 5.61370790e-01
1.91129763e-02 -2.03731108e+00 -3.68964225e-01 1.04703553e-01
-1.56829894e-01 -5.52142322e-01 -7.56778717e-01 7.15423405e-01
-7.38937616e-01 -7.13651001e-01 -2.43648916e-01 -1.23068047e+00
4.65887934e-01 -3.45895201e-01 1.35290861e+00 -6.82854280e-02
4.01550345e-02 -4.55708444e-01 -1.07975975e-02 -4.40430760e-01
-6.20728374e-01 5.33985436e-01 2.81077385e-01 -6.10610843e-01
8.48555148e-01 -8.38365734e-01 -1.13191083e-01 -1.48383483e-01
-7.90075660e-01 -3.48472595e-01 6.47752702e-01 4.99087840e-01
5.64039946e-01 -2.00003296e-01 9.18929696e-01 -9.97361958e-01
9.13816631e-01 -4.87544805e-01 -1.76452637e-01 8.31354409e-02
-4.22574461e-01 -2.12630499e-02 8.99255991e-01 -7.18100190e-01
-8.15389037e-01 -9.22218785e-02 -2.81464666e-01 5.61779365e-02
-5.30492544e-01 6.08630836e-01 -3.67795944e-01 4.87032592e-01
1.02141738e+00 5.54929674e-02 -4.23086166e-01 -5.02000868e-01
4.81294185e-01 3.84969711e-01 9.43226039e-01 -7.93233633e-01
5.93196154e-01 -9.70134363e-02 -3.11958730e-01 -6.62174046e-01
-3.90048325e-01 -5.85890561e-02 -9.82511342e-01 3.28682423e-01
6.84233367e-01 -1.68013424e-01 -1.07266389e-01 4.68842477e-01
-1.61924005e+00 -2.78721631e-01 -6.43182993e-01 4.32805896e-01
-7.39177883e-01 2.04787761e-01 -8.31348538e-01 -2.23723814e-01
-5.41921079e-01 -9.14117157e-01 4.32225943e-01 -1.19942963e-01
-1.16872466e+00 -9.82404530e-01 3.47402364e-01 -3.45941722e-01
4.04745251e-01 1.74551666e-01 1.79510605e+00 -1.06203973e+00
-5.33828214e-02 -7.77088627e-02 1.40438288e-01 3.29573870e-01
3.64120573e-01 3.56137216e-01 -5.18531263e-01 1.12072136e-02
-2.29542494e-01 -3.76198322e-01 1.11414325e+00 3.28319639e-01
7.75379121e-01 -5.45573115e-01 3.30285355e-02 6.56282842e-01
1.05500102e+00 2.78430909e-01 5.35611868e-01 4.60512817e-01
5.42046249e-01 5.90860724e-01 7.30750412e-02 -3.85109931e-01
2.86593556e-01 5.03145278e-01 1.31290093e-01 2.16036424e-01
-5.91012776e-01 -2.22003594e-01 6.64943695e-01 1.09829783e+00
1.04887307e-01 -2.88745731e-01 -1.11262083e+00 1.14580214e+00
-1.38980913e+00 -9.16896701e-01 -4.40242916e-01 2.01217294e+00
1.38164008e+00 5.82218111e-01 2.51491278e-01 2.39866108e-01
9.10932124e-01 7.79870003e-02 -3.27363163e-01 -1.21158814e+00
-5.87067962e-01 6.12780392e-01 3.01166564e-01 4.87337828e-01
-5.83059251e-01 1.47933745e+00 7.64410353e+00 9.85500515e-01
-7.77034044e-01 -2.24249989e-01 7.97245428e-02 -2.51234453e-02
-8.23571861e-01 -1.86665788e-01 -9.59308326e-01 3.92844945e-01
1.26378560e+00 -2.62299359e-01 4.94445562e-01 3.97935003e-01
-2.80395627e-01 2.58670837e-01 -1.16744781e+00 4.66148078e-01
1.50145695e-01 -1.13673317e+00 4.98350531e-01 -4.70837243e-02
6.44870758e-01 3.27025622e-01 1.28491223e-01 3.74632329e-01
6.78108871e-01 -1.17118633e+00 1.11788332e+00 3.97293627e-01
7.75704861e-01 -1.09391117e+00 3.62846762e-01 2.94421047e-01
-6.92061424e-01 1.97061062e-01 -5.30639112e-01 -9.55834687e-02
2.43344650e-01 5.28595269e-01 -7.11827993e-01 2.22792244e-03
2.15265110e-01 3.96496952e-01 -7.90372133e-01 1.14282572e+00
-5.67367494e-01 9.81702507e-01 -6.80845439e-01 -1.24193672e-02
2.12545007e-01 -1.13937780e-01 7.97494113e-01 1.72373021e+00
2.23076344e-01 -1.11109905e-01 -2.46319771e-01 9.37041402e-01
-2.48931006e-01 4.08335835e-01 -7.45662689e-01 1.40420943e-01
7.00276613e-01 1.06936586e+00 -8.23296130e-01 -1.80780485e-01
2.08395887e-02 8.28093350e-01 7.45234966e-01 1.09683871e-01
-6.25909567e-01 -6.71395779e-01 5.75836778e-01 -5.02491444e-02
1.50206819e-01 -3.10461968e-01 -9.86077607e-01 -9.14111793e-01
-4.48308289e-02 -8.88982654e-01 5.87266445e-01 -5.35406232e-01
-1.51658452e+00 9.92418587e-01 1.67307004e-01 -7.11460769e-01
-4.73192096e-01 -6.51194572e-01 -8.14652860e-01 8.43026876e-01
-1.11274588e+00 -1.34783328e+00 4.08817738e-01 2.68668205e-01
5.38534939e-01 -2.50486583e-01 1.00772882e+00 2.68441923e-02
-3.95016909e-01 8.49964976e-01 -5.02348602e-01 3.78802605e-02
8.70010734e-01 -1.75639844e+00 1.15019071e+00 8.92100632e-01
6.02814019e-01 1.07519364e+00 7.19330430e-01 -7.35862732e-01
-5.76495469e-01 -1.09463370e+00 1.56643426e+00 -6.31867111e-01
7.92801023e-01 -1.80796027e-01 -1.07448602e+00 1.07548821e+00
6.83865190e-01 -3.30029786e-01 7.39054799e-01 3.68070304e-01
-6.08394861e-01 3.44959140e-01 -1.00267994e+00 9.07983422e-01
1.41219556e+00 -1.97567567e-01 -1.62613094e+00 2.56088749e-02
7.31809378e-01 -1.34043515e-01 -6.52958393e-01 3.76197010e-01
3.95313889e-01 -6.02113903e-01 6.82394385e-01 -8.19639742e-01
5.16190767e-01 -2.77998060e-01 -1.43769562e-01 -1.85678697e+00
-7.14717925e-01 -7.37905502e-01 -1.32687151e-01 1.31790400e+00
1.00527108e+00 -4.81435806e-01 4.23026055e-01 2.47213870e-01
-6.25220001e-01 -4.85796541e-01 -9.91969466e-01 -1.06733787e+00
9.82939184e-01 -6.51969373e-01 6.91113710e-01 1.03688812e+00
4.35025573e-01 4.34184611e-01 3.08266789e-01 -3.78452927e-01
2.81148106e-01 1.50156051e-01 6.71161637e-02 -1.01483274e+00
-3.22715074e-01 -1.07229602e+00 -3.17953616e-01 -5.62879920e-01
6.82812274e-01 -1.67870486e+00 -5.75747676e-02 -1.22919583e+00
-9.61417109e-02 -2.45500922e-01 -4.10333753e-01 9.64537442e-01
-1.64929956e-01 7.18848169e-01 2.22243682e-01 4.58297320e-02
-1.75434758e-03 6.79668635e-02 6.78443491e-01 -9.61354971e-02
-2.76040792e-01 -2.12965086e-01 -8.57689738e-01 1.08707690e+00
1.23107815e+00 -5.65024078e-01 -1.11244261e-01 -7.32771993e-01
5.52591324e-01 -4.40793335e-01 -4.73144874e-02 -6.43583953e-01
5.92092574e-02 -3.77941877e-01 2.28082627e-01 -7.47801900e-01
1.30007118e-01 -4.23067838e-01 7.36192614e-02 2.85109013e-01
-5.04422367e-01 5.17094016e-01 3.96654427e-01 -1.49304405e-01
-6.92616850e-02 -3.93580705e-01 9.02968287e-01 -2.01060668e-01
-5.12480855e-01 -1.50436431e-01 -6.70495212e-01 3.21146220e-01
6.84984148e-01 -4.47205633e-01 -5.03727973e-01 3.14883590e-01
-7.13972926e-01 -2.49583527e-01 6.50530636e-01 6.22258306e-01
3.82023245e-01 -1.40096343e+00 -9.95226324e-01 2.88977802e-01
4.14632075e-03 -3.93300056e-01 -4.09022629e-01 4.76414502e-01
-2.72243470e-01 2.46499047e-01 -6.20846212e-01 4.99129482e-02
-9.13859963e-01 5.55459321e-01 4.02511984e-01 -1.39300346e-01
-4.95315135e-01 9.23199058e-01 -8.49126428e-02 -7.78428376e-01
-1.57694489e-01 -6.91058099e-01 -2.48803809e-01 2.90991068e-01
3.77992898e-01 3.59383613e-01 5.04189670e-01 -8.02530289e-01
-3.05141240e-01 5.68569720e-01 -2.96699822e-01 -2.47477636e-01
1.22615314e+00 1.79099053e-01 -3.54137748e-01 8.83506536e-01
9.42752600e-01 4.13989753e-01 -5.81877589e-01 9.85760335e-03
5.46258807e-01 1.07773855e-01 -2.56513238e-01 -1.29885352e+00
-9.20983493e-01 5.52492023e-01 -2.55078882e-01 1.46961689e-01
8.46271098e-01 -4.37959209e-02 1.14275217e+00 4.31897283e-01
4.27264571e-01 -1.11046827e+00 -4.59508717e-01 1.22609913e+00
1.09362423e+00 -5.06642342e-01 -3.90621543e-01 -3.24101001e-01
-4.95461315e-01 1.29717100e+00 7.72250235e-01 -6.43169403e-01
4.91846085e-01 4.75552529e-01 7.93101043e-02 -2.04530522e-01
-7.71281004e-01 -7.01272637e-02 3.63563210e-01 7.80206442e-01
6.80430710e-01 1.46274254e-01 -9.33111846e-01 9.61048424e-01
-1.32482100e+00 -4.94404733e-01 6.65097415e-01 5.09669304e-01
-3.97343457e-01 -9.97427464e-01 -1.41143501e-01 5.71665347e-01
-6.80338442e-01 -5.34065068e-01 -1.01862991e+00 8.74867320e-01
4.77762282e-01 4.34496313e-01 4.23295707e-01 -2.50837415e-01
5.38708031e-01 3.77356112e-01 7.54856229e-01 -9.53766763e-01
-1.26545215e+00 -2.91085988e-01 2.64647812e-01 -1.34265497e-01
1.51970342e-01 -1.08659863e+00 -1.65834332e+00 -2.61563838e-01
-9.48712304e-02 -5.28351367e-02 3.96789968e-01 7.94217169e-01
-6.65257499e-02 2.64830083e-01 9.25465524e-02 -7.02110827e-01
-5.67699552e-01 -1.35147798e+00 -6.76900804e-01 4.33627099e-01
7.76701570e-02 -2.30342880e-01 -5.02995312e-01 4.92009148e-02] | [10.718247413635254, 9.728987693786621] |
b46b5859-6833-4a5b-9779-0752a9598f60 | magnificent-minified-models | 2306.10177 | null | https://arxiv.org/abs/2306.10177v1 | https://arxiv.org/pdf/2306.10177v1.pdf | Magnificent Minified Models | This paper concerns itself with the task of taking a large trained neural network and 'compressing' it to be smaller by deleting parameters or entire neurons, with minimal decreases in the resulting model accuracy. We compare various methods of parameter and neuron selection: dropout-based neuron damage estimation, neuron merging, absolute-value based selection, random selection, OBD (Optimal Brain Damage). We also compare a variation on the classic OBD method that slightly outperformed all other parameter and neuron selection methods in our tests with substantial pruning, which we call OBD-SD. We compare these methods against quantization of parameters. We also compare these techniques (all applied to a trained neural network), with neural networks trained from scratch (random weight initialization) on various pruned architectures. Our results are only barely consistent with the Lottery Ticket Hypothesis, in that fine-tuning a parameter-pruned model does slightly better than retraining a similarly pruned model from scratch with randomly initialized weights. For neuron-level pruning, retraining from scratch did much better in our experiments. | ['Hillary Sanders', 'Rich Harang'] | 2023-06-16 | null | null | null | null | ['quantization'] | ['methodology'] | [ 4.68752414e-01 3.16722125e-01 1.05695911e-01 -2.70731270e-01
-4.88103658e-01 -2.95768470e-01 3.13381702e-01 2.95663506e-01
-1.20323801e+00 9.87515628e-01 1.58822879e-01 -1.56736165e-01
-3.21039140e-01 -5.67950010e-01 -7.01971114e-01 -9.82138753e-01
-8.83970112e-02 4.92283553e-01 7.11470425e-01 -7.02984110e-02
7.62968898e-01 7.68994868e-01 -1.56571603e+00 2.16525778e-01
6.96454287e-01 8.36579204e-01 2.68280149e-01 5.81951797e-01
2.15608761e-01 6.74197853e-01 -8.54642153e-01 -4.52033073e-01
2.77793378e-01 -2.28780359e-01 -9.30126607e-01 -2.41053179e-01
8.13352168e-01 -1.36826321e-01 -2.40150928e-01 9.65244293e-01
6.62146449e-01 2.73507148e-01 5.43518305e-01 -8.74021769e-01
-4.10002649e-01 8.06906939e-01 -3.07074219e-01 8.84872377e-01
-4.24848616e-01 2.56042421e-01 5.75076759e-01 -5.70233643e-01
5.53828895e-01 1.28968883e+00 9.56325293e-01 7.60209322e-01
-1.74480867e+00 -5.67074656e-01 2.62275398e-01 -3.61697003e-02
-1.56732285e+00 -6.83651805e-01 2.13623792e-01 -2.00025171e-01
1.62276936e+00 2.40526333e-01 6.32760346e-01 9.20109808e-01
2.74851203e-01 5.23885846e-01 1.04423296e+00 -4.25895631e-01
4.28083271e-01 1.43337741e-01 5.16629457e-01 7.28308141e-01
6.94906354e-01 4.29014601e-02 -2.20707163e-01 -3.05760205e-01
9.42904711e-01 -1.49959251e-01 -3.53202313e-01 -2.29550116e-02
-1.21630037e+00 7.01937973e-01 3.53608966e-01 3.29409599e-01
-3.57504398e-01 4.32111561e-01 4.02987868e-01 6.04429662e-01
1.55138761e-01 9.72737074e-01 -7.98894942e-01 2.06477046e-01
-1.33215928e+00 3.95200342e-01 7.11335301e-01 5.83433211e-01
7.82867372e-01 4.63148743e-01 -4.13448155e-01 1.17312908e+00
-1.88023284e-01 -2.20974803e-01 8.94941866e-01 -1.48811197e+00
4.24921244e-01 2.49927714e-01 -1.70661926e-01 -3.91506910e-01
-5.49341857e-01 -4.80904847e-01 -7.54610419e-01 7.32337415e-01
4.88246560e-01 -3.80905867e-01 -1.35633111e+00 1.92220795e+00
-3.79442930e-01 7.13183880e-02 -7.71005750e-02 3.60928774e-01
4.71038491e-01 3.29534680e-01 2.19495028e-01 -1.53555900e-01
8.53823125e-01 -1.01319170e+00 -3.93622071e-01 -3.31842273e-01
8.38359296e-01 -3.00328493e-01 1.10776854e+00 9.29252207e-01
-1.43772352e+00 -4.07737315e-01 -1.19743669e+00 -3.92388776e-02
-5.19419193e-01 -3.00828833e-02 4.34130549e-01 7.08528280e-01
-1.63152802e+00 1.46799016e+00 -8.46889496e-01 -3.00535202e-01
1.00584161e+00 9.63773847e-01 -3.25269490e-01 1.01072840e-01
-9.76105154e-01 1.45294785e+00 6.58828437e-01 -2.76447624e-01
-8.63474250e-01 -5.99886954e-01 -4.83399153e-01 2.08599776e-01
1.90389648e-01 -8.40070665e-01 1.19707441e+00 -7.38225460e-01
-1.21621990e+00 1.06057608e+00 -1.18874330e-02 -1.08323944e+00
-6.21826947e-02 -1.18006483e-01 -1.15053780e-01 -5.64300381e-02
-4.81704295e-01 1.44167924e+00 8.66547525e-01 -1.00407946e+00
-6.28983319e-01 -4.16329563e-01 -1.93725750e-01 2.89989561e-01
-2.41329968e-01 1.94787338e-01 -1.89360619e-01 -8.56473863e-01
-7.03909993e-02 -5.96159399e-01 -5.41169822e-01 -1.51778117e-01
-5.50584018e-01 -2.80199289e-01 2.90005654e-01 -4.87258226e-01
1.30225921e+00 -1.95037031e+00 2.20175102e-01 8.13959092e-02
3.11022580e-01 3.52767944e-01 -3.62293571e-01 -2.14295477e-01
-2.34529465e-01 5.76018274e-01 -4.27217335e-01 -4.87895221e-01
-2.17569858e-01 4.48894620e-01 -4.07829843e-02 3.06778312e-01
4.17270392e-01 6.13169253e-01 -4.47283506e-01 -5.24513006e-01
-3.30598325e-01 2.77502745e-01 -7.72919118e-01 -4.47903812e-01
1.56529009e-01 -3.27345520e-01 3.91225889e-02 5.54901421e-01
2.96980411e-01 2.70994157e-01 -3.89500082e-01 1.51163861e-02
-9.43645760e-02 4.14652884e-01 -1.03142679e+00 1.35125411e+00
-6.80693164e-02 7.70562470e-01 -9.02411342e-02 -1.03565073e+00
8.72657120e-01 1.67900681e-01 -1.66295059e-02 -4.58431572e-01
3.18005800e-01 2.23453045e-01 5.05545378e-01 -1.57178909e-01
2.51561522e-01 -7.84506127e-02 1.79410845e-01 3.34362030e-01
5.67371368e-01 -2.18034431e-01 3.18210661e-01 1.22997761e-01
1.57682693e+00 -1.87857941e-01 -4.67876121e-02 -4.81325537e-01
-8.21582451e-02 1.24145504e-02 5.61735153e-01 9.96712744e-01
-2.56753713e-01 8.85433078e-01 8.01041305e-01 -2.58129388e-01
-1.10649180e+00 -9.38211203e-01 -3.02757114e-01 1.06450248e+00
-3.37829441e-01 -1.25450313e-01 -1.31371748e+00 -4.55224574e-01
-4.51502055e-02 1.11767936e+00 -8.83904934e-01 -5.25910795e-01
-7.72583485e-01 -1.23342597e+00 9.31338251e-01 4.60414261e-01
3.02922815e-01 -1.59438765e+00 -7.53391147e-01 3.16321760e-01
4.44984496e-01 -4.16598856e-01 -1.29037842e-01 1.25405526e+00
-1.66467619e+00 -6.80416882e-01 -8.28386962e-01 -9.92034972e-01
8.41205716e-01 -2.10072741e-01 1.19296086e+00 4.25575107e-01
-2.39709124e-01 -3.36069912e-01 9.84841436e-02 -5.02838433e-01
-1.37229621e-01 3.50230753e-01 1.52930722e-01 -6.27287567e-01
5.21945238e-01 -7.34109461e-01 -3.17552119e-01 1.19563788e-01
-9.31066811e-01 -3.38524550e-01 8.40309083e-01 7.66719520e-01
7.48018205e-01 1.62316978e-01 5.30789137e-01 -1.20983684e+00
9.89417017e-01 -3.51760924e-01 -1.71413600e-01 2.97501050e-02
-9.15305436e-01 4.86430496e-01 4.92805660e-01 -8.35849285e-01
-5.94314635e-01 -1.82898939e-01 -2.59726912e-01 -5.60250998e-01
-3.91704768e-01 -8.24068859e-02 5.50584495e-02 -2.78269559e-01
9.87852097e-01 -1.05261654e-01 -7.74084851e-02 -7.79924035e-01
4.88244072e-02 7.17016906e-02 6.92230463e-01 -2.99582243e-01
4.67080325e-01 5.97243086e-02 -3.46699834e-01 -2.06074014e-01
-5.67534029e-01 5.28955199e-02 -8.88440847e-01 4.39574987e-01
6.35643542e-01 -4.79378313e-01 -5.66806383e-02 6.83585405e-01
-1.02483904e+00 -9.26413953e-01 -8.20922732e-01 4.67835575e-01
-4.70474452e-01 -1.05080634e-01 -7.93576539e-01 -2.21611619e-01
-3.10535967e-01 -1.11197937e+00 4.91948694e-01 2.35800818e-01
-4.32045370e-01 -8.13873827e-01 1.48515329e-01 -2.70476550e-01
4.17802840e-01 4.82440256e-02 1.29502332e+00 -1.06525373e+00
-3.81051153e-02 -1.75784305e-01 -9.91043746e-02 6.80967748e-01
-1.99499264e-01 -7.10859448e-02 -8.56160581e-01 -1.78772047e-01
1.87144488e-01 -5.50420344e-01 1.57934821e+00 6.39310777e-01
1.48870635e+00 -3.15685958e-01 -4.82461900e-01 6.65175855e-01
1.51714694e+00 1.08390577e-01 9.47966754e-01 6.75742984e-01
4.19289321e-01 5.01065791e-01 6.94807842e-02 2.52906121e-02
-2.24386573e-01 2.69883633e-01 3.78366202e-01 -6.79859594e-02
-3.81374627e-01 1.90041289e-01 2.72285849e-01 5.08283734e-01
-2.68172622e-01 -3.00123423e-01 -6.72854722e-01 6.44084990e-01
-1.37112427e+00 -1.04159701e+00 1.08608425e-01 2.39668846e+00
1.21084905e+00 7.98964739e-01 9.18592364e-02 4.77992088e-01
7.68247068e-01 -1.80264607e-01 -6.43530309e-01 -1.02347314e+00
-3.04782063e-01 6.19176626e-01 8.70300412e-01 5.41788757e-01
-7.21457958e-01 9.94668901e-01 7.83398342e+00 9.68963861e-01
-7.25534081e-01 2.44421601e-01 9.67423201e-01 -7.50885248e-01
6.43294156e-02 -1.72641635e-01 -1.33805370e+00 3.71578991e-01
1.43673313e+00 3.05108666e-01 7.73009539e-01 5.16663849e-01
-2.38343677e-03 -3.28462094e-01 -1.20834029e+00 6.92001760e-01
1.11727417e-01 -1.32113039e+00 7.04953000e-02 9.19619128e-02
7.65756726e-01 4.04378176e-01 -3.76324356e-02 5.37981749e-01
4.75676179e-01 -1.23683906e+00 7.05622494e-01 6.04054511e-01
5.00081778e-01 -8.77871335e-01 5.54667175e-01 3.69723946e-01
-3.48201752e-01 -2.58514434e-01 -9.07241404e-01 6.42568991e-02
-2.09918126e-01 5.72920442e-01 -4.03407693e-01 -5.48292696e-01
9.36623037e-01 1.24555439e-01 -1.18666518e+00 1.61553979e+00
-6.19926769e-03 8.42006803e-01 -7.82472551e-01 7.80558065e-02
2.41471857e-01 5.04313670e-02 4.10643995e-01 1.39878714e+00
1.63653731e-01 -4.38254774e-02 -4.93626475e-01 7.83733070e-01
-2.15284064e-01 -2.45072961e-01 -4.25741792e-01 3.84919912e-01
7.51725495e-01 9.85348523e-01 -9.81187046e-01 -5.94115794e-01
-1.04268394e-01 9.06450927e-01 5.96368670e-01 3.73323172e-01
-5.81043720e-01 -9.46201563e-01 3.11073542e-01 8.40092003e-02
6.59700453e-01 1.14060231e-01 -6.01434767e-01 -6.23069882e-01
-3.72061551e-01 -7.45994270e-01 2.46096909e-01 -7.74735212e-01
-8.82630825e-01 8.63597453e-01 1.66390583e-01 -5.71885943e-01
-8.71042609e-02 -6.62565708e-01 -8.87640774e-01 8.69353116e-01
-1.17709303e+00 -9.28248987e-02 1.83526441e-01 5.31275272e-01
4.82346207e-01 -2.65496373e-01 8.20294142e-01 1.87980354e-01
-9.52239871e-01 8.28339338e-01 1.37377858e-01 -1.32218629e-01
5.40738940e-01 -1.31625211e+00 3.88186008e-01 6.38134360e-01
7.97869042e-02 7.72655606e-01 5.92797399e-01 -4.75108117e-01
-5.70181966e-01 -1.13387692e+00 8.41246903e-01 -3.33344012e-01
3.08445543e-01 -3.00166875e-01 -1.40209532e+00 7.88052440e-01
3.56847972e-01 -4.05537896e-02 4.30633754e-01 -2.61770450e-02
3.00571341e-02 -4.10618633e-02 -1.43913567e+00 6.82950675e-01
1.08874559e+00 5.80618046e-02 -9.94045556e-01 2.33795643e-01
8.31052423e-01 3.57222743e-02 -7.80048370e-01 1.51757509e-01
9.53825489e-02 -8.76389384e-01 1.05015790e+00 -8.24386120e-01
5.35411350e-02 8.41847807e-02 2.32389681e-02 -1.43205237e+00
-7.50006616e-01 -4.29284662e-01 -1.45183668e-01 1.19967365e+00
7.10289299e-01 -6.24555647e-01 8.95564556e-01 6.76132083e-01
-4.15276140e-01 -1.18336165e+00 -8.88416708e-01 -8.44031751e-01
5.38958490e-01 -7.54379183e-02 3.82426709e-01 6.29916430e-01
-3.37655604e-01 3.05574417e-01 2.03024834e-01 -1.08557396e-01
5.60274005e-01 -6.89008951e-01 -8.80607869e-03 -1.44373786e+00
-2.64467746e-01 -1.06886864e+00 -4.30555344e-01 -6.70946836e-01
-5.12608550e-02 -7.64552236e-01 9.93785486e-02 -1.39398801e+00
8.87494460e-02 -5.08391201e-01 -7.35660970e-01 9.26176488e-01
-7.26448819e-02 5.05772531e-01 -2.87046045e-04 3.33269119e-01
-3.26155692e-01 8.76531601e-02 8.82337213e-01 -7.82870725e-02
-3.55501235e-01 -2.08026934e-02 -8.82224321e-01 9.11080837e-01
1.12050605e+00 -1.04403400e+00 -3.53903383e-01 -7.06032217e-01
-5.05346507e-02 -4.03378159e-01 3.97236437e-01 -1.47823536e+00
1.46857426e-01 1.44345388e-01 6.20598674e-01 -2.82988399e-01
2.66739041e-01 -2.03345880e-01 -1.89862221e-01 7.58719385e-01
-7.30231702e-01 1.61987126e-01 5.28603137e-01 9.03479457e-02
1.86454266e-01 -1.24632931e+00 1.14665687e+00 -3.25710446e-01
-6.65575027e-01 3.67406122e-02 -7.99557149e-01 2.09250480e-01
6.75852239e-01 -7.88741529e-01 -3.81352574e-01 2.97396690e-01
-1.18158233e+00 5.12295663e-02 5.25653720e-01 -6.39066026e-02
6.03457689e-01 -1.00578952e+00 -6.61091566e-01 1.19437501e-01
-4.56717074e-01 3.21428329e-02 -3.08785826e-01 8.87128532e-01
-4.50512260e-01 3.38273644e-01 -4.15149122e-01 -1.15832694e-01
-1.20994234e+00 5.33280313e-01 5.69217205e-01 -2.46858910e-01
-5.45683265e-01 1.43120599e+00 -1.52726114e-01 -2.99591422e-01
7.87864983e-01 -7.84639597e-01 -4.03134793e-01 1.98147781e-02
5.31915367e-01 3.93746644e-01 4.69658285e-01 1.06455320e-02
-1.31490707e-01 2.43521422e-01 -4.73033279e-01 -2.12965190e-01
1.42973936e+00 2.01206014e-01 1.04681261e-01 4.85021979e-01
1.11116934e+00 -3.37145060e-01 -1.35203695e+00 -1.37801766e-01
1.98536422e-02 5.77448681e-02 6.52334571e-01 -9.85873938e-01
-1.17904651e+00 9.93511856e-01 7.17976332e-01 2.86833532e-02
1.11901021e+00 -2.03174844e-01 6.43329501e-01 1.02723265e+00
2.85062790e-01 -1.27157032e+00 -6.06849045e-02 4.68869537e-01
7.93568134e-01 -6.85638964e-01 2.32941136e-01 2.94997692e-01
-6.96232677e-01 9.43672657e-01 8.23785126e-01 -6.67588532e-01
6.04477763e-01 3.28296542e-01 -2.65367866e-01 -1.87256888e-01
-1.03909457e+00 -5.72221987e-02 -1.18111841e-01 7.71787882e-01
8.70235860e-02 -4.50691044e-01 -6.13645241e-02 5.75922251e-01
-3.47964227e-01 5.14830835e-02 5.34612894e-01 9.71805573e-01
-1.05730844e+00 -8.70936215e-01 -3.50533068e-01 1.19401979e+00
-6.04829729e-01 -6.15227759e-01 -3.95314664e-01 9.42941606e-01
4.36656594e-01 3.80054742e-01 4.06460792e-01 -3.00045282e-01
2.63863802e-01 4.20027465e-01 8.42933714e-01 -9.42144930e-01
-1.00079238e+00 -8.30220580e-02 4.47946750e-02 -4.37511086e-01
2.66684988e-03 -8.14277649e-01 -1.37479830e+00 -2.78899342e-01
-6.29707456e-01 -1.25163019e-01 4.26468849e-01 9.00731742e-01
1.73980117e-01 5.83274961e-01 1.06563583e-01 -1.15319228e+00
-7.06194162e-01 -1.05265725e+00 -7.87553012e-01 1.29234821e-01
1.47627264e-01 -5.84080040e-01 -5.80898404e-01 8.21380243e-02] | [8.570712089538574, 3.2968051433563232] |
5f3539e5-b663-4f4b-bb3a-d11aa3c010cd | hybrid-bayesian-neural-networks-with | 2107.07014 | null | https://arxiv.org/abs/2107.07014v1 | https://arxiv.org/pdf/2107.07014v1.pdf | Hybrid Bayesian Neural Networks with Functional Probabilistic Layers | Bayesian neural networks provide a direct and natural way to extend standard deep neural networks to support probabilistic deep learning through the use of probabilistic layers that, traditionally, encode weight (and bias) uncertainty. In particular, hybrid Bayesian neural networks utilize standard deterministic layers together with few probabilistic layers judicially positioned in the networks for uncertainty estimation. A major aspect and benefit of Bayesian inference is that priors, in principle, provide the means to encode prior knowledge for use in inference and prediction. However, it is difficult to specify priors on weights since the weights have no intuitive interpretation. Further, the relationships of priors on weights to the functions computed by networks are difficult to characterize. In contrast, functions are intuitive to interpret and are direct since they map inputs to outputs. Therefore, it is natural to specify priors on functions to encode prior knowledge, and to use them in inference and prediction based on functions. To support this, we propose hybrid Bayesian neural networks with functional probabilistic layers that encode function (and activation) uncertainty. We discuss their foundations in functional Bayesian inference, functional variational inference, sparse Gaussian processes, and sparse variational Gaussian processes. We further perform few proof-of-concept experiments using GPflus, a new library that provides Gaussian process layers and supports their use with deterministic Keras layers to form hybrid neural network and Gaussian process models. | ['Daniel T. Chang'] | 2021-07-14 | null | null | null | null | ['probabilistic-deep-learning'] | ['computer-vision'] | [-3.16581726e-01 3.06865811e-01 9.01745632e-02 -8.42098534e-01
-3.55674595e-01 -6.09263897e-01 9.89903152e-01 -2.39768431e-01
-2.82427907e-01 8.06943715e-01 1.05433889e-01 -4.29851949e-01
-3.74081582e-01 -1.30764115e+00 -9.64924276e-01 -7.84235775e-01
2.41734814e-02 7.99207091e-01 1.95677668e-01 1.43979058e-01
4.23953496e-02 4.28061724e-01 -1.43686819e+00 1.96294740e-01
5.32521904e-01 1.17699707e+00 -2.56504640e-02 3.81111234e-01
-3.90679300e-01 6.46720827e-01 -6.38856113e-01 -6.17998481e-01
-1.88404277e-01 2.64444202e-01 -3.54209930e-01 -8.30223083e-01
2.77421176e-02 -6.89130127e-01 -3.42901289e-01 1.04389668e+00
2.81436473e-01 4.12187815e-01 1.39017355e+00 -1.27884769e+00
-9.87117529e-01 1.30077922e+00 -1.92797825e-01 7.87875578e-02
5.76034933e-03 1.25773549e-01 1.10021353e+00 -9.11747754e-01
-1.03690520e-01 1.88734007e+00 7.99675882e-01 5.21169186e-01
-1.50437629e+00 -7.26478875e-01 3.87820154e-01 -1.92046598e-01
-1.43780768e+00 -4.02251542e-01 5.73945105e-01 -5.80149293e-01
9.57777679e-01 -1.17891416e-01 4.60642993e-01 1.69840205e+00
3.89813125e-01 8.55730832e-01 7.12090969e-01 -2.29022652e-02
7.61173606e-01 -2.26386469e-02 3.98063570e-01 4.71761286e-01
3.03875893e-01 5.25443971e-01 -5.57247698e-01 -4.07594889e-01
9.65571404e-01 1.68519765e-01 -1.24907292e-01 -3.97762619e-02
-8.30036283e-01 1.05979240e+00 2.65148759e-01 -2.27964997e-01
-4.16924715e-01 1.12917829e+00 6.14173599e-02 -1.92482948e-01
2.63880670e-01 1.96487948e-01 -6.22163832e-01 -1.74514964e-01
-9.56073225e-01 4.79720503e-01 1.06395817e+00 8.49926412e-01
6.73856378e-01 3.61289203e-01 -5.18555820e-01 6.97678804e-01
1.16588140e+00 8.67127895e-01 -9.36740264e-03 -1.28066325e+00
7.70474523e-02 -2.39902973e-01 1.94764286e-01 -8.39823544e-01
-1.77222744e-01 -1.87323429e-02 -9.60314810e-01 3.61706197e-01
4.11295563e-01 -4.11109120e-01 -1.18871033e+00 1.93899846e+00
-2.18720138e-01 3.10229897e-01 4.61664423e-02 5.53870082e-01
8.68275344e-01 8.12420011e-01 3.26716721e-01 3.84743214e-01
1.42803419e+00 -1.84574768e-01 -6.96612000e-01 -5.71893491e-02
-2.14239389e-01 -3.17781866e-01 8.95291865e-01 5.04453599e-01
-1.00511086e+00 -4.64882165e-01 -9.90636766e-01 7.28415847e-02
-5.01488507e-01 -2.46328935e-01 1.06480050e+00 1.04602969e+00
-1.07611001e+00 9.61245120e-01 -1.29398715e+00 3.15576136e-01
6.32042527e-01 2.92719662e-01 1.10051967e-01 1.33929729e-01
-1.73952687e+00 9.43031669e-01 7.49928236e-01 3.30861121e-01
-1.35354233e+00 -7.31467962e-01 -1.08477163e+00 6.03084803e-01
1.25207290e-01 -8.90298188e-01 1.35067916e+00 -3.39726388e-01
-1.87155509e+00 7.42858425e-02 -1.21534832e-01 -5.05717337e-01
2.91793883e-01 -2.67310709e-01 -1.39965966e-01 -1.44910321e-01
-2.80254364e-01 1.06599307e+00 9.43952203e-01 -1.14595103e+00
-3.85554701e-01 -1.45069137e-01 2.75338858e-01 -2.06473842e-01
1.37647921e-02 -1.15238212e-01 -3.39336932e-01 -2.65518159e-01
7.47826472e-02 -5.61698496e-01 -1.56985164e-01 1.50060162e-01
-5.86676061e-01 -5.65617383e-01 3.46913874e-01 -2.08419830e-01
7.69165754e-01 -2.00276804e+00 -2.22234070e-01 5.08740723e-01
2.63895154e-01 -4.42574978e-01 1.54926062e-01 -1.28117189e-01
1.44481644e-01 4.55490917e-01 -3.92540872e-01 -4.66551512e-01
7.32035220e-01 7.30501413e-01 -7.32513845e-01 1.55984610e-01
5.66538215e-01 9.86307144e-01 -8.57864678e-01 -2.58925200e-01
2.73795456e-01 7.47800529e-01 -5.96411049e-01 6.53819367e-02
-6.60364926e-01 -3.73497866e-02 -3.89595449e-01 5.20783842e-01
7.53573537e-01 -5.13772592e-02 2.13083886e-02 -1.79761797e-01
1.74231470e-01 5.48348963e-01 -1.60678566e+00 1.31994927e+00
-3.69638175e-01 4.77599412e-01 -1.51896730e-01 -7.04697311e-01
7.32087493e-01 4.17947978e-01 3.86396348e-02 2.38579720e-01
2.35330701e-01 -1.05602540e-01 -1.20794676e-01 1.84414908e-01
5.14637053e-01 -2.62772590e-01 -2.39596874e-01 5.68992317e-01
4.23812687e-01 -5.86869299e-01 1.21881448e-01 8.15571379e-03
7.24218071e-01 5.70628941e-01 -1.29668489e-01 -4.05547738e-01
-2.93423701e-02 -6.11163557e-01 6.44021988e-01 1.15427482e+00
6.22535124e-02 6.12830400e-01 8.91600311e-01 -2.29675323e-01
-6.35856152e-01 -2.02758050e+00 -4.78692681e-01 1.15100670e+00
-4.45097834e-01 -1.52322590e-01 -3.83700609e-01 -3.63515496e-01
2.00450495e-01 1.20748055e+00 -7.03333318e-01 -2.74688695e-02
6.30122498e-02 -1.01180816e+00 9.02987480e-01 1.04409385e+00
2.66412169e-01 -9.62076485e-01 -2.24028304e-01 3.77678812e-01
2.80320376e-01 -7.08487511e-01 -1.03751406e-01 6.97372854e-01
-7.69501567e-01 -7.03857303e-01 -4.85363662e-01 -6.11045398e-02
3.54830861e-01 -5.71717322e-01 1.20525527e+00 -6.46479607e-01
3.82159203e-01 3.55281681e-01 3.50858122e-01 -8.41915667e-01
-2.38273159e-01 -2.91815013e-01 2.66573399e-01 -4.91468906e-01
4.08319890e-01 -1.02250957e+00 -3.08156580e-01 1.03574984e-01
-9.55418229e-01 -3.66507918e-01 3.46037298e-01 8.92338336e-01
4.71197188e-01 2.97935784e-01 2.69325227e-01 -7.51387894e-01
9.10637975e-01 -6.36374652e-01 -9.54730570e-01 2.06971571e-01
-4.72066671e-01 6.05203867e-01 2.30749190e-01 -3.93053532e-01
-1.51746559e+00 -1.72999859e-01 -2.43048474e-01 -6.03377640e-01
-2.03910485e-01 7.87547588e-01 -3.49465013e-01 7.20225513e-01
6.82352185e-01 -8.69572908e-02 -2.49428779e-01 -2.74498761e-01
8.70579183e-01 3.53062063e-01 5.44261694e-01 -1.46963167e+00
3.31022322e-01 4.44638222e-01 -3.49559151e-02 -2.69955903e-01
-8.46287191e-01 2.35494033e-01 -3.69398355e-01 2.00931504e-01
1.07155252e+00 -9.21011627e-01 -1.15100515e+00 4.67583865e-01
-1.47642541e+00 -2.48708785e-01 -2.56712288e-01 5.79635262e-01
-5.04542291e-01 1.48535162e-01 -8.46119225e-01 -1.23958850e+00
-2.41036326e-01 -1.28197503e+00 8.92531037e-01 2.64827639e-01
-3.42783242e-01 -1.32827258e+00 -2.87665606e-01 -2.79942989e-01
4.98564988e-01 -1.06167533e-01 9.82737780e-01 -6.93327367e-01
-4.89817262e-01 9.67305377e-02 -3.18679631e-01 6.51659608e-01
-1.08639620e-01 6.05858266e-01 -1.39636397e+00 4.55649287e-01
4.13303301e-02 -1.07927024e-01 1.23221731e+00 9.78015125e-01
1.44144523e+00 -9.73339379e-02 -2.78460205e-01 5.46352327e-01
1.00403416e+00 3.24791446e-02 7.16848791e-01 -3.56174976e-01
5.91429412e-01 4.72079307e-01 -2.62662441e-01 5.49089849e-01
4.57388908e-01 -2.21903902e-02 3.29077661e-01 5.87155879e-01
4.14968193e-01 -4.41521674e-01 5.24427950e-01 3.57407600e-01
-2.41470277e-01 -2.45868847e-01 -9.75095332e-01 1.82231665e-01
-1.85691750e+00 -1.27170730e+00 -6.77289143e-02 1.96397078e+00
1.31424344e+00 5.95009089e-01 -5.63694417e-01 -2.84088314e-01
6.59198821e-01 -9.27850083e-02 -4.16486531e-01 -4.06226963e-01
2.83212326e-02 4.78618801e-01 3.87918353e-01 5.88696301e-01
-1.16310787e+00 7.54914582e-01 7.37820435e+00 1.04053175e+00
-6.06451929e-01 7.94296712e-02 8.25449705e-01 -1.04738399e-01
-7.50346839e-01 -5.27779423e-02 -1.15949082e+00 7.09503651e-01
1.35191095e+00 2.20553651e-01 3.78376395e-01 1.10760176e+00
8.77567530e-02 -2.17633113e-01 -1.66668081e+00 7.71533787e-01
-6.67482913e-01 -1.54748070e+00 1.82135016e-01 -1.53416827e-01
7.20770061e-01 1.43084615e-01 2.61081785e-01 7.13704705e-01
1.55368209e+00 -1.50991762e+00 9.73417461e-01 1.08333588e+00
5.71949363e-01 -8.61848950e-01 8.99927437e-01 1.55104116e-01
-7.99738824e-01 1.53864115e-01 -7.92191207e-01 -1.60193741e-01
2.47253135e-01 1.05637372e+00 -5.03642857e-01 1.02300376e-01
7.85182595e-01 3.82209003e-01 2.07631618e-01 5.83222091e-01
-8.71735871e-01 7.97554553e-01 -8.60479236e-01 -1.11537613e-02
2.48691052e-01 -1.30985528e-01 2.32278228e-01 1.22117591e+00
4.60984558e-01 -2.59389788e-01 1.19460821e-02 1.94919991e+00
-8.39686990e-02 -8.08874547e-01 -4.30155009e-01 -1.10855475e-01
8.88039529e-01 8.67898107e-01 -4.74825412e-01 -3.72459769e-01
-3.06654662e-01 4.10166740e-01 1.46637768e-01 7.55364954e-01
-1.03037524e+00 -1.08066298e-01 9.74251330e-01 -4.01980162e-01
3.91328573e-01 -2.63412774e-01 -5.61868191e-01 -8.64149988e-01
-4.73582476e-01 -3.16855848e-01 2.93086618e-01 -9.90180969e-01
-1.85344219e+00 1.18032917e-01 5.57428658e-01 -4.06081557e-01
-5.18047988e-01 -9.51349020e-01 -7.69393384e-01 1.55082726e+00
-1.34605527e+00 -9.89243209e-01 3.13385755e-01 4.84162539e-01
-1.07763633e-02 -8.78658891e-02 9.84958887e-01 -1.89878002e-01
-4.74181026e-01 3.88625920e-01 1.83181792e-01 2.23949239e-01
3.99242997e-01 -1.62301612e+00 4.80013430e-01 5.36537707e-01
2.67823599e-02 1.25866818e+00 7.23493636e-01 -6.02486670e-01
-1.16511893e+00 -8.51228654e-01 2.16863811e-01 -8.21211874e-01
9.08244133e-01 -4.20697629e-01 -8.57422352e-01 7.51599073e-01
3.83390393e-03 -1.24826049e-02 7.61788845e-01 6.43971026e-01
-6.48638844e-01 1.15326315e-01 -1.27476704e+00 4.71334338e-01
5.52671731e-01 -6.97213769e-01 -9.91600156e-01 -8.82176980e-02
9.70967293e-01 -2.94286370e-01 -9.79001939e-01 1.74335405e-01
7.03573227e-01 -6.79718971e-01 1.05472338e+00 -4.75860715e-01
5.50544262e-01 -4.27384704e-01 -5.22586823e-01 -1.48173916e+00
-2.81564891e-01 -4.30979759e-01 -6.39732063e-01 1.26700687e+00
6.97555244e-01 -8.18125427e-01 6.09643757e-01 1.26263261e+00
-2.76593566e-01 -5.42962492e-01 -9.45023417e-01 -6.76118910e-01
4.92959321e-01 -1.35287678e+00 9.55049336e-01 6.04358137e-01
-3.60055387e-01 8.46403614e-02 6.53579980e-02 3.77725184e-01
6.93833947e-01 -2.60660440e-01 1.19569339e-01 -1.65360868e+00
-5.78289151e-01 -8.49078178e-01 -1.53723285e-01 -1.23562407e+00
3.21444303e-01 -6.79479301e-01 4.96843904e-01 -1.64683974e+00
3.87289263e-02 -4.76728231e-01 -5.00807643e-01 6.44977748e-01
-2.18434751e-01 -1.76814377e-01 -1.73717171e-01 -2.35459246e-02
-1.32791623e-01 7.89844155e-01 7.89980650e-01 -2.24907130e-01
6.59125224e-02 2.74548620e-01 -7.78147578e-01 1.05496550e+00
7.21529543e-01 -4.57933843e-01 -5.94715834e-01 -4.94561642e-01
6.42925501e-01 -2.37813890e-01 5.32623231e-01 -7.38371730e-01
2.24797845e-01 -3.23978812e-01 8.78555715e-01 -8.00254762e-01
6.22433960e-01 -5.70381999e-01 4.31481935e-02 -2.96198279e-02
-2.61061788e-01 -3.10001373e-01 2.90601850e-01 7.58909881e-01
-2.00443659e-02 -5.27489781e-01 4.61297065e-01 -2.27668852e-01
-5.32675862e-01 3.71412963e-01 -7.51950145e-01 -1.25552475e-01
3.49544078e-01 1.04452983e-01 -2.30682895e-01 -4.75277215e-01
-9.85748887e-01 3.80080342e-01 -8.61697644e-02 2.96690427e-02
7.32248545e-01 -1.26801205e+00 -5.32131314e-01 -4.65876386e-02
-2.26388857e-01 5.10380030e-01 7.48314410e-02 3.91144782e-01
-1.53635547e-01 3.41340423e-01 5.02305031e-02 -6.63146973e-01
-1.35366008e-01 2.38032192e-01 4.62856352e-01 2.74941865e-02
-1.46987170e-01 1.21039975e+00 3.96203101e-01 -6.56382322e-01
4.47955668e-01 -1.13610959e+00 -1.04228400e-01 -1.28430733e-02
6.11894429e-01 1.68481737e-01 -3.48125368e-01 2.84628421e-01
-1.86286017e-01 -1.56062722e-01 2.69939899e-01 -5.58312476e-01
1.26978958e+00 1.71009481e-01 -2.39368737e-01 8.37058723e-01
5.38268328e-01 -2.99967021e-01 -1.71035826e+00 -2.05338106e-01
-4.07967120e-02 8.04352537e-02 3.28407139e-01 -9.23489571e-01
-8.92229319e-01 1.34369814e+00 2.75624514e-01 7.88253844e-02
4.68115002e-01 -2.77437251e-02 1.69838324e-01 8.65375161e-01
1.51827127e-01 -1.03429484e+00 -3.21401417e-01 1.03003252e+00
7.93788493e-01 -1.06522965e+00 -1.75321445e-01 -2.30951127e-06
-3.49482983e-01 1.34144485e+00 4.78871912e-01 5.87916970e-02
1.19858503e+00 6.52187169e-01 -4.19016898e-01 -2.54808068e-01
-8.45002234e-01 1.02906451e-02 4.09452796e-01 8.67827535e-01
5.43951571e-01 2.84113586e-01 4.56793934e-01 1.31214190e+00
-3.12794626e-01 -9.08727571e-02 2.86128253e-01 4.68345106e-01
-4.23151374e-01 -8.55478287e-01 -5.31547725e-01 5.85392654e-01
-4.11561728e-01 -6.51199818e-01 3.33170146e-01 3.59208524e-01
6.92758262e-02 8.99812520e-01 3.83617163e-01 5.13363220e-02
-7.62306526e-02 4.22265053e-01 4.79676574e-01 -7.33863831e-01
-4.16062474e-02 -2.73779511e-01 2.12776333e-01 -2.55709261e-01
6.85317665e-02 -5.30826449e-01 -1.46302485e+00 -4.96481955e-01
-1.27845705e-01 -6.89290464e-02 9.10260499e-01 1.03645754e+00
1.56924799e-01 7.83516347e-01 -1.69750229e-01 -9.30000305e-01
-1.04656136e+00 -1.27904916e+00 -6.93934977e-01 -2.33954802e-01
-1.77164733e-01 -1.03918874e+00 -2.83973843e-01 -4.66127731e-02] | [7.2930006980896, 3.848222494125366] |
b9758dc9-db35-44a8-a39c-9c93c1dc7862 | birdsoundsdenoising-deep-visual-audio | 2210.10196 | null | https://arxiv.org/abs/2210.10196v1 | https://arxiv.org/pdf/2210.10196v1.pdf | BirdSoundsDenoising: Deep Visual Audio Denoising for Bird Sounds | Audio denoising has been explored for decades using both traditional and deep learning-based methods. However, these methods are still limited to either manually added artificial noise or lower denoised audio quality. To overcome these challenges, we collect a large-scale natural noise bird sound dataset. We are the first to transfer the audio denoising problem into an image segmentation problem and propose a deep visual audio denoising (DVAD) model. With a total of 14,120 audio images, we develop an audio ImageMask tool and propose to use a few-shot generalization strategy to label these images. Extensive experimental results demonstrate that the proposed model achieves state-of-the-art performance. We also show that our method can be easily generalized to speech denoising, audio separation, audio enhancement, and noise estimation. | ['Jialu Li', 'Youshan Zhang'] | 2022-10-18 | null | null | null | null | ['audio-denoising', 'noise-estimation', 'speech-denoising'] | ['audio', 'medical', 'speech'] | [ 4.56636399e-01 -3.80869001e-01 4.45967376e-01 -2.51899332e-01
-1.24815643e+00 -4.92350698e-01 3.36474866e-01 -8.23151544e-02
-4.57843840e-01 2.26663843e-01 5.98008372e-02 1.43252075e-01
1.18317492e-01 -5.61571300e-01 -5.79816401e-01 -6.12281442e-01
-1.18328311e-01 -1.04830571e-01 3.43949407e-01 -2.07904175e-01
-3.90771218e-02 5.57186566e-02 -1.87732685e+00 4.80218232e-01
7.38644540e-01 1.14202952e+00 9.64129344e-02 1.14442337e+00
3.73675525e-02 7.51768649e-01 -8.30519021e-01 -3.16531092e-01
2.64820784e-01 -5.36518335e-01 -7.04279602e-01 2.01696098e-01
5.15593052e-01 -6.09622002e-01 -3.13154101e-01 1.30307293e+00
9.38689411e-01 3.39965284e-01 4.14829612e-01 -1.25810707e+00
-3.84682655e-01 7.43603826e-01 -5.90959251e-01 1.04325972e-01
2.38264650e-01 1.83397025e-01 8.75132620e-01 -7.85037756e-01
2.69440114e-01 1.34305656e+00 9.05718803e-01 5.98256886e-01
-1.37512493e+00 -8.22003365e-01 1.07069485e-01 4.71192449e-01
-1.18540382e+00 -7.20546424e-01 9.75282490e-01 -3.83620173e-01
6.22392476e-01 2.16470629e-01 7.89393783e-01 1.08257425e+00
-3.16631913e-01 9.52897966e-01 1.12964332e+00 -4.12544668e-01
4.53660309e-01 -3.22886080e-01 -1.62768736e-01 3.68407279e-01
-4.42586839e-01 1.04171120e-01 -5.61148465e-01 7.68735781e-02
4.79292154e-01 -2.92927384e-01 -3.21364254e-01 -3.37995179e-02
-7.06857085e-01 7.42972195e-01 3.82703811e-01 9.87230167e-02
-1.27880752e-01 4.19230819e-01 6.97140038e-01 4.45349604e-01
6.43804193e-01 1.44028023e-01 -1.15087017e-01 -2.52149969e-01
-1.42178380e+00 2.16474995e-01 6.41131103e-01 6.96362853e-01
5.37550330e-01 5.59924603e-01 1.87908243e-02 1.23914492e+00
1.56216741e-01 2.35310584e-01 3.40766013e-01 -1.46340466e+00
-1.10055119e-01 -2.89019972e-01 -2.17453957e-01 -8.24175000e-01
-2.02356920e-01 -4.08167332e-01 -1.01424253e+00 4.42181230e-01
4.65690233e-02 -5.72087988e-03 -1.22511756e+00 1.48727798e+00
1.71093240e-01 5.64139307e-01 -1.98912278e-01 9.45603490e-01
1.07136178e+00 8.63246024e-01 -2.28519291e-02 -1.84198812e-01
1.04118478e+00 -1.30530787e+00 -8.44929755e-01 -3.53054032e-02
-8.75947326e-02 -8.92963707e-01 1.05344784e+00 1.13484466e+00
-1.14778912e+00 -8.74783695e-01 -1.12681413e+00 -1.91343337e-01
-4.39905897e-02 -4.97111194e-02 3.69753212e-01 7.89782643e-01
-1.24146008e+00 6.62008524e-01 -9.39668179e-01 -3.03648710e-01
5.80966115e-01 1.34927467e-01 -6.21076487e-02 -5.04766256e-02
-9.06263053e-01 1.98117167e-01 1.72543496e-01 2.35184789e-01
-1.80549061e+00 -6.11544311e-01 -8.61476123e-01 8.37641880e-02
4.49410498e-01 -5.75950503e-01 1.58236694e+00 -1.14473665e+00
-1.70066822e+00 7.04353213e-01 -1.12626171e-02 -7.21985459e-01
4.74706501e-01 -4.65868682e-01 -4.19728130e-01 5.50068915e-01
3.67512666e-02 9.53150809e-01 1.36558640e+00 -1.53297865e+00
-7.00069666e-01 8.02137982e-03 -3.19206268e-02 2.57829502e-02
-4.33821380e-01 1.79403335e-01 -6.83455586e-01 -1.17766106e+00
-5.45690991e-02 -5.27072728e-01 -2.15983406e-01 1.99950933e-01
-2.87188679e-01 3.05828363e-01 8.44977140e-01 -8.86367500e-01
1.04353166e+00 -2.55963516e+00 2.43175164e-01 -1.04677610e-01
3.60155612e-01 3.90387982e-01 -5.61697185e-01 1.12583116e-01
-3.43758278e-02 1.39194533e-01 -5.32373548e-01 -7.20782757e-01
6.11406602e-02 2.47613803e-01 -4.92589504e-01 2.74315685e-01
6.12260215e-02 4.34233844e-01 -9.06996310e-01 -4.25819457e-01
2.60163218e-01 7.45946646e-01 -6.99142992e-01 1.39047593e-01
-1.10014282e-01 3.45272988e-01 2.13016704e-01 7.61135101e-01
9.41225469e-01 5.58559179e-01 -1.46559998e-01 -1.58250988e-01
5.89093380e-02 4.80451211e-02 -1.37918472e+00 1.94615376e+00
-3.67772400e-01 7.99222827e-01 8.00114393e-01 -1.04072845e+00
7.32403934e-01 3.34378302e-01 3.21695536e-01 -4.92587477e-01
1.12179406e-01 1.49989024e-01 -1.96747899e-01 -5.37529767e-01
4.88597184e-01 -3.93667340e-01 1.76552102e-01 1.60719484e-01
4.99114603e-01 -4.80328918e-01 2.02715054e-01 1.65515825e-01
1.09052229e+00 -7.86636956e-03 -3.08483019e-02 8.49563330e-02
4.00667012e-01 -2.58729130e-01 5.70421338e-01 8.97750258e-01
-4.95992422e-01 1.03721535e+00 4.16494608e-01 -7.58533478e-02
-8.18296969e-01 -1.24847901e+00 3.80562842e-02 1.50242984e+00
5.29353181e-03 -6.22593343e-01 -1.25566721e+00 -4.58350986e-01
-3.02361071e-01 3.98494691e-01 -3.44045997e-01 -1.66043818e-01
-2.88556993e-01 -4.19131547e-01 1.04249740e+00 5.58722079e-01
5.82618713e-01 -9.76788044e-01 -2.94149488e-01 3.51473302e-01
-3.57778609e-01 -8.69926095e-01 -4.31822181e-01 3.95976067e-01
-6.40190959e-01 -7.30267584e-01 -7.50454307e-01 -1.01841509e+00
4.40029847e-03 3.85562927e-01 9.57430542e-01 8.20652470e-02
-4.73364115e-01 7.45232046e-01 -5.25695324e-01 -4.46701586e-01
-5.47687531e-01 -2.28030443e-01 1.36451244e-01 5.75962178e-02
1.19761832e-01 -8.29033256e-01 -6.50541365e-01 9.19177532e-02
-1.24465454e+00 -3.90445411e-01 4.03190196e-01 9.45344448e-01
8.00297558e-01 4.73218769e-01 7.16814041e-01 -3.20123792e-01
8.11998844e-01 -8.48662332e-02 -3.33388329e-01 -1.53291672e-01
-2.51284719e-01 -4.46130812e-01 5.72007298e-01 -6.81544065e-01
-1.09729028e+00 1.93196923e-01 -7.07073689e-01 -5.58932424e-01
-7.15203822e-01 4.65775549e-01 -2.67426670e-01 -8.90878513e-02
5.50378084e-01 1.13477290e-01 -4.48030792e-02 -9.18422759e-01
5.89774549e-01 8.49931061e-01 1.02136803e+00 -3.50366890e-01
6.42395914e-01 6.38310194e-01 -2.93473452e-01 -1.25695050e+00
-8.15666139e-01 -4.59645480e-01 -3.91064107e-01 -3.96095753e-01
8.29310596e-01 -1.20557654e+00 -4.06688720e-01 7.56514549e-01
-9.73897696e-01 -4.51310486e-01 -4.35394675e-01 3.37112486e-01
-5.70498526e-01 7.30165005e-01 -1.00631499e+00 -9.03064728e-01
-2.52417058e-01 -1.32557762e+00 1.13962793e+00 1.66712236e-02
3.13370377e-02 -4.88440931e-01 1.71415582e-01 3.70781422e-01
2.81142086e-01 -1.12245582e-01 5.78963101e-01 -2.85664588e-01
-4.11272883e-01 7.14761764e-02 -8.70442316e-02 9.66632187e-01
-6.49365336e-02 5.90531267e-02 -1.54789937e+00 -3.48388225e-01
2.84384757e-01 -5.57300091e-01 1.59173632e+00 6.47722244e-01
1.43411279e+00 6.44794405e-02 2.10995793e-01 9.34424996e-01
1.06330442e+00 1.47225693e-01 7.92240083e-01 2.70557880e-01
5.32478333e-01 3.95842701e-01 5.11280298e-01 4.53175575e-01
-9.94267873e-03 5.40182233e-01 6.96400106e-01 -3.55295539e-01
-4.71326619e-01 -2.72452265e-01 5.44431090e-01 1.01427734e+00
2.69390345e-02 -1.98528215e-01 -6.82320774e-01 7.87723362e-01
-1.55573046e+00 -9.63248014e-01 2.45590471e-02 1.78161407e+00
9.88819420e-01 5.68390973e-02 2.75660843e-01 6.55971289e-01
6.58465445e-01 1.08995259e-01 -2.68667996e-01 -2.72597462e-01
-1.36490330e-01 6.23640537e-01 -8.13878048e-03 4.50635523e-01
-1.62332249e+00 1.00340104e+00 7.02318239e+00 1.36560071e+00
-9.99553561e-01 1.63639009e-01 4.47652519e-01 -2.12430745e-01
-3.74281108e-02 -2.53917128e-01 -2.25017384e-01 2.39077926e-01
7.51101434e-01 1.97093919e-01 8.33638132e-01 8.49960148e-01
3.11954707e-01 -1.87674370e-02 -8.88554811e-01 1.21458769e+00
5.20865135e-02 -9.38035667e-01 5.78404777e-02 -2.46384487e-01
6.24124587e-01 -2.05074742e-01 2.75562167e-01 3.05599779e-01
7.27547109e-02 -1.05936444e+00 9.32428420e-01 3.70484382e-01
7.93888569e-01 -1.00604284e+00 5.38973033e-01 1.28159910e-01
-1.45127594e+00 -2.45202243e-01 -4.13667679e-01 -1.89919338e-01
1.84138909e-01 5.60518146e-01 -4.61409837e-01 4.71841574e-01
1.35617602e+00 7.93047786e-01 -5.62067986e-01 1.40438843e+00
-3.94151449e-01 1.17966413e+00 -2.72347987e-01 5.07513165e-01
7.65533671e-02 -1.16753921e-01 7.75617480e-01 1.39118290e+00
2.83371210e-01 -1.65196598e-01 1.33453444e-01 6.17251217e-01
-2.09684834e-01 -1.07130617e-01 -3.04307789e-01 1.93354823e-02
2.01639816e-01 1.38973439e+00 -8.98246050e-01 -3.10753286e-01
-2.62773156e-01 1.09990335e+00 -2.53005773e-01 5.51452339e-01
-8.71131480e-01 -7.82749593e-01 8.92426491e-01 -9.49441865e-02
6.62681520e-01 -5.95566165e-03 -2.20991477e-01 -9.23837125e-01
-1.07735842e-01 -1.25365281e+00 1.71728075e-01 -9.48763013e-01
-1.31862462e+00 6.38603568e-01 -1.87931374e-01 -1.20195603e+00
-1.29718214e-01 -3.44285280e-01 -4.97326493e-01 3.63357365e-01
-1.47548079e+00 -1.06914318e+00 -4.03028727e-01 5.82808912e-01
8.95047545e-01 -1.31702706e-01 6.26141071e-01 8.03398192e-01
-3.14882904e-01 4.30005044e-01 2.27025881e-01 1.53417870e-01
8.77890706e-01 -1.33545077e+00 3.29349220e-01 1.00121713e+00
3.41345578e-01 1.95308417e-01 9.11710143e-01 -3.54431450e-01
-9.26687360e-01 -1.16591716e+00 1.31373063e-01 1.18834926e-02
6.98274374e-01 -6.41472280e-01 -1.04058588e+00 3.69434386e-01
5.85054040e-01 -1.49492517e-01 6.72540843e-01 -2.49852479e-01
-2.60183901e-01 -4.25277531e-01 -1.06422424e+00 3.71893913e-01
1.01450324e+00 -6.75019443e-01 -5.39798975e-01 -1.88798487e-01
9.55175817e-01 -2.00081453e-01 -5.34937084e-01 3.26376081e-01
3.43153119e-01 -9.86505449e-01 1.15178967e+00 -3.28000784e-01
4.33486372e-01 -5.56314647e-01 -1.32702440e-01 -1.52528751e+00
-1.16517767e-01 -7.98644722e-01 8.87502283e-02 1.69587338e+00
1.08441869e-02 5.41492626e-02 3.35536361e-01 -2.98772037e-01
-4.37656134e-01 -1.86623260e-01 -1.03852522e+00 -7.70521045e-01
-6.61851913e-02 -1.06444323e+00 3.47505480e-01 7.59125888e-01
-4.36831683e-01 1.94749266e-01 -7.09818304e-01 1.50402158e-01
8.35529327e-01 -2.56618261e-01 7.73409128e-01 -1.22496438e+00
-4.59614098e-01 -3.36642563e-01 -4.30995911e-01 -1.25088501e+00
1.15236104e-01 -4.61772352e-01 6.86877310e-01 -1.50950491e+00
7.86519200e-02 2.19818398e-01 -2.87541598e-01 4.35088485e-01
1.15009800e-01 7.29553103e-01 3.58586349e-02 -1.65363938e-01
-7.67571926e-01 7.85148799e-01 8.37977886e-01 -4.83948797e-01
-2.13388592e-01 -1.04375206e-01 -5.98101497e-01 9.68080521e-01
6.68527305e-01 -5.53103149e-01 -3.41716260e-01 -5.77475190e-01
3.54413278e-02 -2.16033190e-01 7.36574650e-01 -1.11527169e+00
1.69238895e-01 2.80731291e-01 2.15017170e-01 -5.85340321e-01
4.59589630e-01 -7.39440799e-01 -6.23389296e-02 2.30614498e-01
-4.96334434e-01 -5.39126694e-01 4.69738901e-01 9.46726680e-01
-7.25389421e-01 -2.33586818e-01 1.06803763e+00 -9.64855179e-02
-8.09643626e-01 -2.00340655e-02 -8.82594645e-01 6.38515800e-02
5.83468795e-01 -2.48079770e-03 -1.40951291e-01 -7.98315167e-01
-1.06261981e+00 2.99661141e-02 2.49083221e-01 2.21131995e-01
8.79728019e-01 -1.17788088e+00 -5.98911524e-01 1.50977105e-01
-2.48468056e-01 -1.83442831e-02 6.32644176e-01 7.06432998e-01
-6.19418383e-01 -3.25775981e-01 -1.68566421e-01 -8.07110786e-01
-1.58894122e+00 5.92489779e-01 3.09236825e-01 2.69638687e-01
-6.42271817e-01 1.05856109e+00 1.99660555e-01 -2.41107896e-01
8.41513097e-01 -4.26887512e-01 -1.11226134e-01 1.99381620e-01
6.81567192e-01 5.70919216e-01 1.48012489e-01 -4.83414948e-01
-7.52500072e-02 4.72473770e-01 2.42984146e-01 -3.77585083e-01
1.59909058e+00 -2.68503815e-01 -1.24410570e-01 4.10249144e-01
9.48993623e-01 2.15536766e-02 -1.42108703e+00 2.01773494e-02
-3.98287416e-01 -4.65651214e-01 4.84301597e-01 -6.11303091e-01
-1.29685152e+00 1.43237281e+00 1.03242314e+00 4.92303908e-01
1.82281971e+00 -1.38266101e-01 9.12001908e-01 2.49687508e-01
-1.06095918e-01 -1.23448753e+00 4.55725521e-01 4.03786421e-01
8.23121846e-01 -1.04387271e+00 -2.93391556e-01 -3.59924138e-01
-5.40968597e-01 1.09543669e+00 4.33764726e-01 6.11034259e-02
6.12826943e-01 6.60379827e-01 4.67573285e-01 1.82290182e-01
-5.88139534e-01 -5.36811888e-01 1.91035837e-01 1.09719336e+00
1.65853783e-01 -2.68350780e-01 1.99314639e-01 9.20705140e-01
-1.25367627e-01 7.55678341e-02 6.23838246e-01 8.78702700e-01
-7.62351394e-01 -9.78781283e-01 -5.07995486e-01 -7.90686607e-02
-6.00921273e-01 -2.37919554e-01 -4.96442437e-01 3.82733524e-01
2.32993022e-01 1.33293617e+00 -8.65864754e-02 -5.75427771e-01
3.25251251e-01 6.05856143e-02 2.89271951e-01 -4.55546826e-01
-6.31978095e-01 9.83256221e-01 -2.05673531e-01 -3.70878994e-01
-6.94155395e-01 -2.95007259e-01 -9.52614665e-01 -1.26655295e-01
-2.94437766e-01 3.40686850e-02 5.11680841e-01 6.03926122e-01
1.43138215e-01 8.60173821e-01 3.75122249e-01 -1.14800084e+00
-4.79861170e-01 -1.00117290e+00 -8.75874817e-01 2.09678486e-01
5.61418355e-01 -4.49463516e-01 -6.22003019e-01 4.86708403e-01] | [15.194742202758789, 5.559787750244141] |
3384ba0e-8a3a-4e8d-b57c-17d76e279eb8 | disk-learning-local-features-with-policy | 2006.13566 | null | https://arxiv.org/abs/2006.13566v2 | https://arxiv.org/pdf/2006.13566v2.pdf | DISK: Learning local features with policy gradient | Local feature frameworks are difficult to learn in an end-to-end fashion, due to the discreteness inherent to the selection and matching of sparse keypoints. We introduce DISK (DIScrete Keypoints), a novel method that overcomes these obstacles by leveraging principles from Reinforcement Learning (RL), optimizing end-to-end for a high number of correct feature matches. Our simple yet expressive probabilistic model lets us keep the training and inference regimes close, while maintaining good enough convergence properties to reliably train from scratch. Our features can be extracted very densely while remaining discriminative, challenging commonly held assumptions about what constitutes a good keypoint, as showcased in Fig. 1, and deliver state-of-the-art results on three public benchmarks. | ['Michał J. Tyszkiewicz', 'Pascal Fua', 'Eduard Trulls'] | 2020-06-24 | null | http://proceedings.neurips.cc/paper/2020/hash/a42a596fc71e17828440030074d15e74-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/a42a596fc71e17828440030074d15e74-Paper.pdf | neurips-2020-12 | ['image-matching'] | ['computer-vision'] | [-4.50576097e-01 -1.79212764e-01 -4.91565883e-01 -1.41142741e-01
-1.48524606e+00 -6.94646657e-01 9.28849161e-01 2.63587952e-01
-4.58285451e-01 6.55719757e-01 2.75960773e-01 1.66277692e-01
-4.91076887e-01 -5.14643431e-01 -9.26044226e-01 -6.04448855e-01
-4.85553920e-01 4.03773040e-01 1.94767416e-01 -2.37713289e-02
3.99372727e-01 7.35884011e-01 -1.57394302e+00 7.90525675e-02
4.63756025e-01 1.01394498e+00 2.05739722e-01 7.09061146e-01
3.61697197e-01 7.52904773e-01 -1.65299907e-01 -2.84732133e-01
4.47175771e-01 9.51040164e-03 -6.83967173e-01 -2.65215605e-01
8.24437380e-01 -4.28578794e-01 -5.65868080e-01 8.17833126e-01
4.29848701e-01 3.21315080e-01 5.63996911e-01 -1.26453757e+00
-5.94067156e-01 1.26292363e-01 -3.61468583e-01 2.08093226e-01
4.77884442e-01 1.03254072e-01 1.61220360e+00 -1.20608342e+00
6.31951213e-01 1.03956091e+00 9.69195604e-01 1.96104944e-01
-1.33460307e+00 -5.52908480e-01 1.66397974e-01 1.21883690e-01
-1.62103021e+00 -8.50029469e-01 4.50034350e-01 -1.38140410e-01
1.06030095e+00 1.82539195e-01 5.84306836e-01 7.50014842e-01
2.84272939e-01 9.93511438e-01 8.74200761e-01 -3.88581157e-01
1.83068335e-01 -2.51669705e-01 -1.59013942e-01 8.33779812e-01
-1.27658322e-01 5.81417739e-01 -8.70018661e-01 -6.42294466e-01
8.21382940e-01 3.07786584e-01 -9.41336527e-02 -9.42955077e-01
-1.34473658e+00 1.00801849e+00 6.84402764e-01 1.16505519e-01
-5.50666034e-01 5.82711875e-01 2.17165500e-01 2.67201483e-01
1.98653534e-01 3.94899517e-01 -5.67657888e-01 -5.02647400e-01
-1.23062468e+00 7.01333880e-01 6.80020928e-01 1.02004242e+00
1.09749866e+00 -4.35120225e-01 -1.95797086e-01 6.20425284e-01
3.06820095e-01 6.48195803e-01 3.22324187e-01 -1.01841164e+00
1.15677804e-01 7.82494396e-02 3.52854252e-01 -9.79661405e-01
-2.62810498e-01 -3.48122239e-01 -4.91123110e-01 1.85924008e-01
3.40041757e-01 3.00538782e-02 -7.72498310e-01 1.71091342e+00
3.78132403e-01 4.86998826e-01 -1.09973349e-01 6.33930683e-01
2.77257323e-01 4.75243002e-01 1.19735138e-03 7.79327452e-02
1.01102960e+00 -9.77544844e-01 -1.38834015e-01 -1.69080079e-01
5.50866306e-01 -8.44011784e-01 9.90309358e-01 3.71450484e-01
-1.04419744e+00 -4.76896584e-01 -7.94925749e-01 -4.07003500e-02
-1.60017490e-01 -8.63655657e-02 9.55437839e-01 1.42280608e-01
-1.14686573e+00 7.92541504e-01 -1.05696535e+00 -1.31571725e-01
4.90355700e-01 4.94848251e-01 -6.02121234e-01 -2.26582050e-01
-7.71330178e-01 9.03591394e-01 7.71531910e-02 -2.00117469e-01
-1.04573989e+00 -1.10658383e+00 -8.48671615e-01 -1.14145152e-01
5.35235941e-01 -8.32239866e-01 1.50869834e+00 -4.53105539e-01
-1.37089348e+00 6.40940130e-01 -2.99554169e-01 -5.57604134e-01
5.85933983e-01 -6.79040730e-01 -6.05826732e-03 3.17045182e-01
4.13417578e-01 7.67363429e-01 8.42483997e-01 -9.93832052e-01
-9.14065123e-01 -9.94747207e-02 -9.11675021e-02 1.98227853e-01
2.87325811e-02 2.22312566e-02 -5.81905961e-01 -6.51647329e-01
-1.52634561e-01 -9.72631633e-01 -4.35084224e-01 2.64385462e-01
-1.69529662e-01 -5.20716131e-01 5.33876300e-01 -2.23617151e-01
8.72188270e-01 -2.21519995e+00 4.00158837e-02 5.24924815e-01
2.08316728e-01 -1.06691457e-02 -2.11081535e-01 8.35925460e-01
2.35584363e-01 -1.89890176e-01 -4.34871577e-02 -4.88051116e-01
3.71963829e-01 1.79891795e-01 -4.87980753e-01 7.41632164e-01
2.07281783e-01 1.09985971e+00 -1.21915853e+00 -4.16624308e-01
2.99990505e-01 3.12581837e-01 -7.96847820e-01 4.23447549e-01
-6.00715093e-02 -2.64222324e-02 -6.46543920e-01 6.67708516e-01
3.87693882e-01 -3.17598879e-01 -1.60329252e-01 -1.54857547e-03
-2.53492612e-02 2.18192399e-01 -1.42718744e+00 2.06583738e+00
-6.06313288e-01 2.72507668e-01 4.50504059e-03 -6.43628478e-01
7.40304708e-01 2.30796169e-02 6.78477347e-01 -4.73635614e-01
-2.62559682e-01 2.00360790e-01 -4.18203443e-01 -1.93790659e-01
4.11962271e-01 -1.52579412e-01 -2.55296350e-01 3.43442529e-01
2.91750699e-01 -2.49509469e-01 6.09061904e-02 2.31832817e-01
1.22506547e+00 3.02999198e-01 4.60311681e-01 -3.28072965e-01
-1.73626374e-02 -1.08500116e-01 3.68902475e-01 1.13616633e+00
1.51173433e-03 6.84569240e-01 1.51323512e-01 -4.59835291e-01
-1.06727409e+00 -1.29769087e+00 -1.51922360e-01 1.21780396e+00
1.25228763e-01 -6.93157017e-01 -2.65264928e-01 -8.56410801e-01
6.08199000e-01 2.71431416e-01 -7.44802117e-01 -2.11576477e-01
-3.60970229e-01 -2.49792859e-01 5.26946425e-01 5.63446224e-01
9.29985940e-02 -9.09406364e-01 -7.00302899e-01 4.20142680e-01
2.11552933e-01 -1.10596955e+00 -5.30135989e-01 4.62799072e-01
-5.68284273e-01 -1.12826800e+00 -7.51080096e-01 -5.82989454e-01
4.19520617e-01 3.30998689e-01 1.53843939e+00 1.73271194e-01
-4.66316819e-01 6.00233078e-01 -3.31203610e-01 -9.22976583e-02
-6.21716958e-03 5.55743724e-02 1.54076919e-01 -2.68596590e-01
3.58494133e-01 -5.92685938e-01 -7.28893816e-01 1.53527901e-01
-6.49972379e-01 -4.62305099e-01 8.63460958e-01 1.05670202e+00
9.68808830e-01 -2.59552956e-01 4.58478242e-01 -6.33534431e-01
3.13478023e-01 -5.92983902e-01 -5.03236949e-01 2.60265291e-01
-4.82278466e-01 2.50670046e-01 7.43981063e-01 -2.30606079e-01
-3.01805347e-01 1.60856664e-01 -1.58885539e-01 -7.07695484e-01
-2.66609669e-01 2.64181435e-01 2.68803775e-01 -5.12388647e-01
5.97121119e-01 3.47094119e-01 1.14648394e-01 -5.20149529e-01
7.45720983e-01 4.25072074e-01 6.70972943e-01 -1.01318181e+00
9.77364838e-01 5.80725551e-01 -5.72917052e-02 -6.68249369e-01
-9.68324065e-01 -8.79508734e-01 -6.47002995e-01 3.61831427e-01
7.94752911e-02 -1.09231031e+00 -7.49591827e-01 1.37497276e-01
-6.37158751e-01 -3.99914980e-01 -6.48190379e-01 4.39755499e-01
-1.01586318e+00 3.76321167e-01 -4.69362438e-01 -3.80492568e-01
-4.64441597e-01 -8.64798367e-01 1.43246472e+00 4.27912585e-02
-3.62861991e-01 -8.37242365e-01 4.37694371e-01 -3.30365986e-01
4.06168908e-01 2.05512673e-01 5.83048344e-01 -5.83473086e-01
-7.18024552e-01 -4.55554336e-01 -1.40786007e-01 2.10416362e-01
1.97587209e-03 3.19828689e-02 -7.94203937e-01 -5.51537335e-01
-4.60334122e-01 -7.08267450e-01 8.98975730e-01 2.53923148e-01
1.05646276e+00 -2.02712387e-01 -3.79942417e-01 6.98427796e-01
1.55393600e+00 -5.94259501e-01 3.88159394e-01 3.75727773e-01
5.40674031e-01 1.73537165e-01 8.35050941e-01 6.85952187e-01
5.51741362e-01 7.29818046e-01 2.98740119e-01 -8.35635886e-02
-6.34604767e-02 -5.87738693e-01 1.71241716e-01 3.75548571e-01
3.41270655e-01 2.85382003e-01 -7.21197367e-01 7.39825606e-01
-2.08077765e+00 -9.64196384e-01 5.00729799e-01 2.30812097e+00
1.01734340e+00 6.73103705e-02 3.81879419e-01 -1.73530251e-01
2.37964302e-01 1.94288686e-01 -4.47454810e-01 -7.48748332e-02
1.75951980e-02 4.74787086e-01 5.62433839e-01 4.95437175e-01
-1.18567932e+00 1.12191439e+00 7.35514784e+00 1.09634113e+00
-8.84252012e-01 -4.25188504e-02 3.23656619e-01 -2.23562047e-01
-3.32947284e-01 1.87814713e-01 -9.50828910e-01 2.92921305e-01
8.22806358e-01 2.99730022e-02 5.91260195e-01 9.22374129e-01
-8.12088847e-02 -1.10865042e-01 -1.38595903e+00 1.13555968e+00
-1.37286246e-01 -1.54575539e+00 -9.49546993e-02 -9.39585418e-02
7.13533401e-01 5.24326742e-01 1.83581442e-01 3.68853241e-01
7.17821181e-01 -1.10774910e+00 7.54742801e-01 7.40102768e-01
7.18593001e-01 -9.96111274e-01 3.25482488e-01 3.69110852e-01
-1.12177384e+00 -7.16184154e-02 -5.50329447e-01 6.40366897e-02
-1.80226669e-01 6.51080966e-01 -6.73709810e-01 5.76323450e-01
7.54678488e-01 8.11584949e-01 -5.40997624e-01 1.44269240e+00
-1.98079526e-01 2.93424308e-01 -8.03638160e-01 -1.09958336e-01
6.75883472e-01 3.10105294e-01 3.25876832e-01 1.38785386e+00
2.20922828e-01 -2.34014273e-01 6.15600228e-01 6.63755000e-01
-1.60090864e-01 -5.08127101e-02 -7.73635685e-01 3.03313792e-01
7.67844081e-01 1.32404625e+00 -4.11270916e-01 -8.17334428e-02
-4.30178463e-01 8.08073938e-01 8.02496016e-01 2.74233848e-01
-4.75242913e-01 -5.02195120e-01 6.90142989e-01 -5.17199449e-02
6.69153452e-01 -3.61631721e-01 1.39162332e-01 -1.10505474e+00
2.55652249e-01 -9.25537288e-01 4.13671643e-01 -3.18709791e-01
-1.68840075e+00 2.38136992e-01 5.91358468e-02 -1.13781595e+00
-6.27693951e-01 -4.26160961e-01 -5.58123112e-01 7.32807636e-01
-1.82902455e+00 -1.36608851e+00 -6.93800524e-02 8.26697767e-01
1.72907531e-01 3.57903391e-02 1.00998235e+00 -2.13648416e-02
4.87123616e-02 9.15286124e-01 3.92456740e-01 1.50412330e-02
8.50886345e-01 -1.46950877e+00 5.65462232e-01 6.17244005e-01
6.17717683e-01 6.47432685e-01 5.67213356e-01 -2.61660963e-01
-1.77659142e+00 -7.38929152e-01 7.07287729e-01 -6.49253666e-01
8.49982977e-01 -3.57980520e-01 -6.33256733e-01 6.30010843e-01
-7.11340234e-02 7.42219806e-01 6.40583456e-01 5.85276484e-01
-5.85477471e-01 -1.08637139e-01 -1.09533298e+00 4.54723179e-01
9.50463295e-01 -8.57343972e-01 -7.05020070e-01 3.48254889e-01
4.59096164e-01 -4.41342682e-01 -1.06169796e+00 3.24438095e-01
6.01961374e-01 -9.66481745e-01 1.26384342e+00 -5.99305570e-01
1.00795507e-01 -1.29344463e-01 -3.87491763e-01 -1.32643449e+00
-5.64091802e-01 -1.12755239e+00 -5.08733690e-01 9.83283758e-01
1.32381022e-01 -3.42262685e-01 9.31965053e-01 3.83279383e-01
-1.90111306e-02 -1.26454008e+00 -1.02971351e+00 -9.52946961e-01
1.31010294e-01 -2.68444777e-01 6.45099819e-01 7.12498307e-01
5.31345196e-02 -7.88228810e-02 -3.54296029e-01 1.59021318e-01
7.18321681e-01 3.93462300e-01 1.00538504e+00 -1.12530792e+00
-6.00241542e-01 -3.80296409e-01 -5.84159374e-01 -1.40344656e+00
2.18120158e-01 -8.51306379e-01 1.68532759e-01 -1.12915146e+00
2.85461575e-01 -9.68633533e-01 -6.53048754e-01 6.17154896e-01
-3.05928499e-01 3.15291047e-01 2.82445163e-01 4.35755998e-01
-1.09048831e+00 7.14958966e-01 9.64996755e-01 2.53594756e-01
3.12928408e-02 -6.40506819e-02 -7.57729352e-01 5.83804250e-01
3.81427735e-01 -5.40308356e-01 -1.68886945e-01 -2.98116088e-01
1.83197886e-01 5.60008958e-02 5.75811625e-01 -9.34071779e-01
5.24824381e-01 -2.57319421e-01 6.30888999e-01 -5.00829399e-01
3.36200088e-01 -7.64182925e-01 -4.21905220e-01 1.83964536e-01
-3.44973296e-01 2.33431891e-01 1.48526102e-01 8.23382914e-01
-2.47608036e-01 -1.65224418e-01 7.02248514e-01 -6.01057755e-03
-8.66838217e-01 7.07035244e-01 2.33756676e-01 3.72805387e-01
1.09503412e+00 -1.93463545e-02 2.23847348e-02 -3.27141106e-01
-5.18572152e-01 3.30027014e-01 6.34546220e-01 3.02887052e-01
6.76435947e-01 -1.46280384e+00 -7.50701129e-01 2.66296536e-01
3.44069242e-01 1.62855506e-01 8.46094862e-02 6.02411985e-01
-2.99170911e-01 1.99141458e-01 -5.13555259e-02 -7.09417880e-01
-8.00830424e-01 5.75718403e-01 2.80442536e-01 -3.41382682e-01
-9.53498662e-01 9.97839451e-01 -1.24332905e-01 -3.92788172e-01
4.09601748e-01 -6.49441928e-02 4.14218485e-01 -2.39163369e-01
5.88708699e-01 1.70301139e-01 1.96603104e-01 -2.83860713e-01
-4.32834506e-01 6.94616556e-01 -5.13370514e-01 -2.43530311e-02
1.43197632e+00 2.25913953e-02 3.15115392e-01 2.65302032e-01
1.47457600e+00 2.38161325e-01 -1.76842260e+00 -5.75749040e-01
1.71103012e-02 -8.22762609e-01 1.63012087e-01 -6.47605658e-01
-5.80472589e-01 5.29794753e-01 4.91285592e-01 2.79830135e-02
6.00927413e-01 9.64908749e-02 9.60089982e-01 8.12056720e-01
6.00444019e-01 -1.01581585e+00 7.15581700e-02 5.08133113e-01
6.38553321e-01 -1.15271103e+00 8.94475803e-02 9.55208316e-02
-5.47432005e-01 1.05609667e+00 1.07364289e-01 -8.97345126e-01
8.60256612e-01 3.63807797e-01 -1.16830237e-01 -2.55690634e-01
-8.67499053e-01 -2.38322020e-01 3.07261646e-01 5.24934232e-01
5.36090694e-02 -4.72386740e-02 3.89780283e-01 2.62012661e-01
-2.05046907e-01 4.56533534e-03 -1.08021967e-01 1.18808162e+00
-5.42072475e-01 -1.12541449e+00 -9.18245316e-02 4.46844757e-01
-6.27304852e-01 -1.53619453e-01 -9.55087878e-03 9.45377946e-01
-1.72026172e-01 6.71800673e-01 7.78085887e-02 -1.94737375e-01
3.57059360e-01 -2.05459446e-01 6.91216230e-01 -5.68152905e-01
-5.04984856e-01 1.09540873e-04 -2.00785786e-01 -1.04609215e+00
-1.98010534e-01 -7.28484035e-01 -1.24662197e+00 -4.47210282e-01
-3.05134743e-01 3.79631400e-01 2.67881930e-01 9.96637881e-01
6.59695804e-01 2.81361137e-02 9.13203001e-01 -1.03832746e+00
-1.31886947e+00 -6.47139549e-01 -6.20661318e-01 5.15915811e-01
6.19445503e-01 -7.08893895e-01 -1.85288638e-01 -3.23583037e-01] | [8.149181365966797, -1.8818085193634033] |
8f0467ab-43e1-442f-8e35-37983291de51 | aaai-fss-19-human-centered-ai-trustworthiness | 2001.05375 | null | https://arxiv.org/abs/2001.05375v1 | https://arxiv.org/pdf/2001.05375v1.pdf | AAAI FSS-19: Human-Centered AI: Trustworthiness of AI Models and Data Proceedings | To facilitate the widespread acceptance of AI systems guiding decision-making in real-world applications, it is key that solutions comprise trustworthy, integrated human-AI systems. Not only in safety-critical applications such as autonomous driving or medicine, but also in dynamic open world systems in industry and government it is crucial for predictive models to be uncertainty-aware and yield trustworthy predictions. Another key requirement for deployment of AI at enterprise scale is to realize the importance of integrating human-centered design into AI systems such that humans are able to use systems effectively, understand results and output, and explain findings to oversight committees. While the focus of this symposium was on AI systems to improve data quality and technical robustness and safety, we welcomed submissions from broadly defined areas also discussing approaches addressing requirements such as explainable models, human trust and ethical aspects of AI. | ['Ulli Waltinger', 'John Piorkowski', 'Ian McCulloh', 'Florian Buettner'] | 2020-01-15 | null | null | null | null | ['explainable-models'] | ['computer-vision'] | [-8.81447196e-02 6.31582916e-01 -2.21155331e-01 -7.81688154e-01
-3.96656811e-01 -5.46379685e-01 3.29091698e-01 2.35624209e-01
-2.40582287e-01 8.16218913e-01 -3.27999778e-02 -6.56999707e-01
-3.01061124e-01 -5.39819002e-01 -5.09124815e-01 1.41085973e-02
1.14604771e-01 5.72270811e-01 -2.50642926e-01 -1.61827832e-01
2.61317104e-01 7.55774617e-01 -1.30185461e+00 -1.71838738e-02
1.01507246e+00 1.09323776e+00 -6.33950591e-01 5.74750543e-01
3.66907299e-01 1.16387987e+00 -3.28195989e-01 -6.85346544e-01
4.77741748e-01 -2.51894951e-01 -7.11303711e-01 -5.07362604e-01
-1.53333127e-01 -3.28963518e-01 1.32851139e-01 9.05889988e-01
1.03673581e-02 -7.06069693e-02 4.59965169e-01 -1.81055927e+00
-8.02125096e-01 5.30986249e-01 -7.43015260e-02 -3.74986053e-01
1.82914168e-01 8.12779248e-01 8.26639533e-01 -3.90431941e-01
4.66426343e-01 1.05631018e+00 5.25776923e-01 5.37809491e-01
-1.15850997e+00 -9.81500506e-01 2.62383610e-01 2.05679342e-01
-1.22589505e+00 -8.17273438e-01 3.39526325e-01 -6.64004266e-01
9.08373892e-01 5.91383815e-01 7.95309424e-01 7.92919874e-01
1.01810277e+00 4.78909791e-01 8.91055226e-01 -2.07728133e-01
6.96651340e-01 6.03464365e-01 1.99242175e-01 2.57787853e-01
1.08568525e+00 3.92582446e-01 -5.58510363e-01 -1.31544232e-01
2.95348763e-01 -4.06083390e-02 1.85540944e-01 -2.73463100e-01
-1.36122572e+00 7.51285434e-01 2.96987116e-01 2.54084855e-01
-8.84218931e-01 5.61763704e-01 7.05309361e-02 1.25258595e-01
1.77319825e-01 1.08599567e+00 -4.65809584e-01 -4.74725097e-01
-6.77599490e-01 4.93152380e-01 9.08588588e-01 1.07267118e+00
2.70618141e-01 2.74725646e-01 3.79729778e-01 -9.07450765e-02
7.52110362e-01 7.01682806e-01 -2.72418037e-02 -1.49771678e+00
-2.46626303e-01 8.36732566e-01 5.00672281e-01 -1.04355597e+00
-4.13929194e-01 -1.87432930e-01 -5.95818102e-01 4.46857601e-01
-2.22460240e-01 -2.63040483e-01 -9.33113933e-01 1.17598367e+00
1.51825413e-01 -5.23042440e-01 2.70015180e-01 1.10622180e+00
6.83146060e-01 3.63554716e-01 6.16502464e-01 -9.29853786e-03
1.16785443e+00 -2.65343845e-01 -9.80072916e-01 -5.61547160e-01
4.89512980e-01 -4.63498324e-01 6.57605171e-01 5.50920546e-01
-1.11988032e+00 -2.33266801e-01 -1.11540413e+00 -5.12321538e-04
-3.10434908e-01 -4.98314708e-01 8.10162663e-01 6.71196222e-01
-4.57932174e-01 4.18786496e-01 -9.55295265e-01 -4.44445610e-01
6.42502069e-01 4.18459505e-01 -5.58097720e-01 2.56024897e-02
-1.35736334e+00 1.63832307e+00 1.62778616e-01 1.44127458e-01
-7.73331761e-01 -6.82179987e-01 -8.89621079e-01 -1.65031314e-01
4.72529791e-02 -8.07566226e-01 1.48576403e+00 -9.61446762e-01
-1.08819067e+00 4.39234197e-01 5.14499396e-02 -6.60790086e-01
5.33186257e-01 -1.93720207e-01 -8.12609017e-01 -3.84097248e-01
5.69492504e-02 4.91518587e-01 3.29221725e-01 -1.30001116e+00
-6.54291570e-01 -4.41761076e-01 -2.88134098e-01 -1.51320651e-01
2.44711190e-01 1.64980888e-01 4.30433154e-01 2.61166334e-01
-1.83170050e-01 -1.09218001e+00 -6.77704513e-01 2.36773714e-01
-1.01424970e-01 1.14965446e-01 9.37788129e-01 -7.05713272e-01
1.01284230e+00 -1.95613694e+00 -4.32256520e-01 3.90599459e-01
5.35819411e-01 3.29697222e-01 3.11901063e-01 4.06318277e-01
2.43614197e-01 5.16416550e-01 7.29113892e-02 3.09607565e-01
3.41434985e-01 2.99013946e-02 -2.66221225e-01 2.75806278e-01
5.97957432e-01 1.12463605e+00 -8.14807236e-01 -3.45034033e-01
3.98795724e-01 2.59392589e-01 -1.71977863e-01 1.87012821e-01
-2.21017212e-01 5.28903961e-01 -8.16423893e-01 7.69641519e-01
2.64489144e-01 -2.85964608e-01 3.56568173e-02 1.16538405e-01
-1.14694163e-01 8.46306458e-02 -8.61259401e-01 1.12035215e+00
-3.22769970e-01 5.90478241e-01 1.61630571e-01 -3.02944392e-01
9.54005182e-01 5.75992763e-01 4.30136263e-01 -7.74023592e-01
3.68360668e-01 2.53467768e-01 4.73135144e-01 -4.14010078e-01
5.99937856e-01 -3.72968167e-01 -3.23096931e-01 6.93274498e-01
-4.33431029e-01 -7.57752419e-01 -3.11632186e-01 1.83935732e-01
9.23908591e-01 3.06550920e-01 4.84086066e-01 -3.37683707e-01
2.10016444e-02 6.51414275e-01 8.72202635e-01 1.32684737e-01
-6.98563814e-01 2.43761271e-01 1.73210040e-01 -8.66894841e-01
-1.10267758e+00 -5.79041600e-01 -2.00553671e-01 5.79220653e-01
-8.53369683e-02 -1.82579756e-01 -5.69842935e-01 -4.70679760e-01
4.38609302e-01 1.61803770e+00 -6.38473749e-01 -7.25631893e-01
1.34577304e-01 -7.90007040e-02 3.98298979e-01 6.10865116e-01
1.83707044e-01 -7.28025436e-01 -1.25599384e+00 3.49584073e-01
1.86657399e-01 -1.08931661e+00 1.85219258e-01 2.05427185e-01
-4.60680842e-01 -8.17486703e-01 7.85485953e-02 3.29643711e-02
5.01485765e-01 -1.27202317e-01 1.04464149e+00 1.74896400e-02
-2.55308479e-01 4.20554101e-01 -5.08016124e-02 -1.30541468e+00
-9.01434839e-01 -3.53957176e-01 2.22155437e-01 -4.80301529e-01
8.66551816e-01 -1.93509251e-01 -6.00271046e-01 6.54627621e-01
-6.22199535e-01 2.17070207e-01 6.76252186e-01 3.24524641e-01
2.78491944e-01 -1.65676877e-01 9.92982984e-01 -8.07243586e-01
6.65274501e-01 -3.71945322e-01 -5.63673735e-01 5.65833807e-01
-1.45693517e+00 -3.18865210e-01 5.14349163e-01 9.99097526e-02
-8.29821706e-01 1.95198372e-01 3.82287264e-01 -1.56829596e-01
-3.05428684e-01 5.33793390e-01 -2.03389376e-01 -7.39233866e-02
1.08498788e+00 -6.12291276e-01 5.47512293e-01 3.46399307e-01
5.14134049e-01 1.05590737e+00 2.28438720e-01 -4.11938697e-01
5.83071649e-01 1.39989063e-01 -5.64210154e-02 -4.10504997e-01
-1.39362127e-01 1.55018538e-01 -3.02686691e-01 -3.20562899e-01
8.15248013e-01 -9.13465142e-01 -1.03127325e+00 -3.37756485e-01
-8.78214598e-01 -1.81689292e-01 -6.57401204e-01 6.18311882e-01
-6.20359838e-01 -3.03685933e-01 4.83890660e-02 -1.18887341e+00
-5.37508667e-01 -1.23327565e+00 6.94652915e-01 3.38187724e-01
-1.27058160e+00 -7.35199034e-01 -1.32576779e-01 6.88865960e-01
7.44333088e-01 3.71053874e-01 7.11790860e-01 -1.03435612e+00
-6.41436934e-01 -1.10210037e+00 -3.17535661e-02 3.24522734e-01
7.66286626e-02 5.87790370e-01 -1.07566369e+00 2.59573698e-01
-1.94623917e-01 -4.62998331e-01 -1.72499716e-01 2.78575838e-01
5.64618886e-01 -2.78499693e-01 -4.05254751e-01 -2.30876014e-01
7.55098999e-01 7.34475553e-01 4.90561426e-01 1.16352290e-01
3.24714631e-01 1.02954185e+00 9.72407520e-01 2.86962092e-01
7.12764919e-01 6.82762787e-02 7.50559121e-02 -1.45449601e-02
4.26148027e-01 -1.39257222e-01 -8.29978436e-02 3.28726977e-01
1.61618777e-02 2.48890314e-02 -1.38542974e+00 4.38153207e-01
-2.31534505e+00 -7.64748096e-01 -6.47575259e-02 2.21793866e+00
4.89975721e-01 2.64689386e-01 -3.24781276e-02 -1.32725000e-01
2.47016996e-01 -5.87757230e-01 -1.00129151e+00 -9.19464648e-01
3.59883398e-01 -3.10932457e-01 4.84511614e-01 4.01791036e-01
-5.91196299e-01 8.12052011e-01 7.61269426e+00 6.98373478e-04
-8.34048212e-01 -1.07569613e-01 1.03896952e+00 -1.90402940e-01
-7.00993598e-01 2.05265269e-01 -3.03246439e-01 4.18982133e-02
1.55031466e+00 -9.54839528e-01 2.62850732e-01 1.06516552e+00
6.64965391e-01 -3.76498699e-01 -1.41557467e+00 3.68301183e-01
-3.75058472e-01 -1.20250177e+00 -4.25964385e-01 1.10439509e-01
6.07554853e-01 -2.27731958e-01 4.16618623e-02 8.49811174e-03
9.57525909e-01 -1.67603946e+00 7.79676378e-01 8.22699904e-01
4.84808803e-01 -7.96802700e-01 9.24854279e-01 2.03364551e-01
-5.01632273e-01 -1.63128257e-01 -1.71673566e-01 -3.15039486e-01
1.31922916e-01 6.35011435e-01 -9.18119907e-01 4.86918598e-01
4.81523067e-01 4.90734577e-01 -8.80120471e-02 9.00099754e-01
1.75350845e-01 4.91379976e-01 -1.64969847e-01 -1.92738548e-01
-1.20384678e-01 1.18007837e-02 2.30857536e-01 7.24552035e-01
-1.95089489e-01 3.39139909e-01 -3.50153834e-01 1.04561138e+00
5.05553424e-01 -2.42742941e-01 -1.06815970e+00 -9.20989931e-01
5.60423672e-01 1.20469427e+00 -4.67349946e-01 2.35997513e-02
-1.72735825e-01 2.69149899e-01 -3.34521532e-01 2.15868816e-01
-5.05998194e-01 -1.94936693e-01 9.13776219e-01 2.31232345e-01
-5.14299273e-01 -2.30710030e-01 -1.09482300e+00 -4.83490080e-01
-7.34798610e-02 -1.39246452e+00 1.18690819e-01 -8.85794044e-01
-1.21265614e+00 7.63382018e-01 -7.23546743e-02 -1.15726697e+00
-7.81079769e-01 -3.37872833e-01 -2.13004380e-01 1.00049400e+00
-9.52532053e-01 -1.63255918e+00 -2.47056752e-01 -7.85219483e-03
-1.99405342e-01 -2.06689388e-01 1.11060929e+00 -2.57493287e-01
-3.74498457e-01 3.13043535e-01 -2.27073446e-01 -3.83591413e-01
6.71436012e-01 -6.91485763e-01 7.60278404e-01 7.00142205e-01
-2.41070494e-01 1.02819490e+00 8.86184752e-01 -9.67534125e-01
-1.80098379e+00 -1.04328012e+00 8.32882047e-01 -1.05950010e+00
6.87679648e-01 -1.26511633e-01 -6.24143958e-01 9.83605981e-01
8.08443055e-02 -1.54677942e-01 1.00307119e+00 1.53973967e-01
-1.27620563e-01 -3.98028612e-01 -1.48349297e+00 6.14599705e-01
3.90632987e-01 -3.97048950e-01 -5.20341635e-01 3.03098977e-01
8.35494757e-01 -1.89161658e-01 -1.09717488e+00 2.96067953e-01
7.93347299e-01 -5.69212735e-01 3.29022586e-01 -1.03369737e+00
3.02399397e-01 -5.61566472e-01 -7.72375017e-02 -9.65695798e-01
-5.28202713e-01 -8.43202949e-01 1.55086279e-01 8.28697681e-01
8.16374123e-01 -1.01535523e+00 5.73173106e-01 2.38960576e+00
-2.34716386e-02 -6.11691535e-01 -6.93172932e-01 -5.20775855e-01
1.19847313e-01 -7.40588963e-01 8.96740496e-01 8.57888877e-01
4.84067827e-01 9.23719779e-02 -2.40110323e-01 1.37832627e-01
5.78327835e-01 -1.06857277e-01 9.57284033e-01 -1.39227366e+00
4.45189208e-01 -2.03277960e-01 -9.49049950e-01 1.56081408e-01
-2.05646917e-01 -2.49667034e-01 2.34171391e-01 -1.87572849e+00
8.03664327e-02 -5.71913242e-01 -1.54195771e-01 6.10241950e-01
-1.06118761e-01 8.45999923e-03 1.80209935e-01 1.70355633e-01
-5.51631868e-01 2.27039993e-01 1.05942392e+00 8.52099285e-02
1.51501698e-02 -3.86255533e-02 -1.51285553e+00 6.11690342e-01
5.56486607e-01 -3.91775012e-01 -5.34141421e-01 -7.08257630e-02
4.23403889e-01 -6.88162372e-02 5.44998407e-01 -1.03321481e+00
5.09295583e-01 -9.19680178e-01 4.66716260e-01 -7.05509409e-02
2.11284503e-01 -1.52299047e+00 1.03074765e+00 5.59152365e-01
-3.86182219e-01 -1.53380651e-02 3.19087088e-01 2.27561489e-01
6.17469475e-03 1.12859227e-01 5.69921136e-01 1.43494800e-01
-4.61102784e-01 1.25148073e-01 -5.04644871e-01 -4.25508708e-01
1.85782635e+00 -3.68058443e-01 -3.65361005e-01 -8.35721076e-01
-3.53769869e-01 7.35999346e-01 8.74092758e-01 4.78005916e-01
5.02239168e-01 -9.46423471e-01 -7.20172048e-01 1.95620492e-01
3.68852109e-01 2.15291813e-01 1.21637411e-01 6.20071054e-01
-5.88969469e-01 7.05179930e-01 -2.62535304e-01 -2.43864253e-01
-7.32860625e-01 3.87896866e-01 2.95049280e-01 4.65489477e-01
-3.58599901e-01 5.87022960e-01 -1.07236870e-01 -4.10850912e-01
2.79541641e-01 -9.48191285e-02 3.05029869e-01 -6.06846392e-01
6.73203170e-01 2.06015751e-01 2.56758556e-03 -4.79954809e-01
-7.09886670e-01 -8.67854059e-02 -1.33503422e-01 -2.74166703e-01
1.26914561e+00 2.78500300e-02 2.35971659e-02 8.30689549e-01
3.62504125e-01 -3.69397104e-01 -1.14574838e+00 4.19795662e-01
9.04778466e-02 -5.43990314e-01 3.52271914e-01 -1.42069852e+00
-7.05622792e-01 7.49944627e-01 5.14527857e-01 2.11171404e-01
7.27952719e-01 -3.57563913e-01 5.86320400e-01 4.08641487e-01
6.50054157e-01 -1.27145433e+00 -5.70271015e-01 -2.67775387e-01
1.10947561e+00 -1.28324354e+00 5.20982325e-01 -5.30405752e-02
-1.51196694e+00 8.64491105e-01 7.03613937e-01 4.70199972e-01
6.59701645e-01 4.36213762e-01 5.40691137e-01 -1.68312475e-01
-1.00236452e+00 3.14501166e-01 3.46411854e-01 7.30948806e-01
6.32537782e-01 3.76694649e-01 -1.95934117e-01 9.37973976e-01
-1.42163128e-01 6.19440436e-01 4.02417839e-01 1.11851466e+00
-2.94888616e-01 -9.44198668e-01 -3.01478893e-01 5.71436703e-01
-1.65005296e-01 2.54670084e-01 -8.14977944e-01 7.99290836e-01
-1.40380576e-01 1.48973513e+00 -8.02351013e-02 -6.80089414e-01
4.52043802e-01 7.50973970e-02 -1.49533823e-01 -3.73077095e-01
-8.83668602e-01 -4.11662996e-01 6.54312491e-01 -7.37644255e-01
-5.61436266e-03 -6.46358848e-01 -1.49966979e+00 -8.24707627e-01
-3.03945690e-01 4.02654618e-01 1.26709950e+00 7.50773787e-01
9.86460447e-01 3.51165801e-01 4.82070923e-01 -1.52353168e-01
-6.13010943e-01 -6.15443528e-01 -5.11636913e-01 8.26599300e-02
6.59114793e-02 -2.96994388e-01 -7.24838376e-02 -3.37289013e-02] | [8.938403129577637, 6.203271389007568] |
7d02a42e-ff90-4026-b99d-851b0c00aeb7 | mitodet-simple-and-robust-mitosis-detection | 2109.01485 | null | https://arxiv.org/abs/2109.01485v2 | https://arxiv.org/pdf/2109.01485v2.pdf | MitoDet: Simple and robust mitosis detection | Mitotic figure detection is a challenging task in digital pathology that has a direct impact on therapeutic decisions. While automated methods often achieve acceptable results under laboratory conditions, they frequently fail in the clinical deployment phase. This problem can be mainly attributed to a phenomenon called domain shift. An important source of a domain shift is introduced by different microscopes and their camera systems, which noticeably change the color representation of digitized images. In this method description we present our submitted algorithm for the Mitosis Domain Generalization Challenge, which employs a RetinaNet trained with strong data augmentation and achieves an F1 score of 0.7138 on the preliminary test set. | ['Thomas Wittenberg', 'Petr Kuritcyn', 'Volker Bruns', 'Michaela Benz', 'Jakob Dexl'] | 2021-09-02 | null | null | null | null | ['mitosis-detection'] | ['medical'] | [ 5.07298589e-01 -4.83636335e-02 -2.81446636e-01 -1.51155740e-01
-6.34411395e-01 -6.13367140e-01 5.36265075e-01 5.01899838e-01
-6.48491502e-01 9.54664588e-01 -4.78028432e-02 -1.97518125e-01
3.93338889e-01 -3.82907301e-01 -5.39998889e-01 -9.09424782e-01
3.67720813e-01 4.99934971e-01 4.65033352e-01 -5.75626045e-02
2.06114948e-01 5.64183712e-01 -1.09319007e+00 4.09648359e-01
7.23694265e-01 6.64223611e-01 -4.77398522e-02 9.55547988e-01
-1.27416208e-01 4.69158083e-01 -9.95787144e-01 -4.79634106e-01
8.09043646e-02 -4.31898206e-01 -6.94555640e-01 2.73237318e-01
3.93090516e-01 -1.97098538e-01 -2.83275723e-01 1.31080914e+00
6.28036499e-01 -4.18851435e-01 8.70083153e-01 -1.07684064e+00
-3.96781325e-01 1.50919959e-01 -7.95450807e-01 7.40686476e-01
-8.51479266e-03 2.20304161e-01 4.09023643e-01 -5.53594351e-01
1.04880190e+00 9.06186521e-01 5.57762504e-01 7.57902384e-01
-1.57769108e+00 -5.28245091e-01 -3.64154339e-01 1.99522182e-01
-1.46256459e+00 -2.78867930e-01 4.97449636e-01 -5.30874014e-01
5.96205115e-01 -7.14285076e-02 6.87487364e-01 1.02047515e+00
5.02025068e-01 6.79811776e-01 1.30628514e+00 -2.32023999e-01
3.72767627e-01 1.53071001e-01 9.94530767e-02 3.83958131e-01
5.40977061e-01 -2.34457344e-01 -3.58769685e-01 1.48283206e-02
8.86670172e-01 3.61528322e-02 -6.24406576e-01 -1.69327602e-01
-1.06155384e+00 4.95321661e-01 5.56788862e-01 5.06516874e-01
-1.01325579e-01 -9.98476446e-02 4.51594472e-01 2.84174770e-01
2.93360561e-01 5.06059170e-01 -1.57207340e-01 -3.81587483e-02
-8.00059855e-01 -9.92304981e-02 3.70738506e-01 6.32967412e-01
3.47651452e-01 -2.39281505e-01 -9.13682356e-02 5.40760636e-01
-4.87112999e-02 2.87566662e-01 8.14995408e-01 -5.01816690e-01
-1.15251943e-01 9.65217113e-01 -1.43783301e-01 -7.37643540e-01
-7.40345895e-01 -6.68361723e-01 -1.03860235e+00 4.35584962e-01
1.07787037e+00 2.00813115e-01 -1.34331548e+00 1.36546957e+00
2.86589175e-01 2.07960278e-01 1.76981464e-01 9.23565626e-01
1.13228130e+00 7.54785240e-02 3.45987886e-01 -1.37313977e-01
1.41341102e+00 -7.49150097e-01 -8.89403343e-01 -1.80889204e-01
9.24504995e-01 -7.44888186e-01 1.02022445e+00 5.08827507e-01
-7.12073982e-01 -1.25704065e-01 -1.42104006e+00 -2.55985767e-01
-6.26240313e-01 1.75308883e-01 3.39548081e-01 7.64159441e-01
-1.09882319e+00 4.19903308e-01 -7.59788096e-01 -8.87713134e-01
7.98544645e-01 5.45255244e-01 -6.09514952e-01 -1.28325909e-01
-4.06678528e-01 8.17188919e-01 5.02969027e-01 -8.00933689e-02
-5.82915664e-01 -9.42155480e-01 -3.12071830e-01 -2.07198307e-01
6.96073249e-02 -7.35860169e-01 1.28003573e+00 -7.80882597e-01
-1.26066279e+00 1.45215034e+00 7.06978813e-02 -5.98499238e-01
7.34751761e-01 2.75051981e-01 -3.34650189e-01 2.48143390e-01
-1.42725334e-01 7.81160414e-01 5.70210040e-01 -8.71508539e-01
-7.28763223e-01 -5.13000727e-01 -2.26475179e-01 6.08991347e-02
-2.00329512e-01 -2.44078800e-01 -5.59954941e-01 -5.76167941e-01
4.47294451e-02 -9.34824646e-01 -2.44061634e-01 3.01215023e-01
-2.91913241e-01 3.68764959e-02 7.51203954e-01 -3.89820874e-01
7.98472762e-01 -2.23830748e+00 -1.83138743e-01 -1.27742618e-01
4.93664980e-01 5.22068501e-01 3.34151164e-02 -1.74390793e-01
-1.81156516e-01 2.13738769e-01 -1.42627418e-01 -1.30206645e-01
-4.79440719e-01 -1.75079942e-01 -3.50176878e-02 7.67301619e-01
2.47597188e-01 8.18478644e-01 -8.43008637e-01 -5.11749268e-01
-2.41580233e-02 4.53699559e-01 -3.09698135e-01 -8.66910629e-03
-1.45301566e-01 5.15628278e-01 -9.03097354e-03 8.42870414e-01
8.87407064e-01 -6.19819641e-01 2.69497216e-01 -3.06063294e-01
2.08759069e-01 -2.19029158e-01 -6.60933733e-01 1.78767538e+00
2.91049838e-01 9.69328761e-01 3.00534144e-02 -6.99568033e-01
7.47213960e-01 1.88096806e-01 3.43850762e-01 -5.47053039e-01
4.68839675e-01 3.59955817e-01 3.65090311e-01 -1.90306395e-01
4.67378408e-01 -3.55880201e-01 2.99920291e-01 1.15382843e-01
-7.20944628e-02 -1.74390465e-01 2.77458310e-01 2.04960287e-01
1.41643310e+00 -2.89898694e-01 3.98292184e-01 -3.67425144e-01
3.48857105e-01 2.92519957e-01 6.74126148e-01 3.06918442e-01
-6.87954009e-01 8.99920464e-01 8.48047793e-01 -6.02344215e-01
-1.09138429e+00 -8.72791469e-01 -3.20262849e-01 4.54080433e-01
1.20242223e-01 -6.86678737e-02 -6.33148551e-01 -8.65013719e-01
-1.10077448e-01 -9.34396125e-03 -9.70702112e-01 -4.34015363e-01
-2.33106732e-01 -1.08274162e+00 6.72619522e-01 6.26415074e-01
5.68300784e-01 -6.27076089e-01 -6.50015891e-01 1.06075436e-01
7.05350935e-02 -1.23498666e+00 -2.71823823e-01 3.52193177e-01
-8.41300964e-01 -1.35881984e+00 -1.19053531e+00 -9.52276707e-01
9.56782758e-01 2.74652928e-01 1.08482337e+00 2.21492752e-01
-7.05173433e-01 -9.35196429e-02 -2.09670082e-01 -7.31076181e-01
-4.76594299e-01 1.53852776e-01 -1.52885303e-01 -1.01577871e-01
6.79904282e-01 -8.84855092e-02 -8.22368383e-01 2.21883923e-01
-1.13482749e+00 -5.94084933e-02 7.67878115e-01 1.02216077e+00
8.85136127e-01 -3.91476601e-03 4.72512752e-01 -1.12116492e+00
4.33343798e-01 -9.80592966e-02 -5.41275322e-01 5.20617627e-02
-4.76102620e-01 -1.46615595e-01 2.71175683e-01 -5.47596037e-01
-7.50667751e-01 2.65063882e-01 2.03927293e-01 -3.71751755e-01
-2.74042636e-01 2.41846412e-01 -7.41036758e-02 -2.88731992e-01
9.62191164e-01 -5.31755574e-02 2.46230692e-01 -1.73798248e-01
-1.34293512e-01 5.87233305e-01 8.71927083e-01 1.39551342e-01
7.04388440e-01 7.91006565e-01 3.04378420e-01 -1.05546212e+00
-6.07545733e-01 -5.63904762e-01 -4.44629669e-01 -2.84189612e-01
8.50707352e-01 -8.11844230e-01 -4.37089115e-01 6.56904340e-01
-9.84225392e-01 -5.18455923e-01 -3.61143984e-03 3.21706325e-01
-1.93069369e-01 3.03660750e-01 -5.61445594e-01 -2.47375697e-01
-3.34300339e-01 -1.06943798e+00 8.77202690e-01 7.58491278e-01
-2.38757223e-01 -1.01246083e+00 3.84468101e-02 3.03005338e-01
3.26713383e-01 5.50546527e-01 1.08522105e+00 -6.41382575e-01
-4.09130812e-01 -2.40980595e-01 -4.55477625e-01 5.17657883e-02
1.64564356e-01 1.50236860e-01 -1.11572444e+00 -4.94557470e-01
-4.26998615e-01 -2.63883710e-01 9.44539785e-01 5.77540934e-01
1.05201888e+00 5.03292143e-01 -7.63380468e-01 6.50460243e-01
1.44721305e+00 2.95734406e-01 9.60161150e-01 5.33894479e-01
2.21126616e-01 3.54582220e-01 5.24682939e-01 7.27642104e-02
-1.62144899e-01 5.28738797e-01 4.47018534e-01 -6.77771926e-01
-5.48479259e-01 8.35751146e-02 -2.65607744e-01 6.56784400e-02
1.27263039e-01 -2.37525180e-01 -1.14611089e+00 6.62376642e-01
-1.55306339e+00 -4.51762557e-01 -2.70180225e-01 2.21408820e+00
8.31210136e-01 2.89395839e-01 1.90492168e-01 2.27578864e-01
9.46196675e-01 -5.38666725e-01 -6.30605996e-01 -1.38670802e-01
-3.66464764e-01 8.29538032e-02 6.00476801e-01 5.34006618e-02
-1.04863036e+00 6.90917015e-01 7.08585072e+00 6.25951290e-01
-1.42457044e+00 -1.24661237e-01 9.37793314e-01 -1.36529088e-01
1.90871224e-01 -4.45347250e-01 -5.16044080e-01 4.61677819e-01
6.59175456e-01 -3.69659275e-01 -1.59518942e-01 4.30768847e-01
-3.42313270e-03 -3.21846098e-01 -1.16922224e+00 1.19145274e+00
3.82345505e-02 -1.42561626e+00 -9.82874036e-02 4.86066848e-01
7.20926106e-01 -2.19758704e-01 5.02967060e-01 1.10430688e-01
-6.95758685e-02 -1.12606132e+00 -4.47772518e-02 2.49641970e-01
1.23047662e+00 -6.87006772e-01 1.09657776e+00 -1.15374021e-01
-6.91772044e-01 1.71111882e-01 -4.45025176e-01 2.22823605e-01
-2.65608251e-01 6.89555585e-01 -1.58390558e+00 -7.87032321e-02
6.55935407e-01 5.13050258e-01 -8.28518450e-01 1.69735360e+00
-6.12231009e-02 4.40473050e-01 -2.34494179e-01 -7.33258352e-02
-7.91216046e-02 1.60273015e-01 4.56020772e-01 1.17623639e+00
2.54386961e-01 -3.55009474e-02 -2.15010405e-01 5.85152984e-01
-3.58702958e-01 -1.05940634e-02 -5.41220844e-01 -2.37594083e-01
2.17681453e-01 1.38369215e+00 -1.41106641e+00 -2.23280489e-01
-2.58470118e-01 8.41690183e-01 7.87464157e-02 3.49250525e-01
-7.06060648e-01 -4.75529224e-01 5.06622672e-01 4.24337745e-01
1.46276772e-01 2.12400451e-01 -3.75321269e-01 -8.42305124e-01
-2.61692822e-01 -8.92395616e-01 5.33769667e-01 -5.19951224e-01
-1.08296084e+00 4.62530077e-01 -6.61739647e-01 -1.33473992e+00
2.58087039e-01 -9.33516502e-01 -4.61134762e-01 5.90802491e-01
-1.47720683e+00 -9.43472147e-01 -6.46595597e-01 5.11276364e-01
3.61479968e-01 -2.60391295e-01 8.20767462e-01 4.18650120e-01
-5.95476329e-01 8.22544575e-01 2.49425039e-01 5.38571738e-02
1.35785353e+00 -1.53961539e+00 8.60051811e-02 7.48874605e-01
-2.32903242e-01 2.74959803e-01 9.37531948e-01 -4.25455242e-01
-9.61845756e-01 -1.14975464e+00 6.07390761e-01 -3.74279171e-01
5.55944681e-01 -3.78960446e-02 -9.33624923e-01 3.58816832e-01
1.07358143e-01 2.01858819e-01 1.01540291e+00 -7.61235431e-02
-2.73167789e-01 -2.79780570e-02 -1.51696062e+00 6.42549813e-01
4.94867563e-01 -2.65083581e-01 -1.52132303e-01 4.00792301e-01
2.47814581e-01 -6.11042917e-01 -7.18940377e-01 3.72271925e-01
3.60488534e-01 -8.38513553e-01 4.53531593e-01 -5.05732119e-01
4.35178310e-01 -5.41256905e-01 3.42204154e-01 -1.31685150e+00
-3.67710114e-01 -3.34812164e-01 1.17051974e-01 7.93320000e-01
4.71144289e-01 -4.60570931e-01 1.30441964e+00 2.73701340e-01
-1.12214601e-02 -6.38786733e-01 -1.14654279e+00 -6.37366772e-01
3.49993467e-01 1.07664138e-01 3.56026709e-01 7.66600132e-01
2.98522234e-01 4.15275127e-01 1.57021478e-01 -1.34479310e-02
5.35673738e-01 -2.06108123e-01 8.01095188e-01 -1.33503246e+00
-7.18823224e-02 -6.84416115e-01 -1.21278989e+00 -6.64851844e-01
-2.62052298e-01 -7.95697391e-01 -9.75421593e-02 -1.22504056e+00
4.89456177e-01 4.38247472e-02 -4.50344622e-01 2.07645699e-01
-6.66229278e-02 8.74119997e-01 -1.34776309e-01 1.40404508e-01
-6.59727573e-01 -8.08456466e-02 1.33810663e+00 -4.77102727e-01
-2.06729978e-01 -2.28346422e-01 -6.99061811e-01 6.02837324e-01
8.95552278e-01 -3.96676838e-01 -1.65208712e-01 -2.34643534e-01
1.61100388e-01 -3.76281142e-01 2.76999205e-01 -1.33612669e+00
3.63057017e-01 2.19305500e-01 6.81444228e-01 -5.52426398e-01
1.63405657e-01 -6.56058133e-01 1.25803232e-01 8.27318609e-01
-6.62826076e-02 -1.43321723e-01 5.87493360e-01 8.32524180e-01
-9.01851654e-02 4.30384129e-02 1.16640437e+00 2.06267372e-01
-4.78330046e-01 8.10862631e-02 -6.56619370e-01 -2.15680227e-01
1.19753146e+00 -5.58475554e-01 -8.82918596e-01 -1.70316368e-01
-8.10238183e-01 1.31212085e-01 9.85875547e-01 1.65398195e-01
3.26721132e-01 -1.17545843e+00 -6.37200892e-01 1.20367467e-01
3.51467580e-01 2.59541702e-02 3.16223443e-01 1.21006787e+00
-7.76225269e-01 3.33847523e-01 -4.14945990e-01 -9.35672164e-01
-1.58672881e+00 2.77496785e-01 5.46185672e-01 -4.54883099e-01
-5.33222735e-01 9.08344269e-01 3.14674169e-01 2.01034129e-01
2.31854424e-01 -4.36440110e-01 -2.79441297e-01 -3.21475901e-02
7.24162281e-01 2.97820002e-01 4.07774359e-01 -3.62497836e-01
-4.58261877e-01 3.59533340e-01 -6.50276303e-01 2.18461439e-01
8.97535205e-01 1.98458388e-01 1.59653202e-01 1.72954544e-01
8.63120437e-01 -4.09475446e-01 -1.11369038e+00 -1.51835948e-01
-8.70688632e-02 -4.07593310e-01 1.28918499e-01 -1.22529411e+00
-1.02193582e+00 8.89910758e-01 1.13267267e+00 1.71878815e-01
1.19306505e+00 2.32199747e-02 6.00773513e-01 3.68334144e-01
1.34950459e-01 -1.19639027e+00 1.79481477e-01 3.44204694e-01
4.65986192e-01 -1.42681909e+00 1.10373152e-02 -5.12602210e-01
-4.55152541e-01 1.13788903e+00 6.83665395e-01 1.19052537e-01
2.59740651e-01 4.61226255e-01 4.76786315e-01 -1.90403014e-01
-6.96756005e-01 -7.61917531e-02 -3.53218019e-02 9.85388458e-01
6.83435142e-01 -2.23637044e-01 -4.78192955e-01 5.76926470e-01
4.73251343e-02 9.82856899e-02 9.66449320e-01 9.53802586e-01
-4.29834872e-01 -8.92029166e-01 -4.73194301e-01 4.43882704e-01
-7.59139657e-01 4.55827951e-01 -8.00068676e-01 8.36392224e-01
8.04476067e-02 6.69057548e-01 1.92737758e-01 -1.33229420e-01
3.28463852e-01 -3.07951774e-02 5.51347494e-01 -5.71579695e-01
-4.16722745e-01 2.46125206e-01 -2.77346790e-01 -2.94691533e-01
-4.06628340e-01 -5.08980930e-01 -1.43384922e+00 -3.60361367e-01
-2.01099813e-01 -9.75195915e-02 4.64266509e-01 5.91559172e-01
3.69353712e-01 8.72097671e-01 -1.96391102e-02 -4.69207585e-01
-1.21342853e-01 -9.58849311e-01 -9.54050481e-01 4.65190530e-01
4.48639899e-01 -6.51619792e-01 -3.73947501e-01 4.74385083e-01] | [15.115635871887207, -3.146883726119995] |
15f6e321-e824-4485-8e48-2df3b8178e30 | robust-and-explainable-identification-of | 2212.07425 | null | https://arxiv.org/abs/2212.07425v2 | https://arxiv.org/pdf/2212.07425v2.pdf | Robust and Explainable Identification of Logical Fallacies in Natural Language Arguments | The spread of misinformation, propaganda, and flawed argumentation has been amplified in the Internet era. Given the volume of data and the subtlety of identifying violations of argumentation norms, supporting information analytics tasks, like content moderation, with trustworthy methods that can identify logical fallacies is essential. In this paper, we formalize prior theoretical work on logical fallacies into a comprehensive three-stage evaluation framework of detection, coarse-grained, and fine-grained classification. We adapt existing evaluation datasets for each stage of the evaluation. We employ three families of robust and explainable methods based on prototype reasoning, instance-based reasoning, and knowledge injection. The methods combine language models with background knowledge and explainable mechanisms. Moreover, we address data sparsity with strategies for data augmentation and curriculum learning. Our three-stage framework natively consolidates prior datasets and methods from existing tasks, like propaganda detection, serving as an overarching evaluation testbed. We extensively evaluate these methods on our datasets, focusing on their robustness and explainability. Our results provide insight into the strengths and weaknesses of the methods on different components and fallacy classes, indicating that fallacy identification is a challenging task that may require specialized forms of reasoning to capture various classes. We share our open-source code and data on GitHub to support further work on logical fallacy identification. | ['Alain Mermoud', 'Hông-Ân Sandlin', 'Filip Ilievski', 'Himanshu Rawlani', 'Darshan Deshpande', 'Vishnu Priya Prasanna Venkatesh', 'Zhivar Sourati'] | 2022-12-12 | null | null | null | null | ['logical-fallacies', 'propaganda-detection'] | ['miscellaneous', 'natural-language-processing'] | [ 3.55468750e-01 5.20657539e-01 -8.77504289e-01 -2.52030849e-01
-7.75260866e-01 -6.98334932e-01 1.01494873e+00 7.04392374e-01
1.36126637e-01 5.98682582e-01 7.63085723e-01 -8.85493517e-01
-5.02046943e-01 -6.28366411e-01 -7.96623826e-01 1.32421359e-01
1.87603682e-01 3.22570294e-01 2.20498294e-02 -3.94699991e-01
7.70937443e-01 -1.05668955e-01 -1.58120298e+00 8.73784363e-01
1.33495343e+00 7.40624666e-01 -7.89258480e-01 5.12643218e-01
-2.87333339e-01 1.70551932e+00 -6.36272252e-01 -9.22304988e-01
-5.45223653e-02 -4.17124480e-01 -1.21467745e+00 -2.58483320e-01
8.04438412e-01 -3.62171352e-01 1.05319284e-02 1.05783129e+00
1.07471049e-02 -4.00396764e-01 6.83052957e-01 -1.43066335e+00
-8.44928801e-01 1.52623272e+00 -5.18115044e-01 1.97904557e-01
5.62476695e-01 -5.32136112e-02 1.31277955e+00 -5.82287192e-01
7.45506048e-01 1.41721809e+00 1.02952731e+00 3.97310078e-01
-1.06013906e+00 -7.01592505e-01 4.56456393e-01 6.07120395e-01
-7.69684732e-01 -4.36111987e-01 8.57179523e-01 -7.98038125e-01
6.39123678e-01 4.19156402e-01 6.20425820e-01 1.48596060e+00
-1.77591100e-01 9.90994871e-01 1.44970942e+00 -5.48978031e-01
1.24560401e-01 2.35446870e-01 7.50193298e-01 9.12963748e-01
8.91508579e-01 -1.57356516e-01 -8.06474507e-01 -4.76887256e-01
2.64175981e-01 -1.50022522e-01 -1.63349643e-01 -1.72877848e-01
-9.58021939e-01 1.16135335e+00 2.03969344e-01 1.08591929e-01
-5.51886782e-02 6.66642413e-02 3.51396650e-01 4.23961073e-01
4.52571690e-01 7.35164702e-01 -4.68054950e-01 -4.01926845e-01
-1.08363783e+00 6.41371787e-01 1.38126266e+00 6.96195304e-01
3.67257148e-01 -2.21846372e-01 -7.61587024e-02 6.17838085e-01
3.42243761e-01 3.37581843e-01 5.16297780e-02 -1.21860814e+00
7.15542614e-01 9.81930077e-01 4.25724238e-02 -1.40052879e+00
-4.23663974e-01 -3.96058470e-01 -4.45280939e-01 7.25853071e-02
6.85604393e-01 4.24733609e-02 -3.93847287e-01 1.59665179e+00
3.91099662e-01 -1.72258407e-01 -3.49564143e-02 7.84589350e-01
9.76543248e-01 1.72677979e-01 3.16063166e-02 6.35946766e-02
1.44750488e+00 -9.28488016e-01 -8.28380883e-01 -1.87371746e-01
8.92412961e-01 -8.10232282e-01 1.19153631e+00 7.31205940e-01
-1.23777342e+00 1.47751853e-01 -1.14367437e+00 -2.87133902e-01
-3.60107809e-01 -3.58970910e-01 8.14320505e-01 7.53434241e-01
-4.12816972e-01 5.59523344e-01 -6.58286929e-01 -3.76862660e-02
7.72318542e-01 -5.27259707e-01 6.70393035e-02 -6.94406480e-02
-1.44990587e+00 1.05939031e+00 1.50300115e-01 -2.62327105e-01
-6.10528231e-01 -9.92441833e-01 -7.73175061e-01 5.85180856e-02
8.67398679e-01 -5.83884418e-01 1.31647432e+00 -8.54891896e-01
-1.02073896e+00 7.85528958e-01 3.10536325e-01 -4.95627463e-01
7.31797457e-01 -4.24337715e-01 -2.39136234e-01 -1.09127657e-02
2.38890216e-01 -1.43996939e-01 5.94790578e-01 -1.11664021e+00
-3.76373678e-01 -2.45023444e-01 4.25612062e-01 -2.27883264e-01
-3.80002201e-01 2.33285517e-01 2.56386459e-01 -6.93548203e-01
1.58019885e-01 -4.69392747e-01 1.65599927e-01 -5.91920912e-02
-5.80754101e-01 -2.60073811e-01 6.27471745e-01 -8.23380888e-01
1.75187635e+00 -1.54420936e+00 -5.83659671e-02 3.07753980e-01
8.48207057e-01 7.47717172e-02 1.48077846e-01 5.25913954e-01
-2.16877423e-02 8.60734880e-01 2.44736671e-02 -1.38346013e-02
3.28208536e-01 -5.31133413e-02 -4.81280118e-01 2.38391131e-01
2.17840374e-01 8.07907104e-01 -1.00624728e+00 -5.50621033e-01
-2.84572035e-01 2.19645113e-01 -8.18792999e-01 -1.90089345e-01
-5.52759409e-01 1.06008053e-01 -6.06304765e-01 8.60746980e-01
3.89688730e-01 -7.38058984e-01 4.57554042e-01 -3.08576405e-01
-1.31800249e-01 9.81921315e-01 -1.21517277e+00 1.24115503e+00
-2.22294852e-01 5.93178689e-01 1.70816958e-01 -9.30078685e-01
5.26829243e-01 1.05693564e-01 2.27282062e-01 -6.94575310e-01
1.07098922e-01 3.97269577e-01 3.44088860e-02 -6.26165390e-01
3.66904378e-01 2.41375968e-01 -1.25080541e-01 1.10558724e+00
-3.68694574e-01 -6.19621575e-03 3.73085469e-01 7.78690636e-01
1.28031588e+00 -2.06757430e-02 4.40192640e-01 -2.67793626e-01
2.72072017e-01 2.97925770e-01 2.99975753e-01 1.21875072e+00
-8.44684541e-02 1.70982018e-01 1.04818940e+00 -7.14626491e-01
-1.00517499e+00 -6.78747654e-01 -3.72946709e-01 1.19831514e+00
-8.08290765e-03 -1.01212013e+00 -5.32782793e-01 -1.03883648e+00
2.89559722e-01 7.47083485e-01 -7.71087468e-01 7.86186457e-02
-4.93062437e-01 -5.22834480e-01 8.32118928e-01 3.38725388e-01
4.76361603e-01 -7.97903299e-01 -6.95102870e-01 9.23767984e-02
-5.83043456e-01 -9.02177274e-01 1.95599109e-01 3.28286700e-02
-5.44106424e-01 -1.82582867e+00 3.26670647e-01 -1.38448581e-01
3.27873051e-01 2.04889774e-01 1.44582045e+00 7.76913524e-01
-3.17284912e-02 5.75391710e-01 -5.29175878e-01 -4.67976779e-01
-8.25017631e-01 -7.82046691e-02 -1.65987611e-01 -4.96616811e-01
3.21993411e-01 -4.00766343e-01 -3.84571284e-01 -2.07255445e-02
-9.39103961e-01 2.46574894e-01 3.74604553e-01 8.65180731e-01
-2.07851939e-02 -7.55415931e-02 3.43607813e-01 -1.42404675e+00
1.07103622e+00 -8.70669007e-01 -3.06643605e-01 3.48864436e-01
-1.13648355e+00 3.16020697e-01 2.82974422e-01 -1.94712207e-01
-9.92464900e-01 -9.10794973e-01 7.50529543e-02 1.89123675e-01
2.06042364e-01 1.02481759e+00 2.13384211e-01 7.70284981e-02
1.17067230e+00 -4.11130637e-01 4.11787257e-02 -4.13512617e-01
7.98932135e-01 8.19930434e-01 2.97265470e-01 -1.11632884e+00
8.30247462e-01 3.53418201e-01 -4.48737264e-01 -3.16832840e-01
-1.57518971e+00 -9.35874972e-03 -1.47819310e-01 -7.86227062e-02
2.83936441e-01 -8.31125259e-01 -8.91302109e-01 8.18441808e-02
-1.37981796e+00 -3.08385313e-01 -1.62977651e-01 2.33232118e-02
-3.46979618e-01 6.45161271e-01 -7.95434773e-01 -7.71406770e-01
-3.60229999e-01 -8.51238191e-01 6.18746579e-01 -1.36318848e-01
-8.02493095e-01 -1.10985005e+00 -5.41906022e-02 1.12800789e+00
4.03336823e-01 4.39096034e-01 1.26625752e+00 -1.03367078e+00
-5.64806759e-01 -4.51029278e-02 -3.26914966e-01 1.53088108e-01
-2.50979245e-01 1.78367332e-01 -7.36582577e-01 2.00618133e-01
5.69973402e-02 -9.54456270e-01 9.85721767e-01 9.12186503e-02
1.04250741e+00 -1.08604252e+00 1.27924122e-02 2.15851143e-01
9.55946863e-01 -4.80051190e-01 2.22449750e-01 8.08465302e-01
5.45397639e-01 6.31929338e-01 3.43663275e-01 5.81256568e-01
8.88689578e-01 3.59830737e-01 3.90639186e-01 8.87423828e-02
-1.74415424e-01 -4.99612242e-01 1.67355061e-01 4.72370714e-01
-7.17402203e-03 -4.13923152e-03 -1.21736300e+00 1.96337044e-01
-2.20376396e+00 -1.45331383e+00 -3.69818211e-01 1.74355650e+00
1.35915148e+00 3.86734903e-01 9.59265605e-02 5.96379399e-01
4.65301245e-01 1.87956899e-01 -2.21008852e-01 -6.52188182e-01
-1.15363456e-01 1.87091287e-02 -3.41833867e-02 8.54711115e-01
-9.00366545e-01 7.27625728e-01 6.40898514e+00 4.35609192e-01
-7.65152514e-01 1.93083748e-01 5.72957456e-01 2.36536693e-02
-1.00720644e+00 2.01615542e-01 -6.98787749e-01 3.98677200e-01
6.98760986e-01 -9.95343104e-02 4.26709443e-01 7.24719286e-01
1.64503872e-01 1.87223982e-02 -1.28385937e+00 4.21173871e-01
2.14530662e-01 -1.93911517e+00 8.40748996e-02 -2.58761078e-01
9.76478994e-01 -2.35867098e-01 -1.66336913e-02 4.24031109e-01
9.59512472e-01 -1.08786118e+00 1.07979691e+00 3.04934382e-01
2.88708299e-01 -1.26467004e-01 5.39165974e-01 4.33081865e-01
-4.61442083e-01 -4.01007295e-01 2.67223448e-01 -7.10386157e-01
-2.08618939e-01 6.93735600e-01 -4.84035373e-01 3.77169520e-01
5.37347794e-01 1.02151620e+00 -6.18427217e-01 6.56403124e-01
-8.47284436e-01 1.03290904e+00 -2.60878950e-01 -1.62797257e-01
1.46053419e-01 2.69166201e-01 3.70906562e-01 1.19175100e+00
-2.93310732e-01 1.84873771e-02 3.82994503e-01 1.24004364e+00
-1.10697098e-01 -6.87471554e-02 -2.79968709e-01 -1.45810142e-01
8.59750569e-01 1.10737371e+00 -3.05782825e-01 -4.94169980e-01
-4.14238274e-01 2.74248305e-03 5.77309430e-01 2.38601685e-01
-8.04707706e-01 -2.62260642e-02 3.36116970e-01 2.72946030e-01
-8.28377977e-02 5.06298542e-02 -9.94432092e-01 -1.40088618e+00
1.52662143e-01 -1.56329608e+00 8.18500340e-01 -4.56463367e-01
-1.45973718e+00 9.51630846e-02 1.47352114e-01 -5.89499474e-01
-2.47941613e-01 -4.36612070e-01 -4.77661461e-01 4.10067201e-01
-1.64447856e+00 -1.32444715e+00 -5.20230055e-01 2.55194336e-01
3.53940010e-01 1.58459097e-01 5.59818327e-01 7.04052765e-03
-5.80285311e-01 4.88922238e-01 -2.46550843e-01 2.16232404e-01
6.68506324e-01 -1.11099279e+00 7.37820640e-02 6.95768237e-01
2.12381348e-01 9.77942407e-01 8.91537964e-01 -8.69030476e-01
-1.24966049e+00 -5.97808421e-01 1.03230178e+00 -1.07204211e+00
1.41116667e+00 -3.17070395e-01 -1.06412733e+00 8.31946492e-01
1.34461284e-01 -4.61563230e-01 7.33768404e-01 9.24815953e-01
-1.27286303e+00 2.78915465e-01 -9.24777329e-01 6.07395768e-01
1.15766978e+00 -4.21866894e-01 -8.64941537e-01 6.56678677e-01
4.05625373e-01 -4.52390403e-01 -8.01453948e-01 2.81717002e-01
7.69043386e-01 -1.03488529e+00 7.36362815e-01 -1.12489915e+00
1.38205457e+00 9.64778140e-02 7.35406578e-03 -8.42330515e-01
-2.69791394e-01 -7.01583266e-01 -6.03204131e-01 1.17180836e+00
5.13354123e-01 -3.26812804e-01 7.71722913e-01 8.03520322e-01
3.15127671e-02 -7.82449484e-01 -4.90693480e-01 -3.88810396e-01
3.98611486e-01 -7.08532810e-01 4.65397924e-01 1.40701997e+00
6.68658257e-01 4.93416160e-01 -3.15997124e-01 5.04682437e-02
8.52281153e-01 4.46271926e-01 7.91263938e-01 -1.49773800e+00
-2.18541682e-01 -8.73466611e-01 -7.67148212e-02 -8.07350874e-01
3.03851426e-01 -1.04175818e+00 -4.63218182e-01 -1.68387973e+00
5.58636665e-01 -4.33349609e-01 4.32215147e-02 7.92660058e-01
-2.85417318e-01 -1.23209998e-01 1.35711446e-01 3.03107709e-01
-9.12952423e-01 1.94349676e-01 8.25674474e-01 -3.91317338e-01
1.37731908e-02 -2.48620406e-01 -1.29994118e+00 1.06615698e+00
6.22962832e-01 -4.93766993e-01 -3.58617246e-01 -5.37396431e-01
9.22571838e-01 -1.22528933e-01 7.10757732e-01 -6.05912268e-01
2.93712556e-01 -4.45958287e-01 -1.81059595e-02 -2.79676586e-01
-2.26163313e-01 -5.18930316e-01 -1.94034114e-01 5.80156028e-01
-9.91007149e-01 -2.24138945e-01 -6.43425956e-02 4.99796927e-01
3.93855944e-02 -2.72711575e-01 5.48507333e-01 -1.18000858e-01
-3.04858953e-01 -3.04901659e-01 -1.86318383e-01 7.56479442e-01
4.76556212e-01 -1.65552616e-01 -1.21086693e+00 -5.53850710e-01
-5.34092546e-01 2.11635232e-01 2.25074902e-01 4.66032833e-01
4.16064262e-01 -1.08378005e+00 -1.05273008e+00 -1.12587154e-01
1.73930943e-01 -4.09368157e-01 -1.05096651e-02 1.02273428e+00
-3.45403433e-01 2.62587935e-01 -6.01313636e-02 -3.20878983e-01
-1.10271919e+00 3.30929577e-01 1.03105657e-01 -5.20096779e-01
-5.25338352e-01 5.48858821e-01 -4.46962178e-01 -3.55552644e-01
3.88593704e-01 -3.62351447e-01 -2.95270115e-01 5.15957586e-02
6.92620337e-01 6.22188509e-01 -1.76990088e-02 1.01064630e-02
-1.65960461e-01 9.49670970e-02 -3.95245820e-01 3.02505493e-01
1.35401249e+00 2.51315720e-02 -2.64957786e-01 2.30160996e-01
6.47814691e-01 2.81170845e-01 -8.06927741e-01 -4.01821136e-01
3.79965365e-01 -3.64612401e-01 4.71069217e-02 -1.52576959e+00
-5.75636864e-01 6.13971412e-01 -1.68487102e-01 5.46674967e-01
4.68096316e-01 1.73575044e-01 3.59572530e-01 3.18338633e-01
9.79942828e-02 -1.05269587e+00 3.66360277e-01 6.56555951e-01
9.73319590e-01 -1.34920073e+00 4.53075081e-01 -6.23175502e-01
-4.20503259e-01 1.12838483e+00 5.49364090e-01 3.82091910e-01
5.82974613e-01 4.30200279e-01 1.05284572e-01 -6.64567590e-01
-9.90005493e-01 2.20892176e-01 3.02283585e-01 1.99402824e-01
7.11206436e-01 -1.64060563e-01 -8.01410258e-01 1.08854568e+00
-4.27108884e-01 -3.96873057e-03 7.77748346e-01 1.07076991e+00
-4.30299848e-01 -8.36014211e-01 -4.44071561e-01 4.17333454e-01
-5.52611887e-01 -3.96401227e-01 -8.54324877e-01 8.17374408e-01
-2.07877122e-02 1.15330207e+00 -3.39804649e-01 -3.38090509e-01
2.51151770e-01 1.91431046e-02 3.77123058e-01 -4.05773997e-01
-8.60620320e-01 -4.83343422e-01 7.07743764e-01 -7.69747555e-01
-4.82745945e-01 -6.39105856e-01 -1.05368578e+00 -9.08794701e-01
-1.74026132e-01 1.22625075e-01 4.61350918e-01 1.19349861e+00
3.46712142e-01 1.84597075e-01 1.44631103e-01 -9.92042422e-02
-1.16216207e+00 -6.84922993e-01 -2.60017850e-02 7.64052153e-01
2.61154115e-01 -7.43984818e-01 -5.46905279e-01 3.13659199e-02] | [9.74504280090332, 8.136551856994629] |
dfe5ab80-6c1c-4bad-b704-fe4a3c0a350d | unitts-residual-learning-of-unified-embedding | 2106.11171 | null | https://arxiv.org/abs/2106.11171v3 | https://arxiv.org/pdf/2106.11171v3.pdf | UniTTS: Residual Learning of Unified Embedding Space for Speech Style Control | We propose a novel high-fidelity expressive speech synthesis model, UniTTS, that learns and controls overlapping style attributes avoiding interference. UniTTS represents multiple style attributes in a single unified embedding space by the residuals between the phoneme embeddings before and after applying the attributes. The proposed method is especially effective in controlling multiple attributes that are difficult to separate cleanly, such as speaker ID and emotion, because it minimizes redundancy when adding variance in speaker ID and emotion, and additionally, predicts duration, pitch, and energy based on the speaker ID and emotion. In experiments, the visualization results exhibit that the proposed methods learned multiple attributes harmoniously in a manner that can be easily separated again. As well, UniTTS synthesized high-fidelity speech signals controlling multiple style attributes. The synthesized speech samples are presented at https://anonymous-authors2022.github.io/paper_works/UniTTS/demos/. | ['Injung Kim', 'Sungjae Kim', 'Minsu Kang'] | 2021-06-21 | null | null | null | null | ['expressive-speech-synthesis'] | ['speech'] | [-2.46258482e-01 4.36395109e-02 1.16114490e-01 -3.45189333e-01
-7.13382602e-01 -6.69360042e-01 4.01467115e-01 -6.92947730e-02
9.91066173e-02 5.90530336e-01 6.50418878e-01 1.48242623e-01
-6.35574162e-02 -5.64590693e-01 -3.73252571e-01 -8.13351691e-01
-1.82917833e-01 1.51433483e-01 -4.34815437e-01 -2.34428197e-01
-6.93514347e-02 3.65620553e-01 -1.65324795e+00 1.33657068e-01
1.01419866e+00 9.55676496e-01 1.71459511e-01 6.63485706e-01
-1.92157909e-01 5.29485881e-01 -1.08233440e+00 -2.21239105e-01
7.73355216e-02 -6.30162120e-01 -2.16191530e-01 -2.36931458e-01
1.79831102e-01 1.22480415e-01 -5.51116109e-01 1.01797450e+00
7.80118048e-01 3.85232836e-01 5.76855421e-01 -1.47360003e+00
-1.03819883e+00 8.17873776e-01 -1.52556837e-01 -1.96987107e-01
4.34911251e-01 9.40938145e-02 9.70153630e-01 -1.01068985e+00
3.02043825e-01 1.56490099e+00 3.96662444e-01 7.06939101e-01
-1.23545766e+00 -1.12839520e+00 2.44011357e-01 2.20779955e-01
-1.48422110e+00 -8.11571121e-01 9.48245168e-01 -2.67851800e-01
7.08435893e-01 7.25334883e-01 6.46066666e-01 1.54746842e+00
-5.19500077e-02 7.45007515e-01 9.61151183e-01 -2.19040990e-01
1.41585141e-01 4.76389229e-01 -2.11410403e-01 4.05127317e-01
-3.55297863e-01 1.93909019e-01 -5.56420624e-01 -2.60906011e-01
6.21760964e-01 -2.53475368e-01 -5.32944143e-01 1.05211124e-01
-1.60574031e+00 7.57138550e-01 3.05944718e-02 2.86005497e-01
-1.86426803e-01 -3.88936475e-02 4.04430360e-01 4.22787577e-01
4.59205538e-01 4.43405360e-01 -3.61895770e-01 -3.38602334e-01
-7.09721208e-01 4.36138995e-02 6.59854531e-01 1.44372225e+00
5.43033421e-01 7.27469981e-01 -3.32549036e-01 1.20692480e+00
1.10739104e-01 8.75878096e-01 5.12641966e-01 -1.04270029e+00
1.68326885e-01 2.08445728e-01 -6.99613392e-02 -9.43575203e-01
-1.32070154e-01 -3.11875820e-01 -1.16805542e+00 1.10667892e-01
-1.25716537e-01 -3.71772408e-01 -6.88100755e-01 2.19529653e+00
1.18391037e-01 3.24705243e-01 1.96738422e-01 8.22385013e-01
1.16174614e+00 1.11646819e+00 7.96377808e-02 -4.16719615e-01
1.24907839e+00 -1.05503833e+00 -1.42150974e+00 -1.03151398e-02
5.89037314e-02 -1.04199076e+00 1.43192625e+00 3.77041757e-01
-1.21851575e+00 -7.49097526e-01 -1.24406302e+00 -6.46140650e-02
-3.68199229e-01 4.55995500e-01 2.34472916e-01 4.41038877e-01
-9.52010632e-01 6.59946740e-01 -4.94422823e-01 -3.45915072e-02
-8.50998610e-02 2.31753722e-01 -2.99085826e-01 7.54226506e-01
-1.64891827e+00 7.33669758e-01 6.57105260e-03 -1.17068216e-01
-7.15091169e-01 -8.71222556e-01 -8.07287574e-01 2.75689036e-01
1.10217258e-01 -4.56788599e-01 1.13686609e+00 -8.63381028e-01
-2.10643196e+00 2.73280382e-01 -2.65904486e-01 2.25408047e-01
2.98259735e-01 -1.55253559e-01 -1.16106033e+00 -8.87455046e-02
-1.23051450e-01 5.31522691e-01 1.25823009e+00 -1.24247146e+00
-4.58717585e-01 -2.20956594e-01 -3.99342448e-01 4.74840961e-02
-7.31184661e-01 -1.68686986e-01 -4.28126514e-01 -1.24189377e+00
-1.77105889e-01 -8.75197113e-01 2.32985526e-01 -1.17689490e-01
-5.33052862e-01 -1.86692685e-01 9.41832602e-01 -6.19756818e-01
1.61570692e+00 -2.51569414e+00 6.51515186e-01 -3.98737602e-02
1.87318042e-01 6.20949753e-02 -1.52796760e-01 4.26357627e-01
-2.38228276e-01 4.74337459e-01 -1.60528630e-01 -5.71611643e-01
4.29494172e-01 4.89422344e-02 -4.10715401e-01 2.09544361e-01
9.45558473e-02 3.74873877e-01 -7.34141171e-01 -3.44780385e-01
2.62049109e-01 7.72693753e-01 -3.26239258e-01 5.05844295e-01
1.90169923e-02 3.90495867e-01 -2.27153197e-01 3.71405572e-01
6.45937443e-01 2.76118249e-01 -3.25899385e-02 -4.82763141e-01
-2.13189110e-01 3.55906099e-01 -1.46764338e+00 1.61832452e+00
-7.15569198e-01 7.98016965e-01 3.08180213e-01 -3.77996385e-01
1.25290525e+00 7.38652289e-01 3.21045369e-01 -4.91499364e-01
2.48946980e-01 7.37404376e-02 -2.43504927e-01 -4.32098210e-01
4.28519130e-01 -3.43610309e-02 -2.69966125e-01 3.26363713e-01
2.78711885e-01 -4.95565265e-01 -2.10994668e-02 6.55599311e-02
5.59402645e-01 -1.90564081e-01 2.19770983e-01 -2.40355745e-01
5.29542029e-01 -7.74765015e-01 8.56034994e-01 2.00977698e-01
-1.48236394e-01 5.88299811e-01 3.85567546e-01 4.14956063e-02
-1.12253177e+00 -1.32067013e+00 -2.04888418e-01 9.66300726e-01
1.57922715e-01 -7.48982608e-01 -6.60083413e-01 -2.87822992e-01
4.17729318e-02 1.01098418e+00 -7.94073939e-01 -3.65615875e-01
-4.12344247e-01 -1.26505494e-01 8.28359485e-01 4.22415286e-01
1.41757205e-01 -1.06536198e+00 1.02019764e-01 -7.23745348e-03
-2.44617522e-01 -8.26742053e-01 -1.03806949e+00 2.01359019e-01
-3.76045644e-01 -3.76023948e-01 -6.32719576e-01 -6.35504961e-01
2.25918099e-01 -2.29181960e-01 8.19003522e-01 -1.80015564e-01
-1.64876822e-02 1.23734429e-01 -3.30223382e-01 -4.46467161e-01
-6.55605257e-01 1.37571776e-02 6.25648320e-01 1.94905967e-01
2.35428382e-02 -9.60697114e-01 -4.75162327e-01 2.54557610e-01
-6.59204125e-01 1.02702072e-02 2.49796629e-01 7.76180744e-01
4.60653454e-01 -1.64001182e-01 9.08205688e-01 -3.63283902e-01
1.02010763e+00 -4.81964946e-01 -2.89507717e-01 2.46142730e-01
-6.38167918e-01 1.47862643e-01 1.11280417e+00 -7.53653884e-01
-9.81702268e-01 -4.41496044e-01 -1.41335741e-01 -7.81215727e-01
-2.86021203e-01 -4.91325743e-02 -6.64358974e-01 3.84425431e-01
3.89397174e-01 1.46154657e-01 1.30522966e-01 -5.01281202e-01
5.67840576e-01 1.04226410e+00 4.29958552e-01 -5.53532481e-01
8.59246016e-01 -9.54878181e-02 -4.78560388e-01 -1.07576466e+00
-1.89496383e-01 8.15504566e-02 -3.54949981e-01 -3.65007371e-02
7.06212521e-01 -8.62371802e-01 -8.51982713e-01 4.13007647e-01
-1.12653208e+00 -9.63171050e-02 -5.10144234e-01 6.12012446e-01
-6.16058409e-01 2.16375411e-01 -6.50899529e-01 -7.25584805e-01
-4.52400118e-01 -1.22429192e+00 9.06392932e-01 3.47014874e-01
-5.98904431e-01 -1.00751245e+00 9.87793691e-03 -1.42907783e-01
4.17588770e-01 7.11026862e-02 1.08970714e+00 -4.50639457e-01
-3.59557308e-02 1.90699100e-01 3.11685532e-01 5.67118764e-01
5.49279273e-01 4.83775258e-01 -1.35395885e+00 -2.59236246e-01
-6.01186976e-02 1.78380907e-02 4.08899903e-01 2.82240421e-01
1.27219391e+00 -4.65147942e-01 6.73977006e-03 9.63444889e-01
7.79576778e-01 5.37568688e-01 3.74588668e-01 -1.04459271e-01
6.77977800e-01 4.63153064e-01 3.30634326e-01 6.60663009e-01
7.71627277e-02 8.28079939e-01 -3.56562659e-02 -1.53122291e-01
-2.41750240e-01 -3.51452231e-01 5.71958125e-01 1.64781570e+00
4.43796664e-02 -4.32491750e-01 -1.71242476e-01 4.04063970e-01
-1.48031008e+00 -1.07749033e+00 2.07985833e-01 2.08853698e+00
1.03527582e+00 -1.89352408e-01 1.41125426e-01 2.20084533e-01
7.86774039e-01 3.22607130e-01 -6.92685664e-01 -8.03135872e-01
-3.42614233e-01 2.04122096e-01 1.14129849e-01 6.82995200e-01
-9.15006220e-01 7.63363898e-01 6.02880478e+00 1.02473676e+00
-1.25436604e+00 1.70138981e-02 3.63020837e-01 -4.27526325e-01
-7.64904976e-01 -4.21830773e-01 -3.41853410e-01 6.76755786e-01
9.63003814e-01 -5.60791373e-01 7.27863669e-01 6.80381775e-01
4.31736201e-01 6.58534825e-01 -1.09393382e+00 1.11918092e+00
-6.68313727e-02 -1.16705728e+00 2.09406778e-01 -3.26306075e-01
4.13599700e-01 -5.54672003e-01 4.82536882e-01 3.91987145e-01
2.62592994e-02 -1.12038028e+00 9.97224867e-01 5.04451931e-01
1.17595088e+00 -8.71648550e-01 2.01867387e-01 -3.89469825e-02
-1.41204953e+00 -1.21070951e-01 -9.44507793e-02 2.18866527e-01
1.16429403e-01 2.92501479e-01 -5.14630198e-01 5.29481292e-01
7.12593138e-01 7.81564534e-01 -2.59371877e-01 5.81413448e-01
-2.99585938e-01 5.42290866e-01 -7.36975949e-03 -6.57728389e-02
-2.97901720e-01 -4.30833787e-01 9.22887683e-01 1.34903252e+00
8.18102598e-01 1.48531199e-01 -1.66647807e-01 1.11792111e+00
-1.50212720e-01 2.39919811e-01 -3.95637304e-01 -2.66846925e-01
1.22320020e+00 1.26923752e+00 -6.31979629e-02 -3.21703881e-01
-1.04856305e-01 9.95991230e-01 5.20727299e-02 5.53805351e-01
-9.43438649e-01 -7.27654636e-01 1.38690615e+00 -2.46921495e-01
1.27723947e-01 -2.88554770e-03 -3.45058084e-01 -1.19417882e+00
-1.91406496e-02 -1.12431848e+00 -1.90340038e-02 -9.70668495e-01
-1.28144670e+00 1.09104133e+00 -1.92882761e-01 -1.53533208e+00
-2.80761689e-01 -3.16509038e-01 -6.21449590e-01 1.08870196e+00
-8.60957623e-01 -7.87670910e-01 -2.67019570e-01 7.40943432e-01
6.95689380e-01 -4.49844867e-01 1.38992584e+00 4.32631135e-01
-9.16597664e-01 1.21916533e+00 2.32972503e-01 -1.45450756e-01
1.06741178e+00 -1.37856257e+00 3.47567737e-01 4.66478825e-01
1.23815045e-01 6.27368450e-01 9.04416502e-01 -3.10564637e-01
-1.31525564e+00 -8.75232160e-01 6.37003064e-01 -1.22097775e-01
5.71505368e-01 -6.69978559e-01 -9.38129365e-01 5.09415150e-01
6.83801234e-01 -2.42660701e-01 8.26304615e-01 3.17858830e-02
-4.03960854e-01 -3.66380543e-01 -1.03057551e+00 7.65153229e-01
8.75430763e-01 -7.49519765e-01 -5.29083490e-01 1.38914958e-02
1.14687455e+00 -3.69239897e-01 -1.22624719e+00 1.89364240e-01
5.26578248e-01 -7.26221859e-01 8.97836506e-01 -1.76181301e-01
1.05225734e-01 -3.59579146e-01 -2.80118316e-01 -1.93650675e+00
-5.47906816e-01 -9.86935496e-01 -8.07498172e-02 1.62240326e+00
4.84977305e-01 -6.33037210e-01 4.74421680e-02 2.11409330e-01
-3.60920846e-01 -6.28728986e-01 -8.64133775e-01 -7.88463235e-01
3.22609842e-02 -2.05235839e-01 1.18980312e+00 1.11028516e+00
1.21091224e-01 4.85227346e-01 -6.47592664e-01 3.29163164e-01
3.77180934e-01 1.54760294e-02 5.46597302e-01 -1.16229188e+00
-2.08605409e-01 -6.75049543e-01 -1.80999219e-01 -6.19252324e-01
3.10257584e-01 -7.71651745e-01 -1.63188621e-01 -1.16074407e+00
-3.31470490e-01 -4.55262661e-01 -3.13558489e-01 4.29662943e-01
-1.61903694e-01 -1.75844014e-01 3.04948837e-01 -1.91443153e-02
7.82195181e-02 1.14773738e+00 1.34694541e+00 -2.18776986e-01
-3.21261942e-01 -4.01934125e-02 -6.43643022e-01 5.48607111e-01
8.32770705e-01 -2.75507659e-01 -6.30775034e-01 -3.59553277e-01
-3.49908888e-01 3.29636306e-01 -1.84344593e-02 -8.87571931e-01
5.19342646e-02 -1.62074044e-01 4.22483027e-01 -1.56768471e-01
7.73948610e-01 -7.21778929e-01 4.37353879e-01 -4.21302505e-02
-5.06804585e-01 1.96504653e-01 5.15680432e-01 2.60386407e-01
-4.11495239e-01 2.32007146e-01 7.85963893e-01 2.76612282e-01
-3.40885073e-01 4.11916256e-01 -3.33852261e-01 1.73868909e-02
8.34261835e-01 -2.23841406e-02 -2.70979613e-01 -7.04148233e-01
-1.05935919e+00 1.46512732e-01 2.14683980e-01 9.77528214e-01
6.65540636e-01 -1.94510055e+00 -7.12933540e-01 4.95549113e-01
1.07031137e-01 -5.36691189e-01 5.05422056e-01 4.75684047e-01
3.14861089e-02 1.75798461e-02 -2.53603071e-01 -2.74593413e-01
-1.33391666e+00 5.24506867e-01 2.87916422e-01 4.57930356e-01
-4.63162899e-01 8.45286548e-01 1.41859800e-01 -6.03554368e-01
4.90349144e-01 -3.77581716e-01 -1.74215570e-01 2.92627066e-01
5.41000664e-01 5.35363138e-01 -7.62901381e-02 -7.58350492e-01
-3.64252001e-01 6.36689901e-01 2.99478680e-01 -2.88661838e-01
1.12872517e+00 -3.84462297e-01 -3.89850847e-02 8.52220953e-01
1.37628376e+00 4.65147942e-01 -1.09471822e+00 -4.57068421e-02
-4.65438515e-01 -3.95606428e-01 1.39643531e-02 -9.42270339e-01
-1.10701990e+00 9.75277960e-01 7.21443236e-01 3.50162596e-01
1.27668464e+00 -2.73132324e-01 7.87087202e-01 -3.69757228e-03
-1.15322493e-01 -1.19703186e+00 -4.03656438e-02 3.78515005e-01
1.35078335e+00 -9.26734984e-01 -3.88033926e-01 -1.56651810e-01
-8.75405490e-01 1.12132406e+00 6.44480288e-01 1.65320933e-01
6.61575377e-01 6.00416720e-01 4.08630699e-01 3.04510623e-01
-8.81432831e-01 2.30229080e-01 3.53972077e-01 5.81715643e-01
5.52340031e-01 5.67758381e-01 -3.27998400e-02 9.86280560e-01
-6.60875857e-01 -6.54792726e-01 3.09070140e-01 2.12415516e-01
-1.94603875e-01 -1.25511491e+00 -4.17219013e-01 1.06918953e-01
-8.94500539e-02 -1.26773253e-01 -3.80513847e-01 6.28976643e-01
2.07907155e-01 1.13819063e+00 1.40758097e-01 -1.02000916e+00
6.76628828e-01 1.59017578e-01 1.71249405e-01 -3.22496295e-01
-5.36840796e-01 3.20886344e-01 8.68523717e-02 -4.07270402e-01
1.12799570e-01 -4.37098771e-01 -1.34880710e+00 -4.93367732e-01
-3.42740379e-02 5.43982387e-01 5.69495082e-01 4.52044249e-01
5.69833517e-01 8.50860119e-01 1.17302144e+00 -8.70441616e-01
-4.86945093e-01 -1.22507405e+00 -7.20208287e-01 4.74485338e-01
6.44249260e-01 -6.08589828e-01 -8.42629075e-01 -9.88381878e-02] | [15.006948471069336, 6.44118070602417] |
22bf033f-5c41-4640-8fc5-f688b07c7aa5 | short-utterance-compensation-in-speaker | 1810.10884 | null | http://arxiv.org/abs/1810.10884v2 | http://arxiv.org/pdf/1810.10884v2.pdf | Short utterance compensation in speaker verification via cosine-based teacher-student learning of speaker embeddings | The short duration of an input utterance is one of the most critical threats
that degrade the performance of speaker verification systems. This study aimed
to develop an integrated text-independent speaker verification system that
inputs utterances with short duration of 2 seconds or less. We propose an
approach using a teacher-student learning framework for this goal, applied to
short utterance compensation for the first time in our knowledge. The core
concept of the proposed system is to conduct the compensation throughout the
network that extracts the speaker embedding, mainly in phonetic-level, rather
than compensating via a separate system after extracting the speaker embedding.
In the proposed architecture, phonetic-level features where each feature
represents a segment of 130 ms are extracted using convolutional layers. A
layer of gated recurrent units extracts an utterance-level feature using
phonetic-level features. The proposed approach also adopts a new objective
function for teacher-student learning that considers both Kullback-Leibler
divergence of output layers and cosine distance of speaker embeddings layers.
Experiments were conducted using deep neural networks that take raw waveforms
as input, and output speaker embeddings on VoxCeleb1 dataset. The proposed
model could compensate approximately 65 \% of the performance degradation due
to the shortened duration. | ['Ha-Jin Yu', 'Hye-jin Shim', 'Hee-Soo Heo', 'Jee-weon Jung'] | 2018-10-25 | null | null | null | null | ['text-independent-speaker-verification'] | ['speech'] | [ 3.29921812e-01 3.88960928e-01 8.45976770e-02 -6.57114089e-01
-9.84304667e-01 -3.61355335e-01 2.95553476e-01 1.23523679e-02
-6.36979163e-01 2.93769866e-01 4.37806517e-01 -4.45112884e-01
2.22465210e-02 -1.13539599e-01 -4.45642918e-01 -8.17403257e-01
1.46845773e-01 -1.35199025e-01 -1.85569823e-02 -2.12387443e-01
3.97303313e-01 5.32593012e-01 -1.69458795e+00 4.89396639e-02
8.47833753e-01 1.07505453e+00 -8.68434012e-02 9.08621192e-01
5.43611236e-02 6.44620836e-01 -1.04358935e+00 -1.49183080e-01
5.62181808e-02 -3.03041041e-01 -4.64999855e-01 -2.29379222e-01
4.05123591e-01 -4.80257630e-01 -2.75376350e-01 1.03441179e+00
1.00162649e+00 1.83202639e-01 7.66375899e-01 -1.22858322e+00
-5.90549469e-01 1.03469110e+00 -2.81632811e-01 4.40720320e-01
1.21616289e-01 1.49040446e-01 6.51702642e-01 -8.98150206e-01
-2.58056894e-02 1.19633734e+00 6.36005998e-01 6.20472491e-01
-8.89923871e-01 -8.49950314e-01 -5.29465452e-02 3.53271544e-01
-1.34713268e+00 -8.79973352e-01 1.09127116e+00 -3.79327506e-01
1.13300586e+00 2.15394720e-01 5.48032299e-02 8.68620932e-01
1.09482788e-01 6.40293896e-01 8.28742325e-01 -6.13907039e-01
-4.66287322e-02 7.82952130e-01 4.98315185e-01 1.59009278e-01
-4.45775330e-01 2.46140867e-01 -4.71828431e-01 1.90655828e-01
-2.01232266e-03 -2.95876354e-01 -1.59293547e-01 3.07390273e-01
-5.35034359e-01 7.58311272e-01 8.00290108e-02 5.40775359e-01
-1.83858514e-01 -1.97472088e-02 6.86934114e-01 3.61760467e-01
3.54064941e-01 -1.46614045e-01 -3.48649472e-01 -1.76117167e-01
-1.31404781e+00 -1.31654376e-02 5.90968072e-01 6.98635399e-01
1.86621040e-01 8.90129209e-01 -1.90552816e-01 7.07196116e-01
5.36850631e-01 6.73931420e-01 9.59280074e-01 -2.12708846e-01
3.91384870e-01 1.83996290e-01 -7.23559633e-02 -6.57770157e-01
-2.54720241e-01 -6.05511904e-01 -4.95971739e-01 3.08950126e-01
2.57267684e-01 -3.75108600e-01 -8.73844028e-01 1.95763004e+00
2.47193053e-01 1.37889966e-01 4.18311268e-01 9.08413351e-01
1.00946605e+00 8.07713389e-01 3.88135463e-02 -3.73384416e-01
1.42733884e+00 -9.79644656e-01 -1.15412617e+00 6.86570033e-02
2.60452092e-01 -1.07912064e+00 9.95310724e-01 5.28984308e-01
-9.58115578e-01 -9.85851347e-01 -1.40185070e+00 7.08583593e-02
-5.12488246e-01 3.83707821e-01 -3.31533968e-01 1.09510696e+00
-1.05671370e+00 4.21089828e-01 -3.11498284e-01 -1.33844435e-01
-2.69569248e-01 4.43152398e-01 -9.03774276e-02 6.27159357e-01
-1.51875567e+00 1.06940603e+00 1.02641411e-01 2.76238143e-01
-1.05823410e+00 -6.89126253e-01 -8.01123083e-01 4.60798800e-01
-3.15101624e-01 2.49002188e-01 1.44188201e+00 -1.02302122e+00
-1.92368257e+00 4.45790261e-01 -3.11060220e-01 -7.65217543e-01
3.96692783e-01 -1.53532743e-01 -8.31746101e-01 6.25344664e-02
-4.09099907e-01 3.23924690e-01 1.34406960e+00 -8.27702463e-01
-2.77135372e-01 -3.91394585e-01 -5.27904630e-01 1.91793948e-01
-7.59244263e-01 2.92355776e-01 2.53757894e-01 -4.38249767e-01
-6.52480572e-02 -5.31141579e-01 2.72679865e-01 -5.95282376e-01
-4.09524500e-01 -4.16060448e-01 1.28621459e+00 -1.28644907e+00
1.31509483e+00 -2.29963827e+00 -1.51781425e-01 1.08907865e-02
-2.79311508e-01 6.09356225e-01 -1.26845762e-01 4.62260276e-01
-3.94556522e-01 -7.83057287e-02 -2.40938291e-01 -7.40090370e-01
2.04208151e-01 -4.73959863e-01 -6.34108007e-01 6.02363348e-01
1.75523296e-01 3.11101854e-01 -2.89106458e-01 -4.08548772e-01
3.89810592e-01 9.33851302e-01 -3.13112408e-01 4.19483572e-01
2.65103996e-01 2.82570630e-01 1.69080764e-01 2.74520963e-01
1.09377885e+00 8.32582474e-01 -2.17665926e-01 -1.50887012e-01
-3.81628692e-01 5.62352002e-01 -1.26365948e+00 1.35309076e+00
-6.04032397e-01 8.26401770e-01 3.77955794e-01 -1.16744733e+00
1.35753810e+00 9.53404546e-01 1.80870965e-01 -5.50591588e-01
5.60631812e-01 3.24416280e-01 1.46236807e-01 -5.37504196e-01
4.47868794e-01 -2.91842043e-01 8.20084363e-02 4.55046833e-01
1.84715137e-01 -2.26600111e-01 -4.44713324e-01 2.32816115e-02
6.05886400e-01 -1.95638821e-01 3.84650677e-02 -9.60427001e-02
1.03182983e+00 -6.34013414e-01 4.06835407e-01 3.58925283e-01
-7.19872534e-01 3.32020134e-01 2.48498991e-01 8.44528228e-02
-9.72380877e-01 -9.35706079e-01 -3.07238698e-01 1.07929516e+00
-2.51898497e-01 1.63217694e-01 -1.07907963e+00 -3.87989521e-01
6.75906390e-02 1.02621126e+00 -4.27244574e-01 -4.54610854e-01
-5.46102226e-01 -1.52115196e-01 1.29230070e+00 4.46102202e-01
2.95175403e-01 -1.10823321e+00 -4.40223336e-01 2.82768250e-01
6.95590079e-02 -7.76737213e-01 -8.31998765e-01 5.16801476e-01
-6.20357215e-01 -4.27930325e-01 -6.60056531e-01 -9.91095901e-01
3.66574079e-01 -4.86852974e-02 3.36206853e-01 -1.85614184e-01
-2.05816612e-01 -1.00351021e-01 -9.07054916e-02 -6.59678459e-01
-5.47224998e-01 1.40640989e-01 5.61744094e-01 2.94102818e-01
6.40962362e-01 -5.85132301e-01 -4.53394443e-01 1.44826412e-01
-6.70929492e-01 -7.29716718e-01 5.72431445e-01 7.86424160e-01
-6.40913695e-02 -2.97199711e-02 1.23624861e+00 -3.65440816e-01
8.37613404e-01 -2.81495929e-01 -5.85210800e-01 2.50657853e-02
-6.29300654e-01 6.48657382e-02 8.53857577e-01 -5.83730340e-01
-1.27289307e+00 1.06607236e-01 -6.40655279e-01 -4.06946659e-01
-3.97898078e-01 1.84759080e-01 -4.45313781e-01 1.50297925e-01
5.64417779e-01 5.65442562e-01 9.55032930e-02 -5.20807981e-01
3.48191828e-01 1.51927090e+00 6.17141664e-01 3.10709160e-02
7.54535258e-01 -9.94350016e-02 -6.15136623e-01 -1.08789623e+00
-3.79682809e-01 -6.14924431e-01 -6.09216690e-01 -2.28513330e-01
7.77735054e-01 -8.66412580e-01 -1.08100080e+00 5.09403229e-01
-1.32811093e+00 2.60174274e-01 -1.53241018e-02 7.96960831e-01
-3.49755406e-01 3.24624926e-01 -6.40848577e-01 -1.42106497e+00
-9.80962753e-01 -1.35691929e+00 6.64959908e-01 3.41454804e-01
-5.82815483e-02 -6.26267314e-01 1.03541546e-01 3.85449082e-01
8.15808475e-01 -2.86956280e-01 5.63156426e-01 -1.19983578e+00
8.00260925e-04 -4.63553518e-01 7.32157603e-02 1.04193521e+00
1.72017649e-01 1.44327909e-01 -1.81876671e+00 -3.59059781e-01
5.25138199e-01 -2.14899525e-01 8.26043546e-01 4.72039044e-01
8.52980852e-01 -3.92038643e-01 1.00062080e-01 4.33665276e-01
1.26732707e+00 5.40902257e-01 4.13605511e-01 -1.90641090e-01
3.16315979e-01 5.95457673e-01 4.82391953e-01 2.55524844e-01
6.45109266e-02 4.75276858e-01 3.17195266e-01 -4.28259075e-02
-2.59199709e-01 -2.22371727e-01 7.37484038e-01 1.47066319e+00
5.42486727e-01 -1.57178581e-01 -6.12073123e-01 7.59742379e-01
-1.27060330e+00 -1.24398696e+00 -1.05496747e-02 2.27966070e+00
7.98819363e-01 2.42536142e-01 2.17764154e-02 8.33450913e-01
8.90544295e-01 1.73381701e-01 -6.62174702e-01 -1.16321290e+00
-3.93674262e-02 1.72564134e-01 3.16594929e-01 1.05814290e+00
-1.05437827e+00 6.59580350e-01 5.47585154e+00 7.39505351e-01
-1.69478416e+00 2.28953198e-01 1.73823863e-01 4.05960679e-02
-5.32918237e-02 -3.32085967e-01 -9.02243495e-01 4.07365620e-01
1.58789349e+00 -2.10618377e-01 1.56222641e-01 1.07137835e+00
5.49914777e-01 2.65721887e-01 -1.03374887e+00 7.56156147e-01
4.87661391e-01 -7.02247083e-01 -3.44079107e-01 -2.12936208e-01
3.95667106e-01 -2.26897731e-01 6.08723879e-01 4.62080389e-01
-3.73187780e-01 -1.00068176e+00 9.27982450e-01 3.82346749e-01
8.10727417e-01 -9.98617530e-01 8.97563338e-01 3.90601903e-01
-1.26055348e+00 -1.59373909e-01 -2.73484617e-01 -1.70627497e-02
-2.73569524e-02 2.54741609e-01 -1.49257743e+00 1.48562953e-01
3.35523725e-01 1.82679430e-01 -1.79721683e-01 8.38480651e-01
-1.10392652e-01 8.94458055e-01 -1.57532498e-01 -1.14016354e-01
-7.94699937e-02 2.84848779e-01 5.64409792e-01 1.60674524e+00
3.63497287e-01 -2.59538144e-01 -4.49345112e-01 5.67469180e-01
-3.25951818e-03 3.09826404e-01 -6.50688589e-01 1.13701954e-01
8.69313955e-01 1.12443209e+00 -1.21218406e-01 -4.90287811e-01
-2.39817183e-02 7.78127134e-01 4.03317921e-02 2.62795240e-01
-1.08975899e+00 -1.19700241e+00 5.41401863e-01 -1.67381555e-01
4.21781033e-01 2.22309858e-01 -4.52124536e-01 -7.28798985e-01
3.10247373e-02 -7.97863007e-01 -3.49412933e-02 -2.96402603e-01
-8.23912799e-01 8.74746501e-01 -3.11083347e-01 -1.24248171e+00
-4.57323939e-01 -2.30735913e-01 -1.00130534e+00 1.42366827e+00
-1.73242748e+00 -9.68581498e-01 -1.32025629e-01 5.51809907e-01
7.84800112e-01 -3.47604722e-01 9.19281602e-01 5.40355861e-01
-7.05419779e-01 1.26911438e+00 1.40154377e-01 1.71620682e-01
9.35976505e-01 -1.10704923e+00 2.16627553e-01 1.05806780e+00
-2.35641077e-01 6.90070629e-01 9.79626477e-01 -2.63838202e-01
-1.15987766e+00 -7.55725563e-01 1.18027508e+00 -6.46557733e-02
2.85930693e-01 -4.08846349e-01 -8.10067594e-01 5.82709253e-01
7.03617990e-01 -2.73657829e-01 6.97645605e-01 -4.24896121e-01
-4.94785011e-01 -5.73879600e-01 -1.50065029e+00 -8.14130530e-02
6.79814965e-02 -9.88829255e-01 -1.01340055e+00 -9.83275920e-02
9.72141445e-01 -1.94117874e-01 -7.78424919e-01 6.97934255e-02
6.11752808e-01 -8.31275284e-01 7.45784760e-01 -3.18174720e-01
-2.41336301e-02 -3.80124241e-01 -2.22412050e-01 -1.44312930e+00
2.05457583e-01 -4.48270589e-01 1.46080330e-01 1.73379636e+00
6.16509914e-01 -5.31772733e-01 5.64302325e-01 3.55521291e-01
-3.97369117e-01 -3.39338541e-01 -1.09036219e+00 -6.47430658e-01
1.47948712e-01 -4.59468573e-01 7.39629269e-01 5.52883267e-01
2.62930572e-01 2.11540982e-01 -4.71846372e-01 5.94411016e-01
7.72346795e-01 -3.73537004e-01 4.97625560e-01 -8.91150057e-01
-1.88379269e-02 -3.90916705e-01 -2.95760751e-01 -7.89140284e-01
1.90981016e-01 -7.39377260e-01 4.82376397e-01 -1.03999388e+00
-3.06129932e-01 -4.19510575e-03 -5.22501588e-01 2.30864227e-01
-8.61553103e-02 5.95907774e-03 2.42706910e-01 -2.66274363e-01
2.00557962e-01 7.13935435e-01 6.19393170e-01 -4.05904174e-01
-2.62883991e-01 1.38327792e-01 -2.86945254e-01 4.59682167e-01
8.09262156e-01 -4.05092090e-01 -5.31687737e-01 -3.73941511e-01
-8.08708429e-01 2.24783227e-01 1.87251121e-02 -1.09756172e+00
4.48911190e-01 1.99766397e-01 2.21328259e-01 -1.00182724e+00
6.12822175e-01 -9.24063981e-01 -1.59901306e-01 7.70647109e-01
-7.88703322e-01 -2.01011926e-01 2.75995910e-01 8.78064483e-02
-5.13138354e-01 -4.57572162e-01 9.61112380e-01 4.72244292e-01
-2.44324043e-01 -2.09172010e-01 -4.94720578e-01 -4.64261591e-01
9.02713418e-01 -1.13261797e-01 -1.91335931e-01 -3.71091962e-01
-6.95383549e-01 -2.22064674e-01 -3.07843894e-01 5.40132344e-01
8.09994221e-01 -1.17641652e+00 -7.88207769e-01 4.17666554e-01
-1.92190021e-01 -6.28791332e-01 5.79545856e-01 7.86220908e-01
-1.71161145e-01 7.96925426e-01 -1.79200366e-01 -4.57750708e-01
-1.61239147e+00 6.41658962e-01 5.11455894e-01 2.04574779e-01
-1.20919332e-01 1.11741257e+00 -1.99268445e-01 -6.83516800e-01
9.05516326e-01 -4.51497406e-01 -5.53433239e-01 4.28287566e-01
7.34434605e-01 3.24366271e-01 4.31908578e-01 -9.31302786e-01
-5.58470607e-01 5.11841595e-01 -8.13218132e-02 -3.99000704e-01
1.18118632e+00 -1.77054837e-01 2.54660040e-01 4.17339683e-01
1.54672098e+00 3.47185731e-01 -9.49614465e-01 -2.19444573e-01
-1.39758632e-01 -9.19430256e-02 2.41604015e-01 -8.77290964e-01
-9.35825884e-01 1.54487908e+00 1.08717513e+00 1.43193483e-01
1.16401553e+00 -5.35298347e-01 8.83642733e-01 -1.05858982e-01
-3.92997086e-01 -1.17243659e+00 -7.14635178e-02 2.67577142e-01
8.81361008e-01 -1.04128432e+00 -5.23965299e-01 1.77692652e-01
-6.28478587e-01 1.21673250e+00 6.85401142e-01 8.82189944e-02
7.64940679e-01 3.34708512e-01 5.42623222e-01 3.54808182e-01
-6.71331644e-01 7.83679783e-02 1.72082409e-01 5.98795593e-01
6.85043275e-01 1.74245343e-01 -3.47544700e-01 8.82165551e-01
-5.09953737e-01 -2.94208139e-01 5.24074316e-01 5.80636680e-01
-6.84580386e-01 -8.57455373e-01 -6.88901961e-01 -2.59979721e-02
-4.58284706e-01 -1.61583006e-01 -8.72557908e-02 2.00070381e-01
2.33606145e-01 1.39006901e+00 4.03441722e-03 -6.65037394e-01
3.44637454e-01 5.71389496e-01 -1.44189790e-01 -2.15054527e-01
-9.22776103e-01 1.16320111e-01 -2.40036070e-01 -8.16903561e-02
-2.79375687e-02 -6.46506548e-01 -1.38518727e+00 -1.40901104e-01
-4.94538218e-01 4.17731851e-01 1.48405159e+00 8.33579063e-01
4.70416248e-03 7.71589339e-01 1.23895669e+00 -6.49378479e-01
-1.23451316e+00 -1.56361222e+00 -4.85328972e-01 2.03410342e-01
1.06976962e+00 -2.62482941e-01 -8.43181252e-01 6.34116754e-02] | [14.378564834594727, 6.061183452606201] |
c7661106-c6ce-48b3-b361-2308d28332a0 | towards-adversarial-denoising-of-radar-micro | 1811.04678 | null | https://arxiv.org/abs/1811.04678v3 | https://arxiv.org/pdf/1811.04678v3.pdf | Towards Adversarial Denoising of Radar Micro-Doppler Signatures | Generative Adversarial Networks (GANs) are considered the state-of-the-art in the field of image generation. They learn the joint distribution of the training data and attempt to generate new data samples in high dimensional space following the same distribution as the input. Recent improvements in GANs opened the field to many other computer vision applications based on improving and changing the characteristics of the input image to follow some given training requirements. In this paper, we propose a novel technique for the denoising and reconstruction of the micro-Doppler ($\boldsymbol{\mu}$-D) spectra of walking humans based on GANs. Two sets of experiments were collected on 22 subjects walking on a treadmill at an intermediate velocity using a \unit[25]{GHz} CW radar. In one set, a clean $\boldsymbol{\mu}$-D spectrum is collected for each subject by placing the radar at a close distance to the subject. In the other set, variations are introduced in the experiment setup to introduce different noise and clutter effects on the spectrum by changing the distance and placing reflective objects between the radar and the target. Synthetic paired noisy and noise-free spectra were used for training, while validation was carried out on the real noisy measured data. Finally, qualitative and quantitative comparison with other classical radar denoising approaches in the literature demonstrated the proposed GANs framework is better and more robust to different noise levels. | ['Fady Aziz', 'Urs Schneider', 'Sherif Abdulatif', 'Karim Armanious', 'Bin Yang'] | 2018-11-12 | null | null | null | null | ['denoising-of-radar-micro-doppler-signatures'] | ['miscellaneous'] | [ 6.64547384e-01 1.29273552e-02 5.39657891e-01 -1.87633634e-01
-7.06653833e-01 -1.10168867e-01 3.66988569e-01 -3.98770928e-01
-4.60275233e-01 1.05743623e+00 7.93506857e-03 2.69402325e-01
-2.84632295e-01 -1.12854111e+00 -6.30402029e-01 -1.17518473e+00
-2.44975030e-01 3.11502665e-01 -1.52453274e-01 -1.96864203e-01
-1.66790932e-01 4.03950065e-01 -1.58553910e+00 -1.56159371e-01
7.29457438e-01 8.65906596e-01 6.93182349e-02 6.27732754e-01
3.83269519e-01 2.98323393e-01 -1.33764637e+00 -3.02839875e-01
6.60083950e-01 -9.41700816e-01 2.54023075e-01 -5.90300038e-02
1.72603011e-01 -1.72606930e-01 -4.91646439e-01 1.29076505e+00
8.67544353e-01 3.01628381e-01 9.07925844e-01 -8.94907296e-01
-6.18306577e-01 2.68101275e-01 -6.53484464e-01 2.48600885e-01
1.19020805e-01 3.47607732e-01 2.19367012e-01 -2.19259903e-01
5.06285846e-01 9.47031677e-01 6.23921812e-01 6.03459597e-01
-1.36001182e+00 -7.39526033e-01 -4.79322851e-01 3.86921316e-01
-1.16097486e+00 -8.02271888e-02 1.12214446e+00 -3.07708293e-01
1.31083652e-01 3.62077832e-01 7.25200236e-01 1.53067887e+00
4.38542902e-01 -1.56422481e-01 1.55327356e+00 -6.66973948e-01
2.88509250e-01 5.57725430e-02 -6.21742532e-02 4.10217434e-01
5.03685534e-01 8.12364876e-01 -3.26277107e-01 -3.57150584e-02
6.41181231e-01 -3.33436728e-01 -4.88888741e-01 -7.57797956e-02
-1.03249633e+00 8.48335922e-01 4.77479786e-01 3.33658367e-01
-7.24787533e-01 1.17541231e-01 1.28274471e-01 2.92852342e-01
3.01201820e-01 2.41269901e-01 2.09322631e-01 3.10100615e-02
-7.38613844e-01 3.22689474e-01 6.59534693e-01 4.83122408e-01
5.23391008e-01 7.58298993e-01 -2.80235082e-01 6.04604185e-01
1.69621646e-01 1.14714479e+00 2.85438061e-01 -6.45016670e-01
3.23414594e-01 -1.55429378e-01 2.32375383e-01 -1.18004465e+00
-3.39919209e-01 -9.88414347e-01 -1.22206414e+00 6.68052614e-01
4.09690559e-01 -7.76436150e-01 -1.17790306e+00 1.85232496e+00
3.42051387e-01 4.45556432e-01 3.82921904e-01 1.22038317e+00
6.37093782e-01 8.75567496e-01 -1.20462403e-01 -4.66592044e-01
1.12587297e+00 -1.99507833e-01 -8.39254618e-01 -1.94209307e-01
-4.03595179e-01 -8.47531438e-01 8.28901529e-01 6.33665025e-01
-7.18917668e-01 -1.04080975e+00 -1.35056305e+00 7.60197878e-01
2.15381887e-02 -6.23554848e-02 5.35955392e-02 1.32116556e+00
-5.08277059e-01 5.11228323e-01 -5.47842979e-01 1.61744729e-02
1.09160326e-01 -2.36369789e-01 -2.55395978e-04 -2.89350569e-01
-1.45331383e+00 6.84899986e-01 -3.31010967e-02 4.38944131e-01
-9.95507956e-01 -7.97720671e-01 -5.99721789e-01 -3.05797726e-01
7.67758489e-02 -7.80418634e-01 7.20266640e-01 -9.31016088e-01
-1.68673754e+00 6.44038320e-01 3.98389041e-01 -7.66911387e-01
5.15313685e-01 -2.77558446e-01 -9.59449351e-01 2.13369280e-01
-9.07019824e-02 8.48407745e-02 1.42859507e+00 -1.40835810e+00
-4.42335419e-02 -4.28607255e-01 -4.32755917e-01 7.44934455e-02
4.02680635e-01 -5.65869391e-01 3.54236484e-01 -9.55712557e-01
1.53722495e-01 -7.12344170e-01 -8.07214230e-02 -5.69748461e-01
-4.64650780e-01 6.52277708e-01 7.42105186e-01 -7.67930686e-01
6.26178563e-01 -2.25648069e+00 1.19292125e-01 6.09068632e-01
-1.26053691e-01 2.78233469e-01 -9.29172151e-03 5.46076596e-01
-9.71931517e-02 -3.06734383e-01 -4.99857426e-01 4.49536107e-02
-1.49361879e-01 2.33692210e-02 -1.78172335e-01 6.53287351e-01
-7.84228817e-02 3.22081745e-01 -5.59956074e-01 -4.03740555e-02
1.53976724e-01 6.42999589e-01 -1.71179235e-01 4.72111702e-01
-1.15020424e-01 8.54127228e-01 -3.66474330e-01 2.02741086e-01
8.12480748e-01 7.35215306e-01 -1.21983953e-01 -4.50683862e-01
1.02371544e-01 -4.82923299e-01 -1.22118616e+00 1.18228948e+00
-3.75566065e-01 4.76877332e-01 3.43759775e-01 -1.50497353e+00
1.40519583e+00 1.96036264e-01 5.14522314e-01 -9.95675385e-01
2.07513064e-01 -1.54312074e-01 4.37945485e-01 -6.61945164e-01
-1.82915404e-02 -7.78089583e-01 -1.13808870e-01 3.29294026e-01
6.80162609e-02 -4.50544477e-01 -6.92494288e-02 -2.46527225e-01
1.14012289e+00 1.49115641e-02 1.26987740e-01 -6.94583803e-02
4.57733989e-01 4.74500582e-02 5.62777460e-01 8.69686067e-01
-7.53655136e-02 5.32579541e-01 8.84408653e-02 -1.52552634e-01
-9.99534369e-01 -1.45734429e+00 -4.40487564e-02 2.90223151e-01
4.33061421e-02 3.00590187e-01 -9.48168039e-01 -1.96640462e-01
-1.21327825e-01 8.50836515e-01 -5.91991842e-01 -3.98540705e-01
-5.61045647e-01 -1.03707266e+00 7.73773313e-01 -5.36113828e-02
1.07718277e+00 -1.08723080e+00 -9.40764964e-01 2.96492696e-01
-1.88196793e-01 -1.02917194e+00 8.43147561e-02 2.34349936e-01
-5.27596712e-01 -8.06864440e-01 -8.16773593e-01 -3.97123188e-01
5.28634608e-01 -1.58545315e-01 8.87693703e-01 -4.44881648e-01
-9.47302818e-01 5.88758469e-01 -5.14498413e-01 -8.17835689e-01
-4.87514079e-01 -5.14992416e-01 -5.65034598e-02 2.99610794e-01
8.50893110e-02 -8.52983236e-01 -6.81600571e-01 2.18815625e-01
-8.52996886e-01 -3.48782331e-01 6.50543749e-01 1.03870440e+00
4.02349025e-01 6.79077387e-01 7.49134362e-01 -7.69052923e-01
7.23886609e-01 -3.70262623e-01 -6.05702937e-01 -3.72504652e-01
-2.68196851e-01 -1.08118122e-02 7.48537302e-01 -3.44811380e-01
-1.35147238e+00 -2.09157974e-01 -2.64238775e-01 -1.58713639e-01
-5.73661864e-01 3.59567434e-01 -3.25938106e-01 4.23256904e-02
1.12268174e+00 3.86747897e-01 1.20418273e-01 -2.48391867e-01
3.02414209e-01 3.64688754e-01 9.13320541e-01 -3.92071664e-01
1.39540529e+00 3.91147733e-01 3.09350967e-01 -1.24643850e+00
-5.24871886e-01 3.36648971e-02 -2.83507556e-01 -4.02744561e-01
1.10615373e+00 -5.96931577e-01 -3.92902792e-01 7.56473958e-01
-6.47569418e-01 -3.43342900e-01 -4.73687291e-01 7.82191277e-01
-5.62143564e-01 2.68417954e-01 -2.35849112e-01 -9.55841064e-01
-3.72150570e-01 -5.61233222e-01 5.45827448e-01 4.29572284e-01
1.88239850e-02 -6.86524272e-01 1.80552110e-01 4.03207630e-01
5.74955642e-01 8.84733856e-01 1.09960699e+00 -1.54563397e-01
-3.01814288e-01 -2.80254364e-01 2.97724038e-01 7.75521517e-01
3.07530075e-01 -4.72441316e-01 -8.47596884e-01 -1.59421459e-01
6.85147405e-01 -3.46189663e-02 5.71332216e-01 7.08739042e-01
7.30332375e-01 -1.00911073e-01 -7.73073360e-02 5.11237025e-01
1.56784332e+00 5.75665712e-01 1.06908965e+00 3.85047607e-02
1.33521348e-01 4.45043892e-01 6.50282264e-01 3.99814934e-01
-5.46496928e-01 4.55546379e-01 4.70165402e-01 -2.03253582e-01
-3.16992462e-01 -1.31690511e-02 3.64872187e-01 1.69423893e-01
-3.45315307e-01 -4.04007763e-01 -3.22870940e-01 1.97799996e-01
-1.13716924e+00 -1.40116167e+00 -6.12191111e-02 2.41437149e+00
4.27556694e-01 3.36663127e-01 1.21815443e-01 5.76054871e-01
8.06576490e-01 1.88818529e-01 -4.22291636e-01 -4.14509252e-02
-1.07308127e-01 1.01712656e+00 4.92220998e-01 3.96971166e-01
-7.94870436e-01 1.46029681e-01 5.26898146e+00 5.49262762e-01
-1.35233712e+00 5.30640520e-02 3.38522911e-01 4.29685377e-02
-7.43694305e-02 -4.91403997e-01 -3.69948268e-01 5.10862768e-01
8.94497335e-01 -1.90069780e-01 3.54461938e-01 3.94001842e-01
1.88147008e-01 -3.64775836e-01 -4.58579183e-01 9.73998487e-01
1.32168874e-01 -7.18570054e-01 -3.57148677e-01 -3.80920470e-02
5.57661057e-01 -3.77319098e-01 2.36528337e-01 1.54780865e-01
1.55000106e-01 -1.06709170e+00 4.09958661e-01 9.14071381e-01
7.47454643e-01 -9.32076693e-01 9.46404338e-01 5.79009295e-01
-7.68230736e-01 1.37752935e-01 -3.63079309e-01 9.60467383e-02
3.75370800e-01 1.05100894e+00 -8.55505049e-01 8.92602921e-01
6.09086335e-01 -9.00276005e-02 -1.19145714e-01 7.37602353e-01
-3.90151501e-01 9.63881612e-01 -2.71053642e-01 -7.28063472e-03
1.25115812e-01 -6.81555271e-01 9.69109237e-01 9.49991882e-01
6.87479615e-01 3.84103119e-01 -4.31326963e-02 9.86936808e-01
2.28381559e-01 -2.23242283e-01 -7.22201288e-01 1.31790236e-01
2.05068707e-01 1.08914101e+00 -4.37363952e-01 -4.83676940e-02
7.12799653e-02 7.90572882e-01 -5.42309999e-01 5.86665154e-01
-1.13995159e+00 -6.66031837e-01 5.10583758e-01 3.47577661e-01
2.73818016e-01 -3.98239523e-01 -4.14804481e-02 -5.58221936e-01
5.70175797e-02 -8.01385999e-01 2.41133690e-01 -6.92678452e-01
-1.57039213e+00 7.00743377e-01 2.04400778e-01 -1.28021276e+00
-3.84028077e-01 -3.01597238e-01 -5.82296431e-01 1.21745062e+00
-1.09552383e+00 -7.17020571e-01 -8.43914092e-01 6.96314275e-01
2.42424801e-01 -4.74837184e-01 1.00648308e+00 2.38399431e-01
-1.49060771e-01 2.72197574e-01 1.35054290e-01 8.00836831e-02
6.02692902e-01 -9.58043814e-01 2.04458535e-01 9.13337052e-01
2.01924164e-02 1.86031908e-01 1.24415660e+00 -8.64710093e-01
-1.24007177e+00 -1.03606987e+00 -1.41520441e-01 4.59578931e-01
1.50225490e-01 -4.51261729e-01 -5.25331378e-01 1.50241807e-01
4.33968663e-01 -3.56003493e-02 5.48750222e-01 -5.05157292e-01
1.27239570e-01 -4.42795873e-01 -1.74760330e+00 3.14516157e-01
8.19075584e-01 -3.53150927e-02 -5.54629326e-01 1.37424067e-01
3.48496288e-01 -4.51454699e-01 -6.92706525e-01 4.12591249e-01
2.89134562e-01 -1.16317129e+00 1.13630438e+00 -2.81956851e-01
1.10506669e-01 -3.63258004e-01 -3.22003335e-01 -1.93766510e+00
-2.70689070e-01 -6.08972430e-01 2.96274036e-01 1.08915126e+00
-5.83122019e-03 -7.72413492e-01 8.35174620e-01 -2.66212523e-01
-9.85733122e-02 -2.77171731e-01 -1.02628911e+00 -8.44567955e-01
-1.05976067e-01 -2.67054856e-01 2.38613769e-01 4.12446916e-01
-5.96031725e-01 3.51233691e-01 -5.45803487e-01 5.96803963e-01
1.38178527e+00 -1.44239187e-01 8.19970071e-01 -1.13026428e+00
-5.79995334e-01 2.16797799e-01 -7.30420232e-01 -5.07203102e-01
-4.12803739e-02 -4.61769849e-01 2.44090796e-01 -1.35076308e+00
-5.49196124e-01 -4.08254683e-01 1.10210933e-01 -2.44537085e-01
1.43004045e-01 3.69452745e-01 6.09418228e-02 -3.57068628e-01
6.34348214e-01 5.80768824e-01 1.09935820e+00 -2.37325191e-01
2.83167623e-02 5.53759336e-01 -2.68404633e-01 5.39982796e-01
8.32517862e-01 -5.36515057e-01 -6.07746959e-01 -9.18398146e-03
-2.13657960e-01 3.79479051e-01 6.94782615e-01 -1.74187088e+00
-2.38549680e-01 9.01873782e-03 6.60145342e-01 -4.11102176e-01
6.48440838e-01 -9.23048437e-01 6.70585752e-01 6.52075112e-01
5.28377146e-02 -1.80148274e-01 1.04176760e-01 7.52814829e-01
-1.99261770e-01 -3.28025371e-01 1.27017713e+00 -2.98652709e-01
-3.23678583e-01 -1.77805007e-01 -4.60891038e-01 1.22140251e-01
1.17292655e+00 -1.14268422e-01 -3.36424351e-01 -7.42027283e-01
-9.10530090e-01 -5.37711322e-01 -4.69583198e-02 -7.88528770e-02
5.92866898e-01 -1.16425490e+00 -8.78954589e-01 4.27621722e-01
-3.92971218e-01 -3.11114162e-01 6.60785377e-01 4.69228745e-01
-3.35155070e-01 -5.65082356e-02 -6.83886230e-01 -4.42925483e-01
-1.14151633e+00 3.27263802e-01 5.78591466e-01 -1.54941976e-01
-6.47572637e-01 5.81066310e-01 -2.03752488e-01 -4.19671834e-02
7.53160194e-02 -7.15809166e-02 -8.65399539e-02 -1.33815020e-01
2.97573060e-01 5.24631977e-01 1.62276968e-01 -2.81164289e-01
3.02837156e-02 4.99946713e-01 5.47000766e-01 -5.22034407e-01
1.35679531e+00 1.77690342e-01 2.20230445e-01 3.08193088e-01
8.05834413e-01 2.87045240e-01 -1.03984928e+00 -8.05862769e-02
-5.11671901e-01 -4.54237938e-01 -1.37547195e-01 -1.05695128e+00
-1.21796691e+00 7.33292222e-01 1.19778275e+00 4.16574001e-01
1.42818785e+00 -5.85696280e-01 6.33179426e-01 6.04143217e-02
4.55353200e-01 -9.51179385e-01 2.09328815e-01 -3.73079628e-03
8.02561045e-01 -6.10979617e-01 -3.35569829e-02 -4.79703069e-01
-6.23106241e-01 9.57438767e-01 4.23387647e-01 -5.12034178e-01
6.36845052e-01 4.90818083e-01 3.20049286e-01 -1.20483078e-01
-5.23094982e-02 -1.27141237e-01 1.82368178e-02 1.38565314e+00
1.49756432e-01 1.22622490e-01 -2.79503465e-01 5.72311342e-01
-4.07910883e-01 -1.20482206e-01 6.80611849e-01 7.17705965e-01
-3.96561742e-01 -1.08848953e+00 -1.01630783e+00 4.40556049e-01
-4.41822320e-01 2.63356477e-01 3.09051611e-02 9.27656054e-01
4.82981265e-01 9.92797852e-01 -1.52439907e-01 -2.64368653e-01
7.15859711e-01 2.68162727e-01 6.59206092e-01 -3.68504107e-01
-3.75253677e-01 8.50644559e-02 1.24355450e-01 -1.47068337e-01
-4.59195137e-01 -7.98735201e-01 -1.10826969e+00 2.89712325e-02
1.30882606e-01 3.70825022e-01 7.56625414e-01 6.31703138e-01
2.72065550e-02 8.78902793e-01 7.17355490e-01 -7.50371099e-01
-8.34680974e-01 -1.15989852e+00 -9.32526469e-01 3.87339085e-01
2.56426156e-01 -7.65682399e-01 -5.67821801e-01 2.37317085e-01] | [14.080498695373535, 1.6488419771194458] |
5f61d115-9bc2-4e1c-89ab-ea958cf8360c | actively-supervised-clustering-for-open | 2306.04968 | null | https://arxiv.org/abs/2306.04968v1 | https://arxiv.org/pdf/2306.04968v1.pdf | Actively Supervised Clustering for Open Relation Extraction | Current clustering-based Open Relation Extraction (OpenRE) methods usually adopt a two-stage pipeline. The first stage simultaneously learns relation representations and assignments. The second stage manually labels several instances and thus names the relation for each cluster. However, unsupervised objectives struggle to optimize the model to derive accurate clustering assignments, and the number of clusters has to be supplied in advance. In this paper, we present a novel setting, named actively supervised clustering for OpenRE. Our insight lies in that clustering learning and relation labeling can be alternately performed, providing the necessary guidance for clustering without a significant increase in human effort. The key to the setting is selecting which instances to label. Instead of using classical active labeling strategies designed for fixed known classes, we propose a new strategy, which is applicable to dynamically discover clusters of unknown relations. Experimental results show that our method is able to discover almost all relational clusters in the data and improve the SOTA methods by 10.3\% and 5.2\%, on two datasets respectively. | ['Mingming Sun', 'Minlong Peng', 'Zhongyu Wei', 'Tao Gui', 'Qi Zhang', 'Yongxin Zhang', 'Jun Zhao'] | 2023-06-08 | null | null | null | null | ['relation-extraction'] | ['natural-language-processing'] | [ 2.80220598e-01 6.63036764e-01 -4.32905257e-01 -3.44191611e-01
-7.54183292e-01 -7.68356740e-01 3.35524857e-01 5.22769272e-01
-1.82503805e-01 7.19805062e-01 -2.59305626e-01 -1.69392034e-01
-4.45219845e-01 -8.42131257e-01 -3.55139285e-01 -8.22350502e-01
-1.43015072e-01 1.11777568e+00 2.50043988e-01 2.44631290e-01
-2.70523094e-02 3.63833129e-01 -1.47468972e+00 1.45493776e-01
8.90324175e-01 7.67356038e-01 9.05541182e-02 4.01594073e-01
-3.59410703e-01 9.85112965e-01 -3.36714536e-01 -3.21974218e-01
1.60207391e-01 -4.10125226e-01 -1.40569293e+00 4.98456061e-01
-1.96779460e-01 2.87189454e-01 -3.52820046e-02 9.84938085e-01
1.31121829e-01 2.90927082e-01 7.19657660e-01 -1.34266734e+00
-3.09537709e-01 1.20257461e+00 -6.98853791e-01 8.32268670e-02
1.02825850e-01 -4.32832062e-01 1.37435472e+00 -5.46517372e-01
7.18699336e-01 8.27214062e-01 3.34223509e-01 4.06594038e-01
-1.37618577e+00 -5.85092902e-01 3.65295559e-01 2.54686773e-01
-1.89354253e+00 -5.94831645e-01 7.85679281e-01 -5.84298015e-01
7.02462971e-01 3.83752257e-01 4.00807530e-01 5.53427100e-01
-6.48569047e-01 6.34476066e-01 8.46390069e-01 -6.30773365e-01
5.05660951e-01 2.64763951e-01 5.77454865e-01 4.83734071e-01
3.56441468e-01 -3.52261484e-01 -2.95631886e-01 -1.16812676e-01
5.26321352e-01 -1.45057574e-01 -1.66190922e-01 -6.24673784e-01
-1.07310486e+00 6.93059623e-01 3.20480198e-01 3.45812857e-01
-2.47982383e-01 -1.96977034e-01 1.61829144e-01 1.57572091e-01
5.86116076e-01 6.01892769e-01 -7.09638596e-01 1.01868855e-02
-8.23621154e-01 -3.32948826e-02 9.28251863e-01 1.04618776e+00
1.07547188e+00 -7.25084007e-01 7.93504715e-02 9.16208863e-01
3.31625491e-01 -2.26476744e-01 1.58697829e-01 -9.43121731e-01
3.97836357e-01 1.24123979e+00 -2.59292163e-02 -9.74342585e-01
-4.00198638e-01 -4.60570037e-01 -6.50040567e-01 -2.55784631e-01
3.21898788e-01 -8.73361528e-02 -8.41717839e-01 1.46318722e+00
6.47391737e-01 3.70391190e-01 1.90483496e-01 5.71476400e-01
7.87201941e-01 2.53943831e-01 1.33358371e-02 -7.44598687e-01
1.27244830e+00 -8.99398386e-01 -9.49719191e-01 -1.42168999e-01
8.98944438e-01 -5.75127363e-01 6.06047332e-01 3.14630389e-01
-8.12893748e-01 -1.74456075e-01 -1.00681996e+00 9.65845883e-02
-3.75050962e-01 1.73940048e-01 1.02908432e+00 5.20097435e-01
-8.27520251e-01 4.27722752e-01 -1.03587925e+00 -4.87691551e-01
4.96480435e-01 8.26824665e-01 -4.20795619e-01 1.07858934e-01
-8.57034802e-01 5.94452858e-01 8.40474010e-01 1.74490362e-01
-3.68104517e-01 -2.49344990e-01 -7.48532772e-01 1.63225874e-01
1.01860404e+00 -2.72777528e-01 1.01525855e+00 -7.36053050e-01
-1.29462290e+00 1.05787992e+00 -3.64523917e-01 -3.42043966e-01
1.36784494e-01 -2.13155732e-01 -3.24569494e-01 2.51423299e-01
1.42959967e-01 5.17899096e-01 3.13325912e-01 -1.68133748e+00
-6.96943283e-01 -4.64934498e-01 1.44133404e-01 1.88164771e-01
-4.01869774e-01 4.72120643e-02 -8.32979202e-01 -2.27860823e-01
6.80040061e-01 -8.95623684e-01 -5.73804736e-01 -5.62132537e-01
-6.36515796e-01 -6.58927381e-01 6.78688109e-01 -1.42484307e-01
1.44017828e+00 -1.90941048e+00 2.08175942e-01 4.95803475e-01
7.54911244e-01 2.30957698e-02 3.48598510e-01 2.48204172e-01
-1.96439266e-01 2.10318059e-01 -4.02389079e-01 -2.91572839e-01
-2.30945438e-01 4.55066323e-01 -9.13655832e-02 4.07961637e-01
2.68289834e-01 6.89109564e-01 -1.07715702e+00 -8.50956142e-01
9.71034542e-02 4.98403050e-02 -5.33964753e-01 3.37760776e-01
-2.75143951e-01 4.81445104e-01 -3.09745669e-01 4.56350446e-01
5.96992791e-01 -6.93244338e-01 9.54100430e-01 -4.83582243e-02
1.55913264e-01 3.25087070e-01 -1.49823868e+00 1.49364638e+00
-1.89595129e-02 3.26370835e-01 -8.20965245e-02 -1.35817063e+00
1.06194472e+00 5.16744554e-01 8.66483450e-01 7.37544522e-02
2.25513712e-01 1.55187160e-01 1.93077147e-01 -6.03965521e-01
2.81011015e-01 2.80575573e-01 4.10218015e-02 5.17122328e-01
9.11278427e-02 4.21887666e-01 4.60142493e-01 3.44440669e-01
1.22331893e+00 1.18967853e-01 5.90765893e-01 -2.29170784e-01
5.18423498e-01 2.14181393e-01 9.27263081e-01 4.04590040e-01
4.89611477e-02 4.41746026e-01 7.38841236e-01 -3.30729216e-01
-5.23404002e-01 -7.96382487e-01 -6.80161566e-02 9.62893426e-01
2.88870752e-01 -8.22410643e-01 -7.09572375e-01 -1.06571054e+00
-4.40194875e-01 3.88872325e-01 -5.74051380e-01 2.81808358e-02
-5.87185562e-01 -1.00149524e+00 2.14553744e-01 3.58464926e-01
1.18327759e-01 -8.48676682e-01 -3.28170478e-01 1.61499128e-01
-3.11099678e-01 -1.27047467e+00 6.79214820e-02 7.12399065e-01
-8.01621318e-01 -1.51682699e+00 7.50457123e-02 -8.12402725e-01
1.06783020e+00 1.07222587e-01 1.16795683e+00 2.38127217e-01
2.50398610e-02 -9.25729871e-02 -7.58104801e-01 8.97752717e-02
-2.26984084e-01 6.69405758e-01 -4.41052802e-02 2.09841833e-01
8.62126768e-01 -8.08952868e-01 -2.00090721e-01 4.52442884e-01
-6.38984025e-01 -1.09396279e-02 5.54337859e-01 7.30489552e-01
6.34495437e-01 5.54693401e-01 5.29871941e-01 -1.64092660e+00
2.12925777e-01 -7.30130553e-01 -4.61697906e-01 4.22365814e-01
-9.49568689e-01 1.98101014e-01 4.57834035e-01 -4.48802471e-01
-8.73832822e-01 7.18035340e-01 1.76061988e-01 -3.83163303e-01
-4.68852401e-01 6.06048465e-01 -5.45865417e-01 3.50440651e-01
6.33666098e-01 -3.28514397e-01 -4.87121880e-01 -5.76349497e-01
5.98913491e-01 8.51295173e-01 5.50435901e-01 -7.50352979e-01
8.41948926e-01 4.73591477e-01 -1.46170214e-01 -4.31528360e-01
-1.00977468e+00 -7.96014905e-01 -1.33590126e+00 9.76281688e-02
6.95436954e-01 -8.01124632e-01 -7.21830487e-01 -1.51989803e-01
-8.83642673e-01 -1.89381763e-01 -2.88736165e-01 2.54779696e-01
-3.59501213e-01 4.50192541e-01 -4.38126862e-01 -8.73348713e-01
-1.17007736e-02 -1.00588989e+00 7.69389927e-01 1.02836885e-01
-4.77581948e-01 -8.89946401e-01 4.14110832e-02 4.92171586e-01
-4.16025341e-01 2.14371398e-01 8.43773842e-01 -1.14261115e+00
-6.59651518e-01 -8.23808983e-02 -2.01592639e-01 -2.04403594e-01
4.99049425e-01 -1.96508411e-02 -1.05732751e+00 -6.53186142e-02
-3.06441545e-01 -2.52412289e-01 6.96022034e-01 -6.00885078e-02
1.17387962e+00 -2.48920009e-01 -8.66407752e-01 4.54443932e-01
1.17757308e+00 4.48683500e-01 4.30978805e-01 7.85166398e-02
9.76933241e-01 8.19227934e-01 7.27357686e-01 1.73151225e-01
5.17280161e-01 7.91176975e-01 1.28672123e-01 -6.25843704e-02
9.29411426e-02 -9.59878564e-02 -1.00019716e-01 9.87771869e-01
-1.02441840e-01 -1.92181304e-01 -1.17722988e+00 5.90369046e-01
-2.18973351e+00 -6.11261368e-01 -3.11072946e-01 2.07312965e+00
1.25584424e+00 2.19616577e-01 1.01693839e-01 7.58834302e-01
9.10766006e-01 -3.05985838e-01 -3.83379221e-01 1.13959223e-01
2.35812098e-01 3.06787401e-01 2.88203746e-01 5.80581903e-01
-1.30906427e+00 1.19335628e+00 6.10405064e+00 7.55508304e-01
-7.30088413e-01 1.20004281e-01 5.84247649e-01 1.74030080e-01
-3.59350555e-02 5.00915289e-01 -7.83947229e-01 7.00602084e-02
9.36123669e-01 1.15805268e-01 2.51736492e-01 8.09858441e-01
-5.64066414e-03 -1.31175697e-01 -1.38608289e+00 8.83975744e-01
-2.19127625e-01 -1.05932057e+00 -2.31159955e-01 2.56788552e-01
6.78362787e-01 -4.64239478e-01 -5.70876360e-01 1.92799211e-01
8.07040751e-01 -9.12751436e-01 2.35282645e-01 2.44073793e-01
7.10092843e-01 -9.58490789e-01 7.00289309e-01 5.11988938e-01
-1.32363999e+00 1.79555770e-02 -2.34219447e-01 -2.00343683e-01
-1.13111064e-01 7.61472285e-01 -1.05594778e+00 7.01025188e-01
6.27063990e-01 8.20573509e-01 -6.33895576e-01 8.29970956e-01
-3.32651317e-01 8.17932129e-01 -3.84330720e-01 2.16883838e-01
-2.36235440e-01 -1.90935716e-01 2.05365747e-01 9.75866079e-01
-2.62422353e-01 3.82446229e-01 6.21590316e-01 7.22486973e-01
-7.11565539e-02 1.50485754e-01 -3.32956403e-01 -2.12779373e-01
8.64632785e-01 1.59944320e+00 -1.45528126e+00 -2.55803972e-01
-1.00040577e-01 8.63379836e-01 7.78456450e-01 2.27543160e-01
-5.65166593e-01 -3.70239913e-01 2.35854790e-01 7.62766674e-02
2.81590670e-01 -1.74974531e-01 -3.78883004e-01 -1.21131599e+00
-2.44640499e-01 -5.81362545e-01 6.99443102e-01 -2.58493125e-01
-1.14625347e+00 6.80139244e-01 8.96017253e-03 -1.01206470e+00
-3.25791806e-01 -3.04564238e-01 -1.82058245e-01 3.49870324e-01
-1.08912492e+00 -1.12705290e+00 -2.56331414e-01 4.18744057e-01
4.00277972e-01 -2.74617579e-02 9.54604447e-01 5.03775239e-01
-1.02004850e+00 6.40303373e-01 -1.27069309e-01 4.66799527e-01
5.25324404e-01 -1.49247694e+00 1.16036655e-02 8.45647693e-01
5.19664466e-01 7.58667409e-01 5.56195319e-01 -5.75396180e-01
-9.99253750e-01 -9.94232297e-01 1.10928166e+00 -5.68885565e-01
5.34165442e-01 -5.97836196e-01 -9.05804157e-01 8.67701054e-01
2.81201340e-02 1.48884475e-01 1.01877356e+00 6.67388022e-01
-2.29738265e-01 -2.08051205e-01 -9.65538561e-01 3.76714975e-01
1.22486508e+00 -2.07142949e-01 -4.76241559e-01 2.90207803e-01
8.03093910e-01 -2.34264597e-01 -1.10611105e+00 5.27497172e-01
3.64728421e-02 -6.18724763e-01 7.61615336e-01 -5.82660973e-01
2.12411091e-01 -6.18071675e-01 -8.05652738e-02 -8.05067241e-01
-3.11198443e-01 -6.65360034e-01 -5.49568653e-01 1.75461781e+00
6.04719818e-01 -5.37941396e-01 9.58250225e-01 5.45776725e-01
8.70695412e-02 -8.49438131e-01 -5.50899327e-01 -3.75926852e-01
-3.58122438e-01 -3.58273774e-01 5.32487035e-01 1.48703420e+00
3.41833532e-01 9.08754051e-01 -4.95853424e-02 5.29930353e-01
6.09677970e-01 3.63505870e-01 7.23761380e-01 -1.74172592e+00
-3.37779224e-01 -1.68996975e-01 -3.56590420e-01 -9.13688004e-01
3.14414948e-01 -9.85443234e-01 3.60003114e-02 -1.46091473e+00
3.25329691e-01 -1.11161983e+00 -3.46024662e-01 7.83305049e-01
-4.77899730e-01 2.12889463e-01 -2.50684768e-01 5.77084303e-01
-8.71175945e-01 1.31268635e-01 6.91270232e-01 -5.69956601e-02
-4.86246854e-01 1.69779062e-01 -9.87535596e-01 6.73345327e-01
6.59380615e-01 -6.45325601e-01 -7.75081396e-01 9.26303305e-03
3.50027561e-01 -2.51865387e-01 -1.29923016e-01 -7.00161219e-01
4.46396649e-01 -2.48069867e-01 2.28715867e-01 -6.97355688e-01
1.06756628e-01 -9.69187021e-01 3.30746919e-01 -7.55666867e-02
-4.62325722e-01 -4.21343416e-01 -2.97456980e-01 5.84202468e-01
-1.80314437e-01 -2.44852990e-01 6.78863227e-01 -2.43128210e-01
-6.02751672e-01 2.75668532e-01 -1.20407991e-01 -1.14783570e-02
1.21575999e+00 -1.30328745e-01 4.42562532e-03 3.95255536e-02
-1.25776350e+00 4.05853093e-01 3.09381515e-01 2.18341976e-01
2.45722651e-01 -1.14746177e+00 -4.13071275e-01 -2.84825312e-03
1.79100960e-01 6.32996798e-01 -2.86629945e-01 7.20085204e-01
-2.98034251e-01 1.81020558e-01 4.40293014e-01 -6.68187320e-01
-1.31671321e+00 8.23953629e-01 1.51318997e-01 -5.94058931e-01
-5.55115104e-01 8.36404502e-01 4.11602966e-02 -5.66408813e-01
2.55143642e-01 -7.29652047e-02 -6.74041331e-01 2.87679523e-01
1.82395682e-01 2.36432046e-01 2.25379407e-01 -6.41959786e-01
-4.00940418e-01 3.39333206e-01 -2.91217208e-01 1.77583784e-01
1.42753029e+00 -3.20352882e-01 -5.29332817e-01 6.52532160e-01
9.03279305e-01 1.67576130e-02 -8.05615902e-01 -3.92136723e-01
7.58719206e-01 -2.56051511e-01 -7.90306553e-02 -5.53760111e-01
-1.26061153e+00 4.81835008e-01 2.54138261e-01 5.56409180e-01
1.18121469e+00 4.82804418e-01 4.33260828e-01 4.35810715e-01
3.05997878e-01 -8.74977052e-01 -2.26397425e-01 2.29589283e-01
1.44823283e-01 -1.13685691e+00 1.82006285e-01 -1.20258009e+00
-3.89577836e-01 8.03724408e-01 7.99340248e-01 1.47254646e-01
6.90051973e-01 4.23490494e-01 1.63947165e-01 -5.26076376e-01
-8.55345368e-01 -4.63563502e-01 1.22589275e-01 5.21008015e-01
5.79290807e-01 1.72659770e-01 -3.00270259e-01 7.57409453e-01
-1.42113015e-01 -2.67200202e-01 3.45027864e-01 8.73408198e-01
-3.38589787e-01 -1.56418467e+00 -3.56276751e-01 4.93875265e-01
-4.38221395e-01 1.35995165e-01 -6.48805618e-01 4.98299688e-01
5.05627871e-01 1.25956416e+00 1.69786140e-01 -5.69546163e-01
-2.08164956e-02 1.00990206e-01 3.47939104e-01 -1.12364209e+00
-3.58201265e-01 2.59521931e-01 2.51386434e-01 -2.30232850e-01
-8.38425934e-01 -5.96710265e-01 -1.46709394e+00 3.32811661e-02
-1.04435921e+00 6.59417033e-01 6.24340028e-02 1.05209863e+00
3.15021694e-01 5.06365657e-01 6.94805562e-01 -3.85718048e-01
-3.72257055e-05 -9.49492693e-01 -5.05207598e-01 3.33904982e-01
-4.04813923e-02 -9.02844787e-01 -2.97745168e-01 4.37214911e-01] | [9.274344444274902, 8.521838188171387] |
30e10b55-2c8f-4e4b-aec2-27e677134211 | deep-impression-audiovisual-deep-residual | 1609.05119 | null | http://arxiv.org/abs/1609.05119v1 | http://arxiv.org/pdf/1609.05119v1.pdf | Deep Impression: Audiovisual Deep Residual Networks for Multimodal Apparent Personality Trait Recognition | Here, we develop an audiovisual deep residual network for multimodal apparent
personality trait recognition. The network is trained end-to-end for predicting
the Big Five personality traits of people from their videos. That is, the
network does not require any feature engineering or visual analysis such as
face detection, face landmark alignment or facial expression recognition.
Recently, the network won the third place in the ChaLearn First Impressions
Challenge with a test accuracy of 0.9109. | ['Rob Van Lier', 'Umut Güçlü', 'Yağmur Güçlütürk', 'Marcel A. J. van Gerven'] | 2016-09-16 | null | null | null | null | ['personality-trait-recognition'] | ['computer-vision'] | [-4.71572399e-01 1.69674203e-01 2.16743246e-01 -9.08887923e-01
-3.12768549e-01 -1.80551603e-01 3.46926630e-01 -5.07296324e-01
-4.95460838e-01 3.09508443e-01 1.02607869e-01 4.96479303e-01
5.24631813e-02 -4.67042923e-02 -1.33928701e-01 -4.77972746e-01
-2.76334316e-01 2.83116758e-01 -7.40906060e-01 3.55435126e-02
-2.91112289e-02 4.68356997e-01 -1.39260268e+00 1.82369292e-01
2.92162567e-01 1.73391271e+00 -6.03429973e-01 8.19126964e-01
5.18556714e-01 9.21158910e-01 -3.79670560e-01 -7.38339245e-01
1.05939180e-01 -6.11246526e-02 -7.22194612e-01 5.38917243e-01
9.23535824e-01 -7.49380171e-01 -6.14157200e-01 9.89921570e-01
8.36005449e-01 1.30070999e-01 6.05458379e-01 -1.38304567e+00
-8.14456522e-01 -1.38895452e-01 -6.72944903e-01 -2.01156110e-01
6.28259122e-01 2.23465025e-01 8.29930425e-01 -1.45287728e+00
3.85459125e-01 1.47872245e+00 7.68929958e-01 8.21209729e-01
-1.01216042e+00 -6.34782791e-01 -2.86083668e-01 2.43541092e-01
-1.60915017e+00 -9.21227217e-01 8.21170509e-01 -3.61254573e-01
6.13604724e-01 1.19490974e-01 6.55325770e-01 1.48195314e+00
-5.33904531e-04 8.31992328e-01 1.21930707e+00 2.30875120e-01
-2.03319788e-01 3.77070010e-01 -4.73936386e-02 1.20586681e+00
-2.99537539e-01 -2.33518705e-02 -5.29501498e-01 -5.35051338e-02
8.25212181e-01 -1.42993644e-01 1.32186040e-01 1.34723380e-01
-8.36353660e-01 4.11703497e-01 3.12189549e-01 -1.94240678e-02
-5.90746880e-01 4.90570962e-02 6.52983963e-01 5.79136670e-01
2.72812665e-01 3.12024981e-01 -1.37749806e-01 -3.06166887e-01
-9.25818741e-01 7.94340670e-02 4.41744179e-01 6.40158951e-01
4.46267366e-01 4.81408209e-01 -1.85644448e-01 1.04937387e+00
4.08358842e-01 5.83463430e-01 4.08338994e-01 -1.28706765e+00
-1.44868255e-01 5.55888772e-01 -7.86404833e-02 -1.40976727e+00
-7.37389684e-01 -7.15440959e-02 -1.16585553e+00 1.22378819e-01
2.50889540e-01 -4.47590142e-01 -7.03037143e-01 1.49537301e+00
-1.51646495e-01 7.84145221e-02 -3.58040668e-02 9.78978217e-01
1.51047039e+00 3.23597223e-01 -1.70307234e-01 -1.79397669e-02
1.38175929e+00 -7.42417514e-01 -5.68479538e-01 -2.41426334e-01
-1.01054505e-01 -3.90576899e-01 9.23867404e-01 6.62231922e-01
-1.41203737e+00 -8.90878797e-01 -8.55507672e-01 -4.70662080e-02
-1.44977063e-01 5.26135027e-01 7.75858521e-01 5.94061792e-01
-1.36425745e+00 5.43451905e-01 -3.49441767e-01 -3.94593567e-01
6.57296240e-01 6.73802733e-01 -8.73304963e-01 3.12647939e-01
-9.23618913e-01 3.40583324e-01 -2.86479235e-01 7.10256934e-01
-1.05150425e+00 -2.42532834e-01 -8.02134633e-01 3.13320994e-01
-1.00438267e-01 -7.74473310e-01 1.02746177e+00 -1.46138012e+00
-1.88854253e+00 1.31120741e+00 -2.75065243e-01 -8.04109797e-02
5.67983568e-01 -1.88031271e-01 -6.90260053e-01 4.13972199e-01
-1.84146479e-01 9.90633845e-01 1.29644692e+00 -8.42250764e-01
-1.69056386e-01 -4.99034733e-01 -3.13686639e-01 -8.42391551e-02
-6.04379535e-01 5.56528807e-01 -5.30107975e-01 -1.25885800e-01
2.98347417e-02 -9.28353548e-01 7.83421174e-02 -6.06101304e-02
-5.66428244e-01 -5.07691085e-01 5.28826356e-01 -7.61482418e-01
3.92308593e-01 -2.36522794e+00 -6.63807499e-04 3.65543395e-01
6.27390802e-01 -4.32928726e-02 -3.73346716e-01 -1.26610711e-01
-6.42243177e-02 -3.25998664e-02 4.98494089e-01 -8.45443547e-01
4.19067949e-01 -3.77658159e-01 -1.09509677e-01 6.55334413e-01
4.41955686e-01 1.11426127e+00 -5.21423876e-01 -4.04645771e-01
-7.35346526e-02 6.82737172e-01 -5.42437196e-01 5.02126098e-01
4.87068743e-01 5.40512502e-01 -3.76717672e-02 8.90169501e-01
7.18169868e-01 -2.31278628e-01 2.29327798e-01 -2.28144810e-01
2.46844098e-01 -1.96909830e-01 -7.96405494e-01 1.17153323e+00
-1.20908573e-01 9.55505371e-01 3.52577001e-01 -4.72595453e-01
1.34802794e+00 6.43932104e-01 6.73821032e-01 -6.23732090e-01
4.31891352e-01 -4.64864336e-02 -1.89238861e-01 -5.25368810e-01
5.34326196e-01 -9.84359160e-02 -2.34023646e-01 -5.59315011e-02
2.98281461e-01 3.60917628e-01 -2.28184491e-01 1.72297098e-02
7.62878716e-01 -2.78770208e-01 -7.77995959e-03 -8.97282436e-02
7.77015626e-01 -8.01402867e-01 9.20882881e-01 1.62356079e-01
-5.16461849e-01 5.03554761e-01 8.46863687e-01 -6.57271981e-01
-9.36903775e-01 -8.64091098e-01 4.27541845e-02 1.05803466e+00
-3.70506734e-01 -5.89490652e-01 -5.74999690e-01 -7.41931140e-01
1.52652888e-02 -1.95048198e-01 -8.15535188e-01 -1.67552680e-01
-1.52732387e-01 3.09399404e-02 9.80788887e-01 6.48860097e-01
7.21078038e-01 -1.22190893e+00 -1.93959013e-01 -2.82339305e-01
9.42793563e-02 -1.35093641e+00 -4.22891796e-01 -5.87619901e-01
-3.70237559e-01 -8.20159137e-01 -9.22354281e-01 -9.07106280e-01
7.92336881e-01 -2.41280004e-01 9.93484974e-01 -1.91293508e-02
-6.92565888e-02 6.63286030e-01 1.82260796e-01 -8.99930969e-02
2.50820071e-01 -2.95530528e-01 6.41789973e-01 5.55757761e-01
5.01861989e-01 -5.23291886e-01 -5.78458548e-01 5.44419229e-01
4.69839573e-03 -4.07683194e-01 5.60363948e-01 9.38624978e-01
2.97123492e-01 -2.33128648e-02 9.07414734e-01 -1.65828630e-01
8.24159801e-01 6.87258318e-02 -3.13186556e-01 6.46878257e-02
-1.19671650e-01 -6.93897128e-01 7.02781022e-01 -3.32871169e-01
-9.16884065e-01 2.51021743e-01 -3.88419956e-01 -7.74839222e-01
-3.31499338e-01 4.42346752e-01 -2.58894295e-01 -2.57922977e-01
1.09765582e-01 2.89552093e-01 6.82621077e-02 -3.86822641e-01
-5.87733276e-03 8.93355370e-01 9.70662355e-01 -2.40437582e-01
3.79599273e-01 2.64217466e-01 8.52477625e-02 -1.26232862e+00
-7.78225601e-01 -1.87467173e-01 -7.67536700e-01 -4.67720628e-01
9.30325866e-01 -1.12540734e+00 -1.92358267e+00 7.79404104e-01
-9.73640621e-01 2.02130638e-02 2.39331692e-01 4.37226862e-01
-7.67273605e-01 5.37121236e-01 -8.25242996e-01 -9.09274459e-01
-6.87108755e-01 -8.66285086e-01 1.07152152e+00 3.39419723e-01
-4.56401289e-01 -6.83949530e-01 -1.95671141e-01 3.93759102e-01
3.00759614e-01 1.49807721e-01 3.39620560e-01 -3.74576539e-01
3.69673893e-02 -6.03981435e-01 -5.72638750e-01 5.61326206e-01
-7.08447620e-02 2.83603165e-02 -1.42120373e+00 -4.83479232e-01
-1.94898218e-01 -1.02977371e+00 7.04498231e-01 3.05574864e-01
1.53319311e+00 -4.99286443e-01 2.75393486e-01 9.50630248e-01
7.08370090e-01 -1.63646683e-01 8.12909424e-01 -2.16294423e-01
9.16497588e-01 8.24149847e-01 2.72052050e-01 6.75061524e-01
5.07319629e-01 7.70912707e-01 2.80011982e-01 -1.73615105e-02
3.29307348e-01 -1.98955253e-01 5.15472233e-01 5.27889907e-01
-2.20594138e-01 2.79333770e-01 -7.52063811e-01 2.00167924e-01
-1.59326017e+00 -1.23408937e+00 1.42927155e-01 1.97313726e+00
3.16408187e-01 -1.60833910e-01 6.63025737e-01 -2.77250946e-01
5.19587934e-01 1.79657042e-01 -6.17222667e-01 -5.29646516e-01
-1.26364514e-01 -2.91473120e-01 -9.35799628e-02 1.27447888e-01
-1.25710928e+00 1.15579879e+00 6.65858173e+00 3.37815545e-02
-1.36154902e+00 -2.92869002e-01 8.18904996e-01 -3.43271405e-01
4.11455482e-01 -7.70063519e-01 -8.26982915e-01 2.94670641e-01
1.02864301e+00 1.01515330e-01 4.91130710e-01 9.09606159e-01
1.75768316e-01 4.17086482e-01 -1.06694019e+00 1.63886142e+00
3.80940139e-01 -7.59777665e-01 -3.75999242e-01 2.94245183e-01
2.48890087e-01 -3.90624315e-01 5.33235133e-01 4.97569591e-01
-1.48156315e-01 -1.68637490e+00 4.94415849e-01 1.03438997e+00
1.02190733e+00 -9.47024703e-01 8.27101707e-01 -1.25562251e-01
-9.06662285e-01 -3.39148074e-01 -7.74002075e-01 -1.10533491e-01
1.11564785e-01 1.18306763e-01 -8.82634997e-01 -5.38228499e-03
7.69083142e-01 1.08016455e+00 -7.87977695e-01 7.52784669e-01
5.65537401e-02 5.07202923e-01 1.44850656e-01 2.55323108e-02
1.57805249e-01 -4.78845388e-02 4.30649996e-01 1.18023634e+00
3.76188874e-01 -7.26869982e-03 2.35999987e-01 8.29842985e-01
-3.90711218e-01 -1.17212823e-02 -5.98694861e-01 -4.47599262e-01
9.92719159e-02 1.82961714e+00 -1.55809626e-01 -3.42616200e-01
-3.15010279e-01 1.24851835e+00 3.41931164e-01 4.40510869e-01
-6.58969820e-01 -2.66178101e-01 8.67212057e-01 -2.54278064e-01
1.32581756e-01 -1.10802628e-01 -8.79461020e-02 -1.18162107e+00
-9.01879966e-02 -9.85537410e-01 1.03157826e-01 -1.01737797e+00
-1.65863049e+00 9.21753407e-01 -7.42169440e-01 -1.05306685e+00
-4.50006694e-01 -8.45932424e-01 -8.14459085e-01 8.63435984e-01
-8.13945949e-01 -1.38047206e+00 -6.06138885e-01 8.81024182e-01
8.75036940e-02 -7.99424410e-01 1.00820148e+00 4.03190881e-01
-9.78496730e-01 1.20536625e+00 -3.03227603e-01 6.66479707e-01
9.61555958e-01 -1.24339163e+00 2.61174977e-01 2.02180848e-01
-4.05389488e-01 7.31092751e-01 5.72820306e-01 -4.69359875e-01
-1.49088168e+00 -8.93350363e-01 1.09270930e+00 -3.65370899e-01
5.10540605e-01 -5.88227630e-01 -6.92004859e-01 6.81486607e-01
3.45031649e-01 6.47569522e-02 7.43869901e-01 4.93170381e-01
-3.29823554e-01 -4.61875677e-01 -1.17698610e+00 3.49531233e-01
7.37803102e-01 -7.39853084e-01 -1.06961258e-01 2.12906614e-01
3.47748562e-03 -6.38728067e-02 -1.19057965e+00 3.38286161e-01
1.10785186e+00 -8.50965142e-01 1.12068188e+00 -1.01955259e+00
6.91115439e-01 2.19227538e-01 -2.43668288e-01 -1.17578065e+00
-6.60026848e-01 -7.15362072e-01 -1.33543581e-01 1.31012130e+00
6.50518090e-02 -3.30530941e-01 1.03978300e+00 8.92916083e-01
7.90968835e-02 -7.44976163e-01 -7.79442310e-01 -7.13382542e-01
-2.94602901e-01 -1.78638566e-02 5.76885283e-01 1.00215602e+00
2.82904625e-01 7.13100255e-01 -1.10658741e+00 -4.94390093e-02
6.28771305e-01 6.74038427e-03 1.05603337e+00 -1.49166131e+00
-1.28782094e-01 -6.36261225e-01 -9.41013038e-01 -8.44513714e-01
6.79690242e-01 -6.91433370e-01 -3.72349620e-01 -1.11932290e+00
3.40338945e-01 2.35449463e-01 -3.64272594e-01 4.78784472e-01
1.02244414e-01 5.92423022e-01 1.68817759e-01 -2.10326210e-01
-9.42728460e-01 9.81001675e-01 1.14365542e+00 -2.93115586e-01
-1.50090251e-02 1.61335751e-01 -8.98048699e-01 1.01588726e+00
8.80530000e-01 1.92022145e-01 -3.12726423e-02 -1.99209386e-03
2.43453994e-01 3.13903272e-01 8.54885697e-01 -8.33431840e-01
1.06049329e-01 1.62297070e-01 1.30929470e+00 -4.45746839e-01
9.80014920e-01 -8.03826749e-01 -3.20055395e-01 4.69009504e-02
-3.25491965e-01 2.74731755e-01 1.62631243e-01 1.87712416e-01
-1.66351050e-01 1.59661517e-01 8.26910734e-01 -7.40979332e-03
-8.82319570e-01 5.63585460e-01 -3.76095593e-01 -2.86434978e-01
6.93036020e-01 -2.02542409e-01 -3.51485312e-01 -9.66693938e-01
-1.02561677e+00 2.46247783e-01 2.90687174e-01 2.78801560e-01
1.13199055e+00 -1.63912284e+00 -7.65606701e-01 4.60032016e-01
-1.12802945e-02 -8.76371384e-01 5.45918345e-01 1.27133679e+00
1.20295845e-01 3.97296876e-01 -7.83815026e-01 -4.95876342e-01
-1.71682060e+00 3.33671898e-01 7.05154240e-01 2.52704352e-01
-4.44678307e-01 1.08926916e+00 1.96057469e-01 -6.00848973e-01
3.76728594e-01 5.50982118e-01 -6.31798506e-01 5.29099219e-02
8.69965315e-01 5.47040761e-01 -3.02651972e-01 -1.07446194e+00
-4.44967240e-01 4.80421454e-01 -1.38366282e-01 -1.15612008e-01
1.41370642e+00 -4.81441393e-02 -2.65574932e-01 3.67453068e-01
1.36153495e+00 4.60385531e-02 -1.38958752e+00 -5.70582040e-02
-2.04140618e-01 -6.61963522e-01 5.62035516e-02 -8.56573105e-01
-1.16987193e+00 1.16163683e+00 7.33845770e-01 1.54477712e-02
1.04937482e+00 -2.04730973e-01 7.22550869e-01 7.24835217e-01
6.08580373e-02 -1.20142448e+00 4.04922158e-01 8.16885352e-01
1.17655611e+00 -1.46970475e+00 -2.21652225e-01 8.82965624e-02
-1.16702640e+00 1.05334103e+00 8.71688187e-01 -2.41616502e-01
6.16254926e-01 -2.40446493e-01 1.54106990e-01 -5.34390807e-01
-8.33593488e-01 4.69751954e-02 9.81422961e-01 6.08114839e-01
6.40603304e-01 1.79090500e-01 3.61778408e-01 1.01293766e+00
-4.83286798e-01 7.00456053e-02 3.00344586e-01 4.89201508e-02
-3.97994965e-01 -4.03694630e-01 1.11659588e-02 4.97198254e-01
-4.83210266e-01 1.28952488e-01 -1.00823331e+00 3.26129526e-01
-2.89107949e-01 9.65907991e-01 1.29420266e-01 -8.86692405e-01
4.08358693e-01 2.48803586e-01 4.47864801e-01 -1.90593541e-01
-5.75218856e-01 1.14642195e-01 1.91886038e-01 -8.14221501e-01
-6.00402467e-02 -8.95287514e-01 -8.88742447e-01 -8.50507975e-01
5.40754318e-01 -1.47587329e-01 2.31878787e-01 6.78806901e-01
4.86258388e-01 1.62574351e-01 1.06047511e+00 -1.14764297e+00
-5.32409966e-01 -1.17837703e+00 -9.59226429e-01 3.06750208e-01
3.60062480e-01 -4.94322151e-01 -4.30633128e-03 -2.22891331e-01] | [13.490335464477539, 1.7255851030349731] |
fa7f3c94-7770-408b-9c0e-1562e601a4fe | style-guided-domain-adaptation-for-face | 2203.14565 | null | https://arxiv.org/abs/2203.14565v2 | https://arxiv.org/pdf/2203.14565v2.pdf | Style-Guided Domain Adaptation for Face Presentation Attack Detection | Domain adaptation (DA) or domain generalization (DG) for face presentation attack detection (PAD) has attracted attention recently with its robustness against unseen attack scenarios. Existing DA/DG-based PAD methods, however, have not yet fully explored the domain-specific style information that can provide knowledge regarding attack styles (e.g., materials, background, illumination and resolution). In this paper, we introduce a novel Style-Guided Domain Adaptation (SGDA) framework for inference-time adaptive PAD. Specifically, Style-Selective Normalization (SSN) is proposed to explore the domain-specific style information within the high-order feature statistics. The proposed SSN enables the adaptation of the model to the target domain by reducing the style difference between the target and the source domains. Moreover, we carefully design Style-Aware Meta-Learning (SAML) to boost the adaptation ability, which simulates the inference-time adaptation with style selection process on virtual test domain. In contrast to previous domain adaptation approaches, our method does not require either additional auxiliary models (e.g., domain adaptors) or the unlabeled target domain during training, which makes our method more practical to PAD task. To verify our experiments, we utilize the public datasets: MSU-MFSD, CASIA-FASD, OULU-NPU and Idiap REPLAYATTACK. In most assessments, the result demonstrates a notable gap of performance compared to the conventional DA/DG-based PAD methods. | ['Seong-Whan Lee', 'Kyungseo Min', 'Woo-Jeoung Nam', 'Young-Eun Kim'] | 2022-03-28 | null | null | null | null | ['face-presentation-attack-detection'] | ['computer-vision'] | [ 2.07429826e-01 -7.14187801e-01 -2.82986283e-01 -2.95073420e-01
-6.94691062e-01 -9.32074070e-01 5.98286569e-01 -2.71472126e-01
-4.85032275e-02 5.30911863e-01 6.54455321e-03 -4.24798280e-01
-2.72459667e-02 -5.81867754e-01 -3.31839085e-01 -8.36050928e-01
2.64349908e-01 1.48341924e-01 2.45132044e-01 -4.32833195e-01
3.14654559e-01 7.83474624e-01 -1.20449519e+00 3.19832802e-01
9.45729196e-01 9.39559817e-01 -3.24387215e-02 5.46595275e-01
-1.24296747e-01 9.15003493e-02 -9.41635132e-01 -5.41742861e-01
1.99173585e-01 -5.74120522e-01 -4.90156442e-01 -2.99609620e-02
3.28437775e-01 -6.16286337e-01 -3.24558526e-01 9.53680933e-01
1.00957334e+00 4.33134846e-02 7.76520610e-01 -1.38039255e+00
-5.87132871e-01 2.32607629e-02 -7.63992131e-01 3.31152111e-01
3.62405211e-01 3.55543077e-01 3.69414210e-01 -8.95996511e-01
5.73915720e-01 1.51984620e+00 4.28105772e-01 8.50456715e-01
-1.05846989e+00 -1.21289027e+00 2.28752643e-01 5.27304292e-01
-1.17733109e+00 -3.63384128e-01 1.37383723e+00 -2.28373766e-01
3.06922913e-01 1.58583245e-03 6.38988763e-02 1.78578854e+00
-2.36211628e-01 7.33772695e-01 1.33417785e+00 -6.00639343e-01
4.14458424e-01 4.72627878e-01 8.97643417e-02 2.86120594e-01
1.09646782e-01 1.26555458e-01 -5.30156553e-01 -4.96677309e-01
8.62439990e-01 -3.58185738e-01 -1.30901575e-01 -2.87260383e-01
-6.04414761e-01 7.73311555e-01 -1.60847262e-01 1.43229246e-01
3.05062160e-02 -8.38425219e-01 6.18936539e-01 5.03666937e-01
4.01910067e-01 3.54500532e-01 -6.82975590e-01 -6.06142282e-02
-5.43885827e-01 7.20807463e-02 6.80464923e-01 1.00269961e+00
5.73443830e-01 3.60613912e-01 -3.21950436e-01 1.08777750e+00
1.65807605e-02 7.27240205e-01 4.45425212e-01 -4.67057794e-01
6.59393132e-01 3.99995983e-01 -5.34663126e-02 -8.90383482e-01
-1.87203437e-01 -2.98296869e-01 -7.78647482e-01 2.23176062e-01
5.52796543e-01 -3.78136188e-01 -9.68839109e-01 1.88832200e+00
5.50037622e-01 5.12217164e-01 1.92450568e-01 6.40857160e-01
7.93320775e-01 4.90105689e-01 3.44715774e-01 -3.54172260e-01
1.40257156e+00 -4.92953300e-01 -6.60058439e-01 -1.49442777e-01
4.23818171e-01 -1.15547645e+00 1.38405156e+00 7.24445701e-01
-6.52796388e-01 -8.00770581e-01 -1.14081132e+00 3.81356061e-01
-3.38622332e-01 1.56569019e-01 3.02337945e-01 1.31293237e+00
-5.21269441e-01 3.41016382e-01 -2.44451046e-01 -5.28416216e-01
4.38883454e-01 4.47361797e-01 -3.37569773e-01 -1.22945353e-01
-1.40783691e+00 5.89510143e-01 3.97176653e-01 -3.61525536e-01
-9.78416026e-01 -6.72806680e-01 -6.44459426e-01 -1.27010852e-01
5.58925569e-01 -1.73987776e-01 1.02304113e+00 -8.13446581e-01
-2.02718663e+00 7.25478232e-01 -1.52440563e-01 -1.37858987e-01
2.55803734e-01 -1.10986836e-01 -9.91162717e-01 2.38094896e-01
-4.58086967e-01 1.98267072e-01 1.45426965e+00 -1.25961280e+00
-4.26641941e-01 -3.33585769e-01 -3.16987410e-02 2.87411697e-02
-8.67909968e-01 4.09986347e-01 -4.86412823e-01 -1.15102422e+00
-2.74827927e-01 -8.02469492e-01 2.53330886e-01 -1.53266951e-01
-3.25359046e-01 -1.12231664e-01 1.41214263e+00 -9.13245142e-01
1.52899134e+00 -2.53095055e+00 -7.64462948e-02 3.26608330e-01
-1.75563976e-01 1.00472653e+00 -4.60107416e-01 2.05050960e-01
-3.20466518e-01 1.71922743e-02 -1.30333304e-01 -2.01476008e-01
1.23066856e-02 8.91254768e-02 -4.11634088e-01 2.76315957e-01
3.70351702e-01 3.10720921e-01 -5.76468468e-01 -6.01277590e-01
2.88828373e-01 4.41430569e-01 -7.51989961e-01 5.78087687e-01
-1.49122804e-01 7.68573940e-01 -6.03490293e-01 9.12836313e-01
1.28836322e+00 3.11605453e-01 1.53508887e-01 -3.45885456e-01
2.03922436e-01 1.15200281e-01 -1.50369096e+00 1.45170057e+00
-5.04772961e-01 1.09871291e-01 1.03594646e-01 -8.60642493e-01
1.23816860e+00 3.33836198e-01 2.90731996e-01 -7.61144400e-01
2.07484469e-01 9.31633785e-02 6.20979816e-02 -3.28837931e-01
6.03333339e-02 5.23784570e-02 -1.97572395e-01 3.39316308e-01
2.40692541e-01 5.39666414e-02 -9.36941132e-02 1.92157030e-02
7.20990300e-01 6.79373145e-02 3.74924868e-01 -4.98902909e-02
1.00356150e+00 -4.42752898e-01 8.88613880e-01 6.00426257e-01
-5.65693915e-01 4.67449546e-01 5.62238395e-01 -1.35111243e-01
-1.04398823e+00 -1.31635642e+00 -9.90049765e-02 1.17812586e+00
1.71636015e-01 -3.08034807e-01 -9.45503235e-01 -9.93538618e-01
-2.58242965e-01 7.68783748e-01 -3.42629403e-01 -5.92518628e-01
-7.38716304e-01 -8.33922803e-01 7.30197966e-01 5.11317730e-01
8.64280403e-01 -8.61592591e-01 1.86800912e-01 1.19487792e-02
1.13822654e-01 -1.26564312e+00 -6.48910105e-01 -2.06530958e-01
-6.39928341e-01 -8.62230420e-01 -5.07278621e-01 -7.89870381e-01
3.75561833e-01 5.02572134e-02 6.83954835e-01 -1.79152727e-01
-1.55761272e-01 3.86320412e-01 -5.79465508e-01 -4.60090727e-01
-4.28192049e-01 3.59345339e-02 2.72731781e-01 3.53701055e-01
5.99405468e-01 -5.64059258e-01 -5.25862098e-01 6.95293903e-01
-7.72546887e-01 -4.95094597e-01 6.32588685e-01 9.19807732e-01
2.06549302e-01 2.54139960e-01 7.83268452e-01 -9.16578054e-01
7.99312115e-01 -4.42589372e-01 -4.73338127e-01 3.99700582e-01
-4.35483158e-01 -1.17338680e-01 8.37300420e-01 -1.18315852e+00
-1.53362894e+00 -2.35772401e-01 -2.19212487e-01 -4.61631060e-01
-5.75863779e-01 -6.60560131e-02 -1.12823081e+00 -3.14008504e-01
7.70397246e-01 1.80740565e-01 -7.09002316e-02 -6.17928743e-01
1.24171779e-01 8.17131817e-01 5.17174602e-01 -9.04815674e-01
1.26445520e+00 9.78177488e-02 -6.07671626e-02 -7.77897418e-01
-4.88966614e-01 -2.57072538e-01 -6.89596236e-01 1.84432462e-01
4.92101759e-01 -6.86272681e-01 -5.68825901e-01 9.03144121e-01
-1.13575983e+00 -1.45189330e-01 1.80236474e-01 3.66230428e-01
-2.64970869e-01 7.89444089e-01 -5.94121099e-01 -8.21580529e-01
-3.87228638e-01 -1.16555035e+00 7.37181187e-01 3.39580804e-01
-1.40346661e-01 -1.04457736e+00 -1.37892142e-01 2.16198176e-01
4.46400255e-01 1.66580766e-01 1.20525253e+00 -1.19223130e+00
-1.05465464e-02 -1.02852695e-01 -2.29643866e-01 7.19895840e-01
3.12617838e-01 -1.68884233e-01 -1.20974028e+00 -4.82370824e-01
3.26257020e-01 -1.32081255e-01 2.92699963e-01 1.29206657e-01
1.36301529e+00 -2.46636987e-01 -1.75425217e-01 7.31041789e-01
9.95675445e-01 6.50676668e-01 6.58206880e-01 4.08110648e-01
6.34303927e-01 3.77327323e-01 8.40034664e-01 6.92992985e-01
-1.87774599e-01 1.02856100e+00 -2.78013833e-02 -1.74286395e-01
-2.33651027e-01 -3.29777181e-01 5.76680005e-01 5.82267456e-02
2.74738848e-01 -3.25139195e-01 -5.60607195e-01 1.68966115e-01
-1.12288308e+00 -8.09717596e-01 3.01200360e-01 2.34647775e+00
9.08209860e-01 3.01005423e-01 4.35540378e-01 4.15225923e-01
9.24652219e-01 2.48792320e-02 -7.58748412e-01 -6.61421716e-01
-2.11330011e-01 6.12893462e-01 1.14429496e-01 3.29629421e-01
-1.18698394e+00 1.11273801e+00 5.42620897e+00 1.47534239e+00
-1.18222356e+00 1.44486204e-01 4.96630698e-01 2.56716669e-01
-1.04463741e-01 -2.02399090e-01 -1.08039844e+00 5.80959380e-01
7.04662025e-01 7.11456463e-02 4.09893811e-01 9.16096628e-01
7.38743097e-02 3.96233171e-01 -9.16414618e-01 1.10643888e+00
1.86626211e-01 -7.18965769e-01 2.21138835e-01 -8.32059886e-03
4.64177459e-01 -9.87175584e-01 5.85021853e-01 4.76439625e-01
9.69246328e-02 -5.88381469e-01 1.58334836e-01 -9.82728451e-02
1.10281551e+00 -1.09498227e+00 6.38282359e-01 6.02236688e-02
-9.67646182e-01 -3.71981919e-01 -2.69634247e-01 4.77099806e-01
-1.82919562e-01 3.89464200e-01 -1.04176342e+00 6.06417894e-01
5.33238351e-01 3.39637458e-01 -6.94544196e-01 7.22475469e-01
-3.28687400e-01 1.11558950e+00 -1.71693966e-01 3.06523830e-01
-2.39381313e-01 2.59121098e-02 7.61147976e-01 1.08019209e+00
3.52787465e-01 1.09495848e-01 2.03481719e-01 5.40416181e-01
1.29907250e-01 2.73285866e-01 -3.77856612e-01 -3.22568081e-02
9.27076340e-01 1.13857996e+00 -2.03117549e-01 -7.12308735e-02
-3.54416400e-01 9.91982341e-01 -7.06629306e-02 4.17109638e-01
-8.63489985e-01 -4.03577894e-01 9.33805764e-01 1.20730706e-01
2.37467319e-01 -1.41433105e-01 -4.03270453e-01 -1.01286924e+00
-1.15798362e-01 -1.40865982e+00 8.22707951e-01 -3.66910100e-01
-1.61192274e+00 6.85026169e-01 1.89995289e-01 -1.48309445e+00
-1.94978625e-01 -7.96343803e-01 -9.29554462e-01 1.03856480e+00
-1.46508753e+00 -1.44961584e+00 -8.37296471e-02 9.80049968e-01
5.81655383e-01 -7.91432858e-01 7.85922527e-01 5.39871991e-01
-8.47083449e-01 1.52551377e+00 -1.28590465e-01 3.25169593e-01
1.33667672e+00 -8.95682573e-01 3.59046847e-01 1.05309200e+00
-3.67706567e-01 7.18977988e-01 7.74485230e-01 -5.98142624e-01
-1.24907148e+00 -8.46783876e-01 2.30690226e-01 -3.29229265e-01
5.55920839e-01 -5.35369337e-01 -1.17199099e+00 1.51109114e-01
-1.38555849e-02 -2.37769827e-01 9.10518110e-01 1.53422609e-01
-8.12288165e-01 -4.22374994e-01 -1.62788200e+00 7.06943572e-01
9.23631787e-01 -4.86191720e-01 -3.23917538e-01 4.69782688e-02
6.28643274e-01 -2.96213210e-01 -6.91104591e-01 5.73774338e-01
3.84875804e-01 -8.13252211e-01 1.16022801e+00 -6.96300864e-01
-9.14836824e-02 -1.76027507e-01 -1.93449497e-01 -1.25275373e+00
-3.49699497e-01 -8.37095797e-01 -2.17727616e-01 2.01683521e+00
4.15366190e-03 -6.27439976e-01 7.30020344e-01 3.46564561e-01
-3.03512327e-02 -2.88015842e-01 -7.53059566e-01 -9.17304456e-01
2.20446348e-01 -3.13655376e-01 1.03370500e+00 9.91582036e-01
-3.11664373e-01 8.78680050e-02 -5.91082156e-01 5.72893679e-01
6.25037372e-01 -1.70160562e-01 1.04169762e+00 -1.13638651e+00
-4.10413206e-01 -3.74166548e-01 -2.19205037e-01 -6.93742275e-01
3.42965901e-01 -3.33516806e-01 -5.31130493e-01 -6.49553478e-01
-5.82761243e-02 -4.49094951e-01 -3.73067111e-01 4.17509913e-01
-3.81808728e-01 -6.00043163e-02 2.40569171e-02 -1.09806649e-01
-1.95839062e-01 2.89200783e-01 1.22965300e+00 -7.86586925e-02
-3.90135705e-01 3.09867203e-01 -9.12096381e-01 6.00475848e-01
9.90196764e-01 -2.95329243e-01 -6.19355381e-01 -6.73411041e-02
-5.27295053e-01 -1.10336378e-01 2.67067254e-01 -9.66129124e-01
-8.08878243e-02 -4.32091892e-01 5.69274008e-01 -4.20757204e-01
3.11052412e-01 -6.42716408e-01 -3.40390325e-01 2.24765673e-01
-5.46867996e-02 -1.59587309e-01 5.86939394e-01 3.20523024e-01
-8.21225643e-02 -1.44914016e-01 1.06191361e+00 3.61724645e-01
-9.92256701e-01 4.62508440e-01 -1.52118817e-01 9.18615982e-02
8.85884285e-01 -2.71656901e-01 -4.06212360e-01 -2.57963985e-01
-5.33479512e-01 -2.18127847e-01 4.56233680e-01 5.21893322e-01
6.48101926e-01 -1.37987304e+00 -7.75840044e-01 7.36535013e-01
2.31870115e-01 -3.45568627e-01 6.52046025e-01 3.64085138e-01
1.80149972e-02 2.44138595e-02 -4.33656812e-01 -3.28231961e-01
-1.57569599e+00 9.53190506e-01 1.25516029e-02 -2.52234101e-01
-1.68939993e-01 7.32059121e-01 3.96515489e-01 -3.72676998e-01
2.23870590e-01 4.06931311e-01 -3.25849861e-01 -4.70015295e-02
7.09987938e-01 3.17878813e-01 5.92153035e-02 -3.84060711e-01
-3.66468906e-01 5.72275102e-01 -5.10940433e-01 -1.40940938e-02
1.01315641e+00 -1.84218213e-02 2.54232138e-01 -2.01184541e-01
1.08228838e+00 2.63561159e-01 -1.26783371e+00 -3.37227017e-01
-3.78156975e-02 -6.44956052e-01 -2.04588830e-01 -1.02109575e+00
-8.52744997e-01 1.08080542e+00 9.66274500e-01 -2.76678026e-01
1.71999204e+00 -3.28941345e-01 7.77517498e-01 1.20748412e-02
3.15827221e-01 -1.17168605e+00 5.85302413e-01 4.16649461e-01
7.33078122e-01 -1.10079432e+00 -5.19371256e-02 -6.45003021e-01
-8.93373013e-01 1.04033816e+00 9.82615530e-01 2.38660157e-01
4.83266354e-01 2.66532958e-01 3.92680131e-02 3.42195183e-01
-4.71709698e-01 8.41980502e-02 4.76828456e-01 1.19372213e+00
2.39516143e-02 -1.83038637e-01 -6.40374273e-02 8.57982159e-01
-2.04536924e-03 -3.37888271e-01 1.69049487e-01 8.07670712e-01
-2.04620257e-01 -1.76305377e+00 -7.30531037e-01 3.27219516e-02
-3.21290165e-01 3.15722190e-02 -4.85093743e-01 7.03006566e-01
3.04351956e-01 1.12985921e+00 -3.55083257e-01 -6.82762325e-01
4.60054517e-01 1.79897383e-01 5.46126783e-01 -5.43011785e-01
-3.96612614e-01 2.13577092e-01 -1.27525195e-01 -1.00676060e-01
1.08750090e-01 -7.57729709e-01 -7.85405278e-01 -4.96065378e-01
-2.73118585e-01 2.79546417e-02 3.77179861e-01 9.23385441e-01
3.01895320e-01 1.37760639e-01 1.12795508e+00 -4.39725846e-01
-6.42135501e-01 -8.79222929e-01 -5.30656159e-01 6.40170097e-01
6.19677342e-02 -1.00243068e+00 -3.87434453e-01 -1.08031422e-01] | [13.152069091796875, 1.2140318155288696] |
c1512730-32fb-4bf7-bae9-ef71b335dd02 | pengi-an-audio-language-model-for-audio-tasks | 2305.11834 | null | https://arxiv.org/abs/2305.11834v1 | https://arxiv.org/pdf/2305.11834v1.pdf | Pengi: An Audio Language Model for Audio Tasks | In the domain of audio processing, Transfer Learning has facilitated the rise of Self-Supervised Learning and Zero-Shot Learning techniques. These approaches have led to the development of versatile models capable of tackling a wide array of tasks, while delivering state-of-the-art performance. However, current models inherently lack the capacity to produce the requisite language for open-ended tasks, such as Audio Captioning or Audio Question & Answering. We introduce Pengi, a novel Audio Language Model that leverages Transfer Learning by framing all audio tasks as text-generation tasks. It takes as input, an audio recording, and text, and generates free-form text as output. The input audio is represented as a sequence of continuous embeddings by an audio encoder. A text encoder does the same for the corresponding text input. Both sequences are combined as a prefix to prompt a pre-trained frozen language model. The unified architecture of Pengi enables open-ended tasks and close-ended tasks without any additional fine-tuning or task-specific extensions. When evaluated on 22 downstream tasks, our approach yields state-of-the-art performance in several of them. Our results show that connecting language models with audio models is a major step towards general-purpose audio understanding | ['Huaming Wang', 'Rita Singh', 'Benjamin Elizalde', 'Soham Deshmukh'] | 2023-05-19 | null | null | null | null | ['audio-captioning'] | ['audio'] | [ 5.91301262e-01 2.62347609e-01 -1.26759127e-01 -3.63382578e-01
-1.50869608e+00 -5.78396559e-01 7.01105893e-01 1.15596019e-01
-2.55245596e-01 4.77603346e-01 6.18325770e-01 -3.10006827e-01
2.72848904e-01 -6.13381505e-01 -8.11072409e-01 -2.13941112e-01
-2.22950354e-01 3.59464377e-01 3.17395717e-01 -2.11826682e-01
-2.81178206e-02 -3.09721768e-01 -1.70214772e+00 8.94615710e-01
2.57883757e-01 1.19138634e+00 1.93090409e-01 1.03479111e+00
-3.64078879e-01 9.26306903e-01 -5.42947948e-01 -1.96437597e-01
-1.29517049e-01 -4.38249946e-01 -1.02573919e+00 -2.76535034e-01
2.73792922e-01 -3.63681346e-01 -4.65968192e-01 4.11615670e-01
8.15612197e-01 2.74185032e-01 5.73658347e-01 -1.33978724e+00
-7.02182293e-01 9.91430461e-01 -1.03323221e-01 1.23810463e-01
7.48769760e-01 1.35840237e-01 1.46664798e+00 -1.05988514e+00
3.78376544e-01 1.25887823e+00 7.28164077e-01 6.38110995e-01
-1.36185718e+00 -7.37666428e-01 -4.80097458e-02 2.01523706e-01
-1.02923751e+00 -9.03482318e-01 6.58707678e-01 -4.68030423e-01
1.22935128e+00 -3.67766172e-02 3.36946517e-01 1.41295409e+00
1.28297433e-01 8.90792608e-01 5.47725379e-01 -6.25222981e-01
3.07842642e-01 -6.26281695e-03 -1.29724294e-01 2.29604855e-01
-6.70327306e-01 5.67297637e-02 -1.19670379e+00 -2.45908901e-01
3.64801466e-01 -4.01893198e-01 -1.92390919e-01 1.28726155e-01
-1.22700596e+00 8.00473630e-01 6.89196736e-02 2.04944059e-01
-1.48058772e-01 2.87089825e-01 7.99113691e-01 5.58420062e-01
4.50046778e-01 3.57726067e-01 -3.85592252e-01 -5.07877827e-01
-1.08579135e+00 3.04100484e-01 7.31982231e-01 8.89084995e-01
6.03466630e-01 2.40337357e-01 -2.42039636e-01 9.18344080e-01
2.31792182e-01 1.73304349e-01 7.51878798e-01 -8.33590031e-01
7.72646725e-01 1.39843121e-01 1.87048465e-02 -3.11296463e-01
-1.69064030e-01 -1.84848920e-01 -3.96137714e-01 -3.85602340e-02
2.46540844e-01 -3.23375642e-01 -7.45709419e-01 1.87834394e+00
5.59387133e-02 4.02168125e-01 1.64507195e-01 5.46587408e-01
8.06217372e-01 1.29811645e+00 1.28676623e-01 1.45982310e-01
1.26184845e+00 -9.88144815e-01 -5.83424330e-01 -3.74150217e-01
5.78180254e-01 -9.48735297e-01 1.32700777e+00 4.05927837e-01
-1.17565024e+00 -8.40221286e-01 -1.00649071e+00 -3.35412234e-01
-1.70955688e-01 -9.00195241e-02 2.40965649e-01 2.96374202e-01
-1.04507744e+00 4.03855026e-01 -6.64527833e-01 -2.78087199e-01
6.12427182e-02 1.40730888e-01 -2.79735416e-01 2.22941577e-01
-1.39926314e+00 4.16688651e-01 3.90086204e-01 -3.29389960e-01
-1.14975762e+00 -1.19563890e+00 -9.37785268e-01 2.67424852e-01
3.23846698e-01 -5.99019587e-01 1.90325665e+00 -9.28274274e-01
-1.77822697e+00 6.67994559e-01 -7.20542856e-04 -6.43441200e-01
2.22535476e-01 -4.84427333e-01 -6.32266581e-01 2.53575712e-01
2.32556447e-01 1.10545039e+00 1.18592656e+00 -8.02216232e-01
-6.90600395e-01 1.13478996e-01 -3.21891978e-02 6.14241250e-02
-4.90056545e-01 2.77651876e-01 -2.61441797e-01 -7.84351766e-01
-4.95346069e-01 -8.05477560e-01 2.83760548e-01 4.85717542e-02
-2.36576468e-01 -3.21707755e-01 9.51128602e-01 -5.11652887e-01
1.39680552e+00 -2.63425231e+00 -5.18978648e-02 -3.27029973e-01
-1.84664816e-01 1.15884125e-01 -6.05095327e-01 9.43260908e-01
-3.50977063e-01 1.38356944e-03 -8.48276690e-02 -5.70105970e-01
1.66495159e-01 -8.51018447e-03 -9.49747086e-01 -1.76701397e-01
6.33522749e-01 8.13452601e-01 -1.05716038e+00 -3.65154028e-01
3.60519588e-02 5.42085767e-01 -8.39195549e-01 6.10514283e-01
-6.08892262e-01 2.84091234e-01 -9.06219259e-02 2.52544522e-01
-1.39655709e-01 -1.30039901e-01 -1.47467092e-01 1.25603274e-01
-6.64965659e-02 8.95361722e-01 -1.17891955e+00 2.08860254e+00
-8.12853754e-01 7.63864517e-01 3.66361439e-02 -8.17763865e-01
8.82873237e-01 8.87003422e-01 3.12817693e-01 -4.39360499e-01
-6.60514534e-02 1.60563558e-01 -1.18806787e-01 -5.17781973e-01
4.46406364e-01 -3.86681229e-01 -4.52105075e-01 8.55067313e-01
6.44901097e-01 -3.01027566e-01 1.55889625e-02 2.64718622e-01
1.17169642e+00 2.94925362e-01 1.68935314e-01 3.43582749e-01
3.50403786e-01 -1.99127823e-01 9.37848762e-02 5.61886668e-01
6.98189139e-02 8.59082639e-01 3.27142149e-01 -1.59783408e-01
-1.03990829e+00 -1.19408798e+00 2.14875937e-01 1.94676590e+00
-6.05997443e-01 -8.74716520e-01 -6.18104577e-01 -1.60252199e-01
-6.35207742e-02 8.45295966e-01 -5.50854445e-01 -3.00990045e-01
-2.66822606e-01 -9.53928679e-02 9.46893036e-01 7.79797316e-01
-4.79238387e-03 -1.30666912e+00 -5.27175725e-01 6.23622358e-01
-5.03130853e-01 -1.09973550e+00 -6.41520739e-01 4.89986986e-01
-6.58136785e-01 -5.50133646e-01 -7.07185566e-01 -8.81410420e-01
-1.29664972e-01 1.42573118e-01 1.12800527e+00 -3.76157999e-01
-2.86781937e-01 4.52725559e-01 -4.83146727e-01 -5.88178694e-01
-6.51459098e-01 4.02094543e-01 4.83912677e-02 2.40418434e-01
7.77010694e-02 -8.50311220e-01 -2.23622382e-01 -1.86621044e-02
-1.07383037e+00 -1.32678688e-01 3.44232500e-01 7.68826246e-01
4.08966005e-01 -3.73431951e-01 1.06253421e+00 -4.86161619e-01
9.08006251e-01 -5.46317279e-01 -1.17298737e-01 1.33042596e-02
-2.15382185e-02 2.33290359e-01 7.81891763e-01 -6.86931014e-01
-9.93687093e-01 -2.80686151e-02 -3.01239729e-01 -3.58944356e-01
-2.46542245e-01 6.40136242e-01 -4.67940830e-02 6.37724817e-01
8.23622823e-01 1.85710013e-01 5.60417250e-02 -5.99106252e-01
7.16973364e-01 1.17495084e+00 9.23139691e-01 -5.33162832e-01
6.76050425e-01 1.10232033e-01 -5.97190738e-01 -1.02416515e+00
-1.07173467e+00 -5.34034133e-01 -4.69397962e-01 -1.57521218e-01
7.89655745e-01 -1.18768620e+00 -3.55323672e-01 2.49651313e-01
-1.14065313e+00 -6.15104973e-01 -5.42888820e-01 2.94123888e-01
-9.78027165e-01 6.47323206e-02 -5.41388214e-01 -9.08594072e-01
-4.64006960e-01 -8.49769056e-01 1.37340999e+00 -1.61982685e-01
-7.24944174e-01 -7.92060256e-01 3.01909477e-01 2.08578631e-01
4.32314605e-01 -2.85082191e-01 1.19531274e+00 -9.44739282e-01
-3.53454590e-01 -7.57981539e-02 1.01834014e-01 3.70252341e-01
-1.49180636e-01 -5.48563786e-02 -1.46617901e+00 -8.11447501e-02
-3.21283102e-01 -1.00685573e+00 7.94980109e-01 -5.41891716e-03
1.14058959e+00 -1.22680135e-01 1.13172587e-02 4.34001327e-01
9.33500767e-01 1.34634569e-01 4.44174618e-01 9.40340981e-02
7.10907429e-02 5.34846008e-01 3.57443273e-01 5.39288402e-01
3.44188571e-01 6.50718987e-01 2.41694361e-01 2.71498591e-01
-3.23630661e-01 -8.65013063e-01 8.03189874e-01 1.03641343e+00
4.26244020e-01 -1.81429952e-01 -9.71362352e-01 7.80235469e-01
-1.67371738e+00 -1.07959437e+00 2.58795828e-01 2.07464147e+00
1.25455368e+00 2.56118774e-01 2.86766469e-01 5.41932344e-01
4.04749453e-01 2.55662590e-01 -4.12769467e-01 -4.26640391e-01
4.88844424e-01 4.55595464e-01 -3.38815331e-01 5.43234706e-01
-1.06112969e+00 8.87710154e-01 6.80894995e+00 6.35806918e-01
-1.29167771e+00 2.10796654e-01 1.28911853e-01 -2.98953712e-01
-3.25321913e-01 -9.51370150e-02 -7.61668682e-01 3.33381653e-01
1.70358920e+00 -4.23337072e-01 4.42051888e-01 6.28231645e-01
2.38699898e-01 1.71127260e-01 -1.52961671e+00 9.95554030e-01
4.41091973e-03 -1.29029429e+00 1.98968634e-01 -3.28081578e-01
2.34923586e-01 9.18178186e-02 1.22680299e-01 8.13775003e-01
6.23829067e-02 -9.55791593e-01 9.77574348e-01 2.29035959e-01
1.29532933e+00 -5.67954540e-01 2.86191612e-01 3.48110020e-01
-1.21914852e+00 -2.45260566e-01 -1.89282205e-02 -4.20331776e-01
4.72705632e-01 2.54762620e-01 -1.05587029e+00 4.33954716e-01
5.96669912e-01 5.40740848e-01 -2.11819306e-01 8.65033090e-01
-1.78375453e-01 1.15355515e+00 -3.90773624e-01 2.00510874e-01
3.25072229e-01 3.95809114e-01 3.54843795e-01 1.42034042e+00
2.60066539e-01 -1.60644591e-01 2.15292498e-01 6.74026072e-01
-1.74719393e-01 2.15747982e-01 -7.02107131e-01 -4.66225863e-01
6.04746759e-01 8.82525980e-01 -2.46748403e-01 -4.23328519e-01
-5.70688844e-01 8.31882179e-01 1.06237307e-01 2.79790521e-01
-7.21605718e-01 -7.96764076e-01 6.45100594e-01 2.24463910e-01
2.99629450e-01 -1.64557084e-01 -4.78941537e-02 -9.82370973e-01
-1.03050038e-01 -8.93015027e-01 3.62890512e-01 -1.08026433e+00
-1.19639242e+00 6.76978052e-01 -1.69839151e-02 -1.21969330e+00
-9.60973084e-01 -4.21036661e-01 -8.24517071e-01 7.33773947e-01
-1.49404120e+00 -1.06272292e+00 -1.11271799e-01 5.38362026e-01
7.92604864e-01 -3.45730752e-01 1.29917014e+00 4.84813869e-01
-7.57451504e-02 5.34543216e-01 -1.45957381e-01 1.97178692e-01
1.17387223e+00 -1.02925622e+00 6.17400467e-01 4.74408418e-01
5.04850745e-01 1.88893750e-01 6.71533287e-01 -2.53278673e-01
-1.21314919e+00 -1.20087957e+00 1.13744712e+00 -4.17956173e-01
1.21717596e+00 -7.60630608e-01 -9.49388921e-01 8.32021236e-01
3.26450467e-01 -8.14600289e-02 1.13806164e+00 1.90921381e-01
-6.50866449e-01 -1.35865211e-01 -4.33005840e-01 3.24763268e-01
8.08752060e-01 -1.19949734e+00 -1.05265403e+00 1.28972098e-01
1.27649915e+00 -2.35443383e-01 -6.84414923e-01 -1.16898336e-01
5.80644846e-01 -5.84148526e-01 8.44139695e-01 -8.92934561e-01
7.92992413e-01 -1.11569665e-01 -2.06851125e-01 -1.30799353e+00
-3.79183590e-02 -1.11616385e+00 -1.86098948e-01 1.44528258e+00
6.34203792e-01 -2.30582461e-01 3.84222835e-01 2.10491970e-01
-4.63573903e-01 -4.46422130e-01 -9.08615351e-01 -7.59069681e-01
-6.12899400e-02 -1.16273952e+00 4.05719191e-01 5.72057366e-01
4.97458428e-01 9.16228771e-01 -4.43374813e-01 -1.20242521e-01
2.10039794e-01 -5.00968583e-02 8.70575726e-01 -1.16644883e+00
-5.22556365e-01 -1.11738481e-01 -1.89175397e-01 -1.12869036e+00
3.24219078e-01 -1.11129749e+00 1.95188567e-01 -1.34650624e+00
-1.98515967e-01 -1.84188858e-01 -9.57143605e-02 8.03497195e-01
4.23170850e-02 2.90847063e-01 4.18560803e-01 1.54195102e-02
-5.34701169e-01 5.93652427e-01 5.99882483e-01 -1.34788811e-01
-4.40923303e-01 1.03317365e-01 -7.54160345e-01 5.58291256e-01
7.06048429e-01 -4.20121819e-01 -6.47217870e-01 -6.52492285e-01
3.08759362e-01 2.45743424e-01 2.92464674e-01 -1.22153032e+00
2.82714963e-01 2.49083355e-01 1.88537370e-02 -4.02349919e-01
6.60462499e-01 -5.32598197e-01 -1.64910778e-01 -2.31039431e-02
-8.18214536e-01 -5.20220101e-02 3.53084505e-01 5.84349096e-01
-6.50842130e-01 -3.12205940e-01 5.99430144e-01 1.09915972e-01
-5.53729713e-01 7.75889233e-02 -5.88375211e-01 4.28171635e-01
6.45145357e-01 9.59313475e-03 5.19862808e-02 -8.31442714e-01
-1.02412283e+00 9.30094942e-02 -1.35020033e-01 8.47266257e-01
4.23518956e-01 -1.26428747e+00 -7.39384413e-01 3.94840449e-01
3.05642903e-01 -3.16738002e-02 -8.27593170e-03 3.98230940e-01
1.62862107e-01 6.51776731e-01 1.74547092e-03 -6.33979142e-01
-1.02187133e+00 3.50709021e-01 9.74316671e-02 -1.98134661e-01
-6.31820977e-01 8.57117176e-01 5.09719551e-02 -3.43340993e-01
6.46911263e-01 -5.28564274e-01 2.04446167e-01 3.57133925e-01
1.00435460e+00 3.51026803e-02 1.65370047e-01 -1.78030878e-01
-1.22390902e-02 8.44876021e-02 2.88832411e-02 -7.65664876e-01
1.53964055e+00 -1.01276608e-02 3.40323597e-01 1.04383039e+00
1.26269090e+00 -2.06601173e-01 -1.26470685e+00 -3.83299083e-01
6.68488592e-02 1.90230776e-02 7.79136717e-02 -8.47977579e-01
-3.73095125e-01 1.46067214e+00 2.23679215e-01 1.70905530e-01
9.92703378e-01 1.93454653e-01 1.02864933e+00 3.65120739e-01
2.68826395e-01 -1.04827344e+00 5.19185662e-01 8.58461261e-01
9.02383327e-01 -8.96557868e-01 -5.76623738e-01 -6.27830625e-02
-7.15385973e-01 1.21543837e+00 3.24404776e-01 1.60691157e-01
5.29546440e-01 6.89200819e-01 1.18899614e-01 2.87106007e-01
-1.48583889e+00 -9.32700112e-02 2.36260086e-01 7.68396199e-01
6.98314965e-01 -2.57034630e-01 2.90535390e-01 9.63674664e-01
-5.39478898e-01 3.23998988e-01 4.06318724e-01 9.12392855e-01
-5.73235810e-01 -1.13281929e+00 -3.31809610e-01 2.32252494e-01
-4.88069862e-01 -2.21788898e-01 -3.01591396e-01 4.38826650e-01
-2.26754412e-01 9.99085903e-01 3.23456377e-01 -3.89469624e-01
2.99438536e-01 9.64071870e-01 1.86280578e-01 -1.06365776e+00
-6.95607722e-01 1.39451817e-01 1.49924427e-01 -3.37059647e-01
-1.14614718e-01 -4.16043252e-01 -1.34476924e+00 2.95304507e-01
-9.87461060e-02 2.15917617e-01 5.79936147e-01 8.59291732e-01
5.68744123e-01 6.81078315e-01 5.32461286e-01 -1.03161836e+00
-8.35293114e-01 -1.15046751e+00 -2.91769862e-01 2.49477550e-01
5.95177233e-01 -3.27121288e-01 -2.30745375e-01 6.10489547e-01] | [15.25653076171875, 5.036684513092041] |
2098391f-c7da-4729-9d8f-0aac7f40e1cf | large-scale-fast-and-accurate-shot-boundary | 1705.03281 | null | http://arxiv.org/abs/1705.03281v2 | http://arxiv.org/pdf/1705.03281v2.pdf | Large-scale, Fast and Accurate Shot Boundary Detection through Spatio-temporal Convolutional Neural Networks | Shot boundary detection (SBD) is an important pre-processing step for video
manipulation. Here, each segment of frames is classified as either sharp,
gradual or no transition. Current SBD techniques analyze hand-crafted features
and attempt to optimize both detection accuracy and processing speed. However,
the heavy computations of optical flow prevents this. To achieve this aim, we
present an SBD technique based on spatio-temporal Convolutional Neural Networks
(CNN). Since current datasets are not large enough to train an accurate SBD
CNN, we present a new dataset containing more than 3.5 million frames of sharp
and gradual transitions. The transitions are generated synthetically using
image compositing models. Our dataset contain additional 70,000 frames of
important hard-negative no transitions. We perform the largest evaluation to
date for one SBD algorithm, on real and synthetic data, containing more than
4.85 million frames. In comparison to the state of the art, we outperform
dissolve gradual detection, generate competitive performance for sharp
detections and produce significant improvement in wipes. In addition, we are up
to 11 times faster than the state of the art. | ['Sung-Ho Bae', 'Mohamed Hefeeda', 'Ahmed Selim', 'Ahmed Hassanien', 'Wojciech Matusik', 'Mohamed Elgharib'] | 2017-05-09 | null | null | null | null | ['camera-shot-boundary-detection'] | ['computer-vision'] | [ 0.20726568 -0.41458553 -0.04468744 0.1425563 -0.3434368 -0.387747
0.61516553 0.02001347 -0.5070615 0.4498191 -0.09967913 -0.3070643
0.44013706 -0.77441365 -0.8100338 -0.21078812 -0.3795757 0.11544651
1.029617 -0.28334644 0.28930447 0.5851329 -1.5575094 0.43817058
0.5824959 1.2088269 0.19968182 1.0188242 0.02065549 0.9151676
-0.78254306 0.00864787 0.6702474 -0.5000167 -0.5902199 0.10938137
0.7604435 -0.796013 -0.65126395 0.8534111 0.49368876 0.32928967
0.48977315 -1.2057412 -0.55983263 0.2332058 -0.79875314 0.7839258
0.62736195 0.6438905 0.6119691 -1.1639247 1.0804774 1.2098635
0.82604027 0.73504215 -1.0691645 -0.5394524 -0.0079902 0.22413406
-1.1967 -0.37427706 0.64512235 -0.64063776 1.2372575 -0.15688178
1.0439556 1.2993325 0.34384274 0.9631544 0.52377826 -0.41295248
0.12872823 -0.6080804 -0.07907153 0.9151041 0.33669233 0.27816233
-0.6623662 0.41658935 1.0233225 -0.10351423 -0.2506468 -0.38683987
-1.4490237 0.4183061 0.24223503 0.02913332 -0.35869712 0.25342306
0.677891 0.2298716 0.38568974 0.433532 -0.10675715 -0.50251365
-1.3578972 0.53051627 0.74308735 1.0446597 0.50266695 0.20660101
-0.5113554 0.3999077 -0.16320868 0.31923947 0.23032969 -1.4405771
0.41493997 0.42863318 0.41287786 -1.2485212 -0.33988523 0.07929619
-0.7996152 0.68422663 0.6791323 -0.13655187 -1.2247146 1.2053138
0.24631584 0.48191875 -0.2581737 1.0050623 0.8419904 0.83256483
-0.35307142 -0.10126545 1.1315666 -1.293729 -0.85583484 -0.15010223
0.28143424 -0.91264826 1.083987 0.5604151 -1.4029486 -0.71849775
-1.3154213 -0.08517803 -0.30097717 0.02490935 0.485284 0.31068274
-0.9900688 0.6980414 -1.0351372 -0.27779713 0.569238 0.14138661
-0.3037012 -0.03339387 -1.1121031 0.7857858 0.30143243 0.06732561
-1.2090915 -0.79173106 -0.9926707 -0.11894883 0.57700115 -0.5491916
1.4087572 -0.6690496 -1.4437171 0.8266539 -0.02662231 -0.88237256
0.92724544 -0.4791427 -0.12214573 0.5319984 0.10224961 0.8562987
0.9291033 -0.8539271 -0.821683 0.39919034 0.31464174 -0.19552132
0.07216518 0.19690475 -0.7449864 -0.8434746 -0.27930018 -0.717984
-0.07391129 0.358419 -0.28566456 -0.05692008 1.2285029 -0.56856066
1.1924498 -2.0511398 -0.00932011 -0.4651724 0.31027403 0.64311004
-0.12054901 0.17559718 0.18615976 0.03772374 -0.20770141 -0.5279127
-0.10985798 0.00892975 -0.2027448 0.43452898 0.551771 0.8341211
-1.1038868 -0.5585925 0.675056 0.2767556 -0.6252917 0.27782312
-0.20176697 0.14900188 0.0779356 0.8394665 0.64125514 0.02914307
-0.43599278 -0.14864221 -0.34825698 0.09558835 -1.1361539 1.7022709
-0.04573153 1.2888051 -0.11006598 -0.76286244 0.8243199 0.14532362
0.6308132 -0.7424011 0.11695427 0.04686517 -0.21502542 -0.65520245
0.8137953 0.28483534 0.12741429 0.1400531 0.17291409 -0.22712694
0.81252843 0.17619745 1.4369899 0.45635012 0.07316168 -0.01912101
0.18746327 0.13449097 0.74984473 0.9011591 -0.7023217 1.1223232
0.7931255 -0.8548792 -1.2679294 -0.98425287 0.32018077 0.75817645
0.5223575 -0.36002627 -0.97284484 -0.3896104 -0.2409123 0.30493104
-0.656043 -0.16747595 -0.87354845 -0.41937009 0.68962085 0.64096296
0.7992201 -1.2677938 -1.1379437 0.3706582 -0.09215403 -1.4515353
-0.6980677 -0.14469108 -0.6390174 -1.2971951 -0.78134084 -0.73377883
0.21934709 0.60880154 1.2045782 -0.07560816 -0.8072763 0.11312029
-0.5536824 -0.33599445 -0.31130695 -0.07245734 0.12366779 -0.27018338
0.24713781 -0.23483263 -0.75219077 0.2408352 -0.9638646 0.22640501
0.3218322 0.6647222 0.33693397 -0.19037125 0.02435811 -0.37280574
0.6281293 -0.11252277 -0.64788765 -0.00847609 -0.12372326 -0.17185901
0.52129525 -0.5599307 -1.1233768 0.18474625 0.07416153 -0.967416
-0.26080292 0.07645468 0.52823216 -0.12541103 0.9079421 -0.01258676
-0.02524447 -0.21112853 0.27312928 0.2067217 0.85379314 -0.42198175
0.6792933 0.6258311 -0.07347504 -0.9557865 -0.69341165 -0.55186176
-0.8149255 -0.6621157 0.9799218 -0.7114339 -0.5473811 0.91461843
-1.3996712 -0.7249871 -0.35583675 0.5187866 -0.74137014 0.44483516
-1.0447092 -0.6231533 -0.32115236 -1.209083 1.1509014 0.30963477
-0.22777328 -0.5644398 0.05193445 -0.02783694 0.38466 0.85104656
0.30729488 -0.12266616 -0.70860916 -0.37131432 -0.3946701 0.18369614
0.13858636 0.8687257 -0.6387369 -0.20247054 -0.2962745 -0.31039006
1.1818383 0.6699179 0.9965912 0.18411474 -0.20702522 0.5846444
1.0639335 0.4382235 0.7074456 0.5308686 0.48124325 0.16467746
0.993668 0.66256124 -0.07656944 0.6310082 0.58753544 -0.21966301
-0.5700412 -0.077571 0.5712391 0.34496462 -0.5445803 -0.3585905
-1.091381 0.6083717 -1.829577 -1.4292722 -0.18080525 2.0637589
0.5625341 0.6437306 0.2704162 0.1940379 0.8788372 0.42065725
-0.43257385 -0.36869982 0.02136199 0.31230578 0.45493 0.40149477
-1.5135095 1.2557131 6.7583346 0.6556827 -1.1382064 -0.1952648
0.43071377 -0.29535413 0.4213113 -0.314894 -0.77226275 0.33070207
0.5406965 -0.02763265 0.30032647 0.74355084 0.5205595 -0.36785537
-1.0298545 1.0811285 0.14696403 -1.7968426 -0.05329752 -0.14878273
0.87533545 0.03207495 -0.19210273 0.25323147 0.16915306 -0.8158604
1.1328517 0.60432804 0.74108714 -0.6572701 0.645343 0.24136408
-1.3980457 0.07326936 -0.4065652 -0.30321237 0.38161987 0.549411
-0.5929016 0.16106956 0.9928557 1.0236897 -0.46600792 1.3889784
-0.05580636 0.48440033 -0.35610333 0.06614602 0.54383034 -0.04553579
0.70088387 1.5137537 0.2611815 -0.05606626 0.20651248 0.78125846
-0.12597667 -0.46444663 -0.802216 0.02496324 0.37044787 0.9508517
-1.0193236 -0.62050134 -0.40175363 1.2041756 0.13878883 0.35932583
-1.280209 -0.82295114 0.8091846 0.09447645 0.36871582 -0.730888
0.01747951 -1.0646404 0.06877103 -0.6301308 0.14478561 -0.7309419
-0.9380321 0.40725806 -0.04225904 -1.4810897 -0.10441213 -0.6814389
-0.68542665 0.32431772 -1.3566351 -0.63959324 -0.695318 0.37848192
1.1796472 0.11414623 0.33388022 0.49070057 -0.6523159 0.20566724
-0.07782867 0.45554432 0.76863456 -1.0907303 1.0414762 1.2004853
-0.12707359 0.2639853 0.7837297 -0.631073 -1.1987321 -1.0672665
0.37953418 -0.28798822 0.6330067 -0.37901464 -0.8461313 0.49823552
0.28605133 0.4720181 -0.0130497 -0.5443068 -0.03434166 0.14576335
-0.87159413 0.75803614 1.4346555 -0.19486575 -0.69985646 0.09130576
0.85474515 -0.8963173 -0.59273857 0.44306388 0.45117286 -1.3159535
1.1482083 -0.56373656 0.91768074 -0.47701296 0.2907964 -0.9936412
-0.26269794 -0.9337783 -0.65555936 0.9524415 0.13181803 -0.06955444
0.79961866 0.35783806 -0.336815 -0.7486473 -0.8644818 -1.0298072
-0.32695672 -0.5119235 0.19382593 0.8570237 -0.09198564 -0.07007375
-0.65673333 -0.16339095 0.5776493 0.12638915 0.96721774 -0.8003708
0.1487255 -0.59052795 -0.6740612 -1.3029864 -0.21949606 -0.24999453
0.3339486 -1.6224625 -0.02370333 -0.02603796 -0.01291087 0.385469
-0.17126848 0.49451527 0.28001207 -0.06394071 -0.79063666 0.43725666
1.4025816 -0.27160344 -0.42848432 -0.2824447 0.20170264 0.8076352
0.84953934 -0.21745174 -0.05603098 -0.3804306 -0.03363355 0.04508056
0.44147235 -1.4899052 0.22430182 -0.2728439 0.4574762 -0.691961
0.44075197 -0.2869267 -0.16919209 0.76368374 -0.22283976 -0.02814702
0.30593765 0.5446112 -0.20078306 -0.24525046 0.7891546 -0.317299
-1.234017 0.45688272 -0.8546328 0.24251047 1.3751439 -0.64541626
-0.28603742 -0.2357089 -0.57417834 0.09245953 0.51901 0.6729483
0.9617369 -1.2459508 -0.46669436 0.11900592 -0.12746106 0.35534236
0.08424818 0.75051904 -1.3324 0.1952786 -0.56118494 -0.75326747
-1.1749704 0.75737584 0.3659265 0.08694968 -0.977946 0.8248394
-0.08414181 0.22902049 0.28222868 -0.70272577 -0.0283292 0.14609444
0.64246094 0.6435789 -0.02743072 -0.37216282 -0.17277396 0.55985457
0.04704004 0.04092465 1.1842006 0.20571825 0.25830212 0.26088384
1.0006897 -0.20345378 -1.8867478 0.09240659 -0.20713757 -0.86910707
-0.11723498 -0.1602888 -1.1251199 1.0128318 0.46414575 0.24307184
0.91275984 -0.35885432 1.1064068 0.37600946 0.43742803 -1.3089491
0.55616397 0.7819103 0.76867133 -1.2824485 -0.13221884 -0.43541723
-0.3986224 1.4015044 0.8518428 -0.46031913 0.18989621 0.49840555
-0.16729502 -0.12957709 -0.6805849 -0.17441659 -0.03523061 0.5798315
0.23012145 -0.32526127 -0.02068947 -0.13921104 0.15315452 0.36270118
0.8194593 1.172077 -0.62579185 -0.5044209 -0.36179596 0.32186896
-0.35835797 0.0157878 -0.31872 0.94977605 0.05621835 0.8131275
0.31347063 -0.43898594 0.5312925 -0.25344688 0.51272035 -0.41043934
-0.31546086 -0.24198101 0.01266869 -0.9697036 -0.5151561 -0.576303
-1.278453 -0.53210866 -0.25010768 -0.42128083 0.17240103 0.6612665
0.23235816 0.7675675 0.28504956 -1.540198 -0.0397293 -0.6539495
-0.13794595 0.47187287 0.56163126 -0.86222786 -0.28741926 0.43694076] | [8.50622844696045, 0.10755906999111176] |
e06fce1a-da5c-40f6-b9fd-724771be1adf | on-momentum-based-gradient-methods-for | 2303.03944 | null | https://arxiv.org/abs/2303.03944v1 | https://arxiv.org/pdf/2303.03944v1.pdf | On Momentum-Based Gradient Methods for Bilevel Optimization with Nonconvex Lower-Level | Bilevel optimization is a popular two-level hierarchical optimization, which has been widely applied to many machine learning tasks such as hyperparameter learning, meta learning and continual learning. Although many bilevel optimization methods recently have been developed, the bilevel methods are not well studied when the lower-level problem is nonconvex. To fill this gap, in the paper, we study a class of nonconvex bilevel optimization problems, which both upper-level and lower-level problems are nonconvex, and the lower-level problem satisfies Polyak-Lojasiewicz (PL) condition. We propose an efficient momentum-based gradient bilevel method (MGBiO) to solve these deterministic problems. Meanwhile, we propose a class of efficient momentum-based stochastic gradient bilevel methods (MSGBiO and VR-MSGBiO) to solve these stochastic problems. Moreover, we provide a useful convergence analysis framework for our methods. Specifically, under some mild conditions, we prove that our MGBiO method has a sample (or gradient) complexity of $O(\epsilon^{-2})$ for finding an $\epsilon$-stationary solution of the deterministic bilevel problems (i.e., $\|\nabla F(x)\|\leq \epsilon$), which improves the existing best results by a factor of $O(\epsilon^{-1})$. Meanwhile, we prove that our MSGBiO and VR-MSGBiO methods have sample complexities of $\tilde{O}(\epsilon^{-4})$ and $\tilde{O}(\epsilon^{-3})$, respectively, in finding an $\epsilon$-stationary solution of the stochastic bilevel problems (i.e., $\mathbb{E}\|\nabla F(x)\|\leq \epsilon$), which improves the existing best results by a factor of $O(\epsilon^{-3})$. This manuscript commemorates the mathematician Boris Polyak (1935 -2023). | ['Feihu Huang'] | 2023-03-07 | null | null | null | null | ['bilevel-optimization'] | ['methodology'] | [-3.83920252e-01 -8.83814469e-02 -8.35532025e-02 1.23618104e-01
-1.00010383e+00 -2.51806945e-01 -3.03364843e-01 8.20020437e-02
-6.67292774e-01 1.17312288e+00 -4.90633279e-01 -4.70538735e-01
-7.92486191e-01 -7.42394328e-01 -9.99491751e-01 -1.20358288e+00
-2.54306883e-01 3.73524904e-01 1.88540630e-02 -3.26082230e-01
3.37600142e-01 -3.32912318e-02 -1.26707959e+00 -4.64580625e-01
1.38025093e+00 1.30965626e+00 9.79142413e-02 6.01230621e-01
-1.55932447e-02 2.46356785e-01 -2.15640545e-01 -1.95920661e-01
2.27217078e-01 -5.16281068e-01 -4.10257161e-01 -1.10708542e-01
1.80529192e-01 3.67095262e-01 -9.37559307e-02 1.56327176e+00
4.18230325e-01 3.63274425e-01 5.74783742e-01 -1.12644243e+00
-5.55068314e-01 2.97413230e-01 -9.10159767e-01 1.51612505e-01
-2.04567105e-01 1.44459814e-01 8.25391769e-01 -9.88793373e-01
1.40578642e-01 9.18155074e-01 8.54887784e-01 2.18139336e-01
-8.66590500e-01 -7.67574608e-01 5.36071181e-01 -8.93379189e-03
-1.28379416e+00 2.15170354e-01 6.14658654e-01 -4.36441123e-01
4.41546947e-01 5.76959789e-01 7.73360074e-01 1.49605051e-01
2.21640497e-01 6.57877445e-01 1.20132041e+00 -3.42941850e-01
1.44072339e-01 1.15782484e-01 2.02906117e-01 1.17477071e+00
3.60835612e-01 5.88804223e-02 -1.38443157e-01 -1.41281262e-01
9.01726902e-01 -6.02231435e-02 -4.90855306e-01 -1.09611135e-02
-1.10350943e+00 1.16102076e+00 2.97285169e-01 2.81763852e-01
-1.90382972e-01 3.45737219e-01 1.29029453e-01 1.29926339e-01
5.46072900e-01 5.25398016e-01 -4.72939700e-01 -3.07814986e-01
-7.94276655e-01 2.77087063e-01 7.44800508e-01 8.81582797e-01
7.05123425e-01 1.96060255e-01 -4.18711528e-02 9.29859042e-01
3.82174045e-01 9.28359210e-01 2.00596988e-01 -8.79757762e-01
7.74999082e-01 3.72052521e-01 4.00132269e-01 -1.09796095e+00
-5.84477782e-01 -5.69028139e-01 -1.16577220e+00 -1.04465373e-01
6.19564354e-01 -3.67051899e-01 -4.05197382e-01 1.73098850e+00
4.06833619e-01 1.96441039e-01 -2.35188499e-01 9.96742368e-01
4.14976448e-01 1.15371108e+00 -2.95777977e-01 -9.69009817e-01
1.23267198e+00 -9.85334635e-01 -3.21114391e-01 -2.49260999e-02
4.46382314e-01 -8.52099836e-01 1.48013139e+00 5.32302916e-01
-1.59515822e+00 -1.55615747e-01 -8.53746414e-01 3.34479541e-01
1.78853363e-01 1.69774413e-01 7.97974646e-01 6.22541606e-01
-6.38884842e-01 4.62434113e-01 -6.67310119e-01 4.48621541e-01
2.80928284e-01 5.04364967e-01 6.07219525e-02 9.30627584e-02
-9.39533830e-01 3.77722174e-01 1.38457809e-02 5.82403600e-01
-7.61320949e-01 -8.96311641e-01 -6.47116423e-01 -9.50954333e-02
7.65498400e-01 -2.34144986e-01 7.71210611e-01 -4.19797063e-01
-1.53020120e+00 4.51685220e-01 -2.66376257e-01 5.10443114e-02
7.21186817e-01 4.16689180e-02 -3.71138118e-02 -5.75175770e-02
1.47965491e-01 -1.16327710e-01 5.10131299e-01 -1.15541220e+00
-7.63197005e-01 -5.09496033e-01 7.33462423e-02 1.20725431e-01
-3.98025036e-01 -1.15640894e-01 -5.32020271e-01 -7.42236257e-01
3.23747844e-01 -1.10232139e+00 -3.79059106e-01 -3.68173838e-01
-2.76068747e-01 -2.68507689e-01 2.63406664e-01 -5.64397156e-01
1.39879251e+00 -1.95589137e+00 3.70857954e-01 3.64011526e-01
2.54326552e-01 2.06604600e-01 2.33837143e-01 4.57519339e-03
1.82640120e-01 4.50555116e-01 -4.88462299e-01 -2.17102394e-01
1.67392474e-02 -2.41040029e-02 -6.07913807e-02 7.18361139e-01
-4.24980223e-01 6.38622940e-01 -9.32950139e-01 -1.53272107e-01
-9.08132792e-02 3.34234267e-01 -6.42109811e-01 -2.04415709e-01
-2.24999338e-01 3.53320837e-01 -7.22101867e-01 5.62512934e-01
8.53670418e-01 -4.15100902e-01 -1.37291417e-01 -2.51973867e-01
-4.62446064e-01 -2.27352217e-01 -1.60334909e+00 1.10341632e+00
-7.53815651e-01 2.48610184e-01 5.26584268e-01 -1.31042922e+00
7.84790635e-01 3.30018476e-02 7.22340524e-01 -4.67554301e-01
1.87023863e-01 6.01942241e-01 -4.21440244e-01 -4.19345289e-01
2.09710762e-01 -6.76462054e-01 -1.79203257e-01 4.47568804e-01
-4.33608383e-01 -1.42311051e-01 4.96166497e-01 -1.89642027e-01
7.56228507e-01 -1.87227726e-01 -9.50427279e-02 -5.83477318e-01
6.73888147e-01 -1.83991641e-02 9.22015131e-01 7.14716077e-01
-6.01610765e-02 2.40824640e-01 9.15943384e-01 -3.54905486e-01
-7.80149639e-01 -9.32331800e-01 -4.29822773e-01 1.00255275e+00
6.42257452e-01 -2.00152770e-01 -6.42821193e-01 -5.60713351e-01
-8.49355683e-02 5.08875370e-01 -5.81695437e-01 -9.34594944e-02
-8.30436766e-01 -1.54346859e+00 1.79706410e-01 3.53579193e-01
7.72165239e-01 -8.45084906e-01 -2.64174819e-01 2.05875874e-01
-4.16254289e-02 -4.97348070e-01 -9.05823946e-01 2.70421982e-01
-9.00647700e-01 -9.40047145e-01 -1.01798987e+00 -7.17525780e-01
8.23390126e-01 5.33997500e-03 7.33287811e-01 8.41424838e-02
-2.36128107e-01 1.26267150e-01 -1.07795499e-01 -3.08505923e-01
1.83954269e-01 -4.57023643e-02 2.32792199e-01 -1.27985522e-01
-4.55221198e-02 -3.41843009e-01 -6.10951424e-01 6.10746443e-01
-6.79607034e-01 -1.16132684e-01 3.38371575e-01 1.13967192e+00
1.21550798e+00 2.25859955e-01 5.31649232e-01 -6.05735004e-01
6.62510931e-01 -4.04814869e-01 -1.36574817e+00 1.61087990e-01
-7.90999293e-01 1.85709268e-01 9.43578064e-01 -5.49319506e-01
-6.10706210e-01 -4.54948574e-01 -6.45700842e-02 -4.52108055e-01
7.13298738e-01 7.60604560e-01 1.32730082e-02 -2.60443419e-01
3.21165860e-01 3.83102000e-01 -2.85272181e-01 -5.33223093e-01
4.93409112e-02 4.45370853e-01 3.36126894e-01 -8.64222825e-01
6.42648578e-01 4.77129966e-01 2.78913170e-01 -5.98628521e-01
-1.10363793e+00 -1.62279516e-01 5.30226491e-02 -5.69184646e-02
6.66348398e-01 -5.16032994e-01 -1.55843771e+00 2.43252471e-01
-6.29486978e-01 -4.50529754e-01 -1.89524069e-01 7.12103963e-01
-6.03349388e-01 3.24467063e-01 -4.83935863e-01 -1.10658073e+00
-2.88203150e-01 -1.51620305e+00 5.95538735e-01 5.02509058e-01
3.14490438e-01 -1.14926076e+00 -2.13013127e-01 5.84589779e-01
2.25977987e-01 2.85236001e-01 9.39655542e-01 3.84620763e-02
-5.49032807e-01 -9.04124528e-02 -3.14263701e-01 4.23103571e-01
-2.31659964e-01 -2.69360393e-01 -7.20331147e-02 -4.28144455e-01
2.52184182e-01 -7.67918602e-02 6.89790189e-01 6.52862608e-01
1.44392538e+00 -5.89663327e-01 -2.77917683e-01 1.00959432e+00
1.54318118e+00 4.03927296e-01 3.55096012e-01 4.03539598e-01
5.27340591e-01 1.60605431e-01 6.60426617e-01 6.48757637e-01
4.29617912e-01 4.86744583e-01 4.63671416e-01 -4.87158634e-02
3.80959213e-01 -6.99005499e-02 2.54555553e-01 1.07638669e+00
-2.58862317e-01 -5.03158104e-03 -6.88600779e-01 6.08870268e-01
-1.84638941e+00 -5.62521100e-01 -5.00564396e-01 2.38484550e+00
1.12032998e+00 3.68765295e-02 4.51731645e-02 -1.55101810e-03
7.38792181e-01 -1.04889691e-01 -6.85870111e-01 -5.68565607e-01
-5.31486794e-02 4.61300373e-01 5.79847097e-01 7.34585166e-01
-8.35459650e-01 6.86500669e-01 3.81900001e+00 1.29248428e+00
-1.18354988e+00 9.71052721e-02 6.02582693e-01 -4.79447454e-01
-3.63860548e-01 6.86731040e-02 -9.29893672e-01 1.07666814e+00
4.07533437e-01 -3.09863150e-01 6.49242342e-01 9.41826820e-01
3.51988047e-01 -3.76091689e-01 -6.19480550e-01 1.25194573e+00
2.87181903e-02 -1.28050900e+00 -4.86367792e-01 2.62037158e-01
1.06111050e+00 -3.53337497e-01 3.52985233e-01 3.78464580e-01
3.09991449e-01 -1.20097280e+00 4.75081623e-01 2.51383662e-01
8.68373990e-01 -1.03381824e+00 6.48405731e-01 3.89428824e-01
-1.30003941e+00 -2.67737448e-01 -5.11181891e-01 1.17286734e-01
3.98296803e-01 1.10403168e+00 6.96040988e-02 5.50419271e-01
8.44451368e-01 3.00345838e-01 1.98663175e-01 1.11538768e+00
-1.58608541e-01 5.16542554e-01 -5.87135613e-01 -4.33876634e-01
4.96923506e-01 -8.15814495e-01 7.80170143e-01 7.39956796e-01
5.78118801e-01 4.15277719e-01 4.63635981e-01 6.65145934e-01
-3.03081363e-01 5.06036401e-01 3.36867839e-01 1.96635902e-01
2.02382177e-01 1.01683772e+00 -5.69321632e-01 -1.06146157e-01
5.63003495e-02 4.50437814e-01 1.53747395e-01 3.85773897e-01
-1.12547028e+00 -8.43059480e-01 5.93754053e-01 9.89830270e-02
2.99319357e-01 -2.41862297e-01 -6.04103327e-01 -1.33284497e+00
4.22264129e-01 -6.59554541e-01 4.30638403e-01 -3.38696092e-01
-1.04102755e+00 4.31141436e-01 3.84450285e-03 -9.12428975e-01
3.26187611e-01 -5.11868536e-01 -4.47609484e-01 9.66267765e-01
-1.37642467e+00 -5.20864487e-01 -3.72056961e-01 4.83622968e-01
1.79768980e-01 -2.24659324e-01 3.13870460e-01 4.81729537e-01
-8.04563820e-01 6.77490890e-01 7.12579250e-01 -2.45208636e-01
1.97981656e-01 -9.17899013e-01 -5.78051031e-01 5.18128276e-01
-2.14521527e-01 5.08615673e-01 6.58935010e-01 -4.77109760e-01
-1.55068028e+00 -9.29345429e-01 6.30846977e-01 9.49613154e-02
6.34478211e-01 -2.21378971e-02 -8.83266270e-01 3.74565423e-01
-3.81558895e-01 1.96505904e-01 3.68253052e-01 -4.98868860e-02
2.38221675e-01 -3.26355547e-01 -1.13378966e+00 6.60997093e-01
8.65000606e-01 7.32222050e-02 -2.07840770e-01 5.11812449e-01
4.34521973e-01 -6.17500603e-01 -1.08178508e+00 7.66344905e-01
3.84508699e-01 -8.04329753e-01 8.23121667e-01 -4.44184273e-01
2.68188268e-01 -3.82694781e-01 -1.51164398e-01 -1.17847824e+00
1.02402642e-01 -9.65799570e-01 -2.45382220e-01 6.94959164e-01
4.98661041e-01 -1.02108943e+00 7.69778430e-01 3.40594560e-01
-2.43057296e-01 -1.62985861e+00 -1.28126752e+00 -9.24384773e-01
6.40403926e-01 -4.56060320e-01 1.05863087e-01 7.12590814e-01
-7.69041032e-02 -2.20198944e-01 -3.69748682e-01 7.98541587e-03
5.96375465e-01 1.64190024e-01 4.92823958e-01 -8.91316533e-01
-6.14678144e-01 -6.53396785e-01 2.03036919e-01 -1.21654820e+00
-3.97249199e-02 -9.20772612e-01 6.45861104e-02 -1.41063023e+00
3.24672073e-01 -1.33211100e+00 -2.71620125e-01 2.88088202e-01
-3.28832239e-01 1.54548094e-01 1.78464845e-01 1.54772267e-01
-4.32409555e-01 7.30737746e-01 1.44967091e+00 -5.54125123e-02
-5.10173798e-01 1.25953972e-01 -8.06204498e-01 8.65509033e-01
5.98509371e-01 -5.14281869e-01 -2.83074170e-01 -4.33730304e-01
8.77883315e-01 3.69786918e-01 1.98979080e-01 -7.62617767e-01
2.33478844e-02 -5.03382981e-01 -1.06701761e-01 -3.38448524e-01
3.35854471e-01 -7.46402666e-02 -1.64815802e-02 5.44527829e-01
-5.40113635e-02 1.10769428e-01 2.24637777e-01 4.82951701e-01
-1.44170761e-01 -5.57596326e-01 1.03665483e+00 -4.73472588e-02
7.31624216e-02 5.48953891e-01 -5.56487776e-02 5.17509162e-01
1.22192585e+00 -5.61814308e-02 -3.43558937e-01 -2.93672204e-01
-6.12525940e-01 5.97474515e-01 2.24650204e-01 -1.39482126e-01
3.11314940e-01 -1.18402588e+00 -6.94106877e-01 1.36705851e-02
-5.53841949e-01 5.16987920e-01 4.04298842e-01 1.54852152e+00
-5.72073877e-01 3.84168118e-01 4.70678091e-01 -5.36770403e-01
-8.04671764e-01 2.14822412e-01 4.13282186e-01 -5.99081755e-01
-2.37390175e-01 1.41724157e+00 6.36498779e-02 -3.16388577e-01
1.63296629e-02 -3.58222663e-01 2.82333881e-01 -5.44417184e-03
1.72605872e-01 9.84519958e-01 -1.10253595e-01 -2.40874290e-01
-4.08143997e-01 8.55612755e-01 8.94875750e-02 -1.64085373e-01
1.52015829e+00 6.60833418e-02 -5.27732968e-01 3.11509401e-01
1.47226906e+00 3.01528931e-01 -1.30471575e+00 2.10769653e-01
-3.10596555e-01 -3.25741947e-01 1.63793359e-02 -5.71276069e-01
-1.04915023e+00 8.23463559e-01 4.23667461e-01 1.51706219e-01
1.09957218e+00 -2.48035267e-02 1.12751544e+00 8.99160877e-02
3.85557771e-01 -1.47057879e+00 1.84858784e-01 6.83970332e-01
7.95466721e-01 -1.08894181e+00 1.69808641e-01 -3.62248361e-01
-5.29429793e-01 1.09493828e+00 5.28501272e-01 -3.12402517e-01
8.88139427e-01 -2.71023391e-03 -4.31854248e-01 -2.48216037e-02
-3.11268330e-01 1.51408419e-01 2.33945385e-01 -1.98947296e-01
3.72043639e-01 1.00469775e-01 -8.83500993e-01 9.35671628e-01
-2.92541534e-01 -2.86453534e-02 4.20501292e-01 6.33727193e-01
-3.73661846e-01 -9.96231973e-01 -4.98919576e-01 5.82363188e-01
-4.72792357e-01 -1.48211285e-01 4.63813722e-01 6.92525923e-01
3.23246956e-01 7.62609005e-01 -2.20198244e-01 9.70653817e-02
1.54268011e-01 -2.55766630e-01 5.49923956e-01 -3.22660118e-01
-3.42193305e-01 2.06880450e-01 -2.50271320e-01 -2.17024207e-01
-2.09803935e-02 -5.49256206e-01 -1.54912579e+00 -2.66136467e-01
-3.68371069e-01 6.70789003e-01 6.64656103e-01 1.06396186e+00
-4.34943428e-03 3.15272391e-01 7.21763015e-01 -6.20494843e-01
-6.59312189e-01 -6.13685310e-01 -6.75262272e-01 -1.05067670e-01
8.54808986e-02 -8.40641499e-01 -6.76311851e-01 -3.71384501e-01] | [6.499183177947998, 4.570314407348633] |
ab022ffd-affd-4757-a5d7-d291e667ea5d | 3d-aided-data-augmentation-for-robust-face | 2010.01246 | null | https://arxiv.org/abs/2010.01246v2 | https://arxiv.org/pdf/2010.01246v2.pdf | 3D-Aided Data Augmentation for Robust Face Understanding | Data augmentation has been highly effective in narrowing the data gap and reducing the cost for human annotation, especially for tasks where ground truth labels are difficult and expensive to acquire. In face recognition, large pose and illumination variation of face images has been a key factor for performance degradation. However, human annotation for the various face understanding tasks including face landmark localization, face attributes classification and face recognition under these challenging scenarios are highly costly to acquire. Therefore, it would be desirable to perform data augmentation for these cases. But simple 2D data augmentation techniques on the image domain are not able to satisfy the requirement of these challenging cases. As such, 3D face modeling, in particular, single image 3D face modeling, stands a feasible solution for these challenging conditions beyond 2D based data augmentation. To this end, we propose a method that produces realistic 3D augmented images from multiple viewpoints with different illumination conditions through 3D face modeling, each associated with geometrically accurate face landmarks, attributes and identity information. Experiments demonstrate that the proposed 3D data augmentation method significantly improves the performance and robustness of various face understanding tasks while achieving state-of-arts on multiple benchmarks. | ['Wei Xia', 'Yuanjun Xiong', 'Yifan Xing'] | 2020-10-03 | null | null | null | null | ['3d-face-modeling'] | ['computer-vision'] | [ 1.19710341e-01 -2.20817253e-02 -8.37757438e-02 -7.82014012e-01
-5.10292172e-01 -4.34433788e-01 4.62241650e-01 -3.27874184e-01
-4.14044783e-02 4.14279014e-01 -1.40527919e-01 -1.43701918e-02
1.30383298e-01 -3.54528993e-01 -5.69811761e-01 -7.55706787e-01
2.67403454e-01 7.06804872e-01 -3.92112374e-01 -2.11929884e-02
1.66399524e-01 9.84115064e-01 -1.71482027e+00 -2.31896579e-01
6.98160708e-01 1.21028626e+00 -4.49336655e-02 5.48635200e-02
-4.15424615e-01 -1.09383591e-01 -2.76025593e-01 -4.78781998e-01
5.45397758e-01 -1.05784655e-01 -4.53273237e-01 7.02276528e-01
7.97557116e-01 -5.34204185e-01 -1.30406573e-01 1.15736222e+00
5.13729930e-01 6.83499277e-02 4.58620191e-01 -1.52867007e+00
-5.54553568e-01 -2.34697219e-02 -1.26207066e+00 -7.34856129e-02
4.29234654e-01 -1.19183920e-01 3.27030748e-01 -1.41403973e+00
3.19420278e-01 1.55732739e+00 5.28783858e-01 9.03324842e-01
-1.04122841e+00 -8.02230120e-01 2.87871122e-01 5.19710705e-02
-1.70166874e+00 -8.78483117e-01 1.10823870e+00 -2.64199764e-01
4.49434310e-01 1.78800032e-01 3.77661169e-01 9.14449513e-01
-5.56037664e-01 5.38558245e-01 1.18355632e+00 -4.86533433e-01
-1.57030020e-02 3.55189115e-01 -1.03164807e-01 8.22873890e-01
2.43947849e-01 -2.09646091e-01 -5.52027583e-01 -3.61095108e-02
8.61441791e-01 2.43587464e-01 -2.54135430e-01 -4.40192670e-01
-8.11844528e-01 4.67404366e-01 3.65762591e-01 -2.37142980e-01
-3.47344786e-01 -2.51042753e-01 3.29636127e-01 1.66090224e-02
7.32109308e-01 8.83945543e-03 -5.03200173e-01 1.63306147e-01
-6.47275448e-01 6.18203310e-03 2.75512159e-01 1.10799384e+00
7.03298151e-01 3.00200492e-01 1.63707331e-01 9.15979207e-01
4.89697397e-01 6.83709979e-01 3.79393771e-02 -7.46809065e-01
4.03012246e-01 1.03521931e+00 7.60442838e-02 -1.10033405e+00
-3.81475419e-01 -4.04020220e-01 -9.89419341e-01 1.22837871e-01
5.45811892e-01 1.87726796e-01 -1.22488141e+00 1.89979911e+00
8.41903687e-01 3.47877383e-01 -6.86493963e-02 9.92913544e-01
8.94523442e-01 4.22109097e-01 -1.13439122e-02 -5.12392700e-01
1.46232748e+00 -6.79526329e-01 -9.19662893e-01 -1.96059495e-01
4.31404352e-01 -9.34895396e-01 1.10682833e+00 7.62222931e-02
-9.19568717e-01 -7.71585643e-01 -8.41057062e-01 4.26153187e-03
-1.94653139e-01 3.04647565e-01 5.97670138e-01 9.57949042e-01
-8.65106881e-01 -1.65507961e-02 -7.29884148e-01 -3.50679278e-01
8.65730524e-01 7.23638773e-01 -9.57722604e-01 -4.54679102e-01
-6.64003372e-01 7.81253576e-01 -1.04587391e-01 6.23949230e-01
-9.24844146e-01 -5.21083772e-01 -9.84643996e-01 -9.81810540e-02
4.92422223e-01 -1.73218474e-01 7.33141422e-01 -5.67974329e-01
-1.12893319e+00 1.19456148e+00 -3.24043274e-01 9.19425935e-02
4.40472752e-01 -2.38702357e-01 -2.84063458e-01 -1.48253843e-01
-9.14668813e-02 7.26616085e-01 1.10569477e+00 -1.53367174e+00
-1.33679256e-01 -1.13964796e+00 -9.87439528e-02 3.83692563e-01
-6.40285194e-01 1.30518660e-01 -5.82688570e-01 -3.28592956e-01
4.80451494e-01 -9.68228996e-01 -3.61119136e-02 2.70332515e-01
-2.42968842e-01 -2.43571222e-01 1.36417234e+00 -7.16392338e-01
5.55684030e-01 -2.11796546e+00 -7.03570992e-02 1.35951430e-01
4.86690030e-02 5.16153812e-01 -1.30814761e-01 -1.77738279e-01
-1.47899330e-01 2.40229934e-01 -2.60134898e-02 -6.85552955e-01
-1.80422708e-01 1.13121033e-01 -9.02981609e-02 5.79050243e-01
4.04891342e-01 5.93809962e-01 -3.49275798e-01 -5.79342604e-01
1.94094196e-01 8.62472534e-01 -6.90212071e-01 3.45965654e-01
-1.30751193e-01 1.03418911e+00 -4.81564343e-01 1.10059762e+00
1.15745747e+00 -1.44657671e-01 -7.47508407e-02 -3.92070472e-01
2.15497717e-01 -3.24313819e-01 -1.37306654e+00 1.75883532e+00
-4.27123964e-01 2.08058268e-01 3.11458498e-01 -8.45632493e-01
1.24365950e+00 5.21366298e-01 4.70457524e-01 -4.70900685e-01
2.67935038e-01 1.47664845e-01 -1.62076741e-03 -5.55755675e-01
1.04375921e-01 -6.74466789e-02 2.58333862e-01 4.03881103e-01
-9.03622285e-02 -1.62683818e-02 -1.96386456e-01 -9.21073928e-02
4.23788905e-01 3.99069667e-01 1.38232335e-01 -7.30738938e-02
8.15470636e-01 -2.76627451e-01 9.10344660e-01 5.42732142e-02
-3.01072568e-01 6.31356835e-01 2.10412562e-01 -6.17996752e-01
-1.04036510e+00 -7.66432643e-01 -2.23024577e-01 7.31175125e-01
1.59564808e-01 -4.14634384e-02 -7.98409283e-01 -8.61234605e-01
-1.47847638e-01 5.69700226e-02 -5.05463362e-01 -7.65080974e-02
-4.99460310e-01 -7.83767045e-01 3.56714606e-01 5.08910298e-01
7.94877708e-01 -6.60000384e-01 -7.71678537e-02 -2.77322263e-01
-1.32009760e-01 -1.49611592e+00 -6.10030293e-01 -2.87913710e-01
-7.59458721e-01 -1.14919341e+00 -4.87638116e-01 -9.08425510e-01
1.39607096e+00 5.11724234e-01 6.96540773e-01 4.32713330e-01
-4.90656048e-01 2.91700929e-01 -2.85786778e-01 -5.79811454e-01
-5.65083958e-02 -3.00563514e-01 6.32631063e-01 5.30363739e-01
4.84342217e-01 -4.32206005e-01 -5.95800936e-01 6.14941418e-01
-6.63493752e-01 -7.57136121e-02 4.51051146e-01 9.47131634e-01
6.69517994e-01 5.18619306e-02 8.42003882e-01 -7.75154889e-01
1.21314056e-01 -2.51644081e-03 -7.71958590e-01 3.42417032e-01
-3.41461122e-01 -2.47054189e-01 5.16728520e-01 -3.57478172e-01
-1.37252891e+00 4.92794871e-01 -1.56620264e-01 -6.49794400e-01
-4.10281092e-01 7.04307631e-02 -8.31197500e-01 -4.70743448e-01
5.60749233e-01 7.02831075e-02 4.72886801e-01 -6.47555053e-01
1.23008132e-01 8.79619777e-01 3.27499598e-01 -6.31161094e-01
1.06388807e+00 6.11715019e-01 3.30669641e-01 -9.35053766e-01
-9.26326990e-01 -2.89260328e-01 -1.06821644e+00 -2.26954415e-01
7.03671634e-01 -1.01504230e+00 -7.53019392e-01 5.22823215e-01
-1.14909708e+00 4.07848537e-01 9.71849188e-02 3.86063457e-01
-2.70090640e-01 4.21824992e-01 -1.72008485e-01 -1.17043495e+00
-3.47491324e-01 -1.43073499e+00 1.28153133e+00 5.61279535e-01
1.25306383e-01 -7.52761066e-01 -6.30282044e-01 7.82496035e-01
2.51350969e-01 2.93153942e-01 7.96005905e-01 -6.37648106e-01
-5.75063050e-01 -4.76304948e-01 -3.54319453e-01 3.69905621e-01
5.37792265e-01 -1.53695494e-01 -1.31605268e+00 -3.28574449e-01
1.38966471e-01 -4.74762201e-01 1.28645927e-01 1.19406335e-01
1.34182644e+00 -1.10686831e-01 -3.10198873e-01 6.70924723e-01
1.04387152e+00 3.33124697e-01 3.26842993e-01 -2.48817936e-01
9.10666347e-01 9.09006655e-01 9.62140501e-01 5.12600899e-01
1.57120183e-01 8.72519851e-01 6.05330408e-01 -2.97930777e-01
-9.54440236e-02 -2.23537326e-01 -9.21693817e-02 5.45703471e-01
-1.33011177e-01 2.36868560e-02 -9.38694298e-01 3.14944386e-01
-1.44431651e+00 -5.23890674e-01 1.17796045e-02 2.34209681e+00
6.45331979e-01 -1.66962087e-01 -1.10747613e-01 3.10430378e-01
8.71081531e-01 -1.13389023e-01 -7.36927509e-01 1.61859527e-01
1.09379895e-01 -6.66158572e-02 2.79592117e-03 2.32935339e-01
-1.02969444e+00 9.41579640e-01 5.16554928e+00 4.81850386e-01
-9.08567727e-01 3.11168674e-02 9.17864442e-01 1.75928324e-01
5.73061369e-02 -1.58735484e-01 -1.14774585e+00 2.64097482e-01
3.11548799e-01 2.36882418e-01 3.03785264e-01 8.89665484e-01
2.96542317e-01 7.99445063e-02 -1.22179830e+00 1.48724627e+00
3.21930915e-01 -9.03009593e-01 2.48151138e-01 2.69437075e-01
6.96299016e-01 -6.98381841e-01 3.13206077e-01 -5.35962172e-03
-2.86403120e-01 -1.16593230e+00 8.55959207e-02 1.38572469e-01
1.02792275e+00 -6.72433734e-01 7.91447222e-01 3.65033954e-01
-1.06308329e+00 1.35057792e-01 -3.29232931e-01 1.78257406e-01
1.39028072e-01 3.55108052e-01 -8.93902898e-01 3.46937627e-01
6.35643363e-01 3.14426333e-01 -4.97331768e-01 8.01718354e-01
1.10836774e-01 1.00854777e-01 -4.31308091e-01 5.01564682e-01
-2.31755957e-01 -2.18844399e-01 3.49446893e-01 4.74841356e-01
4.45880055e-01 3.46219152e-01 3.93724799e-01 4.70170587e-01
-4.82957691e-01 2.01953501e-01 -8.53510737e-01 6.16774783e-02
6.54099107e-01 1.38803339e+00 -7.60709286e-01 4.78836559e-02
-5.03708124e-01 7.94106007e-01 1.60164207e-01 1.67182326e-01
-7.62712181e-01 1.77246615e-01 8.46203983e-01 2.76014805e-01
-2.28757456e-01 -3.47671837e-01 -2.62739092e-01 -1.01724684e+00
2.19135955e-01 -9.62527335e-01 2.50029176e-01 -4.51871604e-01
-1.07697082e+00 7.53468871e-01 -1.06660672e-01 -1.22709310e+00
-3.54572721e-02 -6.63245201e-01 -2.69600868e-01 8.36717427e-01
-1.49744046e+00 -1.50902128e+00 -7.77339220e-01 8.02828312e-01
6.92091763e-01 -4.43718880e-01 8.49374712e-01 6.15061462e-01
-8.58739555e-01 8.74389768e-01 -3.94472748e-01 8.87603462e-02
7.09491551e-01 -7.39795029e-01 2.68919110e-01 6.69460416e-01
1.96816936e-01 6.78216934e-01 4.77050751e-01 -5.74867427e-01
-1.80831218e+00 -1.00629878e+00 3.99134636e-01 -4.68214691e-01
-3.39732580e-02 -5.49768507e-01 -1.05499637e+00 6.28358901e-01
-2.99092889e-01 4.97173697e-01 8.30624759e-01 8.23414996e-02
-3.24982822e-01 -3.70969683e-01 -1.52990854e+00 5.13119757e-01
1.25911450e+00 -4.09626037e-01 -6.26864731e-02 5.29279351e-01
3.68188292e-01 -6.11708283e-01 -8.78306568e-01 7.59365499e-01
3.31507713e-01 -5.02872229e-01 1.06992197e+00 -6.73206866e-01
-6.35521188e-02 -5.17688274e-01 -2.74290591e-01 -9.65102375e-01
2.78773576e-01 -3.54104012e-01 -2.04153005e-02 1.64710343e+00
1.71317041e-01 -5.39931238e-01 1.20729077e+00 8.89543772e-01
1.59802660e-02 -6.80784702e-01 -1.04128349e+00 -5.50989449e-01
-4.93182451e-01 -7.92524368e-02 8.03611755e-01 1.03120863e+00
-5.00557721e-01 2.77778119e-01 -5.31633139e-01 4.82455254e-01
8.75599980e-01 1.00213759e-01 1.03390849e+00 -1.49150467e+00
3.85943651e-01 4.11899714e-03 -6.04792058e-01 -1.01884210e+00
3.83583754e-01 -7.10814834e-01 -1.14302546e-01 -1.25486720e+00
2.14111298e-01 -8.23669434e-01 1.48813546e-01 5.46056092e-01
-1.76042080e-01 6.82491183e-01 -1.63999740e-02 1.84880093e-01
-1.74093887e-01 8.73613954e-01 1.41841376e+00 -8.61672387e-02
-6.40212670e-02 6.61277100e-02 -6.60586476e-01 9.10913408e-01
6.74092710e-01 -1.81672141e-01 -6.98849499e-01 -7.12680578e-01
-2.49341384e-01 5.98672405e-02 1.45997554e-01 -7.85959005e-01
1.90915287e-01 -1.24286219e-01 7.85072803e-01 -6.91102624e-01
9.23725009e-01 -1.27662563e+00 5.61459474e-02 1.01340219e-01
7.29872733e-02 1.69237390e-01 3.84961724e-01 5.64708412e-01
-3.13134521e-01 -3.51379029e-02 9.36856866e-01 -1.08460054e-01
-7.61128187e-01 8.62428367e-01 4.89106178e-01 -1.62774280e-01
1.30982482e+00 -4.26463097e-01 -1.25676110e-01 -2.62239128e-01
-8.56435716e-01 2.30651602e-01 4.58016306e-01 5.78824222e-01
8.47488523e-01 -1.44563055e+00 -6.93308830e-01 7.35193908e-01
9.19555798e-02 3.55229050e-01 4.19499934e-01 7.79249251e-01
-3.21469784e-01 3.16455185e-01 -4.39634562e-01 -7.84662008e-01
-1.73193729e+00 6.13239110e-01 2.27759153e-01 2.08070219e-01
-2.19320744e-01 9.00142252e-01 4.63595659e-01 -5.98579347e-01
4.43381697e-01 2.78667659e-01 -3.55679125e-01 7.08655640e-02
6.42336667e-01 2.49038897e-02 2.58843333e-01 -1.02081108e+00
-4.50340956e-01 9.94629979e-01 -3.63299429e-01 3.04714620e-01
1.17561245e+00 -3.75128835e-01 -9.01264325e-02 -9.84748527e-02
1.03727162e+00 -2.08926573e-01 -1.37826991e+00 -4.18544769e-01
-2.04381987e-01 -1.03476107e+00 -6.79350868e-02 -5.32498181e-01
-1.42596877e+00 1.22500455e+00 7.95200586e-01 -1.48480341e-01
1.33670855e+00 -1.32385939e-01 5.39950550e-01 2.19927698e-01
6.33350611e-01 -8.87550831e-01 2.26521403e-01 1.24747001e-01
9.85461056e-01 -1.75160444e+00 4.96499091e-02 -9.85757291e-01
-4.69643563e-01 8.83602798e-01 1.20837557e+00 4.28288281e-01
6.50193989e-01 -1.50181185e-02 1.64556354e-01 -1.64183185e-01
-3.66837412e-01 -3.31562236e-02 2.60002315e-01 6.79563344e-01
4.22275901e-01 -1.95664719e-01 1.33726269e-01 1.67301759e-01
1.35554790e-01 -3.68588537e-01 1.50441334e-01 7.97770917e-01
-8.21712986e-02 -1.17248678e+00 -7.76104987e-01 3.45812470e-01
-5.45505762e-01 2.84879833e-01 -3.01823825e-01 8.30259144e-01
-3.79161420e-03 8.92240047e-01 7.60887237e-03 -2.65818298e-01
2.81528890e-01 2.49219328e-01 6.46199346e-01 -7.06209660e-01
6.07253127e-02 8.31338316e-02 -1.87586591e-01 -1.34741127e-01
-4.12636191e-01 -5.26086748e-01 -1.30166864e+00 -3.36266130e-01
-5.65091848e-01 2.91238278e-02 9.88388538e-01 8.83839905e-01
4.29878563e-01 1.61778688e-01 9.09577966e-01 -9.08309400e-01
-5.70534229e-01 -1.03635538e+00 -4.56927747e-01 4.76247847e-01
2.77335256e-01 -1.12597191e+00 -1.08829059e-01 2.19514802e-01] | [13.336557388305664, 0.3359694182872772] |
5741c89b-dbc5-45e5-98e6-b389c7e17d39 | considerations-on-the-evaluation-of-biometric | 2303.13294 | null | https://arxiv.org/abs/2303.13294v3 | https://arxiv.org/pdf/2303.13294v3.pdf | Considerations on the Evaluation of Biometric Quality Assessment Algorithms | Quality assessment algorithms can be used to estimate the utility of a biometric sample for the purpose of biometric recognition. "Error versus Discard Characteristic" (EDC) plots, and "partial Area Under Curve" (pAUC) values of curves therein, are generally used by researchers to evaluate the predictive performance of such quality assessment algorithms. An EDC curve depends on an error type such as the "False Non Match Rate" (FNMR), a quality assessment algorithm, a biometric recognition system, a set of comparisons each corresponding to a biometric sample pair, and a comparison score threshold corresponding to a starting error. To compute an EDC curve, comparisons are progressively discarded based on the associated samples' lowest quality scores, and the error is computed for the remaining comparisons. Additionally, a discard fraction limit or range must be selected to compute pAUC values, which can then be used to quantitatively rank quality assessment algorithms. This paper discusses and analyses various details for this kind of quality assessment algorithm evaluation, including general EDC properties, interpretability improvements for pAUC values based on a hard lower error limit and a soft upper error limit, the use of relative instead of discrete rankings, stepwise vs. linear curve interpolation, and normalisation of quality scores to a [0, 100] integer range. We also analyse the stability of quantitative quality assessment algorithm rankings based on pAUC values across varying pAUC discard fraction limits and starting errors, concluding that higher pAUC discard fraction limits should be preferred. The analyses are conducted both with synthetic data and with real data for a face image quality assessment scenario, with a focus on general modality-independent conclusions for EDC evaluations. | ['Christoph Busch', 'Juan Tapia', 'Christian Rathgeb', 'Torsten Schlett'] | 2023-03-23 | null | null | null | null | ['face-image-quality', 'image-quality-assessment', 'face-image-quality-assessment'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 4.99094695e-01 -3.74102652e-01 -9.66172516e-02 -6.03433132e-01
-8.52826953e-01 -6.19897902e-01 4.74430591e-01 7.42696047e-01
-6.22240603e-01 5.72604418e-01 -1.69752352e-02 -2.84223109e-01
-6.55250728e-01 -6.24718606e-01 -6.06553406e-02 -6.51016295e-01
-1.25433654e-01 4.53098148e-01 1.26304463e-01 1.82697833e-01
4.18225437e-01 8.78985405e-01 -1.76295412e+00 1.52788505e-01
7.94404387e-01 1.09995461e+00 -4.41104650e-01 6.26622915e-01
-3.20536382e-02 -2.38781616e-01 -8.08826804e-01 -5.78752458e-01
2.33712092e-01 -5.25598586e-01 -5.60286283e-01 -6.56188354e-02
4.03299093e-01 -1.06698819e-01 6.35388434e-01 9.62829471e-01
7.06049919e-01 -1.85117587e-01 1.19488752e+00 -1.24592197e+00
-1.79821864e-01 1.97869092e-01 -2.77878374e-01 -9.77899730e-02
7.83573449e-01 3.47013652e-01 6.80140078e-01 -8.00865412e-01
4.76065665e-01 1.09218776e+00 6.95865691e-01 2.65310317e-01
-1.57823622e+00 -4.90382165e-01 -5.28556466e-01 -1.66576564e-01
-1.44469893e+00 -1.96295157e-01 5.38624108e-01 -7.34431922e-01
3.26779634e-01 5.71447730e-01 7.05972075e-01 4.64704305e-01
4.81538922e-01 4.26149964e-02 1.49455619e+00 -5.84764898e-01
4.26778764e-01 1.68710321e-01 1.32748067e-01 4.42403555e-01
4.70903605e-01 2.37054259e-01 -2.59666830e-01 -3.17105055e-01
5.34508646e-01 -4.74678487e-01 -1.37740627e-01 -2.85787076e-01
-6.45580113e-01 5.46150982e-01 1.64227292e-01 4.00186211e-01
-3.28635871e-01 -2.30715573e-01 4.16165829e-01 3.80123138e-01
2.20950082e-01 4.30235684e-01 -2.23410174e-01 -1.05355039e-01
-1.05750871e+00 2.61464745e-01 6.92421198e-01 3.40264738e-01
6.77597106e-01 -1.37721732e-01 -5.43295145e-01 1.00853181e+00
1.73152223e-01 6.41414523e-01 1.90781295e-01 -7.24092066e-01
3.17206942e-02 8.67173791e-01 3.12745214e-01 -8.80905092e-01
-5.23631752e-01 -4.28839028e-01 -6.38282120e-01 4.66866165e-01
8.38758826e-01 1.86356872e-01 -6.78111315e-01 1.42960668e+00
1.76557004e-01 -5.76391399e-01 -1.77273527e-01 8.40194643e-01
5.41481078e-01 2.39536270e-01 2.54621148e-01 -6.20253325e-01
1.43851149e+00 3.23528260e-01 -4.35586751e-01 6.59753501e-01
3.97426933e-01 -8.73985052e-01 1.28612423e+00 6.61040068e-01
-9.35994387e-01 -6.58735931e-01 -9.53350902e-01 3.51135731e-01
-5.08062005e-01 4.45296735e-01 3.98108140e-02 1.17608380e+00
-9.43146229e-01 7.80093253e-01 -4.75109398e-01 -2.97903299e-01
3.96952070e-02 5.52524269e-01 -4.07213897e-01 7.08470941e-02
-9.09377992e-01 8.01871836e-01 6.43973500e-02 2.35858127e-01
-1.79437995e-01 -6.93767011e-01 -6.68517530e-01 -3.21371085e-03
-3.01894754e-01 -2.67216384e-01 4.73159552e-01 -1.01419759e+00
-1.27688158e+00 1.14793837e+00 1.60726666e-01 -1.24899715e-01
5.94855011e-01 2.15845719e-01 -6.86359227e-01 1.14263892e-01
-8.97490010e-02 4.31622416e-01 5.99935651e-01 -1.25941730e+00
-2.47850716e-01 -6.20326519e-01 -3.85940343e-01 -1.61646441e-01
5.54974489e-02 2.52306432e-01 -6.70576692e-02 -4.99131531e-01
2.42094025e-01 -6.03598058e-01 2.28369474e-01 1.63770288e-01
6.68388456e-02 -3.03399950e-01 3.14477503e-01 -7.22833276e-01
1.44622028e+00 -2.11589026e+00 -3.04512322e-01 8.60699177e-01
-3.78042385e-02 2.69545853e-01 -3.51745248e-01 3.49521995e-01
-1.00944400e-01 2.08809972e-02 -4.24739659e-01 2.99860239e-01
-1.68020964e-01 -2.79494405e-01 3.68027240e-01 5.96799493e-01
3.94847035e-01 4.05861259e-01 -8.23203683e-01 -3.53207350e-01
2.36272439e-01 5.68767846e-01 -2.25629717e-01 4.14574193e-03
2.56541878e-01 2.11077407e-01 -9.95963663e-02 7.28322685e-01
7.20842063e-01 1.80585295e-01 1.39334902e-01 -5.63743174e-01
-3.23470563e-01 -2.25162476e-01 -1.41570556e+00 8.43883693e-01
-1.47024468e-01 4.65967208e-01 -1.96855873e-01 -6.56654716e-01
1.65200400e+00 2.68949807e-01 5.49106896e-01 -5.87706804e-01
2.00980082e-01 6.22539341e-01 2.03176603e-01 -2.39651322e-01
2.31082112e-01 -2.90643096e-01 1.95119441e-01 1.52627572e-01
6.22763392e-03 -1.20033391e-01 1.82799354e-01 -3.67550015e-01
6.70386493e-01 1.24217577e-01 4.09670115e-01 -5.95759213e-01
9.01744962e-01 -3.88016641e-01 2.39922345e-01 3.97883058e-01
-2.78619468e-01 6.87680721e-01 7.73873508e-01 -3.51332754e-01
-9.38364625e-01 -1.14228940e+00 -8.27426374e-01 4.86804098e-01
3.04019134e-02 -1.83485061e-01 -7.75105119e-01 -3.86450678e-01
2.46346295e-01 3.21799248e-01 -6.04686499e-01 -1.63067892e-01
-1.84362710e-01 -8.70371580e-01 5.64871490e-01 2.36590549e-01
3.22085857e-01 -8.36125851e-01 -6.41538084e-01 1.90164685e-01
2.16164932e-01 -4.83798742e-01 -6.79340288e-02 -1.11990146e-01
-9.32770133e-01 -1.39204013e+00 -7.24265993e-01 -2.16731831e-01
6.73301697e-01 -4.10761029e-01 9.86240029e-01 2.11366296e-01
-2.62381136e-01 5.40620685e-01 -2.51080692e-01 -2.28075609e-01
-5.90441883e-01 -3.98720980e-01 3.65640298e-02 2.48039812e-01
2.73260742e-01 -2.35197902e-01 -7.71631718e-01 6.16482079e-01
-9.12330508e-01 -6.32081270e-01 4.84426916e-01 8.77695680e-01
6.13111317e-01 -2.69470304e-01 5.39521754e-01 -5.81915081e-01
8.51582229e-01 1.26667903e-03 -7.00256526e-01 5.36123157e-01
-1.07206607e+00 1.50845140e-01 3.71653378e-01 -2.55979627e-01
-6.02635145e-01 -2.31648535e-01 -9.46025178e-02 -2.14620829e-01
-1.29967481e-01 4.28588152e-01 -1.71753004e-01 -1.63213179e-01
9.84843493e-01 -1.54603517e-03 4.48104054e-01 -3.19043815e-01
-1.44936994e-01 7.63583243e-01 5.60966074e-01 -8.27942669e-01
4.41496164e-01 -2.34646536e-02 4.71508175e-01 -7.21968234e-01
9.29491296e-02 -4.94069725e-01 -7.26189792e-01 -8.26041162e-01
5.38311601e-01 -1.78435802e-01 -1.07214820e+00 4.36743528e-01
-6.99976206e-01 7.28395209e-02 -2.49423891e-01 5.48315763e-01
-3.01071733e-01 4.16887641e-01 -6.12499714e-02 -1.30316067e+00
-4.81229156e-01 -1.13714421e+00 1.04435718e+00 3.35162818e-01
-3.79841506e-01 -9.13522303e-01 -5.66682070e-02 1.68027624e-01
2.65514374e-01 6.03683591e-01 1.04039252e+00 -3.87241244e-01
5.58226146e-02 -7.44412839e-01 -2.31288880e-01 4.74292189e-01
1.76938593e-01 5.20740807e-01 -9.49120820e-01 -4.10569310e-01
-2.81307906e-01 1.50821626e-01 4.16175812e-01 3.70524675e-01
9.31468964e-01 -2.88997293e-01 -6.20269291e-02 2.47751459e-01
1.65706766e+00 3.41635793e-01 8.56293559e-01 -1.10563495e-04
2.60844193e-02 9.55761075e-01 5.81782997e-01 3.21303785e-01
-1.87568679e-01 7.28377938e-01 1.96537450e-01 -9.19478480e-03
-1.18155375e-01 -6.93817362e-02 1.70053318e-01 2.82772601e-01
-3.92818481e-01 -7.53107592e-02 -9.33612645e-01 1.59161717e-01
-9.46757972e-01 -6.58189058e-01 -3.00557733e-01 2.89596605e+00
5.56272268e-01 2.24149734e-01 6.26484811e-01 6.98677063e-01
9.85466003e-01 -3.60678762e-01 -2.80374348e-01 -6.06925249e-01
-9.56399217e-02 4.18463975e-01 2.54518360e-01 5.62862098e-01
-5.39118707e-01 2.57684022e-01 6.33240557e+00 8.04712117e-01
-1.30139589e+00 -3.67101192e-01 7.56573796e-01 2.83475190e-01
-2.40845054e-01 2.59373561e-02 -4.99832869e-01 7.12017894e-01
1.03609693e+00 6.53960779e-02 -2.48291846e-02 2.90302306e-01
2.98990726e-01 -5.63149154e-01 -9.80707645e-01 9.96876895e-01
-1.98276743e-01 -7.10707486e-01 2.20931098e-01 2.51953840e-01
3.28085184e-01 -8.13556969e-01 1.00238316e-01 -2.70384759e-01
-3.25508446e-01 -1.01888108e+00 4.76598799e-01 9.94492292e-01
1.28129470e+00 -4.65315312e-01 1.14549315e+00 -2.11624667e-01
-1.06694829e+00 5.19394092e-02 -1.55527040e-01 1.58395439e-01
-3.47823083e-01 7.42457032e-01 -7.48506010e-01 7.20862150e-01
4.00876880e-01 -4.79773581e-02 -6.93216443e-01 1.20040512e+00
2.20577836e-01 4.62514967e-01 -3.58991802e-01 -2.99042910e-01
-2.37914428e-01 -4.72421318e-01 4.62998152e-01 1.23828459e+00
4.01872426e-01 1.82673663e-01 -4.27525818e-01 9.27259028e-01
4.69415188e-01 5.78945398e-01 -1.59675032e-01 2.23057702e-01
7.30928600e-01 1.08779645e+00 -8.25807333e-01 -1.83610469e-02
-1.22842371e-01 4.70324427e-01 -3.16884935e-01 1.54084682e-01
-4.20152903e-01 -3.80413145e-01 2.22822383e-01 5.45742869e-01
-4.57216710e-01 1.57049581e-01 -5.57369709e-01 -4.80858535e-01
1.36070758e-01 -6.80109143e-01 6.03859007e-01 -6.59346282e-01
-1.23048151e+00 5.16670465e-01 3.40532094e-01 -1.55987895e+00
-1.70596153e-01 -7.41144121e-01 -5.40372550e-01 1.26381075e+00
-7.92326689e-01 -9.21027541e-01 -2.99118876e-01 9.99307632e-02
-1.77471131e-01 -1.71472773e-01 8.16712081e-01 2.28406079e-02
-2.51524776e-01 9.52743173e-01 3.67051587e-02 3.09652295e-02
4.86143976e-01 -1.14465451e+00 -3.61273676e-01 5.03377914e-01
-2.38179117e-01 6.23504937e-01 8.06085229e-01 -5.87727189e-01
-9.15940344e-01 -4.49575961e-01 8.29765320e-01 -4.45453733e-01
7.57418498e-02 2.39366628e-02 -8.74183893e-01 -2.44469509e-01
-4.70785767e-01 -1.21888146e-01 8.85881066e-01 1.50753498e-01
-3.41561884e-01 -6.12243950e-01 -1.67519510e+00 3.07411999e-01
5.45327723e-01 -3.52800101e-01 -1.87486574e-01 -2.97613561e-01
-2.69471794e-01 -2.93213073e-02 -1.49339867e+00 6.37602746e-01
1.31017995e+00 -1.19178200e+00 7.29366481e-01 -1.80481762e-01
-2.80798040e-03 -4.37018663e-01 -3.11682299e-02 -9.01231408e-01
-2.60246605e-01 -1.50196984e-01 4.85722661e-01 1.23097360e+00
4.61245388e-01 -6.17406964e-01 5.01116097e-01 7.51762152e-01
2.99811453e-01 -9.88085866e-01 -1.11813962e+00 -8.50220203e-01
-1.18672013e-01 -2.77929306e-01 6.78466439e-01 5.70275247e-01
-9.61272866e-02 -1.37981147e-01 8.74271095e-02 -3.70009779e-03
6.69444621e-01 -1.99075341e-01 6.76921189e-01 -1.60457218e+00
-3.22513725e-03 -6.79396868e-01 -9.56742406e-01 -6.27659187e-02
-5.66342294e-01 -5.22993445e-01 -2.40708470e-01 -1.04650724e+00
2.38790754e-02 -4.30742741e-01 -3.89832467e-01 2.95595527e-01
-8.03110600e-02 4.48859870e-01 1.09424159e-01 2.50120044e-01
2.47257292e-01 6.91793859e-02 9.80435431e-01 4.10716794e-02
-4.58884656e-01 1.23012036e-01 -1.50455222e-01 2.87208170e-01
5.83084404e-01 -2.66916066e-01 -1.76997334e-01 2.62487888e-01
2.43906349e-01 3.48335385e-01 4.65845019e-01 -1.26281273e+00
-1.47471264e-01 -1.61615819e-01 7.43622601e-01 -3.97441506e-01
1.66034624e-02 -7.29941785e-01 6.03630304e-01 7.31125653e-01
-3.16388696e-01 1.95899382e-01 1.04613051e-01 1.23852186e-01
-1.25170976e-01 -3.68571848e-01 1.08497202e+00 4.01792467e-01
-2.17667535e-01 -6.99426010e-02 -8.81343782e-02 -5.38314044e-01
6.94102347e-01 -9.62936759e-01 5.57897706e-03 -1.94918469e-01
-1.00468695e+00 -1.10475749e-01 4.50295538e-01 -2.35031396e-02
6.48586690e-01 -1.30839312e+00 -8.96295249e-01 4.40329552e-01
4.40185130e-01 -6.96540654e-01 1.28592595e-01 8.87328148e-01
-5.79479992e-01 1.47674814e-01 -4.71237272e-01 -7.40923107e-01
-1.57996261e+00 2.44082972e-01 5.60428858e-01 -7.06520751e-02
7.92967901e-02 3.93336594e-01 -2.79585063e-01 -7.52024576e-02
-5.85046522e-02 -4.35595989e-01 -3.23760927e-01 2.11106658e-01
2.57245541e-01 7.86604226e-01 3.61919522e-01 -5.91821074e-01
-3.72662097e-01 8.52128446e-01 4.34846938e-01 -3.45281273e-01
7.61667848e-01 5.29696904e-02 -1.26603276e-01 4.39038664e-01
9.85957026e-01 8.20766166e-02 -8.76218259e-01 1.41047239e-01
6.25731423e-02 -5.21694422e-01 -3.54432315e-01 -1.17424142e+00
-7.35155880e-01 5.13160288e-01 1.24337232e+00 3.39047998e-01
1.44884372e+00 -1.90490469e-01 -1.97152764e-01 -2.71916181e-01
3.11777353e-01 -1.02247357e+00 -2.32723534e-01 -1.21556021e-01
1.14949703e+00 -8.66207302e-01 1.60900339e-01 -3.59514356e-01
-3.02378029e-01 1.50269854e+00 2.02631623e-01 8.42899755e-02
7.11203694e-01 -7.63095468e-02 1.29769191e-01 -3.96145061e-02
-8.52417499e-02 -8.81190412e-03 7.76989579e-01 7.22010136e-01
7.35134959e-01 2.95511633e-01 -1.32801867e+00 5.24485767e-01
-2.76187241e-01 2.36331686e-01 1.35456145e-01 5.03068030e-01
-2.84636855e-01 -1.24903834e+00 -7.21670032e-01 7.88306534e-01
-2.20267996e-01 3.02762419e-01 -3.73801231e-01 9.26194787e-01
4.49187279e-01 7.97317088e-01 2.12643534e-01 -5.64007163e-01
5.97009182e-01 2.51709342e-01 6.46590829e-01 -1.88812062e-01
-8.01073253e-01 4.37594131e-02 -7.23569542e-02 -3.02457750e-01
-4.38248575e-01 -8.52381885e-01 -9.21353638e-01 -3.74827445e-01
-5.50090373e-01 3.34609151e-01 8.43380988e-01 8.75951409e-01
9.66673568e-02 -9.47626978e-02 7.01113105e-01 -2.61335403e-01
-4.22697604e-01 -1.07335544e+00 -6.36644661e-01 5.22520006e-01
1.16037153e-01 -7.88746417e-01 -4.21597749e-01 -1.01690508e-01] | [12.981123924255371, 1.087037444114685] |
29156a0e-edc5-43f9-9588-ae172dba24b7 | m-nca-texture-generation-with-ultra-compact | 2111.13545 | null | https://arxiv.org/abs/2111.13545v1 | https://arxiv.org/pdf/2111.13545v1.pdf | $μ$NCA: Texture Generation with Ultra-Compact Neural Cellular Automata | We study the problem of example-based procedural texture synthesis using highly compact models. Given a sample image, we use differentiable programming to train a generative process, parameterised by a recurrent Neural Cellular Automata (NCA) rule. Contrary to the common belief that neural networks should be significantly over-parameterised, we demonstrate that our model architecture and training procedure allows for representing complex texture patterns using just a few hundred learned parameters, making their expressivity comparable to hand-engineered procedural texture generating programs. The smallest models from the proposed $\mu$NCA family scale down to 68 parameters. When using quantisation to one byte per parameter, proposed models can be shrunk to a size range between 588 and 68 bytes. Implementation of a texture generator that uses these parameters to produce images is possible with just a few lines of GLSL or C code. | ['Eyvind Niklasson', 'Alexander Mordvintsev'] | 2021-11-26 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 6.79910719e-01 7.34182119e-01 2.19019964e-01 -1.97067574e-01
-5.82277656e-01 -4.99726802e-01 8.97281289e-01 -1.52167350e-01
-2.01468334e-01 9.20164943e-01 -2.00082570e-01 -6.27637625e-01
1.03575237e-01 -1.38023329e+00 -9.40125763e-01 -7.56886363e-01
-1.10912696e-01 5.46692729e-01 1.30494878e-01 -2.55321622e-01
3.83707136e-01 5.53057075e-01 -1.76806676e+00 4.52148527e-01
6.23454452e-01 8.67146552e-01 1.90619588e-01 1.23578751e+00
-3.68231982e-01 8.25984955e-01 -7.09050894e-01 -2.28140458e-01
1.42819136e-01 -5.55835068e-01 -7.35425711e-01 9.06242430e-03
2.35106409e-01 -9.07003786e-03 1.03642629e-03 7.20423102e-01
2.63905704e-01 -1.56222254e-01 8.11835825e-01 -7.84105837e-01
-6.56699002e-01 6.28410757e-01 -6.94251209e-02 -3.27473670e-01
2.85801962e-02 3.04689497e-01 5.30787408e-01 -6.31352603e-01
6.09871924e-01 1.15684211e+00 8.49076092e-01 7.22300470e-01
-1.75971198e+00 -2.38254562e-01 -4.83779460e-01 -7.93562710e-01
-1.40300965e+00 -4.50537175e-01 4.55898136e-01 -3.75094473e-01
1.30893946e+00 4.15750921e-01 1.01740539e+00 9.88580525e-01
5.17119110e-01 2.86606818e-01 1.26501358e+00 -7.29820490e-01
3.08635443e-01 2.28333086e-01 -2.79612988e-01 1.13517308e+00
3.54797989e-01 1.93691939e-01 -3.50598581e-02 -3.97615999e-01
1.54303527e+00 -5.28130352e-01 2.08336145e-01 -1.39185384e-01
-1.31082940e+00 9.25053418e-01 -1.52828787e-02 -5.18543087e-02
-9.31953117e-02 9.93616998e-01 4.96119678e-01 3.21123898e-01
2.73439884e-01 8.02518129e-01 -9.12488922e-02 -3.24793220e-01
-9.49330568e-01 3.27486753e-01 1.02450562e+00 1.29495752e+00
1.16505468e+00 8.61671865e-01 6.22252151e-02 8.50570738e-01
-2.54247012e-03 5.59692323e-01 3.64531785e-01 -1.12924111e+00
-5.44889607e-02 3.38007659e-01 -6.09063618e-02 -9.73709166e-01
-4.31530997e-02 -2.07879722e-01 -8.91008258e-01 5.32051146e-01
3.48324478e-01 -3.28531533e-01 -9.72167611e-01 1.34080720e+00
-7.94838518e-02 -1.26234546e-01 -4.16852459e-02 7.51900449e-02
4.02775317e-01 1.02243888e+00 7.05586374e-02 9.09899361e-04
1.09452260e+00 -8.25753033e-01 -2.12199390e-01 -2.46201754e-02
8.02807033e-01 -7.53210008e-01 1.39704704e+00 3.54577184e-01
-1.32408798e+00 -5.74165285e-01 -1.17475665e+00 3.85474674e-02
-3.38489652e-01 1.97290272e-01 1.14347565e+00 1.15960276e+00
-1.49702406e+00 6.11633897e-01 -7.32506514e-01 -2.06992969e-01
3.27050894e-01 5.63038886e-01 6.86486065e-02 4.78277117e-01
-5.18006384e-01 6.52105808e-01 5.44017434e-01 1.02337105e-02
-8.53849113e-01 -6.52686656e-01 -7.89479852e-01 -1.05339456e-02
-8.21331516e-03 -1.04388654e+00 1.06691706e+00 -1.03960311e+00
-2.24646306e+00 9.60339963e-01 -7.53501132e-02 -7.64359295e-01
3.60828310e-01 2.10665315e-01 -2.35318139e-01 -1.06820464e-01
-2.89363384e-01 8.94192219e-01 9.29307878e-01 -1.41484225e+00
-3.37384790e-01 5.77745199e-01 4.01475430e-02 -2.33856842e-01
1.85049996e-01 -2.23105416e-01 -9.06526670e-02 -1.03562915e+00
-1.58771411e-01 -1.14228296e+00 -4.82633173e-01 -8.79289582e-02
-3.18453312e-01 1.98559210e-01 7.10529685e-01 -2.05726668e-01
1.11383128e+00 -1.76921475e+00 -1.39834329e-01 5.43650031e-01
-2.10947707e-01 -9.77722779e-02 -1.14578158e-01 4.58920598e-01
1.41097665e-01 3.08463514e-01 -4.85570401e-01 9.65820849e-02
1.42495126e-01 4.75489855e-01 -5.55781543e-01 -1.97180640e-02
2.99632877e-01 1.04709005e+00 -5.07693529e-01 -3.34673315e-01
-2.68933224e-03 3.89689445e-01 -1.03588974e+00 4.50735800e-02
-6.16994262e-01 2.94913035e-02 -3.76694620e-01 4.41113502e-01
2.53512830e-01 -2.75830179e-01 1.60382047e-01 1.78319320e-01
-4.94645052e-02 1.66189209e-01 -9.58896518e-01 1.50083280e+00
-8.89480531e-01 8.66668284e-01 -3.80724698e-01 -6.82721436e-01
1.44034147e+00 6.05204590e-02 -1.27771914e-01 -4.42235768e-01
7.38566220e-02 4.38802391e-01 4.33707684e-02 -3.92838679e-02
8.19800913e-01 -2.55528033e-01 -3.38890195e-01 7.53704906e-01
-1.76336523e-02 -1.04172647e+00 1.81550920e-01 -9.29480866e-02
1.11346805e+00 3.84496689e-01 2.51601692e-02 -6.88503325e-01
3.52660418e-01 1.84515744e-01 -4.78977375e-02 1.17454398e+00
6.33095860e-01 7.01032698e-01 8.15500796e-01 -7.51588523e-01
-1.82791328e+00 -8.36779654e-01 -6.95760176e-02 1.01498854e+00
-4.30292994e-01 -4.00277287e-01 -1.17748058e+00 -4.19635922e-02
-4.41013843e-01 8.23690236e-01 -7.42109537e-01 4.79988679e-02
-9.15256560e-01 -8.80047560e-01 8.85786533e-01 2.23338634e-01
5.70339143e-01 -1.58693326e+00 -9.12283301e-01 5.99543929e-01
5.36438763e-01 -5.88633657e-01 -3.14053714e-01 3.98329705e-01
-1.07760549e+00 -4.23602670e-01 -7.62220263e-01 -1.00061882e+00
8.46622527e-01 -5.17949343e-01 1.33367848e+00 2.63659358e-01
-4.43149269e-01 2.44688570e-01 4.28205766e-02 -3.14457238e-01
-1.12008238e+00 2.44629815e-01 -4.98539388e-01 -4.11592215e-01
-2.71778673e-01 -6.80279315e-01 -4.45301294e-01 -2.02204704e-01
-9.14364040e-01 5.62295973e-01 6.66467547e-01 1.07551479e+00
7.65626132e-01 -5.54866791e-02 4.68294173e-01 -1.50167930e+00
7.92629540e-01 -4.20780443e-02 -6.37717903e-01 6.17096424e-02
-5.28190315e-01 5.04747093e-01 6.92551792e-01 -5.45309722e-01
-1.24998188e+00 1.46838859e-01 -2.39469446e-02 1.48190320e-01
-2.69456506e-01 4.76855338e-01 3.39943051e-01 -4.41440254e-01
8.86628807e-01 5.15987575e-01 2.36715659e-01 1.46983325e-01
6.02992535e-01 8.95521045e-02 5.17010748e-01 -1.21093822e+00
4.52066064e-01 2.71132797e-01 5.66198304e-02 -8.90542626e-01
1.43096354e-02 6.65190697e-01 -2.11859465e-01 2.46277219e-03
5.13976455e-01 -5.43447435e-01 -5.51596344e-01 4.86530811e-01
-1.06668854e+00 -1.11648333e+00 -7.39159882e-01 -2.24505588e-01
-1.38604653e+00 -1.21472538e-01 -8.26487660e-01 -6.01605296e-01
-3.52526039e-01 -1.16400003e+00 1.06029236e+00 3.89800593e-03
-5.94276547e-01 -1.05288291e+00 1.11256674e-01 -4.06466007e-01
8.57491314e-01 4.26351815e-01 1.39625823e+00 2.18876433e-02
-7.88716316e-01 -1.21625118e-01 -3.92150246e-02 1.13856345e-01
4.56761420e-02 3.45413923e-01 -6.98048353e-01 3.18180732e-02
-5.17504737e-02 -1.63506120e-01 5.56713283e-01 2.88068205e-01
1.43321073e+00 -6.90668762e-01 -2.56003410e-01 6.96578145e-01
1.60129333e+00 5.15941978e-01 1.16662812e+00 5.04145861e-01
4.78819579e-01 2.76959062e-01 1.89179496e-03 3.79868180e-01
4.12965417e-02 3.61996293e-01 -9.60763022e-02 -8.01748261e-02
-2.01997817e-01 -3.26073319e-01 1.97299849e-02 8.67580533e-01
-3.70020926e-01 -4.94545959e-02 -1.09145868e+00 3.92049372e-01
-1.39748287e+00 -9.75876570e-01 6.95481077e-02 1.99341619e+00
1.20512569e+00 5.02478838e-01 -7.26326555e-02 -5.31630144e-02
4.71464843e-01 1.27945721e-01 -7.62931025e-03 -1.28652596e+00
-1.43943653e-01 1.06287789e+00 7.08600283e-01 6.19641781e-01
-7.47041583e-01 1.22477031e+00 7.90625286e+00 1.19288647e+00
-1.36057472e+00 -2.61997879e-01 1.08480406e+00 2.93849278e-02
-7.29531765e-01 2.49409840e-01 -5.34552753e-01 4.29231972e-01
1.41999757e+00 -2.22467855e-01 6.28477752e-01 8.52444887e-01
7.24876821e-02 -2.67144620e-01 -9.03644383e-01 8.92069221e-01
-1.73514709e-01 -1.95145762e+00 5.60206950e-01 1.75091401e-01
1.00746739e+00 -4.68901008e-01 2.64033943e-01 3.80809098e-01
5.81781149e-01 -1.57237363e+00 9.14169073e-01 7.01188087e-01
1.05305266e+00 -9.69555020e-01 2.01501876e-01 1.32912531e-01
-9.32635009e-01 3.28609705e-01 -6.17514372e-01 3.19600627e-02
8.79680179e-03 2.18139201e-01 -8.33187640e-01 -1.61934718e-01
3.22219282e-01 1.12279609e-01 -5.33223569e-01 6.87700987e-01
2.49365672e-01 7.08642304e-01 -4.74338233e-01 -3.57482404e-01
4.25853103e-01 -4.20073062e-01 6.97336346e-02 1.56441998e+00
7.73323178e-01 1.51652634e-01 -5.13700485e-01 1.15190554e+00
1.80674866e-01 3.91468741e-02 -8.56540203e-01 -3.64683382e-02
3.66639495e-01 9.08964038e-01 -1.00062835e+00 -6.60437107e-01
1.46644516e-02 9.30108368e-01 2.61415780e-01 4.71905142e-01
-8.16142023e-01 -7.28440285e-01 3.04335565e-03 3.31707507e-01
4.69569266e-01 -4.85531390e-01 -6.25202239e-01 -6.49262547e-01
-3.66418213e-01 -1.14942622e+00 -4.78337288e-01 -9.66293275e-01
-6.77433252e-01 9.22314584e-01 -1.44514188e-01 -9.36687768e-01
-6.89599514e-01 -6.93410516e-01 -6.43065274e-01 1.12077224e+00
-6.88639998e-01 -1.17870212e+00 2.53674607e-05 4.53149267e-02
3.40623707e-01 -4.27035570e-01 1.26764822e+00 -2.69410402e-01
5.31013235e-02 5.31157792e-01 -4.66323160e-02 -2.23898247e-01
-8.03545490e-02 -1.12967181e+00 7.61531830e-01 2.58177131e-01
5.78040816e-02 7.82565415e-01 9.26425636e-01 -4.73019511e-01
-1.38277543e+00 -9.84264016e-01 6.23504639e-01 -2.90704131e-01
6.04568064e-01 -4.30906087e-01 -8.81891072e-01 6.08872712e-01
4.24810737e-01 -1.07892871e-01 3.46457541e-01 -2.91834354e-01
-1.46542251e-01 2.28898466e-01 -1.04688978e+00 1.01264906e+00
9.71676171e-01 -6.79417789e-01 -1.04305953e-01 1.93907134e-02
4.05405968e-01 -5.45303941e-01 -1.01258671e+00 2.79042870e-01
6.19558632e-01 -9.33142960e-01 7.38822162e-01 -2.04422832e-01
5.21050811e-01 -1.55572072e-01 7.48690497e-03 -1.11608827e+00
-1.96401849e-01 -9.17941570e-01 1.45999685e-01 8.38074625e-01
6.41418874e-01 -7.33951926e-01 1.10352039e+00 5.21485031e-01
-7.80364424e-02 -7.74613321e-01 -7.03587294e-01 -5.26788592e-01
3.80683690e-01 -3.42232466e-01 6.67685032e-01 7.17217267e-01
-1.01736046e-01 3.62240300e-02 -3.79582942e-01 -2.70467609e-01
3.74565482e-01 3.00638239e-05 8.56923699e-01 -8.50107372e-01
-6.59559250e-01 -7.54180968e-01 -2.65685499e-01 -1.10136127e+00
2.77288146e-02 -7.74902403e-01 9.72319022e-02 -1.03421724e+00
3.83036397e-02 -1.10377431e+00 4.86627787e-01 2.70623058e-01
3.80104035e-01 4.16403502e-01 -1.08769082e-01 2.59713791e-02
1.01434149e-01 3.69099200e-01 1.21796346e+00 -8.81875679e-02
-1.13795362e-01 -2.65840262e-01 -5.30333042e-01 6.96324885e-01
8.01686645e-01 -3.90923560e-01 -6.17929876e-01 -3.61949116e-01
5.76995552e-01 3.05792242e-01 4.51411605e-01 -1.10866511e+00
-5.60801178e-02 -1.92805156e-01 2.81480938e-01 -1.74642593e-01
3.07633340e-01 -4.28759843e-01 9.11530614e-01 4.07311291e-01
-6.24889314e-01 1.56528860e-01 5.65927863e-01 4.01737899e-01
-1.27360597e-01 -4.64632213e-01 7.05680549e-01 -5.01588166e-01
-4.34645146e-01 4.33597080e-02 -9.70401108e-01 -2.84775406e-01
8.05089355e-01 -6.57244802e-01 -3.36814255e-01 -3.45344245e-01
-9.48483527e-01 -6.92428529e-01 8.08003843e-01 2.21762862e-02
5.33270061e-01 -1.39449680e+00 -3.47007126e-01 4.59399492e-01
-2.54964709e-01 2.70241257e-02 4.74730097e-02 8.59340355e-02
-1.56566572e+00 4.71550256e-01 -2.93316662e-01 -5.07482529e-01
-8.03528011e-01 2.55002558e-01 5.68531394e-01 -3.53745334e-02
-6.03106797e-01 5.57053566e-01 2.84903228e-01 -4.01903182e-01
-3.73945028e-01 -4.51605201e-01 3.39014918e-01 -6.04961395e-01
1.45487875e-01 3.59135047e-02 -7.47852847e-02 -2.49750793e-01
3.71566325e-01 4.48294759e-01 1.87138006e-01 -3.28910589e-01
1.28648341e+00 3.12454849e-01 -3.21226120e-01 3.18259001e-01
9.37966585e-01 1.02684386e-01 -1.23306990e+00 2.46578991e-01
-8.19266960e-02 -1.19317202e-02 -2.66302437e-01 -5.25072813e-01
-5.96862614e-01 7.55388916e-01 4.61517245e-01 4.53157336e-01
7.62928605e-01 -3.08467925e-01 4.16879207e-01 6.16415977e-01
9.08308148e-01 -9.47782278e-01 1.05565242e-01 6.43786728e-01
9.74062443e-01 -5.56147456e-01 -1.76629320e-01 -4.41286832e-01
-4.95557040e-01 1.38076091e+00 2.68439949e-01 -7.73213804e-01
6.60023272e-01 8.78779948e-01 -1.09162852e-01 -1.36344343e-01
-9.87820148e-01 3.85144442e-01 -1.50266103e-02 5.87526619e-01
6.45228982e-01 2.81128258e-01 -7.94575438e-02 -2.87367050e-02
-7.03587532e-01 6.35279194e-02 9.57567513e-01 1.04345262e+00
-6.11637414e-01 -1.17623985e+00 -2.26275191e-01 6.98947251e-01
-3.53043765e-01 -4.67678040e-01 2.22733185e-01 8.92392457e-01
-1.71887890e-01 2.14870170e-01 3.48342896e-01 -3.46050672e-02
-9.20911059e-02 9.80507880e-02 7.26292849e-01 -7.56659269e-01
-6.35929704e-01 1.41864121e-01 3.55721653e-01 -2.38059297e-01
-1.96045756e-01 -4.09172446e-01 -1.08584630e+00 -6.83524013e-01
1.02848321e-01 -5.63568398e-02 4.21708733e-01 5.32583892e-01
3.71512800e-01 4.57820922e-01 1.28809467e-01 -9.92403567e-01
-4.95228171e-03 -6.45077586e-01 -6.02625489e-01 -1.46813020e-01
-5.49742952e-02 -2.99088001e-01 4.90424521e-02 5.70199311e-01] | [11.49544906616211, -0.46926456689834595] |
8eca2497-f588-4cb3-b897-3ac2ec2dc328 | analysis-of-function-approximation-and | 2206.05997 | null | https://arxiv.org/abs/2206.05997v1 | https://arxiv.org/pdf/2206.05997v1.pdf | Analysis of function approximation and stability of general DNNs in directed acyclic graphs using un-rectifying analysis | A general lack of understanding pertaining to deep feedforward neural networks (DNNs) can be attributed partly to a lack of tools with which to analyze the composition of non-linear functions, and partly to a lack of mathematical models applicable to the diversity of DNN architectures. In this paper, we made a number of basic assumptions pertaining to activation functions, non-linear transformations, and DNN architectures in order to use the un-rectifying method to analyze DNNs via directed acyclic graphs (DAGs). DNNs that satisfy these assumptions are referred to as general DNNs. Our construction of an analytic graph was based on an axiomatic method in which DAGs are built from the bottom-up through the application of atomic operations to basic elements in accordance with regulatory rules. This approach allows us to derive the properties of general DNNs via mathematical induction. We show that using the proposed approach, some properties hold true for general DNNs can be derived. This analysis advances our understanding of network functions and could promote further theoretical insights if the host of analytical tools for graphs can be leveraged. | ['Shih-Shuo Tung', 'Wen-Liang Hwang'] | 2022-06-13 | null | null | null | null | ['mathematical-induction'] | ['reasoning'] | [ 3.54331583e-01 3.93633783e-01 2.86295973e-02 -2.84926564e-01
5.90681672e-01 -7.23689795e-01 5.44840455e-01 1.26830682e-01
-4.78887595e-02 7.57855117e-01 -1.21560141e-01 -7.17648745e-01
-6.81321800e-01 -9.32477236e-01 -8.19295406e-01 -7.13424027e-01
-4.71462190e-01 5.59750460e-02 2.20740825e-01 -5.12648821e-01
-1.41306883e-02 1.03628623e+00 -1.45755923e+00 -8.22875500e-02
7.45859504e-01 8.45530450e-01 -7.93296844e-02 6.74303710e-01
-3.03982466e-01 9.42697167e-01 -2.60135889e-01 -3.72939467e-01
4.45108414e-01 -6.98828340e-01 -8.47410560e-01 -2.70124912e-01
2.95021422e-02 -2.21573398e-01 -3.71152431e-01 1.17717028e+00
1.34474523e-02 9.46085975e-02 7.66026318e-01 -1.49140692e+00
-5.47705948e-01 8.00863326e-01 -1.72371954e-01 1.74765497e-01
-2.25387365e-01 -3.96376818e-01 1.22219348e+00 -4.53869611e-01
4.86935139e-01 1.25836527e+00 6.08562469e-01 6.82026327e-01
-1.41862142e+00 -4.81062651e-01 2.32544288e-01 -3.51294763e-02
-9.72498477e-01 -3.51504833e-01 6.44807398e-01 -5.57285547e-01
6.86511159e-01 1.35985091e-01 9.41365302e-01 6.12394691e-01
2.57430702e-01 3.92862499e-01 7.35339642e-01 -8.09277356e-01
3.11669886e-01 -1.30590320e-01 6.84493482e-01 8.19212496e-01
8.43227088e-01 -1.25204287e-02 -3.07487309e-01 4.79760990e-02
1.04311085e+00 5.54320812e-02 -2.42267415e-01 -6.62804842e-01
-7.30746210e-01 1.00551879e+00 5.20278871e-01 6.39414132e-01
-3.44248801e-01 3.97686273e-01 4.94650304e-01 4.50248837e-01
-2.28866264e-02 4.28973407e-01 -3.33824605e-01 5.06404638e-01
-3.38610023e-01 2.27830216e-01 1.03857958e+00 8.34674478e-01
8.36797178e-01 2.32189342e-01 5.09943128e-01 4.69205171e-01
3.49176377e-01 7.34352972e-03 1.63794503e-01 -9.63271141e-01
1.96786582e-01 8.43847573e-01 -3.05268794e-01 -9.31911826e-01
-6.59118772e-01 -5.58854461e-01 -9.44973886e-01 2.13658005e-01
7.43033767e-01 -2.51222998e-01 -6.23485386e-01 2.11273718e+00
-7.41092637e-02 -4.25041020e-01 1.33939564e-01 4.32816714e-01
5.18019319e-01 4.19615000e-01 1.11714583e-02 -1.46290049e-01
9.22623456e-01 -2.23312899e-01 -7.68612683e-01 1.93385884e-01
8.93881023e-01 1.83065131e-01 7.64352500e-01 5.03297567e-01
-8.81312430e-01 -1.72165513e-01 -1.17528605e+00 -1.03035778e-01
-5.40420294e-01 4.35495703e-03 1.02008903e+00 7.99449205e-01
-1.51492965e+00 6.84837878e-01 -9.19473112e-01 -6.21689916e-01
5.17719269e-01 7.10819840e-01 -2.95018017e-01 8.32714289e-02
-1.12426603e+00 8.59090269e-01 6.20814383e-01 5.36094546e-01
-9.56514955e-01 -5.29792130e-01 -6.87768638e-01 2.99298078e-01
2.38252729e-01 -8.42853129e-01 1.03778398e+00 -1.16078043e+00
-1.31447375e+00 5.00896394e-01 5.82923107e-02 -6.60275102e-01
2.47789353e-01 1.88620865e-01 -8.32515359e-02 3.21159095e-01
-3.14763308e-01 3.77231270e-01 3.04056317e-01 -1.17364967e+00
-3.26042920e-01 -5.97719014e-01 4.88650739e-01 -2.48303771e-01
-4.94472593e-01 -2.67306775e-01 3.18574727e-01 -3.16065252e-01
4.56177175e-01 -7.08930016e-01 -2.08621815e-01 8.02627057e-02
-5.37299633e-01 -1.21027000e-01 5.27327299e-01 -1.78266510e-01
1.19665527e+00 -1.89888120e+00 3.71257275e-01 6.11281693e-01
7.06858218e-01 7.45544434e-02 -4.48366031e-02 7.44926274e-01
-4.74077255e-01 3.60695958e-01 -2.47132480e-01 3.94985527e-01
8.50472450e-02 4.71698493e-01 -4.87953573e-01 3.98451030e-01
3.05309474e-01 9.03370678e-01 -8.24001849e-01 -2.22494319e-01
1.25811026e-01 3.96262378e-01 -5.47967255e-01 -2.23083198e-01
-3.49478871e-01 -5.23506925e-02 -5.85876346e-01 2.91282773e-01
4.44813550e-01 -2.73899566e-02 6.16443932e-01 -1.45879462e-01
-2.04457179e-01 3.21381301e-01 -7.74086893e-01 1.16979337e+00
-2.57858962e-01 8.59590828e-01 6.74065873e-02 -1.65968025e+00
9.87991810e-01 2.06943139e-01 3.22810769e-01 -3.27159226e-01
2.77369708e-01 1.84520707e-01 6.43421412e-01 -1.91416860e-01
-2.22112015e-01 -3.56210887e-01 1.81299388e-01 2.86368728e-01
3.84891748e-01 2.85738826e-01 5.19631624e-01 2.86103159e-01
1.31992781e+00 -2.04744920e-01 4.51097190e-01 -8.41781080e-01
5.76787472e-01 -2.03140259e-01 3.91586423e-01 5.99445343e-01
-1.92112923e-02 -3.16044718e-01 1.39090645e+00 -4.43285525e-01
-1.06856847e+00 -1.22979045e+00 -1.67273074e-01 8.89065266e-01
-3.72383714e-01 -2.85348147e-01 -9.23209965e-01 -1.50335073e-01
-1.43509746e-01 5.13033152e-01 -9.08525646e-01 -2.87208706e-01
-3.02881390e-01 -7.55307436e-01 7.56178319e-01 5.37158251e-01
4.01259124e-01 -7.67801344e-01 -6.07469618e-01 5.21477722e-02
5.42328238e-01 -9.21465337e-01 3.43603611e-01 6.53707683e-01
-1.40070665e+00 -1.23605490e+00 -5.11909127e-01 -8.59465301e-01
9.63676989e-01 1.36707336e-01 7.47880697e-01 9.81501192e-02
1.25703529e-01 2.31407598e-01 -4.18566726e-02 -6.20233059e-01
-5.15200436e-01 1.19663648e-01 1.13547951e-01 -6.73323274e-02
1.00591213e-01 -1.10171437e+00 -2.71982998e-01 1.43941015e-01
-1.35979307e+00 1.09249598e-03 6.20225728e-01 5.46574473e-01
2.70811915e-01 2.41802588e-01 9.21383440e-01 -9.64390397e-01
7.08701193e-01 -3.33116323e-01 -1.01088679e+00 4.20778036e-01
-4.93233263e-01 5.73026836e-01 9.41906393e-01 -1.45447150e-01
-7.30356157e-01 1.29785836e-01 -2.93962508e-02 -4.20541912e-02
1.74434051e-01 6.37078822e-01 -4.90565121e-01 -2.74242043e-01
5.95963180e-01 9.63764042e-02 8.59994590e-02 -2.89337039e-01
4.82793510e-01 1.19218148e-01 3.21845949e-01 -6.34335339e-01
5.55594623e-01 4.19237077e-01 7.65508354e-01 -9.53354120e-01
-5.59900165e-01 1.30069226e-01 -7.46474564e-01 -2.73284763e-01
6.83768272e-01 -3.38922501e-01 -1.16761875e+00 3.86536270e-01
-1.09668696e+00 -3.92839998e-01 -1.50739089e-01 2.60631323e-01
-4.87663060e-01 2.98199832e-01 -6.68928504e-01 -1.16839015e+00
-1.98327780e-01 -8.06477845e-01 1.08558543e-01 -3.15804891e-02
1.88861918e-02 -1.17651927e+00 -6.10799082e-02 -3.38932514e-01
2.93938935e-01 3.84813607e-01 1.63501394e+00 -9.19290066e-01
-4.23113197e-01 -5.53710945e-03 -1.76863015e-01 4.87801343e-01
9.46017727e-02 3.24827850e-01 -8.21187973e-01 2.01452211e-01
9.24751163e-03 -2.29871813e-02 7.80088365e-01 5.33896446e-01
1.11132383e+00 -5.63637435e-01 -3.79039675e-01 3.69798839e-01
1.53665113e+00 3.26501846e-01 6.87323213e-01 2.34997407e-01
6.52848780e-01 8.83677185e-01 -3.41910958e-01 5.65234795e-02
5.30801266e-02 1.85443059e-01 6.23365819e-01 2.71749437e-01
3.03574234e-01 -1.99398518e-01 1.55966207e-01 8.45027149e-01
-2.61580467e-01 -3.38380933e-01 -9.58840728e-01 4.76749480e-01
-1.81694973e+00 -8.35583806e-01 -4.57445830e-01 2.16462159e+00
6.30676806e-01 3.19472224e-01 3.34636748e-01 4.42533255e-01
7.57405877e-01 -3.93834949e-01 -4.64469045e-01 -7.01267540e-01
-2.45436862e-01 3.20671856e-01 6.10416472e-01 3.93473387e-01
-5.01372159e-01 4.77704674e-01 7.26059771e+00 3.98512483e-01
-9.90386426e-01 -2.49217078e-01 4.64532167e-01 2.16430873e-01
-5.96293807e-01 1.92629978e-01 -6.46843135e-01 7.48462677e-02
1.14298403e+00 -2.91639268e-01 6.11780047e-01 7.55926967e-01
1.96571037e-01 2.71904558e-01 -1.30064738e+00 6.16580963e-01
-4.80435580e-01 -1.31686997e+00 2.89048940e-01 2.49784276e-01
5.92739046e-01 -1.45316601e-01 -2.54222512e-01 1.63303278e-02
6.45734310e-01 -9.60684299e-01 5.14325380e-01 4.61077273e-01
3.60526651e-01 -8.01874936e-01 4.74522084e-01 3.71452183e-01
-9.78184104e-01 -2.75707990e-01 -4.36825961e-01 -3.48448306e-01
-1.74592987e-01 7.17934251e-01 -8.17855358e-01 5.41284323e-01
1.23285204e-01 6.04060948e-01 -2.03834295e-01 7.27132142e-01
-3.16799343e-01 5.66475749e-01 -4.31923658e-01 -2.67590463e-01
1.92737490e-01 -4.80674803e-01 9.14062411e-02 1.03250897e+00
2.68996119e-01 -1.96495820e-02 -7.37341762e-01 1.19494355e+00
-1.82243526e-01 5.65805212e-02 -8.94052207e-01 -3.80043775e-01
2.52955019e-01 1.31367457e+00 -9.25542295e-01 -6.74481466e-02
-4.39560771e-01 2.92896301e-01 4.68756467e-01 5.26108623e-01
-5.78182757e-01 -5.13905227e-01 6.55438781e-01 2.57979035e-01
1.79572120e-01 -4.56061661e-01 -3.95739198e-01 -8.60303879e-01
-1.47386283e-01 -6.58088088e-01 2.53596663e-01 -5.56449115e-01
-1.19261730e+00 5.84006011e-01 2.53310770e-01 -6.55208707e-01
-2.86159992e-01 -1.03739750e+00 -5.87367177e-01 6.97626352e-01
-1.26205087e+00 -6.80297136e-01 3.22241671e-02 5.46458960e-01
-1.66843146e-01 1.45204172e-01 7.44712591e-01 1.82871819e-01
-7.05248535e-01 2.07174778e-01 4.00164612e-02 3.51741225e-01
-2.11136907e-01 -1.12512648e+00 1.56969473e-01 1.06252074e+00
-5.67089953e-02 9.84243810e-01 6.50088489e-01 -3.68008673e-01
-1.43161738e+00 -7.67117620e-01 9.17235911e-01 -9.53028277e-02
9.38058436e-01 -7.25263894e-01 -8.70590270e-01 9.69136834e-01
-6.22914769e-02 -4.44313735e-02 5.08544028e-01 6.77404851e-02
-1.90584749e-01 -5.07621050e-01 -8.12779248e-01 7.31756210e-01
1.28343868e+00 -3.26391309e-01 -4.52741444e-01 4.62907068e-02
6.36726081e-01 1.99303940e-01 -7.66320765e-01 2.67675996e-01
6.18738651e-01 -8.64958942e-01 6.49803758e-01 -1.06238782e+00
3.68391007e-01 -2.89538920e-01 -2.07593247e-01 -1.07988572e+00
-2.14781702e-01 -6.67853177e-01 -1.72614634e-01 1.01056266e+00
5.00779152e-01 -1.18924296e+00 3.78613502e-01 7.79094219e-01
-3.60747188e-01 -7.48005927e-01 -7.57776916e-01 -8.12952757e-01
1.53191507e-01 -3.26415777e-01 3.21302325e-01 7.24405289e-01
3.69604379e-01 4.80101883e-01 1.70562118e-01 -1.36198014e-01
4.22626108e-01 -3.40181589e-01 3.10292274e-01 -1.62895644e+00
-1.11336768e-01 -7.66668677e-01 -6.88014627e-01 -8.51616561e-01
3.27093571e-01 -1.07028747e+00 -3.00108820e-01 -1.53696513e+00
-5.09803854e-02 -4.04962540e-01 -2.91061997e-01 5.88334858e-01
6.89509571e-01 -1.51448086e-01 2.70691910e-03 5.07454127e-02
-1.15622990e-01 3.07489485e-01 9.76032972e-01 2.68020302e-01
-1.64088562e-01 1.50689408e-02 -1.03514957e+00 8.90706658e-01
8.39821756e-01 -3.81046116e-01 -7.80873358e-01 -2.69224882e-01
7.93956578e-01 -1.18002594e-01 4.61640477e-01 -8.64080310e-01
1.46769851e-01 1.10606238e-01 1.76443607e-01 -2.09417000e-01
7.47052953e-03 -9.17828977e-01 2.41526186e-01 7.13943124e-01
-6.35612190e-01 1.65923700e-01 2.80516863e-01 4.33154672e-01
3.28958593e-02 -4.06987369e-01 5.29931664e-01 -1.43943623e-01
-3.81112635e-01 6.53325990e-02 -5.56950450e-01 -1.57098040e-01
8.86649370e-01 -2.07207948e-01 -3.16152573e-01 -4.06035483e-01
-7.45389640e-01 -5.58073409e-02 1.50595665e-01 -2.47745544e-01
3.43139499e-01 -1.21258307e+00 -2.16350153e-01 1.03004128e-01
-2.97808498e-01 -1.46237046e-01 2.09548287e-02 9.35633004e-01
-9.36942577e-01 8.67474258e-01 -5.73164701e-01 -3.84664536e-01
-7.64839351e-01 6.99095130e-01 5.60535789e-01 -1.51613355e-01
-3.35105151e-01 4.25278783e-01 6.93445325e-01 -8.77273306e-02
1.91800639e-01 -9.07417774e-01 -1.05484694e-01 -4.59911153e-02
2.73354650e-01 3.62938911e-01 2.27855742e-01 -2.85883933e-01
-3.85802239e-01 2.90203989e-01 1.24090061e-01 2.34161839e-02
1.48226357e+00 -7.03163743e-02 -4.46977109e-01 4.14921463e-01
1.01069629e+00 -3.75352412e-01 -8.45879316e-01 9.49535519e-02
6.49128854e-02 2.37488419e-01 -1.39111597e-02 -4.57851112e-01
-1.23159111e+00 1.03602207e+00 8.83054584e-02 6.89303041e-01
1.37279105e+00 -2.81984434e-02 1.90920815e-01 7.98892081e-01
9.83468294e-02 -6.02376580e-01 -3.31865549e-01 4.61283565e-01
7.35302627e-01 -2.85793602e-01 -2.74909049e-01 -6.27578676e-01
-6.17647022e-02 1.67440629e+00 2.81779051e-01 -3.23194802e-01
8.50094318e-01 3.96833599e-01 -4.21244025e-01 -2.23308757e-01
-8.26226771e-01 -2.71470100e-01 3.71163152e-02 6.00375354e-01
5.18726945e-01 -1.23522393e-01 -5.16036451e-01 6.07527494e-01
-1.57426462e-01 2.51778781e-01 7.05687642e-01 6.52826130e-01
-5.86512625e-01 -1.08666813e+00 6.25049099e-02 4.67078388e-01
-4.03219491e-01 -2.17147455e-01 -6.52885258e-01 1.18791735e+00
-7.38947764e-02 7.09458053e-01 2.58566011e-02 -2.99995273e-01
2.47726277e-01 1.39160708e-01 8.05844665e-01 -4.61505294e-01
-1.15441188e-01 -3.48689735e-01 7.24238753e-02 -6.12665713e-02
-5.30089855e-01 -5.37688553e-01 -1.53527665e+00 -4.93552297e-01
-3.17416191e-01 8.23745970e-04 6.67335451e-01 9.98314857e-01
2.35436782e-01 6.97766602e-01 2.46623054e-01 -4.19606209e-01
-4.84452754e-01 -6.48749530e-01 -9.56094146e-01 -7.82517493e-02
1.97724432e-01 -6.18167698e-01 -3.70939136e-01 6.19291700e-02] | [7.986122131347656, 3.552229404449463] |
aafadb15-d6a3-488b-b804-95821a5f34d7 | learning-progressive-modality-shared | 2212.00226 | null | https://arxiv.org/abs/2212.00226v1 | https://arxiv.org/pdf/2212.00226v1.pdf | Learning Progressive Modality-shared Transformers for Effective Visible-Infrared Person Re-identification | Visible-Infrared Person Re-Identification (VI-ReID) is a challenging retrieval task under complex modality changes. Existing methods usually focus on extracting discriminative visual features while ignoring the reliability and commonality of visual features between different modalities. In this paper, we propose a novel deep learning framework named Progressive Modality-shared Transformer (PMT) for effective VI-ReID. To reduce the negative effect of modality gaps, we first take the gray-scale images as an auxiliary modality and propose a progressive learning strategy. Then, we propose a Modality-Shared Enhancement Loss (MSEL) to guide the model to explore more reliable identity information from modality-shared features. Finally, to cope with the problem of large intra-class differences and small inter-class differences, we propose a Discriminative Center Loss (DCL) combined with the MSEL to further improve the discrimination of reliable features. Extensive experiments on SYSU-MM01 and RegDB datasets show that our proposed framework performs better than most state-of-the-art methods. For model reproduction, we release the source code at https://github.com/hulu88/PMT. | ['Pingping Zhang', 'Xuezhang Zou', 'Hu Lu'] | 2022-12-01 | null | null | null | null | ['person-re-identification'] | ['computer-vision'] | [ 5.01970686e-02 -7.84875154e-01 -1.78415347e-02 -4.59162235e-01
-8.47300410e-01 -4.98730332e-01 4.99360234e-01 -1.23308778e-01
-5.11136472e-01 5.39059997e-01 2.77233541e-01 1.03757210e-01
-8.71162787e-02 -4.22641397e-01 -5.04040062e-01 -7.08473146e-01
4.25749093e-01 -1.16922677e-01 -2.75478289e-02 -6.05283491e-03
1.65953964e-01 3.53359640e-01 -1.51783371e+00 1.78021744e-01
9.78301823e-01 1.03569973e+00 5.50123565e-02 5.62105253e-02
4.03125659e-02 5.41538656e-01 -3.44926625e-01 -5.91912746e-01
5.36779046e-01 -4.43210989e-01 -5.04089057e-01 -4.45822068e-03
5.66067815e-01 -4.80119050e-01 -6.92017913e-01 1.22422910e+00
9.65579569e-01 3.52935791e-01 5.76306522e-01 -1.49725890e+00
-7.60299981e-01 1.37911320e-01 -9.10560310e-01 2.55577028e-01
5.04688978e-01 2.17368737e-01 5.36233187e-01 -9.55920458e-01
4.10812259e-01 1.37738562e+00 6.36222005e-01 5.63798428e-01
-1.07556045e+00 -9.95803356e-01 3.05666417e-01 7.47694016e-01
-1.77069354e+00 -5.85640192e-01 1.02603829e+00 -1.87732086e-01
3.45224321e-01 3.91017407e-01 1.75866425e-01 1.04416990e+00
-2.66096592e-01 8.38774264e-01 1.24992085e+00 -3.94965649e-01
-2.67406672e-01 2.92967379e-01 7.54465088e-02 5.66493750e-01
1.31975397e-01 1.32056579e-01 -4.36742961e-01 -7.47876987e-02
5.88902831e-01 2.80897588e-01 -5.08922517e-01 -3.03282648e-01
-1.00987828e+00 4.09798473e-01 5.56170523e-01 1.48615822e-01
-1.04301170e-01 -1.73035815e-01 4.38480467e-01 2.99528599e-01
3.94424975e-01 -2.49716461e-01 -8.68420452e-02 1.23815805e-01
-6.08492196e-01 1.48498610e-01 1.65624693e-01 8.26786160e-01
7.64751792e-01 -2.83393234e-01 -5.18166006e-01 1.27767122e+00
3.35909992e-01 5.93960345e-01 3.46311212e-01 -6.86985254e-01
6.00553989e-01 5.16942382e-01 6.05531186e-02 -9.48311687e-01
-2.08579019e-01 -5.70723712e-01 -9.43886399e-01 4.59439005e-04
3.05045068e-01 -5.76050133e-02 -8.86128247e-01 1.80861402e+00
4.47007149e-01 2.28263915e-01 -1.42131537e-01 1.21689427e+00
9.88670886e-01 5.73274314e-01 2.80400425e-01 -1.02722548e-01
1.35138631e+00 -8.53484392e-01 -4.85202432e-01 3.74455079e-02
3.15047741e-01 -8.28462660e-01 8.87512684e-01 -4.47261669e-02
-8.84784997e-01 -7.91361690e-01 -7.06690311e-01 -2.14528680e-01
-2.76402354e-01 3.87845278e-01 3.23333532e-01 5.05433142e-01
-8.87032747e-01 2.63423026e-01 -3.81794274e-01 -3.65851343e-01
4.51743722e-01 2.78161854e-01 -6.08630002e-01 -5.14907360e-01
-1.22672999e+00 6.31694734e-01 2.52279580e-01 3.23824435e-01
-4.96094763e-01 -7.51570463e-01 -8.74788284e-01 -8.91789123e-02
3.20125401e-01 -6.77211881e-01 7.98595428e-01 -9.78552282e-01
-1.28931141e+00 8.62282574e-01 -3.77321720e-01 9.92079377e-02
7.17424095e-01 -1.18807346e-01 -5.78997970e-01 6.80044219e-02
1.55345306e-01 6.49540126e-01 7.49883413e-01 -1.50408494e+00
-4.84479278e-01 -4.99672264e-01 -3.30154747e-02 4.82831657e-01
-5.09032667e-01 2.25551620e-01 -8.01788449e-01 -7.09030449e-01
-3.27296406e-02 -9.50546563e-01 1.66986391e-01 2.05131680e-01
-4.98091221e-01 -3.81576926e-01 5.89109540e-01 -9.69631851e-01
9.95434225e-01 -2.36156702e+00 -6.12770556e-04 2.46334925e-01
5.93802258e-02 3.05625111e-01 -4.96136189e-01 2.34063283e-01
-5.87510280e-02 -2.44655102e-01 -1.22338206e-01 -5.98823607e-01
-1.37891853e-02 -2.14252397e-01 5.06646298e-02 5.87380707e-01
-5.42921312e-02 7.35582590e-01 -5.91254354e-01 -5.50697386e-01
2.97889709e-01 7.91904747e-01 -1.17571481e-01 1.48972124e-01
3.89018625e-01 6.40507758e-01 -4.23121363e-01 8.70466590e-01
1.23610938e+00 -7.91711137e-02 -1.74246922e-01 -6.05100155e-01
-6.11376874e-02 -1.47839114e-01 -1.22757757e+00 1.62181938e+00
-3.47700685e-01 3.72300714e-01 5.58541268e-02 -8.38624239e-01
8.46033692e-01 -3.06349993e-02 3.42004806e-01 -1.10936141e+00
1.80357575e-01 3.62559073e-02 -2.47466832e-01 -4.05647129e-01
3.74768734e-01 9.85985324e-02 1.77371472e-01 1.92859367e-01
-1.40063390e-01 6.94647729e-01 -1.42014697e-01 1.04448453e-01
6.61597252e-01 1.39296472e-01 -1.71108067e-01 8.47888142e-02
9.00772572e-01 -5.52565575e-01 8.37971807e-01 5.81089020e-01
-5.61939299e-01 8.35553288e-01 6.09039180e-02 -1.57822981e-01
-8.82420838e-01 -1.09344792e+00 -2.28252053e-01 1.07575274e+00
7.06980169e-01 -1.52280271e-01 -4.58526134e-01 -5.89666069e-01
2.00296000e-01 3.46735090e-01 -5.37961304e-01 -1.23697363e-01
-3.21044266e-01 -6.87901914e-01 3.71831954e-01 5.27921498e-01
1.00545955e+00 -7.32724130e-01 1.30392060e-01 -2.03687027e-01
-4.75839496e-01 -1.07812262e+00 -9.18735504e-01 -6.45897150e-01
-4.72945958e-01 -9.13098633e-01 -1.23770463e+00 -9.13297415e-01
8.06734383e-01 7.91331649e-01 5.93872726e-01 2.01682448e-01
-3.47747654e-01 5.45412302e-01 -4.21557814e-01 -8.32293779e-02
1.53569251e-01 -7.09598064e-02 4.60528508e-02 3.75451088e-01
4.03376579e-01 -3.81132722e-01 -9.76935863e-01 3.98747653e-01
-6.31215215e-01 6.36782572e-02 5.11379361e-01 9.71654177e-01
6.86664343e-01 2.17622500e-02 4.57296669e-01 -3.73234630e-01
5.05257010e-01 -3.37368876e-01 -3.55225086e-01 5.51707506e-01
-5.30013561e-01 -1.55586436e-01 4.22165304e-01 -4.97190624e-01
-1.37324703e+00 -6.29256815e-02 -3.52355167e-02 -5.42039990e-01
-1.23160087e-01 2.46099189e-01 -5.30054986e-01 -3.44695926e-01
1.46928772e-01 5.20111561e-01 -5.69366626e-02 -7.12464690e-01
1.52506188e-01 7.65085399e-01 5.98372579e-01 -4.51630682e-01
9.75973010e-01 4.53229576e-01 -1.44886360e-01 -4.61188525e-01
-5.64561188e-01 -6.60178781e-01 -3.42269748e-01 -2.00055823e-01
5.76014042e-01 -1.32989156e+00 -7.41003454e-01 9.20002937e-01
-8.89129877e-01 -1.18511923e-01 1.47678033e-01 5.21103084e-01
1.20999087e-02 7.47180462e-01 -5.77202737e-01 -7.62107730e-01
-5.55095732e-01 -9.65528786e-01 9.63290393e-01 7.62093604e-01
4.75649148e-01 -6.68929636e-01 -1.21351732e-02 6.96870267e-01
4.02396411e-01 9.31263342e-03 3.60161960e-01 -2.66786784e-01
-4.04208064e-01 -2.99701184e-01 -7.90335596e-01 3.31205815e-01
2.50738919e-01 -3.48013163e-01 -9.63352323e-01 -5.92839599e-01
-3.43654931e-01 -2.19330519e-01 1.17194855e+00 1.56469449e-01
1.38442016e+00 -1.02492027e-01 -4.04858410e-01 9.44694996e-01
1.36469674e+00 1.48614431e-02 7.80414701e-01 4.35167462e-01
1.05744255e+00 5.55828691e-01 6.44840598e-01 4.51242208e-01
7.58698106e-01 8.92188489e-01 4.38009202e-02 -1.86609834e-01
-2.13996291e-01 -3.12130153e-01 3.09169978e-01 4.70492065e-01
-2.17156932e-01 -1.15488239e-01 -5.51719368e-01 4.24769461e-01
-1.81591880e+00 -1.03194129e+00 6.96300268e-02 2.26527619e+00
6.58028841e-01 -2.53151327e-01 2.62084723e-01 -6.31565228e-02
1.14224601e+00 6.93958923e-02 -5.93818247e-01 1.42861620e-01
-4.31618363e-01 -2.55267292e-01 5.07238686e-01 3.35896313e-01
-1.19371557e+00 6.82769358e-01 4.57363415e+00 1.12950122e+00
-1.02630091e+00 3.26480508e-01 5.68202078e-01 -8.85718763e-02
-3.65758419e-01 -1.40096903e-01 -7.35890150e-01 8.43440473e-01
3.58718723e-01 -1.33865491e-01 5.30428827e-01 5.08215129e-01
1.39585182e-01 -4.02429551e-02 -7.78611362e-01 1.44634140e+00
3.75412047e-01 -8.35443795e-01 -4.30746153e-02 -4.16671075e-02
5.65361619e-01 -1.70787781e-01 2.55560517e-01 1.72702864e-01
-2.26659596e-01 -7.59329855e-01 5.73404014e-01 7.02606618e-01
9.36041534e-01 -8.86729658e-01 8.06355774e-01 -1.28171876e-01
-1.43721569e+00 -1.36724621e-01 -4.38627392e-01 3.44236851e-01
-6.92156628e-02 3.24656397e-01 -3.12958919e-02 8.74342978e-01
7.75438964e-01 8.74337852e-01 -9.00597036e-01 1.35153234e+00
-7.36636296e-02 2.12467268e-01 -2.57803738e-01 3.48900735e-01
-3.38566393e-01 -1.37864962e-01 6.07736528e-01 1.12597620e+00
2.26388961e-01 8.60622302e-02 1.26444265e-01 6.85988009e-01
-2.09382221e-01 1.46338210e-01 -2.21072093e-01 3.83138299e-01
7.01433361e-01 1.19237936e+00 -2.61333048e-01 -1.34967536e-01
-5.50327182e-01 1.44110739e+00 2.08349243e-01 6.48149192e-01
-7.54572630e-01 -4.70526665e-01 5.54396927e-01 -1.05623402e-01
1.68411061e-01 6.81075035e-03 1.02421343e-02 -1.42992485e+00
3.89301211e-01 -6.95397317e-01 6.06018305e-01 -7.47077644e-01
-1.67709959e+00 3.92310083e-01 -2.45497152e-02 -1.58981180e+00
2.71450877e-01 -2.32367873e-01 -6.00547194e-01 1.11542141e+00
-1.78608418e+00 -1.60414004e+00 -6.13187253e-01 1.00018454e+00
3.12753111e-01 -2.99629956e-01 3.61645430e-01 7.41964042e-01
-8.16273451e-01 1.37369394e+00 3.73715669e-01 3.09951574e-01
1.17458618e+00 -7.40665674e-01 5.40148579e-02 9.50051248e-01
-3.47226322e-01 6.01979256e-01 4.22940195e-01 -5.36809683e-01
-1.46099818e+00 -1.09793925e+00 6.48722947e-01 -9.74425972e-02
1.76289171e-01 -2.56166816e-01 -9.28831756e-01 4.43771631e-01
3.70645039e-02 2.23303914e-01 7.43996799e-01 8.40057358e-02
-7.09706128e-01 -6.22806251e-01 -1.25227475e+00 4.00474221e-01
1.18415272e+00 -6.90152168e-01 -1.22375660e-01 1.05597794e-01
4.11882043e-01 -2.74444610e-01 -7.52246320e-01 5.51850498e-01
8.53334904e-01 -7.86155283e-01 1.06533396e+00 -1.11387804e-01
-1.25888446e-02 -5.61074555e-01 -6.68235198e-02 -1.08270550e+00
-4.65521395e-01 -2.16133773e-01 1.19153112e-01 1.70075119e+00
-5.93997613e-02 -8.25073123e-01 4.45424616e-01 7.75888622e-01
1.92828894e-01 -3.49263042e-01 -9.70841289e-01 -8.47682714e-01
-1.43930495e-01 -6.30172789e-02 5.64905107e-01 9.98384833e-01
-2.66598016e-01 -7.47563988e-02 -5.98417342e-01 1.82073608e-01
1.05629301e+00 2.10562363e-01 7.33614087e-01 -9.31934774e-01
-1.58015475e-01 -2.83818573e-01 -4.65905905e-01 -9.14370954e-01
1.76128387e-01 -7.85482287e-01 -7.97574744e-02 -1.43601775e+00
8.39745283e-01 -5.22085190e-01 -8.28764081e-01 5.20684481e-01
-5.59225619e-01 4.62345839e-01 5.12722135e-01 4.52214509e-01
-8.61722708e-01 8.54515851e-01 1.14524889e+00 -3.51925611e-01
-2.43709758e-01 -1.24952465e-01 -8.28677416e-01 2.90030420e-01
7.66723573e-01 -2.52369046e-01 -2.99490958e-01 -4.55061197e-01
-2.63359040e-01 -2.22761199e-01 8.09582353e-01 -9.77334857e-01
3.96975011e-01 -1.45954341e-01 8.31902385e-01 -6.39843106e-01
2.84447521e-01 -6.13798976e-01 1.96880102e-01 2.45609134e-01
-4.02058870e-01 -6.76769838e-02 2.07561776e-01 5.12584805e-01
-1.99876502e-01 1.69464275e-01 7.68455923e-01 1.44629255e-01
-8.50169718e-01 5.30980647e-01 1.67579710e-01 -2.10753113e-01
7.56303132e-01 -1.35765895e-01 -6.13296807e-01 -3.32559913e-01
-3.94230276e-01 5.28599083e-01 6.37281299e-01 6.71641111e-01
7.76001751e-01 -1.63926625e+00 -9.60053563e-01 8.70699212e-02
4.19218123e-01 -5.35922587e-01 1.03979886e+00 8.15028191e-01
-1.27522647e-01 1.33690283e-01 -3.37849796e-01 -3.71707380e-01
-1.56560469e+00 6.00974023e-01 4.54570830e-01 3.88245694e-02
-5.61477959e-01 9.04779315e-01 4.26001281e-01 -2.87290782e-01
2.86367267e-01 4.51906919e-01 -2.65862882e-01 -1.14646055e-01
7.80251801e-01 4.22105134e-01 -1.66715875e-01 -1.03292632e+00
-7.40878403e-01 7.26629555e-01 -3.29835624e-01 7.75613412e-02
1.05244577e+00 -5.72468877e-01 -4.11013253e-02 6.74636886e-02
1.36779261e+00 -8.31553265e-02 -1.28831506e+00 -5.30199230e-01
-5.06228328e-01 -8.53935778e-01 3.40174213e-02 -8.86907220e-01
-1.24286807e+00 7.16744721e-01 1.35079408e+00 -3.75679791e-01
1.49725759e+00 -9.41124558e-02 9.25855935e-01 -9.30076989e-04
2.20406294e-01 -1.11623824e+00 -1.45624159e-02 1.18243203e-01
8.19737732e-01 -1.55178535e+00 7.08916336e-02 -4.43335742e-01
-6.35698855e-01 7.88163126e-01 7.25811481e-01 1.61805481e-01
4.89180863e-01 -2.85827488e-01 1.13815323e-01 2.73257107e-01
-2.45620176e-01 -4.19912577e-01 4.60874379e-01 7.13368118e-01
2.48531699e-01 4.82182056e-02 -4.25574511e-01 6.03130043e-01
3.31006676e-01 1.00115955e-01 4.24524061e-02 7.21948564e-01
-3.26666282e-03 -1.21076000e+00 -5.00492871e-01 2.39469111e-01
-3.02043021e-01 -2.65452173e-02 -2.06950173e-01 4.20637518e-01
3.44828933e-01 1.07367837e+00 -7.30038211e-02 -6.70263588e-01
2.80239135e-01 -2.13549092e-01 6.30915940e-01 -9.57704335e-02
-3.29702735e-01 1.41281426e-01 -1.21026501e-01 -4.24742937e-01
-6.00220382e-01 -7.86965489e-01 -8.62090230e-01 -5.51731288e-01
-2.30439216e-01 -1.58859774e-01 4.59172338e-01 7.47240067e-01
5.57515621e-01 1.53201193e-01 9.21942413e-01 -7.59594500e-01
-3.74642998e-01 -8.20510685e-01 -5.66220403e-01 8.05302441e-01
3.86193007e-01 -7.73224056e-01 -2.69885987e-01 -5.89220710e-02] | [14.695296287536621, 0.9333258867263794] |
91db46da-6a03-4f74-a6f8-ca3ba6e686b5 | clustering-images-by-unmasking-a-new-baseline | 1905.00773 | null | https://arxiv.org/abs/1905.00773v1 | https://arxiv.org/pdf/1905.00773v1.pdf | Clustering Images by Unmasking - A New Baseline | We propose a novel agglomerative clustering method based on unmasking, a technique that was previously used for authorship verification of text documents and for abnormal event detection in videos. In order to join two clusters, we alternate between (i) training a binary classifier to distinguish between the samples from one cluster and the samples from the other cluster, and (ii) removing at each step the most discriminant features. The faster-decreasing accuracy rates of the intermediately-obtained classifiers indicate that the two clusters should be joined. To the best of our knowledge, this is the first work to apply unmasking in order to cluster images. We compare our method with k-means as well as a recent state-of-the-art clustering method. The empirical results indicate that our approach is able to improve performance for various (deep and shallow) feature representations and different tasks, such as handwritten digit recognition, texture classification and fine-grained object recognition. | ['Mariana-Iuliana Georgescu', 'Radu Tudor Ionescu'] | 2019-05-02 | null | null | null | null | ['texture-classification', 'handwritten-digit-recognition', 'authorship-verification'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 1.46422565e-01 -2.61673421e-01 1.39686853e-01 -2.88535058e-01
-2.28235111e-01 -5.06216586e-01 8.69190574e-01 6.70049250e-01
-5.18265247e-01 2.49292478e-01 -1.99025899e-01 -1.96786612e-01
-2.36847878e-01 -6.45680189e-01 -7.58140981e-02 -1.04026222e+00
-1.36297643e-01 7.83970118e-01 4.18662280e-01 3.49306732e-01
7.42015243e-01 9.76844251e-01 -1.94336486e+00 7.44315028e-01
5.98927498e-01 1.01097906e+00 -1.23497896e-01 7.64996648e-01
-1.22616217e-01 7.02380538e-01 -6.96577132e-01 -3.27111155e-01
-8.27655867e-02 -6.11921728e-01 -9.38835204e-01 5.68752229e-01
1.57044411e-01 4.65667211e-02 -2.68789604e-02 8.29218626e-01
1.93522125e-01 1.13623962e-01 9.64961946e-01 -1.18623507e+00
-3.39986503e-01 4.35481995e-01 -8.25371623e-01 3.80221903e-01
3.30395609e-01 -3.01642567e-01 7.08197534e-01 -7.20036149e-01
6.53723538e-01 1.08461607e+00 4.46149528e-01 3.01748931e-01
-1.14929569e+00 -4.99190956e-01 5.81238121e-02 5.97976387e-01
-1.47934735e+00 -4.31513608e-01 6.27646506e-01 -5.96767128e-01
7.04573154e-01 3.70166749e-01 4.54896748e-01 8.88838530e-01
-8.12836864e-04 8.02673340e-01 1.30665338e+00 -6.98183179e-01
4.92429942e-01 1.90245181e-01 4.24393177e-01 7.29725718e-01
2.31732935e-01 -3.23411107e-01 -4.04519439e-01 -2.72790700e-01
2.73601383e-01 -6.88751563e-02 -5.84621616e-02 -4.20487255e-01
-1.24701881e+00 7.49725044e-01 -1.36264086e-01 9.51664269e-01
-4.77933407e-01 -1.50044067e-02 4.58381534e-01 2.32036039e-01
3.09371352e-01 3.79300445e-01 -6.59985915e-02 -2.00983316e-01
-1.42305553e+00 -5.74835651e-02 9.78324592e-01 3.21480125e-01
5.26321352e-01 -3.67439300e-01 -1.62708491e-01 5.92986405e-01
2.02684999e-01 -1.21551827e-01 9.41173553e-01 -8.23435009e-01
1.95343625e-02 8.48317444e-01 1.88730340e-02 -1.15394819e+00
-3.67321312e-01 -7.44398236e-02 -8.64213645e-01 3.44149232e-01
8.48922789e-01 1.34801403e-01 -7.85773456e-01 1.06256938e+00
1.46099105e-01 3.34087878e-01 -3.25763077e-02 6.85285985e-01
5.46180665e-01 4.87088412e-01 -2.90465176e-01 -3.83410454e-01
1.31128860e+00 -6.59329951e-01 -4.86126691e-01 3.32761824e-01
5.79568982e-01 -8.26930046e-01 4.54039037e-01 1.03899670e+00
-7.90282726e-01 -5.22428989e-01 -8.95438075e-01 3.84971052e-01
-4.88026351e-01 6.56696379e-01 5.90555191e-01 8.99192989e-01
-8.82289588e-01 1.03893960e+00 -1.12117004e+00 -5.32777905e-01
4.50684071e-01 3.98439705e-01 -4.92056817e-01 -4.35189009e-02
-5.38499773e-01 5.04291713e-01 4.03153896e-01 3.66961062e-02
-4.85092014e-01 6.95655197e-02 -4.48539853e-01 1.27608106e-01
1.37049660e-01 2.68637575e-02 6.70262456e-01 -1.07139480e+00
-1.47513974e+00 1.17107105e+00 -3.08742702e-01 -4.87216145e-01
5.59382915e-01 8.51041749e-02 -3.75160962e-01 4.03342843e-01
-1.88112721e-01 3.63839030e-01 1.14345586e+00 -9.94081795e-01
-7.24999130e-01 -6.70809865e-01 -7.67425537e-01 -1.01417303e-01
-6.50221348e-01 2.45883569e-01 -5.57675600e-01 -6.44651949e-01
2.76718378e-01 -8.94824088e-01 -1.18055483e-02 -3.28723490e-01
-5.53249121e-01 -7.05160141e-01 1.10167944e+00 -6.53723359e-01
1.08587205e+00 -2.28892183e+00 3.76716971e-01 5.78802645e-01
2.77205676e-01 3.90947580e-01 1.15863077e-01 3.10277820e-01
-2.68329084e-01 -2.63010189e-02 -2.94324681e-02 -6.34951234e-01
-9.04647261e-02 1.08989902e-01 -1.72276080e-01 8.00684571e-01
7.51391575e-02 3.79754931e-01 -7.99458802e-01 -6.28147006e-01
3.68959814e-01 3.29067856e-01 -1.41820043e-01 -1.07297696e-01
2.26711884e-01 3.97111595e-01 -1.74364001e-01 3.90668154e-01
5.20695090e-01 -2.08423331e-01 4.02784079e-01 -4.17307466e-02
-4.88155857e-02 -5.68190888e-02 -1.62676716e+00 1.20225155e+00
1.94835916e-01 9.70387340e-01 -2.25070938e-01 -1.41321564e+00
1.07048488e+00 2.73272723e-01 6.00357175e-01 -2.73522913e-01
3.08592290e-01 2.46505693e-01 5.18094189e-02 -3.59758884e-01
6.12419665e-01 3.01687479e-01 9.00662839e-02 6.36554956e-01
9.42330360e-02 4.17065799e-01 4.64253366e-01 2.23797187e-01
1.03507781e+00 -2.40541965e-01 2.25213960e-01 -2.70326853e-01
8.29678714e-01 -3.21366936e-01 1.47661015e-01 8.56421769e-01
-1.86725616e-01 6.22748792e-01 7.95670927e-01 -4.77009505e-01
-7.66014278e-01 -6.84832931e-01 -1.65042624e-01 9.89363611e-01
-1.28725305e-01 -2.91696489e-01 -9.29743648e-01 -9.45016265e-01
1.24532357e-01 3.71711046e-01 -7.48466790e-01 6.34152396e-03
-4.64886636e-01 -9.30017054e-01 5.76583683e-01 3.53728920e-01
2.42716402e-01 -1.02932000e+00 -6.92141354e-01 3.81361842e-02
-3.79085466e-02 -7.63727546e-01 -2.61386507e-03 3.91052157e-01
-8.63825798e-01 -1.31925619e+00 -6.28227353e-01 -7.49071240e-01
7.48083770e-01 -4.24443185e-02 7.80044377e-01 3.30897480e-01
-5.39367795e-01 2.81440020e-01 -6.49297059e-01 -1.27113387e-01
-6.19923592e-01 -5.46107907e-03 9.65830460e-02 7.00293422e-01
6.63121641e-01 -2.40302995e-01 -2.84076601e-01 4.67033118e-01
-8.75163972e-01 -5.39741218e-01 5.33471465e-01 6.02129102e-01
4.19892102e-01 5.68620026e-01 1.70638368e-01 -7.10142791e-01
5.82516789e-01 -2.42882445e-01 -5.25458813e-01 2.89142430e-01
-6.24840021e-01 2.19376624e-01 6.31961465e-01 -5.67037404e-01
-7.35492468e-01 3.87465686e-01 1.49452731e-01 -4.75789368e-01
-6.87485456e-01 1.64495841e-01 1.61879621e-02 -4.55342270e-02
5.82884967e-01 4.08740103e-01 5.35220951e-02 -6.49360836e-01
1.60194993e-01 1.02945566e+00 7.07085073e-01 -2.64294297e-01
5.16974151e-01 7.65523970e-01 1.44376487e-01 -1.03711092e+00
-1.53361231e-01 -7.85709381e-01 -1.02142322e+00 -2.07617149e-01
9.53989625e-01 -3.73382777e-01 -9.57044780e-01 8.01954985e-01
-1.08196473e+00 3.37477066e-02 -1.85326174e-01 4.99523580e-01
-3.18966478e-01 1.08695483e+00 -6.69568777e-01 -9.24483240e-01
1.31758703e-02 -1.02956223e+00 1.16452873e+00 1.05935030e-01
-3.60672504e-01 -7.06338465e-01 1.34446263e-01 2.49149784e-01
-9.39295888e-02 2.46299937e-01 5.79108059e-01 -1.14557290e+00
-1.60889581e-01 -4.71976161e-01 -7.32459426e-02 3.11266571e-01
1.67785048e-01 3.29114318e-01 -9.79032815e-01 -3.32542747e-01
1.57762393e-02 -1.03532404e-01 1.23909497e+00 2.18213499e-01
1.28185213e+00 1.73775721e-02 -6.95334852e-01 2.68325001e-01
9.19876516e-01 4.32741076e-01 9.47846174e-01 3.24840575e-01
5.01885355e-01 6.43049121e-01 5.16405582e-01 5.97794950e-01
-1.00623094e-01 7.29105353e-01 1.62222236e-01 7.62539729e-02
9.65351835e-02 4.03353006e-01 2.46341690e-01 4.01482701e-01
-1.74377874e-01 -4.10707772e-01 -8.87270749e-01 5.17722726e-01
-1.93113101e+00 -1.10748374e+00 -5.17823517e-01 2.15317345e+00
4.67350453e-01 6.01818413e-02 3.75420064e-01 8.58894050e-01
9.56587791e-01 -8.00708756e-02 -9.54438075e-02 -4.30571645e-01
3.19425613e-02 2.77716577e-01 1.82436943e-01 1.36129618e-01
-1.56077850e+00 7.46025622e-01 5.76707602e+00 1.04181457e+00
-1.16943514e+00 -7.90188089e-02 7.40487218e-01 7.29632750e-02
4.93015230e-01 -9.03526098e-02 -7.17489779e-01 7.24052966e-01
8.40024173e-01 3.89100134e-01 2.52329022e-01 6.09050512e-01
-2.06467867e-01 -3.72071415e-01 -1.33935010e+00 1.10017991e+00
4.41672206e-01 -1.16961849e+00 -8.25125352e-02 2.85663098e-01
4.79245722e-01 -3.79723340e-01 -1.57311678e-01 -1.95824996e-01
1.38458848e-01 -1.05227482e+00 5.61145246e-01 5.18440366e-01
1.89212501e-01 -8.60064805e-01 8.88083279e-01 2.49722913e-01
-1.04936886e+00 -1.41236216e-01 -1.64084196e-01 5.99139854e-02
-2.96913266e-01 9.11248803e-01 -7.45680869e-01 5.56537747e-01
8.15568030e-01 7.73736060e-01 -8.37566495e-01 1.12829602e+00
4.83766459e-02 5.97962916e-01 -2.71795571e-01 -1.67399555e-01
3.29204947e-02 -9.90523100e-02 4.51693922e-01 1.47053111e+00
2.55607814e-01 -2.48607621e-01 1.14393748e-01 4.89445120e-01
1.15333326e-01 1.31999239e-01 -3.45297545e-01 -1.19462244e-01
2.54789174e-01 1.35664487e+00 -1.58668637e+00 -6.44927979e-01
-1.36914134e-01 1.47073352e+00 1.48980469e-01 -5.46808392e-02
-5.82930744e-01 -6.86505497e-01 2.01966837e-01 -7.60142505e-02
8.17783475e-01 -3.35234880e-01 -2.16323242e-01 -1.10901153e+00
-6.58248216e-02 -7.09466577e-01 7.11949646e-01 -4.75088894e-01
-1.11620259e+00 6.15174770e-01 -2.23819926e-01 -1.19175339e+00
-3.16016793e-01 -8.60539973e-01 -4.61740136e-01 4.88931596e-01
-1.06436825e+00 -6.63740873e-01 -2.47143045e-01 7.36463606e-01
2.12887108e-01 -4.06404644e-01 8.43318164e-01 2.32223913e-01
-7.20034599e-01 4.52083617e-01 5.60189486e-01 4.81503546e-01
9.07596648e-01 -1.38705838e+00 -1.57676592e-01 9.15628076e-01
3.86954337e-01 2.63267994e-01 5.84591389e-01 -5.40258825e-01
-9.26217973e-01 -8.35973918e-01 9.63790596e-01 -4.31757629e-01
4.35635269e-01 -3.67498994e-01 -9.20690954e-01 2.87366331e-01
1.64880127e-01 -1.80254489e-01 8.36387515e-01 1.11075919e-02
-4.10757840e-01 -6.27208203e-02 -1.17575908e+00 2.23792791e-01
4.23923165e-01 -1.94759905e-01 -5.77922583e-01 3.27062935e-01
-3.35219622e-01 1.61763113e-02 -9.23269033e-01 -4.38604020e-02
5.52421510e-01 -1.29021597e+00 7.99618959e-01 -4.22318876e-01
3.50037962e-01 -3.79675597e-01 4.93672416e-02 -1.00943077e+00
-4.34101999e-01 -4.82519507e-01 -3.32907796e-01 1.36583400e+00
-1.45781618e-02 -4.33063358e-01 9.96406198e-01 4.57114317e-02
2.44477585e-01 -3.52245152e-01 -1.05218709e+00 -8.74422550e-01
-1.19555302e-01 -1.33584842e-01 2.82321155e-01 1.19803381e+00
1.82652250e-01 3.97522524e-02 -2.67988920e-01 9.18384641e-03
5.73517978e-01 4.67196763e-01 5.54879844e-01 -1.75931931e+00
-4.02200639e-01 -7.44431257e-01 -8.79995227e-01 -4.77728605e-01
1.01979367e-01 -7.30512679e-01 -4.24603783e-02 -1.13003886e+00
1.69553682e-01 -3.34131420e-01 -2.57255197e-01 5.56141615e-01
3.13652717e-02 5.73679388e-01 2.89430544e-02 4.87529367e-01
-6.99087203e-01 -7.76998624e-02 5.14207542e-01 -4.90222834e-02
-3.62859219e-01 -9.84206051e-02 -4.43492442e-01 6.57090247e-01
7.00749278e-01 -4.93726254e-01 1.79376289e-01 1.66678980e-01
-1.25225291e-01 -3.14727962e-01 3.58070165e-01 -1.14648700e+00
2.92807072e-01 2.34557822e-01 8.26461732e-01 -6.02379680e-01
5.62915504e-02 -7.72875607e-01 -8.10003653e-02 6.86563432e-01
-2.89060771e-01 -1.05266422e-01 6.66387007e-02 5.85262954e-01
-3.08861583e-01 -4.20557171e-01 9.74206507e-01 -9.58194360e-02
-6.57047153e-01 -2.82033890e-01 -9.98928905e-01 -5.43225825e-01
1.36388910e+00 -4.14529979e-01 -3.21291238e-01 -1.48971274e-01
-9.59149778e-01 -1.10379800e-01 4.71092731e-01 3.65853906e-01
4.49853927e-01 -1.03582585e+00 -6.22790217e-01 2.43796945e-01
-2.19172295e-02 -4.88725543e-01 5.18818311e-02 1.09880853e+00
-5.46319842e-01 4.29744035e-01 -2.09647849e-01 -8.94306958e-01
-1.86067891e+00 8.29067349e-01 1.32757872e-01 -5.51474690e-01
-4.84728366e-01 5.36040008e-01 -3.48637164e-01 2.32009646e-02
4.06420052e-01 -2.66911834e-01 -5.89773059e-01 5.35543561e-01
6.99165940e-01 7.19156086e-01 5.04912198e-01 -6.32257164e-01
-6.48008287e-01 5.44466019e-01 -3.13152701e-01 5.39744683e-02
1.22669482e+00 2.25768924e-01 -4.53487486e-01 3.77109677e-01
1.14164865e+00 -1.54414205e-02 -6.55758142e-01 1.38654098e-01
3.61486495e-01 -4.54013169e-01 -1.78160192e-03 -4.94352579e-01
-1.01686883e+00 8.76383245e-01 8.27313781e-01 7.39677250e-01
1.26256549e+00 1.30136639e-01 2.77066827e-01 5.88327229e-01
1.34411052e-01 -1.24559927e+00 2.47048736e-01 2.57860124e-01
3.93061221e-01 -8.96498024e-01 8.26703310e-02 -2.37931490e-01
-5.05625367e-01 1.44100678e+00 1.68135136e-01 -3.02327812e-01
5.43561816e-01 2.41823584e-01 -1.78216413e-01 -2.31710538e-01
-4.43084270e-01 -7.36427605e-02 3.33260059e-01 3.89760703e-01
3.42004091e-01 -5.54947630e-02 -4.48963434e-01 2.33289212e-01
1.87443823e-01 7.12314248e-02 6.04894042e-01 1.03743029e+00
-4.18359786e-01 -1.34899855e+00 -8.49630713e-01 6.40224993e-01
-5.22647083e-01 2.03324705e-01 -9.60763156e-01 6.83461726e-01
2.51340091e-01 1.01424432e+00 4.80307728e-01 -4.09537911e-01
-3.59025188e-02 4.00665760e-01 5.81205368e-01 -3.41262221e-01
-6.23098016e-01 2.19169497e-01 -1.10284992e-01 -4.47618663e-01
-6.14977896e-01 -1.09027648e+00 -1.06642020e+00 -1.92521840e-01
-3.26888591e-01 2.83549637e-01 5.94248414e-01 9.53150630e-01
3.84407997e-01 3.60083848e-01 6.37707889e-01 -1.09410274e+00
-2.02278197e-01 -8.42842996e-01 -8.36068213e-01 6.96906209e-01
6.26633391e-02 -6.32487357e-01 -3.72868210e-01 1.64864451e-01] | [8.13994026184082, 1.7829595804214478] |
4517a41f-c929-4f54-a00d-cc9c94540cc8 | learning-recommendations-from-user-actions-in | 2211.15360 | null | https://arxiv.org/abs/2211.15360v1 | https://arxiv.org/pdf/2211.15360v1.pdf | Learning Recommendations from User Actions in the Item-poor Insurance Domain | While personalised recommendations are successful in domains like retail, where large volumes of user feedback on items are available, the generation of automatic recommendations in data-sparse domains, like insurance purchasing, is an open problem. The insurance domain is notoriously data-sparse because the number of products is typically low (compared to retail) and they are usually purchased to last for a long time. Also, many users still prefer the telephone over the web for purchasing products, reducing the amount of web-logged user interactions. To address this, we present a recurrent neural network recommendation model that uses past user sessions as signals for learning recommendations. Learning from past user sessions allows dealing with the data scarcity of the insurance domain. Specifically, our model learns from several types of user actions that are not always associated with items, and unlike all prior session-based recommendation models, it models relationships between input sessions and a target action (purchasing insurance) that does not take place within the input sessions. Evaluation on a real-world dataset from the insurance domain (ca. 44K users, 16 items, 54K purchases, and 117K sessions) against several state-of-the-art baselines shows that our model outperforms the baselines notably. Ablation analysis shows that this is mainly due to the learning of dependencies across sessions in our model. We contribute the first ever session-based model for insurance recommendation, and make available our dataset to the research community. | ['Christina Lioma', 'Maria Maistro', 'Simone Borg Bruun'] | 2022-11-28 | null | null | null | null | ['session-based-recommendations'] | ['miscellaneous'] | [ 2.93065608e-01 1.49922878e-01 -7.99533129e-01 -6.64805770e-01
-5.59736013e-01 -5.81442297e-01 3.45117569e-01 -3.00652231e-03
-2.89098769e-01 6.56873405e-01 9.79888201e-01 -6.20721996e-01
-3.47190410e-01 -8.18313777e-01 -9.05236185e-01 -3.38894457e-01
-3.48784715e-01 4.68448460e-01 -1.16703704e-01 -5.02466977e-01
1.22565381e-01 -4.74572890e-02 -1.60845292e+00 7.45082259e-01
5.71714759e-01 8.51042569e-01 1.80768371e-01 4.78532732e-01
-7.73273930e-02 6.94363534e-01 -3.13618153e-01 -2.82403946e-01
3.81015122e-01 -3.69399458e-01 -6.82728589e-01 2.07283959e-01
5.42746484e-01 -8.10992301e-01 -5.22290170e-01 3.23806673e-01
2.76906073e-01 5.73720574e-01 6.14442408e-01 -8.04222167e-01
-1.12640905e+00 1.12491608e+00 -1.72476649e-01 1.68270275e-01
3.26331854e-01 -2.38587841e-01 1.53326917e+00 -5.91679633e-01
4.68744099e-01 8.99348199e-01 5.93840301e-01 6.56819344e-01
-1.40371096e+00 -5.33206284e-01 7.90612161e-01 -1.32794623e-04
-7.30326533e-01 -2.91475326e-01 4.87074494e-01 -2.36479580e-01
1.32093811e+00 3.14709932e-01 4.48662966e-01 1.76253772e+00
-3.78661379e-02 9.86744821e-01 5.18460155e-01 -8.63524452e-02
2.75569201e-01 3.34664911e-01 5.27762771e-01 7.97855464e-05
2.74586201e-01 2.30370387e-01 -5.78035235e-01 -3.06399643e-01
8.08606505e-01 7.78223038e-01 -2.25415230e-01 -2.14836434e-01
-8.96564007e-01 1.05236793e+00 3.82147908e-01 4.05877531e-01
-7.78149128e-01 -1.30908668e-01 2.12699205e-01 7.22314119e-01
5.21632195e-01 3.61864537e-01 -9.97577012e-01 -2.50579834e-01
-7.57606030e-01 5.30382276e-01 1.21937430e+00 1.03541362e+00
3.89803976e-01 -9.61818397e-02 -2.65875518e-01 1.06047571e+00
1.60052732e-01 1.61311209e-01 6.57002270e-01 -7.15363979e-01
6.99913979e-01 4.72421795e-01 4.51447904e-01 -7.26417899e-01
-4.89628553e-01 -3.70681077e-01 -7.31555283e-01 -3.19624245e-01
5.64193726e-01 -3.65554631e-01 -8.26963246e-01 1.59020209e+00
-1.35085210e-01 1.56477407e-01 8.35410655e-02 8.97737801e-01
7.49312341e-01 5.96281826e-01 1.30409867e-01 -4.34035987e-01
9.66173053e-01 -1.03534126e+00 -5.60450137e-01 -4.53181922e-01
7.11450696e-01 -4.89917248e-01 1.10225582e+00 5.80550015e-01
-9.49010193e-01 -4.16216224e-01 -6.47991657e-01 8.09813142e-02
-2.90326029e-01 -1.40060738e-01 1.14402759e+00 5.34022629e-01
-7.92667866e-01 8.97995532e-01 -6.40166283e-01 -7.77999520e-01
2.68074274e-01 3.73697490e-01 2.59109288e-02 -1.82467014e-01
-1.16530097e+00 6.18200362e-01 1.10211177e-02 -9.23269168e-02
-4.91093814e-01 -7.35564530e-01 -6.02425635e-01 1.89405829e-01
5.63412666e-01 -3.84017885e-01 1.58160150e+00 -1.08759451e+00
-1.44043350e+00 1.40282243e-01 -1.05864763e-01 -6.58033669e-01
1.70660108e-01 -4.19205129e-01 -8.15177917e-01 -6.13108873e-01
-2.72791684e-01 1.25746369e-01 5.62729120e-01 -8.29689622e-01
-1.01404452e+00 -2.01825365e-01 3.59419495e-01 3.57907146e-01
-5.98760962e-01 -2.52595931e-01 -2.73387343e-01 -6.09229982e-01
-9.76075381e-02 -9.96854782e-01 -4.63401437e-01 -7.95890987e-01
-7.11753592e-03 -4.24707443e-01 4.95936424e-01 -6.58794463e-01
1.45436633e+00 -2.22497606e+00 -7.67184123e-02 3.54850858e-01
-3.69714051e-01 -1.10718064e-01 -3.07590932e-01 8.81109715e-01
-2.22384587e-01 3.42782401e-02 2.30720907e-01 -3.35044026e-01
5.62985055e-02 5.59206545e-01 -5.85502565e-01 1.21000066e-01
-4.86887127e-01 9.17731643e-01 -7.16153383e-01 3.66250902e-01
3.07212602e-02 2.86600113e-01 -8.75172436e-01 3.15708011e-01
-5.34052014e-01 2.43840024e-01 -4.52999204e-01 5.83117306e-01
2.56609976e-01 -5.60497224e-01 4.99115735e-01 5.32499626e-02
4.94765975e-02 9.47304904e-01 -1.17926621e+00 1.90487063e+00
-6.99714720e-01 -1.53533027e-01 -2.91692942e-01 -1.04807818e+00
6.04450941e-01 6.22389078e-01 7.55962849e-01 -8.69244039e-01
-1.29125088e-01 -2.18687262e-02 -3.36802639e-02 -5.00856042e-01
6.22032344e-01 2.06033662e-01 -1.63000971e-01 9.50873852e-01
1.03127295e-02 5.73993742e-01 3.43876898e-01 4.43115860e-01
1.32676494e+00 -5.88308135e-03 1.77757859e-01 6.07400648e-02
-1.08509913e-01 -2.14751914e-01 5.69832504e-01 1.15959203e+00
2.88653880e-01 4.82889175e-01 2.36977205e-01 -5.65775394e-01
-7.86255121e-01 -7.17619896e-01 -6.65593240e-03 1.74215484e+00
-1.63594663e-01 -5.32670617e-01 -1.27725855e-01 -9.33287501e-01
2.84213901e-01 9.37331080e-01 -7.50188410e-01 -8.64192471e-02
-4.19817835e-01 -7.05313504e-01 -3.12752217e-01 6.67345047e-01
-3.29378508e-02 -1.46409321e+00 -3.56341386e-03 7.07609892e-01
-2.03137398e-01 -7.77579904e-01 -9.03763354e-01 3.28795344e-01
-1.04160404e+00 -8.88962030e-01 -5.62798083e-01 -6.11782730e-01
4.48707432e-01 4.81369704e-01 1.53182387e+00 3.98009941e-02
1.03905901e-01 6.31601334e-01 -5.76046944e-01 -3.14238489e-01
-1.35672539e-01 3.33668828e-01 2.02838838e-01 9.79208425e-02
8.23328137e-01 -9.29374874e-01 -9.06147480e-01 3.88196349e-01
-8.02930593e-01 -1.80141762e-01 7.25041330e-01 8.02656293e-01
2.67560661e-01 3.93369608e-02 9.08238590e-01 -1.64818120e+00
7.06555009e-01 -9.44633842e-01 -2.66100854e-01 -3.47421877e-02
-9.93336022e-01 -2.43911743e-01 6.26704872e-01 -7.82369435e-01
-1.06129766e+00 -9.39863548e-02 -1.00391164e-01 2.60167941e-02
-3.52340639e-01 7.63861477e-01 2.58436471e-01 6.75991476e-01
8.36598456e-01 -9.80436504e-02 -1.77522093e-01 -9.44182217e-01
4.25045341e-01 7.00450242e-01 1.13755144e-01 -2.42732763e-01
4.59648192e-01 1.32336572e-01 -7.99831092e-01 -4.27288592e-01
-1.06276667e+00 -6.78696334e-01 -1.74545348e-01 2.01951250e-01
2.74841458e-01 -8.73398423e-01 -8.56176257e-01 1.62662435e-02
-5.75697482e-01 -5.66245377e-01 -5.97928226e-01 6.97448611e-01
-3.24663192e-01 2.60864245e-03 -1.04900789e+00 -8.94697845e-01
-5.11824310e-01 -5.77419400e-01 4.02482808e-01 1.15269363e-01
-5.91546416e-01 -1.01207650e+00 -4.01601344e-02 3.95752609e-01
6.58996820e-01 -4.06817973e-01 7.67867267e-01 -1.02948248e+00
-4.89465952e-01 -4.15203214e-01 2.19861344e-01 3.35753024e-01
5.67076087e-01 -6.41478240e-01 -5.78337252e-01 -4.57790494e-01
-3.27358633e-01 -2.65280098e-01 8.41192663e-01 6.38122618e-01
1.16398597e+00 -6.38140857e-01 -2.87743986e-01 9.68396291e-02
1.12211525e+00 3.48519146e-01 4.36242700e-01 1.65887401e-01
5.02116382e-01 6.48942590e-01 6.11938119e-01 5.69233119e-01
4.12238300e-01 6.87619686e-01 3.51898313e-01 -1.16109543e-01
3.56932998e-01 -4.38487977e-01 5.45666456e-01 6.35319233e-01
-2.37650901e-01 -4.54804093e-01 -2.09156349e-01 6.11614048e-01
-2.32366562e+00 -1.18060744e+00 4.63369936e-02 2.52096462e+00
7.86146820e-01 2.71887183e-01 5.80964267e-01 -1.51696086e-01
1.37243941e-01 2.10378580e-02 -8.67848694e-01 -5.28035939e-01
2.12150365e-01 1.07243180e-01 7.45796680e-01 2.37398818e-01
-9.97659206e-01 5.78637719e-01 6.94408321e+00 2.90272474e-01
-7.62431085e-01 -4.37258817e-02 5.06319582e-01 -6.68300211e-01
-4.67969060e-01 -2.20829040e-01 -1.00954401e+00 5.98451912e-01
1.29139650e+00 1.00152940e-01 8.86700451e-01 9.32662547e-01
4.34361070e-01 1.47775888e-01 -1.49781716e+00 6.33634150e-01
2.64434889e-03 -1.25258458e+00 -9.80333388e-02 4.35223550e-01
8.11223805e-01 2.73949832e-01 1.87313199e-01 7.65079796e-01
8.70542943e-01 -9.68429685e-01 2.29322717e-01 4.30268437e-01
2.92264938e-01 -7.12079763e-01 6.66934848e-01 5.37686467e-01
-8.88679981e-01 -5.04158735e-01 -3.39981198e-01 -2.83586115e-01
2.79891491e-01 6.17753386e-01 -8.08900356e-01 4.22256887e-01
9.84743714e-01 1.28338134e+00 -2.02667192e-01 7.20394611e-01
1.54730931e-01 1.00870383e+00 -2.94863880e-01 1.48905694e-01
2.34314233e-01 -4.75264549e-01 8.28557312e-02 9.91854668e-01
4.33792621e-01 2.91845083e-01 3.99774790e-01 3.90630782e-01
-2.75103897e-01 2.27723405e-01 -7.18342841e-01 -6.63423017e-02
1.03676938e-01 1.15529180e+00 -2.93520212e-01 -1.57282516e-01
-9.45792377e-01 9.26096499e-01 2.97243565e-01 7.14555264e-01
-5.27755499e-01 2.22331867e-01 8.75130653e-01 3.29360425e-01
6.26074731e-01 1.26145110e-01 -7.22011700e-02 -1.03242123e+00
-4.82750535e-02 -1.14929962e+00 7.98397005e-01 -2.36131206e-01
-1.68267167e+00 2.13295877e-01 -2.71382511e-01 -1.07473207e+00
-6.57632768e-01 -1.56935081e-01 -4.39397931e-01 8.21440578e-01
-1.23543322e+00 -8.61525834e-01 1.88616082e-01 7.32559502e-01
8.78194571e-01 -2.54233360e-01 1.11778045e+00 6.98213577e-01
-3.70033145e-01 7.80970395e-01 3.27524722e-01 -2.97763161e-02
8.62832129e-01 -1.17210674e+00 4.97610271e-01 4.39036340e-01
3.84463251e-01 9.23248529e-01 7.35630453e-01 -5.91006398e-01
-1.46255076e+00 -9.90406096e-01 1.04706740e+00 -7.98767805e-01
4.67511505e-01 -4.74611729e-01 -1.01821804e+00 1.21766603e+00
7.69807845e-02 -3.72651249e-01 9.86400723e-01 1.05311930e+00
-3.60651076e-01 -1.37789264e-01 -8.11222136e-01 6.38755202e-01
1.33375061e+00 -3.66807640e-01 -4.22602326e-01 4.96241122e-01
8.03218961e-01 -2.74726629e-01 -8.51768672e-01 7.95532390e-02
7.39166021e-01 -7.74518847e-01 9.98739839e-01 -9.34275448e-01
3.58614415e-01 3.21710408e-01 -1.09220438e-01 -1.38053679e+00
-7.50295460e-01 -5.65776348e-01 -4.69536304e-01 1.18120694e+00
8.87783527e-01 -5.08217335e-01 9.97197211e-01 1.17408252e+00
1.38313358e-03 -6.49011552e-01 -3.66772026e-01 -5.14276803e-01
-2.74482399e-01 -3.95695657e-01 6.51577652e-01 8.49373043e-01
2.50786841e-01 5.60032070e-01 -1.02092266e+00 -3.54986116e-02
2.17236578e-01 3.32324386e-01 5.65105259e-01 -1.27834558e+00
-8.45283687e-01 -1.53506950e-01 3.86303604e-01 -1.42068815e+00
-1.91363454e-01 -7.35578537e-01 -2.19403625e-01 -1.70698452e+00
6.20503128e-02 -5.83029926e-01 -8.18478584e-01 6.97994173e-01
2.63748050e-01 -3.74825522e-02 -1.65250480e-01 1.27445087e-01
-4.75480497e-01 7.14471787e-02 9.55685437e-01 -3.32372598e-02
-7.93288291e-01 7.33239412e-01 -1.29938388e+00 3.82882625e-01
6.71271861e-01 -3.94989282e-01 -6.79414511e-01 -1.88159838e-01
3.89494598e-01 1.99478656e-01 -3.07436496e-01 -4.09527063e-01
1.13200165e-01 -3.47052991e-01 2.31048271e-01 -6.15440607e-01
3.99390489e-01 -1.10696232e+00 2.93071717e-01 -6.71013538e-03
-7.91897058e-01 -2.98750222e-01 -2.21404240e-01 1.00310063e+00
1.88464984e-01 -2.68172491e-02 3.49865817e-02 -2.42537141e-01
-5.75947165e-01 5.89042723e-01 -5.41886210e-01 -2.06281900e-01
5.64900577e-01 -7.94575736e-02 -1.72694325e-02 -7.75142074e-01
-9.78139758e-01 2.69144416e-01 9.83201060e-03 9.77043331e-01
2.11672738e-01 -1.14106929e+00 -5.76783299e-01 2.27982596e-01
1.63553610e-01 -3.01280856e-01 4.30036902e-01 4.87473369e-01
4.69512105e-01 5.25351167e-01 2.45958045e-01 -1.09368883e-01
-1.19828594e+00 6.82241857e-01 -3.91001105e-02 -5.11115849e-01
-9.33365166e-01 6.77115917e-01 5.20488657e-02 -4.59064573e-01
8.24071348e-01 -4.46727782e-01 -7.14278698e-01 2.31725395e-01
7.09575772e-01 4.80847321e-02 2.31527865e-01 -5.52476235e-02
3.53383496e-02 -1.49805859e-01 -8.10652256e-01 1.80149361e-01
1.74866688e+00 -2.11089492e-01 4.29256409e-01 7.11491644e-01
9.58261609e-01 -2.27505282e-01 -1.21527100e+00 -7.30180681e-01
3.33627462e-02 -3.72225672e-01 -4.02967632e-02 -1.19581378e+00
-1.15306640e+00 1.59290582e-01 5.21515429e-01 5.91758430e-01
8.86973917e-01 3.43607068e-02 1.04666662e+00 4.30763125e-01
3.88525605e-01 -1.28274751e+00 -1.66131020e-01 4.74500299e-01
6.92853630e-01 -1.42570853e+00 -1.02546014e-01 -3.78890000e-02
-8.30373585e-01 6.03282750e-01 4.98173773e-01 -2.52302349e-01
9.43153977e-01 -6.23720884e-02 1.08844392e-01 -1.15404367e-01
-1.18238485e+00 9.35004950e-02 3.19870561e-01 2.22759321e-01
8.76446962e-01 1.71813250e-01 -4.26185489e-01 1.07518697e+00
-1.63572252e-01 2.83429474e-01 4.71278161e-01 9.02938187e-01
-1.34353161e-01 -1.36923707e+00 1.01650327e-01 1.16875386e+00
-6.11953020e-01 -2.27731109e-01 -1.05897347e-02 3.67244214e-01
-1.87286779e-01 1.27138484e+00 1.07658044e-01 -3.32621574e-01
7.16463745e-01 1.89989775e-01 2.71421857e-02 -9.02449846e-01
-1.02510738e+00 1.69799700e-01 3.13325405e-01 -7.90648401e-01
-4.07114208e-01 -6.79300845e-01 -8.41970861e-01 -5.00446379e-01
-1.98565468e-01 3.78056079e-01 4.29773211e-01 8.78131092e-01
5.70909739e-01 5.86003602e-01 7.78809786e-01 -8.00692379e-01
-6.85870171e-01 -1.08442843e+00 -1.08478630e+00 8.05936098e-01
3.43645543e-01 -4.32345152e-01 -5.24438202e-01 -4.82429489e-02] | [10.073685646057129, 5.662058353424072] |
d4f31009-e514-4baf-a96b-2a4131267a93 | locec-local-community-based-edge | 2002.04180 | null | https://arxiv.org/abs/2002.04180v2 | https://arxiv.org/pdf/2002.04180v2.pdf | LoCEC: Local Community-based Edge Classification in Large Online Social Networks | Relationships in online social networks often imply social connections in the real world. An accurate understanding of relationship types benefits many applications, e.g. social advertising and recommendation. Some recent attempts have been proposed to classify user relationships into predefined types with the help of pre-labeled relationships or abundant interaction features on relationships. Unfortunately, both relationship feature data and label data are very sparse in real social platforms like WeChat, rendering existing methods inapplicable. In this paper, we present an in-depth analysis of WeChat relationships to identify the major challenges for the relationship classification task. To tackle the challenges, we propose a Local Community-based Edge Classification (LoCEC) framework that classifies user relationships in a social network into real-world social connection types. LoCEC enforces a three-phase processing, namely local community detection, community classification and relationship classification, to address the sparsity issue of relationship features and relationship labels. Moreover, LoCEC is designed to handle large-scale networks by allowing parallel and distributed processing. We conduct extensive experiments on the real-world WeChat network with hundreds of billions of edges to validate the effectiveness and efficiency of LoCEC. | ['Hongzhao Chen', 'Zongyi Zhang', 'Qian Lin', 'Jun Liao', 'Guohui Ling', 'Chuan Chen', 'Chonggang Song'] | 2020-02-11 | null | null | null | null | ['local-community-detection'] | ['graphs'] | [-5.15255965e-02 -1.43340394e-01 -5.14437258e-01 -4.76284683e-01
2.59648293e-01 -5.29370487e-01 4.11849976e-01 5.01774311e-01
-9.41340327e-02 6.15835369e-01 -6.93277866e-02 -3.33979636e-01
-4.93599772e-01 -1.31073856e+00 -5.62239438e-02 -1.96587101e-01
-5.78412652e-01 5.09303987e-01 3.44393939e-01 -1.25902146e-01
1.28236398e-01 3.38649899e-01 -1.23055339e+00 3.08789462e-01
8.64120007e-01 9.60912108e-01 -2.22964555e-01 3.89256120e-01
-1.08696640e-01 5.99624991e-01 -2.04851031e-01 -6.75335050e-01
3.30767840e-01 -2.89020449e-01 -6.01230264e-01 2.14743957e-01
2.16591015e-01 7.36158788e-02 -3.19707245e-01 9.67789173e-01
1.10585541e-01 -9.87999365e-02 5.20231485e-01 -1.81249297e+00
-3.69150639e-01 8.26414227e-01 -9.05114055e-01 2.15629295e-01
5.34508109e-01 -5.09095907e-01 1.56743908e+00 -7.47906864e-01
8.44259441e-01 1.24785531e+00 9.87170041e-01 -1.85310438e-01
-1.07172632e+00 -9.72600877e-01 2.84172058e-01 2.25947127e-01
-1.52624249e+00 -1.19373322e-01 6.91451252e-01 -5.02355039e-01
6.17631197e-01 3.36602598e-01 1.11348379e+00 6.37291849e-01
1.38085196e-02 4.14397538e-01 9.76224363e-01 -1.86742440e-01
-1.04071600e-02 3.36062253e-01 5.71306169e-01 5.74573338e-01
6.50538325e-01 -3.46087486e-01 -5.67403257e-01 -6.21087909e-01
8.80321205e-01 2.31040657e-01 -1.10960580e-01 -4.63266581e-01
-1.06749892e+00 9.50467527e-01 7.70513177e-01 4.16541070e-01
-1.59233585e-01 -1.38516948e-01 4.01796490e-01 6.57380581e-01
7.84048378e-01 4.28962633e-02 -1.72154948e-01 2.33466610e-01
-5.86482704e-01 -2.20580250e-01 1.17784238e+00 1.10183537e+00
8.04742754e-01 -7.96518385e-01 2.30609238e-01 9.83971477e-01
4.23049778e-01 -3.50634195e-03 -4.26388793e-02 -6.42396152e-01
4.13889438e-01 1.03566217e+00 -3.37492079e-01 -1.78166091e+00
-4.50182885e-01 -5.94507754e-01 -1.04140341e+00 -5.78650951e-01
2.46065736e-01 -1.27137646e-01 -1.25013481e-04 1.35389602e+00
6.28838062e-01 4.49195415e-01 -1.58071503e-01 6.60645127e-01
9.57871020e-01 3.30588013e-01 -2.66133189e-01 -3.94827455e-01
1.37393725e+00 -8.73466790e-01 -7.38631964e-01 -6.87424988e-02
6.50250971e-01 -6.35824919e-01 4.73488897e-01 -5.93741238e-02
-6.71675622e-01 -1.19058527e-01 -7.38534808e-01 2.35260576e-01
-3.29710275e-01 -2.58511841e-01 1.41180611e+00 5.60783684e-01
-8.84360492e-01 4.72149134e-01 -4.40332294e-01 -7.89301693e-01
4.87710327e-01 4.81075525e-01 -6.29122913e-01 -8.80422667e-02
-1.17940080e+00 3.11314613e-01 2.22791746e-01 8.98345187e-02
2.72466354e-02 -2.23823622e-01 -7.41894901e-01 2.67096728e-01
6.27975821e-01 -4.82791126e-01 6.49879217e-01 -8.63209069e-01
-7.98432827e-01 8.49711418e-01 -1.46423787e-01 -2.43642092e-01
2.22634658e-01 3.10395330e-01 -6.64035439e-01 1.60119161e-01
6.84438422e-02 1.62501514e-01 2.82209903e-01 -1.08627212e+00
-8.03318441e-01 -3.75439346e-01 1.55496359e-01 -4.21309669e-05
-9.16328251e-01 2.95910776e-01 -6.14272833e-01 -5.52392781e-01
6.24644101e-01 -1.12467349e+00 -8.69444683e-02 5.74362054e-02
-5.06149352e-01 -4.97222483e-01 7.30370700e-01 -1.85820073e-01
1.39924276e+00 -2.04747701e+00 -8.90960991e-02 8.82808864e-01
8.08579922e-01 -5.93603477e-02 -1.10021122e-01 5.96759856e-01
-3.86000797e-02 2.45168224e-01 1.70372084e-01 -1.93090484e-01
-2.83611596e-01 1.69373900e-01 2.26358369e-01 4.68657583e-01
-1.81189016e-01 8.76929522e-01 -1.13664520e+00 -7.86984086e-01
-3.97436231e-01 3.91169578e-01 -6.85442388e-01 4.12386805e-02
2.96366841e-01 3.24242294e-01 -6.39326692e-01 7.19304502e-01
6.45226717e-01 -5.99139631e-01 8.77108097e-01 -1.78565398e-01
5.31655513e-02 2.46745646e-01 -1.14210057e+00 1.08190179e+00
-1.22542411e-01 6.69690073e-01 4.29176778e-01 -1.05728853e+00
1.05052817e+00 1.47333115e-01 8.84925842e-01 -6.62095547e-02
2.87588805e-01 3.17133039e-01 2.49449149e-01 -3.21840376e-01
3.13715935e-01 3.04600567e-01 7.21883178e-02 7.74954677e-01
-1.95364594e-01 3.88487130e-01 4.08764422e-01 7.00665534e-01
1.35448074e+00 -4.55985904e-01 3.97490561e-01 -3.38118374e-01
3.95036966e-01 -8.91065374e-02 7.70521700e-01 5.23268461e-01
-2.09780499e-01 1.78164214e-01 7.60454237e-01 -2.98622519e-01
-4.72123176e-01 -7.19915688e-01 -1.55648097e-01 9.83842611e-01
3.91108006e-01 -8.80826175e-01 -2.75190592e-01 -9.71385777e-01
2.79735088e-01 -2.90763736e-01 -3.17111820e-01 -3.41567062e-02
-4.25723881e-01 -1.00437975e+00 -5.69575839e-03 2.17801347e-01
6.18905485e-01 -9.11355615e-01 2.84767598e-01 1.69567913e-01
-2.39025697e-01 -1.45183611e+00 -5.69689572e-01 -2.22658649e-01
-7.82620072e-01 -1.40073276e+00 -6.43752748e-03 -1.14725447e+00
9.10337329e-01 8.55191886e-01 1.22363555e+00 6.27421379e-01
-8.23434293e-02 2.08005339e-01 -5.16325593e-01 2.29257867e-01
6.70417910e-03 3.82129639e-01 1.79396689e-01 4.59884316e-01
6.62636518e-01 -1.09820855e+00 -5.79013348e-01 7.72129118e-01
-4.70475882e-01 -8.33351072e-03 5.41005611e-01 4.70250875e-01
1.59269705e-01 4.31155592e-01 8.56030643e-01 -1.47970915e+00
8.42379808e-01 -9.36153531e-01 -3.42434466e-01 2.73333311e-01
-7.61765480e-01 -4.88660127e-01 1.12378851e-01 -6.12696171e-01
-6.32706642e-01 -2.39598393e-01 2.12909922e-01 1.54129833e-01
3.58449847e-01 9.78463292e-01 -4.86315116e-02 -2.80050725e-01
3.09413433e-01 -3.23940784e-01 -2.23827183e-01 -3.35376889e-01
8.92729461e-02 1.07913589e+00 -2.37133652e-02 -4.06262130e-01
9.10372078e-01 4.58255619e-01 -4.40143161e-02 -5.79143763e-01
-8.72091472e-01 -8.86277676e-01 -6.90476060e-01 -3.45686406e-01
3.94250333e-01 -9.31649208e-01 -9.62212205e-01 2.68007189e-01
-9.08921182e-01 -3.62614356e-02 1.91737950e-01 4.28311646e-01
1.15385249e-01 5.01192093e-01 -1.01795757e+00 -7.78888643e-01
-3.02498400e-01 -5.98970413e-01 5.01223385e-01 2.73566451e-02
-2.97319442e-01 -1.13459396e+00 -2.58111767e-02 5.46483278e-01
3.17547202e-01 2.33034700e-01 6.99813485e-01 -7.36290455e-01
-6.55412555e-01 -4.67549115e-01 -8.53523254e-01 -6.74528331e-02
3.23314548e-01 5.14324987e-03 -3.69752765e-01 -3.78872514e-01
-4.75458264e-01 -9.72326249e-02 7.77018070e-01 -9.59442630e-02
8.45939636e-01 -9.07362849e-02 -7.49334872e-01 3.13175112e-01
1.04433978e+00 -7.19218887e-03 3.61448079e-01 -3.44092096e-03
8.32310796e-01 9.90320802e-01 6.58693373e-01 4.39904928e-01
6.02704704e-01 6.94587946e-01 2.96173781e-01 -1.20967925e-01
5.90321682e-02 -1.10731363e-01 1.78786620e-01 1.17646575e+00
-2.00311407e-01 -3.40649754e-01 -7.08780468e-01 3.89465213e-01
-2.22603989e+00 -8.36314857e-01 -6.43025398e-01 1.96872616e+00
8.32692742e-01 3.32321405e-01 2.36797705e-01 2.14072600e-01
1.25824451e+00 -1.71269521e-01 -3.27875525e-01 7.56482333e-02
-8.61138292e-03 -3.36430997e-01 2.18445331e-01 3.47087890e-01
-9.02134717e-01 7.95347095e-01 5.82583714e+00 5.66236556e-01
-5.76948524e-01 8.63927826e-02 5.53372145e-01 2.93167055e-01
-1.77588210e-01 2.74014324e-01 -8.65126073e-01 3.26965332e-01
6.06056690e-01 3.34022124e-03 3.69777590e-01 6.89167321e-01
1.26299381e-01 8.14166963e-02 -9.48255539e-01 9.40905333e-01
-9.62253287e-02 -1.05230582e+00 -2.19351307e-01 4.63015050e-01
7.07014322e-01 -7.45562240e-02 -3.61104697e-01 2.54351646e-01
3.35777819e-01 -6.65246129e-01 1.04506686e-01 9.50533375e-02
5.33554196e-01 -5.46618104e-01 7.57428229e-01 2.41594657e-01
-1.70197856e+00 -1.86881498e-01 -2.34642893e-01 -3.78920734e-01
1.18028902e-01 1.08548784e+00 -6.65012002e-01 6.02104366e-01
7.00842857e-01 1.43190134e+00 -5.11836886e-01 1.17668140e+00
5.11996821e-02 4.80949223e-01 -6.31478667e-01 8.88295658e-03
-9.60981101e-02 -7.14533448e-01 3.68541688e-01 9.96394455e-01
1.68554351e-01 1.31852493e-01 5.01131415e-01 4.83590454e-01
-4.23528641e-01 4.70418602e-01 -5.76070428e-01 -2.54482985e-01
8.11771572e-01 1.69985282e+00 -1.16665113e+00 -2.14676961e-01
-7.04186678e-01 6.16403461e-01 6.11619711e-01 1.83855847e-01
-5.92314720e-01 -3.07657361e-01 5.49514294e-01 5.68705618e-01
1.69211756e-02 -3.31763387e-01 4.18755412e-02 -1.51132727e+00
4.68925051e-02 -5.47034144e-01 6.05891049e-01 -3.58132601e-01
-1.70893919e+00 6.40255094e-01 -8.46708864e-02 -1.12383926e+00
2.83954382e-01 -2.95161009e-01 -7.51874149e-01 2.27164119e-01
-1.37722468e+00 -1.17104197e+00 -6.52766109e-01 6.14947259e-01
-2.41107531e-02 -1.24842346e-01 5.38728535e-01 7.96744585e-01
-8.54039550e-01 5.66274822e-01 -1.90751210e-01 4.65122402e-01
7.29839444e-01 -8.94411147e-01 1.73774421e-01 4.07945067e-01
1.78796127e-01 8.62265825e-01 3.01953256e-01 -9.98308003e-01
-1.00835407e+00 -1.06924069e+00 1.19550478e+00 -6.17109835e-02
1.10650682e+00 -7.09085941e-01 -7.57069647e-01 8.33706260e-01
-4.09002639e-02 4.84208018e-01 9.76567805e-01 6.91611409e-01
-4.40951526e-01 -2.54816562e-01 -1.04831982e+00 5.31895161e-01
1.73208594e+00 -4.72847819e-01 6.26302287e-02 7.03910172e-01
4.68485802e-01 3.34227949e-01 -1.29386199e+00 4.30459380e-01
7.05648243e-01 -8.40880454e-01 9.65677559e-01 -2.72080809e-01
1.25911981e-01 -1.55554920e-01 1.05684683e-01 -1.02223468e+00
-3.71426463e-01 -6.39768720e-01 -1.53512493e-01 1.47707069e+00
3.74651462e-01 -1.13609123e+00 9.90775883e-01 3.87695432e-01
4.67736542e-01 -7.34884977e-01 -7.09196270e-01 -5.70342362e-01
-5.27802706e-01 -3.72589007e-02 4.89352584e-01 1.58163261e+00
3.96456003e-01 9.46691334e-01 -2.89264530e-01 -9.36858505e-02
7.34118044e-01 5.61051011e-01 7.56022811e-01 -1.99494135e+00
-2.74112642e-01 -3.06437075e-01 -6.30816460e-01 -1.02009857e+00
1.90828204e-01 -9.62628365e-01 -5.44447362e-01 -1.49473536e+00
6.48435712e-01 -1.26190543e+00 -1.04510948e-01 5.72176874e-01
-4.05298918e-02 5.78652382e-01 -1.57332197e-01 6.32500768e-01
-9.39384103e-01 1.72595948e-01 1.16997087e+00 -1.62368298e-01
-3.65941346e-01 2.86609739e-01 -7.32212722e-01 5.95080078e-01
7.42993295e-01 -5.62409937e-01 -5.14856279e-01 2.40761116e-02
8.16455483e-01 6.91345185e-02 9.34938416e-02 -6.20503008e-01
4.33105588e-01 -9.66043398e-02 -1.08398259e-01 -4.72601622e-01
2.54575998e-01 -1.14599741e+00 3.29357207e-01 1.79054901e-01
-2.36588612e-01 -2.11380973e-01 -5.70618033e-01 8.66368413e-01
-2.60947436e-01 -7.08496757e-03 4.10879165e-01 3.20410952e-02
-2.88328975e-01 4.61956114e-01 -1.14977203e-01 -2.74876326e-01
1.13981414e+00 -1.60004035e-01 -4.00582820e-01 -5.82471073e-01
-9.59375739e-01 5.50105512e-01 3.94776791e-01 2.77546346e-01
4.16151613e-01 -1.28624403e+00 -8.12284768e-01 1.47146881e-01
2.70050406e-01 -1.97960377e-01 -2.14580029e-01 1.05262232e+00
-2.71902174e-01 4.78637181e-02 9.96280164e-02 -5.09853303e-01
-1.66944611e+00 4.59411353e-01 5.52652441e-02 -4.73665327e-01
-5.79981029e-01 5.89414537e-01 6.02074619e-03 -5.06431699e-01
1.26653790e-01 1.18336141e-01 -5.79809427e-01 4.02357072e-01
1.41696036e-01 1.51205212e-01 -8.67533907e-02 -6.30744219e-01
-2.54002362e-01 4.42206353e-01 -1.97068319e-01 3.78709853e-01
1.40001547e+00 -2.89062381e-01 -7.57418156e-01 2.29380086e-01
1.17794001e+00 1.01723574e-01 -5.25489509e-01 -6.00983024e-01
3.19300383e-01 -6.56227171e-01 -1.50820138e-02 -2.29037881e-01
-1.45911109e+00 3.28189552e-01 -4.35566828e-02 8.64581704e-01
8.87192428e-01 1.79301828e-01 9.70999777e-01 4.50715095e-01
6.29194498e-01 -6.86480880e-01 5.30370921e-02 3.38140339e-01
4.46653187e-01 -1.26103342e+00 3.41193199e-01 -1.57680655e+00
-2.92615116e-01 7.79174566e-01 6.10897481e-01 -7.25474656e-02
1.22248459e+00 1.07470043e-02 -3.50626498e-01 -5.01153648e-01
-6.70121372e-01 -2.25858018e-01 1.77329540e-01 2.75521338e-01
4.16777700e-01 -5.14227571e-03 -6.33836508e-01 4.51725423e-01
1.13586843e-01 -3.34640771e-01 3.32844734e-01 9.34544325e-01
-3.77645999e-01 -1.44829035e+00 -9.61355716e-02 8.03111970e-01
-2.29460388e-01 -1.51837468e-01 -7.93066859e-01 5.70932031e-01
-2.77955784e-03 1.19288635e+00 1.77169710e-01 -5.81545711e-01
3.02575100e-02 -4.04112101e-01 2.87993532e-02 -8.65952194e-01
-7.40110099e-01 1.41505554e-01 5.23009777e-01 -2.36180812e-01
-7.03623533e-01 -5.95884204e-01 -1.04268706e+00 -6.23538733e-01
-8.50953341e-01 5.24208248e-01 3.53665203e-01 6.50342524e-01
4.32445228e-01 2.01151490e-01 9.27422702e-01 -4.96802807e-01
1.90706939e-01 -9.72316921e-01 -7.85761714e-01 6.00145400e-01
-1.78256333e-01 -7.93798506e-01 -6.38860285e-01 -2.66584009e-01] | [7.390580654144287, 6.242337703704834] |
ab5e6a59-ecd7-4409-a42e-93f2abdbad0d | on-the-safety-of-interpretable-machine-1 | 2211.01498 | null | https://arxiv.org/abs/2211.01498v1 | https://arxiv.org/pdf/2211.01498v1.pdf | On the Safety of Interpretable Machine Learning: A Maximum Deviation Approach | Interpretable and explainable machine learning has seen a recent surge of interest. We focus on safety as a key motivation behind the surge and make the relationship between interpretability and safety more quantitative. Toward assessing safety, we introduce the concept of maximum deviation via an optimization problem to find the largest deviation of a supervised learning model from a reference model regarded as safe. We then show how interpretability facilitates this safety assessment. For models including decision trees, generalized linear and additive models, the maximum deviation can be computed exactly and efficiently. For tree ensembles, which are not regarded as interpretable, discrete optimization techniques can still provide informative bounds. For a broader class of piecewise Lipschitz functions, we leverage the multi-armed bandit literature to show that interpretability produces tighter (regret) bounds on the maximum deviation. We present case studies, including one on mortgage approval, to illustrate our methods and the insights about models that may be obtained from deviation maximization. | ['Moninder Singh', 'Elizabeth M. Daly', 'Kush R. Varshney', 'Amit Dhurandhar', 'Rahul Nair', 'Dennis Wei'] | 2022-11-02 | on-the-safety-of-interpretable-machine | https://openreview.net/forum?id=Jt8FYFnyTLR | https://openreview.net/pdf?id=Jt8FYFnyTLR | null | ['additive-models', 'interpretable-machine-learning'] | ['methodology', 'methodology'] | [ 2.87317425e-01 7.11804628e-01 -8.92102540e-01 -5.54455280e-01
-1.18462443e+00 -8.54198992e-01 2.50269175e-01 4.00461793e-01
-1.16304405e-01 9.28760231e-01 4.75873291e-01 -8.50129187e-01
-7.58736551e-01 -4.71086472e-01 -8.32263231e-01 -5.91992974e-01
8.74636695e-03 3.37318212e-01 -7.06706405e-01 2.79284209e-01
1.92795232e-01 4.68521029e-01 -1.00520623e+00 3.58771354e-01
9.38523293e-01 1.04575467e+00 -7.45314181e-01 3.97791922e-01
3.19612145e-01 7.95474231e-01 -2.04453766e-01 -6.29756808e-01
4.21495438e-01 -3.28772902e-01 -1.00656891e+00 2.01833978e-01
6.67233914e-02 -4.88591433e-01 1.16669051e-01 8.14235687e-01
2.29765233e-02 9.73179191e-02 1.06298792e+00 -1.56589699e+00
-6.07762218e-01 9.64306116e-01 -4.33807999e-01 4.84908670e-02
6.35243580e-02 4.35972819e-03 1.55975187e+00 -2.59594113e-01
9.93135497e-02 1.28114784e+00 6.95520997e-01 4.48754758e-01
-1.72466350e+00 -3.97601545e-01 4.95693505e-01 -1.55466959e-01
-9.45788264e-01 -4.97080773e-01 5.77266872e-01 -5.41129947e-01
5.80934048e-01 7.78712153e-01 4.57215786e-01 1.02801442e+00
-6.50434103e-03 8.64757299e-01 1.21134114e+00 -6.01576746e-01
6.33550704e-01 2.38885313e-01 7.80185163e-01 5.79109728e-01
6.86093569e-01 4.11585450e-01 -5.24300992e-01 -4.54238415e-01
4.78969604e-01 1.73563868e-01 -2.80572116e-01 -3.25400054e-01
-5.05069196e-01 1.36485887e+00 6.27870858e-02 -3.37447554e-01
-3.34878176e-01 5.52733004e-01 3.10534418e-01 2.06373125e-01
6.90191805e-01 4.87610936e-01 -4.39076126e-01 -9.50639471e-02
-6.15851939e-01 1.87142015e-01 9.02573764e-01 6.41130686e-01
3.63635629e-01 -1.80897951e-01 -3.63449454e-01 4.04366791e-01
5.80802143e-01 3.78811896e-01 -4.86777335e-01 -1.29246664e+00
6.60920441e-01 3.02988142e-01 6.67637527e-01 -5.98833501e-01
-2.65900791e-01 -5.11907637e-01 -5.07880151e-01 3.57945949e-01
7.23007500e-01 -3.26045334e-01 -6.18577182e-01 1.85651720e+00
1.56837747e-01 -2.64708489e-01 -8.23368356e-02 6.56323016e-01
-2.21647277e-01 5.09333968e-01 3.28135163e-01 -6.59912765e-01
1.02045250e+00 -6.68891311e-01 -5.56134701e-01 -2.14250296e-01
1.07519865e+00 -3.50422077e-02 1.17110825e+00 7.65009403e-01
-1.03230822e+00 3.23767781e-01 -9.67051387e-01 8.47698823e-02
8.19091275e-02 -5.35248876e-01 9.18280900e-01 1.09847188e+00
-6.27219737e-01 6.36946142e-01 -9.23309803e-01 3.82266529e-02
7.33117044e-01 1.94471866e-01 -4.10916433e-02 2.41995946e-01
-8.97366047e-01 1.07999468e+00 -1.50194764e-02 2.17998996e-02
-8.03642869e-01 -8.96258116e-01 -6.47725105e-01 2.69813389e-01
4.37316149e-01 -8.36314976e-01 1.46517241e+00 -1.27997208e+00
-1.12018239e+00 5.69678009e-01 -8.34261328e-02 -9.84844089e-01
7.05892086e-01 -4.41353172e-01 1.15620255e-01 -1.19713321e-01
-1.36953399e-01 2.01582238e-01 6.80322289e-01 -1.22929585e+00
-6.08060598e-01 -4.20148909e-01 4.11534548e-01 9.04933289e-02
-3.59663159e-01 1.63646579e-01 3.73074234e-01 -6.74739659e-01
-2.55515277e-01 -9.60145473e-01 -3.54124337e-01 4.97190617e-02
-6.12622797e-01 -8.42852145e-02 6.20113760e-02 -6.91773713e-01
1.72955287e+00 -1.75802004e+00 -1.44293740e-01 5.06056488e-01
1.51167512e-01 -4.15029317e-01 1.78533286e-01 2.51461834e-01
-3.29020590e-01 6.40121996e-01 -5.15938938e-01 -4.34615880e-01
4.73725796e-01 2.71954805e-01 -6.63542211e-01 7.15288401e-01
6.54257312e-02 8.55307221e-01 -4.97223765e-01 -4.96198945e-02
-1.00297920e-01 3.19977087e-04 -7.02335119e-01 -1.58333555e-01
-1.93647936e-01 3.25999498e-01 -7.43696749e-01 3.35685998e-01
4.04874086e-01 -2.69744068e-01 7.40441233e-02 3.08591515e-01
2.19232425e-01 2.73247570e-01 -1.00466740e+00 1.01306140e+00
-6.25019372e-01 4.00762320e-01 -3.59380692e-02 -1.18734396e+00
4.39106405e-01 2.35342473e-01 2.43799284e-01 -6.14523254e-02
-3.79340560e-03 -2.86700297e-02 -1.31748125e-01 -2.91083694e-01
-1.39427120e-02 -6.73966467e-01 -2.00501770e-01 9.02630448e-01
-8.06487858e-01 7.75961354e-02 -3.51274252e-01 -6.59631044e-02
1.02173793e+00 -1.72106832e-01 5.47284245e-01 -5.62070787e-01
3.31005976e-02 -1.74513876e-01 4.77364898e-01 9.88157034e-01
-5.14183193e-02 4.96908575e-01 9.97403920e-01 -5.78310668e-01
-9.21434343e-01 -1.24552906e+00 -2.39764661e-01 1.02745783e+00
-2.73681849e-01 -1.73123181e-01 -8.32796752e-01 -9.69824195e-01
4.92601633e-01 1.38217115e+00 -9.15508807e-01 -1.44741818e-01
-1.05648443e-01 -8.59557152e-01 4.16210353e-01 6.99869990e-01
-2.52939388e-02 -3.84865820e-01 -6.12519085e-01 -2.99520399e-02
-5.09930588e-02 -6.58994198e-01 -4.59203601e-01 4.19546604e-01
-1.16453099e+00 -9.53341484e-01 -1.55480057e-01 5.05612120e-02
7.26709962e-01 -1.89782113e-01 1.07494748e+00 -2.95469481e-02
2.80240595e-01 6.34718180e-01 -1.83178261e-01 -8.36986661e-01
-5.10055780e-01 -8.86477530e-02 1.51892141e-01 1.05473720e-01
2.03648835e-01 -3.86471421e-01 -4.77794796e-01 1.08965971e-01
-9.15081263e-01 1.44083738e-01 1.33533835e-01 7.70655513e-01
4.60550874e-01 -9.54111591e-02 7.58262694e-01 -1.01004565e+00
7.86607563e-01 -7.18536913e-01 -7.87572503e-01 4.82838243e-01
-1.17232811e+00 5.32925487e-01 5.57216048e-01 -2.00798035e-01
-1.03851128e+00 9.00449231e-02 3.88550192e-01 -1.37090161e-02
5.29258046e-03 6.71621859e-01 -1.74326077e-01 1.94210306e-01
8.71834099e-01 -2.95627087e-01 -1.89577669e-01 -5.33837855e-01
6.19247198e-01 7.50584781e-01 -5.79029620e-02 -9.89986897e-01
6.88759506e-01 4.86890584e-01 1.62434340e-01 -2.59798884e-01
-1.28249562e+00 1.11937039e-01 -2.89082319e-01 -6.94588944e-03
6.35927022e-01 -5.23456812e-01 -7.79218495e-01 -1.60967231e-01
-1.04390121e+00 -4.27579790e-01 -5.42924166e-01 4.98734087e-01
-9.42368984e-01 2.03604758e-01 -3.62068206e-01 -1.61601758e+00
-4.03841734e-01 -8.54699314e-01 9.13573682e-01 -2.35204920e-01
-7.89541185e-01 -1.33657122e+00 -1.45244882e-01 6.44153595e-01
-5.32730371e-02 5.28697968e-01 1.30169809e+00 -1.05795145e+00
-6.68584585e-01 -2.35584721e-01 8.42860192e-02 4.17981207e-01
3.34400404e-03 -1.02054559e-01 -1.00716770e+00 -6.87308162e-02
1.49210826e-01 -2.47198895e-01 7.58844972e-01 8.07625592e-01
1.36515987e+00 -1.14865160e+00 -1.46025717e-01 4.93788064e-01
1.13811338e+00 1.93446130e-01 3.56438696e-01 3.97282064e-01
2.79079050e-01 7.14709878e-01 6.92529142e-01 6.82873130e-01
1.60427690e-01 4.80029762e-01 4.24501479e-01 1.06684133e-01
8.11796188e-01 -3.50736529e-01 3.95655692e-01 1.08363256e-02
-8.83289576e-02 -1.69780388e-01 -8.82721186e-01 3.59608918e-01
-2.06144309e+00 -1.03476012e+00 5.64218275e-02 2.53589010e+00
8.79589200e-01 4.94594306e-01 4.94913638e-01 3.32863122e-01
6.26081228e-01 -1.40507057e-01 -6.66954339e-01 -1.00247514e+00
1.61531731e-01 1.44307896e-01 6.76938653e-01 1.00091052e+00
-9.87578928e-01 3.26635838e-01 7.60411119e+00 9.47767317e-01
-5.29430389e-01 1.10426798e-01 1.57983196e+00 -4.49593872e-01
-9.17612910e-01 1.92304239e-01 -4.55882162e-01 4.43718404e-01
1.07794690e+00 -5.16221523e-01 6.04727745e-01 1.02315331e+00
7.22261429e-01 -1.08116008e-01 -1.93636334e+00 6.60533607e-01
-5.59826076e-01 -1.22848010e+00 -9.04683992e-02 2.41145074e-01
8.01840127e-01 -4.48692352e-01 3.61764252e-01 -5.31284362e-02
6.66975796e-01 -1.51500213e+00 9.47553515e-01 4.83812541e-01
5.77458262e-01 -1.07614779e+00 3.10257554e-01 5.17611921e-01
-5.84921420e-01 -4.65156049e-01 -4.10426185e-02 -4.71544236e-01
8.00854638e-02 6.55264318e-01 -6.41296804e-01 2.52043396e-01
2.22482219e-01 5.17990589e-01 -7.58810490e-02 8.84409904e-01
-3.32927167e-01 9.63484228e-01 -5.51967263e-01 1.40623674e-01
2.38973603e-01 -4.51606065e-01 3.68830025e-01 9.96742427e-01
1.08576082e-01 3.59628856e-01 -2.37767130e-01 1.08163738e+00
6.53721243e-02 -1.12772048e-01 -6.48646712e-01 6.80201575e-02
4.11328226e-01 8.49562407e-01 -5.33945143e-01 -9.80925858e-02
-4.37638283e-01 3.79284441e-01 1.70309454e-01 5.85847378e-01
-1.18270278e+00 8.88449401e-02 9.24113154e-01 6.46049455e-02
-9.29813012e-02 8.72112364e-02 -1.17931175e+00 -8.41786623e-01
8.77866596e-02 -9.34556782e-01 7.99428880e-01 -5.99354863e-01
-1.26384091e+00 -4.49466892e-02 5.04719198e-01 -9.46868718e-01
-2.41759717e-01 -5.06711245e-01 -5.68162441e-01 7.59914696e-01
-9.97349620e-01 -8.63638103e-01 4.21189547e-01 5.85623346e-02
4.01528239e-01 3.18501383e-01 6.49986923e-01 -2.64613599e-01
-7.03510702e-01 9.32732522e-01 3.42828572e-01 -2.25558713e-01
7.06568882e-02 -1.29296529e+00 2.05408543e-01 7.29827225e-01
1.11334972e-01 6.88091993e-01 9.88322318e-01 -4.79549795e-01
-1.22991300e+00 -1.13533342e+00 5.15567183e-01 -9.89066064e-01
7.58977056e-01 -6.95590377e-02 -4.85881269e-01 1.19076991e+00
-1.79050580e-01 -2.51581907e-01 8.69225144e-01 4.39879000e-01
-4.66415077e-01 -2.63544023e-01 -1.36787164e+00 7.22316742e-01
1.19256938e+00 -4.85992104e-01 -4.75600719e-01 4.18491095e-01
6.75391495e-01 4.63399887e-02 -1.00662279e+00 3.19414228e-01
6.84093833e-01 -1.00950050e+00 8.59256268e-01 -1.23947465e+00
3.14847529e-01 2.27215648e-01 -2.22206920e-01 -1.16713297e+00
-3.37507308e-01 -1.08828962e+00 -3.89361560e-01 1.14765644e+00
8.07262182e-01 -9.69578445e-01 8.89041424e-01 1.51897144e+00
1.33255869e-01 -1.13221860e+00 -9.69524145e-01 -1.00071311e+00
5.78076899e-01 -8.22842181e-01 6.36928439e-01 7.55116403e-01
4.15377229e-01 4.22825143e-02 -3.16259623e-01 9.54795182e-02
8.71192634e-01 1.35595039e-01 3.18882734e-01 -1.21787012e+00
-4.49344903e-01 -4.23254609e-01 -6.54401034e-02 -1.08367550e+00
3.08519810e-01 -8.82585943e-01 -3.02301198e-01 -9.99172926e-01
5.09241700e-01 -5.20524025e-01 -3.38161528e-01 6.60180628e-01
-1.41511515e-01 -3.75523299e-01 2.33419225e-01 -4.89259623e-02
-4.42356557e-01 3.06505084e-01 6.87056005e-01 -1.40503347e-01
-1.13879167e-01 4.35416967e-01 -1.50005913e+00 9.58985806e-01
8.75289917e-01 -7.10358441e-01 -7.51267850e-01 -3.52241427e-01
4.35720950e-01 2.19413593e-01 4.09544498e-01 -1.92611039e-01
-3.09404910e-01 -7.50258625e-01 2.35411718e-01 -1.94056690e-01
1.26529410e-01 -9.31496322e-01 5.35956681e-01 5.43104649e-01
-1.05902469e+00 -1.12670928e-01 -2.05584578e-02 8.42431724e-01
2.92337418e-01 -4.31936860e-01 7.86149919e-01 2.46782437e-01
1.83478475e-01 3.10417861e-01 -3.41237456e-01 3.32988232e-01
1.11994100e+00 -8.78014192e-02 -1.40190542e-01 -8.97168159e-01
-9.40422952e-01 1.67279944e-01 4.71058220e-01 4.09416482e-02
4.25448179e-01 -1.31099725e+00 -6.82614148e-01 -3.14728290e-01
-3.40935811e-02 -1.38676479e-01 -1.68565720e-01 8.88322055e-01
-2.36737981e-01 6.64918602e-01 5.01163483e-01 -2.35282019e-01
-1.08644938e+00 6.55830443e-01 4.24651712e-01 -3.30417365e-01
-2.08153695e-01 5.24939001e-01 5.79889119e-01 2.28583932e-01
4.37426865e-01 -8.30020845e-01 2.49739408e-01 -1.05017036e-01
3.30578268e-01 7.06712008e-01 -2.42117956e-01 -1.53398186e-01
-3.42204988e-01 2.93161333e-01 1.80293456e-01 -3.45913559e-01
1.46330702e+00 -2.33848810e-01 1.15419820e-01 3.23276967e-01
1.11295390e+00 -8.55700523e-02 -1.46273065e+00 1.70747072e-01
4.75761294e-01 -4.50072497e-01 -3.14490050e-02 -8.65766108e-01
-8.00549150e-01 7.76177049e-01 2.01745585e-01 5.98417342e-01
1.12406957e+00 -1.09023424e-02 3.84821653e-01 4.07418579e-01
3.24836791e-01 -1.03157783e+00 -4.98110175e-01 -3.25844586e-02
1.00124717e+00 -1.06879258e+00 2.23661438e-01 -3.46441448e-01
-7.51598835e-01 9.85405862e-01 1.55551275e-02 1.46390274e-01
7.51570284e-01 9.80415121e-02 -3.36410850e-01 2.03753904e-01
-8.21358562e-01 1.79142579e-01 2.76516765e-01 5.06067395e-01
9.89161208e-02 2.71636516e-01 -3.24423701e-01 1.32153368e+00
-1.80490583e-01 -4.08602804e-02 5.84297717e-01 6.49642527e-01
-5.58790088e-01 -8.81233394e-01 -5.28845847e-01 6.93596244e-01
-8.54943931e-01 -1.80849880e-01 -3.49115193e-01 5.52711964e-01
-5.22881567e-01 1.28741002e+00 -1.29022226e-01 -1.11559644e-01
1.58779323e-01 1.39951661e-01 5.04545391e-01 -4.16867882e-01
-3.26222599e-01 -1.79134190e-01 6.09704852e-01 -4.55890149e-01
-1.10565946e-01 -7.25314617e-01 -1.06692910e+00 -6.80965781e-01
-5.23222864e-01 3.17613214e-01 1.89212933e-01 1.14866960e+00
1.27316698e-01 -9.61235836e-02 9.50916290e-01 -2.14570373e-01
-1.44117069e+00 -4.54313427e-01 -6.91770077e-01 3.81195396e-01
6.20503426e-01 -4.78572339e-01 -8.47441614e-01 8.85274261e-02] | [8.575719833374023, 5.40773868560791] |
7c730ed3-2f27-44dd-ab5c-65756f13ebd4 | offline-rl-with-no-ood-actions-in-sample | 2303.15810 | null | https://arxiv.org/abs/2303.15810v1 | https://arxiv.org/pdf/2303.15810v1.pdf | Offline RL with No OOD Actions: In-Sample Learning via Implicit Value Regularization | Most offline reinforcement learning (RL) methods suffer from the trade-off between improving the policy to surpass the behavior policy and constraining the policy to limit the deviation from the behavior policy as computing $Q$-values using out-of-distribution (OOD) actions will suffer from errors due to distributional shift. The recently proposed \textit{In-sample Learning} paradigm (i.e., IQL), which improves the policy by quantile regression using only data samples, shows great promise because it learns an optimal policy without querying the value function of any unseen actions. However, it remains unclear how this type of method handles the distributional shift in learning the value function. In this work, we make a key finding that the in-sample learning paradigm arises under the \textit{Implicit Value Regularization} (IVR) framework. This gives a deeper understanding of why the in-sample learning paradigm works, i.e., it applies implicit value regularization to the policy. Based on the IVR framework, we further propose two practical algorithms, Sparse $Q$-learning (SQL) and Exponential $Q$-learning (EQL), which adopt the same value regularization used in existing works, but in a complete in-sample manner. Compared with IQL, we find that our algorithms introduce sparsity in learning the value function, making them more robust in noisy data regimes. We also verify the effectiveness of SQL and EQL on D4RL benchmark datasets and show the benefits of in-sample learning by comparing them with CQL in small data regimes. | ['Xianyuan Zhan', 'Victor Wai Kin Chan', 'Zhaoran Wang', 'Zhuoran Yang', 'Jianxiong Li', 'Li Jiang', 'Haoran Xu'] | 2023-03-28 | null | null | null | null | ['q-learning', 'offline-rl', 'd4rl'] | ['methodology', 'playing-games', 'robots'] | [-1.87841758e-01 5.51631488e-02 -7.67552793e-01 -3.72783005e-01
-9.83183086e-01 -7.29295492e-01 3.26562911e-01 9.56037268e-02
-7.20174491e-01 9.68017220e-01 6.83153272e-02 -6.08193159e-01
-5.51983714e-01 -8.54492962e-01 -1.03634179e+00 -8.68896961e-01
-9.15125534e-02 3.82940233e-01 -9.52486917e-02 -2.18225881e-01
3.47223014e-01 2.63106018e-01 -1.43329942e+00 7.50860125e-02
8.80959988e-01 1.17652440e+00 -9.99092162e-02 2.07147613e-01
-1.01782732e-01 1.02124405e+00 -6.79129720e-01 -1.70345709e-01
7.42217243e-01 -3.49423021e-01 -5.22701085e-01 -1.11699648e-01
3.07521194e-01 -5.77721357e-01 -2.82308191e-01 1.13582420e+00
5.41106343e-01 3.71821225e-01 4.53180611e-01 -1.24872315e+00
-5.61413229e-01 5.53095520e-01 -5.99897802e-01 6.55474886e-02
1.75948575e-01 4.65325177e-01 1.11070669e+00 -3.69070500e-01
3.02979410e-01 1.38357437e+00 4.94543523e-01 5.47795475e-01
-1.40167296e+00 -6.88549101e-01 4.83068317e-01 -6.66472092e-02
-1.01998985e+00 -3.27840388e-01 6.52506351e-01 -2.56807119e-01
8.55535865e-01 -8.20052903e-03 5.01591861e-01 9.75631595e-01
8.81242454e-02 9.64488387e-01 1.61627066e+00 -2.66121328e-01
7.79602766e-01 1.46080866e-01 -1.89251885e-01 4.14677709e-01
2.14878023e-01 6.53583825e-01 -4.67458516e-01 -3.94191146e-01
7.39980757e-01 6.51221946e-02 -9.13114175e-02 -5.29245138e-01
-5.24628758e-01 1.00484383e+00 1.77914530e-01 -1.32664308e-01
-3.25405777e-01 5.78520656e-01 5.69829166e-01 7.98282206e-01
4.20966744e-01 4.86517280e-01 -7.74040699e-01 -5.47046304e-01
-7.43406653e-01 5.90430677e-01 6.12958133e-01 8.53234708e-01
8.56773674e-01 2.74646014e-01 -4.47101921e-01 6.63002253e-01
2.07229018e-01 7.08729804e-01 5.12525380e-01 -1.36957181e+00
5.37216544e-01 3.14459026e-01 5.92875838e-01 -5.05861759e-01
-6.44893795e-02 -3.39300752e-01 -3.26325327e-01 4.55609232e-01
8.74247015e-01 -3.92948270e-01 -8.22010636e-01 2.14438701e+00
2.65449733e-01 7.68740699e-02 9.18925777e-02 7.67449975e-01
-5.17273264e-04 4.04463440e-01 2.09824637e-01 -5.03213763e-01
8.31531942e-01 -4.86851245e-01 -5.58229804e-01 -8.58875513e-02
8.26428771e-01 -2.11732343e-01 1.57699800e+00 6.57846630e-01
-1.04747856e+00 -4.92389917e-01 -6.71709776e-01 3.68596464e-01
-2.05449894e-01 -3.33113641e-01 6.05180681e-01 6.22770667e-01
-1.00461495e+00 8.46150160e-01 -8.02091956e-01 -6.47530751e-03
4.94982034e-01 4.15694237e-01 1.11637756e-01 -3.40831131e-02
-1.23505974e+00 6.30000591e-01 1.78511783e-01 -3.38179976e-01
-1.15548599e+00 -8.34952772e-01 -6.77512705e-01 -3.18722054e-02
1.04028356e+00 -2.20737070e-01 1.49751091e+00 -1.24951148e+00
-1.85034597e+00 2.90552974e-01 -3.52517255e-02 -6.84008181e-01
7.14410245e-01 -2.73345947e-01 7.38733634e-02 3.95068899e-02
-1.25103235e-01 3.34356487e-01 1.09869778e+00 -1.09063601e+00
-5.97543895e-01 -5.49889028e-01 2.56778508e-01 -4.28829081e-02
-5.46003580e-02 -4.50935006e-01 -2.51429938e-02 -5.16408145e-01
-2.85438359e-01 -7.81421304e-01 -2.90370792e-01 -2.01486111e-01
-2.02180929e-02 -5.44026077e-01 7.46751726e-01 -2.15730846e-01
1.41775310e+00 -2.07041097e+00 -2.82980472e-01 4.72858787e-01
-6.23981692e-02 2.99783498e-01 -1.89927980e-01 6.08568966e-01
-1.38845025e-02 1.96680412e-01 -1.32442102e-01 2.58263026e-04
3.39813650e-01 5.84868610e-01 -5.88324130e-01 4.08347160e-01
-5.96190840e-02 8.56009901e-01 -9.92837429e-01 -1.98614985e-01
2.72197872e-02 -9.26100016e-02 -9.71672833e-01 2.59763062e-01
-6.37048244e-01 4.70952123e-01 -7.64028788e-01 4.66148496e-01
4.37611848e-01 1.28992014e-02 3.61196369e-01 1.10151619e-01
-2.32914552e-01 6.96405172e-02 -1.29110122e+00 1.27982640e+00
-3.00560832e-01 -9.50036384e-03 1.15588866e-01 -1.57907724e+00
8.56437504e-01 1.75143450e-01 8.25448275e-01 -1.09299338e+00
-4.60199788e-02 3.55006725e-01 -8.93584713e-02 -6.20169461e-01
1.02683328e-01 -4.62228417e-01 -1.05197750e-01 5.16079664e-01
-1.18460312e-01 -6.39485195e-02 2.76052386e-01 -1.58725725e-03
1.07555532e+00 3.96514326e-01 2.79193580e-01 -4.28384662e-01
2.13398486e-01 -1.97882429e-01 7.47391284e-01 1.22953439e+00
-6.80383384e-01 -1.37254044e-01 1.05203438e+00 -2.73778856e-01
-8.63289416e-01 -1.17427957e+00 -1.61956862e-01 1.25409067e+00
-8.27565417e-02 -2.05842495e-01 -5.99405825e-01 -1.06204152e+00
5.80918789e-01 8.18254232e-01 -6.81338310e-01 -1.87670872e-01
-4.65470552e-01 -6.37247086e-01 2.85181552e-01 5.02042174e-01
3.79000217e-01 -9.52748895e-01 -6.23513281e-01 2.75260150e-01
2.71177024e-01 -6.83254242e-01 -4.13850635e-01 3.51004332e-01
-8.41039360e-01 -1.10909760e+00 -3.88631195e-01 -2.29701012e-01
4.73587096e-01 -1.55443072e-01 1.06861222e+00 -2.55659908e-01
2.06273556e-01 8.40868592e-01 -4.31232631e-01 -4.68995959e-01
-2.15702310e-01 -8.81382450e-02 2.76144117e-01 -4.65236418e-02
4.66633171e-01 -6.17066324e-01 -8.58105838e-01 2.68451452e-01
-1.00985956e+00 -7.30684340e-01 5.13387442e-01 8.79379332e-01
7.46157229e-01 1.21942490e-01 1.13322270e+00 -1.06247973e+00
9.35905278e-01 -5.50024867e-01 -1.00979817e+00 1.06973372e-01
-1.14757764e+00 5.05849421e-01 1.03989875e+00 -5.10571241e-01
-8.96487594e-01 -2.28602633e-01 -2.14960545e-01 -5.66493213e-01
9.19661224e-02 3.62543136e-01 1.35890329e-02 2.30332896e-01
6.08073056e-01 1.70686901e-01 3.22328418e-01 -4.74460721e-01
3.50735068e-01 6.48600757e-01 8.04825202e-02 -1.27510917e+00
6.21125877e-01 5.00986636e-01 -8.06071088e-02 -5.59935689e-01
-1.17696655e+00 -2.73086369e-01 -5.65509014e-02 -1.25309050e-01
5.26436865e-01 -7.71062374e-01 -1.20662582e+00 1.31072611e-01
-3.19230855e-01 -9.39444959e-01 -9.48844850e-01 4.96223718e-01
-1.14594722e+00 2.57034868e-01 -4.11165804e-01 -1.24098730e+00
1.00334939e-02 -1.07581198e+00 8.55277359e-01 1.66457951e-01
2.42649451e-01 -9.08768594e-01 2.00302601e-01 7.96630606e-02
4.00125772e-01 9.70442891e-02 1.06098747e+00 -7.05931008e-01
-3.26099575e-01 1.25363603e-01 7.28954300e-02 6.79202497e-01
4.22595181e-02 -4.65809226e-01 -8.20329785e-01 -8.86007249e-01
1.62921771e-01 -8.56586397e-01 8.39465916e-01 6.29826844e-01
1.62936068e+00 -5.34909427e-01 2.57923543e-01 4.57867265e-01
1.59671640e+00 4.52363640e-01 6.18538737e-01 3.33670229e-01
8.47850740e-02 4.12236392e-01 7.99384117e-01 9.98119771e-01
1.15989089e-01 3.75829071e-01 5.82970321e-01 1.58650517e-01
4.51236129e-01 -5.81982076e-01 7.40485072e-01 2.10256204e-01
2.18673095e-01 -1.52325258e-02 -4.76215601e-01 2.28613079e-01
-2.00818920e+00 -1.12161446e+00 5.10588348e-01 2.58918476e+00
1.21561766e+00 2.34957069e-01 5.33633411e-01 -1.02165081e-01
7.89451301e-02 2.03842431e-01 -1.21003377e+00 -6.47210836e-01
3.10999066e-01 5.26020050e-01 7.22525358e-01 5.65086246e-01
-8.16734612e-01 8.08441162e-01 6.07703781e+00 1.21320021e+00
-1.21138513e+00 2.61293985e-02 7.76573598e-01 -2.47473836e-01
-3.87693673e-01 -4.90354978e-05 -9.06011581e-01 6.43504262e-01
1.04969060e+00 7.29176998e-02 8.34573030e-01 9.37442780e-01
7.22369313e-01 -3.05535942e-01 -1.06145811e+00 8.23566854e-01
-4.64359641e-01 -9.27589655e-01 -1.91125870e-01 1.34373009e-01
6.48952603e-01 4.97846305e-02 2.37865359e-01 9.52471852e-01
6.20167971e-01 -9.82498765e-01 5.92942357e-01 5.11490703e-01
8.47140789e-01 -9.37691271e-01 5.30017316e-01 6.84267223e-01
-8.85982335e-01 -5.63635230e-01 -4.71656680e-01 -1.41815662e-01
-2.22528622e-01 4.99455661e-01 -3.62764299e-01 2.53416270e-01
7.73903489e-01 5.04775405e-01 -2.45127648e-01 5.25704443e-01
-1.36561409e-01 9.39859271e-01 -2.85698295e-01 -1.64598785e-02
4.93346930e-01 -5.69234550e-01 2.65348405e-01 7.15452552e-01
1.13068871e-01 -1.03304595e-01 6.09253526e-01 1.02976954e+00
1.36759281e-01 -4.73847240e-02 -5.20394802e-01 -2.42312297e-01
3.54142427e-01 6.16458774e-01 -8.17670226e-02 -2.20520973e-01
-4.01770204e-01 3.07506144e-01 5.28733850e-01 6.47828102e-01
-6.68695450e-01 -2.99318939e-01 6.97205245e-01 1.82572246e-01
4.06843632e-01 -2.84121722e-01 6.96420074e-02 -9.10595655e-01
6.69147447e-02 -1.20989919e+00 5.02671838e-01 -1.22820951e-01
-1.53697646e+00 -2.45709717e-01 2.50550538e-01 -1.11602437e+00
-4.33145881e-01 -6.48008108e-01 -1.64225399e-01 5.83087981e-01
-1.76257777e+00 -4.06272441e-01 3.68594736e-01 8.11210573e-01
4.38834846e-01 1.56721577e-01 4.64095265e-01 9.32251066e-02
-5.40513098e-01 7.77751327e-01 6.65210426e-01 5.13417237e-02
6.69076800e-01 -1.42956388e+00 -2.94204593e-01 2.57020712e-01
-1.25435427e-01 6.17704630e-01 5.84969044e-01 -5.11225343e-01
-1.63713408e+00 -9.95291173e-01 1.43387437e-01 -3.53660166e-01
7.28894770e-01 -2.30411515e-01 -8.06333899e-01 6.37030244e-01
-1.91726089e-01 1.44726247e-01 6.23405278e-01 2.25675311e-02
-3.36221516e-01 -4.94232744e-01 -1.38050139e+00 5.23117900e-01
9.24852610e-01 -4.25751865e-01 -4.52993006e-01 1.89060122e-01
8.56207848e-01 -1.11400366e-01 -9.85665023e-01 4.20386821e-01
4.37154859e-01 -1.06168020e+00 8.29930961e-01 -1.05851901e+00
2.32912347e-01 -7.24251270e-02 -4.14596140e-01 -1.34026659e+00
1.65808141e-01 -9.46529806e-01 -3.88873219e-01 9.33304429e-01
1.99382409e-01 -1.05385017e+00 8.21112812e-01 6.53918564e-01
1.40316367e-01 -1.22719133e+00 -1.07200575e+00 -1.20191371e+00
6.96137309e-01 -6.46085262e-01 4.92762774e-01 6.14812076e-01
-5.37852719e-02 -1.88929409e-01 -4.21053201e-01 -1.70412764e-01
7.13512003e-01 1.39097884e-01 7.58105934e-01 -7.01492190e-01
-7.50520349e-01 -2.21889123e-01 2.63705999e-01 -1.46068573e+00
3.83374989e-01 -6.39287293e-01 5.24534434e-02 -9.08155024e-01
-7.04026502e-03 -7.39378095e-01 -6.60873473e-01 4.65648055e-01
-7.05118030e-02 -3.65726292e-01 3.40544432e-01 1.40487477e-01
-7.70201504e-01 8.62896323e-01 1.32211864e+00 1.22144066e-01
-4.69078481e-01 2.35388562e-01 -7.60559976e-01 6.65266156e-01
8.63029718e-01 -5.70964038e-01 -7.04661489e-01 -5.70610398e-03
2.70957828e-01 2.66736895e-01 2.22262233e-01 -5.45528054e-01
-1.03673421e-01 -7.20011294e-01 1.27135709e-01 -2.50243425e-01
-3.57779674e-02 -7.43818223e-01 -5.83010674e-01 6.39008164e-01
-6.46905005e-01 -3.96574140e-02 1.21798232e-01 9.23032224e-01
3.63797396e-02 -3.40060681e-01 9.55154836e-01 -3.30359846e-01
-4.40379202e-01 4.81842816e-01 -3.31306010e-01 8.47964108e-01
7.74043202e-01 -1.06420731e-02 -2.97030449e-01 -5.97361684e-01
-5.39933145e-01 5.19157827e-01 1.42787635e-01 -2.94981394e-02
3.51696908e-01 -1.12565470e+00 -3.68529737e-01 2.82511085e-01
-1.21077009e-01 8.02800618e-03 1.24092484e-02 8.34389091e-01
1.57833397e-01 3.29391658e-01 3.68146956e-01 -4.37138349e-01
-4.15720850e-01 6.58495665e-01 5.67071557e-01 -5.31236291e-01
-3.59131962e-01 4.12926078e-01 2.57637352e-02 -5.92753410e-01
5.05677700e-01 -4.11395490e-01 1.74858660e-01 -5.80451153e-02
3.45572978e-01 3.77012730e-01 -2.39514410e-01 8.28211978e-02
6.67171776e-02 4.35630977e-01 -2.85721328e-02 -2.47537196e-01
1.27585912e+00 1.25437245e-01 2.56317079e-01 4.59274441e-01
1.12950373e+00 1.42940059e-01 -1.91668570e+00 -3.73535126e-01
2.76582897e-01 -5.69568396e-01 -2.90148016e-02 -9.60630417e-01
-9.86290872e-01 5.74437201e-01 6.27067566e-01 3.87372106e-01
1.05457604e+00 -1.77218303e-01 6.30295157e-01 5.87459981e-01
5.90081692e-01 -1.70305693e+00 3.15675765e-01 5.21762908e-01
5.71653068e-01 -1.32918727e+00 2.34403517e-02 2.90305436e-01
-6.87897563e-01 7.89190292e-01 6.86985612e-01 -4.45960641e-01
7.67012358e-01 1.53025493e-01 3.14715393e-02 -4.62432951e-02
-8.97397220e-01 -3.66027504e-01 -2.12331131e-01 5.69710612e-01
1.16275251e-01 2.36635543e-02 -3.78011316e-01 5.40595710e-01
1.75191462e-03 1.24296173e-01 2.09490642e-01 1.17203319e+00
-5.23832262e-01 -1.26958549e+00 -2.88910806e-01 6.61232352e-01
-4.31632876e-01 1.77326232e-01 -1.51519567e-01 1.12306631e+00
-5.06157316e-02 1.11167574e+00 -2.87947394e-02 -4.20097597e-02
5.77717543e-01 2.33640268e-01 4.19590801e-01 -3.84818494e-01
-5.99021673e-01 1.41393170e-01 -2.62362391e-01 -1.00601113e+00
-2.56206393e-01 -5.95984101e-01 -1.32303536e+00 -1.52145743e-01
-4.79360064e-03 3.63304168e-01 2.51735896e-01 9.88006294e-01
3.95887524e-01 2.33406276e-01 9.56520557e-01 -3.21074873e-01
-1.89759648e+00 -5.78904629e-01 -9.08940375e-01 6.80537820e-01
4.35461879e-01 -7.20860660e-01 -7.89233625e-01 -5.38483918e-01] | [4.098418235778809, 2.2984869480133057] |
878b2243-63d8-4340-85b1-120a20185149 | personalize-segment-anything-model-with-one | 2305.03048 | null | https://arxiv.org/abs/2305.03048v1 | https://arxiv.org/pdf/2305.03048v1.pdf | Personalize Segment Anything Model with One Shot | Driven by large-data pre-training, Segment Anything Model (SAM) has been demonstrated as a powerful and promptable framework, revolutionizing the segmentation models. Despite the generality, customizing SAM for specific visual concepts without man-powered prompting is under explored, e.g., automatically segmenting your pet dog in different images. In this paper, we propose a training-free Personalization approach for SAM, termed as PerSAM. Given only a single image with a reference mask, PerSAM first localizes the target concept by a location prior, and segments it within other images or videos via three techniques: target-guided attention, target-semantic prompting, and cascaded post-refinement. In this way, we effectively adapt SAM for private use without any training. To further alleviate the mask ambiguity, we present an efficient one-shot fine-tuning variant, PerSAM-F. Freezing the entire SAM, we introduce two learnable weights for multi-scale masks, only training 2 parameters within 10 seconds for improved performance. To demonstrate our efficacy, we construct a new segmentation dataset, PerSeg, for personalized evaluation, and test our methods on video object segmentation with competitive performance. Besides, our approach can also enhance DreamBooth to personalize Stable Diffusion for text-to-image generation, which discards the background disturbance for better target appearance learning. Code is released at https://github.com/ZrrSkywalker/Personalize-SAM | ['Hongsheng Li', 'Peng Gao', 'Hao Dong', 'Junting Pan', 'Shilin Yan', 'Ziyu Guo', 'Zhengkai Jiang', 'Renrui Zhang'] | 2023-05-04 | null | null | null | null | ['personalized-segmentation', 'video-object-segmentation', 'video-semantic-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.40343386e-01 -5.32843880e-02 -3.65131408e-01 -4.21336204e-01
-7.71856785e-01 -7.45221317e-01 3.27385128e-01 -3.36077124e-01
-6.03103042e-01 3.71454090e-01 -1.37982666e-01 -1.61782175e-01
1.17279142e-01 -4.37623024e-01 -7.99325705e-01 -6.19174600e-01
3.34715486e-01 4.40249383e-01 6.88159108e-01 -6.86848611e-02
1.20632023e-01 3.16090733e-01 -1.31706858e+00 3.12015206e-01
1.15319109e+00 8.27615857e-01 8.13045144e-01 4.75924820e-01
-9.22699869e-02 2.56675392e-01 -6.97541475e-01 -4.58442956e-01
3.60427767e-01 -2.98435271e-01 -7.88621545e-01 5.50772369e-01
4.45585579e-01 -3.97174865e-01 -2.57153779e-01 9.04071808e-01
3.96069109e-01 2.70695746e-01 3.53064448e-01 -1.18123257e+00
-7.77038157e-01 7.21097231e-01 -8.89636576e-01 2.39551336e-01
1.09845176e-02 5.02116680e-01 7.06235886e-01 -9.04694617e-01
4.99863893e-01 1.06771636e+00 3.38436544e-01 8.54133070e-01
-1.39991748e+00 -9.16409492e-01 6.54473007e-01 2.28377745e-01
-1.42156410e+00 -4.45298105e-01 6.80284262e-01 -4.03936118e-01
4.36234117e-01 3.70257825e-01 7.10204661e-01 1.37604284e+00
-2.14955121e-01 1.17127967e+00 8.86596978e-01 -1.14652991e-01
3.84767056e-02 2.01611668e-01 1.11965574e-01 5.22237182e-01
3.88026051e-02 -5.56180254e-02 -2.91322589e-01 1.73984736e-01
8.66482675e-01 8.04660991e-02 -4.06455219e-01 -2.92590410e-01
-1.24383724e+00 6.38441563e-01 3.87919486e-01 1.03523508e-01
-2.45320112e-01 1.17826924e-01 1.68587297e-01 -1.03944413e-01
4.27368462e-01 4.33317810e-01 -5.88265538e-01 -7.06074014e-03
-1.16551030e+00 1.32278651e-01 3.51390004e-01 1.16464198e+00
9.52367008e-01 -6.63516298e-02 -6.39390230e-01 9.93589759e-01
9.99816433e-02 4.89127785e-01 6.77192330e-01 -1.02306080e+00
3.65700960e-01 3.82272363e-01 1.68202862e-01 -6.39831841e-01
-3.29861224e-01 -4.15342391e-01 -6.36011302e-01 -1.38373211e-01
1.83186471e-01 -2.87236780e-01 -1.52838981e+00 1.73857117e+00
5.35607517e-01 5.06283581e-01 -2.65742630e-01 1.13099575e+00
6.39900804e-01 7.04431891e-01 2.07939908e-01 2.85308622e-02
1.35624969e+00 -1.54382503e+00 -3.46567839e-01 -6.21945500e-01
1.33691698e-01 -6.58132970e-01 1.47646093e+00 3.58090132e-01
-8.53830934e-01 -5.89269698e-01 -8.48661721e-01 -8.46783165e-03
-1.65311649e-01 2.90387899e-01 6.88917875e-01 5.20431995e-01
-8.70936215e-01 5.09375870e-01 -1.00217116e+00 -3.99354309e-01
7.98544765e-01 2.81432956e-01 -1.46257475e-01 -8.24121088e-02
-8.86051238e-01 4.52531129e-01 5.60722053e-01 -3.61964330e-02
-1.13259280e+00 -7.95838892e-01 -6.54317379e-01 -3.88721898e-02
7.88997412e-01 -7.49700546e-01 1.21074116e+00 -1.33558774e+00
-1.76062047e+00 9.27343845e-01 -1.55113026e-01 -4.01249468e-01
5.93007267e-01 -4.38233286e-01 -2.08224118e-01 1.47219092e-01
2.31327176e-01 1.20250022e+00 1.37320626e+00 -1.46275544e+00
-6.16635621e-01 -4.29055803e-02 1.83975995e-01 1.70448542e-01
-3.87994081e-01 1.56850651e-01 -1.30152941e+00 -9.76255417e-01
-8.27861056e-02 -9.58699286e-01 -5.10604978e-01 -1.05443045e-01
-5.72985291e-01 -1.64495893e-02 8.12489033e-01 -5.83991408e-01
1.20719039e+00 -2.41807652e+00 9.03016254e-02 1.60220832e-01
2.21609786e-01 4.90517765e-01 -6.91604078e-01 5.48077784e-02
-1.83501728e-02 1.94601551e-01 -5.24284065e-01 -5.60907006e-01
-1.57676950e-01 3.45926136e-01 -2.07644030e-01 2.15753555e-01
3.13475192e-01 1.15080893e+00 -7.70140171e-01 -6.02379084e-01
2.13945806e-01 2.64881134e-01 -6.72132075e-01 2.98807234e-01
-5.40655851e-01 6.61825478e-01 -5.03743351e-01 6.56364560e-01
7.72207499e-01 -5.03666282e-01 -4.01189737e-02 -3.75946552e-01
7.01017827e-02 -3.02948266e-01 -1.13029349e+00 1.92974794e+00
-3.73312354e-01 3.39603484e-01 1.36092961e-01 -9.53391850e-01
6.56079948e-01 -7.36743882e-02 3.87493312e-01 -6.72498643e-01
8.58341679e-02 -1.15387999e-01 -2.26719752e-01 -5.27935266e-01
5.06406128e-01 2.90783346e-01 4.84627970e-02 3.05723310e-01
1.17629498e-01 2.98007447e-02 4.09411162e-01 3.64726633e-01
7.96281874e-01 4.10966277e-01 -5.78823918e-03 -3.69152799e-02
2.68088907e-01 2.92744744e-03 7.75047421e-01 7.58186340e-01
-1.94721997e-01 9.88275170e-01 1.48532718e-01 -3.97869684e-02
-7.45744824e-01 -1.12575734e+00 -3.65177989e-02 1.43470728e+00
6.74296439e-01 -5.38030803e-01 -9.90272284e-01 -8.94392967e-01
-1.62010550e-01 7.70108700e-01 -6.29571140e-01 -2.83215456e-02
-6.14959002e-01 -7.77729809e-01 3.02868724e-01 5.02420902e-01
6.10414147e-01 -1.19569123e+00 -5.83330572e-01 1.80825278e-01
-3.37626576e-01 -1.08136356e+00 -1.11821795e+00 -1.32616600e-02
-6.18896723e-01 -9.57807243e-01 -9.97792780e-01 -6.58838451e-01
7.84062862e-01 5.60698450e-01 9.21215177e-01 1.43788263e-01
-2.64605701e-01 4.65959013e-01 -4.39445257e-01 -8.41942504e-02
-7.86726102e-02 2.61509091e-01 -2.17740864e-01 2.22191185e-01
5.03817108e-03 -5.08440495e-01 -9.61160958e-01 6.37335300e-01
-1.00415909e+00 4.52176511e-01 8.11690509e-01 7.36570120e-01
8.84395778e-01 -2.41867647e-01 4.25290138e-01 -9.72648144e-01
4.68449086e-01 -5.13496459e-01 -5.60093343e-01 2.78004348e-01
-3.73751283e-01 -1.36229783e-01 4.21457231e-01 -8.04885089e-01
-1.02327979e+00 2.08528936e-01 -8.98861066e-02 -7.84880757e-01
-2.55521506e-01 2.62489408e-01 -3.73032868e-01 -2.26914808e-02
6.21212184e-01 3.27217728e-01 -1.73563734e-01 -5.24363041e-01
7.19796717e-01 4.54500467e-01 6.77443981e-01 -7.33243406e-01
9.18240666e-01 3.44412833e-01 -7.21501052e-01 -5.46110511e-01
-9.58502054e-01 -5.50484061e-01 -4.47729468e-01 -9.90540627e-03
1.04516602e+00 -9.35585797e-01 -3.86924833e-01 5.04620135e-01
-1.07121801e+00 -8.85858357e-01 -3.04410756e-01 1.90435827e-01
-4.48137671e-01 3.39245111e-01 -4.79024649e-01 -3.52043420e-01
-3.46836478e-01 -1.26123965e+00 1.04160619e+00 5.09532452e-01
-1.03688538e-01 -6.09279156e-01 -1.34788051e-01 5.58384061e-01
3.67473781e-01 -1.09622940e-01 2.69899011e-01 -6.88397288e-01
-9.11178410e-01 1.25687540e-01 -4.69165981e-01 4.61227566e-01
1.42848834e-01 -1.14057669e-02 -9.10297871e-01 -4.23120707e-01
-2.66706258e-01 -2.30966762e-01 1.00384140e+00 4.02357906e-01
1.71216393e+00 -4.33977902e-01 -6.24751091e-01 9.61552024e-01
1.09424949e+00 2.37800419e-01 5.21680236e-01 3.92927498e-01
9.19369340e-01 3.14794302e-01 7.81835675e-01 3.88816863e-01
4.23728377e-01 7.36590505e-01 2.34905332e-01 -3.80833298e-01
-3.09440017e-01 -2.55968302e-01 3.14565539e-01 3.53024036e-01
4.26568231e-03 -4.61203426e-01 -5.97608626e-01 6.08177364e-01
-1.89151776e+00 -7.80945182e-01 3.07607204e-01 1.95155668e+00
9.15311277e-01 4.11464050e-02 1.98063895e-01 -5.59944630e-01
8.96211624e-01 2.57774264e-01 -8.89305055e-01 2.15688691e-01
-1.80763435e-02 2.45720044e-01 6.16263449e-01 4.06075805e-01
-1.22690928e+00 1.48879087e+00 5.22364902e+00 1.40539610e+00
-1.19188380e+00 4.25777137e-01 8.10776711e-01 -1.40619218e-01
-4.17575449e-01 3.88618968e-02 -7.70099223e-01 7.64076233e-01
5.14926136e-01 -5.43045662e-02 6.14196062e-01 9.07394588e-01
3.55284244e-01 -5.68659566e-02 -8.28365922e-01 1.02805960e+00
1.20833360e-01 -1.36413836e+00 9.72717926e-02 -2.47125566e-01
7.09621012e-01 1.15632728e-01 1.98544472e-01 3.60241920e-01
3.13138336e-01 -7.86536515e-01 9.19480622e-01 3.30878377e-01
9.16915119e-01 -5.38241863e-01 2.49766856e-01 2.55168885e-01
-1.09369910e+00 -3.73720266e-02 -2.32107833e-01 5.22379220e-01
3.30255836e-01 3.44480574e-01 -6.20187938e-01 4.91858512e-01
7.73808062e-01 6.97708070e-01 -8.34177077e-01 1.20439029e+00
-3.35931271e-01 7.75082886e-01 -3.95505011e-01 3.32122356e-01
1.65298089e-01 -2.12715760e-01 5.48515916e-01 1.25970948e+00
2.20642492e-01 1.86697781e-01 5.06587923e-01 9.84360397e-01
1.04330946e-02 4.52756844e-02 1.87080745e-02 5.15247770e-02
5.92905879e-01 1.50092530e+00 -9.57358122e-01 -4.19888914e-01
-2.53286570e-01 1.49769509e+00 1.78769037e-01 6.92230403e-01
-1.26901197e+00 -1.87267184e-01 5.88417709e-01 2.25700051e-01
6.42974615e-01 -1.17558986e-01 -1.29070625e-01 -1.27176905e+00
-1.00463040e-01 -9.56310987e-01 3.96913022e-01 -8.12299550e-01
-1.20106030e+00 7.29841888e-01 1.97025821e-01 -1.07099187e+00
1.88042700e-01 -2.64415145e-01 -7.52407432e-01 6.42955780e-01
-1.28013062e+00 -1.35854661e+00 -4.41210210e-01 7.72658467e-01
8.54150712e-01 8.23162869e-02 2.76638955e-01 3.85185361e-01
-9.94415581e-01 8.49529386e-01 -2.22341537e-01 -7.08990311e-03
9.73914266e-01 -1.04808342e+00 6.47078633e-01 1.00200629e+00
2.85249680e-01 4.74112362e-01 5.77497959e-01 -6.33449316e-01
-8.45924258e-01 -1.43783498e+00 9.74122584e-02 -3.45723033e-01
5.62874019e-01 -4.67708051e-01 -1.00550675e+00 7.97493756e-01
2.05098838e-01 3.52651477e-02 4.08787608e-01 -9.68242660e-02
-1.66399598e-01 -1.65021628e-01 -8.93072486e-01 8.91688526e-01
1.29296958e+00 -8.18709806e-02 -3.10487002e-01 3.56582075e-01
1.17075741e+00 -6.32188439e-01 -4.45334792e-01 3.54679495e-01
1.58348396e-01 -8.88760507e-01 9.35622334e-01 -3.39369625e-01
1.23238988e-01 -5.30727029e-01 8.17753375e-02 -1.23177421e+00
-4.63594139e-01 -9.36242104e-01 3.25965136e-02 1.44082880e+00
4.76588368e-01 -6.16410613e-01 9.38481629e-01 5.81875086e-01
-4.27849352e-01 -8.51281226e-01 -6.92633569e-01 -6.37024343e-01
-1.38633355e-01 -5.31577051e-01 6.69855654e-01 7.01921225e-01
-5.52601874e-01 2.88817942e-01 -4.85788226e-01 3.54524076e-01
4.69396442e-01 2.24301308e-01 1.06081855e+00 -7.15316772e-01
-6.35162532e-01 -4.45900440e-01 -3.96306738e-02 -1.61479247e+00
8.07331651e-02 -7.29876876e-01 2.11472273e-01 -1.40468192e+00
3.55849326e-01 -6.13883376e-01 -2.89828420e-01 7.88201392e-01
-4.91877347e-01 4.34301168e-01 4.68792796e-01 3.16892445e-01
-8.88536751e-01 6.17333293e-01 1.46747601e+00 -2.54134178e-01
-3.21278065e-01 8.12865570e-02 -8.42875659e-01 6.24390960e-01
8.90086949e-01 -4.42196369e-01 -6.28584862e-01 -3.97110730e-01
-3.90103906e-01 -1.61396727e-01 4.69880164e-01 -9.52562094e-01
-2.97627524e-02 -4.44828480e-01 3.56765419e-01 -5.19619584e-01
2.48766527e-01 -5.94481468e-01 1.78724855e-01 1.47619054e-01
-1.06502049e-01 -1.87790960e-01 3.28014374e-01 6.25638723e-01
7.60198832e-02 -2.26905137e-01 8.34351361e-01 -1.42347571e-02
-1.12328887e+00 7.95711756e-01 -8.67751837e-02 1.85328320e-01
1.21743417e+00 -3.66511822e-01 -3.23235452e-01 -9.01676118e-02
-8.20097506e-01 5.48331857e-01 6.86605811e-01 5.91944575e-01
5.70935845e-01 -1.05490017e+00 -5.65116286e-01 2.08580032e-01
2.10790977e-01 2.25305706e-01 7.65027821e-01 8.61872375e-01
-2.71141469e-01 -1.16966022e-02 3.41178775e-02 -7.05074370e-01
-1.16085076e+00 8.03552330e-01 2.60667652e-01 -5.70523851e-02
-8.25412273e-01 1.08144915e+00 8.28329921e-01 -1.30428359e-01
1.06635012e-01 -2.26920426e-01 -7.24385381e-02 -1.10681564e-01
4.79209423e-01 2.26410944e-02 -2.62496173e-01 -3.78444195e-01
-2.43002728e-01 5.35199106e-01 -4.93857145e-01 -1.25660881e-01
9.90515530e-01 -4.21791196e-01 1.88494027e-01 2.08887562e-01
8.31158280e-01 7.25770220e-02 -1.83855534e+00 -8.88996348e-02
-2.91264057e-01 -5.86702585e-01 -2.10675478e-01 -9.48094547e-01
-1.38032830e+00 5.23318350e-01 5.62119305e-01 -1.17365941e-01
1.25528204e+00 1.97800398e-01 1.01645398e+00 4.11649793e-02
1.87147349e-01 -9.74777043e-01 3.19102883e-01 2.39562660e-01
8.91655505e-01 -1.28726566e+00 -1.80664092e-01 -4.08392638e-01
-1.06169057e+00 6.83585763e-01 1.04035532e+00 5.25297709e-02
3.99364591e-01 8.73993058e-03 2.87390918e-01 8.32606182e-02
-5.18372297e-01 -2.60701180e-01 3.40247214e-01 7.21527278e-01
-9.93990153e-03 8.18129629e-02 -1.55249357e-01 8.93417001e-01
-2.55909134e-02 -6.89540431e-02 3.02542329e-01 5.85488260e-01
-3.31800491e-01 -1.15442383e+00 -3.16960752e-01 4.08914298e-01
-2.47866943e-01 -2.48237476e-01 -6.49470240e-02 6.35197461e-01
3.40266883e-01 7.91884422e-01 7.10735023e-02 -3.97139221e-01
2.84495890e-01 -3.34146559e-01 3.69535416e-01 -7.73011565e-01
-2.97122687e-01 3.06041002e-01 -1.03958838e-01 -7.90641010e-01
-4.12965029e-01 -5.21532834e-01 -1.17579627e+00 -5.59674278e-02
-3.90919238e-01 6.81818351e-02 2.32173473e-01 8.33890498e-01
6.28226399e-01 6.09049261e-01 4.61586922e-01 -1.21267390e+00
-2.68470377e-01 -7.92528927e-01 -2.88211614e-01 3.09785366e-01
1.14400238e-01 -7.35492468e-01 -1.09753735e-01 2.91845828e-01] | [9.537130355834961, 0.17527590692043304] |
930dbe2b-de5f-4ed4-9fcc-d3fff76de0d5 | single-image-reflection-removal-exploiting | 1904.00637 | null | http://arxiv.org/abs/1904.00637v1 | http://arxiv.org/pdf/1904.00637v1.pdf | Single Image Reflection Removal Exploiting Misaligned Training Data and Network Enhancements | Removing undesirable reflections from a single image captured through a glass
window is of practical importance to visual computing systems. Although
state-of-the-art methods can obtain decent results in certain situations,
performance declines significantly when tackling more general real-world cases.
These failures stem from the intrinsic difficulty of single image reflection
removal -- the fundamental ill-posedness of the problem, and the insufficiency
of densely-labeled training data needed for resolving this ambiguity within
learning-based neural network pipelines. In this paper, we address these issues
by exploiting targeted network enhancements and the novel use of misaligned
data. For the former, we augment a baseline network architecture by embedding
context encoding modules that are capable of leveraging high-level contextual
clues to reduce indeterminacy within areas containing strong reflections. For
the latter, we introduce an alignment-invariant loss function that facilitates
exploiting misaligned real-world training data that is much easier to collect.
Experimental results collectively show that our method outperforms the
state-of-the-art with aligned data, and that significant improvements are
possible when using additional misaligned data. | ['Jiaolong Yang', 'David Wipf', 'Kaixuan Wei', 'Hua Huang', 'Ying Fu'] | 2019-04-01 | single-image-reflection-removal-exploiting-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Wei_Single_Image_Reflection_Removal_Exploiting_Misaligned_Training_Data_and_Network_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Wei_Single_Image_Reflection_Removal_Exploiting_Misaligned_Training_Data_and_Network_CVPR_2019_paper.pdf | cvpr-2019-6 | ['reflection-removal'] | ['computer-vision'] | [ 7.33111560e-01 8.34999681e-02 2.29710788e-01 -4.73207623e-01
-9.97935534e-01 -4.16034758e-01 4.99591917e-01 -3.03126127e-01
-3.94868135e-01 4.04186159e-01 3.26814383e-01 -2.69125700e-01
1.82075072e-02 -4.04842764e-01 -8.85809481e-01 -6.81686819e-01
5.52130342e-02 -5.71518466e-02 1.70635089e-01 -1.90963700e-01
4.33612585e-01 6.27533734e-01 -1.65865898e+00 5.68098724e-01
5.83369136e-01 8.12864125e-01 9.77620408e-02 4.98009592e-01
2.33737320e-01 4.85135108e-01 -4.74207580e-01 1.05956517e-01
5.61589897e-01 -3.53850365e-01 -3.67480993e-01 7.09630847e-02
1.18666065e+00 -4.56975162e-01 -2.83780634e-01 7.75336325e-01
3.20809156e-01 8.12513754e-02 4.66346204e-01 -8.77669454e-01
-4.59947258e-01 -7.25097060e-02 -6.92635059e-01 1.92930743e-01
-9.09188241e-02 4.28195745e-01 1.19901836e+00 -8.16291988e-01
5.36894977e-01 1.06766224e+00 8.75356615e-01 5.53546727e-01
-1.26202536e+00 -4.97910589e-01 4.06115144e-01 4.65173125e-02
-1.08833027e+00 -9.07888412e-01 8.07335675e-01 -4.51497614e-01
1.12951314e+00 2.58326948e-01 4.19298738e-01 1.03037131e+00
-6.70048455e-03 5.68527102e-01 8.39446902e-01 -5.24039865e-01
-3.03341653e-02 -8.45560730e-02 -4.77817021e-02 6.61623895e-01
3.51980239e-01 3.42774957e-01 -6.37547672e-01 2.72700638e-01
6.27651811e-01 1.13391668e-01 -4.56232905e-01 -4.09156799e-01
-9.42567885e-01 5.50078154e-01 6.66597366e-01 1.72373042e-01
-1.01456702e-01 9.59415883e-02 3.56965028e-02 1.17753603e-01
6.41803026e-01 8.55363607e-01 -3.38438690e-01 2.10755065e-01
-1.27009845e+00 2.76811957e-01 4.58046645e-01 5.74877262e-01
7.24538386e-01 3.46939445e-01 -5.63040785e-02 7.41971910e-01
2.25448042e-01 3.34205687e-01 -1.29223198e-01 -8.96997213e-01
4.14418876e-01 5.72243214e-01 8.83630961e-02 -8.47611010e-01
-4.79849458e-01 -6.47273183e-01 -4.82495993e-01 6.73752248e-01
6.12612367e-01 -6.35891780e-03 -1.31430328e+00 1.71799219e+00
1.45394921e-01 2.36523733e-01 -4.63519320e-02 1.08975720e+00
6.50446355e-01 4.83018517e-01 -4.20913875e-01 1.41748905e-01
1.24507594e+00 -1.05737209e+00 -2.81606495e-01 -7.86022067e-01
2.36166880e-01 -9.21489418e-01 1.26959765e+00 6.57050908e-01
-8.17686617e-01 -2.63562769e-01 -1.30865085e+00 -2.74301469e-01
-1.13923557e-01 1.10051662e-01 6.25045240e-01 5.89736521e-01
-1.01400161e+00 4.31340337e-01 -7.57223308e-01 -2.11461276e-01
7.66829371e-01 2.54257828e-01 -4.05622452e-01 -5.54441512e-01
-5.95972300e-01 8.60444903e-01 -2.47822255e-01 5.18249869e-01
-8.87649953e-01 -9.42047238e-01 -9.14199829e-01 -7.83713534e-03
6.27750278e-01 -4.99841511e-01 1.08462393e+00 -9.14732933e-01
-1.11216116e+00 8.31524253e-01 -1.97250247e-01 -2.14200109e-01
4.68946993e-01 -5.00007570e-01 -1.09239526e-01 2.17074007e-01
-1.68176010e-01 3.84094298e-01 1.16280890e+00 -1.65306544e+00
-4.40374225e-01 -3.95662129e-01 -4.59290631e-02 2.45461404e-01
-3.66589665e-01 -6.71831369e-02 -4.62889344e-01 -4.70662802e-01
1.36458620e-01 -1.02820075e+00 -2.62845814e-01 2.49523953e-01
-4.25113380e-01 3.25865537e-01 1.02683318e+00 -6.04008675e-01
8.17117095e-01 -2.27081680e+00 -1.12070017e-01 1.01200238e-01
4.29281443e-01 6.46489024e-01 -3.52734417e-01 3.11317027e-01
-5.56903630e-02 4.90516890e-03 -3.38425934e-01 -5.46603501e-01
-3.54076028e-01 7.34788477e-02 -4.64380294e-01 4.87113744e-01
6.97808206e-01 6.96792185e-01 -6.29711688e-01 -4.25638221e-02
2.35592574e-01 7.74179161e-01 -7.06098378e-01 2.57009119e-01
-3.26511204e-01 5.53838074e-01 -1.99799582e-01 5.81593394e-01
6.16339326e-01 -2.17792138e-01 1.10856228e-01 -4.55699533e-01
-4.13046509e-01 4.22156990e-01 -9.71468270e-01 1.74504197e+00
-5.63780427e-01 9.91992354e-01 2.74961799e-01 -7.77143061e-01
7.50881135e-01 -9.33454558e-02 2.75085241e-01 -7.56580234e-01
-1.57671005e-01 4.35720012e-02 1.05396822e-01 -7.31754601e-01
5.96965075e-01 -1.59747496e-01 2.78142601e-01 4.75295544e-01
-3.94696683e-01 -1.77384481e-01 -2.45938137e-01 -3.41658928e-02
1.31679034e+00 3.73188287e-01 2.65334081e-02 5.77125438e-02
2.43421253e-02 -7.48649687e-02 6.29953086e-01 8.41659248e-01
-1.27172664e-01 1.25663137e+00 2.41246402e-01 -5.94233632e-01
-1.30886734e+00 -1.02255583e+00 1.35747362e-02 9.08678770e-01
4.92795669e-02 -3.37236017e-01 -5.61779201e-01 -3.92444581e-01
-4.59662080e-02 4.69840378e-01 -5.74438632e-01 -1.34058997e-01
-9.08078551e-01 -9.31097865e-01 4.61444616e-01 5.12637377e-01
4.49102521e-01 -8.07942152e-01 -9.87856627e-01 -5.55681344e-03
-4.53566983e-02 -1.38510299e+00 -1.10723101e-01 1.37051046e-01
-7.60226488e-01 -1.06958485e+00 -3.85593176e-01 -5.13927817e-01
7.82721937e-01 7.99506068e-01 1.24186468e+00 4.31720287e-01
-8.36078405e-01 2.60808975e-01 -6.93488345e-02 -3.50766927e-01
2.90109385e-02 -2.04657912e-02 -3.64041656e-01 -7.06833825e-02
1.46674260e-01 -6.56590581e-01 -8.77737999e-01 2.10520104e-01
-1.00734067e+00 1.55527890e-01 7.91476667e-01 1.02677619e+00
2.47133747e-01 -1.66007549e-01 3.18281919e-01 -9.19402778e-01
3.31119239e-01 -4.41337749e-02 -5.74155569e-01 6.78176507e-02
-5.81236422e-01 2.79062539e-01 5.68374097e-01 -1.42546684e-01
-1.20141304e+00 1.77876931e-02 -1.78194214e-02 -3.09080154e-01
-2.34642372e-01 1.63757309e-01 -1.09701701e-01 -2.40322113e-01
7.77989268e-01 -8.22730213e-02 -3.32445353e-02 -4.67841476e-01
2.70900577e-01 3.55138808e-01 6.72934651e-01 -5.42592168e-01
7.56607890e-01 9.07492161e-01 1.81449100e-01 -1.09141695e+00
-1.19601440e+00 -4.16630089e-01 -3.85687709e-01 -1.51727661e-01
6.98388815e-01 -8.87182951e-01 -4.64338154e-01 4.97025937e-01
-1.14546061e+00 -3.91276598e-01 -7.23370835e-02 1.23562567e-01
-2.09828675e-01 2.49126643e-01 -4.68895018e-01 -8.68655205e-01
-3.08097839e-01 -1.20844603e+00 1.20717919e+00 1.29133329e-01
2.64388341e-02 -6.24347746e-01 1.30209019e-02 4.84391123e-01
5.41414499e-01 3.02767605e-01 8.64791870e-01 -3.52420628e-01
-1.02027273e+00 -1.14251733e-01 -6.17006481e-01 5.04743457e-01
1.60371177e-02 4.92215976e-02 -1.55017567e+00 -3.12545389e-01
5.55604100e-02 -3.88646215e-01 1.32456195e+00 2.75429279e-01
1.06601775e+00 -8.44061747e-02 -1.60882518e-01 7.78751433e-01
1.40284967e+00 -1.98370442e-01 9.46068883e-01 4.12123978e-01
9.89770055e-01 8.01181138e-01 3.38213265e-01 1.25435427e-01
3.14550549e-01 7.37914085e-01 7.13965535e-01 -5.87149680e-01
-4.26211178e-01 -6.10478111e-02 1.43346459e-01 1.79672778e-01
-4.45830375e-02 -2.19752505e-01 -8.78715456e-01 6.58501565e-01
-1.87815654e+00 -9.25726533e-01 -7.26230890e-02 2.36508656e+00
6.78804636e-01 2.80762047e-01 -1.96870849e-01 1.54337019e-01
3.64298493e-01 4.65675771e-01 -6.43938541e-01 -1.00872263e-01
-1.73185900e-01 1.90993026e-01 4.34551984e-01 7.06217766e-01
-1.01075888e+00 9.11578059e-01 6.47321653e+00 4.20672864e-01
-1.48266685e+00 -2.10763693e-01 6.37575507e-01 -5.40736735e-01
-4.33276623e-01 4.07772847e-02 -7.36068726e-01 7.89835528e-02
6.41007006e-01 7.05920100e-01 5.34021854e-01 4.51066196e-01
3.69304359e-01 -2.45159000e-01 -1.24048841e+00 9.19719100e-01
4.27038252e-01 -1.35353541e+00 -3.64983082e-01 4.44757864e-02
7.28390276e-01 3.55387032e-01 4.13130552e-01 -2.69075245e-01
3.46670598e-02 -1.32420325e+00 5.64785719e-01 4.91455436e-01
8.15440953e-01 -5.02195120e-01 3.91008943e-01 -4.82502244e-02
-7.45858192e-01 5.68328984e-02 -1.88916281e-01 -1.95772871e-01
-3.61762978e-02 8.94482970e-01 -1.04480171e+00 3.40843946e-01
6.71828389e-01 7.99908042e-01 -5.76279402e-01 1.02188396e+00
-4.24666047e-01 5.81951261e-01 -5.67173362e-01 3.55467945e-01
2.39731133e-01 -6.64639398e-02 5.63465238e-01 1.15238655e+00
1.38735086e-01 -2.54115880e-01 -1.22833781e-01 9.64828372e-01
-4.21222895e-02 -5.22030413e-01 -8.76627624e-01 2.09199950e-01
3.60615969e-01 1.40145123e+00 -5.43095291e-01 8.52619037e-02
-5.88897824e-01 8.84483635e-01 5.10268271e-01 5.99383235e-01
-3.78668666e-01 -2.62889057e-01 8.19808125e-01 3.53060484e-01
4.46742713e-01 -4.39206451e-01 -6.42844379e-01 -1.17576206e+00
4.31650668e-01 -1.01673663e+00 8.41037855e-02 -8.16236258e-01
-1.16609371e+00 5.45585871e-01 -2.52598226e-01 -1.07904327e+00
-1.35588855e-01 -9.08591449e-01 -6.25181019e-01 7.04191864e-01
-2.05914259e+00 -1.42825592e+00 -5.04448593e-01 2.39589140e-01
3.65480036e-01 1.10441998e-01 6.46336317e-01 2.84782588e-01
-5.28397083e-01 5.60182095e-01 1.17332980e-01 -5.42793311e-02
9.80999529e-01 -9.35221612e-01 6.46767676e-01 1.35517204e+00
3.08985740e-01 8.32018733e-01 9.06526923e-01 -2.51822770e-01
-1.51811838e+00 -8.77186775e-01 6.68321609e-01 -5.82471550e-01
3.24341506e-01 -6.94369137e-01 -1.01241171e+00 7.35451341e-01
4.78909202e-02 2.33484626e-01 5.66722393e-01 3.59891534e-01
-1.03759444e+00 -2.22174287e-01 -1.01548564e+00 7.81850040e-01
1.15027094e+00 -6.40195429e-01 -3.66086930e-01 1.18086085e-01
5.31694472e-01 -3.80487621e-01 -3.38415325e-01 3.31770599e-01
7.92979419e-01 -1.34719741e+00 1.15179360e+00 -5.29036522e-01
6.80672288e-01 -4.50765759e-01 -1.48140877e-01 -1.26655650e+00
7.10955933e-02 -6.58021927e-01 3.20193082e-01 8.68094504e-01
4.24672723e-01 -5.33060372e-01 8.56554091e-01 6.83615327e-01
-5.04711390e-01 -7.39673316e-01 -7.95480251e-01 -4.99633521e-01
5.11960462e-02 -5.27556062e-01 3.00446808e-01 7.84318686e-01
-2.74216712e-01 3.74992311e-01 -6.19900823e-01 3.10427070e-01
7.32647896e-01 2.90225685e-01 8.91626716e-01 -1.07481718e+00
-3.23947519e-01 -2.74345487e-01 -2.77733773e-01 -1.09081256e+00
-6.22409321e-02 -3.65544111e-01 5.36180496e-01 -1.40319657e+00
3.99149172e-02 -4.82258052e-01 -6.19082488e-02 7.13380754e-01
-1.81635022e-01 6.18114293e-01 3.24129254e-01 4.47460264e-02
-5.64898252e-01 4.71068025e-01 1.07392991e+00 1.35259047e-01
-7.98816830e-02 -3.75466347e-01 -9.73170638e-01 8.03899407e-01
6.13677979e-01 -3.63785654e-01 -3.90269756e-01 -9.97967303e-01
4.94629532e-01 -5.25903881e-01 7.45132983e-01 -1.03276157e+00
1.41658008e-01 -7.89784864e-02 3.41254979e-01 -3.49478364e-01
5.13327301e-01 -7.08140731e-01 -2.66109109e-01 5.53924497e-03
-3.15847516e-01 -2.63665885e-01 4.14659888e-01 5.19857049e-01
1.08556626e-02 -5.29484078e-02 8.75280976e-01 -2.84669772e-02
-6.22077048e-01 -3.62114944e-02 -7.06982389e-02 2.96816289e-01
4.98544544e-01 -1.76669285e-01 -7.49445617e-01 -3.01497668e-01
-1.48973107e-01 -1.04573287e-01 8.30450535e-01 3.67479950e-01
8.32627177e-01 -8.87467384e-01 -6.56746805e-01 4.92661625e-01
2.96600610e-01 3.07624876e-01 2.11601391e-01 6.40356362e-01
-4.19029087e-01 6.34596944e-02 -1.42096087e-01 -5.92308760e-01
-1.34383917e+00 8.15650299e-02 4.08563226e-01 -1.78907856e-01
-8.41061294e-01 8.55328619e-01 4.15046692e-01 -2.93499440e-01
1.15354091e-01 -3.83613199e-01 1.81251839e-01 -2.83556581e-01
6.31557047e-01 7.04398304e-02 4.71912026e-01 -2.86492765e-01
-3.29067200e-01 5.75198054e-01 -3.68461758e-01 -1.13095947e-01
1.67630887e+00 7.62305781e-03 2.09642611e-02 2.12934196e-01
1.09729242e+00 1.65932804e-01 -1.82280147e+00 -3.28223646e-01
-2.71078229e-01 -7.25567102e-01 3.06350440e-01 -8.22831035e-01
-1.14488268e+00 1.13030875e+00 4.66552377e-01 -8.63822401e-02
1.13965166e+00 -3.12898755e-01 7.02789724e-01 4.58805770e-01
-2.81271543e-02 -8.24194670e-01 2.14107439e-01 3.84559989e-01
8.29087675e-01 -1.40384793e+00 2.97141671e-01 -3.98968995e-01
-4.59805816e-01 1.07063568e+00 5.67564070e-01 -1.65469363e-01
3.33755761e-01 3.86290252e-01 3.96919250e-01 -3.85300159e-01
-7.02607870e-01 -3.55711617e-02 3.90980422e-01 6.18041456e-01
4.26349461e-01 -4.15138930e-01 2.45247677e-01 1.19558960e-01
-5.51616475e-02 -2.57569641e-01 5.04284620e-01 1.13659012e+00
-5.04039168e-01 -9.49723899e-01 -4.24639285e-01 4.49519575e-01
-4.18948710e-01 -4.96151477e-01 -3.28618586e-01 6.77110434e-01
-7.21709654e-02 1.05821514e+00 1.34860829e-01 -2.96515524e-01
1.64612085e-01 -1.88232094e-01 6.64939404e-01 -7.40353227e-01
-5.77541113e-01 2.02123314e-01 2.82483876e-01 -7.61658788e-01
-4.15051401e-01 -6.29673541e-01 -1.02176523e+00 2.09953450e-03
-1.60826787e-01 -5.02077520e-01 8.12925339e-01 1.06120420e+00
6.13792777e-01 6.34143412e-01 4.27512646e-01 -1.26817715e+00
-4.81054872e-01 -6.78230464e-01 -2.19098032e-01 4.76087630e-01
9.92327392e-01 -5.06022573e-01 -4.95786101e-01 1.35951400e-01] | [10.137740135192871, -2.80237078666687] |
3ebfa638-3125-45bb-b029-b928fe44901b | adversarial-robustness-of-prompt-based-few | 2306.11066 | null | https://arxiv.org/abs/2306.11066v2 | https://arxiv.org/pdf/2306.11066v2.pdf | Adversarial Robustness of Prompt-based Few-Shot Learning for Natural Language Understanding | State-of-the-art few-shot learning (FSL) methods leverage prompt-based fine-tuning to obtain remarkable results for natural language understanding (NLU) tasks. While much of the prior FSL methods focus on improving downstream task performance, there is a limited understanding of the adversarial robustness of such methods. In this work, we conduct an extensive study of several state-of-the-art FSL methods to assess their robustness to adversarial perturbations. To better understand the impact of various factors towards robustness (or the lack of it), we evaluate prompt-based FSL methods against fully fine-tuned models for aspects such as the use of unlabeled data, multiple prompts, number of few-shot examples, model size and type. Our results on six GLUE tasks indicate that compared to fully fine-tuned models, vanilla FSL methods lead to a notable relative drop in task performance (i.e., are less robust) in the face of adversarial perturbations. However, using (i) unlabeled data for prompt-based FSL and (ii) multiple prompts flip the trend. We further demonstrate that increasing the number of few-shot examples and model size lead to increased adversarial robustness of vanilla FSL methods. Broadly, our work sheds light on the adversarial robustness evaluation of prompt-based FSL methods for NLU tasks. | ['Srijan Kumar', 'Subhabrata Mukherjee', 'Gaurav Verma', 'Venkata Prabhakara Sarath Nookala'] | 2023-06-19 | null | null | null | null | ['adversarial-robustness', 'few-shot-learning'] | ['adversarial', 'methodology'] | [ 2.97467470e-01 5.52743673e-02 -7.01010600e-02 -9.65660438e-02
-9.58955824e-01 -8.44924688e-01 1.00923419e+00 -6.85950145e-02
-5.80728710e-01 6.13116443e-01 4.76973772e-01 -4.26291198e-01
3.31630148e-02 -7.06578195e-01 -9.35599685e-01 -3.39938343e-01
1.61619693e-01 2.21329704e-01 4.88714784e-01 -7.03637004e-01
1.34096757e-01 3.55120182e-01 -1.18618345e+00 1.38406798e-01
7.22940564e-01 4.95466232e-01 -3.60812098e-01 7.50604510e-01
1.22854963e-01 9.50629473e-01 -7.95521736e-01 -4.56540912e-01
4.32577938e-01 -3.11012685e-01 -8.62951934e-01 -4.79506582e-01
5.64673603e-01 -5.46535194e-01 -8.04737568e-01 1.14731240e+00
8.15044999e-01 7.44775891e-01 7.85732746e-01 -1.42797220e+00
-8.14167738e-01 7.48334765e-01 -1.82051435e-01 6.43130064e-01
3.15129071e-01 9.68613863e-01 8.04690719e-01 -5.32203496e-01
8.63750339e-01 1.55171156e+00 7.92244256e-01 1.05170453e+00
-1.24366117e+00 -7.81815648e-01 2.91120112e-01 1.55458122e-01
-8.92110348e-01 -9.14849699e-01 4.46535349e-01 -5.39615273e-01
1.24013615e+00 8.87990743e-03 -1.17034748e-01 1.72671735e+00
2.31658429e-01 6.23911738e-01 1.03392148e+00 -3.73504817e-01
4.89245057e-01 -6.85881451e-02 4.55683440e-01 3.93458366e-01
1.10419318e-01 7.79556572e-01 -4.82186764e-01 -2.11369663e-01
2.67563611e-01 -2.90707290e-01 -2.31755406e-01 2.77612302e-02
-9.48275924e-01 9.38697636e-01 2.94016093e-01 2.09932074e-01
2.08989143e-01 3.41736048e-01 8.44195962e-01 5.50928950e-01
6.05564296e-01 1.12444806e+00 -4.81649697e-01 -3.37663025e-01
-7.41965473e-01 3.30094993e-01 8.58361185e-01 7.13370442e-01
5.91556013e-01 4.08661604e-01 -8.80074024e-01 6.29690945e-01
-2.04104677e-01 1.84357420e-01 5.39659500e-01 -7.89427936e-01
5.20027101e-01 1.61902279e-01 -5.16794622e-02 -5.49541056e-01
-4.17863965e-01 -1.13742962e-01 -4.46726918e-01 4.21106368e-01
5.26380479e-01 -4.25795168e-01 -1.20781827e+00 2.12623596e+00
1.98974013e-02 4.58255857e-01 2.31321022e-01 7.43919075e-01
9.35073972e-01 4.06788111e-01 3.30758005e-01 2.26484947e-02
1.00001538e+00 -9.90870059e-01 -5.68614364e-01 -4.56081092e-01
7.35765636e-01 -5.84731281e-01 1.39037299e+00 -6.41827583e-02
-8.71993244e-01 -5.81082046e-01 -1.05272257e+00 2.11876798e-02
-6.15969479e-01 -6.79155827e-01 3.60170901e-01 8.66855621e-01
-8.20842147e-01 9.35590923e-01 -7.51933753e-01 -3.81608695e-01
4.13647711e-01 4.61553112e-02 -4.07206923e-01 -2.31416255e-01
-1.81368613e+00 1.22388053e+00 3.52340758e-01 -5.34444690e-01
-1.36086059e+00 -1.30427372e+00 -1.13298178e+00 1.74989760e-01
6.55833483e-01 -6.92819178e-01 1.51586509e+00 -4.72752988e-01
-1.37936771e+00 5.48527837e-01 1.73985824e-01 -6.88573956e-01
7.95789778e-01 -4.69762653e-01 -3.19642127e-01 -6.96739741e-03
-2.18752027e-02 5.85630715e-01 1.07932866e+00 -1.03673851e+00
5.08198440e-02 -1.85642410e-02 4.58515614e-01 -2.69785430e-02
-1.96916357e-01 1.11576013e-01 -6.93263263e-02 -9.24620390e-01
-6.36151433e-01 -8.59891593e-01 -2.31351927e-01 -1.37726232e-01
-4.77373928e-01 -1.41192988e-01 9.00214970e-01 -2.72074163e-01
1.20688736e+00 -2.17740417e+00 -6.39846101e-02 -2.33036265e-01
1.01948462e-01 8.16595137e-01 -4.40180629e-01 5.65453827e-01
-2.31098965e-01 5.58092058e-01 -2.76072502e-01 -1.78052202e-01
2.12160319e-01 1.33183375e-02 -5.98454416e-01 2.72249758e-01
3.86691630e-01 1.30516100e+00 -1.15069234e+00 -1.99364081e-01
3.45231682e-01 1.28758013e-01 -6.79891467e-01 1.58622935e-01
-4.20111209e-01 4.03073043e-01 -2.68142730e-01 5.88378906e-01
2.85988659e-01 1.77882776e-01 -4.81941283e-01 -1.53174728e-01
1.63672835e-01 5.80742881e-02 -8.89464140e-01 1.51140130e+00
-4.45839703e-01 5.71727157e-01 -3.15475345e-01 -5.83210588e-01
5.70295215e-01 4.83466953e-01 6.76565978e-04 -6.26781523e-01
1.79317877e-01 -5.93083352e-02 1.42726377e-01 -5.01067579e-01
2.01560602e-01 -5.52109540e-01 -2.96163738e-01 6.79323614e-01
2.55652934e-01 -3.94870847e-01 3.72567952e-01 5.87850153e-01
1.72567379e+00 -3.05628944e-02 4.22844946e-01 -1.65872306e-01
1.38513699e-01 -1.04943067e-01 4.17308211e-01 1.39680493e+00
-8.22064459e-01 4.78607178e-01 4.90598679e-01 -2.78783828e-01
-1.11928439e+00 -1.15829527e+00 1.16269052e-01 1.44728267e+00
-6.05295859e-02 -3.31404030e-01 -9.09058571e-01 -9.17112768e-01
2.41359934e-01 1.26628602e+00 -8.40931177e-01 -9.46454883e-01
-2.17056468e-01 -6.18824363e-01 1.19225609e+00 5.83597422e-01
4.14993137e-01 -1.32848573e+00 -4.43871975e-01 2.10251715e-02
1.29132178e-02 -1.13266015e+00 -7.32067525e-01 4.91707250e-02
-5.19420862e-01 -1.11236298e+00 -3.83786023e-01 -1.12622105e-01
2.96097189e-01 4.60886918e-02 1.16860855e+00 -1.32568687e-01
-3.68148208e-01 5.67911386e-01 -5.42672992e-01 -5.25083363e-01
-7.76643991e-01 -4.65265810e-02 3.55554491e-01 -4.96886134e-01
2.08079129e-01 -4.79187667e-01 -1.69381201e-01 3.10392439e-01
-1.21726179e+00 -5.62415481e-01 2.53101736e-01 8.74321580e-01
1.03727013e-01 -2.53466338e-01 7.97927618e-01 -1.28350365e+00
1.09570837e+00 -5.21732330e-01 -1.70552641e-01 3.48867059e-01
-5.06707728e-01 3.07420790e-01 8.73753726e-01 -7.36763299e-01
-1.11499667e+00 -4.27623600e-01 -1.63481608e-01 -9.19803500e-01
-2.40031242e-01 1.98904291e-01 -1.08872093e-01 -2.68102348e-01
1.44729602e+00 -2.07602203e-01 -1.41652375e-01 -2.91695535e-01
7.90843666e-01 3.35609466e-01 4.85691696e-01 -6.65005922e-01
1.10113978e+00 2.22334519e-01 -2.48949304e-01 -6.09186828e-01
-1.02122223e+00 -2.94068128e-01 -4.89858747e-01 -6.81432784e-02
7.69420087e-01 -6.90617681e-01 -3.41262162e-01 5.51409781e-01
-8.44929159e-01 -7.67910123e-01 -4.87481296e-01 1.88073203e-01
-7.06311762e-01 3.55628252e-01 -7.00019538e-01 -5.75762272e-01
-4.19002771e-01 -1.34819567e+00 7.60094583e-01 -3.29518132e-02
-6.39183104e-01 -9.97371137e-01 2.86567152e-01 3.32081854e-01
5.55298388e-01 2.56995767e-01 1.09453487e+00 -1.09639549e+00
-1.06831774e-01 -2.54553348e-01 -9.84880328e-02 2.86631525e-01
-4.83545400e-02 -1.18211269e-01 -1.24375677e+00 -4.27545696e-01
-4.43084948e-02 -8.68120193e-01 1.10900986e+00 2.20443442e-01
8.77128363e-01 -3.94125432e-01 -7.55333826e-02 7.49230981e-01
1.17699802e+00 7.68017257e-04 6.53667152e-01 3.24980080e-01
6.69846594e-01 2.90726721e-01 6.51303768e-01 3.00527573e-01
-3.04866970e-01 4.27335262e-01 3.79489481e-01 2.26483449e-01
-3.61255854e-01 -3.65267754e-01 5.40761054e-01 1.20972760e-01
2.54115015e-01 -2.90335268e-01 -9.74294901e-01 4.69438255e-01
-1.67525184e+00 -1.19121432e+00 5.16689122e-01 1.93941116e+00
9.67476249e-01 4.81083095e-01 -2.11715754e-02 5.77772819e-02
6.80547774e-01 5.75919747e-01 -1.03794932e+00 -6.86111331e-01
-8.78989417e-03 1.61134228e-01 3.77655864e-01 6.06831789e-01
-1.16417634e+00 1.42549300e+00 6.87749815e+00 1.05986702e+00
-8.65231693e-01 1.56822398e-01 5.90291321e-01 -5.04308939e-01
-3.69814545e-01 -8.05260241e-02 -7.25047767e-01 3.80950660e-01
8.74463260e-01 -5.13746917e-01 7.64507949e-01 8.39393497e-01
1.02282822e-01 1.30442590e-01 -1.21127462e+00 3.67536783e-01
5.41783832e-02 -1.28550422e+00 7.63546601e-02 -3.60436350e-01
1.02274561e+00 2.31689453e-01 2.82971766e-02 8.79996657e-01
8.90684724e-01 -1.17099369e+00 6.24913812e-01 3.32592398e-01
9.66225386e-01 -6.93748176e-01 5.91643155e-01 4.36636448e-01
-8.14879417e-01 -2.99502462e-01 -3.13382894e-01 -1.49974316e-01
2.20795199e-01 3.65363628e-01 -7.64191985e-01 2.48429373e-01
6.20437324e-01 3.61820549e-01 -6.14938080e-01 5.25886595e-01
-4.72361356e-01 8.50846708e-01 -8.61441940e-02 1.80570513e-01
4.35122609e-01 4.02759701e-01 9.93400216e-01 1.30688179e+00
-2.03441188e-01 3.82705212e-01 9.90034789e-02 9.57556963e-01
-2.99587071e-01 -2.36031637e-01 -1.00240135e+00 -3.63522828e-01
7.15325773e-01 8.59936774e-01 -3.45859736e-01 -2.53227741e-01
-3.12461138e-01 7.22057998e-01 6.00658953e-01 4.90049064e-01
-7.17854023e-01 -3.49280357e-01 9.12135005e-01 5.26140556e-02
-1.37606841e-02 -8.54935199e-02 -3.45470846e-01 -1.05086374e+00
-3.06890815e-01 -1.22390771e+00 4.95864868e-01 -6.07777238e-01
-1.64481115e+00 4.10966158e-01 1.06378444e-01 -8.33612919e-01
-3.22353721e-01 -3.73795539e-01 -1.22534370e+00 7.10447431e-01
-1.21175003e+00 -8.84528220e-01 -2.18049921e-02 6.47816122e-01
7.50501692e-01 -4.84364599e-01 8.02807927e-01 -2.17814103e-01
-8.00982237e-01 1.00832784e+00 8.05341005e-02 1.35812387e-01
1.19878435e+00 -1.02252150e+00 1.15454626e+00 1.26804757e+00
-1.04672723e-01 8.53023469e-01 9.48290348e-01 -9.33837771e-01
-9.96845961e-01 -1.18872321e+00 4.20024484e-01 -8.00906658e-01
9.57574427e-01 -1.75419912e-01 -9.03669357e-01 6.90603375e-01
-1.27577081e-01 1.25796676e-01 5.30519128e-01 5.63723370e-02
-6.93860412e-01 3.49271595e-01 -1.39419580e+00 9.53700483e-01
1.23280621e+00 -6.94182813e-01 -7.78324604e-01 1.74786374e-01
1.37541795e+00 -3.09120327e-01 -6.61397815e-01 5.66110969e-01
3.43567967e-01 -9.75177765e-01 1.08602238e+00 -1.15415251e+00
6.02223873e-01 8.95533189e-02 -1.32884875e-01 -1.75827909e+00
-4.87608135e-01 -7.66700685e-01 -2.42968187e-01 1.02188170e+00
3.71540487e-01 -6.40975237e-01 4.57374722e-01 9.76929367e-01
-2.53405303e-01 -6.90102339e-01 -7.98666060e-01 -1.04150307e+00
3.41435373e-01 -5.08525491e-01 3.51245821e-01 9.21645999e-01
3.54917273e-02 2.76225030e-01 -3.65512818e-01 -1.51730448e-01
3.91546845e-01 -4.53418761e-01 7.75615215e-01 -8.69105339e-01
-4.16230381e-01 -3.06509465e-01 -1.94665685e-01 -1.62811607e-01
6.34091020e-01 -8.45167458e-01 8.99835229e-02 -1.29906166e+00
5.47556169e-02 -1.52005613e-01 -2.87137121e-01 8.20169210e-01
-8.94547582e-01 9.18391943e-02 6.52085543e-01 8.02852288e-02
-5.04589856e-01 4.89485115e-01 9.14633453e-01 -2.67675966e-01
-1.79709256e-01 1.03700755e-03 -7.38812625e-01 6.19779944e-01
7.67002642e-01 -4.14719045e-01 -6.75510168e-01 -3.67513537e-01
-4.72431108e-02 -3.39277923e-01 3.06171656e-01 -9.63131428e-01
1.39187813e-01 -2.21052393e-01 6.82895035e-02 1.44123450e-01
2.79850096e-01 -2.90738255e-01 -3.18125457e-01 6.18458986e-01
-4.96207267e-01 -2.62434185e-01 6.06358409e-01 7.03608274e-01
2.43883505e-01 -4.42803234e-01 1.11136270e+00 -3.82426649e-01
-8.71807694e-01 3.62434715e-01 -3.51267397e-01 7.94976294e-01
1.03205466e+00 -1.69100687e-02 -6.91852093e-01 -4.16243464e-01
-7.30216086e-01 1.79547876e-01 5.45016587e-01 7.16806114e-01
3.97350311e-01 -9.88926888e-01 -7.45088160e-01 1.51698902e-01
2.99938440e-01 -3.04360151e-01 4.48511779e-01 4.11173612e-01
-4.09084894e-02 2.84346879e-01 -2.35842928e-01 5.35938665e-02
-1.02758133e+00 9.05960619e-01 2.83039093e-01 -4.31684047e-01
-4.04702842e-01 1.26825142e+00 2.95302719e-01 -6.92919791e-01
3.49918962e-01 2.20423490e-01 1.22421049e-01 5.77002391e-02
5.93975902e-01 8.45059454e-01 5.34991212e-02 -1.69059753e-01
-3.58448476e-01 1.63265705e-01 -3.50050092e-01 -2.91774303e-01
9.87122476e-01 3.35907131e-01 4.41655964e-01 3.74003828e-01
1.00896764e+00 -2.38276482e-01 -1.45539129e+00 -3.75958622e-01
-2.79017836e-01 -3.40962023e-01 -3.54904830e-02 -9.89066601e-01
-7.11872637e-01 9.15675819e-01 2.14065880e-01 -2.12263063e-01
7.67992973e-01 -1.53551474e-01 7.82181740e-01 4.27339375e-01
3.39413315e-01 -1.04575408e+00 3.12597752e-01 9.02467251e-01
8.07392359e-01 -1.41079366e+00 1.03927450e-02 -1.14622414e-01
-9.53994453e-01 6.06028676e-01 8.86447668e-01 -5.00565171e-02
5.25625229e-01 3.55955690e-01 1.21911965e-01 -6.02161279e-03
-9.64341938e-01 -1.41131133e-01 1.86965168e-01 8.19186687e-01
2.57714763e-02 -1.97961822e-01 1.66005835e-01 5.98959267e-01
-3.43250930e-01 -1.62869930e-01 4.60464269e-01 8.16421688e-01
-4.25400227e-01 -7.45989621e-01 -3.92579287e-01 6.26782417e-01
-4.75512028e-01 -3.57683033e-01 -6.26094222e-01 7.02863157e-01
-2.00342566e-01 1.25644660e+00 -2.97109604e-01 -4.30139095e-01
5.54807961e-01 3.32426429e-01 3.70144784e-01 -9.82753158e-01
-8.58808160e-01 -6.29220426e-01 2.42002398e-01 -7.21395612e-01
3.45667958e-01 -3.87574613e-01 -9.92196262e-01 -4.49412078e-01
-3.21944684e-01 -2.44977146e-01 2.19829246e-01 1.28745496e+00
2.05252349e-01 7.36173034e-01 3.40098053e-01 -9.69568908e-01
-1.16250551e+00 -1.18663549e+00 -2.13566199e-01 7.24414229e-01
3.45114797e-01 -7.28596687e-01 -7.74638116e-01 -1.20420061e-01] | [6.148858547210693, 8.15090560913086] |
58624016-c4cf-4977-9480-f8e9d7229284 | yaso-a-new-benchmark-for-targeted-sentiment | 2012.14541 | null | https://arxiv.org/abs/2012.14541v2 | https://arxiv.org/pdf/2012.14541v2.pdf | YASO: A Targeted Sentiment Analysis Evaluation Dataset for Open-Domain Reviews | Current TSA evaluation in a cross-domain setup is restricted to the small set of review domains available in existing datasets. Such an evaluation is limited, and may not reflect true performance on sites like Amazon or Yelp that host diverse reviews from many domains. To address this gap, we present YASO - a new TSA evaluation dataset of open-domain user reviews. YASO contains 2,215 English sentences from dozens of review domains, annotated with target terms and their sentiment. Our analysis verifies the reliability of these annotations, and explores the characteristics of the collected data. Benchmark results using five contemporary TSA systems show there is ample room for improvement on this challenging new dataset. YASO is available at https://github.com/IBM/yaso-tsa. | ['Noam Slonim', 'Yoav Katz', 'Ranit Aharonov', 'Artem Spector', 'Orith Toledo-Ronen', 'Matan Orbach'] | 2020-12-29 | null | https://aclanthology.org/2021.emnlp-main.721 | https://aclanthology.org/2021.emnlp-main.721.pdf | emnlp-2021-11 | ['aspect-extraction'] | ['natural-language-processing'] | [-2.62119502e-01 -1.59571156e-01 -5.27067721e-01 -5.64661980e-01
-1.11336517e+00 -1.01444042e+00 4.84679788e-01 3.63337487e-01
-1.96256623e-01 7.16782331e-01 2.56374627e-01 -2.81246752e-01
1.97140425e-01 -3.89478147e-01 -2.62096733e-01 -8.04917067e-02
3.90351176e-01 7.11367488e-01 2.76837051e-01 -7.63539910e-01
2.84997433e-01 -2.86298901e-01 -1.09750223e+00 4.18598771e-01
7.92564273e-01 1.03963256e+00 -1.80830285e-01 3.77123564e-01
1.20082781e-01 4.85216767e-01 -7.55671203e-01 -1.02373147e+00
3.94497007e-01 7.42039233e-02 -7.10943639e-01 -2.91901175e-02
4.63484496e-01 -1.82891533e-01 3.01190857e-02 6.46562517e-01
3.75881732e-01 -1.20268732e-01 6.12946987e-01 -1.36884201e+00
-1.15198302e+00 6.02015316e-01 -4.53052193e-01 4.04564977e-01
3.49056691e-01 -1.05331868e-01 1.74715841e+00 -1.05907094e+00
1.12575030e+00 7.75133550e-01 7.45400906e-01 4.64474380e-01
-6.72754109e-01 -6.98589265e-01 7.25432634e-02 -2.08518520e-01
-1.15754271e+00 -4.05199230e-01 4.28446978e-01 -1.83391318e-01
1.28720462e+00 -3.05005815e-02 4.78636563e-01 1.43076372e+00
2.15160221e-01 1.00688434e+00 1.19501340e+00 -4.26771820e-01
1.03399396e-01 5.29204428e-01 5.19197285e-01 2.06025675e-01
4.46613520e-01 -6.49816155e-01 -7.32145786e-01 -3.34778845e-01
-1.29324898e-01 -2.34506309e-01 1.53089417e-02 -2.26728991e-01
-8.26609254e-01 7.97954798e-01 1.57572068e-02 3.91044378e-01
-1.20492645e-01 -3.65854651e-01 1.00915945e+00 6.22241139e-01
1.14424336e+00 7.80287385e-01 -1.16052830e+00 -5.93869448e-01
-6.51938021e-01 3.92073184e-01 1.26986563e+00 1.16484165e+00
5.21755219e-01 -3.95011038e-01 3.92293364e-01 1.19946051e+00
9.61395800e-02 7.00235963e-01 6.53193235e-01 -7.31095016e-01
7.58675754e-01 7.60988653e-01 1.42381102e-01 -8.69906187e-01
-7.92467296e-02 -3.38173568e-01 -9.26509574e-02 -3.44720155e-01
3.86387736e-01 -2.68939316e-01 -6.10832095e-01 1.34242070e+00
2.35931933e-01 -3.45922828e-01 1.11704737e-01 5.71505964e-01
1.11354136e+00 5.31941533e-01 -2.36700699e-01 1.97580338e-01
1.47722781e+00 -1.17767632e+00 -6.35002553e-01 -6.29929423e-01
9.92413938e-01 -1.07840776e+00 1.47578347e+00 5.89654982e-01
-8.65696430e-01 1.17104482e-02 -1.17102695e+00 -3.49837124e-01
-8.85300756e-01 3.88525575e-01 4.87761766e-01 5.57286024e-01
-9.33755934e-01 9.81457978e-02 -4.49082106e-01 -8.06743503e-01
3.52843761e-01 8.79802927e-02 -4.30872619e-01 -1.98098391e-01
-1.45735800e+00 1.02603889e+00 -1.93601087e-01 -1.24854289e-01
-6.49240971e-01 -6.58812582e-01 -8.15194607e-01 -1.12297446e-01
3.93544137e-01 -2.51979709e-01 2.05494022e+00 -1.00599694e+00
-1.12771380e+00 1.07285261e+00 -2.63011158e-01 -4.04837608e-01
8.42392668e-02 -6.63486302e-01 -8.28751266e-01 1.86779052e-02
6.09702945e-01 3.54846418e-01 3.88253152e-01 -8.67559552e-01
-2.85561889e-01 -2.97215909e-01 2.05446959e-01 2.29363993e-01
-6.59889102e-01 3.62960488e-01 -5.30104399e-01 -5.58447361e-01
-5.86159408e-01 -1.15296328e+00 -5.65066971e-02 -5.75220883e-01
-2.92440444e-01 -4.24880058e-01 7.30403244e-01 -5.13897181e-01
1.45025814e+00 -2.19219422e+00 -5.38284779e-01 -1.50115173e-02
2.52921611e-01 3.63552600e-01 -2.22966239e-01 9.77744818e-01
1.29548227e-03 4.50676173e-01 -2.96345521e-02 -7.47148991e-01
9.01931152e-02 -2.60477792e-02 -5.63673794e-01 2.34778851e-01
2.99822837e-01 9.55139458e-01 -8.83418858e-01 -2.58752435e-01
-3.13935369e-01 1.10274203e-01 -1.22129694e-01 -1.97158962e-01
-3.12848836e-01 -1.06500797e-01 -8.33313167e-01 9.13542926e-01
5.18950641e-01 -6.16597354e-01 1.10439166e-01 1.01020634e-01
1.02456607e-01 1.01003289e+00 -5.00949562e-01 1.54112375e+00
-4.52882618e-01 9.30534422e-01 -1.41374946e-01 -4.09548044e-01
1.14185381e+00 3.36388171e-01 4.49786276e-01 -7.44619846e-01
4.65799600e-01 7.20575690e-01 -3.05255175e-01 -1.82280704e-01
1.02355814e+00 6.53913841e-02 -5.28143644e-01 6.05267942e-01
8.33345503e-02 -5.51226974e-01 5.80952585e-01 5.95630825e-01
1.47687149e+00 -2.80944914e-01 4.09165949e-01 -1.95112661e-01
2.24768162e-01 5.47740281e-01 5.24999261e-01 2.41797090e-01
-5.53105593e-01 8.07097435e-01 7.71733642e-01 -2.19497874e-01
-1.36818171e+00 -7.65321493e-01 -3.55256706e-01 9.97744501e-01
-9.82719511e-02 -1.04803967e+00 -4.90403593e-01 -8.33862126e-01
5.79361096e-02 5.29078007e-01 -9.13089216e-01 1.13122419e-01
-7.31676966e-02 -6.13494217e-01 5.39321244e-01 2.92340696e-01
2.16135383e-01 -9.09466803e-01 -1.71578869e-01 9.27866772e-02
-2.51951784e-01 -1.39290142e+00 -6.93482459e-01 1.65505469e-01
-5.47707736e-01 -1.14886332e+00 -4.70438391e-01 -5.73940039e-01
2.96884179e-01 4.07525867e-01 1.80926168e+00 -1.93618834e-01
-4.53760959e-02 3.79890501e-01 -9.57155764e-01 -7.49238312e-01
-3.15379292e-01 5.97755909e-01 -5.72389625e-02 -4.66540486e-01
1.23632693e+00 -2.65246093e-01 -5.45237899e-01 6.02264404e-01
-6.38071537e-01 -4.96407390e-01 3.83788615e-01 6.82310343e-01
1.63816631e-01 -3.25024366e-01 1.16989791e+00 -1.46034372e+00
1.06897271e+00 -9.96552289e-01 -3.22109014e-01 2.33703375e-01
-8.55756700e-01 -5.99021256e-01 4.93462235e-01 -2.06899241e-01
-9.44731414e-01 -4.63874131e-01 9.44732875e-02 7.03318417e-02
1.28890455e-01 8.29401016e-01 2.92012900e-01 5.79442203e-01
9.07934129e-01 -3.60164434e-01 -7.37144053e-02 -4.80544835e-01
2.67627031e-01 9.76200163e-01 4.74091843e-02 -5.69313169e-01
4.72562164e-01 2.07128048e-01 -8.41175318e-01 -3.19311261e-01
-1.18512726e+00 -9.74741757e-01 -3.91847014e-01 1.28005251e-01
4.38396394e-01 -1.18865621e+00 -1.07547298e-01 4.04443324e-01
-7.69841015e-01 -2.72878826e-01 -2.24573225e-01 1.85954347e-01
5.59933782e-02 8.16861764e-02 -8.56931627e-01 -3.82086843e-01
-7.46735394e-01 -9.75292444e-01 1.19347739e+00 6.31667674e-02
-6.65100157e-01 -1.06813955e+00 5.79042494e-01 8.19227874e-01
3.65940124e-01 -3.85136940e-02 2.25124240e-01 -1.26397371e+00
9.95925516e-02 -7.14304149e-01 7.39475712e-03 7.80860782e-01
1.78027675e-01 7.01258898e-01 -9.46076393e-01 -2.44860440e-01
-1.86807200e-01 -1.11987281e+00 5.41689217e-01 1.19406126e-01
5.11340678e-01 1.34941667e-01 -1.79094777e-01 -3.41781676e-01
1.26660955e+00 -8.50101337e-02 2.07561076e-01 5.56696236e-01
3.03910255e-01 5.94593823e-01 1.25697136e+00 6.88064277e-01
8.01701903e-01 3.84000957e-01 6.69758096e-02 2.15315506e-01
2.70572364e-01 8.41746293e-03 7.05828130e-01 1.31549549e+00
4.99601811e-01 -5.56996584e-01 -1.27092421e+00 1.05340445e+00
-1.62296879e+00 -5.53600132e-01 -1.47654921e-01 1.76136327e+00
9.25475955e-01 4.70742524e-01 2.07212895e-01 -2.52657264e-01
5.19085526e-01 2.70522028e-01 -5.14443636e-01 -8.90408635e-01
-3.60246062e-01 4.66225386e-01 4.26503032e-01 2.25750864e-01
-9.17694509e-01 9.55130816e-01 6.40576124e+00 8.19322824e-01
-1.00754762e+00 3.64737809e-01 8.63807738e-01 -3.62314999e-01
-3.62025529e-01 -3.94763332e-03 -9.30927277e-01 3.28004986e-01
1.39649951e+00 -6.72445118e-01 -1.46784157e-01 1.14153242e+00
5.30089103e-02 -1.91412285e-01 -1.03146625e+00 4.94097680e-01
1.38382375e-01 -1.06720436e+00 -4.47670221e-01 -2.12521534e-02
9.39128101e-01 8.39373529e-01 3.89313340e-01 6.62401259e-01
5.58472693e-01 -9.07475948e-01 5.63693285e-01 -2.33150586e-01
7.74629056e-01 -5.94656169e-01 1.06736672e+00 -3.83881889e-02
-8.83146644e-01 1.42837375e-01 -3.14016014e-01 -1.57918900e-01
1.13054425e-01 9.45700288e-01 -1.22894728e+00 4.19769287e-01
9.70657647e-01 1.42537701e+00 -1.00152564e+00 5.20152926e-01
-1.53255790e-01 9.20366049e-01 -3.75595957e-01 -4.61548328e-01
4.38013852e-01 -6.33330122e-02 2.12400004e-01 1.24334693e+00
1.09086744e-01 -2.03855067e-01 -7.74575816e-03 2.74345696e-01
-5.09682238e-01 4.08282489e-01 -9.34078813e-01 -3.04117233e-01
8.69608581e-01 1.66464913e+00 -4.94858772e-01 -4.96544331e-01
-9.48291004e-01 5.20762146e-01 3.86586130e-01 1.93862796e-01
-6.87114358e-01 -2.98262626e-01 6.88934207e-01 1.07866339e-03
2.05440447e-01 -1.71351463e-01 -3.85315806e-01 -1.42708755e+00
4.25985694e-01 -1.38903832e+00 2.77488112e-01 -8.49966049e-01
-1.64522171e+00 7.90014505e-01 -8.14906806e-02 -1.41464937e+00
-2.38963887e-02 -7.80539095e-01 -3.78818303e-01 6.94032252e-01
-1.41237593e+00 -1.09002101e+00 -1.46448947e-02 3.48896801e-01
7.80018389e-01 -1.84593230e-01 6.05631530e-01 6.07520938e-01
-6.44631267e-01 6.01587951e-01 4.10846204e-01 2.85009127e-02
1.56146884e+00 -1.37301970e+00 9.77965415e-01 4.52439904e-01
5.71786426e-02 8.45529377e-01 6.67503774e-01 -5.75643957e-01
-9.68127966e-01 -6.78217590e-01 1.19448423e+00 -1.32223415e+00
1.31279588e+00 -5.21541595e-01 -8.79857540e-01 8.12903941e-01
7.56085813e-01 -3.08154467e-02 1.04160964e+00 5.78975797e-01
-6.18113339e-01 -2.30478227e-01 -9.32459533e-01 4.10766512e-01
4.97168362e-01 -7.17822313e-01 -6.48300946e-01 4.60975021e-01
7.44913161e-01 -4.70894098e-01 -1.01870847e+00 2.85566121e-01
6.33780658e-01 -5.51060438e-01 3.81243289e-01 -4.77550685e-01
9.81834054e-01 1.40774148e-02 -1.30563706e-01 -1.46346021e+00
2.04094812e-01 -3.05879384e-01 2.14187190e-01 1.71109438e+00
1.11804509e+00 -6.44324124e-01 7.01635480e-01 9.07354355e-01
-2.06239983e-01 -1.21814728e+00 -6.18775487e-01 -6.26038253e-01
4.10640657e-01 -5.80842257e-01 5.42194068e-01 1.20021629e+00
3.24057698e-01 6.35321259e-01 -6.77919015e-02 -3.61555845e-01
-6.63487241e-02 -1.21553071e-01 8.82260740e-01 -9.02795315e-01
-3.86438780e-02 -1.55426383e-01 1.63707174e-02 -7.77484000e-01
2.93436825e-01 -6.67849720e-01 -1.46059766e-01 -1.39523506e+00
1.21930249e-01 -6.15231872e-01 -2.57263631e-01 6.03578866e-01
-3.15696187e-02 3.69035989e-01 -6.55385554e-02 1.70832813e-01
-9.45723057e-01 5.29229999e-01 1.04597867e+00 -6.56379536e-02
1.45848528e-01 -2.49305099e-01 -1.06653333e+00 4.13344473e-01
1.09700096e+00 -3.75638276e-01 -7.14027524e-01 -5.82532644e-01
7.80426502e-01 -5.13466001e-01 -2.48064175e-01 -5.27297020e-01
-1.31508540e-02 9.60722789e-02 -2.43151084e-01 -4.77862656e-01
3.17104042e-01 -6.80479407e-01 -3.83840770e-01 -2.27567241e-01
-3.96340489e-01 6.59789741e-01 2.47441098e-01 2.35962540e-01
-6.31517172e-01 -2.85570949e-01 3.99490207e-01 -1.62597284e-01
-6.74334705e-01 9.92004424e-02 -3.33532155e-01 7.82051980e-01
7.54728496e-01 1.51731759e-01 -9.23908412e-01 -4.56256896e-01
-2.50788003e-01 4.80054587e-01 6.27982736e-01 8.44225645e-01
3.21182579e-01 -1.02147067e+00 -7.17093229e-01 -5.48322439e-01
7.73578405e-01 -1.12203978e-01 -8.83258134e-02 6.56977117e-01
-4.04187500e-01 4.60671693e-01 1.12735242e-01 -2.72360444e-01
-1.38012052e+00 2.65905499e-01 -6.60504252e-02 -8.07487905e-01
-3.08768064e-01 6.92036510e-01 -1.14321545e-01 -7.71320522e-01
-3.25573981e-01 -5.08988857e-01 -2.32529879e-01 3.06561917e-01
3.25020283e-01 1.45968094e-01 4.38789070e-01 -5.59452891e-01
-4.92499590e-01 2.00991586e-01 -5.55960417e-01 -2.24683970e-01
1.36019254e+00 -4.23757851e-01 -2.23025963e-01 8.10964048e-01
1.41230488e+00 3.78786415e-01 -5.78368902e-01 -3.85383576e-01
2.11689994e-01 -3.84149283e-01 -3.40972364e-01 -1.04810965e+00
-7.60748267e-01 6.38741076e-01 -2.09067222e-02 3.71103317e-01
1.05705476e+00 4.08537053e-02 1.05287266e+00 5.34189165e-01
3.18171263e-01 -1.35005879e+00 2.44616315e-01 8.67867649e-01
8.66250396e-01 -1.57204938e+00 2.55367637e-01 -3.70678544e-01
-1.23189175e+00 8.14050615e-01 9.61161613e-01 -1.23297125e-01
7.61064351e-01 1.90678984e-01 5.30578792e-01 -5.08561552e-01
-1.52818704e+00 1.78757668e-01 1.56477332e-01 1.02655798e-01
9.22548354e-01 6.94120750e-02 -6.14348948e-01 7.77349770e-01
-4.65962708e-01 2.11599935e-02 9.01925862e-01 1.01267898e+00
-1.61033466e-01 -1.42353773e+00 1.60576761e-01 6.98515654e-01
-8.42703879e-01 -3.55168968e-01 -8.34395766e-01 8.54700089e-01
-3.73932391e-01 1.13411295e+00 -3.31729412e-01 -1.57000542e-01
3.93551826e-01 -4.28211242e-02 -2.58967504e-02 -1.02568102e+00
-9.48900819e-01 -1.67468056e-01 8.56339574e-01 -2.97794491e-01
-4.43161607e-01 -1.04901910e+00 -8.49943519e-01 -4.53305334e-01
-3.29561085e-01 5.44040442e-01 4.74492639e-01 6.73135638e-01
6.05481386e-01 7.07487538e-02 7.35985339e-01 -1.61767736e-01
-4.57359940e-01 -1.17874014e+00 -6.38890803e-01 4.56495553e-01
2.15736151e-01 -5.31144321e-01 -5.16234636e-01 -7.92297125e-02] | [11.29773998260498, 6.839864253997803] |
83722873-364c-421d-91cc-7b3336d34e37 | scprisma-infers-filters-and-enhances | null | null | https://www.nature.com/articles/s41587-023-01663-5 | https://www.nature.com/articles/s41587-023-01663-5.pdf | scPrisma infers, filters and enhances topological signals in single-cell data using spectral template matching | Single-cell RNA sequencing has been instrumental in uncovering cellular spatiotemporal context. This task is challenging as cells simultaneously encode multiple, potentially cross-interfering, biological signals. Here we propose scPrisma, a spectral computational method that uses topological priors to decouple, enhance and filter different classes of biological processes in single-cell data, such as periodic and linear signals. We apply scPrisma to the analysis of the cell cycle in HeLa cells, circadian rhythm and spatial zonation in liver lobules, diurnal cycle in Chlamydomonas and circadian rhythm in the suprachiasmatic nucleus in the brain. scPrisma can be used to distinguish mixed cellular populations by specific characteristics such as cell type and uncover regulatory networks and cell–cell interactions specific to predefined biological signals, such as the circadian rhythm. We show scPrisma’s flexibility in incorporating prior knowledge, inference of topologically informative genes and generalization to additional diverse templates and systems. scPrisma can be used as a stand-alone workflow for signal analysis and as a prior step for downstream single-cell analysis. | ['Mor Nitzan', 'Yonathan Bornfeld', 'Jonathan Karin'] | 2023-02-27 | null | null | null | nature-biotechnology-2023-2 | ['template-matching'] | ['computer-vision'] | [ 5.56653976e-01 -5.13314903e-01 2.69416034e-01 1.62746698e-01
-4.06661570e-01 -1.28765893e+00 8.31075907e-01 3.69534105e-01
-3.21069390e-01 9.92191792e-01 4.37123477e-01 -2.04985082e-01
-3.41604501e-01 -3.02693337e-01 -4.83603567e-01 -1.22289336e+00
-1.62436843e-01 5.70399165e-01 1.80221438e-01 1.34218428e-02
1.14526778e-01 6.96452677e-01 -1.38961577e+00 2.96883732e-01
1.68217957e-01 1.73150152e-01 2.57918328e-01 1.00448000e+00
1.01471543e-02 -2.83683896e-01 -3.26438457e-01 4.37613070e-01
-8.48521814e-02 -6.15460634e-01 -3.02446365e-01 -2.76088417e-01
-9.02841091e-02 8.91819835e-01 8.57271478e-02 5.35041511e-01
8.41655970e-01 -2.95501113e-01 6.66849434e-01 -1.02502644e+00
2.63477653e-01 2.95676678e-01 -5.54811835e-01 4.24762666e-01
3.46560150e-01 3.59255522e-01 6.75760448e-01 -9.12302732e-01
1.14629745e+00 1.31605864e+00 6.12531543e-01 3.47570807e-01
-2.23900437e+00 -4.15092736e-01 -3.36791545e-01 -3.79814953e-01
-1.21496284e+00 -7.05814898e-01 3.10444176e-01 -7.58730114e-01
6.40202463e-01 5.92046797e-01 9.31331933e-01 1.12995398e+00
3.97881866e-01 2.35155344e-01 1.38302112e+00 -1.83227405e-01
5.20019174e-01 -8.12593997e-01 -1.88553840e-01 4.55460519e-01
2.00041741e-01 -1.33613601e-01 -8.15973938e-01 -4.71349418e-01
4.51909572e-01 -7.29843229e-02 -5.36389053e-01 6.18002377e-02
-1.82461536e+00 7.62639642e-02 -5.31666517e-01 4.36074346e-01
4.17942815e-02 2.37443000e-01 5.37647188e-01 1.58890948e-01
1.58321951e-02 3.88713658e-01 -8.84932399e-01 -3.86938035e-01
-1.01626635e+00 -1.61239102e-01 6.95843935e-01 5.25845349e-01
5.56387901e-01 -5.38803458e-01 -3.60973805e-01 6.02688491e-01
3.47591847e-01 4.08032000e-01 3.60319138e-01 -8.72792482e-01
-4.66712117e-01 4.51335490e-01 2.98971273e-02 -7.60791421e-01
-9.29364622e-01 -2.14588791e-01 -1.04459488e+00 -1.63997963e-01
1.00539994e+00 -2.58145034e-01 -8.36269140e-01 1.83163404e+00
5.79137266e-01 4.52606618e-01 -1.23557068e-01 7.32907772e-01
6.04991019e-01 5.66562355e-01 -9.49732438e-02 -6.93252325e-01
2.00377989e+00 1.12211823e-01 -4.78125751e-01 -1.41409067e-02
4.03473616e-01 -8.70280027e-01 7.91109622e-01 4.47196156e-01
-5.14795601e-01 -1.08501771e-02 -6.83923662e-01 7.46199787e-02
-7.09223926e-01 2.44928867e-01 3.42683911e-01 4.80774105e-01
-9.28594887e-01 3.81077558e-01 -8.19298863e-01 -9.05499995e-01
3.02015513e-01 3.54338050e-01 -4.18736458e-01 6.05323017e-02
-8.59295189e-01 2.69392312e-01 1.77905485e-01 1.57519728e-01
-9.17346001e-01 -1.03619254e+00 -5.49355030e-01 3.90525982e-02
6.29307479e-02 -9.32922125e-01 5.04312992e-01 -3.95637378e-02
-1.60039437e+00 1.14077258e+00 -4.94256407e-01 7.12271705e-02
1.23831265e-01 4.04776663e-01 -1.96710363e-01 2.50602979e-02
-2.94628367e-02 4.58074838e-01 2.89243609e-01 -8.77812207e-01
-8.52862075e-02 -5.23240924e-01 -7.33952403e-01 -1.80437744e-01
6.49420559e-01 1.56681955e-01 -3.39887589e-01 -4.53386426e-01
1.60457015e-01 -1.12975788e+00 -1.71103254e-01 -2.77756125e-01
-5.81523955e-01 3.81882071e-01 1.07371569e+00 -2.46124402e-01
7.15570688e-01 -2.24354649e+00 4.85055983e-01 3.66079152e-01
2.87728962e-02 -1.52805805e-01 -1.26940580e-02 7.59422362e-01
-1.38135359e-01 2.78345406e-01 -4.58824277e-01 5.99796772e-02
-4.92141619e-02 4.14936505e-02 1.15678780e-01 1.04302049e+00
2.16089875e-01 9.33725238e-01 -1.03108442e+00 -1.58318937e-01
-4.30620536e-02 7.15018332e-01 1.33089852e-02 -3.83376479e-01
-3.45626593e-01 1.12652600e+00 1.58226430e-01 7.64250875e-01
5.32314897e-01 -3.84716690e-01 7.12848604e-01 -4.63685989e-02
-4.39272761e-01 -1.80911403e-02 -1.20350754e+00 1.69023073e+00
-1.28504217e-01 9.31388199e-01 6.91108048e-01 -7.22232997e-01
7.67236233e-01 1.92411065e-01 5.37427545e-01 -2.79417425e-01
2.90355156e-03 -3.29208151e-02 -5.07229613e-03 -1.61972016e-01
-7.89848343e-02 -2.42466107e-01 -1.60869479e-01 3.67516339e-01
-1.83254592e-02 -1.34964421e-01 5.44962287e-01 9.29658208e-03
1.36504281e+00 1.24505199e-01 3.58255208e-01 -9.50152278e-01
3.31445485e-01 -1.99031323e-01 8.74056756e-01 3.51798832e-01
-2.38093644e-01 7.59382844e-01 1.09155869e+00 -1.50555208e-01
-1.04210114e+00 -9.40453112e-01 -1.41356602e-01 1.06143987e+00
-1.04134917e-01 -3.54835123e-01 -2.94713169e-01 1.52044922e-01
5.37749976e-02 8.00133348e-02 -6.04165196e-01 1.77603438e-01
-1.56364426e-01 -1.60449672e+00 9.84313548e-01 -1.92142606e-01
-1.03502527e-01 -3.58677328e-01 -6.69335544e-01 2.59019285e-01
-1.38866110e-02 -9.47815955e-01 -5.98773539e-01 6.39557779e-01
-4.24260378e-01 -1.45668316e+00 -4.76235986e-01 -3.29959035e-01
7.01672673e-01 1.07808620e-01 8.56257081e-01 -3.81185472e-01
-8.58915150e-01 4.73639131e-01 2.83357233e-01 -4.59467232e-01
-1.91099077e-01 -3.53163034e-01 2.91943640e-01 -1.06529444e-01
3.22550833e-02 -9.74989712e-01 -7.14093983e-01 5.82299292e-01
-9.60744262e-01 -2.29929555e-02 2.08587587e-01 1.02437127e+00
8.68404150e-01 -1.07398391e-01 5.79861522e-01 -8.21945667e-01
5.10973275e-01 -6.82041407e-01 -1.04727852e+00 1.03059903e-01
-2.82691449e-01 1.36683285e-01 6.41555667e-01 -1.54651344e-01
-8.45717311e-01 5.80275238e-01 3.93631786e-01 2.42972225e-01
-4.22514737e-01 7.49006748e-01 -2.82017559e-01 1.52284741e-01
5.41498899e-01 5.20955503e-01 3.21945757e-01 -9.61763039e-02
9.06644464e-02 1.12552240e-01 3.68361801e-01 -5.66579282e-01
2.29333848e-01 1.05295289e+00 8.74713838e-01 -1.31737411e+00
-8.84414166e-02 -9.33839440e-01 -6.81642056e-01 -2.29595885e-01
7.35013366e-01 -8.38168681e-01 -1.26341474e+00 4.32889938e-01
-9.22832131e-01 -6.32115781e-01 1.33877009e-01 3.14736754e-01
-4.95762438e-01 3.15733373e-01 -5.35343170e-01 -7.74149418e-01
-1.26681980e-02 -9.04996395e-01 1.53065741e+00 1.31326422e-01
-3.59784514e-01 -9.78802919e-01 8.91606331e-01 -1.99247122e-01
9.54497233e-02 6.10537350e-01 8.20050478e-01 -4.55111116e-01
-4.56501067e-01 1.78007409e-01 9.54945385e-02 -7.36137629e-01
2.06536815e-01 7.27873445e-01 -1.11018586e+00 -5.81449680e-02
-7.43594170e-01 1.47066757e-01 9.95911717e-01 5.01378298e-01
5.27905345e-01 4.32636327e-04 -8.08233559e-01 5.52275181e-01
1.41081440e+00 2.03849953e-02 6.77726805e-01 -3.32600288e-02
2.15634495e-01 5.41008949e-01 1.33195564e-01 5.19751489e-01
-2.58938640e-01 4.83873487e-01 1.51414305e-01 4.54099812e-02
6.30677864e-02 2.11458653e-01 5.30437827e-01 4.93433088e-01
1.22743197e-01 -1.54986709e-01 -1.08080971e+00 6.08553410e-01
-1.69794202e+00 -1.14850855e+00 -5.26044607e-01 2.23912048e+00
9.17849958e-01 -2.71067739e-01 2.25022256e-01 2.13929843e-02
9.50535178e-01 -1.48716375e-01 -2.90429860e-01 -1.12642102e-01
-7.25659251e-01 4.67887102e-03 4.92802411e-01 5.01505852e-01
-8.02273035e-01 4.57127929e-01 6.93587971e+00 6.70380414e-01
-1.20887852e+00 -2.38298506e-01 7.17428327e-01 -2.52246052e-01
-3.26543540e-01 3.26447636e-01 -7.50714719e-01 6.93145871e-01
9.52028751e-01 -3.19889098e-01 4.95151460e-01 -3.18391956e-02
1.04287970e+00 -4.82923388e-01 -1.11504793e+00 8.19345355e-01
-6.74221635e-01 -1.75615311e+00 -7.29611039e-01 3.10531408e-01
5.65998375e-01 2.87632227e-01 -4.03261155e-01 -4.03060645e-01
4.73324060e-01 -1.17670214e+00 -5.43098412e-02 9.45940733e-01
7.50754178e-01 -3.59758228e-01 6.23008907e-01 3.66499335e-01
-1.13192570e+00 2.47771829e-01 -5.29149286e-02 2.31207833e-02
2.01408397e-02 1.24411011e+00 -1.28735995e+00 3.88119787e-01
5.23601711e-01 4.97112840e-01 -1.45108774e-01 1.16868711e+00
2.32450321e-01 8.70790839e-01 -9.51454461e-01 -1.67779252e-01
-2.23613858e-01 -4.69597310e-01 8.46365631e-01 1.78448749e+00
5.78671575e-01 1.29737094e-01 6.58903196e-02 7.58011758e-01
1.37505651e-01 -3.52846719e-02 -7.66396403e-01 -2.84542918e-01
6.77104175e-01 1.52252364e+00 -1.49326026e+00 -1.38362423e-02
1.40584540e-02 5.46614766e-01 -1.89236045e-01 5.90480387e-01
-3.22618276e-01 -1.90805599e-01 9.66126561e-01 1.25450194e-01
2.68426508e-01 -5.21356881e-01 -2.50871062e-01 -8.72160733e-01
-6.20137453e-01 -4.37386006e-01 4.84278232e-01 -5.96754491e-01
-9.04540598e-01 -3.43483001e-01 -2.05880508e-01 -9.18577433e-01
1.97844297e-01 -6.80511475e-01 -6.66393101e-01 7.11925030e-01
-1.01305926e+00 -7.94411421e-01 -7.45667368e-02 4.63096239e-02
1.13242015e-01 2.01884598e-01 8.27255011e-01 6.16265312e-02
-6.91053271e-01 -3.90100300e-01 6.20846093e-01 -2.30455115e-01
8.17607522e-01 -1.23856556e+00 -1.97787866e-01 7.36503303e-01
-4.83279675e-02 8.87840986e-01 1.05623746e+00 -6.28288150e-01
-1.60873067e+00 -1.05374074e+00 5.89087665e-01 -2.99511969e-01
6.84067249e-01 -8.04855585e-01 -3.43803227e-01 1.78866550e-01
1.28303200e-01 4.75899056e-02 1.23644006e+00 5.02781421e-02
-9.97855514e-02 -1.10363238e-01 -9.68447506e-01 8.12001586e-01
8.03144276e-01 -5.31411707e-01 -9.97845754e-02 3.28713477e-01
2.02150717e-01 -3.29778790e-01 -9.35420036e-01 4.73193154e-02
8.27625096e-01 -8.13405693e-01 6.59887373e-01 -4.40876901e-01
2.64863707e-02 -1.05133975e+00 5.07977977e-02 -1.43413353e+00
-4.42412615e-01 -1.20509160e+00 3.66065562e-01 1.10360754e+00
4.80888337e-01 -5.35364747e-01 5.14770031e-01 -3.93729880e-02
-2.07053781e-01 -7.67890140e-02 -1.48799634e+00 -4.12881076e-01
-4.09502715e-01 -2.44241253e-01 2.71830320e-01 1.10227513e+00
5.83099008e-01 5.83459854e-01 -2.78184470e-02 8.09374154e-02
2.32154042e-01 1.93482433e-02 9.04127061e-01 -1.53609109e+00
-2.28752270e-01 -4.70055431e-01 -4.35072809e-01 -4.49829936e-01
-1.69284716e-01 -7.27004170e-01 2.81057864e-01 -1.25751770e+00
2.09143728e-01 4.34220582e-02 -2.10291713e-01 2.74912059e-01
1.94993809e-01 1.67706594e-01 -3.10116291e-01 -1.69607028e-01
-4.21528757e-01 1.52440161e-01 1.23546886e+00 -1.63086802e-01
-2.15501219e-01 -2.89841652e-01 -6.23445570e-01 3.22516531e-01
5.84460258e-01 -4.65207219e-01 -1.32104561e-01 3.34206283e-01
8.80424678e-01 -6.88386187e-02 4.28254485e-01 -7.66049862e-01
4.11670178e-01 -2.90401727e-01 4.90606487e-01 -5.21293163e-01
2.02379465e-01 -7.60017335e-01 7.89607942e-01 5.87769330e-01
-2.06250444e-01 1.14329550e-02 6.26586378e-01 1.06500697e+00
7.02377409e-02 3.55317861e-01 8.07930231e-01 -3.93527836e-01
-2.36002341e-01 -1.43380821e-01 -1.26141977e+00 -2.01315098e-02
9.16380823e-01 -2.89170325e-01 -6.79725111e-01 -1.02332920e-01
-9.84129846e-01 3.26915264e-01 5.92391491e-01 -2.03058943e-01
2.50045836e-01 -1.07547748e+00 -6.90105975e-01 1.66580789e-02
1.59079105e-01 -3.54175270e-02 1.40129387e-01 1.44833624e+00
-3.86691481e-01 5.32846808e-01 -1.75754115e-01 -1.12035310e+00
-1.60794818e+00 3.42772186e-01 1.82261556e-01 -1.25911027e-01
3.53389396e-03 6.71580851e-01 1.53953686e-01 -4.01641637e-01
-5.12813926e-01 -4.87477094e-01 -9.42612365e-02 4.71407145e-01
2.65298247e-01 3.21814477e-01 -4.37729843e-02 -4.74001408e-01
-6.09391093e-01 5.44611096e-01 6.36991143e-01 -9.83869135e-02
1.31393957e+00 -3.78434420e-01 -7.92836666e-01 9.95044410e-01
8.73198628e-01 5.83869576e-01 -1.04675794e+00 4.25070971e-01
1.81815550e-01 -1.47392586e-01 -3.09131116e-01 -7.69890428e-01
-4.54044968e-01 3.16491693e-01 3.30762744e-01 3.01251411e-01
1.11140919e+00 9.88626946e-03 -8.72206241e-02 2.71933854e-01
2.44584039e-01 -1.12968743e+00 -4.34481829e-01 5.95673680e-01
5.75200021e-01 -6.31103277e-01 9.69989970e-02 -5.23957014e-01
7.31891543e-02 1.37458265e+00 -3.81398760e-02 2.99066901e-01
7.15333879e-01 7.20355153e-01 -1.36121094e-01 -1.93672150e-01
-1.37446523e+00 -1.72708109e-01 -5.06704040e-02 5.90014875e-01
7.92195976e-01 -4.02031578e-02 -5.40871739e-01 3.70926112e-01
2.27217495e-01 1.47538677e-01 7.76618302e-01 7.32159436e-01
-3.05858761e-01 -9.06619370e-01 -5.91164231e-01 2.22828463e-01
-5.99053085e-01 -6.66475147e-02 -6.23851955e-01 5.48691571e-01
9.97257307e-02 6.17971122e-01 9.21893939e-02 -8.77868459e-02
-1.30007789e-01 5.74064612e-01 3.06957394e-01 -4.90270764e-01
-1.11892492e-01 8.31644058e-01 2.75977969e-01 -4.27722454e-01
-5.41126192e-01 -1.16308248e+00 -1.38798356e+00 8.49869549e-02
8.77175480e-02 2.99222946e-01 5.41380405e-01 9.79863584e-01
9.91633773e-01 4.90623981e-01 1.30819827e-01 -6.48185015e-01
6.55503213e-01 -7.60069072e-01 -5.84436715e-01 -6.33378923e-02
5.20785689e-01 -4.44256634e-01 -3.27873111e-01 6.32826090e-01] | [6.569747447967529, 5.156620025634766] |
9a58a0b7-06c5-439d-a3a8-0728eccea907 | multi-center-anatomical-segmentation-with | 2211.07395 | null | https://arxiv.org/abs/2211.07395v1 | https://arxiv.org/pdf/2211.07395v1.pdf | Multi-center anatomical segmentation with heterogeneous labels via landmark-based models | Learning anatomical segmentation from heterogeneous labels in multi-center datasets is a common situation encountered in clinical scenarios, where certain anatomical structures are only annotated in images coming from particular medical centers, but not in the full database. Here we first show how state-of-the-art pixel-level segmentation models fail in naively learning this task due to domain memorization issues and conflicting labels. We then propose to adopt HybridGNet, a landmark-based segmentation model which learns the available anatomical structures using graph-based representations. By analyzing the latent space learned by both models, we show that HybridGNet naturally learns more domain-invariant feature representations, and provide empirical evidence in the context of chest X-ray multiclass segmentation. We hope these insights will shed light on the training of deep learning models with heterogeneous labels from public and multi-center datasets. | ['Enzo Ferrante', 'Diego H. Milone', 'Maria Vakalopoulou', 'Nicolás Gaggion'] | 2022-11-14 | null | null | null | null | ['landmark-based-segmentation'] | ['computer-vision'] | [ 4.83180076e-01 6.20532632e-01 -5.61912298e-01 -4.53766435e-01
-1.26605749e+00 -8.30072701e-01 2.87251890e-01 5.64809680e-01
-3.23237598e-01 6.29056334e-01 4.80105340e-01 -3.28429818e-01
-3.60322684e-01 -7.29456961e-01 -6.17192864e-01 -7.07818389e-01
-1.16941765e-01 8.77745688e-01 2.96940237e-01 1.96185425e-01
-2.33426407e-01 4.61503059e-01 -7.22044230e-01 5.30641675e-01
6.07909024e-01 7.18822718e-01 4.65179645e-02 5.22371769e-01
-6.72715977e-02 8.03862274e-01 -2.30602592e-01 -1.82007968e-01
2.44919062e-01 -4.07011420e-01 -1.29151762e+00 2.86125481e-01
7.16338038e-01 -1.50865898e-01 -3.27594787e-01 9.29966867e-01
3.68845999e-01 -4.94634360e-03 9.97488618e-01 -5.98890066e-01
-5.72593272e-01 5.79010189e-01 -5.17351270e-01 3.94260705e-01
4.13832366e-02 1.62575439e-01 1.09505653e+00 -1.64112344e-01
1.15827978e+00 9.05353308e-01 8.08716774e-01 5.88701725e-01
-1.58642006e+00 -1.59184918e-01 2.61661738e-01 -2.57083058e-01
-1.08940709e+00 5.62706590e-03 6.83074594e-01 -6.98364019e-01
6.30026460e-01 2.37485290e-01 6.00603223e-01 1.18232346e+00
3.48300070e-01 8.23116720e-01 1.22602034e+00 -3.15325677e-01
4.73931953e-02 -3.10409397e-01 4.05312657e-01 1.19531250e+00
2.75046021e-01 -1.52325004e-01 -2.27404699e-01 -3.49655032e-01
9.96801376e-01 2.76860088e-01 -2.23075598e-01 -9.47618723e-01
-1.46354437e+00 9.22860384e-01 9.98543978e-01 5.72588801e-01
-4.94997174e-01 2.26479113e-01 4.54595327e-01 -9.90092382e-02
5.07741749e-01 4.88955230e-01 -3.87919813e-01 4.51044321e-01
-1.23023033e+00 -1.78526074e-01 5.39998174e-01 7.25639403e-01
6.87735200e-01 -4.47976321e-01 -3.22876126e-01 6.39433861e-01
3.14182162e-01 -1.45203276e-02 6.83233261e-01 -9.23436761e-01
2.61134952e-01 7.40828156e-01 -4.94165868e-01 -4.44980502e-01
-1.01287806e+00 -6.02586091e-01 -7.97238111e-01 -2.12334767e-01
6.36905849e-01 -1.54478759e-01 -1.38645208e+00 1.68118167e+00
1.66305810e-01 2.47017220e-01 -9.82344076e-02 7.00039625e-01
9.54077840e-01 -9.69068334e-03 5.22569537e-01 -1.88404229e-02
1.24682140e+00 -1.10312188e+00 -3.82587016e-01 -2.56781012e-01
1.14486945e+00 -3.64440203e-01 8.71950030e-01 5.95320165e-02
-9.17580605e-01 -3.57273668e-01 -7.87533402e-01 -6.36760220e-02
-2.72159636e-01 4.73635867e-02 7.98818827e-01 5.99573135e-01
-1.17833865e+00 5.78470588e-01 -1.16648424e+00 -4.12391484e-01
1.00008571e+00 4.33790147e-01 -4.65450257e-01 -3.60315293e-01
-7.16462135e-01 7.42068172e-01 1.94473535e-01 -1.18914634e-01
-1.09633195e+00 -8.99543643e-01 -8.85919571e-01 -1.66137129e-01
5.04642844e-01 -1.00174773e+00 1.04958284e+00 -8.37907553e-01
-8.82461965e-01 1.47368705e+00 -9.02575105e-02 -3.27376753e-01
5.89619875e-01 2.67527908e-01 -1.04917683e-01 5.40008545e-01
4.25423682e-01 7.78030634e-01 5.25308430e-01 -1.29994583e+00
-1.91561043e-01 -7.36407995e-01 -8.12222213e-02 1.20144449e-01
1.60638347e-01 -5.46915352e-01 -3.52834970e-01 -6.83378577e-01
6.55996978e-01 -1.22668028e+00 -9.00027692e-01 2.16186166e-01
-5.89320123e-01 -7.55522028e-02 3.98026139e-01 -5.94790637e-01
6.48556471e-01 -2.10260630e+00 2.75083601e-01 4.26417112e-01
5.67949474e-01 -1.83065325e-01 1.34031018e-02 -4.48302031e-02
-2.12248236e-01 1.62165418e-01 -4.90570247e-01 -2.23587483e-01
-3.16878915e-01 4.75558370e-01 6.61971644e-02 6.62474751e-01
-1.23742901e-01 1.24986458e+00 -1.06817663e+00 -9.84656811e-01
2.31259823e-01 1.62955746e-01 -5.78300774e-01 -4.57921661e-02
-8.99479091e-02 1.06290257e+00 -7.84539640e-01 6.62981510e-01
2.93379605e-01 -8.77288640e-01 5.54198980e-01 -2.64306307e-01
3.96641433e-01 1.80725694e-01 -6.75201178e-01 2.51462603e+00
-4.48175758e-01 2.59451926e-01 2.50517745e-02 -1.21577477e+00
3.09553504e-01 4.26117212e-01 1.08229506e+00 -5.00175595e-01
3.64227733e-03 1.51359528e-01 -1.05977453e-01 -3.49329978e-01
-6.22600317e-02 -3.74896407e-01 -1.87995002e-01 4.94020164e-01
4.04758692e-01 -3.73927593e-01 -2.21727919e-02 3.82475913e-01
1.25674224e+00 -4.33301330e-02 2.19235405e-01 -3.79853666e-01
1.83814719e-01 2.80166477e-01 3.83497089e-01 1.07982028e+00
-4.31658298e-01 8.69381011e-01 5.56145132e-01 -4.30970073e-01
-6.09033585e-01 -1.44399977e+00 -4.87506479e-01 1.06824744e+00
3.15877143e-03 -2.92113721e-01 -5.54735065e-01 -1.36165881e+00
-1.20626524e-01 2.40688831e-01 -1.04689252e+00 -1.44374475e-01
-5.50344884e-01 -7.94918835e-01 3.44648302e-01 6.05384767e-01
1.00695394e-01 -9.44197834e-01 -6.59131348e-01 2.03163862e-01
-9.83086526e-02 -9.79182005e-01 -2.13629469e-01 4.96985137e-01
-1.35402536e+00 -1.32899988e+00 -1.02509487e+00 -8.99099231e-01
1.08856595e+00 -1.31314427e-01 1.40623462e+00 3.20981383e-01
-7.12327361e-01 9.45241630e-01 -2.28046998e-01 -3.47238220e-02
-3.86107951e-01 5.27596295e-01 -5.49390674e-01 -2.74339497e-01
1.91125143e-02 -2.91285455e-01 -7.47395992e-01 4.01657596e-02
-9.53171909e-01 3.34210470e-02 6.09355450e-01 9.43153799e-01
9.61120248e-01 -3.57978821e-01 2.86217034e-01 -1.73124599e+00
3.35611552e-01 -5.69895744e-01 -3.46555442e-01 5.70274711e-01
-3.08760971e-01 1.21641837e-01 2.10444763e-01 -9.91516337e-02
-8.56133163e-01 4.97293830e-01 -2.14907736e-01 -1.24250069e-01
-4.07302976e-01 5.52385986e-01 4.42046463e-01 -2.90678948e-01
8.29310000e-01 -7.31632039e-02 -1.85529947e-01 -4.27125007e-01
5.47319889e-01 7.70097002e-02 5.55114985e-01 -7.97421813e-01
2.71219760e-01 8.37856293e-01 3.00328493e-01 -4.23422933e-01
-1.24865210e+00 -7.45394945e-01 -1.17764270e+00 4.97270115e-02
1.26819217e+00 -8.69935572e-01 -1.72731608e-01 -1.91584323e-02
-7.08201289e-01 -5.95984221e-01 -7.11341798e-01 4.09496993e-01
-8.58266711e-01 4.48441267e-01 -8.20089757e-01 1.71923377e-02
-9.13116112e-02 -1.54109132e+00 1.38246882e+00 4.89927493e-02
-2.73278177e-01 -1.48846304e+00 3.26531857e-01 4.18322206e-01
9.05286521e-02 3.62319469e-01 1.23818922e+00 -7.99829662e-01
-6.24146104e-01 -1.19951546e-01 -2.09080160e-01 -6.02422059e-02
3.65632713e-01 -4.29177880e-01 -8.78437042e-01 -3.48458856e-01
-1.91340402e-01 -4.16635424e-01 1.17965651e+00 9.48275030e-01
1.57172453e+00 1.49575278e-01 -8.44731987e-01 8.86200190e-01
1.40848148e+00 -3.65960598e-01 1.27756417e-01 -2.27175225e-02
9.87017155e-01 6.30251586e-01 1.28447846e-01 1.40802756e-01
3.29564095e-01 3.73098105e-01 4.46358919e-01 -6.55884326e-01
-3.93424958e-01 -1.52657181e-01 -4.25828457e-01 5.98601401e-01
1.37654513e-01 4.41548638e-02 -1.29653561e+00 7.19427884e-01
-1.70160198e+00 -3.94304782e-01 -2.22305004e-02 1.84026098e+00
8.82012427e-01 4.26738486e-02 -1.11413926e-01 -4.26387191e-01
4.21803296e-01 2.58839250e-01 -6.23313844e-01 4.26809536e-03
8.74870494e-02 5.04468083e-01 9.17585492e-01 3.11951488e-01
-1.44164443e+00 8.54251087e-01 7.40098619e+00 5.80460310e-01
-1.13138115e+00 5.62438250e-01 1.00746739e+00 1.32816806e-01
-4.33166772e-01 -9.66379698e-03 -3.32039684e-01 1.11427933e-01
7.41901398e-01 2.46921077e-01 -4.52260189e-02 7.42730021e-01
-4.15580094e-01 -9.96675342e-02 -1.35565031e+00 8.38269711e-01
3.52857262e-02 -1.70371175e+00 4.58121561e-02 2.36751318e-01
1.05153334e+00 3.68421674e-01 1.70823231e-01 2.92865992e-01
6.39464974e-01 -1.13206649e+00 1.11857586e-01 4.92329657e-01
8.15723181e-01 1.99460946e-02 5.43428004e-01 2.11193308e-01
-9.13069963e-01 6.33735433e-02 -1.35284394e-01 6.38059676e-01
2.27702767e-01 2.53036469e-01 -1.00524497e+00 7.25584984e-01
4.08901453e-01 8.60884130e-01 -8.56527925e-01 1.11151600e+00
-2.06924155e-01 7.45299518e-01 -1.96705982e-01 6.34795904e-01
5.38925707e-01 1.49977684e-01 1.57434225e-01 1.00684631e+00
-2.88509801e-02 1.65698510e-02 6.13177240e-01 8.04883122e-01
-2.20471948e-01 1.05763495e-01 -5.78520715e-01 1.62495971e-01
-2.75617689e-01 1.39829504e+00 -1.29936707e+00 -4.38758105e-01
-5.13727486e-01 8.92134368e-01 3.31874132e-01 3.92071992e-01
-5.60328543e-01 5.22685051e-01 -3.93943004e-02 2.87691802e-01
-2.18855008e-03 -2.71688193e-01 -4.50111955e-01 -1.17418170e+00
-5.49673975e-01 -4.49094266e-01 1.04299557e+00 -6.47598505e-01
-1.63562942e+00 5.11972606e-01 4.51149978e-02 -9.31259513e-01
-1.58686996e-01 -8.10378075e-01 -3.15635532e-01 5.54621935e-01
-1.57817423e+00 -1.44666851e+00 -2.98744589e-01 7.85485148e-01
5.43155491e-01 4.03685831e-02 1.02261233e+00 2.26866990e-01
-2.42611021e-01 4.01453882e-01 1.16634779e-01 4.19534951e-01
7.63488054e-01 -1.43864286e+00 3.43111344e-02 3.87444466e-01
5.65558255e-01 5.07467449e-01 8.58778805e-02 -6.94739819e-01
-8.40632737e-01 -9.64825094e-01 3.56385022e-01 -8.03101301e-01
6.07202351e-01 -2.53267009e-02 -7.88534224e-01 1.10844564e+00
1.37464805e-02 6.86994851e-01 1.06916296e+00 2.89943308e-01
-1.32641509e-01 3.59537929e-01 -1.19833267e+00 3.85102600e-01
1.00562501e+00 -5.90597570e-01 -7.43485928e-01 8.11961591e-01
5.68179131e-01 -6.05940163e-01 -1.10529327e+00 4.83695358e-01
1.22228600e-01 -5.67329228e-01 1.15485334e+00 -8.85394871e-01
1.77047402e-01 2.82148451e-01 -9.40843970e-02 -1.18285453e+00
-2.77356744e-01 -8.43310803e-02 3.74989480e-01 5.12479842e-01
4.32515115e-01 -5.15142798e-01 1.11496389e+00 5.62191248e-01
-3.29285204e-01 -7.37733424e-01 -1.14173543e+00 -3.47512126e-01
5.64561546e-01 -1.37191400e-01 2.06032604e-01 1.15645945e+00
-3.46726775e-01 7.24876449e-02 2.44523928e-01 3.84628139e-02
5.60465455e-01 3.45582515e-01 2.48173445e-01 -1.45942426e+00
-3.64674062e-01 -3.22960556e-01 -3.69050086e-01 -8.47683191e-01
3.60059947e-01 -1.37385368e+00 -4.78198007e-02 -1.94132614e+00
4.79349583e-01 -7.52403021e-01 -7.24105418e-01 6.95571840e-01
-1.32135479e-02 4.83631104e-01 -6.04110509e-02 3.25288147e-01
-9.03577030e-01 5.77953234e-02 1.63207507e+00 -4.42619026e-01
1.10324614e-01 1.09140188e-01 -5.75683177e-01 8.75595868e-01
4.50424463e-01 -8.10659051e-01 -3.92709017e-01 -4.20711040e-01
3.18529606e-02 2.62519717e-01 5.30300796e-01 -8.95887077e-01
1.42547846e-01 -1.15241911e-02 5.95496178e-01 -3.23324472e-01
1.72490343e-01 -9.18972611e-01 -2.54452884e-01 5.20120263e-01
-6.26147151e-01 -2.64086217e-01 1.10181399e-01 7.49577045e-01
-1.28970087e-01 -2.79415518e-01 7.53543139e-01 -7.87295401e-01
-6.67741835e-01 6.33366644e-01 -3.22585613e-01 3.50926787e-01
9.82753456e-01 -6.20014444e-02 -2.70551592e-01 -3.93301323e-02
-1.58314824e+00 1.12891123e-01 3.85604560e-01 1.05744347e-01
2.78326571e-01 -9.58412945e-01 -5.79318821e-01 2.27345094e-01
2.24987924e-01 2.41551027e-01 5.96763313e-01 1.07596803e+00
-6.35734797e-01 4.93155003e-01 -2.46589929e-01 -1.07877767e+00
-9.57471192e-01 5.35162210e-01 5.27030826e-01 -8.38515818e-01
-7.57583559e-01 9.74602401e-01 7.10428178e-01 -6.59056008e-01
1.75426509e-02 -3.45291018e-01 -4.44219671e-02 -1.05150148e-01
-3.62515002e-01 -2.22177222e-01 2.27664933e-01 -7.25101173e-01
-5.10977387e-01 7.68149674e-01 -2.46920332e-01 -5.45391887e-02
1.11252272e+00 7.59859309e-02 1.14784772e-02 3.57943088e-01
1.24909103e+00 -1.22418970e-01 -1.15566754e+00 -4.34570700e-01
2.10659966e-01 -2.10976109e-01 1.78623244e-01 -9.32052970e-01
-1.21535575e+00 1.01672351e+00 8.94735754e-01 -1.40124545e-01
9.56696153e-01 5.53822398e-01 5.96645534e-01 2.15809599e-01
4.03044134e-01 -8.92569482e-01 1.76933050e-01 7.63106868e-02
4.94002372e-01 -1.34909737e+00 1.84363097e-01 -5.39016306e-01
-7.05185354e-01 9.49004948e-01 4.60851520e-01 -2.26525113e-01
8.76908481e-01 -9.03596170e-03 2.09462419e-01 -7.01247334e-01
-3.75122547e-01 -3.33192199e-01 5.54061115e-01 5.53530335e-01
6.26670539e-01 3.14601749e-01 -7.20168799e-02 5.52007496e-01
2.09246337e-01 -1.71207368e-01 2.35210702e-01 1.06967497e+00
-1.44302770e-01 -1.36938798e+00 -1.86728202e-02 7.28618681e-01
-6.55834615e-01 -1.00410007e-01 -3.34534705e-01 7.06335664e-01
2.10920975e-01 3.73715997e-01 -8.98853969e-03 3.18436950e-01
1.60742730e-01 1.60830453e-01 7.45703697e-01 -1.18315542e+00
-7.08430946e-01 1.97563723e-01 -3.56262773e-01 -5.92635155e-01
-4.14756209e-01 -6.08626902e-01 -1.43649757e+00 5.61390996e-01
8.04511458e-03 -3.14000510e-02 4.11827385e-01 9.50129271e-01
1.60193533e-01 9.60427821e-01 1.00863598e-01 -6.44146144e-01
-2.48843506e-01 -5.62039733e-01 -5.92680275e-01 6.55571103e-01
3.49407166e-01 -6.92382395e-01 -3.63963954e-02 7.16181397e-02] | [14.743453025817871, -2.1816747188568115] |
5dead5f7-7faa-4a05-9549-119269c36ed1 | categorisation-of-bulgarian-legislative | null | null | https://aclanthology.org/2020.clib-1.6 | https://aclanthology.org/2020.clib-1.6.pdf | Categorisation of Bulgarian Legislative Documents | The paper presents the categorisation of Bulgarian MARCELL corpus in toplevel EuroVoc domains. The Bulgarian MARCELL corpus is part of a recently developed multilingual corpus representing the national legislation in seven European countries. We performed several experiments with JEX Indexer, with neural networks and with a basic method measuring the domain-specific terms in documents annotated in advance with IATE terms and EuroVoc descriptors (combined with grouping of a primary document and its satellites, term extraction and parsing of the titles of the documents). The evaluation shows slight overweight of the basic method, which makes it appropriate as the categorisation should be a module of a NLP Pipeline for Bulgarian that is continuously feeding and annotating the Bulgarian MARCELL corpus with newly issued legislative documents. | ['Svetla Koeva', 'Martin Yalamov', 'Nikola Obreshkov'] | null | null | null | null | clib-2020-9 | ['term-extraction'] | ['natural-language-processing'] | [-6.26926720e-02 2.72486538e-01 -2.60690749e-01 -3.48916322e-01
-7.71712542e-01 -9.65974987e-01 1.07621372e+00 7.56076813e-01
-1.01914454e+00 8.14869761e-01 7.25628734e-01 -8.95552158e-01
-6.16271436e-01 -5.50557375e-01 -4.16417569e-01 -3.02162081e-01
1.61903962e-01 9.62272227e-01 -1.34863853e-01 -6.04133129e-01
1.57776192e-01 2.81051815e-01 -1.06818295e+00 5.10896027e-01
7.23988116e-01 8.68837178e-01 2.84316838e-01 6.26004040e-01
-3.87617171e-01 6.61998272e-01 -9.73782003e-01 -5.84146142e-01
1.97991952e-01 -1.21956682e-02 -1.21932399e+00 -3.82518321e-01
7.61392653e-01 2.86590487e-01 -1.52066365e-01 1.03134429e+00
7.21127242e-02 -6.77821487e-02 7.22101331e-01 -5.48354328e-01
-3.25078309e-01 1.18921411e+00 1.04017280e-01 2.71691978e-01
2.07026988e-01 -5.13085127e-01 1.37833965e+00 -3.79558474e-01
1.27158725e+00 1.17985201e+00 6.48533285e-01 1.44694448e-01
-7.08379269e-01 -2.77506232e-01 -1.32551536e-01 2.31442943e-01
-1.18031144e+00 -2.93102026e-01 2.57253408e-01 -8.37640822e-01
1.31474829e+00 4.99764472e-01 4.16771561e-01 6.88511431e-01
2.55412132e-01 4.09583747e-01 8.70902121e-01 -1.01236439e+00
2.09336691e-02 4.19880182e-01 4.23573643e-01 1.93147719e-01
3.38550925e-01 -4.77305949e-02 2.61280984e-01 -2.54211366e-01
2.20542938e-01 -5.35647333e-01 8.76879096e-02 -2.58160941e-02
-8.78836513e-01 8.80015850e-01 -2.08501965e-01 9.62199688e-01
-5.53298831e-01 -1.29411981e-01 1.37939608e+00 4.41512674e-01
5.36689639e-01 6.99206829e-01 -9.54009235e-01 -1.23850293e-01
-1.19998610e+00 5.85369706e-01 1.06822908e+00 9.89574373e-01
2.44962275e-01 -1.89110249e-01 3.57953506e-03 1.02132046e+00
1.98276997e-01 2.08858773e-01 5.55518746e-01 -1.09861696e+00
8.18209648e-01 6.46800458e-01 -2.29021665e-02 -6.33062422e-01
-3.74236077e-01 -4.07083452e-01 -3.23386967e-01 -1.35911450e-01
4.58051413e-01 -2.81582057e-01 -9.52231348e-01 1.21078193e+00
2.46120378e-01 -1.01806509e+00 3.36522073e-01 3.77121836e-01
1.14749098e+00 6.68944538e-01 3.79728556e-01 -3.36084068e-01
1.65931845e+00 -6.31340265e-01 -9.51099455e-01 1.60306513e-01
8.15409601e-01 -1.08084595e+00 4.42125499e-01 4.26796198e-01
-9.61840332e-01 -4.70366448e-01 -9.03096080e-01 -4.16768134e-01
-1.10374272e+00 2.11348012e-01 4.88758504e-01 5.89326680e-01
-8.30708742e-01 5.62444925e-01 -1.58108309e-01 -5.71360588e-01
-2.73261040e-01 -1.50305089e-02 -3.25286716e-01 2.27632582e-01
-1.60975373e+00 1.06367052e+00 1.43300402e+00 -3.58285427e-01
-3.81828636e-01 -4.49027598e-01 -1.24976587e+00 2.39979546e-03
4.61382151e-01 7.64549971e-02 1.39118886e+00 -4.88208413e-01
-8.71209204e-01 1.23623180e+00 3.94080728e-01 -7.76287496e-01
3.36579114e-01 -2.48446818e-02 -8.10334861e-01 4.08555456e-02
4.90304470e-01 4.81568843e-01 1.34258822e-01 -8.33163619e-01
-1.34485161e+00 -3.08218628e-01 2.40392372e-01 8.65261331e-02
-7.89116994e-02 5.94281197e-01 -5.14774799e-01 -7.17840672e-01
-2.38636792e-01 -6.03786409e-01 3.59390043e-02 -1.03331840e+00
-1.64372653e-01 -7.58964300e-01 6.27690017e-01 -1.49028146e+00
1.54349184e+00 -1.81892192e+00 -1.91350579e-01 2.73550153e-01
-1.38354257e-01 4.24915761e-01 9.78265554e-02 8.11419010e-01
-2.88915187e-01 2.49979109e-01 9.19110626e-02 2.48457909e-01
3.92494380e-01 4.84752595e-01 -3.85615349e-01 4.63709950e-01
-5.68995357e-01 6.49659038e-01 -7.72641480e-01 -6.97185099e-01
2.33803034e-01 9.83887538e-02 5.49563253e-03 -3.95540416e-01
-4.61411089e-01 1.32827967e-01 -2.87220985e-01 4.68736738e-01
2.15325698e-01 6.70527399e-01 3.56962740e-01 -1.79534294e-02
-5.08124292e-01 8.11452329e-01 -8.28986228e-01 1.78449166e+00
-5.17106473e-01 7.44263291e-01 3.68669629e-01 -1.15219653e+00
7.33742952e-01 8.86309683e-01 5.57610810e-01 -8.86081398e-01
6.11300826e-01 3.79369617e-01 -6.02088422e-02 -5.56486726e-01
1.19488549e+00 1.04400374e-01 -5.63883066e-01 2.59389970e-02
5.24198711e-01 -3.52473676e-01 1.16582060e+00 1.78876668e-01
4.99505073e-01 2.32657090e-01 8.73946011e-01 -5.94726682e-01
9.00400698e-01 4.03878957e-01 3.63197386e-01 5.04777968e-01
3.37949172e-02 3.27073224e-02 5.56574523e-01 -6.44690096e-01
-1.64205265e+00 -1.29470736e-01 -6.35405540e-01 1.23403358e+00
-7.32405126e-01 -8.15683961e-01 -8.62127244e-01 -6.70946658e-01
-2.13552892e-01 1.22415698e+00 -5.43688893e-01 7.37739146e-01
-8.02537441e-01 -2.40669444e-01 9.74282205e-01 8.18951707e-03
3.41728270e-01 -1.12491560e+00 -5.05144954e-01 5.15643537e-01
-2.70435363e-01 -1.28403735e+00 -5.96506819e-02 4.70198840e-01
-5.10695815e-01 -1.14649260e+00 -6.05464756e-01 -8.46159279e-01
-1.85246527e-01 -5.71646631e-01 1.12872314e+00 -2.44706571e-01
-1.41121438e-02 2.15415284e-01 -4.48001087e-01 -8.02443802e-01
-1.08954298e+00 3.69003654e-01 -3.14706087e-01 -1.00369740e+00
6.32243752e-01 -1.48573607e-01 2.02795893e-01 -2.25260437e-01
-1.11630189e+00 -3.41353297e-01 1.98594734e-01 5.16448855e-01
3.41629028e-01 -9.14794207e-02 5.83202988e-02 -9.84787107e-01
9.56325769e-01 -2.49271229e-01 -1.05101502e+00 3.35513204e-01
-6.44993663e-01 -1.57345355e-01 6.08843327e-01 2.30781704e-01
-1.09696364e+00 -2.89158612e-01 -6.23047173e-01 2.32177734e-01
-6.43060744e-01 9.44899261e-01 -1.40624285e-01 4.32742536e-01
7.72307754e-01 -7.54283834e-03 -6.55281305e-01 -7.97883034e-01
7.80137122e-01 1.19597709e+00 9.84885931e-01 -9.67652678e-01
3.88597697e-01 1.35752279e-02 -2.34385088e-01 -7.39454985e-01
-5.46436846e-01 -1.09243071e+00 -8.04028928e-01 -2.24842399e-01
9.64522719e-01 -8.76113892e-01 -6.27792537e-01 1.04548456e-02
-1.60964608e+00 1.12970278e-01 -3.55730802e-01 4.71644521e-01
-2.97584862e-01 3.77017498e-01 -7.88884401e-01 -7.54928470e-01
-6.20213211e-01 -7.60303974e-01 8.73613656e-01 -7.66231194e-02
-5.76772571e-01 -1.23018289e+00 5.23894787e-01 5.00002861e-01
-1.99774038e-02 1.74832001e-01 1.26257515e+00 -1.33820415e+00
4.26883250e-01 -3.36858392e-01 -1.33877069e-01 6.02555573e-01
-3.78047913e-01 -1.30853951e-02 -6.81164265e-01 -1.12285331e-01
6.40138164e-02 -1.54867083e-01 7.50638247e-01 2.81164259e-01
6.16834342e-01 -5.34661233e-01 -1.21955946e-01 2.59204686e-01
1.67299676e+00 9.04503703e-01 4.42279190e-01 1.15696347e+00
1.62540406e-01 7.31304467e-01 4.58274454e-01 9.56000909e-02
3.39151472e-01 6.16825879e-01 1.65463328e-01 9.63347927e-02
4.13762070e-02 7.89192542e-02 1.39768198e-01 1.01920879e+00
-4.13260907e-01 -1.84514984e-01 -1.29951453e+00 1.01563954e+00
-1.68099201e+00 -9.52903986e-01 -1.90902308e-01 1.75142920e+00
9.57679689e-01 1.28218129e-01 -1.38296768e-01 1.00316219e-01
6.34703457e-01 2.94878542e-01 4.01572794e-01 -1.09249854e+00
-1.10495962e-01 5.37292659e-01 8.67466033e-01 6.04492009e-01
-1.50420392e+00 1.06460714e+00 6.24547863e+00 1.26635766e+00
-9.27764118e-01 3.29278946e-01 1.79283813e-01 1.92646965e-01
5.43438457e-02 2.22398657e-02 -1.11061025e+00 3.30172181e-01
1.39084685e+00 -2.25375399e-01 1.88445359e-01 1.05406630e+00
1.29670963e-01 4.64637913e-02 -7.69545078e-01 4.61181164e-01
9.74434763e-02 -1.56313133e+00 -2.13736251e-01 1.17115118e-01
6.29018962e-01 4.94972646e-01 -5.47091901e-01 8.65690708e-01
6.70515954e-01 -8.46406758e-01 1.08532131e+00 -2.85129007e-02
9.28020418e-01 -7.05511034e-01 1.11651444e+00 3.52744073e-01
-9.59591508e-01 3.25907655e-02 -5.06519198e-01 -6.49293698e-03
2.01078624e-01 2.58279681e-01 -9.00100589e-01 1.14472890e+00
6.38762534e-01 3.25648934e-01 -4.61189777e-01 8.79330277e-01
-3.81519079e-01 7.06348002e-01 -2.83605933e-01 3.53552625e-02
9.03097093e-01 -4.55389172e-01 8.17198396e-01 1.77405941e+00
1.69037387e-01 -2.86502481e-01 2.23151609e-01 2.22912550e-01
-4.01661638e-03 8.72683764e-01 -3.13365281e-01 -4.98236082e-02
2.46570379e-01 1.17251527e+00 -5.91575265e-01 -7.48394251e-01
-3.97911221e-01 3.82043868e-01 6.17501996e-02 2.70172000e-01
-4.74818975e-01 -7.69851506e-01 1.19887047e-01 -1.39028162e-01
3.20712000e-01 3.42606269e-02 -2.15893462e-01 -7.24158287e-01
1.23186052e-01 -1.05077875e+00 8.63023579e-01 -7.27358341e-01
-7.29443550e-01 8.35860550e-01 3.93860787e-01 -7.06970155e-01
-5.66063166e-01 -8.47979248e-01 -3.11656892e-01 7.94959664e-01
-1.08037293e+00 -1.21623802e+00 5.97387493e-01 1.76149502e-01
5.73796332e-01 -3.96882772e-01 1.00689399e+00 6.41134143e-01
2.54053506e-04 -2.54034966e-01 5.05359888e-01 6.03960812e-01
6.43222749e-01 -1.51558518e+00 2.96232074e-01 5.66911101e-01
3.08167785e-01 7.55310893e-01 9.89348769e-01 -8.75273228e-01
1.66748241e-02 -9.21540439e-01 1.84892166e+00 -1.30378008e-01
1.18720806e+00 -2.31088504e-01 -6.62752450e-01 7.57401586e-01
1.24746525e+00 -1.07681215e+00 4.97964710e-01 2.48810410e-01
-2.08120212e-01 4.51821536e-02 -7.96420813e-01 1.94982857e-01
5.19096017e-01 -5.37828326e-01 -1.34259713e+00 7.25426316e-01
5.99313557e-01 -7.18727410e-01 -1.11693251e+00 2.09674194e-01
5.03299296e-01 -2.84331769e-01 7.32872367e-01 -7.49481678e-01
3.32414180e-01 -4.75931838e-02 -1.30754247e-01 -1.17113662e+00
6.50049746e-02 -5.12935400e-01 6.66282237e-01 1.45440447e+00
5.91176748e-01 -3.50132406e-01 1.58058137e-01 1.91315070e-01
-2.22409546e-01 1.75898261e-02 -1.42982626e+00 -9.29477930e-01
6.10283613e-01 -7.75263011e-01 4.76711243e-01 1.17413521e+00
3.12203825e-01 4.16882724e-01 -9.32617337e-02 -4.71033990e-01
1.29768059e-01 -4.24740911e-01 3.97097796e-01 -1.53217554e+00
2.18871459e-01 -5.76768458e-01 -3.14619869e-01 -4.09309536e-01
3.47260118e-01 -1.14858758e+00 -7.73890689e-02 -1.72836924e+00
-2.49607205e-01 -7.31344596e-02 -1.75118446e-02 2.97696710e-01
5.49368978e-01 -2.33885154e-01 1.15861818e-01 2.62721419e-01
-4.00428206e-01 -3.96806793e-03 6.39091551e-01 -3.74323696e-01
3.42753390e-03 -2.09642962e-01 -2.96661228e-01 9.50588346e-01
4.97580111e-01 -7.20878601e-01 6.84591755e-02 -3.13708574e-01
4.66078490e-01 -3.91292013e-02 -8.07529762e-02 -8.55016172e-01
2.82266855e-01 -7.46997893e-02 9.83005539e-02 -1.08775055e+00
-1.51700974e-01 -9.83540714e-01 1.73992291e-01 3.00833046e-01
-4.77421075e-01 1.06055692e-01 3.06607634e-01 -1.20086111e-01
-6.13278687e-01 -9.35897946e-01 6.30989909e-01 -5.55014014e-01
-8.65170658e-01 -2.28710935e-01 -7.49602616e-01 2.91111302e-02
7.22737074e-01 8.18852931e-02 -3.24196041e-01 -1.18621089e-01
-9.08221483e-01 1.86314359e-01 2.33531699e-01 3.31271976e-01
-4.10930008e-01 -1.06169534e+00 -8.39941680e-01 -3.15568686e-01
1.86169103e-01 -3.85930598e-01 -1.14058219e-01 4.94449109e-01
-8.88619542e-01 1.56974649e+00 -6.88699856e-02 -1.48536682e-01
-1.51425886e+00 5.37865222e-01 1.98879838e-01 -1.09378350e+00
-3.23072314e-01 1.73206374e-01 -3.97872061e-01 -8.68059993e-01
3.92397434e-01 -4.30865079e-01 -1.21314859e+00 5.57813883e-01
2.22666055e-01 1.16152734e-01 4.88043636e-01 -1.16017199e+00
-2.58194685e-01 2.72141278e-01 -2.25546017e-01 -5.40297985e-01
1.36812806e+00 7.29080103e-03 -4.78864402e-01 2.00436994e-01
1.05037594e+00 2.14371487e-01 3.40505573e-03 -3.17205966e-01
6.54131591e-01 3.50134403e-01 1.80088937e-01 -1.04039717e+00
-5.69052398e-01 4.79756147e-01 3.65173459e-01 4.55989480e-01
7.80495405e-01 -2.15396918e-02 5.76336384e-01 6.53810263e-01
1.59097537e-01 -1.88347936e+00 -1.22754836e+00 1.14123154e+00
1.00468194e+00 -5.88632166e-01 1.92129970e-01 -1.00205727e-01
-5.48344135e-01 1.46207726e+00 -1.18304923e-01 1.54315144e-01
5.38763762e-01 1.45124733e-01 4.51606929e-01 -6.14489913e-01
-4.56263274e-01 -3.43684167e-01 5.30397475e-01 3.22464675e-01
5.84868908e-01 7.99639374e-02 -1.57172239e+00 6.70805871e-01
-5.89567840e-01 9.68972035e-03 3.49185109e-01 8.09879184e-01
-3.82072598e-01 -1.38884783e+00 -5.34613311e-01 2.27213278e-01
-1.24301171e+00 -5.18250525e-01 -5.66706777e-01 1.47284544e+00
6.26442492e-01 7.82067299e-01 1.14142552e-01 6.63601309e-02
3.92832696e-01 4.10429955e-01 -1.74504239e-02 -5.74942052e-01
-1.21617210e+00 4.83332336e-01 8.51730645e-01 -1.59688100e-01
-6.65156186e-01 -6.23201549e-01 -9.44044411e-01 -7.93907209e-04
-1.14040658e-01 9.44405496e-01 1.19979048e+00 1.37961936e+00
-2.86784559e-01 4.50568676e-01 -1.38699129e-01 -2.99647421e-01
-7.57751688e-02 -1.48759806e+00 -4.47001874e-01 4.22689199e-01
-4.70968932e-02 -1.39958963e-01 -2.72112098e-02 3.84616584e-01] | [10.337769508361816, 10.164125442504883] |
594fe800-7f50-4f98-a117-d103b74827ea | two-birds-one-stone-jointly-learning-binary | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Li_Two_Birds_One_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Li_Two_Birds_One_ICCV_2015_paper.pdf | Two Birds, One Stone: Jointly Learning Binary Code for Large-Scale Face Image Retrieval and Attributes Prediction | We address the challenging large-scale content-based face image retrieval problem, intended as searching images based on the presence of specific subject, given one face image of him/her. To this end, one natural demand is a supervised binary code learning method. While the learned codes might be discriminating, people often have a further expectation that whether some semantic message (e.g., visual attributes) can be read from the human-incomprehensible codes. For this purpose, we propose a novel binary code learning framework by jointly encoding identity discriminability and a number of facial attributes into unified binary code. In this way, the learned binary codes can be applied to not only fine-grained face image retrieval, but also facial attributes prediction, which is the very innovation of this work, just like killing two birds with one stone. To evaluate the effectiveness of the proposed method, extensive experiments are conducted on a new purified large-scale web celebrity database, named CFW 60K, with abundant manual identity and attributes annotation, and experimental results exhibit the superiority of our method over state-of-the-art. | ['Xilin Chen', 'Haomiao Liu', 'Yan Li', 'Ruiping Wang', 'Shiguang Shan', 'Huajie Jiang'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['face-image-retrieval'] | ['computer-vision'] | [ 1.58488885e-01 -3.31626743e-01 -2.82435238e-01 -8.23201835e-01
-4.32316929e-01 -5.92173874e-01 5.70450485e-01 8.91573951e-02
-1.54201344e-01 4.85337406e-01 -2.17487980e-02 6.02501370e-02
-2.47385129e-01 -6.85819387e-01 -4.38659191e-01 -7.19361246e-01
1.04779616e-01 4.43339795e-01 -3.63026172e-01 1.33223198e-02
3.02153051e-01 3.38558972e-01 -1.76483238e+00 2.25163549e-01
4.55336869e-01 1.62347937e+00 -3.78998257e-02 -3.03141493e-02
-9.89086404e-02 7.46821702e-01 -2.11681366e-01 -9.55075741e-01
1.06458209e-01 -4.10576075e-01 -7.52044559e-01 2.30896533e-01
6.79899931e-01 -2.80271202e-01 -3.84923965e-01 1.51767707e+00
2.03071222e-01 -2.00585380e-01 8.13011289e-01 -1.47428453e+00
-1.10695469e+00 7.52297118e-02 -5.71771145e-01 -1.70591310e-01
2.82812864e-01 -8.72658342e-02 1.10328043e+00 -9.18833792e-01
3.77574831e-01 1.51001894e+00 4.09644753e-01 6.43009841e-01
-1.12366807e+00 -1.15134144e+00 -9.16804075e-02 5.76672494e-01
-1.92800713e+00 -6.60647392e-01 8.57208133e-01 -5.97116947e-01
-2.14314293e-02 2.68688768e-01 3.34386170e-01 8.34167004e-01
-1.81868821e-01 5.58120072e-01 1.29070151e+00 -1.79497510e-01
3.27561721e-02 3.75166357e-01 -1.79131702e-01 9.71421182e-01
7.43826926e-02 -5.95819612e-04 -4.74591076e-01 -2.69303679e-01
5.07827282e-01 3.65091920e-01 -3.40798706e-01 -4.73043114e-01
-8.69426370e-01 8.79508853e-01 5.99992752e-01 1.16503127e-01
-9.51241329e-02 -6.46282407e-03 3.15850645e-01 2.53290087e-01
4.20615584e-01 -1.79191247e-01 -2.14599222e-01 1.39551774e-01
-9.29532230e-01 1.01379469e-01 6.10674560e-01 9.56704736e-01
1.08589458e+00 -4.85184401e-01 -1.30316302e-01 1.15355587e+00
4.77878898e-01 6.59045994e-01 6.87736630e-01 -6.50564075e-01
2.45657396e-02 6.57967210e-01 -1.08036481e-01 -1.41376078e+00
5.96801229e-02 -4.35696915e-02 -1.01074135e+00 -2.03031139e-03
2.39359900e-01 1.69961035e-01 -6.96608484e-01 1.82392657e+00
3.23551208e-01 2.79178917e-01 -2.20717527e-02 1.09733260e+00
9.12027121e-01 5.17645299e-01 1.40218258e-01 -2.86271006e-01
1.54364419e+00 -7.40749002e-01 -6.72549367e-01 -9.10810232e-02
2.53908336e-01 -8.62995565e-01 7.24592090e-01 2.15814728e-02
-6.29848719e-01 -6.75072730e-01 -7.88265944e-01 1.51300915e-02
-3.87899876e-01 4.92897362e-01 8.18507373e-01 6.72468662e-01
-7.54981935e-01 2.81980157e-01 -3.03537939e-02 -3.44285876e-01
7.50324190e-01 2.84937143e-01 -6.05976880e-01 -4.24886584e-01
-1.17685723e+00 4.56895530e-01 4.47153211e-01 7.68568292e-02
-9.88147676e-01 -3.07714105e-01 -9.97182250e-01 1.62771910e-01
4.21784043e-01 -1.38160437e-01 9.50663626e-01 -1.41006684e+00
-1.07054877e+00 1.48518586e+00 -2.27641091e-01 -1.23956531e-01
2.81415403e-01 1.18412204e-01 -7.14399397e-01 2.49030486e-01
2.69284964e-01 7.34729111e-01 1.05674839e+00 -1.25179327e+00
-5.83792627e-01 -6.53866887e-01 1.49093151e-01 -2.27982868e-02
-7.83060014e-01 1.45768747e-01 -4.10755605e-01 -7.09193528e-01
-5.78332180e-03 -9.65705276e-01 4.09064293e-01 6.33234859e-01
-9.94073674e-02 -6.31419718e-01 7.12472856e-01 -5.37307203e-01
9.94127095e-01 -2.35823250e+00 -3.25212106e-02 2.80592650e-01
8.09539631e-02 2.03129426e-01 -2.04402164e-01 1.02875195e-01
-1.93311721e-01 8.62622708e-02 -3.72601777e-01 -1.32506818e-01
-1.01367034e-01 1.71840906e-01 -3.36441308e-01 5.67479312e-01
3.38570893e-01 7.45518804e-01 -7.54987061e-01 -9.46603775e-01
-2.15215579e-01 3.40247869e-01 -4.17618960e-01 3.70527416e-01
2.00563087e-03 2.72471458e-01 -6.33891642e-01 9.69906569e-01
7.57100642e-01 -2.58276731e-01 7.82376155e-02 -3.00109386e-01
2.09236041e-01 -4.55041856e-01 -1.07809007e+00 1.69069326e+00
-4.45105433e-01 3.98372442e-01 8.44597295e-02 -1.08969653e+00
1.11257195e+00 1.74153551e-01 3.06382984e-01 -7.35836983e-01
1.37437940e-01 1.84388474e-01 -1.90436602e-01 -6.65958822e-01
1.14265352e-01 -1.44571096e-01 5.06861992e-02 3.97959143e-01
1.93318263e-01 2.45071441e-01 9.61014405e-02 1.21348895e-01
3.86921287e-01 -5.13267517e-02 2.34443635e-01 -4.33478743e-01
9.09419537e-01 -2.09416479e-01 6.32903159e-01 3.10808301e-01
-2.75200486e-01 5.16416788e-01 4.95432496e-01 -6.72152877e-01
-8.73634517e-01 -6.81393385e-01 -4.45251197e-01 1.21616626e+00
5.27098835e-01 -4.23653245e-01 -6.67702913e-01 -7.64699996e-01
2.79705137e-01 -1.91081520e-02 -8.76234651e-01 -1.39807999e-01
-7.07450509e-02 -4.20566231e-01 4.97326553e-01 1.89738154e-01
8.50498021e-01 -1.02726340e+00 -1.50681427e-02 -1.39380485e-01
-2.51595855e-01 -9.73954856e-01 -7.90977120e-01 -4.65371460e-01
-2.04529703e-01 -1.30973482e+00 -8.35468531e-01 -1.18838179e+00
8.17422569e-01 3.99948716e-01 1.00580359e+00 5.56489885e-01
-5.28553665e-01 2.80991852e-01 -4.15207654e-01 -1.83321252e-01
-8.30305368e-02 -2.72017568e-01 1.72814563e-01 9.36328053e-01
5.43551564e-01 -2.89005131e-01 -7.75158525e-01 5.07993400e-01
-8.81162643e-01 -8.10262114e-02 7.25467980e-01 1.05536115e+00
5.45795918e-01 1.22588202e-01 4.04704630e-01 -7.12302864e-01
5.05574405e-01 -5.10666370e-01 -3.79571050e-01 6.99872017e-01
-5.61068296e-01 6.93928152e-02 4.37068760e-01 -6.05391800e-01
-8.79852176e-01 2.46276304e-01 6.89316634e-03 -5.48992217e-01
-2.09234864e-01 3.55191022e-01 -4.90603060e-01 -3.96180719e-01
1.84362143e-01 6.04820967e-01 2.40807384e-01 -3.59352708e-01
3.67943764e-01 1.06268275e+00 7.06295729e-01 -8.14026892e-01
7.57248998e-01 4.52140361e-01 1.25687313e-03 -4.48985547e-01
-8.74184251e-01 -4.81142521e-01 -5.86896360e-01 -3.33173573e-01
9.13457572e-01 -9.88313794e-01 -1.08489799e+00 4.54049766e-01
-1.00773513e+00 3.10116887e-01 2.59209245e-01 1.87671155e-01
-4.46321070e-01 3.45167309e-01 -3.86022240e-01 -6.43033326e-01
-1.02153696e-01 -1.02978897e+00 1.17503130e+00 5.22154629e-01
5.77512085e-02 -6.71769381e-01 -2.39487588e-01 4.44556981e-01
2.30079874e-01 -4.27321866e-02 9.58210647e-01 -6.85142159e-01
-2.97494411e-01 -2.48094022e-01 -7.26912141e-01 2.31313452e-01
2.12747246e-01 -1.67073041e-01 -1.01436484e+00 -3.76159817e-01
-1.15333386e-01 -9.16389763e-01 8.15889359e-01 -3.16301495e-01
1.72646201e+00 -6.33414090e-01 -2.85930961e-01 6.54368222e-01
1.44976211e+00 1.84037909e-02 4.08534855e-01 -4.85959314e-02
5.65545082e-01 7.28212059e-01 7.59749234e-01 5.89614272e-01
3.67207378e-01 7.98856199e-01 5.64329326e-01 6.54093176e-02
-6.06323928e-02 -3.38814110e-01 -5.05202748e-02 4.74843264e-01
-6.50446815e-03 1.55164927e-01 -7.39862442e-01 3.87266904e-01
-1.62390244e+00 -1.05414569e+00 3.58042210e-01 2.02169681e+00
1.06474066e+00 -3.35125148e-01 -1.74499422e-01 -9.44512710e-02
1.16438580e+00 1.38667673e-01 -5.92300415e-01 1.09665126e-01
-3.36593240e-02 -7.28833526e-02 5.39118499e-02 3.62192355e-02
-1.32089984e+00 8.60622287e-01 5.01943636e+00 1.32124114e+00
-1.10931373e+00 5.71926944e-02 1.03175676e+00 2.54546791e-01
2.01580841e-02 -1.08172849e-01 -6.45723164e-01 8.97327781e-01
4.38748062e-01 -2.64843792e-01 5.46721220e-01 1.01428211e+00
-4.16351527e-01 3.07974994e-01 -1.11840177e+00 1.67854643e+00
5.30837834e-01 -1.01491594e+00 2.27931395e-01 1.59068033e-01
5.22071064e-01 -8.13099802e-01 2.32627392e-01 2.16356456e-01
-1.21017434e-02 -1.17541313e+00 8.17355990e-01 5.52828610e-01
1.24496341e+00 -6.34353220e-01 5.40118694e-01 1.56883791e-01
-1.40706575e+00 -2.52068222e-01 -6.25604570e-01 8.89663771e-02
-4.03492332e-01 1.97466552e-01 -1.58008859e-01 2.52013862e-01
6.95736647e-01 9.04035568e-01 -8.56342375e-01 9.97814059e-01
-1.25716954e-01 3.55043948e-01 4.96383272e-02 8.48554596e-02
5.07135577e-02 -1.44648135e-01 -1.54248357e-01 1.00870919e+00
4.13340211e-01 5.64841151e-01 2.08969727e-01 7.91871130e-01
-5.47725856e-01 3.65127921e-01 -5.36035061e-01 -4.65574674e-02
5.92440128e-01 1.64590538e+00 -4.29835618e-01 -1.87484354e-01
-5.94368160e-01 1.19878221e+00 3.07448745e-01 2.10893139e-01
-7.46818602e-01 -3.26697081e-01 7.81355917e-01 -3.38011570e-02
1.66222259e-01 1.21342994e-01 2.81425446e-01 -1.12747502e+00
1.39391452e-01 -8.51473927e-01 4.46493626e-01 -7.89044917e-01
-1.59076333e+00 6.38848066e-01 -3.82137030e-01 -1.33905399e+00
1.34450570e-01 -7.08480179e-01 -3.52691978e-01 7.61763513e-01
-1.62550449e+00 -1.52114773e+00 -6.60179973e-01 9.81815159e-01
3.68821114e-01 -5.49220383e-01 9.19846535e-01 5.34389973e-01
-3.88306350e-01 8.29174995e-01 2.77291596e-01 6.91618502e-01
1.06665277e+00 -7.25492537e-01 -1.90344483e-01 2.71391958e-01
4.69321519e-01 5.49601436e-01 2.05643877e-01 -4.19755727e-01
-1.30957520e+00 -1.17936444e+00 7.02559590e-01 -7.61150867e-02
6.13790035e-01 -3.82191956e-01 -8.22897553e-01 2.02670202e-01
-1.74029563e-02 5.14586270e-01 8.63289595e-01 -8.01269412e-02
-7.22628832e-01 -4.62459654e-01 -1.26224506e+00 1.83847949e-01
1.05055785e+00 -8.80282223e-01 -3.63518357e-01 3.37592840e-01
3.68946731e-01 -7.49132857e-02 -8.50574732e-01 4.16806102e-01
8.55422616e-01 -6.89904809e-01 1.04039121e+00 -6.64809287e-01
6.02526665e-01 -1.68892071e-01 -3.66767406e-01 -1.04139698e+00
-3.35471034e-01 -5.45439590e-03 3.01069945e-01 1.73008800e+00
-1.71966597e-01 -4.02699798e-01 5.76172769e-01 6.17101908e-01
4.80307430e-01 -6.03188217e-01 -9.54798341e-01 -6.68763041e-01
-3.69611531e-02 1.55283630e-01 8.27360749e-01 1.20071626e+00
-3.47903758e-01 1.74667090e-01 -6.28973365e-01 1.61270350e-01
8.24460745e-01 6.11705303e-01 4.01537836e-01 -1.60460508e+00
1.18532814e-02 -4.60993201e-01 -8.25598359e-01 -8.79204512e-01
6.11810148e-01 -1.01981294e+00 -5.93735948e-02 -8.23961794e-01
7.36009419e-01 -5.81266165e-01 -4.57581580e-01 7.41716385e-01
-1.63461611e-01 6.56882763e-01 2.67301977e-01 6.73505366e-01
-8.35259318e-01 7.16844678e-01 9.77205098e-01 -6.81164861e-01
6.24504209e-01 -2.53293067e-01 -8.24598014e-01 6.73918605e-01
5.90824485e-01 -5.13267159e-01 -2.72908986e-01 -2.90145457e-01
4.72979993e-02 1.92646801e-01 6.51232421e-01 -8.67530346e-01
3.70837450e-01 -3.21647644e-01 5.70481420e-01 -1.31077822e-02
1.79726407e-01 -1.04988956e+00 2.53626388e-02 2.63875872e-01
-4.47247922e-01 -1.61738023e-01 -1.06183678e-01 6.36893153e-01
-5.39634407e-01 -3.33186656e-01 7.95209646e-01 -8.95511135e-02
-1.09443760e+00 7.66525209e-01 3.12021345e-01 7.24119321e-02
1.15198481e+00 -5.52360117e-02 -2.62396246e-01 -3.28936994e-01
-5.83039939e-01 2.05127642e-01 4.52664793e-01 7.42452860e-01
7.44819582e-01 -1.83999729e+00 -8.34031522e-01 4.45215970e-01
7.06433177e-01 -5.94058573e-01 2.24483147e-01 1.88825473e-01
-2.22560659e-01 2.98768640e-01 -4.70240355e-01 -5.00485420e-01
-1.58729208e+00 8.49060595e-01 1.87763438e-01 4.06945050e-01
2.47863159e-02 8.35374832e-01 3.80343378e-01 -2.02284351e-01
1.50104553e-01 5.69466054e-01 -4.01257217e-01 2.97134876e-01
7.18021810e-01 -1.63170293e-01 -1.99733809e-01 -1.14465392e+00
-4.43477869e-01 9.86019015e-01 -1.19756684e-01 2.85162091e-01
1.00710309e+00 -3.00834060e-01 -5.62648892e-01 8.40353668e-02
1.55436575e+00 -2.27296829e-01 -1.12807763e+00 -6.29378319e-01
-1.35806665e-01 -9.31516171e-01 -1.71346024e-01 -5.55667102e-01
-1.44781816e+00 1.08501685e+00 1.10195017e+00 -7.93005805e-03
1.20927656e+00 2.73895502e-01 4.96963918e-01 3.80943567e-01
6.92804158e-01 -8.72773767e-01 2.15721473e-01 2.52410155e-02
8.71797979e-01 -1.68804300e+00 -1.63794547e-01 -4.06043410e-01
-6.43326104e-01 1.10086465e+00 6.27524018e-01 2.05780432e-01
8.41989696e-01 -3.33957344e-01 -4.95880768e-02 -1.34419650e-01
-3.85333300e-01 -3.14583093e-01 2.36731514e-01 4.14100379e-01
3.74194860e-01 1.15318559e-01 -3.14573139e-01 6.77647293e-01
-4.09405939e-02 -1.47808287e-02 -3.30216996e-02 5.43478072e-01
-3.93848121e-01 -1.08228421e+00 -3.76341254e-01 3.87622118e-01
-4.43873376e-01 -1.57148615e-01 -4.65925574e-01 3.60892713e-01
5.16649842e-01 9.17131007e-01 1.30207345e-01 -3.80140185e-01
-5.08647673e-02 1.71376094e-02 2.89734036e-01 -3.11967909e-01
-3.55003290e-02 -4.09215510e-01 -5.03535807e-01 -4.20451075e-01
-5.52811503e-01 -4.77329046e-01 -8.15908134e-01 -3.51297766e-01
-1.27574772e-01 2.67976373e-01 4.84016836e-01 6.78837478e-01
1.61222890e-01 -1.21781625e-01 8.98728132e-01 -4.58387762e-01
-5.32041788e-01 -6.90981984e-01 -7.71354079e-01 1.12812591e+00
2.40258276e-01 -8.98184538e-01 -7.89229944e-02 4.11276340e-01] | [13.353414535522461, 0.7183812260627747] |
273af3ab-3b7e-4eee-a724-913dbf1f9357 | should-i-run-offline-reinforcement-learning | null | null | https://openreview.net/forum?id=AP1MKT37rJ | https://openreview.net/pdf?id=AP1MKT37rJ | Should I Run Offline Reinforcement Learning or Behavioral Cloning? | Offline reinforcement learning (RL) algorithms can acquire effective policies by utilizing only previously collected experience, without any online interaction. While it is widely understood that offline RL is able to extract good policies even from highly suboptimal data, in practice offline RL is often used with data that resembles demonstrations. In this case, one can also use behavioral cloning (BC) algorithms, which mimic a subset of the dataset via supervised learning. It seems natural to ask: When should we prefer offline RL over BC? In this paper, our goal is to characterize environments and dataset compositions where offline RL leads to better performance than BC. In particular, we characterize the properties of environments that allow offline RL methods to perform better than BC methods even when only provided with expert data. Additionally, we show that policies trained on suboptimal data that is sufficiently noisy can attain better performance than even BC algorithms with expert data, especially on long-horizon problems. We validate our theoretical results via extensive experiments on both diagnostic and high-dimensional domains including robot manipulation, maze navigation and Atari games, when learning from a variety of data sources. We observe that modern offline RL methods trained on suboptimal, noisy data in sparse reward domains outperform cloning the expert data in several practical problems. | ['Sergey Levine', 'Anikait Singh', 'Joey Hong', 'Aviral Kumar'] | 2021-09-29 | null | null | null | iclr-2022-4 | ['robot-manipulation'] | ['robots'] | [-2.80037103e-03 1.05898477e-01 -3.61812085e-01 4.18649577e-02
-9.56294954e-01 -9.18866634e-01 2.75467873e-01 1.30903602e-01
-9.22075748e-01 1.33958888e+00 -2.60018706e-01 -3.65975082e-01
-2.98982054e-01 -5.25635064e-01 -1.09653246e+00 -8.63065779e-01
-5.56475639e-01 6.28526807e-01 -1.11053422e-01 -4.95812118e-01
1.37707293e-01 4.25077558e-01 -1.56134248e+00 -3.14056754e-01
1.10763490e+00 9.73686814e-01 5.31996250e-01 7.48917699e-01
3.36449116e-01 8.65636051e-01 -5.44220209e-01 1.05130553e-01
7.52095819e-01 -3.94131243e-01 -5.47317922e-01 1.69551343e-01
-2.21528202e-01 -7.84329653e-01 -3.65578830e-01 1.06857026e+00
4.78291005e-01 5.20315230e-01 3.79996061e-01 -1.21296084e+00
-4.80900742e-02 7.75077403e-01 -1.15829177e-01 1.09985890e-02
4.03969526e-01 6.26569867e-01 8.72668087e-01 -2.79572159e-01
6.62657201e-01 1.06341028e+00 3.26918542e-01 5.76008379e-01
-1.41927600e+00 -5.79286635e-01 4.18769777e-01 2.21383929e-01
-7.57631719e-01 -2.21092910e-01 5.26930094e-01 -1.83844939e-01
5.70743620e-01 -1.44857779e-01 7.28499174e-01 1.61056757e+00
-9.30705443e-02 1.32053995e+00 1.50311363e+00 -1.52558610e-01
8.02802324e-01 1.55320205e-02 -2.74314046e-01 5.92927516e-01
3.55387509e-01 7.85148799e-01 -3.46198648e-01 -1.95027009e-01
7.51474500e-01 -6.39271289e-02 -3.84628445e-01 -7.84188628e-01
-1.13607204e+00 8.46624434e-01 2.82165408e-01 5.25536872e-02
-6.41506076e-01 2.51826078e-01 2.59643078e-01 1.10085058e+00
-1.53981194e-01 1.04513502e+00 -6.34382010e-01 -6.62857473e-01
-3.03698570e-01 6.63102329e-01 1.08449650e+00 1.16663861e+00
7.29669154e-01 3.82524997e-01 1.10395618e-01 6.42103314e-01
-4.06175792e-01 6.97105646e-01 6.30692422e-01 -1.35291433e+00
6.65614724e-01 1.85401320e-01 9.41156805e-01 -5.38683712e-01
-5.91289878e-01 -4.44938064e-01 -4.78085786e-01 4.99062926e-01
8.09177160e-01 -5.86693168e-01 -6.02951646e-01 1.99404430e+00
2.18467757e-01 -4.89367172e-02 3.48328263e-01 1.10146296e+00
-1.00740321e-01 2.81286389e-01 -3.76615524e-01 -3.71847928e-01
4.33798730e-01 -8.23148549e-01 -5.44955373e-01 -2.88013071e-01
9.26910102e-01 3.44725251e-02 1.24238598e+00 9.78244185e-01
-9.01786625e-01 -3.28041315e-01 -9.92255807e-01 3.83161485e-01
-3.24597985e-01 6.84303045e-02 5.47536671e-01 4.02785569e-01
-9.19235528e-01 1.10778284e+00 -8.63089919e-01 -3.61490667e-01
3.16958159e-01 4.98699069e-01 -3.29425484e-01 -1.05947256e-01
-1.04524732e+00 9.28285360e-01 3.38233173e-01 1.43647626e-01
-1.78437316e+00 -2.39426151e-01 -5.96639931e-01 -1.94371641e-01
1.24462247e+00 -1.61519900e-01 1.76009297e+00 -1.17857385e+00
-1.91769409e+00 1.17453381e-01 2.40842193e-01 -9.12161350e-01
8.49657297e-01 -4.10561562e-01 1.50305033e-01 1.77873120e-01
-4.99543175e-02 4.57412064e-01 8.96746516e-01 -1.39633548e+00
-6.78341925e-01 -3.65531594e-01 6.56829536e-01 4.09162551e-01
-1.03613697e-01 -6.22939885e-01 1.66609049e-01 -1.77384898e-01
-2.23877102e-01 -1.14706659e+00 -6.97600901e-01 -8.58036503e-02
-1.69762209e-01 4.23023887e-02 4.37499642e-01 -5.40470898e-01
6.38117194e-01 -1.83428502e+00 5.46718478e-01 2.53265798e-01
7.47666955e-02 3.38025764e-02 -2.49967113e-01 5.41451514e-01
3.47819686e-01 -2.35359788e-01 -1.62987188e-01 -1.31871969e-01
2.31898963e-01 6.63611650e-01 -5.65035820e-01 5.43245018e-01
-3.06375265e-01 7.62342751e-01 -1.27142859e+00 -5.35433628e-02
-1.73415020e-02 -5.08925736e-01 -8.26802611e-01 3.76955867e-01
-6.46564484e-01 9.58014846e-01 -7.64148057e-01 6.52792871e-01
5.74295148e-02 1.33254528e-02 4.38582927e-01 5.91035962e-01
1.34358173e-02 -6.99886233e-02 -1.12218189e+00 1.73269093e+00
-6.31894469e-01 4.27963227e-01 5.18926024e-01 -1.52699709e+00
7.36673236e-01 2.11726323e-01 5.81341624e-01 -8.01854074e-01
2.86760449e-01 4.14668411e-01 1.79154813e-01 -7.10172236e-01
4.74374920e-01 -2.24396978e-02 -2.96543449e-01 4.41134810e-01
1.39887393e-01 -2.24600554e-01 1.24360286e-01 -7.58102387e-02
1.49325895e+00 3.70379895e-01 2.58071333e-01 8.09073746e-02
8.73661228e-03 3.04860175e-01 5.60114443e-01 1.46743715e+00
-4.21275407e-01 -1.24805523e-02 8.53181243e-01 -2.88658049e-02
-9.96974945e-01 -9.37369823e-01 3.06143641e-01 9.72766817e-01
3.41292113e-01 3.11247166e-02 -4.60699856e-01 -8.53100777e-01
2.76095927e-01 7.01039970e-01 -6.58738732e-01 -1.43891126e-01
-5.41721940e-01 -3.34178418e-01 4.39978778e-01 3.61693025e-01
5.42659581e-01 -1.09687412e+00 -9.45015490e-01 4.80356932e-01
-6.16934933e-02 -1.08042192e+00 1.42328888e-02 6.90268517e-01
-1.01545870e+00 -1.17093754e+00 -7.28946149e-01 -4.41701055e-01
5.43462515e-01 2.13157222e-01 6.85839295e-01 -1.44957408e-01
-4.44134139e-02 8.09763670e-01 -6.47304773e-01 -3.47353667e-01
-3.45161170e-01 2.77891159e-02 4.50389653e-01 -3.99473995e-01
-1.67459711e-01 -6.81467712e-01 -3.72060478e-01 2.66273826e-01
-6.59438431e-01 -2.51178771e-01 6.94792151e-01 1.27946186e+00
3.03063571e-01 1.43873245e-01 6.69184506e-01 -6.67124748e-01
7.08725870e-01 -6.95121944e-01 -9.85587716e-01 8.59169140e-02
-4.35230225e-01 4.30917174e-01 1.30453527e+00 -7.81409800e-01
-7.18370199e-01 1.02928087e-01 1.11103006e-01 -5.44402897e-01
-2.47482270e-01 3.34525734e-01 3.46342698e-02 3.96909527e-02
8.65177274e-01 4.17520136e-01 2.58186996e-01 -5.00531733e-01
2.82234341e-01 5.37776411e-01 2.07598105e-01 -1.33337247e+00
6.16418779e-01 3.60570759e-01 -2.86216419e-02 -7.09822714e-01
-5.89890063e-01 -1.63234353e-01 -3.46245110e-01 -3.74620378e-01
4.84816670e-01 -6.83156371e-01 -1.00956011e+00 3.24022740e-01
-5.54300308e-01 -1.21208358e+00 -5.19810319e-01 5.29072940e-01
-1.31642127e+00 1.33999407e-01 -3.64138186e-01 -1.26132011e+00
4.67609942e-01 -1.31410491e+00 7.48290002e-01 1.92908719e-01
3.09810787e-01 -7.61917531e-01 -3.76731157e-02 -4.06538062e-02
1.81455240e-01 1.74940273e-01 6.89030766e-01 -7.29487538e-01
-5.08706987e-01 1.58104777e-01 3.43746245e-01 2.68600106e-01
-6.72104359e-02 -5.98584950e-01 -6.49795055e-01 -6.37467146e-01
5.64716421e-02 -1.03508377e+00 5.67761600e-01 2.10977212e-01
1.41001749e+00 -4.00824547e-01 -1.27510414e-01 3.99892509e-01
1.39772701e+00 6.10381722e-01 2.02958196e-01 5.45520663e-01
1.08124137e-01 6.61554277e-01 1.08013928e+00 7.51520336e-01
1.75427437e-01 5.66572070e-01 6.95306003e-01 4.25377220e-01
6.93156898e-01 -6.73947990e-01 6.96805656e-01 2.59305924e-01
9.52250138e-03 -5.44029735e-02 -6.17051125e-01 4.14952874e-01
-2.22114921e+00 -8.96740377e-01 5.71255326e-01 2.39613962e+00
9.35044408e-01 7.16999471e-02 6.22693956e-01 2.75753830e-02
2.34199151e-01 -3.27290386e-01 -1.24950409e+00 -2.58240938e-01
-1.37289435e-01 1.30716622e-01 8.66637051e-01 3.84868175e-01
-7.72708237e-01 7.52879024e-01 5.68108940e+00 7.07656622e-01
-9.26474929e-01 8.86573941e-02 5.07866263e-01 -3.11617762e-01
9.25580636e-02 -2.88689189e-05 -5.20141244e-01 4.18904364e-01
8.35330844e-01 -9.41613093e-02 1.24200118e+00 1.29352748e+00
3.18998814e-01 -6.11866474e-01 -1.34118450e+00 9.79417324e-01
-3.84997904e-01 -8.35411251e-01 -7.22521245e-01 1.94202229e-01
8.55141222e-01 1.67397764e-02 1.33581057e-01 9.87560391e-01
1.03341222e+00 -9.43646789e-01 7.30464756e-01 3.82954866e-01
4.52972800e-01 -6.45470262e-01 5.15691936e-01 1.01935005e+00
-6.30890131e-01 -8.25510144e-01 -4.03111756e-01 -2.58689910e-01
-2.57682532e-01 6.30915463e-02 -8.94646525e-01 5.83232224e-01
4.86401320e-01 6.81662917e-01 -2.53545433e-01 1.18357921e+00
-2.29892477e-01 4.93503720e-01 -5.03687501e-01 -6.03319287e-01
5.91225684e-01 -4.52681839e-01 5.68867505e-01 3.54385465e-01
3.39223623e-01 2.36816272e-01 4.31848496e-01 5.90408444e-01
2.50717342e-01 -8.69776756e-02 -1.12813509e+00 -2.65214205e-01
4.22206521e-01 7.36424625e-01 -3.78419459e-01 -2.19003648e-01
-1.80526935e-02 8.95806730e-01 6.80062234e-01 6.18844569e-01
-7.33926237e-01 -1.31139651e-01 6.52646720e-01 -3.01415265e-01
3.61636698e-01 -6.21827722e-01 4.39210497e-02 -1.23359382e+00
-1.20917425e-01 -1.28135478e+00 2.78268307e-01 -5.74363172e-01
-1.06610680e+00 1.32065490e-01 6.18509464e-02 -1.42212927e+00
-4.48431462e-01 -6.60475552e-01 -4.04672734e-02 1.91457585e-01
-1.24201798e+00 -2.18592033e-01 7.87424948e-03 5.49865901e-01
6.63525045e-01 -1.52933136e-01 5.61819434e-01 -1.94887877e-01
-5.16414702e-01 3.15704703e-01 7.47759044e-01 -1.98470980e-01
4.41526383e-01 -1.48421991e+00 -3.35035056e-01 4.87248331e-01
3.99374254e-02 3.18719357e-01 9.85897839e-01 -4.65296328e-01
-1.91015863e+00 -7.22004771e-01 -3.98679584e-01 -1.16374552e-01
8.29491138e-01 -3.12698603e-01 -4.27946329e-01 6.30706906e-01
-2.17995271e-01 -1.98081896e-01 5.74694276e-02 2.45623253e-02
5.22368103e-02 -4.74257544e-02 -1.23543537e+00 1.04091418e+00
1.22710097e+00 -2.05346242e-01 -4.26355213e-01 3.47735822e-01
6.03183746e-01 -5.97353697e-01 -5.77953458e-01 8.06978643e-02
3.95583659e-01 -7.73141980e-01 6.88213825e-01 -9.31612790e-01
2.50359356e-01 -3.82182822e-02 -2.82369375e-01 -1.85392833e+00
2.84909368e-01 -9.47689176e-01 -2.23277166e-01 3.46321702e-01
3.39549363e-01 -9.13858354e-01 7.59772062e-01 3.79382670e-01
1.83592230e-01 -7.73070931e-01 -9.57633913e-01 -1.30952942e+00
3.85722846e-01 -5.25680602e-01 4.43769991e-01 4.96217877e-01
2.96266258e-01 -1.90394044e-01 -5.88602185e-01 -1.71133503e-02
6.96329951e-01 2.65568972e-01 1.16124249e+00 -7.85701513e-01
-7.25119352e-01 -2.83383161e-01 9.71545465e-03 -1.66287446e+00
5.26202857e-01 -5.64350843e-01 5.25285661e-01 -1.02116716e+00
-2.11365208e-01 -8.66656840e-01 1.88719090e-02 3.72287095e-01
1.67962044e-01 -5.44471383e-01 2.31232494e-01 3.08137704e-02
-8.31140578e-01 9.48556006e-01 1.58807003e+00 -4.88124900e-02
-4.16401893e-01 2.17565432e-01 -4.32578355e-01 4.94607717e-01
1.07111001e+00 -3.18774194e-01 -5.53180695e-01 -1.95849657e-01
2.08245769e-01 6.51769936e-01 2.67920047e-01 -9.66074347e-01
5.90032861e-02 -6.16313100e-01 6.03012964e-02 -1.61242753e-01
4.39607084e-01 -9.12915230e-01 -3.28654855e-01 6.60993338e-01
-6.62491441e-01 -1.13036133e-01 1.43695641e-02 1.09231782e+00
1.74430370e-01 -4.55764264e-01 4.66114372e-01 -4.30063277e-01
-6.00822568e-01 2.24814102e-01 -7.11683512e-01 3.39999110e-01
1.14783084e+00 -1.84018654e-03 -2.21877024e-01 -1.01165700e+00
-1.03443718e+00 6.67438805e-01 5.02883315e-01 1.10722899e-01
4.80843782e-01 -9.59566951e-01 -1.93355858e-01 8.24541897e-02
-9.62921008e-02 -2.83506829e-02 8.62875581e-03 1.06622624e+00
-1.77587688e-01 3.02213043e-01 -1.83783382e-01 -3.34354073e-01
-6.88886762e-01 6.99481964e-01 4.03361082e-01 -3.05670053e-01
-6.32749379e-01 4.28048939e-01 -1.90857276e-01 -6.15058064e-01
5.97164392e-01 -6.68466270e-01 3.46714817e-02 -1.94347337e-01
9.77997109e-02 4.63218719e-01 -2.11773276e-01 1.09893836e-01
2.20579609e-01 -5.46395890e-02 1.78003386e-01 -5.82614601e-01
1.33321166e+00 -5.03330566e-02 5.36575377e-01 6.41735137e-01
9.00156021e-01 -2.81682014e-01 -1.86459374e+00 -2.84980744e-01
1.04497470e-01 -5.57672083e-01 -2.80346990e-01 -8.29190433e-01
-7.28465259e-01 6.47997916e-01 3.89468938e-01 3.52098972e-01
9.97661650e-01 -3.10812175e-01 5.36731839e-01 1.18225741e+00
1.33576071e+00 -1.59818685e+00 3.90306532e-01 6.07227385e-01
9.33538616e-01 -1.41285920e+00 -3.25806528e-01 4.10775393e-01
-9.66512799e-01 1.08335888e+00 7.37454295e-01 -5.92402518e-01
3.60457391e-01 1.67787462e-01 -3.92715633e-01 3.21577162e-01
-1.07880318e+00 -6.87463701e-01 -6.13744080e-01 7.54133344e-01
-6.67082191e-01 1.60912305e-01 -5.28444238e-02 6.87496185e-01
-3.48293096e-01 4.71627004e-02 8.20277512e-01 1.28213191e+00
-6.87862813e-01 -1.16761994e+00 -4.36314076e-01 6.50810659e-01
-1.26022264e-01 4.34829623e-01 -3.27391267e-01 9.07672763e-01
-2.68894106e-01 1.05783343e+00 -3.07668865e-01 -3.72324973e-01
9.94883403e-02 -1.25535503e-01 8.87993217e-01 -4.48632330e-01
-4.54324007e-01 -1.51610717e-01 3.00767720e-01 -1.03387928e+00
-6.12116121e-02 -8.09959888e-01 -1.17697954e+00 -9.12481546e-02
-1.63842604e-01 2.44892910e-01 5.69878340e-01 1.07163942e+00
6.50508236e-03 2.27051035e-01 9.64372516e-01 -9.74482954e-01
-1.38266766e+00 -6.97662354e-01 -7.69200206e-01 2.80518290e-02
7.36791432e-01 -1.16227889e+00 -6.40706539e-01 -4.61200535e-01] | [4.186164379119873, 1.9033386707305908] |
599733ab-8899-4a58-b519-0fff54897721 | jepoo-highly-accurate-joint-estimation-of | 2306.01304 | null | https://arxiv.org/abs/2306.01304v2 | https://arxiv.org/pdf/2306.01304v2.pdf | JEPOO: Highly Accurate Joint Estimation of Pitch, Onset and Offset for Music Information Retrieval | Melody extraction is a core task in music information retrieval, and the estimation of pitch, onset and offset are key sub-tasks in melody extraction. Existing methods have limited accuracy, and work for only one type of data, either single-pitch or multipitch. In this paper, we propose a highly accurate method for joint estimation of pitch, onset and offset, named JEPOO. We address the challenges of joint learning optimization and handling both single-pitch and multi-pitch data through novel model design and a new optimization technique named Pareto modulated loss with loss weight regularization. This is the first method that can accurately handle both single-pitch and multi-pitch music data, and even a mix of them. A comprehensive experimental study on a wide range of real datasets shows that JEPOO outperforms state-ofthe-art methods by up to 10.6%, 8.3% and 10.3% for the prediction of Pitch, Onset and Offset, respectively, and JEPOO is robust for various types of data and instruments. The ablation study shows the effectiveness of each component of JEPOO. | ['Gang Wang', 'Yueguo Chen', 'Rui Zhang', 'Jun Yuan', 'Haojie Wei'] | 2023-06-02 | null | null | null | null | ['melody-extraction', 'music-information-retrieval', 'information-retrieval'] | ['music', 'music', 'natural-language-processing'] | [-2.82160550e-01 -8.33639920e-01 -3.74038011e-01 2.95607448e-01
-1.32432783e+00 -7.61260986e-01 -1.29565373e-01 1.39548123e-01
-3.93990248e-01 5.30360341e-01 5.59437498e-02 3.42822254e-01
-5.18294394e-01 -2.05022678e-01 -2.27504596e-01 -6.39229417e-01
-2.48361975e-01 3.63587499e-01 1.60507392e-02 -2.69685596e-01
2.71367162e-01 2.63271242e-01 -1.48383379e+00 1.58145770e-01
4.59719241e-01 1.20317864e+00 4.51751724e-02 7.91623414e-01
2.42078185e-01 4.21025902e-01 -8.18642259e-01 -1.71956226e-01
4.52967077e-01 -3.25119287e-01 -3.39642227e-01 -3.19070131e-01
1.49228185e-01 -5.88776171e-02 -3.82272489e-02 9.29136992e-01
1.22186303e+00 2.43656367e-01 5.61096370e-01 -1.21785307e+00
-9.36848298e-03 8.77533078e-01 -6.66261911e-01 2.02294692e-01
2.84414977e-01 -2.02347875e-01 1.58527601e+00 -9.80472147e-01
-1.58537552e-02 8.92589390e-01 1.05271983e+00 1.60189301e-01
-1.09124279e+00 -1.04472518e+00 -1.78617582e-01 1.56943843e-01
-1.70425117e+00 -2.54369378e-01 1.11960483e+00 -5.37536979e-01
7.63511300e-01 4.74582374e-01 6.30987585e-01 6.08926713e-01
1.31286979e-01 1.01101804e+00 1.01154506e+00 -4.95519042e-01
-1.52197599e-01 -1.59063011e-01 1.97918415e-02 1.71431705e-01
-1.74492866e-01 2.68445790e-01 -9.13477778e-01 -3.51800710e-01
6.26792848e-01 -3.75020117e-01 -1.81614935e-01 2.64150262e-01
-1.21365440e+00 6.99802637e-01 -1.65047273e-01 6.94849864e-02
-2.46342778e-01 -9.03371274e-02 6.10071123e-01 3.02517086e-01
2.82317042e-01 8.21409523e-01 -5.41056871e-01 -5.84184110e-01
-1.35471737e+00 7.26168633e-01 8.06759477e-01 6.36781335e-01
1.06573559e-01 3.46345842e-01 -3.56915236e-01 1.23159409e+00
7.96790868e-02 4.13999170e-01 6.16950929e-01 -6.40239418e-01
5.65534353e-01 2.12737545e-01 2.24559113e-01 -9.27558005e-01
-6.69632971e-01 -6.55363798e-01 -6.45209253e-01 3.39015909e-02
5.34080565e-01 -1.89159140e-01 -4.39275026e-01 1.62135553e+00
2.61684895e-01 1.28584191e-01 -1.43418983e-01 1.08847690e+00
8.74260306e-01 8.05487990e-01 -2.37022907e-01 -6.05139077e-01
1.35643578e+00 -1.17840457e+00 -1.03347981e+00 -5.44806756e-02
-1.54882744e-02 -1.53162873e+00 1.16794562e+00 9.77497756e-01
-1.25876999e+00 -7.76021481e-01 -1.17585623e+00 9.87525806e-02
1.89150855e-01 6.53829932e-01 3.96966606e-01 3.70859116e-01
-6.72389865e-02 7.72689402e-01 -3.28679830e-01 6.66144669e-01
-6.97135255e-02 4.28819895e-01 1.86774671e-01 6.72904432e-01
-1.33845294e+00 4.07522023e-01 6.12436056e-01 -8.20926875e-02
-4.92137134e-01 -9.92634892e-01 -2.83951491e-01 1.26875654e-01
4.83594179e-01 -1.99545667e-01 1.38575459e+00 -5.32439172e-01
-1.74180233e+00 4.49164331e-01 2.43912786e-01 -3.64253432e-01
3.74677479e-01 -6.88809216e-01 -6.35356188e-01 -1.60653576e-01
-1.82993516e-01 1.89450145e-01 1.19944072e+00 -7.16515303e-01
-7.04316080e-01 -1.99472651e-01 -3.59192193e-01 3.88662994e-01
-2.59715825e-01 1.85947970e-01 -4.10169691e-01 -1.13307571e+00
1.80181526e-02 -9.72427189e-01 1.32019088e-01 -6.28500521e-01
-5.71235955e-01 -2.35119939e-01 4.14978057e-01 -8.71534646e-01
2.12951136e+00 -2.37550831e+00 3.11585248e-01 3.97144109e-02
-1.35891646e-01 3.28295738e-01 1.13578305e-01 4.78478700e-01
1.61077082e-02 -1.94621101e-01 -7.87003636e-02 -3.20659876e-01
2.48084247e-01 -1.94458216e-01 -5.44578373e-01 2.54753400e-02
-2.23600522e-01 6.29102468e-01 -6.43344462e-01 -4.25726354e-01
-1.18847676e-01 2.93249756e-01 -5.35507798e-01 1.47824526e-01
1.61599740e-03 3.53088826e-01 -1.50734186e-01 7.27511406e-01
4.34134960e-01 1.39972612e-01 5.09194918e-02 -3.72717619e-01
-3.15272093e-01 4.41108555e-01 -1.74301052e+00 1.64998662e+00
-3.75020742e-01 5.33987463e-01 1.14032574e-01 -4.13889468e-01
1.10784018e+00 5.63322902e-01 7.96206534e-01 -4.40419853e-01
1.50853276e-01 5.25195956e-01 2.44044214e-01 -2.62039125e-01
8.50556791e-01 -4.11784202e-01 -4.06761855e-01 3.89960915e-01
5.12981005e-02 -1.61287844e-01 2.65157104e-01 -3.54157984e-01
5.58360577e-01 -1.27460003e-01 5.08970499e-01 -4.56973612e-02
4.34220701e-01 -2.87382841e-01 1.08559513e+00 5.12390018e-01
-2.57312059e-01 8.00481856e-01 2.89075136e-01 -1.36925802e-01
-1.06844866e+00 -8.03411245e-01 -2.08218515e-01 1.33474839e+00
4.31368910e-02 -9.02262330e-01 -4.73368615e-01 -2.44309083e-01
2.84067094e-01 2.55332083e-01 3.53010260e-02 -7.49051049e-02
-7.17825949e-01 -8.92405808e-01 9.05969262e-01 5.12919128e-01
2.93837667e-01 -1.23559272e+00 -4.42236185e-01 4.79637712e-01
-3.93558353e-01 -1.02059746e+00 -1.13007927e+00 2.07268834e-01
-6.34875357e-01 -8.63779306e-01 -5.88667333e-01 -7.94629097e-01
-3.55537623e-01 -2.38890499e-01 1.06013191e+00 -2.89265931e-01
-5.55763543e-01 1.61158800e-01 -3.52057576e-01 -6.48554265e-01
-8.52594525e-02 2.12318584e-01 3.23257923e-01 2.49982819e-01
-3.40421759e-02 -7.14234531e-01 -6.17136598e-01 4.52473611e-01
-6.03785276e-01 -2.62686163e-01 4.44588572e-01 9.29345727e-01
1.02918911e+00 2.70202905e-01 9.29552138e-01 -5.77412546e-01
1.09503365e+00 -2.22972795e-01 -6.63656175e-01 4.38680984e-02
-7.73565650e-01 -2.46746913e-01 5.27754724e-01 -8.72713864e-01
-5.87429166e-01 1.13194272e-01 -3.13816637e-01 -5.89389086e-01
2.61635602e-01 6.39772177e-01 -1.25368401e-01 1.59589514e-01
5.09444475e-01 8.95142630e-02 -3.84014636e-01 -8.93286347e-01
8.81258398e-02 8.09822977e-01 5.53089738e-01 -4.80198264e-01
5.30707479e-01 7.09068328e-02 1.05291605e-01 -8.43008101e-01
-1.05382097e+00 -8.80559087e-01 -5.00489354e-01 -7.81521276e-02
5.28599501e-01 -9.72356498e-01 -8.22000742e-01 6.24807894e-01
-7.68544137e-01 -7.98280910e-02 -4.96836096e-01 8.34758520e-01
-6.77183986e-01 4.24210876e-01 -7.06005275e-01 -9.86603618e-01
-9.61822391e-01 -1.00805783e+00 1.02108788e+00 1.22239135e-01
-4.07315880e-01 -6.66758835e-01 3.54231358e-01 3.22212040e-01
1.68532487e-02 1.77309420e-02 7.80339062e-01 -6.60494268e-01
-1.39939547e-01 -3.26094896e-01 2.92732954e-01 4.89185870e-01
1.16375901e-01 -2.48423629e-02 -9.32690918e-01 -3.90261650e-01
9.62863415e-02 -4.45981532e-01 7.38506138e-01 4.79254395e-01
1.07207584e+00 -1.98299244e-01 2.33543903e-01 7.28202105e-01
1.16304612e+00 3.01387846e-01 3.09900880e-01 2.65521765e-01
5.35410523e-01 1.50023997e-01 1.14562595e+00 9.24745023e-01
3.91823575e-02 1.16356826e+00 2.35616907e-01 3.67052436e-01
-1.72177389e-01 -2.93968469e-01 4.12406832e-01 1.47636509e+00
-4.73162308e-02 3.35672460e-02 -6.14635646e-01 4.90647048e-01
-1.93700039e+00 -9.67501402e-01 -2.54818667e-02 2.48106694e+00
1.11566222e+00 1.50790796e-01 6.41401827e-01 7.34978437e-01
5.38154900e-01 1.89360246e-01 -4.59836781e-01 -2.49682397e-01
-1.36701763e-01 5.40760219e-01 3.30611914e-01 4.01347518e-01
-1.45749271e+00 7.19164908e-01 6.90579367e+00 1.50046158e+00
-1.10897434e+00 -5.90184219e-02 -1.48274451e-01 -5.46855628e-01
1.22305848e-01 -1.16627276e-01 -1.11863184e+00 4.97727305e-01
5.52525759e-01 -2.18993366e-01 6.30303741e-01 6.28682911e-01
9.48215574e-02 2.59012699e-01 -7.95192063e-01 1.57151699e+00
1.24160023e-02 -8.89122605e-01 -2.59617597e-01 -1.70713216e-01
7.49228418e-01 -3.73336256e-01 1.05698243e-01 5.04904687e-01
-5.99658012e-01 -8.50936294e-01 8.80916834e-01 6.11526668e-01
6.96109414e-01 -1.10905683e+00 4.56100285e-01 5.46209276e-01
-1.46590865e+00 -2.65007079e-01 -1.05008595e-01 -8.47609267e-02
2.65441418e-01 5.63823223e-01 -5.05053401e-01 6.66445971e-01
6.39163017e-01 6.72482491e-01 -1.30001962e-01 1.49578559e+00
-2.66144603e-01 9.02046800e-01 -4.81817007e-01 1.10292330e-01
-1.44018322e-01 -7.40690157e-02 1.01310027e+00 1.21998453e+00
4.41601396e-01 -1.20403863e-01 5.32163560e-01 4.75338548e-01
1.24047948e-02 5.00347912e-01 1.72774374e-01 -1.65567681e-01
6.93763673e-01 1.38987613e+00 -4.06735659e-01 -6.50455430e-02
2.80631091e-02 5.15582561e-01 -2.14958563e-02 4.07711305e-02
-9.34733093e-01 -6.83143079e-01 4.89497095e-01 -1.37107134e-01
4.44167405e-01 -1.52652919e-01 -3.31240565e-01 -1.07208061e+00
1.53145045e-01 -1.17095673e+00 6.25164211e-01 -3.65716726e-01
-1.15870678e+00 4.11428392e-01 -1.56878278e-01 -1.76200223e+00
-4.06989098e-01 -4.11805421e-01 -6.09712601e-01 8.14464867e-01
-1.38881958e+00 -7.31430352e-01 6.24805801e-02 4.96542931e-01
7.37519503e-01 -5.59367061e-01 8.53064775e-01 6.55640602e-01
-5.26126921e-01 8.99562478e-01 2.21544400e-01 -1.60973985e-02
9.64919329e-01 -1.30859184e+00 -3.16528492e-02 3.22241306e-01
6.98829353e-01 1.75297663e-01 6.72840714e-01 -3.27391058e-01
-1.37814164e+00 -7.03788817e-01 7.47857630e-01 -1.56324506e-01
5.78564405e-01 -2.02833191e-01 -8.81498098e-01 1.70293182e-01
-6.12285286e-02 -3.37606639e-01 9.84801829e-01 4.31721598e-01
-2.32894883e-01 -3.69203746e-01 -6.59259081e-01 3.90053391e-01
5.25802016e-01 -5.77026427e-01 -6.14814639e-01 9.37332734e-02
4.20047492e-01 -6.97817028e-01 -1.21249425e+00 7.79553592e-01
9.39463854e-01 -7.81466186e-01 1.17179644e+00 -4.76516724e-01
-6.44856412e-03 -4.44280177e-01 -2.36173302e-01 -1.17231846e+00
-3.45731825e-01 -1.08068871e+00 -5.53617299e-01 1.30247843e+00
3.86616796e-01 -4.67002094e-02 5.26350021e-01 -1.31090060e-01
-2.14716822e-01 -7.94309199e-01 -9.69466209e-01 -1.00636292e+00
-1.41201660e-01 -7.17897534e-01 4.23155546e-01 7.59303272e-01
7.32099637e-02 6.30485594e-01 -1.04841244e+00 -5.84031045e-02
6.82295620e-01 5.90813577e-01 8.57594788e-01 -1.11680901e+00
-8.72324467e-01 -7.24988341e-01 -3.19287091e-01 -8.46249759e-01
-1.49713546e-01 -8.33985031e-01 -2.27454072e-03 -9.11833346e-01
-9.83329564e-02 -2.85035878e-01 -6.82071924e-01 2.50851631e-01
-3.63707215e-01 4.39821988e-01 4.61254776e-01 5.25703907e-01
-3.85566920e-01 5.91846943e-01 1.26881278e+00 -8.91104341e-02
-7.16652811e-01 6.49244070e-01 -3.25766534e-01 9.78911340e-01
9.35026228e-01 -5.53426683e-01 -2.47817084e-01 9.58930701e-02
4.33475941e-01 2.58035809e-01 9.68696699e-02 -1.08105648e+00
1.68352172e-01 6.86580539e-02 1.18255414e-01 -1.00421929e+00
7.72964060e-01 -4.54769969e-01 1.43098071e-01 5.61031625e-02
-3.22748125e-01 2.35113837e-02 4.58038867e-01 4.44153607e-01
-6.98898971e-01 -4.64370996e-01 5.91126323e-01 2.21212566e-01
-6.34853780e-01 1.52881458e-01 9.33295563e-02 3.50103319e-01
7.19888866e-01 1.38521835e-01 2.08518371e-01 -3.63045186e-01
-1.08116198e+00 2.08527684e-01 -4.49359328e-01 4.67208385e-01
6.41369164e-01 -1.75172484e+00 -7.72173941e-01 1.15327112e-01
-6.46170750e-02 -3.96733999e-01 3.93385172e-01 1.03214335e+00
-1.81823090e-01 2.41037488e-01 -8.42039958e-02 -4.63712454e-01
-1.69725788e+00 4.40940291e-01 1.60449952e-01 -6.07440650e-01
-3.82393211e-01 8.31927955e-01 -8.90630111e-02 -3.08895409e-01
5.70461512e-01 -2.27749139e-01 -3.27792197e-01 5.68744183e-01
4.98899788e-01 5.92084289e-01 2.92306226e-02 -5.81547558e-01
-1.58960894e-01 9.12642598e-01 2.97461711e-02 -1.12633780e-01
1.03712249e+00 2.24108756e-01 1.43205598e-02 1.03611624e+00
9.41393733e-01 6.24066114e-01 -1.11539185e+00 -1.08557470e-01
3.79686579e-02 -4.48887199e-01 -2.53917947e-02 -8.35996330e-01
-6.84885204e-01 8.86686504e-01 5.30223072e-01 2.17657298e-01
1.42337716e+00 -4.33172256e-01 1.11720347e+00 1.15983307e-01
1.75560132e-01 -1.42908525e+00 2.49495044e-01 6.81531131e-01
1.01356471e+00 -9.43904281e-01 2.43651971e-01 -2.19565168e-01
-8.46283078e-01 1.02600241e+00 3.24190259e-01 -2.84801960e-01
8.19314778e-01 1.98316917e-01 8.52122381e-02 1.46477506e-01
-7.66861379e-01 -2.37655178e-01 1.00880516e+00 -1.14547081e-01
5.96389174e-01 3.06004614e-01 -5.64187706e-01 1.25870895e+00
-6.01929665e-01 -2.29510590e-01 4.24910430e-03 7.13459373e-01
-4.53767627e-01 -1.36291790e+00 -5.29217303e-01 1.90822467e-01
-1.02879560e+00 -9.96461734e-02 -3.76600474e-01 8.25140595e-01
7.26118162e-02 7.63061523e-01 -3.47391367e-01 -7.43271828e-01
6.36712134e-01 2.73203999e-01 5.47287643e-01 -2.89614171e-01
-1.01414299e+00 8.82000685e-01 -2.94513926e-02 -1.55545980e-01
-2.32268438e-01 -5.44870794e-01 -1.14665794e+00 8.58741999e-02
-7.09326148e-01 4.01480854e-01 6.47896290e-01 6.48320854e-01
1.85994446e-01 7.85528660e-01 9.22857285e-01 -7.98241615e-01
-9.81421590e-01 -1.05496407e+00 -1.16807485e+00 3.39517772e-01
4.28609192e-01 -6.43761396e-01 -3.70021731e-01 -2.13358000e-01] | [15.824742317199707, 5.389732837677002] |
d3127fde-86c5-4a5e-81f9-3dbcdec36370 | improving-sketch-colorization-using | 2301.08590 | null | https://arxiv.org/abs/2301.08590v1 | https://arxiv.org/pdf/2301.08590v1.pdf | Improving Sketch Colorization using Adversarial Segmentation Consistency | We propose a new method for producing color images from sketches. Current solutions in sketch colorization either necessitate additional user instruction or are restricted to the "paired" translation strategy. We leverage semantic image segmentation from a general-purpose panoptic segmentation network to generate an additional adversarial loss function. The proposed loss function is compatible with any GAN model. Our method is not restricted to datasets with segmentation labels and can be applied to unpaired translation tasks as well. Using qualitative, and quantitative analysis, and based on a user study, we demonstrate the efficacy of our method on four distinct image datasets. On the FID metric, our model improves the baseline by up to 35 points. Our code, pretrained models, scripts to produce newly introduced datasets and corresponding sketch images are available at https://github.com/giddyyupp/AdvSegLoss. | ['Pinar Duygulu', 'Emre Akbas', 'Nermin Samet', 'Samet Hicsonmez'] | 2023-01-20 | null | null | null | null | ['panoptic-segmentation', 'colorization'] | ['computer-vision', 'computer-vision'] | [ 5.37732720e-01 -9.35761258e-02 -1.79819882e-01 -3.43337685e-01
-9.53579485e-01 -1.08845413e+00 7.03545809e-01 -5.66820383e-01
-1.32191673e-01 6.42693758e-01 -3.33712071e-01 -3.38462383e-01
4.25096005e-01 -7.94321775e-01 -7.94511795e-01 -4.86796886e-01
4.84977841e-01 3.87444705e-01 -9.40388441e-02 -1.66828617e-01
3.77243042e-01 4.81383353e-01 -9.69552100e-01 6.99823424e-02
1.02521765e+00 9.51915741e-01 -1.36642717e-02 6.53328001e-01
-2.21807778e-01 1.67239428e-01 -5.66592157e-01 -8.31947327e-01
7.20813990e-01 -6.37076139e-01 -6.01892531e-01 7.65883625e-02
8.93688202e-01 -5.66054642e-01 -2.23360285e-01 1.10571682e+00
4.46483582e-01 -4.34824005e-02 5.78800023e-01 -1.58234394e+00
-9.99249041e-01 4.17053580e-01 -5.39086103e-01 -4.59112942e-01
2.11939588e-01 3.52850586e-01 9.34249938e-01 -1.03938723e+00
8.28415155e-01 1.06278193e+00 6.67222023e-01 8.42961073e-01
-1.56825483e+00 -9.50496495e-01 9.19259489e-02 -3.96458447e-01
-1.20793962e+00 -3.21706086e-01 1.15051758e+00 -3.30190778e-01
1.88178971e-01 3.23820949e-01 5.68281353e-01 1.64152253e+00
-2.72413850e-01 8.75931799e-01 1.53049326e+00 -4.48567480e-01
1.25411972e-01 2.38338605e-01 -3.15305591e-01 9.88911211e-01
-2.47291266e-03 1.19775660e-01 -2.30725959e-01 5.54226525e-03
1.13365924e+00 -1.22821689e-01 -1.62306845e-01 -5.33033311e-01
-1.08569157e+00 8.52109194e-01 4.24778432e-01 1.48466364e-01
-1.20692467e-02 4.96526897e-01 5.61687909e-02 3.14853966e-01
5.34524262e-01 5.53957582e-01 -1.83705509e-01 -6.77995160e-02
-1.01713514e+00 3.13327819e-01 5.88703096e-01 1.19750965e+00
8.80602419e-01 1.25911191e-01 -3.06786485e-02 9.20955598e-01
1.13396142e-02 7.81602263e-01 1.02388710e-02 -1.29171109e+00
4.39655870e-01 3.59164208e-01 1.50496870e-01 -8.18227351e-01
9.14567187e-02 -4.52206209e-02 -6.16835296e-01 6.02862775e-01
6.72464907e-01 -3.31024826e-01 -1.28776562e+00 1.86784852e+00
-1.13610039e-03 1.04258945e-02 -3.20066333e-01 9.57311094e-01
5.78932822e-01 4.89802450e-01 7.05170706e-02 5.08337677e-01
1.13652527e+00 -1.09462047e+00 -4.85790908e-01 2.90160608e-02
1.01450734e-01 -1.04923069e+00 1.75877881e+00 5.15712976e-01
-1.19282889e+00 -5.21924078e-01 -1.02552974e+00 -2.27844700e-01
-5.22732317e-01 4.86373454e-01 7.41182327e-01 9.40800250e-01
-1.25840116e+00 7.15581834e-01 -7.20200300e-01 -5.84011853e-01
5.66825628e-01 1.58681259e-01 -2.00493038e-01 4.06239294e-02
-8.43710840e-01 5.64477503e-01 1.36226237e-01 -6.23917915e-02
-8.73759985e-01 -6.68241560e-01 -7.13510215e-01 -2.79365331e-01
2.07221031e-01 -7.36923993e-01 1.02559650e+00 -1.37112463e+00
-2.00736165e+00 9.73380506e-01 1.30962685e-01 -3.83746028e-02
1.00444269e+00 -1.87305599e-01 -8.77334177e-02 3.33047986e-01
8.82466976e-03 1.19117498e+00 1.08003032e+00 -1.57598221e+00
-1.65119469e-01 3.41577418e-02 1.51136935e-01 4.41675298e-02
-2.00791597e-01 -1.59775898e-01 -7.64614820e-01 -1.23127699e+00
-2.70811826e-01 -1.27713490e+00 -1.30538478e-01 4.40634996e-01
-6.12744331e-01 3.86311024e-01 8.85065794e-01 -7.89657235e-01
5.98881781e-01 -1.97184134e+00 1.90080419e-01 4.47674483e-01
-5.91782182e-02 1.63414150e-01 -4.98614609e-01 3.65005314e-01
6.93238080e-02 4.64948386e-01 -7.94472277e-01 -7.04547882e-01
3.25589657e-01 9.83477309e-02 -6.16764605e-01 1.13918334e-01
3.54407787e-01 1.05171931e+00 -7.24636078e-01 -3.59505147e-01
3.39955598e-01 4.31751639e-01 -4.87706691e-01 3.74838084e-01
-4.39571589e-01 6.69878185e-01 -1.66165888e-01 8.28771055e-01
9.36885655e-01 -8.07361156e-02 4.40286174e-02 -1.17677383e-01
1.13947012e-01 -1.51471511e-01 -8.80247176e-01 2.33329511e+00
-4.88514215e-01 6.37406766e-01 1.10364869e-01 -6.33500099e-01
9.84371781e-01 1.41049242e-02 1.37582257e-01 -5.62364042e-01
3.27055193e-02 2.82933623e-01 -5.16756296e-01 -8.20442513e-02
5.28385341e-01 -3.40079740e-02 -1.79056868e-01 6.24682367e-01
3.28406811e-01 -6.40864909e-01 2.59833932e-01 3.31044495e-01
6.97648466e-01 7.04994738e-01 -3.58406544e-01 -1.97262704e-01
2.67604619e-01 -1.27170712e-01 2.92513222e-01 6.78046227e-01
1.60579244e-03 1.09092724e+00 6.32146120e-01 -2.43682638e-01
-1.37591553e+00 -1.36763382e+00 -7.92505518e-02 1.02150333e+00
2.55600005e-01 -2.21958116e-01 -1.09301186e+00 -9.63471353e-01
1.03890948e-01 7.65394926e-01 -7.04129517e-01 1.84271812e-01
-4.44426805e-01 -4.59635347e-01 8.22275937e-01 4.82556254e-01
6.01801097e-01 -1.08512771e+00 -1.37701929e-01 -2.89627910e-01
9.19995271e-03 -1.03253531e+00 -7.64240801e-01 -3.23395759e-01
-6.53135240e-01 -9.76705134e-01 -1.05266237e+00 -6.17672503e-01
9.06940103e-01 -2.44057234e-02 1.03227890e+00 2.12155297e-01
-2.81857669e-01 6.64769948e-01 -1.92431673e-01 -1.92513764e-01
-4.76620317e-01 2.92803496e-01 -3.03935528e-01 2.54770219e-02
-1.03558779e-01 -6.84898198e-01 -8.69206190e-01 4.02134866e-01
-9.97128904e-01 4.18442637e-01 4.90129977e-01 7.55788445e-01
3.84392709e-01 -6.37638032e-01 4.69909698e-01 -1.05552065e+00
7.58442521e-01 -9.87288281e-02 -8.68063152e-01 3.42373878e-01
-7.40387678e-01 7.10276663e-02 6.95289552e-01 -3.52501065e-01
-1.21249092e+00 1.19957946e-01 -2.55533278e-01 -5.39854467e-01
-3.86413753e-01 -6.20457381e-02 -1.98922500e-01 -4.88975257e-01
4.58290547e-01 5.03306054e-02 4.55769747e-02 -4.23579961e-01
9.84314919e-01 3.03357840e-01 5.74941278e-01 -1.07048464e+00
1.08225560e+00 5.22506833e-01 -2.42389113e-01 -4.57682401e-01
-2.06942648e-01 3.18615824e-01 -6.02996588e-01 -2.22875938e-01
7.36353338e-01 -6.65470064e-01 -4.81107265e-01 5.87072611e-01
-9.62632716e-01 -8.06627095e-01 -1.97473049e-01 5.25933616e-02
-6.58399582e-01 3.68524820e-01 -8.43294501e-01 -5.44432640e-01
-5.38107216e-01 -1.15232956e+00 1.09775162e+00 2.86489755e-01
-4.81864214e-02 -9.85997975e-01 -1.08820358e-02 3.79539400e-01
6.97182894e-01 5.33220470e-01 7.38241911e-01 -2.00034678e-01
-7.70342410e-01 8.24050698e-03 -4.24874902e-01 4.55570996e-01
1.72849670e-01 3.91997069e-01 -9.61082220e-01 -2.59757429e-01
-4.82374370e-01 -5.97995996e-01 8.89569879e-01 6.68539703e-02
1.43338239e+00 -1.97289959e-01 -2.02527214e-02 9.64100659e-01
1.48066711e+00 1.06184930e-01 7.37038195e-01 4.19956669e-02
9.72636640e-01 4.86335307e-01 3.24757963e-01 1.88761964e-01
1.53460696e-01 6.52175188e-01 2.42612228e-01 -4.33075041e-01
-3.72101992e-01 -6.06378019e-01 2.43317544e-01 5.19412816e-01
-2.52881169e-01 -2.63850778e-01 -7.21457660e-01 4.15825337e-01
-1.62527847e+00 -7.61982501e-01 2.91660041e-01 2.02790356e+00
8.56317699e-01 -6.00758567e-02 2.40974948e-01 -4.07655358e-01
6.73852324e-01 1.30824760e-01 -4.79525059e-01 -4.52966571e-01
-5.62854558e-02 7.70513892e-01 5.51122904e-01 6.09491765e-01
-1.06184089e+00 1.41270816e+00 6.09223032e+00 8.97191823e-01
-1.40017855e+00 1.71030894e-01 7.88978040e-01 -3.84122282e-02
-7.36972570e-01 1.07690006e-01 -1.95267826e-01 5.77714860e-01
3.89881432e-01 1.13436908e-01 8.26606512e-01 6.38567626e-01
-7.22916201e-02 6.87930062e-02 -8.80820334e-01 9.03171003e-01
-5.82195781e-02 -1.34731424e+00 1.99693218e-01 -1.12013601e-01
9.03106749e-01 -2.75748938e-01 5.22152126e-01 1.29216924e-01
5.07201433e-01 -1.03869665e+00 7.60059536e-01 3.21815282e-01
1.36283994e+00 -6.56917095e-01 1.17761128e-01 -4.33014482e-01
-9.60689008e-01 4.27739918e-01 -1.80405062e-02 2.70250350e-01
1.42438456e-01 1.33601412e-01 -6.31131291e-01 5.29072583e-01
3.35730344e-01 5.47261775e-01 -7.64237463e-01 5.86903334e-01
-7.06137121e-01 6.61219358e-01 -2.30182558e-01 2.49403149e-01
1.28065974e-01 -6.38381422e-01 2.52620578e-01 1.19817543e+00
4.45423990e-01 -9.47986096e-02 -5.10723852e-02 1.45798123e+00
-4.21140462e-01 4.90530580e-02 -7.16126204e-01 -2.50584900e-01
4.32786256e-01 1.49583817e+00 -9.42548156e-01 -3.57806534e-01
-2.75910348e-01 1.61969554e+00 2.33763993e-01 7.45424092e-01
-1.02696407e+00 -7.23536134e-01 6.32321656e-01 -1.17140442e-01
2.65911996e-01 -2.43722335e-01 -6.87572479e-01 -1.36588490e+00
1.26875356e-01 -9.23179388e-01 3.03575699e-03 -8.81690264e-01
-1.34337544e+00 4.99958545e-01 -8.11002478e-02 -1.08046758e+00
-8.86711404e-02 -6.98045373e-01 -9.04478788e-01 8.89904022e-01
-1.16842806e+00 -1.65366638e+00 -5.94771743e-01 6.42633259e-01
3.23000520e-01 -1.89136684e-01 9.46138024e-01 5.05003870e-01
-5.37321329e-01 9.92479324e-01 -2.17333943e-01 2.83700854e-01
9.56173301e-01 -1.49485636e+00 9.38385725e-01 1.00487757e+00
1.97166488e-01 6.31895125e-01 5.76460660e-01 -4.75945383e-01
-1.19613373e+00 -9.67157185e-01 1.73174679e-01 -5.57606161e-01
5.35210431e-01 -8.49377811e-01 -6.79590225e-01 7.43609190e-01
6.85959518e-01 -8.21090341e-02 4.95896697e-01 -1.92978576e-01
-5.62342107e-01 -5.39621674e-02 -1.44175577e+00 8.76423061e-01
1.26560581e+00 -6.30274773e-01 -1.30659968e-01 1.50557816e-01
6.13186479e-01 -5.32205343e-01 -8.26194048e-01 1.82548270e-01
8.05758059e-01 -7.03766763e-01 8.99751306e-01 -3.73334795e-01
6.57017529e-01 -3.01365376e-01 2.36301534e-02 -1.26514244e+00
7.80371577e-02 -8.96167934e-01 3.27215016e-01 1.38070786e+00
5.57565629e-01 -6.77303374e-01 9.70392942e-01 8.59542310e-01
3.07197198e-02 -5.19841194e-01 -5.22102892e-01 -6.28565967e-01
4.23036009e-01 -3.44027370e-01 5.97983301e-01 1.08074617e+00
-4.31213021e-01 -6.39155507e-03 -5.83015859e-01 -1.00803658e-01
8.95707846e-01 3.77447516e-01 1.14513755e+00 -8.13983977e-01
-2.48888642e-01 -7.35204518e-01 7.39158168e-02 -8.62329483e-01
2.38671213e-01 -8.76104116e-01 -1.85579792e-01 -1.30347109e+00
-4.14667577e-02 -7.52071619e-01 -1.64118871e-01 7.22332656e-01
-5.58031239e-02 8.59690607e-01 5.91197610e-01 8.35627243e-02
-1.76255718e-01 5.91921806e-01 1.23069000e+00 -1.83287799e-01
-6.73240349e-02 -2.04054698e-01 -5.93701601e-01 4.92881387e-01
1.10559392e+00 -3.14351916e-01 -4.10296232e-01 -4.37940776e-01
2.07353216e-02 -2.51013309e-01 5.56387365e-01 -8.57157171e-01
-1.90710023e-01 -1.14407584e-01 4.28888500e-01 -7.88805559e-02
4.58722979e-01 -7.88788736e-01 3.26757818e-01 2.70622939e-01
-4.45156932e-01 8.15468729e-02 3.71804833e-01 2.02777684e-01
8.62684473e-02 -1.02892250e-01 8.18423629e-01 -9.52848494e-02
-6.25940323e-01 3.66542727e-01 1.65014207e-01 1.99930612e-02
8.96147788e-01 3.34304050e-02 -5.26098847e-01 -5.33547163e-01
-6.51571333e-01 -3.09080593e-02 1.12711501e+00 5.42128026e-01
5.33996880e-01 -1.60265660e+00 -5.14826834e-01 2.72133112e-01
6.94555342e-02 -3.39513570e-01 1.20711243e-02 5.28421462e-01
-1.00485241e+00 2.50211544e-02 -5.61799049e-01 -3.42976183e-01
-8.98175418e-01 4.91941750e-01 2.32400268e-01 1.05053715e-01
-5.39883256e-01 6.95230544e-01 2.09720612e-01 -7.30002582e-01
7.18941763e-02 -4.06817228e-01 5.52458405e-01 -2.29315653e-01
1.19613320e-01 1.30305305e-01 -3.01081836e-01 -1.79278776e-01
-9.68635082e-02 6.77264154e-01 1.38867319e-01 -6.32057250e-01
1.08348346e+00 3.88576947e-02 1.92813538e-02 1.65394083e-01
1.15651786e+00 2.64967322e-01 -1.61946559e+00 3.39174531e-02
-3.17214131e-01 -7.00534165e-01 -4.01827604e-01 -1.06230271e+00
-1.46172714e+00 8.35062087e-01 6.43363893e-01 -4.63918224e-02
1.21443284e+00 -2.64919400e-01 8.68675292e-01 -1.56705808e-02
3.65132689e-01 -9.62806761e-01 1.79878682e-01 3.50170545e-02
1.03307962e+00 -1.37500119e+00 -1.87349185e-01 -3.19313109e-01
-7.36471176e-01 1.06618488e+00 6.53564394e-01 -4.21956033e-01
2.64054179e-01 3.06736827e-01 3.52338612e-01 1.78996623e-01
-2.28130162e-01 -2.51326840e-02 2.18269959e-01 4.61388499e-01
3.82632911e-01 2.01321557e-01 -3.60892743e-01 1.84783205e-01
-1.62138149e-01 8.06541443e-02 3.41517538e-01 6.89517498e-01
2.31875002e-01 -1.59567595e+00 -2.29063511e-01 1.47175089e-01
-3.27217400e-01 -1.31673783e-01 -6.20102882e-01 8.87635469e-01
3.23515534e-02 5.86159766e-01 4.31099571e-02 -4.31140035e-01
9.35688242e-02 8.75437781e-02 7.30033636e-01 -2.03195661e-01
-4.80074704e-01 -6.67465031e-02 2.41941810e-02 -5.97122908e-01
-2.59365112e-01 -4.43880588e-01 -9.28906381e-01 -5.62738836e-01
9.03108418e-02 -1.21977761e-01 8.28658462e-01 3.50721389e-01
4.74127084e-01 1.30134448e-01 5.47137856e-01 -1.10169637e+00
-8.61003473e-02 -8.81034493e-01 -2.86499858e-01 7.11008787e-01
1.01543311e-02 -4.45774555e-01 -3.31775516e-01 2.04656512e-01] | [11.699539184570312, -0.6074205040931702] |
33cb7f0b-a5cd-48ba-8b3f-30dac3a6f7c2 | a-lightweight-network-for-photovoltaic-cell | 2302.07455 | null | https://arxiv.org/abs/2302.07455v1 | https://arxiv.org/pdf/2302.07455v1.pdf | A lightweight network for photovoltaic cell defect detection in electroluminescence images based on neural architecture search and knowledge distillation | Nowadays, the rapid development of photovoltaic(PV) power stations requires increasingly reliable maintenance and fault diagnosis of PV modules in the field. Due to the effectiveness, convolutional neural network (CNN) has been widely used in the existing automatic defect detection of PV cells. However, the parameters of these CNN-based models are very large, which require stringent hardware resources and it is difficult to be applied in actual industrial projects. To solve these problems, we propose a novel lightweight high-performance model for automatic defect detection of PV cells in electroluminescence(EL) images based on neural architecture search and knowledge distillation. To auto-design an effective lightweight model, we introduce neural architecture search to the field of PV cell defect classification for the first time. Since the defect can be any size, we design a proper search structure of network to better exploit the multi-scale characteristic. To improve the overall performance of the searched lightweight model, we further transfer the knowledge learned by the existing pre-trained large-scale model based on knowledge distillation. Different kinds of knowledge are exploited and transferred, including attention information, feature information, logit information and task-oriented information. Experiments have demonstrated that the proposed model achieves the state-of-the-art performance on the public PV cell dataset of EL images under online data augmentation with accuracy of 91.74% and the parameters of 1.85M. The proposed lightweight high-performance model can be easily deployed to the end devices of the actual industrial projects and retain the accuracy. | ['Kanjian Zhang', 'Haikun Wei', 'Xinyi Chen', 'Jinxia Zhang'] | 2023-02-15 | null | null | null | null | ['defect-detection'] | ['computer-vision'] | [-8.24731961e-02 -4.49130714e-01 1.70413658e-01 9.99300331e-02
-3.52554739e-01 -2.16113850e-01 -1.21069655e-01 -1.12550840e-01
1.90636832e-02 6.78970933e-01 -5.75241029e-01 -2.32815519e-01
-7.16753155e-02 -1.00603211e+00 -6.41907334e-01 -1.17834818e+00
2.54946768e-01 1.32701576e-01 1.92008048e-01 2.07550541e-01
3.62246096e-01 6.95174813e-01 -1.64979136e+00 1.83425605e-01
1.64794719e+00 1.64409292e+00 9.60227132e-01 2.77130455e-01
-1.24992371e-01 5.37461162e-01 -8.82193387e-01 1.25993863e-01
-8.80869180e-02 -1.86497718e-01 -4.37257737e-01 1.22239091e-01
9.88822803e-02 -3.56634632e-02 -3.19444388e-01 1.17707396e+00
1.00545776e+00 -4.11462903e-01 6.24312222e-01 -1.25410104e+00
-9.11169410e-01 -9.74071249e-02 -2.79050022e-01 3.81476521e-01
-2.13983312e-01 3.32019418e-01 4.63525593e-01 -9.54692423e-01
1.45089597e-01 8.51939559e-01 8.46980512e-01 2.45664269e-01
-8.35666895e-01 -8.36525083e-01 -1.66973040e-01 7.91240394e-01
-1.45460391e+00 -7.67986551e-02 1.07539785e+00 -4.53688383e-01
1.27070320e+00 -7.92775825e-02 9.26784456e-01 8.31030309e-01
4.02305037e-01 6.35261536e-01 9.68338788e-01 -2.25975782e-01
2.06469804e-01 1.12022191e-01 -1.83196396e-01 9.50196505e-01
5.86785078e-01 -1.16708986e-01 -2.02815175e-01 6.19805325e-03
7.34772384e-01 1.43407449e-01 -5.78696370e-01 -2.15668023e-01
-6.07411921e-01 6.10337317e-01 8.44318807e-01 5.97756863e-01
-2.92605221e-01 -7.67125860e-02 4.01908487e-01 2.16083735e-01
3.36340219e-01 2.95274854e-01 -7.22075522e-01 6.05839044e-02
-4.88230258e-01 -2.95073569e-01 6.57716513e-01 9.82559681e-01
7.98529685e-01 3.38562101e-01 -3.16950500e-01 1.16273737e+00
2.61167109e-01 6.07088685e-01 8.13554585e-01 -4.70146447e-01
4.14405912e-01 1.20178699e+00 -2.16117978e-01 -7.48283446e-01
-3.75748336e-01 -6.37319267e-01 -1.37099516e+00 1.46086738e-01
-2.83289969e-01 -1.00199960e-01 -1.00428021e+00 1.02507746e+00
8.10948238e-02 1.50268376e-01 1.67815372e-01 5.81484318e-01
9.36645389e-01 8.87336254e-01 3.42015550e-02 -3.64835203e-01
1.32780325e+00 -1.22544897e+00 -6.47542953e-01 -2.22671069e-02
7.47408509e-01 -4.04593110e-01 8.88491869e-01 3.58571142e-01
-6.77411437e-01 -7.67489076e-01 -1.38411570e+00 8.11709389e-02
-5.16196966e-01 7.96995640e-01 4.56615925e-01 6.31254733e-01
-8.64143610e-01 6.78245306e-01 -5.08068740e-01 -4.33329433e-01
1.17327261e+00 4.69365209e-01 -1.24300629e-01 -2.31713414e-01
-7.46033967e-01 9.35349643e-01 4.82385993e-01 5.22886515e-01
-9.85063791e-01 -6.06297851e-01 -5.16217291e-01 3.67435038e-01
1.47276685e-01 -6.72924697e-01 8.34489763e-01 -9.20644581e-01
-1.45366144e+00 3.56296629e-01 1.34810612e-01 -1.32107124e-01
9.37170312e-02 1.06010757e-01 -5.73411465e-01 1.88836977e-01
-1.19035102e-01 3.67606729e-01 6.65823460e-01 -1.22327793e+00
-8.62170100e-01 -3.67663115e-01 -4.96494949e-01 1.37042612e-01
-9.49714184e-01 -3.73914361e-01 -2.87016392e-01 -3.60265315e-01
-1.07983472e-02 -5.91475308e-01 1.98983297e-01 1.58830166e-01
-6.19838983e-02 -6.84939981e-01 1.45276749e+00 -8.39530826e-01
9.06677663e-01 -2.05846715e+00 -1.56947523e-01 5.28571233e-02
-1.01525784e-01 5.87765992e-01 -1.69135377e-01 2.20463779e-02
3.45662795e-02 -1.26454532e-01 -2.77478725e-01 -4.80519012e-02
-2.29510203e-01 3.34799498e-01 2.12804839e-01 3.90181661e-01
3.64950448e-01 1.18568540e+00 -4.42515463e-01 -4.41242427e-01
2.44067162e-01 2.85637885e-01 -2.49603674e-01 3.56170416e-01
-1.09047212e-01 3.73322219e-01 -5.31690776e-01 1.26212204e+00
9.60113049e-01 -2.68931329e-01 -3.48030217e-02 -7.06598043e-01
-1.18617043e-01 -2.59764522e-01 -6.98533058e-01 1.73683667e+00
-4.64277744e-01 5.49536705e-01 2.05018800e-02 -1.36611068e+00
1.22716868e+00 2.00432226e-01 5.42021513e-01 -9.03189301e-01
2.52654344e-01 4.66917753e-01 -8.30581486e-02 -9.71806109e-01
-1.43860981e-01 7.79899582e-02 3.88830066e-01 -2.92394578e-01
2.69337177e-01 -1.40837759e-01 4.21111099e-02 -5.10769546e-01
1.14354837e+00 -1.68950453e-01 -1.15307286e-01 -4.39276069e-01
8.36082041e-01 -1.53724775e-01 7.59297609e-01 3.77492309e-01
8.92981812e-02 2.64077693e-01 2.04386073e-03 -3.69090140e-01
-9.72676516e-01 -6.55413628e-01 -4.12607789e-01 2.98985928e-01
3.07865143e-01 1.28873423e-01 -5.77576876e-01 -9.05158937e-01
2.81170458e-01 1.99883983e-01 -5.97961731e-02 -4.02439415e-01
-3.63996327e-01 -9.93238628e-01 5.43399155e-01 8.64451826e-01
1.20846951e+00 -1.44367599e+00 -4.08793539e-01 2.41789177e-01
8.19227025e-02 -9.63300467e-01 4.32538688e-02 2.39729986e-01
-1.22405195e+00 -1.18052077e+00 -7.79177606e-01 -1.47362566e+00
9.20593143e-01 4.18307222e-02 8.18180621e-01 4.17175889e-01
-5.54187417e-01 2.23497614e-01 -1.93252206e-01 -5.22399962e-01
-1.32835805e-02 1.49834111e-01 -5.31362556e-02 -1.99322239e-01
2.76575953e-01 -6.14144623e-01 -6.66361034e-01 2.57637292e-01
-6.52549326e-01 -2.64586926e-01 1.00798154e+00 1.14385843e+00
3.93584371e-01 8.45527053e-01 1.12209952e+00 -3.84423435e-01
6.10243976e-01 -2.36104488e-01 -8.72008681e-01 4.22089815e-01
-9.87742662e-01 -2.95128137e-01 5.49109936e-01 -4.79459643e-01
-1.16956222e+00 -1.19847365e-01 -3.25586945e-01 -6.43810034e-01
-2.01404750e-01 6.45654142e-01 -5.61864197e-01 -7.23328948e-01
3.00359726e-01 3.04085791e-01 -1.45609900e-01 -7.07146406e-01
-1.68313503e-01 1.03052998e+00 4.01208133e-01 -5.27414739e-01
6.00707352e-01 9.15962681e-02 1.10391408e-01 -8.15840781e-01
-4.00211811e-01 -1.90295756e-01 -3.22990924e-01 3.71354185e-02
6.52485728e-01 -1.18475568e+00 -9.72398877e-01 1.06884682e+00
-1.20837486e+00 -2.86397725e-01 -2.82262504e-01 1.04636125e-01
-1.95181698e-01 6.68624759e-01 -5.61651289e-01 -8.05956304e-01
-9.05799508e-01 -8.93300295e-01 8.11741769e-01 6.87405288e-01
7.28947759e-01 -7.70048141e-01 -3.83813977e-01 3.40552717e-01
5.78135550e-01 -2.12636650e-01 1.45368171e+00 -2.90960819e-01
-8.22277427e-01 -1.52561426e-01 -5.51752210e-01 9.07088757e-01
2.83401042e-01 -2.75782347e-01 -1.02349269e+00 -5.53833067e-01
3.07371557e-01 -4.41944867e-01 1.00922740e+00 2.15667322e-01
1.80016577e+00 -2.15855524e-01 -6.39134884e-01 4.47879255e-01
1.80667150e+00 6.98839605e-01 1.01653981e+00 2.34750852e-01
7.74985850e-01 9.55884084e-02 3.68054569e-01 2.00132385e-01
2.90569007e-01 2.11367652e-01 5.64707875e-01 -1.51558638e-01
4.66467291e-02 -1.04481913e-01 2.57542610e-01 1.07731104e+00
3.58654535e-03 -2.97861874e-01 -6.10340357e-01 8.18745673e-01
-1.75749457e+00 -5.62208712e-01 1.56936958e-01 1.82150996e+00
6.40461385e-01 -6.19233400e-02 -5.23401201e-01 1.80574492e-01
6.84793234e-01 -3.83569926e-01 -7.71694183e-01 -2.30949633e-02
-2.58646339e-01 1.54009372e-01 3.26778710e-01 -3.83838117e-01
-8.99431348e-01 4.12341893e-01 5.19646358e+00 1.28537154e+00
-1.18393111e+00 2.75376946e-01 4.87237304e-01 2.60817170e-01
6.16630912e-02 -2.84720838e-01 -7.47574091e-01 8.40580344e-01
5.83852589e-01 3.00593913e-01 3.46811980e-01 9.60881650e-01
-2.47387365e-01 -3.12200874e-01 -9.91815925e-01 1.42473853e+00
2.31127635e-01 -1.40722203e+00 -8.80893990e-02 -6.24494590e-02
9.85082746e-01 1.31807581e-01 -1.72191992e-01 5.33027768e-01
-4.60158318e-01 -9.80108082e-01 1.23810038e-01 6.30248606e-01
7.85220206e-01 -6.17120922e-01 1.18948090e+00 4.45159614e-01
-1.25072610e+00 -9.17472124e-01 -8.28162491e-01 1.24188177e-01
-1.90396369e-01 8.92131805e-01 -5.25775194e-01 8.78023922e-01
1.13360918e+00 8.11954975e-01 -5.99219024e-01 1.57322514e+00
1.78193431e-02 4.31210011e-01 -2.64527023e-01 -4.71170582e-02
-2.23433554e-01 -2.22096071e-01 3.95329893e-02 8.63788009e-01
1.11880088e+00 -2.02145532e-01 6.26387447e-02 9.12717938e-01
-4.23866302e-01 -3.99167202e-02 -6.76147580e-01 -1.31170973e-01
4.39876288e-01 1.55014312e+00 -3.34196478e-01 -2.76560128e-01
-4.16988462e-01 9.06236708e-01 2.69320190e-01 1.79594934e-01
-5.18792212e-01 -5.00853419e-01 3.61034572e-01 -2.78939307e-01
8.09562266e-01 1.75337583e-01 -2.83993900e-01 -9.17576909e-01
2.74529368e-01 -6.07373595e-01 2.87711471e-01 -1.21638370e+00
-1.73238683e+00 4.27580684e-01 -5.31178713e-01 -1.07094944e+00
5.48435450e-01 -1.00216711e+00 -1.00657952e+00 9.10825133e-01
-1.93949604e+00 -1.37108898e+00 -8.21237981e-01 3.99320930e-01
6.80457830e-01 -4.91909266e-01 8.80543768e-01 5.27580321e-01
-9.26970422e-01 2.81177104e-01 3.11761230e-01 -1.12773620e-01
3.80103588e-01 -9.56856072e-01 -1.80939198e-01 4.95756805e-01
-1.99276164e-01 -1.99882779e-02 -2.52219345e-02 -7.41734385e-01
-1.43901694e+00 -1.27664030e+00 5.37553549e-01 -6.91003054e-02
2.93693453e-01 -7.41745904e-02 -9.76264954e-01 1.98650792e-01
4.40451056e-01 1.71760410e-01 2.39307910e-01 -1.55980915e-01
2.84085661e-01 -6.67796314e-01 -1.42518544e+00 8.35020244e-02
8.47189069e-01 -4.58333760e-01 -3.03933084e-01 4.86170262e-01
5.18446684e-01 -2.96325028e-01 -1.11264193e+00 1.05585408e+00
1.89862609e-01 -5.51968575e-01 7.91103423e-01 5.41339405e-02
7.80379474e-02 -5.22545457e-01 1.44211486e-01 -1.31791675e+00
-3.95426393e-01 -7.13294512e-03 -4.35284138e-01 1.37818825e+00
1.03550941e-01 -9.12943006e-01 7.98792124e-01 8.21049139e-02
-6.13352239e-01 -1.31189823e+00 -1.09483051e+00 -7.59300947e-01
-1.94236532e-01 4.15480472e-02 6.20388389e-01 7.27888048e-01
-3.26392502e-01 2.78666139e-01 8.29790346e-03 4.29993868e-01
1.60888836e-01 2.94544369e-01 3.03498805e-01 -1.69117045e+00
1.54660851e-01 -2.35939354e-01 -5.97360313e-01 -1.05962980e+00
1.01051033e-01 -7.63092995e-01 1.03812218e-01 -1.83778644e+00
3.36456567e-01 -6.83315217e-01 -6.89277649e-01 6.56861961e-01
1.31372079e-01 2.95114845e-01 -1.54766783e-01 1.58616841e-01
-3.28003317e-01 1.19563830e+00 1.33610451e+00 -6.13570929e-01
-6.10784963e-02 -5.84172979e-02 -5.59336424e-01 5.11975527e-01
8.34473848e-01 -2.67213613e-01 -2.78171062e-01 -4.35473710e-01
3.32078114e-02 -3.04593444e-01 6.55478716e-01 -1.44600737e+00
6.06683493e-01 3.39758486e-01 8.84123087e-01 -8.23666751e-01
2.58260518e-01 -1.18281615e+00 -3.34535055e-02 6.30096912e-01
6.21422708e-01 -1.51396856e-01 4.24457818e-01 6.00307524e-01
-2.18454793e-01 -6.06257260e-01 6.29950225e-01 -5.55518754e-02
-8.56161296e-01 4.16681707e-01 -2.13224933e-01 -6.01132870e-01
1.06749260e+00 -3.63796979e-01 -7.89770663e-01 3.14410627e-01
-1.71553358e-01 5.43675303e-01 2.54879296e-01 3.52561921e-01
7.89275646e-01 -1.51298487e+00 -1.83883220e-01 5.47447085e-01
2.29735479e-01 2.81824827e-01 6.33188844e-01 9.73347545e-01
-4.33310837e-01 3.71615022e-01 -5.33333838e-01 -7.61973202e-01
-1.10935378e+00 4.85357314e-01 5.37772179e-01 -1.42325625e-01
-6.34535789e-01 5.81014454e-01 4.99718674e-02 -3.06917608e-01
1.34713635e-01 -5.28870225e-01 -4.28537577e-01 -2.08825454e-01
-4.51211073e-02 4.74333912e-01 4.12119478e-01 -2.22587258e-01
-2.26161495e-01 8.19296002e-01 3.58330578e-01 9.57439423e-01
1.36411524e+00 5.05595878e-02 -3.70358199e-01 4.51740846e-02
1.03369701e+00 -3.98064345e-01 -1.22847164e+00 2.00192649e-02
-2.52839983e-01 -2.45440379e-01 1.41063750e-01 -1.02274942e+00
-1.45461690e+00 9.57243085e-01 1.30414021e+00 1.21977918e-01
1.55011666e+00 -2.42847249e-01 9.13025320e-01 8.30573976e-01
4.71280962e-01 -1.39344895e+00 3.25761735e-01 2.49779895e-01
8.71176243e-01 -1.16827357e+00 -8.36388618e-02 -4.06717867e-01
-3.12157750e-01 1.08528566e+00 1.19266975e+00 7.99505785e-02
6.82073116e-01 1.95545733e-01 -1.01786703e-01 -3.81185919e-01
-5.12297869e-01 2.73987781e-02 1.56608239e-01 8.44746232e-01
-1.82085633e-01 -3.21494728e-01 -3.44771259e-02 8.05473626e-01
5.13379395e-01 2.30436638e-01 -4.82342355e-02 9.31410789e-01
-6.66102529e-01 -9.43549752e-01 -1.69800296e-01 8.59996080e-01
-1.79149747e-01 4.06772681e-02 1.97913051e-01 3.39232296e-01
7.43005931e-01 8.83483708e-01 3.58574395e-03 -2.27006748e-01
5.67822158e-01 1.22155689e-01 5.58758914e-01 -3.17968249e-01
-3.43710721e-01 -1.85902998e-01 -1.86748460e-01 -9.11778286e-02
-4.29771006e-01 -4.41046283e-02 -1.07880759e+00 1.56932548e-01
-9.98652160e-01 7.60836527e-02 6.83281302e-01 1.04811168e+00
7.03663647e-01 8.81463945e-01 7.07111120e-01 -8.30423772e-01
-4.39361572e-01 -1.43083382e+00 -6.98432446e-01 4.17762175e-02
-1.08574398e-01 -8.43005955e-01 -9.19440538e-02 -2.07549945e-01] | [7.337219715118408, 1.8341957330703735] |
cfb5f781-df1e-40a0-91da-3211f83826e2 | iteratively-coupled-multiple-instance | 2303.15749 | null | https://arxiv.org/abs/2303.15749v1 | https://arxiv.org/pdf/2303.15749v1.pdf | Iteratively Coupled Multiple Instance Learning from Instance to Bag Classifier for Whole Slide Image Classification | Whole Slide Image (WSI) classification remains a challenge due to their extremely high resolution and the absence of fine-grained labels. Presently, WSIs are usually classified as a Multiple Instance Learning (MIL) problem when only slide-level labels are available. MIL methods involve a patch embedding process and a bag-level classification process, but they are prohibitively expensive to be trained end-to-end. Therefore, existing methods usually train them separately, or directly skip the training of the embedder. Such schemes hinder the patch embedder's access to slide-level labels, resulting in inconsistencies within the entire MIL pipeline. To overcome this issue, we propose a novel framework called Iteratively Coupled MIL (ICMIL), which bridges the loss back-propagation process from the bag-level classifier to the patch embedder. In ICMIL, we use category information in the bag-level classifier to guide the patch-level fine-tuning of the patch feature extractor. The refined embedder then generates better instance representations for achieving a more accurate bag-level classifier. By coupling the patch embedder and bag classifier at a low cost, our proposed framework enables information exchange between the two processes, benefiting the entire MIL classification model. We tested our framework on two datasets using three different backbones, and our experimental results demonstrate consistent performance improvements over state-of-the-art MIL methods. Code will be made available upon acceptance. | ['Hao Chen', 'Lanfen Lin', 'Hongjie Hu', 'Yen-Wei Chen', 'Ruofeng Tong', 'Fang Wang', 'Luyang Luo', 'Hongyi Wang'] | 2023-03-28 | null | null | null | null | ['multiple-instance-learning'] | ['methodology'] | [ 4.37201440e-01 -6.33961111e-02 -4.55843925e-01 -5.73845029e-01
-1.51620746e+00 -5.77400804e-01 4.55373019e-01 4.90249485e-01
-1.79897264e-01 6.14483714e-01 -9.06981379e-02 -2.28245035e-02
5.85193224e-02 -9.72675681e-01 -7.40351260e-01 -9.61210549e-01
2.01819390e-01 3.77822429e-01 4.24088567e-01 2.71122932e-01
2.95284450e-01 2.16429442e-01 -1.68039405e+00 7.94071317e-01
9.04174387e-01 1.37688613e+00 -1.39815554e-01 4.90527838e-01
-4.09625351e-01 7.05887675e-01 -3.96138608e-01 -3.15831721e-01
1.51755840e-01 -2.61574507e-01 -6.20507300e-01 1.89368606e-01
8.20553243e-01 -1.40930921e-01 1.01967081e-01 7.57428527e-01
2.77573496e-01 -2.81620286e-02 6.96148753e-01 -1.21622300e+00
-2.08545431e-01 4.06745106e-01 -7.43733466e-01 -1.42085627e-01
-6.05419986e-02 1.04820646e-01 1.38357174e+00 -9.77839589e-01
4.22728896e-01 1.00144136e+00 7.41621375e-01 2.90013462e-01
-1.26726043e+00 -8.21515858e-01 3.71256948e-01 1.53382435e-01
-1.52534497e+00 -4.27055180e-01 8.22960436e-01 -4.76861030e-01
7.45257616e-01 4.35892642e-01 6.29503310e-01 4.54287350e-01
1.29871041e-01 1.17017341e+00 1.19261599e+00 -3.56164366e-01
2.44915232e-01 2.77445555e-01 4.55989212e-01 9.00734901e-01
1.09184630e-01 -2.71871984e-01 -6.20794833e-01 -6.81471825e-02
6.18817210e-01 2.41344675e-01 6.46125749e-02 -3.65370780e-01
-1.00255883e+00 7.98691571e-01 8.18510771e-01 3.14659715e-01
-1.60612628e-01 2.12937757e-01 3.10378075e-01 1.66469902e-01
7.42884994e-01 2.92225480e-01 -2.01339751e-01 1.77070215e-01
-1.21407080e+00 2.24485144e-01 5.45511961e-01 7.66210139e-01
1.09295583e+00 -6.49813354e-01 -5.17038405e-01 9.07661438e-01
2.33464435e-01 1.92136895e-02 2.35507533e-01 -3.02193463e-01
6.57402635e-01 1.16501439e+00 -1.13842368e-01 -1.08304095e+00
-2.28510290e-01 -8.60130966e-01 -7.22350597e-01 -3.73978946e-05
3.63579154e-01 4.31772947e-01 -8.67167234e-01 1.32728767e+00
5.93293369e-01 4.95840371e-01 -2.56738216e-01 8.07240903e-01
8.32629442e-01 6.19292736e-01 1.58030182e-01 1.18373618e-01
1.37152421e+00 -1.35657620e+00 -3.94577831e-01 -2.84343958e-01
9.04725254e-01 -5.14884770e-01 1.09562850e+00 2.85946429e-01
-9.43278790e-01 -5.41216910e-01 -1.25910330e+00 -2.14264780e-01
-5.47403812e-01 3.93532693e-01 5.22768259e-01 3.98321867e-01
-8.22112024e-01 4.43175942e-01 -8.54827821e-01 -2.67653596e-02
8.21331620e-01 4.13070470e-01 -3.80293846e-01 -2.95317560e-01
-9.36797380e-01 4.62047428e-01 1.11758113e-01 2.70065784e-01
-7.66447127e-01 -8.76818419e-01 -9.45219338e-01 1.97457090e-01
3.55922759e-01 -3.05725098e-01 9.36498225e-01 -6.63492501e-01
-1.26450670e+00 1.00212097e+00 -2.34458476e-01 -5.25478497e-02
3.50584626e-01 5.23282588e-02 7.70634338e-02 3.47638577e-02
1.53781042e-01 5.98356366e-01 8.84023905e-01 -1.29121459e+00
-1.02063429e+00 -4.38509971e-01 1.71290770e-01 1.51393220e-01
-4.57273275e-01 -5.29700756e-01 -6.01076961e-01 -4.89258617e-01
3.11209649e-01 -7.58379161e-01 1.08365349e-01 3.63039710e-02
-4.91322815e-01 -4.34354275e-01 6.30317032e-01 -5.30155838e-01
1.48817587e+00 -2.29793859e+00 1.05237991e-01 1.19377822e-01
2.84094393e-01 2.51117647e-01 -2.73455292e-01 2.03101709e-01
8.96940008e-02 1.23964317e-01 -2.68020570e-01 -7.05932856e-01
5.54444455e-02 5.09702638e-02 -4.52291891e-02 6.71480536e-01
5.15940547e-01 9.70092118e-01 -9.12507772e-01 -7.28241622e-01
1.46595627e-01 4.02165830e-01 -6.83817327e-01 4.04821336e-01
-1.21477768e-01 4.06233162e-01 -3.62631351e-01 9.70531821e-01
6.62927032e-01 -5.52220404e-01 -5.78036308e-02 -5.89178979e-01
-5.34219295e-03 3.66411507e-01 -1.25284469e+00 1.72293150e+00
-5.89319468e-01 2.32743099e-01 -8.39787498e-02 -1.26777625e+00
9.22050476e-01 5.34833558e-02 3.05529535e-01 -5.13943553e-01
-1.36922866e-01 2.82353938e-01 -2.83047944e-01 -1.74741656e-01
1.49435163e-01 -1.82943135e-01 -3.13532531e-01 2.56923527e-01
-1.61385983e-02 6.69076443e-02 2.69856930e-01 4.44589891e-02
1.06485319e+00 1.20075502e-01 1.89280689e-01 1.26304096e-02
7.12544262e-01 9.05875266e-02 6.90820992e-01 7.12788224e-01
-3.56912240e-02 4.93644953e-01 3.90015900e-01 -4.78928655e-01
-8.42770517e-01 -9.95439351e-01 -4.16154444e-01 1.25310814e+00
2.63007998e-01 -3.63754123e-01 -6.81460559e-01 -1.04851007e+00
6.50922507e-02 2.98773617e-01 -7.85049081e-01 -1.02047101e-01
-4.52075988e-01 -7.79272139e-01 4.05409664e-01 5.92796385e-01
5.23109913e-01 -7.35909879e-01 -2.42700964e-01 3.75573814e-01
-1.44651830e-01 -8.45038116e-01 -4.82294202e-01 1.50930226e-01
-8.12634349e-01 -9.58183944e-01 -4.47945356e-01 -7.53844440e-01
9.41387534e-01 2.05378264e-01 1.02810490e+00 4.27732944e-01
-5.95277488e-01 -1.65079869e-02 -3.86366278e-01 2.04801280e-02
-1.95658565e-01 4.59295243e-01 -4.56145078e-01 3.87223989e-01
3.13155562e-01 -2.62596697e-01 -8.33145380e-01 4.31374580e-01
-9.64945972e-01 2.91928351e-01 7.41179168e-01 1.15842950e+00
9.65337396e-01 6.49397597e-02 6.29658401e-01 -1.02260077e+00
9.21008512e-02 -3.43664855e-01 -6.39474392e-01 4.77359921e-01
-5.83791673e-01 -8.93007964e-02 4.36060339e-01 -2.86690950e-01
-9.42830980e-01 1.51194423e-01 -1.41406387e-01 -1.40026152e-01
-2.47492045e-02 5.19225955e-01 -2.93761790e-01 -1.67000219e-01
3.15638721e-01 2.18329966e-01 -3.52415927e-02 -4.46207762e-01
2.71452755e-01 8.73618603e-01 2.44357020e-01 -4.52657461e-01
7.03246593e-01 3.56121391e-01 -1.49960086e-01 -4.52762365e-01
-1.26971209e+00 -7.70357966e-01 -6.95684493e-01 -1.86502248e-01
6.53799713e-01 -1.06196189e+00 -4.97381389e-01 5.61154664e-01
-7.19167531e-01 -3.88289452e-01 -1.67396933e-01 -2.18790453e-02
-3.19324464e-01 -7.28575587e-02 -7.07133591e-01 -4.13258255e-01
-3.23753089e-01 -1.50716507e+00 1.68976986e+00 3.28740895e-01
-2.06621410e-03 -1.00779176e+00 -6.49428740e-02 7.23176181e-01
3.58500510e-01 2.37259671e-01 1.01076031e+00 -5.89488626e-01
-9.11258876e-01 -7.26586282e-01 -5.20690322e-01 2.92891562e-01
2.43599653e-01 -1.55429602e-01 -1.29678130e+00 -4.63964969e-01
-3.99040967e-01 -4.86992329e-01 1.06600535e+00 2.48044237e-01
1.36639273e+00 -3.46946269e-01 -5.35954535e-01 6.86370254e-01
1.75371516e+00 -1.79826960e-01 5.82135022e-01 2.83088714e-01
8.48841548e-01 5.27468860e-01 9.11483586e-01 2.04249591e-01
4.96945351e-01 7.04791844e-01 2.38983810e-01 -2.63175070e-01
-3.03173214e-01 -2.54883349e-01 1.35489866e-01 8.29142869e-01
3.20549041e-01 -1.72924876e-01 -7.41232276e-01 6.00784540e-01
-1.68488431e+00 -6.25813246e-01 1.16794176e-01 2.27647352e+00
1.11862552e+00 4.85508665e-02 -6.36471361e-02 4.17645156e-01
4.34879750e-01 2.24176958e-01 -5.35966575e-01 -6.17257543e-02
3.88323337e-01 -8.29442218e-03 2.51783580e-01 6.77579701e-01
-1.37552428e+00 9.62605119e-01 4.90368748e+00 1.24780858e+00
-1.18381727e+00 2.34331548e-01 9.59872961e-01 -1.65208578e-01
-1.25175565e-01 -7.47944117e-02 -1.33517861e+00 4.89266515e-01
4.61024463e-01 3.13838869e-01 1.87529519e-01 8.21014822e-01
-3.10214460e-01 -2.50991344e-01 -1.39554989e+00 8.28763783e-01
-3.78336087e-02 -1.60854566e+00 -1.23866148e-01 3.65611091e-02
5.01048982e-01 -2.98957199e-01 3.72364596e-02 4.92941946e-01
-1.13186792e-01 -9.05187666e-01 4.49814409e-01 4.03574795e-01
9.05969560e-01 -7.10078418e-01 7.94513345e-01 2.86837518e-01
-1.55076015e+00 -6.23483560e-04 -2.47410312e-01 1.77396357e-01
-2.74184078e-01 8.71493638e-01 -1.01943612e+00 5.22410333e-01
5.40681839e-01 7.49515891e-01 -6.91507101e-01 9.49768603e-01
-1.57112464e-01 7.06953108e-01 -2.53587723e-01 1.28384000e-02
4.29858357e-01 -1.69941336e-02 1.56190291e-01 1.41905284e+00
7.59306923e-02 -9.53119695e-02 5.15944660e-01 7.50613809e-01
-1.18477777e-01 1.13327317e-01 -1.53464347e-01 -1.92689598e-02
5.01959562e-01 1.62655079e+00 -8.73802483e-01 -3.96698803e-01
-3.81149501e-01 8.45989108e-01 6.51687205e-01 8.39644447e-02
-6.53986514e-01 -3.41693223e-01 4.81230319e-01 1.25968099e-01
3.15776587e-01 2.46685565e-01 -5.12756288e-01 -1.07074273e+00
8.98928866e-02 -6.99052215e-01 5.18684387e-01 -1.92124709e-01
-1.50975406e+00 4.52429652e-01 -1.98961064e-01 -1.38891006e+00
1.03864539e-02 -4.90669847e-01 -4.41325724e-01 7.40675092e-01
-1.86755919e+00 -1.61589408e+00 -5.62609494e-01 2.54456460e-01
5.09711266e-01 1.74922615e-01 7.76265800e-01 4.36583430e-01
-8.29383969e-01 1.10505366e+00 3.34964283e-02 1.92358822e-01
7.67069280e-01 -1.26999438e+00 -1.30428836e-01 4.83585030e-01
3.30018625e-02 4.52399015e-01 2.22448111e-01 -4.64548290e-01
-1.34509325e+00 -1.33948803e+00 8.31561029e-01 -3.37492615e-01
5.26768863e-01 -6.42074525e-01 -1.10960484e+00 4.93920773e-01
-3.69368702e-01 3.51956218e-01 1.05006218e+00 2.09790185e-01
-4.07542348e-01 -5.96003532e-01 -1.21371412e+00 1.42972484e-01
6.76939666e-01 -4.43771005e-01 -2.48202160e-01 3.47930342e-01
5.17269373e-01 -3.69358331e-01 -1.15525138e+00 3.47292989e-01
6.07225239e-01 -6.69548512e-01 9.87398446e-01 -2.28202343e-01
4.92032260e-01 -4.93672311e-01 -1.50192738e-01 -1.16108513e+00
-4.21019256e-01 8.67191851e-02 -4.09144126e-02 1.59116936e+00
5.71440279e-01 -6.43809617e-01 7.80289829e-01 4.83617455e-01
-1.57096252e-01 -1.29691732e+00 -8.60175848e-01 -5.55212200e-01
-7.86173865e-02 -1.30506262e-01 7.52260327e-01 8.17022622e-01
2.56356671e-02 2.44454086e-01 -1.76824063e-01 3.29598188e-01
8.20058227e-01 5.22165298e-01 7.75040329e-01 -1.22648752e+00
-4.71767426e-01 -4.09902364e-01 -5.15079260e-01 -8.13657343e-01
4.84625474e-02 -1.23743618e+00 4.22672689e-01 -1.47547483e+00
5.33093512e-01 -1.17098224e+00 -4.13442105e-01 8.65235031e-01
-6.24358058e-01 6.03733361e-01 4.79423776e-02 3.42340231e-01
-8.14466119e-01 3.82498920e-01 1.20471740e+00 -3.87885541e-01
-1.39798179e-01 -2.27659285e-01 -7.14142621e-01 5.52722216e-01
5.25973678e-01 -5.90272605e-01 -3.34432036e-01 -3.48883450e-01
6.88497797e-02 -1.54033720e-01 3.13065112e-01 -1.09574139e+00
3.74296486e-01 4.22026822e-03 4.58675504e-01 -6.92980051e-01
3.16209048e-01 -6.96548820e-01 -7.35100582e-02 1.02471851e-01
-5.25730729e-01 -6.53955817e-01 1.29880741e-01 5.45489371e-01
-4.06691641e-01 -1.77639037e-01 1.00649869e+00 1.46674976e-01
-4.79891032e-01 6.66549206e-01 9.45592895e-02 -1.84814557e-01
1.12765265e+00 -2.06180856e-01 -2.93353289e-01 1.95232794e-01
-5.18935859e-01 3.73803526e-01 4.91123497e-01 1.55041933e-01
4.57195550e-01 -1.29162169e+00 -6.20429039e-01 4.30733383e-01
3.87054503e-01 5.57516575e-01 3.09060603e-01 1.06983113e+00
-2.89486319e-01 2.58729547e-01 2.43785575e-01 -9.78036642e-01
-1.11679614e+00 3.58742416e-01 2.80628085e-01 -8.78102779e-01
-4.80977267e-01 1.12773347e+00 4.31091189e-01 -5.58132052e-01
4.52383906e-01 -3.23826253e-01 -1.09218627e-01 3.03048104e-01
7.82007158e-01 2.07423434e-01 3.01106542e-01 -3.68320227e-01
-4.31504667e-01 6.64007246e-01 -4.49976295e-01 3.05712223e-01
1.35811329e+00 -8.74551982e-02 -2.47699723e-01 5.52166343e-01
1.48240709e+00 -3.16943884e-01 -1.32234335e+00 -3.88342530e-01
-7.61723444e-02 -5.60422242e-01 2.84344077e-01 -7.56538153e-01
-1.04251254e+00 1.00303495e+00 6.60103500e-01 -1.88956689e-02
1.25512362e+00 8.74828398e-02 9.85949218e-01 7.35355094e-02
3.21038783e-01 -9.30297434e-01 1.15180507e-01 2.79005896e-02
6.83234990e-01 -1.36170149e+00 6.31867871e-02 -6.49028659e-01
-3.51038039e-01 8.99467826e-01 7.14627266e-01 -2.25311935e-01
5.04629135e-01 4.46002036e-01 -8.68988335e-02 -1.42529577e-01
-7.33921230e-01 1.87941771e-02 5.21668971e-01 1.32298946e-01
4.36399251e-01 1.08725145e-01 -9.71987173e-02 7.59968638e-01
3.28569859e-01 1.16082966e-01 -9.15730074e-02 9.58822608e-01
-4.28831309e-01 -1.20317948e+00 -3.72490168e-01 8.16723645e-01
-2.04928622e-01 -4.71110083e-02 -1.26563311e-01 5.12978852e-01
2.64491856e-01 8.57523441e-01 2.45861515e-01 -4.01796758e-01
1.95257694e-01 1.71364799e-01 3.90337735e-01 -7.63313591e-01
-5.64394653e-01 1.24545880e-01 3.76970391e-03 -7.09363043e-01
-3.88171345e-01 -5.48714638e-01 -1.22180033e+00 -9.68144238e-02
-5.41319191e-01 6.60359189e-02 5.14101684e-01 1.13747668e+00
3.54517728e-01 5.75070918e-01 7.85756648e-01 -9.73674715e-01
-5.03887355e-01 -9.16663349e-01 -5.57146311e-01 3.55532199e-01
5.08951366e-01 -7.32817352e-01 -4.35208201e-01 9.16957855e-02] | [15.09524917602539, -2.7476806640625] |
fa299140-0ee4-45ea-94db-2e5e05b43584 | interactive-removal-and-ground-truth-for | 1608.00762 | null | http://arxiv.org/abs/1608.00762v1 | http://arxiv.org/pdf/1608.00762v1.pdf | Interactive Removal and Ground Truth for Difficult Shadow Scenes | A user-centric method for fast, interactive, robust and high-quality shadow
removal is presented. Our algorithm can perform detection and removal in a
range of difficult cases: such as highly textured and colored shadows. To
perform detection an on-the-fly learning approach is adopted guided by two
rough user inputs for the pixels of the shadow and the lit area. After
detection, shadow removal is performed by registering the penumbra to a
normalized frame which allows us efficient estimation of non-uniform shadow
illumination changes, resulting in accurate and robust removal. Another major
contribution of this work is the first validated and multi-scene category
ground truth for shadow removal algorithms. This data set containing 186 images
eliminates inconsistencies between shadow and shadow-free images and provides a
range of different shadow types such as soft, textured, colored and broken
shadow. Using this data, the most thorough comparison of state-of-the-art
shadow removal methods to date is performed, showing our proposed new algorithm
to outperform the state-of-the-art across several measures and shadow category.
To complement our dataset, an online shadow removal benchmark website is also
presented to encourage future open comparisons in this challenging field of
research. | ['Han Gong', 'Darren P. Cosker'] | 2016-08-02 | null | null | null | null | ['shadow-removal'] | ['computer-vision'] | [ 8.59075189e-01 -2.26960167e-01 4.33357120e-01 -3.92069280e-01
-4.45509881e-01 -5.79111040e-01 5.17217875e-01 -8.91917348e-02
-2.38362849e-01 8.83284628e-01 2.81475447e-02 -3.86218816e-01
2.44914427e-01 -3.76037657e-01 -2.87836730e-01 -9.34368849e-01
-8.96410272e-02 5.06206632e-01 8.67603123e-01 -3.00287336e-01
3.06947440e-01 8.68543684e-01 -1.60956895e+00 2.18828276e-01
9.14498210e-01 7.16264546e-01 6.56077623e-01 8.68621886e-01
2.08639234e-01 4.80960846e-01 -9.66173351e-01 1.55765921e-01
4.69747007e-01 -3.54710877e-01 -1.82754807e-02 7.58949965e-02
7.88334310e-01 -4.53764230e-01 -3.79318371e-02 3.98259491e-01
8.71845722e-01 2.69406974e-01 6.84048057e-01 -1.08692360e+00
5.49554676e-02 -2.82729805e-01 -5.94960451e-01 1.43988371e-01
7.16599107e-01 5.47558218e-02 4.22024399e-01 -1.14457309e+00
7.01136708e-01 1.18270087e+00 7.99733996e-01 -1.06980979e-01
-1.03877926e+00 -7.49694645e-01 -1.04392692e-01 2.75161803e-01
-1.19187307e+00 -3.26977223e-01 6.18692875e-01 -4.21888605e-02
6.02618873e-01 9.24101353e-01 7.68182218e-01 9.52259600e-01
5.83075345e-01 7.83374965e-01 1.88080919e+00 -6.70729637e-01
2.26331994e-01 2.18169600e-01 3.29384357e-02 8.04305673e-01
4.66402858e-01 2.22272784e-01 -7.65595496e-01 -3.71551424e-01
1.47535309e-01 7.86119029e-02 -6.43418252e-01 -6.46416306e-01
-1.02146697e+00 3.90868604e-01 2.55596936e-01 -7.45464638e-02
-2.18432039e-01 -6.73979595e-02 1.29441246e-01 -5.04500940e-02
3.97223175e-01 8.60290080e-02 -3.88217300e-01 2.12255418e-01
-1.34596193e+00 1.94010496e-01 1.00578594e+00 6.17961705e-01
7.55971730e-01 7.24557936e-02 -4.60327387e-01 5.07860959e-01
1.61771134e-01 1.36349070e+00 -1.15083091e-01 -7.31541097e-01
5.92367500e-02 2.28463292e-01 3.75352710e-01 -9.74296451e-01
-5.21347642e-01 -2.67315418e-01 -5.26573181e-01 8.35550845e-01
1.88354269e-01 6.24246523e-02 -1.30400693e+00 8.27601314e-01
4.54054862e-01 2.67288119e-01 8.04724321e-02 9.32207823e-01
8.86958659e-01 4.84703153e-01 -5.95823824e-01 -4.40880567e-01
1.29048038e+00 -9.03973043e-01 -9.99531746e-01 -4.06531274e-01
-6.37065545e-02 -1.33753169e+00 1.10667932e+00 6.16509080e-01
-4.34388936e-01 -1.67390078e-01 -1.19413650e+00 1.62708104e-01
-6.97689474e-01 2.15885192e-01 7.43002295e-01 7.37942278e-01
-7.45611489e-01 3.67399573e-01 -4.62924510e-01 -6.42348945e-01
4.40919906e-01 8.18374380e-02 2.88461149e-02 -3.42813969e-01
-7.37976849e-01 1.12881863e+00 -1.94883235e-02 4.32086252e-02
-1.07693052e+00 -7.47405112e-01 -5.28380871e-01 -2.83144623e-01
8.76390934e-01 -2.39735082e-01 9.93513346e-01 -9.24721479e-01
-1.24091828e+00 6.43042922e-01 -4.00930762e-01 -3.69939268e-01
8.22443247e-01 -6.72730625e-01 -2.52889544e-01 -3.67415696e-02
2.34516021e-02 -1.57323942e-01 1.26470375e+00 -2.16970181e+00
-5.07124186e-01 -3.81348692e-02 -1.71173632e-01 5.03782034e-01
1.30028397e-01 6.05719583e-03 -4.47821677e-01 -6.07202232e-01
1.06322803e-01 -1.06430876e+00 1.38955414e-01 1.31455109e-01
-6.82467282e-01 6.17870152e-01 1.66967404e+00 -8.12996924e-01
1.02286017e+00 -1.88712752e+00 -4.14845824e-01 3.34511310e-01
5.07080778e-02 2.51784831e-01 4.30915982e-01 6.04486406e-01
3.77289861e-01 -6.09276891e-01 -7.63482094e-01 -7.69679725e-01
-3.97395417e-02 4.21834469e-01 -6.34395719e-01 9.64706063e-01
-6.44297779e-01 4.20328170e-01 -9.45811808e-01 -5.76422155e-01
6.98157191e-01 6.63877606e-01 4.09971893e-01 3.31680179e-01
9.56477895e-02 2.22983420e-01 1.94591582e-02 1.28599668e+00
1.12822127e+00 4.77688462e-01 3.01871970e-02 -2.54601687e-01
-3.63948196e-01 -2.57389009e-01 -1.25987089e+00 1.08662236e+00
-6.08263493e-01 1.35853553e+00 4.28294152e-01 9.82936621e-02
6.74125314e-01 -4.82418798e-02 2.59992659e-01 -7.97182918e-01
3.20020877e-02 2.74907768e-01 -2.83782095e-01 -1.76545709e-01
7.25394189e-01 9.79280006e-03 1.52488649e-01 5.39566040e-01
-5.83817124e-01 -5.16157031e-01 -9.81501192e-02 3.40404361e-01
1.27564776e+00 2.61215091e-01 4.07864869e-01 -5.22258639e-01
3.69092733e-01 -1.21899962e-01 3.92826885e-01 1.16334856e+00
-1.90463006e-01 9.22437310e-01 -1.32735237e-01 -3.09152961e-01
-4.15066570e-01 -1.24205685e+00 -3.24677438e-01 1.32202673e+00
5.06809771e-01 -8.18646625e-02 -6.28402948e-01 -6.66504920e-01
3.52983177e-01 9.15503561e-01 -7.93245554e-01 3.75703961e-01
-5.21857142e-01 -9.91337776e-01 2.22005248e-01 7.08862692e-02
4.72705573e-01 -1.29419494e+00 -1.42650092e+00 -2.73281038e-01
-1.41686857e-01 -1.09555805e+00 -1.17440395e-01 6.17452800e-01
-4.91139829e-01 -1.30668962e+00 -5.94022155e-01 -1.68790668e-01
6.87610805e-01 1.00134432e+00 1.34298658e+00 3.19652677e-01
-9.18835640e-01 5.04446268e-01 -5.34981668e-01 -9.52670932e-01
-2.06258520e-01 -7.21768618e-01 -5.47948889e-02 6.66496158e-02
-3.94272685e-01 -3.04671973e-01 -7.67946124e-01 4.92526203e-01
-8.51475954e-01 3.32166702e-01 7.52933562e-01 5.14155030e-01
4.98997092e-01 5.10443486e-02 -3.69750172e-01 -1.13770986e+00
4.37079012e-01 -2.16138750e-01 -8.36994231e-01 2.50165164e-01
-7.19882488e-01 -4.19644594e-01 5.01262844e-01 -6.30370677e-02
-1.56283343e+00 2.40455046e-01 6.72388315e-01 -4.50850688e-02
-2.77918488e-01 -2.77251571e-01 -6.89781010e-02 -6.31934106e-01
8.27997327e-01 2.16975629e-01 -5.63927293e-01 -2.95284986e-01
2.59225458e-01 3.80751014e-01 7.05017388e-01 -4.36761640e-02
1.42065358e+00 1.29226947e+00 1.94284365e-01 -1.37652469e+00
-8.76601398e-01 -8.72158110e-01 -9.68113124e-01 -5.81229568e-01
4.63117510e-01 -4.06648517e-01 -1.72722369e-01 7.35134900e-01
-8.94053936e-01 -1.04773498e+00 2.16929819e-02 -2.20774546e-01
-2.18421429e-01 6.00415111e-01 5.61865643e-02 -1.37839448e+00
-4.15894270e-01 -7.51947224e-01 1.39178574e+00 2.60158151e-01
1.87532790e-02 -9.64284837e-01 1.26133084e-01 2.61068940e-01
5.92425227e-01 8.09543073e-01 3.03230166e-01 4.66500223e-02
-7.90626287e-01 -2.19492614e-01 -4.26636517e-01 1.54472783e-01
4.30242956e-01 -1.18231103e-01 -1.48189688e+00 -3.29213798e-01
-2.28298903e-01 -1.44959781e-02 1.20024168e+00 3.38609129e-01
8.46270502e-01 1.63876548e-01 -5.95523655e-01 6.57234788e-01
1.71042562e+00 -9.00495872e-02 7.07031846e-01 4.42683339e-01
8.25538158e-01 2.36599118e-01 1.32222641e+00 6.79297507e-01
6.77238926e-02 7.40627825e-01 6.13690615e-01 -8.09975624e-01
-6.04362130e-01 6.74398303e-01 2.08982378e-01 -6.88562840e-02
-2.19069377e-01 -6.09220684e-01 -8.81821215e-01 1.46781430e-01
-1.68585503e+00 -7.59700179e-01 -5.39774179e-01 2.45852184e+00
3.92696589e-01 2.13875815e-01 -4.43805754e-01 2.84862548e-01
1.99626327e-01 4.78883713e-01 -3.16195190e-01 -2.36536488e-01
-5.09226143e-01 2.61143059e-01 1.12684286e+00 9.38718140e-01
-1.20312560e+00 1.13270938e+00 6.08946133e+00 6.96801603e-01
-9.72984910e-01 1.41613349e-01 7.35889748e-02 1.03506908e-01
-1.38019666e-01 1.12568915e-01 -4.56306219e-01 3.69733125e-01
3.97704363e-01 5.08797228e-01 5.22658646e-01 5.21573186e-01
5.47669232e-01 -1.29736233e+00 -3.42214793e-01 9.65659261e-01
6.07108831e-01 -8.83843958e-01 -5.99767625e-01 -1.23701081e-01
9.51380014e-01 1.67137429e-01 -2.65609682e-01 7.53210410e-02
1.61699951e-01 -9.11518812e-01 7.85861015e-01 8.86688411e-01
6.89855397e-01 -4.32151198e-01 9.75778401e-01 2.59176977e-02
-1.30758202e+00 -8.88492465e-02 -1.10915057e-01 7.43917562e-03
1.83163211e-01 1.02270532e+00 -1.22672439e+00 6.22712553e-01
9.52503264e-01 7.13305771e-02 -8.45504642e-01 1.16222084e+00
-4.08065110e-01 4.60425675e-01 -4.32445258e-01 2.23902255e-01
-1.30032748e-01 -1.52020752e-01 6.06158257e-01 2.00769043e+00
1.67406090e-02 2.14714229e-01 5.18392563e-01 1.65617988e-01
3.16435397e-01 -1.07300915e-01 -5.80962360e-01 6.25211418e-01
3.57874125e-01 1.68521988e+00 -1.46956158e+00 -4.75886673e-01
-2.32417621e-02 1.52652180e+00 -3.25310946e-01 5.20758331e-01
-1.01872516e+00 -3.19585741e-01 3.20040107e-01 1.51216090e-01
-1.39277661e-02 -3.86914492e-01 -5.73297560e-01 -5.20786762e-01
-1.72975942e-01 -7.36369550e-01 1.20775722e-01 -9.66103911e-01
-4.95222807e-01 3.37120324e-01 1.63742602e-01 -9.65283513e-01
3.90036672e-01 -4.56853598e-01 -7.93701768e-01 7.69970179e-01
-1.78002346e+00 -1.23769701e+00 -1.42004597e+00 5.10161996e-01
7.26836562e-01 1.44699169e-02 9.12960052e-01 8.08469430e-02
-1.45085469e-01 1.66440055e-01 4.89573687e-01 -5.26900291e-01
1.24599099e+00 -1.57271969e+00 2.11036488e-01 1.07899559e+00
-1.45408317e-01 -1.73400342e-02 1.33864820e+00 -1.07416379e+00
-1.59876609e+00 -9.17490125e-01 3.96127552e-01 -6.71442151e-01
1.95866436e-01 -7.80099154e-01 -8.03171575e-01 1.18301332e-01
1.64400145e-01 8.13854262e-02 3.55163276e-01 -2.92630285e-01
1.27534524e-01 -3.50614816e-01 -1.17413771e+00 4.99376774e-01
8.28658581e-01 -2.29124725e-01 -2.33675227e-01 7.23681867e-01
7.15137050e-02 -9.79167402e-01 1.24311872e-01 6.02851748e-01
7.96151876e-01 -1.82058167e+00 1.02233803e+00 6.57416284e-01
-1.83108330e-01 -4.40692663e-01 -3.55994433e-01 -9.72615004e-01
5.70825674e-02 -7.10749984e-01 -4.87495005e-01 8.74120474e-01
2.66294498e-02 -4.22148049e-01 7.24833488e-01 -1.06169827e-01
-5.36366522e-01 -8.03599417e-01 -6.77243173e-01 -6.44905329e-01
-1.05357015e+00 -4.59282130e-01 3.87326611e-04 3.96837920e-01
-9.99231458e-01 -2.73225516e-01 -5.56817114e-01 4.97855991e-01
9.56009150e-01 6.75253212e-01 1.23928761e+00 -1.27430844e+00
-1.01649798e-01 1.29592242e-02 9.72644091e-02 -3.22383255e-01
-7.57565945e-02 -2.90733278e-01 7.68833339e-01 -1.90376842e+00
3.66410434e-01 -5.80477178e-01 -1.34245828e-01 5.24431646e-01
-2.34846398e-01 7.27417588e-01 1.82760209e-01 1.35585606e-01
-7.28215277e-01 4.11932707e-01 1.12698841e+00 2.27117345e-01
-2.71367401e-01 1.62321404e-01 -4.31293771e-02 9.44826007e-01
6.85230851e-01 -4.96220499e-01 -3.20968807e-01 -3.89597155e-02
-4.27748144e-01 -3.32320690e-01 6.32353723e-01 -1.26458597e+00
-1.29493311e-01 -3.87705833e-01 7.65656531e-01 -1.00333846e+00
9.80106175e-01 -9.24792349e-01 -1.02697037e-01 7.06571996e-01
5.70825100e-01 -3.30943316e-01 2.24586904e-01 6.52449310e-01
4.66130495e-01 1.62965328e-01 9.42089438e-01 -1.42251357e-01
-9.31824982e-01 -1.67270049e-01 -4.75358933e-01 -2.88679779e-01
1.05521393e+00 -5.47059655e-01 -3.12641501e-01 -6.54186726e-01
-1.13486454e-01 -9.72138867e-02 7.44196653e-01 1.55196413e-01
8.21209669e-01 -6.25994086e-01 -5.70019782e-01 1.33200198e-01
-7.21409842e-02 -2.89822459e-01 9.09736380e-02 9.60407555e-01
-7.51540780e-01 2.75955290e-01 -1.68539643e-01 -5.58186531e-01
-2.01327848e+00 -8.69706646e-03 2.92662710e-01 -6.54547336e-03
-9.06609952e-01 7.22717881e-01 3.82947624e-01 -2.81164140e-01
4.98350650e-01 -1.87614039e-01 1.68744847e-01 1.17962146e-02
3.37866485e-01 7.65247285e-01 2.99345911e-01 -6.76030219e-01
-6.32836759e-01 4.16125715e-01 5.41315258e-01 -1.23625942e-01
9.64416623e-01 -1.93044469e-01 -1.25152141e-01 6.23691797e-01
5.75194597e-01 7.31100142e-01 -1.55115116e+00 1.22131005e-01
-3.08130354e-01 -9.53147292e-01 1.36840314e-01 -1.40681672e+00
-6.30234301e-01 4.28411514e-01 1.28708315e+00 2.52597313e-02
1.31319261e+00 -1.79227948e-01 6.71952069e-01 4.37046230e-01
4.07417417e-01 -1.27895021e+00 1.96219966e-01 6.61121309e-01
1.14682531e+00 -1.47906768e+00 8.59520316e-01 -6.61121547e-01
-5.46597064e-01 1.03424525e+00 3.07417512e-01 -1.97138228e-02
3.11753005e-01 7.08923697e-01 5.08518279e-01 -2.78029352e-01
4.28112224e-02 -3.30965251e-01 3.13168138e-01 8.33910048e-01
3.19494382e-02 2.73412675e-01 -7.60801286e-02 -1.17874406e-01
-2.47363523e-01 -5.95056236e-01 6.87871099e-01 1.33424544e+00
-8.52785349e-01 -8.81161869e-01 -1.29139137e+00 5.96372128e-01
-2.64894962e-02 -1.49750978e-01 -8.20645452e-01 1.02512145e+00
3.38024378e-01 9.49081540e-01 -5.22701323e-01 1.35033834e-03
1.96614295e-01 -1.89638570e-01 4.82103318e-01 -4.93097663e-01
-3.56398523e-01 1.05939455e-01 1.57549724e-01 -8.07962894e-01
-2.72061139e-01 -7.67980635e-01 -1.35223567e+00 -8.27979594e-02
-4.50111896e-01 -2.93005973e-01 9.52758670e-01 9.80494559e-01
-1.56012729e-01 7.74499774e-01 4.13742244e-01 -1.70734191e+00
2.33591214e-01 -7.81152964e-01 -6.43319547e-01 2.47782052e-01
7.43788719e-01 -1.08772433e+00 -4.83881235e-01 -4.33849506e-02] | [10.816753387451172, -4.068728923797607] |
4bb00577-a473-45ac-8597-8aef1e861c09 | tensor-variable-elimination-for-plated-factor | 1902.03210 | null | https://arxiv.org/abs/1902.03210v2 | https://arxiv.org/pdf/1902.03210v2.pdf | Tensor Variable Elimination for Plated Factor Graphs | A wide class of machine learning algorithms can be reduced to variable elimination on factor graphs. While factor graphs provide a unifying notation for these algorithms, they do not provide a compact way to express repeated structure when compared to plate diagrams for directed graphical models. To exploit efficient tensor algebra in graphs with plates of variables, we generalize undirected factor graphs to plated factor graphs and variable elimination to a tensor variable elimination algorithm that operates directly on plated factor graphs. Moreover, we generalize complexity bounds based on treewidth and characterize the class of plated factor graphs for which inference is tractable. As an application, we integrate tensor variable elimination into the Pyro probabilistic programming language to enable exact inference in discrete latent variable models with repeated structure. We validate our methods with experiments on both directed and undirected graphical models, including applications to polyphonic music modeling, animal movement modeling, and latent sentiment analysis. | ['Martin Jankowiak', 'Justin Chiu', 'Alexander Rush', 'Neeraj Pradhan', 'Eli Bingham', 'Noah Goodman', 'Fritz Obermeyer'] | 2019-02-08 | null | null | null | null | ['music-modeling'] | ['music'] | [ 1.05401553e-01 2.36732721e-01 -3.37063640e-01 -1.76674396e-01
-2.31596455e-01 -9.77289498e-01 5.53070366e-01 -2.02638850e-01
2.95632601e-01 3.80635589e-01 -1.37156770e-01 -8.13306391e-01
-5.60060441e-01 -9.47389960e-01 -7.56075084e-01 -6.10996068e-01
-7.26362109e-01 7.39511430e-01 -3.00512969e-01 6.04775473e-02
-2.42062703e-01 4.83779818e-01 -9.93536174e-01 1.78949267e-01
1.88907579e-01 4.63221341e-01 -4.95719165e-01 9.09353316e-01
-4.16417152e-01 6.39798760e-01 -2.39898875e-01 -7.89204359e-01
1.79381862e-01 -1.23996645e-01 -8.51203084e-01 3.48469645e-01
4.14803714e-01 -2.65860911e-02 -3.27128679e-01 8.60638797e-01
-1.42430291e-01 -1.10169187e-01 9.45656121e-01 -1.80001330e+00
-6.75245702e-01 1.35499060e+00 -1.01642263e+00 -1.04114033e-01
4.10565406e-01 -6.27270877e-01 1.61061072e+00 -6.73375070e-01
4.71625328e-01 1.74390507e+00 7.90392518e-01 1.18630178e-01
-2.04385018e+00 -5.27404010e-01 3.11749667e-01 -1.44774318e-01
-1.16341698e+00 1.16411678e-01 8.64173055e-01 -9.08925593e-01
6.78350151e-01 5.63030541e-01 6.40174270e-01 8.93910766e-01
1.50818557e-01 9.87985313e-01 8.46037507e-01 -6.03954613e-01
2.35567018e-01 -2.42768571e-01 8.02852094e-01 1.29401720e+00
3.93134117e-01 -3.85091677e-02 -6.42583549e-01 -8.10481906e-01
7.27926135e-01 1.07494935e-01 1.02176741e-01 -7.04991996e-01
-1.12828577e+00 1.37177336e+00 -4.80309278e-02 -1.48713276e-01
4.00417037e-02 7.85504937e-01 3.16279680e-01 3.06394517e-01
3.34865570e-01 1.42762706e-01 -5.03165603e-01 2.14526176e-01
-8.71196985e-01 3.92624766e-01 1.20377243e+00 1.16093445e+00
9.67589736e-01 2.71345437e-01 8.46532658e-02 6.12573922e-01
4.84208018e-01 8.52634549e-01 -3.95042449e-01 -9.93824959e-01
2.09120423e-01 3.88673574e-01 -3.29781473e-01 -1.34002948e+00
-4.14154232e-01 -5.31211138e-01 -7.81310976e-01 -3.01660951e-02
5.76009631e-01 2.10949481e-02 -8.67822766e-01 1.65225899e+00
1.12085715e-01 -1.47769317e-01 -5.94118536e-01 3.84467483e-01
1.77341223e-01 4.23478842e-01 -2.03448132e-01 -1.84256509e-01
1.27274096e+00 -8.84860694e-01 -7.74251997e-01 -5.02465069e-02
7.76196659e-01 -4.50798690e-01 5.95812380e-01 8.01398396e-01
-6.84499323e-01 6.33506775e-02 -8.96711767e-01 -2.47891918e-01
-1.37621820e-01 4.69619930e-02 1.61463761e+00 1.00761914e+00
-9.78263617e-01 5.34187913e-01 -1.04008830e+00 2.35941112e-01
1.66519970e-01 4.05059218e-01 -4.56744790e-01 -5.79058826e-02
-8.87432694e-01 1.67920917e-01 -1.96395919e-01 1.21268265e-01
-9.11006570e-01 -6.61198497e-01 -9.65526044e-01 -5.77972643e-02
3.44885528e-01 -9.80181038e-01 1.30348325e+00 -7.81091988e-01
-1.11135852e+00 4.23329026e-01 -3.77325863e-01 -3.79573107e-01
1.95988327e-01 4.87893671e-02 1.20418809e-01 6.64513856e-02
1.24738384e-02 9.03408900e-02 1.17609823e+00 -9.51258779e-01
-1.53244480e-01 -6.29672289e-01 6.80939913e-01 -3.82473499e-01
-2.19084606e-01 -4.54474054e-02 -4.96504724e-01 -5.26651442e-01
3.55144054e-01 -1.29333448e+00 -4.74200517e-01 1.06936276e-01
-7.33482301e-01 -1.41378105e-01 6.74726069e-01 -3.69612128e-01
1.44250047e+00 -2.00802612e+00 8.05791914e-01 6.74776375e-01
8.35867405e-01 -7.28379667e-01 3.67005579e-02 3.67827326e-01
-2.84196287e-01 4.49237943e-01 -1.92326665e-01 -5.81518471e-01
3.08337569e-01 4.05604810e-01 -5.28574288e-01 6.84554875e-01
5.36636859e-02 7.18665183e-01 -6.92898810e-01 -3.62945825e-01
1.13873944e-01 3.75141650e-01 -1.02810884e+00 -4.54094857e-01
-6.43496811e-01 -3.68446618e-01 -6.63001895e-01 5.90975106e-01
7.07032979e-01 -4.33139503e-01 5.39803267e-01 -1.23462796e-01
1.70741811e-01 3.94265093e-02 -1.53963065e+00 1.55374336e+00
-4.24288779e-01 5.85323513e-01 3.05665821e-01 -7.22486198e-01
4.84776706e-01 5.94555065e-02 5.06213307e-01 3.59022141e-01
1.04757905e-01 -3.19203436e-01 -3.77850384e-01 -5.74741103e-02
5.01988709e-01 -2.24689960e-01 -4.42721725e-01 5.97900987e-01
2.16961622e-01 -1.83481202e-01 4.05270040e-01 9.51214075e-01
1.12354004e+00 5.77092618e-02 -1.20379411e-01 -2.80566007e-01
8.02360475e-02 -1.46169707e-01 2.54877180e-01 7.55569816e-01
4.60859120e-01 2.07019478e-01 1.19678104e+00 -4.26744640e-01
-7.21791089e-01 -1.41221344e+00 -1.31054953e-01 1.18536103e+00
-4.56218511e-01 -1.36293876e+00 -4.18006778e-01 -3.41337323e-01
1.24782778e-01 5.20175695e-01 -9.24057245e-01 3.66239548e-01
2.20367521e-01 -8.57551098e-01 7.46803522e-01 4.83766973e-01
-2.34286189e-01 2.29927361e-01 1.48066387e-01 5.14316224e-02
-1.42522365e-01 -8.83302331e-01 -2.86875039e-01 5.74565411e-01
-8.78813803e-01 -1.14131629e+00 -2.83708364e-01 -4.54462588e-01
4.41274792e-01 2.59065330e-01 1.24955535e+00 -1.74786910e-01
-2.79578209e-01 8.56199443e-01 6.67567551e-02 -4.01088029e-01
-1.13658972e-01 -1.30077586e-01 8.42826217e-02 -2.25515008e-01
2.39989474e-01 -6.86798155e-01 5.51378094e-02 2.80159861e-01
-8.83072674e-01 2.16212198e-01 3.96669097e-02 8.99936378e-01
5.70843458e-01 3.02108198e-01 -5.82702875e-01 -1.01082921e+00
3.11969995e-01 -5.31306863e-01 -9.17924821e-01 2.60544926e-01
-3.55014443e-01 6.22345507e-01 4.26577896e-01 -5.22189915e-01
-4.97259885e-01 3.33023667e-01 5.18733323e-01 -6.90915346e-01
7.02551603e-01 1.10403228e+00 7.64762685e-02 -3.95664312e-02
6.81214929e-01 -2.15983674e-01 -4.46379445e-02 -3.73578489e-01
7.08064914e-01 -6.17422499e-02 6.61736354e-02 -9.59460199e-01
1.16088581e+00 6.96876407e-01 9.05800760e-01 -9.18318748e-01
-4.86000597e-01 -9.26118270e-02 -7.35947430e-01 -1.28263772e-01
7.60591507e-01 -8.16839278e-01 -9.89257574e-01 2.49194905e-01
-1.14943242e+00 -1.39721185e-01 3.19170237e-01 4.27635074e-01
-5.19459724e-01 6.09509647e-01 -8.50109518e-01 -1.24979949e+00
1.76628947e-01 -9.44055855e-01 1.36256063e+00 -5.29048681e-01
-3.00784141e-01 -1.33975923e+00 4.16235179e-01 3.42960775e-01
-9.98520479e-02 1.87690437e-01 1.33028615e+00 -1.43004298e-01
-9.36707437e-01 -1.99315637e-01 1.95558205e-01 9.78904516e-02
-2.37039223e-01 6.93261564e-01 -5.83449423e-01 1.01271927e-01
-4.88167077e-01 1.19682968e-01 8.56148839e-01 4.23968256e-01
9.49884653e-01 -2.12379888e-01 -5.65363586e-01 6.63436055e-01
1.10962999e+00 -5.68189859e-01 2.50476688e-01 -9.87194180e-02
1.07504869e+00 3.80346984e-01 -1.63438201e-01 4.90932703e-01
2.81938136e-01 6.44017100e-01 2.11879343e-01 1.91825762e-01
1.89266920e-01 -3.49634171e-01 3.84354860e-01 8.75640094e-01
-1.25565469e-01 -3.31303217e-02 -8.92165959e-01 3.79572332e-01
-1.64299750e+00 -1.03205955e+00 -8.49099934e-01 1.95480800e+00
4.85534132e-01 -3.15320902e-02 4.16123658e-01 4.27749991e-01
7.07064629e-01 5.97449876e-02 -4.24111336e-02 -6.23669624e-01
-9.52422246e-02 5.03286302e-01 7.21421599e-01 7.96503007e-01
-1.09006870e+00 5.90441585e-01 7.43239117e+00 7.39696503e-01
-6.14144146e-01 1.12624861e-01 2.45458677e-01 -1.39303520e-01
-1.00130856e+00 3.79963726e-01 -6.34074330e-01 -1.30177392e-02
7.45681047e-01 -1.82510108e-01 8.37357759e-01 1.14811409e+00
-1.78214565e-01 3.16690922e-01 -1.32115757e+00 9.91749346e-01
-2.16981634e-01 -1.22755790e+00 1.22502528e-01 2.97736704e-01
5.64864874e-01 -3.05423498e-01 2.40987852e-01 3.52574974e-01
7.95909047e-01 -1.09761035e+00 6.01926208e-01 1.42881334e-01
5.15971661e-01 -7.76455581e-01 -6.23652153e-02 -6.05878681e-02
-1.33081627e+00 -7.95445964e-03 -1.73900723e-01 -4.28702354e-01
1.93688437e-01 9.15892839e-01 -6.99238241e-01 6.33977294e-01
3.34838182e-01 4.64293778e-01 -3.72100085e-01 5.33392191e-01
-3.32077801e-01 8.91478896e-01 -7.79379010e-01 -3.95536013e-02
-3.06835119e-02 -3.81018549e-01 7.34078348e-01 1.18184817e+00
4.87832427e-02 -7.44509920e-02 3.88811603e-02 1.02653480e+00
1.34885907e-01 -6.34214431e-02 -7.28067815e-01 -4.52662349e-01
1.58434406e-01 1.24409127e+00 -8.17932844e-01 4.03805263e-02
-6.46622360e-01 6.62849545e-01 2.25295320e-01 5.04789531e-01
-9.39786375e-01 -2.08944723e-01 6.88519299e-01 1.36350602e-01
4.33377206e-01 -9.28111911e-01 -2.75986761e-01 -1.49967682e+00
-1.96713731e-01 -8.11808050e-01 5.05346954e-01 -6.99909866e-01
-1.12329781e+00 3.48167062e-01 4.52431977e-01 -6.16091311e-01
-5.12215614e-01 -9.98073220e-01 -9.48675573e-02 8.24483275e-01
-6.26844466e-01 -1.38626659e+00 2.51842737e-01 8.12570274e-01
-5.01346737e-02 2.42866993e-01 1.00769198e+00 1.25542805e-01
-5.34455419e-01 5.20694435e-01 -7.34108761e-02 1.19899392e-01
-4.55544926e-02 -1.47535455e+00 3.84451210e-01 1.00315619e+00
6.90158188e-01 1.07376015e+00 9.96278226e-01 -7.68227220e-01
-2.39842916e+00 -8.06194007e-01 5.01468420e-01 -6.03603840e-01
1.34417641e+00 -6.67073607e-01 -3.04367542e-01 1.35467577e+00
-3.98159742e-01 -1.95496976e-01 8.90483558e-01 1.22011602e+00
-9.89725947e-01 1.79490492e-01 -5.25225103e-01 7.70318627e-01
1.12257111e+00 -8.45308423e-01 -2.97122508e-01 5.99206805e-01
7.81393111e-01 -2.22508743e-01 -9.76418436e-01 2.04708949e-01
8.34331393e-01 -3.43186200e-01 9.25035417e-01 -7.27311850e-01
3.38555247e-01 -6.12720013e-01 -5.71452975e-01 -9.22774494e-01
-7.88281500e-01 -9.67615008e-01 -3.76509577e-01 1.03564751e+00
7.94921696e-01 -2.48462737e-01 7.84411788e-01 8.38051140e-01
2.63855159e-01 -3.69414806e-01 -6.72325909e-01 -6.42197728e-01
8.03825557e-02 -1.16978240e+00 2.43216857e-01 9.29767191e-01
1.05777390e-01 6.36519790e-01 -4.88056958e-01 2.82562226e-01
1.05744207e+00 2.10428953e-01 9.50337529e-01 -1.30570769e+00
-1.18381655e+00 -3.89212132e-01 -8.48079324e-01 -1.08222759e+00
4.82381552e-01 -1.34194100e+00 -5.69887578e-01 -1.18207586e+00
4.80629742e-01 -4.30824071e-01 2.28241444e-01 4.85348910e-01
2.09919035e-01 2.01451257e-01 8.51763710e-02 -1.83206901e-01
-2.89854795e-01 2.38932565e-01 8.15247476e-01 -4.81606305e-01
-2.88798683e-03 1.98850796e-01 -6.86341524e-01 7.22579002e-01
1.85642332e-01 -4.99098241e-01 -6.32812858e-01 -6.10710621e-01
1.08123767e+00 3.27123374e-01 4.45683807e-01 -4.19578940e-01
5.80697805e-02 -2.24922881e-01 6.87668249e-02 -5.05531073e-01
4.59836721e-01 -7.29554534e-01 7.12460756e-01 5.13462387e-02
-4.20038670e-01 1.36027291e-01 2.77301371e-01 8.35610688e-01
1.53434604e-01 -1.18812412e-01 1.03794403e-01 2.03037202e-01
-9.44519066e-04 5.15285373e-01 -6.16949797e-01 -3.39950353e-01
5.94195008e-01 1.38908392e-03 -1.46398664e-01 -6.03340745e-01
-1.15622652e+00 7.52576143e-02 2.52470881e-01 4.54509854e-02
2.49775559e-01 -1.24275517e+00 -3.69673580e-01 1.41522855e-01
-1.05483890e-01 -2.75464922e-01 1.94074243e-01 7.82311380e-01
-4.23875123e-01 3.13098073e-01 3.31437021e-01 -7.87005246e-01
-1.59736764e+00 9.62336183e-01 6.46653995e-02 -2.10927129e-01
-2.84931511e-01 8.92658889e-01 2.79408187e-01 -4.21349227e-01
2.47179434e-01 -6.12761378e-01 4.44725960e-01 -2.48710793e-02
4.87378359e-01 3.96734476e-01 -5.98297939e-02 -3.28890562e-01
-3.76219124e-01 3.87564480e-01 2.12147385e-01 -4.65569377e-01
1.13106096e+00 -8.42124522e-02 -6.11981392e-01 8.56015801e-01
1.18455458e+00 4.89557177e-01 -6.13687992e-01 -8.16425458e-02
-2.07890078e-01 -2.14742303e-01 3.66238877e-02 -1.64595321e-01
-1.03689897e+00 8.57519686e-01 -2.73337793e-02 4.85123903e-01
7.84127295e-01 1.83500469e-01 1.72786657e-02 6.92410111e-01
5.19102871e-01 -5.50768793e-01 -3.01602334e-01 6.73264086e-01
5.43919742e-01 -3.77516687e-01 2.66017884e-01 -8.67401481e-01
-6.40079528e-02 1.34405160e+00 -1.65982559e-01 -1.15808040e-01
1.14168382e+00 8.55867565e-01 -5.18576741e-01 -3.35041732e-01
-9.61417437e-01 8.88109133e-02 2.55258977e-01 4.82756823e-01
5.00678301e-01 5.92208326e-01 -9.24057141e-02 7.58183956e-01
-6.56116068e-01 -9.07948986e-02 6.60856664e-01 5.85009873e-01
1.12103354e-02 -1.22258532e+00 -6.22826278e-01 5.90471208e-01
-7.04932868e-01 -3.94247085e-01 -3.66891295e-01 7.47506618e-01
-2.91209817e-01 1.13772595e+00 -1.03176147e-01 -6.65625155e-01
-3.47017318e-01 1.33500889e-01 9.30126607e-01 -6.61205530e-01
-3.27857137e-02 1.05684936e-01 3.80947977e-01 -3.23973507e-01
-3.17247748e-01 -7.88384497e-01 -8.64909947e-01 -7.35508144e-01
-6.97829068e-01 2.60446697e-01 9.78107750e-01 8.44662011e-01
-1.04605906e-01 4.54269230e-01 3.03947061e-01 -6.67045534e-01
-3.99787933e-01 -6.08293712e-01 -1.03988826e+00 -8.03336352e-02
1.40809819e-01 -9.11877632e-01 -7.41171181e-01 2.14598432e-01] | [7.1196770668029785, 4.783234596252441] |
340f35e7-f814-47ce-b3bf-4711deff7292 | in-domain-representation-learning-for-remote-1 | 1911.06721 | null | https://arxiv.org/abs/1911.06721v1 | https://arxiv.org/pdf/1911.06721v1.pdf | In-domain representation learning for remote sensing | Given the importance of remote sensing, surprisingly little attention has been paid to it by the representation learning community. To address it and to establish baselines and a common evaluation protocol in this domain, we provide simplified access to 5 diverse remote sensing datasets in a standardized form. Specifically, we investigate in-domain representation learning to develop generic remote sensing representations and explore which characteristics are important for a dataset to be a good source for remote sensing representation learning. The established baselines achieve state-of-the-art performance on these datasets. | ['Neil Houlsby', 'Maxim Neumann', 'Xiaohua Zhai', 'Andre Susano Pinto'] | 2019-11-15 | null | https://openreview.net/forum?id=BJx_JAVKDB | https://openreview.net/pdf?id=BJx_JAVKDB | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 7.06460476e-01 -1.69212192e-01 -4.60060358e-01 -4.73903775e-01
-1.02600467e+00 -5.93550801e-01 8.50737214e-01 1.39393643e-01
-1.41414121e-01 7.29490638e-01 4.06904668e-01 -6.37433648e-01
-5.52246511e-01 -1.17907012e+00 -2.98402369e-01 -5.63714206e-01
-4.28434044e-01 1.76679760e-01 -2.55886048e-01 -3.96961540e-01
-8.51578936e-02 9.57249999e-01 -1.51145089e+00 2.43688345e-01
6.92179263e-01 7.02371955e-01 2.85380393e-01 2.69787043e-01
1.09622866e-01 6.64966643e-01 -4.98035043e-01 3.13411504e-01
5.81077635e-01 -3.40260506e-01 -1.03679490e+00 -2.63834685e-01
4.21710998e-01 -1.33546397e-01 -3.82913202e-01 1.01402164e+00
6.15418434e-01 5.51396608e-01 1.02798104e+00 -6.57711267e-01
-9.54535484e-01 8.10705423e-01 -5.88747919e-01 4.95780230e-01
5.52548729e-02 -9.68377292e-02 1.31878150e+00 -5.69731653e-01
4.09981549e-01 1.06809330e+00 7.90441513e-01 3.64945740e-01
-1.25448442e+00 -6.06440961e-01 3.61899734e-01 -1.64089948e-01
-1.62455821e+00 -4.68746394e-01 7.61261106e-01 -4.94702637e-01
9.78164613e-01 3.41488540e-01 3.44052523e-01 8.98676455e-01
-2.31743678e-01 4.48552698e-01 1.10516596e+00 -3.96452278e-01
1.36018172e-01 -3.14055569e-02 3.15134078e-01 1.31549403e-01
3.89057994e-01 1.90881848e-01 6.97855353e-02 -1.42127916e-01
7.45793998e-01 4.82939214e-01 -2.97824174e-01 1.12197556e-01
-9.61940169e-01 1.12744892e+00 1.07682145e+00 6.06195331e-01
-5.92238009e-01 3.25678736e-01 1.14743493e-01 4.20489728e-01
7.32167780e-01 6.15842760e-01 -4.42128509e-01 8.27928856e-02
-8.91735733e-01 2.39098832e-01 3.57892841e-01 3.85822475e-01
1.08556247e+00 3.58697265e-01 2.43637502e-01 1.16152036e+00
2.34219208e-01 7.27047622e-01 1.91502079e-01 -4.45755512e-01
4.55554366e-01 5.73946714e-01 -2.25093946e-01 -9.54189241e-01
-4.10013705e-01 -5.49574435e-01 -1.11567700e+00 2.09785387e-01
-1.08972996e-01 -1.08846001e-01 -8.31802666e-01 1.58186209e+00
-1.76377818e-01 2.68020064e-01 2.99715281e-01 7.33927667e-01
9.11027968e-01 1.01691103e+00 3.68680567e-01 3.74953628e-01
1.03924668e+00 -4.10500228e-01 -1.30187154e-01 -5.65641165e-01
4.67639685e-01 -3.09276909e-01 1.01825404e+00 8.83808210e-02
-1.69252917e-01 -6.76151693e-01 -1.03318417e+00 3.34132314e-01
-7.73630917e-01 1.42020881e-01 1.18615150e+00 5.32300889e-01
-7.89168954e-01 7.07453787e-01 -7.21087575e-01 -6.63585007e-01
4.25480574e-01 -1.63415670e-01 -3.89115244e-01 -3.11624199e-01
-1.43613613e+00 1.11211419e+00 5.94371140e-01 1.54537871e-01
-1.06871092e+00 -8.76289070e-01 -8.44940305e-01 -2.56384201e-02
-1.40442193e-01 -3.35632831e-01 7.63619423e-01 -8.05145383e-01
-9.65462327e-01 9.50975657e-01 3.59858304e-01 -1.84275910e-01
-8.88328701e-02 -8.93542096e-02 -8.01034868e-01 5.99163361e-02
1.08427137e-01 4.52684343e-01 6.57278955e-01 -1.30191183e+00
-2.71289289e-01 -2.97124833e-01 1.45325273e-01 1.06997408e-01
-1.56851873e-01 1.63966700e-01 2.82882154e-01 -8.68065238e-01
2.77826279e-01 -8.17055225e-01 -4.76571023e-01 3.11468802e-02
-3.66087891e-02 2.50223614e-02 7.49928534e-01 -4.87653881e-01
8.83571148e-01 -2.24864316e+00 -2.11057544e-01 3.23269010e-01
-1.83412712e-02 4.70477313e-01 -7.20119894e-01 5.64875364e-01
-4.96766835e-01 6.48857534e-01 -3.92559052e-01 2.01556519e-01
-6.96393251e-02 5.11512995e-01 -8.35418820e-01 6.72310114e-01
6.04392648e-01 7.05180168e-01 -1.04870081e+00 -6.31030947e-02
4.41442013e-01 8.14035833e-01 -2.32972652e-01 4.33395088e-01
-5.32393306e-02 5.13719797e-01 -9.73849297e-01 9.32284296e-01
7.64734626e-01 -1.67118058e-01 2.43795246e-01 8.17609951e-02
3.42234671e-02 5.25233030e-01 -9.69158471e-01 1.42698717e+00
-5.82759082e-01 5.70185244e-01 -2.74368286e-01 -1.15981400e+00
1.26147580e+00 6.13062792e-02 5.58172941e-01 -5.08891165e-01
-2.34756380e-01 8.62570256e-02 -3.50837529e-01 -9.04678926e-02
4.80649352e-01 -3.91960919e-01 -1.37272954e-01 5.93885839e-01
-2.61252135e-01 -2.65470207e-01 -3.18264574e-01 -2.64103055e-01
7.50488997e-01 3.19399923e-01 7.90124834e-01 -2.95655251e-01
3.17183465e-01 2.15205416e-01 3.12241137e-01 8.91896904e-01
1.16922231e-02 8.47364008e-01 -9.40381140e-02 -7.28128195e-01
-4.94317949e-01 -9.49889302e-01 -5.38866222e-01 1.52049959e+00
-2.11340323e-01 -1.77540943e-01 6.31790459e-02 -4.77600247e-01
2.70126015e-01 4.81550723e-01 -9.19100583e-01 -2.90994123e-02
-1.13239847e-02 -1.11065722e+00 7.44853199e-01 7.25476563e-01
4.81061250e-01 -1.17852509e+00 -6.18135512e-01 1.77124783e-01
-1.63690895e-02 -5.22428274e-01 1.97674766e-01 4.15104389e-01
-1.04576492e+00 -1.02234626e+00 -6.53430760e-01 -1.07455246e-01
1.55180439e-01 9.30337846e-01 1.29063857e+00 -4.94360961e-02
-1.37866169e-01 1.36668503e-01 -7.96521187e-01 -5.02534628e-01
-3.33042383e-01 6.08786464e-01 -3.56013417e-01 -2.81957805e-01
4.52189565e-01 -9.12122488e-01 -3.90960842e-01 2.93199599e-01
-1.25821447e+00 -6.36551201e-01 7.92535961e-01 3.90504211e-01
6.01383090e-01 3.22835408e-02 5.94414353e-01 -8.93443525e-01
4.78769541e-01 -1.11758363e+00 -4.14526135e-01 2.99970537e-01
-5.34673572e-01 -1.66071847e-01 4.10407960e-01 -1.63017556e-01
-7.65241742e-01 1.03399031e-01 -8.06799904e-03 -1.47217497e-01
-4.33779567e-01 1.23994493e+00 -6.76248148e-02 -2.92240623e-02
1.18018329e+00 2.07837328e-01 -2.28723153e-01 -8.72099698e-01
6.01296246e-01 9.04577971e-01 2.58043230e-01 -9.19179738e-01
8.93640280e-01 3.73589218e-01 -1.53613478e-01 -1.16425288e+00
-9.87481356e-01 -7.53484726e-01 -6.69765651e-01 3.01116019e-01
6.14461124e-01 -1.41798127e+00 1.40317336e-01 2.79643964e-02
-7.05985725e-01 -3.42057556e-01 -4.67142761e-01 4.67596531e-01
-2.27117375e-01 1.06082648e-01 1.29873365e-01 -9.14785504e-01
-4.45354700e-01 -6.93042636e-01 1.12044120e+00 -1.97998472e-02
-4.77352627e-02 -1.01293588e+00 5.98155200e-01 -6.93592206e-02
1.01102805e+00 4.81176168e-01 7.55717337e-01 -6.07522249e-01
-2.24686876e-01 3.48733738e-03 -6.62228107e-01 3.95665109e-01
5.63676059e-01 -7.95718953e-02 -1.42743003e+00 -2.76919007e-01
-6.24960780e-01 -4.75132048e-01 1.45016944e+00 2.94385612e-01
1.12692380e+00 -8.08767136e-03 -2.95213401e-01 1.03686821e+00
1.68458319e+00 -2.27440018e-02 7.88973868e-01 5.44135213e-01
5.51563799e-01 6.02384746e-01 2.93168366e-01 5.20799100e-01
7.53632635e-02 5.43487012e-01 5.11116028e-01 -2.93475002e-01
-1.88453980e-02 -2.83355594e-01 1.58054605e-01 4.07084733e-01
-5.27732790e-01 -9.30616707e-02 -1.22630095e+00 7.09638357e-01
-1.58481693e+00 -1.28873420e+00 1.64450616e-01 1.80533969e+00
7.00348735e-01 -6.10301197e-01 3.17319520e-02 4.74195667e-02
4.59394157e-01 1.08130634e+00 -3.93211424e-01 -2.88159307e-02
-3.16354185e-01 4.32484418e-01 4.93676662e-01 6.36039525e-02
-1.50628734e+00 1.15475988e+00 7.69700003e+00 3.74572456e-01
-1.38771951e+00 -7.13703362e-03 3.45761836e-01 4.62708980e-01
-4.83420551e-01 6.30579293e-02 -6.87685549e-01 -1.47865057e-01
9.78937745e-01 -8.55331346e-02 3.37395847e-01 1.21372426e+00
2.26858839e-01 2.70132750e-01 -8.44552159e-01 7.48899221e-01
-2.58738220e-01 -1.29153812e+00 4.16436225e-01 1.17369361e-01
8.68630230e-01 7.69674897e-01 5.70321754e-02 5.99104106e-01
6.94247425e-01 -1.55649734e+00 1.89383298e-01 5.12766659e-01
1.09816146e+00 -3.94520342e-01 7.15950668e-01 -1.87648416e-01
-1.69265354e+00 1.53589979e-01 -9.63682413e-01 -2.60008991e-01
-2.57263273e-01 4.52865303e-01 -5.54518998e-01 9.27909195e-01
6.10434711e-01 1.34093177e+00 -5.94442427e-01 7.71261573e-01
-2.02832624e-01 9.13525164e-01 -4.45684999e-01 3.86805743e-01
6.03162825e-01 -2.32581913e-01 7.82278925e-02 1.42277241e+00
3.05711269e-01 2.81028032e-01 5.29005647e-01 8.38739395e-01
-1.05253950e-01 7.08076730e-02 -1.11665726e+00 -4.77967173e-01
5.53929925e-01 1.08153760e+00 -1.65036097e-01 -1.64326038e-02
-3.07875723e-01 3.42971146e-01 2.64544964e-01 5.78104019e-01
-4.57945198e-01 -4.07389700e-01 1.30577815e+00 3.68439592e-02
2.79610634e-01 -3.78090255e-02 -1.61887661e-01 -1.19811666e+00
-3.66130412e-01 -1.03324676e+00 6.69835627e-01 -8.22063982e-01
-1.58485413e+00 7.19849944e-01 2.47793317e-01 -1.37329161e+00
-2.24612117e-01 -6.05987251e-01 -8.10827255e-01 1.35229349e+00
-2.57595515e+00 -1.49008036e+00 -7.11349905e-01 7.09966063e-01
1.30120255e-02 -4.06142801e-01 1.27422953e+00 1.37167320e-01
-1.92732230e-01 2.09769964e-01 3.55287880e-01 3.45035434e-01
3.50221425e-01 -9.07103062e-01 3.93616676e-01 8.53904665e-01
2.80013382e-01 8.11334908e-01 1.54014513e-01 -2.40241215e-01
-1.37415695e+00 -1.71240866e+00 6.34403169e-01 -4.86715227e-01
7.15441823e-01 2.29988381e-01 -1.23493910e+00 9.45251346e-01
-4.68375385e-01 4.23885167e-01 9.49543178e-01 4.20383066e-01
-9.16872203e-01 -3.38333786e-01 -1.01777637e+00 -2.26886030e-02
9.96943891e-01 -9.60732698e-01 -1.01490700e+00 2.17083991e-01
3.33735555e-01 2.79593468e-01 -8.71626258e-01 3.07532489e-01
4.11673933e-01 -4.69834656e-01 1.20552218e+00 -7.53420949e-01
5.23890972e-01 -3.26889366e-01 -7.85477698e-01 -1.48380828e+00
-6.98173404e-01 -2.28015333e-01 5.55769742e-01 1.01498675e+00
3.53723228e-01 -6.93257928e-01 4.62545455e-01 -3.98833342e-02
-8.77460390e-02 -2.03949381e-02 -7.29387283e-01 -8.36456239e-01
6.51871860e-01 -5.67567825e-01 1.09324157e+00 1.49634981e+00
-3.09773326e-01 2.02866852e-01 -4.17345911e-01 5.28847694e-01
3.83205533e-01 4.81050164e-01 8.59281480e-01 -1.77443862e+00
-1.20744646e-01 -7.65208781e-01 -3.63550782e-01 -9.19948816e-01
2.77169704e-01 -1.29561710e+00 -8.43901262e-02 -1.77303016e+00
1.49796337e-01 -9.11739171e-01 -8.10941637e-01 1.08903575e+00
-2.32959718e-01 3.73692840e-01 1.43040732e-01 5.56408048e-01
-2.61560589e-01 6.17322981e-01 6.81241095e-01 -4.95775163e-01
-2.21804902e-01 -5.19407354e-02 -1.28340375e+00 2.76165634e-01
8.54862690e-01 -6.11876965e-01 -5.65731168e-01 -6.96974516e-01
1.42810464e-01 -2.28423312e-01 2.13970289e-01 -1.04282176e+00
-4.66414839e-01 -6.50894761e-01 1.35617822e-01 -3.38364899e-01
1.03624724e-01 -8.31265807e-01 5.29936790e-01 1.32631123e-01
-3.96961898e-01 -1.49672389e-01 2.31414437e-01 3.87314349e-01
-4.04040366e-01 -1.08274654e-01 8.88892531e-01 -2.54298568e-01
-1.24184096e+00 4.50883657e-01 -2.24108621e-01 8.24222416e-02
6.84980333e-01 -1.37520254e-01 -3.37306380e-01 -2.25493401e-01
-3.99455428e-01 -1.21578490e-02 3.10309291e-01 3.02345842e-01
4.87436742e-01 -1.11579406e+00 -1.35093892e+00 2.07407445e-01
7.60768831e-01 -9.93648544e-02 7.40105361e-02 5.76588102e-02
-5.87555051e-01 3.08025122e-01 -3.26705217e-01 -3.68029654e-01
-6.80846930e-01 1.04938790e-01 6.06730878e-01 -3.51040989e-01
-7.75562108e-01 5.02159953e-01 6.64472580e-03 -8.58229935e-01
-2.64130622e-01 -3.56843889e-01 -5.29122651e-01 2.39888325e-01
9.09430385e-01 9.68110636e-02 2.68724337e-02 -7.03264773e-01
-6.46619380e-01 6.14526927e-01 3.51447940e-01 2.17217132e-01
2.11542106e+00 2.34754398e-01 6.37122393e-02 5.31608522e-01
1.06860793e+00 -3.29247653e-01 -1.10978580e+00 -4.91461605e-01
-5.00496030e-02 -7.18021333e-01 4.29355472e-01 -7.91639268e-01
-1.35386157e+00 9.48098421e-01 5.77929974e-01 2.38606825e-01
9.97138023e-01 1.19834401e-01 6.41571209e-02 7.90326834e-01
3.68961930e-01 -4.83538866e-01 -4.52772021e-01 8.28970075e-01
9.90634799e-01 -1.36940491e+00 4.25354779e-01 -1.78215459e-01
-4.98287618e-01 9.27985549e-01 -8.38957876e-02 -2.63758689e-01
1.13855577e+00 -1.56099841e-01 4.35789406e-01 -3.23748559e-01
-3.50323975e-01 -6.98444664e-01 4.64389831e-01 1.26870811e+00
7.68009365e-01 3.69451940e-01 6.33993968e-02 2.46920109e-01
1.03746563e-01 4.50531691e-02 2.08864748e-01 8.82405937e-01
-6.16848588e-01 -9.10478950e-01 -2.67486960e-01 3.68897796e-01
-3.99580866e-01 -2.52142340e-01 -3.92820269e-01 9.35489178e-01
-2.65675455e-01 9.12560165e-01 -4.26367186e-02 -3.11614931e-01
4.10424769e-01 -4.13777083e-02 3.38595100e-02 -1.05005634e+00
-8.05008352e-01 -4.27691281e-01 5.23275286e-02 -3.42353284e-01
-1.03601086e+00 -4.57064629e-01 -7.72084653e-01 -4.03113246e-01
1.64851353e-01 2.39852726e-01 6.51626527e-01 7.19523489e-01
3.70985508e-01 3.48135918e-01 7.68318236e-01 -1.03031504e+00
-8.94787729e-01 -1.35455585e+00 -1.09717035e+00 5.60891330e-01
2.51650512e-01 -6.97482646e-01 -4.48247761e-01 -3.06303293e-01] | [9.638164520263672, -1.4078093767166138] |
3642863a-0a09-45e4-b5be-a9c6e2c5eb18 | bccwj-timebank-temporal-and-event-information | null | null | https://aclanthology.org/Y13-1019 | https://aclanthology.org/Y13-1019.pdf | BCCWJ-TimeBank: Temporal and Event Information Annotation on Japanese Text | null | ['Kikuo Maekawa', 'Hikari Konishi', 'Sachi Yasuda', 'Mizuho Imada', 'Masayuki Asahara'] | 2013-11-01 | bccwj-timebank-temporal-and-event-information-1 | https://aclanthology.org/O14-4001 | https://aclanthology.org/O14-4001.pdf | paclic-2013-11 | ['temporal-information-extraction'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.24799108505249, 3.8141849040985107] |
a9dee558-e3ed-4f10-8378-fcd085b8ae67 | automatic-discourse-segmentation-an | 2002.04095 | null | https://arxiv.org/abs/2002.04095v2 | https://arxiv.org/pdf/2002.04095v2.pdf | Automatic Discourse Segmentation: an evaluation in French | In this article, we describe some discursive segmentation methods as well as a preliminary evaluation of the segmentation quality. Although our experiment were carried for documents in French, we have developed three discursive segmentation models solely based on resources simultaneously available in several languages: marker lists and a statistic POS labeling. We have also carried out automatic evaluations of these systems against the Annodis corpus, which is a manually annotated reference. The results obtained are very encouraging. | ['Juan-Manuel Torres-Moreno', 'Andréa Carneiro Linhares', 'Alejandro Molina-Villegas', 'Rémy Saksik'] | 2020-02-10 | null | null | null | null | ['discourse-segmentation'] | ['natural-language-processing'] | [-1.00463659e-01 3.21765393e-01 -1.41099840e-01 -5.03277957e-01
-9.11119938e-01 -9.00483489e-01 8.32217813e-01 3.84641081e-01
-9.68365848e-01 1.03411174e+00 3.24470073e-01 -3.88543993e-01
-2.89451759e-02 -4.52664584e-01 -2.05869824e-01 -2.96159983e-01
1.62750050e-01 1.15354967e+00 7.81773269e-01 -2.20408395e-01
7.07768798e-01 5.50957143e-01 -1.23999989e+00 2.15087727e-01
1.15909243e+00 4.66575801e-01 1.57585666e-01 7.62804270e-01
-3.90213430e-01 4.91147786e-01 -1.07760966e+00 -6.80251062e-01
-3.78645733e-02 -4.81607258e-01 -1.27725375e+00 2.48380333e-01
1.74685851e-01 2.67670751e-01 1.39326915e-01 1.29436576e+00
3.09680611e-01 -2.36486774e-02 6.61981106e-01 -6.36290848e-01
-4.81414974e-01 1.16274381e+00 -1.97750002e-01 6.33628488e-01
6.07128084e-01 -2.69076526e-01 1.05844796e+00 -7.08638549e-01
1.12869883e+00 1.07166445e+00 5.08171439e-01 3.65332305e-01
-8.63633096e-01 4.26577814e-02 -2.76167452e-01 9.64898430e-03
-1.54409754e+00 -3.20638597e-01 4.63289261e-01 -4.21374768e-01
1.09654832e+00 5.40778577e-01 3.99727106e-01 4.60972250e-01
1.61955982e-01 9.17269349e-01 1.36809659e+00 -1.00044000e+00
1.97165206e-01 3.97136778e-01 5.64734280e-01 4.11765754e-01
2.15326115e-01 -3.90320569e-01 -1.09141082e-01 -1.42270952e-01
5.74047983e-01 -9.28840458e-01 -3.84470612e-01 8.91153887e-02
-1.11568534e+00 6.86031103e-01 -3.06938272e-02 1.19868684e+00
-3.08866590e-01 -3.89261335e-01 6.12110615e-01 5.80164194e-02
5.80845714e-01 5.66428125e-01 -6.04604721e-01 -3.57732117e-01
-1.19653273e+00 2.53831953e-01 1.33696949e+00 1.02095449e+00
3.47474962e-01 -2.67636508e-01 -8.93415883e-02 9.07585144e-01
4.17384177e-01 3.93282861e-01 6.25340104e-01 -7.61217594e-01
5.83338976e-01 5.39066494e-01 2.85068005e-01 -7.72689998e-01
-4.15531874e-01 2.56008655e-01 -1.52264282e-01 -2.64922142e-01
6.58941388e-01 -4.39058542e-01 -1.14695990e+00 1.00074136e+00
2.61714570e-02 -7.63048768e-01 1.13063619e-01 6.48971736e-01
9.53599036e-01 6.39086127e-01 1.52799025e-01 -5.99604249e-01
1.26494062e+00 -8.55293691e-01 -1.25322497e+00 4.22519147e-01
9.29120243e-01 -1.29262006e+00 8.42009664e-01 7.86987364e-01
-1.33996928e+00 -6.05481923e-01 -6.87738121e-01 8.37248117e-02
-5.79201341e-01 3.32900614e-01 3.65029544e-01 1.06080711e+00
-1.22935736e+00 4.54011917e-01 -7.99977422e-01 -5.49622178e-01
-1.13350548e-01 2.84629911e-01 -1.83314487e-01 4.75506812e-01
-1.02808189e+00 8.67547095e-01 9.93761837e-01 2.20752716e-01
-1.27163693e-01 4.87430453e-01 -7.51963496e-01 -3.56609166e-01
4.90322679e-01 2.10886151e-01 1.52869678e+00 -1.11563694e+00
-1.53111172e+00 1.26263738e+00 1.52220828e-02 -4.88451332e-01
5.53585231e-01 -2.62127757e-01 -8.98472488e-01 2.98134387e-01
2.08607867e-01 4.96961713e-01 -8.37572478e-03 -1.32493079e+00
-8.13776255e-01 -2.48660713e-01 -1.30632952e-01 2.29574785e-01
-5.23479693e-02 7.57346451e-01 -8.47913802e-01 -9.21098769e-01
1.97513968e-01 -9.79167104e-01 -2.74332702e-01 -9.55076873e-01
-5.13676107e-01 -4.03055072e-01 4.19100225e-01 -1.19724286e+00
1.59160948e+00 -1.75977445e+00 1.72443032e-01 4.34076846e-01
-2.32066795e-01 5.41673124e-01 2.78964341e-01 4.35614198e-01
6.87521100e-02 4.75368530e-01 -4.24920261e-01 5.85751422e-03
5.36846481e-02 4.99935120e-01 -6.81893453e-02 2.29278550e-01
-1.35245815e-01 7.18277633e-01 -8.43741298e-01 -1.03598404e+00
-5.52917235e-02 1.45554738e-02 -2.39037231e-01 2.77143531e-02
-3.27456772e-01 2.68854260e-01 -5.20226479e-01 6.62150204e-01
4.28813487e-01 4.60304022e-01 2.99891979e-01 1.85363486e-01
-5.22946060e-01 4.52072471e-01 -1.15599906e+00 1.67155063e+00
3.59899439e-02 4.48467255e-01 -1.14717379e-01 -4.96198744e-01
9.93711650e-01 5.39381206e-01 2.28070736e-01 -4.27203953e-01
3.28408122e-01 5.92845559e-01 1.78948283e-01 -4.53878105e-01
1.24190569e+00 1.30256370e-01 -2.83937484e-01 3.44622523e-01
2.39942670e-01 -5.03042638e-01 1.21560061e+00 1.66133177e-02
5.59276640e-01 3.03199589e-01 4.30781156e-01 -7.28108346e-01
8.31802011e-01 7.44484127e-01 4.92253095e-01 3.81246567e-01
-1.84502319e-01 7.13897943e-01 6.53266907e-01 -1.46531194e-01
-8.03786218e-01 -8.02846014e-01 -4.64972675e-01 8.14702570e-01
-6.61938265e-02 -6.43265128e-01 -1.14473271e+00 -1.16837513e+00
-6.36547744e-01 8.79457414e-01 -3.69299382e-01 9.18239951e-01
-9.71120417e-01 -7.80945420e-01 7.05079615e-01 3.59137982e-01
3.43634993e-01 -1.31096721e+00 -2.20829427e-01 3.34344387e-01
1.04040687e-03 -1.05486298e+00 -3.33604127e-01 3.10356230e-01
-7.30219483e-01 -1.10892498e+00 -8.05500805e-01 -1.10321987e+00
3.86691302e-01 -1.89082146e-01 1.19252563e+00 1.12919427e-01
2.32538849e-01 1.34862423e-01 -8.05998087e-01 -5.00641108e-01
-8.64386022e-01 4.49202240e-01 -3.85977715e-01 -7.37799942e-01
6.66629970e-01 1.03494376e-01 6.60572648e-02 4.86980975e-01
-1.25223207e+00 -6.04804814e-01 3.10693294e-01 4.24942344e-01
6.47816777e-01 -3.30993459e-02 3.57979655e-01 -1.54869473e+00
8.17051828e-01 -1.08905368e-01 -7.26119995e-01 2.00509742e-01
-3.44677687e-01 -7.20984489e-02 3.75245303e-01 1.84976995e-01
-1.44347346e+00 4.29156050e-02 -8.38610470e-01 3.48599315e-01
-5.39511204e-01 7.60771215e-01 -3.36694390e-01 -1.58448964e-02
8.72400045e-01 -2.68233895e-01 -3.07045847e-01 -7.69792318e-01
3.22298706e-01 1.09519792e+00 4.23965961e-01 -5.33539593e-01
2.65004814e-01 -1.06407590e-01 -6.59273863e-01 -1.15981519e+00
-5.69319129e-01 -7.63544142e-01 -1.15789759e+00 -1.16496354e-01
9.69433248e-01 -4.99267906e-01 8.53828415e-02 5.00251412e-01
-1.25831020e+00 -1.70791924e-01 -3.96059245e-01 5.98123908e-01
-4.13499713e-01 6.07208192e-01 -8.56883168e-01 -3.99550408e-01
-1.05161004e-01 -1.06805968e+00 6.63833141e-01 3.08997393e-01
-6.66830122e-01 -1.37208498e+00 4.89211380e-01 3.27775121e-01
-1.60562232e-01 -2.20449772e-02 4.74590719e-01 -1.30260479e+00
-4.44715135e-02 -3.01533610e-01 -4.79032937e-03 4.46365356e-01
1.96924563e-02 4.24067825e-01 -6.58333778e-01 6.19915463e-02
-2.92234160e-02 -2.53352046e-01 6.07960284e-01 2.93515712e-01
7.35970020e-01 1.09188296e-02 -3.42935145e-01 -1.46136194e-01
1.28677499e+00 5.81808567e-01 5.78613222e-01 4.64176208e-01
5.25607049e-01 8.09708953e-01 1.09860873e+00 6.65158257e-02
2.06714213e-01 4.82460678e-01 -2.60509551e-01 4.04953994e-02
-5.02444394e-02 1.29894942e-01 1.92597017e-01 1.50638247e+00
-4.31246877e-01 -8.00343692e-01 -1.40368950e+00 6.72356844e-01
-1.76156867e+00 -5.65791011e-01 -5.88586330e-01 1.73763669e+00
9.01760936e-01 5.67010880e-01 3.78019840e-01 2.51052141e-01
9.41829860e-01 3.11489671e-01 4.04858559e-01 -8.86021495e-01
-3.43136698e-01 4.08259839e-01 4.94487971e-01 6.36697292e-01
-1.27777207e+00 1.24343526e+00 8.07364082e+00 8.46904397e-01
-7.88116813e-01 8.28774050e-02 4.88075763e-01 4.76075351e-01
-3.07231575e-01 -1.64665654e-02 -8.60367656e-01 3.14740300e-01
1.22903049e+00 -1.57306835e-01 -2.22020745e-01 6.86795712e-01
5.39355970e-04 -5.87749124e-01 -7.20634460e-01 3.96791160e-01
2.21840739e-01 -1.25313103e+00 5.27741993e-03 -2.73179356e-02
7.63414383e-01 8.84902254e-02 -4.39762205e-01 1.89923756e-02
3.76485378e-01 -6.68540359e-01 7.82717347e-01 2.87750006e-01
4.45583016e-01 -8.22521210e-01 1.20777595e+00 1.73911110e-01
-6.43978775e-01 6.32186949e-01 -2.51816571e-01 6.34092763e-02
3.92072648e-01 3.34376305e-01 -7.71556675e-01 8.92133355e-01
4.57019627e-01 3.52578938e-01 -7.26680934e-01 1.30923545e+00
-6.58400595e-01 8.82941067e-01 -1.46614090e-01 -3.27072740e-01
5.09435594e-01 -4.41334605e-01 7.11377025e-01 1.80565655e+00
2.88797859e-02 -4.83203605e-02 2.47295395e-01 3.56802106e-01
1.66431263e-01 7.41428733e-01 -3.08811218e-01 -2.23333001e-01
1.01949327e-01 1.12176883e+00 -1.52623379e+00 -5.19868612e-01
-4.39678103e-01 9.10991848e-01 -8.15858543e-02 2.34633684e-02
-5.36331177e-01 -7.22806513e-01 -1.13962710e-01 -7.68516138e-02
2.39018440e-01 -3.70213389e-01 -2.52958536e-01 -1.07592368e+00
-1.44823432e-01 -9.43807006e-01 5.49746454e-01 -5.87329209e-01
-8.95185530e-01 1.17654288e+00 3.45933169e-01 -7.64229298e-01
-5.42026758e-01 -7.31022298e-01 -2.32432276e-01 9.35934484e-01
-9.77133870e-01 -7.19335496e-01 4.99730766e-01 2.11044356e-01
5.39629579e-01 -2.02267647e-01 7.48856902e-01 3.47087532e-01
-6.97778940e-01 2.19028726e-01 1.38458401e-01 4.05723900e-01
5.84253073e-01 -1.62294018e+00 3.71651739e-01 1.03354609e+00
6.55581176e-01 5.18935919e-01 9.64126945e-01 -8.61181021e-01
-4.50718969e-01 -5.47442973e-01 1.38392079e+00 -4.08236414e-01
9.18979228e-01 5.76113090e-02 -1.04291713e+00 7.96853960e-01
6.79787755e-01 -5.25045455e-01 7.49734461e-01 9.33796018e-02
4.64701355e-01 5.40565133e-01 -1.14215517e+00 2.84794897e-01
8.38664830e-01 -1.32845834e-01 -1.31019115e+00 4.75456238e-01
4.85030919e-01 -6.93357408e-01 -1.00883734e+00 2.18748510e-01
2.46242471e-02 -8.35454702e-01 1.67706475e-01 -1.71267584e-01
1.75446406e-01 -3.46725971e-01 4.19998504e-02 -1.26849210e+00
1.12122305e-01 -7.89374888e-01 6.50346875e-01 1.64658451e+00
7.65175223e-01 -4.14530128e-01 7.69493103e-01 3.76083374e-01
-4.67368990e-01 -1.13934271e-01 -7.37456024e-01 -6.75742090e-01
9.55638885e-02 -4.48464841e-01 2.63238519e-01 8.06273162e-01
2.38832384e-01 5.03359973e-01 2.25705683e-01 -1.40197963e-01
-2.06838287e-02 2.89419275e-02 4.07466233e-01 -1.20497549e+00
-1.44809768e-01 -4.99392211e-01 -2.60799378e-01 -6.99984372e-01
3.31463695e-01 -7.65073001e-01 3.89205366e-01 -1.37899101e+00
-5.96282594e-02 -3.89242709e-01 -1.92573801e-01 2.85964578e-01
3.46734002e-02 3.98564875e-01 7.71913305e-03 3.66309404e-01
-6.21728003e-01 -3.11381929e-03 1.17860043e+00 2.33346403e-01
-5.48716486e-01 9.03071836e-02 -4.49987590e-01 1.24015558e+00
8.40305150e-01 -4.74444270e-01 -4.97430824e-02 -3.63218397e-01
9.82899368e-02 -7.98101649e-02 -4.89082038e-01 -8.71891797e-01
-4.39938204e-03 -6.39755651e-02 1.20809734e-01 -9.90970969e-01
-3.74348387e-02 -5.03392994e-01 -4.09940705e-02 3.17991018e-01
-6.36251569e-02 3.72542351e-01 3.41912270e-01 8.71626735e-02
-6.89767659e-01 -9.06002045e-01 6.10853016e-01 -2.37216592e-01
-9.40995276e-01 -2.82680064e-01 -5.76551378e-01 2.90138155e-01
1.14333963e+00 -3.58184159e-01 -2.12359145e-01 2.01351658e-01
-9.55563366e-01 8.00879449e-02 8.54382694e-01 4.55566905e-02
1.32851258e-01 -9.14463043e-01 -5.51547825e-01 -7.13430271e-02
6.16950504e-02 -3.53875488e-01 -3.01422775e-01 6.33484304e-01
-1.39208949e+00 8.03167701e-01 -5.56195199e-01 -2.09153339e-01
-1.45448089e+00 5.12910604e-01 -1.21154815e-01 -3.33097041e-01
-1.52156815e-01 6.16897404e-01 -4.85484123e-01 -5.07979631e-01
2.14774817e-01 -4.95775014e-01 -9.69756842e-01 4.48885292e-01
1.70374513e-01 4.53107744e-01 2.72003472e-01 -1.21931148e+00
-3.76801461e-01 2.04028249e-01 -9.09497514e-02 -7.02230394e-01
1.08301747e+00 -1.36054561e-01 -4.61621583e-01 7.66574502e-01
5.76255739e-01 6.51232541e-01 -5.56688845e-01 -2.44763475e-02
7.62079477e-01 -3.42561424e-01 -1.12517543e-01 -9.17385459e-01
-5.93597591e-01 3.75448853e-01 1.40293479e-01 9.48790431e-01
9.22625482e-01 -4.51015867e-02 6.81543350e-01 3.51670712e-01
3.46266091e-01 -1.66472995e+00 -5.27115226e-01 8.49859238e-01
7.55209565e-01 -9.07106996e-01 7.83349872e-02 -8.30861151e-01
-8.54526997e-01 1.02910078e+00 1.77117571e-01 -6.51954710e-02
7.12520778e-01 1.05479218e-01 2.93443710e-01 -1.28675938e-01
-2.27308631e-01 -5.58123887e-01 4.64082628e-01 5.61062455e-01
1.01415896e+00 1.02422722e-01 -1.58787751e+00 3.43365163e-01
-4.46899384e-01 7.62007525e-03 8.32204998e-01 1.25790441e+00
-6.51207805e-01 -1.55621517e+00 -5.90218484e-01 1.80623010e-01
-1.14143491e+00 -9.65062063e-03 -1.01710343e+00 1.42468488e+00
1.77149862e-01 9.22960103e-01 1.45405367e-01 2.82927100e-02
3.58331800e-01 8.48686099e-02 6.24243498e-01 -9.98571396e-01
-8.97542477e-01 5.47481477e-01 7.67153561e-01 -1.61151573e-01
-9.54066455e-01 -9.99366283e-01 -1.22199452e+00 -1.09085038e-01
-4.85317886e-01 8.06417048e-01 5.82528055e-01 9.19651091e-01
-4.49067146e-01 3.23845029e-01 1.92332357e-01 -6.93347216e-01
-1.94597393e-01 -1.18381572e+00 -9.37609196e-01 1.26303032e-01
-4.84535426e-01 -1.70542255e-01 -1.81031182e-01 2.80278236e-01] | [10.405887603759766, 10.137604713439941] |
82639f2b-55c7-49d8-90d8-720bdec8b061 | ood-augmentation-may-be-at-odds-with-open-set | 2206.04242 | null | https://arxiv.org/abs/2206.04242v1 | https://arxiv.org/pdf/2206.04242v1.pdf | OOD Augmentation May Be at Odds with Open-Set Recognition | Despite advances in image classification methods, detecting the samples not belonging to the training classes is still a challenging problem. There has been a burst of interest in this subject recently, which is called Open-Set Recognition (OSR). In OSR, the goal is to achieve both the classification and detecting out-of-distribution (OOD) samples. Several ideas have been proposed to push the empirical result further through complicated techniques. We believe that such complication is indeed not necessary. To this end, we have shown that Maximum Softmax Probability (MSP), as the simplest baseline for OSR, applied on Vision Transformers (ViTs) as the base classifier that is trained with non-OOD augmentations can surprisingly outperform many recent methods. Non-OOD augmentations are the ones that do not alter the data distribution by much. Our results outperform state-of-the-art in CIFAR-10 datasets, and is also better than most of the current methods in SVHN and MNIST. We show that training augmentation has a significant effect on the performance of ViTs in the OSR tasks, and while they should produce significant diversity in the augmented samples, the generated sample OOD-ness must remain limited. | ['Mohammad Hossein Rohban', 'Mohammad Azizmalayeri'] | 2022-06-09 | null | null | null | null | ['open-set-learning'] | ['miscellaneous'] | [ 4.79394317e-01 4.98541929e-02 -2.04809323e-01 -4.58576083e-01
-7.71428585e-01 -4.36160713e-01 6.86177850e-01 -1.38027489e-01
-5.47688782e-01 5.28263390e-01 -1.25514984e-01 -3.40114057e-01
1.76325455e-01 -5.43248653e-01 -7.84722626e-01 -8.20497096e-01
3.88126448e-02 3.86096448e-01 5.28683484e-01 -3.11284602e-01
4.10891771e-02 4.36961859e-01 -1.83492780e+00 5.09445012e-01
6.38790309e-01 1.21313846e+00 1.07549969e-02 6.11185730e-01
-9.30543765e-02 6.84849858e-01 -9.11013782e-01 -4.54452395e-01
5.13860047e-01 -2.23938212e-01 -4.73764509e-01 1.24387853e-01
7.48694956e-01 -1.29369617e-01 -3.87708634e-01 1.05038881e+00
5.47305107e-01 2.14906231e-01 1.06525707e+00 -1.22001946e+00
-9.61979628e-01 6.00762188e-01 -8.82961154e-01 2.84939378e-01
-1.52608544e-01 1.93898063e-02 1.04859471e+00 -1.04787362e+00
2.65680999e-01 1.17618835e+00 6.32019162e-01 7.04311430e-01
-1.16329277e+00 -4.86062706e-01 8.06555673e-02 2.76098669e-01
-1.44079137e+00 -3.42686474e-01 5.99819541e-01 -4.24804866e-01
7.57694125e-01 4.24864382e-01 2.65382528e-01 1.35975742e+00
-3.32072020e-01 1.08547294e+00 1.30563557e+00 -8.11886311e-01
2.14561582e-01 4.46734339e-01 5.29870570e-01 3.73695642e-01
2.08491653e-01 1.20336749e-01 -3.28094602e-01 -1.12597637e-01
7.54044235e-01 -1.32771745e-01 -3.96324933e-01 -1.62991852e-01
-1.02259529e+00 9.40128207e-01 4.97284681e-01 2.58921534e-01
-1.16823353e-01 8.43472853e-02 3.57726127e-01 2.58555681e-01
6.88723743e-01 5.07149220e-01 -4.37740415e-01 1.26279324e-01
-9.73583162e-01 1.54990926e-01 6.14341021e-01 6.63658679e-01
4.96198177e-01 1.80943310e-01 -5.12801647e-01 1.22782159e+00
1.68478727e-01 3.27146649e-01 7.15190053e-01 -6.60671115e-01
2.62580007e-01 4.36980486e-01 2.39870846e-02 -5.84691048e-01
3.43626854e-03 -7.74297535e-01 -7.66511858e-01 4.86200422e-01
7.60934591e-01 -1.26536533e-01 -1.39721060e+00 1.50655508e+00
2.11063683e-01 2.05823839e-01 1.82763547e-01 7.89526343e-01
7.95606732e-01 8.03839684e-01 -2.25751236e-01 6.95771575e-02
1.40351331e+00 -1.06472671e+00 -6.83373809e-01 -5.07412553e-01
5.68729281e-01 -7.27578580e-01 1.13640249e+00 5.85414708e-01
-5.82632244e-01 -6.44402325e-01 -1.21713877e+00 1.08184539e-01
-5.11685848e-01 2.64260769e-01 6.51418746e-01 9.63256478e-01
-7.80572593e-01 6.28985763e-01 -4.43677187e-01 -2.24252224e-01
8.06950986e-01 1.39661118e-01 -2.86965728e-01 -1.81007475e-01
-9.04907346e-01 8.21755886e-01 4.19279784e-01 -5.12905195e-02
-7.28853226e-01 -6.75168335e-01 -7.15668976e-01 1.75022140e-01
6.06361091e-01 -2.81195194e-01 1.24231839e+00 -1.08150721e+00
-1.30104291e+00 9.24936831e-01 1.34102508e-01 -7.88295150e-01
6.07815981e-01 -2.01801956e-01 -2.32870772e-01 -5.47466613e-02
-1.68519244e-01 9.78240967e-01 1.16969323e+00 -1.34242368e+00
-5.14204144e-01 -4.26334202e-01 -9.32501480e-02 3.01793832e-02
-5.15382946e-01 7.25714564e-02 -3.34474206e-01 -9.48329329e-01
3.02477088e-02 -9.45464253e-01 -2.55116284e-01 -1.66731358e-01
-5.19338012e-01 -5.95079541e-01 6.53963029e-01 -2.99343795e-01
8.94553125e-01 -2.35787010e+00 -3.57427150e-01 5.01863323e-02
6.92725033e-02 6.75974607e-01 -2.24167705e-01 3.82804312e-02
-2.33262450e-01 1.73593059e-01 -2.56619543e-01 -5.33848524e-01
1.43588707e-01 3.89581978e-01 -7.35550106e-01 4.14736778e-01
3.82838041e-01 6.80132151e-01 -5.70638657e-01 -2.85620391e-01
2.61973202e-01 5.12537479e-01 -4.19413090e-01 9.30770487e-02
-4.34517682e-01 1.63575754e-01 -4.72240746e-02 4.74032432e-01
6.08497083e-01 -3.07483345e-01 -4.41409171e-01 -1.07593045e-01
1.23558536e-01 7.59592354e-02 -1.36585772e+00 1.07361066e+00
-2.79041138e-02 8.37389827e-01 -4.26667362e-01 -1.18385804e+00
1.01602161e+00 2.27331728e-01 5.68912849e-02 -4.94938552e-01
2.29773924e-01 2.86429107e-01 2.70638883e-01 -3.60442311e-01
5.68396747e-01 -6.83900854e-03 2.56984353e-01 1.42711580e-01
2.79873222e-01 1.49844244e-01 2.76428103e-01 -9.35726911e-02
8.32341433e-01 -3.24646458e-02 3.76686007e-01 -1.97909340e-01
2.80665100e-01 -6.87644631e-02 4.43524152e-01 1.26870966e+00
-2.05950722e-01 9.60827112e-01 3.67227793e-01 -6.33726567e-02
-1.06758976e+00 -9.94816542e-01 -4.93142009e-01 1.06779289e+00
-1.78306296e-01 -9.18293744e-02 -6.95906401e-01 -8.64170790e-01
-1.74020484e-01 6.81294560e-01 -6.54968858e-01 -2.37441305e-02
-2.65871763e-01 -1.05126917e+00 6.24173582e-01 6.62616491e-01
3.21053803e-01 -9.50764358e-01 -3.71813565e-01 -1.19068161e-01
1.66097935e-02 -1.41677547e+00 -3.19184572e-01 6.00110054e-01
-6.92796290e-01 -8.41532528e-01 -1.16287410e+00 -6.94909155e-01
6.60485029e-01 3.37599427e-01 9.55457926e-01 -1.83696270e-01
-2.39894971e-01 1.98721334e-01 -7.86006570e-01 -1.04969513e+00
-4.95724916e-01 3.86271179e-02 -1.54097136e-02 3.63961965e-01
5.63469052e-01 -2.85202831e-01 -3.01579356e-01 4.78587300e-01
-8.53224397e-01 -2.62868911e-01 7.31848001e-01 9.81021047e-01
5.06060302e-01 1.18347909e-02 8.39843869e-01 -1.16175294e+00
3.08643013e-01 -2.17093900e-01 -3.81935090e-01 8.23568106e-02
-4.57935184e-01 3.93685251e-02 6.97055101e-01 -8.38312507e-01
-7.88206875e-01 -2.57380214e-02 -3.64192367e-01 -7.06422806e-01
-6.15969777e-01 2.34207734e-01 9.20403451e-02 -5.07751293e-02
1.09168422e+00 2.63625354e-01 -9.86386985e-02 -5.99947453e-01
2.20513135e-01 1.01311898e+00 4.57601905e-01 -2.39501163e-01
7.63525665e-01 6.02857649e-01 -1.92903474e-01 -1.27031183e+00
-1.31845355e+00 -5.59998691e-01 -2.24950865e-01 5.61154522e-02
6.94500029e-01 -8.62396359e-01 -2.26597399e-01 5.34492254e-01
-9.20396864e-01 -5.04175544e-01 -6.88092053e-01 5.29642642e-01
-2.13753968e-01 3.41893315e-01 -3.66294175e-01 -1.09999347e+00
-7.71367922e-02 -1.28949404e+00 8.77490699e-01 1.84235558e-01
-8.26242268e-02 -7.63423800e-01 -1.51233479e-01 4.50495213e-01
2.73021340e-01 -6.52630348e-03 6.16949737e-01 -1.25979114e+00
-3.77768636e-01 -3.91563296e-01 -1.62763178e-01 8.89725149e-01
1.04857814e-02 3.34328711e-02 -1.58867538e+00 -2.65920669e-01
-7.79566020e-02 -4.89627957e-01 1.31886518e+00 5.27453184e-01
1.36310387e+00 -7.14499801e-02 -1.40015185e-01 4.27871734e-01
1.17234766e+00 1.24082334e-01 8.02826881e-01 1.93566039e-01
6.60578907e-01 5.30445635e-01 5.87723017e-01 2.49289200e-01
-1.34572461e-02 7.09724367e-01 5.29568553e-01 -3.45801324e-01
-3.77313703e-01 1.21981360e-03 5.23664117e-01 2.85407603e-01
-1.58331171e-02 -4.33547050e-01 -6.82311654e-01 5.25227606e-01
-1.72375631e+00 -8.81022871e-01 -3.20531934e-01 2.21819949e+00
8.57980609e-01 4.12337869e-01 1.08474948e-01 5.43814719e-01
7.13111937e-01 2.54633278e-01 -4.13295209e-01 -2.60839969e-01
-2.11107433e-01 4.09849048e-01 5.77060580e-01 2.18339399e-01
-1.20279884e+00 7.42234170e-01 6.28485346e+00 1.27860618e+00
-9.73379850e-01 5.64184971e-02 8.67392182e-01 1.85713783e-01
1.78836629e-01 -2.90269732e-01 -1.19836092e+00 3.49947155e-01
6.35126710e-01 6.04037821e-01 2.47010738e-01 1.14302564e+00
-1.81442484e-01 -2.27728188e-01 -1.29351771e+00 1.26594222e+00
4.07175064e-01 -1.14308429e+00 6.43785149e-02 1.49144903e-01
7.96993911e-01 1.95590645e-01 3.03297460e-01 7.69528806e-01
1.45851254e-01 -1.07990611e+00 6.92120731e-01 7.01349676e-02
6.01479173e-01 -5.85065246e-01 9.33390677e-01 5.11927366e-01
-6.27917886e-01 -5.69485016e-02 -6.43945754e-01 2.23461948e-02
-2.62711376e-01 8.01780105e-01 -9.14399445e-01 2.83236921e-01
7.10435688e-01 4.92044896e-01 -8.13827932e-01 1.16627812e+00
-1.77640945e-01 9.78521824e-01 -3.97712350e-01 -5.32469414e-02
1.60646737e-01 6.21203929e-02 5.92804372e-01 1.07800531e+00
-1.45625528e-02 -3.11960608e-01 1.36202231e-01 7.42172956e-01
-2.27910146e-01 -3.81545797e-02 -5.13776958e-01 -1.15271717e-01
1.30286172e-01 1.11580551e+00 -8.13925505e-01 -4.84015942e-01
-5.16486466e-01 8.96892130e-01 2.27321491e-01 2.21855357e-01
-8.34733844e-01 -3.70368779e-01 5.29038370e-01 -2.78079826e-02
6.99954331e-01 1.48741737e-01 -2.56336331e-01 -1.17922592e+00
-9.62375570e-03 -9.97972131e-01 4.42702144e-01 -5.79922020e-01
-1.70784092e+00 5.68511903e-01 -1.72458798e-01 -1.14420354e+00
-1.82915237e-02 -1.04154694e+00 -4.38440949e-01 6.74359560e-01
-1.78811347e+00 -8.33037496e-01 -1.27459988e-01 3.66841018e-01
8.56930017e-01 -2.26206735e-01 6.26350462e-01 3.76951933e-01
-5.31338215e-01 9.56073344e-01 1.28366932e-01 4.34674114e-01
8.60624135e-01 -1.37225473e+00 3.26117098e-01 1.05345941e+00
7.30918646e-01 3.16849679e-01 8.98058176e-01 -3.24367821e-01
-8.16708922e-01 -1.08610749e+00 4.75744724e-01 -5.45636177e-01
4.50455129e-01 -5.51039457e-01 -1.02866244e+00 7.39720583e-01
4.73172739e-02 4.77495492e-01 6.22273862e-01 1.10492423e-01
-5.62427938e-01 1.34742558e-02 -9.12966549e-01 6.63533568e-01
7.25805104e-01 -2.03512222e-01 -7.63025641e-01 2.82932550e-01
6.65878356e-01 -3.42710167e-01 -5.88953495e-01 5.70124030e-01
2.87292987e-01 -9.68824267e-01 1.16381001e+00 -6.33550942e-01
3.20958346e-01 -2.34231517e-01 -4.66146141e-01 -1.34333920e+00
-8.87173563e-02 -3.03735197e-01 -3.00571740e-01 1.39324439e+00
4.33021396e-01 -6.69906318e-01 8.74564648e-01 2.01698273e-01
-3.63809496e-01 -8.49175930e-01 -7.28765249e-01 -1.14644384e+00
2.41555590e-02 -7.34741569e-01 2.19944954e-01 9.64960277e-01
-4.91943061e-01 4.66286749e-01 -3.64068627e-01 2.79226989e-01
6.61005080e-01 -2.69037873e-01 8.71498048e-01 -1.28220582e+00
-5.43440998e-01 -3.59949142e-01 -4.37176347e-01 -1.14365816e+00
1.36477500e-02 -9.09431398e-01 -3.41027416e-02 -1.32451808e+00
2.77165938e-02 -5.29399395e-01 -4.07922328e-01 6.04605973e-01
-2.87493706e-01 6.54957592e-01 3.02163601e-01 1.60766304e-01
-4.57723081e-01 5.61005414e-01 1.11676002e+00 -7.86112472e-02
-1.59758404e-01 1.49738818e-01 -8.36368680e-01 9.56100881e-01
7.44246602e-01 -5.99319279e-01 -3.03480834e-01 -1.35217816e-01
-2.24899389e-02 -3.87241542e-01 4.94334936e-01 -1.16914117e+00
-1.17889717e-01 2.03366399e-01 4.60705489e-01 -6.67717576e-01
5.55973351e-01 -6.71715915e-01 -4.86629307e-01 3.26454133e-01
-4.97868389e-01 -5.51530480e-01 2.42735278e-02 6.59497380e-01
-3.22661698e-02 -6.14831448e-01 1.08908653e+00 -8.51952583e-02
-7.01179743e-01 2.73616016e-01 -3.57717514e-01 3.80250275e-01
9.64510024e-01 -4.20081764e-01 -3.61168206e-01 -2.54775025e-02
-5.66629946e-01 -1.06601775e-01 7.79050663e-02 4.65744525e-01
3.39437187e-01 -9.83786345e-01 -7.30407953e-01 1.77213520e-01
3.94770026e-01 2.09336132e-01 9.33467150e-02 7.40073800e-01
-6.52442351e-02 3.45578134e-01 2.30555415e-01 -8.58724535e-01
-1.29932082e+00 4.97158676e-01 2.33438909e-01 -2.97123969e-01
-5.12861609e-01 9.64890063e-01 3.28944504e-01 -4.36411172e-01
5.72440624e-01 -3.01086307e-01 -5.29420912e-01 1.63741693e-01
8.60545993e-01 1.69777602e-01 2.06985816e-01 -2.97234476e-01
-1.81674119e-02 1.43998638e-01 -3.99047792e-01 7.63114020e-02
1.54558790e+00 2.07951963e-01 3.64968657e-01 7.35970080e-01
9.72604215e-01 -1.54829130e-01 -1.33666718e+00 -4.42783594e-01
-1.52376980e-01 -6.27439201e-01 2.30939806e-01 -7.46157825e-01
-7.87502289e-01 1.03990293e+00 8.25514495e-01 5.33978224e-01
7.19901800e-01 1.79164782e-01 5.73214054e-01 3.69443834e-01
2.82755077e-01 -9.52329516e-01 2.39640221e-01 5.67533433e-01
7.58722782e-01 -1.64828455e+00 -2.21307039e-01 -5.67915082e-01
-6.81953669e-01 9.36075330e-01 6.87437713e-01 -3.45762938e-01
5.22055507e-01 2.76308984e-01 -9.13314521e-02 1.06627718e-01
-5.13691902e-01 -4.67861503e-01 3.28534007e-01 8.72560322e-01
2.71918476e-01 -1.39227360e-01 1.15134090e-01 4.99232084e-01
-6.18925057e-02 -1.48345158e-01 7.37409472e-01 6.65373623e-01
-4.88541633e-01 -9.64213192e-01 -7.17408895e-01 8.01876545e-01
-7.16960013e-01 -2.42199004e-01 -2.71999270e-01 8.12517762e-01
5.92845231e-02 7.87809193e-01 1.23705469e-01 -2.81484991e-01
1.98410615e-01 1.56339049e-01 6.24587536e-01 -8.25973928e-01
-4.26911920e-01 -2.51456182e-02 6.31314516e-02 -9.68426466e-02
-2.62248516e-01 -5.16322076e-01 -8.50817800e-01 8.45531449e-02
-7.28942811e-01 -1.30384356e-01 5.88223517e-01 9.66723263e-01
1.12980351e-01 7.10361362e-01 4.85542208e-01 -8.54768217e-01
-1.02733970e+00 -1.15834928e+00 -6.80618286e-01 4.08886403e-01
4.85395253e-01 -7.57413030e-01 -6.42049968e-01 7.41541805e-03] | [9.460394859313965, 2.866537094116211] |
03afc3e4-308d-4205-b3f3-7b69c57f81ab | causality-in-neural-networks-an-extended | 2106.05842 | null | https://arxiv.org/abs/2106.05842v1 | https://arxiv.org/pdf/2106.05842v1.pdf | Causality in Neural Networks -- An Extended Abstract | Causal reasoning is the main learning and explanation tool used by humans. AI systems should possess causal reasoning capabilities to be deployed in the real world with trust and reliability. Introducing the ideas of causality to machine learning helps in providing better learning and explainable models. Explainability, causal disentanglement are some important aspects of any machine learning model. Causal explanations are required to believe in a model's decision and causal disentanglement learning is important for transfer learning applications. We exploit the ideas of causality to be used in deep learning models to achieve better and causally explainable models that are useful in fairness, disentangled representation, etc. | ['Abbavaram Gowtham Reddy'] | 2021-06-03 | null | null | null | null | ['explainable-models'] | ['computer-vision'] | [-2.62060583e-01 5.02411067e-01 -8.29190314e-01 -8.80750000e-01
4.42154557e-01 -1.72295704e-01 9.25454021e-01 1.90873146e-01
2.10157350e-01 1.15746653e+00 6.38655245e-01 -6.37440979e-01
-6.15973294e-01 -8.39848578e-01 -5.03327906e-01 -2.93828905e-01
-3.96724999e-01 4.91800040e-01 -2.20278397e-01 -1.25186458e-01
4.40380096e-01 4.95529652e-01 -1.12118745e+00 5.83832741e-01
1.01422167e+00 5.81958771e-01 -4.41429406e-01 7.34339237e-01
-6.35512620e-02 1.87411177e+00 -2.57485032e-01 -7.05544114e-01
5.27097434e-02 -5.97663999e-01 -1.16384196e+00 -7.43943632e-01
-2.97401869e-03 -3.87793630e-01 -3.04828972e-01 8.02677274e-01
-2.60338426e-01 -1.23787582e-01 1.00228250e+00 -2.00564075e+00
-1.21149898e+00 1.22967780e+00 -4.79938984e-01 3.45341176e-01
2.87530899e-01 -6.57796860e-04 1.43023407e+00 -3.21522772e-01
3.85105796e-02 1.84095871e+00 1.55185491e-01 8.69695783e-01
-1.07914674e+00 -1.14100850e+00 8.74431282e-02 9.85905051e-01
-4.04496223e-01 2.57399641e-02 7.30115294e-01 -4.16413724e-01
6.34670019e-01 5.14462948e-01 8.14782858e-01 1.23675430e+00
6.84899926e-01 7.76317477e-01 1.40288186e+00 -3.04141730e-01
2.25811556e-01 6.87989369e-02 3.70737612e-01 9.21747208e-01
5.75428128e-01 9.01176929e-01 -8.79061282e-01 -1.81827582e-02
9.50817049e-01 2.52354652e-01 -9.91227850e-02 -2.97581464e-01
-1.16606009e+00 1.49986720e+00 1.12429738e+00 -1.40723333e-01
-5.79032481e-01 8.21704328e-01 2.19150379e-01 4.44170505e-01
6.07170090e-02 7.63061047e-01 -6.86164677e-01 5.18267378e-02
-4.01151180e-01 3.43497276e-01 7.02617645e-01 4.39276159e-01
3.55895519e-01 1.36313990e-01 -4.10081493e-03 3.18723097e-02
7.77880013e-01 4.49786514e-01 3.01098347e-01 -1.22325432e+00
-1.60085466e-02 8.91704321e-01 -1.54354768e-02 -8.55339706e-01
-4.64982539e-01 -2.25886345e-01 -9.40548420e-01 5.82421124e-01
9.28148329e-02 -7.60801733e-02 -8.15533042e-01 1.58325684e+00
2.49480799e-01 5.29752016e-01 1.73998132e-01 1.22382402e+00
8.41358542e-01 4.37264711e-01 5.52108765e-01 -3.27552259e-02
1.29250050e+00 -7.46627986e-01 -9.04748261e-01 -1.48814663e-01
3.30325305e-01 -4.38169122e-01 9.40281272e-01 4.33092237e-01
-6.40159965e-01 -2.16556951e-01 -1.16291845e+00 -1.17084369e-01
-8.01664442e-02 -4.72855359e-01 1.64336777e+00 4.93216604e-01
-3.88569146e-01 7.76262939e-01 -5.44900239e-01 4.40407433e-02
9.25948501e-01 3.75271350e-01 -5.53763211e-01 9.16347578e-02
-1.61007750e+00 1.39238513e+00 4.32170033e-01 -1.06880598e-01
-1.21039844e+00 -7.30505109e-01 -6.65847182e-01 3.57674301e-01
1.85588658e-01 -1.20970905e+00 1.22511697e+00 -1.03471899e+00
-1.14764929e+00 2.99882889e-01 2.19189916e-02 -7.39283085e-01
4.93352056e-01 -4.63056564e-01 -2.67261356e-01 -2.54468143e-01
-1.70676425e-01 3.78742844e-01 7.87093878e-01 -1.17296994e+00
-6.52764022e-01 -2.96470523e-01 8.09186995e-02 -7.69505044e-03
2.41448224e-01 -1.69019073e-01 9.73873556e-01 -3.16147536e-01
3.58208828e-02 -8.00694823e-01 -1.45467371e-01 -1.46163210e-01
-5.61800957e-01 -6.15851164e-01 7.87832618e-01 -4.36785221e-01
6.32260561e-01 -1.45881259e+00 1.72712162e-01 6.67626783e-02
8.52042139e-01 -3.34217399e-02 1.00093856e-01 2.68304944e-01
-6.79021835e-01 6.08386874e-01 3.93162102e-01 2.69280016e-01
1.04044639e-01 3.94487530e-01 -5.18028319e-01 3.17886889e-01
3.62558931e-01 1.04118335e+00 -1.01374686e+00 -3.62138301e-01
2.85579264e-01 2.43443236e-01 -7.32678235e-01 7.52105713e-01
-2.17251256e-01 5.15666723e-01 -7.15811431e-01 6.35668635e-02
1.24373838e-01 -1.34575471e-01 -3.13697878e-04 -6.10738583e-02
1.97910339e-01 6.73132718e-01 -8.27476561e-01 1.06037498e+00
-4.17932063e-01 9.53019857e-01 -8.04415643e-01 -7.47952521e-01
8.16371799e-01 5.34468114e-01 -6.13862164e-02 -5.64849198e-01
2.64836550e-01 1.68869831e-02 6.62682712e-01 -7.48081386e-01
-5.08811139e-02 -6.85234845e-01 2.74786800e-01 8.89115453e-01
-1.61660731e-01 -1.24453463e-01 -2.00364634e-01 5.20799100e-01
7.30960190e-01 -7.49676824e-02 9.39096272e-01 -3.96257639e-01
1.90304190e-01 8.45021829e-02 6.96879923e-01 5.63917518e-01
-2.81267971e-01 1.34348243e-01 9.31152165e-01 -1.14359021e+00
-9.82415140e-01 -1.04372549e+00 6.08318672e-02 8.45271945e-01
-7.50727281e-02 -1.52654611e-02 -2.28343293e-01 -9.00867879e-01
1.76301673e-01 1.46521080e+00 -1.08547771e+00 -5.95304966e-01
-2.52799839e-01 -4.95626837e-01 2.13019416e-01 7.09125876e-01
3.30787927e-01 -1.10401487e+00 -7.76912570e-01 -2.54592389e-01
-2.94542499e-03 -3.86256099e-01 1.80737525e-01 3.73431772e-01
-8.60527456e-01 -1.44859946e+00 2.64526546e-01 4.89862897e-02
4.51754868e-01 1.87725797e-01 1.23142552e+00 5.56828260e-01
3.37188654e-02 -1.07234009e-01 -2.82096922e-01 -9.85590100e-01
-6.39893174e-01 -2.85246104e-01 2.08744988e-01 -4.67945158e-01
5.20199001e-01 -6.04822457e-01 -7.35594630e-01 1.30227268e-01
-5.13585925e-01 5.21842062e-01 4.43246186e-01 1.12621176e+00
-3.07484716e-01 3.47882099e-02 7.79216826e-01 -1.21792150e+00
9.32431638e-01 -7.42552638e-01 -3.45333546e-01 3.22703511e-01
-1.26304054e+00 5.88766754e-01 5.48342466e-01 -2.81572998e-01
-1.29329383e+00 -5.49483657e-01 4.41247672e-01 -1.28785312e-01
-1.46765649e-01 4.83322680e-01 -1.60262093e-01 3.74130487e-01
9.04567122e-01 -5.98478973e-01 -2.35726416e-01 6.49747858e-03
9.65253294e-01 3.33774924e-01 3.85327861e-02 -6.52220130e-01
6.73470080e-01 2.31420130e-01 3.07334542e-01 9.04858932e-02
-1.06033027e+00 1.15564838e-01 -6.74431860e-01 -2.65685856e-01
8.89658809e-01 -6.45981967e-01 -1.39193249e+00 -2.96705097e-01
-1.42618120e+00 -1.55518368e-01 -2.41324812e-01 8.18069398e-01
-4.93683547e-01 -5.01237214e-01 -2.64352113e-01 -9.31569457e-01
-4.40438502e-02 -7.61352062e-01 1.87128484e-01 6.03082716e-01
-9.91065204e-01 -1.29561591e+00 9.92260724e-02 5.15817106e-01
1.70034349e-01 1.63332522e-01 1.52260494e+00 -9.55654144e-01
-6.15195692e-01 -1.71553984e-01 -4.64059085e-01 -8.45166296e-02
1.50263757e-01 3.04000795e-01 -9.84405637e-01 5.37759423e-01
-1.17826499e-01 -3.61994416e-01 8.44026208e-01 4.42872107e-01
1.02723050e+00 -9.17457640e-01 -2.57926136e-01 3.36731896e-02
9.47516084e-01 2.11931542e-01 6.51602924e-01 6.65999651e-02
8.60062897e-01 1.10456252e+00 4.80065465e-01 2.21457273e-01
6.81326151e-01 7.99989104e-02 8.76183987e-01 -2.59219557e-01
-9.35880765e-02 -6.29949510e-01 -6.72153905e-02 4.27018911e-01
-4.65189368e-01 1.70468744e-02 -9.19450521e-01 1.34952858e-01
-2.29149675e+00 -1.46886110e+00 -9.68996823e-01 1.85213423e+00
8.83626699e-01 -2.38945801e-02 -1.77222639e-01 2.69212693e-01
3.46006840e-01 -1.51029781e-01 -5.49335957e-01 -1.04347050e+00
2.16379300e-01 8.63807276e-03 3.02803725e-01 7.42639422e-01
-6.48862541e-01 9.37486351e-01 6.95148134e+00 1.76943824e-01
-8.81185949e-01 9.46330726e-02 8.98923218e-01 2.01767646e-02
-9.85098481e-01 5.17725766e-01 -4.85812165e-02 4.75380234e-02
9.41894889e-01 -4.15263116e-01 4.29526061e-01 9.12204862e-01
6.29053593e-01 -1.13444790e-01 -1.69754922e+00 6.22072697e-01
-6.38141513e-01 -1.54416883e+00 3.17357332e-01 -1.75355375e-01
9.33849216e-01 -4.99692380e-01 3.58073264e-02 1.01323396e-01
1.30290699e+00 -1.93318832e+00 6.71084940e-01 5.77642381e-01
3.96616478e-03 -8.55305076e-01 9.82356906e-01 1.79101080e-01
-4.41075027e-01 -2.64189392e-01 -5.27075052e-01 -8.54710162e-01
-1.47971243e-01 7.59787261e-01 -1.18024337e+00 4.70431089e-01
5.14328659e-01 7.47684419e-01 -2.55192727e-01 6.18354023e-01
-1.35180831e+00 8.58395040e-01 5.35247803e-01 -4.86677259e-01
-1.41005255e-02 9.34302341e-03 5.11424243e-02 6.62809730e-01
-2.26147339e-01 2.96968997e-01 -5.13034940e-01 1.30519581e+00
-1.43650351e-02 -1.94065526e-01 -6.00234151e-01 -8.06818232e-02
4.37547475e-01 1.05098021e+00 -3.51593703e-01 -2.87365466e-01
-1.25358671e-01 6.30337954e-01 3.65238696e-01 2.91920781e-01
-1.02159429e+00 2.86305040e-01 8.85542989e-01 -9.93534997e-02
-6.16817474e-01 -5.10769635e-02 -1.01479518e+00 -1.04470491e+00
-7.95635760e-01 -8.86694133e-01 5.15305936e-01 -1.13055551e+00
-1.69581103e+00 3.67088050e-01 2.90448554e-02 -5.49828172e-01
-3.75906914e-01 -6.33773446e-01 -1.13911104e+00 1.06041968e+00
-1.68805408e+00 -1.38659418e+00 -1.93258598e-01 5.97893178e-01
4.11491394e-01 -2.71018237e-01 8.97201717e-01 -2.85237521e-01
-3.86508852e-01 -1.81148702e-03 -8.79434824e-01 -4.69719321e-02
5.70238531e-01 -1.68250322e+00 1.90400034e-01 7.25915730e-01
4.18118149e-01 1.02868366e+00 1.11981869e+00 -6.89365745e-01
-1.16768634e+00 -7.38904774e-01 1.01446223e+00 -7.14010239e-01
7.87597775e-01 5.34283817e-02 -8.94336700e-01 9.03369009e-01
6.10474944e-01 -1.90101326e-01 9.96539176e-01 9.02303338e-01
-7.76917338e-01 -2.05235910e-02 -9.83520985e-01 7.42917776e-01
7.52157390e-01 -3.49276423e-01 -1.18590450e+00 1.07121892e-01
9.53010976e-01 2.01029964e-02 -3.81273270e-01 1.01223290e-01
8.63593519e-01 -1.18806708e+00 7.20979691e-01 -1.57103539e+00
1.12601030e+00 -1.48678675e-01 2.33642399e-01 -1.70813596e+00
-6.23933017e-01 -4.35034752e-01 -1.53122500e-01 9.40182388e-01
7.02968001e-01 -7.63308108e-01 5.70612729e-01 1.21848118e+00
4.49300379e-01 -4.03416783e-01 -6.05855525e-01 -2.88638324e-01
3.29549283e-01 -5.44728398e-01 1.05011630e+00 1.51487696e+00
3.82525980e-01 6.39481425e-01 -5.68896949e-01 3.13211709e-01
7.27736592e-01 2.35174656e-01 6.02439225e-01 -1.75159156e+00
7.79677480e-02 -4.45437819e-01 -2.66847610e-01 -2.09617704e-01
3.68296176e-01 -8.34061623e-01 -4.70851004e-01 -1.68743777e+00
6.11980617e-01 -5.28831422e-01 -3.20712924e-01 7.74878800e-01
-5.36768317e-01 -3.16198558e-01 2.64873207e-01 2.64185131e-01
-1.34565411e-02 6.52407348e-01 1.45559454e+00 -7.72917643e-02
2.59927005e-01 4.11192402e-02 -1.09360933e+00 9.21343029e-01
9.89256084e-01 -1.03157401e+00 -7.55179226e-01 -2.58656859e-01
6.52665615e-01 1.46989971e-01 8.28710377e-01 -3.24086010e-01
1.58357665e-01 -1.11290336e+00 6.12592995e-01 1.30892307e-01
-2.30121002e-01 -9.43297863e-01 2.28318989e-01 8.67997646e-01
-8.67756486e-01 1.39330745e-01 -1.71611488e-01 4.36307997e-01
-3.43879350e-02 -1.21061638e-01 5.78645170e-01 -9.03630853e-02
-5.09536922e-01 1.46936223e-01 -2.15908483e-01 -2.80819505e-01
1.04969049e+00 2.46471778e-01 -6.61870003e-01 -6.50170863e-01
-4.12814021e-01 3.98113966e-01 -8.44435021e-02 7.13009179e-01
8.61023664e-01 -1.37500250e+00 -1.00659513e+00 -1.16387226e-01
-1.27527282e-01 -4.22236443e-01 1.53407669e-02 4.58645314e-01
-5.13629675e-01 6.32178128e-01 -6.37361526e-01 -6.13504089e-02
-9.72370505e-01 4.47728872e-01 3.82820278e-01 -2.10890472e-01
-9.50321928e-02 9.07438934e-01 3.95883977e-01 -3.37802172e-01
-1.67294487e-01 -2.43978113e-01 -4.76623684e-01 -2.73841321e-01
4.88678902e-01 4.19569463e-01 -5.59419036e-01 1.17329203e-01
-2.93728113e-01 -1.10516801e-01 3.22428197e-02 -1.10221662e-01
1.40117526e+00 1.37516245e-01 -3.42777967e-01 6.91589355e-01
5.44384003e-01 -2.26935297e-01 -1.18694174e+00 2.11557075e-01
2.09488481e-01 -7.38038361e-01 1.83849409e-01 -1.34904170e+00
-9.62065160e-01 1.33808565e+00 3.38851184e-01 4.97988164e-01
4.94712889e-01 -8.75330493e-02 1.34086862e-01 1.30036403e-03
2.14393318e-01 -4.25114155e-01 2.39787012e-01 -2.40886435e-02
1.30465138e+00 -1.65170193e+00 2.97958285e-01 -2.84831792e-01
-9.27559555e-01 1.36207545e+00 7.10780859e-01 -1.43413082e-01
5.54372251e-01 5.63261956e-02 1.71540648e-01 -5.02287924e-01
-1.19267988e+00 2.05502659e-01 7.01175690e-01 7.24649787e-01
8.73312950e-01 6.33204341e-01 -3.90038073e-01 8.14996243e-01
-4.83027041e-01 -9.60514098e-02 7.10238516e-01 4.11968632e-03
-2.82287687e-01 -1.04174149e+00 -1.69560432e-01 3.86463046e-01
-1.73128724e-01 -1.77516937e-01 -6.38579011e-01 8.05467784e-01
8.28658789e-02 1.24785435e+00 -1.43567115e-01 -3.23901802e-01
-6.09731004e-02 -7.72619173e-02 3.17257017e-01 -4.90319520e-01
-3.62518281e-01 -7.05523372e-01 1.14221953e-01 -7.94625640e-01
-4.55919951e-01 -2.13741079e-01 -1.64841974e+00 -1.11223555e+00
-3.39010477e-01 2.99720824e-01 7.15165913e-01 1.24896348e+00
6.03895076e-02 8.91396284e-01 6.02137268e-01 2.24645454e-02
-3.44131470e-01 -8.35551858e-01 -3.23944777e-01 3.55848312e-01
3.98404241e-01 -8.27325225e-01 -2.99752742e-01 6.08222708e-02] | [8.686554908752441, 5.659814357757568] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.