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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0c898c8c-e511-48ef-a89a-f53a27821204 | hyperbolic-active-learning-for-semantic | 2306.11180 | null | https://arxiv.org/abs/2306.11180v2 | https://arxiv.org/pdf/2306.11180v2.pdf | Hyperbolic Active Learning for Semantic Segmentation under Domain Shift | For the task of semantic segmentation (SS) under domain shift, active learning (AL) acquisition strategies based on image regions and pseudo labels are state-of-the-art (SoA). The presence of diverse pseudo-labels within a region identifies pixels between different classes, which is a labeling efficient active learning data acquisition strategy. However, by design, pseudo-label variations are limited to only select the contours of classes, limiting the final AL performance. We approach AL for SS in the Poincar\'e hyperbolic ball model for the first time and leverage the variations of the radii of pixel embeddings within regions as a novel data acquisition strategy. This stems from a novel geometric property of a hyperbolic space trained without enforced hierarchies, which we experimentally prove. Namely, classes are mapped into compact hyperbolic areas with a comparable intra-class radii variance, as the model places classes of increasing explainable difficulty at denser hyperbolic areas, i.e. closer to the Poincar\'e ball edge. The variation of pixel embedding radii identifies well the class contours, but they also select a few intra-class peculiar details, which boosts the final performance. Our proposed HALO (Hyperbolic Active Learning Optimization) surpasses the supervised learning performance for the first time in AL for SS under domain shift, by only using a small portion of labels (i.e., 1%). The extensive experimental analysis is based on two established benchmarks, i.e. GTAV $\rightarrow$ Cityscapes and SYNTHIA $\rightarrow$ Cityscapes, where we set a new SoA. The code will be released. | ['Fabio Galasso', 'Yu-Teng Li', 'Devin Guillory', 'Konstantinos Kallidromitis', 'Paolo Mandica', 'Luca Franco'] | 2023-06-19 | null | null | null | null | ['active-learning', 'pseudo-label', 'active-learning'] | ['methodology', 'miscellaneous', 'natural-language-processing'] | [ 2.55642951e-01 7.92143464e-01 -4.03291702e-01 -2.01796323e-01
-7.85476506e-01 -6.60045385e-01 4.16883737e-01 4.34542596e-01
-6.59600317e-01 4.23440337e-01 -3.60217541e-01 6.32616808e-04
-3.03381652e-01 -8.41001213e-01 -8.09576154e-01 -1.14070630e+00
-5.11655472e-02 5.05597353e-01 8.57231319e-01 -1.49655730e-01
3.09226096e-01 4.93477374e-01 -1.55440450e+00 3.11789215e-02
1.17757905e+00 1.19205534e+00 9.67927575e-02 7.92028233e-02
-1.77011475e-01 3.75031918e-01 -3.04055154e-01 -2.56343275e-01
5.72447836e-01 -2.68133074e-01 -6.83275461e-01 2.36543238e-01
4.71553892e-01 1.57774583e-01 1.10128857e-01 1.04900420e+00
2.45374098e-01 2.12221697e-01 7.58267283e-01 -1.11448407e+00
-4.42657381e-01 5.97863674e-01 -5.51012576e-01 3.44539098e-02
-1.41929492e-01 3.71925503e-01 1.29672432e+00 -1.08700836e+00
6.99158907e-01 8.62060070e-01 6.76748574e-01 4.26401377e-01
-1.39778030e+00 -6.90472186e-01 5.03293097e-01 1.16981320e-01
-1.69892025e+00 -1.41087413e-01 1.05457199e+00 -4.76509094e-01
4.72186118e-01 2.55492568e-01 8.56142163e-01 8.38643134e-01
-1.57663882e-01 1.01909578e+00 1.25604415e+00 -3.92358720e-01
6.30377769e-01 4.42616910e-01 3.25882554e-01 7.32210755e-01
2.96592683e-01 -1.01538092e-01 -5.44094026e-01 1.78224176e-01
8.69442999e-01 -1.44613966e-01 -3.71730208e-01 -1.13735080e+00
-9.38538551e-01 1.10958576e+00 8.14149618e-01 1.95136577e-01
1.29341260e-01 -2.67424524e-01 -7.91866623e-04 -1.98722780e-02
5.15846968e-01 6.88072205e-01 -5.63890994e-01 1.58943966e-01
-8.95239651e-01 6.48675710e-02 4.14384156e-01 1.07882786e+00
1.15717125e+00 -2.79231131e-01 1.14179760e-01 8.67570162e-01
3.04346025e-01 2.76161194e-01 1.82741135e-01 -7.72857785e-01
3.26337457e-01 1.25079584e+00 -1.68081790e-01 -8.58035743e-01
-4.53595757e-01 -6.09737277e-01 -5.30247450e-01 3.89809489e-01
7.24845171e-01 6.78367093e-02 -1.00877345e+00 1.38847589e+00
4.12756890e-01 -7.88589660e-03 -8.59375894e-02 7.67893195e-01
6.92162097e-01 4.17522162e-01 -5.87879866e-02 -2.66773313e-01
1.34863186e+00 -1.01090634e+00 -3.41399431e-01 -3.63942862e-01
7.26275742e-01 -4.14877921e-01 1.41907263e+00 4.31275159e-01
-9.19776797e-01 -5.97674370e-01 -1.12207592e+00 4.65675034e-02
-6.52816296e-01 -1.42511770e-01 4.90713477e-01 6.94712281e-01
-8.53585243e-01 4.41238493e-01 -8.74745250e-01 -1.18397713e-01
9.11841691e-01 3.12804759e-01 1.74231883e-02 2.88984895e-01
-1.03443253e+00 3.93101305e-01 5.13483167e-01 -9.22322869e-02
-6.09115064e-01 -1.05389416e+00 -1.06301951e+00 -1.09884828e-01
6.33133233e-01 -1.29012764e-03 6.04666591e-01 -1.13946223e+00
-1.46937561e+00 1.10163391e+00 4.22178775e-01 -5.64401984e-01
7.52121806e-01 -8.70708600e-02 -2.01972619e-01 3.34566265e-01
-1.27353834e-03 1.11455858e+00 9.68347132e-01 -1.51661670e+00
-7.42443264e-01 -5.97578585e-01 2.38151595e-01 3.43981057e-01
-5.77817023e-01 -3.80611688e-01 -4.21827078e-01 -5.55547595e-01
4.69582260e-01 -1.12647772e+00 -2.44527534e-01 2.68077075e-01
-3.55080575e-01 -3.20119798e-01 8.67187142e-01 -9.62130129e-02
1.26482332e+00 -2.39884686e+00 4.92962152e-02 4.20392722e-01
4.72655654e-01 2.66937334e-02 2.89457738e-01 -3.34426351e-02
-6.93176910e-02 1.31060302e-01 -8.08648765e-01 -2.27864340e-01
-9.24253166e-02 1.92073748e-01 -1.06369182e-01 5.57061851e-01
4.93273109e-01 7.42380440e-01 -8.85792971e-01 -7.85945237e-01
1.90673128e-01 1.54018447e-01 -4.96534497e-01 -4.80957329e-03
-2.64105856e-01 4.21917379e-01 -3.99045795e-01 3.94513756e-01
8.68318677e-01 -4.01808947e-01 -2.83056885e-01 -5.07612824e-02
-3.75850737e-01 -7.06769899e-02 -1.34345162e+00 1.76981390e+00
-1.91867054e-01 4.64962572e-01 -1.54891565e-01 -1.05646789e+00
1.28332925e+00 -1.54641733e-01 5.99568427e-01 -7.67501354e-01
-1.49913371e-01 3.95541400e-01 -4.47809137e-02 -2.82512009e-01
3.56510907e-01 2.07834795e-01 -2.18725070e-01 1.24689125e-01
-9.95118096e-02 -3.19415390e-01 1.41728073e-01 9.42732915e-02
8.67045879e-01 1.61123276e-01 2.20272988e-01 -6.90018058e-01
3.34926873e-01 1.23689301e-01 7.15544581e-01 6.80132151e-01
-3.96509856e-01 9.64145482e-01 6.18422449e-01 -2.02751905e-01
-9.83787894e-01 -1.33289731e+00 -7.78553843e-01 1.02909386e+00
7.53645897e-01 -7.95085356e-02 -8.40610027e-01 -1.08432531e+00
-1.91615447e-01 6.52658820e-01 -7.87278056e-01 -1.86147779e-01
-7.87652373e-01 -9.33434010e-01 2.53233612e-01 4.55073982e-01
7.33968437e-01 -1.03979802e+00 -9.05998230e-01 -1.17831208e-01
2.35640824e-01 -1.13317800e+00 -2.22350076e-01 5.56080818e-01
-8.03178251e-01 -1.09389293e+00 -8.77335548e-01 -9.11853135e-01
8.37243617e-01 -9.63789523e-02 1.11545968e+00 -1.57972634e-01
-2.64787912e-01 1.24062575e-01 -5.65228462e-01 -2.62958497e-01
-9.53094065e-02 4.11368221e-01 -4.44412649e-01 1.87776655e-01
1.12213463e-01 -2.44437382e-01 -1.01512277e+00 6.16550922e-01
-7.65325963e-01 1.72739953e-01 6.14373922e-01 7.05581546e-01
9.28802133e-01 9.78207216e-02 3.31956118e-01 -1.13748991e+00
-1.75972059e-01 -3.69215906e-01 -7.60545433e-01 9.80906710e-02
-7.46999443e-01 1.68446768e-02 4.91925508e-01 -5.35040855e-01
-9.37724054e-01 3.38066250e-01 9.27824154e-02 -4.59114388e-02
-3.47223133e-01 -2.23961100e-02 -2.94871837e-01 -1.19396351e-01
8.44827116e-01 -3.78236696e-02 -1.88064650e-01 -2.53586650e-01
3.50467086e-01 4.86624271e-01 8.80632848e-02 -4.29663390e-01
7.13396966e-01 7.47477829e-01 -1.29608437e-01 -7.72885561e-01
-8.44599485e-01 -6.39563799e-01 -8.22207451e-01 -1.79388627e-01
9.76266623e-01 -7.05285847e-01 -1.81183487e-01 4.10579026e-01
-5.73014438e-01 -6.69437706e-01 -7.50094473e-01 3.19802135e-01
-6.60550356e-01 1.76289007e-01 -3.67715299e-01 -7.00435162e-01
-1.26987742e-02 -1.24003601e+00 1.19194257e+00 4.23187107e-01
-1.73911914e-01 -1.01738167e+00 -1.51054099e-01 4.14646715e-01
-1.34740949e-01 3.07278037e-01 1.11185336e+00 -6.57263577e-01
-7.56120026e-01 2.74542421e-01 -3.32668245e-01 2.92877197e-01
-1.08980641e-01 -8.97839293e-03 -1.11388898e+00 -1.68199211e-01
-1.67846382e-01 -1.42720059e-01 1.03443205e+00 3.98139060e-01
1.29315412e+00 4.28351909e-02 -3.08907449e-01 6.55085802e-01
1.49294341e+00 3.83801073e-01 7.82899797e-01 4.28970903e-01
6.64460301e-01 6.93309665e-01 7.49086916e-01 1.57771185e-01
8.61684307e-02 7.14067042e-01 5.36412299e-01 -6.98422253e-01
-2.04810426e-01 1.26517385e-01 1.68377206e-01 5.55322528e-01
2.30773002e-01 2.39865296e-03 -1.11598599e+00 5.62269509e-01
-1.70787013e+00 -3.47714454e-01 -2.34133169e-01 2.33724380e+00
9.91019249e-01 7.11230218e-01 2.42651731e-01 4.67338383e-01
5.67658067e-01 2.12530300e-01 -7.51587927e-01 -1.25835374e-01
-3.20166320e-01 4.50898975e-01 6.86155558e-01 3.97134036e-01
-1.23730206e+00 9.85000789e-01 4.50106144e+00 1.28254926e+00
-1.05804062e+00 -9.62006021e-03 8.60990405e-01 1.17880702e-01
-1.81450233e-01 3.58081721e-02 -9.84592915e-01 5.59256852e-01
3.06019157e-01 4.44701552e-01 -9.36489254e-02 9.30304766e-01
-7.15022981e-02 -2.06057847e-01 -1.07504761e+00 8.68953407e-01
7.37251760e-03 -1.22605324e+00 -8.73651356e-02 6.79432005e-02
8.91857624e-01 -1.25678435e-01 3.33944261e-01 1.50230005e-01
1.45783663e-01 -7.28976250e-01 9.68044937e-01 1.59098253e-01
8.06033611e-01 -7.29455531e-01 5.07094622e-01 2.77474970e-01
-1.11692393e+00 -3.55825275e-01 -2.06070080e-01 4.50718194e-01
-1.13447063e-01 5.57385743e-01 -7.11844683e-01 3.73402238e-01
9.65082109e-01 7.17411995e-01 -1.06489491e+00 9.25905526e-01
-2.94672456e-02 5.77649176e-01 -2.51357615e-01 1.19812936e-02
3.37908596e-01 -5.01898527e-01 6.09710932e-01 1.08045638e+00
5.76762557e-02 -8.04211944e-02 1.96564496e-01 1.06011629e+00
1.59661025e-01 2.29119331e-01 -3.53926629e-01 2.10715339e-01
3.27382714e-01 1.16174519e+00 -1.41491401e+00 -9.21943858e-02
-3.01544040e-01 7.12961435e-01 2.58947819e-01 3.43023866e-01
-8.96725357e-01 -4.05814946e-01 2.96598345e-01 5.15228152e-01
3.47418576e-01 -1.21062405e-01 -7.77274370e-01 -7.89316475e-01
-8.22832733e-02 -5.20989478e-01 5.27032733e-01 -4.41836417e-01
-1.01128888e+00 5.34039497e-01 2.92324156e-01 -1.36832571e+00
3.24130803e-01 -6.55948162e-01 -2.27164716e-01 2.56356180e-01
-1.32752907e+00 -9.80430067e-01 -5.54130435e-01 4.98832703e-01
6.88971281e-01 -8.26518834e-02 6.13922000e-01 1.89497545e-01
-7.31039286e-01 7.25646019e-01 1.60146773e-01 2.56539792e-01
4.56354737e-01 -1.51932645e+00 1.18399926e-01 6.98832452e-01
3.75043958e-01 3.31826150e-01 4.73676443e-01 -3.55173290e-01
-6.80037916e-01 -9.97440815e-01 4.53716189e-01 -3.56091708e-01
5.73772311e-01 -9.02513742e-01 -1.02862430e+00 3.76060009e-01
-9.30264778e-03 1.35438547e-01 4.76849079e-01 -9.23442543e-02
-2.39342511e-01 -3.42460454e-01 -1.19622850e+00 7.38645554e-01
1.33975410e+00 -2.10854501e-01 -3.38215888e-01 1.61794826e-01
8.68854642e-01 -2.47443438e-01 -7.17364967e-01 6.65004075e-01
1.36451691e-01 -1.09725082e+00 1.06592584e+00 -1.06751628e-01
3.18939686e-01 -2.64343381e-01 1.71384573e-01 -1.07492805e+00
-2.16350108e-01 -4.19315517e-01 1.44272536e-01 1.27132392e+00
7.57432818e-01 -8.41744781e-01 1.08793807e+00 2.30720565e-01
-3.90113980e-01 -1.08513749e+00 -1.00634789e+00 -7.57121742e-01
3.75030994e-01 -2.83447862e-01 4.46921289e-01 9.38364923e-01
-3.72042835e-01 9.67162773e-02 5.29479146e-01 3.54346111e-02
7.29161978e-01 2.08741814e-01 4.03946847e-01 -1.50358713e+00
-9.27425027e-02 -6.69139564e-01 -4.55728352e-01 -1.21135890e+00
-3.11844032e-02 -7.80863881e-01 1.31897420e-01 -1.02072215e+00
4.42572543e-03 -1.17516184e+00 -3.36002499e-01 3.70911628e-01
-8.10121596e-02 4.34098274e-01 -1.25816176e-02 2.41962478e-01
-5.92257261e-01 5.33872008e-01 1.30965328e+00 -2.57200003e-01
-4.66666162e-01 -1.52546391e-01 -5.06432414e-01 8.58502626e-01
7.90409565e-01 -4.49671775e-01 -5.55937588e-01 -1.05491102e-01
3.49000275e-01 -5.42181253e-01 3.85938495e-01 -1.04199624e+00
4.14469317e-02 -9.06407982e-02 1.80664197e-01 -4.51569289e-01
3.96571428e-01 -9.73352969e-01 -2.20667958e-01 2.57706553e-01
-4.76721138e-01 -4.46164012e-01 -2.66276095e-02 5.51827908e-01
4.80677933e-02 -2.47252956e-01 1.31972647e+00 -2.27711350e-02
-9.86246049e-01 3.29801977e-01 5.03503866e-02 4.40091103e-01
1.29896653e+00 -8.42254996e-01 -2.60877550e-01 2.22920388e-01
-7.45409846e-01 8.61520618e-02 6.10140681e-01 1.98826745e-01
4.19601589e-01 -1.07287848e+00 -4.46454018e-01 4.99428809e-01
5.32788515e-01 7.93645442e-01 1.09577782e-01 9.79275465e-01
-6.73713326e-01 9.40756407e-03 -1.15074683e-02 -1.16690230e+00
-1.04349744e+00 4.15790915e-01 4.13337886e-01 -1.03259481e-01
-8.64547729e-01 1.05456889e+00 7.03197837e-01 -4.07524586e-01
2.51314223e-01 -4.04220849e-01 -4.66289163e-01 4.00865793e-01
-1.27731944e-02 3.82859141e-01 2.23650858e-01 -5.75662911e-01
-1.93765178e-01 8.07770133e-01 -5.37084714e-02 1.28179044e-01
1.17391813e+00 -2.15651020e-01 2.58973986e-01 6.45912468e-01
1.19290328e+00 4.78669256e-02 -1.68597007e+00 -2.11146325e-01
2.10933387e-01 -4.12206173e-01 5.76657914e-02 -5.57789981e-01
-1.29302943e+00 7.76272535e-01 1.06954324e+00 3.67657393e-01
9.91969407e-01 3.94454688e-01 5.65774202e-01 -1.13779485e-01
4.31396395e-01 -1.61667371e+00 3.90783459e-01 1.24807797e-01
7.28058994e-01 -1.26785815e+00 -1.81241520e-02 -9.35029924e-01
-7.44729459e-01 9.17544842e-01 7.22406089e-01 -2.39823088e-01
7.23832369e-01 1.69895634e-01 9.63589475e-02 -2.61125028e-01
-1.38121367e-01 -3.45442474e-01 1.89225763e-01 5.91438770e-01
4.42813560e-02 1.18119411e-01 -1.58910722e-01 3.86108965e-01
-2.24884063e-01 -5.29644489e-01 2.19097644e-01 6.43515468e-01
-4.64461386e-01 -6.75894558e-01 -3.11094105e-01 2.43608296e-01
1.41135836e-02 3.93955745e-02 -4.53970432e-01 1.11663544e+00
7.50317752e-01 7.22343981e-01 5.99409401e-01 2.84548057e-03
2.94233471e-01 -6.10779338e-02 3.97836834e-01 -7.04539776e-01
-5.12413025e-01 1.53406888e-01 -2.26807013e-01 -3.72790545e-01
-2.42543727e-01 -8.07951212e-01 -1.77615166e+00 1.88069180e-01
-4.55669761e-01 1.37124322e-02 1.57949209e-01 6.46320283e-01
1.19930468e-01 1.89339295e-01 7.41044462e-01 -4.01229858e-01
-3.29346985e-01 -5.58085918e-01 -7.50152051e-01 6.76020443e-01
7.88743868e-02 -1.00794613e+00 -7.81555533e-01 1.03941612e-01] | [9.616921424865723, 1.1064866781234741] |
5c74a4f4-a470-407c-92b8-c7879368b908 | rethinking-network-design-and-local-geometry-1 | 2202.07123 | null | https://arxiv.org/abs/2202.07123v2 | https://arxiv.org/pdf/2202.07123v2.pdf | Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework | Point cloud analysis is challenging due to irregularity and unordered data structure. To capture the 3D geometries, prior works mainly rely on exploring sophisticated local geometric extractors using convolution, graph, or attention mechanisms. These methods, however, incur unfavorable latency during inference, and the performance saturates over the past few years. In this paper, we present a novel perspective on this task. We notice that detailed local geometrical information probably is not the key to point cloud analysis -- we introduce a pure residual MLP network, called PointMLP, which integrates no sophisticated local geometrical extractors but still performs very competitively. Equipped with a proposed lightweight geometric affine module, PointMLP delivers the new state-of-the-art on multiple datasets. On the real-world ScanObjectNN dataset, our method even surpasses the prior best method by 3.3% accuracy. We emphasize that PointMLP achieves this strong performance without any sophisticated operations, hence leading to a superior inference speed. Compared to most recent CurveNet, PointMLP trains 2x faster, tests 7x faster, and is more accurate on ModelNet40 benchmark. We hope our PointMLP may help the community towards a better understanding of point cloud analysis. The code is available at https://github.com/ma-xu/pointMLP-pytorch. | ['Yun Fu', 'Haoxi Ran', 'Haoxuan You', 'Can Qin', 'Xu Ma'] | 2022-02-15 | rethinking-network-design-and-local-geometry | https://openreview.net/forum?id=3Pbra-_u76D | https://openreview.net/pdf?id=3Pbra-_u76D | iclr-2022-4 | ['point-cloud-segmentation'] | ['computer-vision'] | [-2.61135012e-01 -8.08476880e-02 -5.01735434e-02 -1.97480366e-01
-8.50898504e-01 -4.47522372e-01 5.52646816e-01 2.01577112e-01
-2.82332033e-01 3.89202774e-01 -2.78635830e-01 -5.68140924e-01
-1.87904790e-01 -1.09476781e+00 -1.23745096e+00 -6.27398491e-01
-1.13389276e-01 6.76256835e-01 2.60341376e-01 -2.07176700e-01
3.95132184e-01 1.08057153e+00 -1.55854964e+00 2.88902074e-02
9.06502485e-01 1.16822183e+00 1.83907092e-01 5.68112910e-01
-3.94181401e-01 2.64720947e-01 -1.68011159e-01 -4.37133878e-01
3.11748743e-01 4.94490534e-01 -7.12951481e-01 -4.81972277e-01
8.65765154e-01 -4.57872361e-01 -3.16276222e-01 9.55721438e-01
4.37158257e-01 -8.41401294e-02 2.86837816e-01 -1.19998240e+00
-5.19564092e-01 5.89420199e-01 -5.16042411e-01 -7.24100620e-02
-6.99124560e-02 3.01387906e-01 1.04779017e+00 -1.27403998e+00
4.86895055e-01 1.01450872e+00 1.04463327e+00 1.24842219e-01
-1.03798485e+00 -7.78732777e-01 1.45205617e-01 2.49915525e-01
-1.73883080e+00 -1.50022358e-01 9.32672918e-01 -3.08565766e-01
1.20860004e+00 2.90606052e-01 7.62673974e-01 9.24209535e-01
1.36222020e-01 7.24651694e-01 6.23234153e-01 2.15247236e-02
1.31986186e-01 -4.09901202e-01 1.55802980e-01 7.19813049e-01
1.08110338e-01 1.38467014e-01 -2.47724488e-01 -2.33885646e-01
1.02511370e+00 2.43909761e-01 -2.93050557e-01 -2.88502753e-01
-1.29773164e+00 7.37566769e-01 9.32325006e-01 1.43722624e-01
-3.54838610e-01 6.21606588e-01 1.70999095e-01 2.13842422e-01
6.93811953e-01 3.45812082e-01 -4.88484859e-01 -2.46344164e-01
-9.81026232e-01 5.02436936e-01 6.57364845e-01 1.22456622e+00
8.73456240e-01 7.87586533e-03 2.41667956e-01 6.17289305e-01
3.76852483e-01 7.33916163e-01 -2.88785487e-01 -8.48287225e-01
6.38019979e-01 7.31265724e-01 -1.91594183e-01 -1.25104797e+00
-5.35412729e-01 -7.77790844e-01 -1.06981230e+00 5.02745152e-01
3.40934396e-01 1.34985656e-01 -8.36003780e-01 1.13122201e+00
2.45790482e-01 4.68654156e-01 -4.81233448e-01 8.34785581e-01
1.05441642e+00 7.04410374e-01 -1.30440891e-01 4.36687320e-01
1.14623511e+00 -8.19626451e-01 -1.70577899e-01 2.97443271e-02
4.89635259e-01 -6.90539062e-01 1.06479156e+00 4.66040760e-01
-1.21315742e+00 -5.39749086e-01 -1.14152193e+00 -3.01695406e-01
-4.54758823e-01 8.05238411e-02 8.89999390e-01 2.06269816e-01
-1.26127875e+00 9.29672837e-01 -1.05637109e+00 -3.03625643e-01
7.11978376e-01 5.97287118e-01 -1.52748212e-01 -1.20557211e-01
-6.39287710e-01 6.09051526e-01 1.51130065e-01 3.07601005e-01
-3.79112095e-01 -1.17620194e+00 -6.83349252e-01 1.68571860e-01
4.16294843e-01 -9.22237575e-01 1.14619088e+00 -1.94049060e-01
-1.47373068e+00 6.11074746e-01 -1.20612741e-01 -4.27023590e-01
6.13215506e-01 -3.37120295e-01 -7.77463317e-02 9.89669114e-02
-2.61483520e-01 7.06065416e-01 6.47604823e-01 -1.23426723e+00
-3.46329212e-01 -4.08488870e-01 9.53145847e-02 -2.37196404e-02
8.93529877e-02 -1.87792733e-01 -8.96875143e-01 -5.80828249e-01
3.30934912e-01 -9.56604004e-01 -2.93679982e-01 2.54273295e-01
-3.80781114e-01 -5.42331219e-01 8.46009374e-01 -3.23068619e-01
8.28958035e-01 -2.12957597e+00 -3.05734798e-02 3.23173165e-01
6.06113434e-01 2.78769583e-01 -1.51845522e-03 4.97892588e-01
2.83284783e-02 2.68464178e-01 -5.83544314e-01 -6.54514492e-01
1.34448946e-01 1.63463861e-01 -5.15861869e-01 5.96726596e-01
3.41060877e-01 1.26640153e+00 -6.22500718e-01 -3.78030151e-01
6.60686910e-01 7.89386272e-01 -7.83361077e-01 -2.91924268e-01
-3.64609182e-01 3.27039540e-01 -5.56994259e-01 9.74590182e-01
1.25703800e+00 -5.20267665e-01 -4.88669395e-01 -2.25133464e-01
-3.38128150e-01 1.30002514e-01 -9.37819719e-01 2.05368543e+00
-4.40806419e-01 6.33460522e-01 4.04838584e-02 -6.94057882e-01
9.17451024e-01 3.22956182e-02 6.21755779e-01 -6.45202279e-01
1.81651309e-01 2.51349211e-01 -1.97827086e-01 -1.80496410e-01
5.20471096e-01 2.67128080e-01 1.66844994e-01 -2.90918425e-02
-6.73334077e-02 -4.34398055e-01 -2.79228240e-01 1.00888290e-01
1.08373690e+00 4.28848207e-01 -5.70346043e-02 -4.22624439e-01
3.35417181e-01 5.68013676e-02 1.96443662e-01 8.23553622e-01
1.95962399e-01 8.72083306e-01 3.10911715e-01 -6.55301988e-01
-1.06822371e+00 -1.21277392e+00 -4.76562828e-01 6.83439136e-01
3.29094797e-01 -6.53832257e-01 -4.11181837e-01 -4.06437069e-01
2.82496065e-01 6.47986591e-01 -2.37804562e-01 2.55031735e-01
-8.94965053e-01 -6.75310254e-01 7.17297912e-01 6.85279250e-01
5.06132066e-01 -9.12946045e-01 -2.44119033e-01 2.69219369e-01
1.23285890e-01 -1.24891222e+00 1.29575096e-02 -1.43280610e-01
-1.24525762e+00 -8.29800487e-01 -3.96501124e-01 -4.00756478e-01
5.19660532e-01 8.67027268e-02 1.46884346e+00 5.08036852e-01
2.20492445e-02 -7.31489738e-04 -2.04150587e-01 -5.68387210e-01
5.61265834e-02 5.91460764e-01 -2.87214458e-01 -3.12121242e-01
3.49179655e-01 -1.04330134e+00 -7.26723909e-01 8.85270014e-02
-7.16140866e-01 1.76128700e-01 7.14235842e-01 3.30769002e-01
7.47231245e-01 -1.42322764e-01 6.10933676e-02 -6.92925811e-01
1.65647328e-01 -4.40615624e-01 -8.64115596e-01 -2.60914564e-01
-4.32185948e-01 4.33519669e-03 7.20080793e-01 6.47322983e-02
-7.00295925e-01 2.70981528e-02 -7.97304988e-01 -6.88674152e-01
-4.15253341e-01 4.77276951e-01 -1.14802066e-02 -5.14730513e-01
4.51722115e-01 6.04485609e-02 -1.73704416e-01 -7.04606235e-01
2.95008838e-01 2.27244571e-01 5.38507700e-01 -7.49948919e-01
1.15346134e+00 8.90733063e-01 1.89981073e-01 -1.01476204e+00
-3.51791114e-01 -4.72317964e-01 -7.74575233e-01 -1.69490725e-01
7.09143877e-01 -8.20864558e-01 -1.11011732e+00 4.96405214e-01
-1.52347887e+00 -2.83917457e-01 -1.60488307e-01 3.00819367e-01
-3.73965025e-01 3.40337515e-01 -6.63369060e-01 -5.62200010e-01
-4.57920492e-01 -1.28064167e+00 1.49211645e+00 -1.29999369e-01
2.06462711e-01 -8.32663774e-01 -2.29946107e-01 7.93441758e-02
5.00896931e-01 5.17561316e-01 6.99660897e-01 -3.23939085e-01
-1.33828413e+00 -2.29998946e-01 -5.91882050e-01 9.54054594e-02
-3.40007782e-01 2.76661456e-01 -9.90373731e-01 -1.20040134e-01
-2.07638405e-02 1.33247733e-01 9.35044050e-01 3.09865087e-01
1.80200851e+00 -9.29199755e-02 -3.97543192e-01 1.27035880e+00
1.77511346e+00 -1.56339288e-01 8.29169333e-01 2.25595742e-01
1.25875747e+00 1.70928150e-01 1.56004414e-01 2.50864089e-01
6.63678825e-01 6.62031829e-01 9.84119177e-01 -3.80835325e-01
-4.51461151e-02 -1.17150165e-01 -2.14090832e-02 1.17990780e+00
-6.90866292e-01 -2.37148955e-01 -1.25828826e+00 3.01114857e-01
-1.85852134e+00 -7.25598752e-01 -7.47829735e-01 1.96846008e+00
3.14654499e-01 3.39219421e-01 -2.42344111e-01 -3.03515084e-02
3.31727415e-01 3.20581049e-01 -4.67141151e-01 -1.47464350e-01
-6.47242218e-02 5.63354075e-01 9.35367823e-01 5.23981094e-01
-1.02857959e+00 1.07122731e+00 5.49213076e+00 1.06476033e+00
-1.28977108e+00 7.12410435e-02 2.07378238e-01 -5.58946468e-02
-2.66453207e-01 1.49724474e-02 -8.13607872e-01 5.18643379e-01
7.24425614e-01 2.12135270e-01 4.31007296e-01 9.43343282e-01
1.28424421e-01 3.11828405e-01 -9.39554274e-01 1.16301942e+00
-9.70773697e-02 -1.58838975e+00 5.87655902e-02 4.35037971e-01
3.00110430e-01 9.72509623e-01 -2.79310439e-02 3.43018830e-01
1.79305017e-01 -1.11650074e+00 6.97656870e-01 7.92227685e-01
7.61672437e-01 -8.43859494e-01 7.41014183e-01 3.83959830e-01
-1.36281013e+00 3.62096131e-01 -5.30698240e-01 -1.37203127e-01
2.31934965e-01 7.68860221e-01 -6.74208343e-01 9.65533197e-01
8.95686507e-01 8.71267796e-01 -6.92797363e-01 1.22832966e+00
-5.02297655e-02 6.44636691e-01 -8.81204665e-01 1.22121491e-01
4.08982754e-01 -2.37181962e-01 6.90372825e-01 1.07992351e+00
4.96820122e-01 9.62083191e-02 1.00101434e-01 1.14126313e+00
-1.36939898e-01 7.66346082e-02 -6.87462568e-01 3.28217715e-01
3.35599542e-01 1.19579303e+00 -8.36511672e-01 -3.04234385e-01
-4.82307613e-01 6.16849244e-01 4.48240280e-01 2.34675527e-01
-1.00222635e+00 -3.85074347e-01 6.96980596e-01 2.00550228e-01
3.84929150e-01 -7.18424857e-01 -6.82866275e-01 -1.07793200e+00
2.15360805e-01 -4.61417377e-01 -2.22572386e-02 -8.10334980e-01
-1.23828101e+00 7.53289700e-01 9.72965509e-02 -1.29676378e+00
9.36484858e-02 -8.77055466e-01 -7.47802198e-01 6.66297197e-01
-1.63356614e+00 -1.30803013e+00 -6.53775513e-01 4.71392363e-01
5.01877010e-01 1.65408164e-01 4.89030421e-01 4.99630153e-01
-2.48355359e-01 4.42953110e-01 3.27052362e-02 5.48131838e-02
2.59030193e-01 -1.22655892e+00 1.08437967e+00 6.24806881e-01
1.25999361e-01 6.44003689e-01 5.04746199e-01 -6.31582737e-01
-1.69268274e+00 -1.18091476e+00 6.34538651e-01 -8.72883081e-01
6.36819243e-01 -5.22942007e-01 -1.21330881e+00 6.81015015e-01
4.93690819e-02 1.42699793e-01 1.21693082e-01 2.02662930e-01
-1.91098452e-01 -5.88537678e-02 -7.35490024e-01 6.04449332e-01
1.47573912e+00 -3.53252590e-01 -4.15615052e-01 3.67960781e-01
1.03715169e+00 -7.67321765e-01 -1.00283515e+00 8.05196464e-01
2.75162339e-01 -9.94369566e-01 1.22603679e+00 -9.53745171e-02
4.28875834e-01 -3.85149032e-01 -7.46171623e-02 -9.88182068e-01
-4.82373625e-01 -3.41812402e-01 -2.36447558e-01 8.96370769e-01
3.26679230e-01 -9.04609203e-01 9.44731593e-01 2.76379585e-01
-6.32661223e-01 -1.03003454e+00 -8.99950266e-01 -8.02531242e-01
2.70409584e-01 -1.11457539e+00 1.23344862e+00 9.50942039e-01
-5.47062397e-01 1.91575196e-02 -3.03250458e-02 5.23329496e-01
7.50186324e-01 2.05248401e-01 1.04408407e+00 -1.50030756e+00
-4.74788584e-02 -7.98128664e-01 -5.26054978e-01 -1.27666140e+00
7.59120658e-02 -1.24122858e+00 -3.14841062e-01 -1.57421041e+00
-3.66687566e-01 -7.44370520e-01 -1.50089622e-01 4.95785862e-01
1.25960410e-01 5.23001194e-01 2.60073543e-01 3.34040016e-01
-3.61961186e-01 7.02211440e-01 1.26791883e+00 -1.68855116e-01
-1.07664123e-01 2.48813555e-02 -4.95304197e-01 8.80052388e-01
8.92559886e-01 -3.96006882e-01 -3.40150632e-02 -9.28341210e-01
4.35947627e-01 -1.79874912e-01 8.48451853e-01 -1.25835657e+00
5.62475979e-01 8.93532559e-02 2.63483465e-01 -1.29755187e+00
5.37658334e-01 -1.01510096e+00 2.59014904e-01 1.70518413e-01
3.62923205e-01 3.08846682e-01 4.47417557e-01 3.70503575e-01
-1.27063096e-01 6.40185699e-02 4.05340165e-01 -1.00915417e-01
-5.62959075e-01 8.94374669e-01 2.76209861e-01 -3.86654913e-01
6.49617970e-01 -2.45538324e-01 -4.26164359e-01 -5.70874885e-02
-5.21194339e-01 1.99772999e-01 6.26911700e-01 3.80326211e-01
6.62766457e-01 -1.35222673e+00 -8.49177599e-01 3.25646639e-01
6.39468133e-02 7.71005988e-01 4.22433406e-01 9.00232673e-01
-1.15147412e+00 3.70161474e-01 1.31217241e-01 -1.21772444e+00
-8.80692899e-01 4.42464173e-01 1.77899688e-01 4.07800376e-02
-1.25369561e+00 7.02822387e-01 2.06081599e-01 -5.87833762e-01
2.62047462e-02 -8.24423552e-01 1.63560331e-01 -2.83950686e-01
1.40184417e-01 3.78043592e-01 5.91154933e-01 -5.82335889e-01
-4.03102219e-01 1.01849294e+00 3.23470756e-02 2.92678654e-01
1.61334658e+00 2.75805891e-01 -2.42491797e-01 2.60886014e-01
1.25238335e+00 2.24151030e-01 -1.17175245e+00 -1.73246726e-01
-1.22973487e-01 -5.44210494e-01 2.02548906e-01 -3.50659370e-01
-1.28074479e+00 8.58016908e-01 3.13349932e-01 2.29565486e-01
8.33102703e-01 2.10304782e-01 8.60563576e-01 5.24055481e-01
6.44500315e-01 -5.86505830e-01 -4.76058036e-01 6.94161177e-01
1.17185819e+00 -1.19482434e+00 1.85248271e-01 -6.50877893e-01
-5.55159040e-02 1.06970787e+00 5.39454281e-01 -6.00874901e-01
8.75766575e-01 3.97867799e-01 -1.65060684e-01 -6.99960768e-01
-5.95706224e-01 -3.76747437e-02 3.98599327e-01 2.82012224e-01
1.23199627e-01 7.06686974e-02 2.35748842e-01 1.82116628e-01
-7.57542133e-01 -1.21190675e-01 1.16684332e-01 6.44478500e-01
-1.57371342e-01 -9.82102394e-01 -4.12646949e-01 5.02609193e-01
-2.65099972e-01 -2.96856284e-01 -1.71491906e-01 1.14793086e+00
9.09648016e-02 4.08748627e-01 3.80456209e-01 -3.32410365e-01
4.86178964e-01 -1.51081011e-01 3.52323681e-01 -2.27146789e-01
-5.14458299e-01 -4.18649390e-02 -3.04183543e-01 -9.90232408e-01
-1.99745536e-01 -5.79565287e-01 -1.21756124e+00 -8.66003811e-01
-2.53112197e-01 -1.34594709e-01 9.64591265e-01 7.31328845e-01
6.37785435e-01 5.85622728e-01 2.56512493e-01 -1.34824312e+00
-9.74622667e-02 -8.08799565e-01 -1.60130829e-01 5.71697624e-03
4.24462438e-01 -5.40800571e-01 -2.72899270e-01 -3.54621708e-01] | [7.976972579956055, -3.5105786323547363] |
23f61eee-c943-4b0a-ba85-3caa5c400b10 | exploiting-language-model-for-efficient | 2107.12168 | null | https://arxiv.org/abs/2107.12168v3 | https://arxiv.org/pdf/2107.12168v3.pdf | Exploiting Language Model for Efficient Linguistic Steganalysis | Recent advances in linguistic steganalysis have successively applied CNN, RNN, GNN and other efficient deep models for detecting secret information in generative texts. These methods tend to seek stronger feature extractors to achieve higher steganalysis effects. However, we have found through experiments that there actually exists significant difference between automatically generated stego texts and carrier texts in terms of the conditional probability distribution of individual words. Such kind of difference can be naturally captured by the language model used for generating stego texts. Through further experiments, we conclude that this ability can be transplanted to a text classifier by pre-training and fine-tuning to improve the detection performance. Motivated by this insight, we propose two methods for efficient linguistic steganalysis. One is to pre-train a language model based on RNN, and the other is to pre-train a sequence autoencoder. The results indicate that the two methods have different degrees of performance gain compared to the randomly initialized RNN, and the convergence speed is significantly accelerated. Moreover, our methods achieved the best performance compared to related works, while providing a solution for real-world scenario where there are more cover texts than stego texts. | ['Xinpeng Zhang', 'Guorui Feng', 'Hanzhou Wu', 'Biao Yi'] | 2021-07-26 | null | null | null | null | ['steganalysis'] | ['computer-vision'] | [ 4.85662997e-01 1.87488824e-01 1.57904595e-01 7.06993937e-02
-4.18508202e-01 -3.09430003e-01 8.31466615e-01 -3.47237051e-01
-2.33093098e-01 5.54264128e-01 4.01609719e-01 -5.46025813e-01
6.75220013e-01 -1.08264470e+00 -6.35458708e-01 -8.51200938e-01
-2.09285435e-03 1.39123529e-01 9.35861766e-02 -5.97763538e-01
2.48337120e-01 1.97373871e-02 -1.38408756e+00 3.68286282e-01
8.36650670e-01 3.97208691e-01 2.95737118e-01 8.29541504e-01
-1.84797943e-01 9.38770533e-01 -9.02670145e-01 -4.53979939e-01
2.13834032e-01 -1.06292510e+00 -5.41838169e-01 -2.34240410e-03
-3.72564077e-01 -3.78178895e-01 -5.39760888e-01 1.49044251e+00
6.02002084e-01 -2.04381034e-01 7.29627967e-01 -7.55165040e-01
-9.49301422e-01 1.41540658e+00 -5.84479094e-01 3.85655016e-02
1.95099428e-01 4.25339550e-01 7.99099028e-01 -5.25083244e-01
3.03831518e-01 1.48003459e+00 6.19047761e-01 6.76295042e-01
-6.23431623e-01 -9.11955357e-01 -3.88430953e-01 -7.22764060e-02
-1.16405475e+00 -3.86918813e-01 8.72241974e-01 -1.56464666e-01
7.88138449e-01 2.93639749e-01 6.52993262e-01 1.54991031e+00
4.51805890e-01 8.79718661e-01 1.00504184e+00 -8.00499558e-01
-2.16352671e-01 4.89785910e-01 -1.74232081e-01 8.22091341e-01
4.07874197e-01 4.12745863e-01 6.65226057e-02 -5.57706878e-02
5.95951259e-01 -1.37048975e-01 -5.22791207e-01 2.23256841e-01
-1.17687237e+00 1.27435017e+00 1.51456371e-01 9.27091718e-01
-2.70149738e-01 2.16094106e-01 4.35789078e-01 5.27374446e-01
3.93096864e-01 4.52154547e-01 7.66426604e-03 7.57238790e-02
-9.80801344e-01 -1.95472717e-01 1.05357611e+00 7.32303798e-01
6.56904876e-01 6.10144913e-01 -4.92637157e-02 3.72858167e-01
5.83985150e-01 8.07662129e-01 1.18846703e+00 -7.73771256e-02
6.18487537e-01 2.92684227e-01 -3.65261614e-01 -1.53416133e+00
5.77525869e-02 -6.00471973e-01 -1.08396780e+00 -1.55577943e-01
1.59758344e-01 -4.19044256e-01 -9.65476751e-01 1.44108033e+00
-1.60330281e-01 3.50531846e-01 5.14884830e-01 6.10050261e-01
6.56913936e-01 1.04100883e+00 -8.14178213e-02 -1.15923353e-01
1.39150584e+00 -9.12321091e-01 -9.06352222e-01 -2.30309010e-01
8.85036826e-01 -8.72321427e-01 8.11735153e-01 1.87135220e-01
-5.62954009e-01 -5.38919568e-01 -1.25179887e+00 4.38873827e-01
-4.23346430e-01 1.57411858e-01 3.19142014e-01 1.21681178e+00
-1.03495038e+00 5.69815099e-01 -4.47782636e-01 -2.50609517e-01
1.17026471e-01 2.35705182e-01 1.48541508e-02 4.70629930e-01
-1.88032711e+00 7.39495635e-01 1.21144927e+00 1.71048447e-01
-1.01465344e+00 9.28046927e-02 -8.58464956e-01 1.95083976e-01
1.60280570e-01 -4.15626019e-01 8.82511795e-01 -1.32605922e+00
-1.72857404e+00 6.28862441e-01 2.05566853e-01 -6.04459763e-01
4.79923546e-01 1.76809341e-01 -7.17587352e-01 9.12903026e-02
-2.26015568e-01 1.79727852e-01 1.30382907e+00 -1.18189108e+00
-5.89219093e-01 9.91814137e-02 -3.17229658e-01 -2.57325321e-02
-6.31757259e-01 1.05964788e-03 -4.84358929e-02 -8.68793726e-01
-1.64550319e-01 -7.92192817e-01 -1.45620897e-01 -9.47907984e-01
-5.52350521e-01 -2.04491705e-01 1.12593305e+00 -9.88831043e-01
1.38597369e+00 -2.04554415e+00 -2.40091845e-01 4.18799073e-01
3.24007690e-01 8.68557692e-01 -1.09846339e-01 6.33206367e-01
-1.77826121e-01 4.54701066e-01 -1.07168399e-01 -1.84790026e-02
-8.76078680e-02 -7.24696741e-02 -5.82739234e-01 4.50520098e-01
-2.24109343e-03 1.18294454e+00 -9.05712068e-01 -6.16411150e-01
2.07110912e-01 4.64233100e-01 -1.87592834e-01 1.15791276e-01
-3.17667067e-01 2.97091335e-01 -5.76913297e-01 2.25501567e-01
4.59858716e-01 -4.34615761e-01 3.85584116e-01 7.38324299e-02
3.07480037e-01 1.58529729e-01 -8.59786749e-01 1.08709204e+00
-3.13613772e-01 9.07821298e-01 -3.53972673e-01 -1.12209761e+00
1.15410638e+00 6.30927503e-01 -2.74138059e-02 -4.45767552e-01
8.95954788e-01 3.18672508e-01 1.02934696e-01 -7.00759888e-01
6.09389007e-01 -2.77634114e-01 -3.20230015e-02 7.99732089e-01
-1.04399316e-01 2.12712716e-02 -1.71667546e-01 3.73326465e-02
8.52619112e-01 -2.11808994e-01 4.36316580e-01 -4.65081781e-02
9.21528578e-01 -2.27322772e-01 -3.56326513e-02 1.04461384e+00
1.21882327e-01 8.11358020e-02 3.57081741e-01 -1.45566985e-01
-1.24812877e+00 -3.14939678e-01 4.55120623e-01 8.60488057e-01
1.69596165e-01 -9.48051438e-02 -1.03620017e+00 -7.10686862e-01
-5.57032883e-01 9.03368711e-01 -4.72616583e-01 -4.63059694e-01
-8.78580093e-01 -9.83827293e-01 1.14640510e+00 8.87049064e-02
8.84773612e-01 -1.52841806e+00 -4.10524756e-01 3.00250351e-01
-3.43816370e-01 -1.03677452e+00 -4.34773475e-01 -1.64152488e-01
-7.02646554e-01 -6.59053147e-01 -7.37337947e-01 -1.13081729e+00
5.18805325e-01 2.03968838e-01 7.56556749e-01 4.21222478e-01
8.53679702e-02 -1.42566085e-01 -8.66472900e-01 -4.09636945e-01
-1.35719204e+00 3.73617321e-01 -1.94412038e-01 1.65896297e-01
4.74591345e-01 -4.13070112e-01 -3.43359530e-01 -6.31035343e-02
-1.25910819e+00 -1.86031442e-02 1.01399064e+00 9.30308044e-01
-2.33998016e-01 6.02594256e-01 3.73356521e-01 -9.48272347e-01
1.04461813e+00 -4.91623819e-01 -3.57725948e-01 3.14157605e-01
-6.35502994e-01 5.02881706e-01 8.72494698e-01 -6.03750765e-01
-9.33553815e-01 -3.77854377e-01 -3.70721102e-01 -1.96878761e-01
-3.30200270e-02 6.24324083e-01 -5.25874458e-02 -2.02102307e-02
6.70786560e-01 9.80955720e-01 1.56195045e-01 -6.16746917e-02
2.06581920e-01 1.04848170e+00 1.52732268e-01 -1.20023335e-03
1.12358713e+00 2.79583544e-01 -4.49228853e-01 -9.09695446e-01
-4.85891938e-01 2.22993493e-02 -1.27776027e-01 4.85673249e-02
8.41686547e-01 -7.62046218e-01 -6.90090537e-01 7.71328270e-01
-1.23755705e+00 -9.97326225e-02 6.97856769e-02 5.32163978e-01
-1.89632758e-01 9.23309207e-01 -7.50967562e-01 -9.46378291e-01
-7.61130989e-01 -1.06281793e+00 7.68709242e-01 -3.52692567e-02
8.03727284e-02 -1.40115285e+00 5.83755299e-02 7.39327520e-02
5.38707972e-01 2.44421110e-01 8.39886189e-01 -1.17662966e+00
-3.23158711e-01 -5.18223047e-01 8.53282982e-04 4.59369063e-01
1.96128562e-01 -1.54387072e-01 -9.23567653e-01 -6.02983057e-01
6.06293380e-01 -5.82921170e-02 1.02277267e+00 4.55169082e-02
8.04374754e-01 -8.55898142e-01 -3.28715533e-01 4.62023526e-01
1.44829786e+00 4.48242307e-01 9.75848377e-01 2.64441401e-01
7.26723671e-01 4.33373034e-01 5.85206645e-03 7.18308389e-02
7.80684277e-02 1.12267084e-01 3.77214253e-01 -9.90334973e-02
2.06248406e-02 -5.41843176e-01 7.64336646e-01 1.27392352e+00
9.50399339e-02 -9.66558933e-01 -6.94025993e-01 2.59084940e-01
-1.58640862e+00 -1.30282426e+00 -1.23189390e-01 1.82178974e+00
7.64668524e-01 3.24342310e-01 -1.06211632e-01 3.68066728e-01
1.13070738e+00 5.37288427e-01 -7.27442726e-02 -3.41702729e-01
-2.18278602e-01 2.09835932e-01 7.09837615e-01 5.20097613e-01
-9.85053182e-01 1.18324518e+00 6.63013124e+00 1.41356015e+00
-1.53214025e+00 1.69315219e-01 5.46059668e-01 5.07612944e-01
-4.81216699e-01 -5.84595837e-02 -7.14150012e-01 7.30010450e-01
1.23234439e+00 -7.65013769e-02 3.72623950e-01 5.92559636e-01
4.14921232e-02 4.52488959e-01 -5.11924088e-01 8.47089350e-01
4.19625252e-01 -1.16809976e+00 4.13096845e-01 1.45902261e-01
6.67224944e-01 -2.13440344e-01 5.04772179e-02 4.73595828e-01
5.71521163e-01 -1.18547070e+00 5.32028675e-01 2.40777761e-01
6.66390717e-01 -7.13766038e-01 1.11443400e+00 6.86601639e-01
-8.72101963e-01 -8.41260143e-03 -4.27243978e-01 6.11470416e-02
-7.57401735e-02 5.04050493e-01 -1.25840938e+00 6.03311360e-01
1.11960471e-01 4.78339970e-01 -3.96490306e-01 3.34239781e-01
-4.40367758e-01 1.04126108e+00 1.76302362e-02 -7.36131907e-01
5.34558237e-01 2.25282386e-02 6.26490712e-01 1.36472857e+00
5.87544143e-01 -9.81315896e-02 -7.10943714e-02 9.07164872e-01
4.23864136e-03 1.66674197e-01 -7.80824363e-01 -5.56868970e-01
2.58242637e-01 1.04633701e+00 -8.44595850e-01 -6.00668311e-01
-1.26312912e-01 1.00530958e+00 -2.80814558e-01 2.40811244e-01
-1.02584517e+00 -8.46383572e-01 -9.65820700e-02 -4.91381474e-02
5.89693069e-01 -8.39245766e-02 -2.70671267e-02 -1.29821241e+00
-2.81387776e-01 -1.32622600e+00 -7.69662783e-02 -4.43959266e-01
-9.68975008e-01 8.86980414e-01 -4.42370743e-01 -1.11853600e+00
-5.39238155e-01 -5.90047836e-01 -5.59502602e-01 7.52663016e-01
-1.35561955e+00 -1.13666677e+00 -3.85102555e-02 5.75267255e-01
4.48419958e-01 -5.68138361e-01 6.96832061e-01 1.13362253e-01
-5.29005945e-01 8.49004626e-01 1.70814544e-01 7.43685067e-01
2.75058687e-01 -7.35533535e-01 5.86676359e-01 1.18051827e+00
9.51087698e-02 4.90994155e-01 1.01244032e+00 -8.52467060e-01
-1.24552512e+00 -1.00976264e+00 8.98561478e-01 -2.28688464e-01
7.40517378e-01 -3.76534194e-01 -8.04339707e-01 8.10790241e-01
5.45778811e-01 -6.96817279e-01 3.59557241e-01 -6.79227054e-01
-1.44453511e-01 3.84531260e-01 -1.09457421e+00 6.26291931e-01
6.45077646e-01 -3.96258891e-01 -7.20019937e-01 1.47811309e-01
9.56210434e-01 -2.97961086e-01 -3.73923421e-01 1.11788370e-01
3.08326066e-01 -9.54008460e-01 5.97112417e-01 -3.46223593e-01
6.15714133e-01 -1.44356832e-01 5.00311255e-02 -1.25890863e+00
-2.58726090e-01 -8.94871831e-01 -1.61694437e-01 1.05751312e+00
4.68957752e-01 -1.00653684e+00 7.02573895e-01 -3.17780763e-01
1.85209766e-01 -4.01563197e-01 -4.22650129e-01 -6.37473941e-01
1.49951115e-01 -2.15244308e-01 8.51655364e-01 1.10118210e+00
-7.32587501e-02 4.19268250e-01 -9.68955874e-01 1.82514593e-01
5.37864149e-01 -1.36834130e-01 7.22145617e-01 -8.96980703e-01
-4.40132618e-01 -5.25735378e-01 -4.38670993e-01 -1.17166293e+00
2.10125849e-01 -8.95170152e-01 2.17817545e-01 -9.71960187e-01
5.52526973e-02 -1.01659507e-01 -2.59955488e-02 2.91349262e-01
-3.52170616e-01 2.27786124e-01 -3.73591892e-02 3.62445146e-01
-2.04863891e-01 4.36077476e-01 1.33136892e+00 -2.40882069e-01
-1.53560087e-01 4.85447189e-03 -9.55029190e-01 5.55605769e-01
9.80626523e-01 -6.20048404e-01 -2.29064360e-01 -2.91037291e-01
3.15936118e-01 1.46856889e-01 1.68196410e-01 -9.60094035e-01
1.27318308e-01 5.75455427e-02 2.83058256e-01 -4.28990155e-01
-1.08054556e-01 -7.71789372e-01 1.84374023e-02 1.02823591e+00
-3.25039506e-01 -2.76087433e-01 -1.62304908e-01 5.89488506e-01
-1.78975672e-01 -5.85194409e-01 6.91218197e-01 -5.14817774e-01
-5.33242404e-01 3.63169499e-02 -8.92964363e-01 -1.08551204e-01
9.65029955e-01 -4.43443835e-01 -1.23659916e-01 -7.70052552e-01
-3.26374888e-01 -1.84447214e-01 1.00896552e-01 4.57560390e-01
6.66039586e-01 -1.20995963e+00 -9.46896672e-01 4.92035449e-01
-4.31351364e-01 -5.53378999e-01 -6.51320741e-02 5.35924196e-01
-8.02715838e-01 6.73214078e-01 2.79537234e-02 -3.50601882e-01
-1.08358717e+00 7.15122640e-01 3.48710835e-01 -7.99481988e-01
-5.83335578e-01 7.22855210e-01 8.34278315e-02 -2.84746379e-01
1.87657233e-02 -1.64354011e-01 -4.06464577e-01 -1.80788770e-01
5.48381448e-01 2.48114139e-01 -3.04579616e-01 -7.04370320e-01
5.87833337e-02 4.25940245e-01 -2.39435911e-01 -3.87109853e-02
1.07563233e+00 -1.96593106e-01 -2.02676222e-01 1.21856011e-01
1.27628458e+00 1.66784942e-01 -5.39042771e-01 -3.30966115e-01
-8.72969106e-02 -1.42343447e-01 -1.41147658e-01 -3.79846573e-01
-1.25163364e+00 8.39935184e-01 4.57125217e-01 7.86957443e-01
1.01819336e+00 -2.74862915e-01 1.19787395e+00 4.45432872e-01
2.71685570e-01 -7.47679353e-01 3.25105697e-01 6.50726080e-01
4.79736656e-01 -1.10049474e+00 -2.94359803e-01 -2.31043354e-01
-6.06574595e-01 1.27672708e+00 2.29424760e-01 -3.85246754e-01
6.06848419e-01 1.50140122e-01 1.15023635e-01 -2.14786127e-01
-3.13166112e-01 -4.06312756e-02 1.45940721e-01 3.77272338e-01
2.67676383e-01 -7.87206590e-02 -3.74357015e-01 2.18376547e-01
-7.86216676e-01 -2.41094559e-01 5.78497648e-01 7.14538574e-01
-8.53577435e-01 -9.54749107e-01 -5.79658687e-01 2.59880126e-01
-7.94504821e-01 -3.78051460e-01 -2.16633186e-01 7.91999757e-01
3.29833999e-02 1.08288598e+00 -1.94991261e-01 -8.54286134e-01
-1.94925562e-01 9.52002853e-02 1.70909062e-01 -4.67635453e-01
-7.32888520e-01 1.09792002e-01 -1.91245034e-01 2.23036353e-02
-3.78121555e-01 -1.32753596e-01 -1.05386984e+00 -5.87958157e-01
-8.68740022e-01 3.59009445e-01 4.24089998e-01 1.18812454e+00
6.59549236e-02 5.65529406e-01 9.10569787e-01 -4.92996156e-01
-7.67393231e-01 -1.29707074e+00 -4.85459745e-01 3.04418981e-01
4.05696452e-01 3.63442861e-02 -7.85986185e-01 1.28298283e-01] | [4.295944690704346, 8.056109428405762] |
86e8e7c7-b5b3-4688-add1-dbfebb1be465 | a-deep-variational-approach-to-clustering | 2106.05763 | null | https://arxiv.org/abs/2106.05763v3 | https://arxiv.org/pdf/2106.05763v3.pdf | A Deep Variational Approach to Clustering Survival Data | In this work, we study the problem of clustering survival data $-$ a challenging and so far under-explored task. We introduce a novel semi-supervised probabilistic approach to cluster survival data by leveraging recent advances in stochastic gradient variational inference. In contrast to previous work, our proposed method employs a deep generative model to uncover the underlying distribution of both the explanatory variables and censored survival times. We compare our model to the related work on clustering and mixture models for survival data in comprehensive experiments on a wide range of synthetic, semi-synthetic, and real-world datasets, including medical imaging data. Our method performs better at identifying clusters and is competitive at predicting survival times. Relying on novel generative assumptions, the proposed model offers a holistic perspective on clustering survival data and holds a promise of discovering subpopulations whose survival is regulated by different generative mechanisms. | ['Bram Stieltjes', 'Timothy Müller', 'Verena Gotta', 'Alexander Sauter', 'Thomas Weikert', 'Julia E. Vogt', 'Marc Pfister', 'Marian C. Neidert', 'Flavio Vasella', 'Michela C. Massi', 'Ričards Marcinkevičs', 'Laura Manduchi'] | 2021-06-10 | a-deep-variational-approach-to-clustering-1 | https://openreview.net/forum?id=RQ428ZptQfU | https://openreview.net/pdf?id=RQ428ZptQfU | iclr-2022-4 | ['time-to-event-prediction'] | ['time-series'] | [-7.08111376e-02 -1.15695328e-01 -2.31062993e-01 -5.15151620e-01
-1.19953275e+00 -2.11313605e-01 5.46485603e-01 1.72380149e-01
-1.77330121e-01 8.46408367e-01 5.34210205e-01 -3.08480084e-01
-3.94326746e-01 -4.86805737e-01 -2.29153201e-01 -1.27321601e+00
-2.08767563e-01 8.48024428e-01 -2.21292838e-01 2.68411219e-01
1.79511216e-02 1.73644513e-01 -9.90856409e-01 -1.26667365e-01
9.94060755e-01 4.79316980e-01 -9.06064957e-02 6.53368890e-01
3.35464478e-01 7.87934065e-01 -2.76317924e-01 -2.58043796e-01
-3.86391312e-01 -6.29651129e-01 -5.94998956e-01 1.77972093e-01
-2.97067851e-01 2.18943924e-01 -5.35058320e-01 7.21298099e-01
5.41803837e-01 -6.47925436e-02 1.35534906e+00 -1.32747769e+00
-5.58290482e-01 6.20848477e-01 -5.41766703e-01 1.96997121e-01
-1.38417944e-01 -1.10352345e-01 6.62504196e-01 -5.07434011e-01
4.33591455e-01 8.85639846e-01 9.84117150e-01 6.89897418e-01
-1.70202959e+00 -2.40781263e-01 -2.46944398e-01 -1.57061473e-01
-1.62598479e+00 -5.44579744e-01 7.95043409e-01 -8.18441212e-01
2.25514621e-01 2.97836930e-01 4.02200013e-01 1.44287884e+00
4.08031702e-01 7.49355972e-01 1.15594876e+00 -1.77185938e-01
7.02593744e-01 5.24793863e-02 3.23051453e-01 6.77192390e-01
-1.10908076e-02 2.61221617e-01 -3.21385562e-01 -7.88730741e-01
5.30726194e-01 4.53863919e-01 -2.70632178e-01 -6.30647540e-01
-1.22086442e+00 1.24439645e+00 -6.54596463e-02 2.40613282e-01
-3.52714211e-01 4.28803802e-01 1.61062524e-01 -2.14353949e-01
8.54111910e-01 -2.88505465e-01 -3.36326510e-01 7.60776848e-02
-1.47027647e+00 3.94224115e-02 6.73472941e-01 5.87066770e-01
1.81822971e-01 9.14343596e-02 -4.13158774e-01 6.39077902e-01
7.40986109e-01 4.54120994e-01 3.20510536e-01 -9.83332396e-01
-1.86974645e-01 3.71793926e-01 5.76303601e-02 -6.37849629e-01
-6.97971642e-01 -6.81849241e-01 -1.18353164e+00 -4.00716141e-02
5.78909874e-01 -2.06576407e-01 -8.45046639e-01 1.95411015e+00
3.58424544e-01 4.49909449e-01 4.80752662e-02 5.77563763e-01
6.82330787e-01 3.76704484e-01 3.52099538e-01 -5.87327123e-01
1.09336209e+00 -4.98727202e-01 -5.99503577e-01 2.40449995e-01
5.92242897e-01 -2.91047364e-01 6.44331396e-01 1.36731505e-01
-8.78854752e-01 -1.80566739e-02 -4.82868701e-01 2.66780347e-01
1.17995508e-01 1.87480718e-01 8.86929393e-01 6.97017193e-01
-1.37785923e+00 5.61501980e-01 -1.36499333e+00 -6.43126965e-01
8.09724331e-01 9.76663902e-02 -4.24427353e-02 -1.91585705e-01
-8.15801382e-01 1.54072255e-01 7.78366327e-02 -1.58851281e-01
-1.24100530e+00 -5.89582801e-01 -7.04521358e-01 3.55732702e-02
1.40185431e-01 -1.33030379e+00 8.14094663e-01 -3.44477564e-01
-1.34915066e+00 8.41057479e-01 -4.22663122e-01 -4.83517557e-01
5.26321709e-01 1.43452212e-01 -1.25743791e-01 7.58270919e-02
1.54521316e-01 2.43535966e-01 6.43430710e-01 -1.21792853e+00
-2.58918703e-01 -7.42106915e-01 -8.54337156e-01 -1.90431848e-01
-3.37738484e-01 -1.59212381e-01 -2.32012227e-01 -8.76608789e-01
1.34443715e-01 -1.05716085e+00 -6.43129587e-01 -3.46188486e-01
-6.06540203e-01 -3.25794160e-01 2.35466719e-01 -5.02405524e-01
1.41356528e+00 -2.15510559e+00 5.35768688e-01 4.37094867e-02
5.50085723e-01 -5.22467792e-01 4.69893247e-01 6.69812381e-01
2.33253479e-01 1.52395919e-01 -8.03809643e-01 -9.73938107e-01
-1.07900530e-01 1.19614050e-01 -1.89052150e-01 8.79084408e-01
-2.76746660e-01 9.15657222e-01 -8.70363235e-01 -8.33777547e-01
-9.47462842e-02 7.22824574e-01 -4.30129409e-01 9.15041119e-02
-4.59068120e-02 9.35685635e-01 -5.49296916e-01 8.61879408e-01
3.39264005e-01 -6.42176926e-01 2.51453251e-01 7.20274508e-01
2.73623884e-01 -5.13212681e-01 -6.70819879e-01 1.75408721e+00
7.28755072e-02 3.77386034e-01 -1.63766176e-01 -1.17651021e+00
5.90879440e-01 4.84459192e-01 8.03989112e-01 1.65916696e-01
2.86313981e-01 -1.60719920e-02 -4.84102279e-01 -4.05374140e-01
-5.11070862e-02 -5.60721755e-01 -3.36609840e-01 3.78951848e-01
5.34177795e-02 3.19933444e-01 -2.39264831e-01 4.40681189e-01
1.13993430e+00 1.06450431e-01 1.22681424e-01 -6.08329594e-01
1.65607631e-01 2.37665921e-01 7.63152003e-01 9.40642297e-01
-3.42788190e-01 5.70283711e-01 7.01397240e-01 -9.92573351e-02
-7.93010592e-01 -1.48245478e+00 -6.16057932e-01 7.86016226e-01
-2.61570394e-01 -2.27877378e-01 -7.61413753e-01 -5.71788728e-01
-4.57212180e-02 7.32530296e-01 -9.91828084e-01 -1.87805295e-01
3.23365591e-02 -1.66599727e+00 5.99824250e-01 6.17775261e-01
-2.08217710e-01 -6.48037791e-01 -1.78487897e-01 1.68121949e-01
-2.37344295e-01 -5.89887083e-01 -3.26066077e-01 1.60582975e-01
-1.32204628e+00 -1.09092617e+00 -8.49379301e-01 -6.84712291e-01
6.15923524e-01 -1.11126438e-01 1.30390227e+00 -9.32697430e-02
-3.87520283e-01 3.40492785e-01 -9.06476602e-02 -1.07225575e-01
-4.74066794e-01 1.88060850e-02 -5.05169481e-02 3.27192992e-02
2.49670550e-01 -5.97012758e-01 -9.55180287e-01 2.17181101e-01
-8.55862617e-01 -1.61164656e-01 3.51054996e-01 1.00828993e+00
7.82093704e-01 2.61592329e-01 8.34393859e-01 -1.08387303e+00
2.87552834e-01 -1.18325746e+00 -3.84147197e-01 4.90985140e-02
-9.16905940e-01 7.80315399e-02 4.54792976e-01 -2.79130429e-01
-9.95113432e-01 9.48015079e-02 1.62335455e-01 -3.94212961e-01
-4.39533085e-01 5.86733639e-01 -3.27572636e-02 4.80630934e-01
3.54690313e-01 2.73396850e-01 1.87754616e-01 -4.39158261e-01
2.06546560e-01 4.02696729e-01 3.96555603e-01 -3.86695951e-01
4.70602721e-01 1.22256958e+00 4.51883912e-01 -4.61248696e-01
-6.15700662e-01 -4.35364932e-01 -6.37830079e-01 -6.99905772e-03
1.11144567e+00 -9.36084688e-01 -7.65666127e-01 5.34872591e-01
-4.63250995e-01 -4.21362877e-01 -6.69037625e-02 5.29494762e-01
-7.67448902e-01 3.94062817e-01 -9.16773379e-01 -1.06391084e+00
-1.17647640e-01 -1.11432409e+00 1.33199430e+00 1.71710141e-02
-3.29250470e-02 -1.62486982e+00 7.24749148e-01 4.51443136e-01
2.81260699e-01 8.03171754e-01 1.03195214e+00 -5.95364630e-01
-1.89887553e-01 -2.50664413e-01 2.03743234e-01 -1.56720579e-01
-2.81862430e-02 7.20791072e-02 -7.59119570e-01 -5.01538336e-01
1.05557799e-01 -1.05731480e-01 1.40882015e+00 9.92954373e-01
1.15542352e+00 3.35925259e-02 -9.24961209e-01 7.55120575e-01
1.38370752e+00 1.67994797e-01 4.50101882e-01 -1.15215905e-01
5.79950929e-01 6.97210014e-01 1.96478605e-01 6.13782644e-01
8.45479012e-01 3.73368531e-01 2.32488945e-01 -1.63973287e-01
2.86701143e-01 -1.83567405e-01 8.90764892e-02 6.05907381e-01
1.18503787e-01 -4.08063650e-01 -1.15561366e+00 8.08151126e-01
-2.17551637e+00 -1.07323325e+00 -4.59631711e-01 2.19688988e+00
6.76628530e-01 -1.53021842e-01 4.80338424e-01 -3.00281812e-02
7.10079134e-01 -7.20661059e-02 -6.33555532e-01 3.13920915e-01
-1.72950089e-01 -1.76000576e-02 2.55644977e-01 3.76529247e-01
-1.16069496e+00 5.91211796e-01 7.37139654e+00 7.20147371e-01
-6.34339869e-01 2.88556933e-01 1.13088894e+00 -3.57733428e-01
-2.92004496e-01 1.94618270e-01 -4.52441037e-01 5.55631578e-01
1.19683766e+00 1.24714404e-01 -2.68670241e-03 5.63359499e-01
3.17053974e-01 -7.69792795e-02 -1.06184733e+00 6.25678301e-01
1.07667387e-01 -1.11326873e+00 -3.34820122e-01 4.07020748e-01
7.70756960e-01 -1.53738424e-01 3.28718722e-01 1.43451661e-01
7.64686704e-01 -1.26978505e+00 3.62990111e-01 1.11020541e+00
6.66909397e-01 -1.02686882e+00 8.01621020e-01 4.72514421e-01
-6.21180117e-01 -2.39575673e-02 -3.95169184e-02 1.81042656e-01
3.31636459e-01 9.34394956e-01 -6.27891541e-01 5.20179212e-01
6.68189645e-01 6.92338407e-01 -6.37755454e-01 8.46887767e-01
2.12197155e-01 1.24206638e+00 -4.52567972e-02 2.45896772e-01
-1.27673030e-01 -1.21985368e-01 3.91951710e-01 1.01535141e+00
3.84255856e-01 1.60457045e-01 1.37264833e-01 8.37043881e-01
2.73717195e-01 -5.70855550e-02 -3.38557839e-01 -8.68607163e-02
3.45420390e-01 1.14698446e+00 -1.03573954e+00 -2.31297567e-01
-6.87876791e-02 8.29151452e-01 3.53439569e-01 4.75857735e-01
-9.17442203e-01 2.04455435e-01 3.33601296e-01 1.93001822e-01
1.79352731e-01 -1.65968612e-01 -4.75479275e-01 -1.38217306e+00
-6.29696131e-01 -1.61732212e-01 7.35373974e-01 -4.88448828e-01
-1.60031474e+00 5.00348091e-01 4.76379730e-02 -7.97269404e-01
-4.21899825e-01 -1.56194031e-01 -6.57726526e-01 6.31094098e-01
-1.17174995e+00 -1.07476711e+00 -6.76118508e-02 7.16112137e-01
3.22997123e-01 -4.75807413e-02 7.55251348e-01 1.90972332e-02
-9.95320559e-01 3.67380589e-01 9.09570515e-01 -9.92438570e-02
6.15393102e-01 -1.42320418e+00 3.83179113e-02 4.94340569e-01
1.99885890e-01 5.93163610e-01 8.49566102e-01 -6.61515415e-01
-1.08343995e+00 -1.11622095e+00 6.04175448e-01 -1.03609097e+00
4.71622705e-01 -4.94882822e-01 -8.27673137e-01 4.73250091e-01
4.95516174e-02 -1.40382156e-01 1.46210098e+00 2.03970209e-01
4.03325520e-02 3.60989213e-01 -1.20883775e+00 3.45312208e-01
1.00642908e+00 -1.75365686e-01 -2.72242159e-01 4.82245237e-01
5.04432559e-01 1.24823213e-01 -1.03741026e+00 3.56896937e-01
2.94751316e-01 -1.03495932e+00 9.33982015e-01 -7.64306366e-01
5.95014155e-01 -1.70850500e-01 -2.52053708e-01 -1.25149357e+00
-6.23675108e-01 -5.59830070e-01 -2.12553725e-01 1.18267250e+00
2.54072070e-01 -4.20262337e-01 9.49770868e-01 3.56236577e-01
-3.34383175e-02 -9.40029621e-01 -1.10541379e+00 -3.83780301e-01
5.39850354e-01 -2.32874006e-01 2.47557029e-01 1.00436115e+00
-5.62688261e-02 1.32943168e-01 -4.95450020e-01 2.58667409e-01
1.42683697e+00 3.00107837e-01 4.75023687e-01 -1.63973534e+00
-4.47981834e-01 -4.04451847e-01 -1.69216454e-01 -4.00841027e-01
4.86542016e-01 -9.31971133e-01 2.51421984e-02 -1.47443914e+00
8.54067981e-01 -2.94586629e-01 -4.52425599e-01 3.33769247e-02
-4.81864125e-01 3.06369513e-01 -4.76965964e-01 5.87863624e-01
-7.16908276e-01 6.94104314e-01 6.19473696e-01 1.82072893e-01
-1.72611829e-02 1.98045641e-01 -9.16704774e-01 6.78311586e-01
7.40572870e-01 -7.51787901e-01 -3.97597462e-01 1.98528226e-02
-1.78958103e-01 8.41492176e-01 6.85548663e-01 -6.44608915e-01
2.45868251e-01 -2.16036513e-01 4.11058635e-01 -6.69139743e-01
7.96067417e-02 -4.08224434e-01 4.88525748e-01 5.19873917e-01
-3.05852443e-01 -6.69676661e-02 -2.12058350e-01 1.24462879e+00
-7.01537961e-03 3.37482959e-01 7.18927264e-01 1.39997816e-02
-1.10980541e-01 3.19519460e-01 -6.64821625e-01 1.77856371e-01
1.08796406e+00 3.04478034e-02 -1.15766123e-01 -4.04390931e-01
-1.25473905e+00 4.34359103e-01 6.64071321e-01 -2.61757597e-02
5.49291015e-01 -1.20201349e+00 -9.66616213e-01 -1.02083229e-01
1.96884125e-01 -1.69457719e-01 6.19205594e-01 1.15182054e+00
-5.19842245e-02 3.19603115e-01 3.63268137e-01 -9.30945694e-01
-9.04554665e-01 8.72483373e-01 2.00495243e-01 -4.21232104e-01
-4.58646536e-01 6.77040935e-01 3.88111383e-01 -2.53118813e-01
2.00666785e-02 1.29104167e-01 -1.88943595e-01 -5.06635047e-02
1.32985622e-01 5.76067150e-01 -4.94710892e-01 -7.70128310e-01
-4.11536098e-01 2.59158194e-01 1.65202960e-01 -2.26399347e-01
1.39785242e+00 -3.95022124e-01 -1.88282639e-01 9.92522359e-01
1.02761233e+00 -1.83051184e-01 -1.29036140e+00 -1.86097562e-01
1.88062876e-01 -9.51205343e-02 1.08578034e-01 -6.58721805e-01
-9.99605775e-01 7.33330250e-01 4.97225761e-01 1.07719727e-01
1.01982188e+00 5.48521698e-01 5.03987551e-01 -3.70548993e-01
3.07499528e-01 -6.80534422e-01 2.87480969e-02 -9.57636908e-02
2.14130566e-01 -1.46660388e+00 -1.96736321e-01 -2.29824692e-01
-6.68211937e-01 6.41121805e-01 5.66273406e-02 -1.81709543e-01
1.04172170e+00 2.12118804e-01 9.08233505e-03 -5.02213240e-01
-8.92790496e-01 -1.12160249e-02 2.50890642e-01 5.29122412e-01
4.66265142e-01 3.36154789e-01 -3.65558118e-01 9.19802845e-01
3.00206076e-02 -1.86407696e-02 4.39167708e-01 6.82704985e-01
-1.48961186e-01 -8.67917061e-01 -3.70126784e-01 5.28025270e-01
-8.04576814e-01 -1.50896057e-01 -2.30561957e-01 4.54991609e-01
-1.93924561e-01 1.11726677e+00 -5.79510815e-02 -7.75337294e-02
-2.77146071e-01 3.57609034e-01 8.40202793e-02 -4.26883847e-01
-1.01097375e-01 3.93510431e-01 -4.78375971e-01 -2.63905823e-01
-5.26298225e-01 -1.22985566e+00 -1.08190024e+00 -2.23895982e-01
-2.70097017e-01 3.47046494e-01 4.90854740e-01 7.62371480e-01
4.86374259e-01 6.57504737e-01 6.95137143e-01 -4.95239258e-01
-4.97013420e-01 -7.28765786e-01 -9.09845233e-01 4.04247522e-01
3.38136107e-01 -7.98712432e-01 -6.60189390e-01 2.59019464e-01] | [7.721401214599609, 5.548935413360596] |
617d2ed9-1a0d-4547-b9bc-7396aff71028 | beamsearchqa-large-language-models-are-strong | 2305.14766 | null | https://arxiv.org/abs/2305.14766v2 | https://arxiv.org/pdf/2305.14766v2.pdf | BeamSearchQA: Large Language Models are Strong Zero-Shot QA Solver | Open-domain question answering is a crucial task that often requires accessing external information. Existing methods typically adopt a single-turn retrieve-then-read approach, where relevant documents are first retrieved, and questions are then answered based on the retrieved information. However, there are cases where answering a question requires implicit knowledge that is not directly retrievable from the question itself. In this work, we propose a novel question-answering pipeline called BeamSearchQA. Our approach leverages large language models to iteratively generate new questions about the original question, enabling an iterative reasoning process. By iteratively refining and expanding the scope of the question, our method aims to capture and utilize hidden knowledge that may not be directly obtainable through retrieval. We evaluate our approach on the widely-used open-domain NQ and WebQ datasets. The experimental results demonstrate that BeamSearchQA significantly outperforms other zero-shot baselines, indicating its effectiveness in tackling the challenges of open-domain question answering. | ['Rangan Majumder', 'Linjun Yang', 'Daxin Jiang', 'Jingwen Lu', 'Anlei Dong', 'Nan Duan', 'Yan Zhang', 'Yeyun Gong', 'Xiao Liu', 'Hao Sun'] | 2023-05-24 | null | null | null | null | ['open-domain-question-answering'] | ['natural-language-processing'] | [ 1.55107915e-01 4.04323965e-01 -5.56270331e-02 -2.28011042e-01
-1.64447606e+00 -9.89725649e-01 6.76299751e-01 2.19334900e-01
-4.23056692e-01 9.11224544e-01 6.28370404e-01 -4.98491108e-01
-1.58487990e-01 -9.72108841e-01 -6.63254559e-01 5.76875806e-02
6.85732186e-01 8.20819557e-01 6.61163032e-01 -6.97865367e-01
3.44323575e-01 -1.99177176e-01 -1.51092577e+00 5.44171333e-01
1.16850030e+00 6.85327411e-01 3.37542355e-01 6.82065964e-01
-8.65462303e-01 1.33405519e+00 -5.19570947e-01 -3.38376969e-01
6.12208713e-03 -5.20372391e-01 -1.73351049e+00 -1.81788906e-01
3.78495097e-01 -7.25082457e-01 -3.75752002e-01 6.39931917e-01
2.41521657e-01 4.59497839e-01 3.47786278e-01 -5.73170662e-01
-1.16238666e+00 4.94717002e-01 2.99117267e-02 5.45396626e-01
1.04425085e+00 1.67238176e-01 1.43097103e+00 -9.71885204e-01
6.07418954e-01 1.26813638e+00 6.57023191e-02 4.66672987e-01
-9.32772160e-01 -3.84053141e-01 2.74812430e-01 4.51590866e-01
-1.19775355e+00 -4.82381761e-01 5.23725212e-01 -4.58189882e-02
9.78194416e-01 3.45845878e-01 1.83376387e-01 8.50983322e-01
-2.37801760e-01 9.27996635e-01 1.00946057e+00 -5.66633701e-01
2.93634623e-01 -2.77140159e-02 6.23214006e-01 5.12246251e-01
-1.22985095e-01 -2.85792559e-01 -6.02255583e-01 -2.92484254e-01
2.74599850e-01 -1.39117479e-01 -3.22426170e-01 7.88756311e-02
-9.73766863e-01 8.46975684e-01 2.61910290e-01 1.78586587e-01
-5.92077613e-01 -2.47937724e-01 5.78492433e-02 5.63491344e-01
4.16604042e-01 7.45109618e-01 -4.34319496e-01 -2.67561018e-01
-8.15059245e-01 4.56228107e-01 1.24073601e+00 9.92694974e-01
9.61199880e-01 -7.73754895e-01 -6.82063162e-01 8.82876217e-01
4.08871531e-01 5.80750763e-01 3.68079245e-01 -1.17967677e+00
6.37370706e-01 8.81909728e-01 5.67861915e-01 -5.96460938e-01
7.92793706e-02 -7.19420537e-02 -3.98911797e-02 -6.18844867e-01
4.62435365e-01 2.46720687e-02 -8.22967947e-01 1.38857377e+00
6.92230999e-01 -1.27093270e-01 2.96008736e-01 8.57526183e-01
1.36021876e+00 9.99959588e-01 -3.07372678e-03 -2.29398105e-02
1.64179647e+00 -1.30389142e+00 -9.82231379e-01 -4.17993844e-01
3.76514643e-01 -6.87971771e-01 1.38356364e+00 6.64329827e-02
-1.06078529e+00 -4.19994891e-01 -6.31363094e-01 -8.21956515e-01
-1.48725435e-01 -2.45779723e-01 -6.70275912e-02 1.75017491e-01
-1.02366340e+00 -9.32886600e-02 -3.43002856e-01 -4.03780043e-01
7.49489591e-02 -2.55160838e-01 6.89403415e-02 -6.95705712e-01
-1.62729228e+00 7.53789783e-01 2.14093044e-01 -1.82068069e-02
-1.06993508e+00 -7.59976566e-01 -6.23511791e-01 2.21925125e-01
1.11008525e+00 -9.14606929e-01 2.01566219e+00 -4.19027060e-01
-1.52399278e+00 7.15699971e-01 -7.05693603e-01 -2.04863161e-01
4.37781066e-02 -6.31867468e-01 -3.50435942e-01 6.04334831e-01
3.92044663e-01 4.97560352e-01 7.21572697e-01 -1.08579838e+00
-5.62120736e-01 -4.25713599e-01 9.30487096e-01 3.52097660e-01
-2.05208138e-01 6.78685158e-02 -7.41158068e-01 -1.87552899e-01
6.76647425e-02 -5.97265959e-01 1.97561122e-02 -2.26629004e-01
-2.51543343e-01 -8.17309499e-01 6.39603615e-01 -1.01852405e+00
1.57726836e+00 -1.80655098e+00 -6.85104504e-02 -8.06098729e-02
4.06282812e-01 3.78750086e-01 -4.58404541e-01 1.02244461e+00
4.22840565e-01 1.41848207e-01 -1.82275042e-01 2.77855456e-01
1.31865233e-01 2.84114391e-01 -9.68160093e-01 -3.96424443e-01
2.46994823e-01 1.34087455e+00 -1.14562821e+00 -5.70978642e-01
-3.65591109e-01 4.14943732e-02 -4.91178721e-01 6.53316855e-01
-9.75069821e-01 2.40473285e-01 -9.19872582e-01 4.71358150e-01
3.15964490e-01 -7.98305213e-01 -2.66389009e-02 9.78343040e-02
2.03855127e-01 9.59065795e-01 -7.70873010e-01 1.80128026e+00
-5.20616174e-01 3.12588692e-01 5.44418357e-02 -5.44187129e-01
9.70475495e-01 4.02119339e-01 -5.45105524e-02 -1.09543002e+00
-3.76755267e-01 1.46946773e-01 -4.76590276e-01 -9.73774076e-01
6.97706580e-01 -1.60420358e-01 7.74658844e-02 9.75818515e-01
-8.02297592e-02 -3.76173913e-01 4.28276688e-01 7.66635120e-01
1.35504687e+00 1.87754914e-01 2.14647308e-01 -1.20161734e-04
6.91504121e-01 3.69751751e-01 3.81082669e-02 1.08790362e+00
-1.54927090e-01 2.43446544e-01 4.34106529e-01 -7.77899772e-02
-6.51922107e-01 -1.20414197e+00 2.24913403e-01 1.39327407e+00
2.58526474e-01 -4.73331690e-01 -6.67271256e-01 -9.79720354e-01
-9.18489471e-02 9.92276669e-01 -3.61019105e-01 -1.27429828e-01
-4.30949688e-01 1.75391793e-01 3.74535710e-01 3.22741091e-01
3.64547580e-01 -1.08154190e+00 -4.91859823e-01 4.17640537e-01
-9.88652050e-01 -1.09075129e+00 -4.90619630e-01 -4.29590166e-01
-8.47077310e-01 -1.26431906e+00 -6.85972452e-01 -6.97006404e-01
4.30594414e-01 7.26800203e-01 1.58516538e+00 3.28318626e-01
2.09703565e-01 9.24734712e-01 -7.50026286e-01 -1.33122429e-01
-3.54132771e-01 3.87496620e-01 -6.51555538e-01 -1.38291165e-01
7.25163579e-01 -3.23560029e-01 -8.46145153e-01 2.31559455e-01
-1.23186767e+00 -1.79682523e-01 5.04351676e-01 6.23071134e-01
5.10954142e-01 -4.21664953e-01 1.22811306e+00 -8.66799474e-01
1.14081597e+00 -9.28353667e-01 -3.89964938e-01 7.66066253e-01
-4.24311459e-01 3.67214590e-01 4.57179755e-01 -1.35139078e-01
-1.56830633e+00 -5.31706810e-01 -2.46903926e-01 1.96561720e-02
-9.22251418e-02 8.56671810e-01 4.87343706e-02 4.47182029e-01
7.51428366e-01 3.56096864e-01 -2.01731160e-01 -6.34404480e-01
8.12456250e-01 7.65623450e-01 3.57383609e-01 -7.83283532e-01
8.48572075e-01 3.89532238e-01 -7.62632549e-01 -6.75228298e-01
-1.57162035e+00 -9.59788918e-01 -2.16115937e-01 -2.60624945e-01
7.45583773e-01 -9.21069741e-01 -4.69551414e-01 -1.50380060e-02
-1.26025081e+00 -1.63720787e-01 -5.17671466e-01 -1.21338144e-02
-1.89207330e-01 5.64236343e-01 -6.41607523e-01 -7.59087324e-01
-5.54850459e-01 -7.07853079e-01 1.17632461e+00 3.16220939e-01
-3.44711214e-01 -8.80486608e-01 3.39730054e-01 1.22036076e+00
5.18680632e-01 -5.75617015e-01 1.04893506e+00 -1.03799009e+00
-1.06023300e+00 -8.76103640e-02 -2.64045209e-01 1.21135555e-01
1.08861886e-01 -5.77664614e-01 -8.76109242e-01 -3.90713774e-02
7.80342147e-02 -1.09220183e+00 7.67191410e-01 -3.63484651e-01
8.58925641e-01 -5.14444649e-01 -7.11833239e-02 -2.36034513e-01
1.07754707e+00 -7.88337588e-02 6.08558953e-01 3.24220121e-01
2.31629789e-01 9.11095858e-01 7.51013041e-01 2.62304962e-01
1.04753029e+00 2.43566632e-01 1.81989729e-01 3.76172841e-01
-8.66670907e-02 -6.14307165e-01 2.00193077e-01 9.98080790e-01
4.67286527e-01 -2.62324572e-01 -1.00139976e+00 8.30074489e-01
-1.55223036e+00 -9.45196390e-01 -1.16823435e-01 1.89190483e+00
1.27402771e+00 -2.04956740e-01 -2.96733886e-01 -3.59917074e-01
4.15482789e-01 3.51566911e-01 -7.83057928e-01 -8.18940476e-02
9.48943496e-02 6.05482101e-01 -3.75990391e-01 7.37598360e-01
-4.47140098e-01 9.26390648e-01 6.34024000e+00 5.34945250e-01
-4.47075903e-01 1.91495329e-01 1.68424845e-01 -9.31185782e-02
-9.32923615e-01 4.07241166e-01 -8.24221909e-01 3.01849786e-02
9.45960760e-01 -5.19942462e-01 4.80015069e-01 4.33840603e-01
6.56140894e-02 -4.01505709e-01 -1.02934802e+00 4.08284098e-01
2.65008301e-01 -1.07832301e+00 3.81004423e-01 -2.38930345e-01
5.17281532e-01 -6.26140833e-02 -2.23131850e-01 9.38006818e-01
5.04719615e-01 -7.71968663e-01 3.83831412e-01 8.30510497e-01
3.51804882e-01 -4.49260920e-01 3.75424445e-01 8.13468397e-01
-9.50594544e-01 -1.94501936e-01 -1.76927030e-01 -1.43368036e-01
4.41063464e-01 3.90255988e-01 -8.82477701e-01 5.97227812e-01
5.75874746e-01 2.37448901e-01 -6.11828744e-01 7.59160221e-01
-6.22807086e-01 9.45363283e-01 -2.59267598e-01 -3.19271803e-01
3.45406532e-01 -7.39675313e-02 3.37096930e-01 7.28379250e-01
1.98832810e-01 8.01631510e-01 2.47193292e-01 1.07993424e+00
-3.91202211e-01 3.43791634e-01 -3.60477477e-01 -1.90343052e-01
6.29914582e-01 1.01293349e+00 -5.85903972e-02 -5.88971972e-01
-6.64104521e-01 8.14547181e-01 6.08070254e-01 5.89267373e-01
-6.05607808e-01 -4.47769016e-01 8.17644149e-02 8.63249078e-02
4.48443383e-01 -6.25076219e-02 4.56154406e-01 -1.32362103e+00
4.96023983e-01 -1.29975188e+00 7.84828126e-01 -1.27121091e+00
-1.28183937e+00 4.42326069e-01 5.88436238e-03 -6.98031843e-01
-5.51146746e-01 -1.12881593e-01 -2.65122771e-01 9.79726076e-01
-2.14780641e+00 -8.91852438e-01 -2.97712594e-01 5.77788174e-01
8.74830902e-01 4.94642287e-01 7.68138587e-01 1.18442401e-01
-1.84226438e-01 9.77818370e-02 2.75296178e-02 2.42437758e-02
7.52854645e-01 -9.90810812e-01 2.95861512e-01 7.31732726e-01
2.63905615e-01 9.41625535e-01 3.43034416e-01 -6.96112812e-01
-1.67967141e+00 -9.13514435e-01 1.27831042e+00 -8.56704056e-01
7.90534258e-01 -9.40951034e-02 -1.55648303e+00 6.46031797e-01
6.19286895e-01 -3.67899954e-01 9.40640092e-01 1.55003637e-01
-6.75905049e-01 -9.26578566e-02 -7.19621301e-01 4.63213444e-01
8.27767313e-01 -1.11746716e+00 -1.61219549e+00 3.82554948e-01
1.23457789e+00 -1.70309201e-01 -6.98669255e-01 1.42186224e-01
1.75123349e-01 -6.75148249e-01 8.96783710e-01 -9.31651890e-01
5.61564267e-01 -3.02673191e-01 -1.62558392e-01 -9.82173026e-01
-1.37891695e-01 -5.52339792e-01 -5.22927523e-01 1.18044639e+00
6.68262124e-01 -5.35636008e-01 3.34873229e-01 9.65623975e-01
4.93191481e-02 -4.99687314e-01 -8.54945719e-01 -5.97982168e-01
1.80475473e-01 -3.60535771e-01 7.20086336e-01 6.56863570e-01
1.82603538e-01 9.11626935e-01 1.24610864e-01 1.22863412e-01
2.84660786e-01 4.49808419e-01 8.53254974e-01 -1.00804842e+00
-2.78598845e-01 1.23037882e-01 5.80004871e-01 -1.92937481e+00
2.33267307e-01 -7.87879884e-01 2.34509483e-01 -2.11730742e+00
3.60408038e-01 -1.15540713e-01 -1.35706827e-01 2.68896937e-01
-7.26067007e-01 -2.62534916e-01 5.54120950e-02 3.02797347e-01
-1.31842649e+00 7.29927719e-01 1.53640831e+00 -1.80721283e-01
-4.24177833e-02 -2.62780249e-01 -1.10478890e+00 3.88465106e-01
4.89385933e-01 -3.58116865e-01 -8.04470837e-01 -7.69744217e-01
6.15810215e-01 4.53914136e-01 4.10537392e-01 -6.19722188e-01
6.12342477e-01 -2.43890986e-01 -4.02445287e-01 -7.00288296e-01
2.75853008e-01 -5.81897318e-01 -4.62906957e-01 1.05267335e-02
-7.80713260e-01 9.62149166e-03 8.65220092e-04 8.07614148e-01
-4.93754357e-01 -5.66800594e-01 1.56842873e-01 -3.27015728e-01
-7.09363639e-01 2.30896369e-01 -5.62169731e-01 9.63963389e-01
6.88949406e-01 2.53607124e-01 -5.12363255e-01 -8.74021709e-01
-4.74357724e-01 6.92308426e-01 4.41044271e-02 7.16937184e-01
7.03198791e-01 -1.13938451e+00 -8.64213467e-01 -2.42426485e-01
6.55883491e-01 2.26420805e-01 3.75961572e-01 3.83615077e-01
-1.03908174e-01 8.76242578e-01 5.43807030e-01 -3.62195998e-01
-8.46373439e-01 7.23856032e-01 1.90336913e-01 -6.51016057e-01
-4.32762295e-01 7.00825155e-01 1.60589114e-01 -6.48597896e-01
6.78242669e-02 -3.04398060e-01 -4.64216262e-01 6.23792708e-02
9.00504529e-01 2.42275432e-01 4.24503768e-03 -2.32676461e-01
4.41440716e-02 2.61363506e-01 -3.46451342e-01 -3.97299945e-01
9.23287332e-01 -6.27945185e-01 -3.90793383e-01 4.10026878e-01
9.34046924e-01 8.14165995e-02 -8.14425111e-01 -9.61385190e-01
3.76723975e-01 -3.29321742e-01 -2.05116868e-01 -1.08557141e+00
-3.11807454e-01 7.87358403e-01 -8.60984623e-02 3.77863318e-01
1.08423209e+00 4.55523938e-01 1.36606312e+00 9.94683206e-01
4.14265275e-01 -1.08496594e+00 4.76792604e-01 9.78441119e-01
1.11655164e+00 -1.18534672e+00 -4.09599155e-01 -2.31739670e-01
-3.08974504e-01 8.94622028e-01 8.53307664e-01 4.66829956e-01
2.66410619e-01 -3.65576059e-01 3.37681025e-01 -4.53195751e-01
-1.15508652e+00 -3.70674759e-01 4.01369631e-01 2.41500363e-01
2.86154330e-01 -4.43082869e-01 -2.18956113e-01 7.01336443e-01
-3.71917449e-02 2.98649341e-01 4.27099794e-01 1.30596590e+00
-8.60273540e-01 -1.03919530e+00 -5.15717745e-01 4.40400183e-01
-3.56819004e-01 -4.04070735e-01 -6.49814069e-01 3.54698420e-01
-5.56890070e-01 1.50539804e+00 -1.67557850e-01 1.57603681e-01
5.05863607e-01 5.29232860e-01 1.88535437e-01 -9.48413432e-01
-4.89767730e-01 -4.59826767e-01 4.01903987e-01 -5.79336882e-01
-3.69692534e-01 -3.17978233e-01 -1.30162585e+00 9.79420766e-02
-4.24275398e-01 7.37658918e-01 1.98208272e-01 1.28055167e+00
7.16950715e-01 2.90821403e-01 4.20673877e-01 3.37192744e-01
-1.03563595e+00 -8.92798483e-01 -6.05737930e-03 3.66545558e-01
5.53894937e-01 -2.85576314e-01 -2.39802822e-01 1.09372899e-01] | [11.311891555786133, 7.949220657348633] |
8b9696f5-4171-4d85-b33d-510796a47639 | part-aware-fine-grained-object-categorization | 1806.06198 | null | https://arxiv.org/abs/1806.06198v2 | https://arxiv.org/pdf/1806.06198v2.pdf | Part-Aware Fine-grained Object Categorization using Weakly Supervised Part Detection Network | Fine-grained object categorization aims for distinguishing objects of subordinate categories that belong to the same entry-level object category. The task is challenging due to the facts that (1) training images with ground-truth labels are difficult to obtain, and (2) variations among different subordinate categories are subtle. It is well established that characterizing features of different subordinate categories are located on local parts of object instances. In fact, careful part annotations are available in many fine-grained categorization datasets. However, manually annotating object parts requires expertise, which is also difficult to generalize to new fine-grained categorization tasks. In this work, we propose a Weakly Supervised Part Detection Network (PartNet) that is able to detect discriminative local parts for use of fine-grained categorization. A vanilla PartNet builds on top of a base subnetwork two parallel streams of upper network layers, which respectively compute scores of classification probabilities (over subordinate categories) and detection probabilities (over a specified number of discriminative part detectors) for local regions of interest (RoIs). The image-level prediction is obtained by aggregating element-wise products of these region-level probabilities. To generate a diverse set of RoIs as inputs of PartNet, we propose a simple Discretized Part Proposals module (DPP) that directly targets for proposing candidates of discriminative local parts, with no bridging via object-level proposals. Experiments on the benchmark CUB-200-2011 and Oxford Flower 102 datasets show the efficacy of our proposed method for both discriminative part detection and fine-grained categorization. In particular, we achieve the new state-of-the-art performance on CUB-200-2011 dataset when ground-truth part annotations are not available. | ['Yabin Zhang', 'Zhixin Wang', 'Kui Jia'] | 2018-06-16 | null | null | null | null | ['object-categorization'] | ['computer-vision'] | [ 1.9953224e-01 1.6156308e-01 -3.0509683e-01 -5.4577696e-01
-6.6965121e-01 -8.7871516e-01 6.2590384e-01 2.6776642e-01
-1.1682148e-02 5.0557780e-01 -1.5777929e-02 1.3689211e-01
-2.4361885e-01 -9.5292211e-01 -8.4645474e-01 -6.0793442e-01
4.4780962e-02 5.6058264e-01 7.6116538e-01 6.7185163e-02
1.0692925e-01 1.0884233e+00 -1.9391749e+00 6.9077522e-01
7.0143741e-01 1.5971386e+00 2.1412764e-01 2.0816986e-01
-3.1968343e-01 4.3064505e-01 -6.0331762e-01 -2.3719613e-01
5.2695966e-01 -1.1670998e-01 -9.1235584e-01 4.1673326e-01
8.3484137e-01 -8.7149188e-02 2.1405157e-02 1.0986507e+00
2.8922020e-02 -2.4535529e-02 9.7986001e-01 -1.3754243e+00
-5.3862780e-01 7.8562367e-01 -5.3796238e-01 1.2834132e-01
-2.4479799e-01 1.5312363e-01 1.0847286e+00 -8.0447108e-01
3.8412291e-01 1.4935131e+00 7.0039779e-01 4.8493838e-01
-1.5567572e+00 -5.9770560e-01 6.7054319e-01 -6.4026108e-03
-1.8136506e+00 -2.9982769e-01 5.9578615e-01 -4.9410483e-01
6.8580210e-01 1.3170360e-01 3.7220973e-01 6.8919551e-01
2.0301765e-01 6.6519928e-01 1.2102500e+00 -3.2197583e-02
3.9101413e-01 2.9235148e-01 4.2651483e-01 6.8960041e-01
4.2642346e-01 -1.5938845e-01 -1.6403346e-01 -2.7168977e-01
6.6579866e-01 3.4481719e-01 -6.8128131e-02 -8.2253301e-01
-1.0082507e+00 6.4324653e-01 1.1213329e+00 6.1994404e-01
-5.4715049e-01 1.4596000e-01 3.2685915e-01 -1.6695529e-02
2.7339283e-01 1.9548079e-01 -7.3192132e-01 5.2947062e-01
-1.0874172e+00 7.3881112e-02 6.4589435e-01 1.0017170e+00
1.1844041e+00 -3.9989698e-01 -8.5231686e-01 9.1430593e-01
4.1332233e-01 8.5253939e-02 4.7571209e-01 -6.9271243e-01
1.9120249e-01 1.1898104e+00 -1.0762882e-01 -6.5648258e-01
-3.2104462e-01 -7.5037915e-01 -1.0660988e+00 2.9103801e-01
4.6303847e-01 4.4273639e-01 -1.2326673e+00 1.7049100e+00
4.3710664e-01 -1.0078437e-01 -4.8339403e-01 7.4822164e-01
9.6184736e-01 2.3378035e-01 4.9147436e-01 2.8858224e-01
1.7614899e+00 -9.5311636e-01 2.7814528e-02 -2.7479291e-01
2.7697453e-01 -4.8356426e-01 9.1876400e-01 -1.4251681e-03
-6.8246019e-01 -1.1971142e+00 -8.6527091e-01 2.7606262e-02
-7.8192502e-01 5.0923550e-01 6.9293684e-01 6.2502617e-01
-1.0130130e+00 7.9715151e-01 -6.0073674e-01 -3.8339305e-01
1.0779592e+00 3.8606566e-01 -5.5798912e-01 -9.4867840e-02
-6.2138593e-01 5.1519066e-01 6.1445272e-01 8.1059806e-02
-1.2183957e+00 -7.0167154e-01 -7.7836108e-01 5.1037014e-01
2.1051268e-01 -6.7398983e-01 9.2699826e-01 -1.2096453e+00
-9.9053550e-01 1.1390694e+00 8.2666174e-02 -3.5134205e-01
2.6702344e-01 3.2768881e-01 -1.3761505e-01 2.1774937e-01
4.8343781e-01 1.2484928e+00 1.0530322e+00 -1.3615884e+00
-9.3517333e-01 -4.0991166e-01 3.4569162e-01 -2.0006293e-01
7.5561027e-03 -3.7267050e-01 -4.1427100e-01 -6.0276616e-01
3.3705840e-01 -5.7871658e-01 -2.2486529e-01 4.2496434e-01
-5.8829516e-01 -7.7894634e-01 7.2848672e-01 -5.9355479e-02
8.6835444e-01 -2.0627811e+00 -3.3089408e-01 1.1978069e-01
2.5645253e-01 1.7716115e-02 -4.1953072e-01 6.7038260e-02
-9.8985456e-02 1.2781106e-01 -1.7290726e-01 -1.5947817e-02
2.8376243e-01 1.3349257e-01 -2.8314474e-01 4.9094212e-01
6.3220495e-01 9.6503085e-01 -4.2515975e-01 -6.7324007e-01
2.5482890e-01 1.9987845e-01 -5.6423622e-01 1.7049434e-02
-2.0208108e-01 8.0650061e-02 -7.2880608e-01 1.0885659e+00
9.8262203e-01 -3.5848480e-01 -8.0588192e-02 -6.2549484e-01
-5.1844511e-02 1.8526860e-01 -1.2790806e+00 1.4882245e+00
-2.9967090e-01 1.4989935e-01 1.7816541e-01 -1.1008642e+00
7.5600010e-01 -1.1287000e-01 1.3138407e-01 -2.4498172e-01
3.0865067e-01 2.3515353e-02 2.0330204e-01 1.3638599e-01
2.7871412e-01 -2.1822333e-01 -4.1021308e-01 2.6543716e-01
5.5666459e-01 -1.8423930e-01 4.3869567e-01 -6.7064255e-02
1.1853975e+00 3.0363369e-01 6.5606135e-01 -5.7795501e-01
6.7691040e-01 1.4002833e-01 5.6590265e-01 1.0558858e+00
-5.8145523e-01 6.2230998e-01 5.0000989e-01 -3.8810959e-01
-7.2561800e-01 -1.1255847e+00 -6.4869606e-01 1.5271138e+00
3.1902567e-01 -1.2485469e-01 -7.6727843e-01 -1.2084689e+00
3.1282556e-01 2.6298717e-01 -1.1373698e+00 -8.1489854e-02
-2.6128289e-01 -4.7078735e-01 5.3221250e-01 7.3506200e-01
6.4227206e-01 -1.3139170e+00 -3.9792827e-01 4.1252859e-02
5.0705347e-02 -9.4343817e-01 -2.7197039e-01 7.1722287e-01
-8.5218543e-01 -1.0777006e+00 -7.8782111e-01 -9.0768015e-01
9.6326274e-01 6.3756013e-01 1.3940892e+00 8.8864006e-03
-5.4526818e-01 1.8442157e-01 -2.8282490e-01 -2.0166442e-01
-2.6253891e-01 7.9858124e-02 -2.1231437e-01 2.9485387e-01
3.6497617e-01 -4.1443408e-01 -6.4141911e-01 7.2308791e-01
-8.6443579e-01 -3.4955087e-01 1.1924695e+00 9.3293387e-01
1.0718101e+00 1.7360587e-01 3.2000291e-01 -7.3875481e-01
-5.0879486e-02 -3.5345736e-01 -6.9530040e-01 4.0881240e-01
-5.7412591e-02 8.6825430e-02 7.2488934e-01 -4.3948206e-01
-9.3448699e-01 2.7959585e-01 4.6425518e-02 -5.3030747e-01
-9.0812296e-01 -1.8856569e-01 -5.6434381e-01 -1.8840681e-01
6.8128717e-01 2.1106952e-01 -5.9730452e-01 -6.2345183e-01
5.5376881e-01 7.1967679e-01 2.9209056e-01 -6.2742591e-01
9.3741983e-01 6.3972008e-01 5.6995302e-02 -4.5790228e-01
-1.0880613e+00 -8.1561244e-01 -1.1297728e+00 1.6643475e-01
8.1763542e-01 -1.0243812e+00 -7.3193675e-01 1.2332249e-01
-8.1515956e-01 -2.2596614e-01 -6.4262986e-01 -7.3645890e-02
-4.4554877e-01 2.4515077e-01 -3.4796602e-01 -2.2465441e-01
-2.4426803e-01 -9.5117986e-01 1.6433332e+00 4.7040313e-01
8.6486347e-02 -4.4846931e-01 -3.4651291e-01 1.2865180e-01
2.1466234e-01 -1.2031192e-01 8.5335886e-01 -6.4145207e-01
-8.0834234e-01 -3.7855995e-01 -7.5869662e-01 4.0483704e-01
1.8408300e-01 -5.0608825e-02 -1.2769110e+00 -2.4288866e-01
-3.7344882e-01 -4.3252316e-01 1.2026204e+00 5.2336454e-01
1.6197587e+00 -1.8641698e-01 -8.3076435e-01 4.2417133e-01
1.4840384e+00 -2.0617129e-01 2.2746772e-01 -1.2268243e-01
5.9572327e-01 6.8692762e-01 7.5689381e-01 4.5941764e-01
-4.9266424e-02 6.6934830e-01 4.2731452e-01 -1.6983639e-02
-4.9455133e-01 -2.3484614e-01 -3.7074994e-02 -1.3730910e-01
1.3716149e-02 -1.1794552e-01 -4.2142415e-01 6.9893509e-01
-1.5401757e+00 -1.0650467e+00 3.5229694e-02 1.9309810e+00
5.8661002e-01 2.6882067e-01 1.1782476e-01 -7.1045075e-04
9.4257289e-01 -9.8753721e-02 -7.2590363e-01 -6.1563354e-02
-4.4678487e-02 3.4252521e-01 5.3991878e-01 -1.4961056e-01
-1.5884833e+00 1.0478764e+00 5.2705660e+00 1.3367018e+00
-8.7618119e-01 1.6670333e-01 8.3415627e-01 3.8466501e-01
1.6108984e-01 1.0011790e-02 -1.2109722e+00 2.2532494e-01
3.6276385e-01 3.6636177e-01 1.1456009e-02 1.3441308e+00
-2.7002239e-01 -1.0803928e-01 -1.2353942e+00 7.7977771e-01
-1.3225441e-01 -1.2815766e+00 3.0936736e-01 -7.2441645e-02
7.8047228e-01 7.7497661e-03 -3.1221595e-01 6.6622382e-01
2.9461554e-01 -9.2465061e-01 1.0538224e+00 3.1233060e-01
8.1294686e-01 -4.5968160e-01 7.0699209e-01 4.1661093e-01
-1.7404563e+00 -3.6769313e-01 -1.0662589e+00 1.8244433e-01
-3.3699074e-01 5.3562671e-01 -8.3584338e-01 3.0266216e-01
8.6075634e-01 4.8571613e-01 -8.3266550e-01 1.1928437e+00
-2.8053144e-01 4.8578838e-01 -3.2556146e-01 1.5357283e-01
3.3830446e-01 2.0460281e-01 -8.9744199e-03 1.3082829e+00
6.3116148e-02 1.3189909e-01 4.2874119e-01 1.2137762e+00
-3.8185734e-01 -2.3563352e-01 -3.0490160e-01 -1.0426581e-02
2.7992368e-01 2.0239596e+00 -1.3862300e+00 -4.9074379e-01
-2.8964469e-01 9.0514547e-01 4.7270933e-01 3.4225121e-02
-6.7442024e-01 -4.4999948e-01 8.9457053e-01 1.4233375e-01
9.1877568e-01 3.0812457e-01 -2.9538319e-01 -1.0411009e+00
-2.3758738e-01 -4.7409818e-01 6.4601743e-01 -5.4313600e-01
-1.8163475e+00 7.5047034e-01 -5.4420847e-02 -1.4670415e+00
1.3184106e-01 -9.0476912e-01 -4.2434463e-01 9.2375976e-01
-1.4180881e+00 -1.6468612e+00 -6.7035329e-01 6.2958276e-01
6.3677275e-01 1.5050080e-01 9.0267664e-01 1.2189474e-01
-3.4992513e-01 6.2957579e-01 -2.1870956e-01 3.2362917e-01
4.9093425e-01 -1.2306705e+00 1.4917688e-01 6.4712971e-01
2.1557593e-01 5.2748352e-01 2.3770151e-01 -4.6723461e-01
-8.5369927e-01 -1.5961111e+00 6.5517539e-01 -3.5988292e-01
5.2295321e-01 -6.0362124e-01 -7.0848852e-01 4.6210381e-01
-2.9521295e-01 7.5581968e-01 3.6919212e-01 -2.4055682e-02
-6.0911536e-01 -3.6835754e-01 -1.4935795e+00 8.2672827e-02
1.1636097e+00 -4.6135861e-01 -6.3069052e-01 3.1505147e-01
4.6725118e-01 9.5240250e-02 -7.4700308e-01 4.2411211e-01
5.1822853e-01 -1.0326749e+00 1.0949026e+00 -5.5376261e-01
2.0363836e-01 -6.7706811e-01 -4.0957057e-01 -9.4289088e-01
-1.0027287e+00 1.1894809e-01 1.4134499e-01 1.4151216e+00
2.2401325e-02 -3.4598556e-01 7.8423524e-01 1.5090875e-01
-3.7542799e-01 -5.7293612e-01 -8.3818358e-01 -9.6476632e-01
1.4968669e-02 -2.5635087e-01 7.3025423e-01 4.0432349e-01
-5.3385383e-01 2.8220692e-01 5.0077754e-01 3.9508304e-01
6.0022461e-01 9.6482432e-01 5.8647990e-01 -1.6426731e+00
-4.4474646e-01 -7.7915674e-01 -7.7435708e-01 -1.0492288e+00
2.7707183e-01 -9.3870533e-01 4.7484502e-01 -1.6893271e+00
6.0640705e-01 -8.2303876e-01 -6.4048296e-01 1.0059686e+00
-4.7960237e-02 9.3397784e-01 1.7239048e-01 3.2803157e-01
-6.8280482e-01 4.4701582e-01 1.2779118e+00 -5.4439479e-01
7.4132837e-02 2.5845376e-01 -8.6211568e-01 8.3104497e-01
4.4501087e-01 -4.9481130e-01 -2.8875539e-02 2.9921505e-01
-5.9491134e-01 -3.9605847e-01 6.4923120e-01 -1.3210523e+00
-1.8820900e-01 -7.8090571e-02 9.2315853e-01 -9.0092880e-01
3.0030674e-01 -8.3378661e-01 -5.4910168e-02 4.5177135e-01
-7.7690765e-02 -7.4627852e-01 3.4208632e-01 4.7467661e-01
-2.3132819e-01 -2.9745129e-01 1.2072977e+00 -2.9974246e-01
-1.0277983e+00 4.8955962e-01 -1.8158762e-01 -7.6211572e-02
1.0603218e+00 -3.9740071e-01 -2.0235853e-01 3.8030908e-01
-6.9593537e-01 -2.8653768e-01 4.4187003e-01 2.6886380e-01
1.8190351e-01 -1.2815727e+00 -4.8144734e-01 2.6848704e-01
6.6035783e-01 -1.7114064e-02 4.7942400e-01 5.8204913e-01
-9.0166405e-02 7.2239435e-01 -5.3446567e-01 -7.7564788e-01
-1.1308506e+00 9.3999928e-01 4.1068947e-01 -2.8937086e-01
-3.3845562e-01 1.1302315e+00 1.0778997e+00 -6.7248881e-01
1.1615611e-01 -7.8856623e-01 -3.4022784e-01 1.5073997e-01
3.8222215e-01 1.0757004e-01 1.5557261e-01 -7.9661787e-01
-6.2573415e-01 6.5781498e-01 -4.1410506e-02 5.6883782e-01
1.2193257e+00 -7.4728876e-03 -6.7979664e-02 7.1330473e-02
1.0026599e+00 -1.3010953e-01 -1.4155319e+00 -2.2523481e-01
-1.4452904e-01 -2.1072124e-01 -1.8598902e-01 -1.1326416e+00
-8.7004060e-01 8.7512261e-01 7.7346861e-01 2.3711610e-01
1.1998608e+00 6.7404050e-01 1.2521824e-01 3.4314749e-01
5.8832610e-01 -6.9922483e-01 -2.1426626e-01 3.4535658e-01
8.3241278e-01 -1.0715443e+00 -1.1629632e-01 -6.1455327e-01
-1.8589531e-01 1.0180330e+00 8.1952089e-01 -3.2274210e-01
7.8144723e-01 -5.6310587e-02 -1.3562091e-01 -1.6442449e-01
-5.1616323e-01 -7.0518076e-01 7.5197691e-01 6.1096227e-01
3.4540561e-01 3.1728077e-01 -7.8920998e-02 9.6087295e-01
1.8720761e-01 -1.9797474e-01 -4.7439564e-02 6.2938482e-01
-7.4045962e-01 -1.0422101e+00 -5.5500114e-01 7.4813110e-01
-3.0719659e-01 9.2647649e-02 -4.7660258e-01 7.9140323e-01
8.9302075e-01 7.6934850e-01 3.3238915e-01 2.2656733e-02
2.5811589e-01 -8.1172511e-02 5.7811791e-01 -8.8902944e-01
-6.9624048e-01 -5.5460293e-02 -2.1267468e-01 -8.4911448e-01
-3.2317159e-01 -4.1425383e-01 -1.1286328e+00 1.4494877e-01
-3.2753941e-01 6.9838494e-02 3.7629876e-01 8.9450032e-01
4.8993567e-01 6.4691222e-01 5.2403456e-01 -1.2654455e+00
-5.2690023e-01 -1.2031775e+00 -9.9919540e-01 4.5139515e-01
-1.6871868e-02 -9.3560702e-01 -1.4079928e-01 -1.0427886e-01] | [9.579010963439941, 1.9908488988876343] |
fa940aff-d1e2-4af0-94d2-c0cef6ced2fe | time-series-anomaly-detection-via-contextual | 2304.07898 | null | https://arxiv.org/abs/2304.07898v1 | https://arxiv.org/pdf/2304.07898v1.pdf | Time-series Anomaly Detection via Contextual Discriminative Contrastive Learning | Detecting anomalies in temporal data is challenging due to anomalies being dependent on temporal dynamics. One-class classification methods are commonly used for anomaly detection tasks, but they have limitations when applied to temporal data. In particular, mapping all normal instances into a single hypersphere to capture their global characteristics can lead to poor performance in detecting context-based anomalies where the abnormality is defined with respect to local information. To address this limitation, we propose a novel approach inspired by the loss function of DeepSVDD. Instead of mapping all normal instances into a single hypersphere center, each normal instance is pulled toward a recent context window. However, this approach is prone to a representation collapse issue where the neural network that encodes a given instance and its context is optimized towards a constant encoder solution. To overcome this problem, we combine our approach with a deterministic contrastive loss from Neutral AD, a promising self-supervised learning anomaly detection approach. We provide a theoretical analysis to demonstrate that the incorporation of the deterministic contrastive loss can effectively prevent the occurrence of a constant encoder solution. Experimental results show superior performance of our model over various baselines and model variants on real-world industrial datasets. | ['Tony S. Wirjanto', 'Mingbin Feng', 'Katrina Chen'] | 2023-04-16 | null | null | null | null | ['one-class-classification', 'time-series-anomaly-detection'] | ['miscellaneous', 'time-series'] | [ 3.16544682e-01 -1.50157958e-01 1.68314084e-01 -4.15940434e-01
-3.29540282e-01 -3.19296420e-01 6.64925575e-01 5.53340375e-01
-1.24884672e-01 4.54576343e-01 -3.37787330e-01 -1.50162518e-01
-1.28826156e-01 -7.98063815e-01 -7.16732681e-01 -1.00007963e+00
-4.27790016e-01 2.38099962e-01 4.03827488e-01 -1.97563142e-01
4.35063571e-01 4.83171165e-01 -1.56077981e+00 8.77198949e-02
1.21304619e+00 1.43403506e+00 -2.38994122e-01 3.37221831e-01
-4.00276065e-01 7.21937835e-01 -9.89648283e-01 -1.75250292e-01
4.59553033e-01 -5.61443985e-01 -4.68876302e-01 -1.49857886e-02
5.55840731e-01 -1.11051731e-01 -2.37677842e-01 1.30631959e+00
1.77251250e-01 3.45310271e-01 6.15376174e-01 -1.46886146e+00
-4.48010504e-01 1.57668710e-01 -7.70402908e-01 7.46903419e-01
-4.10263799e-02 8.47993791e-02 1.11064816e+00 -4.96533304e-01
3.51551086e-01 8.93991411e-01 4.75306809e-01 5.10338724e-01
-1.10279071e+00 -5.11597514e-01 7.61170089e-01 3.78699630e-01
-1.16227686e+00 -1.50218219e-01 1.14754510e+00 -3.62625241e-01
7.55300403e-01 3.02779704e-01 6.20497704e-01 1.08876312e+00
5.35998285e-01 6.88518226e-01 5.72337449e-01 -1.43133268e-01
5.29167235e-01 -8.23624060e-02 1.07629105e-01 5.31663835e-01
1.70168325e-01 9.86317405e-04 -4.69612688e-01 -3.05045396e-01
4.04982001e-01 2.89171398e-01 -2.06118494e-01 -5.46895146e-01
-7.25429296e-01 8.39924932e-01 5.56675434e-01 3.62532020e-01
-4.42458153e-01 -6.84855357e-02 6.01127863e-01 5.62768042e-01
8.09695899e-01 5.64324975e-01 -2.55822569e-01 -1.51373163e-01
-6.98925793e-01 3.74813914e-01 4.70253736e-01 3.56408238e-01
4.32270795e-01 3.37735146e-01 -3.03293556e-01 7.27542460e-01
1.59493729e-01 1.54509589e-01 6.26527548e-01 -4.66836452e-01
4.87278789e-01 9.17686224e-01 -1.17199473e-01 -1.25928688e+00
-2.55288005e-01 -6.75328553e-01 -8.76317143e-01 3.08968335e-01
5.05860388e-01 5.23454249e-02 -8.07217717e-01 1.80126250e+00
4.79460090e-01 5.89431703e-01 1.22823365e-01 8.83598030e-01
5.54500334e-02 7.69017100e-01 -9.60252136e-02 -2.11276993e-01
8.66892755e-01 -8.00889969e-01 -6.81632340e-01 -3.15098673e-01
5.94384909e-01 -2.72989392e-01 1.09404385e+00 4.12702203e-01
-7.81917095e-01 -1.61527544e-01 -1.20854270e+00 4.53286409e-01
-5.73370218e-01 -3.01308006e-01 3.62548470e-01 2.11342528e-01
-5.67126870e-01 8.66230369e-01 -1.10438240e+00 -4.24891442e-01
3.98250878e-01 2.50938553e-02 -4.83508967e-02 2.56920338e-01
-1.10637677e+00 5.98065972e-01 4.49346483e-01 1.49979293e-01
-6.58529580e-01 -6.22635722e-01 -7.89208710e-01 2.99233268e-03
3.64365131e-01 -1.44908950e-01 9.59160089e-01 -1.01188171e+00
-1.16238725e+00 5.02933562e-01 -1.32553831e-01 -9.78693902e-01
5.60290277e-01 -3.88144404e-01 -8.11331093e-01 3.84571962e-02
-4.63870238e-04 -1.15057960e-01 1.14082563e+00 -1.00193310e+00
-7.21317351e-01 -4.82553095e-01 -1.38555810e-01 6.03594668e-02
-8.27873468e-01 -1.64064333e-01 -7.18468949e-02 -8.88884187e-01
3.97409320e-01 -7.95567393e-01 -2.87469923e-01 6.49484843e-02
-4.43160713e-01 -2.06244543e-01 1.32866991e+00 -4.09285694e-01
1.64252424e+00 -2.21708488e+00 -6.45744577e-02 4.43460226e-01
1.15609452e-01 2.82625347e-01 1.50690928e-01 1.86575279e-01
-1.09712347e-01 -9.16933939e-02 -8.37513268e-01 -1.80554137e-01
-1.96038738e-01 3.14620703e-01 -8.84247363e-01 6.06932938e-01
4.95137960e-01 3.72527927e-01 -9.91718531e-01 -8.28971639e-02
2.51497149e-01 2.29250118e-01 -5.56781352e-01 2.30363533e-01
-3.26092929e-01 4.54578847e-01 -5.44704497e-01 6.93047345e-01
6.58354163e-01 -1.03224151e-01 -2.46910676e-01 3.31994772e-01
-9.30102356e-03 2.66159624e-01 -1.02395928e+00 1.47323620e+00
-2.79650658e-01 4.35493886e-01 -1.17307767e-01 -1.49494243e+00
1.11804044e+00 1.25270069e-01 7.84500599e-01 -7.27531910e-01
-3.56703192e-01 4.02867138e-01 -2.74679512e-02 -3.02389205e-01
2.59581119e-01 -3.08860652e-02 8.21123049e-02 4.02298540e-01
-3.49098653e-01 1.26362234e-01 1.43190309e-01 4.43036743e-02
1.43526053e+00 -7.03968629e-02 7.11346278e-03 -2.40542278e-01
7.31905162e-01 -1.71391681e-01 9.36739683e-01 7.35594153e-01
-4.63698387e-01 7.03665435e-01 8.43734860e-01 -8.81158292e-01
-9.08030748e-01 -1.13112772e+00 -4.32788283e-01 8.21471453e-01
2.37260476e-01 -4.14828837e-01 -4.67022508e-01 -1.12159264e+00
1.52198404e-01 9.89054620e-01 -6.42089605e-01 -7.12617218e-01
-7.55713463e-01 -1.08857179e+00 2.83949494e-01 3.82548153e-01
5.49423695e-01 -1.06558621e+00 -8.02751899e-01 3.19666296e-01
-4.47434746e-02 -8.73897851e-01 -2.84821481e-01 2.67323911e-01
-9.57974851e-01 -9.74352241e-01 -4.09622341e-01 -3.14840525e-01
6.70775056e-01 -1.43218279e-01 1.05511677e+00 9.19621587e-02
-3.58960330e-01 1.98766083e-01 -3.43315750e-01 -5.57541192e-01
-3.46518278e-01 6.26811571e-03 2.15828523e-01 4.98558521e-01
5.45006275e-01 -6.88987553e-01 -7.24977434e-01 2.78837085e-01
-1.07419634e+00 -6.34153366e-01 1.73037782e-01 7.94285536e-01
5.62924504e-01 3.22248608e-01 7.18241334e-01 -6.09239697e-01
5.80450416e-01 -8.28066707e-01 -7.26996064e-01 9.95049030e-02
-8.65116358e-01 6.06598593e-02 8.03879619e-01 -3.94722283e-01
-8.99742246e-01 -3.19699496e-02 1.39907628e-01 -6.44454777e-01
1.74250193e-02 3.82720470e-01 1.57261372e-01 2.11089537e-01
6.88312650e-01 4.45679545e-01 1.24595791e-01 -3.26846123e-01
-2.25940585e-01 3.52067202e-01 5.29152572e-01 -6.71064913e-01
8.65219951e-01 5.84459782e-01 2.16723997e-02 -7.79646575e-01
-9.01384473e-01 -4.65193480e-01 -3.48838776e-01 -1.30891725e-01
6.32258654e-01 -5.74766755e-01 -2.75167853e-01 5.89410543e-01
-9.24957752e-01 3.44499238e-02 -6.37642324e-01 2.29337722e-01
-3.02255571e-01 3.79581779e-01 -3.54393482e-01 -9.69498277e-01
-2.80900508e-01 -7.74934411e-01 9.38687742e-01 7.78310522e-02
-3.30345482e-02 -9.75340664e-01 2.85150409e-01 -2.43778020e-01
5.24852812e-01 6.64244413e-01 9.14430797e-01 -1.04050136e+00
-3.26158702e-01 -4.65249836e-01 1.16840184e-01 5.28518796e-01
3.31730515e-01 1.64859414e-01 -9.07445788e-01 -4.44991857e-01
3.69751126e-01 -1.00265987e-01 1.02455997e+00 7.53648356e-02
1.60614264e+00 -4.35119867e-01 -3.65886718e-01 5.60454190e-01
1.17636228e+00 3.74748260e-01 4.57838565e-01 6.29793465e-01
6.23542845e-01 5.18692076e-01 7.20759213e-01 6.82554722e-01
4.35682349e-02 6.20072961e-01 9.02648509e-01 1.53094262e-01
4.32221383e-01 5.09613641e-02 4.86510545e-01 4.82394755e-01
1.71368197e-01 -2.51857907e-01 -1.12466204e+00 6.08454585e-01
-1.94544828e+00 -9.98253345e-01 1.33255705e-01 2.56316781e+00
5.08529782e-01 4.77590412e-01 6.05185106e-02 2.73894161e-01
6.62341714e-01 4.39382434e-01 -9.25163865e-01 -5.14083087e-01
-2.69442081e-01 -1.36656314e-01 5.30314334e-02 8.99167582e-02
-1.41186345e+00 6.17257178e-01 5.61231518e+00 5.46084642e-01
-1.41514897e+00 -2.38364145e-01 5.11176527e-01 -3.30137819e-01
-7.31130689e-02 -2.68003672e-01 -5.17450750e-01 7.89849520e-01
9.46398139e-01 -2.59187073e-01 1.68972746e-01 8.32453907e-01
-5.24176331e-03 2.30972648e-01 -1.16303706e+00 7.22294569e-01
-2.23100279e-02 -9.92809474e-01 1.84834123e-01 8.60720426e-02
6.61934435e-01 4.10311408e-02 2.75308073e-01 3.08134377e-01
-1.51071161e-01 -8.01381111e-01 4.94409859e-01 4.38595116e-01
1.42162278e-01 -9.13815260e-01 5.43522298e-01 1.18086763e-01
-1.16504037e+00 -5.09917200e-01 -4.31710958e-01 3.06891296e-02
-8.57951194e-02 8.57280195e-01 -7.34482944e-01 4.41189438e-01
9.33732927e-01 8.68077040e-01 -5.64040542e-01 1.00112760e+00
1.84134677e-01 7.05579937e-01 -4.43219244e-01 3.63686919e-01
3.55607897e-01 -4.71119016e-01 1.12296081e+00 8.98833752e-01
4.86851007e-01 -3.31183940e-01 2.56782055e-01 1.08183229e+00
2.34044179e-01 1.24335967e-01 -9.24279273e-01 -1.68120816e-01
2.72110313e-01 8.47104311e-01 -5.90285659e-01 -1.26861379e-01
-4.09706920e-01 1.06326568e+00 2.77555466e-01 3.37187707e-01
-8.76615107e-01 -6.11060977e-01 1.00143421e+00 2.51554877e-01
2.38335073e-01 -2.50964565e-03 -2.75878817e-01 -1.24996996e+00
5.84773719e-01 -7.36912549e-01 7.95341313e-01 -1.24892592e-01
-1.48589027e+00 7.00827599e-01 -1.62654638e-01 -1.67706108e+00
-4.83201593e-01 -4.26256686e-01 -1.20749819e+00 5.02339721e-01
-1.42673779e+00 -5.84232330e-01 -1.97310522e-01 6.63643241e-01
5.19375503e-01 -4.14564818e-01 6.12791538e-01 2.85968930e-01
-9.00058329e-01 6.85238361e-01 2.10022986e-01 -1.88405979e-02
6.92256331e-01 -1.58585882e+00 5.04176795e-01 1.23510456e+00
6.75689355e-02 4.54007536e-01 8.19383442e-01 -6.85387611e-01
-8.91803503e-01 -1.32436705e+00 5.36399782e-01 -3.71098340e-01
8.41210544e-01 -2.61674523e-01 -1.60320842e+00 5.86994588e-01
-1.32061258e-01 5.72088957e-01 3.07272345e-01 6.20091930e-02
-2.77524740e-01 -4.28335369e-01 -1.28405261e+00 5.91826737e-01
9.46453214e-01 -3.88926059e-01 -5.58129489e-01 4.51368481e-01
6.30193830e-01 -3.63314301e-01 -6.22833133e-01 7.35060275e-01
2.80054584e-02 -1.04920101e+00 7.39993334e-01 -6.73094451e-01
4.54409331e-01 -4.26838726e-01 -3.53077166e-02 -1.41312671e+00
8.76031443e-02 -7.03998566e-01 -7.08629727e-01 1.14353859e+00
1.94689065e-01 -1.05311954e+00 7.14375615e-01 4.27318633e-01
-3.00667763e-01 -9.58554447e-01 -1.26508474e+00 -1.01314270e+00
1.45917878e-01 -3.83735567e-01 6.39566243e-01 1.09317803e+00
-1.14963777e-01 -1.35997877e-01 -1.62735224e-01 4.35050607e-01
6.11766160e-01 1.65608078e-01 3.91136825e-01 -1.47245288e+00
8.82672518e-03 -6.13895118e-01 -7.39865839e-01 -6.22176945e-01
1.35913357e-01 -7.38653302e-01 3.91729642e-03 -9.35037673e-01
-2.75604337e-01 -5.23990154e-01 -9.68597233e-01 2.73971319e-01
-1.54441699e-01 -2.28732955e-02 -1.36872590e-01 1.88711792e-01
-4.71119702e-01 8.57800961e-01 8.23078573e-01 -1.57410160e-01
-3.37783575e-01 1.93120658e-01 -2.54555821e-01 7.73540080e-01
9.83906507e-01 -5.84151268e-01 -5.80375552e-01 -1.71657458e-01
9.10149738e-02 -3.48452687e-01 4.01935279e-01 -1.30653763e+00
2.62337565e-01 -1.23890795e-01 2.45692953e-01 -5.24693847e-01
1.51240617e-01 -8.75928164e-01 -3.78152847e-01 5.53868175e-01
-3.24163556e-01 3.60211551e-01 2.82753497e-01 1.03549755e+00
-4.17490274e-01 -2.14367621e-02 7.88579822e-01 1.27837777e-01
-6.34354591e-01 3.84466499e-01 -2.15998739e-01 2.14939460e-01
1.28673673e+00 -1.59986556e-01 -1.96713775e-01 -1.84001714e-01
-5.18942297e-01 4.37367857e-01 4.36725497e-01 7.21570969e-01
5.82428694e-01 -1.37720156e+00 -5.49473166e-01 6.34199440e-01
4.21253055e-01 2.57776260e-01 1.03067555e-01 7.31506467e-01
-3.92758489e-01 1.93693355e-01 -1.16793871e-01 -9.55452979e-01
-8.00480664e-01 5.82585394e-01 7.09983289e-01 -2.28833586e-01
-1.19845474e+00 5.96958876e-01 3.31222802e-01 -4.19700891e-01
4.40107763e-01 -3.22255045e-01 -1.69718057e-01 -6.86463714e-02
5.59657931e-01 3.26641053e-01 2.50855088e-01 -3.79090011e-01
-4.51092064e-01 3.92375439e-01 -3.14874589e-01 2.27642909e-01
1.13502252e+00 -3.37670511e-03 -8.37673023e-02 7.44312584e-01
1.02830887e+00 -1.39381930e-01 -1.39314044e+00 -3.46731693e-01
2.87607521e-01 -4.31059986e-01 -2.06621047e-02 -4.78738844e-01
-1.29056191e+00 6.13057911e-01 8.49555373e-01 5.61953306e-01
1.29959452e+00 -2.23401740e-01 8.98348451e-01 5.61931193e-01
-8.92062671e-03 -1.18420208e+00 3.68967772e-01 5.65195262e-01
8.73653412e-01 -1.35047293e+00 -3.10365468e-01 3.63634974e-02
-6.85336471e-01 1.08085656e+00 1.06086540e+00 -3.09588313e-01
7.18690097e-01 4.88390289e-02 7.31441677e-02 -2.98183590e-01
-8.24333668e-01 2.59600997e-01 2.95574665e-01 3.35335135e-01
1.61848679e-01 -2.43068442e-01 -2.93842759e-02 2.45489985e-01
-1.00211397e-01 -6.15242422e-01 3.85390669e-01 1.09402561e+00
-3.53072286e-01 -7.55848050e-01 -2.84638822e-01 5.92812181e-01
-7.19969511e-01 1.10107519e-01 -2.10910916e-01 5.77473938e-01
-7.03224018e-02 6.35513365e-01 6.58336401e-01 -2.10473746e-01
3.55288297e-01 4.37241197e-01 -1.10319749e-01 -4.12134230e-01
-4.17262435e-01 -3.03917795e-01 -4.98917282e-01 -6.98830009e-01
9.58089605e-02 -8.29921901e-01 -1.38434410e+00 4.04205546e-03
1.66619488e-03 2.53211290e-01 5.23847103e-01 8.47260833e-01
4.10909683e-01 7.28620172e-01 9.02402818e-01 -3.79481614e-01
-9.05843556e-01 -5.98303854e-01 -6.10540628e-01 7.69554377e-01
8.28120053e-01 -7.43663549e-01 -8.85875940e-01 -3.82935166e-01] | [7.582258701324463, 2.3880434036254883] |
e88edafc-bc69-4e72-8b0d-c8e0f99ceb7b | intrinseqnet-learning-to-estimate-the | 1906.05893 | null | https://arxiv.org/abs/1906.05893v1 | https://arxiv.org/pdf/1906.05893v1.pdf | IntrinSeqNet: Learning to Estimate the Reflectance from Varying Illumination | Intrinsic image decomposition describes an image based on its reflectance and shading components. In this paper we tackle the problem of estimating the diffuse reflectance from a sequence of images captured from a fixed viewpoint under various illuminations. To this end we propose a deep learning approach to avoid heuristics and strong assumptions on the reflectance prior. We compare two network architectures: one classic 'U' shaped Convolutional Neural Network (CNN) and a Recurrent Neural Network (RNN) composed of Convolutional Gated Recurrent Units (CGRU). We train our networks on a new dataset specifically designed for the task of intrinsic decomposition from sequences. We test our networks on MIT and BigTime datasets and outperform state-of-the-art algorithms both qualitatively and quantitatively. | ['Grégoire Nieto', 'Mohammad Rouhani', 'Philippe Robert'] | 2019-06-13 | null | null | null | null | ['intrinsic-image-decomposition'] | ['computer-vision'] | [ 5.96881688e-01 -1.56116873e-01 4.40279782e-01 -3.61349344e-01
-6.38479173e-01 -2.75487810e-01 2.68827528e-01 -5.95774531e-01
-3.07109028e-01 3.72717768e-01 1.93457529e-01 -1.53642222e-01
3.66359621e-01 -7.65026748e-01 -9.80063736e-01 -8.87165904e-01
2.72248864e-01 1.13879748e-01 -1.68213621e-01 -1.23716220e-01
1.00015044e-01 4.35510188e-01 -1.44241869e+00 6.55474722e-01
3.41824025e-01 1.02631235e+00 2.41027385e-01 1.14035618e+00
2.41969377e-01 1.27222145e+00 -3.61061424e-01 -9.65731144e-02
5.48942387e-01 -3.72706264e-01 -8.99793983e-01 4.50950056e-01
3.89758050e-01 -6.11580074e-01 -5.68837583e-01 8.70926321e-01
4.42348987e-01 2.32581854e-01 5.82035482e-01 -5.50395668e-01
-9.43186462e-01 2.66802251e-01 -7.60767639e-01 8.82828385e-02
2.92118579e-01 2.67623901e-01 1.06421506e+00 -7.37508297e-01
5.25534272e-01 1.16676652e+00 8.41113031e-01 6.35893643e-01
-1.33086765e+00 4.21926007e-02 1.94098845e-01 8.03527534e-02
-1.15317440e+00 -5.42791307e-01 8.73986423e-01 -2.60838956e-01
1.26214576e+00 -1.57991108e-02 4.51966494e-01 1.35021174e+00
-1.08906031e-01 8.93917203e-01 9.98399198e-01 -4.89283383e-01
7.97362775e-02 -4.24228489e-01 1.26495034e-01 8.31884921e-01
-2.31029000e-03 7.88220614e-02 -2.90498406e-01 2.00982824e-01
8.95821929e-01 2.52807755e-02 -5.64160287e-01 -1.67941079e-01
-8.87071252e-01 6.05697453e-01 5.46229482e-01 -7.09898248e-02
-5.47745526e-01 6.45409763e-01 3.07157218e-01 4.04028684e-01
7.60711491e-01 1.45387173e-01 -6.73498392e-01 1.95731834e-01
-6.97362781e-01 -4.40447256e-02 8.54385495e-01 6.43859863e-01
1.02256155e+00 3.59216034e-01 2.48754039e-01 8.80200624e-01
2.55354822e-01 2.49283835e-01 2.20631480e-01 -1.24760520e+00
2.25937784e-01 4.16836977e-01 2.53903240e-01 -6.32345080e-01
-4.93006438e-01 -4.12225187e-01 -1.07381046e+00 3.36816519e-01
2.37049922e-01 -2.87196517e-01 -1.00637901e+00 1.61104631e+00
-3.73913012e-02 1.90708920e-01 2.43403211e-01 1.09001029e+00
9.20741439e-01 7.35814571e-01 -5.98112106e-01 -6.89799115e-02
9.41008687e-01 -1.28286302e+00 -4.22363549e-01 -1.59636483e-01
3.33932489e-01 -7.39787400e-01 1.03494763e+00 7.86393762e-01
-1.26956236e+00 -8.06299984e-01 -1.00695729e+00 -5.71483850e-01
-2.01908976e-01 6.13454938e-01 4.93260711e-01 5.90365887e-01
-1.45563173e+00 7.53975511e-01 -9.14409578e-01 -1.75449759e-01
2.80388385e-01 2.62475282e-01 -6.21158220e-02 -2.41130009e-01
-6.93405271e-01 4.97902453e-01 -2.06326246e-01 7.89588153e-01
-1.54843223e+00 -4.71221775e-01 -7.15291619e-01 7.88770895e-03
1.33823454e-01 -9.39201295e-01 1.32467699e+00 -1.37832606e+00
-1.89491677e+00 1.11211193e+00 -1.72121048e-01 -6.49459839e-01
4.33927178e-01 -5.24131417e-01 3.91998887e-02 3.28850567e-01
-4.17940140e-01 5.17933488e-01 1.13283038e+00 -1.52155268e+00
-1.46249816e-01 -2.93863058e-01 3.69026750e-01 1.16181105e-01
1.35542199e-01 -1.12460554e-01 -3.90736431e-01 -3.48631173e-01
8.25068504e-02 -8.41734290e-01 -3.55747670e-01 -9.82089490e-02
-4.07887995e-01 4.45899032e-02 6.00764096e-01 -7.72193551e-01
3.74939173e-01 -1.83866596e+00 3.64286870e-01 -3.39690804e-01
2.63636619e-01 1.30100623e-01 -5.53317547e-01 2.53672957e-01
-3.49797487e-01 -1.75578475e-01 -1.81037754e-01 -6.99124157e-01
-1.66738048e-01 3.59464705e-01 -3.41029614e-01 7.84242809e-01
9.87183675e-02 7.46952951e-01 -7.90664673e-01 2.11291686e-01
2.51446187e-01 8.89777660e-01 -5.42298734e-01 5.73693514e-01
-5.71319461e-01 4.51684356e-01 7.59496763e-02 5.85283995e-01
7.35727370e-01 -3.92740190e-01 2.26951674e-01 -5.86275995e-01
-1.71454206e-01 3.09860826e-01 -6.34362400e-01 1.99301124e+00
-8.39024425e-01 9.75219011e-01 2.30904579e-01 -1.09272635e+00
1.17784703e+00 2.46840671e-01 4.93128806e-01 -6.21919274e-01
3.71949613e-01 -9.99670997e-02 -1.39574707e-01 -5.38690031e-01
4.84202415e-01 1.17547043e-01 5.29475331e-01 8.07962120e-01
4.88431081e-02 1.61729798e-01 -2.73110569e-02 -9.25081521e-02
1.14094496e+00 7.11777985e-01 -1.30007699e-01 -1.44808412e-01
6.63860142e-01 -4.66703415e-01 3.04670006e-01 5.86080074e-01
7.90371895e-02 1.19369376e+00 5.79797387e-01 -1.16144073e+00
-1.14191723e+00 -8.58627856e-01 3.97542864e-01 1.13098693e+00
-9.37756449e-02 -1.78713799e-01 -7.66053677e-01 -1.21416971e-01
-5.36505640e-01 4.57079440e-01 -9.84884560e-01 8.54773447e-02
-5.78800380e-01 -7.80852079e-01 2.79524118e-01 6.39764428e-01
7.17336535e-01 -1.03678024e+00 -1.05505323e+00 -4.50887997e-03
-4.22042429e-01 -1.47560251e+00 -2.19359547e-01 2.98733294e-01
-6.12398982e-01 -1.09471536e+00 -6.08357966e-01 -5.08958101e-01
7.30812848e-01 7.43849754e-01 1.60366976e+00 5.33334613e-02
-4.10193473e-01 6.65891469e-01 -2.77729243e-01 -1.37448326e-01
-1.24371156e-01 1.35833293e-01 -2.64130682e-01 3.52655381e-01
2.24745840e-01 -7.47171462e-01 -8.73193920e-01 4.77829352e-02
-9.62389529e-01 3.17494571e-01 3.64200890e-01 7.18769550e-01
5.77228487e-01 -2.80161034e-02 -2.01191396e-01 -8.58327210e-01
4.73433435e-01 -6.06051274e-02 -7.44189560e-01 2.27767885e-01
-9.60802287e-02 1.04948610e-01 8.32674623e-01 -1.23983482e-02
-1.24196339e+00 3.62381101e-01 -1.65799499e-01 -8.63865793e-01
-1.86345935e-01 1.72955185e-01 -2.77047120e-02 -1.58958673e-01
7.80403614e-01 2.04753146e-01 -1.96510702e-01 -6.32934332e-01
5.09257436e-01 3.73104036e-01 8.03669810e-01 -7.16815829e-01
4.19863999e-01 8.79387736e-01 2.12940909e-02 -8.52467597e-01
-1.36563814e+00 -3.85097325e-01 -9.05065715e-01 -1.58031762e-01
9.97848809e-01 -1.11824441e+00 -1.08780015e+00 6.04277849e-01
-1.52459419e+00 -9.63559389e-01 -2.13908643e-01 1.04068846e-01
-9.08688664e-01 3.68996918e-01 -8.96468163e-01 -8.15109313e-01
-6.25006974e-01 -1.12481558e+00 1.17864883e+00 1.61905348e-01
2.96136439e-01 -1.10334206e+00 4.23352301e-01 3.50756586e-01
3.13717008e-01 4.91669267e-01 4.21734661e-01 2.67695993e-01
-8.34087312e-01 2.06367940e-01 -6.84885085e-01 8.30997944e-01
2.34444931e-01 2.49918103e-01 -1.44719720e+00 -3.46665740e-01
4.55626428e-01 -5.02276480e-01 1.51351762e+00 7.74376810e-01
1.34806144e+00 -3.56148750e-01 2.83275306e-01 1.15325081e+00
1.88328683e+00 -2.62432307e-01 9.52985585e-01 3.01555634e-01
9.33026731e-01 6.06159627e-01 -1.46092072e-01 4.12975550e-01
3.83723080e-01 2.63869524e-01 8.93659413e-01 -3.68241459e-01
-2.73436546e-01 1.50594532e-01 5.04687965e-01 7.51400232e-01
-4.99003172e-01 -4.28731918e-01 -8.35055411e-01 6.24108970e-01
-1.90317619e+00 -7.57104158e-01 -7.74345919e-02 1.96945167e+00
5.24746418e-01 -8.83932859e-02 9.29783005e-03 -9.42899883e-02
4.77550417e-01 5.35911024e-01 -5.78415155e-01 -7.44383276e-01
-2.77009159e-01 1.91082329e-01 4.67330605e-01 6.98807061e-01
-9.71012712e-01 8.03678870e-01 6.68530560e+00 -3.53847863e-03
-1.11654449e+00 1.07429912e-02 1.01130855e+00 -3.78078260e-02
-1.94741249e-01 -2.73899525e-01 -4.99679416e-01 -2.47912295e-02
9.68165338e-01 4.68905956e-01 1.04401982e+00 5.37225127e-01
2.57133603e-01 -7.58023337e-02 -1.15995967e+00 1.17415583e+00
3.37130100e-01 -1.24379671e+00 1.12694744e-02 -2.53908098e-01
9.17241991e-01 6.61455452e-01 3.07880878e-01 -1.96442321e-01
6.57234311e-01 -1.28812742e+00 4.09946680e-01 8.17796409e-01
8.22728455e-01 -5.31614661e-01 4.63393271e-01 -2.15700013e-03
-1.08053768e+00 6.68380633e-02 -8.65975738e-01 -3.27580810e-01
-1.50053754e-01 5.06612122e-01 -3.93256754e-01 5.84364057e-01
7.81017303e-01 1.11698306e+00 -2.69593537e-01 4.23329055e-01
-3.66232991e-01 5.06974339e-01 -1.72679394e-01 5.27326941e-01
4.82380211e-01 -5.73758543e-01 1.42902479e-01 1.31325078e+00
2.45857891e-02 1.56615123e-01 -2.12968737e-02 1.15571797e+00
-3.63599181e-01 -4.64689374e-01 -6.22942269e-01 1.91865042e-01
-4.38139617e-01 1.57219219e+00 -7.13136673e-01 -2.60003567e-01
-6.14387333e-01 1.16464531e+00 4.18961108e-01 8.46470177e-01
-6.24470174e-01 -9.21586715e-03 7.58920312e-01 -3.13863665e-01
6.39361739e-01 -4.27004814e-01 -1.83900326e-01 -1.47872901e+00
-1.31266281e-01 -6.60638690e-01 1.35092393e-01 -1.31989217e+00
-1.28757918e+00 8.60988617e-01 -6.13141835e-01 -1.01841807e+00
-4.72944826e-02 -1.02110910e+00 -5.65924525e-01 8.80171835e-01
-1.92123878e+00 -1.10534751e+00 -8.03791344e-01 7.48718500e-01
7.30386794e-01 7.57844299e-02 7.42698908e-01 5.61893033e-03
-6.57308221e-01 -7.53624067e-02 1.90510109e-01 3.94716710e-01
3.90643895e-01 -1.26694262e+00 8.14536870e-01 9.43753183e-01
1.54110089e-01 4.99589026e-01 7.38754988e-01 -1.01065084e-01
-1.66876161e+00 -1.17917490e+00 2.77981371e-01 -3.75995338e-01
4.19942111e-01 -4.66471136e-01 -7.64851332e-01 8.82091999e-01
6.24789178e-01 3.66750717e-01 4.57556099e-01 4.32218872e-02
-7.86904752e-01 -2.81277984e-01 -8.48537683e-01 4.01030988e-01
1.07548916e+00 -7.67119110e-01 -1.82993099e-01 5.85095763e-01
7.91280925e-01 -5.93940794e-01 -6.24170542e-01 1.74308479e-01
7.39628673e-01 -1.46180522e+00 1.11399972e+00 -4.74296987e-01
1.04141784e+00 -1.29134387e-01 -4.72894400e-01 -1.32385457e+00
-3.13244879e-01 -8.83518636e-01 -1.06751278e-01 5.59402466e-01
9.31767970e-02 -4.47243303e-01 9.15955722e-01 3.78568172e-01
-1.83348149e-01 -7.69426942e-01 -3.70230287e-01 -5.08217037e-01
-1.41994059e-01 -3.85656655e-01 3.30141962e-01 4.03464645e-01
-8.60815108e-01 6.68757319e-01 -7.19557881e-01 2.09252581e-01
7.97101378e-01 5.04428387e-01 7.68771589e-01 -9.13591921e-01
-4.23798770e-01 -2.72331804e-01 6.67460635e-02 -1.31331313e+00
2.03096032e-01 -4.16157693e-01 1.37875661e-01 -1.56266558e+00
2.47499615e-01 1.72352582e-01 -2.55055130e-01 3.82820308e-01
2.89569408e-01 5.56518257e-01 -1.03741148e-02 7.91272707e-03
-7.45775938e-01 6.40029490e-01 1.11579525e+00 -3.38078797e-01
-1.22202039e-01 -1.31302655e-01 -5.27034163e-01 7.45286584e-01
8.94000471e-01 -1.56695530e-01 -4.10819441e-01 -1.22536755e+00
6.76592231e-01 -2.97387876e-03 4.73739356e-01 -6.29636884e-01
6.82396293e-02 2.18672782e-01 3.83446306e-01 -8.25788081e-01
4.34137970e-01 -6.67949080e-01 -1.10045467e-02 3.87694955e-01
-4.45370495e-01 9.56698358e-02 -2.10807957e-02 4.25275981e-01
-3.58942971e-02 -8.85204375e-02 7.49125421e-01 -6.06334507e-01
-3.31527770e-01 3.30256730e-01 -2.71019518e-01 -1.98750183e-01
2.51510531e-01 -6.38307929e-02 -4.79497075e-01 -4.58142310e-01
-6.54596508e-01 -2.10538775e-01 5.68996549e-01 2.46345382e-02
8.46998572e-01 -9.94249344e-01 -7.90078223e-01 2.77456999e-01
-2.64865160e-01 5.94254211e-02 1.96364433e-01 7.42664516e-01
-8.95550311e-01 2.36174375e-01 -2.97171324e-01 -7.10426092e-01
-9.61965919e-01 6.00330591e-01 7.15357125e-01 -2.40065217e-01
-8.03729236e-01 8.78174901e-01 6.59122109e-01 -5.65714777e-01
1.13352455e-01 -6.45241141e-01 3.72162722e-02 -2.58238882e-01
5.17644703e-01 4.63577628e-01 1.27701834e-01 -4.73386765e-01
6.10750020e-02 6.89520240e-01 8.93362686e-02 8.23667720e-02
1.72164011e+00 -3.07193398e-01 -4.39892679e-01 4.63930368e-01
1.51429343e+00 -4.83807892e-01 -1.73190689e+00 -4.01695251e-01
-4.87828106e-01 -3.52719277e-01 1.50110826e-01 -5.59256077e-01
-1.63536882e+00 1.28457427e+00 4.33906436e-01 1.77444071e-01
1.51008320e+00 -5.34208000e-01 7.93562293e-01 7.11110353e-01
-7.47882351e-02 -8.79895449e-01 4.12439287e-01 6.44388378e-01
8.26501191e-01 -1.18508005e+00 3.50650474e-02 -2.28845701e-02
-3.42119962e-01 1.64119506e+00 3.80544245e-01 -6.21817231e-01
4.83722657e-01 3.34756911e-01 2.26343408e-01 -3.74024272e-01
-1.19778132e+00 -4.14886475e-01 6.55053109e-02 5.21263361e-01
6.74449503e-01 -1.68121397e-01 3.61403257e-01 7.11791813e-02
1.22823127e-01 8.87629688e-02 8.33036661e-01 3.30133259e-01
-2.18803644e-01 -7.08090007e-01 -1.45039439e-01 1.91910677e-02
-4.28580970e-01 -1.87243819e-01 -3.90383869e-01 3.65380555e-01
-9.60420538e-03 6.80219233e-01 3.05906124e-02 -2.72000134e-01
-1.51879981e-01 -3.69124502e-01 8.55480731e-01 -4.58323717e-01
-4.79749173e-01 2.61855394e-01 -1.05238222e-02 -7.39530921e-01
-1.01362574e+00 -6.65325880e-01 -7.57435262e-01 -1.98766783e-01
-4.55332510e-02 -4.40671146e-01 5.61206222e-01 7.76310861e-01
-8.47949684e-02 7.82931089e-01 9.40372229e-01 -1.19295597e+00
-2.79273301e-01 -7.64035463e-01 -4.04898524e-01 2.21991807e-01
7.20607758e-01 -6.80303872e-02 -2.49290645e-01 4.40452814e-01] | [9.757035255432129, -2.8825080394744873] |
2fd256e3-7c49-40e2-b197-42273305b2d1 | pace-unified-multi-modal-dialogue-pre | 2305.14839 | null | https://arxiv.org/abs/2305.14839v2 | https://arxiv.org/pdf/2305.14839v2.pdf | PaCE: Unified Multi-modal Dialogue Pre-training with Progressive and Compositional Experts | Perceiving multi-modal information and fulfilling dialogues with humans is a long-term goal of artificial intelligence. Pre-training is commonly regarded as an effective approach for multi-modal dialogue. However, due to the limited availability of multi-modal dialogue data, there is still scarce research on multi-modal dialogue pre-training. Yet another intriguing challenge emerges from the encompassing nature of multi-modal dialogue, which involves various modalities and tasks. Moreover, new forms of tasks may arise at unpredictable points in the future. Hence, it is essential for designed multi-modal dialogue models to possess sufficient flexibility to adapt to such scenarios. This paper proposes \textbf{PaCE}, a unified, structured, compositional multi-modal dialogue pre-training framework. It utilizes a combination of several fundamental experts to accommodate multiple dialogue-related tasks and can be pre-trained using limited dialogue and extensive non-dialogue multi-modal data. Furthermore, we propose a progressive training method where old experts from the past can assist new experts, facilitating the expansion of their capabilities. Experimental results demonstrate that PaCE achieves state-of-the-art results on eight multi-modal dialog benchmarks. | ['Yongbin Li', 'Fei Huang', 'Min Yang', 'Zhichao Yin', 'Binyuan Hui', 'Yunshui Li'] | 2023-05-24 | null | null | null | null | ['visual-dialogue', 'multimodal-intent-recognition', 'response-generation', 'dialogue-state-tracking', 'visual-dialogue'] | ['computer-vision', 'miscellaneous', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [-8.14010948e-02 2.02947676e-01 5.45392260e-02 -5.21500826e-01
-5.82977057e-01 -8.80683482e-01 1.01903105e+00 1.53169539e-02
-4.13668364e-01 9.06797111e-01 5.79007149e-01 -1.96709201e-01
1.02067113e-01 -5.74130714e-01 -3.63636799e-02 -3.46842855e-01
2.94991285e-01 1.08188355e+00 2.36698821e-01 -9.98251081e-01
5.97750843e-02 5.47706559e-02 -1.26544392e+00 7.47913897e-01
9.43881392e-01 6.84608519e-01 2.66827196e-01 7.25736439e-01
-3.26139808e-01 7.89037943e-01 -6.75285101e-01 -7.49255776e-01
-8.79620388e-02 -5.74783921e-01 -1.44084346e+00 2.43713126e-01
-1.67785257e-01 -5.85752010e-01 -1.43480748e-01 6.64080143e-01
6.29905522e-01 5.15290797e-01 6.33463800e-01 -1.07157397e+00
-4.83294845e-01 7.15226531e-01 -6.77224174e-02 -9.31363404e-02
8.99276793e-01 3.82542193e-01 1.10927355e+00 -7.06587195e-01
5.05910695e-01 1.66611481e+00 3.67201179e-01 1.07574368e+00
-1.05414450e+00 -4.55937907e-03 5.43606933e-03 8.21984708e-02
-5.57266831e-01 -5.81225038e-01 8.67010474e-01 -2.42598012e-01
8.15278947e-01 3.47428322e-01 2.51009822e-01 1.35993493e+00
-8.95034820e-02 9.97204721e-01 1.23975289e+00 -4.89523917e-01
-5.09533249e-02 4.84513521e-01 -1.07352205e-01 3.73134702e-01
-8.63779783e-01 -3.02631795e-01 -6.55462980e-01 -2.85701424e-01
4.78303313e-01 -2.79985040e-01 -1.23414837e-01 -1.61725312e-01
-1.41835678e+00 9.61283863e-01 -1.51358228e-02 5.31513214e-01
-2.55492121e-01 -7.78093159e-01 8.99725378e-01 6.42309010e-01
2.59072602e-01 6.13334060e-01 -5.58658838e-01 -4.46146250e-01
-1.78885922e-01 1.98300287e-01 1.25286400e+00 6.97103322e-01
6.57492578e-01 -1.39122769e-01 -2.74778903e-01 1.28365815e+00
1.71988651e-01 2.25772232e-01 5.00964999e-01 -1.01413977e+00
6.17311716e-01 9.75045741e-01 2.09733576e-01 -6.23698115e-01
-5.98257422e-01 3.16283017e-01 -1.20456815e+00 -2.34885231e-01
6.41339421e-01 -4.78850275e-01 -1.78095624e-01 1.76002121e+00
4.87820745e-01 -4.97233123e-01 7.11123705e-01 8.03027630e-01
1.04796839e+00 8.14315557e-01 1.74759939e-01 -4.40435022e-01
1.42881083e+00 -1.04463851e+00 -7.43435502e-01 -3.78241479e-01
4.11060363e-01 -8.26759219e-01 1.32573688e+00 2.75130630e-01
-1.04998827e+00 -5.56765139e-01 -4.75654483e-01 -6.75876141e-02
-4.72262263e-01 -9.73897949e-02 6.65550292e-01 4.84270751e-01
-8.85216236e-01 7.97340870e-02 -1.72643542e-01 -2.91708022e-01
-4.71744567e-01 2.41122499e-01 -3.11129451e-01 -1.07923793e-02
-1.92684221e+00 1.33435535e+00 7.08346248e-01 1.29382566e-01
-7.31654406e-01 -2.92300195e-01 -8.81243765e-01 -2.52573311e-01
5.65565348e-01 -6.73866808e-01 1.50641346e+00 -7.80968010e-01
-2.04335690e+00 7.68791974e-01 4.33225296e-02 -2.68178314e-01
6.54636741e-01 -1.77817214e-02 -4.12188888e-01 1.42510116e-01
-1.07868966e-02 6.01232886e-01 6.78336024e-01 -1.46228516e+00
-7.35990822e-01 -2.77527869e-01 7.27286994e-01 8.80081415e-01
-5.89625597e-01 -3.50948572e-02 -2.60219812e-01 -1.02308027e-01
-3.40695590e-01 -9.58591223e-01 -2.40380943e-01 -6.31668925e-01
-3.79922628e-01 -6.90046608e-01 8.24892461e-01 -3.50460678e-01
1.07109416e+00 -1.83924794e+00 5.15196323e-01 -3.37651938e-01
1.88494027e-01 3.60967368e-01 -9.51052085e-02 9.16985333e-01
4.01093364e-01 -2.06898272e-01 -1.14555165e-01 -6.41735315e-01
2.94378996e-01 4.36381966e-01 -2.36953497e-01 -7.28155226e-02
7.46091828e-02 7.19496369e-01 -9.84454870e-01 -6.63570583e-01
5.11560798e-01 9.44721326e-02 -2.99899727e-01 7.97167897e-01
-6.48128092e-01 1.10997701e+00 -3.85937482e-01 5.29835343e-01
3.20369631e-01 -3.84481043e-01 4.28176880e-01 -1.28683046e-01
-7.99976215e-02 1.14371233e-01 -1.00532734e+00 2.02568555e+00
-8.09270024e-01 2.96778649e-01 2.79489130e-01 -9.91382539e-01
8.26867402e-01 8.16799998e-01 4.62937146e-01 -6.87526345e-01
2.99292147e-01 1.26360744e-01 9.71798301e-02 -6.28515005e-01
1.01289833e+00 -3.53693843e-01 -5.75559080e-01 7.25581527e-01
2.13626817e-01 -2.53552526e-01 5.00562906e-01 2.56177455e-01
6.10790133e-01 -3.10457587e-01 4.12169188e-01 7.13222027e-02
1.12877512e+00 -4.05981243e-02 4.34067458e-01 5.05931020e-01
-3.48129511e-01 7.74382567e-03 4.54421222e-01 -3.83226514e-01
-6.87248349e-01 -6.68343246e-01 9.43479314e-02 1.89931142e+00
2.05962032e-01 -2.19585955e-01 -4.15143788e-01 -7.84122944e-01
-3.22356194e-01 6.92433119e-01 -4.91722912e-01 4.56637181e-02
-6.12378299e-01 -6.83315575e-01 6.46758616e-01 1.31347515e-02
8.61908734e-01 -1.43052614e+00 -6.64451420e-01 3.91274095e-01
-8.07892382e-01 -1.30899453e+00 -3.08637440e-01 4.86571379e-02
-4.48254853e-01 -9.58301187e-01 -8.47489297e-01 -8.43298078e-01
2.21655890e-01 1.12111531e-01 1.40696073e+00 -1.15162402e-01
2.92169452e-01 7.32303858e-01 -5.40347397e-01 -6.58494309e-02
-1.25966454e+00 2.47010574e-01 3.63852233e-02 7.25473315e-02
2.42118314e-01 -3.99944097e-01 -3.71450245e-01 6.65228546e-01
-9.43557024e-01 2.72887558e-01 5.19852340e-01 1.28556693e+00
-5.60350716e-02 -3.10204085e-02 1.07232273e+00 -1.00769496e+00
1.40483427e+00 -5.23788095e-01 3.85693498e-02 7.20220923e-01
-1.16931744e-01 -4.35332768e-02 9.74998295e-01 -6.75365567e-01
-1.56428766e+00 -7.23350048e-02 -2.98430741e-01 1.96621373e-01
-5.05968869e-01 5.81593633e-01 -3.34550828e-01 1.63127109e-01
6.22479618e-01 3.47130954e-01 1.73256934e-01 -1.95838183e-01
7.21622825e-01 9.29500818e-01 5.52515984e-01 -1.12450969e+00
4.47156668e-01 1.16972171e-01 -3.36042345e-01 -1.06344724e+00
-8.11533511e-01 -7.40807712e-01 -8.62140954e-01 -5.37588835e-01
7.68440902e-01 -7.37315714e-01 -1.09757197e+00 6.17793381e-01
-1.46130335e+00 -4.58348066e-01 1.31927639e-01 -3.92725617e-02
-5.32769620e-01 6.85053647e-01 -7.81619430e-01 -1.03317344e+00
-5.58118343e-01 -1.13729978e+00 7.70828366e-01 3.91882598e-01
-4.24419373e-01 -1.58913064e+00 1.90642849e-01 7.01792240e-01
4.51306999e-01 -6.12922618e-03 9.96026397e-01 -1.07383788e+00
9.50376410e-03 1.02182396e-01 4.33290564e-02 2.58640498e-01
3.66322011e-01 -3.85154247e-01 -8.80263090e-01 -2.53669679e-01
2.09394053e-01 -1.31830359e+00 2.12660611e-01 -2.25957856e-01
5.15411496e-01 -4.75566268e-01 1.57262936e-01 -3.57678026e-01
8.19861770e-01 1.26665741e-01 2.80133486e-01 2.26340100e-01
3.39252084e-01 1.03658652e+00 9.80233431e-01 8.09284985e-01
1.14658153e+00 5.55370033e-01 3.21618915e-01 1.07004568e-02
4.71108228e-01 3.23994346e-02 2.74278909e-01 1.09176385e+00
1.86736703e-01 -2.27466688e-01 -9.99171495e-01 5.13567686e-01
-1.99792325e+00 -1.11488032e+00 1.60963669e-01 1.76380014e+00
1.44567168e+00 3.58089916e-02 4.57583159e-01 -1.03335373e-01
5.72821081e-01 3.72214705e-01 -7.98393369e-01 -3.39598000e-01
-2.31182277e-01 -5.11363506e-01 -4.34739530e-01 4.81101304e-01
-1.20666683e+00 9.60765839e-01 5.45427465e+00 5.62061310e-01
-7.69567072e-01 7.41817430e-02 6.23096228e-01 7.21712470e-01
-3.13748300e-01 -1.17798366e-01 -7.78232932e-01 3.40894014e-01
9.44019079e-01 -3.17953199e-01 4.21994269e-01 6.06557369e-01
-1.47833914e-01 -3.25557500e-01 -1.22469246e+00 8.39109778e-01
1.03832781e-01 -9.45524156e-01 1.63039669e-01 -2.73158550e-01
5.21166742e-01 -4.41056460e-01 1.46002900e-02 7.38918662e-01
5.31283200e-01 -6.99384689e-01 2.11137667e-01 3.08473825e-01
5.98139524e-01 -5.16502440e-01 7.01164305e-01 1.12481999e+00
-1.14528692e+00 -3.86771299e-02 -1.61216572e-01 8.22994560e-02
4.66700464e-01 -1.22580856e-01 -1.15858483e+00 7.34885693e-01
2.37851620e-01 3.41795534e-01 -4.42698568e-01 3.29108685e-01
1.60209045e-01 -6.49603456e-02 -1.18011765e-01 -2.51169890e-01
2.98376322e-01 -1.50922835e-01 4.71071661e-01 1.14246798e+00
-2.15202644e-01 2.63884902e-01 6.73307776e-01 3.33543241e-01
4.10483927e-02 1.38665468e-01 -6.79568291e-01 -5.37381917e-02
5.29618919e-01 1.41413260e+00 -2.09773466e-01 -2.72634298e-01
-6.48175716e-01 1.02127206e+00 2.54709095e-01 1.79947652e-02
-5.64684808e-01 3.09568197e-02 2.72860467e-01 -4.77594346e-01
-5.28648674e-01 -2.30473608e-01 -2.30598655e-02 -1.17286742e+00
-3.18047136e-01 -1.47880852e+00 8.22121084e-01 -4.32756752e-01
-1.66978395e+00 8.77604485e-01 3.77007038e-03 -1.22242618e+00
-9.76499856e-01 -3.92836660e-01 -4.74212170e-01 7.87989616e-01
-1.42604530e+00 -1.52582467e+00 -1.63266897e-01 9.60976481e-01
1.12667513e+00 -4.15595144e-01 1.44653964e+00 1.81277752e-01
-2.55880654e-01 4.39712405e-01 -1.60959870e-01 1.38107136e-01
1.10159004e+00 -1.44405925e+00 -6.94374070e-02 3.54097843e-01
-2.62995988e-01 6.42340064e-01 8.09316754e-01 -3.79473060e-01
-1.52385080e+00 -5.74536741e-01 7.97493339e-01 -4.24032450e-01
9.00962412e-01 -4.35202420e-02 -1.17884576e+00 4.76407260e-01
8.39902699e-01 -7.35127628e-01 1.07196701e+00 4.22943503e-01
1.12292394e-02 2.29346693e-01 -1.01675296e+00 7.77834892e-01
4.76579010e-01 -7.74599195e-01 -1.00218260e+00 4.94966567e-01
5.24506450e-01 -8.19038272e-01 -1.13258672e+00 3.85311931e-01
3.12208831e-01 -1.06899059e+00 9.42224920e-01 -6.80450439e-01
4.31420445e-01 2.48335853e-01 -1.27338603e-01 -1.61574471e+00
2.54319578e-01 -9.27326620e-01 -1.80540651e-01 1.51800561e+00
3.63779247e-01 -5.06101012e-01 5.12503564e-01 1.05264676e+00
-1.24370106e-01 -4.83507067e-01 -8.50199699e-01 -3.32393497e-01
2.87414283e-01 1.11428902e-01 2.95970380e-01 1.05948150e+00
5.40346444e-01 1.08039951e+00 -9.24949825e-01 -1.26772583e-01
2.80635685e-01 3.62732738e-01 1.08472753e+00 -1.37427223e+00
-2.86229700e-01 -3.70588273e-01 4.62236375e-01 -1.58765781e+00
6.01134241e-01 -4.00186419e-01 2.07023293e-01 -1.51232600e+00
-1.46984726e-01 -4.34753567e-01 1.67618677e-01 3.23710084e-01
-3.23206812e-01 -3.56627703e-01 1.45450041e-01 3.05980414e-01
-1.13332498e+00 1.00822711e+00 1.59160733e+00 -1.92195907e-01
-5.02423525e-01 4.75397438e-01 -4.85832125e-01 7.48035371e-01
8.62932324e-01 1.38533801e-01 -6.52482629e-01 -2.83256948e-01
7.09033981e-02 8.83088171e-01 -2.26524640e-02 -6.46349847e-01
6.48634136e-01 -4.53424484e-01 -1.71817839e-01 -7.08604932e-01
9.01599407e-01 -7.76184201e-01 -2.39063770e-01 8.74137878e-02
-5.29266238e-01 2.58390933e-01 1.64129734e-01 5.79948545e-01
-4.83055979e-01 -2.55617112e-01 7.03718960e-01 -5.81652284e-01
-1.03722239e+00 1.19150588e-02 -5.86350977e-01 4.13970292e-01
9.10478711e-01 1.86919853e-01 -5.10406435e-01 -6.47084236e-01
-7.35892057e-01 8.35124850e-01 1.68340042e-01 6.06163859e-01
5.82058907e-01 -1.11588788e+00 -6.80123568e-01 -2.36433402e-01
1.63254306e-01 3.93291377e-02 6.93970859e-01 5.12120605e-01
5.60661890e-02 5.61848879e-01 -4.93108034e-01 -5.55872023e-01
-1.42790473e+00 2.76090324e-01 2.73685992e-01 -7.34598756e-01
-2.67155915e-01 5.23718059e-01 -5.70620522e-02 -1.00090969e+00
3.82670492e-01 3.76369804e-01 -6.73804879e-01 4.83823657e-01
5.13160586e-01 9.62918997e-02 -3.15006167e-01 -8.82249296e-01
-4.97361086e-02 -4.48680222e-02 -1.84887722e-01 -4.36508656e-01
1.11643219e+00 -6.60795033e-01 -3.44505876e-01 1.08434772e+00
6.59453630e-01 -3.34466159e-01 -9.61539567e-01 -7.71921158e-01
7.99918547e-02 -1.08079635e-01 -7.97473788e-01 -9.92122591e-01
-3.25507969e-01 8.76542509e-01 1.37780309e-02 8.16622198e-01
9.70211923e-01 -3.56768295e-02 8.59957039e-01 8.17326069e-01
5.15501440e-01 -1.31681085e+00 7.25823700e-01 1.18538940e+00
1.08259082e+00 -1.73680139e+00 -3.50142121e-01 -9.60778221e-02
-1.37248266e+00 1.19606566e+00 1.13171852e+00 7.16015875e-01
4.60133523e-01 -2.55647629e-01 4.78253692e-01 -1.13637477e-01
-9.59605634e-01 -1.86864689e-01 2.24873066e-01 4.76728886e-01
5.45171380e-01 -5.05906567e-02 -1.11309037e-01 5.76688588e-01
-1.58264581e-02 -5.31024814e-01 4.40874010e-01 8.15347016e-01
-3.31274271e-01 -1.47085297e+00 -2.47201577e-01 1.35602534e-01
-2.04702213e-01 1.28400460e-01 -6.80739522e-01 6.50566876e-01
-3.49831641e-01 1.38502538e+00 -4.59305406e-01 -3.48316073e-01
4.37542439e-01 3.36954623e-01 7.06128627e-02 -6.72850907e-01
-1.00495219e+00 -1.36123627e-01 5.41738808e-01 4.90803830e-02
-7.93406248e-01 -4.28107351e-01 -1.01101351e+00 -4.48749632e-01
-2.47943655e-01 4.44823831e-01 2.65908033e-01 1.16365707e+00
1.13867104e-01 4.23456252e-01 8.31780314e-01 -8.39246452e-01
-1.01418793e+00 -1.30323231e+00 -2.42702648e-01 5.78042150e-01
2.12077796e-01 -6.02779269e-01 -1.73610583e-01 -5.00849783e-02] | [12.827966690063477, 8.009050369262695] |
6950b4e7-e231-48e6-86ae-442457c7f80c | a-multi-task-convolutional-neural-network-for-2 | 2106.09303 | null | https://arxiv.org/abs/2106.09303v3 | https://arxiv.org/pdf/2106.09303v3.pdf | A Multi-task convolutional neural network for blind stereoscopic image quality assessment using naturalness analysis | This paper addresses the problem of blind stereoscopic image quality assessment (NR-SIQA) using a new multi-task deep learning based-method. In the field of stereoscopic vision, the information is fairly distributed between the left and right views as well as the binocular phenomenon. In this work, we propose to integrate these characteristics to estimate the quality of stereoscopic images without reference through a convolutional neural network. Our method is based on two main tasks: the first task predicts naturalness analysis based features adapted to stereo images, while the second task predicts the quality of such images. The former, so-called auxiliary task, aims to find more robust and relevant features to improve the quality prediction. To do this, we compute naturalness-based features using a Natural Scene Statistics (NSS) model in the complex wavelet domain. It allows to capture the statistical dependency between pairs of the stereoscopic images. Experiments are conducted on the well known LIVE PHASE I and LIVE PHASE II databases. The results obtained show the relevance of our method when comparing with those of the state-of-the-art. Our code is available online on https://github.com/Bourbia-Salima/multitask-cnn-nrsiqa_2021. | ['Mohammed El Hassouni', 'Aladine Chetouani', 'Ayoub Karine', 'Salima Bourbia'] | 2021-06-17 | null | null | null | null | ['stereoscopic-image-quality-assessment'] | ['computer-vision'] | [-6.96645156e-02 -5.77580810e-01 3.85430336e-01 -2.12902918e-01
-1.05460513e+00 -1.18168160e-01 3.95285726e-01 -2.30997160e-01
-4.74478573e-01 7.16688275e-01 5.13553202e-01 1.11348346e-01
-3.63509893e-01 -6.60199583e-01 -6.34539783e-01 -8.99280787e-01
5.40589020e-02 -6.90307394e-02 2.66758978e-01 -4.45599586e-01
5.95899820e-01 3.39603841e-01 -2.15150189e+00 5.77415466e-01
9.46088850e-01 1.12608397e+00 5.04863918e-01 7.39305913e-01
3.04177612e-01 8.60254765e-01 -1.68531477e-01 -1.81210309e-01
5.55797756e-01 -4.04296100e-01 -8.70563805e-01 -1.99519724e-01
5.92790782e-01 -4.42505181e-01 -1.67037025e-01 1.38705635e+00
9.75693822e-01 7.14072660e-02 6.09996021e-01 -8.50339770e-01
-4.44563657e-01 -1.59424916e-01 -3.17414612e-01 5.59652448e-01
6.56708121e-01 4.70445544e-01 8.92055511e-01 -7.84491777e-01
4.78671849e-01 1.05356312e+00 4.36592817e-01 5.32348752e-01
-8.11704576e-01 -4.06272680e-01 -4.92746562e-01 9.75210190e-01
-1.20986545e+00 -6.39891148e-01 7.85523474e-01 -6.80930078e-01
4.43600774e-01 1.92594409e-01 6.57069147e-01 9.88956869e-01
2.58368015e-01 6.13598287e-01 1.85850132e+00 -4.68766391e-01
4.92548868e-02 7.57119507e-02 -7.97506701e-03 6.62625849e-01
-1.34503826e-01 5.61726749e-01 -5.28269827e-01 1.94331288e-01
7.17163384e-01 -7.92276263e-02 -8.73463929e-01 -2.22363144e-01
-1.07609046e+00 6.72856629e-01 5.53182185e-01 7.11790800e-01
-4.43903595e-01 -3.31525773e-01 2.78365672e-01 5.24070680e-01
6.14490211e-01 2.93890566e-01 -4.76911664e-01 -1.09296225e-01
-8.56001318e-01 1.35437638e-01 5.70178568e-01 4.07070458e-01
9.02987540e-01 -3.21244508e-01 -3.05643648e-01 9.08951640e-01
2.89923280e-01 3.85003716e-01 7.34058559e-01 -9.23827350e-01
2.60254711e-01 2.09955886e-01 1.71943828e-01 -8.19800794e-01
-3.15240890e-01 -2.69382298e-01 -8.21800470e-01 8.44488859e-01
4.87814963e-01 -9.79968905e-02 -5.92646241e-01 1.40497983e+00
1.14825696e-01 9.86643601e-03 -1.06711537e-01 1.32259762e+00
1.07009506e+00 4.26282018e-01 -3.07751060e-01 -1.92940459e-01
1.16664636e+00 -8.74585330e-01 -6.54569685e-01 2.40178898e-01
4.09834385e-02 -1.03009474e+00 1.23047578e+00 7.26986110e-01
-1.18733406e+00 -9.49067235e-01 -9.55986857e-01 -1.59206390e-01
-3.89779180e-01 2.77550280e-01 1.81907222e-01 5.00741839e-01
-1.53963280e+00 6.88360214e-01 -4.15230542e-01 -2.24214062e-01
1.65037945e-01 2.25345999e-01 -4.90081638e-01 -2.22967088e-01
-1.12980998e+00 1.03237283e+00 1.05194971e-01 -3.31031717e-02
-7.61463821e-01 -4.88857776e-01 -6.93631828e-01 4.61072698e-02
-2.20037755e-02 -9.86933529e-01 1.01518977e+00 -1.26943851e+00
-1.69213510e+00 9.81575787e-01 -4.52341884e-01 -1.66032791e-01
5.87269604e-01 -2.58665353e-01 -4.13249850e-01 3.40915859e-01
1.54975712e-01 2.40917355e-01 1.08940899e+00 -1.45560777e+00
-6.94424510e-01 -6.03251517e-01 1.80347040e-01 4.37443197e-01
-2.41053030e-01 1.36784717e-01 -4.65226114e-01 -4.52239573e-01
-1.92391872e-01 -4.03568983e-01 1.18045032e-01 6.40947074e-02
-2.15991065e-01 -1.66767649e-02 4.50468928e-01 -9.92712736e-01
7.86155105e-01 -2.02250624e+00 3.31484407e-01 -2.00702339e-01
3.39229405e-01 2.86480427e-01 -1.14513136e-01 3.14784199e-01
-1.90021977e-01 -4.05819684e-01 -1.20705649e-01 -3.23983580e-01
-4.36843932e-01 -4.19105113e-01 1.23113073e-01 6.41877890e-01
4.67646793e-02 4.46210176e-01 -6.48120105e-01 -6.05029523e-01
6.58826053e-01 4.81837720e-01 -3.66396695e-01 5.85452914e-01
2.12313458e-01 9.89869833e-01 -2.12611541e-01 5.28773546e-01
9.64012682e-01 -2.35022791e-02 -4.60913420e-01 -5.78446209e-01
-4.22645688e-01 4.60845716e-02 -1.07320201e+00 1.83592939e+00
-7.98644781e-01 6.55084729e-01 5.18083721e-02 -9.90012169e-01
7.32362807e-01 3.66670638e-01 4.87620473e-01 -1.21642315e+00
1.92923531e-01 4.22456533e-01 -3.22555602e-02 -1.05765915e+00
2.07203642e-01 -1.38541937e-01 6.17386401e-01 1.07571617e-01
3.72723371e-01 -3.35839003e-01 1.02718674e-01 -4.16130871e-01
7.51430273e-01 1.30172014e-01 4.85464811e-01 -3.88260841e-01
1.22437930e+00 -4.70550656e-01 3.04815501e-01 3.17854822e-01
-3.95393848e-01 9.19451177e-01 1.03889763e-01 -6.63561285e-01
-1.15937734e+00 -1.02940810e+00 -3.25484693e-01 6.20633066e-01
4.01991546e-01 5.06485142e-02 -6.71898305e-01 -3.59254569e-01
-2.75712967e-01 2.18138352e-01 -6.16452038e-01 1.41080558e-01
-2.85298675e-01 -5.72939456e-01 -9.68115702e-02 -2.02690568e-02
7.53205895e-01 -1.06255913e+00 -5.50271988e-01 -2.29778364e-01
-4.80646968e-01 -9.00706112e-01 -1.87358394e-01 -2.31861874e-01
-6.64259672e-01 -1.46215618e+00 -1.25911582e+00 -7.11351693e-01
2.67241329e-01 2.92245775e-01 1.09412396e+00 -1.29484817e-01
-2.84411281e-01 4.09314305e-01 -4.46722984e-01 -1.51921809e-01
-2.79776424e-01 -4.83582646e-01 -4.08719666e-02 4.24045503e-01
1.97824091e-01 -7.26034582e-01 -9.68217492e-01 2.04285577e-01
-9.27632689e-01 1.15117513e-01 7.35402405e-01 8.74988258e-01
5.26023388e-01 5.65151311e-03 1.24802709e-01 -2.66157150e-01
5.92283189e-01 -2.24123687e-01 -6.25034034e-01 2.48981997e-01
-2.75156617e-01 8.14226568e-02 7.58809268e-01 -1.32287368e-01
-1.06255221e+00 -1.59190431e-01 -4.02736425e-01 -4.30079788e-01
-4.34458345e-01 1.67821243e-01 -2.29332492e-01 -4.74902242e-01
7.11631536e-01 4.28366661e-01 -4.06292686e-03 -5.10205925e-01
-3.10583599e-02 8.92494202e-01 5.66537619e-01 -2.15193614e-01
6.18796587e-01 6.03184998e-01 1.07247643e-01 -9.11924779e-01
-6.46570265e-01 -8.98613334e-01 -5.62973440e-01 -5.68460941e-01
9.55024183e-01 -1.08302796e+00 -9.85136867e-01 8.58711600e-01
-1.24435258e+00 -3.56312841e-01 -2.21417651e-01 6.54162645e-01
-9.81839955e-01 7.11427808e-01 -5.55404603e-01 -7.72571743e-01
-4.47895348e-01 -1.47980821e+00 1.27875912e+00 2.52855361e-01
4.47079480e-01 -9.75821853e-01 4.24907267e-01 6.55336142e-01
4.90731239e-01 7.68965557e-02 6.69752359e-01 -1.97245166e-01
-5.98179877e-01 2.03598708e-01 -4.64925647e-01 7.50465751e-01
2.37322211e-01 -3.89434516e-01 -1.30400240e+00 -3.00275832e-01
6.14736140e-01 -4.96497154e-01 9.72593188e-01 7.73704946e-01
1.25579202e+00 -1.21921137e-01 4.23033088e-01 9.79810715e-01
1.92299688e+00 -4.69762981e-02 8.66799176e-01 5.34401536e-01
3.13307554e-01 6.96950912e-01 6.34219348e-01 4.71219271e-01
2.57253557e-01 9.42980707e-01 6.05397046e-01 -1.11762546e-01
-4.92860764e-01 1.27204180e-01 2.66258538e-01 8.01123977e-01
-5.12981355e-01 -8.07531476e-02 -8.89792085e-01 5.84130049e-01
-1.55613053e+00 -9.70396340e-01 -1.84179038e-01 2.29287696e+00
6.72469735e-01 -2.23774508e-01 3.23028862e-02 4.31252033e-01
7.50766814e-01 1.64381146e-01 -2.65644968e-01 -6.56582937e-02
-3.72637659e-01 3.69029135e-01 3.18567544e-01 6.67515576e-01
-1.21181452e+00 5.57441592e-01 4.83417940e+00 9.50623870e-01
-1.20655954e+00 2.60780871e-01 5.56800663e-01 9.32637304e-02
-9.89257246e-02 -3.36044103e-01 -2.53378212e-01 6.02164447e-01
6.14434540e-01 -9.93537158e-02 6.17036045e-01 4.17610258e-01
5.01045763e-01 -3.69657785e-01 -7.90297270e-01 1.40442932e+00
4.25508171e-01 -1.04020643e+00 1.44195065e-01 -1.11320160e-01
7.62476385e-01 2.89343506e-01 3.05575550e-01 -2.50533283e-01
-1.02371268e-01 -9.70256388e-01 6.88924253e-01 1.12152004e+00
9.71139848e-01 -5.37511587e-01 1.02163255e+00 2.32372612e-01
-1.02333093e+00 -2.87223995e-01 -3.49377066e-01 6.06639720e-02
6.00379035e-02 6.55785382e-01 -1.12568647e-01 9.50487792e-01
1.18766761e+00 9.35957313e-01 -7.38046646e-01 1.55769348e+00
-9.77868587e-02 5.91535643e-02 1.86620593e-01 1.64594933e-01
-1.41494736e-01 -2.64943153e-01 7.76031494e-01 8.43823314e-01
5.65235078e-01 -7.21642300e-02 -4.11569923e-01 7.97560692e-01
2.48561338e-01 2.66696453e-01 -7.95924962e-01 6.01080775e-01
-2.87840396e-01 1.12448692e+00 -9.12197009e-02 -2.13427037e-01
-5.34517467e-01 1.13521469e+00 8.70471448e-03 2.75887221e-01
-3.18826377e-01 -3.28397691e-01 5.23102283e-01 -8.82952586e-02
1.13652334e-01 -3.96052524e-02 -3.37149799e-01 -1.51836169e+00
2.38962665e-01 -1.04476297e+00 9.71006379e-02 -1.19706833e+00
-1.39222002e+00 9.15688515e-01 -3.71287435e-01 -1.74510646e+00
-1.33540154e-01 -8.66870940e-01 -6.86424851e-01 1.31615376e+00
-2.03299022e+00 -1.11264324e+00 -7.04158604e-01 1.37507355e+00
7.35729098e-01 -3.67895752e-01 7.24894583e-01 3.59184623e-01
-3.89027670e-02 2.04123482e-01 3.25224191e-01 -7.82058835e-02
9.66931403e-01 -1.31054294e+00 1.62038933e-02 9.72828567e-01
-2.10601509e-01 3.79408188e-02 8.21976423e-01 -2.57284135e-01
-9.46411908e-01 -7.01628923e-01 7.39637852e-01 -1.92308873e-01
4.83411819e-01 1.56354889e-01 -6.58730149e-01 -2.13849470e-02
3.55598122e-01 7.73949102e-02 3.98614556e-01 -2.37491190e-01
-2.65060365e-01 -2.25683272e-01 -1.08793330e+00 3.12453806e-01
7.18160868e-01 -8.81480038e-01 -4.96961862e-01 4.55743581e-01
3.92232120e-01 -1.25215039e-01 -8.19812059e-01 4.48669612e-01
5.56528270e-01 -1.84304690e+00 9.92663801e-01 -3.28260027e-02
7.32273400e-01 -1.58144295e-01 -3.46414804e-01 -1.54950917e+00
-1.41040489e-01 -3.43398362e-01 2.12932274e-01 6.34851456e-01
9.69215631e-02 -5.12203872e-01 3.75865608e-01 -2.12759096e-02
-8.58007595e-02 -3.53618056e-01 -1.05121875e+00 -7.77169585e-01
-1.33329362e-01 -3.22438449e-01 3.43399465e-01 7.36881852e-01
-2.52286941e-01 4.06324640e-02 -3.62535596e-01 2.33243823e-01
7.69908011e-01 1.63862258e-01 6.00258529e-01 -1.28891480e+00
-4.32880640e-01 -5.46319127e-01 -6.20986462e-01 -5.67438304e-01
9.03289486e-03 -4.45136458e-01 -9.28276405e-02 -1.55288482e+00
2.72836417e-01 8.98367539e-02 -2.24084899e-01 -6.87659904e-02
-1.78669572e-01 2.88799703e-01 2.09582075e-01 3.67691517e-01
-5.79488337e-01 7.91070819e-01 1.50816333e+00 -1.16851509e-01
-1.77456588e-01 3.46978247e-01 -2.60623783e-01 6.23798668e-01
8.62884998e-01 1.04603343e-01 -1.35408074e-01 -5.42272329e-01
7.16259927e-02 4.24846530e-01 7.23198652e-01 -1.50834525e+00
2.62154996e-01 1.73743263e-01 2.11826414e-01 -2.94112444e-01
5.67390740e-01 -8.07511508e-01 -1.55095801e-01 5.34664690e-01
-8.93193707e-02 -5.18830828e-02 1.04786521e-02 3.00420493e-01
-7.44021952e-01 -1.62042201e-01 1.16948831e+00 -2.90998906e-01
-7.93003559e-01 3.57343227e-01 2.02412270e-02 -1.50914058e-01
6.77962363e-01 -9.81519520e-02 -4.34618175e-01 -4.82084721e-01
-5.97814023e-01 -2.32023850e-01 4.81838465e-01 1.92929268e-01
9.36514378e-01 -1.07220387e+00 -8.74601543e-01 2.97361463e-01
3.90001476e-01 -4.41200614e-01 6.86786950e-01 9.30483639e-01
-7.12640822e-01 3.55740249e-01 -9.34677362e-01 -4.69180822e-01
-1.29784465e+00 6.05041742e-01 6.74956739e-01 -2.31579766e-01
-3.80481899e-01 5.76435387e-01 2.88479328e-01 -2.17366114e-01
9.13953260e-02 -7.55044222e-02 -8.85505319e-01 -1.44157708e-01
6.12162650e-01 6.05442166e-01 2.46763051e-01 -8.30414414e-01
-1.44964941e-02 9.42872047e-01 3.61339509e-01 -3.24381948e-01
1.46668065e+00 -3.76892030e-01 -3.62650841e-01 4.25235212e-01
1.47031224e+00 -1.95294879e-02 -1.07309639e+00 -1.91669926e-01
-2.53659904e-01 -7.13531375e-01 2.96251506e-01 -7.82135487e-01
-1.14420259e+00 1.26897562e+00 1.29770136e+00 1.70475975e-01
1.67354476e+00 -2.32501402e-01 5.88348150e-01 6.55015791e-03
5.21077812e-01 -8.17830086e-01 1.02684356e-01 2.38476709e-01
1.27062953e+00 -1.72112048e+00 -2.88003296e-01 -1.12999380e-01
-5.11370420e-01 1.26563227e+00 5.01867294e-01 -1.03289269e-01
7.86819875e-01 -2.72959977e-01 1.41658232e-01 -1.73737779e-01
-3.93331975e-01 -7.57321537e-01 6.14600956e-01 8.74816418e-01
3.08193386e-01 -9.71693769e-02 -3.49650919e-01 2.53069460e-01
-1.82025209e-01 2.47185722e-01 4.91604388e-01 3.52477372e-01
-3.41945112e-01 -8.96677792e-01 -4.62212205e-01 2.78651804e-01
-5.86716294e-01 -3.32488805e-01 -4.23483774e-02 4.01643127e-01
3.79694790e-01 1.25405502e+00 -2.97117889e-01 -5.11646867e-01
2.52751172e-01 -2.73805618e-01 4.68467772e-01 -2.94783086e-01
-4.15100247e-01 -5.68840317e-02 -2.00945377e-01 -9.46399570e-01
-9.64282453e-01 -6.03717446e-01 -2.72582054e-01 -1.74880281e-01
1.35346338e-01 -6.53602108e-02 7.52108872e-01 7.80258238e-01
-8.73951837e-02 3.66117507e-01 9.54369783e-01 -1.17826402e+00
-2.94485480e-01 -1.12182450e+00 -8.35719049e-01 5.49663961e-01
9.59727943e-01 -6.12538040e-01 -6.03160858e-01 1.21430522e-02] | [11.807125091552734, -1.9383623600006104] |
f1418883-ecbf-49e3-8723-39eb9670c757 | co-variance-tackling-noisy-labels-with-sample | null | null | https://openreview.net/forum?id=4ApXq4y81kY | https://openreview.net/pdf?id=4ApXq4y81kY | Co-variance: Tackling Noisy Labels with Sample Selection by Emphasizing High-variance Examples | The sample selection approach is popular in learning with noisy labels, which tends to select potentially clean data out of noisy data for robust training. The state-of-the-art methods train two deep networks simultaneously for sample selection, which aims to employ their different learning abilities. To prevent two networks from converging to a consensus, their divergence should be maintained during training. Typically, the divergence is kept by first locating the disagreement data on which the prediction labels of two networks are different, and then selecting clean data out of such data. However, this procedure is sample-inefficient for network weight updates, which means that a few clean examples can be utilized in training. In this paper, to address the issues, we propose a simple yet effective method called Co-variance. In particular, we select possibly clean data that simultaneously have high-variance prediction probabilities between two networks. As selected data have high variances, the divergence of two networks can be maintained by training on such data. Additionally, the condition of high variances is milder than the condition of disagreement in sample selection, which allows more data to be considered for training, and makes our method more sample-efficient. Moreover, we show that the proposed method enables to mine enough hard clean examples to help generalization. A series of empirical results show that Co-variance is superior to multiple baselines in the robustness of trained models, especially on class-imbalanced and real-world noisy datasets. | ['Tongliang Liu', 'Chen Gong', 'Mingming Gong', 'Jun Yu', 'Yibing Zhan', 'Bo Han', 'Xiaobo Xia'] | 2021-09-29 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [-4.47866693e-02 5.11681028e-02 -2.33517230e-01 -6.75489843e-01
-8.56438100e-01 -2.68522382e-01 2.31275260e-01 7.22898394e-02
-4.73297954e-01 9.36337054e-01 -2.87284076e-01 -7.02597527e-03
-3.37683886e-01 -6.77356184e-01 -6.50826395e-01 -1.16101897e+00
2.39188448e-02 3.17082733e-01 1.32881954e-01 2.61032060e-02
2.87051797e-02 2.13706315e-01 -1.52062988e+00 1.28764480e-01
1.12712169e+00 1.15202427e+00 1.95807859e-01 3.59861068e-02
-1.78240407e-02 4.64320600e-01 -8.42693448e-01 -4.35451150e-01
2.91343391e-01 -5.05059123e-01 -6.37774646e-01 -4.81072403e-02
4.40102994e-01 -1.07603401e-01 1.12400226e-01 1.47405100e+00
6.73292518e-01 3.21655542e-01 3.62580985e-01 -1.15835238e+00
-3.65029454e-01 1.01485860e+00 -4.19228166e-01 1.37987718e-01
-2.56069034e-01 2.67113775e-01 9.54880595e-01 -7.17876792e-01
2.75325537e-01 1.19377899e+00 7.78511763e-01 8.30757380e-01
-1.23386168e+00 -1.04599357e+00 4.51007009e-01 8.19136128e-02
-1.39803445e+00 -5.40429771e-01 9.15800333e-01 -3.06653045e-02
2.33701631e-01 3.93432230e-01 4.21144962e-01 1.31691039e+00
8.29303712e-02 8.32398176e-01 1.12571430e+00 -2.42891341e-01
5.57379186e-01 3.89047176e-01 3.82744998e-01 2.15312555e-01
5.08086920e-01 -2.40835678e-02 -5.40003479e-01 -1.33187562e-01
3.08705628e-01 2.96845436e-02 -4.77998525e-01 5.41173704e-02
-8.62232208e-01 6.47895932e-01 4.83200699e-01 2.62397289e-01
-2.39766598e-01 -2.24897176e-01 4.70277250e-01 5.33143878e-01
6.91845655e-01 4.76058006e-01 -5.02870262e-01 8.17556754e-02
-8.74233305e-01 3.78790088e-02 5.63368261e-01 7.73591697e-01
7.83982515e-01 1.61366537e-02 -1.75911486e-01 1.05032349e+00
2.70011842e-01 3.06556702e-01 7.25818396e-01 -7.12779462e-01
7.58200407e-01 5.33564866e-01 8.04087427e-03 -1.02029955e+00
-2.61197746e-01 -6.37650907e-01 -1.47547710e+00 3.18933725e-02
4.91319776e-01 -4.17824835e-01 -9.73909795e-01 1.93100166e+00
2.85661489e-01 3.14664274e-01 1.40667990e-01 1.00471425e+00
7.09553063e-01 4.79325831e-01 -1.48733482e-01 -5.83869219e-01
9.41698730e-01 -7.26750374e-01 -7.79427290e-01 -4.02490884e-01
5.26129425e-01 -4.04359519e-01 1.00981927e+00 6.25575840e-01
-7.92308092e-01 -6.70351624e-01 -1.07623208e+00 5.61541796e-01
-6.17178492e-02 1.01812787e-01 2.47925863e-01 5.27534246e-01
-5.05807161e-01 1.01171505e+00 -7.47231960e-01 1.19465021e-02
6.17762387e-01 3.29302311e-01 -3.36316705e-01 -8.69520605e-02
-1.26982284e+00 6.85697258e-01 6.75401568e-01 3.27385694e-01
-6.96252525e-01 -4.22766477e-01 -5.48637271e-01 7.02069625e-02
5.06263554e-01 -3.85555029e-01 1.07912517e+00 -1.26808953e+00
-1.37290251e+00 4.52357650e-01 -4.42832708e-02 -4.63536263e-01
8.06277514e-01 -3.55922937e-01 -4.15496439e-01 -2.52273440e-01
3.02863754e-02 3.59249264e-01 9.59258795e-01 -1.56662714e+00
-6.96835577e-01 -3.49611074e-01 -2.37996846e-01 1.28057823e-01
-5.91261148e-01 -6.16918178e-03 -3.13037246e-01 -4.93760645e-01
4.20007378e-01 -8.91733885e-01 -2.29962513e-01 -3.66382748e-01
-7.09029436e-01 -5.04524648e-01 6.29132509e-01 -1.99316517e-01
1.30381536e+00 -2.45405746e+00 -1.14913739e-01 5.73954046e-01
2.51044750e-01 4.08813626e-01 -1.91530406e-01 3.50865945e-02
-1.76077113e-01 5.73720098e-01 -2.32857570e-01 -5.09617984e-01
-2.22888097e-01 2.92017341e-01 -2.39732474e-01 4.61526245e-01
2.87144393e-01 2.99329638e-01 -9.39727664e-01 -3.03141445e-01
-2.53279626e-01 1.70544088e-01 -2.75228713e-02 3.79762292e-01
1.20617151e-01 4.42999542e-01 -4.89643693e-01 3.37252557e-01
9.01519120e-01 -1.77820101e-01 3.46738189e-01 -1.79025322e-01
2.94740617e-01 3.70338976e-01 -1.68967140e+00 9.97407258e-01
-2.31287003e-01 2.78253406e-01 1.45189255e-01 -1.14334238e+00
1.14038026e+00 3.76900703e-01 6.67757392e-02 -4.44569111e-01
2.95598775e-01 3.86014819e-01 1.64358988e-01 -4.67276096e-01
-2.18213769e-03 -1.09367400e-01 2.59936452e-01 3.30632269e-01
-1.46085218e-01 -1.46439493e-01 2.78495550e-01 -1.82863474e-01
7.29374230e-01 -3.41841280e-01 -6.82022572e-02 -2.07004413e-01
3.35135490e-01 -5.92839122e-01 1.24605262e+00 1.15136743e+00
-3.36005479e-01 6.77167296e-01 5.60792267e-01 -3.47052515e-01
-6.82024658e-01 -7.46418536e-01 -2.84273386e-01 7.62397587e-01
2.37575054e-01 -3.31709713e-01 -6.40291631e-01 -1.16371572e+00
-2.34271795e-01 6.93598986e-01 -5.70757687e-01 -3.58527362e-01
-5.72843134e-01 -9.54526663e-01 3.68117422e-01 4.04800713e-01
4.35664117e-01 -1.10740423e+00 -1.38511792e-01 3.89660411e-02
-5.31562082e-02 -5.66390991e-01 -2.02692673e-01 8.14651728e-01
-1.02888918e+00 -1.19094419e+00 -3.26700717e-01 -7.84896433e-01
9.67019796e-01 3.37240964e-01 1.07905114e+00 4.05635148e-01
5.05770147e-01 -4.60405022e-01 -4.29311961e-01 -4.19584990e-01
-4.65252072e-01 2.30545491e-01 4.36051726e-01 6.09701090e-02
3.43799233e-01 -6.76459610e-01 -2.57593215e-01 4.99142021e-01
-1.09682465e+00 -1.03166498e-01 4.87333238e-01 1.19785452e+00
6.91555619e-01 5.08479238e-01 9.39161539e-01 -9.19709742e-01
8.58515561e-01 -5.22988498e-01 -4.21263933e-01 3.46701384e-01
-7.41389930e-01 6.46559820e-02 1.15209925e+00 -6.61688685e-01
-8.74769986e-01 -1.77214921e-01 -5.06558791e-02 -4.37487930e-01
-3.18680614e-01 5.85903406e-01 -2.30932951e-01 2.97594011e-01
6.87963307e-01 -8.54591429e-02 2.74025332e-02 -6.35663748e-01
6.42735809e-02 7.91506052e-01 1.94160253e-01 -5.52490234e-01
7.47305632e-01 1.94941193e-01 -3.39779794e-01 -4.13557261e-01
-1.15714049e+00 -2.08447173e-01 -5.23838818e-01 6.52361214e-02
3.88710648e-01 -7.28813827e-01 -2.64670849e-01 6.63510025e-01
-8.53349864e-01 -1.72532514e-01 -3.05261493e-01 6.50788426e-01
6.98668957e-02 2.77483195e-01 -5.09039462e-01 -1.00916243e+00
-3.51098090e-01 -1.31265473e+00 6.78024054e-01 6.18194759e-01
-3.21597159e-02 -8.30312073e-01 -1.50778890e-01 -5.89475892e-02
2.65231699e-01 -7.97379762e-02 6.56710267e-01 -1.21577048e+00
-1.39028177e-01 -3.45337927e-01 2.36509107e-02 9.15788651e-01
3.66442740e-01 2.21098453e-01 -1.06689346e+00 -4.68083799e-01
3.26753080e-01 -5.62975764e-01 1.02683258e+00 2.89112687e-01
1.52639198e+00 -3.97086203e-01 -3.45440477e-01 4.61949021e-01
1.10009778e+00 1.13277540e-01 3.67692828e-01 4.33670878e-01
6.56080663e-01 5.63571513e-01 7.19662607e-01 1.82152301e-01
-8.32032785e-02 3.47349435e-01 3.92059267e-01 7.31831938e-02
2.69838840e-01 8.74067023e-02 1.77061260e-01 1.17275810e+00
1.49370670e-01 -1.91201612e-01 -7.19906211e-01 4.55673814e-01
-1.96894705e+00 -8.81056070e-01 -3.33978496e-02 2.55408859e+00
1.31575918e+00 5.44498742e-01 -2.46187851e-01 3.61918569e-01
8.83284330e-01 9.29752067e-02 -6.15212023e-01 -8.71537924e-02
-1.31245688e-01 2.08204594e-02 2.40292683e-01 2.21366405e-01
-1.25329447e+00 4.33225840e-01 5.67678642e+00 1.23799753e+00
-1.30356705e+00 -6.91018030e-02 1.04511380e+00 -2.95522809e-01
-3.50700945e-01 -9.04543698e-02 -9.58895802e-01 8.57626140e-01
8.04294169e-01 2.02743679e-01 1.61048844e-01 9.19965982e-01
3.52500379e-01 -1.32133409e-01 -1.07520783e+00 8.77003551e-01
-1.39601305e-01 -1.00374556e+00 -3.00068736e-01 -3.05848002e-01
9.23930168e-01 9.55111980e-02 -6.91593587e-02 3.67817223e-01
4.26666856e-01 -8.62042665e-01 6.15231156e-01 4.71956670e-01
2.37031877e-01 -1.09412849e+00 1.20201468e+00 8.23766172e-01
-6.10905349e-01 -2.40706176e-01 -5.92969835e-01 1.95572719e-01
-4.52195890e-02 1.39815962e+00 -5.50034285e-01 7.13079751e-01
9.46620286e-01 6.38296962e-01 -6.29727960e-01 1.21409285e+00
-3.82781297e-01 9.86522377e-01 -5.19159198e-01 -8.59674066e-03
-1.18588312e-02 -3.69165093e-01 3.67696822e-01 9.05685842e-01
2.14529470e-01 -3.61447453e-01 5.22731900e-01 6.86521411e-01
-3.58036131e-01 3.88532458e-03 -4.47506905e-01 4.13251758e-01
8.41922760e-01 1.21964836e+00 -5.96493423e-01 -5.12953699e-01
-3.77650978e-03 5.58058321e-01 6.06792331e-01 2.84959674e-01
-7.46993303e-01 -3.99312228e-01 4.04000163e-01 -3.60502034e-01
8.03978145e-02 1.27177760e-01 -5.62590480e-01 -1.12042928e+00
2.42801726e-01 -1.17802501e+00 2.54639745e-01 -2.93078244e-01
-1.86744392e+00 7.66627431e-01 -1.43135384e-01 -1.52429426e+00
3.03048287e-02 -1.91109732e-01 -8.42254400e-01 7.53291488e-01
-1.49580634e+00 -4.84950483e-01 -2.86780477e-01 1.18667811e-01
4.65643197e-01 -5.84699698e-02 5.52041888e-01 4.25023824e-01
-1.17828131e+00 8.61916006e-01 4.50121045e-01 2.10243031e-01
1.08677185e+00 -1.24161530e+00 1.26499623e-01 9.66655493e-01
1.04396865e-01 7.38260686e-01 5.44839084e-01 -7.28042185e-01
-8.30384135e-01 -1.20648813e+00 6.62597477e-01 -1.28119051e-01
3.32665592e-01 -2.65776694e-01 -1.35226727e+00 3.55622888e-01
6.75493330e-02 2.45737135e-01 6.43313825e-01 2.64035821e-01
-1.94527596e-01 -5.64208508e-01 -1.08268213e+00 5.90974629e-01
7.15878069e-01 -2.47300357e-01 -5.82059324e-01 4.21264917e-01
5.11329830e-01 -2.88203865e-01 -6.83562279e-01 6.07665002e-01
2.03117475e-01 -1.04593277e+00 4.49179292e-01 -5.93994021e-01
1.64779246e-01 -3.72404397e-01 7.34212101e-02 -1.81102288e+00
-2.37627447e-01 -6.13418341e-01 -8.62366855e-02 1.59706199e+00
6.09955072e-01 -7.02612221e-01 5.88695824e-01 5.70447206e-01
-1.03249267e-01 -1.01021302e+00 -1.03786504e+00 -9.51284409e-01
2.19370387e-02 -4.35719579e-01 5.52246809e-01 1.11566746e+00
-3.48870009e-01 2.63987392e-01 -4.86486763e-01 1.47509784e-01
3.98366690e-01 -4.11061496e-02 6.96276367e-01 -1.46288717e+00
-1.33702725e-01 -5.02912760e-01 6.94188699e-02 -8.63153338e-01
1.87943876e-01 -6.27485931e-01 5.76707065e-01 -1.10228086e+00
1.37729719e-01 -8.19119990e-01 -5.57894647e-01 6.47085905e-01
-7.42651284e-01 1.46544039e-01 1.57375708e-02 4.60583150e-01
-6.36643589e-01 7.19360650e-01 1.15087616e+00 -1.22999109e-01
-3.14871490e-01 3.67852449e-01 -7.47605681e-01 7.01381683e-01
1.06664574e+00 -9.22952235e-01 -4.30421263e-01 -2.26499051e-01
2.49600172e-01 -4.92787689e-01 -7.64869973e-02 -1.00041473e+00
4.39663939e-02 -3.21185172e-01 5.12702823e-01 -3.43836397e-01
-1.00583576e-01 -7.73435235e-01 -7.56670609e-02 3.16525191e-01
-5.33083200e-01 -1.92833647e-01 -1.13782436e-01 4.85596448e-01
-2.28224963e-01 -7.10840762e-01 9.09054518e-01 -1.07065439e-01
-2.32002527e-01 2.25001708e-01 -1.37702286e-01 2.27182865e-01
7.99560905e-01 -2.16872320e-02 -3.44390512e-01 -1.82424814e-01
-5.66433132e-01 4.80509073e-01 2.96582639e-01 5.15298307e-01
4.77636874e-01 -1.36902642e+00 -7.26622522e-01 2.61500150e-01
-8.05659890e-02 6.49182856e-01 9.95118320e-02 1.02775455e+00
8.38958547e-02 -2.13725373e-01 1.81074992e-01 -7.10085273e-01
-1.22449827e+00 3.54447395e-01 4.65008199e-01 -2.19749659e-01
-4.81866688e-01 1.01078367e+00 -1.66903660e-01 -6.21555984e-01
5.84634781e-01 -4.04383779e-01 -3.36837292e-01 3.04709643e-01
6.53415263e-01 1.84856698e-01 2.40645647e-01 -1.58370897e-01
-3.19430709e-01 2.14147508e-01 -3.60897303e-01 3.54666561e-01
1.39944327e+00 -1.59300730e-01 -2.37627268e-01 7.10490465e-01
1.03673661e+00 9.18517560e-02 -1.38675487e+00 -2.53026158e-01
-3.84591818e-02 -3.71753156e-01 -8.49705040e-02 -6.98381841e-01
-1.36385429e+00 6.37749374e-01 6.95528567e-01 7.48980939e-01
1.05820680e+00 -2.78506696e-01 5.00156820e-01 5.33226132e-01
4.79108207e-02 -1.34210157e+00 -1.02853231e-01 5.33736110e-01
5.42070508e-01 -1.55796444e+00 -9.71448943e-02 -2.94260174e-01
-6.05195522e-01 1.01785672e+00 1.01789379e+00 -1.32011622e-01
6.25329375e-01 7.62112811e-02 3.23541999e-01 1.03658307e-02
-7.14455187e-01 1.35374203e-01 3.13111722e-01 4.37633395e-01
2.11565793e-01 -1.20321169e-01 -3.47043961e-01 8.75139475e-01
-1.26708627e-01 -2.51576543e-01 4.33089614e-01 8.41068506e-01
-4.72825527e-01 -9.52287436e-01 -3.31832290e-01 7.20585048e-01
-5.17037272e-01 1.47502462e-04 -2.99546093e-01 4.24106002e-01
3.27682406e-01 1.18253767e+00 -7.74891526e-02 -5.08309424e-01
2.85851330e-01 1.19099393e-01 -5.78180403e-02 -5.32249033e-01
-7.32237160e-01 1.79709084e-02 8.67515244e-03 -4.18508649e-01
-4.84738350e-01 -2.85717219e-01 -1.12899268e+00 -1.38971582e-01
-1.03792524e+00 5.04372716e-01 3.82243007e-01 1.21907187e+00
3.43825549e-01 6.10290408e-01 1.02142143e+00 -7.17703223e-01
-1.14024198e+00 -1.18076134e+00 -6.30213857e-01 6.24986947e-01
3.66972864e-01 -5.44737875e-01 -8.70603800e-01 -3.06098849e-01] | [9.331063270568848, 3.8813915252685547] |
84cfa1fa-9168-456b-91e8-161283b37987 | language-clustering-for-multilingual-named | null | null | https://aclanthology.org/2021.findings-emnlp.4 | https://aclanthology.org/2021.findings-emnlp.4.pdf | Language Clustering for Multilingual Named Entity Recognition | Recent work in multilingual natural language processing has shown progress in various tasks such as natural language inference and joint multilingual translation. Despite success in learning across many languages, challenges arise where multilingual training regimes often boost performance on some languages at the expense of others. For multilingual named entity recognition (NER) we propose a simple technique that groups similar languages together by using embeddings from a pre-trained masked language model, and automatically discovering language clusters in this embedding space. Specifically, we fine-tune an XLM-Roberta model on a language identification task, and use embeddings from this model for clustering. We conduct experiments on 15 diverse languages in the WikiAnn dataset and show our technique largely outperforms three baselines: (1) training a multilingual model jointly on all available languages, (2) training one monolingual model per language, and (3) grouping languages by linguistic family. We also conduct analyses showing meaningful multilingual transfer for low-resource languages (Swahili and Yoruba), despite being automatically grouped with other seemingly disparate languages. | ['Kyle Shaffer'] | null | null | null | null | findings-emnlp-2021-11 | ['multilingual-named-entity-recognition'] | ['natural-language-processing'] | [-3.31429124e-01 -1.67168558e-01 -4.24106866e-01 -5.68777323e-01
-1.21843207e+00 -9.30150807e-01 8.57709408e-01 3.74352247e-01
-1.09727299e+00 7.66592503e-01 6.70868635e-01 -7.73441494e-01
3.75728846e-01 -4.00488913e-01 -7.20475435e-01 -2.83650219e-01
-1.73603624e-01 8.85485828e-01 -2.42397711e-01 -1.34919733e-01
-1.06981188e-01 2.64449865e-01 -7.00387716e-01 1.10466421e-01
1.23349833e+00 -9.77565423e-02 3.10482413e-01 3.66844296e-01
-3.86161834e-01 5.98154187e-01 -3.01967889e-01 -5.15609860e-01
2.83363193e-01 -3.22218150e-01 -9.48296487e-01 -1.94258809e-01
4.40280080e-01 1.02591313e-01 -2.49889549e-02 8.88773501e-01
4.46059197e-01 -8.03452805e-02 7.04737127e-01 -6.67429090e-01
-1.00456691e+00 1.23271990e+00 -6.24677241e-01 2.01950565e-01
2.28670627e-01 6.06231503e-02 1.25695145e+00 -1.07489789e+00
9.93574202e-01 1.51571810e+00 8.64297688e-01 3.00774485e-01
-1.54906166e+00 -8.31104815e-01 2.40131900e-01 -1.45117298e-01
-1.71129894e+00 -7.86584556e-01 3.70293409e-01 -5.49746037e-01
1.28868198e+00 -2.05994830e-01 3.60026881e-02 9.26453114e-01
4.14312333e-02 6.25652134e-01 1.30471027e+00 -7.38067269e-01
-1.58878803e-01 4.24423188e-01 9.47941095e-02 6.85876191e-01
1.58399254e-01 -8.03829655e-02 -3.39090377e-01 -3.13243330e-01
3.12106371e-01 -5.06356657e-01 5.67315556e-02 -2.38175467e-02
-1.70131433e+00 1.03667116e+00 2.93295115e-01 6.73579991e-01
-2.41009310e-01 -2.68615127e-01 4.40341085e-01 4.61751223e-01
7.93232322e-01 6.55822635e-01 -7.53682435e-01 3.24760050e-01
-9.84105885e-01 -1.85034886e-01 8.56591344e-01 7.40243912e-01
1.00936294e+00 4.42113727e-02 2.35186204e-01 1.28419363e+00
2.78564960e-01 5.50285399e-01 5.69551826e-01 -3.50373834e-01
7.63742208e-01 3.26636642e-01 -1.74054980e-01 -4.76409227e-01
-3.49779487e-01 5.60897775e-03 -6.27783656e-01 -1.84733570e-01
5.30580580e-01 -5.27624726e-01 -8.74125659e-01 1.95022798e+00
2.33651504e-01 -1.64653629e-01 4.15309966e-01 5.89787960e-01
4.09179688e-01 7.92970955e-01 5.64426363e-01 8.02733600e-02
1.44332016e+00 -8.63078713e-01 -4.39627796e-01 -4.25119907e-01
9.05607700e-01 -1.09995687e+00 1.01631749e+00 -6.12992123e-02
-7.03104377e-01 -5.69761574e-01 -8.05175364e-01 -4.50326830e-01
-7.57510304e-01 2.82835841e-01 7.04558134e-01 5.16659319e-01
-1.24994636e+00 3.67485732e-02 -8.52482021e-01 -8.03134561e-01
1.98627375e-02 3.59654367e-01 -7.80834794e-01 -3.14403445e-01
-1.37464797e+00 1.14455318e+00 7.09975123e-01 -1.98315591e-01
-7.19902456e-01 -7.24101126e-01 -1.18039358e+00 -2.22384080e-01
-1.18195556e-01 -4.25910503e-01 8.09893787e-01 -9.43575621e-01
-1.05863082e+00 1.33984542e+00 -2.56244570e-01 -3.79275650e-01
2.73875505e-01 -3.18878472e-01 -7.95656085e-01 -2.52405196e-01
7.36031115e-01 9.55311596e-01 2.06806809e-01 -1.09042299e+00
-7.63704717e-01 -3.47618878e-01 -2.60596246e-01 3.76721114e-01
-4.56090540e-01 7.27863371e-01 -3.71710926e-01 -7.27157652e-01
-1.40209243e-01 -9.56315279e-01 -4.98251826e-01 -5.06331801e-01
-3.91526550e-01 -4.93681341e-01 2.67185092e-01 -1.27750242e+00
1.16135979e+00 -1.94788170e+00 4.16180976e-02 1.45597100e-01
-1.40293986e-01 -3.27885263e-02 -3.98957998e-01 3.81125659e-01
-1.30597949e-01 5.04348576e-01 -1.77306548e-01 -4.81653690e-01
3.05917207e-03 2.97485203e-01 -2.50025898e-01 5.27285099e-01
6.76614046e-01 8.33848476e-01 -9.61352825e-01 -6.74812019e-01
2.09412295e-02 6.40967011e-01 -5.52894711e-01 9.57541168e-03
3.55616175e-02 5.66735268e-01 1.51142880e-01 6.85560942e-01
4.18477923e-01 2.51718789e-01 7.05630481e-01 -2.08004080e-02
-5.14421046e-01 7.96618640e-01 -9.48813915e-01 1.75725067e+00
-8.78034115e-01 5.10728419e-01 1.25451803e-01 -6.35816157e-01
7.50824869e-01 4.24646765e-01 1.48819402e-01 -4.94854301e-01
-8.70700181e-02 6.06438696e-01 2.70337284e-01 -2.39610791e-01
5.49972773e-01 -2.90193945e-01 -6.85195029e-01 7.37509131e-01
5.89366913e-01 1.77943423e-01 2.20367432e-01 5.74081317e-02
7.76266277e-01 6.80229366e-02 5.83867490e-01 -6.60949349e-01
3.77194494e-01 1.79619431e-01 7.07244277e-01 5.40981233e-01
-2.77810931e-01 2.41615415e-01 5.86979240e-02 -4.44370568e-01
-1.24912238e+00 -1.34434700e+00 -4.55324292e-01 1.70040846e+00
-3.87271553e-01 -3.07650834e-01 -3.10908675e-01 -8.21086645e-01
3.93016413e-02 7.76203573e-01 -3.79181474e-01 3.59410465e-01
-9.33194935e-01 -1.04328024e+00 9.81491029e-01 2.80852079e-01
-1.47688203e-02 -1.14352000e+00 4.15439725e-01 3.80158633e-01
-2.97092557e-01 -1.23978579e+00 -7.55977571e-01 4.91582662e-01
-4.87074196e-01 -6.45256996e-01 -6.65180743e-01 -1.39127600e+00
6.38154626e-01 -1.11355714e-01 1.39652312e+00 -3.51668626e-01
2.37815008e-02 1.70890670e-02 3.69687863e-02 -1.83479130e-01
-7.45402753e-01 6.19828045e-01 5.88132977e-01 -1.00155301e-01
8.36806953e-01 -3.20233434e-01 2.49557570e-01 4.95823249e-02
-6.66186571e-01 -2.36134887e-01 5.92387617e-01 6.53467238e-01
4.52225715e-01 -2.51651376e-01 7.42606223e-01 -1.19491529e+00
5.16113997e-01 -8.60941052e-01 -5.25846839e-01 5.44565260e-01
-2.94982076e-01 3.40746790e-01 7.20669210e-01 -4.88205314e-01
-9.65226114e-01 2.30951667e-01 -1.63788110e-01 1.52269183e-02
-3.93365413e-01 5.85407138e-01 -2.85149246e-01 1.87058926e-01
7.38274455e-01 3.84183158e-03 -3.72302800e-01 -6.57971501e-01
8.32783043e-01 9.28222537e-01 6.78894758e-01 -8.26373458e-01
9.24502552e-01 1.47550374e-01 -8.78828406e-01 -1.03265560e+00
-6.57924235e-01 -6.58626199e-01 -1.18257475e+00 3.09835881e-01
1.11538815e+00 -1.49523330e+00 -2.81746946e-02 5.01692034e-02
-1.22046554e+00 -1.86875314e-01 6.44710436e-02 8.62739325e-01
3.66168693e-02 7.89650157e-02 -8.19989800e-01 -5.18677235e-01
-1.93732038e-01 -1.01129091e+00 8.96532476e-01 -1.79020744e-02
-5.94699621e-01 -1.58026874e+00 5.75078666e-01 1.89629272e-01
2.39352152e-01 4.41713296e-02 1.19991386e+00 -1.18635046e+00
-1.84450835e-01 2.14099705e-01 -1.16989382e-01 2.03196898e-01
4.51925546e-01 -9.77999344e-02 -9.08795297e-01 -4.82382417e-01
-4.95055974e-01 -5.00471592e-01 7.64245272e-01 3.93324019e-03
4.85305488e-02 -1.86825842e-01 -4.43423629e-01 6.31486535e-01
1.53270078e+00 1.40454695e-02 -4.73842323e-02 4.47234333e-01
1.00827014e+00 7.49529898e-01 -1.50065208e-02 -3.32369000e-01
9.32136834e-01 2.53522038e-01 -5.65744221e-01 -4.62356925e-01
-1.27166227e-01 -4.62986082e-01 8.04322362e-01 1.59510267e+00
3.16726059e-01 1.87547415e-01 -1.46798348e+00 1.15778565e+00
-1.46197343e+00 -7.16090560e-01 3.29409420e-01 2.10284972e+00
1.26901639e+00 -3.41609083e-02 9.32521224e-02 -7.29059100e-01
8.42178822e-01 8.82088095e-02 -3.82078260e-01 -5.82884610e-01
-5.71664214e-01 4.84810412e-01 5.99546075e-01 9.17705297e-01
-1.37310708e+00 1.67121828e+00 6.56321335e+00 4.93311018e-01
-1.17591500e+00 3.71860802e-01 5.89168131e-01 4.12185431e-01
-3.62110168e-01 2.97565073e-01 -1.12956583e+00 1.75985083e-01
1.17800999e+00 -4.35348928e-01 5.08423567e-01 5.84151566e-01
-1.03582460e-02 2.85249919e-01 -1.17495775e+00 6.38997018e-01
2.56571323e-01 -8.95400643e-01 2.12146387e-01 -3.70054459e-03
9.70128834e-01 8.37042391e-01 -2.44122639e-01 6.26621842e-01
1.14588153e+00 -1.07587922e+00 4.59784865e-01 1.00285210e-01
9.90588427e-01 -9.14736748e-01 4.83035564e-01 2.86488622e-01
-1.15554535e+00 2.53976107e-01 -3.84976268e-01 2.68472165e-01
2.45546013e-01 3.44741523e-01 -1.11840463e+00 5.19854307e-01
5.47300160e-01 6.18103147e-01 -7.40308762e-01 6.09913230e-01
-3.55505466e-01 8.82036030e-01 -4.51332659e-01 3.59540284e-01
3.58842134e-01 -2.16253176e-01 3.85435313e-01 1.77407551e+00
4.77798469e-02 -4.82746005e-01 6.59163535e-01 6.61622465e-01
-3.94581735e-01 6.38677001e-01 -9.44914341e-01 -4.24824923e-01
4.59421843e-01 1.28373313e+00 -6.13571942e-01 -4.23353195e-01
-7.29225636e-01 9.82150137e-01 8.58681500e-01 3.34469169e-01
-4.91387457e-01 -5.46750605e-01 7.86965728e-01 -1.41133159e-01
8.75869393e-02 -6.17294967e-01 -1.04812816e-01 -1.47888720e+00
-2.75537282e-01 -8.55922461e-01 5.27813077e-01 -3.47732186e-01
-1.70785940e+00 8.84338677e-01 -2.73452878e-01 -8.38944376e-01
-4.86916870e-01 -5.55826068e-01 -3.20321798e-01 1.30244493e+00
-1.50915849e+00 -1.46905470e+00 6.73681617e-01 5.26257038e-01
3.45114857e-01 -5.02260208e-01 1.16005182e+00 6.32608831e-01
-4.87981439e-01 8.68291795e-01 1.73382327e-01 7.79501796e-01
1.26425469e+00 -1.28161073e+00 6.89855814e-01 1.00745916e+00
6.67204022e-01 1.09451973e+00 1.89840376e-01 -7.21777141e-01
-1.06463909e+00 -1.34526861e+00 1.97028041e+00 -6.82245851e-01
1.23854697e+00 -8.41048002e-01 -1.00361168e+00 1.18737400e+00
7.39040434e-01 -2.99973398e-01 1.12978518e+00 7.30501413e-01
-7.10830927e-01 1.92015842e-01 -8.88985693e-01 7.37606645e-01
7.24518895e-01 -1.07176960e+00 -8.34326625e-01 5.06215990e-01
8.75034392e-01 1.83281258e-01 -1.00557578e+00 4.40854616e-02
3.66857916e-01 -2.26969078e-01 7.21413851e-01 -1.05549145e+00
2.39970256e-02 -4.34496373e-01 -3.81655961e-01 -1.65464664e+00
-3.68848920e-01 -5.41473687e-01 7.65309751e-01 1.66144848e+00
8.33083451e-01 -6.79452300e-01 7.31688365e-02 9.34941992e-02
1.13794655e-01 -4.21336964e-02 -8.53292406e-01 -8.42635870e-01
6.61279440e-01 -3.17296982e-01 2.68590808e-01 1.76785278e+00
-4.36434112e-02 1.01000583e+00 -2.41722122e-01 5.01941323e-01
6.12073541e-01 -1.32457078e-01 6.24926448e-01 -1.13103640e+00
-1.45543650e-01 -2.01141357e-01 -1.52135447e-01 -5.96466422e-01
7.47406125e-01 -1.65845466e+00 2.71782339e-01 -1.41130376e+00
2.86003977e-01 -7.02814937e-01 -4.63176936e-01 8.34940493e-01
-3.35534751e-01 2.12684885e-01 6.76611140e-02 3.14021409e-01
-4.71701771e-01 8.08730349e-02 3.74183297e-01 -1.42822325e-01
-2.60274261e-01 -6.28632784e-01 -8.48563850e-01 7.10400701e-01
6.15700841e-01 -7.33681142e-01 2.12514654e-01 -9.64275420e-01
6.35428205e-02 -3.24313283e-01 -1.91495314e-01 -8.08786809e-01
1.88963518e-01 -1.89579166e-02 4.30660516e-01 -2.17091575e-01
-2.00478137e-01 -5.55504143e-01 -1.14105180e-01 3.11809450e-01
-2.90053517e-01 4.56889242e-01 3.61438066e-01 -2.73604412e-02
-3.43138605e-01 1.18903317e-01 8.44681799e-01 -1.01395927e-01
-8.91597569e-01 1.55700773e-01 -5.25123358e-01 4.16274518e-01
6.22804344e-01 4.40873116e-01 -1.16293989e-01 1.09200217e-01
-7.90467739e-01 3.94899458e-01 5.71012020e-01 8.22279632e-01
-2.01818109e-01 -1.52173078e+00 -1.14481688e+00 4.46976513e-01
3.00490975e-01 -4.55365688e-01 -3.50304633e-01 4.90034014e-01
-4.58715260e-01 4.56239969e-01 -1.03449970e-01 -5.45586765e-01
-8.86478424e-01 4.44620460e-01 1.15956917e-01 -5.73856413e-01
-1.31198943e-01 7.19602287e-01 2.02999949e-01 -1.64274621e+00
-7.74042532e-02 -4.26354483e-02 -1.10549554e-01 2.30174109e-01
1.91653520e-01 -9.81930345e-02 -3.85424383e-02 -1.26612234e+00
-7.34958529e-01 5.47701657e-01 -3.42545897e-01 -4.86194223e-01
1.31844115e+00 -8.01104903e-02 -3.33483845e-01 7.28839099e-01
1.38952720e+00 7.04183340e-01 -5.11402190e-01 -5.39886177e-01
6.40013456e-01 1.69620782e-01 -2.91169822e-01 -7.66547203e-01
-4.73195493e-01 7.61113524e-01 3.50493252e-01 -3.25552195e-01
6.96633399e-01 1.32159144e-01 6.90611064e-01 3.58510286e-01
4.12024915e-01 -1.12715721e+00 -5.89066446e-01 1.04969120e+00
3.38969350e-01 -1.45491707e+00 -2.35330150e-01 1.86639860e-01
-6.57046556e-01 8.31559598e-01 3.95999223e-01 -1.89203233e-01
7.65742838e-01 3.78916174e-01 6.90733731e-01 6.85919374e-02
-6.34296179e-01 -3.53739262e-01 3.73411655e-01 4.14607286e-01
1.01770627e+00 4.26407427e-01 -3.45214546e-01 4.68923241e-01
-4.75982785e-01 -4.99280363e-01 5.77745736e-02 6.61132097e-01
-1.69894278e-01 -1.39694965e+00 -4.39777166e-01 1.18422672e-01
-7.80016780e-01 -7.20027626e-01 -4.41731781e-01 1.04977345e+00
4.57691848e-01 8.41360927e-01 3.00188541e-01 -1.88931003e-01
9.34489071e-02 4.36983258e-01 2.04658911e-01 -9.31849837e-01
-7.98123240e-01 2.68023223e-01 2.64010400e-01 -3.78128588e-02
-3.12002152e-01 -8.57614577e-01 -1.05386651e+00 -1.14232354e-01
-1.13269510e-02 3.08589816e-01 7.20348716e-01 9.10487354e-01
1.07154772e-01 1.06189281e-01 5.55992067e-01 -6.54410899e-01
-1.99449763e-01 -1.07731056e+00 -3.10826778e-01 4.37091142e-01
1.14977740e-01 -1.52232766e-01 -3.06151241e-01 2.00574473e-01] | [10.70944595336914, 9.941396713256836] |
4e2330a0-09a4-4707-b9bf-ba62cca7713e | people-and-places-of-historical-europe | 2305.16718 | null | https://arxiv.org/abs/2305.16718v2 | https://arxiv.org/pdf/2305.16718v2.pdf | People and Places of Historical Europe: Bootstrapping Annotation Pipeline and a New Corpus of Named Entities in Late Medieval Texts | Although pre-trained named entity recognition (NER) models are highly accurate on modern corpora, they underperform on historical texts due to differences in language OCR errors. In this work, we develop a new NER corpus of 3.6M sentences from late medieval charters written mainly in Czech, Latin, and German. We show that we can start with a list of known historical figures and locations and an unannotated corpus of historical texts, and use information retrieval techniques to automatically bootstrap a NER-annotated corpus. Using our corpus, we train a NER model that achieves entity-level Precision of 72.81-93.98% with 58.14-81.77% Recall on a manually-annotated test dataset. Furthermore, we show that using a weighted loss function helps to combat class imbalance in token classification tasks. To make it easy for others to reproduce and build upon our work, we publicly release our corpus, models, and experimental code. | ['Aleš Horák', 'Tereza Vrabcová', 'Michal Štefánik', 'Kristýna Luger', 'Vít Novotný'] | 2023-05-26 | null | null | null | null | ['optical-character-recognition', 'named-entity-recognition-ner', 'information-retrieval'] | ['computer-vision', 'natural-language-processing', 'natural-language-processing'] | [-1.63617060e-01 -1.64476588e-01 -2.32707351e-01 -3.93421710e-01
-1.28865516e+00 -1.03930187e+00 7.53593147e-01 4.03752476e-01
-1.13458943e+00 1.13480151e+00 2.65723974e-01 -3.63023847e-01
2.82186300e-01 -6.80705488e-01 -5.14382660e-01 -2.28180066e-01
2.62182783e-02 4.67752099e-01 6.47472590e-02 -9.42369998e-02
6.71980262e-01 5.80496967e-01 -8.38400245e-01 1.93218499e-01
9.67930257e-01 3.05455983e-01 1.02031350e-01 8.17909777e-01
-1.87841058e-01 8.09080064e-01 -1.28307903e+00 -9.02252495e-01
2.05164701e-02 -2.08421305e-01 -1.19823265e+00 -4.40799981e-01
5.11086226e-01 2.80022919e-02 -3.15957814e-01 7.99946427e-01
5.04491508e-01 2.17658922e-01 8.65959585e-01 -5.62327027e-01
-6.24761641e-01 8.78962696e-01 -3.22628856e-01 4.85558927e-01
1.16783991e-01 -4.24270868e-01 1.20801091e+00 -9.08480585e-01
1.01962745e+00 5.02849638e-01 1.05147314e+00 6.18623078e-01
-9.01764512e-01 -6.24101698e-01 -2.74832815e-01 6.34625405e-02
-1.53457808e+00 -5.67901134e-01 5.42665064e-01 -2.75658309e-01
1.17804873e+00 2.40426078e-01 2.52175868e-01 9.77004707e-01
2.24870756e-01 7.19513416e-01 8.37029457e-01 -7.21381366e-01
8.83017778e-02 1.21130966e-01 1.14145346e-01 4.51492935e-01
3.22542399e-01 -3.66621614e-01 -4.02323961e-01 -1.93856180e-01
3.64369690e-01 -3.46005112e-01 -1.11637972e-01 4.05045182e-01
-9.69415486e-01 7.28893220e-01 7.35066980e-02 9.29660261e-01
-1.74529701e-01 -7.24043250e-02 5.33252776e-01 4.45243493e-02
5.28721035e-01 8.76310647e-01 -6.28724754e-01 -5.14733136e-01
-1.30530846e+00 -8.34008604e-02 9.57904279e-01 9.48857546e-01
5.55311263e-01 1.49814198e-02 2.67811090e-01 1.19460618e+00
-1.58750042e-01 5.77575207e-01 5.98568738e-01 -6.15596473e-01
8.26653004e-01 3.40710938e-01 1.90582633e-01 -8.77770185e-01
-4.66720343e-01 -2.17532933e-01 -5.76361656e-01 -4.19732779e-01
6.92258894e-01 -3.31441134e-01 -1.04400957e+00 1.45566761e+00
5.12716314e-03 -2.46755555e-01 3.06190223e-01 3.90392900e-01
5.39013922e-01 8.40902150e-01 3.41621488e-01 -1.95927033e-03
1.29217136e+00 -8.28139186e-01 -4.11833405e-01 -2.99488604e-01
8.26227903e-01 -9.93282259e-01 6.51451647e-01 2.47111216e-01
-8.60358000e-01 -2.77900904e-01 -9.19695616e-01 -2.28936717e-01
-6.85636103e-01 4.70616192e-01 3.45474958e-01 7.67180920e-01
-7.90160596e-01 7.60635078e-01 -9.09753740e-01 -5.78452647e-01
2.84455150e-01 5.77279972e-03 -4.71738279e-01 7.26423860e-02
-1.26292253e+00 1.11339939e+00 4.60189700e-01 7.96159282e-02
-5.04616559e-01 -6.88852429e-01 -9.33569014e-01 -1.78102896e-01
-1.27539366e-01 -3.92366610e-02 1.37176800e+00 -6.26026571e-01
-1.02231133e+00 1.06408632e+00 -5.68808429e-02 -4.51669574e-01
4.16593462e-01 -4.54272300e-01 -6.69744194e-01 2.95821548e-01
3.64708573e-01 2.63947695e-01 1.66181296e-01 -9.39042687e-01
-7.79962063e-01 -1.02464229e-01 -3.24019372e-01 -5.30765727e-02
-6.01701617e-01 4.99104261e-01 -4.13906842e-01 -8.28713894e-01
-2.69213200e-01 -6.77109361e-01 -3.12404037e-01 -7.18285680e-01
-5.83502769e-01 -8.81457105e-02 1.82153895e-01 -1.26238191e+00
1.43868732e+00 -2.06847119e+00 -4.43893403e-01 1.82704553e-01
-2.02150404e-01 4.29511309e-01 -7.03966990e-02 6.30996525e-01
7.89025351e-02 5.79271436e-01 -3.70148569e-01 -4.22417492e-01
-4.70481440e-02 7.33097345e-02 -2.68252552e-01 5.34268081e-01
1.57712892e-01 5.83702922e-01 -1.05821586e+00 -7.50549793e-01
5.47677651e-03 5.44446647e-01 -1.63560510e-01 -1.98683813e-01
9.99179780e-02 1.69974908e-01 -3.40599209e-01 7.28765547e-01
4.67322379e-01 1.72224984e-01 3.34145576e-01 2.90409327e-01
-3.70771915e-01 6.59130514e-01 -9.49546516e-01 1.67966723e+00
-6.76434875e-01 1.10233188e+00 -2.17437282e-01 -5.99666297e-01
9.61523116e-01 2.39670500e-01 1.12409145e-01 -5.87896407e-01
1.82188079e-01 4.74881232e-01 -3.81061316e-01 -2.81056851e-01
1.23734987e+00 -1.59409076e-01 -5.52745759e-01 2.18451560e-01
1.77116826e-01 6.43521249e-02 5.59834838e-01 6.30697310e-02
1.30873251e+00 -4.37335446e-02 3.96421731e-01 -1.35925800e-01
3.80127758e-01 4.52612072e-01 7.60610282e-01 8.18573952e-01
-1.14073597e-01 6.72570825e-01 3.12281519e-01 -4.30519223e-01
-1.58435464e+00 -8.06016564e-01 -5.53371370e-01 8.92426729e-01
-4.35723335e-01 -6.11243784e-01 -6.56736732e-01 -7.55489349e-01
-4.38587040e-01 1.05980897e+00 -4.01140094e-01 4.13770169e-01
-1.05961084e+00 -8.51732135e-01 1.30144012e+00 7.13024139e-01
3.65915656e-01 -1.05075300e+00 -4.47997123e-01 3.94114077e-01
-1.84597164e-01 -9.57533240e-01 -3.74176234e-01 1.71460837e-01
-5.43352365e-01 -1.06877124e+00 -9.23000574e-01 -1.07257068e+00
5.98382771e-01 -4.76818860e-01 1.15131259e+00 3.02228052e-02
-3.20891351e-01 1.14567831e-01 -5.18309176e-01 -3.60713243e-01
-5.58988333e-01 7.31809199e-01 -1.38917878e-01 -6.05407238e-01
5.05109906e-01 -1.07014500e-01 -1.46547467e-01 2.29608007e-02
-7.50898778e-01 -3.69227171e-01 4.16033059e-01 8.87755871e-01
3.58343035e-01 3.46358940e-02 3.77204508e-01 -1.06710446e+00
1.76245987e-01 -4.54106390e-01 -5.69023907e-01 5.95805943e-01
-4.84577090e-01 -4.41337898e-02 8.75334263e-01 -2.55934805e-01
-1.32482874e+00 1.37140632e-01 -4.93945301e-01 3.45190555e-01
-1.50932327e-01 4.56552297e-01 7.67765194e-02 2.92826056e-01
8.26670766e-01 2.90683284e-02 -8.58587205e-01 -6.84396386e-01
4.65323776e-01 1.15461123e+00 9.03282166e-01 -6.74159765e-01
8.30633104e-01 1.22416012e-01 -5.54735482e-01 -9.81115699e-01
-8.93508315e-01 -4.92335647e-01 -7.95273542e-01 1.41773507e-01
8.00135791e-01 -8.59977722e-01 -1.03121080e-01 5.18172562e-01
-1.31444383e+00 -3.37277591e-01 -2.48997454e-02 6.43614471e-01
2.03710049e-02 3.39852244e-01 -1.03949738e+00 -9.21032965e-01
-5.54262400e-01 -3.80826354e-01 7.39860535e-01 6.13519609e-01
-3.58026087e-01 -1.25539756e+00 4.38498527e-01 2.41138190e-01
2.31328726e-01 2.59931445e-01 6.22672319e-01 -1.22868514e+00
-7.19934404e-02 -3.51533622e-01 4.42295745e-02 2.21733034e-01
3.68905067e-02 2.79768705e-01 -8.93348396e-01 -5.45312325e-03
-3.33498746e-01 -1.91465735e-01 8.96978676e-01 -3.73716950e-02
5.71507215e-01 -2.31295943e-01 -4.35339183e-01 3.83908242e-01
1.51371109e+00 2.54394293e-01 7.61285841e-01 9.48006988e-01
4.73308563e-01 5.38899004e-01 4.37617242e-01 3.29195023e-01
4.18871433e-01 1.25867262e-01 -4.16140765e-01 1.09221362e-01
1.64334431e-01 -4.26519036e-01 2.03220129e-01 8.56339753e-01
-1.08796887e-01 -2.82724977e-01 -1.40432572e+00 1.11951256e+00
-1.31857252e+00 -1.12214923e+00 -1.65942788e-01 2.08073950e+00
1.33940995e+00 1.71673059e-01 -9.59507078e-02 -1.13234930e-01
1.05748630e+00 2.80432962e-02 -1.23746052e-01 -4.73687917e-01
-3.23196113e-01 6.35103405e-01 9.81270432e-01 4.06259060e-01
-1.33696961e+00 1.22167325e+00 6.58897686e+00 8.90453398e-01
-8.83298516e-01 -2.66582868e-03 6.28573418e-01 5.06160259e-02
-2.02379227e-01 2.11453736e-02 -1.10141456e+00 3.90745670e-01
1.39667714e+00 -3.41655761e-01 1.16175361e-01 8.50271404e-01
-2.23366246e-01 -9.29676890e-02 -7.83628523e-01 8.02024245e-01
2.61585742e-01 -1.34843969e+00 -4.76403832e-01 -1.39038116e-01
8.37273479e-01 2.92100340e-01 -4.24449444e-01 4.59248751e-01
6.36399031e-01 -9.82996404e-01 8.32504451e-01 4.50748354e-01
8.03779840e-01 -1.01847672e+00 1.00624347e+00 1.15146481e-01
-1.13142002e+00 1.47598550e-01 -5.51279128e-01 1.01923876e-01
2.92426080e-01 5.16662061e-01 -8.78189385e-01 8.31067562e-01
6.19614422e-01 3.63608718e-01 -7.43730962e-01 1.31029046e+00
-6.48364723e-01 7.46350169e-01 -4.37608063e-01 -3.24949950e-01
2.17843518e-01 2.67236918e-01 2.33273804e-01 1.64963925e+00
2.82490939e-01 -3.77719402e-02 -2.46747196e-01 2.56200314e-01
-5.08082747e-01 4.85843360e-01 -3.84542376e-01 -2.29155436e-01
7.97169209e-01 1.41687942e+00 -9.11950588e-01 -4.12919968e-01
-2.90350258e-01 1.09174180e+00 6.49775803e-01 2.30698287e-01
-7.96477854e-01 -1.25760055e+00 2.35996887e-01 -3.38799000e-01
4.03983206e-01 -4.71450448e-01 -3.88485968e-01 -1.39304495e+00
8.90850946e-02 -5.47917128e-01 6.16285503e-01 -5.50284147e-01
-1.41018236e+00 8.37486267e-01 -2.43962348e-01 -7.64669299e-01
-2.30905160e-01 -6.75683796e-01 -6.53960586e-01 6.63141608e-01
-1.37229121e+00 -9.36663628e-01 4.20854837e-01 -4.66795564e-02
3.98149073e-01 -8.05471838e-02 8.61158252e-01 5.69568932e-01
-7.48929381e-01 7.10709810e-01 5.91552973e-01 1.10946667e+00
9.65471148e-01 -1.42498171e+00 7.89409101e-01 1.04768360e+00
4.15120333e-01 7.84486413e-01 5.60458660e-01 -7.87181795e-01
-8.43500793e-01 -1.08052838e+00 1.78397512e+00 -7.07266688e-01
9.72064257e-01 -4.86029029e-01 -7.59383380e-01 8.19030285e-01
2.38292739e-01 -4.24418628e-01 9.08225060e-01 2.41953611e-01
-4.04956788e-01 1.49493933e-01 -1.16831219e+00 4.91764158e-01
7.82776415e-01 -6.45303607e-01 -1.07098091e+00 3.51467609e-01
3.01049501e-01 -4.88789469e-01 -1.18570292e+00 -1.67831928e-01
5.20284116e-01 -2.67895818e-01 5.66893041e-01 -6.13920629e-01
3.33200365e-01 -2.14722678e-01 -4.62761708e-02 -1.21612811e+00
4.09298688e-02 -6.06196702e-01 5.76978087e-01 1.87819314e+00
1.01038313e+00 -4.57965761e-01 6.34888530e-01 8.08005512e-01
-1.40434340e-01 -2.37886235e-01 -9.88731205e-01 -8.88310730e-01
5.72434366e-01 -5.59206367e-01 3.38162899e-01 1.19274640e+00
1.78023547e-01 4.42113876e-01 -3.26802671e-01 1.69895247e-01
3.67555201e-01 -1.59015834e-01 4.45530206e-01 -1.06835437e+00
1.51252463e-01 -2.22413167e-01 -2.95044094e-01 -5.53490698e-01
2.94459254e-01 -8.06047261e-01 2.24876985e-01 -1.69266164e+00
1.93662092e-01 -6.25261605e-01 -1.01095513e-01 8.33932698e-01
-2.42810566e-02 5.05107284e-01 6.88596442e-03 3.99044752e-01
-6.07121706e-01 2.21575379e-01 4.09465015e-01 2.91163176e-02
-1.07358560e-01 -4.32214230e-01 -5.77455223e-01 5.57252169e-01
1.11753702e+00 -9.63232875e-01 3.92497092e-01 -4.85261708e-01
2.90140420e-01 -3.36949021e-01 -1.19197205e-01 -9.57601428e-01
2.99583197e-01 -1.22843891e-01 8.41983855e-01 -6.01060510e-01
-9.59883910e-03 -2.34147877e-01 -8.50361884e-02 2.85168201e-01
-1.79664403e-01 9.46513861e-02 3.38035375e-01 2.39641964e-01
-1.42472312e-01 -7.88463593e-01 5.17507970e-01 -1.87501907e-01
-7.24714458e-01 -8.40264112e-02 -5.39195955e-01 4.78144974e-01
8.15768659e-01 2.02700615e-01 -7.13931620e-01 1.37650385e-01
-3.10349405e-01 1.58781543e-01 7.32690454e-01 1.99321628e-01
-4.64254580e-02 -1.05705869e+00 -8.14058661e-01 -2.99269438e-01
2.24916749e-02 -2.27057248e-01 8.86472687e-02 3.10074806e-01
-1.04632306e+00 4.67245072e-01 -6.02336265e-02 -4.35842900e-03
-1.00454187e+00 3.32928121e-01 1.32671148e-01 -3.85416269e-01
-5.20967841e-01 6.03240728e-01 -6.84315920e-01 -7.56517887e-01
6.62157908e-02 4.87615801e-02 -3.39070261e-01 1.34197623e-01
7.26894140e-01 3.97879958e-01 1.50758788e-01 -7.92743623e-01
-6.69850171e-01 4.72414762e-01 -1.16839625e-01 -3.83337229e-01
1.41063118e+00 1.00581527e-01 7.10356534e-02 3.87261927e-01
1.21442175e+00 7.87323773e-01 -7.44636238e-01 2.07058210e-02
6.26748562e-01 -1.53048381e-01 -3.10536683e-01 -8.88598979e-01
-6.53746247e-01 4.63001847e-01 1.90771371e-02 1.87010452e-01
9.35832441e-01 3.50890942e-02 9.89950538e-01 5.90405285e-01
1.80287033e-01 -1.53664815e+00 -5.28085649e-01 8.40578020e-01
2.26807386e-01 -9.07685041e-01 9.45276916e-02 5.01749516e-02
-5.67463398e-01 1.28839469e+00 2.43742362e-01 8.35170820e-02
2.88754940e-01 3.49568605e-01 2.04256192e-01 6.43609688e-02
-4.14515048e-01 -9.27492827e-02 1.38721541e-01 5.03947318e-01
6.52746677e-01 -1.15216061e-01 -6.93024933e-01 6.84723973e-01
-7.00040758e-01 -3.22380692e-01 6.40426040e-01 1.10786748e+00
-4.33201432e-01 -1.40147507e+00 -3.64614993e-01 1.37067944e-01
-1.07253385e+00 -4.25428420e-01 -4.33793873e-01 1.02286100e+00
-1.13425285e-01 1.15619636e+00 2.40161091e-01 -2.35969618e-01
1.82627603e-01 3.90322715e-01 2.79095203e-01 -5.44763565e-01
-8.90133619e-01 -1.17776141e-01 7.56393135e-01 2.11305484e-01
-3.47460151e-01 -8.18960309e-01 -1.50216997e+00 -4.55606222e-01
-3.13477993e-01 6.36332810e-01 1.02602434e+00 9.27145958e-01
1.25798553e-01 9.27585289e-02 4.62074637e-01 -3.18627834e-01
-3.46062243e-01 -1.15125525e+00 -5.33634424e-01 1.16243199e-01
-7.41076714e-04 -4.61649299e-02 -4.84934896e-01 2.54438370e-01] | [9.853838920593262, 9.786627769470215] |
229e9a65-7012-41c5-855c-da0b5013a4e8 | energy-efficient-blockchain-enabled-user | 2302.10515 | null | https://arxiv.org/abs/2302.10515v1 | https://arxiv.org/pdf/2302.10515v1.pdf | Energy-Efficient Blockchain-enabled User-Centric Mobile Edge Computing | In the traditional mobile edge computing (MEC) system, the availability of MEC services is greatly limited for the edge users of the cell due to serious signal attenuation and inter-cell interference. User-centric MEC (UC-MEC) can be seen as a promising solution to address this issue. In UC-MEC, each user is served by a dedicated access point (AP) cluster enabled with MEC capability instead of a single MEC server, however, at the expense of more energy consumption and greater privacy risks. To achieve efficient and reliable resource utilization with user-centric services, we propose an energy efficient blockchain-enabled UC-MEC system where blockchain operations and resource optimization are jointly performed. Firstly, we design a resource-aware, reliable, replicated, redundant, and fault-tolerant (R-RAFT) consensus mechanism to implement secure and reliable resource trading. Then, an optimization framework based on alternating direction method of multipliers (ADMM) is proposed to minimize the total energy consumed by wireless transmission, consensus and task computing, where APs clustering, computing resource allocation and bandwidth allocation are jointly considered. Simulation results show superiority of the proposed UC-MEC system over reference schemes, at most 33.96% reduction in the total delay and 48.77% reduction in the total energy consumption. | ['Feng Wu', 'Dan Zhao', 'Zhuojia Gu', 'Yuang Chen', 'Hancheng Lu', 'Langtian Qin'] | 2023-02-21 | null | null | null | null | ['total-energy'] | ['miscellaneous'] | [-3.15063030e-01 -3.25718015e-01 -3.66625935e-01 2.56052315e-01
-2.11258158e-01 -7.73569107e-01 4.08173591e-01 -1.02341138e-01
-2.34457940e-01 9.82553601e-01 -7.41126984e-02 -4.13382351e-01
-4.01903763e-02 -6.82572067e-01 -4.00910944e-01 -1.21632075e+00
-1.27271205e-01 1.64328098e-01 -8.31134841e-02 1.49911493e-01
2.48938352e-01 1.94628417e-01 -6.81747913e-01 -1.31276429e-01
8.92127156e-01 1.41031909e+00 4.33532931e-02 1.80611864e-01
3.99530530e-01 6.00543976e-01 -7.64636323e-02 -3.13742250e-01
2.95875162e-01 -2.80588806e-01 -5.88181615e-01 8.57689083e-02
-9.96448755e-01 -4.27260607e-01 -5.04750237e-02 1.04288447e+00
3.98283511e-01 -2.23006934e-01 5.17126560e-01 -2.02293253e+00
-4.78450179e-01 7.03227401e-01 -1.08962417e+00 -1.67096898e-01
-1.58573881e-01 -1.31294310e-01 5.31439900e-01 -7.15001285e-01
7.74444103e-01 2.82939255e-01 4.26449955e-01 2.90676147e-01
-8.92393768e-01 -8.46252203e-01 -1.82766631e-01 6.15165412e-01
-1.64210343e+00 -7.01045752e-01 6.35046005e-01 1.18938059e-01
7.46422768e-01 7.33026147e-01 8.61075103e-01 1.77377313e-01
6.56798601e-01 4.54797059e-01 1.03491616e+00 -3.94969285e-01
6.15431249e-01 1.94512323e-01 -4.02499110e-01 1.83591276e-01
8.02593231e-01 -2.91186959e-01 -3.33598137e-01 -5.28488696e-01
3.99357557e-01 1.53685272e-01 -5.25886893e-01 -4.85128224e-01
-1.41378081e+00 3.18810672e-01 2.98233390e-01 3.85311753e-01
-8.19380999e-01 5.61395884e-01 5.19650340e-01 2.40347490e-01
2.04241246e-01 -4.10599381e-01 -3.74771714e-01 -4.62895297e-02
-1.00233102e+00 -3.01708192e-01 8.71860385e-01 1.34044564e+00
3.01081300e-01 1.20595634e-01 4.82513802e-03 -2.93780357e-01
7.41687179e-01 6.32193089e-01 5.34317736e-03 -1.39938414e+00
-8.01660642e-02 3.71880293e-01 4.93936539e-01 -9.12909925e-01
-1.58292875e-01 -4.87490296e-01 -1.45776772e+00 -3.61528397e-02
-1.04253359e-01 -5.26710927e-01 -1.85667247e-01 1.64614403e+00
6.22267604e-01 4.13104773e-01 2.49135196e-01 1.01305568e+00
6.77787066e-02 8.85065675e-01 -8.65314677e-02 -1.09032595e+00
1.52895141e+00 -8.68435800e-01 -8.84845495e-01 4.35175389e-01
5.56598485e-01 -6.64312840e-01 -2.11056337e-01 1.73752457e-01
-1.27825773e+00 6.43542051e-01 -1.08507574e+00 4.51761544e-01
6.07654080e-02 -1.87206417e-02 4.36041445e-01 7.72347033e-01
-1.11375225e+00 -3.32654446e-01 -6.72414124e-01 -1.36816114e-01
3.35403621e-01 5.84106445e-01 -2.74793357e-01 2.20835619e-02
-8.68367672e-01 4.77467954e-01 2.77153790e-01 4.96875532e-02
-6.51980340e-01 -5.31895638e-01 -3.00405741e-01 3.98895144e-01
3.17991763e-01 -9.49380875e-01 8.84206235e-01 -8.00702333e-01
-1.41989696e+00 2.06315860e-01 -4.09312453e-03 -4.03217405e-01
5.13219833e-01 6.44646525e-01 -4.45395142e-01 3.08943361e-01
-2.31500104e-01 8.70598108e-02 4.67253685e-01 -1.23610067e+00
-8.51887226e-01 -2.18447119e-01 -3.17543149e-01 1.69040889e-01
-4.00426000e-01 1.66349441e-01 -1.79900393e-01 -3.54224503e-01
-5.80746355e-03 -1.26375055e+00 -1.85409606e-01 -3.21404040e-02
-1.96248859e-01 2.01561861e-02 1.16830695e+00 -7.07419455e-01
1.30516768e+00 -2.02247310e+00 1.20056875e-01 6.90283418e-01
2.26336494e-01 -8.25303197e-02 4.97651786e-01 4.48633313e-01
6.00797713e-01 8.85244980e-02 -1.31001368e-01 1.25409113e-02
-7.97728077e-03 -1.20793149e-01 1.10710390e-01 7.92833328e-01
-6.31078124e-01 5.77280283e-01 -7.18326449e-01 -5.64683795e-01
-1.79813698e-01 4.95521128e-01 -4.58293945e-01 -2.38625214e-01
1.83022514e-01 5.93477428e-01 -5.84929824e-01 9.52084541e-01
9.98064458e-01 -5.20546496e-01 8.43310058e-01 -7.73010850e-02
-2.54804075e-01 -5.23473144e-01 -1.49871433e+00 1.38362205e+00
-4.06500965e-01 3.42080146e-01 9.76639032e-01 -7.27267087e-01
2.63786286e-01 1.00623524e+00 8.50469410e-01 -4.92848158e-01
4.73921746e-01 6.99746192e-01 -1.35457575e-01 -1.50802791e-01
2.45558232e-01 1.99318901e-01 -3.42603438e-02 9.24055755e-01
-5.87298930e-01 5.98027110e-01 -2.58271277e-01 4.85995620e-01
1.10750854e+00 -2.24207029e-01 2.43012518e-01 -5.64698756e-01
7.63579726e-01 -1.39455840e-01 1.08693242e+00 -5.77575353e-04
-4.07779962e-01 -3.11369956e-01 1.15171954e-01 -1.03455372e-01
-9.50628698e-01 -3.76942247e-01 2.16158032e-01 5.10953248e-01
8.04984391e-01 -3.28579754e-01 -7.41242945e-01 -2.89721787e-01
-1.77045599e-01 5.78857660e-01 -7.31977541e-03 -2.74567083e-02
-3.46107185e-01 -7.32217968e-01 2.20411539e-01 2.18731105e-01
9.40180480e-01 -5.54514527e-01 -9.64825988e-01 3.57344955e-01
-5.35514176e-01 -9.65335488e-01 -8.99142444e-01 -3.34816635e-01
-5.62165380e-01 -8.31447363e-01 -3.88130307e-01 -7.25659728e-01
8.99463832e-01 5.13826787e-01 5.55134654e-01 4.13267970e-01
2.12351624e-02 2.32463807e-01 -1.85087770e-01 -7.12993592e-02
4.98696044e-02 -2.13957414e-01 3.64609569e-01 4.93136585e-01
-9.43339542e-02 -7.34009087e-01 -1.36868942e+00 4.20312196e-01
-7.11971045e-01 1.61584899e-01 5.45631945e-01 6.16299808e-01
5.68246782e-01 2.92215407e-01 1.06313014e+00 -4.31975484e-01
3.65107536e-01 -9.69942629e-01 -6.72111630e-01 3.66076589e-01
-1.20680964e+00 -4.05368835e-01 7.05476046e-01 6.53842557e-03
-1.19259274e+00 1.79293126e-01 5.30641139e-01 -1.17678687e-01
6.62259877e-01 2.00760841e-01 -3.87951851e-01 -4.76883948e-01
-2.99473941e-01 4.25114959e-01 -1.23505719e-01 6.45984039e-02
3.37762862e-01 1.10779083e+00 1.31768093e-01 -3.17176998e-01
4.15950060e-01 8.52362216e-01 4.15104210e-01 -1.75491065e-01
6.11747444e-01 -9.29828361e-02 3.75168383e-01 -4.54280138e-01
5.94148457e-01 -1.16973186e+00 -1.44218075e+00 3.50139409e-01
-1.07925153e+00 1.20980285e-01 2.76321352e-01 5.85582137e-01
-3.50424170e-01 3.99414808e-01 -6.12997651e-01 -1.07278633e+00
-1.12726843e+00 -1.10175157e+00 3.08887899e-01 2.67644286e-01
6.12581633e-02 -4.55660522e-01 -5.28356433e-01 2.62352854e-01
9.48998690e-01 3.40003490e-01 6.63264275e-01 -1.28018364e-01
-1.10900128e+00 -5.22717908e-02 -2.73117721e-01 -1.41934365e-01
1.30083561e-01 -4.36642468e-02 -3.08034122e-01 -8.66841197e-01
2.16853451e-02 1.95711896e-01 -8.23840573e-02 3.42803776e-01
1.01179934e+00 -9.34539139e-01 -7.65808225e-01 4.24653143e-01
1.88002241e+00 3.80115002e-01 5.03254175e-01 1.49836123e-01
2.25881115e-01 2.29372159e-01 5.87820232e-01 9.75072145e-01
7.82876074e-01 5.35487413e-01 7.03595996e-01 4.52794224e-01
3.04208010e-01 2.45805815e-01 3.70418876e-01 1.11681950e+00
-3.03851545e-01 -4.99983579e-01 -3.54968995e-01 7.08027959e-01
-2.20369649e+00 -6.93370044e-01 -3.56618762e-01 2.03098440e+00
5.59953034e-01 -2.74117678e-01 1.49425015e-01 4.21670347e-01
8.91291499e-01 -5.23119628e-01 -5.83583474e-01 -3.37133467e-01
1.38311580e-01 -5.02641916e-01 9.03591216e-01 -4.21724766e-02
-1.69761792e-01 1.80029422e-01 4.62468719e+00 9.45932269e-01
-1.10467160e+00 7.88581312e-01 7.28700817e-01 -1.55261487e-01
-4.84607071e-01 4.50821906e-01 9.95078161e-02 1.05626929e+00
1.11771679e+00 -4.96588796e-01 8.05023730e-01 6.71817243e-01
5.10540009e-01 -1.96620405e-01 -5.32143414e-01 1.04184151e+00
-5.80595672e-01 -1.74945509e+00 -5.85700452e-01 4.41340387e-01
1.01789987e+00 -2.85486519e-01 -3.83809209e-01 -5.09225905e-01
1.07769743e-01 -3.11921746e-01 9.37025487e-01 3.28925788e-01
1.01163924e+00 -1.29799044e+00 6.39626384e-01 4.10025507e-01
-1.38021266e+00 -4.41342473e-01 -5.43557256e-02 1.69428274e-01
5.86009860e-01 7.40341544e-01 -5.31592220e-03 7.72541046e-01
8.81052434e-01 -9.20563471e-03 4.72996593e-01 8.22709382e-01
3.13913852e-01 3.74969155e-01 -4.29529697e-01 -9.78909619e-03
-1.54297426e-01 -6.45055532e-01 6.45228863e-01 7.65302956e-01
6.95699692e-01 6.77517593e-01 -3.08703277e-02 4.88978565e-01
-6.24106824e-01 5.35165370e-02 -2.49978513e-01 1.86129078e-01
1.32830882e+00 1.74163330e+00 -8.52861941e-01 -3.08503151e-01
-4.16025937e-01 9.59033906e-01 -2.28358179e-01 3.60839605e-01
-8.04055333e-01 -5.11753798e-01 4.68675613e-01 8.58487040e-02
2.57567495e-01 -3.50554377e-01 -6.31877542e-01 -9.69747066e-01
-1.35431783e-02 -4.69917417e-01 2.47289121e-01 -5.85292399e-01
-6.89375699e-01 1.22112751e-01 -5.41297019e-01 -1.10361683e+00
8.45122933e-02 3.55808884e-01 -7.77312100e-01 4.57492381e-01
-1.64916968e+00 -1.19866216e+00 -1.03510395e-01 8.04132342e-01
-5.59218861e-02 -1.10304751e-01 5.88158011e-01 7.00101435e-01
-5.90985596e-01 6.91395879e-01 6.41197503e-01 -3.97639722e-01
4.09184337e-01 -6.13265872e-01 -5.27560413e-01 1.05492067e+00
-5.27495086e-01 3.88274193e-01 5.56141496e-01 -5.83768487e-01
-2.36159277e+00 -1.00828421e+00 8.31033647e-01 3.79322141e-01
3.34605396e-01 -2.46408746e-01 -1.68403834e-01 6.02008641e-01
6.60685480e-01 1.36433125e-01 7.32464969e-01 -1.06331384e+00
2.81350523e-01 -5.56501210e-01 -1.72676265e+00 7.31270611e-01
9.01009142e-01 -8.90920237e-02 5.98524272e-01 2.51128554e-01
5.88309228e-01 2.81358361e-02 -1.06141758e+00 2.23732084e-01
6.60113931e-01 -5.15934587e-01 3.34530056e-01 1.60266578e-01
-4.79120970e-01 -8.58977795e-01 -2.46441931e-01 -1.00416529e+00
-4.93833572e-01 -1.22816670e+00 -6.94648862e-01 1.39601386e+00
3.31023224e-02 -7.67985225e-01 7.43845761e-01 7.28588283e-01
3.83168319e-03 -7.21858621e-01 -1.26597929e+00 -6.04987979e-01
-4.24295038e-01 1.35068506e-01 1.06461334e+00 9.87262070e-01
5.02515912e-01 -8.33963882e-03 -5.35931528e-01 3.51144373e-01
1.04875684e+00 2.29326263e-01 4.64206904e-01 -9.39916670e-01
3.76756974e-02 -2.38757618e-02 5.68100847e-02 -3.97349417e-01
-1.48837894e-01 -8.14247012e-01 -1.31084323e-01 -1.50223184e+00
4.35522497e-01 -7.88227737e-01 -3.76423538e-01 3.59942555e-01
5.36632180e-01 3.86696905e-01 3.50555390e-01 5.84157765e-01
-8.27405512e-01 3.90292913e-01 8.78025711e-01 1.89263731e-01
1.06621228e-01 -1.24230161e-01 -7.06563294e-01 1.09052077e-01
8.87210429e-01 -5.18991947e-01 -4.49556053e-01 -2.11350918e-01
3.93103510e-01 5.99957466e-01 1.45625398e-02 -6.35737479e-01
7.73227513e-01 -5.22175491e-01 -1.36949662e-02 -3.35885286e-01
-6.00451455e-02 -1.73360384e+00 1.16982281e+00 9.79276836e-01
4.09391910e-01 1.00221001e-01 -6.00235105e-01 8.92841160e-01
1.99261755e-01 2.90599108e-01 5.24966359e-01 1.16674654e-01
-2.56299227e-01 3.31894070e-01 -7.88802505e-01 -7.17409790e-01
1.83918095e+00 -1.82783931e-01 -4.87567574e-01 -3.74411911e-01
-2.24587739e-01 7.63683081e-01 6.78498387e-01 -4.38650921e-02
2.02284783e-01 -1.45071483e+00 -7.45424688e-01 -1.08945303e-01
-3.18660229e-01 -2.57069170e-01 4.76695567e-01 1.52866197e+00
-5.07124364e-01 5.11764288e-01 -3.23759466e-01 -2.64594734e-01
-1.09176219e+00 6.07287586e-01 9.27044526e-02 -1.16589159e-01
-2.22766817e-01 2.00037956e-01 -3.87765706e-01 2.57748485e-01
-3.75327058e-02 2.29326218e-01 4.57385987e-01 -2.36664057e-01
1.48466870e-01 1.14537168e+00 1.30431384e-01 -6.08378768e-01
-6.73636198e-01 1.43963411e-01 3.70121807e-01 -1.66363463e-01
1.07270050e+00 -8.29631090e-01 -8.23866844e-01 -5.10188460e-01
6.70390844e-01 1.40386075e-01 -8.26119542e-01 -3.85653973e-02
-8.25543329e-02 -2.22175300e-01 4.61533129e-01 -6.40119970e-01
-1.44405711e+00 -1.08686108e-02 4.72358286e-01 6.83114976e-02
1.54784691e+00 -6.29305601e-01 1.25647366e+00 -1.66354373e-01
9.50139284e-01 -1.36157644e+00 -4.80402589e-01 -3.39358717e-01
2.49418572e-01 -6.05201781e-01 1.43108116e-02 -4.56728041e-01
-4.68607634e-01 8.46454322e-01 4.96710151e-01 1.29958153e-01
1.00118661e+00 2.39192531e-01 -7.10557401e-01 -4.21326943e-02
-8.71085465e-01 5.03043532e-01 -5.73464572e-01 1.89435720e-01
-1.88543916e-01 5.53810775e-01 -1.09207809e+00 8.78421903e-01
5.15517116e-01 3.43484014e-01 7.93672800e-01 1.40632379e+00
-3.14093739e-01 -1.16190946e+00 -3.15337181e-01 2.28190154e-01
-7.93416202e-01 1.06478974e-01 2.09215999e-01 1.65167749e-01
2.08150595e-01 1.24765968e+00 -2.66576093e-02 -1.77094191e-01
-3.82113397e-01 -1.85273424e-01 -2.64294036e-02 1.36279225e-01
-3.81290346e-01 3.01429272e-01 5.68211414e-02 -1.63951188e-01
-3.11859757e-01 -4.88298625e-01 -1.69527125e+00 -9.93178129e-01
-5.67641258e-01 5.96850097e-01 1.14287162e+00 7.20294952e-01
1.05161572e+00 7.57854655e-02 1.29844213e+00 -2.95088202e-01
-1.36164799e-01 -3.47741842e-01 -8.77972305e-01 -3.03408384e-01
1.03738330e-01 -5.82128726e-02 -2.32976839e-01 1.07975444e-02] | [5.903961658477783, 1.7446110248565674] |
10209627-89ca-451d-8004-b052acf7ce29 | facial-emotion-recognition-using-deep-1 | 2111.02717 | null | https://arxiv.org/abs/2111.02717v1 | https://arxiv.org/pdf/2111.02717v1.pdf | Facial Emotion Recognition using Deep Residual Networks in Real-World Environments | Automatic affect recognition using visual cues is an important task towards a complete interaction between humans and machines. Applications can be found in tutoring systems and human computer interaction. A critical step towards that direction is facial feature extraction. In this paper, we propose a facial feature extractor model trained on an in-the-wild and massively collected video dataset provided by the RealEyes company. The dataset consists of a million labelled frames and 2,616 thousand subjects. As temporal information is important to the emotion recognition domain, we utilise LSTM cells to capture the temporal dynamics in the data. To show the favourable properties of our pre-trained model on modelling facial affect, we use the RECOLA database, and compare with the current state-of-the-art approach. Our model provides the best results in terms of concordance correlation coefficient. | ['Björn W. Schuller', 'Elnar Hajiyev', 'Dénes Boros', 'Panagiotis Tzirakis'] | 2021-11-04 | null | null | null | null | ['facial-emotion-recognition'] | ['computer-vision'] | [ 5.62348552e-02 8.44000503e-02 2.60185540e-01 -6.30374014e-01
-1.15644813e-01 -2.39668190e-01 8.80109727e-01 -1.89541057e-01
-6.82495177e-01 7.06646919e-01 8.64643380e-02 4.41156745e-01
6.25725314e-02 -2.74468541e-01 -3.26775521e-01 -6.96368098e-01
-3.56869072e-01 1.60563901e-01 -1.53424283e-02 -2.76070356e-01
-8.66399147e-03 5.33509493e-01 -2.07399201e+00 7.66841888e-01
1.34939805e-01 1.33790922e+00 -4.55661088e-01 8.69760513e-01
-6.34251684e-02 1.08670866e+00 -6.10955596e-01 -6.63184047e-01
-4.75378633e-02 -3.78150970e-01 -7.95697689e-01 1.69497147e-01
3.70515108e-01 -1.68627113e-01 -7.57248774e-02 6.49916470e-01
6.79465652e-01 8.51586983e-02 7.27943361e-01 -1.36529636e+00
7.65925720e-02 3.08823660e-02 -5.45904756e-01 4.66563433e-01
6.99693203e-01 -1.04553101e-03 5.61115742e-01 -8.53999257e-01
9.08500075e-01 1.30676603e+00 4.97892141e-01 9.03963745e-01
-1.07533419e+00 -5.90588152e-01 1.05419047e-01 4.84924227e-01
-9.59033251e-01 -8.05348814e-01 8.28579545e-01 -5.57154179e-01
9.26938593e-01 1.21227175e-01 1.02717578e+00 1.61371171e+00
2.01059595e-01 1.06306517e+00 1.41105473e+00 -4.30103332e-01
1.03560477e-01 1.85475320e-01 7.28267282e-02 6.77534044e-01
-5.68886936e-01 1.85416758e-01 -1.16474867e+00 3.74972858e-02
6.29916966e-01 -2.10396215e-01 -9.49752107e-02 -2.24652380e-01
-8.40196908e-01 7.28556514e-01 1.86900541e-01 6.36989832e-01
-7.10213125e-01 3.74636739e-01 6.73349142e-01 7.28582084e-01
7.90434659e-01 1.36784688e-01 -3.83457035e-01 -7.05966055e-01
-9.39571381e-01 1.66040316e-01 6.69241846e-01 1.67455822e-01
2.98645914e-01 1.30442316e-02 4.01228629e-02 8.21242690e-01
3.55074942e-01 6.11255169e-02 5.18889248e-01 -1.00530601e+00
-2.16253668e-01 5.90119123e-01 -1.89955309e-01 -8.70646894e-01
-6.55298710e-01 1.39351249e-01 -6.34415448e-01 4.88480061e-01
5.71031511e-01 -3.29738230e-01 -8.04688573e-01 1.60701084e+00
5.86004317e-01 2.47697622e-01 -7.83694610e-02 8.82493138e-01
1.03731167e+00 3.96887183e-01 1.88975647e-01 -5.21315217e-01
1.44810212e+00 -6.36398494e-01 -9.70982611e-01 1.18324324e-01
4.05257910e-01 -6.04114115e-01 4.47553545e-01 8.14663708e-01
-9.38476980e-01 -3.85707021e-01 -6.48556948e-01 1.84119269e-01
-4.68228787e-01 1.96080923e-01 8.70467782e-01 5.63391328e-01
-1.13746238e+00 5.01172900e-01 -7.43944824e-01 -7.72616208e-01
6.45528913e-01 5.81950426e-01 -8.81503820e-01 2.84083009e-01
-1.04262269e+00 8.67083788e-01 -1.13812219e-02 1.93822488e-01
-6.96219206e-01 -3.00610602e-01 -7.56473839e-01 -1.78054914e-01
6.12541847e-02 -3.62174392e-01 1.31980526e+00 -1.76490986e+00
-1.84427810e+00 1.18888211e+00 -1.93001211e-01 -5.55327594e-01
5.67750335e-01 -2.58301198e-01 -3.44534010e-01 5.13759792e-01
-5.41179180e-01 1.03231120e+00 1.20391548e+00 -9.31611717e-01
-4.63939011e-01 -5.83898962e-01 -1.73622683e-01 5.12613319e-02
-4.35504138e-01 4.74355429e-01 -2.20658779e-01 -4.53450888e-01
-5.08656263e-01 -7.94098079e-01 4.18963917e-02 1.22803651e-01
3.15436691e-01 -4.47602123e-01 9.08481002e-01 -4.79218870e-01
1.00777078e+00 -2.20912600e+00 1.05352260e-01 1.34595186e-01
2.69916296e-01 4.36643690e-01 -3.56604047e-02 2.54140377e-01
-4.40582991e-01 -4.28959191e-01 1.51124895e-01 -5.19033313e-01
-1.54387811e-02 3.73835824e-02 -2.60367803e-02 6.73445582e-01
4.79236305e-01 9.16305780e-01 -5.87138414e-01 -6.08175516e-01
2.69676298e-01 8.49771082e-01 -2.26028353e-01 4.76113558e-01
-2.77287490e-03 4.56339478e-01 -3.57553929e-01 4.45161790e-01
1.89641491e-01 2.52041817e-01 -8.22604746e-02 3.21763270e-02
2.27890555e-02 -6.81629851e-02 -7.55340576e-01 1.63737261e+00
-2.03685433e-01 1.04433763e+00 1.68370992e-01 -8.70540619e-01
1.05628586e+00 9.52953517e-01 7.32186913e-01 -8.00153494e-01
5.76398075e-01 -1.27575651e-01 1.90576650e-02 -7.78350830e-01
1.23052210e-01 -4.08520520e-01 2.49863923e-01 4.54024911e-01
6.65176392e-01 1.26189053e-01 2.44247392e-01 -1.84858982e-02
1.15965497e+00 3.30428600e-01 4.11433965e-01 -8.75955373e-02
5.86206734e-01 -3.59157115e-01 1.96776316e-01 -4.71292585e-02
-8.17544520e-01 3.67930502e-01 8.93103004e-01 -9.41240072e-01
-5.65322101e-01 -4.18342292e-01 -6.65575266e-02 1.17817855e+00
-6.94411278e-01 -2.75331914e-01 -1.10794342e+00 -8.63219857e-01
-3.04481834e-01 1.31900206e-01 -1.18971634e+00 -6.30944818e-02
-1.62445247e-01 -5.32087743e-01 6.15382433e-01 2.59252518e-01
3.02147895e-01 -1.73131108e+00 -1.15519309e+00 8.32231715e-02
2.45822798e-02 -1.17101598e+00 1.49398491e-01 2.01458931e-01
-5.01028299e-01 -1.10354245e+00 -6.83786333e-01 -5.05284190e-01
3.48122925e-01 -2.79565006e-01 1.25820386e+00 6.64858520e-02
-6.43537521e-01 7.55647004e-01 -3.88368607e-01 -8.77154410e-01
-5.98250590e-02 -1.67286128e-01 1.02776431e-01 5.48579454e-01
7.66617954e-01 -5.99613011e-01 -3.65369111e-01 4.80179824e-02
-7.45403051e-01 -1.43459946e-01 3.51070255e-01 7.73888230e-01
6.83538243e-02 -3.82441908e-01 6.24802947e-01 -6.80544853e-01
5.06025612e-01 -2.98396200e-01 -3.47404450e-01 -8.14782754e-02
-1.83622450e-01 -1.39974877e-01 2.39811257e-01 -5.27797878e-01
-1.20944476e+00 5.77447355e-01 -2.26659745e-01 -6.78774834e-01
-7.61802375e-01 2.65184760e-01 1.91991284e-01 -1.07166559e-01
5.94170928e-01 -2.21973270e-01 1.55481189e-01 -1.26706660e-01
1.95811227e-01 6.60879016e-01 5.22954106e-01 -2.71449268e-01
1.65085748e-01 5.06689489e-01 1.09373860e-01 -1.11745811e+00
-4.87202883e-01 -6.16444707e-01 -9.90958929e-01 -9.42625523e-01
9.19145286e-01 -8.51560473e-01 -1.24403536e+00 7.21615374e-01
-1.25391948e+00 -3.18744838e-01 -1.80328369e-01 2.67144829e-01
-8.69433999e-01 -1.27081707e-01 -7.22401261e-01 -1.34915352e+00
-3.72739166e-01 -5.44877529e-01 1.15839839e+00 2.92292774e-01
-5.10143578e-01 -9.89203393e-01 5.22777200e-01 2.27428809e-01
2.23402172e-01 6.15726531e-01 2.99727857e-01 -5.71220338e-01
1.91637725e-01 -2.11448580e-01 9.10442770e-02 1.51634708e-01
-1.52380154e-01 3.15751493e-01 -1.58770239e+00 4.84174490e-02
1.35092318e-01 -9.25057411e-01 9.60998356e-01 2.65410811e-01
8.17309558e-01 1.72394201e-01 5.58490083e-02 2.66469270e-01
9.48597789e-01 2.16773897e-01 8.20527196e-01 1.74263269e-01
3.67721587e-01 1.30111706e+00 7.37735331e-01 6.28780007e-01
1.87762275e-01 5.87789059e-01 4.72539634e-01 -1.89937770e-01
1.23054847e-01 2.15373009e-01 6.50979400e-01 5.69307029e-01
-4.12795365e-01 -1.29450792e-02 -8.22461247e-01 5.59885502e-01
-1.99394035e+00 -1.38485265e+00 -9.18003023e-02 1.63361073e+00
7.32006669e-01 1.18314568e-02 3.95341694e-01 3.86210740e-01
3.70312214e-01 1.56520739e-01 -1.22477695e-01 -8.25746298e-01
9.99286249e-02 3.82543325e-01 -3.98062915e-01 2.73691267e-01
-1.30133224e+00 9.88224685e-01 6.30599117e+00 6.96530342e-01
-1.36591184e+00 -1.64568141e-01 8.61878037e-01 -5.19641817e-01
4.66316015e-01 -7.61614978e-01 -3.84584934e-01 2.26594761e-01
1.39403391e+00 1.27003849e-01 2.95488179e-01 5.88897943e-01
3.02840978e-01 -3.55464458e-01 -1.11285329e+00 1.28239095e+00
4.23369825e-01 -8.69440258e-01 -3.69068176e-01 2.71232426e-02
4.23455745e-01 -2.34662414e-01 3.58611927e-03 2.14797825e-01
-6.93752393e-02 -1.41198957e+00 4.37887549e-01 9.07764196e-01
6.07563376e-01 -9.74855542e-01 9.17752624e-01 1.97415158e-01
-9.64775681e-01 1.13070324e-01 1.37762576e-02 -3.22479635e-01
-9.31307673e-02 4.34508115e-01 -8.46199632e-01 1.82058349e-01
9.86975610e-01 9.55022633e-01 -5.87803960e-01 6.69912517e-01
-1.08287990e-01 5.48480392e-01 -3.06130826e-01 -1.60349116e-01
1.29344061e-01 -1.07902527e-01 1.75238088e-01 1.40396261e+00
3.53848226e-02 3.88973653e-01 -4.39260304e-01 4.03573811e-01
-2.79300008e-02 1.66294754e-01 -8.94726217e-01 -2.76683450e-01
-3.52058440e-01 1.78199244e+00 -7.33949244e-01 -2.33924821e-01
-4.52081412e-01 1.12448108e+00 6.06945217e-01 2.14158311e-01
-6.54445648e-01 -5.08855954e-02 8.11890006e-01 -1.14325218e-01
3.30984145e-01 8.87978002e-02 2.04868868e-01 -9.64470506e-01
-8.56953189e-02 -1.15206361e+00 2.92756766e-01 -9.23733532e-01
-1.20752680e+00 9.18133199e-01 -2.36767679e-01 -8.96354973e-01
-7.35871792e-01 -9.60524917e-01 -5.73586166e-01 4.60020244e-01
-1.24045980e+00 -1.28456521e+00 -3.72202635e-01 8.71715963e-01
5.54699123e-01 -1.55746371e-01 1.05555594e+00 2.25715727e-01
-4.91624087e-01 3.26910555e-01 -5.62539577e-01 1.32469311e-01
1.05793452e+00 -1.19714510e+00 3.71210016e-02 5.05591571e-01
3.80572438e-01 1.53844535e-01 7.33674884e-01 -9.25091505e-02
-1.10145807e+00 -7.03953683e-01 1.05583405e+00 -6.40572369e-01
5.63914418e-01 -4.28210467e-01 -8.36708665e-01 5.55293739e-01
6.49690926e-01 1.73123196e-01 8.86754990e-01 1.98411252e-02
-1.83601752e-01 7.88306296e-02 -1.16451609e+00 4.46230531e-01
7.67169714e-01 -5.65950155e-01 -5.25700688e-01 -1.44197136e-01
1.14579789e-01 -2.51908422e-01 -8.86055589e-01 3.22582901e-01
9.46360707e-01 -1.52965403e+00 5.45718610e-01 -7.25857019e-01
6.10763013e-01 2.13526279e-01 1.84108675e-01 -1.35223913e+00
1.48804396e-01 -8.25467288e-01 -3.40351522e-01 1.26462197e+00
5.05321138e-02 -2.85898566e-01 9.15551960e-01 8.10689390e-01
4.54026550e-01 -1.04012954e+00 -1.05111861e+00 -6.77527487e-02
-2.51305640e-01 -6.67697489e-01 2.24732697e-01 9.63557124e-01
2.65834808e-01 5.36676705e-01 -4.59182739e-01 -5.73967397e-01
2.71460295e-01 -2.57909030e-01 8.40143740e-01 -1.54246366e+00
-3.25203761e-02 -4.60078180e-01 -8.76189590e-01 -2.87658811e-01
5.62620580e-01 -2.13067487e-01 -1.00038871e-01 -1.00957680e+00
1.55816615e-01 3.72014344e-01 -2.30290413e-01 6.55565321e-01
-1.46080293e-02 5.86884499e-01 1.12232707e-01 -3.58608752e-01
-7.66784668e-01 7.93821275e-01 9.30907130e-01 1.71952993e-01
-3.01060695e-02 -1.65729463e-01 -2.03721359e-01 9.70987439e-01
6.70774400e-01 -2.50840485e-01 -1.02165222e-01 1.06430046e-01
1.05333827e-01 2.15261281e-02 3.23264599e-01 -8.74340951e-01
1.83306515e-01 9.86062065e-02 6.92812800e-01 -2.12791309e-01
7.24929214e-01 -9.74187732e-01 -1.11570001e-01 2.91945755e-01
-3.39476019e-01 1.22964829e-01 4.05024379e-01 1.33591697e-01
-5.13370514e-01 7.07452325e-03 8.13624859e-01 -8.03355407e-03
-6.99573874e-01 2.91962475e-01 -6.87241554e-01 -3.25238854e-01
1.19046009e+00 -6.52721003e-02 -1.84905663e-01 -7.64714301e-01
-9.16136622e-01 9.70854759e-02 2.25518793e-01 4.63384092e-01
6.10936105e-01 -1.28714323e+00 -5.60991228e-01 3.81405532e-01
1.85592785e-01 -6.28604472e-01 1.79122642e-01 9.27429199e-01
-1.82116330e-01 3.77106607e-01 -6.49105489e-01 -5.42462885e-01
-1.98884916e+00 3.36857677e-01 5.17275810e-01 -2.53541589e-01
-9.90595669e-02 8.61410499e-01 4.77879941e-02 -2.95186281e-01
2.98981190e-01 2.27538124e-02 -8.38595986e-01 6.67910635e-01
7.41966188e-01 2.22529203e-01 6.47844374e-02 -9.27833498e-01
-3.66129965e-01 4.53108311e-01 1.72864094e-01 -4.81978208e-01
1.57230973e+00 -4.01716009e-02 -2.19819501e-01 8.16428840e-01
1.18587232e+00 -3.64612460e-01 -1.30831861e+00 2.84432396e-02
9.62217450e-02 -2.20579311e-01 3.32040936e-02 -6.18734181e-01
-1.05941546e+00 1.20534778e+00 9.30051208e-01 2.77781516e-01
1.36700702e+00 -1.21110104e-01 3.23858261e-01 3.07647496e-01
2.50028968e-01 -1.17649376e+00 4.02295113e-01 5.53551793e-01
8.42230558e-01 -1.43957019e+00 -9.75131467e-02 -1.97611138e-01
-9.32385564e-01 1.26665854e+00 7.16462493e-01 -1.78167149e-01
6.84883535e-01 2.55416662e-01 5.10027587e-01 -6.16647840e-01
-1.49016118e+00 -4.61896032e-01 4.35655206e-01 5.65527201e-01
9.44420695e-01 -2.38170996e-01 -2.39900738e-01 5.89994907e-01
-1.59146801e-01 3.90323848e-01 3.09856713e-01 7.64448404e-01
-1.68128774e-01 -9.62220192e-01 -2.96934336e-01 2.56782800e-01
-9.07752693e-01 2.42348477e-01 -9.67429876e-01 6.85215354e-01
7.67111480e-02 1.00355482e+00 1.44111633e-01 -3.44349921e-01
2.70347595e-01 5.88602901e-01 7.19674647e-01 -2.53237218e-01
-1.01841927e+00 1.37874186e-01 2.86508977e-01 -9.45256650e-01
-1.00461590e+00 -8.89511883e-01 -1.14987326e+00 -1.78524747e-01
-2.10717805e-02 -9.87044424e-02 6.72664881e-01 1.03074884e+00
2.35955939e-01 5.22572577e-01 5.51923990e-01 -1.15623200e+00
1.83926970e-01 -1.06174338e+00 -5.89314699e-01 6.17962658e-01
4.80706453e-01 -7.47065306e-01 -2.75615096e-01 3.63615423e-01] | [13.543023109436035, 2.0682997703552246] |
3760ced2-e6cc-48f5-9c5c-72b4fb81fb32 | towards-context-aware-code-comment-generation | null | null | https://aclanthology.org/2020.findings-emnlp.350 | https://aclanthology.org/2020.findings-emnlp.350.pdf | Towards Context-Aware Code Comment Generation | Code comments are vital for software maintenance and comprehension, but many software projects suffer from the lack of meaningful and up-to-date comments in practice. This paper presents a novel approach to automatically generate code comments at a function level by targeting object-oriented programming languages. Unlike prior work that only uses information locally available within the target function, our approach leverages broader contextual information by considering all other functions of the same class. To propagate and integrate information beyond the scope of the target function, we design a novel learning framework based on the bidirectional gated recurrent unit and a graph attention network with a pointer mechanism. We apply our approach to produce code comments for Java methods and compare it against four strong baseline methods. Experimental results show that our approach outperforms most methods by a large margin and achieves a comparable result with the state-of-the-art method. | ['Dongyan Zhao', 'Yansong Feng', 'Zheng Wang', 'Quzhe Huang', 'Xiaohan Yu'] | 2020-11-01 | null | null | null | findings-of-the-association-for-computational | ['code-comment-generation', 'comment-generation'] | ['computer-code', 'natural-language-processing'] | [ 2.27442473e-01 4.11579549e-01 -2.07138419e-01 -4.00702953e-01
-8.53926182e-01 -4.96485025e-01 3.87228370e-01 4.06799287e-01
2.06611499e-01 3.81500989e-01 4.62076187e-01 -6.79330289e-01
2.84469604e-01 -9.33811724e-01 -7.98312426e-01 -2.17591785e-02
3.71126086e-02 -2.44594976e-01 3.56925189e-01 -2.49472082e-01
5.93168020e-01 -2.39388585e-01 -1.28239715e+00 6.87347651e-01
8.95600796e-01 5.02724685e-02 3.06778342e-01 7.62673199e-01
-4.32841569e-01 1.35223496e+00 -7.98528314e-01 -7.15300083e-01
-1.55451506e-01 -7.71541476e-01 -1.02041852e+00 -1.60981447e-01
5.46492755e-01 -2.90156752e-01 -1.58431847e-02 9.45856392e-01
4.76721019e-01 -1.95259720e-01 2.58367926e-01 -1.26385713e+00
-1.30819821e+00 1.36700439e+00 -6.63123786e-01 -1.13825321e-01
4.46747392e-01 -1.34999573e-01 1.35804272e+00 -9.92794991e-01
8.31314802e-01 9.21558857e-01 8.84335697e-01 6.65358245e-01
-1.32059586e+00 -5.46795607e-01 4.39068079e-01 -7.77991191e-02
-8.48473787e-01 -9.81414840e-02 8.74146461e-01 -8.49937558e-01
1.45315862e+00 2.27155108e-02 5.34908056e-01 8.39545012e-01
3.37304473e-01 4.85422492e-01 4.83876288e-01 -6.49383962e-01
2.04988122e-02 -4.39008959e-02 4.43473160e-01 1.11217225e+00
-8.24971218e-03 -4.83741462e-01 -2.33226359e-01 -4.75655168e-01
3.63269299e-01 3.66858952e-02 -3.43511134e-01 -3.10097128e-01
-1.10271347e+00 8.92870605e-01 5.25401711e-01 4.76926744e-01
-1.76277429e-01 8.52627397e-01 3.85360718e-01 5.02801657e-01
6.01436973e-01 2.61432350e-01 -4.98682529e-01 -4.63658303e-01
-9.01594400e-01 1.74279824e-01 1.14329112e+00 1.41853189e+00
1.04470122e+00 2.50519782e-01 -3.66119832e-01 6.80392623e-01
5.41239738e-01 1.81825544e-05 3.71017098e-01 -7.22602427e-01
6.55151606e-01 1.24671757e+00 -1.11024737e-01 -8.58063817e-01
-3.21937427e-02 -5.56479454e-01 -1.79850966e-01 4.22451496e-01
2.59564575e-02 -3.01792830e-01 -4.97281313e-01 1.48146594e+00
6.97578415e-02 -5.48461340e-02 -2.07755461e-01 3.30892146e-01
1.07454121e+00 7.15684831e-01 -2.33665153e-01 8.20975304e-02
7.76619196e-01 -1.50244963e+00 -4.74853247e-01 -3.42263281e-01
1.02104723e+00 -6.82031512e-01 1.20969284e+00 -3.71039822e-03
-1.09788465e+00 -5.11879802e-01 -9.80808914e-01 -2.10195243e-01
-1.35012493e-01 2.35870898e-01 7.50492930e-01 7.01067984e-01
-1.62026107e+00 4.98462200e-01 -7.80521750e-01 -3.13131064e-01
1.79145575e-01 1.31781310e-01 -1.71903118e-01 3.42061161e-03
-5.07286310e-01 5.52490294e-01 5.73187554e-03 -5.43458387e-02
-7.85363734e-01 -1.00240541e+00 -1.08847010e+00 2.82320857e-01
1.93725541e-01 -8.26341391e-01 1.58454871e+00 -1.18259323e+00
-1.39236987e+00 6.21120512e-01 -2.52493292e-01 -2.90245831e-01
2.95673639e-01 -2.72898793e-01 1.26057982e-01 -2.63088554e-01
1.75278291e-01 4.48858619e-01 6.41850412e-01 -1.21169698e+00
-3.34791720e-01 -9.82120261e-03 5.01553237e-01 -2.92148888e-01
-4.56198305e-01 4.66778278e-01 -5.46381474e-01 -6.46578312e-01
-4.94046360e-01 -8.60062599e-01 -2.61853993e-01 -8.87532160e-02
-3.88514340e-01 -3.82421792e-01 5.55924833e-01 -8.54049265e-01
1.69879556e+00 -1.97582614e+00 2.78390467e-01 -7.56401643e-02
4.01149541e-01 -9.28914174e-02 -2.96130985e-01 9.60726380e-01
-2.69161552e-01 4.50945437e-01 -4.05942142e-01 -4.02607508e-02
2.59622931e-01 -2.59350151e-01 -3.58283430e-01 3.06524634e-01
3.38107824e-01 1.12305784e+00 -1.09893024e+00 -2.12391272e-01
-3.43593597e-01 5.57687163e-01 -1.19720936e+00 3.83933723e-01
-4.65604067e-01 -2.35779479e-01 -3.85982305e-01 4.79455680e-01
2.35294148e-01 -5.95065355e-01 2.97443986e-01 4.45082635e-01
-2.35576078e-01 6.26426637e-01 -7.83403337e-01 1.90978396e+00
-9.23540950e-01 6.71442032e-01 -3.13156880e-02 -7.14390874e-01
1.14493024e+00 3.25381458e-01 -7.02662468e-02 -3.46338093e-01
-4.16432917e-01 1.06232002e-01 7.77420625e-02 -6.59162760e-01
4.95952755e-01 3.26387376e-01 -1.53252274e-01 8.71372402e-01
-3.46702896e-02 -1.16690919e-01 4.23493773e-01 5.84509611e-01
1.69616520e+00 6.87927306e-01 4.17431653e-01 -2.02854529e-01
4.71347421e-01 -2.05572292e-01 4.10735995e-01 8.45976293e-01
3.72843951e-01 5.47487915e-01 1.00440824e+00 -2.45845169e-01
-8.99467051e-01 -6.60872877e-01 3.62248480e-01 1.59814858e+00
-3.39263707e-01 -9.84930754e-01 -8.71348321e-01 -1.19387734e+00
-1.27230242e-01 8.34885716e-01 -9.14138019e-01 1.37054073e-02
-8.26006055e-01 -3.97354484e-01 5.19878566e-01 9.27478969e-01
7.35942349e-02 -1.28177750e+00 -6.45291984e-01 4.10378218e-01
-8.69399775e-03 -4.84695315e-01 -6.86047137e-01 -1.49441093e-01
-6.43125594e-01 -9.81558800e-01 -7.68218637e-01 -1.11450052e+00
9.82781649e-01 2.02044994e-02 1.58731520e+00 8.71880233e-01
-1.35094136e-01 3.26620907e-01 -4.67831612e-01 -1.59678951e-01
-9.12174702e-01 3.83457124e-01 -1.04296005e+00 -3.26074779e-01
1.69827640e-01 -4.60763216e-01 -2.61975795e-01 -1.99849918e-01
-7.46115863e-01 1.74340025e-01 5.12222648e-01 8.93037140e-01
-2.02747852e-01 -4.77161795e-01 7.82607257e-01 -1.39806128e+00
8.00061226e-01 -6.85042441e-01 -6.34915113e-01 3.06620061e-01
-5.90462863e-01 9.98216122e-02 6.50505543e-01 -1.17812216e-01
-1.27053034e+00 -9.62322056e-02 -1.90490544e-01 3.10968280e-01
1.23572409e-01 9.09052074e-01 1.54040173e-01 1.09500058e-01
7.52003610e-01 -9.19799730e-02 -3.01953465e-01 -2.99794495e-01
4.65479970e-01 6.21030509e-01 2.52310008e-01 -6.11873984e-01
8.91553938e-01 9.10549983e-02 -6.40848458e-01 -4.08253580e-01
-2.65973896e-01 -3.43198657e-01 -4.21959758e-01 -2.36943156e-01
4.79510128e-01 -8.25253308e-01 -3.80071908e-01 2.47780517e-01
-1.73607326e+00 -5.92245519e-01 -1.00307225e-03 2.21095812e-02
-4.55990881e-01 3.26671273e-01 -1.03332222e+00 -5.70801497e-01
-4.18029249e-01 -1.28763974e+00 1.09983552e+00 -6.86462075e-02
-4.95628327e-01 -1.09182441e+00 3.45430732e-01 1.39443174e-01
6.98758960e-01 2.09262952e-01 1.17880702e+00 -4.33727413e-01
-9.42951560e-01 -1.06201358e-01 -2.67453015e-01 1.70355633e-01
4.33776826e-02 5.13147831e-01 -5.92441738e-01 -2.06399307e-01
-3.27257633e-01 -2.56320983e-01 7.57534385e-01 -1.27465144e-01
8.64938021e-01 -5.11998653e-01 -2.63188720e-01 3.06858033e-01
1.47150338e+00 -4.50016521e-02 6.70040369e-01 2.72358716e-01
9.15708780e-01 7.54799843e-01 1.09539725e-01 4.20465559e-01
7.33403504e-01 3.88601214e-01 5.98668754e-01 8.53148848e-03
-2.10021347e-01 -2.56193161e-01 7.34535813e-01 1.26275671e+00
2.10843712e-01 -2.33670756e-01 -1.23894835e+00 1.05538249e+00
-2.01621389e+00 -7.90269077e-01 -5.33604085e-01 1.96197760e+00
1.07887316e+00 -1.24063522e-01 -1.32902265e-01 -1.92866147e-01
7.38127351e-01 2.32063010e-02 -1.94584206e-01 -6.00881398e-01
5.27980328e-01 2.14888081e-01 2.59480551e-02 4.70285147e-01
-6.30079925e-01 7.37198532e-01 6.59902477e+00 2.15825245e-01
-7.95487583e-01 1.55214876e-01 1.99914590e-01 3.11500847e-01
-9.32875812e-01 3.76906365e-01 -6.50573790e-01 2.48875886e-01
1.05854499e+00 -4.01929498e-01 3.88051808e-01 1.17509973e+00
-5.81464060e-02 2.25492671e-01 -1.40557683e+00 2.48124436e-01
4.10954654e-01 -1.45880568e+00 -2.03423634e-01 -2.10959405e-01
1.05709076e+00 1.64486930e-01 -3.19619715e-01 4.96723384e-01
9.03753638e-01 -8.82148981e-01 7.84747005e-01 4.33085114e-01
4.30157751e-01 -4.64981437e-01 6.16439104e-01 2.07100019e-01
-1.11148834e+00 2.27100831e-02 -1.44496918e-01 -2.15289697e-01
-1.21682249e-01 5.51878989e-01 -8.90056789e-01 4.73358363e-01
6.09265983e-01 1.07753086e+00 -9.57981944e-01 1.05508268e+00
-5.90519130e-01 8.16088736e-01 3.11967641e-01 -4.80776340e-01
8.94987285e-02 9.58186239e-02 1.55150235e-01 1.71193337e+00
5.02351344e-01 -5.93117893e-01 2.64026105e-01 1.39029491e+00
-4.83446836e-01 4.24972713e-01 -8.16593468e-01 -2.77411014e-01
1.58788115e-01 1.24170399e+00 -6.37979448e-01 -4.33216691e-01
-1.08260357e+00 7.48984516e-01 5.12487829e-01 5.95357239e-01
-8.65408719e-01 -1.00629675e+00 1.65133655e-01 8.43421668e-02
6.42165363e-01 -4.22261506e-02 -8.49538296e-02 -1.11834180e+00
4.78747189e-01 -9.50033069e-01 2.23856926e-01 -9.15627301e-01
-1.00458634e+00 6.94411278e-01 -2.08761334e-01 -9.37238932e-01
-6.43654287e-01 -2.09334955e-01 -9.45041955e-01 8.20081294e-01
-1.40834212e+00 -1.38504136e+00 -3.74059409e-01 5.03276521e-03
5.60349762e-01 -9.77178663e-02 7.78591692e-01 2.48638302e-01
-2.15604976e-01 4.97564048e-01 -2.31725991e-01 3.38426709e-01
5.51724434e-01 -1.43366122e+00 1.03053379e+00 1.18306804e+00
5.61197959e-02 1.26468146e+00 3.83619845e-01 -8.05604637e-01
-1.33760476e+00 -1.15646851e+00 1.11496973e+00 -4.76266652e-01
1.00391495e+00 -5.47593594e-01 -1.09819686e+00 1.15349376e+00
8.06514800e-01 -1.13278866e-01 6.37308478e-01 1.24589071e-01
-7.57784009e-01 1.92783371e-01 -4.58618969e-01 5.98828018e-01
9.65747118e-01 -7.90712059e-01 -6.61323786e-01 3.26118827e-01
1.11578441e+00 -3.93632144e-01 -7.42926359e-01 1.37197897e-01
3.79802555e-01 -1.11634505e+00 4.49748307e-01 -4.58430469e-01
1.12464917e+00 -4.21548218e-01 1.85364440e-01 -1.37183058e+00
-2.91177332e-01 -7.67317593e-01 -3.68072651e-02 1.67710614e+00
7.68396914e-01 -6.05856121e-01 6.16261661e-01 4.08198208e-01
-5.07957041e-01 -5.98026991e-01 -2.60375589e-01 -3.47522348e-01
2.28637502e-01 -3.50361437e-01 5.60388625e-01 1.01895833e+00
3.84798110e-01 4.06228483e-01 -1.22021079e-01 -4.12168913e-02
1.13515198e-01 5.56224048e-01 9.02590930e-01 -9.93565798e-01
-7.98940897e-01 -5.47923863e-01 -2.51513183e-01 -9.60445106e-01
6.80889904e-01 -1.32920289e+00 3.07247400e-01 -2.09431791e+00
4.10450727e-01 -9.22828019e-02 5.95107041e-02 7.59972572e-01
-3.30820143e-01 1.88550159e-01 8.56346264e-02 -1.40850976e-01
-4.95544523e-01 1.95872709e-01 6.31300807e-01 -5.19329846e-01
-4.52288873e-02 -1.49672866e-01 -9.57517564e-01 6.05317056e-01
6.10860467e-01 -8.26976418e-01 -3.65612447e-01 -7.40842104e-01
9.57997084e-01 5.79322018e-02 2.50253022e-01 -6.36898160e-01
3.61932307e-01 1.13673329e-01 -2.06624836e-01 -2.37426579e-01
-3.95094037e-01 -4.48268324e-01 3.93136442e-02 4.00384456e-01
-8.29440296e-01 4.17006284e-01 8.24422762e-02 5.35021305e-01
-2.02983171e-01 -7.57392466e-01 2.81049669e-01 -2.28580430e-01
-5.32294989e-01 -1.46594137e-01 -5.83728850e-01 1.67608693e-01
8.67725194e-01 2.11176887e-01 -8.40979755e-01 -3.07216525e-01
-1.47636265e-01 1.40510261e-01 7.98578978e-01 9.12850201e-01
5.56842983e-01 -1.03684843e+00 -8.60481560e-01 9.49665755e-02
4.87454206e-01 -2.59610564e-01 -2.45896339e-01 5.08604467e-01
-7.38014817e-01 8.07748362e-02 1.24843553e-01 -3.05120766e-01
-1.30851245e+00 4.07981157e-01 1.42775118e-01 -1.99300960e-01
-6.63079977e-01 7.24718392e-01 1.44389376e-01 -9.68658268e-01
1.03199609e-01 -5.49033463e-01 -1.31742194e-01 -1.53221995e-01
5.85857630e-01 1.59129068e-01 2.86809742e-01 -3.78709704e-01
-2.33123198e-01 4.76682872e-01 -1.03624322e-01 1.76399961e-01
1.60010195e+00 3.24660808e-01 -7.92821705e-01 7.00618327e-01
1.18838680e+00 4.73224759e-01 -1.02825344e+00 -7.70897716e-02
3.37153167e-01 -4.64523673e-01 -1.90664992e-01 -7.37652540e-01
-1.12876415e+00 9.04746532e-01 -1.23027399e-01 4.04668301e-01
6.60100639e-01 1.53372765e-01 3.47287327e-01 4.83092636e-01
3.14476758e-01 -7.37945795e-01 3.62203389e-01 5.78282416e-01
9.89040971e-01 -9.20637965e-01 -2.40667775e-01 -5.44517219e-01
-4.31838632e-01 1.37959373e+00 7.49483407e-01 -1.00982554e-01
4.31701392e-01 7.62140095e-01 1.58858314e-01 -1.78547129e-01
-1.35622442e+00 1.78608537e-01 1.40714094e-01 4.71057564e-01
1.28832006e+00 -4.81343508e-01 -3.52727175e-01 3.94661427e-01
1.31088212e-01 1.09027304e-01 1.23933923e+00 1.42370057e+00
-2.01588348e-01 -1.45508075e+00 2.51214635e-02 4.55274761e-01
-7.64382899e-01 -6.99934065e-01 -4.50099468e-01 7.13233888e-01
-3.39365602e-01 9.00075555e-01 -1.04214475e-01 -1.14302650e-01
2.02987775e-01 2.95244485e-01 2.52193242e-01 -1.31052053e+00
-1.35978198e+00 -1.24737613e-01 2.52085090e-01 -4.63544041e-01
-4.76234704e-01 -5.22200584e-01 -1.46756887e+00 2.08289735e-02
-4.40333486e-01 1.71186537e-01 7.17883170e-01 5.13930321e-01
6.75534546e-01 9.67142403e-01 3.33205998e-01 -5.03764033e-01
-4.43456292e-01 -8.63167048e-01 -1.50096621e-02 2.94269770e-01
4.56076026e-01 -2.17667017e-02 -1.93367183e-01 5.32454729e-01] | [7.607490539550781, 7.934596538543701] |
002c4fb5-f8d5-4f05-9c79-6970a9e167bf | greedy-subspace-clustering | 1410.8864 | null | http://arxiv.org/abs/1410.8864v1 | http://arxiv.org/pdf/1410.8864v1.pdf | Greedy Subspace Clustering | We consider the problem of subspace clustering: given points that lie on or
near the union of many low-dimensional linear subspaces, recover the subspaces.
To this end, one first identifies sets of points close to the same subspace and
uses the sets to estimate the subspaces. As the geometric structure of the
clusters (linear subspaces) forbids proper performance of general distance
based approaches such as K-means, many model-specific methods have been
proposed. In this paper, we provide new simple and efficient algorithms for
this problem. Our statistical analysis shows that the algorithms are guaranteed
exact (perfect) clustering performance under certain conditions on the number
of points and the affinity between subspaces. These conditions are weaker than
those considered in the standard statistical literature. Experimental results
on synthetic data generated from the standard unions of subspaces model
demonstrate our theory. We also show that our algorithm performs competitively
against state-of-the-art algorithms on real-world applications such as motion
segmentation and face clustering, with much simpler implementation and lower
computational cost. | ['Constantine Caramanis', 'Sujay Sanghavi', 'Dohyung Park'] | 2014-10-31 | greedy-subspace-clustering-1 | http://papers.nips.cc/paper/5308-greedy-subspace-clustering | http://papers.nips.cc/paper/5308-greedy-subspace-clustering.pdf | neurips-2014-12 | ['face-clustering'] | ['computer-vision'] | [-1.68860823e-01 -2.10594475e-01 -3.70216131e-01 -2.34935552e-01
-7.48954952e-01 -8.93918514e-01 3.53619188e-01 -1.32761016e-01
-3.31391662e-01 3.12729299e-01 -3.22229378e-02 -1.06031969e-01
-3.42491388e-01 -2.12145716e-01 -5.54589450e-01 -1.06669581e+00
-2.82011807e-01 1.02634013e+00 3.46891671e-01 2.31312394e-01
2.44609445e-01 7.62927711e-01 -1.54150581e+00 -7.37116858e-02
7.05106080e-01 3.97114396e-01 -2.73233235e-01 6.85603201e-01
-7.81383365e-02 -2.21685752e-01 -2.15692297e-01 3.26074753e-03
4.99703437e-01 -4.99487251e-01 -9.24706876e-01 7.19480157e-01
5.45577347e-01 2.78206766e-02 -1.82831347e-01 1.09630287e+00
2.75023371e-01 2.86987543e-01 9.71108735e-01 -1.42952740e+00
-5.71895950e-02 2.86964595e-01 -1.04402578e+00 4.66387309e-02
2.82993734e-01 -2.96082824e-01 6.98327541e-01 -9.12607968e-01
6.00748122e-01 1.33722055e+00 7.56188035e-01 7.11806059e-01
-1.62583590e+00 -3.89660478e-01 -7.73129463e-02 -2.47522146e-02
-1.76649380e+00 -6.79285765e-01 5.55420697e-01 -5.13878405e-01
3.75776529e-01 5.56890666e-01 3.70413750e-01 5.43683529e-01
-4.30824727e-01 7.09551930e-01 9.10437346e-01 -4.92162108e-01
5.03553212e-01 1.84637144e-01 3.88747424e-01 5.29822767e-01
3.50048661e-01 -3.22140574e-01 -1.93399802e-01 -6.03346646e-01
8.70626271e-01 -1.29523367e-01 -2.92185813e-01 -1.29576266e+00
-1.21366072e+00 6.84894621e-01 7.28871021e-03 5.93945622e-01
-3.71327624e-02 -3.91778797e-02 2.24952605e-02 -1.43632427e-01
3.47548097e-01 6.04529940e-02 -1.47337377e-01 1.94651023e-01
-1.46582758e+00 2.14893073e-01 9.64717507e-01 1.18552125e+00
6.63958848e-01 -4.43810195e-01 3.67324680e-01 7.45869040e-01
3.00022274e-01 3.97120982e-01 1.78421438e-01 -1.46094739e+00
2.07691416e-01 3.14218193e-01 1.32515341e-01 -1.09879899e+00
-3.69183213e-01 -1.87344417e-01 -8.49686205e-01 7.55663170e-03
8.95126164e-01 1.57807574e-01 -7.84380078e-01 1.67884254e+00
6.54117227e-01 5.06202281e-01 -8.73989891e-03 9.76677835e-01
1.40592113e-01 5.52560866e-01 -1.70064777e-01 -7.57539034e-01
9.29590583e-01 -5.37490666e-01 -5.26048601e-01 1.57913268e-01
6.80407405e-01 -7.89375246e-01 6.90974236e-01 3.72006059e-01
-1.09474432e+00 -4.13391858e-01 -7.83921778e-01 2.72364765e-01
-6.47756159e-02 1.23949885e-01 5.16050160e-01 9.34458554e-01
-1.25036609e+00 7.63085663e-01 -1.11587584e+00 -9.01982546e-01
2.87905037e-01 7.29800820e-01 -4.32417452e-01 -3.80335338e-02
-2.52720594e-01 2.50002384e-01 3.67107272e-01 7.30390847e-02
-4.23804641e-01 -4.03066367e-01 -6.01166308e-01 -1.96603328e-01
2.62274027e-01 -4.49813545e-01 9.27219331e-01 -9.16816711e-01
-9.12934840e-01 1.00433302e+00 -5.65390289e-01 -1.23423390e-01
4.77464676e-01 -1.83098942e-01 -2.64359653e-01 3.42042208e-01
1.61590427e-01 6.26842678e-01 8.02910149e-01 -1.50556409e+00
-4.36915666e-01 -8.14477861e-01 -5.30290663e-01 2.72016972e-01
-5.04284739e-01 1.87777892e-01 -1.02258122e+00 -3.01764905e-01
7.38259315e-01 -1.33996880e+00 -5.41066051e-01 -1.04951374e-01
-6.05531394e-01 -3.14268947e-01 1.04692602e+00 -3.31967920e-01
1.25682056e+00 -2.28420591e+00 4.48700070e-01 8.21078300e-01
2.21423000e-01 7.87053108e-02 1.70690536e-01 2.35471129e-01
-2.83862025e-01 7.20077157e-02 -6.25114202e-01 -2.57482648e-01
-1.07931957e-01 2.24527493e-01 -7.07183927e-02 1.21356714e+00
-4.04405296e-01 3.12945753e-01 -7.14944899e-01 -8.83187234e-01
1.90745145e-01 2.62003809e-01 -5.72990656e-01 -6.17532693e-02
1.74234614e-01 3.75924826e-01 -4.50542450e-01 2.99384207e-01
8.46051931e-01 -1.73392281e-01 3.59558672e-01 -8.34816769e-02
2.85161436e-02 -4.64970380e-01 -1.79680300e+00 1.64616287e+00
4.11404043e-01 4.03288305e-01 5.51211476e-01 -1.00314462e+00
5.36215067e-01 2.87640542e-01 1.06014669e+00 2.80547649e-01
-9.08544511e-02 2.15549782e-01 -1.16970547e-01 -1.59148082e-01
1.87671512e-01 -7.49651417e-02 1.94276366e-02 4.89401728e-01
8.76346678e-02 -4.27917577e-02 2.59879291e-01 4.05863136e-01
8.05223644e-01 2.78465282e-02 2.20097691e-01 -8.19634259e-01
6.05274141e-01 1.19445428e-01 6.12919688e-01 4.61684436e-01
-3.77166301e-01 7.85503864e-01 2.22523525e-01 -5.25383726e-02
-1.17480803e+00 -1.17801571e+00 -3.58077854e-01 8.11110020e-01
4.68814939e-01 -5.08314252e-01 -1.46310186e+00 -4.68347639e-01
8.72799475e-03 3.87659252e-01 -4.71700191e-01 2.80326068e-01
-5.67339242e-01 -8.39910924e-01 2.74900258e-01 5.04218996e-01
1.26224160e-01 -4.50409621e-01 -2.18414500e-01 -2.60637514e-02
-9.38450545e-03 -1.09895480e+00 -7.14234591e-01 -1.76561475e-01
-1.26087952e+00 -1.23961127e+00 -7.57408798e-01 -9.43737686e-01
1.02274477e+00 6.52002096e-01 8.23969543e-01 -6.26200661e-02
-2.30672538e-01 7.79532790e-01 4.21620160e-02 1.72862738e-01
-2.37227142e-01 -8.34245682e-02 4.87367660e-01 1.44915894e-01
7.14162469e-01 -4.27996218e-01 -3.88433635e-01 6.97573364e-01
-7.89308906e-01 -3.73750806e-01 2.89446920e-01 4.54883009e-01
7.85075784e-01 4.04562116e-01 -2.90111937e-02 -8.17878246e-01
3.74062657e-01 -4.92751688e-01 -6.18346334e-01 1.85744792e-01
-3.59283984e-01 1.27398372e-01 3.72627139e-01 -2.64839351e-01
-8.24198604e-01 7.94347703e-01 4.07569289e-01 -7.47247756e-01
-6.65704548e-01 -7.53467754e-02 -4.02763218e-01 5.19507900e-02
5.82957149e-01 7.67240599e-02 3.48897874e-02 -5.78355908e-01
6.11022174e-01 7.51314640e-01 8.12068701e-01 -6.58492982e-01
1.05014455e+00 9.26422954e-01 1.49844289e-01 -1.25594878e+00
-3.08225393e-01 -1.20606971e+00 -1.42494452e+00 -1.69217333e-01
9.16702271e-01 -6.77432239e-01 -5.08876443e-01 1.00911997e-01
-9.98865306e-01 4.66184020e-02 -5.87121099e-02 4.95750278e-01
-1.02432549e+00 7.99870193e-01 -4.65392679e-01 -1.02376568e+00
2.15617940e-01 -9.60046887e-01 1.11247540e+00 8.73479247e-02
-4.21357363e-01 -1.05805874e+00 1.98213592e-01 3.76027077e-01
-2.95727372e-01 2.63357192e-01 6.50901198e-01 -6.54378712e-01
-3.87166500e-01 -2.38515660e-02 1.91466168e-01 -3.93438200e-03
1.51303455e-01 2.00112537e-01 -7.83595502e-01 -5.50233901e-01
2.51384787e-02 1.75406769e-01 7.43172884e-01 6.68138325e-01
1.21050596e+00 -1.45551801e-01 -9.98142183e-01 6.11087263e-01
1.34450686e+00 1.47153914e-01 3.53667915e-01 -8.82025436e-02
9.05053616e-01 1.03236079e+00 5.21686733e-01 1.77333012e-01
-7.73779154e-02 6.85373783e-01 -6.95263892e-02 -1.83249742e-01
2.10476950e-01 6.09128438e-02 2.20687449e-01 9.21544373e-01
-4.06664088e-02 -1.27514331e-02 -1.05780411e+00 8.12140822e-01
-2.19330621e+00 -1.15520322e+00 -4.08889890e-01 2.53446651e+00
5.84932387e-01 -1.88121483e-01 6.29025161e-01 3.52629602e-01
1.09567392e+00 -1.57796368e-01 -4.01925474e-01 -1.81181014e-01
-1.31076232e-01 -1.98524594e-01 6.52250111e-01 5.28878987e-01
-1.44693649e+00 9.81788635e-01 7.14850092e+00 8.73319149e-01
-4.53624427e-01 -8.58751535e-02 7.77983785e-01 -1.46571949e-01
-4.24909182e-02 -6.69277385e-02 -8.33194196e-01 2.07920417e-01
9.00821984e-01 -1.79014266e-01 4.74327832e-01 9.05558288e-01
1.44594163e-01 -5.79344518e-02 -1.35344350e+00 1.15688896e+00
2.05635205e-01 -9.53426063e-01 -2.06710517e-01 4.51964140e-01
1.00200534e+00 -3.39062810e-01 1.98944360e-02 -4.60076004e-01
3.00459445e-01 -1.03703094e+00 3.17304909e-01 2.17476249e-01
6.18994296e-01 -9.45661187e-01 3.01689059e-01 4.81852353e-01
-1.19667304e+00 1.55394450e-01 -5.20227253e-01 4.26614285e-01
-1.04599502e-02 4.75085527e-01 -7.06161499e-01 3.75085831e-01
8.05536330e-01 6.57329619e-01 -4.59949225e-01 1.19179666e+00
3.65642726e-01 6.66194737e-01 -7.36498654e-01 3.33281934e-01
2.29691952e-01 -7.04601288e-01 6.63387537e-01 1.35489428e+00
3.23451608e-01 4.41743076e-01 2.37611234e-01 6.19306445e-01
2.11636037e-01 2.72830516e-01 -8.89950454e-01 8.77731368e-02
4.95113373e-01 1.27095664e+00 -1.29618680e+00 -3.38098407e-01
-4.35803592e-01 1.05975485e+00 3.51569764e-02 4.87820894e-01
-4.62652385e-01 -1.22142537e-03 7.91893303e-01 3.57852846e-01
3.33914995e-01 -6.47034824e-01 -3.37370336e-01 -9.68973041e-01
-1.14477158e-01 -6.52179182e-01 6.61011517e-01 -2.33844608e-01
-1.30861473e+00 2.38035560e-01 3.04629445e-01 -1.26706529e+00
-3.59108567e-01 -5.89957535e-01 -3.95923734e-01 5.57054639e-01
-4.73485529e-01 -6.43251538e-01 8.05164799e-02 9.35474575e-01
3.22968811e-01 -7.19328374e-02 6.45671070e-01 -3.02271210e-02
-5.51320970e-01 2.48098671e-01 6.90006852e-01 2.00204805e-01
6.37630403e-01 -1.45552218e+00 4.34071794e-02 9.76814568e-01
5.50341308e-01 8.72747958e-01 8.31681669e-01 -5.04672527e-01
-1.43948996e+00 -9.49813306e-01 5.61818123e-01 -6.81667149e-01
4.33284521e-01 -4.54690069e-01 -9.96306002e-01 6.36675954e-01
1.50381820e-03 3.50535847e-02 9.75752592e-01 1.63397551e-01
-1.97814599e-01 -2.82214233e-03 -1.14237452e+00 5.54875374e-01
1.21076775e+00 -2.90496856e-01 -4.26551133e-01 5.33223510e-01
1.32681981e-01 1.10739004e-02 -7.51150250e-01 2.87070543e-01
3.55188936e-01 -1.14821875e+00 1.19115520e+00 -7.12261558e-01
-2.28614688e-01 -6.32258773e-01 -2.94613361e-01 -1.03314304e+00
-4.68246013e-01 -7.21749425e-01 -2.96344683e-02 1.20132649e+00
1.29902303e-01 -1.09595872e-01 1.30638599e+00 7.44093060e-01
2.14190930e-01 -3.88837039e-01 -1.06251979e+00 -1.08098483e+00
2.58786738e-01 -2.73894221e-01 2.50819355e-01 1.15994203e+00
2.75795698e-01 2.13232934e-01 -2.29163263e-02 3.51280749e-01
1.21764433e+00 3.82971644e-01 8.20539057e-01 -1.50721264e+00
-1.74158916e-01 -5.44736862e-01 -6.21280909e-01 -9.90595341e-01
4.34279203e-01 -6.88574374e-01 7.97666088e-02 -1.26012695e+00
5.22612333e-01 -4.84639972e-01 2.04295199e-02 1.28715858e-01
-8.02349597e-02 2.38389447e-01 3.10965311e-02 6.18567169e-01
-8.43532622e-01 1.78288817e-01 7.35305846e-01 7.31745213e-02
-4.44937140e-01 5.07924333e-02 -4.56399739e-01 9.83069599e-01
7.03716815e-01 -3.43134344e-01 -4.54668880e-01 -4.20017764e-02
-5.65782189e-01 6.26813397e-02 3.40273045e-02 -1.21020925e+00
3.97479862e-01 -1.77068830e-01 5.20673454e-01 -6.42855644e-01
2.78851092e-01 -1.10960329e+00 3.80870372e-01 2.88381636e-01
-1.73970744e-01 -1.44084796e-01 8.79276991e-02 6.96914077e-01
-1.35732725e-01 -6.44112825e-02 1.07312989e+00 1.39032423e-01
-6.30754948e-01 3.99169326e-01 -1.24627583e-01 -1.06941693e-01
1.49521446e+00 -6.14196360e-01 2.33613297e-01 -4.31200355e-01
-9.31579411e-01 1.90864488e-01 7.93266535e-01 1.44699544e-01
4.61016327e-01 -1.35822999e+00 -4.90105808e-01 1.26735374e-01
-1.68407224e-02 -1.05780102e-01 -9.62888151e-02 8.77261162e-01
-5.03398895e-01 5.65946519e-01 2.80518923e-03 -1.15305519e+00
-1.65424955e+00 9.63337302e-01 3.22328180e-01 4.22836602e-01
-3.91108781e-01 6.17903292e-01 4.77721572e-01 -3.97452176e-01
1.72103509e-01 1.52244493e-01 -4.19849567e-02 -8.36391225e-02
2.92422354e-01 7.46112347e-01 -2.89388478e-01 -1.17777014e+00
-5.75678170e-01 9.55006242e-01 1.20660670e-01 -4.49124098e-01
1.14587414e+00 -1.87667027e-01 -2.60922492e-01 5.22046089e-01
1.36504912e+00 9.00210142e-02 -1.17352355e+00 -9.11043957e-02
3.81087661e-01 -5.91573298e-01 -2.23240048e-01 1.77852344e-02
-9.74398017e-01 8.00184190e-01 5.81914544e-01 3.50624830e-01
1.09170079e+00 4.65575039e-01 4.00634021e-01 2.49299556e-01
4.82116699e-01 -1.24045837e+00 -3.25983375e-01 -7.47732893e-02
4.52886701e-01 -8.99758518e-01 1.65859275e-02 -1.01401997e+00
-3.53988796e-01 1.10290396e+00 5.27446270e-01 -1.62509754e-01
6.89566791e-01 1.99232534e-01 -3.38653289e-02 -3.92571371e-03
-2.66568840e-01 -2.34890774e-01 3.48723173e-01 5.24921477e-01
4.05327410e-01 1.08038731e-01 -3.71001959e-01 2.18673542e-01
-1.04225390e-01 -4.15653974e-01 3.32486331e-01 6.53199315e-01
-7.07009852e-01 -1.01214218e+00 -9.38886881e-01 2.61766672e-01
-3.02384943e-01 3.50126415e-01 -6.94088340e-01 7.73308933e-01
1.48516940e-02 1.13299084e+00 3.04499716e-01 -1.45929381e-01
4.06967998e-02 9.16680545e-02 6.08883381e-01 -4.50777143e-01
-4.36319634e-02 7.00953603e-01 -1.91764310e-01 -8.43543410e-01
-7.67660081e-01 -1.21403611e+00 -1.42695379e+00 -4.31062728e-01
-2.35348269e-01 6.25343502e-01 4.57909673e-01 6.57747746e-01
2.50577658e-01 -3.10873568e-01 5.88441074e-01 -7.70730138e-01
-4.44698244e-01 -4.83901978e-01 -1.00055039e+00 6.44526482e-01
2.51812879e-02 -4.45585966e-01 -4.45690155e-01 4.06704694e-01] | [7.697699069976807, 4.471610069274902] |
f14280c4-f31c-4064-80ad-7cdcd087e84a | exploiting-explainability-to-design | 2305.18585 | null | https://arxiv.org/abs/2305.18585v1 | https://arxiv.org/pdf/2305.18585v1.pdf | Exploiting Explainability to Design Adversarial Attacks and Evaluate Attack Resilience in Hate-Speech Detection Models | The advent of social media has given rise to numerous ethical challenges, with hate speech among the most significant concerns. Researchers are attempting to tackle this problem by leveraging hate-speech detection and employing language models to automatically moderate content and promote civil discourse. Unfortunately, recent studies have revealed that hate-speech detection systems can be misled by adversarial attacks, raising concerns about their resilience. While previous research has separately addressed the robustness of these models under adversarial attacks and their interpretability, there has been no comprehensive study exploring their intersection. The novelty of our work lies in combining these two critical aspects, leveraging interpretability to identify potential vulnerabilities and enabling the design of targeted adversarial attacks. We present a comprehensive and comparative analysis of adversarial robustness exhibited by various hate-speech detection models. Our study evaluates the resilience of these models against adversarial attacks using explainability techniques. To gain insights into the models' decision-making processes, we employ the Local Interpretable Model-agnostic Explanations (LIME) framework. Based on the explainability results obtained by LIME, we devise and execute targeted attacks on the text by leveraging the TextAttack tool. Our findings enhance the understanding of the vulnerabilities and strengths exhibited by state-of-the-art hate-speech detection models. This work underscores the importance of incorporating explainability in the development and evaluation of such models to enhance their resilience against adversarial attacks. Ultimately, this work paves the way for creating more robust and reliable hate-speech detection systems, fostering safer online environments and promoting ethical discourse on social media platforms. | ['Bonnie J. Dorr', 'Ian Perera', 'Suhas Harish', 'Prashanth Thamminedi', 'Sohaib Uddin Syed', 'Pranath Reddy Kumbam'] | 2023-05-29 | null | null | null | null | ['adversarial-robustness', 'hate-speech-detection'] | ['adversarial', 'natural-language-processing'] | [ 1.48203105e-01 3.69934738e-01 4.02720198e-02 1.58117250e-01
-2.71037519e-01 -1.08433330e+00 9.42358911e-01 2.73780644e-01
2.02256456e-01 1.95667386e-01 7.41749763e-01 -8.21312249e-01
-2.21968994e-01 -4.99544829e-01 -3.58082145e-01 -2.35747799e-01
-1.21543989e-01 -1.48721144e-01 -5.28608710e-02 -4.51220661e-01
4.58083779e-01 4.43086118e-01 -1.28204966e+00 5.09376228e-01
9.77116287e-01 1.09383352e-01 -4.36047167e-01 9.55298841e-01
1.58082426e-01 1.24442244e+00 -1.09430373e+00 -1.01963639e+00
-2.29018591e-02 -3.27013910e-01 -7.26278901e-01 -1.94147930e-01
3.87546599e-01 -6.44608378e-01 -2.21319243e-01 1.04813194e+00
5.37019849e-01 -2.52973557e-01 4.42735910e-01 -1.61379814e+00
-1.34423554e+00 7.87253022e-01 5.35489991e-02 3.34825814e-01
6.56444252e-01 5.07000208e-01 7.74538159e-01 -5.11318505e-01
6.15980923e-01 1.51156521e+00 7.07474709e-01 9.85237002e-01
-1.14174545e+00 -8.06273282e-01 -6.39613834e-04 -2.76641380e-02
-8.75522494e-01 -6.60990298e-01 7.90651619e-01 -7.99701810e-01
9.32936311e-01 6.74393654e-01 5.84804833e-01 1.62521422e+00
2.85080433e-01 4.51161534e-01 1.26506639e+00 -3.93959463e-01
1.93534911e-01 4.91903365e-01 -2.42141895e-02 7.15355456e-01
4.24264491e-01 1.44579560e-01 -4.52359676e-01 -7.76815236e-01
1.09344840e-01 -5.06099090e-02 1.46748712e-02 3.21978211e-01
-5.67653120e-01 1.21745968e+00 2.37990141e-01 5.42247057e-01
-2.40894765e-01 3.88913266e-02 5.98624825e-01 5.19781932e-02
9.03684318e-01 9.66640770e-01 7.24693537e-02 -2.42812946e-01
-7.58576512e-01 3.58208477e-01 9.39078331e-01 3.07051718e-01
2.73822516e-01 1.74313754e-01 -7.68031329e-02 4.36767757e-01
6.36908829e-01 7.30085552e-01 -1.49424464e-01 -8.94565225e-01
4.01682794e-01 7.38558650e-01 -4.37469818e-02 -1.71296585e+00
-5.31640239e-02 -2.97726035e-01 -9.54795033e-02 3.19306821e-01
1.53684080e-01 -3.61036956e-01 -5.22024035e-01 1.48631990e+00
2.03046471e-01 -5.88643327e-02 -1.17869467e-01 6.89404070e-01
4.16466862e-01 5.45716226e-01 6.14418924e-01 1.77625626e-01
1.20517206e+00 -5.34001172e-01 -8.84942055e-01 -3.58505845e-01
1.07928419e+00 -8.50510597e-01 1.13188004e+00 3.19531709e-01
-7.71988094e-01 1.19341284e-01 -1.02616203e+00 1.73192948e-01
-6.76110327e-01 -6.12139463e-01 5.74024320e-01 1.50623298e+00
-8.18984270e-01 3.94755393e-01 -5.41260421e-01 -4.40281093e-01
4.85607386e-01 7.97582120e-02 -2.92797565e-01 4.02284600e-02
-1.49077857e+00 1.34127939e+00 -1.52997568e-01 -7.31797665e-02
-9.80759859e-01 -7.02843428e-01 -7.39359975e-01 -2.04126969e-01
2.97243059e-01 -3.62812519e-01 9.66497958e-01 -1.07059336e+00
-9.43745136e-01 4.96839017e-01 2.86793768e-01 -5.27674973e-01
3.36559892e-01 -4.05041009e-01 -6.33350194e-01 1.16028808e-01
9.16256905e-02 1.43924877e-01 8.39720726e-01 -1.31160903e+00
2.08247900e-01 -1.71747118e-01 5.52273929e-01 -1.87439814e-01
-1.08830166e+00 9.13730741e-01 5.48668087e-01 -7.70148695e-01
-8.95962298e-01 -9.51031566e-01 -2.51881387e-02 -1.93079576e-01
-6.57961786e-01 1.56214327e-01 1.39214563e+00 -1.04749227e+00
1.92782426e+00 -1.98254168e+00 -1.52490512e-01 1.20748460e-01
5.70703804e-01 9.88947630e-01 -3.18574533e-02 9.61726844e-01
8.02471042e-02 1.09418356e+00 -1.00092843e-01 -3.80302906e-01
2.93507040e-01 3.29077207e-02 -6.35944128e-01 4.96196002e-01
5.13481617e-01 7.10272193e-01 -9.02939200e-01 -2.25696653e-01
2.60963082e-01 8.26084971e-01 -9.08086538e-01 2.33524621e-01
-1.51892483e-01 1.89668968e-01 -5.27301311e-01 5.27896345e-01
4.11957949e-01 2.90017903e-01 7.21703004e-03 3.70964080e-01
-2.64681280e-01 4.36465949e-01 -6.15246356e-01 6.20554090e-01
-2.97511488e-01 9.98183250e-01 1.78617135e-01 -1.58948347e-01
8.15651298e-01 3.73896003e-01 -1.48431561e-03 -3.62271458e-01
2.55393147e-01 -2.33143978e-02 1.48133814e-01 -9.15880799e-01
5.27965367e-01 -3.77940349e-02 -1.19614996e-01 8.25917959e-01
-3.91145378e-01 1.86872799e-02 -1.50408342e-01 5.69325686e-01
1.31731057e+00 -9.96428579e-02 1.46286339e-01 -1.02986544e-01
4.69919145e-01 -3.97179499e-02 4.13916111e-02 7.42451847e-01
-6.11810565e-01 1.59777716e-01 6.74992740e-01 -4.61851716e-01
-1.12977874e+00 -7.52613723e-01 2.70002395e-01 1.25241148e+00
-3.41798693e-01 -8.01315308e-01 -1.37179732e+00 -9.67822492e-01
-6.98863789e-02 1.16137969e+00 -9.84498024e-01 -4.51944411e-01
-3.35149616e-01 -5.08496940e-01 1.21733236e+00 1.06921874e-01
1.92544341e-01 -9.24030662e-01 -5.42022824e-01 9.87490416e-02
-6.03088513e-02 -9.92509067e-01 -2.69241482e-01 -4.66335565e-01
-2.09050834e-01 -1.29327023e+00 -4.26984951e-02 9.73514840e-03
7.03804255e-01 3.23429942e-01 6.00444913e-01 7.89284289e-01
-2.95143485e-01 5.87437689e-01 -6.54790342e-01 -7.76947320e-01
-1.30114591e+00 8.21102262e-02 9.54979137e-02 -1.41213134e-01
4.04322028e-01 -2.46666417e-01 -3.46385539e-01 3.01202059e-01
-1.47311342e+00 -9.42337662e-02 1.65293828e-01 2.71687150e-01
-6.31352305e-01 -1.05227590e-01 6.28112495e-01 -1.08344162e+00
1.21496797e+00 -1.00593269e+00 -8.97069722e-02 1.78872764e-01
-6.26949728e-01 -2.36417726e-01 7.87407637e-01 -6.16760790e-01
-1.00085020e+00 -6.04482770e-01 -2.26506859e-01 -6.33275807e-02
-2.03745812e-01 4.91935462e-01 5.88108227e-02 -3.03681076e-01
1.16437697e+00 -2.90806085e-01 1.59751445e-01 -2.04460055e-01
4.65695441e-01 8.75938535e-01 -2.89814249e-02 -5.03888965e-01
1.35581040e+00 3.11953902e-01 -4.39307749e-01 -1.00808322e+00
-7.38039970e-01 -1.25644550e-01 -1.69204056e-01 -7.35733032e-01
8.30706358e-01 -4.08996433e-01 -5.21777630e-01 3.71726096e-01
-1.32456672e+00 -2.76986152e-01 4.09886718e-01 -4.44308184e-02
6.28556907e-02 6.53242648e-01 -6.82160556e-01 -1.30602360e+00
-4.85292107e-01 -1.25220144e+00 6.63235128e-01 -1.38196021e-01
-9.18014884e-01 -1.33502054e+00 3.11778218e-01 1.04827821e+00
7.53593266e-01 6.77245438e-01 9.01751757e-01 -1.13705254e+00
-1.41903266e-01 -3.59746635e-01 1.66989975e-02 3.18833321e-01
-7.70648196e-02 6.29583955e-01 -1.04839253e+00 -5.04354313e-02
-8.88512284e-02 -3.60603154e-01 8.17801803e-02 -3.03125292e-01
3.97741467e-01 -1.11495280e+00 4.27225195e-02 -7.54682301e-03
1.04517436e+00 -1.52512461e-01 6.50089383e-01 6.31668687e-01
7.43772030e-01 1.06374621e+00 3.25437486e-01 5.51479638e-01
3.34822327e-01 4.45483148e-01 7.44478762e-01 -1.39911957e-02
-2.25278642e-02 -4.77834970e-01 7.57880628e-01 5.02582431e-01
7.14332908e-02 -2.94485331e-01 -1.21608698e+00 3.53359282e-01
-1.53617096e+00 -1.28004313e+00 -3.69303554e-01 1.66131949e+00
4.49024141e-01 7.13844970e-02 2.21548080e-01 1.69166848e-01
6.72784925e-01 5.37770033e-01 -3.73258404e-02 -1.19378257e+00
-8.55482742e-03 -9.96979848e-02 2.73307383e-01 8.27560604e-01
-9.84822452e-01 1.05725121e+00 6.47617865e+00 3.28006238e-01
-8.91439319e-01 2.09510148e-01 4.22182232e-01 -5.30440286e-02
-6.52617037e-01 2.51579106e-01 -4.33247924e-01 5.06724656e-01
1.36002016e+00 -1.31942630e-01 7.67759442e-01 8.27444136e-01
7.67642140e-01 2.72895157e-01 -6.95848882e-01 -6.84084222e-02
5.04213870e-01 -1.35988653e+00 5.13712727e-02 3.44174057e-01
6.27349973e-01 -3.21706504e-01 4.40523952e-01 1.19736023e-01
5.42393804e-01 -1.13484061e+00 9.28227127e-01 1.70994893e-01
1.67707413e-01 -6.81670010e-01 5.40445328e-01 2.84078091e-01
-5.21405816e-01 -3.05380315e-01 1.85300112e-01 -5.10641813e-01
-6.33587502e-03 9.68008712e-02 -1.18004739e+00 5.08052558e-02
4.09905225e-01 3.51690024e-01 -7.93525875e-01 3.85002136e-01
-4.48855609e-01 1.05355573e+00 8.44597816e-02 -1.83151945e-01
3.01117480e-01 2.73457021e-01 7.23663509e-01 1.46117651e+00
-8.02550092e-02 -1.04109302e-01 -1.49868086e-01 1.16109085e+00
1.87308192e-01 -3.28330174e-02 -1.09078634e+00 -6.46461487e-01
7.81906486e-01 1.12812138e+00 -4.44091618e-01 8.89935493e-02
-3.67709339e-01 6.92173719e-01 2.76847273e-01 1.22824155e-01
-1.06140816e+00 -1.80210665e-01 8.00624132e-01 4.68969852e-01
-3.50765288e-01 -2.59745330e-01 -3.98601174e-01 -8.15055490e-01
-3.37246418e-01 -1.50092840e+00 1.40856028e-01 -7.52875447e-01
-1.20931268e+00 5.06237745e-01 1.45362929e-01 -5.49503922e-01
-2.38487616e-01 -4.11302805e-01 -8.77216220e-01 6.34025097e-01
-1.06987572e+00 -1.60089743e+00 1.99485198e-01 3.17371279e-01
3.07815194e-01 -1.61123220e-02 8.28873217e-01 2.89975461e-02
-8.69282067e-01 4.67561990e-01 -3.89236569e-01 2.33959287e-01
5.28611541e-01 -7.94399738e-01 7.43959844e-01 1.28128576e+00
-2.27005333e-01 1.27558541e+00 1.09212101e+00 -1.11844540e+00
-1.25723100e+00 -9.28460658e-01 1.03294468e+00 -1.13812578e+00
1.43366838e+00 -7.24486291e-01 -1.01134121e+00 5.92622817e-01
4.69534904e-01 -4.78315771e-01 1.18517339e+00 -2.46053073e-03
-6.56174958e-01 6.44495904e-01 -1.30184114e+00 1.00617683e+00
8.04670036e-01 -1.03331614e+00 -5.29788077e-01 3.76411617e-01
8.63045394e-01 1.66807175e-01 -7.96972692e-01 -2.07442865e-01
4.44243163e-01 -1.11270964e+00 6.83334529e-01 -8.89011502e-01
7.22660422e-01 -1.74739674e-01 4.92454395e-02 -1.11742795e+00
-2.27754936e-01 -1.09172189e+00 -2.52192765e-01 1.66551840e+00
4.26082611e-01 -7.80019104e-01 5.69410443e-01 1.21429110e+00
-1.96662936e-02 -5.69607735e-01 -4.66936171e-01 -4.09902513e-01
2.13491142e-01 -7.39469409e-01 5.42390108e-01 1.33262229e+00
2.08643496e-01 -8.02245066e-02 -6.30073488e-01 5.64148962e-01
5.28041363e-01 -8.67784798e-01 9.77553785e-01 -9.55755830e-01
-9.70948860e-02 -2.51159310e-01 -3.38581651e-01 1.40519053e-01
3.71369839e-01 -5.62207699e-01 -4.32533532e-01 -1.31884110e+00
2.77850866e-01 -2.03121677e-02 4.15731013e-01 6.21578753e-01
-3.21283996e-01 3.74821037e-01 7.29278743e-01 2.64922917e-01
-3.70512098e-01 1.08477451e-01 7.57785141e-01 1.01983510e-01
6.55075088e-02 -5.02182662e-01 -1.24251091e+00 8.91818821e-01
9.15431857e-01 -8.51687908e-01 -4.50305670e-01 -3.39770854e-01
5.81943631e-01 -5.48944712e-01 7.03839004e-01 -7.58815765e-01
7.77077377e-02 -3.78453910e-01 -2.08387002e-01 1.94413096e-01
7.66209364e-02 -6.64813817e-01 6.79657981e-02 6.51807904e-01
-5.78007102e-01 2.60255665e-01 3.00539047e-01 3.66201133e-01
3.01435828e-01 -2.58308291e-01 4.33406174e-01 1.11940801e-01
-2.37744123e-01 -9.27936286e-02 -1.06765306e+00 1.79262117e-01
1.24936211e+00 -3.85152668e-01 -1.12557209e+00 -6.34170413e-01
-4.69935626e-01 -1.55553922e-01 7.86260486e-01 6.46885753e-01
5.27736127e-01 -9.79037583e-01 -7.60761917e-01 5.71081825e-02
1.33671209e-01 -1.01516342e+00 1.63019091e-01 4.92018670e-01
-5.61313152e-01 2.05097288e-01 -1.23104468e-01 1.68934256e-01
-1.57724559e+00 6.86163545e-01 2.41102993e-01 -1.24275334e-01
-9.33806673e-02 4.41156656e-01 -7.59922266e-02 -4.28349882e-01
-1.03373947e-02 4.64137197e-01 -1.87097311e-01 -3.16245794e-01
7.78606296e-01 6.95556343e-01 -3.82727116e-01 -9.27086592e-01
-4.26888674e-01 -1.15536965e-01 -1.85494557e-01 -9.14930105e-02
1.21722388e+00 -1.83878653e-02 -3.56948435e-01 -9.78859141e-02
9.89878237e-01 6.78845644e-01 -9.69961882e-01 4.30738866e-01
1.64937064e-01 -6.78343594e-01 -1.64704278e-01 -1.13099861e+00
-4.71370965e-01 7.59867609e-01 1.05510920e-01 9.74868596e-01
6.16691530e-01 -1.75302029e-01 6.73343003e-01 1.02398079e-02
-9.66914073e-02 -8.86390805e-01 3.30151170e-01 4.27984923e-01
1.01511300e+00 -8.57473612e-01 8.24447125e-02 -5.08392632e-01
-7.92978406e-01 1.02245986e+00 5.46692073e-01 -5.07984757e-02
2.22291812e-01 4.09837037e-01 2.54252583e-01 -4.12232548e-01
-6.57016873e-01 3.77320558e-01 2.60685027e-01 9.43595886e-01
5.47954679e-01 1.83710501e-01 -1.73391730e-01 5.82840919e-01
-2.60999680e-01 -4.52626377e-01 8.61482561e-01 8.95018041e-01
-3.90172452e-01 -1.21989083e+00 -9.73522842e-01 -4.47232090e-02
-9.31257367e-01 -3.56158108e-01 -1.50460804e+00 8.25485587e-01
1.14292853e-01 1.62654257e+00 -5.98788917e-01 -9.15012479e-01
1.26835942e-01 -2.57914364e-02 -1.79040730e-01 -5.90637028e-01
-1.42118430e+00 -4.45203632e-01 6.81878984e-01 -3.71270865e-01
-1.82551548e-01 -6.17430985e-01 -7.61829257e-01 -1.17072105e+00
-4.23783749e-01 7.37894922e-02 8.30046654e-01 9.82633889e-01
7.62538850e-01 2.21730307e-01 6.27675354e-01 -5.74958920e-01
-5.34706593e-01 -6.63099349e-01 8.53268951e-02 6.11119688e-01
2.92656362e-01 -4.52238321e-01 -6.09942496e-01 -1.30277023e-01] | [8.64702320098877, 10.442399024963379] |
6f154990-10d0-4555-a185-8d3eb3c597d3 | ultrasound-confidence-maps-of-intensity-and | 2011.11956 | null | https://arxiv.org/abs/2011.11956v4 | https://arxiv.org/pdf/2011.11956v4.pdf | Ultrasound Confidence Maps of Intensity and Structure Based on Directed Acyclic Graphs and Artifact Models | Ultrasound imaging has been improving, but continues to suffer from inherent artifacts that are challenging to model, such as attenuation, shadowing, diffraction, speckle, etc. These artifacts can potentially confuse image analysis algorithms unless an attempt is made to assess the certainty of individual pixel values. Our novel confidence algorithms analyze pixel values using a directed acyclic graph based on acoustic physical properties of ultrasound imaging. We demonstrate unique capabilities of our approach and compare it against previous confidence-measurement algorithms for shadow-detection and image-compounding tasks. | ['John Galeotti', 'Wanwen Chen', 'Alex Ling Yu Hung'] | 2020-11-24 | null | null | null | null | ['shadow-detection'] | ['computer-vision'] | [ 9.11682367e-01 1.19420700e-01 5.69575429e-01 -5.13885736e-01
-9.59696889e-01 -4.28121001e-01 1.75134346e-01 4.81742650e-01
-2.44887367e-01 7.11745918e-01 1.36643827e-01 -4.68351334e-01
-2.64687061e-01 -4.98175234e-01 -4.75140572e-01 -8.62318635e-01
-6.93752766e-01 2.84637898e-01 8.42786610e-01 5.77414572e-01
3.74457300e-01 3.70976329e-01 -1.36906219e+00 6.18604608e-02
6.93239748e-01 9.89096045e-01 7.35914474e-03 1.22035956e+00
2.14220777e-01 7.95939147e-01 -7.78563440e-01 -7.73664415e-02
-4.86792624e-02 -5.51870286e-01 -2.98433185e-01 8.92564431e-02
4.25655961e-01 -6.56617641e-01 2.12552920e-02 1.32879627e+00
5.91993749e-01 -1.32331461e-01 8.43935490e-01 -9.95298564e-01
-7.43994564e-02 4.04470146e-01 -6.40943646e-01 4.73000079e-01
6.27783656e-01 -9.30254087e-02 3.46596569e-01 -6.85019851e-01
4.88696516e-01 1.10304642e+00 1.01164389e+00 1.61658362e-01
-9.99653399e-01 -4.37472314e-01 -4.17156696e-01 1.03511848e-01
-1.27822745e+00 -2.65883595e-01 2.91353196e-01 -4.08671856e-01
4.07095075e-01 4.48825777e-01 4.96709764e-01 6.70207262e-01
7.44424999e-01 3.19916099e-01 1.51154470e+00 -6.11672223e-01
3.37847978e-01 -4.98323664e-02 1.00461878e-01 1.16673899e+00
4.95145500e-01 1.35656476e-01 -5.19603491e-01 -3.82274091e-01
8.47313702e-01 -4.27319467e-01 -6.06676936e-01 -3.95427316e-01
-7.38053143e-01 4.26037967e-01 -1.16116203e-01 2.07543805e-01
-3.58418792e-01 5.42887151e-01 5.46138082e-03 2.88162157e-02
1.95804283e-01 3.06152880e-01 1.29977793e-01 -1.90501496e-01
-1.05808723e+00 -1.93828225e-01 8.15895855e-01 6.45184457e-01
4.34823722e-01 1.46530448e-02 -2.48614788e-01 4.12878722e-01
5.01706123e-01 7.94944525e-01 2.28951382e-03 -1.15185702e+00
-1.77019060e-01 -3.19009304e-01 1.73260182e-01 -1.16644096e+00
-3.92580003e-01 -2.62749106e-01 -4.96729821e-01 6.73278809e-01
3.67538303e-01 -1.72424138e-01 -1.21279502e+00 1.10911250e+00
8.72919187e-02 5.93683243e-01 -4.76584211e-02 7.59655774e-01
7.40720332e-01 2.98330039e-01 -3.21064964e-02 -6.03816450e-01
1.18455720e+00 -1.42755046e-01 -8.82708013e-01 1.50879413e-01
1.32254198e-01 -9.72172976e-01 4.61989850e-01 7.43269622e-01
-1.25520658e+00 -1.49109229e-01 -1.33121896e+00 5.83284259e-01
2.49700293e-01 -5.57887554e-01 5.10301411e-01 1.41406715e+00
-1.13343072e+00 6.22291625e-01 -9.08129156e-01 5.84333241e-02
3.92964184e-01 3.27422768e-01 4.20911983e-02 -1.38889045e-01
-7.16301620e-01 8.89589071e-01 -1.88083619e-01 2.44722173e-01
-7.80903339e-01 -6.90317571e-01 -8.45498800e-01 -2.17740610e-01
4.22726721e-01 -2.78966010e-01 1.17963290e+00 -4.71347183e-01
-1.23195338e+00 4.67220515e-01 -1.01336069e-01 -3.97750974e-01
6.23633265e-01 -2.29491606e-01 -5.54012060e-01 4.86436695e-01
-1.23002641e-01 -3.27725671e-02 9.19281662e-01 -1.73172200e+00
-5.34970045e-01 -6.66191205e-02 -2.47584864e-01 4.48778458e-02
1.74459428e-01 -4.62044775e-02 -3.77159834e-01 -3.64660054e-01
6.90193594e-01 -8.85396481e-01 -5.43966174e-01 2.31109321e-01
-3.28660876e-01 5.27531445e-01 5.21798968e-01 -4.45174307e-01
1.24685085e+00 -1.97137642e+00 -6.36989892e-01 8.79997790e-01
2.78913528e-01 -9.45526734e-02 8.76215249e-02 6.96742237e-02
2.25774437e-01 9.82348844e-02 -4.92518842e-01 1.40656233e-01
-3.79162788e-01 4.51562643e-01 1.17546439e-01 6.74122810e-01
-4.11637686e-02 1.96271226e-01 -1.20355248e+00 -9.40981448e-01
3.97115231e-01 4.47705150e-01 -1.94035336e-01 1.16659418e-01
1.03128232e-01 2.54851490e-01 -4.86597687e-01 7.82788694e-01
8.57822239e-01 -7.73184653e-03 3.13730597e-01 -4.43186074e-01
-1.55563369e-01 -1.40049860e-01 -1.34338665e+00 1.05367899e+00
-4.40338969e-01 8.05091500e-01 1.11660220e-01 -5.06909728e-01
5.01676559e-01 3.98951054e-01 5.34148574e-01 -5.84321737e-01
1.54816076e-01 3.49794656e-01 3.11774343e-01 -7.62521625e-01
2.92131841e-01 -3.20477992e-01 4.51677859e-01 5.47858894e-01
-2.76011944e-01 -6.07580543e-01 -1.08558506e-01 3.25773358e-01
1.61517191e+00 -2.71902800e-01 9.37019140e-02 -6.15459621e-01
2.73821175e-01 -3.29855263e-01 1.90892339e-01 1.19828749e+00
-2.37916306e-01 1.09113336e+00 5.44307411e-01 9.82773751e-02
-7.45262682e-01 -1.39048016e+00 -6.15576327e-01 3.52743268e-01
4.87903088e-01 -1.13373213e-01 -5.96853077e-01 -5.73097646e-01
-1.56477734e-01 5.16225636e-01 -7.05017805e-01 1.45237684e-01
-4.11383480e-01 -8.97709727e-01 3.98248106e-01 4.29125965e-01
3.22949924e-02 -6.50840223e-01 -1.08104205e+00 4.68990713e-01
5.70764802e-02 -1.25853455e+00 -1.33772358e-01 2.73470968e-01
-7.82292128e-01 -1.15356827e+00 -5.80564022e-01 -3.15009803e-01
9.50163066e-01 1.72821924e-01 1.22925949e+00 5.07746935e-01
-6.80373847e-01 9.91696477e-01 -3.86525095e-01 -6.66042924e-01
-7.75678158e-01 -1.05934656e+00 -1.05258599e-01 -1.34495914e-01
-2.69761026e-01 -4.98920560e-01 -9.47626293e-01 2.91727662e-01
-9.72995043e-01 -5.01044750e-01 5.02973378e-01 6.94565237e-01
4.89368826e-01 3.05495054e-01 1.91299543e-01 -1.21989453e+00
6.33836031e-01 -1.66997179e-01 -6.24672711e-01 2.15498224e-01
-7.93478966e-01 9.66884643e-02 -6.44445494e-02 -1.18387900e-01
-1.10635352e+00 -2.21047521e-01 -1.01106048e-01 -2.36905143e-02
2.73488499e-02 4.20534998e-01 4.72673804e-01 -5.68414748e-01
6.96838617e-01 -3.14782746e-02 -4.51098755e-02 1.59709137e-02
-7.98339471e-02 4.63809669e-01 7.49949992e-01 -3.49754483e-01
5.93837798e-01 6.51975155e-01 4.54359949e-01 -1.11077070e+00
-5.09964705e-01 -4.37003553e-01 -2.05116794e-01 -7.41059363e-01
5.81883848e-01 -3.93575788e-01 -6.61712229e-01 2.78346837e-01
-1.03859067e+00 -1.01996645e-01 6.96223155e-02 6.82120264e-01
-1.83071002e-01 9.69325960e-01 -4.38988358e-01 -1.33358943e+00
-1.03527475e-02 -1.13368702e+00 6.77920818e-01 1.50420338e-01
-2.03992426e-01 -1.05844820e+00 1.69130322e-02 9.13330838e-02
5.75617194e-01 4.35817748e-01 7.81309843e-01 -6.47604689e-02
-5.46878397e-01 -1.99915454e-01 -3.17958713e-01 3.98861766e-01
2.60590047e-01 2.26555422e-01 -9.92708921e-01 -1.22586116e-01
2.92366952e-01 -2.12565020e-01 6.92903340e-01 8.63362908e-01
1.24860024e+00 1.13251075e-01 -3.44400316e-01 2.02792987e-01
1.52066445e+00 3.27209413e-01 7.82157660e-01 -4.54512537e-02
1.72475502e-01 4.54624534e-01 6.30720079e-01 5.36319554e-01
-2.16662526e-01 2.18830943e-01 5.68887472e-01 -1.85457751e-01
-3.27734530e-01 4.62165236e-01 -2.28405297e-01 4.14081216e-01
-4.34298992e-01 -4.76768136e-01 -9.83113468e-01 2.46012658e-01
-1.40480149e+00 -6.92203343e-01 -5.50653577e-01 2.25876665e+00
6.38003469e-01 4.94897753e-01 -5.39363086e-01 2.36939564e-01
4.39123392e-01 3.08043770e-02 -1.92868367e-01 -2.37447262e-01
8.55339617e-02 4.71484154e-01 7.26806223e-01 8.49119306e-01
-8.50083590e-01 9.55612361e-02 8.38664627e+00 3.38567346e-01
-7.68162191e-01 -2.34381352e-02 5.85530281e-01 3.46507311e-01
-6.33801639e-01 -1.36694193e-01 -1.14562117e-01 4.04389560e-01
6.43647134e-01 3.68696094e-01 -2.57940143e-01 2.19296962e-01
2.87327133e-02 -1.04855454e+00 -8.12732100e-01 7.57557631e-01
2.91695088e-01 -9.72250819e-01 -3.84396076e-01 -2.30898067e-01
6.72649443e-01 -2.50100225e-01 2.18410179e-01 -5.88029623e-01
4.54669356e-01 -1.07373667e+00 4.60477471e-01 5.52462876e-01
7.00603604e-01 -3.27257723e-01 1.00408745e+00 -3.62250298e-01
-7.37164736e-01 3.72894615e-01 -1.15740649e-01 2.37138450e-01
2.96551138e-01 1.05941200e+00 -1.14565301e+00 3.19450319e-01
5.71362615e-01 -1.13665871e-01 -2.38482997e-01 1.38158691e+00
-1.13952525e-01 7.85815239e-01 -5.00506818e-01 -8.18694010e-02
2.39798158e-01 -5.83869740e-02 5.79705060e-01 1.47190905e+00
5.50154030e-01 4.70137417e-01 -1.94734260e-01 4.30977970e-01
5.16388178e-01 -2.18083620e-01 -3.69722486e-01 3.47081900e-01
3.67575586e-01 1.06986010e+00 -1.20579040e+00 -3.74291480e-01
-6.03795052e-01 8.41898978e-01 -5.08036673e-01 3.46163243e-01
-7.36723602e-01 -2.83217803e-02 4.31001037e-02 2.10488439e-01
1.72942936e-01 -3.82962823e-01 -2.15515465e-01 -4.99131739e-01
-8.09577182e-02 -6.44817770e-01 2.25606754e-01 -8.61955345e-01
-9.01325464e-01 4.45997685e-01 1.26747653e-01 -1.19657648e+00
-2.02127218e-01 -6.63132966e-01 -5.76384306e-01 5.11109173e-01
-1.49207604e+00 -4.90808398e-01 -4.55056697e-01 3.54751617e-01
2.68191606e-01 3.27747732e-01 7.74028540e-01 1.66806981e-01
1.68600798e-01 3.10924411e-01 2.00769350e-01 -1.55653367e-02
5.57233810e-01 -1.38788033e+00 -1.44125745e-01 9.59265888e-01
9.19871181e-02 2.29800954e-01 1.29342508e+00 -7.15892613e-01
-1.42253852e+00 -3.27247620e-01 2.52036452e-01 -5.16343057e-01
8.06454659e-01 7.02207386e-02 -9.93355751e-01 1.82909250e-01
1.87626798e-02 1.95301384e-01 7.79221177e-01 -1.46822259e-02
-1.95189491e-01 8.23531896e-02 -1.13142550e+00 3.51772994e-01
8.62706602e-01 -1.51335657e-01 -3.03894907e-01 3.73302788e-01
-3.48924026e-02 -6.57882094e-01 -7.36377835e-01 5.96499383e-01
1.04578745e+00 -1.38932550e+00 9.25414741e-01 1.58642218e-01
1.77881747e-01 -1.73217624e-01 1.96154006e-02 -9.35796201e-01
-1.69884965e-01 -4.85920578e-01 1.73472106e-01 7.39066541e-01
5.95388651e-01 -4.07459944e-01 8.68147492e-01 6.80568099e-01
-2.43962929e-01 -1.85040876e-01 -1.08856642e+00 -7.73326635e-01
-5.17096162e-01 -7.63177514e-01 -5.38668372e-02 5.14188170e-01
6.26976937e-02 -6.90619051e-02 -2.18533725e-01 7.45143056e-01
1.22373104e+00 -2.60262582e-02 1.24259420e-01 -1.10354185e+00
-5.93397021e-01 -2.00136036e-01 -5.16809225e-01 -7.13379622e-01
-7.18540668e-01 -1.07844427e-01 7.83371389e-01 -1.58935809e+00
1.99451268e-01 -7.64326632e-01 -4.68890667e-01 -5.70444623e-03
-1.98089048e-01 7.05032945e-01 -1.63642660e-01 -8.20044279e-02
-6.81150079e-01 -2.03227967e-01 1.28520215e+00 5.99800870e-02
1.76125914e-01 1.09825127e-01 -2.07184196e-01 9.39405918e-01
4.82824028e-01 -5.90141177e-01 -2.77018130e-01 -3.43694478e-01
3.80454361e-01 5.71897030e-01 3.92062962e-01 -1.34156454e+00
3.83778751e-01 8.78815278e-02 3.75978082e-01 -4.70324993e-01
3.04578185e-01 -1.01953316e+00 2.97293603e-01 9.58427787e-01
-5.81645221e-02 -3.11080098e-01 1.02272861e-01 8.36873353e-01
-5.81393652e-02 -6.91989422e-01 8.10707510e-01 -1.43691942e-01
-6.26926243e-01 -8.41012299e-02 -9.09432173e-01 -3.39220315e-01
6.11544192e-01 -3.22028756e-01 7.09836036e-02 -7.01063752e-01
-7.36966789e-01 3.71215399e-03 7.32382461e-02 -1.53161809e-01
1.01032269e+00 -7.23150194e-01 -7.72454858e-01 8.89331549e-02
-2.26212908e-02 -3.72706860e-01 4.03442591e-01 9.06645417e-01
-9.85102832e-01 -1.56739637e-01 5.16298115e-02 -8.52849782e-01
-1.70737159e+00 7.49004483e-02 2.67483622e-01 -4.89339270e-02
-6.15226030e-01 1.06883919e+00 2.87573226e-02 5.13226151e-01
3.56370479e-01 -4.73892391e-01 1.47522064e-02 -3.13727707e-01
5.71152031e-01 4.79600489e-01 1.57456428e-01 -6.37129098e-02
-4.52352136e-01 7.46748686e-01 1.57468781e-01 -4.51373518e-01
8.20774138e-01 -2.02392712e-01 -1.15061514e-02 2.92254955e-01
6.61381304e-01 1.80026695e-01 -1.09185684e+00 -4.55894507e-02
-1.42863005e-01 -8.03751171e-01 3.51731241e-01 -9.14550483e-01
-7.57641017e-01 7.53466666e-01 9.69437242e-01 6.39346480e-01
1.02475190e+00 -5.80107793e-02 3.98816615e-01 7.40301460e-02
4.08410937e-01 -8.85349095e-01 3.59824687e-01 1.24111492e-02
5.92929363e-01 -1.36859012e+00 4.79466587e-01 -8.55420232e-01
-2.22075805e-01 1.27464759e+00 4.60106879e-02 8.88269320e-02
1.06593502e+00 1.07657707e+00 3.43528032e-01 -1.85201809e-01
-2.59756148e-01 -9.53966230e-02 2.73185998e-01 9.42188382e-01
5.70327461e-01 6.34210408e-02 -4.27119434e-01 -1.30656898e-01
2.17939645e-01 -1.33987060e-02 9.53718185e-01 1.08281147e+00
-8.22956622e-01 -8.28193247e-01 -7.52125561e-01 7.07527220e-01
-7.28112221e-01 -3.33123535e-01 2.14684799e-01 5.86763084e-01
4.17783521e-02 1.07940698e+00 -9.31444205e-03 -2.14182120e-02
1.95351839e-01 -4.96541679e-01 9.31453884e-01 -4.99626756e-01
-1.90544799e-01 3.64604622e-01 3.26429635e-01 -5.79754055e-01
-7.08161831e-01 -8.43567848e-01 -1.39154792e+00 3.96966934e-02
-5.83908916e-01 1.76136747e-01 7.89266706e-01 8.18906486e-01
-2.67758280e-01 7.49345064e-01 3.53236675e-01 -5.21247268e-01
-3.49517047e-01 -7.13855267e-01 -8.42788994e-01 2.14081243e-01
5.51490426e-01 -6.04673207e-01 -6.43384814e-01 1.03177615e-01] | [13.35286808013916, -2.5289251804351807] |
6ae35cab-fa85-4b16-aee0-e8fd6728e862 | table-structure-recognition-using-top-down-1 | 2010.04565 | null | https://arxiv.org/abs/2010.04565v1 | https://arxiv.org/pdf/2010.04565v1.pdf | Table Structure Recognition using Top-Down and Bottom-Up Cues | Tables are information-rich structured objects in document images. While significant work has been done in localizing tables as graphic objects in document images, only limited attempts exist on table structure recognition. Most existing literature on structure recognition depends on extraction of meta-features from the PDF document or on the optical character recognition (OCR) models to extract low-level layout features from the image. However, these methods fail to generalize well because of the absence of meta-features or errors made by the OCR when there is a significant variance in table layouts and text organization. In our work, we focus on tables that have complex structures, dense content, and varying layouts with no dependency on meta-features and/or OCR. We present an approach for table structure recognition that combines cell detection and interaction modules to localize the cells and predict their row and column associations with other detected cells. We incorporate structural constraints as additional differential components to the loss function for cell detection. We empirically validate our method on the publicly available real-world datasets - ICDAR-2013, ICDAR-2019 (cTDaR) archival, UNLV, SciTSR, SciTSR-COMP, TableBank, and PubTabNet. Our attempt opens up a new direction for table structure recognition by combining top-down (table cells detection) and bottom-up (structure recognition) cues in visually understanding the tables. | ['C. V. Jawahar', 'Ajoy Mondal', 'Sachin Raja'] | 2020-10-09 | table-structure-recognition-using-top-down | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/6007_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123730069.pdf | eccv-2020-8 | ['table-recognition', 'cell-detection'] | ['computer-vision', 'computer-vision'] | [ 4.01901513e-01 -1.90650180e-01 -1.81667671e-01 -1.75925508e-01
-8.50779176e-01 -1.20676005e+00 6.94952965e-01 7.65776932e-01
4.16352041e-02 5.45934498e-01 4.92851228e-01 -4.02032375e-01
-1.54672980e-01 -7.40827382e-01 -7.45514989e-01 -3.06499749e-01
3.36907012e-03 7.06568539e-01 2.04181626e-01 -1.58665627e-01
8.07667196e-01 8.17598462e-01 -1.40439606e+00 1.27189863e+00
6.51254475e-01 7.80852556e-01 1.48891404e-01 1.05320799e+00
-5.58853865e-01 1.05292678e+00 -8.80863786e-01 -5.20608187e-01
3.04135066e-02 -3.80290568e-01 -7.68821597e-01 4.63082910e-01
7.50808775e-01 1.05082117e-01 -3.55791926e-01 8.30394208e-01
2.56713867e-01 -4.57006544e-01 9.82428968e-01 -8.54037821e-01
-9.99206126e-01 6.63742900e-01 -7.53325224e-01 3.43425840e-01
9.20191228e-01 -3.55593473e-01 1.24323511e+00 -9.41915691e-01
1.17853212e+00 1.52081645e+00 5.26305676e-01 3.31686065e-02
-1.30121815e+00 -2.62040757e-02 2.32807279e-01 3.59602153e-01
-1.33311844e+00 -4.26295251e-01 4.96682853e-01 -5.01401424e-01
1.19228542e+00 7.05048800e-01 4.16847467e-01 9.95110154e-01
2.57605940e-01 1.19435906e+00 9.77011740e-01 -8.72997820e-01
1.71270400e-01 4.29803729e-01 2.80771196e-01 8.74026537e-01
6.47162199e-01 -5.88118613e-01 -8.22160900e-01 3.70586663e-02
5.28407633e-01 -2.62301713e-01 -8.82852077e-03 -4.40239936e-01
-1.17728925e+00 3.33731920e-01 1.21007524e-01 4.87684548e-01
8.15036818e-02 -4.00815189e-01 4.58209783e-01 1.05532087e-01
9.53472406e-02 6.15942478e-01 -3.88316631e-01 -5.16559631e-02
-9.25288439e-01 6.37218431e-02 9.08477902e-01 1.35713327e+00
4.86610770e-01 -3.65022004e-01 -5.80522001e-01 7.66104758e-01
2.17597589e-01 3.72267753e-01 1.57385364e-01 -4.20842469e-01
1.09910333e+00 1.14192402e+00 -1.57382652e-01 -1.40559971e+00
-3.46070051e-01 -2.57901639e-01 -6.68682814e-01 -1.85758725e-01
5.72529733e-01 5.53759277e-01 -9.94228601e-01 8.59799862e-01
-2.61647161e-02 -7.06591368e-01 -6.36977926e-02 3.62936258e-01
7.60531127e-01 6.57278836e-01 -4.59753245e-01 1.50054693e-01
1.89017165e+00 -6.36601985e-01 -9.71417844e-01 -2.24093691e-01
9.69233453e-01 -1.04238772e+00 1.11941063e+00 6.05456591e-01
-8.74906421e-01 -3.95234436e-01 -1.33881402e+00 -5.00255585e-01
-1.08826172e+00 6.18368208e-01 4.63742107e-01 9.15851951e-01
-6.80713832e-01 8.55441019e-02 -4.98951823e-01 -5.82603276e-01
4.65224147e-01 2.27750212e-01 -5.14291823e-01 -1.62335858e-01
-5.04017770e-01 6.40111029e-01 1.91353858e-01 1.28696427e-01
-2.17924461e-01 -4.70845908e-01 -8.31848204e-01 9.57124382e-02
6.62588596e-01 -2.45119318e-01 5.90095222e-01 -4.84955847e-01
-7.51523852e-01 1.08795881e+00 -3.12040687e-01 -3.28270733e-01
4.23789114e-01 1.23777017e-02 -4.69003737e-01 1.82208166e-01
-1.02879457e-01 4.83418107e-01 4.53032464e-01 -1.66726625e+00
-6.21143639e-01 -6.27796710e-01 -2.94016421e-01 2.01833695e-01
-1.38879240e-01 6.68044090e-02 -9.79926109e-01 -8.33624125e-01
4.59089994e-01 -6.85762107e-01 4.61353958e-01 -1.05798505e-01
-1.14895141e+00 2.62892425e-01 7.96457827e-01 -9.16641057e-01
1.59244430e+00 -1.99749565e+00 -7.05916882e-02 4.80113328e-01
7.63276070e-02 2.49246899e-02 1.97875239e-02 7.46225715e-01
1.35263890e-01 4.95928973e-01 -1.81878299e-01 -1.36088267e-01
1.48846820e-01 6.27570897e-02 -5.42690337e-01 2.98419178e-01
9.51671600e-03 8.49350870e-01 -3.75360012e-01 -9.05138671e-01
1.62844032e-01 4.04291272e-01 -5.63176990e-01 -2.85379052e-01
-2.08217636e-01 -2.76157111e-01 -1.98100820e-01 1.31402993e+00
5.97081363e-01 -2.03309938e-01 6.50688767e-01 -3.11619133e-01
-7.62547478e-02 3.80879551e-01 -1.56192863e+00 1.11072576e+00
5.44253550e-02 8.79649222e-01 -1.38106540e-01 -6.52538121e-01
8.47641468e-01 -8.14939216e-02 5.13071241e-03 -9.83941913e-01
-1.62944958e-01 1.26691625e-01 -1.23493813e-01 -3.59172434e-01
8.81929517e-01 7.78494895e-01 -2.82105774e-01 8.34848955e-02
-5.21299131e-02 1.69988975e-01 7.51136184e-01 4.58736420e-01
1.18568313e+00 -9.09783393e-02 3.99019033e-01 -1.87002152e-01
7.46032596e-01 2.46449336e-01 7.41435662e-02 1.04938257e+00
1.91316798e-01 8.51540804e-01 9.59740281e-01 -2.87973970e-01
-1.07743740e+00 -8.96986365e-01 -2.43567616e-01 1.00045180e+00
-1.14461422e-01 -9.61196065e-01 -8.72839272e-01 -7.28788674e-01
2.46684805e-01 4.40294683e-01 -9.37803864e-01 3.69626075e-01
-8.24557602e-01 -7.40390301e-01 8.26421499e-01 6.99427843e-01
1.23534590e-01 -1.07168519e+00 -1.98628023e-01 4.51403260e-02
-5.80608919e-02 -1.27761436e+00 -4.74129140e-01 4.94983256e-01
-8.35408330e-01 -1.16009820e+00 -3.02319109e-01 -7.97032654e-01
1.06155646e+00 -3.54363322e-02 1.15260386e+00 1.54116631e-01
-8.85044634e-01 4.15358931e-01 -3.24520320e-02 -3.96663010e-01
-3.36502135e-01 3.09450001e-01 -3.90658200e-01 -1.36748999e-01
3.48510176e-01 1.79868340e-01 -6.52239248e-02 4.19342756e-01
-8.34424317e-01 -4.58425842e-02 6.68099344e-01 6.89551771e-01
8.02735448e-01 1.82098746e-01 -1.42170668e-01 -1.31127477e+00
6.28306031e-01 2.74573993e-02 -6.91652775e-01 8.45310628e-01
-5.79219222e-01 2.70847112e-01 6.21987939e-01 -5.28963190e-03
-1.00289404e+00 2.47138113e-01 2.13316411e-01 9.29472446e-02
-2.34465152e-01 4.32999045e-01 -6.64574265e-01 2.85729170e-01
4.99865055e-01 4.50488657e-01 -4.96713281e-01 -6.61386907e-01
2.64510572e-01 7.02985287e-01 6.27384365e-01 -5.73212743e-01
7.91686296e-01 6.52644873e-01 -2.21190471e-02 -1.03005409e+00
-5.90453863e-01 -4.24517184e-01 -1.09495115e+00 3.35800312e-02
8.38912845e-01 -7.02641308e-01 -7.33872831e-01 2.61937380e-01
-8.18745196e-01 1.41624048e-01 -1.06378668e-03 -4.97362837e-02
-3.08449268e-01 5.36994636e-01 -6.80172503e-01 -8.88280034e-01
3.68453935e-02 -1.00621986e+00 1.07970560e+00 -1.35650858e-03
-3.54361624e-01 -9.11599934e-01 -1.20268561e-01 5.69211006e-01
-1.36317149e-01 3.64715606e-02 1.39819229e+00 -5.55619657e-01
-1.16688776e+00 -2.80638576e-01 -3.89982879e-01 -3.19772243e-01
1.49475351e-01 5.18852055e-01 -8.85849655e-01 7.63619244e-02
-6.21815264e-01 -5.84324561e-02 1.02006400e+00 2.86470056e-01
1.30892074e+00 -3.13601226e-01 -5.84154665e-01 6.89183116e-01
1.51291442e+00 6.35851085e-01 9.35187042e-01 7.03189790e-01
8.48250866e-01 7.17491865e-01 5.15781581e-01 3.84084493e-01
2.86192805e-01 5.78423142e-01 7.19715282e-03 -7.92509988e-02
-1.42497554e-01 -5.81677675e-01 1.77518025e-01 4.58429396e-01
2.24382594e-01 -7.72100866e-01 -1.08146560e+00 3.76691610e-01
-1.59389293e+00 -8.71470988e-01 -3.45842183e-01 1.92313969e+00
6.88054800e-01 6.76995575e-01 -4.18824097e-03 5.27402222e-01
6.67253256e-01 5.65489978e-02 -2.80108601e-01 -5.28802752e-01
-6.84391856e-01 -2.68757701e-01 6.74861252e-01 2.89580017e-01
-1.19529188e+00 8.32392752e-01 6.82541513e+00 9.41800475e-01
-5.57484508e-01 -7.23333895e-01 1.05062759e+00 6.14546575e-02
-1.99034438e-01 -1.77075133e-01 -1.33645725e+00 2.12372005e-01
5.73493361e-01 2.82609731e-01 3.46473843e-01 6.72499776e-01
-2.74912208e-01 -5.26836395e-01 -1.30415154e+00 1.15917802e+00
3.86874586e-01 -1.69344938e+00 3.29799920e-01 3.04717302e-01
3.08864087e-01 -8.25629234e-01 3.36441159e-01 3.13676707e-02
-3.93089466e-02 -1.27776718e+00 9.91025865e-01 5.33841610e-01
8.14881325e-01 -6.97439551e-01 5.78742445e-01 -1.06295742e-01
-1.44216406e+00 3.16129774e-02 -3.59827042e-01 3.52393150e-01
-3.47977698e-01 2.68253565e-01 -9.40692604e-01 3.72634917e-01
6.33871853e-01 6.23129904e-01 -1.43934512e+00 7.64309645e-01
6.71755522e-02 3.60835791e-01 -1.35631874e-01 -3.41966510e-01
-9.02928486e-02 -8.74171257e-02 2.26230413e-01 1.79087830e+00
-3.83832841e-04 -3.07525694e-01 -8.67330208e-02 8.01909685e-01
-2.51335293e-01 3.06339383e-01 -5.94429791e-01 -4.50589448e-01
4.47506487e-01 1.19509149e+00 -1.59971738e+00 -3.72961998e-01
-4.13989574e-01 8.54162633e-01 1.82540253e-01 1.05924629e-01
-3.99299324e-01 -5.56789279e-01 2.17645839e-01 2.58491665e-01
5.89405000e-01 -1.95300788e-01 -1.08155954e+00 -1.17574084e+00
2.96130300e-01 -1.38663697e+00 6.47663236e-01 -8.12260926e-01
-1.06932342e+00 4.66187060e-01 -9.88504291e-02 -9.46221113e-01
-2.00976916e-02 -1.04664338e+00 -3.12793814e-02 6.39498115e-01
-8.49033296e-01 -1.07098496e+00 7.48253055e-03 3.24849874e-01
4.85318124e-01 -4.04241294e-01 7.27223992e-01 1.98249623e-01
-7.69193172e-01 8.42664838e-01 4.30363297e-01 6.06681406e-01
7.63929725e-01 -1.84401786e+00 5.00087738e-01 7.70788193e-01
7.47457147e-01 8.46149147e-01 6.82442665e-01 -8.20893526e-01
-1.84895420e+00 -5.85797906e-01 1.00640500e+00 -1.03912199e+00
4.31219995e-01 -1.37286997e+00 -9.52219188e-01 6.60296738e-01
2.07361341e-01 -2.30864987e-01 6.31035089e-01 2.86752373e-01
-7.04138756e-01 -7.15096667e-02 -1.05743241e+00 7.25545526e-01
1.00202882e+00 -7.50356972e-01 -5.40175915e-01 1.62662864e-01
-2.01990232e-02 -5.00980139e-01 -5.43911994e-01 -2.63404232e-02
7.30482578e-01 -9.29031849e-01 1.17415833e+00 -2.65480965e-01
4.87468541e-01 -4.85374540e-01 -4.02616352e-01 -6.30462527e-01
-3.18530202e-01 -2.08479330e-01 -3.36991966e-01 1.47486043e+00
8.60915005e-01 -5.69047080e-03 1.03777254e+00 3.36935103e-01
1.24486282e-01 -7.00519979e-01 -4.92825985e-01 -6.95683122e-01
-3.40848297e-01 -2.77990699e-01 4.66688931e-01 7.24318087e-01
2.14686505e-02 3.31554532e-01 -7.40153044e-02 1.50819004e-01
2.72637606e-01 1.24676652e-01 7.20268786e-01 -1.07671487e+00
-1.61656529e-01 -4.44124103e-01 -3.59533876e-01 -8.83235514e-01
-3.42019945e-01 -8.92920673e-01 -2.74069101e-01 -1.87217832e+00
4.88562495e-01 -5.32369651e-02 -3.11662108e-02 3.19230855e-01
1.67561784e-01 7.67377242e-02 3.31073105e-01 2.68641382e-01
-8.24780762e-01 -1.42292142e-01 1.04027987e+00 -3.46776038e-01
-1.40554652e-01 -5.70232749e-01 -8.71180117e-01 5.73362947e-01
4.16683614e-01 -4.57017869e-01 -2.07099557e-01 -2.01561347e-01
4.27729696e-01 -2.42539262e-03 -7.84406811e-02 -9.94761527e-01
2.95005560e-01 1.35917559e-01 1.37998700e+00 -1.53067696e+00
1.77841961e-01 -5.41307330e-01 -2.38071159e-01 2.57206351e-01
-5.20212889e-01 4.50901479e-01 3.34918976e-01 6.02042973e-01
-1.74555808e-01 -2.42104709e-01 4.63202000e-01 -2.58350879e-01
-5.32116354e-01 -2.94674665e-01 -8.99067104e-01 1.19826250e-01
5.96838236e-01 -7.40196645e-01 -6.71620548e-01 4.62163836e-02
-5.69442034e-01 -4.76413704e-02 6.68391705e-01 5.15727043e-01
5.86165965e-01 -9.13164794e-01 -3.43714267e-01 4.27586913e-01
5.14012933e-01 -3.86518657e-01 6.72031716e-02 2.90913403e-01
-9.78016317e-01 9.72198188e-01 -1.63884327e-01 -6.43569410e-01
-1.68791890e+00 7.71753073e-01 3.05694006e-02 -6.47736192e-01
-5.03963411e-01 7.33558655e-01 1.45750821e-01 -3.66205186e-01
6.54658675e-01 -6.79075599e-01 -3.83093923e-01 5.49316287e-01
6.21588767e-01 1.75577611e-01 4.90631431e-01 -4.74088520e-01
-6.04232430e-01 4.81998950e-01 -5.35274804e-01 -1.00546330e-01
9.16211247e-01 -1.68167129e-01 -1.56118460e-02 5.70097446e-01
9.47280169e-01 5.52888215e-01 -7.55628288e-01 4.51750793e-02
6.80064261e-01 -5.69494545e-01 -3.52972537e-01 -1.20201504e+00
-5.36796391e-01 7.44384646e-01 4.87941504e-01 3.90920728e-01
9.43738639e-01 -1.94789730e-02 1.75197810e-01 8.43602479e-01
1.33019343e-01 -1.31615043e+00 2.73237407e-01 6.72183633e-01
8.99772406e-01 -1.15078843e+00 2.97530264e-01 -7.02274919e-01
-6.15425348e-01 1.29752409e+00 6.01392806e-01 1.96817994e-01
4.09566283e-01 8.52225065e-01 4.54545766e-02 -2.83848077e-01
-9.04705584e-01 -1.16413884e-01 6.45020604e-01 6.31324291e-01
7.56844699e-01 -1.82086900e-01 6.36009965e-03 4.84914929e-01
-2.55193323e-01 -4.88634884e-01 6.73022151e-01 1.33793402e+00
-3.43086928e-01 -1.12006021e+00 -8.78544986e-01 6.86168909e-01
-6.15665376e-01 -1.92308798e-01 -1.05408239e+00 9.90881801e-01
-3.42417359e-02 6.40772641e-01 3.09774011e-01 -1.62320137e-01
4.33030307e-01 1.75424293e-01 7.21375167e-01 -6.73252404e-01
-6.34202659e-01 2.95776814e-01 2.43405446e-01 -2.54719138e-01
7.64400661e-02 -8.85737360e-01 -1.09860659e+00 -3.08099866e-01
-1.15329623e-02 -2.56760031e-01 5.85387945e-01 5.82349122e-01
4.33221072e-01 6.84427500e-01 4.81123365e-02 -1.64364144e-01
5.12595475e-02 -6.59871459e-01 -7.67550707e-01 3.44700545e-01
1.49918795e-01 -4.25755113e-01 -1.06615335e-01 3.93587589e-01] | [11.691003799438477, 2.9907186031341553] |
2f3a5c22-b4cf-49c7-b4e6-55f40bd7a3c9 | multilevel-transformer-for-multimodal-emotion | 2211.07711 | null | https://arxiv.org/abs/2211.07711v2 | https://arxiv.org/pdf/2211.07711v2.pdf | Multilevel Transformer For Multimodal Emotion Recognition | Multimodal emotion recognition has attracted much attention recently. Fusing multiple modalities effectively with limited labeled data is a challenging task. Considering the success of pre-trained model and fine-grained nature of emotion expression, it is reasonable to take these two aspects into consideration. Unlike previous methods that mainly focus on one aspect, we introduce a novel multi-granularity framework, which combines fine-grained representation with pre-trained utterance-level representation. Inspired by Transformer TTS, we propose a multilevel transformer model to perform fine-grained multimodal emotion recognition. Specifically, we explore different methods to incorporate phoneme-level embedding with word-level embedding. To perform multi-granularity learning, we simply combine multilevel transformer model with Albert. Extensive experimental results show that both our multilevel transformer model and multi-granularity model outperform previous state-of-the-art approaches on IEMOCAP dataset with text transcripts and speech signal. | ['Feng Ye', 'Xiaobo Zhu', 'Meng Li', 'Meimei Wu', 'Junyi He'] | 2022-10-26 | null | null | null | null | ['multimodal-emotion-recognition', 'multimodal-emotion-recognition'] | ['computer-vision', 'speech'] | [ 2.17534512e-01 -4.08080041e-01 -7.32425079e-02 -7.15490103e-01
-1.29265440e+00 -4.97115165e-01 4.68559563e-01 -9.47190821e-02
-3.96609604e-01 5.42038321e-01 7.98426032e-01 1.07742704e-01
1.71164066e-01 -4.15125787e-01 -3.27033281e-01 -6.19798601e-01
2.56452829e-01 -2.80326214e-02 -3.08611304e-01 -2.00817883e-01
-1.88756529e-02 4.20817249e-02 -1.49885988e+00 7.44062006e-01
9.46329236e-01 1.53351450e+00 -2.85412639e-01 4.20378804e-01
-6.00546956e-01 8.63615096e-01 -3.79961371e-01 -6.93024874e-01
-3.58827442e-01 -4.26067829e-01 -6.96302593e-01 2.43832171e-01
1.47234917e-01 2.59968396e-02 2.96322871e-02 8.20329249e-01
7.53615558e-01 3.80271494e-01 6.61205351e-01 -1.40002477e+00
-4.95410264e-01 4.76904035e-01 -4.32517797e-01 -1.11709848e-01
5.97006321e-01 -2.74617344e-01 1.06446159e+00 -1.13384748e+00
1.25334427e-01 1.37677085e+00 6.61239624e-01 4.96029764e-01
-9.48159814e-01 -6.21105075e-01 5.80215454e-01 3.82328898e-01
-1.25913250e+00 -4.22565490e-01 1.06267405e+00 -1.32425278e-01
9.08981323e-01 1.74874753e-01 1.83180362e-01 1.51499832e+00
-1.24258464e-02 1.16186845e+00 1.65550542e+00 -4.59859967e-01
8.38330090e-02 1.68227404e-01 1.26534715e-01 6.45399749e-01
-7.72911370e-01 -4.15649712e-01 -7.46019602e-01 -2.83947796e-01
2.94985026e-01 1.76082611e-01 -1.86527193e-01 -1.64112136e-01
-1.20982528e+00 7.05510318e-01 -3.08980402e-02 5.56690812e-01
-4.99739140e-01 5.88673051e-04 8.88022780e-01 4.51812416e-01
5.85560679e-01 -6.11532256e-02 -6.22438371e-01 -7.61503100e-01
-9.10507143e-01 -3.14270675e-01 7.32521594e-01 6.55923545e-01
6.86670125e-01 1.95211232e-01 -2.72158772e-01 1.39662802e+00
3.15814376e-01 2.24079236e-01 7.90280282e-01 -5.43808997e-01
4.56137985e-01 7.27352619e-01 -2.46549964e-01 -8.10977101e-01
-2.35976845e-01 -1.50383607e-01 -9.44151103e-01 -3.08678210e-01
1.63583737e-02 -4.04130936e-01 -7.71181643e-01 1.78273034e+00
2.74385303e-01 4.26104277e-01 2.45810002e-01 8.92039537e-01
9.13381279e-01 9.26052928e-01 4.23145473e-01 -2.43628502e-01
1.58420813e+00 -1.30756438e+00 -1.07694697e+00 7.36761540e-02
6.83222890e-01 -6.15278542e-01 1.19639850e+00 3.59373659e-01
-8.13591361e-01 -4.89197612e-01 -7.75812805e-01 -1.30600229e-01
-7.09954917e-01 2.05664068e-01 5.32519639e-01 7.27395952e-01
-8.47594023e-01 1.49799958e-01 -4.54237878e-01 -4.37689871e-01
4.86580469e-02 2.14673772e-01 -6.97096527e-01 -1.05762422e-01
-1.45995855e+00 8.93529654e-01 1.44422457e-01 3.34973395e-01
-5.96430957e-01 -2.72445500e-01 -1.15450883e+00 2.26800427e-01
2.86946833e-01 -5.43408871e-01 1.24668992e+00 -8.25755060e-01
-2.05239797e+00 5.13330162e-01 -6.06801093e-01 6.91186860e-02
-2.55365167e-02 -2.43963897e-01 -7.39531815e-01 1.64758906e-01
-3.09104532e-01 3.57460231e-01 9.93037343e-01 -1.17642045e+00
-6.11709416e-01 -3.19864631e-01 -8.36777762e-02 4.75528657e-01
-9.16323304e-01 3.60657454e-01 -4.33880985e-01 -5.74185729e-01
-1.17870830e-01 -5.75925827e-01 -1.07699454e-01 -6.10877931e-01
-1.13161117e-01 -6.04303658e-01 9.25033569e-01 -4.41917092e-01
1.49361515e+00 -2.22684002e+00 2.79578149e-01 -7.36672878e-02
-2.56508887e-02 -4.92475107e-02 -3.68825853e-01 6.48471177e-01
2.83823647e-02 1.76900670e-01 -2.61839665e-02 -8.72830987e-01
7.76024580e-01 8.93289968e-02 -3.38171899e-01 2.20562965e-02
2.94862658e-01 9.85138595e-01 -7.38669157e-01 -7.02310383e-01
4.91754264e-01 7.36842155e-01 -2.37385362e-01 6.00108802e-01
2.51377493e-01 2.43542284e-01 -5.99329114e-01 1.05357516e+00
6.19407535e-01 1.44915767e-02 1.87385604e-02 -7.38582611e-01
-4.79570851e-02 8.47207457e-02 -9.59967792e-01 1.81551254e+00
-8.52055430e-01 1.66943327e-01 1.67082429e-01 -9.95671332e-01
8.53938520e-01 9.18847859e-01 4.59683388e-01 -7.19229937e-01
4.71272945e-01 1.90084368e-01 -4.09714699e-01 -6.68989480e-01
4.98185217e-01 -8.57493758e-01 -8.13088179e-01 3.52084339e-01
5.30864775e-01 5.65487072e-02 -2.01292530e-01 -1.10694736e-01
9.86476541e-01 1.46591976e-01 2.38362506e-01 3.11532944e-01
6.63984418e-01 -4.91134346e-01 5.93541145e-01 3.70341390e-01
-5.78081787e-01 5.08613288e-01 3.53341281e-01 -9.23545808e-02
-4.27257776e-01 -7.85677075e-01 1.88094601e-02 1.72815073e+00
3.77669670e-02 -4.92416412e-01 -6.42463088e-01 -9.23398197e-01
-4.21761602e-01 2.88494706e-01 -8.36900711e-01 2.95117870e-02
-1.51252940e-01 -6.72507107e-01 8.62475216e-01 5.61650157e-01
6.21361673e-01 -1.12029707e+00 -2.00696141e-01 3.23455572e-01
-7.49296486e-01 -1.41341817e+00 -4.66898113e-01 3.57243568e-01
-5.62427878e-01 -4.21166122e-01 -7.66282618e-01 -7.04678893e-01
8.09832215e-02 -6.46999404e-02 9.53839600e-01 -4.58827704e-01
1.28037632e-01 6.12086713e-01 -9.58602130e-01 -1.21794611e-01
2.48857319e-01 1.24873146e-01 -2.93627512e-02 6.49995267e-01
6.77464247e-01 -5.67263842e-01 -2.94450283e-01 2.75581658e-01
-9.51263905e-01 -1.85418844e-01 7.76370168e-01 1.04619801e+00
3.68975282e-01 8.52830857e-02 9.28575695e-01 -4.95396852e-01
1.00794947e+00 -4.86290693e-01 1.87919706e-01 5.60382128e-01
-6.65443614e-02 -8.65610689e-02 6.73216462e-01 -6.04276240e-01
-1.38801253e+00 -3.18027958e-02 -4.92141485e-01 -6.61951363e-01
-6.28308058e-01 7.13574350e-01 -5.88473976e-01 -3.62834632e-02
-2.40237504e-01 3.65103930e-01 -5.19144416e-01 -4.60316390e-01
6.71107113e-01 1.00714147e+00 3.87965232e-01 -8.39931130e-01
3.65501612e-01 1.35395736e-01 -5.91662228e-01 -7.08187103e-01
-8.32565725e-01 -5.50209820e-01 -4.92005408e-01 -3.57386440e-01
1.04277098e+00 -9.75978911e-01 -8.76905680e-01 3.95810157e-01
-1.07572186e+00 -7.57528692e-02 1.03451274e-01 5.86703837e-01
-5.53380191e-01 2.84868002e-01 -7.89193928e-01 -1.09528792e+00
-1.81620151e-01 -1.11373830e+00 1.59287131e+00 1.63815841e-01
-1.87169597e-01 -1.31157148e+00 1.28196001e-01 5.75479209e-01
4.80037510e-01 2.90949106e-01 6.85750425e-01 -5.93490958e-01
1.53387180e-02 -2.59554923e-01 -2.33566090e-01 4.45253760e-01
2.65091807e-01 -2.44088843e-01 -1.36798668e+00 2.81177144e-02
1.82890281e-01 -9.78012085e-01 8.02884400e-01 4.19831928e-03
1.10959578e+00 -4.11481820e-02 -1.66026488e-01 4.06133294e-01
1.20284104e+00 5.66839948e-02 5.57903826e-01 1.73966140e-01
7.68622220e-01 7.34512568e-01 7.75176227e-01 5.69585443e-01
8.18865776e-01 5.46881080e-01 1.12505093e-01 -1.38386622e-01
4.30530787e-01 -2.01023787e-01 6.22160017e-01 1.50515151e+00
1.50254771e-01 -3.67445648e-01 -4.62619126e-01 5.96560299e-01
-2.08705449e+00 -1.07352948e+00 4.66870487e-01 1.57104123e+00
7.96766758e-01 -2.02475309e-01 1.04771741e-01 4.65920754e-02
5.13212204e-01 4.86904621e-01 -1.37560323e-01 -9.03759539e-01
-2.33917445e-01 3.85593288e-02 -2.71158814e-01 3.74409497e-01
-1.17962110e+00 1.05311143e+00 5.90725327e+00 1.17813981e+00
-1.25018764e+00 4.59778219e-01 5.58104873e-01 -4.35453728e-02
-4.27798331e-01 -3.99434894e-01 -5.87177694e-01 5.59518456e-01
1.13067067e+00 1.06527247e-01 1.39584973e-01 4.56353903e-01
7.31636807e-02 2.28855148e-01 -1.05790973e+00 1.45959091e+00
4.06055421e-01 -7.39262402e-01 8.61010514e-03 -1.40975431e-01
5.98648727e-01 -2.49587372e-01 5.75284436e-02 9.56633329e-01
-6.52912678e-03 -1.16272819e+00 5.04599750e-01 5.52488804e-01
6.31442249e-01 -8.34015489e-01 9.85120952e-01 1.84926137e-01
-1.68679059e+00 7.05235600e-02 -6.67316932e-03 3.20397578e-02
5.33172727e-01 4.11186427e-01 -1.28320843e-01 8.40180516e-01
7.46640205e-01 6.18058562e-01 -2.82459229e-01 6.24579847e-01
3.43731008e-02 6.22385979e-01 -2.25811854e-01 -1.00072123e-01
4.87888843e-01 -1.69868842e-01 5.52562550e-02 1.67386711e+00
4.00259644e-01 3.22942764e-01 3.18088353e-01 2.80587822e-01
-2.96999753e-01 3.41708332e-01 -4.53917623e-01 -3.64805788e-01
4.50789243e-01 1.75562680e+00 -1.86835408e-01 -4.91032332e-01
-7.88965344e-01 1.43987274e+00 4.99550700e-01 5.22244215e-01
-9.18171942e-01 -4.87886101e-01 7.03849018e-01 -8.01424205e-01
4.43223298e-01 -4.22727391e-02 -4.37368192e-02 -1.51234746e+00
-3.36456746e-02 -9.75593269e-01 5.66069603e-01 -8.18427920e-01
-1.66075027e+00 1.07054412e+00 -3.45548987e-01 -1.29127014e+00
-2.78129876e-01 -6.19096637e-01 -6.02973998e-01 6.99573457e-01
-1.79482949e+00 -1.56908715e+00 -1.23890281e-01 8.24173212e-01
7.10018635e-01 1.67422310e-01 1.16436315e+00 5.39375126e-01
-6.16302669e-01 9.21978712e-01 -3.55212428e-02 3.30536179e-02
1.01290917e+00 -1.13970780e+00 -4.53026831e-01 3.66509616e-01
-1.55204348e-02 6.63660824e-01 4.33627218e-01 -2.04955444e-01
-1.51529515e+00 -8.91597211e-01 1.01747417e+00 -5.95910549e-01
8.08327675e-01 -5.54659724e-01 -8.71944427e-01 5.88322461e-01
7.92251825e-01 -8.87319744e-02 1.32782865e+00 5.90900004e-01
-6.59681916e-01 -2.38894105e-01 -9.97125328e-01 4.67455089e-01
5.46173513e-01 -1.01494706e+00 -9.14333999e-01 -2.47284725e-01
7.41190732e-01 -6.16682954e-02 -1.33148730e+00 8.52948129e-01
7.57117271e-01 -8.00617337e-01 5.24863720e-01 -4.53100562e-01
3.24884206e-01 -4.58157137e-02 -7.13605344e-01 -1.66060781e+00
-9.78512615e-02 -5.15766919e-01 -1.67568475e-01 1.57593238e+00
2.63051242e-01 -4.78750229e-01 4.13974464e-01 5.98513007e-01
-2.56234467e-01 -9.83040392e-01 -9.97515917e-01 -5.60466051e-01
-1.02062291e-02 -6.09086037e-01 6.78777814e-01 1.10897064e+00
7.33921647e-01 7.07843304e-01 -8.61283541e-01 6.70980737e-02
3.98197472e-01 3.43562603e-01 5.88174522e-01 -7.90408373e-01
-3.75763066e-02 -5.06584942e-01 -3.53294998e-01 -1.12553799e+00
5.13622820e-01 -3.64488959e-01 2.14736104e-01 -1.49424183e+00
2.55632371e-01 5.47347888e-02 -9.86621439e-01 6.01525247e-01
-2.50111490e-01 6.03146911e-01 7.23418444e-02 -3.00163478e-01
-1.20758986e+00 1.25295150e+00 9.36421990e-01 -6.67808428e-02
1.22009456e-01 -6.26194417e-01 -7.05826223e-01 7.01463282e-01
5.75072885e-01 -1.33309394e-01 -2.78939724e-01 -3.28867525e-01
-4.89933155e-02 1.22339748e-01 4.76395562e-02 -7.91622877e-01
2.35585555e-01 -1.25619724e-01 1.63419813e-01 -5.90162337e-01
9.39678490e-01 -1.04896641e+00 -2.81396180e-01 -3.91863674e-01
-4.17946309e-01 -4.46426868e-02 3.95042658e-01 4.29252326e-01
-9.55217063e-01 1.41736418e-01 3.94613475e-01 9.98324007e-02
-9.73625541e-01 3.85243982e-01 -6.66723788e-01 -1.51443496e-01
7.66376793e-01 -1.62866607e-01 -2.09807575e-01 -4.57249552e-01
-9.89915013e-01 4.13271129e-01 3.31195327e-03 7.99053907e-01
7.15725064e-01 -1.83998644e+00 -4.50052023e-01 -8.29642490e-02
5.05558610e-01 -7.79579103e-01 7.04416215e-01 1.04896557e+00
5.21859348e-01 3.91677260e-01 -8.94793719e-02 -2.77346760e-01
-1.33255517e+00 3.78089935e-01 2.53669024e-01 -2.98531502e-01
5.90840392e-02 8.32974374e-01 3.81929398e-01 -1.00430048e+00
2.66303509e-01 -3.28539878e-01 -2.67724961e-01 3.60919386e-01
5.30698955e-01 4.26900461e-02 -6.79898113e-02 -9.41298842e-01
-5.92254281e-01 8.41919839e-01 -5.62404767e-02 -3.48061383e-01
1.21867299e+00 -6.31297648e-01 -1.89874530e-01 1.09244096e+00
1.54506290e+00 7.49550611e-02 -8.98049951e-01 -5.04660189e-01
-1.83701128e-01 -3.36498946e-01 7.02229142e-02 -9.63469625e-01
-7.00128198e-01 1.36268866e+00 5.32974720e-01 2.57064164e-01
1.46434915e+00 -8.30491446e-03 1.05496097e+00 2.40928069e-01
3.30473632e-01 -1.28576529e+00 3.61951470e-01 7.34709680e-01
6.34373665e-01 -1.41118383e+00 -6.58811390e-01 -3.00979018e-01
-9.58457351e-01 9.80051339e-01 7.86720216e-01 2.55638331e-01
4.91353989e-01 1.91230923e-01 6.00202858e-01 -1.06907941e-01
-1.04344904e+00 -4.22599047e-01 5.15837848e-01 2.25829124e-01
8.09372365e-01 3.96758430e-02 -1.02761991e-01 1.10298967e+00
1.35318875e-01 8.40500519e-02 -5.92654645e-02 9.28417087e-01
-4.73374486e-01 -1.34069860e+00 -3.15519780e-01 2.34271124e-01
-5.13300002e-01 -2.69675821e-01 -4.01366562e-01 2.30772868e-01
1.64620996e-01 1.29083848e+00 -2.51101822e-01 -7.54852116e-01
3.35479498e-01 6.61414564e-01 4.44502115e-01 -3.52301747e-01
-6.77071095e-01 4.07973409e-01 3.44612092e-01 -6.32363498e-01
-7.25311518e-01 -4.96740043e-01 -9.45698798e-01 -6.31477758e-02
-2.46409640e-01 4.61795658e-01 5.25039494e-01 1.33721721e+00
4.77054030e-01 6.29277647e-01 7.85699308e-01 -9.51078057e-01
-3.30970228e-01 -1.20274782e+00 -6.55869126e-01 4.09244090e-01
3.30882311e-01 -5.92892945e-01 -3.20557028e-01 4.31710761e-03] | [13.235062599182129, 5.250998020172119] |
33d9cbb2-b6a6-453e-9c62-e93a7424dfac | remote-sensing-image-classification-using | 2206.13392 | null | https://arxiv.org/abs/2206.13392v1 | https://arxiv.org/pdf/2206.13392v1.pdf | Remote Sensing Image Classification using Transfer Learning and Attention Based Deep Neural Network | The task of remote sensing image scene classification (RSISC), which aims at classifying remote sensing images into groups of semantic categories based on their contents, has taken the important role in a wide range of applications such as urban planning, natural hazards detection, environment monitoring,vegetation mapping, or geospatial object detection. During the past years, research community focusing on RSISC task has shown significant effort to publish diverse datasets as well as propose different approaches to deal with the RSISC challenges. Recently, almost proposed RSISC systems base on deep learning models which prove powerful and outperform traditional approaches using image processing and machine learning. In this paper, we also leverage the power of deep learning technology, evaluate a variety of deep neural network architectures, indicate main factors affecting the performance of a RSISC system. Given the comprehensive analysis, we propose a deep learning based framework for RSISC, which makes use of the transfer learning technique and multihead attention scheme. The proposed deep learning framework is evaluated on the benchmark NWPU-RESISC45 dataset and achieves the best classification accuracy of 94.7% which shows competitive to the state-of-the-art systems and potential for real-life applications. | ['Alexander Schindler', 'Jasmin Lampert', 'Dat Ngo', 'Khoa Tran', 'Lam Pham'] | 2022-06-20 | null | null | null | null | ['remote-sensing-image-classification'] | ['miscellaneous'] | [ 2.80826062e-01 -5.55596590e-01 -1.70086980e-01 -4.05843943e-01
-5.92283249e-01 -1.17100574e-01 7.53459394e-01 6.98796883e-02
-3.93948853e-01 4.76761818e-01 2.13160723e-01 -5.84410846e-01
-2.97334462e-01 -1.22567272e+00 -3.73479009e-01 -7.06210554e-01
-3.00067693e-01 -7.87538104e-03 7.46745840e-02 -3.65861893e-01
2.69219488e-01 6.38738990e-01 -1.53786123e+00 1.59319565e-01
8.13778937e-01 1.33340812e+00 3.44822973e-01 4.54689443e-01
-1.40968129e-01 9.25975084e-01 -3.79909664e-01 7.75146335e-02
2.99112499e-01 1.39553472e-01 -8.17616940e-01 -2.57784873e-01
1.49651825e-01 -4.46134269e-01 -3.63614202e-01 1.05628192e+00
8.03163528e-01 1.56152472e-01 6.17466092e-01 -9.40759242e-01
-9.51571643e-01 5.49328148e-01 -8.54766071e-01 6.08166218e-01
-4.57851768e-01 -1.23253241e-01 1.07693827e+00 -8.85819674e-01
6.09094948e-02 1.20163310e+00 5.36978126e-01 1.87254906e-01
-7.04150617e-01 -7.77735293e-01 2.83766866e-01 5.28726101e-01
-1.54446781e+00 -2.10222036e-01 4.89523083e-01 -6.10804796e-01
7.59266376e-01 2.26886854e-01 2.50581056e-01 5.56828082e-01
1.68866873e-01 9.28552628e-01 1.16888857e+00 -1.64531350e-01
1.50923282e-01 -2.19628453e-01 3.16404074e-01 2.58827806e-01
3.14195990e-03 -3.60742658e-02 8.40862542e-02 1.50953218e-01
4.79241371e-01 5.03771663e-01 -2.85769731e-01 2.09875450e-01
-9.89924133e-01 9.86179709e-01 1.20556688e+00 3.60881865e-01
-3.91024858e-01 1.30611241e-01 5.73214173e-01 1.27703562e-01
7.68753767e-01 1.63953543e-01 -4.55736250e-01 4.60681170e-01
-5.63213348e-01 6.63770884e-02 8.31270590e-02 5.25399089e-01
7.37899959e-01 2.49978930e-01 -7.38836229e-02 1.16969049e+00
2.17872873e-01 5.38980544e-01 4.76264507e-01 -3.34209949e-01
2.54804343e-01 6.86803699e-01 -4.78838645e-02 -9.48162496e-01
-4.10059392e-01 -6.28148854e-01 -1.34406304e+00 1.53943449e-01
-3.93351734e-01 -4.86168079e-02 -9.20158088e-01 1.25668573e+00
1.65721372e-01 1.78214490e-01 2.37906530e-01 8.69821310e-01
1.04674911e+00 1.17300475e+00 4.03315872e-01 4.64967728e-01
1.08296990e+00 -9.53873575e-01 -3.03806692e-01 -1.76326603e-01
2.26771876e-01 -3.82881552e-01 9.42092955e-01 2.08310917e-01
-2.42016748e-01 -8.35563302e-01 -1.06365228e+00 6.97260024e-03
-7.14475214e-01 4.12572414e-01 7.10585892e-01 3.38276118e-01
-8.86693716e-01 6.23140872e-01 -6.92738414e-01 -7.78708756e-01
8.68300557e-01 -3.41419466e-02 -9.91896018e-02 -1.85401976e-01
-1.15462279e+00 6.74151897e-01 5.50071418e-01 5.79523146e-01
-9.43975329e-01 -6.18312955e-01 -6.25772595e-01 3.12438428e-01
1.09431855e-01 -2.23937500e-02 9.37570512e-01 -8.89267147e-01
-1.03499317e+00 7.60614395e-01 3.50321293e-01 -5.20954192e-01
2.07932428e-01 -3.93214613e-01 -7.81825364e-01 -4.90762442e-02
2.61198580e-01 7.44865179e-01 5.02006412e-01 -9.25527036e-01
-9.13108170e-01 -5.18349528e-01 6.62826374e-02 2.40976304e-01
-5.27796209e-01 1.74122557e-01 -8.48111138e-02 -4.82452452e-01
-1.90446079e-02 -6.54493034e-01 -2.48761579e-01 1.55952066e-01
-2.80571342e-01 -4.35908020e-01 1.24432456e+00 -6.60505772e-01
9.10985470e-01 -2.27919507e+00 -3.21722060e-01 -6.80793598e-02
1.29734546e-01 7.75114059e-01 -1.88791469e-01 4.00771081e-01
-5.75322509e-02 3.80183458e-01 -2.78359920e-01 2.42589116e-01
-9.22311619e-02 1.28345042e-01 -5.95786035e-01 5.38397729e-01
3.58639061e-01 7.67421067e-01 -6.28242016e-01 -1.94229707e-01
4.22540247e-01 4.24041182e-01 -8.35517496e-02 2.77064353e-01
-1.39181495e-01 3.65198880e-01 -7.86859870e-01 7.49472201e-01
8.45142186e-01 -4.14836019e-01 -1.08663559e-01 -1.76598251e-01
-3.48891288e-01 9.73588154e-02 -6.96878374e-01 1.12741160e+00
-3.25023621e-01 9.02292550e-01 -2.31733158e-01 -1.55997181e+00
1.12589514e+00 1.15303643e-01 4.52073961e-01 -8.09593678e-01
3.17927860e-02 1.50923878e-01 1.35001857e-02 -4.05625343e-01
3.81560713e-01 1.10974863e-01 1.08735777e-01 1.96702167e-01
-3.16744298e-01 2.60149539e-01 -2.76601821e-01 -1.21888854e-01
8.20535839e-01 6.54100403e-02 5.47620118e-01 -4.93912250e-01
5.85286558e-01 1.44943625e-01 2.84346819e-01 6.82381034e-01
-5.05565226e-01 3.90389591e-01 1.70815121e-02 -1.01956916e+00
-8.30220461e-01 -7.24294662e-01 -5.31533420e-01 1.26180673e+00
1.92953199e-01 2.96229780e-01 -3.39859337e-01 -3.49898577e-01
1.65707595e-03 3.80348176e-01 -7.44335234e-01 -5.76375909e-02
-1.44698009e-01 -1.14344466e+00 5.71959317e-01 6.49949312e-01
1.32090449e+00 -1.22856557e+00 -7.37042367e-01 1.43512771e-01
-5.31544648e-02 -1.14128745e+00 2.06769019e-01 1.20553989e-02
-6.67840183e-01 -9.04405832e-01 -7.15793133e-01 -7.17984140e-01
6.67272657e-02 8.96369457e-01 9.18017626e-01 1.04501188e-01
-3.80674094e-01 -3.81439808e-04 -6.50952458e-01 -6.04211152e-01
-3.84995118e-02 3.65492493e-01 -3.17676306e-01 2.01257110e-01
4.73963439e-01 -4.83065337e-01 -9.14472103e-01 2.27922991e-01
-1.02630460e+00 5.61895408e-02 7.56603777e-01 6.88799202e-01
3.59880447e-01 3.57072473e-01 8.37640584e-01 -7.30265856e-01
1.54131398e-01 -8.63765597e-01 -6.19142652e-01 2.12658644e-01
-5.01345217e-01 -2.84120560e-01 5.94801009e-01 1.53610677e-01
-9.70139682e-01 -2.45759681e-01 -3.57199311e-01 -1.10263795e-01
-2.31645480e-01 1.04078901e+00 -1.22746423e-01 -6.90916646e-03
7.24194646e-01 4.90222782e-01 -3.83734673e-01 -6.51357353e-01
7.10003078e-02 1.30502236e+00 3.76755267e-01 -2.88950324e-01
5.37579000e-01 5.35506070e-01 -3.46872956e-01 -1.16794658e+00
-1.21195745e+00 -7.84377098e-01 -3.41516584e-01 -1.25464037e-01
1.00050879e+00 -1.28668416e+00 -4.70522881e-01 7.68667579e-01
-9.16802943e-01 -3.01812977e-01 2.94157207e-01 3.22173625e-01
-7.69258430e-03 1.51969880e-01 -3.07363153e-01 -7.68361747e-01
-7.83405662e-01 -9.65266764e-01 1.20739615e+00 2.28427365e-01
4.52757329e-01 -7.28796303e-01 -7.83516839e-02 2.32226178e-01
9.50753510e-01 3.24634492e-01 1.04493845e+00 -5.29105067e-01
-6.11568689e-01 -8.08739364e-02 -8.94653618e-01 5.84033012e-01
2.43145958e-01 -3.64656858e-02 -1.15828872e+00 -2.15976506e-01
-4.65166867e-01 -4.96740252e-01 1.25791144e+00 4.36068445e-01
1.64778149e+00 -8.87255222e-02 -2.23617151e-01 7.11263239e-01
1.95244622e+00 2.28867307e-01 8.61142039e-01 6.74759507e-01
8.22102427e-01 5.17962873e-01 3.86188030e-01 5.26352465e-01
3.86439264e-01 3.76641333e-01 8.71139467e-01 -3.03152770e-01
1.36304393e-01 7.15300292e-02 3.36429402e-02 3.41798842e-01
-2.47636959e-01 -2.56932884e-01 -1.19362628e+00 6.72282338e-01
-1.80136311e+00 -1.15614069e+00 -6.82418346e-02 1.78894079e+00
2.33961448e-01 -1.28299624e-01 -2.60227025e-01 1.58507496e-01
7.74772882e-01 5.95325947e-01 -6.71202660e-01 -1.40506849e-02
-8.61596987e-02 2.22995058e-01 7.31095135e-01 -1.53966481e-02
-1.68428302e+00 1.14327705e+00 5.58621836e+00 7.33573258e-01
-1.70435154e+00 1.16196729e-01 8.42983425e-01 5.99738836e-01
2.27096081e-01 -4.46815103e-01 -6.32308722e-01 3.04527819e-01
7.03466594e-01 2.05450896e-02 1.47750720e-01 9.69812870e-01
3.51092696e-01 6.44561136e-03 -5.24570405e-01 7.75914609e-01
-2.42876738e-01 -1.37222373e+00 3.67798120e-01 -2.68432517e-02
7.47379065e-01 8.59631121e-01 1.90406173e-01 5.97974181e-01
4.11834836e-01 -1.12368476e+00 5.16831875e-01 1.29008412e-01
7.99178660e-01 -5.43001056e-01 9.41100001e-01 2.87271351e-01
-1.46401870e+00 -3.17035288e-01 -7.51000226e-01 -2.53421187e-01
-3.90067846e-01 4.13740903e-01 -4.93323445e-01 6.07046843e-01
1.07495153e+00 1.13460386e+00 -4.45255011e-01 1.00334382e+00
8.20958093e-02 9.12832320e-01 -3.63995135e-02 5.56077287e-02
5.90379596e-01 -1.78465992e-01 1.42776528e-02 1.13899851e+00
4.89772499e-01 3.59231800e-01 3.72039735e-01 6.78698003e-01
-1.63246602e-01 2.42429912e-01 -6.47040844e-01 -2.49472499e-01
3.26280445e-01 1.25810838e+00 -5.70229769e-01 -3.82743955e-01
-3.86871487e-01 6.40572190e-01 1.67905256e-01 3.79108965e-01
-6.81981325e-01 -3.07074964e-01 7.67870128e-01 -1.79507807e-01
3.81096631e-01 -2.17321917e-01 -2.13082001e-01 -1.07639146e+00
-2.78101385e-01 -7.05643952e-01 4.06920433e-01 -9.10018206e-01
-1.15461814e+00 8.46966565e-01 -1.60201415e-01 -1.36988389e+00
3.84132922e-01 -7.10971951e-01 -9.80323970e-01 9.29951727e-01
-2.45803380e+00 -1.27761698e+00 -8.15923393e-01 5.11120677e-01
7.18377471e-01 -3.09108347e-01 8.37494612e-01 4.60658252e-01
-8.08299661e-01 3.47313993e-02 4.53230143e-01 5.03420293e-01
3.45526934e-01 -1.01387894e+00 5.51518083e-01 1.01751268e+00
-1.64279863e-01 1.90575629e-01 1.18191406e-01 -1.01339556e-01
-1.17560148e+00 -1.93328702e+00 5.16632557e-01 3.81050229e-01
6.47763312e-01 1.34609845e-02 -8.80300224e-01 5.43810964e-01
3.27554680e-02 4.21147913e-01 5.48726022e-01 -1.11530259e-01
-3.98276806e-01 -6.28311455e-01 -8.59985113e-01 3.53343576e-01
6.96327150e-01 -5.41525245e-01 -1.75656959e-01 4.53039229e-01
5.69116712e-01 6.66887313e-02 -6.15576088e-01 5.10900736e-01
4.24749166e-01 -8.70304585e-01 8.87237728e-01 -5.55878580e-01
7.85125971e-01 -2.52537549e-01 -5.47331154e-01 -1.25776243e+00
-6.55507922e-01 1.23865508e-01 6.49710953e-01 8.14347446e-01
1.74894586e-01 -5.72407067e-01 4.22307938e-01 -2.96161827e-02
-3.68046284e-01 -4.54790771e-01 -6.53974354e-01 -4.45040077e-01
4.42464128e-02 -4.13491040e-01 7.08338141e-01 1.06194568e+00
-7.25923598e-01 4.09149587e-01 -4.25523221e-01 5.80048501e-01
4.86989766e-01 4.00655061e-01 6.56165600e-01 -1.44563341e+00
1.07560866e-01 -5.82096398e-01 -5.88777304e-01 -9.48187411e-01
8.57692361e-02 -8.10569644e-01 -1.18932826e-02 -1.83144438e+00
2.86296636e-01 -6.68315947e-01 -6.62105858e-01 8.09930563e-01
-2.73736238e-01 4.48429853e-01 -5.35867512e-02 4.89146650e-01
-5.32171905e-01 7.85662413e-01 8.02073836e-01 -6.77503288e-01
9.94176939e-02 1.00031786e-01 -8.98691237e-01 5.09711683e-01
1.02151978e+00 -1.29829839e-01 -1.50743589e-01 -9.33252692e-01
5.05998265e-03 -2.93130219e-01 4.93857294e-01 -1.17982244e+00
-8.15022513e-02 -1.26272887e-01 2.55056977e-01 -6.93103909e-01
-8.95503312e-02 -7.94686317e-01 -6.90821037e-02 5.90790451e-01
-2.20956713e-01 -3.19958717e-01 2.50884205e-01 4.95517552e-01
-5.12144983e-01 2.77108788e-01 9.95029747e-01 -9.16558579e-02
-1.51951993e+00 6.30006135e-01 -3.60099405e-01 -4.65755314e-01
8.44279587e-01 -2.57954467e-02 -4.41650540e-01 -1.46329328e-01
-4.00599450e-01 2.93021709e-01 -2.33336598e-01 5.40186882e-01
5.68731368e-01 -1.21983814e+00 -1.14605105e+00 7.08454251e-02
5.79742074e-01 4.64662910e-02 4.10712689e-01 5.09385943e-01
-7.91295767e-01 3.74718785e-01 -3.73484612e-01 -7.71350145e-01
-7.99615085e-01 2.54857779e-01 4.96302158e-01 -1.35223761e-01
-7.20576346e-01 5.15814364e-01 3.41950685e-01 -5.39992630e-01
-3.30703519e-03 -2.34029293e-01 -7.13961363e-01 -6.17367029e-02
6.39526725e-01 2.85624474e-01 2.74035841e-01 -6.86234474e-01
-2.52151042e-01 5.82706213e-01 1.15479259e-02 4.27071065e-01
1.65976012e+00 -5.54704815e-02 3.31081040e-02 2.39678830e-01
1.28578341e+00 -7.36288249e-01 -1.03754640e+00 -6.91461265e-01
-3.21471989e-02 -3.63663644e-01 6.83978319e-01 -8.48757684e-01
-1.30563927e+00 1.27092505e+00 9.99453127e-01 2.14655414e-01
1.28403687e+00 -2.33202025e-01 7.36868203e-01 6.60790980e-01
2.82504916e-01 -8.09221745e-01 -2.88558662e-01 6.64203584e-01
8.73664081e-01 -1.81644082e+00 -1.87639985e-02 -3.02851163e-02
-4.57544506e-01 9.18331265e-01 3.53846550e-01 -2.21507698e-01
8.79209876e-01 -8.20718929e-02 2.09088221e-01 -1.70234308e-01
-4.43473458e-01 -4.69034344e-01 2.34445646e-01 5.35451293e-01
4.04346555e-01 4.23769057e-01 -7.45236352e-02 2.45002687e-01
2.72888690e-01 -1.69429779e-02 1.57701954e-01 8.54529977e-01
-8.65010798e-01 -5.97565591e-01 -1.99274868e-01 6.76976979e-01
-5.58093309e-01 -3.63148212e-01 -9.02205985e-03 6.90910935e-01
3.85280675e-03 8.80506814e-01 1.66441619e-01 -3.16495001e-01
6.59567118e-02 -4.16998714e-01 -2.16479346e-01 -4.50564474e-01
-3.41556162e-01 -9.09366757e-02 -2.29418427e-01 -2.41749629e-01
-8.28947484e-01 -3.47305775e-01 -8.83581042e-01 -3.86582941e-01
-1.13862753e-01 6.60994053e-02 7.29102731e-01 1.13241184e+00
3.48906487e-01 4.08944041e-01 9.17049229e-01 -9.78169441e-01
-5.92119277e-01 -1.21363974e+00 -7.95196414e-01 2.13914216e-01
2.88626075e-01 -4.92159992e-01 6.99862018e-02 -9.50922295e-02] | [9.719043731689453, -1.3916746377944946] |
abd987c9-30a0-4762-935f-6590854a7a8d | optimizing-the-reliability-of-a-bank-with | 2004.11122 | null | https://arxiv.org/abs/2004.11122v1 | https://arxiv.org/pdf/2004.11122v1.pdf | Optimizing the reliability of a bank with Logistic Regression and Particle Swarm Optimization | It is well-known that disciplines such as mechanical engineering, electrical engineering, civil engineering, aerospace engineering, chemical engineering and software engineering witnessed successful applications of reliability engineering concepts. However, the concept of reliability in its strict sense is missing in financial services. Therefore, in order to fill this gap, in a first-of-its-kind-study, we define the reliability of a bank/firm in terms of the financial ratios connoting the financial health of the bank to withstand the likelihood of insolvency or bankruptcy. For the purpose of estimating the reliability of a bank, we invoke a statistical and machine learning algorithm namely, logistic regression (LR). Once, the parameters are estimated in the 1st stage, we fix them and treat the financial ratios as decision variables. Thus, in the 1st stage, we accomplish the hitherto unknown way of estimating the reliability of a bank. Subsequently, in the 2nd stage, in order to maximize the reliability of the bank, we formulate an unconstrained optimization problem in a single-objective environment and solve it using the well-known particle swarm optimization (PSO) algorithm. Thus, in essence, these two stages correspond to predictive and prescriptive analytics respectively. The proposed 2-stage strategy of using them in tandem is beneficial to the decision-makers within a bank who can try to achieve the optimal or near-optimal values of the financial ratios in order to maximize the reliability which is tantamount to safeguarding their bank against solvency or bankruptcy. | ['Vadlamani Madhav', 'Vadlamani Ravi'] | 2020-03-31 | null | null | null | null | ['electrical-engineering'] | ['miscellaneous'] | [-1.16889171e-01 1.38988391e-01 -1.75885595e-02 2.47938693e-01
-8.87398422e-02 -3.29670846e-01 1.33318156e-01 4.23817039e-01
-2.47055277e-01 9.93791223e-01 -2.84563988e-01 -7.10336566e-01
-5.64118385e-01 -9.96021748e-01 -2.66372561e-01 -9.98686492e-01
2.56441087e-01 4.01570499e-01 -1.06268570e-01 -2.96184253e-02
6.83523715e-01 8.22922468e-01 -1.29118574e+00 -5.03165841e-01
8.80502403e-01 1.23582697e+00 2.40985919e-02 2.58891374e-01
1.76730350e-01 6.32443607e-01 -4.14566249e-01 -2.06576630e-01
7.46891573e-02 -4.02031183e-01 -4.56325769e-01 3.08042496e-01
-9.45664108e-01 -8.08834955e-02 2.17903733e-01 8.29836905e-01
9.39242467e-02 1.27969459e-01 7.24396348e-01 -1.02653897e+00
-3.42103064e-01 8.48111659e-02 -3.92233610e-01 1.18808806e-01
3.26502919e-01 2.08313286e-01 9.63204801e-01 -7.25975573e-01
-1.37947425e-01 6.21296227e-01 1.13683708e-01 -6.59754723e-02
-9.21568632e-01 -3.04400086e-01 -7.31213689e-02 -3.40858176e-02
-1.27431953e+00 -3.99243474e-01 9.23001289e-01 -5.60526788e-01
7.07819700e-01 8.45059603e-02 6.54369891e-01 4.60520893e-01
5.41685462e-01 1.84189389e-03 9.52448905e-01 -4.93437529e-01
6.77300811e-01 2.07301617e-01 -3.28635499e-02 2.47387260e-01
7.62293756e-01 1.10478908e-01 -2.71765977e-01 -9.46473777e-02
5.99618375e-01 -9.68840718e-02 -2.14247182e-01 1.55191356e-02
-6.50166810e-01 8.13803673e-01 3.62187959e-02 5.63121200e-01
-7.65029430e-01 -2.21111581e-01 1.45291597e-01 9.14681479e-02
2.84722000e-01 5.99511862e-01 -2.08776131e-01 -6.85649663e-02
-7.51518726e-01 6.98415264e-02 1.03721941e+00 7.50427768e-02
4.84032631e-01 2.03252167e-01 3.67418259e-01 4.48122770e-01
4.98274148e-01 1.56806901e-01 4.60895598e-01 -4.91045922e-01
4.44126874e-01 7.36663103e-01 6.24212027e-01 -1.07709646e+00
-2.67937422e-01 -6.55395627e-01 -7.58791685e-01 3.42024088e-01
1.60200819e-01 -1.77355498e-01 -2.36502111e-01 1.14281833e+00
3.52635980e-01 2.15714633e-01 2.20323250e-01 8.08350205e-01
3.59127559e-02 6.44343376e-01 -1.28888085e-01 -8.44997764e-01
1.27163744e+00 -3.97369862e-01 -5.01163661e-01 -1.72337294e-01
-3.30708502e-03 -6.76868737e-01 7.63347268e-01 7.78671384e-01
-9.85073447e-01 -2.95475662e-01 -1.27026772e+00 8.20601225e-01
-1.53731301e-01 1.06680535e-01 2.29025960e-01 7.08859086e-01
-5.39688647e-01 4.92806375e-01 -7.70624399e-01 -2.93738935e-02
-8.45831633e-03 2.66169310e-01 -3.53910029e-02 2.13666663e-01
-9.78237748e-01 1.07448733e+00 3.98486912e-01 4.71142203e-01
-7.00515389e-01 -1.10213548e-01 -4.68331695e-01 3.01819623e-01
6.15045011e-01 -4.77087468e-01 7.57073283e-01 -5.34412324e-01
-1.41757369e+00 4.50526237e-01 1.92414239e-01 -3.77475709e-01
3.05214316e-01 -6.95409775e-02 -1.57916173e-01 2.50324886e-02
-1.02849022e-01 -4.62020427e-01 6.64968133e-01 -1.25094914e+00
-5.84279656e-01 -4.65688586e-01 5.72763868e-02 9.74835083e-02
-3.96819592e-01 -1.72853973e-02 -4.03020084e-02 -5.54166377e-01
2.69433588e-01 -7.30054080e-01 -2.71027595e-01 -6.56062782e-01
-4.07462448e-01 -2.19587192e-01 3.93450499e-01 -8.07124257e-01
1.30023777e+00 -2.07997012e+00 3.51704329e-01 5.20829260e-01
-1.86171740e-01 1.23499267e-01 5.26823759e-01 5.04536927e-01
-1.33127784e-02 6.25185221e-02 -3.78186285e-01 -3.73873450e-02
-2.77640790e-01 1.86303541e-01 -1.48389369e-01 3.87424052e-01
6.39432311e-01 2.27637678e-01 -5.24734974e-01 -2.31838256e-01
3.19935262e-01 1.39883652e-01 -2.70463705e-01 4.29675370e-01
3.04402411e-02 4.51729238e-01 -9.16852534e-01 8.16146970e-01
2.99212247e-01 -3.46568003e-02 4.75011408e-01 -8.50366727e-02
-2.30061844e-01 -2.24848487e-03 -1.20444322e+00 5.65670073e-01
-5.82317770e-01 1.28430724e-01 2.17408285e-01 -1.67775631e+00
1.08129895e+00 4.86321270e-01 8.90888453e-01 -5.07281065e-01
2.93931931e-01 4.02883202e-01 -1.99695393e-01 -6.10310733e-01
4.82642621e-01 -4.53138262e-01 -4.78588827e-02 4.06387866e-01
-6.75886154e-01 2.63443664e-02 1.86968341e-01 -2.92069018e-01
9.61762071e-01 -1.79958999e-01 6.39489174e-01 -6.41953126e-02
9.69240844e-01 -2.54010439e-01 4.63369429e-01 9.59256291e-02
4.82020564e-02 8.60944912e-02 8.81987333e-01 -7.72233382e-02
-8.80020380e-01 -7.89698124e-01 -1.15792289e-01 2.17794970e-01
2.47067381e-02 3.45529437e-01 -5.51493645e-01 -2.62165457e-01
1.38974398e-01 7.39005506e-01 -2.90254802e-01 -2.20671132e-01
-4.96123493e-01 -9.30913508e-01 -1.07490733e-01 8.54731128e-02
3.67966175e-01 -9.38285887e-01 -1.11303258e+00 6.46577597e-01
1.71937376e-01 -1.00603390e+00 2.84279138e-01 4.23576176e-01
-9.06026483e-01 -1.24256945e+00 -5.24769068e-01 -2.06378847e-01
6.85590565e-01 -9.07914564e-02 7.33053923e-01 4.53184366e-01
-3.24798152e-02 -1.68904401e-02 -4.89946127e-01 -4.27928679e-02
-4.58942771e-01 -1.77410364e-01 1.11613274e-01 1.43769696e-01
-1.31628871e-01 -5.26690423e-01 -4.88809377e-01 4.47762579e-01
-9.00961518e-01 -6.42946184e-01 6.26790643e-01 6.53686285e-01
3.68781835e-01 8.41879010e-01 1.14244974e+00 -3.52082610e-01
8.48307908e-01 -6.71377718e-01 -9.60007668e-01 4.26954269e-01
-8.85495186e-01 1.00567089e-02 6.84701562e-01 -2.20918074e-01
-8.45206678e-01 -1.23257786e-01 1.05983369e-01 -4.40279655e-02
1.77872535e-02 1.10060108e+00 -4.84820545e-01 1.17533989e-01
8.50113705e-02 3.13600689e-01 1.52164787e-01 -3.63104403e-01
-2.21922845e-01 7.28678644e-01 4.46592778e-01 -5.96541941e-01
7.46323764e-01 5.60764559e-02 3.30003977e-01 -4.99684900e-01
-5.94433963e-01 -2.45781898e-01 -2.27541119e-01 -5.46643019e-01
6.10667467e-01 -3.35515857e-01 -1.05180967e+00 4.15323079e-01
-7.64314890e-01 3.62105310e-01 1.23049663e-02 3.40037197e-01
-3.74244720e-01 5.57247341e-01 -9.14760306e-03 -1.46915448e+00
-3.19813937e-01 -1.28709126e+00 3.53233159e-01 3.94515514e-01
3.94200720e-02 -8.09193969e-01 -1.44945234e-01 5.64826965e-01
2.43459836e-01 5.17421722e-01 8.88537228e-01 -6.20789051e-01
-4.59042907e-01 -5.81516862e-01 6.45263866e-02 7.09815741e-01
3.25402081e-01 2.50226051e-01 -3.32218736e-01 -2.46324986e-01
6.03273571e-01 1.73542108e-02 1.87258616e-01 4.33987826e-01
6.77117527e-01 -3.64539087e-01 6.88046142e-02 2.44715065e-02
1.63542020e+00 7.41738439e-01 7.23046541e-01 6.45112157e-01
-1.25580430e-02 8.78764331e-01 1.07706225e+00 9.47280347e-01
2.24009573e-01 6.26306593e-01 6.21665716e-01 2.90181965e-01
6.63848102e-01 1.61759630e-01 3.78707051e-01 4.79144573e-01
-2.00016662e-01 -2.26213366e-01 -8.92468691e-01 2.87585050e-01
-1.58542681e+00 -6.54206991e-01 1.74370721e-01 2.61214757e+00
4.86617684e-01 6.23338997e-01 1.94874659e-01 6.22739077e-01
6.74172580e-01 -2.55033880e-01 -3.73594642e-01 -4.48662162e-01
1.00779727e-01 4.57329769e-03 3.70174140e-01 2.73760438e-01
-6.18341863e-01 3.90505135e-01 5.15062618e+00 6.50781393e-01
-1.12855208e+00 -4.08788055e-01 1.08164191e+00 4.96050995e-03
-1.88411668e-01 5.05571008e-01 -6.48685396e-01 6.86138690e-01
8.68329525e-01 -2.62474954e-01 5.23029685e-01 6.17269456e-01
7.63046920e-01 -7.33940303e-01 -7.43372977e-01 3.55572313e-01
-2.68506855e-01 -9.10276771e-01 -3.78869623e-01 2.80817360e-01
3.18530589e-01 -7.78684199e-01 -1.71532333e-01 -2.01536864e-01
-1.27992854e-01 -9.73344147e-01 7.57176518e-01 9.01980281e-01
1.46289945e-01 -9.90300298e-01 1.14907527e+00 5.59883535e-01
-9.57768261e-01 -5.70906460e-01 -1.46580741e-01 -1.89682528e-01
5.88504434e-01 1.10761392e+00 -5.80648124e-01 7.47118771e-01
4.55980122e-01 1.19130738e-01 -6.86953366e-02 9.45398629e-01
-2.04236344e-01 4.95405525e-01 -2.16576919e-01 -9.46493521e-02
1.74656603e-02 -6.70656919e-01 3.47498655e-01 1.87915653e-01
5.51067948e-01 3.76956582e-01 -1.21092461e-01 7.64173806e-01
3.15671086e-01 1.04883984e-01 -1.65802985e-01 -3.04037154e-01
6.89330995e-01 1.05186629e+00 -8.84226918e-01 -1.16768375e-01
-2.95078587e-02 2.89483458e-01 8.96551386e-02 2.89873958e-01
-7.57563770e-01 -2.36821368e-01 1.80091649e-01 1.88609183e-01
1.40804112e-01 -3.41774940e-01 -8.24776828e-01 -7.76528060e-01
3.05566221e-01 -8.09441507e-01 1.26214057e-01 -5.33919573e-01
-9.31095302e-01 3.98528576e-01 -9.63848382e-02 -1.02458310e+00
-4.16986257e-01 -5.18915415e-01 -7.58996725e-01 9.79744494e-01
-1.67571533e+00 -5.77351987e-01 2.58302968e-02 9.41026434e-02
9.94721651e-02 -3.15822452e-01 3.74351144e-01 4.96385968e-04
-1.09513938e+00 3.77558917e-02 2.13921309e-01 -2.67684579e-01
1.56156197e-01 -7.36812711e-01 -3.50227296e-01 9.19236481e-01
-4.15458262e-01 4.44472045e-01 1.04338312e+00 -7.54188061e-01
-1.28200817e+00 -2.72803992e-01 9.85800028e-01 2.23801374e-01
9.96070087e-01 2.62121379e-01 -7.52830029e-01 -1.18100129e-01
-2.17481434e-01 -4.63243544e-01 3.42870235e-01 -1.88990742e-01
4.22881991e-01 -3.68428588e-01 -1.28918028e+00 2.90925533e-01
8.85230154e-02 -1.39121532e-01 -4.26310211e-01 1.58763230e-01
2.74697244e-01 1.85345918e-01 -1.11772132e+00 2.93692052e-01
4.56623048e-01 -9.68694448e-01 9.32205081e-01 -3.07048023e-01
3.97747308e-01 -1.49795130e-01 -3.09162438e-02 -8.87683928e-01
1.79054178e-02 -4.15178984e-01 5.36370371e-03 1.20801699e+00
2.75375515e-01 -9.03019547e-01 7.80156016e-01 8.70592117e-01
-1.00294137e-02 -1.31214929e+00 -1.27873707e+00 -6.51763320e-01
-6.68717921e-02 -1.61453351e-01 5.36127150e-01 5.33259869e-01
6.34483099e-02 -1.18694738e-01 -2.75975227e-01 3.42806309e-01
5.97801447e-01 1.01317488e-01 4.79204923e-01 -1.10774672e+00
-5.10311782e-01 -4.40557063e-01 -6.63531870e-02 -3.93268496e-01
8.36649924e-05 -1.57991797e-02 -6.37627766e-02 -1.36302125e+00
-1.04192019e-01 -5.95389068e-01 -4.69173998e-01 1.84410885e-01
-2.24418640e-01 -3.46090764e-01 8.04295540e-02 3.71463686e-01
1.11858517e-01 4.47239757e-01 9.23745394e-01 2.23220751e-01
-2.61822015e-01 3.68229061e-01 -7.97093630e-01 4.69205022e-01
7.18054116e-01 -3.34556788e-01 -2.75323361e-01 1.50093272e-01
5.02963960e-01 8.67821217e-01 4.66233611e-01 -8.25269222e-01
2.57491291e-01 -6.83675289e-01 -8.70088562e-02 -4.53930974e-01
1.46479785e-01 -1.05376482e+00 4.11073208e-01 7.65529156e-01
2.47995585e-01 1.46283269e-01 -2.62301147e-01 3.97920221e-01
-3.53439569e-01 -8.02519679e-01 5.94504297e-01 9.54526514e-02
-7.74805248e-03 -2.35746965e-01 -6.57358110e-01 -4.38452423e-01
1.37234926e+00 -4.82059240e-01 -9.62517038e-02 -2.59053081e-01
-6.60434365e-01 2.92032480e-01 3.06825846e-01 -1.23619668e-01
6.57461107e-01 -6.85243726e-01 -6.27442718e-01 -1.22687645e-01
-3.31224978e-01 -1.50516242e-01 2.39894122e-01 9.17172670e-01
-6.33257866e-01 5.57312489e-01 -1.27498925e-01 -1.49076041e-02
-7.96829164e-01 4.95444238e-01 2.76877463e-01 -6.72751009e-01
6.84765205e-02 4.09289896e-01 -4.75048304e-01 3.63493413e-01
-3.69787030e-02 7.82966018e-02 -4.14931029e-01 1.95367321e-01
-3.81420441e-02 5.55066347e-01 2.64177978e-01 -5.11434555e-01
-5.10809600e-01 4.50028926e-01 3.64784449e-01 -2.14300990e-01
1.51389337e+00 -2.44591966e-01 -3.19452554e-01 1.37481838e-01
7.44272709e-01 5.46013862e-02 -1.11693883e+00 2.73997277e-01
3.49283785e-01 -3.48650903e-01 2.52938122e-01 -6.08328521e-01
-1.06964123e+00 4.13407892e-01 1.64990470e-01 6.21215224e-01
1.49340749e+00 -2.33537138e-01 4.83235240e-01 1.58617437e-01
2.90510207e-01 -1.18498182e+00 2.11869568e-01 2.85630357e-02
7.67206609e-01 -9.78210747e-01 3.57976586e-01 -1.92629308e-01
-5.36591709e-01 1.21232545e+00 3.19968075e-01 -6.48739040e-02
7.05743015e-01 2.58552700e-01 -3.86599928e-01 -3.95174697e-02
-5.93961716e-01 1.75443798e-01 1.75576195e-01 4.12966684e-02
2.69180149e-01 5.07353991e-03 -7.65935779e-01 7.43988216e-01
-3.33509594e-02 3.30041759e-02 5.72339892e-01 1.02448797e+00
-7.25981116e-01 -1.10993290e+00 -7.91885853e-01 3.16847891e-01
-5.13951421e-01 7.87774101e-02 -1.11882590e-01 6.03773534e-01
4.47525643e-02 1.31714845e+00 -1.07489400e-01 -1.94774896e-01
3.99470448e-01 -2.28791654e-01 -1.35220423e-01 -3.22153479e-01
-5.58018744e-01 1.88055083e-01 2.60161042e-01 -5.86490929e-02
-4.31934863e-01 -8.03506553e-01 -1.26449800e+00 -1.33732930e-01
-7.62836933e-01 5.08758128e-01 8.30404103e-01 1.22150254e+00
-2.59842277e-02 4.22729045e-01 1.07055247e+00 -6.26130700e-01
-9.21583176e-01 -7.28981674e-01 -7.41303921e-01 -3.02325916e-02
5.37853164e-04 -7.76949108e-01 -6.58310235e-01 -2.55745828e-01] | [5.8731689453125, 3.405808210372925] |
bfc519fc-4435-4a73-8685-abee734812b6 | dataset-of-propaganda-techniques-of-the-state | 2106.07544 | null | https://arxiv.org/abs/2106.07544v1 | https://arxiv.org/pdf/2106.07544v1.pdf | Dataset of Propaganda Techniques of the State-Sponsored Information Operation of the People's Republic of China | The digital media, identified as computational propaganda provides a pathway for propaganda to expand its reach without limit. State-backed propaganda aims to shape the audiences' cognition toward entities in favor of a certain political party or authority. Furthermore, it has become part of modern information warfare used in order to gain an advantage over opponents. Most of the current studies focus on using machine learning, quantitative, and qualitative methods to distinguish if a certain piece of information on social media is propaganda. Mainly conducted on English content, but very little research addresses Chinese Mandarin content. From propaganda detection, we want to go one step further to provide more fine-grained information on propaganda techniques that are applied. In this research, we aim to bridge the information gap by providing a multi-labeled propaganda techniques dataset in Mandarin based on a state-backed information operation dataset provided by Twitter. In addition to presenting the dataset, we apply a multi-label text classification using fine-tuned BERT. Potentially this could help future research in detecting state-backed propaganda online especially in a cross-lingual context and cross platforms identity consolidation. | ['Chu-Hsing Lin', 'Kai-Lai Chang', 'Chun-Ming Lai', 'Rong-Ching Chang'] | 2021-06-14 | null | null | null | null | ['propaganda-detection'] | ['natural-language-processing'] | [-9.23709869e-02 -1.45240426e-01 -8.23092580e-01 1.40901566e-01
-5.95363736e-01 -8.30317736e-01 1.45414364e+00 7.37519741e-01
-5.73939741e-01 4.53150630e-01 1.03862917e+00 -7.78730750e-01
1.25130313e-02 -9.14563179e-01 -2.01732054e-01 -5.36835194e-01
1.58101454e-01 2.62941748e-01 8.02493840e-02 -7.64574587e-01
7.62541950e-01 1.61457434e-01 -9.70927954e-01 5.23980021e-01
7.67401338e-01 3.12608778e-01 -1.33590147e-01 1.87750384e-01
-5.93787670e-01 1.30713451e+00 -9.82928276e-01 -4.92479354e-01
1.21943526e-01 -5.50889730e-01 -1.06511569e+00 -4.15066957e-01
5.32493353e-01 -1.09028637e-01 -2.28514113e-02 1.17296767e+00
4.53700364e-01 -4.37187314e-01 5.96874297e-01 -6.32422686e-01
-4.51201916e-01 1.34769869e+00 -6.58399045e-01 6.23702765e-01
5.28677464e-01 -2.38820925e-01 7.71593213e-01 -5.66582501e-01
1.13965607e+00 1.42620361e+00 5.17329574e-01 2.30869517e-01
-1.01972258e+00 -8.39402318e-01 -2.14980826e-01 1.28791317e-01
-9.44641113e-01 -1.89285770e-01 1.02411783e+00 -1.07378078e+00
1.31794110e-01 3.83631855e-01 8.37654233e-01 1.48134780e+00
2.44266465e-01 2.91801214e-01 1.81672800e+00 -6.61094189e-01
-1.52785897e-01 3.28276694e-01 2.81352580e-01 5.83750248e-01
2.46399656e-01 -2.80850947e-01 -6.71209037e-01 -4.19605404e-01
1.55407369e-01 -3.48645777e-01 1.01004198e-01 4.79175776e-01
-1.19326866e+00 1.59083760e+00 4.62199420e-01 1.02190459e+00
-3.21858495e-01 -1.84629247e-01 7.59795368e-01 4.85327125e-01
1.17019367e+00 5.10893822e-01 1.61843792e-01 -4.55711961e-01
-1.05393994e+00 1.60197496e-01 8.96137178e-01 -7.17946589e-02
5.45761287e-01 -1.45167977e-01 -5.18942028e-02 6.35427415e-01
3.34607661e-01 6.30281866e-01 3.77796263e-01 -1.42580375e-01
7.20019519e-01 7.67226875e-01 -3.45618159e-01 -1.65251601e+00
-5.16932845e-01 -3.90632242e-01 -3.84275436e-01 1.67288989e-01
5.56306779e-01 -2.67416090e-01 -6.03258669e-01 1.38116169e+00
5.30061543e-01 -4.33048844e-01 -5.25089324e-01 9.56576467e-01
8.62749517e-01 8.14716756e-01 2.86979318e-01 -2.06209436e-01
1.52226341e+00 -5.72947800e-01 -8.18846047e-01 7.99648836e-02
8.78126264e-01 -1.24010170e+00 8.22322607e-01 4.53262008e-04
-6.11096263e-01 -1.09289393e-01 -8.10145617e-01 2.64008164e-01
-7.94401824e-01 -3.55456203e-01 5.53649604e-01 1.05312812e+00
-4.43350136e-01 2.80206829e-01 -4.66938108e-01 -6.73980951e-01
4.58453417e-01 -4.62283045e-01 -2.76091456e-01 3.85699898e-01
-1.49050105e+00 1.33582342e+00 2.89019465e-01 -2.81095415e-01
-6.09142005e-01 -5.47031641e-01 -6.34740531e-01 -6.97380185e-01
3.43511909e-01 -5.28384037e-02 6.73476696e-01 -1.19364357e+00
-1.13096201e+00 1.14345729e+00 4.14170414e-01 -4.40456122e-01
4.93368298e-01 -3.24355811e-02 -6.69849157e-01 2.14641970e-02
5.88395178e-01 -4.92604300e-02 1.06251955e+00 -8.44743609e-01
-5.27082682e-01 -3.98925900e-01 1.61374688e-01 -1.79567590e-01
-5.88009179e-01 9.73779738e-01 5.56616306e-01 -8.94928753e-01
-3.90476100e-02 -8.92439544e-01 2.42182329e-01 -7.81041682e-01
-4.46356624e-01 -2.61304349e-01 1.18180692e+00 -1.13437247e+00
1.64263999e+00 -1.96216547e+00 2.56794430e-02 3.31589580e-01
5.28444171e-01 4.50232059e-01 2.81254470e-01 1.26884711e+00
2.43462116e-01 6.32886827e-01 4.25189376e-01 3.96446228e-01
-2.40275458e-01 -3.33176941e-01 -4.58074361e-01 9.50418770e-01
-1.33546799e-01 6.25649571e-01 -9.12878394e-01 -5.40202320e-01
-1.44387884e-02 1.35610476e-01 -2.64869869e-01 -6.60720825e-01
1.25857681e-01 5.86001158e-01 -7.35628545e-01 5.41738093e-01
5.78562379e-01 -2.01935589e-01 2.71890283e-01 -9.32815298e-02
-1.07655489e+00 7.22473443e-01 -6.70165956e-01 1.26992464e+00
-2.31067270e-01 1.26374054e+00 3.66617888e-01 -7.85619438e-01
5.98827302e-01 1.07531138e-02 3.83104563e-01 -7.06637621e-01
8.26881468e-01 1.47295237e-01 2.80978054e-01 -3.24650586e-01
6.54094398e-01 -1.47794455e-01 -4.89562869e-01 8.63133311e-01
-3.50515336e-01 3.75507325e-01 1.75239190e-01 5.39113939e-01
9.17945206e-01 -2.55812615e-01 5.28589427e-01 -5.00230134e-01
5.84939241e-01 6.20988607e-01 -2.34914068e-02 8.35667908e-01
-3.31097156e-01 -2.54604548e-01 2.17346966e-01 -5.46341598e-01
-7.76461959e-01 -7.49676347e-01 -4.88951892e-01 1.49336505e+00
3.06120627e-02 -6.93707585e-01 -6.13299251e-01 -7.04141319e-01
-2.36065999e-01 4.71510559e-01 -7.01356292e-01 4.31510299e-01
-6.44331992e-01 -5.79544365e-01 7.85607338e-01 -3.88324469e-01
8.16964686e-01 -7.45937526e-01 -6.37563348e-01 3.65832806e-01
-4.76175010e-01 -5.45565069e-01 -4.45717722e-02 -2.75013626e-01
-2.86840796e-01 -1.25587904e+00 -5.00430584e-01 -6.61893010e-01
2.59286672e-01 3.52840364e-01 4.89700258e-01 2.17907596e-02
-1.70983046e-01 9.51279700e-02 -7.50350714e-01 -6.44314110e-01
-1.17387879e+00 3.66308242e-01 6.04879968e-02 -8.06223601e-02
1.46223754e-01 -2.50194937e-01 -6.89513311e-02 6.92960620e-02
-7.92691469e-01 3.76215518e-01 3.61341506e-01 5.51977754e-01
-4.95492637e-01 -1.84387162e-01 6.17604911e-01 -1.15451777e+00
1.08863723e+00 -8.29250157e-01 -2.01328963e-01 -2.17934791e-02
-4.03253555e-01 -2.39952490e-01 2.85346717e-01 -5.63373685e-01
-1.14818215e+00 -7.97069252e-01 -4.21721563e-02 4.33347464e-01
1.35136753e-01 1.09596753e+00 7.75328517e-01 1.35006420e-02
1.16298246e+00 -1.77142665e-01 1.14996657e-01 -3.33385766e-01
5.33679545e-01 1.09119785e+00 -5.22360466e-02 -4.49410707e-01
1.13727450e+00 7.30706513e-01 -3.92324597e-01 -9.44777012e-01
-9.47861969e-01 -7.54998505e-01 -4.86073703e-01 -8.80826414e-01
9.64695215e-01 -8.19767177e-01 -5.91633916e-01 5.46050251e-01
-1.07993329e+00 -2.81232744e-02 2.67652363e-01 4.92266387e-01
2.48720244e-01 3.83113503e-01 -7.45338857e-01 -8.39565098e-01
-3.72955114e-01 -4.41028208e-01 3.17691147e-01 4.72012497e-02
-4.16832477e-01 -1.24404001e+00 5.77342808e-01 7.72320092e-01
7.52555609e-01 5.10856628e-01 7.51212120e-01 -9.56033051e-01
2.25523636e-01 -1.57372162e-01 -1.74777210e-01 -1.16896018e-01
4.71236587e-01 -5.38411215e-02 -5.94512463e-01 -1.29200086e-01
1.12392068e-01 -3.62870485e-01 7.46417761e-01 4.77146655e-02
8.47982615e-02 -9.17592168e-01 -2.32143357e-01 -3.55123401e-01
9.22077358e-01 -1.34214759e-01 1.99653924e-01 1.04349542e+00
4.08289641e-01 8.06901276e-01 6.92346692e-01 7.05283463e-01
2.57830739e-01 5.43097734e-01 1.96474105e-01 -5.13413921e-02
-2.84976572e-01 -4.38148826e-01 5.25431573e-01 9.57913458e-01
-4.70804453e-01 1.93822011e-01 -1.16303539e+00 2.90291518e-01
-1.53600764e+00 -1.64885187e+00 -6.43665016e-01 1.80377638e+00
9.47634816e-01 4.50794734e-02 3.92864823e-01 1.00058585e-01
9.77843583e-01 6.53823376e-01 4.99777704e-01 -5.70288897e-01
-3.57348740e-01 1.45665601e-01 6.52368844e-01 5.95233858e-01
-1.32750452e+00 1.07967842e+00 5.48632479e+00 1.23726666e+00
-1.52010190e+00 5.96428514e-01 1.83185458e-01 1.29561156e-01
-1.94431558e-01 1.82502449e-01 -5.97082973e-01 5.48927546e-01
8.03080976e-01 -1.91117883e-01 2.70685166e-01 7.19518542e-01
5.53870738e-01 -3.06046844e-01 -1.42237201e-01 6.49393380e-01
2.71612078e-01 -2.01568890e+00 -2.33876690e-01 3.92548084e-01
8.01969945e-01 2.67273545e-01 -1.08924881e-02 3.13484102e-01
1.51316106e-01 -5.63285410e-01 1.10109901e+00 -2.15393424e-01
3.15074265e-01 -1.83085188e-01 5.90956151e-01 5.96724868e-01
-6.92369938e-01 3.77004631e-02 2.25211158e-01 -5.34799933e-01
4.74787533e-01 4.22871470e-01 -1.06254411e+00 3.56436163e-01
2.68547982e-01 5.34444511e-01 -6.95251882e-01 4.88518655e-01
-3.30578953e-01 1.23133957e+00 2.15524472e-02 -4.99542743e-01
6.75905764e-01 -1.58821493e-01 9.20229852e-01 1.41780186e+00
-7.47466162e-02 -1.92554027e-01 2.73483545e-01 6.26896620e-01
2.31013656e-01 6.58663273e-01 -5.81986785e-01 -6.16881549e-01
4.31972921e-01 1.37292790e+00 -8.82134378e-01 -1.54320598e-01
-2.78623551e-01 6.28629625e-01 2.43376270e-01 -1.18306272e-01
-1.04193091e+00 9.74658057e-02 1.83135346e-01 7.80118287e-01
-5.58622181e-01 -5.15659928e-01 8.53976142e-03 -9.43951726e-01
-6.95343256e-01 -8.98832858e-01 5.20116925e-01 -3.09644621e-02
-1.43697083e+00 2.57906288e-01 1.41930237e-01 -8.35389912e-01
-2.83336211e-02 -4.33905035e-01 -6.14204884e-01 7.16671467e-01
-9.20858979e-01 -1.59671593e+00 -3.98820527e-02 4.52409267e-01
5.40645778e-01 -2.30896905e-01 5.46320915e-01 3.49134833e-01
-3.69511485e-01 -5.45468815e-02 -1.59193620e-01 5.44423223e-01
9.06099677e-01 -4.57230717e-01 -1.09962218e-01 7.35344708e-01
2.01381370e-01 7.28445590e-01 8.89039159e-01 -1.16342425e+00
-1.23472071e+00 -5.69446683e-01 1.05159974e+00 -4.67952281e-01
1.55339932e+00 -1.76702112e-01 -4.08040106e-01 3.17827374e-01
5.74816883e-01 -9.48566139e-01 9.07788396e-01 6.31492853e-01
-6.34590983e-01 5.26324511e-01 -1.07108104e+00 6.20392263e-01
8.32922220e-01 -5.53521216e-01 -7.41040349e-01 6.66941762e-01
4.66003492e-02 -1.41412854e-01 -5.55260897e-01 -2.35089734e-01
6.18312657e-01 -6.03325009e-01 5.28846920e-01 -9.08016086e-01
7.03809679e-01 -1.35635519e-02 8.00078809e-02 -1.15868974e+00
-4.50947195e-01 -6.95251882e-01 4.04127359e-01 1.36090112e+00
3.84260863e-01 -8.74290049e-01 5.74753106e-01 -2.20046733e-02
-5.75072207e-02 -1.03158681e-02 -1.09385157e+00 -6.14333332e-01
2.71521717e-01 -3.59799981e-01 -1.81712508e-01 1.66589940e+00
3.53201717e-01 5.89274406e-01 -5.13725400e-01 -2.29665220e-01
3.83755893e-01 8.39880258e-02 9.53219652e-01 -1.14055419e+00
1.54472008e-01 -7.45369852e-01 -4.07175452e-01 -4.93215650e-01
1.23750128e-01 -1.11477172e+00 -6.35007024e-01 -1.41328979e+00
3.88765305e-01 -5.33847094e-01 4.00281489e-01 2.70968467e-01
4.25843358e-01 3.62733871e-01 3.55531931e-01 5.79802632e-01
-1.40325487e-01 1.51840439e-02 1.21102905e+00 -4.21865791e-01
-4.00157511e-01 -1.60612375e-01 -7.81338632e-01 8.71531844e-01
9.39549565e-01 -6.80921674e-01 2.55819440e-01 3.59736606e-02
6.04114115e-01 -3.74748081e-01 3.30669612e-01 -5.71924031e-01
3.14114392e-01 -6.20024323e-01 -1.22266188e-01 -6.04137719e-01
-1.13871414e-02 -3.45371276e-01 1.41954944e-01 9.08704579e-01
-1.80541381e-01 -1.58622965e-01 -7.99613893e-02 3.48714828e-01
-2.36776203e-01 -3.11041504e-01 5.77818513e-01 1.26575574e-01
-4.82273251e-01 -2.70646125e-01 -1.09268451e+00 2.08289444e-01
9.73118305e-01 -1.08036309e-01 -1.16817725e+00 -4.74099874e-01
-5.21505475e-01 -3.16158056e-01 2.61909723e-01 5.54431200e-01
-1.06857330e-01 -1.27293313e+00 -1.21447551e+00 -2.67473847e-01
8.97751674e-02 -1.23292673e+00 4.50636558e-02 1.23148823e+00
-8.57310474e-01 3.76992881e-01 -2.44567618e-01 -1.26105323e-01
-1.55068147e+00 1.92900971e-01 -2.32200012e-01 -1.93714485e-01
-5.09595037e-01 4.62679148e-01 -3.91738534e-01 -7.83528537e-02
-3.80641609e-01 3.13896298e-01 -4.80144799e-01 7.93455601e-01
6.87738478e-01 6.95281863e-01 -4.10306305e-01 -1.24815190e+00
-3.92171115e-01 -1.70202472e-03 -2.20645573e-02 -4.50048774e-01
1.00633037e+00 -1.37062166e-02 -5.22641659e-01 5.63009739e-01
8.57801855e-01 1.09551549e+00 -1.07485965e-01 -2.79504843e-02
8.75147134e-02 -7.31620491e-01 -1.55132068e-02 -9.48266387e-01
-3.76602203e-01 3.74696255e-01 3.69697452e-01 9.43706036e-01
3.53865921e-01 5.05343437e-01 3.26130897e-01 -2.90472712e-02
4.45247501e-01 -1.41898155e+00 2.35328421e-01 6.54622376e-01
1.04695404e+00 -1.13675690e+00 1.86380118e-01 -4.80289698e-01
-4.61823255e-01 1.08801556e+00 4.38809916e-02 2.01252863e-01
7.34160841e-01 -9.33180377e-03 3.05401176e-01 -5.66643655e-01
1.37765817e-02 -2.41556793e-01 3.15228254e-01 3.45063120e-01
6.74720049e-01 3.15466732e-01 -1.33911753e+00 4.35566038e-01
-4.01173621e-01 -6.66807294e-01 4.90008414e-01 8.46290827e-01
-9.16642904e-01 -1.26843464e+00 -7.72040486e-01 4.68941957e-01
-8.75781417e-01 -2.42673576e-01 -1.05860865e+00 1.09189236e+00
2.98304021e-01 1.28854477e+00 -2.10848272e-01 -5.87989688e-01
-3.07548910e-01 -2.56027132e-01 4.40623403e-01 -5.72312355e-01
-1.30107141e+00 -6.09231414e-03 8.53043973e-01 1.95694398e-02
-8.69860530e-01 -7.72760272e-01 -8.44847262e-01 -1.07299685e+00
-5.57312071e-01 2.47304574e-01 1.21612287e+00 9.96765077e-01
-5.65139763e-02 -2.73126345e-02 6.65080190e-01 -5.04909098e-01
-4.25886452e-01 -1.26194167e+00 -1.96227044e-01 4.01422262e-01
-3.07767719e-01 -3.84952217e-01 -3.82310629e-01 -2.43418947e-01] | [8.55502700805664, 10.42491340637207] |
a805ba6a-0c11-4d67-904f-dadb8a5b70be | generalizing-brain-decoding-across-subjects | 2205.14102 | null | https://arxiv.org/abs/2205.14102v2 | https://arxiv.org/pdf/2205.14102v2.pdf | Group-level Brain Decoding with Deep Learning | Decoding brain imaging data is gaining popularity, with applications in brain-computer interfaces and the study of neural representations. Decoding is typically subject-specific and does not generalise well over subjects, due to high amounts of between subject variability. Techniques that overcome this will not only provide richer neuroscientific insights but also make it possible for group-level models to outperform subject-specific models. Here, we propose a method that uses subject embedding, analogous to word embedding in Natural Language Processing, to learn and exploit the structure in between-subject variability as part of a decoding model, our adaptation of the WaveNet architecture for classification. We apply this to magnetoencephalography data, where 15 subjects viewed 118 different images, with 30 examples per image; to classify images using the entire 1s window following image presentation. We show that the combination of deep learning and subject embedding is crucial to closing the performance gap between subject- and group-level decoding models. Importantly, group models outperform subject models on low-accuracy subjects (although slightly impair high-accuracy subjects) and can be helpful for initialising subject models. While we have not generally found group-level models to perform better than subject-level models, the performance of group modelling is expected to be even higher with bigger datasets. In order to provide physiological interpretation at the group level, we make use of permutation feature importance. This provides insights into the spatiotemporal and spectral information encoded in the models. All code is available on GitHub: https://github.com/ricsinaruto/MEG-group-decode | ['Mark Woolrich', 'Oiwi Parker Jones', 'Mats Van Es', 'Richard Csaky'] | 2022-05-27 | null | null | null | null | ['brain-decoding', 'brain-decoding'] | ['medical', 'miscellaneous'] | [ 3.19503218e-01 2.21688077e-02 2.59402454e-01 -5.66004217e-01
-6.77822113e-01 -4.35152322e-01 5.35965741e-01 1.58642054e-01
-5.80849707e-01 4.04060245e-01 4.74006534e-01 -3.44261140e-01
-3.67115796e-01 -2.18775243e-01 -5.63881576e-01 -6.41300917e-01
-5.77016413e-01 -6.70078793e-04 -7.22329915e-02 -9.20483749e-03
1.94768697e-01 2.76057184e-01 -1.31746948e+00 3.76133710e-01
5.90892494e-01 7.38947928e-01 3.04921150e-01 4.32521969e-01
4.32494730e-01 1.06194071e-01 -4.41347003e-01 -5.57255074e-02
4.67557460e-02 -6.81475580e-01 -5.54548562e-01 -2.22908050e-01
4.70704675e-01 4.07884717e-02 -3.34960848e-01 1.05945599e+00
8.76825869e-01 1.81293830e-01 8.53495777e-01 -8.09883475e-01
-7.63556540e-01 5.12832403e-01 -3.87455851e-01 6.45760238e-01
1.89349189e-01 2.96830028e-01 9.95904863e-01 -6.80853844e-01
4.20921832e-01 1.17112947e+00 5.15124738e-01 4.53309655e-01
-1.61497295e+00 -8.25902700e-01 1.70047149e-01 4.57973093e-01
-1.24989831e+00 -6.18996918e-01 7.15079010e-01 -7.83193409e-01
9.70952451e-01 1.44060850e-01 9.59836721e-01 1.31928873e+00
6.09577060e-01 3.38073283e-01 1.35029578e+00 -2.65129924e-01
1.43547609e-01 9.03942510e-02 4.00448412e-01 2.97786683e-01
2.85736382e-01 2.77177840e-01 -6.32980764e-01 -1.27420917e-01
7.10911095e-01 -2.81375468e-01 -6.82942867e-01 -2.09117115e-01
-1.49032009e+00 8.71876836e-01 6.72232330e-01 6.56715870e-01
-5.61487913e-01 1.91841617e-01 3.47009301e-01 3.13423097e-01
5.47967136e-01 6.64545953e-01 -4.39657688e-01 -1.07133940e-01
-1.17393398e+00 2.75532156e-01 5.22251427e-01 3.20468962e-01
4.77208436e-01 1.38258442e-01 -7.14914128e-02 1.03798997e+00
3.00719112e-01 2.01654732e-01 8.31097424e-01 -7.94764161e-01
1.07732683e-01 4.76447009e-02 -4.56623495e-01 -1.11451733e+00
-8.39094102e-01 -6.59395337e-01 -7.54983008e-01 1.57154411e-01
4.26090747e-01 -3.28766853e-02 -8.53323877e-01 2.12728667e+00
-2.58354604e-01 2.17441157e-01 -3.34083140e-01 1.09382570e+00
5.46513736e-01 4.50709343e-01 2.89365798e-01 -1.89782411e-01
1.68797171e+00 -2.32109532e-01 -4.99345601e-01 -4.41109061e-01
5.42695224e-01 -4.28377360e-01 9.47520733e-01 4.19091821e-01
-1.06828964e+00 -4.73101348e-01 -1.08717287e+00 3.93182319e-03
-3.72520089e-01 -1.05561517e-01 6.74478471e-01 7.65006244e-01
-1.23418021e+00 5.26208639e-01 -8.52532029e-01 -5.53419650e-01
4.24349815e-01 5.75737953e-01 -4.51259971e-01 8.53889361e-02
-1.32713103e+00 1.12417305e+00 2.69006103e-01 1.05853938e-01
-6.27881944e-01 -1.06806946e+00 -7.94893503e-01 5.65718338e-02
-1.26251653e-01 -6.12000883e-01 7.85724342e-01 -8.48046899e-01
-1.07973540e+00 8.48563135e-01 -2.34772265e-01 -3.61540556e-01
1.13616750e-01 1.38598680e-01 -4.36212093e-01 8.80302563e-02
-3.55172344e-02 8.49437058e-01 8.31664860e-01 -9.22638118e-01
9.38694626e-02 -6.29354119e-01 -3.73803824e-01 4.76299226e-02
-1.88184276e-01 1.00179933e-01 1.31553959e-03 -7.24844396e-01
3.98362070e-01 -8.32170248e-01 -2.01202016e-02 -1.23541318e-01
9.64308809e-03 -9.14018974e-02 2.76185036e-01 -1.09610534e+00
7.98253179e-01 -2.32936931e+00 1.90927431e-01 3.02144796e-01
4.47071642e-01 -5.95051087e-02 -2.92794347e-01 4.52338189e-01
-5.35054803e-01 2.05488712e-01 -4.21164483e-01 -5.13939857e-02
-5.62335476e-02 -2.60116875e-01 -1.01072811e-01 8.16898942e-01
3.51893812e-01 9.21990931e-01 -5.72797060e-01 1.99640796e-01
7.31475577e-02 6.69169903e-01 -8.85847330e-01 -2.69896537e-01
2.20509455e-01 7.53557682e-01 9.16604325e-03 1.99130967e-01
5.74871838e-01 -8.29290971e-02 3.51417065e-02 -4.21942115e-01
9.16356072e-02 5.27224958e-01 -9.03328776e-01 1.57010782e+00
-3.56164455e-01 1.05368495e+00 1.46309718e-01 -1.37216306e+00
4.56833422e-01 3.36280763e-01 5.58693290e-01 -9.38904941e-01
1.70704424e-01 2.15913698e-01 9.95097578e-01 -5.15911102e-01
9.32203699e-03 -3.57496232e-01 1.15830833e-02 5.50178766e-01
4.43353891e-01 3.97662558e-02 -1.10804930e-01 1.16997488e-01
9.79253650e-01 -2.67515033e-01 1.57211959e-01 -6.60807014e-01
1.61521882e-01 -5.66914439e-01 1.46800905e-01 7.38561988e-01
-1.97480887e-01 6.93530798e-01 5.98761857e-01 -7.41665959e-02
-7.83821762e-01 -1.11280882e+00 -7.50745118e-01 8.88594866e-01
-3.03316563e-01 -3.06614578e-01 -5.90799153e-01 1.22112490e-01
-4.97898310e-02 7.18449354e-01 -7.46834159e-01 -5.42488635e-01
-3.22286397e-01 -1.12355828e+00 4.51091141e-01 2.96601951e-01
9.64982286e-02 -9.94540513e-01 -5.38421273e-01 2.70412296e-01
9.02517978e-03 -9.73026693e-01 -2.86073893e-01 3.17891151e-01
-8.90095115e-01 -7.38495588e-01 -7.98519433e-01 -5.50251782e-01
4.78279620e-01 -1.43175691e-01 8.08107316e-01 -3.41600291e-02
-4.75921720e-01 5.39638281e-01 -1.57735139e-01 -4.91864264e-01
-4.53094095e-02 4.26695272e-02 3.24054360e-01 -3.15713398e-02
3.65646660e-01 -8.76473129e-01 -7.58739412e-01 5.62610626e-02
-8.70920956e-01 -1.02618583e-01 6.16701603e-01 8.77242208e-01
7.18791336e-02 -2.77343661e-01 7.18240619e-01 -5.42241633e-01
9.34049785e-01 -6.76537275e-01 -3.05875748e-01 -2.66515166e-01
-4.85518873e-01 4.65921424e-02 3.10614228e-01 -6.31058276e-01
-3.12771618e-01 -5.48075497e-01 -2.41220370e-01 -3.05909719e-02
-1.17261291e-01 8.25045526e-01 8.40794444e-02 -1.96053833e-01
7.80969799e-01 1.97334990e-01 2.56913811e-01 -4.36766654e-01
1.36457250e-01 5.14251411e-01 2.14129135e-01 -4.39531982e-01
3.35599124e-01 1.87164307e-01 -1.54324487e-01 -1.05230784e+00
-2.20234826e-01 -3.02211940e-01 -6.24403358e-01 -1.92541167e-01
9.81067657e-01 -9.37219441e-01 -6.28912091e-01 2.04399735e-01
-9.98456776e-01 -4.90679711e-01 1.47547601e-02 9.17657435e-01
-4.93905187e-01 2.49254070e-02 -4.64325130e-01 -5.64046264e-01
2.15950515e-02 -1.27397561e+00 1.00774741e+00 -1.32202387e-01
-4.81174946e-01 -1.16793215e+00 -1.52852237e-01 1.92701966e-01
4.69698787e-01 6.53010160e-02 1.13554871e+00 -7.04885006e-01
-3.62445742e-01 -1.73158109e-01 -4.26698327e-02 2.91346133e-01
4.23553139e-02 -6.18884563e-01 -1.14312947e+00 -3.99251401e-01
3.16129774e-01 -4.45319898e-03 1.11300766e+00 7.45474160e-01
1.36863196e+00 -1.22336522e-01 -1.35860592e-01 5.86128294e-01
1.15235198e+00 1.08622447e-01 7.88183272e-01 5.80978505e-02
6.22640431e-01 8.22290838e-01 -3.37862790e-01 4.91705239e-02
2.75150567e-01 7.99979925e-01 -2.32822057e-02 7.42671192e-02
-9.60413069e-02 2.35530764e-01 5.42842329e-01 9.51508045e-01
-1.82199124e-02 5.80610894e-02 -1.07610798e+00 6.26092076e-01
-1.34510899e+00 -1.13325226e+00 -2.69747823e-01 2.25079894e+00
7.06365824e-01 -2.98021771e-02 2.01718658e-01 2.06533819e-01
3.51919949e-01 2.18956038e-01 -4.29641336e-01 -5.01070023e-01
-5.99990375e-02 4.43412036e-01 3.46951425e-01 5.59544981e-01
-6.99517012e-01 5.39061427e-01 6.35075474e+00 5.21258533e-01
-1.35822439e+00 3.83081347e-01 5.08774221e-01 -3.16374540e-01
-2.26339951e-01 -2.21456289e-01 -3.58977586e-01 5.84464550e-01
1.41847563e+00 -1.91142514e-01 8.03628743e-01 1.43687874e-01
4.66018826e-01 -2.70108908e-01 -1.28180051e+00 1.09498096e+00
1.62346601e-01 -1.07542372e+00 -2.44452268e-01 3.51900876e-01
3.51681948e-01 1.77746162e-01 2.33312607e-01 2.08225399e-01
-5.09722769e-01 -1.23058856e+00 6.88640773e-01 4.80184555e-01
6.22154593e-01 -3.67658198e-01 4.91442770e-01 2.77112752e-01
-7.36957312e-01 1.01391496e-02 -4.87917930e-01 -1.67349815e-01
8.77762511e-02 6.12419963e-01 -5.82892895e-01 2.53873825e-01
6.74299955e-01 8.77901375e-01 -7.16795504e-01 1.08721256e+00
4.86540347e-02 8.78906429e-01 -1.81952596e-01 5.08386865e-02
-3.24914604e-02 -9.43795592e-02 6.11090243e-01 1.19162226e+00
3.34716171e-01 3.29593360e-01 -2.64005512e-01 1.09659386e+00
1.93174079e-01 1.38604030e-01 -6.23644888e-01 -1.80082157e-01
1.24806054e-01 1.09104264e+00 -6.64310575e-01 -6.86938465e-02
-3.57828200e-01 9.22188461e-01 1.31623372e-01 5.75074136e-01
-5.43343008e-01 -2.38229603e-01 7.53429055e-01 1.96249291e-01
8.82918313e-02 -6.69532776e-01 -5.68293810e-01 -1.12877107e+00
1.34167507e-01 -7.91000247e-01 5.11922128e-02 -8.42292488e-01
-1.40939975e+00 4.14997041e-01 2.98163384e-01 -8.37673545e-01
-3.78221005e-01 -9.39277172e-01 -4.47133482e-01 1.25438833e+00
-1.02716792e+00 -7.44010568e-01 -2.22876687e-02 3.95658374e-01
3.19675535e-01 7.81521127e-02 8.18367004e-01 3.81390899e-01
-2.52087325e-01 5.28636396e-01 7.42933601e-02 8.33624136e-03
7.83925414e-01 -9.85175490e-01 1.49998173e-01 6.44507587e-01
3.54430348e-01 1.16828799e+00 7.56045878e-01 -4.85054433e-01
-1.15303493e+00 -6.88009918e-01 6.21331632e-01 -6.54513657e-01
6.92154706e-01 -9.09028232e-01 -1.21545470e+00 6.38523698e-01
3.38482499e-01 -2.08826005e-01 1.00614679e+00 4.57679927e-01
-1.98749140e-01 3.07415146e-02 -8.38380754e-01 7.44243860e-01
1.09914982e+00 -8.22007239e-01 -6.54808760e-01 3.59992296e-01
3.66563231e-01 -6.72252402e-02 -9.38356400e-01 2.64683455e-01
7.70804346e-01 -7.84489572e-01 8.89100969e-01 -4.77378488e-01
1.56945869e-01 -1.58190411e-02 -8.17123801e-02 -1.78161550e+00
-6.59371793e-01 -1.47652835e-01 2.78285235e-01 8.59193444e-01
5.05710900e-01 -1.06140268e+00 2.35503376e-01 6.11886799e-01
-1.01149403e-01 -7.67923295e-01 -1.05832267e+00 -8.57467711e-01
5.24158478e-01 -8.33365440e-01 1.85466886e-01 8.43995571e-01
1.91876099e-01 3.22167575e-01 -9.78877991e-02 2.27207512e-01
5.12243271e-01 -5.08643389e-01 3.19514908e-02 -1.20640266e+00
-3.75363648e-01 -7.52677739e-01 -8.81914020e-01 -5.96854150e-01
2.15155348e-01 -1.22168291e+00 -4.41079922e-02 -1.45041120e+00
1.03160053e-01 -2.32070580e-01 -6.16931200e-01 4.87254381e-01
-2.10470632e-01 5.25806189e-01 2.45807350e-01 2.37409040e-01
6.16411343e-02 4.21453297e-01 8.43044043e-01 -1.29340906e-02
-1.73283651e-01 -3.18153203e-01 -9.19981658e-01 1.85517699e-01
8.98882866e-01 -4.20689821e-01 -5.38677454e-01 -4.32283074e-01
1.88564435e-02 -1.87954992e-01 7.59757519e-01 -1.31947172e+00
-3.64037347e-04 2.98244536e-01 7.57421374e-01 1.04766339e-01
3.79574209e-01 -5.59865654e-01 1.49217203e-01 4.35950339e-01
-4.44898695e-01 -2.74046957e-02 5.45021772e-01 3.19846392e-01
-7.08947284e-03 -2.73758713e-02 7.33645976e-01 1.73892863e-02
-4.91866052e-01 2.47086480e-01 -7.89582670e-01 1.80232804e-02
4.79689211e-01 -3.56054366e-01 -1.11754842e-01 -4.99164283e-01
-9.71006095e-01 -9.71523672e-02 1.38606578e-01 5.64871788e-01
4.18190449e-01 -1.25815690e+00 -7.21564651e-01 5.94818234e-01
1.08543977e-01 -7.18087375e-01 4.77376461e-01 1.33088219e+00
9.78980120e-03 5.64689100e-01 -3.80162150e-01 -8.09489250e-01
-8.68851244e-01 2.30933115e-01 5.28290153e-01 2.77820975e-01
-5.03681242e-01 9.25840676e-01 5.78353047e-01 -1.94544628e-01
-8.66149291e-02 -4.85098511e-01 -1.70188621e-01 3.91392678e-01
5.93901634e-01 -7.71614090e-02 3.30269039e-01 -7.31014192e-01
-6.64080858e-01 5.06078899e-01 -6.17183857e-02 -5.98878086e-01
1.56142354e+00 5.67502156e-02 -1.62211463e-01 8.08399975e-01
1.43931377e+00 -1.18570298e-01 -1.03265452e+00 1.09716542e-01
-6.38437867e-02 -2.41593689e-01 2.42058471e-01 -9.33173358e-01
-8.73457968e-01 1.27699053e+00 1.01819801e+00 1.77715287e-01
1.04986656e+00 1.39179975e-01 2.57829934e-01 -7.07039684e-02
3.43301922e-01 -7.84595490e-01 -7.51662999e-02 2.85432726e-01
1.19857275e+00 -1.10500467e+00 4.08822447e-02 -9.43187475e-02
-4.76190478e-01 9.76534605e-01 2.22815439e-01 -3.33109558e-01
9.11866128e-01 -8.78726318e-02 -1.48868814e-01 -4.48267937e-01
-6.09637618e-01 -4.68258299e-02 6.64856076e-01 6.26289606e-01
8.34129453e-01 1.36617661e-01 -5.07490218e-01 8.03135157e-01
-4.12295073e-01 -2.09382400e-01 2.77467310e-01 5.81501067e-01
-2.16485411e-01 -1.02392948e+00 -3.52990031e-01 9.13877666e-01
-4.86264437e-01 -3.79582196e-01 -5.48333004e-02 6.18417740e-01
5.23758791e-02 8.07060599e-01 1.86840951e-01 -3.81440639e-01
2.25946516e-01 4.28156793e-01 7.27122724e-01 -8.07505012e-01
-4.31859076e-01 2.94855516e-02 -2.57786531e-02 -4.80393142e-01
-3.24132264e-01 -9.49361026e-01 -9.24478710e-01 -5.19722886e-02
-2.61393469e-02 -4.01183143e-02 8.19284797e-01 8.67651165e-01
4.37026352e-01 7.25360692e-01 1.78566411e-01 -1.22271395e+00
-2.33644068e-01 -1.18984008e+00 -6.23026729e-01 1.67260930e-01
4.86632526e-01 -7.89427638e-01 -4.89530981e-01 6.76215142e-02] | [12.877628326416016, 3.4558260440826416] |
e18b3c32-53e1-41c3-99cd-41c4280c4deb | mel-spectrogram-features-for-acoustic-vehicle | 2204.04013 | null | https://arxiv.org/abs/2204.04013v1 | https://arxiv.org/pdf/2204.04013v1.pdf | Mel-spectrogram features for acoustic vehicle detection and speed estimation | The paper addresses acoustic vehicle detection and speed estimation from single sensor measurements. We predict the vehicle's pass-by instant by minimizing clipped vehicle-to-microphone distance, which is predicted from the mel-spectrogram of input audio, in a supervised learning approach. In addition, mel-spectrogram-based features are used directly for vehicle speed estimation, without introducing any intermediate features. The results show that the proposed features can be used for accurate vehicle detection and speed estimation, with an average error of 7.87 km/h. If we formulate speed estimation as a classification problem, with a 10 km/h discretization interval, the proposed method attains the average accuracy of 48.7% for correct class prediction and 91.0% when an offset of one class is allowed. The proposed method is evaluated on a dataset of 304 urban-environment on-field recordings of ten different vehicles. | ['Slobodan Djukanovic', 'Nikola Bulatovic'] | 2022-04-08 | null | null | null | null | ['vehicle-speed-estimation'] | ['computer-vision'] | [ 1.21327199e-01 -2.35423848e-01 -8.33936036e-02 -5.07175207e-01
-1.11886322e+00 -3.36795002e-01 3.97277921e-01 4.10509050e-01
-7.13919401e-01 5.15471339e-01 -5.03170967e-01 -3.96816969e-01
-1.19363494e-01 -7.20313668e-01 -6.94346726e-01 -8.58993053e-01
-3.37263793e-01 -8.05825926e-03 4.06850159e-01 2.88882822e-01
1.99243039e-01 5.91046154e-01 -1.86270046e+00 -1.54069170e-01
6.35362327e-01 1.18444490e+00 3.09761792e-01 1.21788049e+00
1.87104955e-01 6.02748573e-01 -8.68481338e-01 -1.56804919e-01
7.10462928e-02 1.10368654e-01 -1.90035433e-01 -2.17211694e-01
2.29909703e-01 -4.76486385e-01 -2.20194876e-01 8.26878250e-01
2.45863885e-01 1.64829373e-01 8.36397469e-01 -1.37329733e+00
5.49926996e-01 1.99658334e-01 -1.52696699e-01 3.27280670e-01
2.25645587e-01 -2.26918049e-02 6.10239446e-01 -8.65992308e-01
-1.91786718e-02 7.87868559e-01 7.66686261e-01 1.42886713e-01
-1.02052295e+00 -9.42322135e-01 -3.19658041e-01 7.16973245e-01
-1.93611670e+00 -6.04360580e-01 9.25281227e-01 -7.93491066e-01
9.56548154e-01 2.57730752e-01 5.51115572e-01 3.11200321e-01
1.44820303e-01 3.89785171e-01 5.80450177e-01 -6.43153638e-02
3.79225791e-01 4.12153125e-01 6.18601479e-02 2.10655421e-01
1.87372819e-01 1.72527537e-01 -1.43064111e-01 -1.72194988e-01
2.27280264e-03 -2.27315083e-01 1.94448177e-02 1.02317110e-01
-8.78511548e-01 7.56391585e-01 -2.27759734e-01 5.00931479e-02
-3.28390241e-01 2.70696759e-01 4.60952550e-01 1.57361165e-01
4.74661827e-01 -1.59187376e-01 -4.39137340e-01 -4.90827382e-01
-1.22549367e+00 3.71507972e-01 5.36859393e-01 8.82845521e-01
7.90604115e-01 5.22920012e-01 3.95531952e-01 7.92670310e-01
3.64855081e-01 1.40945566e+00 1.82136759e-01 -9.43021476e-01
5.72555125e-01 -1.00075252e-01 3.00777406e-01 -1.15331483e+00
-5.76962233e-01 -1.33831650e-01 -4.49715763e-01 2.11356699e-01
4.68118995e-01 -4.37980890e-01 -3.38174701e-01 1.35197043e+00
4.24216717e-01 6.65360093e-01 1.81099445e-01 6.56514168e-01
5.18122911e-01 1.16145945e+00 7.69962817e-02 -7.75398314e-01
9.24243808e-01 -5.26182175e-01 -8.83001864e-01 -1.30305835e-03
7.05241621e-01 -7.02054739e-01 5.38088560e-01 6.36237919e-01
-5.93647540e-01 -8.05980623e-01 -1.12522614e+00 7.25557923e-01
-4.25358355e-01 2.52850950e-01 -1.84444174e-01 8.65544081e-01
-5.17991543e-01 3.59673172e-01 -7.85980403e-01 1.05899863e-01
-1.89352438e-01 2.85041004e-01 -4.53954637e-02 3.41276556e-01
-1.19309843e+00 5.69728792e-01 1.04740247e-01 1.85376436e-01
-7.95555174e-01 -8.11213732e-01 -7.63497472e-01 -2.90171951e-01
-3.67888622e-02 5.09500913e-02 1.40437782e+00 -5.95351875e-01
-1.54960787e+00 2.08845183e-01 -4.06014234e-01 -7.35327840e-01
5.12032986e-01 -1.16829403e-01 -1.05748761e+00 4.08961207e-01
-3.67353559e-02 2.21771300e-01 1.11521685e+00 -9.73238349e-01
-1.21562874e+00 -6.15192354e-02 -3.68593395e-01 -9.36960876e-02
-2.01284826e-01 -5.95754460e-02 -9.25450623e-02 -3.39617133e-01
-1.46735445e-01 -9.13598895e-01 -2.65024081e-02 -3.50731879e-01
-1.22239962e-01 -2.10653350e-01 1.13566029e+00 -8.87029767e-01
1.45330644e+00 -2.15006661e+00 -7.13658869e-01 3.83054286e-01
-1.63936108e-01 4.73652691e-01 2.27798000e-02 3.05053890e-01
3.15823734e-01 -3.08936328e-01 -1.47004455e-01 -3.61447334e-01
-2.70873129e-01 -1.26470298e-01 -7.17139170e-02 8.93696904e-01
-1.36681562e-02 1.72250390e-01 -8.43003154e-01 -4.41786617e-01
7.49778509e-01 7.40758359e-01 -4.57346320e-01 2.34226689e-01
3.97803426e-01 1.51268065e-01 -3.51070732e-01 3.66060287e-01
1.00543594e+00 6.35489523e-01 -3.10419679e-01 -1.02171481e-01
-5.13387918e-01 1.19559392e-01 -1.42916632e+00 7.95039415e-01
-8.87140274e-01 1.32948601e+00 7.32890293e-02 -1.13730323e+00
1.23529482e+00 3.51075709e-01 7.07472920e-01 -5.32742500e-01
1.57879889e-01 3.55163276e-01 -2.31722087e-01 -7.74686038e-01
6.03477299e-01 3.00481260e-01 -5.65385036e-02 -2.55216211e-01
-4.19009715e-01 -1.74817592e-01 1.73498318e-01 -1.43735737e-01
8.20811331e-01 -6.80951953e-01 9.77212861e-02 -1.86282188e-01
8.27555895e-01 -1.88630372e-01 4.18415397e-01 4.54795748e-01
-2.17819467e-01 1.32685766e-01 8.73455554e-02 -9.27955955e-02
-9.85966265e-01 -8.79770815e-01 -4.06602740e-01 6.26186609e-01
2.03558832e-01 -2.02772498e-01 -8.20179224e-01 -2.35774621e-01
1.53262779e-01 9.04321671e-01 -2.16589957e-01 -2.90154200e-02
-6.24939203e-01 -2.97758162e-01 7.01050282e-01 4.51361358e-01
6.82434440e-02 -2.94223815e-01 -7.25963891e-01 4.84409332e-01
-1.42437533e-01 -1.54599464e+00 -1.94021419e-01 2.79418565e-02
-6.06410027e-01 -8.52435827e-01 -3.87972742e-01 -6.02309108e-01
3.30686927e-01 3.48398894e-01 3.81798625e-01 -2.33159155e-01
-5.95935397e-02 4.91875380e-01 -2.00799182e-01 -5.22653103e-01
-5.78321576e-01 -2.02359378e-01 3.94493699e-01 4.27699745e-01
4.96141911e-01 -3.54023993e-01 -5.67737341e-01 6.46068215e-01
-2.75931120e-01 -3.50603908e-01 4.45777141e-02 4.95318145e-01
5.90086818e-01 2.83466846e-01 9.31117296e-01 -1.28536105e-01
1.02946624e-01 -5.37063837e-01 -1.18137348e+00 -3.84775102e-01
-7.09305167e-01 -4.83064175e-01 1.08352661e+00 -6.09170794e-01
-7.34582603e-01 4.33038831e-01 -4.45545793e-01 -3.27170432e-01
-4.03801203e-01 1.29053414e-01 -1.62313342e-01 -1.26705632e-01
3.49689156e-01 3.57505262e-01 -3.20696533e-02 -3.83524835e-01
-1.33693824e-02 1.17286968e+00 6.63924813e-01 1.64828256e-01
8.43710601e-01 3.05985391e-01 -1.34722823e-02 -1.72024119e+00
1.01021871e-01 -9.13786650e-01 -2.53773183e-01 -7.31727958e-01
8.52835000e-01 -1.12799251e+00 -9.97540832e-01 6.35823607e-01
-1.18907177e+00 -1.63091406e-01 -3.81896049e-02 1.13876474e+00
-5.54908574e-01 2.60962754e-01 -2.38890976e-01 -1.35940540e+00
-2.98818499e-02 -1.04798281e+00 9.93018270e-01 8.88045058e-02
-1.49784699e-01 -9.24293458e-01 -5.89835308e-02 2.84666598e-01
2.25913376e-01 4.06339653e-02 2.08997712e-01 -6.52583599e-01
-2.68649995e-01 -6.94057703e-01 1.40279189e-01 6.15675092e-01
-9.31456611e-02 2.32167885e-01 -1.18978071e+00 -1.89846650e-01
-2.35725090e-01 2.37920746e-01 5.08932233e-01 6.76521540e-01
8.50988626e-01 -4.09794569e-01 -3.74630392e-01 4.35463637e-01
1.47479236e+00 6.62252307e-01 2.91069120e-01 3.33655626e-02
4.95922387e-01 3.67963910e-01 1.04827726e+00 7.23469317e-01
3.78166109e-01 9.33495045e-01 5.49495339e-01 3.28016728e-01
1.09599397e-01 -2.53380716e-01 4.28466618e-01 1.00049829e+00
7.73366019e-02 -1.90673709e-01 -9.42863703e-01 7.17070520e-01
-1.32777452e+00 -1.16168988e+00 -6.31119668e-01 2.46552753e+00
1.53460518e-01 3.23111176e-01 3.20645958e-01 9.38149929e-01
7.99921751e-01 6.01373799e-03 -1.13679044e-01 -6.05380476e-01
3.74505430e-01 -3.73082846e-01 1.11904335e+00 1.14330935e+00
-1.26237702e+00 3.07190239e-01 6.02968454e+00 1.02077127e+00
-1.36410761e+00 1.85009893e-02 2.61332244e-01 1.13450333e-01
3.45777348e-02 -3.76779675e-01 -1.07017910e+00 7.27940977e-01
1.56658971e+00 -2.31738821e-01 2.50851780e-01 1.01350963e+00
8.73207092e-01 -3.04669172e-01 -9.30109620e-01 1.24185085e+00
-6.40179366e-02 -1.01436484e+00 -4.59704518e-01 -1.02458680e-02
3.30240846e-01 -1.47634163e-01 8.79143104e-02 1.28347248e-01
-7.13022351e-01 -7.30247676e-01 9.48500395e-01 5.04249632e-01
6.90524995e-01 -1.17638576e+00 8.36997747e-01 7.00494826e-01
-1.74659657e+00 -2.67061204e-01 -1.62678599e-01 -2.49789625e-01
5.14099598e-01 6.37810946e-01 -1.27622306e+00 1.62052184e-01
5.29243827e-01 4.70209599e-01 -1.41741142e-01 1.39108896e+00
3.87041688e-01 1.09278166e+00 -6.78480148e-01 -4.35033768e-01
8.36014822e-02 -5.37244007e-02 7.31227160e-01 1.82224786e+00
5.72932720e-01 -1.17682647e-02 -4.95377742e-02 1.32527962e-01
4.88072962e-01 2.72273749e-01 -6.31658077e-01 3.26677054e-01
9.36358690e-01 9.08515573e-01 -4.34524924e-01 -2.70057291e-01
-3.61056656e-01 3.21874887e-01 -4.64768291e-01 5.58551013e-01
-1.32349586e+00 -1.06509280e+00 7.58050203e-01 3.17111939e-01
5.37401795e-01 -2.91512519e-01 9.34094116e-02 -1.45399436e-01
3.74599397e-02 -2.73022860e-01 -1.94252387e-01 -2.92887002e-01
-5.38190722e-01 3.77629876e-01 1.85632601e-01 -1.78498685e+00
-5.68272710e-01 -5.15587449e-01 -6.17853701e-01 5.21895170e-01
-1.37895548e+00 -5.98570883e-01 -1.14823624e-01 4.09045547e-01
7.88877487e-01 -1.70282349e-01 4.83903110e-01 9.01475489e-01
-4.13299143e-01 9.33816433e-01 6.11888528e-01 6.93389848e-02
1.89303339e-01 -8.23248744e-01 1.16926134e-01 6.60847485e-01
-2.76206315e-01 -8.91786963e-02 1.23321366e+00 -2.25246429e-01
-1.46110177e+00 -1.42081356e+00 1.21656024e+00 -1.77756116e-01
7.31003404e-01 -2.49616355e-01 -6.54825509e-01 1.04747444e-01
-4.02988970e-01 3.16298217e-01 4.71643806e-01 -5.71151912e-01
2.05662530e-02 -8.24413955e-01 -1.11732185e+00 -1.03079766e-01
3.59624505e-01 -5.52842677e-01 -3.33522975e-01 1.67702749e-01
4.41088080e-01 -1.02058567e-01 -9.70549285e-01 3.04099023e-01
7.38909364e-01 -6.02342904e-01 9.23699439e-01 9.74877626e-02
-2.71768749e-01 -5.00496149e-01 -4.83123571e-01 -1.07918453e+00
5.03357798e-02 -5.05836010e-01 -1.21957153e-01 1.22642362e+00
6.31159008e-01 -6.42139912e-01 6.39157891e-01 6.63658753e-02
-7.59044811e-02 -5.11698186e-01 -1.41014457e+00 -1.05064809e+00
-2.86512345e-01 -1.35451174e+00 4.77075517e-01 3.32833469e-01
3.37679945e-02 8.43532607e-02 -4.97770131e-01 7.15017140e-01
8.53180587e-01 -4.63328600e-01 9.39106703e-01 -1.11701930e+00
-5.81623949e-02 -2.22738847e-01 -9.27260995e-01 -1.24903631e+00
2.14020953e-01 -3.12832803e-01 4.52297181e-01 -1.06269240e+00
-4.92465883e-01 -3.33686084e-01 -7.59062618e-02 -3.41921061e-01
2.30717778e-01 2.65553117e-01 4.33494076e-02 -3.03450286e-01
-4.33904827e-01 3.62096250e-01 3.91457528e-01 -4.04687554e-01
-2.79554963e-01 8.28350842e-01 3.53660613e-01 1.01825452e+00
8.10920537e-01 -6.27154350e-01 -5.88689804e-01 -1.14356652e-01
-1.48913428e-01 4.18401122e-01 3.90555650e-01 -1.34309852e+00
3.30883533e-01 -2.07018584e-01 1.01546897e-02 -9.29073989e-01
6.60930514e-01 -1.17674100e+00 3.92014861e-01 6.50868237e-01
-1.53390840e-01 -9.39160958e-02 2.48228937e-01 7.66138613e-01
-4.19321746e-01 -9.11523625e-02 8.49982321e-01 7.86422133e-01
-9.04550970e-01 -7.42439367e-03 -1.05539322e+00 -2.53155231e-01
1.24754786e+00 -6.10426366e-01 2.34328300e-01 -5.04608095e-01
-4.42207634e-01 1.69595238e-02 -6.75218999e-02 3.46260190e-01
6.04320168e-01 -1.39414680e+00 -7.54242301e-01 1.30889907e-01
2.17172012e-01 -5.46159029e-01 4.29667681e-01 8.02694976e-01
-5.16545236e-01 7.10620761e-01 3.74021500e-01 -1.12380552e+00
-1.49363458e+00 3.22918683e-01 3.50499719e-01 3.87373507e-01
-2.48043641e-01 5.23872554e-01 -3.03503215e-01 2.08379347e-02
4.71355051e-01 -4.35003459e-01 -3.78740162e-01 8.11578110e-02
8.37112904e-01 9.67531621e-01 2.16715589e-01 -1.15170252e+00
-6.02717876e-01 1.09799337e+00 5.79923630e-01 -2.04605937e-01
7.16875672e-01 -3.50148261e-01 6.45695508e-01 7.24899948e-01
1.72539043e+00 3.35111290e-01 -1.33702898e+00 1.77772149e-01
-2.46671051e-01 -4.52269763e-01 2.17291608e-01 -2.15533495e-01
-8.78243744e-01 9.35711801e-01 1.11676979e+00 4.17256832e-01
1.05345643e+00 -1.69566765e-01 7.50987947e-01 3.58150870e-01
3.39190751e-01 -1.34547293e+00 -5.48399746e-01 4.27839339e-01
3.45439196e-01 -1.14300501e+00 -3.20861489e-01 -2.97531575e-01
-3.40764761e-01 1.02501547e+00 3.92473668e-01 2.76114959e-02
9.92174447e-01 4.20085937e-01 7.22225606e-02 6.94535077e-01
-6.09116673e-01 9.42030475e-02 1.76992454e-02 8.74770880e-01
5.12749292e-02 4.28164810e-01 -3.13084453e-01 5.46634018e-01
-3.11213493e-01 -2.73349851e-01 5.13958037e-01 5.89714944e-01
-9.04514372e-01 -3.03389996e-01 -6.93024099e-01 3.93087864e-01
-4.19286042e-01 2.10963905e-01 4.24348086e-01 6.36723340e-01
2.65603900e-01 1.56322861e+00 4.34268445e-01 -7.92123497e-01
6.11557245e-01 -1.32055655e-01 -2.37106621e-01 1.97569087e-01
1.07174546e-01 3.34290527e-02 3.32621247e-01 -4.72223550e-01
-1.96886837e-01 -8.20740342e-01 -1.46592748e+00 -4.51773703e-01
-3.98684233e-01 6.80967450e-01 1.13133574e+00 7.98336506e-01
-8.74095708e-02 3.67166042e-01 1.34733284e+00 -8.96080792e-01
-5.45673132e-01 -7.19508827e-01 -6.59412801e-01 1.54068461e-02
1.03592324e+00 -5.09466231e-01 -7.64675438e-01 2.68518925e-01] | [7.920884132385254, -1.02773916721344] |
337a37bd-2561-40e4-904e-bd9d086a3b52 | probabilistic-label-trees-for-extreme-multi | 2009.11218 | null | https://arxiv.org/abs/2009.11218v1 | https://arxiv.org/pdf/2009.11218v1.pdf | Probabilistic Label Trees for Extreme Multi-label Classification | Extreme multi-label classification (XMLC) is a learning task of tagging instances with a small subset of relevant labels chosen from an extremely large pool of possible labels. Problems of this scale can be efficiently handled by organizing labels as a tree, like in hierarchical softmax used for multi-class problems. In this paper, we thoroughly investigate probabilistic label trees (PLTs) which can be treated as a generalization of hierarchical softmax for multi-label problems. We first introduce the PLT model and discuss training and inference procedures and their computational costs. Next, we prove the consistency of PLTs for a wide spectrum of performance metrics. To this end, we upperbound their regret by a function of surrogate-loss regrets of node classifiers. Furthermore, we consider a problem of training PLTs in a fully online setting, without any prior knowledge of training instances, their features, or labels. In this case, both node classifiers and the tree structure are trained online. We prove a specific equivalence between the fully online algorithm and an algorithm with a tree structure given in advance. Finally, we discuss several implementations of PLTs and introduce a new one, napkinXC, which we empirically evaluate and compare with state-of-the-art algorithms. | ['Robert Busa-Fekete', 'Kalina Jasinska-Kobus', 'Marek Wydmuch', 'Mikhail Kuznetsov', 'Krzysztof Dembczynski'] | 2020-09-23 | null | null | null | null | ['extreme-multi-label-classification'] | ['methodology'] | [ 4.12856996e-01 4.70801413e-01 -3.76460999e-01 -8.44823360e-01
-1.26175129e+00 -8.21582913e-01 3.03538293e-01 3.20314527e-01
-5.85578799e-01 8.28970134e-01 -3.19710135e-01 -2.58188814e-01
-2.93905973e-01 -5.71592927e-01 -6.95917904e-01 -9.42680180e-01
-6.39136881e-02 8.34124088e-01 -3.19826528e-02 3.47875386e-01
-1.48526415e-01 4.27324772e-01 -1.53283048e+00 3.22102636e-01
4.55632865e-01 1.35550153e+00 -1.41604975e-01 6.53826594e-01
-2.76906341e-01 7.46897042e-01 -4.82348680e-01 -7.31152773e-01
2.70177305e-01 -6.00713901e-02 -1.33552694e+00 1.85022876e-01
5.74955642e-01 9.25131738e-02 2.15707630e-01 1.06508529e+00
4.52957690e-01 2.32581332e-01 7.03796864e-01 -1.56759667e+00
-5.72970808e-02 1.09655035e+00 -4.11660314e-01 -4.56445098e-01
8.58155191e-02 -5.00066102e-01 1.42923784e+00 -5.56540906e-01
6.09423399e-01 1.26220632e+00 8.47006261e-01 6.61076903e-01
-1.40907133e+00 -4.37774837e-01 4.80699629e-01 -8.93290780e-05
-1.17255270e+00 -2.52249628e-01 4.23314095e-01 -4.41770375e-01
5.82866669e-01 4.43334579e-01 -6.52757138e-02 9.84933078e-01
-3.77784222e-01 9.54533041e-01 1.28370214e+00 -7.51634717e-01
5.29629707e-01 3.79327267e-01 5.12436330e-01 7.60237157e-01
-5.91461584e-02 -3.80191952e-02 -3.49232286e-01 -5.05633771e-01
-2.04645712e-02 -1.04747728e-01 1.19405247e-01 -5.49104154e-01
-8.49673688e-01 1.01597011e+00 2.25191787e-01 1.40085801e-01
-1.07059479e-01 4.72177505e-01 4.75875676e-01 2.81074971e-01
7.65093327e-01 4.56346981e-02 -7.86034524e-01 2.87018746e-01
-8.43490481e-01 -1.13645121e-01 1.15068924e+00 1.20785487e+00
7.09758997e-01 -4.93794143e-01 -3.20745111e-01 1.01054573e+00
2.75883049e-01 3.76856536e-01 1.66480079e-01 -1.01330841e+00
3.14830303e-01 8.89427885e-02 1.29929468e-01 -2.79848948e-02
-6.93979442e-01 -5.40246904e-01 -5.62648714e-01 2.20369145e-01
6.42695308e-01 -3.20703864e-01 -6.70327961e-01 2.06787324e+00
7.04168439e-01 1.40498400e-01 -5.84025085e-02 5.19082725e-01
3.90213042e-01 5.72563648e-01 3.96612108e-01 -4.22036648e-01
1.50906622e+00 -1.16170275e+00 -5.63297808e-01 -2.72628248e-01
1.08030641e+00 -5.09669244e-01 6.03481829e-01 4.10598725e-01
-9.59207058e-01 -2.52618819e-01 -6.36079311e-01 -2.39068314e-01
-6.10243499e-01 1.56380489e-01 8.45704734e-01 9.06584740e-01
-1.01258492e+00 9.51705992e-01 -6.57207906e-01 -1.71964392e-01
3.47407043e-01 4.91408676e-01 -2.62796998e-01 -1.46185700e-02
-1.13406730e+00 8.71728659e-01 6.47950709e-01 4.50458564e-03
-7.32564449e-01 -3.70487332e-01 -8.09760690e-01 2.29652151e-01
7.17189729e-01 -3.67849648e-01 1.85096717e+00 -8.60141814e-01
-1.47845793e+00 1.23157513e+00 -1.96653783e-01 -3.91254455e-01
6.49370611e-01 -6.06086217e-02 3.39540020e-02 -1.05872422e-01
1.63652435e-01 6.46710753e-01 6.02086723e-01 -1.22682095e+00
-1.13152695e+00 -1.69406697e-01 3.71556073e-01 7.35579655e-02
-2.00742602e-01 1.76190615e-01 -1.88730940e-01 -3.92187357e-01
5.09544648e-02 -1.14115095e+00 -3.89637828e-01 4.21384349e-02
-6.00857854e-01 -6.56917036e-01 2.69686788e-01 -2.52306134e-01
9.84082937e-01 -2.09801221e+00 6.70208335e-02 2.24343136e-01
5.20885624e-02 -1.32070750e-01 -4.53605056e-02 3.29203099e-01
-1.40044302e-01 2.97287643e-01 -3.74412745e-01 -1.10696077e+00
5.90880036e-01 3.53153765e-01 -1.17751449e-01 6.13256633e-01
-1.94130301e-01 7.97692776e-01 -9.38891292e-01 -8.45169127e-01
-5.79864159e-02 9.82035175e-02 -1.70912296e-01 4.84044440e-02
-5.75494051e-01 1.30937159e-01 -4.85529304e-01 7.19178021e-01
7.29858398e-01 -5.36670566e-01 4.26416755e-01 2.28184715e-01
3.48938331e-02 8.67818221e-02 -1.29154384e+00 1.54427338e+00
-8.20581734e-01 1.13115825e-01 1.60018355e-01 -1.14920759e+00
4.46586162e-01 4.72143888e-01 4.98867631e-01 -1.87172424e-02
3.01109582e-01 3.00234139e-01 -7.01043129e-01 -1.94066137e-01
1.49887487e-01 -5.24463892e-01 -4.70582247e-01 7.56628990e-01
2.23917902e-01 3.60766083e-01 2.56933689e-01 -5.44896349e-02
9.30828333e-01 8.40344802e-02 4.17544186e-01 -1.10114112e-01
2.85611302e-01 -3.87972087e-01 6.33809984e-01 1.17622817e+00
-1.01255476e-01 2.13493213e-01 7.18797803e-01 -2.89042562e-01
-6.20628417e-01 -7.84644961e-01 -5.02216160e-01 1.81741476e+00
-1.59912989e-01 -2.21715674e-01 -6.85056269e-01 -1.20576096e+00
1.29694179e-01 7.36899912e-01 -8.45277190e-01 2.13299930e-01
-2.51882434e-01 -8.20589542e-01 3.58021140e-01 3.82828176e-01
1.07653968e-01 -9.90911543e-01 -5.22322893e-01 2.38795280e-01
-1.07082091e-01 -1.30626023e+00 -3.19469333e-01 9.58475292e-01
-7.46193051e-01 -9.39423561e-01 -4.81559604e-01 -9.86617506e-01
5.15208602e-01 -2.04133019e-01 1.24047983e+00 -1.91299379e-01
-4.22829762e-03 1.61262155e-01 -4.50039148e-01 -2.65210360e-01
-4.92559165e-01 5.46587527e-01 -1.32540762e-01 1.69666335e-01
2.33315621e-02 -4.44338441e-01 -4.64019924e-02 2.24489316e-01
-7.67178893e-01 1.87679067e-01 5.80978751e-01 7.34246075e-01
7.91210651e-01 1.05858371e-01 4.45456475e-01 -1.68456078e+00
9.91682485e-02 -6.66073859e-01 -9.09236848e-01 8.28106225e-01
-6.63555026e-01 4.96992946e-01 6.83493495e-01 -4.44954067e-01
-1.00825250e+00 6.48122907e-01 -2.47510031e-01 -1.16367541e-01
-2.79869944e-01 3.99050087e-01 -2.78865278e-01 -1.24665007e-01
2.60313183e-01 -2.76333511e-01 -5.11447608e-01 -7.95280218e-01
5.63638926e-01 8.00947309e-01 2.80076951e-01 -9.12756443e-01
5.28367996e-01 3.62596959e-01 4.26193357e-01 -1.39366359e-01
-1.85876644e+00 -6.27450466e-01 -7.87212729e-01 -1.15212791e-01
6.10827565e-01 -5.94551146e-01 -1.04240596e+00 3.89399052e-01
-1.05743217e+00 -6.34437263e-01 -4.34208959e-01 1.94326952e-01
-7.72222638e-01 1.96786389e-01 -8.22875679e-01 -1.15024877e+00
-1.73087865e-01 -9.24492776e-01 1.31335104e+00 1.06833741e-01
2.52002090e-01 -1.31308985e+00 -1.07957408e-01 2.64349580e-01
8.48636329e-02 3.67498040e-01 7.66931772e-01 -1.04430878e+00
-3.07709157e-01 -2.62066871e-01 -1.13597170e-01 2.49350980e-01
-2.58242458e-01 -4.28105384e-01 -1.34325540e+00 -4.45234418e-01
-1.44834379e-02 -6.40094936e-01 9.23463225e-01 3.17903191e-01
1.35457480e+00 -3.46602350e-01 -6.37765944e-01 5.56104720e-01
1.59764063e+00 -1.96159989e-01 -2.11258009e-01 2.90664971e-01
4.95691538e-01 7.99443543e-01 6.80326521e-01 4.79761809e-01
3.10566574e-01 7.22922981e-01 4.68341470e-01 8.76819119e-02
3.10696483e-01 -1.28364921e-01 6.69437945e-02 4.44189012e-01
4.05351341e-01 -4.53942060e-01 -7.04056740e-01 2.52968907e-01
-1.95922899e+00 -4.77376878e-01 -3.00668124e-02 2.25303221e+00
1.11662769e+00 1.86850920e-01 1.62959039e-01 1.03037372e-01
1.20128191e+00 3.29147950e-02 -6.40448570e-01 -3.96605611e-01
-5.77591099e-02 3.58333766e-01 8.84519398e-01 7.25373507e-01
-1.68765461e+00 9.23772573e-01 6.22535944e+00 1.06928551e+00
-6.87924981e-01 5.25959492e-01 7.88694859e-01 -2.17370138e-01
9.04996321e-02 2.48178273e-01 -1.14242947e+00 4.01637226e-01
1.29275835e+00 1.28803223e-01 4.55915362e-01 1.13212812e+00
-3.84380519e-01 -2.85847541e-02 -1.61044276e+00 7.27165341e-01
-3.18373382e-01 -8.89865696e-01 -6.32748842e-01 1.11874565e-01
7.45427966e-01 -3.00335251e-02 -3.16513097e-03 5.59228957e-01
8.57880712e-01 -7.48926997e-01 9.18132901e-01 3.06933403e-01
1.02631783e+00 -5.32651305e-01 7.45082021e-01 4.87670094e-01
-1.07346869e+00 -3.30765128e-01 -3.76198530e-01 1.13028847e-01
2.96078950e-01 8.35159540e-01 -3.57938796e-01 6.07431948e-01
4.28970248e-01 3.55356961e-01 -4.56521213e-01 1.13481426e+00
-3.92047048e-01 7.62581885e-01 -7.32107103e-01 -9.80155766e-02
3.48041087e-01 6.96386471e-02 1.20749464e-02 1.37987041e+00
1.16812065e-01 -1.11066597e-02 5.50292850e-01 3.24470431e-01
-4.71032262e-01 1.85581833e-01 -7.60554746e-02 4.03857350e-01
6.96186483e-01 1.53469396e+00 -8.69859815e-01 -3.77216369e-01
-4.78141785e-01 8.45227718e-01 7.63005495e-01 1.16403513e-01
-9.23206329e-01 -1.93958342e-01 2.20847577e-01 -4.62104440e-01
2.71513015e-01 5.30447423e-01 -1.36434972e-01 -9.08498585e-01
2.51305830e-02 -3.14136267e-01 8.48912537e-01 -4.48479325e-01
-1.48350346e+00 3.94480824e-01 1.49792522e-01 -9.11591411e-01
-1.91241592e-01 -6.76641881e-01 -2.01707825e-01 6.65893734e-01
-1.71598554e+00 -1.18074489e+00 4.90733013e-02 1.60092607e-01
3.16254199e-01 4.88498449e-01 9.57390845e-01 2.32371598e-01
-6.00984633e-01 9.54832256e-01 6.82809770e-01 -3.94954123e-02
6.06366396e-01 -1.76983881e+00 2.61704952e-01 4.37196672e-01
2.05165416e-01 4.35346598e-03 5.34933150e-01 -2.18768671e-01
-6.86447561e-01 -1.30449712e+00 1.32912576e+00 -2.77153969e-01
6.22680843e-01 -6.49402678e-01 -6.38929248e-01 1.03799689e+00
-1.92721814e-01 4.43860412e-01 8.52275252e-01 5.62102854e-01
-3.77434969e-01 1.89941912e-03 -1.30711961e+00 1.16101384e-01
1.16165090e+00 -4.29199934e-01 -1.07513078e-01 9.26256657e-01
6.73463106e-01 -4.45091814e-01 -1.09486401e+00 2.77421385e-01
6.06029689e-01 -6.12606466e-01 7.52582908e-01 -6.55272543e-01
1.81300398e-02 1.32737324e-01 -1.52313501e-01 -1.20390821e+00
-3.21129382e-01 -5.84719062e-01 -1.26091659e-01 1.44550049e+00
6.72786057e-01 -8.88334811e-01 8.77881527e-01 6.99261427e-01
2.07986422e-02 -9.05883908e-01 -1.10139954e+00 -8.40830266e-01
3.54828179e-01 -4.72979546e-01 5.04423559e-01 8.62027347e-01
7.78902844e-02 3.51821601e-01 -5.02618968e-01 2.29033992e-01
9.65273678e-01 4.47449595e-01 8.28827620e-02 -1.58209145e+00
-5.45383334e-01 -3.41339290e-01 5.81594072e-02 -9.59268391e-01
7.73772836e-01 -1.31985557e+00 2.89032400e-01 -1.28535998e+00
4.68479902e-01 -1.02619684e+00 -5.34387708e-01 9.06934261e-01
-1.63842365e-01 -4.78907041e-02 2.82033801e-01 8.31801817e-02
-1.16740191e+00 2.49816909e-01 6.52783513e-01 -8.00290052e-03
2.44790941e-01 5.16852438e-01 -5.69699347e-01 6.51315153e-01
8.43419909e-01 -1.14883423e+00 -1.15369275e-01 -2.14678288e-01
3.41975927e-01 3.34084570e-01 3.80236395e-02 -7.61544526e-01
3.33655983e-01 -1.88897789e-01 -2.02061087e-02 -4.13922429e-01
2.56978303e-01 -8.31225157e-01 8.16757232e-02 2.25596741e-01
-1.04424024e+00 -5.11796176e-01 -2.42399171e-01 5.54703057e-01
1.34312123e-01 -1.00466621e+00 8.68067980e-01 -9.34285461e-04
-2.31564730e-01 3.49304795e-01 -2.86999822e-01 2.07560852e-01
1.10783136e+00 1.74685046e-01 -6.34944811e-02 -1.22622617e-01
-1.20839620e+00 6.09069228e-01 2.42676377e-01 -1.07036561e-01
-7.14554787e-02 -1.34575570e+00 -6.24715149e-01 -3.47381473e-01
2.08831459e-01 1.17964104e-01 1.81262806e-01 6.17902100e-01
2.17761695e-02 4.56459433e-01 3.69210720e-01 -2.51789033e-01
-1.30562460e+00 8.36102784e-01 3.02533090e-01 -7.19239354e-01
-2.53285140e-01 9.16596055e-01 1.85029835e-01 -7.08779514e-01
7.10305393e-01 -1.81106210e-01 -2.92122755e-02 1.70064256e-01
1.80922240e-01 3.03794712e-01 -6.64805397e-02 -4.19153959e-01
-3.03826064e-01 4.20770526e-01 1.92099698e-02 -7.02967718e-02
1.18625271e+00 -1.49993032e-01 -1.12903990e-01 7.18438864e-01
1.40497243e+00 -2.52762794e-01 -1.30772114e+00 -7.46407688e-01
5.56151807e-01 3.00920550e-02 -6.23590313e-02 -9.39357817e-01
-1.03843832e+00 7.03880906e-01 5.73422611e-01 4.97535616e-01
1.06233644e+00 4.43036854e-01 6.30755782e-01 5.51956058e-01
5.16062856e-01 -1.18893325e+00 -5.02459586e-01 3.44706148e-01
2.36501262e-01 -1.33481729e+00 -1.78672597e-01 -5.92466176e-01
-4.36073989e-01 1.04186499e+00 1.07692853e-01 2.93588042e-01
8.61722529e-01 3.62663537e-01 -1.39563680e-01 1.00037731e-01
-1.01068962e+00 -5.34042537e-01 -4.75469902e-02 2.86388546e-01
2.91307569e-01 3.64495158e-01 -1.67529717e-01 6.57126129e-01
-3.13636549e-02 -1.63303167e-01 1.24057487e-01 9.47207212e-01
-4.32206124e-01 -1.41360354e+00 -2.53240287e-01 4.23062503e-01
-8.52241755e-01 -6.18885309e-02 -2.38227323e-01 4.75094914e-01
3.24746162e-01 1.05130088e+00 -3.09084833e-01 -4.54059280e-02
1.17061131e-01 5.49299717e-01 3.84271234e-01 -7.00355470e-01
-5.36728203e-01 -1.21413775e-01 3.77239764e-01 -2.22081125e-01
-4.83164430e-01 -9.34322536e-01 -1.06017780e+00 -1.55168265e-01
-7.83292711e-01 4.46171075e-01 8.51656437e-01 1.04328227e+00
-9.73246172e-02 2.42960930e-01 8.41516733e-01 -6.50377631e-01
-1.08067131e+00 -1.04406488e+00 -9.26789463e-01 2.89597899e-01
1.94714576e-01 -6.80245101e-01 -5.00390172e-01 -9.83466208e-02] | [9.387393951416016, 4.306585788726807] |
238d10c7-b921-413e-9272-55d503c95402 | iterative-next-boundary-detection-for | 2212.03022 | null | https://arxiv.org/abs/2212.03022v2 | https://arxiv.org/pdf/2212.03022v2.pdf | Iterative Next Boundary Detection for Instance Segmentation of Tree Rings in Microscopy Images of Shrub Cross Sections | We address the problem of detecting tree rings in microscopy images of shrub cross sections. This can be regarded as a special case of the instance segmentation task with several unique challenges such as the concentric circular ring shape of the objects and high precision requirements that result in inadequate performance of existing methods. We propose a new iterative method which we term Iterative Next Boundary Detection (INBD). It intuitively models the natural growth direction, starting from the center of the shrub cross section and detecting the next ring boundary in each iteration step. In our experiments, INBD shows superior performance to generic instance segmentation methods and is the only one with a built-in notion of chronological order. Our dataset and source code are available at http://github.com/alexander-g/INBD. | ['Uwe Freiherr von Lukas', 'Martin Wilmking', 'Alba Anadon-Rosell', 'Giulia Resente', 'Alexander Gillert'] | 2022-12-06 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Gillert_Iterative_Next_Boundary_Detection_for_Instance_Segmentation_of_Tree_Rings_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Gillert_Iterative_Next_Boundary_Detection_for_Instance_Segmentation_of_Tree_Rings_CVPR_2023_paper.pdf | cvpr-2023-1 | ['boundary-detection'] | ['computer-vision'] | [ 4.20328498e-01 2.17092093e-02 6.38547242e-02 7.81562179e-02
-7.30370358e-02 -9.38668191e-01 4.02708769e-01 3.97012949e-01
-1.84315041e-01 5.98183215e-01 -5.65904737e-01 -4.78715599e-01
-1.09266408e-01 -7.98486054e-01 -5.23439586e-01 -6.60928547e-01
-2.78802186e-01 6.83938324e-01 6.39147639e-01 2.70728301e-03
4.54751998e-01 6.45725012e-01 -1.07932472e+00 -8.88409019e-02
9.05909002e-01 7.18011916e-01 6.19828463e-01 9.48252499e-01
-2.07384139e-01 1.14380546e-01 -4.56130236e-01 -9.86785516e-02
2.02008173e-01 -3.29488009e-01 -1.30864990e+00 3.40152681e-01
1.06442183e-01 -3.42056185e-01 2.82789707e-01 7.90665150e-01
2.65428811e-01 -4.36546177e-01 6.83876872e-01 -8.72421741e-01
-8.01204592e-02 4.97102112e-01 -9.52928126e-01 3.82580698e-01
-5.42754447e-03 -1.74765527e-01 9.64567661e-01 -6.61862552e-01
1.00362623e+00 6.42821133e-01 6.75413132e-01 4.35230881e-01
-1.59996331e+00 -1.51524274e-02 -3.77487652e-02 -5.71586005e-02
-1.36880791e+00 -2.27266729e-01 6.48953915e-01 -7.13672757e-01
3.75270456e-01 4.17465299e-01 7.14966774e-01 2.52181917e-01
-1.44108221e-01 7.32642889e-01 1.13377368e+00 -3.28948587e-01
4.38666880e-01 -3.01098049e-01 1.16107292e-01 6.21510804e-01
2.60770112e-01 -2.81282783e-01 5.50384596e-02 -1.76564515e-01
1.29790258e+00 1.71227138e-02 -3.50546956e-01 -4.23826694e-01
-1.05891907e+00 4.36602682e-01 4.54411954e-01 3.95472407e-01
-1.48878932e-01 3.88202555e-02 2.70558327e-01 -4.93713319e-02
4.08933789e-01 3.36624235e-01 -6.16333783e-01 1.76994935e-01
-1.23635352e+00 3.04568887e-01 8.25557947e-01 7.55778730e-01
7.33555138e-01 -5.81454158e-01 2.70273238e-01 7.33178556e-01
-8.16824138e-02 -1.06226549e-01 -1.80408895e-01 -1.24465334e+00
-2.03612283e-01 7.32890010e-01 1.67687461e-01 -4.86370623e-01
-5.09708762e-01 -3.23380858e-01 -7.98964858e-01 3.97437550e-02
9.76505637e-01 -4.93082628e-02 -7.69214988e-01 1.28053546e+00
7.39124656e-01 1.50916368e-01 -3.44605923e-01 5.70071876e-01
7.86643445e-01 5.28648734e-01 -1.50909856e-01 -3.30305904e-01
1.63780379e+00 -6.38023734e-01 -2.82243133e-01 2.19681896e-02
3.79863560e-01 -8.20409834e-01 6.21020973e-01 2.16740951e-01
-1.15205729e+00 -1.22695476e-01 -6.45313323e-01 4.79944423e-02
-4.16846424e-01 3.03005517e-01 7.26650417e-01 4.27286401e-02
-1.13390756e+00 8.30285013e-01 -6.43792868e-01 -7.37818003e-01
4.99943316e-01 6.41404465e-02 -1.81201816e-01 3.12289268e-01
-3.72948736e-01 3.60715985e-01 1.51679233e-01 1.50292560e-01
-7.63697743e-01 -7.31182992e-01 -3.51179183e-01 -1.03906907e-01
5.97539067e-01 -5.84598184e-01 1.37864685e+00 -5.39835095e-01
-1.15585065e+00 1.50699151e+00 -3.24958295e-01 -4.22707796e-01
7.57485807e-01 8.32561180e-02 4.58948731e-01 3.90811294e-01
7.06809759e-02 6.48333251e-01 6.30249202e-01 -1.48478198e+00
-6.60212338e-01 -7.38401890e-01 2.67118849e-02 -1.49211004e-01
3.03106278e-01 -1.49097994e-01 -3.11427087e-01 -5.10069668e-01
6.19329691e-01 -8.14351082e-01 -4.72245693e-01 1.75027311e-01
-4.63772893e-01 -3.06310147e-01 1.02932143e+00 -6.30894840e-01
1.22242177e+00 -1.71820140e+00 4.94946018e-02 -9.04403031e-02
5.04985332e-01 2.13339001e-01 3.05420786e-01 6.74231231e-01
1.09244496e-01 5.94984531e-01 -6.43609226e-01 4.41648662e-02
-4.74789828e-01 1.41016796e-01 -1.67953089e-01 4.76822793e-01
3.07431184e-02 6.53879762e-01 -9.85761344e-01 -6.13644719e-01
-1.54419720e-01 1.92656100e-01 -5.01271114e-02 6.77814856e-02
-3.29267502e-01 5.93731165e-01 -2.44115651e-01 6.74833119e-01
1.06324422e+00 -7.70671964e-01 2.92892545e-01 1.14663832e-01
-6.61252141e-01 6.02790453e-02 -9.96276677e-01 1.34535015e+00
-1.01293780e-01 4.59544539e-01 5.35458863e-01 -1.09091902e+00
8.18184435e-01 1.19112901e-01 8.10028672e-01 2.08628569e-02
-9.30671170e-02 2.84572303e-01 -3.56703624e-02 -2.24503696e-01
2.85138607e-01 1.04592510e-01 3.27610761e-01 3.46917421e-01
-2.00488836e-01 -5.28224230e-01 5.40083826e-01 2.11476624e-01
1.14750791e+00 2.42438346e-01 4.70840842e-01 -5.86580098e-01
4.51237977e-01 2.36884907e-01 7.84744859e-01 7.07263947e-01
-3.37214887e-01 7.51345873e-01 7.90829778e-01 -5.85058630e-01
-1.35067546e+00 -9.79329944e-01 -6.62605822e-01 5.14601409e-01
3.05756032e-01 -5.35066962e-01 -1.04304087e+00 -4.91676509e-01
-3.09428629e-02 -2.63026040e-02 -6.77961767e-01 6.84768319e-01
-5.15688896e-01 -7.33501077e-01 3.65076393e-01 8.37978125e-02
6.62666917e-01 -8.98005724e-01 -1.07498264e+00 1.85119271e-01
-2.85808265e-01 -1.11578071e+00 -1.17501415e-01 2.62579083e-01
-1.24201953e+00 -1.37313962e+00 -1.01991391e+00 -8.87642503e-01
8.61423373e-01 2.70091742e-01 1.04738069e+00 4.54137206e-01
-8.15762281e-01 -3.18441764e-02 -3.98800403e-01 -3.09627414e-01
-7.01253787e-02 1.50700852e-01 -5.68474114e-01 -4.19029713e-01
-1.15154490e-01 -6.92153037e-01 -9.90383089e-01 5.10555208e-01
-7.57936776e-01 2.89607227e-01 7.92416334e-02 7.45244443e-01
8.77105892e-01 2.40561634e-01 2.57082105e-01 -1.14328492e+00
2.89920121e-01 -2.87576556e-01 -1.01901126e+00 2.46278659e-01
-2.95625657e-01 -3.24838221e-01 7.20346093e-01 -6.81219548e-02
-5.53426921e-01 2.24568292e-01 5.27773723e-02 2.21286178e-01
-4.97925550e-01 2.90024847e-01 2.44656160e-01 6.90672845e-02
3.80015045e-01 2.88694561e-01 8.69836956e-02 -6.32763028e-01
-2.69934256e-02 4.63802159e-01 5.55482447e-01 -6.16377771e-01
5.37278175e-01 1.00348866e+00 4.18211788e-01 -1.26548374e+00
-5.52225649e-01 -7.33754396e-01 -9.32448208e-01 -3.61518055e-01
5.46553969e-01 -2.35012025e-01 -8.43423784e-01 7.30051816e-01
-1.26498759e+00 -8.40905249e-01 -2.63748676e-01 -2.55679756e-01
-6.55976772e-01 7.06080019e-01 -8.78289998e-01 -9.35585141e-01
-5.24115026e-01 -7.42648602e-01 1.16552567e+00 4.92725164e-01
8.27121511e-02 -1.03124380e+00 8.96506906e-02 9.76255015e-02
2.04488292e-01 5.66799760e-01 9.17495549e-01 -2.87783772e-01
-7.02877939e-01 7.38784000e-02 -2.49081865e-01 -7.39456043e-02
9.94199514e-02 7.78328419e-01 -7.29442120e-01 -3.74619849e-03
-1.37083143e-01 2.15271771e-01 9.10549283e-01 8.86079967e-01
1.25358033e+00 -1.96730927e-01 -6.55230880e-01 6.39666736e-01
1.64480233e+00 8.57680812e-02 5.95304906e-01 2.59424657e-01
2.83888787e-01 9.49220836e-01 5.85747182e-01 6.93531215e-01
2.88169444e-01 6.06064677e-01 4.66353625e-01 -2.85631001e-01
7.89851397e-02 1.92884073e-01 -1.79922014e-01 4.53998357e-01
-4.26701993e-01 -5.22850156e-02 -1.27896249e+00 7.78714657e-01
-1.85137904e+00 -1.03626990e+00 -6.43387079e-01 2.32327223e+00
6.96966290e-01 7.64873698e-02 4.61640447e-01 1.97103560e-01
8.84226382e-01 -9.98890772e-02 -5.84490478e-01 -3.35262179e-01
-6.35706633e-02 1.91462427e-01 5.84827483e-01 5.54004669e-01
-1.08868945e+00 1.08185649e+00 6.37645435e+00 7.04265535e-01
-9.52588975e-01 -2.55103916e-01 9.66088235e-01 5.87260365e-01
3.10399141e-02 5.18298626e-01 -8.43053997e-01 3.88709724e-01
4.10298735e-01 -6.57245293e-02 4.07209307e-01 4.67080057e-01
4.68256503e-01 -5.42776942e-01 -8.38204324e-01 5.45031309e-01
-6.08566880e-01 -1.39472067e+00 -3.32016468e-01 3.05140942e-01
5.35837471e-01 -1.71787199e-02 -4.07134354e-01 -5.66155076e-01
3.43155563e-01 -7.53513813e-01 4.30079192e-01 3.03875327e-01
6.87636971e-01 -2.54578710e-01 4.23631847e-01 5.26997983e-01
-1.50027120e+00 1.18663572e-01 -3.95140111e-01 -4.87583950e-02
4.20303941e-02 9.94823396e-01 -1.14323223e+00 5.68289518e-01
8.75411153e-01 6.21253848e-01 -5.60364783e-01 1.42426789e+00
-2.44204462e-01 6.74501896e-01 -6.06423914e-01 2.64777511e-01
9.68358889e-02 -6.99786723e-01 6.19444847e-01 1.03390181e+00
2.36114427e-01 3.04153990e-02 5.53489253e-02 1.04440701e+00
1.22924946e-01 1.07789509e-01 -7.13304698e-01 5.51292002e-02
5.17986059e-01 1.60856295e+00 -1.56708503e+00 -1.94392487e-01
-1.77755542e-02 8.39497089e-01 3.02785158e-01 1.59702912e-01
-4.37053978e-01 -3.13846052e-01 2.53970981e-01 6.57610118e-01
5.52314460e-01 -2.60772854e-01 -3.69463384e-01 -8.25263560e-01
4.07990851e-02 -3.07207555e-01 5.32226264e-01 -6.03850186e-01
-1.00027442e+00 2.72528946e-01 -1.43830702e-01 -8.78883898e-01
1.98813036e-01 -6.39981091e-01 -9.39373612e-01 4.61439133e-01
-1.29607546e+00 -1.15948462e+00 -4.94886518e-01 8.84782895e-02
5.37707806e-01 7.04287112e-01 7.60594130e-01 -1.73245236e-01
-5.88638186e-01 -1.27986342e-01 3.67102116e-01 2.00636517e-02
2.42306754e-01 -1.63123965e+00 4.49691921e-01 7.87849784e-01
-2.39828780e-01 6.12485349e-01 7.26825297e-01 -6.20005071e-01
-7.58177876e-01 -8.43404651e-01 8.17786336e-01 -2.16639861e-01
6.64093912e-01 -5.19233108e-01 -8.20649028e-01 4.16377574e-01
-6.25509210e-03 -4.90052290e-02 5.16852736e-01 -7.76961446e-02
1.33695796e-01 4.68744561e-02 -1.12512183e+00 6.69374287e-01
1.06173217e+00 -1.81352571e-02 8.57092962e-02 6.73298597e-01
1.66806713e-01 -2.84354299e-01 -1.07114065e+00 4.10430044e-01
5.27903199e-01 -9.42792833e-01 8.93219113e-01 -9.43726301e-02
4.96271253e-01 -6.42041147e-01 3.48662317e-01 -8.14085186e-01
-1.99318260e-01 -8.79951239e-01 4.55094129e-02 1.29765761e+00
3.35260957e-01 -4.37666893e-01 9.05272841e-01 -5.95923848e-02
1.77215725e-01 -1.07263744e+00 -8.49274457e-01 -7.72650480e-01
2.12997526e-01 1.73110873e-01 3.90674531e-01 8.17318439e-01
1.43506885e-01 2.07018435e-01 2.12043241e-01 1.29345149e-01
9.22540903e-01 6.08862758e-01 7.65373588e-01 -1.61068237e+00
-5.90453781e-02 -5.44586897e-01 -2.00478494e-01 -1.22202897e+00
-3.31969649e-01 -6.57268286e-01 -6.54779673e-02 -1.80528045e+00
2.07317263e-01 -6.43201530e-01 3.13413501e-01 2.97596782e-01
-7.41058737e-02 -7.95494840e-02 -7.67925903e-02 5.10372162e-01
-4.63175893e-01 -1.13131307e-01 1.38832295e+00 1.41049564e-01
-2.39517525e-01 3.43457401e-01 -3.39174539e-01 8.29827428e-01
1.17750573e+00 -3.02182555e-01 -6.44101351e-02 -1.10592596e-01
-4.08166312e-02 2.59359241e-01 5.42883575e-01 -6.92413568e-01
1.67496368e-01 -2.63542652e-01 8.75174254e-02 -7.50393927e-01
5.02624027e-02 -6.97434723e-01 2.57111579e-01 7.14531183e-01
5.65335900e-02 -2.07311213e-02 1.75585542e-02 3.03526759e-01
1.27814531e-01 -6.01505458e-01 1.01765895e+00 -6.87575758e-01
-4.86247748e-01 2.86939770e-01 -3.68869364e-01 -4.38651368e-02
1.18554330e+00 -5.75717151e-01 -6.95866644e-01 -5.67348190e-02
-9.13088977e-01 3.36482644e-01 9.18357313e-01 -3.81278336e-01
4.14321899e-01 -5.37820458e-01 -8.07643712e-01 -2.41916612e-01
-1.38601273e-01 5.88123918e-01 9.12171826e-02 1.22187793e+00
-1.18496406e+00 2.53371119e-01 -1.43926561e-01 -8.43517244e-01
-1.38964808e+00 4.10218269e-01 4.49565470e-01 -6.33807003e-01
-7.65144229e-01 8.23511720e-01 5.39695621e-01 -3.38107318e-01
-1.32447049e-01 -2.10762545e-01 -2.42098600e-01 -3.28846425e-02
2.24931002e-01 5.03388941e-01 -1.33040950e-01 -4.35503662e-01
-3.72901618e-01 7.93266177e-01 -4.56667691e-02 2.28477627e-01
1.42295277e+00 -1.51693657e-01 -6.53791904e-01 3.63370568e-01
5.27931333e-01 -2.03101620e-01 -1.21700430e+00 2.13994868e-02
2.74197161e-01 -3.45558465e-01 -1.34028316e-01 -6.59252584e-01
-9.82045829e-01 8.39944422e-01 4.05497760e-01 8.30865026e-01
1.15317154e+00 2.38607153e-01 5.55398762e-01 -4.38662469e-02
3.80141109e-01 -1.05750465e+00 -2.77211875e-01 3.75299484e-01
7.45042980e-01 -8.79547954e-01 1.73114464e-01 -9.67835307e-01
-2.04497769e-01 1.20543838e+00 5.55775583e-01 -1.78458676e-01
7.84552813e-01 3.26952904e-01 -5.08506484e-02 -3.70045036e-01
-7.01608956e-01 -4.75760221e-01 -3.96119297e-01 5.29862046e-01
5.30217052e-01 1.12629153e-01 -7.53275692e-01 1.04629576e-01
-1.43029559e-02 -2.13898569e-02 7.20702589e-01 1.10834694e+00
-6.14701748e-01 -1.08709276e+00 -3.29974115e-01 5.09536743e-01
-5.20523548e-01 4.92566340e-02 -8.02057028e-01 6.40911639e-01
1.16097145e-01 6.91550672e-01 1.50961488e-01 -2.93686781e-02
3.29235345e-02 -2.92716652e-01 4.72930044e-01 -5.26401281e-01
-4.14653331e-01 3.90264511e-01 -1.07786819e-01 -1.36966661e-01
-4.33361918e-01 -7.67405152e-01 -1.41627610e+00 -4.14407104e-01
-4.10827309e-01 2.87013739e-01 5.54503620e-01 5.52548409e-01
3.69199008e-01 1.02696985e-01 5.76292217e-01 -6.85799778e-01
7.58327842e-02 -7.37751961e-01 -8.51696908e-01 2.27213174e-01
2.10516006e-01 -4.59284306e-01 -3.48709613e-01 4.13817793e-01] | [14.435789108276367, -3.1395206451416016] |
6d667163-5cf7-42d6-b57d-1840733cdf8a | a-novel-ensemble-deep-learning-model-for | 2007.12620 | null | https://arxiv.org/abs/2007.12620v1 | https://arxiv.org/pdf/2007.12620v1.pdf | A Novel Ensemble Deep Learning Model for Stock Prediction Based on Stock Prices and News | In recent years, machine learning and deep learning have become popular methods for financial data analysis, including financial textual data, numerical data, and graphical data. This paper proposes to use sentiment analysis to extract useful information from multiple textual data sources and a blending ensemble deep learning model to predict future stock movement. The blending ensemble model contains two levels. The first level contains two Recurrent Neural Networks (RNNs), one Long-Short Term Memory network (LSTM) and one Gated Recurrent Units network (GRU), followed by a fully connected neural network as the second level model. The RNNs, LSTM, and GRU models can effectively capture the time-series events in the input data, and the fully connected neural network is used to ensemble several individual prediction results to further improve the prediction accuracy. The purpose of this work is to explain our design philosophy and show that ensemble deep learning technologies can truly predict future stock price trends more effectively and can better assist investors in making the right investment decision than other traditional methods. | ['Yang Li', 'Yi Pan'] | 2020-07-23 | null | null | null | null | ['stock-prediction'] | ['time-series'] | [-4.53747749e-01 -3.68782729e-01 -1.77934483e-01 -3.78017634e-01
-1.78570837e-01 -2.61998266e-01 3.91874045e-01 -1.65880576e-01
-2.17780739e-01 6.75730228e-01 5.11431158e-01 -8.10310721e-01
3.00920725e-01 -1.25442588e+00 -3.54927808e-01 -5.35205662e-01
-1.68268546e-01 -1.04493268e-01 -4.76341061e-02 -7.36629903e-01
6.39490008e-01 5.31061232e-01 -1.52506244e+00 4.12905127e-01
6.45947158e-01 1.65072179e+00 -2.83960134e-01 4.90965396e-01
-7.34370828e-01 1.90564907e+00 -7.80307472e-01 -3.49056631e-01
2.60386109e-01 -2.05482155e-01 -2.15865389e-01 -5.11657536e-01
-5.53386033e-01 -6.55014575e-01 -4.53520805e-01 7.50959456e-01
5.90872824e-01 1.61411420e-01 4.68189746e-01 -1.02600658e+00
-9.34955895e-01 1.06236851e+00 -6.20297909e-01 3.96666229e-01
-3.00137311e-01 -2.11835206e-01 8.28279495e-01 -8.81771684e-01
7.33412802e-02 8.42657328e-01 8.17571402e-01 1.81121200e-01
-3.67913961e-01 -1.03351641e+00 2.68289298e-01 1.44011915e-01
-6.22583508e-01 -7.68059716e-02 9.47159350e-01 -5.13058841e-01
1.33935249e+00 -3.34826261e-02 9.40499842e-01 9.46456790e-01
7.46169388e-01 9.55966234e-01 9.19055641e-01 -2.06219167e-01
3.98772731e-02 1.47377655e-01 4.01405185e-01 2.47282058e-01
1.95277169e-01 1.63614601e-01 -3.28707486e-01 4.48533036e-02
8.39170814e-01 9.03046250e-01 2.02111945e-01 6.81696177e-01
-7.08950102e-01 1.06289732e+00 5.58954060e-01 8.57149661e-01
-7.38433540e-01 1.59572840e-01 5.93610227e-01 4.95471895e-01
1.07156289e+00 5.43402135e-01 -7.74823964e-01 -2.39792198e-01
-1.10699844e+00 1.84049666e-01 7.66563177e-01 2.56883085e-01
2.70842284e-01 9.48469400e-01 1.83138087e-01 6.69142187e-01
2.96954036e-01 4.15982336e-01 1.38430154e+00 -2.03036189e-01
4.65679258e-01 9.67916429e-01 9.36919544e-03 -1.28047979e+00
-5.60077488e-01 -8.65199149e-01 -9.01607633e-01 4.03033674e-01
-2.14483738e-01 -8.38057816e-01 -7.78777182e-01 1.05180430e+00
-3.03779572e-01 2.79374391e-01 3.75473022e-01 3.36425006e-01
9.28413272e-01 1.23241019e+00 -1.26852319e-01 -1.09118029e-01
9.99392152e-01 -9.74429011e-01 -9.57313359e-01 -4.04313207e-02
8.37599754e-01 -4.46928024e-01 1.45950735e-01 3.42772394e-01
-1.03310275e+00 -6.59360588e-01 -1.35071802e+00 1.51402932e-02
-1.08185863e+00 1.20556615e-01 4.81752545e-01 2.75394320e-01
-8.25100005e-01 9.18054283e-01 -5.36783040e-01 6.22058809e-01
1.61618233e-01 3.71289313e-01 1.38989672e-01 6.90478265e-01
-1.70129383e+00 1.17297244e+00 6.38627887e-01 5.39171457e-01
6.26766160e-02 -3.91756386e-01 -8.17336321e-01 2.60781735e-01
-2.70668060e-01 -4.22795683e-01 1.28232002e+00 -1.15037477e+00
-1.72864950e+00 2.26368636e-01 7.66140223e-02 -9.22877192e-01
2.47325614e-01 -3.73917133e-01 -9.97629166e-01 -5.78162134e-01
-2.43613631e-01 -3.80615368e-02 7.05630660e-01 -6.15191460e-01
-8.78832221e-01 -5.30654073e-01 -6.09683514e-01 -6.88061267e-02
-5.35589099e-01 1.36010557e-01 3.34210098e-01 -1.36845386e+00
5.51896580e-02 -3.90302092e-01 -4.30185378e-01 -1.16956437e+00
-1.17408335e-01 -3.67947668e-01 1.07370842e+00 -1.11561739e+00
1.69724238e+00 -1.88240767e+00 -3.35911781e-01 3.39348167e-01
6.08493462e-02 3.60313624e-01 1.19147524e-01 4.10740495e-01
-5.12688041e-01 1.41633853e-01 1.97652474e-01 -1.95516989e-01
-5.22927456e-02 -8.32344666e-02 -1.00502217e+00 -2.40696102e-01
1.23145543e-01 1.34159625e+00 -5.23151219e-01 1.25011161e-01
1.95977837e-01 3.54295731e-01 2.09846169e-01 1.30334958e-01
-1.83075801e-01 2.60202289e-02 -5.53846180e-01 4.38903809e-01
3.90460074e-01 -3.13769102e-01 -9.41348597e-02 -4.95463423e-02
-4.87161249e-01 4.72401947e-01 -1.03847134e+00 6.36963606e-01
-4.85893816e-01 7.03868628e-01 -7.40586281e-01 -1.08734620e+00
1.61461210e+00 3.46775651e-01 4.02849793e-01 -1.15945840e+00
4.55166876e-01 6.54142499e-01 -1.31010905e-01 -2.87012249e-01
8.22342336e-01 -3.04214656e-01 -1.21633895e-01 8.31927538e-01
-7.78587013e-02 4.02105421e-01 -5.98236062e-02 -2.48348475e-01
5.24585962e-01 -1.99057534e-02 1.83535814e-01 1.87508970e-01
3.15237284e-01 -3.85627240e-01 7.39979744e-01 3.67966801e-01
3.89617324e-01 1.52853891e-01 4.77651268e-01 -1.16739678e+00
-8.49380732e-01 -2.87022531e-01 1.24604896e-01 1.10058606e+00
-2.36596256e-01 -1.11749896e-03 -2.06036568e-01 -2.98389405e-01
1.77526876e-01 1.01030266e+00 -7.76522458e-01 -5.78070283e-02
-4.43479210e-01 -9.94761527e-01 3.31123769e-01 1.06566215e+00
8.21471930e-01 -1.55817449e+00 -5.51247776e-01 7.36497819e-01
2.52075464e-01 -5.20615876e-01 2.52868701e-02 2.25141883e-01
-1.02796173e+00 -5.88043809e-01 -9.62579489e-01 -4.56524104e-01
3.82206552e-02 -2.64799446e-01 1.03703547e+00 7.71162584e-02
3.95210385e-01 -4.22431916e-01 -4.99439001e-01 -1.03016984e+00
-6.15280084e-02 1.22346170e-01 2.30087433e-02 2.04080701e-01
5.51084340e-01 -5.62333465e-01 -3.93535495e-01 -1.43879086e-01
-7.24170625e-01 6.25176132e-02 4.95522171e-01 7.17181146e-01
1.44613206e-01 3.14473480e-01 1.06403887e+00 -7.97783017e-01
9.58766460e-01 -7.94295311e-01 -7.08563745e-01 2.69815952e-01
-8.08845699e-01 1.35612100e-01 5.35499811e-01 -6.21833988e-02
-1.08650374e+00 -6.55079424e-01 -3.11351389e-01 -1.60789594e-01
5.47936141e-01 1.28243685e+00 5.18826246e-01 2.73548484e-01
1.55643329e-01 5.14411032e-01 -2.48936564e-02 -5.92492938e-01
-1.50155649e-01 7.90988028e-01 2.89188623e-01 1.40778780e-01
2.33299673e-01 -5.91913909e-02 -3.56945544e-01 -3.60874057e-01
-8.69370341e-01 1.05416283e-01 -4.49570000e-01 -4.17063385e-01
6.03614092e-01 -9.53528821e-01 -5.56311727e-01 1.16135275e+00
-1.03613544e+00 -1.51234433e-01 -1.70342818e-01 6.24413729e-01
-1.93416476e-02 -2.82469898e-01 -1.20716846e+00 -1.25755239e+00
-8.57360125e-01 -8.43166471e-01 4.49897736e-01 6.68614924e-01
-5.03245629e-02 -1.43733454e+00 1.97677627e-01 -5.77407219e-02
7.94752538e-01 6.82164133e-01 6.30175471e-01 -1.25857031e+00
-6.74010813e-02 -7.74048150e-01 -1.25419408e-01 6.17474616e-01
5.92332110e-02 1.86299935e-01 -8.19837034e-01 4.41725180e-02
3.17535698e-01 -6.27238601e-02 1.27713454e+00 6.78008318e-01
1.13732958e+00 -5.37529111e-01 -3.32159474e-02 4.66446877e-01
1.19592857e+00 9.54101443e-01 7.74154425e-01 6.84580624e-01
6.01929009e-01 4.13056076e-01 1.64979454e-02 4.58700538e-01
5.71139157e-01 -5.38495444e-02 1.85497180e-01 -2.23984942e-01
8.99603963e-01 -1.71395361e-01 7.13185608e-01 1.26724613e+00
-5.56080222e-01 -1.45826396e-02 -9.31639314e-01 1.46953210e-01
-2.07722449e+00 -1.31281030e+00 8.68365616e-02 1.73514843e+00
3.44272375e-01 4.87981528e-01 2.89556980e-01 2.51821369e-01
5.35244524e-01 3.60018730e-01 -8.23855102e-01 -6.67460501e-01
-3.89599055e-01 2.55935073e-01 5.88589013e-01 -5.69571704e-02
-1.06861401e+00 4.56519812e-01 6.30382204e+00 5.04887342e-01
-1.61208773e+00 -2.93785989e-01 1.14721513e+00 -8.11625496e-02
-4.24994946e-01 -5.24942279e-01 -8.59792054e-01 9.00404334e-01
1.27755344e+00 -3.52339238e-01 -2.63262987e-01 1.00555956e+00
1.62859961e-01 3.02261561e-01 -4.61388201e-01 7.31398940e-01
-1.73878506e-01 -1.86510277e+00 2.52265364e-01 2.15683892e-01
8.41114283e-01 2.12983906e-01 4.41336751e-01 6.96217239e-01
6.33624256e-01 -1.16652715e+00 5.47767520e-01 1.24977887e+00
1.28560334e-01 -1.19402015e+00 1.44704890e+00 3.02167565e-01
-1.31048763e+00 -4.73989338e-01 -1.58612832e-01 -5.79180777e-01
1.34027690e-01 7.42789984e-01 -1.11471072e-01 8.71700287e-01
8.99888933e-01 1.39274263e+00 -3.58707607e-01 5.59510171e-01
9.61592495e-02 4.04718965e-01 -6.07360154e-02 -4.42954749e-01
5.73658526e-01 -3.63872766e-01 5.57937436e-02 9.87821519e-01
6.76010668e-01 3.73700827e-01 -1.46285921e-01 5.77402234e-01
-1.42695338e-01 1.67514071e-01 -4.73795861e-01 -3.03422511e-01
1.78999752e-01 1.04001677e+00 -6.29905879e-01 -6.59418344e-01
-4.06127810e-01 4.16253120e-01 5.93790784e-02 3.98961842e-01
-4.07576948e-01 -8.05113375e-01 3.31312239e-01 -2.68554240e-01
3.60718548e-01 3.89792062e-02 -7.26820946e-01 -1.48478961e+00
-1.18976898e-01 -5.93038559e-01 4.87752527e-01 -8.76260936e-01
-1.49384773e+00 8.26964319e-01 -5.50820768e-01 -1.38316727e+00
-8.70149791e-01 -1.01914489e+00 -1.27283657e+00 1.26060390e+00
-1.35499704e+00 -8.01033914e-01 2.30976105e-01 3.60092223e-01
3.53042394e-01 -9.96066928e-01 6.35719240e-01 -2.98626348e-02
-1.13718832e+00 1.46042988e-01 5.23722351e-01 8.44363511e-01
1.22114338e-01 -1.19584572e+00 6.42926872e-01 8.35491598e-01
-1.45941854e-01 5.43257415e-01 2.36892030e-01 -7.41034031e-01
-6.21360898e-01 -1.10734737e+00 9.10014927e-01 3.07347327e-02
9.46751833e-01 1.78743377e-01 -1.04929328e+00 9.18675363e-01
2.68070340e-01 -2.79664129e-01 8.83336008e-01 2.63989102e-02
-3.03706586e-01 -1.80120140e-01 -6.87885642e-01 3.95856410e-01
-3.82332355e-02 -3.31520140e-01 -8.81817222e-01 -2.90787339e-01
7.12023079e-01 -3.37794870e-01 -1.02402115e+00 5.32483995e-01
8.36826921e-01 -1.16827476e+00 5.71076453e-01 -7.10175753e-01
8.79869521e-01 8.62907171e-02 3.07334606e-02 -1.49378884e+00
-2.61499107e-01 -1.25848353e-01 -5.20000160e-01 1.10846496e+00
8.36955547e-01 -1.19898760e+00 5.67450225e-01 7.67042935e-01
-1.57881249e-02 -1.16829562e+00 -3.90126735e-01 -3.66053164e-01
2.56491661e-01 -4.16184366e-01 1.16834080e+00 1.14229250e+00
1.53705880e-01 2.82976836e-01 -5.93271613e-01 -3.57860535e-01
9.31929648e-02 4.27553773e-01 3.14455479e-01 -1.35464549e+00
-1.39832394e-02 -1.02133858e+00 -2.83424228e-01 -7.41937160e-01
6.36986196e-02 -3.63342941e-01 -6.81079209e-01 -1.49318027e+00
-2.99497902e-01 -4.87210266e-02 -1.03865004e+00 4.26921159e-01
-1.02501348e-01 -1.23498410e-01 2.16124699e-01 3.21567923e-01
-1.28859162e-01 5.91069162e-01 9.06311452e-01 -1.87634975e-01
-4.51842785e-01 3.72172862e-01 -7.74628580e-01 8.60499561e-01
1.02421522e+00 9.86997113e-02 7.76277557e-02 -2.47696549e-01
6.07074738e-01 2.64448464e-01 -2.69712415e-02 -6.69144928e-01
3.71087641e-01 1.71773165e-01 1.17625022e+00 -1.14274371e+00
-1.57317698e-01 -4.83295172e-01 6.19589686e-02 6.91325307e-01
-3.85504454e-01 5.74474096e-01 2.53881514e-01 4.72959802e-02
-7.85960317e-01 -1.42299205e-01 1.87704578e-01 -2.30856866e-01
-8.77768397e-01 4.24844682e-01 -4.46154892e-01 -5.56004405e-01
9.07799184e-01 -2.27452010e-01 -1.79987669e-01 -6.40842140e-01
-6.98308825e-01 4.81363267e-01 -3.10128391e-01 7.35275865e-01
7.71124005e-01 -1.59556866e+00 -8.36623013e-01 3.66779774e-01
-4.67843443e-01 -3.15010667e-01 3.30796987e-01 4.22286659e-01
-5.54704785e-01 4.76253122e-01 -3.31591487e-01 2.13551953e-01
-5.19449115e-01 2.64383286e-01 7.17302203e-01 -8.09932828e-01
-3.46777290e-01 8.34593534e-01 -3.33961159e-01 -2.72463232e-01
-1.31561374e-02 -3.75520974e-01 -1.02395058e+00 6.39720976e-01
1.00213122e+00 5.55419683e-01 1.56769589e-01 -4.67340916e-01
-7.60344276e-03 5.55494845e-01 -4.44853395e-01 -4.44495343e-02
1.96815848e+00 2.37974599e-01 -2.99276918e-01 1.00710380e+00
1.20795345e+00 -3.93780261e-01 -8.03101838e-01 -3.14293236e-01
4.60219443e-01 2.54507303e-01 3.91444325e-01 -6.43478334e-01
-1.46922326e+00 8.70858848e-01 1.99103087e-01 8.16870630e-01
1.26435518e+00 -7.27217019e-01 1.30854928e+00 2.97642946e-01
7.00916303e-03 -1.41357028e+00 -9.51451808e-02 8.48341286e-01
8.62086475e-01 -1.06130171e+00 -2.38269418e-01 6.18377328e-01
-7.19678223e-01 1.75195694e+00 4.38372910e-01 -4.96704608e-01
1.03215730e+00 4.99337673e-01 2.33966097e-01 -9.92707610e-02
-9.92362797e-01 6.42093047e-02 3.15152109e-01 -1.12647355e-01
6.60914838e-01 -1.34364367e-01 2.09948257e-01 1.21970475e+00
-6.05641484e-01 2.17523068e-01 3.52253973e-01 8.48806202e-01
-5.17744362e-01 -8.93409312e-01 -3.40870291e-01 1.04724097e+00
-9.07180071e-01 -2.31723651e-01 -2.43956655e-01 6.02805972e-01
-1.62862867e-01 7.05959320e-01 8.14639926e-01 -9.78808403e-01
4.25367713e-01 3.11771125e-01 -3.49595815e-01 -2.53096342e-01
-1.04153264e+00 -1.60418805e-02 -2.80525446e-01 -1.74905300e-01
-2.98885763e-01 -4.76628810e-01 -1.12687576e+00 -6.36157215e-01
-2.22063899e-01 2.85119504e-01 5.20677328e-01 1.19857252e+00
3.33667368e-01 9.53867197e-01 1.25736845e+00 -9.51160729e-01
-4.53204453e-01 -1.09759355e+00 -8.77359629e-01 -1.17016099e-01
5.85882425e-01 -3.74412328e-01 -2.97577232e-01 -9.75056738e-02] | [4.435201644897461, 4.2537360191345215] |
4df92dca-197c-4056-bdcf-4994656c5c11 | road-rutting-detection-using-deep-learning-on | 2209.14225 | null | https://arxiv.org/abs/2209.14225v1 | https://arxiv.org/pdf/2209.14225v1.pdf | Road Rutting Detection using Deep Learning on Images | Road rutting is a severe road distress that can cause premature failure of road incurring early and costly maintenance costs. Research on road damage detection using image processing techniques and deep learning are being actively conducted in the past few years. However, these researches are mostly focused on detection of cracks, potholes, and their variants. Very few research has been done on the detection of road rutting. This paper proposes a novel road rutting dataset comprising of 949 images and provides both object level and pixel level annotations. Object detection models and semantic segmentation models were deployed to detect road rutting on the proposed dataset, and quantitative and qualitative analysis of model predictions were done to evaluate model performance and identify challenges faced in the detection of road rutting using the proposed method. Object detection model YOLOX-s achieves mAP@IoU=0.5 of 61.6% and semantic segmentation model PSPNet (Resnet-50) achieves IoU of 54.69 and accuracy of 72.67, thus providing a benchmark accuracy for similar work in future. The proposed road rutting dataset and the results of our research study will help accelerate the research on detection of road rutting using deep learning. | ['Yoshihide Sekimoto', 'Hiroya Maeda', 'Ashutosh Kumar', 'Deeksha Arya', 'Poonam Kumari Saha'] | 2022-09-28 | null | null | null | null | ['road-damage-detection'] | ['computer-vision'] | [ 3.49945426e-01 5.47998734e-02 -3.90273891e-02 -2.84933209e-01
-6.38468564e-01 1.42763346e-01 1.05710119e-01 -1.76651984e-01
-7.23733529e-02 5.85913360e-01 -2.42664009e-01 -4.74611700e-01
1.42498631e-02 -1.49913454e+00 -6.56163216e-01 -4.08979595e-01
9.64060053e-02 5.89045174e-02 8.73922050e-01 -1.73239037e-01
8.99980903e-01 7.13973939e-01 -1.91725183e+00 2.73041148e-02
1.09452331e+00 1.05690396e+00 3.96995991e-01 8.87024164e-01
1.94625884e-01 6.63809180e-01 -6.84837043e-01 3.37607856e-03
2.80136675e-01 9.52689648e-02 -9.38617647e-01 -2.53234114e-02
7.20530570e-01 -6.50975406e-01 -4.74870831e-01 7.11665332e-01
6.99945033e-01 4.89748567e-02 8.77488852e-01 -1.09681726e+00
-6.10397875e-01 1.30082190e-01 -1.05120683e+00 6.89600825e-01
3.57136689e-02 7.64819235e-02 6.33279681e-01 -1.06812310e+00
2.32093796e-01 1.30982208e+00 8.59004796e-01 9.47408974e-02
-5.86130738e-01 -8.37569177e-01 -3.33990097e-01 6.20630443e-01
-1.61158013e+00 -1.70205429e-01 5.51278889e-01 -4.86879081e-01
1.13978302e+00 2.20741928e-01 4.82490867e-01 2.08300903e-01
3.35176975e-01 8.22079957e-01 1.01414430e+00 -2.18890369e-01
4.23175693e-02 -2.22150862e-01 4.60489213e-01 8.63188922e-01
4.77159798e-01 1.79364055e-01 -3.89662199e-02 5.98197401e-01
8.56695950e-01 -9.96352658e-02 4.22176979e-02 4.11308974e-01
-4.78735894e-01 7.17268586e-01 9.96179938e-01 -7.20000789e-02
-2.64641196e-01 2.82604396e-01 3.35124135e-01 -1.30047863e-02
4.40429688e-01 -6.42897934e-02 -2.19077095e-01 4.57059629e-02
-8.97111833e-01 3.16317305e-02 2.00080052e-01 8.26692343e-01
1.06589067e+00 1.98843926e-01 7.76259601e-02 1.03323030e+00
3.22021782e-01 4.97489810e-01 -1.25862285e-01 -9.18913662e-01
5.01737237e-01 7.38561809e-01 4.72711846e-02 -1.22317779e+00
-5.95749497e-01 -1.34326205e-01 -5.68629205e-01 4.41484869e-01
1.21230379e-01 -2.89478034e-01 -1.44098997e+00 5.56227505e-01
3.29661548e-01 3.66366446e-01 -5.55086732e-02 7.14368522e-01
1.16217470e+00 8.28545153e-01 9.98472720e-02 5.03970981e-01
1.11179161e+00 -1.00040126e+00 -4.94991839e-01 -3.63845706e-01
8.00405622e-01 -6.92637861e-01 1.16231382e+00 1.09898105e-01
-6.47559702e-01 -5.87436140e-01 -1.21379113e+00 -1.73766166e-01
-6.40419364e-01 4.76123750e-01 4.71142709e-01 9.59038973e-01
-9.30070817e-01 3.08161557e-01 -4.81599033e-01 -6.11065805e-01
9.81986940e-01 8.25127289e-02 1.07728355e-02 -5.21522641e-01
-1.10720599e+00 1.32975030e+00 2.21400023e-01 6.04086936e-01
-1.18315303e+00 -5.57481349e-01 -7.16053784e-01 -1.43747225e-01
1.58995599e-01 -2.44815573e-01 9.68316972e-01 9.32658836e-03
-8.89907062e-01 9.87237871e-01 2.84462348e-02 -5.45672715e-01
4.14385170e-01 -5.85507452e-01 -5.00957787e-01 2.85657376e-01
3.62240672e-01 6.04592264e-01 4.81044263e-01 -1.34557712e+00
-1.03941488e+00 -2.35252932e-01 6.31744647e-03 8.02245289e-02
6.57509416e-02 3.00358050e-02 -5.51431105e-02 -1.12534299e-01
2.90189236e-01 -5.64664304e-01 -2.49816865e-01 -1.42955065e-01
-7.60019362e-01 -2.82264084e-01 1.48229110e+00 -8.98261726e-01
1.13102090e+00 -1.63111830e+00 -1.01883340e+00 1.68423802e-01
8.23496729e-02 6.60631955e-01 2.17159584e-04 4.94345099e-01
3.04057822e-02 5.36003649e-01 -5.39617121e-01 1.07186444e-01
-5.78027725e-01 2.10980758e-01 -7.67406598e-02 4.60835785e-01
3.89192700e-01 7.98332632e-01 -6.09078228e-01 -6.20040298e-01
6.00018501e-01 4.04218286e-01 -5.14291413e-02 4.37153457e-03
5.19767463e-01 -6.99392334e-02 -5.43287039e-01 1.22614038e+00
1.23271942e+00 4.04121280e-01 -9.34066236e-01 -2.07562789e-01
-5.02461195e-01 1.37996823e-01 -1.02775562e+00 7.57986963e-01
-4.13799018e-01 1.12355042e+00 -3.23525280e-01 -1.03473818e+00
1.31174719e+00 3.61068584e-02 2.42655918e-01 -8.92414153e-01
1.62790582e-01 2.31892109e-01 -2.16546774e-01 -1.09301400e+00
7.37904370e-01 1.39032930e-01 1.97501689e-01 1.07025221e-01
-5.74546576e-01 -3.18813503e-01 5.33183031e-02 5.36332689e-02
1.19155169e+00 -2.94665503e-03 -3.66166443e-01 7.34236315e-02
2.82952577e-01 4.49251801e-01 3.27768773e-01 1.00305879e+00
-4.24325675e-01 8.80456805e-01 1.30817264e-01 -6.09709740e-01
-1.11258841e+00 -9.91714716e-01 -5.83796322e-01 6.97149277e-01
5.29883981e-01 5.03546774e-01 -1.02668476e+00 -3.14932376e-01
1.29989013e-01 4.80187982e-01 -6.54618323e-01 -2.46585920e-01
-6.04853928e-01 -9.79656935e-01 8.81912053e-01 9.53067660e-01
1.24613929e+00 -1.27588928e+00 -7.40477502e-01 1.76412731e-01
-8.97001550e-02 -1.21207094e+00 1.27397373e-01 -4.34841216e-01
-9.75000262e-01 -1.62964201e+00 -6.77162290e-01 -1.11219585e+00
7.57605374e-01 8.72373521e-01 7.67681658e-01 4.86275405e-01
-9.63697433e-01 1.58475846e-01 -3.24758470e-01 -5.59460759e-01
-1.29585072e-01 8.40454027e-02 -5.52877426e-01 -2.53845572e-01
1.95888832e-01 -5.38970232e-02 -9.06832457e-01 7.62283087e-01
-4.84056741e-01 -1.22141307e-02 7.41141200e-01 4.11052674e-01
2.94742584e-01 4.37839627e-01 1.05695283e+00 -7.71419525e-01
5.87392092e-01 -5.22597075e-01 -3.67360234e-01 -3.57507914e-02
-6.33160591e-01 -7.87063062e-01 -2.00804442e-01 1.88939363e-01
-1.32326961e+00 -1.27130806e-01 -4.56390172e-01 1.87082931e-01
-6.87989414e-01 4.74581271e-01 4.29538526e-02 -3.46467227e-01
7.37433016e-01 -1.19982762e-02 -3.39647472e-01 -3.76951873e-01
1.15954466e-01 1.00276268e+00 7.26425707e-01 -2.23765209e-01
7.85270572e-01 6.34412766e-01 3.01258396e-02 -1.40035319e+00
-9.62061465e-01 -8.22070181e-01 -6.62624240e-01 -7.78572977e-01
7.82879710e-01 -8.44300270e-01 -4.60687816e-01 1.13224411e+00
-9.66521919e-01 -2.97643870e-01 1.02676138e-01 7.89624229e-02
-1.55230269e-01 3.74670327e-01 -5.01111627e-01 -1.18766773e+00
-6.67997003e-01 -7.57342756e-01 9.91064727e-01 5.42560935e-01
3.87488306e-01 -5.59200943e-01 -1.10761069e-01 1.04981661e+00
5.10049999e-01 5.03307402e-01 5.43983340e-01 1.78567454e-01
-7.03470051e-01 -8.11195552e-01 -1.13047028e+00 5.40859878e-01
-9.53843445e-02 3.22277993e-01 -1.18085480e+00 3.18261981e-01
-4.96640593e-01 -8.92604664e-02 1.38279641e+00 6.15652978e-01
1.10943913e+00 1.99990183e-01 -6.45497382e-01 2.20004725e-03
1.42121375e+00 4.78179663e-01 1.44051695e+00 4.87630546e-01
1.05730081e+00 7.22902536e-01 1.04935789e+00 5.82426302e-02
6.55669570e-01 9.33638588e-02 1.01398218e+00 -4.77700889e-01
-4.01616126e-01 -6.57482371e-02 2.28836928e-02 8.22189525e-02
-4.74056542e-01 -1.98209569e-01 -1.44042361e+00 1.08448231e+00
-1.57538259e+00 -1.09215605e+00 -1.14787173e+00 1.93830132e+00
1.73386097e-01 3.00332427e-01 -2.73816511e-02 7.00781405e-01
1.21419168e+00 -1.50330976e-01 -5.16128957e-01 -4.17139411e-01
1.98197253e-02 1.42626852e-01 8.74187350e-01 3.10503691e-01
-1.30737174e+00 1.17428899e+00 6.35131645e+00 6.97307825e-01
-1.02885008e+00 -8.59373063e-02 6.29936934e-01 5.98668694e-01
2.51396507e-01 -1.78062506e-02 -8.97840738e-01 1.19787537e-01
6.99550748e-01 4.58312333e-01 -2.53484339e-01 8.23872805e-01
7.30147123e-01 -6.90227211e-01 -1.45753682e-01 4.43412811e-01
-2.59431988e-01 -1.16073418e+00 -3.77495348e-01 -7.14093745e-02
8.61084521e-01 4.02378917e-01 -1.38130158e-01 2.73678213e-01
1.74915344e-01 -1.13284981e+00 3.38619173e-01 4.42358077e-01
5.30609131e-01 -8.96303773e-01 1.05827582e+00 2.51100212e-01
-1.31319785e+00 -3.06193680e-01 -6.30849600e-01 -3.40684623e-01
1.87677667e-01 7.59315729e-01 -1.41174567e+00 2.10464820e-01
1.11512959e+00 1.08858287e+00 -5.78059852e-01 1.66263855e+00
-4.49778348e-01 1.12021995e+00 -4.51068223e-01 1.75098911e-01
3.73826802e-01 3.57190706e-02 2.61193961e-01 1.30477488e+00
3.02674294e-01 -9.76473838e-02 6.85532019e-02 6.06228650e-01
-2.61526406e-02 -2.93860724e-03 -6.55602276e-01 5.76729119e-01
6.73731208e-01 1.12814021e+00 -9.29786563e-01 -3.15336376e-01
-2.69240946e-01 2.97659248e-01 -4.66703549e-02 3.06351721e-01
-9.18208897e-01 -8.43139112e-01 4.48250711e-01 6.76809967e-01
1.82358891e-01 -2.09470719e-01 -9.21465695e-01 -3.59156162e-01
-1.15879618e-01 2.27029487e-01 1.14181526e-01 -1.06581986e+00
-8.25932443e-01 -4.29895101e-03 -3.94627079e-02 -9.41754520e-01
9.40825164e-01 -5.10207176e-01 -1.21025479e+00 6.46213114e-01
-1.97709656e+00 -1.30049789e+00 -9.33849633e-01 2.27829516e-01
8.38511646e-01 3.00495952e-01 1.44375190e-01 6.22170508e-01
-1.23053360e+00 2.69684464e-01 -9.35531482e-02 3.09180140e-01
2.19989300e-01 -1.02114618e+00 6.96744025e-01 1.07224965e+00
-5.97828150e-01 -1.12908958e-02 4.30289596e-01 -8.15664291e-01
-8.69187176e-01 -1.52294040e+00 7.03995466e-01 -1.45260751e-01
3.07970047e-01 2.89577156e-01 -1.14180744e+00 2.21485108e-01
-1.55511633e-01 -6.64309561e-02 -8.06501359e-02 -2.48690337e-01
1.06177896e-01 -6.01736270e-02 -1.53579664e+00 2.64093608e-01
9.32183206e-01 -2.87035495e-01 -2.82413602e-01 3.05552036e-01
3.90703768e-01 -3.02720129e-01 -7.48232305e-01 6.23567820e-01
5.98370135e-01 -9.19070065e-01 1.12975359e+00 1.34121686e-01
5.61420619e-01 -4.06010896e-01 1.52249649e-01 -7.20892072e-01
-1.94023049e-03 9.42734852e-02 2.22006485e-01 1.27492166e+00
4.55218941e-01 -7.24225640e-01 1.08027315e+00 4.89529699e-01
-1.01203144e+00 -8.42178702e-01 -8.76786053e-01 -5.72034061e-01
2.10415758e-02 -8.02905262e-01 3.48869056e-01 6.10553205e-01
-7.41292596e-01 2.81710569e-02 -2.08417758e-01 5.68230391e-01
5.99541008e-01 -2.12916628e-01 7.12996125e-01 -1.35234702e+00
1.03765821e+00 -4.25879300e-01 -6.47827387e-01 -7.73312986e-01
-3.95204425e-01 -4.39839989e-01 3.63032252e-01 -2.24815822e+00
-6.31890744e-02 -7.16068625e-01 -9.93839949e-02 8.22918594e-01
-2.07565770e-01 6.61188304e-01 -4.76274133e-01 1.69350758e-01
-4.00909968e-02 4.26031113e-01 1.53966534e+00 -2.57533133e-01
-6.30609095e-02 2.74892539e-01 -3.75265002e-01 8.48221838e-01
1.30511427e+00 -4.78172749e-01 -2.98675328e-01 -3.32195103e-01
-3.72995809e-02 -1.83096245e-01 7.23219037e-01 -1.41286063e+00
8.91501382e-02 -7.21895695e-02 1.11999646e-01 -1.29129207e+00
1.57483906e-01 -4.76749986e-01 -3.84104282e-01 5.48103690e-01
1.38538942e-01 -4.13000166e-01 2.97640800e-01 5.86907566e-01
-3.50847878e-02 -3.28357279e-01 1.11544812e+00 3.99146900e-02
-1.30353975e+00 1.67941049e-01 -7.10319519e-01 -8.77264962e-02
1.26862681e+00 -1.04535198e+00 -5.25949776e-01 8.07304308e-02
-3.97953659e-01 5.08396566e-01 6.72275573e-02 5.50542414e-01
1.35348940e+00 -7.84454703e-01 -7.78026521e-01 -4.01395336e-02
1.50102958e-01 2.80515045e-01 4.76844221e-01 6.84128821e-01
-1.14634478e+00 2.37998933e-01 -3.31575722e-01 -5.87658286e-01
-1.27056146e+00 -9.60781574e-02 3.77163142e-01 2.20056608e-01
-8.66632402e-01 7.82234848e-01 -2.04734460e-01 -2.13217899e-01
-3.22436094e-02 -2.80237108e-01 -6.17229104e-01 1.12802014e-01
2.26660281e-01 1.34218049e+00 1.48089275e-01 -7.88662314e-01
-2.61959463e-01 9.89452124e-01 -7.54829273e-02 4.15494353e-01
1.22549653e+00 -4.06753033e-01 1.41662195e-01 -9.36788246e-02
9.54492986e-01 -9.72502410e-01 -1.14519572e+00 1.44700900e-01
3.18239480e-02 -4.87334907e-01 6.57076955e-01 -7.97333658e-01
-1.34161687e+00 1.24650991e+00 1.06532502e+00 2.01377392e-01
9.02042925e-01 -9.75686088e-02 1.30114567e+00 4.68365103e-01
3.64259720e-01 -1.32649302e+00 4.64475565e-02 4.79886681e-01
9.00629699e-01 -1.58750916e+00 1.29246106e-02 -6.69123471e-01
-3.23274791e-01 9.62571442e-01 9.25845861e-01 -4.61599827e-01
6.31349564e-01 6.84419945e-02 5.82649149e-02 -7.30086923e-01
5.55263273e-02 -5.42966723e-01 -5.38869109e-03 8.82003009e-01
1.13048300e-01 2.16004178e-01 -1.38778687e-01 -2.04792842e-01
1.31467745e-01 4.22743298e-02 8.15206528e-01 1.19780552e+00
-1.38482487e+00 -4.61287051e-01 -5.99419773e-01 9.61992919e-01
-3.29828531e-01 5.83026409e-02 -2.00263500e-01 9.11883712e-01
2.17679515e-01 1.25131834e+00 -1.06100209e-01 -4.32534337e-01
6.01212561e-01 -6.03861809e-01 2.22625062e-02 -8.00826490e-01
-1.67809755e-01 -4.80869263e-01 4.15675044e-01 -2.40100563e-01
-3.32983822e-01 -4.70643878e-01 -1.54630101e+00 -2.14354515e-01
-7.96824872e-01 -3.62878710e-01 8.51764143e-01 8.76641512e-01
1.53047964e-01 5.19411623e-01 8.14885259e-01 -9.91459489e-01
2.66551048e-01 -1.02879918e+00 -7.59331405e-01 -2.11946562e-01
2.67400622e-01 -9.97056901e-01 -2.81522751e-01 -1.02370642e-02] | [7.424946308135986, 1.1541322469711304] |
3d38d645-a9c5-414c-9e3e-a79b0f733f60 | learning-monocular-dense-depth-from-events | 2010.08350 | null | https://arxiv.org/abs/2010.08350v2 | https://arxiv.org/pdf/2010.08350v2.pdf | Learning Monocular Dense Depth from Events | Event cameras are novel sensors that output brightness changes in the form of a stream of asynchronous events instead of intensity frames. Compared to conventional image sensors, they offer significant advantages: high temporal resolution, high dynamic range, no motion blur, and much lower bandwidth. Recently, learning-based approaches have been applied to event-based data, thus unlocking their potential and making significant progress in a variety of tasks, such as monocular depth prediction. Most existing approaches use standard feed-forward architectures to generate network predictions, which do not leverage the temporal consistency presents in the event stream. We propose a recurrent architecture to solve this task and show significant improvement over standard feed-forward methods. In particular, our method generates dense depth predictions using a monocular setup, which has not been shown previously. We pretrain our model using a new dataset containing events and depth maps recorded in the CARLA simulator. We test our method on the Multi Vehicle Stereo Event Camera Dataset (MVSEC). Quantitative experiments show up to 50% improvement in average depth error with respect to previous event-based methods. | ['Davide Scaramuzza', 'Daniel Gehrig', 'Javier Hidalgo-Carrió'] | 2020-10-16 | null | null | null | null | ['probabilistic-deep-learning'] | ['computer-vision'] | [ 4.01529104e-01 -6.30531684e-02 1.07105095e-02 -5.95394909e-01
-5.15836418e-01 -2.51407355e-01 8.94708216e-01 3.76667641e-02
-5.38336039e-01 5.86440384e-01 2.97202528e-01 -1.17156498e-01
2.09264711e-01 -7.18202412e-01 -9.74961460e-01 -5.55337489e-01
-1.68638572e-01 6.56401664e-02 6.89482152e-01 3.75069641e-02
1.55166909e-01 5.58380902e-01 -1.88497877e+00 4.53935176e-01
1.62064403e-01 1.19352853e+00 1.91199213e-01 9.77489889e-01
1.99507207e-01 1.39719033e+00 -4.04077113e-01 6.06368594e-02
3.34244460e-01 -3.26589286e-01 -4.17071134e-01 -4.71414700e-02
6.29315913e-01 -8.95959973e-01 -1.00846922e+00 5.99443972e-01
4.57530171e-01 7.86032826e-02 1.58781528e-01 -1.43055618e+00
1.37422293e-01 8.85690153e-02 -4.35506016e-01 4.44764793e-01
5.57547748e-01 2.50817627e-01 5.68501711e-01 -6.89282537e-01
8.18743587e-01 1.11595559e+00 6.99095726e-01 5.48502088e-01
-1.11587155e+00 -7.57336497e-01 2.07614526e-01 6.37828648e-01
-8.98481369e-01 -7.34179795e-01 5.31126559e-01 -3.59222382e-01
1.47555554e+00 -3.56417894e-02 8.59448671e-01 1.37149954e+00
5.60231805e-01 9.44592357e-01 9.25423563e-01 -5.26361763e-02
4.75754410e-01 -2.11610928e-01 -4.36120987e-01 3.72641504e-01
-1.42116517e-01 6.14237070e-01 -1.01913035e+00 2.37639204e-01
8.72210383e-01 4.04083610e-01 -4.33758855e-01 -3.81864011e-01
-1.38961661e+00 6.80312574e-01 3.98746103e-01 -3.05257589e-01
-5.28217018e-01 7.06161559e-01 3.68013084e-01 4.32262748e-01
4.20163721e-01 -5.31609245e-02 -3.39431405e-01 -3.66275728e-01
-9.84818816e-01 2.39572540e-01 6.67452276e-01 9.12560999e-01
8.27437878e-01 1.56755716e-01 4.23668995e-02 1.11681029e-01
3.92784387e-01 6.16972089e-01 6.23389959e-01 -1.30958569e+00
5.25749743e-01 3.14179748e-01 2.15823710e-01 -8.65781486e-01
-5.29347062e-01 1.64266139e-01 -7.39883184e-01 7.09543049e-01
2.30308235e-01 -4.45783019e-01 -9.54987466e-01 1.63331163e+00
1.85503677e-01 8.37976456e-01 2.04365253e-01 1.02384007e+00
5.62001824e-01 1.01048386e+00 -2.55590767e-01 -2.67395437e-01
7.22064853e-01 -6.26089990e-01 -6.95020258e-01 -4.17818964e-01
3.26067418e-01 -5.84879279e-01 4.80367154e-01 6.18737221e-01
-1.02851439e+00 -5.04864097e-01 -1.18255377e+00 1.07051849e-01
-1.73057631e-01 -3.21956336e-01 5.71269095e-01 3.41180742e-01
-1.22310686e+00 5.98635316e-01 -1.38303471e+00 -5.01011550e-01
2.35912725e-01 3.36465120e-01 -3.13311934e-01 3.05228028e-03
-9.32438791e-01 6.90566361e-01 2.24435970e-01 -2.82646250e-02
-1.20425677e+00 -7.82578230e-01 -1.10634637e+00 -1.50561586e-01
2.88277656e-01 -5.81611335e-01 1.68265784e+00 -9.78432298e-01
-1.95845270e+00 4.20119405e-01 -4.07540858e-01 -1.06864071e+00
6.99560344e-01 -4.68727976e-01 -2.59656698e-01 3.40038836e-01
-2.59477615e-01 1.09501302e+00 7.62015998e-01 -8.84269655e-01
-1.18945062e+00 7.50928745e-02 1.95329800e-01 2.59176075e-01
-1.47570089e-01 -1.54806823e-01 -2.83614010e-01 -2.97726065e-01
5.68239354e-02 -9.58423615e-01 -3.70777220e-01 3.39828938e-01
8.45661461e-02 2.13179961e-01 1.11502182e+00 -1.06681108e-01
7.81560421e-01 -2.13175941e+00 -9.27267373e-02 -1.28809348e-01
1.50769114e-01 1.11918077e-02 6.36959672e-02 5.35050511e-01
-8.50354433e-02 -4.53856587e-01 5.69977164e-02 -5.34755826e-01
-1.99994564e-01 3.91972452e-01 -6.63366616e-01 5.12111902e-01
3.17338645e-01 7.88898110e-01 -9.73493159e-01 -3.80774796e-01
9.10444856e-01 6.32716715e-01 -5.52230716e-01 4.09463137e-01
-4.17387456e-01 4.53133523e-01 -8.82604066e-03 1.72882572e-01
3.00379753e-01 -4.30857122e-01 3.21204052e-03 -1.62721679e-01
-3.50911885e-01 3.13515931e-01 -1.19798160e+00 1.83029985e+00
-5.51969051e-01 1.24108052e+00 -3.67350399e-01 -6.25266254e-01
6.78730428e-01 4.52683091e-01 6.58798456e-01 -1.07940400e+00
5.43573909e-02 -1.31945238e-01 -3.47407579e-01 -4.91746992e-01
8.53341043e-01 -4.93109263e-02 1.31053701e-01 2.19224632e-01
-8.06222558e-02 -6.26801848e-02 4.61645983e-03 2.68296510e-01
1.49332547e+00 3.91669780e-01 1.20794252e-01 3.33629280e-01
3.62152606e-02 -3.13269645e-02 5.74111402e-01 6.58163190e-01
-1.16374671e-01 9.75296021e-01 2.44423464e-01 -5.25996685e-01
-1.06148899e+00 -1.02443564e+00 6.09775707e-02 6.06239080e-01
5.89927912e-01 -4.87856328e-01 -3.07722211e-01 -1.85263380e-01
-2.60098815e-01 4.18407798e-01 -4.97747749e-01 -6.63648918e-02
-6.74807608e-01 -5.72534442e-01 3.46505076e-01 7.84688711e-01
5.72373569e-01 -8.67959678e-01 -1.44035304e+00 4.64020252e-01
-8.15958455e-02 -1.68859303e+00 -3.92221548e-02 3.34036201e-01
-9.74414706e-01 -8.97283256e-01 -4.13815647e-01 -4.10580605e-01
2.04511389e-01 3.24541837e-01 1.08410692e+00 -5.04690707e-01
-2.24880055e-01 6.06090665e-01 -1.62789539e-01 -4.32440639e-01
-2.11604968e-01 -3.61507982e-01 -1.47958174e-01 2.35906824e-01
2.73814768e-01 -5.63547373e-01 -9.11634326e-01 1.78900704e-01
-1.02686489e+00 5.37073076e-01 4.40280646e-01 6.87282085e-01
5.73087573e-01 -2.23120943e-01 3.09883147e-01 -6.87318146e-01
-1.49888247e-01 -5.17154872e-01 -1.02720177e+00 -3.88245940e-01
-4.86548930e-01 -8.71670917e-02 6.43203974e-01 -3.46998602e-01
-1.32147217e+00 4.76079196e-01 4.37928643e-03 -7.43263721e-01
-2.52439976e-01 3.40108573e-01 3.36906999e-01 5.40873073e-02
6.49799347e-01 -1.45678669e-02 1.62166804e-02 9.90643576e-02
2.52963901e-02 2.98964053e-01 7.67544150e-01 9.06978622e-02
5.32760143e-01 1.04459035e+00 1.76950499e-01 -7.64536142e-01
-4.17524159e-01 -5.36640227e-01 -2.68177330e-01 -4.65931296e-01
7.70518005e-01 -1.45346558e+00 -7.68762290e-01 6.58956945e-01
-1.23626375e+00 -7.90800869e-01 -4.22748506e-01 7.64025986e-01
-8.20225239e-01 1.76075250e-01 -8.84551287e-01 -6.54547989e-01
5.81006780e-02 -9.08522725e-01 1.23097193e+00 3.87339354e-01
-1.58343226e-01 -8.60771894e-01 1.81521356e-01 -2.08523303e-01
5.01934528e-01 5.18847704e-01 2.33815446e-01 1.19012324e-02
-1.15719008e+00 -1.38672411e-01 -2.22768500e-01 2.85688974e-02
-9.69819799e-02 -3.88381295e-02 -1.27796674e+00 -3.51954460e-01
-1.30668283e-02 -3.13885540e-01 8.81003857e-01 4.60001022e-01
1.04044092e+00 2.78417324e-03 -4.73303884e-01 8.36335361e-01
1.58836842e+00 2.29455784e-01 8.73970091e-01 3.84918958e-01
5.14673412e-01 2.53509194e-01 6.33821964e-01 8.88063312e-01
4.51914370e-01 7.16445923e-01 8.59085441e-01 -1.10405430e-01
-2.96881407e-01 -2.93002963e-01 6.97422266e-01 2.89631069e-01
2.27489665e-01 -5.31065643e-01 -9.06711578e-01 7.00929403e-01
-2.20840478e+00 -1.20522404e+00 -1.16878659e-01 2.16070914e+00
5.48195422e-01 3.73134702e-01 -5.33616133e-02 1.46801755e-01
3.88828576e-01 3.06834310e-01 -7.39588976e-01 -1.70632482e-01
-1.49684533e-01 1.06043272e-01 8.03152800e-01 3.91778052e-01
-1.04875243e+00 7.46779501e-01 6.79561281e+00 1.21997766e-01
-1.58993101e+00 -1.69631585e-01 4.71399039e-01 -6.64866328e-01
-6.48728311e-02 -7.02546984e-02 -8.77860844e-01 4.44900990e-01
1.50647402e+00 -1.60420120e-01 1.15389265e-01 5.86583614e-01
6.07351482e-01 -5.28536260e-01 -1.48189449e+00 1.27203310e+00
1.48704290e-01 -1.55173218e+00 -2.07857624e-01 -7.71361291e-02
8.05903494e-01 5.99683881e-01 -1.43875554e-01 4.54907492e-02
4.66676056e-01 -1.03489900e+00 8.43135357e-01 5.24357200e-01
6.88168049e-01 -4.58816737e-01 7.29925871e-01 1.66769505e-01
-1.10543227e+00 -7.52813146e-02 -1.59433380e-01 -4.98975754e-01
6.53038621e-01 4.70609754e-01 -9.36816752e-01 2.49548852e-01
1.04971254e+00 1.23659122e+00 -3.75162333e-01 1.23191249e+00
-8.84089097e-02 7.61104345e-01 -6.87144816e-01 2.39436463e-01
1.66143224e-01 3.28846544e-01 3.94925565e-01 1.16444004e+00
3.13866973e-01 -4.69518155e-02 1.13404103e-01 5.15986621e-01
7.85445124e-02 -5.09507596e-01 -8.46146047e-01 3.87302995e-01
3.43242943e-01 9.54626262e-01 -4.68579620e-01 -4.40668315e-01
-9.06966269e-01 1.06841660e+00 -1.22111171e-01 4.24306512e-01
-1.04204094e+00 -1.77189663e-01 7.47861266e-01 1.43579334e-01
4.73950624e-01 -3.30172956e-01 -8.64395052e-02 -1.04563236e+00
4.38800193e-02 -5.70659876e-01 1.78278729e-01 -1.03134286e+00
-9.01388943e-01 5.12573183e-01 1.18528657e-01 -1.57186139e+00
-9.07405138e-01 -5.06470025e-01 -3.39901149e-01 4.42552328e-01
-1.87769771e+00 -7.23349392e-01 -8.03053916e-01 7.29742587e-01
7.49758780e-01 3.48244272e-02 4.98686910e-01 4.70735878e-01
-2.40053162e-01 1.71775389e-02 -2.47006547e-02 -1.09908797e-01
8.61963272e-01 -1.19884133e+00 5.65413356e-01 1.05904675e+00
4.48995195e-02 -1.95705146e-01 8.44356298e-01 -4.03246731e-01
-1.63299704e+00 -1.22520638e+00 6.63736463e-01 -4.11251396e-01
6.16583884e-01 -3.16893548e-01 -6.33953214e-01 9.44670379e-01
3.79942596e-01 3.39361519e-01 1.59365952e-01 -4.63939905e-01
-7.16199055e-02 -3.57769966e-01 -6.89513505e-01 5.72922826e-01
9.32758927e-01 -5.33380151e-01 -2.83472359e-01 -7.73399370e-04
6.75546944e-01 -8.71396840e-01 -4.93117839e-01 4.99231875e-01
5.08187890e-01 -1.45323443e+00 8.04405272e-01 -9.15760323e-02
7.49897599e-01 -4.43904370e-01 -1.39652625e-01 -1.04929113e+00
7.01268390e-02 -7.24827588e-01 -3.52768421e-01 7.50676692e-01
2.20064744e-01 -5.29472172e-01 1.08993113e+00 4.85089749e-01
-3.06589901e-01 -4.39060569e-01 -9.21052158e-01 -5.84158301e-01
-3.42413694e-01 -7.41616845e-01 2.91059703e-01 4.27032411e-01
-7.94542655e-02 2.84526080e-01 -5.92580140e-01 3.15601587e-01
5.26034534e-01 1.55616209e-01 9.78195488e-01 -9.45207715e-01
-3.69003177e-01 -5.96141107e-02 -8.93993616e-01 -1.46667290e+00
-3.08335777e-02 -3.86721462e-01 3.90116125e-01 -1.45125854e+00
-1.26339877e-02 -6.66031539e-02 -3.90872881e-02 4.73603934e-01
3.07640254e-01 3.48574489e-01 6.61740154e-02 3.22241103e-03
-8.43121111e-01 5.10435641e-01 6.02155149e-01 1.57235339e-01
-8.11684504e-02 -3.05956900e-01 7.22377673e-02 7.53546596e-01
6.68371737e-01 -2.53894925e-01 -5.67159712e-01 -7.01491237e-01
4.43442494e-01 4.95285243e-01 8.30311418e-01 -1.52904177e+00
7.27691889e-01 -2.00064361e-01 5.80033243e-01 -6.91510201e-01
7.33339489e-01 -1.01933122e+00 4.56063300e-01 3.51689786e-01
-2.21486881e-01 3.66475344e-01 4.31472927e-01 8.64801288e-01
-5.45177341e-01 2.77733177e-01 7.31503487e-01 8.00570250e-02
-1.36540294e+00 3.30129623e-01 -7.13454664e-01 -2.43735507e-01
1.09149933e+00 -4.20586646e-01 -2.74050385e-01 -7.59908617e-01
-3.45805615e-01 8.55376720e-02 5.92283666e-01 6.03379250e-01
8.60532761e-01 -1.21664453e+00 -5.17022550e-01 3.43746305e-01
1.90231040e-01 1.62338272e-01 2.25267276e-01 7.12086022e-01
-6.27812803e-01 2.92216986e-01 -2.66714513e-01 -1.21891665e+00
-1.06293404e+00 2.07303524e-01 4.06719953e-01 3.67555842e-02
-1.10341144e+00 4.97866005e-01 4.17770803e-01 2.60569006e-01
3.06654632e-01 -5.31880438e-01 1.10512406e-01 -9.24525708e-02
6.02272332e-01 2.39295051e-01 9.04424042e-02 -3.87558401e-01
-1.71228558e-01 4.40400273e-01 1.66463349e-02 -3.26061070e-01
1.38652265e+00 -4.21834737e-01 5.25932789e-01 7.46449530e-01
1.17319131e+00 -3.74831975e-01 -2.10801840e+00 -2.06967726e-01
-6.17317185e-02 -4.85524505e-01 2.50605255e-01 -5.38504660e-01
-1.15742087e+00 8.34220290e-01 1.00452805e+00 -5.48551939e-02
1.22645712e+00 -2.64743865e-01 9.72064435e-01 3.84496391e-01
5.30941367e-01 -9.18256879e-01 2.00991929e-01 6.01387203e-01
4.97724503e-01 -1.54410267e+00 -2.27771983e-01 -1.74044192e-01
-3.25355053e-01 1.14954066e+00 6.15764022e-01 -2.63726294e-01
4.31395799e-01 8.48949254e-01 1.62859976e-01 3.30401435e-02
-1.41082287e+00 -1.87600702e-01 -3.05908442e-01 4.93709713e-01
2.55504072e-01 -3.81764621e-01 3.50381374e-01 -9.87567976e-02
-2.60449983e-02 5.31466782e-01 9.10701394e-01 1.12689674e+00
-2.81419933e-01 -6.83762491e-01 -1.90478906e-01 3.00772190e-01
-3.03549081e-01 5.14379144e-02 1.71511129e-01 6.87663853e-01
-2.83601552e-01 1.07209218e+00 4.67387199e-01 -3.94270152e-01
4.21267927e-01 -2.08317623e-01 4.06968355e-01 -2.97351092e-01
-3.55248481e-01 -1.32211730e-01 1.18954489e-02 -1.20632470e+00
-6.31904781e-01 -9.37329710e-01 -1.40761113e+00 -4.60778147e-01
9.76861045e-02 -3.75372201e-01 6.95357919e-01 8.56210232e-01
5.35054326e-01 5.80919921e-01 4.87273127e-01 -1.20757139e+00
-1.35890702e-02 -5.66962481e-01 -5.94969653e-03 5.05296707e-01
7.46674359e-01 -3.37353200e-01 -3.72821629e-01 4.44454312e-01] | [8.555281639099121, -1.109528660774231] |
d37057f7-1c11-466d-b849-2418413ab63e | lapar-linearly-assembled-pixel-adaptive-1 | 2105.10422 | null | https://arxiv.org/abs/2105.10422v1 | https://arxiv.org/pdf/2105.10422v1.pdf | LAPAR: Linearly-Assembled Pixel-Adaptive Regression Network for Single Image Super-Resolution and Beyond | Single image super-resolution (SISR) deals with a fundamental problem of upsampling a low-resolution (LR) image to its high-resolution (HR) version. Last few years have witnessed impressive progress propelled by deep learning methods. However, one critical challenge faced by existing methods is to strike a sweet spot of deep model complexity and resulting SISR quality. This paper addresses this pain point by proposing a linearly-assembled pixel-adaptive regression network (LAPAR), which casts the direct LR to HR mapping learning into a linear coefficient regression task over a dictionary of multiple predefined filter bases. Such a parametric representation renders our model highly lightweight and easy to optimize while achieving state-of-the-art results on SISR benchmarks. Moreover, based on the same idea, LAPAR is extended to tackle other restoration tasks, e.g., image denoising and JPEG image deblocking, and again, yields strong performance. The code is available at https://github.com/dvlab-research/Simple-SR. | ['Jiaya Jia', 'Jiangbo Lu', 'Nianjuan Jiang', 'Lu Qi', 'Kun Zhou', 'Wenbo Li'] | 2021-05-21 | lapar-linearly-assembled-pixel-adaptive | http://proceedings.neurips.cc/paper/2020/hash/eaae339c4d89fc102edd9dbdb6a28915-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/eaae339c4d89fc102edd9dbdb6a28915-Paper.pdf | neurips-2020-12 | ['image-deblocking'] | ['computer-vision'] | [ 4.11721975e-01 -2.37951621e-01 -1.81346402e-01 -1.65517539e-01
-1.30837810e+00 -1.18827865e-01 3.80710006e-01 -4.69340861e-01
-1.42353818e-01 6.86598241e-01 4.19912755e-01 -1.05921574e-01
-1.61988065e-02 -7.15158463e-01 -7.44167984e-01 -9.60377395e-01
1.38153449e-01 2.57502310e-02 -1.55351341e-01 -3.65916461e-01
2.42770717e-01 4.50105995e-01 -1.35828805e+00 3.16733629e-01
7.96030819e-01 8.95055175e-01 3.66294831e-01 6.73420906e-01
4.54684466e-01 1.05203080e+00 -2.49047540e-02 -2.42191061e-01
5.19302368e-01 -3.40308636e-01 -7.52763033e-01 4.97323498e-02
7.84492433e-01 -6.92468107e-01 -5.73048532e-01 1.17315841e+00
6.21385753e-01 9.14802700e-02 9.02286991e-02 -6.90135658e-01
-1.01992607e+00 5.04286945e-01 -9.27179873e-01 2.75085330e-01
1.72689676e-01 1.36314958e-01 9.53159511e-01 -1.01488864e+00
5.63945174e-01 1.36394727e+00 8.86303186e-01 4.33814645e-01
-1.67818332e+00 -5.62864900e-01 -7.12313876e-02 3.08707893e-01
-1.43463838e+00 -7.03308940e-01 6.88818812e-01 -2.35325009e-01
6.10015154e-01 1.92938596e-01 2.95654953e-01 1.13719320e+00
1.65424526e-01 4.04656529e-01 1.64114380e+00 -2.51200706e-01
6.30218908e-02 -3.01812798e-01 -9.58033577e-02 3.37254524e-01
3.22654061e-02 1.90008730e-01 -6.59856617e-01 1.90532714e-01
1.28205657e+00 -1.01029791e-01 -6.25743687e-01 -2.00784966e-01
-1.19156742e+00 7.47676373e-01 4.64276195e-01 2.92934716e-01
-3.41693342e-01 1.25315517e-01 3.11594963e-01 5.28381765e-01
6.86644495e-01 3.02670121e-01 -4.34161305e-01 1.22244865e-01
-1.20006096e+00 3.09499949e-01 3.68953377e-01 4.20006394e-01
7.51780450e-01 2.55374014e-01 -4.09799106e-02 1.10692430e+00
-3.18573378e-02 3.69336516e-01 2.33756557e-01 -1.51876891e+00
2.33812615e-01 -4.12199311e-02 1.97252423e-01 -1.00883389e+00
-1.30301356e-01 -7.82464802e-01 -1.51103473e+00 4.31877911e-01
2.26768792e-01 2.82955348e-01 -7.14404106e-01 1.65799737e+00
2.63923794e-01 2.99765408e-01 -2.78326906e-02 1.32082033e+00
6.78578079e-01 8.40252042e-01 -2.46652126e-01 -2.55585939e-01
1.34920192e+00 -1.05641913e+00 -5.91179013e-01 -2.31622756e-01
-6.68793693e-02 -7.77361989e-01 1.12805700e+00 6.33676887e-01
-1.28903365e+00 -7.84477890e-01 -9.72482264e-01 -7.26392746e-01
1.95307806e-01 1.26620159e-01 4.32374328e-01 2.69720763e-01
-1.46354401e+00 8.77249181e-01 -6.94301426e-01 -5.32529466e-02
5.45314491e-01 1.50914878e-01 -4.05843168e-01 -5.18143177e-01
-1.12493920e+00 8.07511806e-01 -5.10351406e-03 1.86730683e-01
-1.04236209e+00 -1.00759614e+00 -6.89381838e-01 -6.96261451e-02
3.91048253e-01 -8.84316683e-01 9.89224732e-01 -9.46879089e-01
-1.63301814e+00 1.02217221e+00 -2.23188326e-01 -6.19497120e-01
7.46732056e-01 -4.54225630e-01 -3.26326519e-01 1.06611162e-01
7.47214779e-02 4.38625515e-01 1.29263425e+00 -1.46738565e+00
-3.42735976e-01 -4.55817342e-01 4.91082966e-02 1.42427593e-01
-1.13390274e-01 1.53984457e-01 -4.96237069e-01 -9.14342642e-01
1.09897584e-01 -6.21347189e-01 -4.17583138e-01 4.81840223e-02
-2.04542786e-01 2.50987113e-01 3.97687942e-01 -1.21246600e+00
1.16150129e+00 -2.18736601e+00 5.69248676e-01 -3.45306396e-01
4.50984061e-01 1.61766306e-01 -3.64807695e-01 3.01033318e-01
-4.16235000e-01 -8.03971514e-02 -4.96580154e-01 -4.97898459e-01
-2.69653916e-01 -6.29623905e-02 -6.61221087e-01 7.29548514e-01
7.67045543e-02 7.56782472e-01 -6.66881979e-01 -2.59884298e-01
3.19668293e-01 1.09315729e+00 -5.05868137e-01 2.11730346e-01
8.00323561e-02 9.06357229e-01 -4.37137298e-02 5.19640744e-01
9.70256984e-01 -4.65680093e-01 1.14007503e-01 -7.06457138e-01
-3.21792662e-01 1.86177686e-01 -1.27424550e+00 1.81365287e+00
-7.69151986e-01 6.10580444e-01 4.17145848e-01 -9.56153214e-01
8.86687100e-01 5.38828671e-02 5.91536522e-01 -9.70138907e-01
-1.51466370e-01 3.13646287e-01 -4.72511142e-01 -1.67560697e-01
6.40133858e-01 -1.26180753e-01 3.06609452e-01 2.51846641e-01
-2.17664912e-01 -2.39510406e-02 -9.16127414e-02 -4.49258974e-03
1.01423407e+00 3.51506174e-01 3.38471323e-01 -2.90227771e-01
6.07481539e-01 -3.29218656e-01 5.81017554e-01 6.88776374e-01
5.40850125e-02 1.08310175e+00 1.71630383e-01 -5.98292172e-01
-1.44002271e+00 -1.12145066e+00 -3.80149364e-01 9.31128263e-01
1.10390224e-01 -2.17391431e-01 -6.06943548e-01 1.49173895e-02
-1.82310089e-01 4.57861394e-01 -5.38438439e-01 6.74813539e-02
-9.59319770e-01 -8.58981133e-01 1.73674449e-01 2.38674954e-01
7.56253660e-01 -7.74287879e-01 -4.18829083e-01 1.69158608e-01
-4.59413111e-01 -1.38007236e+00 -4.16490644e-01 -1.22716069e-01
-9.18678761e-01 -7.37481236e-01 -1.06885588e+00 -6.93565845e-01
3.55258822e-01 5.55504501e-01 1.38445592e+00 8.26266780e-03
-3.49642098e-01 2.00197414e-01 -2.65892535e-01 4.77482319e-01
-4.63995397e-01 -7.48107955e-02 -1.93717495e-01 2.44505212e-01
-2.43549317e-01 -9.08734381e-01 -9.93507445e-01 2.22582936e-01
-1.10126066e+00 3.29249948e-01 6.79113626e-01 9.98226285e-01
9.57066774e-01 1.83342978e-01 4.18823838e-01 -6.31133914e-01
3.80561173e-01 -2.55436480e-01 -7.09886551e-01 8.40546861e-02
-7.62840629e-01 -3.27016413e-02 7.06467032e-01 -2.90212184e-01
-1.11746323e+00 -1.65712819e-01 -2.62030721e-01 -4.41860795e-01
4.40031383e-03 1.81918830e-01 -1.21357394e-02 -2.57767886e-01
5.18100083e-01 4.81538713e-01 2.82989498e-02 -8.35073888e-01
5.80686510e-01 5.02146363e-01 9.86082077e-01 -4.30399090e-01
1.08782804e+00 8.27214420e-01 1.26604334e-01 -8.64448786e-01
-1.08789372e+00 -2.42388442e-01 -4.49673742e-01 -2.84873787e-02
7.72246540e-01 -1.41543067e+00 -4.63399023e-01 6.84970021e-01
-8.70528579e-01 -6.95044816e-01 -3.18945199e-01 3.42465788e-02
-6.65994108e-01 7.11655259e-01 -9.36445415e-01 -5.06962001e-01
-5.46941757e-01 -1.04640460e+00 1.28080773e+00 2.53981743e-02
1.82392478e-01 -7.22665071e-01 1.01584353e-01 6.93173468e-01
7.96144724e-01 1.82476595e-01 6.56682670e-01 3.72306108e-01
-8.39253128e-01 2.07639635e-01 -5.91579139e-01 7.65661478e-01
-4.68672141e-02 -3.87284070e-01 -8.47755671e-01 -7.87869871e-01
3.77490759e-01 -3.67197484e-01 1.27442265e+00 5.76594234e-01
1.42973030e+00 -2.73075521e-01 1.98577732e-01 1.17292571e+00
1.88181984e+00 -4.47822213e-01 9.94034529e-01 5.71848273e-01
6.80284858e-01 2.24491909e-01 5.60525894e-01 4.13837105e-01
5.22828698e-01 1.08698320e+00 4.41864520e-01 -4.21953082e-01
-6.93678796e-01 -4.13639024e-02 4.45740521e-01 6.99923813e-01
-3.31732750e-01 1.52501225e-01 -6.44081712e-01 4.12663400e-01
-1.59334671e+00 -9.74687815e-01 -6.23527393e-02 2.23477530e+00
1.12131035e+00 -2.91764796e-01 -1.68791518e-01 6.85134754e-02
7.32914984e-01 6.57881737e-01 -6.08514547e-01 5.26006743e-02
-4.82850045e-01 2.90696949e-01 6.07942224e-01 7.68308640e-01
-1.16405308e+00 9.41662967e-01 5.17482519e+00 1.03768659e+00
-1.23848045e+00 3.85326505e-01 1.08111012e+00 -8.25053640e-03
-2.24627793e-01 -1.35343477e-01 -7.51222491e-01 3.86188507e-01
7.71274149e-01 7.04320446e-02 1.13741839e+00 4.84828353e-01
4.96823788e-01 -1.11036589e-02 -8.52381766e-01 1.27520978e+00
8.62289667e-02 -1.39951754e+00 -1.41340997e-02 3.02777942e-02
8.84473741e-01 2.43659034e-01 5.41315734e-01 1.35560527e-01
2.11297542e-01 -1.22463548e+00 5.51571310e-01 4.35186297e-01
1.30341685e+00 -4.80546147e-01 3.82250696e-01 3.28698009e-03
-1.11232829e+00 -2.90343404e-01 -5.60872495e-01 6.82564378e-02
2.62821943e-01 8.14693868e-01 7.08144717e-03 6.73572004e-01
1.07823241e+00 1.04333568e+00 -4.36774850e-01 7.78115690e-01
-2.03932837e-01 4.28626657e-01 2.39589531e-02 1.06886446e+00
-1.33198455e-01 -5.10941863e-01 6.80130720e-01 9.99171317e-01
3.33844513e-01 4.87654321e-02 -1.49454609e-01 8.80409598e-01
-2.99000919e-01 -1.44269615e-01 -2.42442667e-01 4.18248624e-01
2.91114122e-01 1.41515458e+00 -4.33338374e-01 -5.72036318e-02
-3.30924243e-01 1.13229609e+00 2.57320821e-01 4.82214957e-01
-8.36355984e-01 2.27693260e-01 7.89299786e-01 2.68466532e-01
4.50906843e-01 -3.06998283e-01 -3.97512525e-01 -1.52316427e+00
1.69652879e-01 -1.38129115e+00 9.22841132e-02 -8.67342710e-01
-1.28194475e+00 6.02767706e-01 -3.73488098e-01 -1.25204444e+00
1.43380448e-01 -3.12753767e-01 -2.45529160e-01 1.01505435e+00
-2.11652207e+00 -1.12812078e+00 -5.73035359e-01 6.18764460e-01
7.56476045e-01 1.88524157e-01 6.44479990e-01 6.05703831e-01
-5.86193025e-01 3.67091894e-01 4.24523622e-01 -5.28874248e-02
9.25306737e-01 -1.02254987e+00 5.40454924e-01 1.07026684e+00
-1.34156048e-02 3.68667811e-01 7.95856059e-01 -3.52295786e-01
-1.59713352e+00 -1.14443159e+00 5.46658158e-01 -1.52885824e-01
5.77851832e-01 -1.25811815e-01 -1.07804930e+00 5.46003938e-01
1.31408170e-01 2.94213593e-01 1.27656356e-01 -1.64510474e-01
-7.23569214e-01 -4.79623735e-01 -1.07431698e+00 5.68389654e-01
1.02903974e+00 -6.45560443e-01 -9.84491780e-02 2.97968924e-01
7.73464382e-01 -5.97493887e-01 -1.12968898e+00 4.48601335e-01
4.73078579e-01 -1.26954007e+00 1.57966590e+00 -1.82067409e-01
9.87819374e-01 -3.94662023e-01 -3.88300717e-01 -1.10034049e+00
-5.06549656e-01 -7.02319860e-01 -2.45954990e-01 9.88722205e-01
-6.60334826e-02 -6.43528044e-01 4.57781792e-01 1.81861579e-01
-1.64213590e-02 -7.72431076e-01 -1.01510894e+00 -6.57608509e-01
1.19285367e-01 -1.69476084e-02 3.61283273e-01 9.80978489e-01
-9.08826888e-01 1.32064193e-01 -8.82505417e-01 4.72172379e-01
1.22279572e+00 2.82884985e-01 6.80343568e-01 -9.36557353e-01
-6.21506393e-01 -3.68086636e-01 -8.84626526e-03 -1.28451848e+00
-4.18377109e-02 -6.53117955e-01 -1.17937669e-01 -1.45102501e+00
3.00476283e-01 -3.42859030e-01 -2.85260499e-01 2.09838927e-01
-1.20054491e-01 8.10830355e-01 2.52790660e-01 4.50441092e-01
-5.81341922e-01 5.10857046e-01 1.35746229e+00 -5.17581478e-02
-4.60298397e-02 -1.34999245e-01 -8.64786565e-01 4.84441519e-01
8.41098666e-01 -2.37996131e-01 3.66471261e-02 -8.07653189e-01
2.39804804e-01 4.55379248e-01 8.14794779e-01 -8.82379770e-01
5.82118966e-02 9.67799574e-02 3.51816058e-01 -2.69508630e-01
5.50672591e-01 -6.10420465e-01 3.89824480e-01 3.49271268e-01
-4.95602697e-01 -1.55963793e-01 -9.93508846e-02 4.57749188e-01
-3.14997315e-01 1.71388447e-01 1.39020801e+00 -1.99042425e-01
-7.05253959e-01 3.94982368e-01 2.45438769e-01 9.02938470e-02
4.70790684e-01 -1.00534722e-01 -4.22708094e-01 -4.20949250e-01
-6.76134765e-01 -1.79842189e-01 7.44384706e-01 2.94922203e-01
6.12792313e-01 -1.10433698e+00 -1.21731877e+00 7.76413083e-02
-3.55778635e-01 1.11109838e-01 6.92309380e-01 1.06281483e+00
-4.97210205e-01 7.84877166e-02 -1.73266038e-01 -5.64028800e-01
-1.23595273e+00 4.41284537e-01 4.57422197e-01 -5.33074319e-01
-1.19959414e+00 7.55201399e-01 2.19427109e-01 -1.39822438e-01
1.39772967e-01 9.40407738e-02 -4.63206917e-02 -1.81805685e-01
8.15706491e-01 6.17339253e-01 -2.03243531e-02 -6.91019058e-01
1.10660316e-02 8.90997648e-01 -7.91115388e-02 1.62587881e-01
1.83069384e+00 -5.62086523e-01 -4.99185860e-01 5.87123670e-02
1.26172161e+00 5.52991405e-02 -1.68929768e+00 -4.86369759e-01
-3.89921993e-01 -6.91128016e-01 5.59385240e-01 -7.61129856e-01
-1.36173892e+00 7.58502722e-01 7.58403599e-01 -3.54843549e-02
1.50768054e+00 -1.79673389e-01 9.18694615e-01 4.19947729e-02
3.99491936e-01 -6.83238566e-01 -3.46422428e-03 3.21975499e-01
1.19526613e+00 -1.47532606e+00 2.67267078e-01 -2.92528898e-01
-3.68040293e-01 1.09083545e+00 1.54545739e-01 -4.06696379e-01
4.32804495e-01 2.73675412e-01 -2.29625087e-02 1.68122590e-01
-5.21683931e-01 3.79237048e-02 6.56355917e-02 5.23876965e-01
3.93406510e-01 -1.22674800e-01 -1.59313411e-01 1.84821188e-01
-8.40178877e-02 2.46754691e-01 6.83198035e-01 2.12517858e-01
-2.55010337e-01 -1.01238310e+00 -4.78858292e-01 1.95110649e-01
-7.08877146e-01 -3.92904401e-01 4.46893364e-01 4.49196815e-01
-1.88470230e-01 8.44034374e-01 -1.68696180e-01 -1.70641169e-02
5.08765914e-02 -5.64323843e-01 4.95352983e-01 -2.25618482e-01
-2.48735875e-01 1.01816319e-01 -2.26093918e-01 -9.27458167e-01
-4.61886108e-01 -5.18884063e-01 -5.63926101e-01 -5.27830899e-01
1.70625448e-01 -2.49159947e-01 5.27474821e-01 4.87600297e-01
3.48418266e-01 5.32641888e-01 7.49129534e-01 -1.19995344e+00
-9.19093609e-01 -7.64686823e-01 -5.62309146e-01 2.74222076e-01
7.92897642e-01 -1.74193695e-01 -4.80014741e-01 1.48149848e-01] | [11.099034309387207, -2.172783374786377] |
f55a42f3-931c-49cf-a706-21af4bf5d6cc | towards-geometry-guided-neural-relighting | 2008.05157 | null | https://arxiv.org/abs/2008.05157v1 | https://arxiv.org/pdf/2008.05157v1.pdf | Towards Geometry Guided Neural Relighting with Flash Photography | Previous image based relighting methods require capturing multiple images to acquire high frequency lighting effect under different lighting conditions, which needs nontrivial effort and may be unrealistic in certain practical use scenarios. While such approaches rely entirely on cleverly sampling the color images under different lighting conditions, little has been done to utilize geometric information that crucially influences the high-frequency features in the images, such as glossy highlight and cast shadow. We therefore propose a framework for image relighting from a single flash photograph with its corresponding depth map using deep learning. By incorporating the depth map, our approach is able to extrapolate realistic high-frequency effects under novel lighting via geometry guided image decomposition from the flashlight image, and predict the cast shadow map from the shadow-encoding transformed depth map. Moreover, the single-image based setup greatly simplifies the data capture process. We experimentally validate the advantage of our geometry guided approach over state-of-the-art image-based approaches in intrinsic image decomposition and image relighting, and also demonstrate our performance on real mobile phone photo examples. | ['Jin Zeng', 'Chengxi Yang', 'Wenxiu Sun', 'Zhanghan Ke', 'Di Qiu'] | 2020-08-12 | null | null | null | null | ['intrinsic-image-decomposition', 'image-relighting'] | ['computer-vision', 'computer-vision'] | [ 7.25676656e-01 -2.19516933e-01 3.71349901e-01 -3.00717175e-01
-5.23560643e-01 -4.29029733e-01 5.65986931e-01 -4.04246092e-01
2.11806335e-02 6.34608448e-01 1.20047055e-01 -2.48573616e-01
2.37996846e-01 -8.72269332e-01 -9.40878570e-01 -8.42500329e-01
4.07155961e-01 2.48930484e-01 1.15177222e-01 -2.41279989e-01
2.87431598e-01 4.41563338e-01 -1.84124529e+00 2.74834424e-01
8.10539961e-01 9.12880778e-01 4.75421727e-01 8.03895593e-01
-8.98036808e-02 7.13010073e-01 -5.08068144e-01 -2.49686986e-01
4.39314395e-01 -4.88812178e-01 -3.19550484e-01 4.13882852e-01
8.48088443e-01 -9.17175710e-01 -3.64184260e-01 7.76612461e-01
3.76025349e-01 -1.51706031e-02 4.74480271e-01 -8.28570485e-01
-5.37529290e-01 -2.69182235e-01 -7.03699946e-01 -3.51145566e-01
6.42705679e-01 3.17578673e-01 5.23709893e-01 -8.03103745e-01
6.30105317e-01 1.04428160e+00 7.92551816e-01 1.47669539e-01
-1.45695210e+00 -4.43398029e-01 -6.84645176e-02 5.48940226e-02
-1.12331605e+00 -3.82919163e-01 1.26287222e+00 -1.63003787e-01
5.44912100e-01 2.84256369e-01 9.78417218e-01 9.70208585e-01
1.98233783e-01 5.55606902e-01 1.66822052e+00 -6.82245612e-01
4.66124043e-02 2.42425129e-01 -2.82369792e-01 8.78408074e-01
4.16416600e-02 4.28494215e-01 -4.60447490e-01 1.50979578e-01
8.21947873e-01 2.63929695e-01 -8.41496348e-01 -4.54148084e-01
-1.14645708e+00 3.25756520e-01 4.85918552e-01 2.79387739e-02
-9.11746845e-02 3.97462696e-01 -2.19770119e-01 1.32688969e-01
4.95332658e-01 2.03438163e-01 -2.42153659e-01 2.19813548e-02
-1.13911879e+00 8.21405351e-02 6.34438276e-01 7.49298930e-01
1.35737944e+00 1.09812245e-01 1.19492054e-01 6.67575598e-01
1.04521975e-01 9.44638729e-01 6.21860735e-02 -1.15775001e+00
2.92365581e-01 3.48038405e-01 4.83326286e-01 -1.07768857e+00
-2.89441645e-01 -1.54666439e-01 -6.90592647e-01 4.52939808e-01
5.36217809e-01 -2.66362331e-03 -9.67747271e-01 1.18763173e+00
3.37096035e-01 1.89477459e-01 -8.41805115e-02 1.01548409e+00
6.95829153e-01 6.83113635e-01 -8.42663109e-01 -3.17486525e-01
1.14842761e+00 -8.29872549e-01 -7.16809034e-01 -1.75717771e-01
2.43867174e-01 -9.40160692e-01 1.61486840e+00 6.83945417e-01
-1.03622377e+00 -5.43304801e-01 -1.07999635e+00 -3.86126131e-01
-9.65506583e-02 2.00776085e-01 7.11988509e-01 8.37672949e-01
-1.00353241e+00 4.90185976e-01 -5.68292379e-01 -3.31504375e-01
1.63768664e-01 8.98753330e-02 -1.31020784e-01 -5.44228017e-01
-9.41444933e-01 6.21124029e-01 -1.63684487e-01 1.53589934e-01
-7.78079271e-01 -9.45991218e-01 -6.56181097e-01 -4.75702025e-02
2.65876770e-01 -6.61848843e-01 8.92349362e-01 -1.06368470e+00
-1.79283273e+00 7.83594012e-01 -3.29940766e-01 -1.00848503e-01
5.26752234e-01 -2.73637529e-02 -5.36162592e-02 4.11330402e-01
-3.82134378e-01 4.31484073e-01 1.25113046e+00 -2.03185439e+00
-8.13122317e-02 -1.63270310e-01 3.22393566e-01 4.45902467e-01
-3.36087346e-01 -4.20931041e-01 -6.19007647e-01 -3.45711946e-01
-1.81234330e-01 -1.00127113e+00 4.20323648e-02 3.08912992e-01
-3.61593246e-01 8.73975992e-01 1.04244518e+00 -6.12894058e-01
9.06787217e-01 -2.00825739e+00 -2.25380331e-01 -1.27230763e-01
-5.91927208e-02 1.13415718e-01 -6.70157745e-02 7.40279496e-01
1.01966679e-01 -2.82130569e-01 -2.32699707e-01 -7.15523839e-01
-2.75985152e-01 3.16399246e-01 -5.78565001e-01 4.14350361e-01
-2.97805667e-01 8.01441431e-01 -8.49167526e-01 -3.62373561e-01
9.09301519e-01 9.66790676e-01 -5.80462933e-01 3.16758126e-01
-2.55891830e-01 6.99910700e-01 -4.37423848e-02 6.77083254e-01
1.09593904e+00 6.64050430e-02 1.04766592e-01 -5.97404361e-01
-1.42656058e-01 -1.61581248e-01 -8.30983043e-01 1.73178542e+00
-1.08421326e+00 8.47473979e-01 3.46566774e-02 -5.98178208e-01
7.97851086e-01 -1.61943465e-01 4.40835953e-01 -9.23450768e-01
6.37944564e-02 1.26043305e-01 -5.92289150e-01 -5.37475348e-01
6.40605509e-01 -2.34925643e-01 2.03448325e-01 6.12256110e-01
-4.22483981e-01 -7.90987074e-01 -2.97159612e-01 4.84252684e-02
7.39481628e-01 5.26720703e-01 -1.42148539e-01 -3.36467698e-02
4.19280231e-01 -1.82125121e-01 1.81931958e-01 6.53429329e-01
3.35515141e-01 1.27101672e+00 2.30433181e-01 -6.38630450e-01
-1.15554130e+00 -9.51399207e-01 -1.09047025e-01 5.58365524e-01
5.87414443e-01 -2.23373234e-01 -9.73612964e-01 -2.87863016e-01
-1.64671496e-01 7.62107134e-01 -6.15814924e-01 -2.70136613e-02
-6.12505972e-01 -7.15632856e-01 2.12634187e-02 -7.40905777e-02
9.32780683e-01 -7.44425058e-01 -8.73085141e-01 -1.16568260e-01
-4.74665463e-01 -1.28030801e+00 -3.87499869e-01 1.50409446e-03
-5.69198191e-01 -1.03256667e+00 -8.29202950e-01 -3.71881068e-01
6.48074806e-01 9.27589536e-01 1.27381730e+00 1.99321479e-01
-8.33670318e-01 6.91538751e-01 -2.59262532e-01 -1.84607327e-01
-3.40323120e-01 -3.96898210e-01 -3.66701633e-01 4.37693149e-01
-2.88517892e-01 -7.94389963e-01 -1.20336998e+00 4.29809451e-01
-1.20613527e+00 7.30612457e-01 4.35941845e-01 8.35184455e-01
4.92134303e-01 2.73205251e-01 -1.86500236e-01 -7.68351972e-01
2.54782230e-01 8.03461149e-02 -7.61580884e-01 2.18921587e-01
-4.55236614e-01 -8.15708563e-02 6.71507180e-01 -3.24545830e-01
-1.65468895e+00 2.19080433e-01 9.11784992e-02 -5.35738170e-01
-3.21599357e-02 -6.45182431e-02 -1.96372002e-01 -4.38542485e-01
6.26253843e-01 4.24138874e-01 -8.57995376e-02 -3.86856526e-01
4.28359360e-01 5.80196500e-01 4.46397424e-01 -5.45275927e-01
9.32360470e-01 1.27493656e+00 2.46398032e-01 -1.21564102e+00
-9.44830060e-01 -1.41274363e-01 -6.31116927e-01 -4.43327367e-01
6.56628013e-01 -8.83867621e-01 -7.78046846e-01 6.62430108e-01
-1.04814327e+00 -8.88882577e-01 -1.38129473e-01 6.16716146e-02
-7.41849542e-01 6.43312693e-01 -2.76386738e-01 -6.94042921e-01
-1.92748532e-01 -1.18802130e+00 1.65483940e+00 1.02894887e-01
2.44688198e-01 -1.03298342e+00 -6.87325671e-02 5.86735308e-01
4.64185089e-01 3.40890884e-01 9.74738300e-01 9.17301238e-01
-1.18023407e+00 -5.28637134e-02 -3.67310643e-01 3.79653722e-01
3.09676677e-01 -4.91515882e-02 -1.24364698e+00 -1.68240234e-01
1.07967220e-01 -3.19925249e-01 8.56284976e-01 3.94229561e-01
1.21980906e+00 -5.73922358e-02 -2.77517378e-01 1.11947012e+00
1.98578131e+00 -1.83458507e-01 9.92426276e-01 1.53610334e-01
1.13275254e+00 6.20685041e-01 7.08997607e-01 3.80576670e-01
3.32838029e-01 1.08520567e+00 6.25662208e-01 -6.76075995e-01
-6.87593520e-01 -2.81472236e-01 3.47018778e-01 2.64565706e-01
-1.78036213e-01 -3.97346914e-01 -5.54435432e-01 2.92824060e-01
-1.47013712e+00 -8.65089953e-01 -1.70158982e-01 2.35904479e+00
7.62673855e-01 -4.91804540e-01 -3.60332072e-01 7.99920484e-02
3.14783514e-01 3.15653563e-01 -5.05812585e-01 -2.29092404e-01
-3.60786021e-01 1.55611530e-01 6.38356507e-01 9.39822674e-01
-6.30477786e-01 9.73039985e-01 6.10415316e+00 6.85404718e-01
-1.30202425e+00 -4.18428034e-02 7.10032523e-01 -4.06813622e-03
-8.58084917e-01 2.24387944e-01 -4.66759890e-01 3.88927460e-01
5.15263498e-01 3.40923041e-01 8.89874876e-01 2.50404030e-01
4.70339447e-01 -7.43323982e-01 -8.43963087e-01 1.30095112e+00
4.21181142e-01 -1.13819742e+00 4.41567693e-03 3.05541664e-01
1.05461562e+00 -1.76503107e-01 2.87091643e-01 -4.63559479e-01
-1.29425060e-02 -8.15644145e-01 5.80673993e-01 7.86734104e-01
1.17409503e+00 -5.49830139e-01 1.21965900e-01 1.63380519e-01
-8.84525359e-01 -2.62427069e-02 -3.80936682e-01 1.23718396e-01
3.44458252e-01 9.96464014e-01 -7.84343779e-01 4.75218743e-01
6.51295781e-01 6.34584486e-01 -4.95156497e-01 5.66576779e-01
-4.08784240e-01 4.14800793e-01 -4.21132356e-01 3.49524677e-01
-1.13359123e-01 -5.47819376e-01 3.17030460e-01 1.03785932e+00
5.02778769e-01 -5.21336570e-02 -6.14519343e-02 8.69620502e-01
1.23063624e-01 -2.75215089e-01 -7.73261189e-01 2.77119666e-01
5.29853031e-02 1.39977932e+00 -6.99900925e-01 -1.16169266e-01
-4.29943919e-01 1.46342790e+00 -4.40027053e-03 8.80984843e-01
-8.58912885e-01 -1.84733599e-01 5.48240721e-01 5.77937126e-01
2.98118740e-01 -4.52049464e-01 -2.60692209e-01 -1.39672685e+00
5.70616536e-02 -5.81445575e-01 -3.69107008e-01 -1.51303947e+00
-7.46033728e-01 4.15166020e-01 -3.54938917e-02 -1.38118565e+00
-7.75801716e-03 -5.23059547e-01 -4.84476119e-01 8.63870203e-01
-2.01012301e+00 -1.58173478e+00 -8.88125002e-01 8.84676278e-01
4.15135115e-01 4.94208366e-01 8.67991984e-01 1.39542609e-01
-1.53908506e-01 2.69478649e-01 3.98716539e-01 -4.26017314e-01
8.43320608e-01 -1.03659606e+00 1.20501176e-01 7.38965094e-01
4.11014035e-02 4.71112281e-01 8.09255838e-01 -3.73661101e-01
-1.83618951e+00 -8.76720965e-01 2.63473630e-01 -3.50206614e-01
2.01181710e-01 -6.65042400e-01 -6.18994176e-01 2.35222846e-01
3.34034294e-01 1.68199793e-01 2.69850701e-01 -3.08454722e-01
-3.24927568e-01 -5.11167824e-01 -1.03253579e+00 8.17645013e-01
1.06968403e+00 -5.95185161e-01 2.10790597e-02 4.80695128e-01
4.94644314e-01 -3.77141535e-01 -4.61170942e-01 2.05866694e-01
8.35266113e-01 -1.68992639e+00 1.33889604e+00 4.41026539e-01
6.00590825e-01 -4.95141357e-01 -1.77110374e-01 -1.31846404e+00
1.57983527e-01 -8.07006717e-01 6.70097768e-02 1.11451566e+00
-1.21541033e-02 -6.44493520e-01 7.87349284e-01 5.08418798e-01
-1.15382046e-01 -5.71419597e-01 -4.41196442e-01 -4.69500929e-01
-2.75619060e-01 -4.24175382e-01 6.89315915e-01 7.54212499e-01
-5.64174891e-01 -1.99468769e-02 -8.85292888e-01 1.90726012e-01
8.44919503e-01 7.58902252e-01 1.28624749e+00 -8.85307848e-01
-5.85792184e-01 1.01168595e-01 2.66004708e-02 -1.07543898e+00
9.00487881e-03 -4.11199600e-01 9.72255394e-02 -1.56976223e+00
2.09236145e-01 -6.07347012e-01 3.13604504e-01 1.69902697e-01
-1.13669910e-01 7.84308314e-01 1.33814096e-01 3.26359808e-01
-1.59625784e-01 6.25420034e-01 1.41352987e+00 3.02247657e-03
-1.59540102e-01 -4.17375833e-01 -5.48619330e-01 7.85052121e-01
4.81147200e-01 -7.38274232e-02 -7.30179310e-01 -7.00779319e-01
4.28362966e-01 -5.56104584e-03 5.44310153e-01 -9.57977355e-01
-1.97783470e-01 -2.31900916e-01 6.05753601e-01 -5.07811844e-01
8.69321167e-01 -1.02335000e+00 4.78095680e-01 2.36036271e-01
6.22488819e-02 -5.45779467e-01 -1.24941925e-02 6.54047549e-01
-3.24508809e-02 -8.91712084e-02 7.31377780e-01 -2.78885007e-01
-4.57956851e-01 1.75482675e-01 -9.26207155e-02 -2.64038593e-01
7.34924436e-01 -6.93674028e-01 -2.87950844e-01 -7.06150472e-01
-7.59726241e-02 -4.17031318e-01 1.22727823e+00 -1.85916200e-02
7.65291989e-01 -1.08682156e+00 -2.21265152e-01 5.54243922e-01
6.97927922e-02 -1.25317015e-02 5.39169908e-01 7.21762359e-01
-8.99380863e-01 1.48939878e-01 1.11427873e-01 -7.69118369e-01
-1.44623792e+00 5.18300831e-01 4.38363135e-01 -7.75603876e-02
-9.45291579e-01 4.77561206e-01 7.54721165e-01 -1.53189272e-01
-1.26616403e-01 -4.32495683e-01 3.22397679e-01 -2.78356910e-01
4.51202750e-01 7.38496706e-02 2.29105890e-01 -5.12783885e-01
1.31052852e-01 1.21594417e+00 1.81061074e-01 -3.41467619e-01
1.17457116e+00 -5.55274189e-01 6.62263110e-02 4.00374174e-01
1.31828165e+00 2.24096507e-01 -1.78529906e+00 -1.49656802e-01
-9.11613464e-01 -1.08958638e+00 3.99348140e-01 -7.61491001e-01
-1.04499304e+00 1.19570100e+00 4.96857077e-01 -7.16296136e-02
1.64643550e+00 -3.24627548e-01 9.79325533e-01 2.93256223e-01
7.08272934e-01 -9.89169657e-01 2.95609862e-01 1.44401968e-01
8.90942752e-01 -1.23038924e+00 2.82574236e-01 -6.85560882e-01
-4.34369832e-01 1.15861785e+00 1.20405525e-01 -2.91314838e-03
5.19215763e-01 3.15535128e-01 1.34310052e-01 -8.55116993e-02
-2.87267178e-01 -1.49347842e-01 1.39113277e-01 7.62969673e-01
1.27097309e-01 -2.42642850e-01 1.12642020e-01 -2.46011317e-01
-4.00177278e-02 -4.68611196e-02 8.09052348e-01 6.53026521e-01
-1.17629565e-01 -1.07796717e+00 -6.29569709e-01 -8.65132548e-03
-2.13271007e-01 -3.33916694e-01 -2.56546944e-01 7.50369608e-01
1.37324885e-01 7.40685940e-01 -8.88283551e-02 -1.95851430e-01
8.82574469e-02 -3.07892203e-01 1.07382107e+00 -3.94131869e-01
-1.58133015e-01 2.37967402e-01 2.86278455e-03 -7.03950465e-01
-5.80302894e-01 -5.55025935e-01 -9.63064432e-01 -1.56794682e-01
-2.77569532e-01 -5.65642297e-01 9.34201121e-01 6.44788265e-01
3.11164141e-01 2.42512956e-01 7.98796773e-01 -1.57067859e+00
3.05171847e-01 -5.85479438e-01 -6.78856611e-01 6.11268461e-01
6.95361257e-01 -7.69601047e-01 -6.50688291e-01 2.88637519e-01] | [9.856095314025879, -2.8990211486816406] |
12288d1b-1212-4f62-a556-410498282125 | on-the-evolution-of-syntactic-information | 2101.11492 | null | https://arxiv.org/abs/2101.11492v2 | https://arxiv.org/pdf/2101.11492v2.pdf | On the Evolution of Syntactic Information Encoded by BERT's Contextualized Representations | The adaptation of pretrained language models to solve supervised tasks has become a baseline in NLP, and many recent works have focused on studying how linguistic information is encoded in the pretrained sentence representations. Among other information, it has been shown that entire syntax trees are implicitly embedded in the geometry of such models. As these models are often fine-tuned, it becomes increasingly important to understand how the encoded knowledge evolves along the fine-tuning. In this paper, we analyze the evolution of the embedded syntax trees along the fine-tuning process of BERT for six different tasks, covering all levels of the linguistic structure. Experimental results show that the encoded syntactic information is forgotten (PoS tagging), reinforced (dependency and constituency parsing) or preserved (semantics-related tasks) in different ways along the fine-tuning process depending on the task. | ['Laura Pérez-Mayos', 'Leo Wanner', 'Miguel Ballesteros', 'Roberto Carlini'] | 2021-01-27 | null | https://aclanthology.org/2021.eacl-main.191 | https://aclanthology.org/2021.eacl-main.191.pdf | eacl-2021-2 | ['constituency-parsing'] | ['natural-language-processing'] | [ 9.26764086e-02 5.08601725e-01 -2.22422704e-01 -6.35861337e-01
-3.52634162e-01 -8.74994159e-01 4.73221630e-01 6.28307581e-01
-5.28136909e-01 6.85649276e-01 7.92127907e-01 -3.21474642e-01
-2.53776044e-01 -8.22772205e-01 -7.36481905e-01 -6.44137800e-01
-2.04737768e-01 4.84607100e-01 2.98483580e-01 -4.80188757e-01
3.76328260e-01 4.17062610e-01 -1.16691315e+00 6.70434415e-01
7.40940154e-01 5.03011882e-01 5.17610013e-01 5.62435925e-01
-7.11718738e-01 7.82181263e-01 -6.09882176e-01 -5.39633691e-01
-6.18486516e-02 -1.21296823e-01 -1.11512125e+00 -2.83256948e-01
4.01153386e-01 3.70099515e-01 -1.29919410e-01 1.10885763e+00
7.58576170e-02 1.18551657e-01 5.74258268e-01 -3.79244447e-01
-8.42339575e-01 1.28967011e+00 -2.69665092e-01 5.53058684e-01
-1.06553466e-03 -1.66598275e-01 1.42347693e+00 -8.60605061e-01
7.41183043e-01 1.43024421e+00 5.21112621e-01 4.95515257e-01
-1.43002987e+00 -2.22858921e-01 4.85527575e-01 1.68563873e-01
-1.07293808e+00 -8.32425207e-02 7.74125576e-01 -6.26091361e-01
1.23844159e+00 -1.64885044e-01 2.93280542e-01 1.04556167e+00
5.86761177e-01 6.50091648e-01 1.14031160e+00 -8.31403852e-01
2.05536280e-02 3.12136263e-01 7.58135796e-01 8.69983912e-01
4.81655121e-01 2.12159932e-01 -6.05836332e-01 1.49614468e-01
6.83670878e-01 -3.72571796e-01 -1.33776188e-01 -4.75959331e-01
-8.54485393e-01 1.17112184e+00 4.79244918e-01 9.65256095e-01
-1.72306702e-01 1.79249629e-01 5.17691016e-01 3.37505668e-01
4.34137136e-01 8.04090619e-01 -1.13979781e+00 6.42203689e-02
-7.80728877e-01 4.05031070e-03 7.26600766e-01 6.35571122e-01
1.08307552e+00 5.25418436e-04 -3.47849518e-01 1.01691246e+00
2.45783195e-01 -1.07524604e-01 7.88928747e-01 -7.00841129e-01
8.96717846e-01 6.83999360e-01 -1.75352409e-01 -9.79296863e-01
-6.55015886e-01 -6.22462809e-01 -6.33862555e-01 -1.55269489e-01
6.67009175e-01 -2.57237196e-01 -7.05016017e-01 2.09487367e+00
-1.03600629e-01 -4.45075899e-01 1.64159521e-01 5.48314810e-01
5.78271389e-01 7.37090945e-01 6.66567206e-01 -8.76019672e-02
1.47369611e+00 -8.97541642e-01 -7.02675343e-01 -6.66190207e-01
1.03794837e+00 -5.74912429e-01 1.30813205e+00 3.63862589e-02
-9.96215463e-01 -7.82116175e-01 -9.46359634e-01 -2.81654269e-01
-6.27052665e-01 -4.69073728e-02 6.52108133e-01 7.38848150e-01
-1.22306800e+00 9.23035920e-01 -4.27154303e-01 -3.85232627e-01
3.09711367e-01 3.19888681e-01 -3.60344529e-01 -5.60127944e-02
-1.52605808e+00 1.23273456e+00 8.36417317e-01 -8.03680867e-02
-5.32723904e-01 -8.49091411e-01 -1.01585138e+00 5.14977753e-01
1.93250448e-01 -6.37738824e-01 1.19512630e+00 -1.10038400e+00
-1.37290490e+00 1.26702988e+00 -1.94271266e-01 -3.40251386e-01
-7.85125047e-02 -1.68161556e-01 -5.08706495e-02 -1.17214739e-01
-3.59857711e-03 5.00891924e-01 7.58630395e-01 -1.15971196e+00
-6.00539327e-01 -6.24796450e-01 4.42901760e-01 1.91285372e-01
-3.02112132e-01 2.36090988e-01 1.60792783e-01 -1.05043864e+00
-1.91428915e-01 -6.42005384e-01 -3.47627640e-01 -5.12691557e-01
-1.36354670e-01 -6.00140333e-01 1.07109010e-01 -5.61403155e-01
1.45847547e+00 -2.09991670e+00 6.65417433e-01 -9.76156816e-02
1.04820073e-01 1.88553751e-01 -3.69921058e-01 5.74003577e-01
-2.61404425e-01 4.92334157e-01 -3.93945217e-01 -3.05143952e-01
2.32046872e-01 6.01097643e-01 -5.49714684e-01 7.57952705e-02
2.61197478e-01 9.36231911e-01 -7.95290709e-01 -4.28184688e-01
-1.29863083e-01 2.34314725e-01 -5.37927091e-01 -4.88923416e-02
-3.93559813e-01 5.01995146e-01 -5.37566841e-01 7.30068535e-02
3.76212448e-01 -1.43208936e-01 3.88720155e-01 -3.50981094e-02
-1.95717037e-01 8.04027796e-01 -7.74266839e-01 1.87360728e+00
-5.67309916e-01 4.55388516e-01 -5.05648647e-03 -1.34260869e+00
7.67112255e-01 2.18152389e-01 -1.41724378e-01 -7.73041129e-01
6.69685602e-02 8.31889082e-03 3.25566649e-01 -4.12311971e-01
5.51283598e-01 -4.48598981e-01 -4.31555629e-01 5.97741306e-01
4.70269293e-01 3.25191878e-02 4.57322508e-01 6.62352368e-02
9.39488292e-01 7.07216933e-02 4.88893032e-01 -6.20540738e-01
6.89215124e-01 1.93139508e-01 3.30295652e-01 7.04965055e-01
-9.21483859e-02 9.27885026e-02 5.47224224e-01 -3.38193655e-01
-8.86094391e-01 -1.07487035e+00 -4.71623778e-01 1.77355599e+00
-3.53293151e-01 -4.68262494e-01 -9.66223240e-01 -7.84789145e-01
-2.44753495e-01 9.62807000e-01 -8.06735694e-01 -3.94395262e-01
-1.09382498e+00 -6.70295894e-01 5.58713377e-01 5.47754109e-01
3.57502341e-01 -1.67124271e+00 -3.80638123e-01 4.35807526e-01
-1.12302937e-01 -1.00960279e+00 -1.50866553e-01 6.96330428e-01
-1.26083469e+00 -8.40857804e-01 -4.54626560e-01 -1.07182252e+00
6.14142299e-01 -1.98472410e-01 1.53620338e+00 1.39197141e-01
-8.17120075e-02 3.79454762e-01 -4.08423036e-01 -2.54088283e-01
-3.82360488e-01 5.42376339e-01 -2.75536120e-01 -1.64620474e-01
3.29722464e-01 -4.07858729e-01 -8.89521018e-02 -1.68427721e-01
-9.12027180e-01 -2.90045649e-01 5.08579135e-01 6.72406971e-01
4.66449887e-01 9.89554897e-02 4.48235601e-01 -1.22798324e+00
8.37451518e-01 -4.12777901e-01 -4.23860967e-01 4.18205440e-01
-2.94080317e-01 8.20435047e-01 7.06128895e-01 -2.37625927e-01
-1.35015237e+00 -1.26944333e-01 -2.44534329e-01 2.13929519e-01
-3.68623435e-01 8.86050403e-01 -1.90304384e-01 4.66393456e-02
5.54651320e-01 5.67939654e-02 -6.18959010e-01 -7.00497448e-01
5.93049347e-01 5.56810061e-03 -2.26609614e-02 -1.06513155e+00
7.13023543e-01 8.53123590e-02 -1.67419806e-01 -8.88554215e-01
-1.54282010e+00 -1.74102575e-01 -1.13968956e+00 2.91316211e-01
1.08401942e+00 -3.77954990e-01 -1.87525630e-01 1.68700457e-01
-1.32244897e+00 -5.62923133e-01 -5.83604634e-01 2.38943413e-01
-4.09650594e-01 3.47008348e-01 -7.74827123e-01 -3.54536116e-01
-1.06129542e-01 -1.05803442e+00 8.63966465e-01 7.08033293e-02
-4.02589947e-01 -1.62955427e+00 3.68576467e-01 1.61769345e-01
3.51331800e-01 -2.53395498e-01 1.78138626e+00 -7.10636377e-01
-2.65588403e-01 3.13238889e-01 -1.77276656e-01 7.64864311e-02
-1.50681464e-02 -3.68209988e-01 -8.90192211e-01 -6.53470308e-02
3.76765490e-01 -2.80821979e-01 1.07775450e+00 4.51307952e-01
1.05112123e+00 -4.63310182e-01 -1.99959382e-01 5.06010056e-01
1.32558203e+00 -1.86760306e-01 5.91201365e-01 6.07699335e-01
4.70104992e-01 1.19304717e+00 3.99659544e-01 -7.31919259e-02
3.25164199e-01 4.69560325e-01 1.31957665e-01 3.48770201e-01
-1.10178500e-01 -3.81514162e-01 5.78569412e-01 1.02078569e+00
2.97426060e-02 -6.76108301e-02 -9.00629103e-01 4.59426075e-01
-1.64936519e+00 -7.54236400e-01 -5.48982620e-02 1.93181479e+00
1.28016531e+00 1.99380219e-01 -3.92359644e-01 5.92595823e-02
6.49594367e-01 4.68529284e-01 -1.74740061e-01 -8.58066857e-01
-2.33204201e-01 7.56882727e-01 2.23743603e-01 7.05849588e-01
-1.06181371e+00 1.49233007e+00 6.64133883e+00 6.36283278e-01
-1.01865447e+00 1.56303525e-01 3.77005637e-01 3.73720318e-01
-2.71056592e-01 9.68670100e-03 -1.16770995e+00 1.15753800e-01
1.00510645e+00 -9.33197588e-02 2.30059028e-01 6.96976006e-01
-6.46105185e-02 -7.25883767e-02 -1.23823667e+00 4.59714919e-01
-7.35749379e-02 -1.28514743e+00 2.58459806e-01 -1.13160290e-01
5.87387204e-01 2.44369544e-02 4.83548902e-02 7.24465966e-01
5.91773391e-01 -1.01593018e+00 8.19486916e-01 3.28128189e-01
4.26059961e-01 -4.84353364e-01 5.18472195e-01 5.97897589e-01
-1.21573269e+00 -2.41514415e-01 -6.66318595e-01 -1.91588640e-01
1.57073900e-01 5.16418397e-01 -5.69060326e-01 2.54381835e-01
7.79736161e-01 4.74969923e-01 -9.67406332e-01 5.46120405e-01
-8.59066129e-01 7.89459586e-01 1.55382067e-01 4.78388071e-02
3.44201952e-01 -6.58749714e-02 4.63907003e-01 1.40057576e+00
6.90346509e-02 5.93398474e-02 9.94957387e-02 8.79634559e-01
-6.48926720e-02 2.36443505e-01 -4.62356746e-01 -2.33859301e-01
3.48454639e-02 1.08909786e+00 -6.75935864e-01 -1.42551541e-01
-3.26967180e-01 6.30047798e-01 9.68250513e-01 3.58777225e-01
-5.16196847e-01 4.33132313e-02 6.91300690e-01 1.42400429e-01
6.48623228e-01 -5.34557879e-01 -3.38811934e-01 -1.16035044e+00
-2.15836421e-01 -4.91099864e-01 6.06298268e-01 -6.05879724e-01
-1.47213507e+00 5.79027891e-01 -6.63590655e-02 -2.10464403e-01
-3.94131318e-02 -8.44667852e-01 -3.70684296e-01 7.45789528e-01
-1.70144689e+00 -8.56998920e-01 3.30572248e-01 5.28438628e-01
7.22704649e-01 -1.22013129e-01 1.08190703e+00 1.09828571e-02
-5.67725122e-01 5.40864110e-01 -1.27184033e-01 3.34566891e-01
4.11597848e-01 -1.33615577e+00 2.52843171e-01 5.63525021e-01
6.71964407e-01 9.81144965e-01 5.12624383e-01 -5.55645049e-01
-9.17670012e-01 -1.05751109e+00 1.42599285e+00 -6.25510454e-01
9.58041012e-01 -3.97653550e-01 -1.25599468e+00 8.87330294e-01
4.73305702e-01 -9.74695534e-02 7.03493655e-01 4.74369198e-01
-6.08372152e-01 -1.62137654e-02 -8.27605665e-01 3.71635586e-01
1.12798655e+00 -6.43145382e-01 -1.43484068e+00 1.26588941e-01
1.00397909e+00 -1.54384151e-01 -6.82010174e-01 2.08369866e-01
-1.40046831e-02 -8.46847892e-01 8.86292219e-01 -1.11703968e+00
4.45959657e-01 5.04430532e-02 -2.08327845e-01 -1.49308074e+00
-7.97410309e-01 -1.71915233e-01 4.30553667e-02 1.20915854e+00
7.01165438e-01 -5.73934138e-01 5.31121790e-01 2.42446557e-01
-3.39577168e-01 -5.84915280e-01 -8.48157644e-01 -5.02262414e-01
6.83610380e-01 -2.55580306e-01 1.84999973e-01 8.94595087e-01
-5.92708588e-02 9.71256733e-01 2.91688800e-01 3.80772650e-02
3.99677694e-01 8.81621987e-02 1.44408822e-01 -1.58589232e+00
-3.67805630e-01 -6.45615935e-01 1.44478884e-02 -9.98784840e-01
5.09550750e-01 -1.41229606e+00 -1.06183857e-01 -1.52346945e+00
5.91503046e-02 -6.11544013e-01 -5.24186611e-01 6.07783794e-01
-9.16359574e-02 -3.25874478e-01 3.01871598e-01 1.11975588e-01
-4.76905227e-01 3.35587233e-01 1.27768731e+00 -5.61783612e-02
-3.04029472e-02 -2.11742893e-01 -8.73636484e-01 1.04626119e+00
9.87130761e-01 -7.51757264e-01 -2.62136310e-01 -1.00353444e+00
6.56072795e-01 -1.54672831e-01 1.66652706e-02 -6.79888368e-01
8.91251713e-02 -7.24888518e-02 1.93357512e-01 -2.68114001e-01
1.00247733e-01 -8.00934374e-01 -6.21843100e-01 4.43235338e-01
-7.66564608e-01 2.13426933e-01 3.90895993e-01 4.31499839e-01
-3.09040487e-01 -7.66517520e-01 8.91324937e-01 -5.62333167e-01
-8.02893043e-01 5.70783615e-02 -4.81240124e-01 6.56060994e-01
7.54564166e-01 -9.82162822e-03 -1.59116179e-01 2.87132025e-01
-1.17071497e+00 -5.20314164e-02 -5.64956525e-03 6.57705784e-01
2.34163493e-01 -9.79623795e-01 -5.71370661e-01 8.00698474e-02
-4.41396721e-02 -2.28929266e-01 1.79088891e-01 5.15674472e-01
-2.48999923e-01 8.89659584e-01 -9.81229022e-02 -3.79803389e-01
-9.34141994e-01 6.55733109e-01 3.25293690e-01 -1.03565478e+00
-3.37848336e-01 1.05389535e+00 5.91563940e-01 -6.39018059e-01
3.77539247e-01 -6.37408376e-01 -7.09774554e-01 3.89573544e-01
3.49447757e-01 -5.32949343e-02 4.03411463e-02 -5.47838271e-01
-1.81410208e-01 9.18680251e-01 -2.39868522e-01 1.24098763e-01
1.52477217e+00 -1.73327938e-01 -4.68095124e-01 7.43502736e-01
8.79564524e-01 1.24636479e-01 -8.93842995e-01 -5.06392419e-01
7.51855850e-01 1.02005489e-01 -2.21297503e-01 -8.75079691e-01
-9.48108792e-01 1.35280406e+00 8.14867020e-02 2.93757647e-01
6.69667482e-01 2.63773799e-01 5.14899850e-01 5.30343890e-01
6.22045457e-01 -1.05812716e+00 1.89234570e-01 1.30744481e+00
1.06716704e+00 -8.07177544e-01 -3.18917215e-01 -5.10586739e-01
-6.05997980e-01 1.17627895e+00 4.15786207e-01 -4.58680242e-01
8.06856215e-01 3.09392959e-01 -2.30065912e-01 -2.63089806e-01
-6.58133447e-01 -2.36413032e-01 1.67424262e-01 5.88053882e-01
7.72996247e-01 -5.16119339e-02 -3.30325842e-01 1.08352864e+00
-5.11992395e-01 -4.08781290e-01 2.64608175e-01 7.01832950e-01
-7.32906640e-01 -1.43527079e+00 -2.37846807e-01 7.98350275e-02
-6.09017611e-01 -5.13027668e-01 -6.22357368e-01 7.14986444e-01
2.95642614e-01 7.57410228e-01 -1.01913027e-02 1.79038122e-01
3.77245277e-01 5.33342600e-01 7.85838068e-01 -1.39757872e+00
-1.06613088e+00 -2.96408921e-01 1.57731712e-01 -4.63249892e-01
-3.39803815e-01 -5.31526446e-01 -1.57166493e+00 1.30446360e-01
-8.47998038e-02 4.14972991e-01 3.23933095e-01 1.24766457e+00
2.39402473e-01 7.86213040e-01 1.54970467e-01 -7.13665128e-01
-6.73576236e-01 -1.04904521e+00 -4.52030182e-01 1.86593577e-01
1.13844745e-01 -7.49678314e-01 -4.04446781e-01 1.02402024e-01] | [10.551362037658691, 9.232505798339844] |
0ee55e90-1328-49f0-87f8-f388a81c7d99 | automatic-identification-of-epileptic | 1910.11183 | null | http://arxiv.org/abs/1910.11183v2 | http://arxiv.org/pdf/1910.11183v2.pdf | Automatic Identification of Epileptic Seizures from EEG Signals using Sparse Representation-based Classification | Identifying seizure activities in non-stationary electroencephalography (EEG)
is a challenging task, since it is time-consuming, burdensome, and dependent on
expensive human resources and subject to error and bias. A computerized seizure
identification scheme can eradicate the above problems, assist clinicians and
benefit epilepsy research. So far, several attempts were made to develop
automatic systems to help neurophysiologists accurately identify epileptic
seizures. In this research, a fully automated system is presented to
automatically detect the various states of the epileptic seizure. The proposed
method is based on sparse representation-based classification (SRC) theory and
the proposed dictionary learning using electroencephalogram (EEG) signals.
Furthermore, the proposed method does not require additional preprocessing and
extraction of features which is common in the existing methods. The proposed
method reached the sensitivity, specificity and accuracy of 100% in 8 out of 9
scenarios. It is also robust to the measurement noise of level as much as 0 dB.
Compared to state-of-the-art algorithms and other common methods, the proposed
method outperformed them in terms of sensitivity, specificity and accuracy.
Moreover, it includes the most comprehensive scenarios for epileptic seizure
detection, including different combinations of 2 to 5 class scenarios. The
proposed automatic identification of epileptic seizures method can reduce the
burden on medical professionals in analyzing large data through visual
inspection as well as in deprived societies suffering from a shortage of
functional magnetic resonance imaging (fMRI) equipment and specialized
physician. | [] | 2020-04-28 | null | null | null | null | ['sparse-representation-based-classification'] | ['computer-vision'] | [ 3.86860400e-01 -4.83168453e-01 5.41452527e-01 -1.35190943e-02
-4.58477587e-01 -4.18973684e-01 1.55888470e-02 1.46341041e-01
-4.24331009e-01 9.16489542e-01 -1.29794613e-01 -1.33024469e-01
-4.66379315e-01 -2.60421693e-01 8.65030736e-02 -8.57397020e-01
-2.78877914e-01 1.45709664e-01 1.73754115e-02 -7.77400061e-02
4.66450125e-01 6.21035457e-01 -1.55559480e+00 -1.00324973e-01
1.04850268e+00 1.15698111e+00 5.47846794e-01 1.46210745e-01
2.93478072e-01 4.16737109e-01 -7.74490714e-01 2.94474572e-01
1.80059850e-01 -5.64340830e-01 -2.10908249e-01 8.76858681e-02
-3.52121621e-01 -4.01589610e-02 -1.90290332e-01 1.26990640e+00
8.96165669e-01 -6.16839938e-02 7.48925388e-01 -1.09770179e+00
-4.15920079e-01 1.73841089e-01 -7.21608579e-01 7.25916445e-01
4.69481260e-01 -2.80438773e-02 -1.17110508e-02 -9.09357905e-01
2.91459658e-03 3.85006696e-01 5.70363164e-01 3.13755721e-01
-8.87662649e-01 -1.27279568e+00 -1.87718287e-01 5.05662680e-01
-1.80871153e+00 -5.04793704e-01 8.83477449e-01 -6.85746968e-01
9.39982116e-01 2.07686082e-01 9.34092045e-01 9.59767222e-01
5.47451615e-01 9.24890786e-02 1.69827557e+00 -4.04679656e-01
3.30983251e-01 2.23166972e-01 2.86903292e-01 3.71863186e-01
5.27728438e-01 7.52682164e-02 -5.26972890e-01 -2.11373076e-01
7.79873431e-01 2.05579817e-01 -7.34487474e-01 -3.93333435e-02
-1.13423800e+00 5.41482747e-01 -1.40601829e-01 9.88391101e-01
-7.83578277e-01 -5.15770674e-01 3.61860543e-01 2.07660839e-01
1.40289024e-01 4.20112222e-01 -2.09441796e-01 -2.72805363e-01
-1.24477208e+00 -3.65027755e-01 5.62164307e-01 6.27635777e-01
3.52383196e-01 5.77279329e-01 1.48626298e-01 6.15654588e-01
9.63554680e-02 4.23551023e-01 9.94250238e-01 -8.04554373e-02
1.22584157e-01 5.36828458e-01 2.89630378e-04 -1.14841163e+00
-7.28228509e-01 -7.39143550e-01 -1.01695943e+00 6.22777082e-03
-1.11441739e-01 -2.28944227e-01 -7.38447368e-01 1.26662040e+00
-1.07019283e-01 3.36959094e-01 4.61250879e-02 9.49145377e-01
7.07061350e-01 4.02649730e-01 1.48671195e-01 -7.16374099e-01
1.56436586e+00 -1.51068792e-01 -1.07988214e+00 -1.62990093e-01
1.56438574e-01 -8.93081725e-01 6.11209631e-01 9.22972560e-01
-7.29135215e-01 -3.01901519e-01 -1.14771163e+00 7.69385338e-01
-1.50725320e-01 4.61919546e-01 6.43216014e-01 7.20151901e-01
-1.01924074e+00 1.18067496e-01 -9.40439999e-01 -5.34797072e-01
2.74388880e-01 7.80499399e-01 -5.89424729e-01 3.06761831e-01
-1.11420226e+00 1.04717541e+00 3.54891002e-01 2.61848569e-01
-7.56041646e-01 -2.85427034e-01 -6.94640040e-01 3.53857763e-02
-1.33624494e-01 -1.49794623e-01 6.38341904e-01 -9.95202482e-01
-1.23442471e+00 5.03548861e-01 -2.26820514e-01 -4.73221302e-01
8.05934668e-02 1.67003013e-02 -9.00701940e-01 4.30263281e-01
3.58146727e-02 9.53331664e-02 7.53566980e-01 -6.59758031e-01
-4.01790321e-01 -7.58297563e-01 -5.83747387e-01 1.54645354e-01
-3.84577513e-01 3.79074782e-01 2.99110770e-01 -7.73138583e-01
4.59198892e-01 -6.56304240e-01 1.06570445e-01 -5.92744648e-01
-1.92617364e-02 -4.49180277e-03 9.03090000e-01 -9.87407625e-01
1.25498176e+00 -2.13758397e+00 -1.47931024e-01 4.12276655e-01
2.09947210e-02 4.23381180e-01 3.74896973e-01 4.23990965e-01
-3.40643674e-01 -3.62238973e-01 -3.97929758e-01 2.08148375e-01
-3.34349036e-01 -2.78145075e-01 6.33425917e-03 8.61393332e-01
-5.18095084e-02 2.75252968e-01 -5.79697549e-01 -1.85511306e-01
5.50438225e-01 7.48301864e-01 1.06075428e-01 2.87962765e-01
7.94208050e-01 8.45827103e-01 -4.42442030e-01 8.44436884e-01
5.96584976e-01 -7.41719687e-03 1.85549874e-02 -2.08470434e-01
-1.14009909e-01 -1.63177941e-02 -1.49333429e+00 1.43196595e+00
-3.07266414e-01 6.61473274e-01 -1.42438952e-02 -1.49486029e+00
1.20580375e+00 9.09750283e-01 6.68456495e-01 -6.97317362e-01
4.07934785e-01 4.27980185e-01 2.80044764e-01 -9.12327170e-01
-2.11432442e-01 -1.72463015e-01 2.23554760e-01 2.88024306e-01
4.46193367e-02 1.59307003e-01 -9.12870020e-02 -2.34801337e-01
1.02365780e+00 -2.43629515e-01 9.57143724e-01 -5.86201906e-01
7.97037363e-01 -3.29649031e-01 6.43078685e-01 2.66309470e-01
-2.26206198e-01 1.88536644e-02 -1.65904135e-01 -2.78755069e-01
-5.09452581e-01 -6.47301376e-01 -5.96939027e-01 1.70711741e-01
1.72879577e-01 -5.55988809e-05 -7.27424562e-01 -1.00949191e-01
-3.03322047e-01 5.15859783e-01 -2.05649510e-01 -4.33216542e-02
-2.14718610e-01 -1.00087976e+00 4.14622337e-01 2.99395025e-01
6.36401236e-01 -1.18130505e+00 -1.09646976e+00 3.86028498e-01
-7.42561072e-02 -1.05736375e+00 -2.44021835e-03 2.90931463e-01
-8.00723732e-01 -1.18610084e+00 -7.88917899e-01 -1.02919304e+00
8.50920677e-01 1.02719612e-01 4.97358948e-01 -7.50415102e-02
-6.79238856e-01 1.55399293e-01 -4.87107128e-01 -6.25607669e-01
1.44695804e-01 -3.58744442e-01 4.77684528e-01 2.75779873e-01
8.88525307e-01 -9.89485025e-01 -7.95596957e-01 2.59373903e-01
-6.06335878e-01 -3.43039781e-01 8.30171227e-01 7.45793819e-01
4.62816715e-01 4.89172846e-01 1.29325747e+00 -4.12436247e-01
9.35958385e-01 -5.44986606e-01 -7.09031463e-01 -1.75653547e-02
-7.25837409e-01 -4.11778241e-01 8.07954073e-01 -4.52458620e-01
-7.97178984e-01 2.49305546e-01 1.17832348e-02 -1.30994007e-01
-4.66658175e-01 2.64922142e-01 -6.49314746e-02 -5.53649902e-01
5.09159923e-01 8.21458220e-01 -2.87984610e-01 -5.26403308e-01
-6.40249848e-01 1.23610842e+00 4.81909245e-01 1.39011666e-01
3.48321021e-01 1.93775788e-01 5.55412285e-03 -9.70625520e-01
-1.30277231e-01 -6.39559865e-01 -4.23350424e-01 -8.31022635e-02
7.52682269e-01 -1.02274179e+00 -5.25069177e-01 5.27714849e-01
-9.04595673e-01 4.60166156e-01 4.80669200e-01 1.11807466e+00
-2.23503083e-01 5.22746384e-01 -2.58895606e-01 -1.25664341e+00
-8.14891517e-01 -1.33156419e+00 6.32104337e-01 1.66175738e-01
-3.38784218e-01 -6.44756734e-01 -2.64945000e-01 3.06341555e-02
5.59334517e-01 4.69351619e-01 6.85397685e-01 -7.84231424e-01
-1.02446288e-01 -4.57680851e-01 4.22991626e-02 3.22939932e-01
6.41514063e-01 -6.58674777e-01 -8.37505400e-01 -5.70928395e-01
8.24213207e-01 -6.64349869e-02 2.36313686e-01 3.89402181e-01
1.14431858e+00 -2.22581759e-01 -3.10079724e-01 5.31149805e-01
1.59073400e+00 1.18288219e+00 6.03237987e-01 1.92968056e-01
2.12622687e-01 4.09251094e-01 2.98802346e-01 5.50295174e-01
-1.00999862e-01 3.15236807e-01 1.56665310e-01 -1.14439882e-01
2.05649689e-01 3.11405629e-01 7.75248408e-02 1.12045431e+00
-6.74432367e-02 3.54900770e-02 -8.57994080e-01 7.26740420e-01
-1.46188831e+00 -9.15011644e-01 -1.69017129e-02 2.31934476e+00
5.69195390e-01 -9.71131027e-02 -1.10864855e-01 8.80363047e-01
7.81597793e-01 -4.45455521e-01 -4.75885361e-01 -1.66626364e-01
-1.14645541e-01 8.22121859e-01 3.38171870e-01 8.53626877e-02
-9.40658092e-01 4.31711733e-01 6.23513937e+00 6.49813652e-01
-1.39932752e+00 3.26207817e-01 5.31999730e-02 9.03272480e-02
2.65047729e-01 -2.15201691e-01 -3.48977804e-01 9.39973950e-01
7.88054585e-01 -2.20009625e-01 5.51441312e-01 6.32673621e-01
4.43387687e-01 -4.30026412e-01 -8.11934114e-01 1.67882299e+00
3.62619638e-01 -8.14978838e-01 -1.47505507e-01 -7.95105845e-02
5.05579054e-01 -3.19484442e-01 -2.15502933e-01 -1.24662474e-01
-7.04234540e-01 -1.21517873e+00 1.19999133e-01 5.89261889e-01
9.26248610e-01 -8.72025013e-01 1.14810491e+00 5.07488608e-01
-1.20907605e+00 -2.17185721e-01 -1.08773261e-01 -3.63566607e-01
3.44645754e-02 5.22481024e-01 -6.11007094e-01 3.94321322e-01
7.15303898e-01 5.88742495e-01 -3.04103464e-01 1.23872888e+00
-2.06356823e-01 6.76669478e-01 -2.46178731e-01 -1.43233508e-01
1.52938836e-03 -2.30074197e-01 5.98258197e-01 1.11457109e+00
7.59363413e-01 5.56928217e-01 -1.37226731e-01 4.67598051e-01
4.35839295e-01 4.26380813e-01 -7.39081264e-01 -1.45279653e-02
6.08985424e-01 1.26197290e+00 -9.00040269e-01 -1.52378336e-01
-4.89081711e-01 8.55902374e-01 -1.76723078e-01 1.62314296e-01
-4.38359231e-01 -8.42197001e-01 4.02200893e-02 1.79577485e-01
-3.49300876e-02 4.88283485e-02 -2.23436683e-01 -1.11827910e+00
1.49706706e-01 -1.12853456e+00 1.03742450e-01 -6.14755988e-01
-1.00934112e+00 1.05236912e+00 4.71663326e-02 -1.49182975e+00
-1.81142777e-01 -5.19462228e-01 -5.70074320e-01 8.70990276e-01
-1.27711093e+00 -5.11680603e-01 -5.24286509e-01 1.02220726e+00
5.73827326e-01 -6.28637135e-01 1.04935896e+00 5.96991539e-01
-5.29331982e-01 3.74577343e-01 1.84203277e-03 -9.41370353e-02
5.01260579e-01 -7.53252804e-01 -4.66792554e-01 1.06556702e+00
-8.37611929e-02 7.59567976e-01 6.65638566e-01 -5.20396590e-01
-1.20254230e+00 -5.35874486e-01 8.04929435e-01 2.67909765e-01
6.22484982e-01 -2.49701574e-01 -7.68096387e-01 3.80202621e-01
3.07918012e-01 -2.16937721e-01 8.89632642e-01 -4.44184095e-01
4.58110005e-01 -1.87676400e-01 -1.33074510e+00 1.17752597e-01
7.18395829e-01 -3.80997568e-01 -1.00469661e+00 4.85583901e-01
-2.42945209e-01 -1.07041851e-01 -8.98209095e-01 4.93403941e-01
4.02594954e-01 -9.47913408e-01 4.72039312e-01 -6.61226064e-02
-3.86945128e-01 -3.92074257e-01 5.17274812e-02 -1.27911270e+00
-3.12488437e-01 -6.73467159e-01 1.78590864e-01 8.15860629e-01
1.80774443e-02 -1.04117095e+00 3.66881460e-01 2.84595996e-01
-1.30050480e-01 -8.71789515e-01 -1.08693683e+00 -7.14155853e-01
-5.76742113e-01 -1.48054034e-01 5.57319582e-01 8.76538932e-01
4.20210093e-01 1.63151175e-01 -4.39675242e-01 2.77470857e-01
6.39192581e-01 -9.44685489e-02 1.26990780e-01 -1.38474941e+00
-1.90906599e-02 -9.12174061e-02 -9.85283732e-01 -2.60491222e-01
-1.49524249e-02 -7.14925706e-01 -3.65193337e-01 -1.51812589e+00
1.73910290e-01 -2.85702765e-01 -7.79851735e-01 4.39419836e-01
1.75451458e-01 4.22847629e-01 -3.07270586e-01 3.85414034e-01
2.35991813e-02 2.67443031e-01 7.70586789e-01 8.04617107e-02
-2.72593200e-01 -1.19037889e-02 -7.46367216e-01 9.10423040e-01
9.75749254e-01 -6.35891140e-01 -6.53548121e-01 -1.99689135e-01
-1.50358334e-01 2.63126642e-01 2.48224720e-01 -1.32325041e+00
4.05842692e-01 1.68200195e-01 6.92817092e-01 -5.70661604e-01
4.36561704e-01 -1.16365254e+00 6.67885065e-01 6.01800501e-01
2.33204961e-01 4.00722057e-01 2.09213361e-01 3.27638000e-01
-4.37737465e-01 -2.33817279e-01 8.28268886e-01 -1.91219136e-01
-6.85181558e-01 -7.89080933e-02 -6.30877018e-01 -2.89605469e-01
1.50730395e+00 -6.27011955e-01 9.23739448e-02 -1.93025321e-01
-7.81022668e-01 -1.46936536e-01 5.48696071e-02 4.34382297e-02
7.98752189e-01 -1.10233128e+00 -5.24568141e-01 6.05951488e-01
6.69784769e-02 -6.14122570e-01 4.40386623e-01 1.40978301e+00
-5.75786233e-01 6.58583641e-01 -7.70958364e-01 -3.87932658e-01
-1.26204216e+00 3.61311913e-01 1.70774713e-01 1.24720186e-01
-7.98261166e-01 5.44007063e-01 -8.72613713e-02 4.81332928e-01
3.24192286e-01 -1.22895911e-01 -6.46838605e-01 -4.11197357e-02
7.10242271e-01 3.88414055e-01 4.52546120e-01 -8.04547846e-01
-8.28504384e-01 8.06697726e-01 1.97466031e-01 1.17959984e-01
1.17524731e+00 4.59717065e-02 -2.48085424e-01 2.23191798e-01
9.35698271e-01 5.86827956e-02 -5.49197137e-01 2.05171287e-01
-2.28290781e-01 -4.06010300e-01 3.08770806e-01 -9.98669565e-01
-9.60718691e-01 7.90973008e-01 1.10015762e+00 1.50227517e-01
1.57261312e+00 -5.30592322e-01 6.53911531e-01 2.60169297e-01
9.15302992e-01 -9.58549142e-01 -4.61269379e-01 -1.90333426e-01
8.11383188e-01 -1.08378136e+00 6.15707971e-02 -6.28869161e-02
-5.58609188e-01 1.11479688e+00 3.74647170e-01 -3.52920771e-01
8.66195083e-01 4.45689410e-01 -8.05891352e-04 -3.25082153e-01
-2.72759467e-01 8.58768523e-02 3.30306888e-01 7.53677547e-01
4.03043330e-01 1.01568498e-01 -1.02540147e+00 9.73816752e-01
-1.96020141e-01 2.87033379e-01 3.49539995e-01 1.04109919e+00
-4.72341776e-01 -7.20403135e-01 -6.21156931e-01 7.73002803e-01
-9.17229116e-01 -2.36456901e-01 -1.02668189e-01 7.57583439e-01
3.59787941e-01 1.33343160e+00 -1.03562310e-01 -3.60528886e-01
1.64307654e-01 5.64250629e-03 4.17171836e-01 -4.99699295e-01
-5.75580001e-01 2.36129627e-01 -2.39686444e-01 -2.74209946e-01
-6.16019726e-01 -5.73706746e-01 -1.09374952e+00 3.27462554e-01
-5.55439353e-01 4.95088577e-01 7.89536119e-01 9.54502225e-01
4.70985532e-01 5.12704372e-01 6.32492125e-01 -5.50747216e-01
-3.01546007e-01 -1.28525460e+00 -1.03357804e+00 1.39735624e-01
1.09894879e-01 -1.09363139e+00 -7.15629399e-01 1.04177324e-02] | [13.190361022949219, 3.447903633117676] |
6d3e26a0-3cd6-4d02-ab94-cdbce56c47a6 | deep-association-learning-for-unsupervised | 1808.07301 | null | http://arxiv.org/abs/1808.07301v1 | http://arxiv.org/pdf/1808.07301v1.pdf | Deep Association Learning for Unsupervised Video Person Re-identification | Deep learning methods have started to dominate the research progress of
video-based person re-identification (re-id). However, existing methods mostly
consider supervised learning, which requires exhaustive manual efforts for
labelling cross-view pairwise data. Therefore, they severely lack scalability
and practicality in real-world video surveillance applications. In this work,
to address the video person re-id task, we formulate a novel Deep Association
Learning (DAL) scheme, the first end-to-end deep learning method using none of
the identity labels in model initialisation and training. DAL learns a deep
re-id matching model by jointly optimising two margin-based association losses
in an end-to-end manner, which effectively constrains the association of each
frame to the best-matched intra-camera representation and cross-camera
representation. Existing standard CNNs can be readily employed within our DAL
scheme. Experiment results demonstrate that our proposed DAL significantly
outperforms current state-of-the-art unsupervised video person re-id methods on
three benchmarks: PRID 2011, iLIDS-VID and MARS. | ['Yanbei Chen', 'Shaogang Gong', 'Xiatian Zhu'] | 2018-08-22 | null | null | null | null | ['unsupervised-person-re-identification'] | ['computer-vision'] | [ 4.66225185e-02 -2.80117154e-01 -2.92459249e-01 -5.64949512e-01
-4.81046200e-01 -4.02308255e-01 8.25473487e-01 -9.81200710e-02
-8.72540593e-01 5.77353895e-01 2.80287713e-01 1.75599962e-01
4.77199033e-02 -4.09651250e-01 -6.96581125e-01 -5.68688035e-01
3.63689661e-02 6.98845804e-01 -6.15385808e-02 1.23153597e-01
-1.48969531e-01 2.68941313e-01 -1.55584717e+00 1.67231530e-01
4.45259064e-01 8.88712049e-01 -2.04552516e-01 6.31023288e-01
4.41521764e-01 8.23719978e-01 -3.69846076e-01 -9.47212756e-01
5.48924625e-01 -2.48320550e-01 -8.04236293e-01 1.38223410e-01
8.87892723e-01 -7.63055146e-01 -8.76195788e-01 1.10953307e+00
8.04076850e-01 4.00342464e-01 5.89447856e-01 -1.44267058e+00
-6.92867815e-01 1.54834360e-01 -7.89172232e-01 4.35420811e-01
5.12296915e-01 -6.39836565e-02 6.75154150e-01 -7.63901472e-01
4.13077891e-01 1.31472051e+00 1.11792803e+00 8.54082882e-01
-1.07296729e+00 -8.38836670e-01 1.98734775e-01 5.67596853e-01
-1.65606260e+00 -6.39311194e-01 5.14576197e-01 -5.93865216e-01
1.02665770e+00 4.70453836e-02 4.55253035e-01 1.20207942e+00
-4.72787768e-01 9.26422000e-01 5.71151912e-01 -3.86390954e-01
-2.45073840e-01 -3.95168290e-02 -3.59629020e-02 5.66202700e-01
2.45482236e-01 2.79646158e-01 -4.80249107e-01 -1.55206263e-01
8.21093619e-01 3.01446289e-01 -5.82329668e-02 -4.38024253e-01
-1.19060159e+00 6.35060906e-01 3.15168917e-01 -8.12880043e-03
-1.78676337e-01 1.04409069e-01 6.54220343e-01 1.60694793e-01
4.29191023e-01 -1.28529340e-01 -1.45689443e-01 4.48127463e-02
-9.62329805e-01 6.23941064e-01 4.01944637e-01 9.28945601e-01
4.83557880e-01 -4.41923618e-01 -3.81879538e-01 1.06979966e+00
3.29169899e-01 3.87558043e-01 4.45095688e-01 -8.43522072e-01
5.60508013e-01 4.28408921e-01 3.45110893e-01 -9.00252759e-01
-3.36214185e-01 -3.16768676e-01 -1.08200693e+00 -8.56036544e-02
4.78775054e-01 -1.69924468e-01 -8.45440209e-01 1.80425394e+00
4.75093424e-01 7.38870740e-01 6.52951598e-02 1.15176857e+00
8.96012902e-01 3.02427024e-01 3.37438703e-01 1.21560663e-01
1.42550969e+00 -1.24681723e+00 -5.18376887e-01 -3.25148284e-01
6.12269282e-01 -5.11421621e-01 2.28472188e-01 8.19011033e-03
-9.67857838e-01 -9.28632557e-01 -9.05561268e-01 -5.89893423e-02
-1.61776975e-01 5.02174437e-01 2.35796586e-01 6.59721732e-01
-1.14307666e+00 2.79058874e-01 -5.69423974e-01 -5.63197076e-01
5.10110259e-01 6.35295987e-01 -9.16143417e-01 -2.85160422e-01
-1.25735104e+00 7.18912363e-01 5.50295234e-01 3.40346158e-01
-9.00508344e-01 -6.90156281e-01 -8.92953277e-01 -1.56516314e-01
3.10074836e-01 -1.01642382e+00 1.16401923e+00 -1.26960576e+00
-1.23168206e+00 1.51713967e+00 -2.11069301e-01 -6.38477147e-01
7.55585849e-01 -4.54572439e-01 -5.55047452e-01 -6.10237457e-02
3.47634435e-01 8.43818188e-01 7.99819648e-01 -1.25132036e+00
-9.55305338e-01 -5.02353370e-01 2.03988284e-01 2.68330187e-01
-4.58718300e-01 4.08844948e-01 -1.06200671e+00 -9.22495127e-01
-4.93336648e-01 -1.11714256e+00 -1.27396703e-01 2.40585883e-03
-2.92376846e-01 -5.41751683e-01 6.15963280e-01 -1.01585901e+00
1.15966964e+00 -2.00122762e+00 2.05916524e-01 8.47608522e-02
2.47079074e-01 7.42136002e-01 -7.83704147e-02 1.18188664e-01
-3.29477787e-01 -2.55683094e-01 -6.96357116e-02 -9.70728755e-01
5.41364737e-02 -1.31339639e-01 2.12818488e-01 7.88990259e-01
-1.20623171e-01 7.84655213e-01 -9.30913806e-01 -5.81163645e-01
4.53158677e-01 6.31541669e-01 -4.47800487e-01 7.68195987e-01
3.73718530e-01 3.82456571e-01 1.30847041e-02 5.70781469e-01
7.59195328e-01 -1.28982261e-01 1.93390906e-01 -5.02593637e-01
1.10512726e-01 -2.30819687e-01 -1.17616010e+00 1.71503878e+00
-9.63207632e-02 6.32863045e-01 -6.20208532e-02 -1.26745665e+00
6.72917306e-01 3.01042229e-01 6.94845021e-01 -6.84882879e-01
-9.02300607e-03 -1.68575525e-01 -3.14765185e-01 -4.38676298e-01
4.34896946e-01 2.55397856e-01 6.88620284e-02 2.79156864e-01
1.65267482e-01 1.14555407e+00 1.00160614e-01 5.54262251e-02
6.73104405e-01 1.16451874e-01 2.77168781e-01 -1.00401357e-01
1.05808628e+00 -4.07386988e-01 6.94128096e-01 8.30428362e-01
-6.60133600e-01 7.33652055e-01 -2.36724652e-02 -9.50685620e-01
-1.31378341e+00 -8.54680598e-01 1.21210888e-01 1.22176957e+00
2.89232314e-01 -2.37123594e-01 -9.93465900e-01 -9.63089406e-01
-2.63685640e-02 8.07141811e-02 -7.32924044e-01 -4.22160663e-02
-7.93879211e-01 -8.12097549e-01 8.49037468e-01 6.64420843e-01
9.22811031e-01 -8.04348886e-01 1.68076549e-02 8.35423023e-02
-5.44776142e-01 -1.34869802e+00 -8.95689189e-01 -5.63914120e-01
-3.00002992e-01 -1.24992394e+00 -1.32295132e+00 -1.10563898e+00
7.95104146e-01 5.17087877e-01 9.84476864e-01 1.79952085e-01
-2.53181428e-01 6.24786079e-01 -2.60177732e-01 3.87042318e-03
5.46359131e-03 1.15291156e-01 4.80499536e-01 4.49179679e-01
1.03488874e+00 -6.76739365e-02 -8.86376321e-01 5.84212720e-01
-5.16455293e-01 -6.86503872e-02 2.26967558e-01 8.59750152e-01
4.53181207e-01 -5.51495031e-02 4.60828900e-01 -4.76207525e-01
2.91088015e-01 -4.32934344e-01 -3.21623981e-01 5.95779836e-01
-3.43156904e-01 -2.17633232e-01 2.92774886e-01 -3.79547894e-01
-1.08751929e+00 2.63871312e-01 -1.63588762e-01 -5.93647540e-01
-4.19128865e-01 1.21328004e-01 -3.65583926e-01 -1.42998427e-01
2.69107133e-01 1.28619835e-01 -8.47421139e-02 -3.94656390e-01
2.26603344e-01 9.21823561e-01 1.14343333e+00 -4.36369866e-01
6.89378858e-01 5.99713087e-01 -3.99313062e-01 -5.96116364e-01
-7.26324379e-01 -1.01779747e+00 -9.56081867e-01 -3.66335034e-01
1.14207721e+00 -1.51297307e+00 -9.11366999e-01 9.59783435e-01
-1.25268781e+00 -1.43817529e-01 1.79692283e-01 4.22532380e-01
-3.24507982e-01 7.99187124e-01 -3.41601640e-01 -7.58540392e-01
-4.70784038e-01 -1.02151871e+00 1.17699051e+00 3.90171170e-01
-2.07279146e-01 -9.32239413e-01 2.57236332e-01 6.83343887e-01
3.68356495e-03 1.77100793e-01 1.10356830e-01 -6.35142326e-01
-1.92549422e-01 -5.50002515e-01 -5.02424955e-01 3.03608537e-01
-1.10710626e-02 -5.23661494e-01 -1.03707206e+00 -7.24770904e-01
-6.22628748e-01 -3.40727121e-01 1.08390749e+00 2.81217635e-01
1.20286739e+00 -3.25265229e-01 -5.24468124e-01 1.01148438e+00
1.19646037e+00 -1.26313530e-02 6.96769774e-01 8.22585821e-01
1.18265438e+00 7.17103362e-01 4.76102740e-01 5.28393269e-01
9.14726317e-01 1.13408995e+00 2.10098550e-01 -2.78145432e-01
-7.73872286e-02 -2.60677934e-01 3.23084295e-01 2.16908589e-01
-5.18552840e-01 -3.70519698e-01 -8.15075815e-01 6.99660361e-01
-2.30771995e+00 -1.30219007e+00 2.30740115e-01 2.38254285e+00
4.06644136e-01 -3.29926312e-01 7.48307943e-01 -9.39011723e-02
1.31592786e+00 2.60407012e-02 -5.72373092e-01 2.88291693e-01
-9.72683579e-02 -4.66766775e-01 7.96374261e-01 2.73881972e-01
-1.89368963e+00 8.04405928e-01 5.51833725e+00 6.68824494e-01
-5.73519349e-01 1.98682308e-01 6.36710465e-01 3.04376651e-02
3.72165114e-01 -4.70605314e-01 -1.20148408e+00 5.74897587e-01
7.10094333e-01 1.87382057e-01 3.57595950e-01 7.62015462e-01
1.74620282e-02 2.96577930e-01 -1.38662076e+00 1.77362001e+00
5.08347332e-01 -1.13055193e+00 5.89928590e-02 -3.67144272e-02
7.61215091e-01 -2.04197615e-01 -3.43480445e-02 2.15170696e-01
2.09251821e-01 -1.06505525e+00 7.45422244e-01 5.52782595e-01
9.35411870e-01 -9.68074977e-01 1.12664223e+00 5.84587120e-02
-1.44490540e+00 -2.28519201e-01 -3.30323249e-01 2.23283127e-01
3.57523382e-01 -2.78184954e-02 -4.80515212e-01 5.99702239e-01
1.18382406e+00 1.06623697e+00 -5.91038346e-01 1.08771944e+00
1.34206206e-01 2.29152441e-01 -1.16770811e-01 5.71828663e-01
9.20253694e-02 1.16178125e-01 3.49553734e-01 1.58035612e+00
1.18093103e-01 -3.63273621e-02 4.56206918e-01 3.09792936e-01
-2.31239215e-01 -3.66406351e-01 -3.11129242e-01 3.41036886e-01
3.67795974e-01 9.30118382e-01 -3.49822462e-01 -4.40261960e-01
-6.34005785e-01 1.36705446e+00 5.03484845e-01 3.67301464e-01
-1.01423514e+00 -8.07327628e-02 1.05830014e+00 9.94237587e-02
3.59778166e-01 -9.97468978e-02 2.81734943e-01 -1.23114991e+00
-1.03396252e-02 -9.79298890e-01 8.84803534e-01 -3.13728333e-01
-1.75250471e+00 4.39227104e-01 2.31635407e-01 -1.23598790e+00
-3.75679612e-01 -5.19319594e-01 -4.10868824e-01 7.41315901e-01
-1.46998894e+00 -1.66446626e+00 -5.36841810e-01 9.50625479e-01
4.19251412e-01 -6.82721317e-01 5.69195092e-01 9.11940575e-01
-9.64736938e-01 1.36747444e+00 2.07341328e-01 8.67836952e-01
1.03703952e+00 -9.97834146e-01 7.46170580e-01 1.14503992e+00
-1.24815442e-01 4.94922042e-01 3.55584323e-01 -5.96827388e-01
-1.01007950e+00 -1.35356295e+00 9.42309499e-01 -4.47984993e-01
2.23229080e-01 -3.65787178e-01 -6.91663563e-01 8.35583210e-01
3.35073024e-02 2.20597118e-01 8.47863615e-01 -1.64124817e-02
-4.72989619e-01 -2.98379242e-01 -1.06082416e+00 3.35736871e-01
1.32789350e+00 -7.10561991e-01 -3.06421906e-01 2.87026823e-01
3.50674421e-01 -2.74665177e-01 -8.65940452e-01 4.20164526e-01
8.63622010e-01 -8.34880590e-01 1.70049655e+00 -7.77831316e-01
2.43605785e-02 -4.87944365e-01 2.16331091e-02 -8.32189023e-01
-7.53004372e-01 -5.50499916e-01 -3.52458209e-01 1.54356539e+00
-2.08672658e-01 -3.65668207e-01 9.10534084e-01 8.57871771e-01
2.70148486e-01 -2.33704537e-01 -1.04078007e+00 -7.13538349e-01
-3.22669923e-01 -1.32473215e-01 6.80467963e-01 1.04588902e+00
-4.66587633e-01 -4.77454700e-02 -1.07453680e+00 4.09512371e-01
1.22212756e+00 -5.48860490e-01 1.07385957e+00 -1.07057500e+00
-2.24795088e-01 -3.30697358e-01 -8.30798388e-01 -1.10869908e+00
4.52554554e-01 -7.26978779e-01 -2.84295995e-02 -1.33061135e+00
6.24350429e-01 -4.92059886e-01 -4.71561968e-01 3.45727026e-01
-5.05570352e-01 4.91366655e-01 2.20242977e-01 5.34941792e-01
-1.05030310e+00 3.60944420e-01 4.58496243e-01 -4.00209337e-01
4.46645655e-02 2.75766645e-02 -5.15636086e-01 7.02926159e-01
6.17722809e-01 -3.56533498e-01 -2.64595181e-01 -7.17926383e-01
-1.34623110e-01 -3.66119593e-01 8.46268773e-01 -1.08989775e+00
6.08453512e-01 1.31940052e-01 8.52498353e-01 -5.47600210e-01
2.13368937e-01 -7.95524120e-01 2.96175122e-01 2.46169448e-01
-4.17923421e-01 2.19858572e-01 4.52549080e-04 7.40174055e-01
-1.50596619e-01 -7.50170797e-02 8.32844138e-01 -1.01925634e-01
-1.14326239e+00 8.30189347e-01 -2.99136266e-02 3.59988734e-02
1.13011241e+00 -3.59470338e-01 -9.16484371e-02 -2.36742243e-01
-5.50654769e-01 3.72976959e-01 4.01103735e-01 6.80396616e-01
4.67161059e-01 -1.61073720e+00 -1.03976464e+00 2.35920042e-01
2.87505567e-01 -1.72026753e-01 6.70369744e-01 4.63053167e-01
-4.82269287e-01 3.97706449e-01 -2.84988016e-01 -7.20185161e-01
-1.58856046e+00 7.70485163e-01 6.29685223e-01 -3.91638547e-01
-5.91612041e-01 1.00870419e+00 4.09757644e-01 -5.00039816e-01
6.64817393e-01 5.02417624e-01 -5.08087158e-01 -1.63212985e-01
1.06427860e+00 6.03265762e-01 -2.18941569e-01 -1.24396718e+00
-6.09985709e-01 6.40691519e-01 -3.63333195e-01 1.03464790e-01
1.03947508e+00 -3.84624749e-01 4.93774749e-02 -2.34001756e-01
1.40307987e+00 -6.51382565e-01 -1.51667082e+00 -4.76292789e-01
-1.57712623e-01 -6.06148839e-01 -1.43754721e-01 -4.95900929e-01
-1.08506632e+00 4.82870728e-01 1.09765494e+00 -3.45443994e-01
1.00886989e+00 -1.49960026e-01 1.07404721e+00 4.03412074e-01
1.69740736e-01 -1.25248897e+00 -5.69351353e-02 2.08717912e-01
5.31630576e-01 -1.67507875e+00 8.14889967e-02 -1.57808214e-01
-5.54904401e-01 8.60836625e-01 8.99039447e-01 1.43930847e-02
6.19129717e-01 -9.84679312e-02 -3.57130147e-03 2.31302246e-01
-6.90493882e-02 -2.27022678e-01 4.71686244e-01 1.09552956e+00
3.05863559e-01 5.10244444e-02 1.59358099e-01 5.79865336e-01
2.44486883e-01 1.15977407e-01 -7.32935742e-02 6.19688869e-01
5.68879629e-03 -1.10750771e+00 -4.92316067e-01 1.75445989e-01
-5.81839204e-01 1.57430008e-01 -2.51809716e-01 5.97395658e-01
5.66332817e-01 9.13398027e-01 3.05250406e-01 -5.48367441e-01
2.69095451e-01 -3.28321427e-01 4.86090690e-01 -7.89334103e-02
-5.34877181e-01 -2.76049107e-01 1.38897896e-01 -4.29727525e-01
-9.60146308e-01 -7.38189280e-01 -6.22608006e-01 -6.93329930e-01
-6.49741888e-02 -1.04459099e-01 1.62299022e-01 1.06987596e+00
3.33905339e-01 6.25201613e-02 3.62197667e-01 -1.13093591e+00
-2.50565737e-01 -8.12125921e-01 -2.24886686e-02 8.25175881e-01
4.87925857e-01 -8.58338356e-01 1.02240844e-02 5.24722993e-01] | [14.681193351745605, 0.9653854966163635] |
b65f251c-796e-4025-8950-865d552db205 | repsum-unsupervised-dialogue-summarization | null | null | https://aclanthology.org/2021.acl-long.471 | https://aclanthology.org/2021.acl-long.471.pdf | RepSum: Unsupervised Dialogue Summarization based on Replacement Strategy | In the field of dialogue summarization, due to the lack of training data, it is often difficult for supervised summary generation methods to learn vital information from dialogue context with limited data. Several attempts on unsupervised summarization for text by leveraging semantic information solely or auto-encoder strategy (i.e., sentence compression), it however cannot be adapted to the dialogue scene due to the limited words in utterances and huge gap between the dialogue and its summary. In this study, we propose a novel unsupervised strategy to address this challenge, which roots from the hypothetical foundation that a superior summary approximates a replacement of the original dialogue, and they are roughly equivalent for auxiliary (self-supervised) tasks, e.g., dialogue generation. The proposed strategy RepSum is applied to generate both extractive and abstractive summary with the guidance of the followed n{\^{}}th utterance generation and classification tasks. Extensive experiments on various datasets demonstrate the superiority of the proposed model compared with the state-of-the-art methods. | ['Zhenglu Yang', 'Changlong Sun', 'Xiaozhong Liu', 'Tianyi Wang', 'Yating Zhang', 'Xiyan Fu'] | 2021-08-01 | null | null | null | acl-2021-5 | ['sentence-compression'] | ['natural-language-processing'] | [ 7.79242575e-01 7.07933784e-01 -5.15084639e-02 -3.76406372e-01
-6.56690300e-01 -2.20879763e-01 8.09359789e-01 4.24551547e-01
-1.58797890e-01 1.43497789e+00 9.81823266e-01 2.58954447e-02
4.62557450e-02 -5.73206365e-01 -2.45454952e-01 -4.58096057e-01
3.72575939e-01 4.41638678e-01 2.94155572e-02 -5.50260961e-01
6.33248866e-01 -2.18198925e-01 -1.44992328e+00 4.16245341e-01
1.51631403e+00 7.21926391e-01 4.84680533e-01 7.32040524e-01
-6.56054199e-01 9.35294747e-01 -1.11195993e+00 -4.38611329e-01
-2.83785164e-01 -1.24984670e+00 -1.04805243e+00 3.51617455e-01
7.52732623e-03 -6.16596282e-01 -5.16507804e-01 9.34418142e-01
6.26509309e-01 3.64148587e-01 9.04768050e-01 -8.42115402e-01
-5.06725073e-01 8.79341125e-01 -2.05927402e-01 1.07946724e-01
7.49020934e-01 -2.64155567e-01 1.08019376e+00 -5.87898910e-01
6.37126625e-01 1.19575214e+00 3.04237455e-01 8.62036586e-01
-8.05622816e-01 -1.66688353e-01 6.19627684e-02 1.15253367e-02
-6.30599141e-01 -7.94462502e-01 1.00513911e+00 5.75331263e-02
9.73043859e-01 5.05895197e-01 5.15911162e-01 1.15756655e+00
1.35766447e-01 1.10264015e+00 8.69472444e-01 -4.95786071e-01
1.89127952e-01 3.11901718e-01 2.89550960e-01 4.53787148e-01
1.63441256e-01 -5.44762671e-01 -7.23696172e-01 -1.46045551e-01
3.04522783e-01 -2.36215249e-01 -4.34007019e-01 2.30202258e-01
-1.09956110e+00 8.55185449e-01 3.50208225e-04 1.71040788e-01
-4.61181700e-01 -4.48256940e-01 6.99406505e-01 2.80461848e-01
6.72459006e-01 6.13572001e-01 -2.34303772e-01 -3.03093404e-01
-1.13461709e+00 4.89455581e-01 1.21946061e+00 1.13711488e+00
4.72597063e-01 3.41626197e-01 -3.93594146e-01 8.44668448e-01
-3.85800973e-02 8.19317177e-02 8.77126038e-01 -7.68463016e-01
7.42267609e-01 6.29869759e-01 1.27329305e-01 -7.76211798e-01
-2.41099194e-01 -2.77153522e-01 -1.19812429e+00 -4.79263246e-01
-2.29558367e-02 -7.10017443e-01 -4.47653562e-01 1.49095917e+00
1.42281607e-01 -1.68499932e-01 8.06225479e-01 4.82183009e-01
1.53594613e+00 1.02456415e+00 -2.21191272e-01 -9.98195410e-01
9.97343957e-01 -1.31747174e+00 -1.25318277e+00 -3.05619597e-01
5.24455190e-01 -7.72452652e-01 8.19692731e-01 1.43060863e-01
-1.23289347e+00 -6.62563443e-01 -1.25476527e+00 -5.36702238e-02
-5.44329211e-02 2.37975940e-01 5.34714162e-01 3.13523442e-01
-8.23966682e-01 7.26030648e-01 -3.47585887e-01 -6.16760254e-01
6.21964373e-02 8.10794085e-02 -2.41604954e-01 3.15280259e-01
-1.33073890e+00 8.72975111e-01 9.36880648e-01 6.53948337e-02
-4.18824375e-01 -1.80793107e-01 -9.08488512e-01 1.39580831e-01
4.88325566e-01 -8.50182772e-01 1.51209843e+00 -8.82938683e-01
-1.95393193e+00 2.14576244e-01 -1.86614767e-01 -7.54681051e-01
4.02948171e-01 -4.12859380e-01 5.77397570e-02 3.22058409e-01
1.33140072e-01 4.76977348e-01 6.60951078e-01 -1.10567999e+00
-7.33948350e-01 -2.33498946e-01 1.08723097e-01 9.41533744e-01
-4.77980942e-01 -1.95980415e-01 9.67008024e-02 -6.31599128e-01
2.90898681e-02 -4.95612323e-01 -2.54310310e-01 -8.71121824e-01
-8.33389282e-01 -4.42520887e-01 9.87372041e-01 -1.04496717e+00
1.46428025e+00 -1.64341092e+00 2.86631376e-01 -5.26246130e-01
3.35450806e-02 4.73572612e-01 1.45462811e-01 1.08786273e+00
2.14628145e-01 -1.24387275e-02 -5.26578128e-01 -5.00052989e-01
-1.10732585e-01 1.22709803e-01 -5.29750109e-01 -1.09511428e-01
1.59566864e-01 7.05078721e-01 -1.14202762e+00 -8.66247177e-01
3.40812020e-02 -1.60675853e-01 -2.63942450e-01 6.72231972e-01
-4.68810260e-01 6.69737101e-01 -7.16847539e-01 2.24944562e-01
3.16174090e-01 7.49964565e-02 2.01715484e-01 -1.09286360e-01
-3.83289792e-02 6.82379723e-01 -8.55596125e-01 1.95834160e+00
-2.48449221e-01 4.50021386e-01 -1.16866857e-01 -1.27521944e+00
1.19420230e+00 5.99998415e-01 1.34864837e-01 -2.62957573e-01
9.89644676e-02 2.28730381e-01 -6.03069440e-02 -6.04947209e-01
1.20741057e+00 -1.53265730e-01 -3.24928582e-01 5.92913270e-01
4.28523123e-01 -4.95511055e-01 4.54960197e-01 5.36813319e-01
9.40834284e-01 1.80307060e-01 7.51962662e-01 -4.57942672e-02
7.80574203e-01 1.20132573e-01 4.45535004e-01 7.81529188e-01
3.76608148e-02 5.59359133e-01 6.71715319e-01 8.55158493e-02
-1.06814110e+00 -6.18542016e-01 2.55405933e-01 7.82734573e-01
1.39293835e-01 -6.38213873e-01 -1.10498846e+00 -8.05749774e-01
-4.85920012e-01 1.07016313e+00 -1.64381132e-01 -2.27435112e-01
-5.76953828e-01 -6.72886491e-01 6.41572356e-01 2.11265296e-01
1.06478620e+00 -1.26438224e+00 -5.15399873e-01 4.79631156e-01
-6.76178694e-01 -1.03678131e+00 -2.42850199e-01 -1.15870699e-01
-1.02910626e+00 -6.83844566e-01 -7.90302992e-01 -8.72937560e-01
4.83000457e-01 1.70755178e-01 8.68397832e-01 -7.34490082e-02
2.95999289e-01 1.42513871e-01 -7.10261583e-01 -4.64560211e-01
-1.01328635e+00 5.59515536e-01 -4.65117618e-02 -9.17225331e-02
-1.83931850e-02 -5.13244569e-01 -4.50445473e-01 -3.11488897e-01
-9.07459974e-01 5.36568403e-01 6.59098983e-01 1.09636557e+00
1.23657212e-01 6.84548169e-02 1.41579807e+00 -1.18562007e+00
1.59055066e+00 -4.42768574e-01 2.92014003e-01 5.01120090e-01
-4.76701528e-01 2.22595796e-01 9.13051963e-01 1.27880853e-02
-1.67652774e+00 -3.05678487e-01 -2.00814575e-01 2.34522790e-01
-1.86203688e-01 7.15905726e-01 -2.39639819e-01 6.92966580e-01
5.07909417e-01 8.37385178e-01 2.92225689e-01 -4.70115870e-01
4.12039399e-01 1.19482791e+00 6.39440179e-01 -3.66519272e-01
3.17864060e-01 -5.42888865e-02 -2.55317837e-01 -1.16048861e+00
-9.74130690e-01 -2.80552417e-01 -5.92418194e-01 -9.99304876e-02
8.06804061e-01 -5.99849820e-01 -1.68269083e-01 3.77594799e-01
-1.55081260e+00 1.81245983e-01 -4.25094128e-01 3.84144187e-01
-8.19165111e-01 1.02555466e+00 -4.65537727e-01 -1.10970104e+00
-1.03340292e+00 -7.50203669e-01 7.78735578e-01 6.51001394e-01
-4.62048203e-01 -8.74845743e-01 -1.53761562e-02 3.44686151e-01
2.08621323e-01 2.86792725e-01 9.65313792e-01 -1.25576949e+00
-1.86753506e-03 -2.34783620e-01 2.81925574e-02 6.99157298e-01
5.76841891e-01 -1.56559393e-01 -6.95975244e-01 -4.31074557e-04
5.57804823e-01 -6.52800679e-01 8.78818095e-01 2.36019418e-02
7.30437338e-01 -9.41594183e-01 6.83155097e-03 -9.93109420e-02
8.71820331e-01 2.39978969e-01 6.63689494e-01 -7.71286115e-02
3.57318223e-01 1.01339042e+00 9.25642431e-01 6.27185225e-01
3.37691009e-01 3.33249390e-01 1.47263214e-01 3.82637322e-01
-7.49656260e-02 -5.06632626e-01 2.97066897e-01 1.54278445e+00
-3.62544656e-02 -5.76173484e-01 -3.52185398e-01 4.44946647e-01
-2.10760474e+00 -1.09956872e+00 2.11807601e-02 1.99309969e+00
1.27908790e+00 2.61124700e-01 -8.26494992e-02 1.69481203e-01
7.97093749e-01 6.20288432e-01 -5.21513879e-01 -6.88655496e-01
-2.17740163e-01 -4.16069403e-02 -1.35714710e-01 3.82559776e-01
-8.69454682e-01 9.48094964e-01 5.53746033e+00 1.05581212e+00
-6.33033991e-01 -1.08129062e-01 5.34030378e-01 2.79778987e-01
-1.84431463e-01 6.02749400e-02 -7.28618205e-01 5.01441479e-01
9.50927734e-01 -7.59643137e-01 1.11916266e-01 6.17858589e-01
2.21889943e-01 -1.91365808e-01 -1.12369359e+00 7.57674158e-01
5.72385907e-01 -1.24298775e+00 5.05724549e-01 -4.38493520e-01
8.00448954e-01 -7.25683868e-01 -3.97366464e-01 5.14023781e-01
-3.89911681e-02 -8.28507662e-01 3.32359970e-01 5.65746009e-01
5.28880358e-01 -7.49277174e-01 1.01635170e+00 1.02047169e+00
-7.61761248e-01 1.53810352e-01 -4.83967185e-01 -3.02114069e-01
3.02009314e-01 2.43463218e-01 -1.16571915e+00 1.12011218e+00
-9.21997651e-02 6.17800653e-01 -2.92071491e-01 5.80016017e-01
-2.97987252e-01 5.19626677e-01 7.44547322e-02 -4.29059476e-01
3.20009977e-01 -4.01847690e-01 8.78112197e-01 1.21173358e+00
2.77962357e-01 4.68465447e-01 1.41170293e-01 5.15567720e-01
-1.03699207e-01 4.84833419e-01 -7.92778432e-01 -3.08781743e-01
5.01971662e-01 1.09100556e+00 -4.05671775e-01 -6.76364958e-01
-1.14288300e-01 1.14761984e+00 1.57057047e-01 2.29631394e-01
-4.27464306e-01 -7.52017975e-01 -1.06498316e-01 -2.92822719e-01
-5.09881154e-02 -5.87322861e-02 -2.90970951e-01 -1.21124792e+00
1.02968365e-01 -9.64169383e-01 2.07397908e-01 -5.76807082e-01
-1.07845318e+00 8.08134794e-01 1.83699369e-01 -1.20283020e+00
-9.56111014e-01 1.39447391e-01 -9.77475822e-01 6.92731023e-01
-1.20607054e+00 -8.70453000e-01 -2.95567751e-01 2.25376636e-01
1.27962995e+00 -4.92692292e-01 1.00519478e+00 -3.10014933e-01
-4.49231803e-01 4.00355041e-01 2.86610454e-01 -5.66973053e-02
6.92410171e-01 -1.32399797e+00 2.32391194e-01 7.35074103e-01
-2.49248877e-01 4.18909371e-01 1.07131195e+00 -8.92875433e-01
-1.18137503e+00 -9.15690184e-01 1.13748109e+00 5.13343252e-02
3.61026555e-01 -1.57419279e-01 -8.43301475e-01 3.87192577e-01
8.52600634e-01 -9.78191376e-01 7.43440330e-01 -3.19473207e-01
4.55059499e-01 1.49177402e-01 -1.05063057e+00 7.92185128e-01
9.23205733e-01 -1.39358506e-01 -1.34834850e+00 4.50484157e-01
1.05315876e+00 -4.87126082e-01 -6.21867716e-01 4.53837097e-01
2.01160401e-01 -9.52305734e-01 4.16960657e-01 -6.65185273e-01
1.02540028e+00 1.42366245e-01 -4.69504371e-02 -1.59712911e+00
3.02325130e-01 -9.66243863e-01 -4.03439730e-01 1.61086953e+00
1.68542519e-01 -4.87987727e-01 6.88474298e-01 3.72704655e-01
-5.48455536e-01 -7.32163787e-01 -8.27033341e-01 -3.41021597e-01
-1.10192366e-01 3.62163097e-01 5.39181411e-01 6.10019922e-01
4.65446293e-01 1.13529503e+00 -6.34603381e-01 -4.22495574e-01
4.74681914e-01 2.02830017e-01 9.74619329e-01 -9.98800874e-01
-2.26703346e-01 -2.79795170e-01 -1.23782821e-01 -1.57046533e+00
2.45061472e-01 -6.43139482e-01 2.28608683e-01 -2.02374673e+00
2.14479268e-01 1.48223639e-02 3.13126624e-01 -1.19021565e-01
-4.49641585e-01 -4.34952050e-01 2.37849683e-01 2.15703249e-01
-7.35885024e-01 1.27046132e+00 1.32343233e+00 -1.84855238e-01
-5.22498727e-01 2.17272907e-01 -9.88613844e-01 6.95186615e-01
8.93509030e-01 -1.42613009e-01 -7.83629000e-01 -1.50911137e-01
-2.76170522e-01 7.81299651e-01 -2.42615446e-01 -8.42214644e-01
3.42414200e-01 -4.43334542e-02 3.29365171e-02 -9.69790518e-01
3.09036553e-01 -1.66441411e-01 -3.08652580e-01 2.83763707e-01
-8.64192367e-01 -9.52592567e-02 -1.87262461e-01 7.13484764e-01
-5.50154269e-01 -8.81221473e-01 3.68908256e-01 -5.58601141e-01
-3.64159584e-01 -7.65802339e-02 -3.94635558e-01 3.20981383e-01
6.40115738e-01 -2.09515408e-01 -4.97557521e-01 -9.13985550e-01
-5.23384988e-01 3.11071366e-01 6.70372397e-02 2.51168281e-01
7.98711479e-01 -1.14467502e+00 -1.15302002e+00 -2.31549114e-01
-2.01104477e-01 3.42893481e-01 3.48924041e-01 5.16511381e-01
-4.72756207e-01 6.73707962e-01 -2.33336508e-01 -2.24498346e-01
-1.17898643e+00 1.45266458e-01 -1.49421051e-01 -6.22096121e-01
-6.27006233e-01 4.29267675e-01 1.94973573e-01 -3.31210732e-01
2.86905468e-01 -1.23866135e-02 -6.64268434e-01 2.20235124e-01
6.48666918e-01 5.20648062e-01 5.11102472e-03 -5.29085457e-01
2.99739391e-01 -1.29876554e-01 -4.32382643e-01 -2.53022969e-01
1.16283250e+00 -4.71729279e-01 -2.62730569e-01 6.26157165e-01
8.92141461e-01 -1.34453073e-01 -8.88381004e-01 -3.64295751e-01
-4.85851392e-02 -1.65146977e-01 -4.10135001e-01 -5.31188607e-01
-3.93759131e-01 7.34633803e-01 -1.36660755e-01 6.67561293e-01
9.38612223e-01 -3.12457740e-01 1.07584357e+00 8.01240504e-01
1.27426133e-01 -1.28733003e+00 2.41973579e-01 6.35994971e-01
1.11342192e+00 -1.07146204e+00 2.60188401e-01 -3.46477717e-01
-9.91244078e-01 1.29551494e+00 6.53672874e-01 -4.87792529e-02
1.10240303e-01 -2.65106320e-01 -3.36970359e-01 9.38192680e-02
-9.12616432e-01 4.44085449e-02 -1.23103842e-01 4.72847641e-01
5.63605070e-01 -1.32743627e-01 -1.01564741e+00 9.33297038e-01
-6.03231132e-01 -1.15703471e-01 1.02428198e+00 9.98777509e-01
-8.27452660e-01 -9.85490382e-01 -7.76119903e-02 8.45689774e-01
-4.95928556e-01 -9.41677615e-02 -7.31430352e-01 5.75442076e-01
-3.43674958e-01 1.40332842e+00 -1.85202584e-01 -2.83730805e-01
3.12394977e-01 3.09134692e-01 3.32705945e-01 -1.02138221e+00
-5.82905054e-01 -2.90887319e-02 6.32511139e-01 1.44239634e-01
-6.32458448e-01 -6.70322120e-01 -1.23126781e+00 -9.78943408e-02
-4.92711335e-01 6.86241031e-01 4.80526656e-01 1.05554366e+00
1.96246594e-01 6.14627600e-01 8.51471841e-01 -7.74434030e-01
-1.12876296e+00 -1.54619598e+00 -5.21927118e-01 3.28627884e-01
1.20102122e-01 -1.93940416e-01 -2.76680350e-01 2.22595204e-02] | [12.494640350341797, 9.190692901611328] |
e180bf3c-dd5e-4b49-b9eb-0405cfd764fb | multi-agent-path-finding-with-awareness-for | 2009.09355 | null | https://arxiv.org/abs/2009.09355v1 | https://arxiv.org/pdf/2009.09355v1.pdf | Multi Agent Path Finding with Awareness for Spatially Extended Agents | Path finding problems involve identification of a plan for conflict free movement of agents over a common road network. Most approaches to this problem handle the agents as point objects, wherein the size of the agent is significantly smaller than the road on which it travels. In this paper, we consider spatially extended agents which have a size comparable to the length of the road on which they travel. An optimal multi agent path finding approach for spatially-extended agents was proposed in the eXtended Conflict Based Search (XCBS) algorithm. As XCBS resolves only a pair of conflicts at a time, it results in deeper search trees in case of cascading or multiple (more than two agent) conflicts at a given location. This issue is addressed in eXtended Conflict Based Search with Awareness (XCBS-A) in which an agent uses awareness of other agents' plans to make its own plan. In this paper, we explore XCBS-A in greater detail, we theoretically prove its completeness and empirically demonstrate its performance with other algorithms in terms of variances in road characteristics, agent characteristics and plan characteristics. We demonstrate the distributive nature of the algorithm by evaluating its performance when distributed over multiple machines. XCBS-A generates a huge search space impacting its efficiency in terms of memory; to address this we propose an approach for memory-efficiency and empirically demonstrate the performance of the algorithm. The nature of XCBS-A is such that it may lead to suboptimal solutions, hence the final contribution of this paper is an enhanced approach, XCBS-Local Awareness (XCBS-LA) which we prove will be optimal and complete. | ['Dipti Deodhare', 'Shyni Thomas', 'M. N. Murty'] | 2020-09-20 | null | null | null | null | ['multi-agent-path-finding'] | ['playing-games'] | [ 1.40082315e-01 1.80143252e-01 -3.24041806e-02 1.24467276e-01
-2.35547140e-01 -6.88577831e-01 7.35384226e-01 4.63238686e-01
-6.39109015e-01 1.23007882e+00 -1.46060407e-01 -4.74342555e-01
-7.92885244e-01 -1.10543752e+00 -3.83399934e-01 -6.42655909e-01
-6.19640052e-01 1.05730176e+00 9.25758958e-01 -4.59191710e-01
5.51728666e-01 8.07590723e-01 -1.23942733e+00 -2.60085136e-01
8.22329760e-01 3.47424328e-01 4.14220333e-01 7.35155880e-01
-6.54525608e-02 5.49091339e-01 -7.31417716e-01 1.01312950e-01
4.61975515e-01 -3.27639699e-01 -1.31994879e+00 -5.33337565e-03
-1.23423316e-01 -6.64951056e-02 1.08152322e-01 7.34569371e-01
1.05650820e-01 3.38563383e-01 5.87276042e-01 -1.85755599e+00
3.68643820e-01 4.75240707e-01 -7.25253165e-01 4.22895879e-01
4.58513528e-01 5.50752245e-02 7.04556882e-01 -2.14350745e-01
7.46521294e-01 1.07249618e+00 3.95444036e-01 1.29559964e-01
-1.14926004e+00 -2.80755252e-01 5.43370128e-01 4.46811229e-01
-1.55967665e+00 -2.45007947e-01 3.62193257e-01 -1.24852844e-01
1.37696385e+00 6.11956060e-01 5.68568885e-01 2.64190525e-01
2.54504025e-01 2.42505103e-01 1.15407801e+00 -5.35884619e-01
6.52476847e-01 -2.01107010e-01 6.24959404e-03 5.34925818e-01
5.41644514e-01 3.48582119e-01 -2.56637156e-01 -3.06474626e-01
8.27808022e-01 -4.10976440e-01 -8.45729634e-02 -5.09253561e-01
-1.32452476e+00 7.46752203e-01 4.19506431e-01 4.85098451e-01
-6.24247670e-01 1.66142672e-01 1.39854386e-01 4.60199028e-01
-1.00320630e-01 5.93492627e-01 -2.22731218e-01 -1.27308935e-01
-5.95505834e-01 5.62648892e-01 9.79520917e-01 9.08020973e-01
9.63384688e-01 -3.59295130e-01 3.29575211e-01 4.00624573e-01
4.54851352e-02 2.46735275e-01 -5.95713295e-02 -1.00549161e+00
5.56381345e-01 7.57331908e-01 4.44085360e-01 -1.21919453e+00
-9.48899806e-01 -2.58584559e-01 -5.76355815e-01 7.74377823e-01
4.67850864e-01 -1.46084219e-01 -5.53750157e-01 1.90635312e+00
5.56459248e-01 1.81473807e-01 4.17857170e-01 8.33269536e-01
6.70374259e-02 7.46854782e-01 -5.85325658e-02 -5.22648752e-01
1.18461978e+00 -1.27831352e+00 -3.54808688e-01 -3.36882472e-01
7.87963033e-01 -4.61445004e-01 3.61921340e-01 2.04333752e-01
-1.16386461e+00 7.78150037e-02 -1.01877844e+00 7.33340025e-01
-5.79993248e-01 -5.47138631e-01 5.24413943e-01 5.29274285e-01
-1.42783797e+00 2.93832928e-01 -8.72466207e-01 -7.92638898e-01
-1.38184890e-01 7.39108503e-01 -3.11142713e-01 1.28900126e-01
-9.44523215e-01 1.22153509e+00 4.48402435e-01 -1.79251865e-01
-6.25128806e-01 -4.36091870e-02 -4.27593350e-01 5.70778735e-02
1.02007675e+00 -6.27653539e-01 1.01273799e+00 -8.20426166e-01
-1.24978364e+00 2.40427703e-01 -2.46801630e-01 -3.93768013e-01
4.32358772e-01 4.40277487e-01 -4.00409997e-01 5.95019795e-02
4.93910760e-01 4.77626652e-01 1.93997294e-01 -1.38503456e+00
-1.13844514e+00 -3.02487046e-01 3.79985482e-01 4.79260892e-01
3.14251840e-01 1.03479743e-01 -3.85312915e-01 -2.36794159e-01
1.45859763e-01 -1.25302505e+00 -7.99346566e-01 -6.59099221e-01
-3.32875848e-01 -4.06272620e-01 5.32539964e-01 1.83050439e-01
1.31669402e+00 -1.62464833e+00 2.10085273e-01 8.20338011e-01
8.42390060e-02 1.40817493e-01 -4.11870867e-01 1.28564322e+00
2.63310850e-01 1.12017460e-01 -2.19863117e-01 6.09509386e-02
-1.15063034e-01 5.78177333e-01 1.92634277e-02 4.76731330e-01
-1.55839667e-01 4.66256261e-01 -8.49041998e-01 -5.15073299e-01
-5.36768176e-02 -1.84017569e-01 -3.96260828e-01 -2.74575114e-01
-1.34013668e-01 -6.50824457e-02 -6.56691313e-01 4.51048106e-01
8.10931981e-01 -1.34828061e-01 3.95262599e-01 3.58969390e-01
-8.01164865e-01 1.78948328e-01 -1.60193300e+00 1.21806467e+00
-6.17485754e-02 3.69935840e-01 3.57266128e-01 -7.90577054e-01
7.05787301e-01 2.13117242e-01 5.74594200e-01 -6.44466400e-01
-3.01735640e-01 3.21666002e-01 9.25560445e-02 -1.06681265e-01
5.98745823e-01 1.07839487e-01 -2.43561491e-02 8.93722475e-01
-9.27722275e-01 2.05909416e-01 6.15816295e-01 3.23862612e-01
1.70812047e+00 -4.30341274e-01 5.18046498e-01 -4.64580715e-01
6.86038375e-01 6.94360793e-01 4.64660466e-01 1.09416890e+00
-4.05511588e-01 -2.50333667e-01 3.72024953e-01 -4.64181989e-01
-7.10282385e-01 -9.21925724e-01 2.95689344e-01 8.45623016e-01
9.31640446e-01 -5.88838816e-01 -5.89513719e-01 -6.04277849e-01
-1.06665134e-01 5.75332522e-01 -3.88518244e-01 3.84114772e-01
-9.88288701e-01 -5.51951408e-01 3.68719608e-01 1.21349290e-01
5.81794620e-01 -9.42720592e-01 -1.46895826e+00 5.47913909e-01
-8.44228193e-02 -7.98620462e-01 -2.20262393e-01 3.89037132e-02
-5.93705654e-01 -1.40540981e+00 -1.98502764e-01 -6.66390836e-01
8.62133384e-01 6.14462197e-01 6.94997132e-01 3.40607435e-01
8.97560492e-02 4.84776586e-01 -4.47397083e-01 -2.06378505e-01
-2.96248585e-01 2.48248637e-01 9.19454396e-02 -3.82725537e-01
1.77481517e-01 -5.43585122e-01 -4.40138727e-01 7.68229723e-01
-7.20990002e-01 8.25870112e-02 6.19403005e-01 4.21642870e-01
3.23282182e-01 8.85767817e-01 6.26220345e-01 -3.94961059e-01
9.21205699e-01 -5.39768875e-01 -9.59073544e-01 3.71721357e-01
-8.32553148e-01 -8.40170383e-02 2.52511382e-01 -1.53392002e-01
-7.59325624e-01 6.97545037e-02 4.46925581e-01 2.62747169e-01
-2.03864977e-01 7.61715233e-01 2.26489708e-01 -2.65481234e-01
3.15970689e-01 -4.32633273e-02 1.63021564e-01 -1.09139383e-01
-8.97068717e-03 2.50897437e-01 2.57873446e-01 -4.12038237e-01
6.56013250e-01 5.29068649e-01 5.63339770e-01 -7.25590408e-01
2.66822368e-01 -4.06170636e-01 -4.18755531e-01 -3.63193065e-01
4.59615707e-01 -3.71434748e-01 -1.03171122e+00 2.52911061e-01
-1.15347922e+00 -4.83166099e-01 1.38214290e-01 3.52571607e-01
-6.54729962e-01 1.48954794e-01 -2.39434421e-01 -1.03940821e+00
3.07288170e-01 -1.18935728e+00 4.17095751e-01 1.02280699e-01
-4.20469910e-01 -9.22766030e-01 5.95849633e-01 1.74593041e-03
4.58800316e-01 3.74332815e-01 7.93566644e-01 -7.26867318e-01
-9.09791946e-01 2.53300756e-01 -1.20325185e-01 -8.28193009e-01
6.37322739e-02 -2.34025910e-01 1.37650341e-01 -5.20552874e-01
-4.07697886e-01 2.41154388e-01 3.31160277e-01 3.38218540e-01
1.61894765e-02 -4.90819395e-01 -1.05525076e+00 -1.75463408e-01
1.74082625e+00 7.85187483e-01 4.90235895e-01 1.16436970e+00
1.74118057e-02 1.12974644e+00 9.83015418e-01 3.78359467e-01
5.60792744e-01 1.05393612e+00 7.86770463e-01 -1.09909154e-01
2.65223563e-01 3.53389919e-01 2.52138644e-01 2.64314741e-01
-2.32642874e-01 -5.41330159e-01 -1.09866786e+00 7.56571174e-01
-2.22741199e+00 -1.05650115e+00 -3.25162113e-01 1.99299908e+00
2.34314978e-01 2.81165703e-03 4.59423155e-01 1.36940658e-01
9.37278986e-01 -1.39916345e-01 -3.37751180e-01 -7.86210895e-01
-2.21026037e-02 -3.35120380e-01 8.46634328e-01 1.07770705e+00
-6.93827450e-01 8.74714673e-01 6.48280430e+00 5.30171096e-01
-7.56459594e-01 9.43232030e-02 8.12085047e-02 -2.34947324e-01
-1.12957075e-01 2.90187091e-01 -5.41528463e-01 2.24582076e-01
7.01938629e-01 -1.25897318e-01 6.25471354e-01 3.17786038e-01
5.12706459e-01 -8.18881392e-01 -8.38038564e-01 4.51828986e-01
-2.77051002e-01 -1.32993853e+00 -2.47918293e-01 5.96813977e-01
6.83111906e-01 1.63286522e-01 -2.97271848e-01 -1.20456085e-01
5.08237243e-01 -9.28483248e-01 5.86901486e-01 1.57530571e-03
8.95721242e-02 -1.09181988e+00 6.31820321e-01 6.36208773e-01
-1.33559382e+00 -2.63016015e-01 -4.77370024e-02 -5.54519892e-01
6.38382673e-01 -6.27572536e-02 -9.32259977e-01 8.82589042e-01
5.54929078e-01 -5.85609488e-02 -2.11296365e-01 1.13361168e+00
1.67436510e-01 2.81632841e-02 -7.31130660e-01 -2.39737734e-01
8.11445415e-01 -3.97282451e-01 9.74240065e-01 8.68009508e-01
3.57954293e-01 4.72976208e-01 4.75156516e-01 6.55323923e-01
7.47127891e-01 9.68683325e-03 -5.70085824e-01 3.57071131e-01
9.05342758e-01 7.45820820e-01 -1.20624673e+00 -1.91881433e-01
-2.61078447e-01 5.43151319e-01 3.00920039e-01 4.58041072e-01
-6.24713540e-01 -3.02653998e-01 5.79419374e-01 2.13043705e-01
1.93383530e-01 -3.06158692e-01 -5.15655763e-02 -3.07842940e-01
-1.28918126e-01 -7.79626906e-01 4.98256713e-01 -4.72442687e-01
-6.59717977e-01 7.86520362e-01 4.26359057e-01 -9.03988421e-01
-4.52857226e-01 -1.48087010e-01 -7.33735919e-01 7.91785717e-01
-1.46446693e+00 -8.48062873e-01 -1.37439623e-01 7.16489434e-01
5.19861758e-01 -2.10247591e-01 6.06935024e-01 -2.97467262e-02
-5.39043784e-01 7.29178190e-02 -3.19841430e-02 -5.24991810e-01
1.70540050e-01 -8.92835855e-01 1.96186841e-01 1.05243528e+00
-9.93978530e-02 5.53075314e-01 9.10071492e-01 -8.88322115e-01
-1.19884288e+00 -6.11229837e-01 1.07813263e+00 4.82509099e-03
5.57719648e-01 2.68465221e-01 -4.73717839e-01 7.30011404e-01
3.91601175e-01 -6.83296323e-01 2.81004816e-01 4.41343337e-02
3.40418130e-01 -2.27236357e-02 -1.16450500e+00 9.40481544e-01
1.01431954e+00 1.37870640e-01 -3.54514688e-01 8.94881561e-02
2.23985210e-01 -1.26348101e-02 -3.70660514e-01 4.68993276e-01
1.81142315e-01 -1.40723681e+00 7.25359440e-01 -1.51554614e-01
-2.31733099e-01 -7.98679769e-01 6.06704205e-02 -1.36733758e+00
-4.67583537e-01 -7.38727331e-01 5.01785994e-01 7.81212449e-01
5.73598802e-01 -1.17887008e+00 9.16768253e-01 4.96154845e-01
-1.02174871e-01 -7.87370265e-01 -1.41600049e+00 -1.13165319e+00
-1.29699618e-01 -1.41075552e-01 9.59649503e-01 9.28371072e-01
4.92141485e-01 1.12366349e-01 -3.92375514e-02 6.52472913e-01
5.41079044e-01 3.13182503e-01 9.94139791e-01 -1.10830057e+00
-2.07226872e-01 -6.81302965e-01 -2.85317361e-01 -9.26832616e-01
-4.57658693e-02 -5.01234949e-01 -1.09373964e-01 -1.80795538e+00
-1.52603805e-01 -1.04881585e+00 -2.47468837e-02 4.94899750e-01
3.93729210e-01 -2.34657779e-01 2.93477744e-01 5.16367793e-01
-7.09462762e-01 -5.41046187e-02 8.98665369e-01 -3.45047773e-03
-6.11683488e-01 1.27248377e-01 -1.96396768e-01 5.35232663e-01
9.65816617e-01 -5.36633551e-01 -7.03404963e-01 -2.31100649e-01
2.99122512e-01 4.48603779e-01 2.36276448e-01 -9.32568312e-01
8.21796358e-01 -6.80869162e-01 -5.84801316e-01 -7.58566201e-01
4.06311214e-01 -1.14528883e+00 8.07270050e-01 1.01111734e+00
-1.35708123e-01 6.75120056e-01 1.28031790e-01 6.00277185e-01
-2.05757573e-01 -5.08560956e-01 2.47619137e-01 -1.53627232e-01
-8.70619833e-01 -1.53570488e-01 -9.68925953e-01 -5.57178617e-01
1.72919536e+00 -6.58576906e-01 -7.09070802e-01 -4.60955560e-01
-5.39426208e-01 7.29769349e-01 7.54589558e-01 -6.26076828e-04
5.08695364e-01 -9.96805668e-01 -5.90182304e-01 -1.38199672e-01
-1.10894382e-01 -1.80162743e-01 3.12625654e-02 1.10586214e+00
-8.51509392e-01 7.57488549e-01 -4.47170287e-01 -2.23672405e-01
-1.59690475e+00 7.97518134e-01 2.79344082e-01 -5.95357955e-01
-5.47368944e-01 3.69963527e-01 1.49128184e-01 2.50102654e-02
-2.22056620e-02 6.15885593e-02 -3.63843501e-01 -1.60365880e-01
3.96183908e-01 9.85773921e-01 -3.44372362e-01 -7.40580678e-01
-8.16636682e-01 6.81038201e-01 7.79667636e-03 -6.27093256e-01
1.15044141e+00 -5.38367867e-01 -5.55419743e-01 -2.59007484e-01
4.09384996e-01 2.19197229e-01 -1.00771224e+00 9.81779099e-02
3.88662159e-01 -4.58511859e-01 -4.69924621e-02 -9.36820567e-01
-6.98417366e-01 -1.58794627e-01 2.34565258e-01 7.06763983e-01
1.15456140e+00 4.61997911e-02 3.86146188e-01 4.81047541e-01
8.37836504e-01 -1.11838496e+00 -1.75061390e-01 5.67978680e-01
7.23738790e-01 -8.85681510e-01 1.86654270e-01 -7.76897430e-01
-5.12769103e-01 8.71055484e-01 6.00925148e-01 6.18148670e-02
2.33467504e-01 3.06197315e-01 -4.82259691e-02 -4.80078459e-01
-9.57512081e-01 -4.84004200e-01 -4.58762437e-01 7.81093001e-01
-6.46795988e-01 1.34182930e-01 -7.49819994e-01 -6.67848140e-02
-1.14934325e-01 -2.34724581e-01 7.91801333e-01 1.37674248e+00
-7.84451902e-01 -1.35538256e+00 -6.65709019e-01 -1.39185667e-01
8.74527469e-02 2.08651870e-01 -5.68307042e-01 1.10019016e+00
6.94937557e-02 1.53351736e+00 3.10356885e-01 -1.43458262e-01
3.08830500e-01 -5.25172770e-01 3.39296997e-01 -2.25911126e-01
-6.33456945e-01 1.61504596e-02 6.54828966e-01 -6.31142974e-01
-7.53139555e-01 -6.50194585e-01 -1.58483374e+00 -6.61152720e-01
-2.00380728e-01 6.59606576e-01 5.86331367e-01 9.06287432e-01
4.44470853e-01 2.52410740e-01 4.72179979e-01 -6.21019185e-01
-1.68234155e-01 -3.86317432e-01 -4.69875455e-01 -2.71198601e-01
2.06140667e-01 -8.21453929e-01 -2.11290583e-01 -5.97280383e-01] | [4.9178571701049805, 1.8346904516220093] |
f1d259fc-ae4f-4389-b21d-441cd2261930 | tracking-by-animation-unsupervised-learning | 1809.03137 | null | http://arxiv.org/abs/1809.03137v3 | http://arxiv.org/pdf/1809.03137v3.pdf | Tracking by Animation: Unsupervised Learning of Multi-Object Attentive Trackers | Online Multi-Object Tracking (MOT) from videos is a challenging computer
vision task which has been extensively studied for decades. Most of the
existing MOT algorithms are based on the Tracking-by-Detection (TBD) paradigm
combined with popular machine learning approaches which largely reduce the
human effort to tune algorithm parameters. However, the commonly used
supervised learning approaches require the labeled data (e.g., bounding boxes),
which is expensive for videos. Also, the TBD framework is usually suboptimal
since it is not end-to-end, i.e., it considers the task as detection and
tracking, but not jointly. To achieve both label-free and end-to-end learning
of MOT, we propose a Tracking-by-Animation framework, where a differentiable
neural model first tracks objects from input frames and then animates these
objects into reconstructed frames. Learning is then driven by the
reconstruction error through backpropagation. We further propose a
Reprioritized Attentive Tracking to improve the robustness of data association.
Experiments conducted on both synthetic and real video datasets show the
potential of the proposed model. Our project page is publicly available at:
https://github.com/zhen-he/tracking-by-animation | ['Hangen He', 'David Barber', 'Jian Li', 'Zhen He', 'Daxue Liu'] | 2018-09-10 | tracking-by-animation-unsupervised-learning-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/He_Tracking_by_Animation_Unsupervised_Learning_of_Multi-Object_Attentive_Trackers_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/He_Tracking_by_Animation_Unsupervised_Learning_of_Multi-Object_Attentive_Trackers_CVPR_2019_paper.pdf | cvpr-2019-6 | ['online-multi-object-tracking'] | ['computer-vision'] | [ 1.32978996e-02 -4.72643793e-01 -2.18256593e-01 -1.42247841e-01
-5.58862448e-01 -3.05706143e-01 3.33340287e-01 -1.29308328e-01
-5.48470378e-01 5.52878797e-01 -2.10248724e-01 -1.07618295e-01
1.04177430e-01 -4.58957553e-01 -9.84199822e-01 -8.32623482e-01
2.09765032e-01 3.20745379e-01 7.35297322e-01 2.39794597e-01
7.51662813e-03 1.26441360e-01 -1.51981604e+00 -1.54063150e-01
6.86242223e-01 9.61243689e-01 4.50352997e-01 5.84224880e-01
1.69715118e-02 9.85176384e-01 -1.68999389e-01 -1.60973817e-01
3.43528599e-01 -3.94041926e-01 -3.49850893e-01 1.86499968e-01
5.43139696e-01 -4.58227932e-01 -4.67685223e-01 1.33201098e+00
4.28184360e-01 1.94476292e-01 2.95672178e-01 -1.50303662e+00
-5.68875849e-01 2.94982970e-01 -7.68986285e-01 2.31499165e-01
3.92407849e-02 3.72251898e-01 8.29309762e-01 -8.49260688e-01
4.64652926e-01 1.31851089e+00 6.19141459e-01 7.56695032e-01
-1.03548777e+00 -9.61357236e-01 3.91837448e-01 2.55081475e-01
-1.35091627e+00 -4.25864428e-01 8.04972053e-01 -6.87910140e-01
1.72785655e-01 2.40843650e-02 7.65943050e-01 8.64814043e-01
-1.00677460e-01 1.11213088e+00 8.51397514e-01 -2.23401561e-01
-5.25658429e-02 3.99426967e-02 -2.97894403e-02 9.29693937e-01
2.26194188e-01 4.27337021e-01 -2.77469575e-01 1.68993562e-01
9.85000908e-01 3.91883582e-01 -2.95470804e-01 -5.77944696e-01
-1.38541687e+00 5.59442043e-01 5.59987903e-01 3.86989899e-02
-3.26310068e-01 4.06985104e-01 3.20764124e-01 7.91442171e-02
3.05761725e-01 -3.06138396e-01 -2.31038377e-01 1.36794016e-01
-8.85113597e-01 3.02286148e-01 2.78598666e-01 9.96302009e-01
7.75335431e-01 1.40267700e-01 -1.80305108e-01 6.33886456e-01
7.47927845e-01 6.22601271e-01 3.32519770e-01 -9.58036184e-01
3.66729379e-01 4.60092813e-01 3.84613574e-01 -1.07286263e+00
-3.32997739e-01 -2.74159282e-01 -6.84545696e-01 3.70251328e-01
6.04742527e-01 -3.36127728e-01 -7.17186630e-01 1.83726871e+00
8.15659463e-01 7.21096575e-01 -1.48746252e-01 1.20778191e+00
7.34591901e-01 7.36457109e-01 2.10156530e-01 -3.41000229e-01
1.13276136e+00 -1.38761127e+00 -8.20454299e-01 -1.58381492e-01
5.23321033e-01 -6.32142961e-01 7.78812587e-01 2.94315100e-01
-8.99588764e-01 -9.01416779e-01 -7.97533631e-01 1.23867109e-01
2.25622999e-03 2.09810391e-01 1.83644757e-01 3.20024908e-01
-6.95199668e-01 4.20240164e-01 -1.16338336e+00 -2.07804158e-01
6.62039816e-01 4.22638953e-01 -6.08963054e-03 3.08439266e-02
-7.93823957e-01 5.04122019e-01 5.64553559e-01 1.74366221e-01
-1.14441371e+00 -4.31740493e-01 -6.82331145e-01 -2.57238537e-01
7.29334712e-01 -6.65936708e-01 1.29695213e+00 -1.12651491e+00
-1.52343738e+00 4.55347329e-01 -1.09595582e-01 -3.14669222e-01
8.82936537e-01 -5.10457337e-01 -2.77120709e-01 2.45008916e-02
1.18124448e-02 7.53191948e-01 9.77035463e-01 -1.23073101e+00
-1.05447304e+00 -6.22737072e-02 3.39507684e-02 1.21941172e-01
-5.57689250e-01 1.84676245e-01 -7.82426715e-01 -6.73126817e-01
-1.41199917e-01 -1.15854931e+00 -1.01148203e-01 5.61470091e-01
-2.24590316e-01 -2.38167167e-01 1.33552969e+00 -5.73455274e-01
1.23264360e+00 -2.12866998e+00 1.01262614e-01 -3.38202477e-01
1.79800391e-01 5.01858652e-01 1.92977972e-02 1.96310189e-02
1.92595765e-01 -3.67355824e-01 -6.40426110e-03 -4.53019321e-01
-6.71096295e-02 -4.98690121e-02 -4.47240286e-02 7.65358746e-01
5.84096573e-02 7.42617428e-01 -1.06934321e+00 -8.45095694e-01
6.25448823e-01 5.53196490e-01 -5.26132107e-01 3.85481149e-01
-5.66190422e-01 8.70841324e-01 -5.92100441e-01 6.48595572e-01
5.93332827e-01 -5.01287937e-01 -1.43870935e-01 -2.16048747e-01
-4.27638441e-01 -2.79990882e-01 -1.40754139e+00 1.52467179e+00
-5.69594279e-02 4.91757989e-01 6.19324408e-02 -8.78986239e-01
6.72413349e-01 3.25178534e-01 7.02164233e-01 -3.66650701e-01
2.08770841e-01 1.73104763e-01 3.15594077e-02 -5.73677242e-01
2.44926184e-01 -1.09560629e-02 1.91475585e-01 2.07739875e-01
-2.09327862e-01 5.48611701e-01 2.74327546e-01 -5.15211858e-02
7.81710744e-01 6.18049562e-01 1.75503612e-01 9.26971734e-02
6.87092066e-01 1.37242407e-01 1.13711929e+00 4.65144545e-01
-4.62544620e-01 4.10529673e-01 -1.64919849e-02 -3.75284344e-01
-9.58402216e-01 -6.58509076e-01 3.46438177e-02 1.13659334e+00
6.68014467e-01 -1.06485769e-01 -6.21698678e-01 -7.50402451e-01
3.36416811e-02 3.30689460e-01 -3.83487403e-01 1.38586853e-02
-7.60398269e-01 -4.19572800e-01 2.89875597e-01 5.70594430e-01
4.81396019e-01 -1.19322920e+00 -7.57255793e-01 3.47979873e-01
-1.40161887e-01 -1.23763072e+00 -7.77614832e-01 -2.46150047e-01
-7.89290249e-01 -1.14046752e+00 -8.91387463e-01 -8.75081718e-01
6.74113095e-01 4.41273779e-01 5.57980478e-01 4.27733839e-01
-9.22154915e-03 1.65933862e-01 -3.07301998e-01 -3.02467227e-01
-2.94568360e-01 -2.44501203e-01 3.06734294e-01 3.47444475e-01
1.83660597e-01 -2.50235915e-01 -6.80336416e-01 6.40472472e-01
-7.17686236e-01 3.12864870e-01 7.32565522e-01 7.52864420e-01
8.83872151e-01 -2.62178462e-02 5.09306908e-01 -7.30240405e-01
-7.11020604e-02 -3.81111056e-01 -1.10898352e+00 2.48322770e-01
-3.93671751e-01 -3.13349336e-01 7.23646939e-01 -7.64772117e-01
-8.79036248e-01 5.61156631e-01 7.42771998e-02 -1.24883568e+00
-1.96579799e-01 1.21004254e-01 -1.63874954e-01 -3.54045480e-01
1.00929581e-01 4.02027398e-01 -1.14150085e-02 -3.11670899e-01
2.01011062e-01 5.41378796e-01 5.41674018e-01 -3.66052330e-01
9.49960411e-01 4.71945643e-01 -1.09759644e-01 -5.94133556e-01
-7.95353472e-01 -5.15120924e-01 -5.83898485e-01 -7.93682694e-01
9.98162091e-01 -1.09415925e+00 -9.13025975e-01 5.62329710e-01
-1.04781115e+00 -3.73103708e-01 1.98347643e-01 6.92188025e-01
-5.95813453e-01 4.06071931e-01 -3.79341632e-01 -9.38875496e-01
-3.06683421e-01 -1.22087908e+00 9.38851058e-01 5.82143605e-01
2.28716299e-01 -9.90613401e-01 -2.85947435e-02 3.04348290e-01
1.61689222e-01 3.77322495e-01 3.52541775e-01 -3.14598799e-01
-1.03143740e+00 -1.09650113e-01 -2.58681387e-01 1.95752770e-01
7.26422295e-02 9.73396152e-02 -7.87894011e-01 -5.04741788e-01
-2.01349780e-01 -3.66439939e-01 7.31172383e-01 5.12047052e-01
1.15411294e+00 -3.61211151e-01 -6.58754289e-01 6.26450121e-01
1.47685266e+00 2.37107411e-01 1.58395901e-01 3.73317391e-01
1.01726127e+00 3.94256294e-01 1.05572188e+00 4.05943662e-01
4.48835045e-01 7.91721404e-01 6.40202880e-01 1.80171393e-02
-2.27715075e-01 -3.04234684e-01 4.53331023e-01 8.32703233e-01
-2.65820306e-02 -2.86553264e-01 -8.24821889e-01 4.95595187e-01
-2.35372400e+00 -9.63027894e-01 -1.46389619e-01 2.08794069e+00
6.78639472e-01 2.57787585e-01 4.52091485e-01 -1.63530484e-01
1.12726927e+00 1.18549034e-01 -8.10617924e-01 5.19326448e-01
1.97505638e-01 -4.86670196e-01 4.34333593e-01 3.33593160e-01
-1.37370777e+00 9.33497310e-01 4.48905659e+00 7.06097662e-01
-1.20350337e+00 2.10031062e-01 9.28833559e-02 -1.09900370e-01
3.20925504e-01 -2.11208388e-02 -9.99207973e-01 8.18217278e-01
5.57115674e-01 -2.78831068e-02 2.67576069e-01 8.80689502e-01
4.05388147e-01 2.11202785e-01 -1.05992162e+00 1.10680592e+00
-1.81025386e-01 -1.19865274e+00 -8.06784928e-02 -1.55948102e-01
5.88067174e-01 -3.09533384e-02 -4.27518748e-02 3.67267400e-01
2.07810804e-01 -4.16084617e-01 9.65679884e-01 6.98899090e-01
5.28175771e-01 -5.18802583e-01 3.91151428e-01 6.39004767e-01
-1.62306345e+00 -2.33345509e-01 -3.59668761e-01 2.05010191e-01
3.62068266e-01 1.95674762e-01 -4.24325347e-01 4.52986419e-01
7.93387294e-01 1.14621329e+00 -4.29226160e-01 1.49748492e+00
-9.19467881e-02 5.06800592e-01 -2.56780118e-01 -1.61969736e-01
2.11753413e-01 -1.78708658e-01 7.00113595e-01 1.14266121e+00
2.03143984e-01 1.40133956e-02 7.75547743e-01 6.91945016e-01
-4.98664342e-02 -6.56250045e-02 -3.65772814e-01 4.28956840e-03
5.43756306e-01 1.44184065e+00 -7.48692632e-01 -2.96309888e-01
-6.56220973e-01 6.19001627e-01 2.42816657e-01 1.67597547e-01
-1.17095339e+00 -1.90962166e-01 3.92929465e-01 7.70740733e-02
6.89507186e-01 -1.05127662e-01 3.00206363e-01 -1.09374321e+00
1.85498223e-02 -7.83751249e-01 4.33789581e-01 -5.77360630e-01
-1.14867210e+00 3.52473468e-01 -1.75057128e-01 -1.76931071e+00
4.90130484e-02 -3.31380904e-01 -4.85724062e-01 3.98226142e-01
-1.54160988e+00 -1.21187925e+00 -5.93547583e-01 7.00848639e-01
6.72204971e-01 7.86263794e-02 3.34198207e-01 7.99826562e-01
-9.68232334e-01 4.94048476e-01 1.88343689e-01 3.84962648e-01
7.45179296e-01 -9.66071725e-01 3.70212682e-02 9.12038803e-01
3.92631739e-02 2.79928118e-01 6.38791621e-01 -7.01869547e-01
-1.42246854e+00 -1.61542535e+00 2.07342833e-01 -2.66491711e-01
6.25885963e-01 -2.58872092e-01 -1.00352657e+00 8.21940005e-01
1.87134221e-01 5.44157326e-01 2.52242118e-01 -4.48884815e-01
1.02371000e-01 -1.75445363e-01 -8.16621959e-01 5.26490927e-01
1.03597355e+00 -8.77087787e-02 -3.00515473e-01 4.68549013e-01
6.88476801e-01 -5.82313716e-01 -6.47596478e-01 3.44183862e-01
5.77048182e-01 -6.96489811e-01 7.90340781e-01 -4.26935196e-01
1.08207725e-01 -9.26441193e-01 -1.77333374e-02 -8.59116256e-01
-3.67910087e-01 -5.67694962e-01 -5.62929094e-01 1.33521140e+00
-1.46048246e-02 -3.57717663e-01 7.92269886e-01 1.97843701e-01
-7.28760511e-02 -8.32771361e-01 -7.51416206e-01 -9.40290689e-01
-3.16542655e-01 -1.78774610e-01 2.23638400e-01 8.10141802e-01
-5.70749462e-01 2.26311371e-01 -7.98359096e-01 4.82993990e-01
1.07934332e+00 3.04713696e-01 1.03622556e+00 -1.12688386e+00
-3.48692477e-01 -3.98062974e-01 -3.20239246e-01 -1.30049014e+00
9.02354792e-02 -7.04695880e-01 2.73936331e-01 -1.34411657e+00
2.08986565e-01 -5.88797748e-01 -4.09175783e-01 4.27504212e-01
-3.73654455e-01 8.50880221e-02 3.68746877e-01 6.52426779e-01
-1.23214316e+00 6.12318456e-01 1.33515620e+00 -9.70501080e-02
-1.02575988e-01 1.47440612e-01 -2.20824823e-01 9.19669926e-01
5.76450169e-01 -8.49907398e-01 -9.38900784e-02 -6.10742450e-01
-3.58598053e-01 3.55231702e-01 6.35729194e-01 -1.22410452e+00
6.53291106e-01 -3.39092135e-01 3.06836069e-01 -8.47461104e-01
4.21569407e-01 -1.00547802e+00 2.36469224e-01 7.87219703e-01
-1.49418131e-01 7.56478403e-04 8.66162777e-02 9.95677412e-01
-3.65739167e-02 -7.29209185e-02 1.07248759e+00 -2.00641416e-02
-8.91805530e-01 7.24104822e-01 1.09797336e-01 2.08852421e-02
1.35060775e+00 -8.31247643e-02 -2.16270939e-01 -1.22139581e-01
-4.77868229e-01 6.14112735e-01 4.24756974e-01 4.46235001e-01
5.70444524e-01 -1.43825161e+00 -6.97992086e-01 -1.07201673e-01
1.33123538e-02 1.60666928e-01 2.30660394e-01 1.14314568e+00
-3.94377738e-01 2.20440730e-01 -1.29371881e-01 -9.77091432e-01
-1.54301548e+00 9.44629192e-01 3.53340447e-01 6.45309761e-02
-8.35028291e-01 7.10673273e-01 3.23115975e-01 -1.68368340e-01
6.06083989e-01 1.66110266e-02 -2.91784495e-01 -1.99677214e-01
5.49598634e-01 3.34707737e-01 -4.80351061e-01 -7.08828807e-01
-4.67721164e-01 7.65112281e-01 -2.14430302e-01 1.66935042e-01
1.20536280e+00 -2.27397114e-01 2.64991254e-01 5.37212551e-01
9.53865886e-01 -2.82826364e-01 -1.75647068e+00 -5.34731567e-01
-5.40887415e-02 -6.47153735e-01 -3.31539661e-03 -3.46753001e-01
-1.35794568e+00 7.75456309e-01 8.42487454e-01 -4.73778564e-05
9.67173398e-01 -2.34188139e-01 9.96123612e-01 4.32312071e-01
3.71235728e-01 -9.41434681e-01 2.99871325e-01 2.63460428e-01
4.83378530e-01 -1.35941541e+00 -1.78753759e-03 -1.97994068e-01
-4.97275114e-01 1.01516640e+00 9.67309415e-01 -2.20445454e-01
6.23559415e-01 1.42910823e-01 1.93820104e-01 7.77416006e-02
-6.78245366e-01 -3.60537648e-01 7.23830462e-02 3.33963633e-01
1.23256341e-01 -2.69047111e-01 -1.49595886e-01 3.56958538e-01
3.90616745e-01 3.72452796e-01 2.22900242e-01 1.04569876e+00
-5.71201384e-01 -8.42664421e-01 -5.27545452e-01 2.54404992e-01
-4.26110327e-01 2.91681856e-01 6.62350431e-02 8.04798484e-01
3.34902823e-01 8.66133451e-01 -1.66879475e-01 -4.44747388e-01
2.02511966e-01 -2.65739828e-01 3.62618804e-01 -5.22966385e-01
-4.06013042e-01 3.65262717e-01 -3.78519565e-01 -3.22403193e-01
-8.20422232e-01 -9.34355855e-01 -1.43291712e+00 -1.66208059e-01
-7.90582180e-01 -5.81898764e-02 3.87472808e-01 8.58961999e-01
2.42366880e-01 6.90493405e-01 6.27258420e-01 -8.99577856e-01
-5.11816442e-01 -7.22480357e-01 -2.54873455e-01 4.13572639e-01
6.29206002e-01 -1.04170370e+00 -1.35764077e-01 2.99751103e-01] | [6.37096643447876, -2.101515531539917] |
91ebd57f-ad39-44dd-be6c-656b628e1f90 | active-exploration-of-multimodal | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wanyan_Active_Exploration_of_Multimodal_Complementarity_for_Few-Shot_Action_Recognition_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wanyan_Active_Exploration_of_Multimodal_Complementarity_for_Few-Shot_Action_Recognition_CVPR_2023_paper.pdf | Active Exploration of Multimodal Complementarity for Few-Shot Action Recognition | Recently, few-shot action recognition receives increasing attention and achieves remarkable progress. However, previous methods mainly rely on limited unimodal data (e.g., RGB frames) while the multimodal information remains relatively underexplored. In this paper, we propose a novel Active Multimodal Few-shot Action Recognition (AMFAR) framework, which can actively find the reliable modality for each sample based on task-dependent context information to improve few-shot reasoning procedure. In meta-training, we design an Active Sample Selection (ASS) module to organize query samples with large differences in the reliability of modalities into different groups based on modality-specific posterior distributions. In addition, we design an Active Mutual Distillation (AMD) module to capture discriminative task-specific knowledge from the reliable modality to improve the representation learning of unreliable modality by bidirectional knowledge distillation. In meta-test, we adopt Adaptive Multimodal Inference (AMI) module to adaptively fuse the modality-specific posterior distributions with a larger weight on the reliable modality. Extensive experimental results on four public benchmarks demonstrate that our model achieves significant improvements over existing unimodal and multimodal methods. | ['Changsheng Xu', 'Chaofan Chen', 'Xiaoshan Yang', 'Yuyang Wanyan'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['few-shot-action-recognition', 'action-recognition-in-videos'] | ['computer-vision', 'computer-vision'] | [ 3.42398852e-01 -2.60090232e-01 -6.21468544e-01 -4.26759034e-01
-1.37285733e+00 -3.98412198e-02 6.39567137e-01 -1.72918469e-01
-4.72432107e-01 6.77603960e-01 7.38979101e-01 4.02895778e-01
7.32737705e-02 -4.88512695e-01 -5.97267628e-01 -1.06986678e+00
4.87543643e-01 4.17845905e-01 3.91709417e-01 5.55198938e-02
2.59217411e-01 -2.21599668e-01 -1.71656382e+00 7.14606643e-01
9.17038560e-01 1.21389961e+00 2.01329768e-01 3.44573110e-01
-2.64576048e-01 1.19739211e+00 -3.32028508e-01 -2.44699761e-01
-2.32800901e-01 -6.93233371e-01 -7.19259322e-01 2.79529661e-01
1.13495126e-01 -6.24052763e-01 -3.67675453e-01 1.01489151e+00
8.41637075e-01 6.84285343e-01 6.21917188e-01 -1.37923622e+00
-5.81050098e-01 7.66085446e-01 -6.58309102e-01 2.16717198e-01
5.85982323e-01 5.65569401e-01 9.71496582e-01 -1.18695533e+00
5.29881716e-01 1.50922525e+00 1.39700040e-01 6.49737477e-01
-1.01128030e+00 -5.71492910e-01 5.51572144e-01 9.79095936e-01
-1.35975158e+00 -6.82837963e-01 1.02248263e+00 -7.80481473e-02
7.17505932e-01 2.03233838e-01 5.96537888e-01 1.46650469e+00
-2.66640872e-01 1.56487656e+00 8.89775276e-01 -2.20959425e-01
5.49351811e-01 -1.03402250e-01 1.93580780e-02 6.03975832e-01
-2.42426515e-01 -3.39553416e-01 -1.00144482e+00 -2.72976726e-01
5.49993813e-01 5.50982833e-01 -3.17179680e-01 -4.67256874e-01
-1.59463966e+00 6.59825981e-01 3.43281448e-01 1.53574854e-01
-4.26090181e-01 8.56909901e-02 4.83329028e-01 2.93608680e-02
1.93172961e-01 -3.67217995e-02 -1.11979946e-01 -4.14201111e-01
-5.69828510e-01 -1.81950733e-01 3.51856738e-01 8.66810262e-01
9.44284081e-01 -2.44123548e-01 -7.37173855e-01 1.08556581e+00
6.38709605e-01 5.87038457e-01 6.62297726e-01 -1.04238904e+00
8.22303772e-01 9.02545750e-01 -1.32243529e-01 -6.89051270e-01
9.06162634e-02 2.89966732e-01 -7.34629154e-01 -9.89843532e-02
-1.25036100e-02 3.13710198e-02 -1.03421009e+00 1.50941491e+00
4.78320926e-01 4.41560596e-01 1.99430391e-01 1.21581733e+00
1.11827791e+00 6.01197779e-01 4.34788942e-01 -2.82199651e-01
1.16677082e+00 -1.10255051e+00 -7.09749758e-01 -5.05571365e-02
4.58300442e-01 -4.38743770e-01 1.12696481e+00 1.50339335e-01
-9.57913935e-01 -4.68794823e-01 -8.14545214e-01 -1.53124630e-01
-3.05272430e-01 1.07181380e-02 5.06936252e-01 2.57707059e-01
-3.06397498e-01 1.56855434e-01 -8.92624438e-01 -3.42405289e-01
8.11251640e-01 1.99863166e-02 -4.22070473e-01 -6.05029106e-01
-1.26650107e+00 5.86606503e-01 6.78268790e-01 9.87055600e-02
-1.27782345e+00 -4.53686714e-01 -1.04127574e+00 -2.07513347e-01
8.89314353e-01 -8.19377959e-01 1.11598575e+00 -9.95589316e-01
-1.71813476e+00 4.56229627e-01 -3.38552296e-01 -8.61929506e-02
3.64264667e-01 -9.75773409e-02 -4.76461083e-01 5.01809716e-01
-9.93147045e-02 8.51375520e-01 9.68188107e-01 -1.25295019e+00
-7.50026107e-01 -4.41133201e-01 2.26003528e-01 7.60071397e-01
-2.64804035e-01 -1.90357715e-01 -7.33130395e-01 -4.27610189e-01
2.70271480e-01 -6.45963669e-01 -5.32338321e-02 1.20429568e-01
-4.33897078e-01 -4.84799981e-01 6.53364599e-01 -3.86626154e-01
1.08073008e+00 -2.18619633e+00 8.25422168e-01 -3.93534899e-02
3.18743996e-02 -1.29377812e-01 -1.77209437e-01 3.23874384e-01
3.19439709e-01 -3.55964810e-01 -4.20812398e-01 -6.30252004e-01
2.03417107e-01 5.06543279e-01 -2.99585909e-01 3.00351381e-01
3.15914601e-01 9.98442411e-01 -1.08754027e+00 -1.00703108e+00
4.76489246e-01 3.57842833e-01 -2.08095819e-01 4.75969017e-01
-1.15456812e-01 5.56327343e-01 -6.01645410e-01 1.32836533e+00
4.48626012e-01 -2.39509225e-01 -7.85880834e-02 -4.25282449e-01
1.94445804e-01 -2.52083838e-01 -1.12171507e+00 2.41708899e+00
-1.43123865e-01 8.44899639e-02 -3.36659521e-01 -9.21197712e-01
5.42035520e-01 2.31723249e-01 4.61794496e-01 -8.60406399e-01
1.69605583e-01 -2.73850244e-02 -2.17395514e-01 -9.08856928e-01
4.20738280e-01 -2.75389254e-01 -8.42525214e-02 2.77286530e-01
3.06448728e-01 2.47205853e-01 9.37017798e-02 3.71574819e-01
9.95605886e-01 4.38540548e-01 1.50648549e-01 4.90812242e-01
5.85841775e-01 -2.62716711e-01 7.92950809e-01 6.70322537e-01
-5.90896726e-01 7.90490031e-01 3.74764174e-01 1.12301953e-01
-4.44599330e-01 -1.15964901e+00 1.98047265e-01 1.48433149e+00
5.59150457e-01 -2.65619338e-01 -2.08632275e-01 -9.01701391e-01
-1.23172767e-01 7.96266139e-01 -9.03238833e-01 -4.67774928e-01
-2.18494594e-01 -6.64795578e-01 3.71953279e-01 8.40762019e-01
7.31767833e-01 -1.17563105e+00 -4.09321487e-01 -1.76481009e-01
-4.52274501e-01 -8.32011223e-01 -1.99261904e-01 -1.20105065e-01
-8.09286058e-01 -9.32608008e-01 -1.15168726e+00 -4.42121118e-01
6.94241703e-01 4.51198846e-01 7.68026173e-01 -3.39179516e-01
-1.10722892e-02 9.05679464e-01 -7.47598171e-01 -1.00661777e-01
2.92438954e-01 -2.36128241e-01 -2.32477099e-01 3.64707947e-01
5.63896775e-01 -3.05327654e-01 -6.94170535e-01 4.56435889e-01
-9.75794435e-01 1.49324909e-01 8.54917765e-01 8.39934707e-01
9.21861887e-01 -2.59298742e-01 4.14080292e-01 -5.10272145e-01
3.03151131e-01 -9.09631670e-01 2.09817097e-01 7.01092362e-01
-1.49970904e-01 5.20404428e-03 2.70875335e-01 -7.12635398e-01
-1.50023794e+00 1.00524478e-01 2.48642236e-01 -7.77511477e-01
-1.79253131e-01 6.05173409e-01 -6.60793722e-01 3.16650093e-01
1.81364715e-01 5.56921780e-01 -9.65198949e-02 -3.46468896e-01
5.45057595e-01 4.63834912e-01 3.84569287e-01 -6.80246294e-01
3.24579954e-01 6.51324272e-01 -3.30738336e-01 -4.28494751e-01
-8.97257864e-01 -6.33891106e-01 -4.62355077e-01 -4.77938682e-01
9.43385303e-01 -1.16611743e+00 -5.44355452e-01 5.26669443e-01
-6.91415906e-01 -1.24878064e-01 -4.45337772e-01 7.29005873e-01
-6.94279432e-01 5.43227494e-01 -2.62960106e-01 -9.71634984e-01
-6.03394844e-02 -1.30757070e+00 1.44984806e+00 5.82099855e-01
1.24761507e-01 -7.03138769e-01 1.97211847e-01 7.14262962e-01
6.33068755e-02 1.04309162e-02 5.60507536e-01 -4.74496812e-01
-6.58188939e-01 -6.93836510e-02 -3.64091814e-01 2.28529409e-01
4.42381538e-02 -2.56788701e-01 -9.70111787e-01 -1.95859466e-02
-2.78175354e-01 -9.48859215e-01 1.41877079e+00 1.39415160e-01
1.16261077e+00 1.35657310e-01 -2.66145617e-01 2.31269002e-01
1.06320953e+00 1.42918834e-02 9.20343220e-01 -8.91002491e-02
9.05730247e-01 3.06139529e-01 9.76435840e-01 6.67752385e-01
5.41477263e-01 3.12321305e-01 3.67732227e-01 3.71137768e-01
-4.37055267e-02 -3.22218359e-01 5.76305330e-01 7.28152514e-01
-4.34691519e-01 -1.53261945e-01 -5.80450952e-01 3.09160769e-01
-2.37585044e+00 -1.34344065e+00 4.61380422e-01 2.20981622e+00
9.02579069e-01 5.36329709e-02 1.91272154e-01 -9.80038792e-02
7.08529711e-01 3.26003611e-01 -9.78856742e-01 5.23628235e-01
-3.23459685e-01 -3.60136330e-01 -1.02460282e-02 1.48586705e-01
-1.05300009e+00 6.72605217e-01 4.98722982e+00 1.15804243e+00
-7.21313596e-01 2.63909638e-01 3.87849867e-01 -3.87226820e-01
-5.18972814e-01 1.30098641e-01 -5.66097379e-01 6.89857781e-01
3.02720100e-01 8.85556042e-02 1.71166912e-01 8.04811299e-01
-1.02745190e-01 -5.37729144e-01 -9.72096384e-01 1.37383974e+00
5.35510421e-01 -1.04826057e+00 3.09873939e-01 -2.73061872e-01
6.25874400e-01 -2.36152872e-01 -5.92905991e-02 8.37538064e-01
-1.92047302e-02 -6.81501269e-01 4.40713555e-01 1.47987032e+00
4.43666637e-01 -7.90457785e-01 7.38157034e-01 4.04049873e-01
-1.31110311e+00 -3.46508503e-01 -5.69446206e-01 2.47697800e-01
2.18480214e-01 1.86068296e-01 -2.32178956e-01 6.88130617e-01
5.48183501e-01 1.07804751e+00 -6.57009184e-01 1.08517599e+00
-1.49530306e-01 5.65880001e-01 -6.14598617e-02 -4.17344011e-02
5.00226207e-02 -3.54899839e-02 4.97666955e-01 7.97604322e-01
1.22013114e-01 4.72127736e-01 3.62211525e-01 5.89923799e-01
1.61155965e-02 3.04312836e-02 -7.68179670e-02 -3.32859397e-01
5.09949565e-01 1.11917257e+00 -4.74896431e-01 -5.67250490e-01
-6.07706368e-01 1.23216772e+00 4.42792088e-01 4.75960046e-01
-9.93376374e-01 -2.06967875e-01 3.20654541e-01 -5.29519022e-01
3.64120215e-01 4.74016964e-02 1.39182404e-01 -1.59034014e+00
-1.81887113e-03 -6.66727006e-01 9.89486694e-01 -1.00537217e+00
-1.67900288e+00 -1.35661922e-02 1.39552489e-01 -1.58799505e+00
-2.79463325e-02 -1.61563605e-01 -7.04486549e-01 6.75364256e-01
-1.26788902e+00 -1.35304987e+00 -5.57832718e-01 8.96993041e-01
7.65427053e-01 -3.57465446e-01 7.29125857e-01 2.29321778e-01
-8.86753976e-01 4.75197524e-01 -1.30845800e-01 -1.06020588e-02
8.62557173e-01 -1.04607487e+00 -3.99252892e-01 4.30337191e-01
-1.57017291e-01 5.87948799e-01 3.66550535e-01 -6.99255884e-01
-1.82976758e+00 -8.19069922e-01 2.24316269e-01 -3.76224458e-01
5.79211771e-01 4.60675582e-02 -8.50015581e-01 4.50851500e-01
3.13018113e-01 8.73844326e-02 1.12203550e+00 1.81284845e-01
-4.93909180e-01 -7.96277970e-02 -9.34077859e-01 4.86078322e-01
1.12325633e+00 -6.33149683e-01 -8.74417722e-01 9.36651155e-02
6.75051272e-01 -1.37255147e-01 -6.80697560e-01 5.23470461e-01
5.95963895e-01 -9.27856386e-01 8.73101592e-01 -7.35153913e-01
5.66173971e-01 -5.18773079e-01 -6.09881699e-01 -1.05255818e+00
-2.66708165e-01 -4.17825393e-02 -7.13661134e-01 1.28527892e+00
2.04931155e-01 -3.03383946e-01 5.60128927e-01 7.24366903e-01
-1.09687299e-01 -8.78691435e-01 -1.10601532e+00 -4.04642522e-01
-4.93141919e-01 -4.89132971e-01 4.46417153e-01 8.30555618e-01
2.16710314e-01 3.88294518e-01 -5.34433424e-01 -1.12855911e-01
5.69571316e-01 2.82334447e-01 6.46280229e-01 -9.06220675e-01
-4.73357111e-01 -2.71208882e-01 -4.85924453e-01 -1.11053586e+00
1.54280066e-01 -6.55122459e-01 2.95952111e-01 -1.65814888e+00
8.27720642e-01 1.84745237e-01 -9.27082777e-01 6.23790622e-01
-5.76562107e-01 1.06219791e-01 1.97749048e-01 3.84616345e-01
-1.62245703e+00 1.27899373e+00 1.22829700e+00 -4.69596535e-01
-4.50170301e-02 -3.05444300e-01 -5.45841455e-01 6.77013218e-01
3.53060395e-01 -2.59866565e-01 -6.38506889e-01 -3.45280468e-01
2.30298400e-01 1.42056525e-01 5.46632826e-01 -1.00023413e+00
4.12567824e-01 -4.73887950e-01 6.66903317e-01 -8.77441943e-01
7.04390526e-01 -8.11982751e-01 -1.91616133e-01 1.52519150e-02
-7.28851020e-01 -5.39644063e-01 -1.96781024e-01 1.14591706e+00
-2.47631177e-01 -1.11594543e-01 4.70356792e-01 -1.64079443e-01
-1.42848778e+00 4.94369835e-01 -5.56811281e-02 9.10586268e-02
9.79774952e-01 -2.42084339e-01 -3.90602380e-01 -1.87895745e-01
-8.21615756e-01 5.35918295e-01 2.78184950e-01 5.09770453e-01
1.04513073e+00 -1.85076666e+00 -5.29793561e-01 -6.74039945e-02
8.01358402e-01 -1.09045260e-01 1.07001555e+00 1.22546184e+00
2.19946966e-01 -2.32645005e-01 -2.15707406e-01 -7.11057186e-01
-1.23582387e+00 4.01333213e-01 1.11900762e-01 1.87304303e-01
-4.00103003e-01 1.10589898e+00 -2.53036954e-02 -1.30315885e-01
4.42596018e-01 6.54670447e-02 -3.38505089e-01 4.45883214e-01
9.10938919e-01 6.07398391e-01 -5.54290295e-01 -7.59926319e-01
-5.28560936e-01 2.91728884e-01 -1.09753013e-01 -3.03281486e-01
1.02064192e+00 -2.72535622e-01 7.21032321e-02 1.10416567e+00
1.16781723e+00 -4.79380518e-01 -1.49490023e+00 -6.23013794e-01
-4.66099620e-01 -7.08780050e-01 7.30788782e-02 -7.65291750e-01
-9.57986057e-01 8.78249526e-01 7.27262795e-01 -3.14467520e-01
1.23438108e+00 1.95275530e-01 7.69653499e-01 6.54845119e-01
3.48952711e-01 -1.53318393e+00 5.09477198e-01 3.35737765e-01
7.47988284e-01 -1.69223428e+00 1.13246650e-01 -2.05198675e-02
-1.25322425e+00 9.31568742e-01 7.95081019e-01 1.95332304e-01
4.69773501e-01 -3.81482124e-01 -2.37190127e-01 -3.93450484e-02
-7.97228038e-01 -6.34484470e-01 4.81311858e-01 4.70000952e-01
6.53850958e-02 1.56333283e-01 -5.77454157e-02 8.71138513e-01
7.70473123e-01 2.25717634e-01 -1.17190093e-01 1.32670081e+00
-7.19407737e-01 -6.42894089e-01 -2.38455400e-01 3.39408845e-01
3.49082142e-01 2.60341540e-02 -4.97825682e-01 2.95379490e-01
2.80622512e-01 9.64664578e-01 1.01390453e-02 -5.82594335e-01
1.74828678e-01 2.82058924e-01 7.14937150e-01 -4.95831877e-01
4.46142405e-02 7.29344785e-02 -5.98517060e-02 -7.11528480e-01
-7.52648711e-01 -8.84036124e-01 -1.41807210e+00 1.33866012e-01
-2.45555639e-01 -1.71211481e-01 1.20171361e-01 1.35770988e+00
3.73260230e-01 4.53939736e-01 4.83136564e-01 -8.92381668e-01
-4.63433653e-01 -1.05033243e+00 -5.98807275e-01 4.87682104e-01
2.43119430e-03 -1.04333782e+00 -2.50988781e-01 1.32100191e-03] | [8.608909606933594, 0.775636613368988] |
4fd0665c-79ee-4fb1-a23b-ab9ce69f2ef9 | auto-encoder-based-model-for-high-dimensional | 2108.02083 | null | https://arxiv.org/abs/2108.02083v2 | https://arxiv.org/pdf/2108.02083v2.pdf | Auto-encoder based Model for High-dimensional Imbalanced Industrial Data | With the proliferation of IoT devices, the distributed control systems are now capturing and processing more sensors at higher frequency than ever before. These new data, due to their volume and novelty, cannot be effectively consumed without the help of data-driven techniques. Deep learning is emerging as a promising technique to analyze these data, particularly in soft sensor modeling. The strong representational capabilities of complex data and the flexibility it offers from an architectural perspective make it a topic of active applied research in industrial settings. However, the successful applications of deep learning in soft sensing are still not widely integrated in factory control systems, because most of the research on soft sensing do not have access to large scale industrial data which are varied, noisy and incomplete. The results published in most research papers are therefore not easily reproduced when applied to the variety of data in industrial settings. Here we provide manufacturing data sets that are much larger and more complex than public open soft sensor data. Moreover, the data sets are from Seagate factories on active service with only necessary anonymization, so that they reflect the complex and noisy nature of real-world data. We introduce a variance weighted multi-headed auto-encoder classification model that fits well into the high-dimensional and highly imbalanced data. Besides the use of weighting or sampling methods to handle the highly imbalanced data, the model also simultaneously predicts multiple outputs by exploiting output-supervised representation learning and multi-task weighting. | ['Chao Zhang', 'Sthitie Bom'] | 2021-08-04 | null | null | null | null | ['sensor-modeling'] | ['computer-vision'] | [ 4.18783516e-01 -4.28528059e-03 -3.66767466e-01 -6.02380037e-01
-2.22762868e-01 -3.70276392e-01 4.39971358e-01 2.90790945e-01
-1.23250425e-01 7.36076236e-01 1.71658486e-01 1.74625069e-01
-6.19615197e-01 -9.50921595e-01 -6.35505557e-01 -6.76837683e-01
1.54252484e-01 6.20445788e-01 -3.16901922e-01 -4.55901235e-01
-1.68318674e-01 4.42384213e-01 -1.68981695e+00 4.32521313e-01
6.51624084e-01 1.54394007e+00 1.03500053e-01 3.00656527e-01
-2.96979938e-02 5.58300436e-01 -8.04045677e-01 -3.27713013e-01
6.12713456e-01 6.03679148e-03 -2.90214002e-01 9.03762206e-02
2.46819220e-02 -1.48723245e-01 -2.04060629e-01 1.08134687e+00
5.89910924e-01 -2.32385755e-01 1.85065508e-01 -1.34871995e+00
-6.50576651e-01 7.01313972e-01 -1.09885678e-01 -2.23752007e-01
3.17225963e-01 6.92849921e-04 8.01847577e-01 -4.26068574e-01
5.18214345e-01 1.03801346e+00 5.22275686e-01 1.96763337e-01
-9.89078343e-01 -6.93258286e-01 -2.21369024e-02 2.34619245e-01
-9.38252509e-01 -2.81429291e-01 1.14176297e+00 -3.70549738e-01
7.94237077e-01 3.60962540e-01 6.79187298e-01 1.74421442e+00
5.32811522e-01 4.62970734e-01 1.05803013e+00 6.95763677e-02
5.30081391e-01 3.45717907e-01 -3.64486165e-02 9.95180607e-02
5.74576318e-01 3.06086749e-01 -4.23111051e-01 6.28172699e-03
4.00048852e-01 9.34059083e-01 1.58665463e-01 -6.94887459e-01
-1.38770425e+00 7.51616836e-01 3.74938339e-01 4.75229442e-01
-5.32709897e-01 7.72418047e-04 7.02225327e-01 9.02977049e-01
4.55287278e-01 5.54490924e-01 -8.53927255e-01 -2.77854383e-01
-8.11860383e-01 1.34160414e-01 8.70875120e-01 1.00943482e+00
5.04438281e-01 3.58537108e-01 2.98863113e-01 4.73323703e-01
7.58844316e-02 6.70913637e-01 5.26883304e-01 -9.44693506e-01
5.89541197e-01 9.85359848e-01 -3.75566706e-02 -1.17744660e+00
-5.17983556e-01 -4.35130060e-01 -1.54584992e+00 4.55507338e-01
2.71256447e-01 -1.29913568e-01 -5.60301840e-01 1.42260039e+00
1.10175610e-01 -4.78599101e-01 5.51459193e-02 7.15683103e-01
3.80390614e-01 4.70075279e-01 -4.69908923e-01 -1.46830127e-01
1.17108142e+00 -2.96432018e-01 -1.27868712e+00 -5.28072640e-02
3.12176257e-01 -5.83198011e-01 8.67883861e-01 8.96448433e-01
-7.01366544e-01 -8.04888546e-01 -1.24155462e+00 1.13144889e-01
-1.06582129e+00 -1.48769334e-01 7.14148521e-01 6.53745413e-01
-4.31356758e-01 6.99160278e-01 -8.67995679e-01 -2.42628783e-01
8.92832577e-01 4.72627640e-01 -5.84041834e-01 -4.07771021e-01
-1.18057847e+00 8.57814848e-01 3.30242932e-01 1.39626086e-01
-6.55816376e-01 -5.76604784e-01 -8.61427009e-01 -1.09620169e-01
7.34731615e-01 -4.68380809e-01 6.74947977e-01 -7.06970215e-01
-1.56728840e+00 2.33499721e-01 4.54273671e-01 -5.28381228e-01
6.07978821e-01 -3.13146979e-01 -7.33610749e-01 -4.63240355e-01
-3.05233955e-01 -2.01184601e-02 1.02497268e+00 -1.24653184e+00
-3.05928737e-01 -7.62578130e-01 -1.54487073e-01 -4.62692529e-01
-4.77314442e-01 -2.53036141e-01 5.84079921e-01 -5.73487699e-01
-3.17396945e-03 -7.55693376e-01 -3.78143728e-01 -1.29770339e-01
-5.89032948e-01 -3.24000679e-02 1.21998632e+00 -4.73434180e-01
9.10793304e-01 -2.23064685e+00 2.48106286e-01 2.01965153e-01
3.77027720e-01 4.57699776e-01 1.44136529e-02 6.96559966e-01
-2.25134507e-01 2.34542042e-02 -2.76458591e-01 -3.38216364e-01
4.52665150e-01 5.82103968e-01 -3.74461591e-01 5.04269123e-01
1.50624186e-01 8.99528682e-01 -9.43425298e-01 1.95188355e-02
7.56004274e-01 2.63561279e-01 -2.52101868e-01 1.70274191e-02
-2.46788576e-01 4.14236993e-01 -4.78316993e-01 9.65272248e-01
6.76362038e-01 3.11925937e-03 2.16052204e-01 -4.61213380e-01
1.25357797e-02 -1.65993124e-01 -1.49884963e+00 1.94504750e+00
-5.67643464e-01 1.38197586e-01 2.92778313e-01 -1.33094776e+00
1.11782801e+00 2.99770653e-01 1.08905447e+00 -9.65337217e-01
3.44410658e-01 2.85959959e-01 -1.22416079e-01 -6.24109566e-01
4.78046805e-01 -1.29360422e-01 -6.06482387e-01 3.49858075e-01
2.68555991e-02 -3.08943003e-01 9.67731774e-02 -2.70850897e-01
1.24127305e+00 -3.71864140e-02 2.06914544e-01 -1.43090691e-02
9.43458825e-02 -3.86736840e-01 9.19393718e-01 4.29935068e-01
-9.00487378e-02 4.14349496e-01 3.09082627e-01 -7.56103814e-01
-1.08835661e+00 -1.00167203e+00 -1.63973689e-01 7.26013720e-01
6.75663294e-04 -4.21282440e-01 -2.48587891e-01 -5.88908792e-01
5.38261831e-01 2.58243173e-01 -3.47309709e-01 -4.36958671e-01
-3.67042243e-01 -5.53376734e-01 1.47857577e-01 4.73714739e-01
2.13522375e-01 -1.19139302e+00 -8.39734256e-01 3.14743787e-01
2.65680760e-01 -1.11723959e+00 2.44485036e-01 7.29816198e-01
-6.67669654e-01 -1.06139100e+00 -1.77882418e-01 -2.24771768e-01
1.85928077e-01 -1.56273782e-01 1.08724868e+00 -5.73707938e-01
-4.19475943e-01 -2.33059563e-02 -4.10416931e-01 -9.94985521e-01
-5.56464732e-01 2.64639169e-01 4.43727463e-01 4.59842868e-02
4.54091519e-01 -6.70036614e-01 -1.12253793e-01 2.21668914e-01
-1.24802053e+00 -4.82479602e-01 7.31370986e-01 7.88305402e-01
5.73889375e-01 6.30710185e-01 1.06537712e+00 -8.47827613e-01
5.98438621e-01 -7.35390902e-01 -5.65807641e-01 -8.15698728e-02
-7.75057793e-01 -8.87992159e-02 9.07796621e-01 -4.75493371e-01
-6.39081240e-01 8.30376446e-02 3.88092622e-02 -5.88953555e-01
-4.07205075e-01 4.11602765e-01 -5.52173674e-01 4.11551327e-01
2.70954281e-01 -2.42673740e-01 3.74838531e-01 -6.47739410e-01
2.73182720e-01 9.51600730e-01 2.35499918e-01 -1.94033116e-01
8.92508686e-01 4.64896321e-01 7.45355785e-02 -7.21721411e-01
-5.34131944e-01 -3.82314958e-02 -5.10225654e-01 7.56014064e-02
8.78800035e-01 -8.03059757e-01 -8.49837542e-01 6.41788363e-01
-7.66817331e-01 -3.89287039e-03 -1.28952765e+00 5.87922633e-01
-4.75349247e-01 3.77438706e-03 -3.31973881e-01 -8.52629900e-01
-1.56526819e-01 -1.33881676e+00 1.24963892e+00 -3.10452491e-01
-3.58329266e-01 -6.77074969e-01 -2.36098394e-01 5.62861562e-01
7.36236334e-01 7.34700799e-01 7.78881848e-01 -7.97976851e-01
-6.15192354e-01 -7.30066538e-01 2.30104432e-01 9.56806540e-01
6.61048234e-01 -2.48974368e-01 -9.00655508e-01 -4.83366013e-01
3.89664501e-01 -5.28362334e-01 4.89435613e-01 2.01318458e-01
1.57823896e+00 -2.43450895e-01 -7.81433433e-02 4.56350446e-01
1.17985332e+00 -4.21704762e-02 4.67122406e-01 -5.81590421e-02
7.95451701e-01 5.51592231e-01 6.56483650e-01 6.69189870e-01
2.57963687e-01 5.70856094e-01 1.13770354e+00 -1.23309091e-01
2.33473435e-01 -8.61061737e-02 3.86568010e-01 1.25713432e+00
1.13888226e-01 -3.20000410e-01 -4.90203112e-01 4.22596902e-01
-1.91549969e+00 -1.02876937e+00 8.41071457e-03 2.08330297e+00
6.50747716e-01 2.35611066e-01 -7.49140233e-02 7.96606123e-01
3.26101869e-01 3.95609468e-01 -8.02802622e-01 -5.68154693e-01
-2.56097823e-01 3.72555137e-01 6.96590364e-01 -2.42864579e-01
-1.05260098e+00 1.91356763e-01 5.49345827e+00 5.11861563e-01
-9.18677866e-01 7.22395703e-02 2.97666043e-01 -4.95987356e-01
-2.26726815e-01 -6.47789419e-01 -3.89346331e-01 8.76291752e-01
1.23493981e+00 9.76313502e-02 6.03603125e-01 9.20412898e-01
6.33537248e-02 2.57435590e-01 -1.37980783e+00 1.19065464e+00
-1.70927376e-01 -1.25466669e+00 -4.41719621e-01 4.91868079e-01
9.08901513e-01 2.52850711e-01 2.81759083e-01 1.56710744e-01
4.33424205e-01 -1.04039454e+00 7.25130081e-01 7.21236885e-01
6.67652190e-01 -6.15136683e-01 1.21186030e+00 3.83004010e-01
-7.34590530e-01 -8.25529754e-01 -3.87139469e-01 -3.48212540e-01
2.22840294e-01 1.32175708e+00 -3.69462490e-01 8.99965942e-01
9.79566634e-01 9.76604700e-01 -1.38814628e-01 3.83067906e-01
3.19696397e-01 1.93597212e-01 -4.34294432e-01 4.43332121e-02
-2.60415733e-01 -2.86146700e-01 2.99771041e-01 5.47219217e-01
4.82970774e-01 -7.13280022e-01 3.93933564e-01 9.18839395e-01
-1.94726557e-01 -2.88573533e-01 -1.30534744e+00 -3.51658165e-01
3.34236234e-01 1.20555627e+00 -1.40752405e-01 1.15363091e-01
-4.36341286e-01 5.46897590e-01 -2.73300767e-01 -4.09738207e-03
-4.31465417e-01 -2.08786041e-01 1.21731746e+00 1.30010977e-01
2.98955411e-01 -3.95957828e-01 -4.36953962e-01 -9.02972400e-01
4.09749806e-01 -1.26238084e+00 6.96256310e-02 -3.66183490e-01
-1.87996197e+00 3.45248938e-01 -1.66607052e-01 -1.23508728e+00
-5.02182961e-01 -6.58089817e-01 1.54781686e-02 4.24901336e-01
-1.41588485e+00 -1.07729506e+00 -3.64015520e-01 6.02488637e-01
4.24775928e-01 -4.15656269e-01 1.05593038e+00 6.78908646e-01
-4.79589045e-01 2.46005386e-01 3.96429092e-01 -7.22929686e-02
6.27526581e-01 -1.16432369e+00 2.00492721e-02 4.52391148e-01
5.84616233e-03 3.06692839e-01 4.28442806e-01 -5.24632752e-01
-1.90744352e+00 -1.23969674e+00 5.14592290e-01 -7.62012064e-01
7.68335640e-01 -8.72675419e-01 -7.61393785e-01 6.13987744e-01
1.37733668e-01 3.80909979e-01 7.65594184e-01 1.54973734e-02
-1.46328509e-01 -8.05569589e-01 -1.52910447e+00 -1.47916213e-01
9.25165355e-01 -5.57682455e-01 -5.20919085e-01 4.19975698e-01
9.35174406e-01 -1.49022192e-01 -1.44393575e+00 6.51457727e-01
2.71376550e-01 -8.31487358e-01 1.01064074e+00 -4.77732629e-01
4.69120562e-01 -1.76427677e-01 -5.94589353e-01 -1.45315349e+00
-6.36470839e-02 -5.38269043e-01 -4.57290173e-01 1.32671845e+00
3.33622172e-02 -8.04388106e-01 6.73579752e-01 3.15862358e-01
-1.85284793e-01 -6.96957350e-01 -1.14544332e+00 -7.78101444e-01
-1.36387005e-01 -4.65952635e-01 1.23425436e+00 1.10050476e+00
-1.26952082e-01 2.49737166e-02 -5.38184166e-01 -1.83070377e-01
4.98563588e-01 2.09395334e-01 7.56562352e-01 -1.76540816e+00
1.68218184e-03 3.45409252e-02 -8.21961641e-01 -2.01846376e-01
7.90470839e-02 -8.78111362e-01 -2.53428072e-01 -1.24717104e+00
-4.04085636e-01 -5.75098813e-01 -7.06349552e-01 4.06891733e-01
6.02378607e-01 1.10176884e-01 2.07355037e-01 -2.03558981e-01
-4.91275162e-01 7.59254456e-01 1.15629518e+00 -5.22938132e-01
1.23886667e-01 2.05883294e-01 -6.76195443e-01 6.48188353e-01
8.50380719e-01 -2.80313641e-01 -5.72867036e-01 -4.22836900e-01
5.07675648e-01 -7.94260502e-02 3.11502010e-01 -1.22317302e+00
-7.90737197e-02 -1.50596187e-01 5.49073458e-01 -5.51112771e-01
4.95846748e-01 -1.82301927e+00 4.96699750e-01 3.97683531e-01
-2.26266474e-01 3.45007598e-01 -2.08712533e-01 7.05823481e-01
-2.35031933e-01 1.97892264e-01 3.18459570e-01 -2.20882639e-01
-4.39817488e-01 5.03673255e-01 -6.78011551e-02 -2.58416772e-01
1.08770394e+00 -3.19999605e-01 -1.90662548e-01 -2.51341492e-01
-5.87653637e-01 1.71026550e-02 4.66592818e-01 9.95861173e-01
4.40187842e-01 -1.47873044e+00 -5.33680320e-01 8.53233933e-01
2.48528987e-01 4.54741150e-01 2.05442429e-01 4.34522212e-01
2.47220352e-01 1.48498103e-01 -4.24617022e-01 -5.68559468e-01
-6.44880116e-01 9.45559621e-01 -1.45593122e-01 -2.99591005e-01
-6.53787673e-01 2.80714512e-01 -5.83136797e-01 -7.90328920e-01
2.64484167e-01 -8.72210681e-01 2.43602261e-01 3.23874116e-01
3.69872242e-01 5.68368018e-01 5.54458022e-01 -2.85938263e-01
-2.54386991e-01 1.47081003e-01 3.19990724e-01 5.78063965e-01
1.98946404e+00 -4.21701372e-02 -1.43726766e-01 9.39851224e-01
1.25828016e+00 -1.88375384e-01 -1.09076548e+00 -1.56900525e-01
-2.61201430e-02 -4.54403013e-01 -1.65472366e-02 -9.08516169e-01
-1.59300482e+00 9.01045740e-01 9.21564281e-01 7.93502748e-01
1.01299965e+00 -1.95740685e-01 9.77414787e-01 3.78502220e-01
6.62575424e-01 -1.51784348e+00 9.86155421e-02 3.36310983e-01
7.66484499e-01 -1.39436340e+00 1.02839274e-02 -1.81175888e-01
-3.64619315e-01 7.04784036e-01 1.61184311e-01 6.79844245e-02
8.50049675e-01 9.10914183e-01 -5.12808338e-02 -1.82186007e-01
-5.04673839e-01 1.14022762e-01 -3.86553913e-01 1.01841831e+00
-4.05599140e-02 3.01763028e-01 2.23624751e-01 7.38915980e-01
-3.92460674e-01 3.45964283e-01 3.26548606e-01 9.51679945e-01
-2.45432705e-02 -1.42561340e+00 -3.50785166e-01 8.36144924e-01
-4.24832672e-01 5.94830751e-01 -1.06446788e-01 5.02194047e-01
5.37951827e-01 1.12833142e+00 1.21511370e-01 -8.02327573e-01
7.08734274e-01 -3.23253381e-03 1.59571514e-01 -3.57988715e-01
-6.52386844e-01 -5.15338778e-01 -1.82031959e-01 -9.48200285e-01
-3.63854080e-01 -7.10695922e-01 -9.24899340e-01 -4.59680855e-01
-8.31368789e-02 -3.63769531e-01 1.24124980e+00 8.72354627e-01
6.40615642e-01 1.21337771e+00 8.44347179e-01 -9.72725689e-01
-1.14148676e+00 -8.87150466e-01 -8.58059466e-01 7.50053585e-01
3.75103295e-01 -7.52098799e-01 -1.14693888e-01 -2.43698895e-01] | [7.256441116333008, 2.3219432830810547] |
257ad5d4-ed8e-4d62-97f9-e39ce9e7e75c | implicit-feedback-for-dense-passage-retrieval | 2204.00718 | null | https://arxiv.org/abs/2204.00718v2 | https://arxiv.org/pdf/2204.00718v2.pdf | Implicit Feedback for Dense Passage Retrieval: A Counterfactual Approach | In this paper we study how to effectively exploit implicit feedback in Dense Retrievers (DRs). We consider the specific case in which click data from a historic click log is available as implicit feedback. We then exploit such historic implicit interactions to improve the effectiveness of a DR. A key challenge that we study is the effect that biases in the click signal, such as position bias, have on the DRs. To overcome the problems associated with the presence of such bias, we propose the Counterfactual Rocchio (CoRocchio) algorithm for exploiting implicit feedback in Dense Retrievers. We demonstrate both theoretically and empirically that dense query representations learnt with CoRocchio are unbiased with respect to position bias and lead to higher retrieval effectiveness. We make available the implementations of the proposed methods and the experimental framework, along with all results at https://github.com/ielab/Counterfactual-DR. | ['Guido Zuccon', 'Hang Li', 'Shengyao Zhuang'] | 2022-04-01 | null | null | null | null | ['passage-retrieval'] | ['natural-language-processing'] | [ 8.98282379e-02 -1.58830583e-02 -4.21175122e-01 -3.07771325e-01
-7.41831660e-01 -6.76986992e-01 9.83265221e-01 -7.66513571e-02
-7.10345089e-01 9.39540923e-01 4.77089614e-01 -2.84494191e-01
-3.59823257e-01 -7.65859246e-01 -1.02059793e+00 -4.46338862e-01
-2.40992814e-01 2.83475041e-01 1.15025774e-01 -2.17508435e-01
6.49827898e-01 3.96391362e-01 -1.64219034e+00 3.75656694e-01
8.08347642e-01 6.62529588e-01 1.28942370e-01 6.23242795e-01
1.83291644e-01 4.62080032e-01 -7.53059864e-01 -4.79163021e-01
6.62820220e-01 4.13298309e-02 -4.86175627e-01 -5.39947689e-01
3.60357612e-01 -7.05254316e-01 -7.76728988e-01 8.72184098e-01
7.99389839e-01 1.99070379e-01 6.44064665e-01 -1.06532931e+00
-7.09662080e-01 8.76399100e-01 -6.44186378e-01 6.02494299e-01
2.97481954e-01 2.82160491e-01 1.53777170e+00 -9.42420661e-01
6.39612377e-01 1.27634573e+00 2.26192683e-01 2.01585844e-01
-1.53591526e+00 -8.84417295e-01 3.60520899e-01 1.56134844e-01
-1.44488525e+00 -6.57852709e-01 7.23894000e-01 -2.00292543e-01
6.41336977e-01 5.85383534e-01 4.33978230e-01 1.33092642e+00
-3.11482307e-02 1.51267374e+00 1.19778812e+00 -4.37429637e-01
2.31363818e-01 3.20637971e-01 5.80635011e-01 4.32828926e-02
7.56306589e-01 9.59951043e-01 -7.67191529e-01 -4.87942964e-01
6.29336596e-01 1.87994018e-01 -4.78290230e-01 -6.29843056e-01
-7.16692090e-01 1.04252076e+00 1.02130556e+00 1.95808902e-01
-5.59515536e-01 3.92278880e-01 -4.77189664e-03 6.09636188e-01
3.39505166e-01 8.58701348e-01 -3.64734501e-01 4.85854745e-02
-9.25050318e-01 7.02558577e-01 8.13114405e-01 8.22468221e-01
9.12409425e-01 -2.94809252e-01 -4.71420854e-01 6.74391627e-01
6.29784316e-02 4.85844553e-01 4.13473994e-01 -8.58651102e-01
4.32829767e-01 9.94251668e-02 6.40616000e-01 -1.01957035e+00
1.97750017e-01 -6.10555470e-01 -1.42138615e-01 -9.58696753e-02
2.75342852e-01 1.18237725e-02 -9.66371238e-01 1.77656615e+00
5.57527505e-02 -1.58184469e-02 -4.14900929e-01 9.91604090e-01
1.29533827e-01 1.91350833e-01 7.88289234e-02 -1.70781940e-01
7.33360469e-01 -5.44308305e-01 -9.78515327e-01 -2.54199117e-01
8.78352046e-01 -6.76742494e-01 1.20331705e+00 1.28306970e-01
-8.80700946e-01 6.31275401e-03 -8.51087272e-01 1.54945313e-03
-5.23961246e-01 -8.65496770e-02 9.04695094e-01 3.89383227e-01
-7.96202779e-01 6.52497888e-01 -3.47135782e-01 -1.04682453e-01
4.17996943e-01 3.90069097e-01 2.58781642e-01 -1.82371408e-01
-1.58040929e+00 6.47247910e-01 6.93831071e-02 9.16171968e-02
-8.72917354e-01 -9.64670897e-01 -3.60905439e-01 1.88636050e-01
8.05833638e-01 -5.72410822e-01 1.62628269e+00 -9.04337108e-01
-9.03247535e-01 5.16790807e-01 -2.04740584e-01 -8.32088470e-01
8.60876679e-01 -1.01683211e+00 5.44405216e-03 -2.19357148e-01
1.85909957e-01 4.16050553e-01 7.19160914e-01 -1.28946507e+00
-3.64148498e-01 -6.29771948e-01 3.86213213e-01 2.96073735e-01
-2.78187003e-02 -4.18220937e-01 -2.95587867e-01 -7.43438959e-01
-4.19854105e-01 -9.95519578e-01 -2.72535145e-01 -2.38103256e-01
-5.77222645e-01 -3.10082644e-01 7.32859194e-01 -3.80265832e-01
1.55914676e+00 -1.98448586e+00 -3.39856863e-01 4.69211340e-01
2.50482649e-01 3.46734405e-01 -2.65114695e-01 5.82308352e-01
1.42156526e-01 3.15812856e-01 1.51971877e-01 7.78101757e-02
1.69320419e-01 -1.98421665e-02 -7.87739217e-01 2.38597944e-01
-1.73228368e-01 1.15202105e+00 -8.85415316e-01 -7.68389925e-02
-1.21689007e-01 8.39524046e-02 -6.41893685e-01 3.27734083e-01
-3.30198288e-01 -9.33832377e-02 -8.33632767e-01 3.91479164e-01
8.09571505e-01 -2.42736593e-01 4.73297566e-01 -1.34056415e-02
-9.82170701e-02 8.43076766e-01 -8.46086204e-01 1.32948792e+00
-4.95609701e-01 5.51195025e-01 1.63806714e-02 -7.07863033e-01
5.95993936e-01 2.88660545e-02 5.57000451e-02 -1.22930384e+00
1.94620155e-02 3.47360015e-01 1.06650233e-01 -1.88644618e-01
6.01253808e-01 1.89651921e-01 9.26030651e-02 7.65621185e-01
-3.30683917e-01 2.12063268e-01 2.11393967e-01 7.40091264e-01
1.42217338e+00 -1.24715470e-01 4.36782986e-01 -2.21267045e-01
-5.37421294e-02 -2.34303206e-01 3.13179016e-01 1.53738952e+00
-9.84129831e-02 2.37899557e-01 4.48246747e-01 -1.56115010e-01
-1.00811005e+00 -1.12631273e+00 2.24253871e-02 1.06327760e+00
8.34842026e-02 -1.89070523e-01 -5.67324534e-02 -8.00834477e-01
5.77265501e-01 1.15869176e+00 -8.87156606e-01 -3.64400715e-01
-5.33269167e-01 -4.42268461e-01 1.65646806e-01 3.90661657e-01
4.44110960e-01 -8.35279882e-01 -3.82136524e-01 -1.53490663e-01
2.19113249e-02 -4.28507149e-01 -6.90587223e-01 9.72329639e-03
-9.27081227e-01 -1.04832506e+00 -8.06702077e-01 6.06210791e-02
3.12653750e-01 6.85296893e-01 1.36284065e+00 2.46930465e-01
-3.66136789e-01 5.55775106e-01 -2.98422605e-01 -4.33676064e-01
3.67417812e-01 7.02377260e-02 -9.59623009e-02 -1.64044499e-01
7.34433293e-01 -5.62509000e-01 -1.26681399e+00 3.09845418e-01
-9.54888463e-01 -3.71564656e-01 9.24411774e-01 8.68655562e-01
3.41244280e-01 -4.56330001e-01 8.24795187e-01 -1.53967321e+00
1.06118941e+00 -8.07592332e-01 -6.46828413e-01 2.15282515e-02
-9.20879960e-01 2.66129434e-01 -1.20820411e-01 -4.93122071e-01
-1.10289133e+00 -2.57947952e-01 3.54331553e-01 -3.64911586e-01
2.90993243e-01 5.94416976e-01 6.97056055e-02 2.70555377e-01
9.86729860e-01 -5.74811734e-02 -2.43360683e-01 -7.62450337e-01
6.80734634e-01 7.53691792e-01 -2.25996822e-01 -5.07100046e-01
8.50819588e-01 4.15909797e-01 -3.88185471e-01 -6.17180288e-01
-1.09634328e+00 -6.79038882e-01 -2.45203763e-01 7.11989701e-02
1.91025794e-01 -7.47516036e-01 -5.00389278e-01 -1.39328688e-01
-8.80694687e-01 -3.74516666e-01 -3.67858320e-01 4.89431351e-01
-5.53009510e-01 -1.45652676e-02 -3.08613747e-01 -1.12084723e+00
-1.88119099e-01 -6.73491895e-01 8.08321834e-01 1.17303856e-01
-2.43751153e-01 -7.67110229e-01 3.69157702e-01 1.40535966e-01
5.95833659e-01 -2.42561057e-01 5.73625743e-01 -1.02730119e+00
-1.37076986e+00 -4.39718485e-01 -2.51361161e-01 1.75638329e-02
-4.61165532e-02 -4.64910299e-01 -1.03196347e+00 -2.98125625e-01
-2.73375124e-01 -4.28165436e-01 1.20698810e+00 5.46630621e-01
1.04313803e+00 -7.91805089e-01 -5.80547631e-01 6.69104829e-02
1.30306375e+00 -8.04260150e-02 7.52860188e-01 3.46942693e-01
7.40150213e-02 5.42857051e-01 1.09537446e+00 5.10771751e-01
-1.09930769e-01 1.00916731e+00 2.93006092e-01 2.05225676e-01
-7.86115080e-02 -6.06648147e-01 2.08354801e-01 1.75721914e-01
1.36023566e-01 -2.40647703e-01 -4.46520388e-01 8.96587133e-01
-1.96962786e+00 -9.61981535e-01 9.02808160e-02 2.60021830e+00
9.23483431e-01 -6.13574162e-02 4.03218232e-02 -2.47199275e-02
7.03144133e-01 3.58745456e-01 -7.00077116e-01 -2.82215178e-01
8.76745954e-02 1.46373957e-01 7.17051864e-01 5.52002668e-01
-7.12973893e-01 8.09528589e-01 6.14833355e+00 8.41851890e-01
-7.37156332e-01 -5.68784028e-02 3.49429339e-01 -6.21658444e-01
-7.85786748e-01 1.71287451e-02 -7.29154229e-01 3.88643295e-01
1.00644958e+00 -5.03247857e-01 5.73651075e-01 7.55333245e-01
3.57103467e-01 -2.75729895e-01 -1.35275126e+00 6.16622984e-01
-3.81703556e-01 -1.36331928e+00 2.23638251e-01 3.74943703e-01
7.54924119e-01 1.94704741e-01 4.83142823e-01 6.66798949e-01
7.01541960e-01 -6.77749395e-01 3.99207324e-01 4.75825220e-01
5.84338963e-01 -4.11023974e-01 4.86158967e-01 3.03655535e-01
-6.10763371e-01 -1.76237151e-01 -4.07952249e-01 6.48909360e-02
-6.17743917e-02 8.86072338e-01 -1.11405444e+00 2.67504424e-01
5.78195632e-01 5.98685503e-01 -5.46525836e-01 1.30560899e+00
-4.82420415e-01 6.86984956e-01 -4.29410368e-01 -3.01909417e-01
-2.26197541e-02 5.19169457e-02 8.64023089e-01 8.79727781e-01
1.68063954e-01 4.32136320e-02 -5.01647770e-01 1.09591925e+00
-4.83424157e-01 -3.47997136e-02 -1.10870934e+00 -3.39824349e-01
8.07535350e-01 7.83852458e-01 3.68123427e-02 -2.94450343e-01
-2.19593033e-01 6.27189457e-01 4.53074664e-01 1.00335276e+00
-4.87017274e-01 -2.70201981e-01 6.46291852e-01 5.44493556e-01
4.83690828e-01 1.39425799e-01 -2.54974216e-01 -1.25439334e+00
2.13900790e-01 -8.66115570e-01 5.79798281e-01 -7.04597175e-01
-1.30217874e+00 -1.35882452e-01 5.80400191e-02 -6.66825652e-01
-1.68141797e-01 -2.60817558e-01 -5.40877998e-01 9.04616416e-01
-1.66606045e+00 -6.71508074e-01 -1.19970534e-02 4.31622922e-01
5.35227776e-01 2.11889952e-01 4.24863845e-01 2.63609588e-01
-7.97057971e-02 6.23570204e-01 2.11594969e-01 -1.73397586e-01
8.33352625e-01 -1.23402023e+00 1.28663331e-01 4.40150172e-01
3.22318673e-01 1.30827522e+00 8.57372522e-01 -8.05450737e-01
-1.46483612e+00 -7.80614436e-01 6.74478412e-01 -5.77231526e-01
8.33109856e-01 -3.92427653e-01 -7.19212174e-01 1.04154217e+00
1.50447324e-01 -3.76706719e-01 5.84784150e-01 7.99678385e-01
-6.74549997e-01 -6.84551001e-02 -9.28296626e-01 8.96702051e-01
1.09124517e+00 -6.26923621e-01 -8.30417871e-01 3.45707029e-01
8.78180265e-01 1.36744738e-01 -2.90010184e-01 4.62188959e-01
7.49472022e-01 -9.40265834e-01 9.52222049e-01 -7.62902141e-01
2.53616571e-01 6.69341013e-02 -2.48730198e-01 -1.22091103e+00
-1.23576246e-01 -5.78390419e-01 -3.58812541e-01 8.43769312e-01
6.77536547e-01 -8.45719218e-01 7.87797511e-01 6.76019549e-01
4.13975090e-01 -6.06895745e-01 -6.87951744e-01 -8.46125662e-01
8.38097334e-02 -1.54636905e-01 4.44039375e-01 6.30747914e-01
-4.03656751e-01 2.66898185e-01 -4.15426731e-01 1.28258048e-02
7.09797621e-01 3.22453767e-01 9.44487393e-01 -1.03079188e+00
-4.04840320e-01 -1.57716200e-01 1.40623664e-02 -1.53378749e+00
-2.19450772e-01 -7.60797381e-01 -1.64600700e-01 -1.04475045e+00
3.89068037e-01 -6.10029161e-01 -7.35090554e-01 -3.63082774e-02
-2.72477537e-01 9.24976543e-03 2.38256574e-01 5.45375586e-01
-7.44405627e-01 7.47207344e-01 1.04100633e+00 -1.35329023e-01
-3.10153753e-01 3.68812531e-01 -1.03407776e+00 2.30409965e-01
8.62705886e-01 -6.66402638e-01 -4.51877177e-01 -3.01622510e-01
4.92197603e-01 -5.44356890e-02 3.50334585e-01 -3.29848498e-01
-1.62715614e-02 5.96742518e-02 2.35124186e-01 -7.13439345e-01
5.75603724e-01 -6.84003949e-01 -1.49189994e-01 4.07322586e-01
-1.09847713e+00 -2.26443872e-01 -8.34340379e-02 1.23078752e+00
-6.69265017e-02 -1.64038077e-01 3.06202710e-01 -1.07584670e-01
-2.27293819e-01 1.77184373e-01 -2.20756814e-01 9.60494131e-02
7.53020585e-01 2.86196858e-01 -6.91308439e-01 -7.25597680e-01
-6.61068618e-01 3.96119624e-01 2.85505861e-01 4.46317315e-01
5.28478563e-01 -1.30583823e+00 -4.73318219e-01 -1.63962290e-01
2.04737231e-01 -6.39998496e-01 5.48153184e-02 8.51975203e-01
9.60944369e-02 9.71773207e-01 2.67658204e-01 -1.19674757e-01
-1.07005012e+00 6.91182375e-01 1.52912855e-01 -5.93744576e-01
-2.32467264e-01 6.83997273e-01 7.48775303e-01 -1.89451069e-01
1.93991289e-01 2.20341861e-01 2.96975020e-02 -4.05571721e-02
5.68683565e-01 4.20918256e-01 -2.33284812e-02 2.30123803e-01
-1.27590254e-01 -2.10148752e-01 -6.88791096e-01 -4.17026490e-01
1.24238288e+00 -2.65623868e-01 2.56188184e-01 4.87253547e-01
1.09396434e+00 3.65506023e-01 -9.13422525e-01 -5.57276607e-01
2.65193760e-01 -1.28616726e+00 2.27806494e-01 -1.23157775e+00
-7.58612752e-01 7.65000761e-01 5.60444593e-01 3.52935553e-01
7.23537147e-01 -9.57675725e-02 3.30783933e-01 6.19561672e-01
4.51856047e-01 -1.15820575e+00 1.91976145e-01 1.41780019e-01
1.09998918e+00 -1.06467617e+00 2.78521150e-01 -9.29520205e-02
-5.43325126e-01 3.81147504e-01 4.34133947e-01 -5.05289793e-01
8.03696752e-01 -3.88393819e-01 -5.91610558e-02 -2.91028738e-01
-1.20626640e+00 -3.51120323e-01 1.80232480e-01 4.89124835e-01
6.27553225e-01 1.66159108e-01 -7.77634501e-01 2.51032144e-01
1.11231357e-02 1.31708577e-01 4.49485540e-01 1.05657196e+00
-3.33271563e-01 -1.10008788e+00 -8.78623426e-02 8.16812873e-01
-5.41353822e-01 -3.41408819e-01 -7.25536287e-01 9.48447049e-01
-7.40142405e-01 7.91920602e-01 -5.63287362e-02 -1.78274676e-01
3.25251758e-01 5.52638806e-02 2.86431611e-01 -6.36696696e-01
-3.39246720e-01 8.31755996e-02 3.35267782e-01 -9.53334749e-01
-1.23900756e-01 -5.37869811e-01 -5.87766290e-01 -4.37192708e-01
-6.05747700e-01 4.83537704e-01 5.89574099e-01 4.93810803e-01
7.26709545e-01 2.01820210e-01 7.48524785e-01 -6.13329291e-01
-1.12476873e+00 -1.11253405e+00 -7.47099519e-01 5.01827896e-01
4.10148174e-01 -9.27981496e-01 -9.22966957e-01 -6.99109674e-01] | [9.732451438903809, 5.5290093421936035] |
dcbd04a0-2132-4398-b6ca-36abc4a49637 | non-parallel-text-style-transfer-with-self-1 | 2204.08123 | null | https://arxiv.org/abs/2204.08123v1 | https://arxiv.org/pdf/2204.08123v1.pdf | Non-Parallel Text Style Transfer with Self-Parallel Supervision | The performance of existing text style transfer models is severely limited by the non-parallel datasets on which the models are trained. In non-parallel datasets, no direct mapping exists between sentences of the source and target style; the style transfer models thus only receive weak supervision of the target sentences during training, which often leads the model to discard too much style-independent information, or utterly fail to transfer the style. In this work, we propose LaMer, a novel text style transfer framework based on large-scale language models. LaMer first mines the roughly parallel expressions in the non-parallel datasets with scene graphs, and then employs MLE training, followed by imitation learning refinement, to leverage the intrinsic parallelism within the data. On two benchmark tasks (sentiment & formality transfer) and a newly proposed challenging task (political stance transfer), our model achieves qualitative advances in transfer accuracy, content preservation, and fluency. Further empirical and human evaluations demonstrate that our model not only makes training more efficient, but also generates more readable and diverse expressions than previous models. | ['Soroush Vosoughi', 'Guangxuan Xu', 'Chenyan Jia', 'Chongyang Gao', 'Ruibo Liu'] | 2022-04-18 | non-parallel-text-style-transfer-with-self | https://openreview.net/forum?id=-TSe5o7STVR | https://openreview.net/pdf?id=-TSe5o7STVR | iclr-2022-4 | ['text-style-transfoer'] | ['natural-language-processing'] | [ 2.60404289e-01 2.59349406e-01 -2.26656541e-01 -7.17004836e-01
-6.00271761e-01 -7.98451722e-01 7.59313226e-01 -2.79700100e-01
-3.41684252e-01 9.05586064e-01 5.37392914e-01 -1.72058389e-01
5.32980323e-01 -7.60027766e-01 -6.79419816e-01 -2.73350567e-01
5.34183502e-01 8.01852047e-01 -1.61768362e-01 -6.64374053e-01
-2.16562841e-02 2.41948422e-02 -7.80876815e-01 5.76623142e-01
9.86016154e-01 6.65520549e-01 6.80484623e-02 4.63148892e-01
-3.69400799e-01 1.00192404e+00 -4.80017573e-01 -8.78716350e-01
1.18352011e-01 -1.02432036e+00 -1.16099596e+00 5.13293128e-03
4.95237559e-01 -4.42784816e-01 -3.25657398e-01 1.02493489e+00
3.65948141e-01 9.20817815e-03 8.92073452e-01 -9.53946769e-01
-1.03304470e+00 7.15424597e-01 -5.76222897e-01 -1.35376692e-01
2.47208744e-01 1.97371989e-01 1.11547112e+00 -8.68959069e-01
1.04496109e+00 1.59673202e+00 5.30141354e-01 1.05641758e+00
-1.52141881e+00 -7.88260520e-01 1.87900946e-01 -9.09862369e-02
-9.08156991e-01 -5.53399384e-01 1.01943207e+00 -2.57755727e-01
5.42350888e-01 -4.65373211e-02 6.21723413e-01 1.84490526e+00
8.92085433e-02 1.01245546e+00 1.58687091e+00 -3.89198542e-01
-1.84876714e-02 3.66545498e-01 -7.64246061e-02 8.23159695e-01
-2.14465156e-01 2.61271019e-02 -8.59843969e-01 2.67249376e-01
6.77750707e-01 -3.45427424e-01 -2.12325618e-01 -3.43445003e-01
-1.33286726e+00 9.11743045e-01 5.03290176e-01 2.33621478e-01
-1.43474760e-02 -1.15784757e-01 6.63233161e-01 8.11325967e-01
7.85909057e-01 6.36105716e-01 -4.26999211e-01 -3.13026100e-01
-7.47670293e-01 2.48651490e-01 9.55175698e-01 1.36722839e+00
8.09339523e-01 -7.60535151e-02 -3.83520722e-01 9.73134637e-01
4.25192602e-02 6.49246335e-01 4.63712662e-01 -7.65293062e-01
7.98891902e-01 6.50376558e-01 -1.69479966e-01 -8.54765713e-01
-1.48882106e-01 -2.54958034e-01 -1.04231906e+00 -2.91460883e-02
5.30186951e-01 -3.34499687e-01 -5.28346598e-01 2.22859526e+00
6.19412959e-02 -5.16550779e-01 1.30786687e-01 6.82799459e-01
7.65736401e-01 6.63165271e-01 1.18418202e-01 3.48983668e-02
1.20512354e+00 -1.19336736e+00 -6.28997207e-01 -4.89843816e-01
9.24025655e-01 -7.18256176e-01 1.61124420e+00 1.12136826e-01
-1.10609233e+00 -5.62360585e-01 -8.73050869e-01 -4.79971766e-01
-2.12891579e-01 9.53869745e-02 2.73051620e-01 2.02189475e-01
-1.00890720e+00 7.34088302e-01 -4.65583980e-01 -5.13970017e-01
7.39003897e-01 8.40826854e-02 -3.49657625e-01 -2.06694663e-01
-1.11222446e+00 1.04234183e+00 1.24857128e-01 -1.47965297e-01
-5.71618676e-01 -8.88347983e-01 -9.90907013e-01 -1.62332609e-01
7.05001503e-02 -1.02795231e+00 1.29292393e+00 -1.71345985e+00
-2.07624626e+00 1.32638180e+00 -1.43477142e-01 -1.17904522e-01
8.99882317e-01 -4.74336267e-01 -1.14182249e-01 -5.10084741e-02
2.46338814e-01 1.02274466e+00 7.95275331e-01 -1.09466243e+00
-2.57530808e-01 -3.00710887e-01 2.05742896e-01 4.64413077e-01
-6.48284435e-01 4.04471457e-02 -1.62149295e-01 -7.57703602e-01
-5.19461811e-01 -1.06914759e+00 1.10657305e-01 1.86238047e-02
-3.93448174e-01 -3.46455306e-01 5.94944000e-01 -7.89294362e-01
8.07063162e-01 -2.17091012e+00 7.65748322e-01 -1.32949695e-01
1.97033852e-01 1.17246635e-01 -4.46974069e-01 3.98308486e-01
2.24286228e-01 2.72447038e-02 -3.64353329e-01 -4.68825996e-01
1.95511356e-02 3.16707104e-01 -5.52311838e-01 2.34532371e-01
2.80095607e-01 1.23638916e+00 -1.10395157e+00 -5.97364902e-01
-2.07056895e-01 9.61887464e-02 -8.11336696e-01 5.69025695e-01
-3.77478391e-01 9.13422346e-01 -4.80521411e-01 1.42824605e-01
2.57552475e-01 -3.54294598e-01 2.35374808e-01 -2.73511142e-01
2.73537308e-01 4.39847410e-01 -4.57477868e-01 2.15925837e+00
-8.57826293e-01 6.80244923e-01 -1.57193199e-01 -7.66859770e-01
1.06567514e+00 6.11743927e-02 -3.68520821e-04 -8.81135941e-01
1.64612815e-01 1.66005865e-01 -1.09665096e-01 -4.01455402e-01
4.29871857e-01 -4.79962468e-01 -2.60525882e-01 7.22933769e-01
3.10536802e-01 -4.80755746e-01 1.31249160e-01 2.34504148e-01
7.76906967e-01 6.32057488e-01 1.53405175e-01 -6.15360200e-01
4.06301171e-01 -1.01239458e-01 5.27954519e-01 3.67290109e-01
-1.21535547e-01 5.13777733e-01 6.98791206e-01 -3.24849546e-01
-1.20489657e+00 -9.46640372e-01 2.17344955e-01 1.53147650e+00
8.11685845e-02 -5.13282597e-01 -9.41764534e-01 -1.10430038e+00
-1.14729039e-01 8.00410986e-01 -9.51112807e-01 -3.49203974e-01
-6.70386016e-01 -2.56854564e-01 7.67815650e-01 5.92779934e-01
6.50016308e-01 -1.32962191e+00 -7.69913793e-02 3.49192135e-02
-5.18941402e-01 -1.11226058e+00 -7.44156241e-01 -2.76652277e-01
-9.90227222e-01 -7.60634720e-01 -6.91273987e-01 -8.21649611e-01
7.96397924e-01 -1.76514387e-01 1.51858735e+00 1.29050026e-02
3.19650620e-01 1.41992062e-01 -3.41597319e-01 -4.63722080e-01
-7.77434886e-01 4.48198974e-01 -1.13370091e-01 7.06938803e-02
3.92743498e-01 -6.92130566e-01 -3.43550175e-01 1.24286748e-01
-5.83973944e-01 5.88556886e-01 5.47829926e-01 1.28476369e+00
2.41178989e-01 -8.35739493e-01 7.42609084e-01 -1.33080280e+00
5.91740310e-01 -2.67355591e-01 -2.68967986e-01 2.38102257e-01
-4.64179873e-01 2.24440962e-01 8.89320552e-01 -3.76191169e-01
-1.38339472e+00 -2.89678305e-01 8.19488913e-02 -3.35775018e-01
-1.73006639e-01 2.67470211e-01 -2.39535451e-01 3.31525922e-01
8.48821759e-01 2.00758025e-01 2.45845392e-01 -5.56248486e-01
5.57662070e-01 5.34420490e-01 3.47064555e-01 -1.08175278e+00
8.47044885e-01 3.25719208e-01 -1.22201860e-01 -8.70628178e-01
-1.43928850e+00 1.22286394e-01 -9.76208210e-01 1.16222762e-01
8.88085961e-01 -9.81055617e-01 -2.53603876e-01 4.20986384e-01
-1.36919689e+00 -8.87883961e-01 -4.63134825e-01 7.28904158e-02
-7.78933406e-01 1.68162033e-01 -1.01443136e+00 -1.99819878e-01
-4.98691231e-01 -8.65777433e-01 1.07262897e+00 -6.38791248e-02
-8.04850459e-01 -1.33229756e+00 2.39088267e-01 4.21672046e-01
4.98114675e-01 1.30357057e-01 1.31660306e+00 -5.32871425e-01
-1.28175035e-01 3.09081674e-01 -3.46008301e-01 4.76692379e-01
2.94655055e-01 -2.43205547e-01 -8.94766748e-01 -3.56347024e-01
-1.73891112e-01 -1.05266345e+00 5.46348035e-01 -2.43505642e-01
1.12656891e+00 -3.99594218e-01 -1.26140779e-02 8.70955646e-01
9.10642207e-01 -3.64173084e-01 4.53474313e-01 2.63786018e-01
1.01350188e+00 1.04182887e+00 3.58709186e-01 -1.02840714e-01
5.65380812e-01 6.10813141e-01 -2.17043892e-01 -4.15997535e-01
-3.81095767e-01 -7.77586758e-01 7.14733899e-01 1.35789514e+00
-2.27008909e-02 2.20114924e-02 -5.30176938e-01 4.77385193e-01
-1.63428950e+00 -8.00439179e-01 1.70127556e-01 1.67265201e+00
1.48733222e+00 1.14639988e-02 1.66221619e-01 -5.18143535e-01
3.63561004e-01 2.15873793e-01 -5.90454936e-01 -5.79712808e-01
-2.47228220e-01 3.37564170e-01 5.20499013e-02 4.78566617e-01
-8.64952505e-01 1.38493741e+00 5.85013580e+00 6.22488141e-01
-1.18122911e+00 1.75799876e-01 8.50112081e-01 -1.69285417e-01
-4.84128118e-01 -1.21407941e-01 -5.85643589e-01 3.17288458e-01
8.46378565e-01 -3.73718947e-01 5.06570220e-01 7.17180133e-01
4.44689207e-02 5.04094839e-01 -1.73833549e+00 8.06673706e-01
1.79571271e-01 -1.08086634e+00 3.86896968e-01 -1.03190735e-01
9.25263166e-01 -5.01798615e-02 2.32391581e-02 6.71285152e-01
5.84804058e-01 -1.06528866e+00 9.07404125e-01 3.21858972e-01
1.10895085e+00 -6.67778671e-01 4.86251056e-01 2.76612103e-01
-5.85773706e-01 3.15815061e-01 -3.80318850e-01 -2.51072884e-01
5.02998084e-02 2.84316868e-01 -4.79094923e-01 4.38426524e-01
5.30590832e-01 1.19288290e+00 -6.61644220e-01 -5.15169911e-02
-7.48597741e-01 7.97905445e-01 9.60558057e-02 -2.12740898e-01
1.87927768e-01 -4.30062115e-01 4.06730473e-01 1.49126422e+00
4.66594435e-02 1.40662780e-02 1.56343654e-01 1.18415380e+00
-6.10617578e-01 3.92979383e-01 -9.96242702e-01 -6.64259642e-02
1.12175070e-01 1.27457535e+00 -3.19224298e-01 -4.98739421e-01
-6.49069309e-01 1.39848661e+00 9.43275630e-01 5.43070734e-01
-4.43112701e-01 -1.42751589e-01 5.99176168e-01 -1.51980266e-01
6.44695610e-02 -1.57824919e-01 -6.30501390e-01 -1.56895602e+00
-3.00258454e-02 -1.28133333e+00 2.29262546e-01 -6.29239559e-01
-1.73622835e+00 6.34780645e-01 -1.51046157e-01 -1.05789948e+00
-4.00055647e-01 -7.11907387e-01 -6.44763589e-01 1.05173302e+00
-1.35391259e+00 -1.54751074e+00 -8.18829834e-02 6.40213251e-01
8.17151725e-01 -4.01454270e-01 9.55883026e-01 1.03939412e-04
-4.24601257e-01 9.00111139e-01 -8.54342431e-02 4.07417804e-01
1.22400999e+00 -1.42288363e+00 4.74860430e-01 5.00832021e-01
8.90597701e-02 6.14300966e-01 6.05281413e-01 -5.61955452e-01
-1.24399185e+00 -1.26544774e+00 9.81125653e-01 -7.50954151e-01
8.25120509e-01 -8.44569564e-01 -1.01223707e+00 9.66143966e-01
6.30832136e-01 -2.64659017e-01 7.35418797e-01 4.68263775e-01
-7.11780965e-01 -4.41951379e-02 -8.52995992e-01 9.86273348e-01
1.37798798e+00 -7.43579268e-01 -8.74177575e-01 3.40803504e-01
7.84219146e-01 -4.09017116e-01 -8.42088282e-01 1.26121014e-01
4.21400249e-01 -7.28096902e-01 5.41857839e-01 -1.09575319e+00
1.30428636e+00 2.38871053e-01 -5.29760197e-02 -1.74711227e+00
-3.84220988e-01 -6.14081919e-01 7.82302320e-02 1.63891685e+00
6.33778691e-01 -3.92054230e-01 5.19702673e-01 4.37008590e-01
3.08525618e-02 -6.33545578e-01 -5.53562224e-01 -6.18284523e-01
7.31127083e-01 1.69725999e-01 5.57511806e-01 1.29359019e+00
-3.55927786e-03 1.20537567e+00 -4.43869501e-01 -5.29111981e-01
5.43131530e-01 4.12755489e-01 1.06403434e+00 -1.02092326e+00
-5.88395715e-01 -6.58800960e-01 1.87192068e-01 -1.13589489e+00
7.70705819e-01 -1.31331360e+00 1.57466177e-02 -1.13941038e+00
3.00974935e-01 -4.02764827e-01 2.05520876e-02 5.41752279e-01
-2.30032846e-01 1.69237569e-01 2.03411400e-01 2.46419474e-01
-5.99891126e-01 1.05301178e+00 1.67698407e+00 -2.33517319e-01
-6.36364967e-02 -2.35609651e-01 -8.56929123e-01 8.79641056e-01
7.29811788e-01 -2.22573966e-01 -5.45125902e-01 -8.14012945e-01
3.48886460e-01 -2.43282095e-01 1.71741709e-01 -3.74344468e-01
-2.94410616e-01 -2.68232167e-01 5.52920401e-01 3.07616293e-02
2.52893809e-02 -3.98425967e-01 -5.56636035e-01 1.50739983e-01
-8.26034069e-01 2.29946718e-01 2.69310892e-01 3.43252063e-01
-1.59002468e-01 6.51045218e-02 9.07830954e-01 -7.98374191e-02
-4.03782219e-01 3.23853791e-01 4.42661569e-02 6.19012952e-01
5.68276048e-01 2.98957109e-01 -2.93875456e-01 -5.23528039e-01
-4.98930067e-01 2.00518698e-01 7.84034908e-01 7.27616072e-01
2.01689824e-01 -1.65060222e+00 -1.04364431e+00 8.44593495e-02
4.28356439e-01 1.74864590e-01 6.85817823e-02 6.67387784e-01
-2.02335984e-01 -1.69598274e-02 -4.98852223e-01 -5.87099731e-01
-9.94439721e-01 4.83077645e-01 2.62590200e-01 -4.36378032e-01
-7.74201155e-01 8.10605824e-01 7.23747373e-01 -9.51853633e-01
-2.01905534e-01 -3.24798860e-02 8.00391566e-03 -6.92108572e-02
3.70200187e-01 1.41923279e-01 -2.66712070e-01 -6.64496124e-01
7.31038824e-02 5.52933633e-01 -3.44074756e-01 -2.92159051e-01
1.33213472e+00 -1.78665698e-01 -3.35916072e-01 8.15163970e-01
1.29242933e+00 1.69182625e-02 -1.35682487e+00 -5.86508691e-01
-7.02809915e-03 -2.03405261e-01 -4.20092851e-01 -9.09707785e-01
-8.84756982e-01 1.19880176e+00 -7.04799816e-02 -3.15324783e-01
9.48567688e-01 2.09816620e-02 9.54310775e-01 4.66711342e-01
1.92146271e-01 -1.18848896e+00 3.37163478e-01 9.58327651e-01
1.23522699e+00 -1.22337425e+00 -3.62262458e-01 -3.35731864e-01
-8.10433328e-01 9.75552320e-01 9.82091844e-01 -1.21749699e-01
2.76902527e-01 3.09258867e-02 2.35262677e-01 -4.70972098e-02
-7.93153286e-01 2.89194554e-01 3.19308251e-01 5.39382100e-01
8.74785364e-01 1.19655430e-01 -1.45723760e-01 6.00037634e-01
-7.07355738e-01 -1.66331697e-02 2.41108760e-01 6.56806529e-01
4.71430980e-02 -1.31513643e+00 9.21516865e-02 2.99821585e-01
-1.73346236e-01 -1.34747773e-01 -8.53470922e-01 7.12376893e-01
-2.71231025e-01 5.52509606e-01 3.81665528e-02 -2.67505705e-01
6.05539739e-01 1.67307705e-01 8.58172238e-01 -7.42825687e-01
-5.91087878e-01 -3.08241755e-01 3.14114392e-01 -6.17597699e-01
-2.34684289e-01 -5.29262543e-01 -1.04451942e+00 -5.18349588e-01
3.30620617e-01 -5.17178811e-02 2.80849129e-01 1.21688068e+00
4.29302126e-01 5.08801401e-01 6.51772082e-01 -6.77687943e-01
-8.05805922e-01 -1.18867922e+00 -3.52510810e-01 1.06036651e+00
1.66112214e-01 -4.34073150e-01 -4.56184968e-02 2.50871420e-01] | [11.612604141235352, 9.545339584350586] |
bbadc28f-0909-4f21-a715-514277165f8f | challenges-in-monocular-visual-odometry | 1705.04300 | null | http://arxiv.org/abs/1705.04300v4 | http://arxiv.org/pdf/1705.04300v4.pdf | Challenges in Monocular Visual Odometry: Photometric Calibration, Motion Bias and Rolling Shutter Effect | Monocular visual odometry (VO) and simultaneous localization and mapping
(SLAM) have seen tremendous improvements in accuracy, robustness and
efficiency, and have gained increasing popularity over recent years.
Nevertheless, not so many discussions have been carried out to reveal the
influences of three very influential yet easily overlooked aspects: photometric
calibration, motion bias and rolling shutter effect. In this work, we evaluate
these three aspects quantitatively on the state of the art of direct,
feature-based and semi-direct methods, providing the community with useful
practical knowledge both for better applying existing methods and developing
new algorithms of VO and SLAM. Conclusions (some of which are
counter-intuitive) are drawn with both technical and empirical analyses to all
of our experiments. Possible improvements on existing methods are directed or
proposed, such as a sub-pixel accuracy refinement of ORB-SLAM which boosts its
performance. | ['Xiang Gao', 'Rui Wang', 'Nan Yang', 'Daniel Cremers'] | 2017-05-11 | null | null | null | null | ['monocular-visual-odometry'] | ['robots'] | [ 4.72282246e-02 -3.23267162e-01 -2.10114047e-01 -5.63200772e-01
-3.62968415e-01 -5.47755122e-01 9.30680811e-01 -2.04658955e-01
-4.38037038e-01 1.11611438e+00 1.00156926e-01 -3.47410768e-01
-2.04705149e-01 -5.37990212e-01 -4.95507717e-01 -5.41628599e-01
-2.87469745e-01 5.83324432e-01 4.04380232e-01 -3.52948755e-01
6.74355805e-01 9.62383568e-01 -1.69782567e+00 -6.87251866e-01
8.40876162e-01 6.78485394e-01 2.24764422e-01 4.92743760e-01
3.44256498e-02 5.13850570e-01 -2.74131060e-01 -2.63874400e-02
3.67957413e-01 -3.76981407e-01 -6.23924673e-01 -4.31073382e-02
5.24128199e-01 -8.82970393e-02 -1.71395361e-01 9.71518934e-01
5.43999434e-01 -6.97658435e-02 3.06888700e-01 -1.19971585e+00
-1.57138839e-01 -2.33229384e-01 -6.76441431e-01 -2.01166072e-03
7.01234877e-01 1.96773127e-01 4.68238950e-01 -8.22275281e-01
8.67956042e-01 9.48784351e-01 1.01105773e+00 -1.78223923e-02
-1.18550110e+00 -6.59425020e-01 -2.34031871e-01 2.98742384e-01
-1.69279122e+00 -6.30284429e-01 5.71713805e-01 -3.46157193e-01
9.13743854e-01 3.62828285e-01 7.18732476e-01 5.33774853e-01
5.44936597e-01 2.14048758e-01 1.40635979e+00 -6.06736004e-01
-8.37200284e-02 4.17642534e-01 -2.20894486e-01 7.33410656e-01
7.12184429e-01 4.47840542e-01 -6.40022635e-01 -2.69244254e-01
8.61702085e-01 -3.25748503e-01 -4.45034862e-01 -1.20782471e+00
-1.20525193e+00 7.73072124e-01 5.06942511e-01 2.34856293e-01
-1.66048720e-01 1.88498288e-01 1.22110613e-01 3.98565263e-01
3.39223593e-01 5.43689013e-01 -2.00916499e-01 -3.56601089e-01
-9.04756904e-01 3.84169728e-01 5.58028221e-01 1.11718249e+00
1.31734061e+00 -8.19640905e-02 7.32005656e-01 4.32220101e-01
5.05545855e-01 7.81253278e-01 1.40114307e-01 -1.09907413e+00
2.10390836e-01 5.35965562e-01 5.98064899e-01 -1.02806735e+00
-5.78145325e-01 -2.33428746e-01 -3.00470144e-01 4.69404578e-01
1.42612055e-01 -2.20877994e-02 -7.25274980e-01 1.12051749e+00
3.11588556e-01 -2.97390036e-02 -1.68765947e-01 9.21458006e-01
3.72073084e-01 2.14879870e-01 -4.95140076e-01 -8.52200910e-02
9.64580357e-01 -6.58313632e-01 -8.21310341e-01 -4.88289267e-01
6.15247071e-01 -1.36922085e+00 6.96211278e-01 2.83031493e-01
-6.56996489e-01 -4.07502651e-01 -1.29181886e+00 -1.43157229e-01
-2.80259848e-01 -1.08718291e-01 9.27517712e-01 7.70156622e-01
-1.12875938e+00 5.09550631e-01 -9.33747530e-01 -1.00003219e+00
-2.16703251e-01 4.81061608e-01 -3.65431368e-01 -2.79204935e-01
-9.92145181e-01 1.48615158e+00 -4.48851734e-02 1.30147219e-01
-2.38082632e-01 -2.66088277e-01 -1.05861139e+00 -7.13384748e-01
1.64709941e-01 -8.32221508e-01 9.38217580e-01 -6.08684123e-01
-1.59157503e+00 1.09884489e+00 -7.00174809e-01 -4.78693634e-01
7.21617103e-01 -1.54680952e-01 -4.75702703e-01 -2.80388951e-01
2.47356132e-01 5.05667627e-01 2.49264911e-01 -1.34822226e+00
-7.08774447e-01 -6.09662175e-01 -1.94541365e-01 3.59660834e-01
6.60962880e-01 -8.26353058e-02 -2.70306081e-01 8.22416320e-02
9.41337109e-01 -1.15417397e+00 -2.27692172e-01 9.67570841e-02
-2.11527050e-02 2.36349717e-01 7.08228528e-01 -3.85802060e-01
7.12701738e-01 -2.00861144e+00 -1.52976707e-01 1.13459148e-01
-2.45389774e-01 -9.13814083e-02 2.94128567e-01 9.62123573e-01
3.66217613e-01 -2.69346863e-01 -1.68686081e-02 -3.19196969e-01
-2.20320836e-01 4.25447553e-01 -2.78304309e-01 1.27895653e+00
-2.83402294e-01 7.79184818e-01 -8.25087368e-01 -2.44813696e-01
9.32815969e-01 4.94731158e-01 -4.81109582e-02 -1.84551522e-01
3.50366980e-01 7.40899384e-01 -7.27893934e-02 8.10803533e-01
9.68079627e-01 1.45467967e-01 7.33930916e-02 1.39944732e-01
-8.26094449e-01 7.52983153e-01 -1.49478269e+00 1.88846743e+00
-4.16445106e-01 1.05123591e+00 1.21710263e-01 -3.29895884e-01
1.18928051e+00 -8.09368342e-02 2.29480535e-01 -9.20182228e-01
-4.66691256e-02 6.75250292e-01 -1.93342730e-01 -2.84395128e-01
1.18917954e+00 -3.34966421e-01 1.68063641e-01 -3.14101167e-02
-1.94536030e-01 -4.87938404e-01 -1.84671044e-01 -2.61525847e-02
5.64833403e-01 7.33786047e-01 8.62170637e-01 -6.13701284e-01
6.49593651e-01 6.40124440e-01 2.51104057e-01 7.31034636e-01
-4.24272925e-01 5.27239740e-01 -9.41928476e-03 -3.67997408e-01
-1.05813909e+00 -8.65114033e-01 -5.80630004e-01 3.15016180e-01
1.01874948e+00 -2.29840577e-01 6.54208288e-02 8.89055207e-02
3.84876609e-01 4.00975376e-01 -3.86652321e-01 2.20957309e-01
-5.02219141e-01 -5.71389675e-01 3.12405080e-01 2.59165317e-01
4.20189559e-01 -4.39712435e-01 -8.07091177e-01 2.94030994e-01
-1.11628346e-01 -9.99832511e-01 3.47508013e-01 1.12377986e-01
-1.23115861e+00 -1.03710949e+00 -5.15847564e-01 -4.66471970e-01
5.61570168e-01 1.02972841e+00 8.34000945e-01 1.67315137e-02
-9.82106626e-02 3.07758361e-01 -3.24891657e-01 -3.59862804e-01
-9.96703878e-02 -9.07235444e-02 2.70676017e-01 -3.68443549e-01
4.80573833e-01 -5.76139867e-01 -4.38266963e-01 7.59896159e-01
-2.14273855e-01 -9.12194848e-02 4.23905790e-01 4.46428418e-01
4.95102018e-01 -6.00404501e-01 -2.10448623e-01 -6.97565734e-01
-1.86205152e-02 -9.46367308e-02 -1.12823069e+00 -3.53908271e-01
-1.03786850e+00 5.09642921e-02 -1.25912484e-03 2.23972037e-01
-9.68074143e-01 7.58258849e-02 -4.13452797e-02 1.61278442e-01
-1.55253053e-01 2.87606627e-01 2.13799044e-01 -1.06159532e+00
8.91583562e-01 3.09318066e-01 1.62807763e-01 -4.99085933e-01
3.03420931e-01 7.36598134e-01 6.89809144e-01 -2.86045045e-01
1.14029968e+00 1.23034751e+00 2.42642462e-01 -1.03105080e+00
-3.17318588e-01 -1.01851439e+00 -8.00724685e-01 -1.33433655e-01
2.79609740e-01 -1.07111788e+00 -2.73485661e-01 5.02890468e-01
-8.91721129e-01 -5.34339361e-02 1.64729799e-03 8.08596015e-01
-6.23300910e-01 6.36379540e-01 -1.70317069e-01 -9.90646303e-01
2.36765176e-01 -1.12300193e+00 1.10532594e+00 3.44491571e-01
-3.37114453e-01 -1.03699780e+00 5.26514828e-01 2.30223596e-01
6.02551460e-01 3.17108959e-01 8.98580179e-02 2.34418616e-01
-9.09062088e-01 -3.37956816e-01 -3.30898523e-01 -3.12990278e-01
2.19731525e-01 -6.70191795e-02 -1.06581259e+00 -6.02339685e-01
-3.85944955e-02 -6.57749251e-02 6.03244066e-01 3.03990394e-01
4.96878922e-02 3.29061180e-01 -7.26717591e-01 8.86992097e-01
1.93721116e+00 1.57807857e-01 9.28137302e-01 1.05224073e+00
3.34902912e-01 5.23265362e-01 1.09189916e+00 2.59458125e-01
4.95744050e-01 1.05151427e+00 7.05633938e-01 -9.82774869e-02
-3.10197175e-02 -2.29369625e-01 7.09174275e-02 4.02760178e-01
-5.19528031e-01 3.48252505e-01 -1.00227356e+00 6.52554929e-01
-1.77303410e+00 -6.57359004e-01 -7.86513090e-01 2.61292458e+00
1.27576753e-01 -6.45488352e-02 -3.24777693e-01 5.55629879e-02
4.85739589e-01 4.44087744e-01 -1.31228179e-01 -2.00139210e-01
-2.39968568e-01 -1.45927951e-01 1.14009976e+00 9.50114191e-01
-8.20679545e-01 9.65342104e-01 6.58833599e+00 7.98124224e-02
-1.51263845e+00 2.08885246e-03 -3.37032825e-01 2.37756655e-01
-2.32945934e-01 5.77590168e-01 -9.58275080e-01 1.08045124e-01
6.05606794e-01 -1.86050028e-01 4.17740732e-01 8.50729883e-01
2.96048313e-01 -9.85029042e-01 -5.87837994e-01 1.10357952e+00
1.80191681e-01 -1.22152936e+00 -5.16970217e-01 4.61031854e-01
9.52810705e-01 6.57466590e-01 -3.78233522e-01 6.74036071e-02
1.93810374e-01 -5.78382313e-01 7.23719716e-01 2.69156605e-01
5.18875957e-01 -6.12538457e-01 1.05591631e+00 3.23163152e-01
-1.23571575e+00 3.05192381e-01 -6.39926910e-01 -7.93181181e-01
3.29239637e-01 5.61397731e-01 -8.08316052e-01 1.23695672e+00
5.67218781e-01 6.37058556e-01 -6.64367616e-01 1.41695774e+00
-4.11036551e-01 -7.60265952e-03 -4.98088211e-01 -1.01376370e-01
4.12875533e-01 -3.53462964e-01 5.99142492e-01 8.44450891e-01
3.05355400e-01 -3.90889645e-01 -1.77241609e-01 6.93341792e-01
6.90423369e-01 -2.26893201e-02 -1.04578960e+00 5.15460789e-01
4.41223264e-01 1.04103196e+00 -5.83110154e-01 -1.27538398e-01
-5.72506905e-01 7.82389164e-01 -3.80869657e-02 1.56370416e-01
-6.47671878e-01 -2.64359206e-01 1.03220558e+00 3.48155528e-01
2.17219722e-02 -8.50275397e-01 -6.57530367e-01 -1.08842695e+00
9.72565114e-02 -3.18954080e-01 1.76948344e-03 -9.14591372e-01
-6.01970851e-01 2.56101340e-01 3.17237191e-02 -1.52361608e+00
-4.00838524e-01 -4.77925867e-01 -1.86201423e-01 1.03793383e+00
-1.93424582e+00 -8.87762010e-01 -6.67033195e-01 2.44034261e-01
1.26043066e-01 2.60570407e-01 6.65872335e-01 1.78966239e-01
2.20091328e-01 -1.04239456e-01 3.77676308e-01 -3.98377746e-01
1.02872908e+00 -8.60807717e-01 6.37112916e-01 1.09349275e+00
2.24356577e-01 7.58793890e-01 1.23760629e+00 -5.84115088e-01
-1.67672765e+00 -4.95799422e-01 1.40713239e+00 -8.12368691e-01
6.94970667e-01 -2.95348823e-01 -5.82534611e-01 9.18788671e-01
-6.65658340e-02 -1.73464715e-01 4.14412692e-02 2.40047842e-01
1.49609193e-01 -8.53740275e-02 -1.05481911e+00 4.30946469e-01
1.05981505e+00 -5.68046689e-01 -6.83000922e-01 8.27880800e-02
2.21717268e-01 -7.97241747e-01 -4.48848933e-01 4.61493462e-01
7.79006720e-01 -1.60521483e+00 7.63262212e-01 1.69306070e-01
-4.27043378e-01 -6.17938280e-01 -2.44816408e-01 -9.89297628e-01
-1.59637794e-01 -6.15493059e-01 4.81278151e-01 8.96772742e-01
4.89404015e-02 -1.31306708e+00 8.08443725e-01 6.26491532e-02
-6.75004274e-02 -3.11017632e-01 -1.04406440e+00 -1.14376032e+00
-4.72052783e-01 -5.08402526e-01 4.42983687e-01 1.05371988e+00
-7.26174787e-02 7.54718529e-03 -5.56162179e-01 4.88919348e-01
6.86362207e-01 3.17067415e-01 1.47312379e+00 -1.31982648e+00
1.93487227e-01 -2.41579283e-02 -9.39995050e-01 -1.21152496e+00
-5.28586805e-01 -2.35530570e-01 5.87134622e-02 -1.58047402e+00
-1.83729246e-01 -3.99857640e-01 1.37393534e-01 -1.18092619e-01
2.60940254e-01 5.35974562e-01 -1.36716992e-01 5.92728794e-01
-2.77362674e-01 3.18488449e-01 1.02170908e+00 3.67384493e-01
-1.84008345e-01 2.58312039e-02 -2.48349845e-01 6.66290998e-01
4.13171917e-01 -6.44844592e-01 -1.92267410e-02 -4.75948840e-01
3.58694702e-01 -7.62647716e-03 4.65482116e-01 -1.26027238e+00
2.41607070e-01 -2.25013584e-01 1.77824095e-01 -8.43144238e-01
4.45279449e-01 -8.86946499e-01 5.37858784e-01 7.49788344e-01
6.41113818e-01 8.87876675e-02 1.16289474e-01 3.76873642e-01
-3.57790500e-01 -1.63182870e-01 8.34886789e-01 -2.62842476e-01
-1.26233578e+00 -2.37615444e-02 -7.29656145e-02 -4.66799647e-01
9.18802738e-01 -7.60298073e-01 -1.74817577e-01 -6.30534887e-01
-3.37301135e-01 1.71781644e-01 1.41593635e+00 4.16792184e-01
1.76970512e-01 -1.15966523e+00 -4.04675424e-01 3.92850399e-01
4.93479788e-01 -1.78133786e-01 -9.87073854e-02 1.34103298e+00
-1.08332515e+00 9.31064010e-01 -4.37900305e-01 -9.63768423e-01
-1.18083894e+00 2.71451592e-01 2.83480525e-01 1.53390974e-01
-5.54643989e-01 5.87145090e-01 -2.72168040e-01 -6.97357893e-01
4.24459912e-02 -1.21733434e-01 2.84914315e-01 -1.73817620e-01
2.38603711e-01 7.54053473e-01 1.27844229e-01 -8.88341010e-01
-7.61373162e-01 1.13149202e+00 4.60720003e-01 -2.82618552e-01
1.05336559e+00 -7.84840405e-01 -2.63351262e-01 6.87259436e-01
7.67272353e-01 3.34535718e-01 -1.14680529e+00 -4.47507426e-02
1.05521038e-01 -9.39262331e-01 -4.67207544e-02 -4.89349782e-01
-2.66145527e-01 8.23351681e-01 7.79915929e-01 4.21844609e-02
7.20953882e-01 -4.53190058e-02 2.36619458e-01 2.87756771e-01
1.29324281e+00 -7.30157375e-01 -6.80247605e-01 6.91380739e-01
5.84485233e-01 -1.37986088e+00 5.75909197e-01 -5.70579171e-01
-2.08921537e-01 1.07200181e+00 1.82092056e-01 -2.95923591e-01
1.35591522e-01 1.49081111e-01 6.06551230e-01 -1.63129881e-01
-5.72907552e-02 -3.85819435e-01 -1.04051173e-01 7.64361739e-01
4.62399334e-01 -1.05243593e-01 -7.19079077e-01 -6.18227661e-01
-1.88646480e-01 1.67198315e-01 6.04731679e-01 1.16371870e+00
-8.01384807e-01 -1.17712450e+00 -8.07710111e-01 -8.76908153e-02
1.52341709e-01 1.43923506e-01 -3.92143130e-01 1.54977834e+00
4.14462313e-02 6.24659240e-01 -4.89763394e-02 -2.89551973e-01
2.36209765e-01 2.11586934e-02 7.24022388e-01 -2.74533093e-01
-1.38449162e-01 -8.88068452e-02 3.79800886e-01 -7.95580566e-01
-6.32702053e-01 -9.57119763e-01 -1.04255211e+00 -6.12250805e-01
-5.61478794e-01 8.75717998e-02 1.36146247e+00 9.37671721e-01
3.26289922e-01 -8.00086334e-02 3.21141571e-01 -1.29592478e+00
-3.87438804e-01 -9.09492075e-01 -7.00074792e-01 -1.79858446e-01
4.06960338e-01 -9.84740496e-01 -4.92869318e-01 -4.40578520e-01] | [7.365614891052246, -2.105556011199951] |
9fda07dd-aa60-46d7-a68b-565a423c93f6 | learning-to-learn-group-alignment-a-self | 2304.07337 | null | https://arxiv.org/abs/2304.07337v1 | https://arxiv.org/pdf/2304.07337v1.pdf | Learning to Learn Group Alignment: A Self-Tuning Credo Framework with Multiagent Teams | Mixed incentives among a population with multiagent teams has been shown to have advantages over a fully cooperative system; however, discovering the best mixture of incentives or team structure is a difficult and dynamic problem. We propose a framework where individual learning agents self-regulate their configuration of incentives through various parts of their reward function. This work extends previous work by giving agents the ability to dynamically update their group alignment during learning and by allowing teammates to have different group alignment. Our model builds on ideas from hierarchical reinforcement learning and meta-learning to learn the configuration of a reward function that supports the development of a behavioral policy. We provide preliminary results in a commonly studied multiagent environment and find that agents can achieve better global outcomes by self-tuning their respective group alignment parameters. | ['Kyle Tilbury', 'David Radke'] | 2023-04-14 | null | null | null | null | ['hierarchical-reinforcement-learning'] | ['methodology'] | [-2.63154745e-01 3.59808028e-01 -1.96955934e-01 -1.57739744e-01
-2.72842616e-01 -5.60364842e-01 6.04236901e-01 3.42454582e-01
-5.33010781e-01 1.21929693e+00 1.29783601e-01 2.10875496e-01
-3.54466140e-01 -7.08594382e-01 -3.76614153e-01 -9.36533868e-01
-2.31136143e-01 9.80591297e-01 2.29801998e-01 -5.99187315e-01
4.31061178e-01 -1.74969696e-02 -1.62795055e+00 -1.23698816e-01
9.45434391e-01 2.13691369e-01 4.00161386e-01 8.10478210e-01
3.18033576e-01 9.44016099e-01 -1.13670087e+00 1.15040570e-01
5.26798785e-01 -6.24142885e-01 -4.95822757e-01 5.99733531e-01
-1.40492961e-01 -3.18952918e-01 3.97112966e-01 6.83815479e-01
6.66510999e-01 5.57282791e-02 6.14421546e-01 -1.56755865e+00
-3.02780420e-01 1.08347845e+00 -5.01254916e-01 3.29420477e-01
2.66054660e-01 3.59790415e-01 1.01546276e+00 1.76651791e-01
4.51529950e-01 1.29592013e+00 3.07655722e-01 5.21615446e-01
-1.37028074e+00 -5.52722573e-01 3.45126688e-01 -3.37257124e-02
-9.58471894e-01 -1.42451316e-01 3.58724415e-01 -3.39493573e-01
9.43127930e-01 -1.82650149e-01 1.08495128e+00 5.88589728e-01
3.92988712e-01 2.99017966e-01 1.30833554e+00 -6.66088760e-01
4.77711856e-01 9.19395015e-02 -4.86501217e-01 5.37207842e-01
5.62189579e-01 1.72081560e-01 -4.17193621e-01 -4.23164755e-01
1.04522240e+00 -4.77553338e-01 1.86883956e-01 -8.02669466e-01
-9.08974111e-01 9.18022037e-01 1.59007236e-01 4.24286991e-01
-7.74087012e-01 4.23246056e-01 -4.04433236e-02 9.75243092e-01
8.51078108e-02 1.11943078e+00 -3.61709595e-01 -2.85267174e-01
-1.66930512e-01 6.15188420e-01 9.03664947e-01 3.70001853e-01
9.40732598e-01 1.71794936e-01 2.10334267e-03 7.93623507e-01
2.78424501e-01 2.40386620e-01 1.93883508e-01 -1.72135389e+00
2.64598846e-01 7.99538076e-01 6.65592790e-01 -4.56635654e-01
-5.30256152e-01 -4.81991947e-01 -7.29802921e-02 7.47256875e-01
5.36332905e-01 -1.05828679e+00 -1.87517256e-01 1.98808229e+00
4.75055337e-01 -2.46126875e-01 3.23873878e-01 5.99954605e-01
-7.80739635e-02 3.33206773e-01 -6.44882694e-02 -5.66349447e-01
8.64921749e-01 -9.35452342e-01 -5.48189521e-01 -2.52500206e-01
6.17010355e-01 -4.93588775e-01 4.71951783e-01 2.67656505e-01
-1.48231125e+00 -1.78793877e-01 -7.91528225e-01 9.98161316e-01
1.51767761e-01 -4.06649351e-01 6.41912818e-01 7.16257274e-01
-1.51082933e+00 5.47368884e-01 -8.30750167e-01 -4.81974214e-01
3.32098603e-02 9.32147324e-01 2.17377558e-01 4.94959086e-01
-8.48244607e-01 1.16737187e+00 2.41018921e-01 -3.31196815e-01
-8.78367662e-01 -1.83511421e-01 -5.18388808e-01 -6.95834914e-03
8.51489305e-01 -1.08643270e+00 1.63929582e+00 -1.45749164e+00
-1.92003834e+00 2.12874919e-01 6.23555243e-01 -3.27844203e-01
2.51555443e-01 3.75748068e-01 3.50992471e-01 -1.01772964e-01
2.71452904e-01 7.98491120e-01 7.49772191e-01 -1.41875136e+00
-9.43098187e-01 -1.22074075e-01 4.40722317e-01 1.13965893e+00
-4.39818889e-01 1.55394375e-01 3.52801234e-01 -2.32656896e-01
-4.65417832e-01 -1.04377961e+00 -8.69749367e-01 -8.70983779e-01
1.42048553e-01 -4.28068340e-01 2.22939253e-01 3.54209810e-01
8.05093527e-01 -1.57501841e+00 6.02602303e-01 2.78594136e-01
2.79198498e-01 -2.55586952e-01 -3.45202148e-01 8.48848403e-01
2.92502850e-01 -4.49023619e-02 4.52467769e-01 -5.24628982e-02
5.81853539e-02 5.14699638e-01 3.65041167e-01 1.56080768e-01
-3.36445719e-02 3.95804852e-01 -9.21496391e-01 -2.43663147e-01
-2.77032644e-01 -1.22447364e-01 -9.25515831e-01 4.26742792e-01
-2.81581402e-01 5.31707048e-01 -5.54336667e-01 2.97045916e-01
7.82504901e-02 -3.44550312e-01 8.74108851e-01 8.14166009e-01
-4.25227612e-01 3.20579931e-02 -1.54338849e+00 7.42486537e-01
-3.72659750e-02 1.24801524e-01 6.29863679e-01 -9.25930560e-01
9.26563680e-01 4.80025738e-01 9.16247666e-01 -2.87378132e-01
3.41512859e-01 1.30855083e-01 7.36156523e-01 -2.35407740e-01
7.19069541e-01 -1.75903499e-01 -7.91997612e-02 1.12859094e+00
-6.67839572e-02 -3.29003632e-01 5.19171298e-01 9.41852704e-02
1.27301455e+00 -1.77583426e-01 3.58782828e-01 -3.34485441e-01
1.45585150e-01 1.94248095e-01 6.01433873e-01 1.31665182e+00
-3.39291304e-01 -1.15124345e-01 7.82941997e-01 -2.79511899e-01
-1.13482130e+00 -6.97415769e-01 2.78551131e-01 1.54131699e+00
3.04028243e-01 -2.15829074e-01 -7.46224761e-01 -3.46483886e-01
1.32228822e-01 2.24053964e-01 -7.59659469e-01 1.55286163e-01
-7.24610567e-01 -1.05013895e+00 -5.85473962e-02 2.48121336e-01
2.96766758e-01 -1.30257487e+00 -1.00577307e+00 5.41545928e-01
1.69268250e-01 -5.02504945e-01 -6.51921809e-01 4.03207839e-01
-6.64010108e-01 -1.14370000e+00 -5.09355426e-01 -7.58122027e-01
9.14712906e-01 3.30102205e-01 1.06303668e+00 4.36958224e-01
6.42336160e-02 9.28590059e-01 -3.24326217e-01 -5.74072480e-01
-8.00770819e-01 3.21774662e-01 3.84893775e-01 -4.46306586e-01
-1.72625333e-01 -6.84953868e-01 -5.08964062e-01 6.32014811e-01
-7.78796315e-01 -6.55284077e-02 5.08665085e-01 8.95820796e-01
-3.41193080e-02 2.43818253e-01 1.01820791e+00 -5.73273420e-01
1.36455035e+00 -4.93385822e-01 -9.79099154e-01 2.73559868e-01
-7.78453410e-01 2.59969801e-01 4.86844391e-01 -7.05985010e-01
-8.67376924e-01 -1.87501073e-01 6.86937809e-01 2.67812401e-01
8.42466727e-02 2.37308726e-01 2.80354857e-01 -2.38055140e-01
6.28501415e-01 -2.42111161e-01 5.86144567e-01 1.05897941e-01
2.60122567e-01 3.20169151e-01 -1.04544036e-01 -9.82674360e-01
6.44912004e-01 3.57414689e-03 -1.44348472e-01 -2.11492285e-01
-3.01013499e-01 -1.54964209e-01 -2.81761676e-01 -6.27160430e-01
3.77288818e-01 -8.86814833e-01 -1.25340724e+00 5.59955597e-01
-7.25384891e-01 -8.24582994e-01 -3.66867870e-01 3.35371047e-01
-9.18672621e-01 4.05179448e-02 -4.74779308e-01 -8.29287052e-01
6.81261718e-02 -1.10703921e+00 4.98756617e-01 6.36059940e-01
-1.69179156e-01 -1.02859235e+00 7.04274237e-01 2.75340319e-01
7.11485565e-01 -4.41233814e-02 3.97204757e-01 -5.06870687e-01
-7.82967269e-01 5.20910323e-01 5.15214503e-01 -1.05055824e-01
3.81740332e-01 4.98826839e-02 -4.05455902e-02 -7.24779010e-01
-2.19238475e-01 -7.25039899e-01 1.35830775e-01 7.49861896e-01
1.68262362e-01 -4.35832828e-01 -2.17889339e-01 -1.03405148e-01
1.16159773e+00 5.22445679e-01 2.51961909e-02 1.01499379e+00
5.52724227e-02 7.11560905e-01 5.58897734e-01 8.75973880e-01
9.55504775e-01 7.99238682e-01 5.69348693e-01 2.91998953e-01
4.03525472e-01 3.30061227e-01 6.14388704e-01 4.59641397e-01
-5.32734931e-01 -8.47240537e-02 -7.19592988e-01 3.78431141e-01
-2.33541107e+00 -1.27398133e+00 4.62158561e-01 1.94419861e+00
1.03188181e+00 1.76916078e-01 1.01791227e+00 -2.40186840e-01
7.97501385e-01 -1.78253189e-01 -5.33164263e-01 -5.39177477e-01
-1.86719373e-01 -1.41101241e-01 5.79658210e-01 7.54343212e-01
-6.81592584e-01 9.36001658e-01 7.50699663e+00 8.74414295e-02
-6.31496966e-01 -1.01780549e-01 4.31708604e-01 -3.49626392e-01
-2.30230898e-01 1.93208665e-01 -8.40603888e-01 2.10224271e-01
8.05593431e-01 -4.34096634e-01 9.63036656e-01 3.99386317e-01
5.10336518e-01 -5.49657881e-01 -8.00476849e-01 1.06775671e-01
-5.85353412e-02 -1.18447638e+00 -4.90769982e-01 4.95418251e-01
1.22591197e+00 -8.48337039e-02 -1.03486657e-01 3.01413953e-01
1.52023852e+00 -8.17125797e-01 5.39886355e-01 2.09209099e-01
-1.66886151e-01 -8.96102488e-01 5.90052664e-01 5.51308095e-01
-8.38413119e-01 -8.24526250e-01 -8.48014727e-02 -8.23414147e-01
-3.84125449e-02 -3.48288685e-01 -1.28281093e+00 1.78669453e-01
5.71060181e-01 1.58598393e-01 -4.54480499e-01 1.11900365e+00
-3.72798890e-02 3.21449459e-01 -2.63708055e-01 -4.21111971e-01
2.68749475e-01 -4.50978816e-01 6.45171762e-01 2.88444370e-01
1.80507958e-01 6.93241358e-02 7.74885416e-01 5.06860793e-01
2.55515814e-01 8.40746313e-02 -4.54437345e-01 2.61853009e-01
8.02928269e-01 1.14786041e+00 -8.60378981e-01 -1.77775696e-01
1.02957236e-02 2.12758273e-01 5.45633316e-01 2.36643359e-01
-5.13771236e-01 6.18147179e-02 6.88664734e-01 -8.16931501e-02
5.04684031e-01 -2.25129098e-01 -7.06814080e-02 -9.30666447e-01
-4.54359859e-01 -1.29612410e+00 4.76636320e-01 -4.65094835e-01
-1.10187447e+00 3.48179817e-01 1.76917136e-01 -9.48649824e-01
-8.13630223e-01 -6.38227863e-03 -7.29398966e-01 3.19941521e-01
-1.13260555e+00 -5.44115484e-01 -5.48252761e-02 3.06518257e-01
2.47578070e-01 -7.09908307e-01 5.15121758e-01 -3.56171399e-01
-7.99312592e-01 2.59457350e-01 1.92288741e-01 -3.67475897e-01
7.09499717e-01 -1.52546489e+00 -4.08300906e-01 2.66680360e-01
-2.15934709e-01 3.09857816e-01 1.08331871e+00 -6.12091959e-01
-1.22533739e+00 -3.80344510e-01 1.32460088e-01 -2.83797562e-01
8.31229568e-01 4.72296514e-02 -4.69242752e-01 5.63737333e-01
9.83473420e-01 -6.03976011e-01 6.69530630e-01 3.32744330e-01
2.61043191e-01 -2.63956249e-01 -1.01041365e+00 8.21275711e-01
7.25619256e-01 2.60522068e-01 -3.49748939e-01 1.27057046e-01
2.19925627e-01 -3.45292598e-01 -8.91194522e-01 1.55495897e-01
2.39608169e-01 -9.99739110e-01 5.04276752e-01 -4.30419326e-01
-7.87640456e-03 -4.06631142e-01 1.76320255e-01 -2.07420063e+00
-5.27331412e-01 -9.39922690e-01 3.40065122e-01 9.63796616e-01
2.66580373e-01 -8.71731639e-01 1.00335741e+00 5.10590434e-01
7.53751174e-02 -4.32126522e-01 -7.74597287e-01 -6.47656500e-01
2.88796067e-01 6.69153810e-01 5.13088107e-01 8.02763820e-01
4.55674917e-01 5.59082866e-01 -3.22992384e-01 2.29883939e-01
7.41489828e-01 1.25284523e-01 9.96602237e-01 -1.19641662e+00
-8.05314422e-01 -8.29090536e-01 -3.80688533e-02 -5.82222164e-01
5.26214838e-02 -2.69525111e-01 1.04086787e-01 -1.43202806e+00
2.93957353e-01 -8.15050125e-01 -4.00378630e-02 6.25307798e-01
-8.72608945e-02 -2.58257180e-01 4.51681465e-01 2.30512351e-01
-1.21696222e+00 4.09723878e-01 1.38526464e+00 1.05367757e-01
-6.95862293e-01 6.86187595e-02 -1.10771048e+00 4.28119808e-01
1.08153403e+00 -5.71127653e-01 -3.96903783e-01 -2.33910352e-01
3.78609300e-01 6.13337755e-01 -2.79660493e-01 -6.97526336e-01
4.19828385e-01 -8.31564069e-01 -1.19743794e-01 8.29275027e-02
8.95225033e-02 -4.12066460e-01 2.40881294e-01 8.46129000e-01
-4.99739707e-01 5.39154768e-01 -8.97370428e-02 4.94033307e-01
1.10268310e-01 -4.36401367e-01 5.29431403e-01 -3.81751060e-01
-1.39122084e-01 -2.65465617e-01 -1.15235603e+00 3.68185900e-02
1.41500771e+00 -2.70713508e-01 -5.23465157e-01 -8.41654122e-01
-5.90254188e-01 9.16387975e-01 8.06435108e-01 1.80719718e-01
1.15735479e-01 -9.58868265e-01 -8.63440573e-01 -2.48724625e-01
-2.48337716e-01 -2.17648193e-01 -1.18621454e-01 5.36394715e-01
-2.06219584e-01 1.76303759e-02 -8.22457850e-01 -3.89185011e-01
-1.39057422e+00 5.69037907e-02 8.10587227e-01 -4.99716431e-01
-3.73586528e-02 4.54513460e-01 -2.15451673e-01 -4.18137938e-01
2.31135845e-01 5.14307283e-02 -5.33077896e-01 2.03265876e-01
2.02194765e-01 1.66049391e-01 -5.00085115e-01 -3.35392088e-01
-5.41983843e-02 3.15715283e-01 -1.03311487e-01 -4.93233114e-01
1.59316659e+00 -2.94586033e-01 -3.05973887e-01 7.64102340e-02
-6.94038812e-04 -5.56740463e-02 -1.65395308e+00 -1.17553689e-01
-3.52292061e-02 -3.52988809e-01 -2.58715481e-01 -6.65116429e-01
-8.08287263e-01 -2.39079222e-01 2.18799695e-01 8.47666740e-01
9.73026395e-01 1.33115530e-01 -2.03064680e-01 6.42145157e-01
5.90026140e-01 -1.53760755e+00 9.49553072e-01 5.99746287e-01
5.31638741e-01 -1.07466519e+00 1.40291184e-01 2.32962659e-03
-9.36739266e-01 1.06918204e+00 1.09838462e+00 -2.16814294e-01
3.09593737e-01 4.96082366e-01 2.84444541e-01 -2.93012500e-01
-1.52784801e+00 -3.42946529e-01 -5.42237520e-01 8.04878235e-01
1.38812676e-01 1.73300192e-01 -5.41568041e-01 -9.19593498e-02
-2.50929981e-01 -3.62479836e-01 1.28720617e+00 1.20620430e+00
-9.75381434e-01 -1.81993842e+00 -8.55362535e-01 4.05627251e-01
-2.40199625e-01 4.57865715e-01 -4.15783972e-01 6.20123982e-01
1.67496786e-01 1.25044692e+00 2.97266632e-01 1.76850721e-01
1.45898223e-01 -3.65204602e-01 7.96349883e-01 -8.15475762e-01
-1.14819133e+00 3.64225745e-01 2.79149979e-01 -7.82090053e-02
-9.12657917e-01 -1.29762602e+00 -1.13162148e+00 -2.58223832e-01
-3.24582666e-01 5.33495843e-01 2.63888747e-01 8.17073703e-01
2.14982882e-01 6.50059700e-01 1.11546063e+00 -8.33576262e-01
-9.44132209e-01 -7.52586782e-01 -5.30802131e-01 7.28622498e-03
2.27949560e-01 -1.11046040e+00 -1.84744179e-01 -3.85841012e-01] | [3.7545526027679443, 2.052415370941162] |
5f75befa-62d9-46a6-99ff-42c2f2ef8b58 | pg-vton-a-novel-image-based-virtual-try-on | 2304.08956 | null | https://arxiv.org/abs/2304.08956v1 | https://arxiv.org/pdf/2304.08956v1.pdf | PG-VTON: A Novel Image-Based Virtual Try-On Method via Progressive Inference Paradigm | Virtual try-on is a promising computer vision topic with a high commercial value wherein a new garment is visually worn on a person with a photo-realistic effect. Previous studies conduct their shape and content inference at one stage, employing a single-scale warping mechanism and a relatively unsophisticated content inference mechanism. These approaches have led to suboptimal results in terms of garment warping and skin reservation under challenging try-on scenarios. To address these limitations, we propose a novel virtual try-on method via progressive inference paradigm (PGVTON) that leverages a top-down inference pipeline and a general garment try-on strategy. Specifically, we propose a robust try-on parsing inference method by disentangling semantic categories and introducing consistency. Exploiting the try-on parsing as the shape guidance, we implement the garment try-on via warping-mapping-composition. To facilitate adaptation to a wide range of try-on scenarios, we adopt a covering more and selecting one warping strategy and explicitly distinguish tasks based on alignment. Additionally, we regulate StyleGAN2 to implement re-naked skin inpainting, conditioned on the target skin shape and spatial-agnostic skin features. Experiments demonstrate that our method has state-of-the-art performance under two challenging scenarios. The code will be available at https://github.com/NerdFNY/PGVTON. | ['Kerui Hu', 'Zili Wang', 'Shuyou Zhang', 'Lemiao Qiu', 'Naiyu Fang'] | 2023-04-18 | null | null | null | null | ['virtual-try-on'] | ['computer-vision'] | [ 6.15132391e-01 1.61285773e-01 -2.27706641e-01 -3.19915086e-01
-8.49424183e-01 -6.71337724e-01 7.14005411e-01 -6.89928472e-01
-7.75138438e-02 3.14123958e-01 1.42656893e-01 -2.95695275e-01
1.96355626e-01 -7.41624594e-01 -7.08676159e-01 -4.99568224e-01
4.45263475e-01 7.28806034e-02 9.02281478e-02 -1.09253630e-01
-2.24025175e-02 3.40756595e-01 -1.18275607e+00 4.09262598e-01
6.83141887e-01 8.94874096e-01 1.67687982e-01 8.12815249e-01
-1.59339455e-03 2.14716166e-01 -3.64034235e-01 -1.08469808e+00
6.56222582e-01 -5.16705930e-01 -5.68043947e-01 2.35754132e-01
9.31392074e-01 -4.87275392e-01 -2.68812031e-01 1.06196558e+00
6.25887632e-01 1.77627094e-02 3.46468747e-01 -1.09552360e+00
-7.36407876e-01 2.14099154e-01 -8.27649355e-01 -3.13749015e-01
5.84867656e-01 4.71066594e-01 8.94716799e-01 -8.25292766e-01
8.05341244e-01 1.36770082e+00 8.44670534e-01 9.46530879e-01
-1.18789983e+00 -7.02591300e-01 2.80459762e-01 9.09373090e-02
-1.20489669e+00 -3.72342825e-01 1.20294487e+00 -2.25055486e-01
4.68773633e-01 6.57753587e-01 8.10573399e-01 1.56153858e+00
1.47386387e-01 8.25073659e-01 1.34769285e+00 -3.80945683e-01
2.03265756e-01 -2.07682982e-01 -4.43598568e-01 9.96028066e-01
-1.08425036e-01 6.02788515e-02 -4.18109804e-01 -3.02335899e-02
8.91462862e-01 -4.50470997e-03 -2.84140170e-01 -1.69970125e-01
-1.07872689e+00 5.93865812e-01 3.51926178e-01 -3.33940327e-01
-2.75514543e-01 1.80653512e-01 3.06370914e-01 9.58239958e-02
5.63961804e-01 1.00506917e-01 -4.72791910e-01 6.76778555e-02
-9.76140082e-01 2.87593156e-01 6.43404603e-01 7.64341235e-01
4.22829032e-01 -1.47589236e-01 -2.73926139e-01 9.65314746e-01
1.86708465e-01 4.75549310e-01 -8.21077153e-02 -1.02351499e+00
4.50219661e-01 3.58035415e-01 -3.26005042e-01 -7.57146657e-01
-1.28574058e-01 -8.92357975e-02 -6.16141498e-01 4.21752930e-01
5.03350735e-01 -3.82494344e-03 -1.51664805e+00 1.68154728e+00
7.04820275e-01 3.37583452e-01 -2.20904961e-01 1.09346712e+00
9.14794683e-01 3.19002986e-01 2.04449043e-01 2.53261030e-01
1.84909916e+00 -1.18641186e+00 -6.39120400e-01 -4.48251665e-01
1.57674745e-01 -8.50466907e-01 1.43042564e+00 5.43421328e-01
-1.09211552e+00 -4.40206766e-01 -8.57040346e-01 -4.05536354e-01
-2.15700433e-01 1.20518677e-01 5.82672119e-01 8.79901886e-01
-8.38935673e-01 5.53900838e-01 -9.45019126e-01 -5.28025031e-01
5.42893171e-01 -1.34834155e-01 -5.07023335e-01 -5.04062295e-01
-9.37725961e-01 7.05590665e-01 2.10839048e-01 2.35929191e-01
-6.91991270e-01 -7.50640035e-01 -1.05845702e+00 -4.88317460e-01
7.51074612e-01 -1.19534445e+00 9.56784904e-01 -9.35997188e-01
-1.86232948e+00 1.10676730e+00 1.58215854e-02 -4.53019403e-02
9.62589025e-01 -3.63756865e-01 -2.69528061e-01 2.98239648e-01
-1.47856846e-01 6.47549152e-01 1.14720416e+00 -1.27163720e+00
1.58234704e-02 -2.89501905e-01 2.46839881e-01 3.71224284e-01
-4.95221512e-03 1.39259994e-01 -9.95775938e-01 -1.10097444e+00
-7.65907317e-02 -1.06332982e+00 2.58637201e-02 6.19189024e-01
-8.42277050e-01 2.18829587e-01 7.57629633e-01 -1.30405521e+00
9.52355862e-01 -2.06863260e+00 3.98460746e-01 3.83087620e-02
1.56811044e-01 3.41896802e-01 -3.61445844e-01 2.57232964e-01
5.47051020e-02 1.51660919e-01 -5.96642196e-01 -7.69734979e-01
-5.95393591e-02 2.93114930e-01 -2.20506251e-01 3.91912729e-01
4.52327073e-01 1.05996644e+00 -8.70344400e-01 -5.82250237e-01
4.88604307e-01 7.25502074e-01 -5.87281942e-01 3.72104257e-01
-2.56736398e-01 4.24963057e-01 -2.56692737e-01 1.32476747e+00
8.97776008e-01 6.55534118e-02 3.09153795e-01 -5.29224873e-01
1.63153544e-01 -1.67187363e-01 -9.04615521e-01 2.21962929e+00
-4.36731696e-01 2.32757017e-01 2.40494505e-01 -5.58935165e-01
6.95507348e-01 4.03127596e-02 1.08209111e-01 -5.42121172e-01
1.41173080e-01 -1.48692355e-01 -2.89688677e-01 -5.46655953e-01
2.86335528e-01 -2.16668800e-01 -8.80321022e-03 1.63902435e-02
5.49778305e-02 -2.61718392e-01 -2.35993117e-01 8.59534070e-02
9.52670574e-01 9.23893392e-01 2.69963771e-01 1.03210554e-01
1.22246154e-01 -2.99796730e-01 5.76300263e-01 4.24089044e-01
-2.25893423e-01 1.23470032e+00 2.75222749e-01 -2.09946394e-01
-1.02382076e+00 -1.22614741e+00 -7.81832859e-02 1.03562868e+00
1.02607295e-01 -3.87934029e-01 -1.13787842e+00 -8.51467788e-01
-2.22063214e-01 7.59515703e-01 -9.53567207e-01 3.27363126e-02
-4.70539480e-01 -6.88721836e-01 7.15931475e-01 5.81395447e-01
8.26565981e-01 -8.81782711e-01 -5.54761112e-01 -2.83669084e-01
-1.95409521e-01 -1.26835322e+00 -6.20311439e-01 -4.12515908e-01
-4.77087408e-01 -1.08690250e+00 -1.00368929e+00 -4.08645749e-01
7.43100107e-01 2.98616607e-02 8.82976234e-01 7.13791847e-02
-5.55365682e-01 5.52467108e-01 -3.59445542e-01 -2.74663985e-01
-2.30816126e-01 -1.78647682e-01 -2.76212662e-01 2.38022193e-01
-8.25495273e-02 -4.88771468e-01 -9.28215683e-01 1.59382015e-01
-8.98099184e-01 5.34896314e-01 5.38807392e-01 7.13191092e-01
5.54142952e-01 -3.41979355e-01 -3.42405476e-02 -7.60944605e-01
3.45095754e-01 -3.37852985e-01 -1.65198758e-01 4.60391015e-01
-2.94905275e-01 -3.19545805e-01 4.91696417e-01 -5.66857159e-01
-1.20232117e+00 -9.20517650e-03 -3.12682480e-01 -7.73033082e-01
-1.42348319e-01 9.24345106e-02 -5.30284405e-01 -7.33490288e-02
3.53512377e-01 2.21957549e-01 1.77500099e-01 -4.90839422e-01
7.98256934e-01 3.63801628e-01 6.60611928e-01 -6.34657264e-01
1.00908566e+00 6.94403052e-01 -7.43863881e-02 -6.06967330e-01
-7.08766818e-01 2.89023109e-02 -5.32063544e-01 -3.51883024e-01
1.09424865e+00 -8.15723419e-01 -4.44123417e-01 8.49995255e-01
-1.12664688e+00 -6.76117718e-01 -1.03849871e-03 1.18228152e-01
-3.12769085e-01 6.73193932e-01 -6.48742497e-01 -7.35907078e-01
-5.44698596e-01 -1.06481886e+00 1.25639427e+00 2.11834013e-01
-2.19707847e-01 -9.76019681e-01 -5.56367226e-02 7.63590932e-01
2.03571215e-01 6.61199093e-01 5.38142383e-01 -3.03790718e-01
-3.92654568e-01 -3.90024879e-03 -4.20620680e-01 3.75756115e-01
7.14403540e-02 2.23406330e-02 -1.19335198e+00 -2.50665069e-01
-2.24572226e-01 -2.15182528e-01 8.91754568e-01 2.68037945e-01
1.37453735e+00 -2.70965874e-01 -1.06958352e-01 1.10309052e+00
1.41767347e+00 -1.80251524e-01 8.69090140e-01 1.95026889e-01
1.08863485e+00 8.41467917e-01 5.60856342e-01 2.30316535e-01
5.20564318e-01 7.30427623e-01 5.67110240e-01 -3.48041296e-01
-6.91590250e-01 -5.12178063e-01 3.34466457e-01 5.81776917e-01
-5.36117136e-01 -2.46664822e-01 -6.85285568e-01 4.53766197e-01
-1.54880095e+00 -7.67317235e-01 2.96816856e-01 2.12358999e+00
9.59554136e-01 -8.58447403e-02 1.37266189e-01 -2.01658458e-01
6.49460137e-01 3.79363835e-01 -5.81538677e-01 -4.41466182e-01
-1.15521764e-02 4.30419594e-01 4.22410071e-01 4.92879301e-01
-9.29165244e-01 9.07243729e-01 5.48435974e+00 9.37187672e-01
-1.19919157e+00 1.46791607e-01 6.67723000e-01 -4.96614911e-02
-5.67336917e-01 -3.83246201e-03 -4.68254656e-01 5.40215254e-01
2.03002661e-01 4.54004526e-01 6.83101892e-01 6.68949306e-01
-4.74075191e-02 6.52655959e-03 -9.05310690e-01 1.05140948e+00
3.08975846e-01 -1.10041189e+00 3.82471073e-04 -7.15228217e-03
3.64991188e-01 -3.32534164e-01 2.11239174e-01 7.41057172e-02
3.91123332e-02 -9.95904148e-01 8.54805529e-01 5.56143939e-01
1.37497616e+00 -4.40429002e-01 3.23146164e-01 -6.93298280e-02
-1.18509281e+00 2.22007334e-01 2.25720480e-01 6.02865927e-02
3.56874168e-01 4.42733616e-01 -5.70746779e-01 8.21790099e-01
6.70628071e-01 4.54948694e-01 -5.24507642e-01 5.61448038e-01
-5.56629300e-01 5.81113279e-01 -2.06797242e-01 3.20640832e-01
-1.63339332e-01 -2.19654456e-01 6.28961205e-01 1.36224413e+00
3.27480495e-01 2.33498037e-01 1.79464385e-01 9.20428872e-01
4.42132633e-03 -9.66177732e-02 -4.58021432e-01 1.39507353e-01
4.37700212e-01 1.38628507e+00 -7.73022473e-01 -2.19932631e-01
-4.59894896e-01 1.63297307e+00 1.00914076e-01 4.30249035e-01
-1.07516062e+00 -2.71411300e-01 6.78748608e-01 2.60995805e-01
2.86869496e-01 -9.27872658e-02 -4.04379308e-01 -1.33455551e+00
2.56430000e-01 -8.65561306e-01 2.66764939e-01 -9.67194676e-01
-1.46890008e+00 4.82789725e-01 2.36389577e-01 -1.03864586e+00
1.31473197e-02 -6.81452215e-01 -7.52221763e-01 7.46112704e-01
-1.54967093e+00 -1.91751957e+00 -5.36730945e-01 6.45825684e-01
7.47853994e-01 1.93869382e-01 7.78155744e-01 1.91367373e-01
-7.34974265e-01 8.43019128e-01 -6.47570968e-01 1.76055461e-01
7.80939579e-01 -1.14469433e+00 7.26037502e-01 1.09741378e+00
-1.07630357e-01 6.88394427e-01 6.59363508e-01 -8.71930838e-01
-1.64646685e+00 -1.13262630e+00 2.74666458e-01 -5.56988358e-01
4.63492930e-01 -5.65925837e-01 -7.82442272e-01 6.72725916e-01
3.81939620e-01 2.37216920e-01 5.22442818e-01 -8.48335624e-02
-6.35589600e-01 9.36844796e-02 -1.39432025e+00 9.94714975e-01
1.39974308e+00 -5.93688905e-01 -4.64218885e-01 1.10241167e-01
6.86294794e-01 -7.27132380e-01 -7.47090816e-01 4.67540443e-01
8.68784428e-01 -8.51976037e-01 1.19156325e+00 -2.60317057e-01
7.43500412e-01 -3.05810243e-01 -2.16392443e-01 -1.12935305e+00
-1.47098750e-01 -8.67048264e-01 -2.16552898e-01 1.28604078e+00
-5.66700846e-02 -6.00031137e-01 6.76931798e-01 8.28916907e-01
-1.33455626e-03 -1.01227570e+00 -7.44915128e-01 -5.05503237e-01
-1.93615541e-01 -6.60797358e-01 3.40598285e-01 1.13562787e+00
-4.19830143e-01 2.97736395e-02 -7.39881933e-01 1.06012367e-01
8.77856016e-01 -1.07108764e-01 9.16231155e-01 -6.90639019e-01
-6.19874656e-01 -1.92587197e-01 -1.91364974e-01 -8.16997945e-01
-2.28765234e-02 -8.28958869e-01 -3.66091914e-02 -1.39314115e+00
3.02361310e-01 -2.17072845e-01 -1.26409531e-01 8.59238803e-01
-3.55612576e-01 7.43763447e-01 4.82906997e-01 -7.10118860e-02
-1.15367129e-01 2.26497963e-01 1.51440299e+00 -2.04328541e-02
-2.12794337e-02 -3.53566647e-01 -7.80813158e-01 8.71850550e-01
8.11049700e-01 -1.62058070e-01 -4.47298676e-01 -4.72961158e-01
1.02026597e-01 -5.24031371e-02 9.71452594e-01 -6.28481448e-01
-1.04315273e-01 -2.87677079e-01 3.13293785e-01 -2.22342715e-01
6.85421467e-01 -4.94356155e-01 3.19707245e-01 2.98011571e-01
-9.92769077e-02 -1.69492230e-01 3.41972232e-01 5.63704789e-01
2.91353673e-01 1.11045137e-01 6.58027530e-01 -1.96244910e-01
-9.22895551e-01 3.29806715e-01 2.45400652e-01 -1.10735439e-01
9.74454701e-01 -5.62525332e-01 -2.34205887e-01 -1.00416869e-01
-6.65978730e-01 -1.25846699e-01 7.85771847e-01 4.81023639e-01
8.89311671e-01 -1.23769999e+00 -8.25318635e-01 2.72111028e-01
1.15333073e-01 3.98513786e-02 5.35021007e-01 8.00311863e-01
-5.51256716e-01 -3.54281455e-01 -2.88975567e-01 -4.62420613e-01
-1.54479563e+00 3.93004954e-01 1.29894182e-01 -1.14877321e-01
-8.68549764e-01 1.01464069e+00 4.48533565e-01 -5.67724109e-01
-9.86712649e-02 -1.49895504e-01 2.49051273e-01 -2.32432246e-01
4.70081091e-01 2.66360134e-01 -1.09855324e-01 -3.95547479e-01
-3.69842857e-01 9.46835458e-01 -2.22729687e-02 -2.56928086e-01
1.01052725e+00 -1.08063728e-01 1.12880222e-01 6.85022697e-02
9.29895878e-01 2.54021466e-01 -1.69546711e+00 -4.25081409e-04
-7.22327232e-01 -8.97464216e-01 -4.44670878e-02 -1.19591379e+00
-1.20345283e+00 7.02224433e-01 5.71273685e-01 -3.35770607e-01
1.38054204e+00 3.23977508e-02 1.06519639e+00 -2.76660293e-01
2.83047318e-01 -9.61972237e-01 9.21146050e-02 -7.08935857e-02
1.09774292e+00 -1.09562004e+00 2.33771548e-01 -8.94035518e-01
-8.35817635e-01 9.34943914e-01 5.13466954e-01 -1.34633839e-01
3.10577035e-01 3.63257587e-01 1.25028223e-01 -3.45429108e-02
-4.26541299e-01 -1.18928343e-01 5.76670408e-01 8.69123042e-01
6.83526993e-02 2.39070132e-01 -2.48855323e-01 5.01348794e-01
-7.28726462e-02 6.23905240e-03 1.40078440e-01 8.45215499e-01
1.86803907e-01 -1.19017899e+00 -4.69704211e-01 7.01619387e-02
-4.85159814e-01 -2.45057017e-01 -4.48840111e-01 8.28652561e-01
3.29654515e-01 5.41676044e-01 -2.44538307e-01 -5.98603964e-01
2.13056952e-01 1.31433588e-02 8.85806680e-01 -4.29339916e-01
-4.42120254e-01 1.20880365e-01 2.69053191e-01 -8.24020803e-01
-3.32868487e-01 -4.67693716e-01 -9.48859990e-01 -2.99151570e-01
-4.90561575e-02 -8.14650118e-01 7.83575237e-01 8.46378624e-01
4.43045557e-01 4.92493570e-01 2.85684347e-01 -9.63395238e-01
-3.44533503e-01 -7.34568179e-01 -3.39866787e-01 5.99238336e-01
1.12469196e-01 -5.83175004e-01 -1.57124192e-01 2.28520751e-01] | [11.923931121826172, -0.866944432258606] |
bdd30313-6896-4c20-ac48-ff6c12fe50c3 | thu_ngn-at-semeval-2019-task-12-toponym | null | null | https://aclanthology.org/S19-2229 | https://aclanthology.org/S19-2229.pdf | THU\_NGN at SemEval-2019 Task 12: Toponym Detection and Disambiguation on Scientific Papers | First name: Tao Last name: Qi Email: taoqi.qt@gmail.com Affiliation: Department of Electronic Engineering, Tsinghua University First name: Suyu Last name: Ge Email: gesy17@mails.tsinghua.edu.cn Affiliation: Department of Electronic Engineering, Tsinghua University First name: Chuhan Last name: Wu Email: wuch15@mails.tsinghua.edu.cn Affiliation: Department of Electronic Engineering, Tsinghua University First name: Yubo Last name: Chen Email: chen-yb18@mails.tsinghua.edu.cn Affiliation: Department of Electronic Engineering, Tsinghua University First name: Yongfeng Last name: Huang Email: yfhuang@mail.tsinghua.edu.cn Affiliation: Department of Electronic Engineering, Tsinghua University Toponym resolution is an important and challenging task in the neural language processing field, and has wide applications such as emergency response and social media geographical event analysis. Toponym resolution can be roughly divided into two independent steps, i.e., toponym detection and toponym disambiguation. In order to facilitate the study on toponym resolution, the SemEval 2019 task 12 is proposed, which contains three subtasks, i.e., toponym detection, toponym disambiguation and toponym resolution. In this paper, we introduce our system that participated in the SemEval 2019 task 12. For toponym detection, in our approach we use TagLM as the basic model, and explore the use of various features in this task, such as word embeddings extracted from pre-trained language models, POS tags and lexical features extracted from dictionaries. For toponym disambiguation, we propose a heuristics rule-based method using toponym frequency and population. Our systems achieved 83.03{\%} strict macro F1, 74.50 strict micro F1, 85.92 overlap macro F1 and 78.47 overlap micro F1 in toponym detection subtask. | ['Suyu Ge', 'Yongfeng Huang', 'Yubo Chen', 'Tao Qi', 'Chuhan Wu'] | 2019-06-01 | null | null | null | semeval-2019-6 | ['toponym-resolution'] | ['natural-language-processing'] | [-3.71221334e-01 -6.78583026e-01 -8.07985812e-02 -1.30329370e-01
-6.40970886e-01 -3.41671497e-01 4.71802503e-01 4.01751995e-01
-7.21921027e-01 7.72212863e-01 -2.73949001e-02 -4.00402784e-01
-2.71427304e-01 -1.25136673e+00 -1.95591599e-01 -4.42543387e-01
9.81931090e-02 6.46575272e-01 -1.29405648e-01 -4.72094268e-01
4.13390130e-01 6.24872804e-01 -1.47314215e+00 2.86120884e-02
9.10724461e-01 8.35724950e-01 4.18961376e-01 5.94105199e-02
-1.06273770e-01 6.50782436e-02 -8.63985717e-01 2.16183156e-01
9.68769714e-02 -3.47723544e-01 -7.26205945e-01 -6.52651489e-01
-7.68708885e-02 2.09897012e-01 -5.14558293e-02 1.29739165e+00
6.54376030e-01 2.83061266e-01 6.94197357e-01 -1.33299184e+00
-4.16165799e-01 8.13329816e-01 -4.48721826e-01 3.17620099e-01
5.88287711e-01 -2.85797656e-01 1.10604978e+00 -1.21361542e+00
4.27326739e-01 1.06326020e+00 3.80246609e-01 2.32546464e-01
-6.43466771e-01 -1.35119009e+00 1.63377136e-01 2.26874083e-01
-2.00256968e+00 3.15290093e-02 4.03143436e-01 -4.77982908e-01
1.21496284e+00 1.35542959e-01 3.66134524e-01 5.40253401e-01
2.43142113e-01 4.40683424e-01 9.76984620e-01 -4.58896250e-01
1.95400044e-01 -8.71009529e-02 3.24339360e-01 6.29878283e-01
3.88958037e-01 -1.71436872e-02 -3.96311522e-01 -3.60412538e-01
7.01298475e-01 2.19670519e-01 -1.50372341e-01 4.45365399e-01
-1.28222895e+00 1.02448332e+00 5.14715970e-01 1.05384541e+00
-3.87781441e-01 1.01867273e-01 2.28346795e-01 4.61950481e-01
1.09648220e-02 9.49445009e-01 -5.85178733e-01 8.15258995e-02
-7.18539178e-01 3.53103369e-01 4.51024741e-01 1.31880581e+00
9.87014413e-01 4.22053821e-02 -1.21730208e-01 1.15954316e+00
4.27454822e-02 6.07822478e-01 8.18058670e-01 -5.05977392e-01
3.36677372e-01 4.98164266e-01 -1.13753136e-02 -9.14652944e-01
-6.32368505e-01 -3.48219156e-01 -7.63761401e-01 -4.80952859e-01
-4.50742990e-02 -4.25579488e-01 -6.26891017e-01 1.73464167e+00
-1.91073239e-01 1.11455388e-01 3.88923846e-02 6.51243389e-01
8.11432779e-01 9.63001549e-01 1.99180126e-01 -5.34787357e-01
1.70017850e+00 -5.99570096e-01 -6.97572827e-01 3.90751883e-02
8.79104435e-01 -1.03631198e+00 7.79317498e-01 5.56257188e-01
-7.49987721e-01 -5.34478843e-01 -7.92303503e-01 6.21078074e-01
-8.60127747e-01 4.69634712e-01 4.42279696e-01 5.71513593e-01
-5.27317166e-01 4.82052743e-01 -3.26135874e-01 -7.46557653e-01
-1.81645095e-01 5.55532038e-01 -2.59308666e-01 -1.91975802e-01
-1.99442899e+00 1.03414667e+00 9.52792645e-01 -1.19262747e-01
-2.22469479e-01 -5.89331150e-01 -7.98394561e-01 -3.07151005e-02
3.18607688e-01 -5.23683488e-01 9.37871158e-01 -3.17752331e-01
-9.61262047e-01 7.33346462e-01 -9.87602547e-02 -1.51702479e-01
-1.97079256e-01 -2.30945125e-01 -1.18819535e+00 -1.16356388e-01
5.78364193e-01 4.58587825e-01 7.59271830e-02 -5.19024909e-01
-9.80721891e-01 -3.33091527e-01 -2.88709491e-01 3.01493257e-01
-5.16793311e-01 6.16610587e-01 -3.69375765e-01 -7.48971939e-01
3.02273128e-02 -9.14575219e-01 -2.27678791e-01 -9.24898207e-01
-5.06425858e-01 -7.68149674e-01 5.41735113e-01 -5.43713272e-01
1.86991429e+00 -1.91621995e+00 -4.53410745e-01 7.40008831e-01
-8.01641867e-03 4.86996919e-01 1.47155851e-01 8.22640598e-01
-3.14520806e-01 2.75097549e-01 -2.38749355e-01 3.80376726e-01
-6.49120584e-02 -5.56781702e-02 -9.81630981e-02 5.16584292e-02
-7.87095279e-02 5.71161628e-01 -9.81024206e-01 -3.55848759e-01
2.73644358e-01 3.71303149e-02 -2.50241309e-01 -5.12251304e-03
3.06138039e-01 1.74689204e-01 -8.36302161e-01 5.39627433e-01
4.43797439e-01 2.56547093e-01 8.24101269e-03 -1.36422828e-01
-3.49986792e-01 1.29697859e-01 -1.53261340e+00 1.45521247e+00
-5.55971444e-01 2.89090574e-01 -4.75228131e-01 -1.07556570e+00
1.35205412e+00 4.30472761e-01 6.05261028e-01 -5.16877413e-01
3.72573882e-01 8.51228535e-01 -3.38423401e-02 -1.46366969e-01
6.70559168e-01 -2.79440969e-01 -7.13398516e-01 1.87513575e-01
-1.60473496e-01 -1.92332759e-01 4.02546167e-01 1.84073865e-01
7.65832543e-01 -3.03453028e-01 8.08455825e-01 -5.17541111e-01
5.82365394e-01 3.73337604e-02 8.42783391e-01 6.98170006e-01
-8.45155343e-02 3.25059056e-01 7.21670613e-02 -3.87240708e-01
-6.24036670e-01 -8.01364839e-01 -5.26591778e-01 8.26597095e-01
5.20397902e-01 -9.03724134e-01 -5.56915998e-01 -3.98622245e-01
-2.01416165e-01 9.04973149e-01 -8.20702240e-02 -3.67942303e-01
-7.55261779e-01 -6.61257207e-01 9.29560602e-01 3.75104517e-01
7.54046559e-01 -1.15843236e+00 -5.05990267e-01 4.59020883e-01
-2.79487342e-01 -1.12621903e+00 -6.79588437e-01 1.34842217e-01
-4.48208362e-01 -8.69786263e-01 -6.84365094e-01 -9.06095147e-01
2.05892593e-01 1.67103985e-03 1.07208848e+00 1.49020910e-01
-3.92475218e-01 1.46617115e-01 -4.28305238e-01 -4.69513297e-01
1.60696894e-01 3.18234295e-01 5.84541440e-01 -1.87888280e-01
9.59180772e-01 -4.09154207e-01 -3.91670287e-01 4.83865976e-01
-7.99493492e-01 -4.92479444e-01 5.64755142e-01 9.30773079e-01
4.91309524e-01 1.99675813e-01 8.99025142e-01 -8.09961855e-01
7.33462572e-01 -4.49680239e-01 -4.44193304e-01 2.83529729e-01
-6.45384133e-01 -2.50968754e-01 7.97471762e-01 -2.05981508e-01
-5.04734814e-01 -5.56996129e-02 -4.53141838e-01 -1.66158408e-01
-4.59941983e-01 6.79348946e-01 -5.89357913e-01 3.13037902e-01
5.74368298e-01 2.59959787e-01 -7.61227548e-01 -4.13767576e-01
6.81633130e-02 1.06423974e+00 3.14355105e-01 -5.30864120e-01
7.02229083e-01 -1.96299314e-01 -2.81955581e-02 -1.18140042e+00
-6.30615950e-01 -6.63582444e-01 -5.33107638e-01 2.02914253e-02
9.65641975e-01 -8.53074014e-01 -5.05209506e-01 3.51388603e-01
-1.09767663e+00 3.75259995e-01 3.43143672e-01 9.40426886e-01
-9.69035998e-02 3.55518967e-01 -4.04632211e-01 -2.21069708e-01
-3.20095539e-01 -1.11478400e+00 6.42143428e-01 5.04808962e-01
-7.62191713e-01 -8.62954676e-01 -1.57443210e-01 -3.13732177e-02
3.14162403e-01 1.86039254e-01 8.90657485e-01 -1.24359119e+00
-1.18753721e-03 -1.42111957e-01 -7.50886574e-02 2.00772554e-01
3.07504147e-01 -5.63690588e-02 -5.90591967e-01 -1.91526607e-01
-4.65923220e-01 1.91062853e-01 6.71277940e-01 2.35567898e-01
1.04503226e+00 3.40791978e-02 -6.75061107e-01 4.13549632e-01
1.33590591e+00 6.85994565e-01 4.38639045e-01 6.20132267e-01
5.03412902e-01 4.17124033e-01 1.00988173e+00 6.00735724e-01
1.52047873e-01 1.06027424e+00 -1.26945466e-01 2.54637718e-01
3.74838114e-01 -3.08956712e-01 2.97475576e-01 7.39940107e-01
1.59498259e-01 -1.10706180e-01 -1.30176747e+00 7.16106951e-01
-1.46228004e+00 -9.37018871e-01 -3.49005818e-01 2.05283475e+00
6.85495079e-01 -1.19361728e-01 2.55815387e-02 5.04854470e-02
1.13264680e+00 -1.74293980e-01 -1.11390151e-01 -6.23991609e-01
-6.11802749e-02 7.55197108e-02 3.20043832e-01 4.25094485e-01
-1.05387759e+00 1.14764214e+00 2.20071292e+00 1.37742722e+00
-9.57639456e-01 -1.90326944e-01 2.10700527e-01 1.38038978e-01
-1.69434547e-01 3.00580621e-01 -1.05545676e+00 7.95085192e-01
7.11736977e-01 -4.52064574e-01 2.58877397e-01 6.06414616e-01
-1.54810920e-01 -7.13648126e-02 -8.85940790e-01 1.51525378e+00
-7.92201459e-02 -1.07338977e+00 1.15467403e-02 -1.17087528e-01
6.82521343e-01 -3.88066098e-03 -2.78343707e-01 4.41281974e-01
1.83566332e-01 -1.05010319e+00 2.79733896e-01 2.81930059e-01
9.42253411e-01 -1.24914730e+00 8.58878970e-01 5.04048824e-01
-1.68893456e+00 -1.37361497e-01 -2.76518971e-01 9.40631330e-02
2.75532424e-01 1.06050849e+00 -4.37191755e-01 9.16611075e-01
7.10060000e-01 5.11468768e-01 -9.89389196e-02 1.10111797e+00
-6.75160706e-01 4.94081527e-01 -4.69030291e-01 -3.68550718e-01
2.63925314e-01 -2.26170361e-01 6.05120301e-01 1.34826350e+00
6.49763823e-01 3.47381175e-01 4.98194337e-01 5.94764948e-01
1.72928488e-03 3.95240754e-01 -8.18150938e-01 1.92509010e-01
1.22552788e+00 9.90504265e-01 -7.06474781e-01 -5.21080911e-01
6.66751713e-02 5.19022703e-01 -2.09209621e-01 4.02509719e-01
-6.51096702e-01 -1.34084487e+00 5.93551278e-01 -1.02898173e-01
-1.22185200e-01 -1.49176687e-01 -1.17285185e-01 -1.16978908e+00
-3.87810975e-01 -6.75879657e-01 6.63155556e-01 -4.17470008e-01
-1.17922902e+00 6.07854068e-01 3.90088230e-01 -1.20614612e+00
-2.60180980e-01 -5.75011253e-01 -8.24279666e-01 1.32298756e+00
-1.17735744e+00 -7.50406802e-01 -1.35002971e-01 5.89740634e-01
4.35738921e-01 -5.60744524e-01 9.38968658e-01 5.53609848e-01
-6.15198672e-01 7.75657594e-01 2.18002170e-01 2.72689402e-01
9.31309402e-01 -1.02178454e+00 2.18838871e-01 7.25255907e-01
4.72977869e-02 9.57742691e-01 5.56754887e-01 -7.36923695e-01
-9.14047182e-01 -1.12315845e+00 1.63729286e+00 -1.68344527e-01
6.07391119e-01 -2.35281423e-01 -7.50664115e-01 4.17049587e-01
6.10857792e-02 -3.21892887e-01 7.14409888e-01 -1.22923493e-01
8.11069608e-02 -1.57905981e-01 -1.01805210e+00 6.66971743e-01
8.13037813e-01 -5.33988953e-01 -6.87752843e-01 3.83314639e-01
5.35683393e-01 -2.30749920e-01 -1.09735239e+00 4.10389453e-01
2.74088800e-01 -2.79308170e-01 8.18551123e-01 -3.64681363e-01
1.24828972e-01 -5.99007130e-01 -2.90427625e-01 -1.29842412e+00
-3.37862819e-01 -5.30645967e-01 4.83954608e-01 1.36518610e+00
3.17821473e-01 -1.11691320e+00 2.18697891e-01 3.29725593e-01
-4.60766584e-01 -5.97100258e-01 -9.87597466e-01 -1.00030863e+00
2.28669107e-01 -3.99519086e-01 8.47062409e-01 1.21592641e+00
-7.31774122e-02 4.24776375e-01 -3.75408679e-01 2.87835032e-01
2.76865780e-01 2.35779107e-01 3.69445741e-01 -1.34403861e+00
-1.10506238e-02 -4.59268481e-01 -2.48433322e-01 -9.49884474e-01
6.67179450e-02 -8.27558100e-01 -2.21420437e-01 -1.56939769e+00
-3.96371096e-01 -4.78013009e-01 -5.00520110e-01 6.15834534e-01
5.22955582e-02 1.76509768e-01 7.72786215e-02 2.62951881e-01
-1.47170186e-01 3.08998644e-01 8.35741222e-01 5.60461804e-02
-1.48383304e-01 1.23211175e-01 -7.55627990e-01 6.35309100e-01
1.22371650e+00 -8.11107814e-01 4.50349264e-02 -2.28595838e-01
3.47458214e-01 -4.99354489e-02 1.60874322e-01 -8.44278097e-01
6.12618446e-01 -2.95407146e-01 5.88740446e-02 -6.03939831e-01
1.99709341e-01 -7.07212865e-01 -2.38907281e-02 7.55932987e-01
-2.89249551e-02 5.39788604e-01 1.79877207e-01 7.04075918e-02
-4.42595631e-01 -2.62446254e-01 6.14094019e-01 -7.29980990e-02
-9.20408249e-01 8.98715034e-02 -7.61744380e-01 1.20781556e-01
1.26066375e+00 -1.89142093e-01 -1.99423641e-01 -3.02781630e-02
-5.29310942e-01 6.97643459e-01 5.15783653e-02 6.14474714e-01
7.24555850e-01 -1.71945047e+00 -7.11888909e-01 7.70817846e-02
3.59157622e-01 -3.48502725e-01 2.67678976e-01 8.70973289e-01
-2.14258358e-01 7.07154214e-01 -1.36608183e-01 -1.30959615e-01
-1.26798975e+00 1.31766975e-01 1.46662682e-01 -1.64399222e-01
-2.11526632e-01 7.54461586e-01 6.74110875e-02 -6.31808877e-01
-6.60426021e-02 -1.38032570e-01 -6.90037131e-01 4.61588323e-01
3.61189634e-01 4.26814139e-01 3.96097243e-01 -6.85501635e-01
-8.21427643e-01 9.31323051e-01 -1.49539756e-02 -1.48409884e-02
9.99614954e-01 2.16151565e-01 -2.76955754e-01 3.84154916e-01
1.41195393e+00 -1.49160437e-02 3.15085471e-01 -2.03898400e-01
2.75737941e-01 -3.52576792e-01 -1.53998718e-01 -8.09555173e-01
-4.84098077e-01 8.96333635e-01 6.13581479e-01 -9.93935838e-02
1.11584485e+00 -1.22286193e-01 1.01862419e+00 7.88943887e-01
6.83701634e-01 -1.43105364e+00 -4.34448749e-01 1.14824283e+00
6.18132174e-01 -8.85170460e-01 -1.99174106e-01 -9.41273347e-02
-5.49549878e-01 1.10648108e+00 5.32670140e-01 -1.57951653e-01
8.12075555e-01 -2.25377694e-01 -8.44047964e-02 -2.40779370e-01
-4.21058416e-01 -1.13454722e-01 3.90637219e-01 2.42180377e-01
3.94284159e-01 1.64743841e-01 -9.02176797e-01 9.93729413e-01
-5.49992859e-01 -3.71575236e-01 4.25989509e-01 8.46481681e-01
-6.41355515e-01 -1.31328487e+00 -6.24037981e-01 5.34504473e-01
-3.13621074e-01 -4.98084575e-01 -4.04808134e-01 8.06417286e-01
5.23896754e-01 8.98173451e-01 -2.20998257e-01 -6.08934164e-01
5.63010931e-01 1.09156564e-01 -7.53398240e-02 -8.86700988e-01
-9.49719191e-01 2.86996141e-02 2.70176649e-01 -3.02466065e-01
-9.29574370e-02 -6.35474920e-01 -1.64720130e+00 -2.34269798e-02
-5.34639776e-01 1.13140440e+00 4.99168843e-01 1.04680932e+00
2.58026958e-01 5.07754445e-01 7.13456333e-01 -2.85077631e-01
-2.77523637e-01 -8.74797642e-01 -7.06860960e-01 1.22820191e-01
-2.52960891e-01 -7.93254316e-01 -6.02373406e-02 -4.59281206e-01] | [9.979114532470703, 9.170188903808594] |
da3b7973-16f1-479f-8573-516fde083f7d | hybrid-clustering-based-on-content-and | 1703.09646 | null | http://arxiv.org/abs/1703.09646v1 | http://arxiv.org/pdf/1703.09646v1.pdf | Hybrid Clustering based on Content and Connection Structure using Joint Nonnegative Matrix Factorization | We present a hybrid method for latent information discovery on the data sets
containing both text content and connection structure based on constrained low
rank approximation. The new method jointly optimizes the Nonnegative Matrix
Factorization (NMF) objective function for text clustering and the Symmetric
NMF (SymNMF) objective function for graph clustering. We propose an effective
algorithm for the joint NMF objective function, based on a block coordinate
descent (BCD) framework. The proposed hybrid method discovers content
associations via latent connections found using SymNMF. The method can also be
applied with a natural conversion of the problem when a hypergraph formulation
is used or the content is associated with hypergraph edges.
Experimental results show that by simultaneously utilizing both content and
connection structure, our hybrid method produces higher quality clustering
results compared to the other NMF clustering methods that uses content alone
(standard NMF) or connection structure alone (SymNMF). We also present some
interesting applications to several types of real world data such as citation
recommendations of papers. The hybrid method proposed in this paper can also be
applied to general data expressed with both feature space vectors and pairwise
similarities and can be extended to the case with multiple feature spaces or
multiple similarity measures. | ['Rundong Du', 'Haesun Park', 'Barry Drake'] | 2017-03-28 | null | null | null | null | ['text-clustering'] | ['natural-language-processing'] | [ 9.45995525e-02 -1.24436304e-01 -3.15717131e-01 -1.30598679e-01
-1.89707294e-01 -5.98549426e-01 6.94574535e-01 2.38556489e-01
-2.28958294e-01 3.99351150e-01 3.96194607e-01 -3.61197025e-01
-9.08439279e-01 -6.09784424e-01 -2.93613464e-01 -9.10107911e-01
-1.31547838e-01 7.95777857e-01 -1.04143701e-01 -8.64134803e-02
5.16228020e-01 4.88242745e-01 -1.45660055e+00 1.84009254e-01
9.62857246e-01 2.62686014e-01 5.38392365e-01 6.12497568e-01
-3.94843012e-01 3.60535115e-01 -2.63046116e-01 -1.76914275e-01
5.63290775e-01 -1.95256576e-01 -9.75674391e-01 3.77490819e-01
4.06424642e-01 4.64331448e-01 -1.48741260e-01 1.14785099e+00
2.60230958e-01 3.96625161e-01 8.98116648e-01 -1.52770007e+00
-4.96782720e-01 6.33165538e-01 -8.87185276e-01 -5.08349836e-02
4.16780353e-01 -8.39866102e-01 1.29867792e+00 -9.39907074e-01
7.98931718e-01 1.41193199e+00 2.81455785e-01 1.34078786e-01
-1.63517427e+00 -2.30412617e-01 4.18049067e-01 3.82413387e-01
-1.33758152e+00 5.56450114e-02 7.99666643e-01 -7.21632481e-01
8.02335620e-01 4.12772059e-01 3.78909677e-01 5.23117006e-01
2.91742850e-02 6.87784851e-01 8.25743496e-01 -6.19592905e-01
9.35439765e-02 3.42096575e-02 6.66256368e-01 6.18078053e-01
5.25487185e-01 -5.34825683e-01 -2.11195260e-01 -6.19180024e-01
4.81526881e-01 1.62301660e-01 -4.68601167e-01 -7.88249433e-01
-1.45364690e+00 1.03796136e+00 1.01479061e-01 6.58070862e-01
-2.43483394e-01 -8.60266387e-02 1.90385386e-01 5.03563404e-01
4.79204446e-01 4.86042172e-01 -1.86254397e-01 8.90171006e-02
-1.16496003e+00 1.71038862e-02 1.00616109e+00 9.73093212e-01
7.93085873e-01 -1.55177265e-01 8.22847933e-02 1.00992119e+00
5.04421711e-01 1.69422179e-01 5.16190648e-01 -1.04652393e+00
4.70913082e-01 8.30428958e-01 1.83017701e-02 -1.40758801e+00
-5.70164263e-01 -3.48382533e-01 -9.64734435e-01 -1.19694613e-01
2.39298254e-01 4.33049686e-02 -4.16700840e-01 1.72174013e+00
1.89491183e-01 1.94329754e-01 6.42604604e-02 7.09381104e-01
6.22607350e-01 7.35723794e-01 -6.40657127e-01 -7.00164557e-01
1.09391439e+00 -1.04340470e+00 -9.21036005e-01 5.41278839e-01
6.43175066e-01 -1.02674854e+00 6.88341498e-01 5.83245397e-01
-9.78208959e-01 -3.23180228e-01 -9.26507652e-01 1.26757473e-01
-4.47499901e-01 2.45595649e-01 6.77588701e-01 5.07285118e-01
-1.62507808e+00 5.31273007e-01 -5.54037571e-01 -6.60720706e-01
-3.23325753e-01 7.77615190e-01 -4.83494371e-01 -1.79388523e-01
-9.79178309e-01 4.10522074e-01 4.11743790e-01 1.28212079e-01
-1.65749282e-01 -2.64604062e-01 -5.44144630e-01 9.65833068e-02
2.62285143e-01 -7.31530726e-01 3.59430969e-01 -8.57445240e-01
-1.16642606e+00 4.11599845e-01 -3.40011656e-01 -8.95555541e-02
2.33962849e-01 4.36834581e-02 -4.45704639e-01 1.38953373e-01
2.64262885e-01 2.56018102e-01 7.54238427e-01 -1.31398332e+00
-2.73339301e-01 -1.89776167e-01 -1.34083912e-01 3.21284562e-01
-8.36928487e-01 -1.27775893e-01 -6.15094960e-01 -7.04135895e-01
5.33856094e-01 -1.09289026e+00 -2.96066195e-01 -5.07951915e-01
-6.06967092e-01 -5.07120609e-01 1.10444164e+00 -2.76662827e-01
1.35257649e+00 -1.92443752e+00 1.01514292e+00 9.15471375e-01
4.50337380e-01 -3.06178257e-02 -3.42570007e-01 9.76097703e-01
-4.65496749e-01 1.82732165e-01 -2.06705138e-01 -4.14773315e-01
-9.20031816e-02 3.22012514e-01 -4.22367454e-03 6.12684250e-01
-4.02489394e-01 5.03541946e-01 -1.04393268e+00 -5.35457134e-01
4.22827713e-02 3.02353740e-01 -5.16219437e-01 -1.63481131e-01
1.68040112e-01 2.93256640e-02 -9.66802612e-02 4.41286527e-02
6.48733497e-01 -4.93929505e-01 5.89980721e-01 -3.75213951e-01
-2.52053112e-01 -2.05925956e-01 -1.77998972e+00 1.77263999e+00
-3.06225270e-01 6.94836318e-01 3.34163725e-01 -1.28334665e+00
9.11504269e-01 3.26879114e-01 1.06701112e+00 1.97045594e-01
-1.10417433e-01 -1.42112775e-02 1.16260238e-01 -3.06655228e-01
5.97875357e-01 1.89566806e-01 3.57051969e-01 7.84317851e-01
1.88852295e-01 5.07760048e-01 7.42721260e-01 1.08168912e+00
1.15291548e+00 -3.35626394e-01 5.23012020e-02 -9.44237113e-01
8.81287694e-01 -2.65556812e-01 4.08357561e-01 5.97481787e-01
5.59799254e-01 3.37553829e-01 5.70714116e-01 2.01719832e-02
-8.82980049e-01 -8.73731017e-01 -2.15054601e-01 7.13932931e-01
1.28838643e-01 -1.15874803e+00 -3.72566402e-01 -6.96163893e-01
-1.08039267e-02 2.32019901e-01 -5.33394754e-01 9.07083675e-02
-2.36426935e-01 -9.97707784e-01 -2.01099008e-01 1.72341019e-01
7.83617422e-02 -4.42448378e-01 4.18660462e-01 1.15731224e-01
-2.90738851e-01 -8.24164689e-01 -5.11269450e-01 1.12850897e-01
-1.03681910e+00 -9.52786922e-01 -5.44713378e-01 -9.09721196e-01
9.90961373e-01 7.39447892e-01 8.17641139e-01 3.16048086e-01
-1.00354150e-01 8.57685447e-01 -6.65607274e-01 4.08722311e-01
-1.85323238e-01 4.76994552e-02 3.70930195e-01 4.81792182e-01
2.72201270e-01 -6.59880459e-01 -3.22485954e-01 5.14521003e-01
-1.07645750e+00 8.67939666e-02 3.24137360e-01 9.65531409e-01
3.13788265e-01 6.11673236e-01 4.62371409e-01 -1.22922301e+00
9.23251927e-01 -7.36470282e-01 -4.99525160e-01 2.79205352e-01
-1.14758015e+00 2.80670792e-01 5.63851297e-01 -3.11615050e-01
-8.14411759e-01 2.15229839e-01 3.81633341e-01 -5.59126735e-01
2.36219347e-01 7.90797114e-01 -3.65303129e-01 -2.72694323e-02
3.86949360e-01 -4.20071110e-02 8.32763314e-03 -7.89987028e-01
5.40894032e-01 7.05760419e-01 2.26831913e-01 -5.71845174e-01
8.25608134e-01 6.46609068e-01 4.08299893e-01 -9.83981788e-01
-2.46893287e-01 -1.23783386e+00 -1.09646225e+00 -1.73120156e-01
9.27515864e-01 -6.78649902e-01 -6.80318534e-01 -1.64465383e-01
-1.06096935e+00 3.31832886e-01 1.73828807e-02 8.60085666e-01
-4.18982357e-01 9.85759497e-01 -5.79365551e-01 -6.92237437e-01
3.59323621e-02 -1.03834915e+00 8.35155904e-01 -4.29803640e-01
-1.65350586e-01 -1.46783304e+00 3.30906123e-01 3.44826579e-01
-1.06769972e-01 -2.34805457e-02 1.24422884e+00 -6.27600610e-01
-3.05673778e-01 -1.25788823e-01 -1.92467481e-01 6.86428547e-02
3.64446074e-01 1.44321263e-01 -3.76184702e-01 -6.99089289e-01
-3.87328044e-02 3.09162229e-01 1.08723962e+00 4.33962673e-01
6.55416846e-01 -3.44590276e-01 -4.98875678e-01 3.92475963e-01
1.56990063e+00 -2.18029469e-02 3.51602048e-01 2.15056866e-01
1.17337680e+00 9.33432400e-01 3.67871255e-01 3.94325584e-01
1.85538203e-01 9.46479857e-01 3.23140085e-01 -1.84642807e-01
2.24415317e-01 2.07776695e-01 3.58066380e-01 1.35485089e+00
-1.89164907e-01 -4.18853313e-01 -8.56103957e-01 4.85173792e-01
-2.58070183e+00 -1.03679872e+00 -8.85181069e-01 2.20417953e+00
4.01456475e-01 -3.03996533e-01 3.01831394e-01 4.29022670e-01
1.20589173e+00 -1.93221912e-01 -1.14610389e-01 -5.06526172e-01
-2.05326900e-01 -2.56269157e-01 3.93061638e-01 7.94144034e-01
-9.73024070e-01 5.50547302e-01 5.91316319e+00 8.95287395e-01
-4.95644331e-01 1.48013011e-01 -1.47233874e-01 -7.56165087e-02
-6.17314339e-01 2.86397725e-01 -5.76690733e-01 3.20337147e-01
7.23594487e-01 -7.34071881e-02 5.41790485e-01 4.67376322e-01
3.37538362e-01 5.34170605e-02 -9.42817807e-01 1.20831609e+00
1.63705423e-01 -1.18621731e+00 2.98592448e-01 5.35245180e-01
1.11616313e+00 -3.13526392e-01 -2.77026221e-02 -3.84999990e-01
3.64289522e-01 -5.98347843e-01 3.04343738e-02 4.83049750e-01
4.08242315e-01 -7.78454065e-01 6.53033495e-01 3.22557539e-01
-1.35384405e+00 -4.54820804e-02 -5.47135055e-01 7.18354527e-03
9.26930010e-02 8.75473917e-01 -6.50876999e-01 9.72005486e-01
3.77929181e-01 1.26970112e+00 -5.75259209e-01 1.06588352e+00
1.02391824e-01 4.47988063e-01 -3.25046211e-01 -6.21839203e-02
3.30374509e-01 -8.46689582e-01 8.38898003e-01 1.18832195e+00
1.40975028e-01 -1.83585748e-01 3.57122600e-01 5.57351470e-01
1.49511546e-01 7.19175696e-01 -5.90840042e-01 -1.38738245e-01
1.68919444e-01 1.60657728e+00 -1.38864648e+00 -2.17860594e-01
-5.33898413e-01 1.09848392e+00 2.90165842e-01 5.13658941e-01
-1.44287169e-01 -1.56672075e-01 3.58633608e-01 1.13856971e-01
1.39032528e-01 -4.99269456e-01 1.02193981e-01 -1.54997528e+00
-1.91274315e-01 -5.17959476e-01 6.17896676e-01 -4.32949901e-01
-1.56854844e+00 6.43434107e-01 2.63122052e-01 -1.25917566e+00
-1.23719856e-01 -5.63433230e-01 -4.86976534e-01 6.67036355e-01
-1.02292180e+00 -9.54235554e-01 4.60823402e-02 1.07551396e+00
3.24190617e-01 -4.63272810e-01 8.20372701e-01 3.81613970e-01
-5.38216949e-01 2.70389736e-01 8.17667902e-01 -1.44237876e-01
6.69109464e-01 -1.65882659e+00 -2.36200213e-01 8.95071924e-01
6.49677098e-01 9.91937935e-01 5.53145170e-01 -8.17807555e-01
-1.38581216e+00 -9.75297332e-01 8.74702513e-01 -2.52856880e-01
6.87506258e-01 -5.94220579e-01 -6.28180027e-01 4.10196871e-01
3.91670436e-01 -4.87200052e-01 1.09426427e+00 5.64458549e-01
-1.94045201e-01 1.25801399e-01 -9.13006544e-01 4.68933731e-01
9.49149668e-01 -3.52315068e-01 -4.12058145e-01 7.94106662e-01
5.66905558e-01 4.77219820e-01 -1.06255698e+00 2.23908290e-01
3.05437386e-01 -5.95577240e-01 7.48598993e-01 -6.97447002e-01
5.77356219e-02 -6.70486987e-01 -3.57727855e-01 -1.27833080e+00
-9.68033254e-01 -7.93358564e-01 -4.40182239e-02 1.38359511e+00
4.81396765e-01 -3.92071068e-01 7.04518020e-01 1.95078969e-01
1.10785134e-01 -5.58877528e-01 -7.17234015e-01 -8.04999948e-01
-2.04187870e-01 -2.33427793e-01 1.08525373e-01 1.39449632e+00
3.56441766e-01 6.46628082e-01 -5.66883981e-01 2.25374941e-02
7.68573403e-01 5.17868221e-01 5.72091699e-01 -1.74057496e+00
-4.77873534e-01 -4.40387189e-01 -5.80992341e-01 -8.35134327e-01
4.69291061e-01 -1.39992344e+00 -4.93774742e-01 -1.81224310e+00
3.96319032e-01 -2.55732626e-01 -3.12727809e-01 2.01466158e-01
7.03563318e-02 1.08696073e-01 3.72066051e-01 6.28134251e-01
-5.91777146e-01 2.30383039e-01 9.72468793e-01 -3.73916849e-02
-4.52297896e-01 1.33282498e-01 -5.04075766e-01 5.27748644e-01
5.28104544e-01 -3.52947772e-01 -7.10540652e-01 -8.15550908e-02
6.40641034e-01 -2.07018480e-02 -1.58735052e-01 -6.70512795e-01
6.18503988e-01 1.06941136e-02 1.01616383e-01 -6.75722599e-01
2.88552403e-01 -1.07732785e+00 3.85656506e-01 1.63580030e-01
-3.92548412e-01 3.98379952e-01 -3.09241414e-01 9.79268909e-01
-2.32536688e-01 -5.16841531e-01 4.62318331e-01 -9.40434858e-02
-4.45885420e-01 1.05341785e-01 -5.83440185e-01 -4.07810926e-01
8.22112560e-01 -2.89607525e-01 -3.53770889e-02 -5.13280630e-01
-1.11103225e+00 4.07906294e-01 4.82780814e-01 4.19825584e-01
7.58963287e-01 -1.65176225e+00 -7.24647343e-01 6.43847957e-02
-1.15086302e-01 -5.54909706e-01 -3.80266160e-02 1.13252866e+00
-1.37851790e-01 4.78727996e-01 2.54723784e-02 -6.79997921e-01
-1.54398692e+00 8.99236083e-01 -2.87682116e-01 -3.92777532e-01
-4.37428415e-01 4.96493816e-01 1.60973728e-01 -3.50703925e-01
1.76648289e-01 1.92915827e-01 -5.91589510e-01 5.09477437e-01
1.98906153e-01 7.24016368e-01 3.38688381e-02 -1.04554617e+00
-2.83505470e-01 7.52654672e-01 -7.18766823e-02 -2.88568288e-01
1.35959303e+00 -4.30532873e-01 -8.20469677e-01 6.05519891e-01
1.57630944e+00 2.35197932e-01 -3.43410671e-01 -2.91358829e-01
1.98337361e-01 -4.19377178e-01 1.84940800e-01 -2.39961639e-01
-1.26627350e+00 6.60987735e-01 4.13888365e-01 5.13907969e-01
1.02590919e+00 -2.18503680e-02 2.21822560e-01 4.72109646e-01
1.53130800e-01 -9.69267070e-01 9.97247323e-02 3.43921483e-01
6.67250693e-01 -7.79244602e-01 2.28662252e-01 -6.32555366e-01
-4.12057310e-01 1.34286773e+00 3.63180220e-01 -2.22765714e-01
1.27206683e+00 -1.03705432e-02 -3.29207063e-01 -3.82254630e-01
-7.61961341e-01 -3.60023737e-01 5.95699489e-01 2.98262507e-01
6.02373540e-01 1.51302880e-02 -1.02922571e+00 2.08228499e-01
3.16680595e-02 -6.89544022e-01 7.14344263e-01 7.04867721e-01
-2.43716419e-01 -1.71741986e+00 -4.00157988e-01 8.56714964e-01
-3.58091563e-01 -2.12889194e-01 -7.79506862e-01 3.40169221e-01
1.92167729e-01 1.25602984e+00 2.37458721e-02 -5.95281124e-01
-1.35583788e-01 -4.44138907e-02 2.12159023e-01 -9.02102470e-01
-4.33752388e-01 7.64445603e-01 -3.39023359e-02 -3.23400855e-01
-9.80412185e-01 -9.97702420e-01 -1.15388370e+00 -2.17187926e-01
-8.29449296e-01 9.81711805e-01 7.85363495e-01 6.21309698e-01
2.52776593e-01 2.35518724e-01 7.96768188e-01 -7.60434508e-01
1.48605809e-01 -8.89896035e-01 -9.75178421e-01 5.28142214e-01
3.27862240e-03 -8.49814475e-01 -4.65332896e-01 8.98971707e-02] | [7.351987361907959, 4.971467971801758] |
e43534fb-6f44-4d02-b6bf-d91756480747 | rodeo-robust-de-aliasing-autoencoder-for-real | 1912.07519 | null | https://arxiv.org/abs/1912.07519v1 | https://arxiv.org/pdf/1912.07519v1.pdf | RODEO: Robust DE-aliasing autoencOder for Real-time Medical Image Reconstruction | In this work we address the problem of real-time dynamic medical MRI and X Ray CT image reconstruction from parsimonious samples Fourier frequency space for MRI and sinogram tomographic projections for CT. Today the de facto standard for such reconstruction is compressed sensing. CS produces high quality images (with minimal perceptual loss, but such reconstructions are time consuming, requiring solving a complex optimization problem. In this work we propose to learn the reconstruction from training samples using an autoencoder. Our work is based on the universal function approximation capacity of neural networks. The training time for the autoencoder is large, but is offline and hence does not affect performance during operation. During testing or operation, our method requires only a few matrix vector products and hence is significantly faster than CS based methods. In fact, it is fast enough for real-time reconstruction the images are reconstructed as fast as they are acquired with only slight degradation of image quality. However, in order to make the autoencoder suitable for our problem, we depart from the standard Euclidean norm cost function of autoencoders and use a robust l1-norm instead. The ensuing problem is solved using the Split Bregman method. | ['Janki Mehta', 'Angshul Majumdar'] | 2019-12-11 | null | null | null | null | ['de-aliasing'] | ['computer-vision'] | [ 4.34431255e-01 2.09369153e-01 2.78494567e-01 -1.53305799e-01
-4.51864511e-01 -2.33289655e-02 2.52527148e-01 -1.34143770e-01
-7.71261632e-01 6.77643180e-01 -1.45303249e-01 -2.21340686e-01
-3.39105040e-01 -7.61736274e-01 -7.07662582e-01 -9.21077549e-01
-2.19667822e-01 6.04680896e-01 1.20942257e-01 -9.98989642e-02
-7.27115050e-02 5.18859625e-01 -1.32347989e+00 5.51060662e-02
5.93799651e-01 1.16026461e+00 5.32547772e-01 7.79716432e-01
2.33001843e-01 1.12599027e+00 -3.83334577e-01 -1.09014593e-01
4.68753934e-01 -7.39823520e-01 -9.48234141e-01 4.21907425e-01
3.88549231e-02 -5.79545975e-01 -6.05178773e-01 1.07012093e+00
6.81336522e-01 3.93070608e-01 5.65666258e-01 -5.13600111e-01
-2.46228233e-01 5.53062797e-01 -6.86975941e-02 2.88102746e-01
1.32161975e-01 -1.75089896e-01 4.73607838e-01 -7.79901624e-01
7.78163016e-01 4.53766942e-01 6.86222434e-01 5.25756896e-01
-1.37434542e+00 -1.32291079e-01 -5.62709391e-01 3.70151311e-01
-1.34918332e+00 -4.94837731e-01 7.78778255e-01 -2.00967818e-01
7.61732876e-01 4.57842946e-01 9.21135604e-01 6.52996004e-01
4.40032333e-01 5.87938845e-01 1.30631506e+00 -6.09287381e-01
3.82056743e-01 9.83471647e-02 -4.10817415e-01 7.47430503e-01
-3.65699921e-03 2.04971254e-01 -4.91598472e-02 -1.91996805e-02
1.19742024e+00 4.46842045e-01 -7.07458496e-01 -2.67959714e-01
-1.29288328e+00 1.05092394e+00 5.42021155e-01 1.04325414e+00
-7.86942303e-01 1.95053622e-01 2.99542069e-01 4.91932929e-01
2.27235034e-01 4.08471555e-01 6.92407042e-02 8.60000774e-02
-1.37273657e+00 -1.77977718e-02 8.25128198e-01 3.65270615e-01
3.00890386e-01 4.12132680e-01 6.60837591e-01 8.12096417e-01
-2.33789477e-02 4.37843025e-01 1.08832777e+00 -1.12159145e+00
-1.97501719e-01 -1.83966830e-01 -1.99967995e-01 -1.08193135e+00
-4.53232497e-01 -5.14915764e-01 -1.03074098e+00 2.57086366e-01
1.81797564e-01 1.92519780e-02 -7.06298053e-01 1.20648980e+00
3.25901985e-01 2.00275838e-01 4.68293056e-02 1.08549941e+00
3.81069958e-01 7.42444456e-01 -5.24022341e-01 -6.58371091e-01
1.05338156e+00 -5.80667496e-01 -1.04573894e+00 -4.27047797e-02
3.30821067e-01 -7.24070251e-01 5.77499688e-01 8.07514548e-01
-1.46428275e+00 -4.35680419e-01 -1.27920461e+00 2.40805686e-01
1.61782742e-01 -1.52435482e-01 6.11572027e-01 5.02923489e-01
-1.14306056e+00 1.25253928e+00 -1.15926576e+00 5.06987190e-03
-2.40832828e-02 5.23864508e-01 -6.26909614e-01 -2.68971294e-01
-9.41332281e-01 1.17042875e+00 5.45597017e-01 2.30689365e-02
-5.09363234e-01 -5.06188154e-01 -8.38252485e-01 -4.08953987e-02
2.48151541e-01 -4.35225636e-01 1.16160917e+00 -9.78298724e-01
-1.90712440e+00 6.17285669e-01 2.80395091e-01 -6.08705759e-01
4.39035535e-01 1.25318393e-01 -5.43485343e-01 8.63456964e-01
-2.15064019e-01 3.45769048e-01 1.22463822e+00 -1.03223777e+00
1.97896045e-02 -2.05600902e-01 -2.71837503e-01 -8.44759196e-02
-2.38124639e-01 -3.42825383e-01 -4.51026969e-02 -7.03708649e-01
6.26785517e-01 -1.07328439e+00 -4.06267911e-01 1.44633520e-02
1.65360384e-02 4.82425511e-01 4.66063827e-01 -9.28678274e-01
9.72261310e-01 -2.16549182e+00 1.43765256e-01 2.86020249e-01
3.87597024e-01 9.63200442e-03 1.35996431e-01 3.85047227e-01
-4.54698861e-01 -6.01431131e-01 -7.08512962e-01 -1.22320384e-01
-4.64930683e-01 5.49325407e-01 -6.47866726e-02 1.00731647e+00
-2.42447168e-01 4.54135835e-01 -6.93762898e-01 -6.31077766e-01
4.55451310e-01 7.02802002e-01 -7.37694800e-01 2.38977715e-01
3.06796342e-01 5.88954151e-01 -2.39382803e-01 1.63653150e-01
6.34079814e-01 -3.85296375e-01 2.29967967e-01 -3.33921045e-01
5.66982105e-02 -2.46396568e-02 -1.24030793e+00 1.88097489e+00
-8.41165543e-01 5.74689746e-01 3.37650925e-01 -1.62594748e+00
6.29864693e-01 7.51667917e-01 1.16103470e+00 -6.89363778e-01
3.78168434e-01 5.41419923e-01 6.47914782e-02 -5.64259589e-01
2.43590340e-01 -1.00362253e+00 3.96680385e-01 5.99690497e-01
2.64608711e-01 -4.15268630e-01 5.02338819e-02 -1.29913883e-02
1.11242616e+00 -2.93758035e-01 4.29811746e-01 -3.71754974e-01
3.75531971e-01 -6.64930716e-02 2.79173851e-01 3.51183563e-01
6.92081153e-02 6.70953691e-01 -1.41340405e-01 -5.84599912e-01
-1.19615376e+00 -7.96651185e-01 -6.41704261e-01 1.80257395e-01
-1.41059443e-01 9.85146016e-02 -8.25950086e-01 -2.68501848e-01
-3.71412814e-01 5.77775240e-01 -5.10507464e-01 -1.44122526e-01
-9.80745316e-01 -7.21265376e-01 2.37436071e-01 3.50317895e-01
2.85622984e-01 -1.01632667e+00 -1.02096653e+00 5.98818481e-01
-2.76754379e-01 -1.19240844e+00 -2.76444107e-01 3.33062857e-01
-1.37755513e+00 -9.90097284e-01 -1.20620024e+00 -6.85251355e-01
7.49010980e-01 3.32699716e-02 9.40377295e-01 7.99842179e-04
-3.92874837e-01 3.95154774e-01 -3.09429914e-01 2.22380236e-02
-5.57721317e-01 -6.72618747e-01 2.18108892e-01 1.05388686e-01
-2.89010525e-01 -9.85842347e-01 -7.65510917e-01 7.68461898e-02
-1.27781713e+00 -1.67843208e-01 6.28477931e-01 1.24195695e+00
8.37058127e-01 4.85152662e-01 1.32215768e-01 -7.89293468e-01
3.14393401e-01 -2.91450351e-01 -5.70132971e-01 -3.19327086e-01
-7.29834497e-01 1.33671373e-01 1.13126457e+00 -4.78832096e-01
-7.17403471e-01 1.48409933e-01 -4.93071258e-01 -7.43533731e-01
7.91410953e-02 4.90251362e-01 5.12340307e-01 -4.56240505e-01
8.19961190e-01 6.65853858e-01 3.53829443e-01 -3.89325440e-01
-9.35243908e-03 3.67545515e-01 7.12218225e-01 -2.79660616e-02
5.63200235e-01 6.92273021e-01 1.40306190e-01 -1.02448297e+00
-1.19375974e-01 -2.82535434e-01 -2.81797498e-01 -2.89431691e-01
9.11858022e-01 -5.04417539e-01 -7.86405206e-01 -6.73666447e-02
-8.46856952e-01 -1.23493172e-01 -6.70925558e-01 1.12372339e+00
-1.06540346e+00 5.71040273e-01 -9.95265245e-01 -5.65177560e-01
-2.61044085e-01 -1.30324686e+00 5.48554838e-01 -3.07012528e-01
1.97960928e-01 -9.67568636e-01 9.42327827e-02 1.70795426e-01
6.45165205e-01 4.41289216e-01 5.55182636e-01 -2.08212867e-01
-3.83800894e-01 -5.29828906e-01 3.74228835e-01 6.57409966e-01
-1.45269781e-01 -6.75249755e-01 -8.05868089e-01 -4.91940767e-01
1.16568410e+00 -2.20559642e-01 6.44489825e-01 7.28485227e-01
1.41715848e+00 -4.19108272e-01 1.49736956e-01 7.81097591e-01
1.72853529e+00 3.13238114e-01 5.85820913e-01 3.17110382e-02
3.04983974e-01 5.68449855e-01 2.61670649e-01 3.13952923e-01
-1.69583097e-01 4.22312468e-01 2.80684978e-01 -4.07283120e-02
-3.32285054e-02 8.16367865e-02 1.86877966e-01 1.41905415e+00
-3.28288287e-01 1.03816077e-01 -7.50837624e-01 5.11971772e-01
-1.38328218e+00 -1.05032349e+00 1.03848092e-01 2.18653059e+00
7.06480801e-01 -1.80484876e-01 -1.12062849e-01 8.98860514e-01
2.81990111e-01 3.48044783e-02 -2.49576122e-01 -2.79684842e-01
2.18454614e-01 4.99310166e-01 5.28158486e-01 6.83019400e-01
-7.62840033e-01 9.51399803e-02 6.39018536e+00 8.41410518e-01
-1.56989884e+00 4.47330058e-01 4.00090694e-01 -1.85640469e-01
-1.12333901e-01 -2.61859983e-01 1.91125706e-01 5.07187784e-01
1.25487721e+00 1.73301473e-01 6.96830571e-01 9.17278469e-01
-9.10097286e-02 -3.15551609e-01 -8.46706927e-01 1.36881328e+00
1.73752442e-01 -1.31812549e+00 -4.85687643e-01 3.85947786e-02
4.74923640e-01 1.15188375e-01 -5.99227659e-02 -1.60474166e-01
-2.66071856e-01 -1.07103324e+00 5.88455796e-01 3.21106076e-01
9.61006045e-01 -7.41692722e-01 6.48915589e-01 5.50224602e-01
-6.38829648e-01 1.23660369e-02 -5.27232528e-01 2.51477927e-01
4.44064766e-01 1.03108251e+00 -5.63584745e-01 3.59215289e-01
4.12109882e-01 4.01066989e-01 2.30176941e-01 9.15537715e-01
2.35800356e-01 5.30440450e-01 -4.79549676e-01 2.21776888e-01
2.73037344e-01 -2.99121857e-01 6.56820655e-01 9.44932163e-01
4.69707876e-01 4.72175717e-01 2.08090946e-01 5.28706551e-01
3.14496487e-01 2.93759629e-02 -6.33557618e-01 -3.66232321e-02
-1.14682175e-01 9.20457602e-01 -8.55431378e-01 -4.50136811e-01
-2.00491041e-01 1.33914423e+00 -3.11701804e-01 1.28821403e-01
-4.57563251e-01 -2.80046165e-01 -1.41404539e-01 4.06668931e-01
3.90899330e-01 -1.69250071e-01 5.99788390e-02 -1.37158608e+00
6.01303093e-02 -8.94865155e-01 1.58889204e-01 -4.85303313e-01
-8.75064850e-01 1.07215166e+00 2.95246430e-02 -1.45841146e+00
-7.56584108e-01 -5.54790378e-01 -2.42884502e-01 6.20775759e-01
-1.41161704e+00 -5.75292230e-01 -2.83668637e-02 9.50416386e-01
4.42846298e-01 9.97425392e-02 1.17047906e+00 6.61256671e-01
7.06737414e-02 6.65028319e-02 3.93247217e-01 -5.08135408e-02
2.52096117e-01 -1.25016987e+00 -2.89825976e-01 7.57923484e-01
1.49982437e-01 4.93133813e-01 9.77876127e-01 -3.29175651e-01
-1.48293746e+00 -4.44986075e-01 7.31941640e-01 3.71881664e-01
5.39742112e-01 5.65482527e-02 -8.76398206e-01 5.42168677e-01
1.28426373e-01 5.87309778e-01 6.24603868e-01 -4.13554370e-01
2.39343837e-01 -1.50033934e-02 -1.40513361e+00 1.48396315e-02
4.82477486e-01 -6.05251610e-01 -4.37674522e-01 5.92161298e-01
3.19957912e-01 -4.38554078e-01 -1.34764528e+00 7.73782730e-02
4.09441292e-01 -1.15633309e+00 1.04972100e+00 -3.03181652e-02
5.81843793e-01 -1.98177204e-01 -2.59435982e-01 -1.35185659e+00
-3.33345473e-01 -4.58297580e-01 -2.15018511e-01 -1.98411960e-02
8.62648040e-02 -6.45518422e-01 7.72599459e-01 3.24351996e-01
-2.20713675e-01 -9.42394018e-01 -1.03386962e+00 -9.47024763e-01
-1.63083866e-01 -4.51404601e-01 1.27905399e-01 1.11934388e+00
1.22342750e-01 -7.82307759e-02 -4.86395717e-01 -7.32883438e-02
9.67439294e-01 2.65583128e-01 2.59723961e-02 -8.86012495e-01
-1.04978919e+00 -2.24782631e-01 -5.43642819e-01 -9.70254719e-01
-2.12250859e-01 -8.47553968e-01 1.37552187e-01 -1.04898095e+00
-6.94542304e-02 -4.19092953e-01 -1.49252906e-01 -2.59522852e-02
2.68971145e-01 3.95467311e-01 2.60222077e-01 3.81320387e-01
1.12476893e-01 4.23152894e-01 1.38329160e+00 -1.16244117e-02
6.04959093e-02 1.06931813e-01 2.63973419e-02 8.11472058e-01
5.51369071e-01 -6.19406164e-01 -5.11308074e-01 -4.52173799e-01
-3.15249376e-02 9.37641680e-01 2.24030182e-01 -1.19889343e+00
3.14184576e-01 1.73194870e-01 4.65551794e-01 -2.54996866e-01
5.62244713e-01 -1.28282344e+00 3.74083906e-01 1.00859010e+00
-7.94805810e-02 1.07152157e-01 -2.18051657e-01 4.55757231e-01
-5.47645926e-01 -6.32900119e-01 1.17225468e+00 -5.09565711e-01
-2.91672140e-01 2.63811171e-01 -5.01096129e-01 -4.24887717e-01
6.91435516e-01 -2.86695451e-01 6.03021026e-01 -5.51059544e-01
-1.25039732e+00 -5.57807267e-01 3.85366261e-01 -4.36981708e-01
9.36126947e-01 -1.26264274e+00 -5.80213428e-01 3.13572049e-01
-4.16837990e-01 -1.56226128e-01 5.86242080e-01 1.26607180e+00
-1.03612292e+00 3.69514942e-01 -4.65792567e-01 -5.67936838e-01
-8.06312859e-01 8.24707091e-01 4.79887843e-01 -3.80144179e-01
-1.23689067e+00 7.02758431e-01 -1.23404130e-01 -1.29448429e-01
-1.92098230e-01 -8.82831737e-02 -7.28093460e-02 -3.42134207e-01
8.14022303e-01 2.42391750e-01 3.49512070e-01 -5.23006856e-01
-1.28726393e-01 3.95469218e-01 1.25440091e-01 -5.02335012e-01
1.77191305e+00 1.43713713e-01 -8.21228027e-02 2.98342496e-01
1.57486677e+00 -7.20132068e-02 -9.95601416e-01 -1.90019801e-01
-4.00555670e-01 -5.13850451e-01 7.36843646e-01 -3.53620827e-01
-1.39623940e+00 9.85749424e-01 7.62401819e-01 3.38749588e-01
1.49663007e+00 -3.33790660e-01 1.12285912e+00 3.94719392e-01
5.03171504e-01 -1.08011067e+00 2.32316460e-02 -2.49409694e-02
1.03167832e+00 -1.09257150e+00 2.13942662e-01 -2.36659661e-01
-3.93604785e-01 1.19325686e+00 -1.87243134e-01 -6.56314909e-01
7.68834472e-01 4.28911805e-01 -6.06159046e-02 -2.27295488e-01
-2.64118701e-01 1.57178640e-01 2.51289546e-01 3.58195901e-01
2.93855548e-01 1.07679993e-01 -5.46420276e-01 -1.70140695e-02
-3.27325523e-01 1.03439666e-01 6.64646029e-01 9.91686463e-01
-4.20561969e-01 -1.03926480e+00 -6.46671832e-01 4.39685494e-01
-8.52820337e-01 3.09380330e-02 4.55189317e-01 6.37495339e-01
-1.72951281e-01 5.41476607e-01 -4.15527038e-02 -1.82201549e-01
2.25799102e-02 -2.03866243e-01 9.05398667e-01 -2.08195344e-01
-3.70450348e-01 4.33928847e-01 -2.48780906e-01 -8.14050019e-01
-5.08155942e-01 -6.87167346e-01 -1.40733290e+00 -4.31170076e-01
-4.22626764e-01 2.72070020e-01 1.12796807e+00 7.25370049e-01
-2.53662348e-01 5.62999904e-01 7.79572487e-01 -8.98919404e-01
-7.07163155e-01 -8.25626135e-01 -8.92148614e-01 3.53612423e-01
5.38302243e-01 -4.30117160e-01 -2.19375581e-01 7.83252642e-02] | [13.437948226928711, -2.466742515563965] |
b3045f39-e9e2-4580-89ea-ba2ddbd83125 | is-rewiring-actually-helpful-in-graph-neural | 2305.19717 | null | https://arxiv.org/abs/2305.19717v1 | https://arxiv.org/pdf/2305.19717v1.pdf | Is Rewiring Actually Helpful in Graph Neural Networks? | Graph neural networks compute node representations by performing multiple message-passing steps that consist in local aggregations of node features. Having deep models that can leverage longer-range interactions between nodes is hindered by the issues of over-smoothing and over-squashing. In particular, the latter is attributed to the graph topology which guides the message-passing, causing a node representation to become insensitive to information contained at distant nodes. Many graph rewiring methods have been proposed to remedy or mitigate this problem. However, properly evaluating the benefits of these methods is made difficult by the coupling of over-squashing with other issues strictly related to model training, such as vanishing gradients. Therefore, we propose an evaluation setting based on message-passing models that do not require training to compute node and graph representations. We perform a systematic experimental comparison on real-world node and graph classification tasks, showing that rewiring the underlying graph rarely does confer a practical benefit for message-passing. | ['Alessio Micheli', 'Domenico Tortorella'] | 2023-05-31 | null | null | null | null | ['graph-classification'] | ['graphs'] | [ 3.26725356e-02 4.12607580e-01 -3.75386924e-02 -3.09224665e-01
-1.74257234e-01 -4.26853269e-01 9.02061284e-01 8.89209330e-01
-3.00734401e-01 5.48540056e-01 1.42966462e-02 -4.35004085e-01
-1.60096914e-01 -1.10507870e+00 -6.70490682e-01 -6.04281068e-01
-8.19733381e-01 1.90427706e-01 4.58156377e-01 -3.40791672e-01
9.31578651e-02 5.61577141e-01 -1.22829843e+00 -7.96970949e-02
5.82793474e-01 5.61549127e-01 -1.26942456e-01 8.26786876e-01
-1.53512567e-01 7.61361361e-01 -6.35927498e-01 -5.10372758e-01
1.74876675e-01 -3.35388601e-01 -5.69530547e-01 -9.09991041e-02
4.67049152e-01 -3.57860774e-01 -5.91546595e-01 1.01858425e+00
6.64022341e-02 2.07891390e-01 5.87174237e-01 -1.42212999e+00
-5.40827990e-01 9.07691658e-01 -6.25757933e-01 2.06210658e-01
-2.90283132e-02 8.77045244e-02 1.30666518e+00 -4.14387792e-01
5.91900885e-01 1.41123676e+00 9.70691741e-01 3.17022115e-01
-1.48165572e+00 -4.88800526e-01 5.14779627e-01 -2.80573189e-01
-1.33879244e+00 -2.05580771e-01 8.82845640e-01 -2.51788437e-01
8.24113727e-01 2.06523180e-01 6.17193103e-01 9.97527838e-01
3.20012450e-01 5.50878584e-01 4.34805036e-01 -5.30904271e-02
3.28916818e-01 -1.04066424e-01 3.98653686e-01 8.42145562e-01
6.07030153e-01 -4.71721999e-02 -2.00865343e-01 -3.00270081e-01
7.95615196e-01 1.15796603e-01 -1.69078171e-01 -5.05507231e-01
-8.22562218e-01 8.78509820e-01 1.30496883e+00 4.87784952e-01
-3.25216025e-01 6.09799862e-01 4.04441714e-01 5.13032854e-01
5.70608914e-01 5.91597199e-01 -2.40612879e-01 1.98805556e-01
-8.50321531e-01 1.32527798e-01 9.09756422e-01 5.68886518e-01
1.07007599e+00 1.83299091e-02 3.22567225e-02 4.33213711e-01
4.95527238e-01 -5.62505610e-02 8.91605914e-02 -3.80579919e-01
3.93224210e-01 8.84892523e-01 -3.74663413e-01 -1.63901770e+00
-6.03113353e-01 -6.96993172e-01 -1.07360446e+00 1.01660050e-01
5.66139638e-01 -1.97335616e-01 -1.01831138e+00 1.95877039e+00
2.68202305e-01 4.60830152e-01 -1.57628253e-01 7.01344550e-01
8.26166809e-01 3.61547977e-01 4.78678852e-01 2.26469696e-01
1.08085525e+00 -8.12740326e-01 -3.01795721e-01 -3.71573299e-01
1.08164525e+00 -1.98995098e-01 7.88101137e-01 -2.50995666e-01
-8.61544251e-01 -1.66330025e-01 -1.14603221e+00 -7.44423792e-02
-6.30016088e-01 -4.45426524e-01 1.02735698e+00 5.34703493e-01
-1.35161340e+00 1.14737797e+00 -1.14449096e+00 -6.04615331e-01
6.12171769e-01 3.88105363e-01 -5.24939775e-01 9.03584436e-02
-1.10493362e+00 7.70557582e-01 2.04497710e-01 1.55212104e-01
-8.78371119e-01 -7.72588313e-01 -9.74120498e-01 5.02501726e-01
2.29996786e-01 -5.93853712e-01 6.98817730e-01 -9.39156771e-01
-1.06103420e+00 5.05929232e-01 9.92535055e-02 -4.53240484e-01
4.09643531e-01 1.02734625e-01 -4.37819064e-02 -3.70757617e-02
-3.21471900e-01 4.49056327e-01 8.17401052e-01 -1.13596630e+00
1.52807264e-02 -3.86875063e-01 4.96857315e-01 1.43326148e-01
-5.41349292e-01 -3.48077625e-01 -3.18717182e-01 -6.27680302e-01
2.39861593e-01 -8.86354506e-01 -5.69102466e-01 9.61289480e-02
-6.63048089e-01 -1.47062317e-01 7.19556630e-01 -1.76187083e-01
1.09207702e+00 -2.07997608e+00 3.65474098e-03 4.01064277e-01
8.54501545e-01 7.60626867e-02 -6.53729916e-01 7.62289524e-01
2.93022059e-02 5.38861275e-01 -4.23545726e-02 -4.64788735e-01
-3.02033186e-01 1.82029024e-01 -1.33504540e-01 6.06480122e-01
3.26984793e-01 1.05717123e+00 -1.01371205e+00 -2.39160016e-01
-1.18817657e-01 7.31078804e-01 -5.90591013e-01 -1.70844905e-02
-1.62567958e-01 -4.28855382e-02 -4.92945313e-01 2.44439498e-01
6.52247369e-01 -7.02226758e-01 4.82009411e-01 7.27026612e-02
4.11176324e-01 6.10865176e-01 -8.71465266e-01 1.46663785e+00
-2.11715490e-01 5.68105102e-01 3.08042318e-01 -1.02664244e+00
6.88526869e-01 -2.68039219e-02 2.90295601e-01 -2.65295088e-01
1.43717453e-01 -1.41767591e-01 1.37268081e-01 1.64218079e-02
5.61728179e-01 1.86339870e-01 2.91602105e-01 5.92124701e-01
-3.94260651e-03 4.35960293e-02 1.40414611e-01 8.49955976e-01
1.65995133e+00 -2.73763299e-01 -6.03939686e-03 -6.23789907e-01
-6.11488074e-02 -3.06751668e-01 2.63207704e-01 1.00384033e+00
-3.26288007e-02 4.11719888e-01 8.89917433e-01 -4.25528854e-01
-7.43061900e-01 -9.41963673e-01 3.39544326e-01 1.25966775e+00
2.12803632e-01 -7.95395911e-01 -7.39799917e-01 -8.01603138e-01
1.10228159e-01 2.97247678e-01 -9.04197872e-01 -5.60540795e-01
-4.19224799e-01 -9.78128254e-01 6.50600731e-01 4.98002023e-01
1.02209792e-01 -6.28179252e-01 -4.43982989e-01 4.11222667e-01
3.21730316e-01 -9.33403075e-01 -2.64364958e-01 1.94781438e-01
-1.24626279e+00 -1.06123102e+00 -5.58915675e-01 -5.79630077e-01
1.06662714e+00 5.89069903e-01 1.26067293e+00 1.18032956e+00
-1.37289733e-01 5.19364513e-02 -2.08091855e-01 2.85713412e-02
-4.10889596e-01 6.82242036e-01 -1.95779487e-01 -1.57582704e-02
8.27013105e-02 -1.04882896e+00 -6.61090612e-01 7.58140013e-02
-9.32068646e-01 3.94368991e-02 6.68193996e-01 6.90296113e-01
-1.47724748e-01 -2.20308006e-02 4.87167448e-01 -1.04759836e+00
8.87856781e-01 -7.02012360e-01 -5.22596776e-01 2.42075041e-01
-6.25102401e-01 3.27654630e-01 4.69769239e-01 -5.42437494e-01
-5.34649193e-01 -1.63263142e-01 2.16138586e-02 -4.81171161e-02
6.80917203e-02 7.95357704e-01 2.88592100e-01 -3.02642614e-01
6.66916490e-01 -2.66607553e-01 1.87304184e-01 -4.26883548e-01
4.27924305e-01 7.97440298e-03 8.22943449e-02 -4.48381931e-01
1.01518440e+00 4.05465573e-01 2.56828547e-01 -1.01497102e+00
-5.56755602e-01 1.44799039e-01 -3.61866146e-01 -5.65238185e-02
5.40226221e-01 -6.94804728e-01 -6.95106208e-01 4.18264419e-01
-1.04082227e+00 -4.95183289e-01 1.09387999e-02 2.22007930e-01
2.27825478e-01 4.33065027e-01 -1.00533986e+00 -5.56816518e-01
-2.21956044e-01 -9.77585614e-01 5.92707515e-01 2.26897895e-01
-2.58277178e-01 -1.44040442e+00 -1.39409890e-02 -3.39421809e-01
9.02151287e-01 3.20634693e-01 1.24242222e+00 -9.15037215e-01
-7.35839844e-01 -3.40476513e-01 -4.77056712e-01 -1.05470113e-01
-2.11225096e-02 1.88864142e-01 -9.39466059e-01 -6.36781156e-01
-5.37904203e-01 -2.31084079e-01 1.08856654e+00 2.16633201e-01
8.67788672e-01 -2.94697642e-01 -6.31884158e-01 4.85433817e-01
1.22444391e+00 -5.63202083e-01 5.07827461e-01 2.92413607e-02
9.35213268e-01 6.21631563e-01 -2.73035049e-01 1.02914423e-01
6.22496009e-01 3.75446230e-01 7.87713349e-01 -3.94723922e-01
-1.78522855e-01 -5.84713697e-01 2.97615398e-02 5.86444974e-01
-1.62033848e-02 -4.05033588e-01 -9.69407082e-01 5.54979444e-01
-1.77278388e+00 -4.93836045e-01 -2.13303000e-01 2.12245250e+00
4.91679668e-01 3.94288868e-01 -5.84471971e-02 -7.45961862e-03
5.77314794e-01 5.79513133e-01 -3.93034905e-01 -1.05040751e-01
1.40667930e-01 -1.53650194e-02 5.38877666e-01 7.65024483e-01
-8.84621143e-01 9.61533368e-01 6.57523537e+00 4.57012326e-01
-1.18411231e+00 -4.98027056e-02 7.40681708e-01 1.60012051e-01
-5.71362376e-01 2.37412006e-01 -4.79943991e-01 1.59813970e-01
7.94094324e-01 1.81968720e-03 5.47479212e-01 6.96564198e-01
-1.89227819e-01 -7.35630654e-03 -1.24269581e+00 4.49526966e-01
-3.20580542e-01 -1.34722841e+00 1.50149822e-01 2.31873363e-01
4.24021482e-01 4.04684275e-01 -1.70256153e-01 3.40570033e-01
7.08705366e-01 -1.16391969e+00 2.73266703e-01 1.48549289e-01
2.07541913e-01 -4.11003917e-01 4.56318676e-01 1.48737133e-01
-1.25877225e+00 1.63168430e-01 -4.11344528e-01 -3.65083814e-01
-2.40614250e-01 9.06212330e-01 -9.96171594e-01 3.12211633e-01
3.69706064e-01 6.89808488e-01 -9.70885456e-01 7.67994761e-01
-2.88694233e-01 6.80130482e-01 -5.92404246e-01 -2.44737476e-01
3.85716587e-01 -4.43410277e-02 5.00197768e-01 9.53639686e-01
1.67623088e-01 -2.76964545e-01 1.84642211e-01 1.07829988e+00
-3.56070578e-01 -9.16464254e-02 -7.74849474e-01 -3.55444103e-01
6.89274013e-01 1.45468605e+00 -1.12392282e+00 -2.82262325e-01
-4.19008672e-01 6.73127234e-01 8.27354133e-01 6.13117993e-01
-5.73150694e-01 -3.98301721e-01 8.66163433e-01 4.80466574e-01
1.94768772e-01 -4.03621912e-01 -1.81006491e-01 -1.12556517e+00
-1.73366740e-01 -4.95778799e-01 3.34326804e-01 -2.80373126e-01
-1.25629508e+00 5.92851996e-01 -2.34119862e-01 -5.25162578e-01
2.70901211e-02 -2.69345254e-01 -1.02652991e+00 5.75686336e-01
-1.47841442e+00 -1.11495459e+00 -2.59540409e-01 3.71410877e-01
9.42889042e-03 2.67172158e-01 6.92889750e-01 1.65225327e-01
-4.58598465e-01 7.05941200e-01 -3.67498904e-01 2.59827435e-01
2.43966162e-01 -1.17413270e+00 9.10909235e-01 8.59919488e-01
1.96505621e-01 9.15440917e-01 7.43025899e-01 -6.93931580e-01
-1.49490702e+00 -9.69106376e-01 8.88919592e-01 -4.18453068e-01
8.50989819e-01 -7.09093034e-01 -1.32470560e+00 8.84442568e-01
-1.60507802e-02 3.05025190e-01 1.89883098e-01 6.56139612e-01
-6.38767481e-01 -1.03664964e-01 -8.88594270e-01 6.77628875e-01
1.18103302e+00 -5.34564972e-01 -9.80849937e-02 1.05135657e-01
7.21954584e-01 -5.36857545e-02 -8.30680668e-01 2.20885068e-01
3.01120222e-01 -7.97170937e-01 9.66353178e-01 -6.96616054e-01
3.56656164e-01 -2.37988699e-02 3.48195821e-01 -1.51058233e+00
-3.96678030e-01 -6.09367192e-01 5.13199791e-02 1.17012310e+00
6.33458674e-01 -8.44479799e-01 9.05541003e-01 7.78155267e-01
2.09426045e-01 -5.16522050e-01 -6.66715801e-01 -4.69088733e-01
2.08487079e-01 -1.18851468e-01 7.59627700e-01 1.12799633e+00
1.93846747e-01 5.97534597e-01 -2.11294949e-01 2.94700831e-01
6.61101341e-01 -1.90505669e-01 7.87584722e-01 -1.57340908e+00
-1.29530206e-01 -6.08699858e-01 -6.34220958e-01 -1.05442417e+00
6.69350326e-02 -1.07000613e+00 -1.80840753e-02 -1.41637707e+00
3.00616007e-02 -6.23186171e-01 -4.14458305e-01 5.81540108e-01
-3.36642712e-01 8.40817168e-02 2.03896075e-01 1.43483356e-01
-5.89419544e-01 4.19892728e-01 1.11210978e+00 -4.43710834e-02
-1.46584198e-01 -2.06451461e-01 -8.33102226e-01 6.37201190e-01
7.82450795e-01 -6.57357454e-01 -5.82395613e-01 -6.97786629e-01
5.77805102e-01 -3.14749219e-02 5.95213354e-01 -9.23199952e-01
3.76312882e-01 1.10458411e-01 4.12583649e-02 -9.08258036e-02
2.45371774e-01 -6.80424750e-01 4.28059548e-02 6.75851405e-01
-5.74488282e-01 1.65201202e-01 -5.25401607e-02 8.16929936e-01
1.33055747e-01 5.29642142e-02 6.22917235e-01 -1.35306880e-01
-2.33390003e-01 3.25543582e-01 -4.01889324e-01 -8.69275182e-02
8.23410034e-01 1.39370143e-01 -5.17474651e-01 -6.30500853e-01
-4.89149839e-01 2.78141886e-01 5.87928236e-01 4.86752421e-01
3.05617511e-01 -9.70826507e-01 -5.96509337e-01 3.69829953e-01
-4.14184742e-02 -6.93472922e-02 -2.25999262e-02 6.91950202e-01
-4.00570005e-01 -6.59608468e-03 1.20234996e-01 -4.80316162e-01
-1.03634143e+00 3.42138559e-01 3.89766186e-01 -5.37418306e-01
-7.70084441e-01 9.04445112e-01 1.91577330e-01 -4.46839482e-01
4.00427610e-01 -5.48063099e-01 -1.73709337e-02 -9.49548185e-02
2.90557265e-01 2.18952954e-01 1.46645427e-01 -2.34169707e-01
-3.72042954e-01 1.62314489e-01 -4.34982300e-01 3.24566752e-01
1.31298208e+00 -7.59775192e-02 -2.14363098e-01 2.01110855e-01
1.15633488e+00 -3.20915371e-01 -1.23179626e+00 -3.20493430e-01
2.49475926e-01 -2.84240991e-01 2.16923892e-01 -4.31574583e-01
-1.37743771e+00 8.81604314e-01 1.38316795e-01 8.13497365e-01
5.33133388e-01 -2.22578030e-02 6.44417584e-01 6.33736193e-01
2.75804728e-01 -7.05706835e-01 1.66895445e-02 4.46576774e-01
6.32017136e-01 -1.18945754e+00 6.86417222e-02 -4.69042957e-01
-7.56191984e-02 8.75463963e-01 6.04353905e-01 -5.12766778e-01
9.77779090e-01 3.52473289e-01 -7.09040314e-02 -5.28859437e-01
-9.63370323e-01 1.32451933e-02 -2.24081278e-02 4.06974643e-01
4.74411488e-01 9.90153924e-02 -2.12968871e-01 1.06302872e-01
3.91054675e-02 -3.25709820e-01 4.75052297e-01 7.69430041e-01
-3.63378316e-01 -9.03598011e-01 1.57025352e-01 5.71961284e-01
-4.01155055e-01 -1.55367583e-01 -5.60336769e-01 9.05532360e-01
-6.42265260e-01 8.46304297e-01 1.77926704e-01 -4.63514715e-01
-8.44161883e-02 -1.67291373e-01 1.80481106e-01 -5.21926284e-01
-7.17749476e-01 -3.67441535e-01 1.78071514e-01 -6.24881804e-01
-1.84759781e-01 -1.22425757e-01 -1.20722663e+00 -7.74563909e-01
-3.65994751e-01 5.22334278e-02 6.96707666e-01 7.41772711e-01
6.59262061e-01 6.84333384e-01 2.70559937e-01 -9.24764216e-01
-7.04580724e-01 -9.65558887e-01 -6.20628774e-01 7.15786219e-01
5.43839931e-01 -4.73201901e-01 -7.10956395e-01 -6.15012825e-01] | [6.9446492195129395, 6.17426872253418] |
7c869e97-55e1-4d97-aa67-24847a95af84 | a-driver-fatigue-recognition-algorithm-based | 2003.08134 | null | https://arxiv.org/abs/2003.08134v1 | https://arxiv.org/pdf/2003.08134v1.pdf | A Driver Fatigue Recognition Algorithm Based on Spatio-Temporal Feature Sequence | Researches show that fatigue driving is one of the important causes of road traffic accidents, so it is of great significance to study the driver fatigue recognition algorithm to improve road traffic safety. In recent years, with the development of deep learning, the field of pattern recognition has made great development. This paper designs a real-time fatigue state recognition algorithm based on spatio-temporal feature sequence, which can be mainly applied to the scene of fatigue driving recognition. The algorithm is divided into three task networks: face detection network, facial landmark detection and head pose estimation network, fatigue recognition network. Experiments show that the algorithm has the advantages of small volume, high speed and high accuracy. | ['Xiaobo Lu', 'Chen Zhang', 'Zhiliang Huang'] | 2020-03-18 | null | null | null | null | ['head-pose-estimation'] | ['computer-vision'] | [-5.07178754e-02 -6.66091800e-01 4.07839380e-02 -4.25060809e-01
7.52959624e-02 2.77037710e-01 1.27958864e-01 -9.80339050e-01
-5.79206467e-01 4.74291205e-01 7.69510567e-02 1.77423488e-02
-2.93224633e-01 -3.87563497e-01 -4.70188558e-02 -8.75350893e-01
2.24020258e-01 7.30780736e-02 3.36090326e-01 -1.73673719e-01
4.70632255e-01 8.48942935e-01 -1.95597816e+00 -2.31374800e-01
6.77702367e-01 9.97874260e-01 1.38527155e-01 2.24958375e-01
1.96271360e-01 3.55055630e-01 -4.16737795e-01 2.28783771e-01
-2.81225145e-01 -3.70401919e-01 -4.78581399e-01 1.29561558e-01
-1.47299111e-01 -3.77242416e-01 -8.42930794e-01 8.13107967e-01
1.00679922e+00 2.72493154e-01 4.45259601e-01 -1.36314023e+00
-3.80815268e-02 -2.45709091e-01 -6.01463675e-01 7.56143272e-01
2.24736094e-01 2.20339626e-01 -9.66166034e-02 -8.69420886e-01
2.56406199e-02 1.34585035e+00 3.15811843e-01 7.24650323e-01
-4.03418243e-01 -1.10914385e+00 -2.79716998e-01 1.07778800e+00
-1.64149368e+00 -6.70183182e-01 8.53516877e-01 -2.80182421e-01
8.21664274e-01 -1.75715033e-02 5.58609366e-01 8.69193554e-01
7.66452610e-01 8.35415959e-01 1.01193726e+00 -1.25590354e-01
-1.47651330e-01 -3.04956466e-01 4.00819331e-01 7.02582836e-01
2.04466611e-01 6.44830406e-01 -4.19166058e-01 5.34718752e-01
5.53105831e-01 1.53584152e-01 -2.10674316e-01 1.65170148e-01
-7.15452731e-01 5.83643258e-01 3.12955886e-01 5.53041816e-01
-2.31703237e-01 8.52521136e-02 5.65548241e-01 1.18355647e-01
3.00566107e-01 -4.31002140e-01 -3.51511128e-02 -4.49737042e-01
-7.76400030e-01 6.49245642e-03 2.62113661e-01 4.09221828e-01
4.26888645e-01 2.84440339e-01 1.46672726e-01 6.35113955e-01
4.60308135e-01 9.17939961e-01 7.98453152e-01 -7.18518138e-01
-1.18966602e-01 4.45285290e-01 -2.65783906e-01 -1.09505630e+00
-9.03937399e-01 -3.00940216e-01 -9.40928876e-01 1.90492406e-01
-8.62837210e-02 -2.75919974e-01 -7.75139987e-01 1.31784844e+00
3.01860094e-01 4.20428932e-01 -3.46825629e-01 1.07703018e+00
1.06571186e+00 4.55284476e-01 4.58116531e-02 -4.99469697e-01
1.33193648e+00 -5.44027090e-01 -1.16170430e+00 -2.56550610e-01
4.11935836e-01 -5.70182920e-01 6.64481640e-01 5.05917251e-01
-6.82629764e-01 -1.00071311e+00 -9.05747712e-01 2.05536798e-01
-3.87394637e-01 2.57185400e-01 5.83605707e-01 8.90771687e-01
-7.25701630e-01 1.03845924e-01 -6.80126190e-01 -3.63259912e-01
4.58632052e-01 6.78193510e-01 -4.90399778e-01 -3.05628777e-01
-1.47922802e+00 1.24690187e+00 3.91907364e-01 8.11894298e-01
-8.20751011e-01 -2.93642823e-02 -8.56085598e-01 -1.08203545e-01
2.93224424e-01 -4.62031990e-01 1.00143051e+00 -6.32453799e-01
-1.23384821e+00 7.05632210e-01 -7.15128064e-01 1.48432598e-01
1.29755363e-01 -2.20884159e-01 -1.11443126e+00 -1.62371814e-01
-6.99417293e-02 3.67879629e-01 9.35447752e-01 -6.08470380e-01
-6.73210382e-01 -7.94616580e-01 -7.17925370e-01 8.87572840e-02
-1.32424831e-01 4.12916660e-01 -2.97293693e-01 2.90277392e-01
-2.52361983e-01 -8.83455217e-01 1.98010519e-01 -4.11143571e-01
-1.48460325e-02 -8.96784186e-01 1.38903069e+00 -5.19913375e-01
1.41283846e+00 -2.36531544e+00 -1.66667387e-01 1.72649011e-01
1.68300405e-01 5.34454584e-01 7.61651397e-02 -2.33359523e-02
-1.74998298e-01 -5.20558536e-01 -6.34810999e-02 1.26597270e-01
-3.54654372e-01 3.52994323e-01 3.09857905e-01 7.53929436e-01
2.42181659e-01 9.63247478e-01 -5.62859535e-01 -5.38323104e-01
4.08188373e-01 5.46947241e-01 -2.28377003e-02 1.22247465e-01
5.74887395e-01 4.87332970e-01 -5.95414758e-01 4.19312954e-01
8.70693922e-01 4.10449475e-01 -5.16994417e-01 -3.04827094e-01
-4.05909359e-01 -2.74953067e-01 -1.13847625e+00 1.19403541e+00
-1.83088243e-01 9.12841141e-01 2.32824888e-02 -1.09118330e+00
1.17070627e+00 4.18294460e-01 4.68073547e-01 -1.16118979e+00
9.24647570e-01 1.08845524e-01 3.14509660e-01 -1.27567554e+00
8.28646645e-02 -2.68721968e-01 3.76559538e-03 4.02231246e-01
-3.06549937e-01 2.70771742e-01 1.92334801e-01 -3.92197669e-01
8.18773210e-01 -3.07716466e-02 -2.96266526e-01 -2.02085033e-01
1.01403594e+00 -3.73391896e-01 6.16038144e-01 -1.65518209e-01
-8.06014061e-01 9.24802721e-02 1.85561642e-01 -5.48150182e-01
-5.19150555e-01 -7.30804682e-01 -1.82457745e-01 8.04783940e-01
2.91840494e-01 3.01857054e-01 -8.37778091e-01 -2.70942003e-01
7.32286274e-02 3.27309817e-01 -7.78343260e-01 -9.15745854e-01
-7.84973502e-01 -4.27991271e-01 5.42968869e-01 6.51483357e-01
9.85304534e-01 -1.56426096e+00 -6.86971962e-01 -1.64584324e-01
-1.42683029e-01 -7.45519400e-01 -6.30282521e-01 -3.49477828e-01
-5.96941352e-01 -1.20306551e+00 -5.89284778e-01 -1.24536586e+00
5.07976115e-01 6.49740636e-01 5.48336208e-01 3.90314817e-01
-5.99688649e-01 2.67457306e-01 7.62163922e-02 -5.99141836e-01
2.00040132e-01 8.85810554e-02 4.07027751e-01 3.28871727e-01
1.00806904e+00 -1.98786899e-01 -5.96490681e-01 6.03380680e-01
-8.00614476e-01 -3.49373162e-01 7.99232662e-01 5.89710534e-01
9.06941295e-02 5.69306433e-01 9.30020690e-01 -4.83862907e-01
8.69460940e-01 -1.78388923e-01 -2.33793333e-01 4.57941554e-02
-5.78327000e-01 -1.82180911e-01 1.44193277e-01 -6.30105436e-02
-1.35217667e+00 -5.02574593e-02 -4.65367079e-01 -3.41855019e-01
-5.72960734e-01 4.93601441e-01 -5.05520761e-01 -1.18447118e-01
4.62277770e-01 3.96614760e-01 7.15312958e-01 -3.21451783e-01
-3.67704667e-02 1.08598280e+00 6.95372105e-01 1.27638653e-01
4.02362585e-01 1.26838818e-01 2.56250709e-01 -1.25159323e+00
-4.08554822e-01 -6.60478473e-01 -8.14664423e-01 -8.37388098e-01
9.12925065e-01 -6.67941988e-01 -1.37494695e+00 8.33186269e-01
-9.24250722e-01 1.82823196e-01 4.12059516e-01 8.08420777e-01
-3.99997562e-01 4.57448483e-01 -3.66961569e-01 -7.84425676e-01
-3.70498925e-01 -1.06865168e+00 6.48436785e-01 6.15150630e-01
6.26321957e-02 -8.18652928e-01 -3.58366258e-02 4.31300938e-01
5.24390697e-01 3.50625776e-02 4.10230756e-01 1.97272539e-01
-2.96226174e-01 -3.74786198e-01 -3.55802625e-01 4.52567756e-01
7.88684413e-02 7.95601904e-02 -9.88339305e-01 -1.68419704e-01
3.07377875e-01 -3.01810130e-02 8.96404147e-01 7.26454437e-01
1.15210021e+00 4.79680300e-01 -5.47112763e-01 3.00229400e-01
1.07957196e+00 7.61545122e-01 1.29120100e+00 5.07484600e-02
5.67434013e-01 7.91702390e-01 7.58847058e-01 1.19876362e-01
2.80784100e-01 3.25286597e-01 2.38035038e-01 -5.90053052e-02
-9.70811620e-02 2.06489787e-01 4.30633545e-01 8.77182245e-01
-1.08671464e-01 1.89571574e-01 -7.91589737e-01 3.65925521e-01
-1.76366937e+00 -1.25490999e+00 -3.42898756e-01 1.87528539e+00
1.70259356e-01 2.29075491e-01 3.63922864e-01 6.29491031e-01
9.02730823e-01 -1.31577224e-01 -6.56769454e-01 -4.62586403e-01
1.00438997e-01 -5.68535179e-02 1.46240577e-01 2.16475874e-01
-8.78367543e-01 6.23765409e-01 6.30353928e+00 6.81256056e-01
-1.34010410e+00 -7.45152170e-03 5.03447115e-01 1.40731499e-01
2.16948137e-01 -6.21787786e-01 -9.45576012e-01 7.72571564e-01
1.16868865e+00 -9.16929729e-03 2.41524622e-01 4.74242479e-01
7.20475912e-01 -4.42429304e-01 -4.94468182e-01 1.30288219e+00
3.45880479e-01 -5.85109532e-01 -5.77312469e-01 -8.01342353e-02
-5.82509264e-02 -1.51377425e-01 7.72314519e-02 3.94228756e-01
-5.51092207e-01 -1.23749149e+00 6.01477474e-02 1.01037312e+00
9.39059556e-01 -1.24744213e+00 1.05984545e+00 5.18316209e-01
-1.24559331e+00 -2.50382841e-01 -4.25199658e-01 -3.82500380e-01
3.80783707e-01 4.48020816e-01 -4.69699800e-01 3.54588956e-01
6.62248790e-01 9.53907430e-01 -3.70148361e-01 1.23756027e+00
-1.64329991e-01 6.18374169e-01 -1.48652017e-01 -3.12167317e-01
-1.16458081e-01 -3.58259901e-02 6.95239902e-02 9.78939235e-01
1.47971168e-01 1.55917764e-01 6.61493167e-02 4.10255194e-01
3.54130298e-01 -2.10977033e-01 -7.54083574e-01 1.97574988e-01
1.43582717e-01 1.30391681e+00 -6.09850287e-01 -1.50943873e-03
-3.94137710e-01 5.23974419e-01 -2.21316859e-01 -3.78719941e-02
-8.71092260e-01 -7.82040298e-01 5.25124252e-01 1.13331556e-01
3.99956992e-03 -4.77478236e-01 -3.20676208e-01 -4.82468575e-01
-2.42274448e-01 -3.31170380e-01 2.57927984e-01 -7.62438536e-01
-8.30073297e-01 3.38735372e-01 -2.48149037e-03 -9.78219330e-01
2.38129627e-02 -7.43480504e-01 -1.02554405e+00 1.09349239e+00
-1.57302213e+00 -8.25085223e-01 -6.07263923e-01 8.19504082e-01
5.48727810e-01 -1.70850113e-01 4.89912242e-01 6.68393970e-01
-1.35309660e+00 5.14033973e-01 -1.58407897e-01 1.39802963e-01
5.98304629e-01 -4.17095900e-01 4.16473635e-02 8.00473988e-01
-6.36878490e-01 4.09969777e-01 4.74524140e-01 -5.62208951e-01
-1.30165803e+00 -8.58560741e-01 1.09867990e+00 -1.16710134e-01
8.20701271e-02 3.74903493e-02 -8.53802741e-01 1.16673373e-01
1.43637702e-01 -1.39063656e-01 3.67550015e-01 -8.13627020e-02
4.67796743e-01 -5.65590501e-01 -1.11251128e+00 1.55270055e-01
9.56523478e-01 -5.17348766e-01 -6.34303212e-01 8.11148286e-02
1.00574076e-01 -2.15529233e-01 -5.01876891e-01 4.65424806e-01
7.99214959e-01 -7.60977030e-01 5.35537720e-01 -1.76605403e-01
-1.54009327e-01 -4.41069692e-01 7.35094070e-01 -1.15039754e+00
-8.02967012e-01 -1.87734246e-01 8.48989338e-02 8.60786617e-01
-1.48571298e-01 -7.97068000e-01 8.65070045e-01 5.23215175e-01
-4.02256012e-01 -6.59984529e-01 -1.23164046e+00 -5.84139705e-01
-4.42395121e-01 -2.96821594e-01 4.25674886e-01 1.61463007e-01
-7.05721136e-03 7.82090306e-01 -3.80749941e-01 -1.75692409e-01
2.71575511e-01 -1.72543287e-01 4.55114394e-01 -1.47216153e+00
6.05490744e-01 -5.64964533e-01 -6.20962322e-01 -9.28766251e-01
4.04903531e-01 -3.51675808e-01 2.83357650e-01 -1.48685980e+00
4.85613942e-03 1.83537998e-03 -3.91462475e-01 3.04377854e-01
-1.68191880e-01 2.93021351e-01 -4.40987021e-01 -4.89137433e-02
-5.63895226e-01 7.20015347e-01 1.67611110e+00 8.66754428e-02
4.48979549e-02 3.30725372e-01 -4.89172339e-01 3.04999113e-01
9.62069392e-01 -5.60400039e-02 -4.69146162e-01 -1.80907428e-01
-1.51919812e-01 1.24507491e-02 1.30135581e-01 -1.26136529e+00
4.68889803e-01 3.75398844e-02 6.61919117e-01 -8.05296659e-01
3.42107028e-01 -8.96722436e-01 -3.80828492e-02 1.04696202e+00
2.72892416e-01 1.31521851e-01 4.49978679e-01 2.90378809e-01
-3.17208588e-01 -1.36939242e-01 7.33798623e-01 1.59308299e-01
-1.20763469e+00 5.20211101e-01 -7.99872577e-01 -4.42596704e-01
1.17207932e+00 -4.73158062e-01 -1.47344336e-01 -1.96896955e-01
-4.43582386e-01 2.97161400e-01 -3.13011408e-01 6.74168706e-01
9.65892076e-01 -1.27807808e+00 -5.28988481e-01 7.37930059e-01
-1.03602931e-01 -2.20580548e-01 7.35437214e-01 1.14370859e+00
-2.72008061e-01 8.68271291e-01 -6.41421616e-01 -6.35009766e-01
-1.36447465e+00 6.59021735e-01 4.04641896e-01 3.54235411e-01
-3.99068534e-01 6.50464714e-01 3.02700102e-02 1.29896253e-01
9.82166678e-02 1.28181741e-01 -9.16722596e-01 4.10804637e-02
7.79496014e-01 7.96844542e-01 2.44720310e-01 -1.16939008e+00
-6.06588840e-01 9.26708579e-01 2.09660176e-02 3.08410704e-01
9.42853749e-01 -2.99047500e-01 -6.87636212e-02 2.38422275e-01
1.23021936e+00 -6.54465616e-01 -1.14671004e+00 3.83977778e-02
-6.88765272e-02 -5.10959685e-01 5.24856567e-01 -6.25051796e-01
-1.32887840e+00 1.27172112e+00 1.24530482e+00 -5.52500710e-02
1.29591799e+00 -2.77052909e-01 1.18479383e+00 1.52374148e-01
3.87785494e-01 -1.19111085e+00 5.80792725e-02 7.18689144e-01
7.78428257e-01 -1.30935538e+00 -2.35348180e-01 -1.78525850e-01
-4.43765372e-01 1.21525276e+00 7.64605165e-01 -1.84568197e-01
1.06614506e+00 1.56433433e-01 7.67074674e-02 -3.94946247e-01
-4.46129560e-01 -5.07774591e-01 5.17855465e-01 6.83671772e-01
3.60417366e-01 -3.13652726e-03 -6.10916793e-01 5.22743225e-01
3.55302021e-02 4.82737422e-01 7.43507445e-02 8.34556401e-01
-1.07917953e+00 -8.61050963e-01 -3.56378943e-01 7.57225990e-01
-2.14131609e-01 5.65727949e-01 1.26750723e-01 5.37786007e-01
3.87194514e-01 1.07112551e+00 1.66200683e-01 -7.22602546e-01
6.63136899e-01 1.95982993e-01 5.12741804e-01 -2.19803914e-01
-1.19631551e-01 -1.28523722e-01 -1.36836097e-01 -5.21750748e-01
-5.11686862e-01 -6.49979651e-01 -1.43908620e+00 -6.33800209e-01
-5.25717497e-01 8.80347490e-02 7.06562459e-01 1.31699586e+00
3.25145930e-01 6.12366199e-01 8.64420533e-01 -7.02727497e-01
7.26005957e-02 -1.13597703e+00 -7.62435794e-01 2.34351918e-01
1.16960429e-01 -1.37854576e+00 -2.12768778e-01 -1.66294724e-01] | [8.103251457214355, -0.5561326742172241] |
d360cb0f-7f13-4db3-ab3f-1c842708ad35 | ontotype-ontology-guided-zero-shot-fine | 2305.12307 | null | https://arxiv.org/abs/2305.12307v1 | https://arxiv.org/pdf/2305.12307v1.pdf | OntoType: Ontology-Guided Zero-Shot Fine-Grained Entity Typing with Weak Supervision from Pre-Trained Language Models | Fine-grained entity typing (FET), which assigns entities in text with context-sensitive, fine-grained semantic types, will play an important role in natural language understanding. A supervised FET method, which typically relies on human-annotated corpora for training, is costly and difficult to scale. Recent studies leverage pre-trained language models (PLMs) to generate rich and context-aware weak supervision for FET. However, a PLM may still generate a mixture of rough and fine-grained types, or tokens unsuitable for typing. In this study, we vision that an ontology provides a semantics-rich, hierarchical structure, which will help select the best results generated by multiple PLM models and head words. Specifically, we propose a novel zero-shot, ontology-guided FET method, OntoType, which follows a type ontological structure, from coarse to fine, ensembles multiple PLM prompting results to generate a set of type candidates, and refines its type resolution, under the local context with a natural language inference model. Our experiments on the Ontonotes, FIGER, and NYT datasets using their associated ontological structures demonstrate that our method outperforms the state-of-the-art zero-shot fine-grained entity typing methods. Our error analysis shows that refinement of the existing ontology structures will further improve fine-grained entity typing. | ['Jiawei Han', 'Xuan Wang', 'Minhao Jiang', 'Tanay Komarlu'] | 2023-05-21 | null | null | null | null | ['entity-typing'] | ['natural-language-processing'] | [ 9.78421718e-02 2.67319620e-01 -5.85027874e-01 -5.71263254e-01
-4.70477730e-01 -5.39212644e-01 5.40015221e-01 4.46561038e-01
-5.18326283e-01 9.80418146e-01 3.17766488e-01 -1.18886329e-01
-2.43377298e-01 -1.22916460e+00 -7.14262784e-01 -1.68376446e-01
1.87579691e-01 9.16052163e-01 4.64409709e-01 -5.04088581e-01
2.52758026e-01 -2.98760474e-01 -2.14345694e+00 7.67354846e-01
1.37399852e+00 1.09582996e+00 3.13755035e-01 2.63099909e-01
-8.13453317e-01 7.24204481e-01 -5.14365375e-01 -7.45136082e-01
-2.50062704e-01 3.00469875e-01 -1.25463212e+00 -5.07490039e-01
4.33020204e-01 7.80965909e-02 4.30054367e-01 1.21566939e+00
2.33212724e-01 2.07126528e-01 7.31722832e-01 -1.24117374e+00
-7.87411749e-01 9.94110525e-01 3.41121368e-02 -1.53091326e-01
6.51781678e-01 -2.35484287e-01 1.48974788e+00 -1.06560099e+00
1.05533135e+00 1.57024586e+00 9.07363474e-01 8.74923527e-01
-1.09164608e+00 -6.26663566e-01 1.91607118e-01 3.48553509e-01
-1.55828989e+00 -2.37454563e-01 2.47905329e-01 -4.93960172e-01
1.18432152e+00 2.45079741e-01 2.44806409e-01 1.10443866e+00
-3.85362208e-02 5.69031954e-01 1.02770948e+00 -8.35915387e-01
3.91574353e-01 3.10681581e-01 5.91517985e-01 8.61668944e-01
5.94115794e-01 -1.34208456e-01 -5.33123016e-01 -4.87637222e-01
4.43493634e-01 -2.01203525e-01 1.98540643e-01 2.95877759e-03
-9.73894954e-01 6.55980825e-01 1.18322030e-01 3.25652421e-01
-2.48146728e-01 -3.48188400e-01 3.18714648e-01 3.92837785e-02
4.68466729e-01 8.29533577e-01 -8.77673805e-01 -1.60612792e-01
-5.62078297e-01 4.10557598e-01 9.84060526e-01 1.40894318e+00
1.13990617e+00 -5.32056749e-01 -5.14583051e-01 1.23721790e+00
3.77312005e-01 2.92688310e-01 6.48606896e-01 -6.97269499e-01
4.63763148e-01 1.44434583e+00 2.51629651e-01 -4.28329200e-01
-4.67573136e-01 2.61486005e-02 -6.23926461e-01 -5.46446592e-02
2.18256176e-01 -1.72516584e-01 -9.71942008e-01 1.59765387e+00
5.73028207e-01 3.90286446e-01 2.68285185e-01 6.45707846e-01
1.04693747e+00 1.84159905e-01 5.56470990e-01 2.96032071e-01
2.01833630e+00 -6.22629404e-01 -7.45234013e-01 -2.81438917e-01
9.00884986e-01 -3.57613593e-01 1.36781824e+00 -4.60655130e-02
-5.74549913e-01 -3.51539135e-01 -9.14604247e-01 -2.60829926e-01
-9.44532573e-01 -1.49771973e-01 5.42236924e-01 4.68515456e-01
-3.69567752e-01 6.22303426e-01 -3.18234295e-01 -4.45890933e-01
2.70787090e-01 -3.61372158e-02 -2.09779695e-01 -4.77764234e-02
-2.05440593e+00 8.13617766e-01 8.35093856e-01 -2.25265086e-01
-3.00440311e-01 -1.02197599e+00 -1.05847454e+00 2.04485357e-01
8.05894971e-01 -1.05749595e+00 1.14496088e+00 -2.97425210e-01
-1.13259530e+00 8.30306530e-01 -4.81740713e-01 -4.79128361e-01
-1.86085925e-01 -2.10714281e-01 -5.69318950e-01 -2.44167536e-01
5.90631187e-01 4.21863586e-01 4.21386272e-01 -1.13054812e+00
-1.23344147e+00 -2.25477904e-01 5.44186771e-01 1.73994064e-01
-2.44680703e-01 1.21994086e-01 -1.89337343e-01 -6.76080525e-01
-2.55701095e-01 -6.28867328e-01 -2.31739357e-01 -5.87457061e-01
-2.44542286e-01 -1.05149591e+00 1.63417891e-01 -1.78475708e-01
1.87946439e+00 -1.75644350e+00 5.34357466e-02 1.18061304e-01
1.83255777e-01 2.56325584e-02 2.95223027e-01 2.43252426e-01
3.10482234e-01 6.05825841e-01 -4.01668213e-02 -1.81770623e-01
5.00630736e-01 6.26222014e-01 -2.33721584e-01 -4.78291124e-01
-1.31775942e-02 7.85872996e-01 -1.25855064e+00 -7.59842813e-01
-1.81213737e-01 -1.50181532e-01 -8.57010424e-01 3.73140484e-01
-5.82508564e-01 -2.15176508e-01 -5.85285842e-01 9.10410404e-01
2.57086247e-01 -4.09447849e-01 3.49817246e-01 -5.05746484e-01
-7.27727264e-02 5.14290631e-01 -1.32093441e+00 1.48532915e+00
-5.98298013e-01 -1.99291989e-01 -3.35905969e-01 -4.95787472e-01
7.87938237e-01 3.15202475e-01 8.78914818e-02 -4.90700454e-01
-1.12860672e-01 4.88026410e-01 -3.90554965e-01 -7.97400951e-01
9.98079300e-01 -3.95765215e-01 -6.96375847e-01 1.49901032e-01
3.48111182e-01 2.47763753e-01 5.16956925e-01 1.92916095e-01
9.29949462e-01 2.65791565e-01 4.75535125e-01 -3.79236311e-01
2.61405885e-01 2.47863263e-01 1.05933297e+00 1.00770509e+00
1.63950756e-01 2.10027635e-01 2.68750042e-01 -4.71245736e-01
-8.61882210e-01 -7.52234280e-01 -3.92253250e-01 1.61934042e+00
4.63713259e-01 -9.37974513e-01 -7.59987593e-01 -9.76001620e-01
1.11634426e-01 7.98058867e-01 -5.74369431e-01 6.41861334e-02
-4.99116093e-01 -6.54585600e-01 9.66881633e-01 4.53601241e-01
2.39123002e-01 -1.36059237e+00 -2.21390858e-01 4.63050514e-01
-7.67982304e-01 -1.30754614e+00 -2.52845794e-01 2.35856876e-01
-3.98103446e-01 -1.13309240e+00 -1.31294504e-01 -8.04536343e-01
5.87115169e-01 -3.98127556e-01 1.36131358e+00 1.99739128e-01
1.32469550e-01 -2.54330873e-01 -7.78289139e-01 -4.55472022e-01
-3.27086389e-01 4.80720431e-01 3.10007989e-01 1.14246979e-02
9.12313223e-01 -1.87203586e-01 -8.95486549e-02 3.58979315e-01
-7.55960405e-01 1.18217520e-01 3.66512090e-01 1.09147596e+00
5.72385550e-01 2.77658492e-01 4.82740909e-01 -1.70971119e+00
5.18388569e-01 -6.02985442e-01 -2.62090296e-01 6.63616776e-01
-1.01183903e+00 5.71335375e-01 7.25666821e-01 -3.53327185e-01
-1.60587478e+00 -5.03593981e-01 -2.43381187e-01 -1.73627466e-01
-4.09663081e-01 9.27094519e-01 -3.03935319e-01 3.05530846e-01
9.55285192e-01 -2.01823384e-01 -8.12917590e-01 -7.62235701e-01
3.85767549e-01 9.44132209e-01 5.30947685e-01 -1.53255510e+00
6.64696038e-01 1.21125154e-01 -4.27857339e-01 -3.48392338e-01
-1.37131381e+00 -6.38694882e-01 -7.68738866e-01 2.40321800e-01
7.68473744e-01 -8.85151207e-01 -3.69790256e-01 4.03311402e-01
-1.04140627e+00 -3.77007276e-01 -2.47722894e-01 4.24308926e-02
-2.49540433e-01 1.89931646e-01 -6.02518082e-01 -5.97438753e-01
-4.66036826e-01 -6.37897491e-01 1.51038170e+00 3.51211280e-01
-4.79068220e-01 -1.10539627e+00 -9.85002443e-02 3.33065242e-01
8.20190236e-02 -1.29522651e-01 1.39521706e+00 -7.62806535e-01
-3.97884756e-01 7.83580840e-02 -1.77157626e-01 -1.95682406e-01
1.63215801e-01 -2.69096971e-01 -7.07225561e-01 3.22957307e-01
-6.70682013e-01 -2.17809856e-01 4.64178741e-01 -1.22718759e-01
1.00285339e+00 -7.90143728e-01 -6.18497849e-01 6.01969898e-01
1.38304687e+00 1.22143459e-02 6.03940129e-01 8.04989576e-01
9.59603608e-01 4.48765695e-01 1.18368053e+00 5.26814044e-01
9.30273533e-01 7.40746856e-01 -5.79156987e-02 2.19327003e-01
8.45736712e-02 -5.44720411e-01 -1.56112760e-01 5.05901158e-01
-3.84502888e-01 -1.15699016e-01 -8.47131252e-01 6.73013031e-01
-2.07096791e+00 -1.17200887e+00 3.10443528e-02 1.99593532e+00
1.57278371e+00 2.66732663e-01 -1.12637751e-01 -6.93401024e-02
8.38740289e-01 -2.21252158e-01 -2.53383785e-01 -3.21246803e-01
-5.58161288e-02 4.60547328e-01 4.77048606e-01 6.02554202e-01
-9.90387857e-01 1.45669901e+00 5.19374132e+00 1.11164951e+00
-5.62316775e-01 6.06411062e-02 -1.40138231e-02 4.51959908e-01
-6.12984836e-01 2.43516654e-01 -1.57471430e+00 6.04312778e-01
7.29579866e-01 -4.98440772e-01 3.73597533e-01 9.29352939e-01
-1.82951972e-01 1.40953526e-01 -1.06173670e+00 6.24720097e-01
-2.45501369e-01 -1.56081903e+00 3.50671887e-01 -1.52008072e-01
5.06637394e-01 -2.75601506e-01 -7.03839362e-01 9.56790209e-01
9.56212819e-01 -8.55387211e-01 9.27954018e-01 4.64401335e-01
1.11588025e+00 -3.48473519e-01 9.53466356e-01 4.40396696e-01
-1.76375771e+00 -1.39116183e-01 -4.77298856e-01 -1.33367941e-01
9.94468406e-02 5.64710319e-01 -8.17031026e-01 9.10570383e-01
9.10984457e-01 5.37080646e-01 -4.45899904e-01 6.04817450e-01
-5.06819606e-01 4.18237150e-01 -1.17346652e-01 -2.56436169e-01
-7.94076920e-02 2.32162178e-01 4.56640959e-01 1.50334322e+00
1.72399983e-01 4.40991640e-01 5.03822267e-01 7.28681386e-01
-1.79749846e-01 -4.26681302e-02 -1.36223212e-01 2.35354498e-01
1.14877880e+00 1.33233261e+00 -2.80843943e-01 -1.05769277e+00
-3.86614949e-01 4.56596524e-01 6.88789189e-01 2.70514637e-01
-6.41880929e-01 -6.70131207e-01 8.74614298e-01 1.78537473e-01
3.40892114e-02 2.28815690e-01 -5.31463981e-01 -1.67934775e+00
9.11235251e-03 -8.45142841e-01 1.14216638e+00 -7.18668520e-01
-1.63121855e+00 6.17543519e-01 2.43194491e-01 -1.07164621e+00
-3.49005967e-01 -4.84973371e-01 -1.79912865e-01 8.67003202e-01
-1.48594594e+00 -1.18423867e+00 -4.00121957e-01 4.54850972e-01
7.37564802e-01 1.73974931e-01 1.09482002e+00 5.64695239e-01
-5.28457701e-01 8.92411292e-01 -5.63695490e-01 3.89795870e-01
8.28668296e-01 -1.67351115e+00 4.31333184e-01 7.68665135e-01
-1.68911546e-01 1.10815692e+00 6.63661003e-01 -9.52124894e-01
-9.70701218e-01 -1.50514126e+00 1.50480914e+00 -6.25758529e-01
8.29866171e-01 -3.82492661e-01 -1.27731490e+00 7.22351432e-01
-2.49850079e-01 -4.94717471e-02 6.58918798e-01 8.62336397e-01
-6.86496615e-01 1.25246644e-01 -1.25366259e+00 5.27771771e-01
1.48513448e+00 -5.35584092e-01 -1.40327132e+00 -1.61938339e-01
8.18569601e-01 -5.91533780e-01 -1.25434756e+00 5.57937443e-01
7.92532682e-01 -3.34272116e-01 8.45520437e-01 -7.97895610e-01
3.25931281e-01 -3.89820099e-01 -2.00817510e-01 -1.51605976e+00
-6.44878387e-01 -2.68613577e-01 -2.82715976e-01 1.54142332e+00
6.76031590e-01 -7.07376420e-01 5.38600385e-01 8.57179165e-01
-3.56489360e-01 -5.93952000e-01 -6.49571419e-01 -9.20693338e-01
-1.55226812e-01 -3.05648446e-01 1.15292335e+00 1.22451615e+00
4.81495291e-01 6.09613776e-01 -3.72493528e-02 4.83108848e-01
6.58382833e-01 2.44989797e-01 5.72788954e-01 -1.70104694e+00
-7.85337687e-02 -4.38992739e-01 -2.56920248e-01 -7.50274539e-01
3.73833865e-01 -1.05972719e+00 2.55227238e-01 -1.60363746e+00
5.68385050e-02 -1.19845331e+00 -4.45351988e-01 1.06918001e+00
-6.58639729e-01 -1.18962023e-03 -2.40798891e-01 1.42737016e-01
-7.03523815e-01 2.08452895e-01 7.63113439e-01 -1.86124042e-01
-7.62304664e-02 -3.18992645e-01 -9.03094232e-01 8.66027057e-01
3.03597897e-01 -6.36232793e-01 -6.31773025e-02 -3.22188437e-01
6.26412928e-01 -2.28336900e-01 8.20692554e-02 -6.06308401e-01
3.35384697e-01 -5.74550807e-01 -1.03993766e-01 -2.31794059e-01
6.29252428e-03 -4.47296917e-01 8.18414688e-02 -2.03445330e-02
-3.25794965e-01 -3.92400503e-01 -6.24285191e-02 9.48900133e-02
-1.89589545e-01 -6.43087626e-01 4.84320223e-01 -3.46847981e-01
-1.43850875e+00 4.22476262e-01 -4.04393673e-02 6.66754127e-01
3.49560201e-01 -1.45737201e-01 -5.05461633e-01 6.35920405e-01
-6.69411182e-01 2.96292752e-01 5.52882195e-01 7.08327711e-01
1.28249124e-01 -1.53512800e+00 -5.56975305e-01 1.52015850e-01
7.76255012e-01 3.91086251e-01 1.09724872e-01 2.47133479e-01
8.27004611e-02 1.36107564e-01 -3.19274291e-02 -4.67984736e-01
-1.25647378e+00 4.19069648e-01 1.79961964e-01 -6.31460249e-01
-5.90649426e-01 8.04643750e-01 2.97625065e-01 -6.97529972e-01
1.84599608e-02 -5.94506025e-01 -6.28383875e-01 2.77337641e-01
6.41610444e-01 1.25847831e-01 2.88034886e-01 -3.32681179e-01
-4.35478657e-01 4.90987539e-01 -9.16716680e-02 1.71162531e-01
1.16177619e+00 -2.18204901e-01 -2.23687693e-01 5.30301630e-01
5.19479632e-01 2.09735215e-01 -6.96086109e-01 -6.13679290e-01
6.43973291e-01 -3.30083549e-01 -4.07011509e-01 -9.42848921e-01
-3.23785275e-01 2.59544075e-01 -4.58344072e-02 4.69935805e-01
7.23679066e-01 2.19447419e-01 8.88927758e-01 4.28959906e-01
1.02996886e+00 -1.14356983e+00 -4.47951376e-01 1.03493929e+00
4.41927224e-01 -1.04767823e+00 -3.28728080e-01 -7.68594444e-01
-3.64795178e-01 8.61004412e-01 9.49803352e-01 2.37438515e-01
3.46401483e-01 5.24610817e-01 -6.51144832e-02 -2.57029533e-01
-1.06639516e+00 -5.87928236e-01 4.60774720e-01 5.53705513e-01
4.70973283e-01 3.64663661e-01 -2.55809426e-01 1.26077485e+00
-4.56338584e-01 2.54232168e-01 3.42430681e-01 9.35718417e-01
-6.06613219e-01 -1.42990839e+00 -2.34084561e-01 6.63774550e-01
-3.83292258e-01 -5.94136655e-01 -5.57469241e-02 4.32035089e-01
6.12406909e-01 8.40504527e-01 -1.48825899e-01 -3.13242435e-01
4.90780979e-01 3.53697091e-01 4.08478349e-01 -1.22037864e+00
-6.21954322e-01 -4.77169812e-01 7.83066452e-01 -3.24618191e-01
-2.37469643e-01 -3.65468740e-01 -1.56154895e+00 -2.50417203e-01
-4.70704526e-01 2.75599539e-01 1.51191458e-01 1.48090065e+00
5.79047322e-01 4.95833755e-01 2.36881047e-01 -3.27751189e-01
-1.90690309e-01 -1.06852281e+00 -3.30127180e-01 8.51733983e-01
9.16615408e-03 -1.31115758e+00 -2.99968988e-01 1.52185187e-01] | [9.634610176086426, 8.759969711303711] |
ef49ed16-60a9-49a6-bd58-3fd47fd46457 | image-based-localization-using-lstms-for | 1611.07890 | null | http://arxiv.org/abs/1611.07890v4 | http://arxiv.org/pdf/1611.07890v4.pdf | Image-based localization using LSTMs for structured feature correlation | In this work we propose a new CNN+LSTM architecture for camera pose
regression for indoor and outdoor scenes. CNNs allow us to learn suitable
feature representations for localization that are robust against motion blur
and illumination changes. We make use of LSTM units on the CNN output, which
play the role of a structured dimensionality reduction on the feature vector,
leading to drastic improvements in localization performance. We provide
extensive quantitative comparison of CNN-based and SIFT-based localization
methods, showing the weaknesses and strengths of each. Furthermore, we present
a new large-scale indoor dataset with accurate ground truth from a laser
scanner. Experimental results on both indoor and outdoor public datasets show
our method outperforms existing deep architectures, and can localize images in
hard conditions, e.g., in the presence of mostly textureless surfaces, where
classic SIFT-based methods fail. | ['Sebastian Hilsenbeck', 'Laura Leal-Taixé', 'Torsten Sattler', 'Daniel Cremers', 'Caner Hazirbas', 'Florian Walch'] | 2016-11-23 | image-based-localization-using-lstms-for-1 | http://openaccess.thecvf.com/content_iccv_2017/html/Walch_Image-Based_Localization_Using_ICCV_2017_paper.html | http://openaccess.thecvf.com/content_ICCV_2017/papers/Walch_Image-Based_Localization_Using_ICCV_2017_paper.pdf | iccv-2017-10 | ['image-based-localization'] | ['computer-vision'] | [-1.23762801e-01 -6.42083228e-01 -1.49334863e-01 -4.21167791e-01
-7.22783625e-01 -5.35222232e-01 4.57978606e-01 -3.09795529e-01
-5.69462836e-01 4.47800219e-01 2.33994603e-01 2.65208632e-02
-1.02588572e-01 -6.87719345e-01 -1.06538045e+00 -5.82970023e-01
-3.40310708e-02 -4.13031653e-02 -9.03098658e-02 -3.40776257e-02
1.17603660e-01 5.48130691e-01 -1.40917075e+00 -3.55419442e-02
5.16709447e-01 1.22214103e+00 3.54428679e-01 6.08838439e-01
1.41795173e-01 8.33742678e-01 -4.20276880e-01 1.02371708e-01
3.87349576e-01 1.06479026e-01 -6.32482708e-01 8.49996358e-02
1.28925586e+00 -5.12560129e-01 -8.15593958e-01 1.04057634e+00
4.25163537e-01 2.97657907e-01 2.22723976e-01 -8.05729926e-01
-1.05980122e+00 -1.03666015e-01 -4.10496086e-01 1.19235769e-01
5.42854726e-01 6.13937713e-02 7.58333921e-01 -1.05285752e+00
6.40010357e-01 1.24047959e+00 1.26092386e+00 4.40638810e-01
-9.18632984e-01 -2.34576017e-01 1.84290901e-01 2.40061998e-01
-1.60038269e+00 -6.12420559e-01 5.78175783e-01 -2.35665128e-01
1.23899245e+00 -4.89479536e-03 4.77731556e-01 1.27147460e+00
5.16455173e-01 7.91226208e-01 8.69290352e-01 -2.88621038e-01
2.33116552e-01 -1.76002026e-01 -2.33440876e-01 1.03825569e+00
2.60483474e-01 9.13522989e-02 -7.53917396e-01 1.14537142e-02
1.07479441e+00 4.15559620e-01 -4.25881952e-01 -6.85386002e-01
-1.62725079e+00 6.42712831e-01 1.26568246e+00 1.93937525e-01
-3.13311458e-01 8.12632501e-01 1.65715739e-01 2.52461344e-01
5.08116364e-01 4.38251436e-01 -5.79805791e-01 7.69551797e-03
-1.06344938e+00 -2.24970374e-03 5.87025762e-01 1.22317004e+00
8.91739547e-01 5.06190807e-02 8.24655220e-02 6.39011681e-01
3.58014911e-01 9.25608754e-01 5.68719149e-01 -1.00236022e+00
4.32907999e-01 1.81024969e-01 3.85706812e-01 -1.21827459e+00
-3.30949694e-01 -4.00259435e-01 -8.47528696e-01 -1.60390154e-01
9.69739035e-02 4.86853085e-02 -9.45250034e-01 1.43174815e+00
-1.11249618e-01 4.13835645e-01 -1.79182306e-01 1.07692277e+00
7.73637474e-01 3.61390233e-01 -2.59461850e-01 5.23561835e-01
9.73484814e-01 -1.18039179e+00 -6.78902507e-01 -5.73515832e-01
5.73637724e-01 -7.10429132e-01 9.12652254e-01 1.39241502e-01
-5.63192546e-01 -7.47135043e-01 -1.11939335e+00 -3.05653781e-01
-6.42436385e-01 5.27568400e-01 8.54894400e-01 4.38222319e-01
-1.42928135e+00 7.68194020e-01 -1.15911984e+00 -7.25421250e-01
4.45933491e-01 5.46790600e-01 -6.78651094e-01 -4.72566277e-01
-8.80792797e-01 9.52738762e-01 -2.01009870e-01 5.23526907e-01
-7.26107776e-01 -2.54874319e-01 -1.40478027e+00 -3.90013643e-02
-6.69707917e-03 -8.19462061e-01 1.23034024e+00 -7.08360016e-01
-1.37717998e+00 5.52821219e-01 -4.29632574e-01 -7.49089837e-01
4.37353611e-01 -7.34934032e-01 -1.35859381e-02 -1.18116826e-01
2.42190927e-01 8.16815078e-01 7.68406391e-01 -9.56249356e-01
-4.76784378e-01 -1.89528167e-01 -2.79120281e-02 5.90118021e-02
-3.01913738e-01 -3.34996194e-01 -5.00581741e-01 -3.47765267e-01
3.72225791e-01 -1.02102077e+00 -4.32295054e-01 1.97779939e-01
-3.91305685e-01 2.33431868e-02 9.97500360e-01 -4.17131484e-01
5.34536779e-01 -2.12272763e+00 -6.33091629e-02 -1.43770650e-01
1.20702140e-01 -5.44582959e-03 -2.87991792e-01 2.62539327e-01
3.18663239e-01 -1.18045276e-02 2.60572553e-01 -8.41093004e-01
1.23398781e-01 2.35393167e-01 -4.56681460e-01 1.02899277e+00
2.13048816e-01 1.13638675e+00 -9.24224973e-01 -7.69920126e-02
7.10793018e-01 9.04546380e-01 -3.42544615e-01 4.79637720e-02
2.42871419e-02 4.94844675e-01 -3.32513601e-01 1.04418230e+00
6.59156501e-01 -3.02946478e-01 -3.38110656e-01 -1.72460020e-01
-3.12614501e-01 4.19769824e-01 -1.00792420e+00 2.27485681e+00
-9.48415041e-01 1.22170889e+00 3.24174091e-02 -3.48244071e-01
8.72334123e-01 -1.47828072e-01 2.48226926e-01 -7.06625581e-01
1.38527393e-01 2.39113435e-01 -5.45632780e-01 -2.24597856e-01
7.30813622e-01 5.13036549e-01 -7.15177283e-02 -1.29688904e-02
2.62017190e-01 -1.50246233e-01 -2.14617714e-01 -2.25699171e-01
1.20829165e+00 4.31706697e-01 2.71652732e-02 -2.51817733e-01
3.15726638e-01 -1.23585328e-01 2.71030694e-01 1.04789770e+00
-2.14876086e-01 8.54900777e-01 -1.95604399e-01 -1.10087025e+00
-8.04886878e-01 -9.33434367e-01 -2.24219654e-02 9.11686361e-01
3.98188144e-01 -3.67693722e-01 -5.38713992e-01 -4.96804446e-01
1.87213458e-02 -1.01646257e-03 -7.18308747e-01 -8.76269862e-02
-6.60666168e-01 -4.35098588e-01 4.31048036e-01 7.98369467e-01
9.74206388e-01 -7.55560458e-01 -8.77475858e-01 6.86854795e-02
-1.13360822e-01 -1.48355162e+00 -3.91023934e-01 4.28072810e-01
-7.81876802e-01 -1.00483429e+00 -5.82439125e-01 -9.76680458e-01
5.96836805e-01 7.34594166e-01 9.85051572e-01 7.53070414e-02
-2.89318413e-01 7.27398872e-01 2.61995643e-02 -1.13630474e-01
4.63789642e-01 2.94251829e-01 3.90970469e-01 -2.64878839e-01
4.63078171e-01 -2.13251278e-01 -5.83272040e-01 1.37678534e-01
-3.35460931e-01 -3.69266868e-01 5.18784404e-01 9.32461560e-01
5.78712761e-01 -1.29765883e-01 -2.68721551e-01 -1.99517503e-01
2.83301175e-01 -2.33360291e-01 -9.07898307e-01 8.49014148e-02
-2.53639311e-01 -1.53228045e-02 5.67257285e-01 -2.56918073e-01
-5.57178915e-01 6.00645244e-01 -3.02458871e-02 -5.70214331e-01
-3.17126691e-01 2.30853409e-01 1.62378207e-01 -9.03907597e-01
5.07619321e-01 2.99400002e-01 -3.36933374e-01 -5.59141338e-01
3.21539581e-01 4.71707076e-01 7.44685233e-01 -2.26818621e-01
8.90002191e-01 7.46029556e-01 6.49180561e-02 -1.03453696e+00
-9.17474926e-01 -5.08767188e-01 -1.00985372e+00 1.37547866e-01
6.01338089e-01 -1.32698667e+00 -7.61628270e-01 5.46153784e-01
-1.28432190e+00 -4.66869712e-01 -1.04773745e-01 6.90798581e-01
-6.24766707e-01 1.57036409e-01 -7.72796988e-01 -4.47339058e-01
-1.85396269e-01 -1.24708354e+00 1.63234043e+00 4.52305466e-01
1.04247574e-02 -1.21724260e+00 8.41559693e-02 -9.93161350e-02
7.19209015e-01 2.05156624e-01 1.17012002e-01 -1.25124112e-01
-1.18533957e+00 -4.89965677e-01 -3.13876778e-01 3.07577461e-01
2.13198140e-01 -1.18566088e-01 -1.14744782e+00 -5.76409101e-01
-5.27959540e-02 -3.54565144e-01 1.26920187e+00 6.90198839e-01
1.10607910e+00 -1.37037307e-01 -5.59812903e-01 1.36744082e+00
1.53792369e+00 -4.01025623e-01 4.92880166e-01 6.82515204e-01
9.68018591e-01 1.59652084e-02 5.47154486e-01 1.39213800e-01
6.08073175e-01 6.44045651e-01 5.71721673e-01 -3.86836499e-01
-3.04565132e-01 -2.66563654e-01 4.02994663e-01 4.82797951e-01
8.21288303e-02 -1.60075068e-01 -8.21299314e-01 5.78592360e-01
-1.86913133e+00 -5.04085422e-01 -2.01339498e-02 2.24875879e+00
2.29887709e-01 -1.39130399e-01 -4.91198897e-01 -3.94083172e-01
4.16029274e-01 3.98837894e-01 -2.94906557e-01 -2.30006054e-02
-3.02492648e-01 2.50157446e-01 1.11579382e+00 5.06948829e-01
-1.80628192e+00 1.30831933e+00 7.12386036e+00 1.00912988e-01
-1.40041125e+00 -2.32991539e-02 9.95277315e-02 -1.22880600e-01
2.11651906e-01 -2.23593280e-01 -7.64446855e-01 3.60260636e-01
8.12273383e-01 3.91090900e-01 5.65828383e-01 1.18980086e+00
-5.60023971e-02 -1.80541933e-01 -1.13759410e+00 1.29574287e+00
1.92625612e-01 -1.48100102e+00 -4.13476080e-01 6.73830137e-02
8.92492056e-01 7.94307649e-01 4.34126109e-01 1.85284808e-01
3.29038560e-01 -1.23827493e+00 6.27318025e-01 5.16996622e-01
8.13198507e-01 -5.36003768e-01 1.07248342e+00 8.24723393e-03
-1.32519281e+00 -8.93921405e-02 -7.55108654e-01 -2.36005440e-01
-1.57977074e-01 4.17922348e-01 -7.05817103e-01 2.50532448e-01
1.14453065e+00 1.27566946e+00 -9.11888957e-01 1.23839164e+00
-3.85678202e-01 4.95498553e-02 -4.37449545e-01 -5.69897816e-02
5.78700542e-01 8.96679163e-02 1.70828834e-01 1.27167284e+00
4.63648260e-01 -5.39496183e-01 3.16086173e-01 9.31657851e-01
-1.88768566e-01 -4.37208951e-01 -1.04603612e+00 2.29556203e-01
4.96253341e-01 1.23231447e+00 -4.77073222e-01 -3.08992956e-02
-4.70027596e-01 1.39702260e+00 4.85946566e-01 5.74599743e-01
-7.12163150e-01 -6.32922113e-01 9.74877954e-01 -2.26882756e-01
5.09493291e-01 -8.63102376e-01 -3.61428782e-02 -1.37709880e+00
5.58715388e-02 -3.72031987e-01 -1.74320653e-01 -7.96465814e-01
-1.03702044e+00 5.53697765e-01 -5.28697670e-01 -1.22726059e+00
-1.94717884e-01 -9.23062384e-01 -5.34278333e-01 6.82703972e-01
-1.75358200e+00 -1.37873018e+00 -6.71163917e-01 5.69160342e-01
5.94269395e-01 -4.30906452e-02 8.63166094e-01 2.44014174e-01
-4.52023417e-01 2.85807282e-01 4.82170433e-01 4.22986299e-01
8.11781108e-01 -1.28453088e+00 1.05421722e+00 1.02732599e+00
5.45722961e-01 1.06216109e+00 5.09336412e-01 -3.25765401e-01
-1.89325988e+00 -1.28978562e+00 8.82504642e-01 -7.90497363e-01
6.12010717e-01 -7.05662727e-01 -4.42593336e-01 9.10544813e-01
1.47106126e-01 5.09411156e-01 2.94492036e-01 1.56868190e-01
-4.88165468e-01 -1.11851297e-01 -9.59743738e-01 3.61580670e-01
1.07583940e+00 -8.64317596e-01 -3.12375993e-01 6.07904792e-01
6.96959257e-01 -8.32233191e-01 -5.68416476e-01 3.14552188e-01
5.03885269e-01 -1.01800883e+00 1.29407299e+00 -1.65161595e-01
4.50433716e-02 -4.64537680e-01 -4.10567433e-01 -1.36520362e+00
-5.38029969e-01 -4.52523887e-01 -2.04116076e-01 6.13890409e-01
-1.26525655e-01 -5.57108283e-01 7.05618680e-01 1.74129143e-01
-1.14615791e-01 -5.86927056e-01 -1.05261302e+00 -9.90934134e-01
-3.26993614e-01 -1.89227343e-01 5.56441545e-01 6.70434296e-01
-6.27642214e-01 -2.17372756e-02 -5.30210316e-01 4.75537390e-01
6.62326753e-01 1.64522290e-01 8.24278533e-01 -9.81088877e-01
1.95807219e-01 -1.10268310e-01 -7.89090335e-01 -1.35888100e+00
4.43818569e-01 -4.00833845e-01 4.19136852e-01 -1.77136910e+00
-2.21524477e-01 -1.16138965e-01 -3.32311481e-01 5.87389708e-01
3.62281084e-01 6.57824278e-01 6.50956407e-02 2.21557751e-01
-1.15391624e+00 6.29531443e-01 7.04110384e-01 -4.06368613e-01
-4.07144204e-02 -8.27245489e-02 -2.41651893e-01 8.69858682e-01
5.80644250e-01 -2.85374373e-01 8.33849609e-02 -1.14920723e+00
1.03429593e-01 -6.58827543e-01 7.54108489e-01 -1.50421107e+00
5.13396680e-01 1.08963214e-01 1.05834115e+00 -5.46245992e-01
5.12340248e-01 -8.54440868e-01 -3.62694740e-01 6.91962183e-01
-9.90873203e-02 2.05390722e-01 2.11830720e-01 6.90080881e-01
-3.56700957e-01 3.07726234e-01 5.71200490e-01 -2.65477628e-01
-1.20644331e+00 3.36947769e-01 -1.61162883e-01 -4.45100039e-01
4.60626453e-01 -2.03188062e-01 -3.98054600e-01 -4.50581282e-01
-3.56139451e-01 1.19391002e-01 7.77226150e-01 7.51809537e-01
7.55931556e-01 -1.60884690e+00 -1.77345708e-01 4.98595923e-01
1.53123289e-01 4.40386645e-02 -2.97174659e-02 8.63496482e-01
-7.65225828e-01 8.38840723e-01 -2.10318938e-01 -9.01358008e-01
-9.14087653e-01 4.29347634e-01 5.60885131e-01 2.50318617e-01
-4.50178981e-01 1.15020692e+00 -2.58636083e-02 -6.56889856e-01
4.94062543e-01 -8.98454130e-01 3.19064468e-01 -3.70259434e-01
5.61852753e-01 1.75492540e-01 2.68087864e-01 -7.77090609e-01
-8.24472070e-01 8.66868138e-01 1.99298635e-01 3.02895695e-01
1.35384595e+00 -3.58160824e-01 -1.89927980e-01 2.86659807e-01
1.36963344e+00 -4.11992818e-02 -1.48988163e+00 -3.82136881e-01
-3.90823511e-03 -7.05550849e-01 3.05633426e-01 -5.02551794e-01
-1.01159692e+00 8.72703850e-01 8.68173838e-01 -3.36971432e-01
8.93108666e-01 -1.06400594e-01 9.17196572e-01 9.78046894e-01
6.99676573e-01 -9.05490637e-01 4.74613681e-02 1.03433681e+00
5.94737351e-01 -1.52618444e+00 3.66503485e-02 8.39145109e-02
-9.59486589e-02 1.33336842e+00 6.77960515e-01 -4.19749975e-01
2.98332036e-01 2.01321453e-01 1.37875438e-01 -2.18859375e-01
-4.12890881e-01 -3.87911350e-01 4.81562972e-01 7.35493183e-01
4.00791526e-01 -1.11670844e-01 6.72918439e-01 1.17277339e-01
-2.43238375e-01 -2.12319493e-01 2.90663004e-01 1.03687108e+00
-3.99533302e-01 -6.09318376e-01 -4.79494780e-01 -3.97796221e-02
-7.91374817e-02 -2.24872380e-01 -6.01302624e-01 8.20925295e-01
2.13036910e-01 7.35854208e-01 2.02215075e-01 -5.58404446e-01
2.13248760e-01 -2.71854311e-01 6.65807664e-01 -4.92368639e-01
-3.08737248e-01 -1.08337626e-01 -4.05770749e-01 -1.18762422e+00
-3.32695216e-01 -4.44323719e-01 -8.10451925e-01 -2.24372819e-01
-3.62880856e-01 -3.51584852e-01 9.68591511e-01 8.16945851e-01
5.49387336e-01 4.27399218e-01 5.67620218e-01 -1.34138739e+00
-4.49881196e-01 -8.19189250e-01 -4.28888112e-01 4.88715107e-03
1.07898879e+00 -6.33099258e-01 -3.29197645e-01 -1.02520168e-01] | [7.67680025100708, -2.1088693141937256] |
6bce8df7-97bd-4d2a-89fa-860bacfae146 | mmdenselstm-an-efficient-combination-of | 1805.02410 | null | https://arxiv.org/abs/1805.02410v2 | https://arxiv.org/pdf/1805.02410v2.pdf | MMDenseLSTM: An efficient combination of convolutional and recurrent neural networks for audio source separation | Deep neural networks have become an indispensable technique for audio source separation (ASS). It was recently reported that a variant of CNN architecture called MMDenseNet was successfully employed to solve the ASS problem of estimating source amplitudes, and state-of-the-art results were obtained for DSD100 dataset. To further enhance MMDenseNet, here we propose a novel architecture that integrates long short-term memory (LSTM) in multiple scales with skip connections to efficiently model long-term structures within an audio context. The experimental results show that the proposed method outperforms MMDenseNet, LSTM and a blend of the two networks. The number of parameters and processing time of the proposed model are significantly less than those for simple blending. Furthermore, the proposed method yields better results than those obtained using ideal binary masks for a singing voice separation task. | ['Naoya Takahashi', 'Yuki Mitsufuji', 'Nabarun Goswami'] | 2018-05-07 | null | null | null | null | ['music-source-separation'] | ['music'] | [-3.52651030e-02 -4.92682308e-01 1.96772188e-01 -4.64724377e-02
-9.25064027e-01 -1.62412047e-01 1.73747659e-01 -3.08985084e-01
-4.69322413e-01 6.21207237e-01 1.86073661e-01 -1.20522352e-02
-8.86861756e-02 -3.91723156e-01 -3.79384905e-01 -7.15123653e-01
-1.58295110e-01 -2.02353626e-01 2.46329099e-01 -1.16270073e-01
7.07184151e-02 4.14590389e-01 -1.69475961e+00 3.57130677e-01
8.23825777e-01 1.38323641e+00 3.90915185e-01 8.40117812e-01
-1.72091007e-01 6.97634459e-01 -1.01043439e+00 -1.38141215e-01
-2.48764884e-02 -5.07704377e-01 -6.60436511e-01 -2.98803449e-01
4.26362246e-01 -1.29312098e-01 -4.38893110e-01 1.14585400e+00
1.27483046e+00 3.53551418e-01 2.87465930e-01 -9.41184938e-01
-5.13228059e-01 8.85427535e-01 -4.92109418e-01 7.51691341e-01
1.21162489e-01 -5.92516847e-02 9.86491144e-01 -1.02843750e+00
-1.32844701e-01 1.12401462e+00 9.21590686e-01 2.41007000e-01
-7.97442853e-01 -1.00961494e+00 4.25676778e-02 5.84995806e-01
-1.62428331e+00 -7.57783651e-01 9.46789026e-01 -8.99352506e-02
1.27730572e+00 1.12002701e-01 5.10428071e-01 1.02299905e+00
1.01162225e-01 7.13892817e-01 9.56291616e-01 -3.95822972e-01
6.99896812e-02 -1.19867846e-01 1.37911767e-01 4.14479047e-01
-1.53099954e-01 1.82673801e-02 -8.40080202e-01 -6.69602826e-02
8.05120826e-01 -3.72521520e-01 -3.92333716e-01 5.53310633e-01
-9.91872668e-01 4.61748838e-01 5.90173900e-01 9.78954136e-01
-4.69300419e-01 1.53492033e-01 5.58547258e-01 3.55682909e-01
6.74050927e-01 4.65802729e-01 -2.70435005e-01 -3.10978264e-01
-1.42877269e+00 1.03814213e-03 6.35738969e-01 5.77596545e-01
1.47847489e-01 9.49274600e-01 -1.27697363e-01 1.26531255e+00
-1.94454938e-02 4.28894401e-01 8.93595934e-01 -7.59396195e-01
3.98855567e-01 -6.63637891e-02 -2.86973298e-01 -9.84110415e-01
-3.93951237e-01 -1.07401705e+00 -1.10119820e+00 -1.23524731e-02
1.71574533e-01 -2.87539065e-01 -8.61830354e-01 1.72001779e+00
-6.09336756e-02 8.67904425e-01 1.38416052e-01 9.31900620e-01
1.18497777e+00 1.02859581e+00 -1.70054451e-01 -2.56950557e-01
9.81137991e-01 -1.23806870e+00 -1.09602869e+00 -1.70454502e-01
9.59229246e-02 -1.05421925e+00 8.35659802e-01 7.19718695e-01
-1.43856871e+00 -1.10463011e+00 -1.21707809e+00 3.31616774e-02
-1.94780067e-01 4.05236930e-01 3.68754059e-01 5.86129546e-01
-1.32629359e+00 9.91986573e-01 -6.31763220e-01 -7.59998783e-02
1.56593293e-01 4.85158652e-01 -3.48571874e-02 5.95501840e-01
-1.38436806e+00 5.57738483e-01 2.95818001e-01 5.33468068e-01
-1.14100742e+00 -4.67537910e-01 -6.95408165e-01 4.17864203e-01
1.42084658e-01 -4.31316763e-01 1.35763514e+00 -8.90287459e-01
-1.88104284e+00 3.82753968e-01 -2.83242077e-01 -6.81353748e-01
-1.94944546e-01 -3.90363514e-01 -7.98461378e-01 2.43330032e-01
-2.44184464e-01 4.17948216e-01 1.15297163e+00 -7.71331728e-01
-5.72624683e-01 -7.45905638e-02 -2.87120324e-02 1.00671008e-01
-6.22286141e-01 3.72568905e-01 -4.53783013e-02 -9.66392100e-01
2.09150970e-01 -5.79113781e-01 6.43902943e-02 -4.35726553e-01
-2.72754788e-01 -2.73576200e-01 7.46703565e-01 -9.17025387e-01
1.50243056e+00 -2.33063483e+00 1.92872435e-01 -2.08759815e-01
6.09945543e-02 8.11734378e-01 -4.23149318e-01 1.55968741e-01
-2.00496614e-01 8.34347233e-02 -1.55453682e-01 -6.22415483e-01
-8.27471316e-02 -1.51649714e-01 -4.93678927e-01 8.20300579e-02
1.19967274e-01 5.65792978e-01 -4.63610739e-01 -3.95853043e-01
7.46149197e-02 7.36565650e-01 -2.56483823e-01 4.11874294e-01
1.42042056e-01 2.28209376e-01 2.50667721e-01 4.30041403e-01
8.90755832e-01 1.20001003e-01 -6.07607551e-02 -1.63115114e-01
-2.98360378e-01 7.14000285e-01 -1.26269853e+00 1.94313788e+00
-7.64142036e-01 7.93423951e-01 5.26041508e-01 -8.27873290e-01
1.07179713e+00 9.87576485e-01 1.44783154e-01 -4.72492009e-01
4.11633849e-01 5.66953778e-01 3.08662176e-01 -4.76127893e-01
3.75740081e-01 -3.60700160e-01 3.59775513e-01 1.93276972e-01
6.41242564e-01 -1.00232899e-01 -3.96973565e-02 -3.08242440e-01
8.36517274e-01 -1.71230510e-01 -6.29574880e-02 -9.90447104e-02
7.41996527e-01 -7.54411995e-01 7.06808686e-01 4.76190925e-01
-2.17000052e-01 7.61200666e-01 1.40132889e-01 -5.85474484e-02
-5.74574828e-01 -9.43239272e-01 5.37249967e-02 1.02643156e+00
-1.89972952e-01 -2.51766384e-01 -8.43250453e-01 -8.23149458e-02
-3.14287364e-01 5.21175742e-01 -7.20629320e-02 -1.14874229e-01
-7.49329150e-01 -5.69603920e-01 1.34192121e+00 7.08377242e-01
8.03030729e-01 -1.09254944e+00 -1.10547081e-01 5.13886452e-01
-3.59981269e-01 -1.13933742e+00 -4.53068256e-01 2.92228162e-01
-9.37070787e-01 -6.22675836e-01 -9.68992114e-01 -1.08318162e+00
-1.82283387e-01 3.32797199e-01 7.57575154e-01 -2.32158676e-01
1.05471350e-01 -3.39737773e-01 -2.75338471e-01 -2.87485719e-01
-1.56312332e-01 3.22958261e-01 2.67573088e-01 1.70863971e-01
1.53245762e-01 -1.22774553e+00 -4.95174468e-01 8.39958265e-02
-7.18879402e-01 -2.94627309e-01 5.84437847e-01 6.87471688e-01
3.53623360e-01 4.02943075e-01 1.16633439e+00 -2.20126241e-01
8.47229064e-01 -3.06797951e-01 -3.09514374e-01 -1.46801978e-01
-3.72265786e-01 -1.98104545e-01 8.41451526e-01 -5.12232363e-01
-1.21077549e+00 -3.58406723e-01 -8.00488055e-01 -7.33005702e-01
-1.62388280e-01 4.73678321e-01 -1.56681389e-01 -1.19602211e-01
2.13062063e-01 2.99910158e-01 -2.86493987e-01 -1.07129574e+00
-9.40783769e-02 1.10893381e+00 5.73108852e-01 -2.70277023e-01
3.74576539e-01 5.04781045e-02 -2.65689075e-01 -1.05570710e+00
-7.85207987e-01 -3.27939779e-01 -6.47558510e-01 6.22574873e-02
8.79001021e-01 -1.05532610e+00 -6.25010729e-01 9.93766189e-01
-1.42406106e+00 -1.34641394e-01 -1.98524117e-01 8.51533949e-01
-1.51737437e-01 1.72003105e-01 -9.25257087e-01 -9.00342464e-01
-5.40253043e-01 -1.03888988e+00 6.45399332e-01 3.24903488e-01
-6.54050857e-02 -9.51753020e-01 7.42535368e-02 1.71595842e-01
7.89905727e-01 -1.49459511e-01 7.29544222e-01 -7.73260295e-01
-1.85909897e-01 -2.10140040e-03 -1.15233250e-01 8.66317689e-01
1.21237233e-01 -1.25429571e-01 -1.73997295e+00 -2.23155707e-01
5.13381183e-01 -1.45118177e-01 1.09844542e+00 5.98588526e-01
1.21826541e+00 -1.98880106e-01 1.43732011e-01 7.82403409e-01
1.01870608e+00 4.96063501e-01 5.80671728e-01 -6.67411089e-02
6.47724271e-01 2.26165503e-01 2.82643735e-01 2.32810169e-01
5.77633120e-02 4.78121310e-01 1.87772244e-01 -4.29954737e-01
-5.93281090e-01 -2.07513958e-01 4.07365859e-01 1.66643965e+00
-1.30746379e-01 -3.57784480e-01 -5.30504704e-01 6.56192720e-01
-1.42000127e+00 -1.04784036e+00 -6.15089061e-03 1.93561041e+00
9.13680673e-01 2.75239646e-01 1.07574895e-01 8.17745447e-01
7.96566725e-01 7.44090259e-01 -2.93497175e-01 -6.74819529e-01
-2.34337166e-01 9.45335507e-01 -4.79900390e-02 4.99882132e-01
-1.11080694e+00 7.80231118e-01 6.78833294e+00 1.42122746e+00
-1.54556715e+00 6.06075644e-01 2.15341538e-01 -3.85710686e-01
4.91216742e-02 -3.96448314e-01 -6.81835651e-01 2.42240772e-01
1.49590790e+00 2.56487969e-02 3.33941758e-01 5.60250700e-01
1.74621031e-01 -1.43192215e-02 -8.39781463e-01 1.13997769e+00
1.40463158e-01 -1.02158237e+00 -1.34863973e-01 -3.57706964e-01
6.04186058e-01 -2.05649920e-02 2.97475189e-01 2.56814510e-01
-3.76728982e-01 -1.03507864e+00 7.18384206e-01 2.79783517e-01
8.53409886e-01 -1.01291049e+00 9.03930664e-01 2.41246626e-01
-1.61486602e+00 -2.53403604e-01 -4.04721081e-01 -3.67760241e-01
1.70087963e-01 6.76829875e-01 -7.83072650e-01 4.99603480e-01
7.06378579e-01 6.02542341e-01 -4.48294759e-01 1.19050658e+00
-2.03338936e-01 1.03862083e+00 -1.81376949e-01 1.29984573e-01
3.03977489e-01 7.22113401e-02 8.19058061e-01 1.33916295e+00
5.05360663e-01 -1.90290868e-01 -3.23724300e-01 7.99256265e-01
-1.53402925e-01 -8.03881907e-04 -4.29342896e-01 -3.09519351e-01
4.60518539e-01 1.17980659e+00 -4.90938962e-01 -2.62879968e-01
-2.54688740e-01 8.25588763e-01 -1.17549896e-02 4.77918625e-01
-8.19025397e-01 -9.74502623e-01 6.57282591e-01 -1.18675634e-01
5.44852674e-01 -3.26452792e-01 -2.41150036e-01 -8.23741078e-01
-7.46088326e-02 -8.57058525e-01 1.30017564e-01 -9.16126966e-01
-9.75935936e-01 1.31923687e+00 -2.88162559e-01 -1.28432834e+00
-2.72914410e-01 -5.16839504e-01 -8.53800833e-01 1.05415332e+00
-1.56925523e+00 -8.75719666e-01 -1.05133459e-01 6.85797155e-01
8.10010791e-01 -3.05286348e-01 8.80238295e-01 7.61492729e-01
-7.24074841e-01 6.70553625e-01 6.44514011e-03 4.74220701e-03
6.56238735e-01 -1.08638728e+00 3.19522291e-01 1.05694497e+00
4.48851645e-01 5.48041463e-01 4.47115451e-01 -2.67040908e-01
-7.21310854e-01 -9.92724299e-01 9.66929495e-01 5.05061030e-01
5.27291656e-01 -2.97053069e-01 -9.12316740e-01 2.86407918e-01
8.42743993e-01 -1.84665680e-01 7.48264372e-01 5.75358756e-02
-2.25316539e-01 -5.37551105e-01 -9.12395537e-01 9.58636999e-02
8.71603429e-01 -9.87329960e-01 -7.95053244e-01 -2.38334104e-01
1.01414859e+00 -3.85887891e-01 -6.65368795e-01 4.64350104e-01
3.87188286e-01 -1.31472456e+00 8.85004759e-01 -1.53531522e-01
1.81276396e-01 -3.06929260e-01 -1.93080753e-01 -1.49022555e+00
-2.63769358e-01 -7.74047434e-01 -1.95827767e-01 1.55464876e+00
3.71855021e-01 -5.96052945e-01 2.09160790e-01 -2.87198663e-01
-6.82343245e-01 -7.33032703e-01 -1.20553529e+00 -9.44740951e-01
-7.12966919e-02 -5.79384804e-01 6.73126817e-01 4.98299718e-01
-3.10888946e-01 3.83542329e-01 -4.39973116e-01 2.13756606e-01
3.14871907e-01 3.95610929e-02 1.23982981e-01 -1.08066511e+00
-3.91278505e-01 -3.94859195e-01 -3.06257457e-01 -1.09075284e+00
4.16410029e-01 -7.38287926e-01 5.96767105e-02 -1.38369131e+00
-2.46876433e-01 -1.13666601e-01 -8.54875267e-01 2.09369600e-01
-2.08086008e-03 4.99001533e-01 1.63723350e-01 -1.06951684e-01
-2.65089303e-01 7.38940775e-01 1.08183372e+00 -1.52922913e-01
-3.62126052e-01 3.38058025e-01 -4.35245126e-01 1.03719246e+00
8.58462393e-01 -5.01241207e-01 -4.19535995e-01 -5.88849366e-01
-1.92578852e-01 2.61543542e-01 3.48609388e-01 -1.59702408e+00
3.38422269e-01 4.28261101e-01 2.59667248e-01 -9.90147829e-01
9.14555788e-01 -4.94302183e-01 4.07313071e-02 1.60226360e-01
-4.60863262e-01 -1.11383013e-01 5.14673531e-01 2.11903661e-01
-9.72565413e-01 -2.60347903e-01 8.98278415e-01 6.69176280e-02
-4.26265061e-01 1.48051530e-01 -4.75116849e-01 -6.39322074e-03
4.47378784e-01 9.31023583e-02 -1.59521431e-01 -4.52801168e-01
-8.40046763e-01 -3.20787132e-01 -3.68993759e-01 2.70553470e-01
8.67962122e-01 -1.51566744e+00 -5.34289718e-01 3.41122866e-01
-6.27567232e-01 -8.71224403e-02 5.62633574e-01 1.10519803e+00
-2.55597264e-01 6.60227835e-01 -2.44181544e-01 -3.56317878e-01
-1.30153024e+00 2.50405729e-01 5.19592106e-01 -1.26811892e-01
-3.07959586e-01 1.43918383e+00 1.05135016e-01 -1.81957141e-01
6.27842069e-01 -4.78487879e-01 -3.21774185e-01 2.49618217e-01
6.14369392e-01 7.70256281e-01 3.07681888e-01 -6.86871171e-01
-3.32911134e-01 5.92884421e-01 2.06325382e-01 -3.77612680e-01
1.49404907e+00 -2.18594447e-01 -2.09821627e-01 7.98377514e-01
1.19152963e+00 1.85174808e-01 -8.88008237e-01 -3.66021603e-01
-2.95327246e-01 -3.45080137e-01 4.98326898e-01 -5.52232504e-01
-1.35992539e+00 1.60824180e+00 8.36161852e-01 3.14353287e-01
1.62170923e+00 -4.29144830e-01 1.44908297e+00 1.38075396e-01
3.82958241e-02 -1.03365374e+00 2.42341667e-01 6.05870545e-01
1.07908094e+00 -7.98604786e-01 -5.15917063e-01 -6.53254380e-03
-3.84948820e-01 1.19464409e+00 6.09609663e-01 -2.58092195e-01
7.72699893e-01 4.98678744e-01 1.51141480e-01 1.51085183e-01
-6.49873495e-01 -4.24052685e-01 4.50734764e-01 3.90376389e-01
6.24849260e-01 -2.21901298e-01 -1.17809065e-01 9.62558389e-01
-3.65890712e-01 6.90563535e-03 3.20598572e-01 5.23451984e-01
-4.39677775e-01 -9.05925989e-01 -4.49554086e-01 1.42970920e-01
-9.17673051e-01 -4.28568482e-01 -3.30657959e-01 3.22991788e-01
3.42133313e-01 1.29215944e+00 1.80885255e-01 -8.42501283e-01
2.42365584e-01 3.38136822e-01 3.12927455e-01 -5.88300288e-01
-9.40949738e-01 6.17936850e-01 9.60607678e-02 -3.77748281e-01
-5.71156621e-01 -2.07838759e-01 -1.17220330e+00 -7.62919858e-02
-6.77826345e-01 4.09633845e-01 5.85102081e-01 8.48216474e-01
4.38848048e-01 1.08769631e+00 7.88436353e-01 -1.07604110e+00
-4.96983528e-01 -1.49026847e+00 -9.84984815e-01 -2.88076401e-01
6.75458789e-01 -5.32770634e-01 -4.92472291e-01 -7.72974938e-02] | [15.331668853759766, 5.6016011238098145] |
b2339332-8bd7-4c17-a7e9-59742feb080d | calip-zero-shot-enhancement-of-clip-with | 2209.14169 | null | https://arxiv.org/abs/2209.14169v2 | https://arxiv.org/pdf/2209.14169v2.pdf | CALIP: Zero-Shot Enhancement of CLIP with Parameter-free Attention | Contrastive Language-Image Pre-training (CLIP) has been shown to learn visual representations with great transferability, which achieves promising accuracy for zero-shot classification. To further improve its downstream performance, existing works propose additional learnable modules upon CLIP and fine-tune them by few-shot training sets. However, the resulting extra training cost and data requirement severely hinder the efficiency for model deployment and knowledge transfer. In this paper, we introduce a free-lunch enhancement method, CALIP, to boost CLIP's zero-shot performance via a parameter-free Attention module. Specifically, we guide visual and textual representations to interact with each other and explore cross-modal informative features via attention. As the pre-training has largely reduced the embedding distances between two modalities, we discard all learnable parameters in the attention and bidirectionally update the multi-modal features, enabling the whole process to be parameter-free and training-free. In this way, the images are blended with textual-aware signals and the text representations become visual-guided for better adaptive zero-shot alignment. We evaluate CALIP on various benchmarks of 14 datasets for both 2D image and 3D point cloud few-shot classification, showing consistent zero-shot performance improvement over CLIP. Based on that, we further insert a small number of linear layers in CALIP's attention module and verify our robustness under the few-shot settings, which also achieves leading performance compared to existing methods. Those extensive experiments demonstrate the superiority of our approach for efficient enhancement of CLIP. | ['Bin Cui', 'Xuming He', 'Xupeng Miao', 'Xianzheng Ma', 'Longtian Qiu', 'Renrui Zhang', 'Ziyu Guo'] | 2022-09-28 | null | null | null | null | ['training-free-3d-point-cloud-classification'] | ['computer-vision'] | [ 1.73668209e-02 -1.35983035e-01 -3.80660534e-01 -3.55134785e-01
-9.62126970e-01 -2.74534762e-01 7.98835516e-01 -2.26791918e-01
-3.78902137e-01 2.91599005e-01 3.60822558e-01 8.27999115e-02
-7.34148622e-02 -6.55217707e-01 -8.65309536e-01 -7.28335738e-01
2.03585550e-01 1.44970194e-01 2.01158047e-01 -3.27280939e-01
7.06118941e-02 5.09724133e-02 -1.74932408e+00 2.71305650e-01
7.67838001e-01 1.04552698e+00 3.91684026e-01 3.65634024e-01
-2.11593777e-01 7.90850282e-01 -1.75498098e-01 -3.73345226e-01
3.12198550e-01 -3.40882987e-01 -5.52968860e-01 1.63647011e-01
4.11331385e-01 -4.83051330e-01 -6.17091894e-01 9.00671303e-01
7.61687636e-01 4.79194015e-01 6.60346746e-01 -1.23615134e+00
-1.11217833e+00 4.74487662e-01 -7.72489607e-01 2.01386690e-01
1.36276066e-01 5.45217454e-01 1.01956618e+00 -1.32962549e+00
5.82494736e-01 1.18928349e+00 4.31768030e-01 5.41675627e-01
-1.02910316e+00 -7.98662901e-01 2.19343767e-01 4.97937351e-01
-1.49720287e+00 -6.30914629e-01 1.02055204e+00 -3.78097296e-01
9.40394938e-01 1.85313404e-01 6.69462502e-01 1.28843951e+00
-1.70101911e-01 7.61168718e-01 6.87728167e-01 -3.51976216e-01
9.70556512e-02 2.86219060e-01 1.51985988e-01 7.45862782e-01
-1.64108053e-01 4.60866243e-02 -5.48545480e-01 2.84355938e-01
6.39875710e-01 4.44658101e-01 -3.62638414e-01 -5.81865907e-01
-1.20580006e+00 8.32185447e-01 8.06809306e-01 3.48853618e-01
-2.11739451e-01 1.39939204e-01 5.49095273e-01 2.19048515e-01
3.41192633e-01 2.85460651e-01 -4.57673110e-02 7.20668435e-02
-8.50602329e-01 -1.77303374e-01 1.63409382e-01 1.13665426e+00
9.21181321e-01 1.76260859e-01 -9.68296885e-01 1.11268151e+00
8.57336968e-02 5.27414143e-01 5.69735646e-01 -7.39836633e-01
5.54832280e-01 6.34709835e-01 -1.34480670e-01 -1.03137684e+00
-1.89285502e-01 -6.36951327e-01 -9.93790388e-01 1.05543602e-02
-5.83365001e-02 1.54414654e-01 -1.18865287e+00 1.74052393e+00
2.46041358e-01 4.28567141e-01 -1.85893308e-02 1.28781259e+00
1.09293389e+00 7.45861351e-01 1.77349329e-01 -1.46476552e-01
1.47360671e+00 -1.35483801e+00 -5.52657068e-01 -2.48817965e-01
4.64188904e-01 -5.41426301e-01 1.54955876e+00 -2.28196055e-01
-8.99325848e-01 -7.51101077e-01 -1.20222819e+00 -3.42743576e-01
-2.85397112e-01 -1.01310097e-01 5.97388089e-01 1.68794498e-01
-5.67075491e-01 4.72907513e-01 -6.18269503e-01 -2.08315507e-01
6.59468174e-01 6.94627464e-02 -3.16835046e-01 -4.00120646e-01
-1.30917490e+00 5.89400887e-01 2.91981697e-01 -1.34956002e-01
-9.95662928e-01 -9.69787896e-01 -1.08140755e+00 3.71476412e-01
4.48337555e-01 -7.73446679e-01 8.66427004e-01 -7.84999907e-01
-1.50788116e+00 8.42935979e-01 5.78092486e-02 -2.13686556e-01
4.66037393e-01 -3.39719802e-02 -2.35074639e-01 2.80559629e-01
1.22422941e-01 9.34658825e-01 1.09409273e+00 -1.20641732e+00
-2.96290815e-01 -2.08980024e-01 2.67625511e-01 3.05980742e-01
-8.62157941e-01 -1.15040921e-01 -1.18287480e+00 -6.91833913e-01
-1.82296231e-01 -6.71732485e-01 1.80284604e-02 1.32623270e-01
-1.23369128e-01 -4.61189710e-02 6.13319874e-01 -4.01642174e-01
1.13025463e+00 -2.40320110e+00 1.46598682e-01 -1.56482443e-01
1.37485877e-01 2.25294635e-01 -5.43673813e-01 2.41150200e-01
-5.55881448e-02 -8.16401765e-02 -2.86156505e-01 -5.83530068e-01
1.37502551e-01 2.92488903e-01 -1.85240030e-01 4.42914933e-01
3.31043124e-01 1.37564921e+00 -8.67536128e-01 -6.60855293e-01
5.80106795e-01 6.39763117e-01 -6.46394193e-01 2.95563698e-01
-2.94152349e-02 3.00083607e-01 -3.51181179e-01 6.06988132e-01
6.55100226e-01 -5.51355779e-01 -1.97234735e-01 -3.73474985e-01
1.00929841e-01 -1.29089579e-01 -9.24325705e-01 2.06159616e+00
-6.74057245e-01 4.66936827e-01 -2.32531101e-01 -1.16192830e+00
8.56166482e-01 1.09101474e-01 3.90456706e-01 -1.19077253e+00
2.95179099e-01 -2.88314611e-01 -2.85954386e-01 -6.63908422e-01
2.96348751e-01 -1.09683815e-02 -6.81002438e-02 1.95768073e-01
3.65008742e-01 2.66431481e-01 6.13109060e-02 3.28369498e-01
7.91180372e-01 1.40428647e-01 1.89802468e-01 -1.62511840e-02
4.65313256e-01 -3.06132555e-01 4.62036043e-01 6.62071526e-01
-1.64987430e-01 8.71044636e-01 1.28112227e-01 -2.29566381e-03
-1.06704307e+00 -9.29236352e-01 -1.84079006e-01 1.45458615e+00
5.21441638e-01 -3.90378356e-01 -4.48429823e-01 -6.37537897e-01
-1.44315526e-01 6.81404531e-01 -9.21661854e-01 -4.37543750e-01
-2.50920564e-01 -5.86087227e-01 1.30096763e-01 6.40970886e-01
5.72480857e-01 -8.39093924e-01 -4.80113417e-01 3.19894613e-03
-2.06629232e-01 -1.10579824e+00 -7.05745578e-01 5.98669313e-02
-4.20531631e-01 -8.62204254e-01 -1.03865886e+00 -7.24480510e-01
6.06786072e-01 8.75134528e-01 8.17708313e-01 1.67881474e-01
-3.67151409e-01 5.08476615e-01 -5.70318103e-01 -7.08371177e-02
1.25378221e-01 -9.38115409e-04 -1.83933228e-01 1.74845278e-01
2.41326213e-01 -8.22679460e-01 -8.60449016e-01 3.11379105e-01
-8.57295632e-01 2.04405516e-01 5.91032326e-01 1.14620411e+00
5.11148334e-01 -4.51929778e-01 5.08598328e-01 -5.43857932e-01
2.53419042e-01 -6.79596961e-01 -2.77845323e-01 2.99785942e-01
-3.78955543e-01 1.22297071e-01 6.11264348e-01 -5.51552296e-01
-1.02131581e+00 -4.56957407e-02 -2.13380307e-02 -1.16411233e+00
5.70415519e-02 3.01493138e-01 -2.95821875e-01 -1.21851183e-01
5.82547307e-01 3.67804676e-01 -2.03139558e-01 -4.12233949e-01
6.75496638e-01 6.65884674e-01 6.05240226e-01 -3.75488311e-01
1.02302003e+00 3.99023890e-01 -3.56913298e-01 -7.46170163e-01
-1.02277958e+00 -6.16968572e-01 -4.51258838e-01 -2.04940364e-01
8.60868573e-01 -1.24427474e+00 -5.24121583e-01 1.69013232e-01
-8.57498705e-01 -1.89885333e-01 -2.58681744e-01 4.53370571e-01
-4.42586720e-01 3.56570840e-01 -4.59969044e-01 -5.77224314e-01
-5.92075109e-01 -1.09304953e+00 1.24047518e+00 2.51573592e-01
1.53576374e-01 -7.20718622e-01 -8.18451643e-02 4.00058359e-01
3.92078698e-01 -2.83484720e-02 7.86045313e-01 -4.52935368e-01
-6.50711477e-01 -1.88449211e-03 -6.43757164e-01 1.15172833e-01
-1.31857097e-01 -2.81310201e-01 -1.25160193e+00 -4.26418364e-01
-1.62392527e-01 -5.58201075e-01 1.13412571e+00 1.97472364e-01
1.30294895e+00 -1.94349885e-01 -1.79950997e-01 1.07128668e+00
1.49055731e+00 -2.12140471e-01 6.31988108e-01 4.03346151e-01
8.35989654e-01 3.55846524e-01 7.07304299e-01 4.92934793e-01
3.60020310e-01 8.46642733e-01 5.05914509e-01 -2.51299322e-01
-3.53069395e-01 -3.20284694e-01 2.67345220e-01 8.96079183e-01
-1.64729938e-01 -3.85029837e-02 -6.09503508e-01 6.33684754e-01
-1.90952301e+00 -1.09348369e+00 3.86238605e-01 2.05544806e+00
7.01088250e-01 5.69238439e-02 -2.05035284e-02 -1.52865469e-01
6.15893841e-01 4.75876004e-01 -6.58863246e-01 1.26261696e-01
-4.99556176e-02 3.08009516e-02 2.55569994e-01 2.11635560e-01
-1.09243989e+00 8.97249162e-01 5.04695892e+00 1.02905047e+00
-1.25510955e+00 4.09437209e-01 5.24502039e-01 -5.21500170e-01
-4.72517580e-01 -2.14381203e-01 -6.94467485e-01 6.02115810e-01
6.57641709e-01 -2.65929222e-01 4.74578619e-01 9.64850247e-01
-3.60284746e-02 3.00902039e-01 -9.88610327e-01 1.30144119e+00
3.64303529e-01 -1.56429160e+00 1.43004209e-01 -1.48983091e-01
7.94741154e-01 2.08120793e-01 9.37725231e-02 8.80973220e-01
-7.34779313e-02 -8.72118890e-01 7.35853255e-01 5.43026507e-01
1.03163922e+00 -7.80587971e-01 5.42055368e-01 2.54159868e-01
-1.31862283e+00 -3.57608885e-01 -7.03354239e-01 1.16809145e-01
1.94837719e-01 2.64109552e-01 -3.91656905e-01 7.69724846e-01
7.51505136e-01 8.75299633e-01 -6.69003963e-01 9.68079209e-01
-5.51428571e-02 3.43179852e-01 -7.45713115e-02 1.26609668e-01
3.86608630e-01 -8.31153430e-03 5.35138667e-01 1.22069299e+00
3.14944357e-01 1.88806966e-01 3.08119833e-01 8.88310611e-01
-2.54260600e-01 1.84956640e-01 -5.02699196e-01 4.15601917e-02
5.06617665e-01 1.35292089e+00 -4.66463834e-01 -4.59375560e-01
-6.76157415e-01 1.07726169e+00 5.24334729e-01 4.85100150e-01
-1.22482884e+00 -4.13636446e-01 6.61444902e-01 1.21848984e-02
5.79699039e-01 -6.21581450e-02 -5.82162812e-02 -1.40097034e+00
2.75103990e-02 -5.94920397e-01 4.17474389e-01 -8.84540141e-01
-1.29058290e+00 5.21498322e-01 4.94350456e-02 -1.48929834e+00
-7.11301565e-02 -3.11320096e-01 -7.40184128e-01 5.85520446e-01
-1.67766821e+00 -1.54425371e+00 -6.38121843e-01 8.25365245e-01
7.95503318e-01 -1.68624490e-01 5.99129200e-01 5.27866364e-01
-7.71465778e-01 1.04963779e+00 7.12673739e-02 1.36627391e-01
8.38657141e-01 -8.42092752e-01 1.26155913e-01 7.76480615e-01
1.84395805e-01 5.58851242e-01 4.00994599e-01 -2.71679431e-01
-1.59611046e+00 -1.21283662e+00 1.88939765e-01 -2.30296999e-01
7.44502008e-01 -5.12933493e-01 -1.22525036e+00 4.72566813e-01
2.65042424e-01 3.15391511e-01 5.94802737e-01 -1.71422958e-02
-5.72376609e-01 -2.08439648e-01 -5.76908112e-01 5.30286968e-01
1.30822241e+00 -6.89442337e-01 -6.95295155e-01 1.79160401e-01
1.12022185e+00 -1.47422120e-01 -8.05401862e-01 3.90859574e-01
4.41533357e-01 -7.94098496e-01 1.15641034e+00 -6.78050876e-01
6.14208221e-01 -1.87966794e-01 -3.19378048e-01 -1.09245777e+00
-6.72172248e-01 -3.19028109e-01 -2.60255426e-01 1.46079266e+00
1.70776621e-01 -2.55074382e-01 4.33760673e-01 3.49422216e-01
-3.63585055e-01 -8.26221943e-01 -7.70979822e-01 -6.41701877e-01
-1.50282338e-01 -3.80822599e-01 5.18982768e-01 1.12911308e+00
-2.10844148e-02 7.66317844e-01 -8.10464978e-01 -6.06416492e-03
6.28099382e-01 4.34585810e-01 9.37080860e-01 -8.80014479e-01
-4.10840601e-01 -4.05543804e-01 -4.26060855e-01 -1.06825173e+00
2.33231351e-01 -1.13177061e+00 8.09590518e-03 -1.39202845e+00
6.25607967e-01 -1.94114640e-01 -6.60283446e-01 6.68369949e-01
-4.92478997e-01 4.40254778e-01 4.49327290e-01 2.94984132e-01
-9.05505538e-01 1.20053411e+00 1.35329592e+00 -4.82379168e-01
-1.59634143e-01 -3.70842129e-01 -6.70416951e-01 4.20779824e-01
4.46516812e-01 -2.52077997e-01 -6.01046264e-01 -4.43553150e-01
-1.56752214e-01 -6.89399466e-02 4.93893236e-01 -9.50358689e-01
1.69311911e-01 -1.51786342e-01 4.99882728e-01 -4.14441496e-01
5.23922503e-01 -6.99595451e-01 -2.04451576e-01 1.08984284e-01
-2.20072284e-01 -5.23765147e-01 1.42273560e-01 6.95533633e-01
-2.11710215e-01 -7.17403591e-02 1.02768576e+00 4.28534187e-02
-1.11433661e+00 6.83071375e-01 2.01255396e-01 -1.96596142e-02
1.14616930e+00 -2.24110425e-01 -3.80146354e-01 -2.96782047e-01
-5.99447787e-01 5.20035923e-01 5.67112029e-01 6.51902735e-01
6.56798363e-01 -1.60986221e+00 -5.49393415e-01 3.69354755e-01
5.88579059e-01 -2.20891386e-01 9.37789798e-01 1.07987010e+00
3.24643925e-02 2.63238907e-01 -4.34739530e-01 -7.48816133e-01
-1.03565407e+00 1.09307039e+00 2.16882706e-01 2.48305965e-02
-1.08285093e+00 6.84394836e-01 5.94389260e-01 -2.36816511e-01
3.32163543e-01 3.86483259e-02 -2.39107698e-01 1.37081385e-01
8.30494821e-01 8.74330923e-02 -1.26738533e-01 -4.64428663e-01
-2.45167062e-01 7.53296971e-01 -7.04563260e-02 1.51419550e-01
1.50315487e+00 -4.02045548e-01 3.41323107e-01 4.70350295e-01
1.42322052e+00 -3.23479682e-01 -1.56113291e+00 -3.80632877e-01
-4.85940784e-01 -5.87780297e-01 2.31347486e-01 -4.17025268e-01
-1.18374193e+00 1.25176787e+00 7.58532882e-01 -2.36408919e-01
1.16198802e+00 1.39575124e-01 7.60861516e-01 3.34701806e-01
5.12900390e-02 -9.57540691e-01 3.94716769e-01 4.06996489e-01
9.15675879e-01 -1.54957676e+00 -1.67075142e-01 -1.88037813e-01
-8.79069388e-01 7.43886769e-01 8.29500139e-01 -1.09195395e-03
3.25647891e-01 -1.18030451e-01 -1.56705767e-01 -9.55539048e-02
-8.38979363e-01 -4.43198889e-01 5.69837332e-01 5.65915287e-01
4.90110032e-02 -2.80757219e-01 -2.14700121e-02 8.13769996e-01
1.66414276e-01 1.68630260e-03 1.35108354e-02 5.37469149e-01
-5.25287330e-01 -4.98858005e-01 -1.51347473e-01 2.45559588e-01
6.92178681e-02 -2.15351343e-01 9.99398604e-02 7.70806134e-01
-8.40150844e-03 5.80378473e-01 1.94736078e-01 -4.32249993e-01
3.32780242e-01 1.20795563e-01 3.45956951e-01 -5.11883497e-01
-2.77058572e-01 1.62347645e-01 -2.12748006e-01 -6.75166726e-01
-2.93803096e-01 -1.79598227e-01 -1.06302309e+00 -2.43811890e-01
-2.53672570e-01 -8.61108601e-02 2.80112356e-01 9.20588195e-01
5.73851943e-01 8.85161936e-01 7.88167596e-01 -1.14276755e+00
-5.33832371e-01 -9.97536004e-01 -3.65631402e-01 6.50957346e-01
1.41468495e-01 -1.05498147e+00 -2.37842500e-01 -7.22814128e-02] | [10.192434310913086, 1.7809630632400513] |
56532c8b-1ec3-44e6-ae14-8dd40a328e56 | semantic-table-detection-with-layoutlmv3 | 2211.15504 | null | https://arxiv.org/abs/2211.15504v1 | https://arxiv.org/pdf/2211.15504v1.pdf | Semantic Table Detection with LayoutLMv3 | This paper presents an application of the LayoutLMv3 model for semantic table detection on financial documents from the IIIT-AR-13K dataset. The motivation behind this paper's experiment was that LayoutLMv3's official paper had no results for table detection using semantic information. We concluded that our approach did not improve the model's table detection capabilities, for which we can give several possible reasons. Either the model's weights were unsuitable for our purpose, or we needed to invest more time in optimising the model's hyperparameters. It is also possible that semantic information does not improve a model's table detection accuracy. | ['Phillip Mortimer', 'Niels Victor', 'Ivan Silajev'] | 2022-11-25 | null | null | null | null | ['table-detection'] | ['miscellaneous'] | [-1.24807067e-01 3.19462270e-01 6.11241646e-02 -3.57068181e-01
-8.87766957e-01 -7.86257446e-01 9.09674108e-01 7.06027329e-01
-3.76479536e-01 5.85843086e-01 4.31865215e-01 -8.49379539e-01
-4.21146542e-01 -7.94624567e-01 -4.46585953e-01 -1.33772731e-01
2.01272383e-01 7.37975776e-01 6.78640783e-01 -3.74298811e-01
8.23627174e-01 4.05146927e-01 -1.34563994e+00 7.22264409e-01
2.84788549e-01 9.63139594e-01 -4.28694487e-03 8.85288239e-01
-8.87804687e-01 1.37394345e+00 -9.65867758e-01 -4.52224731e-01
1.70057416e-01 -5.13907552e-01 -1.23714459e+00 2.10531075e-02
3.14981371e-01 -5.37201427e-02 -2.07318276e-01 9.03285861e-01
-2.95597799e-02 1.12544429e-02 8.16000938e-01 -1.22719622e+00
-2.41043448e-01 7.74078131e-01 -1.41039386e-01 4.17210430e-01
4.04594243e-01 -4.30749148e-01 1.17886734e+00 -9.20712769e-01
9.15575981e-01 1.62711775e+00 5.61514676e-01 1.85442775e-01
-9.86463130e-01 -3.02359790e-01 1.89958125e-01 3.11255902e-01
-1.63169694e+00 -6.52038693e-01 5.00815332e-01 -2.99777001e-01
1.32915986e+00 6.34734750e-01 4.37053382e-01 6.17816925e-01
1.60998166e-01 9.44556594e-01 1.03998554e+00 -7.29412675e-01
3.29280049e-01 7.71446586e-01 -3.99987027e-02 5.77094436e-01
6.32110596e-01 -5.19693792e-01 -5.29162288e-01 -2.91187882e-01
7.52393663e-01 -8.71447027e-01 4.13038135e-01 -3.67223799e-01
-1.17007959e+00 1.10291505e+00 1.75036475e-01 4.84945118e-01
3.12701046e-01 -6.32814169e-02 6.80390656e-01 3.45401376e-01
3.34995598e-01 6.98994398e-01 -6.66062415e-01 -4.05534387e-01
-8.10120404e-01 6.40568614e-01 9.90079701e-01 1.29261982e+00
2.28737056e-01 -1.86421290e-01 1.53326765e-01 7.41775334e-01
3.66782993e-01 1.51966259e-01 2.74258375e-01 -8.17863643e-01
1.09874368e+00 9.16565716e-01 3.11520219e-01 -1.20823741e+00
-5.38764358e-01 1.15522526e-01 -8.67282152e-02 3.65627520e-02
8.40363801e-01 2.83020079e-01 -8.47896993e-01 8.11708152e-01
-1.53631464e-01 -8.95368993e-01 4.15885568e-01 7.98948884e-01
1.13354993e+00 5.49898207e-01 -2.70782299e-02 1.38653949e-01
1.63163531e+00 -6.54230833e-01 -1.02997613e+00 -1.79679766e-01
1.36628222e+00 -1.25171089e+00 1.02399147e+00 4.20974612e-01
-9.97261047e-01 -4.10738796e-01 -1.08281291e+00 5.22196814e-02
-1.07801378e+00 1.16375029e-01 1.01654375e+00 9.89266813e-01
-7.99837470e-01 3.02755743e-01 -5.90403378e-01 -7.37220526e-01
3.68056223e-02 1.10461906e-01 -3.65104862e-02 2.09918335e-01
-1.36908126e+00 1.15158701e+00 6.45638466e-01 3.53146642e-02
-7.33851194e-02 2.18600422e-01 -7.34672427e-01 5.29480726e-02
6.46219432e-01 -2.35296682e-01 1.19555795e+00 -1.79137886e-01
-8.94557476e-01 8.17313731e-01 -1.11347903e-02 -4.69477922e-01
5.68549097e-01 6.99375123e-02 -6.43410087e-01 1.88421890e-01
6.39534593e-02 3.86290103e-01 2.06241578e-01 -1.15590167e+00
-6.03894174e-01 -3.51684064e-01 -3.78170088e-02 3.25475663e-01
-7.60332122e-02 4.08658296e-01 -5.62171698e-01 -8.09681475e-01
4.48464930e-01 -7.07293987e-01 3.91942523e-02 -6.43793046e-01
-4.73056227e-01 -2.68181741e-01 5.65464377e-01 -5.37755072e-01
1.53573954e+00 -1.65231788e+00 -4.78086531e-01 4.65594947e-01
-1.88212633e-01 2.39923105e-01 1.97263151e-01 6.86026752e-01
8.85364637e-02 7.15439498e-01 1.36424974e-01 8.35450813e-02
1.45466074e-01 1.41696975e-01 -3.08099926e-01 1.11086294e-01
4.62107100e-02 8.32178056e-01 -4.40276176e-01 -9.77647722e-01
3.85221720e-01 3.57162982e-01 -4.21631545e-01 -3.58014643e-01
-2.65195221e-01 -4.07155931e-01 -5.72228730e-01 7.66232491e-01
2.94736117e-01 -1.65059894e-01 2.99395174e-01 -1.49734924e-02
-1.25899494e-01 7.56691515e-01 -1.71274984e+00 1.24409461e+00
-3.82583737e-02 6.84058785e-01 -4.00830895e-01 -7.38639712e-01
9.57129180e-01 2.84588307e-01 1.69568375e-01 -9.31072772e-01
2.24207237e-01 4.06665117e-01 -2.12374926e-02 -3.33074003e-01
8.16362560e-01 1.11571632e-01 -1.34277955e-01 2.57833689e-01
-5.15389621e-01 -2.49066606e-01 4.76996005e-01 4.30135310e-01
1.09837615e+00 1.87065944e-01 1.05663687e-01 -7.13505268e-01
8.20508778e-01 5.93797088e-01 9.25180763e-02 9.43634689e-01
3.37898135e-02 5.83005011e-01 6.00026011e-01 -5.87967157e-01
-8.76236498e-01 -7.64175415e-01 -3.20822239e-01 6.88967586e-01
-1.47202417e-01 -1.08659804e+00 -8.40427041e-01 -9.75418091e-01
1.09153084e-01 9.14062738e-01 -5.51412523e-01 3.49039704e-01
-4.82536525e-01 -7.73879111e-01 6.48261070e-01 7.25012243e-01
3.34748119e-01 -1.03994453e+00 -5.61853588e-01 3.87725919e-01
-4.63068753e-01 -1.20695531e+00 -2.56697275e-02 5.01524091e-01
-1.16075039e+00 -1.18000400e+00 -7.64554739e-02 -4.80215549e-01
7.37859070e-01 1.78916067e-01 1.16407311e+00 7.19753578e-02
-2.37911940e-01 2.10049242e-01 -5.65135002e-01 -6.44074202e-01
-7.44412422e-01 2.94890463e-01 -2.62635410e-01 -7.95859694e-01
8.04770470e-01 3.93330872e-01 -2.48021469e-01 3.83444309e-01
-8.62608314e-01 -3.67517650e-01 2.51985580e-01 4.96399909e-01
1.10542126e-01 7.10128367e-01 4.05894011e-01 -1.23852360e+00
6.11778498e-01 1.61419809e-01 -8.77393901e-01 3.07710648e-01
-9.33067322e-01 3.80928129e-01 3.71906102e-01 1.98818028e-01
-9.98929083e-01 -1.20837383e-01 -1.66878447e-01 -9.08811390e-03
7.96782412e-03 4.44531977e-01 -3.73281837e-01 2.01446004e-02
5.18626034e-01 -3.36320847e-02 -3.93782586e-01 -6.50386810e-01
9.45943780e-03 6.92655861e-01 5.38498797e-02 -3.38884652e-01
6.02910101e-01 2.30255619e-01 -5.14821447e-02 -6.76271796e-01
-6.71912134e-01 -6.92979753e-01 -5.87328732e-01 -2.70184819e-02
9.57380593e-01 -9.14711356e-01 -8.27197671e-01 1.05151661e-01
-9.04250681e-01 -1.67071924e-01 6.60145506e-02 3.25754941e-01
-5.13560891e-01 1.85194626e-01 -7.71393061e-01 -9.13925290e-01
5.60448319e-02 -1.04138851e+00 8.46033633e-01 -2.01215655e-01
-7.72663653e-01 -1.33818102e+00 -3.82864326e-01 7.29946494e-01
3.36205393e-01 -9.13549960e-02 1.08663273e+00 -8.08823645e-01
-6.26710117e-01 -2.85952300e-01 -3.76757741e-01 5.99339865e-02
3.02426040e-01 6.87626228e-02 -1.05360210e+00 -6.12932704e-02
-9.00670886e-02 -2.61240751e-01 8.26750994e-01 -1.83491968e-02
1.01927602e+00 -5.35216749e-01 -3.69036824e-01 -2.81528961e-02
1.59783268e+00 5.61328232e-01 7.57817328e-01 1.15560281e+00
7.05797791e-01 7.66986907e-01 9.28431988e-01 2.06777051e-01
4.09726679e-01 8.95870447e-01 1.39320284e-01 -1.54088885e-02
9.53473225e-02 -3.58154893e-01 -1.83483437e-02 6.25963628e-01
2.06002817e-01 -6.77189469e-01 -1.04723489e+00 3.75263721e-01
-1.81086135e+00 -6.76733971e-01 -4.76954818e-01 2.07823396e+00
3.14064592e-01 8.78672659e-01 2.50948608e-01 6.04650080e-01
4.28475916e-01 1.03865184e-01 1.36027157e-01 -1.01952541e+00
-1.98550090e-01 -2.49200583e-01 1.03564060e+00 5.99138081e-01
-1.06502116e+00 1.05047929e+00 7.27433443e+00 6.45834982e-01
-3.23954880e-01 -3.58625323e-01 7.27337062e-01 2.02394664e-01
-3.23751122e-01 1.44072622e-01 -1.26405144e+00 3.00272584e-01
1.15255451e+00 -2.86034532e-02 -2.62651239e-02 7.09775865e-01
-5.14744967e-02 -3.17979574e-01 -9.48548198e-01 6.17786765e-01
1.50512293e-01 -1.55743456e+00 2.14246199e-01 1.89302117e-01
-4.86647524e-02 -5.42410553e-01 -2.26998046e-01 4.30813313e-01
3.30254763e-01 -9.09289837e-01 9.13510084e-01 3.63491848e-02
1.94808930e-01 -1.11847615e+00 9.03790772e-01 1.42173097e-01
-1.01268244e+00 2.29911745e-01 -4.61609989e-01 -1.60660401e-01
-2.15658069e-01 6.54806569e-02 -1.36263669e+00 4.78686899e-01
6.54723406e-01 2.06327021e-01 -1.22589743e+00 5.46437800e-01
5.30422814e-02 6.58307493e-01 -2.43231520e-01 -2.66516209e-01
4.74231839e-01 -5.21983840e-02 3.39499325e-01 1.39887190e+00
3.05067264e-02 -1.32986695e-01 3.40915136e-02 3.26616347e-01
3.31339240e-01 3.43139231e-01 -6.60890937e-01 -1.87505662e-01
4.19130564e-01 9.19499397e-01 -1.80291235e+00 -5.24044454e-01
-6.86445177e-01 6.33121610e-01 -1.58044264e-01 -5.82422912e-02
-5.07891238e-01 -6.35300696e-01 4.18171048e-01 2.62268543e-01
4.43251997e-01 2.69597266e-02 -6.91757202e-01 -9.27517414e-01
8.54935404e-03 -8.71327579e-01 8.89984727e-01 -9.34117854e-01
-5.98890841e-01 6.02396667e-01 2.78271496e-01 -7.56360590e-01
-4.67629880e-01 -8.01765084e-01 2.81111356e-02 7.08170950e-01
-9.15430844e-01 -6.62752748e-01 3.62023890e-01 1.99272275e-01
4.55523431e-01 -3.87935996e-01 9.92633224e-01 1.46554366e-01
-1.01103075e-01 6.08054876e-01 -5.11525534e-02 3.56261343e-01
6.55274749e-01 -1.64214742e+00 9.28324282e-01 9.54014480e-01
4.88801777e-01 5.75594127e-01 1.24009287e+00 -6.76590145e-01
-1.22231317e+00 -6.51004732e-01 1.54033613e+00 -1.27042067e+00
7.07036316e-01 -6.50450170e-01 -9.56310630e-01 6.24324918e-01
2.35615782e-02 -4.44616050e-01 3.53233993e-01 2.01663047e-01
-4.45126086e-01 7.74363130e-02 -1.32178497e+00 5.24373651e-01
7.05159724e-01 -3.92952383e-01 -7.12152243e-01 6.19531453e-01
5.15279472e-01 -6.01713240e-01 -8.89668107e-01 2.90576190e-01
3.77569914e-01 -7.24531412e-01 1.03385806e+00 -4.33812320e-01
3.39933097e-01 -4.38519359e-01 -4.36366767e-01 -7.54694164e-01
-3.15508902e-01 -9.77901835e-03 -8.79707932e-02 1.36160493e+00
7.50594139e-01 -4.50518996e-01 1.14399374e+00 6.31681740e-01
2.01160789e-01 -3.48885208e-01 -7.89701164e-01 -9.86857772e-01
1.96673647e-01 -7.20450222e-01 8.70639086e-01 9.12912905e-01
4.62419689e-02 5.80460846e-01 -5.45597449e-02 -9.79424566e-02
2.20351711e-01 -1.65987745e-01 6.36159480e-01 -1.05088091e+00
1.16059735e-01 -3.90541434e-01 -4.63256538e-01 -4.51387286e-01
-1.80382773e-01 -7.06064105e-01 -2.10310012e-01 -1.86394823e+00
6.25923946e-02 -4.37959135e-01 -3.95457774e-01 5.24530351e-01
1.45690918e-01 1.61848783e-01 4.48246628e-01 1.54056191e-01
-5.92524767e-01 -1.40110299e-01 9.84978199e-01 -3.30891088e-02
-2.00637355e-01 -2.78991219e-02 -1.05456614e+00 5.26627839e-01
7.59091794e-01 -5.97774506e-01 -6.64604127e-01 -1.85558155e-01
5.02977192e-01 1.61258101e-01 4.56552953e-02 -8.95091057e-01
1.68516159e-01 -5.45899607e-02 9.37823772e-01 -9.19238746e-01
1.59985915e-01 -9.22814071e-01 -7.17145056e-02 5.81270874e-01
-4.65596855e-01 4.86985415e-01 5.30577838e-01 2.38574579e-01
-2.58385748e-01 -2.74767846e-01 3.97286475e-01 -4.96574521e-01
-7.17819035e-01 -6.00102127e-01 -1.02024364e+00 9.27693024e-02
5.43388128e-01 -4.29456413e-01 -3.11272025e-01 -1.20568812e-01
-3.53437185e-01 1.19930878e-01 6.84986234e-01 8.35162282e-01
3.97337139e-01 -1.09547353e+00 -4.24348474e-01 1.24572895e-01
5.83858967e-01 -4.52498525e-01 -3.05859596e-01 3.18094254e-01
-9.11174893e-01 1.37456393e+00 1.62896186e-01 -4.63607758e-01
-1.50578833e+00 6.33079231e-01 -1.78326592e-02 -3.14268619e-01
-9.04978096e-01 7.35065103e-01 -2.81289786e-01 -3.95565689e-01
5.05179226e-01 -6.68232083e-01 -4.13750798e-01 3.15973014e-01
6.16040945e-01 3.66020203e-01 5.99670768e-01 -5.81215143e-01
-9.33361292e-01 1.63046554e-01 -5.89733303e-01 -3.66466284e-01
9.07312334e-01 -1.47926450e-01 3.50773066e-01 5.49362302e-01
1.03026485e+00 -2.21666023e-02 -4.76761639e-01 -6.46295678e-03
7.12738395e-01 -6.65107727e-01 1.47722840e-01 -1.00991416e+00
-5.65315485e-01 3.98378998e-01 4.00138438e-01 3.76008928e-01
7.93040872e-01 -4.68307734e-02 4.93505239e-01 8.73267651e-01
4.81259793e-01 -1.69769657e+00 -2.40209401e-01 5.94995797e-01
5.83076179e-01 -1.31972229e+00 4.37077254e-01 -5.92577934e-01
-6.93801582e-01 1.39410424e+00 2.48834893e-01 4.80954707e-01
3.33233833e-01 3.50175977e-01 5.15791237e-01 -6.31841481e-01
-9.58079517e-01 -2.36099124e-01 3.02812099e-01 2.97141343e-01
8.28465879e-01 -1.22201085e-01 -3.53030682e-01 9.25635695e-02
-5.79143822e-01 -2.25581482e-01 8.31119776e-01 1.30103397e+00
-6.24259830e-01 -1.25110233e+00 -6.24101579e-01 5.27885437e-01
-6.67092383e-01 -6.07839637e-02 -8.52730930e-01 1.54762673e+00
-1.76828146e-01 1.10061109e+00 4.67062220e-02 -3.13014567e-01
5.66299975e-01 3.25622976e-01 4.13420439e-01 -6.78226948e-01
-5.75342476e-01 2.06539333e-01 6.42754138e-01 -5.52882552e-01
-2.76359439e-01 -9.80104864e-01 -1.12647748e+00 -4.48787898e-01
-1.30860224e-01 6.36694729e-01 7.92145610e-01 8.73287797e-01
-1.58830702e-01 4.61139292e-01 2.03338265e-01 1.99462831e-01
-2.63460159e-01 -7.64037549e-01 -5.92207313e-01 -2.86468733e-02
-2.02799916e-01 -5.91672540e-01 -5.44860005e-01 -3.14087304e-03] | [9.64558219909668, 7.88485860824585] |
29908b22-2702-4c5a-8723-0a08cb91e062 | a-unified-framework-with-meta-dropout-for-few | 2210.06409 | null | https://arxiv.org/abs/2210.06409v1 | https://arxiv.org/pdf/2210.06409v1.pdf | A Unified Framework with Meta-dropout for Few-shot Learning | Conventional training of deep neural networks usually requires a substantial amount of data with expensive human annotations. In this paper, we utilize the idea of meta-learning to explain two very different streams of few-shot learning, i.e., the episodic meta-learning-based and pre-train finetune-based few-shot learning, and form a unified meta-learning framework. In order to improve the generalization power of our framework, we propose a simple yet effective strategy named meta-dropout, which is applied to the transferable knowledge generalized from base categories to novel categories. The proposed strategy can effectively prevent neural units from co-adapting excessively in the meta-training stage. Extensive experiments on the few-shot object detection and few-shot image classification datasets, i.e., Pascal VOC, MS COCO, CUB, and mini-ImageNet, validate the effectiveness of our method. | ['Rui Zhao', 'Xingyu Zeng', 'Shaobo Lin'] | 2022-10-12 | null | null | null | null | ['few-shot-image-classification'] | ['computer-vision'] | [-3.02696805e-02 -1.21033959e-01 -1.44253209e-01 -5.36478639e-01
-4.13540810e-01 1.79020345e-01 4.31708217e-01 -1.79434925e-01
-5.19395351e-01 7.33928502e-01 -1.35062024e-01 1.72445670e-01
1.77066401e-02 -9.38830435e-01 -8.33708584e-01 -6.51458979e-01
4.26875830e-01 -5.74223101e-02 7.07796156e-01 -2.07175940e-01
3.07793673e-02 1.36501377e-03 -1.81540227e+00 3.86229962e-01
9.85929966e-01 9.03506100e-01 4.08304244e-01 3.45881701e-01
-3.25266868e-01 1.11406147e+00 -4.60957646e-01 -3.82212341e-01
-1.43609613e-01 -7.60522783e-01 -6.17145956e-01 1.87731832e-01
1.88853621e-01 -4.66463298e-01 -6.62369967e-01 1.12947834e+00
4.81952548e-01 7.60902226e-01 5.39575517e-01 -1.10994983e+00
-9.67867136e-01 5.02168059e-01 -2.72070289e-01 5.16878188e-01
-3.67558479e-01 3.65454286e-01 4.60114360e-01 -1.13877487e+00
5.36051154e-01 1.23498416e+00 6.49695456e-01 1.08129549e+00
-6.35171950e-01 -6.03524387e-01 2.28056237e-01 6.78234458e-01
-1.24858677e+00 -6.13222361e-01 6.70175731e-01 -2.07588404e-01
9.24763083e-01 -3.01352888e-01 5.96338749e-01 1.15198851e+00
-1.26850307e-02 9.68832672e-01 8.68069530e-01 -3.13930631e-01
6.31662965e-01 2.69609034e-01 5.36335886e-01 7.72516668e-01
3.26373816e-01 8.84238407e-02 -4.00837272e-01 3.54667641e-02
5.79958022e-01 7.61309087e-01 -8.99500847e-02 -3.03628176e-01
-6.36854112e-01 8.60733390e-01 4.35129583e-01 3.30516696e-01
-2.39256650e-01 2.91696966e-01 6.78035378e-01 2.87273407e-01
6.80974901e-01 2.93288566e-02 -2.96441227e-01 1.08669540e-02
-6.42289102e-01 -1.78494245e-01 4.37479645e-01 1.21699095e+00
9.96638179e-01 5.21026552e-01 -3.10645759e-01 1.20547938e+00
6.78473935e-02 1.11865595e-01 1.08976817e+00 -6.62326753e-01
2.86868811e-01 4.15210843e-01 3.64587107e-03 -4.90235358e-01
-2.64194794e-02 -3.75865966e-01 -8.31777632e-01 -1.14099681e-01
-3.91202509e-01 -3.38975728e-01 -1.55430377e+00 1.61261153e+00
2.07795829e-01 9.08601999e-01 3.09974372e-01 6.96727574e-01
1.11002445e+00 8.21206808e-01 3.70289624e-01 -3.87898237e-01
7.85974443e-01 -1.39362371e+00 -7.84887254e-01 -2.84227759e-01
5.97341478e-01 4.67660911e-02 1.18585300e+00 -5.21054752e-02
-1.10221958e+00 -9.27567601e-01 -1.25765312e+00 1.14410751e-01
-6.11258268e-01 -2.87267327e-01 5.94657481e-01 3.71928662e-01
-6.41926467e-01 9.34505939e-01 -9.51418936e-01 -3.85167867e-01
9.32350516e-01 1.18870676e-01 3.80898304e-02 -4.53173190e-01
-1.33158839e+00 7.60368824e-01 9.84500945e-01 -1.86935797e-01
-1.43177545e+00 -6.81197941e-01 -8.58680069e-01 4.19706911e-01
5.14314353e-01 -6.54087305e-01 1.37615824e+00 -1.06532907e+00
-1.53166068e+00 6.57941043e-01 -1.11205682e-01 -5.85899234e-01
1.95010066e-01 -1.73421204e-01 -4.81846631e-01 1.70810595e-01
-9.82076153e-02 6.51087105e-01 1.11584103e+00 -1.04595780e+00
-8.95530105e-01 -2.16437653e-01 -4.67550345e-02 1.55603617e-01
-7.76522279e-01 -2.19899699e-01 -4.75560844e-01 -5.35653770e-01
-1.35291874e-01 -5.49146175e-01 -1.59654304e-01 -3.23574424e-01
1.86681952e-02 -3.47588152e-01 8.61811936e-01 -1.50229245e-01
1.10627282e+00 -2.33925390e+00 9.43441167e-02 -3.96299869e-01
1.16919518e-01 7.80626357e-01 -4.04704660e-01 -8.00894387e-03
7.73776099e-02 3.13542463e-04 -2.06025064e-01 -3.96321803e-01
-2.19918668e-01 5.62220573e-01 -3.41813177e-01 1.29348978e-01
2.23970830e-01 1.03874815e+00 -1.15959549e+00 -3.89577299e-01
3.46814722e-01 3.11333448e-01 -3.38268578e-01 3.44020307e-01
-2.82760531e-01 2.61605568e-02 -3.95359695e-01 7.16569960e-01
5.06998718e-01 -2.44641349e-01 -3.09504449e-01 8.03985521e-02
1.94780752e-02 -5.48034906e-02 -8.33196640e-01 1.86147761e+00
-4.71551448e-01 3.24138343e-01 -5.08336067e-01 -1.21026015e+00
8.43060553e-01 2.56347299e-01 -4.72904561e-04 -7.94972956e-01
3.88608307e-01 1.52337089e-01 -1.07129760e-01 -6.44861639e-01
3.49311829e-01 -3.88839185e-01 2.83872962e-01 2.17609659e-01
8.37450087e-01 3.76990855e-01 1.63171664e-01 -1.92051642e-02
9.63217616e-01 -1.63096897e-02 1.74600124e-01 1.02713965e-01
1.39576375e-01 -6.44964203e-02 9.57560360e-01 9.55686152e-01
-5.07863045e-01 5.59358239e-01 4.69136648e-02 -4.62367833e-01
-9.68781531e-01 -8.81460190e-01 7.76831061e-02 1.40801406e+00
2.98175305e-01 -2.26445049e-01 -7.11363912e-01 -7.52493143e-01
-1.80917308e-01 9.04995561e-01 -7.79604316e-01 -1.03641713e+00
-3.54977936e-01 -9.30876017e-01 1.99383646e-01 8.13618481e-01
8.02521229e-01 -1.31570971e+00 -7.16217637e-01 4.68937576e-01
2.77700096e-01 -7.29692519e-01 -1.74898684e-01 3.77373904e-01
-1.19261920e+00 -9.02041376e-01 -1.04620206e+00 -9.57776785e-01
5.88344753e-01 7.92210102e-01 7.63514340e-01 8.28885064e-02
-3.70512128e-01 2.28901818e-01 -5.62217474e-01 -4.11979765e-01
-4.79037873e-02 3.54545414e-02 1.41359970e-01 6.41987547e-02
6.63848937e-01 -6.66681647e-01 -5.86912930e-01 1.81368411e-01
-8.68633747e-01 -9.69066024e-02 7.39173234e-01 1.23261249e+00
6.29174590e-01 9.55053791e-02 8.87624502e-01 -1.07161796e+00
5.58081746e-01 -8.48305583e-01 -2.46276706e-01 3.56813997e-01
-5.85308492e-01 -2.00806230e-01 8.64390314e-01 -8.65092576e-01
-1.33437943e+00 -2.18037814e-01 8.24257582e-02 -1.20864749e+00
-1.84552610e-01 5.75709045e-01 -1.90675408e-01 -1.23671010e-01
5.63024104e-01 5.48699737e-01 -4.68803406e-01 -6.87477052e-01
5.66129386e-01 5.93742728e-01 6.05836213e-01 -1.36041597e-01
5.84348500e-01 4.92204756e-01 -5.59674561e-01 -7.48239279e-01
-1.20645487e+00 -5.06617665e-01 -4.62176114e-01 -5.11917956e-02
7.52839446e-01 -9.69679058e-01 -2.81349003e-01 7.24192441e-01
-1.05356228e+00 -4.39010382e-01 -5.23867667e-01 5.11074901e-01
-5.65326512e-01 -8.88091773e-02 -7.79231548e-01 -6.58626854e-01
-3.42325985e-01 -7.38829076e-01 3.64698827e-01 7.57428050e-01
4.67458487e-01 -9.45852339e-01 1.89586520e-01 -9.39719677e-02
4.83471990e-01 -5.60460612e-02 7.91369379e-01 -7.01987565e-01
-3.95164669e-01 -1.39158949e-01 -2.95093030e-01 5.01082063e-01
1.18044913e-01 -1.98140055e-01 -1.22260499e+00 -3.60174745e-01
2.27980152e-01 -9.09203410e-01 1.46395183e+00 2.40376756e-01
1.38656378e+00 -2.45857507e-01 -2.73646772e-01 8.43301713e-01
1.58324087e+00 4.26405698e-01 6.34095311e-01 2.26701483e-01
7.38473475e-01 2.30952218e-01 7.88930893e-01 5.44985890e-01
1.44913182e-01 1.26798525e-01 3.57265383e-01 2.53453642e-01
-1.87226579e-01 -2.30025977e-01 6.00766316e-02 1.18204153e+00
-5.70819415e-02 4.68668528e-02 -6.88663065e-01 7.17916310e-01
-2.17168736e+00 -1.25705075e+00 5.67310154e-01 1.97970760e+00
7.73030341e-01 3.82652789e-01 -1.59434229e-01 -2.88295239e-01
9.06076849e-01 3.29207867e-01 -1.00955534e+00 -2.68661588e-01
1.52576640e-01 1.01528645e-01 2.20895186e-01 6.47134036e-02
-1.10952854e+00 1.18174040e+00 5.97369194e+00 1.03296018e+00
-1.12457752e+00 6.58067048e-01 5.37530124e-01 -3.33260417e-01
3.37884165e-02 -1.14314929e-02 -1.06973147e+00 6.21222556e-01
1.10167885e+00 -3.29584211e-01 2.97697425e-01 1.31904089e+00
-1.81447998e-01 2.42337316e-01 -9.17447627e-01 1.10722375e+00
3.65796655e-01 -1.55112195e+00 2.09889621e-01 -5.31506836e-01
9.53140914e-01 1.79299340e-01 1.69010043e-01 1.15480125e+00
2.36711383e-01 -6.57945931e-01 4.68944550e-01 6.02562845e-01
6.97519064e-01 -8.94379497e-01 6.08036041e-01 5.62910736e-01
-1.03223407e+00 -4.03427094e-01 -1.11976314e+00 1.87407993e-02
-6.15062825e-02 2.95675516e-01 -3.13180387e-01 2.85922438e-01
6.94541216e-01 9.40104604e-01 -5.22289217e-01 1.30212665e+00
-1.90703437e-01 7.05688655e-01 1.11394301e-01 -1.65060401e-01
4.55134928e-01 1.43257156e-01 2.66937375e-01 8.69469047e-01
1.44461185e-01 4.91882145e-01 1.25823706e-01 9.50138271e-01
-3.41145694e-01 -2.17044771e-01 -4.54676419e-01 -9.49390978e-02
6.59761369e-01 1.07979107e+00 -5.06513417e-01 -8.99509668e-01
-6.62873089e-01 1.10766721e+00 6.59870625e-01 5.20438015e-01
-8.60917270e-01 -8.67461324e-01 1.37743101e-01 -2.53809661e-01
7.21476495e-01 1.40120268e-01 8.11513811e-02 -1.49933398e+00
-2.86899805e-01 -5.78935027e-01 4.48685944e-01 -8.46966147e-01
-1.22062290e+00 5.15323997e-01 4.15652767e-02 -1.23295355e+00
-1.33522257e-01 -3.19034576e-01 -1.30722904e+00 5.36277711e-01
-1.74819994e+00 -1.04958498e+00 -5.60924768e-01 7.42978156e-01
1.03862512e+00 -3.42073053e-01 5.84837317e-01 4.14647937e-01
-8.94745886e-01 6.68652356e-01 5.08371115e-01 7.92729557e-02
5.41629374e-01 -8.46776962e-01 3.06534857e-01 7.48884141e-01
-4.59529832e-02 7.06632018e-01 4.74258304e-01 -6.04107261e-01
-1.13432932e+00 -1.66455805e+00 2.79944301e-01 4.82719764e-02
4.60914850e-01 -1.56158969e-01 -1.36561036e+00 7.62436271e-01
5.10581471e-02 2.78571665e-01 6.80438936e-01 1.17914654e-01
-3.21371615e-01 -1.70407489e-01 -9.00505722e-01 5.93489230e-01
1.05101740e+00 -5.10893881e-01 -1.24188602e+00 2.95541584e-01
1.11577678e+00 -4.38999608e-02 -5.02939284e-01 4.41538483e-01
2.21094504e-01 -7.55288005e-01 7.77677476e-01 -1.05155694e+00
4.22171831e-01 1.48809284e-01 -9.21135396e-02 -1.58473659e+00
-6.18385971e-01 -3.51625025e-01 -6.80369735e-01 1.24168468e+00
2.56143063e-02 -4.55529928e-01 8.00586045e-01 3.82973284e-01
-5.73342621e-01 -7.51410961e-01 -8.38945329e-01 -1.14539027e+00
3.33562121e-02 -6.62407139e-04 4.91130918e-01 1.12440097e+00
-7.87024200e-02 4.00249898e-01 -5.37945807e-01 -2.36654609e-01
8.18076074e-01 -4.88413535e-02 4.67864156e-01 -1.08919835e+00
-3.87209713e-01 -2.15442300e-01 -3.62760246e-01 -5.87208450e-01
1.91956609e-01 -7.24506855e-01 1.94633275e-01 -1.30192268e+00
5.22466660e-01 -1.73656195e-01 -1.02240491e+00 7.11112380e-01
-4.35925931e-01 6.43641055e-02 1.33944184e-01 3.93848091e-01
-1.24479985e+00 1.11359179e+00 1.02086556e+00 -3.29575449e-01
-3.56243700e-01 -6.61832020e-02 -4.68332052e-01 9.37737107e-01
8.65984797e-01 -7.66324937e-01 -7.06702828e-01 -5.17672360e-01
-3.11368227e-01 -1.77558005e-01 3.28931630e-01 -1.26729739e+00
3.35459888e-01 -2.52560854e-01 4.41607177e-01 -4.55804586e-01
4.04410273e-01 -4.24844325e-01 -4.47580338e-01 7.35269129e-01
-3.73254746e-01 -5.78130603e-01 1.81672513e-01 8.43646526e-01
-2.62228310e-01 -7.30360389e-01 1.08844543e+00 -5.86793840e-01
-1.42130244e+00 8.44416738e-01 -4.21178229e-02 2.23539084e-01
1.13262904e+00 -1.62504643e-01 -5.93021154e-01 7.40381284e-03
-9.33544219e-01 3.60860914e-01 3.17664415e-01 5.29489040e-01
1.02745521e+00 -1.60033786e+00 -3.44360083e-01 5.87058291e-02
3.42964023e-01 -1.47824585e-01 7.88403213e-01 6.00841999e-01
-6.65593445e-02 1.61587521e-01 -4.53236580e-01 -3.25350732e-01
-8.57891262e-01 1.00968158e+00 3.62945229e-01 2.15490237e-01
-8.38466167e-01 9.84443069e-01 7.63672069e-02 -1.27045572e-01
4.50882941e-01 -2.63749827e-02 -2.90317327e-01 6.61373585e-02
9.50061619e-01 5.63815117e-01 -2.30735466e-01 -1.40422910e-01
-1.73164815e-01 1.74447015e-01 -6.02112949e-01 1.88531071e-01
1.43947589e+00 -1.56142920e-01 2.56125063e-01 9.68380213e-01
1.37708688e+00 -9.69620883e-01 -1.48969960e+00 -4.19039965e-01
-1.14206560e-01 -3.90296847e-01 8.65847468e-02 -4.43444431e-01
-1.04043818e+00 1.33871174e+00 8.54726017e-01 -7.91037902e-02
8.79733980e-01 -1.30364284e-01 1.03207695e+00 9.49982822e-01
3.27416956e-01 -1.49499083e+00 4.34780300e-01 5.97203612e-01
3.15650284e-01 -1.38292468e+00 -2.64668584e-01 1.71465352e-01
-5.86461186e-01 7.91212142e-01 1.11253870e+00 -3.70699048e-01
6.11788452e-01 -3.50281060e-01 -2.89137393e-01 -1.04727700e-01
-1.16979373e+00 -3.18534672e-01 -7.45135769e-02 5.37473500e-01
-8.37097317e-02 -2.95505077e-01 -1.22168399e-01 9.48700309e-01
7.33241677e-01 4.75363314e-01 5.57289124e-01 1.30138743e+00
-1.19535840e+00 -4.20700908e-01 1.55315354e-01 6.44756675e-01
-2.68053234e-01 -1.61824211e-01 2.57989317e-02 5.29191017e-01
1.49815530e-01 7.05672204e-01 2.51636617e-02 -4.66647774e-01
3.72910470e-01 3.19560826e-01 3.24891001e-01 -1.03535068e+00
-2.63975710e-01 -1.72822580e-01 -3.50076795e-01 -4.22274828e-01
-3.37797999e-01 -6.00002222e-02 -1.20307434e+00 -1.80422500e-01
-5.78989983e-01 1.47248164e-01 2.79296100e-01 1.16020513e+00
3.75523955e-01 7.85644054e-01 7.31601775e-01 -7.53431737e-01
-9.51061428e-01 -1.13059962e+00 -7.35883474e-01 5.12862265e-01
1.62792698e-01 -8.87058020e-01 -4.41134691e-01 6.10449575e-02] | [9.954940795898438, 3.022618293762207] |
6d7329b6-380a-4c75-9148-d313df417a19 | ftso-effective-nas-via-first-topology-second-1 | 2303.12948 | null | https://arxiv.org/abs/2303.12948v1 | https://arxiv.org/pdf/2303.12948v1.pdf | FTSO: Effective NAS via First Topology Second Operator | Existing one-shot neural architecture search (NAS) methods have to conduct a search over a giant super-net, which leads to the huge computational cost. To reduce such cost, in this paper, we propose a method, called FTSO, to divide the whole architecture search into two sub-steps. Specifically, in the first step, we only search for the topology, and in the second step, we search for the operators. FTSO not only reduces NAS's search time from days to 0.68 seconds, but also significantly improves the found architecture's accuracy. Our extensive experiments on ImageNet show that within 18 seconds, FTSO can achieve a 76.4% testing accuracy, 1.5% higher than the SOTA, PC-DARTS. In addition, FTSO can reach a 97.77% testing accuracy, 0.27% higher than the SOTA, with nearly 100% (99.8%) search time saved, when searching on CIFAR10. | ['Lei Chen', 'Likang Wang'] | 2023-02-28 | ftso-effective-nas-via-first-topology-second | https://openreview.net/forum?id=7Z29QbHxIL | https://openreview.net/pdf?id=7Z29QbHxIL | null | ['architecture-search'] | ['methodology'] | [-2.61410236e-01 -2.14479566e-01 5.20010479e-02 -9.03126299e-02
-6.72511816e-01 -3.77538592e-01 -4.49957736e-02 -3.09754163e-01
-6.29922867e-01 6.27659440e-01 -3.76156509e-01 -3.54056478e-01
-1.19801611e-01 -8.88533652e-01 -4.25109714e-01 -6.98389351e-01
9.31515917e-02 5.59718251e-01 6.25397921e-01 2.43642703e-02
2.88207412e-01 2.00886071e-01 -1.41540802e+00 -4.12512161e-02
8.91439676e-01 1.50948465e+00 5.20521522e-01 3.82128179e-01
-2.54069000e-01 6.53938651e-01 -8.35024774e-01 -2.76430517e-01
2.32783720e-01 -2.72498906e-01 -7.44797587e-01 -4.74595517e-01
2.80740857e-01 -4.47136194e-01 -4.27260727e-01 1.44406748e+00
6.30677104e-01 1.48484454e-01 7.98564404e-02 -1.17844999e+00
-4.13923055e-01 1.06733096e+00 -6.09641612e-01 5.14508367e-01
-2.99662292e-01 8.86719078e-02 1.21210647e+00 -8.98699701e-01
3.58726919e-01 9.36481774e-01 4.68296736e-01 5.64077795e-01
-8.78474474e-01 -1.17559373e+00 2.50050843e-01 2.88845509e-01
-1.82798493e+00 -4.93798941e-01 5.05014360e-01 -9.69542265e-02
1.12222624e+00 1.51977301e-01 7.62357354e-01 5.78795850e-01
-1.41859144e-01 7.91526616e-01 7.31311262e-01 -3.19880784e-01
6.52494669e-01 -3.33698809e-01 3.40650201e-01 7.74895132e-01
3.32169116e-01 -8.03256482e-02 -3.00128222e-01 -4.57679257e-02
7.95142055e-01 1.27653843e-02 -5.53648286e-02 1.66259393e-01
-1.01373041e+00 6.28719926e-01 8.40879083e-01 5.55810809e-01
-3.47240329e-01 3.40520024e-01 2.32771918e-01 2.89870262e-01
1.71702608e-01 5.86441875e-01 -3.60555798e-01 -2.92332321e-01
-9.14443076e-01 7.28266090e-02 6.05869234e-01 8.62588644e-01
5.42470932e-01 3.33904326e-01 6.18051998e-02 9.45616484e-01
1.33766517e-01 3.50029022e-01 6.88270390e-01 -7.98342526e-01
5.46972692e-01 7.63185680e-01 -2.34694377e-01 -8.82996678e-01
-4.38166171e-01 -9.56442177e-01 -9.46047723e-01 -1.50282666e-01
1.54824674e-01 -2.40277782e-01 -1.07151294e+00 1.75369668e+00
3.93469520e-02 1.93506181e-01 1.71100218e-02 8.36171567e-01
7.28883922e-01 6.68029845e-01 -1.11889206e-01 -2.52243131e-01
1.54907537e+00 -1.29335546e+00 -3.97546023e-01 -4.79561001e-01
7.17459559e-01 -4.63198692e-01 1.10371983e+00 4.20742989e-01
-1.32947361e+00 -4.56906557e-01 -1.26445246e+00 4.03364241e-01
-2.13736251e-01 2.90011972e-01 5.45596242e-01 4.63374972e-01
-1.23142910e+00 2.91479200e-01 -8.79160702e-01 -1.28386930e-01
6.67019069e-01 4.36515242e-01 1.76273406e-01 1.78866927e-02
-1.23537397e+00 5.05215943e-01 4.85139668e-01 -1.36427566e-01
-9.39465761e-01 -6.79507673e-01 -2.63443172e-01 8.53262305e-01
6.15896046e-01 -4.17772859e-01 1.37001181e+00 -6.15390480e-01
-1.26553535e+00 4.83250797e-01 -2.31372759e-01 -7.66995013e-01
2.28935897e-01 -1.00206574e-02 -5.70579469e-01 1.49774358e-01
3.05662621e-02 8.07034791e-01 3.53519648e-01 -6.57259345e-01
-8.24577868e-01 -4.68044609e-01 1.94178626e-01 1.02852541e-03
-8.00277054e-01 2.93546878e-02 -9.54559982e-01 -5.66365063e-01
3.75267059e-01 -1.04148829e+00 -3.77629548e-01 -2.95026422e-01
-3.65991384e-01 -4.68979836e-01 3.63200635e-01 -3.79051358e-01
1.86357450e+00 -2.29016185e+00 -1.58892170e-01 3.46839726e-01
4.81908739e-01 4.51029897e-01 -1.36907563e-01 -1.62808314e-01
1.71325266e-01 2.32086465e-01 -8.81690234e-02 -1.18869707e-01
-1.14855967e-01 -7.05008134e-02 -2.07546756e-01 -1.39108911e-01
-2.08379656e-01 9.31205869e-01 -6.25480831e-01 -3.22367311e-01
-3.25089633e-01 4.77701426e-02 -5.83318591e-01 -4.29376662e-01
-8.00286382e-02 -1.55090094e-01 -6.37128711e-01 7.91042507e-01
5.87567329e-01 -7.62905657e-01 2.53562063e-01 -4.07867990e-02
-1.97526887e-01 2.96485364e-01 -1.02310467e+00 1.73717511e+00
-2.45935217e-01 7.09646821e-01 -2.03365535e-01 -9.62533712e-01
1.13122654e+00 9.56607908e-02 4.66762751e-01 -1.21224236e+00
4.20793444e-01 5.43278635e-01 2.49818459e-01 -2.94814259e-02
9.41945910e-02 4.52776372e-01 1.16842560e-01 5.68328619e-01
-9.10259560e-02 6.34142935e-01 3.38681608e-01 1.25531852e-01
1.29109550e+00 -6.59190357e-01 4.54263166e-02 -4.31668282e-01
4.57539827e-01 1.53678626e-01 7.83836901e-01 6.02362037e-01
-2.73795813e-01 3.29026312e-01 4.71755743e-01 -7.68850684e-01
-8.21511567e-01 -7.70420849e-01 6.16329908e-02 8.21323514e-01
2.85091043e-01 -5.18137813e-01 -1.05921042e+00 -4.12519395e-01
-3.62028301e-01 5.66026449e-01 -3.55462968e-01 -3.44213665e-01
-7.08296239e-01 -7.99288809e-01 7.43886232e-01 6.98926628e-01
1.23847580e+00 -8.69123101e-01 -9.68148530e-01 3.10725331e-01
-1.61014199e-01 -9.27984655e-01 -5.74920058e-01 -8.13340675e-03
-9.10781682e-01 -8.22700262e-01 -8.15580249e-01 -1.02498686e+00
5.88142812e-01 3.48114401e-01 1.03265738e+00 4.89772171e-01
-3.57332617e-01 -4.56589192e-01 -1.65021479e-01 -2.81392843e-01
2.64676273e-01 5.23045659e-01 8.40279683e-02 -1.66434944e-01
4.75375175e-01 -6.83041573e-01 -7.01188266e-01 5.41491032e-01
-3.72783422e-01 -2.15942021e-02 7.85827160e-01 7.47672498e-01
8.26476276e-01 3.78706247e-01 7.80757487e-01 -3.32005531e-01
5.23901045e-01 -2.84479111e-01 -9.71288085e-01 6.39618158e-01
-1.16008699e+00 2.76157726e-02 4.70279336e-01 -6.53720737e-01
-5.80702603e-01 1.13529280e-01 -6.31270232e-03 -8.97422373e-01
3.23897868e-01 6.47818208e-01 4.29801643e-02 -1.99673548e-01
6.57516301e-01 3.68422061e-01 -3.30102891e-02 -6.79320991e-01
-3.23351622e-01 3.50128561e-01 4.13500547e-01 -1.36267483e-01
5.42413712e-01 1.01597067e-02 -2.29193836e-01 -4.00846750e-01
-6.95943892e-01 -1.73368439e-01 -1.70759335e-01 8.18889588e-02
6.88188195e-01 -9.33031142e-01 -9.38700795e-01 3.56348693e-01
-9.82327163e-01 -3.75218511e-01 1.25104129e-01 5.65926790e-01
3.02698910e-02 -1.52655214e-01 -5.17708182e-01 -6.86669588e-01
-9.19121563e-01 -1.23050177e+00 3.93008918e-01 4.94317055e-01
6.73001399e-03 -5.40367782e-01 -2.29354575e-01 2.65214413e-01
6.60858691e-01 -3.83872271e-01 8.79344165e-01 -7.94075668e-01
-7.65503705e-01 -5.71540669e-02 -5.62087834e-01 8.34922418e-02
-3.11836064e-01 -2.45544702e-01 -7.33859062e-01 -3.40643674e-01
-1.04926951e-01 -1.74404308e-01 8.30792189e-01 4.37016457e-01
1.44945085e+00 -2.24794433e-01 -5.00559211e-01 7.12273717e-01
1.45960951e+00 9.36095059e-01 6.20863199e-01 5.15103638e-01
1.76006451e-01 -3.94963250e-02 6.67963564e-01 3.97297025e-01
1.55015424e-01 6.30928516e-01 5.24346590e-01 8.42972770e-02
-2.28413582e-01 -2.90516168e-01 8.70243087e-03 9.79288816e-01
-1.32464796e-01 -1.98657691e-01 -1.27819920e+00 4.58436251e-01
-1.82897210e+00 -7.43114889e-01 2.91484833e-01 1.95322394e+00
7.30067968e-01 3.60894233e-01 2.54159980e-02 1.78687205e-03
7.20435143e-01 6.69523552e-02 -1.04051805e+00 -5.63388318e-02
4.51201089e-02 1.78747088e-01 5.26559412e-01 2.38036051e-01
-7.59509385e-01 9.97031808e-01 6.38140440e+00 1.16816461e+00
-1.18911421e+00 1.44749269e-01 7.96351552e-01 -5.97662270e-01
3.92212309e-02 -1.92273170e-01 -9.56209183e-01 7.18679667e-01
9.60426211e-01 -4.33170915e-01 7.65633047e-01 1.23589182e+00
-3.27686548e-01 1.04433164e-01 -7.26947367e-01 1.46385205e+00
2.34596711e-03 -1.61438620e+00 -6.74057379e-02 2.83219486e-01
7.21388936e-01 2.70059943e-01 -9.50951427e-02 4.88405496e-01
2.86134690e-01 -1.01782811e+00 7.03246772e-01 9.50723290e-02
8.42009127e-01 -1.08141303e+00 7.13209569e-01 4.91261363e-01
-1.60625112e+00 -3.69629771e-01 -6.03148341e-01 1.64808512e-01
1.47502273e-02 5.93596280e-01 -5.96146762e-01 1.43203259e-01
1.11416674e+00 3.46332103e-01 -5.71983516e-01 1.38970149e+00
1.99293531e-02 5.26115596e-01 -4.99643773e-01 -5.32429576e-01
4.27263081e-01 -1.08250320e-01 3.30272198e-01 5.29341400e-01
6.36010826e-01 2.48364851e-01 2.16028944e-01 8.33566606e-01
-5.03467262e-01 -7.32765198e-02 -9.82271209e-02 -9.97410566e-02
1.18786550e+00 1.02018428e+00 -8.72661710e-01 -5.85485280e-01
5.92343099e-02 7.71647215e-01 3.85411441e-01 2.98939943e-01
-1.15459180e+00 -8.27894568e-01 4.01097119e-01 -3.10103506e-01
2.23464057e-01 2.79963724e-02 -4.61145818e-01 -9.14413512e-01
2.08946303e-01 -8.26351881e-01 3.56627703e-01 -7.01536357e-01
-7.13809907e-01 1.06110704e+00 -4.43601876e-01 -1.04646766e+00
1.16697066e-01 -2.03503042e-01 -8.02126586e-01 6.67729735e-01
-1.01893091e+00 -5.83322465e-01 -4.30799961e-01 3.98726285e-01
8.92078996e-01 -5.25831699e-01 6.13409340e-01 7.44986534e-01
-1.00767314e+00 1.04067791e+00 -3.53097469e-02 2.24237189e-01
2.39196435e-01 -6.77630126e-01 5.72196543e-01 6.89006507e-01
2.77007967e-01 8.56579721e-01 2.02173620e-01 -3.39407057e-01
-1.17476761e+00 -8.18582773e-01 9.19389844e-01 2.13408917e-01
6.47139370e-01 -3.79544273e-02 -9.26138759e-01 4.57640350e-01
-1.61635339e-01 -1.15988903e-01 3.40846956e-01 2.44216844e-01
-4.81515437e-01 -6.18052423e-01 -9.22967911e-01 7.23324656e-01
1.41351664e+00 -2.44242057e-01 -3.10964942e-01 2.53141314e-01
9.75433350e-01 -2.22056031e-01 -6.80384219e-01 3.44070911e-01
6.74899518e-01 -7.68741012e-01 1.07085836e+00 -3.42141598e-01
9.28760991e-02 -1.68823361e-01 -1.80182502e-01 -1.13915741e+00
-4.76895571e-01 -3.97122979e-01 -2.01954082e-01 9.93486583e-01
8.55663121e-01 -1.06598091e+00 8.73932362e-01 4.99100834e-01
-2.00297058e-01 -1.20814121e+00 -1.15475500e+00 -1.05426466e+00
-1.05947994e-01 -2.67317533e-01 1.06614935e+00 8.31621170e-01
-3.35397094e-01 3.94992292e-01 -4.82003167e-02 1.77935157e-02
5.49249113e-01 1.26907393e-01 2.01204076e-01 -1.45342112e+00
-2.31774077e-01 -9.70830619e-01 -1.53177679e-01 -1.19898510e+00
-2.50572830e-01 -6.47503495e-01 -1.63008362e-01 -1.30046284e+00
2.07739845e-01 -8.42834294e-01 -7.15984643e-01 7.50069857e-01
1.55337930e-01 1.46756425e-01 1.57307476e-01 5.29870689e-01
-7.02063620e-01 3.79754603e-01 1.12433422e+00 -2.22326010e-01
-1.82692856e-01 -7.24826902e-02 -7.71584451e-01 6.19346678e-01
1.16657984e+00 -3.50507826e-01 -6.96978927e-01 -8.83888006e-01
2.74368197e-01 -1.36742696e-01 8.27358291e-02 -1.46172273e+00
7.32867301e-01 -1.77604500e-02 1.47201151e-01 -7.39214361e-01
3.12803835e-01 -5.37838101e-01 1.13302670e-01 9.93448138e-01
-2.89758027e-01 3.63325298e-01 2.09950298e-01 4.54092205e-01
-2.98208326e-01 -3.50893408e-01 6.89211309e-01 -1.21536858e-01
-7.76995957e-01 4.89529043e-01 3.76674384e-02 -8.12116042e-02
1.13137197e+00 -1.59771591e-01 -6.06481194e-01 8.88629109e-02
-5.21211505e-01 6.35779560e-01 1.22900233e-01 1.99996680e-01
6.41841412e-01 -1.31830883e+00 -2.50846684e-01 1.46628097e-01
-1.88316852e-01 4.99936938e-02 4.38766569e-01 9.63595390e-01
-6.46503091e-01 7.82900870e-01 -1.70063544e-02 -4.60897714e-01
-1.31131780e+00 5.17574608e-01 3.08434516e-01 -2.61014134e-01
-7.18524337e-01 1.26272416e+00 1.31872207e-01 6.60198554e-02
5.52650630e-01 2.61290185e-02 -1.63470834e-01 2.45210733e-02
8.80345464e-01 4.57323641e-01 7.73493126e-02 4.15391587e-02
-4.18973804e-01 6.61669791e-01 -3.15875977e-01 -1.10917337e-01
1.18841112e+00 9.31399316e-02 -1.62619159e-01 5.60016930e-03
1.06309903e+00 -2.85858005e-01 -9.34733272e-01 -4.37832654e-01
1.65798768e-01 -2.84256935e-01 4.67260927e-01 -9.84517694e-01
-1.59464848e+00 8.74764442e-01 7.63305068e-01 3.44242632e-01
1.40848601e+00 4.30379137e-02 1.14519656e+00 6.76799953e-01
6.77446783e-01 -1.20145845e+00 1.12335160e-01 6.27141893e-01
6.65532351e-01 -7.71288931e-01 -2.42049605e-01 -1.95987150e-01
-4.86257762e-01 1.00597715e+00 1.01294351e+00 -8.45510662e-02
5.82850516e-01 1.21084221e-01 -2.61995822e-01 -3.74264657e-01
-1.19027829e+00 5.99447973e-02 5.96517138e-02 -3.32930908e-02
-2.26478085e-01 2.03356296e-02 -1.35414466e-01 8.85656774e-01
-3.89917761e-01 -1.09339975e-01 -8.45825151e-02 6.90642238e-01
-9.00516808e-01 -6.25231743e-01 6.09319136e-02 4.35673684e-01
-2.61048764e-01 -3.96043926e-01 -2.78455377e-01 4.96572703e-01
-3.12968753e-02 8.91580343e-01 5.01362860e-01 -8.56209040e-01
2.08254918e-01 9.63809937e-02 -4.20029126e-02 -2.99897850e-01
-3.60649586e-01 1.25023887e-01 2.37320401e-02 -6.46695018e-01
1.78514034e-01 -1.52486563e-01 -1.52523303e+00 -5.77749610e-01
-5.32277822e-01 2.67030150e-01 7.13002145e-01 6.37332559e-01
7.29328871e-01 6.52983189e-01 6.12441063e-01 -2.25328863e-01
-6.61850214e-01 -8.68669331e-01 -3.33670884e-01 -3.76662642e-01
-6.86084032e-02 -7.62479961e-01 -4.13187176e-01 -5.43665946e-01] | [8.555354118347168, 3.141591787338257] |
9197d93b-99cd-49c5-ab1e-e9149061d1bd | pattern-revising-enhanced-simple-question | null | null | https://aclanthology.org/C18-1277 | https://aclanthology.org/C18-1277.pdf | Pattern-revising Enhanced Simple Question Answering over Knowledge Bases | Question Answering over Knowledge Bases (KB-QA), which automatically answer natural language questions based on the facts contained by a knowledge base, is one of the most important natural language processing (NLP) tasks. Simple questions constitute a large part of questions queried on the web, still being a challenge to QA systems. In this work, we propose to conduct pattern extraction and entity linking first, and put forward pattern revising procedure to mitigate the error propagation problem. In order to learn to rank candidate subject-predicate pairs to enable the relevant facts retrieval given a question, we propose to do joint fact selection enhanced by relation detection. Multi-level encodings and multi-dimension information are leveraged to strengthen the whole procedure. The experimental results demonstrate that our approach sets a new record in this task, outperforming the current state-of-the-art by an absolute large margin. | ['Hao liu', 'Jun Zhao', 'Yanchao Hao', 'Shizhu He', 'Kang Liu'] | 2018-08-01 | pattern-revising-enhanced-simple-question-1 | https://aclanthology.org/C18-1277 | https://aclanthology.org/C18-1277.pdf | coling-2018-8 | ['fact-selection'] | ['natural-language-processing'] | [ 7.30939656e-02 4.91236985e-01 -3.00875694e-01 -2.06500009e-01
-1.18596959e+00 -7.23404527e-01 6.92261100e-01 6.21726632e-01
-4.38262761e-01 1.09674001e+00 3.31502676e-01 -4.55686659e-01
-6.40363276e-01 -1.14278591e+00 -7.49636531e-01 -1.35354489e-01
6.88954443e-02 7.62841105e-01 8.67303789e-01 -5.04730284e-01
1.43335417e-01 2.41214588e-01 -1.57971323e+00 6.49881959e-01
1.33051145e+00 1.15216899e+00 -1.03894109e-03 6.07801527e-02
-7.78612554e-01 1.11813426e+00 -1.99048921e-01 -9.01609719e-01
-8.30231905e-02 -8.05702135e-02 -1.44733942e+00 -1.69818506e-01
4.23573673e-01 7.39678815e-02 -2.53389955e-01 1.01141250e+00
1.56236291e-01 2.12639496e-01 3.41536224e-01 -8.69929790e-01
-5.59671640e-01 4.14905131e-01 -1.18136920e-01 4.55495566e-01
9.98688400e-01 -1.41963810e-01 1.67490029e+00 -1.06526566e+00
7.67705500e-01 1.14763522e+00 1.61238849e-01 1.97878718e-01
-6.94147825e-01 -1.09188847e-01 1.83568150e-01 7.73456872e-01
-1.16377926e+00 -3.55394125e-01 7.59047687e-01 -1.38887390e-01
1.00003374e+00 4.53873366e-01 3.84919584e-01 5.10622501e-01
-1.56121388e-01 8.09181094e-01 9.73864496e-01 -6.93578601e-01
1.69799447e-01 2.91093230e-01 5.86673915e-01 9.18449879e-01
5.00432611e-01 -5.57670653e-01 -6.71280801e-01 -5.24639547e-01
2.10921630e-01 -3.93189877e-01 -4.49401647e-01 -2.06184387e-01
-9.87805068e-01 8.79897833e-01 2.74885714e-01 3.20550174e-01
-8.53698909e-01 -4.32609439e-01 -1.14216059e-02 3.77015620e-01
3.44002619e-02 8.38367999e-01 -7.57965088e-01 9.90205351e-03
-5.34447372e-01 6.32137597e-01 1.26469183e+00 8.55754137e-01
9.69486535e-01 -8.91567469e-01 -5.43160617e-01 8.03259671e-01
1.80528507e-01 1.08453207e-01 3.05550933e-01 -7.75253713e-01
9.29814160e-01 1.40813529e+00 3.08129042e-01 -1.02919018e+00
-3.42019200e-01 -2.60275990e-01 -3.09058160e-01 -6.31768584e-01
5.49032450e-01 2.17855014e-02 -6.23324513e-01 1.38317013e+00
8.61795366e-01 -6.18086755e-02 2.47273535e-01 6.63675547e-01
9.12194788e-01 6.62606120e-01 8.65019783e-02 -2.90846109e-01
2.12664557e+00 -8.35065901e-01 -8.20362806e-01 -2.84867167e-01
5.07880807e-01 -6.41602159e-01 1.08152342e+00 1.26223907e-01
-9.42751229e-01 -1.87168166e-01 -7.77820706e-01 -4.26110625e-01
-4.43003178e-01 -5.99880628e-02 6.76858723e-01 3.66467834e-01
-4.97761548e-01 2.20275030e-01 -3.16966951e-01 -2.88888246e-01
3.68548274e-01 2.23795757e-01 -4.81876045e-01 -5.45782506e-01
-1.71462584e+00 1.08246195e+00 7.55186439e-01 -1.65001661e-01
-2.45009497e-01 -8.22779298e-01 -7.35246539e-01 4.21084613e-01
1.11725199e+00 -1.00746858e+00 1.09803736e+00 -9.05001089e-02
-1.26348805e+00 6.50404692e-01 -4.15710717e-01 -7.29750752e-01
-8.99974331e-02 -5.19751430e-01 -6.83229744e-01 5.55436730e-01
1.26344085e-01 3.27087313e-01 3.73985589e-01 -1.09471357e+00
-8.87159228e-01 -3.05825353e-01 5.15721738e-01 3.27879965e-01
-2.77298301e-01 1.25472337e-01 -7.35019445e-01 -1.36960819e-01
2.35311151e-01 -4.59347934e-01 -9.48732272e-02 -3.92094374e-01
-3.14195782e-01 -8.63219559e-01 4.64763820e-01 -9.67000425e-01
1.51981378e+00 -1.61925590e+00 -1.07489847e-01 4.31004912e-01
1.07763484e-01 4.15819705e-01 7.76228905e-02 6.80540442e-01
3.38962406e-01 -1.69625908e-01 -2.52002388e-01 3.48800570e-01
2.35921234e-01 3.12765539e-01 -6.40960753e-01 -1.01661511e-01
5.71990430e-01 1.13195086e+00 -9.29107130e-01 -7.66822517e-01
-3.22880745e-01 -1.01835027e-01 -5.78195989e-01 2.42053196e-01
-8.35584581e-01 1.85157984e-01 -8.14520538e-01 7.55896807e-01
3.89256746e-01 -4.56936002e-01 8.15315545e-02 -2.42197424e-01
2.55964667e-01 8.95549238e-01 -1.25276566e+00 1.45673728e+00
-2.31791362e-01 -9.78005901e-02 -3.18474084e-01 -1.00114989e+00
8.97282779e-01 4.11426991e-01 2.34899834e-01 -8.09987068e-01
-1.65577188e-01 3.79912466e-01 4.48151007e-02 -1.05404055e+00
4.96889353e-01 -1.23014897e-01 -6.97009265e-02 1.16562389e-01
2.52045691e-01 4.17795144e-02 6.20665252e-01 3.30780119e-01
1.29859030e+00 -1.40583351e-01 7.08742797e-01 -2.00626090e-01
1.05972576e+00 3.71302336e-01 5.31228602e-01 6.87138557e-01
1.98826939e-01 1.14207186e-01 5.55503607e-01 -3.36433649e-01
-5.31447053e-01 -7.31483996e-01 -3.07364967e-02 7.45815754e-01
3.48036513e-02 -3.85912538e-01 -4.73147899e-01 -1.04339123e+00
1.62918881e-01 7.29432344e-01 -3.36824030e-01 -1.05370265e-02
-6.85847521e-01 -5.74692905e-01 3.38595331e-01 1.03838257e-01
5.20432830e-01 -1.19683361e+00 -3.11061591e-01 3.63081098e-01
-6.14977419e-01 -1.38492644e+00 4.06666845e-03 -4.97411788e-02
-6.95728481e-01 -1.51448989e+00 -3.70116085e-01 -8.52642536e-01
3.90433967e-01 -6.42520338e-02 1.42115176e+00 -4.16194722e-02
2.08741613e-02 4.14266616e-01 -5.86145759e-01 -2.61807799e-01
-1.10103525e-01 3.50450456e-01 -2.63012528e-01 8.32018405e-02
6.89894855e-01 -6.14919722e-01 -6.76843464e-01 1.69957995e-01
-1.00605667e+00 -3.21206450e-01 9.21072364e-01 7.30449319e-01
6.51613474e-01 2.23982587e-01 9.93628323e-01 -1.12439823e+00
7.68081725e-01 -7.35118747e-01 -6.29447281e-01 9.42276239e-01
-6.65660381e-01 4.39842612e-01 4.62095708e-01 -4.61079590e-02
-1.35590029e+00 -1.40341341e-01 -3.46559286e-01 3.24689031e-01
-2.89233863e-01 1.09450054e+00 -5.66010058e-01 7.12583140e-02
5.79729319e-01 1.76343903e-01 -5.81027567e-01 -6.46345258e-01
6.28803194e-01 3.06788355e-01 5.36003172e-01 -8.49327803e-01
9.55854356e-01 3.56033415e-01 -1.38062593e-02 -4.00495917e-01
-1.40982652e+00 -9.10409451e-01 -4.72416282e-01 2.14262933e-01
6.86082065e-01 -7.34566629e-01 -7.57484376e-01 -3.28630924e-01
-1.19984412e+00 3.80042046e-01 -4.09226060e-01 3.43471080e-01
-1.47135496e-01 4.80154544e-01 -2.84676373e-01 -7.63698101e-01
-4.69788879e-01 -5.69828331e-01 7.17878282e-01 3.06047291e-01
-1.10314209e-02 -6.79814160e-01 3.44125301e-01 8.87319744e-01
1.71484426e-01 -1.42954126e-01 1.45261657e+00 -1.22361267e+00
-8.71382892e-01 -2.45348662e-01 -3.76139641e-01 -8.04166943e-02
8.49155262e-02 -6.32881224e-01 -7.80229628e-01 2.80754387e-01
1.61161393e-01 -2.81873673e-01 8.41669858e-01 -2.48138726e-01
8.36526871e-01 -7.30284452e-01 -1.86439514e-01 -1.21078201e-01
1.28036571e+00 1.32014388e-02 6.55968964e-01 5.79661727e-01
3.09696168e-01 1.01914239e+00 8.17999661e-01 2.08841711e-01
8.19483221e-01 4.28450763e-01 2.47397527e-01 3.99206668e-01
-1.02012210e-01 -4.66759294e-01 -1.23663127e-01 7.54594147e-01
1.95904270e-01 -1.38165697e-01 -1.01228786e+00 8.60579789e-01
-1.79381812e+00 -8.58227074e-01 -1.35351837e-01 1.94163442e+00
1.30115151e+00 9.18493718e-02 -1.02415755e-01 2.92399138e-01
4.44326222e-01 -5.68230860e-02 -2.95867056e-01 2.78656781e-01
-1.93831161e-01 6.04431093e-01 -5.10961702e-03 5.20232737e-01
-9.28639233e-01 1.00881708e+00 4.77581263e+00 8.80077362e-01
-3.67875010e-01 -8.88347030e-02 2.34358668e-01 5.31871200e-01
-5.76803565e-01 3.97947699e-01 -1.05906665e+00 7.36572295e-02
7.02836454e-01 -2.74523914e-01 1.95196733e-01 6.44113779e-01
-3.74709815e-01 -2.85996646e-01 -9.70561266e-01 6.20454490e-01
5.03829755e-02 -1.37988889e+00 4.89664257e-01 -2.70631671e-01
6.67798877e-01 -3.54098827e-01 -2.98947155e-01 5.36402285e-01
2.64049858e-01 -7.06659377e-01 1.67740256e-01 6.94905221e-01
1.92367017e-01 -6.25097454e-01 8.66981447e-01 5.34316897e-01
-1.16734922e+00 -1.28937662e-01 -3.44983518e-01 1.17502332e-01
3.51442486e-01 9.38774228e-01 -9.21819568e-01 1.12145448e+00
5.84867835e-01 1.01488680e-01 -5.79280615e-01 1.21750736e+00
-6.17726862e-01 6.93686724e-01 -3.75996053e-01 -2.24157959e-01
1.50789469e-01 1.87802892e-02 5.29375374e-01 9.78956163e-01
2.54968882e-01 5.46601653e-01 -6.13389574e-02 6.53092086e-01
-5.59842706e-01 4.69105989e-01 -2.11431369e-01 -1.48903308e-02
4.95061010e-01 1.19915700e+00 -3.02504182e-01 -4.85045969e-01
-5.26414812e-01 6.89979315e-01 6.18314922e-01 2.87549764e-01
-4.41118300e-01 -4.84907568e-01 2.39387944e-01 -9.28946361e-02
4.33161467e-01 1.12544797e-01 2.44787652e-02 -1.36125255e+00
7.00009048e-01 -9.60100412e-01 9.74944890e-01 -4.13254470e-01
-1.42634284e+00 4.47821409e-01 -7.73360282e-02 -9.38983142e-01
-2.16445163e-01 -3.83296698e-01 -3.07410479e-01 7.44613886e-01
-2.34945297e+00 -9.02719140e-01 -5.03588840e-02 7.32420504e-01
1.86130926e-01 1.63190499e-01 8.18388045e-01 5.03235698e-01
-3.07859302e-01 4.04849648e-01 -3.04654866e-01 1.12106152e-01
6.35588944e-01 -1.19161260e+00 6.84780329e-02 8.69697928e-01
4.76144701e-01 9.07158136e-01 5.30804574e-01 -6.67432010e-01
-1.64767063e+00 -8.43194723e-01 1.64968765e+00 -5.21164477e-01
7.20113397e-01 4.47330959e-02 -1.45723450e+00 4.95303452e-01
2.43866801e-01 1.63457431e-02 7.29024529e-01 3.80148679e-01
-5.59950352e-01 -2.43464038e-01 -1.18928707e+00 3.74302566e-01
7.72427022e-01 -5.38065672e-01 -1.42024899e+00 3.39634597e-01
9.70616341e-01 -3.91678572e-01 -8.60299706e-01 6.11122012e-01
1.97914585e-01 -5.25751770e-01 9.76702869e-01 -8.30643296e-01
3.73732656e-01 -5.07220209e-01 -1.27831936e-01 -9.10524011e-01
-6.30995184e-02 -5.41404068e-01 -7.53275275e-01 1.46016979e+00
8.11976433e-01 -6.42068088e-01 7.69347966e-01 8.76155317e-01
3.03710327e-02 -8.18667173e-01 -1.18555057e+00 -5.30907929e-01
-2.14732543e-01 -9.98571888e-02 8.32142770e-01 9.07047808e-01
4.11255211e-01 6.32543027e-01 4.18301411e-02 6.28877759e-01
3.51372182e-01 5.08273840e-01 5.24895489e-01 -1.40062153e+00
-2.95222461e-01 -1.95641369e-01 -7.40711205e-03 -1.27230310e+00
-3.24265435e-02 -9.34070289e-01 -1.57288626e-01 -1.88846231e+00
2.14415908e-01 -3.83109838e-01 -3.42413336e-01 4.97064173e-01
-7.90469885e-01 -2.78372675e-01 -2.29226444e-02 9.93199274e-02
-1.00762856e+00 7.42150247e-01 1.11465740e+00 -2.80692112e-02
-9.26955044e-02 -4.04039398e-02 -9.41152990e-01 5.60798705e-01
5.73604107e-01 -4.34723318e-01 -2.83259600e-01 -2.33864561e-01
6.58794463e-01 3.13174427e-01 2.35247090e-01 -8.95233691e-01
7.56994247e-01 -2.33078897e-01 -1.35256320e-01 -4.61713135e-01
2.87967503e-01 -9.63631570e-01 -3.54810834e-01 2.00931147e-01
-3.90362561e-01 -5.68478219e-02 1.29440129e-01 6.44642949e-01
-6.52136326e-01 -3.56029391e-01 2.45228544e-01 -4.72653285e-02
-7.40037024e-01 3.18622112e-01 2.82178782e-02 6.31573558e-01
6.26144350e-01 4.48094726e-01 -7.03751981e-01 -1.49643689e-01
-4.33199912e-01 4.29061174e-01 -3.06980789e-01 2.25045890e-01
6.24360502e-01 -1.13941300e+00 -6.90161228e-01 -2.11253241e-01
5.13447702e-01 1.70934871e-01 3.41435403e-01 7.26567864e-01
-1.92238197e-01 9.38735783e-01 2.00046033e-01 6.13433383e-02
-9.63833094e-01 5.29804230e-01 -2.07734033e-02 -1.02918398e+00
-4.40667778e-01 8.59848261e-01 -2.59569764e-01 -3.84959310e-01
1.71739638e-01 -2.65923470e-01 -8.23459327e-01 7.50776082e-02
6.20879114e-01 1.55146703e-01 4.54981923e-01 -1.89391106e-01
-3.63858789e-01 2.08293259e-01 -2.62468874e-01 5.42833656e-02
1.21672940e+00 -1.53091118e-01 -4.65608448e-01 -7.52475038e-02
5.69769502e-01 2.57502973e-01 -5.51655769e-01 -9.25252736e-01
8.08169723e-01 -3.47113252e-01 -3.12479794e-01 -1.17306030e+00
-4.58582014e-01 6.36577785e-01 2.39834711e-02 3.97245049e-01
9.21513915e-01 4.84230518e-01 1.22594106e+00 9.59931195e-01
3.24070841e-01 -8.30455005e-01 -7.86308292e-03 6.51666582e-01
9.01216447e-01 -1.15586150e+00 -3.15901786e-02 -8.48141491e-01
-4.80836660e-01 8.93454611e-01 5.65460205e-01 1.90447539e-01
3.86656016e-01 -3.03074986e-01 -1.67961150e-01 -5.23326933e-01
-8.69222105e-01 -6.64522886e-01 7.56296158e-01 1.59460619e-01
1.30638063e-01 -3.57974052e-01 -5.94500422e-01 8.95310581e-01
-1.44237906e-01 2.88745277e-02 -1.60731688e-01 9.95666325e-01
-8.41789722e-01 -1.36720729e+00 -3.68795305e-01 6.58298492e-01
-6.30902410e-01 -3.73567283e-01 -3.00428599e-01 4.70153898e-01
1.56522486e-02 1.14735389e+00 -5.34553766e-01 5.66887036e-02
6.79291129e-01 6.68735027e-01 3.39305907e-01 -5.94210207e-01
-6.30524158e-01 -5.99579871e-01 5.52576005e-01 -2.95299262e-01
-4.07642215e-01 -4.25185651e-01 -1.17372155e+00 2.05269948e-01
-4.25285459e-01 7.47331083e-01 2.34817609e-01 1.36565363e+00
6.75810277e-01 3.10839623e-01 3.34135264e-01 3.58811021e-01
-6.88843191e-01 -8.91575813e-01 -2.19313785e-01 4.07663524e-01
1.98762313e-01 -6.28244817e-01 -1.75128117e-01 -1.53061673e-02] | [10.559683799743652, 7.94567346572876] |
c8841a3c-bde5-47a0-8482-512560f9cca1 | first-steps-toward-camera-model | 1603.01068 | null | http://arxiv.org/abs/1603.01068v2 | http://arxiv.org/pdf/1603.01068v2.pdf | First Steps Toward Camera Model Identification with Convolutional Neural Networks | Detecting the camera model used to shoot a picture enables to solve a wide
series of forensic problems, from copyright infringement to ownership
attribution. For this reason, the forensic community has developed a set of
camera model identification algorithms that exploit characteristic traces left
on acquired images by the processing pipelines specific of each camera model.
In this paper, we investigate a novel approach to solve camera model
identification problem. Specifically, we propose a data-driven algorithm based
on convolutional neural networks, which learns features characterizing each
camera model directly from the acquired pictures. Results on a well-known
dataset of 18 camera models show that: (i) the proposed method outperforms
up-to-date state-of-the-art algorithms on classification of 64x64 color image
patches; (ii) features learned by the proposed network generalize to camera
models never used for training. | ['Paolo Bestagini', 'David Güera', 'Luca Bondi', 'Edward J. Delp', 'Stefano Tubaro', 'Luca Baroffio'] | 2016-03-03 | null | null | null | null | ['image-forensics'] | ['computer-vision'] | [ 5.26559114e-01 -6.49670064e-01 -3.93724330e-02 -2.63124973e-01
-5.89569092e-01 -7.92071700e-01 6.93295598e-01 8.35443139e-02
-4.07109499e-01 2.06619635e-01 -3.49660903e-01 -3.41138333e-01
-1.36670873e-01 -5.32228947e-01 -8.25579166e-01 -4.78901058e-01
1.98073927e-02 1.13292605e-01 4.74031791e-02 3.80164176e-01
9.86171067e-01 1.21156394e+00 -1.53959823e+00 4.11748797e-01
2.14670926e-01 1.24229991e+00 -9.07988176e-02 7.51164913e-01
-2.96146106e-02 1.09125984e+00 -5.47230661e-01 -6.10139072e-01
4.51601982e-01 -1.27063766e-01 -6.91993117e-01 3.17050964e-01
9.16641295e-01 -6.88557386e-01 -5.16970396e-01 1.28276086e+00
-4.87623066e-02 -3.12660694e-01 6.57777309e-01 -1.35473645e+00
-6.63416564e-01 3.32502842e-01 -7.17192292e-01 5.94156742e-01
1.18212290e-01 2.28921145e-01 7.08491027e-01 -6.76940024e-01
5.82747281e-01 8.53962243e-01 7.51214981e-01 3.26138467e-01
-1.11045361e+00 -1.06395042e+00 -4.26010221e-01 4.97947991e-01
-1.44081235e+00 -4.54166353e-01 9.62201416e-01 -7.41353273e-01
6.65750086e-01 1.44103356e-02 4.22214359e-01 9.98490512e-01
1.30701661e-01 6.20013654e-01 1.45506382e+00 -5.15705764e-01
8.97262990e-02 2.83121198e-01 5.91509581e-01 7.21042991e-01
5.99154830e-01 1.51489884e-01 -5.25594532e-01 -3.04650575e-01
5.32403588e-01 5.13620824e-02 -2.90706027e-02 -3.21015328e-01
-6.77169204e-01 7.09767520e-01 -1.31316051e-01 2.92438120e-01
-4.06256706e-01 2.63330907e-01 4.88136470e-01 8.41863900e-02
-6.19123019e-02 5.11788011e-01 -2.87158191e-02 -2.87020132e-02
-1.37295973e+00 9.41798314e-02 6.84536159e-01 8.39312196e-01
9.79539573e-01 -8.95326734e-02 3.73005599e-01 3.29349279e-01
1.50129169e-01 4.21690464e-01 2.34540388e-01 -8.48237574e-01
2.68528461e-01 7.07013667e-01 -6.29919916e-02 -1.19764376e+00
-1.15340315e-01 -6.09734394e-02 -6.03709996e-01 3.27458978e-01
5.31714916e-01 3.17668498e-01 -5.57493269e-01 1.10288405e+00
2.53909156e-02 4.71665859e-01 -1.58675641e-01 6.34787858e-01
4.99247551e-01 3.68472964e-01 -2.07723066e-01 1.08494096e-01
1.34596229e+00 -5.56890786e-01 -3.96880656e-01 3.14193941e-03
1.86249360e-01 -7.87024081e-01 5.21099925e-01 1.01706254e+00
-6.24435961e-01 -6.92780495e-01 -1.25141108e+00 1.39005229e-01
-5.50334930e-01 4.96797085e-01 3.97237718e-01 1.09928477e+00
-6.45427704e-01 7.80929565e-01 -4.16825086e-01 -4.13686484e-01
9.04373765e-01 4.85571176e-01 -6.39398575e-01 -2.87291378e-01
-5.06523073e-01 7.26540804e-01 4.14908469e-01 1.01484269e-01
-1.35448217e+00 -6.85283482e-01 -4.71512914e-01 1.29983127e-01
2.11807117e-01 3.82917821e-02 7.96985209e-01 -1.25387168e+00
-1.22970784e+00 1.41046584e+00 1.96001068e-01 -6.06591761e-01
5.22455633e-01 -4.01966572e-01 -6.61877751e-01 7.10943997e-01
3.79494131e-02 2.46417582e-01 1.48523533e+00 -1.43711770e+00
-6.35727882e-01 -2.74804980e-01 -5.16640395e-02 -7.77165651e-01
-6.66580737e-01 6.33303285e-01 -4.04589415e-01 -4.57136005e-01
-5.40069222e-01 -8.94936562e-01 3.11453402e-01 1.98216528e-01
-7.44758844e-01 2.32971638e-01 1.08180785e+00 -7.79330492e-01
1.20743549e+00 -2.26247787e+00 -3.74159098e-01 4.13561583e-01
9.81933624e-02 6.71787262e-01 -4.31281078e-04 5.43932736e-01
-2.76870370e-01 1.80118248e-01 -3.26484650e-01 -3.62782806e-01
-1.24273442e-01 -8.96614566e-02 -5.14342844e-01 1.00956213e+00
2.44855329e-01 1.90921605e-01 -4.07654107e-01 -6.37076378e-01
5.66896915e-01 1.78298756e-01 -6.59514666e-02 3.99702251e-01
2.92472631e-01 2.32860550e-01 -2.16760248e-01 9.63028729e-01
1.25248206e+00 -7.05219284e-02 2.24356782e-02 -2.22052008e-01
-8.47687274e-02 -4.35865581e-01 -1.15354073e+00 1.36902475e+00
-1.86619818e-01 9.83720362e-01 6.05038330e-02 -8.46375227e-01
9.97579038e-01 -2.04930604e-02 3.98078650e-01 -1.83858484e-01
3.76928985e-01 1.13635927e-01 -1.83411464e-01 -7.11720705e-01
6.19209349e-01 -7.11991489e-02 4.76779826e-02 7.96637654e-01
3.21961462e-01 2.74518549e-01 2.25562468e-01 9.37515199e-02
1.09967268e+00 2.26320084e-02 2.37988830e-01 -1.58892587e-01
1.03139615e+00 7.14261159e-02 3.43060106e-01 9.26234722e-01
-2.69006461e-01 5.93326092e-01 4.94608372e-01 -7.93757319e-01
-1.22484183e+00 -7.49607742e-01 -1.52784720e-01 7.41613448e-01
1.46486253e-01 -2.39611939e-01 -1.09696627e+00 -6.18176699e-01
1.43438593e-01 5.83588600e-01 -8.03131461e-01 9.63202119e-02
-7.32938886e-01 -3.06996703e-01 1.03375542e+00 4.03287888e-01
4.85882670e-01 -1.01351511e+00 -9.59913373e-01 -7.93054178e-02
4.07554239e-01 -1.52122951e+00 -2.69644707e-01 -7.50569552e-02
-7.01469839e-01 -1.78995109e+00 -2.08138108e-01 -2.76371151e-01
5.94348490e-01 4.35337305e-01 8.09750676e-01 3.55716586e-01
-7.09119260e-01 7.20123887e-01 -2.98292547e-01 -3.91596824e-01
-5.64557612e-01 4.31958251e-02 -4.23478261e-02 7.36493945e-01
5.84184349e-01 -2.91818827e-01 -2.25419164e-01 5.32808192e-02
-1.28866875e+00 -5.05003572e-01 6.61731541e-01 5.08746624e-01
1.48647860e-01 2.20353246e-01 -3.14438567e-02 -9.90608275e-01
4.99976993e-01 -4.97037470e-01 -1.00300837e+00 4.92362052e-01
-4.61576670e-01 -1.25798047e-01 7.57198393e-01 -4.25618410e-01
-9.03678060e-01 4.20570672e-01 4.46542561e-01 -1.06262863e+00
-7.13585258e-01 2.65561789e-02 -1.49572566e-01 -4.54600453e-01
2.87290245e-01 3.31179678e-01 1.08300194e-01 -6.87680602e-01
1.97234750e-01 7.15313673e-01 9.01538849e-01 -5.61068416e-01
1.11522865e+00 8.36995602e-01 2.51649886e-01 -1.07102513e+00
-4.09473836e-01 -6.98326051e-01 -1.06740928e+00 -5.26742697e-01
9.77209449e-01 -5.92111945e-01 -7.22320795e-01 1.01860774e+00
-1.34446657e+00 2.78950214e-01 1.75281152e-01 3.50752860e-01
-4.31035012e-01 9.79380906e-01 -4.59021270e-01 -9.64364827e-01
-2.53126949e-01 -1.09850252e+00 1.02934146e+00 2.80004621e-01
1.44385677e-02 -8.52807641e-01 3.47185805e-02 4.76375103e-01
1.12111099e-01 3.30122292e-01 7.77550340e-01 -7.66814768e-01
-7.94579923e-01 -5.92740655e-01 -4.28072840e-01 7.69924939e-01
-1.46860957e-01 4.33484793e-01 -1.35928929e+00 -2.60088056e-01
3.49186957e-02 -2.24565804e-01 6.98768377e-01 -6.03247434e-03
1.31041861e+00 4.77244630e-02 -2.46400982e-01 9.89842296e-01
1.76697421e+00 5.56084551e-02 8.27845097e-01 7.82402515e-01
6.26746714e-01 4.55334663e-01 2.77655661e-01 5.72289705e-01
-1.97529614e-01 4.11471903e-01 6.39754713e-01 1.99024513e-01
1.20586485e-01 -1.74661219e-01 3.72715592e-01 8.39565620e-02
6.79615736e-02 -2.52853241e-02 -1.01389885e+00 3.48898858e-01
-1.49687994e+00 -1.03224874e+00 -3.27372402e-01 2.08377647e+00
3.95869650e-03 6.84619024e-02 9.36122909e-02 1.88600123e-01
1.25602734e+00 1.02235004e-01 -4.28295076e-01 -3.03547531e-01
-1.08494259e-01 4.20746386e-01 8.27048421e-01 -8.52810666e-02
-1.46964598e+00 8.49168897e-01 6.57306957e+00 8.96884143e-01
-1.22312093e+00 8.66004303e-02 2.71186620e-01 2.26410940e-01
4.23149586e-01 2.79933274e-01 -8.38421404e-01 5.82557440e-01
8.82364750e-01 1.20535735e-02 3.31289887e-01 1.05760574e+00
-1.18287586e-01 -1.08008213e-01 -1.21920788e+00 1.28486884e+00
6.21794760e-01 -1.49825323e+00 1.24851853e-01 5.12464643e-01
3.24470431e-01 -3.93652141e-01 1.67720318e-01 -2.04815641e-01
-8.93981680e-02 -1.09719718e+00 9.29823697e-01 7.64528871e-01
7.30802953e-01 -7.37606943e-01 8.01735342e-01 2.22298741e-01
-8.71547222e-01 -4.96357828e-01 -4.73031014e-01 7.48592690e-02
1.34315059e-01 1.84686929e-01 -6.89490616e-01 3.99245769e-01
7.20703125e-01 8.88099551e-01 -1.21959662e+00 1.21751416e+00
-1.58877701e-01 7.15793788e-01 -3.39590833e-02 3.71117562e-01
2.50644386e-01 -1.97123706e-01 3.70479643e-01 1.28702497e+00
4.25048113e-01 -9.88021418e-02 -3.38219613e-01 1.00962996e+00
-1.66941792e-01 -1.35323808e-01 -8.90937150e-01 -2.02255607e-01
5.11404216e-01 1.44057143e+00 -8.14519882e-01 -1.75363109e-01
-3.88190955e-01 8.66164029e-01 1.18032411e-01 -4.50387262e-02
-8.03776979e-01 -2.32516617e-01 3.53301704e-01 8.09716433e-02
7.52064407e-01 -8.03743303e-02 -1.15179896e-01 -1.09361720e+00
-1.09107770e-01 -9.05939400e-01 4.85496908e-01 -8.09995949e-01
-1.48965120e+00 5.12617230e-01 6.07732348e-02 -1.52584755e+00
2.47293264e-02 -1.10447705e+00 -8.91190529e-01 3.99086475e-01
-1.54286146e+00 -1.53603482e+00 -4.09766048e-01 6.54981375e-01
1.97139978e-01 -8.52118850e-01 6.48885727e-01 3.15589339e-01
-6.49865866e-01 5.36622584e-01 3.31827134e-01 6.07856810e-01
7.26740658e-01 -8.18646967e-01 6.15926050e-02 1.20321429e+00
2.88998514e-01 7.39517152e-01 3.12185287e-01 -3.37171406e-01
-1.73944914e+00 -9.49075520e-01 3.78280103e-01 -6.00904763e-01
8.17393184e-01 -1.46875903e-01 -8.32776368e-01 6.34146154e-01
3.75040293e-01 1.73833385e-01 9.62763369e-01 -3.22850287e-01
-8.17140698e-01 -4.34662670e-01 -1.23654759e+00 -3.36386025e-01
2.44573891e-01 -8.34464788e-01 -6.20018423e-01 -1.09415203e-02
-3.26070666e-01 1.56575054e-01 -8.21723819e-01 -3.02748889e-01
7.62960792e-01 -1.21421897e+00 1.02677643e+00 -6.65841460e-01
5.82389772e-01 -2.98505902e-01 -2.99691796e-01 -4.59454268e-01
1.69211794e-02 -5.48506320e-01 -1.54267311e-01 1.44284964e+00
-2.19801202e-01 -3.24746281e-01 8.12752187e-01 4.35345471e-01
1.44981027e-01 -4.00962681e-02 -1.09281743e+00 -8.19027543e-01
1.40641063e-01 -4.86656010e-01 6.27801657e-01 1.03908837e+00
-5.96763551e-01 -1.09601229e-01 -7.26781249e-01 3.26102108e-01
1.16539931e+00 1.61401361e-01 1.09951532e+00 -1.32424390e+00
-2.67065138e-01 -5.14751017e-01 -7.83952773e-01 -4.24832940e-01
5.18696547e-01 -4.99451071e-01 -4.21353608e-01 -6.12774670e-01
5.52051187e-01 -3.37292284e-01 -2.73598552e-01 1.29747853e-01
2.17105776e-01 5.15017688e-01 6.34744167e-01 5.94275057e-01
-5.20834506e-01 -2.92151392e-01 4.44123745e-01 -2.91634351e-01
5.04486263e-01 -2.45999783e-01 -4.81150746e-01 9.10156429e-01
5.63719928e-01 -8.05430174e-01 1.11127153e-01 -2.98466861e-01
4.32437360e-02 -5.13606220e-02 7.62085319e-01 -1.40573204e+00
4.61702496e-01 2.44119987e-02 5.00249565e-01 -7.34375060e-01
2.84375191e-01 -1.09939349e+00 3.65187764e-01 4.54988092e-01
-1.94261655e-01 -3.13459081e-03 1.04538865e-01 6.55513287e-01
-2.68700838e-01 -9.53344524e-01 1.04679537e+00 -2.51270115e-01
-1.02069950e+00 5.96191883e-02 -6.26870811e-01 -2.96974272e-01
1.32615674e+00 -4.53734726e-01 -4.04996991e-01 -7.30847567e-02
-3.62377495e-01 -4.17470485e-01 8.40474367e-01 1.87967539e-01
5.97317100e-01 -1.07338345e+00 -4.51528221e-01 1.78490907e-01
3.03852320e-01 -7.50548422e-01 3.24290812e-01 5.46570599e-01
-1.03549767e+00 2.08059207e-01 -6.03243053e-01 -6.67469501e-01
-1.57224691e+00 9.79946196e-01 2.10822254e-01 -1.53286591e-01
-4.78164405e-01 4.15842324e-01 -2.48312280e-01 -1.22153029e-01
-7.33736753e-02 5.93391992e-02 -1.34344921e-01 2.24368289e-01
7.19805241e-01 6.83475375e-01 1.96052901e-02 -1.12644720e+00
-4.76182997e-01 7.65688956e-01 -1.88428536e-01 2.22811252e-01
1.59495091e+00 2.55035788e-01 -3.09779286e-01 1.76039301e-02
1.37007368e+00 -9.54245329e-02 -1.06575871e+00 1.57185402e-02
1.60237670e-01 -9.21165228e-01 6.24261498e-02 -5.05636096e-01
-1.13182378e+00 1.30765569e+00 9.08803582e-01 1.90357298e-01
1.08938301e+00 -2.90476978e-01 6.76059723e-01 5.73152125e-01
3.63742799e-01 -1.16388893e+00 1.21396989e-01 9.41712931e-02
3.24943155e-01 -1.10697556e+00 3.76115054e-01 -1.51691854e-01
-5.10490894e-01 1.78313911e+00 2.81819880e-01 -4.99916047e-01
5.64897835e-01 1.79980814e-01 -4.34967577e-02 -3.28282416e-01
-1.95069343e-01 1.55340438e-03 -7.52255097e-02 7.53435969e-01
-1.36480793e-01 -1.68714374e-01 2.79866487e-01 4.91926938e-01
2.03323543e-01 1.75972342e-01 9.64254439e-01 9.44077492e-01
-2.89764136e-01 -1.23017800e+00 -9.84691083e-01 3.30240935e-01
-6.83566153e-01 7.68536255e-02 -8.11433673e-01 1.21683490e+00
3.79121840e-01 8.72757018e-01 9.66959968e-02 -3.99323493e-01
2.45469976e-02 -1.04364149e-01 5.00093877e-01 -1.88882262e-01
-8.45942914e-01 -2.50233561e-01 -1.11776888e-01 -4.96107846e-01
-7.31085777e-01 -1.18090034e+00 -5.69896162e-01 -3.57528031e-01
-3.71361583e-01 -1.38749644e-01 7.04914749e-01 8.23551297e-01
2.43594855e-01 1.06376357e-01 7.24822164e-01 -9.67393100e-01
-6.32925391e-01 -8.62173319e-01 -9.07529712e-01 7.65617490e-01
3.55980635e-01 -6.75799072e-01 -4.06240940e-01 5.04002035e-01] | [12.385390281677246, 1.0024831295013428] |
3f3e5688-d597-40af-8621-10aafce4a354 | slam-a-unified-encoder-for-speech-and | 2110.10329 | null | https://arxiv.org/abs/2110.10329v1 | https://arxiv.org/pdf/2110.10329v1.pdf | SLAM: A Unified Encoder for Speech and Language Modeling via Speech-Text Joint Pre-Training | Unsupervised pre-training is now the predominant approach for both text and speech understanding. Self-attention models pre-trained on large amounts of unannotated data have been hugely successful when fine-tuned on downstream tasks from a variety of domains and languages. This paper takes the universality of unsupervised language pre-training one step further, by unifying speech and text pre-training within a single model. We build a single encoder with the BERT objective on unlabeled text together with the w2v-BERT objective on unlabeled speech. To further align our model representations across modalities, we leverage alignment losses, specifically Translation Language Modeling (TLM) and Speech Text Matching (STM) that make use of supervised speech-text recognition data. We demonstrate that incorporating both speech and text data during pre-training can significantly improve downstream quality on CoVoST~2 speech translation, by around 1 BLEU compared to single-modality pre-trained models, while retaining close to SotA performance on LibriSpeech and SpeechStew ASR tasks. On four GLUE tasks and text-normalization, we observe evidence of capacity limitations and interference between the two modalities, leading to degraded performance compared to an equivalent text-only model, while still being competitive with BERT. Through extensive empirical analysis we also demonstrate the importance of the choice of objective function for speech pre-training, and the beneficial effect of adding additional supervised signals on the quality of the learned representations. | ['Yu Zhang', 'Alexis Conneau', 'Jason Riesa', 'Melvin Johnson', 'Jonathan H. Clark', 'Ye Jia', 'Anmol Gulati', 'Nan Wu', 'Yu-An Chung', 'Ankur Bapna'] | 2021-10-20 | null | null | null | null | ['unsupervised-pre-training'] | ['methodology'] | [ 6.81953073e-01 4.47113246e-01 -4.50458229e-01 -7.15197563e-01
-1.36223233e+00 -4.71825927e-01 9.48888183e-01 2.56636608e-02
-6.58614218e-01 4.11040574e-01 9.12655413e-01 -8.50493968e-01
3.02046686e-01 -8.84897411e-02 -9.98981893e-01 -2.06323847e-01
5.68080604e-01 7.30128884e-01 -2.69929677e-01 -3.58186811e-01
-2.96676278e-01 6.61668703e-02 -1.04764318e+00 4.69906569e-01
7.57178783e-01 4.95745480e-01 3.70130032e-01 7.68467188e-01
-2.50697225e-01 7.34907985e-01 -4.24009234e-01 -6.06624842e-01
2.08499104e-01 -6.01397514e-01 -9.29752707e-01 1.47037938e-01
5.15413046e-01 -2.60043532e-01 -7.65280366e-01 7.50947833e-01
6.85074031e-01 1.75100788e-01 5.51430464e-01 -8.00309241e-01
-8.61121595e-01 1.19818878e+00 -2.89274335e-01 4.17935193e-01
2.62447540e-02 1.29333422e-01 1.33599102e+00 -9.79616165e-01
5.20090401e-01 1.30813849e+00 4.55544233e-01 8.43781292e-01
-1.49593341e+00 -5.90426087e-01 9.30046067e-02 -1.11156985e-01
-9.61582065e-01 -1.52214611e+00 2.54098177e-01 -1.75543219e-01
1.64829731e+00 2.61768192e-01 -1.09392069e-01 1.26557672e+00
-5.97326271e-02 9.10684168e-01 6.94379210e-01 -8.81191075e-01
-2.98392117e-01 1.95536062e-01 -4.68450151e-02 2.87070155e-01
-2.07027465e-01 2.23384634e-01 -8.47115219e-01 3.14744234e-01
3.67927790e-01 -3.10236394e-01 -2.64531821e-01 7.69157112e-02
-1.33981895e+00 6.06741488e-01 1.31066501e-01 3.65541160e-01
6.37076497e-02 2.39245221e-01 6.10693455e-01 7.40147114e-01
6.38157368e-01 2.49897540e-01 -6.50354326e-01 -3.84997487e-01
-1.13649535e+00 -3.85781407e-01 5.84010482e-01 1.23430419e+00
4.88759249e-01 2.31459886e-01 -3.52421552e-01 1.04159188e+00
4.15860027e-01 1.00225449e+00 8.46662283e-01 -5.78343868e-01
1.12764990e+00 1.33753017e-01 -3.42558444e-01 -9.71027389e-02
-7.15936199e-02 -4.73947942e-01 -5.37156761e-01 -3.05867314e-01
4.01676178e-01 -3.11225653e-01 -1.34417415e+00 2.06032801e+00
-2.34720334e-01 -1.86166987e-01 4.03282791e-01 5.22108972e-01
5.58491409e-01 8.87067139e-01 2.08224192e-01 -2.33532727e-01
1.06230056e+00 -1.18376863e+00 -7.58193493e-01 -6.75632417e-01
9.40963268e-01 -1.06380939e+00 1.26557291e+00 -1.41710177e-01
-1.23677075e+00 -5.48990071e-01 -9.09782410e-01 -5.11920810e-01
-1.46983087e-01 2.39339039e-01 1.84120104e-01 7.87640870e-01
-1.37159145e+00 4.94434834e-01 -9.60459054e-01 -6.09200716e-01
1.40377611e-01 5.61903596e-01 -3.86083543e-01 -1.14584684e-01
-1.05295753e+00 1.14891696e+00 6.23874143e-02 -2.10267693e-01
-7.42708802e-01 -5.56654155e-01 -8.63726437e-01 3.60120416e-01
1.22701533e-01 -6.25164747e-01 1.61653507e+00 -1.36898434e+00
-1.81461942e+00 8.72745395e-01 -6.25484705e-01 -5.94541132e-01
2.22376972e-01 -2.19904080e-01 -6.77616835e-01 -2.53519386e-01
-1.06061205e-01 6.52261615e-01 9.75391686e-01 -7.42116213e-01
-3.63986731e-01 -2.32075229e-01 -3.73372018e-01 6.22382939e-01
-4.58222687e-01 4.33861911e-01 -5.53605795e-01 -6.68351769e-01
-4.77384962e-02 -8.20024252e-01 1.03454843e-01 -5.85072339e-01
-3.76325428e-01 -5.79560548e-02 6.06674016e-01 -9.61464822e-01
1.19859815e+00 -2.20251298e+00 3.77037853e-01 -7.54670007e-03
-1.68141380e-01 3.03650916e-01 -6.43384993e-01 5.72278321e-01
-2.98544496e-01 1.85566202e-01 -1.28183901e-01 -1.02845681e+00
1.00814864e-01 3.03156912e-01 -5.35623431e-01 3.32697123e-01
3.50339353e-01 1.17098331e+00 -4.91925001e-01 -1.03409305e-01
1.30081445e-01 3.55131269e-01 -5.98822892e-01 2.11383954e-01
-1.90753058e-01 4.13444519e-01 1.38065308e-01 3.59548718e-01
9.13392827e-02 -2.70044059e-01 3.52657109e-01 9.04478580e-02
1.20612234e-01 1.18850601e+00 -6.12212479e-01 1.80356431e+00
-7.69949913e-01 9.49253976e-01 2.48847201e-01 -9.96295333e-01
5.36686122e-01 7.28975296e-01 1.94354743e-01 -9.78055656e-01
2.93947816e-01 3.47710818e-01 2.14378953e-01 -1.90603122e-01
5.15062690e-01 -4.99923587e-01 6.13774620e-02 6.77195728e-01
6.83318853e-01 -5.86452298e-02 -1.09706849e-01 3.11995089e-01
9.94483113e-01 -3.23987044e-02 -1.79252159e-02 -1.08340092e-01
3.44221860e-01 -1.08063467e-01 1.10860139e-01 6.80845499e-01
6.18409961e-02 6.39638305e-01 1.12850599e-01 3.39676678e-01
-1.30170619e+00 -8.72718275e-01 4.55946755e-03 1.77247691e+00
-2.85367548e-01 -4.72497374e-01 -7.61467278e-01 -6.48078203e-01
-3.82093294e-03 9.64524865e-01 -2.02998936e-01 -2.89503127e-01
-8.67496014e-01 -5.67483604e-01 9.11079109e-01 6.03560627e-01
-1.81991026e-01 -7.92708755e-01 2.61360615e-01 2.14546323e-01
-2.50232667e-01 -1.23639190e+00 -9.32291448e-01 5.66521108e-01
-8.26349854e-01 -2.29655519e-01 -7.48076022e-01 -7.34432876e-01
5.21919310e-01 2.06253409e-01 1.09148228e+00 -2.74302550e-02
3.00192088e-01 4.19356465e-01 -3.12022865e-01 -2.10793018e-01
-1.08178699e+00 5.29525578e-01 2.87366569e-01 -3.92922834e-02
3.19043964e-01 -4.44800675e-01 6.42379969e-02 2.09651977e-01
-6.61137760e-01 1.45745710e-01 8.06657851e-01 1.06314552e+00
1.47483006e-01 -6.93965316e-01 6.09165788e-01 -8.12808990e-01
4.97279733e-01 -3.83450836e-01 -3.14114600e-01 2.84181148e-01
-7.91838229e-01 4.26759154e-01 5.52289307e-01 -3.53619277e-01
-1.12986243e+00 -1.64874271e-01 -4.27421242e-01 -4.26553637e-01
1.50421048e-02 5.04367352e-01 -3.33109498e-01 2.10437000e-01
6.25276268e-01 3.04639935e-01 3.21523577e-01 -6.70283079e-01
8.21563959e-01 1.15901041e+00 5.56158662e-01 -5.26930869e-01
6.97250009e-01 -2.81529631e-02 -6.06264412e-01 -7.76339054e-01
-7.03038752e-01 -5.25717735e-01 -5.36999464e-01 2.90892899e-01
7.27725089e-01 -1.16710925e+00 -2.59161860e-01 2.01825887e-01
-1.15486801e+00 -5.77026904e-01 -4.79760647e-01 6.08186662e-01
-5.62067807e-01 3.85116220e-01 -7.86291301e-01 -6.18835688e-01
-3.54780436e-01 -1.22630024e+00 1.06993246e+00 -2.82017857e-01
-2.33284727e-01 -9.85250235e-01 -5.28582856e-02 7.38961577e-01
7.15981424e-01 -1.00421119e+00 9.61688697e-01 -1.18348765e+00
-4.52917457e-01 1.93901919e-02 -1.99729994e-01 6.32354617e-01
1.30229592e-01 -3.71268511e-01 -1.27229726e+00 -4.86328065e-01
-1.36899024e-01 -3.50359738e-01 9.34738100e-01 2.53734618e-01
4.17785645e-01 -2.55398154e-01 -1.57414764e-01 6.94345415e-01
9.27616000e-01 1.05566848e-02 3.71545643e-01 2.20034160e-02
7.19374716e-01 5.55572271e-01 2.25447714e-02 -1.32009506e-01
2.28961274e-01 6.92824960e-01 -2.51601618e-02 -1.64527506e-01
-5.97735822e-01 -4.22567546e-01 8.67531538e-01 1.43870211e+00
3.33053052e-01 -6.56941712e-01 -9.36656415e-01 4.94929463e-01
-1.57499778e+00 -7.86158323e-01 2.02187270e-01 2.22597742e+00
9.57879663e-01 2.03700989e-01 4.35975306e-02 -2.34341756e-01
6.01361811e-01 1.37475446e-01 -4.43825752e-01 -4.97155428e-01
-2.67726034e-01 4.12804574e-01 8.28396261e-01 9.61027563e-01
-8.64688814e-01 1.24577165e+00 6.86045265e+00 7.62310982e-01
-1.40517199e+00 5.43075919e-01 4.96007383e-01 -3.06210607e-01
-5.88938415e-01 8.07895064e-02 -9.97022510e-01 3.77366841e-01
1.78576529e+00 -3.52875113e-01 7.76127577e-01 4.92869258e-01
1.97026163e-01 3.43489915e-01 -1.45160127e+00 7.79324293e-01
1.85983300e-01 -1.24276054e+00 1.10677741e-01 1.52889136e-02
5.15144110e-01 8.99927080e-01 7.94917494e-02 5.44457734e-01
4.63261664e-01 -1.22287822e+00 8.75943244e-01 -4.12064092e-03
1.25226486e+00 -4.52098101e-01 6.08279347e-01 3.98067892e-01
-7.38669395e-01 -3.75922732e-02 -1.30507529e-01 3.56950872e-02
3.36638123e-01 1.26417831e-01 -1.01948512e+00 5.10541856e-01
1.57652885e-01 5.43879867e-01 -2.70375669e-01 3.87551725e-01
-1.74894661e-01 9.90149200e-01 -5.17035544e-01 2.23758638e-01
1.85748249e-01 9.92713347e-02 5.56726933e-01 1.55170476e+00
2.89094061e-01 -2.73678333e-01 -1.10072210e-01 5.56829691e-01
-4.98803586e-01 3.09173375e-01 -3.73197883e-01 -5.71804523e-01
4.58009183e-01 6.38635457e-01 -1.62280321e-01 -5.88584960e-01
-7.31041670e-01 1.13463104e+00 3.41891885e-01 5.36172867e-01
-4.85587835e-01 -1.89515293e-01 7.91529894e-01 8.69465619e-02
2.44213879e-01 -3.39630604e-01 -4.63285685e-01 -1.46835089e+00
-6.88338131e-02 -1.12021017e+00 1.88583821e-01 -6.06952608e-01
-1.07498217e+00 7.10211873e-01 -3.00179034e-01 -8.16354156e-01
-6.45269632e-01 -5.16647637e-01 -3.25262159e-01 1.41236424e+00
-1.55400813e+00 -1.20911157e+00 3.79914761e-01 4.33607221e-01
9.16375935e-01 -3.63923430e-01 7.64116466e-01 5.65277457e-01
-6.75796926e-01 9.67237413e-01 5.09407282e-01 1.81735933e-01
9.89360809e-01 -1.04433882e+00 9.21376109e-01 1.16700327e+00
5.81979215e-01 7.71877706e-01 5.26354730e-01 -5.61105907e-01
-1.53856802e+00 -8.92345965e-01 1.35227048e+00 -8.13026786e-01
8.06867898e-01 -7.07387924e-01 -8.43635380e-01 1.21569836e+00
6.02057934e-01 -2.88493007e-01 5.89010358e-01 3.94453496e-01
-3.46133620e-01 -3.44019383e-02 -6.04913175e-01 6.19739711e-01
1.05561852e+00 -1.06970179e+00 -7.84145057e-01 3.09706658e-01
1.24432504e+00 -4.21049595e-01 -5.81734598e-01 1.90568656e-01
2.37565085e-01 -3.71727765e-01 7.16095924e-01 -9.06548560e-01
2.74123520e-01 1.89258605e-01 -5.01307249e-01 -1.38772118e+00
-1.59066290e-01 -1.02034700e+00 9.14652199e-02 1.43006611e+00
8.82217646e-01 -5.07522941e-01 6.26410723e-01 4.54186827e-01
-5.68572760e-01 -2.36502632e-01 -1.14278722e+00 -7.46947289e-01
3.58006746e-01 -6.54744804e-01 3.82426709e-01 9.18910623e-01
3.36579800e-01 9.57399964e-01 -3.12750608e-01 1.13008037e-01
2.24081844e-01 -4.43540215e-01 7.57034957e-01 -6.69347644e-01
-5.95895171e-01 -5.16278386e-01 1.71370443e-03 -1.55906701e+00
3.49373132e-01 -1.46330321e+00 3.23962033e-01 -1.26147902e+00
5.97949289e-02 -2.51724690e-01 -1.98703691e-01 6.59418941e-01
-7.70590305e-02 -4.44998555e-02 2.96296775e-01 3.25505733e-01
-2.03733966e-01 6.15048647e-01 8.45876157e-01 -1.35810092e-01
-2.01576814e-01 -1.05139531e-01 -7.34140694e-01 2.24407449e-01
4.23157007e-01 -4.79015976e-01 -4.61139888e-01 -1.14749026e+00
-1.37435094e-01 8.45771730e-02 -2.01248690e-01 -7.28919864e-01
3.33654672e-01 6.18578792e-02 -2.50558387e-02 -2.12177023e-01
4.80139703e-01 -7.11490691e-01 -2.75331557e-01 3.01521532e-02
-7.47982979e-01 -2.35563833e-02 4.06822860e-01 4.57185417e-01
-2.16154322e-01 -9.94456559e-02 7.09547698e-01 1.54694885e-01
-4.51172620e-01 -5.20552993e-02 -4.34715480e-01 2.69966274e-01
2.32048795e-01 8.56746957e-02 -3.12971592e-01 -7.21575201e-01
-8.78314853e-01 7.96500146e-02 3.43584359e-01 6.41390145e-01
1.89083531e-01 -9.90172267e-01 -8.26344669e-01 6.83656573e-01
6.89958781e-02 -4.40436095e-01 -8.45544934e-02 9.58663464e-01
4.24652807e-02 7.80172467e-01 2.59934098e-01 -6.32315397e-01
-1.22438133e+00 2.72620440e-01 2.74852186e-01 -1.35701686e-01
-4.52832192e-01 9.46091473e-01 6.76477402e-02 -7.83272207e-01
4.45966780e-01 -5.00689745e-01 3.43234926e-01 -1.85674727e-01
2.86750048e-01 2.78259479e-02 4.92407560e-01 -8.86257827e-01
-3.03948969e-01 2.08275542e-01 -2.64611840e-01 -7.34223545e-01
1.15659297e+00 -5.41733861e-01 2.49178514e-01 3.50631922e-01
1.26263261e+00 2.45528132e-01 -1.01063287e+00 -5.96512437e-01
2.51069546e-01 -2.59443074e-02 1.91618547e-01 -9.13939357e-01
-7.82628715e-01 1.22037101e+00 2.61675000e-01 -2.26147845e-01
8.55472624e-01 2.18071893e-01 8.97139609e-01 4.99184579e-01
-8.23248476e-02 -1.08178020e+00 -8.48383680e-02 9.75238264e-01
6.53861463e-01 -1.28986454e+00 -4.43794578e-01 -2.25254148e-01
-6.86258435e-01 8.83316934e-01 2.07090557e-01 3.89599830e-01
3.73049378e-01 5.52291930e-01 3.23697478e-01 2.40617841e-01
-9.54569697e-01 -2.85156012e-01 4.79305714e-01 3.33709925e-01
8.07963371e-01 -2.31533684e-02 1.71741843e-01 5.06177843e-01
-2.65217036e-01 -1.49674982e-01 2.95509666e-01 8.01953971e-01
-4.06991392e-01 -1.39963639e+00 -1.38167709e-01 3.94805580e-01
-6.31217062e-01 -7.25731075e-01 -2.83905238e-01 5.75021684e-01
-3.63441288e-01 1.13961375e+00 3.13832223e-01 -4.11146283e-01
2.80261368e-01 5.93364835e-01 3.04054350e-01 -1.04188335e+00
-6.29448593e-01 4.37664121e-01 5.46745896e-01 -3.24290752e-01
-9.72578302e-02 -6.82446480e-01 -1.10743046e+00 -3.65670264e-01
-5.96332073e-01 5.00420369e-02 8.41885328e-01 1.20465302e+00
5.82209468e-01 4.68689710e-01 5.01558900e-01 -6.90002680e-01
-9.89103854e-01 -1.26081944e+00 -1.68210089e-01 2.87714690e-01
6.29450440e-01 -1.14091277e-01 -4.87670273e-01 3.22998017e-01] | [14.377996444702148, 7.128015995025635] |
fb77dad9-172c-4a35-8246-5fd128d6eac2 | ji-yu-zhu-ti-ti-shi-xue-xi-de-ling-yang-ben | null | null | https://aclanthology.org/2022.ccl-1.48 | https://aclanthology.org/2022.ccl-1.48.pdf | 基于主题提示学习的零样本立场检测方法(A Topic-based Prompt Learning Method for Zero-Shot Stance Detection) | “零样本立场检测目的是针对未知目标数据进行立场极性预测。一般而言,文本的立场表达是与所讨论的目标主题是紧密联系的。针对未知目标的立场检测,本文将立场表达划分为两种类型:一类在说话者面向不同的主题和讨论目标时表达相同的立场态度,称之为目标无关的表达;另一类在说话者面向特定主题和讨论目标时才表达相应的立场态度,本文称之为目标依赖的表达。对这两种表达进行区分,有效学习到目标无关的表达方式并忽略目标依赖的表达方式,有望强化模型的可迁移能力,使其更加适应零样本立场检测任务。据此,本文提出了一种基于主题提示学习的零样本立场检测方法。具体而言,受自监督学习的启发,本文为了零样本立场检测设置了一个代理任务框架。其中,代理任务通过掩盖上下文中的目标主题词生成辅助样本,并基于提示学习分别预测原样本和辅助样本的立场表达,随后判断原样本和辅助样本的立场表达是否一致,从而在无需人工标注的情况下判断样本的立场表达是否依赖于目标的代理标签。然后,将此代理标签提供给立场检测模型,对应学习可迁移的立场检测特征。在两个基准数据集上的大量实验表明,本文提出的方法在零样本立场检测任务中相比基线模型取得了更优的性能。” | ['Ruifeng Xu', 'Bin Liang', 'Zixiao Chen'] | null | null | null | null | ccl-2022-10 | ['stance-detection'] | ['natural-language-processing'] | [-7.06793725e-01 -1.08910787e+00 8.04058909e-01 2.17016280e-01
5.46010315e-01 -1.40819919e+00 3.30559686e-02 6.02229416e-01
1.88348874e-01 1.51442230e+00 1.48498222e-01 -2.03226045e-01
-3.66683751e-01 -1.19703507e+00 7.98188671e-02 -1.28747332e+00
-3.17240596e-01 1.60136974e+00 5.50615191e-01 -2.19347119e-01
5.66035867e-01 1.01602626e+00 -1.20807755e+00 2.96678483e-01
1.01524377e+00 1.19973052e+00 9.66719806e-01 4.20278281e-01
3.55818331e-01 4.89520311e-01 -1.41106129e+00 2.65989035e-01
-1.45696372e-01 1.13332503e-01 -3.88195425e-01 8.72698188e-01
8.18622291e-01 -5.08623719e-01 -1.13811231e+00 1.23397923e+00
9.27757859e-01 3.03489655e-01 7.46148348e-01 -6.37343228e-01
-7.37249792e-01 4.47795182e-01 1.68008283e-01 2.90025115e-01
-1.98098734e-01 2.83341140e-01 5.31926155e-01 -9.30858910e-01
2.82023966e-01 1.64167535e+00 1.47822845e+00 6.05654895e-01
-1.40662313e+00 -1.27012205e+00 1.91369668e-01 8.78366008e-02
-1.86767662e+00 4.39281702e-01 3.34063858e-01 -3.25692356e-01
9.89298880e-01 1.34032536e+00 1.55534816e+00 1.77035606e+00
5.09412766e-01 1.02989566e+00 1.66890001e+00 6.92900479e-01
4.82579440e-01 1.87096477e+00 -3.48355174e-01 2.25332975e-01
1.10502386e+00 -2.54108489e-01 1.58985168e-01 1.60673156e-01
2.31702119e-01 5.23869276e-01 -7.94617087e-02 5.32250822e-01
-9.98884022e-01 9.53672707e-01 2.62073368e-01 1.57359242e+00
-1.95889845e-01 -1.00029135e+00 -8.87305975e-01 5.91211140e-01
-1.04406014e-01 -1.50918057e-02 4.08539116e-01 1.15668273e+00
-1.65827528e-01 -3.23154271e-01 9.53494012e-01 1.41357112e+00
8.56627047e-01 3.83706242e-01 -1.96793333e-01 8.03180873e-01
1.34544742e+00 1.15083587e+00 1.26470160e+00 -7.33524382e-01
-5.15291393e-01 9.81473863e-01 9.75082591e-02 -1.78550828e+00
-1.04465890e+00 -1.34527707e+00 -9.53776300e-01 -9.82749462e-01
-5.21910004e-02 -6.94589674e-01 -7.89348662e-01 1.36630571e+00
1.91866800e-01 -3.46499205e-01 6.26350641e-01 1.02332449e+00
8.46118748e-01 1.70024514e+00 -3.34629178e-01 -2.60454893e-01
1.73112667e+00 -7.52710521e-01 -8.32082510e-01 -2.75036424e-01
-1.30248857e+00 -1.66366959e+00 7.16857195e-01 -4.34890181e-01
-8.04716170e-01 -3.61258507e-01 -2.21639419e+00 6.39673412e-01
-6.90342784e-01 -2.65605748e-01 1.10805213e+00 1.22815776e+00
-1.71283007e+00 4.04320776e-01 -4.59344387e-01 -8.91220391e-01
3.30942303e-01 3.83160770e-01 1.38829991e-01 -7.30289072e-02
-9.46184039e-01 1.20780468e+00 -7.09035695e-01 5.17972052e-01
-2.09211946e+00 -5.99627674e-01 -1.07312337e-01 -2.91385174e-01
8.74190688e-01 -9.16995108e-01 1.22700953e+00 -8.17115366e-01
-7.79327273e-01 7.49263823e-01 -9.52166557e-01 5.88075966e-02
7.04411983e-01 1.00479662e+00 -4.76552039e-01 2.77556390e-01
-2.49928162e-01 1.10873662e-01 6.99610353e-01 -2.62918282e+00
-7.06414878e-01 -1.77913833e+00 1.56803891e-01 -3.20425183e-01
5.99267185e-01 7.97123015e-01 3.65546703e-01 -7.57637471e-02
2.26361468e-01 -1.64960742e+00 -1.31076932e-01 -1.57691801e+00
-5.87466478e-01 5.77551723e-02 1.94698584e+00 1.50283694e-01
1.37453938e+00 -1.90228760e+00 -8.54475498e-01 8.73449326e-01
5.06585360e-01 1.08218944e+00 1.48737922e-01 2.04258180e+00
5.13940990e-01 9.98849392e-01 -1.16320717e+00 5.33209801e-01
-2.93284982e-01 1.82917356e-01 1.10613823e+00 1.58524305e-01
-1.97042316e-01 -1.62927061e-01 -2.74775177e-01 2.90277183e-01
6.94985092e-01 8.29495370e-01 7.49115407e-01 2.11502999e-01
8.19362164e-01 3.65222365e-01 -2.12029949e-01 1.34664381e+00
1.41056514e+00 -6.07907474e-01 8.32800508e-01 3.56748819e-01
-3.78885657e-01 -8.31795871e-01 -2.68037629e+00 4.30696160e-01
-1.65574491e-01 6.81649685e-01 8.97270620e-01 -9.62260783e-01
1.49793589e+00 1.28730834e+00 7.58523524e-01 3.70803140e-02
2.14491829e-01 8.78150523e-01 4.70471680e-01 -6.89162910e-01
5.86067557e-01 -7.51993954e-01 8.81665230e-01 1.19097106e-01
-8.50939095e-01 -1.38041928e-01 6.08400106e-01 -2.50410557e-01
1.17597616e+00 -1.48847973e+00 -3.01454663e-01 -1.11121178e+00
1.46928215e+00 -3.93660218e-01 2.39155993e-01 3.27198327e-01
-6.88285232e-01 3.82523447e-01 -8.03176045e-01 4.08207506e-01
-5.21119297e-01 -1.72076094e+00 -4.88270223e-01 4.45575058e-01
-2.92118430e-01 -9.90780652e-01 -9.40735102e-01 -4.23691273e-02
-1.27210781e-01 4.17033672e-01 1.43491700e-01 2.29254812e-01
-6.42944098e-01 -1.45766485e+00 -6.00308716e-01 -2.96930224e-01
1.24709702e+00 -1.58499920e+00 -3.17480803e-01 8.99288803e-02
-3.59124064e-01 -7.74668157e-01 -5.92275977e-01 -3.03381503e-01
-1.81647265e+00 -1.47260094e+00 1.12378318e-02 -1.70224130e+00
1.60569108e+00 2.62154043e-01 7.22167134e-01 7.92063475e-01
-7.52453744e-01 1.14990306e+00 4.76611733e-01 -1.02962244e+00
2.34539017e-01 -4.67611462e-01 1.67201832e-01 1.15454972e+00
6.91514254e-01 -6.25821650e-01 -1.28544676e+00 1.09368455e+00
-1.60706735e+00 -8.33535910e-01 2.56825387e-01 6.56283975e-01
8.02945793e-01 3.34597409e-01 -1.09903544e-01 -6.30412996e-01
1.12040484e+00 6.60423994e-01 -9.69523787e-01 7.09823251e-01
-1.17960501e+00 -9.66081023e-01 1.08166730e+00 -6.50535524e-01
-8.38410795e-01 -8.37972403e-01 -3.49202454e-01 2.30424821e-01
3.30027223e-01 -2.29040638e-01 -3.71485263e-01 2.30878800e-01
-4.95849520e-01 -1.51703227e-03 6.66119635e-01 -6.16165221e-01
-4.09157008e-01 1.55751109e+00 1.79217383e-01 -1.03997362e+00
4.10319865e-01 6.25348687e-01 1.08335428e-01 -5.71560264e-01
2.70990819e-01 -8.79095942e-02 -9.44992542e-01 7.11700499e-01
7.99248636e-01 -1.52533567e+00 -3.99336517e-01 1.14233053e+00
-2.48197421e-01 7.94798195e-01 -6.51014924e-01 1.67041790e+00
-1.72140298e-03 -1.58661813e-01 -4.41966474e-01 -8.40720952e-01
-3.59379053e-01 -2.79603148e+00 3.26710463e-01 2.16134524e+00
1.83513120e-01 -8.88158917e-01 -9.52190533e-02 5.84966421e-01
9.13058221e-01 1.07787093e-02 1.31851327e+00 -1.32315683e+00
-9.63392615e-01 -4.83109206e-01 3.28232795e-01 1.59930587e+00
6.40257120e-01 6.63784683e-01 -4.79974644e-03 -3.17894012e-01
1.07794553e-01 8.99382830e-01 5.97264826e-01 6.63111925e-01
1.61068320e-01 -6.51853144e-01 -2.59992450e-01 6.27430603e-02
2.57529879e+00 9.05499935e-01 8.45193028e-01 1.64399922e+00
-9.02888358e-01 1.59448177e-01 1.02976024e-01 4.19627249e-01
-1.80433746e-02 1.16320908e+00 -1.47107512e-01 1.13811159e+00
-9.68087435e-01 -1.38175204e-01 8.49937081e-01 1.47389483e+00
-1.19570438e-02 -6.30231872e-02 5.63855827e-01 6.39222145e-01
-2.49221492e+00 -1.92728353e+00 -2.17685103e-01 2.97941709e+00
9.81722549e-02 -9.37034786e-01 -1.21923976e-01 6.60494387e-01
1.73908257e+00 -3.66440177e-01 -1.65998012e-01 -5.37539423e-01
-6.59324154e-02 1.02705562e+00 4.46955860e-01 2.82072574e-01
-8.67404759e-01 3.55395824e-02 1.44021064e-01 1.03717434e+00
-1.36219001e+00 7.98375607e-02 -3.20140831e-02 -7.50491202e-01
-8.05512965e-01 2.50275522e-01 -1.00431871e+00 1.73130858e+00
1.04997241e+00 -7.12499797e-01 -1.26710013e-01 3.07696313e-01
9.80732739e-01 -1.57093465e-01 -9.59032536e-01 1.55203605e+00
3.29707384e-01 -1.42919874e+00 1.97783280e-02 2.96949446e-01
1.02048612e+00 -9.37188268e-02 -1.58008799e-01 6.57836318e-01
8.70431960e-02 -9.86837864e-01 3.42399985e-01 -3.89739573e-01
-4.52858359e-01 -1.25477910e+00 1.62050331e+00 6.11877162e-03
-1.73408890e+00 -8.02354336e-01 -1.55540675e-01 -2.34048799e-01
4.80483115e-01 3.02913010e-01 -6.90495789e-01 2.74762630e-01
4.83165622e-01 -2.25057416e-02 -1.31538558e+00 2.04303169e+00
-8.65894735e-01 -3.56222183e-01 -8.54610443e-01 -2.01023746e+00
1.13987684e+00 -8.33398461e-01 1.00228369e+00 1.16158855e+00
2.05456272e-01 5.93764126e-01 5.33797979e-01 4.60628241e-01
-4.41914827e-01 2.54955202e-01 -4.34146404e-01 4.85372692e-01
9.36819673e-01 2.02268744e+00 -1.29387617e+00 -7.11728990e-01
-2.54152063e-02 -1.14917386e+00 -4.79140311e-01 -4.37299371e-01
-5.22429585e-01 -7.11836398e-01 5.34150898e-01 2.28728488e-01
-1.55452594e-01 -6.30683482e-01 5.11573792e-01 -4.00073528e-01
1.24480657e-01 -1.09012580e+00 -2.03275427e-01 -6.19010568e-01
-1.42991996e+00 3.47783744e-01 -1.39358833e-01 -1.35207939e+00
1.06649542e+00 -1.22653008e+00 -5.12029111e-01 2.56931454e-01
-1.17739849e-01 -6.97029591e-01 -5.06862588e-02 1.21089017e+00
1.19881725e+00 -3.11958432e-01 6.27381563e-01 1.08370411e+00
-8.12302351e-01 5.66937208e-01 1.61453950e+00 4.59590554e-01
2.81998098e-01 -1.49415970e+00 -1.02756393e+00 1.51076898e-01
4.05269682e-01 2.66523212e-01 -7.46378303e-02 -1.82059050e-01
-2.12744713e+00 -8.65967512e-01 7.30919063e-01 -4.07354027e-01
6.03349268e-01 6.83792541e-03 -1.83594167e-01 1.28851473e+00
4.09558386e-01 -1.80039763e-01 6.94872975e-01 -3.86050373e-01
3.86260688e-01 -1.12432396e+00 -2.22815514e+00 1.24403667e+00
9.93809104e-01 2.28054062e-01 -6.50105655e-01 5.09394348e-01
-4.32645440e-01 -3.94956917e-01 -1.27802038e+00 9.45457757e-01
1.61268795e+00 -7.57816970e-01 8.62184823e-01 -9.99835730e-01
-6.62739336e-01 -2.82312244e-01 -1.06005299e+00 -1.47839856e+00
1.95848137e-01 -4.40165460e-01 1.14589942e+00 1.50628340e+00
1.08968556e+00 -1.24035013e+00 1.85545951e-01 1.25363195e+00
-5.95372319e-01 -7.26777792e-01 -2.04216695e+00 -1.37048364e+00
-2.99370468e-01 4.44351472e-02 3.18264574e-01 -1.72381420e-02
-5.56351483e-01 6.09513700e-01 -2.80484390e-02 3.15252066e-01
3.35529864e-01 -4.16046202e-01 4.33819443e-01 -1.08449435e+00
1.12855649e+00 -8.99697304e-01 -4.50777084e-01 -3.49748768e-02
-8.83300364e-01 -3.13676775e-01 -1.41513216e+00 -2.18834758e+00
2.16387972e-01 -5.22870421e-01 -2.61762857e-01 -9.02676404e-01
1.59328771e+00 3.72191101e-01 7.23765612e-01 7.72000670e-01
8.33853260e-02 -5.86936712e-01 1.58130002e+00 -1.15964496e+00
-3.93846095e-01 6.86477244e-01 -2.09745169e-01 1.06782603e+00
5.64309001e-01 -7.01347291e-02 -2.48644724e-01 -6.74585342e-01
-6.69091940e-02 -1.04403114e+00 -1.09418726e+00 -1.25852191e+00
6.78034425e-01 2.82663405e-01 2.31421381e-01 -2.04480201e-01
5.69889069e-01 -1.00354934e+00 7.06211448e-01 6.35056078e-01
6.37193799e-01 7.93910027e-01 -2.51930863e-01 -3.69799107e-01
-8.05858448e-02 -1.83761978e+00 9.34687972e-01 -6.34094119e-01
-5.85751057e-01 -2.40747869e-01 -7.85652041e-01 -2.63145119e-01
1.86261976e+00 -2.25230694e-01 -6.40505016e-01 7.60775268e-01
-1.15875101e+00 -5.41969895e-01 3.13222319e-01 1.39802253e+00
1.38288844e+00 -1.11032379e+00 -2.08549929e+00 1.30403325e-01
-1.01980424e+00 7.55377579e-03 6.53423429e-01 9.95106161e-01
-7.07075000e-01 4.90687102e-01 -4.23776418e-01 -7.64345944e-01
-2.59251499e+00 8.37341428e-01 -2.24748909e-01 -5.58445215e-01
-4.29184824e-01 6.47871614e-01 -1.15873539e+00 1.12436652e-01
1.73670679e-01 -6.73775524e-02 -7.89248824e-01 3.64725858e-01
1.08490765e+00 9.47296858e-01 1.81761473e-01 -1.23583686e+00
-9.68881965e-01 2.01616788e+00 -5.20797074e-01 -9.69424546e-02
4.83747214e-01 -6.85269177e-01 -6.02417648e-01 -4.51310813e-01
9.46350574e-01 1.45434201e+00 4.08623442e-02 7.74551928e-01
-6.87949181e-01 -6.68098271e-01 -8.71359706e-01 -1.71490240e+00
-9.59951162e-01 1.96852997e-01 1.07658935e+00 -4.16779250e-01
8.04171085e-01 -6.02589428e-01 1.33648586e+00 3.58571857e-01
1.64415801e+00 -2.09922647e+00 -8.93137097e-01 5.38548589e-01
1.18846142e+00 -1.58683348e+00 2.99900353e-01 -4.33759034e-01
1.34319469e-01 1.63284588e+00 6.38313711e-01 -4.69327569e-01
1.41806626e+00 1.53054968e-01 -3.10034901e-01 7.39227310e-02
-1.19555801e-01 -3.32935631e-01 -4.69304532e-01 1.17544627e+00
5.72928488e-01 1.81702271e-01 -1.53108871e+00 3.92393291e-01
-1.08862889e+00 3.84349972e-01 1.50932252e+00 8.93736064e-01
-7.80243158e-01 -6.47310793e-01 -1.42175329e+00 -2.91327596e-01
-7.04716325e-01 6.52245462e-01 2.82352567e-01 9.33980823e-01
6.82008386e-01 2.72446823e+00 -6.41783118e-01 -9.54552591e-01
1.92570329e+00 -2.02808514e-01 9.78820562e-01 -1.04227468e-01
-1.39255548e+00 -4.48493063e-01 2.01067448e-01 9.19879973e-01
-7.50622153e-01 -2.34311387e-01 -7.53527641e-01 -1.17160761e+00
4.13362771e-01 2.14052105e+00 1.70877910e+00 4.01733696e-01
3.09423264e-02 1.08934450e+00 1.41014695e+00 -4.00307178e-01
-1.31948566e+00 -9.97947335e-01 -9.18803751e-01 -3.27492654e-01
-3.33579816e-02 -1.03700113e+00 -1.44769287e+00 -1.03474545e+00] | [-3.316145896911621, 6.907724857330322] |
ad1dbe7e-1c12-4875-a72e-227cdd8d3793 | erasing-appearance-preservation-in | null | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4677_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123510052.pdf | Erasing Appearance Preservation in Optimization-based Smoothing | Optimization-based image smoothing is routinely formulated as the game between a smoothing energy and an appearance preservation energy. Achieving adequate smoothing is a fundamental goal of these image smoothing algorithms. We show that partially ""erasing"" the appearance preservation facilitate adequate image smoothing. In this paper, we call this manipulation as Erasing Appearance Preservation (EAP). We conduct an user study, allowing users to indicate the ""erasing"" positions by drawing scribbles interactively, to verify the correctness and effectiveness of EAP. We observe the characteristics of human-indicated ""erasing"" positions, and then formulate a simple and effective 0-1 knapsack to automatically synthesize the ""erasing"" positions. We test our synthesized erasing positions in a majority of image smoothing methods. Experimental results and large-scale perceptual human judgments show that the EAP solution tends to encourage the pattern separation or elimination capabilities of image smoothing algorithms. We further study the performance of the EAP solution in many image decomposition problems to decompose textures, shadows, and the challenging specular reflections. We also present examinations of diversiform image manipulation applications like texture removal, retexturing, intrinsic decomposition, layer extraction, recoloring, material manipulation, etc. Due to the widespread applicability of image smoothing, the EAP is also likely to be used in more image editing applications. | ['Tien-Tsin Wong', 'Yi Ji', 'Lvmin Zhang', 'Chunping Liu', 'Chengze Li'] | null | null | null | null | eccv-2020-8 | ['image-smoothing'] | ['computer-vision'] | [ 7.12764382e-01 -1.26579928e-03 4.19416398e-01 -8.53325054e-02
-1.18941769e-01 -3.31788838e-01 5.04711211e-01 -1.77006394e-01
-1.62867233e-01 5.22694588e-01 8.51750094e-03 -3.72971088e-01
-5.00720218e-02 -5.27023494e-01 -4.75352407e-01 -1.01969397e+00
8.57744962e-02 -2.29187891e-01 3.63774657e-01 -3.97288442e-01
6.00153387e-01 7.60203004e-01 -1.86033452e+00 3.10903668e-01
1.17186499e+00 7.39413559e-01 5.93230784e-01 9.34201479e-01
-3.74762654e-01 4.09636825e-01 -7.38766670e-01 -2.94264853e-01
3.26636255e-01 -2.80918479e-01 -3.16541106e-01 7.25552797e-01
6.46494448e-01 -2.71555960e-01 2.37864181e-01 1.38980782e+00
3.37930620e-01 3.76792580e-01 5.76929152e-01 -1.14578843e+00
-7.76515186e-01 7.33582824e-02 -1.05429876e+00 -1.87104061e-01
3.89368623e-01 1.21960960e-01 3.07653785e-01 -7.37970829e-01
6.33832633e-01 1.27129281e+00 5.98974645e-01 3.79406750e-01
-1.40200174e+00 -3.23011726e-01 3.38761508e-02 -1.77576020e-01
-1.20776999e+00 -5.30852735e-01 9.59522545e-01 -3.69792104e-01
5.32635212e-01 8.99578273e-01 4.70966727e-01 5.15299976e-01
4.37231660e-01 6.83066189e-01 1.57572293e+00 -8.52887213e-01
7.55097345e-02 5.49717009e-01 2.46130332e-01 8.88229072e-01
1.66982248e-01 2.03806967e-01 -3.08051318e-01 -1.54157817e-01
1.01662326e+00 -3.02389890e-01 -5.68148494e-01 -2.43722677e-01
-1.04435921e+00 3.61392707e-01 8.35476369e-02 2.03738343e-02
-2.91721463e-01 -5.76429665e-02 4.55350690e-02 1.94256872e-01
6.41192317e-01 5.92338443e-01 -9.27153900e-02 2.64323473e-01
-8.44646513e-01 2.68343627e-01 6.69347167e-01 9.12068665e-01
7.45193720e-01 2.43823096e-01 -1.41393945e-01 1.29803264e+00
-2.97209732e-02 7.37173319e-01 1.14429541e-01 -1.15091956e+00
-5.76812178e-02 3.74666035e-01 5.91709197e-01 -1.15559566e+00
-3.34173858e-01 7.01019838e-02 -8.76461089e-01 1.11794746e+00
4.88101602e-01 -1.33601561e-01 -1.10120356e+00 1.37498617e+00
3.54772538e-01 -2.60982931e-01 -8.57546926e-02 9.05640543e-01
6.47244453e-01 7.80057490e-01 -2.61668377e-02 -4.14080381e-01
1.61497498e+00 -8.86390507e-01 -1.00323272e+00 6.66678026e-02
1.71898440e-01 -1.52250218e+00 1.47869837e+00 6.87545180e-01
-1.30626142e+00 -6.74266875e-01 -8.28665674e-01 3.96036282e-02
-3.04096878e-01 5.53383410e-01 7.70388365e-01 9.31405008e-01
-1.15732145e+00 6.64829731e-01 -4.71281707e-01 -2.69279540e-01
-1.21883735e-01 1.20862104e-01 -2.03925252e-01 3.33455145e-01
-5.11559308e-01 9.27981138e-01 -1.25668108e-01 2.21870646e-01
-8.78838450e-02 -6.90864503e-01 -6.91298366e-01 -4.07981947e-02
3.75311852e-01 -4.44403976e-01 7.90467560e-01 -1.37895942e+00
-1.75088704e+00 1.14738071e+00 -4.49414968e-01 -1.05862498e-01
4.54097807e-01 -1.62769273e-01 -6.87297404e-01 -1.03201203e-01
-3.73080939e-01 5.05729854e-01 1.49407125e+00 -1.87832630e+00
-4.72993195e-01 4.94677760e-02 -4.72065695e-02 2.46895924e-01
-1.04020789e-01 1.95272118e-01 -3.94803554e-01 -1.09929216e+00
1.43188730e-01 -8.32925797e-01 -1.94614261e-01 1.95726588e-01
-6.23888016e-01 1.42229080e-01 1.03793657e+00 -9.75447655e-01
1.31993008e+00 -2.31112742e+00 -9.21393838e-03 5.35761535e-01
1.59604579e-01 2.60294676e-01 -2.46758744e-01 3.65477681e-01
-2.51910210e-01 -1.04056068e-01 -2.42441341e-01 -3.46414387e-01
-3.06572206e-02 -2.71883584e-03 -3.42760503e-01 2.87522405e-01
-1.04403988e-01 3.85453910e-01 -4.78923738e-01 -3.59985173e-01
4.93857056e-01 3.96770895e-01 -4.54763770e-01 1.13649540e-01
-7.74187446e-02 7.55422562e-02 2.39475816e-02 8.78927231e-01
1.16109419e+00 -2.19738185e-02 -4.80839089e-02 -7.74115801e-01
-5.16596675e-01 -4.62817162e-01 -1.38920021e+00 9.76357102e-01
-3.30311805e-01 8.87467504e-01 5.77887416e-01 -3.43439698e-01
1.02030611e+00 -6.47000149e-02 1.90477580e-01 -6.16305292e-01
2.69514602e-03 5.97471278e-03 -2.11337477e-01 -5.69800675e-01
1.19677329e+00 -4.00320143e-02 4.63424176e-01 2.97198176e-01
-7.40021706e-01 -5.05715668e-01 2.48356476e-01 2.29670092e-01
6.16284311e-01 1.21797256e-01 3.71597737e-01 -7.59914756e-01
6.04385018e-01 -1.93226449e-02 1.22846924e-01 7.69828975e-01
-4.26752232e-02 7.65091658e-01 5.32720983e-01 -2.67870903e-01
-1.11274290e+00 -1.00938940e+00 -1.81006491e-01 1.02362490e+00
4.59838986e-01 -1.15562916e-01 -1.03532219e+00 -2.15997979e-01
-8.03425536e-02 8.75223219e-01 -5.22082448e-01 2.86438018e-01
-5.21295488e-01 -7.80042946e-01 3.95786352e-02 1.53000534e-01
6.48107708e-01 -1.26449335e+00 -6.00577652e-01 4.82804850e-02
-1.35765905e-02 -8.26808989e-01 -7.87336767e-01 -1.10819675e-01
-5.88103235e-01 -9.36628759e-01 -8.89166832e-01 -7.42095172e-01
1.25875521e+00 6.44454002e-01 9.69463408e-01 5.29971659e-01
-4.92347211e-01 5.31147420e-01 -3.35019141e-01 -3.22171509e-01
-6.05145752e-01 -6.05295956e-01 -1.74905226e-01 1.11996107e-01
-1.45801410e-01 -2.71511555e-01 -7.08840191e-01 6.57617390e-01
-1.15287471e+00 5.69905758e-01 4.12672102e-01 6.15197182e-01
5.24173737e-01 4.40656275e-01 -1.64625682e-02 -9.53014731e-01
1.03341317e+00 3.70344132e-01 -8.20083320e-01 4.49191391e-01
-4.76782918e-01 1.16049930e-01 3.84608179e-01 -7.17451334e-01
-1.62259495e+00 -1.92298606e-01 -2.77043544e-02 -2.38694444e-01
-2.84717560e-01 -1.43787749e-02 -6.97694123e-02 -4.63672370e-01
6.24229193e-01 2.10949197e-01 2.14199811e-01 -3.57136667e-01
4.09779876e-01 4.10954237e-01 7.32637823e-01 -7.03917921e-01
8.20760667e-01 6.81477487e-01 -1.68623760e-01 -1.34756279e+00
-3.51698339e-01 -1.73286825e-01 -6.70175552e-02 -4.46858317e-01
7.58012414e-01 -2.41542637e-01 -8.88535678e-01 7.64522254e-01
-1.15973675e+00 -6.75457478e-01 -2.12536767e-01 1.76590420e-02
-3.31783295e-01 8.79779994e-01 -3.94130796e-01 -9.37840044e-01
-1.71198800e-01 -1.27412510e+00 1.00745571e+00 4.95781898e-01
-2.85402209e-01 -1.01829433e+00 -2.30178505e-01 2.10678130e-01
5.21554708e-01 2.10968837e-01 9.73101676e-01 3.56556833e-01
-5.55764377e-01 9.18497741e-02 -5.39147198e-01 3.75077724e-01
3.10393184e-01 5.36015689e-01 -8.67249250e-01 -1.73949346e-01
-6.50106072e-02 3.86443049e-01 7.50751138e-01 6.51808560e-01
1.29375589e+00 -2.90053904e-01 -7.60220662e-02 7.11870551e-01
1.38394284e+00 4.19502407e-01 1.38532126e+00 3.07482988e-01
3.97988111e-01 8.78439844e-01 5.15071630e-01 3.15307349e-01
-2.33380660e-01 7.28180945e-01 1.52919784e-01 -7.61074305e-01
-5.28817117e-01 2.16695637e-01 3.38875979e-01 4.69519079e-01
-4.70891595e-01 -3.19437504e-01 -3.03189397e-01 1.40427142e-01
-1.38857460e+00 -9.19138789e-01 -4.52882290e-01 2.19944406e+00
7.49510407e-01 4.45628446e-03 -1.76192224e-01 -7.19061717e-02
9.38592196e-01 1.65677354e-01 -3.03467568e-02 -8.46472383e-01
-2.94995427e-01 1.99703306e-01 7.60778308e-01 9.71500516e-01
-9.42422211e-01 9.19649541e-01 6.52193832e+00 1.24194813e+00
-1.13273990e+00 -3.18307042e-01 6.34000957e-01 3.90459299e-01
-5.39595544e-01 3.77658159e-02 -6.72760129e-01 6.19547546e-01
-8.61151889e-02 1.76455215e-01 6.18837655e-01 5.30585766e-01
5.23150265e-01 -8.00132036e-01 -5.85764885e-01 9.49576437e-01
1.46282492e-02 -1.27252614e+00 1.40844151e-01 -1.15345754e-01
8.10458720e-01 -9.05502260e-01 1.60302907e-01 -2.61977077e-01
8.13746825e-02 -8.49935889e-01 7.23824680e-01 7.01522768e-01
7.51891077e-01 -3.06977689e-01 1.95956782e-01 -8.43099281e-02
-1.21163535e+00 2.62599707e-01 -3.58664304e-01 2.44291246e-01
4.02162164e-01 7.02653348e-01 -4.60203707e-01 2.59531289e-01
4.36278641e-01 -6.54889345e-02 -4.68074679e-01 1.03456140e+00
-1.94202170e-01 2.50286460e-02 -3.09383482e-01 -9.11346674e-02
-1.53316349e-01 -8.55497599e-01 7.77735949e-01 1.44633460e+00
3.69898587e-01 4.64510024e-01 -1.43104658e-01 9.53170717e-01
4.00014251e-01 1.20338000e-01 -2.63523310e-01 1.84833497e-01
1.77484468e-01 1.36431479e+00 -1.23902202e+00 -4.32035238e-01
-2.40460545e-01 1.26086688e+00 -2.71824956e-01 7.55497396e-01
-8.62027705e-01 -5.21025836e-01 5.17652035e-01 2.85859019e-01
6.01945743e-02 -1.53748155e-01 -6.53231859e-01 -9.33098018e-01
-1.47724718e-01 -1.05456591e+00 -7.20436051e-02 -1.51029456e+00
-1.05350089e+00 5.63856006e-01 5.63075542e-02 -1.08052063e+00
4.58073825e-01 -7.64721096e-01 -8.79102290e-01 9.44039822e-01
-1.27126563e+00 -8.46444905e-01 -6.65483236e-01 4.83692676e-01
7.25808322e-01 -3.65896802e-03 4.58064735e-01 7.20241889e-02
-3.67134511e-01 3.27364266e-01 3.38953324e-02 -4.36531633e-01
7.89956093e-01 -1.22092009e+00 1.47672832e-01 1.03933334e+00
-2.00010598e-01 6.85618877e-01 1.30452383e+00 -8.09606850e-01
-1.20356679e+00 -3.77359748e-01 5.40339947e-01 8.36688373e-03
4.21912909e-01 -1.06861062e-01 -1.08773386e+00 2.47692749e-01
3.96108210e-01 -6.39777362e-01 1.02065541e-01 -8.49661976e-02
-8.95045921e-02 -1.81964755e-01 -1.17071474e+00 1.13640225e+00
8.60980451e-01 -1.34427533e-01 -4.00873095e-01 1.84600368e-01
1.50034055e-01 -5.97146571e-01 -5.25339007e-01 3.73099416e-01
6.99727952e-01 -1.31529629e+00 1.23679388e+00 -4.25431915e-02
4.03592825e-01 -4.69070822e-01 1.47203803e-01 -1.31613684e+00
-3.09067428e-01 -8.93449605e-01 3.96853775e-01 9.81348753e-01
2.95222819e-01 -7.84134507e-01 7.12887108e-01 8.58667791e-01
-1.68467417e-01 -2.82102257e-01 -3.06564152e-01 -4.94081408e-01
-4.71765101e-01 8.63933237e-04 2.54781634e-01 9.31442618e-01
-1.86071724e-01 -3.50028247e-01 -6.76782787e-01 3.56244087e-01
9.95419621e-01 2.32495084e-01 1.00675654e+00 -7.82403469e-01
-4.50440168e-01 -7.04080880e-01 2.02932358e-01 -1.17944705e+00
-2.97586769e-01 -2.05788210e-01 5.39282672e-02 -1.29782307e+00
8.59022364e-02 -5.93796134e-01 2.05330834e-01 2.00401798e-01
-3.69749069e-01 1.78749308e-01 2.35446125e-01 -2.81751249e-02
-3.47669125e-02 1.08474925e-01 1.58714652e+00 1.40370607e-01
-5.19633830e-01 -4.53186408e-02 -5.71666896e-01 1.07397175e+00
7.36093581e-01 7.58544579e-02 -4.16405529e-01 -2.75998592e-01
1.46165177e-01 -3.60605642e-02 5.48125565e-01 -5.95595479e-01
-6.62270701e-03 -5.44727981e-01 3.41548562e-01 -5.45214117e-01
4.05735910e-01 -7.64476717e-01 4.93516058e-01 3.57169271e-01
-1.84225291e-01 -8.94615725e-02 2.72255123e-01 1.40682042e-01
1.26217663e-01 -3.91430557e-01 1.07739782e+00 -2.12058127e-02
-1.02308393e+00 -2.74465472e-01 -4.12218004e-01 -4.65684086e-01
7.78820515e-01 -6.75690114e-01 -5.33582449e-01 -5.23886204e-01
-7.50558734e-01 -1.78724006e-01 7.19478607e-01 7.05244914e-02
6.80220544e-01 -9.72013295e-01 -4.30057257e-01 5.06446958e-01
-3.69325072e-01 -5.37886322e-01 4.84142989e-01 9.17480111e-01
-1.01229930e+00 -3.66910487e-01 -2.29325920e-01 -3.50447029e-01
-1.87900841e+00 5.08842468e-01 2.29989931e-01 2.46047298e-03
-8.49254429e-01 7.20948815e-01 6.95930958e-01 2.44276613e-01
1.53897598e-01 -3.46059293e-01 1.09437607e-01 -2.58518517e-01
5.79916120e-01 6.81322455e-01 -1.12869307e-01 -5.44795580e-02
9.13006514e-02 6.58811152e-01 1.22839864e-02 -2.22871099e-02
8.29721808e-01 -4.21473742e-01 -3.44748557e-01 -1.51570141e-01
6.81104541e-01 7.44540632e-01 -1.47614253e+00 2.10379019e-01
-3.55450034e-01 -8.72859478e-01 -1.59837008e-01 -7.84447670e-01
-1.00434470e+00 4.29188013e-01 5.32951534e-01 6.72837138e-01
1.41583991e+00 -3.59242916e-01 5.10762274e-01 1.27112597e-01
4.90609705e-02 -1.22867417e+00 -4.01911847e-02 2.28603799e-02
1.27360487e+00 -1.01404703e+00 3.74888003e-01 -1.00788856e+00
-7.71273077e-01 1.14884984e+00 4.48143512e-01 -1.39802888e-01
4.94928569e-01 8.17992806e-01 2.80495524e-01 -8.77891481e-02
-2.45429203e-01 -1.03681967e-01 4.33327675e-01 6.09497130e-01
2.99238741e-01 1.67525500e-01 -6.74414635e-01 3.39875892e-02
-9.81131494e-02 -2.96646506e-01 6.38773441e-01 6.10585213e-01
-8.84588540e-01 -1.11335170e+00 -1.01453781e+00 2.25394756e-01
-9.43361297e-02 -2.43519470e-01 -1.94570512e-01 9.02253628e-01
1.80158056e-02 8.08922946e-01 -3.88199948e-02 4.69749384e-02
3.57300013e-01 -2.28083402e-01 5.41214585e-01 -1.83568448e-02
-4.34551656e-01 5.44197321e-01 3.35572153e-01 -4.35673505e-01
-4.85658824e-01 -3.32526565e-01 -9.31636274e-01 -6.48610711e-01
-3.67850095e-01 -9.28331241e-02 7.31893778e-01 5.25235176e-01
9.29923076e-03 4.89351541e-01 3.86429697e-01 -1.02334440e+00
-1.85264796e-01 -5.83929241e-01 -7.96039999e-01 8.64318252e-01
1.39557868e-01 -5.51927805e-01 -5.68611741e-01 5.67333877e-01] | [11.077857971191406, -2.3587615489959717] |
68f80bcf-0789-4d78-9862-d54bf48b3795 | superpixelgraph-semi-automatic-generation-of | 2304.05661 | null | https://arxiv.org/abs/2304.05661v2 | https://arxiv.org/pdf/2304.05661v2.pdf | SuperpixelGraph: Semi-automatic generation of building footprint through semantic-sensitive superpixel and neural graph networks | Most urban applications necessitate building footprints in the form of concise vector graphics with sharp boundaries rather than pixel-wise raster images. This need contrasts with the majority of existing methods, which typically generate over-smoothed footprint polygons. Editing these automatically produced polygons can be inefficient, if not more time-consuming than manual digitization. This paper introduces a semi-automatic approach for building footprint extraction through semantically-sensitive superpixels and neural graph networks. Drawing inspiration from object-based classification techniques, we first learn to generate superpixels that are not only boundary-preserving but also semantically-sensitive. The superpixels respond exclusively to building boundaries rather than other natural objects, while simultaneously producing semantic segmentation of the buildings. These intermediate superpixel representations can be naturally considered as nodes within a graph. Consequently, graph neural networks are employed to model the global interactions among all superpixels and enhance the representativeness of node features for building segmentation. Classical approaches are utilized to extract and regularize boundaries for the vectorized building footprints. Utilizing minimal clicks and straightforward strokes, we efficiently accomplish accurate segmentation outcomes, eliminating the necessity for editing polygon vertices. Our proposed approach demonstrates superior precision and efficacy, as validated by experimental assessments on various public benchmark datasets. A significant improvement of 8% in AP50 was observed in vector graphics evaluation, surpassing established techniques. Additionally, we have devised an optimized and sophisticated pipeline for interactive editing, poised to further augment the overall quality of the results. | ['Qing Zhu', 'Zhendong Wang', 'Qisen Shang', 'Bo Xu', 'Han Hu', 'Haojia Yu'] | 2023-04-12 | null | null | null | null | ['superpixels'] | ['computer-vision'] | [ 6.85504377e-01 1.17994145e-01 1.47412494e-01 -2.65174210e-01
-5.95236838e-01 -5.81419468e-01 5.59384048e-01 5.24108708e-01
-3.58860761e-01 6.21186554e-01 -1.45977838e-02 -1.86112761e-01
4.81022112e-02 -1.31056321e+00 -6.15213215e-01 -5.18040657e-01
-9.83636975e-02 1.44018725e-01 3.65796924e-01 -2.25882262e-01
4.80854720e-01 7.77644038e-01 -1.57131159e+00 4.31654491e-02
1.44321489e+00 8.87221992e-01 8.24567825e-02 6.95435226e-01
-4.01312768e-01 1.49083033e-01 -6.78736746e-01 -2.28405103e-01
1.86717317e-01 -1.12731129e-01 -4.44837540e-01 2.84318835e-01
6.09835327e-01 -2.92910069e-01 6.15611635e-02 8.49488974e-01
3.70117962e-01 7.32444003e-02 8.09186876e-01 -9.24993515e-01
-7.08043873e-01 4.29667234e-01 -8.65313411e-01 -1.38347089e-01
1.13644391e-01 1.56349435e-01 1.05247903e+00 -8.32841694e-01
4.69471157e-01 9.45299089e-01 8.50587547e-01 5.19458316e-02
-1.44277155e+00 -4.86019045e-01 2.50533432e-01 -1.74414679e-01
-1.71460831e+00 -3.80362682e-02 1.04424155e+00 -5.30465841e-01
7.33048141e-01 8.78911376e-01 8.98936033e-01 6.21716261e-01
-1.66249320e-01 7.63988793e-01 8.33410501e-01 -3.17634761e-01
1.40910476e-01 3.72440964e-02 1.71511456e-01 7.44921923e-01
3.13592434e-01 -1.65658504e-01 -2.37718627e-01 2.26997174e-02
9.41022217e-01 -3.62607278e-02 -3.62430662e-01 -4.04320955e-01
-9.92427945e-01 6.91715240e-01 8.09934974e-01 2.33337268e-01
-5.93907833e-01 2.89851040e-01 1.93752393e-01 -4.13358033e-01
6.43519461e-01 3.77552271e-01 5.34160733e-02 1.98187441e-01
-1.33319497e+00 1.75718755e-01 3.20990950e-01 9.42206264e-01
8.97918105e-01 2.55705237e-01 -5.86682081e-01 7.72804797e-01
1.99256614e-01 3.61813098e-01 -5.76988123e-02 -8.87256086e-01
4.19098228e-01 1.13817012e+00 1.83436736e-01 -1.42282701e+00
-3.75968248e-01 -5.08312345e-01 -1.00794661e+00 3.53836000e-01
3.57857913e-01 1.22013755e-01 -1.14108574e+00 1.38425839e+00
4.55111057e-01 4.93857823e-02 -4.46110398e-01 5.89737058e-01
6.77428544e-01 6.20273173e-01 4.29051757e-01 9.56660211e-02
1.42052937e+00 -7.71033645e-01 -4.35854316e-01 -8.97436440e-02
4.74912822e-01 -5.17422497e-01 1.35533762e+00 2.96692938e-01
-8.37825596e-01 -4.49327499e-01 -1.19022119e+00 -1.25683844e-01
-6.54057264e-01 3.26694548e-01 8.09237361e-01 9.40541089e-01
-1.01850486e+00 6.14794850e-01 -6.60182893e-01 -1.82113960e-01
8.15114856e-01 3.92952055e-01 -1.59724616e-02 5.43989301e-01
-8.60191166e-01 4.45791513e-01 4.06636298e-01 2.76237607e-01
-4.02291119e-01 -9.83872890e-01 -7.96680331e-01 1.65298328e-01
8.68788883e-02 -6.20938003e-01 6.68221653e-01 -8.53518963e-01
-1.27714610e+00 7.30088532e-01 5.28155454e-02 -3.23192686e-01
7.62736559e-01 -1.10163338e-01 -9.41868722e-02 1.56994656e-01
1.49338737e-01 8.66641879e-01 7.90674210e-01 -1.52385235e+00
-9.38362062e-01 -2.48622298e-01 -1.10597856e-01 2.56904423e-01
-4.80385095e-01 -3.32816005e-01 -8.90993237e-01 -9.40025389e-01
2.41945505e-01 -6.21893585e-01 -5.01091480e-01 2.08546251e-01
-7.44092166e-01 -8.89395922e-02 8.72371793e-01 -8.48179877e-01
1.55598605e+00 -2.00788975e+00 -1.67538300e-01 5.50006866e-01
3.49288881e-01 2.63662040e-01 1.62169158e-01 8.05799142e-02
1.47982627e-01 4.14264888e-01 -7.33546019e-01 -2.62292802e-01
-1.40796512e-01 -1.13894112e-01 -2.85901576e-01 2.31899694e-01
2.58809328e-01 8.89034986e-01 -8.26458633e-01 -6.81635022e-01
6.56682432e-01 8.33584249e-01 -4.87358242e-01 -1.17585026e-01
-1.84234977e-01 2.97831953e-01 -5.61434269e-01 6.60173118e-01
7.81480551e-01 -1.00114428e-01 -7.95016438e-02 -3.71537119e-01
-2.95603544e-01 -2.53467172e-01 -1.31573749e+00 1.38402617e+00
-4.37403589e-01 5.70644796e-01 -2.19786823e-01 -8.25008035e-01
1.05420697e+00 -9.71624702e-02 5.37801087e-01 -7.51637459e-01
4.09541652e-03 7.16256872e-02 -7.38817990e-01 -1.03014261e-01
7.60160983e-01 4.10928845e-01 -1.42709330e-01 8.24646056e-02
-4.98193473e-01 -4.95797127e-01 2.57248014e-01 1.11143075e-01
5.95159471e-01 3.18336755e-01 1.87162668e-01 -2.70550042e-01
4.48257953e-01 2.15540692e-01 1.72901466e-01 6.90304875e-01
1.85299784e-01 7.90299296e-01 3.12020272e-01 -3.75842988e-01
-8.77669573e-01 -1.17094684e+00 -2.03324035e-01 9.13777113e-01
2.75425911e-01 -3.97667021e-01 -1.17844105e+00 -4.17376548e-01
-6.92405030e-02 9.28957522e-01 -5.81020594e-01 1.45215392e-01
-6.88797176e-01 -7.99413264e-01 7.14895964e-01 6.37122095e-01
6.90588713e-01 -7.78908253e-01 -8.60390246e-01 3.50735664e-01
1.55700706e-02 -8.87064457e-01 -4.59536403e-01 3.66525315e-02
-9.37357664e-01 -9.63341236e-01 -7.81911254e-01 -5.43823481e-01
9.76966500e-01 2.34350890e-01 9.07987058e-01 3.15967113e-01
-5.79710960e-01 9.54079777e-02 -6.41467348e-02 -3.28934222e-01
-6.48492426e-02 6.31784573e-02 -4.88259137e-01 2.21675232e-01
-4.30324078e-02 -5.88470817e-01 -9.24780130e-01 2.01163292e-01
-8.35681200e-01 5.56166410e-01 5.31447828e-01 5.09214461e-01
9.93156254e-01 7.59936646e-02 2.09543735e-01 -1.04061055e+00
5.92547297e-01 -1.26298875e-01 -7.51356781e-01 2.84176141e-01
-6.35834634e-01 -2.83864904e-02 7.05409944e-01 -4.19769846e-02
-1.15461373e+00 3.29063833e-01 -1.77927949e-02 7.40507916e-02
-2.48400286e-01 2.81129062e-01 -1.36017367e-01 -2.00033948e-01
7.43310094e-01 2.81000435e-01 -4.39975291e-01 -2.59424567e-01
6.66306794e-01 5.98040104e-01 6.98325038e-01 -5.21419168e-01
8.70728374e-01 7.15564907e-01 -2.44296566e-02 -1.26180434e+00
-2.95142770e-01 -3.69979978e-01 -6.78848505e-01 -3.98894042e-01
9.41024959e-01 -6.00058138e-01 -5.13218462e-01 5.60883403e-01
-1.06282139e+00 -4.05657351e-01 -2.08749309e-01 -9.54046473e-02
-2.78312981e-01 4.64186549e-01 -3.07070911e-01 -7.82131732e-01
-5.46125054e-01 -9.98670101e-01 1.24856544e+00 5.28259397e-01
-3.75350565e-01 -8.92718852e-01 -1.49944931e-01 3.46478522e-01
3.63592505e-01 7.84712017e-01 1.00798786e+00 -1.70885310e-01
-8.89292896e-01 -4.25511122e-01 -5.71351707e-01 1.32802457e-01
2.65908301e-01 5.15150368e-01 -1.14259577e+00 2.03893617e-01
-6.53562307e-01 2.99729258e-01 6.93260372e-01 4.78038341e-01
1.40577936e+00 -1.43799469e-01 -5.56894481e-01 7.87078559e-01
1.50682390e+00 8.96072835e-02 6.76529765e-01 3.34625334e-01
1.16220176e+00 5.87449968e-01 4.81227428e-01 5.05805492e-01
3.84831101e-01 5.88907421e-01 4.28533912e-01 -5.50315499e-01
-2.68101305e-01 -3.83646309e-01 -1.08364448e-01 2.89794385e-01
-4.62647855e-01 -2.51747161e-01 -9.50149536e-01 6.30251765e-01
-1.56077468e+00 -6.23021603e-01 -5.85752845e-01 2.18813419e+00
7.74203956e-01 2.08904848e-01 -1.20086018e-02 2.45879814e-01
7.08328187e-01 2.73746103e-01 -4.01501238e-01 -3.53275776e-01
-1.44050524e-01 4.57744002e-01 7.56045938e-01 5.84254265e-01
-1.26474273e+00 1.17027617e+00 5.86481714e+00 9.52324092e-01
-1.16911638e+00 -1.83326304e-01 8.24188292e-01 9.87273082e-02
-5.49378693e-01 -1.30890504e-01 -5.84603727e-01 3.04581404e-01
4.70774233e-01 7.74637610e-02 2.21014857e-01 8.72959852e-01
5.99007010e-01 -2.80427814e-01 -7.35203981e-01 8.24008584e-01
-2.86173105e-01 -1.58654964e+00 2.02659726e-01 -2.09704693e-02
8.11714828e-01 -3.17075491e-01 2.94549167e-02 4.21275012e-02
3.28198403e-01 -1.04514599e+00 8.61360908e-01 3.95083696e-01
8.21359038e-01 -8.22441876e-01 1.82810128e-01 1.02533795e-01
-1.64068115e+00 2.36073077e-01 -9.14676040e-02 1.25544608e-01
8.96816254e-02 5.02555668e-01 -7.16697872e-01 5.32447457e-01
6.25679135e-01 3.83905858e-01 -8.33470702e-01 1.12753129e+00
-3.33840430e-01 5.66245377e-01 -6.03161693e-01 -6.21212758e-02
4.46300745e-01 -4.98084038e-01 3.29832703e-01 1.56317031e+00
2.55526900e-01 -1.12970330e-01 5.37929940e-04 1.11371684e+00
5.58299497e-02 5.11354804e-01 -5.30716836e-01 3.30176875e-02
5.47389209e-01 1.35648203e+00 -1.38478720e+00 -3.77985656e-01
-1.96159199e-01 1.08198130e+00 2.88583219e-01 5.81009150e-01
-1.03307021e+00 -6.16045892e-01 3.38343203e-01 3.04607481e-01
2.72575140e-01 -3.05451512e-01 -9.44636941e-01 -4.91512179e-01
7.12198243e-02 -4.29351836e-01 2.13713899e-01 -6.01495981e-01
-5.87730587e-01 4.48230118e-01 -1.83981881e-02 -1.07126427e+00
2.14197144e-01 -3.66594613e-01 -8.40558350e-01 7.06795990e-01
-1.28590524e+00 -1.54275107e+00 -7.65076220e-01 2.23119855e-01
5.14371634e-01 5.50231755e-01 7.80146062e-01 2.64741927e-01
-7.56246984e-01 5.11163056e-01 2.80061793e-02 7.66069219e-02
5.72878420e-02 -1.33298588e+00 6.32560551e-01 1.02609706e+00
3.31449717e-01 4.64009881e-01 5.86488366e-01 -7.38255978e-01
-8.84963572e-01 -1.34335482e+00 6.45381510e-01 -2.10118756e-01
3.05676192e-01 -4.73998815e-01 -7.82085359e-01 3.18039447e-01
-6.40250444e-02 -2.48235792e-01 3.64174187e-01 -2.15321079e-01
-9.37200934e-02 -8.90685711e-03 -9.72625196e-01 1.00209260e+00
1.05081117e+00 -3.71664554e-01 -1.64626285e-01 2.75765896e-01
4.18234825e-01 -2.90699601e-01 -6.81306839e-01 3.75971973e-01
3.93634349e-01 -1.03019822e+00 1.32776093e+00 -1.29977465e-01
2.09848762e-01 -5.96680701e-01 -2.12824289e-02 -9.68110144e-01
-2.47765362e-01 -3.75179917e-01 1.38884246e-01 1.39474702e+00
4.03332800e-01 -2.91905612e-01 9.63048100e-01 7.11531818e-01
-2.42665440e-01 -7.76076198e-01 -6.84116721e-01 -4.81336087e-01
-3.42869043e-01 -6.26539528e-01 6.39780581e-01 8.21043611e-01
-5.39403379e-01 -7.42696375e-02 -2.42058448e-02 3.66667688e-01
7.20819235e-01 6.88939989e-02 7.02660263e-01 -1.20104170e+00
1.74249023e-01 -9.61273015e-01 -3.93544197e-01 -8.64136338e-01
-2.52787143e-01 -8.05085242e-01 -1.66227315e-02 -1.95708203e+00
-1.91829830e-01 -7.23777950e-01 -6.85030594e-02 4.32831675e-01
-4.21363860e-01 7.29010880e-01 -1.18209735e-01 4.07687910e-02
-2.12119237e-01 3.97194088e-01 1.21196651e+00 -3.34011555e-01
-6.14485979e-01 -6.05350807e-02 -7.75457382e-01 8.84058475e-01
9.12325025e-01 -1.78165108e-01 -5.61503172e-01 -3.94544691e-01
2.32330672e-02 -4.87957954e-01 5.35268486e-01 -1.00377047e+00
1.06494203e-02 -2.29802743e-01 4.05015975e-01 -7.72961080e-01
1.84831679e-01 -7.12124705e-01 4.21037108e-01 3.83140773e-01
2.79149599e-02 -4.77634519e-01 2.41843253e-01 5.94155610e-01
2.20474750e-02 -1.89770630e-03 7.76540995e-01 2.55565271e-02
-9.38343525e-01 2.48907968e-01 -2.87895650e-01 -1.25195786e-01
1.28900063e+00 -8.30478013e-01 -3.37010548e-02 -9.59881768e-02
-5.07358253e-01 2.89776791e-02 5.17919838e-01 9.28622633e-02
5.98594546e-01 -1.19663763e+00 -4.64116454e-01 2.45180324e-01
8.77201036e-02 4.05596524e-01 3.97160679e-01 5.98549664e-01
-1.17164087e+00 2.49592856e-01 -1.79436550e-01 -7.38499165e-01
-1.20544052e+00 2.51054555e-01 2.02765882e-01 -1.08081251e-01
-7.78873742e-01 9.56658483e-01 3.77653301e-01 -1.57870427e-01
2.23282889e-01 -7.45784223e-01 -3.24219614e-01 3.61351162e-01
2.13726759e-01 5.34485579e-01 -2.09250730e-02 -6.25108600e-01
-2.64530063e-01 7.96928227e-01 1.96577922e-01 2.23901987e-01
1.22928286e+00 -6.65162206e-02 1.31632760e-01 7.40217790e-02
7.74557650e-01 1.68853790e-01 -1.43171334e+00 8.14335942e-02
1.64054871e-01 -4.69670445e-01 1.80729717e-01 -5.06699204e-01
-1.09596944e+00 8.40358257e-01 6.40893638e-01 2.69475251e-01
1.05784249e+00 -3.08659792e-01 7.67852962e-01 -2.34598946e-02
2.92231888e-01 -1.34230614e+00 -1.11373976e-01 -6.07239567e-02
9.48191822e-01 -9.57342267e-01 1.31144717e-01 -8.80358219e-01
-4.36277568e-01 1.03644001e+00 5.05135417e-01 -1.44610733e-01
3.99374127e-01 2.68985927e-01 -2.44235992e-02 -3.13154161e-01
1.93529457e-01 -1.68155357e-01 4.73380893e-01 6.63434327e-01
2.81449348e-01 4.04972076e-01 -1.54120222e-01 5.36714673e-01
-1.74004495e-01 -2.10327387e-01 1.35729417e-01 8.43612134e-01
-5.76178491e-01 -7.65268505e-01 -4.28613663e-01 4.97725666e-01
3.00707985e-02 -2.83217013e-01 -4.00044978e-01 7.94272423e-01
1.12057306e-01 5.99295616e-01 2.47828767e-01 -1.97874755e-01
5.31619608e-01 -1.89885929e-01 1.82218313e-01 -4.44395840e-01
-6.22313559e-01 7.35288709e-02 1.05473079e-01 -5.42967141e-01
-1.64381981e-01 -3.62262756e-01 -1.43694603e+00 -2.26275861e-01
-3.32814425e-01 -3.13405812e-01 6.14514709e-01 5.80188990e-01
3.99129331e-01 7.84530759e-01 4.65981036e-01 -9.99285221e-01
7.03823492e-02 -5.37999988e-01 -3.88051808e-01 4.35083330e-01
-1.52841479e-01 -5.90212166e-01 -8.77282098e-02 2.67457336e-01] | [9.393410682678223, -0.14888212084770203] |
95da890d-d22d-4f2e-b872-dd291d7c21d7 | implementation-of-a-practical-markov-chain | 2109.14445 | null | https://arxiv.org/abs/2109.14445v1 | https://arxiv.org/pdf/2109.14445v1.pdf | Implementation of a practical Markov chain Monte Carlo sampling algorithm in PyBioNetFit | Bayesian inference in biological modeling commonly relies on Markov chain Monte Carlo (MCMC) sampling of a multidimensional and non-Gaussian posterior distribution that is not analytically tractable. Here, we present the implementation of a practical MCMC method in the open-source software package PyBioNetFit (PyBNF), which is designed to support parameterization of mathematical models for biological systems. The new MCMC method, am, incorporates an adaptive move proposal distribution. For warm starts, sampling can be initiated at a specified location in parameter space and with a multivariate Gaussian proposal distribution defined initially by a specified covariance matrix. Multiple chains can be generated in parallel using a computer cluster. We demonstrate that am can be used to successfully solve real-world Bayesian inference problems, including forecasting of new Coronavirus Disease 2019 case detection with Bayesian quantification of forecast uncertainty. PyBNF version 1.1.9, the first stable release with am, is available at PyPI and can be installed using the pip package-management system on platforms that have a working installation of Python 3. PyBNF relies on libRoadRunner and BioNetGen for simulations (e.g., numerical integration of ordinary differential equations defined in SBML or BNGL files) and Dask.Distributed for task scheduling on Linux computer clusters. | ['Richard G. Posner', 'William S. Hlavacek', 'Ye Chen', 'Abell T. Duprat1', 'Joshua Colvin', 'Ely F. Miller', 'Abhishek Mallela', 'Yen Ting Lin', 'Jacob Neumann'] | 2021-09-29 | null | null | null | null | ['numerical-integration'] | ['miscellaneous'] | [ 3.74722704e-02 -5.52154183e-01 1.88927248e-01 -2.42800504e-01
-5.57447076e-01 -6.46801353e-01 6.53132558e-01 -8.82821307e-02
-4.36504126e-01 1.05887938e+00 -4.63564068e-01 -9.32995796e-01
1.82679892e-02 -6.04628980e-01 -5.63418806e-01 -1.13177800e+00
4.72522303e-02 1.27939582e+00 2.83075005e-01 5.42673528e-01
1.12513579e-01 8.35045695e-01 -1.22800219e+00 -1.37237618e-02
4.19449836e-01 1.82570413e-01 5.38931310e-01 1.34559405e+00
6.04527593e-02 3.29845063e-02 -3.68676573e-01 6.13824725e-02
-2.29147166e-01 -4.52539325e-01 -3.37489009e-01 -4.99914408e-01
-4.12821561e-01 -4.12759870e-01 4.78033453e-01 6.88614011e-01
6.89554393e-01 -2.08428666e-01 1.07400477e+00 -1.17070866e+00
3.06738675e-01 2.14781672e-01 -2.61713445e-01 1.33283123e-01
1.10312797e-01 6.68327808e-01 1.09146625e-01 -8.21008146e-01
7.24251986e-01 1.57976925e+00 8.52144778e-01 4.14638162e-01
-1.97398353e+00 -5.67375064e-01 -6.07586443e-01 -3.18511277e-01
-1.51760614e+00 -3.45962286e-01 -1.64703876e-01 -9.74636436e-01
1.15129209e+00 2.06654936e-01 7.43916988e-01 1.16652811e+00
8.78128767e-01 2.11975783e-01 1.35935748e+00 -3.54921632e-02
8.49161327e-01 -2.09686875e-01 2.05183640e-01 3.51818651e-01
3.92728239e-01 2.58720249e-01 -1.55128678e-03 -1.08238566e+00
8.05589020e-01 6.15780987e-02 8.28004628e-02 -7.70971477e-02
-1.27161586e+00 1.05494606e+00 -6.89649284e-01 -1.44281268e-01
-6.26502275e-01 3.26872647e-01 2.65445173e-01 -2.39964008e-01
4.46034223e-01 -1.77153334e-01 -8.22251141e-01 -3.26963305e-01
-1.04765081e+00 7.37293601e-01 1.07028317e+00 8.36333513e-01
6.41591489e-01 -9.75332409e-02 -1.46752715e-01 4.15092260e-01
8.56343567e-01 1.11919904e+00 1.07264714e-02 -1.30925703e+00
-5.92772603e-01 -3.12219650e-01 6.20004833e-01 -2.87715405e-01
-6.02934480e-01 6.23094589e-02 -7.61303067e-01 2.15353563e-01
4.46768522e-01 -5.77781677e-01 -8.19567859e-01 1.41255593e+00
9.23658550e-01 4.02098417e-01 -3.78931426e-02 2.80244678e-01
5.79100549e-01 1.00677919e+00 3.70286137e-01 -6.80589557e-01
1.54080141e+00 -1.48036286e-01 -7.02259600e-01 2.21953481e-01
5.53559840e-01 -1.04480565e+00 2.86370367e-01 4.94397134e-01
-9.15610135e-01 -1.49986580e-01 -5.68307698e-01 4.63858128e-01
-3.82953674e-01 -3.45785320e-01 5.90996146e-01 9.00629580e-01
-1.14036167e+00 5.84072948e-01 -1.59922671e+00 -7.44016826e-01
1.23304933e-01 1.40233804e-02 1.29081547e-01 -1.91209316e-01
-8.22530270e-01 7.40129471e-01 3.68922919e-01 4.33533415e-02
-1.35025978e+00 -7.12970436e-01 -6.43534660e-01 -3.02203029e-01
-2.94136375e-01 -1.16924930e+00 1.07059669e+00 4.12682705e-02
-1.75841928e+00 7.45393276e-01 -2.56078750e-01 -2.68804342e-01
4.96857613e-01 3.73697579e-01 -3.66673619e-02 3.41472998e-02
2.46841833e-02 6.56296313e-01 5.29795587e-01 -8.88490856e-01
-1.12963133e-02 -1.83493003e-01 -7.67502010e-01 -3.66871655e-01
7.84624338e-01 5.90426922e-01 -2.31683180e-01 -2.97399372e-01
-2.03454852e-01 -1.07871795e+00 -4.09152925e-01 -1.91730097e-01
-2.58705080e-01 -3.50550413e-02 5.12838423e-01 -7.24593878e-01
8.91632080e-01 -1.62660158e+00 -1.00951321e-01 2.24281833e-01
-3.08377117e-01 2.04832166e-01 -9.50342119e-02 8.81623805e-01
1.46119580e-01 6.40742928e-02 -5.09409964e-01 -2.76835978e-01
8.02660361e-02 2.85142839e-01 -9.60356221e-02 9.84371543e-01
-6.27878904e-02 4.53790039e-01 -9.23160434e-01 -5.70774078e-01
3.32639545e-01 8.25861633e-01 -5.45660973e-01 3.17473501e-01
-5.05268693e-01 8.62093449e-01 -3.79854679e-01 4.90692377e-01
1.31966698e+00 -5.73685646e-01 6.33500874e-01 3.90136868e-01
-5.38656056e-01 4.40576300e-02 -1.37245798e+00 1.27933896e+00
1.02967106e-01 9.08836871e-02 4.09018129e-01 -4.44153577e-01
6.02897406e-01 3.31384212e-01 2.87240297e-01 6.49644673e-01
4.90105674e-02 -5.54649495e-02 -5.95206432e-02 -1.27856478e-01
-1.97113026e-02 -1.84549615e-01 3.05536807e-01 8.14422369e-01
1.11069987e-02 -6.08111501e-01 3.66774827e-01 2.35207468e-01
1.08289385e+00 5.66672325e-01 6.15062118e-01 -6.63129389e-01
1.65771917e-01 3.00029248e-01 7.36699402e-01 1.04826403e+00
-1.48930654e-01 2.13798597e-01 6.46890342e-01 -1.83715761e-01
-1.28159678e+00 -1.23896635e+00 -1.12170327e+00 8.58393490e-01
-6.25100136e-01 -3.90339762e-01 -8.07483673e-01 2.37150788e-01
2.00677633e-01 7.12912440e-01 -2.35607550e-01 5.22009552e-01
-1.22387409e-01 -1.60177946e+00 5.91583490e-01 -3.68948542e-02
-2.64732987e-01 -9.82355893e-01 -7.38293767e-01 5.03103435e-01
1.58206671e-01 -7.02577174e-01 1.43767774e-01 2.29161009e-01
-1.02740479e+00 -9.28092360e-01 -7.42494881e-01 -8.60276818e-02
4.56919611e-01 -1.62467867e-01 9.06364083e-01 3.24076749e-02
-5.17035961e-01 4.58351493e-01 9.78288148e-03 -5.44671714e-01
-8.05507541e-01 -2.43630901e-01 1.73755080e-01 -6.07258499e-01
4.77034599e-01 -4.73158598e-01 -6.00945890e-01 4.11773384e-01
-1.02183557e+00 3.09863627e-01 -8.38418007e-02 9.09675598e-01
7.67980576e-01 -4.06491935e-01 4.10666257e-01 -6.61053598e-01
3.32438380e-01 -8.10940027e-01 -1.31726456e+00 7.17488974e-02
-2.77734756e-01 -1.89645767e-01 9.31020156e-02 -1.80928148e-02
-1.05006337e+00 2.69465059e-01 -4.81433183e-01 -8.79658312e-02
-4.46263105e-01 7.79431164e-01 1.80233464e-01 4.39885229e-01
1.97207943e-01 8.08276236e-02 6.29594147e-01 -5.35428762e-01
1.73848104e-02 6.62833750e-01 1.76546965e-02 -6.39972091e-01
1.06650889e-01 6.55560613e-01 3.73344690e-01 -1.01386857e+00
-1.57631844e-01 -4.37870532e-01 -2.48774096e-01 -2.02073246e-01
1.04181838e+00 -9.82363164e-01 -1.13612771e+00 1.04678810e+00
-1.25047743e+00 -8.55396509e-01 4.58928555e-01 8.64647686e-01
-6.63044691e-01 3.73092651e-01 -7.45646715e-01 -1.04628980e+00
-2.26967320e-01 -1.13990414e+00 1.00162137e+00 1.15301818e-01
-2.99424142e-01 -1.01429248e+00 7.90824056e-01 -1.26363605e-01
4.50716645e-01 3.68159711e-01 7.36122489e-01 -5.26306391e-01
-4.39773560e-01 -1.57863855e-01 1.55425817e-02 -1.35548664e-02
-2.43337601e-01 9.84662354e-01 -9.24693108e-01 -5.95370889e-01
-3.82762402e-02 -8.45500827e-02 6.90968573e-01 1.11368620e+00
7.05587089e-01 1.35481194e-01 -8.45853329e-01 6.34010136e-01
1.25962305e+00 1.31718844e-01 5.45722544e-01 -1.23867802e-01
-6.88471347e-02 3.63506943e-01 5.18191516e-01 1.11752343e+00
1.08874738e-01 4.02303725e-01 2.21988246e-01 2.39120245e-01
6.19721949e-01 2.77512461e-01 2.38083094e-01 5.78029037e-01
3.93133759e-02 -1.56563938e-01 -1.25361145e+00 9.15789083e-02
-1.84536648e+00 -1.15366578e+00 -7.31679738e-01 2.37785077e+00
1.14614999e+00 -2.70651549e-01 1.81522533e-01 -5.84228992e-01
8.89544845e-01 -5.52487373e-01 -5.40224075e-01 -2.98676282e-01
3.96291167e-03 1.37372062e-01 5.89799225e-01 8.11578155e-01
-8.68313432e-01 7.11510777e-01 7.33227348e+00 8.68693471e-01
-5.73844492e-01 3.89834195e-01 5.81244171e-01 3.78737859e-02
-4.92941812e-02 2.99969375e-01 -1.40351021e+00 7.79038787e-01
1.62724018e+00 -6.31613135e-02 4.73511040e-01 5.17702222e-01
8.57851684e-01 -7.50167012e-01 -7.89835751e-01 4.86447990e-01
-5.89689910e-01 -1.38351321e+00 -4.53039914e-01 4.13769484e-01
6.10025764e-01 5.10034621e-01 -3.44222903e-01 1.40707389e-01
1.02626765e+00 -7.17245519e-01 2.34532937e-01 9.74035800e-01
6.84892058e-01 -7.46288061e-01 8.40902448e-01 6.38706923e-01
-6.14565551e-01 6.94946468e-01 -7.13087440e-01 -3.27059850e-02
6.48892462e-01 1.18589246e+00 -1.18520546e+00 1.14220619e-01
6.70754433e-01 2.98989773e-01 -9.54203531e-02 1.23522747e+00
1.45888217e-02 1.04259145e+00 -8.28275502e-01 -1.59490153e-01
-1.71422407e-01 -5.21809697e-01 5.17864645e-01 1.57385194e+00
3.73137772e-01 4.46028635e-02 3.38727646e-02 1.03993738e+00
9.67930496e-01 -2.81479269e-01 -2.87934184e-01 -1.61879510e-01
7.19522059e-01 1.16740370e+00 -1.12741745e+00 -6.55073881e-01
1.47752047e-01 3.70497912e-01 -1.37761086e-01 4.88774836e-01
-1.09069455e+00 -2.96601385e-01 8.16311181e-01 -2.20825166e-01
5.94972789e-01 -4.11712587e-01 1.53030857e-01 -8.02712440e-01
-9.82655168e-01 -7.74468780e-01 5.13820410e-01 -1.02315998e+00
-1.35654449e+00 8.57533813e-02 5.89910448e-01 -4.78700161e-01
-6.35466933e-01 -7.92496979e-01 -4.74721164e-01 1.38708842e+00
-8.65770996e-01 -7.09486961e-01 1.90340012e-01 2.23810807e-01
-1.42986700e-01 2.82998800e-01 1.12217546e+00 -6.57303408e-02
-7.53994942e-01 -2.72122860e-01 9.44662869e-01 -6.09219909e-01
6.45898998e-01 -9.41906452e-01 5.07040977e-01 4.80218291e-01
-8.78150463e-01 1.23306668e+00 1.11524987e+00 -1.33250761e+00
-1.38587987e+00 -9.09979105e-01 9.15533364e-01 -4.17611241e-01
7.09265947e-01 -4.89530623e-01 -7.61705935e-01 7.68679082e-01
1.09801926e-01 -3.54060918e-01 9.86018181e-01 -2.18926907e-01
9.52921212e-02 4.96862352e-01 -1.34945452e+00 3.89719188e-01
1.10749409e-01 -1.86195314e-01 -1.47932097e-01 7.18591213e-01
3.27539086e-01 -1.73450679e-01 -1.14762485e+00 2.00978175e-01
8.18790615e-01 -6.48137152e-01 8.59490275e-01 -5.79990506e-01
-2.53462076e-01 -9.08102214e-01 -1.76486418e-01 -8.90293002e-01
-2.21847937e-01 -1.11114609e+00 1.67668089e-01 1.03298736e+00
1.70363829e-01 -9.44979191e-01 1.41668558e-01 3.68432343e-01
1.97686687e-01 -1.84633493e-01 -1.16273475e+00 -7.50466347e-01
3.01962376e-01 -5.55706382e-01 5.59485018e-01 5.30069888e-01
-1.65617988e-01 -8.47062394e-02 -2.61531591e-01 3.00327748e-01
1.09268975e+00 -1.66182876e-01 9.81873155e-01 -1.29899216e+00
-6.62968218e-01 -6.50087968e-02 5.39075024e-02 -6.96611106e-01
-7.57469907e-02 -5.39690614e-01 3.46877635e-01 -1.44812000e+00
6.02429330e-01 -1.62509710e-01 1.84721753e-01 3.16404819e-01
-3.42085101e-02 3.19355838e-02 -3.68859619e-01 2.00086251e-01
-1.40877664e-01 1.39348000e-01 6.76558077e-01 3.09688538e-01
8.94336253e-02 6.02555610e-02 2.89371490e-01 6.58394873e-01
8.14483404e-01 -1.02555013e+00 1.42046854e-01 1.78996950e-01
3.09451401e-01 2.38621548e-01 7.40342557e-01 -5.97069502e-01
-9.38065276e-02 -5.06432474e-01 3.15998018e-01 -1.32235110e+00
3.32817465e-01 -3.67651254e-01 1.22677755e+00 9.97785151e-01
2.88721353e-01 1.65780470e-01 4.25846457e-01 6.42901540e-01
7.88301289e-01 -5.70037007e-01 1.05610001e+00 -2.47373059e-01
3.26495469e-01 6.34060353e-02 -1.56767154e+00 -3.67527515e-01
9.02465761e-01 8.19344223e-02 -5.47899663e-01 -1.27369896e-01
-9.31829631e-01 2.01027945e-01 8.91542912e-01 -6.18636370e-01
1.39114201e-01 -8.31353247e-01 -9.09879982e-01 -2.44343355e-02
-1.83235720e-01 -4.13676590e-01 5.86932421e-01 1.24224949e+00
-1.17545891e+00 5.53328693e-01 -4.98455577e-02 -1.05558753e+00
-1.08204401e+00 6.80551767e-01 1.97159052e-01 -2.47445479e-01
-1.87319845e-01 4.41267937e-01 4.26010899e-02 -7.09875882e-01
-2.40684092e-01 -2.40167692e-01 2.01013759e-01 -9.91591960e-02
8.46578598e-01 5.38095176e-01 -2.34984860e-01 -2.43268907e-01
-5.67778528e-01 1.38357073e-01 2.64781833e-01 -4.40743297e-01
1.42476702e+00 -1.10724814e-01 -6.20374024e-01 6.33455515e-01
7.81037211e-01 -4.33890253e-01 -1.32628083e+00 3.79233390e-01
-1.75484374e-01 -8.92345533e-02 -5.17471954e-02 -8.00417423e-01
-1.99483275e-01 7.04339802e-01 6.17205143e-01 -5.65961413e-02
3.13754976e-01 -6.01773113e-02 9.66174826e-02 2.67198473e-01
2.30507851e-01 -7.45873809e-01 -7.37576365e-01 6.23054445e-01
5.35406351e-01 -8.42058122e-01 3.97527277e-01 -2.01962609e-02
-1.07734382e-01 9.03102756e-01 -6.93120016e-03 2.60582805e-01
1.15625274e+00 6.47959173e-01 -1.89364061e-01 -2.39402384e-01
-1.18639410e+00 7.38849491e-02 -4.72582668e-01 7.42666066e-01
4.98217255e-01 2.90544063e-01 -6.30535960e-01 3.44455481e-01
3.59820038e-01 3.38148177e-01 6.81860626e-01 1.25880456e+00
-4.50579673e-01 -1.08505523e+00 -9.37332034e-01 3.19788665e-01
-4.48608816e-01 -4.83686984e-01 1.96641594e-01 3.62569630e-01
-1.79315850e-01 9.22143757e-01 3.31001014e-01 4.15232450e-01
-5.95951319e-01 3.31196219e-01 3.89789164e-01 -4.81458187e-01
-3.13612908e-01 7.94404805e-01 1.43046394e-01 -2.22213596e-01
-4.05885398e-01 -1.38125551e+00 -1.08511424e+00 -6.31866097e-01
-2.22815350e-01 3.22408706e-01 1.11295068e+00 6.85035467e-01
5.67593396e-01 -1.00057304e-01 3.65222804e-02 -1.13119853e+00
-4.21505272e-01 -1.27186859e+00 -9.59481776e-01 -6.07923806e-01
-7.21697956e-02 -5.01605630e-01 -6.10915422e-01 3.40457529e-01] | [6.62697172164917, 4.053398132324219] |
6244fd97-2c16-4b39-bc78-8069b5211f97 | gvcnn-group-view-convolutional-neural | null | null | http://openaccess.thecvf.com/content_cvpr_2018/html/Feng_GVCNN_Group-View_Convolutional_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Feng_GVCNN_Group-View_Convolutional_CVPR_2018_paper.pdf | GVCNN: Group-View Convolutional Neural Networks for 3D Shape Recognition | 3D shape recognition has attracted much attention recently. Its recent advances advocate the usage of deep features and achieve the state-of-the-art performance. However, existing deep features for 3D shape recognition are restricted to a view-to-shape setting, which learns the shape descriptor from the view-level feature directly. Despite the exciting progress on view-based 3D shape description, the intrinsic hierarchical correlation and discriminability among views have not been well exploited, which is important for 3D shape representation. To tackle this issue, in this paper, we propose a group-view convolutional neural network (GVCNN) framework for hierarchical correlation modeling towards discriminative 3D shape description. The proposed GVCNN framework is composed of a hierarchical view-group-shape architecture, i.e., from the view level, the group level and the shape level, which are organized using a grouping strategy. Concretely, we first use an expanded CNN to extract a view level descriptor. Then, a grouping module is introduced to estimate the content discrimination of each view, based on which all views can be splitted into different groups according to their discriminative level. A group level description can be further generated by pooling from view descriptors. Finally, all group level descriptors are combined into the shape level descriptor according to their discriminative weights. Experimental results and comparison with state-of-the-art methods show that our proposed GVCNN method can achieve a significant performance gain on both the 3D shape classification and retrieval tasks. | ['Yue Gao', 'Zizhao Zhang', 'Yifan Feng', 'Xibin Zhao', 'Rongrong Ji'] | 2018-06-01 | null | null | null | cvpr-2018-6 | ['3d-shape-retrieval', '3d-shape-recognition', '3d-shape-representation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-2.63887972e-01 -5.46615005e-01 -1.56871751e-01 -6.20662570e-01
-6.68678701e-01 -6.21393383e-01 8.21284711e-01 5.69808744e-02
2.40730956e-01 -1.47365347e-01 3.42475146e-01 2.06989288e-01
-2.26991802e-01 -9.86149311e-01 -2.80561715e-01 -8.85844409e-01
3.75914514e-01 4.10278708e-01 8.19867179e-02 7.01509416e-02
1.51082918e-01 9.94770050e-01 -1.40037572e+00 3.50316465e-01
3.46139342e-01 1.36310816e+00 -8.76579881e-02 8.55548829e-02
-2.16799706e-01 1.62430733e-01 -2.07479835e-01 -8.29014778e-02
3.04485679e-01 -2.42373511e-01 -1.71809942e-01 3.75159502e-01
5.66627383e-01 -5.17879009e-01 -3.96294683e-01 8.34484100e-01
9.38698828e-01 -4.14663441e-02 8.02514017e-01 -1.11851025e+00
-9.48783457e-01 -8.45540911e-02 -5.54639220e-01 -1.10649861e-01
1.48886979e-01 -4.13056649e-03 1.13069212e+00 -1.44630659e+00
5.34929991e-01 1.47854733e+00 2.78877079e-01 3.68126303e-01
-1.16468287e+00 -6.82060957e-01 3.67722660e-01 4.15076129e-02
-1.34196234e+00 -3.53024274e-01 1.18107569e+00 -4.66305882e-01
9.65950191e-01 3.37350555e-02 8.56777787e-01 7.16684639e-01
-8.95532370e-02 1.02375472e+00 9.62019861e-01 1.21421866e-01
2.25384310e-02 -2.75486916e-01 -1.48719355e-01 7.59248972e-01
1.97554290e-01 -4.97571714e-02 -3.40788186e-01 -1.23663329e-01
9.82999623e-01 4.82286006e-01 -1.29387215e-01 -1.00721979e+00
-1.02411246e+00 7.33737469e-01 6.24097466e-01 4.79492128e-01
-3.40580046e-01 -1.10487811e-01 5.79221666e-01 8.29073861e-02
7.10067451e-01 -1.98761955e-01 -3.16509485e-01 2.62148291e-01
-7.87664294e-01 1.63246736e-01 5.70554614e-01 1.04385138e+00
6.57189190e-01 1.14735983e-01 -4.35230970e-01 1.15098059e+00
5.54004848e-01 6.73699439e-01 1.02664083e-01 -4.63122010e-01
5.63392639e-01 1.23710978e+00 -3.31846237e-01 -1.34031332e+00
-4.08012778e-01 -5.89738190e-01 -1.23919475e+00 1.45075843e-01
-6.96569011e-02 5.05953312e-01 -9.02811944e-01 1.56153667e+00
2.76863545e-01 -3.24421041e-02 -9.31376889e-02 1.08334446e+00
1.28237474e+00 4.26632315e-01 -1.75514713e-01 4.15139943e-02
1.33759916e+00 -8.29801142e-01 -3.03723842e-01 2.97288805e-01
4.34127301e-01 -6.93363488e-01 5.95783174e-01 9.83969588e-03
-1.18935561e+00 -8.00325394e-01 -1.11233640e+00 -2.02764943e-01
-3.34531188e-01 4.18458164e-01 4.17518497e-01 4.18776095e-01
-9.03250933e-01 2.59024054e-01 -7.77925313e-01 -2.96846032e-01
7.01780021e-01 3.00878227e-01 -6.44710362e-01 -3.69271040e-01
-7.08578527e-01 4.42977518e-01 2.83545583e-01 1.27867237e-01
-8.28463793e-01 -4.83281791e-01 -1.06185031e+00 2.61495650e-01
1.60636380e-01 -8.99414241e-01 6.59688056e-01 -3.55969280e-01
-1.27601159e+00 1.20487833e+00 -2.16144010e-01 2.59789139e-01
1.78417981e-01 3.07164133e-01 -3.06397051e-01 1.50783494e-01
1.01814181e-01 4.34494525e-01 8.63233864e-01 -1.53247571e+00
-4.23241824e-01 -8.54311705e-01 2.84378491e-02 3.52491140e-01
-2.81689554e-01 -1.80728957e-01 -7.96046019e-01 -5.31222463e-01
7.57666409e-01 -7.50106215e-01 2.02559739e-01 1.57864928e-01
-1.82546198e-01 -6.08664751e-01 7.85015285e-01 -2.41986826e-01
1.00098312e+00 -2.22444797e+00 2.88255930e-01 2.04105586e-01
4.44229424e-01 3.08785588e-01 -2.92018473e-01 5.87483764e-01
-7.24358782e-02 4.71973494e-02 -4.00636196e-02 -5.19285738e-01
1.10481665e-01 4.85326499e-02 -6.49390817e-02 5.23449719e-01
2.97444582e-01 1.02829826e+00 -7.41709590e-01 -2.44304612e-01
3.83880675e-01 6.39776647e-01 -6.07233047e-01 4.13717657e-01
1.87157810e-01 4.67244714e-01 -8.49709153e-01 9.26075697e-01
1.19282103e+00 -2.87000656e-01 1.00526903e-02 -6.42851293e-01
-2.29864076e-01 -1.71600636e-02 -8.68645191e-01 1.78402030e+00
-5.59053898e-01 -1.77488942e-02 4.22027409e-02 -1.11601174e+00
1.51093328e+00 1.93158075e-01 7.18854785e-01 -4.72168177e-01
1.68164931e-02 4.90592808e-01 -1.26940966e-01 -2.05996484e-01
2.38699391e-02 -2.10355476e-01 -1.02418318e-01 4.67658252e-01
3.66573989e-01 -2.88980752e-01 -3.15513760e-01 -1.66781873e-01
7.16373324e-01 1.12920232e-01 3.88836622e-01 -1.84972752e-02
1.08412647e+00 -5.58635235e-01 3.79120886e-01 3.26765150e-01
-1.22916363e-01 8.51849735e-01 4.02911663e-01 -8.18941593e-01
-1.02785051e+00 -1.09098840e+00 -1.43457681e-01 6.78685367e-01
2.46187285e-01 -2.94562697e-01 -3.48124743e-01 -8.63867581e-01
2.93332428e-01 7.25407302e-02 -5.77399373e-01 -1.58204153e-01
-6.07829571e-01 -3.12592596e-01 3.68772000e-01 6.33817196e-01
7.67491698e-01 -9.06174481e-01 -3.23115885e-01 6.45733774e-02
3.43473345e-01 -1.13684118e+00 -7.38833547e-01 -7.20088854e-02
-9.41112101e-01 -8.58180821e-01 -8.46483052e-01 -8.63286078e-01
7.90153384e-01 7.81362593e-01 9.13417935e-01 1.56041905e-01
7.26970360e-02 6.54942095e-01 -3.27551901e-01 -2.25588098e-01
2.14935895e-02 8.14875662e-02 -1.18374433e-02 4.03140694e-01
6.42179906e-01 -9.73874152e-01 -9.09811199e-01 4.04598445e-01
-9.64755535e-01 -8.45181644e-02 8.07365596e-01 1.00510943e+00
9.02666986e-01 -4.27181333e-01 4.08723116e-01 -4.00779754e-01
3.98823887e-01 -2.71338195e-01 -5.32855809e-01 2.93187588e-01
-3.46470475e-01 -1.36984966e-03 7.38965392e-01 -5.19496500e-02
-7.74312139e-01 1.77451596e-01 -2.79783845e-01 -8.46235394e-01
-4.37002569e-01 5.05116343e-01 -6.85517251e-01 -1.57866150e-01
-7.47042447e-02 8.11658919e-01 1.19635105e-01 -6.25492513e-01
4.06657487e-01 7.22798526e-01 3.13338861e-02 -4.20401961e-01
9.47021484e-01 7.34564424e-01 2.15834945e-01 -6.59326434e-01
-1.01034248e+00 -6.59650922e-01 -1.03302562e+00 -2.43097082e-01
9.62780178e-01 -1.08603895e+00 -8.37748230e-01 4.74462301e-01
-1.23616087e+00 3.90893996e-01 -1.15427971e-02 3.65211874e-01
-6.71895087e-01 4.65699583e-01 -2.98373938e-01 -6.17347419e-01
-4.51516300e-01 -1.24806583e+00 1.66069913e+00 1.57016620e-01
3.91867310e-01 -9.32499051e-01 4.12587635e-02 3.25910449e-01
3.92255336e-01 2.81541139e-01 1.13978016e+00 -9.49166834e-01
-6.67214870e-01 -3.11274886e-01 -6.87901855e-01 3.54660034e-01
3.33424300e-01 -4.46074545e-01 -9.64710951e-01 -6.14504278e-01
1.91080555e-01 -3.23312938e-01 9.50424194e-01 2.30073258e-01
1.21474075e+00 1.09334610e-01 -2.26653799e-01 8.18200052e-01
1.52435970e+00 2.97196001e-01 2.96220630e-01 -8.06068927e-02
9.31225955e-01 3.70142013e-01 2.32945636e-01 5.29469669e-01
3.85414720e-01 8.58737826e-01 4.96413469e-01 1.09368144e-02
-1.52787253e-01 -4.45442706e-01 -3.80529799e-02 1.43040431e+00
-2.33772025e-01 -1.46772146e-01 -6.52497947e-01 3.44091624e-01
-1.77411914e+00 -8.68222177e-01 1.79777339e-01 2.17301822e+00
1.26013726e-01 7.73605570e-05 -5.92348278e-02 -1.96964487e-01
6.10122204e-01 7.43977606e-01 -6.32513046e-01 -3.07347387e-01
-8.07551146e-02 -5.35812378e-02 -6.01037294e-02 2.30945069e-02
-1.17346108e+00 6.86901629e-01 4.72105169e+00 1.06308782e+00
-1.19261646e+00 -1.04904026e-01 4.62206095e-01 3.38064462e-01
-4.71706003e-01 -1.50537238e-01 -9.10092175e-01 7.82363340e-02
-3.93814854e-02 -1.35589223e-02 1.06469214e-01 7.82507718e-01
8.32156539e-02 5.11592984e-01 -1.10402656e+00 1.31373608e+00
5.49806118e-01 -1.01406574e+00 6.93897009e-01 2.29508027e-01
6.52309418e-01 -2.03184485e-01 8.84779468e-02 3.57919902e-01
-2.99064487e-01 -9.75660741e-01 3.54802996e-01 6.71079159e-01
9.13722396e-01 -8.39922667e-01 7.22018600e-01 4.61815774e-01
-1.84278071e+00 1.10407189e-01 -6.89618051e-01 2.74545372e-01
6.63109943e-02 4.40049678e-01 -2.10592195e-01 1.14273059e+00
2.63048679e-01 1.26994646e+00 -3.74663144e-01 9.24844265e-01
8.59017968e-02 3.72521318e-02 -3.85263488e-02 -3.79858315e-02
4.51883614e-01 -4.48229432e-01 5.89268923e-01 9.31713998e-01
5.47353864e-01 3.43986601e-01 5.19582033e-01 9.95150328e-01
-8.59603286e-02 3.00363779e-01 -1.02073967e+00 9.72844362e-02
3.31672966e-01 1.49234688e+00 -6.49439812e-01 -2.42537931e-01
-8.55029941e-01 8.35933149e-01 4.54858720e-01 1.25114068e-01
-3.36506635e-01 -1.46695107e-01 5.29283881e-01 -9.03424621e-02
8.13757181e-01 -3.90014589e-01 -6.74097687e-02 -1.38999152e+00
9.09806192e-02 -6.97689712e-01 2.23403499e-01 -5.36321938e-01
-1.75357580e+00 5.71531296e-01 1.79320928e-02 -1.78548586e+00
2.13159308e-01 -7.88885713e-01 -5.17613947e-01 1.05119359e+00
-1.41285682e+00 -1.71508467e+00 -5.01104832e-01 4.54778224e-01
6.01920068e-01 -5.49567521e-01 9.46013629e-01 3.04162443e-01
-1.69596136e-01 5.89836240e-01 1.17244003e-02 3.56614351e-01
5.43971539e-01 -1.01273763e+00 3.51703852e-01 1.65053219e-01
2.23531440e-01 6.59668148e-01 -2.80124068e-01 -4.63201076e-01
-1.58698666e+00 -1.07516706e+00 8.35942626e-01 -3.34211439e-01
3.72048765e-01 -4.59513813e-01 -9.32858527e-01 1.78122371e-01
-1.02620982e-01 4.98774469e-01 7.77459562e-01 -7.96667933e-02
-7.40986288e-01 -3.75794351e-01 -8.95034790e-01 3.17699283e-01
1.36531198e+00 -8.12713146e-01 -4.85979795e-01 -7.89343938e-02
3.52592915e-01 -3.60368550e-01 -1.19491756e+00 6.65188134e-01
8.49831820e-01 -1.20853722e+00 1.11083364e+00 -5.47618151e-01
5.19946456e-01 -3.20764393e-01 -4.28805798e-01 -1.26006198e+00
-7.00169563e-01 1.78968787e-01 3.56982648e-02 1.32309628e+00
-6.69091716e-02 -4.89319563e-01 6.37438834e-01 8.57691765e-02
-2.47306317e-01 -1.24691308e+00 -1.02907622e+00 -7.78048754e-01
1.43733934e-01 -2.64900237e-01 7.35906005e-01 7.69049346e-01
-5.10055840e-01 5.33674955e-01 -6.89034238e-02 7.47205094e-02
6.93072259e-01 9.00358796e-01 7.17359722e-01 -1.38933969e+00
-7.79586807e-02 -7.23640800e-01 -9.00591671e-01 -1.41995525e+00
2.16783404e-01 -1.29649198e+00 -3.18233937e-01 -1.68731689e+00
6.38219357e-01 -2.89331406e-01 -5.91354311e-01 2.44917676e-01
7.31777586e-03 2.45982498e-01 5.08094370e-01 2.62997806e-01
-6.58372402e-01 8.87160361e-01 1.72692025e+00 -4.40830380e-01
2.36188322e-02 1.25281781e-01 -4.36910242e-01 6.43138409e-01
4.77532238e-01 -8.96159485e-02 -3.72225821e-01 -4.26828384e-01
-1.04523553e-02 2.24384353e-01 4.33941096e-01 -8.08907926e-01
2.33880550e-01 2.29508325e-01 5.87615550e-01 -1.13764453e+00
5.14389515e-01 -9.06419277e-01 -1.00272238e-01 3.23683590e-01
-1.48709729e-01 -1.93931293e-02 5.68363592e-02 7.03844965e-01
-6.30268037e-01 1.01922057e-03 6.47133350e-01 -2.92829156e-01
-3.96405309e-01 9.04785454e-01 1.08937956e-01 -1.76135823e-01
7.08525121e-01 -3.41175824e-01 1.67526435e-02 -3.96916658e-01
-7.25872040e-01 7.72052407e-02 3.34773660e-01 5.31638086e-01
9.52049375e-01 -2.16894531e+00 -7.89588869e-01 5.74260533e-01
5.32781780e-01 -5.08762673e-02 4.75152671e-01 7.65762866e-01
-2.50296071e-02 6.49500012e-01 -3.22920501e-01 -8.75400066e-01
-1.21280909e+00 5.95400631e-01 2.69176394e-01 -4.78310823e-01
-5.01400888e-01 5.53812504e-01 7.05996811e-01 -7.15987206e-01
6.18838184e-02 -2.71692276e-01 -5.73607028e-01 3.08186382e-01
3.29894155e-01 2.90560871e-02 6.54422864e-02 -1.03223109e+00
-4.35878098e-01 1.41632998e+00 1.30282283e-01 2.77326643e-01
1.65101039e+00 2.06865016e-02 -2.51049489e-01 3.99325609e-01
1.77704978e+00 -1.00266278e-01 -1.04345107e+00 -4.18519586e-01
-2.26070851e-01 -5.17104387e-01 -7.50951543e-02 -3.77780110e-01
-1.29667699e+00 1.30509162e+00 5.72644591e-01 3.05930972e-01
1.12891769e+00 1.12937026e-01 8.51479352e-01 2.32147574e-01
4.53256547e-01 -5.39210260e-01 1.53411597e-01 7.81208456e-01
1.13666975e+00 -1.26343560e+00 5.96150346e-02 -3.42508107e-01
-2.90744275e-01 1.40895081e+00 5.48083961e-01 -4.04994041e-01
9.75926042e-01 -1.95575073e-01 -1.18184753e-01 -6.20190799e-01
-5.42746305e-01 -3.83484036e-01 8.41875553e-01 5.58845043e-01
4.81385618e-01 5.03267534e-02 -2.41596326e-01 8.75158787e-01
4.51689005e-01 -3.14767152e-01 -2.22188279e-01 4.77071762e-01
-3.36971790e-01 -1.22026801e+00 -7.83293396e-02 4.75144833e-01
-1.28626972e-01 4.91186678e-02 -4.55045611e-01 6.32771611e-01
1.71354592e-01 4.45232004e-01 -4.80833761e-02 -4.98758972e-01
7.37210333e-01 2.40902621e-02 4.97879118e-01 -7.22485602e-01
-5.24368405e-01 3.35140735e-01 -2.36917347e-01 -4.67780262e-01
-6.80857718e-01 -5.70632935e-01 -7.95130551e-01 3.79083455e-02
-1.30431131e-01 -1.95583314e-01 2.61185646e-01 8.43551457e-01
4.80161965e-01 1.41551197e-01 1.05496657e+00 -1.23088312e+00
-5.90924621e-01 -6.81340456e-01 -6.87253237e-01 6.56698704e-01
1.01684511e-01 -9.21032071e-01 -2.83587545e-01 -4.47046369e-01] | [8.15119457244873, -3.865414619445801] |
8f6f0c08-9af3-4d23-8a74-f92c7287a6ad | hengam-an-adversarially-trained-transformer | null | null | https://aclanthology.org/2022.aacl-main.74/ | https://aclanthology.org/2022.aacl-main.74.pdf | Hengam: An Adversarially Trained Transformer for Persian Temporal Tagging | Many NLP main tasks benefit from an accurate understanding of temporal expressions, e.g., text summarization, question answering, and information retrieval. This paper introduces Hengam, an adversarially trained transformer for Persian temporal tagging outperforming state-of-the-art approaches on a diverse and manually created dataset. We create Hengam in the following concrete steps: (1) we develop HengamTagger, an extensible rule-based tool that can extract temporal expressions from a set of diverse language-specific patterns for any language of interest. (2) We apply HengamTagger to annotate temporal tags in a large and diverse Persian text collection (covering both formal and informal contexts) to be used as weakly labeled data. (3) We introduce an adversarially trained transformer model on HengamCorpus that can generalize over the HengamTagger’s rules. We create HengamGold, the first high-quality gold standard for Persian temporal tagging. Our trained adversarial HengamTransformer not only achieves the best performance in terms of the F1-score (a type F1-Score of 95.42 and a partial F1-Score of 91.60) but also successfully deals with language ambiguities and incorrect spellings. Our code, data, and models are publicly available at https://github.com/kargaranamir/Hengam. | ['Ehsaneddin Asgari', 'Hinrich Schütze', 'Amir Hossein Kargaran', 'Sajad Mirzababaei'] | 2022-11-20 | null | null | null | proceedings-of-the-2nd-conference-of-the-asia | ['text-anonymization', 'temporal-tagging'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.94757320e-02 4.70137239e-01 -3.06537330e-01 -1.50803000e-01
-1.37809575e+00 -1.21113276e+00 8.71369064e-01 -8.03879648e-03
-3.27598929e-01 9.27055836e-01 4.13834870e-01 -3.53227377e-01
2.34650031e-01 -4.97364789e-01 -5.81803679e-01 -4.89449352e-01
-1.46067977e-01 8.39456916e-01 5.22963822e-01 -4.62206334e-01
-3.17133993e-01 1.03079760e-02 -6.66346252e-01 3.72873932e-01
1.10053885e+00 6.99363172e-01 1.20621743e-02 6.73075438e-01
2.84873489e-02 1.17189705e+00 -8.07166755e-01 -9.12393808e-01
-4.22026739e-02 -4.87607777e-01 -1.38446712e+00 -5.82606912e-01
1.12358987e-01 1.32732363e-02 -4.82259691e-01 6.45186484e-01
2.72026598e-01 3.43050919e-02 4.89604354e-01 -9.76149857e-01
-5.26377976e-01 1.18266022e+00 -2.52474874e-01 3.71533096e-01
3.86718065e-01 1.10477500e-03 1.22042286e+00 -3.83723527e-01
9.55584645e-01 1.09856868e+00 7.15707064e-01 8.09178948e-01
-1.10240865e+00 -6.65459573e-01 1.01271132e-02 2.27115244e-01
-1.14122987e+00 -2.80562669e-01 6.47158921e-01 -4.12719637e-01
1.13352454e+00 3.74406993e-01 5.83958805e-01 1.42882824e+00
6.04702115e-01 9.59906518e-01 8.26318026e-01 -4.03888971e-01
-6.44115210e-02 -3.32013965e-01 1.68757588e-01 5.43338835e-01
-4.17730331e-01 1.18059935e-02 -6.03537261e-01 -1.61839887e-01
2.91337162e-01 -4.30817962e-01 -1.45236805e-01 4.80683267e-01
-1.32208741e+00 6.90498769e-01 1.42799482e-01 9.03086364e-01
-3.10352564e-01 2.31574938e-01 6.71959579e-01 2.99542934e-01
6.04767084e-01 5.79483271e-01 -7.67800033e-01 -5.64241588e-01
-9.07635331e-01 2.15032414e-01 8.59948933e-01 1.07087791e+00
4.15537298e-01 3.23319323e-02 -5.79514444e-01 9.43823874e-01
-2.77303547e-01 8.20145726e-01 5.49255967e-01 -1.16864395e+00
5.05714417e-01 2.31492266e-01 2.88790446e-02 -5.76108336e-01
-2.19809726e-01 -1.58992156e-01 -7.94703901e-01 -7.33751833e-01
4.39807236e-01 -2.94710845e-01 -8.51438999e-01 2.10351205e+00
1.31394923e-01 -1.78827584e-01 2.87294835e-01 3.80781829e-01
7.73218513e-01 1.07687366e+00 3.04345131e-01 -3.72342408e-01
1.67944288e+00 -8.73760343e-01 -8.63612652e-01 -6.11866951e-01
7.80704916e-01 -7.83952475e-01 9.07116234e-01 2.59767652e-01
-9.81275439e-01 -3.78096700e-02 -6.99684560e-01 -3.63461852e-01
-1.96588710e-01 1.89194456e-01 6.32442892e-01 2.77645066e-02
-9.59180176e-01 6.90118611e-01 -1.16612494e+00 -5.59555113e-01
-1.79451387e-02 1.91027343e-01 -4.86887395e-01 1.42131448e-01
-1.54252827e+00 9.77682710e-01 6.61908150e-01 -1.28355458e-01
-7.48543382e-01 -5.01138806e-01 -1.00152111e+00 -2.85487235e-01
4.09950823e-01 -4.05568182e-01 1.97027147e+00 -9.86010969e-01
-1.40759468e+00 1.12662232e+00 -2.80330509e-01 -9.47987854e-01
3.83321971e-01 -4.42722231e-01 -5.74423134e-01 1.92841023e-01
4.21738803e-01 5.18107355e-01 4.50517803e-01 -6.99411273e-01
-4.46107537e-01 -5.86825283e-03 -1.17543131e-01 -2.13901505e-01
-4.24439870e-02 3.15807939e-01 -5.56007087e-01 -1.03626561e+00
-2.37144023e-01 -1.10520065e+00 1.58492904e-02 -7.27784693e-01
-5.58093905e-01 -6.06953204e-01 8.27192187e-01 -1.21936762e+00
1.38940644e+00 -2.00961208e+00 1.57610491e-01 -2.78187990e-01
-1.03265665e-01 5.37973225e-01 -1.73145205e-01 7.77935147e-01
2.69567613e-02 2.02740341e-01 -4.45227176e-01 -3.92886847e-01
6.80849701e-02 4.72310334e-01 -7.00122476e-01 1.61673531e-01
1.76872462e-01 1.22492230e+00 -1.22859895e+00 -6.41004920e-01
-8.49351753e-03 1.38897792e-01 -3.33525360e-01 3.14905316e-01
-5.59993804e-01 4.69456881e-01 -4.39469725e-01 5.70908248e-01
1.14680707e-01 8.34248737e-02 4.45115566e-01 2.24706642e-02
-6.81251734e-02 8.55934560e-01 -2.74906904e-01 1.73890936e+00
-6.11779809e-01 7.00943768e-01 -2.94946104e-01 -5.97986996e-01
7.99810708e-01 6.24443769e-01 5.46583235e-01 -5.21096051e-01
1.91806480e-01 3.20498228e-01 -1.80526838e-01 -2.83445954e-01
6.62401915e-01 -3.11938792e-01 -9.80873227e-01 5.63702106e-01
4.83796000e-01 -2.90853560e-01 4.23946172e-01 5.48561335e-01
1.59059465e+00 1.38452753e-01 7.56112218e-01 -1.16008796e-01
4.61702377e-01 2.64774561e-01 1.22841382e+00 4.58454818e-01
-2.27569744e-01 4.44909424e-01 7.01600373e-01 -4.10589069e-01
-9.51559246e-01 -9.40561295e-01 1.93390369e-01 1.14511311e+00
-3.58349711e-01 -7.88310528e-01 -6.63008869e-01 -1.11552751e+00
-4.00167406e-01 1.10322368e+00 -6.92036271e-01 -5.84088750e-02
-1.03159344e+00 -3.32234532e-01 1.21442795e+00 5.39884269e-01
6.31272733e-01 -1.41429949e+00 -2.83246845e-01 3.54573131e-01
-9.26012516e-01 -1.39254248e+00 -9.39784944e-01 2.06160977e-01
-4.84549880e-01 -1.19227326e+00 -3.62951905e-01 -8.02406073e-01
-7.60513023e-02 -3.44865739e-01 1.34505057e+00 -4.69924569e-01
2.58061796e-01 3.28939915e-01 -6.58615768e-01 -4.87747550e-01
-6.62633359e-01 3.97649139e-01 -1.22664459e-01 -3.78907651e-01
4.19921100e-01 -6.27141654e-01 -1.45902392e-02 1.31898835e-01
-6.95895195e-01 -1.03775010e-01 4.04952347e-01 7.43699431e-01
5.24014533e-01 -2.13437602e-01 4.80878055e-01 -1.12786520e+00
4.12682086e-01 -3.60298693e-01 -5.28544247e-01 4.14514631e-01
-1.82615533e-01 2.69520283e-01 7.27075934e-01 -6.43764317e-01
-1.21038496e+00 -1.34568214e-01 -3.37245613e-01 -1.09301783e-01
4.98118773e-02 5.10082066e-01 -1.58456028e-01 5.43711007e-01
7.12189019e-01 2.73360312e-01 -5.20038605e-01 -3.32591832e-01
5.09711087e-01 4.63809520e-01 1.04825568e+00 -7.56452918e-01
8.99595380e-01 1.74259514e-01 -3.14084858e-01 -7.24227786e-01
-1.19069207e+00 -4.50431168e-01 -5.04085064e-01 8.38658437e-02
9.62970853e-01 -8.00694108e-01 -3.70759994e-01 6.76601112e-01
-1.62304544e+00 -9.19755995e-01 -4.43312168e-01 1.46844000e-01
-5.33332109e-01 4.46533740e-01 -9.28541541e-01 -6.82764769e-01
-8.65084291e-01 -2.94657975e-01 1.12969697e+00 -2.56669186e-02
-8.32746089e-01 -1.07552600e+00 4.07803565e-01 3.16220284e-01
9.34042037e-04 3.68402332e-01 7.49463379e-01 -1.04500723e+00
-2.34680593e-01 -1.01008527e-01 3.83241177e-01 2.85617679e-01
-1.12018943e-01 -1.05149418e-01 -7.94654429e-01 -3.97961065e-02
-1.91759244e-01 -4.45919752e-01 6.71447456e-01 2.18777061e-01
6.16540968e-01 -8.77547443e-01 -3.82514834e-01 3.83946836e-01
8.39437902e-01 4.88120526e-01 7.59526908e-01 3.40833336e-01
4.01901573e-01 5.39245844e-01 9.80978966e-01 1.91200525e-01
5.83918333e-01 7.52236724e-01 -7.70300478e-02 3.02549809e-01
7.22040460e-02 -7.09243178e-01 7.93923438e-01 1.13685274e+00
8.06030631e-02 -5.73270500e-01 -1.19352770e+00 9.43776250e-01
-2.06935668e+00 -1.08643532e+00 1.22623546e-02 1.68075132e+00
1.16713595e+00 1.87310651e-02 5.78441545e-02 5.45218494e-03
6.12018347e-01 4.45729792e-01 -8.26234445e-02 -4.51030284e-01
-2.62575865e-01 4.44791198e-01 4.69951093e-01 5.91696978e-01
-1.31892395e+00 1.63353145e+00 5.36863136e+00 9.73587632e-01
-9.53325629e-01 4.77925628e-01 1.64850473e-01 2.67452031e-01
-2.67298162e-01 2.70589828e-01 -6.59431458e-01 4.24071401e-01
1.14443982e+00 -5.15624225e-01 4.46880162e-01 4.63957369e-01
1.59675896e-01 1.13866657e-01 -9.37433958e-01 7.53834426e-01
1.25859510e-02 -1.16304362e+00 -2.73563892e-01 -1.74102515e-01
6.19157612e-01 1.75960496e-01 -3.35682124e-01 6.86669648e-01
9.34171259e-01 -7.36015320e-01 1.04320908e+00 3.14756632e-01
9.01588261e-01 -6.35691643e-01 9.94056284e-01 5.49668968e-01
-1.32988453e+00 9.34435204e-02 2.34363392e-01 2.17376322e-01
3.82821858e-01 5.32503009e-01 -1.08075583e+00 8.06334317e-01
6.94089592e-01 9.24762011e-01 -2.58706421e-01 4.36672390e-01
-9.48304772e-01 1.20302522e+00 -3.82931292e-01 2.23923344e-02
2.87949413e-01 2.16334816e-02 7.79408097e-01 1.50402582e+00
4.33084160e-01 4.22015041e-01 2.82024201e-02 5.79777062e-01
-2.08250716e-01 8.07451606e-02 -6.95963740e-01 -4.75964725e-01
8.16750944e-01 1.14662695e+00 -3.79781246e-01 -2.80037880e-01
5.53582832e-02 1.29412186e+00 2.78238982e-01 2.60102659e-01
-9.39128876e-01 -2.50085622e-01 3.17393303e-01 1.37054905e-01
2.49548733e-01 -3.26281846e-01 1.25099644e-01 -1.27945650e+00
1.46141248e-02 -9.76897776e-01 9.33750510e-01 -8.85828197e-01
-1.11260104e+00 9.22932208e-01 1.73492372e-01 -1.13452840e+00
-8.24128449e-01 -2.26227283e-01 -7.01924980e-01 5.53915322e-01
-9.08985436e-01 -1.65748870e+00 1.45410433e-01 5.00239253e-01
5.22241890e-01 -2.17389822e-01 9.27194297e-01 2.40067784e-02
-4.12062943e-01 5.40117443e-01 -3.00188869e-01 5.39351642e-01
1.00246286e+00 -1.34828198e+00 5.71580708e-01 1.27046096e+00
9.83445644e-02 5.53672373e-01 8.63153279e-01 -6.73116267e-01
-8.93792033e-01 -1.60743105e+00 1.94413960e+00 -5.89432001e-01
9.17991281e-01 -4.14012581e-01 -8.29093039e-01 1.36621618e+00
3.91968936e-01 -3.36321950e-01 4.38996762e-01 1.83810517e-01
-4.34449553e-01 -8.84365886e-02 -1.08513129e+00 5.61934948e-01
1.10076511e+00 -6.91980660e-01 -1.10354900e+00 4.18893844e-01
8.94809783e-01 -4.82239932e-01 -9.68249142e-01 2.60895520e-01
2.33431399e-01 -5.10713875e-01 4.96787995e-01 -3.70940834e-01
3.27234417e-01 -2.43671805e-01 -1.07754104e-01 -1.28412223e+00
-2.36400858e-01 -1.34704876e+00 -9.50732827e-02 1.70995760e+00
4.13127989e-01 -7.07826853e-01 3.12167645e-01 1.85403436e-01
-3.25622439e-01 -2.48579204e-01 -1.18153179e+00 -9.88475919e-01
3.52865547e-01 -7.25027263e-01 3.61683577e-01 1.00296044e+00
4.28548574e-01 7.09819257e-01 -5.89503109e-01 -8.05792958e-02
1.60134003e-01 -7.78709166e-03 5.66188812e-01 -9.89685893e-01
-3.66803378e-01 -7.96824321e-02 -2.27224812e-01 -8.96569192e-01
5.97603619e-01 -1.04665434e+00 2.84334004e-01 -1.50639570e+00
1.88525189e-02 -2.41042048e-01 2.03548670e-01 1.25042188e+00
7.78908357e-02 6.44595250e-02 1.33985698e-01 2.78317928e-01
-6.98989213e-01 6.02490485e-01 9.20799851e-01 -3.99965465e-01
-2.13175282e-01 -9.25978422e-02 -5.31937182e-01 6.19340658e-01
9.25924599e-01 -7.18491912e-01 -1.80705771e-01 -2.04770267e-01
1.09625220e-01 3.49274039e-01 2.09205613e-01 -6.96144879e-01
7.45126233e-02 -2.72384554e-01 -3.31965446e-01 -4.83621657e-01
2.88328320e-01 -3.45725238e-01 2.36222789e-01 4.29809451e-01
-9.76998284e-02 2.05646262e-01 4.85382199e-01 1.98596686e-01
-3.16144556e-01 2.31814478e-03 6.30184174e-01 -6.96434677e-02
-7.75561213e-01 1.41342089e-01 -5.71568668e-01 6.21245265e-01
8.88626993e-01 4.34854001e-01 -5.34826756e-01 -6.23448968e-01
-5.09697080e-01 2.77755409e-01 2.79984057e-01 4.96767759e-01
1.37804896e-01 -1.19332874e+00 -8.03044438e-01 -2.09958971e-01
8.75412375e-02 -1.22972913e-01 2.29514074e-02 8.97082984e-01
-2.94161469e-01 6.01290703e-01 2.07172349e-01 -3.03385347e-01
-1.29806113e+00 3.07253391e-01 4.03298646e-01 -9.91825044e-01
-5.61388969e-01 6.56081140e-01 8.43706802e-02 -5.60306013e-01
-2.04029337e-01 -4.39161122e-01 8.65792707e-02 -8.77784267e-02
2.54188150e-01 4.19193543e-02 -1.32202543e-02 -7.16116488e-01
-6.35375679e-01 2.41877243e-01 1.88472532e-02 -5.42546928e-01
1.26492703e+00 2.18611807e-01 -3.35538179e-01 4.71535504e-01
7.71102488e-01 3.95738184e-01 -7.27671564e-01 -2.94158041e-01
1.99448571e-01 2.37522617e-01 -3.97393107e-01 -1.04829681e+00
-6.10699832e-01 4.61408883e-01 -9.51275155e-02 1.46839783e-01
1.03580630e+00 5.00187397e-01 1.20980823e+00 4.06933755e-01
4.32776868e-01 -9.55335081e-01 -7.35330358e-02 1.28519726e+00
9.82154429e-01 -7.23848641e-01 -5.02800941e-01 -3.00936669e-01
-1.01449215e+00 7.51166701e-01 3.94326955e-01 1.17237553e-01
2.68891811e-01 4.29819584e-01 3.16346258e-01 -1.14631258e-01
-1.09564519e+00 -5.87283522e-02 3.73072624e-01 5.61859071e-01
6.66560888e-01 2.82639027e-01 -6.72785938e-01 1.18261194e+00
-7.64922857e-01 -1.09656379e-01 2.68831193e-01 8.32856596e-01
-7.58605748e-02 -1.42013633e+00 -1.61702737e-01 6.68184482e-04
-7.55749702e-01 -2.51219243e-01 -8.60428095e-01 8.86964023e-01
1.29214942e-01 9.79672551e-01 -1.99540630e-01 -5.62127829e-01
3.42757702e-01 2.56776571e-01 4.22083557e-01 -5.89453816e-01
-7.54613817e-01 -1.98923424e-02 7.61246622e-01 -4.35411841e-01
-2.95466065e-01 -9.57200527e-01 -1.33169687e+00 -3.29558074e-01
1.93192646e-01 6.05804861e-01 -1.14379004e-01 1.19329977e+00
1.08642668e-01 3.76912773e-01 2.15367526e-01 -3.49774390e-01
-2.55095690e-01 -1.34752226e+00 -3.75377595e-01 4.03140992e-01
1.49063587e-01 -2.64623195e-01 -1.71328142e-01 2.57909626e-01] | [10.224520683288574, 9.297150611877441] |
f715982f-2ed4-4bad-9656-2ebcbc8d6e95 | use-your-head-improving-long-tail-video | 2304.01143 | null | https://arxiv.org/abs/2304.01143v1 | https://arxiv.org/pdf/2304.01143v1.pdf | Use Your Head: Improving Long-Tail Video Recognition | This paper presents an investigation into long-tail video recognition. We demonstrate that, unlike naturally-collected video datasets and existing long-tail image benchmarks, current video benchmarks fall short on multiple long-tailed properties. Most critically, they lack few-shot classes in their tails. In response, we propose new video benchmarks that better assess long-tail recognition, by sampling subsets from two datasets: SSv2 and VideoLT. We then propose a method, Long-Tail Mixed Reconstruction, which reduces overfitting to instances from few-shot classes by reconstructing them as weighted combinations of samples from head classes. LMR then employs label mixing to learn robust decision boundaries. It achieves state-of-the-art average class accuracy on EPIC-KITCHENS and the proposed SSv2-LT and VideoLT-LT. Benchmarks and code at: tobyperrett.github.io/lmr | ['Dima Damen', 'Majid Mirmehdi', 'Tilo Burghardt', 'Saptarshi Sinha', 'Toby Perrett'] | 2023-04-03 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Perrett_Use_Your_Head_Improving_Long-Tail_Video_Recognition_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Perrett_Use_Your_Head_Improving_Long-Tail_Video_Recognition_CVPR_2023_paper.pdf | cvpr-2023-1 | ['video-recognition'] | ['computer-vision'] | [ 3.22860032e-01 -4.51641262e-01 -8.38748753e-01 -6.11979961e-01
-1.27224636e+00 -6.05329514e-01 6.85352266e-01 -6.31013393e-01
-9.44445506e-02 6.51307702e-01 2.32402846e-01 -2.32627451e-01
6.44752523e-03 -1.13287129e-01 -1.08578658e+00 -7.70244122e-01
5.47683276e-02 3.85967702e-01 5.27940512e-01 6.40168607e-01
-4.50501144e-02 6.77370373e-03 -1.88802266e+00 7.66272426e-01
2.81796694e-01 1.48451746e+00 -5.67699932e-02 1.07553720e+00
-1.43007845e-01 1.41783750e+00 -2.83873945e-01 -5.39747596e-01
2.41291150e-01 -5.93520284e-01 -3.90833855e-01 4.08117563e-01
1.08421648e+00 -5.64407289e-01 -6.14436567e-01 9.83258963e-01
4.38618153e-01 2.24188402e-01 1.07207191e+00 -1.63877821e+00
-2.33688742e-01 5.15694022e-01 -7.00378895e-01 4.31024939e-01
5.98310411e-01 3.02928388e-01 9.62925673e-01 -1.18658960e+00
7.53059983e-01 1.00713468e+00 9.28543806e-01 6.89066410e-01
-1.24591970e+00 -9.98875797e-01 9.97147113e-02 4.10993934e-01
-1.30190670e+00 -9.40714121e-01 2.93280303e-01 -7.57644475e-01
7.20293760e-01 4.70191300e-01 2.86492705e-01 1.91920817e+00
1.18093438e-01 1.35814583e+00 1.07580256e+00 7.88000152e-02
3.36749911e-01 -1.53711811e-01 3.69444817e-01 5.87676823e-01
2.86917895e-01 1.15400977e-01 -8.10047925e-01 -2.17982665e-01
3.07263374e-01 8.43169987e-02 -4.13750529e-01 -4.46299940e-01
-1.05000842e+00 6.26660407e-01 -4.66430336e-01 -3.81036013e-01
-7.49814659e-02 4.15943027e-01 8.21257949e-01 2.86638349e-01
5.56085825e-01 -4.02610362e-01 -3.67736816e-01 -5.10089338e-01
-1.57651424e+00 1.01839229e-01 8.41794431e-01 1.63165247e+00
5.73420048e-01 1.85843185e-01 -6.77085280e-01 6.95074737e-01
1.39385059e-01 6.09349310e-01 4.89251077e-01 -1.29307938e+00
3.47861022e-01 -1.32976755e-01 -5.69605567e-02 -3.06468666e-01
-3.61130349e-02 -2.93509692e-01 -5.79014003e-01 1.14182532e-01
3.42563778e-01 -1.58991013e-02 -1.28751695e+00 1.35131586e+00
-7.24203289e-02 9.69802618e-01 -1.35577977e-01 8.12461436e-01
1.29823565e+00 7.32383370e-01 2.46748179e-01 -3.50598991e-01
1.10247886e+00 -1.45822215e+00 -4.26814169e-01 -1.38618559e-01
5.66691279e-01 -4.99623954e-01 1.00466168e+00 5.04853249e-01
-1.13523114e+00 -6.07998848e-01 -7.01465428e-01 2.84153402e-01
2.52511371e-02 -2.27613598e-02 4.25520182e-01 7.24750459e-01
-7.70438492e-01 5.23886681e-01 -5.82168162e-01 -4.18435991e-01
7.40242064e-01 -1.65833101e-01 -3.27589899e-01 -3.68943810e-01
-6.20748937e-01 2.35405624e-01 6.71928227e-02 -4.82735425e-01
-1.63038766e+00 -8.39670420e-01 -8.51511776e-01 -4.48414087e-02
6.91124022e-01 -5.25701582e-01 1.59762824e+00 -1.18046057e+00
-1.16504824e+00 1.13740122e+00 -1.86259121e-01 -5.04347086e-01
8.10403943e-01 -2.83727467e-01 -5.63683569e-01 3.61564428e-01
1.20343126e-01 5.35326838e-01 1.25596130e+00 -1.16161978e+00
-7.64801025e-01 4.49957699e-02 -4.01864290e-01 -1.83442727e-01
-9.62460041e-02 2.22131133e-01 -8.51314664e-01 -8.50477636e-01
-5.97477078e-01 -9.24579799e-01 2.27596730e-01 -4.21429388e-02
-3.16244036e-01 -1.54683694e-01 8.56593072e-01 -4.12531734e-01
1.29452920e+00 -2.30809736e+00 -1.18338451e-01 -2.17165872e-01
2.39922673e-01 2.66401887e-01 -3.62336218e-01 1.34976670e-01
-9.93894339e-02 -2.89807469e-02 9.60091129e-02 -6.96093559e-01
2.68802822e-01 2.76821017e-01 -4.99401450e-01 6.95604086e-01
-2.01477230e-01 8.81705821e-01 -8.50171506e-01 -7.48974800e-01
1.43262565e-01 6.66280910e-02 -3.93833131e-01 4.17303443e-01
-4.50605482e-01 4.43487018e-02 -4.06222716e-02 9.97133493e-01
7.42834330e-01 -3.34042162e-01 -2.37815082e-01 -2.46730685e-01
3.00901234e-01 -3.30298930e-01 -7.93384492e-01 1.58733475e+00
1.04648814e-01 8.61081183e-01 -1.10673439e-02 -8.63469779e-01
3.92719954e-01 3.61209661e-01 4.95976448e-01 -3.80139738e-01
2.61079401e-01 1.32019907e-01 -6.90910995e-01 -7.23348141e-01
3.09490889e-01 -1.20857999e-01 4.78980504e-02 1.11168779e-01
4.54701334e-01 2.33451232e-01 5.40198684e-01 3.58885556e-01
1.15032423e+00 5.39023221e-01 5.52221574e-02 -2.08322369e-02
1.61442503e-01 -3.15470099e-01 6.97311223e-01 1.02873492e+00
-5.11349916e-01 1.24531317e+00 5.17503381e-01 -1.89154208e-01
-9.65774357e-01 -1.38648105e+00 -5.45751713e-02 1.57435131e+00
-9.95582938e-02 -6.22440696e-01 -7.68612385e-01 -9.66939688e-01
5.92427775e-02 1.00177419e+00 -6.79693758e-01 -2.39480332e-01
-3.21093313e-02 -3.86190504e-01 9.20038044e-01 7.33164608e-01
9.93676782e-02 -8.11845660e-01 -5.29267848e-01 -3.09118688e-01
-2.10982144e-01 -1.45620608e+00 -8.39837015e-01 4.10122424e-01
-6.02746785e-01 -1.12746370e+00 -1.12071347e+00 -6.77544951e-01
2.30152711e-01 3.99640799e-01 1.57483923e+00 -2.26303548e-01
-3.63327205e-01 7.70569921e-01 -6.28066838e-01 3.10754795e-02
-2.39196628e-01 -2.63127238e-01 8.99669304e-02 1.37324408e-01
6.42723739e-01 -2.83762246e-01 -4.71629471e-01 6.88280761e-01
-7.16489732e-01 -3.62804644e-02 4.28348988e-01 7.35400438e-01
7.35187173e-01 -3.02142531e-01 4.67442244e-01 -9.08458233e-01
-6.44028410e-02 -8.82419229e-01 -2.68754452e-01 2.95801908e-01
-3.76248926e-01 -4.26332444e-01 3.77782553e-01 -7.28163004e-01
-1.07701159e+00 1.49213988e-02 -1.36164268e-02 -1.18904364e+00
-3.85387659e-01 1.48468792e-01 5.89560764e-03 1.63987771e-01
4.65412647e-01 4.19118702e-01 -1.00951612e-01 -3.90613198e-01
2.12554649e-01 7.41111159e-01 7.35544205e-01 -7.18933702e-01
3.31423491e-01 3.93748075e-01 -3.07329744e-01 -1.07436526e+00
-1.15516865e+00 -8.90771687e-01 -2.14586079e-01 -5.97555339e-01
9.21703398e-01 -1.27882457e+00 -3.88880998e-01 7.25416303e-01
-5.01170158e-01 -8.48184049e-01 -5.62083185e-01 4.22736228e-01
-1.13002455e+00 5.34793913e-01 -7.74902344e-01 -8.33219111e-01
-1.30899087e-01 -9.98905122e-01 1.33744502e+00 1.02300830e-01
-1.86025605e-01 -5.90543330e-01 3.99774015e-02 5.51419437e-01
3.72961164e-02 4.45109494e-02 2.99416304e-01 -6.86307788e-01
-5.09904683e-01 -2.72565454e-01 -3.51320088e-01 6.78356111e-01
-3.72669458e-01 3.18077385e-01 -1.10542023e+00 -5.05873978e-01
-3.90640587e-01 -8.17806423e-01 1.37558627e+00 6.26893580e-01
1.35908723e+00 -3.77563797e-02 -2.79464960e-01 1.07824886e+00
1.19532585e+00 -4.71893176e-02 7.43808031e-01 6.14012331e-02
5.81183255e-01 9.97848958e-02 5.79759836e-01 7.86339462e-01
1.19782604e-01 4.28596258e-01 2.54280746e-01 3.93760800e-01
-6.15315378e-01 -2.59363800e-01 9.30519581e-01 6.04367375e-01
-2.62372624e-02 -6.70596540e-01 -6.69853568e-01 3.31913859e-01
-1.93177712e+00 -1.32070792e+00 -2.71460652e-01 2.10904884e+00
4.21515763e-01 3.70907128e-01 5.89665234e-01 -2.03367412e-01
6.77287281e-01 3.96676838e-01 -5.48947990e-01 1.07768334e-01
-2.77583748e-01 -1.95679754e-01 7.51116991e-01 -8.46162215e-02
-1.35195589e+00 7.92808175e-01 6.83320808e+00 1.32346463e+00
-5.63146174e-01 2.06738174e-01 7.30523288e-01 -5.70814371e-01
-5.26395924e-02 -2.26463765e-01 -1.07898808e+00 8.32229137e-01
1.29345274e+00 -7.24464878e-02 2.72647202e-01 1.03747797e+00
-1.12682395e-01 -1.23784229e-01 -1.35734940e+00 1.30836701e+00
5.64150214e-01 -1.31395924e+00 1.25420451e-01 -1.43414989e-01
8.99919510e-01 3.24423790e-01 1.48248419e-01 7.22452819e-01
2.35378742e-01 -7.71042705e-01 1.25932276e+00 7.38007009e-01
1.24669564e+00 -4.92590666e-01 6.44147336e-01 3.87152791e-01
-1.23575127e+00 -1.97810918e-01 -4.02307689e-01 1.54020786e-01
1.60454929e-01 4.53958482e-01 -1.40073985e-01 1.80679426e-01
8.70139897e-01 9.76885557e-01 -6.37950182e-01 1.30517328e+00
1.49729609e-01 1.26110852e+00 -1.44812996e-02 2.02293068e-01
1.87560096e-01 -6.79136366e-02 4.85343844e-01 1.76806116e+00
4.80610520e-01 -5.51929884e-02 5.81196427e-01 2.84026146e-01
-1.98763326e-01 -3.01367581e-01 -6.12947702e-01 -2.02989519e-01
9.82889533e-02 1.12203157e+00 -8.89807403e-01 -8.71231735e-01
-9.16419148e-01 9.52995360e-01 4.76339981e-02 5.70479512e-01
-1.36518633e+00 -4.80876826e-02 7.47106791e-01 2.13206574e-01
7.56245434e-01 9.85598713e-02 1.48406014e-01 -1.67308426e+00
-3.80470932e-01 -9.68478560e-01 7.35504627e-01 -9.42877889e-01
-1.48256254e+00 2.53753155e-01 1.79112807e-01 -1.45765042e+00
-8.86167884e-02 -6.23820782e-01 -4.95662391e-01 -1.24535300e-02
-1.10973001e+00 -1.16270053e+00 -5.98359942e-01 6.41991913e-01
1.12788928e+00 -3.99043947e-01 4.39206839e-01 5.82184732e-01
-6.41366363e-01 7.87915289e-01 3.92688632e-01 2.42343973e-02
9.67178166e-01 -1.02987671e+00 2.99043804e-01 7.33305156e-01
1.81121737e-01 -7.91771114e-02 7.82754362e-01 -6.49620771e-01
-1.46067762e+00 -1.31085670e+00 6.22836202e-02 -6.05152011e-01
7.40940273e-01 -4.82474327e-01 -6.72247469e-01 1.02506316e+00
1.82284042e-01 5.02402127e-01 8.03705037e-01 -2.84137547e-01
-7.10430086e-01 -4.18422520e-02 -7.94105291e-01 4.16100323e-01
1.32185578e+00 -5.83969057e-01 -4.72568929e-01 2.79301673e-01
5.07439554e-01 -2.64121946e-02 -6.55848861e-01 3.19138974e-01
8.25623274e-01 -1.14310181e+00 9.32106376e-01 -7.87353754e-01
5.02588749e-01 -1.13969512e-01 -7.66445994e-01 -9.51388538e-01
-3.69724393e-01 -6.60172641e-01 -7.47237027e-01 1.34727240e+00
1.20581716e-01 -2.08849818e-01 1.33773291e+00 3.48490477e-01
-2.25942269e-01 -3.79636139e-01 -7.62754381e-01 -1.45386660e+00
-1.50948510e-01 -7.08383560e-01 -2.32555382e-02 5.40066779e-01
-2.47221321e-01 2.29693130e-01 -8.69433165e-01 -3.42776865e-01
1.17424607e+00 1.32056445e-01 1.04256749e+00 -9.00705159e-01
-4.44122016e-01 -3.96142870e-01 -5.06698549e-01 -1.35361624e+00
4.32842225e-01 -8.30661893e-01 4.92676526e-01 -1.22458243e+00
8.44976425e-01 6.79499805e-02 -4.74063426e-01 1.04038738e-01
-2.46527791e-02 7.09460676e-01 2.32229948e-01 9.05815363e-02
-1.51966321e+00 5.37007749e-01 8.53322566e-01 -1.83827922e-01
5.01075029e-01 1.20178089e-01 -1.75429583e-01 9.21117127e-01
4.60458249e-01 -4.32417989e-01 -4.12686020e-01 -1.61519665e-02
-4.03320929e-03 1.94928616e-01 4.13195342e-01 -1.33691442e+00
2.14448258e-01 -2.61096507e-01 4.12851900e-01 -9.14101899e-01
4.36244309e-01 -7.03244865e-01 3.77842516e-01 1.49359837e-01
-4.13325161e-01 -3.47978085e-01 -1.40352279e-01 9.18914378e-01
-1.92975178e-02 -3.84999305e-01 8.41724336e-01 -1.83299154e-01
-1.12772810e+00 5.50589502e-01 -7.25925088e-01 6.23246849e-01
1.32619488e+00 -4.44980741e-01 -6.03172719e-01 -6.11660540e-01
-6.89322591e-01 2.56137669e-01 5.98256648e-01 3.96264613e-01
6.81611001e-01 -1.37136245e+00 -7.40209281e-01 1.39162838e-01
3.85655224e-01 -4.76944000e-01 5.52815855e-01 1.18552911e+00
-3.75881076e-01 1.19849585e-01 -6.00962900e-02 -8.03818107e-01
-1.57837903e+00 8.63686860e-01 9.48948599e-03 3.16900611e-01
-7.32443392e-01 9.82309997e-01 5.32634616e-01 -3.46976556e-02
6.90360367e-01 -2.24777758e-01 1.69753864e-01 3.34043145e-01
7.60439634e-01 7.13717937e-01 -2.05690265e-01 -7.70499945e-01
-3.24231058e-01 4.77383107e-01 1.55521974e-01 2.21938998e-01
1.20683575e+00 -1.56523332e-01 4.64597732e-01 8.35533559e-01
1.55521774e+00 -1.37965441e-01 -1.79506791e+00 -8.76036286e-02
-3.49330008e-02 -6.67110622e-01 -1.20802641e-01 -5.03359616e-01
-9.76748824e-01 7.72839367e-01 5.21996796e-01 2.45346874e-02
9.80913281e-01 2.48836145e-01 1.09669542e+00 1.64812699e-01
2.63694197e-01 -1.03420663e+00 3.71618301e-01 4.46600467e-01
4.61610109e-01 -1.34185922e+00 4.55169566e-02 -2.10943282e-01
-7.84090459e-01 1.01661277e+00 4.68203157e-01 -1.92732200e-01
6.11685514e-01 5.34873903e-01 -9.29473788e-02 3.66539434e-02
-1.21531272e+00 -2.96343476e-01 3.43424082e-01 6.17042065e-01
1.88165411e-01 1.98574498e-01 2.03022823e-01 7.21283436e-01
3.43398929e-01 3.93160254e-01 6.49598897e-01 7.99931347e-01
-6.25356138e-01 -3.87256980e-01 -1.79630294e-01 1.01694298e+00
-5.75335264e-01 4.40802015e-02 4.68031988e-02 3.79723012e-01
-7.42787793e-02 8.72134387e-01 2.41293795e-02 -6.53019428e-01
4.27885093e-02 3.18162322e-01 4.09309953e-01 -4.53309596e-01
-9.17283371e-02 2.77862519e-01 5.04997253e-01 -7.88592756e-01
-3.85511696e-01 -1.04364181e+00 -6.56897247e-01 -4.62618053e-01
-2.88488299e-01 1.33570191e-02 2.42673934e-01 5.45693696e-01
9.08894315e-02 5.11889040e-01 3.58307302e-01 -1.25891578e+00
-7.87961543e-01 -7.08679557e-01 -8.35863829e-01 6.37897849e-01
4.39729124e-01 -7.67512500e-01 -7.40583003e-01 4.94672865e-01] | [9.031224250793457, 0.9206150770187378] |
6fbc0767-b1eb-4aa0-8ddc-ef7ff9aea093 | joint-multiview-segmentation-and-localization | null | null | http://openaccess.thecvf.com/content_cvpr_2016/html/Zhang_Joint_Multiview_Segmentation_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Zhang_Joint_Multiview_Segmentation_CVPR_2016_paper.pdf | Joint Multiview Segmentation and Localization of RGB-D Images Using Depth-Induced Silhouette Consistency | In this paper, we propose an RGB-D camera localization approach which takes an effective geometry constraint, i.e. silhouette consistency, into consideration. Unlike existing approaches which usually assume the silhouettes are provided, we consider more practical scenarios and generate the silhouettes for multiple views on the fly. To obtain a set of accurate silhouettes, precise camera poses are required to propagate segmentation cues across views. To perform better localization, accurate silhouettes are needed to constrain camera poses. Therefore the two problems are intertwined with each other and require a joint treatment. Facilitated by the available depth, we introduce a simple but effective silhouette consistency energy term that binds traditional appearance-based multiview segmentation cost and RGB-D frame-to-frame matching cost together. Optimization of the problem w.r.t. binary segmentation masks and camera poses naturally fits in the graph cut minimization framework and the Gauss-Newton non-linear least-squares method respectively. Experiments show that the proposed approach achieves state-of-the-arts performance on both tasks of image segmentation and camera localization. | ['Zhiwei Li', 'Yong Rui', 'Rui Cai', 'Hongyang Chao', 'Chi Zhang'] | 2016-06-01 | null | null | null | cvpr-2016-6 | ['camera-localization'] | ['computer-vision'] | [ 3.21904182e-01 -1.12754531e-01 -8.05190280e-02 -5.83498538e-01
-1.03941596e+00 -8.25571060e-01 1.21436819e-01 5.05632162e-03
-4.39611375e-01 3.22705835e-01 -4.09185886e-01 2.70651244e-02
6.86893836e-02 -4.87871230e-01 -8.40520859e-01 -5.82736313e-01
5.46823442e-01 4.58259493e-01 3.80555093e-01 5.87673709e-02
4.22326028e-01 7.09120214e-01 -1.26880848e+00 -3.62299919e-01
9.24525380e-01 1.08771789e+00 3.21694672e-01 6.31028473e-01
2.31686383e-02 3.21512163e-01 -6.78259283e-02 -4.89160717e-01
6.27579808e-01 -2.05532685e-01 -5.38240731e-01 9.55778897e-01
7.99200475e-01 -3.87317985e-01 3.98281515e-02 1.17310059e+00
2.15060741e-01 2.59875834e-01 4.89351749e-01 -1.27958405e+00
-2.69104652e-02 -1.07806496e-01 -1.28701186e+00 -1.59475535e-01
7.40789115e-01 -9.47569385e-02 8.73789907e-01 -8.84105027e-01
6.89259648e-01 1.07970059e+00 6.40163064e-01 2.29091123e-01
-1.25603199e+00 -2.19278857e-01 5.06842732e-01 -2.54675448e-01
-1.55060744e+00 -2.57173151e-01 1.12789547e+00 -4.33732569e-01
4.87273276e-01 2.43959546e-01 6.85306013e-01 4.50378120e-01
-5.41211888e-02 6.41985118e-01 9.70011294e-01 -4.94577080e-01
1.76276848e-01 2.00405374e-01 -1.27747893e-01 9.33851123e-01
2.15456769e-01 -2.33641103e-01 -3.51204962e-01 -4.55505326e-02
1.24660981e+00 3.03645194e-01 -3.51737261e-01 -1.07868075e+00
-1.23531067e+00 7.27807999e-01 3.20573598e-01 -5.82920648e-02
-2.51256615e-01 2.31347203e-01 -1.02712298e-02 -3.02911907e-01
3.28123897e-01 6.39708266e-02 -3.69855970e-01 2.28350192e-01
-1.10370719e+00 7.98273608e-02 4.77164537e-01 1.30760038e+00
1.11298347e+00 -9.96152163e-02 5.02844036e-01 5.94215870e-01
5.75969279e-01 5.81915021e-01 -1.46579981e-01 -1.36951745e+00
5.39816916e-01 7.13141322e-01 2.41458759e-01 -1.32343185e+00
-8.58795643e-02 -1.16457105e-01 -6.32751346e-01 1.35414585e-01
5.58906376e-01 -3.62066552e-02 -8.99559677e-01 1.26297271e+00
9.37821805e-01 2.20601350e-01 -2.26382524e-01 1.17341971e+00
3.00205410e-01 4.39971656e-01 -4.42022502e-01 -3.23794425e-01
1.17678475e+00 -8.90054107e-01 -6.95450425e-01 -4.40102845e-01
1.35598406e-01 -9.74128187e-01 7.11767256e-01 4.09319699e-01
-1.35707295e+00 -4.03100342e-01 -1.00246239e+00 -2.38822386e-01
-1.12212539e-01 2.34322831e-01 5.86269498e-01 5.78834116e-01
-9.76682663e-01 3.01840037e-01 -9.73826468e-01 -2.28507519e-01
-4.93686497e-02 5.71385264e-01 -5.08065283e-01 -1.19046859e-01
-5.30580461e-01 7.73568034e-01 2.25998774e-01 3.16026300e-01
-6.47440076e-01 -2.35034809e-01 -1.00369155e+00 -3.55944395e-01
7.78122604e-01 -6.96281731e-01 1.03533673e+00 -9.11373556e-01
-1.49896693e+00 1.05937469e+00 -2.68223852e-01 9.80871450e-03
6.75508797e-01 -1.48017734e-01 3.60825747e-01 5.18791497e-01
8.12592134e-02 6.59290016e-01 9.11997259e-01 -1.65276945e+00
-7.24482536e-01 -4.99708742e-01 1.28478080e-01 5.87606013e-01
-2.18261816e-02 -1.73955947e-01 -1.19300103e+00 -2.76351899e-01
6.98055565e-01 -8.99957180e-01 -4.93626803e-01 2.51966506e-01
-4.57796365e-01 1.84964418e-01 7.26723254e-01 -7.25779116e-01
9.22846973e-01 -1.80896735e+00 5.45181870e-01 4.54997480e-01
1.79195613e-01 -3.08272660e-01 3.67146075e-01 8.71171132e-02
1.66141436e-01 -1.65638998e-01 -3.63048404e-01 -7.35183895e-01
-2.30473951e-01 2.19845787e-01 2.68642694e-01 1.04978180e+00
-1.41382050e-02 5.29750526e-01 -8.14014792e-01 -8.72880876e-01
7.05158770e-01 5.74251056e-01 -5.28204322e-01 2.84732372e-01
-1.82326481e-01 4.52053487e-01 -6.16461217e-01 8.12471747e-01
1.00539315e+00 -1.52057573e-01 5.23222005e-03 -5.56553185e-01
-2.36286715e-01 -4.30930883e-01 -1.58951354e+00 2.15658450e+00
-4.03590620e-01 2.11621866e-01 5.06408513e-01 -8.31240177e-01
7.79891193e-01 1.26398325e-01 7.45868146e-01 -1.07479384e-02
3.61969292e-01 2.70119280e-01 -6.24964893e-01 -1.81993827e-01
3.56122702e-01 -2.58165924e-03 -2.07645819e-03 1.22534253e-01
-6.38291612e-02 -5.78388870e-01 4.38313298e-02 1.14881262e-01
4.86643255e-01 4.50050503e-01 5.21463811e-01 -5.47184469e-03
6.84807658e-01 1.17314458e-01 7.22323298e-01 2.24070951e-01
3.84243988e-02 1.02197373e+00 1.95551425e-01 -2.58039147e-01
-1.12269843e+00 -8.87218058e-01 -9.65295658e-02 3.08909208e-01
6.81091726e-01 -9.29604620e-02 -9.91411865e-01 -7.21777380e-01
-1.67726547e-01 2.47418776e-01 -3.49593461e-01 4.80817109e-01
-6.33238375e-01 -2.83841282e-01 -1.52476311e-01 3.73329312e-01
4.44204032e-01 -2.15607539e-01 -8.61483037e-01 4.03637625e-02
-1.35442108e-01 -1.47872353e+00 -9.27086651e-01 1.85627207e-01
-8.92216563e-01 -1.20206428e+00 -9.02250707e-01 -6.71681762e-01
1.16664624e+00 5.70512950e-01 7.95762599e-01 9.83292460e-02
-3.29954356e-01 7.11872101e-01 -2.32791200e-01 1.94874734e-01
9.02129188e-02 -1.08810820e-01 -9.06408057e-02 1.43710330e-01
-8.64179656e-02 -5.23972154e-01 -7.31105030e-01 5.80912888e-01
-8.68537366e-01 2.96854913e-01 4.99011248e-01 5.58472812e-01
1.16078198e+00 -7.67914876e-02 -1.29577503e-01 -7.84214854e-01
-1.22148313e-01 -1.27333533e-02 -1.02587295e+00 4.05925125e-01
-3.54988188e-01 -2.17608243e-01 3.43077511e-01 -3.24445158e-01
-9.53978598e-01 9.10536349e-01 1.03909619e-01 -9.68170464e-01
-1.21054254e-01 8.16236064e-02 -4.34418619e-01 -4.97921139e-01
5.45423143e-02 2.26153716e-01 -8.60220119e-02 -2.94430494e-01
6.67772949e-01 5.22559822e-01 5.27329087e-01 -4.66390073e-01
8.65614235e-01 6.99737132e-01 3.97115536e-02 -6.85716093e-01
-6.92748606e-01 -8.76401246e-01 -1.16900337e+00 -5.07054269e-01
1.21029925e+00 -9.72807348e-01 -6.50104642e-01 2.97713578e-01
-1.39717257e+00 1.92275956e-01 7.24586844e-02 3.71690214e-01
-7.31663585e-01 7.07984746e-01 -2.88697660e-01 -1.02921557e+00
-9.17665213e-02 -1.48155630e+00 1.61057270e+00 2.08430797e-01
2.46815681e-02 -1.07108867e+00 -9.69013497e-02 5.69573522e-01
-2.27471158e-01 6.60579979e-01 3.61428499e-01 -2.76885349e-02
-1.02395260e+00 -3.27616364e-01 -2.22750828e-01 3.19139421e-01
6.82314411e-02 1.53086588e-01 -7.60856271e-01 -1.73127070e-01
1.72081262e-01 1.43733606e-01 3.11129242e-01 5.89720607e-01
8.74694645e-01 -1.46784708e-01 -3.57889593e-01 9.52840507e-01
1.84813428e+00 2.75852352e-01 3.20103467e-01 1.20863356e-01
1.19682217e+00 5.90722919e-01 8.05579662e-01 4.24218684e-01
5.92338860e-01 9.45363998e-01 6.15129292e-01 -3.58355522e-01
2.05602437e-01 -1.87012076e-01 3.63707133e-02 9.85463977e-01
6.68123960e-02 -1.90647259e-01 -9.28213239e-01 4.26171094e-01
-1.87158203e+00 -4.10351306e-01 -4.42927688e-01 2.40009570e+00
7.16460824e-01 -1.09207273e-01 7.39443526e-02 -1.67003479e-02
8.75598907e-01 7.07533862e-03 -6.54350936e-01 -4.42632623e-02
5.28060310e-02 -3.04718971e-01 9.07043517e-01 9.69614446e-01
-1.02342856e+00 8.89306366e-01 4.95069551e+00 7.75459588e-01
-8.36060107e-01 6.69423267e-02 7.79565752e-01 1.78894177e-01
-3.74604642e-01 2.51473576e-01 -8.35948944e-01 2.81795830e-01
1.48621216e-01 2.81142563e-01 5.21390498e-01 8.65365148e-01
9.40624997e-02 -6.19248092e-01 -1.01793873e+00 1.44461036e+00
3.69017512e-01 -9.74214494e-01 -1.94098040e-01 4.55591343e-02
8.87845039e-01 -3.94877613e-01 -1.42246649e-01 -5.67475200e-01
-2.12434772e-02 -5.35948753e-01 8.45904350e-01 4.00330007e-01
7.62149930e-01 -8.13780308e-01 4.24475700e-01 3.31496030e-01
-1.43660223e+00 3.42390805e-01 -2.81609744e-01 3.04853439e-01
7.10439980e-01 6.57312512e-01 -3.99703175e-01 7.43690610e-01
4.75848049e-01 7.44304240e-01 -3.26074064e-01 1.07520759e+00
-1.76053882e-01 -3.19907248e-01 -5.23997724e-01 3.69463295e-01
1.24217816e-01 -7.57385671e-01 4.52083290e-01 7.95264065e-01
3.31294954e-01 1.93910673e-01 5.25047958e-01 8.74642372e-01
1.86882943e-01 -2.20689503e-03 -4.78285521e-01 2.63691247e-01
4.37337667e-01 1.48939681e+00 -1.24473870e+00 -1.16340570e-01
-5.32572627e-01 1.47384059e+00 9.66311321e-02 3.82472247e-01
-9.40508902e-01 -9.88309830e-03 3.31577152e-01 2.60367811e-01
1.77851200e-01 -5.43296218e-01 -3.53510171e-01 -1.32151902e+00
8.23067352e-02 -6.67743027e-01 4.20128787e-03 -7.25707531e-01
-1.02505529e+00 5.48757851e-01 1.52890578e-01 -1.30680525e+00
-1.66134834e-01 -4.92012233e-01 -3.41042548e-01 8.14861953e-01
-1.48920250e+00 -1.25040662e+00 -4.74163532e-01 7.52921402e-01
7.20195711e-01 4.64489222e-01 3.69820893e-01 3.22253376e-01
-6.72842681e-01 2.45941952e-01 -2.85716683e-01 -5.09331711e-02
4.58886415e-01 -1.25279915e+00 1.76388383e-01 9.81844902e-01
1.62109844e-02 5.09508669e-01 5.68979442e-01 -6.60292208e-01
-1.76061833e+00 -8.22258294e-01 4.94039387e-01 -4.53701824e-01
3.65142494e-01 -4.07490730e-01 -4.10699308e-01 6.74783707e-01
1.17314748e-01 9.70682949e-02 3.07946950e-01 -5.58111370e-01
1.57950804e-01 -1.26087651e-01 -1.10875952e+00 4.02391970e-01
8.73799384e-01 -4.80115175e-01 -1.08622052e-01 3.61359298e-01
5.88627338e-01 -9.47284997e-01 -8.37819219e-01 2.65790194e-01
4.10399228e-01 -9.31388080e-01 1.22350121e+00 2.18675256e-01
4.98724617e-02 -7.17476726e-01 -3.06633413e-01 -8.51801336e-01
3.90740573e-01 -7.20415115e-01 1.49193972e-01 1.33092558e+00
1.88695595e-01 -3.02649289e-01 8.61502767e-01 8.84757221e-01
8.19781870e-02 -8.69792640e-01 -9.47100222e-01 -4.29152340e-01
-5.39701819e-01 -3.66983414e-01 1.86360985e-01 9.06638026e-01
-2.71553785e-01 2.39518166e-01 -3.63954097e-01 4.85125452e-01
1.01095545e+00 1.52371302e-01 8.55734527e-01 -1.05455518e+00
-3.21508318e-01 -1.14485435e-01 -3.02417040e-01 -1.53663337e+00
-1.81389842e-02 -4.55044091e-01 2.46977851e-01 -1.51108909e+00
1.74482718e-01 -4.80695516e-01 3.98586810e-01 5.33340909e-02
-1.77761868e-01 1.78165540e-01 1.67055815e-01 8.83630440e-02
-6.92173362e-01 3.71195555e-01 1.18955553e+00 1.80283383e-01
-2.97425061e-01 -8.89183581e-03 -3.33071858e-01 9.57072377e-01
5.61118007e-01 -2.67977625e-01 -5.48811555e-01 -6.15920246e-01
2.42050812e-01 5.90398371e-01 2.37015888e-01 -7.22570956e-01
3.82757097e-01 -2.15875119e-01 4.08079535e-01 -7.24644542e-01
8.11209023e-01 -1.31411195e+00 1.95756987e-01 4.74391785e-03
6.31623492e-02 3.91021937e-01 -2.13692740e-01 7.04053342e-01
-2.20528573e-01 -2.80746907e-01 7.76845634e-01 -4.25769657e-01
-4.96705621e-01 5.40151298e-01 -2.12945733e-02 -7.60436058e-02
1.41890144e+00 -7.51927495e-01 2.56204844e-01 -3.89006674e-01
-7.39763439e-01 2.42961779e-01 9.28445280e-01 6.74550608e-02
9.41988289e-01 -1.17721307e+00 -2.24713489e-01 2.39835173e-01
-2.09787205e-01 4.96946812e-01 1.46714658e-01 1.08351958e+00
-9.01194692e-01 2.74460942e-01 2.68492311e-01 -9.80541646e-01
-1.34585571e+00 5.41590154e-01 4.94251668e-01 1.12899184e-01
-2.61822164e-01 8.61545682e-01 1.96881652e-01 -4.06257689e-01
4.05752510e-01 -5.76129735e-01 1.32003009e-01 6.95366710e-02
5.72221614e-02 4.25213993e-01 -1.46602526e-01 -9.85133052e-01
-4.88951266e-01 1.40941060e+00 1.19945966e-01 -2.23829508e-01
1.03597736e+00 -6.71709955e-01 -1.22538529e-01 3.47951621e-01
1.35459268e+00 1.55247077e-02 -1.66933274e+00 -1.24502573e-02
-1.77796245e-01 -8.29699993e-01 3.54846150e-01 -3.31701726e-01
-1.28664303e+00 9.33958888e-01 4.67313945e-01 -1.00925580e-01
1.12214589e+00 -1.29989177e-01 7.78329968e-01 -9.50531587e-02
5.32005906e-01 -1.06970847e+00 -2.93446891e-02 1.05112560e-01
6.17967486e-01 -1.27661431e+00 2.77427047e-01 -1.05092263e+00
-6.16702974e-01 1.13505316e+00 7.09123433e-01 -2.09288567e-01
3.21279079e-01 1.94975182e-01 3.46160047e-02 -2.72252653e-02
-5.55665884e-03 -1.61547557e-01 4.61370617e-01 3.28576028e-01
2.34693214e-01 -1.74563050e-01 -1.29555106e-01 3.14095281e-02
2.73934454e-01 -3.58170182e-01 3.29754770e-01 8.03480506e-01
-4.52377260e-01 -1.20050204e+00 -7.31083035e-01 -4.07079272e-02
-2.44215623e-01 1.09369531e-01 -3.53024811e-01 7.34015942e-01
2.42832050e-01 9.40983236e-01 -3.38352248e-02 -9.56713110e-02
3.82105917e-01 -3.24165970e-01 6.71684146e-01 -6.05526865e-01
-2.78232008e-01 6.79029286e-01 -2.65641153e-01 -6.63066328e-01
-7.97384679e-01 -7.64147878e-01 -1.26167214e+00 -1.02420077e-01
-7.17018604e-01 -1.09600171e-01 7.31799126e-01 7.27010846e-01
-4.51067425e-02 1.64799526e-01 6.97713792e-01 -1.35369003e+00
-2.09333897e-01 -4.17233527e-01 -6.11352801e-01 9.81814861e-02
3.77983838e-01 -4.70360518e-01 -4.45360124e-01 2.12512493e-01] | [7.798305511474609, -2.3468387126922607] |
11520337-b846-43dc-8dc7-f4690361bdab | neural-nonnegative-matrix-factorization-for | 2303.00058 | null | https://arxiv.org/abs/2303.00058v1 | https://arxiv.org/pdf/2303.00058v1.pdf | Neural Nonnegative Matrix Factorization for Hierarchical Multilayer Topic Modeling | We introduce a new method based on nonnegative matrix factorization, Neural NMF, for detecting latent hierarchical structure in data. Datasets with hierarchical structure arise in a wide variety of fields, such as document classification, image processing, and bioinformatics. Neural NMF recursively applies NMF in layers to discover overarching topics encompassing the lower-level features. We derive a backpropagation optimization scheme that allows us to frame hierarchical NMF as a neural network. We test Neural NMF on a synthetic hierarchical dataset, the 20 Newsgroups dataset, and the MyLymeData symptoms dataset. Numerical results demonstrate that Neural NMF outperforms other hierarchical NMF methods on these data sets and offers better learned hierarchical structure and interpretability of topics. | ['Deanna Needell', 'Denali Molitor', 'Jamie Haddock', 'Joshua Vendrow', 'Mengdi Gao', 'Eli Sadovnik', 'Runyu Zhang', 'Tyler Will'] | 2023-02-28 | null | null | null | null | ['document-classification'] | ['natural-language-processing'] | [ 1.75837755e-01 3.21911097e-01 -4.92539614e-01 -4.34516966e-01
-4.90958899e-01 -2.53133953e-01 2.49281392e-01 1.63259670e-01
-8.51115808e-02 6.49025142e-01 8.82448792e-01 -3.78820479e-01
-6.17611170e-01 -4.96751785e-01 -5.38171589e-01 -6.12142324e-01
-6.07894599e-01 5.66087008e-01 -4.78470206e-01 2.16136411e-01
1.32681355e-01 -9.32724588e-03 -1.37237799e+00 8.64469469e-01
8.88063431e-01 6.05055630e-01 1.64352864e-01 5.42750418e-01
-1.58250816e-02 8.31985354e-01 -6.40563548e-01 -1.77406713e-01
1.51825219e-01 -2.00971574e-01 -8.70694876e-01 5.40038347e-01
5.02338052e-01 2.63278075e-02 -3.02455068e-01 1.02871931e+00
1.08091436e-01 1.20231971e-01 7.34989107e-01 -1.16677749e+00
-6.46472931e-01 9.19282496e-01 -7.37096012e-01 5.01147881e-02
1.11376740e-01 -4.43422079e-01 1.51839375e+00 -8.61756861e-01
6.13135099e-01 1.75881803e+00 7.09448814e-01 4.36905921e-01
-1.45067668e+00 -6.77210987e-01 3.60257506e-01 3.16795819e-02
-1.06230938e+00 1.84951365e-01 4.70604837e-01 -9.20905888e-01
7.23882198e-01 2.33712167e-01 5.61275840e-01 1.02406609e+00
5.29957414e-01 1.05992079e+00 1.10969138e+00 -8.80288109e-02
2.87292570e-01 -3.26361150e-01 7.38529205e-01 7.59936333e-01
1.54461130e-01 -1.15231834e-01 -6.39195979e-01 -6.18624032e-01
6.75345361e-01 1.39350280e-01 1.39819477e-02 -5.41988760e-02
-1.36491346e+00 1.25226855e+00 1.48445606e-01 -4.99484967e-03
-4.94718045e-01 -2.24101260e-01 3.67436737e-01 3.87253582e-01
5.19529700e-01 7.47662902e-01 -4.96589810e-01 3.97778749e-01
-9.84367311e-01 5.94818778e-02 6.17841840e-01 4.40984726e-01
5.39196491e-01 -1.02901869e-01 -2.60400534e-01 1.01821792e+00
5.12576401e-01 7.85815790e-02 6.72782898e-01 -1.04580557e+00
2.80096292e-01 9.98513997e-01 -2.92358935e-01 -1.07773304e+00
-9.59280729e-01 -5.62880337e-01 -1.61401665e+00 -2.42010608e-01
-1.80117220e-01 -1.30723253e-01 -1.07809663e+00 1.71752131e+00
2.99899191e-01 -6.32388368e-02 1.55375436e-01 5.82789600e-01
1.09227991e+00 8.56898129e-01 -1.25741959e-01 -4.78224039e-01
1.34545875e+00 -9.95871067e-01 -8.15305114e-01 7.90102482e-02
4.76856738e-01 -4.28746611e-01 1.00789750e+00 8.97964358e-01
-8.56091321e-01 -5.71745038e-01 -7.89604008e-01 -1.43985689e-01
-2.22344577e-01 2.55123526e-01 1.21383333e+00 1.94005832e-01
-1.28545618e+00 5.04682899e-01 -7.04359055e-01 -2.94406861e-01
2.57848054e-01 6.27844810e-01 -4.45674807e-01 2.42401838e-01
-1.52203453e+00 1.48736894e-01 9.10246670e-01 5.58077917e-02
-1.09637582e+00 -6.02218211e-01 -8.05088401e-01 1.23794422e-01
2.31391154e-02 -1.08229768e+00 9.10806179e-01 -6.14627123e-01
-9.16096747e-01 7.76275575e-01 -3.19727749e-01 -5.87241471e-01
1.15464337e-01 -1.14990368e-01 -2.91567892e-01 -1.30252406e-01
4.49750394e-01 1.00967598e+00 1.07787037e+00 -1.01998889e+00
-6.43698514e-01 -6.80173993e-01 5.84835708e-02 1.82909444e-01
-8.42611372e-01 -2.42758393e-02 -2.71609157e-01 -7.66761184e-01
5.13856471e-01 -9.82060492e-01 -6.50892973e-01 -8.50416303e-01
-9.34560657e-01 -7.19491243e-01 5.84725201e-01 -4.93188381e-01
1.46359503e+00 -1.93218005e+00 6.49143219e-01 1.63765550e-01
8.85520697e-01 -4.60071445e-01 -2.87083592e-02 3.00703108e-01
-3.94570172e-01 1.95756361e-01 -8.79184809e-03 -4.06373382e-01
-1.83750018e-01 2.61965692e-01 -5.14384151e-01 2.41672307e-01
-1.93231046e-01 8.66870821e-01 -7.09500790e-01 -3.94954532e-01
-1.63090274e-01 1.72043934e-01 -8.36320639e-01 8.27033520e-02
-2.77496010e-01 3.13100696e-01 -9.29857865e-02 5.87165594e-01
2.87493646e-01 -8.53531003e-01 3.94860595e-01 -8.22930709e-02
1.59039661e-01 1.52643234e-01 -8.55195165e-01 1.72661996e+00
3.57120186e-02 7.55088627e-01 -5.28477579e-02 -8.40969563e-01
6.66091681e-01 2.42696539e-01 6.54330671e-01 -2.38181546e-01
-1.57576904e-01 -4.26320583e-01 2.15028763e-01 -5.36472619e-01
8.72316718e-01 -4.98663262e-02 -2.56619036e-01 5.51559329e-01
7.63224512e-02 5.02339959e-01 4.73444760e-01 6.44804597e-01
1.14320612e+00 -7.63627291e-01 1.92122683e-01 -5.69150865e-01
1.54874921e-01 -1.50173426e-01 7.19575107e-01 9.01555955e-01
2.42613271e-01 3.71142030e-01 6.33856595e-01 -5.96201718e-01
-8.24654400e-01 -1.06736207e+00 -3.90684724e-01 1.62248540e+00
-3.88178349e-01 -1.06123197e+00 -3.71698380e-01 -6.86175227e-01
1.50623158e-01 1.57258928e-01 -1.19999588e+00 3.89811434e-02
-1.25822380e-01 -1.39379776e+00 3.25832516e-01 4.74882126e-01
1.02210209e-01 -1.02360070e+00 1.08257614e-01 -1.91453546e-01
-1.99484676e-01 -7.81352103e-01 -7.38341436e-02 6.34295106e-01
-1.23948896e+00 -7.90524840e-01 -2.36726269e-01 -7.90548325e-01
9.94107246e-01 3.26819599e-01 1.19100928e+00 -3.17627251e-01
-1.25595942e-01 3.39223854e-02 -1.00532226e-01 -2.23851353e-01
-5.46774685e-01 4.74273562e-01 3.23904932e-01 -2.25002706e-01
2.27364555e-01 -6.54662251e-01 -3.26126486e-01 2.26424381e-01
-1.10356009e+00 4.12932485e-01 6.48247540e-01 1.06075752e+00
5.84448218e-01 6.40557945e-01 5.24623454e-01 -1.23602509e+00
1.06314611e+00 -7.02776551e-01 -5.10274827e-01 1.98812142e-01
-6.90592349e-01 2.98942924e-01 4.90992278e-01 -3.72451395e-01
-6.25639796e-01 -4.30597365e-02 1.27861738e-01 -5.14948726e-01
-1.92425385e-01 9.41421628e-01 -2.30347916e-01 6.40788019e-01
6.81500673e-01 2.33023971e-01 -4.03686762e-01 -5.31166732e-01
4.30615813e-01 6.84213877e-01 5.86892366e-01 -4.45417374e-01
4.91040766e-01 6.59232318e-01 -1.08043976e-01 -9.31389332e-01
-1.35722065e+00 -7.29636967e-01 -7.66164958e-01 -3.88605893e-02
1.09443271e+00 -1.12560308e+00 -7.93605626e-01 1.88698713e-02
-1.20168114e+00 2.56704371e-02 4.56710979e-02 4.02539521e-01
-3.69569451e-01 3.21343660e-01 -9.95089173e-01 -4.61087048e-01
-5.53552985e-01 -9.68903005e-01 1.34560454e+00 -1.91726536e-01
-5.55656254e-01 -1.19839966e+00 3.05280328e-01 6.57214582e-01
-3.21208626e-01 6.67572320e-02 1.60263824e+00 -4.55442816e-01
-1.45286530e-01 1.37381673e-01 -8.81340802e-02 2.45916277e-01
2.55140424e-01 9.18797869e-03 -8.87578309e-01 -4.80292261e-01
-1.02466360e-01 -4.48539585e-01 1.44634497e+00 6.05621219e-01
1.35169220e+00 -5.12803495e-01 -2.95939386e-01 7.45888233e-01
7.99901962e-01 2.41276175e-01 3.27479929e-01 1.85242280e-01
9.30792332e-01 7.27427959e-01 4.08969596e-02 4.90535557e-01
4.27982002e-01 9.95966718e-02 4.38608944e-01 -3.99933398e-01
5.12641668e-01 -1.94569871e-01 3.58099371e-01 1.27340782e+00
1.97328687e-01 -1.21380731e-01 -9.43984032e-01 2.32472360e-01
-2.21879268e+00 -6.83043480e-01 -8.44136700e-02 1.74678302e+00
7.73668885e-01 1.82848573e-01 2.08547756e-01 2.15573441e-02
8.02571833e-01 4.30060960e-02 -7.81446815e-01 -1.81321710e-01
-3.22587162e-01 -4.95618612e-01 -5.10133170e-02 2.28700861e-01
-1.55935049e+00 9.17847633e-01 7.36175060e+00 5.97725093e-01
-7.49252975e-01 -7.17251049e-03 7.82487988e-01 -1.07227780e-01
-3.33036929e-01 -3.37428272e-01 -1.01501763e+00 2.55536795e-01
9.63614225e-01 -1.29250988e-01 1.58840790e-01 6.52514338e-01
4.28767979e-01 3.94061685e-01 -1.12106621e+00 9.74615812e-01
1.25122815e-01 -1.33894837e+00 6.24382019e-01 4.13815826e-01
1.22207487e+00 7.24065453e-02 2.98917741e-01 2.85523891e-01
9.00099576e-01 -1.27058804e+00 -3.29380706e-02 3.15401226e-01
2.64427990e-01 -7.18093693e-01 3.59072477e-01 5.70205152e-01
-7.68995941e-01 -4.14027184e-01 -6.80476606e-01 -8.78254101e-02
-1.62789226e-01 9.48103905e-01 -1.14260459e+00 2.54518211e-01
7.47851014e-01 9.37255442e-01 -6.43939495e-01 8.57475758e-01
-3.46319079e-01 8.85755539e-01 6.03185780e-02 2.10258171e-01
2.52635539e-01 -1.47491256e-02 4.43979353e-01 1.10289764e+00
-1.32492825e-01 -2.50278860e-01 5.21990538e-01 4.43140030e-01
-3.63206208e-01 3.35312366e-01 -4.56310511e-01 -3.38455707e-01
-2.87523009e-02 1.44780457e+00 -1.11573899e+00 -3.80214393e-01
-1.36329263e-01 7.37986267e-01 4.74193186e-01 2.29108199e-01
-5.24647653e-01 1.39520481e-01 5.05485475e-01 5.37743755e-02
-1.96538240e-01 -2.60742128e-01 -1.85459614e-01 -1.51252949e+00
-5.89862049e-01 -1.11244965e+00 1.01638842e+00 -5.66172123e-01
-1.40831947e+00 6.66620195e-01 -2.46265009e-02 -1.00924456e+00
-5.62275410e-01 -5.44937730e-01 -1.84147641e-01 2.13530660e-01
-6.52795076e-01 -9.43471551e-01 -9.88366455e-02 6.19005084e-01
8.20997119e-01 -3.36044908e-01 9.21773553e-01 1.43998712e-01
-8.85029674e-01 2.37388089e-01 4.44362134e-01 2.98936814e-01
5.15360773e-01 -1.52823186e+00 3.94007474e-01 5.65761268e-01
3.73492748e-01 1.21699917e+00 4.99454409e-01 -8.51590216e-01
-1.13796532e+00 -1.42253101e+00 6.50475681e-01 -4.35334563e-01
8.51418853e-01 -8.07227373e-01 -8.84261549e-01 6.50666416e-01
3.68469983e-01 -6.60171747e-01 1.23726594e+00 8.98640811e-01
-3.50350142e-01 3.04380268e-01 -5.19303918e-01 5.18246889e-01
9.50200558e-01 -6.26479268e-01 -7.43539095e-01 8.96260202e-01
1.08205593e+00 2.28011906e-02 -1.21648920e+00 7.18220353e-01
6.63261652e-01 -7.57428229e-01 6.74540818e-01 -1.38601494e+00
8.42874289e-01 -2.09856763e-01 -2.18017891e-01 -1.39464843e+00
-9.12216187e-01 -6.51433349e-01 -3.61878008e-01 8.07270050e-01
6.22384250e-01 -2.83117384e-01 1.19051027e+00 9.89238247e-02
4.31479746e-03 -8.87173057e-01 -5.68693995e-01 -4.34993863e-01
7.60657117e-02 -5.29482424e-01 2.08228752e-01 1.36358953e+00
7.86532536e-02 1.22382474e+00 -6.55518472e-01 1.29149705e-01
7.08172858e-01 6.93032503e-01 6.31721675e-01 -1.75024533e+00
-5.02270520e-01 -2.43408173e-01 -3.13403636e-01 -1.01206779e+00
5.92576861e-01 -1.13216650e+00 -2.63406992e-01 -1.72930157e+00
8.34731877e-01 2.03016326e-01 -5.54680407e-01 6.69649541e-01
-2.64546216e-01 1.35046199e-01 -6.89526424e-02 4.81507093e-01
-1.05225623e+00 4.05548424e-01 1.08595860e+00 -4.24706966e-01
-3.89630437e-01 1.29597280e-02 -9.21296716e-01 1.07893252e+00
6.27182424e-01 -4.30745929e-01 -6.68899834e-01 -3.00978512e-01
8.10997605e-01 2.07689852e-01 -2.09738731e-01 -6.51620746e-01
2.86916077e-01 -8.96989107e-02 8.14285755e-01 -1.17449069e+00
1.93992674e-01 -4.24584448e-01 7.99561664e-02 4.74850357e-01
-9.28082466e-01 3.51042449e-01 -5.54569007e-04 4.59737271e-01
-3.51288736e-01 1.38517290e-01 3.16512465e-01 -2.13114083e-01
-4.89689261e-01 2.81067640e-01 -6.13866985e-01 -9.86121595e-02
3.86237949e-01 1.18152961e-01 -3.73434544e-01 -4.78251398e-01
-1.26923513e+00 5.30520916e-01 -4.55150288e-03 6.85328841e-01
7.79593706e-01 -1.23402977e+00 -7.93420494e-01 4.12668288e-01
-1.42761573e-01 -5.23244850e-02 2.89714426e-01 8.93794298e-01
2.02152282e-01 7.30154037e-01 1.81504078e-02 -8.80737603e-01
-1.48868394e+00 7.24190116e-01 -9.00406763e-02 -7.12917924e-01
-1.93239987e-01 7.07250714e-01 1.13488412e+00 -6.63449168e-01
4.55058545e-01 -4.77447927e-01 -4.40205634e-01 5.76402962e-01
5.57407439e-01 3.53744447e-01 -8.79529193e-02 -4.00862008e-01
5.22301197e-02 3.35336998e-02 -5.30109882e-01 1.25390396e-01
1.24105418e+00 -7.20108151e-02 -6.55165017e-01 7.45191753e-01
1.29733324e+00 -3.24546874e-01 -6.17917299e-01 -1.77188531e-01
3.48989815e-01 1.37659192e-01 2.80965567e-02 -5.71232200e-01
-7.81026185e-01 7.57005394e-01 4.42561299e-01 7.10912883e-01
1.10108685e+00 1.53773367e-01 3.52315515e-01 8.75125587e-01
-1.35147655e-02 -8.74493599e-01 1.79435372e-01 8.96117747e-01
7.97856748e-01 -1.14695776e+00 1.16236433e-01 -2.26714253e-01
-2.25127280e-01 1.05928850e+00 5.78945935e-01 2.18360513e-01
8.18430543e-01 2.99139291e-01 -1.06428057e-01 -5.01624823e-01
-1.39160931e+00 -5.78446873e-02 6.09406769e-01 5.78209106e-03
5.90899646e-01 3.26690614e-01 -1.79702625e-01 7.48657882e-01
-6.64851487e-01 -2.40479618e-01 2.46708050e-01 3.66107821e-01
-5.98955691e-01 -6.31447971e-01 -5.50003946e-01 9.94046152e-01
-7.60236204e-01 -4.37787026e-01 -9.31316316e-01 4.42163497e-01
2.29873076e-01 6.43503964e-01 1.43150061e-01 -8.01202357e-01
-2.58380651e-01 -9.26004946e-02 1.55280475e-02 -7.58531868e-01
-4.19711262e-01 5.25316477e-01 -8.72067437e-02 -2.97092825e-01
-4.63602364e-01 -6.57816589e-01 -1.20784283e+00 1.27527192e-01
-1.20970391e-01 6.60771906e-01 5.17470062e-01 7.67430782e-01
4.22848254e-01 6.57276630e-01 6.98497355e-01 -5.07946074e-01
-1.66576058e-01 -1.24276924e+00 -6.00386024e-01 4.62089956e-01
1.72480702e-01 -4.26873416e-01 -2.99073070e-01 1.14017382e-01] | [9.540022850036621, 4.397922515869141] |
20f7ad7b-4063-4f8f-a9b6-71e6369e682f | 3d-object-recognition-with-deep-belief-nets | null | null | http://papers.nips.cc/paper/3872-3d-object-recognition-with-deep-belief-nets | http://papers.nips.cc/paper/3872-3d-object-recognition-with-deep-belief-nets.pdf | 3D Object Recognition with Deep Belief Nets | We introduce a new type of Deep Belief Net and evaluate it on a 3D object recognition task. The top-level model is a third-order Boltzmann machine, trained using a hybrid algorithm that combines both generative and discriminative gradients. Performance is evaluated on the NORB database(normalized-uniform version), which contains stereo-pair images of objects under different lighting conditions and viewpoints. Our model achieves 6.5% error on the test set, which is close to the best published result for NORB (5.9%) using a convolutional neural net that has built-in knowledge of translation invariance. It substantially outperforms shallow models such as SVMs (11.6%). DBNs are especially suited for semi-supervised learning, and to demonstrate this we consider a modified version of the NORB recognition task in which additional unlabeled images are created by applying small translations to the images in the database. With the extra unlabeled data (and the same amount of labeled data as before), our model achieves 5.2% error, making it the current best result for NORB. | ['Vinod Nair', 'Geoffrey E. Hinton'] | 2009-12-01 | null | null | null | neurips-2009-12 | ['3d-object-recognition'] | ['computer-vision'] | [ 1.66226879e-01 8.09604153e-02 -1.86771125e-01 -6.79693639e-01
-7.68577456e-01 -5.63898921e-01 9.61028636e-01 -4.07081068e-01
-8.40563238e-01 6.98634326e-01 3.69638801e-02 -2.90019333e-01
2.54578292e-01 -5.90422213e-01 -1.02442813e+00 -1.00099075e+00
-4.02514376e-02 1.01894879e+00 3.80991608e-01 -3.39840241e-02
1.51771471e-01 9.23835397e-01 -1.36249948e+00 3.13877016e-01
1.15441933e-01 1.31836414e+00 1.23174503e-01 6.70175374e-01
1.49725556e-01 6.06336892e-01 -4.29729968e-01 -2.67081678e-01
4.65295523e-01 -8.50717723e-02 -1.04700899e+00 -3.99469882e-02
1.11401534e+00 -7.15357542e-01 -4.49079812e-01 7.53415227e-01
5.90571165e-01 2.22905234e-01 9.96122479e-01 -1.07309496e+00
-8.77981663e-01 1.89773768e-01 -2.41234973e-01 4.37662214e-01
-7.28252679e-02 8.68707076e-02 8.21853578e-01 -1.01338041e+00
5.58832228e-01 1.26399350e+00 7.38687336e-01 9.59687531e-01
-1.59850168e+00 -2.83786952e-01 -2.73924410e-01 3.01250726e-01
-1.08187699e+00 -5.21131933e-01 3.73666286e-01 -5.36714256e-01
1.69999552e+00 1.87078670e-01 6.82368577e-01 1.51950896e+00
1.65426701e-01 8.66314292e-01 1.62919831e+00 -5.31866789e-01
4.07388628e-01 2.48627454e-01 1.74182624e-01 6.32721364e-01
6.80746213e-02 3.78174603e-01 -5.61616898e-01 -1.61096230e-01
7.34777331e-01 -6.44715950e-02 -1.58852980e-01 -8.85286927e-01
-1.26226866e+00 7.29319811e-01 7.81771600e-01 2.21407220e-01
-1.65584683e-01 2.00077355e-01 -6.19945824e-02 2.55583264e-02
3.30134213e-01 2.24147677e-01 -6.05879128e-01 -4.26938124e-02
-1.02558446e+00 9.60295200e-02 1.00729132e+00 8.34397256e-01
8.31124663e-01 2.02440128e-01 -2.64360756e-02 1.10490835e+00
5.14906645e-01 6.87047184e-01 7.51252472e-01 -6.99201763e-01
1.26132414e-01 2.40493670e-01 9.95459855e-02 -6.33711159e-01
-3.68337870e-01 -3.69251579e-01 -8.15666318e-01 5.41066587e-01
4.09772903e-01 3.25696439e-01 -1.59163368e+00 1.70067501e+00
8.66527483e-02 7.18568340e-02 1.54190928e-01 9.17407870e-01
1.00320137e+00 7.58016467e-01 -3.60047847e-01 3.16588670e-01
1.02077699e+00 -1.09665847e+00 -1.68919876e-01 -3.44026953e-01
1.22234307e-01 -5.68315268e-01 6.36789680e-01 3.32156390e-01
-8.60959768e-01 -6.38708055e-01 -1.39189613e+00 -2.06343487e-01
-7.66145110e-01 -1.50949597e-01 4.84408945e-01 8.92000079e-01
-1.58512163e+00 6.98750198e-01 -1.04326165e+00 -6.10272765e-01
5.08317411e-01 4.32936102e-01 -5.36760449e-01 -1.98606357e-01
-9.67631221e-01 1.49222875e+00 4.25448567e-01 1.70876041e-01
-1.47814083e+00 -3.32294345e-01 -9.35374379e-01 -1.37965158e-01
-2.13237286e-01 -4.78325069e-01 1.37934744e+00 -9.21864808e-01
-1.70302379e+00 1.46240318e+00 -8.98859054e-02 -4.92959291e-01
4.44709122e-01 4.44751047e-03 -1.08472720e-01 -1.38695434e-01
-1.40878230e-01 1.18019891e+00 9.26737726e-01 -1.36117196e+00
-2.71048069e-01 -6.22640669e-01 -7.96636194e-03 5.95769323e-02
-1.92611381e-01 -2.67652005e-01 -3.65383327e-01 -3.69430900e-01
2.73753613e-01 -1.10847151e+00 -1.39381483e-01 -2.22035691e-01
-2.11426198e-01 -1.94382727e-01 6.87635839e-01 -6.96242213e-01
2.18643099e-01 -2.04793310e+00 1.89107850e-01 9.18688476e-02
-2.08848249e-02 1.59645468e-01 -2.98007816e-01 1.08880624e-01
-2.52690077e-01 -8.05675238e-03 -1.80877268e-01 -4.75765347e-01
2.06360906e-01 5.26183248e-01 -2.48592958e-01 6.84734762e-01
2.40741745e-01 9.69773650e-01 -6.25759721e-01 -2.22814396e-01
2.72345364e-01 4.03253853e-01 -4.57796633e-01 1.96610048e-01
-4.54866327e-02 1.58927113e-01 2.61685342e-01 7.44236767e-01
7.41735578e-01 -1.53796032e-01 -5.60025498e-02 -2.92228580e-01
1.73767030e-01 2.82276243e-01 -8.06427002e-01 1.72615230e+00
-3.30393583e-01 9.64180231e-01 -1.30125582e-01 -1.28986013e+00
1.01937187e+00 9.54751819e-02 1.33361235e-01 -6.91246629e-01
-4.46126685e-02 2.40415394e-01 6.49098456e-02 -2.59192526e-01
2.44484767e-02 -4.08128232e-01 7.24913627e-02 2.62594968e-01
9.09056485e-01 -5.29340148e-01 -5.99342920e-02 -8.16964656e-02
9.80384707e-01 2.96979517e-01 1.56790957e-01 -3.93089116e-01
1.27233118e-01 -1.43884107e-01 3.88402045e-01 1.12537026e+00
-5.01290083e-01 7.40919292e-01 2.36848563e-01 -9.35803711e-01
-1.18470633e+00 -1.31082201e+00 -4.83105898e-01 9.10686433e-01
4.42911610e-02 1.42068252e-01 -5.53067029e-01 -5.70768833e-01
1.40750453e-01 6.17296278e-01 -8.94320130e-01 -1.77477941e-01
-2.33060107e-01 -1.11934125e+00 3.93062830e-01 7.61121750e-01
6.09995902e-01 -1.14158154e+00 -3.75251979e-01 9.80477333e-02
1.98055468e-02 -9.61741626e-01 2.25959837e-01 7.92257309e-01
-9.28721249e-01 -8.14313889e-01 -8.67848694e-01 -9.76424456e-01
4.54370916e-01 -2.25698352e-01 1.26633608e+00 -5.49089909e-01
-3.60425264e-01 5.38174152e-01 -1.15478754e-01 -4.93201613e-01
-4.16905046e-01 1.16327494e-01 1.75944850e-01 -1.80560574e-01
5.29714823e-01 -5.81932724e-01 -4.95462865e-01 4.55350637e-01
-6.61186874e-01 6.42472506e-02 6.39071882e-01 1.41363990e+00
4.74604994e-01 -3.27426046e-01 -5.11595327e-03 -1.34627149e-01
-1.62900630e-02 -9.73173678e-02 -7.64443755e-01 9.29996297e-02
-5.17221451e-01 5.98431518e-03 1.99664727e-01 -3.99100184e-01
-9.45064366e-01 7.27107599e-02 -3.76070708e-01 -3.92203271e-01
-5.87451041e-01 1.19444549e-01 -7.72905722e-02 -3.23546052e-01
8.20561111e-01 5.25829434e-01 -9.79354754e-02 -6.99679673e-01
1.94391683e-01 6.94309592e-01 4.16625172e-01 -5.55032611e-01
5.29062808e-01 5.77431440e-01 -2.98867151e-02 -7.17309356e-01
-9.74212110e-01 -3.23269635e-01 -9.49884713e-01 -2.24454895e-01
8.54894757e-01 -5.69899738e-01 -5.64754725e-01 8.81591558e-01
-1.02081525e+00 -7.85424709e-01 -1.61317945e-01 5.92310965e-01
-9.83623862e-01 -3.24997343e-02 -8.18892241e-01 -5.48295557e-01
6.36894628e-03 -1.22387850e+00 1.08521461e+00 3.57539989e-02
1.93159178e-01 -8.93592596e-01 4.15132225e-01 4.94009346e-01
4.37374592e-01 4.62088212e-02 9.42371249e-01 -9.35862899e-01
-4.58940893e-01 -2.06267372e-01 -2.16489956e-01 9.75330651e-01
-1.69114899e-02 -2.71925658e-01 -1.44053197e+00 -5.20246387e-01
3.55328619e-02 -9.34075952e-01 1.19588137e+00 3.72977465e-01
1.06472564e+00 4.29777354e-02 -2.40136027e-01 5.66385567e-01
1.32258749e+00 3.16103607e-01 6.02792144e-01 5.46251833e-01
5.59003353e-01 2.12118655e-01 -1.74838789e-02 -9.38856043e-03
1.19889855e-01 7.36367047e-01 6.65375710e-01 -2.27481335e-01
-1.85145855e-01 3.64908315e-02 2.92924762e-01 6.18680477e-01
-2.29900584e-01 -4.06127051e-02 -1.28182602e+00 4.89437282e-01
-1.58947837e+00 -9.57987905e-01 3.90458226e-01 2.02030373e+00
8.62199843e-01 3.73738050e-01 -6.48764521e-02 -3.91482338e-02
4.63864565e-01 2.82413870e-01 -6.23521030e-01 -5.53767323e-01
-1.13748193e-01 5.43312371e-01 5.00547528e-01 4.43455875e-01
-1.44666576e+00 9.04050946e-01 7.78619194e+00 7.89476633e-01
-1.33138561e+00 2.29225144e-01 8.59639049e-01 -1.99136227e-01
1.25840098e-01 -3.43917519e-01 -9.12400663e-01 2.82015085e-01
1.06765103e+00 4.41539228e-01 6.73234344e-01 1.09659708e+00
-3.73935997e-01 -8.67601410e-02 -1.41497493e+00 1.05255711e+00
5.84119737e-01 -1.31260097e+00 -7.80141875e-02 3.64538692e-02
8.84594083e-01 6.97187006e-01 2.44059831e-01 7.30330110e-01
4.13108706e-01 -1.16491318e+00 7.90952802e-01 3.99860203e-01
5.86312473e-01 -5.06959975e-01 8.91605258e-01 4.05640721e-01
-3.73460978e-01 2.13310838e-01 -6.62617445e-01 -1.90462135e-02
-2.25477800e-01 3.77278775e-01 -1.06162131e+00 1.06171835e-02
1.11590719e+00 5.26948333e-01 -6.67536438e-01 9.36920881e-01
-9.26152766e-02 7.06958771e-01 -6.17418587e-01 -2.36208424e-01
5.16855657e-01 -1.62818078e-02 2.63293326e-01 1.34421706e+00
1.25601694e-01 -1.83262244e-01 -4.57546581e-03 7.27258623e-01
-2.19321340e-01 -1.83257028e-01 -6.45584583e-01 1.93060398e-01
-4.36220085e-03 1.16010654e+00 -6.23093426e-01 -4.98284817e-01
-3.73163611e-01 1.21300459e+00 7.87368834e-01 4.61880863e-01
-7.63079286e-01 -9.69234947e-03 4.37213182e-01 -4.72890973e-01
7.03842223e-01 -1.00570850e-01 -3.47620510e-02 -1.12795246e+00
-2.02900231e-01 -7.22358048e-01 8.68227854e-02 -1.20935881e+00
-1.41853917e+00 6.36311769e-01 7.58089200e-02 -7.78199911e-01
-1.92057818e-01 -1.50182831e+00 -7.37232193e-02 1.00943518e+00
-1.23877871e+00 -1.23556757e+00 -2.69167721e-01 5.27713418e-01
2.42032930e-01 -2.01546371e-01 1.31840396e+00 6.16928004e-02
-3.70581627e-01 3.84810865e-01 4.93848294e-01 3.02572638e-01
7.46068537e-01 -1.60558856e+00 4.60458905e-01 5.20020425e-01
5.77356577e-01 5.44216752e-01 4.35704142e-01 -9.25165266e-02
-1.29629922e+00 -7.87392259e-01 7.89405525e-01 -7.75404811e-01
4.56763476e-01 -6.82861030e-01 -5.75531542e-01 1.06713963e+00
3.10252577e-01 3.06505919e-01 8.86107743e-01 3.35222840e-01
-6.59206808e-01 -2.77724355e-01 -1.23477995e+00 4.49672848e-01
8.68788779e-01 -5.75668454e-01 -1.15760708e+00 3.30225348e-01
2.68609285e-01 -6.50970876e-01 -7.31647134e-01 7.23870277e-01
7.78303564e-01 -1.21028543e+00 9.74851310e-01 -8.17237914e-01
2.43577123e-01 -1.43282888e-02 -5.77928603e-01 -1.43275094e+00
-5.98911703e-01 5.75576909e-02 -1.31255984e-01 7.69288599e-01
4.93004590e-01 -6.65463805e-01 8.78477871e-01 4.96821433e-01
9.13027078e-02 -8.60324264e-01 -1.27432549e+00 -1.11863148e+00
1.37877688e-01 -4.97675627e-01 1.49584591e-01 7.96889186e-01
-2.88210720e-01 2.80927688e-01 -3.21554720e-01 -1.13324381e-01
5.65138280e-01 3.55986357e-01 6.62554324e-01 -1.11510420e+00
-3.13729793e-01 -4.58498806e-01 -8.63541007e-01 -1.24372959e+00
3.12533647e-01 -1.02674389e+00 2.81211197e-01 -1.39426064e+00
6.22543275e-01 -2.55113959e-01 -7.04019487e-01 8.86715293e-01
3.72419536e-01 8.46790612e-01 -9.81190130e-02 1.46596104e-01
-2.72875875e-01 5.50582230e-01 9.25335169e-01 -5.99811137e-01
3.59321177e-01 -1.30259797e-01 -1.11450471e-01 8.47686827e-01
7.81179070e-01 -2.59040117e-01 -9.39190239e-02 -3.53371650e-01
-3.26145828e-01 -4.06499177e-01 6.58268332e-01 -9.94923651e-01
-9.51228216e-02 -1.86610725e-02 1.01022363e+00 -7.86658585e-01
7.06090271e-01 -7.58756161e-01 -1.42493188e-01 4.78654653e-01
-2.35647470e-01 -2.65126348e-01 4.51047152e-01 4.01807576e-01
-1.71282426e-01 -3.45307082e-01 1.05571306e+00 -2.63779074e-01
-9.11656260e-01 3.33881348e-01 -5.77838242e-01 -2.59996414e-01
7.63046265e-01 -3.35286915e-01 -1.22253917e-01 -3.18614721e-01
-1.08025467e+00 -4.25973147e-01 4.64441508e-01 3.01015884e-01
5.11555016e-01 -1.54153597e+00 -4.63828266e-01 3.96102071e-01
2.81755358e-01 -1.06468454e-01 -4.35299277e-02 5.68589568e-01
-5.62536538e-01 6.08073056e-01 -5.31596005e-01 -9.39929128e-01
-9.72429335e-01 5.57530880e-01 6.70477211e-01 -1.42219961e-01
-2.69129008e-01 1.21361494e+00 4.14410494e-02 -6.66001141e-01
2.92222470e-01 -4.17549789e-01 1.28782868e-01 -1.68174937e-01
2.77087837e-01 8.08641315e-03 4.34925735e-01 -5.12504697e-01
-6.57806098e-01 1.40500337e-01 -8.75508189e-02 -4.27889109e-01
1.40988398e+00 3.27707291e-01 -2.02461079e-01 6.83617592e-01
1.48914945e+00 -4.78507936e-01 -1.21343470e+00 -1.35184348e-01
-2.20437199e-01 -2.98389196e-01 2.57188797e-01 -1.07715976e+00
-7.37348139e-01 1.00466168e+00 9.16013420e-01 -2.25252062e-02
8.39883626e-01 1.08096413e-01 2.34246209e-01 8.77571285e-01
6.12101257e-01 -8.45263064e-01 1.87250674e-01 7.66065896e-01
1.01365066e+00 -1.45695484e+00 -1.45243064e-01 3.32561880e-01
-5.80053590e-02 1.00905764e+00 7.79985905e-01 -2.18408212e-01
7.34968126e-01 3.17095309e-01 2.43893385e-01 4.07717228e-02
-5.58123350e-01 -1.81447472e-02 4.77839470e-01 7.96299040e-01
2.12739035e-01 -5.33019453e-02 2.80317396e-01 2.08273798e-01
-1.73147067e-01 -2.40043804e-01 -2.75236499e-02 9.65050042e-01
-3.07309210e-01 -9.02499199e-01 -3.41834664e-01 3.17772150e-01
-1.76109180e-01 -2.46254697e-01 -5.44339716e-01 8.03711832e-01
5.00446483e-02 4.93777275e-01 2.19268307e-01 -2.23237649e-01
9.41720232e-02 3.62206370e-01 1.11124456e+00 -4.84843284e-01
-8.29767436e-02 -4.44032513e-02 -9.93598904e-03 -7.89400220e-01
-6.48754358e-01 -4.84251320e-01 -5.29854834e-01 -1.48370519e-01
-1.74893007e-01 -2.13952795e-01 8.91045749e-01 7.90806472e-01
2.93615414e-03 1.25762045e-01 3.93343985e-01 -1.57665706e+00
-7.47115791e-01 -1.24323308e+00 -5.82515478e-01 5.12181342e-01
4.17159438e-01 -9.14439678e-01 -4.89098549e-01 1.51866004e-01] | [9.457414627075195, 2.3341968059539795] |
cfbe8349-c061-4479-b3c2-0ceea552989a | robust-filtering-and-smoothing-with-gaussian | 1203.4345 | null | https://arxiv.org/abs/1203.4345v1 | https://arxiv.org/pdf/1203.4345v1.pdf | Robust Filtering and Smoothing with Gaussian Processes | We propose a principled algorithm for robust Bayesian filtering and smoothing in nonlinear stochastic dynamic systems when both the transition function and the measurement function are described by non-parametric Gaussian process (GP) models. GPs are gaining increasing importance in signal processing, machine learning, robotics, and control for representing unknown system functions by posterior probability distributions. This modern way of "system identification" is more robust than finding point estimates of a parametric function representation. In this article, we present a principled algorithm for robust analytic smoothing in GP dynamic systems, which are increasingly used in robotics and control. Our numerical evaluations demonstrate the robustness of the proposed approach in situations where other state-of-the-art Gaussian filters and smoothers can fail. | ['Carl Edward Rasmussen', 'Uwe D. Hanebeck', 'Marco F. Huber', 'Ryan Turner', 'Marc Peter Deisenroth'] | 2012-03-20 | null | null | null | null | ['gaussian-processes'] | ['methodology'] | [-1.47617266e-01 1.05587365e-04 3.65214229e-01 1.89734235e-01
-4.15887177e-01 -3.33460748e-01 6.89498484e-01 3.47710326e-02
-1.99827150e-01 8.37747514e-01 -1.43539712e-01 -2.30344236e-01
-4.31471795e-01 -4.47768420e-01 -6.80719137e-01 -1.06945980e+00
-2.74203479e-01 4.14610088e-01 3.66854012e-01 -4.46499214e-02
1.14703596e-01 6.76943779e-01 -1.16693151e+00 -7.41944373e-01
7.53591478e-01 7.17515111e-01 2.25778326e-01 8.82978797e-01
3.38723302e-01 -2.11340804e-02 -4.87536192e-01 1.40790969e-01
2.82462835e-01 5.54660931e-02 3.85019695e-03 -1.28882506e-03
-7.29237050e-02 2.37586021e-01 -5.87238371e-01 1.44962513e+00
4.38468158e-01 7.30856538e-01 1.06802416e+00 -8.71986985e-01
-4.19971168e-01 9.72583890e-02 -3.46680731e-01 -5.50472736e-02
2.35482082e-01 2.92024553e-01 3.90016176e-02 -6.20323718e-01
1.75099626e-01 1.76808488e+00 9.34887409e-01 2.75427073e-01
-1.46466804e+00 -3.58103842e-01 -5.87869324e-02 -2.10934982e-01
-1.49636054e+00 -2.92990357e-01 4.83746707e-01 -8.96960437e-01
2.79741615e-01 -6.85517266e-02 3.21854740e-01 7.97666967e-01
1.00794446e+00 4.47661728e-01 9.77190733e-01 -1.28331184e-01
4.19238538e-01 -5.30783571e-02 2.54460841e-01 7.22454607e-01
3.71426672e-01 6.31135345e-01 -1.19218349e-01 -5.68531752e-01
1.15839326e+00 -2.46485472e-02 -3.12364787e-01 -4.78711724e-01
-1.05233181e+00 8.36864591e-01 3.80429104e-02 8.85712132e-02
-7.86787450e-01 5.29901147e-01 1.98123250e-02 5.77807724e-02
6.15680754e-01 2.13144049e-01 -3.37553620e-01 -2.50328898e-01
-5.98154962e-01 4.45279658e-01 1.11447549e+00 1.12797153e+00
3.77486944e-01 5.21875679e-01 -1.68025017e-01 4.73988384e-01
9.00262058e-01 1.23157454e+00 2.50518382e-01 -9.61597085e-01
-1.54809445e-01 -1.85239196e-01 7.51704633e-01 -9.22628880e-01
-5.40023029e-01 -2.83830851e-01 -8.60253513e-01 4.49945390e-01
6.38330221e-01 -3.84623170e-01 -1.06184840e+00 1.32392728e+00
4.25772458e-01 6.48181915e-01 1.47668228e-01 6.07240260e-01
2.26904899e-02 8.85200560e-01 -6.95673600e-02 -3.49764913e-01
1.19085991e+00 -9.05491635e-02 -1.07010603e+00 -6.46359175e-02
-1.51096895e-01 -9.03869629e-01 5.61586022e-01 6.48432434e-01
-8.89177740e-01 -6.22547805e-01 -8.61368716e-01 4.36835408e-01
-1.18004374e-01 7.00341463e-02 1.84892267e-01 6.83299065e-01
-8.48607600e-01 9.71728802e-01 -1.33900547e+00 -2.25197807e-01
-8.72600824e-02 2.05092430e-01 -3.52731650e-03 2.95185089e-01
-9.62098837e-01 1.24108481e+00 2.63910085e-01 5.00501752e-01
-1.01373100e+00 -6.52799487e-01 -6.30633891e-01 4.46388014e-02
2.64445126e-01 -4.60440159e-01 1.27661991e+00 7.68045336e-02
-2.13720894e+00 -5.89992777e-02 -2.96931863e-01 -6.83721066e-01
3.78106117e-01 -6.07216537e-01 -4.99799579e-01 -1.09951511e-01
-4.11899269e-01 -1.63258359e-01 1.69965887e+00 -9.03202057e-01
-3.70980918e-01 -2.34290257e-01 -6.72648311e-01 -1.92722440e-01
6.55547976e-01 -1.23975702e-01 1.38422042e-01 -4.79583621e-01
5.45214176e-01 -1.07086897e+00 -6.34905636e-01 3.11571062e-02
-3.36405993e-01 -1.12590708e-01 8.81894529e-01 -8.68067682e-01
8.26361299e-01 -2.23656845e+00 2.26442501e-01 6.31363332e-01
-1.92170694e-01 3.19421232e-01 5.05706429e-01 4.89156306e-01
2.72212178e-01 -2.80192524e-01 -2.90033340e-01 -2.41002575e-01
2.93090969e-01 1.08421713e-01 -6.17594123e-01 1.04437578e+00
1.34912148e-01 3.21515828e-01 -8.83013487e-01 -1.68953501e-02
5.75096726e-01 7.63983071e-01 4.19206284e-02 -1.24186523e-01
5.53997699e-03 6.31037474e-01 -5.45636296e-01 3.71693403e-01
4.61287826e-01 4.01316494e-01 -2.97462732e-01 -2.08869204e-01
-2.26358607e-01 -2.81045139e-01 -1.78658640e+00 1.05086732e+00
-3.64924163e-01 7.34151721e-01 5.22836447e-01 -8.54682565e-01
1.26418245e+00 4.79852408e-01 1.45651147e-01 3.34533572e-01
4.53554988e-01 2.14321345e-01 -2.81166267e-02 -2.55381256e-01
4.31746960e-01 -2.48545840e-01 9.38116759e-02 -2.05554754e-01
1.53769627e-01 -9.32290673e-01 4.32115458e-02 -2.27784574e-01
8.23495984e-01 2.81264007e-01 5.01183033e-01 -7.46152043e-01
6.58405662e-01 -3.86733890e-01 5.30374289e-01 7.95552135e-01
-2.98666328e-01 1.80925429e-01 1.76567644e-01 3.29171345e-02
-8.67829919e-01 -1.20180333e+00 -4.72071141e-01 2.83667177e-01
-4.34992313e-02 1.43617451e-01 -4.75019008e-01 2.33019128e-01
4.56557959e-01 7.38549471e-01 -3.75997514e-01 -2.84758359e-01
-3.74455005e-01 -6.51328027e-01 1.83131620e-01 3.18640977e-01
-1.02750435e-02 -3.69988918e-01 -1.92519292e-01 8.50539982e-01
5.68182588e-01 -9.09746289e-01 -2.54394323e-01 -2.01448295e-02
-1.12308097e+00 -9.77019429e-01 -6.91115916e-01 -3.58890563e-01
5.69249332e-01 -6.74133599e-02 3.12915355e-01 -6.34970188e-01
-1.50613904e-01 8.84362817e-01 1.27610043e-01 -8.57340157e-01
-6.92979515e-01 -6.61017776e-01 6.70304239e-01 2.37114221e-01
-8.60703588e-02 -5.21783948e-01 -1.97316855e-01 4.48897779e-01
-4.08287138e-01 -7.35277593e-01 2.49657840e-01 6.30654812e-01
6.28775179e-01 4.75267172e-01 3.76723349e-01 -4.61448520e-01
1.13615537e+00 -3.62464070e-01 -1.38418519e+00 -9.66546312e-02
-3.10391188e-01 1.42700538e-01 5.08582711e-01 -7.52448797e-01
-1.22267985e+00 3.48691523e-01 1.65991113e-01 -5.03673911e-01
-2.37844169e-01 6.26030922e-01 1.80765819e-02 -2.81071752e-01
8.65997255e-01 -6.78070239e-04 5.57570338e-01 -6.53378963e-01
5.65998435e-01 4.13788676e-01 9.68091786e-01 -7.25731313e-01
9.82484579e-01 6.43707216e-01 6.97415054e-01 -1.35707498e+00
-2.47935832e-01 -6.77339733e-01 -5.45387328e-01 -1.91896483e-01
5.94162226e-01 -8.02556753e-01 -9.82514918e-01 6.92435086e-01
-1.03161323e+00 -1.52942717e-01 -4.69029069e-01 9.68543112e-01
-1.00566423e+00 4.84426767e-01 -5.83777189e-01 -1.68954194e+00
1.96617097e-02 -9.67757046e-01 9.84880984e-01 5.58437049e-01
-2.97125317e-02 -1.27914703e+00 8.37503746e-02 -7.24980414e-01
4.80955780e-01 3.10111940e-01 2.26296887e-01 -3.28542948e-01
-3.10654402e-01 -7.45579302e-01 3.59654576e-01 4.17532712e-01
7.32143298e-02 3.85591120e-01 -7.34710813e-01 -2.92703211e-01
5.38129270e-01 7.74865806e-01 4.99051213e-01 1.08323026e+00
3.15585852e-01 -4.32574265e-02 -5.08993924e-01 3.13733220e-01
1.21555805e+00 4.53628063e-01 3.26291144e-01 -1.69450819e-01
4.91858900e-01 6.08560324e-01 7.13504791e-01 4.55969334e-01
-2.35473946e-01 4.75401670e-01 8.12943429e-02 4.94977266e-01
4.44324255e-01 -2.35198587e-02 5.82316339e-01 4.55451906e-01
5.38704311e-03 1.23667382e-01 -8.37932169e-01 4.72004533e-01
-2.04697633e+00 -9.07500148e-01 -5.60252070e-01 2.47723961e+00
4.37610537e-01 9.41772386e-02 -1.37123272e-01 6.44944282e-03
1.09386706e+00 -4.21673566e-01 -2.65851378e-01 -2.16842756e-01
1.84391528e-01 1.48927540e-01 8.62820983e-01 8.14803064e-01
-1.30151260e+00 5.90474546e-01 6.48067379e+00 8.32246721e-01
-8.69447827e-01 -5.60719594e-02 -2.88936526e-01 5.86384892e-01
5.11853755e-01 -2.31640730e-02 -1.08111930e+00 6.57461643e-01
1.20964003e+00 -4.87258911e-01 2.59895027e-01 8.52456689e-01
8.96364212e-01 -4.36390430e-01 -7.17579484e-01 8.54260027e-01
-5.70156872e-01 -7.98682153e-01 -6.09727085e-01 1.17051423e-01
7.40211546e-01 -1.52066231e-01 1.76662341e-01 1.65860355e-01
5.78843355e-01 -5.97488880e-01 6.22759998e-01 1.50073540e+00
-1.24129616e-01 -5.71444988e-01 8.20127189e-01 6.00544453e-01
-1.14328063e+00 -2.04859793e-01 -6.72767699e-01 -2.58368999e-02
8.91173363e-01 1.03420758e+00 -6.29025042e-01 4.74310607e-01
3.25460255e-01 5.92878878e-01 4.38657627e-02 1.77256787e+00
-4.22751367e-01 8.11776340e-01 -9.14800167e-01 -1.18444808e-01
1.85956061e-02 -6.48579240e-01 1.33365631e+00 9.79023397e-01
7.87576735e-01 8.05343091e-02 3.03109467e-01 8.21868420e-01
7.95130372e-01 -2.47377247e-01 -5.06569922e-01 -1.55361548e-01
4.68027353e-01 8.55795681e-01 -7.26417780e-01 -3.05560648e-01
-7.98694119e-02 1.77258059e-01 -6.16394579e-01 6.31693900e-01
-5.41958749e-01 -4.27468359e-01 7.38392591e-01 -7.74070472e-02
4.23488528e-01 -9.55164611e-01 -3.18044960e-03 -7.72161424e-01
-4.09419119e-01 -2.67168283e-01 -8.02589580e-03 -5.86294711e-01
-1.35028756e+00 1.99279543e-02 4.38054413e-01 -1.30690026e+00
-5.93710482e-01 -8.48444521e-01 -7.22586155e-01 1.17937839e+00
-9.34237659e-01 -7.87563086e-01 2.66032964e-01 4.59290802e-01
1.96388304e-01 -1.06388852e-01 6.12894714e-01 -1.86799258e-01
-2.76839465e-01 -3.05188745e-01 1.09055531e+00 -4.99915391e-01
5.40318370e-01 -1.40457249e+00 4.44087684e-01 1.12493300e+00
-1.89152598e-01 8.82576704e-01 1.67575812e+00 -9.23486352e-01
-1.62182450e+00 -8.00392687e-01 9.93012786e-02 -3.14189404e-01
1.16462505e+00 -8.44684839e-02 -1.18466794e+00 4.80016619e-01
-4.21581268e-01 -3.13681923e-02 -4.73959967e-02 -1.66898057e-01
4.37850296e-01 2.01192778e-02 -1.26045191e+00 4.67286736e-01
1.90848693e-01 -3.53632182e-01 -8.28774810e-01 2.40822002e-01
4.62434441e-01 -5.61223149e-01 -1.05318987e+00 3.15675199e-01
4.64503318e-01 -5.81711344e-02 9.89380956e-01 -2.11208031e-01
-7.81467617e-01 -8.71477664e-01 -2.90311649e-02 -1.87739682e+00
-2.19884485e-01 -1.41797864e+00 -5.83811939e-01 9.04059887e-01
1.48696965e-02 -1.13852692e+00 4.47817087e-01 5.42460322e-01
-3.70593280e-01 -1.82446286e-01 -1.17186153e+00 -1.20440209e+00
-9.50197801e-02 -5.61821938e-01 1.23380631e-01 1.59310967e-01
-4.62745093e-02 -8.49047527e-02 -2.85151869e-01 8.43420088e-01
1.08527052e+00 -4.80995536e-01 7.76724756e-01 -1.60454655e+00
-3.12962234e-01 -3.47996622e-01 -6.05466783e-01 -1.02126133e+00
8.91598538e-02 2.04446800e-02 8.32770824e-01 -1.36719549e+00
-8.79451036e-01 -5.06451018e-02 9.08325166e-02 -4.88752902e-01
-1.72103927e-01 -3.43078941e-01 -3.89441773e-02 3.54736410e-02
2.21812889e-01 6.39927566e-01 9.17382181e-01 1.21929325e-01
-4.89225924e-01 8.50573182e-01 1.18963398e-01 1.16226602e+00
6.80408955e-01 -4.30221081e-01 -3.72765750e-01 2.03937501e-01
-1.76669464e-01 1.75551876e-01 4.54782993e-01 -1.32716238e+00
5.56208551e-01 -6.21484295e-02 5.99468239e-02 -6.06873155e-01
6.18277848e-01 -1.04688585e+00 4.47409838e-01 5.57288766e-01
2.87079424e-01 -2.00998381e-01 3.30880642e-01 1.24690914e+00
-1.41304150e-01 -4.17415082e-01 9.69306767e-01 1.12378851e-01
-6.59099758e-01 1.61362365e-01 -9.57488179e-01 -5.38106680e-01
6.52695954e-01 -3.57053913e-02 -2.18304276e-01 -6.52642608e-01
-1.26129758e+00 8.21952000e-02 1.04390308e-01 -2.35337857e-02
4.46233749e-01 -8.01871061e-01 -5.05184829e-01 3.07588913e-02
-5.91536939e-01 -1.32965252e-01 2.20700070e-01 1.05177498e+00
-3.80316645e-01 4.12897408e-01 8.03463161e-02 -7.07690954e-01
-8.13939750e-01 4.79453146e-01 3.88685614e-01 1.34864703e-01
-5.71876168e-01 6.32037222e-01 -4.02375124e-02 -2.41695330e-01
1.54489920e-01 -8.43109369e-01 1.13887526e-02 -3.55963767e-01
4.79816824e-01 8.93368959e-01 -1.65811941e-01 -6.86171412e-01
2.22503170e-02 5.14296591e-01 4.25532579e-01 -3.66382211e-01
8.87863040e-01 -3.64144534e-01 7.18001351e-02 8.95862341e-01
4.58245963e-01 -1.40177295e-01 -1.70216846e+00 -1.87494069e-01
3.08382124e-01 -2.98148334e-01 2.19602332e-01 -2.54380822e-01
-2.61685133e-01 8.06730807e-01 7.20912516e-01 5.50326765e-01
4.70097095e-01 -4.21233505e-01 3.09869766e-01 5.64140081e-01
5.81913233e-01 -1.04854631e+00 -5.70737243e-01 8.85487199e-01
8.74988973e-01 -6.02364242e-01 1.48997977e-01 -6.07444346e-01
-2.40681648e-01 1.09827518e+00 -1.46263540e-01 -9.08151746e-01
1.25941753e+00 3.76054823e-01 -2.71139473e-01 1.83471903e-01
-3.92350376e-01 -2.67890006e-01 4.79519755e-01 1.06531417e+00
5.37826121e-02 2.46674210e-01 -3.40909064e-01 4.47928697e-01
-7.93562979e-02 -2.28895679e-01 6.92847431e-01 9.45279896e-01
-9.17627394e-01 -6.99385345e-01 -1.05861795e+00 3.04265559e-01
-4.78107840e-01 2.21905261e-01 5.17257571e-01 8.14977109e-01
-5.43666422e-01 9.00975823e-01 -8.22502524e-02 2.03339860e-01
7.68692970e-01 2.32352033e-01 4.13941413e-01 -4.44265634e-01
-6.34984225e-02 5.66402972e-01 -4.66913506e-02 -3.83300960e-01
-5.32943681e-02 -1.12528765e+00 -1.10905957e+00 -1.16540186e-01
-5.72904229e-01 2.79024810e-01 1.15157413e+00 9.76930380e-01
1.50460467e-01 4.74807680e-01 1.55742332e-01 -1.19071269e+00
-1.11378884e+00 -1.13906109e+00 -1.08050227e+00 -3.31123769e-01
3.45413566e-01 -1.20106256e+00 -5.90549231e-01 2.36108035e-01] | [6.439033508300781, 3.4814810752868652] |
823b8937-c09e-4574-8c47-4ddd63a63b4b | deep-neural-network-approximation-theory | 1901.02220 | null | https://arxiv.org/abs/1901.02220v4 | https://arxiv.org/pdf/1901.02220v4.pdf | Deep Neural Network Approximation Theory | This paper develops fundamental limits of deep neural network learning by characterizing what is possible if no constraints are imposed on the learning algorithm and on the amount of training data. Concretely, we consider Kolmogorov-optimal approximation through deep neural networks with the guiding theme being a relation between the complexity of the function (class) to be approximated and the complexity of the approximating network in terms of connectivity and memory requirements for storing the network topology and the associated quantized weights. The theory we develop establishes that deep networks are Kolmogorov-optimal approximants for markedly different function classes, such as unit balls in Besov spaces and modulation spaces. In addition, deep networks provide exponential approximation accuracy - i.e., the approximation error decays exponentially in the number of nonzero weights in the network - of the multiplication operation, polynomials, sinusoidal functions, and certain smooth functions. Moreover, this holds true even for one-dimensional oscillatory textures and the Weierstrass function - a fractal function, neither of which has previously known methods achieving exponential approximation accuracy. We also show that in the approximation of sufficiently smooth functions finite-width deep networks require strictly smaller connectivity than finite-depth wide networks. | ['Helmut Bölcskei', 'Dmytro Perekrestenko', 'Dennis Elbrächter', 'Philipp Grohs'] | 2019-01-08 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [-1.18175872e-01 4.82360780e-01 2.01992288e-01 6.10957555e-02
-1.30931646e-01 -4.79838252e-01 3.78594548e-01 7.29458481e-02
-5.04347980e-01 9.93024290e-01 -9.35370624e-02 -2.73666710e-01
-2.86724329e-01 -1.17723501e+00 -9.29607153e-01 -8.76990259e-01
-5.22904098e-01 4.76991385e-01 7.32174963e-02 -3.84789079e-01
9.70600247e-02 8.80830824e-01 -1.34729981e+00 -1.51408289e-03
4.92744356e-01 1.69755340e+00 -4.52050805e-01 1.14117670e+00
-1.10899739e-01 9.27618623e-01 -4.72618729e-01 -4.41622913e-01
3.60833168e-01 -4.62293327e-01 -9.66298997e-01 -3.99286389e-01
4.97642815e-01 -4.28594351e-01 -1.00143158e+00 1.21231306e+00
3.19349289e-01 4.02331740e-01 1.07013786e+00 -7.20678031e-01
-1.01847291e+00 7.17127919e-01 3.83365415e-02 5.71631074e-01
-3.83512616e-01 -5.19632474e-02 7.81366110e-01 -6.52073026e-01
1.33014500e-01 8.96798253e-01 1.44843769e+00 6.83856130e-01
-1.38356078e+00 -2.04575196e-01 -5.53204298e-01 1.34710893e-01
-1.33673835e+00 -5.96448720e-01 5.42433560e-01 -3.28729659e-01
8.35811973e-01 2.34579176e-01 9.75325942e-01 6.62490666e-01
3.91691923e-01 1.90459862e-01 6.64672196e-01 -5.71146250e-01
4.67722893e-01 -9.01854411e-03 3.27856988e-01 9.91727531e-01
4.63255435e-01 1.07225706e-03 1.81448638e-01 -8.85148346e-02
1.37134480e+00 -2.12267727e-01 -6.28241122e-01 1.12680577e-01
-7.23847389e-01 8.41125250e-01 5.20918190e-01 7.25795150e-01
-3.64894032e-01 6.14144027e-01 5.37715137e-01 6.53927565e-01
4.49444264e-01 5.83549738e-01 -4.03622448e-01 -6.52880669e-02
-8.86472523e-01 6.99182302e-02 1.07044339e+00 7.58215070e-01
5.51420331e-01 5.80866158e-01 2.36477137e-01 5.21088004e-01
-4.42129642e-01 1.67675585e-01 5.36687374e-01 -1.22728372e+00
-1.16562508e-01 -8.23098943e-02 2.24621117e-01 -9.94370639e-01
-5.32626271e-01 -6.33892894e-01 -1.37294960e+00 1.14079334e-01
9.82677519e-01 6.32382259e-02 -2.73596704e-01 1.83719039e+00
4.20829058e-02 4.63920571e-02 5.93615025e-02 6.13055587e-01
4.00606126e-01 7.23118663e-01 -5.67686260e-01 -2.83102155e-01
9.83835518e-01 -5.57840049e-01 -6.35808885e-01 4.37329113e-01
7.02473283e-01 -2.74087727e-01 1.18627405e+00 4.17702019e-01
-1.65447366e+00 -6.62533820e-01 -1.26850390e+00 -2.36830801e-01
-4.36125994e-01 -2.39871234e-01 3.75037581e-01 5.10574877e-01
-1.24749470e+00 1.44927824e+00 -6.42571926e-01 1.07136168e-01
5.95824957e-01 3.46032351e-01 -2.56948948e-01 2.57608622e-01
-1.26432419e+00 9.01979268e-01 5.58835089e-01 1.04715124e-01
-4.52121526e-01 -8.79250824e-01 -4.74146515e-01 5.59552491e-01
-6.13185823e-01 -6.00188673e-01 1.18990016e+00 -1.14986122e+00
-1.43853462e+00 5.46551645e-01 2.27356374e-01 -9.91781473e-01
4.78954673e-01 4.38180082e-02 -3.47212851e-01 4.77685630e-01
-5.97205222e-01 3.71426761e-01 7.51223922e-01 -5.90072036e-01
-2.73705304e-01 -2.41224706e-01 3.08170199e-01 -3.19297850e-01
-7.14650631e-01 -6.94190919e-01 5.75664520e-01 -5.93318880e-01
-6.39462397e-02 -4.29908514e-01 -4.55352478e-02 5.96668303e-01
1.73084170e-01 -4.70415026e-01 5.61188519e-01 -6.64830565e-01
1.07018137e+00 -2.40317130e+00 2.36355457e-02 1.22282945e-01
3.46866727e-01 1.02300785e-01 -1.54856872e-02 9.83938873e-02
-1.52613267e-01 1.56994779e-02 -9.17658433e-02 7.75063559e-02
2.16513097e-01 3.86045039e-01 -4.26522255e-01 1.06540740e+00
-1.87346395e-02 7.92164743e-01 -5.93608618e-01 -2.13349044e-01
-8.28390047e-02 7.40891457e-01 -6.00925088e-01 -2.28058964e-01
-2.03031197e-01 1.07380569e-01 -2.60736316e-01 -9.37049836e-02
6.10714555e-01 -3.76632690e-01 -4.60841358e-01 -1.48299381e-01
7.53461346e-02 2.46365219e-01 -8.24135721e-01 1.20359373e+00
-6.75135732e-01 1.21904588e+00 3.50188673e-01 -1.65313733e+00
7.88488865e-01 4.02629495e-01 2.97208428e-01 -3.32295805e-01
3.28093022e-01 6.27922535e-01 1.28409276e-02 -3.04074347e-01
2.29256257e-01 -3.93709123e-01 3.85807782e-01 1.84853241e-01
8.98503065e-02 -4.01528291e-02 2.01987982e-01 -2.96038806e-01
9.04308081e-01 -5.96409678e-01 7.73615688e-02 -8.87008607e-01
5.12305856e-01 -1.76074088e-01 -5.60617745e-02 8.75396729e-01
-8.96874815e-02 2.26858765e-01 5.74535310e-01 -9.79659379e-01
-1.64392436e+00 -9.21782076e-01 -7.36566305e-01 7.57553756e-01
-2.17702195e-01 8.40078965e-02 -1.06185102e+00 9.39696729e-02
-1.12778306e-01 6.26820743e-01 -8.79635453e-01 -3.34164113e-01
-7.38387287e-01 -3.49465460e-01 1.09373522e+00 5.73466122e-01
6.92140937e-01 -9.34642076e-01 -5.76448441e-01 2.03830481e-01
2.31102437e-01 -1.08771753e+00 -3.10880840e-01 3.61845016e-01
-1.06981969e+00 -1.03032243e+00 -7.92879820e-01 -1.08888805e+00
5.22446334e-01 -5.52530348e-01 1.09993458e+00 1.14063181e-01
-1.57911107e-01 3.87037396e-01 3.16805393e-01 -2.66945623e-02
-4.97902542e-01 -2.57454533e-02 3.18204463e-01 -1.59392819e-01
9.40375626e-02 -9.03590620e-01 -7.52537608e-01 -6.62564039e-02
-9.20056760e-01 -2.92629421e-01 9.77823809e-02 8.89119029e-01
4.78242755e-01 5.19905150e-01 5.37101209e-01 -9.87184271e-02
6.49939060e-01 -2.13601634e-01 -7.05825806e-01 -9.70363840e-02
-1.57052934e-01 2.33911008e-01 1.45203352e+00 -7.45983839e-01
-1.99640378e-01 -4.44207251e-01 -4.15936232e-01 -2.78556168e-01
1.34280592e-01 2.46917099e-01 4.55970407e-01 -5.80134094e-01
9.94660616e-01 5.09591043e-01 -1.40766412e-01 -2.72363275e-01
2.32129663e-01 6.01207137e-01 8.32098603e-01 -6.35682762e-01
3.11999410e-01 4.87338245e-01 4.27280903e-01 -1.29853666e+00
-8.90562296e-01 2.65499502e-01 -3.68322849e-01 -4.16327566e-02
6.56194210e-01 -3.82683694e-01 -1.05693269e+00 4.46005732e-01
-1.48527193e+00 -5.21613002e-01 -1.06808400e+00 4.25432533e-01
-9.81622279e-01 1.56040639e-01 -1.36637795e+00 -1.01264322e+00
-4.12433118e-01 -6.74040496e-01 1.44494727e-01 -6.27766103e-02
-1.07061908e-01 -1.34879005e+00 -1.55759007e-01 -4.78836387e-01
8.66758823e-01 3.98811251e-01 1.35381448e+00 -5.72355866e-01
-5.87498350e-03 -5.37924230e-01 -2.79551387e-01 8.23571622e-01
-2.27595434e-01 -2.27648184e-01 -8.37687254e-01 -1.65145755e-01
6.69807613e-01 -3.82192343e-01 8.09578776e-01 7.33612776e-01
1.60077119e+00 -9.92344975e-01 2.40082443e-01 8.45710576e-01
1.34689939e+00 -2.06283987e-01 4.37079370e-01 5.16021363e-02
2.95709968e-01 4.23614204e-01 -6.71343505e-01 2.83140808e-01
-2.69349933e-01 2.43612543e-01 1.80428401e-01 2.94250160e-01
-1.76684201e-01 5.49699180e-02 -1.25538856e-02 1.13668203e+00
-2.55904734e-01 2.24172622e-01 -6.50951624e-01 3.85969520e-01
-1.45101452e+00 -1.11458230e+00 -4.06346433e-02 2.16249251e+00
1.02142942e+00 4.90514606e-01 1.17249973e-01 6.98831379e-01
6.47806168e-01 -1.29783273e-01 -6.36501849e-01 -7.37934470e-01
-3.84624511e-01 6.27622664e-01 4.97406393e-01 8.64318728e-01
-7.71204591e-01 2.92960942e-01 6.95081091e+00 1.15647244e+00
-1.00862348e+00 2.49049366e-01 7.53642619e-01 -5.48939481e-02
-1.49630889e-01 -6.59172475e-01 -3.04701120e-01 4.97078508e-01
1.24783623e+00 -1.09093882e-01 8.61087859e-01 8.81507635e-01
-1.23553254e-01 4.69810426e-01 -1.24250567e+00 9.56048429e-01
-4.29200470e-01 -1.93340218e+00 9.51135904e-02 2.34274104e-01
6.20117426e-01 -2.99379742e-03 9.20416340e-02 1.67901829e-01
1.73001103e-02 -1.30422711e+00 7.88028598e-01 6.27627194e-01
9.45852697e-01 -9.42854404e-01 5.55937350e-01 4.81707096e-01
-1.08232164e+00 -2.47767001e-01 -1.00342536e+00 -3.43500108e-01
-3.47638428e-01 6.95057511e-01 1.31481821e-02 -1.06124200e-01
4.51755792e-01 2.50646323e-01 1.88123912e-01 9.40675676e-01
3.96345109e-01 5.70423067e-01 -8.84352922e-01 -3.84409517e-01
5.16987443e-01 -2.81701982e-01 2.58724540e-01 1.23918414e+00
4.43346679e-01 4.78504926e-01 -1.22167647e-01 8.70197535e-01
-2.53644586e-01 -1.16747141e-01 -5.51910877e-01 -9.84963402e-02
4.78462845e-01 7.96624124e-01 -6.18358672e-01 -3.74029875e-01
-5.41736633e-02 7.30112791e-01 4.83083427e-01 5.10925353e-01
-5.71915269e-01 -7.69747674e-01 5.40595710e-01 4.65241283e-01
5.87856412e-01 -4.36925232e-01 -5.36811709e-01 -8.93699348e-01
2.60696951e-02 -3.63275379e-01 7.34405220e-02 -5.50565541e-01
-1.11887980e+00 8.28806221e-01 -4.78626192e-01 -5.96929669e-01
6.38986006e-02 -1.03401232e+00 -5.98520637e-01 8.35153043e-01
-1.08694112e+00 -4.61261898e-01 1.66183040e-01 6.89210057e-01
1.47679389e-01 -1.55382812e-01 1.02723658e+00 2.36712500e-01
-1.97470710e-01 6.34316444e-01 6.34033322e-01 2.94200689e-01
-4.86788601e-01 -1.17070425e+00 2.25241885e-01 3.22643727e-01
9.17817950e-02 2.44679451e-01 1.01276159e+00 3.51806916e-02
-1.23490560e+00 -6.62573934e-01 6.64125323e-01 3.11952550e-02
1.08072710e+00 -2.13743955e-01 -1.12693834e+00 4.81586993e-01
3.62495817e-02 6.60033226e-01 4.25882012e-01 -3.80087852e-01
-4.04744208e-01 -7.52291828e-03 -1.28602266e+00 3.81002128e-01
8.85407090e-01 -6.99070930e-01 -3.33739191e-01 4.83985722e-01
6.93628728e-01 -3.55892092e-01 -1.05342579e+00 1.70268074e-01
6.97407484e-01 -9.67405260e-01 8.63842070e-01 -9.70445096e-01
5.35724401e-01 1.87681884e-01 -4.42123085e-01 -1.07444179e+00
-6.53628469e-01 -7.07443655e-01 -8.38167369e-01 3.63218486e-01
7.69987628e-02 -7.73604095e-01 6.82470500e-01 2.05330133e-01
-1.88247025e-01 -1.41417611e+00 -1.33248067e+00 -1.05634987e+00
8.02194297e-01 -2.19364047e-01 4.77478862e-01 8.40673923e-01
1.48134187e-01 3.51454429e-02 1.06803074e-01 -7.92111903e-02
5.32509685e-01 -2.99975276e-01 5.87952808e-02 -1.40134966e+00
-4.26447153e-01 -1.06457853e+00 -6.68696404e-01 -1.27238166e+00
2.80019045e-01 -8.30666959e-01 -3.64023328e-01 -1.15553737e+00
-5.34319341e-01 -4.39361632e-01 -2.18060046e-01 -3.72841805e-02
6.50153399e-01 3.06760192e-01 -3.71882111e-01 1.97604626e-01
-1.65032789e-01 5.46811104e-01 1.16090810e+00 5.40040843e-02
7.12650195e-02 9.08974558e-02 -2.09807545e-01 1.02778053e+00
7.88987160e-01 -1.59961462e-01 -2.90413111e-01 -3.55763286e-01
3.55890512e-01 6.25031739e-02 5.05156457e-01 -1.31781161e+00
2.87172139e-01 3.19641739e-01 6.85057104e-01 -1.32258549e-01
2.81035334e-01 -6.95432246e-01 -1.73051342e-01 7.81195045e-01
-7.79299438e-01 9.64431316e-02 2.26355016e-01 4.43391114e-01
1.33015275e-01 -5.60391068e-01 1.32641244e+00 -3.93068306e-02
1.01687208e-01 4.83044386e-01 -4.01047647e-01 4.68511224e-01
5.09167373e-01 -3.83497417e-01 -2.58762330e-01 -4.68403548e-01
-9.58900094e-01 -4.30157155e-01 2.08473012e-01 -4.52010632e-01
4.53804106e-01 -1.70225608e+00 -6.73185349e-01 1.69291601e-01
-6.62502527e-01 8.35733488e-02 9.56889912e-02 7.31351793e-01
-9.41113651e-01 3.22279990e-01 -2.94416487e-01 -2.35999703e-01
-4.48609293e-01 4.85277385e-01 9.97915268e-01 -8.96153450e-02
-9.26319182e-01 8.90377045e-01 -8.36050604e-03 1.52653143e-01
5.51769674e-01 -6.05012238e-01 3.10109794e-01 -2.37761423e-01
7.55923033e-01 5.10350108e-01 6.54296055e-02 -2.41784945e-01
9.94868726e-02 5.17092407e-01 2.21371964e-01 1.16179690e-01
1.23978233e+00 2.06591710e-01 -3.19351584e-01 5.48210621e-01
1.69087470e+00 -3.79623353e-01 -1.31632566e+00 -2.78541982e-01
-3.47973049e-01 1.67012483e-01 1.65435493e-01 -1.56584412e-01
-9.81875181e-01 1.11072600e+00 3.22263509e-01 1.10152221e+00
7.91886508e-01 -1.66964769e-01 1.04477715e+00 7.68851578e-01
1.82075605e-01 -1.10289216e+00 -7.81810805e-02 8.22863519e-01
8.77949834e-01 -4.94849354e-01 -4.50374931e-01 2.45659992e-01
4.95546401e-01 1.73093784e+00 2.38682374e-01 -7.69800246e-01
1.08077502e+00 3.22063476e-01 -7.15410411e-01 -1.83302239e-02
-4.21381354e-01 3.87508452e-01 3.74633700e-01 3.65843117e-01
3.79216224e-01 -1.60107329e-01 -4.77554575e-02 5.65038383e-01
-7.07549036e-01 -8.79319161e-02 4.61782336e-01 1.57152057e-01
-9.45924580e-01 9.29519348e-03 -1.24682158e-01 5.44583857e-01
-5.13634145e-01 -2.19266981e-01 5.99872991e-02 7.11187303e-01
-4.36521880e-03 2.91593820e-01 4.90829796e-01 -1.07658021e-01
4.64161895e-02 1.62228048e-01 1.05996418e+00 6.33296669e-02
-1.64141759e-01 -5.74173391e-01 -1.16565160e-01 -1.58134416e-01
-7.46877119e-02 -7.23505616e-02 -1.22223771e+00 -1.00937450e+00
-2.98211038e-01 4.09631580e-01 3.42030376e-01 1.12037098e+00
-1.57574117e-02 4.34460998e-01 4.72582370e-01 -1.01663935e+00
-1.07041025e+00 -9.21861589e-01 -7.86443293e-01 5.53644523e-02
1.03720069e+00 -1.93539083e-01 -7.77735770e-01 -3.12119395e-01] | [7.699287414550781, 3.617872714996338] |
4d752fd1-fdaf-4f2b-882a-98e91fc56ba6 | what-is-it-like-to-be-a-bot-simulated | 2103.12638 | null | https://arxiv.org/abs/2103.12638v1 | https://arxiv.org/pdf/2103.12638v1.pdf | What is it Like to Be a Bot: Simulated, Situated, Structurally Coherent Qualia (S3Q) Theory of Consciousness | A novel representationalist theory of consciousness is presented that is grounded in neuroscience and provides a path to artificially conscious computing. Central to the theory are representational affordances of the conscious experience based on the generation of qualia, the fundamental unit of the conscious representation. The current approach is focused on understanding the balance of simulation, situatedness, and structural coherence of artificial conscious representations through converging evidence from neuroscientific and modeling experiments. Representations instantiating a suitable balance of situated and structurally coherent simulation-based qualia are hypothesized to afford the agent the flexibilities required to succeed in rapidly changing environments. | ['S. Rogers', 'M. Molineaux', 'O. Larue', 'H. S. Clouse', 'C. Cox', 'J. Culbertson', 'K. Schmidt'] | 2021-03-13 | null | null | null | null | ['2d-human-pose-estimation', '2d-semantic-segmentation'] | ['computer-vision', 'computer-vision'] | [ 3.20375383e-01 2.22693592e-01 3.52497905e-01 -2.10250765e-01
4.48533833e-01 -5.41544676e-01 1.13827264e+00 1.78703308e-01
-2.94010937e-01 4.29455042e-01 7.67602623e-01 -3.70787054e-01
-2.62871653e-01 -7.57367849e-01 -3.63718390e-01 -4.74693507e-01
-3.03865492e-01 -1.73119307e-01 -2.99811691e-01 -5.19005477e-01
5.04878581e-01 6.52533710e-01 -1.94194317e+00 2.30752766e-01
9.52530503e-01 4.62631017e-01 5.73370636e-01 5.26709616e-01
1.00886397e-01 9.81744349e-01 -3.69455338e-01 2.85456121e-01
1.42451480e-01 -1.17406249e+00 -4.42377955e-01 1.81851581e-01
2.02369779e-01 1.22306131e-01 -3.92131865e-01 1.14604282e+00
5.15243933e-02 2.97540743e-02 5.91002345e-01 -6.35446191e-01
-1.00930083e+00 5.40861011e-01 4.10767257e-01 5.53475320e-01
6.67543828e-01 5.82640588e-01 5.81178248e-01 -6.90439761e-01
6.44682705e-01 1.36566889e+00 8.89028013e-02 7.08965540e-01
-1.39240205e+00 -3.58175859e-02 -5.26113296e-03 1.14128897e-02
-1.25596321e+00 -7.05419481e-01 3.36950302e-01 -5.54018855e-01
1.22991598e+00 3.78896981e-01 1.63768303e+00 7.87598014e-01
1.01011479e+00 -1.48035601e-01 1.72273707e+00 -7.26889372e-01
5.87831318e-01 3.63222957e-02 -3.57553624e-02 5.22447348e-01
7.69608259e-01 7.93245673e-01 -7.20174611e-01 5.27266748e-02
1.36830366e+00 1.32273994e-02 -3.78532916e-01 -3.89171690e-01
-1.59774113e+00 4.38365102e-01 6.36064470e-01 7.11554885e-01
-8.70957315e-01 4.75831836e-01 3.60868126e-02 2.67432928e-01
-3.22694629e-01 1.22589445e+00 1.02066427e-01 1.44349243e-02
-5.52182138e-01 1.26672417e-01 5.49967766e-01 2.92008966e-01
3.75535309e-01 3.95323426e-01 -1.25812236e-02 -2.45983809e-01
5.84951818e-01 5.32429338e-01 7.02430904e-01 -8.91208649e-01
-4.00634825e-01 8.24094653e-01 1.56151146e-01 -8.30378115e-01
-2.97450393e-01 -2.75878072e-01 -1.39200464e-01 6.21188164e-01
-6.38163388e-02 3.58441100e-02 -8.49128664e-01 1.87733507e+00
5.65143721e-03 -1.92939371e-01 5.18647611e-01 9.74330604e-01
5.26862502e-01 5.52401781e-01 4.91816044e-01 -3.78129482e-01
1.30634344e+00 -1.89693421e-01 -8.34695041e-01 -3.58716130e-01
2.73721159e-01 -2.85072207e-01 1.01829791e+00 -1.01769790e-01
-1.37179339e+00 -3.26236904e-01 -1.34804189e+00 1.81453496e-01
-2.99703509e-01 -6.26332104e-01 1.10376418e+00 6.55025065e-01
-1.28892553e+00 5.75584590e-01 -6.27691925e-01 -5.88297784e-01
7.35196471e-02 -1.10331602e-01 -2.13361517e-01 4.93842006e-01
-1.00902402e+00 1.55999017e+00 6.39267564e-01 2.47008219e-01
-1.02375436e+00 -8.48705173e-02 -8.96962285e-01 5.09625413e-02
-2.30616361e-01 -1.00384521e+00 9.99384642e-01 -1.29058397e+00
-1.31002402e+00 1.14753771e+00 -1.60474062e-01 -5.03243089e-01
7.67700449e-02 2.35110417e-01 -5.45575142e-01 2.57705897e-01
-1.56950653e-01 3.97060335e-01 4.70114917e-01 -1.56122947e+00
1.31557688e-01 -3.39869022e-01 5.56470230e-02 5.69331288e-01
2.20010847e-01 2.40542181e-02 7.25509763e-01 -5.03178872e-02
8.92970264e-01 -5.42341173e-01 -5.38850963e-01 -1.84298292e-01
8.43445361e-02 1.52206644e-01 2.03344718e-01 -6.93984628e-02
6.76924050e-01 -2.16787124e+00 2.82198966e-01 1.19219922e-01
5.39080977e-01 -1.07989758e-01 -1.65075529e-02 1.04089308e+00
-7.04864562e-02 4.52868752e-02 1.33128077e-01 5.21036327e-01
3.99328321e-01 1.07920423e-01 -4.37626332e-01 5.71990073e-01
5.81438281e-02 1.02301049e+00 -8.85109007e-01 -2.87584543e-01
4.74488467e-01 4.61232126e-01 -4.45824176e-01 2.52870888e-01
3.24273184e-02 5.88705242e-01 -3.09255868e-01 3.16814870e-01
2.69114017e-01 -1.97171897e-01 7.21306622e-01 2.27903515e-01
-6.71985269e-01 8.43912840e-01 -8.03137481e-01 1.42081892e+00
-2.51848996e-01 6.70613170e-01 -3.73287126e-02 -4.95461941e-01
7.97967374e-01 4.79430914e-01 -1.51161060e-01 -1.18985331e+00
4.28123921e-01 7.49616995e-02 9.24947023e-01 -5.76457500e-01
5.83736360e-01 -7.24622786e-01 -1.23856828e-01 6.76475823e-01
-1.06007405e-01 -7.42720306e-01 -2.22788267e-02 3.28158557e-01
7.59395182e-01 3.15264344e-01 6.76352680e-01 -9.17118013e-01
1.11649826e-01 -2.04144344e-01 3.39924157e-01 6.22205019e-01
-4.41747308e-01 7.78121054e-02 3.63465361e-02 -6.40135407e-01
-1.01185846e+00 -1.46177399e+00 -4.13707763e-01 5.17194390e-01
6.20226085e-01 -2.08389595e-01 -8.73741150e-01 5.67756414e-01
-4.72273380e-01 1.48109663e+00 -8.37699890e-01 -4.42717999e-01
-1.90040395e-01 -4.11691695e-01 -5.01272082e-02 1.38284430e-01
2.64196038e-01 -1.80705380e+00 -2.15835619e+00 3.11376512e-01
7.32728466e-02 -8.89502466e-01 4.21172887e-01 3.19745868e-01
-9.75448012e-01 -6.76015317e-01 3.11285436e-01 -4.69079494e-01
7.95964360e-01 5.25561929e-01 1.02910399e+00 3.79649192e-01
-5.44999540e-01 6.07999563e-01 6.53732792e-02 -4.52576876e-01
-6.30714297e-01 -8.16593051e-01 2.19739065e-01 -4.72603142e-01
4.76896726e-02 -9.86926317e-01 -5.31640112e-01 -1.98825523e-01
-9.52184141e-01 6.67857587e-01 4.11338627e-01 4.04666096e-01
3.92970383e-01 -2.43443832e-01 4.28489178e-01 -1.69393882e-01
8.44160914e-01 -4.30644870e-01 -2.44213983e-01 3.23998705e-02
-4.25278008e-01 -2.25226134e-02 -1.96072776e-02 -3.48535031e-01
-1.18885767e+00 -5.48442662e-01 4.38054264e-01 3.94502670e-01
-2.62339354e-01 6.22352540e-01 -3.20025593e-01 -2.57094149e-02
9.62419271e-01 5.85326612e-01 3.58829573e-02 2.40097076e-01
6.82912469e-01 1.15997728e-03 5.11446774e-01 -6.68790460e-01
4.34653163e-01 5.29475152e-01 -1.01296462e-01 -6.86047912e-01
-3.61220241e-01 4.09425855e-01 -7.69398928e-01 -5.01131177e-01
7.84370005e-01 -9.02921200e-01 -6.35174692e-01 -6.34190291e-02
-8.60297561e-01 -3.15757126e-01 -9.41269815e-01 7.41346776e-01
-1.00680375e+00 7.08885537e-03 -3.03857923e-01 -1.03737485e+00
-2.09891975e-01 -9.95174825e-01 2.20451683e-01 4.78402108e-01
-6.97409451e-01 -8.77139330e-01 2.30313689e-01 -1.99328557e-01
6.49288595e-01 6.80871129e-01 9.01598334e-01 -3.03873986e-01
-8.20777774e-01 -1.53601155e-01 9.04470831e-02 -2.84898847e-01
9.41855460e-02 3.41245756e-02 -8.50356698e-01 -3.40157636e-02
4.84791458e-01 -4.51538175e-01 4.91498679e-01 1.64310232e-01
3.33393484e-01 -3.88250351e-01 -2.21446052e-01 4.25736874e-01
1.68532503e+00 4.59455729e-01 1.10727906e+00 4.24512267e-01
-8.79623517e-02 8.44545305e-01 1.62334070e-01 2.66373992e-01
2.88858712e-01 7.96787217e-02 3.29550415e-01 1.78636506e-01
-2.95587089e-02 -3.15720052e-01 3.68254304e-01 1.00320745e+00
-2.94917136e-01 3.37928951e-01 -7.69302726e-01 4.61732894e-01
-1.44879997e+00 -1.49880302e+00 -1.81085849e-03 2.17026830e+00
7.38385737e-01 2.74368167e-01 -4.86639559e-01 -1.39001593e-01
6.46184862e-01 1.30461887e-01 -4.56686288e-01 -9.41430748e-01
-2.71367431e-01 1.37830555e-01 -2.00100273e-01 2.55318820e-01
-2.60447800e-01 9.26746786e-01 8.58045959e+00 -4.81199101e-02
-9.17209566e-01 -2.48025265e-02 -7.23730922e-02 -1.26736820e-01
-7.67287076e-01 3.37111622e-01 -2.20301338e-02 1.94805443e-01
1.30974257e+00 -6.60260022e-01 7.15071023e-01 4.09564912e-01
4.80412185e-01 -5.78203321e-01 -1.02392113e+00 3.86253506e-01
1.76065043e-02 -1.51739073e+00 1.55795708e-01 3.88214737e-01
5.89361727e-01 -3.62433374e-01 2.09007770e-01 -1.57731757e-01
3.39448780e-01 -1.51738381e+00 1.49200177e+00 7.43065715e-01
6.23602629e-01 -2.20215216e-01 2.07730487e-01 2.05143467e-01
-7.20631480e-01 -1.35370597e-01 -3.72544855e-01 -9.62067783e-01
2.36863077e-01 5.29682711e-02 -3.23579818e-01 -1.52120560e-01
7.79021010e-02 8.07712451e-02 -3.30035299e-01 8.82147193e-01
-3.42098206e-01 1.84362411e-01 1.47276700e-01 -2.23087624e-01
1.24166124e-01 -3.08185577e-01 6.87372863e-01 9.09986794e-01
1.83791652e-01 6.14846051e-01 -3.23277503e-01 1.57696795e+00
6.12255991e-01 -3.25695649e-02 -8.89488399e-01 -4.10904914e-01
8.36665154e-01 1.08910048e+00 -1.06311035e+00 -5.09593904e-01
2.75043666e-01 6.21145666e-01 1.34802600e-02 2.26119071e-01
-6.20913267e-01 1.83069348e-01 7.09331691e-01 9.12340060e-02
-9.98961851e-02 -7.25838482e-01 -5.64171672e-01 -1.11788821e+00
-6.35217786e-01 -9.71580863e-01 -4.37041789e-01 -8.72895181e-01
-8.08047414e-01 6.10498250e-01 2.55256370e-02 -5.55759013e-01
-1.47211283e-01 -3.51752788e-01 -8.89904201e-01 1.11404216e+00
-7.86195219e-01 -1.22394288e+00 -3.02396208e-01 5.83807588e-01
-7.69145368e-03 1.83705211e-01 1.35988498e+00 -8.89310241e-01
-3.42125177e-01 -2.23742560e-01 -2.64164418e-01 -5.40802717e-01
-1.90635413e-01 -1.11564016e+00 5.40718555e-01 9.68163431e-01
-8.38841200e-02 1.32605088e+00 1.09363627e+00 -5.94027102e-01
-1.69439840e+00 -3.33878130e-01 7.82684803e-01 -5.69253027e-01
5.59000254e-01 -2.41437450e-01 -7.74708688e-01 8.03093314e-01
5.20226598e-01 -4.33674306e-01 8.57844234e-01 -8.00472572e-02
-4.69072759e-01 4.58564937e-01 -1.33730483e+00 1.05590761e+00
1.15861559e+00 -6.34451389e-01 -1.33842289e+00 -1.64651155e-01
6.41487896e-01 -8.31799507e-02 -4.46097136e-01 6.56841770e-02
6.64159894e-01 -1.21462727e+00 7.74681568e-01 -3.26138824e-01
8.02801475e-02 -3.96453232e-01 -4.02383745e-01 -9.95422781e-01
-7.14868367e-01 -6.34572327e-01 1.56655923e-01 4.25535709e-01
2.13423409e-02 -1.17034662e+00 -1.48482651e-01 8.07605267e-01
-2.70045638e-01 -1.08569570e-01 -8.40715349e-01 -4.17170554e-01
1.01919860e-01 -1.10309184e-01 7.52923489e-01 9.14367318e-01
9.12937164e-01 2.69905895e-01 5.46305478e-01 -1.42720237e-01
6.86127782e-01 2.68520057e-01 2.08984464e-01 -1.33909452e+00
-2.73316000e-02 -5.44957995e-01 -6.72560632e-01 -1.46281883e-01
-1.23642065e-01 -8.94535124e-01 1.65240392e-01 -1.85200179e+00
4.27547961e-01 4.16340157e-02 -3.51648510e-01 3.32595930e-02
3.89992118e-01 -1.63210377e-01 6.27617180e-01 3.75667393e-01
-3.27450424e-01 5.64478397e-01 1.42438614e+00 7.59352863e-01
-3.69330496e-01 -1.09648848e+00 -1.24063289e+00 9.03710425e-01
8.84895384e-01 -4.09614146e-02 -5.43983698e-01 9.72132608e-02
3.29878420e-01 1.29126146e-01 7.80202925e-01 -1.21455669e+00
-4.89234850e-02 -7.21553385e-01 5.57924390e-01 -2.07890850e-02
5.25386631e-01 -5.41522682e-01 4.95287538e-01 1.05843961e+00
-4.67029005e-01 2.21824795e-01 2.55735368e-01 3.44516218e-01
3.27664107e-01 7.85393044e-02 9.12277818e-01 -6.01472557e-01
-7.64376521e-01 -5.49933612e-01 -1.20023847e+00 -1.90038681e-01
1.01219440e+00 -7.31168628e-01 -3.09450448e-01 -1.04750648e-01
-7.56232977e-01 -5.36997080e-01 1.05866408e+00 3.35276015e-02
7.66693950e-01 -1.12835371e+00 -4.53376204e-01 2.83829272e-01
-2.46232059e-02 -8.27495873e-01 1.02210358e-01 4.59489226e-01
-6.27702415e-01 4.61242110e-01 -1.04023325e+00 -1.16136178e-01
-5.67389190e-01 6.02193117e-01 6.87574148e-01 4.70413595e-01
-8.76343787e-01 5.48810720e-01 4.86888796e-01 9.32174399e-02
-6.07877970e-01 -2.28866205e-01 -1.99538052e-01 -3.07016224e-01
7.10872948e-01 -1.28470892e-02 -6.11043513e-01 -8.45695794e-01
-3.92734081e-01 1.44949213e-01 4.66389716e-01 -7.34124839e-01
8.78849804e-01 -3.08257014e-01 -7.63530374e-01 8.25987875e-01
1.18118986e-01 -7.47081712e-02 -1.19332135e+00 3.77378255e-01
-2.08065376e-01 -6.22873127e-01 -1.95497088e-03 -9.66794491e-01
-2.87058540e-02 8.75545800e-01 5.78119874e-01 3.81127417e-01
7.67306864e-01 -3.57848033e-02 2.71222051e-02 1.72183737e-01
7.02693522e-01 -7.92553365e-01 3.59200090e-01 2.93156266e-01
1.14273870e+00 -3.15604419e-01 -4.20273340e-04 1.93912983e-02
-5.74951887e-01 8.95401597e-01 6.03705764e-01 -4.91312742e-01
2.80987471e-01 1.43015400e-01 -2.83915307e-02 -6.90245688e-01
-1.06736636e+00 -3.85666341e-01 8.44661519e-02 8.33179891e-01
7.15781033e-01 5.59908509e-01 -7.21478701e-01 1.98509037e-01
-6.36212468e-01 -2.55427390e-01 8.23472559e-01 1.01460576e+00
-1.10881412e+00 -4.01668847e-01 -5.88637590e-01 -1.22867040e-01
6.61868544e-04 -2.73640096e-01 -4.20726627e-01 6.77558661e-01
5.10914326e-01 1.00399101e+00 3.84069353e-01 -2.26003584e-02
-5.91631643e-02 1.36478305e-01 1.19626760e+00 -7.77560651e-01
-6.53616548e-01 -2.03863960e-02 -1.16574718e-03 -5.66671669e-01
-3.73747915e-01 -1.00732517e+00 -1.83376873e+00 -4.94711399e-01
-7.25555941e-02 1.91890463e-01 7.53934920e-01 8.02853525e-01
3.72643501e-01 6.78383946e-01 1.24235496e-01 -1.06122530e+00
-2.36717537e-01 -8.76309514e-01 -6.50429785e-01 3.09188753e-01
1.50478676e-01 -5.50238669e-01 -3.50538969e-01 9.51595418e-03] | [8.817514419555664, 6.283960819244385] |
ec282c0d-bbd9-4b25-9ad5-d491333dd4bc | enhancing-covid-19-diagnosis-through-vision | 2306.06914 | null | https://arxiv.org/abs/2306.06914v2 | https://arxiv.org/pdf/2306.06914v2.pdf | Enhancing COVID-19 Diagnosis through Vision Transformer-Based Analysis of Chest X-ray Images | The advent of 2019 Coronavirus (COVID-19) has engendered a momentous global health crisis, necessitating the identification of the ailment in individuals through diverse diagnostic modalities. Radiological imaging, particularly the deployment of X-ray imaging, has been recognized as a pivotal instrument in the detection and characterization of COVID-19. Recent investigations have unveiled invaluable insights pertaining to the virus within X-ray images, instigating the exploration of methodologies aimed at augmenting diagnostic accuracy through the utilization of artificial intelligence (AI) techniques. The current research endeavor posits an innovative framework for the automated diagnosis of COVID-19, harnessing raw chest X-ray images, specifically by means of fine-tuning pre-trained Vision Transformer (ViT) models. The developed models were appraised in terms of their binary classification performance, discerning COVID-19 from Normal cases, as well as their ternary classification performance, discriminating COVID-19 from Pneumonia and Normal instances, and lastly, their quaternary classification performance, discriminating COVID-19 from Bacterial Pneumonia, Viral Pneumonia, and Normal conditions, employing distinct datasets. The proposed model evinced extraordinary precision, registering results of 99.92% and 99.84% for binary classification, 97.95% and 86.48% for ternary classification, and 86.81% for quaternary classification, respectively, on the respective datasets. | ['Sultan Zavrak'] | 2023-06-12 | null | null | null | null | ['covid-19-detection'] | ['medical'] | [ 4.95740831e-01 -3.66452426e-01 1.22321337e-01 -1.43927401e-02
-5.34628630e-01 -5.50137162e-01 8.04319143e-01 3.67162824e-01
-4.23027575e-01 6.45245492e-01 -2.61828452e-01 -5.37758887e-01
-6.04653716e-01 -4.62656379e-01 -1.53851151e-01 -9.61462259e-01
-4.90188152e-02 9.25556898e-01 -3.56824964e-01 2.32118323e-01
1.25562206e-01 7.11644530e-01 -1.39840209e+00 1.67518541e-01
8.43805850e-01 1.02452397e+00 4.12766814e-01 9.66777325e-01
1.29653394e-01 5.95166564e-01 -5.44825315e-01 -3.86992991e-01
-2.21424043e-01 -3.00246984e-01 -5.28778791e-01 8.12597275e-02
3.37081328e-02 -2.43949786e-01 8.60425755e-02 5.27438939e-01
2.51445234e-01 -2.59616345e-01 1.17290664e+00 -1.04690731e+00
-5.58415890e-01 -2.32610434e-01 -3.74946475e-01 6.22419894e-01
4.30106252e-01 3.78315300e-01 6.61621094e-01 -6.84005380e-01
4.94714558e-01 6.70284510e-01 6.21038437e-01 3.56031001e-01
-8.13683927e-01 -2.56410092e-01 -5.31228960e-01 4.05120313e-01
-1.15093195e+00 1.24622211e-01 5.94240904e-01 -1.04893100e+00
8.83591056e-01 5.80320716e-01 9.53700900e-01 1.13032317e+00
5.95169187e-01 3.45805228e-01 1.57906449e+00 -3.02858651e-01
3.79018001e-02 2.20838428e-01 2.81263977e-01 6.66721284e-01
5.09171844e-01 1.57699555e-01 1.98643044e-01 -3.64878267e-01
5.44931233e-01 4.81558353e-01 -2.63968915e-01 -7.16499165e-02
-1.11967432e+00 7.03587174e-01 1.54175013e-01 5.76533079e-01
-8.21489871e-01 -4.88097519e-01 6.74802482e-01 -7.86399171e-02
6.43932372e-02 3.88257712e-01 -2.87373513e-01 1.12707270e-02
-7.90914178e-01 4.47969064e-02 3.36254001e-01 2.62958020e-01
1.43801227e-01 -7.96867162e-02 -1.50190935e-01 7.62036145e-01
3.45807314e-01 9.94643390e-01 6.80360615e-01 -3.48320276e-01
2.48339444e-01 8.13900292e-01 -1.09770477e-01 -9.49582815e-01
-5.55455983e-01 -5.33762336e-01 -1.20241868e+00 -1.64858401e-01
8.34990963e-02 3.53209232e-03 -9.89587843e-01 1.36444771e+00
3.32492977e-01 1.05094925e-01 1.74200177e-01 8.84314477e-01
5.68034351e-01 6.28803194e-01 3.01604897e-01 -3.64383340e-01
1.76472032e+00 -2.34066904e-01 -4.21411186e-01 2.78853744e-01
2.10031226e-01 -5.87209165e-01 7.98071623e-01 5.20767570e-01
-6.29114807e-01 -4.50456083e-01 -9.63833034e-01 5.80436170e-01
-2.57835954e-01 3.66312355e-01 5.97376049e-01 7.38302112e-01
-6.94824815e-01 7.00271428e-02 -7.00367332e-01 -5.11419296e-01
3.84079844e-01 2.89518833e-01 -3.79785150e-01 -3.07046250e-02
-8.94094527e-01 9.97356534e-01 1.49289042e-01 2.27649808e-01
-8.89481246e-01 -5.84643722e-01 -3.61729950e-01 -1.49722084e-01
1.05655650e-02 -1.04229176e+00 7.54619002e-01 -3.03513348e-01
-9.34527338e-01 1.06261957e+00 1.63971692e-01 -4.50852811e-01
4.39398885e-01 -5.44852465e-02 -7.06349850e-01 6.34285986e-01
2.48811711e-02 1.83933556e-01 8.44002306e-01 -1.13821375e+00
-5.02787292e-01 -8.07061672e-01 -2.92093426e-01 4.91957255e-02
-1.58758268e-01 -5.37909381e-02 2.11826265e-01 -5.70456743e-01
-6.78115040e-02 -8.31085742e-01 1.16798110e-01 -2.40471557e-01
-2.27735937e-01 -3.95485878e-01 9.75173771e-01 -7.72346020e-01
7.83713281e-01 -1.86687362e+00 -1.29630223e-01 3.55650246e-01
4.87669289e-01 9.37265873e-01 4.14553076e-01 3.96104306e-01
4.68726791e-02 -8.69596824e-02 -3.52999121e-01 8.55797306e-02
-1.26850039e-01 3.22832435e-01 -3.26540112e-01 6.73292458e-01
4.56457704e-01 7.56473601e-01 -5.98818898e-01 -6.26487553e-01
5.64364791e-01 8.06672335e-01 -1.35453925e-01 4.44209069e-01
-9.53742936e-02 5.53703368e-01 -6.32735133e-01 8.96003962e-01
3.97829920e-01 -5.20693600e-01 1.27520368e-01 -2.58537065e-02
9.48960558e-02 -4.99559879e-01 -6.19977355e-01 6.15876853e-01
-3.05327743e-01 5.12573242e-01 -1.07942671e-01 -1.11103261e+00
8.14153373e-01 6.33277893e-01 6.01478279e-01 -7.49346375e-01
4.94984686e-01 1.17287189e-01 -1.78157255e-01 -1.15515387e+00
-4.00039181e-02 -3.03908765e-01 1.80575192e-01 3.78698885e-01
-3.35010290e-01 3.59459706e-02 -4.09550965e-02 -1.72178298e-01
9.17489827e-01 -3.39871615e-01 3.10405225e-01 1.34321153e-01
8.96792948e-01 1.69129863e-01 -1.75805807e-01 6.11583829e-01
-2.30066657e-01 3.94993335e-01 3.18579078e-02 -4.12526518e-01
-8.96808267e-01 -1.30237353e+00 -6.46891654e-01 3.92020077e-01
-8.02568346e-02 3.69275510e-01 -7.18585491e-01 -4.30536449e-01
-1.54533416e-01 3.65231723e-01 -4.79535371e-01 -3.11927795e-02
-5.62118351e-01 -9.19775963e-01 6.97669864e-01 3.76716971e-01
5.18681705e-01 -9.66317832e-01 -1.33109355e+00 2.65951129e-03
-4.27983031e-02 -9.03390288e-01 1.67189807e-01 -1.54185509e-02
-9.40301895e-01 -1.54993880e+00 -9.86786306e-01 -6.72768295e-01
5.65358996e-01 1.87129617e-01 7.90401101e-01 3.66665512e-01
-9.45995688e-01 6.04913294e-01 -2.24495381e-01 -3.55587631e-01
-4.79148835e-01 -2.89993316e-01 7.88991451e-02 -5.02288789e-02
2.35901415e-01 -3.26157123e-01 -7.08434105e-01 1.77453786e-01
-8.67539883e-01 -1.68232769e-01 9.48693037e-01 7.60835588e-01
5.66122472e-01 -2.01466635e-01 5.82310140e-01 -7.16647923e-01
6.08413458e-01 -7.29869962e-01 -3.03605020e-01 2.42254868e-01
-9.19990063e-01 -4.91543442e-01 7.04994202e-01 -2.06058219e-01
-8.87710631e-01 -3.88711691e-01 -2.94754028e-01 -4.14260536e-01
-5.49392164e-01 3.85123253e-01 3.28185439e-01 2.13291183e-01
3.99119467e-01 4.56768543e-01 6.81310371e-02 -3.13225865e-01
-1.02298193e-01 1.02930069e+00 8.74813020e-01 -3.35162312e-01
4.90491331e-01 5.47130823e-01 1.00606896e-01 -1.16933239e+00
-3.66567433e-01 -6.65662169e-01 -4.60626960e-01 -4.47449207e-01
1.30185628e+00 -5.80258906e-01 -7.28781998e-01 6.16799176e-01
-8.73759329e-01 5.37164748e-01 1.16484627e-01 6.52988315e-01
-3.67305100e-01 5.09906769e-01 -4.90104288e-01 -8.00892889e-01
-7.85630167e-01 -1.26028562e+00 8.45213652e-01 1.23349100e-01
-4.13419753e-01 -8.51360917e-01 3.18421572e-01 8.39767933e-01
4.21888918e-01 5.17427027e-01 1.46975756e+00 -9.26155567e-01
-4.01654541e-01 -5.64017355e-01 -4.29917455e-01 3.14592332e-01
8.61243159e-02 4.68875244e-02 -9.79525805e-01 -2.43283480e-01
2.64902025e-01 -1.53809473e-01 3.89230818e-01 3.33738059e-01
6.72091067e-01 -1.40954822e-01 -3.81882787e-01 4.34262693e-01
1.37816167e+00 8.06035936e-01 3.20144773e-01 2.39234850e-01
5.35916209e-01 3.26159209e-01 5.27076721e-01 4.65190917e-01
1.86110765e-01 4.16178048e-01 6.73674345e-01 -1.78166226e-01
-3.27420980e-03 3.20225537e-01 -3.16343337e-01 5.19510806e-01
-3.53207797e-01 -1.53164461e-01 -1.12911057e+00 4.20731068e-01
-1.07568109e+00 -8.84515464e-01 -4.01422054e-01 1.68174338e+00
3.15407455e-01 -5.89243472e-02 3.28239016e-02 5.55647194e-01
7.57415414e-01 -2.14655455e-02 -3.70926857e-01 -4.93330479e-01
8.68472680e-02 3.22194278e-01 -1.63252026e-01 -9.23682842e-03
-1.12405872e+00 1.04719782e-02 6.05663872e+00 1.51657730e-01
-1.47837818e+00 1.66344754e-02 5.58468401e-01 3.52755487e-01
-9.17612016e-02 -4.79569376e-01 -4.56092656e-01 6.08947635e-01
7.48326659e-01 1.95030645e-01 -1.07074708e-01 7.94203460e-01
1.30919963e-01 -2.45883670e-02 -6.60126805e-01 9.02374804e-01
3.22513938e-01 -1.21753597e+00 1.14910461e-01 1.64473448e-02
4.40222561e-01 -8.30071867e-02 2.64811963e-01 8.27354286e-03
-4.90214080e-01 -1.02288997e+00 1.74164489e-01 6.71382546e-01
8.49324405e-01 -6.01389170e-01 1.19529295e+00 2.48186067e-01
-7.52158880e-01 1.07511170e-01 1.19269684e-01 8.44526067e-02
1.31997272e-01 4.45392579e-01 -1.43294668e+00 4.43603456e-01
6.47415102e-01 2.93569386e-01 -3.21037441e-01 9.41605508e-01
1.20344579e-01 7.06103146e-01 -1.57579035e-01 -1.35014102e-01
2.16884583e-01 -3.39155465e-01 6.45421445e-01 1.27664721e+00
1.73230529e-01 2.41577208e-01 -1.97750881e-01 6.21467352e-01
4.48435396e-01 2.00031966e-01 -5.49113512e-01 -1.73668548e-01
2.17780709e-01 9.92397726e-01 -8.15040171e-01 -5.36784172e-01
-1.93010777e-01 5.98984361e-01 -2.40201578e-01 1.55620024e-01
-1.02209044e+00 -1.66754410e-01 3.27423453e-01 2.34361395e-01
3.69295508e-01 7.36013800e-02 -3.15309942e-01 -6.68865442e-01
-5.49047254e-03 -7.28998482e-01 4.12822455e-01 -8.00474703e-01
-1.14163160e+00 8.63570452e-01 1.68754578e-01 -1.08613491e+00
-4.48707193e-01 -7.28056252e-01 -6.79255068e-01 8.03570032e-01
-1.27849340e+00 -1.19943619e+00 -4.87560630e-01 4.50369269e-01
1.70574248e-01 -3.63826364e-01 9.77871656e-01 3.08146417e-01
-4.90590006e-01 3.01245004e-01 1.36986926e-01 -1.60873383e-01
1.81348369e-01 -1.04201877e+00 -3.20829004e-01 6.19195402e-01
-3.52450430e-01 6.02409363e-01 5.88124633e-01 -5.21274388e-01
-1.38832200e+00 -9.75345612e-01 8.63828421e-01 -3.07503790e-01
3.41828376e-01 2.00662985e-01 -8.57493877e-01 3.01034003e-01
-1.11592390e-01 -2.43232325e-01 8.24574232e-01 -5.20237207e-01
-1.45280629e-01 2.07561806e-01 -1.43511713e+00 2.22629145e-01
4.16307390e-01 -6.56841457e-01 -9.14708376e-01 2.71862149e-01
2.61872500e-01 -1.13750644e-01 -1.09023488e+00 6.57539606e-01
6.72444403e-01 -9.78939235e-01 1.28850365e+00 -5.21300375e-01
4.52118039e-01 -1.53431594e-01 -2.45736837e-01 -5.91960609e-01
-1.65875822e-01 1.89743891e-01 -6.50491714e-02 8.51323724e-01
-7.50146359e-02 -7.77726114e-01 5.94206154e-01 1.23695798e-01
-1.06152356e-01 -1.15438831e+00 -9.71276879e-01 -2.68750757e-01
-4.43653643e-01 -3.98001164e-01 3.72350872e-01 9.29275334e-01
-6.11845911e-01 6.01028763e-02 -3.96803692e-02 2.93917984e-01
3.95456731e-01 2.32296720e-01 3.70203853e-01 -1.37962294e+00
-4.30459291e-01 -3.73524427e-01 -7.73336828e-01 -1.96046114e-01
-4.03412193e-01 -7.59434700e-01 -4.78435814e-01 -1.57693923e+00
2.89668530e-01 -4.30384368e-01 -3.73449504e-01 2.40384355e-01
-1.46008506e-01 6.89592659e-01 1.12740643e-01 4.57702309e-01
-2.30563153e-02 1.32231817e-01 1.08665586e+00 -1.78195626e-01
1.49146393e-01 2.90309191e-01 -3.98340613e-01 7.28516340e-01
7.14859247e-01 -3.00539345e-01 -4.38743412e-01 -2.14844849e-02
-3.43883829e-03 3.85392576e-01 7.85810709e-01 -1.14949405e+00
-2.57767409e-01 -9.43697095e-02 5.12877166e-01 -9.57556188e-01
4.46406871e-01 -8.37491989e-01 4.64879811e-01 1.03184116e+00
6.31545624e-03 2.73103982e-01 -4.97496575e-02 5.50988257e-01
-9.22624394e-02 -2.14817241e-01 7.81520247e-01 1.24765068e-01
-6.30943179e-01 -5.47925346e-02 -7.00044930e-01 6.76671881e-03
1.43939137e+00 -4.86933470e-01 -3.27964395e-01 2.99846411e-01
-6.52585924e-01 -1.70671761e-01 2.14702696e-01 2.09695905e-01
8.46392035e-01 -7.06903875e-01 -6.32517815e-01 5.12834370e-01
1.66259944e-01 -1.66970044e-01 4.98431742e-01 1.13960218e+00
-8.76546323e-01 6.61760688e-01 -3.67919236e-01 -1.02665401e+00
-1.79822671e+00 6.19451404e-01 1.70082837e-01 -3.20019990e-01
-7.16883540e-01 5.02559662e-01 6.13759421e-02 9.37753543e-02
1.01231895e-02 -2.08334550e-01 -5.59061944e-01 9.38302949e-02
4.62476999e-01 5.29912651e-01 3.13096195e-01 -7.14397252e-01
-4.31670398e-01 9.33186173e-01 -1.86246678e-01 3.90019327e-01
1.19487298e+00 4.04985771e-02 -1.67512268e-01 1.02649413e-01
1.16022265e+00 -3.76011223e-01 -4.65504080e-01 1.10797174e-01
-1.96262211e-01 -1.62132561e-01 -1.57071784e-01 -1.09460104e+00
-5.40790915e-01 8.87565136e-01 9.82072771e-01 2.49839246e-01
1.15722489e+00 1.64187074e-01 6.52799666e-01 2.56581813e-01
3.17880988e-01 -4.18732852e-01 -5.15957326e-02 1.52369842e-01
5.72061658e-01 -1.00158179e+00 4.75348309e-02 -1.60331532e-01
-6.10750496e-01 9.55102801e-01 1.02262005e-01 -1.15779778e-02
3.91417563e-01 6.13842160e-02 1.52140975e-01 -6.32464647e-01
-6.06569409e-01 2.39193529e-01 2.32712805e-01 8.71231496e-01
1.90639272e-01 5.05993605e-01 -3.79067093e-01 5.65167904e-01
6.94338456e-02 1.00276977e-01 1.52595937e-01 9.70217407e-01
-4.27888304e-01 -6.15257800e-01 -8.06995511e-01 8.70194733e-01
-5.39015830e-01 1.08119912e-01 -2.19536841e-01 1.01223290e+00
4.20046717e-01 7.27842629e-01 2.70072073e-02 -1.95758313e-01
1.51933923e-01 9.51953530e-02 3.42130989e-01 -2.14482352e-01
-4.46121544e-01 -1.73542663e-01 -8.25234503e-02 1.18692508e-02
-3.69268715e-01 -5.92416704e-01 -1.21418715e+00 5.58262281e-02
-1.38166294e-01 3.59701008e-01 8.51762056e-01 1.26178873e+00
9.02620032e-02 6.75557613e-01 6.38561487e-01 -4.24177498e-01
-6.99682236e-01 -7.41871715e-01 -3.59980553e-01 3.02860320e-01
5.05276144e-01 -6.39310002e-01 -3.14295828e-01 1.05341122e-01] | [15.554816246032715, -1.6875970363616943] |
4210f546-1ffb-4221-973c-ca4bc9a076ac | general-e2-equivariant-steerable-cnns | null | null | http://papers.nips.cc/paper/9580-general-e2-equivariant-steerable-cnns | http://papers.nips.cc/paper/9580-general-e2-equivariant-steerable-cnns.pdf | General E(2)-Equivariant Steerable CNNs | The big empirical success of group equivariant networks has led in recent years to the sprouting of a great variety of equivariant network architectures. A particular focus has thereby been on rotation and reflection equivariant CNNs for planar images. Here we give a general description of E(2)-equivariant convolutions in the framework of Steerable CNNs. The theory of Steerable CNNs thereby yields constraints on the convolution kernels which depend on group representations describing the transformation laws of feature spaces. We show that these constraints for arbitrary group representations can be reduced to constraints under irreducible representations. A general solution of the kernel space constraint is given for arbitrary representations of the Euclidean group E(2) and its subgroups. We implement a wide range of previously proposed and entirely new equivariant network architectures and extensively compare their performances. E(2)-steerable convolutions are further shown to yield remarkable gains on CIFAR-10, CIFAR-100 and STL-10 when used as drop in replacement for non-equivariant convolutions. | ['Maurice Weiler', 'Gabriele Cesa'] | 2019-12-01 | null | null | null | neurips-2019-12 | ['rotated-mnist'] | ['computer-vision'] | [-1.93912778e-02 3.67489606e-01 3.95051479e-01 -7.22708821e-01
1.21882871e-01 -6.95701063e-01 9.97491241e-01 -1.05287862e+00
-5.90524793e-01 3.41141433e-01 5.87773263e-01 -3.87814760e-01
-4.61601973e-01 -6.40283942e-01 -7.04990089e-01 -7.51723230e-01
-2.06336498e-01 1.19556502e-01 1.86417371e-01 -8.06727469e-01
-3.53115678e-01 1.08699429e+00 -1.10095704e+00 2.16397151e-01
1.00785807e-01 4.42381531e-01 -1.30211592e-01 9.97284949e-01
4.32736903e-01 6.30065799e-01 1.32672668e-01 -2.60004878e-01
7.43721664e-01 4.18481790e-02 -9.59957302e-01 3.35326754e-02
1.07375908e+00 -9.06971917e-02 -1.11508000e+00 1.20968032e+00
3.77280265e-01 4.18733299e-01 6.91163123e-01 -9.31132376e-01
-1.03788733e+00 9.63447988e-01 2.81925723e-02 2.82402337e-01
-6.73821151e-01 -3.41263384e-01 1.20484936e+00 -1.19276428e+00
6.32770538e-01 1.57367682e+00 8.07160497e-01 6.60435379e-01
-1.28153086e+00 -4.05840129e-01 1.84279352e-01 1.15234271e-01
-1.22863078e+00 -6.29921138e-01 5.11431396e-01 -2.01429591e-01
1.05696344e+00 3.88643831e-01 4.14382279e-01 1.00416863e+00
2.56047457e-01 2.28886709e-01 7.67540455e-01 -3.92255962e-01
3.46502773e-02 -2.97187209e-01 4.12406087e-01 8.63857388e-01
3.95204127e-01 -2.65192002e-01 2.07676012e-02 1.04315370e-01
1.42611122e+00 1.52614772e-01 -2.16490224e-01 -7.14443564e-01
-1.54364944e+00 1.00863731e+00 8.83006752e-01 1.77696735e-01
7.08225416e-03 5.70919037e-01 3.07722896e-01 5.49730659e-01
6.63795993e-02 4.30302143e-01 -3.57675403e-01 6.57974482e-01
-1.05076544e-01 2.63936132e-01 6.88416243e-01 1.21510136e+00
6.84507430e-01 7.20851481e-01 5.29255532e-02 6.95370793e-01
3.22254688e-01 3.35865796e-01 2.45021388e-01 -9.88047302e-01
2.50830680e-01 2.26257578e-01 -1.94503218e-01 -8.69279444e-01
-7.44908154e-01 -4.12446350e-01 -9.64140415e-01 3.32151830e-01
5.67172766e-01 -1.37027264e-01 -1.11834812e+00 1.98504269e+00
-1.36567175e-01 4.86286692e-02 2.10021228e-01 1.02790892e+00
6.31058097e-01 2.91289091e-01 -1.80174530e-01 5.22659361e-01
1.34368551e+00 -7.02225447e-01 -4.29111093e-01 2.62058377e-01
4.60391134e-01 -7.84973145e-01 7.06732035e-01 -1.26658589e-01
-9.44300413e-01 -3.66000354e-01 -1.28654861e+00 -2.15368599e-01
-5.21558583e-01 3.36499922e-02 1.10121739e+00 6.84771836e-01
-1.55792296e+00 6.71104312e-01 -1.01930904e+00 -4.92181510e-01
3.33032250e-01 8.29971671e-01 -7.42097735e-01 1.70991927e-01
-1.13901365e+00 7.00653493e-01 1.64364681e-01 3.37633461e-01
-8.72853577e-01 -7.06000388e-01 -7.31033385e-01 3.51408087e-02
-2.62188017e-01 -8.08659434e-01 1.10665762e+00 -1.20081341e+00
-1.28912091e+00 7.65325129e-01 3.02663088e-01 -6.26001000e-01
4.85676140e-01 -7.52086043e-02 -5.53172648e-01 -6.52280673e-02
-4.10083026e-01 7.17080116e-01 1.03944767e+00 -5.72905719e-01
-3.43426555e-01 -3.89705032e-01 4.56414938e-01 3.19796056e-01
-4.86901641e-01 2.25589663e-01 5.18955588e-02 -7.10115016e-01
5.15183330e-01 -1.39255953e+00 -4.15097952e-01 -7.23572373e-02
-3.49294037e-01 -1.22048661e-01 1.05992997e+00 1.77141279e-02
3.18935066e-01 -2.19903469e+00 3.98722053e-01 2.99307019e-01
4.68147427e-01 2.28430524e-01 -4.05513436e-01 -9.60191563e-02
-7.80862153e-01 5.53338602e-02 3.47575217e-01 1.89554498e-01
8.59021395e-02 3.51251513e-01 -6.12552106e-01 8.86673987e-01
1.65087536e-01 9.47264433e-01 -5.24506927e-01 1.42647743e-01
1.58184290e-01 1.10210419e+00 -8.30731869e-01 -4.15132880e-01
3.16787660e-01 2.14268733e-02 -4.40780073e-01 -8.81217478e-04
7.23640978e-01 -4.92607094e-02 1.03134941e-02 -5.80744147e-01
-3.84901494e-01 1.57026812e-01 -1.16508353e+00 1.48235345e+00
-1.65767819e-01 1.02901006e+00 2.43415684e-01 -9.19285297e-01
7.13843346e-01 3.31591398e-01 2.52856821e-01 -3.47823165e-02
2.60822266e-01 5.80559596e-02 4.59406316e-01 1.78819269e-01
6.23526692e-01 2.10647807e-02 6.63811713e-02 1.29782647e-01
5.69880962e-01 1.97990879e-01 -2.18062803e-01 2.61790395e-01
9.99304473e-01 2.80566439e-02 -5.77194244e-03 -9.84449089e-01
4.79740381e-01 -4.74284351e-01 2.49992937e-01 7.46883392e-01
-5.62402979e-02 8.67633462e-01 3.63905281e-01 -1.05277717e+00
-1.37592351e+00 -1.41092682e+00 -4.51989591e-01 1.17295337e+00
-2.12924480e-01 -1.10621773e-01 -9.80893552e-01 -2.54193485e-01
-2.38524526e-01 2.73190945e-01 -9.35903788e-01 -2.76326686e-01
-8.22746038e-01 -8.99061263e-01 8.50443959e-01 6.44715905e-01
7.99877346e-01 -8.17733824e-01 -2.56522268e-01 3.95192355e-02
4.86898720e-01 -1.18367159e+00 -5.03549576e-01 4.75723267e-01
-1.00724542e+00 -9.52555597e-01 -6.70569777e-01 -9.25833285e-01
9.85188305e-01 6.34719849e-01 5.35467267e-01 -4.87740517e-01
-3.92125309e-01 7.20728636e-01 -1.75384656e-02 -4.16564792e-01
-1.24974228e-01 1.90738678e-01 6.63459003e-01 3.15420866e-01
1.23017870e-01 -8.32459092e-01 -5.57545424e-01 8.09403121e-01
-9.34571743e-01 -1.21760562e-01 2.33653858e-01 6.20506883e-01
5.11270881e-01 -5.16567111e-01 2.79630840e-01 -9.39020157e-01
3.85023117e-01 -3.56273577e-02 -6.79630578e-01 9.04108658e-02
-1.82306677e-01 4.26453650e-01 5.54349124e-01 -4.96006519e-01
-1.06169152e+00 2.73753345e-01 8.23630244e-02 -5.97756386e-01
4.71772850e-02 1.09945436e-03 -1.66021392e-01 -6.11726582e-01
1.04627109e+00 -1.90266341e-01 -9.79528651e-02 -3.07975471e-01
7.11015821e-01 2.30651841e-01 8.71248841e-01 -5.35390794e-01
1.09211910e+00 9.15064871e-01 4.82157826e-01 -1.28931355e+00
-4.74366874e-01 -6.40506744e-02 -1.02063847e+00 1.52329460e-01
9.82306123e-01 -1.11464155e+00 -4.80808705e-01 5.36322355e-01
-1.09333825e+00 -2.54914075e-01 -2.87144631e-01 7.23998427e-01
-7.23532736e-01 4.70668031e-03 -5.73087156e-01 -2.51781121e-02
-1.94723278e-01 -1.29296649e+00 4.12638307e-01 9.25697088e-02
-3.05575524e-02 -1.29640162e+00 -1.00389384e-01 -2.77719349e-01
8.85338247e-01 7.65988156e-02 8.73728156e-01 -7.62558222e-01
-7.92517066e-01 -2.55832046e-01 -3.72951716e-01 3.70998621e-01
-8.58429894e-02 -8.58784914e-02 -1.09295499e+00 -2.89298743e-01
-2.61127919e-01 1.64172441e-01 1.07324910e+00 4.61723566e-01
1.06621349e+00 -2.79789358e-01 1.79043069e-01 1.26753402e+00
1.07155180e+00 -2.77810872e-01 7.41161942e-01 2.86673963e-01
1.09248233e+00 1.76460996e-01 -2.91408241e-01 -7.03008249e-02
9.65200141e-02 7.21544981e-01 7.30889380e-01 -1.30566195e-01
-2.89641500e-01 -1.06195277e-02 4.08356488e-01 1.01034844e+00
-9.60814416e-01 2.93834925e-01 -3.68617326e-01 3.79019290e-01
-1.82471907e+00 -8.96031618e-01 -3.76235723e-01 2.02437520e+00
3.27740937e-01 -2.20798939e-01 -1.47537038e-01 -4.18369681e-01
8.34880114e-01 3.78008336e-01 -2.61784524e-01 -6.50261164e-01
-4.26545620e-01 3.62987041e-01 1.15725803e+00 5.38337231e-01
-1.46783304e+00 1.18860221e+00 7.22693634e+00 3.69675428e-01
-1.20770693e+00 9.85407755e-02 1.46596208e-01 -2.63343491e-02
-2.96178728e-01 -1.81163885e-02 -9.48305964e-01 -5.08841455e-01
8.22018862e-01 -2.31634125e-01 6.53654873e-01 9.47530806e-01
3.63742560e-02 9.54732537e-01 -1.27153671e+00 8.38415325e-01
-8.28452781e-02 -1.57489645e+00 4.07342672e-01 2.71621644e-01
9.29808140e-01 7.42123306e-01 4.55739707e-01 6.87400624e-02
6.48150504e-01 -1.17684281e+00 6.11476898e-01 4.00802553e-01
8.49769831e-01 -8.96949708e-01 2.70847738e-01 -5.18971741e-01
-1.17161381e+00 1.36219725e-01 -1.32160151e+00 -2.67903153e-02
-3.13492864e-01 -4.35575396e-02 -6.78221345e-01 2.11736843e-01
7.00979829e-01 8.99461746e-01 -3.32709432e-01 6.43816590e-01
-3.48744690e-02 3.47528696e-01 -2.33538449e-01 1.83135286e-01
6.61279440e-01 -3.03774625e-01 7.93471336e-01 1.38030922e+00
1.79325342e-01 1.13647275e-01 -3.73853356e-01 8.60114276e-01
-4.45757300e-01 -4.45243111e-03 -9.31100190e-01 2.94326544e-01
-5.32046296e-02 1.69203556e+00 -6.63158596e-01 6.59587830e-02
-5.88045359e-01 5.78758121e-01 4.44538593e-01 6.34597063e-01
-6.29937172e-01 -6.08281732e-01 1.41939378e+00 8.74390453e-02
5.60438514e-01 -5.46914756e-01 1.79547533e-01 -1.40892661e+00
-4.11972046e-01 -6.02165282e-01 1.39638245e-01 -7.67391801e-01
-8.09892654e-01 6.68732285e-01 -2.55108308e-02 -9.35321569e-01
8.95705670e-02 -1.31786311e+00 -5.78370988e-01 7.89398551e-01
-1.36034656e+00 -1.13135958e+00 -2.37890616e-01 1.25033128e+00
3.43447894e-01 -6.69301212e-01 1.06515121e+00 2.73776352e-01
-2.18191445e-01 6.79561257e-01 2.15867683e-01 6.44892454e-01
6.44713342e-01 -1.18361843e+00 7.22865164e-01 9.41519499e-01
3.06489527e-01 1.01170087e+00 5.03173053e-01 9.34712067e-02
-1.78804886e+00 -1.62233210e+00 3.38925391e-01 -5.39564848e-01
9.86000597e-01 -4.24771816e-01 -7.39665866e-01 1.54599023e+00
8.44189674e-02 6.31840885e-01 1.32432163e-01 -2.98226718e-02
-1.10300767e+00 -1.68311641e-01 -6.62586510e-01 1.15542412e+00
1.38016176e+00 -6.17430508e-01 -3.64210457e-01 2.76599795e-01
8.50045621e-01 -4.28448468e-01 -8.37837100e-01 2.99072057e-01
6.67735875e-01 -6.39990866e-01 1.30312753e+00 -1.30600870e+00
-6.29705265e-02 -2.74610519e-01 -5.21157444e-01 -1.32793784e+00
-7.90785372e-01 -1.03122103e+00 5.15471280e-01 3.95920724e-01
3.43908459e-01 -9.60681021e-01 4.24055368e-01 5.55226982e-01
-5.56439102e-01 -4.86517180e-04 -1.00372863e+00 -7.35272467e-01
3.59958172e-01 -2.96206146e-01 2.82566905e-01 1.21073806e+00
-2.36590281e-01 1.12480238e-01 -2.60242015e-01 3.55520725e-01
5.82198799e-01 -4.51221377e-01 6.41532958e-01 -1.26951396e+00
-8.69989097e-02 -5.72102308e-01 -1.03290987e+00 -8.70921552e-01
3.17915738e-01 -1.52779043e+00 -3.76676291e-01 -7.47514009e-01
6.96181729e-02 -1.86036572e-01 -4.49281693e-01 6.50846958e-01
5.16963303e-01 4.70637590e-01 6.97005615e-02 8.45838487e-02
-5.54942906e-01 1.68241039e-01 1.24679351e+00 1.13228723e-01
-2.11951206e-03 -7.04845935e-02 -6.79192364e-01 1.14290035e+00
8.81018579e-01 4.91818897e-02 -5.36053598e-01 -5.98328769e-01
4.32115734e-01 -6.38122439e-01 6.53557956e-01 -9.89934444e-01
1.34575710e-01 8.26667398e-02 3.50948483e-01 7.97748566e-02
1.63335562e-01 -6.46366417e-01 1.63846269e-01 2.66641639e-02
-6.13919556e-01 3.25836539e-01 4.43292707e-02 5.60564399e-01
7.54303038e-02 9.36123133e-02 1.16243768e+00 -2.34101161e-01
-7.75723875e-01 6.64671123e-01 -2.67317057e-01 3.61765781e-03
5.74795425e-01 -5.82700856e-02 -5.27384520e-01 -2.72113293e-01
-1.23267496e+00 -4.20315444e-01 2.39371091e-01 6.33619785e-01
6.15724385e-01 -1.49462867e+00 -7.80590773e-01 4.56816405e-01
-5.75236194e-02 -4.22615036e-02 -1.96205769e-02 7.77043998e-01
-8.10105383e-01 6.31604493e-01 -7.80442119e-01 -5.04615366e-01
-1.34336948e+00 6.19977750e-02 9.50882792e-01 3.60601842e-01
-1.04586363e+00 6.24386072e-01 7.57753551e-01 -5.78906298e-01
8.72948393e-02 -2.66438574e-01 -1.74834743e-01 -3.44907761e-01
5.03348291e-01 3.33787441e-01 1.81862056e-01 -9.72028553e-01
-1.35631591e-01 5.20571113e-01 -2.95179009e-01 -6.05206750e-02
1.76235127e+00 2.63299525e-01 -4.35490161e-01 -8.76469240e-02
1.36747015e+00 -2.30646029e-01 -1.25439048e+00 -3.91614169e-01
-3.92125309e-01 4.06734198e-02 1.15700737e-01 8.00795779e-02
-1.27027905e+00 8.57945383e-01 4.10985678e-01 -8.69036838e-02
5.08498847e-01 4.31507640e-02 -3.51517722e-02 1.12282979e+00
2.15035349e-01 -7.15058684e-01 -1.73552856e-01 1.17644441e+00
1.23666871e+00 -5.13749480e-01 -2.18970954e-01 -3.34492326e-01
-1.74615636e-01 1.54843652e+00 1.84718817e-01 -8.87598336e-01
1.00649261e+00 6.75166398e-02 -5.72203249e-02 -1.86663657e-01
-4.64067310e-01 1.71218887e-01 4.29347754e-01 7.01380074e-01
4.56501842e-01 2.89031923e-01 7.45453164e-02 8.42710584e-02
-5.34516871e-01 -6.46923244e-01 8.50262880e-01 3.07429284e-01
-4.97874886e-01 -8.23893368e-01 -3.07977527e-01 1.54528961e-01
-5.57746708e-01 -1.21377543e-01 -3.16474140e-01 1.18030500e+00
-4.30656314e-01 1.89527526e-01 2.27090150e-01 -2.38899067e-01
3.59145910e-01 5.53275719e-02 6.81379557e-01 -4.26252276e-01
-2.13999435e-01 -4.85505462e-02 -3.96779031e-02 -5.27386308e-01
-4.16565239e-01 -7.82037199e-01 -1.31356263e+00 -4.50870484e-01
5.21832779e-02 -2.98336327e-01 8.81653488e-01 5.47311544e-01
3.33511502e-01 6.26055241e-01 4.66012299e-01 -1.10590589e+00
-9.92739618e-01 -8.00247252e-01 -7.73166299e-01 2.33739570e-01
3.93624544e-01 -4.56240386e-01 -3.86442661e-01 8.27797353e-02] | [8.886938095092773, 2.4075276851654053] |
84f6a80a-6fc4-4132-9744-9b16f76f9f87 | i-spy-a-metaphor-large-language-models-and | 2305.14724 | null | https://arxiv.org/abs/2305.14724v1 | https://arxiv.org/pdf/2305.14724v1.pdf | I Spy a Metaphor: Large Language Models and Diffusion Models Co-Create Visual Metaphors | Visual metaphors are powerful rhetorical devices used to persuade or communicate creative ideas through images. Similar to linguistic metaphors, they convey meaning implicitly through symbolism and juxtaposition of the symbols. We propose a new task of generating visual metaphors from linguistic metaphors. This is a challenging task for diffusion-based text-to-image models, such as DALL$\cdot$E 2, since it requires the ability to model implicit meaning and compositionality. We propose to solve the task through the collaboration between Large Language Models (LLMs) and Diffusion Models: Instruct GPT-3 (davinci-002) with Chain-of-Thought prompting generates text that represents a visual elaboration of the linguistic metaphor containing the implicit meaning and relevant objects, which is then used as input to the diffusion-based text-to-image models.Using a human-AI collaboration framework, where humans interact both with the LLM and the top-performing diffusion model, we create a high-quality dataset containing 6,476 visual metaphors for 1,540 linguistic metaphors and their associated visual elaborations. Evaluation by professional illustrators shows the promise of LLM-Diffusion Model collaboration for this task.To evaluate the utility of our Human-AI collaboration framework and the quality of our dataset, we perform both an intrinsic human-based evaluation and an extrinsic evaluation using visual entailment as a downstream task. | ['Smaranda Muresan', 'Marianna Apidianaki', 'Yue Yang', 'Artemis Panagopoulou', 'Olivia Winn', 'Arkadiy Saakyan', 'Tuhin Chakrabarty'] | 2023-05-24 | null | null | null | null | ['visual-entailment'] | ['reasoning'] | [ 8.07779208e-02 2.10927248e-01 1.08526357e-01 -8.90273675e-02
1.91887151e-02 -9.13758397e-01 1.51555550e+00 1.08241692e-01
-2.35315338e-01 1.95739701e-01 5.80181420e-01 -9.90972102e-01
1.09460227e-01 -8.52123141e-01 -4.27173823e-01 -1.73399031e-01
1.18230879e-01 6.05875075e-01 -2.67492592e-01 -5.03089786e-01
5.24114490e-01 3.87313485e-01 -1.11042047e+00 7.57206500e-01
7.83842087e-01 8.05966437e-01 3.69296789e-01 6.61325336e-01
-9.82073009e-01 1.20019984e+00 -9.84772682e-01 -4.56205547e-01
6.39041290e-02 -6.67446613e-01 -7.96544373e-01 -1.10427007e-01
3.50080073e-01 7.62229413e-02 4.12276909e-02 8.17943633e-01
3.17198157e-01 2.79340195e-03 1.08173418e+00 -1.23914433e+00
-1.65495181e+00 8.09642911e-01 -6.91458106e-01 1.60289466e-01
6.03381217e-01 6.38386667e-01 9.05477643e-01 -1.18687713e+00
1.16297424e+00 2.10516453e+00 1.84219182e-01 4.79272872e-01
-1.72993934e+00 -6.11149430e-01 1.65336877e-01 9.03267115e-02
-1.16177320e+00 -1.12678185e-01 9.52327728e-01 -8.71495247e-01
1.10084808e+00 4.34382647e-01 1.02758646e+00 1.42283583e+00
-1.66999415e-01 5.30789554e-01 1.58274472e+00 -6.44940257e-01
1.68862626e-01 3.76284063e-01 -2.79763937e-01 8.72934282e-01
-2.59094805e-01 2.25124747e-01 -7.46715307e-01 -1.63763851e-01
1.00553203e+00 -2.25368306e-01 1.80655066e-02 2.96902154e-02
-1.48817515e+00 1.16745043e+00 9.96048868e-01 5.46598732e-01
-4.50315863e-01 4.33323890e-01 2.52373546e-01 3.92756492e-01
6.11306608e-01 9.13175583e-01 4.89326179e-01 -2.48966087e-02
-6.74642920e-01 4.34528589e-01 5.78562975e-01 7.66966343e-01
3.63746732e-01 2.50617024e-02 -4.77638960e-01 8.37367296e-01
3.87548715e-01 7.43000150e-01 3.53505790e-01 -8.79977047e-01
1.70703322e-01 9.09876108e-01 1.11784898e-01 -1.41673803e+00
-1.06826797e-01 1.20986000e-01 -7.27339804e-01 6.02480710e-01
1.82175577e-01 1.50340095e-01 -8.55533361e-01 1.76148486e+00
1.76383570e-01 -2.63792425e-01 -1.90665692e-01 1.15548861e+00
1.12122965e+00 8.45922828e-01 5.72995961e-01 7.28396848e-02
1.77330470e+00 -9.41321552e-01 -6.78410292e-01 -3.31959695e-01
7.19386339e-01 -8.98395061e-01 1.99313867e+00 6.67591467e-02
-1.01214564e+00 -6.54502988e-01 -8.10225487e-01 -5.91200173e-01
-6.94086373e-01 -5.42553961e-02 4.66728091e-01 1.30621880e-01
-1.14780259e+00 8.56361538e-03 -9.95351374e-02 -3.98857892e-01
5.29980540e-01 -5.15008748e-01 -1.27949953e-01 2.21218839e-01
-1.14616072e+00 1.11867177e+00 1.18998438e-01 -2.58403480e-01
-8.21428478e-01 -7.34302104e-01 -8.87816310e-01 -1.00337103e-01
6.00547269e-02 -1.27805388e+00 9.33508098e-01 -1.22302151e+00
-1.16991508e+00 1.22419941e+00 1.79410782e-02 -5.49762249e-01
7.60122955e-01 -1.11830778e-01 -2.78850615e-01 2.25384936e-01
4.86066341e-01 1.28530347e+00 1.08483303e+00 -1.58221734e+00
-2.11222395e-01 1.13930367e-01 3.92073780e-01 1.62390947e-01
-3.92865032e-01 6.66367449e-03 -1.24545246e-01 -1.24558055e+00
-4.74284768e-01 -6.55566812e-01 6.69438615e-02 5.08784711e-01
-5.41175961e-01 -4.04014468e-01 1.05802238e+00 -5.15731692e-01
1.32668889e+00 -2.35049510e+00 3.93333018e-01 1.85460374e-01
6.10582232e-01 -2.45932341e-01 -3.81741017e-01 6.79167330e-01
-2.84443460e-02 4.73272085e-01 -1.98310256e-01 -6.08909011e-01
3.49013120e-01 1.84150293e-01 -8.23790789e-01 -1.41587079e-01
1.01797253e-01 1.43809152e+00 -1.08258104e+00 -6.63172722e-01
3.63752007e-01 6.06315374e-01 -2.21273467e-01 1.96774855e-01
-5.86597919e-01 3.32988441e-01 -3.01541630e-02 2.76135653e-01
1.49088845e-01 -5.43644071e-01 9.52627063e-02 -2.13110551e-01
-1.43506557e-01 1.01879172e-01 -4.10217375e-01 1.89759779e+00
-8.63866985e-01 1.07772732e+00 -3.36012691e-01 -5.08033216e-01
1.03645086e+00 -4.59930813e-03 -2.32479379e-01 -1.16609120e+00
-9.62719023e-02 2.46239882e-02 -9.57044065e-02 -5.42429268e-01
5.18186629e-01 -6.90850198e-01 -3.17162395e-01 9.24334586e-01
-4.97581393e-01 -5.79095304e-01 2.31319472e-01 7.71942377e-01
7.83699811e-01 1.90303475e-02 1.41986996e-01 -5.35802484e-01
1.13208584e-01 3.24187666e-01 -4.32321876e-01 5.91633677e-01
5.02547085e-01 1.59741521e-01 5.76874018e-01 -7.12812245e-01
-1.06202853e+00 -1.30894935e+00 2.84395009e-01 1.00763130e+00
1.71814099e-01 -6.80653632e-01 -6.32356644e-01 -4.52647954e-01
1.01443045e-01 1.59381402e+00 -1.05689645e+00 -7.32866004e-02
-2.96594530e-01 -1.28286153e-01 6.05868638e-01 3.34540904e-01
2.08328277e-01 -1.51428783e+00 -1.35784221e+00 -3.33329849e-02
-2.28801295e-01 -9.24106061e-01 -3.63372117e-01 -4.17552561e-01
-4.49832231e-01 -8.15724015e-01 -8.16722095e-01 -5.32129407e-01
7.23672330e-01 2.45149687e-01 1.50932956e+00 3.76484305e-01
-5.23318410e-01 3.75702828e-01 -1.94111601e-01 -5.08287191e-01
-9.09772217e-01 -5.61156988e-01 -2.56404072e-01 -2.88721383e-01
1.57448366e-01 -4.63749439e-01 -6.32575631e-01 6.45318106e-02
-1.09769964e+00 1.01931393e+00 3.28708649e-01 6.91741765e-01
2.43154585e-01 -4.84813184e-01 1.29205003e-01 -6.80412829e-01
1.46531641e+00 -3.15508187e-01 -2.74624258e-01 1.95816681e-01
-4.78018433e-01 1.25883222e-01 4.23498511e-01 -7.90289462e-01
-9.68585253e-01 -6.46650612e-01 5.17902732e-01 -6.64851427e-01
1.59956694e-01 6.42214119e-01 3.62904042e-01 3.90366286e-01
1.11504805e+00 -3.50493123e-03 1.16617739e-01 -4.38034654e-01
1.37486136e+00 4.33947831e-01 5.19780278e-01 -8.54531705e-01
6.53646290e-01 6.34594798e-01 -2.85523757e-02 -9.22939420e-01
-3.59255910e-01 2.41057456e-01 -1.00225076e-01 -2.93112785e-01
9.45036232e-01 -6.24476969e-01 -7.67302632e-01 -2.00217729e-03
-1.66251087e+00 -6.31381452e-01 -7.43903935e-01 -2.32373849e-01
-5.81903458e-01 2.57938087e-01 -4.92529005e-01 -7.63831615e-01
-5.67579031e-01 -9.76655722e-01 1.09251118e+00 -1.57779843e-01
-1.04065299e+00 -1.27605677e+00 -1.99406669e-01 1.00528084e-01
5.90846777e-01 6.07375383e-01 1.68198657e+00 -1.02997027e-01
-2.13189125e-01 3.19116563e-01 -6.52521133e-01 -1.17085174e-01
2.29136378e-01 -4.40015011e-02 -6.85924411e-01 1.84731275e-01
-1.94117635e-01 -4.24680889e-01 5.93403459e-01 1.70298517e-01
9.33541775e-01 -6.57907069e-01 -4.18780237e-01 2.58904159e-01
1.24161673e+00 2.19344303e-01 3.83809507e-01 2.98113376e-01
7.42206693e-01 8.04854870e-01 2.76109397e-01 3.92693222e-01
4.52336460e-01 6.05417013e-01 1.19384423e-01 -5.31058729e-01
-4.43817466e-01 -6.37071729e-01 2.45624736e-01 5.15902460e-01
4.75769714e-02 -2.81189293e-01 -1.32727051e+00 4.86640066e-01
-1.82594025e+00 -8.69336367e-01 -9.20378566e-02 1.54014587e+00
8.09298217e-01 1.78890407e-01 1.07129425e-01 -2.48924226e-01
2.28016123e-01 3.42059165e-01 -2.52533853e-01 -8.21177721e-01
-3.98594409e-01 9.11303535e-02 -3.58164966e-01 7.41160750e-01
-3.55475396e-01 1.17436600e+00 6.24923277e+00 1.06339264e+00
-1.14738929e+00 1.79854244e-01 7.32131183e-01 -3.31714064e-01
-1.11843646e+00 1.96839809e-01 1.32066637e-01 3.96963745e-01
5.56518555e-01 -2.39968404e-01 6.79427564e-01 4.57236737e-01
4.29473311e-01 -2.34646231e-01 -1.15514266e+00 1.24607491e+00
1.11230657e-01 -1.91627371e+00 9.27461743e-01 -3.65005769e-02
4.51132238e-01 -7.19611347e-01 4.06078875e-01 2.38089766e-02
4.09240246e-01 -1.45154846e+00 1.32936370e+00 7.32222795e-01
1.13582599e+00 -3.24492395e-01 -1.95281908e-01 2.57625103e-01
-1.12049878e+00 -2.50173882e-02 4.21014391e-02 -3.98661822e-01
5.82880199e-01 3.13190132e-01 -8.82397830e-01 1.78558066e-01
4.20145780e-01 6.80773973e-01 -6.89267159e-01 -6.72826543e-02
-6.12902641e-01 2.48215884e-01 -1.71255544e-01 -2.62995064e-01
3.79762322e-01 -1.34951845e-01 7.28484571e-01 1.55361938e+00
3.26560408e-01 -6.83528720e-04 1.97681673e-02 1.91561353e+00
-6.53554406e-03 1.57758445e-01 -1.01006663e+00 -7.30531871e-01
5.53640425e-01 1.12950206e+00 -8.42580020e-01 -7.89318264e-01
-2.56408378e-02 1.18328965e+00 3.88083130e-01 5.93098044e-01
-7.38299847e-01 1.32716564e-03 4.47315902e-01 4.42938477e-01
-3.40509772e-01 -3.39791030e-01 -6.61055923e-01 -6.56338990e-01
-2.89822400e-01 -9.42368686e-01 1.82991311e-01 -1.65044975e+00
-1.58635008e+00 6.08181536e-01 3.38568568e-01 -7.52873540e-01
-3.07984829e-01 -6.60278022e-01 -7.64060616e-01 1.25175107e+00
-8.93451214e-01 -1.58055925e+00 -2.35101819e-01 3.47193688e-01
7.55762815e-01 -4.66373414e-02 8.54387224e-01 -2.23657370e-01
2.08465219e-01 4.30510677e-02 -7.71405518e-01 -1.16204098e-01
5.18961430e-01 -1.28136742e+00 8.78761947e-01 4.32032913e-01
5.18875003e-01 9.20975864e-01 1.04488003e+00 -7.41799235e-01
-1.17270541e+00 -5.07814586e-01 7.89394200e-01 -8.55820000e-01
9.74014223e-01 -5.34999251e-01 -6.52819991e-01 3.37267667e-01
7.79030621e-01 -5.43237925e-01 6.19776368e-01 -1.83087438e-01
-6.76485598e-01 7.41744161e-01 -7.00998068e-01 1.20083976e+00
1.43720901e+00 -8.87169361e-01 -1.01093483e+00 2.94795781e-01
1.05233967e+00 -1.41376510e-01 -4.99808013e-01 -3.30597878e-01
4.14034367e-01 -7.42507339e-01 1.21092427e+00 -7.05551863e-01
1.02953303e+00 -2.86671579e-01 -7.82588273e-02 -1.47536993e+00
-2.36950308e-01 -7.44207144e-01 -1.32239051e-02 1.03066027e+00
3.74215811e-01 -4.01624203e-01 -1.56412005e-01 4.94053572e-01
2.68721908e-01 -4.17053521e-01 -8.07518303e-01 -5.62630832e-01
1.23471484e-01 -4.93163437e-01 6.26961887e-01 1.26232255e+00
2.67807126e-01 8.00599039e-01 -1.63331687e-01 -6.89629078e-01
2.23807707e-01 3.88119370e-01 8.90820146e-01 -1.01327515e+00
-2.25045964e-01 -1.10835385e+00 9.42417905e-02 -8.85609686e-01
2.00036481e-01 -1.08755243e+00 -4.22527432e-01 -1.90633023e+00
6.54370338e-02 -2.80290723e-01 1.70902908e-01 5.07670045e-01
1.47401929e-01 1.05257519e-01 7.39357829e-01 4.21146631e-01
-1.90051362e-01 5.24699152e-01 1.67487776e+00 -4.97535467e-01
-1.91874474e-01 -1.17196774e+00 -8.60940933e-01 7.71822989e-01
1.76873505e-01 -2.37677470e-01 -6.48886263e-01 -6.03369474e-01
6.95838928e-01 -1.57224059e-01 9.96646881e-01 -3.69650751e-01
-1.26737818e-01 -3.81813228e-01 3.22240978e-01 -2.53152382e-02
5.24697185e-01 -6.18682981e-01 3.04320812e-01 6.45189941e-01
-6.17860913e-01 6.19893014e-01 4.30098206e-01 3.88151586e-01
7.38840103e-02 2.78888643e-01 3.59267026e-01 -3.40462089e-01
-8.02825093e-01 -4.71046090e-01 -3.82940054e-01 9.69465300e-02
7.81759501e-01 -3.28781128e-01 -7.86739409e-01 -5.42079926e-01
-6.30381942e-01 -2.71112919e-02 7.28380978e-01 9.05621290e-01
8.97002935e-01 -1.54033709e+00 -6.53626621e-01 -8.54990503e-05
3.71558368e-01 -2.27209553e-01 5.72483763e-02 6.11282110e-01
-5.52679896e-01 5.43324165e-02 -3.72583866e-01 -5.26143730e-01
-1.04616416e+00 1.03721642e+00 3.81135605e-02 1.46300346e-01
-8.97156477e-01 9.31405783e-01 8.54235768e-01 4.05568518e-02
-3.78783166e-01 -3.92635018e-01 -5.61700724e-02 9.74591896e-02
5.56481659e-01 1.55203536e-01 -9.06207323e-01 -7.16721773e-01
-1.91675827e-01 5.00785351e-01 3.84424716e-01 -8.29751194e-01
8.35019886e-01 -1.15705997e-01 -3.79816800e-01 6.86672032e-01
8.61986339e-01 -1.46173835e-01 -8.02945435e-01 -5.33453934e-02
2.00255234e-02 -6.30284488e-01 -2.35616133e-01 -1.22908032e+00
-3.80459458e-01 1.38162351e+00 4.83106524e-01 4.00722086e-01
7.22145915e-01 2.94572562e-01 4.50212061e-01 -1.23372544e-02
9.12499987e-03 -8.72416556e-01 7.22368956e-01 2.03891262e-01
1.92142391e+00 -8.41854513e-01 -2.64978737e-01 -1.46702349e-01
-9.44775641e-01 1.02461767e+00 5.10189235e-01 1.65803805e-01
1.98405400e-01 1.72919124e-01 4.18517202e-01 -8.04720938e-01
-8.83666992e-01 1.16580896e-01 5.74862361e-01 5.61121106e-01
5.32350183e-01 4.29453343e-01 -4.23542589e-01 3.63687158e-01
-5.34334838e-01 -5.16786054e-02 -5.43572828e-02 7.78817058e-01
-1.28609940e-01 -7.59103239e-01 -5.18751681e-01 1.09813325e-02
3.85899693e-01 -4.33549762e-01 -9.74680781e-01 6.87837303e-01
4.22102064e-02 8.11267257e-01 2.79603511e-01 -2.40700170e-01
1.83000445e-01 9.75558832e-02 3.39135528e-01 -4.14690793e-01
-6.92816854e-01 -4.67464626e-02 2.39284456e-01 -3.71687144e-01
-3.57595891e-01 -3.55913378e-02 -1.45276475e+00 -5.00889957e-01
5.12073100e-01 -6.32943437e-02 5.52945793e-01 8.85255456e-01
5.78488648e-01 5.47518194e-01 -8.86262506e-02 -7.45089948e-01
-1.44330725e-01 -1.09488142e+00 -6.13243729e-02 1.10672510e+00
-3.34094130e-02 -7.46814668e-01 -1.07792422e-01 2.17174962e-01] | [10.989360809326172, 1.4034862518310547] |
cc888e99-6214-4031-8d0a-3621a5f3a212 | anoshift-a-distribution-shift-benchmark-for | 2206.15476 | null | https://arxiv.org/abs/2206.15476v4 | https://arxiv.org/pdf/2206.15476v4.pdf | AnoShift: A Distribution Shift Benchmark for Unsupervised Anomaly Detection | Analyzing the distribution shift of data is a growing research direction in nowadays Machine Learning (ML), leading to emerging new benchmarks that focus on providing a suitable scenario for studying the generalization properties of ML models. The existing benchmarks are focused on supervised learning, and to the best of our knowledge, there is none for unsupervised learning. Therefore, we introduce an unsupervised anomaly detection benchmark with data that shifts over time, built over Kyoto-2006+, a traffic dataset for network intrusion detection. This type of data meets the premise of shifting the input distribution: it covers a large time span ($10$ years), with naturally occurring changes over time (eg users modifying their behavior patterns, and software updates). We first highlight the non-stationary nature of the data, using a basic per-feature analysis, t-SNE, and an Optimal Transport approach for measuring the overall distribution distances between years. Next, we propose AnoShift, a protocol splitting the data in IID, NEAR, and FAR testing splits. We validate the performance degradation over time with diverse models, ranging from classical approaches to deep learning. Finally, we show that by acknowledging the distribution shift problem and properly addressing it, the performance can be improved compared to the classical training which assumes independent and identically distributed data (on average, by up to $3\%$ for our approach). Dataset and code are available at https://github.com/bit-ml/AnoShift/. | ['Marius Dragoi', 'Florin Brad', 'Andrei Manolache', 'Emanuela Haller', 'Elena Burceanu'] | 2022-06-30 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [-1.75702035e-01 -4.25142020e-01 -3.03347617e-01 -4.21811372e-01
-4.59827363e-01 -5.87365925e-01 5.81805110e-01 4.90073740e-01
-4.49849308e-01 8.04215014e-01 -3.76622021e-01 -6.66848898e-01
-5.35648525e-01 -8.66408110e-01 -8.87650967e-01 -8.47930610e-01
-6.53050303e-01 7.77793348e-01 2.67540932e-01 -2.27024689e-01
1.36254981e-01 7.48302639e-01 -1.51989675e+00 1.28670060e-03
5.85850120e-01 1.14815557e+00 -6.43517792e-01 4.89768058e-01
-1.37524039e-01 5.50228357e-01 -9.81912553e-01 -5.17708242e-01
3.98260415e-01 -1.88241318e-01 -5.65942764e-01 -1.61947727e-01
3.31399381e-01 -1.73757061e-01 -6.60986960e-01 9.42474067e-01
3.80033463e-01 -7.67728686e-02 6.41180396e-01 -1.90642762e+00
-1.71915099e-01 8.57555747e-01 -5.78603923e-01 8.69768500e-01
-4.19299351e-03 1.49180144e-01 8.35525095e-01 -2.06894979e-01
3.44628990e-01 9.34625566e-01 7.47148395e-01 3.22527856e-01
-1.34036517e+00 -7.31923044e-01 2.39090830e-01 7.21396148e-01
-1.28328836e+00 -9.18266699e-02 8.79213750e-01 -4.96095985e-01
7.33838201e-01 2.47190043e-01 2.51183361e-01 1.63263810e+00
2.16453254e-01 6.58627629e-01 8.62231016e-01 -2.66189873e-01
4.06403899e-01 1.13542318e-01 4.35131848e-01 7.46886507e-02
3.13486189e-01 1.10679246e-01 -1.90842196e-01 -1.87390298e-01
-4.76723053e-02 2.34103590e-01 -4.84739244e-02 -4.19763595e-01
-7.75442123e-01 6.27192676e-01 1.92409672e-03 5.37056565e-01
-3.08614969e-01 -5.15050534e-03 8.57501447e-01 8.58556390e-01
5.67397952e-01 1.94224492e-02 -1.02105093e+00 -5.53845108e-01
-9.02787685e-01 2.35272467e-01 9.58287835e-01 7.36109495e-01
8.27683389e-01 2.63659477e-01 -3.90239842e-02 6.16120458e-01
-1.88669950e-01 6.34193301e-01 5.93018651e-01 -5.75445175e-01
5.40512919e-01 3.94717455e-01 -2.32013091e-01 -9.17796910e-01
-6.15300119e-01 -6.42331958e-01 -1.03147042e+00 -4.61080559e-02
8.03467810e-01 -3.28636587e-01 -5.55034816e-01 1.90915442e+00
2.74563786e-02 3.39017272e-01 -2.98789203e-01 2.56597977e-02
-3.52899730e-02 2.66410559e-01 -9.42667648e-02 -3.06507796e-01
8.94758224e-01 -2.70708054e-01 -6.44617438e-01 1.83863416e-01
9.46199358e-01 -2.69414842e-01 9.77626324e-01 7.01748788e-01
-6.77782595e-01 -3.52434605e-01 -8.06354344e-01 7.69565821e-01
-9.46760893e-01 -4.13495213e-01 3.34180921e-01 1.01740324e+00
-8.80886257e-01 8.88496578e-01 -8.32523763e-01 -5.42106807e-01
5.48094809e-01 3.00110996e-01 -2.69291401e-01 1.98380481e-02
-1.39849269e+00 5.31713545e-01 4.14729625e-01 -2.47716904e-01
-6.73376322e-01 -9.93524969e-01 -5.28595507e-01 5.12324534e-02
4.64532226e-01 -1.45327136e-01 1.03558552e+00 -6.98287010e-01
-1.12390912e+00 6.51617408e-01 1.34016693e-01 -1.01061487e+00
6.70738637e-01 -1.11059904e-01 -9.83993113e-01 -2.38280952e-01
-1.26219913e-01 -1.73859164e-01 7.03637660e-01 -9.68484402e-01
-6.69205010e-01 -6.11246586e-01 -2.00165182e-01 -7.20018327e-01
-6.59865201e-01 -2.46582478e-02 -1.07444458e-01 -7.31231511e-01
-4.18280542e-01 -8.27016234e-01 -1.89169440e-02 -6.96495056e-01
-6.01929903e-01 -2.84695983e-01 1.19955385e+00 -5.32705128e-01
1.73319173e+00 -2.28256488e+00 -2.49815419e-01 5.82192540e-01
6.56237379e-02 9.22905579e-02 -2.60493755e-02 7.93756306e-01
-5.03835022e-01 1.86746672e-01 -6.13806903e-01 -5.68000615e-01
3.46328437e-01 3.75593334e-01 -5.09797692e-01 7.67309666e-01
-5.03824353e-02 6.02828801e-01 -5.59723318e-01 1.46956787e-01
1.85409963e-01 -1.50107797e-02 -4.78968799e-01 -8.12384412e-02
-1.90061465e-01 4.50333446e-01 -1.86049994e-02 7.85754323e-01
8.87051821e-01 -1.02184098e-02 -3.28846136e-03 8.36387500e-02
-4.68742587e-02 -4.84354161e-02 -1.07460070e+00 1.37250972e+00
-1.89747438e-01 6.41243100e-01 -2.48094127e-01 -1.69908929e+00
9.10454392e-01 -7.71139795e-03 9.57796216e-01 -1.02767432e+00
1.86547920e-01 4.67058778e-01 2.98561871e-01 -3.59933943e-01
6.00722171e-02 1.40808761e-01 -3.56136680e-01 6.09914660e-01
6.71174303e-02 3.09742451e-01 5.87502897e-01 -7.74810649e-03
1.58478820e+00 -4.91488993e-01 6.09476343e-02 -2.91354001e-01
5.73477685e-01 -3.82867336e-01 5.62479377e-01 8.31886411e-01
-3.99713248e-01 1.65386528e-01 9.57974374e-01 -5.13596892e-01
-9.96701658e-01 -1.35967648e+00 -4.57510889e-01 9.49366868e-01
-2.31151730e-01 -2.72064716e-01 -7.01718569e-01 -1.03673530e+00
4.28552240e-01 9.69933271e-01 -6.76560998e-01 -4.10629004e-01
-6.81178451e-01 -1.18191767e+00 9.01122868e-01 3.09584141e-01
3.76806945e-01 -1.04813230e+00 -2.04571128e-01 6.85323104e-02
1.27459690e-01 -1.25315750e+00 -5.77204265e-02 5.09058774e-01
-7.93404639e-01 -1.15018952e+00 -2.08155766e-01 -5.26277944e-02
1.74762234e-01 -3.22787911e-01 1.30946958e+00 -1.64461359e-01
-4.13647205e-01 5.86510599e-01 -2.94966102e-01 -5.40585637e-01
-3.83796632e-01 4.46680099e-01 4.39356595e-01 2.82508850e-01
7.38504052e-01 -1.13686478e+00 -4.98874158e-01 3.67108673e-01
-1.14518726e+00 -1.09587777e+00 5.53329170e-01 3.66566926e-01
1.32326096e-01 4.49482739e-01 7.90244579e-01 -9.79001462e-01
4.72967744e-01 -1.07959592e+00 -4.69018161e-01 -1.53829291e-01
-8.12428415e-01 -9.55399200e-02 8.89431238e-01 -3.73904109e-01
-4.42328244e-01 -6.95176363e-01 -1.52410030e-01 -5.06337762e-01
-7.65753686e-01 2.29171142e-01 -4.28144932e-01 3.51892799e-01
5.60658872e-01 3.37656796e-01 1.43173058e-02 -6.06227040e-01
1.32500201e-01 7.31923103e-01 4.86880302e-01 -8.26438606e-01
1.09862697e+00 5.44371367e-01 -7.06416145e-02 -8.90822232e-01
-4.79190171e-01 -4.80051994e-01 -6.97572827e-01 -6.30021021e-02
3.22672278e-01 -3.36263686e-01 -6.30119681e-01 8.42094839e-01
-7.11069703e-01 -5.32000184e-01 -7.43284047e-01 3.73016447e-01
-5.59212744e-01 6.07849658e-01 -5.31057715e-01 -6.64287031e-01
-1.29134674e-02 -7.77461648e-01 6.21937692e-01 -1.71236470e-01
-9.58542079e-02 -1.32244408e+00 3.70481879e-01 4.90159616e-02
6.40843332e-01 4.52401370e-01 1.27623940e+00 -1.52745056e+00
-2.16318712e-01 -3.87017727e-01 7.58182199e-04 7.20667720e-01
3.15873295e-01 1.28512874e-01 -9.51554954e-01 -5.35784304e-01
-9.93546918e-02 7.87342489e-02 6.29822314e-01 2.78586686e-01
1.74070358e+00 -1.62418813e-01 -2.92321205e-01 5.28624892e-01
1.22422957e+00 3.97008836e-01 7.90795386e-01 6.79608464e-01
3.98843497e-01 5.43465674e-01 2.29416549e-01 7.69296467e-01
2.22861066e-01 7.23924160e-01 6.38682425e-01 1.16282562e-02
1.92880660e-01 1.32919610e-01 5.14483809e-01 6.66757822e-01
2.15989530e-01 -4.46836799e-01 -1.10056686e+00 5.89167058e-01
-1.62625980e+00 -1.28392398e+00 -1.90463126e-01 2.63624430e+00
4.83258724e-01 5.60635805e-01 6.21284664e-01 7.22363293e-01
5.93523741e-01 1.40641436e-01 -7.51181662e-01 -4.10957515e-01
-2.79094994e-01 5.35578907e-01 6.33313775e-01 1.32226616e-01
-1.22535706e+00 5.07870257e-01 5.86003923e+00 1.08061194e+00
-1.27571678e+00 2.53654644e-02 7.00569212e-01 -7.28515610e-02
-9.11202282e-02 -2.52654284e-01 -5.56407034e-01 9.37944531e-01
1.59091365e+00 -2.05274343e-01 2.39524946e-01 6.56869590e-01
2.82295674e-01 2.61462748e-01 -1.16290748e+00 9.38102543e-01
1.24216564e-01 -8.34270179e-01 -5.40763363e-02 4.21460718e-01
3.77808571e-01 2.82508075e-01 2.38669738e-01 7.31698990e-01
9.16145816e-02 -7.98764825e-01 4.81222123e-01 5.99208355e-01
4.51869041e-01 -1.05776715e+00 9.61558402e-01 4.81814981e-01
-8.88355732e-01 -4.81036633e-01 1.01868577e-01 8.05541128e-02
1.16033433e-02 1.13686931e+00 -3.40200961e-01 9.03932273e-01
9.85853255e-01 8.64787519e-01 -8.07839394e-01 1.01526880e+00
3.27108920e-01 9.79594231e-01 -4.55598444e-01 3.89647096e-01
4.61593233e-02 -2.15276256e-01 7.19439626e-01 1.26213384e+00
5.32185733e-01 -7.77166367e-01 1.11159310e-01 5.07239759e-01
-1.48442134e-01 -3.03623956e-02 -7.26024866e-01 5.75202815e-02
5.35495281e-01 9.31644917e-01 -5.24611235e-01 -5.89220934e-02
-3.86674702e-01 6.84511721e-01 1.48990070e-02 4.68416661e-01
-1.07926488e+00 -5.16985595e-01 8.17503452e-01 3.76662761e-01
3.02795470e-01 -1.76542327e-01 -1.90672167e-02 -1.07382238e+00
2.50387102e-01 -9.53034282e-01 7.43826807e-01 6.43469915e-02
-1.80364823e+00 5.05191326e-01 2.72017092e-01 -1.32682800e+00
-2.77008981e-01 -7.51794815e-01 -6.66884124e-01 3.02069068e-01
-1.42219806e+00 -7.30014324e-01 -9.42927897e-02 7.68642962e-01
2.56117553e-01 -4.87612754e-01 5.37859619e-01 9.08869624e-01
-1.00456548e+00 1.01998615e+00 3.93564939e-01 2.16499105e-01
8.45646441e-01 -1.48829758e+00 3.95929515e-01 8.28706682e-01
1.16855979e-01 4.19694066e-01 6.70373559e-01 -3.03089738e-01
-1.01581061e+00 -1.16113317e+00 5.36527872e-01 -6.73785269e-01
1.17328978e+00 -3.60206664e-01 -1.12625444e+00 6.90575957e-01
1.28369749e-01 1.05037525e-01 9.33187246e-01 1.97364375e-01
-5.19797683e-01 -5.37163675e-01 -1.37357748e+00 3.36281955e-01
1.13143229e+00 -3.53350818e-01 -2.68200189e-01 1.88059822e-01
4.20417666e-01 4.97105010e-02 -9.25661087e-01 6.37575150e-01
2.01561958e-01 -1.38172114e+00 8.15193534e-01 -8.08616221e-01
-1.28219664e-01 -1.81816056e-01 -3.15363020e-01 -1.29424453e+00
1.94101520e-02 -6.75481379e-01 -5.18713772e-01 1.37408125e+00
4.02463436e-01 -1.05172372e+00 8.05370390e-01 6.21489063e-02
-1.46099523e-01 -7.07909226e-01 -1.32486033e+00 -1.14924514e+00
4.64192718e-01 -8.98132026e-01 8.27897727e-01 1.06459498e+00
-3.16070676e-01 -1.20009206e-01 -1.27144948e-01 1.98854119e-01
6.98926210e-01 -1.41523540e-01 8.87117505e-01 -1.49125695e+00
-2.46993765e-01 -8.79464030e-01 -8.08888972e-01 -6.63186133e-01
3.20608139e-01 -8.34980369e-01 -4.47870433e-01 -7.94281006e-01
-2.59463847e-01 -4.48823333e-01 -6.79114521e-01 3.07257384e-01
3.15779388e-01 -3.87610458e-02 -1.16722867e-01 1.80939481e-01
-6.04470849e-01 4.96755004e-01 3.72363776e-01 -2.45188802e-01
5.82550690e-02 4.58615273e-01 -3.69907051e-01 5.95054328e-01
1.11299157e+00 -5.39195657e-01 -2.81445682e-01 8.83440673e-02
6.81435689e-02 -5.42206526e-01 3.70661199e-01 -1.38664782e+00
-3.27433981e-02 -2.05555046e-03 -7.96850026e-03 -6.98127151e-01
-1.20113090e-01 -1.00285649e+00 -5.21967001e-02 4.72797245e-01
3.79400849e-02 4.61674184e-01 3.87859195e-01 6.94822788e-01
-3.67086202e-01 6.77338988e-02 6.19289398e-01 3.05365711e-01
-5.49436331e-01 7.95129001e-01 -3.65274638e-01 3.48867804e-01
1.22815633e+00 -6.35388047e-02 -4.20247346e-01 -2.75001794e-01
-7.99569070e-01 1.78365499e-01 2.84554124e-01 4.59466904e-01
7.63163343e-02 -1.28119218e+00 -5.88394701e-01 2.49193266e-01
2.77029783e-01 -3.72270495e-01 5.74123859e-01 1.12469816e+00
-2.84768105e-01 2.69923359e-01 -1.47852227e-01 -7.66676426e-01
-8.00218403e-01 7.93708086e-01 4.93901253e-01 -5.03276646e-01
-4.40310478e-01 2.12490052e-01 -3.80900979e-01 -6.57998800e-01
4.33003128e-01 -3.46054524e-01 -1.57891922e-02 3.88715148e-01
4.37560320e-01 7.33187497e-01 5.66615105e-01 -2.24118277e-01
-3.94189954e-01 3.89343530e-01 -1.37935936e-01 2.37505063e-01
1.31326866e+00 3.01894755e-03 -2.31984228e-01 1.00498962e+00
1.24168217e+00 1.90565109e-01 -8.35777879e-01 -3.20622206e-01
4.47501212e-01 -1.39756337e-01 -5.85646987e-01 -5.73147237e-01
-1.22950327e+00 9.46914852e-01 7.85841346e-01 9.82679009e-01
1.17018628e+00 -1.44470274e-01 7.80078650e-01 4.06869948e-01
3.64598781e-01 -9.26742852e-01 9.64752734e-02 7.24172473e-01
3.81570846e-01 -9.86785352e-01 -4.60476875e-01 7.27932602e-02
-8.63034651e-02 1.02919674e+00 5.75299323e-01 7.09662447e-04
1.15401769e+00 4.90145326e-01 -1.41940610e-02 -5.29781915e-02
-6.00247383e-01 -1.57860801e-01 -1.99276626e-01 6.46230698e-01
9.57266986e-02 8.26344565e-02 -4.93104346e-02 6.12534702e-01
-3.63750070e-01 -3.36554974e-01 4.69027609e-01 7.41121829e-01
-1.06137939e-01 -1.40888166e+00 -1.73696980e-01 6.12224460e-01
-5.46652734e-01 1.85191154e-01 -2.16754869e-01 1.06479740e+00
2.93252259e-01 7.72735417e-01 3.91002238e-01 -4.56782490e-01
7.15717137e-01 3.92544150e-01 6.95289746e-02 -2.33630523e-01
-4.14850682e-01 -5.24889708e-01 -1.54062659e-01 -6.13759995e-01
-5.74584268e-02 -8.28127384e-01 -8.85648787e-01 -7.31614113e-01
2.83960879e-01 1.96478546e-01 4.05573636e-01 1.02908373e+00
4.11456615e-01 6.72644019e-01 9.93808389e-01 -5.47865093e-01
-7.46798277e-01 -1.13002610e+00 -8.16533446e-01 7.08491027e-01
4.06275690e-01 -7.82109976e-01 -1.14511180e+00 -2.42237955e-01] | [7.490002155303955, 2.552647352218628] |
ebf7fca1-32d8-4638-807d-f3f56d776edc | prompt-learning-for-domain-adaptation-in-task | 2211.05596 | null | https://arxiv.org/abs/2211.05596v1 | https://arxiv.org/pdf/2211.05596v1.pdf | Prompt Learning for Domain Adaptation in Task-Oriented Dialogue | Conversation designers continue to face significant obstacles when creating production quality task-oriented dialogue systems. The complexity and cost involved in schema development and data collection is often a major barrier for such designers, limiting their ability to create natural, user-friendly experiences. We frame the classification of user intent as the generation of a canonical form, a lightweight semantic representation using natural language. We show that canonical forms offer a promising alternative to traditional methods for intent classification. By tuning soft prompts for a frozen large language model, we show that canonical forms generalize very well to new, unseen domains in a zero- or few-shot setting. The method is also sample-efficient, reducing the complexity and effort of developing new task-oriented dialogue domains. | ['Christopher Parisien', 'Makesh Narsimhan Sreedhar'] | 2022-11-10 | null | null | null | null | ['intent-classification', 'task-oriented-dialogue-systems'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.38099530e-01 4.89842296e-01 -1.77919984e-01 -6.89578295e-01
-8.96332622e-01 -9.01444435e-01 8.17504346e-01 1.48532718e-01
-4.41417485e-01 7.66945481e-01 6.73664033e-01 -2.24338949e-01
6.77177235e-02 -5.26251674e-01 1.17952801e-01 8.11681896e-02
1.61849722e-01 8.61750841e-01 1.53284818e-01 -7.93852031e-01
3.21026057e-01 3.28967534e-03 -1.58785903e+00 5.79544783e-01
9.13508832e-01 5.66408753e-01 3.23639363e-01 7.73489118e-01
-7.64195025e-01 7.17030525e-01 -9.00549412e-01 -5.25239050e-01
-1.00756682e-01 -5.62794566e-01 -1.37002945e+00 2.21986905e-01
1.49817243e-01 -4.50045556e-01 -1.89342368e-02 5.38533807e-01
3.49831104e-01 4.41860706e-01 4.35766160e-01 -1.33141494e+00
-1.91467240e-01 5.18249333e-01 2.63451576e-01 -1.57836482e-01
1.24030221e+00 7.28678629e-02 1.11775780e+00 -7.97307312e-01
9.68481719e-01 1.33892107e+00 7.33664930e-01 1.02043462e+00
-1.39223826e+00 -5.86224645e-02 2.46020600e-01 -2.67726988e-01
-8.48817110e-01 -1.00966990e+00 5.74473798e-01 -3.98751676e-01
1.30182552e+00 4.86386269e-01 5.64915240e-01 1.29110372e+00
-1.20646581e-01 8.80757511e-01 9.20991242e-01 -6.13474667e-01
4.95106786e-01 5.77861011e-01 4.30591524e-01 5.82614303e-01
-1.21663041e-01 -5.03029287e-01 -7.58632123e-01 -6.10452652e-01
4.73639250e-01 -2.95108527e-01 -1.24238245e-01 -4.86522257e-01
-8.14577460e-01 9.46642697e-01 -3.01570654e-01 2.26006091e-01
-1.65725857e-01 -1.54154599e-01 7.82057762e-01 6.00169718e-01
4.96529043e-01 1.10494518e+00 -4.78897035e-01 -1.11981463e+00
-3.03177476e-01 7.55152702e-01 1.64570904e+00 1.37113523e+00
3.60975921e-01 -2.34346747e-01 -6.01448044e-02 1.31511676e+00
1.77879222e-02 -1.16054090e-02 5.64114332e-01 -1.24574077e+00
2.85773039e-01 7.61781931e-01 4.97635096e-01 -4.78878856e-01
-5.70101678e-01 9.83921662e-02 -2.17584178e-01 -7.02573061e-02
5.49933970e-01 -4.17972773e-01 -4.04960215e-01 1.74257016e+00
3.10166687e-01 -9.08336103e-01 3.24421465e-01 4.59791481e-01
6.69581652e-01 5.51874399e-01 2.69806981e-01 -4.83119965e-01
1.59854317e+00 -9.18029785e-01 -9.25819814e-01 -5.66609800e-01
9.59831417e-01 -7.56713450e-01 1.64312577e+00 5.77594578e-01
-1.30655301e+00 -2.69962013e-01 -7.57532716e-01 -3.82374525e-01
-4.15811688e-01 -4.21027839e-01 1.08932602e+00 8.59009504e-01
-8.85592997e-01 4.03818697e-01 -2.00280368e-01 -7.06117749e-01
-2.03235168e-02 1.45576805e-01 -2.92868704e-01 -2.33158376e-02
-1.29167938e+00 1.05828071e+00 1.25868186e-01 -4.07584667e-01
-3.12024802e-01 -6.06312394e-01 -9.68224168e-01 -8.81082565e-02
6.43890977e-01 -6.91189945e-01 2.08351660e+00 -9.63439524e-01
-1.80362988e+00 6.68703973e-01 -1.87767312e-01 -2.02266231e-01
3.49914193e-01 -2.40448043e-01 -2.12751076e-01 -1.64702490e-01
1.54019117e-01 5.56935549e-01 5.52217722e-01 -9.96798515e-01
-7.72761166e-01 -3.11847609e-02 5.51149189e-01 7.40571737e-01
-5.02440095e-01 1.56097546e-01 -8.68846923e-02 -4.12809432e-01
-2.68283069e-01 -7.85904884e-01 -3.24270487e-01 7.19301449e-03
9.19395015e-02 -6.90121412e-01 6.19858205e-01 -3.87606382e-01
1.31134236e+00 -1.83019829e+00 6.71811849e-02 -2.39469111e-01
3.73910636e-01 2.07307756e-01 -1.80623941e-02 8.20134997e-01
3.60497653e-01 1.75639898e-01 9.87410843e-02 -2.25078374e-01
2.57691920e-01 2.96258926e-01 -1.70700952e-01 -3.33931059e-01
-3.94946374e-02 7.09899068e-01 -1.15339017e+00 -3.58757079e-01
1.25744581e-01 -1.01948224e-01 -7.08333135e-01 6.62303507e-01
-6.88590050e-01 -2.38800868e-01 -4.37731981e-01 2.18150109e-01
1.26319751e-01 -1.61471441e-01 5.13003230e-01 7.75502473e-02
-7.72991702e-02 9.68984008e-01 -9.88134027e-01 2.10254526e+00
-1.09610999e+00 4.88907725e-01 1.87596753e-01 -5.15764117e-01
9.13821459e-01 3.97794276e-01 1.40931278e-01 -6.18988633e-01
-8.67188573e-02 4.80318740e-02 -2.21261203e-01 -7.40714014e-01
9.88806844e-01 -6.29473150e-01 -7.72014081e-01 8.99975121e-01
2.37840518e-01 -5.66585720e-01 3.55047137e-01 3.95432025e-01
1.16201687e+00 -1.29601374e-01 5.78341722e-01 -2.49663875e-01
4.38849889e-02 4.03039217e-01 2.79222667e-01 7.23519087e-01
-1.04652278e-01 2.77878344e-01 7.18135536e-01 -5.97972870e-01
-1.04014480e+00 -7.54261911e-01 9.05423164e-02 1.70169675e+00
-1.19881734e-01 -8.54680121e-01 -8.52492750e-01 -8.28832805e-01
-1.48971573e-01 1.11224008e+00 -1.72617346e-01 -1.66923702e-01
-5.32535732e-01 -2.04198807e-01 6.75482213e-01 2.85826445e-01
4.06661004e-01 -9.19469237e-01 -7.40462840e-01 6.18056655e-01
-5.91516316e-01 -1.03422999e+00 -5.26221275e-01 2.98130929e-01
-7.43995607e-01 -7.47561634e-01 -3.82770598e-01 -7.29825497e-01
3.46208930e-01 3.49762946e-01 1.26917958e+00 -7.94173591e-03
-2.69089520e-01 5.16877472e-01 -4.70993131e-01 -2.27238685e-01
-7.74051249e-01 2.47407600e-01 9.54489708e-02 -5.54081559e-01
5.69204807e-01 -3.11023980e-01 -9.07561705e-02 3.37454498e-01
-5.35333276e-01 4.07719523e-01 5.25226817e-02 1.05594862e+00
-3.70966375e-01 -2.27679431e-01 8.10796022e-01 -1.29752469e+00
1.52149296e+00 -4.13647979e-01 -1.30161613e-01 1.83542386e-01
-7.33069181e-01 2.61137724e-01 8.17528188e-01 -3.76721799e-01
-1.52247882e+00 -5.80850877e-02 -2.87207842e-01 4.52309638e-01
-3.63517195e-01 3.32484156e-01 -4.14414331e-02 1.57051787e-01
1.09042192e+00 -1.68031752e-02 1.81846201e-01 -5.03718376e-01
6.32549405e-01 1.06025064e+00 2.09574386e-01 -9.19748664e-01
3.66824955e-01 -4.69236821e-02 -7.39175260e-01 -1.17475593e+00
-5.88747859e-01 -6.65948808e-01 -4.15949494e-01 -2.51680076e-01
5.07030904e-01 -6.92832410e-01 -6.50013208e-01 1.74140319e-01
-1.22829998e+00 -5.91213226e-01 -5.12715220e-01 -1.49238348e-01
-8.23110342e-01 4.42498326e-01 -5.52203715e-01 -1.01868856e+00
-3.03813130e-01 -7.86821127e-01 8.99399340e-01 1.92430943e-01
-1.28952801e+00 -1.14078116e+00 1.15204498e-01 6.90518439e-01
5.82280040e-01 -2.64837205e-01 1.17791569e+00 -9.20095682e-01
-3.18361744e-02 -1.10874899e-01 -3.01872045e-02 7.39592761e-02
2.77532995e-01 -4.08023745e-01 -9.41512465e-01 -2.77888030e-02
1.27439395e-01 -9.73922908e-01 7.12453350e-02 -3.45147461e-01
5.96965373e-01 -8.02823782e-01 -1.42565072e-01 -5.47005087e-02
8.78884196e-01 4.50357288e-01 3.23669463e-01 4.68924940e-02
2.23422229e-01 9.11226630e-01 8.37781966e-01 7.29026794e-01
5.62649012e-01 9.50707793e-01 -4.83482778e-01 6.76320866e-02
2.10698098e-02 -4.22288716e-01 2.83049285e-01 5.91579437e-01
2.02853188e-01 -3.21596682e-01 -1.01381874e+00 6.18792355e-01
-1.83226979e+00 -8.80683720e-01 2.69438237e-01 1.89707720e+00
1.28435814e+00 2.78178245e-01 4.18499678e-01 -1.69069961e-01
2.23453388e-01 2.66396731e-01 -3.36389333e-01 -9.45806324e-01
2.47890338e-01 2.10878834e-01 -9.00706053e-02 7.10275829e-01
-6.88753664e-01 1.12707233e+00 7.38220882e+00 5.56149840e-01
-6.12540364e-01 1.00018628e-01 2.82784969e-01 -1.27152160e-01
-6.39997184e-01 8.21941644e-02 -4.89916921e-01 1.54604167e-01
1.03039718e+00 -6.05507493e-01 5.77337742e-01 1.21537626e+00
1.30897895e-01 -3.34902674e-01 -1.37972236e+00 9.18378472e-01
1.43891796e-01 -1.35855258e+00 -1.40179932e-01 -2.12813526e-01
2.37303451e-01 -5.25884449e-01 -3.76375645e-01 6.02007687e-01
7.03480780e-01 -8.61104786e-01 4.19382662e-01 1.35273561e-01
8.76024842e-01 -4.88846749e-01 4.49647069e-01 5.40694892e-01
-8.33361983e-01 -3.74681763e-02 -2.12099105e-01 -5.27309299e-01
3.83201510e-01 4.66738977e-02 -1.25386894e+00 -3.14538479e-02
3.18640023e-01 8.54851604e-02 -2.40829036e-01 3.41898620e-01
2.21425131e-01 1.56470999e-01 -3.53747666e-01 -5.60682893e-01
1.38071969e-01 4.45188209e-02 5.84375560e-01 1.38743520e+00
-8.36878344e-02 4.15065676e-01 4.07195002e-01 5.97842216e-01
1.74513794e-02 8.08820352e-02 -1.02770638e+00 -3.80425364e-01
6.14190698e-01 1.22115636e+00 -4.42491949e-01 -4.31685805e-01
-3.19740534e-01 1.09126461e+00 4.07078654e-01 1.08762175e-01
-1.92079172e-01 -5.21168768e-01 8.57065082e-01 1.77062526e-01
-3.76723170e-01 -2.54927784e-01 -4.40411180e-01 -1.13505101e+00
-1.73026063e-02 -1.44820058e+00 3.68740797e-01 -6.26134396e-01
-1.26337039e+00 5.80751121e-01 1.26889750e-01 -8.24839652e-01
-8.80301833e-01 -4.32524025e-01 -4.11428869e-01 7.02113688e-01
-6.36637747e-01 -7.87394047e-01 -3.78008485e-01 2.74732143e-01
1.27674532e+00 -2.44013146e-01 1.45975387e+00 -7.58891255e-02
-2.39739388e-01 5.78499019e-01 -3.24858546e-01 -2.31895208e-01
8.71951520e-01 -1.36767721e+00 7.34643281e-01 3.15888375e-01
-1.73269242e-01 1.05160940e+00 1.00543845e+00 -5.09055197e-01
-1.61098170e+00 -5.21893680e-01 1.18603790e+00 -6.00764811e-01
5.33475280e-01 -9.55078423e-01 -8.68219733e-01 4.45688397e-01
3.14420193e-01 -8.45986426e-01 9.80287313e-01 7.78183222e-01
-3.60986620e-01 2.52607316e-01 -1.33047056e+00 9.10716891e-01
1.31988978e+00 -8.48782241e-01 -1.10309291e+00 5.46268344e-01
8.95189643e-01 -3.88723046e-01 -7.91063786e-01 -1.89244717e-01
7.27023065e-01 -8.54585052e-01 7.27580369e-01 -7.97959328e-01
3.01141918e-01 2.94656813e-01 1.14974096e-01 -1.29374886e+00
-2.40229368e-01 -1.26828456e+00 6.60272315e-02 1.09157300e+00
5.36171317e-01 -4.67368662e-01 8.30576360e-01 1.41347742e+00
-1.87633276e-01 -4.37333494e-01 -7.02992916e-01 -5.79119921e-01
-2.08478514e-02 -5.23509562e-01 4.73887682e-01 9.25751746e-01
1.13446844e+00 1.03908932e+00 -3.30636770e-01 -7.21079230e-01
2.63715118e-01 -8.94295350e-02 9.85997260e-01 -1.27481484e+00
-3.28117549e-01 -4.66637135e-01 4.50960584e-02 -1.35847437e+00
-4.28849552e-03 -4.41396654e-01 3.92865419e-01 -1.49387038e+00
-5.91939129e-02 -7.45607853e-01 4.52193439e-01 4.36201513e-01
3.53447311e-02 -3.32541555e-01 4.52345796e-02 2.07734890e-02
-7.93267250e-01 4.07966524e-01 8.27114046e-01 7.90429786e-02
-5.03917873e-01 1.29283294e-01 -9.82806623e-01 8.64972889e-01
7.55616486e-01 -1.58118233e-01 -9.60739553e-01 -1.85786441e-01
3.82865816e-01 4.34526294e-01 -1.90188557e-01 -7.81084418e-01
3.04100275e-01 -5.35430014e-01 -1.61717817e-01 -1.84356168e-01
4.85981822e-01 -5.66079855e-01 -6.46819398e-02 1.44795492e-01
-8.35015357e-01 3.53658348e-02 1.84429958e-01 2.50891685e-01
-6.46204278e-02 -6.04153395e-01 5.59753776e-01 -3.81152570e-01
-9.35725212e-01 -3.24168652e-01 -9.78848636e-01 5.31165898e-01
9.43890154e-01 -4.34985906e-01 -2.96780646e-01 -7.50919580e-01
-5.26750386e-01 2.51531512e-01 6.52577162e-01 7.46939600e-01
6.11649632e-01 -1.04637194e+00 -1.51448950e-01 2.25472480e-01
5.26901901e-01 -1.38855755e-01 1.73769996e-01 3.22886333e-02
-3.66082400e-01 4.66780066e-01 -2.74202734e-01 -2.41489723e-01
-1.46806133e+00 3.34010780e-01 2.47738600e-01 -2.59755880e-01
-5.97118258e-01 8.46729279e-01 1.14373401e-01 -6.49806082e-01
4.69057947e-01 -3.32771949e-02 -1.66834399e-01 3.10795277e-01
6.08788908e-01 3.16421002e-01 -6.41703084e-02 5.72685152e-02
-1.49159521e-01 -1.20617568e-01 -4.29740578e-01 -6.24476790e-01
1.03589439e+00 -2.59947896e-01 6.38610274e-02 7.67641783e-01
9.47506785e-01 -2.85357028e-01 -8.57669771e-01 -2.97480732e-01
3.76456141e-01 -4.70311493e-01 -2.64609873e-01 -9.14777160e-01
6.38801083e-02 5.67257464e-01 2.26158634e-01 6.80172801e-01
6.59633338e-01 -4.25174609e-02 9.59081471e-01 9.93292928e-01
6.59660697e-01 -1.68110037e+00 3.92344743e-01 8.82893562e-01
9.85012054e-01 -9.90700483e-01 -1.27760172e-01 -3.80555063e-01
-1.03348541e+00 1.20696914e+00 7.73922682e-01 5.38946331e-01
1.60530373e-01 4.57022280e-01 2.43104219e-01 -2.16414213e-01
-1.31310582e+00 4.58970368e-02 -2.35784099e-01 8.27213407e-01
6.95165455e-01 7.01282769e-02 -3.89159828e-01 6.87111855e-01
-3.48069787e-01 8.41520578e-02 8.12637687e-01 1.28403282e+00
-6.82690203e-01 -1.56159508e+00 1.15822069e-01 6.05770648e-01
-2.47026812e-02 -1.21390626e-01 -8.04785132e-01 6.42381728e-01
-3.87829095e-01 1.16049695e+00 -8.79553985e-03 -3.32837284e-01
6.61883056e-01 8.00588667e-01 3.40285748e-01 -1.32183504e+00
-8.67846847e-01 -2.49759674e-01 1.26912832e+00 -5.65881133e-01
-9.14025120e-04 -5.82291365e-01 -1.02482498e+00 -5.28106868e-01
-2.10478634e-01 4.33599740e-01 4.77423221e-01 8.92791927e-01
3.54992002e-01 4.96499427e-02 6.52141035e-01 -4.17745501e-01
-8.60916555e-01 -9.92406070e-01 -1.65550545e-01 6.26028597e-01
1.07743256e-01 -6.21683776e-01 -7.48997629e-02 8.33591148e-02] | [12.763280868530273, 7.956464767456055] |
5d31a879-02ef-450c-b30a-7763b1fdef54 | intersectional-bias-in-hate-speech-and | 2005.05921 | null | https://arxiv.org/abs/2005.05921v3 | https://arxiv.org/pdf/2005.05921v3.pdf | Intersectional Bias in Hate Speech and Abusive Language Datasets | Algorithms are widely applied to detect hate speech and abusive language in social media. We investigated whether the human-annotated data used to train these algorithms are biased. We utilized a publicly available annotated Twitter dataset (Founta et al. 2018) and classified the racial, gender, and party identification dimensions of 99,996 tweets. The results showed that African American tweets were up to 3.7 times more likely to be labeled as abusive, and African American male tweets were up to 77% more likely to be labeled as hateful compared to the others. These patterns were statistically significant and robust even when party identification was added as a control variable. This study provides the first systematic evidence on intersectional bias in datasets of hate speech and abusive language. | ['Sarah Nam', 'Jae Yeon Kim', 'Carlos Ortiz', 'Sarah Santiago', 'Vivek Datta'] | 2020-05-12 | null | null | null | null | ['abuse-detection'] | ['natural-language-processing'] | [-3.19737583e-01 1.88413247e-01 -4.39810932e-01 -2.34316170e-01
-2.72962630e-01 -9.77160037e-01 1.11365080e+00 6.16968989e-01
-5.80918550e-01 6.88605070e-01 6.73399270e-01 -1.87346324e-01
5.04085541e-01 -7.13385820e-01 -2.52061576e-01 -4.57686305e-01
-8.07596445e-02 8.03753436e-02 -2.97114044e-01 -1.62539974e-01
5.74806750e-01 3.82946044e-01 -1.13905764e+00 7.02128187e-02
7.09391475e-01 4.76768166e-01 -7.98517287e-01 5.91847718e-01
-1.10849261e-01 1.23323452e+00 -8.88389230e-01 -1.01679647e+00
-2.39170603e-02 -2.78346717e-01 -7.48221636e-01 -8.45967457e-02
8.18803310e-01 -3.88876706e-01 -3.78078818e-01 1.12120140e+00
6.42554820e-01 -1.71509579e-01 7.53724098e-01 -1.14486456e+00
-7.43496954e-01 1.13990188e+00 -8.49671185e-01 6.12584770e-01
4.23476458e-01 2.74849147e-01 6.63923502e-01 -8.54576945e-01
9.08785582e-01 1.62442076e+00 5.86592734e-01 4.98327464e-01
-1.33333218e+00 -1.45756292e+00 -2.38425598e-01 -3.24370652e-01
-1.40010059e+00 -6.90632403e-01 6.90768540e-01 -1.10625648e+00
3.01124722e-01 2.82129914e-01 6.44554436e-01 1.59601307e+00
-1.17684357e-01 4.13445830e-01 1.44417739e+00 -5.04642539e-02
-2.78481215e-01 5.46834111e-01 6.78422928e-01 5.65171063e-01
5.76290727e-01 -1.30917788e-01 -5.38734078e-01 -8.83975744e-01
-9.07009989e-02 -8.16702917e-02 2.24600524e-01 4.86429870e-01
-9.84420300e-01 1.65786707e+00 3.42615455e-01 5.28676987e-01
-6.42490610e-02 -2.78333575e-01 9.27894533e-01 8.61847773e-02
1.08060277e+00 7.96970844e-01 3.37697685e-01 -1.81654185e-01
-8.98719907e-01 3.41399401e-01 7.81820893e-01 2.82131255e-01
4.73975807e-01 -1.71179678e-02 -1.12938635e-01 9.69937921e-01
-2.66897958e-02 7.77717650e-01 4.20912623e-01 -7.88051486e-01
5.15618682e-01 5.09612858e-01 3.19212154e-02 -1.65654385e+00
-5.75524986e-01 -7.14279711e-02 -6.55433297e-01 -1.91943854e-01
7.28190124e-01 -5.42719722e-01 -6.46702230e-01 1.70427728e+00
3.36499751e-01 -2.32594535e-01 -4.94603246e-01 6.73783004e-01
7.57194757e-01 5.71365833e-01 7.30853796e-01 -2.47344971e-01
1.46119785e+00 -2.84006000e-01 -9.56924319e-01 -1.52699858e-01
7.49926388e-01 -8.67087007e-01 9.03159738e-01 -1.90562010e-02
-8.06279480e-01 -1.78409174e-01 -5.59238136e-01 -1.24259572e-02
-7.09859133e-01 -7.40779340e-01 4.43265110e-01 1.19147754e+00
-4.34197336e-01 3.84563267e-01 -1.76591501e-01 -4.10589725e-01
7.29255795e-01 -2.76411653e-01 -3.99154991e-01 1.61260113e-01
-1.39648986e+00 1.05890393e+00 8.86242837e-02 -4.44564372e-01
-7.25882709e-01 -7.34404325e-01 -8.29570413e-01 -6.86748445e-01
9.38968211e-02 1.25540048e-01 8.64537418e-01 -1.17372525e+00
-6.95210934e-01 1.59204078e+00 -1.19973823e-01 -5.16700387e-01
5.52589715e-01 -2.97094136e-01 -7.31033146e-01 1.04902111e-01
5.67209005e-01 3.78663540e-01 1.03379869e+00 -1.06904423e+00
-3.04740429e-01 -6.32867098e-01 9.06466544e-02 -2.89167166e-01
-6.38054788e-01 9.97240424e-01 8.50108564e-01 -8.67136955e-01
-3.99746925e-01 -1.10044563e+00 4.48855847e-01 -4.42048043e-01
-8.88812482e-01 -2.97604352e-01 9.78588700e-01 -1.03676701e+00
1.86500573e+00 -2.21329498e+00 -1.20181784e-01 1.23509720e-01
7.19229102e-01 3.70769083e-01 5.75520337e-01 5.91260731e-01
1.26101360e-01 9.49846804e-01 -9.06325728e-02 -3.20292443e-01
1.49745479e-01 -4.18728776e-02 -5.22268414e-01 1.14159536e+00
3.25428247e-02 4.46167290e-01 -9.67991233e-01 -7.43511617e-01
2.32256334e-02 4.07090127e-01 -3.33084702e-01 1.15561048e-02
6.43189609e-01 4.13939029e-01 6.28266484e-02 6.89880252e-01
6.45164073e-01 3.49536926e-01 -1.98898926e-01 2.00481445e-01
-7.41674840e-01 4.49076355e-01 -3.36216301e-01 3.95953655e-01
-1.24448299e-01 1.43146122e+00 3.18946958e-01 -1.08539701e-01
9.17696238e-01 1.43413126e-01 4.16188203e-02 -2.95447379e-01
6.57686293e-01 1.84331670e-01 4.40670371e-01 -5.22464454e-01
6.57745957e-01 -4.46098536e-01 -5.19859195e-01 5.29404342e-01
-3.69845897e-01 -1.38894692e-01 1.63522035e-01 3.43583524e-01
7.13575065e-01 -7.51702487e-01 3.22478682e-01 -4.60456401e-01
5.01608729e-01 2.43694149e-03 3.82812709e-01 6.74995184e-01
-7.07665145e-01 2.05312699e-01 8.85688365e-01 -5.29663503e-01
-1.13674104e+00 -7.15320289e-01 -4.51347411e-01 1.55374753e+00
-3.15074563e-01 -3.16571385e-01 -8.38483214e-01 -8.04944158e-01
2.71866828e-01 1.07267749e+00 -9.12444890e-01 -1.46457911e-01
-3.06919605e-01 -7.76176453e-01 1.24260426e+00 -1.11199252e-01
4.21182722e-01 -7.92952299e-01 -3.99874628e-01 -2.14399308e-01
-4.93519679e-02 -1.07276201e+00 -3.87465268e-01 -3.54396880e-01
-5.27659059e-02 -1.17264247e+00 -4.88588572e-01 -1.95164829e-01
4.79479760e-01 7.17466772e-02 7.37015486e-01 3.76702040e-01
-5.35924286e-02 -1.82238966e-01 -4.09261763e-01 -7.55092442e-01
-8.67553890e-01 2.25568250e-01 1.72010705e-01 2.97278583e-01
5.21443784e-01 -2.12410852e-01 -3.06689888e-01 8.78490135e-02
-8.92040670e-01 -1.95822954e-01 -3.01184207e-01 2.94388592e-01
-5.11620879e-01 -3.45301628e-01 3.73865306e-01 -1.38245237e+00
5.03156602e-01 -1.15633476e+00 -4.18409407e-02 -5.61645806e-01
-3.17714483e-01 -6.67190850e-01 6.84274316e-01 -7.55813718e-01
-6.93951368e-01 -5.09361506e-01 2.76716892e-02 -4.22605425e-02
-3.54841024e-01 8.28207284e-02 3.44513059e-01 1.14370115e-01
7.04159856e-01 -5.26329100e-01 7.91881047e-03 -2.40902379e-01
-6.01902269e-02 9.75167930e-01 -4.58866637e-03 -2.63845980e-01
1.12705123e+00 7.50296295e-01 -3.68096918e-01 -1.26793301e+00
-1.16615665e+00 -5.31853616e-01 -7.87986875e-01 -5.49966097e-01
1.25951004e+00 -8.31467152e-01 -6.71908438e-01 8.35245430e-01
-1.01201594e+00 -2.30091408e-01 5.57562590e-01 2.99916089e-01
2.52172470e-01 3.30984235e-01 -8.34626019e-01 -1.15868616e+00
-4.61205661e-01 -6.20105982e-01 5.08636355e-01 -1.35998324e-01
-1.13961816e+00 -9.64761913e-01 2.83944488e-01 6.18515790e-01
3.01415473e-01 9.79953051e-01 7.69228876e-01 -1.19362855e+00
7.53542602e-01 -3.04166824e-01 -3.89552474e-01 1.70808509e-01
1.49015695e-01 6.98929906e-01 -1.05585063e+00 -1.07579418e-01
-2.50449002e-01 -5.09822607e-01 5.80285013e-01 -1.70482785e-01
7.53554106e-01 -9.02261257e-01 -2.87583172e-01 1.29173711e-01
1.00408661e+00 -1.22416407e-01 2.26827979e-01 3.85792285e-01
1.00333571e+00 9.21378195e-01 2.84566611e-01 6.29346251e-01
1.69404775e-01 3.93378794e-01 1.62267551e-01 1.79975763e-01
1.17909901e-01 -4.65907007e-01 4.32792127e-01 5.00906050e-01
1.91069860e-02 -1.44960694e-02 -1.43690252e+00 6.79124236e-01
-1.26514363e+00 -1.71809065e+00 -6.82501554e-01 1.87341964e+00
7.04615891e-01 7.33831376e-02 7.61250198e-01 2.77744859e-01
1.23486471e+00 6.38147652e-01 -1.66340470e-01 -9.86397922e-01
-1.61149219e-01 -7.84610510e-02 7.48126864e-01 6.11704707e-01
-1.32472801e+00 8.39807987e-01 6.92700958e+00 5.42377949e-01
-1.06079435e+00 3.89296234e-01 7.92476654e-01 -4.17227060e-01
-1.33413181e-01 -3.24855864e-01 -6.51016235e-01 9.11475122e-01
1.24304295e+00 -2.49732316e-01 2.45370120e-01 6.37405574e-01
4.22461778e-01 -1.15513869e-01 -5.62732399e-01 7.18241513e-01
4.41791654e-01 -9.21038866e-01 -3.95322800e-01 4.48178232e-01
7.30890155e-01 -1.52499659e-03 2.74948031e-01 3.54768783e-01
4.20963645e-01 -1.15507841e+00 1.07835567e+00 1.95140243e-02
5.05690038e-01 -9.43072677e-01 6.91839099e-01 5.03069937e-01
-3.01297754e-01 -3.18439156e-01 -9.34605747e-02 -6.45322740e-01
2.02157080e-01 5.11976838e-01 -7.13730931e-01 -4.09181714e-01
6.96162224e-01 6.70550287e-01 -8.82082403e-01 1.80819809e-01
-1.86199889e-01 1.13574886e+00 -1.81823879e-01 -7.35281780e-02
4.98680174e-01 1.67757533e-02 6.40398741e-01 1.63892555e+00
-8.52932408e-02 -4.71463427e-02 -2.14409724e-01 6.10094249e-01
-1.75493270e-01 4.08346318e-02 -1.26768363e+00 -6.60662055e-01
5.96814990e-01 1.40596199e+00 -7.34077573e-01 -4.18723732e-01
-4.55748647e-01 6.78866923e-01 6.73426807e-01 -4.28070128e-02
-1.19285309e+00 -1.91916913e-01 9.24042344e-01 6.10040069e-01
-4.44225520e-01 -2.35709220e-01 -2.97398895e-01 -8.78115296e-01
-8.79993141e-01 -9.82404590e-01 6.27498865e-01 -4.61767316e-01
-1.75062513e+00 1.47461131e-01 1.16695829e-01 -5.06293774e-01
-9.65580791e-02 -4.23737019e-01 -5.00139296e-01 7.57543206e-01
-8.84794593e-01 -1.06725717e+00 -1.61603898e-01 3.87809962e-01
1.12505019e-01 -6.08774908e-02 6.44195437e-01 4.20855194e-01
-1.02675843e+00 4.12996411e-01 -3.27169001e-01 8.99031341e-01
8.14804316e-01 -8.01210523e-01 7.30628148e-02 5.83064795e-01
-4.31363493e-01 7.50655830e-01 8.77987444e-01 -9.03044283e-01
-6.99145675e-01 -9.36151981e-01 1.11264157e+00 -7.94109941e-01
1.45134687e+00 -4.88640755e-01 -7.32450128e-01 8.70855033e-01
4.66021985e-01 -2.96146482e-01 1.41562772e+00 1.07770942e-01
-9.16087627e-01 6.39624655e-01 -1.46360707e+00 5.09245992e-01
1.06977928e+00 -9.00968432e-01 -6.01335347e-01 3.33916605e-01
4.79629606e-01 -1.53289378e-01 -9.50063050e-01 -8.20203051e-02
6.72275245e-01 -1.12298799e+00 5.71998060e-01 -8.27193737e-01
7.99889803e-01 2.63289958e-01 3.16356905e-02 -1.01960766e+00
-4.89368021e-01 -5.61771750e-01 1.36499196e-01 1.72419024e+00
3.09590429e-01 -7.07975984e-01 1.89440399e-01 7.55855322e-01
1.48849830e-01 -6.91734478e-02 -9.24488306e-01 -4.50804889e-01
6.32595658e-01 -1.98293358e-01 3.27242285e-01 1.82557237e+00
8.00971836e-02 3.25568855e-01 -4.65953767e-01 3.85307558e-02
6.14894807e-01 -3.24006230e-01 9.04233098e-01 -1.15261543e+00
5.32513976e-01 -8.09920013e-01 -2.96515226e-01 3.12545300e-01
7.55313575e-01 -8.84418845e-01 -5.35316110e-01 -9.64822948e-01
4.75821316e-01 -2.38723144e-01 3.25684607e-01 1.71482325e-01
-1.20392986e-01 8.59952211e-01 4.99717653e-01 2.20569104e-01
-1.15288571e-01 -6.56732917e-02 8.73836279e-01 -1.66279614e-01
9.45790391e-03 -2.91027427e-01 -7.71376252e-01 1.09012473e+00
9.83238101e-01 -6.38958156e-01 2.55604833e-01 -5.64724393e-02
3.77433449e-01 -6.19703352e-01 5.24967194e-01 -6.31829321e-01
-3.86784673e-01 -2.63594270e-01 3.59181911e-01 -4.05066580e-01
3.36867571e-01 -6.66727722e-01 5.26510477e-02 7.25759685e-01
-5.11291921e-01 8.30580667e-02 2.08608717e-01 1.08204201e-01
2.51082294e-02 -2.27545813e-01 9.99053955e-01 4.50045988e-02
-4.96323034e-02 1.22372344e-01 -9.45525348e-01 5.13947546e-01
1.26386416e+00 -5.03510758e-02 -8.16695213e-01 -5.96916139e-01
-3.75767738e-01 -2.07146659e-01 6.08396411e-01 3.11392754e-01
-7.62664527e-02 -1.22709572e+00 -9.34257925e-01 -1.68270692e-01
2.11761430e-01 -9.40969288e-01 -7.96874911e-02 8.59751582e-01
-3.79868776e-01 8.98086503e-02 -2.53378570e-01 8.23744237e-02
-1.31557667e+00 7.53750443e-01 -1.19650876e-02 2.62560517e-01
-5.83830737e-02 5.65361440e-01 -1.62104934e-01 -4.07233715e-01
-2.90643573e-01 4.40119117e-01 -1.98844448e-01 9.43111420e-01
6.84161603e-01 9.88107681e-01 -7.64768362e-01 -1.72749043e+00
-5.26185691e-01 2.21336201e-01 6.76011220e-02 -9.29818451e-02
1.02687216e+00 -1.40538096e-01 -4.98491377e-01 8.48776281e-01
1.53906596e+00 8.80105555e-01 -3.51100832e-01 1.28384188e-01
8.13363045e-02 -8.46706033e-01 -9.84042138e-02 -3.93980771e-01
-6.57283843e-01 6.14412844e-01 -1.05365571e-02 1.11071289e+00
1.14517428e-01 6.13420568e-02 7.55197406e-01 -4.08968627e-01
2.02147201e-01 -1.14836562e+00 -1.24520756e-01 6.06749237e-01
8.81025493e-01 -1.30458069e+00 2.04027668e-01 -4.92036909e-01
-7.17942834e-01 8.10001731e-01 6.69961452e-01 -2.63979316e-01
6.55549347e-01 -1.03918880e-01 2.37395853e-01 -4.27250117e-01
-9.49084684e-02 -1.03946760e-01 1.54497862e-01 4.46917355e-01
8.53543401e-01 3.56196910e-01 -8.95526588e-01 2.34460354e-01
-8.03100109e-01 -7.61157572e-01 5.71144998e-01 4.64250416e-01
-2.83653706e-01 -2.95035183e-01 -8.84557545e-01 3.88071060e-01
-1.12985992e+00 1.22765414e-01 -1.24574935e+00 1.08874869e+00
2.80285120e-01 1.17049837e+00 3.88656557e-01 -5.25144994e-01
-1.45887315e-01 5.05832911e-01 -1.27894850e-02 -6.71116710e-01
-1.22679675e+00 -4.83073622e-01 7.11656392e-01 -2.10822955e-01
-5.29212058e-01 -7.26261497e-01 -8.34624350e-01 -1.35287023e+00
-8.25048089e-02 -8.00272673e-02 4.32132781e-01 7.08889902e-01
1.58192888e-01 -1.92193314e-01 7.84947991e-01 -7.95187771e-01
-1.58014409e-02 -1.18968952e+00 -4.26957935e-01 9.46401536e-01
6.03392303e-01 -7.57984698e-01 -9.36063170e-01 -3.11142176e-01] | [8.715365409851074, 10.469146728515625] |
9ff64218-1d4e-4d1d-806c-8740c819e0dd | a-survey-of-automated-data-augmentation | 2206.06544 | null | https://arxiv.org/abs/2206.06544v2 | https://arxiv.org/pdf/2206.06544v2.pdf | A Survey of Automated Data Augmentation Algorithms for Deep Learning-based Image Classification Tasks | In recent years, one of the most popular techniques in the computer vision community has been the deep learning technique. As a data-driven technique, deep model requires enormous amounts of accurately labelled training data, which is often inaccessible in many real-world applications. A data-space solution is Data Augmentation (DA), that can artificially generate new images out of original samples. Image augmentation strategies can vary by dataset, as different data types might require different augmentations to facilitate model training. However, the design of DA policies has been largely decided by the human experts with domain knowledge, which is considered to be highly subjective and error-prone. To mitigate such problem, a novel direction is to automatically learn the image augmentation policies from the given dataset using Automated Data Augmentation (AutoDA) techniques. The goal of AutoDA models is to find the optimal DA policies that can maximize the model performance gains. This survey discusses the underlying reasons of the emergence of AutoDA technology from the perspective of image classification. We identify three key components of a standard AutoDA model: a search space, a search algorithm and an evaluation function. Based on their architecture, we provide a systematic taxonomy of existing image AutoDA approaches. This paper presents the major works in AutoDA field, discussing their pros and cons, and proposing several potential directions for future improvements. | ['Qiuhong Ke', 'James Bailey', 'Richard O. Sinnott', 'Zihan Yang'] | 2022-06-14 | null | null | null | null | ['image-augmentation'] | ['computer-vision'] | [ 2.58421332e-01 3.94724756e-02 -3.62240881e-01 -2.77919859e-01
-1.49976656e-01 -4.44962561e-01 6.50865257e-01 -4.63615842e-02
-4.59184140e-01 4.86582100e-01 -7.03804865e-02 -2.78271079e-01
7.16512874e-02 -8.71364474e-01 -4.48717713e-01 -9.90963995e-01
2.98047692e-01 5.47506452e-01 -4.26008999e-02 -1.91888511e-01
1.92376599e-01 7.62947679e-01 -1.71938813e+00 1.09928541e-01
8.89951766e-01 1.15549767e+00 2.49930039e-01 4.99384969e-01
-5.12946010e-01 6.33785009e-01 -8.08779657e-01 -2.35021904e-01
5.99081099e-01 -4.70863372e-01 -7.35745370e-01 5.97233176e-01
1.71098843e-01 -2.05594435e-01 -1.90545827e-01 9.79428411e-01
4.18303192e-01 -2.62837484e-02 6.94432378e-01 -1.60054314e+00
-6.53730035e-01 2.27355704e-01 -5.28171659e-01 2.21654072e-01
-4.10434633e-01 2.24690810e-01 6.23701811e-01 -6.34130955e-01
3.01830232e-01 1.02026343e+00 3.84370476e-01 7.65560567e-01
-1.23451173e+00 -5.31136870e-01 1.84344262e-01 2.04800993e-01
-1.12663627e+00 -2.60506809e-01 1.15043473e+00 -6.07183039e-01
5.89979947e-01 2.54689872e-01 6.96745038e-01 9.30370986e-01
-2.84219235e-01 1.03559935e+00 1.40084302e+00 -8.62265110e-01
4.24499750e-01 4.10667419e-01 9.24402028e-02 3.77394110e-01
2.23727450e-01 2.32492968e-01 -2.92765349e-01 -1.12768002e-01
1.12695420e+00 -3.21176425e-02 -2.18687323e-03 -7.87866890e-01
-8.94621611e-01 9.71064806e-01 4.91246492e-01 2.22987726e-01
-5.84311843e-01 -2.08361104e-01 4.95361209e-01 3.40227038e-01
2.91049838e-01 8.01303208e-01 -5.81602395e-01 1.51326433e-01
-5.18387496e-01 4.35878456e-01 4.00506020e-01 6.87483966e-01
7.03321755e-01 3.38929623e-01 -3.27658430e-02 9.58424747e-01
2.06411153e-01 1.02281399e-01 8.13989699e-01 -7.49771953e-01
1.43584222e-01 9.08573329e-01 4.88212891e-02 -1.00056577e+00
-1.90313578e-01 -6.35135591e-01 -1.10977781e+00 5.90670228e-01
4.16137099e-01 -3.10444441e-02 -1.16554523e+00 1.64150751e+00
4.01666343e-01 -6.47567436e-02 2.73095593e-02 8.19680750e-01
5.28989434e-01 5.68053901e-01 2.18306392e-01 -2.42578074e-01
1.20818031e+00 -1.06879902e+00 -7.50233650e-01 -4.80058402e-01
5.69250405e-01 -5.92059791e-01 1.33259189e+00 5.41561544e-01
-8.38793039e-01 -7.11092949e-01 -1.11511230e+00 1.65923849e-01
-4.80497360e-01 2.98781842e-01 6.95978701e-01 6.87154710e-01
-8.70382726e-01 1.57638177e-01 -7.41265833e-01 -2.94429243e-01
5.45722365e-01 4.89044011e-01 -2.27657884e-01 7.51772225e-02
-9.06326592e-01 8.79749000e-01 7.14479327e-01 2.38155220e-02
-8.07979226e-01 -3.19133371e-01 -6.44911349e-01 -1.12685084e-01
5.11165679e-01 -5.60344696e-01 1.45706642e+00 -1.42026389e+00
-1.31936741e+00 9.59199011e-01 9.39983502e-02 -8.21723521e-01
3.96704435e-01 -2.24025503e-01 -3.38767529e-01 -1.63259506e-01
-3.07092339e-01 9.09039617e-01 1.08807564e+00 -1.61022973e+00
-7.15581834e-01 -3.22323769e-01 1.26551360e-01 2.65032321e-01
-7.37797856e-01 -1.02761105e-01 -5.03922105e-01 -9.37791467e-01
-1.69311315e-01 -8.88608456e-01 -6.27291381e-01 -1.94281433e-02
-1.67791799e-01 -1.69032276e-01 1.04108703e+00 -2.53464997e-01
1.44163835e+00 -2.11315084e+00 3.90688851e-02 7.64345527e-02
2.25972518e-01 7.15774894e-01 -2.02926576e-01 1.45018861e-01
-1.70765370e-01 1.36844188e-01 -3.16156179e-01 -1.04429172e-02
-4.54740912e-01 4.41321820e-01 -3.63301992e-01 4.30436060e-02
3.63905251e-01 7.45290220e-01 -7.03594148e-01 -3.90945613e-01
5.02435744e-01 2.33317345e-01 -4.42202181e-01 4.74335909e-01
-4.89647925e-01 4.84805524e-01 -4.78232026e-01 5.23227453e-01
5.80088258e-01 -4.11801875e-01 -4.92324494e-02 -2.50006706e-01
-1.32272929e-01 -6.54193759e-02 -1.02639639e+00 1.28898430e+00
-3.79123956e-01 6.28055394e-01 -1.51419640e-01 -1.28212178e+00
1.12527752e+00 2.56370038e-01 6.58726573e-01 -5.49965441e-01
3.78115922e-01 1.37374505e-01 2.68308729e-01 -5.24548173e-01
3.69229227e-01 7.20520169e-02 3.24075848e-01 2.99880564e-01
-8.15361887e-02 2.20365208e-02 2.06180975e-01 -1.69122085e-01
7.65023172e-01 -1.63759798e-01 5.71400166e-01 -8.55384916e-02
7.14945018e-01 3.89905959e-01 6.02458894e-01 7.20879316e-01
-2.34285116e-01 5.92510402e-01 2.36826658e-01 -9.42268789e-01
-1.42226696e+00 -6.00150764e-01 -4.51148972e-02 7.48268545e-01
-5.97049557e-02 -8.44725221e-03 -9.52504158e-01 -8.69858444e-01
-2.84578860e-01 4.44266886e-01 -7.13928521e-01 -2.00408563e-01
-5.75817585e-01 -9.97852802e-01 2.54769266e-01 5.95429659e-01
8.48317146e-01 -1.32885396e+00 -6.68643951e-01 1.99826166e-01
4.69863042e-02 -1.18170142e+00 -1.22565076e-01 1.66801900e-01
-1.09165847e+00 -8.99312198e-01 -6.63676560e-01 -8.04499090e-01
1.05380821e+00 3.55206728e-01 1.09464049e+00 3.54309887e-01
-2.45861426e-01 1.54343307e-01 -5.84029257e-01 -8.83265793e-01
-6.08031690e-01 1.79029077e-01 8.26644674e-02 1.09380655e-01
6.52522862e-01 -2.42293179e-01 -5.64851105e-01 4.00777370e-01
-1.17473173e+00 2.32742742e-01 9.68842924e-01 1.07360208e+00
7.27929771e-01 2.36156002e-01 6.72959387e-01 -7.69472659e-01
7.28876173e-01 -3.09744954e-01 -8.19246650e-01 1.51319668e-01
-7.88378358e-01 9.40808058e-02 5.36397278e-01 -5.56775808e-01
-8.71344447e-01 4.36380535e-01 -1.73221931e-01 -6.57038152e-01
-4.98302162e-01 4.16121125e-01 -3.34600419e-01 -1.47018552e-01
9.58553374e-01 3.83629918e-01 2.89039344e-01 -5.47852755e-01
3.49856257e-01 7.14726090e-01 4.47965294e-01 -4.88576591e-01
7.48023689e-01 3.22062284e-01 -8.96114632e-02 -9.38678324e-01
-9.48873580e-01 -3.95772070e-01 -7.06691682e-01 -1.72220469e-01
7.99607933e-01 -5.51760316e-01 -1.50140896e-01 7.81413078e-01
-1.03091133e+00 -3.89642388e-01 -5.13472378e-01 3.12626004e-01
-5.38249671e-01 1.34376973e-01 -1.90475151e-01 -6.88899636e-01
-3.76651734e-01 -1.48906231e+00 5.11833251e-01 5.39978206e-01
-1.08971082e-01 -8.12369585e-01 -1.68853238e-01 3.69867384e-01
3.85776788e-01 1.75088346e-01 1.10410202e+00 -8.48267198e-01
-5.59796929e-01 -3.51682454e-01 -1.30769327e-01 7.83911288e-01
4.38582957e-01 -2.04403624e-01 -1.09338188e+00 -1.05682947e-01
1.96353972e-01 -4.02235955e-01 3.73632133e-01 5.44245541e-01
1.74366212e+00 -4.43747222e-01 -1.46037698e-01 3.67956579e-01
1.33206594e+00 6.32111371e-01 5.98976731e-01 7.18445539e-01
4.69515353e-01 4.99694139e-01 8.46688986e-01 2.76686311e-01
7.83976614e-02 6.59775317e-01 6.55961514e-01 -3.70911896e-01
-6.29156157e-02 -1.33889854e-01 -9.06031877e-02 5.94770253e-01
-5.13703264e-02 -3.56818646e-01 -9.66048419e-01 5.39671183e-01
-1.75176966e+00 -6.14683747e-01 1.09462902e-01 2.08574390e+00
7.63856232e-01 1.35098830e-01 3.91934365e-01 5.00458360e-01
6.19193375e-01 -1.55248463e-01 -6.63204968e-01 -3.65104556e-01
-1.10144265e-01 -8.77630413e-02 3.38553041e-01 1.63494170e-01
-1.32141221e+00 9.79532659e-01 6.82006216e+00 7.25888312e-01
-1.23064864e+00 -1.26508087e-01 9.85804677e-01 3.06240410e-01
1.37433410e-01 -2.51237720e-01 -7.13542640e-01 3.96746099e-01
4.96763766e-01 -3.65073010e-02 2.74320692e-01 1.21543670e+00
1.97274238e-01 -1.17152696e-02 -1.02166772e+00 1.09883201e+00
3.89540307e-02 -1.51573682e+00 3.28767747e-01 3.38279992e-01
8.37451398e-01 -1.49409562e-01 1.29696935e-01 2.20816076e-01
2.81268924e-01 -9.35469270e-01 2.49148563e-01 1.15381710e-01
4.45444912e-01 -7.66369164e-01 7.10814297e-01 4.64940935e-01
-7.61914492e-01 -2.82629967e-01 -2.04745695e-01 -3.18701230e-02
-1.93193898e-01 4.80491251e-01 -8.02168906e-01 1.61682099e-01
6.28307760e-01 2.69787639e-01 -4.90935892e-01 1.24651551e+00
-1.57036141e-01 5.85394084e-01 -1.38879344e-01 1.75279707e-01
2.79530108e-01 -2.20223382e-01 4.02713686e-01 7.94670165e-01
1.18196823e-01 1.08149081e-01 2.50207871e-01 5.71976721e-01
-1.10297196e-01 6.21319935e-02 -7.00120807e-01 -2.61374384e-01
6.76170707e-01 1.22295821e+00 -7.03523099e-01 -3.15544635e-01
-5.10448813e-01 8.04627299e-01 1.18964270e-01 1.88081786e-01
-5.10461450e-01 -5.48311397e-02 6.83385432e-01 2.12631762e-01
-1.03953242e-01 -2.78986573e-01 -7.22043157e-01 -8.30504537e-01
-8.34673271e-02 -1.34743273e+00 4.87672359e-01 -5.89798689e-01
-1.28692973e+00 7.58854389e-01 -6.47420157e-03 -1.44917142e+00
-2.98237383e-01 -7.29561031e-01 -3.31891268e-01 6.95353746e-01
-1.41088712e+00 -9.70535517e-01 -5.07848859e-01 4.00989145e-01
9.77253318e-01 -6.24877572e-01 6.79188728e-01 3.90694082e-01
-5.43600082e-01 5.36348939e-01 -1.31267175e-01 2.54193783e-01
3.12996149e-01 -1.06152129e+00 4.70323533e-01 8.96902442e-01
1.98530465e-01 3.06836396e-01 7.68603206e-01 -3.30483288e-01
-1.13242614e+00 -1.00999784e+00 4.89345819e-01 -1.65042222e-01
3.21784109e-01 -8.35172683e-02 -1.17220807e+00 5.44776380e-01
1.58429205e-01 1.26272395e-01 5.77838004e-01 -2.18747988e-01
5.12877889e-02 -1.78145379e-01 -1.18575370e+00 7.53688693e-01
6.86536849e-01 -1.14680357e-01 -2.85393715e-01 5.64615428e-02
4.47154403e-01 -3.74392092e-01 -5.13484716e-01 6.08474255e-01
1.75869837e-01 -8.08888257e-01 9.46495473e-01 -6.95014894e-01
2.14566752e-01 -4.74245638e-01 9.76539552e-02 -1.37721384e+00
-2.76874840e-01 -2.57990092e-01 -3.67090344e-01 1.11590827e+00
1.04927346e-01 -4.19061929e-01 1.05554926e+00 7.75621176e-01
3.14464569e-02 -1.02494061e+00 -4.58294302e-01 -9.26213026e-01
-8.27596188e-02 -4.10570323e-01 7.86636889e-01 1.05871403e+00
-5.82008183e-01 3.36365640e-01 -4.86804485e-01 1.86195299e-01
3.92298281e-01 -2.21162587e-01 9.98963535e-01 -1.32310343e+00
-2.97685135e-02 -6.38284802e-01 -6.01370633e-01 -1.03857648e+00
-2.11567327e-01 -5.45494318e-01 -1.02839269e-01 -1.39350688e+00
-3.21617685e-02 -7.70119071e-01 -3.16180408e-01 6.78012609e-01
-5.64043373e-02 4.77845632e-02 1.40745372e-01 3.27167243e-01
1.18630812e-01 6.05391324e-01 1.24052024e+00 -8.06443766e-02
-4.62542057e-01 1.36435047e-01 -7.60288894e-01 1.01922202e+00
1.22013307e+00 -3.41319442e-01 -8.20015669e-01 -6.31739199e-01
3.35887484e-02 -4.14292514e-01 2.13730931e-01 -9.61818695e-01
-1.58404857e-02 -4.48302656e-01 3.14290255e-01 -5.33233762e-01
2.05325544e-01 -1.15996897e+00 -1.64965421e-01 4.20825541e-01
-3.25021535e-01 2.39345640e-01 2.67897457e-01 2.99942851e-01
-4.50690418e-01 -3.15374196e-01 1.07978714e+00 -2.04485089e-01
-8.66089344e-01 4.77256209e-01 -4.03942227e-01 -1.63413882e-01
1.23343909e+00 -2.72844613e-01 -2.58676219e-03 -1.23456120e-01
-5.25823295e-01 6.20065704e-02 3.20488721e-01 6.08744204e-01
5.08994460e-01 -1.32575524e+00 -3.68830621e-01 3.00304502e-01
9.70012620e-02 3.64714205e-01 1.80696212e-02 3.17418665e-01
-3.58226717e-01 2.31596068e-01 -4.54455256e-01 -5.85401058e-01
-1.29400611e+00 7.29713678e-01 4.49848771e-01 -2.60543287e-01
-3.66206318e-01 6.40834153e-01 3.59380424e-01 -2.00755224e-01
3.71018231e-01 -5.03212735e-02 -3.97538841e-01 -1.03317633e-01
5.77112496e-01 1.70347661e-01 1.44101173e-01 -3.38550270e-01
1.39077842e-01 2.86394358e-01 -4.07487154e-01 -7.01454878e-02
1.20676672e+00 -5.42602204e-02 1.50547653e-01 1.12154596e-01
7.42950439e-01 -5.84439039e-01 -1.25416684e+00 -4.84464973e-01
3.49232703e-02 -6.08874381e-01 1.93524674e-01 -8.67616653e-01
-1.28406322e+00 9.87615943e-01 8.48545015e-01 3.30730855e-01
1.64802313e+00 -1.51754454e-01 5.02192974e-01 3.80987972e-01
1.89653799e-01 -1.24938965e+00 3.53336692e-01 3.03731561e-01
1.09348333e+00 -1.46726573e+00 -1.78347692e-01 -3.00667942e-01
-8.71971071e-01 9.74353731e-01 1.12913072e+00 6.71120659e-02
6.08442724e-01 2.06064254e-01 3.25314552e-01 -1.14969760e-01
-3.82195473e-01 -1.62003100e-01 2.82263964e-01 7.58080781e-01
2.18373001e-01 -2.14033902e-01 -3.89203787e-01 3.77713054e-01
-6.50029778e-02 5.38547449e-02 2.96646386e-01 1.08968914e+00
-3.38059187e-01 -1.60578775e+00 -4.36452866e-01 5.68943799e-01
-2.95770556e-01 3.03555727e-01 -4.40815806e-01 8.55835795e-01
2.62091070e-01 7.43932009e-01 1.54862776e-01 -3.74275714e-01
2.47599795e-01 -1.99926607e-02 2.95369625e-01 -4.92579937e-01
-3.21744531e-01 -4.47279140e-02 -2.86944985e-01 4.29749414e-02
-4.80633497e-01 -5.32534480e-01 -1.08536518e+00 9.11904592e-03
-3.29481751e-01 -2.07365602e-02 8.28132689e-01 9.85034287e-01
4.16366339e-01 3.66893500e-01 8.36751640e-01 -6.44154251e-01
-3.86859298e-01 -8.43470275e-01 -6.51897863e-02 3.70492786e-01
3.91763479e-01 -7.20556855e-01 5.25435768e-02 2.81559706e-01] | [9.348225593566895, 2.241117238998413] |
3d4263f1-34f4-48e7-98a8-5898e7d40f35 | retrieval-augmented-generative-question | 2211.07067 | null | https://arxiv.org/abs/2211.07067v1 | https://arxiv.org/pdf/2211.07067v1.pdf | Retrieval-Augmented Generative Question Answering for Event Argument Extraction | Event argument extraction has long been studied as a sequential prediction problem with extractive-based methods, tackling each argument in isolation. Although recent work proposes generation-based methods to capture cross-argument dependency, they require generating and post-processing a complicated target sequence (template). Motivated by these observations and recent pretrained language models' capabilities of learning from demonstrations. We propose a retrieval-augmented generative QA model (R-GQA) for event argument extraction. It retrieves the most similar QA pair and augments it as prompt to the current example's context, then decodes the arguments as answers. Our approach outperforms substantially prior methods across various settings (i.e. fully supervised, domain transfer, and fewshot learning). Finally, we propose a clustering-based sampling strategy (JointEnc) and conduct a thorough analysis of how different strategies influence the few-shot learning performance. The implementations are available at https:// github.com/xinyadu/RGQA | ['Heng Ji', 'Xinya Du'] | 2022-11-14 | null | null | null | null | ['generative-question-answering'] | ['natural-language-processing'] | [ 4.52537626e-01 1.99060827e-01 -3.10719460e-01 -4.37781543e-01
-1.79673100e+00 -8.26091707e-01 1.00902700e+00 1.04519188e-01
-5.34758270e-01 8.27592909e-01 6.03148639e-01 -4.95239556e-01
3.96718383e-02 -7.55875468e-01 -9.59203959e-01 -5.73589206e-01
3.09326112e-01 7.82880247e-01 2.31706753e-01 -9.74176601e-02
4.28551793e-01 -4.60246392e-02 -1.42998028e+00 5.77016771e-01
9.41819727e-01 5.90941429e-01 2.93014765e-01 9.23730910e-01
-2.50534624e-01 1.06100214e+00 -7.61375725e-01 -7.99077034e-01
-9.64889750e-02 -7.65029073e-01 -9.76530433e-01 -4.01358962e-01
1.06562734e-01 -4.20299292e-01 -3.48660052e-01 7.59930789e-01
8.50958943e-01 2.30911642e-01 7.36835182e-01 -1.24032617e+00
-7.16099679e-01 1.16676176e+00 -8.76277909e-02 5.56144595e-01
5.04838407e-01 1.78161815e-01 1.28958607e+00 -1.09801459e+00
8.94599319e-01 1.30430186e+00 3.55883330e-01 5.69245517e-01
-1.16064298e+00 -5.63342154e-01 2.54500777e-01 5.84174871e-01
-5.98178506e-01 -5.47131419e-01 8.89404833e-01 -2.14970380e-01
1.25611722e+00 3.21802013e-02 5.21854341e-01 1.95682704e+00
-5.45319580e-02 1.34951091e+00 8.95234823e-01 -6.00593567e-01
3.31241459e-01 -3.97288382e-01 4.10377413e-01 4.56603765e-01
-1.53336218e-02 8.93869177e-02 -8.55665743e-01 -4.59555894e-01
2.96943933e-01 -1.57602295e-01 -5.46057560e-02 -9.81123969e-02
-1.26873112e+00 1.11814344e+00 -1.11471258e-01 -1.09998450e-01
-6.43397093e-01 4.05051589e-01 5.69014013e-01 3.98091942e-01
3.49793315e-01 2.67662734e-01 -6.17115378e-01 -5.66533506e-01
-6.99723423e-01 6.44175529e-01 9.61514413e-01 1.07208562e+00
6.36022508e-01 -2.06123769e-01 -5.37248194e-01 7.76664555e-01
2.29360744e-01 3.16357851e-01 5.21851480e-01 -7.77911961e-01
8.46167326e-01 2.27348819e-01 1.71057254e-01 -2.34875694e-01
4.31966260e-02 -7.06304312e-02 -2.31205933e-02 -2.53476351e-01
5.41163146e-01 -4.27631766e-01 -8.19805503e-01 1.90020752e+00
4.79853898e-01 1.67364717e-01 4.05626386e-01 8.00789535e-01
9.04721856e-01 7.56286085e-01 5.25946736e-01 -1.71349883e-01
1.54097342e+00 -1.31779194e+00 -7.36220956e-01 -4.54775423e-01
4.92004097e-01 -8.43251467e-01 1.24806488e+00 2.73269594e-01
-1.26948309e+00 -4.06343758e-01 -8.57604444e-01 -3.71637374e-01
-2.04005092e-01 2.40047574e-01 6.54773772e-01 1.86397314e-01
-4.68867362e-01 5.61147332e-01 -8.53956044e-01 -1.13118865e-01
3.73769492e-01 -7.50622153e-02 1.52253628e-01 -3.38363498e-02
-1.55187130e+00 8.09959054e-01 4.89500672e-01 -3.73535573e-01
-1.36207747e+00 -7.29220212e-01 -9.82848883e-01 2.41466202e-02
6.76053643e-01 -9.78638768e-01 1.91097212e+00 -5.76816738e-01
-1.48658323e+00 5.52770793e-01 -4.16417927e-01 -7.50787973e-01
3.98398072e-01 -8.76420796e-01 -1.59864962e-01 3.43286097e-01
4.28741753e-01 6.43246472e-01 1.10216129e+00 -7.92584419e-01
-6.00762308e-01 -6.55778050e-02 2.05081269e-01 3.77853721e-01
2.02271223e-01 2.67437369e-01 -3.07135522e-01 -7.46199071e-01
-1.12488575e-01 -8.23784649e-01 -1.56616822e-01 -6.44100666e-01
-3.88219208e-01 -8.86520326e-01 4.94805843e-01 -7.22125053e-01
1.09371352e+00 -2.10309148e+00 2.76553214e-01 -2.79966235e-01
-1.32936358e-01 2.06293121e-01 -3.35058123e-01 9.99589205e-01
-1.48918718e-01 -8.80580693e-02 -1.62273601e-01 -4.41210270e-01
3.05918097e-01 2.27675393e-01 -8.48831475e-01 7.29443207e-02
5.99465907e-01 1.19664061e+00 -1.23930025e+00 -7.31471062e-01
1.60672388e-03 3.89931440e-01 -5.59521377e-01 5.93674958e-01
-6.85492575e-01 1.32685691e-01 -7.10000277e-01 5.74725986e-01
1.90163076e-01 -5.68391144e-01 3.87347758e-01 -3.41268659e-01
1.56966195e-01 1.15422475e+00 -8.38196754e-01 2.05606508e+00
-2.74744034e-01 4.11080837e-01 -5.67399204e-01 -1.09436107e+00
6.21554434e-01 5.10636926e-01 -4.53491323e-03 -6.82264805e-01
1.02222227e-01 1.52093053e-01 -5.01087867e-02 -6.55496776e-01
5.23489416e-01 6.96480796e-02 -3.28032434e-01 7.10268497e-01
4.88554448e-01 1.16887912e-01 4.42012876e-01 7.54408956e-01
1.22416556e+00 8.47719550e-01 4.89793211e-01 3.24869573e-01
6.67498708e-02 2.90624589e-01 5.08206964e-01 9.56607878e-01
-7.47255310e-02 5.11224270e-01 4.33265209e-01 -1.81127682e-01
-1.02674377e+00 -1.22983897e+00 3.26910436e-01 1.34334671e+00
1.47541687e-01 -6.89752162e-01 -6.49096072e-01 -1.15244102e+00
-1.18750930e-01 1.09325039e+00 -4.16597664e-01 -7.30607286e-02
-8.60150874e-01 -6.71173632e-01 6.31634712e-01 7.90395379e-01
1.59775153e-01 -1.45227206e+00 -9.48842347e-01 4.29315954e-01
-7.06219673e-01 -8.32990468e-01 -2.83081383e-01 5.19441664e-01
-7.42431343e-01 -1.20813584e+00 -5.41840374e-01 -6.88618064e-01
2.50912040e-01 -4.95549515e-02 1.40337121e+00 -1.25515059e-01
-2.70309538e-01 5.47721505e-01 -7.20199883e-01 -4.26023513e-01
-4.68065858e-01 4.39262092e-02 -3.55662137e-01 -5.33510566e-01
5.43741047e-01 -6.32663548e-01 -6.88469708e-01 -1.03396960e-01
-5.80465674e-01 4.80862968e-02 8.48701954e-01 1.03050339e+00
7.61145175e-01 -5.51959097e-01 9.44269121e-01 -1.10999739e+00
1.01497531e+00 -7.44034469e-01 -2.79126436e-01 3.21831018e-01
-5.98677456e-01 2.89261490e-01 3.97227705e-01 -7.82356203e-01
-1.44904280e+00 -1.98884279e-01 -1.92939192e-01 -2.89723784e-01
-3.60799402e-01 4.83254403e-01 6.84935460e-03 9.17789102e-01
7.04507411e-01 3.73939127e-01 -1.98506594e-01 -4.52455103e-01
7.15030015e-01 2.86597818e-01 4.50833112e-01 -1.06509054e+00
6.64321601e-01 2.07554653e-01 -6.27697647e-01 -4.13249075e-01
-1.17195487e+00 -4.21152443e-01 -2.69189477e-01 -4.69991080e-02
8.91537011e-01 -9.95103598e-01 -3.41643393e-01 7.32608587e-02
-1.43919468e+00 -5.24408996e-01 -3.35035086e-01 6.21682525e-01
-8.19801033e-01 4.39360440e-01 -1.03080904e+00 -8.59642804e-01
-6.76972389e-01 -8.91376972e-01 1.16141307e+00 2.65959531e-01
-6.04754984e-01 -7.49194145e-01 3.76005858e-01 3.64924788e-01
1.58950388e-02 -1.77194625e-02 1.09572601e+00 -1.15363550e+00
-6.16909683e-01 1.32691547e-01 5.56725403e-03 -2.00046189e-02
-1.43982023e-01 -2.71742970e-01 -8.73241663e-01 6.43893033e-02
-2.81714275e-02 -6.16839230e-01 1.00204718e+00 2.92442501e-01
7.22829759e-01 -4.08404469e-01 -5.31192541e-01 8.03373083e-02
1.02615738e+00 3.16760719e-01 6.01316512e-01 4.68296587e-01
2.34290406e-01 4.79209989e-01 1.11589682e+00 4.29009467e-01
4.68331575e-01 3.11566025e-01 2.52930462e-01 3.69406611e-01
-4.25519019e-01 -7.21571743e-01 5.99063933e-01 6.78572237e-01
1.36496797e-01 -4.88524705e-01 -6.02832437e-01 8.44159961e-01
-2.15946007e+00 -1.28613126e+00 1.90185551e-02 1.69156480e+00
1.06062067e+00 1.72492519e-01 1.80961728e-01 -1.56637952e-01
4.75689232e-01 2.50946522e-01 -7.54900753e-01 -1.55103773e-01
-2.26012757e-03 4.77990180e-01 -1.03985507e-03 2.68621325e-01
-1.10641170e+00 1.09601581e+00 5.87808657e+00 9.67164755e-01
-5.29056013e-01 1.74606398e-01 4.13172692e-01 -1.06799342e-01
-5.09531379e-01 4.49237734e-01 -1.02365983e+00 4.92857397e-01
1.13766134e+00 -1.89611748e-01 2.42344648e-01 1.02081490e+00
-2.25558117e-01 -5.41704595e-02 -1.37910461e+00 6.81954622e-01
1.70635074e-01 -1.42468023e+00 9.87915993e-02 -2.65039861e-01
3.18596125e-01 1.23407811e-01 1.41945444e-02 6.82356358e-01
6.76835239e-01 -5.03005803e-01 8.23693097e-01 4.18259352e-01
3.05991203e-01 -5.87551951e-01 3.15550297e-01 4.75620002e-01
-1.06153381e+00 -6.85241893e-02 -2.20905870e-01 4.11532223e-02
5.67774475e-01 3.12700182e-01 -9.82073307e-01 5.13768494e-01
5.99878132e-01 6.23359025e-01 -3.58342946e-01 6.60321057e-01
-9.72895145e-01 1.12848175e+00 -1.62385941e-01 -1.94765538e-01
1.77051127e-01 -6.73402324e-02 6.50586843e-01 1.31253922e+00
3.16388607e-01 4.78210062e-01 5.37307821e-02 1.00864518e+00
-8.81593451e-02 -3.90537232e-02 -3.47521663e-01 -3.54687482e-01
8.18282783e-01 1.28657079e+00 -5.64983010e-01 -8.73985648e-01
-3.74402583e-01 1.03702068e+00 5.67493081e-01 3.86164844e-01
-1.15751863e+00 -3.73608649e-01 4.90814477e-01 -3.48322839e-01
6.63398802e-01 -1.00884624e-01 2.97061980e-01 -1.28810894e+00
-1.99287124e-02 -1.13421118e+00 8.81956458e-01 -8.48122180e-01
-1.41132236e+00 3.86348635e-01 2.91917086e-01 -1.13294733e+00
-8.36106300e-01 -3.34198594e-01 -8.00570011e-01 6.26168847e-01
-1.59402800e+00 -1.07574952e+00 2.08694309e-01 4.80912983e-01
1.19985199e+00 -1.96734723e-03 8.29479754e-01 2.24537969e-01
-4.21311468e-01 4.59682375e-01 -4.57727611e-01 2.78654128e-01
8.86755526e-01 -1.35993528e+00 7.39411294e-01 9.35900807e-01
5.30634463e-01 8.53070199e-01 6.96971655e-01 -8.75063777e-01
-1.56649959e+00 -8.12129915e-01 1.08463514e+00 -7.84626961e-01
7.67978728e-01 -1.91413924e-01 -7.86629200e-01 9.11290169e-01
6.55873179e-01 -4.23876673e-01 8.12424898e-01 1.88808098e-01
-6.21133983e-01 4.81893212e-01 -6.66509748e-01 7.42878616e-01
1.14771307e+00 -6.49782062e-01 -1.28793991e+00 4.60346401e-01
1.03711712e+00 -3.63275379e-01 -6.20794535e-01 1.09023161e-01
3.81488949e-01 -7.55885184e-01 1.20259452e+00 -9.06665802e-01
7.43362904e-01 -1.26709357e-01 -1.07440889e-01 -1.11861491e+00
-8.14497322e-02 -7.80572712e-01 -7.24442601e-01 1.19585526e+00
6.86455369e-01 -3.63850027e-01 5.89546740e-01 3.85168046e-01
-3.50363672e-01 -7.00280905e-01 -5.33691645e-01 -6.76810622e-01
-9.53391939e-02 -4.41747546e-01 4.84142482e-01 6.41584516e-01
1.27698734e-01 9.65888679e-01 -3.19059491e-01 1.98583588e-01
6.97592437e-01 6.10576153e-01 8.27266037e-01 -8.57222021e-01
-6.72974527e-01 -9.72596332e-02 3.82164150e-01 -1.35608089e+00
2.52106160e-01 -8.69598925e-01 2.30403095e-01 -1.65401638e+00
2.57593006e-01 -1.28494874e-01 -2.40369275e-01 4.61161733e-01
-5.87215662e-01 -1.69772804e-01 -3.49899828e-02 2.70860761e-01
-9.39923108e-01 5.23152530e-01 8.00450206e-01 7.80208260e-02
-7.99502730e-02 -4.51985896e-02 -5.27925074e-01 6.78931534e-01
9.21884716e-01 -7.01704443e-01 -5.58488071e-01 -3.81948888e-01
4.27328944e-01 3.54887933e-01 4.29078192e-01 -6.43137038e-01
2.73010015e-01 -1.17883254e-02 2.91594684e-01 -1.03886080e+00
4.73595828e-01 -1.77746892e-01 -1.71641290e-01 3.20241094e-01
-6.89122975e-01 2.56968170e-01 -5.81446588e-02 8.87003601e-01
-1.31047770e-01 -5.36353350e-01 9.95958447e-02 -2.75018752e-01
-8.17287624e-01 1.91861764e-02 -3.33809197e-01 5.34952044e-01
6.95510864e-01 9.68043357e-02 -6.73815846e-01 -4.06550080e-01
-6.68923855e-01 1.34298250e-01 -1.04996927e-01 3.71872485e-01
7.36931205e-01 -1.14613998e+00 -7.49299288e-01 -2.10013404e-01
8.14132467e-02 -9.29981023e-02 1.72840342e-01 7.87775517e-01
1.55941755e-01 2.58078456e-01 2.45801032e-01 -3.52188855e-01
-1.32607877e+00 7.20169544e-01 -2.95914978e-01 -4.85122114e-01
-8.55431676e-01 1.05204010e+00 2.25629658e-02 -2.17927977e-01
1.80956125e-01 -2.06982225e-01 -2.11350605e-01 2.42209673e-01
4.42273051e-01 2.79995739e-01 -1.62211210e-01 -8.19579288e-02
-1.63362011e-01 7.81337395e-02 -4.68733102e-01 -4.28785652e-01
1.31676185e+00 9.09371153e-02 2.98450589e-01 4.17473078e-01
9.09506202e-01 3.71932797e-02 -1.43619537e+00 -3.42952639e-01
3.88521224e-01 -1.28461659e-01 -5.99759519e-01 -8.90365660e-01
-2.30412498e-01 8.74761522e-01 9.20771584e-02 7.60778710e-02
7.87503839e-01 5.31606615e-01 1.03273010e+00 6.08231425e-01
2.32087642e-01 -1.14660954e+00 6.19795918e-01 5.16424179e-01
6.12541676e-01 -1.12247133e+00 -1.31166965e-01 -2.94782847e-01
-8.68807793e-01 9.60258961e-01 4.63384777e-01 -2.92391747e-01
1.93792790e-01 1.26653090e-01 -5.51250502e-02 -3.54607373e-01
-1.25854039e+00 -3.19694996e-01 1.15467697e-01 5.35182059e-01
5.16592801e-01 -9.77643803e-02 -5.24421632e-01 8.59715164e-01
-2.11004242e-01 -1.31356254e-01 1.91902399e-01 1.18315458e+00
-3.59333873e-01 -1.45797861e+00 5.07118320e-03 3.40693593e-01
-4.88882482e-01 -4.33292300e-01 -3.40299040e-01 7.64056921e-01
-2.75119364e-01 9.23700690e-01 -7.54110739e-02 3.00582629e-02
2.33482897e-01 7.09983766e-01 6.01167917e-01 -8.47425282e-01
-6.47130013e-01 2.68137932e-01 6.05904758e-01 -5.58062971e-01
-4.98009205e-01 -1.01968157e+00 -1.46466005e+00 3.68923753e-01
-3.24258029e-01 5.09649277e-01 5.15480876e-01 1.00459826e+00
5.92273057e-01 5.54170251e-01 1.28436103e-01 -5.78405738e-01
-9.92216229e-01 -9.69399452e-01 2.86399245e-01 5.62681317e-01
5.66985905e-02 -6.42957926e-01 -3.72100234e-01 2.48394012e-01] | [10.757536888122559, 8.479764938354492] |
5f408e4a-4832-4040-bc9b-b8f3145c5117 | andrejjan-at-semeval-2019-task-7-a-fusion | null | null | https://aclanthology.org/S19-2190 | https://aclanthology.org/S19-2190.pdf | AndrejJan at SemEval-2019 Task 7: A Fusion Approach for Exploring the Key Factors pertaining to Rumour Analysis | The viral spread of false, unverified and misleading information on the Internet has attracted a heightened attention of an interdisciplinary research community on the phenomenon. This paper contributes to the research efforts of automatically determining the veracity of rumourous tweets and classifying their replies according to stance. Our research objective was to investigate the interplay between a number of phenomenological and contextual features of rumours, in particular, we explore the extent to which network structural characteristics, metadata and user profiles could complement the linguistic analysis of the written content for the task at hand. The current findings strongly demonstrate that supplementary sources of information play significant role in classifying the veracity and the stance of Twitter interactions deemed to be rumourous. | ['Andrej Janchevski', 'Sonja Gievska'] | 2019-06-01 | null | null | null | semeval-2019-6 | ['rumour-detection'] | ['natural-language-processing'] | [-3.08509474e-03 2.84788400e-01 -4.36493129e-01 5.85145764e-02
-8.89511853e-02 -6.31942868e-01 1.11406446e+00 7.03715682e-01
-9.15025175e-02 5.88464618e-01 9.69920218e-01 -6.57042623e-01
1.53437257e-01 -5.50344169e-01 -3.31660546e-02 -2.75403440e-01
2.48493224e-01 9.91934240e-02 -1.09682493e-02 -5.66914141e-01
9.15807664e-01 3.37939501e-01 -1.23372638e+00 3.92990112e-01
7.37620294e-01 7.41462350e-01 -3.31628293e-01 2.79792428e-01
-2.93074667e-01 1.38558710e+00 -6.73531830e-01 -4.32599634e-01
-4.60232407e-01 -5.93139589e-01 -9.09486055e-01 3.07004929e-01
1.62555084e-01 -9.72144902e-02 -3.61036569e-01 1.08468187e+00
1.57843024e-01 -1.71096697e-01 4.03439581e-01 -1.16833770e+00
-7.02736139e-01 9.43307757e-01 -5.74086666e-01 1.00834012e+00
5.40393829e-01 1.33016899e-01 8.27001214e-01 -6.38038397e-01
9.20097291e-01 1.10486937e+00 8.25943768e-01 2.00338755e-02
-9.54320967e-01 -6.93631172e-01 -4.24676031e-01 1.91829558e-02
-1.17919028e+00 -7.89135695e-01 7.86552370e-01 -1.09165502e+00
3.77908021e-01 4.07469004e-01 6.22457981e-01 1.41163599e+00
1.98867872e-01 7.51947761e-02 1.42087746e+00 -1.52377352e-01
-5.19383559e-03 6.25179052e-01 2.93880910e-01 5.22531033e-01
4.73852098e-01 -2.93315172e-01 -7.54411936e-01 -7.68423140e-01
1.86603770e-01 -2.15822667e-01 -4.74860251e-01 5.11135280e-01
-9.20617163e-01 1.33801484e+00 2.53497839e-01 9.54998314e-01
-6.93087161e-01 -4.95933264e-01 8.37026715e-01 5.15839458e-01
1.20358324e+00 6.21332467e-01 4.01655883e-02 -4.80997354e-01
-1.03087926e+00 1.90342478e-02 1.16914344e+00 1.44318074e-01
5.57127118e-01 -9.96677652e-02 2.11717054e-01 7.08896756e-01
2.52050161e-01 3.64382327e-01 5.81196845e-01 -4.26826000e-01
4.34673607e-01 6.61445498e-01 3.73358607e-01 -1.95734501e+00
-5.11415184e-01 -4.80927914e-01 -2.56682277e-01 -5.36327362e-01
2.22728059e-01 -4.65083510e-01 2.08412725e-02 1.26001036e+00
2.50754297e-01 -2.03233123e-01 -3.49088222e-01 7.90249765e-01
7.40110576e-01 5.10917962e-01 3.91860120e-03 -5.81409156e-01
1.22570336e+00 -2.13872492e-01 -8.30255747e-01 -2.62409836e-01
5.45172095e-01 -1.26647353e+00 7.44057953e-01 -9.93223041e-02
-1.02385688e+00 9.44339558e-02 -9.91228521e-01 2.60799795e-01
-2.97654063e-01 -6.20932817e-01 1.93405703e-01 8.31262231e-01
-8.73308659e-01 5.31498790e-01 -2.32191816e-01 -6.34684503e-01
4.18425500e-01 -5.94799042e-01 -4.81726229e-02 6.31836876e-02
-1.23591352e+00 1.06829584e+00 3.41316938e-01 2.50181444e-02
-2.27779955e-01 -3.38713467e-01 -4.00326490e-01 -5.09689033e-01
4.50969338e-01 -1.36261702e-01 1.01572037e+00 -1.26066470e+00
-1.00561833e+00 1.08399856e+00 -1.55928627e-01 -3.24631810e-01
5.52270412e-01 3.63117993e-01 -8.72245848e-01 3.37774426e-01
5.26449680e-01 -5.16167760e-01 8.36344063e-01 -1.05127203e+00
-4.24259871e-01 -5.02069831e-01 -3.10066372e-01 -1.44927591e-01
-4.93040651e-01 5.68885326e-01 3.44506711e-01 -6.51110530e-01
2.13985279e-01 -7.29192793e-01 4.24103945e-01 -7.48329043e-01
-4.80938733e-01 -3.24576467e-01 8.71748209e-01 -8.16420317e-01
1.58569121e+00 -2.10555720e+00 -4.61924404e-01 3.12118441e-01
7.96560407e-01 1.83117145e-03 4.96169895e-01 1.23628414e+00
3.98673207e-01 8.52630377e-01 3.87486607e-01 3.12092274e-01
-1.90073669e-01 -1.36698365e-01 -4.81372118e-01 1.13526118e+00
-2.59222716e-01 7.93762863e-01 -1.14414370e+00 -3.35914940e-01
-3.68334532e-01 2.60055304e-01 -4.84551415e-02 -8.34319219e-02
-5.11687361e-02 4.17304933e-01 -7.23058581e-01 3.91502887e-01
3.71384203e-01 -6.88211143e-01 3.02139431e-01 4.11600731e-02
-6.72536969e-01 8.68197858e-01 -2.41016895e-01 2.18583167e-01
-1.17796436e-01 1.43467832e+00 3.27248782e-01 -4.11889732e-01
8.35618317e-01 3.27904224e-01 2.58962840e-01 -5.58654666e-01
6.08339310e-01 2.38838255e-01 1.26928866e-01 -8.02381277e-01
8.76471579e-01 -6.28154397e-01 -7.21533075e-02 9.77666557e-01
-7.21435964e-01 2.63032675e-01 -8.56805295e-02 4.27186042e-01
8.61369073e-01 -7.09893048e-01 5.09042919e-01 -3.71212125e-01
2.23609954e-01 3.45001191e-01 5.55027574e-02 5.63558578e-01
-3.22822183e-01 8.04541807e-04 8.51131737e-01 -9.99351144e-02
-9.72353935e-01 -3.53822261e-01 -3.22418630e-01 1.11003876e+00
9.87808481e-02 -3.52814883e-01 -5.50321698e-01 -2.99175650e-01
7.87450448e-02 8.17913771e-01 -8.19466233e-01 5.89739010e-02
-2.45056733e-01 -8.68708253e-01 7.73373783e-01 -2.39837199e-01
3.54034394e-01 -8.26700568e-01 -4.96601492e-01 3.61153543e-01
-8.31642926e-01 -1.20217931e+00 -3.08273196e-01 -5.01330018e-01
-7.28284299e-01 -1.38429737e+00 -4.78946678e-02 -2.75165945e-01
4.13748860e-01 6.36271536e-01 8.67849469e-01 4.91564125e-01
3.04751307e-01 8.96140635e-02 -6.18536472e-01 -2.88278013e-01
-1.02029574e+00 -1.80301461e-02 1.01774768e-03 2.07946360e-01
3.97061169e-01 -4.96150404e-01 -1.40519783e-01 4.31511641e-01
-1.00542247e+00 -2.83213943e-01 5.00282906e-02 3.04809630e-01
-3.94379944e-01 1.67021051e-01 8.60734105e-01 -1.25270581e+00
1.42220604e+00 -1.52155626e+00 1.26739949e-01 -4.86676306e-01
-6.82730079e-01 -4.92104709e-01 2.79950768e-01 -2.47126967e-01
-8.37514400e-01 -1.20680499e+00 9.24603492e-02 2.04527527e-01
7.98012465e-02 1.14528310e+00 6.86857045e-01 -1.74491376e-01
1.03996587e+00 1.57665052e-02 4.35898423e-01 -3.92118394e-01
-2.93613877e-02 1.18213236e+00 5.85544556e-02 -1.18859364e-02
6.59865737e-01 8.05613816e-01 -5.47915459e-01 -1.26341534e+00
-1.14417851e+00 -7.67145634e-01 -2.43605182e-01 -4.91966873e-01
3.32556397e-01 -7.46764779e-01 -7.28167236e-01 3.64483595e-01
-9.26953912e-01 1.05673648e-01 4.08194840e-01 8.36302787e-02
-1.01867609e-01 6.29829407e-01 -8.56339931e-01 -7.93686152e-01
-2.57778406e-01 -5.04455209e-01 -3.20023485e-02 -2.84035027e-01
-1.00184643e+00 -1.16846275e+00 1.45327106e-01 9.32031929e-01
6.90416157e-01 7.27706492e-01 6.53026760e-01 -1.12783766e+00
-3.24067473e-02 -4.05337393e-01 -3.26204479e-01 -2.60906339e-01
2.43308917e-01 -2.07617953e-02 -5.90252757e-01 -2.25203652e-02
5.69211304e-01 -4.55022484e-01 4.33054775e-01 -1.71487838e-01
5.94507232e-02 -1.08593488e+00 -2.95591295e-01 -3.76274467e-01
1.22828531e+00 -3.21047753e-01 2.68361807e-01 6.68136179e-01
4.16509569e-01 1.03168166e+00 1.96193054e-01 8.98804247e-01
2.52502203e-01 3.53829116e-01 3.65201235e-01 5.80364227e-01
1.32557765e-01 -4.10439849e-01 3.96176845e-01 9.11311209e-01
2.28009686e-01 -3.50794017e-01 -1.22401011e+00 6.50420010e-01
-1.47005844e+00 -1.34730971e+00 -8.45201612e-01 1.70346522e+00
6.68037355e-01 2.82927126e-01 4.00091231e-01 1.75693095e-01
9.33656156e-01 7.83231080e-01 4.10246290e-02 -5.59876561e-01
-8.61774012e-02 -6.70995831e-01 4.52325523e-01 5.72079480e-01
-4.67717946e-01 3.86568129e-01 6.46780396e+00 3.03383559e-01
-1.20765078e+00 3.19813132e-01 5.17679513e-01 7.45909140e-02
-6.11354291e-01 -2.49617144e-01 -4.16004807e-01 8.34459424e-01
1.30012763e+00 -3.63830626e-01 3.46661270e-01 4.02232230e-01
1.00285864e+00 -2.69732565e-01 -6.26376510e-01 2.59515584e-01
5.16838431e-01 -1.45942330e+00 -3.18019986e-01 3.79669607e-01
6.74338281e-01 5.05805969e-01 1.44361556e-01 -2.28369445e-01
2.57890616e-02 -6.99906528e-01 9.70353663e-01 1.78897291e-01
3.45690787e-01 -3.58254224e-01 8.87271225e-01 6.64792359e-01
-1.56483278e-01 -1.26848053e-02 7.74933724e-03 -4.96335924e-01
4.68645751e-01 7.64475524e-01 -1.38468981e+00 -1.26650438e-01
2.51939893e-01 6.95284724e-01 -5.21199822e-01 6.92609906e-01
-2.05416717e-02 1.10842657e+00 5.92862628e-03 -3.26541096e-01
6.06251895e-01 -7.89502934e-02 9.37824786e-01 1.21773612e+00
-1.33937567e-01 1.42887101e-01 -1.21188000e-01 8.58025074e-01
-1.50924549e-01 2.79798329e-01 -6.04143798e-01 -9.39409077e-01
6.63634598e-01 9.62064922e-01 -9.37803924e-01 -1.86578467e-01
-2.63931155e-01 3.94624352e-01 3.74922901e-01 -3.51658836e-02
-3.71401340e-01 2.21169174e-01 2.66398489e-01 9.75430071e-01
8.94059390e-02 -2.30593756e-01 -5.81879854e-01 -9.42595184e-01
-1.83424264e-01 -8.26134443e-01 1.69543043e-01 -5.44963837e-01
-1.56286776e+00 5.69663763e-01 -3.57127786e-01 -6.79051936e-01
1.41968047e-02 1.41578734e-01 -6.96455419e-01 9.66095388e-01
-1.33732021e+00 -6.43915117e-01 -4.00452197e-01 1.48662522e-01
1.64342806e-01 2.58859098e-01 2.92406142e-01 7.01942071e-02
-4.56433475e-01 -6.46916553e-02 1.50262922e-01 1.01037167e-01
4.76879835e-01 -5.16168773e-01 2.60357261e-01 3.78657043e-01
-2.51886010e-01 7.07343161e-01 1.36962426e+00 -1.13915122e+00
-1.01712525e+00 -6.67482555e-01 1.55458581e+00 -6.05927408e-01
1.73631036e+00 2.27477863e-01 -9.51778948e-01 6.11842752e-01
3.82392555e-01 -5.55150151e-01 9.74696100e-01 3.42057347e-01
-4.20386583e-01 7.29346991e-01 -1.21850181e+00 4.24929559e-01
6.96508169e-01 -8.44400287e-01 -7.98104763e-01 3.65108609e-01
3.00186872e-01 7.80246109e-02 -8.76741588e-01 -3.76451880e-01
3.75370651e-01 -1.03387773e+00 3.16848159e-01 -5.62516332e-01
7.88184226e-01 2.15147480e-01 1.56821683e-01 -1.14101982e+00
-4.30850089e-01 -7.18164742e-01 1.85283124e-01 1.18241167e+00
3.44599903e-01 -9.02232766e-01 6.37133479e-01 3.88678789e-01
9.02551338e-02 -5.34463763e-01 -6.93621874e-01 -1.57769457e-01
-2.85038650e-02 -2.20774978e-01 1.76249370e-01 1.50464094e+00
6.99085236e-01 5.20889521e-01 -3.47847372e-01 -1.34925976e-01
3.60947758e-01 1.23434469e-01 4.39299047e-01 -1.41328824e+00
2.32043549e-01 -6.93815410e-01 -3.45929444e-01 -4.61554080e-01
9.88006890e-02 -7.85134614e-01 -3.34154904e-01 -1.08041358e+00
3.34698707e-01 -3.54271382e-01 2.74848342e-01 -1.61614865e-01
8.01544785e-02 3.70647967e-01 1.61192194e-01 9.58265185e-01
-4.11580950e-01 8.28154907e-02 1.01619935e+00 9.72746462e-02
-2.71585733e-01 3.71633843e-02 -1.30835128e+00 7.68239915e-01
9.46744263e-01 -5.21262527e-01 -1.59638897e-01 6.93561211e-02
9.93091285e-01 2.08918005e-01 4.02132004e-01 -2.50571668e-01
5.38930148e-02 -2.73567885e-01 -1.97346002e-01 -2.98162460e-01
-6.13119751e-02 -4.94284898e-01 6.10331818e-02 3.22832495e-01
-4.81053442e-01 2.82700986e-01 -1.23670986e-02 7.96519041e-01
-1.84787408e-01 -2.75392592e-01 5.71296930e-01 -2.33256951e-01
-1.50208026e-01 -1.77075177e-01 -9.90608454e-01 6.44412935e-01
7.39909291e-01 -3.94087344e-01 -7.45344758e-01 -8.34449530e-01
-5.11344850e-01 -3.35053980e-01 7.26962864e-01 3.87356132e-01
2.90644407e-01 -9.08717990e-01 -1.02977383e+00 -3.00385475e-01
1.36602344e-03 -9.71287429e-01 3.75347747e-03 1.50055254e+00
-4.32530344e-01 5.13970554e-01 -1.10389374e-01 7.27149025e-02
-1.03834283e+00 4.93802786e-01 4.96957153e-02 2.27265358e-01
-6.62652969e-01 3.70473236e-01 -5.58795631e-01 1.93698823e-01
-4.20665652e-01 4.91350412e-01 -4.50265855e-01 9.36770260e-01
6.62838399e-01 1.03517842e+00 -6.10343041e-03 -1.46061933e+00
-1.30436048e-01 -4.22373265e-01 -6.11763783e-02 -1.59592822e-01
1.10275447e+00 -7.11715043e-01 -6.39122248e-01 9.42261159e-01
1.45624113e+00 5.40520847e-01 -3.63682449e-01 -2.99941927e-01
3.24023843e-01 -7.87038505e-01 2.77299076e-01 -7.11966157e-01
-4.92011428e-01 2.06626967e-01 -4.72234547e-01 1.30791044e+00
2.19017103e-01 3.08407605e-01 9.63271976e-01 -2.54612356e-01
7.90749118e-02 -8.87559056e-01 1.39612839e-01 5.09319663e-01
8.71544778e-01 -1.12061787e+00 1.71667054e-01 -5.28177679e-01
-7.41728485e-01 1.04527736e+00 6.58028945e-02 2.00811446e-01
6.60410047e-01 -9.55090299e-02 1.23814307e-01 -7.88447976e-01
-4.93545532e-01 1.32164478e-01 1.08050190e-01 1.91459790e-01
6.81439996e-01 7.97600076e-02 -1.08099961e+00 2.07720682e-01
-5.53833961e-01 -1.58885852e-01 1.22247064e+00 6.58933282e-01
-9.12509620e-01 -3.06220800e-01 -4.69333678e-01 6.29717290e-01
-1.13729298e+00 -1.63063243e-01 -9.94066477e-01 5.45122683e-01
-2.58046329e-01 1.42373741e+00 -9.79271904e-02 -4.43527162e-01
-2.92924702e-01 1.92286577e-02 -2.68745035e-01 -5.33103704e-01
-9.55400586e-01 -2.25608826e-01 8.71970117e-01 -2.45974422e-01
-5.78061819e-01 -1.04774833e+00 -7.64564097e-01 -1.00422573e+00
-5.65234303e-01 5.14209747e-01 6.60348654e-01 1.19475734e+00
3.55179459e-01 -1.31917045e-01 7.68415451e-01 -2.49960199e-01
-5.90025365e-01 -1.11716390e+00 -5.35958052e-01 4.70967978e-01
6.77059889e-01 -3.72338086e-01 -7.94263065e-01 -3.06574136e-01] | [8.339365005493164, 10.085081100463867] |
975499d2-cd03-4232-b9f6-534587be7013 | bottrinet-a-unified-and-efficient-embedding | 2304.03144 | null | https://arxiv.org/abs/2304.03144v4 | https://arxiv.org/pdf/2304.03144v4.pdf | BotTriNet: A Unified and Efficient Embedding for Social Bots Detection via Metric Learning | The rapid and accurate identification of bot accounts in online social networks is an ongoing challenge. In this paper, we propose BOTTRINET, a unified embedding framework that leverages the textual content posted by accounts to detect bots. Our approach is based on the premise that account personalities and habits can be revealed through their contextual content. To achieve this, we designed a triplet network that refines raw embeddings using metric learning techniques. The BOTTRINET framework produces word, sentence, and account embeddings, which we evaluate on a real-world dataset, CRESCI2017, consisting of three bot account categories and five bot sample sets. Our approach achieves state-of-the-art performance on two content-intensive bot sets, with an average accuracy of 98.34% and f1score of 97.99%. Moreover, our method makes a significant breakthrough on four content-less bot sets, with an average accuracy improvement of 11.52% and an average f1score increase of 16.70%. Our contribution is twofold: First, we propose a unified and effective framework that combines various embeddings for bot detection. Second, we demonstrate that metric learning techniques can be applied in this context to refine raw embeddings and improve classification performance. Our approach outperforms prior works and sets a new standard for bot detection in social networks. | ['Yanyuet Man', 'Xuesong Ye', 'Jun Wu'] | 2023-04-06 | null | null | null | null | ['metric-learning', 'metric-learning'] | ['computer-vision', 'methodology'] | [-8.70656371e-02 -2.93718278e-01 -4.11650777e-01 2.13598564e-01
-1.88894838e-01 -6.53128624e-01 7.89696991e-01 1.54557943e-01
-6.32477283e-01 3.97186339e-01 8.99967402e-02 -1.92185655e-01
2.31938779e-01 -9.52932775e-01 1.37332097e-01 -2.66363204e-01
5.61911575e-02 5.03521800e-01 7.94370472e-01 -3.93004239e-01
6.48240864e-01 2.58890718e-01 -1.03346419e+00 1.43471137e-01
6.26233637e-01 7.51415133e-01 -1.17140993e-01 9.31875646e-01
-1.58126488e-01 1.18244684e+00 -8.72013152e-01 -1.18530583e+00
2.49471396e-01 -2.08164468e-01 -1.05491245e+00 -9.50661451e-02
5.76697737e-02 -6.35398805e-01 -6.92691445e-01 1.14999855e+00
3.71807486e-01 -3.58842939e-01 7.05711305e-01 -1.30616927e+00
-1.22712076e+00 7.94639230e-01 -7.78940856e-01 4.64859664e-01
1.52602568e-01 5.52989423e-01 1.43976974e+00 -7.01301277e-01
5.28074205e-01 1.17529047e+00 1.13805020e+00 6.67257428e-01
-1.32797849e+00 -6.02229476e-01 -1.25719067e-02 4.04477000e-01
-9.00784850e-01 -6.32477850e-02 4.78371799e-01 -8.09467137e-01
9.76958990e-01 2.83073783e-02 4.67449456e-01 1.35753143e+00
-2.63713837e-01 6.67941511e-01 1.01857424e+00 -1.13567479e-01
-3.09431165e-01 6.04315579e-01 3.68217260e-01 9.01205361e-01
7.79396892e-01 -3.80794555e-01 -2.27102295e-01 -4.02650177e-01
5.40304720e-01 2.44987339e-01 2.77201682e-01 2.18702629e-01
-1.17768312e+00 1.42442000e+00 6.20612800e-01 5.29024720e-01
-2.89896309e-01 2.40988508e-01 8.17297697e-01 1.59441993e-01
8.29669952e-01 7.66095221e-01 -1.45606637e-01 -3.87002259e-01
-5.38993239e-01 1.63917914e-01 9.66219425e-01 6.30050480e-01
7.44274199e-01 -1.40266374e-01 -3.91664416e-01 9.02243435e-01
2.26440519e-01 3.99632305e-01 5.26990175e-01 -7.68434584e-01
3.50221246e-01 8.82796705e-01 1.49628893e-01 -1.45407414e+00
-7.21111521e-02 -8.70418623e-02 -4.91471082e-01 -1.73324198e-01
5.65922201e-01 -1.15148976e-01 -2.97695309e-01 1.36653364e+00
2.76533455e-01 1.82172969e-01 -5.27904868e-01 3.95702988e-01
5.87847054e-01 9.70715433e-02 -5.75408340e-02 1.08013064e-01
1.49730825e+00 -1.20196939e+00 -6.04061246e-01 -2.49683276e-01
5.88862002e-01 -6.91267550e-01 1.17793429e+00 -1.09213866e-01
-5.72389603e-01 -1.89453185e-01 -7.28240788e-01 1.36480749e-01
-7.05746293e-01 -6.27198219e-02 5.12482524e-01 1.06769466e+00
-1.01997817e+00 8.39982867e-01 -4.92004961e-01 -6.89868987e-01
8.34328532e-01 2.29924738e-01 -1.26524135e-01 2.40528211e-01
-9.70056772e-01 8.31007719e-01 2.34748721e-01 -5.68166673e-01
-6.26454532e-01 -6.26716614e-01 -4.47818577e-01 -1.65187508e-01
2.41586119e-01 -2.95224518e-01 1.29067934e+00 -3.40596348e-01
-1.26380193e+00 1.12289119e+00 -2.83772359e-03 -6.19744301e-01
5.06586611e-01 2.92044640e-01 -2.68730640e-01 2.31034577e-01
6.03758693e-01 1.04941048e-01 1.06899440e+00 -9.73592222e-01
-5.94605207e-01 -2.45377153e-01 3.92829806e-01 -7.05047548e-01
-1.45227885e+00 5.45917094e-01 -1.23016588e-01 -5.77280581e-01
-3.81769806e-01 -8.83641481e-01 -8.12659785e-02 -3.08554202e-01
-5.30641735e-01 -6.80526078e-01 1.08743382e+00 -7.13022530e-01
1.54997277e+00 -1.76809430e+00 -9.69470590e-02 -1.40473351e-01
1.08768582e+00 8.65719974e-01 -1.89142212e-01 6.59290493e-01
2.81375408e-01 7.59263754e-01 -1.38637185e-01 -8.23275208e-01
4.10619676e-01 -1.52061641e-01 -1.98884621e-01 4.34923321e-01
6.38063908e-01 1.14764595e+00 -1.18458200e+00 -6.16423249e-01
2.75473874e-02 3.75719726e-01 -9.13778722e-01 2.09966287e-01
1.16708361e-01 2.56644666e-01 -3.13820094e-01 8.21391821e-01
5.42105019e-01 -5.34934759e-01 2.53747255e-01 4.89879064e-02
3.67322303e-02 3.60463977e-01 -6.29746020e-01 7.24715114e-01
-3.71238500e-01 9.65545893e-01 -1.59269154e-01 -6.82436347e-01
7.84318566e-01 6.48307893e-03 6.60083652e-01 -2.87819058e-01
5.90475976e-01 2.99992412e-01 -1.09264664e-01 -5.00160575e-01
7.35829532e-01 3.20041209e-01 -7.61625990e-02 1.07390237e+00
1.90507732e-02 1.52327776e-01 6.22082174e-01 4.38631594e-01
1.68383420e+00 -6.02567315e-01 3.41779262e-01 -8.17089155e-02
6.92825437e-01 -5.50372228e-02 2.71754652e-01 8.50911021e-01
-1.01643932e+00 2.18033299e-01 1.11163270e+00 -5.28325796e-01
-1.15305221e+00 -6.68059170e-01 1.22038670e-01 1.24755907e+00
7.50458166e-02 -7.21948504e-01 -8.54288757e-01 -1.51912010e+00
2.97961205e-01 -1.19518805e-02 -8.11165392e-01 1.55384332e-01
-8.32369387e-01 -9.23496068e-01 7.26139009e-01 3.43521804e-01
5.78989208e-01 -1.24713469e+00 1.50624961e-01 1.90605134e-01
-4.45639074e-01 -1.46232438e+00 -7.43298471e-01 -5.60000837e-01
-5.06501317e-01 -1.62336969e+00 -3.78108382e-01 -6.05911434e-01
2.86783785e-01 6.34995461e-01 1.13925385e+00 5.81931353e-01
-1.44492030e-01 8.70035738e-02 -6.05393231e-01 -1.27938420e-01
-7.08979547e-01 5.18034458e-01 1.94176123e-01 2.30229363e-01
7.11605728e-01 -7.46912360e-01 -5.59865892e-01 5.31925738e-01
-6.16917610e-01 -7.51876652e-01 4.66776192e-01 7.59182155e-01
-2.54888654e-01 -1.43537581e-01 8.55165541e-01 -1.18292725e+00
1.04305291e+00 -9.00505602e-01 -1.93449885e-01 -3.09377730e-01
-8.43593657e-01 -3.61366004e-01 6.41147614e-01 -6.95997775e-01
-3.58131975e-01 -4.41245764e-01 -1.53512672e-01 -2.18564823e-01
3.41901064e-01 -3.27059403e-02 4.89709795e-01 -2.66418666e-01
1.02146220e+00 2.09860113e-02 1.95425525e-01 -5.92779696e-01
3.03494841e-01 1.29876435e+00 1.92417786e-01 -1.40432119e-01
1.30441797e+00 4.67163444e-01 -4.40864414e-01 -6.90898418e-01
-7.97604740e-01 -1.02236784e+00 -8.38444829e-01 -2.22907141e-01
8.46760154e-01 -5.39467394e-01 -1.22413635e+00 8.27709317e-01
-1.42325914e+00 -2.41370499e-01 -7.57352188e-02 -1.32622560e-02
-2.21939832e-01 8.55644584e-01 -1.05609274e+00 -7.89617896e-01
-5.46670616e-01 -9.80189085e-01 7.21556723e-01 -1.68360010e-01
-4.21692431e-01 -1.16950858e+00 2.78676569e-01 5.99171698e-01
8.74835074e-01 2.54500657e-01 4.43717271e-01 -1.01855302e+00
-3.29183012e-01 -5.07924974e-01 -9.52622533e-01 5.64987361e-01
3.74625176e-01 -6.78699017e-02 -1.02847135e+00 -3.62189978e-01
-4.05556560e-01 -3.91382486e-01 1.16976452e+00 -4.78437036e-01
8.95251215e-01 -7.05472827e-01 -2.85846025e-01 8.55604745e-03
1.31722510e+00 -5.18321633e-01 4.02524978e-01 4.41355199e-01
1.00640476e+00 4.19473112e-01 6.50647804e-02 7.00500607e-01
4.00530517e-01 6.66335940e-01 8.33296478e-01 5.55544198e-01
-3.27589363e-01 -3.55629116e-01 4.32195723e-01 1.01898396e+00
-1.23392425e-01 -7.15935379e-02 -7.54409075e-01 9.09280002e-01
-1.79128766e+00 -1.34010828e+00 -5.39834440e-01 1.59278095e+00
7.55651414e-01 7.71204829e-02 1.14826655e+00 3.61859262e-01
1.05030620e+00 4.65313166e-01 -4.14131254e-01 -4.67552662e-01
4.27828848e-01 2.13182285e-01 6.93638504e-01 5.25226593e-01
-1.34436977e+00 1.16064668e+00 6.18696594e+00 8.50244522e-01
-6.40306652e-01 7.87272036e-01 1.10368781e-01 2.92938381e-01
2.77019143e-01 -3.77470642e-01 -9.02406275e-01 7.66910791e-01
1.13004148e+00 -1.73943132e-01 7.57005215e-01 9.68606234e-01
6.99836910e-02 4.79619682e-01 -5.04055500e-01 8.18902493e-01
2.58283317e-01 -1.42647862e+00 -2.96698928e-01 5.92630684e-01
7.94629693e-01 2.55810231e-01 1.17396064e-01 5.22338986e-01
7.02794909e-01 -7.80610502e-01 5.47982812e-01 -5.96654741e-03
8.59762490e-01 -4.30968583e-01 8.40116501e-01 2.71179676e-01
-1.35023320e+00 -6.94560647e-01 -3.52088720e-01 -2.28989944e-01
-3.35431541e-03 5.49777687e-01 -1.24101067e+00 1.61680534e-01
5.18750072e-01 1.34156799e+00 -8.53774428e-01 9.07602668e-01
-3.70345831e-01 8.09058726e-01 8.90605077e-02 -5.49118161e-01
2.98842341e-01 1.57149091e-01 6.74586594e-01 1.45400214e+00
-2.32740328e-01 -4.45138365e-01 1.50276661e-01 9.92691875e-01
-4.71696883e-01 3.53557095e-02 -7.07095385e-01 -5.15524864e-01
6.14432812e-01 1.59922755e+00 -6.94751740e-01 -3.51367235e-01
-2.26406395e-01 1.10575604e+00 7.47873843e-01 -2.46456698e-01
-1.00789297e+00 -6.47111297e-01 1.12122703e+00 1.76003292e-01
4.15494323e-01 -1.99737579e-01 -3.73538107e-01 -1.21698809e+00
-9.26375687e-02 -7.97929764e-01 2.04530701e-01 1.78741977e-01
-1.91287732e+00 7.17111588e-01 -3.33451003e-01 -1.05561829e+00
-2.27992162e-01 -1.04979813e+00 -7.98359931e-01 6.82400107e-01
-1.51972103e+00 -1.33567858e+00 -3.82222205e-01 2.73640364e-01
4.23457772e-01 -4.36677098e-01 7.57260203e-01 5.17023921e-01
-9.57568645e-01 8.32356095e-01 1.35704190e-01 7.18178868e-01
4.82571036e-01 -1.32533395e+00 7.51415253e-01 7.13119984e-01
2.13850569e-03 4.67333227e-01 3.84982646e-01 -6.03676438e-01
-8.78875732e-01 -1.48157680e+00 1.28534079e+00 -1.30257261e+00
1.49428964e+00 -5.29241204e-01 -5.49326837e-01 5.96705139e-01
4.12456654e-02 -4.90672328e-02 4.84423459e-01 7.32802674e-02
-8.92323494e-01 2.30654597e-01 -1.55799484e+00 4.59069610e-01
1.28472626e+00 -8.58457327e-01 -7.87815452e-02 7.94533014e-01
8.70430708e-01 3.07712555e-01 -9.71254408e-01 -7.85287693e-02
6.55770719e-01 -1.06596935e+00 9.22829270e-01 -6.92984045e-01
5.70558131e-01 6.75664842e-02 -1.09703198e-01 -1.09830999e+00
-6.69242084e-01 -7.39624083e-01 -4.59757030e-01 1.32772410e+00
3.16691786e-01 -1.07682610e+00 8.90734076e-01 -3.32066491e-02
2.60091960e-01 -6.54858828e-01 -6.91159010e-01 -9.74353671e-01
1.94166437e-01 -2.52410740e-01 6.55426204e-01 9.51809645e-01
3.43890458e-01 2.65585542e-01 -4.35940593e-01 -1.02519102e-01
6.01008356e-01 -2.62293547e-01 1.00715411e+00 -1.41646361e+00
-2.20915958e-01 -6.66904926e-01 -6.64000452e-01 -7.64507532e-01
4.24198449e-01 -8.77278924e-01 -4.03079987e-01 -1.55157614e+00
6.55954301e-01 -3.39642704e-01 -1.16492018e-01 4.64130610e-01
-2.69159645e-01 1.21262240e+00 4.18922156e-01 5.96066594e-01
-6.90532327e-01 2.46325865e-01 9.81819093e-01 -4.55490559e-01
1.13008395e-01 1.02992110e-01 -9.47625756e-01 7.39607990e-01
1.25981879e+00 -6.86457813e-01 3.05091310e-02 -3.52080137e-01
6.46191537e-02 -8.66253912e-01 3.59879404e-01 -7.92813420e-01
-8.14339742e-02 1.70747899e-02 -3.22470456e-01 -1.84194461e-01
3.41326654e-01 -3.85167390e-01 -6.87341154e-01 7.51622438e-01
1.70637205e-01 2.01938450e-01 -6.49057850e-02 9.19573843e-01
2.97574908e-01 -4.00647789e-01 9.13987160e-01 -1.92354143e-01
-3.75474930e-01 4.44608659e-01 -3.51188928e-01 3.84497702e-01
9.02487636e-01 -1.28850833e-01 -7.62951672e-01 -1.45882547e-01
-2.19454050e-01 -1.99896425e-01 4.83787268e-01 5.29486954e-01
2.79004008e-01 -1.34789920e+00 -8.71298373e-01 6.84205070e-02
2.50729173e-01 -1.12853789e+00 -2.58128613e-01 1.00261033e+00
-4.95908529e-01 2.76015490e-01 -6.83549643e-02 -4.81391937e-01
-1.27443373e+00 4.52737480e-01 1.28550500e-01 -7.57489085e-01
-2.21978784e-01 8.04518938e-01 -6.71903014e-01 -7.76829839e-01
-1.49245679e-01 2.42340744e-01 -5.98206103e-01 3.36813986e-01
7.90130556e-01 9.14829791e-01 -9.63226780e-02 -9.46994960e-01
-3.11991751e-01 1.77723184e-01 -1.68057486e-01 1.15159027e-01
1.28151333e+00 5.40598966e-02 -4.95464325e-01 1.04328640e-01
1.44550753e+00 3.69037747e-01 -6.41642332e-01 -5.13463140e-01
1.78947002e-01 -1.12796807e+00 -5.23178518e-01 -4.98096019e-01
-1.00870502e+00 7.13536024e-01 2.55422473e-01 1.09921527e+00
4.04167712e-01 4.06586342e-02 1.27910435e+00 2.88622707e-01
2.64099479e-01 -8.00793111e-01 8.48867059e-01 1.00246549e+00
3.89947146e-01 -1.57599461e+00 -2.91527867e-01 -4.57125872e-01
-3.28811288e-01 8.56221080e-01 6.12458706e-01 -1.85318828e-01
5.89869857e-01 -2.60811448e-01 -2.57481217e-01 -4.04607952e-01
-3.17376703e-01 -6.09951854e-01 -1.74584448e-01 8.21791351e-01
2.15066120e-01 1.47345200e-01 -4.04817402e-01 3.91923726e-01
-1.63788751e-01 -5.06907165e-01 7.18500257e-01 5.12295604e-01
-5.78203857e-01 -1.28205144e+00 -2.16258001e-02 6.35255337e-01
-6.72428966e-01 -4.31297533e-02 -1.02153909e+00 6.59584820e-01
5.70578761e-02 1.48268127e+00 -2.40471855e-01 -1.10530388e+00
2.53911406e-01 -8.63754898e-02 -8.92833341e-03 -1.00521386e+00
-8.65688920e-01 -9.55430269e-01 2.33957469e-01 -4.25069451e-01
-2.47266218e-01 -3.75923693e-01 -4.95236158e-01 -1.33644843e+00
-5.73867619e-01 1.01408616e-01 4.08774108e-01 7.39937305e-01
4.18484330e-01 3.07797045e-01 1.23929310e+00 -8.86395752e-01
-8.92772675e-01 -1.48277164e+00 -4.34890002e-01 5.71516514e-01
1.50585398e-01 -7.06306159e-01 -7.55924881e-01 -3.41016322e-01] | [8.126452445983887, 10.131301879882812] |
a7a23235-624c-4adb-ad7f-82c2d937fe2c | source-free-video-domain-adaptation-with | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Li_Source-Free_Video_Domain_Adaptation_With_Spatial-Temporal-Historical_Consistency_Learning_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Source-Free_Video_Domain_Adaptation_With_Spatial-Temporal-Historical_Consistency_Learning_CVPR_2023_paper.pdf | Source-Free Video Domain Adaptation With Spatial-Temporal-Historical Consistency Learning | Source-free domain adaptation (SFDA) is an emerging research topic that studies how to adapt a pretrained source model using unlabeled target data. It is derived from unsupervised domain adaptation but has the advantage of not requiring labeled source data to learn adaptive models. This makes it particularly useful in real-world applications where access to source data is restricted. While there has been some SFDA work for images, little attention has been paid to videos. Naively extending image-based methods to videos without considering the unique properties of videos often leads to unsatisfactory results. In this paper, we propose a simple and highly flexible method for Source-Free Video Domain Adaptation (SFVDA), which extensively exploits consistency learning for videos from spatial, temporal, and historical perspectives. Our method is based on the assumption that videos of the same action category are drawn from the same low-dimensional space, regardless of the spatio-temporal variations in the high-dimensional space that cause domain shifts. To overcome domain shifts, we simulate spatio-temporal variations by applying spatial and temporal augmentations on target videos, and encourage the model to make consistent predictions from a video and its augmented versions. Due to the simple design, our method can be applied to various SFVDA settings, and experiments show that our method achieves state-of-the-art performance for all the settings. | ['Martin Renqiang Min', 'Erik Kruus', 'Deep Patel', 'Kai Li'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['source-free-domain-adaptation', 'unsupervised-domain-adaptation'] | ['computer-vision', 'methodology'] | [ 1.66102946e-01 -5.38941801e-01 -6.26330853e-01 -3.25480819e-01
-4.06245679e-01 -5.28532267e-01 6.52907968e-01 -3.35855216e-01
-4.07916665e-01 7.39443958e-01 4.43436205e-01 8.46207291e-02
1.30421249e-02 -4.69903499e-01 -7.54776239e-01 -7.80407071e-01
1.39766664e-03 1.68208241e-01 4.14209843e-01 -1.22439608e-01
-8.61670636e-03 3.30727637e-01 -1.35279691e+00 1.99517444e-01
6.93830252e-01 8.21690798e-01 2.66961724e-01 4.18167740e-01
-1.53581530e-01 7.69077003e-01 -3.94920886e-01 -1.13130942e-01
4.96045053e-01 -8.79468143e-01 -6.55648828e-01 6.16038740e-01
5.04380524e-01 -4.09951270e-01 -6.69926226e-01 9.51299906e-01
2.90990561e-01 4.31520551e-01 8.08346868e-01 -1.47131443e+00
-9.06358242e-01 1.48466274e-01 -5.61118305e-01 5.41978300e-01
3.82683605e-01 1.88131824e-01 4.94591981e-01 -8.95988584e-01
8.82424772e-01 1.15539098e+00 5.08112609e-01 8.08779836e-01
-1.24068379e+00 -6.56714261e-01 6.61343157e-01 5.74654400e-01
-1.19883478e+00 -3.54009569e-01 1.02988446e+00 -6.04863644e-01
6.13092542e-01 -2.85717726e-01 5.89572430e-01 1.62480187e+00
-2.70289853e-02 8.11263621e-01 9.62035477e-01 -3.98291409e-01
4.40591395e-01 2.35911652e-01 -2.72394568e-01 2.46063843e-01
-3.20725292e-02 -4.79950570e-02 -6.11619174e-01 -8.85067359e-02
9.65390384e-01 1.99428752e-01 -3.16275895e-01 -9.90482569e-01
-1.41315174e+00 8.05845022e-01 2.23947927e-01 2.80181110e-01
-3.55426282e-01 -3.03634137e-01 5.65785110e-01 4.49942052e-01
4.45737004e-01 1.84426516e-01 -6.14597857e-01 -1.67313859e-01
-7.15803266e-01 2.88610905e-01 4.76686925e-01 1.24437153e+00
6.86044931e-01 2.89805651e-01 -4.82525975e-02 8.35471988e-01
3.81758995e-02 4.39497828e-01 9.31439161e-01 -1.05177820e+00
4.28880811e-01 4.30941314e-01 2.11715430e-01 -1.00640428e+00
9.11152083e-03 -9.22316685e-02 -8.20031047e-01 -2.41394136e-02
5.91711700e-01 -4.47038710e-02 -9.79892194e-01 1.89156878e+00
4.93001759e-01 7.41404831e-01 1.95229411e-01 1.00452292e+00
4.73514050e-01 6.53370261e-01 2.12517202e-01 -4.15007383e-01
8.43985975e-01 -9.71384406e-01 -6.61067307e-01 -2.40049973e-01
4.01339710e-01 -5.41699827e-01 1.26292348e+00 2.75134206e-01
-7.28516579e-01 -8.27041090e-01 -9.03707266e-01 2.28679880e-01
-3.15244853e-01 -2.19229445e-01 3.25774074e-01 2.51629949e-01
-7.32720971e-01 3.56808931e-01 -7.60290682e-01 -7.90752649e-01
3.23304504e-01 1.58491775e-01 -6.03432596e-01 -3.31195682e-01
-1.23538494e+00 7.18337655e-01 5.69146216e-01 -3.98386687e-01
-8.65642190e-01 -6.16660714e-01 -9.88180041e-01 -3.19562316e-01
7.66334057e-01 -6.28674388e-01 1.25812185e+00 -1.43917799e+00
-1.50612569e+00 5.65724313e-01 -3.28154564e-01 -5.20889699e-01
4.39450443e-01 -1.01723574e-01 -8.29983413e-01 2.46073306e-01
2.00878099e-01 4.86491591e-01 1.30083752e+00 -1.06225300e+00
-7.26454973e-01 -1.87494919e-01 -3.64889130e-02 2.91000992e-01
-6.66786909e-01 -1.09014928e-01 -6.63024664e-01 -1.13731086e+00
-2.99058110e-02 -1.12719095e+00 -1.20167017e-01 1.56073451e-01
1.63219959e-01 -5.13592549e-02 1.13245845e+00 -5.43103039e-01
1.32946944e+00 -2.41023111e+00 4.51171488e-01 -2.18049716e-02
-2.74173647e-01 3.26425374e-01 -3.67896408e-01 3.42022717e-01
-4.46003191e-02 -1.05892055e-01 -3.35964143e-01 5.87091455e-03
-1.71748295e-01 4.46298301e-01 -3.35029125e-01 4.30040598e-01
2.49232799e-01 5.08981645e-01 -1.23491359e+00 -5.85797966e-01
2.56979138e-01 4.00703043e-01 -6.13922775e-01 2.83780009e-01
-2.52053976e-01 7.93788195e-01 -4.35847580e-01 5.28600037e-01
5.23808360e-01 -2.12141156e-01 2.61603653e-01 -8.23653266e-02
1.46870732e-01 -8.30584764e-02 -1.24923718e+00 1.88113034e+00
-2.26771638e-01 5.38723230e-01 -2.23461792e-01 -1.21663439e+00
6.89181030e-01 3.38274390e-01 7.58255184e-01 -6.37784660e-01
-2.66886711e-01 -1.68170147e-02 -1.49080172e-01 -6.74981534e-01
2.49101296e-01 -1.84931502e-01 -6.78812563e-02 9.46612433e-02
3.99315119e-01 6.37228191e-02 2.48406351e-01 2.86135018e-01
8.61559272e-01 5.80607235e-01 4.50042903e-01 2.61786133e-02
6.08267188e-01 1.05179138e-01 9.27171826e-01 4.62722361e-01
-4.43365067e-01 4.69033569e-01 1.79424420e-01 -4.82641906e-01
-1.15517628e+00 -1.20279372e+00 8.71152729e-02 1.13860035e+00
3.14937770e-01 -2.94273704e-01 -5.52026272e-01 -1.01206541e+00
9.53844341e-04 4.37916756e-01 -7.21500158e-01 -3.29309046e-01
-6.48767054e-01 -3.64708006e-01 7.89044574e-02 6.73851907e-01
5.95662177e-01 -8.49508464e-01 -1.59543544e-01 3.03909242e-01
-3.31795931e-01 -1.23986304e+00 -8.58637512e-01 -2.37711787e-01
-1.01699853e+00 -1.03486764e+00 -9.86676395e-01 -8.72003794e-01
6.07857049e-01 5.00993192e-01 8.99210870e-01 -5.42995453e-01
2.29770258e-01 7.01475680e-01 -5.08869648e-01 -2.57868260e-01
-4.67098862e-01 -3.36053669e-02 5.29243827e-01 3.78912807e-01
6.65177941e-01 -5.14384568e-01 -4.36180264e-01 7.19093978e-01
-1.11646700e+00 -2.13674903e-01 3.84619772e-01 8.43823552e-01
7.69491851e-01 2.03843750e-02 7.77026474e-01 -7.98858225e-01
2.71183729e-01 -8.95588219e-01 -3.10415059e-01 1.19909130e-01
-4.94859368e-01 -1.48920029e-01 7.80590713e-01 -1.04587877e+00
-1.18585443e+00 2.96935141e-01 3.84905130e-01 -9.74377692e-01
-3.89715403e-01 3.74052763e-01 -3.38458836e-01 1.63687006e-01
6.87142015e-01 3.34111303e-01 1.29689440e-01 -5.09380817e-01
3.23197216e-01 4.42161411e-01 6.70062006e-01 -3.96918386e-01
9.60893393e-01 5.72360635e-01 -2.19843790e-01 -9.18314517e-01
-7.58653283e-01 -7.12076187e-01 -1.00977135e+00 -1.77185118e-01
7.26431370e-01 -1.12283075e+00 1.26657605e-01 4.53035861e-01
-6.06634974e-01 -4.57937658e-01 -2.53825963e-01 7.29034066e-01
-7.49001265e-01 6.40977859e-01 -1.56442374e-01 -3.29759270e-01
3.88551652e-01 -8.42015684e-01 4.97522593e-01 1.59445569e-01
-2.03076437e-01 -1.23743927e+00 1.64443240e-01 -1.70829594e-02
2.20453769e-01 2.59675264e-01 7.14588284e-01 -7.25546718e-01
-3.16834569e-01 4.06462997e-02 1.90520346e-01 5.99519968e-01
7.22233593e-01 -1.60470709e-01 -6.07402861e-01 -3.99867624e-01
-2.26811180e-03 -2.05556974e-01 6.32621408e-01 5.09226203e-01
1.18940449e+00 -4.87542659e-01 -2.72378474e-01 5.42712808e-01
1.29464948e+00 3.80492330e-01 4.99450058e-01 6.13367856e-01
7.04639316e-01 4.14531112e-01 9.44192111e-01 4.89814997e-01
3.77790332e-01 9.38186228e-01 2.64940023e-01 1.68778986e-01
-2.38287136e-01 -5.03753841e-01 6.33608878e-01 6.60292685e-01
-5.40659130e-02 -1.73816785e-01 -6.55921757e-01 1.00718021e+00
-2.04235578e+00 -1.38649094e+00 2.34953269e-01 2.30531549e+00
8.29036236e-01 -4.69252840e-02 6.38564169e-01 -1.55157298e-01
7.69640744e-01 1.66114092e-01 -9.30761695e-01 -2.89707072e-02
-1.60631776e-01 -1.56458884e-01 2.76139855e-01 5.02739567e-03
-1.38806999e+00 8.83041084e-01 6.39970684e+00 7.66710162e-01
-1.06796896e+00 1.00461021e-01 2.73244947e-01 -2.64572322e-01
3.46775129e-02 -2.39994615e-01 -5.07418931e-01 7.30507672e-01
8.38926852e-01 -2.43318692e-01 4.00032192e-01 1.05000997e+00
3.27137083e-01 1.50537476e-01 -1.26412606e+00 1.07083893e+00
1.56976923e-01 -1.10383260e+00 2.35805571e-01 -5.28878607e-02
1.10037816e+00 -1.15457267e-01 1.57060727e-01 4.19147432e-01
6.44311979e-02 -5.04547715e-01 5.18406689e-01 3.19222450e-01
8.03675354e-01 -6.40753031e-01 4.17931378e-01 2.92665333e-01
-1.17015707e+00 -3.02181512e-01 -4.48260278e-01 4.96123731e-02
8.87861252e-02 4.15151976e-02 -7.03800440e-01 3.51352781e-01
8.51704955e-01 1.22812998e+00 -3.60220551e-01 9.63753641e-01
1.31980171e-02 7.65983999e-01 -4.67293076e-02 4.43953276e-01
3.04957956e-01 -1.61688268e-01 5.65926015e-01 1.15177751e+00
4.35681790e-01 1.81647256e-01 4.14438367e-01 2.94361383e-01
-6.86973184e-02 1.04464248e-01 -9.60541904e-01 1.06168792e-01
5.81251144e-01 6.79229975e-01 -3.62961352e-01 -4.85340804e-01
-8.81823957e-01 1.22186732e+00 8.45293105e-02 6.09303534e-01
-8.89983594e-01 -1.05069596e-02 9.07400608e-01 3.53404790e-01
6.38134539e-01 -3.23361814e-01 9.84148979e-02 -1.42255819e+00
1.23458482e-01 -1.03569722e+00 7.27002561e-01 -6.04969144e-01
-1.53576946e+00 4.33820009e-01 3.40892285e-01 -1.85289955e+00
-5.75115681e-01 -3.62321228e-01 -2.56188124e-01 5.41018367e-01
-1.53900409e+00 -1.01265645e+00 -2.97636330e-01 1.19987571e+00
1.08460736e+00 -5.37166178e-01 5.78044832e-01 2.15617537e-01
-3.46599787e-01 5.70249200e-01 4.52270001e-01 2.36804709e-01
1.14707243e+00 -1.01472843e+00 1.88598692e-01 1.07813239e+00
8.58712643e-02 3.79974365e-01 6.32874966e-01 -6.42618775e-01
-1.26961362e+00 -1.41614246e+00 5.16606748e-01 -5.23183346e-01
7.86584556e-01 -3.28490958e-02 -1.25851595e+00 9.02601123e-01
1.86506748e-01 1.82777762e-01 6.96719527e-01 -2.24423841e-01
-4.67806906e-01 -1.69504017e-01 -1.16517413e+00 6.63158894e-01
1.25497389e+00 -4.70733017e-01 -7.12280929e-01 2.34749824e-01
6.16213381e-01 -2.83742726e-01 -9.27062988e-01 2.35546723e-01
3.59635860e-01 -7.83654392e-01 1.08637452e+00 -8.99452925e-01
3.63210499e-01 -5.54635823e-01 -2.28638694e-01 -1.58152044e+00
-7.01206744e-01 -3.27355891e-01 -4.89400864e-01 1.25526929e+00
2.46285107e-02 -4.49822843e-01 5.94014943e-01 5.86292744e-01
6.46874914e-03 -2.50239551e-01 -8.58932257e-01 -1.29697883e+00
8.43027458e-02 -2.60508806e-01 6.21152401e-01 1.24729180e+00
-1.00703940e-01 1.79809734e-01 -7.59118915e-01 2.87498862e-01
4.55345690e-01 -3.68172787e-02 9.94975150e-01 -9.78345990e-01
-2.64114618e-01 -2.18442664e-01 -4.95941728e-01 -1.10776305e+00
2.78951645e-01 -4.46057260e-01 -5.00969589e-02 -1.19988656e+00
1.69743598e-01 -2.44509339e-01 -5.13707161e-01 3.66431564e-01
-1.47919938e-01 3.09220314e-01 2.58914411e-01 5.07527828e-01
-5.60060740e-01 5.83686173e-01 1.08448768e+00 -2.00663164e-01
-3.65458220e-01 -4.09808382e-02 -5.21965504e-01 8.18575799e-01
7.54252374e-01 -3.70192081e-01 -8.53001118e-01 -5.11319995e-01
-3.57551515e-01 -1.84794933e-01 3.07288170e-01 -1.03406549e+00
-1.02868877e-01 -6.82048023e-01 4.60833877e-01 -1.96436882e-01
2.57670820e-01 -1.10673165e+00 2.17007771e-01 1.01343133e-01
-3.38510007e-01 5.01818918e-02 1.40436649e-01 1.07861733e+00
-4.27343696e-01 -8.78682174e-03 9.66337323e-01 -2.53675491e-01
-1.43462729e+00 5.93062520e-01 -4.47856784e-01 1.14551432e-01
1.26046777e+00 -5.34237802e-01 7.55428597e-02 -5.03787458e-01
-7.96696723e-01 9.74561274e-02 9.14182603e-01 7.99436927e-01
5.53772628e-01 -1.77710998e+00 -6.14145875e-01 2.24296123e-01
5.51033258e-01 -5.78137264e-02 4.71340299e-01 5.56257248e-01
-1.19720362e-02 2.07793355e-01 -3.87673646e-01 -7.70442486e-01
-1.13407552e+00 1.22670317e+00 9.54583213e-02 7.04183988e-03
-6.80305660e-01 5.03683746e-01 4.73273218e-01 -1.83989465e-01
2.25530013e-01 -2.01720044e-01 -2.05108896e-01 -4.17071246e-02
6.40897989e-01 2.98428327e-01 -2.72159398e-01 -8.93704534e-01
-3.93838763e-01 6.70530438e-01 -8.10636729e-02 3.50221619e-02
1.22672272e+00 -4.60042179e-01 5.10754168e-01 5.92444897e-01
1.17740500e+00 -1.25981897e-01 -1.80895305e+00 -8.05113494e-01
-1.05789222e-01 -8.94941151e-01 -2.41112143e-01 -6.94710612e-01
-1.02474689e+00 6.29837036e-01 5.45452356e-01 -5.68187460e-02
1.46279979e+00 6.35515992e-03 6.12964392e-01 9.09522548e-02
4.08966392e-01 -1.35567343e+00 4.30543274e-01 3.65754783e-01
8.38683903e-01 -1.44562685e+00 -8.26147497e-02 -8.83428082e-02
-1.20352316e+00 8.49959612e-01 8.56180489e-01 -1.08315282e-01
5.63089788e-01 -2.66856194e-01 -1.90334804e-02 3.36678207e-01
-6.58392251e-01 -9.56890583e-02 2.69593924e-01 1.20062697e+00
1.38679752e-02 -2.11663112e-01 -2.22572200e-02 4.35560465e-01
4.32095319e-01 2.26209193e-01 6.16095245e-01 8.99125755e-01
-2.06078440e-01 -1.17780578e+00 -4.35443193e-01 2.74026006e-01
-3.14443529e-01 2.68918216e-01 -1.43970981e-01 1.08229446e+00
2.23706841e-01 6.87153339e-01 1.18809856e-01 -2.45274574e-01
3.99496764e-01 2.09151268e-01 4.52809274e-01 -6.25870109e-01
1.11233937e-02 1.77526504e-01 -2.64240533e-01 -5.26000798e-01
-9.37066913e-01 -1.09484148e+00 -1.03789067e+00 -1.51254028e-01
1.24584869e-01 1.53178060e-02 2.75163770e-01 8.36048722e-01
6.21597111e-01 1.55439809e-01 8.19810867e-01 -8.87839496e-01
-3.49009335e-01 -8.83827865e-01 -6.43098295e-01 9.04364109e-01
5.08915424e-01 -1.05070162e+00 -1.80869967e-01 7.09127307e-01] | [8.870343208312988, 0.983022153377533] |
d6a0af82-7ae4-4e2f-8b72-72435363c45d | self-supervised-contrastive-representation-1 | 2208.06616 | null | https://arxiv.org/abs/2208.06616v2 | https://arxiv.org/pdf/2208.06616v2.pdf | Self-supervised Contrastive Representation Learning for Semi-supervised Time-Series Classification | Learning time-series representations when only unlabeled data or few labeled samples are available can be a challenging task. Recently, contrastive self-supervised learning has shown great improvement in extracting useful representations from unlabeled data via contrasting different augmented views of data. In this work, we propose a novel Time-Series representation learning framework via Temporal and Contextual Contrasting (TS-TCC) that learns representations from unlabeled data with contrastive learning. Specifically, we propose time-series specific weak and strong augmentations and use their views to learn robust temporal relations in the proposed temporal contrasting module, besides learning discriminative representations by our proposed contextual contrasting module. Additionally, we conduct a systematic study of time-series data augmentation selection, which is a key part of contrastive learning. We also extend TS-TCC to the semi-supervised learning settings and propose a Class-Aware TS-TCC (CA-TCC) that benefits from the available few labeled data to further improve representations learned by TS-TCC. Specifically, we leverage robust pseudo labels produced by TS-TCC to realize class-aware contrastive loss. Extensive experiments show that the linear evaluation of the features learned by our proposed framework performs comparably with the fully supervised training. Additionally, our framework shows high efficiency in few labeled data and transfer learning scenarios. The code is publicly available at \url{https://github.com/emadeldeen24/CA-TCC}. | ['Cuntai Guan', 'XiaoLi Li', 'Chee-Keong Kwoh', 'Min Wu', 'Zhenghua Chen', 'Mohamed Ragab', 'Emadeldeen Eldele'] | 2022-08-13 | null | null | null | null | ['semi-supervised-time-series-classification'] | ['time-series'] | [ 2.71059811e-01 -2.74638146e-01 -4.02538627e-01 -6.73319757e-01
-1.09184587e+00 -6.23539031e-01 7.99869716e-01 -7.69890994e-02
-1.95066109e-01 5.95768511e-01 1.62854403e-01 -5.46969958e-02
-2.46008545e-01 -5.14022768e-01 -6.12878740e-01 -8.11144650e-01
-2.55639791e-01 5.68935014e-02 -1.35663792e-01 -1.13079615e-01
-9.82635096e-02 3.72944504e-01 -1.48852754e+00 2.25124627e-01
9.41956103e-01 1.14313948e+00 2.14747507e-02 2.98505455e-01
6.70326352e-02 7.25507379e-01 -1.21445730e-01 1.19460151e-01
4.27994579e-01 -5.03700018e-01 -5.84662497e-01 1.56720117e-01
2.22490206e-01 -1.79276004e-01 -3.95256937e-01 5.91213644e-01
3.69062304e-01 4.25458431e-01 6.99045360e-01 -1.50643504e+00
-6.55945063e-01 2.02092662e-01 -9.35580194e-01 5.13531148e-01
2.93369088e-02 1.56699434e-01 8.73742282e-01 -1.13870418e+00
4.49757457e-01 9.62004781e-01 6.03726566e-01 5.45522928e-01
-1.18962622e+00 -9.17344928e-01 6.25720620e-01 4.60702062e-01
-1.15513003e+00 -3.29400539e-01 1.25227976e+00 -3.34723204e-01
6.21249616e-01 2.18939528e-01 4.45723087e-01 1.23820400e+00
-3.44606727e-01 1.06087637e+00 1.56522703e+00 -3.65080565e-01
2.39308923e-01 2.38522794e-02 4.16075975e-01 6.19137287e-01
-3.15232724e-01 3.33771765e-01 -4.66433436e-01 -4.34382306e-03
6.68585420e-01 5.97766161e-01 -1.51533112e-01 -5.13041079e-01
-1.26486778e+00 6.49367332e-01 5.19278944e-01 3.30133647e-01
-2.78252453e-01 -5.70584685e-02 7.21301079e-01 6.14432275e-01
8.27519476e-01 2.31453478e-02 -7.32664466e-01 3.29859480e-02
-7.86188006e-01 -1.41749889e-01 1.87916949e-01 9.71790433e-01
8.15215170e-01 2.92445093e-01 -2.82542557e-01 9.46884692e-01
-4.75214235e-02 4.27228570e-01 8.47687006e-01 -7.13977933e-01
5.76046407e-01 5.24153829e-01 -1.70254022e-01 -5.70111871e-01
-3.85184199e-01 -6.76821291e-01 -8.62581372e-01 4.63600765e-04
1.76658228e-01 -2.02394679e-01 -9.68479633e-01 2.10690022e+00
3.07776630e-01 6.91311002e-01 1.12496905e-01 7.49345601e-01
6.01281703e-01 7.26528704e-01 1.13863915e-01 -7.60846257e-01
1.03921807e+00 -1.18116081e+00 -6.77238345e-01 -2.33407295e-03
7.28398263e-01 -4.60364401e-01 1.15922189e+00 1.82159126e-01
-7.63891339e-01 -7.19798028e-01 -1.08894873e+00 7.39654303e-02
-3.99442405e-01 4.43580866e-01 5.91874778e-01 1.83199704e-01
-5.69632471e-01 7.01220453e-01 -1.14557683e+00 -2.94556618e-01
5.34939468e-01 2.64477760e-01 -5.00412345e-01 -8.56757164e-02
-1.12580454e+00 5.74365437e-01 2.87286252e-01 1.59885809e-01
-9.10925806e-01 -7.13342547e-01 -1.01037014e+00 -1.90477461e-01
4.47462440e-01 -1.26204506e-01 1.19776881e+00 -9.91387844e-01
-1.26227486e+00 7.55274177e-01 -3.34693551e-01 -3.48452747e-01
4.88228440e-01 -3.01455140e-01 -6.72578812e-01 3.56285781e-01
2.63853699e-01 4.84865725e-01 9.15270507e-01 -1.24622250e+00
-4.32484120e-01 -5.20063519e-01 -1.23780258e-01 1.73201486e-01
-6.15278542e-01 -3.34157050e-01 -2.67017275e-01 -1.09179533e+00
3.04373264e-01 -1.03977275e+00 -3.08641881e-01 8.04274231e-02
-1.33814812e-01 -2.82543629e-01 1.27166736e+00 -5.83269775e-01
1.08443344e+00 -2.22706819e+00 -1.12159021e-01 6.95364773e-02
6.12041075e-03 2.32461870e-01 -2.81143457e-01 5.30631781e-01
-5.75804293e-01 -1.54436618e-01 -4.84996676e-01 -3.36121917e-01
-2.40852728e-01 1.66842327e-01 -4.61529434e-01 5.41467607e-01
4.95281786e-01 8.39901984e-01 -1.08404386e+00 -3.79073471e-01
4.49219823e-01 3.54204565e-01 -2.44908229e-01 3.15168947e-01
-2.48394180e-02 9.29537594e-01 -6.57041430e-01 6.89939260e-01
6.77008569e-01 -3.94299448e-01 1.41163722e-01 -3.21429521e-01
-6.52026981e-02 1.07181013e-01 -7.65736639e-01 1.71747613e+00
-6.86522722e-01 4.45898563e-01 -5.35476148e-01 -1.52311254e+00
1.09124601e+00 4.84443784e-01 7.24424303e-01 -9.38011348e-01
-1.97921637e-02 2.19150737e-01 -4.48358059e-01 -5.55007815e-01
5.96909667e-04 -2.25950629e-01 -5.12033887e-03 5.39072096e-01
2.79736102e-01 3.63691114e-02 1.45850420e-01 8.57273340e-02
8.32427561e-01 7.62199461e-01 3.72538954e-01 2.89385393e-02
5.78954756e-01 -1.69446960e-01 9.25728977e-01 2.38732085e-01
-4.95864868e-01 6.64848149e-01 1.81987092e-01 -4.69802111e-01
-8.51047099e-01 -1.01600814e+00 -1.14450864e-01 1.05891681e+00
5.89132495e-02 -2.84804016e-01 -8.28923956e-02 -1.18955135e+00
-2.25946948e-01 5.73973238e-01 -8.33858371e-01 -2.89433151e-01
-7.15596855e-01 -6.43738091e-01 2.41015516e-02 9.77625787e-01
5.54941118e-01 -9.87526000e-01 -3.35668325e-01 7.91787356e-02
-8.84919092e-02 -9.41655517e-01 -6.46051764e-01 5.40385962e-01
-1.30156422e+00 -9.37036455e-01 -8.25960875e-01 -9.38318014e-01
8.30577433e-01 6.04939103e-01 6.19833648e-01 -3.65573943e-01
-1.94999486e-01 5.45603395e-01 -5.63909292e-01 -2.67589092e-01
1.78422719e-01 -1.15272962e-01 8.12032893e-02 2.35013813e-01
3.26369852e-01 -9.97936785e-01 -7.06194878e-01 4.35183793e-01
-8.55649352e-01 -6.26398399e-02 4.37730283e-01 1.13850701e+00
8.62250388e-01 -3.00660670e-01 1.04442585e+00 -1.01267910e+00
2.05543011e-01 -8.29542398e-01 -4.20372158e-01 2.07076594e-01
-7.15168178e-01 1.29674256e-01 8.70782733e-01 -7.06562400e-01
-1.24571908e+00 1.88858077e-01 1.69957161e-01 -1.01253438e+00
-3.18179727e-02 5.96252084e-01 1.13683164e-01 2.36921668e-01
5.83108544e-01 3.66879314e-01 5.26794158e-02 -5.62251687e-01
4.99934167e-01 3.66798073e-01 4.55265939e-01 -7.89890945e-01
1.03249371e+00 7.27782965e-01 -1.48495600e-01 -4.04696345e-01
-1.08240032e+00 -6.95296824e-01 -7.34874666e-01 -1.12609759e-01
3.65410179e-01 -1.13736773e+00 -2.68818617e-01 1.99928403e-01
-5.20477891e-01 -3.08482766e-01 -5.87823689e-01 7.47580767e-01
-6.78196073e-01 4.39250857e-01 -5.29181421e-01 -8.46217453e-01
-3.05633813e-01 -7.80198097e-01 1.04347992e+00 2.58510530e-01
3.02674361e-02 -1.12823808e+00 2.56965250e-01 2.60542929e-01
1.89878643e-01 5.62548399e-01 5.78537464e-01 -9.10565794e-01
-2.43088081e-01 -1.36924043e-01 -1.38569295e-01 4.82633740e-01
5.46891212e-01 -2.28705481e-01 -1.29929423e+00 -6.26183510e-01
8.86849985e-02 -7.05909669e-01 1.01689219e+00 3.79093856e-01
1.52072608e+00 -1.72398180e-01 -3.15393925e-01 6.05384588e-01
1.32388604e+00 3.97619843e-01 2.56495923e-01 3.31976712e-01
6.92241728e-01 8.08517873e-01 1.11913872e+00 6.25517845e-01
2.96421081e-01 5.29915035e-01 1.65872738e-01 -1.91060096e-01
3.56661119e-02 -3.76405895e-01 3.22863728e-01 1.28089833e+00
-1.80277988e-01 1.58755824e-01 -6.32304668e-01 7.55857050e-01
-1.94232988e+00 -1.10845923e+00 1.68279588e-01 2.15820408e+00
8.14646006e-01 1.69262979e-02 1.20035268e-01 3.76361132e-01
7.24634826e-01 4.75038230e-01 -9.14854050e-01 -1.61769167e-02
1.77442543e-02 3.29767615e-01 1.54813319e-01 -2.75720041e-02
-1.49505043e+00 5.87607086e-01 5.42272186e+00 8.45207989e-01
-1.36396420e+00 2.78202266e-01 7.29820549e-01 -1.62786186e-01
-3.70741695e-01 4.73500957e-04 -2.94369549e-01 4.07460392e-01
8.85916591e-01 -5.03832817e-01 -3.10025341e-03 7.95745492e-01
4.27213162e-01 4.06581521e-01 -1.22010076e+00 1.06770349e+00
1.15518679e-03 -1.06587243e+00 -7.12567344e-02 -1.89948514e-01
8.28690588e-01 -2.04696774e-01 4.04496640e-01 7.57123828e-01
6.58982247e-02 -5.91692328e-01 5.66438019e-01 3.59603018e-01
8.65127146e-01 -5.43859184e-01 4.48622853e-01 1.43724248e-01
-1.60939169e+00 -2.42593348e-01 9.91997309e-03 9.87333134e-02
-4.26076502e-02 3.81499380e-01 -5.54982245e-01 9.00444508e-01
5.89625359e-01 1.45848095e+00 -5.81833422e-01 8.07800651e-01
-8.29183236e-02 8.96111012e-01 -4.18005325e-02 4.24726963e-01
2.25128382e-01 -2.70658046e-01 3.70683581e-01 8.89872611e-01
3.24411780e-01 2.25658178e-01 3.63785923e-01 5.93998671e-01
-2.35000849e-02 1.53196156e-01 -6.54533565e-01 -9.50967707e-03
5.94586134e-01 1.22620332e+00 -5.20568132e-01 -3.53946507e-01
-6.00551367e-01 8.25077713e-01 4.66362208e-01 5.39016306e-01
-8.47906709e-01 -6.60521761e-02 2.25823671e-01 -2.44741857e-01
3.21881741e-01 2.04865239e-03 2.93250158e-02 -1.42513430e+00
2.77247727e-01 -7.87542880e-01 8.75004351e-01 -6.73894584e-01
-1.67488623e+00 6.04106426e-01 2.43859619e-01 -2.12591004e+00
-2.96105713e-01 -3.14621359e-01 -7.60845184e-01 5.65391779e-01
-1.72489774e+00 -1.41307354e+00 -3.13464344e-01 8.47072184e-01
9.04805660e-01 -2.17096820e-01 6.51268721e-01 3.08165014e-01
-7.54441500e-01 6.84917986e-01 2.02364996e-01 1.56206805e-02
8.29732955e-01 -1.16941679e+00 1.71406135e-01 8.20136011e-01
1.09779648e-01 6.49193823e-01 2.25458592e-01 -3.16899002e-01
-1.17836440e+00 -1.47296047e+00 3.47811013e-01 -3.09486315e-03
5.69716990e-01 -1.65376469e-01 -1.11040795e+00 9.34861183e-01
7.58542940e-02 6.31007671e-01 8.11474264e-01 2.83254143e-02
-7.91875005e-01 -4.34508562e-01 -1.16001534e+00 4.22390103e-01
1.00251949e+00 -8.85388732e-01 -7.72971511e-01 3.67616594e-01
8.20302248e-01 -1.61824480e-01 -8.15394282e-01 6.49995685e-01
5.19120216e-01 -6.45295143e-01 8.54302585e-01 -5.11521459e-01
3.08298707e-01 -2.95510471e-01 -2.44461857e-02 -1.38700664e+00
-1.78012311e-01 -4.46024269e-01 -2.08164901e-01 1.24134648e+00
2.29653224e-01 -7.60557711e-01 7.27742314e-01 2.12619215e-01
-2.36899346e-01 -8.40134442e-01 -6.94826841e-01 -1.13332474e+00
1.07579134e-01 -2.64389157e-01 3.47182959e-01 1.54185283e+00
6.74178749e-02 1.71194479e-01 -5.28276920e-01 4.64147441e-02
6.67037725e-01 4.87631530e-01 3.86304975e-01 -1.06005561e+00
-1.98916271e-01 4.57620202e-03 -2.20245555e-01 -8.86811376e-01
2.12138489e-01 -9.94564056e-01 -1.20785587e-01 -1.19312513e+00
2.74257421e-01 -5.97255707e-01 -8.89869928e-01 7.56993890e-01
-3.49505156e-01 2.03439787e-01 6.45246580e-02 4.95308399e-01
-5.50963283e-01 9.57169533e-01 1.25242710e+00 -2.23097160e-01
-2.65171468e-01 9.79136899e-02 -4.17433649e-01 4.70305443e-01
9.12942111e-01 -3.41424495e-01 -8.99918377e-01 -1.44596949e-01
-4.52307522e-01 2.42055237e-01 2.47311547e-01 -8.52717102e-01
-2.43736571e-03 -9.33615863e-02 3.87058496e-01 -8.03397775e-01
4.06172842e-01 -7.92936444e-01 -5.20329773e-02 3.05597752e-01
-6.83218777e-01 -1.07430192e-02 2.78169185e-01 8.82841647e-01
-5.76744854e-01 8.12716782e-02 7.43569791e-01 -8.81544650e-02
-9.02888775e-01 6.57311261e-01 1.29974306e-01 1.49377912e-01
1.14489508e+00 -9.20878425e-02 -2.78747201e-01 -3.29311907e-01
-1.01781762e+00 4.83663112e-01 5.34179062e-02 4.97821242e-01
6.55162394e-01 -1.74298859e+00 -4.61146683e-01 1.56983852e-01
5.92245638e-01 -2.62631804e-01 5.20357788e-01 9.42113817e-01
1.12529345e-01 1.60955802e-01 -3.57945740e-01 -6.77709818e-01
-1.07286167e+00 8.80881250e-01 7.70799965e-02 -2.71018296e-01
-7.39844263e-01 5.47053814e-01 4.84393716e-01 -5.65832436e-01
8.67801979e-02 -2.87055254e-01 -4.04503584e-01 2.40432203e-01
4.11456734e-01 3.49054694e-01 8.04539025e-02 -5.00748694e-01
-3.28060657e-01 6.27458692e-01 -3.16487402e-01 -1.03014834e-01
1.65947998e+00 -1.20882154e-01 3.80400687e-01 8.07729840e-01
1.63399076e+00 -3.11046720e-01 -1.55362201e+00 -6.65444970e-01
8.97786319e-02 -3.35062891e-01 -1.24923036e-01 -5.27782440e-01
-1.26833296e+00 9.39812303e-01 8.96291256e-01 -1.33708939e-01
1.57091129e+00 -2.40563512e-01 6.39587939e-01 2.03756660e-01
3.42415482e-01 -8.81171227e-01 5.14997542e-01 2.09769845e-01
9.46878195e-01 -1.60225654e+00 -7.52656981e-02 -4.03526276e-01
-7.82225013e-01 9.81865108e-01 8.38778257e-01 -2.51350194e-01
7.18364596e-01 -1.15628183e-01 1.28308505e-01 -1.63584605e-01
-1.06784952e+00 -3.71971935e-01 3.83222848e-01 5.43629229e-01
5.01767755e-01 -1.58889145e-02 -3.18316609e-01 5.24288177e-01
3.35989177e-01 -2.50076912e-02 1.82871759e-01 1.36670029e+00
1.21421039e-01 -1.18723011e+00 3.13815288e-03 4.79838222e-01
-1.88381582e-01 1.31346881e-01 -3.34183406e-03 8.15948963e-01
-2.80123185e-02 8.17781270e-01 -3.95164713e-02 -4.28397447e-01
3.98529112e-01 1.37212992e-01 3.43658864e-01 -6.51277363e-01
-4.90685105e-01 3.55761886e-01 -1.00562513e-01 -5.70286274e-01
-9.25858498e-01 -7.47626126e-01 -1.25472236e+00 2.53742576e-01
-2.97446996e-01 1.74434707e-01 1.61638409e-01 8.82262230e-01
2.60518163e-01 5.37038863e-01 1.41442525e+00 -8.48424554e-01
-6.26688361e-01 -9.01827812e-01 -5.58969498e-01 6.76406622e-01
5.91450453e-01 -8.76547098e-01 -5.64977705e-01 2.99772412e-01] | [7.41714334487915, 2.939249038696289] |
4cd305f9-5986-422b-80de-fe9378d97cf8 | learning-differentiable-logic-programs-for | 2307.00928 | null | https://arxiv.org/abs/2307.00928v1 | https://arxiv.org/pdf/2307.00928v1.pdf | Learning Differentiable Logic Programs for Abstract Visual Reasoning | Visual reasoning is essential for building intelligent agents that understand the world and perform problem-solving beyond perception. Differentiable forward reasoning has been developed to integrate reasoning with gradient-based machine learning paradigms. However, due to the memory intensity, most existing approaches do not bring the best of the expressivity of first-order logic, excluding a crucial ability to solve abstract visual reasoning, where agents need to perform reasoning by using analogies on abstract concepts in different scenarios. To overcome this problem, we propose NEUro-symbolic Message-pAssiNg reasoNer (NEUMANN), which is a graph-based differentiable forward reasoner, passing messages in a memory-efficient manner and handling structured programs with functors. Moreover, we propose a computationally-efficient structure learning algorithm to perform explanatory program induction on complex visual scenes. To evaluate, in addition to conventional visual reasoning tasks, we propose a new task, visual reasoning behind-the-scenes, where agents need to learn abstract programs and then answer queries by imagining scenes that are not observed. We empirically demonstrate that NEUMANN solves visual reasoning tasks efficiently, outperforming neural, symbolic, and neuro-symbolic baselines. | ['Kristian Kersting', 'Devendra Singh Dhami', 'Viktor Pfanschilling', 'Hikaru Shindo'] | 2023-07-03 | null | null | null | null | ['program-induction', 'visual-reasoning', 'visual-reasoning'] | ['computer-code', 'computer-vision', 'reasoning'] | [ 1.22519895e-01 3.02936256e-01 2.52630692e-02 -3.45124841e-01
-5.53682968e-02 -6.61849499e-01 9.02216315e-01 2.90550172e-01
-3.13100576e-01 2.66640931e-01 4.61340025e-02 -9.56540704e-01
-2.21443340e-01 -1.19203258e+00 -9.53713894e-01 -2.73633122e-01
-1.55926615e-01 7.11125672e-01 2.16495574e-01 -3.41362208e-01
5.26339352e-01 5.16146183e-01 -1.68553817e+00 8.59137177e-01
1.02525449e+00 6.96975946e-01 2.87144303e-01 7.66283810e-01
-6.09633327e-01 1.91720665e+00 -2.40020379e-01 -5.76842189e-01
-1.09058253e-01 -6.13147199e-01 -1.17016065e+00 -3.05425823e-01
5.63225448e-01 -4.06294167e-01 -2.30936989e-01 1.23369598e+00
-4.26645041e-01 3.20988089e-01 5.97543061e-01 -1.52625692e+00
-1.12955177e+00 7.47531414e-01 -7.16055632e-02 4.14499901e-02
7.92100012e-01 6.31358922e-01 1.23347974e+00 -7.14769721e-01
6.57308877e-01 1.77574182e+00 4.46552098e-01 6.91817284e-01
-1.59198213e+00 -1.38534769e-01 6.69226110e-01 6.78256094e-01
-7.94075549e-01 -2.07242966e-02 6.90224767e-01 -5.71008086e-01
1.31176889e+00 2.96436042e-01 1.04216444e+00 8.07754517e-01
-2.38481224e-01 1.01938057e+00 1.18221569e+00 -3.62032562e-01
5.56533039e-01 2.10674815e-02 3.15566212e-01 1.54473650e+00
7.08702207e-02 3.14694643e-01 -4.75894570e-01 2.67682783e-03
5.49075961e-01 9.43254828e-02 -2.43815035e-01 -7.17169285e-01
-1.28370690e+00 9.90189373e-01 9.98126388e-01 6.27030283e-02
-3.06199849e-01 4.79918391e-01 4.82166827e-01 6.13404453e-01
-2.59370357e-01 7.07454681e-01 -3.08148772e-01 1.68939725e-01
-5.05832493e-01 3.48633885e-01 1.21040475e+00 7.69310534e-01
8.15051317e-01 2.28034616e-01 -2.14047611e-01 1.64184123e-01
5.57544827e-01 4.81489152e-01 2.29893371e-01 -1.36341655e+00
4.68925744e-01 1.14449763e+00 -2.32096761e-01 -9.55162585e-01
-4.24554765e-01 8.48280732e-04 -5.15856445e-01 7.61687160e-01
5.39980710e-01 2.84163982e-01 -8.10779631e-01 1.73424864e+00
1.73843250e-01 4.18431722e-02 4.42745686e-01 9.07006085e-01
9.01060522e-01 7.11679816e-01 2.83108532e-01 1.49747971e-02
1.40850639e+00 -1.29977810e+00 -2.69312888e-01 -3.63925725e-01
9.60033596e-01 1.99285403e-01 1.73340571e+00 3.32706392e-01
-1.17605507e+00 -3.95563036e-01 -9.02215838e-01 -4.08871561e-01
-7.09693909e-01 -3.65759760e-01 1.23098946e+00 4.15529549e-01
-1.20768487e+00 4.05289233e-01 -6.48009062e-01 -2.57803530e-01
6.54721558e-01 8.25898200e-02 -1.35586902e-01 3.20853032e-02
-8.19242001e-01 1.11963189e+00 5.33118784e-01 -7.26513192e-02
-1.26399314e+00 -8.44239473e-01 -1.20709920e+00 3.80681098e-01
6.21505558e-01 -1.25905633e+00 1.41719496e+00 -1.21314013e+00
-1.59429038e+00 1.06587875e+00 -2.52691448e-01 -7.88300693e-01
4.83076632e-01 7.46044666e-02 3.43048461e-02 3.14922333e-01
-9.90718901e-02 6.21259391e-01 5.25735974e-01 -1.40854740e+00
-3.89507025e-01 -4.09465015e-01 1.00154400e+00 1.49554521e-01
3.33927095e-01 -3.90517622e-01 1.42234966e-01 -3.61433788e-03
2.39005849e-01 -7.90750623e-01 -2.95957118e-01 2.86948651e-01
-2.46658623e-01 -4.21386719e-01 5.26731610e-01 -5.22563994e-01
4.46072012e-01 -2.10233593e+00 5.12201250e-01 1.93032488e-01
5.44309139e-01 7.55708739e-02 -1.42056093e-01 2.91862875e-01
1.62742019e-01 -2.98455767e-02 -2.56847322e-01 8.14392641e-02
5.37757695e-01 4.57220137e-01 -7.02449858e-01 5.94386505e-03
6.10769056e-02 1.50139630e+00 -1.08202279e+00 -3.64347458e-01
2.89848655e-01 5.86826354e-02 -1.02490926e+00 2.62538582e-01
-1.14196897e+00 3.09571326e-01 -2.82295346e-01 3.70578736e-01
3.67547423e-01 -6.24925196e-01 2.85091966e-01 1.03750765e-01
1.79528818e-02 3.78130406e-01 -8.76909196e-01 1.83565414e+00
-6.78379655e-01 5.85488856e-01 2.91712899e-02 -1.36748159e+00
5.23557544e-01 -8.53622034e-02 -4.32954282e-01 -9.50922608e-01
-8.53147805e-02 2.16618124e-02 8.23910290e-04 -8.14858794e-01
-1.37466220e-02 -2.52044857e-01 4.14828658e-02 2.32898831e-01
-1.10303678e-01 -6.91831231e-01 4.45026755e-01 4.93698508e-01
1.07564914e+00 4.76808161e-01 2.54313499e-01 4.93235216e-02
7.64959097e-01 4.84821260e-01 -1.52985461e-03 1.04079008e+00
3.62522937e-02 -1.56900346e-01 9.23310280e-01 -8.23061764e-01
-6.00000501e-01 -1.43330729e+00 7.56268799e-01 1.58787537e+00
1.25492036e-01 -3.34615350e-01 -7.23855674e-01 -5.49127460e-01
6.10761065e-03 1.21504343e+00 -5.77155113e-01 -2.78746754e-01
-5.29504716e-01 -2.49887258e-02 5.15567899e-01 4.05369103e-01
7.54948139e-01 -1.45665002e+00 -1.14734256e+00 -1.96293190e-01
-5.95829869e-03 -8.47858965e-01 2.57150531e-01 1.81148991e-01
-9.45785642e-01 -1.31593132e+00 -1.09086163e-01 -9.89553750e-01
7.37951994e-01 6.11770600e-02 1.46293592e+00 5.62394261e-01
-6.09954335e-02 8.33791077e-01 -1.90919656e-02 -2.46187314e-01
-5.32806993e-01 -2.92678177e-01 -4.61885214e-01 -3.13330859e-01
2.50852168e-01 -6.62099957e-01 -2.72736311e-01 -3.53289515e-01
-6.95431113e-01 6.16153777e-01 6.58695817e-01 7.92481661e-01
1.91411167e-01 -4.80920374e-02 -1.77816730e-02 -9.26128209e-01
5.95169067e-01 -2.63556421e-01 -1.07619071e+00 5.28014660e-01
-3.29933196e-01 6.47526383e-01 9.92299199e-01 -3.47098559e-01
-1.34314954e+00 -2.77614314e-02 1.17303781e-01 -1.59198582e-01
-3.28320146e-01 8.86982024e-01 -5.47908619e-02 -2.47220695e-02
8.42374325e-01 5.29958129e-01 -1.37075726e-02 1.08559214e-01
9.34423506e-01 -1.85425952e-01 6.15318716e-01 -1.03194082e+00
9.16840792e-01 6.90739155e-01 4.29848105e-01 -6.27853155e-01
-1.02895689e+00 1.31101096e-02 -3.70129019e-01 1.06941923e-01
1.06198323e+00 -6.22081280e-01 -1.72957325e+00 -8.49986356e-03
-1.32042050e+00 -8.19620788e-01 -3.43433917e-01 2.01999500e-01
-1.12028039e+00 1.86627403e-01 -6.72839046e-01 -8.50827932e-01
8.10206980e-02 -1.16600752e+00 5.98205447e-01 -3.60794887e-02
-3.54024798e-01 -1.11158359e+00 -4.24608635e-03 3.16624284e-01
3.32009971e-01 -3.25431174e-04 1.70480478e+00 -5.48260570e-01
-1.16919816e+00 4.82297063e-01 -5.02539039e-01 -2.01761603e-01
-5.83105922e-01 -3.23960006e-01 -9.16951299e-01 2.62332819e-02
-6.94559142e-02 -6.68034375e-01 9.75567520e-01 1.08082652e-01
1.29239690e+00 -6.05936646e-01 -1.84112415e-01 7.92375803e-01
1.33656669e+00 2.90763795e-01 6.36760473e-01 4.73453552e-01
5.83729148e-01 7.37148166e-01 6.36343136e-02 1.19248465e-01
7.30384707e-01 2.38671288e-01 7.38541424e-01 7.91666284e-02
-6.41484112e-02 -7.40589321e-01 2.59727418e-01 1.36446595e-01
-1.75169826e-01 3.54316443e-01 -1.04865897e+00 2.12689310e-01
-1.93081236e+00 -1.45691502e+00 -2.35173941e-01 1.67626631e+00
7.93235242e-01 9.48779359e-02 6.87008053e-02 -4.22907621e-02
1.19062670e-01 8.57451856e-02 -6.82318568e-01 -6.02008045e-01
7.75197819e-02 1.79719061e-01 -1.39670774e-01 7.79325008e-01
-6.99851274e-01 1.15503931e+00 6.07373381e+00 1.70405611e-01
-8.34145248e-01 -1.44825935e-01 1.55122355e-01 1.24410272e-01
-5.80675602e-01 3.78211647e-01 -1.79882184e-01 -8.72047469e-02
8.10414493e-01 2.16023922e-02 1.22442722e+00 1.15822315e+00
-2.44906738e-01 -1.86521500e-01 -1.77320862e+00 9.70709562e-01
4.82005402e-02 -1.63184702e+00 4.89010870e-01 -3.57741028e-01
3.02253395e-01 -3.05866510e-01 1.73031762e-01 9.91495848e-01
8.39258969e-01 -1.27617204e+00 9.99975920e-01 6.85597062e-01
4.87830937e-02 -3.13927084e-01 2.99377143e-02 5.00270784e-01
-1.06938636e+00 -5.10114551e-01 -2.08781391e-01 -6.63066924e-01
-2.05178931e-01 5.00412993e-02 -6.38239384e-01 1.36883289e-01
5.96141756e-01 4.01179969e-01 -6.33280814e-01 7.62181222e-01
-7.33420730e-01 1.31770194e-01 9.69330817e-02 -6.06136382e-01
2.83893794e-01 -1.47584245e-01 2.99391091e-01 1.03834212e+00
3.87258157e-02 1.27478600e-01 2.78263271e-01 1.66730428e+00
1.52743116e-01 -3.44316602e-01 -9.14375782e-01 -9.95244011e-02
1.14025258e-01 8.76922667e-01 -7.38400936e-01 -6.90215111e-01
-6.89927757e-01 1.11784005e+00 7.15235770e-01 7.92965293e-01
-8.28918993e-01 -3.42546463e-01 4.34724510e-01 -1.00164577e-01
2.30650112e-01 -3.88528466e-01 -3.29402804e-01 -1.15836859e+00
-4.57491726e-02 -1.04307544e+00 5.41564584e-01 -1.28568041e+00
-9.48072493e-01 1.70722723e-01 7.12779313e-02 -4.98206377e-01
-4.08566594e-01 -1.26273739e+00 -5.20716846e-01 5.51924288e-01
-1.57318962e+00 -1.26741183e+00 -4.64970499e-01 8.37207198e-01
4.13086742e-01 -1.74043640e-01 8.96907210e-01 -4.40843016e-01
1.06111512e-01 -5.60257360e-02 -6.49886966e-01 1.71718836e-01
-2.49808267e-01 -1.53629637e+00 2.22955674e-01 6.04549885e-01
3.20632756e-01 8.95570576e-01 7.62827277e-01 -2.95977294e-01
-1.95055497e+00 -7.38043666e-01 5.61533213e-01 -5.53258717e-01
8.43465745e-01 -2.93621421e-01 -9.79475737e-01 1.14273083e+00
2.99917847e-01 -1.47719473e-01 2.83146918e-01 3.69857609e-01
-1.01810896e+00 1.50298476e-01 -9.63050783e-01 1.22934568e+00
1.37925208e+00 -9.56133306e-01 -1.30429626e+00 3.48493934e-01
1.04400671e+00 -3.07349950e-01 -1.45671189e-01 -4.59018052e-02
3.01362753e-01 -1.17263043e+00 1.25923669e+00 -9.82047498e-01
6.09998047e-01 -5.56010544e-01 -2.62704700e-01 -1.36389720e+00
-3.37018460e-01 -2.65566498e-01 -3.30333501e-01 5.16998053e-01
2.53716826e-01 -8.53049517e-01 7.23339796e-01 6.01929367e-01
-5.09068742e-02 -2.44917333e-01 -5.60206413e-01 -6.50454760e-01
1.79887507e-02 -4.89513874e-01 5.46024024e-01 7.99554527e-01
5.08306980e-01 6.34426057e-01 3.98011595e-01 3.45879942e-01
5.50568223e-01 6.75859571e-01 8.07724416e-01 -1.35076606e+00
-6.13593459e-01 -8.99218321e-01 -2.75876433e-01 -1.07840931e+00
7.68169463e-01 -1.46931505e+00 -7.60667678e-03 -2.00126028e+00
2.35537767e-01 4.88528274e-02 1.03873603e-01 5.63937604e-01
1.47341222e-01 -3.79866153e-01 3.18556637e-01 -3.08507234e-02
-9.57057476e-01 3.30955803e-01 1.35790718e+00 -6.16934836e-01
-2.68738102e-02 -3.41103405e-01 -5.47944009e-01 1.17403662e+00
5.65750897e-01 -7.57316919e-03 -8.21000993e-01 -5.23461163e-01
9.53224063e-01 2.30739653e-01 1.34879732e+00 -1.03542459e+00
5.33854365e-01 -6.43963456e-01 3.01172882e-01 -3.79339129e-01
3.52468163e-01 -8.87427330e-01 -1.34720087e-01 9.27226841e-01
-6.69218481e-01 2.21696630e-01 1.35036707e-01 3.84710342e-01
-1.00632094e-01 -3.72446835e-01 5.18456280e-01 -7.91996956e-01
-1.24540699e+00 1.13800392e-02 -5.70575833e-01 2.37324998e-01
8.38089645e-01 -5.74240973e-03 -6.72495723e-01 -3.84163231e-01
-8.48993421e-01 1.91743582e-01 3.90316814e-01 -8.28999206e-02
8.64057720e-01 -1.01508093e+00 -2.02824563e-01 1.98320493e-01
3.34944636e-01 -1.57908559e-01 1.26974583e-01 8.08670938e-01
-9.07242715e-01 4.48170364e-01 -2.35065594e-01 -4.60501611e-01
-1.01403177e+00 1.26350594e+00 5.72545111e-01 -9.26305056e-02
-7.16545641e-01 9.12085950e-01 5.85418284e-01 -8.76280010e-01
4.09353554e-01 -8.67411435e-01 -1.46197304e-01 -3.65155697e-01
6.43907785e-01 9.90966558e-02 -3.55953753e-01 -2.11602431e-02
-2.37928122e-01 3.10758322e-01 3.95996332e-01 -2.61657476e-01
1.12352240e+00 2.41430417e-01 -5.74398994e-01 5.17048240e-01
7.30935037e-01 -3.40974629e-01 -1.12602162e+00 -1.73791111e-01
2.89039075e-01 -2.67663866e-01 -3.60016257e-01 -9.15959418e-01
-5.69733858e-01 1.25435090e+00 2.16237351e-01 4.77893353e-01
8.04731071e-01 2.10724309e-01 1.10509368e-02 1.28817308e+00
3.78699183e-01 -7.90975988e-01 6.50142312e-01 8.91097128e-01
1.02439952e+00 -1.08886290e+00 -2.35366315e-01 -2.68051028e-01
-5.36783814e-01 1.10193229e+00 7.39309192e-01 -9.74271819e-02
1.26551569e-01 1.25046030e-01 -1.51806712e-01 -5.94881654e-01
-9.85455573e-01 -3.93873185e-01 1.18132003e-01 9.82614636e-01
1.33356139e-01 -8.05688500e-02 4.39192802e-01 1.51044965e-01
-3.82132590e-01 8.90929773e-02 1.65286690e-01 8.50009024e-01
-6.11353576e-01 -4.49900568e-01 -3.19754183e-01 1.47980019e-01
1.40230641e-01 -3.27320158e-01 -3.06952655e-01 8.20295632e-01
-1.36349112e-01 7.57850468e-01 -4.87532839e-03 1.42929986e-01
3.60061884e-01 3.33817810e-01 9.53221917e-01 -4.46485490e-01
-2.66556919e-01 -9.18447196e-01 -5.43121900e-03 -8.93061519e-01
-3.37273747e-01 -2.47903928e-01 -1.77659345e+00 -5.12354076e-01
5.42261183e-01 3.25975306e-02 3.94477725e-01 1.09378183e+00
4.23692502e-02 7.77389109e-01 -1.95192650e-01 -7.84386635e-01
-7.07271695e-01 -3.49159867e-01 -7.69384801e-02 6.72891200e-01
5.74488342e-01 -5.28453469e-01 -4.12652075e-01 1.45990476e-01] | [10.608108520507812, 2.2625999450683594] |
7418f808-3e64-40e3-b968-8463310e0de8 | cue-an-uncertainty-interpretation-framework | 2306.03598 | null | https://arxiv.org/abs/2306.03598v1 | https://arxiv.org/pdf/2306.03598v1.pdf | CUE: An Uncertainty Interpretation Framework for Text Classifiers Built on Pre-Trained Language Models | Text classifiers built on Pre-trained Language Models (PLMs) have achieved remarkable progress in various tasks including sentiment analysis, natural language inference, and question-answering. However, the occurrence of uncertain predictions by these classifiers poses a challenge to their reliability when deployed in practical applications. Much effort has been devoted to designing various probes in order to understand what PLMs capture. But few studies have delved into factors influencing PLM-based classifiers' predictive uncertainty. In this paper, we propose a novel framework, called CUE, which aims to interpret uncertainties inherent in the predictions of PLM-based models. In particular, we first map PLM-encoded representations to a latent space via a variational auto-encoder. We then generate text representations by perturbing the latent space which causes fluctuation in predictive uncertainty. By comparing the difference in predictive uncertainty between the perturbed and the original text representations, we are able to identify the latent dimensions responsible for uncertainty and subsequently trace back to the input features that contribute to such uncertainty. Our extensive experiments on four benchmark datasets encompassing linguistic acceptability classification, emotion classification, and natural language inference show the feasibility of our proposed framework. Our source code is available at: https://github.com/lijiazheng99/CUE. | ['Yulan He', 'Lin Gui', 'Bin Liang', 'Zhaoyue Sun', 'Jiazheng Li'] | 2023-06-06 | null | null | null | null | ['emotion-classification', 'natural-language-inference', 'sentiment-analysis', 'linguistic-acceptability', 'emotion-classification'] | ['computer-vision', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [ 2.82879531e-01 4.35287923e-01 -3.31418455e-01 -8.30074310e-01
-1.01053929e+00 -3.96904677e-01 8.93486977e-01 2.79003859e-01
5.10850847e-02 8.28465044e-01 4.54950780e-01 -1.90883383e-01
-2.28541400e-02 -6.78148925e-01 -7.75038302e-01 -6.16452694e-01
3.54213923e-01 3.10780734e-01 -3.06021899e-01 1.49183542e-01
4.43924844e-01 -1.48735613e-01 -1.70040011e+00 5.13706088e-01
1.14359450e+00 1.25594294e+00 8.36583748e-02 3.33462924e-01
-4.41240340e-01 8.05046022e-01 -4.87964541e-01 -7.03721106e-01
-2.40143239e-01 -2.25173771e-01 -6.42344058e-01 -1.73461154e-01
-2.79373348e-01 1.32894322e-01 1.00306291e-02 1.13921797e+00
1.81735918e-01 2.48276830e-01 1.18836117e+00 -1.30170441e+00
-9.09957707e-01 1.09434295e+00 -2.93365985e-01 3.87748778e-02
3.49308997e-01 -2.86940604e-01 1.19806218e+00 -1.00389838e+00
1.66022435e-01 1.58355772e+00 5.42739570e-01 6.55815482e-01
-1.16567171e+00 -5.90836048e-01 3.93570244e-01 1.92278907e-01
-1.19968569e+00 -5.01901388e-01 1.00432909e+00 -5.04035413e-01
7.92284548e-01 2.07909793e-01 1.37273120e-02 1.53169668e+00
6.97718203e-01 1.03227580e+00 1.08149433e+00 -4.86374468e-01
4.84990567e-01 6.80791914e-01 1.57803252e-01 2.98061222e-01
3.54650579e-02 -5.51782586e-02 -7.61143684e-01 -3.22700411e-01
-6.23924471e-02 -8.02254155e-02 -2.70587444e-01 -3.03793848e-02
-6.74888730e-01 1.11878288e+00 1.18064195e-01 1.55236959e-01
-2.47195512e-01 2.05396742e-01 3.40414196e-01 6.47936985e-02
1.07996809e+00 3.23103279e-01 -5.72002590e-01 -2.81675637e-01
-6.68426752e-01 7.07884431e-02 6.48216605e-01 8.07841599e-01
4.14986759e-01 -1.61102593e-01 -2.62881339e-01 8.06348562e-01
6.92431509e-01 2.52565086e-01 7.29444325e-01 -7.03434467e-01
3.89227837e-01 4.64032173e-01 1.73754930e-01 -1.12387884e+00
-1.01902366e-01 -1.52127549e-01 -7.44417667e-01 -2.54949003e-01
-6.47095144e-02 -2.09254399e-01 -5.17152786e-01 1.70166171e+00
1.10270180e-01 1.21604517e-01 3.64061773e-01 6.04559422e-01
6.68527544e-01 1.06935263e+00 2.34897807e-01 -1.53008729e-01
1.16591465e+00 -6.99847877e-01 -8.17682028e-01 -2.04823509e-01
4.73311663e-01 -6.82835698e-01 1.15040839e+00 2.95081735e-01
-7.36596882e-01 -3.71483475e-01 -9.69415545e-01 8.57246667e-02
-3.30854893e-01 7.21021518e-02 5.44529200e-01 7.04472661e-01
-4.92631495e-01 6.93929613e-01 -9.22216356e-01 3.73724960e-02
3.56116295e-01 4.92975749e-02 -1.96594372e-02 1.56507611e-01
-1.49009216e+00 7.28053451e-01 4.20195073e-01 3.52730781e-01
-5.59963286e-01 -4.43576157e-01 -1.01158929e+00 -1.09399913e-03
3.61601353e-01 -4.75044340e-01 1.31257093e+00 -8.80453110e-01
-1.74385262e+00 4.26393032e-01 -3.97298068e-01 -3.75385672e-01
4.41637456e-01 -3.34888101e-01 -3.62029493e-01 -4.24044251e-01
4.60809236e-03 4.34326768e-01 9.85816181e-01 -1.21670997e+00
-2.99658060e-01 -3.17940146e-01 -1.75175339e-01 1.86396077e-01
-2.16735333e-01 -4.00110520e-02 -1.12451777e-01 -6.56088173e-01
4.91291918e-02 -8.08000624e-01 -1.01880588e-01 -2.77849257e-01
-5.92391670e-01 -5.85787535e-01 4.28460300e-01 -2.17866704e-01
1.08342898e+00 -2.03050542e+00 2.37624586e-01 2.22554747e-02
-7.94237033e-02 -5.66816665e-02 1.62093729e-01 4.40455288e-01
1.88692063e-01 4.05237079e-01 -3.24587077e-01 -7.88248360e-01
3.86065006e-01 3.21755588e-01 -1.07380378e+00 2.38778517e-01
4.41290230e-01 7.53927052e-01 -6.33616507e-01 -3.84747326e-01
1.76452339e-01 5.52934647e-01 -5.68715632e-01 2.27006823e-01
-7.11615741e-01 4.09178406e-01 -7.33009577e-01 6.19746149e-01
5.60263097e-01 -2.26804912e-01 2.86514987e-03 1.10009640e-01
8.85368511e-02 4.89320993e-01 -9.64638054e-01 1.34382999e+00
-4.57049966e-01 5.20039618e-01 -4.89283741e-01 -1.07540512e+00
1.02366316e+00 1.79609835e-01 4.21630479e-02 -1.55623972e-01
4.34381098e-01 4.34059873e-02 -4.58919138e-01 -3.73805970e-01
6.04007721e-01 -2.44346514e-01 -4.02605593e-01 3.49792570e-01
2.83005759e-02 -3.09763283e-01 -2.58532584e-01 1.01234525e-01
3.33881825e-01 1.93084836e-01 1.83540732e-01 -1.67436093e-01
4.63112384e-01 -3.27994227e-01 6.25737369e-01 7.63733029e-01
-2.32451931e-01 4.84064192e-01 9.10869777e-01 -1.95411578e-01
-6.07280791e-01 -9.53433990e-01 -5.19790709e-01 9.92806733e-01
-1.08634913e-02 -5.44775426e-01 -5.27209520e-01 -5.47285259e-01
7.93834031e-02 1.39759398e+00 -9.55940604e-01 -4.43891466e-01
2.87327200e-01 -9.06380594e-01 3.90272111e-01 4.71379727e-01
5.85174039e-02 -9.54073429e-01 -2.92049199e-01 1.62523657e-01
-3.07940006e-01 -8.92179012e-01 -2.29899809e-01 2.78824151e-01
-7.44675219e-01 -5.49710870e-01 -1.45525977e-01 -1.89440668e-01
4.79468137e-01 -4.97531325e-01 9.19104517e-01 -4.68997836e-01
6.35584518e-02 2.97791302e-01 -5.01285851e-01 -7.64955997e-01
-5.13772011e-01 2.14129803e-03 2.79909134e-01 1.21666372e-01
5.20954907e-01 -2.77686536e-01 -2.59892076e-01 1.88455835e-01
-9.29961681e-01 8.65287930e-02 3.11154395e-01 8.14042628e-01
6.21834576e-01 6.23087995e-02 7.24251807e-01 -1.02218950e+00
9.52592731e-01 -8.80028605e-01 -5.27619541e-01 3.71726781e-01
-7.97320485e-01 5.32062888e-01 4.71626997e-01 -3.45140368e-01
-1.33865047e+00 -2.86877543e-01 -2.05198362e-01 -2.70972341e-01
-1.12103611e-01 8.22117150e-01 -1.25958920e-01 6.94128990e-01
4.81526732e-01 7.34363347e-02 -1.88718364e-01 -3.67454320e-01
5.11233985e-01 9.33050454e-01 6.03525266e-02 -8.77028763e-01
2.84884155e-01 1.93145588e-01 -3.21919113e-01 -5.43065548e-01
-1.32562160e+00 4.54257093e-02 -2.94901997e-01 -9.19819996e-02
6.37153685e-01 -7.33920693e-01 -5.33795655e-01 2.62693048e-01
-1.03746104e+00 4.39817011e-02 -1.16718918e-01 4.49780583e-01
-5.70754409e-01 2.03433082e-01 -3.72068524e-01 -1.28970492e+00
-4.26066369e-01 -1.28090465e+00 1.17317736e+00 2.86589921e-01
-4.16703939e-01 -1.15410292e+00 6.12618923e-02 4.02116001e-01
3.88028264e-01 1.08598612e-01 1.07650363e+00 -6.86562896e-01
-2.61438131e-01 -2.72853196e-01 6.58358485e-02 4.06238407e-01
2.73446813e-02 2.25751355e-01 -1.22324514e+00 7.24891350e-02
1.67218521e-01 -5.57655692e-01 1.02852237e+00 6.35843515e-01
1.45228279e+00 -2.49352723e-01 -3.27848494e-01 3.25885236e-01
1.10835242e+00 1.86939328e-03 2.99161553e-01 7.03627616e-02
2.49748662e-01 8.01303506e-01 6.14624977e-01 7.22672999e-01
3.18139851e-01 4.28525239e-01 2.88499355e-01 7.72618651e-01
6.90404475e-01 -6.63317978e-01 6.01031959e-01 8.81107926e-01
2.83215016e-01 -5.16790628e-01 -8.92067373e-01 2.73324937e-01
-1.99946809e+00 -6.81107402e-01 2.87794381e-01 1.77546167e+00
8.53229344e-01 2.97554791e-01 -6.22295141e-01 2.37991661e-02
4.93270516e-01 2.33851299e-01 -7.81418085e-01 -6.93713725e-01
-4.89592664e-02 -1.83959156e-01 -9.47844684e-02 6.76438808e-01
-1.11281168e+00 8.81855011e-01 5.49321079e+00 8.63714874e-01
-8.81670594e-01 -1.54986009e-01 9.93231654e-01 -3.69414277e-02
-9.84481275e-01 -1.16363645e-01 -8.30371559e-01 5.70775986e-01
1.12192535e+00 -4.48889881e-01 1.62733808e-01 1.01837659e+00
1.48291111e-01 -1.69297889e-01 -1.12226176e+00 7.97579169e-01
1.43223837e-01 -1.20782924e+00 2.51921594e-01 -2.42557690e-01
7.10889757e-01 -1.78882405e-01 6.01157129e-01 5.70226014e-01
1.72692657e-01 -1.22637010e+00 7.51448214e-01 8.67015302e-01
5.48969030e-01 -9.11724627e-01 9.76590753e-01 6.64399028e-01
-6.96002901e-01 -3.65629382e-02 -5.91945648e-01 -1.32347673e-01
-1.16210729e-01 9.62871850e-01 -8.46128345e-01 3.43320847e-01
6.15091980e-01 5.24887323e-01 -2.85305917e-01 1.13661066e-01
-2.83441812e-01 8.29310358e-01 -1.12198673e-01 -3.94957572e-01
5.32349683e-02 -1.63986802e-01 4.31171864e-01 8.80755842e-01
3.78653705e-01 -1.90629868e-03 -1.90552920e-01 1.36481643e+00
-1.83141664e-01 1.39766157e-01 -4.90915626e-01 -4.67956722e-01
4.37139839e-01 8.38486791e-01 -3.82106632e-01 -1.59053221e-01
-1.99363157e-01 8.94858062e-01 4.54304665e-01 2.27511346e-01
-9.01072204e-01 -1.03774711e-01 7.52480507e-01 -4.54659849e-01
-3.13172936e-02 4.99396026e-02 -4.69802171e-01 -1.52863050e+00
2.57488135e-02 -6.04286492e-01 2.66999841e-01 -8.28482568e-01
-1.67883027e+00 8.79495084e-01 7.28354529e-02 -7.85015762e-01
-6.71704412e-01 -5.67791104e-01 -5.31159818e-01 1.06151199e+00
-1.34004164e+00 -7.50511289e-01 4.66226712e-02 2.39123315e-01
6.85721099e-01 -1.52312234e-01 1.08010948e+00 -3.41963589e-01
-7.47681081e-01 6.52477384e-01 4.72850382e-01 -2.42261514e-01
6.62519515e-01 -1.13841367e+00 2.61891097e-01 5.32689571e-01
1.60644531e-01 5.22464216e-01 9.69918430e-01 -5.29931068e-01
-1.14421272e+00 -1.02138650e+00 1.03936529e+00 -9.53233719e-01
7.15903223e-01 -5.25424361e-01 -8.74494731e-01 6.87279463e-01
1.35482848e-01 -1.83585137e-01 9.78609979e-01 3.06524873e-01
-3.50281715e-01 1.91498339e-01 -1.08424342e+00 4.40819174e-01
3.51215035e-01 -6.38751388e-01 -7.30985582e-01 1.59173071e-01
9.48124886e-01 -2.66736448e-01 -6.43984556e-01 4.15890992e-01
4.33676809e-01 -8.16497386e-01 4.12450880e-01 -7.37748623e-01
8.92038047e-01 8.54810178e-02 -5.89894712e-01 -1.48849738e+00
4.52641994e-02 -2.92389601e-01 -8.02621692e-02 1.45011258e+00
6.53033674e-01 -7.51705766e-01 6.79160714e-01 9.40179586e-01
3.52953300e-02 -1.25182557e+00 -8.70278776e-01 -2.76440442e-01
4.14467216e-01 -1.04271495e+00 5.56263387e-01 6.93097293e-01
1.26577333e-01 1.46631509e-01 -4.08536762e-01 3.15398663e-01
5.77074468e-01 2.98262417e-01 2.30848625e-01 -1.14720058e+00
-2.49439195e-01 -3.47479433e-01 -2.15171576e-01 -8.64097834e-01
6.15851521e-01 -9.00531292e-01 1.62913814e-01 -1.12665641e+00
2.69139677e-01 -4.28068787e-01 -3.24943453e-01 2.21254975e-01
-3.70032221e-01 -1.92237377e-01 7.26344250e-03 2.28060648e-01
-3.94318283e-01 1.09408295e+00 6.06825531e-01 -9.72702578e-02
-1.05507463e-01 1.81231454e-01 -9.96512294e-01 8.92145395e-01
1.19216073e+00 -5.38861454e-01 -5.32730281e-01 -2.86484540e-01
4.17315364e-01 3.01523935e-02 8.49884748e-02 -5.28147221e-01
-8.16744938e-02 -3.28842938e-01 3.60109746e-01 -5.37931621e-01
5.03798187e-01 -6.02262735e-01 -5.31044491e-02 -2.97719780e-02
-9.19714510e-01 -3.17359179e-01 2.64322162e-01 7.91203737e-01
-3.69381458e-01 -4.79542375e-01 5.37312508e-01 1.73832849e-01
-3.64869267e-01 9.71766487e-02 -4.59602922e-01 1.20856024e-01
8.35668921e-01 4.10804391e-01 -1.43281236e-01 -5.56050122e-01
-6.92798316e-01 1.54622138e-01 1.01519659e-01 7.41917193e-01
6.28437102e-01 -1.21273220e+00 -6.05220199e-01 2.51307935e-01
2.62670249e-01 -1.08314693e-01 3.42220604e-01 4.03304398e-01
-3.44670983e-03 7.51273572e-01 2.70114362e-01 -6.45994842e-01
-7.42610991e-01 4.21984553e-01 2.86524206e-01 -1.54495299e-01
-1.94525197e-01 1.02197051e+00 6.35676235e-02 -5.57426691e-01
3.92713189e-01 -5.73453724e-01 -1.74763337e-01 1.58259109e-01
3.92523378e-01 8.54710117e-02 -1.19439699e-01 -4.89772052e-01
-3.50001246e-01 3.02800506e-01 -8.73486102e-02 -1.20844014e-01
1.10004497e+00 -3.49842399e-01 -7.61789444e-04 1.15523601e+00
1.18620205e+00 -8.22527111e-02 -1.16611969e+00 -2.98198402e-01
2.14037195e-01 -3.84596467e-01 1.98092088e-01 -6.27583802e-01
-5.34418166e-01 1.11809015e+00 3.04949492e-01 1.97893813e-01
7.95147479e-01 2.61488140e-01 3.51673454e-01 3.40878814e-01
2.94610679e-01 -9.39692259e-01 -7.40181804e-02 4.91953522e-01
1.10127628e+00 -1.54370880e+00 -3.18385869e-01 -2.05265760e-01
-1.14504993e+00 9.88006949e-01 4.14400697e-01 1.87198162e-01
9.50168908e-01 2.02511609e-01 8.25905204e-02 1.49798408e-01
-1.31929266e+00 3.18882138e-01 5.51915765e-01 2.17194632e-01
5.97724736e-01 3.78198862e-01 3.11811101e-02 1.27386248e+00
-3.59381199e-01 -1.19458094e-01 4.40136939e-01 5.88720202e-01
-4.84059036e-01 -9.10236299e-01 -2.95700639e-01 5.30456305e-01
-4.31624115e-01 -1.08077303e-01 -3.52940202e-01 6.40812144e-02
2.59117726e-02 1.12676883e+00 1.18749566e-01 -4.17213082e-01
-3.70810330e-02 5.65812647e-01 9.62739512e-02 -5.84583819e-01
8.19848403e-02 -1.90031856e-01 3.25162746e-02 -3.13806385e-01
-2.76975393e-01 -8.12434554e-01 -1.14951885e+00 -9.34733227e-02
-3.83634150e-01 4.99407351e-01 8.13012481e-01 9.47124958e-01
4.65366036e-01 4.95733351e-01 5.65767884e-01 -5.99983752e-01
-9.12079990e-01 -1.17262447e+00 -5.68727374e-01 1.73521847e-01
1.86487362e-01 -7.91900039e-01 -7.58116424e-01 -1.47203028e-01] | [11.535274505615234, 8.721552848815918] |
1622e6c5-4ebd-4def-966f-f8a6c2089739 | target-active-speaker-detection-with-audio | 2305.12831 | null | https://arxiv.org/abs/2305.12831v3 | https://arxiv.org/pdf/2305.12831v3.pdf | Target Active Speaker Detection with Audio-visual Cues | In active speaker detection (ASD), we would like to detect whether an on-screen person is speaking based on audio-visual cues. Previous studies have primarily focused on modeling audio-visual synchronization cue, which depends on the video quality of the lip region of a speaker. In real-world applications, it is possible that we can also have the reference speech of the on-screen speaker. To benefit from both facial cue and reference speech, we propose the Target Speaker TalkNet (TS-TalkNet), which leverages a pre-enrolled speaker embedding to complement the audio-visual synchronization cue in detecting whether the target speaker is speaking. Our framework outperforms the popular model, TalkNet on two datasets, achieving absolute improvements of 1.6% in mAP on the AVA-ActiveSpeaker validation set, and 0.8%, 0.4%, and 0.8% in terms of AP, AUC and EER on the ASW test set, respectively. Code is available at https://github.com/Jiang-Yidi/TS-TalkNet/. | ['Haizhou Li', 'Zexu Pan', 'Ruijie Tao', 'Yidi Jiang'] | 2023-05-22 | null | null | null | null | ['audio-visual-synchronization', 'audio-visual-synchronization'] | ['audio', 'computer-vision'] | [ 7.06183491e-03 1.20116130e-01 -2.41094425e-01 -3.19761425e-01
-1.32672477e+00 -6.35313570e-01 3.72976571e-01 -2.80556120e-02
-4.16882895e-02 1.44728750e-01 4.05288398e-01 -8.01830888e-02
5.04076362e-01 -1.34371653e-01 -3.81446302e-01 -6.17801011e-01
1.11166671e-01 -2.44799033e-01 2.55632490e-01 2.60888517e-01
5.17206118e-02 7.51551762e-02 -1.40107262e+00 4.21115875e-01
5.27560234e-01 1.25800598e+00 -6.49031401e-02 7.64650643e-01
1.70604542e-01 5.71476877e-01 -6.79700017e-01 -3.33750933e-01
7.82677308e-02 -4.50449139e-01 -2.03447670e-01 -5.43471910e-02
6.23598218e-01 -2.93890238e-01 -4.89106447e-01 1.05056393e+00
9.26614106e-01 -1.81725606e-01 1.74828187e-01 -1.52532935e+00
-2.80763149e-01 6.38788104e-01 -6.53221548e-01 3.16158146e-01
7.39482760e-01 4.53997642e-01 9.71215367e-01 -1.18193352e+00
8.79273936e-02 1.41870677e+00 5.16809404e-01 7.54206181e-01
-1.10488963e+00 -1.16276610e+00 5.71037456e-02 3.89206678e-01
-1.67487156e+00 -1.42301857e+00 9.07004416e-01 -3.57193261e-01
4.92067665e-01 3.59884769e-01 5.37228584e-01 1.28826356e+00
-1.73136353e-01 9.36178207e-01 8.19686592e-01 -3.05349022e-01
6.32222071e-02 4.21107799e-01 1.49574518e-01 3.31229657e-01
-2.72841841e-01 2.15064734e-01 -1.06047809e+00 -1.86773375e-01
2.71439880e-01 -3.74725878e-01 -6.43087208e-01 8.46674293e-02
-1.09238577e+00 6.67146027e-01 2.02797264e-01 2.57474221e-02
-1.32681638e-01 -2.28039455e-02 3.06318790e-01 1.45623088e-02
6.64774477e-01 -9.93483141e-02 8.81230012e-02 -4.84794855e-01
-1.13586998e+00 -2.14688256e-01 6.06062472e-01 6.51113093e-01
2.30412379e-01 1.58331662e-01 -3.17397058e-01 9.75876927e-01
5.78438580e-01 7.26019561e-01 2.82677352e-01 -9.47383463e-01
4.58210319e-01 3.12711984e-01 3.58343907e-02 -6.11462891e-01
-2.94933133e-02 -1.61619887e-01 -3.22369516e-01 1.11949004e-01
2.01074153e-01 -1.04614779e-01 -6.42043233e-01 1.82145393e+00
5.35978913e-01 5.65813363e-01 -1.41230449e-01 1.05974996e+00
1.25088274e+00 5.79528928e-01 -1.18013129e-01 -3.50646913e-01
1.41853631e+00 -8.90410542e-01 -8.43451798e-01 -2.46490866e-01
9.76798907e-02 -9.37702656e-01 1.10606039e+00 3.05466592e-01
-1.17161715e+00 -7.74785459e-01 -9.15073812e-01 3.13186437e-01
2.07469493e-01 3.28506261e-01 2.63443701e-02 8.83024633e-01
-1.25025165e+00 -1.83567971e-01 -8.24245751e-01 -3.59778732e-01
2.36506060e-01 2.05224380e-01 -2.56089896e-01 7.72312731e-02
-1.09309328e+00 3.00692469e-01 -3.77274513e-01 3.43003385e-02
-1.25245965e+00 -7.66333699e-01 -8.67610514e-01 -4.66300249e-02
2.81147897e-01 -2.90977418e-01 1.50685310e+00 -1.24400353e+00
-1.64514422e+00 8.77606094e-01 -7.10422099e-01 -2.85920352e-01
7.46306717e-01 -1.52358145e-01 -7.88137317e-01 4.62920249e-01
2.89184554e-03 7.20377326e-01 1.21202242e+00 -1.01072073e+00
-4.57897246e-01 -3.91396850e-01 -3.57675888e-02 1.62139431e-01
-3.08906525e-01 4.84545916e-01 -7.01537728e-01 -5.69429338e-01
-1.03459239e-01 -9.65939999e-01 6.59784913e-01 3.71817440e-01
-5.49718440e-01 -3.37393910e-01 1.05936038e+00 -9.41967845e-01
1.05817807e+00 -2.78028250e+00 -3.86564195e-01 -9.94957685e-02
3.88317764e-01 3.50900024e-01 -1.53171569e-01 7.89513066e-02
-1.36339039e-01 -3.24172787e-02 2.03077227e-01 -6.28283620e-01
-2.02664271e-01 -4.70444739e-01 -2.20381364e-01 7.47315049e-01
1.24680735e-01 5.24533868e-01 -5.99164367e-01 -4.42717314e-01
9.64206606e-02 8.59797657e-01 -2.31003508e-01 3.32059383e-01
3.81435335e-01 4.66776162e-01 2.36994890e-03 7.29597092e-01
6.88148916e-01 -8.44121501e-02 -7.12968260e-02 -4.50584665e-02
-1.04392305e-01 7.12137222e-01 -1.28592491e+00 1.41041207e+00
-3.86967152e-01 1.16140711e+00 5.89246333e-01 -4.26886737e-01
9.28404987e-01 8.07793498e-01 1.65300578e-01 -7.09044993e-01
1.91610362e-02 5.16862199e-02 2.65596092e-01 -4.74645287e-01
8.82196352e-02 2.51798898e-01 3.46964180e-01 1.35654166e-01
-1.88134566e-01 3.89941216e-01 -1.14384480e-01 3.32497895e-01
1.03986001e+00 -4.08198863e-01 -6.94114389e-03 -1.49897575e-01
7.21869111e-01 -7.96321571e-01 5.46238005e-01 3.75345558e-01
-8.05738091e-01 8.98649156e-01 4.46731985e-01 4.13864702e-01
-5.39990485e-01 -1.29217780e+00 -6.43673167e-02 1.11848819e+00
1.59646824e-01 -6.79641604e-01 -8.73008132e-01 -4.19726402e-01
-1.25949025e-01 6.94577217e-01 -4.88453716e-01 -1.17088258e-01
-3.78171355e-01 1.29143586e-02 7.46023476e-01 4.58695352e-01
3.93591702e-01 -7.24866271e-01 -2.07645342e-01 -7.18854740e-02
-4.85038698e-01 -1.29038012e+00 -9.95955110e-01 -4.46419984e-01
-2.86808819e-01 -9.39270318e-01 -9.11080301e-01 -4.75456446e-01
3.17273796e-01 5.12413383e-01 7.05999374e-01 -1.67778656e-01
-1.80976301e-01 6.90055668e-01 -1.97724849e-01 -4.44976866e-01
-4.89924014e-01 -3.29338998e-01 2.57146806e-01 6.69334590e-01
2.13051394e-01 -3.09979677e-01 -9.54957545e-01 6.09225333e-01
-1.59706220e-01 -1.28898313e-02 2.74197049e-02 4.27533567e-01
1.80459097e-01 -3.21178943e-01 5.57911694e-01 -1.18027784e-01
4.03413713e-01 -3.03051054e-01 -5.31878173e-01 -2.47584768e-02
-3.28999311e-01 -6.89618409e-01 2.73239426e-02 -8.59534681e-01
-8.50559592e-01 1.51786227e-02 -2.57169574e-01 -8.01124573e-01
-1.78269267e-01 -5.25307469e-02 -5.18121779e-01 1.52039915e-01
5.48548877e-01 1.87145621e-01 1.71150178e-01 -3.13216478e-01
1.34497031e-03 1.23631012e+00 4.22846526e-01 -1.26725763e-01
4.55341607e-01 5.43338299e-01 -5.36120236e-01 -1.17940938e+00
-4.81068283e-01 -7.84408867e-01 -7.84809738e-02 -5.14729977e-01
6.63020432e-01 -1.30192542e+00 -8.12074125e-01 6.33461535e-01
-1.25998223e+00 -2.97125787e-01 1.34748489e-01 5.52997530e-01
-2.93111503e-01 3.73621255e-01 -4.20120537e-01 -1.24381006e+00
-4.76369560e-01 -1.40385258e+00 1.13172603e+00 1.96420357e-01
-4.36179668e-01 -5.51493526e-01 -3.55913974e-02 6.39899433e-01
5.42016506e-01 -1.37089133e-01 1.33175284e-01 -6.20818555e-01
-2.78742880e-01 -2.24255174e-01 -2.03950666e-02 4.23342168e-01
3.43270034e-01 1.81910202e-01 -1.68910348e+00 -4.73743737e-01
-9.93115529e-02 -1.09157190e-01 6.77142262e-01 5.84332466e-01
8.51231277e-01 -3.76587808e-01 -4.06877548e-01 4.38125521e-01
7.30001569e-01 3.78667861e-01 5.11233926e-01 -3.19178283e-01
4.12387043e-01 7.20453262e-01 3.78630430e-01 3.46005857e-01
2.79348552e-01 9.86285985e-01 1.95861697e-01 -1.48149759e-01
-6.93967938e-01 -2.75371611e-01 9.45647776e-01 6.23312056e-01
3.53113323e-01 -3.83184969e-01 -1.01955831e+00 6.18527472e-01
-1.48490071e+00 -8.35417747e-01 -5.23354523e-02 2.42269897e+00
6.96672142e-01 2.52252698e-01 5.98763108e-01 4.07402694e-01
1.27917099e+00 1.89821690e-01 -7.15253830e-01 -1.55052776e-02
-5.04055694e-02 -1.68091342e-01 -1.57867391e-02 6.46202743e-01
-1.01812899e+00 5.13162673e-01 5.10840368e+00 6.11220717e-01
-1.74427593e+00 3.13829899e-01 7.23846972e-01 -5.99060953e-01
1.36597112e-01 -3.85012329e-01 -9.33831871e-01 8.39473426e-01
1.31091583e+00 -1.98329955e-01 1.71641558e-01 6.13426208e-01
7.84181476e-01 -3.80376503e-02 -1.32101333e+00 1.41476655e+00
4.32136834e-01 -8.16693902e-01 -4.59946126e-01 1.64899319e-01
1.96316801e-02 1.04815260e-01 3.80202681e-01 1.16823576e-01
-2.50748903e-01 -9.37708199e-01 9.33619082e-01 1.27315089e-01
1.08034766e+00 -6.18660748e-01 3.15684766e-01 1.31021410e-01
-1.41919816e+00 6.54355586e-02 3.17900389e-01 3.43220711e-01
9.04583931e-02 2.82215923e-01 -1.08283603e+00 1.98594145e-02
1.00386882e+00 5.00310719e-01 -4.86025810e-01 1.19279170e+00
-4.05794770e-01 1.27740276e+00 -4.82160300e-01 2.38863453e-01
-3.68844986e-01 4.62653190e-01 8.83640170e-01 1.21709728e+00
2.38876939e-01 -8.28308910e-02 -8.42305049e-02 8.38332176e-01
-7.37773031e-02 2.66545620e-02 -4.38635886e-01 1.08988374e-03
7.21551299e-01 1.23508775e+00 -2.54221618e-01 1.38902972e-02
-4.85612810e-01 8.80978405e-01 -3.29428881e-01 4.21710908e-01
-9.79288280e-01 -5.34421146e-01 8.49427640e-01 4.83990371e-01
1.24698140e-01 7.58039802e-02 1.11063972e-01 -9.46329236e-01
2.91559696e-01 -1.06410348e+00 2.63233066e-01 -8.37277293e-01
-9.07151699e-01 6.49763942e-01 -3.44304085e-01 -1.28560066e+00
-3.39368194e-01 -2.67073870e-01 -9.69785035e-01 1.07672095e+00
-1.16160131e+00 -1.15821242e+00 -4.48491842e-01 7.53310323e-01
7.56502151e-01 -1.60626799e-01 5.80104291e-01 4.03557718e-01
-8.02490115e-01 1.28226495e+00 -2.29420513e-01 5.52474797e-01
1.01023281e+00 -7.37549782e-01 3.81209135e-01 9.34908926e-01
1.57431349e-01 5.06824434e-01 7.43512690e-01 -1.45607397e-01
-1.30195558e+00 -8.72095525e-01 8.01842391e-01 -3.19735587e-01
6.29983366e-01 -7.57434011e-01 -8.88618052e-01 4.69756633e-01
3.09043705e-01 9.78698209e-02 7.93005645e-01 2.63285637e-02
-6.05827868e-01 -4.16367114e-01 -1.01289582e+00 3.53407711e-01
6.56026065e-01 -8.30161870e-01 -3.23921442e-01 -5.50355949e-02
6.49690628e-01 -2.61039048e-01 -4.96726781e-01 2.31657729e-01
7.25396097e-01 -8.93415093e-01 8.87273848e-01 -3.92452143e-02
-3.97102013e-02 -1.83038682e-01 -1.02757715e-01 -1.05438244e+00
2.67482977e-02 -8.55760574e-01 -1.93542525e-01 1.82725346e+00
4.30249512e-01 -7.55481422e-01 5.66152453e-01 4.51337963e-01
6.30277246e-02 -4.46944535e-01 -1.42706466e+00 -7.97400832e-01
-3.82852465e-01 -7.15569377e-01 2.82789558e-01 7.28773117e-01
1.02401830e-01 4.21604782e-01 -2.73763567e-01 3.13374519e-01
5.66427052e-01 -3.07765007e-01 8.42013299e-01 -1.02731824e+00
-3.53270054e-01 -4.02874649e-01 -4.28224653e-01 -1.08689487e+00
2.35937104e-01 -6.15942121e-01 3.40806283e-02 -1.22618568e+00
1.66184455e-01 8.03172737e-02 -5.35870910e-01 4.50031340e-01
-5.64695932e-02 3.35842550e-01 2.99554408e-01 3.09561402e-01
-4.45392817e-01 5.48374593e-01 7.60567129e-01 -4.30310249e-01
-3.46575469e-01 3.49197447e-01 -6.67998612e-01 7.36170650e-01
7.91992188e-01 -3.50455552e-01 -2.96865433e-01 -1.16503499e-01
-2.76341081e-01 4.26742285e-01 4.76207346e-01 -9.70631540e-01
3.65960062e-01 2.28981972e-01 1.17327146e-01 -4.55951244e-01
1.00473595e+00 -3.86804909e-01 -1.62342995e-01 3.17147791e-01
-4.77179080e-01 -2.28741169e-01 4.61229891e-01 3.88745099e-01
-2.87244171e-01 1.91732317e-01 1.08961904e+00 3.18437904e-01
-2.26558432e-01 2.32740268e-01 -3.11241001e-01 5.57019636e-02
7.83970177e-01 -1.91781491e-01 -5.65071106e-01 -8.63012850e-01
-5.80844939e-01 1.23065844e-01 9.54314321e-02 6.40129089e-01
7.98984110e-01 -1.15583146e+00 -9.39149022e-01 3.17995816e-01
3.42199057e-01 -5.09317696e-01 2.91446477e-01 1.22792053e+00
3.55898403e-02 3.24971437e-01 2.04096898e-01 -9.52169955e-01
-1.93794620e+00 2.72683114e-01 5.20487428e-01 7.11196244e-01
-4.34694439e-01 1.02418554e+00 5.66687882e-01 2.03810290e-01
6.93284869e-01 -8.15750808e-02 6.75674081e-02 1.35735393e-01
8.77824187e-01 5.33927143e-01 4.86424826e-02 -9.19572175e-01
-7.23468900e-01 3.09505254e-01 3.55295576e-02 -4.77561802e-01
7.25341916e-01 -3.82869244e-01 2.78989255e-01 6.47657037e-01
1.28398132e+00 5.30863702e-01 -1.18153834e+00 -3.61432344e-01
-3.75445515e-01 -4.37973529e-01 2.21917048e-01 -8.08722556e-01
-1.17121828e+00 1.25184882e+00 1.19489408e+00 1.53517157e-01
1.04469955e+00 3.29050004e-01 5.15979528e-01 -1.62992164e-01
-2.39487328e-02 -6.36464536e-01 3.04217815e-01 -1.22415498e-02
1.10801053e+00 -1.28199279e+00 -4.44116086e-01 -4.73602206e-01
-6.75591588e-01 9.75425184e-01 5.76991022e-01 4.29038107e-01
7.98740089e-01 2.23584548e-01 4.86118317e-01 2.14749500e-01
-7.51466870e-01 -1.65305570e-01 4.65766490e-01 5.28719068e-01
4.57938850e-01 1.36355266e-01 3.70462894e-01 6.38698101e-01
-3.07359666e-01 -3.67803574e-01 1.54325008e-01 4.57319468e-01
-3.02151710e-01 -6.27085030e-01 -5.81172109e-01 -1.13796055e-01
-6.52789116e-01 -3.97263244e-02 -6.89431071e-01 4.40083563e-01
-1.45800322e-01 1.66357684e+00 2.85521001e-01 -4.34154928e-01
2.92364210e-01 1.63361311e-01 2.05442518e-01 -4.72058922e-01
-3.13521057e-01 6.57442391e-01 4.40386795e-02 -5.65776527e-01
-3.16298217e-01 -7.92534173e-01 -9.48982418e-01 -1.92327753e-01
-2.95657545e-01 8.47239643e-02 6.21685326e-01 3.54407281e-01
5.19897342e-01 2.94942528e-01 8.02934170e-01 -5.96854150e-01
-2.59353012e-01 -1.25490272e+00 -5.24245262e-01 7.42352977e-02
7.88089097e-01 -5.71194947e-01 -8.28900218e-01 -8.32665339e-02] | [14.438026428222656, 5.192652225494385] |
750f3080-091e-43d8-a3d4-2a8a4fd3d126 | towards-understanding-and-improving-knowledge | 2305.08096 | null | https://arxiv.org/abs/2305.08096v1 | https://arxiv.org/pdf/2305.08096v1.pdf | Towards Understanding and Improving Knowledge Distillation for Neural Machine Translation | Knowledge distillation (KD) is a promising technique for model compression in neural machine translation. However, where the knowledge hides in KD is still not clear, which may hinder the development of KD. In this work, we first unravel this mystery from an empirical perspective and show that the knowledge comes from the top-1 predictions of teachers, which also helps us build a potential connection between word- and sequence-level KD. Further, we point out two inherent issues in vanilla word-level KD based on this finding. Firstly, the current objective of KD spreads its focus to whole distributions to learn the knowledge, yet lacks special treatment on the most crucial top-1 information. Secondly, the knowledge is largely covered by the golden information due to the fact that most top-1 predictions of teachers overlap with ground-truth tokens, which further restricts the potential of KD. To address these issues, we propose a novel method named \textbf{T}op-1 \textbf{I}nformation \textbf{E}nhanced \textbf{K}nowledge \textbf{D}istillation (TIE-KD). Specifically, we design a hierarchical ranking loss to enforce the learning of the top-1 information from the teacher. Additionally, we develop an iterative KD procedure to infuse more additional knowledge by distilling on the data without ground-truth targets. Experiments on WMT'14 English-German, WMT'14 English-French and WMT'16 English-Romanian demonstrate that our method can respectively boost Transformer$_{base}$ students by +1.04, +0.60 and +1.11 BLEU scores and significantly outperform the vanilla word-level KD baseline. Besides, our method shows higher generalizability on different teacher-student capacity gaps than existing KD techniques. | ['Yufeng Chen', 'Jinan Xu', 'Jian Liu', 'Wenjuan Han', 'Shuaibo Wang', 'Yunlong Liang', 'Songming Zhang'] | 2023-05-14 | null | null | null | null | ['model-compression'] | ['methodology'] | [ 1.35913059e-01 3.17347705e-01 -6.15593195e-01 -1.89766139e-01
-1.07520950e+00 -6.42817616e-01 1.82751134e-01 1.90033615e-02
-7.13428855e-01 1.04761291e+00 8.23800266e-02 -8.33183825e-01
-2.45716944e-01 -7.01066375e-01 -1.20098364e+00 -5.82930684e-01
4.84388292e-01 6.43246472e-01 1.17307544e-01 -4.57560807e-01
-1.16703503e-01 3.21749188e-02 -1.21170354e+00 2.88004637e-01
1.47247005e+00 5.08117676e-01 6.32026255e-01 3.52902949e-01
-1.95329696e-01 6.70004904e-01 -6.45483255e-01 -9.96592760e-01
1.39609963e-01 -3.85335147e-01 -7.89419770e-01 -6.18635535e-01
6.38032973e-01 -3.19036126e-01 -4.41166580e-01 1.17618513e+00
6.24967933e-01 3.82314287e-02 6.44109905e-01 -9.11206901e-01
-9.28278863e-01 1.28066730e+00 -7.41688013e-01 2.74375141e-01
9.82707599e-04 1.16407789e-01 1.07543015e+00 -9.98467684e-01
4.55823839e-01 1.19164979e+00 5.11114299e-01 7.50076175e-01
-1.09100568e+00 -9.67522681e-01 1.19534418e-01 4.25137371e-01
-1.48000848e+00 -1.76648527e-01 5.09611428e-01 -3.19234848e-01
8.80190730e-01 2.35674500e-01 5.16830921e-01 1.12300861e+00
-1.48228198e-01 1.25430548e+00 1.36794651e+00 -7.69215703e-01
-2.65350878e-01 3.43687683e-01 2.15878353e-01 7.91848719e-01
1.99772939e-01 1.67202756e-01 -8.35345447e-01 2.12325484e-01
4.56398606e-01 -4.49825108e-01 -3.91693562e-01 -1.81374177e-02
-9.77088928e-01 7.15650260e-01 1.18455358e-01 1.58740014e-01
6.98286444e-02 7.11555034e-02 5.74479811e-02 4.49451655e-01
4.32159394e-01 3.86250675e-01 -9.32569921e-01 -6.23218775e-01
-1.01843798e+00 2.28668287e-01 5.93666673e-01 1.10640538e+00
7.86654770e-01 1.24379098e-01 -4.20586348e-01 1.08180261e+00
1.09752260e-01 8.57349396e-01 6.68372631e-01 -6.75127506e-01
8.39278400e-01 3.92021954e-01 -2.18464494e-01 -4.44842607e-01
7.40686059e-02 -6.35902166e-01 -7.54651248e-01 -2.86274374e-01
5.16700923e-01 -2.06351608e-01 -1.06202710e+00 2.07804084e+00
2.06265092e-01 4.39305454e-01 9.46347415e-03 6.54922605e-01
8.15591693e-01 5.52785695e-01 -2.75533311e-02 -2.89190412e-01
1.34834886e+00 -9.28826869e-01 -6.81869566e-01 -1.79005042e-01
8.73630464e-01 -7.55872667e-01 1.38630211e+00 5.66424906e-01
-1.22708976e+00 -5.56125224e-01 -8.57293367e-01 -1.82184070e-01
-4.11067277e-01 1.87292293e-01 4.34251517e-01 9.14532959e-01
-9.45931137e-01 4.83451515e-01 -5.00998795e-01 1.67280167e-01
3.80048335e-01 4.47284877e-01 -8.38246942e-02 -4.42754030e-01
-1.58669508e+00 1.07922971e+00 5.32694161e-01 -1.37623742e-01
-9.31965947e-01 -1.19873428e+00 -6.96015358e-01 5.36153503e-02
4.44723606e-01 -5.12672782e-01 1.43299079e+00 -4.71545458e-01
-1.61034381e+00 6.47627652e-01 -1.37297615e-01 -4.40318167e-01
5.11633337e-01 -3.62831295e-01 -8.84659067e-02 -2.02403128e-01
-1.97360162e-02 7.14546144e-01 4.79302168e-01 -1.04591024e+00
-6.51064336e-01 -2.25785270e-01 1.73762795e-02 3.06371212e-01
-5.97450733e-01 -2.28705615e-01 -5.34474254e-01 -7.82194495e-01
-6.66535571e-02 -8.62473130e-01 7.18614385e-02 -7.57168889e-01
-4.02213544e-01 -6.03715420e-01 2.78144240e-01 -6.66757166e-01
1.68905282e+00 -1.94776022e+00 1.24551937e-01 1.07918955e-01
1.79039344e-01 6.28981352e-01 -3.66712034e-01 2.45107397e-01
5.18231420e-03 3.02727282e-01 1.07852504e-01 -7.37613663e-02
6.55828416e-02 5.50126195e-01 -3.41598958e-01 -2.99555645e-03
1.16760634e-01 1.08836508e+00 -9.24490392e-01 -5.65568805e-01
-1.63596198e-01 1.96353152e-01 -7.84705162e-01 1.14231685e-03
-3.22899759e-01 3.20713252e-01 -3.33909661e-01 4.79072481e-01
5.95427811e-01 7.37302303e-02 9.68319923e-02 1.91732526e-01
-2.80042458e-02 6.60050154e-01 -1.00008965e+00 1.77113867e+00
-3.42257589e-01 3.49231541e-01 8.44823848e-03 -1.22550607e+00
6.30762637e-01 2.47598186e-01 1.74307063e-01 -7.30694950e-01
-7.54733309e-02 3.80896568e-01 3.41347337e-01 -3.60170633e-01
5.99092662e-01 -2.68466502e-01 -1.15097528e-02 2.94684261e-01
3.26337337e-01 -2.78787255e-01 1.11727312e-01 2.19180107e-01
8.54276896e-01 2.78332621e-01 -1.81029543e-01 -3.27668130e-01
3.04095984e-01 4.32009287e-02 7.85323560e-01 8.02630663e-01
-9.04392824e-02 1.36232242e-01 3.30738574e-01 1.05561428e-01
-8.22187662e-01 -9.82869387e-01 -2.16617495e-01 1.37530208e+00
-4.81345057e-02 -4.89809781e-01 -9.49025333e-01 -9.52235401e-01
2.22062413e-02 9.87421930e-01 -3.98029774e-01 -3.77639532e-01
-7.97467530e-01 -7.02748179e-01 9.52981174e-01 4.26399261e-01
3.81070971e-01 -6.71526730e-01 -3.31329852e-02 3.76357138e-02
-5.72201908e-01 -1.00859809e+00 -5.48435271e-01 3.17249894e-01
-7.72880435e-01 -6.11862004e-01 -6.17376626e-01 -6.93020701e-01
5.51836729e-01 1.74248368e-01 1.20033193e+00 -6.68808371e-02
1.08139805e-01 5.55618256e-02 -2.27610350e-01 -7.16011524e-01
-3.63907069e-01 4.72268701e-01 3.47675890e-01 -6.74544692e-01
8.18874717e-01 -4.37155008e-01 -2.32501626e-01 3.13468762e-02
-6.21303678e-01 3.10395181e-01 9.81249511e-01 1.11238146e+00
6.30123019e-01 2.25225002e-01 6.71711028e-01 -9.05187428e-01
5.42574465e-01 -2.76808321e-01 -4.82577652e-01 3.71611625e-01
-9.28612232e-01 2.98977137e-01 5.64079583e-01 -7.17371464e-01
-1.00812507e+00 -3.63455981e-01 -2.01443911e-01 -6.59945130e-01
2.24108338e-01 6.46688640e-01 -2.79986411e-01 1.28686830e-01
6.22649312e-01 4.37028617e-01 -3.02433193e-01 -7.47336924e-01
6.14973426e-01 6.52268410e-01 4.67676401e-01 -1.25970030e+00
9.47846651e-01 -3.46375465e-01 -3.97153586e-01 -6.00319624e-01
-1.16330087e+00 -1.59667328e-01 -4.42266852e-01 2.20637679e-01
4.38697278e-01 -1.17591333e+00 -7.77820706e-01 2.32993945e-01
-1.07663417e+00 -4.97911364e-01 -4.67638761e-01 7.24150538e-01
-3.68473411e-01 3.16648245e-01 -8.24237764e-01 -5.19681633e-01
-1.13755763e-01 -1.40350425e+00 5.12393653e-01 3.06768417e-01
-6.43306412e-03 -7.31319785e-01 -9.68402848e-02 9.09625709e-01
2.00068846e-01 -6.72311723e-01 1.54739809e+00 -8.67448211e-01
-6.59542322e-01 2.83457488e-01 -1.03381477e-01 7.10331321e-01
-2.68536925e-01 -3.66034031e-01 -9.87011850e-01 -2.49604955e-01
-1.99551225e-01 -5.38222492e-01 8.71362031e-01 2.38726273e-01
1.20486021e+00 -5.21836400e-01 -1.97415158e-01 4.46129233e-01
1.22463405e+00 -4.60967682e-02 3.58001709e-01 1.57930806e-01
7.96187401e-01 4.55881178e-01 4.95457798e-01 5.79080209e-02
6.83606625e-01 6.92110479e-01 2.26587579e-01 1.05820470e-01
-3.77466291e-01 -8.12179446e-01 6.66545689e-01 1.55113423e+00
-7.74324238e-02 -8.05963725e-02 -9.23520148e-01 8.28161001e-01
-1.47490895e+00 -6.50140107e-01 -1.15084521e-01 2.28929615e+00
1.76266420e+00 1.64516881e-01 -1.76111966e-01 1.35860341e-02
5.67914903e-01 -2.81554222e-01 -3.02427858e-01 -5.44594049e-01
-3.30546647e-02 7.23610997e-01 6.55132830e-01 6.06428087e-01
-5.34897387e-01 1.32894242e+00 5.41330671e+00 1.68878770e+00
-8.54201019e-01 4.16827679e-01 4.73896325e-01 -4.92119193e-02
-6.37068033e-01 -1.56123087e-01 -1.43384075e+00 5.42911410e-01
1.11747789e+00 -1.76117748e-01 5.51877797e-01 6.52370512e-01
1.07730245e-02 6.23398721e-02 -9.48590755e-01 7.03949630e-01
-1.01834178e-01 -8.65494132e-01 1.76825851e-01 2.45524004e-01
1.12665176e+00 4.23652586e-03 3.94608617e-01 1.06709087e+00
8.40257466e-01 -1.13632500e+00 6.86083794e-01 1.72762170e-01
9.70929384e-01 -9.14459050e-01 5.61618268e-01 9.19438362e-01
-6.77545130e-01 1.25314057e-01 -5.34770727e-01 -8.87435675e-02
-4.12059814e-01 8.22025180e-01 -1.24033403e+00 6.98961079e-01
5.06897748e-01 5.32225855e-02 -3.45468611e-01 6.30475163e-01
-7.36540377e-01 1.18535984e+00 -3.36663395e-01 -1.16248121e-02
2.94221371e-01 -9.44391191e-02 2.80516624e-01 1.24134791e+00
4.36940044e-01 2.62612134e-01 1.21586159e-01 7.77932525e-01
-5.34611404e-01 1.54417381e-01 -4.79590893e-01 -6.29140511e-02
7.06523776e-01 8.62315178e-01 1.64824259e-02 -3.54164064e-01
-4.60910290e-01 8.29385698e-01 6.87734127e-01 3.43552768e-01
-8.80356252e-01 -2.64708996e-01 7.03089952e-01 1.30223054e-02
2.95511931e-01 -1.32007733e-01 -2.18911842e-01 -1.22226322e+00
-6.56812042e-02 -1.14562333e+00 9.67839733e-02 -5.38422942e-01
-1.05978608e+00 1.24070600e-01 1.56607851e-01 -7.68128097e-01
-2.98446745e-01 -5.83408594e-01 -1.55112743e-01 1.20292544e+00
-1.80620778e+00 -1.01917827e+00 2.83780038e-01 5.59395432e-01
5.00384629e-01 -1.13405161e-01 7.41974831e-01 5.23082197e-01
-6.27358496e-01 1.28816473e+00 2.95043141e-01 2.17023522e-01
7.99226642e-01 -1.37594724e+00 5.20133711e-02 8.20288002e-01
3.90138656e-01 6.74974859e-01 5.40838480e-01 -6.57952070e-01
-1.43913329e+00 -8.94361854e-01 1.41107321e+00 -6.98608935e-01
5.92764914e-01 -2.30052695e-01 -8.87107015e-01 6.09640658e-01
1.00189269e-01 -3.63072544e-01 8.37585926e-01 2.96277136e-01
-3.92099828e-01 -3.20362784e-02 -9.97688890e-01 6.10459864e-01
9.45689499e-01 -5.27390063e-01 -8.54337275e-01 2.08806425e-01
1.12096512e+00 -6.70549631e-01 -9.35315669e-01 5.41840613e-01
3.55505139e-01 -5.09895682e-01 8.75727117e-01 -7.10775018e-01
6.87653005e-01 8.20150226e-02 -1.41644835e-01 -1.45785725e+00
-3.33023854e-02 -6.18235230e-01 -3.27650160e-01 1.50727332e+00
6.96302414e-01 -2.69876033e-01 8.95665050e-01 3.40486795e-01
-3.71143520e-01 -9.56052840e-01 -1.11742306e+00 -9.10136342e-01
8.67163360e-01 -7.09364653e-01 7.13734329e-01 1.31682765e+00
3.06604002e-02 3.06513220e-01 -2.93570966e-01 -1.43128298e-02
3.68184686e-01 -1.47233337e-01 6.05748534e-01 -1.08668613e+00
-4.93135571e-01 -3.80365700e-01 1.30632132e-01 -1.54244137e+00
3.03904146e-01 -1.22856247e+00 7.01738894e-02 -1.23571277e+00
3.82160544e-01 -8.08054924e-01 -5.38826823e-01 9.01806533e-01
-4.68307644e-01 7.31666163e-02 1.62207842e-01 -1.38618857e-01
-2.28722692e-01 5.26091039e-01 1.71323359e+00 -5.99608012e-02
4.30583432e-02 -6.71194643e-02 -8.41115475e-01 4.72225696e-01
6.12946510e-01 -5.67109644e-01 -6.07008815e-01 -7.11683869e-01
2.57585108e-01 2.02841628e-02 1.00573696e-01 -4.97338295e-01
2.52047777e-01 -3.91520411e-01 9.52110663e-02 -5.70973158e-01
2.37602770e-01 -6.71421528e-01 -4.34267461e-01 3.42765212e-01
-3.15217465e-01 -1.01483218e-01 3.62759709e-01 2.35050127e-01
-5.54321818e-02 -5.33308685e-01 6.15439713e-01 -2.11631954e-01
-4.73653346e-01 1.48708493e-01 -4.14289460e-02 5.45038283e-01
5.40639997e-01 3.23231611e-03 -5.10319233e-01 -1.06122747e-01
-3.95639300e-01 3.45540136e-01 3.89883220e-02 3.57511908e-01
3.39853048e-01 -1.33684838e+00 -9.26024258e-01 4.45673883e-01
-8.99349377e-02 2.76656151e-01 2.05452994e-01 9.43224072e-01
-1.13823362e-01 7.78420150e-01 2.37912744e-01 -4.39957947e-01
-1.12791502e+00 3.90959471e-01 2.63319742e-02 -5.27987599e-01
-1.72072142e-01 1.35443997e+00 1.45734876e-01 -7.52693772e-01
3.10301453e-01 -5.14405251e-01 1.11259393e-01 -2.41573807e-02
2.28706300e-01 3.69566172e-01 2.18340635e-01 -4.35090810e-01
3.85321379e-02 3.52221876e-01 -3.95344585e-01 -2.19035506e-01
9.61642385e-01 7.27122054e-02 5.09039760e-02 1.67053565e-01
7.96704829e-01 2.88825691e-01 -8.95980239e-01 -5.76776981e-01
3.97493392e-02 -3.16125244e-01 5.84869683e-02 -1.28358018e+00
-9.40741301e-01 1.27420378e+00 2.90150493e-01 -3.46703470e-01
1.02045584e+00 -1.68247610e-01 8.68912399e-01 3.44337940e-01
3.25205028e-01 -1.16145980e+00 -8.37083831e-02 7.94055283e-01
2.87425131e-01 -1.05792320e+00 -1.55102164e-01 -3.58753979e-01
-4.00300026e-01 6.30924821e-01 8.68136048e-01 5.98244190e-01
2.82775849e-01 2.22330406e-01 -2.20656782e-01 2.39227921e-01
-8.83907616e-01 -3.44276518e-01 3.61863106e-01 4.25654948e-01
6.52154922e-01 3.16251397e-01 -5.37719250e-01 1.14212346e+00
-7.06311464e-01 1.30056992e-01 4.36192662e-01 6.17106318e-01
-6.75594509e-01 -1.53659785e+00 -1.57648355e-01 3.33643287e-01
-9.21649039e-01 -6.92939043e-01 -1.24597915e-01 8.03593695e-01
4.81253147e-01 8.43107581e-01 -5.08994579e-01 -5.34791648e-01
3.29486161e-01 5.55357754e-01 7.41873622e-01 -7.77834952e-01
-6.87098265e-01 -3.61862220e-02 -1.94680374e-02 3.44787119e-03
9.41355154e-02 -3.64164084e-01 -1.13006270e+00 -4.93002385e-01
-6.05002224e-01 4.73917454e-01 6.08603597e-01 1.04413652e+00
1.27083570e-01 4.21863407e-01 2.84156889e-01 7.15354085e-02
-1.05781078e+00 -1.17726278e+00 -3.95277441e-01 -3.98253053e-02
1.05148152e-01 -5.56973755e-01 -3.08722615e-01 -9.68279243e-02] | [10.972457885742188, 8.342584609985352] |
b637e8d4-a080-4c0b-8bc2-943c888e8883 | an-improved-approach-of-intention-discovery | 2009.09354 | null | https://arxiv.org/abs/2009.09354v1 | https://arxiv.org/pdf/2009.09354v1.pdf | An Improved Approach of Intention Discovery with Machine Learning for POMDP-based Dialogue Management | An Embodied Conversational Agent (ECA) is an intelligent agent that works as the front end of software applications to interact with users through verbal/nonverbal expressions and to provide online assistance without the limits of time, location, and language. To help to improve the experience of human-computer interaction, there is an increasing need to empower ECA with not only the realistic look of its human counterparts but also a higher level of intelligence. This thesis first highlights the main topics related to the construction of ECA, including different approaches of dialogue management, and then discusses existing techniques of trend analysis for its application in user classification. As a further refinement and enhancement to prior work on ECA, this thesis research proposes a cohesive framework to integrate emotion-based facial animation with improved intention discovery. In addition, a machine learning technique is introduced to support sentiment analysis for the adjustment of policy design in POMDP-based dialogue management. The proposed research work is going to improve the accuracy of intention discovery while reducing the length of dialogues. | ['Ruturaj Raval'] | 2020-09-20 | null | null | null | null | ['dialogue-management'] | ['natural-language-processing'] | [-1.11010328e-01 5.09231031e-01 -1.08388849e-01 -4.57165152e-01
1.07384905e-01 -5.21447599e-01 6.80260122e-01 2.48054206e-01
-1.96238682e-01 4.47378188e-01 4.46829915e-01 -2.15059727e-01
3.74084748e-02 -6.45495117e-01 4.63076055e-01 -3.40956181e-01
2.15300828e-01 4.34475064e-01 -3.14270675e-01 -9.47894335e-01
4.66316670e-01 4.23260540e-01 -1.49837506e+00 2.87947387e-01
3.72687161e-01 6.24111176e-01 -7.83069730e-02 7.20181465e-01
-3.61349642e-01 1.02450836e+00 -7.20592678e-01 -3.45207900e-01
-2.84332961e-01 -5.69040656e-01 -1.00767696e+00 3.86535585e-01
-5.82060397e-01 -4.25918877e-01 2.45622724e-01 5.81478655e-01
5.90965748e-01 4.57176983e-01 4.68325198e-01 -1.40960634e+00
4.79993969e-02 4.47855264e-01 -2.30634913e-01 -1.44575477e-01
1.08037961e+00 -6.47603646e-02 5.80379367e-01 -2.06258059e-01
6.15904212e-01 1.49540007e+00 3.22108388e-01 7.71280646e-01
-5.16941726e-01 -4.11861300e-01 3.41017574e-01 1.79260895e-02
-7.78512359e-01 -3.35159630e-01 8.71419132e-01 -2.65398294e-01
1.02352023e+00 4.48221922e-01 8.84174168e-01 8.91997695e-01
-3.51590589e-02 6.73969924e-01 1.07352173e+00 -8.94280016e-01
4.50837076e-01 6.97265625e-01 5.24378836e-01 6.85443521e-01
-4.11500722e-01 -5.57846308e-01 -5.51588595e-01 -4.77197021e-01
2.30049759e-01 -2.90335476e-01 3.49893942e-02 -1.20579556e-01
-6.78948283e-01 1.17576575e+00 -2.63487577e-01 4.65426654e-01
-5.72260320e-01 -6.69252872e-01 7.35544026e-01 1.91805795e-01
5.40966511e-01 5.56617796e-01 -4.68789607e-01 -9.59680319e-01
-1.57443076e-01 3.00336272e-01 1.34083211e+00 3.20802122e-01
1.84985578e-01 -4.71628904e-02 -3.08032110e-02 7.76773036e-01
6.15169406e-01 6.01600893e-02 6.54281259e-01 -1.21450293e+00
-2.55081236e-01 1.17784286e+00 3.48248988e-01 -1.22736490e+00
-7.05572367e-01 1.93518028e-01 -1.26733109e-01 4.55404133e-01
2.63396889e-01 -5.32742381e-01 -9.74601060e-02 1.51333225e+00
7.65086293e-01 -4.85219419e-01 3.10759157e-01 5.29896557e-01
8.18878591e-01 7.40484416e-01 1.17861189e-01 -5.96610904e-01
1.70440876e+00 -7.98511088e-01 -1.19049942e+00 1.35268927e-01
8.73920381e-01 -6.34208202e-01 9.35326934e-01 4.18274701e-01
-8.35676193e-01 -1.44033013e-02 -8.49453092e-01 2.44676054e-01
-6.09957457e-01 1.42334951e-02 8.05542409e-01 1.05574918e+00
-7.93237090e-01 1.99631214e-01 -5.16501188e-01 -7.05022573e-01
-3.14549327e-01 5.99717319e-01 -3.72821726e-02 5.63772976e-01
-1.14612365e+00 1.16414511e+00 1.06771514e-02 -3.67869921e-02
3.51841539e-01 -1.42423689e-01 -9.24886227e-01 -2.08352566e-01
1.61079407e-01 -3.97138149e-01 1.57090044e+00 -1.19567502e+00
-2.24279881e+00 7.82647014e-01 2.50612316e-03 -2.85062164e-01
3.29374999e-01 9.29422453e-02 -2.84682721e-01 -5.46532124e-02
-4.68649827e-02 5.80690145e-01 3.72493416e-01 -9.55581248e-01
-7.95921147e-01 -5.42594373e-01 4.39763844e-01 8.50261152e-01
-4.13461328e-01 3.79934788e-01 -2.49072865e-01 -1.05683312e-01
-3.98008019e-01 -9.97687876e-01 -1.11299306e-01 -4.14078951e-01
1.89365298e-01 -6.93138957e-01 1.26451492e+00 -6.55484498e-01
1.14832377e+00 -1.85627282e+00 2.32045054e-02 3.35877240e-01
-6.75461292e-02 3.62300426e-01 2.66835809e-01 5.96567273e-01
1.76809296e-01 -1.32052660e-01 2.70551354e-01 -1.68806404e-01
3.17347348e-01 2.33067319e-01 1.24233045e-01 1.79256529e-01
-4.23135728e-01 3.02623421e-01 -7.99086571e-01 -2.81164944e-01
5.72527111e-01 5.38286388e-01 -2.37230122e-01 1.37155160e-01
-1.60880432e-01 4.28669870e-01 -4.58263457e-01 2.45076254e-01
1.36382297e-01 8.80242288e-02 5.47596753e-01 1.63410425e-01
-4.28633869e-01 3.31182152e-01 -8.79846096e-01 1.15639901e+00
-6.93096280e-01 6.28780067e-01 5.66789865e-01 -8.14403892e-01
1.02544141e+00 6.63146973e-01 7.73014963e-01 -5.80416679e-01
6.91327810e-01 -2.83739656e-01 -5.78627177e-02 -7.61098683e-01
6.32133603e-01 2.05593809e-01 -2.75448054e-01 9.17278945e-01
-5.75424671e-01 1.43658826e-02 2.75708586e-02 1.53307930e-01
7.15865791e-01 1.11317290e-02 4.78491604e-01 -4.14023884e-02
8.51235867e-01 3.37965414e-02 1.47786975e-01 7.58363754e-02
-5.28912246e-01 -4.70545709e-01 5.81146002e-01 -3.95810932e-01
-4.06911522e-01 1.39876038e-01 2.83767909e-01 1.49884927e+00
1.83761939e-02 -2.80043125e-01 -1.17440474e+00 -4.26320165e-01
-4.77557480e-01 1.02323878e+00 -4.37430292e-01 1.02995791e-01
-1.44725800e-01 -6.64672136e-01 2.10092589e-01 -1.41067833e-01
7.34738410e-01 -1.24826324e+00 -9.74001706e-01 3.03174525e-01
-3.65584552e-01 -9.74661112e-01 -3.99862826e-02 -1.71127766e-01
-4.06190515e-01 -7.99618065e-01 -2.56718516e-01 -8.22826207e-01
4.33717787e-01 1.92371786e-01 4.68620420e-01 1.92474201e-02
3.34492922e-02 9.41673994e-01 -7.96794772e-01 -8.38004410e-01
-6.90795779e-01 5.52972928e-02 9.37205851e-02 -8.66280720e-02
5.96930504e-01 -3.15680623e-01 -3.54916930e-01 3.71931583e-01
-4.91587162e-01 3.06313217e-01 -2.31138632e-01 3.37338507e-01
-4.50078547e-01 3.27743977e-01 5.55144668e-01 -7.29896307e-01
1.55786312e+00 -3.22930634e-01 -1.87896058e-01 8.23292360e-02
-7.58949339e-01 -1.38467371e-01 1.29052997e-01 -2.46976286e-01
-1.44066119e+00 -2.81174891e-02 -2.33217433e-01 4.90533680e-01
-3.62786114e-01 4.43930656e-01 -6.87766075e-02 -1.50913984e-01
3.19631726e-01 -2.88456619e-01 7.39159703e-01 3.58778387e-02
9.26688835e-02 1.21618986e+00 8.60582814e-02 -3.49115640e-01
-9.83415693e-02 2.92618603e-01 -4.04319286e-01 -1.20326197e+00
-3.04539055e-01 -5.64647734e-01 -4.52760398e-01 -9.60649252e-01
8.42518806e-01 -5.50428152e-01 -1.32320738e+00 5.45969248e-01
-1.01261353e+00 -4.09145415e-01 1.84116155e-01 3.39479685e-01
-3.01612645e-01 4.06860650e-01 -5.67172587e-01 -1.29232311e+00
-6.48138225e-01 -9.03595388e-01 4.09015208e-01 5.64409018e-01
-1.09857357e+00 -1.24877632e+00 1.51392281e-01 1.07119918e+00
2.76882529e-01 2.29835913e-01 8.09713900e-01 -7.43866026e-01
4.87764418e-01 -4.22421813e-01 3.60487431e-01 5.62999174e-02
2.77433485e-01 2.37989545e-01 -7.11377442e-01 5.45497686e-02
2.78325170e-01 -3.28720987e-01 -4.60730076e-01 3.15012455e-01
1.89997941e-01 -4.81118798e-01 -5.57999536e-02 -4.29594368e-01
5.74827611e-01 9.46358740e-01 4.31183219e-01 6.94702566e-01
-3.52321612e-03 1.14891946e+00 1.12851036e+00 9.07677710e-01
8.11264575e-01 7.39359856e-01 2.35041961e-01 1.35330241e-02
5.63786089e-01 1.24927081e-01 6.35601342e-01 7.55707979e-01
-3.05410266e-01 -2.04020470e-01 -6.83931828e-01 8.44233036e-02
-2.05771518e+00 -9.45148289e-01 -9.32227150e-02 1.58620143e+00
4.46287602e-01 -1.12240650e-01 4.55277443e-01 4.19725806e-01
6.96277261e-01 6.12161793e-02 -2.43710175e-01 -1.42935491e+00
4.27588284e-01 -6.47069663e-02 -7.84999516e-04 7.37642169e-01
-8.64314377e-01 8.27080607e-01 5.72054291e+00 2.28931561e-01
-1.11184335e+00 -7.60237202e-02 6.14517808e-01 3.37265193e-01
9.89159718e-02 -3.45292360e-01 -3.82401466e-01 1.71699315e-01
8.60299468e-01 -2.13324472e-01 5.31075418e-01 9.05855119e-01
8.29181314e-01 -5.62598526e-01 -5.06438613e-01 7.17304766e-01
1.33603781e-01 -9.88181293e-01 -3.70069265e-01 5.77761009e-02
3.22340220e-01 -7.82933414e-01 -9.64924693e-02 3.24794680e-01
9.14643332e-02 -5.44955850e-01 3.06632340e-01 2.73031086e-01
2.46626902e-02 -9.41440582e-01 1.06262827e+00 4.82001990e-01
-7.82046676e-01 1.29650405e-03 3.06576580e-01 -7.84546256e-01
8.13992620e-02 -5.51827013e-01 -1.22003257e+00 5.61510138e-02
4.74394143e-01 9.04231295e-02 -1.31360861e-02 2.61699796e-01
1.92508861e-01 2.49504268e-01 -1.72071412e-01 -7.94205308e-01
1.25883296e-01 -4.35159683e-01 4.95937407e-01 9.63881612e-01
-8.58201161e-02 7.66935468e-01 1.90533608e-01 4.98916321e-02
3.70574743e-01 5.85460126e-01 -5.12988389e-01 -1.26026481e-01
3.75118643e-01 1.42008364e+00 -1.01872361e+00 -4.59926501e-02
-4.13931072e-01 9.65924561e-01 -2.58280784e-01 2.25501414e-02
-5.17215073e-01 -2.42219970e-01 5.57420731e-01 6.61337078e-02
-2.92510927e-01 -1.23104960e-01 -2.51629680e-01 -4.24378067e-01
-4.26662356e-01 -1.14382935e+00 2.84918159e-01 -7.64776409e-01
-5.60710728e-01 3.45849633e-01 4.52067517e-02 -6.63744211e-01
-7.10985243e-01 -4.83797133e-01 -8.32505643e-01 4.74571675e-01
-6.61013603e-01 -1.09745693e+00 -2.25951672e-01 4.31920886e-01
7.40501583e-01 -3.44928980e-01 1.38260400e+00 -2.52186418e-01
-6.61165178e-01 2.13014498e-01 -3.72841150e-01 -2.23813951e-02
6.16959691e-01 -8.00416231e-01 -2.95547605e-01 2.70314336e-01
-6.31845832e-01 3.60831141e-01 1.09783745e+00 -4.04441357e-01
-1.27013588e+00 -1.79634243e-01 9.08354223e-01 -2.65928600e-02
5.15494406e-01 5.20820469e-02 -4.92677957e-01 1.90185547e-01
7.27911770e-01 -1.13783479e+00 1.10902631e+00 2.82334536e-01
4.12888974e-01 5.71963452e-02 -1.50841308e+00 7.61367440e-01
1.16723768e-01 -4.34709847e-01 -6.04102254e-01 1.64893806e-01
2.24308550e-01 -4.67606634e-01 -7.69204557e-01 -8.89331996e-02
7.28498995e-01 -1.13369191e+00 5.83665431e-01 -2.95382559e-01
-1.97618902e-02 8.02665204e-02 2.52800018e-01 -9.73328650e-01
1.83419496e-01 -1.05993986e+00 2.55547404e-01 1.24893892e+00
2.97199398e-01 -6.34682178e-01 9.39437032e-01 1.55364907e+00
1.95481449e-01 -6.80552602e-01 -4.79479879e-01 1.32516831e-01
-4.17428344e-01 -4.42093045e-01 3.53240520e-01 1.10970283e+00
1.01256788e+00 6.45595014e-01 -2.77907431e-01 -6.14162125e-02
7.72115290e-02 -2.29819998e-01 8.73693228e-01 -1.25095737e+00
1.49492860e-01 -4.43641841e-01 -2.78280944e-01 -5.16496301e-01
2.52114236e-01 -2.06094235e-01 -1.77746221e-01 -1.55815768e+00
-1.66948885e-01 -1.13488995e-02 3.26856017e-01 4.81868714e-01
3.37074846e-01 -2.65545636e-01 1.69882774e-01 -1.61396474e-01
-4.15904939e-01 3.53974491e-01 9.50544477e-01 2.41861701e-01
-8.43920410e-01 4.57879364e-01 -7.23399878e-01 1.04557824e+00
1.17745364e+00 3.19931805e-02 -6.76468551e-01 5.01253068e-01
4.09291178e-01 4.97362524e-01 -1.38810515e-01 -5.09497404e-01
2.79032350e-01 -3.16812485e-01 -3.48585621e-02 -4.98357475e-01
4.64532316e-01 -9.26380277e-01 -9.64124948e-02 3.84005845e-01
-4.50469494e-01 2.02835619e-01 1.78284138e-01 1.27443150e-01
-6.06840216e-02 -6.22798562e-01 5.17216980e-01 -1.46916300e-01
-7.08216250e-01 -4.90243226e-01 -1.48664439e+00 -5.65769553e-01
1.35667968e+00 -4.51017857e-01 4.87871468e-03 -1.24976146e+00
-6.28997803e-01 3.72531623e-01 3.31675142e-01 5.59216499e-01
3.26697737e-01 -7.60508060e-01 -7.29988962e-02 -1.09551542e-01
-2.63782889e-01 -4.33517814e-01 1.86029613e-01 5.67912400e-01
-5.67389071e-01 4.51860964e-01 -5.29311955e-01 9.89868771e-03
-2.08313036e+00 2.29388420e-02 1.50655523e-01 -1.90290034e-01
-2.33593166e-01 3.96101505e-01 -5.42172432e-01 -5.35609722e-01
6.77695751e-01 2.03730226e-01 -9.67066050e-01 5.72511196e-01
5.52460134e-01 5.88799357e-01 -1.41645759e-01 -8.06896269e-01
-1.94494218e-01 6.75834045e-02 6.89908490e-02 -5.63712120e-01
1.23171508e+00 -5.70594370e-01 -3.09993625e-01 5.23940444e-01
6.68141246e-01 9.70166773e-02 -6.60547376e-01 2.91885108e-01
2.95237321e-02 5.32143712e-02 9.19179693e-02 -9.17115867e-01
-4.34230417e-01 3.69087368e-01 6.77685797e-01 5.96382618e-01
9.72944081e-01 -4.26711291e-01 4.42135990e-01 4.97624725e-01
1.93910241e-01 -1.72129536e+00 1.43970355e-01 5.95403314e-01
6.35345757e-01 -1.21955323e+00 1.27803817e-01 -2.32464820e-01
-1.29417145e+00 1.43443322e+00 5.03801346e-01 6.04938269e-01
6.25334084e-01 2.92180389e-01 4.88247871e-01 -3.92081112e-01
-7.05288529e-01 -1.55435465e-02 -1.48818776e-01 6.48116708e-01
6.49672329e-01 1.49155006e-01 -6.94288909e-01 6.73550785e-01
-2.89355636e-01 -6.76987246e-02 5.79397798e-01 1.15667713e+00
-5.50104916e-01 -1.22316003e+00 -4.39808577e-01 1.33976698e-01
-4.47305441e-01 4.94615793e-01 -9.13800716e-01 7.93078125e-01
-4.08353768e-02 1.52653670e+00 -1.13602644e-02 -4.96513665e-01
2.91033506e-01 2.86620349e-01 -5.97180091e-02 -2.57599205e-01
-8.78019869e-01 3.81988264e-03 7.53528476e-01 -1.90633029e-01
-1.00266206e+00 -8.38569820e-01 -1.46230602e+00 -6.70289338e-01
-2.04577163e-01 3.87809128e-01 1.06146574e+00 1.10725141e+00
1.12309627e-01 2.03311369e-01 8.02112937e-01 -7.35694051e-01
-9.39704105e-03 -9.81464505e-01 -3.36736351e-01 2.03605175e-01
-2.13178068e-01 -3.71382087e-01 -2.42343768e-01 -2.38385513e-01] | [13.121038436889648, 7.6531596183776855] |
0cbc4f22-30c0-4da3-8dfe-893ae666a7dc | learning-dynamic-point-cloud-compression-via | 2305.05356 | null | https://arxiv.org/abs/2305.05356v2 | https://arxiv.org/pdf/2305.05356v2.pdf | Learning Dynamic Point Cloud Compression via Hierarchical Inter-frame Block Matching | 3D dynamic point cloud (DPC) compression relies on mining its temporal context, which faces significant challenges due to DPC's sparsity and non-uniform structure. Existing methods are limited in capturing sufficient temporal dependencies. Therefore, this paper proposes a learning-based DPC compression framework via hierarchical block-matching-based inter-prediction module to compensate and compress the DPC geometry in latent space. Specifically, we propose a hierarchical motion estimation and motion compensation (Hie-ME/MC) framework for flexible inter-prediction, which dynamically selects the granularity of optical flow to encapsulate the motion information accurately. To improve the motion estimation efficiency of the proposed inter-prediction module, we further design a KNN-attention block matching (KABM) network that determines the impact of potential corresponding points based on the geometry and feature correlation. Finally, we compress the residual and the multi-scale optical flow with a fully-factorized deep entropy model. The experiment result on the MPEG-specified Owlii Dynamic Human Dynamic Point Cloud (Owlii) dataset shows that our framework outperforms the previous state-of-the-art methods and the MPEG standard V-PCC v18 in inter-frame low-delay mode. | ['Zhu Li', 'Jenq-Neng Hwang', 'Yiling Xu', 'Tingyu Fan', 'Shuting Xia'] | 2023-05-09 | null | null | null | null | ['motion-compensation', 'motion-estimation'] | ['computer-vision', 'computer-vision'] | [-1.22559734e-01 -5.99519789e-01 -4.29688603e-01 -2.03607053e-01
-2.76611865e-01 -4.20285622e-03 2.07428455e-01 -2.46767685e-01
-1.94275841e-01 1.96046948e-01 5.79610825e-01 2.11632866e-02
-2.93681800e-01 -6.13905847e-01 -5.68927109e-01 -7.09688663e-01
-4.24296856e-01 1.21220909e-01 5.30400872e-01 -3.80189791e-02
6.84854329e-01 4.86336648e-01 -1.72321904e+00 6.27188921e-01
8.37276638e-01 1.23562908e+00 6.38765037e-01 7.81153381e-01
-1.16090462e-01 9.44692194e-01 -3.04982550e-02 -1.31429747e-01
5.57162225e-01 -1.86388955e-01 -6.50848389e-01 3.83711867e-02
6.01047277e-01 -9.46739316e-01 -7.56992638e-01 7.67655015e-01
4.90636498e-01 1.92629784e-01 3.08478147e-01 -1.31808960e+00
-2.60273695e-01 4.04145420e-02 -5.26813447e-01 5.72931290e-01
2.88125932e-01 4.71089661e-01 9.70809817e-01 -9.51264203e-01
9.12703276e-01 1.40854645e+00 4.46789235e-01 4.54340786e-01
-7.55197942e-01 -6.43222034e-01 1.08092494e-01 9.13998783e-01
-1.29468620e+00 -3.97307962e-01 9.37776268e-01 -5.35421193e-01
1.16883063e+00 1.13358513e-01 9.59529579e-01 6.82022512e-01
3.63453329e-01 8.03713083e-01 3.86258066e-01 2.07695179e-02
9.53090116e-02 -7.36496210e-01 -2.84265697e-01 7.06396580e-01
-1.55217379e-01 3.79880130e-01 -8.22473407e-01 7.81098679e-02
1.02292633e+00 2.68180877e-01 -5.80356658e-01 -4.21975881e-01
-1.32267594e+00 5.29620469e-01 3.74381185e-01 1.95290148e-02
-3.40644002e-01 3.18429738e-01 5.41876078e-01 6.98450506e-02
3.74270111e-01 -1.49434090e-01 -5.17785132e-01 -5.67910373e-01
-1.31875145e+00 3.28541577e-01 4.81426775e-01 1.22548294e+00
7.09024608e-01 -8.97646099e-02 -1.46014005e-01 5.97161889e-01
4.52172101e-01 3.46683979e-01 6.10461175e-01 -1.58898103e+00
9.28280473e-01 4.74432468e-01 -6.86545670e-03 -1.61102271e+00
-1.45094514e-01 -2.25353494e-01 -9.47405398e-01 -5.78054227e-03
-2.36188620e-01 2.67458320e-01 -5.10598302e-01 1.46832919e+00
4.11968976e-01 8.57613921e-01 -1.57197878e-01 1.25987720e+00
5.47425926e-01 9.69982028e-01 -5.44826686e-02 -5.11762798e-01
9.60350215e-01 -1.17886651e+00 -6.23400211e-01 1.56755492e-01
5.84843874e-01 -7.52878845e-01 7.66425610e-01 1.25831172e-01
-1.24127150e+00 -9.34566438e-01 -1.02804148e+00 -5.47589362e-01
2.65803307e-01 -1.95985034e-01 3.32508445e-01 -3.71214859e-02
-8.35548460e-01 9.62037623e-01 -1.01163387e+00 1.62628755e-01
3.91822934e-01 3.25664043e-01 -1.05163246e-01 -3.21637958e-01
-1.09971845e+00 3.23759735e-01 4.14358795e-01 1.58850223e-01
-6.06801510e-01 -9.96858120e-01 -7.85909951e-01 3.65084618e-01
1.99992791e-01 -9.36467648e-01 7.81701028e-01 -7.29398966e-01
-1.46449244e+00 3.93531561e-01 -4.14195299e-01 -5.07977366e-01
4.99075949e-01 -2.55745769e-01 -2.20781103e-01 7.81028986e-01
1.53866166e-03 1.17166603e+00 9.86964762e-01 -8.33661139e-01
-9.93102252e-01 -1.31749645e-01 -1.47287458e-01 3.70650709e-01
-5.07475920e-02 -2.98631519e-01 -9.23754692e-01 -8.26892257e-01
1.64203510e-01 -8.78182650e-01 -1.00447871e-01 2.90538251e-01
1.33226931e-01 -1.25808731e-01 1.22987282e+00 -7.15457201e-01
1.60546112e+00 -2.22868013e+00 3.41323048e-01 4.27893512e-02
3.50725353e-01 2.75340438e-01 -1.22595221e-01 6.81364238e-02
7.99910054e-02 -1.29200622e-01 -9.24635231e-02 -5.76441944e-01
-1.48896262e-01 3.98332685e-01 -3.40524077e-01 4.60735083e-01
4.36473936e-02 7.57218182e-01 -8.75387669e-01 -9.81011629e-01
6.73975110e-01 5.86517453e-01 -1.17631567e+00 2.53520399e-01
-8.56616274e-02 4.35459614e-01 -3.04815799e-01 6.10489428e-01
1.00537372e+00 -4.38468039e-01 -1.50094435e-01 -5.27300119e-01
-3.56262356e-01 1.48671582e-01 -1.07707071e+00 2.12746429e+00
-2.56820112e-01 6.61266625e-01 -1.27379030e-01 -7.74669051e-01
5.54664552e-01 1.51480854e-01 1.23350775e+00 -5.39553523e-01
-2.68576853e-02 3.57916355e-01 -1.73952073e-01 -6.55729830e-01
7.32839525e-01 4.55536962e-01 4.03693438e-01 -7.81316012e-02
-1.23942800e-01 3.19961935e-01 1.82442710e-01 1.87096223e-01
1.06569147e+00 3.49806160e-01 3.29605266e-02 -7.58234411e-02
9.76373971e-01 -2.74779171e-01 1.12593901e+00 1.02588303e-01
-6.92136943e-01 7.78180480e-01 1.90343991e-01 -5.88256717e-01
-1.24432421e+00 -6.77061617e-01 -5.25681861e-02 5.18724978e-01
6.24107063e-01 -7.92320311e-01 -4.76537019e-01 -3.30948502e-01
-1.15015917e-02 2.42651269e-01 -2.40450874e-01 -2.12602794e-01
-1.16493309e+00 -8.32156837e-02 -7.95889571e-02 3.86940807e-01
8.48455727e-01 -1.05513918e+00 -9.07166541e-01 4.69431251e-01
-7.66988516e-01 -1.51636195e+00 -9.82269406e-01 -5.93523920e-01
-1.13745904e+00 -1.01454699e+00 -5.82178593e-01 -6.84829056e-01
3.12727749e-01 5.84298015e-01 1.02132618e+00 3.21758181e-01
-1.27818108e-01 3.40117157e-01 -4.49822068e-01 4.67395276e-01
-5.85094839e-02 5.11031859e-02 -2.00393707e-01 2.22320873e-02
3.92605841e-01 -8.36344361e-01 -1.44957376e+00 3.50700498e-01
-8.78776073e-01 5.01241565e-01 3.97159547e-01 5.72778940e-01
8.90223801e-01 -1.03578508e-01 -3.83860543e-02 -1.85677826e-01
8.08587857e-03 -5.43711901e-01 -6.74716532e-01 2.25272942e-02
-7.44313240e-01 2.59770099e-02 4.86332685e-01 -3.23555082e-01
-8.39246750e-01 1.25118911e-01 -7.11276308e-02 -1.21533835e+00
2.65977144e-01 2.20443383e-01 -1.74413696e-01 -1.60058662e-01
-1.20061927e-01 5.06802619e-01 -2.62091517e-01 -3.13524604e-01
1.39127523e-01 3.30541462e-01 7.82948971e-01 -3.89439940e-01
5.33208013e-01 5.79697192e-01 3.56696010e-01 -5.86358190e-01
-4.68714386e-01 -7.81732678e-01 -6.94815993e-01 -4.21615988e-01
1.08726311e+00 -1.07028806e+00 -1.02955449e+00 4.66236651e-01
-1.44839072e+00 -1.46451637e-01 -6.10352792e-02 7.06978917e-01
-9.71723616e-01 9.17621315e-01 -8.51254880e-01 -3.28944296e-01
-4.99646544e-01 -1.47161937e+00 1.27578497e+00 -3.86549644e-02
2.38340095e-01 -6.44837320e-01 1.49886161e-01 4.16944057e-01
2.06863031e-01 1.15183026e-01 6.83051646e-01 2.45744407e-01
-1.48242450e+00 2.97616154e-01 -3.45990539e-01 2.88283467e-01
-1.56150937e-01 1.54978275e-01 -3.76105487e-01 -3.94907862e-01
5.98189719e-02 1.62458673e-01 8.35725129e-01 6.43802583e-01
1.55107701e+00 -5.06627381e-01 -2.18029767e-01 1.49300480e+00
1.60762811e+00 -2.86886338e-02 8.34798753e-01 2.69491702e-01
9.69935417e-01 3.60181004e-01 6.93097532e-01 8.85833383e-01
7.34834135e-01 8.56422067e-01 7.29494154e-01 4.84073699e-01
-2.90313244e-01 -3.79935443e-01 4.11698818e-01 1.38837564e+00
-2.70960957e-01 -2.35301018e-01 -4.90402490e-01 4.40612406e-01
-2.03480697e+00 -1.30498469e+00 -9.36184917e-03 1.82517982e+00
6.84803188e-01 3.73081714e-02 -3.39136064e-01 4.77658659e-02
6.41074181e-01 4.52188730e-01 -6.29169703e-01 -1.00739568e-01
-1.66164130e-01 -1.29813269e-01 5.73912740e-01 5.81638992e-01
-1.01087165e+00 7.91425586e-01 4.55649710e+00 9.62094426e-01
-1.14664865e+00 9.82549414e-02 3.55756342e-01 -3.06518734e-01
-2.31796592e-01 1.20795675e-01 -9.37452197e-01 1.06970608e+00
9.13600922e-01 2.34157015e-02 4.98384804e-01 7.71415114e-01
5.31647265e-01 1.80525810e-01 -8.92359197e-01 1.48635948e+00
-1.04270279e-01 -1.82698238e+00 2.36882031e-01 3.66488099e-01
7.03205585e-01 1.65489048e-01 -6.10508621e-02 1.17043249e-01
-3.80060822e-01 -4.11535501e-01 8.43917429e-01 7.97739983e-01
8.22759151e-01 -7.20936477e-01 3.32421452e-01 3.02686334e-01
-1.66117013e+00 -3.20103467e-01 -6.48093164e-01 2.98758239e-01
4.25549567e-01 4.27453190e-01 -1.49615601e-01 5.46439707e-01
9.22270477e-01 1.34330165e+00 -2.41192937e-01 1.07117999e+00
2.20842302e-01 4.13267076e-01 -3.20504457e-01 4.98040825e-01
3.08056265e-01 -1.78550377e-01 7.75832593e-01 1.09869170e+00
5.89277565e-01 4.83130395e-01 1.40782773e-01 5.50707400e-01
8.69454369e-02 1.60594493e-01 -5.18223457e-02 4.65476722e-01
6.00221336e-01 9.07577813e-01 -3.71941388e-01 -4.11480427e-01
-5.88067114e-01 1.26844859e+00 1.16319329e-01 1.96884274e-01
-8.70349169e-01 6.48780987e-02 8.88817906e-01 2.56713420e-01
7.90832877e-01 -5.02552688e-01 -1.28743025e-02 -1.61166477e+00
2.24121496e-01 -5.29666007e-01 3.85645926e-01 -9.33993816e-01
-9.64285314e-01 2.57458925e-01 5.43137081e-02 -1.91540730e+00
-4.53943312e-01 -2.87258297e-01 -3.57224494e-01 6.23377204e-01
-1.89278400e+00 -9.45397735e-01 -5.94975471e-01 1.02681983e+00
8.72602463e-01 -7.90399164e-02 2.67241746e-01 7.71040916e-01
-2.48693779e-01 4.25929278e-01 -5.78271560e-02 -6.50088489e-02
6.71461642e-01 -6.33122206e-01 2.95248091e-01 1.01868320e+00
-3.04674685e-01 3.78225744e-01 2.67916679e-01 -8.80270362e-01
-1.47122312e+00 -1.30231857e+00 8.66609395e-01 1.46308631e-01
2.28593096e-01 1.83853786e-02 -9.60399628e-01 3.20839942e-01
-1.45135999e-01 4.97373968e-01 3.44706059e-01 -7.70635605e-01
-1.55447483e-01 -3.50194544e-01 -8.81465793e-01 4.64447737e-01
1.36496019e+00 -4.41395819e-01 -2.41396248e-01 6.50934279e-02
1.24251235e+00 -4.96791720e-01 -9.41997826e-01 7.08537340e-01
5.88240445e-01 -1.26243770e+00 1.28497386e+00 -2.37324253e-01
9.67880845e-01 -4.74173337e-01 -4.18385446e-01 -5.48765361e-01
-6.02515519e-01 -8.02993834e-01 -8.20056617e-01 9.40601289e-01
-4.44381416e-01 -5.34169842e-03 1.01682675e+00 5.85196495e-01
-4.04875517e-01 -8.61421406e-01 -1.09258556e+00 -3.86370093e-01
-3.89559567e-01 -6.04900122e-01 6.92902446e-01 8.67609262e-01
-3.54799539e-01 -1.20019846e-01 -5.74323654e-01 1.71728939e-01
6.83235407e-01 3.21304202e-01 7.28155553e-01 -8.08082104e-01
-3.63996893e-01 -4.45277125e-01 -7.42799342e-01 -1.68508589e+00
1.45022333e-01 -7.87628055e-01 -1.78495839e-01 -1.08251369e+00
-1.02912061e-01 -3.36541533e-01 -2.37147093e-01 -1.21554382e-01
-1.28044844e-01 -1.66409925e-01 4.63221461e-01 9.11547065e-01
-7.49966502e-01 9.49767768e-01 1.54310381e+00 7.40104495e-03
-1.77766353e-01 -1.61526620e-01 7.72890225e-02 4.98924673e-01
4.27352697e-01 -1.79154843e-01 -5.64020991e-01 -6.06216133e-01
-2.87081786e-02 5.90140343e-01 4.03334886e-01 -1.24619746e+00
5.76868892e-01 -3.28604013e-01 2.83641398e-01 -1.31006777e+00
4.01243657e-01 -1.01332033e+00 3.60610373e-02 5.55761218e-01
-2.46897817e-01 3.98366272e-01 -1.92811027e-01 8.13228667e-01
-3.29737514e-01 -2.21811254e-02 6.70847774e-01 -2.92032287e-02
-1.19942307e+00 1.16588950e+00 4.27221954e-02 1.54601678e-01
6.58746481e-01 -3.56671244e-01 -1.48119986e-01 -2.78030217e-01
-3.25363666e-01 3.25516135e-01 5.80498636e-01 5.13858795e-01
1.12537336e+00 -1.60508072e+00 -6.57579303e-01 4.19402212e-01
1.87519914e-03 2.67650157e-01 7.74097919e-01 7.38363564e-01
-9.44114864e-01 4.08845991e-01 -3.61694902e-01 -1.00351071e+00
-1.10419273e+00 7.59389579e-01 1.64045468e-01 -3.39551389e-01
-9.62655425e-01 5.25115490e-01 1.51637018e-01 2.59059161e-01
1.82854235e-01 -4.48314816e-01 -8.73527154e-02 -3.60315859e-01
4.83452290e-01 6.64688528e-01 -2.17970893e-01 -8.35753202e-01
-2.35007152e-01 8.59003901e-01 1.79611295e-01 7.57537559e-02
1.17397738e+00 -5.52375376e-01 -1.96010634e-01 -8.54189917e-02
1.75865149e+00 -2.22626194e-01 -1.87334824e+00 -1.97432384e-01
-3.04031253e-01 -1.02722788e+00 1.56056449e-01 3.50693837e-02
-1.45569956e+00 9.14295018e-01 7.23136008e-01 -4.93683100e-01
1.31938326e+00 -5.71373165e-01 1.55980873e+00 9.49656889e-02
2.45154783e-01 -1.08318996e+00 6.19900227e-02 5.02478302e-01
7.61957288e-01 -9.90426958e-01 2.90029496e-01 -4.54495549e-01
-3.52646828e-01 1.36680162e+00 7.02273786e-01 -3.11006576e-01
1.04296339e+00 3.00261788e-02 -4.70972538e-01 -9.09889489e-03
-1.12392831e+00 1.84427589e-01 4.02328223e-01 3.69081736e-01
-3.12481518e-03 -3.01030129e-01 -4.56558257e-01 4.31801975e-02
-2.24282116e-01 3.36660296e-01 3.24102223e-01 7.59303927e-01
-4.06211853e-01 -9.79610324e-01 -3.09382789e-02 1.73216373e-01
-2.35990599e-01 -1.85955152e-01 5.07054031e-01 4.61260587e-01
4.37470108e-01 6.07060730e-01 3.26715022e-01 -6.48332298e-01
4.05236110e-02 -4.04505402e-01 3.66601050e-01 8.43309537e-02
-2.50764012e-01 2.99318373e-01 -4.37864363e-01 -1.25223494e+00
-8.51938248e-01 -8.08373690e-01 -1.38650084e+00 -6.59895480e-01
1.73489228e-01 -1.87849760e-01 4.19459254e-01 6.69290960e-01
8.47952724e-01 2.26416752e-01 8.99589300e-01 -1.08568811e+00
-2.01930866e-01 -4.93080378e-01 -1.30475357e-01 6.88240290e-01
4.56726581e-01 -6.06043339e-01 -4.53714043e-01 3.13556403e-01] | [10.964595794677734, -1.6671537160873413] |
b87ef0d0-eab8-4814-bf6e-b410a49c52ae | alternating-between-spectral-and-spatial | 1911.07953 | null | https://arxiv.org/abs/1911.07953v3 | https://arxiv.org/pdf/1911.07953v3.pdf | Sequential Multi-Frame Neural Beamforming for Speech Separation and Enhancement | This work introduces sequential neural beamforming, which alternates between neural network based spectral separation and beamforming based spatial separation. Our neural networks for separation use an advanced convolutional architecture trained with a novel stabilized signal-to-noise ratio loss function. For beamforming, we explore multiple ways of computing time-varying covariance matrices, including factorizing the spatial covariance into a time-varying amplitude component and a time-invariant spatial component, as well as using block-based techniques. In addition, we introduce a multi-frame beamforming method which improves the results significantly by adding contextual frames to the beamforming formulations. We extensively evaluate and analyze the effects of window size, block size, and multi-frame context size for these methods. Our best method utilizes a sequence of three neural separation and multi-frame time-invariant spatial beamforming stages, and demonstrates an average improvement of 2.75 dB in scale-invariant signal-to-noise ratio and 14.2% absolute reduction in a comparative speech recognition metric across four challenging reverberant speech enhancement and separation tasks. We also use our three-speaker separation model to separate real recordings in the LibriCSS evaluation set into non-overlapping tracks, and achieve a better word error rate as compared to a baseline mask based beamformer. | ['Zhuo Chen', 'Shinji Watanabe', 'Desh Raj', 'Zhong-Qiu Wang', 'Scott Wisdom', 'John R. Hershey', 'Kevin Wilson', 'Hakan Erdogan'] | 2019-11-18 | null | null | null | null | ['speaker-separation'] | ['speech'] | [ 5.38656890e-01 -6.40095472e-01 5.68392336e-01 -3.55674833e-01
-1.48902464e+00 -7.67719507e-01 4.18136269e-01 -2.35542595e-01
-6.58596814e-01 4.07591522e-01 7.98250079e-01 -4.67633933e-01
-3.11704844e-01 -1.18919358e-01 -4.44744080e-01 -9.43159401e-01
-2.74734616e-01 -5.30278683e-01 8.15518275e-02 -1.82617530e-01
-1.20196581e-01 5.89150727e-01 -1.25348556e+00 4.02620971e-01
6.77092433e-01 1.01295245e+00 1.87482938e-01 1.22422349e+00
3.83368045e-01 2.64612406e-01 -9.41579640e-01 8.05615038e-02
4.74696189e-01 -4.82487381e-01 -2.18146637e-01 -1.58491269e-01
9.85845387e-01 -2.20461965e-01 -4.77819324e-01 9.05899823e-01
1.10412037e+00 6.36986613e-01 3.50907683e-01 -6.41330421e-01
-3.50518942e-01 7.11070299e-01 -5.24603009e-01 7.54909277e-01
2.93309718e-01 -1.48754772e-02 7.74057746e-01 -1.07400501e+00
-1.33592129e-01 1.11751568e+00 1.06317985e+00 2.03708664e-01
-1.30919802e+00 -8.60924184e-01 -1.95765514e-02 2.44379684e-01
-1.47481084e+00 -1.12964213e+00 5.60392857e-01 -2.73971200e-01
1.47135413e+00 6.35909736e-01 2.43914157e-01 8.07046592e-01
5.31544536e-02 4.71663624e-01 1.02153158e+00 -6.78309083e-01
-6.83108196e-02 -5.12260139e-01 1.93323508e-01 2.72144020e-01
-1.72142267e-01 3.42373073e-01 -8.50641966e-01 -2.21744142e-02
4.83828336e-01 -3.82362902e-01 -7.71072030e-01 3.21631640e-01
-1.16619122e+00 4.27293330e-01 3.02866399e-01 4.97248858e-01
-1.31413251e-01 1.73603103e-01 1.84490457e-01 2.22498655e-01
6.87487662e-01 5.26853204e-01 -5.33169925e-01 -5.43771945e-02
-1.59247029e+00 1.20113522e-01 5.95768869e-01 4.23736989e-01
1.23460405e-01 7.10332334e-01 -5.56977928e-01 1.16227102e+00
3.95256281e-01 7.40726411e-01 5.16207814e-01 -9.68759179e-01
4.89058852e-01 -6.21178985e-01 6.59677014e-02 -8.52973759e-01
-6.18110716e-01 -1.06843495e+00 -6.63946927e-01 2.40563110e-01
4.42950726e-01 -5.71247220e-01 -1.14840400e+00 2.05927324e+00
5.87375462e-02 5.87798893e-01 5.02057560e-02 1.17677367e+00
7.28024662e-01 7.65882254e-01 -3.84116322e-01 -4.52766269e-01
1.18907988e+00 -1.27631235e+00 -9.57078695e-01 -2.05777168e-01
-5.35904383e-03 -1.17925954e+00 7.03186870e-01 6.34326041e-01
-1.51206481e+00 -6.62612796e-01 -1.35536945e+00 4.47715484e-02
-1.55206874e-01 2.00825721e-01 3.30526605e-02 1.17010176e+00
-1.49717736e+00 5.22993684e-01 -8.16446126e-01 4.71542738e-02
4.49274434e-03 3.66259366e-01 -2.59131998e-01 4.62384149e-02
-1.16502857e+00 6.73324585e-01 -1.94834560e-01 2.03324765e-01
-6.42835140e-01 -9.39052641e-01 -9.12549257e-01 4.89501417e-01
-5.33517497e-03 -5.05659759e-01 1.38309228e+00 -9.35156584e-01
-1.44194448e+00 1.15703829e-01 -6.28050566e-01 -6.78389072e-01
1.75995737e-01 -5.39999068e-01 -1.06674862e+00 6.15982898e-02
-2.22308069e-01 2.01777503e-01 1.22523451e+00 -9.35981274e-01
-4.79736626e-01 2.26227157e-02 -3.21737051e-01 3.25547069e-01
-5.48938990e-01 4.23066676e-01 -2.40833730e-01 -1.35086346e+00
2.95606405e-01 -5.77835977e-01 -8.29038247e-02 -3.68131518e-01
-7.96069577e-02 6.57546520e-01 5.06195188e-01 -1.33397639e+00
1.53517747e+00 -2.44404650e+00 1.07792906e-01 1.26156673e-01
1.65245354e-01 3.97590101e-01 -5.31679213e-01 5.41597009e-02
-4.87378478e-01 -3.66889685e-02 -2.84130901e-01 -6.07695758e-01
-1.79931015e-01 -5.60988426e-01 -3.12032044e-01 4.66405243e-01
1.63899645e-01 1.95454806e-01 -6.41899824e-01 2.33426914e-01
1.26172632e-01 1.05949342e+00 -7.23904312e-01 4.39692363e-02
6.16815984e-01 2.57441849e-01 5.87096810e-01 3.36666077e-01
8.47072005e-01 3.46640050e-01 -4.79756929e-02 -5.14963388e-01
-2.24433780e-01 5.82128763e-01 -1.39478827e+00 1.73273408e+00
-7.22039938e-01 1.18594062e+00 8.14757466e-01 -4.89108324e-01
5.28669894e-01 5.63044667e-01 3.06931138e-01 -5.17424107e-01
2.93347128e-02 3.43650311e-01 3.50332350e-01 7.20748305e-02
4.25962776e-01 -2.74729580e-01 4.79873389e-01 2.25813121e-01
2.68790185e-01 1.60496924e-02 6.36984259e-02 -1.44055650e-01
1.42448390e+00 -2.78996736e-01 -1.66571572e-01 -4.03333336e-01
5.25329530e-01 -7.44397342e-01 2.87028342e-01 6.04697526e-01
-2.25222230e-01 1.10680807e+00 -2.65442014e-01 1.44302726e-01
-5.90987563e-01 -1.40279543e+00 -1.05446205e-01 1.22756493e+00
-2.37214863e-01 -4.95749503e-01 -7.98297644e-01 -8.23872387e-02
-3.06119680e-01 9.17679071e-01 -2.70145118e-01 -8.50479752e-02
-8.66252005e-01 -8.19518805e-01 7.15257585e-01 6.31169319e-01
2.81405985e-01 -3.54365587e-01 -3.24549705e-01 3.74607176e-01
-3.09979647e-01 -1.02087080e+00 -1.12676275e+00 5.76156914e-01
-3.79134476e-01 -4.31906104e-01 -9.15434062e-01 -6.05148494e-01
2.41284221e-01 7.68516183e-01 7.22432971e-01 -4.42972660e-01
-1.48532957e-01 4.28521335e-01 -3.14819217e-01 -3.47880393e-01
-4.11666222e-02 -1.78287402e-01 3.30415040e-01 1.55144557e-01
-1.23530753e-01 -6.77036524e-01 -7.06640005e-01 4.07338917e-01
-7.39953995e-01 -1.67927682e-01 3.86603147e-01 9.73341107e-01
4.76049781e-02 1.84684485e-01 3.92842889e-01 1.11406185e-01
9.38541472e-01 -1.02872312e-01 -3.80021244e-01 -5.22925071e-02
-2.69798607e-01 -4.98565063e-02 3.41609925e-01 -5.67493498e-01
-1.27005255e+00 -1.23333856e-01 -4.10438210e-01 -5.24263084e-01
7.58653358e-02 4.79124635e-01 -4.68909144e-02 -2.99746636e-02
7.85427392e-01 1.12219796e-01 -1.22137047e-01 -5.59306204e-01
4.15903270e-01 8.11180592e-01 6.60340011e-01 -1.98430076e-01
8.29539537e-01 1.72327995e-01 -2.54876405e-01 -9.79850709e-01
-5.25541365e-01 -7.60344625e-01 -3.11147571e-01 -2.92964075e-02
7.64826834e-01 -1.07707012e+00 -1.57727391e-01 5.64560354e-01
-1.13438487e+00 -4.67068315e-01 -3.09288297e-02 8.87083232e-01
-4.62609753e-02 1.48273632e-01 -6.83941185e-01 -8.28785956e-01
-4.66111660e-01 -1.23133945e+00 9.88001108e-01 -3.15740518e-02
-3.86283517e-01 -6.89998090e-01 5.48142754e-02 2.18311265e-01
1.08634281e+00 -2.78576106e-01 4.10223663e-01 -5.15582025e-01
-8.05786178e-02 6.61939830e-02 -1.41128615e-01 6.24027073e-01
3.89695555e-01 -2.18744501e-01 -1.30586421e+00 -6.05383694e-01
3.24252009e-01 3.14859718e-01 1.06838310e+00 8.79959762e-01
6.95354521e-01 -2.02433184e-01 -4.78238054e-02 9.48504806e-01
1.06263804e+00 5.05786300e-01 4.77073401e-01 6.27734214e-02
5.65802634e-01 2.83830255e-01 1.15481086e-01 2.45266035e-01
-2.18412504e-01 7.76441813e-01 -7.46505857e-02 -3.90318245e-01
-8.50151420e-01 5.04321456e-01 6.14296257e-01 1.08599043e+00
8.46130252e-02 -4.65305656e-01 -8.69273543e-01 5.73257506e-01
-1.32387030e+00 -1.20367265e+00 -3.53910588e-02 2.21585560e+00
8.24347615e-01 7.04116672e-02 3.23676839e-02 5.97497582e-01
6.22078359e-01 3.18827420e-01 -2.17083648e-01 -4.34023827e-01
-4.36031997e-01 7.76745737e-01 6.28917933e-01 1.11731303e+00
-1.23474801e+00 4.20754194e-01 6.68530655e+00 9.86231029e-01
-1.45216858e+00 3.74765396e-01 4.78188396e-01 -8.15427184e-01
-1.01283163e-01 -6.12056553e-01 -4.73552257e-01 6.86564147e-02
1.40587723e+00 -8.82692076e-03 7.98970342e-01 7.13523626e-02
4.42057818e-01 1.15730897e-01 -8.87889266e-01 1.12904727e+00
3.27060223e-01 -1.02793396e+00 -4.66129243e-01 -2.45583951e-01
5.33233881e-01 2.10169390e-01 3.90210688e-01 5.36870472e-02
-1.31725684e-01 -1.20343208e+00 1.21615076e+00 3.05908471e-01
8.76175880e-01 -6.11842215e-01 3.79186928e-01 -2.19414473e-01
-1.51257622e+00 -2.64112860e-01 3.13714385e-01 -4.70720418e-02
1.61244363e-01 6.79017425e-01 -6.27073109e-01 5.34577847e-01
9.04670477e-01 2.46437430e-01 -1.67374998e-01 1.32116282e+00
5.24507388e-02 9.28210855e-01 -4.89903063e-01 2.98813492e-01
2.05275957e-02 1.11299038e-01 9.12186444e-01 1.78161013e+00
5.19782305e-01 1.16220616e-01 -2.78576553e-01 4.37546581e-01
1.26270428e-01 -2.51935273e-01 3.17993835e-02 2.32445613e-01
5.89842677e-01 1.18323052e+00 -5.28288007e-01 9.09315143e-03
-5.01014650e-01 8.87356997e-01 -3.22054505e-01 8.28987360e-01
-9.31216657e-01 -8.59588385e-01 9.40906048e-01 6.57955334e-02
6.47876561e-01 -6.68564498e-01 -3.48779082e-01 -8.31765950e-01
1.47275915e-02 -1.15275705e+00 -4.45265397e-02 -7.91256845e-01
-8.72432768e-01 9.35301006e-01 -7.55084530e-02 -9.62553620e-01
-1.21239744e-01 -6.72728419e-01 -5.88426471e-01 1.48352504e+00
-1.31932545e+00 -8.38179886e-01 -5.39857782e-02 5.19451678e-01
7.37512350e-01 -2.41571099e-01 7.59340346e-01 7.82620847e-01
-5.38861811e-01 9.45134819e-01 3.18677306e-01 1.46005321e-02
1.01274478e+00 -1.16014862e+00 5.48090458e-01 1.46031272e+00
3.24013263e-01 9.09127533e-01 8.96032155e-01 -1.60151735e-01
-1.10028303e+00 -9.54935431e-01 6.26725733e-01 -3.22252095e-01
6.03805482e-01 -5.98931789e-01 -7.79134333e-01 3.00174534e-01
5.44285953e-01 -9.38314125e-02 8.02653968e-01 4.13131937e-02
-5.05911648e-01 -3.90646547e-01 -9.29629147e-01 6.57256603e-01
8.62070203e-01 -5.41319609e-01 -5.48228860e-01 4.16854918e-02
9.58771050e-01 -5.29145420e-01 -4.61622387e-01 4.64167893e-01
7.83944905e-01 -1.05678523e+00 1.21491301e+00 -1.90191954e-01
2.71640830e-02 -5.33606410e-01 -5.59315085e-01 -1.65130866e+00
-6.29806936e-01 -1.02364707e+00 -4.64711823e-02 1.24552429e+00
7.41984189e-01 -6.34655178e-01 3.29375148e-01 2.87286937e-01
-5.56688309e-01 -5.10900915e-01 -9.91681635e-01 -9.04585302e-01
-4.47042212e-02 -6.48396671e-01 2.77269453e-01 6.59939945e-01
-2.56530702e-01 3.38375360e-01 -2.02419564e-01 4.51401830e-01
3.86363536e-01 -3.86703253e-01 2.21921414e-01 -5.10591805e-01
-6.52405500e-01 -7.27169573e-01 -5.22070378e-03 -1.08508587e+00
-6.82868883e-02 -5.32891750e-01 4.02498215e-01 -1.41077876e+00
-3.71810794e-01 1.47849157e-01 -6.96169853e-01 2.53716826e-01
-2.42697820e-01 2.23695949e-01 2.57708788e-01 -2.86311120e-01
-4.96069528e-02 3.39958459e-01 6.06682181e-01 -3.29388380e-01
-3.32500309e-01 8.92827567e-03 -6.23766363e-01 6.69402122e-01
6.40900850e-01 -2.48616174e-01 -4.81571287e-01 -8.64415526e-01
-2.97283888e-01 1.19010121e-01 1.86468691e-01 -1.47452843e+00
3.02366406e-01 2.28314698e-01 5.39763629e-01 -3.06008458e-01
7.43478119e-01 -6.95136368e-01 1.09314531e-01 2.57677495e-01
-4.12668973e-01 1.06424406e-01 8.37730646e-01 3.75801146e-01
-2.69646198e-01 1.54137120e-01 8.89748096e-01 3.57570559e-01
-1.97467506e-01 -2.60480821e-01 -6.38663948e-01 -1.39853150e-01
3.61034989e-01 -1.67937025e-01 -1.98585019e-01 -6.40439868e-01
-7.57584989e-01 -3.11312437e-01 -3.12432081e-01 2.77984053e-01
5.49131393e-01 -1.14882898e+00 -9.94640231e-01 3.78947258e-01
-5.19533217e-01 -5.57204187e-01 3.80848110e-01 1.08800256e+00
-3.35727215e-01 3.46679538e-01 5.34062013e-02 -4.61633563e-01
-1.65138566e+00 8.37251246e-02 7.06623018e-01 8.92774910e-02
-8.27377141e-02 1.55630624e+00 1.44379660e-01 5.00238501e-02
5.22843540e-01 -5.70127606e-01 1.39657855e-01 -2.66662296e-02
8.21478665e-01 6.79596543e-01 5.79772174e-01 -7.90367842e-01
-5.51172793e-01 5.00492692e-01 1.70626998e-01 -8.58144045e-01
1.11031604e+00 -2.84719616e-01 1.00210637e-01 3.27217251e-01
1.32283652e+00 7.06830740e-01 -1.17295563e+00 -1.87075928e-01
-3.39220017e-01 -4.58918929e-01 6.06402099e-01 -1.26681352e+00
-1.02051198e+00 9.33148146e-01 1.10859573e+00 2.89953589e-01
1.74145651e+00 -6.78828359e-01 7.38415003e-01 3.69745307e-02
-1.58021346e-01 -6.80478394e-01 -1.95661671e-02 7.49673963e-01
1.21137881e+00 -7.70210624e-01 -3.01798750e-02 -1.62384570e-01
-2.50641312e-02 1.08933985e+00 2.01216251e-01 9.39358547e-02
8.38885248e-01 8.97217333e-01 4.19787735e-01 4.34673488e-01
-4.53962535e-01 -1.39805913e-01 8.09855044e-01 4.78432953e-01
7.39400685e-01 -8.36555362e-02 -2.94838101e-02 8.04177821e-01
-4.75972027e-01 -6.50153995e-01 7.66693652e-02 5.68299830e-01
-4.75230932e-01 -9.35467184e-01 -9.72763956e-01 1.83936894e-01
-6.76777303e-01 -7.63483644e-01 -1.69252947e-01 1.85088590e-01
-8.85401294e-02 1.53787100e+00 1.55797020e-01 -5.81586123e-01
4.13556635e-01 1.23772323e-01 5.48928678e-01 -4.36959922e-01
-9.38281059e-01 9.25699532e-01 2.89466113e-01 -3.64167362e-01
-4.04038876e-01 -8.10841322e-01 -9.76575553e-01 -7.99895674e-02
-5.95841825e-01 -1.04469724e-01 7.89145529e-01 7.60029972e-01
3.27879101e-01 1.24194133e+00 4.02135134e-01 -1.30856085e+00
-3.73761207e-01 -1.16064525e+00 -3.35270286e-01 4.89285551e-02
9.84688342e-01 -2.99069881e-01 -6.00746810e-01 1.04008235e-01] | [15.040948867797852, 5.876130104064941] |
e1e2a6db-ba01-479b-b3a3-664676671c2b | a-zero-shot-adaptive-quadcopter-controller | 2209.09232 | null | https://arxiv.org/abs/2209.09232v2 | https://arxiv.org/pdf/2209.09232v2.pdf | Learning a Single Near-hover Position Controller for Vastly Different Quadcopters | This paper proposes an adaptive near-hover position controller for quadcopters, which can be deployed to quadcopters of very different mass, size and motor constants, and also shows rapid adaptation to unknown disturbances during runtime. The core algorithmic idea is to learn a single policy that can adapt online at test time not only to the disturbances applied to the drone, but also to the robot dynamics and hardware in the same framework. We achieve this by training a neural network to estimate a latent representation of the robot and environment parameters, which is used to condition the behaviour of the controller, also represented as a neural network. We train both networks exclusively in simulation with the goal of flying the quadcopters to goal positions and avoiding crashes to the ground. We directly deploy the same controller trained in the simulation without any modifications on two quadcopters in the real world with differences in mass, size, motors, and propellers with mass differing by 4.5 times. In addition, we show rapid adaptation to sudden and large disturbances up to one-third of the mass of the quadcopters. We perform an extensive evaluation in both simulation and the physical world, where we outperform a state-of-the-art learning-based adaptive controller and a traditional PID controller specifically tuned to each platform individually. Video results can be found at https://youtu.be/U-c-LbTfvoA. | ['Mark W. Mueller', 'Jitendra Malik', 'Ashish Kumar', 'Xiangyu Wu', 'Antonio Loquercio', 'Dingqi Zhang'] | 2022-09-19 | null | null | null | null | ['drone-controller'] | ['robots'] | [-4.45844382e-01 1.22462243e-01 1.18509941e-01 3.43507916e-01
6.48604333e-02 -8.59081089e-01 1.27490819e-01 -1.05963156e-01
-4.46529061e-01 7.58434236e-01 -6.76537573e-01 -3.12246680e-01
-2.42902860e-01 -5.52199841e-01 -1.08032119e+00 -8.02093923e-01
-6.40955091e-01 6.84556305e-01 3.60096961e-01 -6.21078730e-01
-1.16410829e-01 6.77819312e-01 -1.55309618e+00 -5.97200632e-01
5.58063567e-01 6.39554918e-01 2.35941187e-01 1.29692471e+00
1.05963922e+00 3.80497336e-01 -6.83660567e-01 4.11047012e-01
6.29225075e-01 1.68995568e-04 -1.42256692e-01 1.19881816e-01
1.04760416e-01 -3.43792379e-01 -2.88835466e-01 7.73104668e-01
6.50302649e-01 5.96455872e-01 3.18969190e-01 -1.43967021e+00
1.85271874e-01 3.48172605e-01 -1.54274061e-01 -1.44040495e-01
1.70595329e-02 7.75343597e-01 1.96710035e-01 -1.57041252e-01
2.20945567e-01 1.04058695e+00 8.54669750e-01 4.26972777e-01
-8.79476905e-01 -7.28062093e-01 1.17712483e-01 -2.45814145e-01
-1.38627934e+00 -2.90682524e-01 1.32495686e-01 -4.63749111e-01
1.05064023e+00 -1.80067524e-01 7.26429999e-01 9.13923740e-01
6.82292759e-01 -2.97469627e-02 3.81397784e-01 -1.15136392e-01
4.03185546e-01 -5.91421612e-02 -7.51030564e-01 8.44198823e-01
4.52186197e-01 5.31421542e-01 1.53239891e-01 -1.30744770e-01
8.76843989e-01 -9.80443582e-02 -4.83421892e-01 -5.56795418e-01
-1.37716937e+00 5.50400555e-01 3.38554591e-01 -2.46341005e-01
-4.05975223e-01 6.31812155e-01 4.61369902e-01 5.44443667e-01
-3.57321203e-01 8.55975091e-01 -9.78782833e-01 -4.85484004e-01
-1.51525453e-01 5.59647083e-01 1.14834023e+00 1.16332281e+00
4.19712633e-01 6.18733287e-01 3.63236994e-01 2.92070389e-01
6.50990829e-02 8.76916230e-01 5.34321964e-01 -1.33919930e+00
4.75938439e-01 2.67023832e-01 7.74433792e-01 -8.77226591e-01
-7.41996884e-01 -4.30111349e-01 -4.34677541e-01 7.10933566e-01
9.27341580e-02 -1.28678250e+00 -6.35700226e-01 1.63528025e+00
6.01853192e-01 4.59753349e-02 1.18750066e-01 1.07495034e+00
-3.72580737e-01 6.11821353e-01 -5.92722833e-01 -5.79920895e-02
9.15866435e-01 -7.86529183e-01 -4.04298276e-01 -4.45948124e-01
6.34451509e-01 -4.67311323e-01 9.62912440e-01 5.54561317e-01
-9.04582798e-01 -6.88909769e-01 -1.51405525e+00 7.80629039e-01
-3.31588328e-01 4.17145193e-01 -1.56695560e-01 2.19272584e-01
-9.13047850e-01 9.67327356e-01 -1.31112921e+00 -3.39988530e-01
-6.25577152e-01 6.76477551e-01 -7.11513236e-02 7.19789565e-01
-1.21119428e+00 1.24459636e+00 6.77082956e-01 -3.07778381e-02
-1.34125936e+00 -4.97812957e-01 -7.47778773e-01 -1.30027294e-01
7.84993768e-01 -8.34708810e-01 1.65746963e+00 -6.03983283e-01
-2.13299656e+00 -8.11996982e-02 5.90609848e-01 -7.01786160e-01
3.13784182e-01 -4.83054340e-01 -1.77135527e-01 1.25527948e-01
-1.20027095e-01 3.68763089e-01 1.13386095e+00 -1.11054695e+00
-8.71993005e-01 2.69492287e-02 3.06663156e-01 4.62590009e-01
-6.79964870e-02 -6.11481965e-01 -1.76567629e-01 -3.43862295e-01
-3.48708838e-01 -1.62989652e+00 -2.57731944e-01 -9.41849053e-02
-7.51401410e-02 2.34296396e-01 8.93323004e-01 -1.08435445e-01
8.03932309e-01 -1.97228193e+00 4.65529591e-01 1.25971502e-02
-4.43671405e-01 2.67346203e-01 9.27832425e-02 7.85068393e-01
-4.83054928e-02 -1.76334202e-01 1.77590311e-01 -8.15255642e-02
1.06985129e-01 2.50935405e-01 -3.73148590e-01 7.31911957e-01
4.41903099e-02 3.91740829e-01 -9.97959018e-01 1.55147493e-01
1.47058308e-01 2.08953276e-01 -5.80994487e-01 4.40859914e-01
-3.73728037e-01 3.78284425e-01 -5.22832572e-01 2.60702699e-01
1.83357432e-01 3.22727591e-01 3.50745082e-01 1.22324578e-01
-5.61103284e-01 -1.48514330e-01 -1.36609709e+00 1.14064634e+00
-7.25166976e-01 4.90970403e-01 6.91750884e-01 -7.73570776e-01
9.01466727e-01 2.52550930e-01 3.37220639e-01 -2.57177949e-01
6.51224613e-01 1.53119192e-01 8.74179602e-02 -4.85909551e-01
4.18268770e-01 -1.49319738e-01 -2.45486036e-01 8.78807455e-02
6.04382828e-02 -8.19840252e-01 2.16227949e-01 -1.89606011e-01
1.14805508e+00 2.05888823e-01 3.88782173e-02 -1.47795320e-01
3.98285270e-01 2.90685713e-01 5.08039653e-01 4.91637319e-01
5.81167080e-02 8.16944614e-02 3.55589092e-01 -1.98837951e-01
-1.10383987e+00 -7.01217353e-01 3.35420668e-01 9.47765410e-01
3.68164837e-01 -3.02276611e-01 -6.27953053e-01 -1.14749193e-01
3.94832164e-01 7.89238453e-01 -4.47479039e-01 -6.18316948e-01
-6.43888116e-01 -2.42076829e-01 3.93853456e-01 5.31550527e-01
3.53792727e-01 -5.69246054e-01 -1.47044134e+00 3.31409365e-01
4.73075420e-01 -1.12915766e+00 -4.12014902e-01 4.52888131e-01
-5.40466666e-01 -1.31245279e+00 -6.31111562e-02 -5.71801901e-01
6.11893833e-01 5.76106198e-02 5.47630370e-01 2.19466984e-01
-1.54453605e-01 7.50734746e-01 -2.74576604e-01 -6.24908268e-01
-4.89134848e-01 8.66760090e-02 8.71454954e-01 -3.48797023e-01
-8.70530367e-01 -4.85553712e-01 -4.31269556e-01 7.33827412e-01
-7.27327585e-01 -1.99872106e-01 2.98814476e-01 8.00248802e-01
4.16382909e-01 6.47662640e-01 -4.00927290e-02 -2.69822568e-01
5.66541374e-01 -4.46794927e-01 -1.41725767e+00 -4.01461124e-01
-3.28591168e-01 -7.86311552e-02 1.32712531e+00 -9.00663257e-01
-4.71019775e-01 4.44382548e-01 1.63536474e-01 -8.57961774e-01
-1.30475974e-02 3.99869740e-01 1.82788670e-01 -1.83072507e-01
6.25610590e-01 -1.57225236e-01 3.08907598e-01 -1.23532765e-01
2.78403044e-01 3.40434909e-01 5.67702532e-01 -4.39508617e-01
1.44231892e+00 1.61831573e-01 2.86279619e-01 -6.15613639e-01
-2.74401993e-01 -3.87604684e-02 -5.09571910e-01 -2.44449794e-01
3.89656842e-01 -9.93245721e-01 -1.18969643e+00 4.68238652e-01
-7.24804938e-01 -1.22180426e+00 -1.98645994e-01 5.90859354e-01
-8.77549767e-01 -1.31947815e-01 -3.07811767e-01 -7.12664723e-01
-1.67838871e-01 -9.22272444e-01 9.53204632e-01 4.89016980e-01
1.17661014e-01 -9.20843065e-01 4.90968406e-01 -5.16820908e-01
5.24322689e-01 5.42120397e-01 4.54562396e-01 -1.40813544e-01
-1.45902842e-01 -4.97621924e-01 6.48196101e-01 2.75110275e-01
5.22021875e-02 1.67306796e-01 -2.75360227e-01 -1.25617731e+00
7.11728483e-02 -2.72037566e-01 8.21044818e-02 2.27565706e-01
4.78082627e-01 -4.78834629e-01 -6.06861234e-01 6.86030686e-01
1.46662390e+00 3.57997894e-01 -1.34941489e-01 6.06009007e-01
4.26788002e-01 2.36987203e-01 1.17712259e+00 7.10050881e-01
1.50612012e-01 8.82901847e-01 1.11073697e+00 2.36168206e-01
5.34342885e-01 -2.68517911e-01 8.15189660e-01 4.37660873e-01
2.90655904e-02 -2.92893946e-01 -9.00918245e-01 5.22944987e-01
-1.86742723e+00 -4.88311350e-01 2.28231400e-01 2.48726177e+00
4.23588932e-01 3.12518120e-01 2.37433508e-01 -2.02264905e-01
7.25789666e-01 -4.18479294e-01 -7.86122084e-01 -5.42613506e-01
5.48255622e-01 -2.98545122e-01 1.19397461e+00 6.06332183e-01
-1.02778733e+00 6.51078880e-01 5.99570179e+00 2.20114276e-01
-1.68129253e+00 -3.06401819e-01 -2.13843733e-01 -4.60502058e-01
7.44017482e-01 1.39075354e-01 -8.81588280e-01 5.26249588e-01
1.33787334e+00 -4.85982656e-01 7.96382248e-01 1.21893859e+00
4.76808548e-01 -8.37314054e-02 -9.08780932e-01 5.26966453e-01
-1.14022128e-01 -8.17425251e-01 -5.70148766e-01 -4.75632064e-02
7.34937549e-01 3.84615660e-01 -8.45351722e-03 6.79259956e-01
3.37516129e-01 -4.00529593e-01 8.88779998e-01 3.46740127e-01
5.54949999e-01 -8.99858832e-01 8.90065134e-01 9.10463572e-01
-1.11786795e+00 -6.78698063e-01 -4.10366863e-01 -3.78689110e-01
2.13934213e-01 -2.18258977e-01 -1.23318875e+00 2.97156632e-01
6.49254322e-01 3.56119335e-01 -3.35784495e-01 1.07293355e+00
-3.11139882e-01 4.78651792e-01 -6.85099006e-01 -3.73290688e-01
3.46292704e-01 8.62814020e-03 9.06518817e-01 6.37306333e-01
5.72603405e-01 -1.66165233e-01 6.40676260e-01 4.82023537e-01
4.49294358e-01 -3.53358418e-01 -6.78608060e-01 1.14523567e-01
4.10093576e-01 1.32659614e+00 -2.85057187e-01 -1.66718543e-01
2.36286864e-01 5.34791052e-01 9.73697081e-02 3.04348737e-01
-1.36512709e+00 -9.12687182e-01 8.50025475e-01 3.13210428e-01
4.69226271e-01 -7.83041418e-01 6.93513453e-01 -7.16932952e-01
-1.48798451e-01 -7.75377095e-01 -2.10351899e-01 -8.76826704e-01
-7.06319451e-01 5.05982101e-01 3.15880477e-01 -1.76444232e+00
-7.46869564e-01 -6.95934474e-01 -6.98760331e-01 5.83420575e-01
-1.11330998e+00 -4.84968394e-01 -3.43082815e-01 3.41919690e-01
4.10505950e-01 -1.83785677e-01 6.13978326e-01 -5.35874367e-02
-7.18743026e-01 1.05949245e-01 3.38633686e-01 -2.59464771e-01
9.05363381e-01 -1.18620849e+00 4.48509306e-01 6.75019920e-01
-7.41645217e-01 5.47882795e-01 1.24002767e+00 -6.96651578e-01
-2.05399275e+00 -1.40369630e+00 -1.16299406e-01 -2.38088384e-01
1.14501607e+00 -3.26518774e-01 -6.48981392e-01 8.07841122e-01
5.67209199e-02 -4.96771710e-04 -2.71073133e-01 -4.58395928e-01
6.46821320e-01 -2.40815118e-01 -8.24799716e-01 8.51351202e-01
5.89194238e-01 1.12963922e-01 -2.43207023e-01 4.73160684e-01
9.39538598e-01 -1.19459665e+00 -8.32652807e-01 4.26566035e-01
5.18805146e-01 -5.20428538e-01 6.31925166e-01 -4.20814902e-01
-3.47959578e-01 -6.91139281e-01 1.73366874e-01 -1.96556771e+00
-2.40066424e-01 -1.13349307e+00 -1.98551655e-01 6.47463322e-01
2.30164796e-01 -9.24794078e-01 4.48364139e-01 5.32001723e-03
-4.24948454e-01 -7.98062980e-01 -1.03128552e+00 -1.06749058e+00
4.47042584e-02 -2.95723341e-02 3.74885648e-01 4.37873304e-01
7.01582730e-02 3.73531431e-01 -4.54140157e-01 1.02871370e+00
1.62163869e-01 -1.04249306e-01 1.27390766e+00 -7.97138870e-01
-6.82804108e-01 -5.59453256e-02 -2.38346532e-01 -7.98108220e-01
1.72999531e-01 -1.74155906e-01 5.61758220e-01 -1.04296029e+00
-7.70271420e-01 -2.80939132e-01 1.16042651e-01 5.87711513e-01
3.13555747e-02 -1.83726504e-01 1.82400256e-01 -8.84667784e-02
-5.38694143e-01 6.35912776e-01 9.91011798e-01 -6.36156350e-02
-6.08260989e-01 4.69887882e-01 -1.07015688e-02 7.73809016e-01
1.22133994e+00 -4.86541867e-01 -7.45779634e-01 -5.96514761e-01
3.27700317e-01 3.61494988e-01 2.45270938e-01 -1.93938470e+00
3.51202875e-01 -3.02076280e-01 7.19355941e-02 -1.30570337e-01
4.03266549e-01 -9.98327494e-01 2.74594873e-01 9.61689711e-01
7.94964284e-02 6.14189327e-01 9.03796554e-01 5.65132856e-01
1.52732596e-01 -9.14652720e-02 8.20119381e-01 -2.14517061e-02
-2.87016124e-01 2.00301617e-01 -7.93783665e-01 -1.69635490e-01
1.24887586e+00 7.26249516e-02 -2.71158248e-01 -4.97556031e-01
-5.35411894e-01 7.78328419e-01 8.55096161e-01 4.84111071e-01
3.26395452e-01 -7.94432402e-01 -3.10592383e-01 2.35113487e-01
-1.92769662e-01 -4.35147658e-02 9.67612341e-02 8.30317736e-01
-7.62245715e-01 3.33121210e-01 -5.40923595e-01 -4.90573883e-01
-1.03234076e+00 7.11401403e-01 7.85228372e-01 2.28142530e-01
-4.42477852e-01 2.47239754e-01 -2.79298991e-01 -5.39513111e-01
1.11304343e-01 -6.27020538e-01 1.47717655e-01 -4.39027727e-01
1.19904704e-01 2.01872900e-01 6.55925944e-02 -3.12478691e-01
-2.21720859e-01 8.83837759e-01 4.77048934e-01 -3.94222997e-02
1.09544384e+00 -1.43017154e-02 3.49166453e-01 4.42236006e-01
5.64814389e-01 1.05684020e-01 -1.72262621e+00 5.60121894e-01
-5.71490347e-01 -9.35358778e-02 7.41691738e-02 -6.12470686e-01
-7.79267430e-01 3.79014343e-01 6.84185386e-01 1.84706524e-01
9.28308368e-01 -5.95715702e-01 4.64830041e-01 9.03729260e-01
7.85548747e-01 -1.18063319e+00 1.30482078e-01 9.40269113e-01
9.69734967e-01 -7.10003376e-01 1.14717491e-01 1.59407258e-01
-7.34860837e-01 1.11872780e+00 9.22313929e-01 -8.09297442e-01
3.13747138e-01 5.58255851e-01 1.61664873e-01 2.18213573e-01
-1.22102845e+00 6.54355139e-02 -1.80151224e-01 6.06073916e-01
-2.78240502e-01 1.08441180e-02 2.33115748e-01 1.69483542e-01
-1.56146407e-01 -2.66643137e-01 7.85199940e-01 1.17471278e+00
-7.56446004e-01 -8.11786592e-01 -6.21039391e-01 -1.25260144e-01
9.19446796e-02 5.14880657e-01 -6.30782023e-02 1.29812098e+00
4.15758401e-01 5.78733325e-01 9.61309597e-02 -5.91545463e-01
9.36553001e-01 -3.82140428e-01 2.28471607e-01 -6.88202441e-01
-6.73229337e-01 -3.65433171e-02 2.16420516e-01 -7.84529865e-01
3.09857309e-01 -4.12318677e-01 -1.53619158e+00 -6.65607676e-02
-3.68871927e-01 3.55007380e-01 6.19779825e-01 4.31812853e-01
5.13175964e-01 7.19924629e-01 7.95703948e-01 -1.50767231e+00
-1.05006707e+00 -9.54893112e-01 -4.83828664e-01 -1.97959960e-01
6.93951845e-01 -1.21328425e+00 -7.07415223e-01 -1.26845360e-01] | [4.855281829833984, 1.8723255395889282] |
9563d841-c202-4837-83d1-976d259fc1ba | semi-supervised-offline-reinforcement-1 | 2210.06518 | null | https://arxiv.org/abs/2210.06518v3 | https://arxiv.org/pdf/2210.06518v3.pdf | Semi-Supervised Offline Reinforcement Learning with Action-Free Trajectories | Natural agents can effectively learn from multiple data sources that differ in size, quality, and types of measurements. We study this heterogeneity in the context of offline reinforcement learning (RL) by introducing a new, practically motivated semi-supervised setting. Here, an agent has access to two sets of trajectories: labelled trajectories containing state, action and reward triplets at every timestep, along with unlabelled trajectories that contain only state and reward information. For this setting, we develop and study a simple meta-algorithmic pipeline that learns an inverse dynamics model on the labelled data to obtain proxy-labels for the unlabelled data, followed by the use of any offline RL algorithm on the true and proxy-labelled trajectories. Empirically, we find this simple pipeline to be highly successful -- on several D4RL benchmarks~\cite{fu2020d4rl}, certain offline RL algorithms can match the performance of variants trained on a fully labelled dataset even when we label only 10\% of trajectories which are highly suboptimal. To strengthen our understanding, we perform a large-scale controlled empirical study investigating the interplay of data-centric properties of the labelled and unlabelled datasets, with algorithmic design choices (e.g., choice of inverse dynamics, offline RL algorithm) to identify general trends and best practices for training RL agents on semi-supervised offline datasets. | ['Aditya Grover', 'Brandon Amos', 'Mikael Henaff', 'Qinqing Zheng'] | 2022-10-12 | null | null | null | null | ['d4rl'] | ['robots'] | [-1.41074210e-01 1.89280268e-02 -5.69570482e-01 -1.44706160e-01
-8.35509658e-01 -1.04651165e+00 1.03755903e+00 2.57071167e-01
-8.00576031e-01 8.14902306e-01 2.25242615e-01 -3.79164398e-01
-2.39674821e-01 -6.46088660e-01 -7.57507622e-01 -7.24657536e-01
-3.85348827e-01 8.65615368e-01 3.73230204e-02 -2.36065745e-01
-2.88099498e-02 3.99012387e-01 -1.52419150e+00 -2.00035572e-01
4.36615258e-01 6.65858865e-01 -8.40635896e-02 7.57088542e-01
4.05709088e-01 1.14763272e+00 -4.73852605e-01 -6.89767823e-02
6.34058952e-01 -6.38495028e-01 -5.83157182e-01 9.23338383e-02
-6.01055324e-02 -5.51069498e-01 -5.70147634e-01 7.51823723e-01
4.77585673e-01 2.07261622e-01 5.72188556e-01 -1.40160012e+00
-2.31912449e-01 6.83143377e-01 -7.76886418e-02 3.11087787e-01
2.22917363e-01 1.18330336e+00 9.52021182e-01 6.22140802e-02
8.22509229e-01 9.58871365e-01 5.67319930e-01 6.79765940e-01
-1.44628775e+00 -4.19265926e-01 2.75425836e-02 -1.57596707e-01
-6.48657203e-01 -5.05000234e-01 3.58012259e-01 -5.21412373e-01
7.80719876e-01 -2.39621192e-01 7.56439984e-01 1.46810043e+00
-1.47632137e-01 7.84124017e-01 1.48820782e+00 3.44455242e-04
6.82225883e-01 9.18931663e-02 9.47872326e-02 7.40916431e-01
-6.57294765e-02 1.11135793e+00 -4.18681949e-01 -2.69102156e-01
6.09226584e-01 -1.30380511e-01 3.51623744e-02 -4.97915030e-01
-1.55867445e+00 7.77248621e-01 2.63091713e-01 -3.54910977e-02
-3.64213496e-01 4.72179800e-01 5.45944095e-01 8.33702981e-01
1.56945705e-01 6.37654662e-01 -6.84676468e-01 -5.93980551e-01
-5.75731933e-01 6.42288208e-01 9.40364480e-01 8.52277577e-01
8.54229748e-01 1.50493711e-01 -2.10884303e-01 1.21994779e-01
1.42755777e-01 7.27221251e-01 6.00943506e-01 -1.33274972e+00
4.65698183e-01 4.55624819e-01 5.50800979e-01 -3.32826465e-01
-7.35563755e-01 -3.36932778e-01 -3.13464105e-01 2.21469909e-01
7.84797013e-01 -6.76585376e-01 -8.01765025e-01 1.97150874e+00
5.36162078e-01 3.89913052e-01 4.75182027e-01 9.04367983e-01
2.72790521e-01 3.82146209e-01 4.00050096e-02 -2.83146143e-01
6.35782123e-01 -9.34134781e-01 -2.58421183e-01 -1.31298617e-01
1.12852430e+00 9.83177200e-02 1.12982500e+00 2.19324738e-01
-8.94256890e-01 -4.48196232e-01 -8.39313865e-01 3.73674929e-01
-2.30588183e-01 2.16203943e-01 4.44885671e-01 4.06223744e-01
-9.56779599e-01 1.05642807e+00 -1.09698534e+00 -1.88943878e-01
9.29031074e-02 4.57320035e-01 -1.66296050e-01 8.84017199e-02
-9.21388149e-01 8.86865973e-01 4.14384961e-01 -1.96018234e-01
-1.69129848e+00 -6.03745639e-01 -6.49926782e-01 -4.46966320e-01
5.40366709e-01 -2.51067251e-01 1.65716147e+00 -1.10298693e+00
-1.72572422e+00 4.76635545e-01 2.52574861e-01 -7.91940570e-01
7.40383685e-01 3.62459011e-02 -2.27278158e-01 6.15475513e-02
1.98626980e-01 6.14557326e-01 9.09074306e-01 -1.12974191e+00
-8.48382413e-01 -2.42624879e-01 2.50494063e-01 2.47077942e-01
1.49734631e-01 -4.10206348e-01 1.04261460e-02 -2.49864012e-01
-6.10232174e-01 -1.31083608e+00 -3.57247680e-01 -5.33334255e-01
-7.76225701e-02 -2.00680882e-01 7.66637087e-01 -3.42211008e-01
8.22019100e-01 -2.11495709e+00 4.00407016e-01 9.79286209e-02
7.48215690e-02 2.54092097e-01 -3.77105027e-01 6.85002744e-01
2.35190377e-01 -1.57043844e-01 -1.90150872e-01 -3.85998279e-01
2.17264339e-01 4.49938178e-01 -5.10963917e-01 7.51040459e-01
9.53807011e-02 1.09158325e+00 -1.42223644e+00 -6.23637103e-02
2.16829255e-01 -1.37122080e-01 -4.11980063e-01 4.71015871e-01
-7.85864294e-01 9.77535248e-01 -5.26332855e-01 3.03194314e-01
-1.85977370e-01 -2.64480591e-01 2.53101200e-01 3.70210171e-01
-1.56592205e-01 3.02064449e-01 -9.64299679e-01 1.57823408e+00
-2.68672138e-01 6.14482343e-01 -6.87648430e-02 -7.94017196e-01
5.47563493e-01 3.98536801e-01 7.01722026e-01 -8.28456163e-01
2.09115386e-01 1.90147787e-01 2.69032288e-02 -5.78265011e-01
2.90406287e-01 -6.83722943e-02 -2.15691596e-01 9.18753028e-01
1.53017133e-01 -1.26494527e-01 3.26259583e-01 1.22931957e-01
1.55911505e+00 5.26129723e-01 5.84125668e-02 2.14978941e-02
-3.76040190e-02 6.70636237e-01 3.86745185e-01 1.00864518e+00
-4.59967405e-01 -2.09603459e-02 6.79979861e-01 -4.65641916e-01
-1.26374781e+00 -8.63306105e-01 2.72936761e-01 1.21361256e+00
7.35602230e-02 -3.41111809e-01 -6.13234997e-01 -1.06341898e+00
1.75998569e-01 7.18864322e-01 -7.26508975e-01 -3.76797438e-01
-5.23693621e-01 -7.46335506e-01 8.00348401e-01 3.23740244e-01
3.41057003e-01 -1.35691142e+00 -1.12047732e+00 3.46347362e-01
2.71363944e-01 -1.02400935e+00 -2.11541265e-01 3.93459439e-01
-8.11926246e-01 -1.16024542e+00 -2.04891175e-01 -2.17322186e-01
4.46345985e-01 -1.11625493e-01 1.11158168e+00 -1.37543336e-01
1.39981896e-01 7.95283258e-01 -5.64649463e-01 -1.15179375e-01
-7.91370392e-01 1.66800022e-01 2.72139311e-01 -1.45961761e-01
2.60155797e-02 -4.82503444e-01 -4.35416222e-01 2.62689024e-01
-7.47261643e-01 -1.64233342e-01 5.44064462e-01 7.10713446e-01
3.15101147e-01 -4.46433984e-02 7.31201828e-01 -8.05032969e-01
6.97373331e-01 -8.26220393e-01 -9.04843450e-01 -1.92524828e-02
-5.61393440e-01 4.72881436e-01 9.20774937e-01 -7.53152847e-01
-5.94276905e-01 1.28076494e-01 2.38429666e-01 -5.93546391e-01
-3.04055721e-01 3.19448411e-01 3.32770109e-01 2.11222321e-01
8.18600476e-01 3.53942931e-01 2.54046619e-01 -2.87438840e-01
6.53646946e-01 5.32336891e-01 4.56149429e-01 -7.80968904e-01
7.42550731e-01 5.28501213e-01 -6.04940671e-03 -5.51344395e-01
-8.15567017e-01 -2.90459991e-01 -5.38806796e-01 -2.19248608e-01
6.21211827e-01 -8.96714389e-01 -9.59211707e-01 4.70753938e-01
-4.50407475e-01 -1.47794056e+00 -8.52823913e-01 5.64668655e-01
-1.10821664e+00 -1.78670794e-01 -5.60156882e-01 -7.58777559e-01
1.78818777e-01 -1.36011887e+00 1.01120126e+00 2.47199051e-02
2.76606251e-02 -1.17561162e+00 7.54605711e-01 2.29490139e-02
2.14221478e-01 2.68431276e-01 6.64883435e-01 -9.40079033e-01
-5.37030220e-01 2.85800528e-02 1.00274689e-01 1.71260118e-01
-6.20428659e-02 -1.07844256e-01 -7.93478489e-01 -6.22586668e-01
-2.04080045e-01 -9.50246036e-01 6.40467107e-01 7.26977214e-02
5.30468345e-01 -3.83701384e-01 -1.51959702e-01 3.60652030e-01
1.26719642e+00 1.19579174e-01 1.01284776e-02 2.94793576e-01
4.50001031e-01 4.99580383e-01 7.83625007e-01 5.10380566e-01
6.22299671e-01 5.67585826e-01 4.06305999e-01 3.83754939e-01
1.93668753e-01 -5.88980377e-01 8.86532724e-01 5.93041956e-01
2.10957587e-01 -2.69654393e-01 -8.62926722e-01 5.34968317e-01
-2.02097011e+00 -9.49720025e-01 2.80449808e-01 2.32702637e+00
9.16275918e-01 1.33233070e-01 8.03078949e-01 -1.10946320e-01
1.56948343e-01 2.45197520e-01 -1.09940600e+00 -1.99769698e-02
1.94534972e-01 -3.04385871e-02 9.31877494e-01 3.28429818e-01
-1.01223266e+00 9.88392651e-01 6.51865053e+00 6.42318606e-01
-1.17730284e+00 7.12274462e-02 3.77260149e-01 -3.08445543e-01
-1.05543725e-01 1.68093145e-01 -7.58337319e-01 6.02754056e-01
1.69305634e+00 9.31456685e-02 9.92833316e-01 5.52195728e-01
4.51810986e-01 -2.08655655e-01 -1.42113209e+00 5.82129896e-01
-5.03487170e-01 -1.31995642e+00 -4.34618264e-01 2.28983611e-01
7.76570380e-01 6.18830025e-01 8.40570256e-02 7.45369196e-01
1.23860776e+00 -9.83258843e-01 7.69945264e-01 4.66199428e-01
7.07337439e-01 -5.96506357e-01 2.82297343e-01 9.28376019e-01
-9.21112597e-01 -4.68449920e-01 2.03046143e-01 -2.36686319e-01
-1.51489452e-01 -1.44455403e-01 -7.66676724e-01 3.37926149e-01
4.08887804e-01 1.14154279e+00 -4.60380226e-01 5.93219161e-01
-1.21030986e-01 9.48864758e-01 -4.97091621e-01 -3.43449414e-02
6.31304026e-01 -3.93799782e-01 4.07717317e-01 7.98496366e-01
-1.22863054e-01 8.44084024e-02 4.16148245e-01 6.29822791e-01
3.64749506e-02 -4.11107689e-01 -8.29524100e-01 -4.42329854e-01
4.11914378e-01 1.03488064e+00 -6.32034898e-01 -2.96591759e-01
-2.42303014e-01 5.45893133e-01 6.11609459e-01 4.41195726e-01
-7.99405158e-01 3.52037162e-01 5.75698972e-01 -9.91393924e-02
1.53033987e-01 -5.73928773e-01 3.47347260e-01 -1.20676959e+00
-4.26801234e-01 -1.23910677e+00 3.24813634e-01 -4.11321223e-01
-1.13121474e+00 1.97228596e-01 1.78770095e-01 -1.18820786e+00
-8.35151315e-01 -2.61807173e-01 -4.26268637e-01 2.73357779e-01
-1.33314240e+00 -7.34003186e-01 1.21203251e-01 5.60154438e-01
4.33804959e-01 -2.59265989e-01 3.98359925e-01 -6.82016239e-02
-7.55367219e-01 1.57230183e-01 5.87603152e-01 1.93943307e-01
4.01267499e-01 -1.38646722e+00 4.32994336e-01 3.75652611e-01
3.71668190e-01 7.00836107e-02 6.76291347e-01 -7.39066660e-01
-1.83539915e+00 -1.27110231e+00 -7.01254010e-02 -7.95453787e-01
9.25430536e-01 -8.93119201e-02 -4.49652910e-01 1.07954454e+00
-7.11653233e-02 8.51967931e-02 2.42345080e-01 -3.11019868e-01
-6.09894358e-02 1.26130477e-01 -1.04938459e+00 5.80750048e-01
1.12262392e+00 -5.42794168e-01 -3.64044458e-01 3.94209474e-01
7.58085191e-01 -4.06066716e-01 -8.68082523e-01 2.04862297e-01
2.46275082e-01 -9.02061164e-01 6.99212074e-01 -8.77034068e-01
1.24347605e-01 -2.44184464e-01 -5.05740158e-02 -1.62493074e+00
2.34142900e-01 -9.25039411e-01 -3.37597638e-01 8.66347790e-01
2.43389964e-01 -6.90811276e-01 6.88728809e-01 4.85336185e-01
2.12209508e-01 -7.39253521e-01 -7.55640149e-01 -8.80539477e-01
1.16542004e-01 -4.17228997e-01 5.36047459e-01 6.02104187e-01
-6.60806522e-02 2.03671113e-01 -3.38840723e-01 6.46135509e-02
6.95419431e-01 2.84314871e-01 1.08742893e+00 -7.85852551e-01
-5.11619091e-01 -2.48367429e-01 -7.14554871e-03 -1.06096137e+00
4.32428598e-01 -7.35195398e-01 5.47387935e-02 -9.67759311e-01
-2.00359434e-01 -8.60378563e-01 8.23455304e-03 5.02240241e-01
1.29594356e-01 -2.37705320e-01 2.77068466e-01 5.30371726e-01
-9.53318179e-01 7.19842196e-01 1.11513996e+00 2.26795346e-01
-5.64438105e-01 1.55172274e-01 -4.74534333e-02 5.64635336e-01
8.03979039e-01 -6.00367069e-01 -7.74784446e-01 -2.15650201e-01
2.11470068e-01 5.62471926e-01 5.28500199e-01 -9.81745541e-01
1.59528121e-01 -4.95250195e-01 -7.39038810e-02 -1.73528194e-01
2.02892303e-01 -7.52326071e-01 6.91822693e-02 5.30837893e-01
-8.65841687e-01 2.63112962e-01 -3.73113006e-02 1.02090609e+00
2.73623466e-01 -1.55121073e-01 5.44912875e-01 -3.25190693e-01
-5.34878433e-01 4.43093181e-01 -4.04076904e-01 6.88778698e-01
1.22885239e+00 1.72538102e-01 -3.30466986e-01 -4.77544099e-01
-6.22994006e-01 5.57316184e-01 6.47202015e-01 1.87932819e-01
1.88168228e-01 -9.19052839e-01 -5.05916059e-01 1.81926981e-01
-1.84896693e-01 2.40008323e-03 -2.38895699e-01 8.57099891e-01
-1.24648936e-01 2.97007620e-01 -9.63483453e-02 -4.56728935e-01
-4.36008096e-01 7.65056074e-01 6.53527737e-01 -5.35432339e-01
-6.75725579e-01 1.24155454e-01 -3.71056467e-01 -8.60832334e-01
2.38463625e-01 -4.09740031e-01 -8.82257000e-02 1.78896144e-01
8.25970769e-02 5.66800296e-01 -2.34321952e-01 -5.81626117e-01
1.05835654e-01 1.71023652e-01 2.23307490e-01 -6.03891432e-01
1.21086776e+00 -3.74996453e-03 4.79233682e-01 7.88239062e-01
1.04714692e+00 -2.61308044e-01 -2.12424278e+00 -3.20815712e-01
1.75641611e-01 -7.78049901e-02 -1.39592156e-01 -7.60951757e-01
-9.89504337e-01 4.89604563e-01 4.56283152e-01 3.55854958e-01
6.98533714e-01 -6.06105477e-02 6.24590576e-01 5.27074814e-01
7.46047497e-01 -1.05127168e+00 2.16333196e-01 5.13183713e-01
3.57629627e-01 -1.36404502e+00 -1.52675956e-01 7.57657886e-01
-9.85048175e-01 7.78218091e-01 4.37315345e-01 -5.31426251e-01
4.71650511e-01 1.99695095e-01 2.65771467e-02 -2.64391661e-01
-1.23826802e+00 -5.85534096e-01 -2.20018744e-01 5.98889112e-01
-4.11093593e-01 9.79941338e-02 2.43944809e-01 1.76808536e-01
-2.42544398e-01 1.24469474e-01 4.90833282e-01 1.12758553e+00
-1.78602159e-01 -1.04937851e+00 5.97384945e-02 5.45867920e-01
3.15633453e-02 3.83325338e-01 -3.41907769e-01 9.55776691e-01
-7.11431205e-02 9.51089501e-01 1.00514062e-01 -4.71384764e-01
2.17713714e-01 -2.66687609e-02 3.12918276e-01 -5.29278576e-01
-9.27902102e-01 -2.88967371e-01 1.49465948e-01 -8.12502265e-01
-5.55211663e-01 -9.94441748e-01 -1.46588206e+00 -3.00304979e-01
3.64033394e-02 2.13716760e-01 4.22082126e-01 1.25101566e+00
4.74443167e-01 1.34821102e-01 8.86826813e-01 -1.08673120e+00
-1.12883186e+00 -6.73694134e-01 -3.16697448e-01 5.08342862e-01
8.12402189e-01 -7.00767756e-01 -6.14426553e-01 -1.06462494e-01] | [4.095372200012207, 1.6736195087432861] |
2d79a0ab-418f-4823-a99a-fe5347f2d2e6 | a-sequence-matching-network-for-polyphonic | 2002.05865 | null | http://arxiv.org/abs/2002.05865v1 | http://arxiv.org/pdf/2002.05865v1.pdf | A Sequence Matching Network for Polyphonic Sound Event Localization and Detection | Polyphonic sound event detection and direction-of-arrival estimation require
different input features from audio signals. While sound event detection mainly
relies on time-frequency patterns, direction-of-arrival estimation relies on
magnitude or phase differences between microphones. Previous approaches use the
same input features for sound event detection and direction-of-arrival
estimation, and train the two tasks jointly or in a two-stage transfer-learning
manner. We propose a two-step approach that decouples the learning of the sound
event detection and directional-of-arrival estimation systems. In the first
step, we detect the sound events and estimate the directions-of-arrival
separately to optimize the performance of each system. In the second step, we
train a deep neural network to match the two output sequences of the event
detector and the direction-of-arrival estimator. This modular and hierarchical
approach allows the flexibility in the system design, and increase the
performance of the whole sound event localization and detection system. The
experimental results using the DCASE 2019 sound event localization and
detection dataset show an improved performance compared to the previous
state-of-the-art solutions. | [] | 2020-02-14 | null | null | null | null | ['direction-of-arrival-estimation', 'sound-event-localization-and-detection'] | ['audio', 'audio'] | [-7.84701705e-02 -4.84866619e-01 5.13401747e-01 -2.04715073e-01
-1.15088689e+00 -5.98496735e-01 2.25945637e-01 4.84402359e-01
-4.03670251e-01 1.00333408e-01 3.91338132e-02 -1.37562737e-01
1.23959705e-02 -6.61913514e-01 -4.74349916e-01 -6.46265566e-01
-2.26239339e-01 -1.12713777e-01 8.10768366e-01 2.58479148e-01
2.13617250e-01 3.77503753e-01 -1.71169126e+00 4.88605142e-01
2.88637727e-01 1.40535617e+00 3.24856818e-01 1.33520865e+00
-1.63065791e-02 6.75893605e-01 -8.08441937e-01 2.38518372e-01
7.72382915e-02 -5.44158876e-01 -1.46440402e-01 -4.20720637e-01
2.38889143e-01 -4.26588297e-01 -3.60168725e-01 7.41297841e-01
1.09341645e+00 9.01188515e-03 5.33901989e-01 -1.12227046e+00
1.88511327e-01 5.59521556e-01 -3.78994271e-02 5.05599976e-01
7.29290068e-01 -2.32734699e-02 8.49106550e-01 -1.21579838e+00
-9.60944220e-02 8.52702916e-01 9.20981050e-01 1.26929373e-01
-7.49729991e-01 -8.72211814e-01 -1.74651906e-01 3.39672506e-01
-1.48401463e+00 -6.73968077e-01 9.34686482e-01 -6.30074382e-01
1.05695665e+00 6.21378720e-02 5.70259988e-01 6.25616968e-01
9.70942304e-02 6.66788399e-01 4.36610430e-01 -6.87716305e-01
3.19032907e-01 -3.14404108e-02 -3.44351158e-02 3.76217633e-01
-2.66961187e-01 2.90041149e-01 -7.79912889e-01 -1.49851367e-01
6.46403015e-01 1.21701546e-01 -3.42561960e-01 1.67299137e-02
-1.12047887e+00 4.77386087e-01 2.49861509e-01 4.95912373e-01
-5.62539577e-01 3.52762818e-01 5.15578330e-01 2.97321290e-01
3.96776736e-01 2.05661997e-01 -4.79350746e-01 -3.66907924e-01
-1.28913212e+00 1.05017491e-01 1.03538501e+00 4.19692934e-01
4.34675545e-01 1.98941082e-01 -1.54157817e-01 7.89105833e-01
5.46857655e-01 5.90385675e-01 4.38347310e-01 -4.25505549e-01
3.93865526e-01 9.77195874e-02 2.43184090e-01 -9.09426868e-01
-6.66652381e-01 -5.49609482e-01 -4.21874464e-01 6.96399137e-02
5.69182336e-01 -6.90336287e-01 -6.73070252e-01 1.60505664e+00
4.98697549e-01 5.90883613e-01 -2.10544840e-01 8.81324351e-01
5.78409791e-01 9.21282232e-01 -8.62313509e-02 -1.79998800e-01
1.42579293e+00 -8.36654067e-01 -9.23013926e-01 -1.52216151e-01
7.91861489e-02 -1.07873547e+00 4.61110890e-01 4.45121586e-01
-1.14076591e+00 -9.48173940e-01 -1.04910660e+00 2.34735459e-01
-5.08597612e-01 6.06203079e-01 2.30805337e-01 6.41298771e-01
-6.71957493e-01 2.18680784e-01 -9.21700299e-01 2.34189462e-02
-2.21607774e-01 2.43676499e-01 8.25826675e-02 6.24679744e-01
-1.33077860e+00 3.58494312e-01 1.44260228e-01 2.61230260e-01
-9.85070884e-01 -8.46565962e-01 -6.36004508e-01 4.21135306e-01
-3.46589573e-02 -3.57122570e-01 1.69303393e+00 -5.36340296e-01
-1.71651173e+00 3.02974373e-01 -2.25105822e-01 -5.11848927e-01
3.66767555e-01 -4.95550245e-01 -8.67171645e-01 3.41353178e-01
8.63082632e-02 1.99912027e-01 1.16968012e+00 -7.56101251e-01
-1.43828189e+00 1.32729113e-01 -2.93650299e-01 -1.60686105e-01
-2.50097126e-01 2.35807642e-01 -3.60585362e-01 -5.42721212e-01
2.98923820e-01 -5.54656565e-01 8.87546688e-02 3.66183440e-03
-1.35760024e-01 -2.03960180e-01 7.28498161e-01 -6.06605649e-01
1.57502151e+00 -2.54043412e+00 -2.82938421e-01 -7.19312392e-03
-9.24573019e-02 3.39163572e-01 -2.74346322e-02 6.21998608e-01
-2.51757532e-01 -5.87751687e-01 3.34511735e-02 -5.52017033e-01
-8.61969311e-03 -4.72476900e-01 -5.97828984e-01 3.99501085e-01
3.42894673e-01 1.01875685e-01 -1.06204140e+00 -3.28033924e-01
5.12976766e-01 6.29453778e-01 -5.28031051e-01 6.90730572e-01
1.81607231e-01 4.27572459e-01 -2.48237789e-01 3.25959861e-01
5.51649749e-01 2.86370069e-01 -1.31576151e-01 -3.28785926e-01
-4.05666649e-01 7.78499007e-01 -1.73372889e+00 1.51207268e+00
-8.32268357e-01 7.74646461e-01 2.85316646e-01 -7.13107228e-01
8.74109387e-01 1.07515121e+00 5.36957979e-01 -5.46729088e-01
1.05766907e-01 5.97467422e-01 -2.18972918e-02 -7.48398185e-01
3.45313579e-01 -1.10236302e-01 -1.05325967e-01 5.13752162e-01
2.14288861e-01 -1.63101852e-01 1.62222221e-01 -3.17129016e-01
1.06380451e+00 -2.06389710e-01 2.67303437e-01 3.38948220e-01
7.43569255e-01 -8.04676294e-01 3.99442434e-01 7.60676682e-01
-1.17410138e-01 7.92487383e-01 1.16285190e-01 -3.26945752e-01
-4.14547265e-01 -1.34318841e+00 -3.85123864e-02 1.55945706e+00
-1.32455960e-01 -3.99703205e-01 -7.90383279e-01 -5.25396466e-01
-1.56725362e-01 6.71129107e-01 -1.61242425e-01 -1.78100258e-01
-8.30001771e-01 -4.58278865e-01 7.81285822e-01 6.24103367e-01
2.32795954e-01 -1.10495079e+00 -8.53064656e-01 6.00136161e-01
-2.00453386e-01 -1.08621871e+00 -5.65800011e-01 7.31683910e-01
-2.98066109e-01 -6.77706242e-01 -5.42856991e-01 -1.02468920e+00
1.90718219e-01 1.65891293e-02 9.14955616e-01 -3.60577375e-01
-2.06828594e-01 5.71553707e-01 -4.09458697e-01 -7.26469457e-01
-1.87430367e-01 -5.82888424e-02 1.08008943e-01 4.84886676e-01
2.49908328e-01 -9.16812062e-01 -8.62392306e-01 2.73182511e-01
-6.95049286e-01 -6.35421932e-01 4.60416555e-01 4.70702112e-01
3.29244554e-01 1.43481553e-01 9.59410429e-01 -1.59469515e-01
6.25316918e-01 -4.34268266e-01 -5.71728945e-01 -7.86265433e-02
-7.94464275e-02 -4.50534016e-01 8.55530560e-01 -7.08159447e-01
-9.49329078e-01 5.59346437e-01 -6.35700226e-01 -2.58511096e-01
-3.57675403e-01 1.48959115e-01 -1.67502061e-01 2.98221111e-01
5.03451407e-01 3.27817559e-01 -7.12792516e-01 -6.53958857e-01
1.83577776e-01 1.12914217e+00 7.06702769e-01 -9.16108340e-02
5.75501800e-01 2.44973332e-01 -2.92248309e-01 -8.97636116e-01
-5.73445678e-01 -8.66076410e-01 -5.24958611e-01 -2.70927608e-01
7.86645293e-01 -1.07805681e+00 -6.99492812e-01 7.89607227e-01
-1.57054496e+00 6.42468631e-02 -2.66833305e-01 9.76880312e-01
-3.44615072e-01 -2.30516903e-02 -5.65901577e-01 -1.18959665e+00
-3.84165972e-01 -9.83570874e-01 1.36190462e+00 2.69600391e-01
-1.72688141e-01 -6.30565286e-01 3.53886992e-01 -3.28418791e-01
5.19360662e-01 -1.92905143e-01 4.24062878e-01 -7.64595509e-01
-3.19133580e-01 -8.02908778e-01 3.05564459e-02 3.68861198e-01
2.91746140e-01 -8.10578018e-02 -1.44179022e+00 2.74777375e-02
2.80451983e-01 7.56237507e-02 6.97764754e-01 7.07059562e-01
8.02799821e-01 2.62455270e-02 -3.01705748e-01 5.24549663e-01
1.00200307e+00 5.25663197e-01 3.13095063e-01 -1.06662154e-01
4.68329757e-01 2.70255297e-01 5.31495631e-01 7.37789094e-01
2.17525572e-01 7.53080666e-01 2.24001721e-01 -1.49198055e-01
-4.26484406e-01 -3.72229636e-01 5.34845650e-01 1.05346310e+00
3.60702962e-01 -3.89180005e-01 -6.19832993e-01 7.12108076e-01
-1.60527289e+00 -9.81366158e-01 -1.41808033e-01 2.30202842e+00
8.03655326e-01 5.49035706e-02 2.51491517e-01 7.49928772e-01
8.90633106e-01 2.05961317e-01 -1.01184674e-01 -5.02796412e-01
4.47471917e-01 3.19868535e-01 4.97019403e-02 5.01861274e-01
-1.30640280e+00 5.90867996e-01 6.43299246e+00 7.45630503e-01
-1.60528588e+00 1.85558662e-01 -9.59667861e-02 -2.61980891e-01
3.00951183e-01 -2.93371886e-01 -7.95901895e-01 4.58334565e-01
1.13085604e+00 4.18294728e-01 2.56143302e-01 7.01776087e-01
3.45657527e-01 -2.76478976e-02 -1.44135773e+00 1.05764318e+00
-2.22926944e-01 -7.82487571e-01 -5.23997962e-01 -5.88474631e-01
2.04849482e-01 -7.94013068e-02 -1.09274313e-01 3.09718609e-01
-4.39513117e-01 -3.83926660e-01 1.21775556e+00 4.97997642e-01
5.29685795e-01 -6.59659624e-01 4.67264026e-01 3.53699267e-01
-1.79106152e+00 -2.35351697e-01 -4.00597695e-03 -4.91008580e-01
5.99491596e-01 9.62441742e-01 -1.02951896e+00 2.50526965e-01
8.81110072e-01 2.84494132e-01 -1.59828484e-01 1.51639283e+00
-5.19633055e-01 1.04800355e+00 -5.72314918e-01 -1.58207759e-01
-5.92339896e-02 4.08374310e-01 7.58958697e-01 1.82959461e+00
4.58385497e-01 -1.69371054e-01 1.81669891e-01 4.98896450e-01
1.71018213e-01 -7.67630711e-02 -2.08138183e-01 2.08454832e-01
7.87922263e-01 1.22285473e+00 -6.44053340e-01 -1.49152264e-01
-3.60174119e-01 6.63155198e-01 -1.01532489e-01 3.15095067e-01
-1.06224298e+00 -1.13728917e+00 4.24620748e-01 5.18287206e-03
8.74128759e-01 -2.05623657e-01 7.46456385e-02 -6.30970120e-01
1.78088158e-01 -5.50337076e-01 3.31716210e-01 -6.32935882e-01
-1.07212663e+00 6.17571294e-01 -2.89731830e-01 -1.48053133e+00
-7.36893773e-01 -5.49513221e-01 -1.04708052e+00 8.06425691e-01
-1.36912894e+00 -7.84239650e-01 9.68721285e-02 5.13150156e-01
5.18466175e-01 1.20005555e-01 8.94425809e-01 7.96883821e-01
-4.23149407e-01 6.17470741e-01 -4.42906693e-02 2.84745783e-01
8.17210376e-01 -1.18421817e+00 4.13993359e-01 9.88149047e-01
3.77437532e-01 2.11324066e-01 5.98260820e-01 -2.68766880e-01
-1.19894540e+00 -9.56243694e-01 1.12530887e+00 -1.20641567e-01
6.64324641e-01 -6.98172748e-01 -6.39839947e-01 3.41464043e-01
6.46460354e-02 1.96260303e-01 6.52590752e-01 -1.30778402e-01
-1.88974574e-01 -5.51787019e-01 -8.01209688e-01 2.10186303e-01
4.12999004e-01 -9.10476625e-01 -7.87674308e-01 1.84785292e-01
6.40890241e-01 -3.35614681e-01 -5.04678905e-01 1.77461833e-01
7.30357885e-01 -9.90380824e-01 1.13868773e+00 -1.51271746e-02
-3.25378589e-02 -6.31105542e-01 -2.22148150e-01 -1.09561408e+00
-2.79861987e-01 -8.30016971e-01 -2.51554579e-01 1.45377624e+00
4.82203007e-01 -4.79723662e-01 7.28286654e-02 -1.75938636e-01
-2.08219349e-01 -3.64641815e-01 -9.60904300e-01 -4.88402873e-01
-5.08526027e-01 -1.07595718e+00 5.46117842e-01 3.20994884e-01
4.13084775e-02 5.67024291e-01 -3.04625273e-01 7.74283528e-01
2.11817980e-01 2.50881940e-01 4.36267704e-01 -9.56541121e-01
-6.05540335e-01 -3.17815125e-01 -1.83099970e-01 -1.26919651e+00
-4.07019049e-01 -4.45969313e-01 8.06768775e-01 -1.32874906e+00
-6.54649913e-01 2.67200749e-02 -7.62756467e-01 2.81667858e-01
-2.88305014e-01 1.22897252e-01 -3.32519040e-02 -1.32332876e-01
-6.79004848e-01 3.36283326e-01 5.46245277e-01 2.11949676e-01
-5.20161092e-01 5.71667194e-01 5.19945063e-02 1.03606033e+00
4.36543405e-01 -8.19921613e-01 -2.35827789e-01 -5.87774575e-01
5.33477776e-02 4.86894369e-01 4.31768030e-01 -1.56332159e+00
7.55947411e-01 4.16752934e-01 4.18614209e-01 -8.47772837e-01
4.63977069e-01 -9.99440491e-01 -2.08829686e-01 5.61803341e-01
-3.36070806e-01 -1.35395646e-01 2.75642693e-01 5.17054856e-01
-4.63046730e-01 -2.93023467e-01 6.85422778e-01 2.17377737e-01
-3.98313284e-01 -1.85135320e-01 -9.08588231e-01 1.32518644e-02
6.24984860e-01 3.94529514e-02 1.11594379e-01 -3.33095044e-01
-5.79135537e-01 -3.41543198e-01 -5.81622779e-01 3.94694567e-01
5.59419215e-01 -1.18670225e+00 -7.40768075e-01 4.98406291e-01
-2.04401463e-01 -1.70533240e-01 2.54200935e-01 8.22474480e-01
-3.02131027e-01 3.18302661e-01 9.61781815e-02 -7.67198205e-01
-9.52653527e-01 3.48058581e-01 5.79459548e-01 -1.14321813e-01
-2.48749211e-01 1.17883325e+00 1.92285851e-01 -2.57996380e-01
6.29960656e-01 -5.78706980e-01 -1.73809797e-01 2.07425907e-01
8.26337636e-01 6.24538660e-01 1.80613607e-01 -2.50650197e-01
-5.87270677e-01 7.09735453e-01 2.57311910e-01 -4.61133987e-01
1.19149649e+00 -2.70752281e-01 8.36598575e-02 7.28071570e-01
1.33388507e+00 3.78911167e-01 -9.93557155e-01 -1.02964371e-01
-1.96544841e-01 -1.71870932e-01 3.27663958e-01 -7.80046940e-01
-1.07038224e+00 1.31783354e+00 9.65690911e-01 6.47852600e-01
1.58133268e+00 -1.84739858e-01 8.98539186e-01 2.41056412e-01
4.48433571e-02 -9.44780052e-01 1.31744161e-01 5.04435718e-01
6.69022381e-01 -7.39369810e-01 -3.99390608e-01 -2.45081440e-01
-6.43169973e-03 1.20476246e+00 2.88511395e-01 -1.81048185e-01
1.05719519e+00 8.07163596e-01 2.21513227e-01 8.93515572e-02
-5.66044331e-01 -1.95260271e-01 5.09139419e-01 4.77295190e-01
6.03451610e-01 7.25182379e-03 -8.08065757e-02 9.64478135e-01
-2.48629779e-01 -1.25946864e-01 -3.54133435e-02 8.94636571e-01
-6.24152422e-01 -9.05481577e-01 -6.56031609e-01 -1.15980051e-01
-6.47219241e-01 -1.89044267e-01 -2.10198030e-01 2.75189072e-01
4.19328779e-01 1.43534970e+00 3.05406570e-01 -5.51276207e-01
6.63403332e-01 2.96377122e-01 1.60699069e-01 -4.58998561e-01
-9.48265254e-01 6.42950058e-01 -7.17247352e-02 -3.85355234e-01
-4.34893603e-03 -6.39987051e-01 -1.35574162e+00 4.34929192e-01
-6.99253798e-01 2.73084581e-01 8.31272185e-01 8.21100652e-01
3.88574332e-01 1.03944635e+00 1.06352735e+00 -1.05615509e+00
-4.48195785e-01 -9.98688698e-01 -5.35603344e-01 3.03665735e-02
7.25380599e-01 -2.40228072e-01 -5.62554836e-01 2.47260645e-01] | [15.154099464416504, 5.302259922027588] |
53dd4c65-c0c2-45e5-8dbf-03bfd6c42f52 | seeg-semantic-energized-co-speech-gesture | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Liang_SEEG_Semantic_Energized_Co-Speech_Gesture_Generation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Liang_SEEG_Semantic_Energized_Co-Speech_Gesture_Generation_CVPR_2022_paper.pdf | SEEG: Semantic Energized Co-Speech Gesture Generation | Talking gesture generation is a practical yet challenging task which aims to synthesize gestures in line with speech. Gestures with meaningful signs can better convey useful information and arouse sympathy in the audience. Current works focus on aligning gestures with the speech rhythms, which are hard to mine the semantics and model semantic gestures explicitly. In this paper, we propose a novel method SEmantic Energized Generation (SEEG), for semantic-aware gesture generation. Our method contains two parts: DEcoupled Mining module (DEM) and Semantic Energizing Module (SEM). DEM decouples the semantic-irrelevant information from inputs and separately mines information for the beat and semantic gestures. SEM conducts semantic learning and produces semantic gestures. Apart from representational similarity, SEM requires the predictions to express the same semantics as the ground truth. Besides, a semantic prompter is designed in SEM to leverage the semantic-aware supervision to predictions. This promotes the networks to learn and generate semantic gestures. Experimental results reported in three metrics on different benchmarks prove that SEEG efficiently mines semantic cues and generates semantic gestures. In comparison, SEEG outperforms other methods in all semantic-aware evaluations on different datasets. Qualitative evaluations also indicate the superiority of SEEG in semantic expressiveness. | ['Yi Yang', 'Pan Pan', 'Li Hu', 'Linchao Zhu', 'Qianyu Feng', 'Yuanzhi Liang'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['gesture-generation'] | ['robots'] | [ 2.98099697e-01 4.19744343e-01 -3.45352829e-01 -8.01440358e-01
-6.12662673e-01 -3.98699462e-01 9.10791099e-01 -7.42791593e-01
-2.82304622e-02 4.04949129e-01 1.11377478e+00 3.70135128e-01
9.19084065e-03 -4.28755343e-01 -3.64263237e-01 -7.44873166e-01
3.04403305e-01 4.91258711e-01 -2.36931667e-01 -4.09398258e-01
1.91351861e-01 4.46973294e-02 -1.54134572e+00 6.67659700e-01
4.53512549e-01 1.11549306e+00 1.73927218e-01 5.35679579e-01
-5.89125454e-01 8.41617107e-01 -8.12789679e-01 -3.11169982e-01
1.58876300e-01 -9.37463641e-01 -8.89558733e-01 -1.05806097e-01
-2.00867459e-01 -3.81592542e-01 -1.01487942e-01 7.55147338e-01
7.08165586e-01 3.34696144e-01 6.44555688e-01 -1.62052619e+00
-2.65682042e-01 1.31347704e+00 -2.53929257e-01 -4.99845862e-01
5.53873003e-01 4.35200781e-01 1.25015736e+00 -7.05886066e-01
7.83752084e-01 1.76201308e+00 2.34011248e-01 1.34729993e+00
-6.07842684e-01 -1.06438959e+00 4.56117511e-01 -1.31634390e-02
-8.47489893e-01 -4.79445368e-01 1.18737090e+00 -1.52626380e-01
5.75494647e-01 6.20496988e-01 6.70851290e-01 1.95363450e+00
-9.86558020e-01 1.30874324e+00 9.49523926e-01 -4.00682949e-02
3.48347217e-01 -1.22369125e-01 3.32577005e-02 1.95212245e-01
-3.75891358e-01 3.65886509e-01 -1.19842041e+00 1.08126976e-01
5.77760756e-01 1.03356671e-02 -1.98450699e-01 1.44436330e-01
-1.53445530e+00 8.86204064e-01 4.81037915e-01 4.80719626e-01
-3.97318751e-01 4.83830422e-01 4.68441993e-01 1.11926585e-01
8.20095167e-02 6.88661277e-01 -1.57626197e-01 -5.30035913e-01
-7.92014718e-01 2.43673488e-01 1.02771556e+00 9.65795934e-01
2.78089881e-01 2.47601107e-01 -3.61057222e-01 6.95408344e-01
4.36302513e-01 7.40513444e-01 8.85694563e-01 -1.15633619e+00
4.24164444e-01 7.81402826e-01 -1.54683709e-01 -9.69092309e-01
-3.78044009e-01 -1.65499985e-01 -6.93430781e-01 -1.38792709e-01
3.10190022e-01 -2.21644521e-01 -6.77289665e-01 2.08799696e+00
3.14294130e-01 3.63963693e-01 3.71735722e-01 1.58088171e+00
1.47555053e+00 5.14055192e-01 3.89993608e-01 1.29943890e-02
1.24942350e+00 -1.05754209e+00 -7.82553911e-01 -1.68940663e-01
4.40207154e-01 -6.95176244e-01 1.60556877e+00 1.84934691e-01
-6.81065857e-01 -2.99840629e-01 -8.65912139e-01 1.10508583e-01
-3.52900140e-02 2.45476723e-01 7.95809865e-01 3.11634868e-01
-5.74848056e-01 4.90219533e-01 -7.99543083e-01 -5.94899893e-01
3.70739311e-01 2.21282646e-01 -1.13876173e-02 5.67797542e-01
-1.19248617e+00 5.02635658e-01 3.82537842e-01 -7.41201192e-02
-8.79496157e-01 -3.64575535e-01 -7.07760215e-01 -1.09189212e-01
4.04970795e-01 -6.26218498e-01 1.43022299e+00 -1.44646728e+00
-2.07622218e+00 6.93261683e-01 -2.08949208e-01 -2.82369196e-01
6.71885073e-01 -3.74161810e-01 -1.18941844e-01 2.41445586e-01
-1.03817090e-01 1.19890416e+00 8.45131814e-01 -1.44067407e+00
-4.12762612e-01 -3.07064861e-01 -2.46951648e-04 5.50166428e-01
-4.79505360e-01 8.24795738e-02 -2.26339310e-01 -9.93513882e-01
2.07483068e-01 -8.15294981e-01 -1.77801382e-02 -2.83506095e-01
-7.37586141e-01 -4.00511354e-01 9.94443417e-01 -4.80520040e-01
1.02501392e+00 -2.04475331e+00 3.32340419e-01 3.17213804e-01
1.33031324e-01 -1.04357369e-01 -3.13937962e-01 3.09130937e-01
2.08920464e-01 -1.98363513e-01 -3.06227095e-02 -2.41310671e-01
3.12559187e-01 4.81107324e-01 -7.97794521e-01 -2.08116412e-01
1.27065564e-02 1.29257202e+00 -1.10224104e+00 -4.29965168e-01
2.63319582e-01 3.94640684e-01 -7.15462744e-01 5.83053410e-01
-5.72372675e-01 1.01403022e+00 -9.66406763e-01 7.18533635e-01
2.48847716e-02 -2.80441970e-01 3.54174703e-01 -1.37810886e-01
3.06235611e-01 5.91174066e-01 -9.70896482e-01 2.14971089e+00
-5.75806499e-01 3.46264690e-01 -1.39622346e-01 -9.97728527e-01
1.30874836e+00 4.52638835e-01 5.35344899e-01 -5.79241872e-01
4.66520071e-01 2.75912404e-01 -1.69920847e-01 -7.98076034e-01
1.42188877e-01 -3.97031516e-01 -4.04168874e-01 8.69494081e-01
-2.20870785e-02 -3.55080873e-01 -4.15890992e-01 2.00969175e-01
7.58215487e-01 5.71048558e-01 6.55108243e-02 1.75236061e-01
1.28663823e-01 -1.80887748e-02 4.29427862e-01 3.40665132e-01
3.74516621e-02 3.96673113e-01 2.70662546e-01 -1.28780916e-01
-4.06552136e-01 -9.99109924e-01 6.94512904e-01 1.40513146e+00
6.77672029e-01 -5.23074746e-01 -9.55010831e-01 -6.94402158e-01
-3.36721957e-01 9.23862219e-01 -4.92081434e-01 -1.39737308e-01
-6.72880948e-01 -5.00123203e-01 8.60392272e-01 8.01680326e-01
4.09232110e-01 -1.64911282e+00 -9.74014401e-01 1.43858735e-02
-7.91294992e-01 -1.03853869e+00 -3.34459990e-01 -3.28709409e-02
-7.14299977e-01 -9.60801482e-01 -6.08880341e-01 -4.59859699e-01
4.41667825e-01 1.01301178e-01 7.09707260e-01 4.47745100e-02
-6.44398779e-02 2.00608820e-01 -8.58725369e-01 -5.66124260e-01
-3.48882407e-01 7.36275017e-02 -1.31337181e-01 1.63643196e-01
7.08777487e-01 -8.04309249e-01 -5.37463784e-01 3.97542566e-01
-5.86975157e-01 5.56493521e-01 5.88986158e-01 9.08647716e-01
1.35314912e-01 -7.76836157e-01 6.31804883e-01 -3.48304838e-01
6.15882039e-01 -4.26965535e-01 1.43066257e-01 -7.25495145e-02
-2.79211342e-01 4.14491504e-01 3.88603240e-01 -5.84007382e-01
-1.23084259e+00 2.79562503e-01 -8.55650753e-02 -6.39476538e-01
-3.30063045e-01 1.52360427e-03 -5.02544582e-01 5.22393882e-01
5.67934811e-01 1.76253691e-01 5.01531102e-02 -6.96463704e-01
7.78915167e-01 8.54560971e-01 8.73101532e-01 -8.63628626e-01
7.70291090e-01 5.64070642e-01 -2.39067242e-01 -5.47449887e-01
-8.48922610e-01 -5.07285178e-01 -3.52381915e-01 -4.69327688e-01
9.61100817e-01 -7.40765095e-01 -9.64335263e-01 3.33152324e-01
-1.30249345e+00 -5.08294046e-01 -5.71174741e-01 5.80192804e-01
-9.07450438e-01 -8.68707597e-02 -3.60284239e-01 -8.58252883e-01
-5.74765742e-01 -9.98357475e-01 1.54498935e+00 2.65622586e-01
-9.79491949e-01 -5.52019179e-01 -2.73542225e-01 5.48905551e-01
2.78536022e-01 4.98357594e-01 5.37197590e-01 -1.04508328e+00
-4.55096453e-01 3.32674295e-01 -2.26630583e-01 2.79986747e-02
1.64658859e-01 -3.16927820e-01 -1.35997725e+00 4.64675605e-01
-1.30194649e-01 -7.47253537e-01 7.65859663e-01 2.55499575e-02
1.17971969e+00 -8.10881436e-01 -2.61341780e-01 7.56689847e-01
6.64259791e-01 2.61784554e-01 2.26841003e-01 7.40362927e-02
7.91672349e-01 1.17605889e+00 8.75256658e-01 5.89219630e-01
3.32168967e-01 9.06372964e-01 3.51219177e-01 -6.30483702e-02
-4.95951205e-01 -8.68609905e-01 6.07670486e-01 8.24969769e-01
-1.75326616e-01 4.07855213e-02 -5.82562029e-01 2.82652587e-01
-2.15484762e+00 -1.16604590e+00 1.49978429e-01 1.47164524e+00
1.16041625e+00 -4.04718697e-01 2.98599660e-01 2.78451502e-01
5.02502978e-01 2.84910709e-01 -6.76559091e-01 -1.39407754e-01
-2.59012818e-01 3.84081990e-01 -1.40050769e-01 3.84808213e-01
-8.56764734e-01 1.51910257e+00 4.95101213e+00 8.46548975e-01
-1.40035248e+00 2.86115140e-01 4.07027990e-01 -5.02386689e-01
-5.80319107e-01 -1.04944341e-01 -6.21231258e-01 4.07685727e-01
4.45450515e-01 1.15540981e-01 5.38826168e-01 1.03888535e+00
5.44159949e-01 5.20208299e-01 -1.21799040e+00 1.18803287e+00
1.38178945e-01 -1.16386962e+00 4.82497960e-01 -4.30188149e-01
5.11459768e-01 -4.17783052e-01 -1.36740685e-01 2.32296094e-01
6.41278803e-01 -1.26779222e+00 1.04304576e+00 4.31072056e-01
8.67823303e-01 -4.35536921e-01 4.82046753e-01 2.74978399e-01
-1.07597291e+00 -4.31855805e-02 1.78188637e-01 -2.16951877e-01
3.19350511e-01 4.27892581e-02 -1.01495636e+00 2.45209441e-01
5.61904907e-01 8.52313101e-01 2.51669623e-02 1.81491181e-01
-8.56251538e-01 7.00496912e-01 -3.78444493e-01 -4.81750190e-01
5.44650197e-01 5.21773957e-02 7.93742776e-01 1.30685151e+00
2.57681549e-01 3.41383487e-01 2.02203259e-01 1.10689330e+00
4.75305505e-02 3.24724078e-01 -3.42454910e-01 -3.52702349e-01
7.12619603e-01 1.15233696e+00 -6.34304821e-01 -4.40736413e-01
1.00950859e-01 1.02571309e+00 -2.52607137e-01 3.95911664e-01
-7.86131978e-01 -1.38528898e-01 8.82465184e-01 -1.75559655e-01
-2.13656396e-01 2.01590821e-01 -7.99477935e-01 -1.06453717e+00
3.13246646e-03 -9.76810455e-01 3.07574004e-01 -8.44408035e-01
-1.06733167e+00 5.96068561e-01 -1.67372152e-02 -1.26857746e+00
-8.02369237e-01 -3.63663435e-01 -8.28031123e-01 5.40752828e-01
-1.28369260e+00 -1.54874671e+00 -8.00725281e-01 7.60862470e-01
9.61623132e-01 -2.47131020e-01 8.50584805e-01 -1.76134527e-01
-1.59313977e-01 6.17212713e-01 -8.39410603e-01 3.18146646e-01
5.98940551e-01 -1.02106655e+00 1.97733596e-01 4.18019950e-01
4.08194512e-01 5.29623330e-01 6.77465260e-01 -6.00118935e-01
-1.26620483e+00 -1.02747405e+00 8.87417018e-01 -3.43013883e-01
5.92914820e-01 -2.33081475e-01 -4.02167201e-01 4.60725725e-01
-1.15622751e-01 -6.49638236e-01 7.74732053e-01 -7.61022568e-02
-6.27821624e-01 2.46099100e-01 -7.37097323e-01 7.13595092e-01
1.79154861e+00 -3.68327707e-01 -9.83775496e-01 1.14779651e-01
1.07118428e+00 -2.88177580e-01 -2.40840897e-01 3.64000589e-01
8.84603262e-01 -8.19442153e-01 9.22388673e-01 -8.29038680e-01
6.69936121e-01 -1.61318034e-01 -4.35183316e-01 -1.27157474e+00
5.37946045e-01 -9.88536358e-01 -8.29865485e-02 1.28185594e+00
2.45325699e-01 -1.55836225e-01 9.12684858e-01 3.60247195e-01
-1.70425937e-01 -5.36794722e-01 -5.94349742e-01 -7.67201483e-01
-4.05854315e-01 -7.32568741e-01 1.10781121e+00 1.09898078e+00
3.74160826e-01 5.51800966e-01 -5.75517118e-01 -3.30870599e-01
3.79622608e-01 5.36680102e-01 1.11588192e+00 -1.13723433e+00
-5.32390237e-01 -9.29342210e-01 -1.70148566e-01 -1.11562371e+00
4.37139720e-01 -1.27677834e+00 2.09824741e-01 -1.38883126e+00
1.47835881e-01 -4.32739615e-01 -2.82374118e-02 9.45300996e-01
-8.53171498e-02 2.18194410e-01 5.29790342e-01 4.40047473e-01
-5.07075608e-01 8.45492184e-01 1.35711527e+00 -1.86166018e-01
-5.51872492e-01 8.51723482e-04 -9.81047213e-01 1.08308208e+00
8.41042697e-01 -3.42632115e-01 -6.72321916e-01 -2.02382922e-01
-9.05565843e-02 -9.31762718e-03 6.99164629e-01 -6.95321977e-01
-7.64308171e-03 -4.94567126e-01 7.43793547e-02 -3.86556536e-01
6.89454257e-01 -5.63365698e-01 1.00758597e-02 4.33650166e-01
-8.68166268e-01 -5.66053212e-01 -3.92442346e-01 3.93894464e-01
-3.76106262e-01 2.05905810e-01 4.89299208e-01 -4.17692885e-02
-8.96533132e-01 -4.12086099e-02 2.29131356e-01 1.95467293e-01
7.96560466e-01 -2.94373184e-01 -2.16546208e-02 -8.46768796e-01
-8.65580916e-01 2.31861323e-01 -4.40360978e-02 1.03535581e+00
7.71985710e-01 -1.57364309e+00 -6.32909834e-01 -9.54594314e-02
2.54976183e-01 -6.65166080e-02 -6.18349351e-02 8.11448753e-01
6.14819266e-02 2.49518186e-01 -9.51606557e-02 -5.67206919e-01
-1.29708326e+00 7.59474561e-02 5.78188105e-03 4.28769439e-02
-7.49856770e-01 1.24559736e+00 4.71722692e-01 -5.54657459e-01
5.95235646e-01 -4.17679876e-01 -2.30611280e-01 1.77915871e-01
5.83830416e-01 2.68209606e-01 -5.86656749e-01 -7.12116241e-01
-2.54531622e-01 4.94747519e-01 6.38913214e-01 -6.65659547e-01
1.26639950e+00 8.69020447e-02 2.34030977e-01 5.00452936e-01
9.44963217e-01 -1.50941402e-01 -1.27226675e+00 -2.35570014e-01
2.85775036e-01 -3.01094413e-01 -4.17817771e-01 -1.26584649e+00
-1.16410041e+00 1.20621800e+00 1.13925770e-01 -5.79309285e-01
1.18793440e+00 4.24981505e-01 1.20788944e+00 4.34868664e-01
2.73832113e-01 -1.17869830e+00 7.14386225e-01 3.18959504e-01
1.28416800e+00 -9.68844831e-01 -6.55914426e-01 -4.28301632e-01
-1.23266625e+00 9.24553156e-01 7.86286533e-01 1.04428880e-01
2.30107173e-01 4.20978367e-01 4.72375900e-01 -3.80356103e-01
-6.79812849e-01 -3.12481344e-01 3.07962358e-01 6.74734890e-01
2.70348191e-01 5.04692614e-01 -4.04002517e-01 1.44740975e+00
-9.78581071e-01 -2.61684619e-02 -1.86646074e-01 5.04893899e-01
-4.46678281e-01 -1.06429350e+00 -2.79219270e-01 3.18525992e-02
1.29314303e-01 -9.67464000e-02 -1.20527399e+00 5.73465824e-01
1.48998303e-02 1.11403573e+00 -1.46710843e-01 -8.60725224e-01
5.33341095e-02 3.44648302e-01 2.63426274e-01 -5.44015944e-01
-7.38066435e-01 3.28807652e-01 2.18436837e-01 -1.12302959e+00
-7.29215920e-01 -5.37097931e-01 -1.97496235e+00 -1.27700232e-02
3.37301716e-02 2.18424216e-01 8.20583701e-01 1.14582670e+00
2.37553954e-01 4.88245249e-01 6.29826367e-01 -8.08478296e-01
-8.15522492e-01 -1.13511443e+00 -2.33196720e-01 9.18606162e-01
-1.73839375e-01 -8.08326364e-01 -3.44855756e-01 2.37210393e-01] | [5.642239093780518, -0.133840411901474] |
51539e48-1265-4029-8c70-a8c0c62b1028 | learning-better-intent-representations-for | 2210.14304 | null | https://arxiv.org/abs/2210.14304v1 | https://arxiv.org/pdf/2210.14304v1.pdf | Learning Better Intent Representations for Financial Open Intent Classification | With the recent surge of NLP technologies in the financial domain, banks and other financial entities have adopted virtual agents (VA) to assist customers. A challenging problem for VAs in this domain is determining a user's reason or intent for contacting the VA, especially when the intent was unseen or open during the VA's training. One method for handling open intents is adaptive decision boundary (ADB) post-processing, which learns tight decision boundaries from intent representations to separate known and open intents. We propose incorporating two methods for supervised pre-training of intent representations: prefix-tuning and fine-tuning just the last layer of a large language model (LLM). With this proposal, our accuracy is 1.63% - 2.07% higher than the prior state-of-the-art ADB method for open intent classification on the banking77 benchmark amongst others. Notably, we only supplement the original ADB model with 0.1% additional trainable parameters. Ablation studies also determine that our method yields better results than full fine-tuning the entire model. We hypothesize that our findings could stimulate a new optimal method of downstream tuning that combines parameter efficient tuning modules with fine-tuning a subset of the base model's layers. | ['Stephen W. Thomas', 'Xiaodan Zhu', 'Will Aitken', 'Xianzhi Li'] | 2022-10-25 | null | null | null | null | ['intent-classification'] | ['natural-language-processing'] | [-1.59004331e-01 5.45563638e-01 -4.24847424e-01 -6.51021540e-01
-8.00716221e-01 -1.02769458e+00 8.02794397e-01 3.19228232e-01
-4.90399420e-01 4.85704929e-01 5.08400381e-01 -8.46484303e-01
1.26653209e-01 -8.31624031e-01 -3.41744632e-01 -2.46168319e-02
1.14680327e-01 1.02620149e+00 -1.54393807e-01 -5.48629165e-01
3.70529354e-01 4.54851985e-01 -1.01777923e+00 6.59708738e-01
7.11442173e-01 1.09194601e+00 -1.25890270e-01 4.92189586e-01
-5.20984709e-01 7.37337470e-01 -7.91182935e-01 -5.77342927e-01
4.14697826e-01 2.02599078e-01 -1.27491963e+00 -2.75782734e-01
1.69033810e-01 -4.99374628e-01 -6.40032589e-02 4.32064712e-01
3.46530735e-01 2.25533426e-01 8.41106951e-01 -1.08809960e+00
-9.73934293e-01 9.12028670e-01 -3.48007500e-01 4.80713129e-01
5.76053500e-01 1.33537710e-01 1.51022816e+00 -7.60341167e-01
6.60672486e-01 1.04118001e+00 6.46186888e-01 4.52047914e-01
-1.24871790e+00 -4.82019186e-01 4.36859250e-01 -2.41346415e-02
-1.26721168e+00 -4.25721169e-01 5.72732151e-01 -5.67553103e-01
1.66756034e+00 3.02916676e-01 2.68764049e-01 1.43045998e+00
2.95465328e-02 9.02327836e-01 7.10716128e-01 -4.08435285e-01
3.05896908e-01 6.67883337e-01 7.19984829e-01 3.76145542e-01
2.62858778e-01 -4.77526411e-02 -2.07744673e-01 -6.06508136e-01
4.84464169e-01 -1.66893527e-01 6.83628544e-02 -6.11029826e-02
-8.95893455e-01 1.41736960e+00 4.84949023e-01 3.44473451e-01
-5.99768281e-01 -2.83582628e-01 7.12264299e-01 4.34546292e-01
4.26202953e-01 9.19144094e-01 -8.33035588e-01 -1.68480694e-01
-8.48349750e-01 2.52154887e-01 1.36444902e+00 6.71379268e-01
6.25188351e-01 -2.48837769e-01 -4.23749357e-01 9.26655829e-01
4.39746559e-01 -2.77834654e-01 8.22279811e-01 -5.89571238e-01
7.09881485e-01 9.36685681e-01 2.47374892e-01 -4.91103768e-01
-5.28854966e-01 -6.54078007e-01 -3.01831424e-01 1.69348065e-03
3.91506284e-01 -3.83588970e-01 -1.02928960e+00 1.43305683e+00
1.41036242e-01 -2.48633534e-01 2.69536406e-01 6.84282839e-01
1.01679492e+00 6.98261082e-01 2.68751860e-01 1.38068851e-03
1.53325725e+00 -1.03942955e+00 -3.81127208e-01 -5.82321882e-01
1.04037416e+00 -6.89720035e-01 1.18655288e+00 4.38984454e-01
-7.18888104e-01 -2.96948761e-01 -1.02964664e+00 -1.65915281e-01
-6.43502772e-01 -1.21822849e-01 1.14063811e+00 7.88695335e-01
-9.35790062e-01 2.02819437e-01 -3.08021516e-01 -4.64825749e-01
4.14688408e-01 4.27607328e-01 -9.62746963e-02 3.04366201e-01
-1.46521175e+00 1.01347995e+00 3.24660212e-01 -2.42227003e-01
-5.67163348e-01 -9.23063040e-01 -9.52888787e-01 3.26036066e-01
1.74129009e-01 -8.17196310e-01 1.51334226e+00 -8.60461116e-01
-1.29818857e+00 9.87136543e-01 -1.85838729e-01 -7.97557116e-01
5.40468872e-01 -3.83266687e-01 -3.66011798e-01 -2.63834685e-01
2.62076080e-01 7.05141187e-01 6.86262369e-01 -1.08624578e+00
-5.60153723e-01 -1.61203623e-01 4.39668834e-01 1.64603144e-01
-4.79044825e-01 8.55787396e-02 -2.43591443e-01 -5.92810392e-01
-6.35595247e-02 -5.87141156e-01 -3.35922122e-01 -6.15994811e-01
-3.33949178e-01 -7.16147780e-01 6.13211513e-01 -6.69561148e-01
1.37122142e+00 -1.80588007e+00 -3.28264058e-01 2.84118414e-01
1.15240514e-01 4.03207451e-01 -2.87362665e-01 4.42017853e-01
-1.20065778e-01 3.90097648e-01 1.82932720e-01 -4.85212028e-01
1.56292573e-01 1.58245727e-01 -8.50441754e-01 7.29887858e-02
3.31933312e-02 1.17106700e+00 -4.14129823e-01 -3.43719095e-01
1.08589455e-01 3.71059328e-01 -9.26870584e-01 2.31704384e-01
-3.40842515e-01 6.75392374e-02 -4.49839503e-01 8.13283563e-01
5.96134782e-01 -4.91955638e-01 1.87808812e-01 -1.35468347e-02
1.30377471e-01 9.95090961e-01 -1.05647624e+00 1.46739674e+00
-7.26291537e-01 4.88242775e-01 3.26793790e-02 -7.58224607e-01
8.76955628e-01 1.70387492e-01 3.28661978e-01 -6.45098507e-01
3.07296336e-01 8.02344754e-02 -1.81960464e-01 -3.26087445e-01
6.57822013e-01 -1.26489982e-01 -4.60153848e-01 7.45701134e-01
-3.96705642e-02 1.92586824e-01 -2.80954745e-02 2.88032383e-01
1.20138967e+00 -1.52032197e-01 5.52474201e-01 -9.97225642e-02
5.11239231e-01 4.12628092e-02 4.21883494e-01 1.03049052e+00
-3.99714142e-01 4.43667054e-01 5.79965532e-01 -6.19155586e-01
-4.89758492e-01 -7.42988765e-01 -3.37249100e-01 1.50876868e+00
-2.51924157e-01 -4.75908905e-01 -5.24984360e-01 -1.23678112e+00
5.21881461e-01 1.24311209e+00 -6.71298206e-01 5.30717671e-02
-6.12751901e-01 -7.02918768e-01 6.37171507e-01 5.43464780e-01
2.88899332e-01 -1.17409110e+00 -3.91391844e-01 3.20483625e-01
-2.45293409e-01 -1.07597518e+00 -3.13339114e-01 4.49546576e-01
-6.55122280e-01 -6.96291625e-01 -4.37825412e-01 -7.16781199e-01
3.67158830e-01 -7.59076104e-02 1.26639593e+00 -7.52627701e-02
4.03349251e-02 2.03845456e-01 -4.57146198e-01 -2.35436484e-01
-5.46405315e-01 7.22938180e-01 9.04610753e-02 -4.92223911e-02
7.41430998e-01 -4.42909092e-01 -5.05959630e-01 2.89372116e-01
-5.15710711e-01 -2.08438233e-01 6.38704300e-01 5.81036210e-01
6.59645498e-02 -3.42844516e-01 8.25283587e-01 -1.25929487e+00
9.55182910e-01 -7.66472816e-01 -3.61814678e-01 1.55635789e-01
-9.14433360e-01 8.95499438e-02 4.86143500e-01 -4.19368297e-01
-1.09323132e+00 -3.59976083e-01 -4.13242370e-01 -1.31182417e-01
-3.46631676e-01 6.21042252e-01 4.47339099e-03 5.68119623e-02
7.79911220e-01 -1.64175838e-01 -2.68631876e-01 -5.52373886e-01
4.12976682e-01 1.13400769e+00 1.62040845e-01 -3.47990692e-01
5.30085683e-01 2.31170714e-01 -8.16724658e-01 -4.62890685e-01
-9.55153883e-01 -8.18831861e-01 -4.70904022e-01 3.38390023e-01
4.99335736e-01 -8.69873285e-01 -8.34131718e-01 1.50810227e-01
-1.12431133e+00 -3.79490048e-01 -3.05447698e-01 1.51706532e-01
-3.08256030e-01 4.00028110e-01 -8.63190293e-01 -6.04425013e-01
-7.21202731e-01 -9.59180892e-01 8.84939790e-01 2.36800373e-01
-9.21800196e-01 -1.04567850e+00 1.76288441e-01 9.34931815e-01
4.74954337e-01 -3.87554646e-01 1.01468992e+00 -1.62887585e+00
-1.82291642e-01 -3.91778141e-01 -2.34446540e-01 2.20672861e-01
6.32811412e-02 -3.79934847e-01 -1.16052985e+00 -3.06541115e-01
8.53352696e-02 -3.48067462e-01 9.91737247e-01 3.02998632e-01
7.27030218e-01 -5.78488946e-01 -4.76528913e-01 6.70199871e-01
1.10544872e+00 1.27011850e-01 4.16553080e-01 8.06832910e-01
3.98652732e-01 4.78091598e-01 6.97241187e-01 6.46031618e-01
6.06903553e-01 6.15277052e-01 3.49713594e-01 6.00415282e-02
3.18426877e-01 -2.02600628e-01 5.53442240e-01 3.93681936e-02
2.71267653e-01 -3.72181982e-01 -1.05633569e+00 4.36911017e-01
-1.62807214e+00 -7.09425509e-01 3.22250217e-01 1.97605717e+00
7.72243559e-01 5.24880826e-01 3.00562978e-02 -1.94964036e-01
4.38516855e-01 2.83008277e-01 -5.75681925e-01 -1.02605557e+00
1.27177790e-01 1.27934739e-01 4.64119822e-01 7.27161705e-01
-1.20944560e+00 1.23034048e+00 6.95842886e+00 4.79262143e-01
-9.66918170e-01 6.98646680e-02 8.30503166e-01 -1.75609127e-01
-5.75522542e-01 3.53637934e-02 -1.35290194e+00 3.26690882e-01
1.08498836e+00 9.80109870e-02 3.39780688e-01 1.18848407e+00
-5.42259291e-02 7.85340667e-02 -1.37656379e+00 6.61743164e-01
1.49761960e-01 -1.58871305e+00 2.62566358e-01 3.60704601e-01
4.88822967e-01 2.20937550e-01 -6.85460940e-02 1.18161201e+00
6.34669304e-01 -1.00715244e+00 5.95236957e-01 1.93361491e-01
5.03437340e-01 -5.14805496e-01 8.57054114e-01 4.96559054e-01
-9.42405820e-01 -5.38923562e-01 -1.49537668e-01 -1.64353982e-01
3.64641666e-01 4.08484936e-01 -1.41424298e+00 1.65597141e-01
4.82938141e-01 4.20366108e-01 -5.73260605e-01 6.04622543e-01
-1.08192228e-01 7.08338499e-01 -3.46303016e-01 -1.53661035e-02
4.71950889e-01 -1.36381477e-01 5.57136297e-01 1.30165732e+00
-1.29395843e-01 3.63153778e-02 2.05805823e-01 9.45155263e-01
-2.35919029e-01 1.33971944e-01 -5.61090529e-01 -2.45646194e-01
4.26221967e-01 1.34888148e+00 -4.62782741e-01 -3.56493145e-01
-5.00006676e-01 9.65562820e-01 6.85956299e-01 2.94363558e-01
-9.74538028e-01 -1.57610223e-01 8.82128298e-01 1.20634817e-01
4.39296842e-01 -1.39790233e-02 -5.92604697e-01 -1.06112850e+00
-1.42221734e-01 -8.21013570e-01 1.06098676e+00 -5.30062795e-01
-1.35771632e+00 6.40013933e-01 -2.47467756e-01 -7.49369085e-01
-5.97879767e-01 -7.14080155e-01 -8.05045664e-01 9.27985013e-01
-1.65945804e+00 -1.30445313e+00 3.90154868e-02 2.45079294e-01
8.20601583e-01 -3.51830482e-01 9.52015221e-01 7.38122463e-02
-6.48534238e-01 9.87094045e-01 -2.67765313e-01 3.69275153e-01
7.73763955e-01 -1.08724475e+00 6.79527402e-01 5.27566671e-01
1.42447203e-01 9.63789821e-01 6.45432353e-01 -7.11926460e-01
-1.07685959e+00 -8.25687826e-01 1.21161413e+00 -9.63644147e-01
6.55898690e-01 -5.46840906e-01 -8.67663801e-01 1.18867421e+00
2.73534507e-01 -1.90434739e-01 9.61086154e-01 6.89808547e-01
-4.93839115e-01 2.48625994e-01 -1.24067068e+00 4.07022744e-01
7.63873458e-01 -4.30859774e-01 -1.18083084e+00 3.10745299e-01
1.01354551e+00 -2.71604002e-01 -7.83218563e-01 3.29068184e-01
3.78709972e-01 -8.81474972e-01 1.09627962e+00 -1.04223847e+00
1.99540127e-02 3.52414489e-01 -2.68800378e-01 -9.45908010e-01
-4.06461596e-01 -7.36010134e-01 -3.95572364e-01 1.42957914e+00
7.86308229e-01 -1.00911903e+00 9.90121484e-01 8.88944507e-01
-5.87597629e-03 -9.60183620e-01 -7.67852545e-01 -3.87477607e-01
2.88125634e-01 -5.79083025e-01 6.72648430e-01 1.01578391e+00
2.68360287e-01 6.14797175e-01 1.05029285e-01 1.78270519e-01
7.91798607e-02 3.17827582e-01 5.74941158e-01 -1.27922225e+00
-4.11397636e-01 -5.59947610e-01 -2.57488694e-02 -1.15407813e+00
4.03183997e-01 -1.04784870e+00 -5.19618273e-01 -1.69640815e+00
1.74454022e-02 -7.78473556e-01 -4.76413548e-01 6.95194781e-01
1.11479819e-01 -2.44320273e-01 -5.86249866e-02 3.65966707e-01
-4.58170921e-01 3.48920286e-01 6.17306709e-01 -1.78071648e-01
-7.41787076e-01 3.39241743e-01 -1.23614788e+00 6.30458236e-01
9.00983393e-01 -2.47762233e-01 -3.88575882e-01 -9.62885246e-02
3.85832161e-01 -1.54129043e-01 1.41322240e-01 -4.69493896e-01
3.11737657e-02 -2.24583685e-01 2.51658797e-01 -6.01368368e-01
3.36624354e-01 -4.75619107e-01 -5.29272437e-01 3.26986551e-01
-5.44939578e-01 -2.79294908e-01 3.12832713e-01 2.82009333e-01
4.21647355e-02 -5.23763001e-01 4.11502361e-01 -1.88648835e-01
-8.77521217e-01 -1.30096227e-02 -6.04014337e-01 2.82079220e-01
9.98703718e-01 -2.42103547e-01 -4.37619239e-01 -3.24711353e-01
-6.58441663e-01 3.74279231e-01 1.94668487e-01 6.69545710e-01
3.83158773e-01 -8.00356865e-01 -5.74923992e-01 5.51491916e-01
3.08011398e-02 -1.38423502e-01 3.77499834e-02 6.20199621e-01
-3.48707855e-01 9.83450472e-01 1.61427706e-01 -2.10512385e-01
-1.04999924e+00 4.73009050e-01 3.10656846e-01 -8.12826455e-01
-6.01897359e-01 1.17367685e+00 2.80738860e-01 -8.74053478e-01
4.64559317e-01 -4.14040238e-01 -6.58152223e-01 5.54471552e-01
4.07563388e-01 1.85512498e-01 3.08440983e-01 -5.51053405e-01
-6.00561976e-01 -4.20913612e-03 -7.83942342e-01 -1.65176898e-01
1.24297297e+00 -9.71825272e-02 2.26880342e-01 2.47796625e-01
1.09963858e+00 -1.53549982e-03 -7.82407045e-01 -1.82108060e-01
1.27475904e-02 -1.95846915e-01 1.92878470e-01 -1.08435535e+00
-5.99398315e-01 4.86991286e-01 1.29077286e-01 3.92206818e-01
5.91101587e-01 2.69538999e-01 8.00574601e-01 6.45807743e-01
2.63901561e-01 -1.17346430e+00 -5.37314489e-02 7.72347510e-01
1.03273427e+00 -1.42089617e+00 -8.79902095e-02 -2.10106760e-01
-9.92131293e-01 8.95998359e-01 6.32603824e-01 6.45923913e-02
5.78663349e-01 2.09340215e-01 4.20776814e-01 -1.88678220e-01
-9.99439657e-01 1.17249563e-01 2.58855641e-01 2.56725550e-01
4.78771448e-01 1.07157670e-01 -8.65918249e-02 1.20948112e+00
-4.73728657e-01 -2.20312059e-01 4.45886463e-01 9.66643095e-01
-3.75274479e-01 -1.04663122e+00 -2.08014101e-01 7.79983997e-01
-4.16216344e-01 -3.88162166e-01 -5.43922782e-01 7.55128503e-01
-1.90486833e-01 9.89639580e-01 2.39367157e-01 -4.91484284e-01
2.63641357e-01 5.41402400e-01 -1.67919442e-01 -9.27528143e-01
-1.10150218e+00 -3.30809653e-01 3.80899876e-01 -6.96244538e-01
2.52128571e-01 -8.50355625e-01 -1.45523202e+00 -2.39992425e-01
-4.49141562e-01 2.47913793e-01 4.35463041e-01 9.04286563e-01
6.52927995e-01 3.24362844e-01 5.65045774e-01 -6.32104337e-01
-1.03817856e+00 -1.05676115e+00 -2.66643375e-01 3.45160574e-01
2.96150446e-01 -6.68003559e-01 -6.83373332e-01 -4.74190742e-01] | [11.280313491821289, 7.491511344909668] |
d80b6f38-83bf-4a59-845f-0f9026b889a6 | text-infilling | 1901.00158 | null | http://arxiv.org/abs/1901.00158v2 | http://arxiv.org/pdf/1901.00158v2.pdf | Text Infilling | Recent years have seen remarkable progress of text generation in different
contexts, such as the most common setting of generating text from scratch, and
the emerging paradigm of retrieval-and-rewriting. Text infilling, which fills
missing text portions of a sentence or paragraph, is also of numerous use in
real life, yet is under-explored. Previous work has focused on restricted
settings by either assuming single word per missing portion or limiting to a
single missing portion to the end of the text. This paper studies the general
task of text infilling, where the input text can have an arbitrary number of
portions to be filled, each of which may require an arbitrary unknown number of
tokens. We study various approaches for the task, including a self-attention
model with segment-aware position encoding and bidirectional context modeling.
We create extensive supervised data by masking out text with varying
strategies. Experiments show the self-attention model greatly outperforms
others, creating a strong baseline for future research. | ['Zhiting Hu', 'Wanrong Zhu', 'Eric Xing'] | 2019-01-01 | null | https://openreview.net/forum?id=r1zmVhCqKm | https://openreview.net/pdf?id=r1zmVhCqKm | null | ['text-infilling'] | ['natural-language-processing'] | [ 8.14760029e-01 2.17891321e-01 -2.42336616e-01 -2.06186146e-01
-9.36171293e-01 -6.95729733e-01 8.72880995e-01 1.04137562e-01
-2.83048540e-01 1.12834918e+00 7.32459009e-01 -7.88038194e-01
4.40290928e-01 -6.62022412e-01 -9.52655613e-01 -4.01602507e-01
4.98499215e-01 4.97842848e-01 1.18738003e-01 -4.89219874e-01
4.95975405e-01 -4.88548167e-02 -1.25107348e+00 5.59300363e-01
8.68742645e-01 2.15975240e-01 5.35565853e-01 7.82660306e-01
-7.47507215e-01 4.44174796e-01 -9.53549504e-01 -4.45257157e-01
2.27067187e-01 -6.36729181e-01 -1.03970790e+00 8.72940272e-02
5.74121594e-01 -2.33241931e-01 -1.69041932e-01 7.11032450e-01
6.49029493e-01 -4.34391052e-02 4.46601391e-01 -8.79648566e-01
-1.03100955e+00 1.27583146e+00 -6.77401662e-01 1.76891461e-01
6.10254705e-01 -6.17160536e-02 1.13390958e+00 -1.00073087e+00
7.46902287e-01 1.16663516e+00 3.97578895e-01 7.67905772e-01
-1.08743334e+00 -3.64206344e-01 2.97723174e-01 -1.64045319e-01
-1.13923895e+00 -5.19598782e-01 6.30381823e-01 -2.49146461e-01
1.12356412e+00 4.12558377e-01 2.22675920e-01 1.30952621e+00
2.79500782e-01 9.22794938e-01 8.79022539e-01 -1.05790389e+00
-4.91766520e-02 -2.50889864e-02 4.30929549e-02 2.84531206e-01
1.91395476e-01 -3.29729885e-01 -3.27373832e-01 -7.34571815e-02
5.91105402e-01 -6.29358515e-02 -2.28550643e-01 9.23551247e-02
-1.42418754e+00 7.15395749e-01 -4.05503325e-02 2.95647621e-01
-3.00888587e-02 1.26012072e-01 5.32320619e-01 3.52346957e-01
6.72578335e-01 4.12207067e-01 -4.10449237e-01 -2.27426440e-01
-1.27268088e+00 5.39144039e-01 9.01353300e-01 1.54346192e+00
6.32751048e-01 -1.58569634e-01 -7.63779938e-01 9.20338988e-01
-5.84889352e-02 3.42337102e-01 6.28899813e-01 -3.13240767e-01
1.04503977e+00 4.78356630e-01 3.68513823e-01 -3.39664131e-01
-4.45397794e-02 -4.25362587e-01 -7.54338861e-01 -1.22660778e-01
2.85853714e-01 -6.50492787e-01 -1.06225407e+00 1.78844774e+00
2.36605138e-01 4.97534573e-02 -6.08703531e-02 4.93425399e-01
5.69054723e-01 7.58397162e-01 -1.90265954e-01 -1.29256472e-01
1.16640842e+00 -1.35451877e+00 -8.00784707e-01 -3.67570996e-01
6.58036649e-01 -1.16401100e+00 1.23174572e+00 8.12073052e-02
-1.27472401e+00 -4.27408814e-01 -9.46631968e-01 -5.63563049e-01
-5.91494441e-01 1.54317603e-01 2.44824067e-01 4.19219583e-01
-1.26307452e+00 5.02860367e-01 -4.35754776e-01 -4.09101248e-01
7.17410669e-02 3.42444554e-02 -1.57201126e-01 -1.65964335e-01
-1.21496058e+00 7.19428897e-01 1.42798781e-01 1.34266332e-01
-3.16323906e-01 -7.13763535e-01 -8.82359385e-01 1.26061589e-01
5.23402929e-01 -1.12511837e+00 1.33762527e+00 -8.61062646e-01
-1.25758517e+00 5.49025595e-01 -5.56087136e-01 -4.06050771e-01
6.86026216e-01 -5.30022681e-01 -2.96817988e-01 -3.47070575e-01
3.79125714e-01 7.48549163e-01 9.20623302e-01 -1.19137335e+00
-4.22319114e-01 -6.68058470e-02 6.24387190e-02 4.06895220e-01
-2.46328548e-01 1.30411714e-01 -5.63223720e-01 -1.14044309e+00
-2.90155351e-01 -7.82010913e-01 -3.39493513e-01 -3.57629210e-01
-9.46214199e-01 -2.55648375e-01 7.86631346e-01 -7.51588523e-01
1.71356070e+00 -1.68220985e+00 1.26518011e-01 -2.35845417e-01
-2.59030759e-01 1.80007741e-01 -3.89323622e-01 1.17274272e+00
-1.42619178e-01 6.74507737e-01 -5.07265031e-01 -8.27865303e-01
4.17082496e-02 1.31007314e-01 -7.70263195e-01 -2.12522641e-01
3.58853281e-01 1.10678875e+00 -9.37099934e-01 -5.22693396e-01
-1.66447878e-01 3.16494763e-01 -6.14071488e-01 -1.87231340e-02
-4.91731793e-01 1.29273236e-01 -3.63821328e-01 3.91882896e-01
4.16164309e-01 -1.83740869e-01 -2.00728932e-03 2.96711713e-01
-2.20119342e-01 5.82798004e-01 -1.10346913e+00 1.92280543e+00
-5.85338950e-01 6.35468721e-01 -1.50507122e-01 -4.94315118e-01
6.47296488e-01 2.62765825e-01 5.39587773e-02 -3.67992431e-01
-1.13692522e-01 -4.22679000e-02 -1.68811262e-01 -4.77198154e-01
1.28586459e+00 9.86814033e-03 -2.15314269e-01 8.32220674e-01
-2.83406556e-01 -9.90103111e-02 6.46370292e-01 5.03612280e-01
1.28172970e+00 3.51734668e-01 1.91741273e-01 -8.54579434e-02
2.46835694e-01 3.18798907e-02 1.02137610e-01 1.15531540e+00
2.88787007e-01 1.13688552e+00 4.60807770e-01 7.54572898e-02
-1.44065821e+00 -5.77210903e-01 4.24612015e-02 1.09072232e+00
-2.97337957e-02 -5.66226780e-01 -9.35347021e-01 -7.61253536e-01
-1.21393934e-01 8.66146922e-01 -6.18269384e-01 9.82107222e-02
-1.02733374e+00 -5.28356016e-01 4.37631011e-01 6.73050642e-01
1.38464002e-02 -1.31943703e+00 -2.63328612e-01 2.99715221e-01
-4.31883752e-01 -9.00922537e-01 -1.08964789e+00 2.97109131e-02
-7.97011614e-01 -5.17002344e-01 -8.98429275e-01 -1.05367422e+00
8.58621836e-01 4.46626425e-01 1.19507313e+00 2.81003207e-01
-2.08306521e-01 1.18950069e-01 -5.97543299e-01 -4.75897312e-01
-3.98083001e-01 5.54924369e-01 -4.38244790e-01 -3.72635238e-02
8.07813555e-02 -3.87849152e-01 -4.39962596e-01 -3.08495983e-02
-1.28280699e+00 3.69054198e-01 7.25664198e-01 1.05305064e+00
4.94488895e-01 -4.56655085e-01 8.43295217e-01 -1.33712041e+00
1.13775718e+00 -7.11528540e-01 5.27846664e-02 3.50542694e-01
-3.19345742e-01 2.04011783e-01 8.64515364e-01 -3.05843651e-01
-1.03527498e+00 -3.06800365e-01 -3.02210599e-01 -1.14201419e-01
-2.29692653e-01 6.18850172e-01 -8.71748254e-02 6.64705336e-01
4.73014683e-01 3.23944539e-01 -1.67769715e-01 -5.25715411e-01
6.74461484e-01 7.62363434e-01 4.52511787e-01 -6.49718225e-01
7.68277526e-01 2.15796560e-01 -3.68762404e-01 -7.42817283e-01
-7.56177425e-01 -4.48946059e-01 -5.86481035e-01 2.65960842e-01
4.80396062e-01 -7.15429664e-01 2.23173961e-01 8.64143968e-02
-1.38704419e+00 -5.71003020e-01 -6.09761119e-01 -1.11220688e-01
-2.33419999e-01 7.42217124e-01 -5.17398119e-01 -5.76561809e-01
-6.16128922e-01 -9.92503285e-01 1.47809601e+00 2.68336758e-02
-5.09419918e-01 -9.83159900e-01 1.15939856e-01 1.68311030e-01
4.41879064e-01 2.11392287e-02 9.61742043e-01 -6.72572970e-01
-4.63365465e-01 -2.97154158e-01 3.11325071e-03 7.03417584e-02
1.44463748e-01 -5.11347735e-03 -6.37861013e-01 -1.67237714e-01
-3.15861732e-01 -1.07801400e-01 1.06417978e+00 1.08662583e-01
1.17681539e+00 -3.83675992e-01 -4.48131830e-01 2.25967452e-01
1.24870813e+00 1.38936535e-01 7.93710291e-01 9.59344655e-02
6.21740758e-01 4.33029741e-01 3.89582634e-01 4.54208344e-01
5.29185534e-01 4.38563615e-01 3.78604561e-01 -1.66550547e-01
-3.83226454e-01 -4.62650567e-01 2.43684903e-01 9.27746296e-01
3.80901068e-01 -1.04732955e+00 -5.92809081e-01 8.53653967e-01
-1.81828690e+00 -1.07858634e+00 7.29261115e-02 2.05066323e+00
1.01416540e+00 1.28758103e-01 -1.35520637e-01 2.70118088e-01
9.67923164e-01 2.88520545e-01 -3.20931762e-01 -6.67073011e-01
-1.86421350e-01 3.90424579e-01 2.16633305e-01 6.33311749e-01
-8.97504747e-01 1.10245323e+00 7.08341646e+00 7.67437458e-01
-9.76335108e-01 -6.40913695e-02 5.29108882e-01 -1.96548358e-01
-8.36476386e-01 2.21620306e-01 -1.23768401e+00 7.13919103e-01
7.15644062e-01 -1.33212626e-01 3.54161471e-01 4.33262259e-01
3.37587506e-01 -6.75208569e-02 -1.16430581e+00 5.15180826e-01
4.10537571e-01 -1.34850287e+00 4.24155831e-01 -4.60863374e-02
9.91087496e-01 -9.78853852e-02 -1.46255657e-01 4.10496145e-01
3.52457047e-01 -7.67460585e-01 8.16912055e-01 5.05611181e-01
9.77446556e-01 -3.85406971e-01 5.63488483e-01 4.94689792e-01
-9.02687967e-01 1.04654543e-01 -2.81926513e-01 -1.54473037e-01
4.52629119e-01 5.59171975e-01 -8.41092467e-01 7.23876476e-01
2.12902755e-01 6.07239425e-01 -5.82805872e-01 9.88517463e-01
-2.11828396e-01 7.59431660e-01 -2.93510854e-01 -2.67657667e-01
3.14797521e-01 -1.64345011e-01 5.90885878e-01 1.52145767e+00
6.31522954e-01 -2.63258815e-01 2.08524004e-01 7.90022910e-01
-3.62629801e-01 3.69505018e-01 -6.63840353e-01 4.84179743e-02
5.21122813e-01 1.23442638e+00 -5.73225439e-01 -4.83089507e-01
-5.72247207e-01 1.13053668e+00 3.50741595e-01 5.25939703e-01
-7.95800090e-01 -7.64548004e-01 3.62971365e-01 1.90417632e-01
5.70010722e-01 -1.81171313e-01 -5.84075391e-01 -1.27521980e+00
4.17852461e-01 -6.50305390e-01 1.91688299e-01 -1.02131581e+00
-1.15575576e+00 5.52915335e-01 -6.17081337e-02 -1.08674467e+00
-2.68346906e-01 -1.88345999e-01 -1.01246274e+00 1.16733992e+00
-1.74421561e+00 -1.09377956e+00 2.31871232e-02 2.16334745e-01
1.20084000e+00 1.33323774e-01 5.01159787e-01 2.04274431e-01
-6.36654854e-01 7.20842361e-01 2.65766561e-01 1.90612316e-01
9.28995371e-01 -1.32660317e+00 8.92040849e-01 1.12138379e+00
1.41183987e-01 8.85312438e-01 7.29803622e-01 -9.20766592e-01
-1.33930254e+00 -1.22181296e+00 1.56545913e+00 -5.81562877e-01
6.08313441e-01 -6.81311071e-01 -8.81270587e-01 8.35854173e-01
7.94372380e-01 -3.47465873e-01 5.41998029e-01 -1.31981924e-01
-5.74743487e-02 2.20723420e-01 -6.38291061e-01 1.18337774e+00
1.20620084e+00 -3.17650884e-01 -7.84135282e-01 4.04392540e-01
1.19002378e+00 -6.95357978e-01 -4.28707778e-01 1.21939942e-01
2.78214216e-01 -5.67551553e-01 6.74367726e-01 -6.81114793e-01
9.28977609e-01 -2.86535740e-01 1.81439728e-01 -1.42571354e+00
-1.75935522e-01 -1.05174196e+00 -9.80007201e-02 1.62853003e+00
7.19524026e-01 -3.26290637e-01 7.52860129e-01 3.87367368e-01
-6.11579478e-01 -8.12550902e-01 -6.89551950e-01 -4.72856909e-01
4.61657673e-01 -2.71465510e-01 9.01545525e-01 7.16818690e-01
2.31701732e-01 6.65734172e-01 -6.07783556e-01 -3.42315674e-01
-4.55940887e-02 2.30280563e-01 8.41601372e-01 -7.40780950e-01
-1.56133384e-01 -5.85770071e-01 4.01374817e-01 -1.56885529e+00
8.98988768e-02 -1.06069934e+00 2.20048308e-01 -1.96898437e+00
1.87340274e-01 -3.32583427e-01 1.88691899e-01 4.65243965e-01
-6.36596560e-01 7.37394169e-02 8.45901445e-02 -3.92952897e-02
-4.46989715e-01 4.84561026e-01 1.34054959e+00 -1.27684474e-01
-1.76139459e-01 -9.48508456e-02 -1.01087010e+00 1.48397282e-01
8.20835233e-01 -3.04514110e-01 -5.04472494e-01 -8.14818680e-01
3.82542729e-01 1.55910134e-01 9.86165833e-03 -6.51457369e-01
2.62943000e-01 -2.03793734e-01 1.85933679e-01 -8.51728082e-01
5.40043265e-02 -6.11464560e-01 -1.08094163e-01 -8.36806297e-02
-8.20618927e-01 4.76952523e-01 1.24035463e-01 5.68673253e-01
-8.50983858e-02 -5.57018757e-01 2.09391400e-01 -3.46806437e-01
-2.98933029e-01 1.61416739e-01 -4.22843486e-01 3.62399369e-01
7.21285939e-01 -3.38194191e-01 -4.83326346e-01 -2.99066067e-01
-5.83939672e-01 3.02740216e-01 4.00963336e-01 7.14686096e-01
4.55245167e-01 -1.12493181e+00 -8.21785867e-01 2.34664991e-01
-5.35048246e-02 1.69191986e-01 2.77367476e-02 4.24317747e-01
-3.35584372e-01 5.29676855e-01 2.59397477e-01 -2.86967963e-01
-9.94606793e-01 6.81780219e-01 -1.99254408e-01 -5.59382260e-01
-7.60867834e-01 4.87424821e-01 5.67791499e-02 -3.09720337e-01
1.49165243e-01 -5.54225087e-01 -1.16036847e-01 -2.08950713e-02
6.67400539e-01 2.70335227e-01 1.43159315e-01 -4.09962863e-01
2.09536046e-01 3.98520499e-01 -3.97296339e-01 -2.54808903e-01
9.96427059e-01 -5.89082539e-01 -1.50309503e-01 3.77732515e-01
8.93648386e-01 1.62137553e-01 -8.84058833e-01 -3.38013798e-01
7.66265439e-03 -2.03088194e-01 -3.52526098e-01 -8.48604023e-01
-5.35780370e-01 9.95484054e-01 -3.08367997e-01 3.39848846e-01
8.17010045e-01 -7.98312649e-02 1.02487981e+00 2.39115342e-01
1.77091449e-01 -1.07023358e+00 -3.35705504e-02 8.90238047e-01
1.06237626e+00 -1.15243840e+00 -1.53797150e-01 -2.38543510e-01
-5.02259970e-01 1.03499544e+00 6.28674448e-01 8.96227509e-02
3.74543190e-01 5.24225593e-01 -1.29356921e-01 3.23728353e-01
-1.06210339e+00 -8.34141001e-02 8.12802762e-02 4.34621274e-01
7.62719452e-01 -2.22080365e-01 -6.19399965e-01 4.41198796e-01
-3.94515365e-01 -1.58744127e-01 8.26233625e-01 1.22219753e+00
-2.97912747e-01 -1.49312305e+00 -3.25464308e-01 6.87682152e-01
-6.95790052e-01 -7.26137161e-01 -5.04037261e-01 5.63958824e-01
-3.07440404e-02 9.50170636e-01 2.03760415e-01 4.09185067e-02
1.82252869e-01 5.59176922e-01 3.00897390e-01 -1.11204016e+00
-8.33226562e-01 3.45231324e-01 1.73156559e-01 -1.34359691e-02
-1.09854527e-01 -8.27319682e-01 -1.08049810e+00 -9.32835713e-02
-4.69105273e-01 1.66829467e-01 5.05940557e-01 9.25049543e-01
5.74536264e-01 5.71800649e-01 3.85938406e-01 -8.58515620e-01
-6.10832512e-01 -1.19209933e+00 -2.40994111e-01 3.56894642e-01
4.39716250e-01 -9.44190249e-02 -1.01955317e-01 4.08771664e-01] | [11.765270233154297, 9.088308334350586] |
31a63b66-9aa1-4aac-ace6-fa1054271e40 | a-simple-temporal-information-matching | 2209.09677 | null | https://arxiv.org/abs/2209.09677v1 | https://arxiv.org/pdf/2209.09677v1.pdf | A Simple Temporal Information Matching Mechanism for Entity Alignment Between Temporal Knowledge Graphs | Entity alignment (EA) aims to find entities in different knowledge graphs (KGs) that refer to the same object in the real world. Recent studies incorporate temporal information to augment the representations of KGs. The existing methods for EA between temporal KGs (TKGs) utilize a time-aware attention mechanism to incorporate relational and temporal information into entity embeddings. The approaches outperform the previous methods by using temporal information. However, we believe that it is not necessary to learn the embeddings of temporal information in KGs since most TKGs have uniform temporal representations. Therefore, we propose a simple graph neural network (GNN) model combined with a temporal information matching mechanism, which achieves better performance with less time and fewer parameters. Furthermore, since alignment seeds are difficult to label in real-world applications, we also propose a method to generate unsupervised alignment seeds via the temporal information of TKG. Extensive experiments on public datasets indicate that our supervised method significantly outperforms the previous methods and the unsupervised one has competitive performance. | ['Man Lan', 'Jianchao Zhu', 'Hao Yuan', 'Meirong Ma', 'Xin Mao', 'Li Cai'] | 2022-09-20 | null | https://aclanthology.org/2022.coling-1.181 | https://aclanthology.org/2022.coling-1.181.pdf | coling-2022-10 | ['entity-alignment', 'entity-embeddings', 'entity-alignment'] | ['knowledge-base', 'methodology', 'natural-language-processing'] | [-3.23297948e-01 1.73036326e-02 -6.42263710e-01 -2.97543705e-01
-7.94373974e-02 -3.84404510e-01 6.28517151e-01 5.66184342e-01
-4.14036423e-01 4.09081995e-01 2.30047688e-01 -2.05686659e-01
-3.41526538e-01 -1.16259170e+00 -7.62625933e-01 -4.23906475e-01
-2.88510650e-01 5.16151547e-01 5.46483457e-01 -2.31816322e-01
1.24630488e-01 3.92010033e-01 -1.02247214e+00 -2.10049972e-01
1.17223847e+00 7.89000630e-01 2.24880502e-01 1.36991993e-01
-3.03785354e-01 8.56346309e-01 -3.86251926e-01 -6.78250015e-01
2.91940987e-01 -1.86290875e-01 -1.00903416e+00 -3.50558907e-01
1.84415832e-01 -9.68616754e-02 -8.31672192e-01 1.00134385e+00
3.41583997e-01 4.51179922e-01 5.58490098e-01 -1.66209543e+00
-1.28583610e+00 8.35360706e-01 -6.44963562e-01 2.99827456e-01
3.08861017e-01 -3.02435189e-01 1.13727772e+00 -6.24266088e-01
7.17927754e-01 8.53611112e-01 7.90024340e-01 6.89765513e-02
-8.18579376e-01 -4.60258931e-01 5.31519711e-01 7.56080449e-01
-1.59270394e+00 -4.57057916e-02 8.82430494e-01 -2.14481995e-01
1.22374880e+00 -1.05150834e-01 7.34446824e-01 8.54684532e-01
1.01746187e-01 6.47403538e-01 4.70803261e-01 -5.57338893e-01
-2.93612499e-02 -2.00024262e-01 2.33032346e-01 8.15099061e-01
4.23521817e-01 -1.57802090e-01 -5.25425851e-01 -8.93974304e-02
7.33900070e-01 1.96603179e-01 -8.18607435e-02 -6.09833956e-01
-1.43044448e+00 7.12692320e-01 9.30292964e-01 5.57951868e-01
-4.87040132e-01 3.86643946e-01 4.36220348e-01 3.24934572e-01
3.99262905e-01 4.55797315e-01 -5.33931196e-01 2.66413003e-01
-5.57270586e-01 5.84382117e-02 4.89399016e-01 1.37393951e+00
7.48868108e-01 -8.57760310e-02 -1.55000925e-01 5.40518522e-01
3.73109072e-01 1.33310929e-01 6.36872232e-01 -3.93075585e-01
7.46042430e-01 1.06766844e+00 2.13903651e-01 -1.50616360e+00
-4.08115178e-01 -3.04498523e-01 -6.57974184e-01 -6.06399715e-01
1.71552882e-01 2.23926485e-01 -1.10386086e+00 1.77347243e+00
4.05180901e-01 7.30164170e-01 1.56103507e-01 6.08638704e-01
9.22733128e-01 6.07537329e-01 1.53927013e-01 -1.92986727e-01
1.15760934e+00 -1.30861282e+00 -9.92421746e-01 -2.03493372e-01
1.06492424e+00 -4.47665155e-01 8.85603130e-01 -2.86763012e-01
-6.79052234e-01 -4.71777588e-01 -1.18228447e+00 -1.31485596e-01
-9.55273390e-01 1.41269729e-01 1.00685441e+00 2.85661906e-01
-9.64107513e-01 8.63115132e-01 -9.92238045e-01 -7.93018401e-01
2.90706977e-02 3.96240145e-01 -4.91669774e-01 5.65720834e-02
-1.74452388e+00 1.01070631e+00 9.92013574e-01 2.45970577e-01
-1.89498350e-01 -3.72536272e-01 -1.18232882e+00 -9.47154015e-02
5.18742442e-01 -7.07465470e-01 8.71412873e-01 -6.21769786e-01
-1.12666190e+00 6.72496915e-01 -1.95697695e-01 -6.50120437e-01
1.02056161e-01 -2.10135102e-01 -1.01171792e+00 7.44394809e-02
3.06358695e-01 5.85376859e-01 4.46042657e-01 -9.04306591e-01
-4.69014674e-01 -1.05541989e-01 2.84607440e-01 3.26115251e-01
-5.61347485e-01 -1.33041754e-01 -9.36160386e-01 -7.29516149e-01
3.81470412e-01 -8.85366082e-01 -4.69669729e-01 -3.48023057e-01
-4.73710656e-01 -5.41615129e-01 6.49191260e-01 -7.10483611e-01
1.72981250e+00 -2.07097578e+00 2.00943694e-01 2.46423200e-01
1.65769935e-01 1.94266975e-01 -1.87109023e-01 7.64734805e-01
-2.12346032e-01 2.20003366e-01 1.53614908e-01 -6.79166093e-02
8.45864713e-02 3.19027126e-01 -1.51145920e-01 2.49231532e-01
2.27026492e-02 1.37327826e+00 -1.22058427e+00 -5.78465641e-01
-7.21627176e-02 -3.34510915e-02 -2.85326421e-01 6.65436462e-02
-2.07692206e-01 3.91769819e-02 -5.23563623e-01 5.81483006e-01
4.11632717e-01 -6.72319114e-01 4.36496705e-01 -5.76258957e-01
9.85191092e-02 4.17707980e-01 -1.13413668e+00 1.93907654e+00
-2.82368045e-02 1.87001705e-01 -8.88329625e-01 -1.05033946e+00
8.75039935e-01 3.99307787e-01 5.88908017e-01 -7.99865365e-01
-3.31308134e-02 2.69356996e-01 -8.47311839e-02 -4.27381605e-01
9.84234095e-01 4.25559729e-01 -2.13547394e-01 3.25630099e-01
1.23586819e-01 4.42746162e-01 4.07999337e-01 5.99463522e-01
1.32762837e+00 1.63850665e-01 4.54460472e-01 1.18902199e-01
2.71503955e-01 7.45949149e-02 9.14176524e-01 5.48034072e-01
-2.74165511e-01 2.50344127e-01 2.24121138e-01 -5.33372164e-01
-8.94639969e-01 -9.07027066e-01 3.52627575e-01 8.92583907e-01
7.34905362e-01 -8.23436856e-01 -2.61539668e-01 -1.10201263e+00
1.90002814e-01 4.37603712e-01 -6.93156064e-01 -4.56143200e-01
-6.99731171e-01 -5.55195510e-01 4.36259210e-01 1.06148124e+00
4.21885133e-01 -9.40912247e-01 6.76433146e-02 4.40972596e-01
-3.27251494e-01 -1.29599345e+00 -7.56119609e-01 8.03140700e-02
-8.61740947e-01 -9.74201441e-01 -3.33424062e-01 -1.01689911e+00
9.26406026e-01 4.02042449e-01 9.34147298e-01 4.66913879e-02
-6.80339336e-02 3.77612740e-01 -6.96902394e-01 -1.33043498e-01
5.43046445e-02 3.28295320e-01 2.86753327e-01 -2.44320169e-01
6.42763793e-01 -6.66341245e-01 -5.27606785e-01 3.98882598e-01
-7.88908899e-01 5.32568730e-02 5.78242421e-01 7.03319430e-01
5.31458378e-01 2.13782787e-01 7.50635684e-01 -8.94153297e-01
5.41565299e-01 -6.65453732e-01 -5.01988173e-01 6.59842789e-01
-9.94017720e-01 2.46072128e-01 5.10617673e-01 -6.61564469e-01
-8.33269417e-01 -2.67229855e-01 4.06276643e-01 -7.58911967e-01
2.85687715e-01 9.94188607e-01 -6.23570830e-02 -1.69900320e-02
4.78518367e-01 6.97570890e-02 -5.27704954e-01 -4.06277716e-01
7.36747146e-01 2.15987474e-01 5.76974750e-01 -5.78411996e-01
1.04809856e+00 2.40710437e-01 -1.51157379e-01 4.59558377e-03
-7.39084601e-01 -6.25532985e-01 -8.72098982e-01 1.54279266e-02
6.58605933e-01 -9.24696147e-01 -3.80118251e-01 3.15312177e-01
-1.10405421e+00 -7.77027756e-02 -1.86006784e-01 6.86608434e-01
-3.25928330e-01 5.61117053e-01 -6.12013578e-01 -3.63744617e-01
-2.45067194e-01 -6.07587039e-01 6.49445355e-01 4.45972502e-01
-1.89117700e-01 -1.08921182e+00 5.06048501e-02 6.28179312e-02
3.56358528e-01 2.51355171e-01 9.36743498e-01 -9.45882857e-01
-8.26480389e-01 -1.62668630e-01 -2.90820271e-01 -1.71636850e-01
6.04668736e-01 6.57001808e-02 -3.74936372e-01 -2.85520941e-01
-5.23561180e-01 -1.03350608e-02 7.37063646e-01 2.71880567e-01
8.92182171e-01 -4.64275539e-01 -8.41151118e-01 5.17441928e-01
1.51894081e+00 4.83646184e-01 6.36721611e-01 6.46543682e-01
1.05956972e+00 3.76944333e-01 1.08288324e+00 1.45534396e-01
7.70571649e-01 7.20889091e-01 2.55091906e-01 -2.62348596e-02
5.10078073e-02 -5.11955261e-01 3.25902790e-01 1.40028739e+00
-1.98800296e-01 -4.87511545e-01 -9.08708692e-01 1.18450212e+00
-2.35725522e+00 -9.68579173e-01 -1.45809159e-01 2.00659084e+00
8.93424094e-01 7.36793056e-02 -1.53237239e-01 -1.28467560e-01
9.21749532e-01 3.22914541e-01 -4.43052769e-01 -5.43668978e-02
-1.39680961e-02 -2.22560298e-02 6.88643396e-01 9.07494724e-02
-1.27875137e+00 1.21881318e+00 6.04618406e+00 6.82189047e-01
-8.05756807e-01 -6.23676069e-02 6.30404428e-02 2.62432426e-01
-3.76183391e-01 2.41665915e-01 -7.57424414e-01 5.01331925e-01
1.01816499e+00 -6.79819107e-01 1.02617539e-01 8.07296813e-01
-1.02303252e-01 2.11699679e-01 -1.06793010e+00 7.93552697e-01
-4.68778498e-02 -1.29281676e+00 -4.00406532e-02 -3.08576167e-01
7.88023829e-01 -2.11065896e-02 -2.14426771e-01 3.65845859e-01
7.01022804e-01 -7.08964765e-01 3.32532108e-01 6.41709745e-01
5.05882859e-01 -7.10533917e-01 8.95851493e-01 -7.20783696e-02
-1.88886261e+00 1.69104233e-01 -4.70875084e-01 2.76547521e-01
3.10082436e-01 3.93887281e-01 -1.04937458e+00 1.50991476e+00
6.94410086e-01 1.16329229e+00 -8.37007284e-01 1.23603439e+00
-3.78397942e-01 2.90786833e-01 -2.10488051e-01 1.05813622e-01
2.06979141e-01 -3.44796747e-01 3.12057257e-01 9.10405338e-01
4.15016174e-01 -7.39466697e-02 3.03562939e-01 4.97575760e-01
-2.55236119e-01 2.74488926e-01 -8.32477152e-01 -5.66425085e-01
9.48755980e-01 1.14335275e+00 -8.81477594e-01 -1.94701508e-01
-5.59365809e-01 1.19891870e+00 5.98355412e-01 3.86763543e-01
-1.11869609e+00 -6.25578225e-01 2.70579964e-01 -1.95521802e-01
4.05892640e-01 -3.38034630e-01 2.46357530e-01 -1.13258123e+00
2.48836115e-01 -4.43156332e-01 9.44335341e-01 -8.42441440e-01
-1.53814936e+00 5.25954008e-01 1.62127182e-01 -1.36410058e+00
-1.50147334e-01 -2.02128351e-01 -4.90080267e-01 5.65151930e-01
-1.43173385e+00 -1.42132688e+00 -2.70429701e-01 6.03697002e-01
-6.49852380e-02 -6.05234616e-02 6.69670165e-01 5.20819008e-01
-6.62474096e-01 6.21993005e-01 -1.64902268e-03 5.19184649e-01
9.34563696e-01 -1.33226776e+00 7.61830568e-01 1.12577307e+00
4.30605620e-01 1.03392947e+00 4.42704946e-01 -1.01322091e+00
-1.21396005e+00 -1.45713997e+00 1.27198696e+00 -2.60719657e-01
9.26040888e-01 6.95217550e-02 -1.10963345e+00 1.20794916e+00
4.05095577e-01 1.68374643e-01 6.17921948e-01 3.19405317e-01
-5.18195331e-01 -2.42125928e-01 -8.63718033e-01 7.00121760e-01
1.51904905e+00 -6.00629985e-01 -9.06231046e-01 3.67827535e-01
1.27554381e+00 -4.55937088e-01 -1.21289647e+00 7.08772600e-01
8.60577077e-02 -2.72611409e-01 8.75620544e-01 -8.49102736e-01
3.62164788e-02 -9.28680122e-01 5.33832721e-02 -1.27861369e+00
-4.78443116e-01 -3.80221426e-01 -5.44931829e-01 1.50973332e+00
2.91100413e-01 -8.18207145e-01 7.21117318e-01 4.80287015e-01
-1.71388954e-01 -6.48120940e-01 -6.52188301e-01 -1.23745012e+00
-3.71398240e-01 3.39693911e-02 9.10154939e-01 1.58721924e+00
1.16850391e-01 2.05649078e-01 -3.45831811e-01 6.68297589e-01
2.76126146e-01 4.28700209e-01 5.53700626e-01 -1.28683209e+00
6.87539428e-02 -1.66947782e-01 -7.64833212e-01 -9.47861731e-01
2.32104361e-01 -1.04221845e+00 -1.84050903e-01 -1.93477726e+00
1.47007644e-01 -6.14463508e-01 -7.32540786e-01 7.82475710e-01
-3.93381685e-01 -1.87270194e-01 -3.09743583e-01 2.74502039e-01
-9.26539004e-01 7.51411378e-01 8.71276677e-01 -2.78198808e-01
-2.04013467e-01 -4.11446631e-01 -5.75322866e-01 4.60096598e-01
7.05740869e-01 -6.00006938e-01 -8.06423426e-01 -5.00423729e-01
4.64687794e-01 -2.24706501e-01 -1.45548610e-02 -8.20021391e-01
8.79801452e-01 -3.55336279e-01 9.45948437e-02 -9.52627301e-01
1.13372160e-02 -7.41753221e-01 5.46270669e-01 1.86196461e-01
1.62655767e-02 4.66872424e-01 8.76856968e-02 9.98716831e-01
-5.00589490e-01 -1.73720747e-01 3.11372668e-01 1.49080679e-01
-1.18107533e+00 7.30334520e-01 1.21743947e-01 -1.09994099e-01
1.17476177e+00 -2.15144843e-01 -4.98669058e-01 -1.56280190e-01
-6.70763433e-01 6.08440816e-01 5.09405971e-01 7.21214771e-01
4.70590860e-01 -1.89650202e+00 -1.16032332e-01 -2.50752509e-01
4.86420423e-01 1.60474658e-01 1.08229093e-01 8.86526823e-01
-4.24185902e-01 3.49782616e-01 -2.14950360e-05 -2.88982421e-01
-1.05436230e+00 9.51639950e-01 1.17782861e-01 -5.98307788e-01
-5.98979414e-01 7.62045622e-01 -1.35850847e-01 -5.01497388e-01
2.25787833e-01 -2.87500799e-01 -4.70906764e-01 1.58212274e-01
8.66727605e-02 6.74918368e-02 8.72760788e-02 -4.59215522e-01
-4.83131021e-01 4.23619330e-01 -5.71550488e-01 2.00873107e-01
1.34996343e+00 -7.63059705e-02 -1.90430805e-01 4.85090643e-01
1.00761127e+00 9.29675475e-02 -5.64741015e-01 -7.69705474e-01
6.32091999e-01 -4.94635582e-01 -3.59886050e-01 -3.71585816e-01
-1.16734231e+00 2.38173112e-01 2.09299326e-01 2.66302675e-01
1.08101845e+00 -4.84127030e-02 8.56800914e-01 5.17533779e-01
5.70297599e-01 -1.20278788e+00 2.53222287e-01 5.21922350e-01
5.22668123e-01 -1.04520333e+00 1.00741901e-01 -5.54473400e-01
-6.49742007e-01 9.81370032e-01 1.00862598e+00 1.71548110e-02
6.14396453e-01 -2.12757781e-01 -1.41647622e-01 -3.80958140e-01
-7.45361030e-01 -3.55984181e-01 3.85910064e-01 5.34435928e-01
1.33335933e-01 -1.73135355e-01 -5.82059562e-01 6.28552139e-01
9.37382355e-02 -1.47895038e-01 3.38882536e-01 1.05066991e+00
-1.05340786e-01 -1.56056166e+00 3.05526137e-01 3.84514004e-01
-2.60300964e-01 -2.11292267e-01 -3.21953923e-01 8.94694090e-01
9.16535128e-03 6.51784718e-01 1.32587773e-03 -5.21876752e-01
2.58889526e-01 1.52679205e-01 3.72877955e-01 -6.21389270e-01
-2.86105514e-01 -3.09203237e-01 1.81422830e-01 -5.85463166e-01
-6.96354032e-01 -4.36006218e-01 -1.47956395e+00 -4.80438828e-01
-7.42610276e-01 4.04800713e-01 1.46418780e-01 8.67525458e-01
7.04024971e-01 5.97981811e-01 6.70456946e-01 -2.34994620e-01
8.37394595e-02 -8.52329850e-01 -7.03877091e-01 5.71102262e-01
-2.12608472e-01 -8.86925101e-01 -9.15019214e-02 3.91768403e-02] | [8.647992134094238, 7.85780143737793] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.