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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
8af71914-6910-4ef4-aa6e-588192228c66 | large-scale-unsupervised-person-re | 2105.07914 | null | https://arxiv.org/abs/2105.07914v1 | https://arxiv.org/pdf/2105.07914v1.pdf | Large-Scale Unsupervised Person Re-Identification with Contrastive Learning | Existing public person Re-Identification~(ReID) datasets are small in modern terms because of labeling difficulty. Although unlabeled surveillance video is abundant and relatively easy to obtain, it is unclear how to leverage these footage to learn meaningful ReID representations. In particular, most existing unsupervised and domain adaptation ReID methods utilize only the public datasets in their experiments, with labels removed. In addition, due to small data sizes, these methods usually rely on fine tuning by the unlabeled training data in the testing domain to achieve good performance. Inspired by the recent progress of large-scale self-supervised image classification using contrastive learning, we propose to learn ReID representation from large-scale unlabeled surveillance video alone. Assisted by off-the-shelf pedestrian detection tools, we apply the contrastive loss at both the image and the tracklet levels. Together with a principal component analysis step using camera labels freely available, our evaluation using a large-scale unlabeled dataset shows far superior performance among unsupervised methods that do not use any training data in the testing domain. Furthermore, the accuracy improves with the data size and therefore our method has great potential with even larger and more diversified datasets. | ['Yin Wang', 'Ming Feng', 'Xinbo Zhao', 'Qiuyu Ren', 'Yan Bai', 'Weiquan Huang'] | 2021-05-17 | null | null | null | null | ['self-supervised-image-classification', 'unsupervised-person-re-identification'] | ['computer-vision', 'computer-vision'] | [ 7.28414133e-02 -3.41902584e-01 -3.51820320e-01 -6.06987178e-01
-8.37939382e-01 -8.24057996e-01 5.25033712e-01 -1.11766763e-01
-7.44057417e-01 9.50170457e-01 2.07450584e-01 -4.06976528e-02
2.52163887e-01 -4.17306900e-01 -6.83757722e-01 -6.37896657e-01
1.48514092e-01 4.92186099e-01 2.16491759e-01 4.76296544e-02
-1.36780426e-01 3.60523969e-01 -1.67823696e+00 -1.94673412e-04
6.47126973e-01 6.77823126e-01 -1.66995153e-01 5.91543853e-01
1.52073190e-01 7.77089596e-01 -4.87555832e-01 -7.38059461e-01
7.05521226e-01 -3.57737184e-01 -5.94647646e-01 3.40916127e-01
1.11969757e+00 -5.81686676e-01 -6.47875249e-01 1.12556791e+00
6.75911188e-01 2.09470153e-01 5.95268130e-01 -1.33745372e+00
-8.06346595e-01 1.95335105e-01 -8.01348746e-01 4.14720833e-01
4.43854690e-01 1.74600914e-01 6.21510684e-01 -6.73205197e-01
8.14600825e-01 1.01035535e+00 9.61769283e-01 7.18474746e-01
-1.40205681e+00 -1.08694780e+00 3.26212555e-01 2.20695436e-01
-1.58244050e+00 -7.45731592e-01 8.16924512e-01 -5.22597313e-01
5.88474095e-01 8.26783478e-02 3.20183277e-01 1.47303724e+00
-4.10837352e-01 8.26679528e-01 1.16006541e+00 -5.48189402e-01
9.71371233e-02 4.57852066e-01 4.41273600e-01 6.34527147e-01
5.93786359e-01 2.89490014e-01 -4.35266167e-01 -1.50846541e-01
4.86914486e-01 5.18203713e-02 -1.80803776e-01 -6.29828870e-01
-1.15057874e+00 5.98241627e-01 3.54909331e-01 -1.18087701e-01
-3.25221494e-02 -1.54764384e-01 5.03375947e-01 4.55449551e-01
4.82362896e-01 1.77558288e-01 -3.27342451e-01 5.81570156e-02
-9.81712282e-01 9.45876166e-02 6.66184843e-01 1.32010221e+00
8.34612548e-01 -1.43253446e-01 1.55642197e-01 8.36784244e-01
-1.99368168e-02 6.74432218e-01 3.04505020e-01 -8.72570753e-01
5.20794570e-01 4.99238789e-01 3.05068046e-01 -7.38784730e-01
-3.65468621e-01 -3.22458506e-01 -6.71299100e-01 2.01438606e-01
8.53403151e-01 -3.98674607e-01 -8.93200338e-01 1.91603470e+00
3.98993284e-01 2.91556120e-01 3.87402065e-02 8.48457813e-01
6.97876036e-01 1.59175590e-01 3.55253339e-01 -9.94740613e-03
1.40945673e+00 -8.70803237e-01 -3.61667722e-01 -1.91391200e-01
5.99814475e-01 -6.00793064e-01 5.97520053e-01 1.93370342e-01
-6.76952362e-01 -6.90911949e-01 -1.04842794e+00 6.08138554e-02
-5.63600063e-01 3.89013290e-01 5.59104204e-01 1.09654629e+00
-1.17382252e+00 1.43420234e-01 -5.65368533e-01 -1.04570425e+00
5.50681770e-01 5.96626461e-01 -9.32999253e-01 -5.86828649e-01
-8.08033347e-01 7.26099670e-01 2.39354700e-01 -1.50032565e-01
-6.79442286e-01 -5.05257666e-01 -9.50246990e-01 -2.43902683e-01
3.13941717e-01 -8.41907263e-01 1.09786367e+00 -1.22030807e+00
-1.07395196e+00 1.11328471e+00 -1.72582597e-01 -3.54558587e-01
7.86546767e-01 -4.16414514e-02 -4.11667228e-01 2.73808569e-01
3.40555221e-01 1.03331184e+00 8.51208270e-01 -1.33283532e+00
-8.88114214e-01 -4.54855919e-01 2.20582783e-01 9.74306315e-02
-6.04729176e-01 2.17405811e-01 -5.50076067e-01 -6.66072726e-01
-3.36991608e-01 -1.19252050e+00 -1.34907141e-01 5.09346426e-02
-2.43035987e-01 -7.19270185e-02 8.59928131e-01 -6.53661132e-01
5.96394479e-01 -2.25308847e+00 -2.00335443e-01 6.23966716e-02
2.40399793e-01 2.79869437e-01 -3.39954913e-01 1.61795780e-01
-2.27157071e-01 -1.96744680e-01 9.96191502e-02 -5.22822618e-01
-1.40927732e-01 -2.44414136e-02 5.04321791e-02 8.28469634e-01
6.71523884e-02 5.90186119e-01 -1.07278407e+00 -7.03554749e-01
1.60748124e-01 2.06001341e-01 -5.16862333e-01 1.89174011e-01
3.70297194e-01 5.91232479e-01 -4.17753905e-01 7.13036299e-01
7.28629589e-01 -2.02546582e-01 6.57870770e-02 -2.49848142e-01
-1.29090235e-01 -2.81104684e-01 -1.16569269e+00 1.71031177e+00
8.58204812e-02 7.50465453e-01 1.04967415e-01 -1.14826787e+00
6.07340097e-01 1.74432099e-01 5.81709862e-01 -4.65559721e-01
1.17661975e-01 -4.61659245e-02 -2.54166454e-01 -3.81706655e-01
5.70222437e-01 2.65138783e-02 5.10074524e-03 4.70811397e-01
1.86894536e-01 6.11090600e-01 4.11975294e-01 3.38322163e-01
1.16289401e+00 1.78870559e-01 1.71499312e-01 -2.07952663e-01
4.03393120e-01 4.12724286e-01 6.82915092e-01 9.43380773e-01
-6.45173550e-01 7.32365251e-01 1.02139063e-01 -5.31284630e-01
-1.20640135e+00 -1.02836120e+00 -2.09365249e-01 1.34712350e+00
3.30685288e-01 7.26098707e-03 -9.12978947e-01 -9.89036262e-01
1.96612682e-02 2.42350653e-01 -6.69552028e-01 1.03292920e-01
-6.00314736e-01 -9.99284089e-01 6.97908819e-01 8.01821530e-01
6.79782271e-01 -5.81263185e-01 -7.10628256e-02 1.20027110e-01
-8.56217742e-02 -1.35603797e+00 -7.12700248e-01 -2.07673430e-01
-6.26291096e-01 -1.40923548e+00 -1.03765893e+00 -1.06265569e+00
1.10983002e+00 8.00888658e-01 8.93825114e-01 -9.61764753e-02
-4.35597092e-01 9.50722516e-01 -4.74848002e-01 -3.80975515e-01
-1.44399166e-01 -2.01451108e-02 4.24928159e-01 -2.90100239e-02
7.24522114e-01 -1.99282348e-01 -5.77486932e-01 6.09696805e-01
-4.60421622e-01 -1.55205891e-01 4.46695298e-01 9.07135963e-01
3.07350039e-01 -4.95167188e-02 7.38213956e-01 -7.47346878e-01
1.47069484e-01 -4.39075589e-01 -6.71723604e-01 4.13895637e-01
-3.35413009e-01 -2.36136615e-01 4.02688026e-01 -6.20476902e-01
-1.25832629e+00 2.09039360e-01 2.43047446e-01 -3.22349131e-01
-4.68100160e-01 -4.31670807e-02 -2.00236306e-01 -2.47497916e-01
7.96491086e-01 5.36257923e-02 1.99459996e-02 -5.02810001e-01
2.93184072e-01 7.27902353e-01 8.09306562e-01 -5.35046339e-01
1.03591108e+00 5.86446404e-01 -3.43158394e-01 -7.58977532e-01
-6.98281169e-01 -8.73660505e-01 -8.84721518e-01 -2.98106790e-01
8.76355767e-01 -1.31152296e+00 -6.10445559e-01 4.20031518e-01
-7.96443284e-01 2.00865325e-02 -2.50782371e-01 7.67245948e-01
-2.82031119e-01 6.71822965e-01 -4.79470611e-01 -8.11386585e-01
-2.70486534e-01 -1.03080034e+00 9.84504044e-01 3.13136250e-01
-3.45336013e-02 -7.00126171e-01 -7.05612227e-02 6.05359077e-01
2.34734073e-01 -2.87271990e-03 4.74590629e-01 -8.66926849e-01
-6.61014259e-01 -6.93806231e-01 -5.44674814e-01 3.59674037e-01
1.98345691e-01 -2.83816576e-01 -1.17198610e+00 -6.51870608e-01
-4.25424486e-01 -4.31227058e-01 8.18672597e-01 2.52631009e-01
8.67954493e-01 1.02857985e-02 -5.22437692e-01 5.34482896e-01
1.13264203e+00 -4.76161316e-02 3.01734954e-01 4.95921105e-01
8.41193378e-01 7.28576899e-01 4.18780148e-01 3.23285580e-01
6.71122789e-01 6.97035730e-01 -1.16981417e-01 -1.70273691e-01
-4.36781019e-01 -2.59387016e-01 4.90500569e-01 4.08921719e-01
-2.53869385e-01 -1.80998340e-01 -8.20362985e-01 4.85472500e-01
-1.82119679e+00 -1.30055618e+00 2.19298452e-01 2.47801828e+00
5.04092455e-01 -7.64580593e-02 5.22392392e-01 -1.61483318e-01
1.11260831e+00 -1.38426319e-01 -6.43268228e-01 3.10154915e-01
-2.51237273e-01 -3.25818270e-01 9.84141827e-01 5.81944138e-02
-1.60324132e+00 8.10012877e-01 6.66172218e+00 4.77255225e-01
-8.51438046e-01 1.33137301e-01 5.40669858e-01 -2.04408228e-01
2.93548495e-01 -3.09653014e-01 -1.06574345e+00 4.43516463e-01
9.52920735e-01 -8.38674903e-02 2.65481353e-01 1.04575527e+00
-5.37668169e-02 -1.75664555e-02 -1.28103411e+00 1.27644157e+00
3.46612900e-01 -1.13370633e+00 -2.62101084e-01 4.67982143e-02
7.88338721e-01 2.25280955e-01 -3.56131643e-02 4.17788714e-01
5.38756132e-01 -5.05748212e-01 6.07123792e-01 2.35631272e-01
9.40745890e-01 -6.64457619e-01 8.51066053e-01 2.08253458e-01
-1.21880579e+00 -2.01991856e-01 -5.57720363e-01 1.70230538e-01
-3.93341854e-02 -6.01139478e-02 -6.47401094e-01 4.01118368e-01
8.01425219e-01 8.38514030e-01 -8.39110255e-01 1.33872855e+00
3.94642986e-02 4.30239201e-01 -3.76382530e-01 3.28074723e-01
-3.92148457e-03 -4.82285917e-02 3.76724958e-01 1.33171606e+00
1.28415748e-01 1.81079462e-01 6.47420108e-01 2.92438120e-01
-4.26560014e-01 -7.66361952e-02 -7.41141856e-01 1.87754944e-01
3.46776187e-01 1.09459233e+00 -6.69204772e-01 -4.56968486e-01
-9.04434562e-01 9.10451591e-01 2.72159696e-01 6.65241599e-01
-6.19564950e-01 -1.41625097e-02 8.42173278e-01 1.02147996e-01
2.75176316e-01 -1.80308655e-01 5.86596392e-02 -1.55324876e+00
-1.28049597e-01 -7.47014523e-01 6.57416165e-01 -4.16231483e-01
-1.83130360e+00 4.76296365e-01 2.67646104e-01 -1.48267484e+00
-3.86052787e-01 -6.55843496e-01 -3.51868272e-01 7.22013414e-01
-1.65259135e+00 -1.48897624e+00 -4.31726396e-01 8.72518539e-01
4.98199850e-01 -4.69568104e-01 7.78756559e-01 6.15656495e-01
-9.16303575e-01 1.16471994e+00 3.73458743e-01 6.72902286e-01
1.21785414e+00 -1.01484203e+00 3.28792930e-01 1.12310946e+00
5.21377428e-03 6.83423579e-01 5.47469139e-01 -6.62113011e-01
-1.52464318e+00 -1.08449543e+00 4.85155851e-01 -8.51145685e-01
5.97682536e-01 -5.06719470e-01 -6.71260834e-01 8.99866998e-01
1.26883024e-02 4.46926087e-01 1.03090441e+00 2.40533963e-01
-6.96018517e-01 -2.01942042e-01 -1.44876361e+00 3.17637533e-01
1.30532134e+00 -5.13720870e-01 -5.44505835e-01 3.31783980e-01
2.90785015e-01 -3.05829421e-02 -7.91414440e-01 2.22472250e-01
6.34850979e-01 -6.47661924e-01 1.06011808e+00 -7.78601170e-01
-2.02906087e-01 -3.93680632e-01 -1.30901143e-01 -8.46947610e-01
-3.19346875e-01 -1.64409056e-01 3.10743630e-01 1.59475827e+00
1.98299721e-01 -6.78726017e-01 1.21295404e+00 1.12140167e+00
8.23950022e-02 1.73651129e-01 -8.00757885e-01 -1.02948022e+00
-1.28248483e-01 -1.28381819e-01 3.53475541e-01 1.04489589e+00
-8.26318190e-02 1.23932093e-01 -5.67891836e-01 4.13409412e-01
1.29302621e+00 2.78058946e-02 1.18638551e+00 -1.39275849e+00
-2.19064150e-02 -6.61444515e-02 -8.12614918e-01 -9.53206599e-01
4.42648977e-01 -6.57468438e-01 6.32544085e-02 -1.06145048e+00
7.03526497e-01 -6.41611218e-01 -5.25700569e-01 4.68869239e-01
-2.81687051e-01 5.60304463e-01 2.72078305e-01 5.18539548e-01
-9.14091170e-01 2.57325053e-01 6.56617939e-01 -4.81513619e-01
-3.57553475e-02 -1.70920700e-01 -6.47166312e-01 7.21115351e-01
5.80697775e-01 -4.59476709e-01 -4.01877880e-01 -4.08145398e-01
-5.29957116e-01 -3.45124871e-01 4.41816807e-01 -1.08098960e+00
4.64911491e-01 2.01052547e-01 8.42908323e-01 -3.41829777e-01
3.00207138e-01 -8.91499996e-01 6.66529387e-02 7.64073059e-02
-2.52311051e-01 4.67197895e-02 1.61124453e-01 8.00742626e-01
-4.76251245e-02 -1.46028325e-01 8.71074796e-01 -6.37901574e-02
-1.12558591e+00 4.17506337e-01 -2.52671778e-01 3.21617379e-04
1.14016366e+00 -5.44698119e-01 -5.91622353e-01 -2.45932892e-01
-4.54833001e-01 1.35338277e-01 9.76118207e-01 5.15089333e-01
2.85147667e-01 -1.11897981e+00 -7.01554537e-01 2.92218864e-01
5.67200124e-01 -4.37852204e-01 2.46854737e-01 6.57271087e-01
-1.52830064e-01 2.65265405e-01 -4.07438308e-01 -5.21267295e-01
-1.38663256e+00 9.16881859e-01 5.48314750e-02 1.81363583e-01
-7.16821909e-01 6.93919897e-01 3.22513610e-01 -2.98396289e-01
3.63902718e-01 2.71365345e-01 -2.30142564e-01 1.86255515e-01
9.18324769e-01 4.12316620e-01 -1.66893482e-01 -9.82263923e-01
-4.28899556e-01 4.93911743e-01 -3.42456639e-01 1.20163046e-01
1.18054926e+00 -2.91310072e-01 3.31607044e-01 -2.15489224e-01
1.21548223e+00 -2.90263612e-02 -1.50198507e+00 -3.80557835e-01
5.87951615e-02 -6.16753280e-01 -2.35088393e-01 -4.81090784e-01
-9.89212751e-01 4.22891647e-01 9.13253367e-01 -2.69886822e-01
9.97077703e-01 -2.52361894e-02 6.10894322e-01 5.91637671e-01
8.47150147e-01 -1.00763893e+00 -1.65519983e-01 8.03872943e-02
2.27157637e-01 -1.82348680e+00 1.61590531e-01 -3.11647743e-01
-6.30696893e-01 7.66125560e-01 6.21268213e-01 1.16884839e-02
4.18280989e-01 5.94867095e-02 2.81431407e-01 2.54956871e-01
-1.58125401e-01 -3.65962446e-01 1.37326553e-01 1.03618896e+00
2.49279901e-01 -1.09290041e-01 6.70127422e-02 3.94962460e-01
6.58751875e-02 4.61678803e-02 5.70312977e-01 9.79200065e-01
-2.01673672e-01 -1.37177861e+00 -6.08846366e-01 4.39509869e-01
-3.85771215e-01 1.77318841e-01 -2.48179674e-01 9.09078896e-01
8.84902328e-02 1.26145029e+00 -2.58418638e-02 -3.35182011e-01
3.73888850e-01 -6.09622225e-02 4.91532296e-01 -5.88426054e-01
-3.91340613e-01 -3.05477262e-01 2.07204476e-01 -4.20236111e-01
-8.45223188e-01 -9.63339806e-01 -6.40457571e-01 -5.60723245e-01
-2.69907624e-01 5.65811172e-02 4.67572868e-01 8.26301396e-01
3.12724113e-01 -1.73677891e-01 4.28170800e-01 -1.06463408e+00
-4.20599729e-01 -7.15147913e-01 -5.42784631e-01 6.63084745e-01
3.44224870e-01 -8.20685863e-01 -1.13072224e-01 5.09284854e-01] | [14.72867488861084, 0.9807615876197815] |
bbdec1c7-59ac-4d2e-8866-e2078b605ffa | vdtr-video-deblurring-with-transformer | 2204.08023 | null | https://arxiv.org/abs/2204.08023v1 | https://arxiv.org/pdf/2204.08023v1.pdf | VDTR: Video Deblurring with Transformer | Video deblurring is still an unsolved problem due to the challenging spatio-temporal modeling process. While existing convolutional neural network-based methods show a limited capacity for effective spatial and temporal modeling for video deblurring. This paper presents VDTR, an effective Transformer-based model that makes the first attempt to adapt Transformer for video deblurring. VDTR exploits the superior long-range and relation modeling capabilities of Transformer for both spatial and temporal modeling. However, it is challenging to design an appropriate Transformer-based model for video deblurring due to the complicated non-uniform blurs, misalignment across multiple frames and the high computational costs for high-resolution spatial modeling. To address these problems, VDTR advocates performing attention within non-overlapping windows and exploiting the hierarchical structure for long-range dependencies modeling. For frame-level spatial modeling, we propose an encoder-decoder Transformer that utilizes multi-scale features for deblurring. For multi-frame temporal modeling, we adapt Transformer to fuse multiple spatial features efficiently. Compared with CNN-based methods, the proposed method achieves highly competitive results on both synthetic and real-world video deblurring benchmarks, including DVD, GOPRO, REDS and BSD. We hope such a Transformer-based architecture can serve as a powerful alternative baseline for video deblurring and other video restoration tasks. The source code will be available at \url{https://github.com/ljzycmd/VDTR}. | ['Yujiu Yang', 'Jue Wang', 'Yong Zhang', 'Yanbo Fan', 'Mingdeng Cao'] | 2022-04-17 | null | null | null | null | ['video-restoration'] | ['computer-vision'] | [-1.67603850e-01 -8.99539709e-01 -2.91091651e-01 1.36939203e-02
-7.00924158e-01 -2.90047348e-01 3.57582718e-01 -5.87339997e-01
2.72602551e-02 5.57741821e-01 6.98699355e-01 -3.48676354e-01
7.62334839e-03 -3.14561874e-01 -6.96544111e-01 -6.72262013e-01
-7.89187998e-02 -3.67647588e-01 3.96549612e-01 -5.00980914e-02
2.29627565e-01 4.74331170e-01 -1.28676593e+00 3.24470758e-01
9.01147723e-01 8.54680836e-01 4.83680099e-01 8.79076421e-01
3.29847962e-01 1.10637403e+00 -4.01734769e-01 -4.08058567e-03
6.22317009e-02 -3.86401474e-01 -7.52992272e-01 -4.26704250e-02
7.01907873e-01 -9.77042973e-01 -1.14088547e+00 9.54720020e-01
6.04201615e-01 2.13296100e-01 4.23064291e-01 -1.01585555e+00
-1.05793476e+00 2.18062460e-01 -8.07092667e-01 1.01316702e+00
2.58409709e-01 1.84004650e-01 6.15228713e-01 -1.01226008e+00
3.97120416e-01 1.11793804e+00 8.46905649e-01 5.73354602e-01
-8.04934323e-01 -7.94872463e-01 5.42934947e-02 1.00735152e+00
-1.43627250e+00 -6.86198235e-01 6.11428678e-01 -3.75677854e-01
1.12107670e+00 2.92695284e-01 4.75256264e-01 1.20966899e+00
4.25425202e-01 8.17768157e-01 8.75581503e-01 -8.05118703e-04
-7.30613396e-02 -6.57093108e-01 2.77138650e-01 3.46992970e-01
-1.22599207e-01 3.37139636e-01 -6.28921926e-01 -4.73616160e-02
1.16946304e+00 9.77904648e-02 -8.57166886e-01 6.27595931e-02
-1.24173927e+00 3.43673736e-01 3.89012694e-01 3.24984491e-01
-3.57076943e-01 5.37212491e-01 4.82871503e-01 3.10804933e-01
7.21903503e-01 9.23803151e-02 -5.01710951e-01 -4.33851421e-01
-1.40609765e+00 2.02248544e-01 1.27666980e-01 8.79323006e-01
4.43916976e-01 2.00479746e-01 -3.91734451e-01 9.85609531e-01
2.06231073e-01 2.91100532e-01 6.11482441e-01 -1.12843442e+00
3.98090780e-01 -1.68714479e-01 1.79580554e-01 -1.00272226e+00
4.41509001e-02 -4.26846981e-01 -1.25969195e+00 -8.40959996e-02
1.18471704e-01 1.23062976e-01 -9.48754549e-01 1.50289035e+00
4.44512852e-02 1.11952484e+00 -1.90207332e-01 1.25987017e+00
7.96791255e-01 9.33884680e-01 -1.59117848e-01 -3.42549652e-01
1.35872710e+00 -1.24979937e+00 -1.06822431e+00 -5.86428344e-02
3.56439143e-01 -8.50185275e-01 7.40171671e-01 2.07841843e-01
-1.16966021e+00 -4.38284576e-01 -9.68463778e-01 -3.62173736e-01
1.43763661e-01 2.13268444e-01 2.65958399e-01 2.71958679e-01
-1.39247441e+00 4.70649511e-01 -9.83471930e-01 -2.68543273e-01
4.57925260e-01 4.84485701e-02 -2.07707569e-01 -5.16150594e-01
-1.49509466e+00 8.68946850e-01 -1.84874944e-02 3.77753139e-01
-1.14628804e+00 -8.21831107e-01 -8.73322546e-01 1.72359556e-01
1.79659650e-01 -8.16687107e-01 1.16770339e+00 -6.62299931e-01
-1.24658060e+00 3.52610439e-01 -5.97607136e-01 -4.57299680e-01
5.13577759e-01 -5.05317390e-01 -4.69779342e-01 1.62947282e-01
-4.78231795e-02 3.69868755e-01 1.26311684e+00 -9.97929573e-01
-3.51261318e-01 -1.11949615e-01 -2.97024567e-03 2.68846095e-01
-4.31790799e-01 4.13841456e-01 -8.01891685e-01 -1.40283215e+00
-2.95822714e-02 -6.11828446e-01 4.82384562e-02 -1.04699381e-01
-1.59260347e-01 1.00396901e-01 1.31592309e+00 -1.27951801e+00
1.68882620e+00 -2.10438275e+00 4.44038868e-01 -6.46658182e-01
4.67696488e-01 5.50033987e-01 -2.97072887e-01 2.60809690e-01
-3.34960550e-01 1.34787545e-01 -3.42471972e-02 -3.91407639e-01
-3.13924789e-01 -1.27986029e-01 -4.43062335e-01 5.81563592e-01
-7.50545785e-02 9.48479593e-01 -7.49895930e-01 -8.37879702e-02
4.83482480e-01 8.10769677e-01 -4.61292952e-01 3.49553853e-01
9.29501876e-02 4.66493726e-01 -1.66515782e-01 6.11118376e-01
9.60613668e-01 -3.60877961e-01 -2.12395981e-01 -6.57725394e-01
-7.88743347e-02 1.78805292e-01 -8.11483920e-01 1.72636914e+00
-4.68947977e-01 1.22989821e+00 7.91150928e-02 -6.62335634e-01
3.29868436e-01 5.77463567e-01 4.91833478e-01 -6.28657818e-01
1.03772469e-01 1.55848429e-01 -4.84625071e-01 -7.45151281e-01
8.49792480e-01 1.84120744e-01 3.92342627e-01 4.20040190e-01
-1.38610676e-01 2.90929794e-01 -1.27291456e-01 3.07601571e-01
1.32072020e+00 1.47774801e-01 -4.62700017e-02 -2.59379476e-01
5.54209530e-01 -3.04605514e-01 6.23017550e-01 4.68876302e-01
-3.20901811e-01 1.10190368e+00 6.42274618e-02 -6.00502074e-01
-1.13206518e+00 -8.58150423e-01 -1.51284290e-02 7.64955223e-01
4.59783942e-01 -5.06516218e-01 -8.25678349e-01 -3.91249865e-01
-4.20915961e-01 5.11194408e-01 -4.15559649e-01 -2.52897501e-01
-9.43363547e-01 -9.61885393e-01 6.39805973e-01 4.88552064e-01
7.03368545e-01 -4.80673969e-01 -1.68375984e-01 2.23952949e-01
-8.19034278e-01 -1.26878834e+00 -1.15473878e+00 -4.04271454e-01
-8.81752491e-01 -1.03032625e+00 -1.31321061e+00 -7.48361826e-01
2.50713736e-01 9.95408952e-01 9.73897517e-01 1.99485451e-01
-1.36311769e-01 1.27374291e-01 -5.08475304e-01 5.40092170e-01
-2.56628275e-01 -2.07364082e-01 1.56366173e-02 -1.24984875e-01
8.40434730e-02 -7.72942126e-01 -8.90481472e-01 6.51005208e-01
-1.07505870e+00 2.17414036e-01 1.16036557e-01 9.37765956e-01
1.21011935e-01 1.84641153e-01 4.76190031e-01 -1.27941580e-03
6.62215054e-01 -6.49121761e-01 -3.59297901e-01 2.19835788e-01
-3.53108466e-01 -3.18132132e-01 3.70946914e-01 -6.61046982e-01
-9.96926486e-01 -6.13875985e-01 -6.39303699e-02 -1.11656833e+00
-2.04065908e-02 4.34269190e-01 -4.62798476e-02 -6.28860220e-02
4.28698689e-01 5.70047319e-01 -1.67539328e-01 -8.34071279e-01
1.84170172e-01 7.35363305e-01 5.37839353e-01 -3.71504813e-01
6.68383420e-01 2.38226756e-01 -4.67364222e-01 -8.02812397e-01
-5.04172742e-01 -2.98234195e-01 -3.49228412e-01 -1.68720499e-01
8.60500097e-01 -1.38210857e+00 -3.17137748e-01 1.21445858e+00
-1.52841330e+00 -5.11253834e-01 2.63675988e-01 4.96282399e-01
-3.64039063e-01 1.03893173e+00 -1.24489105e+00 -2.78460264e-01
-3.80887479e-01 -1.43854213e+00 9.47037995e-01 4.61005606e-03
6.33989200e-02 -9.98295784e-01 -4.31666225e-02 6.47585154e-01
8.39099109e-01 -2.86299497e-01 6.12515330e-01 1.06316745e-01
-9.33823466e-01 5.21254912e-02 -6.13495290e-01 4.84302938e-01
4.34792340e-01 -2.14933038e-01 -7.34746337e-01 -5.68049848e-01
1.85169846e-01 -4.89986502e-02 1.14724088e+00 7.92182565e-01
1.26147366e+00 -3.48665237e-01 -1.84267655e-01 9.45938945e-01
1.18772757e+00 2.09496915e-02 1.03049159e+00 4.88601029e-01
1.09156203e+00 -3.60890329e-02 4.72440541e-01 4.04409707e-01
6.96326494e-01 1.15577710e+00 3.51281613e-01 -3.31304036e-02
-7.05293238e-01 1.62640169e-01 6.91069961e-01 9.02190149e-01
-1.02246381e-01 -5.84953904e-01 -8.51137996e-01 6.81046367e-01
-1.97163618e+00 -1.11638689e+00 -3.26688737e-01 1.94881892e+00
8.90032828e-01 -4.46443528e-01 -2.23146588e-01 -1.68560565e-01
9.26452816e-01 5.65766275e-01 -4.27859306e-01 1.26040086e-01
-3.18295330e-01 -2.72439003e-01 5.04185438e-01 6.80137992e-01
-1.30765617e+00 9.73451197e-01 6.00501156e+00 1.17926490e+00
-1.09576154e+00 4.89812762e-01 7.31254876e-01 -1.58919334e-01
-1.18903741e-01 -1.42893270e-01 -6.17314041e-01 8.07792127e-01
8.53281081e-01 -4.84175682e-02 7.21369684e-01 9.22567025e-02
8.08603346e-01 -2.49021929e-02 -8.43392670e-01 1.25367606e+00
1.27669111e-01 -1.63347089e+00 -5.05484231e-02 -1.15485035e-01
8.41236174e-01 2.31457829e-01 1.07038453e-01 -7.45470598e-02
2.98548373e-03 -1.06886899e+00 8.02859485e-01 4.84070718e-01
9.62264478e-01 -2.97555596e-01 7.65098810e-01 1.92782417e-01
-1.34077609e+00 3.14976983e-02 -1.74284354e-01 9.34777111e-02
4.17434245e-01 6.74519181e-01 -1.60287723e-01 6.34422481e-01
1.00871813e+00 1.46954560e+00 -4.16578025e-01 1.21767855e+00
4.44987640e-02 6.78852558e-01 7.58669227e-02 8.51661086e-01
-8.69908556e-03 2.14747083e-03 8.08118224e-01 1.31078231e+00
7.99652398e-01 1.72006994e-01 -3.19911420e-01 6.86663508e-01
-2.45174952e-02 -4.55571204e-01 -1.65046558e-01 1.77426904e-01
5.43702781e-01 7.44421303e-01 -1.11774459e-01 -4.78660226e-01
-4.93574113e-01 1.40052521e+00 -6.61235526e-02 7.12798476e-01
-1.30250943e+00 5.95911816e-02 1.15081346e+00 1.23109169e-01
4.85984325e-01 -5.57989955e-01 -7.46310875e-02 -1.87319970e+00
9.08912271e-02 -1.03978956e+00 1.14406675e-01 -1.13066900e+00
-1.25443244e+00 7.22415924e-01 4.45495534e-05 -1.49892414e+00
-6.37463704e-02 -1.53591260e-01 -4.86832410e-01 1.08442771e+00
-1.86986244e+00 -1.14920473e+00 -5.05602479e-01 7.35174119e-01
1.05791974e+00 -5.18956557e-02 3.18865269e-01 7.51069486e-01
-7.85399973e-01 4.51625794e-01 2.85009623e-01 8.93952325e-02
9.83878791e-01 -7.41279304e-01 7.01261401e-01 1.27877319e+00
-3.83444965e-01 4.10454690e-01 7.87286401e-01 -7.31983066e-01
-1.47113240e+00 -1.32033134e+00 6.67545676e-01 -2.77491957e-01
6.94320261e-01 9.68288407e-02 -1.29123712e+00 6.42771900e-01
4.38288212e-01 3.53438854e-01 2.60356367e-01 -4.51312572e-01
-4.65888053e-01 3.09192389e-02 -8.14399481e-01 5.54459393e-01
1.02741456e+00 -6.07214451e-01 -3.28327060e-01 1.42762035e-01
8.37807477e-01 -5.78289211e-01 -9.21274722e-01 4.48349059e-01
4.06061530e-01 -1.12386763e+00 1.28428483e+00 -2.41715953e-01
7.34308064e-01 -4.95091200e-01 -1.32191315e-01 -1.26950562e+00
-6.56878352e-01 -7.27379620e-01 -6.74463272e-01 1.11972415e+00
-2.01272577e-01 -5.03802657e-01 5.42783380e-01 3.11616808e-01
-3.17440003e-01 -6.21648312e-01 -9.96795356e-01 -7.98548281e-01
6.78281859e-02 -4.63877350e-01 4.62728053e-01 1.14791918e+00
-4.06250298e-01 -2.13185214e-02 -1.16559935e+00 4.63908672e-01
7.39620924e-01 -1.14393301e-01 3.38309884e-01 -4.25024480e-01
-2.82779127e-01 -3.98757666e-01 -3.73694569e-01 -1.72153544e+00
9.07899663e-02 -4.37097430e-01 -7.85259753e-02 -1.54920793e+00
3.89073849e-01 -3.66799265e-01 -2.96791524e-01 2.22535223e-01
-3.73247951e-01 4.35345799e-01 1.79212868e-01 6.54236257e-01
-3.99343133e-01 8.04858088e-01 1.36434591e+00 -2.03507513e-01
3.53716910e-02 -2.57246345e-01 -3.34117323e-01 4.14918721e-01
8.12528133e-01 -2.37881929e-01 -2.98309714e-01 -1.13513649e+00
-1.12925567e-01 5.27828097e-01 8.01138997e-01 -8.72040689e-01
3.75112832e-01 -2.19344839e-01 2.09435776e-01 -5.37916243e-01
4.35960084e-01 -6.35560930e-01 4.54030365e-01 9.73610356e-02
-2.57955268e-02 2.52492398e-01 3.18037868e-01 6.07026458e-01
-4.75690097e-01 2.04445839e-01 8.18616450e-01 1.19287632e-01
-8.69363844e-01 6.13904357e-01 -7.02068031e-01 -1.67953596e-01
7.37543702e-01 -3.04947346e-01 -6.53537810e-01 -6.14474416e-01
-6.12919450e-01 2.66625062e-02 7.70587981e-01 6.87130809e-01
9.11226034e-01 -1.32548440e+00 -8.24062049e-01 1.09671682e-01
-4.06818479e-01 -3.04047137e-01 9.35413957e-01 1.22047484e+00
-6.26222312e-01 3.25421095e-01 -2.68859625e-01 -4.64652359e-01
-1.50255811e+00 4.95951384e-01 6.14116192e-01 -1.30167842e-01
-7.10214674e-01 8.91674459e-01 5.23051620e-01 3.59214693e-01
8.98521096e-02 -2.34776124e-01 -7.26727908e-03 -3.05449635e-01
8.72770131e-01 6.02526724e-01 7.96951279e-02 -9.10020769e-01
-2.80565381e-01 5.87317944e-01 -1.41923606e-01 2.44641453e-01
1.26876366e+00 -7.93035328e-01 -3.47492993e-01 -1.08461857e-01
1.25189793e+00 -1.79442063e-01 -1.57004058e+00 -4.49977726e-01
-2.74325341e-01 -9.25954759e-01 5.53699791e-01 -6.42881989e-01
-1.42745411e+00 9.56453502e-01 6.14419937e-01 -1.44863859e-01
1.51741505e+00 -9.90857705e-02 1.25022197e+00 -3.27935308e-01
1.89737409e-01 -4.61426318e-01 2.61789709e-01 7.31080592e-01
1.10161293e+00 -1.03655910e+00 9.91839096e-02 -4.39065278e-01
-3.44604135e-01 1.11275005e+00 5.34759045e-01 -5.93248121e-02
6.51822329e-01 3.79982203e-01 -5.96509278e-02 9.86537412e-02
-9.04747248e-01 3.31738025e-01 4.12412852e-01 3.49772245e-01
4.48584110e-01 -2.27466822e-01 -9.69660804e-02 3.62572521e-01
4.26528096e-01 3.09704214e-01 6.89178109e-01 6.19038463e-01
-2.39066362e-01 -1.03829253e+00 -5.98148406e-01 1.90034986e-01
-5.89511096e-01 -5.13320565e-01 2.59712994e-01 1.71460956e-01
-7.38126487e-02 9.53967929e-01 -4.71838899e-02 -5.65786064e-01
-1.79948375e-01 -4.42243457e-01 4.55673665e-01 -1.25875115e-01
-2.85381585e-01 3.71508896e-01 -1.10222753e-02 -7.41781294e-01
-4.96782422e-01 -5.75193465e-01 -7.10636497e-01 -6.80341899e-01
-2.49534726e-01 -2.06913292e-01 1.30182534e-01 8.48263919e-01
7.09586263e-01 6.28271878e-01 4.75544840e-01 -1.20465863e+00
-3.62793744e-01 -1.04296482e+00 -3.32918525e-01 7.23619014e-02
9.05906320e-01 -5.98363698e-01 -3.77932131e-01 3.99687648e-01] | [11.318318367004395, -2.286377191543579] |
1eff5b83-b07c-4737-8b71-8d4685230e6d | analysis-of-adversarial-image-manipulations | 2305.06307 | null | https://arxiv.org/abs/2305.06307v1 | https://arxiv.org/pdf/2305.06307v1.pdf | Analysis of Adversarial Image Manipulations | As virtual and physical identity grow increasingly intertwined, the importance of privacy and security in the online sphere becomes paramount. In recent years, multiple news stories have emerged of private companies scraping web content and doing research with or selling the data. Images uploaded online can be scraped without users' consent or knowledge. Users of social media platforms whose images are scraped may be at risk of being identified in other uploaded images or in real-world identification situations. This paper investigates how simple, accessible image manipulation techniques affect the accuracy of facial recognition software in identifying an individual's various face images based on one unique image. | ['Michael C. King', 'Gabriella Pangelinan', 'Ahsi Lo'] | 2023-05-10 | null | null | null | null | ['image-manipulation'] | ['computer-vision'] | [ 2.22708687e-01 6.57055080e-02 -2.02855691e-01 -4.63609099e-01
-4.25050050e-01 -9.95169044e-01 3.69408131e-01 7.96158537e-02
-5.63194990e-01 5.31084418e-01 -1.33361787e-01 -2.29068279e-01
1.11303613e-01 -4.65305150e-01 -2.26803586e-01 -3.86202812e-01
9.79047939e-02 -1.56272128e-01 -8.91454220e-02 8.44425410e-02
4.38703954e-01 8.02327275e-01 -1.75619388e+00 1.35634735e-01
3.66264611e-01 1.14624524e+00 -2.08961636e-01 4.32585716e-01
-3.05405974e-01 5.35625935e-01 -7.47471571e-01 -1.24404788e+00
6.56857967e-01 -1.50078326e-01 -5.57424366e-01 1.94042742e-01
7.47632146e-01 -9.11019802e-01 -1.02772459e-01 1.17575955e+00
4.01385427e-01 -4.02936161e-01 -9.53199267e-02 -1.43559146e+00
-9.48174119e-01 -4.26976755e-03 -8.07064652e-01 5.77332340e-02
8.02872837e-01 2.03405261e-01 3.05138737e-01 -7.37015486e-01
8.02071989e-01 9.25884247e-01 6.99388623e-01 4.48276281e-01
-1.56502128e+00 -1.03664148e+00 -3.07169795e-01 -1.78239256e-01
-1.99388003e+00 -1.12012768e+00 4.64959741e-01 -5.63734591e-01
2.92192817e-01 5.20891309e-01 7.26458907e-01 9.61965263e-01
1.77867547e-01 1.07941739e-01 1.06682491e+00 -2.71219045e-01
-1.80056229e-01 8.25277805e-01 -2.05480725e-01 4.65274870e-01
5.16656399e-01 -2.33865380e-01 -7.29559422e-01 -6.09278798e-01
8.09336305e-01 -2.80513763e-02 -1.21501848e-01 -3.22426736e-01
-7.23054767e-01 5.11208415e-01 -2.90553868e-01 2.43920028e-01
-2.13811889e-01 -3.92519295e-01 4.58687246e-01 3.33395839e-01
4.36984807e-01 2.62374312e-01 1.29941821e-01 -3.61232400e-01
-7.06585586e-01 1.19503453e-01 6.76064789e-01 6.78795040e-01
6.91904724e-01 -4.72427756e-01 4.76719171e-01 8.81759405e-01
1.48049951e-01 2.73676395e-01 6.10841393e-01 -7.49259233e-01
2.10432708e-01 5.56636333e-01 3.83594811e-01 -1.55145729e+00
2.14531794e-01 2.52639562e-01 -2.89605707e-01 4.69778359e-01
7.61560678e-01 -1.69733942e-01 -3.37415397e-01 1.20651162e+00
4.13700879e-01 -2.90854871e-01 -2.52470434e-01 7.27711797e-01
5.06288528e-01 9.98176485e-02 3.84617090e-01 -1.32004276e-01
1.48175669e+00 8.73546004e-02 -9.88538980e-01 -9.01212841e-02
-6.76449109e-03 -1.14213538e+00 8.11045885e-01 3.36086094e-01
-1.20696485e+00 -3.05076957e-01 -1.03769588e+00 -2.12617144e-01
-6.57044291e-01 4.10988592e-02 5.56996703e-01 1.49678648e+00
-1.22944808e+00 8.01422536e-01 -4.86684382e-01 -6.76479578e-01
7.99493372e-01 6.79024100e-01 -8.93000662e-01 2.58696198e-01
-7.88577616e-01 6.25773966e-01 -8.01954418e-02 4.92032170e-02
-3.28293331e-02 -4.63329852e-01 -6.29819155e-01 -2.93061435e-01
6.10771403e-02 -8.31651017e-02 1.04479885e+00 -1.51620293e+00
-1.08899939e+00 1.63147438e+00 -8.91921520e-02 -1.34591714e-01
8.09203029e-01 -2.35654667e-01 -1.06594276e+00 3.85999799e-01
2.74794608e-01 2.97073364e-01 1.28549051e+00 -8.57985377e-01
-4.44225013e-01 -8.52388382e-01 -1.45325791e-02 -3.52869667e-02
-6.03742838e-01 5.97489178e-01 -3.18618342e-02 -4.21836257e-01
7.70489722e-02 -5.23082852e-01 6.04430676e-01 5.69088697e-01
-1.63835317e-01 2.33532220e-01 1.38503826e+00 -9.70620394e-01
1.05409062e+00 -2.61871195e+00 -6.15769506e-01 3.28014374e-01
3.07048440e-01 4.92042720e-01 2.14941606e-01 5.22329330e-01
-1.92237571e-02 8.94863188e-01 6.47755623e-01 -7.27264807e-02
-9.97268632e-02 -2.27670133e-01 -8.86504352e-03 6.35517478e-01
-2.27857485e-01 4.16165471e-01 -6.99312627e-01 -6.45339608e-01
-4.93175238e-02 7.03747153e-01 -3.12491693e-02 -3.14722173e-02
6.68677449e-01 2.62732506e-01 -2.86073923e-01 7.83166766e-01
1.18943775e+00 -9.10617113e-02 4.23060238e-01 2.25811034e-01
-3.02131414e-01 2.78483685e-02 -1.05451930e+00 1.02972579e+00
-9.51664075e-02 9.37596798e-01 6.67220473e-01 -2.71379873e-02
6.61063910e-01 3.91753316e-01 3.42697501e-01 -5.95623136e-01
1.86063752e-01 4.65652980e-02 -4.31494445e-01 -5.83749652e-01
7.86606789e-01 -7.39101246e-02 4.17327315e-01 7.93274641e-01
-4.07765269e-01 4.51410890e-01 -2.60529011e-01 1.32863015e-01
5.20836592e-01 -1.07019849e-01 3.05643857e-01 -2.23912388e-01
1.89587861e-01 -2.88607597e-01 2.27046341e-01 4.11037892e-01
-7.33888447e-01 5.57906389e-01 3.76974642e-01 -4.18914735e-01
-9.15207088e-01 -1.16345668e+00 -4.74574447e-01 9.05026674e-01
1.12273917e-01 -3.69691432e-01 -8.71629000e-01 -3.95722866e-01
4.63705659e-01 3.60458791e-01 -4.30055201e-01 2.04366997e-01
9.54620540e-02 -1.38651654e-01 5.92758775e-01 -4.88721579e-03
5.01151681e-01 -5.43652892e-01 -7.61078000e-01 -4.55678441e-02
-6.89868480e-02 -1.09584689e+00 -6.80530608e-01 -8.15147638e-01
-3.86473119e-01 -1.10505283e+00 -6.58942580e-01 -4.80632365e-01
9.72125649e-01 5.94254315e-01 4.94575530e-01 3.07047725e-01
-5.79387665e-01 8.29098105e-01 -2.12318420e-01 -5.31178415e-01
-2.43913114e-01 -1.35909364e-01 1.62878484e-01 8.42423558e-01
7.21043587e-01 -5.41251898e-01 -5.36699116e-01 5.16737103e-01
-8.19463789e-01 -2.08248585e-01 -2.29238540e-01 9.34506655e-02
5.18338792e-02 9.53626856e-02 5.69160223e-01 -8.25491488e-01
7.04360604e-01 -4.66989040e-01 -4.91029531e-01 2.45866403e-01
-3.35379392e-01 -8.55454385e-01 1.85154259e-01 -3.96198630e-01
-1.00806475e+00 4.93808798e-02 3.20761412e-01 -2.93674439e-01
-2.85150409e-01 -6.92077726e-02 -1.46840632e-01 -7.12032199e-01
5.43236434e-01 1.14573538e-02 7.40900040e-01 -3.19451511e-01
-2.29447395e-01 1.25056398e+00 3.13686311e-01 5.99050596e-02
8.86693239e-01 7.26474404e-01 -6.53111637e-01 -1.31483054e+00
-1.41810268e-01 -3.49425405e-01 -6.87891364e-01 -6.31832480e-01
8.19600880e-01 -8.02647352e-01 -1.23076916e+00 8.58547568e-01
-6.39928281e-01 2.85408050e-01 9.58490521e-02 2.26642609e-01
4.51716520e-02 5.43106318e-01 -2.77732521e-01 -1.04731750e+00
2.23966315e-01 -8.00594091e-01 6.52493775e-01 5.26694894e-01
-6.33917928e-01 -6.32883728e-01 -4.89389181e-01 7.33851612e-01
6.57772481e-01 3.94900024e-01 2.53604084e-01 -4.01837617e-01
-5.22080064e-01 -1.02187312e+00 -4.10720736e-01 2.01987907e-01
8.07318866e-01 4.35918123e-01 -1.26752543e+00 -1.73634380e-01
-8.47540274e-02 -3.30753177e-01 -1.79434121e-01 -8.77542943e-02
1.10524690e+00 -7.48784244e-01 -3.26495856e-01 2.74006039e-01
1.37331510e+00 4.87658322e-01 8.58606577e-01 3.59198809e-01
3.78686607e-01 8.88840556e-01 2.73835033e-01 6.67864084e-01
2.90533938e-02 4.79025066e-01 1.79939121e-01 2.86853015e-01
5.74070215e-01 -4.46844548e-01 1.28534660e-01 -4.59750928e-02
-6.65682182e-02 7.02812076e-02 -8.07778835e-01 2.35530213e-01
-1.17186522e+00 -1.20298016e+00 2.24400580e-01 2.71219063e+00
4.74440187e-01 -2.18324974e-01 4.15895134e-01 -1.64124548e-01
1.28640056e+00 -3.82718518e-02 -5.44056654e-01 -4.96688604e-01
6.71403334e-02 6.27510995e-02 7.33586371e-01 3.69845659e-01
-9.20649529e-01 5.34628212e-01 7.05263472e+00 3.54691744e-01
-1.17543936e+00 7.98439533e-02 1.05157530e+00 -4.43183333e-01
-1.62796453e-01 -8.24460834e-02 -6.67548418e-01 5.23041189e-01
7.75176227e-01 -5.48031271e-01 5.49659133e-01 8.40692937e-01
2.77514905e-01 -6.19194806e-01 -8.97061706e-01 1.21558857e+00
2.72464573e-01 -1.22785997e+00 -1.48324341e-01 5.82906604e-01
1.90356866e-01 -6.61752939e-01 5.79025328e-01 -4.85284656e-01
-5.56241758e-02 -1.10505557e+00 7.15037942e-01 4.40434635e-01
9.95484114e-01 -7.01636851e-01 2.41477758e-01 -5.12063466e-02
-8.99822474e-01 -4.47923169e-02 -1.39269441e-01 -2.01977462e-01
6.41167909e-02 -4.56560887e-02 -6.91347778e-01 -7.02733966e-03
7.69615054e-01 1.94438726e-01 -7.76301503e-01 7.92024255e-01
3.55658352e-01 1.46370471e-01 -3.48167598e-01 1.01242602e-01
-2.75191665e-01 -4.93837237e-01 4.74526197e-01 6.42361045e-01
1.47106141e-01 2.02176213e-01 -6.33484483e-01 6.51564538e-01
9.57822055e-02 2.51280099e-01 -8.87913704e-01 -7.55424023e-01
5.06728292e-01 1.26855588e+00 -8.04150939e-01 1.41122386e-01
-4.90323722e-01 1.17577314e+00 -3.75625342e-02 2.49585688e-01
-4.52737331e-01 -3.89769495e-01 1.09991229e+00 8.58119726e-01
-5.26657999e-02 -1.15904614e-01 -3.93234678e-02 -7.92539001e-01
2.92385250e-01 -9.88217473e-01 1.82983086e-01 -8.53716731e-01
-1.05836105e+00 4.04996514e-01 -1.94654420e-01 -1.11013675e+00
-2.87698489e-02 -3.02714407e-01 -2.96289265e-01 8.98120403e-01
-4.97540295e-01 -9.88070130e-01 -2.57001728e-01 3.76232952e-01
-1.52559474e-01 -2.47781411e-01 8.51379871e-01 2.95432687e-01
-5.84354699e-01 9.67743456e-01 2.34212294e-01 3.49320710e-01
7.44525671e-01 -4.50501055e-01 3.85860443e-01 4.50256497e-01
-1.57354027e-01 1.02920353e+00 2.71743834e-01 -7.46604204e-01
-1.53188765e+00 -4.43396330e-01 9.45190191e-01 -6.80812359e-01
5.83193898e-01 -6.29024148e-01 -6.44363046e-01 9.39705133e-01
2.23447531e-01 -1.88514411e-01 1.21274519e+00 -4.20268476e-02
-4.49524820e-01 -2.48486817e-01 -1.94691443e+00 7.83156037e-01
1.01699114e+00 -1.16904426e+00 7.19901770e-02 3.36192518e-01
-1.01874992e-01 -1.10385314e-01 -9.48870599e-01 -3.55991393e-01
1.26582563e+00 -1.21694803e+00 8.67569447e-01 -4.58086789e-01
1.01835802e-01 -2.04934161e-02 1.70029998e-01 -4.52403426e-01
2.81058159e-03 -1.04301286e+00 6.54906154e-01 1.62057328e+00
1.83238789e-01 -1.15172744e+00 1.04000425e+00 1.79383147e+00
8.60022664e-01 -8.13227519e-02 -8.57550383e-01 -6.56377852e-01
-5.15421152e-01 -4.80619296e-02 7.37463117e-01 1.22583580e+00
5.28208077e-01 -4.08692986e-01 -2.62719572e-01 1.83206230e-01
6.72780871e-01 -5.84759653e-01 8.73379350e-01 -1.19362640e+00
1.46811336e-01 -3.04137319e-01 -9.34822619e-01 -8.85733590e-02
-1.82145014e-01 -4.51795697e-01 -8.66743326e-01 -8.27271760e-01
9.76250991e-02 -1.44888163e-01 1.85563743e-01 3.35774750e-01
5.33689380e-01 5.38810849e-01 4.18301940e-01 4.07998979e-01
-1.15429990e-01 -2.20692426e-01 8.80323052e-01 2.09167138e-01
-2.37502605e-01 -1.09472960e-01 -1.02971315e+00 5.15932798e-01
8.47594678e-01 -2.09098935e-01 -4.04706597e-01 3.01495139e-02
2.43518889e-01 -7.75793493e-02 5.35759449e-01 -9.40879583e-01
1.01976119e-01 -1.52837306e-01 7.16806650e-01 -1.88352272e-01
4.64100331e-01 -1.13488078e+00 7.27872133e-01 1.65272832e-01
-1.89363450e-01 1.67112410e-01 4.54526663e-01 3.66458744e-01
-7.36640170e-02 -3.16700995e-01 8.33341479e-01 -1.90820873e-01
-4.85825360e-01 -3.23642790e-03 -6.18460059e-01 -2.75091469e-01
1.54403627e+00 -1.08705056e+00 -2.12693274e-01 -7.18990207e-01
-9.06793714e-01 -1.97441429e-01 1.08562636e+00 3.74448717e-01
5.08716226e-01 -1.02227783e+00 -1.88009828e-01 5.85257053e-01
-4.72062267e-02 -8.13131869e-01 3.70040983e-01 5.92097878e-01
-7.65844822e-01 -8.79730582e-02 -6.98308468e-01 -9.76883695e-02
-1.69018233e+00 4.49516594e-01 2.13773400e-01 8.42826903e-01
-5.45171201e-01 7.42786169e-01 -6.94676936e-02 1.68417409e-01
6.96524382e-02 6.20037079e-01 2.94401143e-02 5.06334543e-01
1.02587819e+00 5.17219007e-01 -1.44651547e-01 -1.09610355e+00
-3.78118634e-01 2.73918748e-01 -5.43455243e-01 -3.09337497e-01
7.72609711e-01 -4.36324239e-01 -2.84659207e-01 2.37306386e-01
1.40907180e+00 3.56116980e-01 -1.02799749e+00 4.33425367e-01
-4.87372011e-01 -1.42286217e+00 -2.16628551e-01 -3.69985342e-01
-1.19218910e+00 3.21863413e-01 8.14680636e-01 8.36110353e-01
9.31871414e-01 -3.22640568e-01 5.31350791e-01 -1.75677687e-01
6.22193694e-01 -1.28105772e+00 -2.99239844e-01 -3.36550444e-01
7.43430316e-01 -1.17557979e+00 3.07513833e-01 -6.40588403e-01
-6.44890189e-01 1.06196928e+00 3.43034923e-01 3.32401454e-01
8.88187170e-01 1.94389462e-01 2.85024196e-01 -1.14688806e-01
-3.67183425e-02 4.24063087e-01 -1.99629098e-01 9.10270572e-01
3.85698020e-01 2.36408636e-02 -3.99975359e-01 2.78044075e-01
-3.33012968e-01 2.86805406e-02 7.33207464e-01 1.08812666e+00
-1.21528447e-01 -1.16774940e+00 -7.16511130e-01 6.23975277e-01
-9.26426888e-01 2.64159858e-01 -1.05186605e+00 8.91674399e-01
-1.13554504e-02 8.32294762e-01 3.95202041e-01 -3.15009743e-01
-1.48802131e-01 3.89908552e-01 1.16093263e-01 -2.90339857e-01
-7.12788582e-01 -4.44335043e-01 2.58247197e-01 -6.09415054e-01
-2.95129806e-01 -1.36401451e+00 -6.32659435e-01 -8.22220623e-01
-1.36891454e-01 -1.34314314e-01 9.62067246e-01 3.55845809e-01
7.47568250e-01 -8.22292268e-01 4.86930192e-01 -5.74954808e-01
-1.78130761e-01 -1.85244381e-01 -9.80648935e-01 3.36682618e-01
3.44614297e-01 -2.34450832e-01 -4.74940091e-01 2.29020208e-01] | [12.814892768859863, 0.9786702990531921] |
8468ed60-8ebc-4090-9e08-3fb683af89f1 | causal-effect-estimation-with-variational | 2304.11969 | null | https://arxiv.org/abs/2304.11969v1 | https://arxiv.org/pdf/2304.11969v1.pdf | Causal Effect Estimation with Variational AutoEncoder and the Front Door Criterion | An essential problem in causal inference is estimating causal effects from observational data. The problem becomes more challenging with the presence of unobserved confounders. When there are unobserved confounders, the commonly used back-door adjustment is not applicable. Although the instrumental variable (IV) methods can deal with unobserved confounders, they all assume that the treatment directly affects the outcome, and there is no mediator between the treatment and the outcome. This paper aims to use the front-door criterion to address the challenging problem with the presence of unobserved confounders and mediators. In practice, it is often difficult to identify the set of variables used for front-door adjustment from data. By leveraging the ability of deep generative models in representation learning, we propose FDVAE to learn the representation of a Front-Door adjustment set with a Variational AutoEncoder, instead of trying to search for a set of variables for front-door adjustment. Extensive experiments on synthetic datasets validate the effectiveness of FDVAE and its superiority over existing methods. The experiments also show that the performance of FDVAE is not sensitive to the causal strength of unobserved confounders and is feasible in the case of dimensionality mismatch between learned representations and the ground truth. We further apply the method to three real-world datasets to demonstrate its potential applications. | ['Kui Yu', 'Lin Liu', 'Jixue Liu', 'Jiuyong Li', 'Debo Cheng', 'Ziqi Xu'] | 2023-04-24 | null | null | null | null | ['causal-inference', 'causal-inference'] | ['knowledge-base', 'miscellaneous'] | [ 1.64416641e-01 4.61638011e-02 -6.06794000e-01 -2.72641271e-01
-6.14397645e-01 -4.25079137e-01 6.97853327e-01 2.76724864e-02
-1.10374682e-01 8.82839799e-01 6.29883647e-01 -4.55594867e-01
-4.49900329e-01 -1.00727248e+00 -9.21219885e-01 -8.30090702e-01
-1.08934425e-01 5.11412561e-01 -4.59915221e-01 2.19178032e-02
-1.76221475e-01 3.01384598e-01 -1.25621772e+00 -1.59220323e-01
7.70096481e-01 3.21333379e-01 7.65108019e-02 2.68919438e-01
1.24210648e-01 6.69657767e-01 -4.22875851e-01 -2.43094414e-01
2.10235775e-01 -5.49671650e-01 -4.93292540e-01 -2.92242080e-01
1.05213590e-01 -4.64253962e-01 -3.47086161e-01 8.65493119e-01
4.54303622e-01 1.35251626e-01 1.02539802e+00 -1.37343872e+00
-7.60629833e-01 7.22301483e-01 -5.39216220e-01 2.48902068e-01
4.09872271e-02 2.39886977e-02 1.13603699e+00 -7.40443885e-01
4.51191694e-01 1.55978489e+00 5.63142121e-01 3.93903702e-01
-1.30557191e+00 -1.00153422e+00 3.66520345e-01 -4.82808463e-02
-1.10383642e+00 -4.59805131e-01 6.94311738e-01 -6.91854179e-01
5.06418765e-01 1.00795075e-01 2.30530441e-01 1.45704007e+00
2.03126773e-01 3.48076224e-01 8.00096869e-01 -2.89485604e-01
3.31919968e-01 -2.47750297e-01 1.15386888e-01 5.87917268e-01
3.94675970e-01 6.75367415e-01 -2.38301694e-01 -5.93616843e-01
9.37624753e-01 3.57736290e-01 -2.70319343e-01 -3.78427953e-01
-1.31141996e+00 1.39771223e+00 6.11026406e-01 -1.09109674e-02
-8.41333747e-01 4.27856505e-01 2.17890859e-01 9.48315337e-02
4.37893987e-01 3.60244870e-01 -2.77595669e-01 2.79132009e-01
-8.04165959e-01 3.13056916e-01 4.79085147e-01 6.28644645e-01
5.13910592e-01 2.19685622e-02 -3.96639407e-01 5.07871926e-01
3.39245170e-01 6.37181938e-01 2.42107630e-01 -8.64987075e-01
5.36312580e-01 5.51485360e-01 3.67475718e-01 -1.01155376e+00
-4.16109562e-01 -1.75477237e-01 -1.25591958e+00 9.32222698e-03
4.44485903e-01 -3.92263532e-01 -9.80055094e-01 2.28843641e+00
6.64781034e-01 4.49369341e-01 6.66195303e-02 1.08914173e+00
8.33933413e-01 6.53778672e-01 2.02648073e-01 -4.23717439e-01
1.11470187e+00 -3.69819015e-01 -9.55326259e-01 -2.71108359e-01
3.65510881e-01 -3.44138682e-01 7.37519503e-01 -1.69502854e-01
-8.68324339e-01 -3.64551961e-01 -6.99767888e-01 -7.45239258e-02
3.61007378e-02 7.75201172e-02 8.74455333e-01 4.08031076e-01
-4.01719481e-01 4.43862557e-01 -8.31647396e-01 -1.99920796e-02
2.55587876e-01 5.91835499e-01 -4.76538062e-01 -1.26409000e-02
-1.53707314e+00 5.25782466e-01 4.74670045e-02 2.22461417e-01
-1.26380837e+00 -9.37091231e-01 -7.86724985e-01 4.49813366e-01
4.57386166e-01 -1.08836508e+00 9.84411120e-01 -9.28013027e-01
-1.16554284e+00 3.13375145e-01 -5.26279867e-01 -1.04126371e-01
4.41406906e-01 -9.54561457e-02 -1.20636255e-01 -2.48607889e-01
3.27205449e-01 1.94571599e-01 7.67646134e-01 -1.09379053e+00
-4.44219679e-01 -5.77217400e-01 2.12411359e-01 -2.95867715e-02
-9.23213288e-02 -1.81136757e-01 -1.63678691e-01 -6.75411820e-01
8.08093473e-02 -9.47767138e-01 -2.12276444e-01 -2.11655945e-01
-5.22413671e-01 -4.36740965e-01 5.15209258e-01 -4.65962321e-01
1.09070969e+00 -2.14882898e+00 2.91075528e-01 1.65828690e-01
3.01589847e-01 -1.08984619e-01 -4.48898487e-02 4.75283653e-01
-4.41235274e-01 1.79595724e-01 -4.13132131e-01 -3.63774121e-01
-5.96800260e-02 4.16153431e-01 -5.62004626e-01 7.28071690e-01
2.37803221e-01 8.10652792e-01 -8.50937068e-01 -3.17006439e-01
-3.48271686e-03 4.62757677e-01 -7.96280980e-01 5.22610605e-01
-7.99020603e-02 7.41908967e-01 -6.72028244e-01 2.19005257e-01
4.95991319e-01 -4.69143391e-01 3.70397210e-01 1.02668993e-01
2.08495045e-03 3.74520093e-01 -1.49131501e+00 1.09128976e+00
-3.01414043e-01 3.85177195e-01 -1.56269297e-01 -1.21858907e+00
6.14875197e-01 6.24797404e-01 3.17911893e-01 -4.40918207e-01
1.17035083e-01 -3.97374593e-02 7.61767849e-02 -6.62910163e-01
-9.82742161e-02 -6.05523646e-01 -1.90726355e-01 3.43031198e-01
-1.16136789e-01 4.04822320e-01 -2.46245638e-01 8.32393393e-02
1.01809657e+00 -1.87412426e-01 5.45076787e-01 -1.11006580e-01
1.04097970e-01 -1.10613532e-01 1.02264559e+00 9.95319545e-01
2.00970024e-01 4.79784369e-01 8.12801003e-01 -3.73801112e-01
-8.45583618e-01 -1.25193143e+00 -7.60717019e-02 9.41773057e-01
-8.36836547e-02 9.46190879e-02 -3.92844468e-01 -5.62905610e-01
1.67980611e-01 7.20320523e-01 -1.23724508e+00 -2.80645579e-01
-4.25544173e-01 -1.00917172e+00 3.18537891e-01 6.66596353e-01
7.80911818e-02 -7.31260717e-01 -3.52092683e-01 1.61305651e-01
-4.33960319e-01 -5.41724384e-01 -2.66346544e-01 1.16484597e-01
-8.20627332e-01 -1.30940473e+00 -5.73253632e-01 -4.84090179e-01
6.25238776e-01 7.90892988e-02 9.29568946e-01 -6.48666471e-02
1.47825554e-01 2.94320304e-02 1.95279732e-01 -5.81780136e-01
-4.58167136e-01 -1.71267331e-01 5.88509850e-02 -1.88874137e-02
2.22943038e-01 -4.56337065e-01 -7.96141744e-01 1.23141266e-01
-7.36777008e-01 -1.83985114e-01 2.54758924e-01 1.21253800e+00
4.67803091e-01 -5.81410639e-02 8.66781890e-01 -1.01749289e+00
4.92587924e-01 -9.73550379e-01 -7.38798738e-01 5.10311089e-02
-5.24413824e-01 2.23835766e-01 5.27200937e-01 -7.20214665e-01
-1.07059968e+00 -1.25128359e-01 1.02350250e-01 -5.96798956e-01
5.34735583e-02 7.24169672e-01 -3.57914656e-01 8.40949297e-01
5.60335696e-01 -4.09235239e-01 -7.66687617e-02 -4.77572560e-01
3.66824061e-01 6.15302980e-01 2.70087868e-01 -4.21068609e-01
4.40921366e-01 8.27497780e-01 1.83370367e-01 -2.73566037e-01
-8.60086739e-01 -2.04065695e-01 -2.88157403e-01 2.81646430e-01
9.30435479e-01 -1.04986477e+00 -1.01693010e+00 -8.05216804e-02
-1.02025545e+00 -2.23898306e-01 -2.44716778e-01 8.56447220e-01
-4.75656360e-01 -1.88263319e-02 -2.86918432e-01 -7.34677017e-01
-1.30536333e-01 -1.45617211e+00 1.01840019e+00 -1.28638253e-01
-3.91861290e-01 -1.11667371e+00 3.00101191e-01 8.17462504e-02
-1.11540578e-01 3.40458006e-01 1.48419893e+00 -3.71745586e-01
-4.20101374e-01 -1.70015708e-01 4.33233567e-02 -1.41806975e-01
4.09773141e-01 -2.34700851e-02 -9.00152266e-01 -2.57534385e-01
-7.25251855e-03 1.09584458e-01 9.27064180e-01 1.00131679e+00
1.01491690e+00 -5.86709797e-01 -5.00523984e-01 5.82650363e-01
1.19755256e+00 1.31141260e-01 4.76845950e-01 6.62610168e-03
7.84306407e-01 6.45598054e-01 3.34855556e-01 4.91892487e-01
5.77456176e-01 6.36131585e-01 6.81818604e-01 -4.25163954e-01
2.45688006e-01 -4.88516778e-01 6.31209239e-02 2.29020670e-01
-2.34247502e-02 -4.89492416e-01 -8.20021629e-01 7.69766390e-01
-1.86290359e+00 -1.03275037e+00 -3.50762248e-01 2.36379671e+00
8.05313230e-01 -3.58639210e-01 9.46339518e-02 -9.93054360e-03
7.38290846e-01 -2.25547105e-01 -6.70527101e-01 -3.00649226e-01
-2.08828617e-02 9.96545702e-02 5.16003013e-01 5.05616248e-01
-1.05900252e+00 4.80417311e-01 6.67939234e+00 2.57478386e-01
-1.03747690e+00 2.30473623e-01 5.34387827e-01 -6.24170490e-02
-5.07050097e-01 1.91210255e-01 -6.38128519e-01 5.39662719e-01
8.40559840e-01 -4.67919447e-02 2.15439215e-01 3.11406970e-01
6.81256533e-01 1.05655625e-01 -1.45357645e+00 5.72724819e-01
-5.61326087e-01 -1.24250603e+00 -1.78757478e-02 1.98096246e-01
8.32078755e-01 -7.59424418e-02 1.61162749e-01 2.71171510e-01
6.76841855e-01 -1.24563658e+00 4.54898059e-01 5.97103298e-01
7.03683794e-01 -6.82563841e-01 6.67852938e-01 4.73435849e-01
-6.56878352e-01 -3.59034836e-01 -4.18431252e-01 -4.03066397e-01
-7.52847223e-03 6.78362370e-01 -7.83636808e-01 3.61725688e-01
4.73597258e-01 5.60234308e-01 2.45522577e-02 6.49182260e-01
-4.90630567e-01 9.19445217e-01 -2.41181359e-01 2.02633470e-01
9.51409787e-02 -1.19398326e-01 3.71915936e-01 8.61557305e-01
5.18120885e-01 2.52893507e-01 -4.38789204e-02 1.10800278e+00
-2.09936380e-01 -1.14950776e-01 -8.04380476e-01 4.00318876e-02
5.41812479e-01 5.69293559e-01 -1.85521185e-01 -3.61473203e-01
-4.08557922e-01 5.75892091e-01 3.91036600e-01 7.08434701e-01
-9.90115345e-01 3.29532325e-01 8.62225890e-01 2.33843848e-02
2.43554473e-01 1.58378273e-01 -2.22826019e-01 -9.57136273e-01
-2.53517747e-01 -9.97104228e-01 6.87301338e-01 -6.05440915e-01
-1.37249565e+00 -6.32045567e-02 1.35672629e-01 -9.21227694e-01
-4.05158937e-01 -6.36151507e-02 -6.44748151e-01 1.16371608e+00
-1.19937944e+00 -9.49968755e-01 -3.46129872e-02 7.01037169e-01
1.88228056e-01 1.17179587e-01 9.82097447e-01 2.00107798e-01
-8.30858886e-01 3.41693759e-01 4.34688926e-01 1.27565444e-01
5.98805785e-01 -1.15664947e+00 1.14130355e-01 5.80283344e-01
-4.18871604e-02 9.54574764e-01 8.41953397e-01 -8.22644353e-01
-1.41911304e+00 -1.15173852e+00 8.60541344e-01 -2.95238346e-01
5.07140875e-01 -4.46450293e-01 -9.76612091e-01 1.01781225e+00
-3.73516567e-02 -8.05592611e-02 6.74750805e-01 5.92337906e-01
-3.11162889e-01 6.62433431e-02 -9.29597855e-01 5.95356822e-01
8.77384841e-01 -3.47915113e-01 -7.70177960e-01 3.25641572e-01
6.96703315e-01 -2.20657080e-01 -7.07690001e-01 5.35510123e-01
5.12808442e-01 -5.66571593e-01 1.03865469e+00 -1.14601898e+00
7.54698277e-01 -1.97673202e-01 -6.29382804e-02 -1.48691332e+00
-5.23535967e-01 -2.97643602e-01 -1.31946029e-02 1.25361919e+00
2.77824849e-01 -6.75607920e-01 4.24996883e-01 6.46080852e-01
3.81036311e-01 -3.92652690e-01 -1.15425622e+00 -5.06011844e-01
3.41789693e-01 -2.39212587e-01 1.02822065e+00 1.28424871e+00
-3.20567608e-01 5.25946498e-01 -6.06159925e-01 7.65736639e-01
5.89994490e-01 4.46536005e-01 8.51809084e-01 -1.32647574e+00
-4.71396416e-01 -3.99073325e-02 -1.40567675e-01 -6.98026478e-01
4.21669096e-01 -8.22204828e-01 -3.79570872e-02 -1.71624756e+00
4.86151814e-01 -6.08298123e-01 -3.51755470e-01 4.43899572e-01
-7.22992778e-01 -3.10314327e-01 -6.78348765e-02 2.01054722e-01
1.72661796e-01 7.52957284e-01 1.20405412e+00 -2.77179748e-01
-3.59551609e-01 1.30572304e-01 -8.40273559e-01 7.33194709e-01
6.41337335e-01 -9.62835431e-01 -5.68185031e-01 -4.96882051e-01
2.30690509e-01 5.12736082e-01 7.65223444e-01 -2.31275856e-01
-6.81578368e-02 -5.30070961e-01 3.83704841e-01 -2.66908914e-01
2.37543836e-01 -7.40509927e-01 5.61440051e-01 4.71146673e-01
-4.88392681e-01 1.44943938e-01 5.63341565e-02 8.19084764e-01
2.76951641e-02 -1.37249455e-01 4.47589308e-01 5.78971766e-02
-2.14029685e-01 2.37540022e-01 -2.58031994e-01 8.22949186e-02
8.07854772e-01 3.31767380e-01 -2.21413389e-01 -5.00824094e-01
-7.47778833e-01 3.75743657e-01 1.03166088e-01 4.51126337e-01
5.28496385e-01 -1.24021351e+00 -8.79944801e-01 2.71810085e-01
-1.49974367e-02 -6.58485293e-02 2.36091286e-01 7.58797288e-01
2.47690067e-01 3.09071898e-01 3.43045890e-01 -5.77227712e-01
-1.10569561e+00 1.04142570e+00 3.98855597e-01 -1.99282542e-01
-6.40889466e-01 6.38394773e-01 1.11177337e+00 -2.87477523e-01
7.03981742e-02 -2.72776365e-01 -1.91897631e-01 1.04046524e-01
3.60090435e-01 3.31647843e-01 -2.31331810e-01 -4.14607018e-01
-2.50469238e-01 1.71903178e-01 3.45369935e-01 6.70128874e-03
1.49542463e+00 -7.00943246e-02 -4.21506278e-02 6.64410830e-01
1.21999621e+00 -5.01027405e-02 -1.18681276e+00 -2.20423236e-01
-2.81060249e-01 -4.17731106e-01 1.95615575e-01 -5.06848454e-01
-1.08609486e+00 9.97766793e-01 7.39299297e-01 -4.76549752e-02
9.69215274e-01 3.45024839e-02 2.26648808e-01 2.59639360e-02
1.75580271e-02 -5.53179085e-01 -3.13095212e-01 5.89367747e-02
9.43518639e-01 -1.38689709e+00 -1.01810724e-01 -2.91196734e-01
-3.70911360e-01 6.02234244e-01 1.51149943e-01 -1.12919129e-01
6.49255157e-01 1.62889749e-01 -2.40728483e-02 -4.05025095e-01
-7.92322159e-01 6.60847872e-03 2.66280949e-01 3.94644767e-01
5.03016412e-01 4.03755546e-01 -2.47704729e-01 5.37970126e-01
-6.69587031e-02 9.84058157e-02 4.20728892e-01 4.29140925e-01
1.53664425e-01 -9.39908803e-01 -6.65981948e-01 5.23835301e-01
-6.05116069e-01 -7.15465173e-02 -3.34738530e-02 9.87752140e-01
2.14245185e-01 1.12789106e+00 1.92921132e-01 3.07197422e-01
4.52724725e-01 -9.95644927e-02 1.94392398e-01 -5.54902017e-01
-4.44628417e-01 1.66656494e-01 -2.72657394e-01 -3.43873769e-01
-5.69606125e-01 -6.88078225e-01 -1.18624794e+00 -3.72259557e-01
-4.31321234e-01 1.70819357e-01 3.68238539e-01 9.93646145e-01
3.24852198e-01 7.69014537e-01 7.38486111e-01 -2.56953359e-01
-7.27523208e-01 -8.08965564e-01 -4.08857256e-01 5.58464468e-01
8.71881723e-01 -1.13103199e+00 -4.10435170e-01 -3.23879011e-02] | [8.000360488891602, 5.358336448669434] |
2526b735-7b87-4669-99bf-e35a563d721a | graph-representation-learning-for-audio-music | 1910.11117 | null | https://arxiv.org/abs/1910.11117v1 | https://arxiv.org/pdf/1910.11117v1.pdf | Graph Representation learning for Audio & Music genre Classification | Music genre is arguably one of the most important and discriminative information for music and audio content. Visual representation based approaches have been explored on spectrograms for music genre classification. However, lack of quality data and augmentation techniques makes it difficult to employ deep learning techniques successfully. We discuss the application of graph neural networks on such task due to their strong inductive bias, and show that combination of CNN and GNN is able to achieve state-of-the-art results on GTZAN, and AudioSet (Imbalanced Music) datasets. We also discuss the role of Siamese Neural Networks as an analogous to GNN for learning edge similarity weights. Furthermore, we also perform visual analysis to understand the field-of-view of our model into the spectrogram based on genre labels. | ['Vasudev Singh', 'Shubham Dokania'] | 2019-10-23 | null | null | null | null | ['genre-classification'] | ['computer-vision'] | [ 9.82437357e-02 -2.09790319e-01 -8.40723068e-02 1.70441747e-01
-5.73731542e-01 -6.94633067e-01 3.31221282e-01 1.21167056e-01
1.09059252e-02 1.59978986e-01 4.80971605e-01 -1.40020490e-01
-5.38611948e-01 -7.09279358e-01 -4.05542344e-01 -5.35950482e-01
-5.45674384e-01 2.78391242e-01 -4.12848502e-01 -3.20157737e-01
4.19709414e-01 4.50848818e-01 -1.58808815e+00 6.14426911e-01
2.21669704e-01 1.06509876e+00 -2.01205119e-01 6.95657015e-01
-1.77198723e-01 6.87010169e-01 -8.87625515e-01 -2.91306704e-01
1.98824227e-01 -6.27261043e-01 -6.54727340e-01 2.44007744e-02
9.19990897e-01 4.40452427e-01 -6.58930182e-01 1.06254375e+00
7.79424667e-01 4.98150676e-01 9.17984664e-01 -1.46684349e+00
-9.12323892e-01 1.05379069e+00 -8.22946906e-01 4.63844627e-01
4.88395244e-01 1.24989428e-01 1.56278610e+00 -5.29502809e-01
7.33157098e-01 1.17683673e+00 9.52044070e-01 2.02305883e-01
-1.27710617e+00 -9.05254185e-01 4.30522375e-02 6.87572420e-01
-1.21695507e+00 -2.05574885e-01 1.33159602e+00 -3.79322380e-01
1.00493252e+00 2.97953486e-01 1.15491378e+00 1.42851460e+00
-2.59404898e-01 7.83726335e-01 8.49993646e-01 -5.83895147e-01
-4.10166271e-02 -3.75896126e-01 -3.90419662e-02 6.07604742e-01
-1.55031001e-02 -6.78085834e-02 -1.17572582e+00 -7.10327625e-02
8.79878938e-01 -3.18823099e-01 -3.58332932e-01 -5.43573558e-01
-1.10948813e+00 8.63767982e-01 7.57655025e-01 5.37337244e-01
8.22928846e-02 6.55776203e-01 7.90033340e-01 4.97348964e-01
6.36895120e-01 9.25307155e-01 6.57961443e-02 -2.55647898e-01
-1.12508261e+00 7.46548623e-02 5.87196469e-01 5.04628956e-01
1.23457991e-01 7.06652045e-01 -1.47130102e-01 9.80715811e-01
1.07804902e-01 -9.14681405e-02 4.09134120e-01 -8.26767147e-01
4.66459632e-01 5.35352945e-01 -8.77629161e-01 -1.26560462e+00
-6.45667017e-01 -1.04236543e+00 -9.60425675e-01 4.59935606e-01
5.75348079e-01 3.45033109e-01 -8.68842661e-01 1.53313816e+00
-1.75203577e-01 4.91310835e-01 -4.63419050e-01 1.01640868e+00
1.02089500e+00 2.26621225e-01 -1.06763266e-01 2.39466578e-01
1.32958710e+00 -7.73157775e-01 -5.89208245e-01 3.18813883e-02
3.47391546e-01 -6.31161809e-01 1.40509057e+00 7.37373888e-01
-9.56533611e-01 -7.21306562e-01 -1.18083978e+00 -1.69252291e-01
-4.77873623e-01 3.38967353e-01 8.96936595e-01 5.42561054e-01
-9.20139372e-01 1.22995186e+00 -5.31187773e-01 -3.47481310e-01
8.13707411e-01 3.13641310e-01 -2.79833943e-01 4.18972671e-01
-9.97207761e-01 4.18280423e-01 3.29581529e-01 -1.53737962e-01
-8.44602048e-01 -7.63020337e-01 -8.58159423e-01 2.58256823e-01
-4.21412401e-02 -7.04425573e-01 8.46893728e-01 -9.95817900e-01
-1.29990661e+00 9.40494597e-01 6.16300106e-01 -4.20249879e-01
2.24782079e-01 1.26665339e-01 -5.94439149e-01 3.74821663e-01
-2.01248929e-01 6.20642960e-01 9.75039542e-01 -9.01487410e-01
-3.02413106e-01 -4.63219911e-01 2.82152057e-01 3.78938109e-01
-7.24870741e-01 -2.99636900e-01 -3.09675217e-01 -1.13342690e+00
1.27782270e-01 -8.22805822e-01 2.42758945e-01 2.03345194e-02
-4.65340197e-01 -2.76783884e-01 8.14801931e-01 -5.67702830e-01
1.41877913e+00 -2.41498590e+00 3.93820167e-01 3.74308586e-01
4.21925306e-01 -2.18933269e-01 -4.22167629e-01 6.54810607e-01
-4.43904519e-01 2.00440526e-01 1.33604556e-01 -2.58531272e-01
1.65937722e-01 -2.16932461e-01 -1.90131709e-01 6.31659210e-01
-5.35357147e-02 8.04178119e-01 -7.99401283e-01 -1.43791586e-01
1.43749729e-01 6.00252509e-01 -5.35040200e-01 -2.28487924e-01
-1.06578037e-01 4.50886875e-01 7.32819661e-02 6.65854037e-01
2.28052303e-01 -2.98639864e-01 3.78409207e-01 -6.20887220e-01
2.22019657e-01 6.34621084e-01 -1.21282876e+00 2.32286978e+00
-4.24749970e-01 1.27290332e+00 -2.04937145e-01 -1.16131806e+00
7.47714758e-01 3.05776805e-01 4.32142019e-01 -7.49757588e-01
1.87765360e-01 -1.14972837e-01 3.40634644e-01 -2.66090035e-01
4.98495787e-01 -2.08705202e-01 -3.54706869e-02 5.23549676e-01
6.04408681e-01 1.36668399e-01 2.23975211e-01 2.34166622e-01
9.03867900e-01 3.33136797e-01 1.61174104e-01 -1.05252005e-01
1.08652517e-01 -1.49224401e-01 5.28889783e-02 6.04321718e-01
4.17784639e-02 8.45590889e-01 6.55256450e-01 -3.44655216e-01
-1.03720891e+00 -6.90575540e-01 1.40717709e-02 1.50187254e+00
-1.06121466e-01 -7.36396730e-01 -4.25616562e-01 -5.70455194e-01
1.01620644e-01 4.99270886e-01 -8.83782625e-01 -2.77565449e-01
-3.38905334e-01 -4.28525686e-01 1.01910293e+00 6.87585711e-01
4.39767390e-02 -1.28076446e+00 -1.24203928e-01 6.70852810e-02
-5.99976629e-02 -7.76841104e-01 -2.73944974e-01 2.71067053e-01
-8.52819264e-01 -1.23211563e+00 -6.02960825e-01 -7.02391028e-01
-3.48855145e-02 1.62387758e-01 1.47757983e+00 -6.69774935e-02
-6.07808590e-01 5.10496497e-01 -3.86557162e-01 -4.15032744e-01
-1.19124964e-01 3.04650038e-01 -4.16430905e-02 -7.07176998e-02
2.40708351e-01 -1.19762897e+00 -8.84290338e-01 1.96615569e-02
-7.14227498e-01 -1.00936303e-02 3.68384756e-02 6.13722205e-01
5.25374353e-01 -1.34573385e-01 6.43876910e-01 -7.52778053e-01
9.54732299e-01 -2.95982689e-01 -2.94828899e-02 -8.00323337e-02
-3.98205042e-01 -3.20843995e-01 5.06318808e-01 -6.54582024e-01
-2.18426898e-01 -2.47227639e-01 3.39233242e-02 -9.45751607e-01
-1.39114514e-01 4.80262220e-01 2.53456056e-01 -2.17680603e-01
9.02129471e-01 -1.73669830e-01 -2.98010528e-01 -7.52902508e-01
5.62037349e-01 5.58065116e-01 6.20112419e-01 -3.78980905e-01
5.90238035e-01 5.90527117e-01 2.31356025e-01 -7.68765569e-01
-7.83954084e-01 -5.91298997e-01 -5.36138475e-01 -5.37121594e-01
9.50647175e-01 -8.36678445e-01 -1.16984630e+00 5.30006029e-02
-6.50895596e-01 -1.48430169e-01 -6.49503827e-01 5.11849582e-01
-7.90151715e-01 3.19800705e-01 -7.39156723e-01 -8.30036700e-01
-4.13546622e-01 -5.36817849e-01 1.03873825e+00 2.91823968e-02
-4.11811888e-01 -1.12166691e+00 1.77584752e-01 3.24391931e-01
5.09876311e-02 4.24042910e-01 1.28731287e+00 -6.99925005e-01
-2.69853979e-01 6.64607584e-02 -1.90008610e-01 4.25963551e-02
-1.09279223e-01 -6.41575605e-02 -1.54099464e+00 -2.96844155e-01
-5.06845295e-01 -5.54496646e-01 1.21136236e+00 5.14581501e-01
1.62330651e+00 1.91682532e-01 -6.03605136e-02 9.39677358e-01
1.22159147e+00 -7.39158615e-02 5.88260829e-01 4.67297465e-01
1.16091573e+00 3.95006448e-01 1.26576930e-01 3.82946610e-01
-2.56092716e-02 8.36369395e-01 6.26481831e-01 -2.64010787e-01
-8.51154685e-01 -5.33183992e-01 1.24936730e-01 9.94600236e-01
-3.68066788e-01 -3.69050533e-01 -7.22855210e-01 4.56782848e-01
-1.82125330e+00 -1.00969040e+00 -1.26819789e-01 1.89802194e+00
3.51233721e-01 -1.81484688e-02 6.05043948e-01 8.39425862e-01
6.94424748e-01 3.95995885e-01 -3.99346411e-01 -4.16112959e-01
-2.46954173e-01 5.34012437e-01 1.16863828e-02 -2.79960930e-01
-1.29767275e+00 7.71617472e-01 6.71914339e+00 1.30461359e+00
-1.07368231e+00 -7.65550733e-02 3.54437858e-01 -4.02833670e-01
-2.43218780e-01 -2.70185858e-01 -5.81773520e-02 -1.61638800e-02
6.93091512e-01 1.37688845e-01 7.91182876e-01 5.41962564e-01
-1.41128510e-01 3.16174150e-01 -1.18670738e+00 1.40014350e+00
3.02044094e-01 -1.60119760e+00 1.82990208e-01 1.22707710e-01
5.24765372e-01 -8.85025710e-02 3.17842603e-01 2.80309558e-01
-9.41127315e-02 -1.27659035e+00 7.02631474e-01 2.43128419e-01
1.18141484e+00 -1.04425049e+00 1.87337786e-01 -2.88003653e-01
-1.52996612e+00 -1.20506398e-01 -3.59468609e-01 -2.63701081e-01
-1.06125802e-01 1.54982448e-01 -7.29599893e-01 7.52450347e-01
9.64804292e-01 1.28348303e+00 -6.46876156e-01 1.23724949e+00
-1.26040116e-01 8.19850564e-01 5.86888008e-03 1.55856326e-01
1.57898173e-01 -2.35864535e-01 7.87699759e-01 1.31130731e+00
3.81787360e-01 -5.59503376e-01 1.07467555e-01 7.82369614e-01
-3.86766255e-01 2.97211617e-01 -8.84617090e-01 -5.17599940e-01
-5.97349294e-02 1.51994455e+00 -1.19729626e+00 -2.24502068e-02
-1.63132593e-01 6.74373031e-01 4.94262487e-01 4.02011156e-01
-4.82270658e-01 -4.61374611e-01 7.85950422e-01 1.51701644e-01
3.53100210e-01 -1.59621552e-01 -4.43734854e-01 -9.77108121e-01
-4.58814770e-01 -1.13516366e+00 7.77655900e-01 -1.12097633e+00
-1.55922771e+00 5.02599061e-01 -4.55328345e-01 -1.73910737e+00
-1.66371390e-01 -8.66070330e-01 -6.34906411e-01 5.04956543e-01
-1.05528128e+00 -1.22558796e+00 -1.99800730e-01 7.23841846e-01
5.50696313e-01 -5.73879778e-01 1.02209651e+00 5.24956942e-01
-2.69174665e-01 6.30567133e-01 -9.03288573e-02 3.36665660e-01
6.60387337e-01 -1.52877223e+00 5.51287234e-01 3.04059118e-01
1.42641544e+00 2.02196926e-01 5.26308477e-01 -4.04572397e-01
-1.23370671e+00 -7.51678884e-01 7.31184557e-02 -3.22595984e-01
9.85854805e-01 -6.45539641e-01 -7.79567361e-01 2.85416007e-01
5.03484845e-01 -3.64157110e-01 1.19089246e+00 7.95947373e-01
-6.99086726e-01 9.55867618e-02 -5.95895290e-01 6.04269505e-01
1.56016040e+00 -1.00513184e+00 -4.09668446e-01 3.06842208e-01
3.36434424e-01 -3.47526014e-01 -7.60196745e-01 3.00302058e-01
6.52543843e-01 -9.75673020e-01 1.12578285e+00 -8.90847683e-01
5.25340319e-01 -2.79058874e-01 9.39337537e-02 -1.49747670e+00
-7.21861899e-01 -5.29202998e-01 -2.54297733e-01 1.11438382e+00
3.26753318e-01 3.98650579e-02 7.73445964e-01 -6.51645899e-01
-1.38609275e-01 -5.60219824e-01 -8.27705324e-01 -7.53302872e-01
-2.26364240e-01 -9.70579863e-01 2.47633070e-01 1.32797432e+00
2.53302723e-01 5.72271466e-01 -5.04820824e-01 -2.53681064e-01
3.33734751e-01 4.78905559e-01 6.35586739e-01 -1.43281937e+00
-6.05067849e-01 -7.97021747e-01 -9.12424266e-01 -2.96264559e-01
2.67553240e-01 -1.48207843e+00 -6.65039718e-01 -1.71788442e+00
8.48945677e-02 -5.27888499e-02 -8.52615952e-01 3.73305172e-01
3.31886888e-01 1.28172660e+00 5.17023444e-01 8.54688659e-02
-5.67012668e-01 2.89800376e-01 1.47357512e+00 -4.10398066e-01
-6.10394068e-02 -4.47392426e-02 -7.63500810e-01 8.18180263e-01
7.45074689e-01 -3.98239493e-01 -5.19904554e-01 -2.21338183e-01
7.61128366e-01 -1.39854893e-01 3.44350189e-01 -1.17376089e+00
5.36173023e-02 3.73078465e-01 7.73798347e-01 -3.76605123e-01
6.31719887e-01 -4.69619036e-01 9.10377726e-02 4.05237973e-02
-5.81087351e-01 4.22499515e-02 4.18222070e-01 8.06851029e-01
-4.69418854e-01 -2.67733578e-02 2.68054456e-01 -3.60882133e-02
-4.79012519e-01 2.17678696e-01 -1.92970321e-01 2.87751287e-01
2.25530401e-01 -4.37228173e-01 -1.48934841e-01 -8.49279881e-01
-1.09343886e+00 -2.64799029e-01 6.02283850e-02 6.88730240e-01
3.50470603e-01 -1.66746879e+00 -5.64735770e-01 4.90261689e-02
2.75693387e-01 -8.34130943e-01 4.13310438e-01 7.30474591e-01
-4.42286193e-01 5.79842553e-02 -4.81706738e-01 -6.59241438e-01
-1.52166319e+00 5.78317046e-01 1.98410422e-01 -2.27566913e-01
-8.68478477e-01 8.63835812e-01 2.82633066e-01 -6.32730424e-02
6.25220060e-01 -1.92656383e-01 -7.19928443e-01 5.74363589e-01
2.34225571e-01 5.28775752e-01 9.62959975e-02 -4.33316559e-01
-3.16543043e-01 8.25889826e-01 4.05022055e-01 -2.34862968e-01
1.47708583e+00 1.80868700e-01 2.01695859e-01 7.90439606e-01
9.71841037e-01 1.54087096e-01 -7.68005133e-01 2.24806517e-01
-6.55405447e-02 -4.01077330e-01 1.33178711e-01 -7.99107432e-01
-1.35308433e+00 1.40807974e+00 8.26780677e-01 5.70464432e-01
1.23290348e+00 -7.18371719e-02 3.47330630e-01 1.30845472e-01
-8.43163282e-02 -1.24050140e+00 3.88064682e-01 3.14341784e-01
1.03496981e+00 -8.16273212e-01 5.00642881e-02 -2.34342083e-01
-4.80545819e-01 1.43298888e+00 3.21049154e-01 -4.87393171e-01
5.63026547e-01 5.67622110e-03 2.68799037e-01 -5.57455242e-01
-4.31659281e-01 -4.92735982e-01 1.12107766e+00 8.05266917e-01
7.35019743e-01 9.79012772e-02 -5.99334687e-02 3.90807182e-01
-6.19727969e-01 -3.37915957e-01 1.85856447e-01 3.26873571e-01
6.31792322e-02 -1.01977420e+00 -6.95742518e-02 4.43103284e-01
-8.13007414e-01 -2.07989827e-01 -7.19985664e-01 9.43393171e-01
4.12519947e-02 6.94948256e-01 2.70007998e-01 -6.42596006e-01
3.06744426e-01 2.46708184e-01 8.38985860e-01 -5.98263025e-01
-1.12959146e+00 5.07265747e-01 2.32198030e-01 -4.51961249e-01
-4.91220087e-01 -2.30592951e-01 -9.90545869e-01 -1.14398107e-01
-2.76755214e-01 -7.36386701e-02 7.59639323e-01 6.42038286e-01
3.35060328e-01 1.10333490e+00 3.37764621e-01 -1.11878705e+00
4.61662747e-02 -1.19030321e+00 -1.08248425e+00 7.71291137e-01
6.74795285e-02 -6.51071489e-01 -3.80903780e-01 -2.25738555e-01] | [15.729440689086914, 5.2683258056640625] |
a88422ee-ef49-47b7-a681-4260c8c1aa49 | data-integration-in-systems-genetics-and | 2207.03540 | null | https://arxiv.org/abs/2207.03540v1 | https://arxiv.org/pdf/2207.03540v1.pdf | Data integration in systems genetics and aging research | Human life expectancy has dramatically improved over the course of the last century. Although this reflects a global improvement in sanitation and medical care, this also implies that more people suffer from diseases that typically manifest later in life, like Alzheimer and atherosclerosis. Increasing healthspan by delaying or reverting the development of these age-related diseases has therefore become an urgent challenge in biomedical research. Research in this field is complicated by the multi-factorial nature of age-related diseases. They are rooted in complex physiological mechanisms impacted by heritable, environment and life-style factors that can be unique to each individual. Although technological advances in high-throughput biomolecular assays have enabled researchers to investigate individual physiology at the molecular level, integrating information about its different components, and accounting for individual variations remains a challenge. We are using a large collection of omics and phenotype data derived from the BXD mouse genetic diversity panel to explore how good data management practices, as fostered by the FAIR principles, paired with an explainable artificial intelligence framework, can provide solutions to decipher the complex roots of age-related diseases. These developments will help to propose innovative approaches to extend healthspan in the aging global population. | ['Johan Auwerx', 'Maroun Bou Sleiman', 'Alexis Rapin'] | 2022-07-07 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [ 4.38771933e-01 -3.54014039e-01 -3.00478071e-01 -1.33774072e-01
8.42374787e-02 -3.10604960e-01 -9.23011976e-04 5.73648751e-01
-4.65250880e-01 9.24241245e-01 2.04751685e-01 -2.85785139e-01
-4.91875619e-01 -4.14928585e-01 -4.02980506e-01 -5.13282895e-01
-3.54881287e-01 3.44506264e-01 -2.48698860e-01 -1.98818237e-01
2.91327368e-02 2.29510233e-01 -1.58448327e+00 -4.92116287e-02
1.31871998e+00 6.93990052e-01 1.96476743e-01 5.26898265e-01
-2.11617630e-02 -1.84521869e-01 -2.66560704e-01 -1.54351518e-01
-3.84082347e-02 -7.34624326e-01 -3.03672075e-01 -4.33196455e-01
8.78801569e-02 -1.57321230e-01 1.37595981e-01 6.75324023e-01
6.32562816e-01 -4.10830021e-01 3.90320003e-01 -7.83539474e-01
-6.98005378e-01 1.42725348e-01 -2.25191832e-01 -5.04708961e-02
1.99720532e-01 2.12570384e-01 6.51421309e-01 -3.60517770e-01
5.27030170e-01 1.26899981e+00 5.26618421e-01 6.87908411e-01
-1.55316961e+00 -2.46048480e-01 -6.90889508e-02 4.42570269e-01
-8.86759639e-01 -5.05650938e-01 2.46550933e-01 -6.81133389e-01
5.04763126e-01 4.92270470e-01 1.27148974e+00 6.50944352e-01
6.99169517e-01 -2.99800690e-02 9.99403536e-01 -3.16541910e-01
2.24556968e-01 -4.93070453e-01 -1.46517307e-01 5.13114691e-01
8.82427454e-01 9.13664252e-02 -5.17107368e-01 -1.91911921e-01
5.77442825e-01 1.65764078e-01 -3.97603363e-01 -3.04296225e-01
-1.49616838e+00 2.03989074e-01 1.42207921e-01 3.31748009e-01
-2.87805408e-01 2.42806926e-01 3.43191683e-01 2.67541140e-01
2.13757455e-01 7.39922583e-01 -1.00912857e+00 -2.11694062e-01
-6.11944616e-01 2.75712997e-01 5.56681514e-01 2.21176922e-01
4.24680501e-01 2.64162617e-03 2.27747172e-01 7.27797329e-01
5.56814253e-01 6.73526168e-01 5.03262758e-01 -1.33510005e+00
-2.10650533e-01 9.96691763e-01 -2.41901018e-02 -8.37927163e-01
-8.70140195e-01 -5.96907377e-01 -8.17152441e-01 4.23505634e-01
9.17407095e-01 -2.08158717e-01 -6.65078819e-01 1.98166203e+00
6.83077455e-01 -2.83720911e-01 -3.21712255e-01 8.09431970e-01
1.85736805e-01 2.77213484e-01 6.88982964e-01 -1.72066033e-01
1.67259371e+00 -8.32709670e-02 -3.75007302e-01 -2.58732259e-01
5.14356673e-01 -3.10065687e-01 6.60567045e-01 3.43738645e-01
-8.74718249e-01 -3.49752665e-01 -1.06655490e+00 -5.52194752e-02
-5.59781492e-01 -2.55406857e-01 7.50295699e-01 9.95100260e-01
-6.94140613e-01 8.97715747e-01 -1.04348862e+00 -9.55327213e-01
4.19936597e-01 3.32320839e-01 -3.94400626e-01 -3.47346365e-02
-1.12652552e+00 1.18204319e+00 9.33103636e-02 1.46481112e-01
-4.86144304e-01 -1.37381053e+00 -5.20863533e-01 -4.51016128e-02
3.22126448e-02 -1.35868561e+00 4.31799114e-01 -2.72558451e-01
-9.95055854e-01 5.89314818e-01 -2.50529256e-02 -1.09471701e-01
1.01073496e-01 -5.25854170e-01 -4.48653489e-01 8.12640637e-02
-1.08446151e-01 5.47065020e-01 1.34886056e-01 -5.76773643e-01
-3.44923407e-01 -1.06933820e+00 -5.75684130e-01 -1.00332305e-01
-1.85095072e-01 2.34986156e-01 3.79477441e-01 -4.42463011e-01
8.20852295e-02 -9.25765932e-01 -4.84730572e-01 6.76927269e-01
-1.13988556e-01 3.82411271e-01 5.66547573e-01 -9.08380449e-01
9.49144423e-01 -1.71242833e+00 6.12998903e-01 -2.94060260e-01
4.11503702e-01 4.72197622e-01 -1.65399257e-02 7.34544218e-01
4.72901762e-03 6.52009726e-01 -1.79723889e-01 2.94758826e-01
-1.03940643e-01 -1.36035830e-01 3.33313555e-01 5.21520793e-01
1.60106212e-01 8.05318713e-01 -8.50564778e-01 -2.98955720e-02
2.11367279e-01 6.32832170e-01 -3.94744217e-01 -2.15646863e-01
-4.69976664e-01 9.00411725e-01 -4.82519746e-01 8.81394982e-01
3.09901774e-01 -3.93570848e-02 5.72697997e-01 3.28117311e-02
-3.27793211e-01 -2.96375364e-01 -4.59175169e-01 1.80604100e+00
1.35158792e-01 8.84343535e-02 2.50353128e-01 -9.48369801e-01
5.85181355e-01 2.58021414e-01 5.95605135e-01 -8.76220345e-01
1.61180958e-01 3.25517863e-01 6.26144350e-01 -6.39838636e-01
-2.30771974e-01 -2.86384553e-01 2.15567961e-01 1.94986567e-01
-3.23452741e-01 3.77498031e-01 4.05347168e-01 -1.52997598e-01
1.16241050e+00 4.98700231e-01 2.78106749e-01 -4.31674600e-01
3.19147080e-01 -2.98875314e-03 9.76112604e-01 8.62935111e-02
-6.61013961e-01 3.24402243e-01 5.83176970e-01 -3.48252624e-01
-1.28295672e+00 -1.04331172e+00 -4.06071216e-01 7.12606609e-01
-2.42475927e-01 -2.61190265e-01 -4.31076497e-01 -3.82918827e-02
4.86372620e-01 3.95197511e-01 -6.25385821e-01 -5.24555802e-01
-2.82085299e-01 -1.14434028e+00 5.20339549e-01 8.67174193e-02
3.91563296e-01 -3.52186412e-01 -7.64061570e-01 5.56933641e-01
2.12859184e-01 -6.77274764e-01 1.14107408e-01 4.80171815e-02
-9.76666927e-01 -1.35521388e+00 -1.01642036e+00 -4.01272237e-01
6.00314200e-01 -1.46308035e-01 9.12380934e-01 4.14956599e-01
-1.07621062e+00 3.39250006e-02 -2.89621383e-01 -6.44845843e-01
-4.45503742e-01 -1.82877287e-01 2.14232504e-01 -4.51792687e-01
7.40243196e-02 -9.64635253e-01 -1.08580625e+00 2.67854214e-01
-6.33091152e-01 -1.94746740e-02 8.42172265e-01 4.36934173e-01
4.21037734e-01 -1.28754169e-01 1.19127464e+00 -5.92856526e-01
1.43220395e-01 -7.58312583e-01 -3.07428986e-01 3.80677700e-01
-8.38468432e-01 1.20016433e-01 3.93233031e-01 -3.05718988e-01
-8.18489194e-01 -1.71258882e-01 -7.67581910e-02 7.01241255e-01
-3.70136738e-01 5.61428308e-01 -4.71471965e-01 1.87926814e-01
4.11687642e-01 -6.29699081e-02 5.86347640e-01 -6.42973363e-01
2.09566072e-01 3.68536562e-01 2.52744794e-01 -6.75690055e-01
5.65843999e-01 3.24256182e-01 5.46155810e-01 -1.13038182e+00
-5.03425777e-01 -4.43613939e-02 -3.95592749e-01 -3.56797308e-01
1.17065966e+00 -7.39709556e-01 -8.24691057e-01 5.93757331e-01
-5.08321702e-01 -3.27167273e-01 1.26198113e-01 6.14731967e-01
-1.78003177e-01 1.60481393e-01 -3.69466186e-01 -5.98954380e-01
-5.29872596e-01 -7.42936075e-01 6.70584500e-01 2.99729288e-01
-6.68729603e-01 -8.99580002e-01 4.64365274e-01 4.67508197e-01
6.95437491e-01 7.84882307e-01 1.74548554e+00 -3.55605930e-01
-4.49176520e-01 5.57866432e-02 -9.35496669e-03 9.53080282e-02
4.23629493e-01 2.01450825e-01 -2.18114391e-01 1.76100343e-01
-3.17608804e-01 -6.26016781e-02 4.05697107e-01 4.70234573e-01
4.31552410e-01 6.13003671e-02 -3.35644513e-01 3.89607519e-01
1.29884577e+00 3.63341004e-01 7.83367276e-01 3.64182740e-01
4.89976585e-01 9.16733265e-01 4.03720945e-01 1.76151797e-01
4.27873462e-01 4.63724613e-01 3.52369457e-01 -7.53170177e-02
-9.85593870e-02 2.41588414e-01 1.33663237e-01 3.40717256e-01
-2.57581383e-01 1.70425087e-01 -7.52568901e-01 3.10543865e-01
-1.43432438e+00 -1.04973793e+00 -6.04801714e-01 2.43882012e+00
1.03854001e+00 -1.06481969e-01 3.56726438e-01 1.83044039e-02
4.39266801e-01 -4.32869673e-01 -9.35630977e-01 -3.85468811e-01
-3.66550803e-01 1.72856644e-01 2.27379352e-01 -1.26507208e-01
-3.14466536e-01 2.19402760e-01 7.13144112e+00 -2.09745288e-01
-9.99457419e-01 -5.13029814e-01 7.41800964e-01 -1.32416114e-02
-2.79759973e-01 9.54935327e-02 -6.33906603e-01 5.91845095e-01
1.30872607e+00 -3.64702046e-01 4.48060870e-01 3.34611386e-01
7.00889647e-01 -3.47247630e-01 -1.04057980e+00 2.09769726e-01
-3.88338238e-01 -1.00484192e+00 -4.09105510e-01 3.67531061e-01
2.94483155e-01 -1.13449655e-01 -2.26562601e-02 -7.12941736e-02
-1.49279743e-01 -9.04490530e-01 3.97121161e-01 9.91165340e-01
5.96441627e-01 -4.76132691e-01 2.70703703e-01 8.00937638e-02
-8.13260198e-01 5.51932938e-02 -7.28055984e-02 -2.08782926e-01
6.83837086e-02 1.18588066e+00 -5.47169387e-01 3.47723514e-01
5.53627193e-01 3.09467733e-01 -6.87292516e-01 1.37239683e+00
4.61232290e-02 3.93180192e-01 -1.76061109e-01 2.18653545e-01
-8.46216023e-01 -3.98588240e-01 3.89863908e-01 4.05200720e-01
5.61729789e-01 -2.00004667e-01 -3.82031530e-01 1.06870532e+00
4.44614649e-01 1.33512132e-02 -9.62164253e-02 -7.71719694e-01
3.84623885e-01 1.17661500e+00 -6.15454257e-01 -4.26886901e-02
-4.06965733e-01 6.24426305e-01 3.54468897e-02 2.19659314e-01
-4.45395261e-01 -2.81148046e-01 1.21835089e+00 3.56830001e-01
-1.35936156e-01 -5.59983194e-01 -1.87945560e-01 -8.38261545e-01
-1.81861371e-01 -1.08103979e+00 1.62310258e-01 -4.60072458e-01
-8.30020964e-01 -3.98907274e-01 -9.48238671e-02 -5.60859919e-01
5.09073660e-02 -4.17013675e-01 -3.57469767e-01 7.67698109e-01
-1.08292139e+00 -9.73481536e-01 7.21934512e-02 -2.78787434e-01
-4.89747338e-02 4.79778707e-01 1.08392429e+00 4.30148214e-01
-9.59785759e-01 3.58021520e-02 3.17464203e-01 -5.52951455e-01
8.40613127e-01 -1.23992121e+00 1.99706972e-01 5.30963540e-01
-6.97443485e-01 1.11942649e+00 8.43348503e-01 -1.09704900e+00
-1.70074511e+00 -8.60012472e-01 8.39455962e-01 -1.67849407e-01
5.17650187e-01 -2.38193125e-01 -7.31773317e-01 4.58131701e-01
-2.45094314e-01 -5.30482769e-01 1.23381412e+00 3.39497209e-01
-2.89422065e-01 -2.16360316e-01 -1.23838699e+00 8.28763664e-01
1.02740407e+00 1.07698338e-02 -2.21907780e-01 3.51314209e-02
6.45985126e-01 2.25348428e-01 -1.51293087e+00 5.01578569e-01
1.17545938e+00 -6.97423339e-01 1.02613425e+00 -6.99114263e-01
3.05228919e-01 -5.74663997e-01 6.93078339e-02 -1.27258921e+00
-2.72789061e-01 -3.92556190e-01 -6.60142154e-02 1.25843692e+00
3.36231500e-01 -7.15073526e-01 5.72905898e-01 1.11034155e+00
-1.66220054e-01 -8.11977565e-01 -8.13547134e-01 -7.24532247e-01
2.42304146e-01 1.13133296e-01 6.31626189e-01 5.15247285e-01
1.94807455e-01 -2.66851224e-02 -2.12803900e-01 -3.14980149e-01
8.14359903e-01 -2.73431927e-01 6.98252320e-01 -1.78276002e+00
-1.96599513e-02 -3.71889204e-01 -7.19455004e-01 2.58278847e-02
-5.30052364e-01 -5.50930321e-01 -3.77162188e-01 -1.58705831e+00
2.44131088e-01 -2.99438447e-01 -2.97932714e-01 6.83443546e-02
-2.33019277e-01 -9.48728994e-02 -4.57994267e-02 -2.02689230e-01
-1.87337577e-01 3.05005938e-01 1.17388594e+00 1.53203622e-01
-1.49537474e-01 -4.53466326e-01 -1.23072481e+00 4.67595190e-01
1.03682172e+00 -3.77115965e-01 -1.93032458e-01 -2.23614022e-01
6.26713991e-01 -2.23255921e-02 2.93822080e-01 -1.25966513e+00
-1.87587351e-01 -3.28788847e-01 8.19864750e-01 -1.79744974e-01
1.97697669e-01 -7.92195678e-01 7.62664676e-01 1.08880281e+00
-1.49419323e-01 -2.15053841e-01 1.16902135e-01 5.54374754e-01
6.11040413e-01 2.95590729e-01 8.08663785e-01 -1.50679186e-01
-2.58581638e-01 1.08757436e-01 -7.30647087e-01 -5.18477298e-02
1.16158831e+00 -1.45047069e-01 -6.93820238e-01 1.52314574e-01
-8.95396471e-01 2.20089749e-01 9.34436083e-01 3.01484972e-01
1.30812347e-01 -1.00486922e+00 -7.79820561e-01 -5.71374185e-02
4.06808034e-02 -5.93062341e-01 5.70565999e-01 1.16290331e+00
-6.31496787e-01 3.68607372e-01 -5.29713750e-01 -2.51887590e-01
-1.14231932e+00 5.78151941e-01 3.20318282e-01 5.98805211e-02
-3.61995757e-01 4.05505031e-01 -8.97076577e-02 -4.84825857e-02
-1.92263439e-01 -1.47777557e-01 -1.15734562e-01 8.57816264e-02
7.63182044e-01 8.03799570e-01 -2.55313963e-01 -1.91460967e-01
-3.24515522e-01 5.87539375e-01 1.50857260e-02 4.03797448e-01
1.49653304e+00 -4.75310713e-01 -4.72722828e-01 5.66238642e-01
7.20068157e-01 -9.36878994e-02 -7.21830845e-01 4.58760470e-01
9.78876650e-02 -1.67166471e-01 -3.38978678e-01 -1.12789965e+00
-8.69902253e-01 4.01913345e-01 9.13985729e-01 4.75416146e-02
9.86686707e-01 -1.38464600e-01 9.44060445e-01 1.24873132e-01
3.84779274e-01 -9.31409240e-01 -4.53255475e-01 -1.91097260e-01
8.42047095e-01 -7.47825980e-01 3.76510143e-01 -3.21790665e-01
2.18792021e-01 9.58253086e-01 4.17670518e-01 2.68603384e-01
5.48639357e-01 -8.46220553e-02 9.15177446e-03 5.04320748e-02
-8.86887491e-01 -1.65432855e-01 -6.60548881e-02 7.90224254e-01
7.42234051e-01 -5.24533130e-02 -1.15610993e+00 5.16470850e-01
5.34297787e-02 3.02491546e-01 1.93470165e-01 1.09317374e+00
-7.98423707e-01 -1.81512189e+00 -5.50702512e-01 8.05773139e-01
-7.57050574e-01 6.70387074e-02 -4.51874614e-01 7.97825158e-01
3.54790509e-01 9.67122138e-01 -2.15666935e-01 -7.28264526e-02
2.26454996e-02 4.43143368e-01 5.44847250e-01 -4.02714580e-01
-1.99784547e-01 -3.70042063e-02 4.02550280e-01 -4.68480140e-01
-2.99146324e-01 -1.11562324e+00 -1.40933383e+00 -3.79884958e-01
-1.04447622e-02 -1.25516430e-01 1.21923661e+00 8.46898675e-01
8.66114438e-01 7.60945559e-01 -4.41746116e-02 -4.35935915e-01
-3.47379565e-01 -6.06351495e-01 -6.82137489e-01 9.68779251e-02
1.63956694e-02 -5.02118111e-01 -2.56207466e-01 3.11747253e-01] | [6.6127777099609375, 5.464190483093262] |
a61e6015-23d7-4872-95f4-d9b9bedde31a | hetergraphlongsum-heterogeneous-graph-neural | null | null | https://aclanthology.org/2022.coling-1.545 | https://aclanthology.org/2022.coling-1.545.pdf | HeterGraphLongSum: Heterogeneous Graph Neural Network with Passage Aggregation for Extractive Long Document Summarization | Graph Neural Network (GNN)-based models have proven effective in various Natural Language Processing (NLP) tasks in recent years. Specifically, in the case of the Extractive Document Summarization (EDS) task, modeling documents under graph structure is able to analyze the complex relations between semantic units (e.g., word-to-word, word-to-sentence, sentence-to-sentence) and enrich sentence representations via valuable information from their neighbors. However, long-form document summarization using graph-based methods is still an open research issue. The main challenge is to represent long documents in a graph structure in an effective way. In this regard, this paper proposes a new heterogeneous graph neural network (HeterGNN) model to improve the performance of long document summarization (HeterGraphLongSum). Specifically, the main idea is to add the passage nodes into the heterogeneous graph structure of word and sentence nodes for enriching the final representation of sentences. In this regard, HeterGraphLongSum is designed with three types of semantic units such as word, sentence, and passage. Experiments on two benchmark datasets for long documents such as Pubmed and Arxiv indicate promising results of the proposed model for the extractive long document summarization problem. Especially, HeterGraphLongSum is able to achieve state-of-the-art performance without relying on any pre-trained language models (e.g., BERT). The source code is available for further exploitation on the Github. | ['Khac-Hoai Nam Bui', 'Ngoc-Dung Ngoc Nguyen', 'Tuan-Anh Phan'] | null | null | null | null | coling-2022-10 | ['document-summarization'] | ['natural-language-processing'] | [ 3.55833709e-01 5.71516395e-01 -2.10658893e-01 -8.63379799e-03
-4.81179625e-01 -3.21996361e-01 5.75395763e-01 1.05645549e+00
-3.39115620e-01 8.82778645e-01 8.23522508e-01 -1.31440714e-01
-1.52812868e-01 -9.88818586e-01 -5.32714903e-01 -4.91568297e-01
1.03863880e-01 2.75028229e-01 5.29785268e-02 -4.43959057e-01
5.18388987e-01 3.35937381e-01 -1.03268576e+00 2.85688967e-01
1.35426593e+00 5.47134876e-01 2.93818295e-01 6.31929278e-01
-8.44764292e-01 6.48580968e-01 -7.78201044e-01 -3.56807411e-01
-4.23582971e-01 -5.34078062e-01 -8.83533418e-01 -9.54444259e-02
3.70279998e-01 -2.10460704e-02 -3.95910382e-01 1.29241276e+00
6.33508384e-01 4.20856804e-01 6.08889818e-01 -6.19114935e-01
-6.14192367e-01 1.03211260e+00 -5.64547300e-01 2.12655097e-01
3.95408750e-01 -3.27312797e-01 1.18001199e+00 -6.33176506e-01
7.87308514e-01 1.35002053e+00 3.69065881e-01 3.40724915e-01
-8.88302326e-01 -1.79698333e-01 3.83770168e-01 2.03041002e-01
-9.95395780e-01 -1.34255290e-01 9.90384459e-01 -9.96449441e-02
1.24688220e+00 3.75907719e-01 6.12835348e-01 9.56900001e-01
5.40273726e-01 9.10449326e-01 3.70878518e-01 -3.08115751e-01
1.20513640e-01 -3.33921790e-01 9.17904794e-01 6.75788283e-01
7.94773042e-01 -7.78486192e-01 -3.26885462e-01 -3.72740179e-02
2.37079561e-01 -1.73779093e-02 -4.13744986e-01 1.23266958e-01
-1.03294647e+00 8.31473172e-01 7.49975622e-01 7.15191305e-01
-7.87540436e-01 -6.22619735e-03 8.45444083e-01 -3.21452580e-02
8.88326824e-01 8.40814769e-01 -5.27738780e-02 1.22753307e-01
-1.01228786e+00 1.90220147e-01 8.30708265e-01 8.47119629e-01
5.20244539e-01 2.44910613e-01 -7.25810885e-01 8.96719873e-01
-1.10157855e-01 6.67697787e-02 7.45729923e-01 -3.93514693e-01
8.32628787e-01 8.89778793e-01 -4.00401890e-01 -1.34132934e+00
-6.80578530e-01 -9.10651803e-01 -1.60269737e+00 -7.26365209e-01
-3.38808507e-01 -2.91260809e-01 -6.82912588e-01 1.34647989e+00
1.81706771e-01 1.28006637e-01 3.42788100e-01 4.18608159e-01
1.72360504e+00 1.13602197e+00 5.24177328e-02 -4.77078140e-01
1.41684914e+00 -1.13051653e+00 -1.04717016e+00 -2.90565878e-01
7.74297416e-01 -3.32689404e-01 6.76331401e-01 5.25847264e-02
-1.14213765e+00 -5.95671296e-01 -1.01128137e+00 -3.14058363e-01
-5.91904163e-01 7.79812410e-02 5.28256178e-01 5.22103831e-02
-1.19497406e+00 6.49758995e-01 -6.70526922e-01 -7.77122080e-01
4.22753662e-01 1.87534615e-01 -3.90107900e-01 -7.04439506e-02
-1.28952861e+00 7.19454765e-01 1.18795240e+00 1.79655075e-01
-1.09648801e-01 -4.79231626e-01 -1.08278906e+00 5.72050214e-01
4.16610599e-01 -1.21236551e+00 8.59228134e-01 -5.05430281e-01
-1.18145919e+00 4.63653862e-01 -2.59173691e-01 -6.83958828e-01
-3.09585147e-02 -4.03748378e-02 -1.98079422e-01 4.69427228e-01
5.50551293e-03 3.99736911e-01 3.99451166e-01 -1.05860007e+00
-3.67499232e-01 -5.66765964e-01 9.04691666e-02 5.25158286e-01
-8.09335947e-01 -1.92797095e-01 -5.02578795e-01 -7.65006185e-01
2.99153361e-03 -4.58078980e-01 -2.16500416e-01 -9.81679201e-01
-8.50564003e-01 -7.88260460e-01 5.09883285e-01 -1.04041815e+00
1.72514665e+00 -1.72747219e+00 5.83749712e-01 -9.67317373e-02
5.68504095e-01 5.54089665e-01 -3.60814542e-01 8.97615016e-01
1.15440771e-01 2.42445588e-01 -3.46937239e-01 -4.51939940e-01
-1.51086152e-01 -3.43542881e-02 7.36397551e-03 -4.02335748e-02
8.68284032e-02 1.16059422e+00 -1.08705533e+00 -6.62566245e-01
-2.87506655e-02 2.60524213e-01 -3.92609358e-01 -2.59918743e-04
-2.51755565e-01 9.96172950e-02 -6.64124489e-01 2.66821891e-01
6.49128914e-01 -1.37807280e-01 1.50622681e-01 -3.21915954e-01
2.97816121e-03 2.57080048e-01 -6.82435215e-01 1.75254595e+00
-3.59182030e-01 4.75020111e-01 -1.50556803e-01 -1.33709800e+00
9.77584779e-01 1.23287544e-01 3.35197002e-01 -5.35011590e-01
1.82033896e-01 1.12131551e-01 -1.23684093e-01 -4.76331204e-01
1.18377399e+00 6.72494397e-02 -1.84290618e-01 2.26534605e-01
2.27409035e-01 -1.52665794e-01 7.98567474e-01 7.24805295e-01
1.09846663e+00 -3.60879004e-01 5.86846709e-01 -2.08305210e-01
7.83562005e-01 -2.94740759e-02 6.72784671e-02 7.50906050e-01
3.25592846e-01 3.94371599e-01 8.06490362e-01 5.73368445e-02
-7.26205230e-01 -6.75351501e-01 3.68940860e-01 7.16113985e-01
1.16027847e-01 -7.17317998e-01 -9.07827318e-01 -5.96268773e-01
-1.09138817e-01 1.11592150e+00 -3.86034310e-01 -5.80937445e-01
-5.90899706e-01 -5.58584988e-01 5.81393301e-01 4.23703730e-01
5.98307490e-01 -1.34499240e+00 3.22654694e-02 3.96219969e-01
-4.39653844e-01 -1.04273903e+00 -5.55999994e-01 -1.24689162e-01
-1.09907508e+00 -6.71079516e-01 -9.25873935e-01 -9.24616754e-01
7.46066928e-01 3.99718016e-01 1.01282966e+00 1.12287261e-01
5.94298802e-02 3.50671232e-01 -5.81661940e-01 -4.19586748e-01
-4.71364141e-01 7.10291922e-01 -2.88281173e-01 -1.09455414e-01
1.25541940e-01 -4.34743702e-01 -4.30715084e-01 -5.46512842e-01
-1.20206726e+00 1.87066510e-01 6.20124400e-01 7.17853248e-01
5.05736589e-01 1.46465540e-01 1.05921197e+00 -9.32041585e-01
1.50072956e+00 -5.76637506e-01 -7.17129782e-02 3.53556275e-01
-3.91917467e-01 6.04941063e-02 9.72953856e-01 -4.91186455e-02
-1.08769357e+00 -6.40317738e-01 -2.89268345e-01 1.18311398e-01
-2.17563361e-02 1.25448918e+00 -1.05414428e-01 4.04311508e-01
3.60327363e-01 6.94821179e-01 -1.05918229e-01 -4.50299650e-01
5.08298874e-01 7.82246888e-01 4.87698704e-01 -3.11047912e-01
3.35285097e-01 1.09979309e-01 1.76466003e-01 -1.32165658e+00
-9.21792805e-01 -7.56863713e-01 -4.60337013e-01 -9.63836908e-02
9.57226872e-01 -6.03581011e-01 -5.04477203e-01 3.63895118e-01
-1.52913165e+00 1.68776691e-01 -3.31816465e-01 3.35221887e-01
-2.15941474e-01 9.35230196e-01 -5.64530015e-01 -4.13638622e-01
-1.23954535e+00 -6.50160074e-01 1.02462280e+00 5.52319288e-01
-2.91591972e-01 -1.27987838e+00 4.42506699e-03 3.34734082e-01
2.42196515e-01 3.34092021e-01 1.22238815e+00 -1.02672100e+00
-1.58871949e-01 -5.33442795e-01 -4.02642369e-01 5.37333071e-01
2.20413357e-01 -3.23788285e-01 -2.77998745e-01 -3.30356091e-01
-2.06817508e-01 1.60124898e-02 1.45880008e+00 7.50716448e-01
1.22228277e+00 -4.78889793e-01 -2.69760877e-01 1.90926984e-01
1.25095558e+00 -8.53542536e-02 5.92739761e-01 6.65143430e-02
1.04215991e+00 6.36270583e-01 2.79546022e-01 4.35528934e-01
4.87232327e-01 1.96595371e-01 2.59695172e-01 1.13137625e-02
-3.22785437e-01 -2.85323858e-01 2.81560510e-01 1.28940046e+00
-4.82318327e-02 -8.81977975e-01 -6.73244238e-01 3.99959445e-01
-1.94516468e+00 -9.78970170e-01 -4.10562217e-01 1.86766052e+00
6.05385363e-01 1.14879519e-01 -2.18233362e-01 -5.58856912e-02
7.80011952e-01 6.98563457e-01 -3.25583071e-01 -7.58921325e-01
-2.45483875e-01 9.45423394e-02 3.50274816e-02 4.48086172e-01
-8.66979837e-01 1.19302011e+00 3.84265661e+00 1.08420050e+00
-7.30402291e-01 -2.43826762e-01 3.65382791e-01 1.80884480e-01
-4.10992742e-01 -2.81419843e-01 -8.47955167e-01 2.62291282e-01
7.76113391e-01 -6.96266890e-01 1.34917036e-01 3.80206943e-01
4.47953433e-01 -8.17399770e-02 -7.21793890e-01 8.29728723e-01
6.28039241e-01 -1.62948060e+00 7.97450423e-01 -2.06807271e-01
7.57361233e-01 -1.53866351e-01 -5.20655811e-01 5.74723423e-01
-2.15059206e-01 -6.42959118e-01 1.76558390e-01 6.17727995e-01
4.21249807e-01 -8.34007144e-01 1.18293118e+00 5.36506712e-01
-1.15976310e+00 1.12043813e-01 -6.28639936e-01 2.01793224e-01
2.31616780e-01 8.73001277e-01 -7.62293100e-01 1.48893750e+00
9.83340144e-02 1.13914907e+00 -7.36647248e-01 1.01819825e+00
-2.56787002e-01 5.23204029e-01 8.71411040e-02 -5.30969203e-01
5.53901315e-01 -3.68540794e-01 9.71265316e-01 1.48297262e+00
4.75542188e-01 1.17893457e-01 2.21321583e-01 6.13170743e-01
-5.89606166e-01 5.26925266e-01 -7.07824707e-01 -5.37240505e-01
5.51395584e-03 1.43120944e+00 -8.98834407e-01 -6.66505039e-01
2.14034002e-02 9.21148598e-01 4.01248842e-01 4.58640069e-01
-2.76727617e-01 -9.17764366e-01 -1.22338928e-01 -5.87099157e-02
3.94539833e-02 -1.82244480e-01 -2.08002493e-01 -1.21011722e+00
1.12063557e-01 -5.65183520e-01 6.48508132e-01 -7.83510268e-01
-1.05535746e+00 8.13951254e-01 1.66728631e-01 -8.57775211e-01
-9.48730856e-02 -2.14758739e-01 -7.92626083e-01 6.92739487e-01
-1.41021228e+00 -1.06812429e+00 -3.64454508e-01 2.90548295e-01
8.18024278e-01 -1.36660591e-01 6.81869090e-01 -1.85914082e-03
-6.92152917e-01 2.29570299e-01 2.74951547e-01 3.52276415e-02
2.94667959e-01 -1.25856793e+00 3.68308187e-01 8.06791663e-01
4.76753153e-02 6.01220548e-01 6.48493767e-01 -9.61826026e-01
-1.26901102e+00 -1.34595406e+00 1.23387408e+00 1.75818607e-01
6.05602801e-01 -1.27997473e-01 -9.77445900e-01 3.65024924e-01
7.03534186e-01 -6.54924929e-01 4.45746303e-01 1.98424682e-01
1.82335600e-01 -6.92635626e-02 -7.03213811e-01 6.49977863e-01
8.43344033e-01 -1.55971363e-01 -7.58773744e-01 6.39890671e-01
1.08403087e+00 -2.75056988e-01 -6.97608948e-01 2.82215476e-01
3.48606370e-02 -5.18691301e-01 7.30985224e-01 -7.26873577e-01
8.03674579e-01 -6.42580688e-02 2.90055633e-01 -1.87458766e+00
-2.86769837e-01 -3.46368104e-01 -4.21676725e-01 1.38116288e+00
2.34387338e-01 -8.35316539e-01 4.48423773e-01 -1.23970747e-01
-6.78163230e-01 -8.81196499e-01 -5.96405149e-01 -6.68531418e-01
-2.78053829e-03 1.25719393e-02 4.68676865e-01 6.50655508e-01
2.43428349e-01 9.00652587e-01 -6.90329298e-02 -1.53011307e-01
4.88972425e-01 1.59473076e-01 5.48521638e-01 -1.37996626e+00
1.43996835e-01 -8.26293230e-01 -3.84483755e-01 -9.40881610e-01
6.11625016e-01 -1.47038352e+00 -2.22626984e-01 -2.79402351e+00
3.87162536e-01 5.59576809e-01 -2.93314129e-01 1.59552500e-01
-6.16321325e-01 -4.27637935e-01 2.50827312e-01 -1.95682913e-01
-9.44950700e-01 9.67165172e-01 1.53972256e+00 -6.43868744e-01
-3.88982952e-01 3.27081867e-02 -1.02644908e+00 4.68400657e-01
9.87923920e-01 -2.79578596e-01 -4.49520111e-01 -3.38858426e-01
1.81860059e-01 1.30290687e-01 8.32254514e-02 -7.91275322e-01
4.22093004e-01 2.76167721e-01 5.53043233e-03 -9.72034216e-01
-3.04872054e-04 -2.54523665e-01 -4.52197075e-01 5.28596759e-01
-5.75322628e-01 -4.02341969e-02 3.18944812e-01 7.20581889e-01
-5.54830432e-01 -4.76965725e-01 3.46048057e-01 -1.88096657e-01
-4.63579446e-01 2.94208050e-01 -3.56661230e-01 1.72131479e-01
6.37786031e-01 -9.13569033e-02 -6.60347223e-01 -5.35592973e-01
-4.45770055e-01 5.28621733e-01 -1.50112465e-01 4.19890463e-01
8.18768263e-01 -1.04980612e+00 -1.15319788e+00 -3.56680006e-01
1.81410491e-01 2.24929810e-01 6.41781390e-01 7.66026258e-01
-5.28922737e-01 8.14152360e-01 1.21035010e-01 -2.43971467e-01
-1.31504333e+00 4.46923375e-01 -7.96606392e-02 -9.27572608e-01
-7.37547219e-01 5.43651223e-01 1.63435787e-01 -3.13126713e-01
9.79326144e-02 -4.86996591e-01 -9.14595425e-01 5.60196400e-01
5.30366421e-01 5.82852185e-01 1.68807998e-01 -5.38100243e-01
-2.12499090e-02 3.52060169e-01 -3.19773316e-01 3.87091756e-01
1.42965722e+00 -7.80867115e-02 -7.66493082e-01 2.42144689e-01
1.27825248e+00 -1.39980078e-01 -2.03192160e-01 -2.39327967e-01
1.98794737e-01 3.36442202e-01 1.64208546e-01 -3.51707220e-01
-8.37272644e-01 9.20866191e-01 -1.22245684e-01 5.30609131e-01
1.05317771e+00 7.04695843e-03 1.09292710e+00 7.95611739e-01
-1.06304310e-01 -1.08213353e+00 5.62052093e-02 7.29463160e-01
1.28921437e+00 -8.60036314e-01 2.71724969e-01 -3.09803396e-01
-4.45023894e-01 1.36022365e+00 4.07035410e-01 -6.65781647e-02
2.65208393e-01 -4.63536531e-01 -6.31423712e-01 -4.39224750e-01
-3.81041020e-01 -1.52454898e-01 6.20315015e-01 1.77980304e-01
4.54930067e-01 1.82550102e-01 -8.35018873e-01 6.97338223e-01
-1.48146823e-01 -2.30974689e-01 7.32216835e-01 5.19323587e-01
-6.03885472e-01 -7.57993817e-01 -1.03797473e-01 8.36351037e-01
-2.57247061e-01 -2.88878828e-01 -5.47902942e-01 4.58442658e-01
-3.99229765e-01 1.13564312e+00 -9.44580659e-02 -4.90046814e-02
4.97292042e-01 -2.02736035e-02 2.00462267e-01 -1.00795376e+00
-7.99629569e-01 4.93763853e-03 3.65114510e-01 -1.44903749e-01
-4.12974656e-01 -3.44415337e-01 -1.49593306e+00 -7.72082657e-02
-2.35355318e-01 3.98878068e-01 6.00051403e-01 8.68850887e-01
5.52525759e-01 1.29117942e+00 3.58017057e-01 -8.08351398e-01
-3.75518441e-01 -1.37043500e+00 -6.31344199e-01 3.04599434e-01
1.75070614e-01 -7.81477019e-02 -1.59062311e-01 -4.38422322e-01] | [12.645139694213867, 9.568184852600098] |
55fc2324-9ce3-48d5-9e69-6afe1777aa59 | road-damage-detection-based-on-unsupervised | 1910.04988 | null | https://arxiv.org/abs/1910.04988v1 | https://arxiv.org/pdf/1910.04988v1.pdf | Road Damage Detection Based on Unsupervised Disparity Map Segmentation | This paper presents a novel road damage detection algorithm based on unsupervised disparity map segmentation. Firstly, a disparity map is transformed by minimizing an energy function with respect to stereo rig roll angle and road disparity projection model. Instead of solving this energy minimization problem using non-linear optimization techniques, we directly find its numerical solution. The transformed disparity map is then segmented using Otus's thresholding method, and the damaged road areas can be extracted. The proposed algorithm requires no parameters when detecting road damage. The experimental results illustrate that our proposed algorithm performs both accurately and efficiently. The pixel-level road damage detection accuracy is approximately 97.56%. | ['Rui Fan', 'Ming Liu'] | 2019-10-11 | null | null | null | null | ['road-damage-detection'] | ['computer-vision'] | [ 4.24510568e-01 6.24466687e-04 1.05439387e-01 -1.65131122e-01
-5.19376636e-01 -5.72509281e-02 2.22277455e-02 -1.88514858e-01
-6.33519113e-01 6.68730140e-01 -1.10304639e-01 -3.68629366e-01
4.74579036e-01 -1.30766106e+00 -4.00103211e-01 -6.28842115e-01
4.83559340e-01 2.46477947e-02 6.14236832e-01 3.60784903e-02
1.02732122e+00 6.73554420e-01 -1.77477980e+00 -5.62513113e-01
1.48428214e+00 7.89305687e-01 3.02268744e-01 7.76774168e-01
-4.08735387e-02 3.60400558e-01 -3.64700615e-01 -1.55583888e-01
4.85461891e-01 -2.58248150e-01 -6.85059071e-01 5.02462983e-01
4.98769343e-01 -8.82281661e-01 -3.04049253e-01 1.32583046e+00
5.24139762e-01 2.98910439e-02 9.04981434e-01 -8.38492990e-01
-3.23467776e-02 -5.11783361e-01 -1.07117414e+00 2.87002504e-01
3.21898460e-01 6.52727857e-02 4.83426929e-01 -9.49852228e-01
4.76270616e-01 1.01034880e+00 4.83374536e-01 5.42194843e-02
-1.08110690e+00 -4.67825085e-01 -3.82345378e-01 2.42664039e-01
-1.55401897e+00 -2.11170137e-01 9.16550815e-01 -6.61571980e-01
9.02639270e-01 2.09829733e-01 6.59289420e-01 -1.41470522e-01
3.73838663e-01 5.79366684e-01 1.32922602e+00 -5.12556314e-01
2.69393809e-02 -1.90611407e-01 3.51773471e-01 9.75556016e-01
5.05641460e-01 4.14623320e-01 -4.88209873e-02 2.15267241e-01
8.51394475e-01 -2.13928878e-01 -2.74827808e-01 -1.54256403e-01
-7.32383251e-01 5.20639718e-01 2.67477363e-01 -1.89142793e-01
-3.80093992e-01 -1.68679561e-02 1.70104519e-01 2.21596919e-02
4.85103816e-01 -1.03747338e-01 3.90503407e-01 -8.29268172e-02
-1.03690374e+00 -1.02270253e-01 3.82994115e-01 4.87572283e-01
1.30889821e+00 1.38755053e-01 5.04201710e-01 9.05374050e-01
4.62999046e-01 8.23047042e-01 1.09231263e-01 -1.22105598e+00
5.05198300e-01 6.65701866e-01 5.47753647e-02 -1.33368599e+00
-1.33146316e-01 3.98532063e-01 -6.89405739e-01 9.80757236e-01
3.10244083e-01 -2.04984605e-01 -1.03562295e+00 7.83833027e-01
4.63767350e-01 1.25514185e-02 -3.54932174e-02 8.69163334e-01
5.30173302e-01 6.96887136e-01 -3.64796996e-01 -2.72760242e-01
8.99637103e-01 -6.23713374e-01 -7.50929236e-01 -2.20835909e-01
2.13122472e-01 -7.67873049e-01 1.03138411e+00 5.19503951e-01
-1.12621975e+00 -2.69693017e-01 -1.21597707e+00 -4.29804951e-01
-1.01714805e-01 8.02755356e-02 -2.12368183e-02 8.01294446e-01
-1.15403342e+00 1.48016900e-01 -5.16862869e-01 -2.83941180e-01
1.77143961e-01 3.11854661e-01 -5.52895628e-02 -4.79415320e-02
-8.41310859e-01 1.00456488e+00 2.77894348e-01 3.67212534e-01
-3.89803886e-01 -4.69190218e-02 -9.09716368e-01 -2.73730606e-01
3.59758325e-02 -5.69955945e-01 8.88791442e-01 -1.35253653e-01
-1.73525143e+00 1.32200229e+00 -6.36256039e-01 -1.16975710e-01
5.10427833e-01 -3.07929039e-01 -1.94147423e-01 5.11301398e-01
3.09991896e-01 4.68012333e-01 6.98544860e-01 -1.32599926e+00
-1.04186010e+00 -2.34716803e-01 -2.86307871e-01 5.77596903e-01
5.44792004e-02 4.96493503e-02 -4.46251929e-01 -1.97674617e-01
6.53231621e-01 -5.10453105e-01 -2.27873266e-01 2.01434344e-01
-7.92587101e-01 2.92372733e-01 9.59051430e-01 -9.06185150e-01
1.42150354e+00 -1.85509431e+00 -3.49645555e-01 6.29617214e-01
1.88280031e-01 1.48865461e-01 5.12651801e-01 2.03565717e-01
2.68684149e-01 3.33779119e-03 -8.81996036e-01 -1.51563194e-02
-4.38862383e-01 6.54325187e-02 -2.09087044e-01 6.51838064e-01
-1.84761211e-01 3.79003882e-01 -6.18662775e-01 -9.08606589e-01
7.46584773e-01 2.92250603e-01 -3.31754804e-01 1.67436991e-02
6.15254819e-01 1.45629808e-01 -4.06274021e-01 9.33777630e-01
1.14745784e+00 6.05122328e-01 -4.30749953e-01 -4.50777113e-02
-7.25559473e-01 1.20863624e-01 -1.36793375e+00 9.49678481e-01
-2.38876477e-01 1.09246993e+00 1.06361076e-01 -8.56118798e-01
1.49797440e+00 -1.56263247e-01 4.67059463e-01 -7.25022674e-01
1.41586512e-01 3.87020409e-01 -5.60013294e-01 -6.87499285e-01
9.00208473e-01 -3.39586377e-01 7.73007795e-02 4.26723361e-01
-8.30855370e-01 -5.95516443e-01 5.49112670e-02 -1.01592988e-01
7.65871644e-01 -1.33086532e-01 1.83384567e-01 -3.32280159e-01
7.50506997e-01 3.84473950e-01 5.13154089e-01 2.20048606e-01
-2.74205357e-01 7.70807743e-01 3.27042848e-01 -2.67807484e-01
-9.53891397e-01 -1.11002159e+00 -5.81392467e-01 1.31733373e-01
9.37478781e-01 2.98672795e-01 -1.03548110e+00 -5.87790087e-02
5.67072704e-02 5.13266742e-01 -1.97007909e-01 7.36880898e-02
-7.81700313e-01 -8.22234511e-01 4.46852207e-01 2.03508615e-01
1.15441418e+00 -8.61141086e-01 -9.11305189e-01 1.81856096e-01
-3.64988953e-01 -7.25352466e-01 -3.86014611e-01 -4.91520494e-01
-1.25325751e+00 -1.28755021e+00 -5.65817714e-01 -1.12675989e+00
9.68487859e-01 8.82905841e-01 7.45911598e-01 2.02388406e-01
-5.71431458e-01 2.41627604e-01 2.52498299e-01 -7.56799728e-02
-4.17196192e-02 -4.22420442e-01 -2.61034667e-01 -1.66552830e-02
3.48333925e-01 -5.11349916e-01 -9.24749136e-01 5.95667720e-01
-5.71464539e-01 4.73607183e-02 4.76176143e-01 4.82133746e-01
1.03477025e+00 4.10755694e-01 6.82588816e-02 -5.42964101e-01
6.66745603e-01 4.21342961e-02 -8.69191766e-01 -2.07960576e-01
-7.51278639e-01 -3.29974532e-01 -9.38283503e-02 2.07065672e-01
-1.34353471e+00 1.49843633e-01 -1.02421865e-01 1.69582553e-02
-1.90638825e-01 2.35383049e-01 -2.34184504e-01 -2.98094124e-01
5.96411586e-01 2.69898951e-01 2.23384440e-01 -3.45854521e-01
2.23236069e-01 8.30789924e-01 1.21319771e+00 -1.91129461e-01
1.02186298e+00 9.36554372e-01 2.50401869e-02 -1.21829116e+00
-1.33855462e-01 -6.92651272e-01 -8.46073627e-01 -7.72668242e-01
1.08771014e+00 -6.66952372e-01 -6.62502885e-01 1.15480733e+00
-1.01628053e+00 -1.55842990e-01 -1.00985497e-01 5.17952740e-01
-5.37187219e-01 9.43633854e-01 -6.80892587e-01 -9.32774663e-01
-3.87040585e-01 -9.54204440e-01 8.66680801e-01 3.80301416e-01
2.42632687e-01 -9.33578610e-01 2.48490229e-01 5.31935155e-01
3.22775543e-03 5.42546451e-01 5.75502276e-01 6.82400346e-01
-7.66421318e-01 -3.26670051e-01 -5.47900796e-01 3.35628331e-01
2.04659209e-01 4.39116120e-01 -9.58833456e-01 2.19455779e-01
2.64681578e-01 1.52593911e-01 1.03683019e+00 6.81214929e-01
8.19393635e-01 5.60154244e-02 -3.65893185e-01 8.33569407e-01
1.72163606e+00 2.86412984e-01 1.24383831e+00 7.23436594e-01
8.16478491e-01 8.16802144e-01 1.08602071e+00 4.59961966e-02
5.96778750e-01 3.69977355e-01 5.43998122e-01 -4.69528794e-01
-1.87786162e-01 -2.86724687e-01 3.18707228e-01 6.70767188e-01
-1.05489314e-01 1.15130275e-01 -1.16634786e+00 6.45906091e-01
-1.45594287e+00 -1.07011318e+00 -9.81804013e-01 2.32262158e+00
5.42784631e-01 1.70088977e-01 6.36272430e-02 5.17497897e-01
1.29974067e+00 -1.50523633e-01 -5.75786233e-01 -7.43429482e-01
-2.31664672e-01 -9.14513096e-02 9.20949042e-01 1.17585707e+00
-9.52504933e-01 9.55572128e-01 7.10193062e+00 5.17170787e-01
-1.07618237e+00 -2.93879330e-01 3.64990056e-01 3.63703847e-01
-2.90215254e-01 -9.64399148e-03 -6.52725458e-01 4.95213479e-01
1.86603755e-01 -4.23396111e-01 2.25580689e-02 4.19274390e-01
7.06435323e-01 -1.15060663e+00 -1.69466332e-01 1.01166916e+00
-5.73934130e-02 -7.95294940e-01 -2.47916698e-01 4.13861156e-01
7.29406953e-01 -2.08914489e-01 2.09447462e-02 -3.19477648e-01
1.47895366e-01 -7.54539251e-01 4.50419664e-01 6.86918139e-01
9.93827760e-01 -8.33116770e-01 5.44167519e-01 3.81416351e-01
-1.36199784e+00 1.18705519e-01 -6.66216969e-01 -2.99139440e-01
5.95848918e-01 9.11944509e-01 -8.00269246e-01 2.38491192e-01
5.89773059e-01 5.93107224e-01 -2.36975431e-01 1.43953681e+00
-5.61399102e-01 3.98964047e-01 -5.32889485e-01 4.28148270e-01
7.38828331e-02 -1.03331947e+00 8.05466294e-01 1.05800450e+00
4.62802649e-01 2.22680867e-01 -1.22800343e-01 6.51730537e-01
2.94746578e-01 2.79767364e-01 -8.43536437e-01 7.45004177e-01
4.14609671e-01 8.29171538e-01 -8.54634702e-01 -3.46619070e-01
-2.65437126e-01 9.32783246e-01 -2.99942661e-02 3.85676324e-01
-5.00690579e-01 -7.25861311e-01 4.60636795e-01 2.44405553e-01
-8.64140093e-02 -2.61930466e-01 -1.06359482e+00 -8.68194938e-01
1.85063198e-01 -3.01266257e-02 1.24383882e-01 -8.89651358e-01
-6.86285734e-01 1.57387242e-01 -7.78165609e-02 -1.53845716e+00
7.08629563e-02 -4.03571397e-01 -1.04354227e+00 1.06561494e+00
-1.72459793e+00 -5.26925981e-01 -5.82383275e-01 6.23475850e-01
3.81363869e-01 2.62848556e-01 2.14935958e-01 2.61317968e-01
-9.14541602e-01 2.32874781e-01 2.32960805e-01 -7.59695098e-02
4.68344867e-01 -1.27093339e+00 1.96164787e-01 1.41928172e+00
-8.45946670e-01 1.11102805e-01 6.36207819e-01 -9.29488659e-01
-6.99859262e-01 -9.03799772e-01 1.14850616e+00 6.44632578e-02
2.66884297e-01 1.89589828e-01 -1.03926611e+00 1.07039228e-01
-3.49609926e-02 -3.66201043e-01 1.97502524e-01 -6.27083778e-01
2.05209613e-01 -8.40095729e-02 -1.64158082e+00 5.51622152e-01
9.62144196e-01 -6.19426966e-01 -6.72034919e-01 3.90387848e-02
1.76532447e-01 -5.60191751e-01 -6.26716197e-01 4.31710124e-01
5.26429951e-01 -1.23085308e+00 1.00849903e+00 4.43172336e-01
3.57351035e-01 -7.42890000e-01 9.34944302e-02 -8.49098444e-01
-1.05687976e-03 -5.10627627e-01 3.72545689e-01 1.02069092e+00
2.55927384e-01 -8.77658248e-01 9.15111184e-01 6.24508858e-01
-4.13183391e-01 -5.61678648e-01 -9.14976656e-01 -5.51334560e-01
-2.09869355e-01 -3.55606824e-01 1.54063776e-01 6.77797318e-01
-6.21236041e-02 -4.48072217e-02 -1.67867213e-01 4.10684854e-01
9.54982996e-01 2.44739026e-01 8.08872521e-01 -1.17328954e+00
5.24124146e-01 -5.68163633e-01 -5.71539104e-01 -1.31083179e+00
-1.73849061e-01 -4.84492153e-01 4.72985983e-01 -1.92706096e+00
-2.93178074e-02 -2.77720928e-01 8.22835118e-02 1.21154994e-01
-4.19888973e-01 4.28036392e-01 -3.57893705e-01 5.65751135e-01
2.42448270e-01 4.19532239e-01 1.29279268e+00 3.40532735e-02
-4.81992602e-01 6.85629845e-02 -2.84442008e-01 9.70885158e-01
1.09541011e+00 -2.68992662e-01 -3.75748098e-01 -3.77085537e-01
-7.09567964e-02 3.27489898e-02 3.99373770e-01 -1.05325603e+00
3.04076254e-01 -1.93497360e-01 -1.01348765e-01 -1.13799810e+00
1.30896181e-01 -6.10202909e-01 -1.26194879e-01 6.41694605e-01
3.70101601e-01 -1.43440604e-01 7.71508887e-02 4.96938795e-01
-3.84849697e-01 -4.54811633e-01 1.08326352e+00 3.42545211e-01
-9.44167256e-01 -3.10395062e-02 -7.54496694e-01 -2.14505509e-01
1.14710355e+00 -1.29264975e+00 -4.09467489e-01 -2.41206333e-01
-4.49468434e-01 2.96854198e-01 1.05380356e+00 -2.60733932e-01
1.11527860e+00 -1.18133903e+00 -5.37108421e-01 3.18114936e-01
-1.18149132e-01 3.09049606e-01 2.63890684e-01 8.29651475e-01
-1.44302213e+00 -3.42301838e-02 -4.42602187e-01 -8.01798701e-01
-1.24811506e+00 8.37792158e-02 5.19048452e-01 1.38249239e-02
-9.15398002e-01 5.09218335e-01 1.26053214e-01 -1.58551902e-01
-1.98088437e-01 -1.01359919e-01 -3.34091485e-01 -5.32687223e-03
2.07708746e-01 1.16541481e+00 -2.04051882e-01 -9.28874731e-01
-1.68320060e-01 1.50471938e+00 4.25582290e-01 -5.25545478e-01
7.99708068e-01 -7.85223424e-01 -1.34851933e-01 7.11007416e-02
1.15569615e+00 1.99857920e-01 -1.26865101e+00 8.48513693e-02
-2.89990604e-01 -8.86418164e-01 4.45676595e-01 -2.72979617e-01
-1.14017451e+00 1.06118190e+00 7.46499538e-01 7.77507424e-02
1.46732962e+00 -5.56971371e-01 1.16053188e+00 1.37388378e-01
3.77048045e-01 -1.42060721e+00 -1.92047730e-01 3.89100194e-01
5.76198757e-01 -1.05966079e+00 2.67188400e-01 -8.62650096e-01
-4.29520965e-01 1.10146356e+00 6.38765037e-01 -4.54375654e-01
5.02652407e-01 2.42063016e-01 2.38176674e-01 -3.07636440e-01
1.55354768e-01 -5.46758056e-01 3.72504443e-03 6.86547399e-01
-1.77646220e-01 -1.24838702e-01 -6.82816565e-01 -4.46016908e-01
-3.51302445e-01 -1.56503066e-01 8.33534002e-01 9.12003696e-01
-1.39655173e+00 -7.70331085e-01 -7.35886455e-01 4.10858363e-01
-1.18411541e-01 1.36311710e-01 -2.77248621e-01 8.26942444e-01
5.22649474e-02 1.07305837e+00 2.40506023e-01 -5.26018202e-01
5.55404007e-01 -3.10011566e-01 8.18092376e-02 -4.31222051e-01
1.14243135e-01 6.91938624e-02 1.78842425e-01 -4.79178786e-01
-3.42868060e-01 -6.60759747e-01 -1.59093118e+00 -3.42384994e-01
-3.42789710e-01 -2.05870628e-01 6.11400247e-01 6.08889937e-01
-1.11719877e-01 -1.97097778e-01 1.10100985e+00 -7.74896979e-01
2.33888745e-01 -3.99846882e-01 -6.86214387e-01 1.19498800e-02
1.92454845e-01 -7.44078398e-01 -6.74413621e-01 1.73907250e-01] | [9.020442008972168, -2.3682498931884766] |
8792e741-37d9-4329-8c18-12fd21515d1f | hoi4d-a-4d-egocentric-dataset-for-category | 2203.01577 | null | https://arxiv.org/abs/2203.01577v3 | https://arxiv.org/pdf/2203.01577v3.pdf | HOI4D: A 4D Egocentric Dataset for Category-Level Human-Object Interaction | We present HOI4D, a large-scale 4D egocentric dataset with rich annotations, to catalyze the research of category-level human-object interaction. HOI4D consists of 2.4M RGB-D egocentric video frames over 4000 sequences collected by 4 participants interacting with 800 different object instances from 16 categories over 610 different indoor rooms. Frame-wise annotations for panoptic segmentation, motion segmentation, 3D hand pose, category-level object pose and hand action have also been provided, together with reconstructed object meshes and scene point clouds. With HOI4D, we establish three benchmarking tasks to promote category-level HOI from 4D visual signals including semantic segmentation of 4D dynamic point cloud sequences, category-level object pose tracking, and egocentric action segmentation with diverse interaction targets. In-depth analysis shows HOI4D poses great challenges to existing methods and produces great research opportunities. | ['Zhoujie Fu', 'Weikang Wan', 'Li Yi', 'He Wang', 'Boqiang Liang', 'Hao Shen', 'Kangbo Lyu', 'Che Jiang', 'Yun Liu', 'Yunze Liu'] | 2022-03-03 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Liu_HOI4D_A_4D_Egocentric_Dataset_for_Category-Level_Human-Object_Interaction_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Liu_HOI4D_A_4D_Egocentric_Dataset_for_Category-Level_Human-Object_Interaction_CVPR_2022_paper.pdf | cvpr-2022-1 | ['motion-segmentation'] | ['computer-vision'] | [-1.15504965e-01 -2.66012400e-01 6.54159114e-02 -2.62243629e-01
-3.39626193e-01 -7.42506325e-01 2.62035400e-01 -4.36439991e-01
-3.17761227e-02 3.50275859e-02 5.55730879e-01 5.38985014e-01
-1.21600211e-01 -3.95334154e-01 -8.13984513e-01 -3.55423301e-01
-1.98447794e-01 9.66808259e-01 3.67277950e-01 -2.36357730e-02
3.22825015e-01 6.88587427e-01 -1.67649543e+00 4.02779281e-01
1.99621946e-01 1.04402435e+00 3.70876849e-01 1.10264564e+00
2.15424240e-01 4.70712543e-01 -4.68946666e-01 3.88020501e-02
5.16475379e-01 -2.89319679e-02 -1.05246317e+00 5.69391847e-01
8.99426699e-01 -6.91945493e-01 -4.11031514e-01 8.58031929e-01
6.48638785e-01 5.05379498e-01 6.38121843e-01 -1.70427382e+00
-1.25181034e-01 1.70791134e-01 -5.80119014e-01 1.31895244e-01
9.55848992e-01 6.90749705e-01 7.53495336e-01 -1.04491568e+00
1.24542117e+00 1.76384687e+00 5.29124498e-01 6.64324522e-01
-1.14086103e+00 -5.48240483e-01 1.61383465e-01 1.19027719e-01
-1.25783587e+00 -8.04411620e-02 9.09598470e-01 -8.12632561e-01
1.26326263e+00 2.70405143e-01 1.11512458e+00 1.77474165e+00
-4.02785316e-02 1.37268329e+00 6.66732311e-01 1.65683180e-01
2.47140869e-01 -3.66122633e-01 4.57913168e-02 4.33955997e-01
-6.72852322e-02 -2.76436090e-01 -8.24804008e-01 1.01772837e-01
1.23420346e+00 1.91701874e-01 -1.10447116e-01 -1.06923950e+00
-1.92537522e+00 1.34489730e-01 4.31208819e-01 5.66922761e-02
-4.08040702e-01 5.90469182e-01 6.25415742e-01 -1.98406264e-01
4.38283324e-01 2.23519206e-01 -8.52063298e-01 -4.10707682e-01
-2.65021294e-01 8.39783788e-01 4.16088313e-01 1.70630419e+00
4.23699766e-01 -1.01394348e-01 -4.25324142e-01 4.66869324e-01
4.09282386e-01 7.17564046e-01 -8.97815973e-02 -1.86234331e+00
7.16108143e-01 8.33678544e-01 4.35947746e-01 -1.00991786e+00
-6.82271898e-01 9.32899863e-02 -3.76498431e-01 -2.00630650e-02
5.93146324e-01 1.26763269e-01 -7.95197189e-01 1.33243370e+00
8.72801125e-01 5.71939647e-02 -3.88365477e-01 1.41014099e+00
1.06133687e+00 2.46105343e-01 2.31920913e-01 4.06844527e-01
1.46737373e+00 -1.02159345e+00 -5.79694629e-01 1.37779610e-02
5.23603499e-01 -5.57968557e-01 1.26168871e+00 4.18374330e-01
-1.29349530e+00 -8.84206057e-01 -3.35896850e-01 -2.84220189e-01
-1.66017875e-01 -1.43571898e-01 7.38467276e-01 2.96610117e-01
-8.07080030e-01 3.82803887e-01 -7.12555766e-01 -5.75302720e-01
5.19422829e-01 2.33066857e-01 -6.43033862e-01 -5.44638671e-02
-5.92507899e-01 3.05404186e-01 4.93844628e-01 2.78520025e-02
-1.38574314e+00 -1.11015105e+00 -9.09423649e-01 -2.88666517e-01
5.77282786e-01 -9.19869065e-01 1.32384634e+00 -2.39750594e-01
-1.33277833e+00 1.30784130e+00 6.55306280e-02 3.78097109e-02
8.11019063e-01 -5.88382661e-01 1.54096901e-01 2.57448405e-01
2.89085418e-01 1.05383360e+00 7.44609892e-01 -1.34330940e+00
-7.20692575e-01 -9.62809622e-01 1.08419158e-01 6.05025530e-01
2.60823458e-01 -1.31048769e-01 -1.06497049e+00 -6.85183287e-01
2.59600341e-01 -1.18905914e+00 -1.16528168e-01 1.98115930e-02
-5.71151793e-01 -7.14002252e-01 1.13691258e+00 -4.55784261e-01
4.40453440e-01 -2.19200301e+00 7.75261641e-01 -1.48488447e-01
4.01490480e-01 -2.57258952e-01 2.55000014e-02 -1.11882493e-01
2.37837341e-02 -2.45255768e-01 1.19310662e-01 -6.23842597e-01
2.45652750e-01 9.13218632e-02 -5.66700846e-03 5.57661653e-01
-3.38473052e-01 1.35194170e+00 -1.31041384e+00 -5.75670838e-01
9.39866245e-01 3.86376649e-01 -9.38829780e-01 2.90764391e-01
-5.44984937e-01 9.85146761e-01 -6.36288345e-01 1.15000165e+00
6.36668921e-01 -1.85513198e-01 -2.40590721e-01 -4.90497112e-01
8.14701393e-02 -1.10366933e-01 -1.22039151e+00 2.54134512e+00
-5.89006394e-02 5.17586946e-01 7.15270638e-02 -5.34832776e-01
2.64587581e-01 2.79867649e-01 1.18857157e+00 -4.21772242e-01
5.77908695e-01 -1.56787291e-01 -5.75074136e-01 -6.96563423e-01
5.16345024e-01 6.48702085e-01 -4.65730369e-01 1.23176977e-01
2.59030193e-01 -7.26829052e-01 -1.29666671e-01 1.65859908e-01
8.59030128e-01 5.27886510e-01 -1.51356593e-01 -2.17234626e-01
2.15462014e-01 3.60599637e-01 2.23662928e-01 7.28495359e-01
-7.09527016e-01 8.84839058e-01 1.45488590e-01 -4.77322996e-01
-1.18853307e+00 -1.20458186e+00 -1.15246490e-01 1.13518965e+00
6.24010742e-01 -2.92364120e-01 -9.40034151e-01 -5.56120217e-01
1.67438850e-01 5.60110688e-01 -6.81834579e-01 3.65203857e-01
-5.55787086e-01 -1.92237228e-01 9.65094939e-02 8.68092477e-01
6.92779839e-01 -1.53089178e+00 -8.94034028e-01 7.84894973e-02
-5.99677622e-01 -1.57409346e+00 -6.97272599e-01 -1.22748315e-01
-8.78413916e-01 -1.41804743e+00 -1.04322505e+00 -5.58350444e-01
3.96706760e-01 6.71595871e-01 1.36732578e+00 -7.31050551e-01
-7.50783741e-01 1.42244494e+00 -4.01050568e-01 -3.35794449e-01
1.88617483e-01 -1.60287514e-01 4.01309758e-01 -2.63416439e-01
4.34268177e-01 -4.26738411e-01 -9.25010443e-01 8.52590680e-01
-1.10119253e-01 2.44758148e-02 -1.82516828e-01 3.23023950e-03
7.80390859e-01 -2.65963823e-01 -3.78316402e-01 -4.47484642e-01
-3.01070474e-02 -2.37404317e-01 -4.06492442e-01 -9.72001478e-02
3.18444043e-01 -6.53350413e-01 -1.45087406e-01 -2.74020404e-01
-1.00897408e+00 2.93981135e-01 2.49314293e-01 -1.14058816e+00
-5.42911828e-01 -6.46137655e-01 -4.91640449e-01 2.32478634e-01
7.78331697e-01 -1.87949196e-01 -3.41834486e-01 -4.79384810e-01
6.70159638e-01 3.58709097e-01 9.36641514e-01 -7.05184996e-01
6.05954289e-01 7.74735153e-01 -2.65519649e-01 -8.83922160e-01
-8.04191768e-01 -9.56595659e-01 -1.16145778e+00 -6.33533001e-01
1.68506634e+00 -1.14762866e+00 -1.49145341e+00 8.89916599e-01
-1.42986643e+00 -8.98199618e-01 -5.09620905e-01 4.47769731e-01
-1.23312604e+00 8.64891633e-02 -2.93988764e-01 -6.56490684e-01
-1.64310306e-01 -1.29045725e+00 1.90687263e+00 -9.93127823e-02
-6.57966971e-01 -6.28944874e-01 -2.33359158e-01 8.27054441e-01
-3.59322637e-01 5.02538264e-01 1.89553380e-01 -4.50823754e-02
-1.04332185e+00 7.50763714e-02 -1.36208549e-01 -4.73660603e-02
1.06262207e-01 -1.96681142e-01 -1.01696813e+00 -1.50522962e-01
-1.36967078e-01 -2.96730816e-01 2.62544185e-01 8.18732083e-01
1.68175948e+00 8.32826346e-02 -3.46145302e-01 8.37347806e-01
7.69954860e-01 1.16661519e-01 3.97663921e-01 2.65623748e-01
1.55940652e+00 7.21972108e-01 1.02224541e+00 6.77582204e-01
4.60370153e-01 1.08633077e+00 8.54227424e-01 3.31938148e-01
-2.66526520e-01 -1.30618021e-01 4.76603322e-02 4.59666550e-01
-6.07455730e-01 -1.61970139e-01 -9.71034110e-01 7.30513453e-01
-1.69237733e+00 -9.54048216e-01 -5.09580731e-01 1.83965969e+00
2.43534252e-01 -5.62588908e-02 5.77752292e-01 9.71553102e-02
5.16456544e-01 1.55396864e-01 -9.87963080e-01 2.57433951e-01
1.89197868e-01 -3.43514502e-01 5.14014363e-01 2.19276637e-01
-1.26225996e+00 1.16404355e+00 5.80216312e+00 4.79877561e-01
-5.46226025e-01 1.90917850e-01 6.26618043e-02 -3.41813624e-01
8.02297369e-02 -3.69857311e-01 -7.48305619e-01 3.46670270e-01
2.31997564e-01 3.31889540e-01 5.45998752e-01 1.56640887e+00
2.73405522e-01 -1.24669991e-01 -1.34895265e+00 1.62085891e+00
-1.20922796e-01 -1.15246725e+00 -1.50715008e-01 1.35937676e-01
9.42460060e-01 2.28865072e-01 7.85827041e-02 9.68958884e-02
3.46043169e-01 -6.93494081e-01 1.01909184e+00 4.38725412e-01
8.14956009e-01 -5.12621284e-01 1.04453258e-01 2.49410570e-01
-1.57758677e+00 -1.04403637e-01 -2.93915812e-02 2.68313825e-01
2.41120487e-01 -1.10797852e-01 -7.47814834e-01 3.17204505e-01
1.67470860e+00 1.10361469e+00 -3.20860922e-01 6.51811361e-01
8.27455819e-02 2.50310190e-02 -4.21114445e-01 3.39341342e-01
2.18850970e-01 7.53112584e-02 1.02185178e+00 1.09263957e+00
-4.98266928e-02 3.77546728e-01 2.37354860e-01 7.81359792e-01
1.17094800e-01 -2.97253907e-01 -7.70775974e-01 1.88307494e-01
2.28349805e-01 1.04665434e+00 -1.04385591e+00 -5.38152397e-01
1.12217799e-01 1.52111399e+00 -3.68712395e-01 3.44606847e-01
-9.42964613e-01 -6.29961491e-03 1.15504336e+00 2.95914799e-01
1.52284458e-01 -8.01218748e-01 -2.23528177e-01 -1.26937151e+00
-1.35578483e-01 -7.17057884e-01 2.42086679e-01 -1.38249826e+00
-1.02084851e+00 6.29759356e-02 4.96451229e-01 -1.29327452e+00
-2.46349856e-01 -8.11375141e-01 -7.96391070e-02 4.26816255e-01
-5.12743115e-01 -1.19729888e+00 -1.36915052e+00 1.11999393e+00
1.40803587e+00 9.08351690e-02 3.83055270e-01 2.69130617e-01
-3.16411674e-01 2.48134449e-01 -3.40117514e-01 6.31001070e-02
4.67562586e-01 -1.34588301e+00 8.25795054e-01 1.47771627e-01
1.13551781e-01 4.58977669e-01 5.92628539e-01 -8.14959288e-01
-1.84514785e+00 -1.14876878e+00 1.72137767e-01 -1.75828362e+00
2.86309361e-01 -7.00098574e-01 -4.59986329e-01 9.51413751e-01
-4.23394352e-01 2.64904946e-01 2.23973058e-02 -1.34640262e-01
8.82316977e-02 2.67454058e-01 -1.23097801e+00 7.19401240e-01
2.21462226e+00 -5.16107023e-01 -4.16850507e-01 7.60101974e-01
8.79119694e-01 -1.10055041e+00 -8.99786294e-01 4.83950436e-01
6.09289289e-01 -9.24925864e-01 1.57195103e+00 -5.67950368e-01
2.76997417e-01 -4.17097002e-01 -5.08662462e-01 -6.47550225e-01
-2.14384481e-01 -5.20704687e-01 -3.66525292e-01 8.59737992e-01
-5.78858137e-01 1.33327916e-01 1.16359365e+00 8.48232687e-01
-3.05586278e-01 -2.47988045e-01 -7.78097272e-01 -7.70910382e-01
-6.07174873e-01 -1.11770761e+00 5.59715033e-01 6.53139293e-01
-4.75534469e-01 -2.65348796e-02 -2.26660684e-01 1.17859028e-01
1.11122835e+00 -1.25573084e-01 1.65689504e+00 -1.31003892e+00
1.66627973e-01 -3.84864718e-01 -7.01780140e-01 -1.40052247e+00
3.01719099e-01 -8.03254664e-01 -9.06022731e-03 -1.39750540e+00
1.97413161e-01 -3.01370233e-01 1.87251523e-01 3.13170522e-01
1.23261765e-01 5.26614845e-01 2.73069531e-01 5.64497292e-01
-8.38121474e-01 5.53252518e-01 1.65128601e+00 -2.66685456e-01
-4.29200530e-01 -8.69698152e-02 1.40215367e-01 1.06823742e+00
2.26907566e-01 -1.41190305e-01 -5.39819479e-01 -3.80184174e-01
-2.53287256e-01 3.50523219e-02 1.01829720e+00 -1.30099249e+00
-6.56659156e-02 -3.67134959e-01 6.35027707e-01 -1.35109007e+00
8.41161609e-01 -1.14511466e+00 3.55704039e-01 4.08205003e-01
-4.78258319e-02 -2.01776430e-01 1.67442262e-01 7.97259569e-01
3.08683068e-01 4.46448594e-01 3.85518819e-01 -6.06152534e-01
-1.33644223e+00 6.36214137e-01 -4.11766639e-04 3.45876902e-01
1.31550467e+00 -6.63129270e-01 2.63269663e-01 -2.20432431e-02
-1.03649735e+00 4.85310853e-01 5.96422017e-01 9.84295189e-01
6.00666702e-01 -1.19125462e+00 -3.31835091e-01 4.08909880e-02
2.75910079e-01 7.54637063e-01 7.81093180e-01 8.07639003e-01
-6.04694068e-01 5.98288774e-01 -3.38953495e-01 -1.52820599e+00
-1.34205818e+00 5.20652473e-01 2.92553246e-01 4.07966286e-01
-1.20916891e+00 1.15183878e+00 7.29717672e-01 -7.19775081e-01
5.27056754e-01 -6.56524837e-01 1.36769459e-01 5.76098934e-02
1.68634117e-01 8.35967183e-01 -3.53303909e-01 -8.34419489e-01
-5.90478539e-01 1.02785754e+00 5.34839571e-01 -3.86284515e-02
1.07952845e+00 -2.90378898e-01 1.60227001e-01 6.31134510e-01
1.07516754e+00 -4.67081040e-01 -1.89634252e+00 3.75567079e-02
-3.71121466e-01 -8.74807954e-01 -3.49968374e-01 -3.98895353e-01
-8.79541099e-01 8.96995962e-01 6.74885213e-01 -3.63091320e-01
6.12802625e-01 5.02691686e-01 8.02357614e-01 3.54965061e-01
1.09253144e+00 -1.20509648e+00 7.87845552e-01 5.69970131e-01
1.19831252e+00 -1.40951180e+00 -2.20577359e-01 -5.08754313e-01
-8.23219478e-01 6.45996153e-01 1.05721116e+00 -1.86323777e-01
3.09219271e-01 -2.79548049e-01 -2.54525125e-01 -6.73899889e-01
7.08662570e-02 2.76927128e-02 4.15640652e-01 6.44284368e-01
-1.68535948e-01 5.90013936e-02 4.90425020e-01 2.73996800e-01
-3.52025777e-01 -8.99045244e-02 -2.27179259e-01 7.46360481e-01
1.60041766e-03 -3.66378874e-01 -6.31770790e-01 -1.78894419e-02
8.63582194e-02 6.27483964e-01 -1.63586318e-01 1.00582671e+00
2.59009778e-01 6.27049148e-01 3.25351864e-01 -3.95764142e-01
7.16695428e-01 -8.77521187e-02 8.28006685e-01 -5.85977256e-01
-4.96384263e-01 -7.10482374e-02 -3.00927460e-01 -1.30294585e+00
-5.07061481e-01 -8.20211828e-01 -1.29144764e+00 -3.80770981e-01
1.96065903e-01 -4.70796168e-01 8.64451170e-01 6.50390804e-01
4.75938261e-01 7.00578570e-01 3.19235563e-01 -2.06817889e+00
1.73816115e-01 -8.18245232e-01 -5.35010338e-01 8.01329613e-01
1.77390978e-01 -1.08617210e+00 -2.43250549e-01 1.97812095e-01] | [6.8852338790893555, -0.9579700827598572] |
96274ec7-42e3-4d48-82c6-2103a74c2437 | using-deep-neural-network-for-android-malware | 1904.00736 | null | http://arxiv.org/abs/1904.00736v1 | http://arxiv.org/pdf/1904.00736v1.pdf | Using Deep Neural Network for Android Malware Detection | The pervasiveness of the Android operating system, with the availability of
applications almost for everything, is readily accessible in the official
Google play store or a dozen alternative third-party markets. Additionally, the
vital role of smartphones in modern life leads to store significant information
on devices, not only personal information but also corporate information, which
attract malware developers to develop applications that can infiltrate user's
devices to steal information and perform harmful tasks. This accompanied with
the limitation of currently defenses techniques such as ineffective screening
in Google play store, weak or no screening in third-party markets. Antiviruses
software that still relies on a signature-based database that is effective only
in identifying known malware. To contrive with malicious applications that are
increased in volume and sophistication, we propose an Android malware detection
system that applies deep learning technique to face the threats of Android
malware. Extensive experiments on a real-world dataset contain benign and
malicious applications uncovered that the proposed system reaches an accuracy
of 95.31%. | ['Abdelmonim Naway', 'Yuancheng LI'] | 2019-01-16 | null | null | null | null | ['android-malware-detection'] | ['miscellaneous'] | [-2.67543048e-02 -3.06889266e-01 -5.23558736e-01 4.34053421e-01
-3.43661815e-01 -9.16042507e-01 5.91679633e-01 -3.83667976e-01
-2.79774386e-02 4.01339084e-01 -3.57621133e-01 -9.52156425e-01
2.78114051e-01 -7.35960126e-01 -6.72602773e-01 -2.59678900e-01
-2.80196499e-02 -8.56606215e-02 7.04505682e-01 -4.47707504e-01
5.59656799e-01 1.21445358e-01 -1.83418310e+00 4.93668318e-01
8.78067195e-01 1.38914752e+00 1.01622634e-01 6.84610426e-01
-1.23382360e-01 6.00831985e-01 -7.28624940e-01 -6.89926207e-01
3.93237323e-01 4.48843464e-02 -3.11563462e-01 -3.35330516e-01
-1.09546974e-01 -7.20862091e-01 -1.81713656e-01 1.43350267e+00
3.15330476e-02 -6.90842032e-01 2.74318457e-01 -1.42621851e+00
-7.65632153e-01 3.33234179e-03 -1.02071893e+00 2.95650154e-01
4.45829332e-01 2.84622729e-01 2.42660895e-01 -2.84805030e-01
1.79619923e-01 9.96906281e-01 9.04072106e-01 4.88010734e-01
-2.82127261e-01 -9.63659644e-01 -2.75578797e-01 -3.94090600e-02
-1.01690483e+00 -3.58695358e-01 6.79934561e-01 -5.92328966e-01
7.77530253e-01 4.83453900e-01 6.30298853e-01 1.39330482e+00
9.10962164e-01 2.26162612e-01 1.25426733e+00 6.04462698e-02
6.86626881e-02 6.87768340e-01 4.43436205e-01 8.00770700e-01
8.30807865e-01 2.75500864e-02 -1.01667583e-01 -7.35811591e-01
1.23905353e-01 5.62395155e-01 9.05144215e-02 2.16864198e-02
-4.09361362e-01 8.46681714e-01 -1.42749786e-01 3.26025188e-01
-2.50629783e-01 -2.25492314e-01 7.31712997e-01 1.07386753e-01
2.12827370e-01 1.52350217e-01 -8.09527814e-01 -6.14072621e-01
-6.23460531e-01 -1.48270026e-01 1.01778829e+00 4.59906906e-01
5.56882143e-01 2.02546090e-01 7.40744889e-01 4.86308694e-01
7.21754849e-01 6.40948951e-01 1.12443650e+00 -6.80883229e-01
2.63455272e-01 1.29100811e+00 -7.44180679e-02 -1.43002617e+00
3.23164433e-01 -1.63368732e-01 -6.19167387e-01 1.62380651e-01
7.38958567e-02 -1.03093095e-01 -6.47109091e-01 1.09252191e+00
3.58949870e-01 3.59587282e-01 -3.43086362e-01 3.95071879e-02
4.61305350e-01 6.11248016e-01 -1.02370791e-01 -2.16157436e-01
1.53366041e+00 -4.93746310e-01 -5.46860278e-01 -1.15517482e-01
3.77653986e-01 -6.52600646e-01 1.51819384e+00 6.97236478e-01
-5.29747069e-01 -4.81946141e-01 -1.36008894e+00 2.25334451e-01
-1.04957879e+00 -1.41952559e-01 6.27318203e-01 1.67142367e+00
-9.51678336e-01 2.54000843e-01 -9.00206327e-01 -2.20298573e-01
8.63407373e-01 7.97369003e-01 -2.89335907e-01 3.52342516e-01
-8.74690115e-01 4.89399493e-01 3.46851237e-02 -3.53342593e-01
-1.21821487e+00 -2.97837287e-01 -4.18718249e-01 -7.45785683e-02
5.43788850e-01 5.81761524e-02 1.04684067e+00 -1.09639275e+00
-1.51516950e+00 6.83527350e-01 2.25815818e-01 -3.88044685e-01
2.65829355e-01 -3.50988150e-01 -6.68138325e-01 -1.60908297e-01
1.19385384e-01 -2.86711663e-01 1.30638146e+00 -9.77392614e-01
-5.34156442e-01 -8.73175263e-01 2.64991671e-01 -4.47237313e-01
-8.85426104e-01 1.14134111e-01 -2.60359108e-01 -7.70363435e-02
-4.00042176e-01 -1.14246106e+00 2.81328321e-01 -7.29601204e-01
-5.94464540e-01 -3.43869217e-02 2.02648091e+00 -1.06540728e+00
1.47195101e+00 -2.20300651e+00 -4.26547885e-01 9.66586620e-02
3.36490244e-01 9.25003290e-01 4.52674180e-01 1.20647490e-01
1.21815555e-01 8.84400606e-01 -2.49766074e-02 3.04552138e-01
-4.04356629e-01 -6.26641288e-02 -4.75703955e-01 3.62304837e-01
-3.06183726e-01 7.60439098e-01 -4.91977334e-01 -2.54357845e-01
-2.80414782e-02 4.22982961e-01 -4.05978799e-01 -1.59860790e-01
-1.01951487e-01 2.67727733e-01 -8.78565431e-01 1.27292490e+00
7.22263217e-01 -3.46004277e-01 1.23957008e-01 2.80091107e-01
-1.94259658e-02 2.99792767e-01 -4.65224773e-01 8.45573187e-01
-4.00482029e-01 3.56113762e-01 3.03161412e-01 -7.14146435e-01
3.30980957e-01 -3.97299118e-02 2.50035107e-01 -3.03762704e-01
6.28451467e-01 3.90090764e-01 1.24973334e-01 -6.17529392e-01
2.75536776e-01 6.91341698e-01 -1.77536123e-02 3.94013017e-01
-5.07359982e-01 4.11909670e-01 -4.35074449e-01 2.60430664e-01
1.49875379e+00 -6.94130510e-02 2.97173649e-01 -3.29099670e-02
1.04225349e+00 -2.10639089e-01 1.13067590e-01 3.99402291e-01
-5.68989336e-01 -2.90562570e-01 7.57679284e-01 -1.82900026e-01
-5.50644875e-01 -7.32643604e-01 3.38598192e-02 1.19956815e+00
1.72731519e-01 -3.55314136e-01 -1.22153342e+00 -1.41131806e+00
1.18629746e-02 1.50356153e-02 -4.95561391e-01 -4.87629563e-01
-2.95357823e-01 -6.16372108e-01 7.92619824e-01 -1.64889485e-01
1.14679575e+00 -9.00865912e-01 -6.19236171e-01 -2.78534800e-01
3.51910979e-01 -9.01666224e-01 -2.90727466e-01 -2.69535452e-01
-8.48653197e-01 -1.58545518e+00 -2.43464768e-01 -4.69500095e-01
3.47325593e-01 4.04825896e-01 5.92815816e-01 5.43148637e-01
-1.95887342e-01 3.57280910e-01 -2.56336063e-01 -5.27035356e-01
-6.89370871e-01 3.57643038e-01 2.56296009e-01 1.12585887e-01
6.57155514e-01 -4.33523774e-01 -3.65804285e-01 2.06572101e-01
-7.22138047e-01 -8.39838743e-01 6.16049469e-01 3.87971818e-01
7.89893940e-02 5.15208781e-01 6.25737667e-01 -1.22157991e+00
9.48744476e-01 -9.51432705e-01 -6.06490016e-01 -1.40995845e-01
-8.64394009e-01 -5.78677773e-01 6.24417365e-01 -7.01984286e-01
-9.08887267e-01 -4.62373272e-02 -2.72150576e-01 -1.28827572e-01
5.66820316e-02 1.47595912e-01 -5.53387225e-01 -4.19779807e-01
5.52945018e-01 3.96528423e-01 3.04446548e-01 -3.26983064e-01
-3.27947587e-02 1.36103916e+00 1.16629802e-01 7.71627203e-02
7.65797496e-01 4.95910674e-01 -3.14904660e-01 -1.09242427e+00
-2.27581680e-01 -2.69029111e-01 2.45019644e-02 -2.30317842e-02
7.95179844e-01 -5.73114455e-01 -1.21483636e+00 9.65623736e-01
-1.16837895e+00 2.27940455e-01 6.40546739e-01 -1.84223622e-01
1.81782633e-01 7.19090521e-01 -6.92285836e-01 -9.89788353e-01
-2.92498291e-01 -1.56519151e+00 9.44369316e-01 1.39806852e-01
-1.82323121e-02 -6.01328075e-01 -1.84862986e-02 8.18603456e-01
5.68622053e-01 1.30248532e-01 7.54720151e-01 -8.82272780e-01
-8.41256142e-01 -5.34076691e-01 -1.69419691e-01 7.78000236e-01
5.66596806e-01 2.41368398e-01 -9.20526981e-01 -1.30653977e-01
7.99358666e-01 6.76283464e-02 4.70326930e-01 2.24194452e-02
1.41694331e+00 -7.61694014e-01 -6.92431211e-01 4.47476178e-01
1.22008038e+00 7.84894645e-01 9.23069239e-01 3.54814261e-01
1.04976594e+00 4.25294936e-01 3.47237587e-01 2.42132604e-01
2.21906096e-01 1.94187075e-01 9.11445200e-01 5.25966048e-01
4.24690932e-01 -1.97057784e-01 6.17399216e-01 6.51866794e-01
-8.77484586e-03 9.86548588e-02 -7.84536004e-01 3.33787836e-02
-1.29992461e+00 -7.96966374e-01 -2.54649609e-01 2.26632738e+00
6.10970080e-01 4.69204277e-01 3.61427486e-01 2.25079432e-01
7.48060107e-01 3.50876972e-02 -7.72469640e-01 -5.53444266e-01
3.81153196e-01 1.65518105e-01 7.56719232e-01 1.42638117e-01
-1.10016835e+00 8.68700743e-01 5.80058479e+00 9.14018691e-01
-1.37649095e+00 8.72397721e-01 8.60984147e-01 3.14676672e-01
7.42858462e-03 -3.05971742e-01 -8.66980016e-01 1.26740634e+00
1.18286777e+00 2.73424447e-01 5.13803065e-01 1.47864270e+00
1.85089018e-02 -1.85972244e-01 -3.64949912e-01 1.12758374e+00
-5.04897954e-03 -1.24724638e+00 -9.80393365e-02 8.85612607e-01
5.17885685e-01 2.15314254e-02 7.15835094e-01 4.44819778e-01
-2.10837498e-01 -8.25259030e-01 6.23700656e-02 -5.41726947e-02
6.64050162e-01 -8.56834471e-01 8.44731212e-01 4.52059329e-01
-7.60591865e-01 -5.75384676e-01 -9.59860906e-02 -2.78256148e-01
-4.73217130e-01 1.87646195e-01 -6.65495932e-01 -2.52274334e-01
8.83132100e-01 5.55013001e-01 -9.26455617e-01 2.49695063e-01
1.92131341e-01 8.10309708e-01 -1.03340641e-01 -2.80340791e-01
1.91466603e-02 -1.55517504e-01 3.87109697e-01 5.52522302e-01
3.35573256e-01 -3.24493796e-01 -1.43260434e-01 4.98660147e-01
-1.93521813e-01 9.64431167e-02 -1.39743233e+00 -6.57113731e-01
4.39004630e-01 1.37862599e+00 -7.22410619e-01 -1.82586670e-01
-4.86825734e-01 1.00482357e+00 -3.12867522e-01 -6.40081242e-02
-9.82779205e-01 -3.08243573e-01 7.43073106e-01 8.89187396e-01
-1.68327034e-01 -2.57164389e-01 -1.97823197e-01 -7.87247360e-01
4.79710102e-02 -1.54175365e+00 -1.76784873e-01 -2.21413106e-01
-1.08579004e+00 5.01916230e-01 -3.64495039e-01 -9.22306776e-01
-1.73933059e-01 -9.06438053e-01 -7.32661605e-01 3.24652344e-01
-8.79211247e-01 -1.22818232e+00 -4.77364063e-02 6.98180556e-01
5.88126957e-01 -8.80846739e-01 5.15186727e-01 4.63044703e-01
-6.15221083e-01 3.98490876e-01 1.05912179e-01 -7.07924515e-02
1.93565652e-01 -7.88553417e-01 2.98566997e-01 7.02072799e-01
-2.81826496e-01 1.11552858e+00 1.55407131e-01 -1.17623365e+00
-1.99297249e+00 -7.35430062e-01 -2.25759186e-02 -1.14540660e+00
7.91210353e-01 -4.18500453e-01 -7.05086410e-01 7.14468360e-01
2.40209952e-01 -3.45084041e-01 6.69533312e-01 -3.22473586e-01
-3.67671371e-01 -2.61317164e-01 -1.34376419e+00 3.88480723e-01
6.82906449e-01 -7.49955475e-01 -4.13632095e-02 4.02924538e-01
8.32299471e-01 -8.63542594e-03 -6.46476969e-02 1.31412923e-01
7.89290607e-01 -1.54703236e+00 8.08392525e-01 -4.28096265e-01
2.30045304e-01 -6.28581718e-02 -1.39708817e-01 -2.69757539e-01
4.18636918e-01 -9.36622202e-01 -6.83704853e-01 1.18862319e+00
2.04442590e-01 -1.04186070e+00 1.14745271e+00 4.00033444e-01
1.42911568e-01 -8.85276735e-01 -8.32736433e-01 -5.30535400e-01
-3.37925136e-01 -4.07551050e-01 6.85103118e-01 9.07110274e-01
-3.89609665e-01 2.14702517e-01 -2.77633607e-01 9.84245017e-02
4.31860179e-01 -7.36296713e-01 9.43322718e-01 -1.34598601e+00
-4.90883112e-01 -1.58569112e-01 -4.42652345e-01 -7.29675531e-01
1.67816013e-01 -2.85426140e-01 -6.49174929e-01 -6.48130238e-01
1.95016533e-01 -2.65926123e-01 -7.10254610e-02 2.92358816e-01
2.66009301e-01 5.31951010e-01 -2.47581914e-01 3.70075136e-01
-5.33483803e-01 -1.66765779e-01 6.37108862e-01 -3.15471828e-01
-5.24196923e-01 3.59015316e-01 -1.11605382e+00 1.19351482e+00
7.78669834e-01 -2.27709427e-01 -7.61744976e-01 1.80604413e-01
4.91223723e-01 -3.33613724e-01 2.21392766e-01 -9.41957653e-01
-2.35771671e-01 1.20526090e-01 1.89120427e-01 -3.76414746e-01
2.73822993e-01 -9.67665076e-01 -1.19513236e-02 8.53678107e-01
5.74152410e-01 2.89603025e-01 1.97865114e-01 7.91699350e-01
1.72071487e-01 -1.51831985e-01 7.62663364e-01 -3.33742976e-01
-4.89434540e-01 1.54712379e-01 -5.22807002e-01 -3.11140060e-01
1.44086218e+00 -7.06713021e-01 -5.59713662e-01 -3.80433410e-01
-1.14436828e-01 -4.66669530e-01 6.28124058e-01 7.93399453e-01
4.72957015e-01 -8.00507784e-01 1.70261800e-01 5.07997096e-01
-3.40310156e-01 -7.43063509e-01 -2.59742178e-02 6.68453634e-01
-7.03956962e-01 3.72165322e-01 -3.93507153e-01 -4.33551013e-01
-1.66748643e+00 8.55311215e-01 4.95409109e-02 -2.24703148e-01
1.88786358e-01 4.69911307e-01 -8.00569076e-03 -2.22483709e-01
1.27827063e-01 -1.13092363e-01 -4.00262982e-01 -1.85261279e-01
7.59219468e-01 7.83574462e-01 2.81605609e-02 -9.53602970e-01
-5.83050787e-01 3.41101825e-01 -2.52232969e-01 3.07549566e-01
7.71839619e-01 -4.59870771e-02 -6.18076801e-01 1.75334349e-01
1.26007211e+00 7.87699819e-01 -5.83794415e-01 8.97708714e-01
4.04183343e-02 -6.83566868e-01 -1.65311083e-01 -8.58133197e-01
-1.04587412e+00 7.77868330e-01 1.16008818e+00 1.12350905e+00
8.15611422e-01 -2.47607559e-01 1.27467477e+00 3.51613998e-01
6.52885914e-01 -6.02371275e-01 3.81652087e-01 2.40609705e-01
3.33408237e-01 -1.59178019e+00 -8.21280256e-02 -4.13242280e-01
-2.55224884e-01 4.85385537e-01 8.10494721e-01 -9.86748468e-03
1.15946770e+00 2.60080010e-01 -1.32701531e-01 -2.44949564e-01
-6.35781512e-02 4.96960044e-01 -1.33465156e-01 1.01667857e+00
2.42877111e-01 -7.61877820e-02 -4.32744741e-01 8.95060956e-01
1.70341022e-02 -7.64688700e-02 4.93071169e-01 8.93434644e-01
-6.04903221e-01 -1.35923350e+00 -3.38127166e-01 8.59864950e-01
-1.47162378e+00 -3.10060848e-02 -9.76369500e-01 8.04266095e-01
6.75520122e-01 1.18074727e+00 -4.19855207e-01 -9.35158432e-01
-1.19078726e-01 -5.82589209e-02 -1.82430848e-01 -5.85843265e-01
-6.63437545e-01 -2.31060579e-01 -2.04641327e-01 -6.02310002e-01
2.09209487e-01 -2.78589487e-01 -9.11535084e-01 -4.52750444e-01
-3.92649382e-01 -1.21877089e-01 1.07682240e+00 7.53163218e-01
7.96035826e-01 9.09989104e-02 7.18610406e-01 -4.96558875e-01
-7.13754833e-01 -8.36979806e-01 -3.29883307e-01 -4.54473728e-03
3.65831673e-01 -7.55448222e-01 -4.12302941e-01 -1.74840659e-01] | [14.423316955566406, 9.680148124694824] |
f70cc143-103f-4118-909c-5c704bbdf208 | random-feedback-alignment-algorithms-to-train | 2306.02325 | null | https://arxiv.org/abs/2306.02325v1 | https://arxiv.org/pdf/2306.02325v1.pdf | Random Feedback Alignment Algorithms to train Neural Networks: Why do they Align? | Feedback alignment algorithms are an alternative to backpropagation to train neural networks, whereby some of the partial derivatives that are required to compute the gradient are replaced by random terms. This essentially transforms the update rule into a random walk in weight space. Surprisingly, learning still works with those algorithms, including training of deep neural networks. This is generally attributed to an alignment of the update of the random walker with the true gradient - the eponymous gradient alignment -- which drives an approximate gradient descend. The mechanism that leads to this alignment remains unclear, however. In this paper, we use mathematical reasoning and simulations to investigate gradient alignment. We observe that the feedback alignment update rule has fixed points, which correspond to extrema of the loss function. We show that gradient alignment is a stability criterion for those fixed points. It is only a necessary criterion for algorithm performance. Experimentally, we demonstrate that high levels of gradient alignment can lead to poor algorithm performance and that the alignment is not always driving the gradient descend. | ['Florian Bacho', 'Dominique Chu'] | 2023-06-04 | null | null | null | null | ['mathematical-reasoning'] | ['natural-language-processing'] | [ 1.08097970e-01 -6.57429844e-02 -1.91765413e-01 -5.59900463e-01
2.54281074e-01 -2.90631056e-01 5.63435316e-01 8.26012120e-02
-6.88151777e-01 8.66828740e-01 -1.45234257e-01 -5.25351644e-01
-1.27878025e-01 -8.42880189e-01 -1.05342233e+00 -8.99669349e-01
-1.40105918e-01 9.64021459e-02 3.44126463e-01 -4.74883765e-01
5.84200621e-01 4.32675332e-01 -1.38469458e+00 -4.70861234e-02
7.75861621e-01 7.74774671e-01 9.85361934e-02 9.38551247e-01
-3.89806002e-01 6.89434826e-01 -4.73867774e-01 -3.26625615e-01
4.49288756e-01 -8.32108021e-01 -8.82158279e-01 -3.94360542e-01
4.41726387e-01 -3.90321255e-01 -1.41493574e-01 1.14000571e+00
2.85371870e-01 5.70735075e-02 7.70892322e-01 -1.03910446e+00
-3.73553693e-01 5.25196493e-01 -4.11575615e-01 2.55361110e-01
7.60103613e-02 -2.35222746e-02 1.14512229e+00 -7.74819851e-01
7.88234711e-01 7.42552638e-01 1.03177702e+00 5.17298758e-01
-9.15743530e-01 -2.91611016e-01 -1.57763110e-03 2.62248665e-01
-1.20530283e+00 -1.72481701e-01 8.01388860e-01 -2.06230313e-01
9.36382532e-01 2.86689639e-01 9.67498779e-01 5.96736372e-01
5.71695507e-01 6.12268209e-01 9.05349195e-01 -8.63376141e-01
2.83272892e-01 2.13807628e-01 3.71556133e-01 9.29757118e-01
3.66867810e-01 2.28180125e-01 -6.03505850e-01 9.58242342e-02
8.13572049e-01 -1.62723213e-01 -5.32281756e-01 -5.69017291e-01
-5.06156862e-01 7.79194713e-01 8.65271270e-01 3.81072611e-01
-2.72006631e-01 3.61931175e-01 -1.42767251e-01 7.44353950e-01
1.79528985e-02 5.94618440e-01 -4.00456280e-01 -1.40883863e-01
-8.96937132e-01 3.39200407e-01 1.00310981e+00 3.15325320e-01
1.01114392e+00 5.06440438e-02 1.48906857e-01 6.04838490e-01
3.60947460e-01 3.34692776e-01 8.28862727e-01 -8.81380439e-01
2.68874168e-01 5.41244805e-01 1.87097639e-01 -1.11143351e+00
-5.49589634e-01 -7.70864844e-01 -6.16486251e-01 6.68059707e-01
1.05711901e+00 -3.88109654e-01 -6.28125310e-01 1.91794777e+00
1.24145590e-01 -9.01454538e-02 -1.83546647e-01 9.80192661e-01
9.71319377e-02 5.15763640e-01 -5.48605680e-01 -2.66299337e-01
5.43886721e-01 -8.32063437e-01 -5.21188796e-01 -4.09266680e-01
6.45690739e-01 -6.01391852e-01 1.26559746e+00 -2.66718790e-02
-1.20091009e+00 -2.07481995e-01 -1.60007393e+00 3.97544771e-01
-2.48950318e-01 -2.98052579e-01 3.41971844e-01 4.77621734e-01
-1.10191131e+00 1.48791504e+00 -9.22690034e-01 -2.43689597e-01
-7.60804862e-02 4.08078253e-01 -1.47214979e-02 4.59309816e-01
-1.20202744e+00 1.20075059e+00 1.25208303e-01 3.35893303e-01
-6.64537176e-02 -4.04266894e-01 -3.38331640e-01 -6.19385280e-02
-2.59538114e-01 -5.39850056e-01 1.40248370e+00 -1.32316685e+00
-1.75046265e+00 6.02112949e-01 -3.53722751e-01 -6.76015854e-01
8.68045807e-01 -3.08323681e-01 3.49771529e-02 -2.67063737e-01
-3.64185572e-01 5.27612567e-01 7.82442331e-01 -1.03184903e+00
-4.32030857e-01 -2.13184848e-01 -2.04724208e-01 4.07633603e-01
-4.43996370e-01 -3.50956768e-01 -3.25460657e-02 -2.37670466e-01
3.84790540e-01 -9.14878488e-01 -2.40598455e-01 1.22353090e-02
-3.16616476e-01 8.02110732e-02 4.76311684e-01 -1.71474993e-01
1.36208546e+00 -1.69212604e+00 -1.09680936e-01 6.63219690e-01
7.17725903e-02 1.44303575e-01 2.39128873e-01 3.02664608e-01
-7.87217841e-02 1.53749473e-02 -2.62942612e-01 3.62216085e-02
-3.67287323e-02 2.08375245e-01 -3.90148193e-01 6.71375990e-01
1.21457931e-02 9.06076014e-01 -9.18196499e-01 4.53351764e-03
-1.79936960e-01 3.45074445e-01 -6.40738547e-01 -2.61870376e-03
-8.43679626e-03 -3.99775542e-02 -1.21677406e-01 1.75277442e-02
4.87141848e-01 -2.07868516e-01 7.34889582e-02 6.41537681e-02
-4.75561261e-01 5.04922390e-01 -1.07698619e+00 1.13549352e+00
-1.36456117e-01 1.07864332e+00 -2.14877918e-01 -1.02911472e+00
9.77452397e-01 -1.42360851e-01 2.79045373e-01 -6.34893596e-01
1.47928014e-01 5.74832559e-01 4.56786841e-01 -1.49225950e-01
4.79088783e-01 -1.69378057e-01 5.64915895e-01 6.50039136e-01
-1.41282767e-01 -1.19317636e-01 3.39534014e-01 -1.70605451e-01
1.01303804e+00 4.06671353e-02 4.21669222e-02 -3.89845580e-01
3.42953891e-01 -1.33218272e-02 4.64243472e-01 1.03578985e+00
-9.52917039e-02 6.81197047e-01 4.94129926e-01 -6.57366574e-01
-1.23760951e+00 -9.27011669e-01 -2.29078487e-01 6.61535442e-01
2.42796630e-01 -3.38611722e-01 -9.61620867e-01 -3.51753891e-01
-6.30179420e-04 7.10959077e-01 -5.94505787e-01 -4.41852182e-01
-8.65599573e-01 -9.63021934e-01 4.42828327e-01 4.14492339e-01
6.01894796e-01 -1.01300228e+00 -9.63690758e-01 4.09542739e-01
1.67399958e-01 -3.43801022e-01 -2.99142718e-01 6.07238650e-01
-1.13525009e+00 -9.60022867e-01 -7.18151093e-01 -5.35515845e-01
9.35656369e-01 -2.32605129e-01 1.10523307e+00 4.54271317e-01
-9.20624584e-02 -1.73602983e-01 1.31459713e-01 -4.07194287e-01
-4.86187309e-01 8.24454650e-02 1.16865896e-01 -3.37715477e-01
1.61924407e-01 -7.92462409e-01 -8.54327440e-01 4.72225726e-01
-4.61015463e-01 -1.27308205e-01 4.84916359e-01 9.37435806e-01
3.24895680e-01 -4.32620086e-02 2.62537599e-01 -6.17041945e-01
9.49148893e-01 -6.17311969e-02 -6.52673423e-01 8.39705318e-02
-1.16998374e+00 5.65310180e-01 5.92268288e-01 -4.13034469e-01
-7.32792735e-01 3.29887904e-02 -3.20704699e-01 1.61880045e-03
2.80640006e-01 4.94841605e-01 4.95393544e-01 -3.02524507e-01
1.01817334e+00 1.95231259e-01 1.77669615e-01 -2.53918558e-01
3.77542555e-01 3.81299555e-01 4.50927228e-01 -4.33985919e-01
7.43797958e-01 3.53186816e-01 6.32239059e-02 -5.40940583e-01
-7.10138440e-01 -1.02467552e-01 -5.19499838e-01 -4.74447489e-01
3.15114498e-01 -8.99650902e-02 -4.61156428e-01 6.94412768e-01
-1.11892927e+00 -7.03179896e-01 -4.71694291e-01 3.87523353e-01
-5.68399549e-01 -6.37654960e-02 -9.00680006e-01 -7.34590530e-01
-3.32353950e-01 -9.08395171e-01 1.37426347e-01 6.11562788e-01
-4.03567404e-01 -1.18114829e+00 3.30484837e-01 -6.29297137e-01
8.20598841e-01 -1.72656730e-01 7.70894468e-01 -3.89376879e-01
-1.65216535e-01 -2.75360405e-01 4.14894931e-02 3.00411880e-01
8.38017184e-03 2.88028955e-01 -7.15366721e-01 -3.48351523e-02
2.52560407e-01 8.02623406e-02 8.66934836e-01 6.13178253e-01
5.54696143e-01 -3.99437934e-01 -2.34034389e-01 6.33479357e-01
1.37647259e+00 2.61044830e-01 5.85358858e-01 7.05933213e-01
4.37804997e-01 2.44680211e-01 6.64691329e-02 2.26253390e-01
-1.08055457e-01 5.62884629e-01 2.60542423e-01 -2.40537412e-02
-2.81382706e-02 -3.19671333e-01 4.26670372e-01 9.70444083e-01
-8.92325044e-02 1.84775621e-01 -1.07060945e+00 2.88199157e-01
-1.88435125e+00 -9.78877306e-01 -3.03293079e-01 2.45458055e+00
1.31595194e+00 6.87209010e-01 -9.93296131e-02 3.53858292e-01
6.33258581e-01 -1.73678771e-01 -6.25551164e-01 -6.55926764e-01
-2.89153820e-03 1.33654147e-01 9.56267238e-01 9.42536533e-01
-6.14249468e-01 6.14666104e-01 7.62570333e+00 3.39443386e-01
-1.68375587e+00 -2.53294230e-01 4.02955115e-01 -2.37398036e-02
-2.67761201e-01 -7.41073862e-02 -8.05108309e-01 4.85711634e-01
8.03057790e-01 -3.06039959e-01 3.91598821e-01 7.58634329e-01
2.54926588e-02 -3.54021713e-02 -9.43898499e-01 5.84698319e-01
-3.26199740e-01 -1.38944435e+00 -2.36646816e-01 -1.57974496e-01
8.04636836e-01 1.94218010e-01 1.10819563e-02 -6.54187351e-02
4.30272043e-01 -8.62949789e-01 6.93915725e-01 5.54455042e-01
2.58084506e-01 -6.21577740e-01 7.32982278e-01 3.72561485e-01
-8.47049236e-01 -9.88722816e-02 -4.59643453e-01 -4.73215580e-01
9.99332443e-02 9.14209187e-01 -7.88786948e-01 -5.95835038e-02
4.33122903e-01 3.48590523e-01 -4.34555739e-01 1.47623289e+00
-5.17225802e-01 6.38297319e-01 -6.72395647e-01 -5.25942504e-01
3.18055809e-01 -4.43555892e-01 5.96040845e-01 1.06542039e+00
3.49326909e-01 -4.90957469e-01 -3.99431437e-01 8.83746386e-01
4.43043076e-02 1.57953173e-01 -5.45937538e-01 4.33801971e-02
3.92992884e-01 7.34887719e-01 -8.04833889e-01 -3.06961201e-02
-3.90405953e-02 1.01028788e+00 5.75872838e-01 4.96492594e-01
-6.39472485e-01 -6.94349587e-01 6.64085209e-01 2.41253123e-01
1.45254090e-01 -2.63665199e-01 -6.54065490e-01 -8.04333746e-01
2.68261582e-01 -3.26427788e-01 -2.07830459e-01 -5.26288092e-01
-9.41527128e-01 3.92907888e-01 -2.79796511e-01 -9.24239218e-01
-7.40719020e-01 -5.69415569e-01 -1.00448334e+00 8.87816608e-01
-1.43434429e+00 -2.05623120e-01 -1.53606817e-01 2.53718317e-01
6.22046366e-02 8.00889805e-02 5.87998867e-01 1.66060224e-01
-4.53967184e-01 7.40089238e-01 4.35697615e-01 7.85494149e-02
4.28242445e-01 -1.49210966e+00 3.83627594e-01 7.83077359e-01
2.66239673e-01 9.25760150e-01 1.17752647e+00 -4.25142348e-01
-1.10016549e+00 -5.52542984e-01 1.07536781e+00 -1.89626887e-01
9.02648926e-01 4.41461243e-02 -1.04787183e+00 4.62542295e-01
8.25496688e-02 -1.35047227e-01 1.94546148e-01 5.07385209e-02
-8.87844190e-02 -2.20810428e-01 -8.33417773e-01 8.31956923e-01
9.90975916e-01 -3.32505971e-01 -5.13596654e-01 7.43142962e-02
2.68970728e-01 -4.38097864e-01 -2.87524462e-01 3.83578926e-01
8.90694022e-01 -1.16626310e+00 7.04520941e-01 -5.49035549e-01
4.78425354e-01 -2.61736482e-01 1.39950365e-01 -1.61783779e+00
-1.09140530e-01 -7.31107950e-01 -6.99120685e-02 8.47609341e-01
8.32222223e-01 -7.48000503e-01 1.05552411e+00 6.42272651e-01
9.63644534e-02 -1.02887583e+00 -8.31499994e-01 -8.14700305e-01
4.11009908e-01 -2.67062813e-01 2.44019538e-01 8.39398444e-01
2.25738749e-01 2.56805032e-01 6.09457679e-03 -3.14036727e-01
4.93274719e-01 -1.46548152e-01 3.97862822e-01 -1.02001774e+00
-3.22964191e-01 -1.11572945e+00 -4.20311958e-01 -1.24920869e+00
-4.72460538e-02 -7.44542241e-01 1.88482165e-01 -1.46652508e+00
-3.56036216e-01 -6.22675121e-01 -4.52686608e-01 1.42241240e-01
-1.45382077e-01 -4.81486022e-02 2.35931724e-02 3.00133079e-01
-1.34585872e-01 2.38657266e-01 9.88229036e-01 1.65962487e-01
-6.05666220e-01 3.68591547e-01 -3.98895115e-01 9.66598451e-01
1.19097364e+00 -5.26009381e-01 -3.78273696e-01 -4.06449527e-01
6.77668452e-01 -5.00896335e-01 3.94092426e-02 -1.13368714e+00
5.51104963e-01 3.70000601e-02 4.26582992e-01 -2.25636035e-01
3.38493002e-04 -6.16867423e-01 -7.21038431e-02 8.41125548e-01
-5.58626711e-01 2.88223594e-01 -5.99111691e-02 1.30934790e-01
-4.82536964e-02 -8.76305640e-01 8.43984783e-01 1.17697701e-01
-3.75752389e-01 -1.58058077e-01 -4.96836275e-01 -6.04871511e-02
5.51536977e-01 -2.35577539e-01 -1.03810474e-01 -5.55189669e-01
-4.66112047e-01 1.00187898e-01 5.80779731e-01 -4.56983857e-02
4.26570982e-01 -1.33884740e+00 -4.82912332e-01 3.09364974e-01
-4.92893696e-01 -2.40134344e-01 -5.62581658e-01 7.61545658e-01
-8.51868570e-01 1.20819777e-01 -2.50167876e-01 -5.77544928e-01
-1.10711324e+00 -9.99166742e-02 9.78539288e-01 -1.75013825e-01
-5.15961885e-01 1.09912193e+00 -4.93931264e-01 -5.42246878e-01
4.22614038e-01 -3.98983777e-01 9.23771709e-02 -1.23825513e-01
6.30164623e-01 1.40177280e-01 2.41512507e-01 -5.96069172e-02
-1.93324551e-01 7.19969809e-01 -4.23343889e-02 -3.50015759e-01
1.12867248e+00 2.96150129e-02 -6.15422353e-02 7.30126679e-01
1.23123491e+00 -7.52903372e-02 -1.38825953e+00 -1.27126276e-01
2.20531821e-01 -2.96736091e-01 -6.54990971e-02 -6.23733699e-01
-9.95386541e-01 8.69640827e-01 6.83465600e-01 4.33962762e-01
7.11181045e-01 -5.31200945e-01 6.32178068e-01 7.62020886e-01
2.22702935e-01 -1.34872866e+00 -7.92896226e-02 7.79191375e-01
7.01242447e-01 -8.77138793e-01 -8.22326392e-02 2.65042912e-02
-1.72018319e-01 1.39389908e+00 6.59930944e-01 -4.05813634e-01
8.67362738e-01 4.35366035e-01 2.71564752e-01 9.66155380e-02
-7.29923844e-01 1.44936079e-02 5.66738322e-02 1.76840261e-01
6.42306328e-01 -1.93936318e-01 -6.78522944e-01 4.39080829e-03
-7.18106687e-01 2.30118856e-01 3.49517226e-01 9.83694255e-01
-8.27469409e-01 -1.10871923e+00 -1.96488827e-01 3.40822279e-01
-3.81875634e-01 -1.75296769e-01 -4.65052485e-01 6.76687539e-01
-2.68199623e-01 6.27277672e-01 2.88907856e-01 -4.47221071e-01
2.00836584e-01 3.50458175e-01 6.24170005e-01 1.63915604e-01
-4.86953139e-01 -4.16019261e-01 -1.51823938e-01 -4.26640153e-01
-5.94932884e-02 -3.70878845e-01 -1.71704924e+00 -4.87447262e-01
-5.16436517e-01 4.11190063e-01 1.05289817e+00 1.05452943e+00
-3.99313308e-02 3.34175378e-01 5.94976068e-01 -6.86502934e-01
-8.12834024e-01 -6.96677446e-01 -3.73944551e-01 5.14555812e-01
3.57424617e-01 -4.57842201e-01 -6.93659604e-01 -1.54167309e-01] | [7.893991470336914, 3.5364742279052734] |
47cc4119-ec0b-4463-9e61-c3ffb2674885 | lit-tuned-models-for-efficient-species | 2302.10281 | null | https://arxiv.org/abs/2302.10281v1 | https://arxiv.org/pdf/2302.10281v1.pdf | LiT Tuned Models for Efficient Species Detection | Recent advances in training vision-language models have demonstrated unprecedented robustness and transfer learning effectiveness; however, standard computer vision datasets are image-only, and therefore not well adapted to such training methods. Our paper introduces a simple methodology for adapting any fine-grained image classification dataset for distributed vision-language pretraining. We implement this methodology on the challenging iNaturalist-2021 dataset, comprised of approximately 2.7 million images of macro-organisms across 10,000 classes, and achieve a new state-of-the art model in terms of zero-shot classification accuracy. Somewhat surprisingly, our model (trained using a new method called locked-image text tuning) uses a pre-trained, frozen vision representation, proving that language alignment alone can attain strong transfer learning performance, even on fractious, long-tailed datasets. Our approach opens the door for utilizing high quality vision-language pretrained models in agriculturally relevant applications involving species detection. | ['Chinmay Hegde', 'Benjamin Feuer', 'Andre Nakkab'] | 2023-02-12 | null | null | null | null | ['fine-grained-image-classification'] | ['computer-vision'] | [ 7.02208996e-01 -3.42304051e-01 -9.54061225e-02 -8.43179896e-02
-7.47807980e-01 -8.24473858e-01 9.20912206e-01 2.29203310e-02
-6.79152906e-01 4.51703131e-01 -2.08328918e-01 -6.71937525e-01
3.63170415e-01 -7.04500794e-01 -1.14592206e+00 -6.90535486e-01
-9.75900795e-03 4.86731052e-01 3.36160779e-01 -3.47476870e-01
2.12375581e-01 2.42435291e-01 -1.75942063e+00 3.96989107e-01
4.73319679e-01 7.97841907e-01 5.56424379e-01 1.33169007e+00
-3.53033513e-01 8.38167727e-01 -4.45027798e-01 -4.27553415e-01
3.35383713e-01 -3.25868756e-01 -6.58146381e-01 6.00562319e-02
9.44333315e-01 -4.20932740e-01 8.04963782e-02 9.20447171e-01
5.49924612e-01 -1.57200634e-01 9.27409410e-01 -8.25279891e-01
-1.30683255e+00 2.65250087e-01 -7.37483740e-01 2.29113206e-01
-2.95880511e-02 5.39235115e-01 8.52731168e-01 -8.85368526e-01
6.75256133e-01 1.16870880e+00 9.72400904e-01 7.60538459e-01
-1.41727757e+00 -4.99778181e-01 -8.35478604e-02 -7.29756206e-02
-1.11613977e+00 -5.63774467e-01 1.61775108e-02 -8.39576185e-01
1.36686182e+00 -1.58553183e-01 3.95642459e-01 1.27940166e+00
3.18675429e-01 4.10156876e-01 1.18758571e+00 -6.59383535e-01
1.85949981e-01 -1.18485978e-02 9.51269362e-03 1.04832184e+00
3.91816527e-01 2.35956386e-01 -5.21056712e-01 -9.71902534e-02
6.83737218e-01 -9.94702354e-02 1.09288655e-01 -5.74597836e-01
-1.32630062e+00 9.95064914e-01 4.68445152e-01 9.44760740e-02
2.33945549e-02 2.08594292e-01 6.90493107e-01 4.19625252e-01
3.71551365e-01 4.30041939e-01 -7.11734354e-01 1.36805266e-01
-9.30188119e-01 -1.36070743e-01 8.16421449e-01 7.94621944e-01
1.03399909e+00 2.96952456e-01 -1.37625232e-01 8.12118292e-01
5.88146746e-02 1.02508557e+00 4.53148514e-01 -9.66065884e-01
-9.25505832e-02 1.45807654e-01 -6.83744177e-02 -6.72204971e-01
-1.57662153e-01 -2.18856812e-01 -9.34292972e-01 2.95951962e-01
4.12798703e-01 -6.90179542e-02 -1.39160502e+00 1.64790845e+00
6.33404553e-02 2.31610686e-01 4.83502388e-01 4.03183997e-01
9.38155591e-01 8.69983196e-01 3.94252807e-01 1.92625877e-02
1.35290623e+00 -1.07909536e+00 4.74615060e-02 -6.23992145e-01
4.88176882e-01 -8.54322851e-01 1.33836794e+00 1.78793043e-01
-6.56384587e-01 -7.55427837e-01 -9.01972950e-01 -9.53913573e-03
-9.43171144e-01 -3.22930217e-02 9.37561929e-01 6.40659750e-01
-1.23630095e+00 2.44220003e-01 -5.38301826e-01 -1.13015306e+00
4.90156889e-01 1.13426790e-01 -5.08607984e-01 -3.24840993e-01
-7.24726379e-01 8.93870950e-01 3.26990992e-01 -2.87914604e-01
-1.70588136e+00 -8.22430849e-01 -8.37894917e-01 -1.77458495e-01
8.08477327e-02 -8.60339046e-01 1.25864577e+00 -1.16095638e+00
-1.44979024e+00 1.65005124e+00 1.16071932e-01 -7.55262613e-01
2.48185232e-01 -6.58121258e-02 -5.36335185e-02 2.02651933e-01
3.44780415e-01 1.25220180e+00 1.20900226e+00 -1.31851888e+00
-6.31191194e-01 -2.38944501e-01 6.66174144e-02 -8.59962851e-02
-5.85747242e-01 8.15099627e-02 -2.84359574e-01 -5.26024163e-01
-6.82361603e-01 -8.20902288e-01 -3.19184065e-01 5.07881641e-01
1.34865880e-01 1.94170564e-01 5.71312308e-01 -5.61075091e-01
3.63231927e-01 -1.94979620e+00 -1.14699462e-02 -4.40825075e-01
-3.82608213e-02 7.21969545e-01 -9.57369804e-01 4.13188666e-01
6.70274347e-02 3.11298762e-02 -5.63549399e-01 -1.97732858e-02
-1.63221046e-01 4.87401664e-01 -4.43839341e-01 3.74254376e-01
4.69695538e-01 1.11451709e+00 -9.60495949e-01 -3.53394806e-01
4.09544051e-01 2.72549719e-01 -5.34231246e-01 5.11765838e-01
-4.67312008e-01 2.27976620e-01 -1.26497358e-01 6.96484268e-01
6.73070431e-01 -4.18779254e-01 -7.00242370e-02 7.20420778e-02
-2.19580889e-01 -7.07333326e-01 -1.73153639e-01 2.03380203e+00
-5.77341497e-01 6.27925634e-01 1.49799600e-01 -1.20862067e+00
9.42841530e-01 -2.39894137e-01 3.06232184e-01 -7.59952307e-01
-9.91000235e-02 5.42908208e-03 -1.22916801e-02 -5.22929490e-01
2.50832617e-01 -2.40362156e-02 -8.50344449e-02 1.25691846e-01
6.07458949e-01 -6.37739897e-01 2.26117551e-01 1.26675859e-01
1.13565540e+00 4.45486128e-01 3.58286679e-01 -4.54390466e-01
3.51691365e-01 4.53746736e-01 6.95271939e-02 1.25032198e+00
-5.24528265e-01 5.28759778e-01 -2.03701723e-02 -6.37627900e-01
-1.33349323e+00 -1.16801929e+00 -8.40422511e-02 1.85212731e+00
-2.50021130e-01 -9.54920501e-02 -8.89039338e-01 -4.26404357e-01
1.03442170e-01 4.63379860e-01 -7.93916047e-01 -3.57886583e-01
-5.47203310e-02 -9.63136613e-01 9.60864305e-01 4.02696669e-01
5.60114741e-01 -1.03549159e+00 -7.88385749e-01 7.40079656e-02
2.13305622e-01 -1.16453290e+00 -2.91255444e-01 6.23262405e-01
-4.80554849e-01 -9.00188386e-01 -1.21407080e+00 -1.30879533e+00
4.29559231e-01 7.31468081e-01 1.50204229e+00 -1.89572349e-02
-8.70885670e-01 7.27242410e-01 -3.59767288e-01 -6.36894405e-01
-6.41256869e-01 4.68403567e-04 -4.13782336e-02 -4.11387712e-01
6.72199011e-01 -1.38757810e-01 -3.15822333e-01 -1.09514087e-01
-8.43781531e-01 -1.40095413e-01 7.84759223e-01 1.32568300e+00
5.21428347e-01 -5.69549382e-01 5.83621562e-01 -6.10598564e-01
2.76864052e-01 -3.43427181e-01 -8.89525890e-01 7.63807774e-01
-3.82314801e-01 7.01380223e-02 6.47200167e-01 -5.90387225e-01
-9.68861401e-01 4.39184010e-01 -7.81018585e-02 -4.26791370e-01
-4.46043104e-01 4.09256816e-01 3.77744436e-01 -5.65005422e-01
1.10352325e+00 6.86093330e-01 1.09629311e-01 -5.94367236e-02
8.73474121e-01 6.40001535e-01 9.86935616e-01 -6.32349789e-01
8.62318933e-01 4.74369735e-01 -6.03023060e-02 -1.35051286e+00
-1.02279496e+00 -6.33446515e-01 -7.50415683e-01 1.18953094e-01
1.04199636e+00 -1.36938632e+00 -5.04408658e-01 7.44027257e-01
-1.13914502e+00 -7.09774852e-01 -3.10676247e-01 6.88637868e-02
-7.70048320e-01 3.91833514e-01 -5.43789744e-01 -5.06296635e-01
-6.42983019e-01 -8.57077956e-01 1.43205523e+00 1.71598598e-01
3.79853517e-01 -8.98973227e-01 4.78567570e-01 2.31001109e-01
5.12123704e-01 7.33329449e-03 8.46240044e-01 -3.50840986e-01
-3.11131328e-01 5.11882827e-02 -6.30774081e-01 4.43680376e-01
2.01609060e-01 2.86464125e-01 -1.20861101e+00 -5.29050052e-01
-3.92202973e-01 -1.05983353e+00 1.31020379e+00 4.08311158e-01
9.55335021e-01 1.16704702e-01 -9.08614025e-02 8.47159505e-01
1.57167327e+00 -1.86996862e-01 5.14172316e-01 2.89174616e-01
7.18493462e-01 4.21179295e-01 4.50375229e-01 3.56748879e-01
3.18239510e-01 1.85300678e-01 4.93739694e-01 -3.83113712e-01
-3.10768843e-01 -1.55927554e-01 5.31504273e-01 6.18015110e-01
-2.26407945e-02 -2.30742663e-01 -1.03449833e+00 7.09219158e-01
-1.74176300e+00 -1.06568658e+00 3.38937551e-01 2.08086395e+00
8.83889139e-01 -1.34369269e-01 -8.61041620e-02 -5.84779561e-01
5.33544719e-01 8.86215866e-02 -8.54411006e-01 -4.88702655e-01
-4.49110240e-01 3.86381835e-01 8.97444546e-01 1.90076366e-01
-1.41554284e+00 1.55351460e+00 7.53061867e+00 7.65633464e-01
-1.18934202e+00 8.67703483e-02 6.07152283e-01 4.90148991e-01
3.72969806e-02 -2.58084297e-01 -8.20516646e-01 -7.34068900e-02
1.22769129e+00 -1.14590861e-01 5.91143787e-01 9.54212427e-01
-1.40777901e-01 -6.52254522e-02 -8.32733333e-01 9.48640168e-01
5.20496368e-01 -1.55142498e+00 5.21634102e-01 -2.02168077e-01
9.31399465e-01 5.56344628e-01 2.47184873e-01 5.36910772e-01
8.11317384e-01 -1.16471195e+00 4.74130660e-01 3.68087679e-01
1.16378188e+00 -4.64328617e-01 4.69498277e-01 2.74883181e-01
-1.02021515e+00 -3.30721810e-02 -9.57365513e-01 -1.97918229e-02
-4.05167729e-01 9.06710327e-02 -8.55439186e-01 2.85438597e-01
1.04976809e+00 8.82042289e-01 -1.11037922e+00 7.64664412e-01
2.61640847e-01 6.06559932e-01 -1.07081071e-01 -6.52572466e-03
4.77560580e-01 8.84358808e-02 -1.83007009e-02 1.53902960e+00
3.01223040e-01 -3.99805665e-01 4.48944092e-01 5.31615973e-01
-6.18420839e-02 -4.99825887e-02 -1.27419829e+00 -4.21850175e-01
-5.60265733e-03 1.28815484e+00 -7.32251763e-01 -4.49218065e-01
-7.75328815e-01 1.08429980e+00 4.90908712e-01 1.64655864e-01
-7.09896743e-01 -1.63287312e-01 7.47637510e-01 -4.26743805e-01
7.09532082e-01 -2.60221124e-01 7.24507198e-02 -1.32113135e+00
-6.34474218e-01 -1.07147908e+00 3.70420784e-01 -8.46809983e-01
-1.84236133e+00 4.52506930e-01 -2.92442530e-01 -9.52991009e-01
-3.50389034e-01 -1.36727571e+00 -3.66713613e-01 5.38109958e-01
-1.73147726e+00 -1.91902387e+00 -4.74110901e-01 6.81540370e-01
7.46290445e-01 -5.90416014e-01 1.40133917e+00 -5.97218126e-02
-2.03160703e-01 4.07837510e-01 4.43899125e-01 6.79683983e-02
9.76249754e-01 -1.16662967e+00 5.56261241e-01 9.20207918e-01
3.59480619e-01 4.17790890e-01 6.07664883e-01 -3.51552546e-01
-1.81123340e+00 -1.45901191e+00 3.51701498e-01 -6.75896823e-01
1.06949103e+00 -4.68937337e-01 -7.10188925e-01 6.42456472e-01
6.23701751e-01 2.37149209e-01 7.20533907e-01 -2.13562325e-01
-9.21318710e-01 -1.55128598e-01 -9.35789764e-01 4.54721719e-01
1.03414083e+00 -8.80648851e-01 -7.16976821e-01 6.03484452e-01
9.11937356e-01 1.05307341e-01 -6.24606729e-01 2.69153029e-01
5.01961052e-01 -5.13148785e-01 1.32609200e+00 -9.41529989e-01
6.15414560e-01 -2.38083392e-01 -5.66274822e-01 -1.18144143e+00
-5.04163921e-01 -3.63772064e-01 2.09398195e-01 1.20805955e+00
2.08999738e-01 -3.42771649e-01 5.61482012e-01 -2.54932255e-01
-9.31171551e-02 4.21149805e-02 -5.04981339e-01 -9.78998780e-01
6.14606380e-01 -1.82984341e-02 4.54060659e-02 8.16134512e-01
-5.17202795e-01 6.39234543e-01 -6.20823681e-01 5.45870587e-02
8.15558851e-01 2.22152874e-01 1.03190053e+00 -1.03120911e+00
-4.06976372e-01 -3.63481075e-01 -3.80200297e-01 -7.47518897e-01
3.06148320e-01 -9.02344763e-01 5.92158437e-01 -1.32353401e+00
5.84979296e-01 2.51910444e-02 -1.88630074e-01 6.92228973e-01
-7.15463236e-02 6.53427124e-01 -1.47500803e-04 3.92450541e-02
-6.22779906e-01 2.47823760e-01 9.54500675e-01 -7.18714356e-01
3.15633744e-01 -4.26077574e-01 -6.44537628e-01 6.35841846e-01
6.32313490e-01 -1.17989153e-01 -3.67571741e-01 -8.37218523e-01
-2.04310328e-01 -5.74790001e-01 6.61778927e-01 -1.02071154e+00
9.05392021e-02 -4.92477864e-01 4.41699088e-01 -2.84518898e-01
2.48208478e-01 -6.42394722e-01 -1.72158390e-01 8.14254701e-01
-3.50441962e-01 -8.57870877e-02 6.16272926e-01 6.21114016e-01
6.27182797e-02 -2.12383091e-01 1.08781493e+00 -6.40134096e-01
-1.46096814e+00 3.60396326e-01 -5.80477774e-01 1.77088797e-01
9.74785209e-01 5.11656106e-02 -6.92538261e-01 2.07494106e-02
-3.53316784e-01 -1.11536294e-01 7.54094243e-01 5.42395949e-01
4.31507051e-01 -6.09846771e-01 -9.89065349e-01 2.58688301e-01
8.62124920e-01 -3.48310322e-01 6.65390790e-02 1.50276870e-01
-8.56226385e-01 4.61494237e-01 -7.26801872e-01 -1.07298696e+00
-1.15465260e+00 1.03667319e+00 2.52054870e-01 -1.07525744e-01
-2.40770787e-01 1.08083642e+00 6.32542133e-01 -7.29763985e-01
-2.43982207e-02 -5.64507991e-02 2.33082790e-02 -8.45005810e-02
6.97592616e-01 -2.84903109e-01 -8.86195898e-02 -5.72511613e-01
-3.81382674e-01 9.64929461e-01 -1.04686534e-02 1.64140195e-01
1.41274095e+00 2.93661170e-02 -8.04470032e-02 4.65733021e-01
1.04876745e+00 -5.45415938e-01 -1.41179597e+00 -1.10180870e-01
-6.08294904e-02 -2.15716824e-01 -1.20028354e-01 -7.85853386e-01
-7.50876606e-01 1.40127265e+00 9.93301868e-01 -4.88912351e-02
9.45398033e-01 -5.48812374e-02 2.33250558e-01 1.07114816e+00
5.92717946e-01 -7.61161506e-01 4.04510260e-01 8.83465528e-01
6.12427831e-01 -1.77334011e+00 -1.87386662e-01 7.03796968e-02
-5.50929964e-01 1.04356098e+00 7.74845123e-01 -7.35512450e-02
4.34929281e-01 4.71280813e-01 3.52732748e-01 1.61917716e-01
-9.00098681e-01 -6.33391380e-01 -9.21838582e-02 1.41183138e+00
3.21175486e-01 1.17426710e-02 3.57140422e-01 -1.09516226e-01
2.14518577e-01 1.77451074e-01 3.65082741e-01 9.57366943e-01
-9.18338060e-01 -7.30208635e-01 -1.32382303e-01 2.58240491e-01
-2.24572033e-01 -5.41468322e-01 -4.45607543e-01 4.08419997e-01
3.89599651e-02 8.33419561e-01 9.77557674e-02 -1.63118526e-01
-1.72111311e-03 -9.54658464e-02 6.51296377e-01 -8.30519974e-01
-4.68008518e-01 -2.05483928e-01 -1.02929726e-01 -3.56221318e-01
-6.44646823e-01 -3.79153520e-01 -7.70795763e-01 -2.91176707e-01
-4.24790457e-02 -2.78383076e-01 6.39599323e-01 6.87520385e-01
3.61625701e-01 2.51452208e-01 3.43975276e-01 -1.15010810e+00
-6.88114166e-01 -6.93582773e-01 -2.95394808e-01 3.32064062e-01
4.97038603e-01 -3.17652851e-01 -1.23831160e-01 7.76422381e-01] | [10.053557395935059, 2.0209317207336426] |
cff263b9-0076-476f-99dc-c406dd09e0b2 | why-having-10000-parameters-in-your-camera-1 | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Schops_Why_Having_10000_Parameters_in_Your_Camera_Model_Is_Better_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Schops_Why_Having_10000_Parameters_in_Your_Camera_Model_Is_Better_CVPR_2020_paper.pdf | Why Having 10,000 Parameters in Your Camera Model Is Better Than Twelve | Camera calibration is an essential first step in setting up 3D Computer Vision systems. Commonly used parametric camera models are limited to a few degrees of freedom and thus often do not optimally fit to complex real lens distortion. In contrast, generic camera models allow for very accurate calibration due to their flexibility. Despite this, they have seen little use in practice. In this paper, we argue that this should change. We propose a calibration pipeline for generic models that is fully automated, easy to use, and can act as a drop-in replacement for parametric calibration, with a focus on accuracy. We compare our results to parametric calibrations. Considering stereo depth estimation and camera pose estimation as examples, we show that the calibration error acts as a bias on the results. We thus argue that in contrast to current common practice, generic models should be preferred over parametric ones whenever possible. To facilitate this, we released our calibration pipeline at https://github.com/puzzlepaint/camera_calibration, making both easy-to-use and accurate camera calibration available to everyone.
| [' Torsten Sattler', ' Marc Pollefeys', ' Viktor Larsson', 'Thomas Schops'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['stereo-depth-estimation'] | ['computer-vision'] | [-1.38561383e-01 -2.88526993e-02 1.90560278e-02 -3.58105630e-01
-4.36213702e-01 -9.67857420e-01 5.06327569e-01 -2.06094012e-01
-3.39627236e-01 4.44922298e-01 8.08049738e-02 -2.81472892e-01
1.33579627e-01 -4.02975440e-01 -8.55733633e-01 -3.16092640e-01
6.61846399e-01 7.26737022e-01 3.57279003e-01 6.36105612e-02
4.57828432e-01 6.74374163e-01 -1.37549305e+00 -3.43693316e-01
6.48853481e-01 3.59900177e-01 3.84774953e-01 6.89337254e-01
9.32381153e-02 4.10232395e-01 -4.08562720e-01 -7.15708673e-01
5.29546857e-01 -2.00981632e-01 -6.16117656e-01 5.05748391e-01
6.36155188e-01 -7.66819060e-01 -1.94171995e-01 8.84075701e-01
5.20187914e-01 -2.52293348e-01 3.79657835e-01 -1.13338220e+00
-3.22522640e-01 1.76520392e-01 -5.71087480e-01 -1.85425162e-01
6.04070306e-01 1.33458123e-01 7.40842044e-01 -6.65958166e-01
8.18148911e-01 9.30537224e-01 9.97023046e-01 4.85513985e-01
-1.29099143e+00 -3.68067741e-01 1.68762729e-01 -1.20990001e-01
-1.33157754e+00 -5.62968433e-01 8.89114678e-01 -5.39037347e-01
7.28565276e-01 1.14708088e-01 9.92953598e-01 1.31178761e+00
-1.23492546e-01 5.34931123e-01 9.97007608e-01 -6.67018414e-01
1.73001513e-01 2.81364828e-01 1.47655699e-02 4.60349262e-01
6.64069593e-01 6.87290281e-02 -3.53970617e-01 -7.50108659e-02
1.45915675e+00 -1.82378292e-01 -5.12291789e-01 -1.07556248e+00
-1.22010827e+00 7.70472944e-01 1.07750967e-01 2.84019820e-02
-2.17302777e-02 2.67225534e-01 2.86340620e-02 1.71571225e-01
6.36430979e-02 5.17899632e-01 -3.77636522e-01 -4.72744137e-01
-7.86319196e-01 4.38245744e-01 8.26086164e-01 1.18084407e+00
6.95125878e-01 -4.26954806e-01 6.00669503e-01 8.65993917e-01
2.55644888e-01 2.80310780e-01 1.20444059e-01 -1.51953232e+00
7.19065070e-02 3.31730723e-01 2.76956350e-01 -9.37127829e-01
-3.68298978e-01 -3.91329676e-01 -4.11647677e-01 5.11417150e-01
6.97689593e-01 9.60237533e-02 -4.11480129e-01 1.44682837e+00
3.10411036e-01 7.43871778e-02 -3.36743534e-01 9.48470414e-01
4.21037555e-01 5.88147901e-02 -5.91397405e-01 1.51067395e-02
1.14791942e+00 -8.48665178e-01 -5.07536352e-01 -4.49890912e-01
6.41566515e-01 -1.30094743e+00 1.33008742e+00 6.68205559e-01
-1.09926033e+00 -2.19041020e-01 -1.16936612e+00 -3.49850059e-01
2.82062758e-02 1.02424800e-01 6.70160472e-01 8.61387730e-01
-1.09189463e+00 4.96766329e-01 -8.69971693e-01 -8.04262459e-01
-4.39320169e-02 2.60998666e-01 -5.42157292e-01 -2.27706626e-01
-4.11227822e-01 1.23061085e+00 2.10777640e-01 -1.55289009e-01
-2.34219700e-01 -6.31301403e-01 -6.10814571e-01 -3.17086518e-01
5.20194948e-01 -1.10047364e+00 1.69204473e+00 -8.62981617e-01
-1.76675642e+00 1.13453043e+00 -5.83373904e-02 -2.03856796e-01
9.53697443e-01 -3.86719406e-01 2.36209363e-01 1.66919991e-01
-1.08631782e-01 7.77649343e-01 6.35623097e-01 -1.42470288e+00
-2.50214458e-01 -6.63448200e-02 4.86141950e-01 2.67769843e-01
1.25902265e-01 -4.66187038e-02 -9.93329048e-01 -3.71345550e-01
3.13032448e-01 -1.24351001e+00 -1.36886373e-01 3.88158619e-01
-1.14298485e-01 4.75882679e-01 7.29845047e-01 -5.20529270e-01
1.03017163e+00 -1.94683945e+00 1.00548953e-01 4.95325513e-02
2.32412629e-02 -1.93642955e-02 2.41455287e-01 3.00992310e-01
9.10780802e-02 -6.07703812e-02 -2.54870713e-01 -7.26264358e-01
7.56122693e-02 3.95220459e-01 -6.17776066e-02 6.59808576e-01
-1.34150439e-03 6.15860224e-01 -6.43915832e-01 -2.81767815e-01
7.29302585e-01 6.80120111e-01 -7.19582319e-01 1.33126214e-01
-8.87224674e-02 5.69758236e-01 -2.41743606e-02 6.37762666e-01
1.01942933e+00 -1.33824334e-01 1.83488905e-01 -6.29877746e-02
-2.84325421e-01 1.83925018e-01 -1.48878694e+00 1.87725329e+00
-3.86065215e-01 7.15740025e-01 2.20745459e-01 -3.80680561e-01
8.53416741e-01 1.75105520e-02 1.51884615e-01 -2.62969792e-01
1.35821208e-01 4.66305792e-01 -2.59853974e-02 -9.44452360e-02
5.41974545e-01 2.49175765e-02 2.84232050e-01 3.96088481e-01
-1.52607858e-01 -9.79816020e-01 1.43542469e-01 1.61913946e-01
6.91582203e-01 6.46804094e-01 6.94718897e-01 -1.45663500e-01
2.22327098e-01 2.05454990e-01 4.03930843e-01 4.57554191e-01
-2.89258584e-02 1.37312913e+00 5.28381109e-01 -3.16069067e-01
-1.44121695e+00 -7.73761094e-01 -1.45335227e-01 2.13153064e-01
2.54914820e-01 -5.86166024e-01 -1.01629400e+00 -2.02313751e-01
-1.48266554e-01 6.62923515e-01 -5.46161771e-01 2.52795249e-01
-6.55019164e-01 -2.76985198e-01 2.19008714e-01 4.10390645e-01
3.18433493e-01 -4.64909613e-01 -1.08211982e+00 5.75443320e-02
2.25776155e-02 -1.28943002e+00 -6.07035398e-01 9.75634065e-03
-1.01586974e+00 -1.31245065e+00 -8.36162508e-01 -4.32018936e-01
8.32261264e-01 7.51338899e-01 1.25833905e+00 2.04632416e-01
2.53527462e-02 8.83446097e-01 -4.30741847e-01 -3.17676008e-01
-3.35259259e-01 -3.14645618e-02 -1.20270774e-01 -4.29480642e-01
2.32165426e-01 -6.32393539e-01 -7.01581657e-01 5.85125387e-01
-8.06453943e-01 2.70359755e-01 4.28958595e-01 6.69438183e-01
4.49690133e-01 -4.98772979e-01 -5.16030073e-01 -8.19297254e-01
4.87928659e-01 1.93574037e-02 -1.05841100e+00 -4.09458503e-02
-5.52683115e-01 -6.15684316e-02 3.79368842e-01 -3.94444019e-01
-7.08341122e-01 2.82132238e-01 -1.63219780e-01 -6.80388272e-01
-3.04065347e-01 2.42192671e-01 -7.82871395e-02 -4.16061550e-01
6.47163332e-01 -1.93986714e-01 1.38606250e-01 -5.98436058e-01
2.04051599e-01 3.39157701e-01 5.76214731e-01 -6.38533354e-01
8.79982948e-01 5.44504941e-01 3.48603576e-02 -8.99320900e-01
-4.18708146e-01 -4.04646009e-01 -9.74708259e-01 -2.60853678e-01
4.82246995e-01 -8.15015018e-01 -4.49457139e-01 3.99425924e-01
-1.21743333e+00 -3.88142526e-01 -1.23136073e-01 4.84790742e-01
-8.14582765e-01 7.28748262e-01 -2.35215187e-01 -5.34919739e-01
2.69957542e-01 -1.56352174e+00 1.08133471e+00 2.49828711e-01
-4.76023942e-01 -1.06550002e+00 1.20187260e-01 3.18381816e-01
3.47447336e-01 3.00188243e-01 4.53658700e-01 7.84997866e-02
-6.34170473e-01 -2.14059651e-01 3.87757272e-02 2.34043419e-01
2.75152251e-02 6.00268543e-01 -1.05120599e+00 -1.78835005e-01
1.84220821e-02 6.67294264e-02 2.28017136e-01 5.43730378e-01
9.05362010e-01 1.06782392e-01 -7.84819648e-02 8.78972471e-01
1.48718977e+00 -9.12089571e-02 7.62735903e-01 8.80611002e-01
6.20281577e-01 5.57505727e-01 5.40394604e-01 3.56249422e-01
5.76007664e-01 1.16803694e+00 5.29984295e-01 8.25935379e-02
-1.40228063e-01 -2.52390355e-01 2.34984532e-01 7.28383780e-01
-1.31682783e-01 1.32097572e-01 -1.11267328e+00 3.33668798e-01
-1.78234720e+00 -6.03893816e-01 -5.35469770e-01 2.61699915e+00
4.91913766e-01 1.51512548e-01 3.47852707e-01 1.56077549e-01
6.53356016e-01 -2.94465631e-01 -1.82011083e-01 -5.17768264e-01
-2.14368582e-01 -1.11399785e-01 6.93544686e-01 7.36148417e-01
-7.80529082e-01 6.93066418e-01 6.25667953e+00 3.23844045e-01
-1.21790850e+00 1.53418601e-01 2.16929585e-01 -1.66322306e-01
-3.69322956e-01 4.76352125e-01 -6.04706883e-01 4.68344688e-01
6.02755666e-01 4.16301191e-02 5.03630519e-01 8.22883427e-01
2.74103522e-01 -5.23823261e-01 -1.27109647e+00 1.41479743e+00
-4.03103717e-02 -1.18629968e+00 -3.32649350e-01 2.21079081e-01
5.24857402e-01 -1.88196599e-01 -7.07590580e-02 -2.52360493e-01
1.51770398e-01 -8.48549604e-01 1.00442672e+00 2.62550473e-01
5.86570024e-01 -6.07021093e-01 6.33111179e-01 3.92950863e-01
-8.01706076e-01 1.98272154e-01 -5.86684883e-01 -1.03164241e-01
3.52511227e-01 5.19039929e-01 -7.01116741e-01 3.61674070e-01
6.65303707e-01 5.16362429e-01 -7.00797141e-01 1.39269841e+00
-4.55698103e-01 1.94380268e-01 -5.60624897e-01 4.08696532e-01
6.29861727e-02 -5.79108477e-01 3.64877552e-01 1.04966116e+00
5.28574944e-01 1.12640718e-02 -1.49545714e-01 6.10920846e-01
2.07682431e-01 1.35865630e-04 -8.53021801e-01 3.13970357e-01
5.84717512e-01 1.01711297e+00 -9.51928020e-01 9.08231512e-02
-6.38973653e-01 1.16729426e+00 2.91598588e-01 -3.32074836e-02
-8.84276211e-01 7.70451650e-02 6.26334786e-01 3.97325933e-01
3.02272230e-01 -5.92115283e-01 -5.02896190e-01 -1.44499886e+00
2.29552701e-01 -8.15765142e-01 -1.63087193e-02 -1.16565681e+00
-8.11596572e-01 2.18382165e-01 2.45664433e-01 -1.54456162e+00
-4.27449584e-01 -7.63628960e-01 -3.29257607e-01 6.21003032e-01
-1.39931285e+00 -1.24270451e+00 -6.50224507e-01 4.12283987e-01
5.29818237e-01 4.37295318e-01 7.49867022e-01 1.82104394e-01
-3.69886875e-01 5.39465427e-01 1.37174055e-02 -2.72976041e-01
1.05880320e+00 -1.21065438e+00 5.71326315e-01 9.99405444e-01
9.86836851e-02 9.93149281e-01 1.01417124e+00 -3.52807403e-01
-1.52214313e+00 -4.41765070e-01 5.76927602e-01 -7.81126499e-01
4.92467999e-01 -4.20437723e-01 -7.30924189e-01 8.34065497e-01
2.13307306e-01 -2.41710588e-01 4.75091636e-01 -3.58841345e-02
-4.15774107e-01 1.13975152e-01 -1.06668925e+00 7.16958463e-01
1.04019868e+00 -3.86999846e-01 -4.98200208e-01 2.23013818e-01
4.42182809e-01 -7.95974493e-01 -7.18383789e-01 1.02626281e-02
7.44505644e-01 -1.60812116e+00 9.41300452e-01 3.35785836e-01
2.23893836e-01 -5.72483420e-01 -2.07083061e-01 -1.16347432e+00
-2.91520804e-01 -8.10407519e-01 1.03786364e-01 1.13703084e+00
9.32270437e-02 -7.42384434e-01 7.48861611e-01 7.86761701e-01
-1.74571678e-01 -3.15598369e-01 -8.79305422e-01 -7.62639105e-01
1.44880906e-01 -6.16580784e-01 4.63105798e-01 1.01820314e+00
-2.94520110e-01 4.29841839e-02 -4.88900036e-01 2.16188937e-01
5.94578028e-01 -9.86453816e-02 1.42341745e+00 -1.35157013e+00
-5.96784055e-01 -4.84306216e-01 -3.96705031e-01 -1.22855961e+00
-2.06367210e-01 -2.54820943e-01 -3.19791436e-01 -1.33934700e+00
1.80060819e-01 -4.15537149e-01 6.93064451e-01 1.57344446e-01
1.56789914e-01 4.94920045e-01 2.84263998e-01 4.09603417e-01
-2.92196237e-02 -4.57357541e-02 1.03752089e+00 3.91247660e-01
-1.35288611e-01 -2.55886018e-02 -8.85148406e-01 8.42985511e-01
8.39830697e-01 -2.61795074e-01 -3.47586185e-01 -8.25374544e-01
3.49657416e-01 -2.41021827e-01 4.68887955e-01 -1.00629902e+00
2.84133464e-01 -9.32017714e-02 1.53160989e-01 -4.16971385e-01
5.37768006e-01 -1.09761977e+00 6.91067040e-01 2.43928775e-01
2.71435857e-01 2.66694576e-01 2.80055553e-01 -1.25974938e-02
-4.39749518e-03 -6.19418740e-01 9.13124323e-01 -3.92987579e-01
-6.59986794e-01 7.02603087e-02 -5.07950336e-02 -3.58087212e-01
9.77114201e-01 -9.63721633e-01 -2.03014642e-01 -6.09747350e-01
-4.41617697e-01 -1.12142600e-01 1.57157707e+00 2.48880327e-01
4.01310712e-01 -1.05198658e+00 -4.67404425e-01 1.73729032e-01
1.81163788e-01 2.87928522e-01 7.79774413e-02 8.34205449e-01
-1.21948564e+00 5.37584066e-01 -2.13173866e-01 -8.32776725e-01
-1.40265679e+00 5.01936197e-01 3.23316991e-01 3.00169379e-01
-7.22744405e-01 6.05104625e-01 1.10619804e-02 -5.56546032e-01
1.56113416e-01 -5.22683024e-01 2.41385475e-01 -1.88131273e-01
2.67980337e-01 3.17880422e-01 2.41316244e-01 -5.87590694e-01
-3.36011201e-01 1.20936418e+00 1.52273506e-01 -3.30181777e-01
1.24707448e+00 -5.28877139e-01 1.14369998e-03 3.82543325e-01
8.20096910e-01 4.50779229e-01 -1.62416804e+00 1.65865690e-01
-1.92738533e-01 -9.50569332e-01 -6.19295090e-02 -5.99366665e-01
-8.52970302e-01 8.92270863e-01 3.11102003e-01 1.31866559e-02
1.02105510e+00 -3.08095906e-02 2.20945403e-01 1.53477311e-01
7.32048213e-01 -9.07424688e-01 -3.46626699e-01 4.24776316e-01
8.94021690e-01 -1.11533368e+00 3.24780792e-01 -7.75565803e-01
-5.78135133e-01 1.39297616e+00 5.26295245e-01 -2.82042623e-01
1.88584238e-01 4.77601439e-01 2.89991677e-01 1.18133649e-01
-4.67248768e-01 4.68556061e-02 -8.51560161e-02 9.76541519e-01
4.66224164e-01 -2.25973830e-01 -2.30720684e-01 1.36937406e-02
-5.09096622e-01 6.49411976e-02 1.11666453e+00 9.83585000e-01
-1.26052529e-01 -1.63813651e+00 -7.88054585e-01 -1.34273022e-01
-7.44756162e-02 1.29453719e-01 -5.24926484e-01 1.17996991e+00
-1.59362614e-01 8.15092444e-01 -1.43836975e-01 -1.31338090e-01
4.94252145e-01 -1.66946277e-01 1.00218701e+00 -4.28797424e-01
-4.57334280e-01 2.26439863e-01 2.04839222e-02 -5.66285431e-01
-3.94530475e-01 -8.88570249e-01 -6.15546882e-01 -5.97479761e-01
-3.06979418e-01 -1.72902226e-01 9.43548977e-01 6.00257993e-01
3.28632295e-01 9.64927208e-03 2.91501582e-01 -1.26886070e+00
-4.73422766e-01 -6.40136242e-01 -2.80180126e-01 2.93385625e-01
2.57758677e-01 -9.25040126e-01 -4.53378648e-01 1.53421521e-01] | [8.237353324890137, -2.3849596977233887] |
e7dde1a7-0c0b-4a22-86cc-d66f2f2e23be | wifi-based-multi-task-sensing | 2111.14619 | null | https://arxiv.org/abs/2111.14619v1 | https://arxiv.org/pdf/2111.14619v1.pdf | WiFi-based Multi-task Sensing | WiFi-based sensing has aroused immense attention over recent years. The rationale is that the signal fluctuations caused by humans carry the information of human behavior which can be extracted from the channel state information of WiFi. Still, the prior studies mainly focus on single-task sensing (STS), e.g., gesture recognition, indoor localization, user identification. Since the fluctuations caused by gestures are highly coupling with body features and the user's location, we propose a WiFi-based multi-task sensing model (Wimuse) to perform gesture recognition, indoor localization, and user identification tasks simultaneously. However, these tasks have different difficulty levels (i.e., imbalance issue) and need task-specific information (i.e., discrepancy issue). To address these issues, the knowledge distillation technique and task-specific residual adaptor are adopted in Wimuse. We first train the STS model for each task. Then, for solving the imbalance issue, the extracted common feature in Wimuse is encouraged to get close to the counterpart features of the STS models. Further, for each task, a task-specific residual adaptor is applied to extract the task-specific compensation feature which is fused with the common feature to address the discrepancy issue. We conduct comprehensive experiments on three public datasets and evaluation suggests that Wimuse achieves state-of-the-art performance with the average accuracy of 85.20%, 98.39%, and 98.725% on the joint task of gesture recognition, indoor localization, and user identification, respectively. | ['Kang Yin', 'Yasong An', 'Chengpei Tang', 'Xie Zhang'] | 2021-11-26 | null | null | null | null | ['indoor-localization'] | ['computer-vision'] | [ 3.66034299e-01 -6.48514688e-01 -1.70749605e-01 -3.43344033e-01
-8.81949842e-01 -1.94522202e-01 2.83361793e-01 -6.50518596e-01
-2.79516190e-01 5.28091729e-01 3.18365604e-01 5.25735617e-02
-2.92420298e-01 -3.22481573e-01 -4.27300245e-01 -8.97323489e-01
2.61177123e-01 -2.70417333e-01 2.31773838e-01 2.12396875e-01
-6.00282550e-02 -1.68511331e-01 -1.33698034e+00 2.55932547e-02
9.45888758e-01 1.31641972e+00 2.37628341e-01 3.95785272e-01
2.03831419e-02 4.17325079e-01 -6.03465199e-01 1.57605931e-01
1.18512653e-01 -3.42170775e-01 -2.63548255e-01 -2.36309186e-01
-1.52670154e-02 -3.12803775e-01 -4.61219996e-01 8.88073683e-01
9.28682804e-01 6.43775836e-02 3.31120282e-01 -1.42243516e+00
-7.76828900e-02 3.71248752e-01 -7.11126983e-01 8.47332180e-02
6.95718050e-01 8.11826065e-02 4.77644950e-01 -8.22708666e-01
-5.65989390e-02 1.04443157e+00 8.82448971e-01 5.36637068e-01
-7.20325053e-01 -1.30438507e+00 2.34242961e-01 1.09491356e-01
-1.74004555e+00 -4.56387192e-01 7.45156586e-01 -2.60178894e-01
4.13374752e-01 4.66068923e-01 4.16175425e-01 1.42662549e+00
-3.76938768e-02 8.52807462e-01 1.00306559e+00 -4.49695326e-02
-8.92965347e-02 -1.48176819e-01 9.01111290e-02 2.76505083e-01
2.63127178e-01 1.58625364e-01 -8.58531296e-01 -8.62351134e-02
7.12761998e-01 5.42706251e-01 -4.43666488e-01 1.35192797e-01
-1.46566260e+00 1.07124053e-01 3.32429290e-01 5.70831418e-01
-4.48615581e-01 1.42590448e-01 1.75640017e-01 3.39688770e-02
3.70308645e-02 -1.51777923e-01 -5.17428875e-01 -5.78426242e-01
-8.76628995e-01 -2.60261092e-02 6.25440717e-01 1.01107287e+00
6.21921420e-01 -1.00377031e-01 -5.16617954e-01 7.24515796e-01
8.13327551e-01 1.07276952e+00 6.77258730e-01 -4.98142481e-01
8.04809809e-01 4.15258944e-01 2.64347792e-01 -1.07995951e+00
-4.74279165e-01 -6.48362994e-01 -1.02357686e+00 -5.49761474e-01
3.68540853e-01 -6.19865537e-01 -6.29071057e-01 1.96022737e+00
3.83845896e-01 1.01205742e+00 -1.73652798e-01 1.02236271e+00
7.97810376e-01 4.28117394e-01 8.10987577e-02 -3.56294513e-01
1.34479237e+00 -6.87490284e-01 -8.07896197e-01 -3.77688378e-01
3.43177348e-01 -7.76661098e-01 9.55481112e-01 2.96463788e-01
-3.09452921e-01 -8.56234014e-01 -9.40842867e-01 5.94645023e-01
8.87043625e-02 3.69104296e-01 5.69546759e-01 1.04838359e+00
-2.95786560e-01 6.50899410e-02 -1.03329349e+00 -3.28614622e-01
1.99204326e-01 4.87944692e-01 -1.51318694e-02 -1.53379813e-01
-1.36533225e+00 2.39723951e-01 4.83716950e-02 4.64368939e-01
-4.76230234e-01 -4.30986762e-01 -5.68866909e-01 -1.39754921e-01
4.70622152e-01 -3.40102077e-01 1.19904792e+00 -6.92539573e-01
-1.50147700e+00 2.31027305e-02 -4.75184143e-01 1.33082062e-01
4.37264711e-01 -3.29512358e-01 -1.01927209e+00 -4.84242618e-01
1.13000453e-01 -6.85689449e-02 6.93069100e-01 -9.23046708e-01
-7.98458338e-01 -4.98958826e-01 -2.71786690e-01 2.38103271e-01
-5.37072599e-01 -1.10786296e-01 -7.66627550e-01 -6.49549246e-01
5.67154229e-01 -1.02877378e+00 3.35084647e-02 -3.99264842e-01
-4.96511549e-01 -7.60920644e-02 8.95822406e-01 -8.03752363e-01
1.75946736e+00 -2.42084146e+00 -1.72456324e-01 5.05193353e-01
3.87521498e-02 1.77866846e-01 -2.46924609e-02 8.49322677e-02
3.49021524e-01 -1.21932939e-01 -6.11394420e-02 -2.79393792e-01
1.79793127e-02 1.30841985e-01 -8.37198645e-02 3.61455947e-01
-3.65675569e-01 8.17541957e-01 -9.42703843e-01 -4.45709378e-01
3.21084380e-01 4.81955349e-01 -2.75859743e-01 1.74248800e-01
4.92397457e-01 8.40169728e-01 -1.12072897e+00 8.93972516e-01
8.88235271e-01 -1.83142647e-01 1.10845983e-01 -5.74884653e-01
-8.69088247e-02 1.72779724e-01 -1.63654006e+00 2.05166984e+00
-6.37731910e-01 1.85325161e-01 9.71011668e-02 -6.85321331e-01
8.34144771e-01 5.30673742e-01 7.33306348e-01 -7.35820472e-01
7.83347338e-02 3.58764708e-01 2.99773477e-02 -7.38273621e-01
-1.09786436e-01 1.49966925e-01 -2.98095644e-01 5.07804334e-01
-2.29899794e-01 5.76746345e-01 -4.84467119e-01 -1.95985541e-01
1.31837904e+00 3.32093388e-01 1.53662309e-01 4.05066982e-02
6.04854047e-01 -5.77982545e-01 8.74497354e-01 8.23964775e-01
-4.73558575e-01 4.04095858e-01 -3.03633153e-01 -6.82128072e-02
1.29771261e-02 -1.01635265e+00 3.58155966e-02 1.09017205e+00
6.45836592e-01 -4.33440208e-01 -4.82770503e-01 -6.75042570e-01
1.05349518e-01 7.35768005e-02 -4.59273666e-01 -2.92948574e-01
-5.67933440e-01 -9.17879343e-01 8.49489808e-01 6.16438627e-01
1.19748914e+00 -8.38700056e-01 -6.57724142e-01 3.74359041e-01
-6.94379807e-01 -1.07098949e+00 -6.14632130e-01 -9.02458876e-02
-4.90427017e-01 -9.71865296e-01 -8.24016988e-01 -4.04844373e-01
2.20337152e-01 7.07871377e-01 4.21076626e-01 4.39797044e-02
6.77333400e-02 3.12932283e-01 -3.62498105e-01 -2.13558689e-01
5.87913036e-01 2.95868546e-01 2.23042846e-01 5.75033069e-01
5.69975495e-01 -7.05432117e-01 -6.67048991e-01 7.87490785e-01
-5.26791692e-01 -1.98441654e-01 8.30442607e-01 6.26170933e-01
2.54209459e-01 -7.60836713e-03 4.92861688e-01 -3.40543538e-01
4.73288119e-01 -6.18043780e-01 -7.23657608e-02 2.62515366e-01
-4.64022100e-01 -1.39318019e-01 2.79336214e-01 -7.12341070e-01
-1.21721566e+00 3.31044197e-01 -2.38227725e-01 -3.07639092e-01
-2.28306547e-01 5.44286668e-01 -7.04346657e-01 -2.35319123e-01
3.68054360e-01 5.34799933e-01 -4.80566651e-01 -5.89929819e-01
-2.80368030e-01 1.25284183e+00 5.15383065e-01 -4.64746237e-01
8.14393640e-01 1.67337000e-01 -2.79563308e-01 -6.48220241e-01
-7.55638301e-01 -8.14269245e-01 -1.82368457e-01 -3.50933760e-01
6.07289970e-01 -1.13337994e+00 -9.16116416e-01 8.77652407e-01
-9.64795589e-01 -1.86809629e-01 3.98894191e-01 9.03646469e-01
-5.05065694e-02 2.94055402e-01 -2.79176861e-01 -1.15287805e+00
-2.48791859e-01 -1.13096178e+00 1.29360712e+00 5.82466900e-01
-8.08216184e-02 -3.43684018e-01 -3.58416028e-02 3.68466944e-01
6.18772566e-01 1.09842971e-01 1.23009272e-02 -2.17501983e-01
-4.80587155e-01 -2.04565868e-01 -3.87794167e-01 -1.51066661e-01
5.19416094e-01 -5.29782295e-01 -1.27756965e+00 -1.66484177e-01
1.82039980e-02 -2.04174034e-02 5.87705970e-01 3.49308401e-01
1.16395104e+00 -1.22808993e-01 -8.65701556e-01 7.71561027e-01
1.06613564e+00 3.15642506e-01 5.85936487e-01 5.43653518e-02
9.11701024e-01 6.61252066e-02 8.08184028e-01 5.98788798e-01
4.09888327e-01 1.04869390e+00 1.56209201e-01 1.88566014e-01
1.55638577e-02 -3.46524060e-01 3.61855000e-01 5.98503172e-01
-2.07368240e-01 -1.19261563e-01 -8.21568072e-01 1.59366429e-01
-2.13965845e+00 -7.40618408e-01 -2.13660330e-01 2.27840686e+00
5.49691021e-01 7.31488913e-02 6.00082567e-03 2.86023408e-01
7.71611333e-01 2.22347870e-01 -7.25964010e-01 5.26569247e-01
2.02850565e-01 -1.01401232e-01 6.06119275e-01 1.99417487e-01
-1.22897005e+00 6.33564949e-01 4.90828419e+00 9.93161440e-01
-1.24947691e+00 4.03245091e-01 1.16175465e-01 -7.70122111e-02
2.97455825e-02 -3.19539756e-01 -7.29479253e-01 1.04370058e+00
5.38896203e-01 2.83629388e-01 5.00644326e-01 6.52998149e-01
2.17432708e-01 -1.14870213e-01 -6.40905917e-01 1.55447280e+00
-7.11406544e-02 -5.56492627e-01 -5.53994834e-01 1.52404502e-01
3.71643722e-01 4.09575254e-02 -1.67766567e-02 5.37451029e-01
-1.39323980e-01 -7.75982618e-01 4.53447223e-01 6.01510823e-01
9.67108190e-01 -4.72153157e-01 9.08921301e-01 5.13064146e-01
-1.99896717e+00 -2.26989508e-01 5.98202311e-02 -3.29782099e-01
2.35860467e-01 8.55391622e-01 -1.19614802e-01 8.25069845e-01
8.25741768e-01 7.81556666e-01 -2.86343634e-01 9.85332668e-01
-2.46616825e-01 7.85026908e-01 -6.65503502e-01 -9.77999717e-02
-8.38117450e-02 -4.67625223e-02 2.63974994e-01 1.06601179e+00
6.68960035e-01 1.91012233e-01 4.55022275e-01 4.51197296e-01
1.71581283e-01 -9.72443745e-02 -3.43430042e-01 3.29359144e-01
8.89100015e-01 1.18640447e+00 -2.69807845e-01 -9.66492146e-02
-4.53563899e-01 9.93533552e-01 -3.28022271e-01 6.04929745e-01
-1.06839728e+00 -5.32816648e-01 8.48722339e-01 -2.73008525e-01
1.03253640e-01 -2.60849983e-01 -3.43225151e-01 -1.29318309e+00
3.08910251e-01 -6.16913021e-01 1.50889814e-01 -4.89421964e-01
-9.63112235e-01 2.82329768e-01 -3.40988100e-01 -1.50049353e+00
-1.72637388e-01 -1.11848973e-01 -7.91760921e-01 9.91939306e-01
-1.34286165e+00 -1.15720260e+00 -9.88809407e-01 9.13723946e-01
2.46499524e-01 1.26943335e-01 7.72258937e-01 8.16972673e-01
-9.45694208e-01 1.05969048e+00 1.78909451e-02 2.74953932e-01
7.97650576e-01 -6.68733299e-01 1.31786123e-01 8.60207140e-01
-1.12867095e-01 7.52444506e-01 4.05117393e-01 -6.99412346e-01
-1.63463390e+00 -1.10454726e+00 8.39205980e-01 -2.97362447e-01
2.79526681e-01 -4.57711190e-01 -4.79475826e-01 5.88610530e-01
-5.10774553e-01 2.28893727e-01 7.25213945e-01 1.79551765e-01
-1.70557976e-01 -4.91395384e-01 -1.04926908e+00 2.98885345e-01
1.56929088e+00 -5.48770487e-01 -2.66346782e-01 1.09646656e-02
3.43662709e-01 -4.93119389e-01 -6.28538907e-01 4.96745110e-01
1.22160423e+00 -7.03107774e-01 1.06486392e+00 1.19078360e-01
-2.76621908e-01 -5.92713654e-01 -3.79226506e-01 -9.97715294e-01
-2.27808669e-01 -5.61808348e-01 -3.25932175e-01 1.33339596e+00
1.02498151e-01 -8.73363316e-01 7.35754371e-01 5.65153599e-01
7.45003000e-02 -6.05823278e-01 -1.05872107e+00 -7.89534211e-01
-9.29189563e-01 -6.48566306e-01 1.07785118e+00 8.53662610e-01
-2.39416696e-02 3.03491265e-01 -7.83748984e-01 2.39502862e-01
5.55758297e-01 1.01998881e-01 9.59783614e-01 -1.39120305e+00
-4.33245361e-01 -7.15820491e-02 -7.88909122e-02 -1.55346382e+00
-2.36540526e-01 -3.85576040e-01 2.42048144e-01 -1.32994449e+00
3.49269211e-02 -8.20940554e-01 -7.12751627e-01 5.13456047e-01
-3.54567587e-01 1.70322806e-01 1.24424644e-01 3.25162977e-01
-7.93535948e-01 5.11511028e-01 1.11664999e+00 -6.40345067e-02
-3.94191951e-01 4.70550537e-01 -7.47700214e-01 6.00712121e-01
7.56761134e-01 -3.31316382e-01 -2.74557322e-01 -5.01514435e-01
-4.07369584e-02 1.68773904e-01 3.65234315e-01 -1.51601648e+00
4.30822432e-01 -3.36627990e-01 6.11422896e-01 -3.06576550e-01
4.73523349e-01 -1.09576178e+00 2.98177391e-01 7.18383789e-01
1.22744873e-01 -3.35100949e-01 -7.21699297e-02 6.71790659e-01
-5.46992682e-02 3.27445775e-01 6.03252873e-02 2.39724234e-01
-7.61167884e-01 4.69911546e-01 -1.61361501e-01 -1.62199363e-01
6.75971866e-01 -3.23574513e-01 -1.09119065e-01 -4.14098948e-01
-3.14734221e-01 3.25222939e-01 -5.84208630e-02 5.31211376e-01
4.52068716e-01 -1.62806833e+00 -4.26635206e-01 5.36709845e-01
2.11457402e-01 -1.41492411e-01 5.23400605e-01 1.10762203e+00
3.63365978e-01 4.08185303e-01 -2.35544667e-02 -7.22630382e-01
-1.06468964e+00 -6.55783415e-02 2.73157805e-01 -2.03602985e-01
-1.56770363e-01 8.00883293e-01 1.45289127e-03 -4.46198225e-01
4.31644380e-01 -4.36242700e-01 -9.85345542e-02 -1.12643175e-01
6.25004649e-01 4.52989876e-01 -5.05984090e-02 -5.79083920e-01
-1.02763581e+00 1.03527856e+00 4.54717219e-01 -5.68831526e-02
7.68523335e-01 -3.38842928e-01 4.55704898e-01 2.62306422e-01
1.00414109e+00 1.93703353e-01 -1.23169219e+00 -5.25674999e-01
-1.59616724e-01 -6.24717236e-01 2.50218175e-02 -9.88366663e-01
-1.10556054e+00 7.63820350e-01 1.09219229e+00 -9.06082168e-02
1.20879757e+00 -2.41769060e-01 1.16338480e+00 2.84022152e-01
8.59301329e-01 -8.57244074e-01 -1.78059936e-01 4.63572502e-01
3.54470581e-01 -1.20292318e+00 -2.77360141e-01 -4.89380419e-01
-2.07532361e-01 5.47577620e-01 6.91546679e-01 2.74932027e-01
8.64670813e-01 2.67501980e-01 1.64562151e-01 1.99460968e-01
6.47502542e-02 -3.56075376e-01 3.63529623e-01 7.79821455e-01
3.43502253e-01 3.00002992e-01 -2.75815904e-01 1.33713603e+00
2.54875515e-02 4.22465384e-01 -3.91314447e-01 1.03684580e+00
-2.78199881e-01 -9.86774743e-01 -5.66730797e-01 6.06541216e-01
-2.22700238e-01 1.02478325e-01 -4.28950600e-02 3.30873787e-01
5.00624597e-01 1.36695242e+00 -4.25463468e-01 -1.18046343e+00
3.48174870e-01 2.86822282e-02 2.59727001e-01 -2.88674325e-01
-4.87153739e-01 2.90472418e-01 -1.23425141e-01 -9.70116854e-01
-5.21949828e-01 -6.56414032e-01 -1.25313544e+00 -1.27739355e-01
-4.55661893e-01 1.34567350e-01 5.85062861e-01 1.36713326e+00
5.36257744e-01 8.67327571e-01 6.24572277e-01 -7.97262013e-01
-4.87249911e-01 -1.14007723e+00 -6.49850488e-01 3.33393514e-01
2.06696734e-01 -9.69593883e-01 -1.45953685e-01 -3.69525284e-01] | [6.694449424743652, 0.7050118446350098] |
5c14d6f6-4cf4-4517-9954-c0175238fb5a | why-does-chatgpt-fall-short-in-answering | 2304.10513 | null | https://arxiv.org/abs/2304.10513v2 | https://arxiv.org/pdf/2304.10513v2.pdf | Why Does ChatGPT Fall Short in Providing Truthful Answers? | Recent advancements in Large Language Models, such as ChatGPT, have demonstrated significant potential to impact various aspects of human life. However, ChatGPT still faces challenges in aspects like truthfulness, e.g. providing accurate and reliable outputs. Therefore, in this paper, we seek to understand why ChatGPT falls short in providing truthful answers. For this purpose, we first analyze the failures of ChatGPT in complex open-domain question answering and identifies the abilities under the failures. Specifically, we categorize ChatGPT's failures into four types: comprehension, factualness, specificity, and inference. We further pinpoint three critical abilities associated with QA failures: knowledge memorization, knowledge recall, and knowledge reasoning. Additionally, we conduct experiments centered on these abilities and propose potential approaches to enhance truthfulness. The results indicate that furnishing the model with fine-grained external knowledge, hints for knowledge recall, and guidance for reasoning can empower the model to answer questions more truthfully. | ['Kevin Chen-Chuan Chang', 'Jie Huang', 'Shen Zheng'] | 2023-04-20 | null | null | null | null | ['memorization', 'specificity', 'open-domain-question-answering'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [-3.03507090e-01 5.29608846e-01 9.57085714e-02 -3.20654571e-01
-9.08102512e-01 -6.90526128e-01 2.58044839e-01 3.73432398e-01
9.63983033e-03 8.07703555e-01 3.57214272e-01 -5.66402256e-01
-4.05763507e-01 -9.21804011e-01 -4.79394019e-01 5.97591326e-02
5.10071218e-01 4.42427993e-01 2.53233880e-01 -5.88818371e-01
7.12165952e-01 -4.04285677e-02 -1.23330069e+00 5.35632908e-01
1.31977403e+00 5.27516007e-01 -8.94550607e-02 5.42034745e-01
-4.68156248e-01 1.76627433e+00 -7.19202101e-01 -7.92265356e-01
-4.25271571e-01 -4.71664876e-01 -1.75545537e+00 -4.80871439e-01
2.76967466e-01 -6.12640381e-01 7.84871951e-02 7.29396045e-01
1.32873967e-01 3.12115818e-01 4.03041929e-01 -1.12951362e+00
-1.11762798e+00 7.68981814e-01 1.24945834e-01 3.05089176e-01
9.98510301e-01 5.02337575e-01 9.51446235e-01 -6.21590972e-01
4.35984850e-01 1.21982729e+00 8.79781127e-01 6.80293143e-01
-8.92451763e-01 -4.14606541e-01 1.12603277e-01 7.39605069e-01
-1.10147834e+00 -6.62893236e-01 4.74159569e-01 -3.55789125e-01
1.29744947e+00 2.91084975e-01 5.00432074e-01 8.39779735e-01
2.44591609e-01 6.31122410e-01 1.37126970e+00 -4.52195078e-01
2.89995790e-01 2.40918308e-01 6.24290943e-01 7.02664256e-01
3.03418249e-01 -2.08929688e-01 -1.11051500e+00 -1.03743896e-01
6.87622666e-01 -5.15774786e-01 -2.82299072e-01 1.16238885e-01
-9.69513059e-01 6.20436013e-01 3.13255191e-01 2.83410251e-01
-4.38499302e-01 2.22960755e-01 1.05691157e-01 5.58056951e-01
2.22753018e-01 1.06040394e+00 -4.45558488e-01 -6.26540422e-01
-5.63847244e-01 3.44057471e-01 1.21792746e+00 6.75824404e-01
7.71175563e-01 -1.11983158e-01 -4.70863670e-01 6.59102559e-01
3.58316571e-01 5.29192746e-01 2.95904100e-01 -1.48088646e+00
5.44551671e-01 8.32485378e-01 3.76046032e-01 -1.25740516e+00
-3.43153805e-01 -2.90597320e-01 -1.24383248e-01 -4.37367558e-01
6.27201200e-01 -1.27213597e-01 -2.50128210e-01 1.89641857e+00
2.14355573e-01 -1.97213143e-01 2.63430387e-01 8.42146099e-01
9.41494524e-01 4.65671778e-01 3.82159948e-01 3.04880105e-02
1.31879199e+00 -8.59059691e-01 -9.69257176e-01 -5.28086603e-01
8.61641049e-01 -5.79942644e-01 1.48247170e+00 2.81629324e-01
-1.46375465e+00 -2.94652760e-01 -6.04873896e-01 -5.30844808e-01
-2.81027853e-01 -3.55948150e-01 6.53534055e-01 6.63909912e-01
-1.18131733e+00 3.72703880e-01 -5.55760562e-01 -3.06474715e-01
2.96834469e-01 -1.16902895e-01 -5.30974977e-02 -4.05607134e-01
-1.69046128e+00 1.44659376e+00 9.85445976e-02 7.63830990e-02
-7.49731004e-01 -5.51731288e-01 -7.88560390e-01 3.99082601e-01
6.12820387e-01 -1.05041790e+00 1.84382021e+00 -6.56847179e-01
-1.60045552e+00 6.94312990e-01 -4.03629363e-01 -3.31259668e-01
1.41201422e-01 -4.53607917e-01 -6.07725382e-02 3.41311783e-01
3.08538705e-01 3.92262459e-01 2.99576283e-01 -9.03067112e-01
-2.21928269e-01 -3.67459476e-01 6.67242110e-01 2.37612009e-01
-3.08751553e-01 -7.14072660e-02 6.37259632e-02 -1.45995155e-01
2.78646976e-01 -3.78990978e-01 1.76998198e-01 -2.62518376e-01
-2.35182986e-01 -5.32716453e-01 2.59755552e-01 -8.98148239e-01
1.21024597e+00 -1.65052843e+00 -2.25228414e-01 6.61908984e-02
5.20298541e-01 1.97954983e-01 -2.61505097e-02 8.56707156e-01
4.28153247e-01 4.00771081e-01 2.44203776e-01 -2.67310180e-02
2.68503100e-01 1.98037982e-01 -4.34427291e-01 -2.99783468e-01
3.40676457e-01 1.48426569e+00 -1.04272699e+00 -5.86172402e-01
-1.51544675e-01 4.98427711e-02 -5.90783298e-01 2.95332938e-01
-6.77577317e-01 2.30619922e-01 -5.94149053e-01 5.81482887e-01
3.94925684e-01 -7.24244356e-01 2.56381869e-01 2.72876114e-01
3.67702693e-02 8.73474956e-01 -6.73875332e-01 1.36907840e+00
-4.70141232e-01 4.46760237e-01 -8.87746140e-02 -4.57825959e-01
7.70146132e-01 2.59894013e-01 -4.43714261e-01 -8.57462525e-01
5.10183442e-03 1.20433420e-01 4.80413530e-03 -9.72647607e-01
6.88467503e-01 -4.04661536e-01 5.28331399e-02 9.07428265e-01
-7.45420009e-02 -3.52471620e-01 -3.76041457e-02 7.69090474e-01
9.47471201e-01 8.52683261e-02 2.62436837e-01 -1.95394337e-01
3.12117964e-01 3.72797012e-01 1.24418102e-01 9.63566601e-01
-4.22517806e-01 -4.21158858e-02 7.18664408e-01 -2.96475850e-02
-3.53453696e-01 -8.88224304e-01 4.62218791e-01 1.21279752e+00
1.16658852e-01 -5.96590102e-01 -9.21375632e-01 -3.55259746e-01
-3.43217790e-01 1.30682969e+00 -4.22134578e-01 -5.72655380e-01
-2.68060178e-01 -9.47227478e-02 9.47386563e-01 7.22378850e-01
9.76429045e-01 -1.05035007e+00 -8.64828348e-01 7.62234032e-02
-1.06084085e+00 -9.09765720e-01 -1.33132011e-01 -2.34381482e-01
-1.01603854e+00 -1.23584902e+00 -1.17342703e-01 -5.07395029e-01
2.14660704e-01 5.60690165e-01 1.46723139e+00 5.71215928e-01
3.99543911e-01 9.64394033e-01 -5.02206326e-01 -2.00840876e-01
-3.81614238e-01 -3.64233851e-02 -3.87789726e-01 -6.11791968e-01
5.76437414e-01 -3.61147016e-01 -3.83481055e-01 4.01680946e-01
-6.07411146e-01 1.81732997e-01 4.14103895e-01 6.61915779e-01
7.06581101e-02 -1.60308659e-01 1.04647160e+00 -8.66866946e-01
1.23523211e+00 -7.05874085e-01 -6.42838329e-02 6.28788173e-01
-6.47836804e-01 3.99169736e-02 2.84201205e-01 -3.69575769e-02
-1.57467103e+00 -7.76902556e-01 -2.52294809e-01 2.34421209e-01
-6.70316964e-02 9.59530950e-01 6.98640570e-02 -1.65268838e-01
1.01645386e+00 3.39570105e-01 9.71002355e-02 -1.36483550e-01
4.97197360e-01 4.29197878e-01 4.12076563e-01 -1.09633017e+00
4.56533819e-01 -2.14360803e-02 -5.84426343e-01 -5.91736436e-01
-1.38911569e+00 -2.71212667e-01 6.58051372e-02 -3.59910280e-01
5.40595233e-01 -7.37386048e-01 -1.28659999e+00 3.53640676e-01
-1.17890596e+00 -6.33132875e-01 -2.13455096e-01 4.11392115e-02
-7.23870993e-01 4.77575839e-01 -8.65449309e-01 -9.10766900e-01
-4.49139029e-01 -6.87230229e-01 4.86101031e-01 5.08186519e-01
-7.66437829e-01 -1.17150474e+00 1.13689303e-02 1.20530796e+00
9.13039148e-01 -3.40389669e-01 1.28672862e+00 -7.07845688e-01
-8.07329178e-01 8.24375972e-02 -1.87464938e-01 2.85140365e-01
-2.74263442e-01 -4.21215594e-01 -9.92470443e-01 3.33934963e-01
4.44525450e-01 -9.49167192e-01 4.07306850e-01 1.71568077e-02
9.04601455e-01 -5.29988706e-01 3.02961394e-02 3.84941362e-02
1.04463840e+00 -8.99181440e-02 8.27536762e-01 3.21109414e-01
4.88736369e-02 7.84985960e-01 7.46667624e-01 1.09965034e-01
1.13005412e+00 2.11203113e-01 1.43863887e-01 6.19478166e-01
-8.16008970e-02 -6.64776981e-01 1.97613597e-01 6.59224927e-01
3.26297954e-02 -8.71295184e-02 -1.27030897e+00 7.18036473e-01
-1.89182663e+00 -9.08918381e-01 -2.15561464e-01 1.74298537e+00
1.22942960e+00 -9.09442827e-03 -3.26167643e-01 3.98832932e-02
2.89134413e-01 -1.18989117e-01 -5.29762685e-01 -5.48101366e-01
-2.97306813e-02 2.10641101e-01 -4.13519204e-01 7.22267389e-01
-1.33966208e-01 1.28277385e+00 7.08558416e+00 4.27809626e-01
-6.58143818e-01 2.31915995e-01 5.31010032e-01 2.57268876e-01
-6.09529316e-01 1.61128461e-01 -3.71826649e-01 -4.21143770e-02
9.92672741e-01 -3.10001761e-01 5.42910278e-01 5.65553069e-01
1.36000440e-01 -5.65256834e-01 -9.60985422e-01 4.41184342e-01
6.02980182e-02 -1.36914206e+00 1.22823827e-01 -6.35358870e-01
5.64444959e-01 -4.83705401e-01 -1.20299682e-01 6.62111282e-01
3.05373698e-01 -1.19607604e+00 6.45168841e-01 8.77748787e-01
1.71728551e-01 -5.31761885e-01 8.71148109e-01 8.38917434e-01
-4.35143262e-01 6.84165210e-02 -8.42024907e-02 -8.63476098e-01
1.42718583e-01 2.03095779e-01 -8.34238946e-01 4.53808337e-01
5.34454346e-01 2.79626131e-01 -6.69176519e-01 7.05117583e-01
-7.62739241e-01 8.18948448e-01 3.71226147e-02 -3.38162690e-01
-1.40351832e-01 1.60925701e-01 1.45173222e-02 7.08596587e-01
-6.49968982e-02 6.57629669e-01 -1.65350374e-03 1.23993683e+00
1.41893122e-02 -7.88646713e-02 -2.14286804e-01 -3.24715823e-01
8.36531937e-01 7.05393553e-01 -3.15979153e-01 -5.15896022e-01
-1.47455826e-01 6.78610623e-01 7.42138743e-01 4.18917030e-01
-5.96236825e-01 -2.15488672e-01 4.69697684e-01 7.25340694e-02
-3.91321272e-01 -2.03121275e-01 -8.56355011e-01 -1.20783889e+00
1.03289314e-01 -1.12391925e+00 3.10368717e-01 -1.44105697e+00
-1.15805328e+00 1.30267948e-01 -5.36175035e-02 -9.91373658e-02
-3.45216781e-01 -3.74978870e-01 -6.21976018e-01 1.06233871e+00
-1.48811448e+00 -1.03412104e+00 -6.05662704e-01 6.42170131e-01
3.76753569e-01 4.96105582e-01 1.02760458e+00 -2.39165232e-01
-3.00366461e-01 4.67811078e-01 -4.99287665e-01 -2.01332686e-03
7.54994631e-01 -1.03907609e+00 3.16902101e-01 5.49838901e-01
-1.78307950e-01 1.37657952e+00 7.47295797e-01 -7.95002580e-01
-1.58770776e+00 -6.12010300e-01 1.47723889e+00 -9.82697725e-01
6.83222353e-01 1.71658516e-01 -1.20225680e+00 9.75699306e-01
2.49894857e-01 -7.99064815e-01 8.13748538e-01 5.02443612e-01
-6.37004852e-01 2.89007246e-01 -1.25779760e+00 5.89408636e-01
9.02454674e-01 -8.93074274e-01 -1.41896248e+00 4.18067604e-01
7.92388260e-01 -4.92011696e-01 -6.87309206e-01 -2.25264817e-01
3.56063128e-01 -1.14288652e+00 7.92423129e-01 -7.96205819e-01
7.49904335e-01 -9.13279280e-02 -1.53157767e-02 -1.22061670e+00
-3.39723617e-01 -4.57264513e-01 -2.17821762e-01 1.11171806e+00
3.78184408e-01 -9.21838880e-01 5.97200871e-01 1.19995034e+00
-1.40371189e-01 -5.37143171e-01 -6.73983395e-01 -5.29401004e-01
4.06955093e-01 -6.67755723e-01 6.29774511e-01 9.87124562e-01
7.06180036e-01 6.47605658e-01 -1.78657278e-01 1.41355082e-01
2.56525785e-01 2.01019451e-01 6.68744922e-01 -1.14465129e+00
-1.71895951e-01 -1.40708864e-01 2.98282981e-01 -1.24216199e+00
1.30885452e-01 -6.00552082e-01 3.39931473e-02 -2.02710390e+00
3.04724157e-01 -2.78145432e-01 2.26809457e-01 8.33776712e-01
-3.97360384e-01 -1.73381031e-01 2.10950941e-01 1.45088494e-01
-1.06374323e+00 3.25882763e-01 1.31680822e+00 1.85385287e-01
5.52658997e-02 -2.40322471e-01 -1.40892041e+00 6.38280332e-01
1.06913137e+00 -1.60003364e-01 -6.15469038e-01 -7.08519638e-01
7.45829642e-01 4.89548087e-01 6.96567118e-01 -8.10986757e-01
5.99951565e-01 -4.87834871e-01 -5.61257228e-02 -2.26749405e-01
3.85562122e-01 -3.54402483e-01 -1.82245433e-01 3.31770480e-01
-5.66610932e-01 4.40485552e-02 4.11911130e-01 2.93855339e-01
-2.34328896e-01 -3.35360199e-01 3.61807168e-01 -3.66103917e-01
-7.05903232e-01 -6.43691182e-01 -7.15138376e-01 5.86392522e-01
5.95378280e-01 -1.93419963e-01 -9.21609402e-01 -8.91445279e-01
-6.32753372e-01 6.30812168e-01 3.28588039e-01 3.12104881e-01
8.39507461e-01 -8.85027766e-01 -4.28372800e-01 -1.65333480e-01
2.10233137e-01 -1.81783542e-01 6.79495692e-01 9.08328295e-01
-3.66803348e-01 8.78539920e-01 -1.88734367e-01 -1.65921688e-01
-9.14362848e-01 2.75703490e-01 5.04971147e-01 -1.66706234e-01
-1.57723010e-01 9.36334372e-01 -1.02370180e-01 -5.06176591e-01
6.97886199e-02 -3.72313857e-01 -1.92203850e-01 -2.02496707e-01
6.82905853e-01 4.80969101e-01 9.01174620e-02 -3.96751724e-02
-3.53189349e-01 2.52482891e-01 2.91063148e-03 -7.81812817e-02
8.14964652e-01 -4.21635777e-01 -4.25562054e-01 4.61399138e-01
3.49130332e-01 -1.14389800e-01 -7.40247130e-01 -2.66455382e-01
2.18918622e-01 -4.67375994e-01 -1.87750623e-01 -1.64910185e+00
-3.84625614e-01 9.33048010e-01 -2.20301375e-01 2.77796477e-01
8.60734820e-01 1.49196625e-01 6.67918086e-01 9.61596251e-01
7.01888382e-01 -9.77195919e-01 4.79264498e-01 9.37743604e-01
8.68779182e-01 -1.06818032e+00 -3.41604233e-01 -4.15749222e-01
-6.74238622e-01 9.33808088e-01 1.14039302e+00 4.07382786e-01
3.00524756e-02 -2.55159497e-01 3.39899719e-01 -5.25499940e-01
-1.26620066e+00 -9.54788364e-03 -8.29322413e-02 5.73112428e-01
6.26056433e-01 -6.63819760e-02 -3.23167741e-01 9.09619391e-01
-5.99248588e-01 4.60799664e-01 6.83895648e-01 1.02141678e+00
-8.49161267e-01 -5.30819476e-01 -6.49371803e-01 3.46018314e-01
-2.35300705e-01 -3.65099460e-01 -8.61308932e-01 4.97680068e-01
-3.36380601e-01 1.61531270e+00 -4.34897035e-01 -2.33594090e-01
4.75398064e-01 5.35411239e-01 4.94806051e-01 -6.42901182e-01
-9.60315287e-01 -8.45710695e-01 5.06109178e-01 -6.24119282e-01
-1.62394956e-01 -2.83628225e-01 -1.23997009e+00 -8.30778241e-01
-3.67689371e-01 5.09763896e-01 1.66813061e-01 1.32878387e+00
5.19722700e-01 3.17694575e-01 -2.65526265e-01 1.49394363e-01
-9.47519243e-01 -1.02514398e+00 2.56995019e-02 3.74337174e-02
7.42236450e-02 -4.50659931e-01 -3.32739860e-01 -1.85522884e-01] | [10.797511100769043, 7.893683910369873] |
25d5bc66-a528-4a85-8770-c14a2996bfba | ubernet-training-a-universal-convolutional-1 | null | null | http://openaccess.thecvf.com/content_cvpr_2017/html/Kokkinos_Ubernet_Training_a_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Kokkinos_Ubernet_Training_a_CVPR_2017_paper.pdf | Ubernet: Training a Universal Convolutional Neural Network for Low-, Mid-, and High-Level Vision Using Diverse Datasets and Limited Memory | In this work we train in an end-to-end manner a convolutional neural network (CNN) that jointly handles low-, mid-, and high-level vision tasks in a unified architecture. Such a network can act like a `swiss knife' for vision tasks; we call it an "UberNet" to indicate its overarching nature. The main contribution of this work consists in handling challenges that emerge when scaling up to many tasks. We introduce techniques that facilitate (i) training a deep architecture while relying on diverse training sets and (ii) training many (potentially unlimited) tasks with a limited memory budget. This allows us to train in an end-to-end manner a unified CNN architecture that jointly handles (a) boundary detection (b) normal estimation (c) saliency estimation (d) semantic segmentation (e) human part segmentation (f) semantic boundary detection, (g) region proposal generation and object detection. We obtain competitive performance while jointly addressing all tasks in 0.7 seconds on a GPU. Our system will be made publicly available.
| ['Iasonas Kokkinos'] | 2017-07-01 | null | null | null | cvpr-2017-7 | ['human-part-segmentation'] | ['computer-vision'] | [ 2.86086172e-01 2.11003140e-01 1.53338566e-01 -2.56714404e-01
-7.40341306e-01 -4.44363594e-01 5.74776769e-01 1.83303967e-01
-6.65830433e-01 4.22060430e-01 -2.33315557e-01 -4.30464059e-01
5.28363347e-01 -4.94531959e-01 -1.09788990e+00 -4.07998592e-01
6.49131387e-02 5.13706028e-01 9.11660552e-01 -2.67446607e-01
9.08569917e-02 5.18542647e-01 -1.79741967e+00 1.87622786e-01
8.00160050e-01 1.22483122e+00 6.53201342e-01 9.60117042e-01
1.07197858e-01 5.14566660e-01 -4.25824851e-01 -3.21582824e-01
4.47693139e-01 -1.03889152e-01 -1.02806830e+00 2.79357493e-01
7.16775298e-01 -4.14504379e-01 2.69301027e-01 9.33655858e-01
4.90079612e-01 3.86937857e-02 4.10212606e-01 -1.18266344e+00
-2.32286617e-01 3.57268929e-01 -6.29573584e-01 2.37572789e-01
3.16586755e-02 2.99864650e-01 7.89429128e-01 -8.35217416e-01
7.94881165e-01 1.06756079e+00 6.63463771e-01 5.54823458e-01
-1.11755598e+00 -2.08200514e-01 3.16272676e-01 -1.38152003e-01
-1.10131299e+00 -4.28558290e-01 5.13589501e-01 -3.69674474e-01
9.93781090e-01 4.03189734e-02 6.44103110e-01 8.53610039e-01
-2.93632410e-02 1.15337145e+00 6.79814458e-01 -2.91680723e-01
4.82731849e-01 -1.87414452e-01 2.32754797e-01 7.97223985e-01
2.73870617e-01 -2.07965165e-01 -2.57550567e-01 1.04498595e-01
8.68313491e-01 -1.43032700e-01 7.54780546e-02 -3.72661918e-01
-1.41331470e+00 7.27453530e-01 8.86341333e-01 2.92991698e-01
-3.94831032e-01 4.85619664e-01 6.25812054e-01 8.26906487e-02
3.34887296e-01 3.27438623e-01 -6.02146506e-01 2.07490757e-01
-1.02832365e+00 4.13957119e-01 7.84124196e-01 9.77977097e-01
8.23223948e-01 1.71385594e-02 -1.62031338e-01 8.15043151e-01
1.47379845e-01 1.67907894e-01 3.39817286e-01 -8.99606705e-01
2.40117654e-01 3.73748362e-01 3.25834304e-01 -4.60367620e-01
-7.46500134e-01 -5.20673096e-01 -8.15883279e-01 5.34496605e-01
5.38525760e-01 -4.06704873e-01 -1.40636313e+00 1.72660339e+00
5.49870253e-01 2.33252168e-01 -1.68822750e-01 1.11226761e+00
9.43124354e-01 3.70611250e-01 3.05185378e-01 4.97554660e-01
1.75655425e+00 -1.44608855e+00 -1.35113090e-01 -5.90416014e-01
3.96183640e-01 -9.19519901e-01 9.06778574e-01 2.04393223e-01
-1.34907687e+00 -7.78813362e-01 -9.83469129e-01 -5.53201437e-01
-4.95005131e-01 2.78015614e-01 6.12072051e-01 3.78195345e-01
-1.64255691e+00 4.83154655e-01 -1.06571162e+00 -4.35703486e-01
6.81613743e-01 4.85439271e-01 -2.79304445e-01 1.83263242e-01
-6.89994752e-01 7.59990752e-01 5.35930634e-01 1.67054057e-01
-9.02487099e-01 -5.59803545e-01 -9.87715781e-01 5.13998866e-02
5.65625727e-01 -1.20007062e+00 1.47625625e+00 -1.23191416e+00
-1.22012424e+00 1.16766930e+00 -9.48307291e-02 -6.57815754e-01
6.61850035e-01 -2.95831531e-01 1.77899987e-01 2.42238477e-01
3.94483238e-01 1.32094395e+00 8.73421371e-01 -1.26870096e+00
-1.00526917e+00 -3.01403493e-01 2.93910038e-02 1.51827604e-01
3.28518569e-01 3.30151141e-01 -7.41731763e-01 -7.67391443e-01
1.65404111e-01 -9.09158945e-01 -3.69136482e-01 3.51438999e-01
-6.06974483e-01 -2.42609814e-01 9.17204261e-01 -5.35772145e-01
5.76502621e-01 -1.98821163e+00 1.80434301e-01 -2.90730357e-01
4.11616951e-01 6.23083353e-01 -1.16284221e-01 8.06367956e-03
8.72802660e-02 -4.35202345e-02 -4.99514878e-01 -8.14013541e-01
-1.56485096e-01 -1.06690854e-01 1.02943137e-01 3.85817856e-01
4.03798789e-01 1.15090466e+00 -7.84970403e-01 -3.55044067e-01
3.63641649e-01 3.65729123e-01 -4.36108083e-01 2.76282609e-01
-4.94914562e-01 4.00312096e-01 -2.91675329e-01 8.33103478e-01
6.54843688e-01 -5.19157290e-01 -2.36176804e-01 -8.54384378e-02
-2.86702484e-01 7.99602568e-02 -1.02707851e+00 1.95106292e+00
-3.88078034e-01 6.83302283e-01 7.45169461e-01 -8.61992359e-01
6.68426812e-01 8.16359222e-02 2.01291040e-01 -4.36032087e-01
3.64940226e-01 3.97314727e-01 -3.95228080e-02 -1.92362547e-01
5.70028484e-01 1.23281687e-01 -7.23089576e-02 4.01417136e-01
3.06366891e-01 5.95584884e-02 1.74947053e-01 5.29841594e-02
9.18318748e-01 2.36165926e-01 2.77349293e-01 -5.37578762e-01
4.59216207e-01 1.75421789e-01 3.41920286e-01 8.23845804e-01
-4.89072680e-01 8.38021696e-01 4.69145060e-01 -7.55511642e-01
-1.28067267e+00 -1.03460717e+00 2.58068517e-02 1.42971659e+00
2.95574367e-01 1.04844451e-01 -9.65358496e-01 -4.35732663e-01
-5.40140551e-04 2.69288927e-01 -7.75327563e-01 1.77908733e-01
-7.42315054e-01 -4.54928756e-01 4.90154028e-01 8.77001703e-01
8.51192772e-01 -1.38847613e+00 -1.21873236e+00 2.17671350e-01
5.82650490e-02 -1.19155967e+00 -4.23132360e-01 4.35169816e-01
-7.49981105e-01 -9.15205061e-01 -1.02355063e+00 -1.14972723e+00
5.32466292e-01 4.90609139e-01 1.27903616e+00 8.84043500e-02
-4.65973049e-01 7.56418332e-02 -2.00826347e-01 -5.04258335e-01
-1.69714585e-01 2.66507000e-01 -3.19172055e-01 -1.03910387e-01
-1.21764280e-02 -3.20799798e-01 -9.53904867e-01 2.35833690e-01
-9.81808424e-01 3.23445946e-01 7.19868541e-01 6.43329918e-01
6.32907391e-01 -6.28833592e-01 4.39820170e-01 -7.15623796e-01
2.93907434e-01 -4.19909477e-01 -7.58943260e-01 1.69063047e-01
-9.59322602e-02 -1.36651263e-01 4.28142011e-01 -5.11129536e-02
-8.98105979e-01 3.03488225e-01 -5.22618413e-01 -3.29988807e-01
-5.18062472e-01 9.49619710e-02 5.24637923e-02 -1.55619711e-01
5.23637176e-01 6.99595436e-02 1.42363906e-01 -5.31651020e-01
5.52246034e-01 5.35073400e-01 8.98948789e-01 -4.49020177e-01
5.30298293e-01 6.56485438e-01 4.62916121e-02 -8.05025101e-01
-9.86442268e-01 -7.12185919e-01 -8.85330617e-01 -6.62652105e-02
1.39695680e+00 -1.22325921e+00 -7.00708389e-01 8.30211937e-01
-1.16128218e+00 -7.44536936e-01 -9.59742293e-02 -9.54676978e-03
-6.04407430e-01 2.10819036e-01 -5.99240661e-01 -4.61082816e-01
-6.90582275e-01 -1.22050345e+00 1.64485013e+00 5.59725881e-01
1.68415345e-02 -9.32650089e-01 -2.54650205e-01 5.16227484e-01
6.09004855e-01 6.54284954e-01 3.10194731e-01 -7.16976345e-01
-7.65924692e-01 9.89966094e-02 -7.98780382e-01 2.42739543e-01
-2.89017141e-01 -1.57577977e-01 -1.09859979e+00 -3.32923204e-01
-2.22799703e-01 -5.63847959e-01 1.16034281e+00 5.26675224e-01
1.14136577e+00 1.15487337e-01 -3.02674323e-01 8.68073642e-01
1.50649607e+00 -2.87979305e-01 4.30015087e-01 5.63534439e-01
8.47998083e-01 5.63056886e-01 4.19436187e-01 1.59620687e-01
4.86961424e-01 7.05293000e-01 6.80181503e-01 -6.18427634e-01
-3.93893152e-01 1.48612753e-01 1.50884374e-03 -2.76774745e-02
-2.01889854e-02 -6.59174025e-02 -1.01872063e+00 9.14330900e-01
-2.06047320e+00 -7.04547405e-01 -3.79362285e-01 1.84673285e+00
6.60732448e-01 3.16809952e-01 4.83754277e-01 -3.18351865e-01
8.96806300e-01 1.01716578e-01 -6.05188906e-01 -5.08457005e-01
9.67611074e-02 3.36583823e-01 4.55734611e-01 2.46476710e-01
-1.49505663e+00 1.21997857e+00 5.58155298e+00 6.71443045e-01
-1.25053620e+00 1.54993221e-01 7.79958487e-01 2.01142058e-01
1.62176847e-01 -2.29670241e-01 -7.39263415e-01 3.47588271e-01
6.64854825e-01 3.49948615e-01 1.06581591e-01 1.06505823e+00
-1.16061509e-01 -3.36947113e-01 -8.92467320e-01 9.22742665e-01
-2.09726896e-02 -1.43874669e+00 -1.70067608e-01 -1.82157248e-01
7.15606809e-01 6.17272496e-01 -1.71364948e-01 1.37393147e-01
3.23162645e-01 -9.79026914e-01 1.10451901e+00 2.28703856e-01
7.23891675e-01 -5.31339109e-01 6.99989319e-01 3.28510314e-01
-1.23338246e+00 -7.35144224e-03 -3.17631721e-01 1.67944282e-01
2.08924502e-01 5.73781133e-01 -7.49467909e-01 5.12304127e-01
8.07896197e-01 3.12428266e-01 -6.83034539e-01 1.30997598e+00
-1.75508469e-01 2.25524187e-01 -5.21186948e-01 1.35818720e-01
7.00829446e-01 7.85623789e-02 4.00867224e-01 1.45391154e+00
2.66866907e-02 -2.10021257e-01 3.58938873e-01 1.04622066e+00
-1.77201018e-01 -2.76434064e-01 -2.74351954e-01 3.62846613e-01
1.89658344e-01 1.42025661e+00 -1.49653924e+00 -4.68180478e-01
-3.63477647e-01 1.14761889e+00 3.91156405e-01 1.41758710e-01
-7.77416945e-01 -4.71192062e-01 7.29374647e-01 -2.25466453e-02
6.62367046e-01 -2.18568534e-01 -5.42646766e-01 -1.01097906e+00
4.79726307e-02 -3.72112721e-01 1.75965518e-01 -9.34765339e-01
-1.01103222e+00 7.02950001e-01 -2.39921674e-01 -7.50272572e-01
-1.53268725e-01 -6.76576674e-01 -8.00710499e-01 8.46661925e-01
-1.69998264e+00 -1.33411634e+00 -6.15273178e-01 4.51278210e-01
7.17550457e-01 7.64149576e-02 5.39296031e-01 5.72619475e-02
-6.21608377e-01 3.23872894e-01 -2.15395257e-01 3.03033590e-01
4.10492569e-01 -1.58907688e+00 9.14800048e-01 9.52817738e-01
-1.99659184e-01 4.13889110e-01 6.57703638e-01 -4.63826507e-01
-9.32929516e-01 -1.39628541e+00 8.66315067e-01 -2.72293866e-01
4.71809804e-01 -4.90482360e-01 -8.22656035e-01 6.56156600e-01
2.95291930e-01 4.03192282e-01 2.45090187e-01 -1.68134365e-02
-1.89114690e-01 2.02740625e-01 -1.11476517e+00 6.42707586e-01
9.90036964e-01 -2.34713465e-01 -5.84349036e-01 3.33144546e-01
8.47759128e-01 -7.74796188e-01 -3.77986073e-01 2.28610978e-01
3.87363672e-01 -1.28178132e+00 1.03872907e+00 -4.25716579e-01
6.07340515e-01 -4.25208986e-01 2.45260149e-01 -8.74507606e-01
-1.93660632e-01 -7.91508675e-01 -7.92276487e-02 7.10137367e-01
2.97089607e-01 -4.26216602e-01 5.98124325e-01 5.02219081e-01
-5.35351217e-01 -1.01191473e+00 -9.04659986e-01 -5.98570168e-01
-6.56683892e-02 -2.92173862e-01 4.44044143e-01 4.29533303e-01
-6.41541600e-01 2.59587377e-01 -4.50381450e-02 8.96982402e-02
5.05292952e-01 3.60074431e-01 6.62786067e-01 -1.33790076e+00
-1.00217223e-01 -7.55802870e-01 -4.29729998e-01 -1.21932471e+00
-2.44974047e-02 -7.82153428e-01 2.35380679e-01 -1.66296566e+00
2.29504332e-01 -2.93194443e-01 -1.87127843e-01 5.70010900e-01
-2.44861722e-01 5.69714427e-01 2.77147323e-01 1.27088279e-01
-9.20325816e-01 2.15952858e-01 1.10208082e+00 1.78350419e-01
-1.81180071e-02 1.55469105e-01 -7.91696489e-01 7.22664952e-01
7.62304127e-01 -1.46646515e-01 -4.40551415e-02 -6.99011385e-01
2.42710244e-02 -1.13199756e-01 9.11305428e-01 -1.16129732e+00
2.63786197e-01 2.54054457e-01 3.86252433e-01 -5.30067503e-01
2.28472650e-01 -4.81292844e-01 -3.39170486e-01 5.10321498e-01
-1.63750187e-01 -7.53904954e-02 5.10858715e-01 4.60515708e-01
-1.71392754e-01 -2.10867509e-01 1.20113099e+00 -3.07074517e-01
-9.94741619e-01 2.48352826e-01 -1.91558748e-01 1.08089438e-02
1.18156278e+00 -2.85772353e-01 -3.14309984e-01 6.63807848e-03
-8.43223929e-01 3.63886356e-01 8.20980608e-01 2.56982505e-01
3.10395211e-01 -9.18724716e-01 -6.78634167e-01 2.26227149e-01
4.47635725e-02 5.18731773e-01 2.30469793e-01 6.87210083e-01
-9.65048969e-01 4.36980158e-01 -4.75144804e-01 -7.77167618e-01
-1.18099737e+00 4.14534390e-01 3.52825612e-01 -1.78731352e-01
-8.51896107e-01 1.31335843e+00 1.28175676e-01 -1.31522179e-01
2.99420536e-01 -3.17515880e-01 -7.20001087e-02 -3.26710567e-03
4.91578400e-01 1.51202559e-01 1.67175621e-01 -5.89661598e-01
-4.76286918e-01 3.72253239e-01 -1.22645698e-01 9.62080210e-02
1.35316288e+00 -5.13255373e-02 -1.29657403e-01 1.02783151e-01
1.05904245e+00 -4.98353601e-01 -1.64499414e+00 -4.71082181e-02
-1.03165023e-01 -1.27663285e-01 2.28128374e-01 -8.33766997e-01
-1.08225513e+00 8.71162653e-01 4.49948788e-01 3.11325818e-01
1.11550891e+00 1.05170958e-01 1.10979044e+00 2.52127200e-01
3.15180451e-01 -1.13668895e+00 -3.85860540e-02 5.98836839e-01
7.07206845e-01 -1.48625255e+00 -1.65723532e-01 -3.71373802e-01
-6.77664876e-01 1.18532491e+00 5.28764725e-01 -4.18315411e-01
5.74863195e-01 3.44408989e-01 7.58919418e-02 -3.10049504e-01
-6.88637614e-01 -8.46535861e-01 3.38461816e-01 5.25619566e-01
3.07711601e-01 -9.29071661e-03 -7.27100670e-02 2.25263864e-01
7.16495886e-02 -5.40320650e-02 5.05904317e-01 1.03907990e+00
-7.86211073e-01 -9.52547014e-01 -2.73174822e-01 2.66674012e-01
-5.27879655e-01 7.13812560e-02 -4.49732274e-01 8.54154468e-01
4.40112084e-01 8.24314058e-01 2.69670904e-01 3.21080610e-02
1.37619555e-01 -8.71888362e-03 2.36384541e-01 -9.00630116e-01
-8.75022769e-01 9.14300755e-02 2.80419383e-02 -7.62806475e-01
-3.94664735e-01 -4.99779910e-01 -1.20392120e+00 4.54074843e-03
7.59035051e-02 -3.65858853e-01 8.67216468e-01 9.86890018e-01
5.70954323e-01 7.00763106e-01 6.70211539e-02 -1.37673295e+00
-3.92050326e-01 -9.09501493e-01 -3.30803841e-01 3.31504226e-01
5.32078803e-01 -5.80472112e-01 2.41772295e-03 1.58084139e-01] | [9.44040298461914, 0.15048018097877502] |
6abc75a4-7501-4b75-99ea-6788fd68d74a | stimulating-student-engagement-with-an-ai | 2304.11376 | null | https://arxiv.org/abs/2304.11376v1 | https://arxiv.org/pdf/2304.11376v1.pdf | Stimulating student engagement with an AI board game tournament | Strong foundations in basic AI techniques are key to understanding more advanced concepts. We believe that introducing AI techniques, such as search methods, early in higher education helps create a deeper understanding of the concepts seen later in more advanced AI and algorithms courses. We present a project-based and competition-based bachelor course that gives second-year students an introduction to search methods applied to board games. In groups of two, students have to use network programming and AI methods to build an AI agent to compete in a board game tournament-othello was this year's game. Students are evaluated based on the quality of their projects and on their performance during the final tournament. We believe that the introduction of gamification, in the form of competition-based learning, allows for a better learning experience for the students. | ['Quentin Lurkin', 'Ken Hasselmann'] | 2023-04-22 | null | null | null | null | ['board-games'] | ['playing-games'] | [-2.11462498e-01 1.37776002e-01 3.70358340e-02 -1.55582666e-01
-9.91071761e-02 -4.48035687e-01 1.68469250e-01 4.56559330e-01
-4.23058927e-01 5.17413855e-01 -4.51463073e-01 -7.75811613e-01
-4.65544134e-01 -1.27641356e+00 -3.34924400e-01 -7.79187754e-02
-6.22575358e-02 5.22961617e-01 3.80846649e-01 -1.06863368e+00
9.45942163e-01 5.48420250e-01 -1.55525982e+00 6.25544414e-02
9.83170867e-01 3.05563122e-01 1.02947421e-01 5.49380958e-01
-2.73514539e-01 1.14877558e+00 -6.07038915e-01 -6.17279828e-01
2.21098468e-01 -9.25461531e-01 -1.05346096e+00 -5.41708767e-01
-1.14477433e-01 1.04464535e-02 -8.29665437e-02 8.39158595e-01
8.73411968e-02 6.34962678e-01 1.76190615e-01 -1.30340922e+00
1.66100934e-02 7.34443188e-01 -3.39849383e-01 3.50238442e-01
6.81274712e-01 1.37118012e-01 9.79681432e-01 -2.69677609e-01
4.27809149e-01 8.95520687e-01 6.04969323e-01 6.07634008e-01
-1.08224487e+00 -8.10265601e-01 9.80914664e-03 3.39568734e-01
-1.17655718e+00 3.37046683e-01 6.80624068e-01 -4.51931179e-01
7.98043966e-01 1.14528380e-01 1.74445164e+00 -1.22460201e-01
7.37853870e-02 7.78568149e-01 8.46637428e-01 -7.25220084e-01
4.80174959e-01 4.30545866e-01 -5.06593920e-02 6.31145954e-01
6.18400037e-01 3.66132885e-01 -3.70211303e-01 1.53089494e-01
9.87141967e-01 -1.19464599e-01 9.82628465e-02 -4.61676598e-01
-1.15411770e+00 1.15003240e+00 4.69424427e-01 5.79647183e-01
-3.69709700e-01 4.20637608e-01 1.87480986e-01 6.28648043e-01
-3.80664676e-01 1.27135956e+00 -1.83207691e-01 -6.95655167e-01
-8.64573479e-01 6.22753382e-01 1.27174008e+00 4.62849349e-01
4.29705888e-01 2.94844300e-01 4.42936361e-01 3.68990362e-01
4.61602032e-01 -3.65446299e-01 2.78446585e-01 -1.26627135e+00
-1.01642482e-01 9.03129399e-01 -2.54908264e-01 -1.09310770e+00
1.11018769e-01 -3.86930853e-01 -6.22735098e-02 1.11871243e+00
5.22001266e-01 -4.80770856e-01 -7.05292284e-01 1.33127344e+00
2.35271707e-01 1.54607892e-01 8.00278038e-02 6.68686628e-01
6.40041649e-01 6.04336500e-01 1.39492109e-01 1.31145000e-01
1.38085699e+00 -9.61441398e-01 -2.14035548e-02 -2.48540565e-02
7.22145200e-01 -6.49320006e-01 7.18780696e-01 1.10546601e+00
-1.47552311e+00 -6.34671032e-01 -1.14570439e+00 4.04578596e-01
-3.70336205e-01 -6.47297800e-01 9.61795926e-01 1.39990699e+00
-8.48542392e-01 8.26352060e-01 -7.52537370e-01 -3.76480252e-01
1.46690756e-01 6.34207308e-01 1.55093506e-01 -8.99418592e-02
-1.15597272e+00 1.09651411e+00 4.46303219e-01 -2.98199505e-01
-5.39841950e-01 -9.59568977e-01 -7.00240374e-01 3.84531289e-01
4.30393368e-01 -6.45490646e-01 1.38721550e+00 -5.98477006e-01
-1.64793873e+00 9.27508354e-01 6.88998580e-01 -1.98376730e-01
4.47157711e-01 4.65698302e-01 2.26732031e-01 2.39063520e-02
-1.34970129e-01 6.61244631e-01 -3.20864677e-01 -8.48706186e-01
-9.40252006e-01 -1.36107296e-01 8.23027253e-01 6.23855770e-01
1.15198150e-01 1.23199970e-01 1.47626743e-01 -4.01619226e-01
2.49795437e-01 -5.32883823e-01 -5.81868887e-01 -2.53766119e-01
6.02385163e-01 -5.48579872e-01 1.27892539e-01 -2.91018374e-02
9.70551729e-01 -1.69359434e+00 -1.67172909e-01 6.17106676e-01
2.90706187e-01 3.17563564e-02 2.43574440e-01 7.64799535e-01
-4.71480340e-02 1.82921588e-01 6.57628059e-01 3.58203620e-01
2.84502387e-01 3.59211229e-02 3.10811430e-01 -2.89895058e-01
-5.03569841e-01 5.40477633e-01 -1.11072779e+00 -2.08535250e-02
3.00821871e-01 -1.01087511e-01 -5.94382703e-01 -4.75014448e-02
-4.23228033e-02 2.34211698e-01 -6.49139941e-01 2.88179129e-01
1.54735312e-01 1.01630546e-01 1.58362672e-01 9.03377593e-01
-2.41301805e-01 5.84108531e-01 -1.25465071e+00 1.79409087e+00
-4.62461710e-01 8.49323332e-01 -9.77478921e-02 -1.36187577e+00
1.03121161e+00 4.92766768e-01 3.17381382e-01 -5.33091903e-01
1.98645115e-01 3.64372969e-01 7.38972008e-01 -3.98179814e-02
4.13992137e-01 -2.85156906e-01 7.00650066e-02 8.27102005e-01
-3.76649469e-01 -1.02405071e+00 9.13751245e-01 4.20466334e-01
1.08455312e+00 1.50280118e-01 8.00507292e-02 -4.54387426e-01
2.74794668e-01 4.54408765e-01 3.79383504e-01 1.04846191e+00
8.49046484e-02 1.51933562e-02 4.13230628e-01 -8.22918415e-01
-8.69428158e-01 -4.09101367e-01 4.05580908e-01 1.24069881e+00
-1.46112114e-01 -6.86945796e-01 -5.41376412e-01 -1.77313134e-01
-2.98245341e-01 8.71565461e-01 -2.92589486e-01 -2.26434544e-01
-5.22958159e-01 -4.72967267e-01 7.75950998e-02 4.39882576e-01
4.86816317e-01 -1.39571130e+00 -1.18489420e+00 4.10496354e-01
4.42346424e-01 -3.20797205e-01 2.41633356e-01 2.53289431e-01
-8.66775215e-01 -5.88622928e-01 -5.90313435e-01 -1.03407371e+00
5.35728991e-01 1.63750544e-01 1.12521160e+00 8.02197218e-01
-5.83177567e-01 5.61653137e-01 -3.02885085e-01 -7.85620987e-01
-5.53201079e-01 1.95126101e-01 -4.71077800e-01 -1.20134902e+00
7.26293921e-01 -8.43138337e-01 -3.36602449e-01 -1.85640827e-02
-6.66486681e-01 3.32020700e-01 2.77694762e-01 4.52596813e-01
-2.27555603e-01 7.13386774e-01 1.08600453e-01 -5.48360884e-01
8.26025009e-01 -1.91271175e-02 -9.39099312e-01 8.74993578e-02
-8.93694758e-01 3.44929434e-02 2.02137277e-01 -3.74866426e-01
-5.38427353e-01 -3.25154752e-01 -7.78038427e-02 3.79871130e-01
2.82576922e-02 9.15229976e-01 1.59003630e-01 -8.11164737e-01
9.44731176e-01 4.28224243e-02 1.03046664e-03 1.91636115e-01
-3.43546271e-01 1.74474437e-02 1.28670245e-01 -1.04238117e+00
6.28536284e-01 -4.27931815e-01 6.27899915e-02 -4.35607553e-01
-2.07360327e-01 -3.41777533e-01 -5.21339066e-02 -3.19286406e-01
4.56291378e-01 -6.86964333e-01 -1.50106239e+00 2.22117662e-01
-6.18466139e-01 -7.08948910e-01 -7.45032787e-01 9.83201742e-01
-4.73984599e-01 -2.03928292e-01 -3.75939578e-01 -7.45532095e-01
2.22498327e-01 -1.31913018e+00 -1.10876910e-01 1.01420927e+00
-4.98474389e-01 -9.69378710e-01 5.28948784e-01 7.97675788e-01
3.35143417e-01 1.16569094e-01 7.72654712e-01 -6.85620666e-01
-6.17802799e-01 -5.32627285e-01 4.17602450e-01 5.20284846e-02
-4.31810766e-01 4.90119271e-02 -2.75809914e-01 1.08970255e-01
-1.10304542e-01 -4.04220760e-01 3.47435057e-01 2.66749412e-01
6.00767076e-01 2.83419192e-01 -4.79678154e-01 -4.27921936e-02
1.52750540e+00 8.83698344e-01 7.53617644e-01 9.61305618e-01
-5.17954081e-02 5.20965040e-01 2.45610818e-01 2.17109308e-01
3.01665217e-01 4.90431309e-01 1.51605187e-02 3.28100324e-01
3.82212430e-01 -5.46995401e-02 1.94289580e-01 6.50130093e-01
-5.14478505e-01 1.32575899e-01 -1.33971524e+00 7.12079704e-01
-1.68687773e+00 -1.02991343e+00 -9.22534466e-02 2.02068162e+00
8.41396928e-01 6.25503123e-01 7.25986958e-01 3.57691020e-01
5.37049055e-01 -3.91990483e-01 1.23683780e-01 -1.13246882e+00
5.69778919e-01 9.96580720e-01 2.18053702e-02 7.38704383e-01
-3.91519427e-01 8.30309212e-01 6.46321058e+00 6.95251524e-01
-6.98283613e-01 -3.98849070e-01 4.17653561e-01 7.95898214e-02
-6.40971422e-01 3.95865560e-01 -6.52655721e-01 4.98731807e-03
8.42509449e-01 -6.42370582e-01 6.39837801e-01 9.08940673e-01
-2.30688810e-01 -4.27855432e-01 -7.46300101e-01 3.56682986e-01
-2.16248363e-01 -1.59878719e+00 -4.77677763e-01 3.30513269e-01
9.62890804e-01 -4.39976007e-01 -2.34866545e-01 5.61175406e-01
1.22937071e+00 -1.41587710e+00 3.59418958e-01 -1.32915452e-01
-3.07183772e-01 -1.15834343e+00 6.46082520e-01 3.42814803e-01
-8.17498147e-01 -4.92319763e-02 -2.33278662e-01 -1.12658823e+00
-3.00912589e-01 -1.87734738e-01 -8.74954402e-01 5.37156045e-01
4.37485993e-01 -1.27637625e-01 -4.71422463e-05 1.73028803e+00
-3.47813010e-01 2.29112923e-01 -6.94348276e-01 -9.02368307e-01
3.07573795e-01 -2.77393341e-01 2.41483733e-01 5.78464448e-01
2.24943742e-01 9.58465874e-01 2.68678427e-01 1.03730190e+00
4.95862931e-01 1.30636811e-01 -9.36314091e-02 -1.54809996e-01
4.50937748e-01 1.08800852e+00 -1.26500809e+00 5.70102260e-02
-1.03134796e-01 3.61493379e-01 -3.56980622e-01 -7.99332559e-02
-2.51145065e-01 -9.39063549e-01 3.92054051e-01 2.48743638e-01
2.51119584e-01 -2.74214864e-01 -7.38462985e-01 -6.62610531e-01
-4.39667612e-01 -1.34467947e+00 1.23860404e-01 -7.40120649e-01
-6.52496397e-01 4.97008830e-01 6.42305613e-02 -3.78237367e-01
-4.12812799e-01 -4.62797046e-01 -1.25387228e+00 1.02746999e+00
-6.18831217e-01 -4.44515079e-01 -6.38036728e-01 1.85638368e-01
2.78297752e-01 -4.48826194e-01 8.10449421e-01 -1.96642593e-01
-3.03622603e-01 4.23204303e-01 -3.05731803e-01 1.65970683e-01
-1.65514275e-01 -1.25715935e+00 4.12615478e-01 4.47850466e-01
4.98643637e-01 5.73075116e-01 7.79882312e-01 -3.74482363e-01
-9.41216528e-01 2.74890661e-01 6.68638825e-01 -8.41143057e-02
4.42406356e-01 2.39318937e-01 -2.93703526e-01 3.44672233e-01
4.01887357e-01 -5.55811524e-01 8.74596834e-01 2.00800925e-01
2.78281659e-01 -1.27800062e-01 -9.25597787e-01 4.18116271e-01
5.75337768e-01 -4.52296063e-02 -9.14193690e-01 1.41077429e-01
1.13442928e-01 -5.97076893e-01 -7.07909644e-01 1.08373828e-01
7.23856390e-01 -8.17238331e-01 9.60415661e-01 -6.45518780e-01
5.28788388e-01 -1.86063707e-01 2.36954719e-01 -1.15455759e+00
-2.43055001e-01 -9.50479329e-01 7.77430534e-01 7.28048146e-01
3.85689050e-01 -5.10628223e-01 1.56692481e+00 1.02286959e+00
4.09561694e-02 -5.99705398e-01 -3.65593225e-01 -6.30428195e-01
6.43952549e-01 -3.93660665e-01 4.05770510e-01 9.47303116e-01
8.66862953e-01 -4.03399169e-02 3.80346328e-01 -4.65574324e-01
3.02487552e-01 5.43434992e-02 8.21196616e-01 -1.76099491e+00
-4.50646222e-01 -1.03865838e+00 -9.35671687e-01 -6.09655321e-01
-3.50593328e-01 -4.54423755e-01 -6.03044853e-02 -1.56816006e+00
-3.78250688e-01 -6.51233554e-01 -9.32630748e-02 4.24310446e-01
1.45931887e-02 3.66156101e-01 3.04921299e-01 -2.07966730e-01
-2.33187243e-01 -3.10958773e-01 1.15901029e+00 3.61975640e-01
-5.20738602e-01 2.96002984e-01 -1.03412712e+00 7.71337867e-01
9.38598454e-01 -4.24190968e-01 -6.66398823e-01 2.52987444e-02
9.68346477e-01 7.87090138e-02 -3.26174721e-02 -1.43458331e+00
7.72321761e-01 -6.75888360e-01 2.50887632e-01 9.32427719e-02
2.75135875e-01 -6.05758548e-01 2.05367833e-01 9.55871701e-01
-5.15351832e-01 1.39178589e-01 6.15497112e-01 -5.00355661e-01
-2.91593164e-01 -9.75663900e-01 5.17416060e-01 -7.51429617e-01
-3.99558157e-01 -3.65248710e-01 -7.53175735e-01 1.31579518e-01
1.34764743e+00 -7.64311671e-01 1.31243616e-01 -8.22688282e-01
-1.12464297e+00 2.78564274e-01 2.87829667e-01 -2.78323799e-01
3.08027089e-01 -9.00657177e-01 -6.64646566e-01 6.51713833e-02
-4.23868924e-01 -5.90134189e-02 2.19578966e-02 6.03991508e-01
-1.50978160e+00 4.86783445e-01 -8.42045903e-01 -4.65704016e-02
-1.32677257e+00 5.96906617e-03 3.45189005e-01 -5.15429020e-01
-3.53158563e-01 1.53907943e+00 -1.26703735e-02 -5.16776085e-01
4.10117596e-01 9.97931734e-02 -7.43076950e-02 -1.34510875e-01
7.12391317e-01 2.89347649e-01 -1.11004941e-01 4.49556708e-01
-1.84835106e-01 2.81918615e-01 -1.80523902e-01 -5.35189807e-01
1.49845517e+00 3.32113743e-01 1.57783762e-01 1.49976045e-01
3.65963042e-01 4.48478200e-03 -3.77764404e-01 2.93290734e-01
-2.48186439e-02 -4.02642488e-01 1.39541730e-01 -1.11403358e+00
-8.55197430e-01 8.99106801e-01 1.77147076e-01 4.17383939e-01
9.81174409e-01 -6.78412259e-01 2.60176271e-01 5.13365328e-01
3.96949500e-01 -9.45303380e-01 1.06791437e-01 3.99059057e-01
3.87344718e-01 -5.91852009e-01 3.21604431e-01 -3.66804570e-01
-4.64224219e-01 1.59468687e+00 9.82603848e-01 -3.65166366e-01
4.56153870e-01 8.76941308e-02 -1.95292324e-01 -2.56997794e-01
-7.98168123e-01 -4.80316952e-02 8.96880701e-02 5.15819132e-01
6.16209686e-01 -2.89717227e-01 -7.26603210e-01 1.39094591e-01
-7.12473929e-01 5.06045163e-01 9.47226107e-01 1.30499482e+00
-8.09978783e-01 -1.59304833e+00 -5.20668447e-01 -8.54395926e-02
-5.36481857e-01 -3.71989399e-01 -5.53099573e-01 1.02340806e+00
2.70514399e-01 5.22857904e-01 8.02865848e-02 -2.16013342e-01
1.44789740e-02 5.30531779e-02 1.17099643e+00 -7.36271381e-01
-1.22251880e+00 -3.69027972e-01 -3.27388942e-02 3.94012332e-02
-1.52955070e-01 -5.09477258e-01 -1.25504327e+00 -1.24048710e+00
-3.88699919e-01 1.20865691e+00 8.53358746e-01 8.81717741e-01
3.50049101e-02 4.72482443e-01 2.64298264e-02 -3.21067154e-01
-4.61051106e-01 -5.90945542e-01 -4.20575470e-01 -5.81085145e-01
-4.25657719e-01 -3.99989933e-01 -1.04829721e-01 -4.24493253e-01] | [3.4597327709198, 1.4891256093978882] |
8e18831d-b5db-4047-8019-f2b9ed9e98f1 | gfpose-learning-3d-human-pose-prior-with | 2212.08641 | null | https://arxiv.org/abs/2212.08641v1 | https://arxiv.org/pdf/2212.08641v1.pdf | GFPose: Learning 3D Human Pose Prior with Gradient Fields | Learning 3D human pose prior is essential to human-centered AI. Here, we present GFPose, a versatile framework to model plausible 3D human poses for various applications. At the core of GFPose is a time-dependent score network, which estimates the gradient on each body joint and progressively denoises the perturbed 3D human pose to match a given task specification. During the denoising process, GFPose implicitly incorporates pose priors in gradients and unifies various discriminative and generative tasks in an elegant framework. Despite the simplicity, GFPose demonstrates great potential in several downstream tasks. Our experiments empirically show that 1) as a multi-hypothesis pose estimator, GFPose outperforms existing SOTAs by 20% on Human3.6M dataset. 2) as a single-hypothesis pose estimator, GFPose achieves comparable results to deterministic SOTAs, even with a vanilla backbone. 3) GFPose is able to produce diverse and realistic samples in pose denoising, completion and generation tasks. Project page https://sites.google.com/view/gfpose/ | ['Yizhou Wang', 'Fangwei Zhong', 'Hao Dong', 'Xiaoxuan Ma', 'Wentao Zhu', 'Mingdong Wu', 'Hai Ci'] | 2022-12-16 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Ci_GFPose_Learning_3D_Human_Pose_Prior_With_Gradient_Fields_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Ci_GFPose_Learning_3D_Human_Pose_Prior_With_Gradient_Fields_CVPR_2023_paper.pdf | cvpr-2023-1 | ['multi-hypotheses-3d-human-pose-estimation', '3d-human-pose-estimation', 'monocular-3d-human-pose-estimation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-6.90187365e-02 1.67924121e-01 9.66406912e-02 -2.56119728e-01
-9.18780267e-01 -3.15068007e-01 4.03891295e-01 -6.80907488e-01
-3.03351969e-01 5.90363681e-01 5.04703045e-01 4.43494499e-01
8.14709887e-02 -2.87276566e-01 -7.86270738e-01 -4.50547338e-01
-9.69872251e-02 9.02824640e-01 4.24623303e-02 -3.86076748e-01
-2.17923760e-01 3.53035927e-01 -1.36916912e+00 -1.97277859e-01
7.17165530e-01 7.24252343e-01 1.21462703e-01 8.84246409e-01
5.73517084e-01 3.45156848e-01 -6.45478964e-01 -5.60117185e-01
4.36528772e-01 -3.06913108e-01 -5.20144224e-01 1.63147315e-01
7.98457623e-01 -3.25477391e-01 -3.87252510e-01 7.53204226e-01
9.55124199e-01 2.98178047e-01 6.84825838e-01 -1.42103374e+00
-3.06265473e-01 1.85289219e-01 -5.08976579e-01 -2.12734833e-01
7.17176378e-01 5.70743382e-01 7.30992675e-01 -1.08665764e+00
7.98802435e-01 1.52638018e+00 1.04475784e+00 7.63165057e-01
-1.31183827e+00 -6.31765783e-01 7.48755187e-02 -2.54840195e-01
-1.47764885e+00 -4.52323407e-01 6.45529747e-01 -4.71661448e-01
5.94825149e-01 2.36317024e-01 1.03485692e+00 1.88377988e+00
4.49729025e-01 1.00421262e+00 7.78844237e-01 9.65868980e-02
1.29954919e-01 -8.00770760e-01 -2.86605895e-01 8.84882152e-01
6.52137697e-02 1.67570949e-01 -9.99981880e-01 -2.02574357e-01
9.81912613e-01 -2.03775153e-01 -1.10933892e-01 -7.71699131e-01
-1.43351138e+00 5.61880112e-01 4.53957886e-01 -4.19007599e-01
-5.34080327e-01 8.25805604e-01 2.58275062e-01 -2.01590255e-01
4.01172400e-01 3.47054213e-01 -3.67039829e-01 -2.53730386e-01
-9.23625767e-01 9.77239311e-01 7.49621809e-01 1.02302253e+00
2.50368863e-01 1.28647223e-01 -4.31184798e-01 6.25407159e-01
4.97636676e-01 8.83752823e-01 -3.61533538e-02 -1.46531320e+00
4.06013429e-01 7.97501653e-02 2.40366250e-01 -9.36291873e-01
-5.49442589e-01 -7.71343172e-01 -8.55975449e-01 1.75312161e-01
3.65624309e-01 -2.98192650e-01 -1.19908774e+00 2.00235438e+00
7.62618184e-01 1.77180603e-01 -4.34972435e-01 1.19016457e+00
7.60548115e-01 3.85234237e-01 2.28616536e-01 3.31199288e-01
1.13782370e+00 -1.03919792e+00 -5.84423304e-01 -6.97178721e-01
1.32636592e-01 -5.45860589e-01 9.55116510e-01 7.76391506e-01
-1.24220729e+00 -7.34447181e-01 -8.64065230e-01 -2.89801568e-01
2.14082912e-01 1.78437889e-01 7.36665010e-01 4.55513299e-01
-1.01669502e+00 6.88018560e-01 -1.18465602e+00 -2.80952245e-01
2.56149530e-01 2.42496833e-01 -5.66006243e-01 -8.05953741e-02
-1.12257743e+00 9.35070276e-01 1.75606340e-01 4.75088060e-01
-1.26631355e+00 -5.20354271e-01 -1.17311227e+00 -3.31748545e-01
4.58392352e-01 -1.41141856e+00 1.36398029e+00 -4.28294033e-01
-1.62598395e+00 7.71837175e-01 -1.19810574e-01 -3.53026003e-01
1.16759944e+00 -9.66562569e-01 1.35052845e-01 3.89172211e-02
2.80981272e-01 1.17561495e+00 1.00727665e+00 -1.22059619e+00
-1.47211567e-01 -4.31192547e-01 -4.03406650e-01 5.71914792e-01
3.79975110e-01 -2.88991123e-01 -1.06211734e+00 -9.22854006e-01
1.87484801e-01 -1.22084486e+00 -5.35335958e-01 4.03826386e-01
-6.93545461e-01 -3.99144776e-02 2.20517904e-01 -8.93117487e-01
8.96343768e-01 -1.74382973e+00 8.88461113e-01 2.76898950e-01
3.22011650e-01 -2.05691755e-01 -4.85737287e-02 2.65673041e-01
1.38308734e-01 -2.67722756e-01 -3.72536272e-01 -1.00511706e+00
4.59419549e-01 1.91492602e-01 7.93669671e-02 7.10088432e-01
-7.55326124e-03 1.24791026e+00 -8.56168926e-01 -5.85187793e-01
3.60915601e-01 9.79783237e-01 -7.35504389e-01 1.07969493e-01
-2.92430133e-01 8.32859814e-01 -3.50660622e-01 7.60608375e-01
5.03757238e-01 -1.20758556e-01 1.45704165e-01 -2.84426183e-01
3.56952697e-01 -1.54154465e-01 -1.31650126e+00 2.39190555e+00
-2.49400754e-02 2.65091896e-01 2.85547584e-01 -3.07801336e-01
7.06488609e-01 2.49693722e-01 4.67704117e-01 -2.24429965e-01
2.98962355e-01 4.62011360e-02 -4.29114938e-01 -2.54629403e-01
4.73136723e-01 1.12149920e-02 -4.80044842e-01 -2.46189963e-02
2.35650390e-01 -3.95085692e-01 9.35080498e-02 1.36556089e-01
9.70234811e-01 9.27595615e-01 2.60022074e-01 -1.56288296e-01
5.43487445e-02 -2.72627085e-01 7.52581060e-01 6.20383263e-01
-4.40119445e-01 1.12062132e+00 2.83701360e-01 -4.35336679e-01
-9.35202062e-01 -1.70345342e+00 9.83287692e-02 1.13109756e+00
1.66318148e-01 -5.01652300e-01 -7.83574402e-01 -5.13120890e-01
3.43510389e-01 4.84155685e-01 -7.38424778e-01 -4.03985083e-02
-6.21826828e-01 -5.54682791e-01 6.99829102e-01 7.15380430e-01
3.70227277e-01 -1.07133055e+00 -7.01446414e-01 5.40838093e-02
-4.74709898e-01 -8.63057792e-01 -8.24983776e-01 3.35908569e-02
-7.07155049e-01 -9.69847202e-01 -1.05311370e+00 -4.68313307e-01
5.34864426e-01 -2.39181191e-01 1.34942496e+00 -2.52658993e-01
-4.99442190e-01 6.40342295e-01 -1.26287177e-01 -3.41264158e-01
-1.68620825e-01 -6.19847029e-02 3.60030919e-01 -3.87098163e-01
-5.23324050e-02 -6.07284248e-01 -8.15740287e-01 3.22163910e-01
-4.67448264e-01 9.35336500e-02 4.06501651e-01 6.99677527e-01
7.32384682e-01 -6.09450817e-01 1.91768542e-01 -4.29853827e-01
5.74378371e-01 -2.01727942e-01 -3.01276058e-01 -3.43115106e-02
-8.73457491e-02 -3.14213894e-02 -8.99448544e-02 -3.86863053e-01
-9.63686049e-01 5.85059226e-01 -3.75249565e-01 -5.64464390e-01
-2.28320047e-01 1.19576767e-01 -3.16374093e-01 2.45704636e-01
9.63006616e-01 2.00903155e-02 2.58726031e-01 -5.67566276e-01
5.79128146e-01 5.69955297e-02 1.06947386e+00 -8.76394212e-01
9.55600560e-01 4.43426758e-01 7.29991794e-02 -5.15323758e-01
-9.02605236e-01 -3.31757516e-01 -8.37273419e-01 -5.01192033e-01
1.11278403e+00 -1.26498699e+00 -9.04453337e-01 5.53938091e-01
-1.10891628e+00 -6.38757586e-01 -1.64824739e-01 4.13900405e-01
-9.17507827e-01 3.63544166e-01 -7.09165037e-01 -7.45482802e-01
-5.38775861e-01 -1.11825633e+00 1.72547019e+00 -3.23364213e-02
-8.58006001e-01 -7.03280926e-01 1.95514038e-01 6.21966183e-01
-7.23754466e-02 8.30919325e-01 -6.05670363e-02 5.73836342e-02
-2.80473024e-01 -1.95228726e-01 3.75259042e-01 2.42895827e-01
-1.69338301e-01 -9.92697254e-02 -7.81827748e-01 -4.29112732e-01
-3.97224844e-01 -6.03836060e-01 6.90235317e-01 7.27120876e-01
9.61102188e-01 7.68340528e-02 -2.85760671e-01 9.51928854e-01
7.42222369e-01 -3.87036979e-01 6.27676547e-01 1.15406245e-01
9.22126532e-01 5.41755438e-01 7.17115939e-01 5.83926260e-01
4.25425738e-01 7.62206733e-01 4.28808540e-01 -6.96329325e-02
-1.19887978e-01 -6.26764059e-01 5.69159329e-01 4.90161002e-01
-3.48235875e-01 -2.26175696e-01 -9.37811732e-01 2.39872336e-01
-1.97252882e+00 -8.71053398e-01 5.16900532e-02 1.96584272e+00
7.75158584e-01 2.16170013e-01 3.92934352e-01 -2.91529536e-01
5.82241893e-01 2.68560976e-01 -8.09570730e-01 1.25995189e-01
2.02632393e-03 3.76944214e-01 3.12731028e-01 4.92853314e-01
-1.27822816e+00 9.74028051e-01 6.61801672e+00 6.92500830e-01
-4.68465000e-01 2.38457248e-02 3.66803497e-01 -4.45867747e-01
5.36636859e-02 -3.44255120e-01 -7.07769096e-01 3.42172772e-01
4.03320968e-01 2.24232934e-02 2.24019393e-01 9.08885181e-01
3.43917429e-01 -2.89741624e-02 -1.09597778e+00 1.14735091e+00
1.37803599e-01 -8.73635411e-01 -4.26926166e-02 4.31157909e-02
5.65382004e-01 -1.20685166e-02 1.67935804e-01 2.39398062e-01
3.79037440e-01 -1.25908208e+00 1.25361466e+00 6.91708982e-01
7.28213072e-01 -9.00294065e-01 4.09811288e-01 4.59686428e-01
-1.16156137e+00 2.85786927e-01 2.42380835e-02 2.60358322e-02
6.77831590e-01 3.64671856e-01 -5.59901953e-01 4.25513327e-01
8.46100569e-01 6.11464381e-01 -5.31692982e-01 1.05322874e+00
-7.35941947e-01 2.46969104e-01 -4.80466545e-01 3.25660765e-01
-6.83984756e-02 -1.03652857e-01 8.24431539e-01 1.14335394e+00
2.43895456e-01 -1.19079784e-01 5.32335281e-01 8.07622731e-01
9.48023200e-02 -2.56546170e-01 -3.47898692e-01 4.45032924e-01
4.09934938e-01 9.65252876e-01 -5.01132667e-01 -1.89931050e-01
3.98671985e-01 1.56101632e+00 2.60920316e-01 3.41478765e-01
-1.24947250e+00 -1.70742685e-03 8.94598544e-01 1.12039320e-01
1.05827726e-01 -5.43346345e-01 -1.57197714e-01 -1.11718142e+00
1.76695243e-01 -1.10504508e+00 3.37095529e-01 -1.00455070e+00
-1.20832813e+00 3.70252222e-01 2.45152667e-01 -1.06602383e+00
-4.65680510e-01 -4.64836270e-01 -3.04661214e-01 7.20489025e-01
-5.67402065e-01 -1.35322356e+00 -5.10462105e-01 5.78381717e-01
5.44860542e-01 3.76314044e-01 6.66495919e-01 1.33322448e-01
-6.15403295e-01 5.23692429e-01 -4.67931181e-01 6.36475161e-02
9.14586067e-01 -1.41210079e+00 1.08278525e+00 7.62211859e-01
3.28576826e-02 8.05190027e-01 1.09138548e+00 -1.10564399e+00
-1.37602091e+00 -1.01014209e+00 4.46170419e-01 -9.12851095e-01
2.33083755e-01 -5.95820546e-01 -3.88402313e-01 8.87370765e-01
-1.15853816e-01 2.96627451e-02 3.70128512e-01 1.59960717e-01
-2.39624634e-01 2.64093190e-01 -1.02928245e+00 9.24675107e-01
1.75971687e+00 -1.94827706e-01 -6.86097920e-01 4.90932941e-01
5.60153723e-01 -1.06380570e+00 -8.98779750e-01 4.61581647e-01
8.38204086e-01 -9.30406809e-01 1.36075687e+00 -4.08286065e-01
3.69650215e-01 -4.59048063e-01 -8.61004069e-02 -1.31378829e+00
-5.26779532e-01 -9.72304165e-01 -5.04360199e-01 5.87606251e-01
1.24140427e-01 -2.06922829e-01 1.20777380e+00 5.42935908e-01
-8.48188400e-02 -7.33784497e-01 -9.62352931e-01 -8.36874425e-01
-2.24359948e-02 -6.95018888e-01 2.13843480e-01 3.20361257e-01
-5.32858968e-01 2.10278943e-01 -9.34330404e-01 2.57116646e-01
1.23657823e+00 -3.57174993e-01 1.25846708e+00 -1.11639893e+00
-5.36725283e-01 -1.57942563e-01 -3.74772906e-01 -1.33629310e+00
1.41290084e-01 -6.65941238e-01 4.53460842e-01 -1.50744772e+00
1.16250262e-01 -7.37911277e-03 2.09319457e-01 4.58363652e-01
-3.99756700e-01 4.87373710e-01 3.58866900e-01 -7.40340501e-02
-6.67797029e-01 8.29845250e-01 1.35223126e+00 4.30044858e-03
-2.17890963e-01 3.12376469e-02 -4.04137343e-01 9.25339937e-01
6.30070686e-01 -2.76365250e-01 -4.12897259e-01 -4.36751783e-01
2.22464606e-01 -2.68007927e-02 9.06815171e-01 -1.30839968e+00
2.22582817e-02 -4.55608182e-02 8.92371774e-01 -7.61926830e-01
8.99579227e-01 -3.06505680e-01 7.03092754e-01 6.26431882e-01
-1.59968324e-02 1.62016675e-01 1.53063303e-02 6.95772946e-01
1.24023259e-01 3.24270159e-01 6.42159998e-01 -2.65254766e-01
-8.56921256e-01 5.33351064e-01 5.71384619e-04 3.50675374e-01
8.13286304e-01 -3.77560198e-01 2.77689487e-01 -4.92466092e-01
-1.04882169e+00 4.79754716e-01 5.02084732e-01 4.52980250e-01
7.38587141e-01 -1.40753484e+00 -7.45086253e-01 -9.86451060e-02
1.01473257e-02 4.19225663e-01 3.77746999e-01 6.57265782e-01
-7.39172876e-01 -3.68791819e-02 -1.07889637e-01 -8.71200323e-01
-1.22512805e+00 1.23820528e-01 4.72637504e-01 -7.51037002e-02
-7.99843431e-01 1.20461583e+00 1.68107748e-01 -7.38559484e-01
5.16324699e-01 -8.32067207e-02 3.78657967e-01 -1.34405568e-01
2.91116416e-01 5.65846682e-01 -2.89288133e-01 -8.16230118e-01
-5.38792312e-01 6.41838729e-01 3.78152162e-01 -4.37230945e-01
1.40165353e+00 -9.20340302e-04 3.15040767e-01 1.91950649e-01
9.48252082e-01 -1.36413695e-02 -1.78665984e+00 2.22713113e-01
-3.22537899e-01 -2.86174387e-01 -2.95032263e-01 -8.08631837e-01
-8.92117023e-01 5.07542431e-01 3.66244197e-01 -8.11385393e-01
8.20753813e-01 -3.06184497e-02 8.45186234e-01 2.45832384e-01
6.71315789e-01 -1.07938290e+00 3.07450205e-01 5.25717854e-01
1.34617698e+00 -9.17204857e-01 3.30268562e-01 -5.28645217e-01
-8.78108323e-01 6.19479656e-01 7.67922640e-01 -3.40114564e-01
3.41771424e-01 2.39742130e-01 1.53411373e-01 -3.83438855e-01
-5.03821254e-01 -1.66143939e-01 6.81437254e-01 7.36030161e-01
4.12711799e-01 9.95470956e-02 -6.23852573e-02 4.45665032e-01
-6.69861436e-01 -1.11899506e-02 -6.53378218e-02 9.91425395e-01
-2.39819676e-01 -8.88471365e-01 -7.51908064e-01 5.01210168e-02
-2.15604752e-01 1.85736910e-01 -4.85114068e-01 8.74252439e-01
1.92154601e-01 5.80929279e-01 -3.19475979e-01 -4.04962450e-01
5.81765890e-01 1.00306660e-01 7.81530559e-01 -5.08222938e-01
-5.40540516e-01 2.10609227e-01 1.50801763e-01 -1.13492417e+00
-1.83721632e-01 -7.77759016e-01 -1.20170462e+00 -2.00046182e-01
5.89394867e-02 -1.79508388e-01 4.51419502e-01 5.94781458e-01
4.14838403e-01 4.69803065e-01 -2.17569359e-02 -1.64020562e+00
-6.48344576e-01 -9.44618464e-01 -3.89087647e-01 6.89000487e-01
1.71677887e-01 -1.04514766e+00 -3.02775502e-01 9.30261984e-02] | [7.024331092834473, -0.877502977848053] |
a6ae194d-dafc-4900-bb4c-f02f1b6fb37c | from-alignment-to-entailment-a-unified | 2305.11501 | null | https://arxiv.org/abs/2305.11501v1 | https://arxiv.org/pdf/2305.11501v1.pdf | From Alignment to Entailment: A Unified Textual Entailment Framework for Entity Alignment | Entity Alignment (EA) aims to find the equivalent entities between two Knowledge Graphs (KGs). Existing methods usually encode the triples of entities as embeddings and learn to align the embeddings, which prevents the direct interaction between the original information of the cross-KG entities. Moreover, they encode the relational triples and attribute triples of an entity in heterogeneous embedding spaces, which prevents them from helping each other. In this paper, we transform both triples into unified textual sequences, and model the EA task as a bi-directional textual entailment task between the sequences of cross-KG entities. Specifically, we feed the sequences of two entities simultaneously into a pre-trained language model (PLM) and propose two kinds of PLM-based entity aligners that model the entailment probability between sequences as the similarity between entities. Our approach captures the unified correlation pattern of two kinds of information between entities, and explicitly models the fine-grained interaction between original entity information. The experiments on five cross-lingual EA datasets show that our approach outperforms the state-of-the-art EA methods and enables the mutual enhancement of the heterogeneous information. Codes are available at https://github.com/OreOZhao/TEA. | ['Xiaojie Yuan', 'Haiwei Zhang', 'Ying Zhang', 'Xiangrui Cai', 'Yike Wu', 'Yu Zhao'] | 2023-05-19 | null | null | null | null | ['entity-alignment', 'entity-alignment'] | ['knowledge-base', 'natural-language-processing'] | [-4.92809922e-01 2.73708016e-01 -3.90603662e-01 -3.75363648e-01
-3.72787774e-01 -6.42465711e-01 4.52175707e-01 5.69893479e-01
-4.55752492e-01 2.71788061e-01 4.99609768e-01 -2.10145861e-01
-1.66670710e-01 -1.15368569e+00 -9.00792003e-01 -5.46122372e-01
-1.63059011e-01 6.02038324e-01 8.69285539e-02 -3.80221725e-01
-1.35352775e-01 4.37728800e-02 -1.14772761e+00 2.57229865e-01
1.16527939e+00 7.71703899e-01 1.18951321e-01 2.28573620e-01
-5.68429053e-01 8.81043494e-01 -6.29903749e-02 -1.08567595e+00
1.89946145e-02 -4.05431688e-02 -9.12590921e-01 -4.24082637e-01
4.19951797e-01 -7.34268501e-02 -8.99105549e-01 1.17135739e+00
5.04299760e-01 -1.25981450e-01 7.45202541e-01 -1.58556414e+00
-1.55817437e+00 9.36699033e-01 -5.21953464e-01 3.23760323e-02
4.48457330e-01 -3.48862499e-01 1.69080222e+00 -1.13012791e+00
6.99670732e-01 1.08182681e+00 7.97788322e-01 1.26309559e-01
-9.32930648e-01 -4.67047721e-01 -1.37239145e-02 9.16039050e-01
-1.75725853e+00 -1.25659049e-01 7.35764742e-01 -5.12080669e-01
1.13709986e+00 1.57827929e-01 6.84869289e-01 8.41677785e-01
6.66993335e-02 8.87202084e-01 2.30094850e-01 -4.03507650e-01
-3.11207056e-01 2.88162440e-01 3.80577445e-01 7.61976779e-01
5.58093309e-01 -2.67968446e-01 -4.50945973e-01 -2.51290351e-01
2.06943616e-01 -2.41569318e-02 -4.04560834e-01 -5.78828454e-01
-1.20793569e+00 7.42270350e-01 6.46417558e-01 6.15606666e-01
-2.65767694e-01 -5.18351272e-02 4.34675545e-01 9.78592187e-02
1.69350475e-01 3.91306907e-01 -6.63029611e-01 2.89406121e-01
-2.82291055e-01 7.02686012e-02 8.12772274e-01 1.51231217e+00
9.70174968e-01 -4.97669458e-01 -6.39086068e-02 5.82894206e-01
7.18797565e-01 2.90766507e-01 5.11506736e-01 -1.35576501e-01
8.86077166e-01 1.02291763e+00 6.18748963e-02 -1.57910848e+00
-2.00897634e-01 -3.80349010e-01 -7.06063569e-01 -8.04233372e-01
7.14196116e-02 9.53714177e-02 -3.13594013e-01 1.91936100e+00
6.41613007e-01 3.78502131e-01 3.10975134e-01 5.53893030e-01
1.10770905e+00 8.13066065e-01 1.52562425e-01 6.95940703e-02
1.63511562e+00 -1.12937832e+00 -9.50459361e-01 -2.56622344e-01
1.09228861e+00 -6.30825877e-01 8.90437484e-01 -6.13206029e-01
-8.29738140e-01 -4.90613580e-01 -9.92783725e-01 -5.08752942e-01
-9.60075498e-01 3.63510311e-01 3.19706947e-01 1.89782888e-01
-6.01073980e-01 5.76984644e-01 -6.09696031e-01 -1.90021694e-01
6.33054301e-02 1.33579224e-01 -6.29325449e-01 -1.59216940e-01
-2.05900693e+00 1.10858667e+00 8.49493444e-01 3.51793438e-01
1.50624424e-01 -1.01325774e+00 -1.42203820e+00 2.22647265e-01
3.81056994e-01 -8.08278441e-01 6.12467170e-01 -4.20802325e-01
-8.33738089e-01 9.58698452e-01 -2.24425718e-01 -6.40858933e-02
1.54315338e-01 -4.11813319e-01 -9.80664790e-01 -2.35307589e-01
3.80845249e-01 2.26553634e-01 2.64052451e-01 -1.02401137e+00
-5.76508284e-01 -4.21859443e-01 8.90506804e-02 4.33568180e-01
-5.60048997e-01 -1.45984903e-01 -7.80589819e-01 -5.44516206e-01
-9.33226869e-02 -8.37235093e-01 2.26941034e-01 -3.49662781e-01
-6.81523561e-01 -4.70702618e-01 6.56530797e-01 -9.18983698e-01
1.82068813e+00 -2.27294755e+00 4.13531810e-01 1.19404592e-01
3.08349609e-01 1.51412040e-01 -2.48014942e-01 9.46094036e-01
-2.81715751e-01 3.21812600e-01 -6.40171841e-02 -2.08379805e-01
4.57456589e-01 3.11834991e-01 -2.53428698e-01 2.98658073e-01
3.11900705e-01 1.37503827e+00 -1.20206749e+00 -6.56821728e-01
-1.44041523e-01 4.11465138e-01 -2.78175950e-01 4.07779008e-01
1.44230634e-01 -5.15845679e-02 -4.46533889e-01 2.58239865e-01
6.95366919e-01 -4.41987246e-01 6.68223858e-01 -9.70812738e-01
1.38669938e-01 5.19323409e-01 -1.24613094e+00 1.61276710e+00
-3.08988839e-01 2.34790921e-01 -6.92296445e-01 -8.30883741e-01
6.67150795e-01 2.52261043e-01 4.27297235e-01 -5.69992244e-01
-1.13435484e-01 3.48786473e-01 -8.11925828e-02 -6.79355741e-01
8.83647442e-01 1.05386160e-01 -3.25017065e-01 9.34209004e-02
3.30415308e-01 3.74046564e-01 3.14173102e-01 5.28629363e-01
9.52709615e-01 5.97925112e-02 5.73711157e-01 -6.84503168e-02
5.68254709e-01 -3.02412748e-01 6.91220939e-01 3.71031165e-01
1.10217683e-01 -3.37122828e-02 3.98706794e-01 -3.39124262e-01
-1.16009963e+00 -1.17376232e+00 -6.63915947e-02 8.65997553e-01
7.70717740e-01 -9.45475042e-01 -2.93639898e-01 -1.01901650e+00
3.38815570e-01 8.10357988e-01 -7.77985215e-01 -4.66216058e-01
-5.06856024e-01 -6.87299311e-01 6.54317319e-01 6.01608038e-01
3.73500675e-01 -6.04531348e-01 3.13293785e-01 1.96743071e-01
-5.15304863e-01 -1.21047437e+00 -9.06059980e-01 1.46003226e-02
-3.52505594e-01 -1.05917990e+00 -2.41072133e-01 -1.14529502e+00
6.95044994e-01 -9.55564454e-02 1.31794608e+00 9.65484306e-02
-3.27575132e-02 3.07809412e-01 -4.21568066e-01 1.50723338e-01
-9.39176008e-02 2.22937673e-01 2.37268239e-01 1.69303939e-01
8.21667314e-01 -4.74253863e-01 -3.82551819e-01 1.92461520e-01
-9.38183188e-01 1.10706270e-01 5.11805296e-01 9.84046102e-01
7.20248580e-01 8.75617191e-02 3.54249179e-01 -9.64800537e-01
4.17878836e-01 -9.67601895e-01 -3.06930304e-01 8.76831889e-01
-5.97558081e-01 4.98160690e-01 6.68342710e-01 -4.25198913e-01
-8.23741555e-01 -3.20089877e-01 -3.31375003e-03 -3.94884288e-01
2.58647710e-01 9.95524347e-01 -6.07593894e-01 2.82086402e-01
1.16713494e-01 2.49230489e-01 -8.01894009e-01 -3.63534868e-01
7.85822809e-01 5.02822757e-01 4.93630797e-01 -7.16574609e-01
9.50878322e-01 -1.00010233e-02 -3.31506789e-01 -3.42932165e-01
-8.34481776e-01 -6.26836360e-01 -7.94457614e-01 1.60186857e-01
7.56610036e-01 -9.23662841e-01 -4.17606860e-01 3.10750753e-01
-1.39428067e+00 1.54583335e-01 -2.77764916e-01 4.97745007e-01
-1.47374958e-01 5.94308853e-01 -7.40484774e-01 -1.28404155e-01
-1.07951805e-01 -7.72980630e-01 9.11317766e-01 3.09261829e-01
-1.68450907e-01 -1.28418875e+00 4.42674130e-01 1.46200955e-01
-7.40749435e-03 -1.25477403e-01 1.36005521e+00 -1.03455329e+00
-5.58600783e-01 -2.87649184e-01 -2.56409138e-01 2.80048735e-02
3.81142795e-01 2.20155627e-01 -4.86849695e-01 -1.34680375e-01
-4.68750626e-01 -3.90973613e-02 4.88502383e-01 -2.02328220e-01
6.32171750e-01 -6.33538127e-01 -5.85902810e-01 5.53460062e-01
1.49813390e+00 -6.21198937e-02 7.03932703e-01 3.68814290e-01
1.23841619e+00 5.52303791e-01 5.43095350e-01 1.46680906e-01
1.09474576e+00 8.46252203e-01 2.34671250e-01 1.65922821e-01
1.42436787e-01 -9.75386441e-01 2.88093567e-01 1.62778449e+00
1.44107491e-01 -4.51166689e-01 -8.48528028e-01 9.51152563e-01
-2.04643941e+00 -1.07868981e+00 -3.33483666e-01 1.89542997e+00
1.11264431e+00 -3.24173540e-01 -3.84712338e-01 -3.53531033e-01
8.67606103e-01 1.94128051e-01 -4.11821276e-01 -1.49498452e-02
-2.92702764e-01 -2.58466721e-01 5.61451435e-01 5.19626141e-01
-1.10486758e+00 8.41533303e-01 4.64048147e+00 9.61720288e-01
-5.11568487e-01 1.07679829e-01 2.67839860e-02 2.91929543e-01
-9.19654965e-01 1.44385979e-01 -1.02488494e+00 8.41114640e-01
7.13735640e-01 -7.21699774e-01 1.08972982e-01 7.30493665e-01
-5.82981884e-01 4.52877134e-01 -1.25562930e+00 8.38113070e-01
-6.96432739e-02 -1.30432105e+00 2.20927626e-01 -5.67424186e-02
6.70679867e-01 2.63958983e-03 -7.43071288e-02 6.35808647e-01
4.43344921e-01 -7.21762896e-01 6.78574979e-01 5.88474512e-01
4.71992403e-01 -6.75414681e-01 1.08554173e+00 6.20025434e-02
-1.90102613e+00 3.43758792e-01 -5.07765710e-01 4.19976473e-01
3.09365004e-01 6.36119604e-01 -5.55841267e-01 1.46492219e+00
7.12108493e-01 1.08698416e+00 -4.82340455e-01 6.19850039e-01
-5.52328765e-01 1.06853314e-01 -1.25215068e-01 8.01048577e-02
6.55982643e-02 -5.21774471e-01 4.86127436e-01 1.33009958e+00
4.58616942e-01 -4.39864434e-02 -8.55866596e-02 8.86384368e-01
-5.05577981e-01 4.92827684e-01 -7.59528041e-01 -3.96603227e-01
9.73668337e-01 1.24575138e+00 -4.44604363e-03 -4.24494505e-01
-7.53769040e-01 1.23899853e+00 1.02534318e+00 2.39935756e-01
-1.00580823e+00 -6.66711926e-01 8.40890944e-01 -3.04352701e-01
5.47028720e-01 9.32585448e-03 1.06963143e-01 -1.58156240e+00
4.64370459e-01 -6.27789676e-01 6.59750581e-01 -7.67246246e-01
-1.79522717e+00 6.20901406e-01 -2.24506497e-01 -1.30005968e+00
1.13360651e-01 -4.71214235e-01 -5.25677323e-01 8.16632330e-01
-1.54925334e+00 -1.31858385e+00 -7.36082643e-02 4.73327130e-01
-1.97451472e-01 8.50496348e-03 9.08612311e-01 8.10880363e-01
-8.36715698e-01 9.46076810e-01 3.28927547e-01 6.65082097e-01
7.97961056e-01 -1.26075947e+00 5.02522767e-01 8.44048560e-01
4.33846653e-01 1.06946790e+00 4.97747451e-01 -8.13015640e-01
-1.52567768e+00 -1.23491275e+00 1.60046220e+00 -5.78544140e-01
1.16971099e+00 -3.14390808e-01 -1.18646538e+00 1.09135485e+00
4.13648039e-01 1.17962532e-01 1.07339323e+00 2.88856477e-01
-8.52136195e-01 5.72089478e-02 -6.97700381e-01 7.54868031e-01
1.22525072e+00 -1.02970243e+00 -1.06254745e+00 2.60195956e-02
1.06811595e+00 -4.33997750e-01 -1.38789284e+00 4.91128981e-01
4.91311491e-01 -5.02414405e-01 9.58982527e-01 -1.12248230e+00
6.21156752e-01 -6.25613332e-01 -3.66404504e-01 -1.53863251e+00
-4.77883786e-01 -1.21589832e-01 -5.76868057e-01 1.60914814e+00
6.35872722e-01 -5.96306324e-01 3.76592517e-01 5.13562679e-01
-4.36521359e-02 -7.98293829e-01 -7.01347828e-01 -8.80948901e-01
-1.35186628e-01 -8.59099105e-02 9.32587445e-01 1.64360023e+00
4.88972068e-01 4.96068895e-01 -3.17810148e-01 7.48306632e-01
3.70628983e-01 3.71728212e-01 4.86697197e-01 -9.41356122e-01
-2.18393683e-01 -1.59364760e-01 -7.83962965e-01 -1.05728471e+00
4.78349000e-01 -1.36888659e+00 -2.72027224e-01 -1.52754772e+00
3.25666845e-01 -4.87941653e-01 -4.12541777e-01 5.36738992e-01
-6.93937600e-01 -3.73102427e-01 -1.68376863e-01 1.64527550e-01
-6.18651986e-01 1.00584674e+00 7.79221356e-01 -3.71388167e-01
1.72906250e-01 -7.29560256e-01 -7.06373870e-01 5.16250432e-01
5.40093243e-01 -5.60334802e-01 -4.04469550e-01 -6.67657018e-01
7.30573118e-01 -3.50102186e-01 1.47456959e-01 -4.07444119e-01
6.32335424e-01 -7.27551803e-02 2.08239742e-02 -4.76692259e-01
5.13078012e-02 -1.08336508e+00 5.11078775e-01 9.69834328e-02
-2.39370167e-01 2.23808303e-01 1.73173204e-01 6.87577546e-01
-5.81697404e-01 -2.17844874e-01 2.28968218e-01 3.25383335e-01
-8.75278473e-01 6.58069134e-01 2.93204576e-01 3.78615201e-01
1.03038144e+00 2.24815428e-01 -6.08156085e-01 -4.22289334e-02
-6.50007069e-01 5.33998489e-01 4.85064477e-01 6.45925164e-01
4.40434098e-01 -2.11511540e+00 -6.02824330e-01 8.72490183e-02
6.18863225e-01 -5.77126024e-03 5.89982688e-01 8.84076536e-01
-1.50799662e-01 1.84795842e-01 -9.45280716e-02 -4.39936966e-02
-1.27212977e+00 7.71169364e-01 3.86891961e-01 -6.98324978e-01
-2.70622164e-01 8.57493579e-01 2.49917597e-01 -8.90012085e-01
-1.02254488e-01 3.53039079e-03 -2.40260705e-01 2.94946730e-01
3.81623298e-01 2.45149583e-01 -1.50723569e-02 -8.94617498e-01
-6.18662298e-01 5.74709773e-01 -4.30410266e-01 4.06889528e-01
1.32103908e+00 -2.05651835e-01 -5.08271635e-01 4.80354726e-01
1.63542998e+00 3.24232250e-01 -4.32924867e-01 -7.13301778e-01
3.26230675e-01 -3.90390396e-01 -4.02819008e-01 -1.44583240e-01
-1.04390049e+00 6.18345857e-01 -3.13357785e-02 1.32681876e-01
7.23540664e-01 3.15727025e-01 7.83884525e-01 3.71120542e-01
1.74577683e-01 -8.72671664e-01 -1.85659811e-01 5.41257083e-01
5.79635739e-01 -1.01861799e+00 -1.91119149e-01 -6.59390509e-01
-7.35835195e-01 8.90019536e-01 7.46784866e-01 1.51278645e-01
6.81311488e-01 3.69717553e-02 -1.92495957e-01 -4.18913424e-01
-5.96200168e-01 -2.65411884e-01 6.58108890e-01 4.29162115e-01
3.40986937e-01 3.55545320e-02 -4.20059204e-01 1.05670679e+00
-5.87223582e-02 -3.00986320e-01 1.08415455e-01 7.69241095e-01
1.48569614e-01 -1.30527103e+00 2.94358373e-01 2.31092572e-01
-1.49469227e-01 -5.83225906e-01 -4.33357060e-01 8.08895528e-01
3.20955813e-01 5.31224728e-01 -7.42004439e-02 -6.77678883e-01
4.87173289e-01 2.41423056e-01 1.73982337e-01 -4.10454541e-01
-2.12378055e-01 -6.54487491e-01 2.30066776e-01 -3.32287580e-01
-1.71619847e-01 -4.34099555e-01 -1.24298251e+00 -5.90231001e-01
-5.36110759e-01 3.94482642e-01 1.25446305e-01 8.45965087e-01
8.11069608e-01 4.62466627e-01 6.61929965e-01 -1.47277951e-01
-3.15309078e-01 -7.20992684e-01 -8.32517087e-01 8.58716667e-01
-7.49406070e-02 -6.61592662e-01 -2.75782913e-01 -5.92939323e-03] | [8.746574401855469, 7.962504863739014] |
7952ec54-eafd-4187-83f5-84afd90e7e79 | hierarchically-refined-label-attention | 1908.08676 | null | https://arxiv.org/abs/1908.08676v3 | https://arxiv.org/pdf/1908.08676v3.pdf | Hierarchically-Refined Label Attention Network for Sequence Labeling | CRF has been used as a powerful model for statistical sequence labeling. For neural sequence labeling, however, BiLSTM-CRF does not always lead to better results compared with BiLSTM-softmax local classification. This can be because the simple Markov label transition model of CRF does not give much information gain over strong neural encoding. For better representing label sequences, we investigate a hierarchically-refined label attention network, which explicitly leverages label embeddings and captures potential long-term label dependency by giving each word incrementally refined label distributions with hierarchical attention. Results on POS tagging, NER and CCG supertagging show that the proposed model not only improves the overall tagging accuracy with similar number of parameters, but also significantly speeds up the training and testing compared to BiLSTM-CRF. | ['Yue Zhang', 'Leyang Cui'] | 2019-08-23 | hierarchically-refined-label-attention-1 | https://aclanthology.org/D19-1422 | https://aclanthology.org/D19-1422.pdf | ijcnlp-2019-11 | ['ccg-supertagging'] | ['natural-language-processing'] | [ 1.31978944e-01 3.35695744e-01 -4.49423760e-01 -6.71307385e-01
-6.87205255e-01 -6.23418093e-01 3.55719626e-01 2.82417685e-01
-6.82238281e-01 9.97585177e-01 4.24882203e-01 -4.23064351e-01
6.98022306e-01 -4.84030157e-01 -3.66966546e-01 -7.71755338e-01
-7.11196510e-04 4.70779717e-01 2.52795577e-01 9.35301483e-02
-8.21164846e-02 2.01520383e-01 -9.13974166e-01 1.87096968e-02
6.44914091e-01 7.79684007e-01 3.40487361e-01 5.51956475e-01
-4.11839962e-01 1.08326304e+00 -3.50338191e-01 -4.62150097e-01
-3.00148845e-01 -3.61721218e-01 -1.16744769e+00 -2.20280960e-01
3.62876505e-01 -1.64697275e-01 8.08043685e-03 1.07037985e+00
4.19111103e-01 2.68527359e-01 6.03116572e-01 -6.72042370e-01
-8.30299854e-01 9.47431922e-01 -4.59485471e-01 4.81571779e-02
4.26622108e-02 -1.34230852e-01 1.33041894e+00 -5.98973870e-01
4.55767810e-01 1.26179016e+00 1.03893638e+00 7.66665161e-01
-1.20926452e+00 -5.93883395e-01 3.10369402e-01 -8.96283761e-02
-1.22288859e+00 1.95100456e-01 2.89611489e-01 -2.48361617e-01
1.28289294e+00 -2.16430083e-01 3.09277236e-01 1.12089837e+00
3.18050474e-01 6.60818338e-01 1.08190954e+00 -5.55950165e-01
-1.40792895e-02 -1.04093775e-01 7.44899333e-01 8.32396567e-01
-1.50990589e-02 6.06312007e-02 -1.78664714e-01 3.30137499e-02
5.59148133e-01 -5.56807406e-03 1.56011641e-01 2.61695266e-01
-9.84521329e-01 9.69965994e-01 4.97985214e-01 7.20796287e-01
-4.10621971e-01 6.45159543e-01 5.27306497e-01 5.70716290e-03
6.98940873e-01 4.07955974e-01 -8.33565176e-01 -1.89757481e-01
-8.35973680e-01 -3.81538868e-01 5.22770464e-01 8.90867174e-01
1.00134313e+00 2.54925638e-01 -5.31965017e-01 8.97470415e-01
4.33720112e-01 3.95134866e-01 5.44126868e-01 -8.10839415e-01
1.48261040e-01 2.64593035e-01 -4.42900658e-02 -3.50192934e-01
-7.40552783e-01 -6.66856647e-01 -7.06969440e-01 -2.29958043e-01
2.13987932e-01 -3.76754373e-01 -1.38822937e+00 2.02605605e+00
-1.13381129e-02 4.78340060e-01 -1.84058938e-02 5.41892350e-01
6.73953354e-01 7.38169134e-01 9.57765818e-01 -1.19301334e-01
1.67909491e+00 -1.30576873e+00 -1.01076543e+00 -3.70928288e-01
1.34520996e+00 -5.33988833e-01 9.39717710e-01 -1.15018249e-01
-6.77462637e-01 -6.65999889e-01 -7.45344102e-01 -3.39475751e-01
-3.83217514e-01 1.87717378e-02 8.07184994e-01 8.19622636e-01
-1.30691624e+00 4.96702522e-01 -1.06723332e+00 -4.95266110e-01
3.25830251e-01 3.86047065e-01 -2.99090356e-01 -2.43696705e-01
-1.42796016e+00 1.24821651e+00 8.66628230e-01 8.76614228e-02
-7.33470380e-01 -5.01618266e-01 -1.08690882e+00 2.41130292e-01
-6.11728504e-02 -3.40353191e-01 1.51129127e+00 -5.51203310e-01
-1.62546444e+00 7.19896197e-01 -3.06699455e-01 -5.85923612e-01
-3.49569440e-01 -3.10049087e-01 -2.77098149e-01 -1.74071476e-01
-1.60712413e-02 1.25865030e+00 1.32535905e-01 -9.69115555e-01
-6.51956439e-01 6.58598766e-02 -1.52037501e-01 1.37973890e-01
-3.53779048e-01 2.44714782e-01 -4.45794798e-02 -4.68706876e-01
-2.79384941e-01 -9.94925082e-01 -6.12787604e-01 -9.32453036e-01
-2.56439000e-01 -6.64239168e-01 4.95979309e-01 -8.54086637e-01
1.19906783e+00 -1.83200204e+00 -5.10739028e-01 -1.64764360e-01
-2.53364772e-01 3.98570269e-01 -4.59169865e-01 3.61200392e-01
-3.83029766e-02 4.42652881e-01 -1.77423149e-01 -6.99930310e-01
1.17420234e-01 6.23316467e-01 -1.52986914e-01 1.80762216e-01
3.81962985e-01 1.28213429e+00 -1.04787982e+00 -6.68118298e-01
5.80092147e-02 5.04881322e-01 -3.77818018e-01 2.35346228e-01
-3.76029015e-01 6.54024005e-01 -2.73052961e-01 4.65833515e-01
5.71829796e-01 -4.87832189e-01 5.04893005e-01 -1.72250252e-02
3.28397602e-02 6.73800051e-01 -3.90383393e-01 1.89144957e+00
-6.29760563e-01 2.94065773e-01 -4.82805461e-01 -9.61181223e-01
9.16297317e-01 6.51282609e-01 1.97329089e-01 -7.26636052e-01
2.42382556e-01 3.49881388e-02 -1.01105765e-01 -2.54004419e-01
5.46862602e-01 -6.47588372e-01 -4.22890186e-01 4.17891026e-01
4.78405476e-01 4.39970851e-01 2.22749040e-02 -8.66634399e-03
1.17414665e+00 4.29241776e-01 3.55850101e-01 -3.18902642e-01
2.57129401e-01 -1.15535602e-01 7.86470056e-01 6.98579371e-01
-1.16935067e-01 3.45612198e-01 2.09400520e-01 -2.79566288e-01
-7.65505254e-01 -8.14537525e-01 -1.10855922e-01 1.68078315e+00
-9.37508047e-02 -1.47282749e-01 -8.05353343e-01 -1.16796136e+00
-3.61841202e-01 8.31190526e-01 -5.48479021e-01 -3.68083431e-03
-7.07143724e-01 -8.49687159e-01 9.03497577e-01 8.44255209e-01
3.00343633e-01 -1.56282377e+00 -1.70498893e-01 5.23599565e-01
-3.27634901e-01 -1.28024805e+00 -6.79374993e-01 9.42347229e-01
-1.11001134e+00 -3.84200811e-01 -5.69401145e-01 -1.28868079e+00
5.41555524e-01 -2.49436155e-01 1.24229729e+00 1.96957439e-01
3.72579277e-01 -1.57402188e-01 -6.36275172e-01 8.51867497e-02
-4.90560204e-01 6.52816355e-01 -1.76951721e-01 -3.59525532e-01
6.10727847e-01 -4.78829205e-01 -2.13097602e-01 5.49468882e-02
-7.13122427e-01 -1.02997631e-01 7.03135490e-01 1.01256669e+00
2.79671192e-01 -2.78048605e-01 9.64048743e-01 -1.24386966e+00
2.40861967e-01 -3.52217466e-01 -2.56838053e-01 3.31212759e-01
-7.61506617e-01 4.98228222e-01 5.16986430e-01 -3.86306345e-01
-1.28173232e+00 1.45389989e-01 -8.74375880e-01 5.70544079e-02
-4.64161128e-01 7.62480915e-01 2.16530804e-02 1.57970205e-01
2.56004542e-01 -5.75181097e-03 -4.14613813e-01 -8.67161870e-01
5.53401053e-01 6.23305440e-01 4.93500501e-01 -3.81186515e-01
3.83448720e-01 -6.46079779e-02 -4.91697863e-02 -3.86821985e-01
-1.36194277e+00 -5.73562503e-01 -9.14469600e-01 2.53486753e-01
1.36039340e+00 -9.21481609e-01 -7.67521501e-01 4.30015892e-01
-1.27160966e+00 -7.69925952e-01 -1.29645154e-01 5.63958108e-01
-2.69564837e-01 4.39330965e-01 -1.39167535e+00 -7.73454547e-01
-3.25736701e-01 -8.46240699e-01 8.91906559e-01 2.53697842e-01
-2.08994672e-01 -1.49772513e+00 4.22910154e-01 3.84338945e-02
4.52884525e-01 1.28198624e-01 9.46366906e-01 -8.32948327e-01
-1.04555659e-01 -1.92792386e-01 -2.68090844e-01 4.94639963e-01
8.76300409e-02 -4.50035542e-01 -1.20056510e+00 -3.60238969e-01
-5.51562190e-01 -5.60852230e-01 9.63819623e-01 4.12035137e-01
5.61558425e-01 -1.91129386e-01 -3.92570525e-01 3.03097785e-01
1.49278474e+00 2.54081666e-01 6.46654248e-01 1.78499341e-01
1.07088625e+00 4.41458493e-01 6.26700997e-01 -2.38054171e-02
5.62397897e-01 5.72021306e-01 2.91172117e-01 -4.15454656e-01
-1.92420721e-01 -4.45742428e-01 5.45224428e-01 1.20485759e+00
2.16208220e-01 -4.44477767e-01 -8.77793849e-01 5.51753342e-01
-1.70652890e+00 -5.42258501e-01 -2.37321004e-01 1.79622495e+00
1.12858641e+00 1.96399298e-02 3.27784643e-02 -2.82361329e-01
9.05096412e-01 7.18325600e-02 -1.05142571e-01 -6.90241516e-01
6.83933794e-02 5.17288864e-01 8.40841174e-01 7.16647983e-01
-1.12962854e+00 1.56131768e+00 6.73490953e+00 8.82375658e-01
-1.01700664e+00 7.89599478e-01 5.80249667e-01 6.08141840e-01
-2.01704815e-01 1.57455653e-01 -1.13525939e+00 4.55177218e-01
1.68887115e+00 7.69633591e-01 -4.58914004e-02 6.57305181e-01
-8.56966227e-02 7.24071115e-02 -8.45435441e-01 3.91710073e-01
-2.84037888e-01 -9.94977593e-01 -1.67047083e-01 2.76254535e-01
8.40881348e-01 4.44710314e-01 -2.32673839e-01 7.25633025e-01
1.20404387e+00 -1.11165524e+00 6.79374695e-01 2.50074744e-01
9.39146459e-01 -8.68541896e-01 1.39898324e+00 3.55966777e-01
-1.20673406e+00 1.28442973e-01 -5.46773374e-01 -2.14931637e-01
6.21786833e-01 5.75859666e-01 -1.06265938e+00 4.51789707e-01
3.17091495e-01 6.27383828e-01 -6.30378187e-01 7.96343982e-01
-6.38620913e-01 1.25240982e+00 -1.34964928e-01 -9.84316841e-02
6.05775654e-01 -1.87257994e-02 -2.62033373e-01 1.89939797e+00
2.64586091e-01 -7.19772950e-02 4.74310666e-01 3.10952604e-01
-2.17773199e-01 3.79862711e-02 -3.18892270e-01 -2.02864006e-01
4.96136189e-01 1.37261391e+00 -9.71742988e-01 -4.48088616e-01
-6.04402125e-01 1.05352652e+00 8.24157298e-01 3.58114392e-01
-8.86542976e-01 -3.99710648e-02 4.03485179e-01 -4.01248544e-01
5.65671623e-01 -3.14878017e-01 -3.03709656e-01 -8.08175087e-01
-7.91282535e-01 -2.60548204e-01 4.57544088e-01 -5.01603663e-01
-1.12208378e+00 7.68116295e-01 -4.38777149e-01 -6.46567822e-01
-3.49914998e-01 -4.52454001e-01 -3.81450206e-01 9.40190852e-01
-1.75360894e+00 -1.53577757e+00 1.44859195e-01 -6.23149574e-02
2.32104450e-01 4.08011317e-01 1.24522507e+00 4.27162409e-01
-5.91094553e-01 7.17431962e-01 -1.63436513e-02 2.74428338e-01
7.46916473e-01 -1.55816472e+00 6.56299889e-01 8.36082876e-01
2.67126769e-01 4.92612511e-01 4.66700137e-01 -6.63649023e-01
-3.80881518e-01 -1.43715823e+00 1.47470176e+00 -4.69634920e-01
5.19191861e-01 -4.97886211e-01 -1.07503712e+00 1.00737083e+00
5.76017797e-01 1.25936361e-03 8.02554250e-01 3.79511744e-01
-4.97720778e-01 3.60053986e-01 -1.00741029e+00 2.32573286e-01
9.50357080e-01 -6.43664658e-01 -4.68621761e-01 1.08495221e-01
1.19157314e+00 2.49386299e-02 -1.06941438e+00 4.62842494e-01
2.34755635e-01 -3.57993573e-01 6.47603869e-01 -4.27924186e-01
1.80464283e-01 -1.36918217e-01 -3.92468199e-02 -1.25950527e+00
-9.91988719e-01 -1.90579757e-01 1.46165580e-01 1.61911905e+00
7.44902074e-01 -6.73458517e-01 8.54076862e-01 2.58799553e-01
-4.89558965e-01 -4.89575028e-01 -6.92686319e-01 -8.92253995e-01
3.10866982e-01 -3.13895077e-01 4.90896851e-01 1.17297542e+00
4.90351953e-02 6.89564347e-01 -6.78602159e-01 1.18122555e-01
3.67842674e-01 -3.04277211e-01 -2.69556954e-03 -1.23566735e+00
-1.32542953e-01 -2.09181026e-01 -1.69014931e-01 -1.25565183e+00
6.99573576e-01 -1.18461180e+00 6.78605974e-01 -1.74847877e+00
3.66728604e-01 -6.63550735e-01 -8.95355880e-01 1.15973198e+00
-4.94028717e-01 8.70971978e-01 -1.00696906e-01 -4.66192141e-02
-9.29553092e-01 3.39548916e-01 1.07412696e+00 2.07405776e-01
1.60995930e-01 -3.55727643e-01 -6.30612671e-01 4.48013306e-01
7.93874681e-01 -9.52805817e-01 -1.73122361e-02 -3.61593962e-01
2.34235808e-01 4.03647684e-02 -2.57525772e-01 -8.53016198e-01
-2.13692314e-03 3.33622470e-02 6.78232312e-02 -4.84620512e-01
-1.63971223e-02 -6.17024422e-01 -5.83917052e-02 4.60087985e-01
-5.64495742e-01 -3.33870798e-02 7.31123909e-02 5.26108742e-01
-2.63709813e-01 -7.14662433e-01 9.23242211e-01 5.69119072e-03
-7.65495121e-01 2.29229614e-01 -7.92951822e-01 -1.09730355e-01
4.41331953e-01 -1.98428966e-02 -1.47344977e-01 -1.47450566e-01
-8.02897811e-01 3.37949067e-01 3.40165645e-01 2.74715632e-01
-1.65836647e-01 -1.41359591e+00 -5.50157964e-01 -1.61654502e-01
-3.06650791e-02 -1.22188546e-01 1.70066997e-01 8.12461972e-01
-3.38826299e-01 7.87684798e-01 -3.78996246e-02 -4.43595111e-01
-1.05885708e+00 5.67390323e-01 6.08182326e-02 -9.63066697e-01
-4.38633114e-01 1.03005648e+00 3.38529199e-01 -8.30500305e-01
2.11582720e-01 -5.92523217e-01 -6.13373518e-01 -3.58040445e-02
2.97772914e-01 -5.16215980e-04 -1.18652144e-02 -7.86415577e-01
-3.67686629e-01 5.31696796e-01 -1.88419849e-01 -3.68398167e-02
1.20129740e+00 -3.62344116e-01 -4.20046262e-02 5.64672470e-01
1.29281890e+00 -2.61381924e-01 -1.35138094e+00 -3.69909227e-01
7.14795053e-01 3.09862345e-01 9.69245136e-02 -9.77396667e-01
-8.89979362e-01 1.08292902e+00 4.38492507e-01 1.08141191e-01
7.71075010e-01 -3.75749581e-02 1.19748533e+00 1.71263739e-01
4.36250448e-01 -7.65422344e-01 -2.82978062e-02 1.13209891e+00
2.97754910e-02 -1.15594149e+00 -5.83529353e-01 -2.85928130e-01
-7.81776607e-01 7.97021627e-01 5.61060369e-01 -5.84119968e-02
4.95581895e-01 5.34015536e-01 3.45003963e-01 -5.33743203e-02
-6.55214131e-01 -4.95852143e-01 -1.06364489e-03 6.55117810e-01
9.78514731e-01 1.83721274e-01 -3.53935719e-01 4.70967412e-01
9.13998112e-02 9.41708870e-03 1.99055672e-01 7.03236103e-01
-7.17613459e-01 -1.51416600e+00 2.14580614e-02 2.36787945e-01
-9.12408471e-01 -5.61338305e-01 3.39844190e-02 4.49320078e-01
1.95223093e-01 9.38920856e-01 1.44028559e-01 -4.19788957e-01
-2.07756355e-01 6.99555635e-01 3.78410429e-01 -1.04569912e+00
-8.44372869e-01 1.67478502e-01 3.35079789e-01 -1.44034907e-01
-5.40564537e-01 -4.38068479e-01 -1.75233150e+00 4.03583050e-04
-6.83362842e-01 5.39103389e-01 5.22144973e-01 1.24704647e+00
1.10693932e-01 6.29592597e-01 3.28584045e-01 -7.52112746e-01
-3.76219809e-01 -1.59973943e+00 -5.35835087e-01 2.98107475e-01
1.58248067e-01 -5.10696530e-01 -9.13597941e-02 1.42132312e-01] | [10.010204315185547, 9.743517875671387] |
8633667e-dcae-4ac4-87ac-652e08c40a53 | multi-label-zero-shot-learning-with | 1711.06526 | null | http://arxiv.org/abs/1711.06526v2 | http://arxiv.org/pdf/1711.06526v2.pdf | Multi-Label Zero-Shot Learning with Structured Knowledge Graphs | In this paper, we propose a novel deep learning architecture for multi-label
zero-shot learning (ML-ZSL), which is able to predict multiple unseen class
labels for each input instance. Inspired by the way humans utilize semantic
knowledge between objects of interests, we propose a framework that
incorporates knowledge graphs for describing the relationships between multiple
labels. Our model learns an information propagation mechanism from the semantic
label space, which can be applied to model the interdependencies between seen
and unseen class labels. With such investigation of structured knowledge graphs
for visual reasoning, we show that our model can be applied for solving
multi-label classification and ML-ZSL tasks. Compared to state-of-the-art
approaches, comparable or improved performances can be achieved by our method. | ['Yu-Chiang Frank Wang', 'Chih-Kuan Yeh', 'Chung-Wei Lee', 'Wei Fang'] | 2017-11-17 | multi-label-zero-shot-learning-with-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Lee_Multi-Label_Zero-Shot_Learning_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Lee_Multi-Label_Zero-Shot_Learning_CVPR_2018_paper.pdf | cvpr-2018-6 | ['multi-label-zero-shot-learning'] | ['computer-vision'] | [ 3.56486470e-01 3.80672872e-01 -3.22585911e-01 -5.99585354e-01
-2.60524631e-01 -3.99299145e-01 6.88132226e-01 6.09099925e-01
-1.45272702e-01 3.53754967e-01 3.68595943e-02 -8.64618644e-02
-3.41853827e-01 -1.13339293e+00 -5.53519130e-01 -3.48200470e-01
3.15837920e-01 7.08071351e-01 5.11412621e-01 7.04054758e-02
8.10047090e-02 3.36588264e-01 -1.83851910e+00 8.26882303e-01
2.12595627e-01 9.32226658e-01 -1.17085353e-01 3.32850993e-01
-5.44603944e-01 1.51980591e+00 -2.26180226e-01 -6.42404079e-01
-2.00384736e-01 -3.48826855e-01 -1.42342150e+00 2.63972282e-01
5.99466145e-01 -1.02820866e-01 -1.54502705e-01 1.07900667e+00
1.04729876e-01 3.91280979e-01 1.16547894e+00 -1.57328296e+00
-9.47024405e-01 4.58160013e-01 -3.21772546e-01 -1.79050416e-01
3.42831850e-01 -2.52877027e-01 1.28856063e+00 -7.79518008e-01
1.01683640e+00 1.50481594e+00 5.42474747e-01 7.18204200e-01
-1.16159821e+00 -2.97522575e-01 2.96413481e-01 9.13009167e-01
-1.16740966e+00 -5.65725006e-02 7.07685053e-01 -7.07725942e-01
7.18549192e-01 -2.72479840e-02 5.06819785e-01 1.18820739e+00
-6.03713617e-02 9.43439841e-01 1.05536902e+00 -8.65828574e-01
3.53135407e-01 2.71955162e-01 6.92779958e-01 1.08728743e+00
9.14705396e-02 -6.98659569e-02 -6.32364035e-01 -2.03858782e-02
4.75639671e-01 2.91849434e-01 -6.60523772e-03 -9.03419852e-01
-1.02175081e+00 9.54384208e-01 7.33335257e-01 4.60168719e-01
4.41421662e-03 3.65700036e-01 4.52947140e-01 1.64982408e-01
4.92767870e-01 3.57414782e-01 -2.34368145e-01 4.61907685e-01
-5.73253870e-01 -1.28627107e-01 7.22492754e-01 1.05738449e+00
1.06028509e+00 -3.52608442e-01 -4.93304700e-01 9.03209686e-01
5.85903466e-01 4.82870750e-02 2.54245996e-01 -1.03375566e+00
-8.45400989e-02 9.60642695e-01 4.19978751e-03 -8.32219779e-01
-7.28805780e-01 -7.87881017e-02 -6.04626238e-01 3.52823645e-01
1.87709317e-01 3.39631110e-01 -1.09922504e+00 1.63876903e+00
3.16576600e-01 5.67009866e-01 1.85901612e-01 8.01000416e-01
1.26627052e+00 3.25639665e-01 6.05457067e-01 -1.28274247e-01
1.48581326e+00 -1.17748702e+00 -8.01285684e-01 -3.43197316e-01
8.50688934e-01 -1.10905267e-01 8.54423761e-01 -6.44495040e-02
-7.15590537e-01 -4.90358591e-01 -9.37198341e-01 -3.76886308e-01
-1.00827050e+00 -3.19004297e-01 6.65617108e-01 2.55396515e-01
-1.12922692e+00 6.97112620e-01 -5.04754841e-01 -7.51527250e-01
6.89220428e-01 2.19331421e-02 -2.96384156e-01 -4.07496929e-01
-1.28100133e+00 1.22545218e+00 7.11402476e-01 -2.60070950e-01
-1.06267071e+00 -6.04279459e-01 -1.15784085e+00 4.18360025e-01
6.99806750e-01 -7.93914318e-01 1.28686440e+00 -8.01018417e-01
-9.88930106e-01 1.41379869e+00 -1.91063449e-01 -2.00353399e-01
9.35501754e-02 1.79346442e-01 -4.73869383e-01 2.51645774e-01
1.71244144e-01 7.08983958e-01 6.14286959e-01 -1.80225778e+00
-6.34712160e-01 -4.66589659e-01 3.17922473e-01 1.31111532e-01
-3.49497736e-01 -4.73363623e-02 -1.10250376e-01 -2.55823523e-01
2.76701078e-02 -6.39860153e-01 -1.88654661e-01 1.33384019e-01
-3.51247609e-01 -6.23498321e-01 8.71590376e-01 -9.55099389e-02
8.74893367e-01 -1.92392099e+00 3.42127502e-01 8.11244026e-02
5.34229755e-01 1.19207516e-01 -4.09861088e-01 5.24023890e-01
-4.81630750e-02 9.66401175e-02 -1.17472142e-01 -2.43518338e-01
3.30196112e-01 4.99412298e-01 -2.44025186e-01 1.68774351e-01
4.13624980e-02 1.28448486e+00 -1.31580234e+00 -5.61956704e-01
3.37436885e-01 3.33377331e-01 1.12555802e-01 3.83511633e-01
-4.41833109e-01 2.34764367e-01 -2.24727973e-01 7.16766059e-01
3.77104789e-01 -8.43883753e-01 5.05231321e-01 -2.67148197e-01
4.04107988e-01 -3.35952222e-01 -9.30740237e-01 1.68965530e+00
-5.70987403e-01 4.60562944e-01 -4.46069032e-01 -1.09043491e+00
7.02243328e-01 4.19537723e-01 1.25093997e-01 -6.99802756e-01
2.62987852e-01 -2.29245037e-01 -5.74725747e-01 -7.27405548e-01
-2.71924194e-02 -5.18350303e-01 -7.00091645e-02 5.70509970e-01
7.10773528e-01 -2.14740485e-02 3.05797160e-01 3.22841138e-01
9.25513566e-01 8.58986527e-02 7.87140608e-01 -6.66630864e-02
5.64878225e-01 -1.84824273e-01 3.77918810e-01 7.52079308e-01
-2.88156301e-01 1.16929889e-01 5.54495335e-01 -8.62520814e-01
-7.71606326e-01 -1.25068533e+00 -2.42887624e-02 1.55878639e+00
5.44215500e-01 -3.19236308e-01 -4.64879751e-01 -1.13691473e+00
4.68819886e-02 1.05657876e+00 -9.18648005e-01 -2.38662675e-01
1.27854198e-02 -1.93807498e-01 2.34720170e-01 6.09324455e-01
1.27840444e-01 -1.44289446e+00 -6.06119514e-01 1.10562563e-01
-1.87833030e-02 -1.21667194e+00 2.50983179e-01 2.75531262e-01
-5.35981655e-01 -1.50823903e+00 -2.78406173e-01 -9.54647481e-01
5.75621068e-01 1.71506673e-01 1.34187663e+00 1.89027846e-01
-4.82045323e-01 8.03082705e-01 -4.90123004e-01 -3.45728457e-01
-5.28157532e-01 -3.00513923e-01 -2.94716507e-01 2.54837632e-01
6.00583434e-01 -4.82114166e-01 -3.42708677e-01 1.86779529e-01
-1.00989032e+00 2.00740352e-01 1.09093614e-01 7.51841128e-01
6.01418257e-01 -7.15513900e-02 7.90481865e-01 -1.40637958e+00
3.58284324e-01 -6.26708567e-01 -3.08980674e-01 1.00658619e+00
-6.93063438e-01 3.64446789e-01 5.11591196e-01 -2.65753537e-01
-1.16152263e+00 5.15606403e-02 2.10235208e-01 -6.01218045e-01
-6.11369133e-01 2.18243986e-01 -9.07958001e-02 -2.98087597e-01
4.34079647e-01 -1.84428230e-01 -4.00991499e-01 -3.14950913e-01
1.12631571e+00 5.09493530e-01 2.58607417e-01 -2.86925048e-01
2.64134407e-01 7.79502153e-01 4.36414778e-01 -3.29068899e-01
-1.78196573e+00 -8.54096830e-01 -1.02544236e+00 -5.89182436e-01
1.38050890e+00 -6.78406537e-01 -8.79079163e-01 3.09022456e-01
-1.26886988e+00 -1.83286861e-01 -3.21757674e-01 9.35517251e-03
-8.68222773e-01 3.39971542e-01 -7.48766184e-01 -6.00626469e-01
-8.56035110e-03 -8.70370448e-01 1.07902443e+00 2.95295835e-01
-1.53531075e-01 -1.60709798e+00 2.64058948e-01 4.72903997e-01
1.87282205e-01 1.11609198e-01 1.44496167e+00 -1.00936782e+00
-5.24653435e-01 1.25570670e-01 -6.00862145e-01 2.01964984e-03
-8.20624232e-02 -4.60986763e-01 -1.22425604e+00 -2.29021639e-01
-2.28356823e-01 -8.30218494e-01 9.75085020e-01 -6.49689417e-03
1.14157462e+00 -6.19815700e-02 -6.60105288e-01 2.06573904e-01
1.74171746e+00 1.41699865e-01 4.32880282e-01 3.03398609e-01
9.78220582e-01 7.82212973e-01 5.10452509e-01 3.42134506e-01
6.63846135e-01 5.75947285e-01 6.79003000e-01 -2.34751571e-02
-3.17282498e-01 -2.77029306e-01 -3.72154295e-01 5.02080739e-01
3.48675065e-02 -3.52291435e-01 -1.07028055e+00 5.95084429e-01
-2.30007911e+00 -1.10381532e+00 -1.32502735e-01 1.74110019e+00
4.22500521e-01 -2.41023287e-01 -2.83110261e-01 -1.83381468e-01
8.99679005e-01 2.99498379e-01 -7.14373469e-01 -5.34908295e-01
1.56261250e-01 2.35687301e-01 1.66683510e-01 5.10519445e-01
-1.32405412e+00 1.15133333e+00 6.66890478e+00 7.53877699e-01
-4.67994541e-01 6.15042210e-01 3.43141377e-01 2.28343546e-01
-2.98020840e-01 -4.49432395e-02 -5.60738623e-01 -1.38941899e-01
9.33824062e-01 -1.04276739e-01 3.05302829e-01 8.96450639e-01
-5.61406016e-01 9.11829248e-02 -1.34822285e+00 8.61571670e-01
4.06245440e-01 -1.42586756e+00 3.66070539e-01 -1.79084614e-01
8.12084973e-01 -4.40815926e-01 -3.10135603e-01 5.12805939e-01
7.95925081e-01 -9.22807455e-01 3.86779338e-01 8.05367231e-01
6.59245789e-01 -6.94934309e-01 5.75754404e-01 3.84315431e-01
-1.30954587e+00 -3.99295032e-01 -5.56740701e-01 -6.93649650e-02
1.99617550e-01 4.42930490e-01 -8.79142702e-01 7.31910169e-01
3.93504143e-01 9.66707587e-01 -6.91098571e-01 7.39595354e-01
-7.96816409e-01 1.17825493e-01 4.00064081e-01 3.34681757e-02
2.06812918e-01 7.02343211e-02 1.15408925e-02 1.08822143e+00
8.58520716e-02 8.51797611e-02 3.33227962e-01 8.81553233e-01
-2.10141584e-01 1.13950752e-01 -1.02511454e+00 1.82869032e-01
2.56868213e-01 1.38539410e+00 -9.40025210e-01 -7.01786280e-01
-7.48000026e-01 1.15361106e+00 1.05267787e+00 5.19474089e-01
-7.19546735e-01 -3.93479019e-02 4.25886691e-01 -1.35284200e-01
2.48817325e-01 2.00089559e-01 -6.71200380e-02 -1.15604508e+00
-5.19620538e-01 -1.46630540e-01 8.42948377e-01 -1.09541595e+00
-1.70114601e+00 4.66795504e-01 -9.29139629e-02 -1.06346691e+00
-1.81864545e-01 -7.84286797e-01 -3.51014555e-01 4.39282358e-01
-1.69030797e+00 -1.55319512e+00 -2.53811449e-01 5.66617846e-01
5.70072651e-01 -6.01693913e-02 1.31453490e+00 -9.54865590e-02
-9.45393369e-02 1.66632593e-01 1.51655069e-02 -7.12883705e-03
5.77292502e-01 -1.33634889e+00 2.15329796e-01 4.62015092e-01
7.13248551e-01 1.28195837e-01 2.59619594e-01 -5.17183125e-01
-6.61215425e-01 -1.28749621e+00 1.05544400e+00 -5.79786658e-01
8.36021841e-01 -3.24693799e-01 -1.01117146e+00 9.43381369e-01
3.15099180e-01 2.73566216e-01 1.18514776e+00 2.78877407e-01
-9.50210392e-01 3.28224540e-01 -1.06711352e+00 3.98446113e-01
1.34628272e+00 -8.14663589e-01 -9.53729451e-01 5.74263453e-01
8.61160934e-01 4.48702164e-02 -8.77294242e-01 4.22262251e-01
2.86926746e-01 -9.64754939e-01 1.11659408e+00 -1.03759515e+00
5.52384257e-01 -1.71459481e-01 -1.31654173e-01 -1.60111213e+00
-8.55413914e-01 4.56248790e-01 -4.09174353e-01 9.51030195e-01
2.76026964e-01 -1.61599129e-01 4.60045874e-01 6.20790243e-01
5.67503944e-02 -5.05065024e-01 -8.80395293e-01 -6.06094122e-01
-1.97681785e-01 -2.72327155e-01 4.06820923e-01 1.42121994e+00
2.64780551e-01 7.20274508e-01 -4.52159613e-01 2.11971864e-01
9.56306398e-01 4.30850238e-01 -2.74726115e-02 -1.80706012e+00
-1.27551109e-01 -2.58982062e-01 -7.13357091e-01 -3.23963493e-01
8.91091883e-01 -1.36227989e+00 -1.67920694e-01 -2.21853018e+00
6.11722112e-01 -9.27470699e-02 -7.73801565e-01 1.02210975e+00
-1.88642010e-01 3.30780715e-01 5.68988681e-01 2.80148704e-02
-1.31590974e+00 3.61027151e-01 1.31471801e+00 -3.84665221e-01
3.65353584e-01 -3.64709318e-01 -4.37011421e-01 7.81440258e-01
4.72644448e-01 -5.73336244e-01 -5.38497329e-01 -2.61437178e-01
3.12393099e-01 9.13314670e-02 4.93725926e-01 -8.71678054e-01
3.26591581e-01 -2.30752155e-01 2.02591985e-01 -1.75974324e-01
2.80538619e-01 -1.10297763e+00 4.38186303e-02 2.99491674e-01
-7.61175573e-01 -5.70975542e-01 -8.74944776e-02 8.95379305e-01
-2.26732045e-01 -5.71872532e-01 8.33000600e-01 -4.60516483e-01
-1.42983389e+00 2.44878605e-01 -3.27374399e-01 5.56328408e-02
1.39872444e+00 1.21031500e-01 -6.93082809e-01 -1.29785731e-01
-1.31209910e+00 3.73726487e-01 3.41042638e-01 6.02348685e-01
7.74863958e-01 -1.69389236e+00 -2.72538543e-01 2.76039932e-02
7.41984785e-01 -3.60797554e-01 4.96710479e-01 1.60986707e-01
-9.94048044e-02 3.83649558e-01 -3.30979556e-01 -2.74202436e-01
-1.15902352e+00 1.23397744e+00 1.81202665e-01 -4.33284700e-01
-5.59394240e-01 7.59676337e-01 4.09992874e-01 -5.96002579e-01
2.27245644e-01 1.77791357e-01 -5.85217714e-01 3.70676994e-01
6.26258194e-01 3.98722619e-01 -1.14451125e-01 -6.37319744e-01
-4.02162284e-01 7.91612566e-01 -5.57207651e-02 2.40307540e-01
1.21016872e+00 -1.98568583e-01 -3.32773179e-01 1.08157170e+00
1.39276254e+00 -7.99410462e-01 -9.06251192e-01 -4.39050078e-01
3.75288159e-01 -2.46252283e-01 -1.32796735e-01 -9.32539463e-01
-8.40906024e-01 1.05528486e+00 6.25079453e-01 2.67322838e-01
8.63849401e-01 6.49055481e-01 3.46093446e-01 5.05346477e-01
5.67176700e-01 -9.32436347e-01 3.04481834e-01 4.64143991e-01
4.98949766e-01 -1.58128846e+00 -3.24293911e-01 -6.16245031e-01
-7.53155529e-01 1.31554341e+00 7.64112532e-01 2.21806005e-01
6.68962955e-01 -7.55524710e-02 2.64162362e-01 -8.30048800e-01
-8.61082435e-01 -6.00696027e-01 4.34917659e-01 6.39035881e-01
2.05241337e-01 2.31865168e-01 -9.76680815e-02 1.99425504e-01
6.82996809e-01 1.09555371e-01 4.53662544e-01 8.88720453e-01
-7.10078120e-01 -1.10291207e+00 1.95218157e-03 3.05680692e-01
1.47270700e-02 2.10521910e-02 -3.80442858e-01 5.35094440e-01
2.79281408e-01 9.74018157e-01 8.58457163e-02 -1.70946151e-01
3.58955413e-01 7.58245051e-01 5.49832106e-01 -9.74379838e-01
-2.92987913e-01 -4.96962249e-01 8.44536163e-03 -7.19222426e-01
-8.67295742e-01 -1.06170654e-01 -1.36811137e+00 1.88366711e-01
-6.61170781e-02 -2.15326667e-01 3.48051220e-01 1.15409744e+00
1.18451230e-01 7.18940139e-01 2.56389439e-01 -5.73157549e-01
-1.45197704e-01 -6.83638930e-01 -9.09146726e-01 9.98206437e-01
1.65672190e-02 -9.99743998e-01 -2.66161144e-01 1.14862770e-01] | [10.055002212524414, 2.503145217895508] |
9cf6c0f1-3fd0-448a-8432-21f1c16ebc67 | hard-samples-rectification-for-unsupervised | 2106.07204 | null | https://arxiv.org/abs/2106.07204v1 | https://arxiv.org/pdf/2106.07204v1.pdf | Hard Samples Rectification for Unsupervised Cross-domain Person Re-identification | Person re-identification (re-ID) has received great success with the supervised learning methods. However, the task of unsupervised cross-domain re-ID is still challenging. In this paper, we propose a Hard Samples Rectification (HSR) learning scheme which resolves the weakness of original clustering-based methods being vulnerable to the hard positive and negative samples in the target unlabelled dataset. Our HSR contains two parts, an inter-camera mining method that helps recognize a person under different views (hard positive) and a part-based homogeneity technique that makes the model discriminate different persons but with similar appearance (hard negative). By rectifying those two hard cases, the re-ID model can learn effectively and achieve promising results on two large-scale benchmarks. | ['Shao-Yi Chien', 'Tsai-Shien Chen', 'Man-Yu Lee', 'Chih-Ting Liu'] | 2021-06-14 | null | null | null | null | ['unsupervised-person-re-identification'] | ['computer-vision'] | [ 3.39960814e-01 -9.83737931e-02 -1.49251848e-01 -3.93815517e-01
-6.30039275e-01 -2.80541390e-01 7.23653376e-01 -1.69124767e-01
-3.46219122e-01 6.96951270e-01 2.52925336e-01 3.36170256e-01
-1.48677185e-01 -3.97332460e-01 -3.59689236e-01 -8.58997226e-01
3.60940427e-01 7.39043236e-01 1.69749185e-01 -1.82065219e-02
1.16533719e-01 1.53940305e-01 -1.77616966e+00 4.69754547e-01
8.72651935e-01 7.54177511e-01 -2.29939461e-01 4.00677323e-01
1.21589102e-01 5.48751295e-01 -7.02509224e-01 -6.71826601e-01
5.10118723e-01 -4.39702749e-01 -7.22337484e-01 3.51178497e-01
5.44969440e-01 7.07931668e-02 -3.32744718e-02 1.17295706e+00
7.13882267e-01 1.84673116e-01 9.01179075e-01 -1.49964273e+00
-7.66807556e-01 1.96728811e-01 -1.10139823e+00 1.05062231e-01
3.90780807e-01 -2.75537252e-01 4.45106983e-01 -7.31898904e-01
3.18968028e-01 1.45888305e+00 8.55980873e-01 1.03027034e+00
-1.24724483e+00 -8.50959003e-01 7.04113171e-02 4.63089436e-01
-1.64512300e+00 -3.99550647e-01 7.76630700e-01 -5.61086953e-01
6.98428035e-01 3.37376386e-01 3.78206402e-01 1.16792655e+00
-1.54538840e-01 7.11985171e-01 1.29658961e+00 -4.87877578e-01
9.00341198e-02 5.40215790e-01 2.15383381e-01 1.38334855e-01
3.36166173e-01 7.13112131e-02 -1.28494635e-01 7.22685754e-02
4.64737922e-01 5.45450807e-01 -2.18309984e-01 -3.81762445e-01
-8.68657410e-01 5.02514005e-01 1.46302611e-01 1.71531349e-01
6.77561685e-02 -6.31027222e-01 4.00774062e-01 2.24668741e-01
3.98631603e-01 1.80044889e-01 -3.81406844e-01 2.75420815e-01
-7.77782619e-01 9.45089012e-02 5.69333613e-01 8.83825839e-01
6.12219512e-01 -2.99870640e-01 -2.06993207e-01 1.33831644e+00
1.14955194e-01 5.50893724e-01 9.46951449e-01 -5.30394435e-01
3.79236639e-01 9.13967431e-01 -7.63496431e-03 -1.08738077e+00
-3.79734188e-01 -4.95485365e-01 -1.37087393e+00 1.69816732e-01
4.30075228e-01 -3.26409116e-02 -8.08376372e-01 1.52477348e+00
3.33582520e-01 2.81377345e-01 2.62794584e-01 7.63130188e-01
8.64231408e-01 3.23724210e-01 4.43548337e-02 -1.55705065e-01
1.33947265e+00 -1.24614167e+00 -5.47175288e-01 -4.72943366e-01
4.38248456e-01 -5.08468449e-01 5.04525065e-01 4.45832044e-01
-7.18809366e-01 -9.71681654e-01 -1.01566195e+00 3.26782554e-01
-4.38371032e-01 2.64784932e-01 -3.53380181e-02 8.67298722e-01
-8.72455716e-01 3.06295067e-01 -1.99977998e-02 -6.46278977e-01
4.93723154e-01 5.23613930e-01 -8.12510550e-01 -4.04068083e-01
-7.75230527e-01 6.72601938e-01 4.05957162e-01 -1.91872910e-01
-4.18810338e-01 -5.41955590e-01 -6.89787745e-01 -1.56349093e-01
3.93998563e-01 -5.07538795e-01 8.32766414e-01 -1.44387937e+00
-1.19789529e+00 1.33925688e+00 -2.24610180e-01 -4.04444069e-01
7.46367395e-01 -1.93946180e-03 -9.19066191e-01 -7.28169680e-02
3.75512064e-01 6.05535865e-01 1.32101166e+00 -1.71073997e+00
-9.63295877e-01 -8.78517151e-01 -6.01082146e-01 2.48314321e-01
-3.48847330e-01 -3.33110914e-02 -7.23979354e-01 -8.80619228e-01
-5.12372963e-02 -1.07718813e+00 2.46748719e-02 -4.73607928e-01
-5.20322561e-01 -4.34111565e-01 8.78528118e-01 -7.65359521e-01
1.02134490e+00 -2.20672917e+00 1.90298840e-01 2.23867372e-01
3.04815412e-01 5.81113458e-01 -1.12207562e-01 5.62717132e-02
-5.70207715e-01 -7.38719199e-03 -8.22927430e-02 -5.94167769e-01
-2.42076084e-01 7.00134858e-02 1.62828080e-02 3.88758659e-01
7.32411351e-03 5.95890164e-01 -5.63533366e-01 -6.31523371e-01
1.89457387e-01 1.85604736e-01 -1.44496918e-01 3.46162766e-01
5.09066463e-01 4.41022575e-01 6.59948587e-02 7.20124125e-01
9.90290940e-01 -1.24633297e-01 4.04151618e-01 -5.09039581e-01
1.04738139e-01 -6.03792310e-01 -1.48936486e+00 9.64488685e-01
1.16666570e-01 1.06952094e-01 5.81678227e-02 -1.28638828e+00
9.63940144e-01 4.42667492e-02 4.83171999e-01 -9.42586243e-01
2.87505966e-02 -3.36551256e-02 -2.24366486e-01 -3.96914333e-01
3.09659749e-01 -2.11408034e-01 2.69077402e-02 4.21165794e-01
-1.42567739e-01 6.79314613e-01 6.07587509e-02 8.13268423e-02
6.41241014e-01 4.05146256e-02 3.23523343e-01 -1.98690534e-01
9.05414462e-01 -1.42433122e-01 9.12001789e-01 7.12439537e-01
-6.71782792e-01 9.11228836e-01 -9.08473432e-02 -4.59171206e-01
-1.02704906e+00 -8.76468062e-01 1.84472315e-02 9.75424528e-01
5.68973124e-01 -2.56144196e-01 -8.42273593e-01 -1.09958792e+00
1.50906220e-01 1.67966992e-01 -6.37027264e-01 -3.17963034e-01
-4.56340879e-01 -1.10826302e+00 4.58638549e-01 7.64013410e-01
8.13386738e-01 -7.38046706e-01 3.55624668e-02 -2.51382500e-01
-5.23779809e-01 -1.09132195e+00 -6.54913306e-01 9.69452709e-02
-5.36169529e-01 -1.30524898e+00 -1.08343542e+00 -1.15893388e+00
8.97281826e-01 8.58871758e-01 9.41281915e-01 5.39282635e-02
-3.16551656e-01 4.67171013e-01 -3.99413437e-01 -3.11647743e-01
-2.55154341e-01 -2.25668639e-01 4.82131064e-01 6.26446128e-01
8.68272841e-01 -2.15390399e-01 -4.35947031e-01 7.70969689e-01
-5.34452617e-01 -3.13818492e-02 6.75483644e-01 8.53902578e-01
7.21015871e-01 6.11035466e-01 7.78539538e-01 -9.01453376e-01
2.87781835e-01 -4.38627690e-01 -4.57850397e-02 5.30423701e-01
-9.73163128e-01 -1.14390641e-01 5.69336414e-01 -7.23246753e-01
-1.24575460e+00 1.91742942e-01 1.69225529e-01 -2.87851840e-01
-5.60081363e-01 -2.65092254e-01 -5.10174394e-01 -6.24016335e-04
3.98908317e-01 3.52287412e-01 8.88498724e-02 -7.11089075e-01
-1.22244649e-01 1.02004218e+00 7.38545120e-01 -2.59858876e-01
1.00604713e+00 6.86438560e-01 -3.67304206e-01 -8.12677085e-01
-8.40778470e-01 -1.01615250e+00 -9.18220520e-01 -1.44618064e-01
1.02747476e+00 -1.26110160e+00 -4.61669445e-01 7.45549202e-01
-6.65681183e-01 1.77159514e-02 -3.14242989e-01 2.19832793e-01
-1.31145686e-01 7.44079292e-01 -1.68626934e-01 -8.41706097e-01
-3.40825230e-01 -7.35367000e-01 9.97019231e-01 6.62625611e-01
-1.87232241e-01 -7.50773907e-01 1.61669686e-01 8.14081013e-01
-2.09053652e-03 3.58771868e-02 6.18185461e-01 -9.52518284e-01
1.84293482e-02 -4.93130624e-01 -4.34314251e-01 5.64836919e-01
3.16255808e-01 -4.14501876e-01 -1.07439208e+00 -5.13232112e-01
-2.43146360e-01 -1.46595031e-01 9.31411922e-01 1.25197336e-01
1.02534902e+00 -2.38283455e-01 -6.97419465e-01 6.00827754e-01
1.33707964e+00 2.05081195e-01 6.76823854e-01 6.17072940e-01
8.93564641e-01 8.28877866e-01 5.14924049e-01 4.38561171e-01
6.46180332e-01 9.12138343e-01 1.05273873e-01 -5.28634369e-01
-2.33164132e-01 -1.72322646e-01 3.57072979e-01 3.29148382e-01
-3.10194701e-01 -5.66920191e-02 -6.43071473e-01 5.26194394e-01
-2.13021588e+00 -1.29043460e+00 -1.38519213e-01 2.36476254e+00
6.18680596e-01 -1.34648159e-01 6.42082632e-01 4.28002983e-01
1.33841002e+00 -2.36689761e-01 -5.56949914e-01 6.53100014e-02
-4.56555873e-01 -2.46700332e-01 5.59346318e-01 3.08372127e-03
-1.45937538e+00 5.43998063e-01 5.81185436e+00 9.37742472e-01
-5.43725610e-01 1.27810434e-01 7.17954695e-01 2.89098948e-01
3.63720477e-01 -3.76439154e-01 -1.29518712e+00 6.69886649e-01
4.12228405e-01 2.31810302e-01 3.13744485e-01 8.68773222e-01
-1.15689814e-01 -1.88524704e-02 -1.02372706e+00 1.72938895e+00
7.32741058e-01 -7.95148730e-01 1.46938995e-01 -2.98527582e-03
8.46413910e-01 -5.30641437e-01 -4.77535762e-02 4.50704575e-01
2.10715130e-01 -9.65681016e-01 4.79916751e-01 5.26628613e-01
8.20001721e-01 -8.38720262e-01 1.00375175e+00 5.40012062e-01
-1.32795954e+00 -6.52912438e-01 -3.22663128e-01 1.39919773e-01
-6.91937581e-02 3.67326736e-01 -5.25464177e-01 7.92387724e-01
1.28310990e+00 1.02472174e+00 -1.01918054e+00 9.40356553e-01
1.95839167e-01 2.24274069e-01 6.03426918e-02 6.04263365e-01
-4.74975377e-01 -2.52560616e-01 3.03169280e-01 1.23645484e+00
2.63118464e-02 -1.27254009e-01 3.40695709e-01 3.82034719e-01
7.58980364e-02 -1.26328096e-01 -2.61328757e-01 4.54796851e-01
1.56829312e-01 1.26254880e+00 -5.89152038e-01 -4.34290379e-01
-4.92845148e-01 1.48710287e+00 3.30926001e-01 2.58990079e-01
-6.86994970e-01 -8.25094134e-02 5.60337961e-01 3.37487280e-01
2.57936746e-01 1.64268106e-01 -2.42482111e-01 -1.29734170e+00
1.60446465e-02 -1.06533027e+00 9.53530729e-01 -4.59138632e-01
-1.95856690e+00 4.95940000e-01 -5.45034036e-02 -1.46356356e+00
-4.39197347e-02 -5.64730823e-01 -4.56765383e-01 6.48173690e-01
-1.58846533e+00 -1.44081247e+00 -5.99963665e-01 9.51108873e-01
6.11350596e-01 -7.08550096e-01 5.74995995e-01 5.94097793e-01
-1.10506415e+00 1.09312499e+00 2.77519464e-01 4.33575422e-01
1.30082405e+00 -1.28271031e+00 -5.47940247e-02 8.39272499e-01
-2.16441691e-01 4.92133617e-01 2.82309353e-01 -8.34099233e-01
-9.77474332e-01 -1.37965298e+00 7.65309215e-01 -4.94729221e-01
-5.17816544e-02 -1.99594632e-01 -7.85986900e-01 5.67961693e-01
-1.43816024e-01 -1.50808156e-01 9.63827789e-01 1.11777121e-02
-5.59835374e-01 -4.21642989e-01 -1.44145691e+00 2.78237909e-01
9.45956767e-01 -2.51282364e-01 -6.53932512e-01 2.34822348e-01
4.24749516e-02 1.46035299e-01 -6.66309953e-01 4.14388210e-01
5.84499836e-01 -1.01213264e+00 1.42309177e+00 -4.93302613e-01
5.89846494e-03 -5.12578547e-01 -4.20557745e-02 -1.13563168e+00
-6.36052132e-01 -1.87082678e-01 -2.81973258e-02 1.80683899e+00
-1.01164140e-01 -5.78065634e-01 7.62258887e-01 6.24974549e-01
3.11087340e-01 -5.60910031e-02 -6.86516225e-01 -1.02237833e+00
-1.16509266e-01 1.67298153e-01 5.48900664e-01 1.17330420e+00
-1.48589030e-01 4.44634378e-01 -8.09656620e-01 1.96312830e-01
9.93089318e-01 4.14065458e-02 9.25440192e-01 -1.63435423e+00
-2.49870613e-01 -2.63265163e-01 -4.56054479e-01 -8.31178010e-01
1.20962925e-01 -8.25323641e-01 -3.72890383e-02 -1.25645399e+00
9.15179253e-01 -5.05544782e-01 -1.74787983e-01 4.39350963e-01
-3.57686192e-01 6.10338867e-01 1.71817273e-01 6.95674121e-01
-9.79123414e-01 3.51005644e-01 6.98914766e-01 -4.38958973e-01
-2.60255814e-01 1.79799989e-01 -1.00425434e+00 7.91567266e-01
7.39614248e-01 -4.62966532e-01 -1.42785445e-01 -1.27887949e-01
-3.23972762e-01 -5.18349469e-01 6.08926356e-01 -1.30797207e+00
2.94761837e-01 1.08297482e-01 9.59301293e-01 -6.03969157e-01
1.29307583e-01 -9.02550042e-01 2.49945179e-01 4.39875960e-01
-2.13511690e-01 5.30168526e-02 -1.44843668e-01 9.50049281e-01
-1.57324225e-01 -1.17721204e-02 1.22459924e+00 -2.79073894e-01
-9.62940991e-01 1.31005242e-01 8.80433060e-03 1.59571916e-01
1.27855730e+00 -6.79779649e-01 -3.54918361e-01 -1.80295676e-01
-6.13622665e-01 2.53258377e-01 6.12987518e-01 6.28929257e-01
6.11811936e-01 -1.37914491e+00 -8.37175727e-01 3.79353344e-01
3.62895578e-01 -2.30574101e-01 5.89557648e-01 7.06302106e-01
1.37099728e-01 8.53775591e-02 -3.32946926e-01 -6.21719480e-01
-1.74743068e+00 8.88934672e-01 4.98987883e-01 -3.29509258e-01
-6.07008338e-01 6.79005861e-01 3.95503104e-01 -5.42745352e-01
3.67880195e-01 6.67071402e-01 -7.66144216e-01 2.21926391e-01
9.51060295e-01 7.37635016e-01 -2.23512091e-02 -1.18888843e+00
-5.58228135e-01 9.08882141e-01 -3.39739740e-01 3.58407229e-01
1.14410758e+00 -4.21104968e-01 6.16356544e-02 1.18684158e-01
1.08067024e+00 -1.25892058e-01 -1.16820884e+00 -4.01766211e-01
1.78795438e-02 -4.86757427e-01 -3.94978374e-01 -8.97922337e-01
-9.15289879e-01 5.72069824e-01 1.19620121e+00 1.70901746e-01
1.08695960e+00 6.52795583e-02 9.40060139e-01 2.82394700e-02
1.20794930e-01 -1.64557862e+00 2.68324524e-01 2.25613087e-01
4.64224994e-01 -1.75119328e+00 7.42513984e-02 -4.15910184e-01
-9.93392050e-01 7.05148995e-01 8.27925742e-01 1.21448055e-01
4.94397193e-01 1.22720547e-01 8.30179732e-03 2.25581482e-01
-9.61723924e-02 -4.11652982e-01 4.19685066e-01 1.28462338e+00
-1.22934461e-01 1.69553638e-01 6.61732107e-02 1.16604531e+00
2.15660095e-01 -1.53004229e-01 1.60661429e-01 6.79265916e-01
-2.87759453e-01 -1.18463683e+00 -8.24389517e-01 3.64662319e-01
-2.84346789e-01 3.06132942e-01 -7.89425850e-01 5.68259597e-01
6.09701812e-01 1.16731632e+00 1.64406627e-01 -8.14837217e-01
3.69402945e-01 1.64139806e-03 3.26394647e-01 -5.03487706e-01
-6.29162729e-01 -5.95141426e-02 -1.68601722e-01 -2.63913721e-01
-7.86443353e-01 -7.24827707e-01 -9.23056781e-01 -2.68694997e-01
-1.90311015e-01 -1.39100384e-02 5.18727563e-02 1.09124649e+00
4.06094760e-01 1.97198644e-01 7.77851522e-01 -8.38222623e-01
-5.83401799e-01 -7.60160863e-01 -7.06503868e-01 1.01793754e+00
2.97067970e-01 -6.81831181e-01 -4.69994217e-01 4.32793826e-01] | [14.788588523864746, 1.1114357709884644] |
4af620f6-15d0-4824-bb79-0526150eb70d | using-contextual-information-to-improve-blood | 1909.01735 | null | https://arxiv.org/abs/1909.01735v1 | https://arxiv.org/pdf/1909.01735v1.pdf | Using Contextual Information to Improve Blood Glucose Prediction | Blood glucose value prediction is an important task in diabetes management. While it is reported that glucose concentration is sensitive to social context such as mood, physical activity, stress, diet, alongside the influence of diabetes pathologies, we need more research on data and methodologies to incorporate and evaluate signals about such temporal context into prediction models. Person-generated data sources, such as actively contributed surveys as well as passively mined data from social media offer opportunity to capture such context, however the self-reported nature and sparsity of such data mean that such data are noisier and less specific than physiological measures such as blood glucose values themselves. Therefore, here we propose a Gaussian Process model to both address these data challenges and combine blood glucose and latent feature representations of contextual data for a novel multi-signal blood glucose prediction task. We find this approach outperforms common methods for multi-variate data, as well as using the blood glucose values in isolation. Given a robust evaluation across two blood glucose datasets with different forms of contextual information, we conclude that multi-signal Gaussian Processes can improve blood glucose prediction by using contextual information and may provide a significant shift in blood glucose prediction research and practice. | ['Rumi Chunara', 'Mohammad Akbari'] | 2019-08-24 | null | null | null | null | ['value-prediction'] | ['computer-code'] | [ 4.01180208e-01 -2.53652275e-01 -4.58531380e-01 -1.04621100e+00
-6.45365596e-01 -1.98541105e-01 6.22357368e-01 7.61159420e-01
-3.41325343e-01 9.66847718e-01 9.27847087e-01 1.26586467e-01
-3.72960806e-01 -1.11129081e+00 -4.02495563e-01 -7.97670007e-01
-2.51235306e-01 2.69156337e-01 -2.28912517e-01 2.56568760e-01
1.71489596e-01 -3.84943531e-04 -1.13262534e+00 -3.46296504e-02
1.10028064e+00 1.13629448e+00 -3.52112293e-01 5.75723052e-01
-1.69315174e-01 3.85959089e-01 -6.44965708e-01 -3.26131105e-01
1.40650481e-01 -6.95014834e-01 1.05626412e-01 2.19213471e-01
5.38780689e-01 7.62620494e-02 1.07242763e-01 9.98990059e-01
9.85445440e-01 3.18190083e-03 4.49367434e-01 -1.00149584e+00
-7.61827528e-01 6.12941802e-01 -3.47952127e-01 2.90341616e-01
1.86169147e-01 4.98118430e-01 5.04607141e-01 -2.15171933e-01
2.07957655e-01 1.33216000e+00 9.62328315e-01 1.27504215e-01
-1.90386510e+00 -3.53874624e-01 3.75574738e-01 -1.27193287e-01
-1.06672752e+00 -6.11679375e-01 5.65020621e-01 -6.41563773e-01
3.86728108e-01 3.93498212e-01 1.00119352e+00 1.86154938e+00
4.65036213e-01 2.45349526e-01 1.55666244e+00 -6.23123012e-02
4.52819824e-01 2.59711534e-01 3.64844084e-01 2.11596861e-01
4.41381544e-01 2.31352076e-01 -5.99563062e-01 -6.74670041e-01
7.35111713e-01 5.28699100e-01 -2.38417342e-01 -6.88935816e-02
-1.33529699e+00 6.95155084e-01 -2.90050749e-02 -1.47419214e-01
-9.46264625e-01 -1.23882843e-02 2.20642149e-01 2.64921337e-01
1.05445182e+00 2.05660015e-01 -7.21053064e-01 -4.24981207e-01
-9.43258584e-01 3.40596944e-01 1.05336595e+00 5.08982539e-01
6.27275705e-01 2.03436427e-03 -4.00138050e-01 7.62045205e-01
4.62153256e-01 6.84104741e-01 4.70707029e-01 -9.26818788e-01
7.40850419e-02 5.17903626e-01 3.05454016e-01 -1.36032379e+00
-7.18648136e-01 -5.77610612e-01 -9.13552403e-01 -1.45656466e-01
8.48448098e-01 -7.95420408e-01 -7.56235778e-01 1.84971702e+00
4.83975291e-01 5.75113952e-01 -1.04067445e-01 1.13972795e+00
6.79979563e-01 1.09477691e-01 8.70446563e-01 -5.41491091e-01
1.63574934e+00 -2.62174100e-01 -8.18672955e-01 -2.45765880e-01
7.75766298e-02 -5.16733706e-01 4.86538380e-01 4.54381078e-01
-8.83539379e-01 -5.15391409e-01 -6.59328699e-01 2.89453715e-01
-4.36142594e-01 -6.26318336e-01 1.06087267e+00 9.48879421e-01
-6.85418665e-01 8.43701482e-01 -1.01138556e+00 -7.88135290e-01
4.53985333e-01 -1.42113879e-01 -2.86298525e-02 -2.63238966e-01
-1.15184295e+00 7.64804780e-01 -1.79768115e-01 1.51737705e-01
-1.93684235e-01 -1.41797626e+00 -8.21885705e-01 -2.86092013e-01
1.73057497e-01 -1.21850049e+00 6.69854105e-01 -9.68621910e-01
-1.41671562e+00 5.74735701e-01 -3.74150872e-01 -5.01903474e-01
7.42855966e-01 -2.50053614e-01 -7.98817277e-01 7.40744248e-02
8.46515894e-02 3.04954648e-01 5.59287608e-01 -5.83236098e-01
-4.07972425e-01 -7.93172777e-01 -7.69299209e-01 -1.07056670e-01
1.38546705e-01 1.17100202e-01 1.72107488e-01 -4.57721680e-01
1.56574756e-01 -6.89168751e-01 -7.34280050e-01 1.32515922e-01
-2.67386347e-01 6.88891932e-02 3.30840647e-02 -6.30778968e-01
8.54846597e-01 -1.82029045e+00 -2.72470504e-01 2.66142547e-01
2.39603102e-01 -2.43272126e-01 1.37198985e-01 4.04049158e-01
1.60257384e-01 3.45325589e-01 7.60123730e-02 -2.50933707e-01
1.28402546e-01 3.16414684e-01 6.13456517e-02 7.19732821e-01
4.80995089e-01 1.16990185e+00 -1.20080900e+00 -2.30372682e-01
2.31007144e-01 8.37422192e-01 -4.97424960e-01 -2.57159203e-01
-1.88391894e-01 8.15119326e-01 -3.36534083e-01 7.30596900e-01
5.13266146e-01 -1.42061651e-01 3.35890859e-01 -2.43875116e-01
-1.34054020e-01 9.23162177e-02 -1.12856758e+00 1.46635270e+00
1.69449702e-01 3.13010693e-01 -6.76625296e-02 -1.02091444e+00
1.22594118e+00 5.57308555e-01 8.96649778e-01 -8.14347565e-01
-6.08334430e-02 -2.00172856e-01 5.90012483e-02 -8.32098782e-01
-1.06572367e-01 -6.05959117e-01 1.89647332e-01 5.56083694e-02
-2.48698324e-01 5.00926316e-01 1.19145855e-01 -3.62444162e-01
1.06921220e+00 3.25737923e-01 3.58417720e-01 -4.57045466e-01
-9.98623595e-02 -1.26543388e-01 1.15380549e+00 9.77198064e-01
-7.65168548e-01 4.36697841e-01 6.61337197e-01 -5.13518810e-01
-6.09013140e-01 -1.05784988e+00 -6.87997580e-01 8.94031942e-01
-3.87693495e-01 -4.50576723e-01 1.09794497e-01 -3.54684144e-01
7.24645495e-01 2.98189133e-01 -7.41736054e-01 -2.11572886e-01
-2.28811562e-01 -1.86681342e+00 1.32982135e-01 5.27067721e-01
1.30768254e-01 -4.02017802e-01 -2.88525671e-01 8.34379137e-01
1.43459421e-02 -9.21283364e-01 -1.93998605e-01 2.84581631e-01
-1.10266376e+00 -1.05923951e+00 -5.33456922e-01 -1.15938433e-01
2.90989220e-01 -2.68155158e-01 1.50751340e+00 -6.70413673e-01
-4.49709892e-01 5.18002212e-01 -1.23076990e-01 -7.93477833e-01
-9.78458226e-02 -5.44854283e-01 9.94180366e-02 4.85715806e-01
1.26103556e+00 -1.21491516e+00 -1.19969201e+00 1.56569034e-01
-3.07960361e-01 -2.19448119e-01 4.73570198e-01 3.30460608e-01
7.94888258e-01 -1.77941173e-01 1.10995257e+00 -1.09797418e+00
6.13743067e-01 -1.07854271e+00 -1.55993387e-01 -3.58835578e-01
-9.59267616e-01 -2.87010401e-01 -9.46798176e-02 -7.62021244e-01
-6.48424923e-01 -3.93130273e-01 3.57240468e-01 8.53209347e-02
-6.15943670e-01 8.43333066e-01 -3.43706347e-02 3.29195052e-01
9.60631371e-01 1.22097798e-01 4.22122031e-01 -6.26497686e-01
9.91394818e-02 3.05103898e-01 2.40738213e-01 -5.90984583e-01
9.14777443e-02 3.23267400e-01 3.58675241e-01 -8.74265194e-01
-8.08012426e-01 -3.89157742e-01 -3.95519048e-01 -2.35579573e-02
8.84167910e-01 -1.17847323e+00 -6.98040366e-01 2.91477084e-01
-4.26237643e-01 -2.90716112e-01 -1.59785241e-01 8.80275428e-01
-3.31199169e-01 1.05835401e-01 -6.83050275e-01 -1.00256324e+00
-1.63383558e-01 -6.48883820e-01 8.52634132e-01 1.32068068e-01
-5.47675788e-01 -1.51661921e+00 1.41209766e-01 2.79682696e-01
9.06997740e-01 1.27886784e+00 7.33027756e-01 -6.95592105e-01
-4.35831726e-01 -2.90335685e-01 -9.51738432e-02 2.00876832e-01
6.82358563e-01 -2.65214235e-01 -7.13849306e-01 1.42705217e-01
2.25091189e-01 4.33913544e-02 5.89550376e-01 9.00597811e-01
5.37393749e-01 -3.26426834e-01 -1.64001808e-01 5.06114364e-01
1.52608645e+00 7.32744262e-02 5.21902144e-01 -1.28693483e-03
3.48814547e-01 5.65481782e-01 -2.28009447e-02 6.61682069e-01
6.99151874e-01 4.58936185e-01 9.75726247e-02 -3.14042509e-01
1.06006056e-01 8.48797411e-02 1.47367954e-01 3.08403045e-01
-1.92087829e-01 2.63296723e-01 -6.83576465e-01 1.00564010e-01
-1.97932696e+00 -1.19400287e+00 -6.45041049e-01 2.17483640e+00
1.12605715e+00 -7.19610229e-02 4.10073489e-01 -2.33375177e-01
4.63471979e-01 9.03172567e-02 -5.29052675e-01 -8.64127949e-02
-4.68205750e-01 6.82254955e-02 6.06454909e-01 2.24155605e-01
-1.15510201e+00 1.49365067e-01 7.02702904e+00 -3.20564598e-01
-1.06129324e+00 1.36850446e-01 8.36935818e-01 -4.07103360e-01
1.30032346e-01 -3.12941343e-01 -5.49737573e-01 9.09158766e-01
1.55172086e+00 -2.34159529e-01 9.50194076e-02 4.84247535e-01
1.13881600e+00 -3.49649966e-01 -1.32614696e+00 9.48632300e-01
2.20553055e-02 -8.57098281e-01 -4.39551830e-01 2.52162546e-01
6.46621108e-01 1.84699789e-01 -4.14885581e-02 2.37830341e-01
2.83312857e-01 -8.58570814e-01 9.21352655e-02 1.26492655e+00
1.30098835e-01 -3.40784818e-01 5.97387195e-01 -2.98527386e-02
-6.00139022e-01 1.38732031e-01 -3.35678577e-01 -3.41337562e-01
2.75777042e-01 1.36305785e+00 -6.20035112e-01 3.77836019e-01
2.84656942e-01 1.12990344e+00 -7.40504503e-01 1.57074893e+00
1.51600838e-01 9.27860260e-01 -2.08959609e-01 1.20990299e-01
-8.13906863e-02 -5.86038768e-01 4.18574423e-01 1.37426209e+00
4.19120580e-01 9.99867637e-03 6.60291076e-01 9.33287024e-01
5.57116628e-01 3.41715068e-01 -2.87904084e-01 -1.11271836e-01
8.35716724e-02 1.13654888e+00 -4.11766648e-01 -4.17775929e-01
-9.52176213e-01 6.15752161e-01 -3.21236819e-01 7.57863224e-01
-5.13924420e-01 2.33366311e-01 1.41989195e+00 4.55039114e-01
-2.21892148e-01 -3.51068705e-01 -4.06654119e-01 -1.24969435e+00
-1.27992809e-01 -7.40847051e-01 4.53634024e-01 -4.23726916e-01
-1.89906323e+00 -2.95477033e-01 -3.51429939e-01 -8.15268874e-01
-3.21966633e-02 -3.98458540e-01 -5.89968920e-01 1.33481026e+00
-1.75784099e+00 -8.50104928e-01 -4.81799096e-01 4.16932911e-01
3.04714262e-01 2.18811810e-01 1.13429713e+00 3.64814132e-01
-6.20438337e-01 1.42570194e-02 -5.96096292e-02 4.02918719e-02
1.23352444e+00 -1.57831740e+00 -4.83166799e-02 4.81383562e-01
-2.35452577e-01 7.93585837e-01 8.77949059e-01 -1.04695749e+00
-1.66441798e+00 -1.44314313e+00 7.48601913e-01 -8.63922238e-01
7.13130355e-01 -2.67572910e-01 -7.60937214e-01 5.48600495e-01
-1.02913246e-01 1.61127210e-01 1.61616445e+00 7.39552677e-01
-1.15971893e-01 -3.68001521e-01 -1.31904650e+00 4.63789403e-01
1.01520467e+00 -1.32022440e-01 -4.08393174e-01 3.60739410e-01
4.04159218e-01 -9.85249653e-02 -1.63205278e+00 2.33275577e-01
5.44453859e-01 -7.76293278e-01 1.00396776e+00 -8.43480945e-01
2.03854799e-01 -1.75763275e-02 -3.52032274e-01 -1.58923769e+00
-5.43044031e-01 -7.54034281e-01 -3.62284303e-01 1.35000956e+00
3.63398254e-01 -9.93299842e-01 7.76881397e-01 1.46118903e+00
3.70247625e-02 -4.76200849e-01 -5.67851067e-01 -4.11282122e-01
-1.91971753e-02 -4.80319709e-01 5.46908677e-01 1.16381776e+00
2.01053515e-01 1.79272950e-01 -5.26761770e-01 2.25662217e-01
9.77841079e-01 1.63573921e-01 6.85355783e-01 -1.76830614e+00
-2.18597159e-01 -2.92077571e-01 -8.73359025e-01 -4.19285208e-01
-2.56529689e-01 -8.25516701e-01 -4.57215428e-01 -1.45868087e+00
1.22545943e-01 -4.41432506e-01 -7.04365194e-01 1.41610563e-01
-5.73993683e-01 2.39129215e-01 -3.02557558e-01 -3.17635596e-01
-3.11704755e-01 2.70568162e-01 1.12776220e+00 5.75540029e-02
-6.13481641e-01 1.94773003e-02 -1.21188796e+00 3.11790556e-01
7.84555852e-01 -2.41344467e-01 -3.56922358e-01 -7.20077455e-02
2.04090551e-01 2.16028452e-01 5.41028440e-01 -8.56577098e-01
8.83116275e-02 -7.02465892e-01 1.31318176e+00 6.37800097e-02
3.47516716e-01 -6.54877365e-01 3.91940475e-01 2.92123079e-01
-2.64375418e-01 -1.25123799e-01 -1.97810501e-01 1.05733609e+00
6.46548718e-02 4.69691306e-01 4.08170521e-01 -6.10977709e-01
-2.04754904e-01 5.12665749e-01 -4.40781772e-01 9.31457505e-02
6.37338102e-01 -1.88101858e-01 -4.52593595e-01 -2.84901291e-01
-1.33522844e+00 3.75366658e-01 9.30508375e-02 5.45037806e-01
2.31857494e-01 -1.35766447e+00 -1.13735330e+00 3.85752887e-01
7.04887584e-02 -4.90172178e-01 2.93085098e-01 1.47985899e+00
3.24317515e-01 1.43873751e-01 -3.16606686e-02 -7.51193523e-01
-8.31391692e-01 5.95300138e-01 3.39526534e-01 5.06531715e-01
-9.52984512e-01 1.68417022e-01 -2.18020499e-01 5.88537008e-02
2.76953280e-01 -6.44338131e-01 -7.25782365e-02 6.34832203e-01
8.87980878e-01 4.19806749e-01 -1.89708591e-01 -9.84561667e-02
6.45081326e-03 2.25445196e-01 4.20596868e-01 2.61398822e-01
1.65852332e+00 -5.33615828e-01 -6.57537356e-02 9.40389872e-01
7.53685176e-01 -3.90711538e-02 -1.40939450e+00 -5.39115429e-01
2.73668587e-01 -5.29983342e-01 -3.98983657e-02 -1.21614349e+00
-6.98348522e-01 2.99999356e-01 8.33730102e-01 4.68784153e-01
7.73950815e-01 -3.49275649e-01 2.54178375e-01 -1.03748292e-01
2.28530273e-01 -8.77506018e-01 -1.28411293e-01 -1.42503247e-01
6.04882061e-01 -1.42800891e+00 -4.62589264e-02 -3.08547527e-01
-5.97640872e-01 8.87438953e-01 1.33682773e-01 -4.41676863e-02
1.08987713e+00 6.11314252e-02 4.62167650e-01 2.55385060e-02
-1.03163016e+00 -2.73721129e-01 -6.00335700e-03 1.03095090e+00
6.59354627e-01 3.84317040e-01 -4.25297707e-01 8.84929538e-01
-1.89451918e-01 3.57901484e-01 3.76533002e-01 7.58865297e-01
-2.81887203e-01 -1.06131077e+00 -3.79044294e-01 1.03387308e+00
-8.58565807e-01 -1.54704317e-01 1.00694537e-01 2.37407550e-01
2.10755467e-01 1.19749129e+00 1.95689321e-01 2.59052932e-01
3.70791703e-01 1.04585564e+00 2.46481165e-01 -6.84338272e-01
-4.94526356e-01 2.16166392e-01 3.98367465e-01 -8.18133950e-01
-6.67422295e-01 -1.01488543e+00 -4.52549487e-01 -2.25688010e-01
4.07656044e-01 -1.10047586e-01 7.86236286e-01 7.55028725e-01
6.41179621e-01 5.24622858e-01 3.44392031e-01 -7.66426921e-01
-4.50639576e-01 -1.16691029e+00 -8.01583290e-01 7.68787980e-01
4.88412023e-01 -4.06558096e-01 -3.15131456e-01 3.07377458e-01] | [13.62446403503418, 3.2205724716186523] |
253f9582-c8ed-43a9-8c81-71c3290afa0d | solving-diffusion-odes-with-optimal-boundary | 2305.15357 | null | https://arxiv.org/abs/2305.15357v2 | https://arxiv.org/pdf/2305.15357v2.pdf | Solving Diffusion ODEs with Optimal Boundary Conditions for Better Image Super-Resolution | Diffusion models, as a kind of powerful generative model, have given impressive results on image super-resolution (SR) tasks. However, due to the randomness introduced in the reverse process of diffusion models, the performances of diffusion-based SR models are fluctuating at every time of sampling, especially for samplers with few resampled steps. This inherent randomness of diffusion models results in ineffectiveness and instability, making it challenging for users to guarantee the quality of SR results. However, our work takes this randomness as an opportunity: fully analyzing and leveraging it leads to the construction of an effective plug-and-play sampling method that owns the potential to benefit a series of diffusion-based SR methods. More in detail, we propose to steadily sample high-quality SR images from pretrained diffusion-based SR models by solving diffusion ordinary differential equations (diffusion ODEs) with optimal boundary conditions (BCs) and analyze the characteristics between the choices of BCs and their corresponding SR results. Our analysis shows the route to obtain an approximately optimal BC via an efficient exploration in the whole space. The quality of SR results sampled by the proposed method with fewer steps outperforms the quality of results sampled by current methods with randomness from the same pretrained diffusion-based SR model, which means that our sampling method "boosts" current diffusion-based SR models without any additional training. | ['Jiaying Liu', 'Jianlong Fu', 'Wenhan Yang', 'Huan Yang', 'Yiyang Ma'] | 2023-05-24 | null | null | null | null | ['image-super-resolution', 'efficient-exploration'] | ['computer-vision', 'methodology'] | [ 2.54960746e-01 1.88113824e-01 5.74296713e-03 2.07665339e-01
-7.89759099e-01 -1.28131881e-01 8.12446177e-01 -4.36419994e-01
-1.49432212e-01 8.15501273e-01 2.00590163e-01 1.15797698e-01
-3.97807419e-01 -9.73871827e-01 -3.83775562e-01 -1.03971171e+00
-1.17375039e-01 4.93147850e-01 4.99355555e-01 -3.53747964e-01
2.91449964e-01 6.07551098e-01 -1.52015734e+00 5.97441196e-02
1.12264752e+00 7.75763750e-01 7.11237073e-01 6.55909717e-01
-3.29221845e-01 7.93838918e-01 -5.10746062e-01 -3.49620497e-03
3.35770011e-01 -8.91340137e-01 -6.97605014e-01 2.58816108e-02
4.33573239e-02 -3.40091705e-01 -3.32399696e-01 1.14872921e+00
6.26554072e-01 1.22097671e-01 5.57668805e-01 -5.82440972e-01
-1.09893906e+00 7.70332634e-01 -7.87149549e-01 5.47960043e-01
1.66198164e-01 3.93948555e-01 5.48034668e-01 -7.38443017e-01
1.21144676e+00 1.20996928e+00 6.50765121e-01 8.10072541e-01
-1.56203294e+00 -4.26354766e-01 -2.70408895e-02 -6.35702386e-02
-1.37646842e+00 -5.31551123e-01 9.04264450e-01 -4.02603984e-01
3.92901570e-01 1.68411374e-01 9.56999362e-01 1.19933879e+00
1.05913326e-01 7.10829854e-01 1.54783094e+00 -2.73497820e-01
4.97927278e-01 3.20485443e-01 -1.03272632e-01 4.23927456e-01
1.13000289e-01 2.17901543e-01 -5.65279365e-01 -1.62413269e-01
1.38941371e+00 -3.07330996e-01 -4.46227074e-01 -4.38180327e-01
-1.07739162e+00 8.65293920e-01 4.84943867e-01 6.21435940e-01
-5.46281397e-01 1.14908129e-01 -1.67480960e-01 3.87085885e-01
7.85632789e-01 4.46022511e-01 8.09193030e-02 -1.13911949e-01
-1.34246373e+00 4.10968721e-01 4.70480770e-01 6.56379640e-01
6.35537148e-01 2.96063989e-01 -1.70629814e-01 9.54761982e-01
-2.49546096e-02 5.46483099e-01 5.08748651e-01 -1.09895563e+00
-1.54665440e-01 1.87589064e-01 2.02045932e-01 -8.69076788e-01
-7.29365572e-02 -8.49352419e-01 -1.14793110e+00 3.99627328e-01
4.87940580e-01 5.92090078e-02 -7.77492881e-01 1.54290235e+00
5.31137347e-01 1.63127556e-01 -1.33398324e-01 8.08086574e-01
4.52057004e-01 6.62535489e-01 -3.02256376e-01 -5.51308632e-01
8.13116074e-01 -8.13213527e-01 -7.70958662e-01 2.69634664e-01
3.40929896e-01 -5.17679989e-01 1.20588446e+00 6.42942548e-01
-1.33177900e+00 -6.00495815e-01 -9.79165077e-01 1.34583771e-01
-1.44085763e-02 -2.46031627e-01 5.43646157e-01 6.37976468e-01
-1.47262478e+00 1.15630829e+00 -7.47512877e-01 -1.34271786e-01
7.86746740e-01 1.06391557e-01 1.75097138e-01 -1.03176899e-01
-1.05904508e+00 7.54559100e-01 -2.36050591e-01 1.88727528e-02
-1.07232642e+00 -9.83794272e-01 -3.18161100e-01 -3.05865437e-01
1.38348013e-01 -7.35997081e-01 7.55566299e-01 -1.01206863e+00
-1.91002297e+00 5.76879680e-01 -2.54738599e-01 -5.59097052e-01
1.06551683e+00 4.09572572e-02 -1.88812956e-01 2.93248892e-01
2.86864932e-03 5.61146021e-01 1.21860635e+00 -1.41897237e+00
-2.57578582e-01 -2.28426307e-01 -1.28114164e-01 7.24716485e-02
-2.31822029e-01 -1.49364114e-01 -1.37041539e-01 -6.15698516e-01
1.15674771e-01 -7.29671776e-01 -6.70146704e-01 2.32198089e-01
-2.27829367e-01 2.36412529e-02 6.61289394e-01 -4.42938298e-01
1.22273970e+00 -2.12755919e+00 4.29779291e-01 5.09941243e-02
5.08292973e-01 1.66378364e-01 -1.07601054e-01 2.91729689e-01
2.65274227e-01 1.70987025e-01 -3.57365191e-01 -5.47154784e-01
-3.77293348e-01 -9.93209705e-03 -2.49684840e-01 5.72712660e-01
6.85120234e-03 8.69767308e-01 -1.12766588e+00 -4.96272266e-01
8.46978873e-02 8.88525665e-01 -4.05008614e-01 -2.87704598e-02
-2.25207970e-01 9.41401303e-01 -5.60053349e-01 2.57431805e-01
8.49730253e-01 -5.31772733e-01 -1.11765705e-01 8.17781761e-02
-1.90697074e-01 -6.58469722e-02 -1.32464015e+00 1.67091918e+00
-5.83681643e-01 5.53463459e-01 1.01741344e-01 -6.38570368e-01
1.01648617e+00 3.19648981e-02 5.26351988e-01 -8.52999210e-01
-2.77987421e-01 3.73264909e-01 -1.55495703e-01 -2.15514660e-01
4.79816079e-01 -2.91994750e-01 6.48707151e-01 7.38921642e-01
-2.48064980e-01 -3.96121114e-01 9.41550881e-02 3.70027035e-01
8.94245863e-01 2.17487603e-01 -6.63660243e-02 -4.62105453e-01
4.62008178e-01 -3.14502120e-02 1.65780589e-01 1.02768219e+00
1.49074234e-02 8.53982568e-01 2.76100069e-01 -3.80360097e-01
-1.30498159e+00 -1.02456045e+00 -4.21697050e-01 2.55550176e-01
1.97125465e-01 -1.73662961e-01 -8.19697738e-01 -4.17078495e-01
-3.48570913e-01 5.33873975e-01 -9.62617636e-01 9.91346687e-02
-5.52198052e-01 -9.72983778e-01 2.97579885e-01 1.22801222e-01
7.21736014e-01 -1.04124093e+00 -6.19679868e-01 2.94293463e-01
-3.32825705e-02 -8.07852387e-01 -1.67126268e-01 -1.94784060e-01
-1.19153464e+00 -6.63674951e-01 -1.40604734e+00 -1.90522254e-01
5.60392022e-01 2.60995388e-01 9.67778385e-01 1.50839046e-01
-2.07478255e-01 1.42159268e-01 -1.39487162e-01 1.36128794e-02
-7.78868020e-01 2.48407433e-03 -2.18036234e-01 1.87266082e-01
-4.26184237e-01 -8.67574513e-01 -1.01541770e+00 2.58479089e-01
-9.61058915e-01 1.75681338e-01 4.28613693e-01 6.34354234e-01
7.48079956e-01 3.57554913e-01 6.67811036e-01 -8.95975113e-01
8.16084564e-01 -4.68817085e-01 -6.36823893e-01 -1.38823956e-01
-1.00136781e+00 2.36178666e-01 6.28202677e-01 -8.58133972e-01
-1.19754291e+00 -2.76880860e-01 -1.02499470e-01 -4.99461979e-01
2.75385916e-01 1.08318821e-01 2.90222794e-01 -2.57400483e-01
9.24975038e-01 6.25600576e-01 2.93608904e-01 -6.03422046e-01
4.99956399e-01 3.51709396e-01 3.61792669e-02 -4.48177069e-01
7.50517905e-01 1.06375897e+00 9.51890424e-02 -1.09772503e+00
-5.83546042e-01 -7.34344125e-02 -4.06206071e-01 -2.89544195e-01
7.22242534e-01 -5.90733707e-01 -2.69459337e-01 7.59322166e-01
-8.31823230e-01 -7.07150280e-01 -9.35693860e-01 7.91612417e-02
-7.02526629e-01 4.56887305e-01 -7.29567111e-01 -1.19013286e+00
-1.65943936e-01 -1.31627142e+00 9.39024210e-01 1.73904613e-01
-1.02578565e-01 -1.14452875e+00 1.63660318e-01 1.27295509e-01
1.09733284e+00 1.84448302e-01 6.33384168e-01 2.42608320e-02
-9.96993840e-01 2.22774386e-01 -2.23620743e-01 2.61557877e-01
3.96053642e-02 -1.15590401e-01 -8.37521076e-01 -2.53310174e-01
4.36743587e-01 -4.66625988e-02 9.64282334e-01 8.86302352e-01
9.26782966e-01 -2.09449843e-01 -4.12741572e-01 7.23472118e-01
1.63915801e+00 -9.53346118e-02 9.06573176e-01 4.09224898e-01
4.19045895e-01 4.64668065e-01 2.04737991e-01 3.46026361e-01
1.92513093e-02 7.16253698e-01 1.08734339e-01 -2.37772450e-01
-6.76136434e-01 -2.60730118e-01 2.16591045e-01 4.31533098e-01
-3.58261228e-01 -1.88964568e-02 -5.27827084e-01 5.60272396e-01
-1.45751846e+00 -1.03350425e+00 -2.32540473e-01 2.29586625e+00
1.11988962e+00 7.97366500e-02 2.98395842e-01 1.31814212e-01
5.28460741e-01 4.58681524e-01 -6.08555734e-01 -1.32129997e-01
-3.94066513e-01 2.69333035e-01 3.37697417e-01 5.84976017e-01
-5.18088996e-01 8.23013127e-01 6.92482805e+00 1.13318384e+00
-1.27316022e+00 3.51338863e-01 7.47024536e-01 -1.91424817e-01
-7.58912325e-01 1.22505454e-02 -8.05651605e-01 5.36821365e-01
8.64601016e-01 -2.42033064e-01 7.13754833e-01 7.17446625e-01
6.14009440e-01 -3.76437873e-01 -6.50382578e-01 7.88642704e-01
-2.14231297e-01 -1.73985493e+00 1.69483885e-01 4.22520518e-01
1.06813312e+00 7.56933838e-02 3.85902554e-01 -3.05428922e-01
6.17092252e-01 -1.04361773e+00 7.35597074e-01 6.82818413e-01
7.45136440e-01 -4.53661412e-01 2.32512057e-01 4.54885781e-01
-7.69399822e-01 -5.45714982e-03 -5.37813187e-01 2.10496917e-01
4.73748505e-01 1.19566131e+00 -3.18004072e-01 4.51626420e-01
6.94006681e-01 7.59310722e-01 -3.74011010e-01 7.73519635e-01
2.44296361e-02 5.70387185e-01 -3.42857599e-01 8.71844739e-02
3.80389877e-02 -5.22563994e-01 9.31368649e-01 9.11075592e-01
4.35004056e-01 9.44603682e-02 -4.45060104e-01 1.43364489e+00
2.97875881e-01 7.68404827e-02 -6.73917890e-01 9.15388837e-02
1.47112936e-01 9.43394840e-01 -9.19761896e-01 -4.56555426e-01
8.08833688e-02 1.01571119e+00 1.74393147e-01 4.88722295e-01
-6.69887125e-01 3.04217786e-01 3.21524918e-01 6.55921638e-01
4.40213025e-01 -4.18096900e-01 -4.18641984e-01 -1.07887268e+00
-2.03195244e-01 -8.96601439e-01 5.37234731e-02 -6.97496176e-01
-1.26884401e+00 9.95168626e-01 -2.20112000e-02 -1.03104007e+00
-4.11086939e-02 2.49052554e-01 -3.47301155e-01 1.08746421e+00
-1.70932662e+00 -8.86714876e-01 -1.96065888e-01 5.67248464e-01
7.63577700e-01 1.19449645e-01 4.17763174e-01 1.07392237e-01
-1.79650933e-01 2.39581630e-01 2.89559156e-01 -4.16506410e-01
1.68496653e-01 -1.02210832e+00 3.99739057e-01 7.63655305e-01
1.14222363e-01 6.80171430e-01 9.41508770e-01 -8.20243776e-01
-1.22889137e+00 -6.50085926e-01 5.34671068e-01 -4.29774582e-01
6.65937901e-01 -1.36711583e-01 -1.09444380e+00 7.33257681e-02
1.09436266e-01 1.02313673e-02 -8.12192261e-02 -1.23469509e-01
4.53614555e-02 5.24317846e-02 -1.33145285e+00 7.25990295e-01
1.43522727e+00 -2.84533173e-01 -8.70083645e-02 3.34370703e-01
5.81538141e-01 -3.11830014e-01 -7.32829273e-01 1.08630516e-01
4.08528388e-01 -1.36415982e+00 1.24496984e+00 -7.22998977e-02
6.92150056e-01 -1.42866641e-01 2.64221907e-01 -1.36541533e+00
-2.30097055e-01 -8.59665275e-01 -1.57079905e-01 1.00074351e+00
3.33171755e-01 -8.18565845e-01 6.64503932e-01 3.59794647e-01
4.21921432e-01 -7.92872727e-01 -8.16999435e-01 -1.04225588e+00
2.83773333e-01 -2.87975341e-01 5.57500064e-01 8.30183804e-01
-5.99940777e-01 -3.98495905e-02 -3.14832956e-01 -1.15732491e-01
1.16741455e+00 9.73982811e-02 6.31302357e-01 -1.14816260e+00
-5.04572630e-01 -6.90067947e-01 1.32903442e-01 -1.38216972e+00
-3.34606469e-01 -6.16052449e-01 -2.65286803e-01 -1.51593888e+00
1.04025707e-01 -1.08334231e+00 1.77830935e-01 -4.31083262e-01
-4.37904522e-02 2.61929989e-01 1.23331845e-01 7.22362578e-01
-1.44858360e-01 6.39992714e-01 1.96900845e+00 7.65085816e-02
-5.63954532e-01 6.72913268e-02 -4.64309365e-01 5.05308688e-01
3.87533456e-01 -5.07788718e-01 -6.92312956e-01 -2.07892537e-01
4.00520116e-01 3.62679571e-01 3.34795713e-01 -9.90060687e-01
1.33580193e-01 -1.56909540e-01 2.47599453e-01 -2.92317748e-01
3.74205649e-01 -3.78367841e-01 5.01742065e-01 3.84268522e-01
-4.95040506e-01 -2.59020984e-01 -1.87486693e-01 7.09129155e-01
8.53565335e-02 -2.39820927e-01 1.10943413e+00 -4.31443751e-01
-2.99898356e-01 2.55702645e-01 -3.38088810e-01 1.06942452e-01
6.86012983e-01 -6.08866274e-01 9.72565040e-02 -6.13547087e-01
-1.04345119e+00 -2.91288048e-01 7.46852338e-01 8.44148993e-02
5.48126340e-01 -9.64450061e-01 -8.04053366e-01 2.90018260e-01
-4.65647101e-01 1.99671373e-01 4.49521750e-01 1.01160693e+00
-3.42665285e-01 -1.80264518e-01 5.64128160e-03 -7.29421556e-01
-6.96587145e-01 6.17552280e-01 4.55472559e-01 -5.57044804e-01
-1.18355286e+00 8.23841631e-01 1.55330926e-01 1.33268088e-01
-7.72770718e-02 -8.66102278e-02 -1.92689940e-01 -5.26296161e-02
6.77113712e-01 5.42904675e-01 -2.61752337e-01 -2.00456500e-01
1.11889310e-01 7.29091406e-01 2.77765021e-02 -4.63567168e-01
1.53594661e+00 -5.03091037e-01 -3.80209903e-03 3.60412270e-01
7.66372919e-01 2.61010170e-01 -1.57373786e+00 -2.09749848e-01
-2.46139526e-01 -6.96597934e-01 3.48137587e-01 -5.68098128e-01
-1.31380093e+00 7.90958464e-01 5.22843361e-01 4.52098072e-01
9.67160285e-01 5.38684987e-02 9.34833109e-01 -4.75494564e-01
5.98238528e-01 -8.76258373e-01 1.45824313e-01 2.13619582e-02
8.97265553e-01 -8.67666304e-01 3.16480137e-02 -4.45200831e-01
-5.98413765e-01 8.34352434e-01 -7.02461647e-03 -3.49656165e-01
7.66086340e-01 4.69764471e-01 -3.95653844e-02 -2.36327931e-01
-6.21328473e-01 -1.06232561e-01 -5.11280484e-02 8.61998737e-01
-1.90474617e-04 -2.70583063e-01 -4.35486674e-01 2.81882100e-02
-1.14921164e-02 4.13181454e-01 6.77357852e-01 5.79051733e-01
-3.04207891e-01 -1.03771687e+00 -2.43620202e-01 1.95972234e-01
-2.13877067e-01 -2.74837494e-01 7.67652364e-03 6.89264953e-01
-1.63414791e-01 6.66777134e-01 -2.04929948e-01 1.11168861e-01
-4.47772294e-02 -3.59050751e-01 6.33217335e-01 -2.69893706e-01
-4.25129831e-01 9.66297165e-02 -1.56489983e-01 -6.20982170e-01
-4.67204005e-01 -8.07581961e-01 -1.06836939e+00 -4.15263295e-01
-2.89671391e-01 1.29880413e-01 5.91373503e-01 8.70267093e-01
5.08393764e-01 3.79349440e-01 5.99527121e-01 -8.85212779e-01
-6.97920561e-01 -7.98099756e-01 -7.25133896e-01 2.31692046e-01
3.35847408e-01 -6.31042182e-01 -6.64875507e-01 -7.24053457e-02] | [11.39459228515625, -2.009979724884033] |
efb03827-aa7a-47de-96c8-02092c42a004 | melanoma-detection-using-adversarial-training | 2004.06824 | null | https://arxiv.org/abs/2004.06824v2 | https://arxiv.org/pdf/2004.06824v2.pdf | Melanoma Detection using Adversarial Training and Deep Transfer Learning | Skin lesion datasets consist predominantly of normal samples with only a small percentage of abnormal ones, giving rise to the class imbalance problem. Also, skin lesion images are largely similar in overall appearance owing to the low inter-class variability. In this paper, we propose a two-stage framework for automatic classification of skin lesion images using adversarial training and transfer learning toward melanoma detection. In the first stage, we leverage the inter-class variation of the data distribution for the task of conditional image synthesis by learning the inter-class mapping and synthesizing under-represented class samples from the over-represented ones using unpaired image-to-image translation. In the second stage, we train a deep convolutional neural network for skin lesion classification using the original training set combined with the newly synthesized under-represented class samples. The training of this classifier is carried out by minimizing the focal loss function, which assists the model in learning from hard examples, while down-weighting the easy ones. Experiments conducted on a dermatology image benchmark demonstrate the superiority of our proposed approach over several standard baseline methods, achieving significant performance improvements. Interestingly, we show through feature visualization and analysis that our method leads to context based lesion assessment that can reach an expert dermatologist level. | ['A. Ben Hamza', 'Hasib Zunair'] | 2020-04-14 | null | null | null | null | ['skin-lesion-classification'] | ['medical'] | [ 9.20091212e-01 2.20362023e-01 -2.11655736e-01 -3.06968421e-01
-1.03919888e+00 -4.51151192e-01 4.61658895e-01 3.08931470e-01
-4.66952115e-01 6.75913393e-01 -2.59504080e-01 -1.12013690e-01
1.07119605e-01 -7.28777707e-01 -6.47087812e-01 -1.16791296e+00
3.39481175e-01 2.39904955e-01 1.55446799e-02 6.85570985e-02
2.74947677e-02 4.94231492e-01 -1.35108316e+00 6.01663411e-01
1.12722576e+00 1.05829597e+00 -1.05923161e-01 7.60024786e-01
5.87801784e-02 6.80121124e-01 -5.35542011e-01 -3.55583549e-01
1.94924310e-01 -5.55552185e-01 -5.67937911e-01 4.30721521e-01
6.50266767e-01 -2.41398931e-01 -4.65105958e-02 1.16018534e+00
5.71561635e-01 -3.00842583e-01 9.30025458e-01 -1.08935845e+00
-5.02153039e-01 -3.06604784e-02 -6.67735875e-01 -1.99160472e-01
2.08538055e-01 2.79399872e-01 7.69120514e-01 -6.58951521e-01
7.98376203e-01 7.59183228e-01 5.43223977e-01 8.08461607e-01
-1.45453644e+00 -4.30866599e-01 -1.87676009e-02 1.04010321e-01
-1.32505655e+00 -1.62212744e-01 8.42437148e-01 -6.05993748e-01
3.40363294e-01 5.57812333e-01 6.07726455e-01 1.22038591e+00
8.85922834e-02 7.52159715e-01 1.50701642e+00 -6.56538069e-01
2.37291589e-01 4.61260915e-01 -1.05507977e-01 6.53442383e-01
-1.43200420e-02 -7.97802433e-02 -8.19021910e-02 -1.46420568e-01
4.80280221e-01 1.16442397e-01 -3.37662011e-01 -4.59438622e-01
-8.12335372e-01 6.95917666e-01 6.07836306e-01 1.56129509e-01
-4.32196319e-01 -2.66204834e-01 2.98104733e-01 1.36888444e-01
5.11035085e-01 1.24499589e-01 -7.59050669e-03 4.02009845e-01
-8.91390026e-01 1.13330856e-01 5.29533982e-01 2.70889640e-01
4.70998019e-01 -3.02543193e-01 -4.75025892e-01 1.01968718e+00
-8.78554955e-02 3.76342595e-01 6.26036465e-01 -4.29135561e-01
2.40350023e-01 7.66974032e-01 -5.29093742e-02 -8.13619673e-01
-2.30722174e-01 -5.05261660e-01 -1.03342533e+00 6.27951443e-01
6.63758337e-01 6.30134419e-02 -1.25544536e+00 1.77240705e+00
5.37039816e-01 -1.57291204e-01 1.13899812e-01 8.57372642e-01
3.84043455e-01 2.37201110e-01 3.08763862e-01 -2.60698590e-02
1.15830767e+00 -9.15954113e-01 -4.06170607e-01 -1.65242806e-01
4.24010158e-01 -5.88463843e-01 1.15580654e+00 3.86453986e-01
-9.22114611e-01 -2.79392451e-01 -1.03286815e+00 1.62546277e-01
-3.31562281e-01 5.22357881e-01 3.28016728e-01 5.34462154e-01
-7.74443567e-01 4.37545419e-01 -6.77340746e-01 -4.23567802e-01
9.76226211e-01 1.73965573e-01 -7.15044558e-01 -2.10475221e-01
-7.66905129e-01 6.89355671e-01 3.14143270e-01 -4.62275464e-03
-6.98317885e-01 -9.52072084e-01 -5.97431779e-01 -1.80184901e-01
1.63774431e-01 -5.74036658e-01 8.58208835e-01 -1.57309496e+00
-1.31103349e+00 1.19802678e+00 1.12188859e-02 -3.01761806e-01
9.96270537e-01 2.79262990e-01 -3.98435414e-01 3.51891667e-01
1.17093846e-02 6.08053863e-01 1.06852472e+00 -1.39495575e+00
-6.27454460e-01 -4.52410996e-01 -8.27449262e-02 8.33058208e-02
-5.08513033e-01 -4.93605286e-01 -2.81973451e-01 -6.21900260e-01
-1.10053264e-01 -8.95937026e-01 -1.52103201e-01 6.17477953e-01
-5.74774623e-01 2.80888140e-01 5.20492971e-01 -8.25432301e-01
9.77579892e-01 -2.26462889e+00 8.55450928e-02 4.47017342e-01
1.55121922e-01 3.66265148e-01 -3.12904716e-01 2.43153229e-01
-3.98104668e-01 -7.18209967e-02 -5.56606293e-01 -1.56096727e-01
-1.44924417e-01 -5.03639085e-03 -1.17506228e-01 4.59118754e-01
7.97467053e-01 7.17583954e-01 -9.04641986e-01 -3.95294905e-01
2.59044528e-01 5.46767354e-01 -3.49896461e-01 2.36658499e-01
-1.47897571e-01 5.71704090e-01 -6.53782338e-02 8.29489470e-01
7.94955373e-01 -2.34304056e-01 2.07070649e-01 -3.59709024e-01
4.13185060e-01 -2.80951917e-01 -7.38016367e-01 1.46980143e+00
-6.41624928e-01 5.08580387e-01 1.10725397e-02 -9.08933342e-01
5.34966886e-01 1.60155267e-01 3.43118608e-01 -7.84551203e-01
4.47398387e-02 2.72452116e-01 2.24215120e-01 -6.55926287e-01
-3.86296250e-02 -3.43861550e-01 2.88905114e-01 2.37488270e-01
-1.58726647e-01 -8.02496150e-02 1.71245202e-01 3.94349732e-02
9.91048455e-01 -7.48168379e-02 3.20201784e-01 1.89719751e-01
6.11223757e-01 -6.49733394e-02 2.53875405e-01 4.38319862e-01
-3.09520185e-01 7.34062910e-01 8.68588746e-01 -2.27348521e-01
-9.76901770e-01 -1.24628115e+00 -4.00449663e-01 6.00695193e-01
-1.05979271e-01 3.23520958e-01 -1.09620535e+00 -1.06225336e+00
1.57397352e-02 5.51593542e-01 -1.10272491e+00 -3.00640613e-01
-3.00912440e-01 -9.60011542e-01 4.70191836e-01 4.38016474e-01
2.40358442e-01 -8.79006207e-01 -4.53071177e-01 -8.86810273e-02
1.52236205e-02 -9.28038061e-01 -3.34764719e-01 1.60614416e-01
-4.80077326e-01 -1.35734987e+00 -1.03873992e+00 -8.83954406e-01
1.35833836e+00 -9.37778875e-02 8.43074441e-01 2.84080580e-02
-1.07121527e+00 1.10656723e-01 -1.95144072e-01 -3.09309989e-01
-6.49936199e-01 -1.52197585e-01 -4.13336545e-01 6.02779686e-01
1.74706548e-01 -3.15274298e-01 -9.08308923e-01 -2.29537040e-02
-1.26312530e+00 1.08547427e-01 9.28579986e-01 1.32435846e+00
8.27923417e-01 -1.03006363e-02 6.16085708e-01 -1.20577908e+00
3.81966531e-01 -3.83308232e-01 -2.06167340e-01 3.92340690e-01
-4.47161943e-01 -1.99416682e-01 6.68529034e-01 -5.09685874e-01
-9.71548259e-01 2.87115157e-01 -4.42902744e-02 -2.48323426e-01
-1.56997710e-01 2.48307407e-01 -1.66378349e-01 -2.53893524e-01
8.65150690e-01 3.25425029e-01 4.00249720e-01 -5.22989072e-02
2.27311105e-01 6.67851508e-01 5.93075335e-01 -2.46925235e-01
6.56073153e-01 7.48676419e-01 4.81961221e-02 -6.51364207e-01
-8.45281959e-01 -2.52680033e-01 -5.94838679e-01 -2.27762535e-01
7.52255321e-01 -7.31946945e-01 -2.57780164e-01 7.32598543e-01
-6.66975439e-01 -4.24623728e-01 -5.84450126e-01 8.39605406e-02
-3.56567979e-01 2.95208037e-01 -3.80511314e-01 -6.19767249e-01
-3.40700954e-01 -1.12699294e+00 1.24544621e+00 2.91872174e-01
-1.28429323e-01 -1.19641626e+00 -4.62097973e-02 3.87651324e-01
2.61833102e-01 7.60865390e-01 1.22575212e+00 -6.53089821e-01
-1.84848502e-01 -7.87497461e-01 -2.78890193e-01 7.79096067e-01
3.79740000e-01 1.26479968e-01 -1.12178850e+00 -4.30539817e-01
-3.07598859e-01 -5.66795170e-01 9.36629117e-01 3.21429342e-01
1.45531237e+00 -2.62394566e-02 -3.03489983e-01 5.59305370e-01
1.54321492e+00 -1.87230587e-01 6.11630619e-01 6.37410283e-02
4.93558735e-01 8.86527359e-01 6.47350550e-01 1.35979727e-01
1.86761729e-02 4.58679110e-01 5.25631189e-01 -6.10342443e-01
-5.15692949e-01 -2.37288728e-01 5.41417254e-03 1.19136378e-01
2.75373131e-01 -1.21038340e-01 -7.41766095e-01 6.03091180e-01
-1.47450900e+00 -7.05633163e-01 7.80433193e-02 2.34439945e+00
1.04873776e+00 1.41338527e-01 1.55735433e-01 2.27804393e-01
8.58425021e-01 -8.15356001e-02 -6.99825525e-01 -1.47068679e-01
-1.91767752e-01 4.95609343e-01 3.32884014e-01 4.30780977e-01
-1.30608284e+00 6.31441295e-01 5.20998621e+00 1.10571039e+00
-1.63367510e+00 -9.21636522e-02 1.04535139e+00 -2.19687074e-01
-2.02500060e-01 -3.43145549e-01 -2.19552845e-01 5.20181477e-01
5.25335610e-01 -4.52737175e-02 1.64560422e-01 5.92770100e-01
6.47662058e-02 -1.58593565e-01 -1.13056099e+00 8.18846822e-01
2.32668668e-01 -1.09884334e+00 2.12863550e-01 -1.53010473e-01
8.65931332e-01 -4.10660356e-01 4.86224204e-01 -8.54958221e-02
-2.44007096e-01 -1.11561811e+00 4.24824327e-01 6.61091685e-01
1.18633342e+00 -7.00641155e-01 7.42264807e-01 1.28386363e-01
-6.53802991e-01 3.35570388e-02 -7.28609338e-02 3.63206416e-01
-4.09515351e-01 6.74699485e-01 -1.12210524e+00 5.11970043e-01
2.41190597e-01 4.71609622e-01 -7.12124705e-01 9.36887503e-01
-2.21456245e-01 3.41724306e-01 -7.02610612e-02 1.80594504e-01
1.16848044e-01 -2.04008967e-01 3.34675044e-01 1.16734231e+00
8.65919665e-02 -4.05140430e-01 -8.59183667e-04 8.77279341e-01
-6.55888766e-02 2.16510892e-01 -4.79726911e-01 -2.89734565e-02
1.65984213e-01 1.56691647e+00 -6.50745511e-01 -2.54406512e-01
-7.91286156e-02 1.28594244e+00 3.19860816e-01 3.16116631e-01
-6.03514135e-01 -6.18623376e-01 3.40074986e-01 1.79544076e-01
5.83170131e-02 6.32831395e-01 -3.33661228e-01 -9.35794711e-01
4.20548409e-01 -9.27588165e-01 3.76340359e-01 -5.74720800e-01
-1.49884403e+00 6.54106379e-01 -5.44952154e-01 -1.34998488e+00
-1.81862488e-01 -7.40977764e-01 -7.88019180e-01 9.87513661e-01
-1.74576640e+00 -1.51841938e+00 -6.03844285e-01 4.39635813e-01
2.84774899e-01 -3.39292064e-02 9.84607816e-01 2.76334137e-01
-5.68055809e-01 1.05186617e+00 1.67543665e-01 1.58666641e-01
9.51173484e-01 -1.51971567e+00 -5.45968972e-02 6.72435522e-01
-3.11359346e-01 1.49331644e-01 3.50356549e-01 -3.63592088e-01
-8.40758562e-01 -1.37324524e+00 5.22218823e-01 -2.33506531e-01
6.20855749e-01 -3.25217187e-01 -8.45718265e-01 2.11765319e-01
8.34773388e-03 4.38840777e-01 9.19649482e-01 -5.05600095e-01
-5.52501202e-01 -2.79614031e-01 -1.57816827e+00 7.78978527e-01
4.36703831e-01 -5.06421268e-01 -9.10529718e-02 6.00392044e-01
1.05550602e-01 -3.82229716e-01 -8.76133561e-01 3.95616442e-01
7.50942826e-01 -8.22426438e-01 9.10447896e-01 -7.73179710e-01
9.44834769e-01 -4.99475151e-02 -1.91418864e-02 -1.43932092e+00
6.22013919e-02 -1.89725205e-01 3.14330369e-01 1.15233898e+00
3.80326718e-01 -5.33447146e-01 1.00456309e+00 3.97727400e-01
1.88177302e-01 -1.14537013e+00 -8.90040576e-01 -5.10489464e-01
3.46099734e-01 3.09296288e-02 8.67538080e-02 7.89788365e-01
-2.83314407e-01 -1.81254044e-01 -7.18285590e-02 1.15282938e-01
7.83002257e-01 1.22559473e-01 5.98610640e-01 -7.88077712e-01
-3.96972418e-01 -3.62155706e-01 -5.96508622e-01 -2.67468125e-01
1.48664996e-01 -1.19848871e+00 -7.54111782e-02 -1.16258264e+00
4.76961374e-01 -4.38110471e-01 -4.94363755e-01 5.40300906e-01
-5.33939302e-01 6.95565283e-01 -8.68798122e-02 -7.31353229e-03
-2.33573735e-01 2.20341951e-01 1.45902514e+00 -4.82784897e-01
5.08705936e-02 1.25518367e-01 -7.35417724e-01 6.60596669e-01
6.13140941e-01 -3.33261251e-01 -2.58247167e-01 -1.56573087e-01
-2.46349230e-01 -1.33220926e-01 8.50145876e-01 -9.38171983e-01
1.11946119e-02 -8.48055705e-02 6.91901505e-01 -2.24330992e-01
3.21378171e-01 -9.13183510e-01 -1.49310663e-01 8.74427199e-01
-6.06811345e-01 -6.38798475e-01 1.07798092e-01 7.55459070e-01
-2.54315883e-01 -2.51933545e-01 1.24668169e+00 1.21389516e-01
-3.33463788e-01 2.83460736e-01 -8.80886763e-02 3.79086696e-02
1.32413268e+00 -2.24491864e-01 -3.22281480e-01 -1.06256373e-01
-8.99104178e-01 -1.08109802e-01 5.68525195e-01 1.85736850e-01
4.58137363e-01 -1.30293441e+00 -8.64139915e-01 4.50096250e-01
4.84707445e-01 -5.31890765e-02 6.32049203e-01 9.71666873e-01
-6.66290820e-01 -4.97063883e-02 -3.23129207e-01 -7.79520810e-01
-1.26196134e+00 2.71595120e-01 6.19351983e-01 -4.54634458e-01
-2.50925541e-01 7.90935755e-01 3.49070519e-01 -4.47662711e-01
1.62972122e-01 -7.01851547e-02 8.69080722e-02 -1.97221618e-02
5.31961799e-01 1.06143363e-01 3.23714018e-01 -3.01849365e-01
-2.38666847e-01 5.08209825e-01 -5.29061973e-01 1.30253047e-01
1.14781666e+00 2.75875777e-01 -1.13933936e-01 2.55429626e-01
1.41785610e+00 6.81393081e-03 -1.30655587e+00 -5.89715876e-02
-2.16053903e-01 -6.22239888e-01 -9.16179046e-02 -1.25691736e+00
-1.14449656e+00 9.28994775e-01 1.15139627e+00 -6.97633717e-03
1.48665643e+00 -3.11652094e-01 4.99772906e-01 -1.58791374e-02
-1.04334675e-01 -1.06908047e+00 2.37679288e-01 -1.63658261e-01
8.22211623e-01 -1.45283926e+00 -1.46652505e-01 -5.48783064e-01
-6.59325063e-01 1.05344594e+00 5.26768565e-01 -3.34987730e-01
4.12676841e-01 1.43799096e-01 3.34482431e-01 3.21824364e-02
-6.42312527e-01 -8.55747536e-02 2.93468684e-01 8.58827770e-01
2.49725506e-01 2.21242860e-01 -3.10511649e-01 4.64282125e-01
1.70158476e-01 -1.00557981e-02 3.87301743e-01 7.72799671e-01
-1.02149248e-02 -1.28410339e+00 -1.55593410e-01 6.33364618e-01
-5.60367584e-01 1.75690018e-02 -5.87060928e-01 9.02224481e-01
1.69903591e-01 5.77960610e-01 8.44504535e-02 -2.27958992e-01
3.32651019e-01 9.70549956e-02 6.57811701e-01 -5.41881561e-01
-5.82078636e-01 3.11262370e-03 -1.82533085e-01 -3.75796258e-01
-1.13340594e-01 -5.26503026e-01 -8.52249503e-01 2.53187507e-01
-2.45538101e-01 -2.13912964e-01 7.82738328e-01 7.08108008e-01
1.99125037e-01 8.01114500e-01 9.38013852e-01 -6.09692633e-01
-9.30184245e-01 -7.16913044e-01 -6.35116100e-01 8.74494970e-01
4.81991678e-01 -3.52054358e-01 -3.90759796e-01 2.37180308e-01] | [15.498924255371094, -2.7835607528686523] |
6c6eecd8-fb37-420e-8244-9085c6cd6a2e | time-discretization-invariant-safe-action | 2111.03941 | null | https://arxiv.org/abs/2111.03941v6 | https://arxiv.org/pdf/2111.03941v6.pdf | Time Discretization-Invariant Safe Action Repetition for Policy Gradient Methods | In reinforcement learning, continuous time is often discretized by a time scale $\delta$, to which the resulting performance is known to be highly sensitive. In this work, we seek to find a $\delta$-invariant algorithm for policy gradient (PG) methods, which performs well regardless of the value of $\delta$. We first identify the underlying reasons that cause PG methods to fail as $\delta \to 0$, proving that the variance of the PG estimator can diverge to infinity in stochastic environments under a certain assumption of stochasticity. While durative actions or action repetition can be employed to have $\delta$-invariance, previous action repetition methods cannot immediately react to unexpected situations in stochastic environments. We thus propose a novel $\delta$-invariant method named Safe Action Repetition (SAR) applicable to any existing PG algorithm. SAR can handle the stochasticity of environments by adaptively reacting to changes in states during action repetition. We empirically show that our method is not only $\delta$-invariant but also robust to stochasticity, outperforming previous $\delta$-invariant approaches on eight MuJoCo environments with both deterministic and stochastic settings. Our code is available at https://vision.snu.ac.kr/projects/sar. | ['Gunhee Kim', 'Jaekyeom Kim', 'Seohong Park'] | 2021-11-06 | null | http://proceedings.neurips.cc/paper/2021/hash/024677efb8e4aee2eaeef17b54695bbe-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/024677efb8e4aee2eaeef17b54695bbe-Paper.pdf | neurips-2021-12 | ['policy-gradient-methods'] | ['methodology'] | [-5.28226839e-03 -5.41820168e-01 -1.57944351e-01 -7.35297054e-02
-6.96159363e-01 -7.12468565e-01 5.64352393e-01 -2.21416295e-01
-8.76742005e-01 1.06285417e+00 -3.06487501e-01 -4.60397571e-01
-2.33960748e-01 -7.22117782e-01 -6.22719705e-01 -8.61315668e-01
-4.49167788e-01 4.70167771e-02 3.54914486e-01 -4.43406641e-01
4.55065459e-01 3.90181184e-01 -1.70532000e+00 -1.40003428e-01
7.97246635e-01 9.22282994e-01 3.42221975e-01 8.17529321e-01
6.93863109e-02 5.30558944e-01 -7.44587481e-01 2.65442193e-01
5.77623308e-01 -7.84344852e-01 -5.47587037e-01 -1.24818265e-01
5.82919503e-03 -1.74436644e-01 -1.28014237e-01 1.32788110e+00
6.70051277e-01 6.49546504e-01 4.59067166e-01 -1.08362389e+00
-2.45824277e-01 5.67734540e-01 -4.57136989e-01 4.98539388e-01
3.96336734e-01 6.47689283e-01 6.86460018e-01 -5.31018674e-01
5.30147254e-01 1.33809459e+00 4.56746876e-01 7.81380653e-01
-1.14696610e+00 -6.76223636e-01 7.40858614e-01 9.23736170e-02
-1.24737811e+00 -2.72090644e-01 7.83577681e-01 -3.19304138e-01
1.02467585e+00 -1.02436775e-02 6.88628793e-01 1.04367554e+00
4.53257173e-01 7.32592285e-01 1.55323970e+00 -2.85534799e-01
8.98713112e-01 -2.06427038e-01 -2.94751912e-01 7.57500112e-01
-1.60726547e-01 7.31822550e-01 -4.79887456e-01 -9.23667997e-02
6.79888070e-01 -2.73532003e-01 -6.53553456e-02 -2.51917124e-01
-9.45999205e-01 8.66730094e-01 1.63317293e-01 1.42496794e-01
-3.59249949e-01 6.10528529e-01 4.17924643e-01 7.14421332e-01
1.88280955e-01 4.56730455e-01 -3.68299812e-01 -6.27855837e-01
-4.06939715e-01 5.08824646e-01 5.15624404e-01 6.48509622e-01
6.39249325e-01 6.81426346e-01 -1.80287063e-01 5.91829777e-01
6.40672669e-02 7.93938696e-01 6.88899577e-01 -1.18302953e+00
2.58747846e-01 1.47567481e-01 6.53635502e-01 -6.78654373e-01
-3.46199661e-01 -4.08027947e-01 -5.97170770e-01 6.07200503e-01
5.47620475e-01 -3.83245796e-01 -6.90029681e-01 2.27814078e+00
3.77439708e-01 2.01582704e-02 5.97614236e-02 6.87722206e-01
-8.63996707e-03 3.88210237e-01 -2.17384286e-02 -6.29099250e-01
8.60712945e-01 -4.49723691e-01 -7.87577808e-01 -4.30456907e-01
5.12468696e-01 -5.36657989e-01 1.58217490e+00 4.29474503e-01
-9.51642811e-01 -3.34676504e-01 -9.80665267e-01 8.24503243e-01
-3.36910844e-01 -3.07439923e-01 5.88920414e-01 7.17427671e-01
-1.16720212e+00 7.60163665e-01 -9.73291695e-01 -2.19632134e-01
3.39120924e-02 2.70399600e-01 2.07166255e-01 3.07277501e-01
-1.32565427e+00 9.69327211e-01 3.99289548e-01 3.21323089e-02
-1.14458299e+00 -2.20929191e-01 -6.48712397e-01 -4.57972497e-01
7.37421691e-01 -1.95278853e-01 1.66640210e+00 -9.12703335e-01
-2.02596879e+00 1.43469021e-01 -2.52074242e-01 -7.01442659e-01
8.51118922e-01 -3.92773673e-02 -2.26324454e-01 -4.11033556e-02
2.44369265e-02 2.81604767e-01 1.06089044e+00 -1.02772748e+00
-7.53815174e-01 -3.92391115e-01 6.30803779e-02 4.47061479e-01
-6.06108420e-02 -3.73998769e-02 1.55864581e-01 -4.01030421e-01
-9.65859592e-02 -1.04309523e+00 -5.81091821e-01 -2.81186223e-01
2.57860888e-02 -1.68007493e-01 7.48063862e-01 -4.76234332e-02
1.35630488e+00 -1.92102897e+00 -1.88120101e-02 1.01428844e-01
-3.59071404e-01 2.32659802e-01 -8.08837637e-02 4.66786861e-01
2.68248647e-01 -2.34696418e-02 -3.24043989e-01 7.32646231e-03
1.18744239e-01 3.96184891e-01 -5.04573166e-01 7.54668951e-01
-2.60911316e-01 5.35490155e-01 -1.31225049e+00 -1.65426925e-01
2.99379557e-01 6.81179091e-02 -4.65354294e-01 3.99316214e-02
-3.90726745e-01 4.10817564e-01 -6.18578196e-01 5.07587254e-01
3.03101867e-01 1.95818096e-01 4.93953750e-02 5.47579288e-01
-4.24412519e-01 -9.23489220e-03 -1.52829659e+00 1.19846082e+00
-4.97391641e-01 1.98151156e-01 8.70981663e-02 -1.13731933e+00
1.01661038e+00 1.64551824e-01 5.25134861e-01 -7.61975110e-01
1.74105823e-01 3.08944762e-01 1.74117684e-02 -2.31405899e-01
3.66140306e-01 -3.14786524e-01 -1.57527596e-01 5.04327595e-01
-3.63016665e-01 -5.63014805e-01 3.64408106e-01 -1.33266792e-01
1.31225252e+00 3.18166107e-01 3.11745971e-01 -5.40405869e-01
3.94774944e-01 -1.44666553e-01 9.23245728e-01 1.24604094e+00
-8.48103285e-01 -9.21143293e-02 4.37019944e-01 -4.65357006e-01
-6.89774573e-01 -1.20688570e+00 1.77154154e-01 1.14239180e+00
3.79454374e-01 -7.60754496e-02 -6.41203403e-01 -7.73673892e-01
1.46432640e-02 8.55074584e-01 -6.14758790e-01 -4.40971822e-01
-5.08814037e-01 -8.63735378e-01 3.76340955e-01 4.73615319e-01
8.08879852e-01 -1.30197179e+00 -1.19389367e+00 5.75003862e-01
1.68742537e-01 -7.69294262e-01 -6.04207039e-01 4.72001076e-01
-7.97718942e-01 -9.65334594e-01 -4.23846662e-01 -1.31097972e-01
4.37362731e-01 1.08709000e-01 8.29610705e-01 -3.15049052e-01
-1.66634753e-01 6.99425042e-01 -3.77245486e-01 -4.94378865e-01
-4.30053502e-01 -2.86521196e-01 5.17863870e-01 -2.32209623e-01
5.28655536e-02 -5.92867017e-01 -8.25784981e-01 5.37811875e-01
-1.00342786e+00 -4.08708483e-01 1.70185953e-01 9.32304680e-01
7.83613801e-01 2.63292134e-01 7.40758777e-01 -4.78992462e-01
1.02245855e+00 -2.33551338e-01 -9.38628614e-01 2.36902069e-02
-8.32023621e-01 3.34005505e-01 1.11386037e+00 -8.36117446e-01
-8.92642200e-01 9.56418440e-02 -1.68213561e-01 -4.11889553e-01
-3.93555267e-03 2.58985788e-01 2.64108837e-01 8.36087449e-04
9.86796677e-01 6.18987679e-01 -2.66337476e-04 -9.69671682e-02
2.62299567e-01 2.21269816e-01 1.88741148e-01 -8.95755768e-01
5.71920395e-01 5.78435183e-01 1.80022996e-02 -6.64266348e-01
-6.55398250e-01 -1.34070113e-01 -5.47672063e-02 -4.01912808e-01
4.44635719e-01 -3.86032999e-01 -8.81806374e-01 6.58842981e-01
-5.14302969e-01 -9.90317822e-01 -5.60200810e-01 5.26382804e-01
-1.10929441e+00 3.26278269e-01 -4.30640906e-01 -1.25961816e+00
-1.52941868e-01 -1.18644917e+00 4.87761647e-01 4.06168461e-01
7.02077672e-02 -7.67415047e-01 2.74809450e-01 -2.75587887e-01
6.29560590e-01 2.74129868e-01 3.82208467e-01 -2.30048507e-01
-8.61772746e-02 1.77954696e-02 4.10515130e-01 4.13962185e-01
1.72349557e-01 -7.49005796e-03 -5.35781205e-01 -7.17077017e-01
1.80346221e-01 -2.66033500e-01 8.00300956e-01 5.03358960e-01
1.05980837e+00 -4.83533412e-01 9.64090228e-02 3.88249546e-01
1.39944565e+00 8.69701087e-01 3.69608134e-01 7.06627071e-01
-2.55168267e-02 1.99729219e-01 8.29684794e-01 8.82037997e-01
1.48132384e-01 5.30879676e-01 6.25932992e-01 3.04368526e-01
3.79918098e-01 -2.08484352e-01 9.38487411e-01 3.58829051e-01
-2.65272167e-02 -9.69448239e-02 -8.38937759e-01 5.64822972e-01
-1.91440797e+00 -1.28203678e+00 4.97057289e-01 2.41346955e+00
9.75311875e-01 3.64091098e-01 2.79797375e-01 -4.06827889e-02
6.75712764e-01 2.64722437e-01 -1.09935391e+00 -7.94660032e-01
2.42317952e-02 2.13370413e-01 6.41641200e-01 6.66185319e-01
-9.74500060e-01 1.20523882e+00 6.39913940e+00 9.21233296e-01
-1.15550244e+00 -3.36977243e-02 4.59998906e-01 -1.70198977e-01
-1.92800444e-02 -4.73839603e-02 -7.08998859e-01 8.42761815e-01
9.82257843e-01 -4.51462388e-01 6.73920214e-01 1.03299415e+00
6.32524669e-01 -4.86764967e-01 -6.94997668e-01 7.80237675e-01
-4.33541894e-01 -9.19477046e-01 -4.54425067e-01 -1.64398909e-01
8.78221571e-01 5.23127383e-03 4.23512131e-01 5.86629093e-01
9.44447517e-01 -8.19708407e-01 6.62533462e-01 2.10028872e-01
5.33593535e-01 -9.99622941e-01 4.18107957e-01 6.06080770e-01
-1.22430193e+00 -4.74005669e-01 -2.38162160e-01 -3.51368278e-01
-4.77455929e-02 2.90027410e-01 -5.86376905e-01 1.72158018e-01
8.29907060e-01 2.40644157e-01 -1.92288876e-01 7.17119873e-01
-3.74551326e-01 4.73371148e-01 -5.52480936e-01 -4.74757195e-01
5.73850632e-01 -2.31523931e-01 6.31890833e-01 9.72244978e-01
5.20131588e-01 1.29355922e-01 3.44136059e-01 5.25455475e-01
5.60761571e-01 -8.64629298e-02 -8.36783051e-01 6.03278354e-02
5.38693905e-01 6.09865427e-01 -7.74798274e-01 -1.73903167e-01
7.34247938e-02 9.43011642e-01 2.27365613e-01 6.06081605e-01
-8.67738247e-01 -4.96902406e-01 8.52592170e-01 -2.01153323e-01
4.01449621e-01 -7.66347289e-01 9.12772492e-02 -1.14539063e+00
3.24689560e-02 -9.95366037e-01 6.01075888e-01 -2.74282038e-01
-1.03694475e+00 5.57464480e-01 8.90614316e-02 -1.26763237e+00
-5.55757761e-01 -3.47008348e-01 -5.21944404e-01 4.15419877e-01
-1.41114831e+00 -3.03535044e-01 1.49961933e-01 8.45042467e-01
7.71915913e-01 -5.70520759e-02 7.26648986e-01 -4.12294894e-01
-6.07466221e-01 5.25990427e-01 5.89911938e-01 -2.36576125e-01
5.59299886e-01 -1.42918825e+00 -2.16132123e-02 1.01020634e+00
-2.93151811e-02 5.21656752e-01 1.25956476e+00 -5.11559188e-01
-1.48850775e+00 -8.50424886e-01 1.20138839e-01 -4.67806794e-02
7.37633049e-01 -1.54355377e-01 -5.41567385e-01 4.74028587e-01
-1.20007813e-01 1.50889739e-01 2.03593075e-01 -1.39407396e-01
-1.79911613e-01 -2.38899752e-01 -1.40210521e+00 1.00945759e+00
1.10468566e+00 -4.23847258e-01 -3.86423409e-01 1.40972823e-01
3.54117304e-01 -4.62990314e-01 -5.20943046e-01 4.58152384e-01
3.73314768e-01 -1.12320411e+00 7.25635409e-01 -4.72276568e-01
-1.99264541e-01 -3.64423841e-01 -2.69594133e-01 -1.63191354e+00
1.51558602e-02 -1.10414720e+00 -3.04617316e-01 5.04610956e-01
2.12624654e-01 -1.10019898e+00 6.23067141e-01 3.64131451e-01
5.69729283e-02 -9.07247961e-01 -1.39184237e+00 -1.42446482e+00
4.05257970e-01 -6.81383967e-01 3.08897793e-01 5.26378751e-01
8.51898193e-02 -2.91159600e-01 -2.87586033e-01 1.18321525e-02
6.20115638e-01 1.55383557e-01 4.64603603e-01 -3.31556499e-01
-5.15804231e-01 -7.37293482e-01 -1.82595044e-01 -1.03926766e+00
1.16472371e-01 -2.07703233e-01 4.29880142e-01 -1.17210495e+00
-3.02793622e-01 -4.28128630e-01 -7.68112540e-01 3.70462775e-01
-2.06615925e-01 -1.26426801e-01 7.83804432e-02 -5.71728721e-02
-7.88387716e-01 1.01727855e+00 1.18063653e+00 -3.11968662e-02
-6.62786543e-01 3.29123139e-01 -3.51593763e-01 8.37613821e-01
1.27349555e+00 -3.25864226e-01 -6.31727397e-01 -1.11173868e-01
1.90386578e-01 1.40412495e-01 1.18382327e-01 -1.08993971e+00
2.02974696e-02 -8.43411624e-01 6.15055449e-02 -2.63855994e-01
2.36846820e-01 -4.31882858e-01 -1.86331317e-01 8.35434794e-01
-5.43107688e-01 3.86666268e-01 2.84836203e-01 6.97333038e-01
6.10691868e-02 -1.06793605e-01 1.05035877e+00 -4.26848233e-01
-8.39139462e-01 3.05816531e-01 -8.81413817e-01 2.78097272e-01
1.21606171e+00 -1.64594054e-01 -1.20710038e-01 -6.04164898e-01
-5.82461894e-01 4.36447740e-01 4.28084850e-01 2.10169271e-01
5.70299208e-01 -1.04950917e+00 -3.42670798e-01 6.32872013e-03
-2.18407467e-01 -4.01953906e-01 2.36881942e-01 5.78049302e-01
-2.03027129e-01 7.20808133e-02 -1.20243169e-01 -4.70734477e-01
-1.04068220e+00 6.04090393e-01 6.21341765e-01 -2.86378801e-01
-4.34437633e-01 8.93506527e-01 -3.24956656e-01 -2.29163736e-01
3.81160647e-01 -3.30760956e-01 1.18448898e-01 -2.25179717e-01
6.14260972e-01 2.94278920e-01 -2.30471343e-01 -8.67519602e-02
-3.49991471e-01 5.47665656e-01 3.52102369e-02 -7.03938484e-01
1.10499322e+00 -1.24124356e-01 4.51713771e-01 6.44848645e-01
7.70531178e-01 -3.62420559e-01 -1.85510659e+00 -2.08427727e-01
-4.88426499e-02 -5.68604231e-01 -1.14209816e-01 -7.24515736e-01
-8.30619693e-01 6.31057918e-01 8.53415489e-01 2.40016475e-01
1.15290654e+00 -3.65049154e-01 3.79606158e-01 6.60553455e-01
8.81946027e-01 -1.68921256e+00 2.00204074e-01 9.58545327e-01
7.59676933e-01 -1.12627816e+00 -1.07776307e-01 2.99908519e-01
-7.99761832e-01 8.47151875e-01 8.09424520e-01 -2.53487110e-01
5.70458293e-01 3.78700078e-01 7.33422488e-02 1.90795526e-01
-9.11925077e-01 -4.10099387e-01 -3.92778248e-01 5.82422495e-01
-4.24839742e-02 6.87354729e-02 -6.88579261e-01 6.83154687e-02
-1.37962133e-01 -1.09394304e-01 4.11450416e-01 1.37563229e+00
-6.67612195e-01 -1.08637452e+00 -3.12680751e-01 1.08710885e-01
-3.88820767e-01 1.77447721e-01 -9.39346403e-02 8.30611050e-01
-9.39571392e-03 1.00110006e+00 -1.74237028e-01 -3.70916426e-01
2.86599904e-01 1.13597684e-01 4.54997540e-01 -1.59722432e-01
-4.74280626e-01 -1.20324017e-02 -1.50954172e-01 -9.50476885e-01
-2.02805713e-01 -8.32609177e-01 -1.50130761e+00 -1.62399307e-01
8.23759362e-02 2.30044156e-01 5.35424471e-01 7.24335194e-01
2.21339256e-01 2.83521116e-01 1.02862084e+00 -6.17592216e-01
-1.29466736e+00 -7.19769597e-01 -4.97750938e-01 2.39868790e-01
2.71660537e-01 -9.40422773e-01 -7.49207616e-01 -3.40181112e-01] | [4.0578718185424805, 2.2182810306549072] |
f78ae9a9-38c7-46d8-ada5-1cffc3ac6542 | learning-in-a-single-domain-for-non | 2305.06200 | null | https://arxiv.org/abs/2305.06200v1 | https://arxiv.org/pdf/2305.06200v1.pdf | Learning in a Single Domain for Non-Stationary Multi-Texture Synthesis | This paper aims for a new generation task: non-stationary multi-texture synthesis, which unifies synthesizing multiple non-stationary textures in a single model. Most non-stationary textures have large scale variance and can hardly be synthesized through one model. To combat this, we propose a multi-scale generator to capture structural patterns of various scales and effectively synthesize textures with a minor cost. However, it is still hard to handle textures of different categories with different texture patterns. Therefore, we present a category-specific training strategy to focus on learning texture pattern of a specific domain. Interestingly, once trained, our model is able to produce multi-pattern generations with dynamic variations without the need to finetune the model for different styles. Moreover, an objective evaluation metric is designed for evaluating the quality of texture expansion and global structure consistency. To our knowledge, ours is the first scheme for this challenging task, including model, training, and evaluation. Experimental results demonstrate the proposed method achieves superior performance and time efficiency. The code will be available after the publication. | ['Zhen Zhu', 'Zhiliang Xu', 'Zijie Wu', 'Xudong Xie'] | 2023-05-10 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 6.26228034e-01 -3.31687629e-01 -3.74056511e-02 -9.34331398e-03
-7.84691155e-01 -5.82615793e-01 4.48648095e-01 -2.60653794e-01
2.18431681e-01 6.28011763e-01 -3.09481651e-01 2.20250543e-02
1.68656521e-02 -7.36217499e-01 -7.27250099e-01 -1.04134774e+00
3.47915292e-01 4.25461948e-01 5.12216747e-01 -3.57239008e-01
1.97735384e-01 4.94182467e-01 -1.79402983e+00 5.00384390e-01
9.40441370e-01 1.02061832e+00 6.00842595e-01 7.46905386e-01
1.51106477e-01 4.58272308e-01 -7.22727001e-01 -5.11485279e-01
1.27642170e-01 -7.45064616e-01 -4.88735437e-01 5.73607385e-01
6.92223608e-01 1.06724717e-01 1.57286629e-01 1.21494853e+00
5.64868808e-01 1.50294087e-04 5.30480802e-01 -8.66959810e-01
-6.67227447e-01 5.12733757e-01 -5.98147511e-01 -8.34136233e-02
6.47003129e-02 3.39094289e-02 8.03138614e-01 -6.70913637e-01
6.91596091e-01 1.36424088e+00 4.44022954e-01 5.51592886e-01
-1.18965089e+00 -3.98067087e-01 1.93283245e-01 -8.23921561e-02
-1.33123493e+00 -4.32013363e-01 9.93232667e-01 -1.97627679e-01
2.84759402e-01 5.88541090e-01 4.09305781e-01 1.39967644e+00
2.58354843e-01 7.23633051e-01 1.43752766e+00 -5.84981978e-01
1.81232825e-01 -3.60661820e-02 -5.22357702e-01 5.28589249e-01
2.57698685e-01 -1.28195897e-01 -4.08108175e-01 7.52898753e-02
1.03550160e+00 -3.82452458e-01 -2.16917157e-01 -4.64218289e-01
-1.32679880e+00 4.48503554e-01 -7.53495544e-02 6.02426291e-01
-1.28740549e-01 1.49019007e-02 1.90861106e-01 3.97833735e-01
5.91335654e-01 5.27630925e-01 -4.66995537e-01 -2.11518869e-01
-8.55179071e-01 3.37774694e-01 6.45631373e-01 1.12270248e+00
6.50253534e-01 2.69508064e-01 -2.93946177e-01 1.23416376e+00
-3.02474856e-01 8.89006913e-01 6.87614620e-01 -6.62581265e-01
2.70759255e-01 2.12480053e-01 3.82984802e-02 -1.23331821e+00
-1.54772416e-01 -4.50852841e-01 -1.14976907e+00 8.76248479e-02
4.10748124e-01 -1.20119929e-01 -8.43731463e-01 1.65665376e+00
4.17915732e-01 2.79765338e-01 -3.73139977e-02 6.75919056e-01
6.02565110e-01 5.51400602e-01 -3.40973854e-01 -3.34952980e-01
1.28461826e+00 -1.28188992e+00 -6.55901551e-01 -4.13785353e-02
1.24164663e-01 -1.32704699e+00 1.25156891e+00 4.78303134e-01
-1.15372646e+00 -8.84236932e-01 -1.09003282e+00 4.03791428e-01
-7.38783274e-03 4.30053711e-01 5.82672596e-01 7.11776257e-01
-9.11706507e-01 6.55485928e-01 -8.44972908e-01 -4.64011252e-01
1.45981655e-01 1.34360045e-01 -1.27900064e-01 -3.60323973e-02
-8.16365719e-01 6.42665803e-01 3.08903933e-01 4.39419085e-03
-6.00641429e-01 -4.52678591e-01 -6.38344944e-01 -5.03050387e-02
5.51115215e-01 -5.80702543e-01 1.04168200e+00 -1.25862002e+00
-2.07238460e+00 5.39248943e-01 -3.65212679e-01 -5.32990471e-02
5.90504587e-01 8.29474777e-02 -7.34175980e-01 -4.07883674e-02
3.11517343e-02 3.74503851e-01 1.39745224e+00 -1.56204581e+00
-6.87810063e-01 5.81474155e-02 -1.60714284e-01 9.54678655e-02
-5.04565597e-01 -1.35337472e-01 -7.45655477e-01 -1.22040224e+00
1.82976395e-01 -1.14831531e+00 -7.06362799e-02 -4.16752189e-01
-3.38416308e-01 1.06134623e-01 7.95879543e-01 -3.31422180e-01
1.25560689e+00 -2.20658112e+00 3.44335556e-01 2.31625199e-01
-1.15016021e-01 2.12028354e-01 -5.82684040e-01 1.62511125e-01
1.61339492e-01 1.34000525e-01 -2.22184733e-01 -2.61023194e-01
-9.63278934e-02 1.85294271e-01 -3.31514984e-01 2.82690153e-02
4.15476620e-01 6.86227381e-01 -7.64601469e-01 -3.99807721e-01
2.06620842e-02 2.37566128e-01 -6.50124013e-01 -1.35576224e-03
-3.18313450e-01 5.86061835e-01 -4.45776492e-01 8.79425645e-01
8.92711103e-01 -4.50026989e-01 3.71863633e-01 -2.87320912e-01
1.38417473e-02 -3.83708537e-01 -1.43750870e+00 1.55362320e+00
-7.29055822e-01 2.99597800e-01 -1.59927085e-01 -9.45346177e-01
1.18391097e+00 9.34236497e-02 2.75864035e-01 -7.15974092e-01
4.64532338e-02 5.58085322e-01 7.56375641e-02 -3.18924457e-01
6.78358912e-01 1.74802903e-03 -1.44607916e-01 1.21263690e-01
-6.91991821e-02 -4.08706665e-01 6.49711072e-01 -5.55263519e-01
8.24012935e-01 2.98053294e-01 4.31974605e-02 -2.79973805e-01
6.30098522e-01 -3.20294112e-01 6.43043339e-01 8.24288905e-01
2.46092886e-01 9.50508952e-01 4.37815934e-01 -4.60981220e-01
-1.35526729e+00 -8.43627512e-01 -1.27457544e-01 1.02258646e+00
4.10703480e-01 -2.98759043e-01 -6.64447606e-01 -3.73941332e-01
-1.91790015e-01 2.22974628e-01 -6.43307924e-01 5.38299866e-02
-8.38359773e-01 -9.97442544e-01 2.83562958e-01 9.89743620e-02
6.03492975e-01 -1.07824481e+00 -3.78486753e-01 2.11838305e-01
-2.75936604e-01 -1.23047018e+00 -5.61703444e-01 -1.54521734e-01
-7.90338635e-01 -8.76446068e-01 -1.00623024e+00 -1.12401128e+00
7.08383739e-01 1.90865353e-01 1.08597767e+00 1.62444830e-01
-9.69042927e-02 -9.21378136e-02 -5.27900815e-01 -1.65187135e-01
-7.75863171e-01 4.06666249e-01 -1.66599751e-01 4.89907414e-01
-2.39336267e-01 -6.45906329e-01 -4.26315516e-01 7.55595505e-01
-1.05423164e+00 4.18734193e-01 8.03929746e-01 1.06457162e+00
1.02751243e+00 5.40396869e-01 5.30975819e-01 -8.90637875e-01
5.59565127e-01 -1.24512918e-01 -6.07781768e-01 5.98934293e-01
-2.52460092e-01 8.60626176e-02 8.82168174e-01 -9.15813684e-01
-1.21022987e+00 6.05752580e-02 3.69926803e-02 -2.84826517e-01
-2.31281474e-01 2.18738317e-01 -2.98473209e-01 -2.18025386e-01
5.45878708e-01 5.20193458e-01 -2.69995302e-01 -5.83170652e-01
2.55429089e-01 2.78765053e-01 4.68779325e-01 -9.08416986e-01
8.99450243e-01 3.99909228e-01 1.84959173e-01 -9.45098877e-01
-7.11535215e-01 2.83291135e-02 -5.04385829e-01 -1.11866072e-01
5.42722344e-01 -7.48683155e-01 -3.20115566e-01 9.49215293e-01
-9.44048822e-01 -6.05864048e-01 -2.03780979e-01 1.16036735e-01
-7.03135729e-01 3.98907155e-01 -4.59425956e-01 -4.33436662e-01
-1.35599017e-01 -1.27872717e+00 1.38689733e+00 1.81201056e-01
-1.40125453e-02 -9.27654862e-01 -1.04952313e-01 1.93672851e-02
6.43132448e-01 3.61241579e-01 8.31637800e-01 -5.11134416e-02
-5.77082157e-01 3.73232216e-02 -8.07645731e-04 2.24462911e-01
6.55395091e-01 2.26139665e-01 -6.51053131e-01 -4.65541005e-01
-7.11270422e-02 -3.56089205e-01 7.32911885e-01 3.36959690e-01
1.43733871e+00 -2.30354309e-01 -7.28495568e-02 7.16927409e-01
1.41321778e+00 3.56737286e-01 6.24544084e-01 4.67992574e-01
6.98021531e-01 3.57232958e-01 6.09773040e-01 4.81999934e-01
1.01935036e-01 9.73448157e-01 -4.44569662e-02 -1.99847296e-01
-4.63979304e-01 -9.71395075e-02 9.98661220e-02 1.03708220e+00
-2.69685328e-01 -5.57538629e-01 -5.94729662e-01 3.37403893e-01
-1.78326249e+00 -1.02869904e+00 7.57158697e-02 2.05396032e+00
9.06471372e-01 2.00883299e-01 -1.00730218e-01 1.73951253e-01
8.86099100e-01 2.31278643e-01 -4.71069694e-01 -1.47403196e-01
-6.13102734e-01 4.46938336e-01 3.04680616e-01 3.11577678e-01
-1.29475772e+00 1.27616608e+00 6.07623053e+00 1.44412935e+00
-1.50696719e+00 -1.59801900e-01 7.31748641e-01 1.13854490e-01
-4.13691342e-01 -1.15531079e-01 -8.12883258e-01 6.69363439e-01
4.39219385e-01 -1.12155333e-01 5.34313321e-01 7.31429815e-01
7.39404783e-02 3.00702751e-02 -6.12622023e-01 9.76198196e-01
1.24113388e-01 -1.28952301e+00 2.89884716e-01 -2.27136552e-01
1.27789307e+00 -7.08615422e-01 3.50582957e-01 5.10098487e-02
-1.85686431e-03 -7.24333405e-01 9.25334036e-01 5.34602106e-01
9.31675732e-01 -7.56331086e-01 5.72958350e-01 -1.86024234e-02
-1.39824522e+00 1.19670883e-01 -4.61889058e-01 3.63475680e-01
7.23211020e-02 5.38602889e-01 -4.06775177e-01 7.79903531e-01
4.57418799e-01 6.69742644e-01 -8.64703655e-01 9.77616727e-01
-1.48353735e-02 3.92178029e-01 -9.32520553e-02 -4.87302504e-02
-8.61120149e-02 -1.29044786e-01 5.15468538e-01 1.04045701e+00
9.96581316e-01 -2.08263949e-01 3.50294292e-01 5.23799777e-01
1.88257173e-01 3.24753582e-01 -3.62223029e-01 -4.46233712e-03
3.84491026e-01 1.19731557e+00 -1.15452421e+00 -3.28870952e-01
-1.74782768e-01 1.21075654e+00 1.05772115e-01 3.01253736e-01
-8.48348141e-01 -3.25882554e-01 3.57620806e-01 -6.55894727e-02
6.49133086e-01 -1.59168318e-01 -3.98395032e-01 -1.44444752e+00
1.46621957e-01 -1.42658901e+00 -2.62364764e-02 -6.06193125e-01
-1.25867033e+00 1.01116478e+00 -1.61435172e-01 -1.50829816e+00
-1.87388510e-01 -5.09906948e-01 -3.12952936e-01 5.90070367e-01
-1.24955738e+00 -1.44821358e+00 -3.64120960e-01 5.03222406e-01
7.54559577e-01 -4.32685822e-01 7.04765797e-01 2.98413783e-01
-6.55632973e-01 8.96887779e-01 2.39806980e-01 -4.59633112e-01
8.91455054e-01 -1.22871494e+00 4.52819824e-01 1.01245761e+00
-4.31165732e-02 2.82697380e-01 7.50762343e-01 -6.45520210e-01
-1.36641014e+00 -1.10180080e+00 7.17712402e-01 -1.21731207e-01
6.64194465e-01 -3.54801595e-01 -9.06087816e-01 5.45199886e-02
6.41628876e-02 -1.68218225e-01 3.75986725e-01 -8.62189457e-02
1.40213629e-03 -3.74148041e-01 -7.22199321e-01 8.61758351e-01
1.17770112e+00 -1.44467935e-01 1.07110828e-01 3.18559647e-01
5.67530334e-01 -7.89443195e-01 -8.92991245e-01 7.04050720e-01
5.45003951e-01 -8.45030606e-01 7.19103754e-01 -1.70913160e-01
7.00964987e-01 -4.17904645e-01 -2.62359411e-01 -1.36498785e+00
-5.02626956e-01 -8.06424558e-01 2.25554761e-02 1.41805446e+00
4.85843539e-01 -4.27097529e-01 7.10637510e-01 -9.00771916e-02
-1.48173735e-01 -6.13406777e-01 -5.86800098e-01 -1.01165187e+00
-6.21246435e-02 -4.74298112e-02 9.73168731e-01 8.26382756e-01
-6.05841398e-01 2.08575293e-01 -8.71164858e-01 3.56194794e-01
5.19875705e-01 6.96459770e-01 1.04406869e+00 -9.82217431e-01
-5.99817514e-01 -5.98463058e-01 -2.02119887e-01 -1.09487545e+00
1.94140505e-02 -4.84718204e-01 1.79097250e-01 -9.94143903e-01
1.53904587e-01 -6.88488483e-01 -5.82657158e-02 2.11580604e-01
-4.12105590e-01 4.22944129e-01 2.02466652e-01 1.99717879e-01
-5.13679266e-01 6.13210142e-01 2.02262402e+00 -1.74252301e-01
-1.93524897e-01 9.58604217e-02 -7.12885737e-01 4.81017053e-01
9.87598419e-01 -2.38539904e-01 -6.53326213e-01 -3.67602855e-01
8.28963798e-03 -2.02889033e-02 9.16964784e-02 -1.19429743e+00
-1.16642840e-01 -5.03684700e-01 3.29126030e-01 -2.76052922e-01
9.03615355e-02 -5.38990080e-01 6.21143878e-01 2.23479480e-01
-1.52632371e-01 1.32161111e-01 2.36868277e-01 4.09311652e-01
-4.25770074e-01 -5.80113865e-02 1.01745582e+00 5.66690937e-02
-6.77396238e-01 4.36122686e-01 -2.25664660e-01 -2.86974572e-02
8.16161275e-01 -6.11789450e-02 -2.46756986e-01 -1.47641614e-01
-6.90733552e-01 -2.89128155e-01 7.96798825e-01 7.01830924e-01
4.38145965e-01 -1.58508658e+00 -7.28941500e-01 3.78119707e-01
1.01497419e-01 -3.34076136e-02 5.48664272e-01 5.19223332e-01
-6.00520849e-01 4.24405485e-02 -3.64346772e-01 -6.51144803e-01
-1.23365891e+00 5.02025187e-01 2.55346417e-01 -5.03024340e-01
-4.01298016e-01 6.25422537e-01 4.23773468e-01 -2.36373216e-01
-1.47474855e-01 -3.08944374e-01 -8.00848827e-02 -1.32430077e-01
5.24821818e-01 4.26960364e-02 2.31835410e-01 -5.70440590e-01
3.83821055e-02 9.30665076e-01 -1.02160824e-02 -2.07584407e-02
1.11420929e+00 -1.05337977e-01 -1.45323515e-01 3.82172287e-01
9.49046075e-01 1.75068453e-01 -1.36351430e+00 -1.82464778e-01
-1.94058806e-01 -6.21021926e-01 -3.70279372e-01 -5.61876655e-01
-1.24697578e+00 4.03145373e-01 6.75248682e-01 3.07482570e-01
1.44036317e+00 -3.48914057e-01 8.69488955e-01 2.01295063e-01
4.97578591e-01 -1.12794602e+00 5.06908000e-01 5.88678002e-01
1.16358829e+00 -1.10529315e+00 -2.57668912e-01 -5.65301716e-01
-6.48953497e-01 1.22263801e+00 7.37995148e-01 5.42548578e-03
4.83440638e-01 3.67035061e-01 5.43743372e-02 1.96548447e-01
-7.59168863e-01 -1.68626711e-01 4.35663462e-01 4.74639326e-01
3.44461471e-01 1.30001009e-01 -3.53997588e-01 2.01839209e-01
-3.95434946e-01 -8.46205875e-02 3.00767183e-01 8.00602317e-01
-3.27559292e-01 -1.65572190e+00 -2.99885243e-01 2.06676647e-01
-4.86145973e-01 -1.41473999e-03 -4.98331562e-02 5.84102273e-01
2.94427186e-01 8.24190795e-01 -4.10072133e-02 -4.65383589e-01
4.10528481e-01 -2.27951303e-01 7.13356137e-01 -3.41850460e-01
-2.33705223e-01 5.03108323e-01 -2.78286021e-02 -2.78542787e-01
-6.10729337e-01 -6.05720997e-01 -4.14417595e-01 -3.38410646e-01
-1.98665828e-01 -6.07501641e-02 3.33137155e-01 5.78741014e-01
3.41390848e-01 6.63070202e-01 9.74628925e-01 -8.42940331e-01
-3.66027832e-01 -8.10799658e-01 -6.65320754e-01 6.12429142e-01
5.28035797e-02 -6.77030087e-01 -8.53633434e-02 4.02605057e-01] | [11.524518013000488, -0.6496699452400208] |
dda7b8dd-56bf-4993-86a1-217e9d4e3977 | topological-data-analysis-guided-segment | 2306.17400 | null | https://arxiv.org/abs/2306.17400v1 | https://arxiv.org/pdf/2306.17400v1.pdf | Topological Data Analysis Guided Segment Anything Model Prompt Optimization for Zero-Shot Segmentation in Biological Imaging | Emerging foundation models in machine learning are models trained on vast amounts of data that have been shown to generalize well to new tasks. Often these models can be prompted with multi-modal inputs that range from natural language descriptions over images to point clouds. In this paper, we propose topological data analysis (TDA) guided prompt optimization for the Segment Anything Model (SAM) and show preliminary results in the biological image segmentation domain. Our approach replaces the standard grid search approach that is used in the original implementation and finds point locations based on their topological significance. Our results show that the TDA optimized point cloud is much better suited for finding small objects and massively reduces computational complexity despite the extra step in scenarios which require many segmentations. | ['Shusen Liu', 'Ruben Glatt'] | 2023-06-30 | null | null | null | null | ['zero-shot-segmentation', 'topological-data-analysis'] | ['computer-vision', 'graphs'] | [ 1.17822565e-01 1.55724529e-02 2.09745049e-01 -4.26481366e-01
-6.23130500e-01 -8.02675784e-01 8.53904009e-01 5.88005602e-01
-4.97948736e-01 6.23172581e-01 -2.82499075e-01 -3.78679484e-01
-3.90559733e-01 -7.65254498e-01 -7.16709197e-01 -5.26817501e-01
-2.12350264e-01 1.18610871e+00 6.57271802e-01 -2.31270775e-01
7.80258834e-01 1.00877082e+00 -1.39925563e+00 1.19007841e-01
7.27972269e-01 8.74624848e-01 5.46074152e-01 7.98985660e-01
-5.77600777e-01 -2.14971006e-02 -4.09484327e-01 -4.89794314e-02
3.21310014e-01 5.99843897e-02 -1.08169436e+00 -5.11962995e-02
4.85664785e-01 2.53357351e-01 3.08837414e-01 7.06322610e-01
3.20893228e-01 2.25254372e-01 8.03207099e-01 -1.14112687e+00
-1.48611128e-01 1.44345060e-01 -5.39093733e-01 2.33900562e-01
1.40474871e-01 6.30153120e-02 9.53131795e-01 -1.02552569e+00
8.37041080e-01 1.24528670e+00 1.00774169e+00 3.00014973e-01
-1.66162694e+00 2.93455720e-02 -2.69415993e-02 7.49650821e-02
-1.25002992e+00 -1.11017317e-01 6.59968972e-01 -5.88121593e-01
1.15656233e+00 3.97499830e-01 7.79022694e-01 4.19009298e-01
1.31238997e-02 5.54428399e-01 1.12024915e+00 -5.07477403e-01
5.23267150e-01 4.51090075e-02 3.47867198e-02 6.96641505e-01
3.56474519e-02 -5.60614131e-02 -3.90250564e-01 -1.59397826e-01
1.03109145e+00 -2.20329121e-01 2.56381571e-01 -7.78223932e-01
-1.41302299e+00 7.72433043e-01 6.46954417e-01 3.30015481e-01
-4.11733717e-01 1.66981548e-01 2.99957484e-01 7.74885702e-04
4.36049253e-01 1.05251777e+00 -7.36845136e-01 -4.94100228e-02
-1.27609670e+00 4.31725204e-01 5.73535025e-01 8.90068352e-01
9.81379271e-01 -4.17417765e-01 2.12689593e-01 8.65220428e-01
1.47922963e-01 4.67586927e-02 2.29409054e-01 -1.09224880e+00
1.53274566e-01 8.89037251e-01 6.94611594e-02 -1.24335194e+00
-7.66224623e-01 -3.54946762e-01 -5.33327103e-01 5.18624723e-01
5.99120378e-01 3.78997236e-01 -1.05549371e+00 1.21141219e+00
4.78552312e-01 -4.42821793e-02 -2.15238109e-01 7.27383912e-01
5.49905241e-01 7.58689284e-01 -6.32785680e-03 1.51013747e-01
1.19004488e+00 -6.54669404e-01 -7.70087168e-02 -1.95467234e-01
9.08363461e-01 -6.17344558e-01 1.13898647e+00 5.30212343e-01
-1.04829359e+00 -5.29716253e-01 -9.08718884e-01 -2.91058958e-01
-9.41787004e-01 -6.76365271e-02 7.30255842e-01 2.36895442e-01
-1.41684926e+00 8.64910722e-01 -8.31570148e-01 -8.51644039e-01
9.22961831e-01 6.22649491e-01 -3.04376155e-01 2.39224628e-01
-3.93368632e-01 1.04961193e+00 6.03890955e-01 -9.82764587e-02
-5.38977861e-01 -6.72632277e-01 -5.19711852e-01 -1.96595803e-01
2.63658851e-01 -7.60230601e-01 1.06223369e+00 -4.68030602e-01
-1.21223569e+00 1.21336889e+00 -2.00177833e-01 -6.31318510e-01
5.17031848e-01 1.20723642e-01 5.16312599e-01 2.98541725e-01
2.02278927e-01 1.34974325e+00 2.62696266e-01 -1.41208088e+00
-6.12063944e-01 -5.52758098e-01 -2.86168922e-02 9.63358730e-02
1.67610228e-01 -2.16668636e-01 -3.23346704e-01 -4.08947796e-01
6.72791779e-01 -7.67731428e-01 -7.42183924e-01 3.39057624e-01
-4.62366402e-01 -2.82753885e-01 9.06105757e-01 -2.66409725e-01
7.80206680e-01 -1.64798439e+00 1.88823178e-01 4.24986988e-01
1.17008142e-01 1.29690066e-01 6.89731389e-02 5.03095508e-01
1.21045180e-01 4.40950006e-01 -4.29814965e-01 -2.59953707e-01
-2.62760893e-02 4.33074772e-01 -2.36811861e-01 3.74774665e-01
2.63824135e-01 1.04302812e+00 -8.07059467e-01 -9.31818902e-01
5.35898328e-01 7.80770257e-02 -4.60491449e-01 -1.67401329e-01
-6.79092526e-01 4.96263742e-01 -4.79273587e-01 6.49297893e-01
6.64189041e-01 -5.17886877e-01 -2.88863599e-01 -2.08393008e-01
-5.34420073e-01 2.08147988e-01 -9.63570178e-01 1.99231434e+00
-2.51337379e-01 7.26662993e-01 -9.30735171e-02 -1.22887325e+00
1.01198506e+00 -1.17543496e-01 6.60265982e-01 -3.62711042e-01
3.46550681e-02 1.65550232e-01 -1.39527678e-01 -3.86401743e-01
3.42006147e-01 -9.22094882e-02 -1.90364458e-02 2.85621136e-01
5.72671555e-02 -8.47780228e-01 2.04315498e-01 2.33966067e-01
1.03428376e+00 2.86883056e-01 5.57577372e-01 -6.67057037e-01
3.66098881e-01 9.61679935e-01 2.05335885e-01 8.01530540e-01
-1.07307546e-01 7.92477667e-01 3.43609959e-01 -8.16123247e-01
-1.24185562e+00 -1.00859785e+00 -4.08526629e-01 9.68660831e-01
4.38604176e-01 -3.80925626e-01 -7.03923345e-01 -3.55364889e-01
-2.29190402e-02 5.90258360e-01 -6.22135818e-01 4.55057025e-01
-6.80215001e-01 -5.02682388e-01 2.70110101e-01 5.07361710e-01
3.56943220e-01 -1.03981256e+00 -1.01785576e+00 1.91592693e-01
1.77720696e-01 -9.15088713e-01 -6.78662509e-02 1.94519728e-01
-1.36242294e+00 -9.24791396e-01 -7.00139284e-01 -8.69688272e-01
9.37698364e-01 1.02689132e-01 1.08450651e+00 2.42592975e-01
-6.84404671e-01 4.33759123e-01 -1.60249993e-01 -6.37829423e-01
-2.32480858e-02 1.73889816e-01 -5.66855706e-02 -4.90798950e-01
1.30724981e-01 -6.72597110e-01 -4.01233554e-01 4.70779210e-01
-7.43744850e-01 5.33068955e-01 5.30228257e-01 6.58367395e-01
9.29811656e-01 -2.01286688e-01 4.61811751e-01 -6.75323904e-01
4.10759658e-01 -2.01181993e-01 -7.66977131e-01 2.29273081e-01
-4.81479168e-01 1.63217455e-01 4.42184567e-01 -2.89085954e-01
-5.87094009e-01 4.62404549e-01 -2.62445416e-02 -3.56678754e-01
-6.18534386e-01 5.27411103e-01 2.12926939e-01 -5.85604787e-01
7.62323380e-01 1.58892751e-01 7.89327621e-02 -5.13954818e-01
4.19827819e-01 1.39523908e-01 6.30277753e-01 -7.57808030e-01
7.32875764e-01 7.79906809e-01 6.15008175e-01 -1.00600493e+00
-4.09714222e-01 -5.21026373e-01 -1.36485028e+00 -3.07469785e-01
8.44690084e-01 -7.25569203e-02 -6.43936098e-01 8.86356682e-02
-1.38889909e+00 -2.80444592e-01 -3.53008419e-01 2.99846500e-01
-1.00688446e+00 1.79578260e-01 -1.34969518e-01 -6.38407171e-01
-2.76009124e-02 -9.22089756e-01 1.23959935e+00 3.06688398e-01
-2.69005060e-01 -1.29906762e+00 1.63173616e-01 6.65228888e-02
4.35092539e-01 5.42671859e-01 1.05994570e+00 -6.77147210e-01
-9.65636790e-01 -6.80176690e-02 -2.84418076e-01 -3.59678447e-01
-1.16365954e-01 9.35958624e-02 -7.83600807e-01 3.65224667e-02
-3.29571635e-01 -2.35054120e-01 7.05153644e-01 7.07051337e-01
1.32950664e+00 -5.65232895e-02 -8.12409818e-01 7.09676147e-01
1.54916823e+00 9.21757147e-02 3.20742190e-01 5.89539707e-01
5.23773015e-01 8.59273612e-01 7.71515012e-01 5.70082068e-02
2.85828948e-01 7.54392147e-01 6.60190880e-01 -3.46131772e-01
5.00841923e-02 -1.64052024e-01 -3.96420836e-01 4.11708921e-01
-1.03697523e-01 -7.00945184e-02 -1.43887198e+00 7.26490617e-01
-1.89150953e+00 -7.69680679e-01 -2.27480412e-01 1.83892691e+00
6.35741651e-01 1.73094288e-01 1.91433772e-01 -8.70776474e-02
4.38536704e-01 -6.95848763e-02 -5.93579650e-01 -5.16383946e-01
-1.14950739e-01 2.54019946e-01 5.89860976e-01 5.33570468e-01
-9.91576016e-01 1.13862431e+00 7.11508703e+00 7.35171974e-01
-9.17869270e-01 -1.93759158e-01 6.18072093e-01 3.83314416e-02
-5.85189722e-02 1.82313442e-01 -4.58570302e-01 1.14394017e-01
6.30010128e-01 -2.62849182e-01 3.49258333e-01 7.39330232e-01
3.50531429e-01 -4.54660922e-01 -1.30188060e+00 1.15292919e+00
-1.98090822e-01 -1.88237083e+00 9.94828064e-03 1.89084247e-01
3.66096139e-01 2.47116581e-01 -2.22866312e-02 -1.85672060e-01
3.38464141e-01 -1.12881076e+00 6.86756432e-01 6.62625134e-01
5.47285438e-01 -3.93417478e-01 3.65462095e-01 7.06503332e-01
-1.02863300e+00 1.12722479e-01 -5.21045029e-01 2.26033032e-02
2.57473677e-01 3.87675256e-01 -1.48541808e+00 2.89611667e-01
7.37090766e-01 5.49277306e-01 -9.23477590e-01 1.48829174e+00
4.48218107e-01 4.24239576e-01 -8.36210668e-01 -2.51740634e-01
5.21639705e-01 -2.90469170e-01 7.56937504e-01 1.37892866e+00
3.07552963e-01 1.52300969e-01 1.17118418e-01 1.14804494e+00
4.48052645e-01 1.41376421e-01 -9.51192975e-01 2.68778950e-01
4.08801973e-01 1.20700407e+00 -1.38494873e+00 -2.83304185e-01
1.71228275e-01 6.65366888e-01 3.33006710e-01 1.66905522e-01
-4.34203893e-01 -1.80839717e-01 2.59896725e-01 4.01313454e-01
2.44703472e-01 -5.89306951e-01 -8.21070850e-01 -6.31314516e-01
-2.95095026e-01 -2.56018728e-01 1.82027996e-01 -1.25314963e+00
-1.17834806e+00 4.50889051e-01 3.53068888e-01 -1.04952860e+00
-1.13893576e-01 -8.38925421e-01 -6.89990819e-01 8.67181003e-01
-1.14189041e+00 -1.17932367e+00 -1.09369650e-01 2.16067582e-01
7.55552292e-01 1.20728843e-01 7.61203408e-01 -1.65736049e-01
-1.29628569e-01 -4.36594486e-02 3.19963098e-02 -9.79265869e-02
1.84711710e-01 -1.50892425e+00 5.59196711e-01 4.23036397e-01
2.97749221e-01 7.44489193e-01 8.52905452e-01 -5.64996481e-01
-1.09282005e+00 -7.48659730e-01 7.62369335e-01 -6.26502395e-01
3.80066484e-01 -3.73956710e-01 -1.00528777e+00 3.91094059e-01
-2.87700025e-03 9.12046954e-02 2.49562219e-01 6.68532103e-02
2.43173521e-02 1.69696584e-01 -1.37244487e+00 4.15417910e-01
9.95832443e-01 -1.72568783e-01 -6.16019964e-01 3.98081273e-01
3.54205728e-01 -4.91973251e-01 -6.03185296e-01 4.15450633e-01
2.99504220e-01 -9.20052707e-01 1.23853648e+00 -7.79240549e-01
8.04001391e-02 -6.44070864e-01 -1.10749729e-01 -1.39423108e+00
-2.67348826e-01 -4.12682533e-01 3.45935583e-01 8.96499872e-01
5.24678290e-01 -4.79540259e-01 8.95159483e-01 4.76803303e-01
-3.31591398e-01 -9.62187231e-01 -1.21143723e+00 -7.10501969e-01
4.50455137e-02 -4.03636187e-01 5.30425310e-01 9.45418477e-01
-4.95574577e-03 2.52271473e-01 3.77079308e-01 2.34919563e-01
6.70749605e-01 2.72518486e-01 7.88389981e-01 -1.81160188e+00
1.42084882e-01 -7.29098082e-01 -7.55146682e-01 -1.05955863e+00
-1.65301457e-01 -1.01483703e+00 1.19453080e-01 -1.81302512e+00
-9.23864320e-02 -8.31431448e-01 1.29885748e-01 4.43141460e-01
3.18175733e-01 1.00067697e-01 1.23345777e-01 3.57146740e-01
-6.33172572e-01 7.99372569e-02 1.06023216e+00 -4.76215780e-02
-2.96559572e-01 -3.04197017e-02 -3.71724576e-01 1.01977015e+00
8.93547893e-01 -3.86480600e-01 -1.88477606e-01 -4.89323825e-01
2.75171876e-01 -2.06643656e-01 6.97715521e-01 -1.08948147e+00
5.59788227e-01 -3.82922113e-01 3.91006052e-01 -1.09101367e+00
4.71282214e-01 -8.75231147e-01 9.18230936e-02 2.81784624e-01
-5.27284503e-01 2.51754999e-01 3.63312721e-01 4.60120410e-01
1.19278766e-01 -3.35892886e-01 9.06876862e-01 -5.78188539e-01
-7.54574120e-01 2.51775801e-01 -7.44627267e-02 -2.65119106e-01
1.09764743e+00 -7.37729013e-01 -1.55616492e-01 -7.25357756e-02
-1.04570150e+00 3.41428876e-01 8.67285967e-01 -2.30631568e-02
7.48390496e-01 -1.00726211e+00 -4.75899994e-01 1.31828068e-02
-3.38404179e-02 3.96955222e-01 -1.76432028e-01 8.92288327e-01
-1.13114035e+00 7.10241973e-01 -3.29304904e-01 -1.18339241e+00
-1.26863229e+00 5.47007918e-01 4.64233130e-01 9.33844876e-03
-8.00569952e-01 9.38582957e-01 4.06268053e-02 -5.62595844e-01
4.24198695e-02 -6.50037587e-01 -2.20069826e-01 -2.57254969e-02
9.17689651e-02 4.33018059e-01 1.23538606e-01 -7.84970880e-01
-4.06506330e-01 9.66431499e-01 1.41781271e-01 -2.65703738e-01
1.55688858e+00 -9.92507786e-02 -2.12400109e-01 5.62500179e-01
9.89723921e-01 -5.01023173e-01 -1.16410446e+00 -1.72578633e-01
4.42642540e-01 -5.24544418e-01 -1.94082335e-02 -8.55471969e-01
-5.51360726e-01 1.18905771e+00 5.79110801e-01 5.04015744e-01
8.20731342e-01 5.22424102e-01 4.59711164e-01 4.43159550e-01
6.20894313e-01 -1.10868621e+00 -2.59929989e-02 3.60803306e-01
1.03366399e+00 -1.16592646e+00 1.77267477e-01 -3.49086016e-01
-3.93950880e-01 1.31783187e+00 4.63504940e-01 -1.99532002e-01
6.61461115e-01 2.02753901e-01 -1.53131172e-01 -6.88242137e-01
-6.73237681e-01 -2.78408617e-01 2.02690706e-01 8.86573076e-01
-2.24262662e-02 -1.03115715e-01 -1.07463934e-01 -1.33424364e-02
-3.45189452e-01 -5.90378568e-02 2.40966484e-01 9.10549223e-01
-8.72048795e-01 -8.91568124e-01 -5.55718005e-01 5.98946273e-01
-5.59361279e-02 8.69266763e-02 -6.92374766e-01 8.07416737e-01
2.54981786e-01 5.80228925e-01 2.18948349e-01 -1.40029222e-01
2.44210124e-01 3.34979296e-02 5.62189400e-01 -6.64378941e-01
-4.42085952e-01 -4.53899466e-02 -2.41484478e-01 -5.63624501e-01
-5.14901161e-01 -8.46576929e-01 -1.62927079e+00 1.72500256e-02
-2.88046479e-01 1.03579968e-01 9.88599181e-01 1.03025186e+00
4.53748614e-01 2.16563344e-01 1.73601761e-01 -1.42593896e+00
7.62341218e-03 -7.47307122e-01 -2.63025016e-01 7.54863247e-02
3.90835434e-01 -8.08736801e-01 -1.49164781e-01 2.27033719e-01] | [8.104736328125, -3.0969181060791016] |
48c3a1e3-c7b0-4903-9207-e996e72758fa | deanet-decomposition-enhancement-and | 2209.06823 | null | https://arxiv.org/abs/2209.06823v1 | https://arxiv.org/pdf/2209.06823v1.pdf | DEANet: Decomposition Enhancement and Adjustment Network for Low-Light Image Enhancement | Images obtained under low-light conditions will seriously affect the quality of the images. Solving the problem of poor low-light image quality can effectively improve the visual quality of images and better improve the usability of computer vision. In addition, it has very important applications in many fields. This paper proposes a DEANet based on Retinex for low-light image enhancement. It combines the frequency information and content information of the image into three sub-networks: decomposition network, enhancement network and adjustment network. These three sub-networks are respectively used for decomposition, denoising, contrast enhancement and detail preservation, adjustment, and image generation. Our model has good robust results for all low-light images. The model is trained on the public data set LOL, and the experimental results show that our method is better than the existing state-of-the-art methods in terms of vision and quality. | ['Hongbing Ma', 'Yuan Xue', 'Liangliang Li', 'Yonglong Jiang'] | 2022-09-14 | null | null | null | null | ['low-light-image-enhancement'] | ['computer-vision'] | [ 1.34311348e-01 -7.68674552e-01 2.08479747e-01 -2.46168926e-01
-1.65452898e-01 -2.98011992e-02 2.09552079e-01 -4.67879057e-01
-5.20430088e-01 6.67903185e-01 1.27639815e-01 -6.21488988e-02
2.42213309e-01 -1.01121354e+00 -4.62335825e-01 -1.10960317e+00
4.87452745e-01 -7.37983167e-01 3.67755085e-01 -4.54581857e-01
3.26966763e-01 5.27886093e-01 -1.78056884e+00 4.06352878e-01
1.00157785e+00 8.72340143e-01 3.82498652e-01 6.75328493e-01
2.78865136e-02 7.84867942e-01 -3.19002628e-01 -8.70917812e-02
3.93800229e-01 -2.66543984e-01 -2.08825618e-01 2.97650754e-01
5.57230949e-01 -6.31915510e-01 -3.64444345e-01 1.48337209e+00
1.02912605e+00 1.75767511e-01 3.80525440e-01 -6.53363407e-01
-1.15137851e+00 -2.50665426e-01 -9.77157652e-01 3.41023743e-01
1.13828152e-01 4.31017429e-01 2.57503986e-01 -1.08858204e+00
2.79632479e-01 1.28650069e+00 7.01116383e-01 5.45772314e-01
-9.09122407e-01 -8.25659394e-01 -2.99305677e-01 5.96236944e-01
-9.58990216e-01 -5.78927338e-01 6.16353035e-01 -1.68660153e-02
6.22013450e-01 2.16238592e-02 7.39226937e-01 4.62782174e-01
6.77310050e-01 4.49530691e-01 1.50024176e+00 -5.86321890e-01
-2.57549375e-01 -2.26817629e-03 2.81675961e-02 7.59524047e-01
3.97932976e-01 5.73103964e-01 -3.16791892e-01 4.58185762e-01
7.63435185e-01 1.38512790e-01 -5.86312175e-01 3.66702527e-01
-8.05974066e-01 2.71383852e-01 6.04874194e-01 3.23139280e-01
-2.73070127e-01 -5.36221266e-02 -6.00143597e-02 2.00645238e-01
7.14816153e-01 1.64414704e-01 -1.75238818e-01 5.04029453e-01
-4.09182608e-01 1.94797032e-02 -2.39948202e-02 3.75497192e-01
6.81047142e-01 1.56695321e-01 -3.95355582e-01 1.23723149e+00
4.58041340e-01 7.99476862e-01 2.42250890e-01 -9.50643659e-01
2.45212659e-01 3.76850635e-01 2.65996844e-01 -1.06654024e+00
-4.75053728e-01 -4.03130293e-01 -1.29069948e+00 6.89105332e-01
-4.18588594e-02 6.62222803e-02 -9.45735991e-01 1.43332112e+00
3.47959638e-01 1.85890451e-01 1.76814329e-02 1.10513306e+00
1.45326495e+00 1.17293930e+00 -1.13123521e-01 -6.25142813e-01
1.38745391e+00 -9.51370060e-01 -1.17101729e+00 -1.95576385e-01
-3.17473561e-01 -1.33417618e+00 1.10498679e+00 6.13023758e-01
-1.54009652e+00 -1.17276442e+00 -1.17502093e+00 -4.84567612e-01
-7.39422366e-02 4.63589072e-01 5.16637146e-01 7.23227143e-01
-1.24627769e+00 2.61691093e-01 -2.32873514e-01 -7.59294583e-03
3.63653928e-01 -5.47773428e-02 -1.51652813e-01 -5.29004574e-01
-1.16746998e+00 8.58970225e-01 1.23818249e-01 4.98963356e-01
-6.26429379e-01 -5.38882732e-01 -7.74359882e-01 -8.69635046e-02
1.16467871e-01 -7.42156446e-01 8.27914298e-01 -4.47547644e-01
-1.41453207e+00 9.55827534e-01 -3.44501108e-01 8.68698359e-02
1.50935993e-01 1.05573192e-01 -6.92157328e-01 1.49724707e-01
-2.74108723e-02 3.26639593e-01 1.05982435e+00 -1.47884226e+00
-8.72282982e-01 -2.95769215e-01 -1.11350581e-01 2.88475245e-01
-2.00745165e-01 1.65938303e-01 -8.63133490e-01 -5.78535020e-01
9.40098688e-02 -4.24964190e-01 1.66267939e-02 1.47456706e-01
3.40996236e-02 -1.10220045e-01 8.46797049e-01 -9.33728337e-01
1.13031948e+00 -2.30038452e+00 -4.53657389e-01 -7.14333579e-02
2.74436831e-01 7.86739409e-01 -3.02524447e-01 -2.22173288e-01
-1.11636154e-01 -1.39430523e-01 -6.18645847e-02 -7.68982992e-02
-4.60804462e-01 -2.47093756e-02 1.23920560e-01 7.02502966e-01
1.44116301e-02 6.31616652e-01 -6.86933875e-01 -6.15428686e-01
5.91191947e-01 8.67128074e-01 -4.37554866e-01 3.40543419e-01
1.86550871e-01 3.12164217e-01 6.47170320e-02 7.75786340e-01
1.23300612e+00 3.92287225e-02 -3.56575221e-01 -8.39905798e-01
-3.02991837e-01 -3.06345731e-01 -1.13930738e+00 1.11888814e+00
-6.52058721e-01 8.71489108e-01 1.71380147e-01 -2.26991326e-01
8.98720145e-01 7.74054453e-02 2.39470914e-01 -1.17851818e+00
2.91914016e-01 4.82388362e-02 -2.27819830e-01 -1.06511080e+00
4.61736441e-01 -2.36633316e-01 8.35958600e-01 2.95152217e-01
-3.79780620e-01 -2.20777959e-01 4.38977271e-01 -1.37545660e-01
3.11356932e-01 -6.19000904e-02 1.87806219e-01 1.23203576e-01
9.28407669e-01 -7.91018903e-01 7.47006893e-01 4.46625262e-01
-3.02092910e-01 5.17796576e-01 -2.05891594e-01 -3.65498275e-01
-1.01692235e+00 -1.06160688e+00 -3.13517749e-01 8.69487464e-01
7.34086096e-01 -1.73163004e-02 -7.00302362e-01 -1.06918246e-01
-3.90843093e-01 4.50564772e-01 -2.45663047e-01 -3.78932744e-01
-4.75032300e-01 -1.19834030e+00 9.17400271e-02 1.71454221e-01
1.43889678e+00 -1.24151278e+00 -1.93356186e-01 -1.16537854e-01
-6.81086421e-01 -1.08486509e+00 -5.08320093e-01 -4.90696043e-01
-5.79350531e-01 -1.26121426e+00 -6.38625562e-01 -9.09994125e-01
8.21385741e-01 1.01933932e+00 1.11999738e+00 5.39313674e-01
-5.51884472e-01 1.10982165e-01 -1.43042967e-01 -5.23371518e-01
-3.71028692e-01 -6.53076172e-01 -1.35284871e-01 2.48712793e-01
2.55091548e-01 -3.12231123e-01 -1.01700366e+00 3.25791061e-01
-1.03723645e+00 1.09985419e-01 8.17259729e-01 9.04029846e-01
7.54000485e-01 9.18775737e-01 1.03528030e-01 -5.42163908e-01
8.82186651e-01 2.68067956e-01 -6.81412101e-01 3.26561034e-01
-1.03648460e+00 -3.08672547e-01 6.10006630e-01 -1.83524147e-01
-1.90148032e+00 -3.98768902e-01 -2.83515126e-01 7.56996199e-02
-1.84040889e-01 -8.80668312e-02 -3.11887294e-01 -6.84301555e-01
7.71425247e-01 3.92584205e-01 -1.08717093e-02 -6.40314221e-01
8.47471431e-02 7.92186499e-01 8.80171716e-01 -7.55306557e-02
7.39565492e-01 6.07846975e-01 3.98749948e-01 -9.10683990e-01
-8.36672425e-01 -5.80598056e-01 -1.31694347e-01 -6.22115076e-01
9.10199702e-01 -9.63147700e-01 -1.05431414e+00 1.03369153e+00
-1.11750209e+00 -1.14842705e-01 1.81887504e-02 4.88413125e-01
-2.51396805e-01 6.17847145e-01 -9.23547983e-01 -6.83292687e-01
-5.19831002e-01 -1.16938710e+00 6.84360743e-01 8.70301306e-01
9.43677008e-01 -8.20866108e-01 -6.28365800e-02 4.93596286e-01
4.58426476e-01 -7.90518075e-02 8.43381047e-01 8.90208066e-01
-7.23917663e-01 6.51777908e-02 -8.37387919e-01 9.09818470e-01
2.30743930e-01 7.62451738e-02 -1.16871607e+00 -3.77775222e-01
3.41008276e-01 -8.20981860e-02 1.25912642e+00 8.63017499e-01
1.36921775e+00 -5.27451225e-02 7.82840401e-02 1.16010201e+00
1.80275249e+00 3.33484620e-01 1.26667941e+00 4.17821050e-01
4.61855352e-01 3.93493235e-01 6.69923127e-01 1.66952461e-01
2.72689104e-01 3.76065701e-01 5.97469211e-01 -1.05765665e+00
-7.71103084e-01 2.02368021e-01 2.96445578e-01 5.95540106e-01
-5.57450473e-01 -1.99319974e-01 -5.14183402e-01 3.88234973e-01
-1.47859025e+00 -1.38427031e+00 -6.18203163e-01 1.83454573e+00
1.03530133e+00 -1.53941318e-01 -3.53064895e-01 2.20099121e-01
8.19009304e-01 1.80484682e-01 -3.23896110e-01 -2.38826573e-01
-4.13712651e-01 -3.84972133e-02 5.15877783e-01 5.94669878e-01
-1.05460310e+00 6.58466756e-01 6.73373222e+00 8.38197529e-01
-9.31404352e-01 1.35520592e-01 8.63234341e-01 7.96440616e-02
6.59883916e-02 -5.10110676e-01 -8.50163817e-01 6.55417442e-01
5.92664555e-02 1.46979406e-01 7.54512489e-01 2.60579884e-01
7.08631873e-01 -4.84663367e-01 -4.33729440e-01 1.38642347e+00
4.21939641e-01 -1.08340812e+00 2.55878847e-02 -1.85323328e-01
1.00181210e+00 -1.24556147e-01 4.29427415e-01 -1.68250889e-01
-3.87386531e-02 -9.17671025e-01 1.59879312e-01 9.42589998e-01
1.06655765e+00 -8.18605125e-01 9.81000900e-01 2.57875264e-01
-1.14178789e+00 -5.80479279e-02 -7.36709237e-01 1.85848296e-01
4.87054326e-02 7.87313759e-01 -2.00500280e-01 4.26047415e-01
1.04494369e+00 9.26512063e-01 -7.60617375e-01 1.35004711e+00
-5.82361281e-01 4.07207191e-01 5.36278486e-02 2.64752209e-01
-1.19105749e-01 -5.28514326e-01 3.92775536e-01 1.16268027e+00
-1.66618656e-02 3.86684120e-01 -2.82045566e-02 6.79222941e-01
-3.26177515e-02 -4.59094755e-02 -2.95290232e-01 5.89986980e-01
-1.12119503e-01 1.34561563e+00 -4.30321425e-01 -2.14459836e-01
-5.44739366e-01 5.19360900e-01 -4.80476975e-01 5.89312375e-01
-7.32102633e-01 -7.20654964e-01 3.05090100e-01 1.01054333e-01
-1.82457402e-01 -1.23109613e-02 -4.49690223e-01 -1.06672382e+00
1.44399568e-01 -9.04533327e-01 2.09452122e-01 -1.55515385e+00
-1.19508052e+00 4.30801362e-01 -3.55158806e-01 -1.20771599e+00
3.58098298e-01 -7.29752600e-01 -6.07855916e-01 1.18173850e+00
-2.42375088e+00 -9.85606909e-01 -9.30944204e-01 9.65303719e-01
6.10249102e-01 -2.78332442e-01 2.26051345e-01 5.55449843e-01
-5.01240909e-01 7.20504150e-02 2.26723313e-01 -8.93326104e-02
1.03562236e+00 -9.00635183e-01 -3.91468580e-04 1.48528397e+00
-3.45056087e-01 3.74024034e-01 5.75677872e-01 -5.15272737e-01
-8.59713674e-01 -1.08076835e+00 7.66228259e-01 1.99048758e-01
-1.31394029e-01 2.26208955e-01 -7.91398704e-01 1.85956061e-02
3.16686064e-01 1.08941153e-01 4.08311576e-01 -3.56549412e-01
-1.96446050e-02 -6.66126609e-01 -1.32140207e+00 6.18204534e-01
8.10636342e-01 -2.96147287e-01 -3.71376127e-01 2.71476775e-01
4.86973852e-01 -3.75623196e-01 -4.70142663e-01 5.19707561e-01
5.01165688e-01 -1.47170627e+00 1.42412007e+00 -4.10328154e-03
6.65406585e-01 -6.26456380e-01 7.43701681e-02 -1.36852682e+00
-7.45549262e-01 -2.96139836e-01 1.87983885e-01 1.20683372e+00
-3.70260887e-02 -6.83635116e-01 2.26244405e-01 2.05735117e-01
4.41774074e-03 -5.38764596e-01 -2.85705596e-01 -3.61148566e-01
-6.08316898e-01 -1.56600371e-01 6.16944194e-01 6.77594483e-01
-7.74081528e-01 3.19014609e-01 -6.22822106e-01 2.24187106e-01
1.15672660e+00 4.21020277e-02 6.71727836e-01 -9.88008380e-01
-1.07253194e-01 -3.06122422e-01 -8.31210241e-03 -7.57633150e-01
-4.10205685e-02 -5.00151455e-01 1.16888754e-01 -2.06740165e+00
4.78235662e-01 -9.46598053e-02 -2.69728243e-01 4.51733530e-01
-5.70545316e-01 8.13405514e-01 1.51966929e-01 9.62607637e-02
-4.00687397e-01 4.54907119e-01 1.88157547e+00 -3.37418675e-01
-2.33364120e-01 4.28448617e-02 -7.30519176e-01 8.68522406e-01
8.54274631e-01 -3.07910424e-02 -3.44559640e-01 -6.01073682e-01
4.02190268e-01 -2.27510303e-01 4.97638315e-01 -9.48243797e-01
3.11116785e-01 -4.03651714e-01 9.13759410e-01 -6.97317302e-01
2.81065136e-01 -9.26644564e-01 -1.09135330e-01 4.86912280e-01
-9.72908884e-02 -2.35268369e-01 1.21714786e-01 3.08339655e-01
-2.72612423e-01 -2.83135921e-01 1.57754564e+00 -3.11603427e-01
-8.53917301e-01 3.85438591e-01 -1.69183061e-01 -3.25356394e-01
8.31870317e-01 -1.93134099e-01 -8.44419956e-01 -4.11945820e-01
-3.46232086e-01 4.61023524e-02 3.66271079e-01 2.37362254e-02
1.14446151e+00 -1.30014038e+00 -8.44798028e-01 7.08612919e-01
-3.94848436e-02 -1.98653102e-01 6.23284519e-01 6.04884267e-01
-7.67532885e-01 -7.08605647e-02 -5.74640274e-01 -4.10184890e-01
-1.73267388e+00 4.24012005e-01 5.83404183e-01 -9.98151079e-02
-5.46598256e-01 6.25717342e-01 3.24416697e-01 1.25187963e-01
1.04672939e-01 -5.36837801e-02 -5.43946385e-01 -5.48830807e-01
1.07929039e+00 7.85950840e-01 4.62843291e-02 -6.57684386e-01
9.80605185e-02 1.04646945e+00 -3.00285146e-02 3.63396496e-01
1.50520313e+00 -5.60044825e-01 -5.96378624e-01 4.12238799e-02
8.52412343e-01 1.17941566e-01 -1.08162761e+00 -2.48653114e-01
-1.06437635e+00 -1.06137335e+00 6.98907852e-01 -1.00415516e+00
-1.47312975e+00 1.08699417e+00 1.05934775e+00 2.51575857e-01
1.94018006e+00 -4.82526541e-01 8.48500967e-01 2.58644372e-01
1.49329435e-02 -1.27675915e+00 3.90441209e-01 2.49501705e-01
8.40752423e-01 -1.43913686e+00 3.11268747e-01 -5.17952502e-01
-1.13016084e-01 1.17659676e+00 7.59882748e-01 1.38665915e-01
5.12747347e-01 2.45634764e-01 3.90017897e-01 1.71862058e-02
-4.73273575e-01 -4.34110224e-01 4.49096978e-01 1.00893152e+00
3.51602286e-01 -4.31839079e-01 -4.06740069e-01 1.57587782e-01
2.54434854e-01 2.05379054e-01 6.69827342e-01 3.20836365e-01
-7.76304066e-01 -7.02040613e-01 -9.49291527e-01 2.27156863e-01
-8.14549148e-01 -2.95348465e-01 1.12812474e-01 3.23735088e-01
6.52480841e-01 1.41954851e+00 -2.31102675e-01 -1.10172100e-01
3.69283795e-01 -6.36912644e-01 6.06328964e-01 -1.46538630e-01
-3.75092328e-01 3.88400376e-01 -2.90646166e-01 -5.37898958e-01
-9.12482023e-01 2.23885421e-02 -7.43427575e-01 -5.91176808e-01
-3.52676123e-01 -4.35869873e-01 7.03634799e-01 4.01271313e-01
-2.68495409e-03 8.01236093e-01 9.53083158e-01 -8.77101064e-01
-5.80093786e-02 -8.22210073e-01 -8.65846932e-01 5.56582391e-01
6.53469265e-01 -5.65995991e-01 -3.98365021e-01 4.54427302e-01] | [10.809139251708984, -2.470363140106201] |
2f5f1e69-2971-475a-aa67-9f8987ea4002 | excalibr-expected-calibration-of | 2304.12311 | null | https://arxiv.org/abs/2304.12311v1 | https://arxiv.org/pdf/2304.12311v1.pdf | ExCalibR: Expected Calibration of Recommendations | In many recommender systems and search problems, presenting a well balanced set of results can be an important goal in addition to serving highly relevant content. For example, in a movie recommendation system, it may be helpful to achieve a certain balance of different genres, likewise, it may be important to balance between highly popular versus highly personalized shows. Such balances could be thought across many categories and may be required for enhanced user experience, business considerations, fairness objectives etc. In this paper, we consider the problem of calibrating with respect to any given categories over items. We propose a way to balance a trade-off between relevance and calibration via a Linear Programming optimization problem where we learn a doubly stochastic matrix to achieve optimal balance in expectation. We then realize the learned policy using the Birkhoff-von Neumann decomposition of a doubly stochastic matrix. Several optimizations are considered over the proposed basic approach to make it fast. The experiments show that the proposed formulation can achieve a much better trade-off compared to many other baselines. This paper does not prescribe the exact categories to calibrate over (such as genres) universally for applications. This is likely dependent on the particular task or business objective. The main contribution of the paper is that it proposes a framework that can be applied to a variety of problems and demonstrates the efficacy of the proposed method using a few use-cases. | ['Pannagadatta Shivaswamy'] | 2023-04-24 | null | null | null | null | ['movie-recommendation'] | ['miscellaneous'] | [-3.38485204e-02 -3.26034054e-02 -3.16679686e-01 -6.37315452e-01
-8.01889420e-01 -6.42781198e-01 4.63408887e-01 6.28901199e-02
-4.53036070e-01 7.55553305e-01 2.45322093e-01 -3.05180937e-01
-6.94830537e-01 -6.36514843e-01 -5.54652989e-01 -8.16933036e-01
6.44247383e-02 5.99426925e-01 1.10420123e-01 -6.15941584e-01
5.39590418e-01 2.21682996e-01 -1.56554949e+00 2.87492365e-01
8.21912289e-01 1.05158532e+00 3.47706228e-01 3.47779721e-01
-6.24888279e-02 5.62661886e-01 -3.86321425e-01 -7.74042547e-01
6.62034690e-01 -2.48992532e-01 -7.24090278e-01 2.23573029e-01
3.03390592e-01 -2.02846825e-01 3.54915947e-01 1.18244183e+00
5.09178519e-01 4.78197694e-01 8.74812841e-01 -1.12741899e+00
-5.92447877e-01 9.12986457e-01 -7.05092669e-01 1.45657718e-01
2.15216950e-01 -2.97169358e-01 1.60571492e+00 -6.15485907e-01
2.46164665e-01 1.17737472e+00 3.29306871e-01 1.74100250e-01
-1.25490880e+00 -2.61329830e-01 4.03016806e-01 8.31217244e-02
-1.12400973e+00 -2.72407383e-01 6.90535605e-01 -5.64141512e-01
3.41942757e-01 6.92600787e-01 2.68901050e-01 7.92177796e-01
1.59071997e-01 6.32705152e-01 1.24753463e+00 -4.39602882e-01
4.14145410e-01 8.01529884e-01 2.60591716e-01 1.87245727e-01
3.29038918e-01 -1.92471698e-01 -3.17057729e-01 -3.09466332e-01
4.06927466e-01 7.15544866e-03 -3.67333978e-01 -5.91032088e-01
-9.06906486e-01 9.59898710e-01 2.26283953e-01 2.56299496e-01
-4.34271723e-01 -8.24191123e-02 1.53245360e-01 4.87878084e-01
2.36494198e-01 6.63569093e-01 -3.07178348e-01 -5.71662281e-03
-9.89521921e-01 4.90905583e-01 1.20543146e+00 7.00730026e-01
4.49779600e-01 -2.80742735e-01 -3.75251234e-01 1.15744925e+00
5.13830483e-01 1.36742532e-01 3.21690798e-01 -1.07145846e+00
4.48715538e-01 -1.30180065e-02 6.63338125e-01 -1.06941032e+00
-1.74899787e-01 -6.88368917e-01 -6.54815018e-01 1.93077743e-01
5.84315658e-01 -1.86960429e-01 -2.02281192e-01 1.67544305e+00
3.52176249e-01 -1.13013752e-01 -2.89291441e-01 1.37599134e+00
3.71435881e-01 6.35567725e-01 -2.09274903e-01 -4.61933047e-01
1.35296321e+00 -9.66686547e-01 -7.97085941e-01 -1.09877653e-01
1.28101900e-01 -1.03933465e+00 1.27237844e+00 6.65547073e-01
-1.17589819e+00 -2.58113831e-01 -9.45850730e-01 1.55401304e-01
-1.81029573e-01 1.90310583e-01 5.95925748e-01 1.00191700e+00
-9.84003365e-01 7.21450746e-01 -2.11861745e-01 -4.11653847e-01
-3.29679221e-01 5.35971463e-01 1.88081041e-01 3.31782699e-01
-1.25048316e+00 9.10254180e-01 2.57149428e-01 -8.77119154e-02
-4.76656884e-01 -4.82648969e-01 -3.09914619e-01 2.71354556e-01
5.26480019e-01 -8.47106695e-01 1.34178913e+00 -1.05688107e+00
-1.75709248e+00 7.21390188e-01 2.87786752e-01 -5.13875842e-01
6.99833393e-01 -2.09642977e-01 -1.88156769e-01 -3.37532192e-01
4.48982045e-02 1.75303757e-01 7.13586450e-01 -1.20544720e+00
-9.31596816e-01 -1.78630352e-01 5.99090397e-01 5.44958115e-01
-6.17894828e-01 1.89207867e-01 -6.21051669e-01 -6.64134920e-01
-1.06201217e-01 -9.59899247e-01 -3.89412224e-01 -2.66956329e-01
-2.38478035e-01 -6.14869408e-02 1.03615545e-01 -5.22826731e-01
1.52820361e+00 -1.84856844e+00 3.94541085e-01 3.49602669e-01
-2.00477093e-01 -1.62847117e-01 1.18501879e-01 5.44636250e-01
1.27329186e-01 6.71665594e-02 8.79543498e-02 -2.51467824e-01
3.79606724e-01 9.90408212e-02 -3.50731015e-01 5.55501819e-01
-3.07772100e-01 3.80087912e-01 -8.11186194e-01 -2.88475841e-01
8.68461281e-03 2.37122640e-01 -8.25300813e-01 9.88411084e-02
-1.96979940e-01 6.11461140e-02 -4.92941499e-01 2.69129574e-01
4.11977828e-01 -2.81370074e-01 4.33391660e-01 -2.52724677e-01
-7.67630637e-02 1.42053500e-01 -1.70048761e+00 1.19161773e+00
-5.67220747e-01 1.06344879e-01 3.06341827e-01 -1.32018685e+00
8.53646696e-01 1.49870545e-01 6.35961831e-01 -3.88197571e-01
3.91997784e-01 9.35826674e-02 3.97044308e-02 -3.18076491e-01
8.74275804e-01 -3.83033663e-01 -7.97195658e-02 5.45707881e-01
-2.82846302e-01 -8.31338614e-02 2.67083734e-01 1.39422327e-01
4.86323029e-01 4.00817096e-02 4.76215929e-01 -7.74180949e-01
6.72767103e-01 -2.45375693e-01 5.14714599e-01 8.54966819e-01
5.48050255e-02 4.52825546e-01 6.33158386e-01 -3.26048359e-02
-1.00336969e+00 -6.60893023e-01 -2.53978223e-01 1.38489819e+00
2.94405639e-01 -1.70776173e-01 -6.29169822e-01 -4.48936492e-01
-3.79336439e-02 8.66356194e-01 -3.40284824e-01 6.23599999e-02
-1.57851264e-01 -9.78247762e-01 -2.32563764e-01 -1.85132176e-02
1.98393807e-01 -6.01079166e-01 -3.50332648e-01 1.71888486e-01
-2.45480984e-01 -8.43480468e-01 -8.29481959e-01 1.23676926e-01
-8.05629015e-01 -7.28287995e-01 -8.46531510e-01 -6.30807936e-01
4.83121842e-01 5.42681158e-01 1.19213820e+00 -2.28545442e-01
4.64683384e-01 4.87978637e-01 -4.60805923e-01 -3.25833470e-01
-1.90343738e-01 1.94283724e-01 3.06531459e-01 5.50656557e-01
1.15963735e-01 -6.34783745e-01 -7.09164619e-01 6.07466519e-01
-8.84339035e-01 -1.60862580e-01 3.95262629e-01 8.69695604e-01
5.67019880e-01 9.71363410e-02 5.57415128e-01 -1.28762293e+00
1.21933687e+00 -7.17726350e-01 -7.49811828e-01 4.09739196e-01
-8.50463390e-01 3.20401430e-01 7.47755289e-01 -6.09748423e-01
-9.47779179e-01 -2.38286242e-01 -1.38412476e-01 -6.91523999e-02
4.07690495e-01 4.71238792e-01 -2.36166775e-01 -2.27651112e-02
6.91935718e-01 -1.41790807e-01 -9.81184915e-02 -5.12680948e-01
4.85621125e-01 9.89417255e-01 9.00096297e-02 -8.70267153e-01
4.20808047e-01 2.81681687e-01 -2.22966924e-01 -5.71431041e-01
-9.83357370e-01 -6.98232174e-01 -2.03856960e-01 -1.82199210e-01
4.15627033e-01 -4.72854286e-01 -7.82453835e-01 -1.92507356e-01
-6.65853918e-01 -2.22799048e-01 -3.14828426e-01 4.75968122e-01
-8.07693958e-01 4.94718462e-01 -4.15495217e-01 -9.60247695e-01
-3.47017109e-01 -1.13897491e+00 6.49845123e-01 3.03085655e-01
-2.98759490e-02 -1.07400107e+00 7.77242184e-02 3.86299253e-01
4.27486897e-01 -6.08625934e-02 5.40638864e-01 -6.76524520e-01
-4.61437702e-01 -2.82743108e-02 1.45491529e-02 3.63739669e-01
9.61570293e-02 -1.08703347e-02 -5.75391054e-01 -4.14617389e-01
2.08297223e-01 -2.73113847e-02 5.61495900e-01 4.66401577e-01
1.00639117e+00 -6.09070182e-01 4.44472097e-02 5.06290495e-01
1.51911449e+00 1.65413097e-01 4.47548985e-01 5.95860839e-01
2.59277940e-01 8.39959145e-01 1.09231544e+00 7.19939649e-01
3.48902047e-01 1.11301088e+00 4.40684944e-01 1.60642311e-01
3.50685954e-01 3.67601658e-03 3.85142922e-01 8.50812793e-01
2.82547455e-02 -2.87673950e-01 -3.28652024e-01 4.85534221e-01
-2.13222122e+00 -1.09476876e+00 6.25468567e-02 2.63396192e+00
7.22605407e-01 6.59851879e-02 4.89181012e-01 2.69082468e-02
7.84637392e-01 -8.20370950e-03 -3.57871860e-01 -6.24644160e-01
3.84639055e-02 -2.82761782e-01 6.66981220e-01 7.11103857e-01
-9.63021100e-01 5.37348390e-01 6.36146498e+00 1.04397476e+00
-1.07396817e+00 2.77044177e-01 7.75616109e-01 -2.02241883e-01
-5.98158181e-01 9.92628187e-02 -8.96863341e-01 5.79763472e-01
7.52076983e-01 -4.01789725e-01 6.71091437e-01 8.47294688e-01
4.43965167e-01 3.30370404e-02 -1.00894666e+00 1.11507285e+00
2.19159070e-02 -1.05147123e+00 -2.82754570e-01 2.54461020e-01
8.52561176e-01 -4.02664483e-01 2.78015256e-01 2.24240839e-01
5.80483794e-01 -6.16086006e-01 8.58086407e-01 5.35101414e-01
3.21230441e-01 -6.67178094e-01 6.00029290e-01 6.56066120e-01
-7.07863092e-01 -2.72131979e-01 -6.65190995e-01 -1.08555466e-01
3.43670100e-01 7.31095672e-01 -5.59201241e-01 5.12445927e-01
6.35144830e-01 3.09076220e-01 -9.39822719e-02 1.24162948e+00
3.13898847e-02 4.60944772e-01 -4.01850462e-01 -4.42141682e-01
1.73584744e-01 -8.38677704e-01 4.89972919e-01 1.09539771e+00
5.30768812e-01 1.52118921e-01 1.49749056e-01 5.39697051e-01
5.21836057e-02 9.00156379e-01 -2.89413273e-01 3.51797998e-01
1.71868682e-01 1.41716170e+00 -6.37426913e-01 -1.88108593e-01
-4.12664056e-01 7.61140168e-01 3.39393973e-01 3.31423283e-01
-9.49736118e-01 -4.16422635e-02 6.30553842e-01 2.99035400e-01
2.58669436e-01 1.24129876e-01 -3.03405195e-01 -1.26490891e+00
-2.38345824e-02 -1.15126538e+00 5.97715616e-01 -4.60837722e-01
-1.35872889e+00 5.44119358e-01 1.45469725e-01 -1.32196772e+00
-2.04252340e-02 -4.69663203e-01 -3.33998471e-01 8.43267322e-01
-1.50727010e+00 -5.62414885e-01 5.79712652e-02 3.86159897e-01
4.92099494e-01 -2.96313554e-01 2.73152858e-01 6.00967705e-01
-4.74400640e-01 5.94484687e-01 5.50321937e-01 -6.65115058e-01
1.06449926e+00 -1.43201125e+00 -3.26145619e-01 8.29957068e-01
2.81308502e-01 8.28755975e-01 1.19218290e+00 -1.44924492e-01
-1.11014426e+00 -6.97302878e-01 7.28060424e-01 -2.90222913e-01
6.86844349e-01 -1.09782787e-02 -5.05454779e-01 3.12875807e-01
2.45298192e-01 -5.31161368e-01 7.49796987e-01 4.93265450e-01
-6.89648986e-02 -5.98346174e-01 -1.08986020e+00 7.49332368e-01
7.21572995e-01 -1.12721607e-01 -4.40799803e-01 6.80193961e-01
3.80061775e-01 -3.16939712e-01 -8.16529989e-01 1.17170520e-01
7.26982117e-01 -1.27291369e+00 9.69054759e-01 -6.40064955e-01
3.19925427e-01 -3.08683366e-01 -3.74763697e-01 -1.43204105e+00
-5.83644569e-01 -7.83920646e-01 1.74254328e-02 1.34120405e+00
4.26639676e-01 -5.65354407e-01 6.69339001e-01 1.03780854e+00
3.07855338e-01 -8.88620496e-01 -5.84232628e-01 -6.03295207e-01
4.77973185e-02 -3.07394534e-01 5.91981649e-01 7.15583980e-01
4.99235950e-02 4.89289612e-01 -9.81198847e-01 1.89934775e-01
5.16175210e-01 4.70082909e-01 7.37123907e-01 -9.48778212e-01
-9.51519668e-01 -6.45890117e-01 5.87415993e-02 -1.41100633e+00
-2.87481397e-01 -8.39424193e-01 -7.16888830e-02 -1.43576145e+00
3.06750447e-01 -8.97797048e-01 -6.66328549e-01 -2.10613627e-02
-1.29519448e-01 1.23676039e-01 4.22671288e-01 2.12263048e-01
-6.06247246e-01 1.98613897e-01 1.14050853e+00 5.71700977e-03
-3.01314116e-01 6.87518060e-01 -1.42676556e+00 4.82431531e-01
8.22808504e-01 -3.51276904e-01 -6.43013418e-01 -2.33409062e-01
6.70274854e-01 2.02683389e-01 -2.15710759e-01 -4.93566841e-01
3.03211600e-01 -5.33404291e-01 -4.15711492e-01 -2.22276017e-01
2.70650685e-01 -9.95734751e-01 4.74719584e-01 1.46761775e-01
-4.06458199e-01 -1.32954761e-01 -4.16060865e-01 6.01067364e-01
-6.97917715e-02 -7.83649147e-01 8.23861301e-01 -7.14965686e-02
-3.97899657e-01 1.34935826e-01 5.64243644e-02 -4.49874848e-02
1.04591179e+00 -1.95979387e-01 8.03102627e-02 -8.49739373e-01
-6.07211590e-01 4.21803772e-01 3.76914054e-01 4.17343259e-01
1.28759488e-01 -1.25321424e+00 -6.60992205e-01 -3.70891631e-01
-1.18020989e-01 -6.56825483e-01 1.34557962e-01 9.09412265e-01
-4.28002208e-01 4.27545458e-01 -1.56077638e-01 -2.76915222e-01
-1.20325208e+00 7.44485736e-01 2.53946334e-01 -5.76874793e-01
-7.31608868e-02 7.17561483e-01 3.39137197e-01 -2.85375476e-01
5.00163317e-01 -1.20982997e-01 -7.69403219e-01 3.72574300e-01
5.17046750e-01 3.04146558e-01 -1.68925757e-03 -5.94992042e-01
-6.14194572e-02 5.63747585e-01 3.30689326e-02 -2.89943308e-01
1.31378531e+00 -5.49421191e-01 -1.70040876e-02 3.34764630e-01
8.16268027e-01 4.31602299e-01 -1.05482244e+00 -3.04941922e-01
-8.97136927e-02 -6.69327974e-01 1.14423856e-01 -5.53044140e-01
-1.15457356e+00 6.34084463e-01 4.55826372e-01 6.51284993e-01
1.18806946e+00 -2.83159882e-01 4.25056219e-01 2.63173491e-01
6.24386072e-01 -1.20135915e+00 -1.37885943e-01 9.62185785e-02
8.70091021e-01 -1.19781208e+00 2.42582291e-01 -4.12373155e-01
-1.01953256e+00 9.59876597e-01 2.38504365e-01 -2.09187135e-01
7.81574965e-01 4.90821414e-02 -2.63825841e-02 1.31297648e-01
-6.69449329e-01 -3.03391278e-01 4.91353542e-01 2.26534247e-01
6.14953637e-01 1.92034245e-01 -1.16790855e+00 8.22647572e-01
-2.30589733e-01 -2.92922229e-01 5.49525499e-01 3.23530376e-01
-5.25494397e-01 -1.64593625e+00 -5.88557243e-01 6.70991182e-01
-9.51721847e-01 -1.06441379e-02 -2.34797243e-02 2.17363164e-01
1.84898451e-02 1.13871992e+00 -4.20885414e-01 -2.04934001e-01
4.15089995e-01 -3.03732485e-01 2.46588305e-01 -6.26893997e-01
-7.46865749e-01 3.78610849e-01 2.91768640e-01 -2.42000565e-01
-5.38511276e-01 -8.48445296e-01 -4.56076086e-01 -3.80600423e-01
-5.95256209e-01 4.97894168e-01 5.37092745e-01 7.05354750e-01
1.00358233e-01 3.64093244e-01 7.73286939e-01 -5.59896767e-01
-1.13212788e+00 -6.43080950e-01 -9.95223939e-01 5.07265031e-01
-1.23870142e-01 -8.02057326e-01 -2.28495479e-01 -8.48996546e-03] | [9.666354179382324, 5.601218223571777] |
accdf14d-e04a-44a7-8ab1-ead42cfe1843 | multimodal-learning-using-optimal-transport | 2110.10949 | null | https://arxiv.org/abs/2110.10949v1 | https://arxiv.org/pdf/2110.10949v1.pdf | Multimodal Learning using Optimal Transport for Sarcasm and Humor Detection | Multimodal learning is an emerging yet challenging research area. In this paper, we deal with multimodal sarcasm and humor detection from conversational videos and image-text pairs. Being a fleeting action, which is reflected across the modalities, sarcasm detection is challenging since large datasets are not available for this task in the literature. Therefore, we primarily focus on resource-constrained training, where the number of training samples is limited. To this end, we propose a novel multimodal learning system, MuLOT (Multimodal Learning using Optimal Transport), which utilizes self-attention to exploit intra-modal correspondence and optimal transport for cross-modal correspondence. Finally, the modalities are combined with multimodal attention fusion to capture the inter-dependencies across modalities. We test our approach for multimodal sarcasm and humor detection on three benchmark datasets - MUStARD (video, audio, text), UR-FUNNY (video, audio, text), MST (image, text) and obtain 2.1%, 1.54%, and 2.34% accuracy improvements over state-of-the-art. | ['Vishal M. Patel', 'Aniket Roy', 'Shraman Pramanick'] | 2021-10-21 | null | null | null | null | ['humor-detection'] | ['natural-language-processing'] | [ 7.78173581e-02 -1.79044485e-01 -8.93528312e-02 -3.68081070e-02
-9.81857657e-01 -3.01569283e-01 6.90447867e-01 -1.27671743e-02
-2.96871454e-01 4.53254938e-01 6.15353882e-01 3.29992175e-01
4.08920556e-01 6.55402169e-02 -4.01432395e-01 -6.55227304e-01
3.52257639e-01 -4.58677337e-02 5.51381242e-03 -3.44070703e-01
4.23241556e-01 -1.37974575e-01 -1.27878821e+00 9.85467613e-01
5.77636898e-01 8.90655398e-01 -2.07870975e-02 9.83560920e-01
7.16871815e-03 1.47084594e+00 -2.39560962e-01 -8.05463433e-01
-4.18451816e-01 -8.67703080e-01 -8.37469995e-01 3.55450809e-01
5.13607144e-01 -2.64443099e-01 -3.82946730e-01 1.02605951e+00
7.05838442e-01 8.29046592e-02 5.93414843e-01 -1.43396580e+00
-4.06732649e-01 6.76973164e-01 -1.07695460e+00 -4.53741848e-02
6.10022902e-01 2.57794380e-01 9.32095647e-01 -1.22544849e+00
4.34311479e-01 1.50724435e+00 5.38793325e-01 6.59632087e-01
-7.89605439e-01 -3.10187697e-01 -2.83245921e-01 5.91669738e-01
-1.03713906e+00 -5.39053798e-01 1.05756867e+00 -4.93808657e-01
6.59390271e-01 3.32811445e-01 4.77396339e-01 1.45035851e+00
-1.19601920e-01 1.49753177e+00 1.05568087e+00 -4.24475014e-01
-3.41335922e-01 3.46969694e-01 7.21757561e-02 6.90742135e-01
-7.79616117e-01 -5.02296269e-01 -1.16849887e+00 3.37499776e-03
3.11043352e-01 4.80907597e-02 -4.03731167e-01 -8.01242441e-02
-1.36824834e+00 9.35132325e-01 3.37597966e-01 2.79114753e-01
-2.30994299e-01 -4.22550477e-02 9.14568245e-01 2.50401139e-01
2.11542636e-01 1.58351541e-01 -1.40198111e-03 -4.75704819e-01
-7.90568590e-01 -1.33334780e-02 5.47725439e-01 5.85024238e-01
3.02410036e-01 7.26396451e-03 -2.78917223e-01 1.35874772e+00
2.54894733e-01 6.38817966e-01 6.71635032e-01 -7.22846508e-01
6.74043238e-01 7.66695440e-01 -1.65145904e-01 -1.35132110e+00
-6.49982631e-01 -5.37619442e-02 -1.04561090e+00 -2.79049724e-01
2.33232409e-01 -1.51632667e-01 -3.07927549e-01 1.55574942e+00
2.39383474e-01 1.94896445e-01 4.93255466e-01 1.29215002e+00
1.42886794e+00 8.18485737e-01 9.26282182e-02 -4.60719883e-01
1.47250652e+00 -1.60034013e+00 -8.48695517e-01 -9.84200090e-02
6.23372972e-01 -1.25553131e+00 1.36182225e+00 4.83155817e-01
-1.36046398e+00 -5.61715245e-01 -7.51822889e-01 -2.36375839e-01
-4.79297005e-02 3.90906334e-01 1.66727379e-01 1.81626409e-01
-3.87345254e-01 2.04942584e-01 -3.63163471e-01 -6.29921734e-01
2.00583711e-01 -7.73189962e-02 -3.70886534e-01 -9.64790136e-02
-1.03601313e+00 7.57266223e-01 1.27424628e-01 1.22024767e-01
-9.11736488e-01 -3.47890168e-01 -7.05964565e-01 -1.72063693e-01
2.90347964e-01 -4.71789420e-01 1.26022279e+00 -1.44396198e+00
-1.49590707e+00 9.99536455e-01 -1.59304757e-02 -2.42663652e-01
4.58771706e-01 -3.40512931e-01 -4.58727002e-01 5.39870381e-01
-1.69339195e-01 6.97410464e-01 1.09624124e+00 -1.35875309e+00
-4.32780921e-01 -3.46718341e-01 8.82075876e-02 5.97473919e-01
-8.15022171e-01 5.01838386e-01 -3.51756305e-01 -4.14050967e-01
-1.68374151e-01 -8.94747376e-01 4.14241374e-01 -4.14187163e-01
-7.30876803e-01 -1.48040846e-01 1.26807749e+00 -1.00352168e+00
1.22248328e+00 -2.24324298e+00 6.22193098e-01 -4.73651260e-01
2.97066212e-01 1.86986580e-01 -1.57289460e-01 6.79827511e-01
1.45622671e-01 -2.51964182e-01 -2.90593266e-01 -7.56777525e-01
7.88564011e-02 -9.02178288e-02 -3.92946862e-02 4.69730914e-01
1.01216406e-01 9.35580194e-01 -6.80636704e-01 -9.66246843e-01
2.76168138e-01 4.79329109e-01 -3.31748784e-01 5.35200715e-01
1.11738838e-01 5.32454669e-01 -1.28646150e-01 9.70592499e-01
5.17395973e-01 -2.83486724e-01 4.58223410e-02 -7.03089952e-01
2.85109784e-03 -4.27344143e-01 -6.42463803e-01 1.73229873e+00
-5.37210941e-01 8.80629599e-01 1.31635413e-01 -9.28045392e-01
7.85393417e-01 4.51227158e-01 4.75475848e-01 -6.91980958e-01
4.89933789e-01 4.93846349e-02 -1.76206872e-01 -1.24992299e+00
6.12029731e-01 -3.59333634e-01 -2.28425533e-01 3.98608327e-01
6.30619898e-02 8.52308869e-02 1.21223599e-01 4.47135508e-01
8.96292925e-01 -2.33991832e-01 1.77585423e-01 2.27149308e-01
9.57743227e-01 -1.59417689e-02 4.99166921e-02 3.02445382e-01
-2.43633404e-01 6.86981559e-01 6.83522224e-01 -1.33516937e-01
-1.04791987e+00 -7.77670383e-01 1.40668914e-01 1.45576000e+00
2.13220313e-01 -4.24028754e-01 -7.42243290e-01 -7.09716082e-01
-2.70823687e-01 2.73965031e-01 -4.35923368e-01 -1.02951705e-01
-4.35462236e-01 -8.03709149e-01 5.51412344e-01 3.87341291e-01
7.08327651e-01 -1.32054639e+00 -6.31048858e-01 -1.76889867e-01
-8.08316469e-01 -1.40364790e+00 -6.89646423e-01 -3.62659395e-01
-6.02316260e-01 -1.22588968e+00 -8.19864750e-01 -8.36593747e-01
3.49050194e-01 3.70228827e-01 1.02286148e+00 5.19498810e-02
-1.85999215e-01 8.13854635e-01 -6.16721869e-01 1.31851047e-01
-4.35869783e-01 -1.50909007e-01 -2.56058604e-01 5.27242422e-01
2.09504828e-01 -2.31190100e-01 -4.92192745e-01 4.10772473e-01
-8.13898146e-01 4.20990825e-01 6.72504127e-01 1.39774871e+00
1.08542420e-01 -4.52563524e-01 7.17376828e-01 -6.15134180e-01
6.84690475e-01 -6.75234139e-01 2.32516274e-01 2.26405352e-01
1.09229438e-01 -4.41199780e-01 6.37359738e-01 -7.71202624e-01
-1.10681009e+00 1.65913627e-01 2.41071224e-01 -7.68547237e-01
-1.13476239e-01 6.62243187e-01 -9.66738909e-02 9.93574187e-02
5.14596403e-01 1.72484085e-01 1.90484717e-01 -4.19704914e-01
3.74260604e-01 1.17391622e+00 8.52092564e-01 -2.77020931e-01
2.81630993e-01 3.65366399e-01 -2.13947281e-01 -1.14013648e+00
-9.40020382e-01 -7.95502305e-01 -5.69080293e-01 -7.38893688e-01
1.09144115e+00 -1.01602769e+00 -1.02886617e+00 7.29744732e-01
-1.24775767e+00 -8.43718052e-02 2.70415843e-01 5.47831953e-01
-7.95639455e-01 8.52609813e-01 -8.68984878e-01 -1.00841117e+00
-6.19220138e-01 -1.03717399e+00 1.03356123e+00 2.15718359e-01
-2.17968345e-01 -8.83066773e-01 7.18199536e-02 1.25510740e+00
1.19541548e-02 3.01797479e-01 6.20084643e-01 -4.92918253e-01
-1.54261276e-01 -1.46470487e-01 -6.58448100e-01 4.16782469e-01
-2.37157315e-01 -5.19153215e-02 -9.85006213e-01 -2.22391531e-01
-1.01253562e-01 -1.21407485e+00 8.65552962e-01 3.41652930e-02
9.06325936e-01 -4.34761941e-01 1.93469957e-01 9.97306854e-02
9.42693710e-01 -3.09021413e-01 5.67427695e-01 1.28474459e-01
1.00688720e+00 8.36760938e-01 7.05752015e-01 8.69867802e-01
6.08091176e-01 7.30150938e-01 6.79550946e-01 -1.97820202e-01
-1.90186307e-01 -2.11349383e-01 8.13837409e-01 1.16446221e+00
8.75629112e-02 -2.10965216e-01 -6.87136710e-01 5.57803094e-01
-2.32842779e+00 -1.31647277e+00 -5.87477326e-01 1.95863032e+00
6.69502854e-01 -4.33934540e-01 5.97320497e-01 1.05635516e-01
7.51412392e-01 1.87625483e-01 -3.95928025e-01 -3.08234811e-01
-4.41807836e-01 -6.28272831e-01 -1.23620965e-01 2.71938860e-01
-1.26533139e+00 8.73190224e-01 4.98026514e+00 8.20214450e-01
-1.11591077e+00 2.90737778e-01 6.45420253e-01 -2.00346693e-01
6.77506998e-02 -4.41828936e-01 -4.09963846e-01 6.46328032e-01
8.36412251e-01 2.41214857e-01 3.20049345e-01 6.67405784e-01
4.13607985e-01 -1.74557626e-01 -8.28110695e-01 1.50433254e+00
8.74576986e-01 -9.07683372e-01 -2.62052476e-01 -4.72279429e-01
7.91161656e-01 -3.06885600e-01 1.97080523e-01 3.73749644e-01
-2.95581013e-01 -1.02007246e+00 5.40409267e-01 5.63617408e-01
4.40125585e-01 -8.27615440e-01 9.55962300e-01 4.03677076e-01
-8.92391086e-01 -2.42548913e-01 -1.37077838e-01 2.37588465e-01
4.46800143e-01 2.65498698e-01 -5.32136738e-01 3.54082823e-01
7.62375474e-01 1.18062401e+00 -6.22327507e-01 8.41335535e-01
1.63450595e-02 5.83942652e-01 1.58172622e-02 -2.63523459e-01
3.11161309e-01 7.04792961e-02 4.94467676e-01 1.58019316e+00
2.41146997e-01 -5.06040379e-02 3.36847782e-01 4.66371387e-01
-1.88951418e-01 5.44879317e-01 -2.95456171e-01 -1.64659411e-01
-1.38491929e-01 1.65586495e+00 -1.85927764e-01 -2.65486091e-01
-5.65691650e-01 1.26286626e+00 2.51335949e-01 1.36599958e-01
-1.14200354e+00 2.55768895e-02 1.04111910e-01 -2.79308945e-01
-1.31567016e-01 8.07645619e-02 -1.05193639e-02 -1.28750873e+00
-1.58583105e-01 -1.20935750e+00 7.42350459e-01 -1.10446322e+00
-1.57352912e+00 3.96603674e-01 -1.34239659e-01 -1.27998400e+00
-1.10426702e-01 -5.06990969e-01 -6.56611502e-01 4.06321317e-01
-1.21257496e+00 -1.73361790e+00 -7.59398460e-01 9.48066533e-01
8.99119973e-01 -3.70872498e-01 4.21698242e-01 3.79090399e-01
-6.91388547e-01 6.76191807e-01 -1.49544075e-01 -2.84408852e-02
9.83819127e-01 -9.58978891e-01 -6.62174463e-01 4.07909364e-01
-3.32823358e-02 3.85419205e-02 7.28855312e-01 -2.06354231e-01
-1.60493100e+00 -6.62654936e-01 7.31468976e-01 -2.79259175e-01
1.03964961e+00 -6.07380504e-03 -8.38212788e-01 1.65179640e-01
8.41227889e-01 -2.56178528e-01 7.68264711e-01 -5.30098937e-02
-4.75890845e-01 9.77212042e-02 -1.00189245e+00 6.43949747e-01
6.50133908e-01 -5.70323288e-01 -3.35991204e-01 4.70310479e-01
3.69561315e-01 -2.54333019e-01 -9.71797585e-01 3.30237001e-01
6.54077470e-01 -1.15492666e+00 7.98117816e-01 -7.82331884e-01
1.21573329e+00 -1.48075372e-02 -3.67125571e-01 -1.09661520e+00
2.59015083e-01 -6.28475845e-01 -2.58825600e-01 1.35412335e+00
2.28881821e-01 1.54943123e-01 6.86573923e-01 4.60180081e-02
-2.91189134e-01 -6.07157707e-01 -8.43072236e-01 -1.49827987e-01
-1.80796355e-01 -1.65229112e-01 -1.30524501e-01 1.24967921e+00
7.27953494e-01 8.84658039e-01 -1.34246802e+00 -2.38444194e-01
6.91111088e-01 2.54664004e-01 1.10227597e+00 -6.44552648e-01
-3.08490872e-01 -5.32671630e-01 -2.51716346e-01 -1.11412942e+00
2.55241424e-01 -6.78909242e-01 -3.95531766e-03 -1.12345862e+00
9.17146564e-01 4.21126038e-01 -8.59871134e-02 4.51278508e-01
-1.68419816e-02 5.84114194e-01 5.72097003e-01 3.21214855e-01
-1.14805889e+00 9.21874702e-01 1.36763382e+00 -2.63945132e-01
-3.81227657e-02 -3.11413854e-01 -3.07759494e-01 8.67009223e-01
8.86517048e-01 1.95303429e-02 -1.92927688e-01 -6.85384050e-02
6.08655848e-02 5.50175548e-01 7.19298005e-01 -8.29414606e-01
2.78492659e-01 -1.36229843e-01 2.84525752e-02 -8.78441155e-01
9.18338537e-01 -6.08581424e-01 -2.90816694e-01 2.00071484e-01
-6.70547128e-01 1.19375393e-01 1.93203371e-02 4.50365484e-01
-4.61485922e-01 -3.16634417e-01 1.04871595e+00 9.78313107e-03
-6.45410717e-01 -1.65381908e-01 -2.65263796e-01 9.09609571e-02
8.95903766e-01 1.01851910e-01 -7.37689257e-01 -7.62721956e-01
-7.40905762e-01 3.82524163e-01 1.25765011e-01 5.02042890e-01
9.83768106e-01 -1.52186024e+00 -9.32822645e-01 -3.16570252e-01
5.15951097e-01 -6.61669314e-01 1.00674152e+00 1.64401674e+00
-1.11929812e-01 6.62771910e-02 -2.59065032e-01 -8.03520977e-01
-1.95831215e+00 5.20087004e-01 2.48950407e-01 -1.18933976e-01
-2.68232167e-01 5.98970711e-01 1.21964961e-01 -4.62226361e-01
2.26690114e-01 4.45874602e-01 -5.19441783e-01 3.28470141e-01
5.44578314e-01 5.64985573e-01 -4.62428868e-01 -1.18302667e+00
-5.13517670e-02 6.02524221e-01 4.02771346e-02 1.00319579e-01
9.90061581e-01 -4.32488114e-01 -2.28420764e-01 7.73845434e-01
1.32754970e+00 -1.92436218e-01 -1.00128245e+00 -2.15570986e-01
-2.48001546e-01 -5.04434347e-01 -1.51602939e-01 -7.46015608e-01
-1.03461993e+00 1.25036812e+00 5.88284910e-01 3.21450293e-01
1.13173997e+00 1.66820481e-01 9.62775707e-01 1.72151104e-01
-3.85948330e-01 -1.29139900e+00 1.04906535e+00 4.05723274e-01
1.13601148e+00 -1.65427482e+00 -2.14587405e-01 -2.13800773e-01
-1.38067067e+00 1.19676745e+00 7.96305239e-01 3.13604236e-01
1.38587475e-01 -1.01201475e-01 1.06491193e-01 -2.76866555e-01
-8.18917751e-01 -2.04700321e-01 3.60958457e-01 2.21782878e-01
5.80549300e-01 -1.61672398e-01 -3.68349314e-01 5.90637922e-01
1.26799926e-01 -2.92983055e-01 6.68917000e-01 4.80167985e-01
-6.25350893e-01 -6.18212283e-01 -5.60767651e-01 1.20147713e-01
-4.97189701e-01 -1.02533609e-01 -7.06689894e-01 5.93172729e-01
-2.50944138e-01 1.31988060e+00 -2.59194374e-01 -6.74344778e-01
2.16286570e-01 -4.60383901e-03 2.39447221e-01 -6.33965433e-02
-7.40331292e-01 3.97541583e-01 2.77568758e-01 -3.46427530e-01
-9.22716498e-01 -6.85026705e-01 -1.03050160e+00 -4.87150192e-01
-9.77724418e-02 -2.00640447e-02 5.08131087e-01 8.64227712e-01
1.00894816e-01 2.37981543e-01 8.57375085e-01 -8.82948458e-01
-5.38353145e-01 -1.12339509e+00 -3.76295537e-01 8.13459456e-01
1.32709071e-01 -5.08863866e-01 -4.79661316e-01 3.27947080e-01] | [13.091118812561035, 5.1366167068481445] |
40fe4ec6-513f-4e21-a4c5-38b23643b091 | deep-decomposition-and-bilinear-pooling | 2205.05880 | null | https://arxiv.org/abs/2205.05880v2 | https://arxiv.org/pdf/2205.05880v2.pdf | Deep Decomposition and Bilinear Pooling Network for Blind Night-Time Image Quality Evaluation | Blind image quality assessment (BIQA), which aims to accurately predict the image quality without any pristine reference information, has been extensively concerned in the past decades. Especially, with the help of deep neural networks, great progress has been achieved. However, it remains less investigated on BIQA for night-time images (NTIs) which usually suffers from complicated authentic distortions such as reduced visibility, low contrast, additive noises, and color distortions. These diverse authentic degradations particularly challenges the design of effective deep neural network for blind NTI quality evaluation (NTIQE). In this paper, we propose a novel deep decomposition and bilinear pooling network (DDB-Net) to better address this issue. The DDB-Net contains three modules, i.e., an image decomposition module, a feature encoding module, and a bilinear pooling module. The image decomposition module is inspired by the Retinex theory and involves decoupling the input NTI into an illumination layer component responsible for illumination information and a reflection layer component responsible for content information. Then, the feature encoding module involves learning feature representations of degradations that are rooted in the two decoupled components separately. Finally, by modeling illumination-related and content-related degradations as two-factor variations, the two feature sets are bilinearly pooled together to form a unified representation for quality prediction. The superiority of the proposed DDB-Net has been well validated by extensive experiments on several benchmark datasets. The source code will be made available soon. | ['Xiongkuo Min', 'Wei Zhou', 'Yudong Mao', 'Guangtao Zhai', 'Jiawu Xu', 'Qiuping Jiang'] | 2022-05-12 | null | null | null | null | ['blind-image-quality-assessment'] | ['computer-vision'] | [-1.78768989e-02 -7.30997562e-01 4.19112235e-01 -3.44720155e-01
-7.07962215e-01 -1.77330375e-01 4.50458884e-01 -2.54977763e-01
-1.74394682e-01 5.96881509e-01 3.95598441e-01 -1.76880304e-02
-1.32284462e-01 -7.31569111e-01 -5.00378132e-01 -1.15113997e+00
1.22176759e-01 -4.66295987e-01 7.77175277e-02 -2.02427745e-01
1.56668276e-01 4.41140741e-01 -1.53291154e+00 2.12804601e-01
1.15028036e+00 1.43781102e+00 1.36724571e-02 3.74742955e-01
1.21542737e-01 6.93818331e-01 -6.70863926e-01 -4.29195613e-01
2.73762912e-01 -4.87178862e-01 -3.22366178e-01 2.08907828e-01
1.99375257e-01 -6.43396080e-01 -7.47269809e-01 1.21323216e+00
8.69877219e-01 6.65803403e-02 4.01246786e-01 -1.13614273e+00
-7.72229314e-01 -1.36428803e-01 -5.35699964e-01 2.96008140e-01
4.68981899e-02 4.93386477e-01 7.08155870e-01 -8.16448867e-01
-3.01730074e-03 1.30414927e+00 4.75011051e-01 1.40010834e-01
-9.24053550e-01 -5.28847158e-01 -9.39624682e-02 7.65140474e-01
-1.16941595e+00 -4.62614089e-01 8.19491446e-01 -2.42343724e-01
6.54822469e-01 1.59571767e-01 6.81021750e-01 7.29563713e-01
2.78440148e-01 7.79902518e-01 1.44573867e+00 -2.64197811e-02
-8.51987973e-02 6.67356327e-02 3.91116599e-03 4.63963062e-01
2.65198529e-01 2.67544538e-01 -1.74850062e-01 2.28073329e-01
7.03099728e-01 5.89660592e-02 -7.78155744e-01 -1.37262732e-01
-9.81802464e-01 4.32948798e-01 9.56738293e-01 1.21972606e-01
-4.19982463e-01 -1.50814831e-01 2.94016778e-01 3.08253437e-01
3.77332509e-01 9.01090354e-02 -1.93803355e-01 2.08444640e-01
-7.08627343e-01 8.56608450e-02 3.52749825e-01 5.81773579e-01
7.51374543e-01 4.63934839e-02 -5.06125093e-01 1.18192708e+00
4.83472288e-01 6.20062709e-01 4.79000241e-01 -7.37021863e-01
5.07319093e-01 5.97622275e-01 2.87352741e-01 -1.18396580e+00
-3.92932802e-01 -7.54701555e-01 -1.21895528e+00 4.35366571e-01
2.37094373e-01 -1.12099797e-01 -8.41298699e-01 1.61936605e+00
9.98755321e-02 -6.28735423e-02 -2.55669206e-02 1.44578040e+00
9.34982300e-01 9.59390879e-01 -6.24526106e-02 -2.42612481e-01
1.40752947e+00 -9.94284451e-01 -8.00518513e-01 -7.58285001e-02
-1.97425127e-01 -8.86586130e-01 9.07383621e-01 5.72494268e-01
-1.24453366e+00 -7.75136292e-01 -1.37034249e+00 -3.56868088e-01
-3.25259864e-01 4.58594471e-01 3.08451086e-01 6.99258327e-01
-1.14207494e+00 3.79437804e-01 -3.99029493e-01 -1.40885800e-01
5.56841910e-01 2.02642843e-01 -2.47632042e-01 -5.00646532e-01
-1.25739467e+00 7.43390739e-01 1.28129469e-02 7.04388559e-01
-1.10544813e+00 -2.50714123e-01 -6.17315710e-01 1.20431416e-01
-6.95805252e-03 -7.54527390e-01 9.20177817e-01 -9.50576782e-01
-1.71607661e+00 4.75895643e-01 -2.06047177e-01 1.15797040e-03
4.74946350e-01 -7.54953399e-02 -8.18875611e-01 1.78414434e-01
-1.77935734e-01 1.16113663e-01 1.10133266e+00 -1.40673113e+00
-5.29316366e-01 -6.66084468e-01 1.95175961e-01 4.24232572e-01
-4.40358162e-01 2.15980142e-01 -7.61492729e-01 -6.78760827e-01
2.08615944e-01 -4.30566669e-01 1.59938008e-01 1.56816706e-01
-2.67340034e-01 1.02959700e-01 5.84267259e-01 -1.18791044e+00
1.23656094e+00 -2.16266561e+00 5.82721792e-02 3.50363851e-02
2.32333422e-01 5.45113862e-01 -3.46504211e-01 1.81597516e-01
-1.92447424e-01 -1.47971645e-01 -3.49409074e-01 -2.05462232e-01
6.51704669e-02 -7.90260285e-02 -1.44941688e-01 6.80220425e-01
3.48429888e-01 7.34406412e-01 -7.12079585e-01 -1.76569954e-01
2.23898411e-01 6.89796805e-01 -2.35343024e-01 5.88766038e-01
1.57360658e-01 4.51227307e-01 -7.32039511e-02 7.34589458e-01
1.31344128e+00 -4.34405953e-02 -2.93589473e-01 -6.54745698e-01
-3.25842530e-01 5.04832193e-02 -1.07211483e+00 1.47171402e+00
-4.02189732e-01 5.12628376e-01 2.99506456e-01 -8.75620127e-01
9.79677737e-01 3.69640350e-01 1.29930332e-01 -1.17451060e+00
2.93760687e-01 2.33946711e-01 1.62082743e-02 -7.37351656e-01
2.41659030e-01 -2.03994945e-01 3.76728326e-01 2.51005560e-01
5.89111075e-02 1.96332797e-01 -3.63920704e-02 -2.40355104e-01
8.58470440e-01 -1.92237757e-02 -7.02350885e-02 9.23605636e-02
1.04848230e+00 -5.43940604e-01 7.69819856e-01 1.73368856e-01
-5.51081598e-01 8.33180428e-01 3.49572331e-01 -3.89923245e-01
-9.71462607e-01 -1.06651545e+00 -2.30655849e-01 5.87335408e-01
6.58523381e-01 -7.29004890e-02 -5.85636318e-01 -2.53225386e-01
-1.95684969e-01 1.48852900e-01 -3.99352163e-01 -3.01889092e-01
-3.72130722e-01 -1.10060024e+00 2.56543130e-01 3.00387681e-01
1.29657793e+00 -1.09896696e+00 -7.93337524e-02 -2.85738986e-02
-5.25851309e-01 -8.76992285e-01 -4.34908748e-01 -2.24246413e-01
-6.14757717e-01 -9.67491746e-01 -1.14258850e+00 -7.24123240e-01
5.40844202e-01 7.40534484e-01 8.66647005e-01 2.83231232e-02
-2.35167280e-01 7.22498596e-02 -2.16066435e-01 -2.85851866e-01
7.68767344e-03 -4.56749111e-01 -8.18163380e-02 5.33026159e-01
1.61277726e-01 -5.76389313e-01 -1.16333985e+00 5.48257053e-01
-1.11802161e+00 -8.36424530e-02 8.35171103e-01 9.14903164e-01
4.75065351e-01 4.84786421e-01 3.48292142e-01 -1.21476367e-01
6.68598056e-01 -3.38955492e-01 -6.16829097e-01 3.44192356e-01
-3.49163204e-01 -2.18185291e-01 6.30521953e-01 -4.51079980e-02
-1.48361659e+00 -5.09807229e-01 -2.10990757e-01 -2.93054193e-01
-1.17586724e-01 3.67843688e-01 -8.91111732e-01 -3.14244628e-01
3.48204434e-01 4.23215955e-01 -5.93121797e-02 -6.76807940e-01
1.18976951e-01 9.67584550e-01 6.72322571e-01 -3.40945601e-01
9.74051893e-01 3.50637764e-01 -2.56622016e-01 -6.63048863e-01
-5.53432345e-01 -3.80950123e-01 -1.54985234e-01 -3.29766482e-01
7.69140124e-01 -1.13919520e+00 -7.31620789e-01 1.28882146e+00
-1.16352916e+00 -6.97100982e-02 1.23984374e-01 4.60916847e-01
-2.74915665e-01 5.95887959e-01 -6.67754412e-01 -7.23776460e-01
-4.61327046e-01 -1.37050092e+00 6.60113454e-01 7.71773875e-01
7.88511395e-01 -6.08287990e-01 -7.36472756e-02 5.47573745e-01
6.15743756e-01 -9.71768983e-03 9.10147905e-01 1.97063550e-01
-7.92959630e-01 -1.82049870e-01 -9.90619719e-01 1.02436411e+00
2.75966018e-01 -2.87073612e-01 -1.21402323e+00 -5.32922328e-01
3.73091877e-01 -2.57883817e-01 9.80601072e-01 3.99786204e-01
1.20118594e+00 -2.84850031e-01 3.05866748e-01 9.76830661e-01
1.72581720e+00 2.61950821e-01 1.12033665e+00 5.43331206e-01
6.10405266e-01 4.54121709e-01 3.42908561e-01 4.03695613e-01
4.52773750e-01 6.04558945e-01 6.79457009e-01 -3.84673089e-01
-4.30399895e-01 1.14083149e-01 4.35720176e-01 8.78196895e-01
-1.86928153e-01 -2.12491512e-01 -5.58658540e-01 5.20146430e-01
-1.37408388e+00 -7.78474033e-01 -2.56892070e-02 2.19620180e+00
7.96186745e-01 -1.46165997e-01 -6.41609654e-02 2.69190967e-01
7.30346262e-01 2.75492609e-01 -7.53683686e-01 -2.01859340e-01
-4.34472293e-01 9.67236795e-03 3.21329951e-01 6.85091168e-02
-1.15913904e+00 3.47173423e-01 4.82764959e+00 7.17739522e-01
-1.09772301e+00 1.43154010e-01 7.13329673e-01 1.90906316e-01
-1.91681311e-01 -2.06387088e-01 -3.70548397e-01 6.53364241e-01
5.56963503e-01 1.32625893e-01 8.17037046e-01 2.14828029e-01
4.68424320e-01 -6.54805452e-02 -3.76646876e-01 1.21465468e+00
1.54974237e-01 -6.58477068e-01 1.45186439e-01 -2.49846652e-03
6.56860888e-01 -3.43548879e-02 4.24559385e-01 5.13367988e-02
-1.42393172e-01 -9.11871433e-01 5.94147742e-01 9.65608597e-01
8.45582545e-01 -7.13206112e-01 9.96147156e-01 6.45083711e-02
-1.12142563e+00 -5.06242990e-01 -6.53355479e-01 8.54722783e-03
-6.71474040e-02 7.76705205e-01 9.05895606e-02 9.61192191e-01
1.02578843e+00 8.47352982e-01 -6.46501720e-01 1.70308530e+00
-4.76324052e-01 5.39132357e-01 8.21119621e-02 4.25081581e-01
4.90701795e-02 -4.32306737e-01 6.10266745e-01 8.78762007e-01
4.04680997e-01 1.78113565e-01 -2.99057871e-01 8.53013098e-01
-1.16046295e-01 7.12263212e-02 -4.01388779e-02 2.08151340e-01
-4.62949760e-02 1.41112423e+00 -1.61565840e-01 -1.25045642e-01
-6.47513866e-01 1.09434676e+00 -4.25871648e-02 7.41401672e-01
-6.78791583e-01 -7.27434695e-01 9.38696921e-01 -2.62857556e-01
3.23344886e-01 -1.12479351e-01 -2.13768482e-01 -1.41931415e+00
3.33727986e-01 -1.12807500e+00 7.85085373e-04 -1.09457564e+00
-1.46535861e+00 7.16978669e-01 -4.54751581e-01 -1.53273463e+00
5.64107716e-01 -7.65459895e-01 -7.18014359e-01 1.53942263e+00
-2.18337774e+00 -1.09463847e+00 -9.15282786e-01 7.75786459e-01
1.76153958e-01 -3.12523730e-02 5.78296721e-01 6.06448233e-01
-1.16678655e+00 6.39209270e-01 4.62836802e-01 1.75831437e-01
9.23469424e-01 -1.06162202e+00 6.04798757e-02 1.14243448e+00
-4.72032785e-01 4.22248125e-01 3.01939279e-01 -2.61253268e-01
-1.30685532e+00 -1.14657736e+00 5.58097839e-01 1.93944961e-01
3.72316301e-01 -1.77032147e-02 -1.10345232e+00 7.07997829e-02
2.13628188e-01 1.84045479e-01 3.65182132e-01 -3.15580547e-01
-4.88609314e-01 -8.07538033e-01 -1.11532331e+00 3.06111723e-01
7.41836250e-01 -6.25598907e-01 -3.54107976e-01 1.16217852e-01
4.88892406e-01 -2.28300601e-01 -8.01933408e-01 3.68740737e-01
5.38876474e-01 -1.26459074e+00 9.88788664e-01 -1.64299142e-02
4.66139585e-01 -6.93241119e-01 -1.25133082e-01 -1.41371608e+00
-4.97516066e-01 -2.49895200e-01 8.94156173e-02 1.30840719e+00
-9.02794302e-02 -7.37695336e-01 1.64629981e-01 2.91404307e-01
-2.90162206e-01 -5.89857340e-01 -7.64031768e-01 -6.33957386e-01
-1.82984874e-01 -1.33476675e-01 8.40683758e-01 4.57940102e-01
-4.48552549e-01 6.67789653e-02 -4.78660643e-01 5.04313529e-01
7.23584712e-01 8.31963718e-02 5.34189165e-01 -1.00715029e+00
-2.10062742e-01 -5.11801124e-01 -3.80369723e-01 -1.06811941e+00
-4.27020967e-01 -5.76993763e-01 2.42412001e-01 -1.90740275e+00
2.44752049e-01 -2.77978450e-01 -6.76028728e-01 2.02412993e-01
-4.42584068e-01 4.85998631e-01 5.98000586e-02 3.93638909e-01
-4.66543585e-01 9.67030644e-01 1.55252695e+00 -5.03490090e-01
-8.05918947e-02 -4.42434661e-02 -7.80403972e-01 4.11790282e-01
8.10010672e-01 5.10679521e-02 -3.54772627e-01 -7.75541663e-01
1.46511033e-01 6.22071773e-02 5.92732847e-01 -1.41064072e+00
1.95218310e-01 1.91088006e-01 6.76912665e-01 -3.15908611e-01
2.97034264e-01 -8.08958232e-01 6.22402004e-04 1.85728952e-01
7.78805390e-02 -2.22670570e-01 2.29392543e-01 5.26277602e-01
-6.53361797e-01 3.23482417e-02 9.97745752e-01 -1.40799740e-02
-6.54595733e-01 6.65871143e-01 -1.00596048e-01 -3.56985778e-01
5.76066077e-01 -7.65538663e-02 -6.28371656e-01 -3.38738590e-01
-4.06654090e-01 1.96614146e-01 3.90043944e-01 3.49999726e-01
8.42688799e-01 -1.45794737e+00 -9.15210545e-01 4.03348744e-01
1.92012280e-01 -2.22509176e-01 8.73341143e-01 9.11346257e-01
-5.45374513e-01 2.05460355e-01 -5.48642457e-01 -2.49221474e-01
-9.18301880e-01 5.08309424e-01 6.10252559e-01 -1.93538263e-01
-3.39104772e-01 7.40002215e-01 3.17243665e-01 -7.23888502e-02
2.96266168e-01 -1.45120278e-01 -6.06519163e-01 -4.70077526e-03
8.61237168e-01 5.84315360e-01 3.42072546e-01 -8.85472000e-01
-1.50685430e-01 6.92813635e-01 1.13576002e-01 1.50370374e-01
1.35065985e+00 -5.60689986e-01 -5.54335058e-01 1.57074168e-01
1.34438801e+00 -3.28821868e-01 -1.35275853e+00 -4.60939884e-01
-4.76076305e-01 -6.35189235e-01 4.10322458e-01 -1.22457922e+00
-1.41550016e+00 1.33167708e+00 1.08628845e+00 1.59885913e-01
1.86959183e+00 -5.56820154e-01 1.02461624e+00 -4.73061986e-02
1.22249678e-01 -7.35338271e-01 1.07306674e-01 3.34688932e-01
1.19085455e+00 -1.25654292e+00 -1.08002536e-01 -1.35916220e-02
-2.57316381e-01 1.07892871e+00 5.16742051e-01 8.74569640e-02
5.91108680e-01 -4.14937645e-01 2.78654784e-01 6.67095855e-02
-1.91094592e-01 -2.01286957e-01 5.88883519e-01 6.37149334e-01
1.63098246e-01 2.38075797e-02 -3.08701158e-01 9.26892996e-01
1.35605067e-01 1.04595080e-01 2.11636439e-01 3.42435658e-01
-3.60352695e-01 -8.88633728e-01 -7.05274165e-01 3.50065321e-01
-3.08624476e-01 -1.53146923e-01 1.38747513e-01 2.40872338e-01
4.87446904e-01 1.30776322e+00 -1.37425363e-01 -5.73919415e-01
4.65226442e-01 -3.26126099e-01 2.73005605e-01 -6.98802546e-02
-4.39039469e-01 5.31910807e-02 -3.43943417e-01 -6.73290670e-01
-4.77453142e-01 -3.51677537e-01 -7.24762499e-01 -3.24222982e-01
-1.07443310e-01 -1.07190423e-01 7.11706579e-01 7.00863361e-01
1.31222993e-01 6.05900347e-01 1.00913692e+00 -9.71670151e-01
-3.58328789e-01 -1.03422165e+00 -8.07943642e-01 3.69152606e-01
6.99519336e-01 -6.33838952e-01 -4.79470581e-01 -1.53465435e-01] | [11.82862377166748, -1.887738585472107] |
64c707f6-e4b7-406e-8611-c6fdb4817121 | image-search-with-text-feedback-by | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Chen_Image_Search_With_Text_Feedback_by_Visiolinguistic_Attention_Learning_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Chen_Image_Search_With_Text_Feedback_by_Visiolinguistic_Attention_Learning_CVPR_2020_paper.pdf | Image Search With Text Feedback by Visiolinguistic Attention Learning | Image search with text feedback has promising impacts in various real-world applications, such as e-commerce and internet search. Given a reference image and text feedback from user, the goal is to retrieve images that not only resemble the input image, but also change certain aspects in accordance with the given text. This is a challenging task as it requires the synergistic understanding of both image and text. In this work, we tackle this task by a novel Visiolinguistic Attention Learning (VAL) framework. Specifically, we propose a composite transformer that can be seamlessly plugged in a CNN to selectively preserve and transform the visual features conditioned on language semantics. By inserting multiple composite transformers at varying depths, VAL is incentive to encapsulate the multi-granular visiolinguistic information, thus yielding an expressive representation for effective image search. We conduct comprehensive evaluation on three datasets: Fashion200k, Shoes and FashionIQ. Extensive experiments show our model exceeds existing approaches on all datasets, demonstrating consistent superiority in coping with various text feedbacks, including attribute-like and natural language descriptions.
| [' Loris Bazzani', ' Shaogang Gong', 'Yanbei Chen'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['deep-attention', 'deep-attention'] | ['computer-vision', 'natural-language-processing'] | [ 2.61998713e-01 -4.37654525e-01 -2.19204843e-01 -4.70238149e-01
-4.12557423e-01 -6.14467204e-01 6.12011850e-01 1.21262342e-01
-4.54895794e-01 2.86813706e-01 4.57375139e-01 -1.24898620e-01
-8.38495418e-02 -5.47472656e-01 -8.24243963e-01 -3.98751229e-01
6.66653693e-01 1.45039737e-01 2.13173658e-01 -4.64614362e-01
2.88027942e-01 2.02917457e-01 -1.50615180e+00 5.86796820e-01
8.97184670e-01 1.28578353e+00 7.00185418e-01 2.10317612e-01
-2.06566691e-01 8.22595954e-01 -1.53073668e-01 -6.84432983e-01
1.63013816e-01 -2.46161073e-01 -8.41605842e-01 3.58068526e-01
5.62951624e-01 -4.14109558e-01 -4.37433511e-01 1.17895579e+00
5.75191736e-01 3.11874032e-01 4.70937103e-01 -1.03433251e+00
-1.49662697e+00 4.47007537e-01 -7.10739017e-01 1.04620375e-01
4.41336781e-01 4.68615115e-01 1.10935807e+00 -9.77561831e-01
6.46897078e-01 1.69307220e+00 2.21462604e-02 4.86340344e-01
-1.25260556e+00 -7.01587975e-01 5.14204502e-01 4.93763685e-01
-1.19659138e+00 -3.67197633e-01 1.06649637e+00 -2.78052837e-01
6.69203877e-01 3.89501631e-01 6.19483829e-01 9.99206543e-01
1.39099494e-01 1.22583389e+00 1.09344351e+00 -3.20780933e-01
-2.05253720e-01 5.26650727e-01 -1.90465853e-01 7.89973557e-01
-2.74084717e-01 6.05784636e-03 -7.92211771e-01 2.45705947e-01
8.00857842e-01 3.13935459e-01 -1.99663341e-01 -5.41932046e-01
-1.33972168e+00 6.72628939e-01 7.93262124e-01 1.08466066e-01
-5.01770437e-01 2.77485520e-01 5.81998765e-01 4.58824784e-01
4.58219022e-01 3.51233631e-01 -4.03240710e-01 3.51242334e-01
-5.69423318e-01 2.61896670e-01 2.28170469e-01 1.03165257e+00
7.65984654e-01 -1.47279367e-01 -7.25104868e-01 1.03115416e+00
3.58252257e-01 6.92748666e-01 5.61778188e-01 -8.18144619e-01
4.60350245e-01 8.68783891e-01 6.17472343e-02 -1.35482359e+00
-2.41929330e-02 -4.87341166e-01 -9.19442773e-01 -1.36520583e-02
-7.36454353e-02 5.06533682e-01 -1.02105153e+00 1.90591609e+00
2.71888226e-01 -3.76882911e-01 -1.12723082e-01 1.17717564e+00
1.10717762e+00 6.38955712e-01 2.25073904e-01 -2.55570766e-02
1.59134817e+00 -1.21192336e+00 -7.88953841e-01 -4.01051074e-01
1.83727905e-01 -8.08049619e-01 1.70282769e+00 1.44344732e-01
-1.09899139e+00 -7.39453614e-01 -8.30656290e-01 -5.62056541e-01
-5.08479834e-01 3.91318649e-01 3.94213736e-01 1.56988755e-01
-1.16810763e+00 9.90150347e-02 -1.65231422e-01 -5.24239838e-01
3.13574344e-01 2.14590088e-01 -3.61581415e-01 -2.59775758e-01
-1.38559103e+00 7.34109640e-01 3.82422715e-01 2.52980292e-01
-7.47556627e-01 -4.24918622e-01 -9.57155704e-01 1.50599539e-01
6.14583075e-01 -8.60465884e-01 1.12174320e+00 -1.16355002e+00
-1.23660564e+00 1.13457787e+00 -2.05550298e-01 -2.07612723e-01
6.73557699e-01 -2.84684807e-01 -2.34060466e-01 1.54038399e-01
3.44194919e-01 9.80305374e-01 9.58495975e-01 -1.39666200e+00
-4.90680009e-01 -5.84446251e-01 4.71206665e-01 6.25894248e-01
-6.93163753e-01 1.42033949e-01 -9.15333211e-01 -8.27326715e-01
-1.21473946e-01 -6.46436453e-01 -2.78121866e-02 4.96480674e-01
-3.74538213e-01 -4.10042226e-01 7.61671364e-01 -7.74530292e-01
1.12388968e+00 -2.14145494e+00 3.15351039e-01 -8.35346803e-02
1.80467412e-01 1.33751974e-01 -1.93485618e-01 3.78921360e-01
2.67900497e-01 1.84173793e-01 1.57861039e-01 -1.94503710e-01
1.07583992e-01 9.83698294e-02 -4.78558987e-01 4.77943644e-02
1.74224630e-01 1.52518022e+00 -9.66535091e-01 -6.99890852e-01
3.79071057e-01 1.79962724e-01 -5.23283124e-01 1.66054294e-02
-5.66503763e-01 3.80622149e-01 -6.30490601e-01 7.16681302e-01
4.21224356e-01 -4.41184074e-01 -2.18937427e-01 -5.95123291e-01
9.26904380e-02 -2.36107692e-01 -7.12733746e-01 1.94424832e+00
-5.22504687e-01 5.61266840e-01 -1.01297796e-02 -8.83108616e-01
9.49270666e-01 -8.62113461e-02 1.49865359e-01 -1.60922098e+00
-1.15513802e-04 -5.84858991e-02 -4.59358722e-01 -6.75343335e-01
6.46919370e-01 9.66810510e-02 -2.69461632e-01 4.39218640e-01
-5.37655205e-02 2.74890307e-02 5.42133264e-02 3.44201595e-01
4.90367621e-01 1.00011177e-01 1.07888073e-01 -2.22562581e-01
7.60045230e-01 -9.42610428e-02 2.62279242e-01 8.02958071e-01
-1.40313581e-01 2.67578661e-01 2.27068365e-01 -5.49540043e-01
-9.12330210e-01 -9.06059265e-01 3.02493840e-01 1.56390774e+00
7.75236428e-01 -2.14963973e-01 -4.27834004e-01 -5.09918749e-01
1.64569393e-02 4.89867270e-01 -8.22523355e-01 -5.26291370e-01
-2.44736969e-01 -2.29089037e-01 2.57976383e-01 5.49768388e-01
9.80546474e-01 -1.51685822e+00 -4.99876708e-01 -6.74147829e-02
-4.35547024e-01 -1.12561119e+00 -1.10762346e+00 -7.65742585e-02
-4.34909999e-01 -7.19127655e-01 -6.69469774e-01 -1.14564836e+00
5.89785278e-01 5.27775168e-01 1.14752913e+00 1.44771457e-01
-4.52897042e-01 5.42482078e-01 -4.48388606e-01 -1.94283321e-01
-1.65554911e-01 -9.54179391e-02 -2.40426660e-01 2.48646542e-01
1.83113918e-01 -5.31496778e-02 -7.86703050e-01 3.62200707e-01
-1.23165917e+00 4.97662872e-01 7.05153048e-01 1.12076783e+00
6.90383554e-01 -1.01888604e-01 4.49060231e-01 -6.35376394e-01
9.83105004e-01 -2.13842332e-01 -2.84478396e-01 8.01729679e-01
-6.43190622e-01 2.18286335e-01 5.37734449e-01 -5.11160970e-01
-1.25874805e+00 5.59120513e-02 1.26966700e-01 -7.05054104e-01
1.98999688e-01 5.15740335e-01 -2.77776599e-01 -2.41878796e-02
3.65708798e-01 7.84170508e-01 -1.34802520e-01 -4.29290980e-01
6.10397637e-01 5.33285618e-01 6.19243860e-01 -5.74444652e-01
4.63142544e-01 4.57673669e-01 -3.59664768e-01 -5.75600684e-01
-8.28878641e-01 -3.55967700e-01 -5.07012486e-01 -3.81046146e-01
9.00937319e-01 -8.88667703e-01 -9.16853309e-01 5.39715290e-01
-1.15499282e+00 -1.59030974e-01 -1.07444778e-01 1.37413095e-03
-4.88939822e-01 3.63912135e-01 -4.14627045e-01 -5.01308143e-01
-5.55597603e-01 -1.31173015e+00 1.27932990e+00 3.41143847e-01
2.52424747e-01 -7.85634279e-01 -4.13676500e-01 4.60646749e-01
3.33857715e-01 -9.43539739e-02 1.21239018e+00 -4.17426556e-01
-7.18605042e-01 1.76825523e-01 -6.74283683e-01 2.70633757e-01
1.82221770e-01 -3.65592688e-01 -6.94151163e-01 -2.63383955e-01
-2.44086400e-01 -7.94429541e-01 1.20676732e+00 1.32889301e-01
1.36848927e+00 -4.08423662e-01 -2.29811355e-01 5.41430354e-01
1.28850985e+00 1.74688265e-01 4.63310301e-01 2.53763944e-01
6.70441449e-01 6.84687853e-01 7.59656847e-01 3.13028395e-01
6.74003065e-01 7.75539279e-01 6.46524966e-01 -3.01211327e-01
-2.48059526e-01 -6.56040490e-01 1.41990677e-01 4.96395379e-01
2.96014845e-01 -2.81142205e-01 -5.53349376e-01 5.47798038e-01
-2.05822873e+00 -8.75982225e-01 3.63305688e-01 1.77507496e+00
1.03047526e+00 -1.31623089e-01 -2.43824363e-01 -4.27232385e-01
6.56761050e-01 3.89134854e-01 -1.12221611e+00 -2.82946765e-01
-2.20858023e-01 -3.09715003e-01 2.28395164e-01 1.76554292e-01
-1.01420462e+00 1.23518682e+00 5.82509804e+00 9.60538208e-01
-1.16251206e+00 1.06805891e-01 7.76001334e-01 -1.77675188e-02
-6.83521807e-01 -2.31258571e-01 -4.41949725e-01 3.07100922e-01
4.68233181e-03 -2.96648055e-01 6.29087567e-01 4.48429048e-01
2.37453654e-01 1.40013754e-01 -8.43825877e-01 1.17782950e+00
2.68758982e-01 -1.24758554e+00 6.47024214e-01 -2.02326164e-01
6.07943296e-01 -3.47409278e-01 4.85330015e-01 3.49331260e-01
2.66637266e-01 -9.64527249e-01 1.05210972e+00 6.34952545e-01
9.58553433e-01 -4.70263362e-01 2.95962453e-01 3.22394043e-01
-1.28240454e+00 -2.41113067e-01 -2.37943962e-01 2.20285028e-01
7.74292601e-03 -6.50062552e-03 -4.51496392e-01 4.66281056e-01
9.86752927e-01 9.21322703e-01 -8.31528604e-01 7.27271318e-01
-1.64716505e-02 2.74049751e-02 4.58289459e-02 -2.68647671e-01
3.43268007e-01 -9.89723280e-02 3.34369034e-01 9.83861566e-01
3.96673679e-02 4.12719250e-02 4.15065348e-01 1.24818647e+00
-3.18254888e-01 4.78066593e-01 -4.54589516e-01 -1.23032853e-01
2.71905422e-01 1.09660637e+00 -6.12774670e-01 -3.83072257e-01
-4.60810721e-01 1.46317458e+00 4.60343361e-01 6.55654192e-01
-6.41152322e-01 -1.46992609e-01 5.27424216e-01 -6.50259107e-02
2.53405005e-01 1.15430988e-01 1.52768614e-02 -1.18586874e+00
3.57756853e-01 -8.26541901e-01 3.30457538e-01 -1.31574166e+00
-1.16888762e+00 5.52091718e-01 -7.08935931e-02 -9.90420997e-01
3.20584364e-02 -4.71964002e-01 -2.46683821e-01 8.64321113e-01
-1.66226447e+00 -1.72300172e+00 -5.30850470e-01 9.14467394e-01
1.01018918e+00 -2.04062238e-02 5.17532825e-01 2.87748396e-01
-2.70974308e-01 6.26226425e-01 -7.68014938e-02 -4.82643628e-03
8.37575972e-01 -1.03072488e+00 2.69474149e-01 5.14618218e-01
1.67180568e-01 6.11880720e-01 5.62750101e-01 -6.74496949e-01
-1.46920490e+00 -1.12790823e+00 6.38589203e-01 -1.19064577e-01
5.19716680e-01 -3.58286291e-01 -7.60575593e-01 6.24537230e-01
4.39372689e-01 -3.32486220e-02 1.61783680e-01 -2.39653051e-01
-4.76025760e-01 -3.36238474e-01 -7.54858732e-01 8.49219978e-01
1.23450029e+00 -7.92528927e-01 -4.66936678e-01 3.36076796e-01
1.15052617e+00 -3.59022588e-01 -5.39413333e-01 3.27304959e-01
7.21327841e-01 -8.34747255e-01 1.15166521e+00 -7.01081991e-01
5.11666715e-01 -2.15006411e-01 -1.92349255e-01 -1.20122135e+00
-4.60067481e-01 -3.11804801e-01 2.64326543e-01 1.29273093e+00
1.55005455e-01 -4.10424113e-01 3.76691788e-01 5.55797040e-01
6.71484098e-02 -7.09667742e-01 -5.78985751e-01 -3.27384233e-01
-1.65663153e-01 -1.11558773e-01 5.62144518e-01 7.82673120e-01
-1.73443660e-01 4.68803465e-01 -7.65739799e-01 -1.79482162e-01
3.59964907e-01 4.04275179e-01 5.73320150e-01 -1.00753164e+00
-2.21714750e-02 -5.52252591e-01 -3.18110883e-01 -1.50036871e+00
2.08848417e-01 -1.02659094e+00 2.00410765e-02 -1.68322504e+00
7.32815623e-01 -1.98725730e-01 -5.37586629e-01 6.18078172e-01
-3.11996877e-01 2.29942217e-01 3.54158401e-01 2.01873913e-01
-1.00645578e+00 8.14098656e-01 1.72945106e+00 -6.24992013e-01
2.02935692e-02 -2.86662668e-01 -1.08777988e+00 3.63289088e-01
4.61908877e-01 6.59240335e-02 -7.36571610e-01 -7.83494174e-01
3.92759502e-01 1.62664026e-01 7.46860921e-01 -3.40441823e-01
4.23433721e-01 -3.44592780e-01 5.40551722e-01 -7.43110716e-01
2.46554598e-01 -8.68446469e-01 -1.51179314e-01 3.49729121e-01
-9.30262029e-01 3.60558867e-01 2.03762338e-01 8.17107916e-01
-4.33423847e-01 2.13509262e-01 4.59822834e-01 -3.64494681e-01
-1.38546991e+00 5.13367534e-01 5.17147630e-02 -2.35588625e-02
9.39856052e-01 -1.48619935e-01 -3.42511296e-01 -5.83516657e-01
-6.72764659e-01 6.89668000e-01 5.25501668e-01 9.83854234e-01
9.37524676e-01 -1.56564665e+00 -6.53362513e-01 3.35282713e-01
6.22961402e-01 -2.61940122e-01 5.03641427e-01 6.08126879e-01
-1.60282239e-01 5.41133583e-01 -3.24160159e-01 -6.25581861e-01
-1.24583483e+00 8.64284873e-01 2.91458398e-01 -9.69473049e-02
-4.53270525e-01 7.22212315e-01 8.93503129e-01 -4.40161735e-01
2.85534322e-01 -2.86824435e-01 -2.59821445e-01 -2.30417959e-02
4.40826654e-01 -2.16474339e-01 -2.43599519e-01 -6.84103012e-01
-2.27132306e-01 6.39100254e-01 -3.38707566e-01 -4.39117029e-02
1.00290906e+00 -7.38183856e-01 -1.37995258e-01 2.84223914e-01
1.20557356e+00 -3.63904446e-01 -1.15280461e+00 -8.37435246e-01
-1.41212091e-01 -6.40718222e-01 2.94873025e-02 -1.07224798e+00
-1.24622667e+00 9.06764150e-01 6.80895925e-01 -5.12782373e-02
1.49699938e+00 2.83542216e-01 6.72914445e-01 4.59476829e-01
1.70815602e-01 -1.05774605e+00 6.54506981e-01 2.15426758e-01
1.30821586e+00 -1.40672135e+00 -1.98841095e-01 -1.84807003e-01
-9.71763492e-01 8.23625088e-01 9.09519732e-01 1.35945335e-01
3.48324537e-01 -3.38966697e-01 1.41463233e-02 -4.03710335e-01
-1.01150203e+00 -3.51032436e-01 5.51977098e-01 3.32000196e-01
1.90059498e-01 -6.82682618e-02 -1.29306525e-01 5.05498946e-01
1.34718135e-01 -4.71189916e-02 -9.28194970e-02 6.49350047e-01
-4.62864906e-01 -7.87828326e-01 -6.41695857e-02 4.74864811e-01
-2.04392701e-01 -4.23251897e-01 -6.59218550e-01 5.28677464e-01
-7.89392591e-02 7.80867994e-01 -6.98633566e-02 -4.64244902e-01
4.29945052e-01 -1.13791101e-01 3.61831099e-01 -3.59734267e-01
-4.85728323e-01 4.17497419e-02 -3.65361094e-01 -6.13890171e-01
-4.55639869e-01 -4.94172513e-01 -9.25056458e-01 -3.14294659e-02
-3.17188323e-01 -2.18153834e-01 5.42573094e-01 8.36457014e-01
4.49307173e-01 6.26963198e-01 4.25060213e-01 -4.43365932e-01
-5.67409098e-01 -8.39290261e-01 -5.35137117e-01 8.37227106e-01
3.32319319e-01 -7.40882516e-01 1.05515763e-01 3.73585016e-01] | [10.782570838928223, 1.3958505392074585] |
31ef3674-6cf2-434a-a240-a9802ae08dc6 | a-morphable-face-albedo-model | 2004.02711 | null | https://arxiv.org/abs/2004.02711v2 | https://arxiv.org/pdf/2004.02711v2.pdf | A Morphable Face Albedo Model | In this paper, we bring together two divergent strands of research: photometric face capture and statistical 3D face appearance modelling. We propose a novel lightstage capture and processing pipeline for acquiring ear-to-ear, truly intrinsic diffuse and specular albedo maps that fully factor out the effects of illumination, camera and geometry. Using this pipeline, we capture a dataset of 50 scans and combine them with the only existing publicly available albedo dataset (3DRFE) of 23 scans. This allows us to build the first morphable face albedo model. We believe this is the first statistical analysis of the variability of facial specular albedo maps. This model can be used as a plug in replacement for the texture model of the Basel Face Model (BFM) or FLAME and we make the model publicly available. We ensure careful spectral calibration such that our model is built in a linear sRGB space, suitable for inverse rendering of images taken by typical cameras. We demonstrate our model in a state of the art analysis-by-synthesis 3DMM fitting pipeline, are the first to integrate specular map estimation and outperform the BFM in albedo reconstruction. | ['Joshua Tenenbaum', 'William A. P. Smith', 'Bernard Tiddeman', 'Alassane Seck', 'Hannah Dee', 'Bernhard Egger'] | 2020-04-06 | a-morphable-face-albedo-model-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Smith_A_Morphable_Face_Albedo_Model_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Smith_A_Morphable_Face_Albedo_Model_CVPR_2020_paper.pdf | cvpr-2020-6 | ['art-analysis'] | ['computer-vision'] | [ 3.86214942e-01 -3.20535935e-02 6.56672776e-01 -5.93891203e-01
-5.74201286e-01 -3.50116044e-01 4.84926552e-01 -5.87200701e-01
1.29594862e-01 1.45595878e-01 2.98392996e-02 4.90178429e-02
1.48143157e-01 -4.74852860e-01 -7.54388511e-01 -8.44367385e-01
4.18449044e-01 6.93123698e-01 -1.16803301e-02 -1.89750269e-01
-2.77306080e-01 9.59360123e-01 -1.94070029e+00 7.15007782e-02
6.56874299e-01 1.13780892e+00 -2.43897676e-01 5.72925866e-01
-2.27831416e-02 6.59166098e-01 -1.88525364e-01 -5.81352830e-01
5.49590945e-01 -3.38227600e-01 -2.35964701e-01 3.60946327e-01
1.28813589e+00 -5.97947299e-01 4.76024114e-02 5.29807031e-01
5.96172929e-01 -2.66465575e-01 6.68060899e-01 -1.08138192e+00
-2.68476844e-01 -1.45560130e-01 -7.07517147e-01 -5.31156659e-01
4.17160720e-01 4.10451144e-01 4.50459749e-01 -9.03111458e-01
7.14610100e-01 1.36709261e+00 7.59220302e-01 6.52048409e-01
-1.40442193e+00 -6.80700302e-01 -3.77459168e-01 -2.91177511e-01
-1.26979005e+00 -1.12314892e+00 9.14119959e-01 -4.49602395e-01
6.10928178e-01 4.22170401e-01 1.03624666e+00 8.87975514e-01
-1.27676710e-01 2.15510130e-01 1.66391468e+00 -7.95898318e-01
2.20568292e-02 3.70169163e-01 -2.82225013e-01 9.13894296e-01
5.93751818e-02 8.23414773e-02 -7.98191607e-01 -3.31366241e-01
6.37696743e-01 -2.04586670e-01 -2.01880485e-01 -4.16370362e-01
-5.92471838e-01 2.61282325e-01 1.51782230e-01 -4.40893739e-01
-2.49671683e-01 2.19452873e-01 -4.51453149e-01 3.14738363e-01
9.84286010e-01 -1.86267514e-02 -3.61966968e-01 3.10738832e-01
-1.11601961e+00 3.77055436e-01 7.39397049e-01 8.00411344e-01
1.17993474e+00 1.44472346e-01 4.30016696e-01 1.15372753e+00
7.43022203e-01 1.24810660e+00 -2.15571940e-01 -1.47377050e+00
-2.53319144e-01 2.56206304e-01 1.27443066e-02 -4.92306978e-01
-1.50601223e-01 2.63695605e-02 9.00009833e-03 6.23687088e-01
4.00681704e-01 -4.90723178e-02 -9.47080851e-01 1.32036746e+00
9.62683737e-01 3.07133701e-02 -3.57239038e-01 8.43644857e-01
9.96332347e-01 1.64646283e-01 -3.66923809e-01 -1.62873343e-01
1.38181651e+00 -5.92433870e-01 -5.22343874e-01 -6.87451512e-02
3.03335667e-01 -1.26136959e+00 8.19021225e-01 4.84455884e-01
-1.25305104e+00 -1.17973067e-01 -7.69660354e-01 -1.03064738e-01
1.20650396e-01 7.55991563e-02 7.88593471e-01 1.12329733e+00
-1.31045461e+00 3.95602256e-01 -7.40045011e-01 -4.64178324e-01
4.74677265e-01 1.16490074e-01 -4.90127116e-01 -2.28109181e-01
-4.56638068e-01 7.07452059e-01 -6.75845861e-01 1.13850638e-01
-8.00718904e-01 -1.03600311e+00 -7.39854217e-01 -4.81697917e-01
1.19950034e-01 -7.44526982e-01 1.16677535e+00 -1.35162663e+00
-1.89178920e+00 1.46121418e+00 -5.50614059e-01 2.60448426e-01
4.58681643e-01 -1.20242499e-01 -4.04502690e-01 2.70051152e-01
-3.99517357e-01 4.39383477e-01 1.21002066e+00 -1.54433548e+00
1.93256021e-01 -6.59607291e-01 -3.15316647e-01 1.00870803e-01
1.64346278e-01 4.60481673e-01 -4.14262950e-01 -3.17523926e-02
3.21290642e-01 -1.01155663e+00 1.03436448e-01 4.73264605e-01
-1.16746947e-01 6.10127389e-01 5.99928081e-01 -8.38629961e-01
4.71857607e-01 -2.06349206e+00 -3.07613611e-01 3.15658987e-01
6.56803921e-02 5.22373915e-02 -1.20373823e-01 3.11032712e-01
-1.89008638e-01 -2.69121319e-01 -2.87497282e-01 -7.60567665e-01
2.16422370e-03 -1.60070881e-01 -1.19743183e-01 9.12774205e-01
2.14494273e-01 7.62570500e-01 -1.50343016e-01 -3.21991146e-01
2.41594493e-01 9.33046222e-01 -6.00392461e-01 2.82179832e-01
-3.17861438e-01 6.79778397e-01 -1.61710493e-02 1.05776715e+00
1.37329364e+00 3.63338351e-01 6.80159852e-02 -2.18573138e-01
-3.09921622e-01 7.46186525e-02 -1.06894052e+00 1.50328374e+00
-6.31264448e-01 6.84357047e-01 7.78776407e-01 4.88118157e-02
1.08267415e+00 1.60377808e-02 6.24477565e-01 -6.81354105e-01
1.52660623e-01 4.20036376e-01 -2.61911482e-01 -3.23951393e-01
1.06993578e-01 -4.87717181e-01 7.48087645e-01 3.83792073e-01
9.29523110e-02 -1.00709987e+00 -5.56920946e-01 -1.40069962e-01
5.38382351e-01 7.97914684e-01 -2.27926686e-01 -4.75318968e-01
4.33975160e-01 -2.19285175e-01 1.12797111e-01 8.36525708e-02
1.70837224e-01 1.20088387e+00 3.12651128e-01 -5.03612220e-01
-1.11231244e+00 -1.07701385e+00 -5.98259687e-01 6.89475298e-01
-5.39608479e-01 -1.88960359e-01 -1.24526513e+00 4.74414416e-02
2.27026656e-01 3.98028195e-01 -5.52182317e-01 3.97281349e-01
-3.84577721e-01 -8.27290237e-01 3.75213385e-01 -2.12827280e-01
2.85282016e-01 -5.04665494e-01 -4.28560078e-01 -3.06131899e-01
3.49827558e-01 -1.17457545e+00 -2.43634164e-01 -3.42358470e-01
-5.53912938e-01 -1.26943099e+00 -4.75167304e-01 -2.47834593e-01
4.98099327e-01 2.11953342e-01 1.31370234e+00 1.56297639e-01
-7.67481565e-01 1.00264692e+00 -4.09048274e-02 -9.28798199e-01
-5.38716495e-01 -4.91323978e-01 -3.15694734e-02 5.37302494e-01
2.48155415e-01 -7.66632676e-01 -8.46959889e-01 3.87556285e-01
-7.93172240e-01 8.45205635e-02 -1.47362966e-02 6.14064187e-02
5.24162948e-01 -7.43480861e-01 -2.43717358e-01 -1.07080293e+00
-1.45026997e-01 -1.67374060e-01 -1.14348316e+00 2.96530202e-02
-5.73770106e-01 -1.75718829e-01 2.30367016e-02 9.74508002e-02
-1.51343930e+00 2.86694616e-01 -3.61200422e-01 -5.21882832e-01
-3.53418142e-01 -1.70063525e-01 -3.24441195e-01 -5.89617193e-01
6.85724974e-01 -8.63125920e-02 5.40670991e-01 -7.10213900e-01
2.52047747e-01 6.76565230e-01 2.58468032e-01 -5.81965864e-01
9.44620848e-01 1.13287508e+00 5.64465046e-01 -1.39531255e+00
-6.51759624e-01 -4.12592828e-01 -8.37312341e-01 -5.45423090e-01
5.42676270e-01 -1.16537416e+00 -8.26526999e-01 9.54284430e-01
-9.54547763e-01 -6.33400142e-01 -2.12032050e-01 9.54624116e-02
-5.67744792e-01 2.19950855e-01 -2.10205495e-01 -1.19866192e+00
-2.07847148e-01 -1.07082367e+00 1.64224219e+00 8.45893398e-02
2.44686812e-01 -9.27705646e-01 2.96441525e-01 6.62799478e-01
4.47949260e-01 6.33947849e-01 3.51341903e-01 4.58459407e-01
-6.78561509e-01 4.04278375e-02 -1.09988138e-01 4.02384877e-01
-3.79296988e-02 7.05435514e-01 -1.87849164e+00 -7.97685385e-02
2.29391813e-01 -2.42585570e-01 8.97345662e-01 5.21385550e-01
8.97813380e-01 2.50588544e-02 5.86745329e-02 1.28745496e+00
1.56497669e+00 -4.72523212e-01 9.17875290e-01 4.90766317e-02
9.69520390e-01 1.08533907e+00 3.54168206e-01 4.10486996e-01
4.08760577e-01 9.89534557e-01 6.23293340e-01 -2.96240926e-01
-6.53480053e-01 3.77092883e-02 4.09651339e-01 4.55138564e-01
-3.89074624e-01 2.59865463e-01 -8.49507153e-01 6.60445392e-02
-1.19324875e+00 -7.76407063e-01 -7.15322018e-01 2.56173277e+00
6.89128637e-01 -6.60730183e-01 8.46449286e-03 -3.24981749e-01
9.13874060e-02 1.47881612e-01 -7.26429895e-02 -5.13544619e-01
-3.72913301e-01 7.39449024e-01 6.65444016e-01 1.02089751e+00
-7.40602255e-01 9.43225920e-01 6.68846226e+00 3.02509010e-01
-1.26043320e+00 2.68897831e-01 5.70987046e-01 -3.98872882e-01
-8.93131673e-01 1.14884160e-01 -8.22506189e-01 1.15755178e-01
1.30522633e+00 6.68171942e-01 8.66380334e-01 5.03715694e-01
3.77482176e-01 -4.01352078e-01 -7.95154870e-01 1.04298878e+00
4.32234764e-01 -8.41394544e-01 -2.57658094e-01 3.71681869e-01
6.55324161e-01 2.26229399e-01 7.57873729e-02 -4.26257133e-01
1.53262824e-01 -1.17658889e+00 8.26840937e-01 9.84747291e-01
1.18983424e+00 -3.89411211e-01 -1.68716051e-02 -1.08109035e-01
-7.68635213e-01 5.00086606e-01 -4.33486879e-01 2.53294170e-01
-4.10177419e-03 7.92183757e-01 -7.99795866e-01 3.79228473e-01
7.09833324e-01 4.93050009e-01 -6.76799297e-01 6.86424911e-01
-1.10215001e-01 6.17913246e-01 -6.84232414e-01 5.12488961e-01
-4.00177747e-01 -6.86774194e-01 4.51295197e-01 8.82427275e-01
3.08943838e-01 4.13126610e-02 -4.00332361e-01 1.03354824e+00
-7.42624281e-03 1.35446787e-01 -4.98692244e-01 2.47457013e-01
4.13850807e-02 1.55857956e+00 -1.54608116e-01 2.08341807e-01
-4.45290565e-01 8.89297128e-01 1.62850529e-01 3.65759075e-01
-6.38215005e-01 2.92459935e-01 1.04932427e+00 7.20510304e-01
9.36381612e-03 -1.38042048e-01 -4.00511660e-02 -1.19254100e+00
9.71886441e-02 -8.21161449e-01 -1.93146810e-01 -1.23538220e+00
-1.04859138e+00 2.75536865e-01 4.08666991e-02 -6.07649863e-01
-2.96314992e-02 -1.00692046e+00 -4.03812677e-01 1.30363190e+00
-1.95713794e+00 -1.77148402e+00 -7.05883741e-01 4.66206402e-01
-4.92379419e-04 1.16283342e-01 9.30016339e-01 1.63391277e-01
-3.64872634e-01 3.37021589e-01 2.26888642e-01 -3.46778363e-01
8.94375563e-01 -1.13424790e+00 3.72945249e-01 4.97069150e-01
4.21470031e-02 6.41650259e-01 7.35038459e-01 -2.28967816e-01
-1.91187775e+00 -7.04419732e-01 4.24454451e-01 -1.07774210e+00
2.62536675e-01 -6.63022101e-01 -6.35369718e-01 6.93095624e-01
9.04246196e-02 3.60310256e-01 7.81217813e-01 -1.52808338e-01
-5.51625848e-01 -6.03282869e-01 -1.24979174e+00 3.54728580e-01
9.38998938e-01 -4.94389266e-01 6.86961710e-02 3.64870399e-01
2.64805764e-01 -5.31333029e-01 -9.51756954e-01 1.98724836e-01
1.17120028e+00 -1.57505763e+00 9.71767724e-01 -1.12555221e-01
1.66919351e-01 -2.50903279e-01 -3.09015542e-01 -1.17762613e+00
2.62514502e-01 -9.38886166e-01 2.61864632e-01 1.46346295e+00
2.29274720e-01 -9.10324991e-01 6.72775626e-01 9.23339963e-01
-7.67193958e-02 -3.59651685e-01 -8.44054461e-01 -3.33308309e-01
4.87509891e-02 -5.33136427e-01 9.19375002e-01 7.59994864e-01
-8.42805028e-01 -3.04759085e-01 -2.80876070e-01 2.46580735e-01
1.00047255e+00 1.02745906e-01 1.29540992e+00 -1.55663455e+00
-3.01851422e-01 -6.72041103e-02 -3.37476679e-03 -4.68433142e-01
2.57104427e-01 -7.04267740e-01 -1.28075093e-01 -1.06121075e+00
2.26353839e-01 -5.43973267e-01 4.72564191e-01 2.68099129e-01
2.97383696e-01 6.57357574e-01 -4.05314155e-02 3.46386135e-01
1.50594085e-01 3.10985327e-01 1.06056511e+00 2.55515546e-01
-1.20325923e-01 -2.69463152e-01 -3.60994577e-01 8.34296346e-01
4.13991332e-01 -2.76732832e-01 -1.41617298e-01 -5.35304844e-01
4.09610868e-01 -2.88455993e-01 6.50787592e-01 -8.76252115e-01
-3.54420841e-01 -2.59998709e-01 4.91840392e-01 -1.66871145e-01
7.79021859e-01 -1.04849875e+00 7.74047852e-01 -6.17570281e-02
2.30029806e-01 -5.88987827e-01 2.59433538e-01 2.59733517e-02
2.73880839e-01 -1.25444040e-01 1.07520056e+00 -3.08970541e-01
-1.48288995e-01 5.13307571e-01 6.64418936e-02 -2.66577929e-01
4.98610884e-01 -2.74081022e-01 -3.48362833e-01 -3.79214823e-01
-3.67866129e-01 -4.61281389e-01 1.30942106e+00 -2.13567823e-01
3.64697516e-01 -1.11098921e+00 -9.54977214e-01 5.40894210e-01
1.46209344e-01 -6.64045140e-02 2.75483012e-01 9.36573327e-01
-1.13210034e+00 3.70213427e-02 1.82956252e-02 -6.13112926e-01
-1.64390421e+00 -1.21548802e-01 8.78991604e-01 4.98843491e-01
-2.24094734e-01 7.74838030e-01 2.33829439e-01 -7.89399564e-01
-4.68027651e-01 1.10863172e-01 2.39899144e-01 3.19923498e-02
4.24527347e-01 3.72222453e-01 3.61608863e-01 -1.12956023e+00
-4.01225060e-01 1.00397420e+00 6.74274743e-01 -2.63327360e-01
1.65643072e+00 -2.33146444e-01 -7.44986355e-01 4.87904131e-01
1.02296793e+00 8.33114505e-01 -1.43083048e+00 6.10030182e-02
-7.31888711e-01 -7.66369760e-01 2.99358010e-01 -6.52346849e-01
-1.10979331e+00 8.72393966e-01 5.97429752e-01 -2.81284988e-01
1.16817224e+00 8.91087949e-02 2.90416598e-01 -1.26141906e-01
2.88007468e-01 -7.49619544e-01 -3.83470386e-01 1.20111033e-01
9.90603805e-01 -1.08914208e+00 4.75834191e-01 -8.94528568e-01
-3.50463748e-01 1.19018161e+00 3.62310857e-01 6.49770424e-02
8.31351221e-01 3.78983885e-01 4.06345725e-01 -5.55760801e-01
-4.67799783e-01 -3.19510311e-01 5.70172608e-01 8.59477699e-01
6.76607668e-01 -6.44075125e-02 2.84972578e-01 -9.67468470e-02
-4.84615177e-01 -2.10973471e-01 5.51957071e-01 5.66568732e-01
-2.49604195e-01 -1.16564190e+00 -7.60738313e-01 1.46016181e-01
-3.09575528e-01 -1.73255652e-01 -5.50127566e-01 5.97907722e-01
9.91655067e-02 7.11974859e-01 1.34876803e-01 5.32689206e-02
2.79517591e-01 4.97272491e-01 1.00571036e+00 -4.82004225e-01
-2.77572453e-01 3.62489015e-01 2.41079926e-01 -6.48890197e-01
-6.32919073e-01 -9.92164433e-01 -5.40033817e-01 -4.10637826e-01
-4.59566861e-01 -5.27983963e-01 1.21063542e+00 5.26933610e-01
2.58352786e-01 -1.55518195e-02 8.22146714e-01 -1.06084538e+00
6.23012893e-02 -8.53001177e-01 -9.02664661e-01 3.31003785e-01
4.33883011e-01 -5.15359402e-01 -4.80210900e-01 1.55278161e-01] | [12.75777530670166, -0.34688320755958557] |
e00368e7-fbfd-484b-943b-6f1cb554465f | an-end-to-end-neural-network-for-polyphonic | 1508.01774 | null | http://arxiv.org/abs/1508.01774v2 | http://arxiv.org/pdf/1508.01774v2.pdf | An End-to-End Neural Network for Polyphonic Piano Music Transcription | We present a supervised neural network model for polyphonic piano music
transcription. The architecture of the proposed model is analogous to speech
recognition systems and comprises an acoustic model and a music language model.
The acoustic model is a neural network used for estimating the probabilities of
pitches in a frame of audio. The language model is a recurrent neural network
that models the correlations between pitch combinations over time. The proposed
model is general and can be used to transcribe polyphonic music without
imposing any constraints on the polyphony. The acoustic and language model
predictions are combined using a probabilistic graphical model. Inference over
the output variables is performed using the beam search algorithm. We perform
two sets of experiments. We investigate various neural network architectures
for the acoustic models and also investigate the effect of combining acoustic
and music language model predictions using the proposed architecture. We
compare performance of the neural network based acoustic models with two
popular unsupervised acoustic models. Results show that convolutional neural
network acoustic models yields the best performance across all evaluation
metrics. We also observe improved performance with the application of the music
language models. Finally, we present an efficient variant of beam search that
improves performance and reduces run-times by an order of magnitude, making the
model suitable for real-time applications. | ['Simon Dixon', 'Emmanouil Benetos', 'Siddharth Sigtia'] | 2015-08-07 | null | null | null | null | ['music-transcription'] | ['music'] | [ 2.41943806e-01 -1.68524116e-01 6.82240278e-02 -2.76471496e-01
-7.41245031e-01 -3.73568535e-01 3.99512738e-01 -2.47890621e-01
-3.47951829e-01 1.81392461e-01 3.30631524e-01 -2.45154887e-01
-3.08060735e-01 -6.44262195e-01 -5.75717390e-01 -7.91028559e-01
-2.35838354e-01 4.34150726e-01 1.65321872e-01 -2.34719533e-02
2.04106569e-01 2.89429635e-01 -1.75199270e+00 4.65234429e-01
-4.38335799e-02 1.07968986e+00 3.92772645e-01 1.38009655e+00
1.67240221e-02 7.60973513e-01 -8.29753518e-01 3.09493132e-02
1.33874789e-01 -6.18500173e-01 -4.87779498e-01 -3.43009025e-01
1.82718202e-01 3.96093503e-02 -3.92686695e-01 8.77712786e-01
8.53116274e-01 4.57055539e-01 5.22201777e-01 -9.27822530e-01
-9.64509547e-02 1.34769213e+00 2.96481028e-02 2.54612714e-01
2.30290890e-01 -9.38408375e-02 1.30348206e+00 -8.47430885e-01
1.77027415e-02 1.38788450e+00 8.50384176e-01 1.49078920e-01
-9.82283473e-01 -8.77910972e-01 -1.72590241e-01 5.05606174e-01
-1.54326499e+00 -6.15756691e-01 9.45136011e-01 -2.87976682e-01
1.31998885e+00 3.43493670e-01 5.11685014e-01 8.33979726e-01
1.47142276e-01 6.45735204e-01 7.05215752e-01 -8.85109007e-01
2.36564979e-01 -1.18408024e-01 5.01752533e-02 4.97559637e-01
-5.37491441e-01 3.32321912e-01 -1.13784897e+00 -3.63865554e-01
6.57871485e-01 -4.06399965e-01 2.08931658e-02 2.00309142e-01
-1.00952899e+00 7.36725032e-01 -3.07272561e-02 2.99002588e-01
-4.75868583e-01 6.00509107e-01 5.21082103e-01 1.62760645e-01
1.26177087e-01 3.85658324e-01 -4.03625518e-01 -4.35299575e-01
-1.45486975e+00 1.78412095e-01 1.13269687e+00 5.71478724e-01
1.66632682e-01 6.11006975e-01 -1.77121490e-01 1.27390707e+00
5.76957405e-01 3.99993449e-01 7.69486070e-01 -9.71506834e-01
1.48416981e-01 -1.99544117e-01 -2.17834294e-01 -7.76966929e-01
-5.08774579e-01 -6.63638890e-01 -8.12607169e-01 1.58523306e-01
2.28233039e-01 -2.02234715e-01 -8.19038033e-01 1.68266666e+00
-2.58968860e-01 6.87877715e-01 4.60365079e-02 6.92490101e-01
8.21508408e-01 1.00132608e+00 -1.53131053e-01 -3.07576478e-01
1.27692962e+00 -9.90474522e-01 -1.07846200e+00 1.07763916e-01
2.59281278e-01 -1.27572691e+00 9.83283460e-01 9.13365066e-01
-1.34659421e+00 -8.97916377e-01 -1.18408227e+00 2.84152269e-01
-2.66221255e-01 5.96399188e-01 2.78978407e-01 8.47155392e-01
-1.07550120e+00 8.75085354e-01 -8.19251776e-01 -7.07335584e-03
-2.96301812e-01 7.74583519e-01 3.89192820e-01 6.84392810e-01
-1.38421571e+00 4.90865588e-01 6.38899267e-01 3.60264897e-01
-9.32054043e-01 -4.63923365e-01 -6.08655512e-01 4.85882223e-01
-1.76967829e-01 -4.04224664e-01 1.80051005e+00 -8.75123024e-01
-2.18134642e+00 2.32444122e-01 -2.43784368e-01 -8.31368208e-01
-1.45779818e-01 -3.93463552e-01 -6.57289565e-01 -3.77777480e-02
-6.02964103e-01 7.03769028e-01 8.76932621e-01 -7.72740364e-01
-5.71638048e-01 3.72793019e-01 -3.51768672e-01 2.20422760e-01
-2.03214228e-01 1.21918060e-01 -4.34445024e-01 -9.82332110e-01
8.75497237e-02 -1.28576803e+00 -9.40994471e-02 -7.72163033e-01
-4.75792408e-01 -1.72785059e-01 7.05954432e-01 -5.74885845e-01
1.64972842e+00 -2.27398801e+00 2.55442932e-02 4.00841981e-01
-5.79639792e-01 1.65834844e-01 -1.25174180e-01 5.49414039e-01
-3.58438015e-01 -1.25424847e-01 -6.54106587e-02 -5.73618650e-01
2.57484913e-01 2.45005742e-01 -7.73102939e-01 -7.03304186e-02
-1.10182770e-01 5.58246076e-01 -3.22814614e-01 -8.88752118e-02
2.93203235e-01 6.35171771e-01 -6.56328380e-01 3.03351432e-01
-3.49420756e-01 8.66718367e-02 2.53511876e-01 2.26291165e-01
2.03914106e-01 7.24352822e-02 2.79248536e-01 -2.28142425e-01
5.53229544e-03 1.01646483e+00 -1.53117669e+00 1.71140778e+00
-6.47235334e-01 7.83053994e-01 -2.15542898e-01 -6.37650728e-01
1.16419959e+00 8.77081931e-01 1.78081781e-01 -1.20791093e-01
8.28902870e-02 9.92810056e-02 4.72579569e-01 -3.02058607e-01
5.77584326e-01 -3.59436363e-01 1.47101516e-02 5.92052102e-01
3.67853850e-01 -3.04182649e-01 2.86168922e-02 -2.58211464e-01
8.06887925e-01 3.74367498e-02 1.26857311e-01 -1.78856850e-02
5.04044354e-01 -4.94113773e-01 4.10882682e-01 8.92595112e-01
2.36308500e-01 5.41118026e-01 1.41031191e-01 -3.41620922e-01
-8.13058853e-01 -9.63800788e-01 2.15172723e-01 1.57463419e+00
-4.34273154e-01 -7.40086377e-01 -6.00148559e-01 -1.15774255e-02
-4.59160715e-01 8.83197248e-01 -2.31521204e-01 -4.27010581e-02
-5.33629417e-01 -8.24076712e-01 9.67274189e-01 5.62617838e-01
2.02153489e-01 -1.58556688e+00 -3.12901616e-01 3.07590038e-01
-1.21180713e-01 -8.60787153e-01 -3.11279535e-01 5.39465129e-01
-7.92925417e-01 -4.97970551e-01 -3.99606973e-01 -9.13231790e-01
-3.11311930e-01 -3.92601997e-01 1.09954703e+00 -3.66295010e-01
6.29242137e-02 3.77059698e-01 -3.59317631e-01 -8.37632596e-01
-7.01298952e-01 2.81203985e-01 4.68456775e-01 1.30704835e-01
3.60642642e-01 -9.70215023e-01 -4.48616445e-01 1.90475196e-01
-7.62871206e-01 -1.08099699e-01 4.76876885e-01 6.38116002e-01
6.28999472e-01 3.68320495e-01 5.87984025e-01 -5.61667740e-01
8.63714099e-01 -1.60690770e-01 -5.01466691e-01 -1.42232090e-01
-2.87610292e-01 4.21045162e-02 4.06915516e-01 -6.68352008e-01
-8.70065928e-01 4.18913454e-01 -5.59406996e-01 -6.35004461e-01
-1.99485481e-01 6.56066060e-01 1.81885645e-01 3.07472050e-01
5.02146602e-01 1.24018282e-01 -4.35153514e-01 -7.74749279e-01
2.48034731e-01 8.49858820e-01 6.80871964e-01 -2.58274913e-01
3.06157470e-01 6.44925386e-02 -1.06604695e-01 -1.05439281e+00
-6.69493556e-01 -5.64540267e-01 -5.38784921e-01 -2.79730678e-01
7.95327663e-01 -7.76057959e-01 -8.02184224e-01 3.68269622e-01
-1.24401224e+00 -4.07022983e-01 -3.31741422e-01 1.10086644e+00
-9.41009521e-01 8.67721289e-02 -8.74909282e-01 -1.28208566e+00
-5.06872296e-01 -1.07314456e+00 8.63610446e-01 8.29750523e-02
-6.80816233e-01 -8.71040285e-01 5.07212698e-01 -9.82363448e-02
2.90739328e-01 -4.94739383e-01 9.29466546e-01 -8.56528223e-01
-2.91974217e-01 -8.14402476e-02 5.33237934e-01 4.47922438e-01
-9.56160575e-03 1.17700286e-01 -1.55919659e+00 -4.89358790e-02
5.75666241e-02 -1.61804229e-01 1.03744507e+00 7.29755640e-01
1.22386634e+00 -1.56816289e-01 1.53773993e-01 4.43379104e-01
9.52992678e-01 5.74872911e-01 5.52130878e-01 5.02225719e-02
4.39258784e-01 2.89848655e-01 2.59880483e-01 4.56647098e-01
-1.15690611e-01 8.79318655e-01 2.58891940e-01 1.94082081e-01
-2.61521727e-01 -3.75919849e-01 7.04326808e-01 1.78275061e+00
-4.66092192e-02 -2.90095329e-01 -9.55986321e-01 4.30571705e-01
-1.85338426e+00 -9.33755338e-01 5.33531271e-02 2.22256303e+00
7.46131718e-01 2.70378560e-01 2.59860039e-01 6.96631849e-01
6.27990484e-01 7.17073306e-02 -1.43340379e-01 -9.94752109e-01
4.69750352e-02 8.18584085e-01 2.58712560e-01 6.86466932e-01
-1.34316611e+00 8.78880322e-01 7.48599911e+00 1.02221715e+00
-1.14634979e+00 4.30552065e-02 2.20274717e-01 -3.87640029e-01
1.64367437e-01 -2.56022185e-01 -8.21252644e-01 1.71925277e-01
1.63696134e+00 3.16809595e-01 5.83468854e-01 6.49496138e-01
5.03903687e-01 2.16200173e-01 -1.17482686e+00 1.23741972e+00
2.89969984e-02 -1.32313192e+00 1.89802125e-01 -1.92075893e-01
5.41538358e-01 2.48431489e-01 2.77022660e-01 3.79235983e-01
2.97391772e-01 -1.16988206e+00 8.10468912e-01 7.50731051e-01
5.13912618e-01 -1.00407290e+00 7.39642203e-01 2.82057852e-01
-1.28343141e+00 -1.45614922e-01 -1.70724213e-01 -3.65370840e-01
2.30610996e-01 3.83522600e-01 -1.11466181e+00 1.69791877e-01
7.50224710e-01 4.12770122e-01 -1.21052615e-01 1.36630011e+00
-2.29440942e-01 1.32529104e+00 -4.86449331e-01 -1.12857349e-01
2.70481765e-01 -5.95819093e-02 7.38424599e-01 1.72638929e+00
5.73962808e-01 -2.73039132e-01 3.55895534e-02 6.35601103e-01
1.39269173e-01 1.84556246e-01 -3.29948843e-01 -2.01065838e-01
5.39083600e-01 9.98701394e-01 -7.32714832e-01 -3.50975186e-01
-1.22840554e-02 6.36721492e-01 -1.61600500e-01 4.49069500e-01
-6.17136121e-01 -4.17795718e-01 3.53160620e-01 -3.20917368e-01
5.86335063e-01 -4.25583005e-01 -2.69535452e-01 -4.69623774e-01
-4.35930550e-01 -8.87402117e-01 4.13428903e-01 -9.81317997e-01
-9.58952248e-01 7.56524444e-01 -1.88353091e-01 -1.20834637e+00
-8.89688849e-01 -5.56799471e-01 -7.03974307e-01 1.00501585e+00
-1.13816690e+00 -6.12268925e-01 2.97059447e-01 2.77079016e-01
8.54552746e-01 -5.50039709e-01 1.40749013e+00 1.49714544e-01
-3.19969863e-01 5.08646965e-01 -9.44704711e-02 5.19147813e-02
6.16723776e-01 -1.24093091e+00 4.02618587e-01 5.31045675e-01
9.55098808e-01 5.08781791e-01 7.99521506e-01 -3.52414578e-01
-9.65174437e-01 -9.69042003e-01 1.01748967e+00 -1.28729075e-01
6.06916368e-01 -4.03027534e-01 -8.33668113e-01 4.78011161e-01
4.57029611e-01 -5.73093116e-01 1.12143588e+00 3.23727429e-01
-3.01891893e-01 -1.45181581e-01 -3.62369657e-01 4.93214697e-01
3.48768800e-01 -9.06750917e-01 -7.56243408e-01 2.51005381e-01
7.79901981e-01 -3.45287502e-01 -7.57944345e-01 2.91313738e-01
8.40242326e-01 -9.13410127e-01 8.14374268e-01 -2.30456397e-01
1.31393090e-01 -5.01754105e-01 -4.37848687e-01 -1.12603319e+00
-3.40409666e-01 -1.05086493e+00 -3.75312120e-01 9.74412203e-01
7.72962391e-01 -4.26620729e-02 6.67582989e-01 -8.56442526e-02
-1.01594210e-01 -4.41173077e-01 -1.02417457e+00 -6.64433599e-01
-2.57836223e-01 -1.26695526e+00 3.84518445e-01 5.79620063e-01
2.04123762e-02 5.29302597e-01 -6.27747774e-01 4.66640860e-01
1.27790347e-01 -2.74627320e-02 4.76431131e-01 -9.44266558e-01
-9.72208977e-01 -6.61432385e-01 -3.77738595e-01 -1.08171904e+00
3.31730507e-02 -8.80687118e-01 2.84681350e-01 -1.04948926e+00
-2.08971530e-01 -1.65155053e-01 -9.20631170e-01 4.44178671e-01
3.86687547e-01 5.27386367e-01 3.90488952e-01 1.09014235e-01
-2.65828162e-01 3.64573389e-01 4.88671303e-01 -1.41925830e-02
-6.62843585e-01 5.09705663e-01 4.83308807e-02 1.10036039e+00
1.07024980e+00 -5.74520528e-01 -4.74870056e-01 -1.16773374e-01
4.81909364e-01 5.49024828e-02 1.40266076e-01 -1.32198632e+00
5.17006516e-01 4.52329844e-01 2.10523620e-01 -8.62015188e-01
8.66255105e-01 -5.53786874e-01 2.70881772e-01 3.32138300e-01
-7.06775784e-01 1.48838773e-01 4.72570062e-01 4.70109135e-01
-4.62742001e-01 -3.89802903e-01 6.19804263e-01 1.97368726e-01
-2.96566755e-01 -1.76045403e-01 -7.36230493e-01 -5.76410294e-01
9.12449509e-02 4.91674915e-02 5.33901036e-01 -8.35142672e-01
-1.27198744e+00 -5.07247210e-01 -4.86024052e-01 6.30725205e-01
5.27160585e-01 -1.37801754e+00 -5.04056513e-01 4.13858593e-01
-4.32646811e-01 -4.75532532e-01 -1.93729550e-02 6.40489340e-01
-3.89942914e-01 6.38595760e-01 1.66764826e-01 -7.34386206e-01
-1.59743929e+00 2.26153687e-01 4.90910649e-01 -2.29334101e-01
-3.21378022e-01 9.61296260e-01 4.48883958e-02 -4.16096449e-01
8.44411790e-01 -6.93675637e-01 -4.57047790e-01 -6.00923896e-02
5.68331540e-01 3.66413116e-01 1.78621620e-01 -5.67870200e-01
-2.01867491e-01 4.91351664e-01 2.03105018e-01 -7.03749955e-01
1.21970999e+00 6.69929981e-02 3.43444049e-02 1.33442986e+00
7.27352798e-01 1.31687969e-01 -6.49473369e-01 -1.80355459e-01
1.46527737e-01 3.01808774e-01 3.91683668e-01 -9.41962302e-01
-7.15948880e-01 1.07247019e+00 8.20448101e-01 3.33504319e-01
1.31623721e+00 -3.35339308e-01 5.99484682e-01 6.63837254e-01
-1.17748402e-01 -1.15210986e+00 -1.79840609e-01 9.39747095e-01
8.21183324e-01 -5.18237531e-01 -3.16579282e-01 -1.13524735e-01
-4.22043204e-01 1.32926142e+00 -1.29915223e-01 -3.13279301e-01
1.03840542e+00 5.57175875e-01 3.34300905e-01 3.34845819e-02
-9.87297595e-01 -5.90109937e-02 6.75580561e-01 2.60050774e-01
6.58660412e-01 2.69754976e-01 9.12685040e-03 9.65768218e-01
-1.01318812e+00 -2.17914041e-02 3.15111339e-01 4.66696262e-01
-3.72202218e-01 -1.16218734e+00 -6.00336313e-01 5.15958481e-02
-6.13031030e-01 -4.04536992e-01 -4.65477198e-01 3.29776853e-01
1.21966310e-01 1.14096165e+00 3.15063596e-01 -5.99933207e-01
2.42404252e-01 6.46218121e-01 2.63057202e-01 -6.64553285e-01
-8.85035396e-01 7.28344798e-01 1.98016912e-01 -3.27209383e-01
-4.76224035e-01 -4.62380797e-01 -1.16840124e+00 2.59559631e-01
-4.04023141e-01 3.41868907e-01 1.05010140e+00 8.84842396e-01
2.34335840e-01 9.15479720e-01 4.85804468e-01 -1.06917429e+00
-3.40513766e-01 -1.37574637e+00 -7.23979354e-01 -1.67066336e-01
2.05888689e-01 -3.34640801e-01 -2.11810425e-01 3.51234078e-01] | [15.619244575500488, 5.5186543464660645] |
5d55226d-cd6c-4ac6-ae36-7245a79e1ec2 | on-regularization-and-inference-with-label | 2307.03886 | null | https://arxiv.org/abs/2307.03886v1 | https://arxiv.org/pdf/2307.03886v1.pdf | On Regularization and Inference with Label Constraints | Prior knowledge and symbolic rules in machine learning are often expressed in the form of label constraints, especially in structured prediction problems. In this work, we compare two common strategies for encoding label constraints in a machine learning pipeline, regularization with constraints and constrained inference, by quantifying their impact on model performance. For regularization, we show that it narrows the generalization gap by precluding models that are inconsistent with the constraints. However, its preference for small violations introduces a bias toward a suboptimal model. For constrained inference, we show that it reduces the population risk by correcting a model's violation, and hence turns the violation into an advantage. Given these differences, we further explore the use of two approaches together and propose conditions for constrained inference to compensate for the bias introduced by regularization, aiming to improve both the model complexity and optimal risk. | ['Dan Roth', 'Piyush Kumar', 'Tin D. Nguyen', 'Hangfeng He', 'Kaifu Wang'] | 2023-07-08 | null | null | null | null | ['structured-prediction'] | ['methodology'] | [ 5.62418580e-01 8.92215967e-01 -4.73149508e-01 -6.69491231e-01
-4.91264939e-01 -5.54175496e-01 5.19898653e-01 4.41436410e-01
-3.87399107e-01 7.16914654e-01 2.98209310e-01 -4.62276280e-01
-3.72346133e-01 -6.16417825e-01 -8.77155900e-01 -3.79751861e-01
3.55369717e-01 4.37139034e-01 3.04268152e-01 2.90919214e-01
2.94668347e-01 3.61625224e-01 -1.49678242e+00 4.05019462e-01
1.00429690e+00 8.12211812e-01 -1.62047774e-01 -5.83566017e-02
-9.30262879e-02 7.98753858e-01 -3.75436276e-01 -6.30334973e-01
2.48671934e-01 -4.15912688e-01 -7.46236861e-01 -2.22693890e-01
4.95036751e-01 -4.21174057e-02 1.99848309e-01 1.06114614e+00
1.46069765e-01 -4.64450289e-03 8.63653779e-01 -1.11247718e+00
-1.57473847e-01 9.60335076e-01 -9.15914699e-02 -2.14861393e-01
1.41671166e-01 1.23499833e-01 1.18513322e+00 -5.54737985e-01
7.47580707e-01 1.30286348e+00 9.23063517e-01 6.46929741e-01
-1.68810451e+00 -3.77610296e-01 4.78488773e-01 -1.69518948e-01
-1.25328040e+00 -5.14245808e-01 5.94201505e-01 -7.33178854e-01
1.05266833e+00 4.46772099e-01 4.32897538e-01 1.01250219e+00
-1.31058291e-01 3.48654598e-01 9.09951627e-01 -6.52626395e-01
4.85263407e-01 4.54762280e-01 4.15504873e-01 7.63955116e-01
5.16204953e-01 2.54972279e-01 -5.78691661e-01 -3.40023786e-01
3.06260943e-01 -4.05692697e-01 -8.03044438e-02 -4.37637269e-01
-7.07368731e-01 8.47683847e-01 7.93117061e-02 1.76864758e-01
2.11846437e-02 6.19957000e-02 2.91446596e-01 -2.09276509e-02
5.02185583e-01 7.69990861e-01 -7.66255200e-01 1.31105214e-01
-9.43638444e-01 3.48770589e-01 8.40226710e-01 8.66259634e-01
9.14347470e-01 -3.28213602e-01 -4.32268351e-01 8.36198926e-01
5.68744361e-01 -5.78528419e-02 8.30714777e-02 -1.11349261e+00
3.42796326e-01 1.00058091e+00 1.50048584e-01 -7.53053665e-01
-5.61442852e-01 -3.70135427e-01 -2.30346337e-01 2.34394535e-01
8.18999529e-01 -7.49460757e-02 -6.74638510e-01 2.05050945e+00
1.21764377e-01 9.55978557e-02 -9.71423015e-02 5.29800415e-01
4.86115336e-01 1.27019078e-01 3.81657004e-01 -4.97493804e-01
1.13566196e+00 -7.55293310e-01 -5.61247885e-01 -5.89761555e-01
1.14002287e+00 -2.98014015e-01 1.06767130e+00 4.14050281e-01
-8.71755302e-01 -1.09355390e-01 -8.80583227e-01 -9.42051783e-02
-1.12632841e-01 4.09898125e-02 7.24815965e-01 7.77034163e-01
-5.31956971e-01 1.04181659e+00 -7.94799507e-01 -1.93736091e-01
2.75653392e-01 3.22361052e-01 -1.18109070e-01 1.46383420e-01
-1.06920910e+00 1.24704099e+00 5.31537294e-01 -4.53355862e-03
-4.70362186e-01 -9.13542211e-01 -8.53425741e-01 3.18102837e-01
6.46314859e-01 -4.20100659e-01 9.26513970e-01 -9.38131690e-01
-1.40807045e+00 9.65550959e-01 -1.55054525e-01 -4.39728290e-01
7.94852734e-01 -2.13510394e-01 1.80834364e-02 -5.54731548e-01
-2.12202698e-01 5.56930780e-01 5.40216029e-01 -1.09182155e+00
-4.35097307e-01 -2.47385651e-01 2.02861860e-01 -1.42815828e-01
-1.95364028e-01 2.11204246e-01 -4.60292220e-01 -3.02475244e-01
1.89360291e-01 -1.07420444e+00 -4.88977462e-01 -1.94707990e-01
-5.48343241e-01 -2.34634116e-01 2.70081431e-01 -4.09571171e-01
1.46978164e+00 -2.17422152e+00 2.60772496e-01 4.82628435e-01
-8.07678252e-02 2.11756095e-01 1.63326338e-01 5.44756688e-02
-4.35397774e-02 5.01879990e-01 -6.33725286e-01 -4.91854191e-01
1.79536879e-01 6.00712657e-01 -2.21849814e-01 2.80162245e-01
4.63052958e-01 6.48295760e-01 -6.57042265e-01 -5.28129935e-01
-4.18310352e-02 3.06122065e-01 -1.11600578e+00 4.87088412e-02
-6.91677809e-01 5.87904036e-01 -2.54786998e-01 3.20936412e-01
6.09424710e-01 -9.76034552e-02 7.89650083e-01 -1.27617165e-01
-2.18136683e-01 6.85319245e-01 -1.26743031e+00 1.30368352e+00
-2.28630483e-01 2.36601636e-01 8.60093569e-04 -9.58338916e-01
8.84854078e-01 1.48568675e-03 3.82529385e-02 -1.24807842e-01
5.59577383e-02 2.88367242e-01 1.35459706e-01 -6.11661971e-01
1.59657001e-01 -2.55347550e-01 2.57915165e-02 9.45643932e-02
-8.34195688e-02 -4.76335846e-02 2.36242369e-01 -1.40462905e-01
8.58220875e-01 6.21665061e-01 3.44305158e-01 -5.27653337e-01
4.00299251e-01 -1.64866105e-01 1.06273746e+00 8.10778320e-01
1.06696278e-01 2.96498030e-01 1.13042855e+00 -2.99459308e-01
-6.03548467e-01 -6.61777139e-01 -4.23539221e-01 1.14533854e+00
-2.16943011e-01 -7.70585299e-01 -6.54580534e-01 -9.73473608e-01
3.04766655e-01 1.10744762e+00 -6.36713445e-01 -3.36210161e-01
-6.36523783e-01 -8.16190541e-01 6.99164152e-01 3.62678498e-01
-2.07945034e-01 -7.84378886e-01 -7.79051065e-01 -1.12905845e-01
-1.14440089e-02 -9.14537549e-01 -8.83837044e-02 6.38852835e-01
-9.70256329e-01 -1.08304346e+00 -8.75100493e-02 -2.98037559e-01
9.20706153e-01 -7.81792879e-01 1.20084882e+00 5.37384093e-01
1.12043656e-01 -1.96981505e-01 -3.21209699e-01 -4.00098264e-01
-6.80692673e-01 3.65501158e-02 1.02262177e-01 -1.81922883e-01
4.32710767e-01 -3.36280972e-01 -5.31361476e-02 2.42119163e-01
-7.38210857e-01 1.16963871e-01 3.75616878e-01 8.76803637e-01
4.73941535e-01 -3.22172225e-01 2.29593694e-01 -1.46237981e+00
3.07575136e-01 -3.38221818e-01 -9.15946484e-01 4.27977502e-01
-1.14641750e+00 5.77235281e-01 3.78983349e-01 -3.48711073e-01
-1.12185228e+00 3.20280284e-01 -4.60309871e-02 -1.87250182e-01
-2.42206156e-01 8.14848959e-01 -2.85774678e-01 5.48124760e-02
8.35949719e-01 -4.37476546e-01 -1.46065399e-01 -7.64103651e-01
2.19932437e-01 3.34006906e-01 -2.48336475e-02 -9.93068278e-01
3.43241960e-01 3.43543664e-02 2.07503438e-01 -2.12450877e-01
-1.10337400e+00 1.06275372e-01 -6.95304990e-01 -9.34430361e-02
7.00494766e-01 -4.86361951e-01 -5.84997714e-01 -2.28832867e-02
-1.02680945e+00 -4.70230877e-01 -3.34350079e-01 4.26223755e-01
-4.50455546e-01 3.81983519e-01 -3.34789276e-01 -9.01528418e-01
3.06443244e-01 -1.39083171e+00 6.72232568e-01 -5.52519318e-03
-6.89267933e-01 -1.02309132e+00 -3.71238887e-02 1.66607257e-02
1.15442276e-01 2.06938311e-01 1.37352395e+00 -1.01423168e+00
-3.03735226e-01 7.62805194e-02 -1.46199152e-01 3.23746830e-01
-1.32333159e-01 1.69822589e-01 -1.02927291e+00 7.72927403e-02
-1.40998662e-01 -1.87056065e-01 9.60185289e-01 2.36105263e-01
1.07941866e+00 -3.95826489e-01 -4.10257459e-01 6.57617331e-01
1.38083863e+00 -7.62156099e-02 5.52345395e-01 2.73922712e-01
5.24922371e-01 1.06993544e+00 5.22082090e-01 3.46552938e-01
5.15449084e-02 9.33406234e-01 3.26059788e-01 2.63734102e-01
8.71402472e-02 -3.58477026e-01 2.22804934e-01 2.11468071e-01
-7.37673556e-03 1.08900387e-02 -1.13741696e+00 2.09243178e-01
-2.06222057e+00 -5.09564161e-01 -2.51132637e-01 2.24051189e+00
1.14258051e+00 4.99952614e-01 -6.32985607e-02 9.11296755e-02
4.59409356e-01 -2.04554543e-01 -4.28095073e-01 -6.33928239e-01
-4.96761128e-02 -6.65228367e-02 4.95996594e-01 9.29987073e-01
-7.66214371e-01 9.73391891e-01 7.47484016e+00 7.08058894e-01
-7.56154299e-01 -1.07636899e-01 4.96963501e-01 -1.42700449e-01
-5.25127530e-01 4.38023388e-01 -1.03557169e+00 5.54257333e-01
8.94926667e-01 2.09701583e-01 3.74513060e-01 8.83846760e-01
5.60971461e-02 -3.28727365e-01 -1.89979851e+00 3.00504476e-01
-2.27490708e-01 -1.29958868e+00 1.03267759e-01 -1.38379699e-02
3.70886087e-01 -3.08596551e-01 -1.72930345e-01 3.41161817e-01
2.05715910e-01 -1.10546529e+00 9.54568326e-01 5.84601998e-01
5.44657469e-01 -5.09339392e-01 6.71361148e-01 5.94723880e-01
-4.61427629e-01 -1.72902212e-01 -1.73080355e-01 -5.13296604e-01
3.16704884e-02 8.29117477e-01 -5.76120198e-01 2.70633847e-01
4.07336146e-01 5.00594795e-01 -6.97874546e-01 6.21005774e-01
-4.00792599e-01 8.18253756e-01 -4.35192257e-01 2.54826277e-01
-1.86578527e-01 -4.42178577e-01 4.70457613e-01 1.37927198e+00
-2.15084665e-02 -1.65508986e-02 1.03565723e-01 1.21494615e+00
2.23745599e-01 -1.02610268e-01 -5.05039811e-01 1.17953680e-01
5.92136264e-01 7.62906253e-01 -7.23356605e-01 -3.20999891e-01
-3.32696021e-01 4.41150993e-01 6.57615364e-01 2.81640470e-01
-8.49536836e-01 7.84849226e-02 4.32484537e-01 -1.94955226e-02
5.96568286e-02 2.52780169e-01 -7.78457463e-01 -1.05495679e+00
5.45729548e-02 -7.90350854e-01 3.41307521e-01 -3.66200358e-01
-1.01076865e+00 1.30600825e-01 2.99436331e-01 -6.69512153e-01
-2.64689118e-01 -7.44138896e-01 -3.92532259e-01 7.85150528e-01
-1.31798649e+00 -9.02310312e-01 1.42517120e-01 5.56859151e-02
4.69726734e-02 1.29303515e-01 8.36237788e-01 2.36603841e-01
-7.93612719e-01 7.91529298e-01 -3.45182627e-01 -3.16994071e-01
7.77700007e-01 -1.11742854e+00 -8.77515376e-02 7.50150502e-01
-1.57709092e-01 9.99143124e-01 9.80708957e-01 -7.55260885e-01
-9.05347884e-01 -9.46718991e-01 1.11085308e+00 -7.00194538e-01
4.01931763e-01 -3.69184047e-01 -1.11114168e+00 9.00102198e-01
-3.57584655e-01 -1.09156013e-01 8.06816399e-01 6.43256545e-01
-7.40624428e-01 2.12229460e-01 -1.35566616e+00 4.98192996e-01
1.21729565e+00 -3.32163483e-01 -5.97935498e-01 1.20264798e-01
6.81679904e-01 -3.07752013e-01 -8.59523356e-01 6.48261487e-01
4.81374800e-01 -1.17109036e+00 7.54627168e-01 -8.36287141e-01
4.62526262e-01 -1.45619795e-01 -2.02724069e-01 -1.05096626e+00
-4.39264685e-01 -3.06112707e-01 -1.97339639e-01 1.45857942e+00
9.72701728e-01 -6.37736142e-01 8.04565430e-01 1.24223387e+00
-1.57339856e-01 -8.13859701e-01 -9.64601517e-01 -6.66989386e-01
3.36849988e-01 -4.73374069e-01 4.99667883e-01 1.11526072e+00
3.07504356e-01 9.56068411e-02 -3.27280521e-01 1.65710703e-01
3.82907152e-01 -2.22298041e-01 3.22515100e-01 -1.49174440e+00
-4.82773483e-01 -5.69637001e-01 -9.07482803e-02 -5.88930845e-01
5.36717951e-01 -1.05051112e+00 7.54013881e-02 -1.14536262e+00
2.12570116e-01 -6.86354041e-01 -9.25854221e-02 8.03666472e-01
-2.39693061e-01 -2.34571602e-02 2.53523588e-01 1.03118561e-01
-3.55670720e-01 4.63331677e-02 4.64915186e-01 9.14279073e-02
-2.66141474e-01 -1.50509164e-01 -7.00962365e-01 1.15821457e+00
6.70769751e-01 -7.64431298e-01 -1.22134045e-01 -3.71467143e-01
8.73505771e-01 -2.82070071e-01 2.31650844e-01 -6.63532138e-01
-4.65229563e-02 -3.83404225e-01 7.45725483e-02 -3.20605487e-02
7.06050172e-02 -8.70147705e-01 3.53495240e-01 6.38416171e-01
-8.94617856e-01 -5.74192762e-01 2.82253832e-01 4.10652518e-01
1.71831474e-01 -8.52902710e-01 8.80268514e-01 4.86280546e-02
-2.86085993e-01 -3.91618609e-01 -3.66420776e-01 5.15123606e-02
8.03681791e-01 9.59976856e-03 -7.04964325e-02 1.87828884e-01
-1.07111228e+00 2.78079748e-01 6.00637257e-01 1.46454856e-01
1.00163147e-01 -8.49892497e-01 -5.73944569e-01 1.96411520e-01
1.76558077e-01 -1.18228957e-01 -2.88765550e-01 8.96549165e-01
-2.08568141e-01 3.87761116e-01 1.29100621e-01 -4.56797689e-01
-1.22773755e+00 6.07357264e-01 3.89271736e-01 -4.82944578e-01
-3.39216232e-01 8.29378426e-01 -1.55692231e-02 -7.02212453e-01
5.35141051e-01 -6.31594241e-01 -1.84074327e-01 2.12359741e-01
1.05724953e-01 3.50219637e-01 1.67445168e-01 -2.50325352e-01
-6.33289456e-01 5.06580055e-01 3.23112751e-03 1.22140780e-01
1.18193161e+00 1.55364737e-01 -3.14318776e-01 5.25536120e-01
6.25099063e-01 3.18685055e-01 -1.39569771e+00 -1.33379057e-01
5.70863903e-01 -4.13067549e-01 -9.57755279e-03 -1.09128594e+00
-6.51429296e-01 5.08549929e-01 3.04369688e-01 1.99724868e-01
7.05569327e-01 1.66883379e-01 -1.46390483e-01 5.00114262e-01
2.83164918e-01 -1.10342848e+00 -4.33914781e-01 6.96691632e-01
7.64843762e-01 -1.13832259e+00 2.17918649e-01 -9.53540027e-01
-5.12722909e-01 8.89135718e-01 5.49123406e-01 6.04567826e-02
4.57721055e-01 4.09290403e-01 -9.82382521e-02 -9.55720842e-02
-7.43322909e-01 -6.52125403e-02 2.18920335e-01 5.61774731e-01
6.58974826e-01 -4.07444797e-02 -7.01103687e-01 1.05644131e+00
-1.06129356e-01 1.34739444e-01 3.84594619e-01 6.58037663e-01
-2.56970435e-01 -1.32995677e+00 -2.71302611e-01 4.85582739e-01
-5.45274615e-01 -1.59312353e-01 -5.37083149e-01 5.43692827e-01
5.42533576e-01 8.54068041e-01 -1.28615219e-02 -2.41659239e-01
3.73305261e-01 5.37145138e-01 5.41643918e-01 -8.37175548e-01
-8.00125420e-01 -9.11004543e-02 5.93808651e-01 -4.68896508e-01
-2.87706763e-01 -6.84139669e-01 -1.20358968e+00 8.89042094e-02
-5.32144845e-01 1.00211911e-01 5.43794632e-01 1.17004418e+00
3.79385531e-01 5.66684604e-01 2.08797157e-02 -5.26711643e-01
-7.83080161e-01 -9.10052001e-01 -2.95959562e-01 5.30999124e-01
7.98330680e-02 -9.54840899e-01 -6.53758883e-01 -3.78801604e-03] | [8.79807186126709, 6.30372428894043] |
65ec636d-f79b-40fd-87e9-fd64c6cf9e20 | end-to-end-compressed-video-representation | 2203.15336 | null | https://arxiv.org/abs/2203.15336v1 | https://arxiv.org/pdf/2203.15336v1.pdf | End-to-End Compressed Video Representation Learning for Generic Event Boundary Detection | Generic event boundary detection aims to localize the generic, taxonomy-free event boundaries that segment videos into chunks. Existing methods typically require video frames to be decoded before feeding into the network, which demands considerable computational power and storage space. To that end, we propose a new end-to-end compressed video representation learning for event boundary detection that leverages the rich information in the compressed domain, i.e., RGB, motion vectors, residuals, and the internal group of pictures (GOP) structure, without fully decoding the video. Specifically, we first use the ConvNets to extract features of the I-frames in the GOPs. After that, a light-weight spatial-channel compressed encoder is designed to compute the feature representations of the P-frames based on the motion vectors, residuals and representations of their dependent I-frames. A temporal contrastive module is proposed to determine the event boundaries of video sequences. To remedy the ambiguities of annotations and speed up the training process, we use the Gaussian kernel to preprocess the ground-truth event boundaries. Extensive experiments conducted on the Kinetics-GEBD dataset demonstrate that the proposed method achieves comparable results to the state-of-the-art methods with $4.5\times$ faster running speed. | ['Libo Zhang', 'Tiejian Luo', 'Dexiang Hong', 'Longyin Wen', 'Xinyao Wang', 'CongCong Li'] | 2022-03-29 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Li_End-to-End_Compressed_Video_Representation_Learning_for_Generic_Event_Boundary_Detection_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Li_End-to-End_Compressed_Video_Representation_Learning_for_Generic_Event_Boundary_Detection_CVPR_2022_paper.pdf | cvpr-2022-1 | ['boundary-detection'] | ['computer-vision'] | [ 2.29582474e-01 -2.81415850e-01 -1.49242491e-01 -2.21116111e-01
-5.40059209e-01 -2.81474620e-01 4.82624024e-02 -7.90134296e-02
-4.76157069e-01 3.57086241e-01 1.75528854e-01 1.04422241e-01
2.57510871e-01 -5.04591763e-01 -8.25989723e-01 -7.33227849e-01
-5.22784829e-01 -4.37798321e-01 5.80308080e-01 4.01910782e-01
2.50537813e-01 2.30929643e-01 -1.60774589e+00 4.95173246e-01
4.65756238e-01 1.73302484e+00 3.61108720e-01 5.83225131e-01
7.94235244e-02 1.07133460e+00 -3.81623507e-01 1.10741928e-01
2.02345148e-01 -4.98935163e-01 -4.62110311e-01 4.38868314e-01
2.79907763e-01 -7.83311725e-01 -1.11054230e+00 8.81122589e-01
2.64746547e-01 4.27274883e-01 3.46451670e-01 -1.12306571e+00
-1.48875415e-01 2.28108421e-01 -5.62396288e-01 6.36932015e-01
3.71836811e-01 1.44978181e-01 8.05004120e-01 -1.07834947e+00
7.93286562e-01 8.90774786e-01 4.51367319e-01 2.67547399e-01
-6.21958137e-01 -7.25681484e-01 2.43733257e-01 7.41661191e-01
-1.57640123e+00 -4.21252429e-01 7.70610571e-01 -4.66060936e-01
1.04831481e+00 -3.76635082e-02 7.22480059e-01 9.55696464e-01
2.69907147e-01 8.84508491e-01 3.05519909e-01 -1.23568922e-01
3.06910455e-01 -6.93513036e-01 -5.81486598e-02 7.96388268e-01
-1.32921636e-01 3.62888798e-02 -7.85800099e-01 3.02242070e-01
8.54654789e-01 3.22490960e-01 -5.51725864e-01 -5.60602434e-02
-1.12915087e+00 4.84589219e-01 2.71367788e-01 3.63690816e-02
-6.52963877e-01 3.64461064e-01 6.85159743e-01 -1.56595428e-02
3.34224284e-01 -4.13940936e-01 -3.81496638e-01 -2.39735901e-01
-1.25434911e+00 -2.13279836e-02 4.95741189e-01 8.87318373e-01
7.44427681e-01 3.53200138e-02 -4.54397619e-01 4.80790913e-01
2.65039355e-01 -4.21839990e-02 3.89429688e-01 -1.06114435e+00
5.91486096e-01 5.25677741e-01 6.73378929e-02 -1.06114888e+00
-1.89248055e-01 -2.26313323e-02 -7.67059386e-01 -1.98602244e-01
1.69497281e-01 -1.40345320e-01 -8.94611299e-01 1.39083874e+00
3.82923067e-01 1.02604389e+00 -1.53334178e-02 1.18784571e+00
7.75591612e-01 1.00200462e+00 9.61376131e-02 -3.85538757e-01
1.35393429e+00 -9.32216644e-01 -6.56887412e-01 -2.06773758e-01
2.61667639e-01 -5.31546116e-01 4.95982587e-01 2.14216366e-01
-1.15883970e+00 -7.50100851e-01 -1.18133831e+00 -5.65095209e-02
1.30866855e-01 4.29783791e-01 3.13275397e-01 4.73430939e-02
-7.75001049e-01 7.10552573e-01 -1.13972580e+00 3.01884077e-02
5.32292008e-01 2.04724848e-01 -2.41375640e-01 -2.74432659e-01
-1.20218575e+00 1.99695244e-01 7.98122883e-01 3.00108910e-01
-1.32996917e+00 -5.16236544e-01 -1.15201545e+00 2.45260343e-01
5.68761408e-01 -2.49801487e-01 8.78465295e-01 -9.91267622e-01
-1.14466906e+00 4.55119073e-01 -2.44787678e-01 -5.77280521e-01
1.82428822e-01 -2.19648987e-01 -2.86155283e-01 8.47226679e-01
-6.70013130e-02 7.90974677e-01 1.04182911e+00 -6.40584350e-01
-1.07310116e+00 -2.57365387e-02 -9.95030254e-02 2.42395833e-01
-2.22741425e-01 2.82464385e-01 -1.10204124e+00 -7.48123109e-01
1.68586791e-01 -6.71676934e-01 1.05017973e-02 1.04610875e-01
-1.69059843e-01 -2.68586218e-01 9.64377880e-01 -1.18465030e+00
1.36211038e+00 -2.70189953e+00 1.70418024e-01 -2.54547950e-02
1.74428403e-01 1.85375229e-01 -4.67481576e-02 -4.95572314e-02
-1.40015513e-01 -2.37365991e-01 -4.45905998e-02 -2.70598531e-01
-1.43099278e-01 1.44417241e-01 -3.60970736e-01 7.47039676e-01
3.63256335e-01 6.38707161e-01 -7.49809325e-01 -6.76423669e-01
3.80375177e-01 5.65794349e-01 -5.76618016e-01 4.00007486e-01
-1.00051060e-01 4.41686332e-01 -3.63658845e-01 5.90703905e-01
5.15076220e-01 -2.44666159e-01 9.79233533e-04 -4.62742388e-01
-2.40078926e-01 3.16438049e-01 -1.16087580e+00 1.98184776e+00
6.50676191e-02 7.85938859e-01 2.51027625e-02 -1.12227821e+00
5.47054291e-01 4.38265800e-01 8.58242750e-01 -6.03703499e-01
3.33660901e-01 1.01155698e-01 -3.26424748e-01 -6.33388758e-01
3.86625916e-01 4.18416440e-01 -7.40959309e-03 6.00045435e-02
1.59798652e-01 5.90471387e-01 4.92864996e-01 -5.75111993e-03
1.21719980e+00 4.58914757e-01 1.17505200e-01 2.07102343e-01
6.45651221e-01 -2.93695360e-01 9.88800585e-01 2.63855219e-01
-3.74268115e-01 7.07666218e-01 5.18077731e-01 -4.50585604e-01
-9.56314087e-01 -1.02898610e+00 2.14433610e-01 9.91191387e-01
5.04302204e-01 -5.75629115e-01 -7.85791099e-01 -5.19332111e-01
-3.14520329e-01 2.72222608e-01 -4.52455044e-01 -3.04681361e-01
-9.70108986e-01 -3.39082211e-01 3.33551645e-01 8.21952581e-01
7.70184577e-01 -1.00300992e+00 -1.11890566e+00 4.75853652e-01
-4.71213460e-01 -1.52536440e+00 -7.06121862e-01 1.44194022e-01
-7.59435058e-01 -1.11457133e+00 -6.06143594e-01 -1.07478118e+00
4.58062768e-01 2.83880353e-01 6.56214654e-01 -3.50532942e-02
-4.05152380e-01 1.32171124e-01 -5.06595254e-01 2.85056651e-01
1.79296046e-01 -4.80136067e-01 -2.79904753e-01 2.09897339e-01
3.26915205e-01 -4.77748752e-01 -1.02306330e+00 3.74074012e-01
-9.30624485e-01 1.92616284e-01 4.83124137e-01 7.25798249e-01
8.96861672e-01 2.48039797e-01 2.15694070e-01 1.34009346e-02
-1.62661403e-01 -5.42210281e-01 -6.36030197e-01 1.38727963e-01
1.04219899e-01 -2.26271659e-01 5.18856764e-01 -4.95419383e-01
-8.82846773e-01 1.99620545e-01 5.58232889e-02 -9.52695489e-01
-5.44187762e-02 5.12873232e-01 -1.89548135e-01 1.90061063e-01
1.26569480e-01 4.92448211e-01 -3.59787583e-01 -3.07212979e-01
1.27269030e-01 5.68502843e-01 9.32892144e-01 -2.54976690e-01
2.79137164e-01 5.44609785e-01 -2.86596656e-01 -6.88891947e-01
-7.32515037e-01 -6.88571453e-01 -4.19054627e-01 -4.57928777e-01
1.19865370e+00 -1.33675146e+00 -5.31784356e-01 4.45627749e-01
-1.22788095e+00 -2.32213587e-01 -1.77503139e-01 7.83509433e-01
-7.09094644e-01 6.74076617e-01 -1.00996161e+00 -4.06452119e-01
-3.91482770e-01 -1.13166189e+00 1.10001564e+00 3.96459818e-01
3.89919728e-02 -4.18325901e-01 -3.13164949e-01 1.01741076e-01
-1.49390828e-02 2.47173727e-01 6.39172196e-01 -3.08535516e-01
-9.02189255e-01 -8.15711021e-02 -4.46904659e-01 4.71713066e-01
-1.30750507e-01 -3.07645816e-02 -7.28664100e-01 -3.01300853e-01
1.28150970e-01 -8.29166248e-02 1.05474555e+00 5.95850348e-01
1.57603109e+00 -1.68896407e-01 -3.53606820e-01 9.74252045e-01
1.17639697e+00 4.23906624e-01 8.26199055e-01 5.78978695e-02
6.41489208e-01 2.62081146e-01 8.54028285e-01 9.63317633e-01
3.18592370e-01 4.84815389e-01 5.14357448e-01 2.10714906e-01
-2.87258208e-01 -8.71898308e-02 7.25758553e-01 7.09058166e-01
2.14488227e-02 -3.49375844e-01 -5.89118481e-01 5.60022116e-01
-1.92355490e+00 -9.96281803e-01 2.13586614e-01 2.01073360e+00
6.86236501e-01 1.00485943e-01 -1.91046730e-01 8.39090943e-02
1.04695451e+00 3.96218568e-01 -6.43674493e-01 2.45714694e-01
1.64982527e-02 1.55571893e-01 4.11883533e-01 -6.26596585e-02
-1.40089726e+00 8.51950109e-01 5.28955793e+00 7.96176255e-01
-1.15671849e+00 1.33920014e-01 6.19617343e-01 -4.47713077e-01
3.32534790e-01 -5.80191659e-03 -7.38030672e-01 8.03653777e-01
1.01026785e+00 1.87260330e-01 5.81823349e-01 7.08103776e-01
3.73085439e-01 -2.13884622e-01 -1.22620559e+00 1.40381336e+00
1.84590667e-01 -1.37370574e+00 -3.13346922e-01 -2.80787528e-01
5.96924961e-01 1.54015943e-01 -3.07354093e-01 2.93249488e-01
-3.69950354e-01 -7.03227758e-01 1.12461174e+00 5.49200654e-01
9.86409068e-01 -7.93911457e-01 6.21924162e-01 6.82062209e-02
-1.65108955e+00 -2.57142246e-01 -5.14275908e-01 -4.97000292e-04
2.75589108e-01 4.34083879e-01 -3.92225623e-01 4.49054301e-01
9.48784947e-01 1.09586096e+00 -2.49766052e-01 1.17304349e+00
-2.33029783e-01 6.39445186e-01 -3.52105767e-01 5.31009912e-01
3.13206911e-01 -1.71097443e-02 3.15078646e-01 1.22821093e+00
6.28819346e-01 2.34778956e-01 3.37061644e-01 6.63930595e-01
-6.64805993e-02 -2.18495518e-01 -2.02205718e-01 -1.07207552e-01
3.62837225e-01 1.18913651e+00 -9.21576262e-01 -5.10719478e-01
-7.57646799e-01 1.37653220e+00 1.39640719e-01 6.04824543e-01
-1.28372097e+00 -4.13505256e-01 6.65994704e-01 -1.27770051e-01
1.07382667e+00 -4.26534027e-01 2.88972199e-01 -1.25676847e+00
2.33668119e-01 -5.98390937e-01 5.11709094e-01 -6.81125760e-01
-9.09325063e-01 3.30011308e-01 -6.56924397e-02 -1.31707263e+00
-3.68373901e-01 -4.81465191e-01 -5.60209036e-01 6.16149962e-01
-1.49630702e+00 -7.46821404e-01 -6.46056712e-01 7.78326452e-01
7.68210173e-01 5.96190542e-02 2.34690368e-01 5.39189816e-01
-8.36549878e-01 3.83720636e-01 -4.68719453e-02 5.36468625e-01
5.18991947e-01 -5.78153431e-01 3.19306076e-01 1.23505223e+00
-6.77960813e-02 1.48074031e-01 3.05487156e-01 -7.84706771e-01
-1.50301898e+00 -1.37709475e+00 4.78431642e-01 2.87317276e-01
5.56261122e-01 -2.69576609e-01 -8.36366355e-01 6.00692749e-01
-1.09152300e-02 5.80274284e-01 4.71859336e-01 -6.39765143e-01
-1.02062255e-01 -4.89074811e-02 -6.75079346e-01 3.64424348e-01
1.09823382e+00 -6.18407309e-01 -4.95765328e-01 4.33714092e-02
7.22417474e-01 -5.61214149e-01 -7.97685206e-01 3.75351310e-01
2.93702781e-01 -8.14478278e-01 9.50447679e-01 -2.73447186e-01
5.22456884e-01 -5.88625014e-01 -4.01296675e-01 -6.70560896e-01
-2.22879380e-01 -5.47224879e-01 -5.59042692e-01 1.07413840e+00
-2.70213306e-01 1.45701086e-02 6.43897593e-01 3.42621922e-01
-5.42100668e-01 -8.13201129e-01 -1.18145156e+00 -4.96407568e-01
-6.69497252e-01 -6.99345469e-01 1.93423808e-01 5.57212710e-01
-2.79766396e-02 8.69664401e-02 -4.16791379e-01 3.40375483e-01
4.78778660e-01 1.79216802e-01 2.63359189e-01 -7.37663865e-01
-2.51742393e-01 -1.42299980e-01 -6.49230301e-01 -1.41683090e+00
2.20592201e-01 -7.20490575e-01 3.39239359e-01 -1.34103191e+00
1.98508233e-01 -1.21442974e-01 -5.26710987e-01 4.37648684e-01
-1.88724667e-01 3.58097017e-01 2.15422139e-01 2.53668338e-01
-1.10192716e+00 6.19625330e-01 9.78334844e-01 -8.60192329e-02
-1.43869042e-01 -5.90247393e-01 -7.99345076e-02 8.32843363e-01
5.25008440e-01 -3.87041330e-01 -3.15823674e-01 -3.32450509e-01
-3.19500268e-02 3.70706111e-01 5.57219505e-01 -1.30082059e+00
4.43942368e-01 -4.88499701e-02 7.35550284e-01 -9.89642143e-01
3.70021105e-01 -7.15959311e-01 6.82900399e-02 2.86328614e-01
-1.27052531e-01 -9.51365978e-02 1.26951873e-01 7.15078950e-01
-4.06096786e-01 -2.30889499e-01 6.07630491e-01 -1.91207044e-02
-1.14015460e+00 6.09447002e-01 -4.71079201e-01 4.44558039e-02
1.30287385e+00 -2.95489967e-01 -4.28901576e-02 -9.86128673e-03
-6.18961036e-01 3.21825504e-01 4.48281407e-01 3.08655381e-01
1.05125403e+00 -1.19114137e+00 -4.38899636e-01 4.43400145e-01
-1.55394733e-01 1.98446035e-01 6.44739568e-01 9.58907664e-01
-6.83003187e-01 2.11272359e-01 -3.15077007e-01 -6.84647083e-01
-1.09696364e+00 7.50530839e-01 2.30933264e-01 4.83975336e-02
-9.84199703e-01 8.19070578e-01 4.37263727e-01 6.91732526e-01
5.12343287e-01 -5.77589333e-01 -1.84631944e-01 2.61021405e-01
8.09698045e-01 2.98201382e-01 -2.06731170e-01 -6.22917533e-01
-4.46467936e-01 3.67838144e-01 -9.74125490e-02 8.11388865e-02
1.18863130e+00 -3.02059084e-01 7.36188665e-02 1.85464352e-01
1.40373921e+00 -3.97843093e-01 -2.11580133e+00 -1.76595300e-01
-1.33107647e-01 -4.56465364e-01 2.65873462e-01 -2.30702996e-01
-1.37428653e+00 8.66793454e-01 7.17385709e-01 -2.68885285e-01
1.51635504e+00 1.03703178e-02 1.17257524e+00 4.56139781e-02
3.09646368e-01 -1.14055693e+00 1.78299561e-01 4.50731814e-01
5.30031919e-01 -9.13187981e-01 -8.24728906e-02 -4.09093827e-01
-4.22219247e-01 1.24089289e+00 4.54690814e-01 -2.66157866e-01
4.90677834e-01 2.10494250e-01 -3.54437292e-01 9.27447341e-04
-7.42452025e-01 -5.58271408e-02 3.38742554e-01 3.57117802e-01
1.33723810e-01 -2.64541596e-01 -2.11956531e-01 7.13694692e-01
4.12196755e-01 2.67728508e-01 3.05912524e-01 9.66929972e-01
-4.30427969e-01 -5.00958562e-01 -2.20154166e-01 3.58167082e-01
-5.70501924e-01 -1.59278497e-01 1.62302494e-01 4.26194161e-01
3.60951662e-01 8.68852019e-01 5.25028944e-01 -5.23662567e-01
7.48452768e-02 5.24862632e-02 4.44480360e-01 -2.51194179e-01
-3.33567500e-01 4.34376061e-01 -7.18598813e-02 -1.09794188e+00
-5.25641024e-01 -6.97031498e-01 -1.78137314e+00 -3.26247923e-02
-4.90742773e-02 -1.45076543e-01 3.91439736e-01 8.90516758e-01
5.16742527e-01 7.18456745e-01 5.82596660e-01 -1.22335982e+00
-1.66397139e-01 -7.95698881e-01 -4.23115015e-01 4.44293410e-01
3.08278263e-01 -7.47007847e-01 -2.76612818e-01 4.98592973e-01] | [8.813159942626953, 0.26419568061828613] |
3abab6ce-f14a-4493-8bdf-4182183f67ec | dsrn-an-efficient-deep-network-for-image | 2102.09242 | null | https://arxiv.org/abs/2102.09242v2 | https://arxiv.org/pdf/2102.09242v2.pdf | DSRN: an Efficient Deep Network for Image Relighting | Custom and natural lighting conditions can be emulated in images of the scene during post-editing. Extraordinary capabilities of the deep learning framework can be utilized for such purpose. Deep image relighting allows automatic photo enhancement by illumination-specific retouching. Most of the state-of-the-art methods for relighting are run-time intensive and memory inefficient. In this paper, we propose an efficient, real-time framework Deep Stacked Relighting Network (DSRN) for image relighting by utilizing the aggregated features from input image at different scales. Our model is very lightweight with total size of about 42 MB and has an average inference time of about 0.0116s for image of resolution $1024 \times 1024$ which is faster as compared to other multi-scale models. Our solution is quite robust for translating image color temperature from input image to target image and also performs moderately for light gradient generation with respect to the target image. Additionally, we show that if images illuminated from opposite directions are used as input, the qualitative results improve over using a single input image. | ['Himanshu Kumar', 'Saikat Dutta', 'Nisarg A. Shah', 'Sourya Dipta Das'] | 2021-02-18 | null | null | null | null | ['image-relighting'] | ['computer-vision'] | [ 5.16407371e-01 -1.95464194e-01 3.62211376e-01 -2.61682600e-01
-5.16156256e-01 -4.12734091e-01 3.72175336e-01 -4.37579125e-01
-5.32504201e-01 6.72795236e-01 -2.09108949e-01 -3.84619772e-01
4.83619034e-01 -8.71231139e-01 -1.19412899e+00 -9.31754291e-01
3.38821679e-01 -1.25704736e-01 3.77771914e-01 -3.18696678e-01
4.65646416e-01 7.35347092e-01 -1.72035170e+00 2.28144646e-01
6.83329999e-01 8.51549447e-01 4.03538346e-01 1.25471282e+00
-4.60244976e-02 8.58575881e-01 -6.98772430e-01 -2.75845498e-01
4.90008712e-01 -3.50927770e-01 -7.03083932e-01 1.67297766e-01
9.12528336e-01 -7.93655753e-01 -3.52417558e-01 1.00398231e+00
7.88061500e-01 1.76172420e-01 1.75536305e-01 -9.38478827e-01
-7.75449097e-01 1.45267308e-01 -1.00729918e+00 6.75435662e-02
2.04935312e-01 3.94483000e-01 2.59236366e-01 -8.28425467e-01
6.19059443e-01 1.04644549e+00 4.84081686e-01 4.10658687e-01
-1.32228637e+00 -5.42625606e-01 -2.26489007e-01 1.87053591e-01
-1.19979823e+00 -3.84484261e-01 7.10680008e-01 -5.21601699e-02
1.12789011e+00 4.43035066e-01 6.18480504e-01 7.90725589e-01
5.60459077e-01 2.90303707e-01 1.61398804e+00 -6.30049348e-01
9.22900289e-02 2.44734585e-01 -4.17744756e-01 8.88773382e-01
-7.33507350e-02 2.83343345e-01 -6.75540924e-01 2.75928646e-01
1.16114700e+00 -1.29468441e-01 -1.75687835e-01 -3.05007659e-02
-1.22462428e+00 4.49850082e-01 6.84451044e-01 4.35931943e-02
-3.39830928e-02 6.97081566e-01 2.79040217e-01 2.60288119e-01
5.02660453e-01 3.57569158e-01 -4.08169806e-01 1.07700557e-01
-1.08273518e+00 -3.16455066e-02 3.34084630e-01 9.04446363e-01
1.27644825e+00 3.94809484e-01 -4.34412919e-02 6.04774177e-01
-9.40431133e-02 7.27615833e-01 3.82222235e-01 -1.33133340e+00
1.82352051e-01 3.74321640e-01 2.88213640e-01 -9.18766022e-01
-4.40364331e-01 -2.01683417e-02 -1.12051272e+00 8.73092532e-01
2.87793607e-01 -2.76205331e-01 -1.18261635e+00 1.30386090e+00
6.33333743e-01 2.26336002e-01 1.18800133e-01 8.25837433e-01
8.22276890e-01 1.03233492e+00 -3.42272937e-01 -4.92680334e-02
1.37909162e+00 -1.19591272e+00 -5.67214727e-01 -2.89678454e-01
3.18419069e-01 -1.07111239e+00 1.41390538e+00 5.79863310e-01
-1.31334841e+00 -7.80809641e-01 -1.11445987e+00 -7.63876021e-01
-5.51476479e-01 2.16928318e-01 6.07541621e-01 7.62975097e-01
-1.62716377e+00 5.54314494e-01 -5.14721274e-01 -2.51013666e-01
2.15643659e-01 3.94808054e-01 -2.18514442e-01 -1.10994570e-01
-8.48567843e-01 8.69775116e-01 1.76289037e-01 1.76845700e-01
-8.65397036e-01 -7.23413944e-01 -7.94650555e-01 -2.13378683e-01
1.33649424e-01 -6.39012873e-01 9.79710281e-01 -1.25351417e+00
-2.01545095e+00 9.12860513e-01 -3.17994475e-01 -3.18722695e-01
6.62058949e-01 -8.11315402e-02 -2.45391399e-01 2.80772120e-01
-3.32624406e-01 9.33029294e-01 1.27041101e+00 -1.39461267e+00
-5.17049253e-01 -7.86543116e-02 7.28962272e-02 2.31185779e-01
-3.05712819e-01 1.74086899e-01 -6.23944044e-01 -4.18013424e-01
-3.50351244e-01 -7.84074962e-01 -2.23337337e-01 4.04679805e-01
-3.61624181e-01 5.46368957e-01 9.11962509e-01 -7.38355279e-01
8.56897116e-01 -2.02408242e+00 -9.75134000e-02 -1.23328440e-01
-8.87557492e-02 3.61352056e-01 -3.35127264e-01 3.11829448e-01
-6.74878210e-02 -7.16462061e-02 -2.26609930e-01 -3.94838333e-01
-2.91515678e-01 2.07193166e-01 -4.29910839e-01 5.51195502e-01
-2.71664537e-03 9.43989873e-01 -6.41153693e-01 -5.60344279e-01
7.49488950e-01 8.31674039e-01 -4.66284335e-01 2.44940221e-01
-9.93558466e-02 3.92106920e-01 3.78743231e-01 5.08562386e-01
1.15838253e+00 2.53029214e-03 1.15341647e-02 -6.04764640e-01
-5.56721151e-01 -2.21283168e-01 -1.08129740e+00 1.71526682e+00
-8.52476180e-01 9.82991636e-01 -6.88030794e-02 -4.63549972e-01
8.25350404e-01 -1.12958618e-01 1.08065434e-01 -9.35611904e-01
2.19713822e-01 -8.22940469e-02 -4.26696092e-01 -3.97804081e-01
9.12954450e-01 1.14041500e-01 2.26970866e-01 6.41643524e-01
-2.34627843e-01 -7.24968016e-01 2.55674243e-01 4.71375212e-02
6.90801919e-01 5.13235688e-01 1.78421393e-01 -1.03224002e-01
4.12946463e-01 -2.67087907e-01 1.70256510e-01 7.73369968e-01
2.96577737e-02 7.62118280e-01 2.46254668e-01 -7.58361697e-01
-1.47083068e+00 -8.65813792e-01 -7.24020898e-02 1.16032195e+00
3.38250428e-01 -1.14698866e-02 -9.84648466e-01 -1.40523404e-01
-3.81810635e-01 6.92116857e-01 -7.22241938e-01 1.47570506e-01
-6.40795827e-01 -9.27649200e-01 4.08401638e-01 4.17223960e-01
1.09308088e+00 -1.10071909e+00 -1.06021094e+00 1.37370583e-02
-7.34762326e-02 -1.09638488e+00 -3.98611933e-01 1.12095468e-01
-8.27572942e-01 -7.36848295e-01 -7.03943610e-01 -7.19798803e-01
8.20806026e-01 5.96099496e-01 1.24071836e+00 1.46028653e-01
-6.96115732e-01 2.02006996e-01 -1.15455471e-01 -1.71460256e-01
-4.15586799e-01 -3.92012149e-01 -2.21431866e-01 -9.26152021e-02
-1.75212190e-01 -4.39511359e-01 -9.63542402e-01 3.70288998e-01
-1.42299867e+00 6.43006265e-01 3.76466513e-01 7.78039694e-01
6.64841294e-01 2.56519675e-01 -1.87210709e-01 -6.76025867e-01
9.09891352e-02 2.87817329e-01 -9.03661668e-01 2.81150728e-01
-5.15887678e-01 2.37204917e-02 6.85473561e-01 -3.73425752e-01
-1.54089344e+00 1.74389139e-01 1.62614081e-02 -6.02449402e-02
-2.65046936e-02 -2.41787374e-01 1.86761498e-01 -6.77289188e-01
7.00861156e-01 3.76626134e-01 -1.24411918e-01 -2.84906384e-02
6.91917062e-01 4.45417166e-01 6.14532173e-01 -3.12195659e-01
9.19097185e-01 9.13149059e-01 2.30798364e-01 -9.50804412e-01
-6.51407659e-01 8.61084163e-02 -5.98906100e-01 -4.51904088e-01
8.54972720e-01 -8.48339558e-01 -8.13147664e-01 9.99658704e-01
-1.09676790e+00 -1.00826907e+00 -2.35457063e-01 -1.14276797e-01
-5.25377214e-01 4.64937717e-01 -6.59656286e-01 -5.49322903e-01
-5.32762527e-01 -1.22427607e+00 1.24876654e+00 6.13132000e-01
3.85725439e-01 -1.00655770e+00 -1.31558985e-01 3.87200803e-01
6.48238659e-01 2.49923125e-01 6.41696751e-01 7.14539707e-01
-7.68404424e-01 4.17134240e-02 -7.82953203e-01 5.08558571e-01
1.65496752e-01 3.55201334e-01 -1.15543699e+00 -3.12557787e-01
-1.11337215e-01 -4.93980974e-01 8.79191041e-01 4.61181551e-01
1.43576312e+00 -2.24797308e-01 2.46181674e-02 9.70094860e-01
1.87033761e+00 -3.22648957e-02 1.16845214e+00 5.51373899e-01
9.26807225e-01 3.52114350e-01 6.04673624e-01 4.34675395e-01
1.90630674e-01 5.55464685e-01 5.87691247e-01 -9.58624959e-01
-4.20077771e-01 1.44274801e-01 4.97890323e-01 3.02465081e-01
-7.10395202e-02 -4.52727675e-01 -6.04015172e-01 2.95165449e-01
-1.44943488e+00 -8.82948518e-01 -2.57947683e-01 2.16980338e+00
1.01956820e+00 -2.37552106e-01 -3.71953487e-01 -1.54286951e-01
6.42056286e-01 2.66551852e-01 -7.84970224e-01 -7.84663260e-01
-3.62124890e-01 3.44764411e-01 9.45258558e-01 7.50716865e-01
-8.03798676e-01 1.10430801e+00 6.85981131e+00 5.59157431e-01
-1.33703995e+00 -3.75053771e-02 1.01374137e+00 -2.60895848e-01
-1.66595384e-01 -1.93882436e-01 -6.03453994e-01 2.94978499e-01
8.42034221e-01 2.46965021e-01 9.30670857e-01 5.07676303e-01
4.32922572e-01 -6.83258772e-01 -6.79087877e-01 1.15099192e+00
3.68749768e-01 -1.38790905e+00 -5.98298311e-02 -4.74326238e-02
9.78703022e-01 1.29942909e-01 4.11229968e-01 -2.19665498e-01
5.03715336e-01 -6.86437130e-01 5.42945087e-01 4.30523902e-01
1.23382461e+00 -8.14501643e-01 4.54951048e-01 -7.59454295e-02
-9.63492632e-01 5.60858510e-02 -7.03226089e-01 1.60879150e-01
2.22522303e-01 6.44721270e-01 -6.95969641e-01 3.56440336e-01
1.06019318e+00 4.66021478e-01 -9.13764894e-01 5.85157692e-01
-2.17801318e-01 3.40072483e-01 -4.14183408e-01 2.84817427e-01
1.05203755e-01 -2.85766751e-01 4.67119599e-03 1.26220679e+00
6.11117065e-01 -1.88749775e-01 -3.61909777e-01 6.10648930e-01
-1.12296186e-01 -1.00418657e-01 -6.50062799e-01 3.33321065e-01
-7.93835055e-03 1.50022840e+00 -8.11387956e-01 -5.55680037e-01
-3.18533450e-01 1.65991271e+00 1.97441012e-01 5.75529099e-01
-1.09671557e+00 -5.50586879e-01 3.99818182e-01 -7.09370449e-02
3.58954787e-01 -1.97160736e-01 -2.61187911e-01 -1.05026352e+00
-1.69433966e-01 -6.32824659e-01 -8.57641473e-02 -1.60008514e+00
-7.96701193e-01 6.57307744e-01 -2.06827372e-01 -1.06769252e+00
1.33087784e-01 -7.80840516e-01 -5.12926996e-01 8.32857966e-01
-1.94369876e+00 -1.25461221e+00 -7.43368566e-01 8.55373263e-01
4.49644059e-01 2.60245144e-01 6.92385018e-01 1.41803071e-01
-5.24202704e-01 4.37923670e-01 3.87926430e-01 -1.34029046e-01
1.01188219e+00 -1.28071749e+00 6.26601160e-01 1.29452097e+00
-3.08568440e-02 3.58648866e-01 9.07960296e-01 -3.14808637e-01
-1.63255167e+00 -1.16446733e+00 4.50615853e-01 -1.65262297e-01
4.79872823e-01 -3.49776000e-01 -6.76631451e-01 4.51673210e-01
8.49611580e-01 1.67570502e-01 4.26233411e-01 -5.15073180e-01
-3.90369892e-01 -5.78236282e-01 -1.32800758e+00 9.42811012e-01
7.63599515e-01 -4.41017747e-01 2.21402496e-01 4.74195510e-01
7.92278707e-01 -7.64509082e-01 -6.45388484e-01 -1.87028460e-02
5.64697802e-01 -1.43450511e+00 1.09014094e+00 5.45976833e-02
5.86949110e-01 -5.51584423e-01 -6.47500530e-02 -1.15324569e+00
-2.13895082e-01 -9.04992342e-01 2.11208031e-01 1.09268808e+00
1.23560704e-01 -6.98157430e-01 5.28226972e-01 4.89204466e-01
3.68425739e-03 -2.88648576e-01 -7.65652597e-01 -3.86161655e-01
-2.38920569e-01 -3.38776588e-01 5.53600311e-01 4.70664829e-01
-5.91657102e-01 5.49400635e-02 -8.10937822e-01 2.26438016e-01
7.51889169e-01 2.99594522e-01 1.13490498e+00 -4.49540496e-01
-2.44639292e-01 -8.68766308e-02 -1.54949710e-01 -8.80840778e-01
-1.98757619e-01 -3.33150089e-01 1.47890016e-01 -1.49087167e+00
1.57218739e-01 -1.75928935e-01 2.61753667e-02 5.91942847e-01
-1.08674750e-01 9.99287724e-01 1.29541263e-01 -2.29994714e-01
-2.23054036e-01 3.19174021e-01 1.52632463e+00 -7.88637027e-02
8.07169883e-04 -5.92265904e-01 -5.16277075e-01 5.80537379e-01
9.74187791e-01 -1.69793218e-01 -3.99974227e-01 -7.58697629e-01
5.63735366e-01 -2.05372602e-01 5.44918478e-01 -9.44728673e-01
-4.70871292e-02 -2.79737592e-01 6.63303971e-01 -6.02790058e-01
4.80279624e-01 -7.39357352e-01 2.35711008e-01 3.42795014e-01
-1.77275553e-01 9.52801108e-02 4.35139716e-01 3.32369715e-01
1.20294571e-01 -1.12528123e-01 1.20129395e+00 -1.89822465e-01
-8.25053096e-01 1.67854890e-01 -5.01827121e-01 -3.00791740e-01
1.04723489e+00 -4.49502736e-01 -5.40188432e-01 -3.45530301e-01
-1.23595737e-01 -3.44759345e-01 9.25060987e-01 5.18174320e-02
7.23161399e-01 -1.09933698e+00 -4.31062728e-01 2.64844805e-01
-1.35333672e-01 -1.85308754e-02 6.42979145e-01 5.99033296e-01
-1.23516476e+00 -4.16094996e-02 -3.51585388e-01 -5.79080284e-01
-1.48481357e+00 6.40439689e-01 4.27573979e-01 -3.65832224e-02
-6.80203557e-01 8.93175483e-01 2.67859817e-01 -1.80788636e-01
-1.33707017e-01 -4.29613233e-01 3.48527580e-01 -2.82904834e-01
8.46205890e-01 4.53518689e-01 1.12132050e-01 -4.40833449e-01
-1.46661429e-02 9.42864895e-01 -9.88327339e-02 -2.46707231e-01
1.32404268e+00 -4.52221692e-01 -4.31360930e-01 1.56327665e-01
1.21108115e+00 -1.10824257e-01 -1.68542063e+00 -5.14660254e-02
-9.20361400e-01 -7.05773830e-01 4.61055011e-01 -9.28319156e-01
-1.32523608e+00 9.95788753e-01 1.00953138e+00 -7.18142241e-02
1.56179047e+00 -5.17359257e-01 7.96871960e-01 4.22311157e-01
3.29417259e-01 -1.37626588e+00 2.92652398e-01 2.31074616e-01
8.14320982e-01 -1.36589110e+00 3.83259147e-01 -8.72982666e-02
-6.56766295e-01 1.21876395e+00 5.82085729e-01 -7.82689750e-02
1.97857141e-01 5.91049075e-01 4.19140935e-01 1.90230552e-02
-5.14832079e-01 8.95469561e-02 -1.65830292e-02 7.20966458e-01
3.00933599e-01 -2.00823307e-01 2.01334894e-01 -8.14557254e-01
-5.12073822e-02 -2.71287188e-02 9.26066220e-01 7.57237792e-01
-4.08366770e-01 -8.90237153e-01 -5.63090205e-01 3.10039651e-02
-2.28264555e-01 -4.12852287e-01 -8.89146701e-02 6.05181873e-01
7.52523914e-02 8.70101035e-01 9.79672745e-02 -1.35800347e-01
-1.19474195e-01 -3.46537620e-01 8.39043856e-01 -2.18562379e-01
-6.11529708e-01 1.43329035e-02 -2.70511985e-01 -1.01644456e+00
-7.32830822e-01 -2.88092613e-01 -1.07331455e+00 -5.10191500e-01
-9.46794003e-02 -6.04979336e-01 1.01693916e+00 4.81568992e-01
3.25620681e-01 4.96155411e-01 7.94825196e-01 -1.47643876e+00
2.20950723e-01 -6.10536039e-01 -5.29395938e-01 8.79034996e-02
4.54736143e-01 -2.21933305e-01 -4.09028649e-01 4.65942115e-01] | [10.611298561096191, -2.3420965671539307] |
1d030f82-2748-49d0-a4ca-d08906481978 | flame-facial-landmark-heatmap-activated | 2110.04828 | null | https://arxiv.org/abs/2110.04828v3 | https://arxiv.org/pdf/2110.04828v3.pdf | FLAME: Facial Landmark Heatmap Activated Multimodal Gaze Estimation | 3D gaze estimation is about predicting the line of sight of a person in 3D space. Person-independent models for the same lack precision due to anatomical differences of subjects, whereas person-specific calibrated techniques add strict constraints on scalability. To overcome these issues, we propose a novel technique, Facial Landmark Heatmap Activated Multimodal Gaze Estimation (FLAME), as a way of combining eye anatomical information using eye landmark heatmaps to obtain precise gaze estimation without any person-specific calibration. Our evaluation demonstrates a competitive performance of about 10% improvement on benchmark datasets ColumbiaGaze and EYEDIAP. We also conduct an ablation study to validate our method. | ['Francois Bremond', 'Michal Balazia', 'Neelabh Sinha'] | 2021-10-10 | null | null | null | null | ['gaze-estimation'] | ['computer-vision'] | [-3.23282391e-01 1.31118596e-01 -8.42700377e-02 -5.38349271e-01
-2.68229008e-01 -3.74748588e-01 4.69273627e-01 -3.29812318e-01
-6.25989556e-01 7.12360978e-01 2.00172037e-01 -4.23792489e-02
1.45629212e-01 -6.59334846e-03 -4.26392913e-01 -4.59692389e-01
2.01902404e-01 -2.71970443e-02 1.41834855e-01 6.84597045e-02
6.89124346e-01 4.58631545e-01 -1.73612821e+00 -1.87415853e-01
7.44680226e-01 9.88903046e-01 -3.37909728e-01 5.51487446e-01
7.90470093e-02 2.32949525e-01 -5.66735387e-01 -3.94929945e-01
3.77063662e-01 -3.63690913e-01 -7.84740925e-01 1.64454095e-02
1.24962616e+00 -3.06369573e-01 4.78083789e-02 8.20660710e-01
7.58885801e-01 -5.11131957e-02 8.06077540e-01 -1.60602725e+00
-5.42007983e-01 -2.37505957e-01 -1.35477448e+00 4.75559622e-01
8.26568484e-01 2.94642895e-01 4.28087503e-01 -8.01747084e-01
6.10821664e-01 1.09980226e+00 7.90187240e-01 8.64377558e-01
-1.13441730e+00 -8.32161486e-01 1.74944878e-01 2.21442625e-01
-1.82312608e+00 -7.77738392e-01 6.52847528e-01 -4.53644216e-01
8.31005335e-01 3.30705136e-01 7.04976082e-01 1.02656913e+00
5.63883260e-02 4.68000740e-01 1.46188867e+00 -6.08912706e-01
-2.11545318e-01 3.25718164e-01 2.86078781e-01 8.99670780e-01
2.09402204e-01 2.50012279e-01 -9.86780763e-01 6.18369430e-02
7.96536267e-01 -2.80524433e-01 -6.88968003e-01 -5.76324582e-01
-1.20194435e+00 4.21964020e-01 9.10856783e-01 -1.48670524e-01
-1.98242530e-01 -6.07507154e-02 -5.90663217e-02 1.88968629e-02
5.20723403e-01 3.84032518e-01 -1.21657073e-01 -1.40294001e-01
-1.09614408e+00 1.44258067e-01 4.31569338e-01 1.09647393e+00
7.32719183e-01 -4.55225348e-01 -3.86454701e-01 3.51731032e-01
7.00514019e-01 6.12659693e-01 3.17464143e-01 -8.32746625e-01
2.43539065e-01 7.91229606e-01 2.44840309e-01 -7.05274403e-01
-9.02883470e-01 -9.83625352e-02 -2.94703305e-01 6.97277367e-01
8.98935556e-01 -2.61984229e-01 -1.01224005e+00 1.60212743e+00
5.47696531e-01 2.43177805e-02 -5.70713580e-01 1.32655072e+00
9.14235473e-01 -2.35270321e-01 1.52422890e-01 1.02143260e-02
1.58743703e+00 -9.27876055e-01 -6.14298224e-01 -2.20549837e-01
4.99927670e-01 -5.26411295e-01 1.23512399e+00 2.21338183e-01
-1.18319166e+00 -3.56571436e-01 -8.76088023e-01 -3.15358758e-01
-3.94311965e-01 1.74380764e-01 7.60560036e-01 1.18124223e+00
-1.31249046e+00 -1.08077243e-01 -6.33260608e-01 -7.67635882e-01
6.86994374e-01 6.94001734e-01 -7.67965198e-01 4.97178882e-01
-6.90318167e-01 1.24759674e+00 2.23581176e-02 2.00352758e-01
-2.39032581e-01 -1.02188921e+00 -9.13782418e-01 -2.23767966e-01
2.38034919e-01 -1.05747974e+00 9.60943818e-01 -7.75535882e-01
-1.56504858e+00 1.52299857e+00 -9.18532073e-01 -1.23285949e-01
5.72101951e-01 -2.18641758e-01 -3.89955610e-01 7.27007017e-02
-3.30737412e-01 1.03202176e+00 1.10886896e+00 -9.81236696e-01
-6.64191663e-01 -8.15731585e-01 1.58358470e-01 4.31507915e-01
-8.67434740e-02 3.29388231e-01 -5.15164435e-01 1.17711850e-01
1.25298873e-01 -9.74921048e-01 3.34021688e-01 4.58618164e-01
-6.96906865e-01 -3.94992590e-01 3.87878478e-01 -4.74236935e-01
1.33312619e+00 -1.87971640e+00 9.72766429e-02 3.42345893e-01
9.75440919e-01 -9.35681686e-02 1.41020179e-01 -2.69445390e-01
-2.33897954e-01 8.34463313e-02 1.98678046e-01 -6.46377802e-01
-1.35369092e-01 -6.54377341e-01 2.51529843e-01 8.26921284e-01
2.82846484e-02 1.01313794e+00 -5.77914774e-01 -6.84677124e-01
6.31893128e-02 6.55032337e-01 -3.90907645e-01 -1.50208071e-01
4.52644795e-01 5.73630989e-01 -1.54941797e-01 7.50393569e-01
7.01065898e-01 -3.46133649e-01 -4.29216385e-01 -3.70007664e-01
-2.19075069e-01 -3.99637930e-02 -7.25658715e-01 1.92129326e+00
-2.57302701e-01 9.76282299e-01 -3.49523127e-01 1.16003282e-01
6.82577729e-01 5.21650212e-03 -2.32913699e-02 -5.95210910e-01
5.66274822e-01 -2.18618080e-01 -2.19675060e-02 -2.90554821e-01
3.85913670e-01 1.81289852e-01 2.60241747e-01 4.49787468e-01
4.24103253e-02 3.18499804e-01 -8.96841660e-02 -3.23923267e-02
5.83518684e-01 4.62452382e-01 3.67325664e-01 -3.00005198e-01
6.86894357e-01 -3.81215632e-01 -6.21654429e-02 3.35718751e-01
-8.07461441e-01 8.40709746e-01 5.78393161e-01 -4.96226698e-01
-6.43372118e-01 -9.80138540e-01 -3.53356808e-01 8.34749401e-01
3.44739258e-01 -6.01873755e-01 -1.19015765e+00 -1.02217901e+00
-1.55509338e-01 5.31860352e-01 -1.23865330e+00 5.26793040e-02
-1.07966855e-01 -5.49067378e-01 3.49763304e-01 3.09952080e-01
5.32854974e-01 -5.45677543e-01 -9.07255411e-01 -8.84442270e-01
-5.61088994e-02 -9.44305778e-01 -8.04979146e-01 -5.93841553e-01
-3.85522872e-01 -1.41674733e+00 -1.26828313e+00 -3.72392595e-01
1.08700371e+00 3.95795286e-01 1.08792925e+00 8.63538235e-02
-1.38979509e-01 4.99022514e-01 1.06644727e-01 -6.72494411e-01
4.15983051e-01 3.23632777e-01 3.30165237e-01 1.01375781e-01
1.10164249e+00 -2.27505028e-01 -1.04247272e+00 6.12049103e-01
1.34770632e-01 -1.34885041e-02 3.10232222e-01 4.13354129e-01
3.82630765e-01 -5.69160461e-01 -1.03148162e-01 -6.52061045e-01
7.24054098e-01 5.91437668e-02 -7.01277196e-01 5.13312280e-01
-6.42549455e-01 -2.41880510e-02 -3.83854896e-01 -3.18525314e-01
-9.63422477e-01 1.22736283e-01 3.83234084e-01 -4.57063675e-01
-4.63850290e-01 -1.27170712e-01 1.02321483e-01 -7.64411271e-01
1.25050604e+00 -1.39356866e-01 2.74585724e-01 -2.80854166e-01
2.10878760e-01 7.56871223e-01 5.50239325e-01 -9.77433249e-02
6.86571896e-01 5.90820789e-01 3.19888234e-01 -6.23045623e-01
-1.00742567e+00 -4.44814712e-01 -1.26908958e+00 -5.36915064e-01
9.07605827e-01 -8.60349238e-01 -1.45733786e+00 3.54165792e-01
-9.18034673e-01 -6.21431656e-02 2.36897707e-01 6.27246559e-01
-5.12340724e-01 1.77426964e-01 2.65983939e-01 -8.03430140e-01
-4.45863694e-01 -9.14379418e-01 1.28473341e+00 6.51145875e-01
-5.59113204e-01 -9.18263674e-01 2.09197357e-01 1.65517718e-01
2.16297939e-01 3.05723637e-01 1.48483932e-01 -1.44971922e-01
-5.46152711e-01 -3.75140995e-01 -5.34960568e-01 -2.85248876e-01
6.54455498e-02 6.28732815e-02 -1.44970953e+00 -9.83527079e-02
-4.12234783e-01 -1.07891254e-01 4.58463162e-01 6.73689365e-01
9.34865236e-01 1.69167891e-01 -6.75391734e-01 1.05310273e+00
1.11440778e+00 -5.51896155e-01 8.01351547e-01 3.56052697e-01
7.89237797e-01 7.67349958e-01 4.12592351e-01 4.49741520e-02
6.69589937e-01 8.13705206e-01 1.61335737e-01 -2.21991450e-01
-4.40350175e-01 -1.98142797e-01 -6.97766840e-02 -2.16446504e-01
-7.24413812e-01 7.23128617e-02 -1.06598949e+00 1.61313057e-01
-1.23955917e+00 -8.64964008e-01 -2.66553223e-01 2.37689662e+00
6.23669624e-01 6.82719983e-03 6.94064319e-01 -5.50456531e-02
6.36155665e-01 -2.55244225e-01 -5.81849515e-01 -8.37258101e-02
1.31652534e-01 -1.14769161e-01 6.32287681e-01 4.57558244e-01
-9.65294421e-01 9.60450649e-01 7.40239000e+00 3.35611224e-01
-1.30385566e+00 9.20303762e-02 3.91286165e-01 -5.71386099e-01
1.12385146e-01 -2.62859434e-01 -1.13833654e+00 4.77831244e-01
7.72566140e-01 5.38186841e-02 4.09057379e-01 3.47432494e-01
6.07744679e-02 -5.17968714e-01 -1.14672709e+00 1.66193902e+00
5.45667231e-01 -9.51081395e-01 -5.23054183e-01 4.50050026e-01
2.12857947e-01 9.77542773e-02 3.18471253e-01 -1.64070010e-01
-3.77807558e-01 -1.04289496e+00 5.73740244e-01 1.02556777e+00
1.25960064e+00 -4.95435923e-01 3.90665650e-01 1.80196818e-02
-6.97360039e-01 2.41178319e-01 -1.78616345e-01 -7.53526855e-03
-8.56197625e-03 -2.45115608e-01 -9.12303209e-01 5.54178394e-02
9.19946909e-01 5.82856596e-01 -1.19822681e+00 1.58868170e+00
-4.21929568e-01 2.60302816e-02 -5.84263921e-01 -1.63019150e-02
-3.70718747e-01 2.25330308e-01 6.00015461e-01 6.23850822e-01
1.22385725e-01 -2.10080311e-01 -7.71420896e-01 1.01574147e+00
1.92726091e-01 7.76882768e-02 -4.57172632e-01 6.16878629e-01
3.26735765e-01 1.29607034e+00 -5.14629662e-01 3.66190106e-01
-4.76113498e-01 1.09296870e+00 5.55877209e-01 6.46226108e-01
-8.77977967e-01 -4.12743479e-01 8.46070588e-01 4.10477728e-01
-8.58322605e-02 -9.37097594e-02 -4.88858879e-01 -1.11553848e+00
1.30903081e-03 -3.19289297e-01 2.32954845e-01 -1.30706549e+00
-1.01771975e+00 7.85358131e-01 1.19654667e-02 -1.37701762e+00
-2.50623256e-01 -7.12069571e-01 -4.66745049e-01 1.47814333e+00
-1.69069481e+00 -1.46818674e+00 -9.15164471e-01 1.03452671e+00
-1.54295996e-01 -1.92527086e-01 8.61135840e-01 -9.14126188e-02
-8.16565990e-01 1.34045506e+00 -6.59248710e-01 -5.01535013e-02
1.22621238e+00 -1.35715652e+00 4.00585681e-01 7.01681137e-01
2.97879577e-02 9.61307347e-01 5.41288078e-01 -2.61530459e-01
-8.63465965e-01 -3.58435541e-01 1.09897578e+00 -1.52951777e+00
1.78628266e-01 -5.40238976e-01 -4.68376905e-01 7.41266131e-01
3.00849825e-01 2.71094032e-02 8.54425073e-01 7.52462566e-01
-5.21786511e-01 -8.96588117e-02 -1.33533478e+00 8.48857522e-01
1.31145835e+00 -4.46412057e-01 -5.20160139e-01 5.07326461e-02
2.83739001e-01 -7.37576306e-01 -7.87577510e-01 1.38749443e-02
7.89489567e-01 -1.30298936e+00 9.13069963e-01 -5.40633261e-01
-2.22316682e-01 -4.31209475e-01 5.09688020e-01 -1.10534012e+00
-2.52589613e-01 -7.92608261e-01 -3.41701150e-01 9.29854870e-01
4.60375458e-01 -7.86785424e-01 8.80026400e-01 1.23211420e+00
3.46462131e-01 -4.65682447e-01 -7.18480885e-01 -2.16802806e-01
-1.42091587e-01 -1.59503698e-01 8.80592048e-01 5.76098621e-01
3.93946171e-01 3.05619746e-01 -2.54277110e-01 1.28698796e-01
7.59378791e-01 -1.49604216e-01 1.28558040e+00 -1.35519600e+00
4.42707628e-01 -5.63968122e-01 -8.27950656e-01 -9.68645871e-01
1.15768649e-01 -3.32139999e-01 -3.76428485e-01 -1.05642486e+00
1.38435125e-01 -1.60426453e-01 -2.54342407e-01 5.40705442e-01
-2.48999029e-01 8.00690770e-01 -4.77111986e-04 2.73952633e-01
-5.61638534e-01 1.67122453e-01 1.45752394e+00 2.48564154e-01
-3.75441879e-01 1.42640740e-01 -9.14760232e-01 6.43321991e-01
4.87389356e-01 -6.04027063e-02 -4.71326530e-01 -4.71168458e-01
3.83126110e-01 -5.00795841e-01 6.72600091e-01 -9.81127441e-01
5.17578363e-01 2.15393394e-01 8.12383771e-01 -6.40038431e-01
4.57864493e-01 -7.94913948e-01 -3.52631509e-01 -3.46564092e-02
-6.11891188e-02 -9.34615806e-02 3.89282227e-01 3.51718068e-01
1.26748353e-01 1.12539984e-01 6.40391350e-01 2.39573300e-01
-6.83326066e-01 5.05557775e-01 2.23268524e-01 -1.49680749e-01
1.10545993e+00 -8.50981057e-01 -4.75014687e-01 -3.63795489e-01
-7.49213994e-01 2.67964780e-01 9.49097931e-01 4.50693667e-01
3.69529486e-01 -9.92905498e-01 -5.15090942e-01 7.35459089e-01
5.12946308e-01 -3.30057681e-01 2.22574145e-01 1.48355436e+00
-5.34360111e-01 7.47125566e-01 -5.15395105e-01 -8.76099944e-01
-1.52732229e+00 6.28622651e-01 8.92900229e-01 6.51196420e-01
-3.66472632e-01 1.13609684e+00 4.32662725e-01 -7.11631849e-02
2.97733933e-01 -2.68345743e-01 -5.75506806e-01 9.43797268e-03
9.79255497e-01 3.75331163e-01 3.88818607e-02 -1.04846799e+00
-5.75289130e-01 1.05222392e+00 -5.31011298e-02 4.47602291e-03
6.67191625e-01 -8.09977889e-01 1.91639900e-01 2.10466594e-01
7.68493772e-01 3.28704029e-01 -1.35983741e+00 -1.50715085e-02
-3.26247901e-01 -8.31536412e-01 2.60879219e-01 -1.10749352e+00
-9.36580062e-01 9.74769533e-01 1.03788602e+00 -1.19812615e-01
1.29791915e+00 -2.25965567e-02 2.74809569e-01 -3.78674216e-04
3.76153499e-01 -6.30672693e-01 -6.67097747e-01 -4.11308790e-03
8.50158751e-01 -1.63106024e+00 9.53991562e-02 -4.78805780e-01
-6.68553829e-01 8.33637357e-01 9.85753417e-01 1.50560036e-01
9.38725770e-01 -2.79530227e-01 3.69688153e-01 -5.47926426e-01
-2.48583615e-01 -6.03211641e-01 1.09766638e+00 1.10541475e+00
4.30598825e-01 -3.24699342e-01 -3.76998298e-02 2.81214833e-01
-4.48444515e-01 1.65669963e-01 7.18281716e-02 4.21812117e-01
2.30993126e-02 -7.92850852e-01 -4.31492150e-01 2.50688612e-01
-3.02992314e-01 -9.41672027e-02 -5.48173487e-01 1.01215613e+00
-1.05354466e-01 6.80455565e-01 4.39147055e-01 -2.72625089e-01
3.60767543e-01 2.12393895e-01 1.11287963e+00 -4.57560867e-01
-2.83663273e-01 -2.07647588e-02 -1.61999688e-01 -8.93103302e-01
-6.97180748e-01 -6.34791672e-01 -8.43410790e-01 -5.46712041e-01
-4.55524117e-01 -3.48406821e-01 6.25947893e-01 6.64448023e-01
6.37009025e-01 5.87676093e-02 2.49963537e-01 -1.14752734e+00
1.12129897e-01 -1.10612178e+00 -6.33480430e-01 2.75584728e-01
5.60540617e-01 -1.20142531e+00 -3.10659975e-01 -2.58180127e-02] | [14.112130165100098, 0.09601867198944092] |
6d5d8005-2324-4c9f-bab2-a76f0a3fd30b | incremental-outlier-detection-modelling-using | 2305.09907 | null | https://arxiv.org/abs/2305.09907v1 | https://arxiv.org/pdf/2305.09907v1.pdf | Incremental Outlier Detection Modelling Using Streaming Analytics in Finance & Health Care | In this paper, we had built the online model which are built incrementally by using online outlier detection algorithms under the streaming environment. We identified that there is highly necessity to have the streaming models to tackle the streaming data. The objective of this project is to study and analyze the importance of streaming models which is applicable in the real-world environment. In this work, we built various Outlier Detection (OD) algorithms viz., One class Support Vector Machine (OC-SVM), Isolation Forest Adaptive Sliding window approach (IForest ASD), Exact Storm, Angle based outlier detection (ABOD), Local outlier factor (LOF), KitNet, KNN ASD methods. The effectiveness and validity of the above-built models on various finance problems such as credit card fraud detection, churn prediction, ethereum fraud prediction. Further, we also analyzed the performance of the models on the health care prediction problems such as heart stroke prediction, diabetes prediction and heart stroke prediction problems. As per the results and dataset it shows that it performs well for the highly imbalanced datasets that means there is a majority of negative class and minority will be the positive class. Among all the models, the ensemble model strategy IForest ASD model performed better in most of the cases standing in the top 3 models in almost all of the cases. | ['Vivek', 'Ch Priyanka'] | 2023-05-17 | null | null | null | null | ['diabetes-prediction', 'outlier-detection', 'fraud-detection'] | ['medical', 'methodology', 'miscellaneous'] | [-3.95367920e-01 -2.29189694e-01 1.33898601e-01 -2.74893522e-01
2.55239218e-01 1.04252296e-03 1.56222492e-01 8.59039962e-01
-2.61480480e-01 6.64963722e-01 2.71864593e-01 -4.09048110e-01
-6.41026616e-01 -7.26236224e-01 -4.02889609e-01 -2.49851078e-01
-6.29455268e-01 8.35005879e-01 4.95174944e-01 -3.13792795e-01
7.26754785e-01 5.62694967e-01 -1.80837286e+00 7.53362715e-01
1.05657756e+00 9.99683499e-01 -6.18214309e-01 8.35793197e-01
-2.83641219e-01 1.16484821e+00 -4.17386502e-01 -1.94689840e-01
6.45316064e-01 -2.16794968e-01 -3.55397671e-01 -6.13822043e-01
-1.40446261e-01 -4.08794582e-01 1.09785467e-01 4.15619582e-01
6.83997512e-01 -8.11662972e-02 6.22392535e-01 -1.91118479e+00
3.06250572e-01 3.11161995e-01 -6.11886263e-01 9.59648132e-01
3.20520610e-01 -1.43146619e-01 4.01254803e-01 -8.02187264e-01
6.14514112e-01 8.93567502e-01 1.22479057e+00 -1.25033706e-01
-4.98674929e-01 -5.81597447e-01 -1.81312606e-01 7.06438363e-01
-8.50643873e-01 2.73211347e-03 3.77078265e-01 -4.33487743e-01
1.23967600e+00 2.79782891e-01 1.02919614e+00 5.81577897e-01
7.52026737e-01 5.06267846e-01 9.03176665e-01 -4.33289468e-01
4.14693266e-01 8.52407366e-02 5.47580421e-01 1.40013739e-01
8.37693632e-01 3.15871119e-01 -8.55643272e-01 -6.95373654e-01
1.80865467e-01 8.18216980e-01 3.47600728e-01 6.27864227e-02
-8.40878248e-01 8.89802396e-01 -1.47367254e-01 1.99444175e-01
-7.18144536e-01 -4.40098405e-01 9.80011761e-01 7.46645570e-01
5.13217866e-01 -5.10877594e-02 -8.14952552e-01 -4.19420600e-01
-1.20203221e+00 6.18615448e-01 9.94647741e-01 7.00836778e-01
4.89661507e-02 1.63297683e-01 1.87642828e-01 2.98120320e-01
3.18011999e-01 9.97705758e-02 1.04586935e+00 -4.29629445e-01
5.87553442e-01 1.03682423e+00 2.63393402e-01 -1.12085962e+00
-6.13624692e-01 -3.75817448e-01 -7.60456622e-01 4.56324786e-01
3.28090787e-01 -1.69826910e-01 -8.52207124e-01 8.76711905e-01
4.76134568e-01 5.88801801e-01 3.23837131e-01 2.86801249e-01
5.77448130e-01 3.96516293e-01 4.73068319e-02 -3.38491291e-01
9.51804459e-01 -7.05111027e-01 -8.01179588e-01 3.76868486e-01
8.73345077e-01 -8.76451015e-01 3.73424977e-01 1.01587939e+00
-6.64868295e-01 -1.97088778e-01 -8.98847103e-01 6.76474333e-01
-6.30891442e-01 -3.44898224e-01 7.01262534e-01 7.62549698e-01
-6.11071229e-01 8.87192965e-01 -1.04162192e+00 -7.71222651e-01
4.20711547e-01 3.28473240e-01 -1.62059978e-01 -2.08947256e-01
-9.57787395e-01 9.14950430e-01 5.05377531e-01 -2.33177528e-01
-4.73736554e-01 -7.06196368e-01 -3.18447202e-01 -1.39843509e-01
-4.06305075e-01 -5.36903381e-01 6.09293461e-01 -1.14542437e+00
-7.88275182e-01 4.71646011e-01 -1.02695242e-01 -1.14025748e+00
1.02465701e+00 -3.08446825e-01 -7.80088544e-01 9.26898941e-02
6.02850206e-02 -3.35226953e-01 4.24140692e-01 -3.98168564e-01
-8.84733975e-01 -8.30047131e-01 -6.89007461e-01 1.25891462e-01
4.26409058e-02 3.88529330e-01 9.81039822e-01 -6.99836016e-01
4.53807265e-01 -5.76181293e-01 -2.20332801e-01 -5.01042604e-01
6.02684505e-02 -1.93326727e-01 1.26950049e+00 -8.96841168e-01
1.56595802e+00 -2.06990457e+00 -6.33441687e-01 6.19393289e-01
-3.24439764e-01 2.51495779e-01 7.49031246e-01 1.03675520e+00
-3.97588342e-01 9.80298445e-02 -4.90379073e-02 3.55737239e-01
-4.36047971e-01 4.52163190e-01 -3.23662937e-01 5.18012762e-01
-1.50305286e-01 1.79766461e-01 -4.61139381e-01 -5.25706947e-01
1.34011880e-01 -2.31177837e-01 -6.59020483e-01 2.21776754e-01
2.18483418e-01 4.06254716e-02 -1.91059098e-01 1.08216977e+00
8.24625373e-01 2.68984020e-01 -3.49092782e-01 2.38725394e-01
-7.79274106e-02 -5.66753626e-01 -1.93088830e+00 8.40410233e-01
3.67270499e-01 2.24703297e-01 -5.90206802e-01 -1.31649768e+00
1.15714717e+00 5.73053956e-01 8.37454259e-01 -3.88403237e-01
-2.95725018e-01 8.11939240e-01 1.59271777e-01 -1.12175405e+00
1.76395983e-01 -2.79532731e-01 5.00946939e-01 3.56540054e-01
-1.85288012e-01 7.01325476e-01 1.05226964e-01 2.04507649e-01
1.29522681e+00 2.86750682e-02 7.59040833e-01 -1.81155056e-01
3.97311538e-01 4.24492091e-01 8.23703945e-01 6.57758951e-01
-7.80902088e-01 6.48868680e-01 7.59777129e-01 -1.00262213e+00
-1.09315836e+00 -9.20097768e-01 -2.59330064e-01 6.67619467e-01
-2.20159233e-01 -7.73539841e-02 -2.01980993e-01 -4.89985824e-01
5.23024142e-01 6.10063493e-01 -3.16335231e-01 -5.64710423e-02
-5.21829128e-01 -1.27397215e+00 5.39690554e-01 3.13830495e-01
4.81616557e-01 -1.06757760e+00 -1.01464844e+00 4.54314798e-01
2.36447707e-01 -5.54797292e-01 3.96870971e-01 2.06172988e-01
-1.59035695e+00 -1.33015966e+00 -2.25811198e-01 -3.34572554e-01
2.48465672e-01 -2.60752648e-01 9.75725949e-01 1.54857561e-01
-6.40401602e-01 -1.14339724e-01 -8.11461270e-01 -1.26959622e+00
-1.38795704e-01 -3.39502960e-01 2.55685419e-01 1.89967323e-02
8.67696345e-01 -8.21774900e-01 -7.74956822e-01 -2.35097948e-02
-8.97009313e-01 -6.17374659e-01 2.91228741e-01 6.24339044e-01
2.37506196e-01 3.20025533e-01 1.33921885e+00 -1.31842613e+00
5.30672908e-01 -1.26128495e+00 -5.34094870e-02 -1.95062324e-01
-1.06520534e+00 -4.28849727e-01 6.34591401e-01 -1.38075054e-01
-5.79817414e-01 -1.83491647e-01 2.61862259e-02 -1.69159472e-01
-3.14932734e-01 3.70625615e-01 2.78580546e-01 4.35135335e-01
9.03616726e-01 -1.01438969e-01 2.13314503e-01 -4.19449240e-01
-5.40416300e-01 1.04950225e+00 -1.61320224e-01 -8.30791146e-02
2.93098003e-01 7.00365365e-01 1.13647208e-02 -7.92657077e-01
-2.09411740e-01 -1.02079284e+00 -4.95301753e-01 -1.76515236e-01
5.31057239e-01 -8.68021369e-01 -3.39052260e-01 8.81250322e-01
-7.32925475e-01 2.31546044e-01 -2.42174000e-01 6.74557030e-01
-3.17805260e-01 3.23032498e-01 -4.77184415e-01 -1.39560592e+00
-5.65047681e-01 -4.43405837e-01 3.06291491e-01 2.86949247e-01
-3.31820339e-01 -7.77653575e-01 2.61515677e-01 1.86548486e-01
5.39642453e-01 1.06559849e+00 8.76588404e-01 -1.68560612e+00
-1.95087448e-01 -5.78807473e-01 1.30538538e-01 3.10224056e-01
-1.64905563e-01 3.01331878e-01 -4.90702003e-01 -4.47009541e-02
1.00062661e-01 2.49899685e-01 4.71518636e-01 3.58505130e-01
7.93353498e-01 -3.13981503e-01 -1.85511202e-01 5.29968500e-01
1.86344326e+00 4.60860521e-01 8.54700327e-01 1.01338363e+00
8.25999528e-02 2.15292081e-01 9.62631285e-01 1.09559667e+00
8.54740664e-02 1.76612325e-02 5.74870527e-01 3.29350471e-01
5.85888267e-01 5.16586043e-02 5.27735949e-01 6.29925907e-01
-1.87321261e-01 2.44663339e-02 -1.15047002e+00 5.78468502e-01
-2.20187974e+00 -1.40298295e+00 -8.21049571e-01 2.19416523e+00
1.93332881e-01 3.32351565e-01 5.83144903e-01 7.95683861e-01
5.46606600e-01 -4.03855890e-01 -5.15065908e-01 -1.07492828e+00
-1.99454483e-02 3.03071320e-01 5.49805284e-01 -1.85762402e-02
-8.72629702e-01 2.12961048e-01 5.69692802e+00 9.87988636e-02
-7.31792152e-01 1.18772751e-02 5.76054215e-01 -1.50651902e-01
2.07455963e-01 1.68960914e-01 -6.08313203e-01 1.14878380e+00
1.31838274e+00 -1.96394995e-01 -8.72266665e-02 1.19744825e+00
7.17070103e-01 -3.87028039e-01 -6.48969233e-01 6.67014003e-01
3.55997384e-02 -8.86258960e-01 8.22179914e-02 -1.60009086e-01
6.28507614e-01 2.95687467e-01 -7.12923050e-01 2.60379076e-01
-1.00964397e-01 -9.61723268e-01 2.65448928e-01 1.01842260e+00
-1.27074257e-01 -1.06066549e+00 1.62952721e+00 6.36218548e-01
-7.07457483e-01 -7.00427115e-01 -3.67620319e-01 -5.10270119e-01
1.07337385e-01 9.61683989e-01 -6.63793147e-01 8.35866928e-01
1.28419256e+00 7.12838650e-01 -5.00263631e-01 1.69290555e+00
6.87537313e-01 7.86581635e-01 -8.78732383e-01 1.61318615e-01
-6.40741736e-02 -2.14750469e-01 6.40967250e-01 1.02343714e+00
6.73586249e-01 2.31655389e-01 -3.12196743e-02 -2.75713243e-02
6.67871654e-01 7.11568058e-01 -7.57187247e-01 5.54913044e-01
3.34962577e-01 6.63609326e-01 -6.50205374e-01 -5.07276952e-01
-3.01117063e-01 5.14468372e-01 -3.11791867e-01 1.41914710e-01
-7.07089245e-01 -6.98653638e-01 2.59261072e-01 6.50052965e-01
1.36439204e-01 8.34909678e-02 -8.32726657e-01 -1.05291760e+00
2.05468103e-01 -1.01021349e+00 1.07030427e+00 -4.61599767e-01
-1.36767757e+00 1.61891863e-01 2.04427782e-02 -1.56674504e+00
-7.73486421e-02 -5.64211011e-01 -1.23566902e+00 5.37173450e-01
-1.36690617e+00 -5.90901852e-01 -4.56087857e-01 8.18337381e-01
3.77259225e-01 -7.63452828e-01 4.38443750e-01 5.51837564e-01
-4.50986952e-01 2.54048109e-01 4.76520419e-01 -2.72798181e-01
7.16882408e-01 -1.12425828e+00 -3.67182672e-01 9.95868802e-01
-4.28622901e-01 2.68072575e-01 9.79628801e-01 -1.05188024e+00
-6.54566526e-01 -1.02124262e+00 9.79836404e-01 -4.06142503e-01
3.70349258e-01 4.21638340e-01 -9.68474865e-01 6.18311226e-01
-3.98254097e-02 2.55593628e-01 7.71668255e-01 -2.37824202e-01
2.37512380e-01 -3.61391157e-01 -1.85969245e+00 -1.92998722e-01
5.88376522e-01 4.99590904e-01 -7.88823485e-01 5.66081941e-01
1.41808093e-01 -2.47141391e-01 -1.01653528e+00 5.14411867e-01
6.66213393e-01 -1.60611069e+00 7.50342131e-01 -1.13825738e+00
3.22846979e-01 -1.45424426e-01 -1.77772731e-01 -7.07832158e-01
1.90425560e-01 -3.42951149e-01 -5.25263786e-01 1.07941878e+00
2.58634150e-01 -1.19325364e+00 1.02510893e+00 4.83230531e-01
9.11654085e-02 -1.05843318e+00 -1.16235805e+00 -6.18210137e-01
-2.06361949e-01 -1.92785814e-01 6.22853458e-01 1.11655402e+00
2.51537934e-03 -3.80862951e-01 -1.56997070e-01 1.01934224e-01
6.33598924e-01 -8.93755481e-02 6.33590400e-01 -1.63540423e+00
-1.56598225e-01 2.23240361e-01 -1.03263295e+00 3.58633190e-01
-7.87372112e-01 -5.81559241e-01 -1.02643061e+00 -1.48335850e+00
-5.97241260e-02 -4.87544566e-01 -4.40716743e-01 1.11196361e-01
-1.12554692e-01 -4.51610684e-01 -4.30218596e-03 4.81181651e-01
-1.78708583e-01 -1.93420369e-02 2.81393111e-01 6.28525734e-01
-4.19877082e-01 5.35048604e-01 -2.00006545e-01 8.35794687e-01
1.08618426e+00 -9.11851168e-01 -2.26210192e-01 2.32319430e-01
5.30080378e-01 5.61533868e-02 2.38084808e-01 -1.24293649e+00
2.23743200e-01 -3.30953956e-01 7.12859213e-01 -1.09846473e+00
-4.74900186e-01 -1.04551077e+00 2.70161867e-01 1.12319291e+00
2.94579327e-01 9.27412152e-01 -1.20607056e-01 8.62312555e-01
-3.49648803e-01 -4.82369512e-01 6.62928820e-01 -3.85999709e-01
-5.42428434e-01 1.89198181e-01 -1.99250162e-01 2.63548464e-01
1.54650009e+00 -8.28969836e-01 -8.19850340e-02 -9.85887349e-02
-9.90455270e-01 3.41676652e-01 1.92541316e-01 6.21926263e-02
6.25067472e-01 -1.15952456e+00 -9.26125884e-01 3.75171304e-01
6.77015632e-02 -9.72865336e-03 2.36631498e-01 1.14524651e+00
-1.32374871e+00 8.77176300e-02 -6.20859802e-01 -4.73360419e-01
-1.39164448e+00 4.29709733e-01 4.02904630e-01 -5.21046460e-01
-7.14871347e-01 3.00588012e-01 -1.00773048e+00 -2.69539237e-01
1.90984458e-01 -3.45500112e-01 -5.27361095e-01 2.78578132e-01
3.27837914e-01 1.22299731e+00 3.97260606e-01 -1.80421293e-01
-4.88339573e-01 1.22628480e-01 8.75857286e-03 1.90966263e-01
1.60388756e+00 2.45586857e-01 -3.15634847e-01 6.26922190e-01
7.27245212e-01 -2.13420287e-01 -5.27980626e-01 3.60599101e-01
4.69866812e-01 -4.07278240e-01 -5.54788470e-01 -7.58183479e-01
-5.69176972e-01 5.51648200e-01 1.13431406e+00 4.40217644e-01
1.07204604e+00 -8.44987810e-01 8.91157925e-01 1.26963839e-01
1.70535937e-01 -1.57533216e+00 -1.83027551e-01 3.16976011e-01
4.49690253e-01 -1.21247900e+00 3.18744659e-01 1.48325190e-01
-7.79674113e-01 1.35050583e+00 5.44237196e-01 -6.43524408e-01
1.28954935e+00 2.03604043e-01 -2.49662459e-01 -1.40349269e-01
-1.04675114e+00 4.17517483e-01 -5.86556673e-01 6.84281349e-01
2.45176941e-01 4.06726338e-02 -8.45612168e-01 8.11842024e-01
-2.29804948e-01 6.58409357e-01 8.15032303e-01 1.15708470e+00
-7.85494030e-01 -7.92656660e-01 -5.41220129e-01 1.21440125e+00
-8.91113698e-01 1.16186075e-01 4.37144898e-02 8.60580862e-01
5.80783427e-01 7.22869277e-01 -7.80783147e-02 -2.49278406e-03
5.68127930e-01 6.35140657e-01 -2.76261330e-01 -2.60974944e-01
-9.46254075e-01 -7.30329394e-01 2.19606832e-02 -4.34708506e-01
-1.84627831e-01 -9.49853182e-01 -1.24109757e+00 -3.47748816e-01
-1.74596772e-01 2.28565216e-01 6.94341958e-01 8.98846924e-01
4.85849321e-01 2.65292019e-01 5.49100280e-01 -1.45293539e-02
-7.56044745e-01 -1.11239767e+00 -7.99532473e-01 5.82280934e-01
1.64925084e-01 -4.51363713e-01 -7.61259079e-01 -5.43924421e-02] | [8.147663116455078, 4.919399738311768] |
36a8944c-369d-498e-a9de-8addad133f41 | an-embarrassingly-simple-baseline-for-extreme | 1912.08140 | null | https://arxiv.org/abs/1912.08140v2 | https://arxiv.org/pdf/1912.08140v2.pdf | On-the-fly Global Embeddings Using Random Projections for Extreme Multi-label Classification | The goal of eXtreme Multi-label Learning (XML) is to automatically annotate a given data point with the most relevant subset of labels from an extremely large vocabulary of labels (e.g., a million labels). Lately, many attempts have been made to address this problem that achieve reasonable performance on benchmark datasets. In this paper, rather than coming-up with an altogether new method, our objective is to present and validate a simple baseline for this task. Precisely, we investigate an on-the-fly global and structure preserving feature embedding technique using random projections whose learning phase is independent of training samples and label vocabulary. Further, we show how an ensemble of multiple such learners can be used to achieve further boost in prediction accuracy with only linear increase in training and prediction time. Experiments on three public XML benchmarks show that the proposed approach obtains competitive accuracy compared with many existing methods. Additionally, it also provides around 6572x speed-up ratio in terms of training time and around 14.7x reduction in model-size compared to the closest competitors on the largest publicly available dataset. | ['Yashaswi Verma'] | 2019-12-17 | null | null | null | null | ['extreme-multi-label-classification'] | ['methodology'] | [ 4.22600478e-01 2.06193570e-02 -2.61648864e-01 -7.94469059e-01
-1.39753377e+00 -6.98792636e-01 5.42380869e-01 4.93728131e-01
-5.71010947e-01 5.41574776e-01 1.88339036e-02 -1.75990149e-01
-1.30507797e-01 -5.66649675e-01 -6.23951435e-01 -8.04890573e-01
1.00751929e-01 6.35435462e-01 1.81445211e-01 1.22343130e-01
2.41629407e-01 1.20025709e-01 -1.67571878e+00 1.92715764e-01
2.74792403e-01 1.34768915e+00 -9.72578079e-02 2.77351171e-01
-3.74395192e-01 8.35959852e-01 -8.96121785e-02 -4.80377465e-01
3.80861580e-01 1.13941185e-01 -1.02844512e+00 -8.45516697e-02
8.35596263e-01 2.68533006e-02 1.90980464e-01 9.07681942e-01
6.63669646e-01 -2.31347363e-02 7.55349398e-01 -1.18923640e+00
-5.18228054e-01 7.54285216e-01 -7.05688059e-01 -2.66389668e-01
2.06335604e-01 -4.91204500e-01 1.46851969e+00 -1.10582125e+00
5.01820505e-01 9.80277479e-01 8.49255204e-01 6.38986886e-01
-1.33092344e+00 -8.49575460e-01 9.24469531e-02 -3.15806344e-02
-1.46518803e+00 -3.40873033e-01 7.46706724e-01 -1.45018324e-01
7.41689503e-01 4.55397904e-01 -3.67230996e-02 8.95166099e-01
-8.60043019e-02 6.33129537e-01 1.39105952e+00 -6.62474811e-01
1.82379499e-01 4.65062588e-01 5.73366821e-01 7.30567217e-01
-1.33031523e-02 -2.24619821e-01 -5.11864007e-01 -4.95644718e-01
-7.52233192e-02 -1.81057364e-01 5.22748344e-02 -5.97029388e-01
-1.24366391e+00 8.83921683e-01 1.54866412e-01 2.10279599e-01
-1.03380240e-01 2.98767567e-01 7.00886369e-01 3.51103395e-01
6.06276631e-01 3.37537289e-01 -8.74498248e-01 -1.02227991e-02
-7.75833249e-01 4.16479446e-03 7.55550325e-01 1.00702572e+00
8.57562065e-01 -4.10171747e-01 1.63986966e-01 1.05301225e+00
3.88402790e-01 3.02627534e-01 5.32787681e-01 -9.71566200e-01
5.12332261e-01 6.58358335e-01 -1.24948487e-01 -6.26237810e-01
-5.71880996e-01 -2.69665122e-01 -7.13509023e-01 4.56287153e-02
2.94824302e-01 -2.14951038e-02 -4.69719738e-01 1.76064324e+00
8.19357514e-01 3.93336564e-01 4.08889018e-02 3.51983309e-01
5.95378280e-01 6.77658498e-01 2.74227560e-01 -2.67969728e-01
1.50182915e+00 -1.23154247e+00 -4.14165944e-01 -1.42104747e-02
1.20878911e+00 -7.78158009e-01 1.26514614e+00 3.88078064e-01
-7.35592365e-01 -5.72854638e-01 -1.01265860e+00 -1.05177045e-01
-5.00157773e-01 4.41887416e-02 7.35704303e-01 7.61787832e-01
-9.43170607e-01 5.00449240e-01 -5.18560886e-01 -4.30285364e-01
2.88412273e-01 6.54984355e-01 -5.43166339e-01 -4.22416627e-02
-9.37339902e-01 6.49345756e-01 5.81757486e-01 -4.91723567e-01
-5.77782691e-01 -9.32015300e-01 -6.33156359e-01 -1.16564119e-02
5.24452567e-01 -2.28398368e-01 1.15927684e+00 -6.31320059e-01
-1.33963704e+00 9.28875029e-01 -4.73810025e-02 -3.11492771e-01
1.82667539e-01 -2.64113754e-01 -3.87654692e-01 -8.61103609e-02
2.96688769e-02 9.16020691e-01 5.38884163e-01 -1.28377140e+00
-1.01803565e+00 -3.65274191e-01 5.75921237e-02 1.19276561e-01
-1.17286170e+00 2.65356630e-01 -3.03202242e-01 -4.33026761e-01
2.24414751e-01 -1.28711438e+00 -1.01269685e-01 6.44310117e-02
-1.93650201e-01 -7.76654303e-01 8.09279382e-01 -2.99094588e-01
1.28535402e+00 -2.14878201e+00 -1.46548310e-02 4.23907414e-02
2.09545553e-01 6.89065754e-02 -1.22084528e-01 5.28507113e-01
-1.10071912e-01 1.50516465e-01 -1.71330109e-01 -5.92480302e-01
3.35053265e-01 1.13379024e-01 -3.32666636e-01 6.60751641e-01
-7.78134689e-02 8.68970990e-01 -8.09952319e-01 -9.02988791e-01
1.12098053e-01 4.76772785e-01 -4.21852589e-01 -4.19671610e-02
-2.37770572e-01 2.53606349e-01 -4.35279429e-01 6.52665138e-01
5.07296920e-01 -5.56475043e-01 3.91279042e-01 -1.17595717e-01
1.36531875e-01 8.63667130e-02 -1.39823413e+00 1.89837897e+00
-7.78346419e-01 1.57502487e-01 -4.34763193e-01 -1.00415456e+00
7.92424619e-01 4.72151488e-01 8.45995963e-01 -3.99005324e-01
1.11888111e-01 3.30622077e-01 -5.44732928e-01 -1.23098791e-01
1.74296454e-01 -1.42873988e-01 -4.20799285e-01 8.09684753e-01
1.80465251e-01 4.40750122e-01 1.37438059e-01 -8.57975855e-02
1.14855552e+00 1.34597287e-01 4.53435004e-01 -2.89238244e-01
7.43607402e-01 -2.99630612e-01 4.76117432e-01 3.71404558e-01
-3.32195982e-02 2.71198571e-01 1.62307605e-01 -7.77986705e-01
-9.47787106e-01 -7.70761371e-01 -3.78648728e-01 1.63605213e+00
-7.02664210e-03 -7.85620809e-01 -4.88334298e-01 -1.30757940e+00
3.48287001e-02 7.54857421e-01 -6.67607546e-01 9.20798555e-02
-7.65971780e-01 -8.07333350e-01 5.72574973e-01 5.03261626e-01
3.77851039e-01 -8.86599779e-01 -4.62344825e-01 3.12145650e-01
3.37896645e-02 -1.06894600e+00 -4.01330322e-01 4.57289815e-01
-7.84546256e-01 -8.49322557e-01 -4.55616415e-01 -1.06463993e+00
5.29893994e-01 5.78397177e-02 1.11441016e+00 -1.98644117e-01
-2.30383594e-02 2.24038199e-01 -4.53118354e-01 -2.15263069e-01
-4.43384200e-01 4.16054487e-01 1.27594292e-01 7.84531012e-02
5.81387818e-01 -6.58956885e-01 -3.05633724e-01 3.62291783e-01
-9.04104352e-01 -1.58052087e-01 5.51734567e-01 8.32579851e-01
7.93293893e-01 1.46660805e-02 7.14343488e-01 -1.33191109e+00
2.74725825e-01 -5.33961058e-01 -5.41531503e-01 6.47994995e-01
-1.36380792e+00 3.03903073e-01 8.67557704e-01 -4.61083144e-01
-8.09047103e-01 5.64319253e-01 -2.89109915e-01 -4.86373417e-02
-1.91700637e-01 3.26134622e-01 -4.21074182e-02 -2.61471510e-01
5.51420569e-01 2.71771461e-01 -3.04289728e-01 -7.98585951e-01
6.27308607e-01 1.01741338e+00 2.07754560e-02 -5.89408278e-01
6.67901635e-01 4.38432515e-01 2.20705315e-01 -3.11408013e-01
-1.25326252e+00 -8.56857181e-01 -8.06567967e-01 -4.09473814e-02
6.15068734e-01 -8.42095673e-01 -6.39502227e-01 1.90502584e-01
-6.33680105e-01 5.77838384e-02 -2.39886433e-01 2.62510598e-01
-6.48707449e-01 4.88886654e-01 -4.71328348e-01 -4.56488073e-01
-5.84451020e-01 -1.00776470e+00 1.31976378e+00 -1.62748367e-01
-8.09506848e-02 -9.94345725e-01 3.17260236e-01 3.29156041e-01
3.10875773e-01 1.26717478e-01 1.15365899e+00 -1.10424566e+00
-1.21511236e-01 -2.63498217e-01 -1.10529751e-01 4.73558813e-01
2.20298350e-01 -3.35377991e-01 -1.15934479e+00 -6.11059725e-01
-3.59631032e-02 -7.57952392e-01 6.78437233e-01 -1.08251318e-01
1.15858364e+00 -9.63937566e-02 -5.23172498e-01 4.12019879e-01
1.90092981e+00 -1.06877536e-01 1.83901098e-02 4.23015028e-01
6.90216601e-01 6.50138855e-01 8.36072862e-01 4.44639027e-01
4.93876845e-01 1.04371989e+00 2.58371174e-01 1.26994222e-01
-2.44002387e-01 -1.95221364e-01 9.42490175e-02 1.16868258e+00
4.08805370e-01 -7.74559751e-02 -1.08604705e+00 4.00248289e-01
-1.73457348e+00 -5.24435461e-01 4.29930240e-02 2.27353191e+00
9.50167298e-01 1.27469763e-01 8.60416591e-02 3.47173601e-01
5.65924883e-01 2.07932368e-01 -5.30206263e-01 -1.89790338e-01
1.03436589e-01 1.59945935e-01 6.69411182e-01 3.04735333e-01
-1.48843956e+00 8.68860543e-01 5.87734938e+00 9.91969526e-01
-1.06247640e+00 4.44092572e-01 5.61906278e-01 -1.78979561e-01
-8.13590139e-02 -6.32295059e-03 -1.30791128e+00 4.13207561e-01
1.50094783e+00 -2.17171088e-02 1.71519563e-01 1.03856790e+00
-5.33336937e-01 3.71593535e-01 -1.35149539e+00 9.43136036e-01
2.35965073e-01 -1.15015936e+00 -1.00162655e-01 1.95431724e-01
8.50174725e-01 6.47946522e-02 1.77908123e-01 5.47395527e-01
3.86824042e-01 -7.55585909e-01 5.49373925e-01 2.26647809e-01
9.90734100e-01 -8.60800564e-01 7.48043895e-01 5.14840066e-01
-1.38921189e+00 -3.43784273e-01 -5.17736971e-01 1.77670449e-01
1.07037246e-01 4.97327209e-01 -6.83591366e-01 5.67553341e-01
6.32993639e-01 4.63703215e-01 -7.83419549e-01 7.48678386e-01
5.04375100e-02 7.67725706e-01 -5.54328918e-01 -1.37391448e-01
3.73560667e-01 7.85197169e-02 -1.36191368e-01 1.11009836e+00
2.38273010e-01 -2.10489169e-01 2.68835366e-01 -7.70625174e-02
-4.49692994e-01 7.80156374e-01 -6.10822797e-01 2.58583546e-01
5.51230907e-01 1.51667643e+00 -6.21360242e-01 -3.67401153e-01
-8.66078734e-01 8.76036823e-01 8.01139355e-01 -1.89074367e-01
-1.06207192e+00 -1.24777898e-01 3.05408657e-01 -1.67272374e-01
4.02594000e-01 1.68301329e-01 -1.30830809e-01 -7.97670543e-01
2.32947543e-01 -8.47712874e-01 4.91391659e-01 -9.79316160e-02
-1.31052911e+00 5.36838293e-01 -1.49086565e-01 -1.32941103e+00
-1.37413755e-01 -4.54057515e-01 4.10456173e-02 4.70754176e-01
-1.56826091e+00 -1.50739706e+00 -1.33585688e-02 2.37418413e-01
6.21450603e-01 -1.39987811e-01 1.46879470e+00 5.69207132e-01
-3.12460721e-01 9.32982922e-01 7.26312816e-01 -1.30442753e-01
1.01358616e+00 -1.47803330e+00 3.29857230e-01 2.49342576e-01
8.38976502e-01 3.06045264e-01 3.66326511e-01 -1.53437629e-01
-1.23347998e+00 -1.35445201e+00 1.25177073e+00 -7.17563212e-01
6.12725496e-01 -5.85826695e-01 -6.29901707e-01 8.19310486e-01
1.13523833e-01 4.36374038e-01 1.21325290e+00 3.51315379e-01
-6.47649586e-01 -4.80736822e-01 -1.13168693e+00 2.24100888e-01
9.28224742e-01 -3.25759917e-01 -3.94906998e-01 6.19213641e-01
7.93567121e-01 -7.75116682e-02 -1.30463159e+00 5.42271435e-01
7.43395329e-01 -6.40167177e-01 9.70966458e-01 -6.38141453e-01
2.62833327e-01 -1.74129188e-01 -5.47868788e-01 -8.90672803e-01
-3.64821196e-01 -2.72015870e-01 -2.49951065e-01 1.62091768e+00
5.62736034e-01 -4.75167960e-01 9.35310900e-01 4.83451128e-01
1.47427440e-01 -1.24695158e+00 -1.01204503e+00 -8.62967193e-01
3.63941133e-01 -3.38142425e-01 7.01485634e-01 1.07193184e+00
-2.53261596e-01 5.67931175e-01 -6.05781138e-01 1.51746854e-01
8.18021595e-01 3.39459330e-01 6.15892231e-01 -1.48368001e+00
-1.05256028e-01 9.12291929e-02 -3.42263967e-01 -1.02610970e+00
5.03693759e-01 -1.28898430e+00 -2.40922615e-01 -1.30446088e+00
3.67321700e-01 -9.12342966e-01 -9.74427819e-01 6.35540903e-01
-1.01137459e-01 4.55720723e-01 7.42769167e-02 2.27575719e-01
-1.03361905e+00 3.51093978e-01 6.02789700e-01 -1.33249134e-01
2.76721299e-01 1.97831495e-03 -7.72635281e-01 6.22662723e-01
8.05515110e-01 -1.01209974e+00 -5.42209506e-01 -4.05993134e-01
3.01703751e-01 -2.40128607e-01 -2.05313042e-01 -1.18428838e+00
3.06349635e-01 1.54722929e-01 -2.13380363e-02 -4.66656834e-01
3.21141183e-01 -1.17418253e+00 2.11209476e-01 1.83216751e-01
-7.47278094e-01 7.46654943e-02 -7.45189935e-02 7.33430922e-01
-1.18358925e-01 -5.86570203e-01 7.33710408e-01 7.04369470e-02
-8.20984721e-01 2.19470590e-01 1.86009660e-01 -3.98117341e-02
1.51289463e+00 1.30368099e-01 -7.41558224e-02 1.76372454e-01
-5.21681428e-01 7.82192424e-02 4.40323651e-01 3.80607307e-01
2.34446138e-01 -1.61882603e+00 -6.77654862e-01 -5.12557812e-02
5.16075015e-01 -3.13844770e-01 9.82531160e-03 5.02240956e-01
-2.32385606e-01 6.67991638e-01 1.45587221e-01 -4.78616774e-01
-1.52041459e+00 9.64335203e-01 -1.82438955e-01 -7.44554281e-01
-5.71672320e-01 1.06604576e+00 -9.22315419e-02 -9.18410361e-01
4.83225971e-01 -2.36068293e-02 -2.36665830e-01 2.72503853e-01
5.43282807e-01 3.96573663e-01 1.64237455e-01 -7.31256962e-01
-4.54027653e-01 9.85207856e-01 -2.68149585e-01 1.49579391e-01
1.54202247e+00 -1.10446811e-01 -4.53900034e-03 8.08082879e-01
1.78191113e+00 7.47831836e-02 -1.05648541e+00 -5.92168748e-01
4.79086399e-01 -2.80523330e-01 2.65976507e-02 -7.57906079e-01
-9.88479257e-01 7.51331925e-01 1.05812109e+00 2.60723591e-01
1.10465825e+00 1.22909158e-01 8.93948913e-01 5.83001018e-01
7.20769525e-01 -1.01552892e+00 -4.08776812e-02 9.58655868e-03
3.98594290e-01 -1.54425085e+00 1.14788651e-01 -3.34930271e-01
-4.84390169e-01 9.96184826e-01 3.48284066e-01 5.03525920e-02
8.65103722e-01 2.08405167e-01 2.40842164e-01 -1.76129386e-01
-1.05968976e+00 1.14415877e-01 1.85465276e-01 1.72922179e-01
4.53940570e-01 -3.18455994e-02 -4.79280233e-01 2.34959379e-01
7.70941675e-02 -1.82697326e-01 -1.96798146e-02 9.26871717e-01
-5.33679903e-01 -1.67777431e+00 4.70205732e-02 7.27176130e-01
-7.68550813e-01 -1.70140117e-01 4.78024743e-02 7.53947616e-01
1.67746872e-01 8.20921659e-01 -3.08776081e-01 -4.72422719e-01
2.63666064e-01 6.60242319e-01 1.85511693e-01 -6.97952211e-01
-5.76437533e-01 -1.86977312e-01 7.28525445e-02 -5.23540795e-01
-5.43497860e-01 -7.76520908e-01 -1.14239419e+00 -6.33633807e-02
-6.22777998e-01 1.90831959e-01 7.83594847e-01 6.74494863e-01
3.41825545e-01 1.61987275e-01 8.90605032e-01 -2.33305335e-01
-1.07271504e+00 -1.00625718e+00 -6.04174018e-01 4.42351252e-01
-8.72556195e-02 -6.51436985e-01 -3.59553665e-01 1.26492772e-02] | [9.528292655944824, 4.381275653839111] |
796b3c8b-80d7-4173-8e3b-12cbf5e092a6 | egocentric-activity-recognition-with | 1601.06603 | null | http://arxiv.org/abs/1601.06603v1 | http://arxiv.org/pdf/1601.06603v1.pdf | Egocentric Activity Recognition with Multimodal Fisher Vector | With the increasing availability of wearable devices, research on egocentric
activity recognition has received much attention recently. In this paper, we
build a Multimodal Egocentric Activity dataset which includes egocentric videos
and sensor data of 20 fine-grained and diverse activity categories. We present
a novel strategy to extract temporal trajectory-like features from sensor data.
We propose to apply the Fisher Kernel framework to fuse video and temporal
enhanced sensor features. Experiment results show that with careful design of
feature extraction and fusion algorithm, sensor data can enhance
information-rich video data. We make publicly available the Multimodal
Egocentric Activity dataset to facilitate future research. | ['Vijay Chandrasekhar', 'Ngai-Man Cheung', 'Sibo Song', 'Jie Lin', 'Bappaditya Mandal'] | 2016-01-25 | null | null | null | null | ['egocentric-activity-recognition'] | ['computer-vision'] | [ 9.16957259e-02 -6.11359000e-01 -4.05910760e-01 -5.28095603e-01
-5.38630247e-01 -4.68702406e-01 6.28363788e-01 -1.64406434e-01
-4.46026355e-01 6.25157773e-01 9.70032752e-01 3.00210208e-01
-3.98791611e-01 -4.99464244e-01 -3.90757978e-01 -6.05059147e-01
-5.60874701e-01 -5.55375218e-01 7.60078356e-02 2.35352620e-01
2.91230381e-01 2.87141919e-01 -1.58548295e+00 1.82413727e-01
6.61274910e-01 1.11025500e+00 -7.67569337e-03 9.18293715e-01
5.07016599e-01 9.26015615e-01 -1.36330232e-01 -1.67190760e-01
2.31791269e-02 -4.56574053e-01 -5.15801191e-01 3.01619202e-01
2.00508788e-01 -5.99676728e-01 -8.96901608e-01 6.66337013e-01
6.66752219e-01 5.20814180e-01 6.08853161e-01 -1.42459130e+00
-3.38917285e-01 3.42622608e-01 -4.05468732e-01 7.30625629e-01
1.03544426e+00 -1.62051693e-02 5.83210826e-01 -7.83836305e-01
3.94688159e-01 7.51322508e-01 4.74086583e-01 2.10451692e-01
-7.42809594e-01 -4.30719614e-01 1.95538461e-01 6.12909198e-01
-1.47879434e+00 -6.93318665e-01 9.38595593e-01 -5.35387695e-01
9.52176988e-01 2.75884639e-04 1.04142606e+00 1.53653049e+00
1.78600013e-01 1.17005122e+00 6.67634249e-01 5.10498602e-03
3.06609988e-01 -3.71501535e-01 1.55856907e-02 4.77952808e-01
2.73545772e-01 -1.56587407e-01 -9.03754950e-01 -3.69423240e-01
8.62175703e-01 5.47601759e-01 -2.87129343e-01 -6.95997357e-01
-1.55912375e+00 5.73909044e-01 1.46618620e-01 3.20806712e-01
-7.73018420e-01 4.28061187e-01 5.75654447e-01 1.51137754e-01
4.01675522e-01 -7.65665472e-02 -1.82147041e-01 -1.06073368e+00
-5.96906006e-01 1.93466812e-01 5.76898932e-01 1.02208126e+00
6.94692135e-01 1.03258282e-01 -8.22059661e-02 6.16999805e-01
1.36327803e-01 5.94108880e-01 6.54371679e-01 -1.26884127e+00
4.95705754e-01 6.92770779e-01 8.60422403e-02 -1.12857640e+00
-4.32325393e-01 8.95737484e-02 -4.48407173e-01 -7.47512400e-01
1.21780500e-01 -5.18149912e-01 -2.58248627e-01 1.50197375e+00
4.02587414e-01 7.65100718e-01 3.01339328e-02 1.03775001e+00
5.84969580e-01 3.62367541e-01 3.78032953e-01 -2.32970208e-01
1.21430218e+00 -5.60343623e-01 -9.78721797e-01 2.33124435e-01
7.65755892e-01 -2.67667860e-01 6.58288717e-01 2.91344434e-01
-1.08278787e+00 -4.12488431e-01 -1.04580152e+00 2.37272710e-01
-2.56215096e-01 -1.46667371e-02 9.62868512e-01 7.46469617e-01
-7.07505703e-01 4.04431254e-01 -1.22286892e+00 -8.68872285e-01
3.98392677e-01 1.86138093e-01 -7.48552322e-01 -1.69610590e-01
-9.97921705e-01 5.00492930e-01 5.01471698e-01 -2.68245578e-01
-8.46302986e-01 -4.81515676e-01 -1.21949756e+00 -4.09422666e-02
3.63131315e-01 -7.41174340e-01 1.22566319e+00 -8.37884665e-01
-1.35213423e+00 2.19861940e-01 -3.26171815e-01 -2.60524482e-01
1.43538907e-01 -4.74532783e-01 -7.95576930e-01 5.73060811e-01
-1.02118455e-01 6.05234206e-02 9.58025634e-01 -5.36443412e-01
-8.23590696e-01 -6.09895527e-01 1.40350834e-01 4.86374021e-01
-7.75557935e-01 3.31131965e-02 -4.34357971e-01 -6.99606836e-01
-6.21691681e-02 -8.33223522e-01 6.42576292e-02 -4.09991771e-01
1.25619054e-01 -3.75038445e-01 9.35369134e-01 -3.95267695e-01
1.45840263e+00 -2.22055912e+00 2.77873129e-01 -1.53514650e-02
2.74138659e-01 -2.11268917e-01 3.78315523e-02 6.43921018e-01
3.39885093e-02 -3.76629025e-01 1.22740954e-01 -2.03850597e-01
-3.21348831e-02 1.49563909e-01 2.98752636e-02 9.02490020e-01
-4.32410054e-02 1.05401981e+00 -1.24198580e+00 -5.85728049e-01
5.53177655e-01 5.84509194e-01 -6.11695766e-01 4.28919673e-01
4.92294341e-01 4.45090353e-01 -9.83067989e-01 1.02592289e+00
2.83763558e-01 -1.73783183e-01 -2.90590972e-02 -3.91064495e-01
-6.94682747e-02 -8.04532170e-02 -1.00209105e+00 2.25351191e+00
-1.38102576e-01 6.80495799e-01 -2.76000202e-01 -9.96427000e-01
4.72743869e-01 4.40501183e-01 1.30187213e+00 -6.10109746e-01
3.31570238e-01 -3.79209369e-01 -4.47568148e-01 -9.71576512e-01
7.80266345e-01 3.47123027e-01 -5.02757728e-01 4.21621650e-01
4.35307473e-01 4.50272799e-01 2.63469905e-01 2.29149938e-01
1.39875102e+00 3.79951745e-01 5.64724207e-01 1.94426961e-02
6.64207101e-01 -2.28295878e-01 5.44729590e-01 5.87316573e-01
-7.76338279e-01 3.22738320e-01 -2.30525876e-03 -2.85873890e-01
-7.18112767e-01 -1.40787005e+00 7.96374008e-02 1.41598845e+00
2.15950295e-01 -8.96507204e-01 -6.15697086e-01 -6.09610856e-01
-2.42101029e-02 2.01879218e-01 -5.79958618e-01 -2.44363233e-01
-6.10261977e-01 -4.86704290e-01 6.15055561e-01 8.97575200e-01
6.51468754e-01 -6.10920250e-01 -5.57364404e-01 9.55625698e-02
-4.15405631e-01 -1.10539353e+00 -7.04319596e-01 -2.08916500e-01
-1.00474393e+00 -1.34056056e+00 -6.84347391e-01 -2.57497609e-01
4.10602838e-01 8.83624911e-01 7.17969477e-01 -7.06031144e-01
-7.71796927e-02 1.52368283e+00 -6.48694098e-01 -4.92834710e-02
6.67859495e-01 -5.55619895e-02 5.74772298e-01 3.58515084e-01
9.08965468e-01 -7.49998629e-01 -8.90220881e-01 3.99427831e-01
-8.12586188e-01 -2.93028176e-01 2.03407809e-01 3.60607564e-01
2.81816095e-01 -4.07120466e-01 4.98188466e-01 -1.05463333e-01
7.43507922e-01 -1.08083272e+00 -9.32961032e-02 -9.53723863e-03
-1.83190212e-01 -2.91034192e-01 1.74964264e-01 -5.84689200e-01
-1.11665916e+00 9.46632400e-02 2.96474338e-01 -6.33452833e-01
-2.50757307e-01 6.55640483e-01 -3.56943250e-01 -5.17071523e-02
4.08141345e-01 3.83823425e-01 -3.49583447e-01 -4.10043091e-01
4.00757968e-01 7.96641469e-01 6.34938240e-01 -5.92566967e-01
4.84917849e-01 7.68257916e-01 -3.17225814e-01 -1.18429446e+00
-3.45991522e-01 -8.09572399e-01 -5.61287105e-01 -6.45449340e-01
8.48625124e-01 -1.38747990e+00 -1.01606905e+00 4.31078434e-01
-6.03736401e-01 1.16490476e-01 -2.07397118e-01 1.17595851e+00
-1.12118137e+00 7.07987368e-01 -3.77504736e-01 -8.38707566e-01
2.17370335e-02 -7.13147402e-01 1.07887161e+00 4.60709453e-01
-4.19281989e-01 -9.99842465e-01 4.66074526e-01 3.79159063e-01
2.60189563e-01 2.82942325e-01 -2.03053012e-01 -2.73619741e-01
-6.19383931e-01 -4.28667486e-01 -6.12945408e-02 4.11278531e-02
5.42124689e-01 -3.52657586e-01 -8.12042892e-01 -3.37701827e-01
2.49023661e-02 -1.92450956e-01 6.66947246e-01 5.65556407e-01
1.16086042e+00 -3.37455086e-02 -3.75161856e-01 8.21428776e-01
9.11874354e-01 2.25109562e-01 6.65732145e-01 1.86969906e-01
7.81979620e-01 3.67799610e-01 7.47970283e-01 1.03082383e+00
7.35815406e-01 4.86512810e-01 1.96308956e-01 5.07369995e-01
4.53300655e-01 -3.77705514e-01 6.89292908e-01 8.10305893e-01
-5.56086719e-01 -1.92457825e-01 -4.06018317e-01 7.80212879e-01
-2.25876427e+00 -1.47580111e+00 2.82073200e-01 1.81859398e+00
7.14713782e-02 -4.90641564e-01 4.61088210e-01 1.19731277e-02
4.78091121e-01 4.78742838e-01 -6.36259615e-01 -6.50203228e-02
-5.75863644e-02 -4.49118674e-01 4.61520821e-01 2.42400845e-03
-1.61679280e+00 6.21634066e-01 7.17728472e+00 4.97598737e-01
-7.80271113e-01 1.48066491e-01 -2.50301272e-01 -4.84253436e-01
-1.05883367e-02 -3.22230160e-01 -4.29775000e-01 6.81346536e-01
1.28381586e+00 -5.75016737e-01 3.32237124e-01 1.11305225e+00
6.83423579e-01 -4.20290202e-01 -1.05348897e+00 1.70430255e+00
3.47993344e-01 -1.04723632e+00 -1.97579309e-01 1.60437226e-01
6.21859848e-01 2.25973636e-01 -3.13215494e-01 2.62286752e-01
1.34791091e-01 -6.10662699e-01 2.21921951e-01 1.00787508e+00
4.02823180e-01 -7.45301485e-01 4.50950682e-01 1.32626250e-01
-1.70916331e+00 -3.34027112e-01 -8.50092396e-02 -3.52343857e-01
4.40715700e-01 2.75261819e-01 -3.03590178e-01 6.60241187e-01
1.02674091e+00 1.65470576e+00 -4.62563217e-01 1.22665942e+00
1.98738605e-01 5.55887938e-01 -2.99939752e-01 2.05828585e-02
2.64906380e-02 -3.86141717e-01 6.33013070e-01 1.20234787e+00
6.25755489e-01 3.58753592e-01 1.89984381e-01 1.37055963e-01
9.73156840e-02 -3.10828462e-02 -1.10585129e+00 -5.62098622e-01
1.54689118e-01 1.19142306e+00 -1.31213680e-01 -3.77786487e-01
-9.14849102e-01 1.17613220e+00 1.63211927e-01 5.65385938e-01
-9.04859364e-01 -2.84717500e-01 1.32299149e+00 -1.56628296e-01
2.23202154e-01 -7.34654605e-01 3.14349204e-01 -1.92965603e+00
-5.46248779e-02 -2.96442807e-01 7.55918324e-01 -7.63171673e-01
-1.14751029e+00 9.03977361e-03 3.56491148e-01 -1.53825843e+00
-6.53791606e-01 -3.41345817e-01 -5.55235326e-01 2.78644234e-01
-1.03205872e+00 -1.07600021e+00 -7.11518526e-01 1.23802197e+00
5.10200739e-01 -7.43681490e-02 5.62767684e-01 6.30415678e-01
-5.44026792e-01 2.90834486e-01 1.18741684e-01 2.75162488e-01
7.16756761e-01 -8.79709721e-01 -1.02329031e-01 6.72685862e-01
-1.32216305e-01 8.49757254e-01 5.36572516e-01 -6.55530334e-01
-2.17434239e+00 -1.00808299e+00 6.32903516e-01 -6.53429687e-01
7.71011293e-01 -1.71594441e-01 -4.09542859e-01 1.01509166e+00
1.73825592e-01 -8.57139826e-02 1.25179827e+00 -7.50304013e-03
1.96570046e-02 -2.63370544e-01 -9.78061020e-01 6.15335703e-01
1.46317565e+00 -7.32151628e-01 -9.42967176e-01 -1.49722114e-01
2.68932283e-01 9.21251625e-02 -1.26168704e+00 5.69241464e-01
9.18689787e-01 -7.26576865e-01 8.79936337e-01 -6.23261154e-01
-9.75844264e-02 -4.01689589e-01 -5.64052403e-01 -1.20518589e+00
-3.73517960e-01 -6.34879649e-01 -8.66270483e-01 8.47750008e-01
-5.32640517e-01 -4.62046981e-01 8.63601327e-01 6.83390260e-01
-9.81600285e-02 -1.84648618e-01 -9.77415264e-01 -6.77769303e-01
-7.86920905e-01 -7.07502782e-01 5.20907700e-01 7.84648180e-01
7.97391057e-01 -1.06070921e-01 -7.83726573e-01 -1.57397464e-01
5.06644607e-01 -6.19580150e-02 1.14263356e+00 -9.25888062e-01
1.87893912e-01 2.81132571e-02 -9.76994395e-01 -1.34211564e+00
1.66971586e-03 -3.28658998e-01 -4.43320811e-01 -1.16123617e+00
3.95412147e-01 4.42483455e-01 -7.23104537e-01 9.42819566e-02
1.58557016e-02 4.49726045e-01 -1.30166441e-01 -5.08505255e-02
-1.15118170e+00 1.02472782e+00 9.89052713e-01 -9.04514119e-02
-2.06733361e-01 -1.10365979e-01 -5.65923691e-01 7.62806296e-01
7.22319782e-01 -8.47694930e-03 -7.42026746e-01 -2.41366789e-01
6.32162914e-02 5.61042055e-02 2.54163563e-01 -1.25058925e+00
2.01397240e-01 -3.58984560e-01 6.28971934e-01 -7.94630349e-01
7.93490708e-01 -1.00730121e+00 1.55208245e-01 1.34123012e-01
-1.46363415e-02 1.42159954e-01 -1.25041589e-01 1.02643228e+00
-3.42003018e-01 3.99945736e-01 9.65882689e-02 8.44698176e-02
-1.11028957e+00 5.65149784e-01 -8.99356186e-01 -2.95308292e-01
1.25046015e+00 -6.16701961e-01 -1.37832060e-01 -7.31184185e-01
-7.76620746e-01 3.70997488e-01 5.53515077e-01 8.29491019e-01
7.27397442e-01 -1.74063861e+00 -3.21134806e-01 2.29618549e-01
5.84939480e-01 -7.83994019e-01 7.04858303e-01 1.14877319e+00
1.33712376e-02 5.70998430e-01 -4.09216821e-01 -6.81469619e-01
-1.10820520e+00 4.69908088e-01 5.68304583e-02 4.06847715e-01
-5.23049831e-01 1.45091608e-01 5.78449778e-02 6.33555502e-02
1.25627294e-01 -2.25676298e-01 -4.73349571e-01 2.68079340e-01
9.48793948e-01 8.35412562e-01 -3.48249763e-01 -8.57125342e-01
-5.78272581e-01 5.28963864e-01 2.93822825e-01 4.58280407e-02
1.20320857e+00 -7.26700187e-01 4.26542789e-01 6.25917614e-01
1.29843688e+00 -1.40954718e-01 -1.41852868e+00 -1.20828368e-01
4.85563576e-02 -8.57463002e-01 1.54768705e-01 -5.52926622e-02
-8.21299791e-01 4.78574157e-01 7.73406267e-01 -6.28662901e-03
1.27076316e+00 4.89501945e-05 8.05493653e-01 4.49548155e-01
6.56853259e-01 -1.21907103e+00 2.26423994e-01 5.23180425e-01
4.41680640e-01 -1.27832663e+00 6.97526559e-02 -3.76310796e-02
-7.58685529e-01 8.02546203e-01 4.32243228e-01 -3.54148507e-01
7.66606688e-01 -5.56798391e-02 -2.90112793e-01 -1.43188536e-01
-6.02883220e-01 -4.71140981e-01 2.41809815e-01 8.73534560e-01
3.95180434e-01 5.20501845e-02 -4.60246921e-01 8.57543051e-01
2.83721566e-01 3.99000108e-01 3.37340564e-01 1.35525298e+00
-5.10520756e-01 -8.00933957e-01 -2.62542307e-01 4.16044265e-01
-3.19840878e-01 3.96506310e-01 -1.67415857e-01 5.41636705e-01
-2.58528322e-01 1.10677052e+00 1.54604837e-01 -5.99903703e-01
2.27989852e-01 1.46843776e-01 6.47978663e-01 -1.33874705e-02
-1.86686933e-01 3.38011086e-02 2.50135958e-01 -1.26849508e+00
-9.83924508e-01 -1.01488423e+00 -8.67681444e-01 -5.02533317e-01
1.47909954e-01 1.65206715e-01 6.98189199e-01 9.20471966e-01
6.78402483e-01 2.85740793e-01 7.36713052e-01 -1.26632202e+00
2.55700462e-02 -9.63161230e-01 -7.92355359e-01 4.70219105e-01
3.23422641e-01 -1.09332442e+00 -1.32992849e-01 2.49140903e-01] | [8.051386833190918, 0.5512800216674805] |
67462d97-df05-4f96-aafe-dab55ca9e702 | edione-lt-edi-eacl2021-pre-trained | null | null | https://aclanthology.org/2021.ltedi-1.11 | https://aclanthology.org/2021.ltedi-1.11.pdf | EDIOne@LT-EDI-EACL2021: Pre-trained Transformers with Convolutional Neural Networks for Hope Speech Detection. | Hope is an essential aspect of mental health stability and recovery in every individual in this fast-changing world. Any tools and methods developed for detection, analysis, and generation of hope speech will be beneficial. In this paper, we propose a model on hope-speech detection to automatically detect web content that may play a positive role in diffusing hostility on social media. We perform the experiments by taking advantage of pre-processing and transfer-learning models. We observed that the pre-trained multilingual-BERT model with convolution neural networks gave the best results. Our model ranked first, third, and fourth ranks on English, Malayalam-English, and Tamil-English code-mixed datasets. | ['Radhika Mamidi', 'Suman Dowlagar'] | null | null | null | null | eacl-ltedi-2021-4 | ['hope-speech-detection'] | ['natural-language-processing'] | [-2.94037998e-01 7.80612975e-02 -5.86615145e-01 -1.91394404e-01
-7.67457545e-01 1.81777418e-01 7.51600981e-01 5.41099012e-01
-4.87224430e-01 7.04301775e-01 9.96915340e-01 -4.28937554e-01
-3.43302190e-02 -8.80408287e-01 -9.18027088e-02 -1.18786164e-01
-3.68192822e-01 2.13669956e-01 -3.97172660e-01 -6.17494166e-01
4.16914612e-01 1.53543368e-01 -7.78349102e-01 6.46938503e-01
9.70505059e-01 3.48815858e-01 -1.15938105e-01 6.73516154e-01
-2.75174141e-01 1.18116879e+00 -3.50173533e-01 -6.87990367e-01
-4.55791742e-01 -6.90584183e-01 -9.70727205e-01 -3.06093395e-01
-2.41372824e-01 -4.31069911e-01 -3.53332043e-01 1.02335346e+00
7.30772793e-01 -2.95145243e-01 4.70511585e-01 -7.94522524e-01
-1.06193697e+00 1.10659051e+00 -4.24608320e-01 4.09078300e-01
7.08996177e-01 7.82104582e-02 5.90405345e-01 -9.22424972e-01
8.13266277e-01 1.37066436e+00 8.81130457e-01 5.93224108e-01
-6.19887531e-01 -7.68199563e-01 -4.31184828e-01 3.28725189e-01
-1.16936004e+00 -7.60974228e-01 8.56716812e-01 -6.48877561e-01
1.10027695e+00 -2.19620168e-02 1.07213974e+00 1.38392580e+00
3.38073045e-01 6.08429492e-01 1.10562813e+00 -2.93258339e-01
-1.24251597e-01 1.91449732e-01 4.06625792e-02 8.71506393e-01
-4.07222182e-01 -3.24064255e-01 -8.37952733e-01 -7.25884676e-01
2.11422682e-01 -9.63579211e-03 8.76542181e-02 6.69781566e-01
-9.45987761e-01 1.26767206e+00 4.43978578e-01 6.03207231e-01
-6.48177147e-01 -3.18539768e-01 4.47198033e-01 4.22575660e-02
1.08598924e+00 -6.43730164e-04 1.55826077e-01 -7.00786114e-01
-1.04985416e+00 -2.04270020e-01 5.38590550e-01 7.92780370e-02
2.05160771e-02 -4.40684222e-02 -2.94940591e-01 1.31131136e+00
3.34218591e-01 5.38041949e-01 7.75676668e-01 -6.19573176e-01
5.56353092e-01 6.56266749e-01 -3.29909146e-01 -1.13208890e+00
-7.17955470e-01 -7.02332139e-01 -8.20637465e-01 -4.03845876e-01
1.06601678e-02 -4.48104471e-01 -3.98240268e-01 1.62921834e+00
3.72982211e-02 1.69814676e-01 1.54577374e-01 3.60945195e-01
9.65010285e-01 5.76873779e-01 1.09174833e-01 -4.78796959e-01
1.21853399e+00 -8.54248166e-01 -8.57804596e-01 -2.38987401e-01
5.45227468e-01 -7.56070733e-01 8.67558599e-01 1.75454900e-01
-1.37812006e+00 4.53530625e-02 -6.20615780e-01 7.10515231e-02
-4.10403341e-01 -1.02869235e-01 7.14613140e-01 9.22668874e-01
-1.19216466e+00 4.50488061e-01 -5.65865815e-01 -9.25911248e-01
6.90633535e-01 -4.59050313e-02 -3.60615581e-01 1.29079461e-01
-1.21387708e+00 7.73106992e-01 6.23921677e-02 -1.48874849e-01
-6.46985710e-01 -1.94635525e-01 -5.70115626e-01 -1.57778431e-02
-3.47435921e-01 -3.35116982e-01 1.01076496e+00 -8.82266700e-01
-1.21923876e+00 8.64510953e-01 -1.75704882e-01 -4.57359791e-01
2.05861479e-01 -7.12140724e-02 -8.10394645e-01 2.23831281e-01
1.43515959e-01 3.55146706e-01 5.83740890e-01 -4.07863289e-01
-2.79406101e-01 -5.49559236e-01 -2.78929055e-01 1.11326110e-02
-1.26260412e+00 6.04318321e-01 1.59099668e-01 -3.42662573e-01
1.77088715e-02 -5.86722314e-01 4.96291108e-02 -4.65564579e-01
-3.93569946e-01 -3.29835832e-01 3.82384926e-01 -1.23991632e+00
1.41635144e+00 -1.90516043e+00 -3.14696759e-01 -7.49837384e-02
3.11003923e-01 2.12780774e-01 -1.59356996e-01 7.68401980e-01
-6.95458651e-02 4.85185325e-01 3.10071707e-01 -3.14815462e-01
-4.49762642e-01 -1.75577372e-01 -9.88868549e-02 4.96250004e-01
2.41552249e-01 6.70875311e-01 -1.17787957e+00 -4.89233196e-01
-2.91910857e-01 7.75422871e-01 -7.39210844e-01 5.83849885e-02
3.07403564e-01 2.14906424e-01 -3.12619478e-01 8.21164727e-01
4.52744544e-01 -4.55068439e-01 9.78787914e-02 5.99217832e-01
-1.60803720e-01 5.81624687e-01 -3.21937174e-01 1.19609916e+00
-2.92100996e-01 6.95541263e-01 6.92996979e-02 -6.29615963e-01
8.68090153e-01 4.06219631e-01 5.87279081e-01 -8.91619086e-01
2.78148085e-01 1.17463529e-01 1.75944895e-01 -9.30468082e-01
4.69645321e-01 -2.33728483e-01 2.59625494e-01 4.66931850e-01
-1.84736520e-01 3.94236773e-01 -2.05207050e-01 6.65434182e-01
1.03718483e+00 -7.60523200e-01 4.16557580e-01 9.32891220e-02
4.42188650e-01 -4.75971073e-01 4.39763039e-01 6.13681912e-01
-6.34294331e-01 3.30537498e-01 8.25388610e-01 -2.68992394e-01
-7.09942460e-01 -6.27110958e-01 -2.15529799e-01 1.22358084e+00
-7.29802966e-01 -5.20729184e-01 -7.21839190e-01 -4.28918213e-01
-5.27675927e-01 8.57252002e-01 -5.46538830e-01 -1.38797149e-01
5.63493930e-02 -1.09302831e+00 7.91578054e-01 -6.09131679e-02
6.87857628e-01 -1.21317792e+00 5.16872592e-02 3.30986828e-01
-6.94726467e-01 -7.28976309e-01 -2.05470428e-01 -3.89032453e-01
-7.61415482e-01 -8.52690876e-01 -8.55986059e-01 -5.40555596e-01
2.53391474e-01 2.21103132e-02 6.79259837e-01 6.82087541e-02
6.85406383e-03 -2.51026452e-02 -3.96707833e-01 -3.20880741e-01
-6.01489663e-01 1.87601388e-01 8.50434527e-02 -9.94889587e-02
3.80864590e-01 -5.22887170e-01 -4.40144122e-01 -5.80623567e-01
-4.17392731e-01 2.34781504e-01 2.75714159e-01 8.08014214e-01
-2.74039537e-01 -2.43837819e-01 9.96016145e-01 -8.21290970e-01
1.41337478e+00 -1.32622147e+00 1.61980554e-01 6.40626326e-02
-8.93506467e-01 -4.38290119e-01 7.83384219e-02 -3.12425256e-01
-6.43429995e-01 -1.80936120e-02 -7.96770573e-01 4.16131437e-01
5.84556945e-02 1.11302495e+00 4.74329770e-01 3.27604711e-01
9.20632899e-01 3.42987061e-01 3.77330668e-02 -5.40079594e-01
1.84039518e-01 1.43824518e+00 1.78476959e-01 9.91865918e-02
2.00910076e-01 3.84120464e-01 -6.23197258e-01 -1.15208888e+00
-6.75090134e-01 -4.67110753e-01 -8.48270878e-02 -4.06571776e-01
8.96022499e-01 -8.81787598e-01 -6.11928403e-01 9.06939447e-01
-1.36060476e+00 -1.57306552e-01 4.15628999e-01 4.75958318e-01
-4.91694137e-02 2.06031576e-01 -1.03812468e+00 -1.01305187e+00
-9.26426411e-01 -7.35030949e-01 4.12754923e-01 2.96811968e-01
-1.72770545e-01 -8.99802029e-01 5.62038124e-01 6.73461497e-01
8.27308893e-01 1.24590108e-02 8.73682499e-01 -8.29055727e-01
4.81344372e-01 -1.32547781e-01 -4.05471861e-01 6.68056980e-02
7.39233941e-02 -7.56742666e-04 -8.23012769e-01 3.15906294e-02
9.76238698e-02 -6.40679777e-01 7.81165302e-01 3.38370591e-01
8.57521355e-01 -7.97832191e-01 -6.14485480e-02 1.57940507e-01
8.60224307e-01 -3.56812999e-02 7.89207399e-01 2.86835194e-01
3.24882865e-01 5.45845747e-01 1.01378264e-05 8.93918872e-01
7.48281360e-01 2.39794284e-01 2.93968350e-01 -4.96132560e-02
-6.31581992e-02 -3.00997674e-01 8.31397057e-01 1.13757515e+00
-1.74567521e-01 -7.87972063e-02 -1.42980373e+00 7.76171625e-01
-1.57585084e+00 -1.45895743e+00 -2.51592368e-01 1.69052553e+00
9.65136468e-01 2.40408242e-01 4.90540951e-01 -1.51717946e-01
5.02656639e-01 1.85960591e-01 -1.79694965e-01 -7.06272423e-01
-6.25624731e-02 1.58758104e-01 -6.35867985e-03 5.80715775e-01
-8.83599699e-01 7.94083595e-01 6.96916103e+00 5.26146770e-01
-1.35652757e+00 7.81136394e-01 9.39162970e-01 -1.75947160e-01
-2.92100817e-01 -3.41112912e-01 -4.37270790e-01 4.22316104e-01
1.44256806e+00 -1.14409596e-01 6.63295805e-01 7.19959080e-01
6.35795593e-01 1.51432872e-01 -3.08045387e-01 9.51314390e-01
3.62488031e-01 -1.46107376e+00 -3.88428450e-01 7.77688771e-02
6.34416580e-01 6.01276219e-01 3.67800802e-01 3.00995082e-01
2.37078682e-01 -1.01016593e+00 3.72399837e-01 6.12736344e-01
5.10589302e-01 -7.75918782e-01 8.27095866e-01 4.78043079e-01
-4.45462674e-01 -2.55297661e-01 -5.20644560e-02 -2.30660185e-01
1.61544178e-02 8.40585828e-01 -1.29409897e+00 -1.03416122e-01
3.62135112e-01 8.95397663e-01 -6.26145959e-01 8.74438703e-01
-2.35831186e-01 1.02367520e+00 1.51747450e-01 -3.85346174e-01
2.08399430e-01 2.53753811e-01 3.64420116e-01 1.40576410e+00
5.20686209e-01 4.18264158e-02 -1.93673968e-02 7.40169823e-01
-3.39535356e-01 6.03271723e-01 -6.56163931e-01 -6.68868363e-01
7.89189562e-02 1.15961838e+00 -6.43080115e-01 -3.92898947e-01
-4.36541975e-01 7.82130063e-01 4.84663367e-01 1.49421081e-01
-6.20167077e-01 -1.31723478e-01 1.95732668e-01 4.34648067e-01
-4.39080417e-01 -1.96407840e-01 -2.97942698e-01 -9.79745805e-01
-6.45180821e-01 -9.24256206e-01 4.27759349e-01 -6.71884596e-01
-1.31377578e+00 6.38655603e-01 -4.50003922e-01 -5.13956666e-01
-5.03506184e-01 -8.64334777e-02 -8.25843036e-01 8.02793145e-01
-1.27158129e+00 -1.02634072e+00 4.74464260e-02 7.29210019e-01
5.61697960e-01 -5.18557489e-01 9.21464622e-01 3.70272934e-01
-7.67765760e-01 4.30620611e-01 1.04640059e-01 3.09350282e-01
5.04662097e-01 -6.30360961e-01 7.11999387e-02 6.15263700e-01
-3.25027972e-01 6.43423736e-01 4.12257880e-01 -1.07666588e+00
-1.03813553e+00 -8.61787260e-01 1.60928786e+00 3.39214876e-02
1.06726432e+00 -2.86016643e-01 -5.97307265e-01 2.76692033e-01
6.54288292e-01 -7.06801891e-01 9.40855443e-01 7.88615584e-01
-2.97119468e-01 1.49160340e-01 -1.35999477e+00 5.47523260e-01
7.91663170e-01 -9.75884676e-01 -3.57463688e-01 9.71233845e-01
4.80894536e-01 1.74649313e-01 -6.65422916e-01 -3.11980784e-01
5.38564742e-01 -1.02639925e+00 7.77518928e-01 -7.32434154e-01
1.14155364e+00 8.03231597e-01 5.52362055e-02 -1.33591080e+00
-3.91975522e-01 -5.55505693e-01 9.43809673e-02 1.16400850e+00
6.18180633e-01 -6.82466567e-01 5.21726251e-01 4.81251150e-01
1.68999225e-01 -6.00231767e-01 -1.10974717e+00 -1.95122957e-01
2.18487248e-01 -4.35875565e-01 3.41696292e-01 1.32096720e+00
9.35387731e-01 4.82621789e-01 -7.02434838e-01 -1.77363813e-01
1.32988721e-01 -5.18665791e-01 3.41167808e-01 -9.52984452e-01
-1.41077638e-01 -6.22464418e-01 -1.64188847e-01 -1.06594749e-02
1.73224986e-01 -1.11303473e+00 -4.18693930e-01 -1.70507538e+00
9.44794118e-01 -1.15218997e-01 -3.53630692e-01 8.27320993e-01
-7.12235570e-02 1.92094266e-01 -1.17445409e-01 1.09246477e-01
-2.67030209e-01 5.72178245e-01 9.30895746e-01 -2.11770535e-01
-4.17051882e-01 2.95185633e-02 -9.10227358e-01 6.93390191e-01
1.29886448e+00 -6.52364969e-01 -2.71810055e-01 -1.85196429e-01
5.35226405e-01 3.70206326e-01 1.95875734e-01 -8.30279231e-01
1.71652094e-01 -4.13260818e-01 1.46410584e-01 -3.79356712e-01
5.01792669e-01 2.26031635e-02 -2.20004693e-01 8.93488228e-01
-6.02888048e-01 4.96379286e-02 -5.40409535e-02 -2.81835943e-02
1.42111823e-01 -3.81895840e-01 5.83013952e-01 -1.30213350e-01
-6.63535018e-03 1.42479733e-01 -6.54258907e-01 -1.95285003e-03
3.85518640e-01 4.06767637e-01 -8.22066844e-01 -1.02893722e+00
-7.61665583e-01 -2.92207271e-01 1.08435964e-02 5.56190908e-01
1.06551862e+00 -1.24198854e+00 -1.23298728e+00 8.51331800e-02
3.97305340e-02 -1.32678652e+00 1.09024696e-01 1.13902497e+00
-2.45041087e-01 2.40663588e-01 -2.98531264e-01 9.01327282e-02
-1.25221443e+00 3.80377054e-01 5.03949784e-02 -2.19089210e-01
-2.82813609e-01 7.00790346e-01 -6.12479687e-01 -3.69884104e-01
1.12544574e-01 1.69306070e-01 -7.43154585e-01 2.43540853e-01
9.34735239e-01 4.90841180e-01 -1.49012759e-01 -9.25608218e-01
-2.47555912e-01 -2.84157515e-01 -8.78864899e-02 -3.81695271e-01
1.73373926e+00 6.70491979e-02 -5.33072829e-01 3.38697821e-01
1.16179216e+00 2.69157946e-01 -8.91386718e-02 6.08519427e-02
-4.44057025e-02 -3.18910629e-01 4.01694000e-01 -7.80801952e-01
-9.22453463e-01 9.63594019e-01 1.06623054e+00 4.09992516e-01
8.27458978e-01 2.05330536e-01 7.57427752e-01 1.31958485e-01
-6.32437468e-02 -1.03801036e+00 4.25224453e-01 7.49032795e-01
9.84054148e-01 -1.23726368e+00 -3.26135159e-01 1.26660258e-01
-5.35230160e-01 1.12052250e+00 2.39367202e-01 3.86529952e-01
8.76848817e-01 -8.92982632e-02 -3.60263810e-02 -3.87059808e-01
-6.89779460e-01 2.37958338e-02 7.22626830e-03 6.46234334e-01
8.99836123e-01 3.95051062e-01 -8.71195436e-01 6.98736906e-01
-1.90863594e-01 1.62354708e-01 6.31931007e-01 5.11697352e-01
-8.32810938e-01 -1.08650219e+00 -4.01893735e-01 6.75260723e-01
-8.16321135e-01 -5.61306477e-01 -8.16162944e-01 1.79160491e-01
1.46347687e-01 1.35480785e+00 -3.09329987e-01 -7.24398613e-01
-1.77132264e-01 5.23418784e-01 -2.81445175e-01 -4.84437764e-01
-7.49497056e-01 1.68391168e-01 4.58504409e-01 -3.01941186e-01
-2.66796380e-01 -7.43532658e-01 -1.25802875e+00 -7.32027471e-01
-1.11432904e-02 -4.88189384e-02 8.03992152e-01 7.38190651e-01
5.08343875e-01 2.37417787e-01 6.39359355e-01 -5.07282242e-02
-2.90534705e-01 -1.49480295e+00 -4.54502255e-01 1.88020557e-01
2.21820548e-01 -9.98960137e-02 8.37908164e-02 -3.59048456e-01] | [8.993998527526855, 10.697962760925293] |
68208786-90bf-43f8-adfa-eac1b83f78c6 | weakly-supervised-silhouette-based-semantic | 1811.11985 | null | https://arxiv.org/abs/1811.11985v3 | https://arxiv.org/pdf/1811.11985v3.pdf | Weakly Supervised Silhouette-based Semantic Scene Change Detection | This paper presents a novel semantic scene change detection scheme with only weak supervision. A straightforward approach for this task is to train a semantic change detection network directly from a large-scale dataset in an end-to-end manner. However, a specific dataset for this task, which is usually labor-intensive and time-consuming, becomes indispensable. To avoid this problem, we propose to train this kind of network from existing datasets by dividing this task into change detection and semantic extraction. On the other hand, the difference in camera viewpoints, for example, images of the same scene captured from a vehicle-mounted camera at different time points, usually brings a challenge to the change detection task. To address this challenge, we propose a new siamese network structure with the introduction of correlation layer. In addition, we collect and annotate a publicly available dataset for semantic change detection to evaluate the proposed method. The experimental results verified both the robustness to viewpoint difference in change detection task and the effectiveness for semantic change detection of the proposed networks. Our code and dataset are available at https://kensakurada.github.io/pscd. | ['Weimin WANG', 'Ken Sakurada', 'Mikiya Shibuya'] | 2018-11-29 | null | null | null | null | ['scene-change-detection'] | ['computer-vision'] | [ 2.84082979e-01 -6.08063400e-01 3.01826239e-01 -5.74630499e-01
-3.97043705e-01 -4.53286231e-01 4.43462491e-01 -8.61747712e-02
-7.63712645e-01 5.09966016e-01 -1.44979209e-01 1.98807977e-02
1.49220794e-01 -7.53858566e-01 -7.20258296e-01 -7.75486529e-01
4.29123044e-01 -4.02610265e-02 6.12279773e-01 -1.72051355e-01
1.61982447e-01 3.31375241e-01 -1.45802736e+00 -1.00588575e-01
8.95242393e-01 9.56672549e-01 6.34365261e-01 1.32472023e-01
1.23144045e-01 3.64136428e-01 -1.76645055e-01 -2.13703170e-01
4.94508356e-01 -3.59167278e-01 -5.57488441e-01 3.52997184e-01
5.36547422e-01 -3.18371534e-01 -4.09673154e-01 1.44386828e+00
4.85385448e-01 3.06128174e-01 7.66668692e-02 -1.20858836e+00
-1.97016001e-01 3.11289877e-01 -5.44085026e-01 4.50268447e-01
3.53505947e-02 2.98118562e-01 7.52446949e-01 -7.93914676e-01
6.69339120e-01 1.06766570e+00 4.60633248e-01 3.18800360e-01
-8.12226415e-01 -8.13802123e-01 4.54488456e-01 5.55771887e-01
-1.35182750e+00 -5.40354609e-01 1.24825442e+00 -3.90167117e-01
4.36369419e-01 2.61248858e-03 5.83751202e-01 9.35898960e-01
-2.37909332e-01 7.49688685e-01 9.95261967e-01 -4.32588335e-04
2.20706776e-01 1.02789789e-01 1.07709497e-01 5.53966641e-01
3.14690471e-01 -2.41788983e-01 -2.67690923e-02 3.38358492e-01
5.00120163e-01 4.59516644e-01 -5.57276070e-01 -6.77935064e-01
-1.33107293e+00 4.64341909e-01 7.45959044e-01 3.81409407e-01
-1.35221407e-01 -7.26340488e-02 6.03629053e-01 3.89461577e-01
4.23017651e-01 3.41884755e-02 -4.93993163e-01 -1.16844654e-01
-6.12358630e-01 -4.13813889e-02 4.54786152e-01 7.52437651e-01
7.94256985e-01 -2.14727670e-01 1.86517626e-01 8.35651457e-01
1.55208245e-01 4.21455771e-01 4.91560429e-01 -4.28141981e-01
6.81795061e-01 5.31591117e-01 2.67327249e-01 -1.50827110e+00
-4.26134914e-01 -5.12765050e-01 -1.02343333e+00 2.83461716e-02
4.15195197e-01 -7.39965737e-02 -8.07485104e-01 1.81342125e+00
6.81106806e-01 4.30654317e-01 -3.92929502e-02 1.38102448e+00
6.24650717e-01 6.46470487e-01 -1.76282018e-01 -2.63739407e-01
1.25613809e+00 -1.21337998e+00 -6.70999944e-01 -2.17876196e-01
6.46996021e-01 -6.74050868e-01 1.09087276e+00 8.14129934e-02
-6.42168164e-01 -6.54396474e-01 -1.09819627e+00 -5.21691740e-02
-3.31690192e-01 3.03514481e-01 4.01465416e-01 8.46135020e-02
-7.20336080e-01 3.54067653e-01 -1.05345750e+00 -7.10780442e-01
3.58527392e-01 1.16502158e-01 -3.95416647e-01 -3.41791093e-01
-1.30735481e+00 5.61512947e-01 5.80916107e-01 5.04267752e-01
-7.40687370e-01 -2.97167718e-01 -7.83353209e-01 6.40290184e-03
7.31721163e-01 -4.27173376e-01 1.08929932e+00 -1.33413649e+00
-1.38577771e+00 5.72350204e-01 -3.49198058e-02 -2.16264784e-01
8.22890103e-01 -1.31365657e-01 -4.69336778e-01 2.37936080e-02
1.71782196e-01 2.97072053e-01 7.90531158e-01 -1.08762443e+00
-7.70510435e-01 -4.42556739e-01 2.84526169e-01 3.33887070e-01
-3.80955398e-01 -4.97347899e-02 -8.17158461e-01 -5.26970327e-01
1.92532688e-01 -1.06613445e+00 -1.45840198e-01 3.55527326e-02
-2.00027823e-01 -1.43953666e-01 1.12074816e+00 -8.13045084e-01
9.18428898e-01 -2.42523766e+00 2.11719703e-02 1.46398144e-02
-3.28160147e-03 3.90497148e-01 -1.22121245e-01 1.88735530e-01
-4.30757776e-02 -2.55069196e-01 -5.45636475e-01 -2.78342932e-01
-2.91913986e-01 -8.11360702e-02 1.05397023e-01 6.37428582e-01
1.55610098e-02 4.82760012e-01 -1.10041881e+00 -4.07284439e-01
3.11507463e-01 3.39808881e-01 -5.13944030e-01 2.35935360e-01
-1.68668538e-01 8.20801795e-01 -4.73045021e-01 4.21393722e-01
8.92638624e-01 -1.91215947e-01 -3.04472186e-02 -4.50567067e-01
-1.50229841e-01 1.82657450e-01 -1.45744073e+00 2.10044837e+00
-5.75656235e-01 5.31374574e-01 -1.26379812e-02 -1.30130768e+00
6.99274898e-01 1.00160159e-01 4.72397596e-01 -7.18908906e-01
2.23293290e-01 3.42374861e-01 8.30656737e-02 -7.16088831e-01
3.36363822e-01 -6.46599904e-02 -1.35707140e-01 8.92342925e-02
-2.53607810e-01 6.82050362e-02 3.46956164e-01 -3.88954617e-02
1.08510625e+00 1.93906680e-01 2.33986631e-01 6.95740506e-02
8.35598648e-01 4.25743908e-02 1.14373326e+00 2.69600660e-01
-4.58599478e-01 4.86160010e-01 7.81008601e-02 -4.79129732e-01
-7.53749013e-01 -6.72498405e-01 2.82299537e-02 5.99081397e-01
7.01641798e-01 -2.15027571e-01 -5.58400333e-01 -8.54405105e-01
-1.78600177e-01 3.46027613e-01 -2.42179394e-01 -2.38380000e-01
-5.71732402e-01 -7.52450645e-01 6.47439882e-02 2.49367535e-01
1.21837354e+00 -8.11736882e-01 -5.51740885e-01 1.89899012e-01
-4.69869107e-01 -1.39650548e+00 -7.11729467e-01 -2.68673271e-01
-7.68859684e-01 -1.18535364e+00 -5.46569645e-01 -1.00361729e+00
7.82498121e-01 7.48033941e-01 6.14413977e-01 -4.99852411e-02
-1.85512573e-01 3.30739212e-03 -4.16229039e-01 4.42754813e-02
-5.16737998e-02 1.57221779e-01 1.72671732e-02 3.97667497e-01
2.17746019e-01 -6.32002950e-01 -9.39270675e-01 4.42067593e-01
-1.09133315e+00 2.26066127e-01 6.58683896e-01 6.80493355e-01
4.15918410e-01 4.54562396e-01 4.48993951e-01 -6.91333055e-01
1.65692493e-01 -3.67523313e-01 -8.18275690e-01 4.88903001e-02
-4.15339291e-01 -2.91386276e-01 7.80550778e-01 -2.52483130e-01
-1.28050184e+00 4.34744209e-01 -6.25189990e-02 -3.90402406e-01
-3.27137262e-01 5.08098006e-01 -5.66716731e-01 3.54183577e-02
2.25515425e-01 3.61898601e-01 -1.01125151e-01 -5.33453226e-01
2.46230021e-01 7.15266228e-01 6.07139826e-01 9.66756418e-03
1.16989422e+00 6.46460176e-01 -3.08525175e-01 -6.31589949e-01
-7.28248358e-01 -6.70733154e-01 -6.99113727e-01 -1.02436543e-01
8.35205436e-01 -1.22700167e+00 -1.88310251e-01 1.01740730e+00
-1.10867286e+00 -8.31849873e-02 1.56041756e-01 6.21907175e-01
-2.26024821e-01 6.33206189e-01 -3.32261384e-01 -2.55341262e-01
-2.53668636e-01 -1.06822658e+00 8.35478306e-01 3.54673326e-01
3.75362188e-01 -7.89113641e-01 1.20764636e-01 5.36998510e-01
3.58515650e-01 4.25001770e-01 5.08733690e-01 -4.37969387e-01
-7.72417724e-01 -1.79365814e-01 -4.17091668e-01 4.17929053e-01
7.02111900e-01 -1.90414920e-01 -6.95182443e-01 -4.90643412e-01
2.54509002e-01 -3.99422422e-02 8.56358111e-01 -2.79518142e-02
1.05867517e+00 -2.04514250e-01 -3.48605454e-01 6.26688242e-01
1.72066510e+00 2.04401359e-01 2.53777862e-01 5.38156390e-01
1.09048927e+00 3.52626652e-01 8.86197567e-01 4.04634386e-01
5.55866063e-01 9.88431454e-01 4.19859678e-01 -1.02342561e-01
-2.04833075e-01 -8.46392140e-02 3.40434074e-01 1.11106789e+00
2.90837348e-01 -2.26851821e-01 -8.48987162e-01 7.35098183e-01
-2.01104927e+00 -8.91038120e-01 -4.99831028e-02 2.15649438e+00
9.05628920e-01 1.19235821e-01 -8.26274976e-02 4.50078063e-02
1.00495350e+00 3.07667315e-01 -7.88178205e-01 2.65175492e-01
5.70938475e-02 -4.45467323e-01 2.68521518e-01 2.16465876e-01
-1.20443869e+00 1.05803907e+00 3.84474134e+00 6.53486967e-01
-1.46132481e+00 3.65256250e-01 3.04506838e-01 -2.00618073e-01
-5.80187216e-02 9.30432230e-02 -5.19352257e-01 9.51401472e-01
4.02870774e-01 -2.56286804e-02 3.91869098e-01 6.86785161e-01
5.41516244e-01 -1.56868964e-01 -9.14011419e-01 1.29015183e+00
2.25597247e-01 -8.62849355e-01 -1.60865128e-01 -5.11300266e-01
8.30428481e-01 2.31091484e-01 -3.95905584e-01 2.68266082e-01
2.43842471e-02 -3.29255551e-01 4.62300718e-01 3.88435006e-01
4.13436383e-01 -6.01221442e-01 8.62399220e-01 3.11474085e-01
-1.31657767e+00 -2.95227747e-02 -3.32683951e-01 -1.43840507e-01
1.92270249e-01 8.86496425e-01 -5.67836106e-01 6.28183424e-01
8.61707330e-01 1.26539278e+00 -7.09842503e-01 1.23150504e+00
-1.93117350e-01 3.11685622e-01 -3.81243557e-01 2.74494976e-01
1.69719577e-01 -3.22918624e-01 6.11564815e-01 9.67108607e-01
2.95911759e-01 -4.80174012e-02 3.90079051e-01 5.49998283e-01
-2.37380624e-01 1.02249458e-01 -5.77177465e-01 1.79811552e-01
4.32982385e-01 1.47147751e+00 -8.31296742e-01 -2.38097519e-01
-6.74335361e-01 1.31373298e+00 2.52991825e-01 2.54053950e-01
-9.97341275e-01 -6.39588416e-01 6.85415685e-01 -1.41105816e-01
4.74095732e-01 -3.01944941e-01 1.78955406e-01 -1.61359727e+00
5.14806330e-01 -7.70031810e-01 2.97791362e-01 -6.46363616e-01
-1.13109577e+00 3.96014720e-01 -1.35815501e-01 -1.57602906e+00
2.07669616e-01 -1.02794334e-01 -5.94710469e-01 5.06889701e-01
-1.78569484e+00 -8.67751956e-01 -9.94198084e-01 7.78693378e-01
6.46398306e-01 1.78831980e-01 6.18837848e-02 6.67987168e-01
-9.79851782e-01 5.11453331e-01 2.72114486e-01 2.51684248e-01
9.70889807e-01 -9.43601012e-01 3.64147097e-01 1.30858481e+00
-1.46764517e-01 2.10935831e-01 5.30904710e-01 -4.60605443e-01
-1.30375385e+00 -1.42979574e+00 5.53022683e-01 -2.47409660e-02
6.21568024e-01 -2.98245728e-01 -1.00827324e+00 5.37850201e-01
6.11877851e-02 1.34713128e-01 1.78365603e-01 -1.78538114e-01
-1.27752662e-01 -5.64737678e-01 -1.04474449e+00 6.16993845e-01
1.41635525e+00 -3.84177476e-01 -4.86028224e-01 3.63478810e-01
8.32200229e-01 -5.77308297e-01 -7.08305538e-01 5.50336719e-01
1.40136346e-01 -7.63669193e-01 5.05601346e-01 -9.05142948e-02
2.77799398e-01 -9.03858900e-01 9.90827940e-03 -1.52308154e+00
-1.95203513e-01 -2.46811539e-01 4.22168612e-01 1.40328991e+00
-4.47443090e-02 -8.46468925e-01 4.50110644e-01 2.63428509e-01
-1.10904388e-01 -4.23128247e-01 -8.29490364e-01 -8.41015875e-01
-3.24934989e-01 -2.21169904e-01 5.84074080e-01 1.09591901e+00
-3.66041183e-01 4.80864346e-01 -3.52348596e-01 5.45816720e-01
4.57758129e-01 4.20335501e-01 9.41257417e-01 -1.15440285e+00
-6.43583462e-02 -1.60119757e-01 -7.61571288e-01 -9.77448761e-01
7.86466449e-02 -8.59179437e-01 2.91461319e-01 -1.51222825e+00
3.59012276e-01 -5.29876828e-01 -4.88546818e-01 4.56405967e-01
-2.89034158e-01 1.10548541e-01 2.00359255e-01 4.79357094e-01
-7.82212794e-01 8.52271914e-01 1.18410659e+00 -2.17851773e-01
-2.10421324e-01 -1.36318177e-01 -3.89809340e-01 7.15845466e-01
1.05464339e+00 -4.82121229e-01 -6.91246927e-01 -7.23444819e-01
1.07649960e-01 -1.43250629e-01 5.49564719e-01 -1.22699559e+00
2.79810011e-01 -1.41112059e-01 -1.75518952e-02 -4.59754616e-01
1.79715361e-02 -1.16985261e+00 1.54536724e-01 4.99296367e-01
-6.43485412e-02 7.17555806e-02 1.05192944e-01 7.63540506e-01
-4.89646941e-01 -1.74388550e-02 8.98220718e-01 -2.34692767e-02
-1.10541773e+00 5.25731862e-01 1.12700358e-01 2.30920836e-02
1.20705616e+00 1.09398894e-01 -3.51922274e-01 -1.62759528e-01
-3.91587019e-01 3.86152923e-01 7.58037806e-01 7.79806256e-01
3.17325562e-01 -1.43167651e+00 -5.43027163e-01 1.10674575e-01
4.16658103e-01 4.47884411e-01 4.82211947e-01 1.09796286e+00
-3.48880529e-01 1.12617522e-01 -2.75889099e-01 -7.07884550e-01
-1.25737441e+00 5.73031187e-01 4.53075796e-01 -6.52853474e-02
-6.81398749e-01 4.35865521e-01 1.73919916e-01 -5.33489287e-01
4.72103879e-02 -5.36756217e-01 -8.13053027e-02 7.14295283e-02
2.54350871e-01 1.93796247e-01 1.75960794e-01 -5.51505089e-01
-5.17738819e-01 7.28602886e-01 -1.22060947e-01 2.25717902e-01
1.23180544e+00 -5.53761721e-01 -5.13314381e-02 4.85044330e-01
1.41473889e+00 -2.73488969e-01 -1.44386673e+00 -6.62138581e-01
-2.61373013e-01 -6.00067914e-01 9.82215181e-02 -6.29223168e-01
-1.57709599e+00 6.53680503e-01 8.98188412e-01 -6.13753796e-02
1.33919322e+00 -3.35366517e-01 1.17844772e+00 6.13507807e-01
4.35400039e-01 -1.25258541e+00 1.50684267e-02 3.87762815e-01
7.36521542e-01 -1.72087562e+00 -1.98900774e-01 -4.26681310e-01
-4.88979161e-01 8.23598862e-01 7.74896204e-01 -6.72012642e-02
6.95664227e-01 -3.62490654e-01 1.48109078e-01 -3.16311559e-03
-4.44543064e-01 -2.15419069e-01 -6.21490143e-02 1.01161651e-01
2.97456868e-02 -9.25027132e-02 -4.37066138e-01 1.17252596e-01
2.96279527e-02 1.61482781e-01 5.02481461e-01 9.44657207e-01
-2.29119763e-01 -8.30243647e-01 -8.89695212e-02 6.72408044e-02
-1.81534588e-01 1.52677640e-01 -8.10555294e-02 6.72100008e-01
2.63935059e-01 1.00745201e+00 6.26942702e-03 -3.78457725e-01
4.02191758e-01 -1.38042629e-01 4.99655046e-02 -4.55344856e-01
-1.85486272e-01 -1.92659795e-01 -7.77043030e-02 -6.23457372e-01
-7.32777357e-01 -9.36487615e-01 -1.30462015e+00 -9.58493277e-02
-2.14649484e-01 -7.73814619e-02 6.08765841e-01 9.07784998e-01
4.39494878e-01 4.75942463e-01 1.03962922e+00 -6.00301087e-01
-4.32822138e-01 -1.00644660e+00 -4.17576790e-01 8.69571984e-01
3.58010888e-01 -6.63330495e-01 -4.55040067e-01 2.07707047e-01] | [9.634237289428711, -1.1036251783370972] |
92ead621-fb86-45f4-b173-c6d6c441ef18 | learning-deep-representations-of-medical | 1711.08490 | null | http://arxiv.org/abs/1711.08490v2 | http://arxiv.org/pdf/1711.08490v2.pdf | Learning Deep Representations of Medical Images using Siamese CNNs with Application to Content-Based Image Retrieval | Deep neural networks have been investigated in learning latent
representations of medical images, yet most of the studies limit their approach
in a single supervised convolutional neural network (CNN), which usually rely
heavily on a large scale annotated dataset for training. To learn image
representations with less supervision involved, we propose a deep Siamese CNN
(SCNN) architecture that can be trained with only binary image pair
information. We evaluated the learned image representations on a task of
content-based medical image retrieval using a publicly available multiclass
diabetic retinopathy fundus image dataset. The experimental results show that
our proposed deep SCNN is comparable to the state-of-the-art single supervised
CNN, and requires much less supervision for training. | ['Wei-Hung Weng', 'Yu-An Chung'] | 2017-11-22 | null | null | null | null | ['medical-image-retrieval', 'medical-image-retrieval'] | ['computer-vision', 'medical'] | [ 4.19638366e-01 2.37675801e-01 -6.20491743e-01 -6.33802474e-01
-8.70756507e-01 5.20654880e-02 4.73555177e-01 9.19339210e-02
-6.88310206e-01 5.79048336e-01 2.43239984e-01 -5.61837442e-02
-3.05412412e-01 -5.83831489e-01 -6.27675474e-01 -7.28479743e-01
5.91906756e-02 5.17817497e-01 8.72542674e-04 2.04798296e-01
7.47750103e-02 3.93895835e-01 -1.47668779e+00 7.23494172e-01
8.03658009e-01 1.17767835e+00 4.08215225e-02 5.29238760e-01
8.81668925e-02 1.39137495e+00 -5.54655015e-01 -1.59689993e-01
2.24385723e-01 -4.77815121e-01 -1.00621581e+00 1.89936996e-01
8.71890485e-01 -4.40501004e-01 -5.78249514e-01 1.24520004e+00
5.93172133e-01 -1.61027238e-01 7.89432824e-01 -1.00482309e+00
-1.04011130e+00 2.35758305e-01 -1.01749167e-01 4.12692577e-01
-2.68389523e-01 7.83804804e-03 8.30983698e-01 -5.38179696e-01
7.97160625e-01 1.18370414e+00 5.01175284e-01 4.54170078e-01
-1.12564385e+00 -3.28584015e-01 -1.83782727e-01 2.18199000e-01
-1.15169513e+00 -2.39596859e-01 5.35849094e-01 -4.39535052e-01
1.06152737e+00 -1.11294560e-01 6.45822644e-01 1.02240324e+00
1.92521244e-01 1.06728625e+00 1.05177164e+00 -2.52527356e-01
3.99083681e-02 -1.42536536e-01 1.84122160e-01 1.01168227e+00
3.78815889e-01 1.07604846e-01 -2.36683562e-01 -7.61993900e-02
8.03672314e-01 4.53384787e-01 -2.40867153e-01 -6.13441646e-01
-1.39507985e+00 1.19314480e+00 1.14645028e+00 3.06034982e-01
-2.64300793e-01 3.83614451e-01 7.92253613e-01 4.21762109e-01
6.93468750e-01 3.55457544e-01 -3.66559684e-01 2.23110139e-01
-1.14887130e+00 1.79215148e-02 4.80242997e-01 7.87789285e-01
6.33194149e-01 -2.47259781e-01 -2.66541660e-01 9.01932955e-01
2.42392570e-01 3.35378021e-01 1.06802952e+00 -8.82770360e-01
4.03660506e-01 9.83565748e-01 -2.26788327e-01 -8.49826872e-01
-4.83728081e-01 -5.56683600e-01 -1.16997707e+00 2.83051580e-01
1.86191693e-01 1.10864304e-01 -1.35314465e+00 1.08108759e+00
-3.88851196e-01 2.15747610e-01 4.72258389e-01 1.05056906e+00
1.33542776e+00 2.30862767e-01 -4.73824814e-02 2.20842268e-02
1.10791862e+00 -1.71095729e+00 -7.34322846e-01 -9.49260406e-03
9.54553246e-01 -6.13474071e-01 7.09741592e-01 3.63589823e-01
-9.49933648e-01 -6.74785614e-01 -9.96814728e-01 -5.25289178e-01
-5.48583210e-01 7.74401307e-01 8.12742114e-01 1.25109628e-01
-1.06152260e+00 5.90131164e-01 -1.09617913e+00 -4.34742123e-01
9.99583364e-01 4.08215731e-01 -6.38574243e-01 -4.31601375e-01
-7.76274741e-01 9.50613916e-01 5.82003653e-01 3.21758121e-01
-1.08595181e+00 -5.39903879e-01 -1.11080885e+00 2.08767690e-02
-1.87692121e-01 -8.10692132e-01 8.82468820e-01 -1.35226655e+00
-1.28635752e+00 1.32533562e+00 1.77932277e-01 -9.62662697e-01
5.69263637e-01 -1.99952200e-01 -3.10226768e-01 7.92928696e-01
1.38466448e-01 1.20454335e+00 7.59903252e-01 -8.46510708e-01
-3.29097509e-01 -1.11330725e-01 5.30642420e-02 -2.08473578e-01
-3.63106638e-01 9.66393501e-02 -3.31128359e-01 -7.75033593e-01
8.83386061e-02 -9.97561276e-01 -4.36075240e-01 5.26434183e-01
-4.33792233e-01 -4.06232208e-01 7.84767449e-01 -2.84166038e-01
6.79674923e-01 -2.01958036e+00 2.43763208e-01 -5.23266830e-02
2.04317600e-01 5.73470652e-01 -5.75940847e-01 2.13229105e-01
-3.40802938e-01 -1.45566687e-01 -3.84043097e-01 -7.31072545e-01
-3.59852463e-01 5.37809551e-01 -2.44769454e-01 7.03090608e-01
4.33085382e-01 1.18573761e+00 -1.08945978e+00 -6.46718323e-01
1.91214055e-01 5.73680818e-01 -5.88029206e-01 2.94873595e-01
-9.47897285e-02 2.36482069e-01 -2.13097408e-01 9.95238364e-01
4.01394039e-01 -9.45375681e-01 1.09954484e-01 -4.33372647e-01
2.04702169e-01 -1.58954803e-02 -5.53071320e-01 2.23050261e+00
-2.53294885e-01 8.97904515e-01 -5.26250482e-01 -1.55555129e+00
6.39368236e-01 3.39357466e-01 6.34876847e-01 -7.02465892e-01
2.71475017e-01 3.57841015e-01 -8.41953903e-02 -8.28031898e-01
3.13685313e-02 2.22669721e-01 2.68060833e-01 1.28788814e-01
6.17584944e-01 1.96190104e-01 3.09000313e-01 7.82698393e-02
1.06971467e+00 1.69259205e-01 3.09195995e-01 -1.29785284e-01
3.96196246e-01 1.07589111e-01 4.20118988e-01 6.86232090e-01
-3.94185871e-01 1.06296289e+00 5.48695385e-01 -1.03522015e+00
-9.14729655e-01 -6.46663368e-01 -5.79252303e-01 7.06541836e-01
9.80363563e-02 -3.31700206e-01 -4.42798138e-01 -8.60651672e-01
6.46601319e-02 -1.65205151e-01 -1.03828979e+00 -6.63196621e-03
-3.64836723e-01 -8.32375824e-01 5.38879097e-01 5.17700970e-01
6.49614573e-01 -1.29588807e+00 -6.93746686e-01 -8.74486566e-03
1.45398244e-01 -1.14162326e+00 -8.26866329e-02 4.92935516e-02
-1.30342042e+00 -1.47254181e+00 -1.19396269e+00 -1.11885858e+00
1.19006479e+00 3.33615951e-02 1.06809771e+00 1.84440568e-01
-9.05848742e-01 3.01780432e-01 -4.37256306e-01 -3.59162748e-01
-2.34813765e-01 1.95012376e-01 -2.65273839e-01 1.95028573e-01
3.44571620e-01 -1.98920533e-01 -9.98459399e-01 1.52272768e-02
-1.10147512e+00 5.05511556e-03 1.13818300e+00 1.39103591e+00
7.80536175e-01 -3.08396071e-01 4.03985023e-01 -1.20308280e+00
3.11197579e-01 -4.81664121e-01 -6.20358646e-01 3.10857713e-01
-6.62985742e-01 7.50575066e-02 2.80844986e-01 -2.98048347e-01
-5.03446102e-01 3.16198319e-01 1.66214868e-01 -8.02980900e-01
-2.75848180e-01 8.25510204e-01 6.06265247e-01 -3.35782856e-01
5.89149833e-01 2.41106883e-01 3.30830783e-01 -6.41441703e-01
3.38438809e-01 6.45955384e-01 5.42222619e-01 4.09725457e-02
2.13240176e-01 6.52588785e-01 2.21551713e-02 -3.85902703e-01
-1.28146648e+00 -6.92945898e-01 -7.98827946e-01 2.91383177e-01
1.17657602e+00 -1.28519297e+00 -4.36353177e-01 4.15502697e-01
-1.02100539e+00 -2.22165555e-01 -2.35058069e-01 8.41387153e-01
-6.64007604e-01 3.65177631e-01 -7.40175247e-01 -1.17485523e-01
-3.69216651e-01 -1.20627785e+00 1.26510715e+00 2.42447034e-02
2.15537101e-02 -1.09215403e+00 3.39194477e-01 7.30793118e-01
5.07942498e-01 2.83438623e-01 8.80213022e-01 -5.90986252e-01
-7.74997234e-01 -4.77933019e-01 -7.11798251e-01 8.24820578e-01
3.92668664e-01 -1.61766499e-01 -9.10332203e-01 -7.60429084e-01
-4.33767617e-01 -1.07752597e+00 1.56975102e+00 4.80644941e-01
1.60157120e+00 -3.58443558e-01 -4.25273925e-01 9.87705410e-01
1.63400662e+00 -3.72736663e-01 8.23164225e-01 4.07745510e-01
5.24882078e-01 3.50663751e-01 2.62122303e-01 3.24154994e-03
2.06167772e-01 4.21837002e-01 7.14834750e-01 -8.38575959e-01
-3.61752182e-01 1.28543496e-01 -7.36413002e-02 8.71443808e-01
-2.97621101e-01 -1.74366273e-02 -6.91410482e-01 9.07208502e-01
-2.08354783e+00 -7.08399892e-01 4.14759308e-01 1.62505484e+00
8.56381893e-01 -1.25622392e-01 -2.82073230e-01 -2.75689304e-01
1.90151572e-01 1.88926324e-01 -5.80458283e-01 7.93332160e-02
-1.35697901e-01 2.86171973e-01 6.27644181e-01 7.12453574e-02
-1.53879690e+00 7.36321867e-01 7.26621532e+00 5.47259569e-01
-1.36077368e+00 9.93169993e-02 6.59421861e-01 -2.83366710e-01
2.68032342e-01 -4.62444752e-01 -2.92890668e-01 4.08208966e-01
9.43010449e-01 3.65196258e-01 -2.70044357e-01 9.58212256e-01
-1.26590103e-01 1.03772640e-01 -1.11934471e+00 1.22945893e+00
3.71041536e-01 -1.70574474e+00 6.50470495e-01 -3.49369496e-02
1.13679266e+00 6.44391775e-01 3.35252374e-01 4.57231142e-02
1.24779277e-01 -1.56065786e+00 -5.40986843e-02 8.86348903e-01
8.49083960e-01 -3.90028059e-01 1.36729884e+00 -1.51832089e-01
-6.43443465e-01 -1.61934733e-01 -7.55213499e-01 2.51073092e-01
-2.46529981e-01 3.23896915e-01 -6.30297482e-01 4.82728839e-01
9.33608711e-01 1.43309319e+00 -8.45044851e-01 1.39760530e+00
-4.84204553e-02 6.56434298e-01 1.41130880e-01 3.40706021e-01
6.81499720e-01 -3.55892368e-02 3.57392319e-02 1.16965902e+00
1.56547323e-01 -1.04931228e-01 3.91244769e-01 8.75585437e-01
-3.81058276e-01 2.55437493e-01 -7.26778865e-01 -3.37115407e-01
-3.96727532e-01 1.21716261e+00 -6.40238762e-01 -5.49239695e-01
-6.18587017e-01 1.04488289e+00 5.06554246e-01 4.33793694e-01
-3.60116869e-01 -3.66324723e-01 4.62519050e-01 -2.65483648e-01
4.86952990e-01 1.33479267e-01 2.96129342e-02 -1.40736187e+00
-2.46163547e-01 -7.51788437e-01 5.41079879e-01 -9.01058197e-01
-1.70443094e+00 8.57932448e-01 -1.44434616e-01 -1.67032361e+00
-1.00030556e-01 -9.59240437e-01 -2.48209849e-01 6.05052352e-01
-2.30743265e+00 -1.51318860e+00 -4.24926281e-01 8.12754810e-01
3.91655892e-01 -8.66754115e-01 1.22597873e+00 4.08135086e-01
-4.54111397e-01 4.00633454e-01 2.46141359e-01 6.21568084e-01
9.00220215e-01 -1.15023994e+00 -1.99785933e-01 4.53751534e-01
3.51011723e-01 8.58018816e-01 5.76109923e-02 -2.85444945e-01
-8.61974239e-01 -1.45053291e+00 7.77490795e-01 -3.19619447e-01
3.64556342e-01 1.83839247e-01 -7.25685656e-01 7.63138473e-01
5.54371893e-01 1.02258933e+00 9.38297451e-01 -2.35147953e-01
-3.54210764e-01 -1.72606021e-01 -1.12013578e+00 1.09679475e-01
7.47887373e-01 -7.43460596e-01 -6.48570061e-01 7.61687279e-01
7.38236606e-01 -3.81755978e-01 -1.05894911e+00 5.86992085e-01
2.60532230e-01 -6.24852836e-01 1.03934336e+00 -1.16614890e+00
8.70150089e-01 -2.60432422e-01 2.04722919e-02 -9.88379061e-01
-1.37896389e-01 -1.52888805e-01 3.15107554e-02 3.45452070e-01
3.23602170e-01 -5.88673711e-01 7.05087602e-01 4.25253212e-02
-2.16094509e-01 -1.08711660e+00 -9.10808146e-01 -5.16251266e-01
3.94890793e-02 1.62653953e-01 6.28806278e-02 1.02497506e+00
-3.89396191e-01 1.58844627e-02 -4.86436337e-01 1.62617162e-01
5.64328432e-01 4.43723619e-01 3.67426366e-01 -1.20658553e+00
1.64214708e-02 -2.93149412e-01 -1.11536777e+00 -6.59966290e-01
3.15567583e-01 -1.29741895e+00 -2.01469272e-01 -1.78712344e+00
4.86940980e-01 -1.80204511e-01 -1.01541638e+00 8.72668207e-01
1.02363005e-01 9.44866002e-01 -1.19841650e-01 4.99177039e-01
-1.06008470e+00 6.18155897e-01 1.20913172e+00 -7.19243586e-01
2.21360028e-01 -1.90269738e-01 -4.13852602e-01 6.90720081e-01
4.63395566e-01 -6.54318273e-01 -5.22779882e-01 -5.92161715e-01
5.62782884e-02 -2.40082115e-01 6.01097465e-01 -1.18203628e+00
3.27218115e-01 3.13046455e-01 5.25382280e-01 -4.67350423e-01
1.54517189e-01 -8.02503884e-01 -4.11507785e-01 7.93718934e-01
-8.68443489e-01 -2.50610471e-01 8.12839195e-02 5.68063319e-01
-9.28745985e-01 -1.57061994e-01 7.55270064e-01 -4.16576385e-01
-5.62545538e-01 7.11016595e-01 -1.42843336e-01 -1.54515237e-01
7.85697639e-01 9.82246399e-02 -4.55331147e-01 5.44045307e-02
-8.64455283e-01 8.66445154e-02 1.84052035e-01 4.43680555e-01
8.40733826e-01 -1.39470470e+00 -7.02160537e-01 2.99156070e-01
6.33058786e-01 -3.43305245e-02 1.33551762e-01 7.90105700e-01
-1.02007127e+00 8.25078189e-01 -6.43989503e-01 -9.26654577e-01
-1.29054081e+00 6.96105003e-01 6.76734030e-01 -3.74829412e-01
-9.97159898e-01 8.57194245e-01 7.42976069e-02 -3.83269429e-01
4.07766134e-01 -9.10284579e-01 -5.92550099e-01 1.24242984e-01
5.36029696e-01 -5.21769710e-02 5.48111461e-02 -3.85917991e-01
-1.29707277e-01 6.75168455e-01 -2.63200521e-01 5.44036210e-01
1.64894474e+00 3.38198930e-01 -5.63315511e-01 2.21622914e-01
1.66018224e+00 -8.16576600e-01 -1.04755783e+00 -5.78696072e-01
-8.57990310e-02 -3.87017369e-01 3.84018153e-01 -7.18910694e-01
-1.46899390e+00 1.02452934e+00 1.19985282e+00 -2.44347572e-01
1.03786075e+00 -7.47507438e-02 8.21252942e-01 9.95046496e-01
1.97879970e-01 -9.18389201e-01 3.50233674e-01 1.76988557e-01
1.05924261e+00 -1.90111244e+00 2.58217007e-01 -4.57176603e-02
-5.39378166e-01 1.46975088e+00 4.41339403e-01 -6.35174036e-01
8.17646444e-01 -2.80614555e-01 4.22083557e-01 -4.98668492e-01
-6.62502229e-01 -2.59144485e-01 8.20487380e-01 2.90097326e-01
4.52586800e-01 -2.34674141e-01 -1.39154509e-01 1.44798473e-01
2.48687387e-01 5.38956463e-01 3.38230103e-01 9.01345968e-01
4.25625555e-02 -1.05544591e+00 3.60727608e-01 6.52325988e-01
-7.30791688e-01 -2.29178384e-01 -6.65024966e-02 5.31909704e-01
3.24730903e-01 4.20722038e-01 1.98059022e-01 3.42752248e-01
8.45640078e-02 -2.83664435e-01 4.43824291e-01 -9.34889138e-01
-4.29134399e-01 -2.11418122e-01 -2.13376790e-01 -9.67728615e-01
-1.28430128e+00 -1.38096660e-01 -9.31489646e-01 5.45916796e-01
4.80138883e-02 -8.98192078e-02 3.51216137e-01 8.93925488e-01
4.43876833e-01 7.70649254e-01 1.91648826e-01 -8.21805358e-01
-4.68302578e-01 -1.12738919e+00 -5.67928851e-01 7.01558769e-01
8.23399961e-01 -5.86682141e-01 -1.39209433e-02 3.13549817e-01] | [14.83836555480957, -2.436047077178955] |
698401a0-af79-4d76-a4eb-cd2789f0c0c8 | 3d-registration-with-maximal-cliques-1 | 2305.10854 | null | https://arxiv.org/abs/2305.10854v1 | https://arxiv.org/pdf/2305.10854v1.pdf | 3D Registration with Maximal Cliques | As a fundamental problem in computer vision, 3D point cloud registration (PCR) aims to seek the optimal pose to align a point cloud pair. In this paper, we present a 3D registration method with maximal cliques (MAC). The key insight is to loosen the previous maximum clique constraint, and mine more local consensus information in a graph for accurate pose hypotheses generation: 1) A compatibility graph is constructed to render the affinity relationship between initial correspondences. 2) We search for maximal cliques in the graph, each of which represents a consensus set. We perform node-guided clique selection then, where each node corresponds to the maximal clique with the greatest graph weight. 3) Transformation hypotheses are computed for the selected cliques by the SVD algorithm and the best hypothesis is used to perform registration. Extensive experiments on U3M, 3DMatch, 3DLoMatch and KITTI demonstrate that MAC effectively increases registration accuracy, outperforms various state-of-the-art methods and boosts the performance of deep-learned methods. MAC combined with deep-learned methods achieves state-of-the-art registration recall of 95.7% / 78.9% on 3DMatch / 3DLoMatch. | ['Yanning Zhang', 'Shikun Zhang', 'Jiaqi Yang', 'Xiyu Zhang'] | 2023-05-18 | 3d-registration-with-maximal-cliques | http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_3D_Registration_With_Maximal_Cliques_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_3D_Registration_With_Maximal_Cliques_CVPR_2023_paper.pdf | cvpr-2023-1 | ['point-cloud-registration'] | ['computer-vision'] | [-1.74842075e-01 2.66419679e-01 -1.79745153e-01 -7.44184181e-02
-7.95644760e-01 -3.72079790e-01 4.03282821e-01 1.71575889e-01
-1.85717568e-01 2.25660726e-01 -2.06098109e-02 3.77601665e-03
-2.03373820e-01 -8.02776933e-01 -8.30481887e-01 -5.39432883e-01
-3.85060757e-01 1.11776781e+00 3.92253637e-01 -2.07983047e-01
2.19558671e-01 8.56513381e-01 -1.13753784e+00 -2.17774108e-01
9.13832188e-01 7.69491494e-01 2.55336493e-01 7.01732934e-02
5.28271124e-02 -3.14052962e-02 -3.74805123e-01 -1.58716276e-01
6.75810397e-01 -6.31810203e-02 -8.13393891e-01 2.07230866e-01
8.73953521e-01 -9.02879462e-02 -2.58434147e-01 1.05131483e+00
4.76235688e-01 1.60676558e-02 6.11124754e-01 -1.26694500e+00
-1.57489955e-01 2.39523947e-01 -1.21290267e+00 -1.94525495e-01
5.46398997e-01 -1.73955128e-01 9.26629484e-01 -1.20233607e+00
9.88795340e-01 1.19626772e+00 7.14077234e-01 2.75218785e-01
-1.10861087e+00 -8.95100176e-01 -1.13440685e-01 1.71573237e-02
-1.95805156e+00 -1.70990407e-01 8.05081427e-01 -3.05498749e-01
8.56444240e-01 1.37381569e-01 9.77802098e-01 5.32477796e-01
3.96924794e-01 2.56736577e-01 8.93501997e-01 -3.57062489e-01
1.42947197e-01 -5.89963377e-01 3.11050806e-02 9.99162972e-01
3.58458847e-01 -3.63789387e-02 -6.71565533e-01 -4.62336361e-01
1.14988279e+00 -8.63506049e-02 -8.47260132e-02 -6.55823886e-01
-1.43475437e+00 7.37643421e-01 9.53603625e-01 1.23722509e-01
-5.06280839e-01 2.04536647e-01 -1.21693611e-01 1.80808067e-01
3.85512948e-01 2.69021243e-01 -1.89770967e-01 4.29881096e-01
-9.26407158e-01 3.35092306e-01 6.55762732e-01 1.27755857e+00
1.32478940e+00 -4.45740849e-01 1.51230380e-01 6.49206460e-01
6.38605416e-01 6.65277660e-01 -2.47232690e-01 -1.03088725e+00
4.47459996e-01 8.84835005e-01 -2.92197376e-01 -1.30691600e+00
-5.22301495e-01 -4.29042071e-01 -9.47098911e-01 2.25520089e-01
6.29349500e-02 2.02560425e-01 -1.13869369e+00 1.31211138e+00
7.78979003e-01 7.86779463e-01 -1.68553889e-01 1.07546890e+00
9.33588922e-01 3.30772221e-01 -3.01905274e-01 -2.08348975e-01
1.36814773e+00 -5.67263067e-01 -2.98503995e-01 -3.53241384e-01
2.93927968e-01 -1.08974183e+00 4.59789634e-01 -1.29778191e-01
-7.63356507e-01 -4.66068119e-01 -1.04054630e+00 2.36743093e-02
2.20374748e-01 -2.03389913e-01 5.51882386e-01 8.36014971e-02
-1.13562977e+00 5.37630737e-01 -9.20279205e-01 -3.85944724e-01
3.76628250e-01 8.16128016e-01 -6.61829412e-01 -1.93466172e-01
-7.50558972e-01 8.81583929e-01 2.42394537e-01 2.00023502e-01
-4.28460687e-01 -5.21455109e-01 -7.28223741e-01 -3.79980117e-01
3.68029624e-01 -1.01605284e+00 7.52008379e-01 -3.41032773e-01
-1.03992927e+00 1.37981009e+00 -2.86653876e-01 -2.10042730e-01
3.89962524e-01 3.39425988e-02 1.03195652e-01 1.64029300e-01
3.72461617e-01 9.45932686e-01 5.81929386e-01 -1.58839750e+00
-4.65249121e-01 -5.96320987e-01 -2.45584756e-01 4.34550524e-01
3.88519317e-01 -1.32795438e-01 -1.10149777e+00 -1.66652247e-01
1.20028698e+00 -1.50083363e+00 -5.69826186e-01 -8.11170712e-02
-5.92198789e-01 -4.86061633e-01 8.66300762e-01 -5.45982122e-01
7.63248920e-01 -1.84435356e+00 2.99963832e-01 9.35813248e-01
5.98154247e-01 -1.54090405e-01 -1.61929503e-01 3.90902549e-01
1.81660011e-01 -2.21929058e-01 -6.08936064e-02 -5.54442644e-01
-1.24838874e-01 2.48434693e-01 1.18013054e-01 9.39206481e-01
-1.92577794e-01 9.15580750e-01 -7.58294642e-01 -7.83102870e-01
4.45493519e-01 3.92459899e-01 -5.32013297e-01 -1.53324027e-02
-8.51640552e-02 6.27373576e-01 -6.65829241e-01 7.59476840e-01
1.19712090e+00 -4.26327854e-01 2.25701973e-01 -6.14290774e-01
-2.98274737e-02 -1.05052199e-02 -1.55341852e+00 2.08267164e+00
9.27911401e-02 2.90321708e-01 -5.06532229e-02 -3.89197797e-01
1.32606757e+00 -4.04117741e-02 9.70493615e-01 -3.79789472e-01
2.45900795e-01 3.53196710e-01 -1.38703257e-01 1.02815535e-02
4.46826160e-01 2.99438775e-01 -1.43956155e-01 2.02139005e-01
-9.39517766e-02 -3.34949821e-01 -3.27595115e-01 2.44390965e-01
8.96215737e-01 5.03395051e-02 3.39712977e-01 -4.04275835e-01
3.88702333e-01 1.73598439e-01 6.83352590e-01 5.25889635e-01
9.11503211e-02 7.92958975e-01 7.84246251e-02 -4.52428758e-01
-8.15138400e-01 -1.10807467e+00 -6.38008639e-02 2.95616359e-01
8.23631227e-01 -4.93441045e-01 -6.41521931e-01 -4.66519445e-01
1.50784418e-01 3.31147648e-02 -3.40050906e-01 -5.54758720e-02
-9.01448548e-01 -3.72899324e-01 1.72303081e-01 1.93426460e-01
5.55154562e-01 -6.37820601e-01 -2.62853622e-01 7.63056129e-02
-3.29614609e-01 -1.15897691e+00 -6.11915231e-01 -2.58576751e-01
-1.13953471e+00 -1.37699568e+00 -1.80580169e-01 -1.20765460e+00
1.13082874e+00 5.90770304e-01 1.15876913e+00 5.01012087e-01
3.53569202e-02 2.24252880e-01 -2.51474708e-01 -5.82000986e-02
-1.87324733e-01 2.24654749e-01 2.57820338e-01 -1.36664942e-01
4.13082540e-01 -6.75816059e-01 -6.44555271e-01 5.60339808e-01
-3.32491249e-01 6.68116957e-02 7.43941069e-01 3.85642380e-01
1.23717856e+00 -3.72881293e-01 -3.13450024e-02 -5.95456779e-01
2.57383436e-01 -6.51153773e-02 -7.62625873e-01 1.39275491e-01
-4.67338204e-01 -9.02389511e-02 -1.59835175e-01 -2.46849090e-01
-4.27374423e-01 6.79046452e-01 -1.07024750e-03 -9.64927316e-01
5.44461329e-03 5.24887800e-01 -1.55344814e-01 -6.94419265e-01
6.93206906e-01 -1.10122986e-01 1.32081211e-01 -4.12882090e-01
3.87393147e-01 3.67411464e-01 6.37206852e-01 -6.65233612e-01
1.31362939e+00 6.51581228e-01 4.61987287e-01 -7.87782371e-01
-6.11266077e-01 -6.50576830e-01 -1.15997636e+00 -4.59244370e-01
9.96829867e-01 -1.07185888e+00 -8.13211977e-01 3.39488268e-01
-1.41572344e+00 3.10331844e-02 8.64632651e-02 6.56890094e-01
-3.55694711e-01 4.87869412e-01 -4.12589133e-01 -3.88335764e-01
-4.77913648e-01 -1.21040308e+00 1.50843394e+00 1.36575148e-01
-1.97611719e-01 -6.94173872e-01 2.52969563e-01 4.19601023e-01
-2.30531991e-01 7.17176676e-01 5.69022834e-01 -4.31788951e-01
-1.16185617e+00 -3.29931676e-01 -1.06996752e-01 -2.78134018e-01
6.32907450e-02 -3.56659070e-02 -6.79331362e-01 -5.97711325e-01
-2.93974072e-01 2.10009113e-01 4.73089933e-01 4.32519585e-01
6.56330645e-01 2.51758303e-02 -8.06642354e-01 8.25045526e-01
1.30215383e+00 -1.49483070e-01 7.52340436e-01 2.81516850e-01
1.05563617e+00 1.30504414e-01 7.19418466e-01 2.26602912e-01
5.80495059e-01 9.93572712e-01 7.19563603e-01 -2.77537942e-01
-1.91502944e-01 -3.16199839e-01 -7.97084346e-02 1.22444201e+00
-3.95140409e-01 2.71028399e-01 -1.14230096e+00 2.24982262e-01
-2.01011682e+00 -4.12933171e-01 -5.95662594e-01 2.26814437e+00
6.43275738e-01 9.70727652e-02 -1.69845998e-01 -2.82863647e-01
9.23049569e-01 2.00771675e-01 -3.17057043e-01 5.38815141e-01
1.92224216e-02 2.75786787e-01 6.87901139e-01 6.88520133e-01
-1.17682195e+00 1.32400525e+00 5.24228430e+00 7.15682983e-01
-8.23389471e-01 -6.68399110e-02 2.26356134e-01 2.73510486e-01
4.12635272e-03 3.00631016e-01 -8.70227814e-01 -7.10786879e-02
7.97459185e-02 7.42607266e-02 2.50528514e-01 7.49501765e-01
1.09418616e-01 -5.48936166e-02 -1.21201468e+00 1.13028550e+00
1.14474744e-01 -1.49203086e+00 3.11042797e-02 5.17755687e-01
8.91543210e-01 4.52773035e-01 -3.62678528e-01 -1.72777429e-01
2.53479183e-01 -8.13371480e-01 3.84178519e-01 4.76418912e-01
6.87114179e-01 -7.58318841e-01 6.70337796e-01 3.93524975e-01
-1.64101565e+00 6.09776258e-01 -6.77033484e-01 3.48861873e-01
1.02797873e-01 5.26150644e-01 -1.05416179e+00 8.68673623e-01
6.42378211e-01 6.66288793e-01 -4.95362937e-01 1.23473227e+00
-1.78872734e-01 1.51222153e-02 -6.81117535e-01 2.13868260e-01
1.66772410e-01 -5.77227950e-01 8.68891835e-01 8.41397583e-01
3.86307448e-01 3.06872606e-01 6.56981051e-01 7.46670544e-01
-2.83535600e-01 8.51469934e-02 -4.84034330e-01 7.28872418e-01
9.91584778e-01 1.32289934e+00 -1.10110569e+00 8.41544047e-02
-2.43606627e-01 9.29303944e-01 3.78444791e-01 3.70980501e-02
-7.22721696e-01 1.72434837e-01 8.27164710e-01 2.25620061e-01
8.06771591e-02 -6.14707112e-01 -1.71359152e-01 -9.32309806e-01
1.82968192e-02 -8.04323256e-01 3.55298698e-01 -5.16976118e-01
-1.27866232e+00 4.87157077e-01 4.25276831e-02 -1.63370633e+00
5.85107133e-02 -2.16616258e-01 -6.42879486e-01 7.74348378e-01
-1.20073235e+00 -1.36724174e+00 -8.29837024e-01 7.88425326e-01
2.20169812e-01 5.15769050e-02 6.25815094e-01 3.07562172e-01
-9.61195156e-02 3.14179808e-01 -3.37115079e-01 3.70487988e-01
6.16110325e-01 -9.12845194e-01 6.89936996e-01 7.06921101e-01
4.72406626e-01 7.50662863e-01 2.92595476e-01 -1.16493666e+00
-1.78221297e+00 -1.05592668e+00 9.20230508e-01 -3.48498881e-01
2.65786082e-01 -3.35529208e-01 -8.00228000e-01 6.24571741e-01
-7.01989233e-02 9.67353582e-04 2.86407828e-01 1.01531073e-01
-2.56747961e-01 -1.36057019e-01 -1.03691411e+00 5.88850200e-01
1.33191609e+00 -2.74700731e-01 -5.81204295e-01 6.48226500e-01
7.80931771e-01 -1.17914748e+00 -1.03505874e+00 4.93173033e-01
2.87474543e-01 -5.42030990e-01 1.24000406e+00 -1.48407936e-01
-1.10955566e-01 -7.96021879e-01 -5.33905812e-02 -1.14074588e+00
-5.58748782e-01 -7.03202665e-01 8.79846737e-02 8.04748416e-01
1.95476994e-01 -4.16574359e-01 1.11612546e+00 2.96155035e-01
-3.15901756e-01 -6.75042272e-01 -1.38079357e+00 -6.30739033e-01
-3.03560823e-01 -2.50200540e-01 8.62695515e-01 1.24990618e+00
-4.54386473e-01 2.93838352e-01 -1.43841162e-01 8.70099664e-01
1.13329172e+00 2.77685791e-01 1.33468521e+00 -1.64691675e+00
1.25571817e-01 -1.87686831e-01 -7.16623127e-01 -1.25271535e+00
1.47479579e-01 -1.09307909e+00 -5.50252832e-02 -1.80128336e+00
5.31552993e-02 -8.15880656e-01 9.39329341e-02 4.91969377e-01
1.28576815e-01 4.04266417e-01 3.12969118e-01 7.12083340e-01
-6.57789409e-01 2.73241729e-01 1.47013521e+00 -3.50088701e-02
-4.37226027e-01 1.82816405e-02 -3.35871339e-01 6.28047466e-01
5.10704160e-01 -5.60177088e-01 -7.52431974e-02 -3.88985187e-01
9.11752209e-02 1.54983893e-01 4.25951928e-01 -1.08282173e+00
6.36369169e-01 -1.33447841e-01 4.00813133e-01 -1.11986709e+00
4.83525246e-01 -1.02124512e+00 7.00167060e-01 5.85674047e-01
3.65057796e-01 1.06288381e-01 -8.36914405e-02 5.83606362e-01
-7.17777200e-03 1.13231950e-01 7.53231525e-01 -3.10296342e-02
-7.45054066e-01 8.71244431e-01 3.14517975e-01 -2.14085728e-01
1.11291897e+00 -4.72021610e-01 -1.06034629e-01 -2.10353911e-01
-6.69948220e-01 4.93894935e-01 6.09793544e-01 4.99095768e-01
9.84660268e-01 -1.60720944e+00 -9.28545654e-01 1.10085919e-01
1.56009167e-01 7.16768026e-01 -1.60257250e-01 8.22252691e-01
-8.10819685e-01 -1.19716689e-01 -5.60477097e-03 -1.37204695e+00
-1.53187466e+00 -4.96763103e-02 4.23376858e-01 -1.90083298e-03
-8.36132467e-01 8.06262255e-01 -6.35508075e-02 -5.62084079e-01
1.25373051e-01 -2.36551821e-01 -4.34377901e-02 -1.21079370e-01
-1.07559264e-01 2.70449311e-01 2.83599079e-01 -1.05815423e+00
-8.42986047e-01 1.23975575e+00 1.08842580e-02 1.44691557e-01
1.37055564e+00 8.71768892e-02 -4.59652483e-01 -1.84945881e-01
1.23424435e+00 7.18574077e-02 -1.02072048e+00 -4.00511622e-01
-6.94796368e-02 -6.64554000e-01 2.36061029e-03 -1.59210488e-01
-1.35746789e+00 1.67336285e-01 5.96175969e-01 -2.22220331e-01
7.67738760e-01 4.84599531e-01 7.89527535e-01 4.49493110e-01
8.87187958e-01 -8.68701220e-01 7.07861111e-02 6.03845656e-01
9.02928114e-01 -1.02909148e+00 5.46440840e-01 -8.88921380e-01
-2.83230871e-01 1.06871164e+00 7.67180681e-01 -5.43270707e-01
7.60382831e-01 1.34092689e-01 1.16007023e-01 -8.11288774e-01
-4.48860377e-02 -1.61904991e-01 6.47585094e-01 4.90389049e-01
-2.25509629e-02 3.57790552e-02 -2.34278932e-01 -4.41266224e-02
-4.75192666e-01 -3.78664255e-01 1.50970370e-01 8.08414996e-01
-5.55418670e-01 -1.22042429e+00 -7.21494138e-01 3.74832541e-01
1.85335472e-01 2.14963481e-01 -6.71199441e-01 1.00601161e+00
1.41009673e-01 7.20513403e-01 2.29298979e-01 -6.83993280e-01
5.45631707e-01 -3.47486794e-01 5.64427912e-01 -6.33646131e-01
-4.85432953e-01 5.90287447e-01 -6.50064722e-02 -7.42227733e-01
-5.53845704e-01 -8.06359470e-01 -1.53810942e+00 -5.21469235e-01
-6.05539441e-01 5.50597571e-02 5.08471429e-01 8.91808450e-01
5.66524684e-01 1.38560617e-02 7.53228307e-01 -1.14383626e+00
-1.39743671e-01 -6.61576331e-01 -3.15448254e-01 3.49148691e-01
-2.03890596e-02 -8.58133137e-01 -1.36820421e-01 -1.77281097e-01] | [7.64498233795166, -3.028937816619873] |
a1deb9f1-7c04-43d0-a9df-cea8aef5ecca | a-station-data-based-model-residual-machine | null | null | https://doi.org/10.1007/ | https://doi.org/10.1007/ | A station-data-based model residual machine learning method for fine-grained meteorological grid prediction | Fine-grained weather forecasting data, i.e., the grid data with high-resolution,
have attracted increasing attention in recent years, especially for some specific applications
such as the Winter Olympic Games. Although European Centre for Medium-Range
Weather Forecasts (ECMWF) provides grid prediction up to 240 hours, the coarse data
are unable to meet high requirements of these major events. In this paper, we propose
a method, called model residual machine learning (MRML), to generate grid prediction
with high-resolution based on high-precision stations forecasting. MRML applies model
output machine learning (MOML) for stations forecasting. Subsequently, MRML utilizes
these forecasts to improve the quality of the grid data by fitting a machine learning (ML)
model to the residuals. We demonstrate that MRML achieves high capability at diverse
meteorological elements, specifically, temperature, relative humidity, and wind speed. In
addition, MRML could be easily extended to other post-processing methods by invoking
different techniques. In our experiments, MRML outperforms the traditional downscaling
methods such as piecewise linear interpolation (PLI) on the testing data. | ['†', 'Pingwen ZHANG1', 'Jiangjiang XIA3', 'Chen YU1', 'Haochen LI2', 'Chuansai ZHOU1'] | 2021-12-16 | a-station-data-based-model-residual-machine-1 | https://doi.org/ | https://paperswithcode.com/ | appl-math-mech-engl-ed-43-2-155-166-2022-2021 | ['machine-learning', 'weather-forecasting', 'machine-learning'] | ['methodology', 'miscellaneous', 'miscellaneous'] | [-3.93243521e-01 -4.52355921e-01 1.79020599e-01 -5.23279488e-01
-7.91500866e-01 -2.31258869e-01 7.34510005e-01 1.34309873e-01
-1.32301703e-01 1.39064050e+00 2.69500762e-01 -6.79979503e-01
6.00927100e-02 -1.28479218e+00 -3.91592711e-01 -8.72075260e-01
-2.47021973e-01 1.60828218e-01 1.83492433e-02 -6.44593596e-01
2.03240544e-01 6.01362765e-01 -1.50909460e+00 2.07961470e-01
1.38795996e+00 8.21464181e-01 2.83064336e-01 6.44045234e-01
-1.33879662e-01 6.46396399e-01 -5.07382572e-01 2.95335919e-01
2.14518249e-01 -3.03575367e-01 -1.68469056e-01 -3.77886683e-01
4.84659299e-02 -3.12010139e-01 2.33199105e-01 7.00078309e-01
3.60335708e-01 5.71450591e-01 5.29708624e-01 -1.11105442e+00
-2.26492330e-01 2.64083177e-01 -6.60700917e-01 1.77548707e-01
-4.36850861e-02 -1.29813492e-01 4.93404150e-01 -1.05290198e+00
-5.01183746e-03 1.14958405e+00 9.63085473e-01 8.63073543e-02
-1.28731489e+00 -7.77186215e-01 6.85318187e-02 -1.44143729e-02
-1.46584082e+00 -3.23574960e-01 5.03012002e-01 -7.33691692e-01
9.10200655e-01 5.60683310e-01 4.34977502e-01 3.76016378e-01
6.39017105e-01 1.10206511e-02 1.58862603e+00 -3.61722589e-01
3.34913164e-01 5.63481152e-02 1.64483432e-02 6.27114996e-02
1.36615723e-01 6.36578143e-01 -3.01431924e-01 -4.29295868e-01
8.96121681e-01 1.82473704e-01 -2.45137930e-01 5.16905308e-01
-9.73144591e-01 9.73773956e-01 3.71729076e-01 1.20712370e-01
-5.69917440e-01 -3.12997729e-01 1.95020869e-01 3.28054845e-01
1.29817498e+00 2.88821012e-01 -8.94543290e-01 3.93952467e-02
-1.34800518e+00 5.91589987e-01 6.15977108e-01 5.58807194e-01
9.23972070e-01 6.61308706e-01 1.63996860e-01 6.94855452e-01
2.23399177e-01 8.11204076e-01 4.74008292e-01 -7.23407269e-01
3.47238541e-01 4.98701036e-02 6.90639138e-01 -1.00439453e+00
-8.24035764e-01 -4.20850158e-01 -1.65008223e+00 4.60239738e-01
1.85469449e-01 -6.80389166e-01 -4.78728622e-01 1.22289824e+00
3.89577866e-01 7.06458986e-01 2.68502802e-01 1.00692844e+00
6.38352513e-01 1.59657812e+00 2.29272917e-01 -5.06667674e-01
8.42129290e-01 -8.89533043e-01 -7.38781929e-01 1.09257460e-01
7.59557486e-01 -7.38156736e-01 9.43430126e-01 2.94234693e-01
-6.97631598e-01 -1.06753647e+00 -6.03120089e-01 3.34562600e-01
-5.55308938e-01 -5.63948415e-03 5.12961328e-01 2.51271158e-01
-1.05372977e+00 7.99792171e-01 -8.85499358e-01 6.38664365e-02
-5.26216090e-01 -1.14320971e-01 -2.25431517e-01 3.17904949e-01
-1.60632193e+00 9.01916504e-01 3.72632146e-01 5.64060271e-01
-3.28035504e-01 -1.03019392e+00 -1.09836686e+00 1.89567938e-01
-4.32032347e-01 -3.12320352e-01 9.06521618e-01 -5.65209150e-01
-1.54459071e+00 4.12942730e-02 -5.41595578e-01 -3.79121006e-01
2.19133064e-01 -1.85010791e-01 -1.01286757e+00 -5.33993065e-01
1.07162327e-01 2.02164397e-01 6.07615471e-01 -1.17259300e+00
-9.91479218e-01 -1.80632383e-01 -3.78754675e-01 3.32908542e-03
7.48283714e-02 1.42713739e-02 2.26432562e-01 -8.98076713e-01
4.89157811e-02 -7.64869094e-01 -7.69371986e-01 -7.06555307e-01
2.84333620e-02 -1.56883627e-01 6.03976369e-01 -1.18177927e+00
1.39731514e+00 -2.01969337e+00 -2.03013197e-01 3.78665000e-01
-3.59438092e-01 8.79385248e-02 6.47277460e-02 5.87998092e-01
-1.21371351e-01 1.53007984e-01 -4.25695360e-01 -3.52281004e-01
-2.40572274e-01 5.07168055e-01 -9.85530198e-01 4.51006114e-01
1.06146134e-01 4.01834279e-01 -6.10729396e-01 -1.27134085e-01
4.60206687e-01 4.82263476e-01 -2.96333194e-01 4.21717346e-01
-1.17253758e-01 1.11554193e+00 -1.37669370e-01 3.15210313e-01
1.16426790e+00 -9.80620906e-02 -1.58322766e-01 1.74376637e-01
-8.08689117e-01 -6.11997023e-02 -1.24025083e+00 1.05106997e+00
-7.27767467e-01 3.65810335e-01 -1.76193379e-02 -7.02512920e-01
1.34688199e+00 3.75652939e-01 2.34101146e-01 -7.77910531e-01
-6.97151959e-01 3.20876181e-01 -3.73044044e-01 -3.62239510e-01
8.22579682e-01 -2.89475560e-01 -1.99834853e-02 1.86795682e-01
-6.74809456e-01 -1.78230017e-01 2.69025993e-02 -2.26497695e-01
-2.11913511e-02 3.89010608e-01 4.37158585e-01 -5.73585212e-01
5.29878974e-01 2.32029572e-01 9.62187171e-01 6.01747870e-01
3.21858168e-01 5.20288348e-01 7.85712898e-02 -9.13452804e-01
-1.12242639e+00 -6.16064489e-01 -5.97061574e-01 1.17979681e+00
-2.83769339e-01 -2.27906808e-01 -3.28212887e-01 2.51403963e-03
2.44519368e-01 8.12630534e-01 -3.91928464e-01 4.92936879e-01
-4.93846297e-01 -1.29195988e+00 2.84541875e-01 3.81404638e-01
5.35640478e-01 -1.02998173e+00 -9.75915715e-02 4.20162678e-01
-1.54193223e-01 -8.09793770e-01 4.68357019e-02 -6.53986633e-02
-1.08955824e+00 -3.12140048e-01 -5.99083841e-01 -3.37482512e-01
4.36151296e-01 5.45902140e-02 1.10762990e+00 -5.24269342e-02
3.45215499e-01 -3.93662959e-01 -4.08389986e-01 -2.71156579e-01
-2.36690268e-01 3.45385522e-02 2.61556566e-01 4.88871336e-02
2.00689249e-02 -6.10501826e-01 -3.18314463e-01 3.28764558e-01
-8.66721809e-01 3.31130326e-01 1.51495978e-01 8.91935945e-01
7.05004275e-01 4.68019783e-01 8.99960935e-01 -8.48779261e-01
5.37297904e-01 -8.08042049e-01 -1.11889064e+00 3.87208946e-02
-7.30198741e-01 -1.69436127e-01 1.07160234e+00 -2.22933549e-03
-1.44260168e+00 -2.30763242e-01 -5.15461326e-01 5.22714630e-02
-4.69556451e-01 1.11633384e+00 2.22213611e-01 -9.34517905e-02
5.40553272e-01 4.02332544e-01 -4.79179472e-01 -9.72901762e-01
8.54638144e-02 7.41708159e-01 6.11737251e-01 -4.93680775e-01
9.71698105e-01 2.63780564e-01 1.44652218e-01 -9.78210092e-01
-8.69588912e-01 -3.11462164e-01 -4.55531120e-01 2.37182677e-02
6.05178475e-01 -1.36133587e+00 -4.01962817e-01 6.27350807e-01
-8.94450605e-01 -6.26809478e-01 9.23869833e-02 8.55167150e-01
-1.28799900e-01 1.80115744e-01 -6.56338871e-01 -1.03591681e+00
-3.96589935e-01 -7.48794377e-01 8.76799762e-01 3.71449560e-01
6.08858606e-03 -1.32683015e+00 5.46449661e-01 -2.73817450e-01
8.91636550e-01 6.20859385e-01 7.61875749e-01 -1.99568167e-01
-3.43082845e-02 -7.88847059e-02 -1.68442711e-01 1.59285277e-01
2.14394435e-01 2.08520457e-01 -8.80337894e-01 -4.18106198e-01
1.59283839e-02 -3.78835313e-02 7.71751285e-01 6.64751768e-01
1.24644160e+00 -4.28050548e-01 -6.10355809e-02 9.98169601e-01
1.45086598e+00 9.08037573e-02 5.41530073e-01 5.15471816e-01
5.50959051e-01 4.82834339e-01 8.34971130e-01 9.28915083e-01
4.95246768e-01 5.12701154e-01 1.01712182e-01 -6.56242549e-01
6.41480565e-01 -6.44726679e-02 9.57986787e-02 9.21249807e-01
-5.61894178e-01 2.25980431e-01 -1.10904908e+00 1.86189115e-01
-1.96198285e+00 -1.00742698e+00 -6.57186568e-01 2.32561874e+00
7.86894262e-01 -9.69535187e-02 -3.16352725e-01 -8.85626115e-03
4.40491319e-01 3.94964099e-01 -3.87963206e-01 -5.50324380e-01
-1.87263459e-01 3.69285256e-01 6.68232560e-01 9.08384740e-01
-1.18791819e+00 9.66174781e-01 6.03836536e+00 5.93951702e-01
-1.25471115e+00 1.40266180e-01 7.34400392e-01 4.70620930e-01
-2.98079699e-01 4.54531126e-02 -1.23738539e+00 6.49366617e-01
1.36527622e+00 -2.13121533e-01 3.81119013e-01 8.56459856e-01
1.09804213e+00 -6.91623911e-02 -2.81367272e-01 6.77020371e-01
-5.19547582e-01 -1.54202688e+00 -1.33125186e-01 -8.34319293e-02
1.16541994e+00 8.24258551e-02 -2.55922914e-01 7.65711665e-01
3.09461713e-01 -1.06771421e+00 1.14742331e-01 9.68253613e-01
8.88897955e-01 -9.44517910e-01 1.00212920e+00 7.76177704e-01
-1.44382727e+00 1.56389132e-01 -8.36293459e-01 -7.60781884e-01
2.64843374e-01 1.15272248e+00 -2.26257086e-01 8.79162490e-01
8.77565265e-01 7.60532975e-01 -1.66740999e-01 9.89761412e-01
-1.20833218e-01 1.01899910e+00 -3.42168629e-01 5.38433313e-01
2.45115444e-01 -5.46746850e-01 6.48355857e-02 1.23522937e+00
7.80148327e-01 4.62465972e-01 5.40221035e-01 2.49637589e-01
5.16256869e-01 3.39779258e-01 -4.03195381e-01 6.60919189e-01
3.57520580e-01 1.24720943e+00 -9.10623521e-02 -5.38692832e-01
-4.42625910e-01 4.20809686e-01 2.01854426e-02 7.88524568e-01
-6.59351051e-01 -2.67616928e-01 8.90644073e-01 5.76786548e-02
-5.74304201e-02 -5.99790931e-01 -4.82072264e-01 -1.39774024e+00
-3.71485233e-01 -7.81395376e-01 2.99792081e-01 -1.00987291e+00
-1.27955973e+00 7.57718623e-01 -5.08718491e-02 -1.40950930e+00
-5.99414408e-01 -2.19512671e-01 -9.28728819e-01 1.82490742e+00
-2.15778947e+00 -9.66152191e-01 -5.63842773e-01 5.72531223e-01
4.58795369e-01 1.69991627e-02 1.21768951e+00 1.35794446e-01
-5.34374535e-01 -1.11828139e-02 9.12090182e-01 -7.55666867e-02
8.35540652e-01 -1.20973575e+00 6.95900857e-01 8.62377822e-01
-3.97857219e-01 2.35045418e-01 7.99189210e-01 -7.72123039e-01
-9.44590330e-01 -1.69617522e+00 1.32425034e+00 2.84566045e-01
4.98047233e-01 -4.49709361e-03 -1.34774172e+00 7.23247349e-01
7.12185130e-02 2.84215957e-01 6.73242807e-01 1.79000899e-01
1.16145656e-01 -5.20022154e-01 -9.85501528e-01 3.01986873e-01
-1.58550348e-02 -1.64253339e-01 -3.43467385e-01 4.26695913e-01
5.96458793e-01 -5.99732041e-01 -1.27885950e+00 8.03681493e-01
3.86393875e-01 -9.19686615e-01 5.75581849e-01 -7.57843971e-01
2.81964719e-01 -6.61489785e-01 -2.59654492e-01 -1.90747619e+00
-7.42676795e-01 -4.36301649e-01 5.30180968e-02 1.03820455e+00
4.16842997e-01 -9.68431473e-01 3.49028498e-01 7.34125018e-01
-1.65200546e-01 -3.47866178e-01 -9.50652063e-01 -6.25381827e-01
3.63229275e-01 -4.26144838e-01 1.07647002e+00 1.26228964e+00
-2.16200426e-01 -2.81119704e-01 -1.02617228e+00 9.07713890e-01
5.63840508e-01 6.49200678e-01 9.12538230e-01 -1.46498036e+00
-2.68925786e-01 -2.18442604e-02 1.83566570e-01 -8.20922852e-01
1.45753682e-01 -3.97146821e-01 -4.57560457e-02 -1.37547207e+00
-4.46810663e-01 -7.14292824e-01 -4.08343196e-01 4.90008146e-01
-4.11043972e-01 2.30895579e-01 -1.25921711e-01 3.97612393e-01
2.16850966e-01 6.33387804e-01 9.76156116e-01 2.78688759e-01
-5.34530640e-01 2.75319904e-01 -9.38524157e-02 6.89141631e-01
1.15019476e+00 -3.54179829e-01 1.42445937e-01 -3.91933382e-01
2.43484959e-01 5.95240057e-01 5.12762107e-02 -1.08868992e+00
2.23279418e-03 -7.55320728e-01 8.01434457e-01 -8.41136873e-01
-5.14594000e-03 -5.66837430e-01 7.02200472e-01 2.79956907e-01
-8.35802332e-02 3.24099451e-01 5.48349023e-01 9.93011519e-02
-6.40772283e-01 1.98144436e-01 6.96983635e-01 -4.88132164e-02
-8.81134093e-01 4.21688020e-01 -5.14894187e-01 -2.86084145e-01
6.66068792e-01 2.48702109e-01 -3.25798720e-01 -3.60531390e-01
-8.47560883e-01 5.63165367e-01 2.32607976e-01 1.22840717e-01
3.48848313e-01 -1.24936783e+00 -1.32591176e+00 7.26963580e-01
-1.45148188e-01 5.49268872e-02 5.17520547e-01 7.48810470e-01
-5.66590488e-01 4.08726931e-01 -5.67395128e-02 -4.70552772e-01
-8.06326449e-01 2.94702232e-01 3.55026573e-01 -5.71980953e-01
-7.42131889e-01 2.59444088e-01 1.16357796e-01 -8.36555064e-01
-3.68082851e-01 -5.33748984e-01 -3.66561949e-01 2.77534639e-03
9.78296816e-01 3.78139317e-01 1.51201338e-01 -5.83035290e-01
5.97337112e-02 5.73259354e-01 5.18975794e-01 -1.87138822e-02
1.32113957e+00 -2.29455009e-01 -2.37228960e-01 6.57344103e-01
7.63798952e-01 1.72911227e-01 -1.40544450e+00 -2.23445371e-01
-8.92291069e-02 -5.50182700e-01 3.59987497e-01 -5.40492356e-01
-9.91714001e-01 9.04014885e-01 3.84541988e-01 4.08970952e-01
1.20539737e+00 -7.50162661e-01 8.24447632e-01 3.91788065e-01
3.23788971e-01 -8.29976439e-01 -1.09870780e+00 7.78340459e-01
8.16588283e-01 -1.40253258e+00 -1.12622090e-01 1.01655602e-01
-6.17470980e-01 1.21315801e+00 3.49249721e-01 -1.53701782e-01
1.05757511e+00 4.89030868e-01 3.13094646e-01 5.20212114e-01
-8.30098093e-01 8.06493871e-03 1.21267997e-01 3.33651274e-01
5.85558295e-01 4.60149318e-01 -2.85299331e-01 4.80947137e-01
-3.38974595e-01 1.58845827e-01 4.83401179e-01 5.34981847e-01
-7.18872070e-01 -7.81819105e-01 -9.43504751e-01 5.54532409e-01
-2.96702683e-01 -5.27635813e-01 7.10715353e-01 5.28003871e-01
2.28808820e-02 1.04956508e+00 3.73445004e-01 -1.21235743e-01
9.58611518e-02 1.63806621e-02 -4.84157413e-01 -3.12881500e-01
-4.04440612e-01 2.21490875e-01 -1.93047635e-02 -3.18157017e-01
-3.26803476e-01 -5.26606262e-01 -1.13485897e+00 -8.77642095e-01
-5.38240038e-02 8.70160222e-01 6.82059586e-01 1.00817490e+00
1.88932329e-01 2.86757082e-01 1.09426296e+00 -1.35578060e+00
-3.59845877e-01 -1.20182383e+00 -1.09316158e+00 4.43895869e-02
5.36168873e-01 -3.77326280e-01 -5.11833429e-01 1.00981303e-01] | [6.596577167510986, 2.9866318702697754] |
d9c1d6f2-8fff-4dbf-b1a6-b5a99882d2cf | reducing-annotation-need-in-self-explanatory | 2206.13608 | null | https://arxiv.org/abs/2206.13608v2 | https://arxiv.org/pdf/2206.13608v2.pdf | Reducing Annotation Need in Self-Explanatory Models for Lung Nodule Diagnosis | Feature-based self-explanatory methods explain their classification in terms of human-understandable features. In the medical imaging community, this semantic matching of clinical knowledge adds significantly to the trustworthiness of the AI. However, the cost of additional annotation of features remains a pressing issue. We address this problem by proposing cRedAnno, a data-/annotation-efficient self-explanatory approach for lung nodule diagnosis. cRedAnno considerably reduces the annotation need by introducing self-supervised contrastive learning to alleviate the burden of learning most parameters from annotation, replacing end-to-end training with two-stage training. When training with hundreds of nodule samples and only 1% of their annotations, cRedAnno achieves competitive accuracy in predicting malignancy, meanwhile significantly surpassing most previous works in predicting nodule attributes. Visualisation of the learned space further indicates that the correlation between the clustering of malignancy and nodule attributes coincides with clinical knowledge. Our complete code is open-source available: https://github.com/diku-dk/credanno. | ['Sune Darkner', 'Michael Bachmann Nielsen', 'Kenny Erleben', 'Oswin Krause', 'CHONG YIN', 'Jiahao Lu'] | 2022-06-27 | null | null | null | null | ['clinical-knowledge'] | ['miscellaneous'] | [ 4.68196943e-02 8.90568912e-01 -5.76099336e-01 -4.49255198e-01
-9.70727324e-01 -4.08904612e-01 4.06390578e-01 3.38349581e-01
-4.93313111e-02 6.99567676e-01 3.77079368e-01 -3.14825773e-01
-3.91200602e-01 -4.63875115e-01 -3.52748871e-01 -7.07792580e-01
2.90540382e-02 8.59791160e-01 1.23444267e-01 2.53707469e-01
-1.67909294e-01 3.97426588e-03 -1.30695069e+00 4.90556955e-01
8.70606840e-01 1.06350183e+00 1.48935989e-01 6.45895779e-01
1.15184426e-01 1.05904377e+00 -4.81850393e-02 -5.88626623e-01
-1.67887937e-02 -5.94136417e-01 -1.17955494e+00 1.53530285e-01
8.27822760e-02 -9.45037603e-02 -1.95516005e-01 6.42130017e-01
3.07426125e-01 -5.63339412e-01 7.89681077e-01 -1.28489065e+00
-5.96896350e-01 7.05806494e-01 -9.38451514e-02 -2.17867680e-02
-7.68322572e-02 2.89494004e-02 1.12778854e+00 -8.08423758e-01
6.72887564e-01 5.39595366e-01 9.72094417e-01 7.86479175e-01
-1.00759006e+00 -3.61829728e-01 -4.42451417e-01 9.24684703e-02
-1.37111199e+00 -4.28041160e-01 3.89967531e-01 -5.47899604e-01
2.57188588e-01 6.05425715e-01 8.29645574e-01 9.04391587e-01
-2.06221081e-02 7.58355916e-01 8.67364109e-01 -3.77902776e-01
2.19519302e-01 5.68295717e-01 -1.98096167e-02 1.20046663e+00
5.85071981e-01 -7.23135993e-02 -4.10505533e-01 -5.18174291e-01
6.98380113e-01 2.15139419e-01 -1.42505333e-01 -6.48371279e-01
-1.36748922e+00 7.84768462e-01 6.35259211e-01 4.24563974e-01
-2.45640963e-01 2.66933322e-01 4.08165187e-01 -2.61552241e-02
5.95823884e-01 6.64852023e-01 -5.43008029e-01 4.68464755e-02
-8.33444297e-01 -3.04107785e-01 8.74001563e-01 8.94845724e-01
5.24295866e-01 -3.09692949e-01 4.67873514e-02 5.38891613e-01
1.14803962e-01 1.20224394e-01 8.34606588e-01 -9.70802248e-01
-4.02307391e-01 8.62096548e-01 -1.69643059e-01 -4.82548505e-01
-4.85548735e-01 -8.95508409e-01 -9.85043228e-01 -1.76498681e-01
3.74397457e-01 4.00410173e-03 -6.77834272e-01 1.24197817e+00
3.68269712e-01 1.55290887e-01 1.72929242e-02 7.15063572e-01
1.01243603e+00 -1.95533082e-01 2.74480611e-01 -5.39512634e-02
1.65415955e+00 -1.02935421e+00 -6.04925692e-01 -9.96162519e-02
1.01850498e+00 -5.13598561e-01 8.86414826e-01 -5.83447376e-03
-7.83879936e-01 -2.24470764e-01 -6.10015213e-01 7.46782646e-02
-2.88527310e-02 5.04974961e-01 1.07267261e+00 3.75964105e-01
-8.73474896e-01 5.81277549e-01 -1.18853641e+00 -4.91102844e-01
9.87494290e-01 4.85613555e-01 -6.96461499e-01 7.99440295e-02
-7.65123129e-01 7.77570426e-01 4.54803407e-01 -1.50012583e-01
-6.52878642e-01 -1.14507234e+00 -6.93045795e-01 1.00840449e-01
6.71510220e-01 -1.12726939e+00 1.35614777e+00 -9.70717013e-01
-8.25840056e-01 1.10534894e+00 -3.13717455e-01 -4.71735030e-01
7.71103501e-01 -4.59578745e-02 -1.00284629e-01 3.60886991e-01
2.92015076e-01 7.30684519e-01 7.83782184e-01 -1.21514404e+00
-5.24900854e-01 -1.99434698e-01 -3.73680413e-01 3.03079542e-02
-5.19943953e-01 -5.42092979e-01 -2.92979896e-01 -3.20424885e-01
2.37696707e-01 -1.09071255e+00 -5.92855930e-01 4.64258224e-01
-5.94931066e-01 -2.18847811e-01 6.42740488e-01 -3.41856927e-01
9.12708938e-01 -2.18311954e+00 -3.81715655e-01 4.26738039e-02
9.56683338e-01 9.28921998e-02 2.87097454e-01 2.69489259e-01
-1.45159557e-01 3.11160386e-01 -3.48135829e-01 -2.26675957e-01
-2.53867954e-01 3.07189614e-01 2.83971369e-01 5.54472089e-01
5.30222118e-01 1.31260490e+00 -1.07826042e+00 -1.06158185e+00
2.52131045e-01 4.17696476e-01 -5.10704696e-01 4.50320877e-02
-9.11513716e-02 7.56506503e-01 -7.03116953e-01 6.49591684e-01
1.40627608e-01 -1.06834495e+00 2.69604087e-01 -1.61359459e-01
3.31668198e-01 1.58580109e-01 -6.26397312e-01 1.56102705e+00
-2.48709410e-01 5.52500963e-01 -3.66950572e-01 -8.10897231e-01
7.08380997e-01 5.92694521e-01 8.72849524e-01 -2.76983473e-02
1.80789441e-01 5.65251410e-01 9.92602706e-02 -6.34898663e-01
-7.24833608e-02 -2.52845794e-01 2.14994639e-01 3.25001180e-01
9.42247063e-02 -1.70783728e-01 -1.73900321e-01 3.15795183e-01
1.37600160e+00 -5.56379892e-02 9.44703758e-01 -4.82072085e-01
3.65702212e-01 5.45343518e-01 3.35341781e-01 5.51110029e-01
-1.46874934e-01 6.68448925e-01 5.54282069e-01 -6.26122236e-01
-1.03700471e+00 -6.26381993e-01 -4.29337800e-01 7.06301570e-01
-1.29620880e-01 -4.86914873e-01 -6.99730754e-01 -1.01760244e+00
1.82863027e-01 5.43662548e-01 -1.20352018e+00 -1.87535375e-01
-1.35522231e-01 -5.53435504e-01 3.22309881e-01 5.97493768e-01
2.07903847e-01 -7.79799938e-01 -6.88173234e-01 1.67756714e-02
-3.06603521e-01 -9.84509170e-01 -2.57140726e-01 5.45633197e-01
-9.97931600e-01 -1.41453671e+00 -5.34660637e-01 -7.05149829e-01
1.18956399e+00 1.04308635e-01 1.25453591e+00 6.32502139e-01
-8.61663163e-01 2.84368336e-01 -4.06193942e-01 -5.70043862e-01
-7.53958046e-01 4.37222242e-01 -4.17247355e-01 -2.23051533e-01
4.22906011e-01 -3.16586256e-01 -8.42684329e-01 4.46264595e-01
-6.64619863e-01 3.48750561e-01 9.00840878e-01 1.04363513e+00
7.90709436e-01 -1.83806702e-01 4.91176546e-01 -1.43172860e+00
-2.71213781e-02 -6.50984645e-01 -9.34680626e-02 8.16004053e-02
-8.68552029e-01 1.22664668e-01 4.20503646e-01 -2.57551432e-01
-7.03882754e-01 6.21286869e-01 1.35803983e-01 -3.34387213e-01
-1.98336869e-01 4.48392719e-01 3.80369067e-01 3.71174589e-02
9.21729982e-01 4.81170714e-02 4.08051819e-01 -1.96712762e-01
2.56275028e-01 6.41761482e-01 6.27837002e-01 5.23924157e-02
9.04923916e-01 6.49603665e-01 1.87284186e-01 -4.45850730e-01
-1.50168729e+00 -6.74144745e-01 -8.83837402e-01 1.17832705e-01
7.19287992e-01 -8.26900423e-01 -6.80895925e-01 -3.01106244e-01
-5.75770974e-01 -9.04877484e-02 -9.23290491e-01 6.65167034e-01
-8.45644414e-01 1.44246414e-01 -2.99596578e-01 -4.33338672e-01
-3.81309450e-01 -8.66241753e-01 1.12651920e+00 -8.20717439e-02
-6.87022686e-01 -1.12572348e+00 -1.38743162e-01 5.32673359e-01
3.96123260e-01 2.32019961e-01 9.65271473e-01 -1.20886838e+00
-4.99722004e-01 -6.33540869e-01 -2.57764071e-01 3.97893451e-02
2.89931059e-01 -3.09928060e-01 -1.17025387e+00 6.57085190e-03
2.31093857e-02 -3.84928674e-01 7.03552544e-01 4.99824345e-01
1.33749318e+00 -4.79839027e-01 -7.14390278e-01 5.24434865e-01
1.26664758e+00 -4.42669630e-01 1.58466876e-01 2.31064677e-01
5.19770443e-01 5.88241100e-01 6.17912710e-01 4.66301292e-01
3.42443645e-01 2.36968532e-01 6.89499915e-01 -5.29794455e-01
-3.14042121e-01 -3.53810340e-01 -4.90893602e-01 5.64433396e-01
-1.80541098e-01 3.55758071e-02 -1.19126737e+00 5.71773231e-01
-1.93718088e+00 -6.72770619e-01 -4.08406526e-01 1.96157229e+00
8.55216205e-01 -1.17013626e-01 -1.17858229e-02 8.71226564e-02
4.62171078e-01 -4.52560008e-01 -6.17478430e-01 3.11545640e-01
1.18060537e-01 -1.67370185e-01 5.94614923e-01 3.08781236e-01
-1.16881895e+00 6.95086837e-01 6.17515707e+00 7.30879128e-01
-8.46820772e-01 3.19449812e-01 1.00974774e+00 1.22007310e-01
-3.10858101e-01 4.69264276e-02 -3.51845205e-01 2.04619139e-01
8.48012686e-01 -3.01014751e-01 -1.47678271e-01 1.22205997e+00
1.12422347e-01 -1.87609911e-01 -1.39236307e+00 7.42647767e-01
-4.31977101e-02 -1.56447756e+00 -2.26816356e-01 2.29999110e-01
5.80962241e-01 -2.13750601e-02 -6.52390048e-02 -1.16300501e-01
1.26846284e-01 -1.25184274e+00 2.91879326e-01 4.56508577e-01
1.12670910e+00 -2.31334493e-01 1.16131651e+00 1.98292017e-01
-8.78162861e-01 -3.06576099e-02 -2.41497651e-01 3.04817051e-01
-3.46633762e-01 6.42675996e-01 -1.60895312e+00 5.51698089e-01
5.56976378e-01 7.20898688e-01 -9.23730791e-01 9.49596345e-01
-9.00186449e-02 8.81309569e-01 -7.73220882e-02 7.82179013e-02
5.04253283e-02 3.72767359e-01 1.62132800e-01 1.00651872e+00
2.67755896e-01 1.94406793e-01 1.31166488e-01 6.61147654e-01
1.01746701e-01 1.80915982e-01 -3.99149984e-01 -2.07649231e-01
4.13994193e-01 1.58223808e+00 -9.00730848e-01 -4.74409848e-01
-1.06797256e-01 9.03786719e-01 1.50990725e-01 -3.05585682e-01
-6.74343169e-01 8.07050467e-02 2.98106167e-02 5.86773753e-01
1.05023295e-01 3.55698287e-01 -4.48990434e-01 -8.36747944e-01
-1.11963972e-01 -5.30025542e-01 4.64740187e-01 -6.78877056e-01
-1.22487831e+00 7.14820862e-01 -3.49058628e-01 -1.59927464e+00
-3.99677038e-01 -3.94460618e-01 -3.01151782e-01 4.04795378e-01
-1.41375899e+00 -1.49244177e+00 -8.66727054e-01 3.22901011e-01
3.99139225e-01 -1.30758703e-01 1.30266023e+00 -8.98183435e-02
-8.92786682e-02 5.30754030e-01 -2.13384014e-02 1.75445020e-01
8.29477727e-01 -1.39104092e+00 2.16063149e-02 -3.36582102e-02
-3.62131856e-02 3.13153207e-01 6.59063220e-01 -5.44762313e-01
-1.05864370e+00 -1.13946283e+00 1.07362413e+00 -8.53242397e-01
8.00307810e-01 -1.63656157e-02 -9.63020623e-01 6.54180229e-01
-1.90942153e-01 6.78945065e-01 1.18350840e+00 5.60249910e-02
-2.06986874e-01 1.51072785e-01 -1.15589917e+00 2.62138098e-01
1.10628164e+00 -2.74179399e-01 -1.64454609e-01 7.88632691e-01
6.07577920e-01 -2.21788153e-01 -1.17700160e+00 4.02095377e-01
5.01813114e-01 -8.07839692e-01 6.86686754e-01 -6.02919936e-01
7.74260759e-01 -1.22706242e-01 1.16341241e-01 -9.74787533e-01
-4.75210220e-01 -4.34462845e-01 1.70946568e-02 7.78116167e-01
7.06701756e-01 -5.16219139e-01 1.24854708e+00 6.43954456e-01
-1.67509928e-01 -1.17369306e+00 -7.05207765e-01 -5.59827864e-01
-1.16861872e-01 -3.89026254e-02 3.40691537e-01 1.15013659e+00
2.08137438e-01 7.42297471e-02 -1.43445328e-01 1.52818603e-03
6.71678722e-01 -1.95917040e-02 4.58775908e-01 -1.59489322e+00
-4.24701750e-01 -1.49310485e-01 -6.74732804e-01 -1.00400262e-01
-3.86692621e-02 -1.30923176e+00 -4.42797765e-02 -1.58294702e+00
8.58252704e-01 -6.68398976e-01 -2.65075974e-02 1.00183475e+00
-3.94047856e-01 4.33827490e-01 -1.68707848e-01 8.16604555e-01
-7.58603513e-01 2.52085775e-01 1.33890975e+00 7.15971589e-02
1.51026741e-01 2.49174625e-01 -9.96120214e-01 8.67891610e-01
9.47498560e-01 -8.20329726e-01 -3.76737267e-01 -2.28237882e-02
-5.62451594e-02 9.57724974e-02 6.49008453e-01 -8.72792006e-01
2.53058702e-01 7.71604106e-02 5.19765973e-01 -3.58291209e-01
1.93779200e-01 -1.01531684e+00 3.72909665e-01 1.04812908e+00
-5.89055836e-01 -4.13983613e-01 -1.58124939e-01 5.91205478e-01
-2.15291396e-01 -5.03660023e-01 7.12665260e-01 -4.16707844e-01
-3.37624878e-01 3.24191302e-01 -2.32809067e-01 9.41326097e-02
1.12375903e+00 -3.05000156e-01 -2.68584341e-01 -3.92303616e-01
-1.13576245e+00 1.40662268e-01 4.95304435e-01 -7.54186660e-02
3.17389011e-01 -9.82693017e-01 -8.90691340e-01 -3.39753032e-02
4.72279519e-01 1.73155800e-01 2.00408787e-01 1.17103803e+00
-4.30707633e-01 6.08595788e-01 8.18103179e-02 -8.98227096e-01
-1.28042173e+00 4.31543827e-01 3.38547796e-01 -3.54474545e-01
-6.83074534e-01 8.26942146e-01 2.79700011e-01 -3.74085009e-01
6.16880506e-02 1.87590003e-01 1.15415305e-01 -1.45201057e-01
2.12110117e-01 2.33948112e-01 8.28204602e-02 -1.56938538e-01
-4.34085727e-01 1.71914548e-01 -1.87784716e-01 2.11777523e-01
1.40715873e+00 1.63284481e-01 -6.39045238e-02 3.71161908e-01
9.81685936e-01 -1.72336206e-01 -9.90551233e-01 -3.56196851e-01
2.50601560e-01 -2.97019422e-01 -4.61404808e-02 -1.09746802e+00
-8.93826306e-01 6.18315637e-01 5.82726836e-01 1.75723266e-02
8.27898860e-01 6.84654653e-01 2.64338285e-01 4.16728824e-01
6.34049252e-02 -5.11668324e-01 1.69959381e-01 -1.10000789e-01
6.68573081e-01 -1.76882243e+00 2.52786309e-01 -1.00293136e+00
-9.52699780e-01 1.02457976e+00 5.00660300e-01 2.91421451e-02
7.75190532e-01 3.42310667e-01 3.09203804e-01 -4.29692149e-01
-8.81018341e-01 -2.06872731e-01 5.24784505e-01 6.24176323e-01
7.32064307e-01 3.25740814e-01 -2.17986137e-01 7.20470548e-01
-4.54152524e-01 1.74840778e-01 3.48495007e-01 8.07780147e-01
-5.10334671e-01 -9.68407273e-01 5.49798198e-02 9.52292621e-01
-5.03425777e-01 -2.74553418e-01 -5.39377272e-01 9.16894019e-01
8.25862121e-03 5.61462998e-01 2.69405241e-03 -9.34092626e-02
-1.34872288e-01 1.40185028e-01 6.77987859e-02 -8.39475811e-01
-5.17617226e-01 2.04019129e-01 1.36850968e-01 -4.68451172e-01
-3.86938930e-01 -6.87314153e-01 -1.39902818e+00 4.92034629e-02
-5.17242372e-01 3.83953780e-01 6.54231548e-01 7.87060142e-01
4.73349094e-01 5.97237170e-01 5.80880702e-01 -2.02631965e-01
-7.28443742e-01 -8.16035271e-01 -3.61028969e-01 4.08580363e-01
3.06010634e-01 -4.11210358e-01 -3.18892866e-01 3.28796744e-01] | [15.014144897460938, -2.215553045272827] |
87d69ff8-d475-4868-b704-1a3437e4d635 | computational-complexity-of-observing | 1808.03387 | null | http://arxiv.org/abs/1808.03387v1 | http://arxiv.org/pdf/1808.03387v1.pdf | Computational Complexity of Observing Evolution in Artificial-Life Forms | Observations are an essential component of the simulation based studies on
artificial-evolutionary systems (AES) by which entities are identified and
their behavior is observed to uncover higher-level "emergent" phenomena.
Because of the heterogeneity of AES models and implicit nature of observations,
precise characterization of the observation process, independent of the
underlying micro-level reaction semantics of the model, is a difficult problem.
Building upon the multiset based algebraic framework to characterize
state-space trajectory of AES model simulations, we estimate bounds on
computational resource requirements of the process of automatically discovering
life-like evolutionary behavior in AES models during simulations. For
illustration, we consider the case of Langton's Cellular Automata model and
characterize the worst case computational complexity bounds for identifying
entity and population level reproduction. | ['Misra Janardan'] | 2018-06-24 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [ 2.78609723e-01 8.91795903e-02 1.99744925e-01 3.84463072e-01
2.85433710e-01 -8.61784220e-01 9.46994960e-01 7.79255092e-01
-2.21500918e-01 7.95846105e-01 -2.63920754e-01 -4.70455796e-01
-4.05568928e-01 -9.12464321e-01 -5.13508022e-01 -9.41929162e-01
-7.72002101e-01 5.80263317e-01 -8.97192489e-03 -4.03679371e-01
9.33711827e-02 7.14120328e-01 -1.66277134e+00 -2.81312287e-01
7.00286806e-01 5.37795365e-01 -2.36836851e-01 1.28216529e+00
1.46114260e-01 3.19307029e-01 -7.92665064e-01 5.10841124e-02
2.40418702e-01 -8.60389888e-01 -4.44998235e-01 3.47471803e-01
-6.02472067e-01 4.71166760e-01 -2.33332291e-01 1.30246472e+00
1.73942715e-01 -1.15058452e-01 8.91323030e-01 -1.54729509e+00
-2.21067294e-01 6.90899193e-01 -2.59752721e-01 3.58392715e-01
2.55852103e-01 2.91318834e-01 5.95071018e-01 -3.28513741e-01
6.84746921e-01 7.18274295e-01 3.91255051e-01 4.18514788e-01
-1.59669137e+00 -3.88367534e-01 -9.07768309e-02 -2.17549443e-01
-1.68608427e+00 -4.53883141e-01 3.41460973e-01 -4.79601532e-01
8.89399886e-01 6.31435513e-01 1.05408120e+00 1.07405746e+00
6.15538776e-01 2.66308576e-01 1.11440456e+00 -5.22708774e-01
8.92386436e-01 3.01247299e-01 1.54925227e-01 7.77888238e-01
8.48389268e-01 7.18042493e-01 -3.89677107e-01 -7.46689022e-01
5.88473797e-01 -1.51505396e-01 -1.70490831e-01 -4.00755197e-01
-9.09447789e-01 2.21835047e-01 -3.38115871e-01 2.50209302e-01
-5.16162872e-01 4.42112386e-01 7.25325793e-02 5.57986140e-01
2.49259606e-01 8.39792192e-01 -6.24401689e-01 -2.99982458e-01
-5.05889356e-01 2.55656928e-01 1.15380955e+00 9.36499119e-01
7.04785347e-01 1.16869196e-01 4.66065913e-01 -1.98889434e-01
1.76508771e-03 5.12547612e-01 4.02085751e-01 -4.99684453e-01
-5.16601861e-01 1.07644296e+00 3.02523166e-01 -4.57106650e-01
-3.74356776e-01 -8.22483659e-01 -1.01275456e+00 9.77594703e-02
2.79175311e-01 -4.05357122e-01 -5.96588552e-01 1.84916699e+00
4.64036435e-01 4.03550893e-01 3.67486507e-01 1.68610886e-01
-1.51972681e-01 7.70080447e-01 6.76184054e-03 -6.74250960e-01
1.29323077e+00 -2.16302603e-01 -3.29995930e-01 2.89522588e-01
8.41359437e-01 -5.58623783e-02 5.11150718e-01 2.13056076e-02
-1.25325572e+00 1.57486480e-02 -1.01674628e+00 8.45510244e-01
-6.27095759e-01 -2.82529712e-01 6.59270167e-01 9.30652499e-01
-1.04721308e+00 2.98875123e-01 -1.02860987e+00 -8.73078763e-01
-2.69178301e-01 6.62583888e-01 1.10418908e-01 9.54006195e-01
-9.00329769e-01 6.43365860e-01 2.62858212e-01 -5.33174202e-02
-1.25547564e+00 -8.19325745e-01 -4.28639531e-01 5.12625277e-01
3.05490851e-01 -9.28658903e-01 1.00246823e+00 -1.01521194e+00
-1.44983673e+00 5.22919893e-01 -3.44843149e-01 -5.69700360e-01
2.96893984e-01 7.73938537e-01 -5.85722208e-01 -2.04216808e-01
-5.00964165e-01 4.34446074e-02 3.97992402e-01 -1.51852810e+00
-6.36847198e-01 -2.63062984e-01 1.65799838e-02 -3.74578014e-02
-2.88508147e-01 -2.16421545e-01 3.06583941e-01 -1.71695113e-01
-3.34111661e-01 -1.18670619e+00 -7.35685587e-01 -6.06135130e-01
-2.04099879e-01 2.68940981e-02 4.94686007e-01 2.35880926e-01
1.41368234e+00 -1.91883540e+00 3.94601583e-01 6.62304580e-01
3.15942764e-01 -2.63519794e-01 1.62548088e-02 1.16145515e+00
-3.37212272e-02 3.22087198e-01 -3.04367155e-01 -1.45781592e-01
9.28113386e-02 -1.85996950e-01 -5.75015426e-01 5.53686261e-01
9.35601071e-02 8.55978787e-01 -7.95698106e-01 -2.79892348e-02
-2.37996522e-02 4.78368485e-03 -2.29556993e-01 1.21733375e-01
-5.18686235e-01 4.31229234e-01 -5.00138640e-01 4.84830618e-01
1.85990855e-01 -6.18834615e-01 4.84990746e-01 5.39950311e-01
-4.67459679e-01 -1.11732334e-01 -1.06184256e+00 9.25633430e-01
-5.58204412e-01 4.96970028e-01 -1.10524110e-01 -6.73627555e-01
5.22109210e-01 4.83752370e-01 4.43937451e-01 9.81588848e-03
5.00770807e-01 1.29656017e-01 5.88971674e-01 -6.72803298e-02
5.23176789e-01 1.02167539e-01 -4.97836351e-01 9.41282690e-01
-2.08916947e-01 1.41362354e-01 2.46600911e-01 1.62053540e-01
1.48928189e+00 -5.69705009e-01 7.63858259e-01 -9.98349547e-01
5.10582328e-01 3.27652186e-01 2.76252538e-01 9.06733513e-01
-2.86635421e-02 -4.26811665e-01 4.76473242e-01 -4.01126742e-01
-1.07644796e+00 -9.99749303e-01 -1.62916571e-01 4.26119924e-01
5.77627540e-01 -6.03924930e-01 -9.17108357e-01 -1.29724115e-01
-2.37365127e-01 6.08781576e-01 -1.16899383e+00 -5.26813090e-01
-2.33426109e-01 -1.21948206e+00 4.78622973e-01 -8.84217024e-02
2.24079952e-01 -1.02649176e+00 -1.20379579e+00 2.92860806e-01
3.95239234e-01 -8.56564105e-01 -5.86015657e-02 3.20224643e-01
-7.34977722e-01 -9.42434251e-01 -2.59830415e-01 -3.23186249e-01
1.08298624e+00 -3.43279243e-01 9.54676032e-01 2.77995318e-01
-5.62748253e-01 4.62313771e-01 5.18820770e-02 -6.04425192e-01
-9.14786160e-01 1.04196802e-01 3.62440854e-01 5.67117780e-02
3.57757479e-01 -8.37066650e-01 -4.69170272e-01 1.09422348e-01
-9.64658260e-01 -1.10110886e-01 4.22572225e-01 4.70897704e-01
2.34667987e-01 6.76160753e-01 2.90244669e-01 -5.42695224e-01
9.34518158e-01 -7.49961078e-01 -1.03135395e+00 8.02772880e-01
-7.15768158e-01 2.49018401e-01 9.36282277e-01 -6.94375753e-01
-5.21005690e-01 6.77093193e-02 7.47712791e-01 6.39886111e-02
-2.86773443e-01 3.84315133e-01 1.42179234e-02 -7.79050961e-02
4.40799296e-01 8.77949417e-01 -2.60270357e-01 5.15388660e-02
-1.71759665e-01 4.01483208e-01 3.16780448e-01 -7.39554584e-01
8.88244629e-01 5.29821873e-01 6.83347106e-01 -1.24541461e+00
2.11472958e-01 6.81979069e-03 -1.64067373e-01 -2.33400300e-01
4.77814823e-01 -5.87568879e-01 -1.41196632e+00 3.61738473e-01
-8.96452665e-01 -5.45287073e-01 -7.38926649e-01 2.53276825e-02
-5.50840378e-01 5.42040728e-02 -2.72329509e-01 -1.48504364e+00
-1.02026746e-01 -9.87139821e-01 8.31904292e-01 3.68732959e-01
-5.79242587e-01 -1.19081879e+00 6.64291263e-01 -7.28322029e-01
3.83268237e-01 2.72422701e-01 1.25917447e+00 -6.18897617e-01
-7.12281287e-01 -3.01090062e-01 3.80426049e-01 -9.04228270e-01
6.36267886e-02 5.70921183e-01 -5.27650237e-01 -4.68599468e-01
7.50988945e-02 4.81023759e-01 1.36839688e-01 3.68926257e-01
9.32691574e-01 -3.93891931e-01 -8.08993816e-01 5.37231743e-01
1.54919839e+00 4.26206052e-01 2.35112578e-01 1.43328026e-01
-2.31105462e-01 6.39570355e-01 -5.78544401e-02 7.84143209e-01
1.88554257e-01 4.27456617e-01 1.64435595e-01 1.54467836e-01
5.63621104e-01 -1.08542472e-01 5.33452690e-01 8.24934602e-01
-2.32801750e-01 -8.01512659e-01 -9.65963542e-01 6.03203952e-01
-1.76115191e+00 -9.07492638e-01 9.43539143e-02 2.28632784e+00
4.25098598e-01 -1.13663906e-02 5.01057565e-01 -7.88641870e-02
8.15041363e-01 -4.85172987e-01 -7.57341027e-01 -5.85762858e-01
-2.45203510e-01 1.91723496e-01 6.56828046e-01 5.16039610e-01
-2.84902036e-01 5.22548676e-01 7.40997124e+00 4.87212598e-01
-8.38071644e-01 -7.55034685e-02 4.42441851e-01 1.77526996e-02
-3.70080352e-01 1.94317356e-01 -5.93888938e-01 6.03201747e-01
1.59499407e+00 -9.32522416e-01 7.42771447e-01 2.03244448e-01
1.80269450e-01 -2.34756470e-01 -1.24669278e+00 4.73266900e-01
-6.44260526e-01 -1.39778304e+00 2.78184433e-02 6.82402670e-01
8.72226417e-01 -2.48614192e-01 1.38453200e-01 -5.02475910e-02
1.13171923e+00 -8.90825272e-01 6.11672640e-01 3.10266078e-01
6.95753813e-01 -7.31332123e-01 2.37976894e-01 6.12294436e-01
-1.27397192e+00 -2.32304275e-01 -9.67434198e-02 -2.10726812e-01
1.31255537e-01 2.43273020e-01 -7.12304711e-01 3.53727102e-01
3.30100864e-01 1.12931319e-01 -3.24835837e-01 8.66710603e-01
2.52171099e-01 6.59260690e-01 -7.31161773e-01 -6.07033849e-01
-4.16106656e-02 -3.52996022e-01 8.73451293e-01 8.98703992e-01
6.58148944e-01 3.82769018e-01 -3.25400114e-01 1.05539823e+00
-6.21344000e-02 -2.96334296e-01 -7.04669416e-01 -4.96150553e-01
6.58468008e-01 1.11421537e+00 -1.15850484e+00 -3.36831272e-01
2.00987801e-01 7.07699537e-01 -1.43227190e-01 5.05124211e-01
-8.04579616e-01 -1.41225055e-01 9.21277940e-01 1.16769217e-01
4.60265167e-02 -4.97903258e-01 -4.46625203e-01 -1.15114391e+00
-3.71089846e-01 -9.03413653e-01 4.37539965e-02 -3.52899730e-01
-8.69786501e-01 6.79611385e-01 -1.32093588e-02 -9.38161135e-01
-1.82284713e-01 -2.31297910e-01 -8.69568765e-01 5.74285090e-01
-6.51468992e-01 -4.22942132e-01 -8.88331383e-02 3.57828468e-01
2.98550725e-01 -2.15801269e-01 1.05773187e+00 -4.63195950e-01
-9.01335061e-01 5.27304471e-01 5.00666976e-01 -5.48472285e-01
-3.87780428e-01 -1.13049340e+00 6.20908558e-01 9.64303017e-01
1.21962152e-01 7.94592440e-01 1.38385212e+00 -7.75448382e-01
-1.70385861e+00 -8.25453579e-01 6.35007799e-01 -6.53767228e-01
9.96023118e-01 -7.76212513e-01 -3.47220033e-01 6.00815475e-01
3.49540174e-01 -4.10074353e-01 5.43973148e-01 -3.83320414e-02
1.31537855e-01 3.57828587e-01 -8.94883096e-01 1.30236316e+00
1.24042618e+00 -5.42115092e-01 -1.10892236e-01 1.05204813e-01
3.95926088e-01 1.75065681e-01 -7.65924394e-01 3.54267299e-01
4.09573019e-01 -5.11182427e-01 5.96751809e-01 -7.28357792e-01
1.60981923e-01 -4.44987923e-01 8.13619792e-02 -1.29183924e+00
-2.36486748e-01 -1.31026602e+00 -4.98852193e-01 7.94117749e-01
6.50247335e-01 -9.82158184e-01 6.35360241e-01 4.53328162e-01
6.32469237e-01 -5.60092628e-01 -6.20268285e-01 -9.40831006e-01
1.30862566e-02 2.38301739e-01 9.76932108e-01 7.61755526e-01
4.65160787e-01 1.52928039e-01 8.11232179e-02 7.28544235e-01
8.52892935e-01 2.57335991e-01 7.89963484e-01 -1.23079264e+00
-6.19391620e-01 -7.07716525e-01 -5.46780825e-01 -5.12603700e-01
6.93188831e-02 -4.92450476e-01 -1.21696733e-01 -4.14657623e-01
3.07703584e-01 -3.17857295e-01 -3.55674118e-01 -3.64982426e-01
2.23276258e-01 -3.14653993e-01 -1.98802471e-01 2.72748053e-01
-4.69084352e-01 5.88366032e-01 4.60495085e-01 1.88981041e-01
-3.20843339e-01 -1.12015843e-01 -4.20746297e-01 4.43607420e-01
7.14375615e-01 -6.75196052e-01 -5.45492947e-01 2.12587193e-01
7.29494214e-01 3.81714463e-01 3.48959088e-01 -8.71744514e-01
4.91168350e-01 -4.16078061e-01 -8.46861303e-02 -2.76487023e-01
2.67163843e-01 -7.79351830e-01 1.03287017e+00 1.05075276e+00
-6.40652537e-01 5.69323778e-01 1.48394763e-01 8.23215306e-01
1.15803361e-01 -6.26682788e-02 4.93889391e-01 5.32081835e-02
-1.73900455e-01 4.50120211e-01 -1.23778284e+00 -1.87329635e-01
1.51603305e+00 -3.98693621e-01 -4.44165885e-01 -3.37467641e-01
-5.82163870e-01 1.46403506e-01 1.15501165e+00 -1.30128503e-01
7.78809339e-02 -5.96527278e-01 -4.85146314e-01 1.78986624e-01
1.07606724e-01 -4.03193861e-01 6.06993586e-02 9.20736849e-01
-4.17741418e-01 3.25331599e-01 -2.74305850e-01 -3.28657985e-01
-1.15445769e+00 6.64855301e-01 5.65862834e-01 -3.09518784e-01
-2.75989503e-01 4.30026531e-01 5.09283125e-01 -4.14377488e-02
-3.55220556e-01 -7.25671574e-02 3.24385762e-01 -5.04692197e-01
1.29585564e-01 4.25022990e-01 -2.75078535e-01 -2.49999046e-01
-3.69219959e-01 2.44183004e-01 1.83785141e-01 -4.24272269e-01
1.11765575e+00 -3.34047884e-01 -4.41967726e-01 6.79296613e-01
6.62032664e-01 3.48038003e-02 -7.94385493e-01 1.28486112e-01
2.24031255e-01 1.12977773e-01 -3.31329525e-01 -5.25443494e-01
-4.05821800e-01 6.71462119e-01 3.80117983e-01 9.41290081e-01
1.08631790e+00 -1.13386706e-01 2.68736511e-01 5.93671083e-01
6.89449608e-01 -7.36262798e-01 -4.95514810e-01 9.86255854e-02
3.42424870e-01 -4.96344686e-01 -1.84964567e-01 -3.01020563e-01
5.06708287e-02 9.69778538e-01 3.80239457e-01 -3.64553511e-01
9.19773281e-01 7.69207001e-01 -5.45957983e-01 -4.75891024e-01
-1.53777862e+00 -3.84902023e-03 -2.88587093e-01 3.22856635e-01
-1.75668955e-01 2.45457724e-01 -3.02191406e-01 4.00847405e-01
-1.85394436e-01 -2.19402194e-01 1.07931149e+00 8.60852838e-01
-2.82090724e-01 -9.29835498e-01 -4.01605479e-02 4.94410396e-02
-3.33848357e-01 -2.91940309e-02 -8.39697838e-01 7.68982232e-01
-1.88204706e-01 7.16287673e-01 2.46047080e-01 -2.93956339e-01
-1.33544073e-01 -1.01541854e-01 4.69053268e-01 -3.49499822e-01
-8.22503805e-01 -4.07877058e-01 5.19698821e-02 -7.41523430e-02
-1.87657952e-01 -7.47774541e-01 -1.06955338e+00 -7.12538660e-01
-5.90309620e-01 6.60587490e-01 5.77605247e-01 9.24173176e-01
7.16005564e-01 3.31327379e-01 8.13491940e-01 -2.78040081e-01
-4.14378285e-01 -3.98473293e-01 -7.42178500e-01 -7.08460249e-03
2.25132361e-01 -3.43618184e-01 -6.97666764e-01 5.13890348e-02] | [5.609847068786621, 4.150233268737793] |
b2f1e8a0-f142-4c96-9ec7-7ec57a7f5c81 | an-in-depth-investigation-of-user-response | 2304.07944 | null | https://arxiv.org/abs/2304.07944v1 | https://arxiv.org/pdf/2304.07944v1.pdf | An In-depth Investigation of User Response Simulation for Conversational Search | Conversational search has seen increased recent attention in both the IR and NLP communities. It seeks to clarify and solve a user's search need through multi-turn natural language interactions. However, most existing systems are trained and demonstrated with recorded or artificial conversation logs. Eventually, conversational search systems should be trained, evaluated, and deployed in an open-ended setting with unseen conversation trajectories. A key challenge is that training and evaluating such systems both require a human-in-the-loop, which is expensive and does not scale. One strategy for this is to simulate users, thereby reducing the scaling costs. However, current user simulators are either limited to only respond to yes-no questions from the conversational search system, or unable to produce high quality responses in general. In this paper, we show that current state-of-the-art user simulation system could be significantly improved by replacing it with a smaller but advanced natural language generation model. But rather than merely reporting this new state-of-the-art, we present an in-depth investigation of the task of simulating user response for conversational search. Our goal is to supplement existing works with an insightful hand-analysis of what challenges are still unsolved by the advanced model, as well as to propose our solutions for them. The challenges we identified include (1) dataset noise, (2) a blind spot that is difficult for existing models to learn, and (3) a specific type of misevaluation in the standard empirical setup. Except for the dataset noise issue, we propose solutions to cover the training blind spot and to avoid the misevaluation. Our proposed solutions lead to further improvements. Our best system improves the previous state-of-the-art significantly. | ['Vivek Srikumar', 'Qingyao Ai', 'Zhichao Xu', 'Zhenduo Wang'] | 2023-04-17 | null | null | null | null | ['user-simulation', 'conversational-search'] | ['natural-language-processing', 'natural-language-processing'] | [ 8.44154432e-02 3.21936607e-01 1.23304680e-01 -3.48725319e-01
-1.00792336e+00 -8.20754290e-01 7.20096648e-01 -3.87505949e-01
-2.93427289e-01 7.33749449e-01 3.54992539e-01 -6.39981151e-01
-1.13096483e-01 -2.75980920e-01 -2.69693732e-01 -2.52645314e-01
2.55414784e-01 8.85205328e-01 9.25740823e-02 -6.62452936e-01
1.36391044e-01 -2.79298499e-02 -1.71375108e+00 4.55453992e-01
1.12707174e+00 4.90184903e-01 4.71201450e-01 1.01966441e+00
-1.63508534e-01 7.92734087e-01 -8.30796003e-01 -4.10043210e-01
5.79660535e-02 -6.42855108e-01 -1.29775059e+00 -1.47242695e-01
2.16406912e-01 -5.15793800e-01 -7.28251114e-02 7.02245831e-01
9.74719465e-01 2.46697605e-01 2.44921044e-01 -1.39397490e+00
-7.62242138e-01 3.62805963e-01 4.36587036e-02 -1.72291785e-01
9.88035381e-01 2.94580191e-01 9.07019794e-01 -7.25710511e-01
4.93888736e-01 1.32425106e+00 3.12614292e-01 1.12171590e+00
-9.65194464e-01 -5.22044241e-01 2.11240292e-01 1.75918698e-01
-1.12600315e+00 -6.27459586e-01 4.39257443e-01 -3.34083885e-01
9.36032891e-01 7.31483817e-01 3.30565572e-01 1.62290919e+00
-6.35273039e-01 8.96758258e-01 1.12892401e+00 -7.19645798e-01
6.81428090e-02 6.87856495e-01 1.32858709e-01 5.51491201e-01
-3.40510279e-01 1.00071259e-01 -5.18565238e-01 -4.91062909e-01
3.78520608e-01 -2.56835014e-01 -6.25718057e-01 -2.41709501e-01
-1.03572440e+00 7.51705945e-01 1.74913481e-01 2.84129441e-01
-2.31191278e-01 -2.55689621e-01 -2.57792957e-02 5.73817790e-01
3.37394893e-01 8.98447573e-01 -5.20644486e-01 -5.62973261e-01
-7.80290902e-01 4.95673925e-01 1.57203925e+00 9.42223549e-01
4.85347599e-01 -5.00114858e-01 -4.34939474e-01 1.17216969e+00
2.09682330e-01 4.22475040e-01 5.67984521e-01 -8.59646261e-01
4.00438756e-01 6.47837520e-01 5.90931416e-01 -7.09813297e-01
-3.66053730e-01 -3.57248425e-01 -4.79574710e-01 -1.27272621e-01
6.42359614e-01 -3.54036391e-01 -6.33185565e-01 1.93386710e+00
1.14741445e-01 -5.29137589e-02 1.40644133e-01 1.13164306e+00
1.00540793e+00 4.65604037e-01 -2.73697585e-01 -3.08120102e-01
1.52953494e+00 -1.26123357e+00 -7.26244211e-01 -3.19255978e-01
7.56119490e-01 -9.46782410e-01 1.58183384e+00 2.26742253e-01
-1.22355199e+00 -3.01310897e-01 -6.92130387e-01 -1.02603838e-01
-3.34014863e-01 1.19497709e-01 5.70066750e-01 8.55101049e-01
-1.40607095e+00 3.14042605e-02 -3.72571498e-01 -7.76448190e-01
-3.41257304e-01 3.66081357e-01 -5.84700182e-02 1.47665301e-02
-1.46674013e+00 1.13432801e+00 -3.40212971e-01 1.23849638e-01
-5.60433567e-01 -4.34148878e-01 -4.15797651e-01 2.29881555e-01
6.26110613e-01 -9.58968878e-01 2.05912995e+00 -7.62676775e-01
-1.84617400e+00 7.66479671e-01 -4.72808838e-01 -1.14220373e-01
6.15982592e-01 -1.48179814e-01 -3.40046197e-01 -1.86799884e-01
-5.43309115e-02 3.27458233e-01 2.56189287e-01 -1.54906154e+00
-4.46222544e-01 -1.80162922e-01 4.64142293e-01 5.36284089e-01
-1.97854787e-01 2.57191896e-01 -7.47759104e-01 -2.24012285e-01
-1.90628991e-01 -1.16294956e+00 -3.12994659e-01 -4.71136928e-01
-2.53749222e-01 -4.72110599e-01 6.16276383e-01 -5.40374815e-01
1.46711957e+00 -1.75330913e+00 -1.25783131e-01 2.20977608e-03
1.03037603e-01 5.14240146e-01 -2.67022640e-01 8.43232870e-01
1.39064759e-01 3.46407473e-01 1.17608704e-01 -6.44136667e-01
9.43578556e-02 -1.61910191e-01 -3.57321143e-01 -1.09576665e-01
-1.93347961e-01 9.59709644e-01 -1.10948372e+00 -2.82713443e-01
-1.56105578e-01 2.23396152e-01 -6.04729116e-01 6.92171991e-01
-3.73551488e-01 5.41036606e-01 -3.80385876e-01 4.95586157e-01
3.08881640e-01 -3.86473924e-01 5.89289255e-02 2.28446528e-01
-8.70888978e-02 6.67696893e-01 -1.04995477e+00 1.58819878e+00
-7.62835920e-01 5.11477888e-01 3.85073334e-01 -7.25709379e-01
6.37543440e-01 5.35308063e-01 1.86133713e-01 -7.63608217e-01
-3.27645615e-02 2.56795853e-01 1.17891990e-01 -8.44080746e-01
7.04740763e-01 1.78272903e-01 -1.66522071e-01 9.39007699e-01
-2.13652253e-01 -1.29789904e-01 4.03169878e-02 3.30775261e-01
1.16375256e+00 -3.38974625e-01 3.24004181e-02 -1.06582381e-01
5.65646112e-01 -6.00486435e-02 8.30957890e-02 1.29399574e+00
-2.55174398e-01 5.87609828e-01 3.58365446e-01 2.36043986e-02
-5.04610479e-01 -6.31202698e-01 2.96238393e-01 1.34508407e+00
2.17222601e-01 -4.39202458e-01 -1.02832711e+00 -6.29989386e-01
-4.56923187e-01 7.98259437e-01 -2.19959289e-01 9.11153704e-02
-4.19364691e-01 -6.09520435e-01 7.35364497e-01 1.96810756e-02
4.11698133e-01 -1.25387645e+00 -4.78417337e-01 1.09009594e-01
-8.38497579e-01 -1.16523135e+00 -6.51937723e-01 -2.88623959e-01
-4.07319188e-01 -1.10089028e+00 -8.46278250e-01 -8.85237455e-01
4.93506700e-01 6.07738495e-01 1.21156931e+00 3.96818072e-01
1.02243596e-03 7.14237809e-01 -5.46373427e-01 -4.31165040e-01
-5.93526483e-01 2.46874660e-01 8.69701058e-02 -8.18509012e-02
5.20711720e-01 -4.01294112e-01 -9.27158475e-01 8.53584290e-01
-5.58951855e-01 6.58490360e-02 5.27091980e-01 9.19803977e-01
-2.99936056e-01 -3.81034732e-01 7.85651505e-01 -9.05605674e-01
1.66363943e+00 -3.54580790e-01 -2.18529090e-01 7.08936572e-01
-8.06015551e-01 7.61010423e-02 3.53065401e-01 -5.30104518e-01
-1.13705838e+00 -2.09058583e-01 -2.38946587e-01 7.58921430e-02
-1.32396027e-01 4.17195737e-01 4.63056266e-02 -1.61718950e-02
1.01585817e+00 2.66696960e-01 9.46328565e-02 -4.38002378e-01
4.04511720e-01 1.30313373e+00 3.15258682e-01 -6.60869896e-01
4.56644565e-01 -4.96810786e-02 -7.61805892e-01 -8.14320266e-01
-4.96328324e-01 -8.05609703e-01 -4.99289110e-03 -3.00443113e-01
4.75392848e-01 -8.11782897e-01 -1.24252915e+00 3.61489773e-01
-1.39050508e+00 -4.78011876e-01 1.68094978e-01 2.90478319e-01
-5.29130578e-01 4.34485227e-01 -4.77479756e-01 -1.40282595e+00
-5.78858197e-01 -1.38660145e+00 9.18516040e-01 3.35792661e-01
-6.34181559e-01 -7.65837312e-01 2.56525934e-01 8.50660384e-01
8.63699377e-01 -5.41697860e-01 6.87860191e-01 -8.44429910e-01
-5.67377865e-01 -3.48357320e-01 -1.23719536e-01 9.99178886e-02
3.06113977e-02 -3.08147520e-01 -1.19865513e+00 -3.01648676e-01
3.14912558e-01 -5.22741795e-01 2.90988922e-01 -1.01682030e-01
1.01116860e+00 -4.74036396e-01 -3.84533823e-01 4.41226214e-02
7.79118717e-01 2.57965207e-01 5.27502835e-01 7.88753405e-02
1.69726834e-01 9.45446551e-01 4.35922265e-01 1.07809789e-01
7.35346079e-01 9.69671965e-01 1.55656815e-01 -1.12928338e-01
6.49314299e-02 -3.53031844e-01 2.72498161e-01 7.12977767e-01
-4.87257689e-02 -7.56600380e-01 -9.08356786e-01 5.59987783e-01
-2.13230801e+00 -1.00338840e+00 4.39467318e-02 2.24908900e+00
9.06065762e-01 -1.74577594e-01 2.55940229e-01 -1.19626909e-01
5.06484866e-01 -1.43421441e-01 -2.40683049e-01 -4.89747792e-01
1.26584336e-01 3.15075740e-02 -1.22694187e-01 9.14617360e-01
-5.38188159e-01 9.33342278e-01 6.58167791e+00 3.47358704e-01
-9.24412370e-01 1.38408214e-01 3.36635351e-01 -1.32445484e-01
-3.22154194e-01 2.26931900e-01 -7.10337102e-01 4.21592027e-01
8.83059084e-01 -1.12252071e-01 8.62496138e-01 8.30010831e-01
3.73887360e-01 -1.72126397e-01 -1.33979642e+00 1.10072267e+00
3.13422441e-01 -9.22096789e-01 -8.57172608e-02 3.27385520e-03
3.87350589e-01 -1.02079615e-01 -8.00266638e-02 8.34647238e-01
5.25580943e-01 -1.02733541e+00 1.67629018e-01 4.29291844e-01
5.31437814e-01 -1.02818765e-01 7.51896262e-01 1.04052269e+00
-6.35191500e-01 -1.52298808e-01 1.41899288e-01 -2.19366714e-01
2.94964880e-01 1.73520476e-01 -1.09981549e+00 2.94319212e-01
4.77632016e-01 -8.73544514e-02 -3.79550040e-01 1.02305794e+00
-1.09913178e-01 4.54357445e-01 -3.80719036e-01 -6.19685054e-01
1.47183791e-01 -2.22742949e-02 3.81991476e-01 1.21221280e+00
3.28245878e-01 3.70716333e-01 2.05614313e-01 8.31014335e-01
-1.07486509e-01 1.26564637e-01 -5.44550955e-01 -4.10403274e-02
5.76301336e-01 1.05589080e+00 -1.69918045e-01 -2.35007182e-01
-3.39100599e-01 1.13640058e+00 4.11868542e-01 7.25609422e-01
-4.30548161e-01 -1.71379879e-01 5.63282430e-01 2.00379163e-01
-3.01230520e-01 7.80718178e-02 -1.19059592e-01 -1.22314489e+00
2.45355740e-01 -1.56668460e+00 2.06795588e-01 -8.52110267e-01
-1.17222142e+00 8.05703461e-01 -7.17127547e-02 -9.50911403e-01
-9.27149296e-01 -2.28015915e-01 -5.84580839e-01 1.21637845e+00
-1.20953441e+00 -8.36179912e-01 -4.56065953e-01 5.50562739e-01
8.35060596e-01 -1.84940875e-01 1.24858916e+00 4.47543144e-01
-3.90313208e-01 8.89630675e-01 -1.75531507e-01 -2.20027283e-01
1.01906884e+00 -1.05178750e+00 4.74183768e-01 4.91116256e-01
-4.87453165e-03 1.06427085e+00 9.25546765e-01 -3.55425388e-01
-1.31352687e+00 -2.70660818e-01 1.42742908e+00 -9.80122089e-01
4.84665960e-01 -8.03399503e-01 -8.50807846e-01 3.69808346e-01
3.26700598e-01 -5.88101268e-01 6.08621895e-01 5.70403516e-01
-5.80623597e-02 1.91322669e-01 -9.62792993e-01 9.29837227e-01
1.03231215e+00 -8.60098898e-01 -5.64300776e-01 6.10868514e-01
6.76164627e-01 -5.28972030e-01 -1.82432011e-01 2.11198688e-01
7.15167165e-01 -9.52197969e-01 8.55891824e-01 -6.41661704e-01
5.58376275e-02 3.86855472e-03 1.36129841e-01 -1.45695555e+00
4.87981290e-02 -1.21756029e+00 -6.19585477e-02 1.13789666e+00
7.18912005e-01 -6.71705544e-01 7.38804758e-01 1.19452894e+00
9.94011089e-02 -7.34620512e-01 -6.24124050e-01 -5.45209229e-01
-1.43129155e-01 -2.91207373e-01 5.91801167e-01 8.40223372e-01
5.49674988e-01 9.31663513e-01 -6.41819954e-01 2.23644879e-02
4.94020917e-02 -4.78683747e-02 1.19924724e+00 -1.12268758e+00
-5.06382048e-01 -5.28057218e-01 4.19577569e-01 -1.68286467e+00
-8.27935562e-02 -4.88119006e-01 4.22236353e-01 -1.67206466e+00
2.55699128e-01 -4.84605849e-01 2.08488047e-01 1.13654502e-01
-3.57508928e-01 -2.92714149e-01 1.93078041e-01 3.53452712e-01
-6.13304317e-01 4.39298719e-01 1.19130766e+00 1.06397837e-01
-5.84572971e-01 6.18502378e-01 -1.01599491e+00 3.88544202e-01
6.89664602e-01 -2.51623005e-01 -8.40058684e-01 -4.38753039e-01
4.18775588e-01 4.19002265e-01 3.57760966e-01 -6.18488371e-01
7.79397547e-01 2.99655776e-02 -4.52370763e-01 -3.82336825e-01
5.18583477e-01 -6.87793970e-01 -1.79475039e-01 1.95488647e-01
-5.64286828e-01 1.20914482e-01 -1.41627658e-02 4.64439452e-01
-2.16839999e-01 -3.24651599e-01 1.63091153e-01 -3.33394766e-01
-2.47952476e-01 -1.43191069e-01 -5.26830852e-01 5.65573052e-02
6.35330021e-01 9.95120965e-03 -6.32784367e-01 -1.21018624e+00
-5.46678662e-01 6.74450457e-01 2.19822139e-01 7.95111835e-01
2.95497626e-01 -7.47563183e-01 -6.68121874e-01 1.79868281e-01
2.64478594e-01 -2.43512139e-01 1.85642079e-01 4.12734061e-01
-2.41349965e-01 6.62425578e-01 4.04722810e-01 -3.40189695e-01
-1.40801096e+00 3.94198656e-01 4.47240770e-01 -5.09835780e-01
-1.68599258e-03 8.50910485e-01 6.00728206e-02 -8.90247941e-01
7.51946270e-01 -8.92980397e-03 -3.02170157e-01 -4.17044535e-02
5.72207510e-01 4.63602319e-02 1.26311451e-01 -2.83445865e-01
-2.72712916e-01 1.74916744e-01 -2.07158640e-01 -4.43104923e-01
7.64572978e-01 -3.13844651e-01 9.48486254e-02 2.13094667e-01
8.52967024e-01 1.05264196e-02 -4.50278699e-01 -3.84157568e-01
-8.43837485e-02 -2.97255516e-01 -2.44030163e-01 -1.37493229e+00
-3.18733454e-01 6.83413327e-01 5.47902882e-01 5.53424299e-01
9.21621323e-01 7.01265261e-02 7.98553705e-01 8.99767399e-01
3.44698578e-01 -1.07580745e+00 1.36172310e-01 6.19458079e-01
1.14310062e+00 -1.64142871e+00 -3.13633144e-01 -4.02976155e-01
-8.17825556e-01 6.86726034e-01 7.55091667e-01 5.41064858e-01
4.70825225e-01 -1.01219527e-01 5.02579927e-01 -2.76653290e-01
-1.02773106e+00 -3.65188032e-01 2.75224358e-01 4.27498639e-01
6.88609958e-01 -1.21644676e-01 -4.78029460e-01 7.14848459e-01
-4.24892724e-01 1.52495936e-01 4.13119078e-01 8.45154703e-01
-7.96706453e-02 -1.33953202e+00 -2.02517733e-01 2.02703968e-01
-2.46471584e-01 -2.04174772e-01 -1.02202117e+00 6.73452437e-01
-4.68891203e-01 1.70432913e+00 -3.61217767e-01 -5.56476474e-01
7.03498840e-01 2.83886760e-01 9.31016076e-03 -7.12626636e-01
-9.04409051e-01 -3.06899607e-01 5.23895025e-01 -5.28845251e-01
-3.01129729e-01 -3.59188408e-01 -5.82611024e-01 -2.79463232e-01
-5.84342659e-01 6.02677763e-01 7.74174631e-01 9.08562481e-01
6.44435704e-01 1.63326278e-01 6.14939392e-01 -4.59632516e-01
-9.21724558e-01 -1.34113002e+00 6.03419505e-02 4.05821443e-01
4.32043970e-01 -3.81895632e-01 -6.07940912e-01 -3.43120724e-01] | [12.167961120605469, 7.773397922515869] |
28e2f87d-f04c-487b-b52c-56d238526284 | learning-a-smooth-kernel-regularizer-for | 1903.01882 | null | http://arxiv.org/abs/1903.01882v1 | http://arxiv.org/pdf/1903.01882v1.pdf | Learning a smooth kernel regularizer for convolutional neural networks | Modern deep neural networks require a tremendous amount of data to train,
often needing hundreds or thousands of labeled examples to learn an effective
representation. For these networks to work with less data, more structure must
be built into their architectures or learned from previous experience. The
learned weights of convolutional neural networks (CNNs) trained on large
datasets for object recognition contain a substantial amount of structure.
These representations have parallels to simple cells in the primary visual
cortex, where receptive fields are smooth and contain many regularities.
Incorporating smoothness constraints over the kernel weights of modern CNN
architectures is a promising way to improve their sample complexity. We propose
a smooth kernel regularizer that encourages spatial correlations in convolution
kernel weights. The correlation parameters of this regularizer are learned from
previous experience, yielding a method with a hierarchical Bayesian
interpretation. We show that our correlated regularizer can help constrain
models for visual recognition, improving over an L2 regularization baseline. | ['Reuben Feinman', 'Brenden M. Lake'] | 2019-03-05 | null | null | null | null | ['l2-regularization'] | ['methodology'] | [ 2.15293512e-01 2.98913330e-01 -2.90525228e-01 -9.76475418e-01
-1.44842446e-01 -4.63046849e-01 6.75616324e-01 -1.59856424e-01
-6.93326652e-01 3.55932832e-01 3.81039500e-01 -1.62662849e-01
-1.85061581e-02 -5.66417575e-01 -9.07843173e-01 -7.01963902e-01
-1.63284257e-01 2.71887362e-01 4.70369041e-01 1.58505589e-01
2.85795867e-01 6.60400450e-01 -1.17370915e+00 5.56110144e-01
3.54085654e-01 1.05588543e+00 3.31926614e-01 6.67298257e-01
-9.07354206e-02 7.41075814e-01 -1.55603722e-01 -5.70610985e-02
3.71805549e-01 2.34703664e-02 -8.56906354e-01 9.07785892e-02
8.16087723e-01 -2.15147108e-01 -5.09117603e-01 1.02495384e+00
-7.27091625e-04 3.90689671e-01 9.66398358e-01 -6.35905564e-01
-8.72202456e-01 5.27128339e-01 -4.09284592e-01 3.46496791e-01
-5.85394681e-01 1.14608824e-01 1.13175726e+00 -9.63498771e-01
3.61731499e-01 1.29848969e+00 8.09879124e-01 7.35846043e-01
-1.80121017e+00 -3.75302374e-01 5.28288007e-01 6.38709450e-03
-1.28523922e+00 -5.39789617e-01 4.90812570e-01 -6.28964543e-01
1.24341393e+00 -1.37970656e-01 6.20611787e-01 9.91185546e-01
-5.01821861e-02 6.38610482e-01 7.56249487e-01 -4.76428181e-01
4.22905862e-01 1.51477858e-01 4.63391781e-01 8.62687647e-01
3.65575671e-01 9.80852470e-02 -5.25872409e-01 -1.74150437e-01
1.19642460e+00 2.25783080e-01 -1.91239357e-01 -6.30303860e-01
-9.73506153e-01 9.27729368e-01 8.81693900e-01 1.22614175e-01
-7.12827966e-02 5.82764685e-01 1.83478653e-01 1.01973992e-02
2.50128508e-01 5.31687200e-01 -8.57075930e-01 2.92859703e-01
-5.99711895e-01 -6.64963573e-02 6.75823867e-01 8.95301521e-01
9.54712749e-01 2.85408795e-01 7.90205747e-02 1.10412788e+00
5.16896784e-01 1.20121188e-01 4.42315102e-01 -1.05486488e+00
2.24213004e-01 4.21189189e-01 -2.31035858e-01 -8.08331966e-01
-4.49350744e-01 -5.11288345e-01 -1.00064039e+00 4.76520300e-01
7.70676672e-01 4.48285043e-02 -1.36466551e+00 1.77235687e+00
-1.51716471e-01 2.75803238e-01 -7.60846734e-02 7.67367482e-01
2.97990054e-01 5.02787411e-01 2.00612724e-01 2.53704697e-01
1.19423318e+00 -7.21359611e-01 -3.19793403e-01 -6.77056670e-01
6.46476984e-01 -6.69171572e-01 1.04196870e+00 4.09637481e-01
-8.23204994e-01 -5.12663722e-01 -9.78650749e-01 -2.50189900e-01
-3.58627081e-01 2.98508406e-01 9.73961830e-01 5.00852942e-01
-1.01882720e+00 8.08490992e-01 -1.15979540e+00 -8.58757049e-02
1.06270111e+00 6.14752293e-01 -3.08257550e-01 -1.48970544e-01
-5.01756489e-01 1.02357292e+00 5.79962671e-01 1.75446630e-01
-7.49465764e-01 -7.12782025e-01 -8.74651194e-01 1.91535130e-01
9.15296823e-02 -4.73963052e-01 1.20413506e+00 -1.08448374e+00
-1.35116005e+00 7.16712236e-01 -1.74860850e-01 -3.65678132e-01
-2.21268520e-01 -2.90521026e-01 1.64487794e-01 -3.91233303e-02
-4.52671230e-01 9.41993594e-01 1.02816427e+00 -1.11436450e+00
-4.24357206e-01 -2.34416872e-01 -1.17373392e-01 -2.43329480e-01
-4.54015046e-01 -5.27311489e-02 -4.21303421e-01 -8.40748608e-01
3.93899947e-01 -9.29575324e-01 -6.00311041e-01 2.48724937e-01
-2.32668072e-01 -3.67130309e-01 4.48502094e-01 -3.42151612e-01
7.26245344e-01 -2.22224951e+00 2.11423382e-01 5.88119745e-01
3.28239352e-01 2.73265153e-01 -4.36483651e-01 -3.45678926e-01
-3.45414951e-02 2.15362996e-01 -1.41046107e-01 -1.44917682e-01
-6.03318922e-02 6.18578970e-01 -4.14943963e-01 4.42748994e-01
7.35679150e-01 7.70378888e-01 -7.17566073e-01 2.29478758e-02
-3.30165327e-02 6.79504335e-01 -8.53530884e-01 1.68927208e-01
-3.90308350e-01 4.67901900e-02 -2.50989228e-01 2.47299612e-01
4.94280666e-01 -6.72796786e-01 1.64033607e-01 -3.36778522e-01
1.65673286e-01 5.67515790e-01 -1.01421142e+00 1.48896778e+00
-3.20529163e-01 8.23417306e-01 -3.67011242e-02 -1.26785767e+00
8.14984202e-01 2.66284160e-02 1.26135290e-01 -1.24088861e-01
1.73389316e-01 1.83031410e-02 3.36699575e-01 -3.40954870e-01
2.70262100e-02 -1.59255475e-01 4.50770706e-01 3.51819187e-01
5.91478765e-01 -3.62994475e-03 -1.41034558e-01 1.38765216e-01
1.17225075e+00 -1.64836973e-01 1.74170814e-03 -4.63668406e-01
-1.27661020e-01 -2.63383865e-01 8.49011242e-01 7.97179163e-01
1.15639381e-01 7.51685917e-01 4.84473407e-01 -9.53793406e-01
-1.20218158e+00 -9.91090536e-01 -4.63732511e-01 1.06339073e+00
-5.62856257e-01 -3.58785272e-01 -5.73274493e-01 -5.00638962e-01
-6.34417981e-02 2.42057517e-01 -9.25229728e-01 -2.71290272e-01
-4.70738471e-01 -8.82417262e-01 5.75700045e-01 1.01568520e+00
3.48284692e-01 -8.53701770e-01 -4.70441818e-01 1.91695184e-01
4.98956800e-01 -1.17593479e+00 -3.82015258e-01 9.12360966e-01
-1.26879418e+00 -9.39588904e-01 -7.00079501e-01 -9.03760254e-01
1.20209551e+00 9.28760022e-02 1.19440949e+00 3.42536867e-01
-6.28366530e-01 3.09874833e-01 9.78375413e-03 -4.27465230e-01
8.29729345e-03 1.56880289e-01 2.73207784e-01 -1.10546730e-01
5.86923659e-01 -7.28227615e-01 -4.25026894e-01 3.17220509e-01
-9.72625792e-01 -8.72937515e-02 7.80874789e-01 1.22286677e+00
4.66358811e-01 -2.85547376e-01 2.01940298e-01 -1.06817317e+00
1.84379742e-01 -3.24542969e-01 -7.22884536e-01 1.48842141e-01
-3.00243199e-01 7.05573142e-01 6.43552065e-01 -8.99180949e-01
-9.00381982e-01 3.87891531e-01 -2.87258606e-02 -4.10806775e-01
-3.71962607e-01 4.24893588e-01 1.59845695e-01 -4.04011995e-01
1.08174419e+00 -2.39017308e-01 -1.00943469e-01 -5.75392544e-01
4.36262786e-01 3.07304170e-02 2.85832763e-01 -7.84885526e-01
6.53555155e-01 4.43620414e-01 -2.77521908e-02 -8.94682050e-01
-1.21613085e+00 -3.08196485e-01 -9.64756131e-01 3.91079456e-01
8.23670149e-01 -7.96271741e-01 -6.28134608e-01 2.60779917e-01
-1.27598596e+00 -8.19340169e-01 -2.66318768e-01 8.15727115e-01
-3.99437726e-01 -1.18280917e-01 -6.34000301e-01 -5.35902500e-01
4.67683911e-01 -8.67617071e-01 4.81011957e-01 1.77547082e-01
-2.17619330e-01 -1.18878257e+00 -1.13523379e-02 -1.50725931e-01
5.57051718e-01 -3.81249577e-01 1.07146776e+00 -6.62632227e-01
-7.70501435e-01 -3.47311422e-02 -5.36502004e-01 6.63752735e-01
1.74513817e-01 9.37858969e-02 -1.30822766e+00 -1.41215563e-01
-1.61110774e-01 -6.73224807e-01 1.56726086e+00 6.77574217e-01
1.54284441e+00 -4.34020907e-01 -1.89501375e-01 9.10674393e-01
1.18579364e+00 -1.38519809e-01 5.70941448e-01 4.01773304e-02
8.82785261e-01 7.28431523e-01 -3.15873891e-01 2.08896041e-01
-9.83952507e-02 3.50411564e-01 2.69896567e-01 -7.58464038e-02
-7.31912553e-02 1.37129351e-01 1.50209710e-01 8.24957132e-01
-4.57608193e-01 4.05518323e-01 -1.06429052e+00 4.38088298e-01
-1.78516400e+00 -9.18222845e-01 1.33334577e-01 2.09121037e+00
9.79984760e-01 3.72502685e-01 -2.79545516e-01 -8.82575139e-02
2.32894570e-01 9.46255624e-02 -7.21137822e-01 -3.96681368e-01
-1.08868726e-01 4.31713551e-01 6.40483439e-01 6.18235290e-01
-9.48578060e-01 8.68210256e-01 7.76450968e+00 6.46241248e-01
-9.48900223e-01 -3.06163698e-01 7.08826363e-01 -7.51829371e-02
-2.15158761e-01 1.37743324e-01 -1.18276286e+00 -1.18073523e-01
8.77544343e-01 4.39041555e-01 5.95898926e-01 9.55802858e-01
-1.67683542e-01 -6.63653314e-02 -1.38948417e+00 8.79777551e-01
-9.84737203e-02 -1.75773382e+00 -1.70781333e-02 2.69904256e-01
7.62149811e-01 5.16517699e-01 3.35193276e-01 1.92494407e-01
9.63275373e-01 -1.38526905e+00 3.12826484e-01 5.80823481e-01
4.71447438e-01 -4.51355428e-01 3.59044343e-01 3.05135459e-01
-8.62688243e-01 -9.66854915e-02 -1.22865605e+00 -1.39957532e-01
-3.61041337e-01 7.81814158e-01 -8.20371211e-01 -4.44514334e-01
7.04739273e-01 8.24326396e-01 -6.58939242e-01 1.05069160e+00
-7.62155354e-02 8.64974499e-01 -4.02604222e-01 -1.02629781e-01
3.57788652e-01 -4.93926555e-02 -7.23436289e-03 1.37549508e+00
6.65426105e-02 2.01041162e-01 2.70742718e-02 1.02081418e+00
-2.85536647e-01 -2.13943601e-01 -7.30403125e-01 -6.40223101e-02
1.77168310e-01 1.26244938e+00 -8.33041966e-01 -4.03029889e-01
-5.72200954e-01 5.82523823e-01 8.76139224e-01 6.92054868e-01
-1.84024379e-01 -9.23080221e-02 9.25190270e-01 1.46938721e-02
6.13722086e-01 -5.98123789e-01 -5.49855173e-01 -1.28557479e+00
-1.97340846e-01 -6.41935468e-01 9.13263038e-02 -5.05604804e-01
-1.62087071e+00 5.74798167e-01 -8.12374353e-02 -7.89685786e-01
-4.57128175e-02 -1.50229955e+00 -4.59677041e-01 9.77844596e-01
-1.35156500e+00 -8.41999292e-01 1.17406547e-01 5.67849874e-01
4.81902570e-01 -5.15507221e-01 9.79161501e-01 -5.21374047e-02
-3.18093032e-01 4.24423128e-01 1.04760423e-01 5.49923122e-01
5.76981306e-01 -1.32424891e+00 4.69387501e-01 5.00213206e-01
6.33196712e-01 1.18107271e+00 4.42619294e-01 -2.68473148e-01
-1.13911283e+00 -9.64826405e-01 4.26097035e-01 -6.09919012e-01
7.19424367e-01 -4.62215126e-01 -1.31304646e+00 9.33160543e-01
1.14433885e-01 6.92195415e-01 9.53527510e-01 8.21137786e-01
-1.12263870e+00 -8.76966342e-02 -6.28990769e-01 5.16587138e-01
8.54793131e-01 -6.81291580e-01 -6.17621481e-01 1.69301823e-01
3.49838018e-01 -6.91097155e-02 -5.08112311e-01 2.57077157e-01
6.65321231e-01 -6.83935940e-01 1.03281558e+00 -1.19306338e+00
1.56036869e-01 -6.73334152e-02 -3.28010231e-01 -1.46234679e+00
-6.87245786e-01 -3.01564693e-01 -1.34363294e-01 7.13569939e-01
6.42528534e-01 -5.04066467e-01 9.57635820e-01 9.94799137e-01
-1.62600651e-01 -7.84137607e-01 -5.31028807e-01 -8.63176525e-01
4.00544852e-01 -5.71351647e-01 -6.69771954e-02 9.03372765e-01
-2.16189250e-01 4.40858364e-01 -1.28392369e-01 2.61455774e-01
8.41758907e-01 -4.72214639e-01 4.33195859e-01 -1.53848481e+00
-4.79852557e-01 -6.72832906e-01 -4.83357787e-01 -1.39195991e+00
3.14018339e-01 -9.29389417e-01 1.18671499e-01 -1.15240061e+00
2.79893547e-01 -6.72137141e-01 -4.31036115e-01 8.82108212e-01
3.22517492e-02 3.35115492e-01 -4.03605551e-02 2.26778343e-01
-3.66696030e-01 3.64674181e-01 9.58779097e-01 -1.58299938e-01
-3.28813568e-02 -5.71899973e-02 -5.67926705e-01 1.38600266e+00
8.46059740e-01 -6.11478806e-01 -4.04342413e-01 -8.83973002e-01
4.93985891e-01 -8.09401572e-01 4.94952589e-01 -9.45754766e-01
3.68364781e-01 -3.51159990e-01 9.14405227e-01 -8.90478566e-02
3.95888418e-01 -7.45424092e-01 -3.59082967e-01 1.11623630e-01
-8.16284120e-01 -3.01914662e-01 3.23751181e-01 6.01254404e-01
-6.72741234e-02 -4.89295602e-01 1.08727288e+00 -3.47262710e-01
-6.72076225e-01 3.94804597e-01 -7.33159065e-01 1.93026260e-01
3.92374247e-01 1.55112026e-02 -1.56796515e-01 3.49000171e-02
-1.02978861e+00 -4.55600433e-02 2.34670281e-01 2.27554277e-01
8.22909474e-01 -1.46762228e+00 -3.58840644e-01 3.97863358e-01
-6.48944974e-02 1.76643461e-01 -2.00311258e-01 4.77980316e-01
-2.91021079e-01 2.01489404e-01 -3.75015765e-01 -6.42769396e-01
-9.15182412e-01 3.05861801e-01 4.19205219e-01 -6.48717061e-02
-5.74753404e-01 1.36246645e+00 4.62748021e-01 -4.00722742e-01
5.67570925e-01 -8.04506719e-01 -3.20228301e-02 -1.25268415e-01
7.73592412e-01 -1.81334764e-01 -3.37615646e-02 -1.96727797e-01
-2.07921252e-01 6.06366634e-01 -4.78318006e-01 -1.52342305e-01
1.68190753e+00 3.49801958e-01 -1.23269357e-01 5.01018107e-01
1.25413322e+00 -2.75351197e-01 -1.86418569e+00 -6.57954752e-01
3.95963341e-01 -4.41764116e-01 3.34943533e-01 -3.99124116e-01
-1.08849227e+00 1.29027033e+00 2.76635647e-01 8.52846131e-02
6.17645621e-01 1.35839954e-01 1.97015911e-01 1.18282878e+00
1.50405169e-01 -1.14658511e+00 4.96282011e-01 1.01084745e+00
8.33069086e-01 -1.38879430e+00 -1.00365758e-01 -1.16799198e-01
-3.96800011e-01 1.43612814e+00 4.81397033e-01 -6.04123116e-01
1.15677476e+00 5.42289019e-01 2.20101904e-02 -1.90203056e-01
-9.84980881e-01 -2.21517697e-01 7.17160940e-01 7.24641562e-01
6.34627044e-01 -2.26573765e-01 4.92397606e-01 3.82139623e-01
1.83700040e-01 -2.02822790e-01 2.20951885e-01 7.04240620e-01
-8.00413251e-01 -1.10109270e+00 -1.99083298e-01 6.00230217e-01
-4.00209904e-01 -4.32737231e-01 -2.58502722e-01 2.95810074e-01
1.26835242e-01 4.09328282e-01 2.41777003e-01 -2.28087962e-01
-6.56538233e-02 2.82217234e-01 7.10754573e-01 -1.11902380e+00
-1.58165261e-01 9.13883001e-02 -2.18492553e-01 -5.14611602e-01
-4.81493324e-01 -3.33315104e-01 -1.25620925e+00 -3.46875601e-02
-2.46837765e-01 -1.70996785e-01 5.67499757e-01 1.01902807e+00
9.41109434e-02 4.43042994e-01 3.23109061e-01 -1.13504934e+00
-6.23914063e-01 -1.01529884e+00 -5.81692517e-01 1.29459530e-01
5.41057646e-01 -7.01231480e-01 -4.22215015e-01 5.82002521e-01] | [9.278127670288086, 2.7124645709991455] |
5566cd02-570e-439f-93de-0116a8a7b248 | meta-learners-for-estimation-of-causal | 2201.12692 | null | https://arxiv.org/abs/2201.12692v1 | https://arxiv.org/pdf/2201.12692v1.pdf | Meta-Learners for Estimation of Causal Effects: Finite Sample Cross-Fit Performance | Estimation of causal effects using machine learning methods has become an active research field in econometrics. In this paper, we study the finite sample performance of meta-learners for estimation of heterogeneous treatment effects under the usage of sample-splitting and cross-fitting to reduce the overfitting bias. In both synthetic and semi-synthetic simulations we find that the performance of the meta-learners in finite samples greatly depends on the estimation procedure. The results imply that sample-splitting and cross-fitting are beneficial in large samples for bias reduction and efficiency of the meta-learners, respectively, whereas full-sample estimation is preferable in small samples. Furthermore, we derive practical recommendations for application of specific meta-learners in empirical studies depending on particular data characteristics such as treatment shares and sample size. | ['Gabriel Okasa'] | 2022-01-30 | null | null | null | null | ['econometrics'] | ['miscellaneous'] | [-7.74498284e-02 1.37296543e-01 -1.07978058e+00 -4.20724303e-01
-8.65181029e-01 -2.66008884e-01 4.61729765e-01 3.17293882e-01
-5.26577175e-01 9.86323893e-01 4.64616984e-01 -5.95127404e-01
-5.47772884e-01 -6.50407195e-01 -7.51112223e-01 -6.33392274e-01
-2.15768948e-01 2.48584032e-01 -2.65071005e-01 2.72486567e-01
4.71832365e-01 1.40568316e-01 -1.33186722e+00 1.28392607e-01
1.54973829e+00 2.41403118e-01 -9.06435251e-02 2.31149241e-01
2.08104268e-01 3.99102420e-01 -4.01277274e-01 -3.96847725e-01
1.88804120e-01 -5.18779755e-01 -4.52431947e-01 1.27454594e-01
3.69008154e-01 -1.95054010e-01 -2.64044479e-02 9.05530155e-01
6.54827416e-01 4.30452116e-02 1.21713257e+00 -1.26593864e+00
-3.84141922e-01 1.07544291e+00 -9.31272328e-01 1.09937191e-01
-7.40083260e-03 2.31710985e-01 7.74942935e-01 -5.14809906e-01
3.29200774e-01 1.63398588e+00 7.15703964e-01 4.66780365e-02
-1.28764510e+00 -1.12051570e+00 3.53996418e-02 1.44948587e-02
-9.66778457e-01 -4.53300923e-01 4.89333749e-01 -6.88038111e-01
1.65746108e-01 1.03098050e-01 5.06545603e-01 1.10022414e+00
4.45286989e-01 4.91756320e-01 1.53135574e+00 -7.08782077e-01
4.85669881e-01 4.18229640e-01 2.78373033e-01 1.04905359e-01
7.67599285e-01 4.66385573e-01 -4.90380153e-02 -6.30195796e-01
1.04932177e+00 -1.41389579e-01 -4.94974442e-02 -3.94800842e-01
-8.38232040e-01 1.37252045e+00 1.82526201e-01 5.37368916e-02
-6.97665632e-01 1.66745499e-01 6.51758552e-01 2.46844903e-01
8.52710962e-01 4.87150759e-01 -6.31011903e-01 2.36251011e-01
-8.91865492e-01 5.02262115e-01 4.37977701e-01 5.72341144e-01
2.82958090e-01 3.29423174e-02 -3.41538638e-01 9.81304049e-01
6.88302591e-02 5.94849586e-01 5.54677844e-01 -1.18889678e+00
7.02584922e-01 4.01503295e-01 5.14858007e-01 -5.98906696e-01
-4.19737190e-01 -3.82167906e-01 -9.12702799e-01 2.07499377e-02
9.17375922e-01 -7.76648343e-01 -5.97702980e-01 1.94576001e+00
5.80070615e-01 2.97240913e-01 -1.93603098e-01 5.52417994e-01
3.12221140e-01 6.79052830e-01 6.53318942e-01 -7.99167097e-01
1.31774235e+00 -4.63798374e-01 -7.40365803e-01 -8.26043710e-02
9.61905003e-01 -4.48390484e-01 1.02682984e+00 1.35430217e-01
-1.31336117e+00 -5.67885876e-01 -5.68364322e-01 3.58123332e-01
6.40501007e-02 1.63444981e-01 7.29332387e-01 8.77516627e-01
-3.10440302e-01 6.90228283e-01 -3.46963853e-01 -3.12210768e-01
5.41276395e-01 5.54503918e-01 -8.91391784e-02 2.73231268e-02
-1.20068538e+00 6.55157566e-01 3.15635294e-01 -2.24581540e-01
-6.47278070e-01 -1.38884556e+00 -6.41286135e-01 3.68290126e-01
3.37779224e-01 -1.04952633e+00 1.22780430e+00 -1.23966002e+00
-1.27221239e+00 4.38016683e-01 1.10745952e-01 -3.74133766e-01
6.62315786e-01 8.89508873e-02 -5.12959622e-02 -3.29397589e-01
1.97529674e-01 4.70760077e-01 7.03407466e-01 -1.24872541e+00
-9.90492702e-01 -7.08165228e-01 -9.46905240e-02 2.17654392e-01
-1.63799897e-01 1.38813242e-01 2.34947652e-01 -7.04994559e-01
-2.65200943e-01 -7.62462795e-01 -5.26726186e-01 -8.61695886e-01
-1.42860040e-01 -3.69141072e-01 2.66328335e-01 -3.97170037e-01
1.35581040e+00 -1.92865419e+00 -2.23895654e-01 2.72458136e-01
-2.36984678e-02 -8.13679919e-02 -9.03741941e-02 4.14851040e-01
-4.60537821e-01 9.39799771e-02 3.17357853e-02 2.48487860e-01
-1.78362012e-01 -3.32677126e-01 -4.07753065e-02 7.22817183e-01
-2.02624738e-01 4.57025647e-01 -6.51268840e-01 -6.50192320e-01
2.87322640e-01 -1.22291399e-02 -8.68309140e-01 7.48829767e-02
3.51537496e-01 2.83914238e-01 -5.60445964e-01 1.00871630e-01
8.48632514e-01 -1.19560167e-01 5.36307216e-01 6.07889239e-03
-2.77955770e-01 5.09782910e-01 -1.31764650e+00 8.13510120e-01
-8.10040295e-01 3.27932358e-01 1.19481660e-01 -1.49694145e+00
3.78970563e-01 4.20484781e-01 2.42647007e-01 -6.30076766e-01
1.38675168e-01 1.92821816e-01 4.39997703e-01 -6.65692151e-01
2.10748985e-01 -6.54423594e-01 -6.43416366e-04 2.68487513e-01
-1.58574522e-01 1.34046480e-01 3.22693139e-01 -2.05000311e-01
3.78776073e-01 -2.63210326e-01 6.47025526e-01 -7.58092582e-01
1.46673426e-01 -2.67524123e-01 6.20852053e-01 9.55537081e-01
8.55163336e-02 2.18243292e-03 8.28673542e-01 3.30028906e-02
-1.09445274e+00 -6.58442318e-01 -5.86596012e-01 1.21093202e+00
-3.28747004e-01 3.31873208e-01 -9.04690862e-01 -6.91516817e-01
4.68765467e-01 1.21313453e+00 -7.77637124e-01 -2.62798727e-01
-2.93440789e-01 -1.24504888e+00 2.53272265e-01 6.78518832e-01
2.87490413e-02 -9.21555638e-01 -4.41880405e-01 -1.18013192e-03
6.70074373e-02 -4.71095771e-01 -2.60065287e-01 1.78849012e-01
-1.37632608e+00 -1.04376554e+00 -8.92504811e-01 -4.54253107e-01
5.68202674e-01 3.27017367e-01 8.00076067e-01 -2.71847546e-01
5.38920224e-01 1.03910707e-01 -5.27656376e-02 -7.17551947e-01
-5.86581171e-01 1.31181121e-01 -2.19276641e-02 -2.67188519e-01
2.09560201e-01 -5.61857879e-01 -5.74882269e-01 1.28387481e-01
-6.09012961e-01 -1.95638448e-01 5.55632353e-01 1.08260965e+00
-8.08604434e-02 2.06137449e-01 1.04230249e+00 -1.41693020e+00
7.17569232e-01 -7.40702987e-01 -9.06178355e-01 2.77225763e-01
-7.45142639e-01 1.07606344e-01 4.59474623e-01 -8.75225842e-01
-1.56789196e+00 -7.22089410e-01 3.65498215e-01 5.88066466e-02
-1.07745163e-01 8.13544214e-01 -2.58293092e-01 3.41590554e-01
7.14887083e-01 -5.92061400e-01 1.47587592e-02 -3.11676651e-01
3.49879600e-02 7.70869911e-01 -1.64880291e-01 -7.41072237e-01
2.74469525e-01 1.08538538e-01 3.40085663e-02 -6.12016559e-01
-7.62011647e-01 -2.02405989e-01 -2.69684881e-01 1.05510857e-02
5.55345595e-01 -1.07830024e+00 -7.70206988e-01 2.41602868e-01
-5.72177172e-01 -7.32992470e-01 -2.24824741e-01 1.14424276e+00
-4.78482157e-01 6.95806667e-02 -3.98984700e-01 -9.99376476e-01
3.03954799e-02 -1.33103347e+00 8.42153251e-01 3.29171211e-01
-3.02839369e-01 -1.44073308e+00 1.79406732e-01 4.11057651e-01
-2.43190005e-01 5.27760610e-02 1.29584658e+00 -6.60234332e-01
4.18925025e-02 -2.92644978e-01 -1.52922481e-01 -1.90166116e-01
1.45582408e-01 9.20713246e-02 -9.21392024e-01 -3.12571883e-01
-2.99415886e-01 -2.34120771e-01 6.84337318e-01 1.69977498e+00
1.36060536e+00 -4.22781229e-01 -6.29684567e-01 2.38899574e-01
1.28243959e+00 3.92470062e-01 6.39860809e-01 2.13846907e-01
3.65741611e-01 1.19172800e+00 8.23456824e-01 7.26069868e-01
3.65835458e-01 4.31632072e-01 3.68334912e-02 -1.77024320e-01
4.96083230e-01 -3.73704642e-01 1.46816790e-01 2.26705045e-01
-1.13471083e-01 -1.59336880e-01 -6.80881560e-01 4.49408054e-01
-1.84183276e+00 -1.13533449e+00 -5.26021242e-01 2.82956457e+00
1.00597203e+00 6.95780143e-02 7.11437166e-01 -5.52310497e-02
1.07415295e+00 -2.22071022e-01 -1.81488290e-01 -4.89196330e-01
2.08758593e-01 2.20488664e-02 9.94336545e-01 5.67947686e-01
-9.22213435e-01 5.12956381e-01 7.08177376e+00 1.17412400e+00
-9.31167483e-01 2.08564445e-01 1.08421862e+00 1.42646730e-01
-4.00552422e-01 3.40799004e-01 -6.93134844e-01 5.79139829e-01
1.35822296e+00 -2.17507049e-01 -1.50016621e-01 5.49783051e-01
9.21640515e-01 -5.17387629e-01 -9.59840417e-01 4.38197762e-01
-6.57289088e-01 -1.15964985e+00 -1.97887823e-01 3.94030333e-01
1.17699075e+00 -6.42395437e-01 6.52700141e-02 5.42301118e-01
4.27903205e-01 -1.04515553e+00 5.65298438e-01 6.18647486e-02
5.92810333e-01 -9.94436026e-01 1.01173401e+00 5.67421079e-01
-3.87645185e-01 -6.65786207e-01 -6.04432940e-01 -2.54832357e-01
-8.71692076e-02 9.98594701e-01 -5.79627752e-01 4.23273593e-01
4.27971810e-01 2.07740143e-01 -2.89158791e-01 1.11527312e+00
3.10969502e-01 1.29042375e+00 -6.90856352e-02 1.34540021e-01
2.97493041e-02 -5.52810371e-01 -3.97778861e-02 1.02756202e+00
5.03466964e-01 2.29515716e-01 -1.90052226e-01 6.54889107e-01
9.09443870e-02 3.21153790e-01 -5.40914237e-01 3.15414555e-02
7.75008678e-01 8.35184038e-01 -4.90001559e-01 -3.92643034e-01
-3.73695165e-01 -3.00178770e-02 1.12753496e-01 4.28630650e-01
-7.46700823e-01 -3.72141646e-03 2.69651681e-01 4.12015289e-01
1.20998817e-02 4.57153022e-01 -7.75089562e-01 -8.26509356e-01
-5.60757458e-01 -1.08977461e+00 7.19059169e-01 -4.39988852e-01
-1.17524898e+00 -6.07477129e-01 8.16428006e-01 -9.02086437e-01
-2.53993779e-01 -2.56759852e-01 -8.25890303e-01 6.82486653e-01
-9.17101264e-01 -7.83576429e-01 2.11289406e-01 6.35901093e-03
4.83755350e-01 1.33922398e-01 2.68558979e-01 2.28433967e-01
-8.43189478e-01 7.54704237e-01 4.29644138e-01 -3.55316907e-01
8.99831355e-01 -1.05887806e+00 -3.75717938e-01 2.55918890e-01
-5.22527575e-01 7.15226889e-01 1.02225971e+00 -7.98748195e-01
-8.21071088e-01 -6.84311032e-01 8.24404180e-01 1.09123103e-01
5.20351589e-01 -6.85499385e-02 -8.17821622e-01 7.00297117e-01
4.37308431e-01 -4.59254146e-01 6.81857109e-01 9.18397009e-01
-8.89534727e-02 -3.21904987e-01 -1.23816550e+00 7.39935815e-01
5.66522777e-01 -2.41392036e-03 -3.36471647e-01 2.81102568e-01
3.17366481e-01 -2.91586779e-02 -1.08766437e+00 5.89872837e-01
6.33484304e-01 -1.09587801e+00 8.32861423e-01 -8.98544431e-01
6.13637030e-01 5.18826425e-01 2.86612093e-01 -1.66284633e+00
-6.27626300e-01 -1.17647909e-01 6.79224133e-01 1.39854670e+00
4.84663635e-01 -6.41987085e-01 7.00818837e-01 5.85816205e-01
2.04725131e-01 -4.83836710e-01 -6.98780715e-01 -5.29188514e-01
8.19922686e-01 -3.09921861e-01 5.22401989e-01 1.18618631e+00
6.16124906e-02 4.66351658e-01 -3.57152909e-01 4.91879322e-02
7.97351301e-01 2.02207953e-01 8.95362496e-01 -1.31769669e+00
-2.58024544e-01 -4.31499511e-01 3.37876856e-01 -4.16680127e-01
6.15254164e-01 -2.84753382e-01 -3.74683887e-01 -1.10491765e+00
6.40078366e-01 -5.39601028e-01 -5.18170521e-02 1.88416764e-02
-6.70029163e-01 -4.73967314e-01 -3.54851224e-03 -1.51768073e-01
2.32814923e-02 5.66446543e-01 1.22366905e+00 1.66194052e-01
-5.82455873e-01 7.04753876e-01 -6.40904248e-01 7.86471665e-01
8.38262737e-01 -7.68874645e-01 -5.94390154e-01 2.34054938e-01
-2.35757139e-02 8.24543178e-01 2.49175429e-01 -2.62204945e-01
-3.89600605e-01 -9.01522040e-01 3.63415480e-01 -3.22836131e-01
-3.48465860e-01 -7.40093589e-01 2.29583785e-01 5.94012082e-01
-8.40796471e-01 5.80192581e-02 2.60842800e-01 4.43764567e-01
1.22192159e-01 -5.90345442e-01 8.55783880e-01 -9.27729309e-02
1.73837304e-01 -2.00363725e-01 -6.16354108e-01 3.90776515e-01
7.73193955e-01 -2.69405544e-02 -1.83269493e-02 -6.29499197e-01
-1.01830967e-01 1.97889656e-01 1.35191083e-01 1.89905204e-02
-6.45279214e-02 -1.30723560e+00 -9.64469671e-01 -2.03589246e-01
-2.65962109e-02 -6.00582600e-01 5.14583290e-01 1.22080064e+00
4.34214659e-02 5.43990195e-01 -9.27199498e-02 -1.69712320e-01
-1.10726559e+00 6.01508141e-01 2.44404092e-01 -4.13901448e-01
1.59729123e-02 3.76258731e-01 8.37514162e-01 -4.28994805e-01
1.28524750e-01 -4.18521702e-01 -2.03866646e-01 3.47413599e-01
1.91704810e-01 1.05039001e+00 -3.20470661e-01 -1.07349582e-01
7.39463568e-02 1.53009593e-01 9.62440521e-02 -1.87378079e-01
1.28709006e+00 -1.75049007e-01 -4.97658327e-02 6.19383752e-01
9.48206425e-01 2.82225519e-01 -1.18214178e+00 8.25879872e-02
7.60428309e-02 -5.79664648e-01 2.42257744e-01 -4.84099120e-01
-8.74253869e-01 5.09871185e-01 4.99450922e-01 1.94823429e-01
1.00508523e+00 -3.71522993e-01 -1.05797574e-01 -6.94684833e-02
1.63277686e-01 -1.26850295e+00 -3.06347817e-01 -1.14074118e-01
6.24768376e-01 -1.43106902e+00 4.28757399e-01 -3.73049945e-01
-6.37217045e-01 7.31978357e-01 4.92201567e-01 -2.70086855e-01
7.06973612e-01 -9.61461216e-02 -2.08159640e-01 5.63970879e-02
-9.88007367e-01 1.89006224e-01 4.18634713e-01 5.05140901e-01
8.04975569e-01 3.05370480e-01 -1.12971103e+00 5.24144411e-01
6.43990859e-02 5.11475839e-03 6.20804906e-01 3.09159786e-01
-1.43844217e-01 -1.07205236e+00 -8.42633128e-01 7.90585876e-01
-6.45304680e-01 -2.53023542e-02 -4.44646813e-02 1.49870133e+00
9.90321487e-02 9.17816281e-01 2.59319544e-01 2.45564044e-01
3.11122835e-01 3.52966636e-02 6.23878598e-01 -4.03057516e-01
-7.32954085e-01 5.63647449e-01 2.87188083e-01 2.84112003e-02
-7.52978325e-01 -1.01855803e+00 -7.83603072e-01 -5.02099216e-01
-8.67072940e-01 4.20068175e-01 1.61822274e-01 1.07895648e+00
1.37351379e-01 2.04010099e-01 8.97781253e-01 -6.52391911e-01
-1.26280236e+00 -1.42004001e+00 -8.72609377e-01 2.39481598e-01
4.07413989e-02 -9.44987416e-01 -5.36273956e-01 -5.33718109e-01] | [7.954464435577393, 5.227576732635498] |
d823a1f1-caa0-426b-b466-cb7a70751a0b | improving-negation-detection-with-negation-1 | 2205.04012 | null | https://arxiv.org/abs/2205.04012v1 | https://arxiv.org/pdf/2205.04012v1.pdf | Improving negation detection with negation-focused pre-training | Negation is a common linguistic feature that is crucial in many language understanding tasks, yet it remains a hard problem due to diversity in its expression in different types of text. Recent work has shown that state-of-the-art NLP models underperform on samples containing negation in various tasks, and that negation detection models do not transfer well across domains. We propose a new negation-focused pre-training strategy, involving targeted data augmentation and negation masking, to better incorporate negation information into language models. Extensive experiments on common benchmarks show that our proposed approach improves negation detection performance and generalizability over the strong baseline NegBERT (Khandewal and Sawant, 2020). | ['Karin Verspoor', 'Trevor Cohn', 'Timothy Baldwin', 'Thinh Hung Truong'] | 2022-05-09 | null | https://aclanthology.org/2022.naacl-main.309 | https://aclanthology.org/2022.naacl-main.309.pdf | naacl-2022-7 | ['negation-detection'] | ['natural-language-processing'] | [ 6.85082823e-02 -9.72099081e-02 -6.15034878e-01 -7.27080405e-01
-5.19591808e-01 -8.04544091e-01 6.63315177e-01 4.22203898e-01
-7.53598392e-01 1.16719270e+00 2.67049134e-01 -4.59908664e-01
4.64967221e-01 -8.54272962e-01 -7.07884431e-01 -7.71382526e-02
1.96716160e-01 4.06846553e-01 3.07761550e-01 -9.13932562e-01
2.75226366e-02 3.81970525e-01 -1.24155700e+00 6.71836913e-01
1.02765369e+00 6.28472924e-01 -3.62973452e-01 4.11108017e-01
-5.60213923e-01 1.02581811e+00 -6.83682382e-01 -7.79818296e-01
6.91262111e-02 -3.52390945e-01 -6.68698609e-01 -7.02950776e-01
5.13597429e-01 7.21378904e-03 -4.51391637e-02 1.44384289e+00
3.63659650e-01 -1.21785082e-01 6.10521197e-01 -1.07816732e+00
-9.31995153e-01 8.86267245e-01 -5.55019975e-01 3.93518686e-01
4.65308666e-01 4.55258600e-03 1.41312969e+00 -1.13675535e+00
6.55678451e-01 1.50437140e+00 8.18623245e-01 8.74378741e-01
-8.53905797e-01 -8.67317021e-01 3.22208524e-01 2.65088588e-01
-1.15282583e+00 -1.97526544e-01 6.72239184e-01 -6.58282340e-02
1.62681890e+00 -1.77690700e-01 4.18629527e-01 1.19890440e+00
4.21474010e-01 9.84960914e-01 1.00756705e+00 -6.23363018e-01
-2.17846781e-02 4.16843072e-02 2.30739802e-01 6.07394040e-01
5.82938969e-01 -2.49282777e-01 -7.35509396e-01 -2.64781445e-01
7.20600411e-02 -3.74948233e-01 -3.40670824e-01 7.90807530e-02
-1.08589840e+00 9.48120475e-01 6.76395595e-02 4.81217206e-01
-3.33962709e-01 -3.38921994e-02 9.77961838e-01 7.55353212e-01
5.53023517e-01 3.84827435e-01 -1.20810556e+00 -1.60988957e-01
-7.26023793e-01 5.68446100e-01 1.20372343e+00 8.60941231e-01
6.29541397e-01 -1.06278904e-01 -1.72550678e-01 7.08413720e-01
1.90578848e-01 7.13222980e-01 4.25697386e-01 -5.93978345e-01
7.53096759e-01 9.25574660e-01 -7.57986307e-02 -7.15713084e-01
-3.88519347e-01 -3.26526910e-01 -8.80038738e-01 -2.15837732e-01
3.09235394e-01 -4.01181489e-01 -9.22306657e-01 1.90206325e+00
8.86833146e-02 -4.60253991e-02 6.49391174e-01 4.33609962e-01
9.95264173e-01 5.24203360e-01 3.15295517e-01 -3.50028165e-02
1.26817226e+00 -7.85958767e-01 -1.20901775e+00 -8.98069561e-01
1.23918962e+00 -7.02320874e-01 1.13288987e+00 6.79825485e-01
-9.14761066e-01 -1.78614706e-01 -1.13521504e+00 -2.55108505e-01
-7.15773284e-01 -1.67300671e-01 9.23214734e-01 7.99612641e-01
-5.95149934e-01 1.72898635e-01 -4.87404644e-01 -3.40352505e-01
5.16734004e-01 2.42645204e-01 -6.65950775e-01 -4.86114264e-01
-1.90367532e+00 1.15254462e+00 9.34036374e-01 1.69120193e-01
-2.98814058e-01 -5.69985569e-01 -1.23900032e+00 -1.17813759e-01
5.65096378e-01 -4.60550398e-01 1.28933406e+00 -8.00586343e-01
-1.10401666e+00 1.07994509e+00 -4.34787989e-01 -6.97355151e-01
3.32820266e-01 -6.27343357e-01 -6.72719955e-01 -2.22323298e-01
1.86409373e-02 6.87929511e-01 3.84644449e-01 -8.45171988e-01
-6.99093342e-01 -5.06428666e-02 2.12484241e-01 3.02292496e-01
-4.05191392e-01 3.65768939e-01 -2.79803067e-01 -5.27281821e-01
-1.30622670e-01 -4.32612360e-01 -1.22714788e-01 -9.36310887e-02
-3.48270744e-01 -7.33751416e-01 9.14757550e-01 -3.63516569e-01
1.14136446e+00 -1.89171553e+00 -2.09209308e-01 4.55904286e-03
-1.48785576e-01 7.31291950e-01 -1.94820046e-01 4.22412038e-01
-1.32735446e-01 3.35993916e-01 -3.03089529e-01 -2.50566870e-01
7.76357204e-02 8.24647009e-01 -4.65487421e-01 2.25647941e-01
6.28447354e-01 1.14278948e+00 -1.14491987e+00 -4.06829596e-01
-1.16437592e-01 3.54037166e-01 -6.91327214e-01 -1.48862898e-01
-7.02833772e-01 2.98774000e-02 -1.50942668e-01 1.19767058e+00
8.06933224e-01 -1.29022123e-02 3.07662159e-01 -2.13045076e-01
2.79446363e-01 5.77404141e-01 -1.13628519e+00 1.44239974e+00
-6.68459907e-02 4.02527124e-01 1.15740560e-01 -1.12935412e+00
8.13624680e-01 4.06348020e-01 -2.81546772e-01 -7.42546201e-01
2.21104711e-01 5.15664816e-01 2.69563198e-01 -4.69369173e-01
4.51667458e-01 -5.18008411e-01 -3.64729762e-01 4.33328673e-02
2.15571582e-01 -4.74995553e-01 6.75270557e-01 5.74039817e-02
1.14804077e+00 -1.52287915e-01 7.49404550e-01 -3.06487679e-01
1.03781450e+00 1.39586389e-01 9.61757660e-01 8.35494637e-01
-5.50228953e-01 2.65519053e-01 8.03428948e-01 -2.26943135e-01
-4.95108366e-01 -7.27123260e-01 -1.86962448e-02 1.17954838e+00
-2.76300341e-01 -4.43019122e-01 -2.39781097e-01 -1.18997777e+00
1.97851837e-01 9.17250335e-01 -5.97713768e-01 -2.82926828e-01
-9.48900759e-01 -7.79583454e-01 1.10439634e+00 7.43222713e-01
5.60636342e-01 -1.39394701e+00 1.45996958e-01 1.90803304e-01
-2.00281501e-01 -1.42676044e+00 -8.69213641e-02 7.25699723e-01
-6.66715741e-01 -1.09057426e+00 -1.61093399e-01 -1.13125241e+00
4.03550386e-01 -3.58535737e-01 1.49893391e+00 8.27483181e-03
1.65147498e-01 7.03803748e-02 -2.88925737e-01 -8.08850706e-01
-6.53529823e-01 9.62328389e-02 1.15481727e-02 -6.71207905e-01
1.07978594e+00 -1.14359953e-01 3.41373496e-03 -2.04146028e-01
-9.73056555e-01 -6.51347697e-01 6.12394512e-01 1.15860665e+00
5.39089799e-01 4.69833128e-02 8.19226742e-01 -1.39130116e+00
1.00943625e+00 -4.22930241e-01 -4.41647559e-01 2.82907248e-01
-2.70453990e-01 -7.33502954e-02 7.79055059e-01 -2.79384524e-01
-9.54374194e-01 -2.10501090e-01 -4.33493376e-01 3.67955770e-03
-9.44878086e-02 1.03793097e+00 -4.00169134e-01 -6.91013634e-02
5.83799720e-01 8.85771662e-02 -3.02073747e-01 -2.21441835e-01
5.14636457e-01 4.15905029e-01 4.59098577e-01 -5.63064814e-01
5.93494654e-01 3.91286403e-01 -8.35151598e-02 -8.71571243e-01
-1.47913587e+00 -4.58570063e-01 -6.38546824e-01 5.37042737e-01
4.00598228e-01 -1.14719820e+00 -4.15780813e-01 6.64625406e-01
-1.43003559e+00 -1.18516624e-01 -2.80924678e-01 3.07698935e-01
8.46125782e-02 4.63546306e-01 -8.22594225e-01 -5.98053038e-01
-6.59551203e-01 -4.74002063e-01 7.38747954e-01 1.22865677e-01
-3.18386316e-01 -1.29406428e+00 2.32064471e-01 3.03936660e-01
2.10777938e-01 -1.39381588e-01 1.17865753e+00 -1.30671537e+00
-4.37990651e-02 -3.09407920e-01 -2.53641158e-01 1.10000515e+00
1.96219042e-01 -1.79211810e-01 -8.03972483e-01 -1.30781040e-01
1.53687865e-01 -8.51509750e-01 1.05686474e+00 1.51197175e-02
8.11330795e-01 -1.11877285e-01 -2.68371642e-01 1.79822698e-01
1.27742255e+00 -1.66416571e-01 4.69878703e-01 3.67378354e-01
5.52521408e-01 3.86194706e-01 8.73431385e-01 1.52522745e-02
4.20861125e-01 -2.41161183e-01 3.69426161e-01 -7.51457661e-02
1.16857037e-01 -8.76443982e-02 4.75984454e-01 6.93634927e-01
5.48150659e-01 -6.91439748e-01 -1.12182212e+00 8.40880156e-01
-1.74250579e+00 -8.24751556e-01 -1.27037585e-01 1.49046528e+00
1.33443224e+00 7.88631856e-01 -6.02080286e-01 1.68208539e-01
3.57391864e-01 6.66105300e-02 -4.49660569e-01 -7.34661579e-01
-9.94760633e-01 5.00189900e-01 3.18386853e-01 5.74549854e-01
-1.31976032e+00 1.34619367e+00 7.22490263e+00 7.12932408e-01
-1.00361454e+00 3.82337645e-02 2.26363495e-01 2.58630574e-01
-3.59220654e-01 -1.43815681e-01 -1.19735801e+00 -9.46099386e-02
6.82061791e-01 -3.30583900e-01 -1.04676850e-01 6.87992930e-01
-5.77212172e-03 -1.02415979e-01 -1.31047726e+00 6.11000419e-01
4.89568830e-01 -1.14092338e+00 3.10211927e-01 -5.48095107e-01
8.96712065e-01 5.03365040e-01 -1.78740993e-01 8.76986027e-01
2.76972175e-01 -1.22985566e+00 3.33379447e-01 1.65370658e-01
2.73096532e-01 -8.71158302e-01 1.51692164e+00 6.29455030e-01
-9.28259611e-01 1.19870985e-02 -2.67088622e-01 -5.01885653e-01
1.76611304e-01 8.13533783e-01 -9.30478215e-01 4.55231220e-01
4.39291269e-01 8.29198837e-01 -4.76058394e-01 5.06969333e-01
-8.67185175e-01 8.23600888e-01 -4.93629903e-01 -2.87573367e-01
3.10296386e-01 1.16645373e-01 2.73141593e-01 1.69822764e+00
-1.16905987e-01 -6.41192272e-02 3.15744281e-01 6.65411472e-01
-6.36342108e-01 4.12756264e-01 -6.92835271e-01 -3.81957173e-01
3.39715242e-01 1.02840328e+00 -3.96531671e-01 -4.79557276e-01
-1.07416284e+00 6.98331594e-01 4.36810881e-01 2.73353308e-01
-6.24591768e-01 -6.03123605e-01 7.97364712e-01 -1.85894936e-01
6.35538280e-01 -1.35669455e-01 -2.29645342e-01 -1.36680830e+00
3.36628169e-01 -1.15708911e+00 5.38410604e-01 -4.64612722e-01
-1.89131474e+00 2.94026583e-01 -2.06989884e-01 -6.66008115e-01
-2.72765487e-01 -1.05577946e+00 -4.37248796e-01 9.88940537e-01
-2.21321988e+00 -1.22184432e+00 2.24806398e-01 5.79276562e-01
1.88901395e-01 -1.57887861e-02 1.00590611e+00 4.53056037e-01
-2.76373923e-01 7.26123512e-01 -2.28835464e-01 6.24647021e-01
1.07633233e+00 -1.33804131e+00 4.67148334e-01 9.31040406e-01
-5.04154451e-02 8.56140196e-01 7.92299390e-01 -9.34104741e-01
-1.35317647e+00 -1.10079861e+00 1.75580549e+00 -4.14315224e-01
1.07077479e+00 -4.76721525e-01 -1.40123498e+00 9.46056843e-01
4.69380051e-01 2.77738690e-01 7.39839673e-01 2.68803686e-01
-6.32367313e-01 1.20920897e-01 -1.36881065e+00 6.02851987e-01
8.92607927e-01 -8.74918103e-01 -1.28720832e+00 4.41287488e-01
7.53033698e-01 -4.80097264e-01 -4.39218879e-01 8.76722097e-01
8.18388462e-02 -6.85258090e-01 6.32547379e-01 -8.74465704e-01
2.76444376e-01 -2.85080522e-01 -3.33410680e-01 -9.44115341e-01
2.77851105e-01 -4.13923323e-01 -2.31334090e-01 1.25494897e+00
6.72129452e-01 -7.77406454e-01 9.25093353e-01 2.11329401e-01
3.02358419e-02 -7.41037309e-01 -8.86395872e-01 -7.06954598e-01
5.98315179e-01 -6.54758394e-01 1.04237035e-01 1.14686227e+00
1.02767363e-01 6.79919779e-01 -6.47547329e-03 1.92263663e-01
2.04600200e-01 -2.00437903e-01 3.97666574e-01 -1.20000005e+00
1.26051486e-01 -2.98728228e-01 -4.66314375e-01 -1.15956652e+00
7.78479099e-01 -1.00546813e+00 -8.84835701e-03 -1.59037673e+00
5.50287105e-02 4.87164184e-02 -4.16699708e-01 1.04350281e+00
-3.72768283e-01 2.41994634e-01 -3.10771286e-01 -4.78581429e-01
-6.56788409e-01 7.53339887e-01 1.06690133e+00 -3.46197009e-01
1.93530992e-02 -4.00783688e-01 -8.10005665e-01 1.19501293e+00
5.18108845e-01 -6.08323872e-01 -2.70591050e-01 -3.98549139e-01
8.64402235e-01 -5.86167514e-01 1.01268210e-01 -6.87450469e-01
2.90778428e-01 -1.44686729e-01 2.72333205e-01 -1.01783121e+00
3.26177210e-01 -6.08696461e-01 -9.41414654e-01 6.49012506e-01
-9.78576615e-02 2.62848049e-01 7.25397527e-01 4.34512109e-01
-6.45641208e-01 -4.27891493e-01 4.71322447e-01 -1.33802786e-01
-8.91447544e-01 -1.29742295e-01 -2.89707094e-01 8.84979784e-01
6.77229106e-01 9.14447457e-02 -4.24751908e-01 -1.33437485e-01
-4.21624124e-01 5.69597721e-01 1.84900686e-01 4.28528607e-01
5.36232650e-01 -9.28382397e-01 -9.03099060e-01 2.89641678e-01
1.38585985e-01 1.03836901e-01 -2.15377778e-01 5.10345101e-01
-4.15354908e-01 5.31139553e-01 2.40840465e-01 -5.24699807e-01
-1.29792738e+00 1.78033337e-01 3.94837260e-01 -6.30111694e-01
-2.35612035e-01 1.24112999e+00 -1.77837342e-01 -1.07447338e+00
3.41455460e-01 -8.45593929e-01 -3.89183015e-01 9.64316800e-02
5.14473200e-01 -2.21758723e-01 4.90562052e-01 -4.14367616e-01
-7.78853834e-01 2.90986747e-01 -4.29907680e-01 2.16252819e-01
1.12350738e+00 3.09398979e-01 -4.69112724e-01 5.74517429e-01
8.84606540e-01 2.39285246e-01 -2.76630461e-01 -4.87572730e-01
4.63935584e-01 -2.82846857e-02 -1.44295946e-01 -1.16602004e+00
-6.54607534e-01 6.90719128e-01 9.70184058e-02 -5.24895824e-02
1.10827863e+00 -1.44791827e-01 7.91383088e-01 1.28728056e+00
1.21858962e-01 -1.00862718e+00 -1.28445312e-01 1.59324503e+00
6.66652143e-01 -1.69836271e+00 -3.71427611e-02 -7.28242517e-01
-5.10621369e-01 9.65939701e-01 7.71638453e-01 1.18164578e-02
8.38088810e-01 6.78266048e-01 4.77452546e-01 -2.24519476e-01
-1.10722351e+00 -7.67086148e-02 4.98261750e-02 3.95417273e-01
8.96243811e-01 -2.30930954e-01 -5.47629356e-01 8.96488547e-01
-3.49910893e-02 2.54143447e-01 4.28260773e-01 1.51684189e+00
-3.03654730e-01 -1.50795853e+00 2.63434667e-02 4.04951602e-01
-8.21137905e-01 -6.30766571e-01 -5.49552977e-01 1.01923418e+00
3.37274373e-01 8.08260024e-01 -3.43405396e-01 1.84760571e-01
7.10045993e-01 4.85954732e-01 4.55645680e-01 -1.07606542e+00
-8.97884071e-01 -2.59032458e-01 4.19438750e-01 -3.80707473e-01
-5.59261084e-01 -4.62824106e-01 -1.65844834e+00 -2.05719396e-01
-4.83580589e-01 2.42642909e-01 2.53905416e-01 1.34377122e+00
8.57294425e-02 3.84721249e-01 -2.65369713e-01 -5.52399792e-02
-8.36807728e-01 -9.87626672e-01 -4.34645921e-01 4.02975827e-01
4.49103355e-01 -3.19514155e-01 -6.22235477e-01 -1.84693575e-01] | [10.400712966918945, 9.170385360717773] |
d63c30de-edc6-4323-a749-482927c7e7f0 | graph-self-supervised-learning-with-accurate | 2202.02989 | null | https://arxiv.org/abs/2202.02989v5 | https://arxiv.org/pdf/2202.02989v5.pdf | Graph Self-supervised Learning with Accurate Discrepancy Learning | Self-supervised learning of graph neural networks (GNNs) aims to learn an accurate representation of the graphs in an unsupervised manner, to obtain transferable representations of them for diverse downstream tasks. Predictive learning and contrastive learning are the two most prevalent approaches for graph self-supervised learning. However, they have their own drawbacks. While the predictive learning methods can learn the contextual relationships between neighboring nodes and edges, they cannot learn global graph-level similarities. Contrastive learning, while it can learn global graph-level similarities, its objective to maximize the similarity between two differently perturbed graphs may result in representations that cannot discriminate two similar graphs with different properties. To tackle such limitations, we propose a framework that aims to learn the exact discrepancy between the original and the perturbed graphs, coined as Discrepancy-based Self-supervised LeArning (D-SLA). Specifically, we create multiple perturbations of the given graph with varying degrees of similarity, and train the model to predict whether each graph is the original graph or the perturbed one. Moreover, we further aim to accurately capture the amount of discrepancy for each perturbed graph using the graph edit distance. We validate our D-SLA on various graph-related downstream tasks, including molecular property prediction, protein function prediction, and link prediction tasks, on which ours largely outperforms relevant baselines. | ['Sung Ju Hwang', 'Jinheon Baek', 'DongKi Kim'] | 2022-02-07 | null | null | null | null | ['protein-function-prediction'] | ['medical'] | [ 4.23271865e-01 3.89543027e-01 -4.66839463e-01 -3.32650065e-01
-3.74443561e-01 -4.50977862e-01 4.29521739e-01 8.59673142e-01
6.66544512e-02 6.14184797e-01 9.70166922e-02 -1.53996825e-01
-3.02592158e-01 -1.05609059e+00 -9.32935119e-01 -7.34875083e-01
-3.94151300e-01 3.92944694e-01 1.94756821e-01 -2.96600610e-01
9.96494889e-02 4.54411060e-01 -1.07747853e+00 1.30326882e-01
1.11215734e+00 6.30653620e-01 -1.08191473e-02 3.84288728e-01
-7.12330043e-02 9.65190113e-01 -2.84277916e-01 -2.43830070e-01
7.12309480e-02 -6.90131128e-01 -8.56796622e-01 5.27885463e-03
4.89846140e-01 2.64341086e-01 -6.91211343e-01 1.30331433e+00
2.73521215e-01 5.84560521e-02 9.14094210e-01 -1.47050202e+00
-8.90765190e-01 7.46823490e-01 -4.78811324e-01 2.86839515e-01
3.24134380e-01 7.41593912e-02 1.34076202e+00 -4.15123373e-01
6.96416020e-01 1.11544049e+00 7.14054406e-01 4.76726234e-01
-1.54426825e+00 -3.86576563e-01 3.52211356e-01 2.87991792e-01
-1.19342065e+00 -2.46469095e-01 1.03676736e+00 -4.44846630e-01
7.57133782e-01 1.52266294e-01 5.96956313e-01 1.04108822e+00
2.68821090e-01 4.37269211e-01 8.87849212e-01 -1.41991809e-01
3.17485929e-01 -3.32358807e-01 1.64080620e-01 1.00324023e+00
3.20506752e-01 5.25336675e-02 -4.59057838e-01 -1.94631770e-01
3.58378261e-01 1.70522839e-01 -5.05346656e-01 -7.20289230e-01
-1.05486286e+00 6.91244304e-01 8.44924450e-01 4.34495896e-01
-1.50656030e-01 3.28889601e-02 4.65915561e-01 5.98861098e-01
6.09187722e-01 6.26947701e-01 -3.86790216e-01 2.66586781e-01
-3.99546087e-01 -1.54216483e-01 7.13144302e-01 7.83795774e-01
1.08268118e+00 -8.74067992e-02 -1.01958185e-01 5.49278140e-01
2.25017473e-01 -7.39040524e-02 6.31181836e-01 -3.49657536e-01
5.16701818e-01 1.13107932e+00 -5.39993525e-01 -1.61043906e+00
-3.66247684e-01 -4.63573366e-01 -1.26706994e+00 -5.74554913e-02
4.40714419e-01 -2.42169220e-02 -7.35517919e-01 1.94216275e+00
2.29566664e-01 4.47860152e-01 -4.13880274e-02 4.33993608e-01
1.00416589e+00 4.78220969e-01 4.92105111e-02 -2.62092650e-01
5.84798872e-01 -8.91949296e-01 -3.24217021e-01 -4.24529761e-01
1.08817267e+00 -1.57183170e-01 8.26578438e-01 -5.27848341e-02
-8.05047452e-01 -3.03372920e-01 -1.03423560e+00 4.21444848e-02
-4.07570839e-01 -2.91636556e-01 3.27453852e-01 1.78470060e-01
-1.07136548e+00 1.25395906e+00 -6.88340664e-01 -3.22078139e-01
4.41889524e-01 1.46961465e-01 -6.11953020e-01 -1.88429892e-01
-1.08745551e+00 6.79646730e-01 5.67264140e-01 -1.02050997e-01
-6.81078732e-01 -7.56949365e-01 -1.11948836e+00 2.32009336e-01
3.75416875e-01 -4.71960127e-01 5.07716954e-01 -1.11459351e+00
-1.04562402e+00 9.10099506e-01 5.03780209e-02 -5.14822185e-01
3.34543288e-01 4.45119977e-01 -3.75204802e-01 9.61343944e-03
-3.76324952e-02 3.46075475e-01 7.76244044e-01 -1.06807017e+00
-1.14475124e-01 -5.26234686e-01 -1.45806223e-01 1.49938315e-01
-5.31559467e-01 -6.37990415e-01 -2.36844733e-01 -6.57891691e-01
3.53329122e-01 -8.14505160e-01 -2.81726301e-01 6.56064376e-02
-7.05866516e-01 -1.81186736e-01 7.49038696e-01 -4.43958104e-01
1.06178212e+00 -2.12880421e+00 4.73583043e-01 4.23519582e-01
7.90406287e-01 2.70412445e-01 -5.59195876e-01 6.23107910e-01
-4.50598985e-01 1.29698619e-01 -3.63239944e-01 -1.17505866e-03
-3.07239115e-01 2.38697752e-01 1.97933502e-02 5.69066525e-01
3.04979622e-01 1.08197188e+00 -1.34796989e+00 -2.29699805e-01
-1.48756042e-01 2.89909929e-01 -3.00920367e-01 2.91333675e-01
-3.60672832e-01 5.15680134e-01 -4.14325953e-01 2.56997198e-01
4.78816509e-01 -5.43519020e-01 6.77578807e-01 -3.60191166e-01
4.99966085e-01 1.88049749e-01 -8.30853403e-01 1.37230515e+00
-1.15190245e-01 6.16541803e-01 -2.73959637e-01 -1.74026370e+00
1.22127628e+00 -3.10877878e-02 5.85585415e-01 -7.36576557e-01
-1.30497575e-01 8.06862488e-02 8.44896957e-02 -2.88856000e-01
-2.77335327e-02 5.76789603e-02 1.54707387e-01 4.08556461e-01
8.71633515e-02 2.07832858e-01 1.62535608e-01 4.80372906e-01
1.48468876e+00 -2.29735136e-01 5.02609849e-01 -8.61180201e-02
5.11190593e-01 -3.63522679e-01 7.22581208e-01 4.82847631e-01
-1.89697504e-01 4.10810530e-01 1.03964710e+00 -4.82978135e-01
-6.97157860e-01 -1.01171410e+00 4.02358323e-01 1.00019693e+00
3.25135320e-01 -6.37306869e-01 -4.86819059e-01 -1.03196275e+00
2.00222850e-01 2.47504145e-01 -5.56130707e-01 -9.52066958e-01
-3.05297375e-01 -5.89635491e-01 4.03765649e-01 2.76396245e-01
2.99120724e-01 -7.43666410e-01 3.86145800e-01 2.71264855e-02
-9.12528634e-02 -8.78736198e-01 -7.04689980e-01 1.67877257e-01
-1.07364118e+00 -1.70258248e+00 -3.20098758e-01 -1.17221892e+00
1.06848037e+00 3.22949409e-01 1.14090383e+00 3.36306423e-01
-1.12963110e-01 8.33148360e-02 -3.75763088e-01 3.44189219e-02
-7.08486140e-01 1.02902211e-01 -1.07908346e-01 1.56843185e-01
8.13027471e-02 -9.61067617e-01 -3.78273100e-01 3.17549348e-01
-9.29061174e-01 -1.52290255e-01 5.59417367e-01 8.30648780e-01
8.71422291e-01 1.28196836e-01 6.79366648e-01 -1.29011226e+00
9.62934554e-01 -6.96767271e-01 -2.33373731e-01 5.05307198e-01
-9.30330515e-01 4.11333531e-01 1.07052636e+00 -2.85584003e-01
-3.02443177e-01 -6.81900978e-02 1.36719272e-01 -3.94368380e-01
-4.43145726e-03 8.77835810e-01 -3.50826442e-01 -2.67121583e-01
8.73323262e-01 4.31950659e-01 2.02257425e-01 -3.29150677e-01
4.69376892e-01 9.57556292e-02 4.37406272e-01 -2.41121858e-01
8.20116341e-01 1.53891549e-01 3.10548604e-01 -5.49915016e-01
-8.01048994e-01 -3.88841063e-01 -7.96378970e-01 -5.01921289e-02
4.09706593e-01 -6.20336473e-01 -4.98776585e-01 4.69790161e-01
-8.09334517e-01 -3.97047430e-01 -2.74630159e-01 1.28989786e-01
-4.41379696e-01 8.41485798e-01 -5.30823112e-01 -1.92912355e-01
-3.44655424e-01 -8.58254611e-01 6.51826859e-01 -3.78438458e-02
-7.21461102e-02 -1.41941535e+00 3.47822726e-01 4.75374684e-02
3.70672680e-02 7.74152160e-01 1.56629086e+00 -1.03521657e+00
-2.49876633e-01 -3.16392303e-01 -1.85949638e-01 3.32143724e-01
7.32877791e-01 1.40687600e-02 -4.23347622e-01 -6.81016624e-01
-6.40328288e-01 -4.14520621e-01 1.00129449e+00 8.58405530e-02
1.21750569e+00 -6.93078816e-01 -6.06378973e-01 6.28125846e-01
1.20364189e+00 -2.06789300e-01 4.82347161e-01 -7.70969763e-02
9.91978526e-01 4.91035461e-01 2.81892955e-01 -1.80106238e-02
2.71142095e-01 4.89888340e-01 6.47082806e-01 -3.40641737e-02
-8.98478031e-02 -7.23788738e-01 3.69662136e-01 9.52289581e-01
1.18252926e-01 -3.90055060e-01 -7.92289019e-01 3.59727502e-01
-2.05739522e+00 -7.91676939e-01 -1.34080246e-01 2.25396347e+00
7.86772728e-01 2.07266897e-01 1.53016061e-01 1.00780670e-02
9.90263641e-01 4.68335599e-01 -9.36513841e-01 -2.01593578e-01
-1.55169263e-01 -6.00520931e-02 2.39324301e-01 3.92592996e-01
-8.66044223e-01 8.28674316e-01 5.92933226e+00 6.47843361e-01
-1.13446295e+00 -3.36484671e-01 7.46052563e-01 2.38012135e-01
-4.99905050e-01 6.35985136e-02 -1.60924822e-01 5.11496246e-01
8.88987362e-01 -4.33265984e-01 3.78912419e-01 7.54081249e-01
1.30290970e-01 5.20847201e-01 -1.42420912e+00 8.52248907e-01
8.35272297e-02 -1.53362226e+00 3.59011441e-01 1.60409361e-02
7.95887828e-01 1.52641013e-01 -9.42977592e-02 2.88012624e-01
3.66638124e-01 -1.10431206e+00 2.29239967e-02 7.92621672e-01
6.35644138e-01 -7.48085201e-01 5.31163216e-01 4.84253407e-01
-1.25188518e+00 1.66894063e-01 -4.89581943e-01 4.41378392e-02
-3.67578059e-01 7.95475245e-01 -9.55804884e-01 6.65304542e-01
3.48764539e-01 1.35905111e+00 -7.59451568e-01 9.01475728e-01
-3.36629868e-01 5.50741434e-01 2.69852966e-01 5.21722902e-03
-9.96327400e-03 -5.69323838e-01 6.62026465e-01 7.30993092e-01
1.01095745e-02 -3.42039764e-01 3.74155521e-01 8.80060792e-01
-6.19418263e-01 2.59168565e-01 -8.73213708e-01 -4.43124264e-01
4.50299799e-01 1.03937066e+00 -4.75843698e-01 -7.68127441e-02
-2.28653923e-01 1.16513968e+00 1.01512718e+00 2.16676101e-01
-4.72985238e-01 -4.98246104e-01 8.22847903e-01 1.14215314e-01
-4.82000075e-02 -1.31552309e-01 1.63250044e-01 -1.03680480e+00
1.13748543e-01 -9.17445660e-01 6.45588875e-01 -5.51904678e-01
-1.68562567e+00 5.45403004e-01 -3.67027640e-01 -1.28605831e+00
-1.93052128e-01 -4.88887221e-01 -9.41427588e-01 3.77056211e-01
-1.54892302e+00 -8.83919477e-01 -4.77700561e-01 6.22839689e-01
8.36976469e-02 -2.81178743e-01 8.44632328e-01 7.58690387e-02
-7.05767989e-01 8.00501883e-01 5.49365878e-01 3.54548544e-01
7.36740112e-01 -1.28120863e+00 6.06923401e-01 5.23943126e-01
4.02838260e-01 3.85931551e-01 5.48343420e-01 -7.69717216e-01
-1.38258493e+00 -1.68256927e+00 7.32961833e-01 -5.09307943e-02
8.12988997e-01 -2.26329625e-01 -1.33027315e+00 7.09525466e-01
-3.82056624e-01 5.83757102e-01 6.36707604e-01 3.53139103e-03
-6.98865056e-01 -1.93055958e-01 -1.00236523e+00 6.30492687e-01
1.41967547e+00 -6.66067839e-01 -1.91218853e-01 7.08287716e-01
8.52794230e-01 -9.73406136e-02 -1.05609870e+00 3.42952579e-01
4.05845270e-02 -9.86860335e-01 8.08591664e-01 -1.14433753e+00
5.06643116e-01 -9.58913043e-02 2.22382575e-01 -1.92616010e+00
-4.31976050e-01 -6.63234890e-01 -3.78262550e-01 1.07352114e+00
5.58338761e-01 -9.25354064e-01 1.03797638e+00 2.23401815e-01
-4.17838357e-02 -9.05467033e-01 -7.44086444e-01 -9.54300642e-01
1.36851087e-01 6.25820830e-02 6.21697485e-01 1.35291326e+00
3.21336061e-01 5.71327329e-01 -2.85888135e-01 1.60805762e-01
5.84117532e-01 1.78106621e-01 5.79251707e-01 -1.51883602e+00
-2.83494234e-01 -5.33185303e-01 -9.91845548e-01 -8.22865129e-01
6.25495255e-01 -1.55153513e+00 -9.10433978e-02 -1.50776196e+00
1.77317694e-01 -2.99136192e-01 -3.70637357e-01 5.75973928e-01
-4.11267877e-01 -3.86024676e-02 -2.64085144e-01 2.22464412e-01
-6.81504011e-01 7.64909923e-01 1.17282414e+00 -5.95048428e-01
-2.75441915e-01 2.31881029e-04 -8.55519772e-01 5.53965032e-01
9.21994925e-01 -4.23214376e-01 -6.08686447e-01 -2.25106820e-01
3.08002174e-01 -1.54198363e-01 3.11868727e-01 -8.13101470e-01
3.22619170e-01 -2.31671691e-01 1.46343231e-01 -1.15883620e-02
-1.70261234e-01 -5.43988049e-01 4.52364720e-02 7.09466338e-01
-6.77532077e-01 -5.91371357e-02 -1.89482138e-01 1.25014114e+00
-2.61134118e-01 -3.26106325e-02 7.95488775e-01 -5.96231781e-02
-6.26228511e-01 8.80398333e-01 -1.35139395e-02 4.18325841e-01
9.80261147e-01 -1.60180494e-01 -4.35067087e-01 -5.50403595e-01
-7.69135058e-01 1.84187829e-01 5.18975317e-01 3.83347213e-01
8.18004489e-01 -1.39956784e+00 -7.39255607e-01 9.81960148e-02
5.50051510e-01 -9.65470821e-02 7.24426210e-02 6.91845000e-01
-2.78624594e-01 4.51852866e-02 -1.76142573e-01 -3.84776860e-01
-1.28666806e+00 6.53845906e-01 5.57376385e-01 -5.02441406e-01
-5.89432120e-01 6.72767043e-01 5.23139723e-02 -6.86178029e-01
6.77625611e-02 -1.69524178e-01 -3.04362625e-01 -1.55995846e-01
2.23046675e-01 3.73237193e-01 2.37773061e-01 -5.97366095e-01
-1.50039718e-01 3.89309555e-01 -2.12041676e-01 8.66772413e-01
1.34891582e+00 1.79305360e-01 -3.80938828e-01 2.30263740e-01
1.62862837e+00 -1.92530349e-01 -1.18347633e+00 -6.63624823e-01
2.74039239e-01 -3.82223308e-01 -1.15919806e-01 -3.46664697e-01
-1.22751558e+00 6.11794293e-01 2.57193953e-01 4.23761308e-01
9.55600619e-01 1.08195618e-01 6.45880759e-01 6.79192603e-01
1.29741237e-01 -7.74063349e-01 4.70501423e-01 4.83972073e-01
8.27701330e-01 -1.27237236e+00 6.29531890e-02 -6.71077549e-01
-3.99798542e-01 1.13703549e+00 6.82740092e-01 -2.05731601e-01
6.17448032e-01 -2.21105203e-01 -3.76510739e-01 -4.42568898e-01
-7.19757080e-01 -1.67097852e-01 6.24659717e-01 1.05290699e+00
4.57508266e-01 -1.32324221e-02 -8.92709419e-02 1.91155121e-01
2.09850371e-02 -4.64354813e-01 3.49381268e-01 6.25734508e-01
-4.04218316e-01 -1.16367662e+00 2.39219800e-01 7.98777699e-01
1.52501136e-01 -4.93070446e-02 -1.07486725e+00 4.56907630e-01
-2.20310807e-01 7.79440463e-01 -7.52228200e-02 -7.45882988e-01
3.01986754e-01 -1.34544417e-01 3.06195170e-01 -8.72058749e-01
-2.24056274e-01 -6.01268709e-01 -1.47880185e-02 -5.38157105e-01
-2.02937648e-01 -3.00007105e-01 -1.23000622e+00 -3.61212403e-01
-1.64744571e-01 2.17086822e-01 6.74244240e-02 1.01993203e+00
6.68524027e-01 5.08922637e-01 9.67845440e-01 -4.29418057e-01
-5.85496128e-01 -7.14263260e-01 -7.96836257e-01 9.18674171e-01
2.85584629e-01 -3.46843302e-01 -4.44670141e-01 -1.79441303e-01] | [7.210498332977295, 6.291930198669434] |
8d754e4d-30e8-4ba6-8184-19b4bb9c9cb0 | detecting-denial-of-service-attacks-from | null | null | https://aclanthology.org/N18-1147 | https://aclanthology.org/N18-1147.pdf | Detecting Denial-of-Service Attacks from Social Media Text: Applying NLP to Computer Security | This paper describes a novel application of NLP models to detect denial of service attacks using only social media as evidence. Individual networks are often slow in reporting attacks, so a detection system from public data could better assist a response to a broad attack across multiple services. We explore NLP methods to use social media as an indirect measure of network service status. We describe two learning frameworks for this task: a feed-forward neural network and a partially labeled LDA model. Both models outperform previous work by significant margins (20{\%} F1 score). We further show that the topic-based model enables the first fine-grained analysis of how the public reacts to ongoing network attacks, discovering multiple {``}stages{''} of observation. This is the first model that both detects network attacks (with best performance) and provides an analysis of when and how the public interprets service outages. We describe the models, present experiments on the largest twitter DDoS corpus to date, and conclude with an analysis of public reactions based on the learned model{'}s output. | ['James McMasters', 'Ben Fry', 'Nathanael Chambers'] | 2018-06-01 | null | null | null | naacl-2018-6 | ['computer-security'] | ['miscellaneous'] | [-2.21879750e-01 9.08362046e-02 -5.08692443e-01 -5.14202178e-01
-9.35770035e-01 -8.07658076e-01 8.09672713e-01 4.69285131e-01
-1.65887758e-01 6.04600310e-01 2.22962320e-01 -8.69649351e-01
-2.96864063e-01 -8.71985793e-01 -2.52315104e-01 -5.38365722e-01
-5.61372280e-01 1.12064481e+00 4.07840163e-01 -7.93870613e-02
3.95291775e-01 1.10908663e+00 -5.00092864e-01 5.48182607e-01
-8.53214115e-02 9.51412380e-01 -7.91904211e-01 6.29133105e-01
-2.33980522e-01 1.26342750e+00 -1.34005570e+00 -3.13831300e-01
-7.73368850e-02 7.29748234e-02 -8.30505669e-01 7.74004161e-02
-2.76081443e-01 -5.72422147e-01 -5.77901125e-01 5.40989339e-01
5.32333672e-01 -3.37798595e-01 4.15640205e-01 -1.72429442e+00
-1.03172235e-01 8.67128134e-01 -4.53321427e-01 1.10233033e+00
5.35050094e-01 1.11479037e-01 1.18920791e+00 -2.74543315e-01
4.66927975e-01 1.32457507e+00 7.35624373e-01 1.16689198e-01
-1.34508288e+00 -9.67503011e-01 3.66003662e-01 1.99586064e-01
-1.16580319e+00 -4.26205605e-01 6.72233105e-01 -5.05297422e-01
1.38298810e+00 1.98696703e-01 2.27075875e-01 1.46603060e+00
-3.20507661e-02 5.65207481e-01 1.06827462e+00 -1.24858461e-01
1.78106666e-01 4.86423820e-01 2.67903358e-01 4.48869109e-01
-5.38892709e-02 -1.81604579e-01 -3.21259379e-01 -1.13789964e+00
3.23569149e-01 1.28414780e-01 -2.10286099e-02 4.34415162e-01
-6.02428436e-01 1.10410321e+00 1.90171048e-01 3.01866829e-01
-6.59838080e-01 -2.38751806e-02 4.30913717e-01 5.90910912e-01
1.08386922e+00 3.12554628e-01 -6.92156971e-01 -2.85954475e-01
-9.81516480e-01 1.76425159e-01 1.77045536e+00 1.32893234e-01
6.39783978e-01 1.18612692e-01 1.48718581e-01 6.05429828e-01
2.98793197e-01 4.43384469e-01 -1.36554435e-01 -7.96508670e-01
2.62088031e-01 3.40741575e-01 1.94369912e-01 -9.92262900e-01
-8.13954771e-01 -6.35913849e-01 1.33480942e-02 -1.85696393e-01
3.59426707e-01 -6.55215979e-01 -2.08949029e-01 1.56967318e+00
9.35507119e-02 6.70521855e-01 -2.21952647e-01 3.67501825e-01
4.04324383e-01 7.93356359e-01 2.13071242e-01 -6.09065890e-01
1.21507382e+00 -6.50697887e-01 -6.19144082e-01 -1.98550716e-01
7.37754583e-01 -6.24164820e-01 2.81592727e-01 4.70714301e-01
-8.20794642e-01 5.16335547e-01 -4.63751942e-01 8.02787960e-01
-3.56524229e-01 -5.30656934e-01 7.32396305e-01 9.72836912e-01
-9.09388363e-01 4.65472758e-01 -7.76447833e-01 -7.85732746e-01
5.40371418e-01 2.79111296e-01 2.48163044e-01 1.57452404e-01
-1.34783530e+00 7.95895815e-01 -1.99129507e-01 -3.78537625e-01
-1.18886685e+00 -4.45404083e-01 -7.44200638e-03 3.71298075e-01
3.93890321e-01 2.31460650e-02 1.31651402e+00 -4.90865380e-01
-1.18785906e+00 7.85559475e-01 -2.27453172e-01 -8.02204967e-01
1.59462631e-01 5.22408076e-02 -1.00814068e+00 4.70237672e-01
1.30941689e-01 -2.42194086e-01 5.75531602e-01 -1.04336977e+00
-5.62632442e-01 -1.07651278e-01 2.48476744e-01 -5.34834385e-01
-4.08873290e-01 9.26597536e-01 2.18952760e-01 -1.48915827e-01
5.14400266e-02 -7.54822254e-01 -8.53979215e-02 -5.54029882e-01
-6.00596488e-01 -6.89208686e-01 1.17999959e+00 -4.44703102e-01
1.42463624e+00 -1.51746356e+00 -6.93091094e-01 6.22532427e-01
4.26373094e-01 2.34736562e-01 1.09030977e-01 1.25227261e+00
-2.67231941e-01 5.66378236e-01 3.25766265e-01 -4.49289501e-01
1.77853689e-01 3.08565706e-01 -9.03871953e-01 6.29734755e-01
2.22133100e-01 2.31600747e-01 -7.25098073e-01 -1.39789894e-01
-3.02999645e-01 2.89512783e-01 -5.85787416e-01 3.57193589e-01
-2.25639492e-01 3.98682684e-01 -6.55193746e-01 9.42778707e-01
1.84310153e-01 -6.98611200e-01 2.69394577e-01 1.31693125e-01
-4.84302491e-02 8.08823526e-01 -6.10860705e-01 5.88810325e-01
-2.13553488e-01 7.45805442e-01 1.24165356e-01 -1.26962018e+00
9.39079762e-01 7.91978776e-01 9.47743535e-01 -2.55335838e-01
2.54052222e-01 5.99626377e-02 -6.24004379e-02 -6.77164197e-01
-5.12486875e-01 -2.05792889e-01 -2.05028504e-01 1.39805031e+00
-6.94318190e-02 1.67394921e-01 1.49863837e-02 6.26794875e-01
1.86615455e+00 -7.76620984e-01 1.55945003e-01 1.48179820e-02
2.98031420e-01 -4.49151881e-02 6.32114053e-01 1.21759975e+00
-5.95425844e-01 -9.13610160e-02 1.29841840e+00 -7.88121760e-01
-7.90338874e-01 -1.06367743e+00 -1.02477923e-01 1.42487907e+00
-4.25133616e-01 -4.98278737e-01 -5.09944022e-01 -9.09440517e-01
-9.60218906e-02 8.79757822e-01 -1.88531399e-01 2.72884935e-01
-5.65032184e-01 -1.52494133e+00 1.03333461e+00 1.17997721e-01
9.54727754e-02 -1.06395471e+00 2.71216482e-01 4.70265329e-01
-5.56082428e-01 -1.67260909e+00 3.02095413e-01 1.04605846e-01
-4.87056315e-01 -1.16961789e+00 2.50813961e-01 -2.76127994e-01
2.14822069e-01 -9.18492377e-02 1.14620304e+00 7.62916505e-02
-1.11327149e-01 6.45076931e-01 -1.24940485e-01 -5.50476015e-01
-6.39532983e-01 6.16106903e-03 5.90952456e-01 2.67284334e-01
1.05026424e+00 -1.27589488e+00 -3.10055345e-01 4.89153475e-01
-5.95253944e-01 -9.13109541e-01 3.29535812e-01 1.12151496e-01
-3.87847632e-01 1.70399517e-01 8.83005142e-01 -1.19567716e+00
1.10054660e+00 -1.31615305e+00 -3.38287860e-01 -2.06383660e-01
-7.16084063e-01 -5.10036826e-01 2.29342222e-01 -3.68026316e-01
-6.98330820e-01 -7.01704741e-01 -3.59251857e-01 -3.83720398e-02
-4.54849452e-01 4.85170454e-01 3.61758798e-01 -6.84859650e-03
1.08771527e+00 -9.25042257e-02 -2.06670150e-01 -4.88041610e-01
-7.01613203e-02 9.51030612e-01 -4.45517711e-03 -6.60638392e-01
9.06707346e-01 8.99007261e-01 -3.34961146e-01 -9.66256142e-01
-1.12755108e+00 -7.42148995e-01 -1.95380837e-01 -2.73705125e-01
2.37930432e-01 -6.60983562e-01 -1.18446803e+00 2.31636912e-01
-1.37948060e+00 -6.12410717e-02 1.02013841e-01 5.03415108e-01
-2.65522480e-01 3.32677990e-01 -1.45399141e+00 -1.12839615e+00
-2.63051480e-01 -6.00751638e-01 5.12911916e-01 -2.42099725e-02
-3.75278801e-01 -1.36827254e+00 3.19935113e-01 4.79871184e-01
7.48346567e-01 7.76797682e-02 8.82421076e-01 -1.83828771e+00
-4.14958388e-01 -8.09106886e-01 -6.32644355e-01 1.33178681e-01
-1.89306125e-01 -8.04129057e-03 -1.18265498e+00 -4.54605311e-01
3.15154344e-01 -3.62071902e-01 4.04676467e-01 1.91981524e-01
8.83472145e-01 -7.38605440e-01 -5.71764708e-01 7.32015520e-02
1.03424037e+00 2.14154813e-02 4.43170756e-01 4.81716096e-01
2.14125276e-01 7.03505874e-01 -2.09825933e-01 8.59690547e-01
4.50820178e-01 3.34266633e-01 6.09293640e-01 -7.78963491e-02
4.03382182e-01 -1.31153911e-01 6.51574373e-01 5.11959314e-01
1.60683006e-01 -7.77038038e-01 -1.18066943e+00 3.53138983e-01
-1.52266228e+00 -1.14977121e+00 -5.86141311e-02 1.69788456e+00
3.67852479e-01 7.04417109e-01 4.53839064e-01 1.70171678e-01
8.19285095e-01 5.64015925e-01 -1.59788311e-01 -6.36135578e-01
-2.52897944e-02 1.22968301e-01 4.70692903e-01 7.68797040e-01
-1.06631863e+00 8.10871482e-01 7.59243059e+00 5.57262659e-01
-1.00885630e+00 3.31842542e-01 4.80651408e-01 -2.03883588e-01
1.02714419e-01 4.87645835e-01 -1.04961610e+00 2.89531618e-01
1.87966251e+00 5.21070836e-03 2.45030656e-01 6.52620733e-01
7.06069350e-01 2.55798697e-01 -8.22515786e-01 3.62288415e-01
2.50185460e-01 -1.23351252e+00 -1.26446873e-01 3.91288102e-01
3.23704571e-01 7.06329048e-01 -3.29472482e-01 3.69629085e-01
7.02126741e-01 -6.30029857e-01 -5.20537980e-02 4.32820618e-01
2.97597609e-02 -5.32596946e-01 6.67253435e-01 6.27610683e-01
-4.26620126e-01 -5.81151605e-01 8.49114433e-02 -3.00409168e-01
7.52962708e-01 9.32610273e-01 -1.37299383e+00 -1.15622602e-01
4.78100449e-01 6.29557192e-01 -1.71686649e-01 8.81431341e-01
-5.09270489e-01 1.45149922e+00 -6.17537260e-01 9.98494402e-02
5.20668805e-01 3.47061336e-01 1.20601904e+00 1.47147810e+00
-1.49693489e-01 2.10688010e-01 8.13968837e-01 6.42741323e-01
-1.40227135e-02 -1.60686955e-01 -3.21981460e-01 -4.10505533e-01
6.03624046e-01 1.31783128e+00 -6.64861500e-01 -3.63757849e-01
-3.25124532e-01 3.10571522e-01 1.46047115e-01 6.22568905e-01
-6.71247900e-01 -1.76543891e-01 6.71290874e-01 7.04100609e-01
-1.99444562e-01 -2.05038235e-01 1.61231115e-01 -1.10438669e+00
-4.96307164e-01 -6.98226988e-01 6.82707727e-01 -5.53537726e-01
-1.75204480e+00 7.44451344e-01 -1.02470309e-01 -8.13994765e-01
-4.28588808e-01 -4.96690065e-01 -1.03502679e+00 6.51715338e-01
-1.61158299e+00 -9.98241365e-01 2.87859738e-01 3.85772645e-01
3.42894018e-01 -3.04502040e-01 1.00084257e+00 6.47121429e-01
-8.74233305e-01 2.09366307e-01 -3.19292903e-01 5.02472758e-01
7.14336157e-01 -9.72427189e-01 3.30448061e-01 6.93681121e-01
1.61458135e-01 3.53695065e-01 9.65016007e-01 -6.15869284e-01
-7.23421037e-01 -6.88575923e-01 1.38496125e+00 -7.92769134e-01
1.43637359e+00 -3.62653226e-01 -7.88042307e-01 1.22774589e+00
-5.05652614e-02 6.19828282e-03 1.15092826e+00 5.51794946e-01
-6.04219317e-01 -1.45012662e-01 -1.27129638e+00 1.75766617e-01
5.17687023e-01 -9.00803030e-01 -2.83682942e-01 1.28013504e+00
6.36327386e-01 2.17744291e-01 -8.51838291e-01 2.07691565e-02
3.22540522e-01 -9.96195912e-01 8.61925244e-01 -1.21294665e+00
-1.84308082e-01 2.02012941e-01 -1.41390219e-01 -9.85952437e-01
-2.77217776e-01 -9.31857586e-01 -3.16020846e-01 1.16145945e+00
6.32445931e-01 -1.13684249e+00 8.53838623e-01 4.28745091e-01
4.20039505e-01 -5.06103158e-01 -7.10497618e-01 -5.16900480e-01
-1.01922847e-01 -8.25662494e-01 2.93308586e-01 1.26279473e+00
3.04470360e-01 2.29011863e-01 -3.48784655e-01 8.23711812e-01
5.13645351e-01 -1.41654208e-01 5.56644619e-01 -1.59518898e+00
-2.59663254e-01 -3.30155313e-01 -5.53689897e-01 -9.29980218e-01
3.51733357e-01 -5.09227693e-01 -6.84788227e-01 -1.00024652e+00
8.80731121e-02 -5.78698277e-01 -5.30954599e-01 5.29424965e-01
7.44554222e-01 4.41910148e-01 -3.74430597e-01 5.78423142e-01
-6.47171497e-01 -2.97180921e-01 5.70604540e-02 3.23739499e-02
-1.23113342e-01 5.91621637e-01 -6.68015361e-01 8.91067982e-01
1.00689149e+00 -9.95125949e-01 4.54546995e-02 -2.05669910e-01
4.66240078e-01 3.55914503e-01 5.49832225e-01 -5.61906755e-01
4.40540880e-01 -3.55512425e-02 3.40244137e-02 -4.25350547e-01
4.30544734e-01 -3.45683306e-01 -4.33822364e-01 4.19776827e-01
-3.36605668e-01 1.16914533e-01 -7.71599710e-02 8.46932173e-01
2.09625345e-02 8.98766816e-02 5.93852162e-01 -1.72074154e-01
-6.33851066e-02 4.21606421e-01 -9.59713757e-01 -3.21065681e-03
6.78179979e-01 4.23445970e-01 -7.67365456e-01 -1.11547816e+00
-1.33683276e+00 2.23868951e-01 -2.60261059e-01 3.01062733e-01
1.08775347e-01 -7.19775140e-01 -8.95925820e-01 -1.12165570e-01
-3.11633736e-01 -1.18132317e+00 1.89228300e-02 1.07621586e+00
-3.29697937e-01 5.74511290e-01 1.49773404e-01 -5.09561658e-01
-9.64047551e-01 3.57720494e-01 4.41950798e-01 -5.81984937e-01
-4.13730472e-01 7.00669825e-01 -5.17369151e-01 -3.89561057e-01
3.81153405e-01 4.68107402e-01 -4.15108442e-01 3.82965147e-01
8.29093337e-01 4.79869246e-01 1.24451861e-01 -4.44816828e-01
-5.57551265e-01 -2.86009938e-01 -3.91529411e-01 -2.27132544e-01
1.46533346e+00 -2.78653949e-01 -4.66526359e-01 5.67960382e-01
1.23960018e+00 2.19811276e-01 -6.95793986e-01 -8.99677515e-01
2.85893857e-01 -3.90444219e-01 5.34870513e-02 -8.97202253e-01
-8.32177758e-01 8.10339272e-01 1.75600171e-01 1.22983849e+00
6.08956039e-01 1.99506149e-01 7.81972349e-01 5.62604725e-01
1.24192201e-01 -7.58942127e-01 2.09651262e-01 3.78682822e-01
2.90729046e-01 -9.43733096e-01 8.96309838e-02 -3.21725339e-01
-2.31378287e-01 1.18927586e+00 2.13802427e-01 -3.26610029e-01
1.13209105e+00 3.56761247e-01 1.57528549e-01 -7.70301044e-01
-1.13680911e+00 -2.84324661e-02 -5.07718325e-01 3.10638577e-01
1.57291844e-01 6.39918726e-03 -1.10146061e-01 5.42748988e-01
3.51657830e-02 -4.88555849e-01 8.25958729e-01 8.81613314e-01
-8.20967734e-01 -1.14598763e+00 -3.00112426e-01 5.22947133e-01
-1.06730258e+00 2.39791479e-02 -6.75263286e-01 5.07210612e-01
-3.25805724e-01 1.46518910e+00 1.18346997e-01 -3.81350547e-01
2.32143968e-01 4.27847981e-01 -4.21865910e-01 -6.82844877e-01
-5.73556185e-01 9.02691409e-02 4.99849200e-01 -6.95710003e-01
-2.21261904e-01 -6.96702778e-01 -7.39634752e-01 -7.76239336e-01
-3.36577088e-01 6.19530857e-01 8.07554781e-01 1.18390441e+00
1.61499575e-01 2.19282404e-01 1.39457285e+00 -4.86944944e-01
-4.79488462e-01 -9.93396580e-01 -7.18844593e-01 1.44328952e-01
2.75490433e-01 -4.04251188e-01 -1.18273628e+00 -5.23374140e-01] | [8.106654167175293, 9.557048797607422] |
ffcad9e7-282e-4837-904f-51c1a94d22d9 | domain-adversarial-training-of-self-attention | 2104.00564 | null | https://arxiv.org/abs/2104.00564v2 | https://arxiv.org/pdf/2104.00564v2.pdf | Domain-Adversarial Training of Self-Attention Based Networks for Land Cover Classification using Multi-temporal Sentinel-2 Satellite Imagery | The increasing availability of large-scale remote sensing labeled data has prompted researchers to develop increasingly precise and accurate data-driven models for land cover and crop classification (LC&CC). Moreover, with the introduction of self-attention and introspection mechanisms, deep learning approaches have shown promising results in processing long temporal sequences in the multi-spectral domain with a contained computational request. Nevertheless, most practical applications cannot rely on labeled data, and in the field, surveys are a time consuming solution that poses strict limitations to the number of collected samples. Moreover, atmospheric conditions and specific geographical region characteristics constitute a relevant domain gap that does not allow direct applicability of a trained model on the available dataset to the area of interest. In this paper, we investigate adversarial training of deep neural networks to bridge the domain discrepancy between distinct geographical zones. In particular, we perform a thorough analysis of domain adaptation applied to challenging multi-spectral, multi-temporal data, accurately highlighting the advantages of adapting state-of-the-art self-attention based models for LC&CC to different target zones where labeled data are not available. Extensive experimentation demonstrated significant performance and generalization gain in applying domain-adversarial training to source and target regions with marked dissimilarities between the distribution of extracted features. | ['Mauro Martini', 'Marcello Chiaberge', 'Aleem Khaliq', 'Vittorio Mazzia'] | 2021-04-01 | null | null | null | null | ['crop-classification'] | ['miscellaneous'] | [ 4.65116888e-01 -2.44830027e-01 -2.66476661e-01 -3.71497303e-01
-7.36876070e-01 -8.09717000e-01 5.85496724e-01 2.91648865e-01
-5.25497556e-01 1.01388586e+00 -1.75840314e-02 -4.98920918e-01
-2.50782192e-01 -1.05547225e+00 -6.50123775e-01 -8.27461541e-01
-1.21648476e-01 2.89107233e-01 -3.16668302e-02 -6.44785762e-01
-1.86557114e-01 7.19668269e-01 -1.45313871e+00 -4.27308567e-02
1.23076892e+00 8.73581648e-01 4.13684875e-01 4.98815149e-01
-1.13330767e-01 2.50352979e-01 -6.08693361e-01 -5.96217923e-02
3.13806951e-01 -2.69244254e-01 -6.57442331e-01 7.97457621e-02
2.28922740e-01 -3.75882000e-01 -1.52477205e-01 1.14828122e+00
6.27137780e-01 2.39907935e-01 7.90682077e-01 -9.98563170e-01
-7.96607494e-01 3.60083282e-01 -4.42512244e-01 3.99707526e-01
-2.56300330e-01 1.18750846e-03 6.30105913e-01 -4.92201835e-01
3.98226976e-01 6.53283298e-01 7.07054734e-01 2.94886976e-01
-1.24606955e+00 -5.41596174e-01 5.44955254e-01 1.51820585e-01
-1.45141375e+00 -1.99237764e-01 8.40872347e-01 -6.08877420e-01
7.96923399e-01 1.63223505e-01 4.03330117e-01 1.23780537e+00
-6.43444657e-02 5.36173761e-01 9.76946831e-01 -2.30974346e-01
4.18897539e-01 2.69255102e-01 -2.14712694e-01 -1.07698753e-01
1.24616519e-01 4.09771442e-01 5.24385981e-02 -1.28410840e-02
4.98365819e-01 6.78413063e-02 -2.31409475e-01 -3.84350985e-01
-9.58422244e-01 9.24427807e-01 8.74416292e-01 5.56930900e-01
-7.00615346e-01 -3.74049902e-01 5.64940512e-01 2.33427331e-01
1.11289346e+00 3.83005798e-01 -6.85248911e-01 3.09110761e-01
-1.14342904e+00 1.08838171e-01 2.65932411e-01 8.66572559e-01
5.75087428e-01 4.67754751e-01 2.98426539e-01 7.71946549e-01
-1.97850928e-01 9.92081404e-01 4.95499849e-01 -3.60621154e-01
3.67855936e-01 4.87623394e-01 2.36083627e-01 -1.05178440e+00
-5.00224113e-01 -7.59979010e-01 -1.28709328e+00 1.75151780e-01
3.45052987e-01 -2.83606410e-01 -9.78122890e-01 1.76937163e+00
2.05954388e-01 7.62249983e-04 4.98131126e-01 9.16842997e-01
4.78884488e-01 7.30174959e-01 5.55103362e-01 1.01959363e-01
1.03877878e+00 -4.10177708e-01 -4.45719689e-01 -6.74366713e-01
3.41319084e-01 -2.43864223e-01 9.99297321e-01 -7.36128837e-02
-3.61126602e-01 -7.49169946e-01 -1.04908168e+00 3.73765916e-01
-1.09520197e+00 1.60475329e-01 5.47446609e-01 3.62927616e-01
-7.40991652e-01 5.85446596e-01 -6.13664150e-01 -7.61121511e-01
5.15956759e-01 3.31575833e-02 -4.62215692e-01 -1.75225865e-02
-1.56386590e+00 9.73519385e-01 7.27330685e-01 4.45987433e-01
-8.96082282e-01 -8.96092355e-01 -8.89838934e-01 2.14857593e-01
1.68776140e-01 -1.17246378e-02 8.12273860e-01 -1.27986586e+00
-1.02546656e+00 9.67547655e-01 1.41255423e-01 -7.25805700e-01
4.12956923e-01 -2.68220007e-01 -8.49198103e-01 3.53659205e-02
3.37093920e-01 6.35836005e-01 8.16200793e-01 -1.12898219e+00
-7.10411251e-01 -4.47666317e-01 3.03915590e-02 2.55830348e-01
-3.93900573e-01 -1.40325412e-01 3.39421928e-01 -7.92322338e-01
4.03562700e-03 -7.81658828e-01 -3.34167391e-01 -7.80417398e-02
1.00686485e-02 1.62105381e-01 8.69079471e-01 -8.31191421e-01
8.41140270e-01 -2.43573499e+00 -2.18383577e-02 -7.17110783e-02
-2.95765847e-01 7.52448201e-01 -3.52040976e-01 4.66773123e-01
-2.28437647e-01 1.56561419e-01 -7.28042305e-01 5.01784623e-01
-2.20356986e-01 2.60995269e-01 -6.66115463e-01 4.92783427e-01
5.63733995e-01 7.66784310e-01 -1.00254023e+00 -1.46890879e-01
4.40022737e-01 3.90717417e-01 -1.52784094e-01 2.95654178e-01
-4.67287093e-01 5.30223191e-01 -5.17110825e-01 5.40020049e-01
8.53274763e-01 -8.97071436e-02 7.39619210e-02 4.78322804e-02
-1.55219510e-01 -1.01610094e-01 -7.21242487e-01 1.44484270e+00
-6.24475777e-01 7.75898993e-01 6.58949688e-02 -1.39367676e+00
1.04092264e+00 3.09778035e-01 3.92077357e-01 -7.53168344e-01
-1.20661594e-01 3.04279178e-01 -1.16159007e-01 -4.71982747e-01
6.28417611e-01 -3.12874615e-01 -2.02760532e-01 1.21609820e-02
-4.28705709e-03 -1.11094929e-01 -5.24574742e-02 -3.18301231e-01
4.99825954e-01 3.45906407e-01 4.20540363e-01 -3.94382864e-01
5.93484223e-01 5.42886019e-01 3.92830551e-01 5.49805760e-01
-4.42501217e-01 5.70772886e-01 9.97912437e-02 -5.62785327e-01
-1.14777160e+00 -7.92533338e-01 -3.70701909e-01 1.32321298e+00
-8.06789696e-02 6.06829822e-01 -4.51757848e-01 -7.06132531e-01
1.92006096e-01 8.52518022e-01 -8.42512071e-01 -2.95726836e-01
-2.33940661e-01 -1.15565670e+00 6.43734694e-01 6.47367835e-01
8.64265621e-01 -1.22921801e+00 -7.69782901e-01 4.68608111e-01
-2.53490537e-01 -1.06356013e+00 9.19092670e-02 4.95440215e-01
-8.29253018e-01 -9.02863622e-01 -1.09200835e+00 -5.52453220e-01
4.67030495e-01 3.52213770e-01 1.09366274e+00 -5.06136954e-01
-4.30736039e-03 1.76142268e-02 -4.63471532e-01 -6.70767546e-01
-4.79498446e-01 6.57596469e-01 -1.40240774e-01 5.27757891e-02
5.55755973e-01 -6.75271153e-01 -3.87323260e-01 3.03803414e-01
-1.02951527e+00 -2.56830543e-01 5.86271405e-01 8.43447328e-01
3.72244418e-01 1.63735792e-01 9.89880323e-01 -6.23515606e-01
2.09090352e-01 -8.67639840e-01 -6.95296466e-01 2.63618320e-01
-3.05066377e-01 -4.51549771e-04 8.62687349e-01 -3.99734676e-01
-1.22987449e+00 1.03751369e-01 -9.44882482e-02 -1.26417622e-01
-7.66139090e-01 9.72852767e-01 -2.20201999e-01 1.78448990e-01
1.08314455e+00 5.50521016e-01 -1.21979050e-01 -3.37429702e-01
2.60385424e-01 9.78749335e-01 7.16529369e-01 -1.76932842e-01
8.84295702e-01 5.85405111e-01 -1.88813999e-01 -1.11406791e+00
-1.05847025e+00 -5.17245173e-01 -8.75416994e-01 3.45391557e-02
7.44622529e-01 -1.17905235e+00 9.93513837e-02 6.77773774e-01
-7.68892050e-01 -5.83756924e-01 -3.27865899e-01 3.66764724e-01
-3.66665959e-01 1.26514420e-01 9.11148712e-02 -7.74278998e-01
-3.54730546e-01 -5.84763229e-01 8.92145097e-01 8.02733228e-02
-1.72831230e-02 -1.18421888e+00 4.44153473e-02 -5.59031740e-02
9.06048059e-01 6.15696013e-01 8.42378378e-01 -6.85429871e-01
-1.06554121e-01 -4.18772012e-01 -4.04180855e-01 5.00975788e-01
5.43169796e-01 -2.03129277e-01 -1.18437469e+00 -2.66416609e-01
-2.36263305e-01 -2.34487712e-01 7.46415973e-01 5.94123244e-01
9.02251542e-01 1.13194650e-02 -1.72031462e-01 6.90328896e-01
1.54479134e+00 2.00483188e-01 3.87119293e-01 4.72487926e-01
4.81127471e-01 7.31729507e-01 8.20114493e-01 3.45787585e-01
1.43678263e-01 4.36125100e-01 6.77894056e-01 -5.03963947e-01
1.82271793e-01 -9.47328061e-02 -1.87989175e-02 -4.00063731e-02
-1.99496727e-02 -3.89228910e-01 -1.02795994e+00 1.16360688e+00
-1.70149577e+00 -1.22077787e+00 1.58684865e-01 2.21013880e+00
7.07209527e-01 -9.52537730e-02 6.08111732e-03 1.31945699e-01
7.65122294e-01 5.67027569e-01 -9.94550705e-01 -3.90617773e-02
-5.49395919e-01 1.22973524e-01 1.04536963e+00 1.66614905e-01
-1.66577113e+00 1.19585443e+00 5.78242588e+00 6.70098126e-01
-1.50558901e+00 -7.57133812e-02 4.11212265e-01 2.64197111e-01
2.77505945e-02 -4.89993662e-01 -4.82329369e-01 3.52571577e-01
1.21853983e+00 -2.24523172e-01 2.73917496e-01 1.00401330e+00
2.35214561e-01 -3.18896994e-02 -7.18357265e-01 4.31456804e-01
-3.29241723e-01 -1.00906157e+00 -9.47348699e-02 -2.47886702e-02
7.58616567e-01 5.06476223e-01 1.84219569e-01 3.52192611e-01
4.16061997e-01 -8.81041884e-01 5.17777205e-01 2.20286459e-01
9.65846658e-01 -6.72865450e-01 8.73100579e-01 3.82962316e-01
-1.13639319e+00 -1.31695837e-01 -5.45738876e-01 6.54355362e-02
-1.74794868e-01 3.18962425e-01 -7.01652586e-01 8.82187247e-01
7.46412396e-01 9.21671927e-01 -4.32015538e-01 7.00614035e-01
-8.73121545e-02 7.00476408e-01 -4.05344605e-01 2.48394907e-01
5.31778157e-01 -1.03741191e-01 3.63970757e-01 1.01254010e+00
3.62889498e-01 1.82855800e-01 -3.28698643e-02 7.30658233e-01
2.13615447e-01 3.92540134e-02 -9.52761054e-01 -1.33125052e-01
3.85069042e-01 8.50283265e-01 -4.19778496e-01 -1.50590524e-01
-3.45942378e-01 9.55186844e-01 1.27585143e-01 5.97673774e-01
-6.20267630e-01 -3.21365714e-01 9.60825801e-01 1.75146051e-02
4.33621377e-01 -2.56857932e-01 -3.16583037e-01 -9.62879837e-01
-1.27696976e-01 -8.37365866e-01 3.56078327e-01 -6.75396323e-01
-1.26200879e+00 7.06028044e-01 1.66996166e-01 -1.42483950e+00
-3.47115159e-01 -3.75380874e-01 -3.81986201e-01 1.24325097e+00
-2.04225278e+00 -1.39997542e+00 -5.77165961e-01 5.27929306e-01
5.06907701e-01 -3.53395283e-01 1.21687222e+00 1.69693649e-01
-3.53294611e-01 3.05328965e-01 6.38704777e-01 2.21299738e-01
5.85298181e-01 -1.03824019e+00 6.19146764e-01 1.11679101e+00
-1.84904858e-01 1.23493746e-02 6.24252260e-01 -4.69754547e-01
-7.44493902e-01 -1.60563624e+00 8.01025033e-01 8.43253639e-03
6.42881751e-01 -1.09605245e-01 -1.31365633e+00 5.74818313e-01
-5.01721213e-03 1.82520702e-01 5.75434685e-01 -4.03622128e-02
-2.85659313e-01 -3.08496594e-01 -1.22659755e+00 3.98863018e-01
7.52488971e-01 -6.97338820e-01 -4.04952943e-01 2.98002362e-01
4.99491721e-01 -1.37049451e-01 -7.43611038e-01 4.61170226e-01
2.73548216e-01 -7.15683997e-01 8.84064078e-01 -7.34611571e-01
4.51267451e-01 -2.03525811e-01 -4.14564043e-01 -1.63777459e+00
-5.07697046e-01 1.32664829e-01 5.04466712e-01 1.09575403e+00
4.11536485e-01 -6.56532228e-01 7.18994141e-01 3.76393139e-01
-2.89311223e-02 -4.48274016e-02 -7.14212418e-01 -8.55695248e-01
5.42606533e-01 -3.33503962e-01 8.45056772e-01 1.14170301e+00
-4.24546003e-01 -1.56715870e-01 -3.25459033e-01 7.47694373e-01
3.95693392e-01 3.05103630e-01 4.88023281e-01 -1.41187716e+00
1.65508553e-01 -3.11031073e-01 -3.39012206e-01 -4.90646601e-01
3.83603722e-01 -7.34680414e-01 -5.27959615e-02 -1.29909706e+00
-2.97805697e-01 -5.68445206e-01 -4.14830446e-01 6.61451936e-01
-3.32253605e-01 1.64373770e-01 -2.49828622e-02 9.90874097e-02
1.58909753e-01 7.75357306e-01 7.30597734e-01 -5.43131053e-01
-3.15169364e-01 2.16028765e-01 -6.08027399e-01 4.16514695e-01
1.17474031e+00 -3.97434324e-01 -3.60544652e-01 -5.48189044e-01
-1.96917299e-02 -1.26035139e-01 5.08564889e-01 -1.15936279e+00
-2.80792236e-01 -3.95485759e-01 3.04487765e-01 -6.41672671e-01
-3.96777056e-02 -1.14387071e+00 1.08686164e-01 2.99874306e-01
-3.04529995e-01 -4.28027391e-01 7.07255304e-01 5.17402053e-01
-4.02304798e-01 -9.44242999e-02 8.78601432e-01 -7.55145624e-02
-1.26472366e+00 3.54955763e-01 -4.86182511e-01 6.65542260e-02
1.06971252e+00 -1.91631868e-01 -5.91549650e-02 -2.62308896e-01
-7.16033459e-01 2.24929452e-01 3.86435628e-01 6.09885633e-01
1.55371875e-01 -1.10668516e+00 -1.10894036e+00 4.95308876e-01
4.72605884e-01 1.81538820e-01 5.53772032e-01 1.81191042e-01
-4.32029188e-01 5.55603027e-01 -5.19483685e-01 -5.56704044e-01
-9.61474895e-01 7.82288909e-01 5.43979168e-01 -2.10444659e-01
-4.46733624e-01 5.16912222e-01 1.62406534e-01 -7.22724736e-01
-2.39251778e-01 -2.76005894e-01 -2.80645251e-01 4.25835103e-01
2.46001542e-01 -1.78258475e-02 2.20767900e-01 -7.69211650e-01
-3.28902900e-01 2.55422503e-01 2.06336081e-01 1.59040024e-03
1.54950452e+00 -9.92804766e-02 4.27687168e-01 3.08264673e-01
9.83724833e-01 -5.32167435e-01 -1.54406333e+00 -5.89675963e-01
2.23371871e-02 -2.57887602e-01 2.85004646e-01 -9.66985881e-01
-1.13545430e+00 1.03243542e+00 8.36787641e-01 3.51764143e-01
1.41749370e+00 -3.12390417e-01 4.12772208e-01 2.87191093e-01
2.46960193e-01 -1.01112568e+00 -5.91382861e-01 5.12736022e-01
9.85122144e-01 -1.64945722e+00 -2.97704875e-01 -5.84956706e-02
-6.99207008e-01 7.37154007e-01 3.86124849e-01 -1.79979466e-02
7.85650015e-01 -2.06935871e-02 4.03150886e-01 1.76361531e-01
-2.32092351e-01 -5.87898850e-01 9.89971235e-02 1.18439555e+00
2.45698556e-01 2.90535778e-01 1.40756443e-01 3.17426413e-01
4.61827442e-02 6.57161325e-02 3.10117483e-01 8.25588763e-01
-3.61319631e-01 -8.41019034e-01 -4.03854787e-01 1.87232405e-01
-3.80489260e-01 -1.49630800e-01 -6.92736879e-02 1.02467716e+00
-8.05347562e-02 8.23025644e-01 1.20878227e-01 -1.05400190e-01
3.86536241e-01 1.42809257e-01 -1.40161946e-01 -4.81807798e-01
-4.24414217e-01 -2.38196611e-01 -1.72725946e-01 -1.53910518e-01
-6.75670445e-01 -7.83125222e-01 -7.83392608e-01 -2.59079039e-01
-1.27983287e-01 7.28771016e-02 7.82381058e-01 9.37270045e-01
4.96523649e-01 4.73015189e-01 7.90725410e-01 -1.10958576e+00
-7.73939490e-01 -1.19437361e+00 -7.79679537e-01 4.28037703e-01
6.42190874e-01 -4.48502839e-01 -2.09713131e-01 4.72026877e-02] | [9.586417198181152, -1.4788589477539062] |
2d58e70a-2457-4037-9bb4-6d42905e7db3 | learning-blind-video-temporal-consistency | 1808.00449 | null | http://arxiv.org/abs/1808.00449v1 | http://arxiv.org/pdf/1808.00449v1.pdf | Learning Blind Video Temporal Consistency | Applying image processing algorithms independently to each frame of a video
often leads to undesired inconsistent results over time. Developing temporally
consistent video-based extensions, however, requires domain knowledge for
individual tasks and is unable to generalize to other applications. In this
paper, we present an efficient end-to-end approach based on deep recurrent
network for enforcing temporal consistency in a video. Our method takes the
original unprocessed and per-frame processed videos as inputs to produce a
temporally consistent video. Consequently, our approach is agnostic to specific
image processing algorithms applied on the original video. We train the
proposed network by minimizing both short-term and long-term temporal losses as
well as the perceptual loss to strike a balance between temporal stability and
perceptual similarity with the processed frames. At test time, our model does
not require computing optical flow and thus achieves real-time speed even for
high-resolution videos. We show that our single model can handle multiple and
unseen tasks, including but not limited to artistic style transfer,
enhancement, colorization, image-to-image translation and intrinsic image
decomposition. Extensive objective evaluation and subject study demonstrate
that the proposed approach performs favorably against the state-of-the-art
methods on various types of videos. | ['Ming-Hsuan Yang', 'Jia-Bin Huang', 'Wei-Sheng Lai', 'Ersin Yumer', 'Eli Shechtman', 'Oliver Wang'] | 2018-08-01 | learning-blind-video-temporal-consistency-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Wei-Sheng_Lai_Real-Time_Blind_Video_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Wei-Sheng_Lai_Real-Time_Blind_Video_ECCV_2018_paper.pdf | eccv-2018-9 | ['intrinsic-image-decomposition', 'video-temporal-consistency'] | ['computer-vision', 'computer-vision'] | [ 3.81891370e-01 -5.51274896e-01 1.61989719e-01 -2.80159056e-01
-5.66633761e-01 -5.85516095e-01 3.33849043e-01 -3.74102563e-01
-5.98920286e-01 6.37191772e-01 -1.50015175e-01 3.43379453e-02
4.18521650e-02 -3.26008379e-01 -9.61252511e-01 -5.87265551e-01
-9.23057571e-02 -2.39901021e-01 2.05832273e-01 -8.48975778e-02
1.54478326e-01 5.03565550e-01 -1.51338851e+00 3.72334957e-01
5.54485500e-01 1.25254858e+00 2.63735116e-01 9.50777948e-01
3.24328840e-01 1.06047583e+00 -3.14928502e-01 -3.06857586e-01
5.88531315e-01 -6.57704234e-01 -6.96492255e-01 4.40949768e-01
8.62723887e-01 -6.95325911e-01 -5.68515241e-01 1.16002285e+00
4.55913454e-01 5.72344840e-01 2.88427263e-01 -1.23810911e+00
-8.91518235e-01 1.44195557e-01 -7.13388085e-01 4.04076070e-01
2.66306579e-01 2.92868495e-01 6.00916207e-01 -9.11425769e-01
8.38371813e-01 1.07160878e+00 5.64315438e-01 6.79931343e-01
-1.36192381e+00 -3.96937996e-01 3.34028423e-01 3.90299261e-01
-1.13537812e+00 -7.15867937e-01 8.90709877e-01 -3.00617874e-01
8.11159611e-01 6.84927702e-02 6.42515242e-01 1.10208130e+00
3.70144963e-01 5.10477841e-01 1.06504631e+00 -3.24324459e-01
2.07857206e-01 -2.24121124e-01 -2.10375816e-01 6.46560252e-01
-3.34032103e-02 2.70624995e-01 -6.14185572e-01 4.29166228e-01
1.04741001e+00 9.74851567e-03 -4.88863170e-01 -3.76514286e-01
-1.34856761e+00 4.35709298e-01 2.37324461e-01 2.63867974e-01
-6.15909994e-01 5.67316413e-01 5.74324608e-01 4.58858699e-01
4.69072133e-01 9.65431482e-02 -4.55379039e-01 -1.82476968e-01
-1.24250162e+00 1.69731975e-01 3.61652821e-01 8.84758532e-01
5.02638400e-01 4.35860991e-01 -1.64446309e-01 7.11917818e-01
-1.92214537e-03 2.68076301e-01 5.07746756e-01 -1.68712127e+00
3.66444141e-01 -2.29938313e-01 3.84567589e-01 -1.21139157e+00
-1.41080260e-01 -2.77912825e-01 -1.00938928e+00 5.74562013e-01
3.68471205e-01 -1.84826076e-01 -9.10864174e-01 2.14746571e+00
1.68187320e-01 4.47769880e-01 9.94515419e-03 1.19624567e+00
4.10855949e-01 7.91993737e-01 1.07466735e-01 -6.60122097e-01
9.54567730e-01 -1.19644332e+00 -9.03167784e-01 -1.11663267e-01
-7.90882036e-02 -8.97385240e-01 9.61522222e-01 4.01061684e-01
-1.74408674e+00 -8.18123639e-01 -9.89757836e-01 -2.70844460e-01
8.82004425e-02 -7.94284120e-02 2.29927525e-01 2.59581536e-01
-1.46574521e+00 8.70478153e-01 -1.08433127e+00 -3.28665614e-01
2.06651762e-01 3.27508062e-01 -4.62636888e-01 1.20422971e-02
-1.01358724e+00 7.82850683e-01 2.76169121e-01 3.50783944e-01
-9.19944823e-01 -6.95989370e-01 -8.78475964e-01 2.63939472e-03
4.62407023e-01 -1.11797976e+00 1.22647822e+00 -1.68715310e+00
-1.75445366e+00 7.04904854e-01 -3.02666992e-01 -5.63552737e-01
7.55519271e-01 -4.18869257e-01 -3.50298971e-01 4.14899111e-01
-1.19687896e-03 8.84033203e-01 1.40060925e+00 -1.07793200e+00
-6.63522422e-01 4.94566411e-02 3.08135208e-02 2.98872083e-01
-3.13755929e-01 2.12219641e-01 -7.81389475e-01 -1.13573778e+00
-4.40385677e-02 -1.01840675e+00 -1.88292950e-01 5.57084143e-01
1.78516865e-01 3.69965345e-01 9.74370003e-01 -8.69242072e-01
9.76261079e-01 -2.16559768e+00 3.54882151e-01 -1.88994870e-01
-2.26074848e-02 1.80254266e-01 -4.38138187e-01 -5.56709431e-02
-2.84951955e-01 -1.68412045e-01 -3.23184252e-01 -3.39812875e-01
-3.16381395e-01 1.38724640e-01 -3.01275700e-01 6.65029049e-01
2.74748951e-01 7.95511603e-01 -1.05120897e+00 -4.97340381e-01
5.38646579e-01 8.08174014e-01 -8.64580035e-01 3.61815721e-01
-7.59300590e-02 5.30598640e-01 -4.74431925e-02 4.03972656e-01
6.43962622e-01 -1.87505826e-01 -3.52762709e-03 -5.07823408e-01
-1.32130727e-01 -1.52542159e-01 -1.15168381e+00 2.17281294e+00
-6.61715925e-01 9.73133504e-01 2.79903948e-01 -9.72721159e-01
3.40843886e-01 4.03557241e-01 7.58551598e-01 -7.53057957e-01
9.87902209e-02 2.10190922e-01 -1.06308259e-01 -5.40936410e-01
6.48606122e-01 -1.54911444e-01 3.81084234e-01 2.39060521e-01
1.39819920e-01 1.84810027e-01 4.37806666e-01 2.01526750e-02
7.90790915e-01 6.08396351e-01 2.90208589e-02 -1.37387574e-01
5.62692761e-01 -4.10191745e-01 7.94003904e-01 5.00231028e-01
-5.48852324e-01 8.59852731e-01 1.14901945e-01 -4.32082772e-01
-1.30290663e+00 -1.20239222e+00 2.64309227e-01 1.05063140e+00
4.01485056e-01 5.87917631e-03 -6.20729089e-01 -3.06852192e-01
-4.11788583e-01 3.59906793e-01 -5.55649519e-01 -7.10896179e-02
-7.52156854e-01 -2.60309756e-01 2.02809855e-01 4.86566722e-01
7.24947274e-01 -1.08881736e+00 -9.73509610e-01 4.37004745e-01
-3.88203323e-01 -1.55122983e+00 -1.14107037e+00 -2.96824515e-01
-9.81987655e-01 -7.98891842e-01 -1.03486252e+00 -9.57819700e-01
6.10738397e-01 4.59605962e-01 1.08173406e+00 -1.12999231e-01
-2.41467893e-01 4.82352048e-01 -2.14603633e-01 3.04703563e-01
-2.86892414e-01 -5.59148788e-01 1.62384495e-01 4.48972285e-01
-3.57917130e-01 -7.26265132e-01 -1.02533472e+00 3.29487830e-01
-1.26940286e+00 1.58751622e-01 3.28993112e-01 9.78112161e-01
6.10376596e-01 8.11125785e-02 2.73873359e-01 -4.32958573e-01
4.06847537e-01 -1.20331347e-02 -5.36437273e-01 2.60148227e-01
-4.16108161e-01 -6.15987740e-03 8.59798610e-01 -6.83785617e-01
-1.26255727e+00 2.32194111e-01 2.04182744e-01 -1.14664447e+00
1.27230749e-01 1.30725160e-01 1.09395079e-01 -1.83884159e-01
4.21425730e-01 3.19848567e-01 1.73358336e-01 -9.90745872e-02
5.46946406e-01 8.96483138e-02 1.08373523e+00 -4.67084587e-01
7.60205209e-01 6.55157089e-01 7.59776607e-02 -6.19933248e-01
-6.04552031e-01 -2.35463232e-01 -5.59719503e-01 -3.96545887e-01
9.49970484e-01 -1.04559183e+00 -7.37324655e-01 6.15128517e-01
-1.39679909e+00 -3.33754301e-01 -2.78605223e-01 5.99652886e-01
-9.39745069e-01 5.81200182e-01 -8.71676981e-01 -5.03138423e-01
-4.85361785e-01 -1.15738773e+00 8.39109242e-01 8.02925751e-02
-1.29160574e-02 -9.64724600e-01 -1.53399795e-01 4.97245044e-02
5.97027421e-01 3.83438975e-01 4.44191039e-01 2.53892034e-01
-7.14401603e-01 1.88427478e-01 -3.09463084e-01 8.09405684e-01
2.06557959e-01 2.56030291e-01 -7.45444357e-01 -6.02363884e-01
1.85025692e-01 -2.71593183e-01 8.74121487e-01 6.69024646e-01
1.20245314e+00 -4.85516608e-01 3.68035704e-01 8.50373328e-01
1.54814768e+00 2.01736912e-01 7.28551745e-01 3.76269907e-01
6.66040301e-01 6.23475194e-01 6.20513737e-01 3.57265383e-01
-6.52041808e-02 8.53550434e-01 3.46311152e-01 -2.88081229e-01
-3.37861389e-01 1.10048287e-01 7.77373195e-01 6.04303479e-01
-3.17314893e-01 -2.69996703e-01 -3.53013575e-01 5.35229981e-01
-1.91965425e+00 -1.44765317e+00 2.49602705e-01 2.08694911e+00
7.32881606e-01 -8.46043378e-02 8.41677263e-02 -4.87295389e-02
9.44452286e-01 2.30514377e-01 -6.62389278e-01 -3.79833877e-01
-1.67653903e-01 1.11322463e-01 3.76206964e-01 5.61321437e-01
-1.18069887e+00 8.69257390e-01 6.13386917e+00 5.83171129e-01
-1.39108825e+00 1.58821478e-01 8.31773520e-01 -4.52238619e-01
6.93865195e-02 -1.50162578e-01 8.19160044e-02 4.35067624e-01
8.00514460e-01 -2.61190325e-01 8.57483685e-01 5.81280410e-01
6.49655759e-01 1.16618015e-01 -1.34189725e+00 1.30240595e+00
1.82697594e-01 -1.45142913e+00 1.23054720e-01 -3.42250854e-01
9.09180403e-01 -1.81351990e-01 3.42019171e-01 -1.63309991e-01
-1.60124719e-01 -7.59175777e-01 1.19982946e+00 4.63416994e-01
9.97909307e-01 -6.04574442e-01 3.91037285e-01 -2.72340447e-01
-1.21430326e+00 -2.29599457e-02 -6.32640794e-02 1.19434550e-01
3.86761785e-01 1.59749150e-01 -4.64007594e-02 4.36661303e-01
9.11878765e-01 1.12199342e+00 -2.61282682e-01 8.45356762e-01
1.42893881e-01 1.07608721e-01 -1.19449511e-01 5.65258920e-01
3.58888716e-01 -2.76970893e-01 5.64011514e-01 1.19424093e+00
4.50436771e-01 1.54847413e-01 4.47017187e-03 8.32989812e-01
-2.17322618e-01 -1.38174474e-01 -5.34340441e-01 1.05110534e-01
1.04237363e-01 1.17284691e+00 -6.69268191e-01 -3.63461226e-01
-5.18246233e-01 1.54320812e+00 1.54789373e-01 7.12707400e-01
-1.07146859e+00 -2.59732395e-01 7.16531396e-01 -4.14755195e-02
4.41945374e-01 -3.90124440e-01 -3.61058973e-02 -1.39482689e+00
3.41760755e-01 -9.29652810e-01 3.97046387e-01 -1.03811121e+00
-1.20154667e+00 7.41957724e-01 -1.67620093e-01 -1.59381652e+00
-4.68272001e-01 -4.40734118e-01 -4.63480115e-01 6.31111383e-01
-1.64074695e+00 -8.65873754e-01 -3.99616271e-01 1.08163393e+00
8.31210673e-01 -3.86722572e-02 3.13756466e-01 6.49289072e-01
-4.83755559e-01 4.97176260e-01 1.21955030e-01 1.08232908e-01
9.77136374e-01 -7.03849852e-01 2.50999629e-01 1.42351973e+00
-5.44322946e-04 4.69583154e-01 9.33047354e-01 -2.19185606e-01
-1.42758834e+00 -1.41898990e+00 4.94477719e-01 -6.43534511e-02
6.42771482e-01 4.98278327e-02 -8.79246235e-01 5.46299338e-01
5.09057224e-01 4.36955184e-01 1.01817168e-01 -5.28600514e-01
-3.67923826e-01 -3.80124032e-01 -1.00519240e+00 7.57151008e-01
1.06871831e+00 -5.68834007e-01 -2.40719318e-01 1.13008194e-01
7.40459740e-01 -5.25717199e-01 -7.46888936e-01 2.47436538e-01
6.03614867e-01 -1.15200269e+00 1.00024307e+00 -6.09745502e-01
8.56395781e-01 -4.60963905e-01 -2.01912776e-01 -1.01018667e+00
-3.71725827e-01 -1.03735304e+00 -1.13407120e-01 1.02869582e+00
1.75135493e-01 -2.51518160e-01 5.58855653e-01 8.85041952e-01
-1.11062042e-01 -4.32279110e-01 -8.77059340e-01 -9.46374655e-01
-2.06614956e-01 -5.10419548e-01 5.70037076e-03 8.93806636e-01
-4.41141099e-01 2.99090776e-03 -8.41344595e-01 1.85458973e-01
8.62391531e-01 2.43419603e-01 4.81005043e-01 -4.97415215e-01
-5.27966559e-01 -4.58415151e-01 -4.40229863e-01 -9.83889937e-01
2.29345575e-01 -4.97629821e-01 1.04013719e-01 -1.08648467e+00
1.99743181e-01 -4.43430524e-03 -4.92188036e-01 3.05690944e-01
4.44846740e-03 6.31256938e-01 4.99714822e-01 1.47408724e-01
-6.93661451e-01 5.39002955e-01 1.39409912e+00 -2.09102094e-01
-2.75863916e-01 -3.18087041e-01 -3.66922885e-01 6.19755983e-01
4.73755807e-01 -2.81606913e-01 -5.06305575e-01 -7.88699329e-01
-6.04890883e-02 4.08818752e-01 6.09462678e-01 -9.92985010e-01
2.31917486e-01 -3.80224854e-01 5.35799563e-01 -1.20118655e-01
4.05700088e-01 -9.24326777e-01 3.28643203e-01 4.96437520e-01
-5.41921258e-01 4.48913097e-01 2.76144475e-01 7.75366962e-01
-3.83697927e-01 5.22987805e-02 1.24160159e+00 -1.03442505e-01
-9.24226701e-01 5.61453760e-01 -1.36198908e-01 1.13042861e-01
1.13541329e+00 -3.94550085e-01 -8.67420509e-02 -6.35947764e-01
-7.94143081e-01 -1.13038067e-02 6.76182687e-01 5.50731957e-01
8.73469234e-01 -1.27649915e+00 -8.31807256e-01 9.38387439e-02
-3.86255324e-01 -4.24506217e-01 6.27880514e-01 7.71835089e-01
-7.69958496e-01 1.45971283e-01 -7.60617852e-01 -6.31313503e-01
-1.32347929e+00 9.08131838e-01 4.80858415e-01 1.04360906e-02
-7.79332757e-01 5.93170702e-01 3.78278285e-01 3.79261136e-01
3.71326804e-01 -2.22963139e-01 1.98936448e-01 -1.52120173e-01
4.79267091e-01 2.85719424e-01 -9.29191634e-02 -7.11492777e-01
-2.69908339e-01 8.72510672e-01 -1.00888841e-01 -4.39708054e-01
1.24097788e+00 -4.83134151e-01 -4.59661111e-02 3.03913116e-01
1.49817371e+00 -4.03978795e-01 -1.81454825e+00 -1.39715761e-01
-4.62877512e-01 -7.88728654e-01 1.69816703e-01 -5.68752110e-01
-1.64600730e+00 7.97077537e-01 7.94739604e-01 -2.15438470e-01
1.67716360e+00 -5.85808635e-01 9.64984298e-01 2.05006078e-01
2.43851617e-01 -1.14394796e+00 2.70947516e-01 2.32155368e-01
9.91607189e-01 -1.31013763e+00 -1.23594515e-01 -1.29970461e-01
-6.93184376e-01 1.13954759e+00 5.55565894e-01 -2.47885466e-01
4.04728383e-01 2.50848737e-02 6.50494546e-02 4.18610007e-01
-1.03897524e+00 1.44289553e-01 3.59219432e-01 4.61281508e-01
4.79687929e-01 -4.77973342e-01 -2.34164461e-01 -1.09369546e-01
2.84033746e-01 2.50138819e-01 7.68946648e-01 8.41598272e-01
6.01054989e-02 -8.57254505e-01 -2.81396985e-01 -2.20235623e-03
-6.87321246e-01 -1.14961155e-01 2.03211859e-01 6.55938208e-01
-3.54079753e-02 8.96603346e-01 1.12100862e-01 -7.08999336e-02
2.44493708e-01 -2.79993355e-01 7.43156552e-01 2.02647094e-02
-4.74568307e-01 3.36943358e-01 -1.44856706e-01 -9.68679130e-01
-9.93489027e-01 -5.99322319e-01 -1.11854792e+00 -4.38831121e-01
1.50296301e-01 -2.36560032e-01 3.93193781e-01 5.42863846e-01
4.71704155e-01 6.11481547e-01 7.99294055e-01 -1.28337872e+00
-3.33572298e-01 -4.32365596e-01 -4.94752437e-01 9.50835705e-01
7.50946879e-01 -4.43602830e-01 -2.80256152e-01 7.15116143e-01] | [10.851390838623047, -1.3979383707046509] |
994671d9-5f67-4b8e-a360-2344528da146 | autodepthnet-high-frame-rate-depth-map | 2305.14731 | null | https://arxiv.org/abs/2305.14731v1 | https://arxiv.org/pdf/2305.14731v1.pdf | AutoDepthNet: High Frame Rate Depth Map Reconstruction using Commodity Depth and RGB Cameras | Depth cameras have found applications in diverse fields, such as computer vision, artificial intelligence, and video gaming. However, the high latency and low frame rate of existing commodity depth cameras impose limitations on their applications. We propose a fast and accurate depth map reconstruction technique to reduce latency and increase the frame rate in depth cameras. Our approach uses only a commodity depth camera and color camera in a hybrid camera setup; our prototype is implemented using a Kinect Azure depth camera at 30 fps and a high-speed RGB iPhone 11 Pro camera captured at 240 fps. The proposed network, AutoDepthNet, is an encoder-decoder model that captures frames from the high-speed RGB camera and combines them with previous depth frames to reconstruct a stream of high frame rate depth maps. On GPU, with a 480 x 270 output resolution, our system achieves an inference time of 8 ms, enabling real-time use at up to 200 fps with parallel processing. AutoDepthNet can estimate depth values with an average RMS error of 0.076, a 44.5% improvement compared to an optical flow-based comparison method. Our method can also improve depth map quality by estimating depth values for missing and invalidated pixels. The proposed method can be easily applied to existing depth cameras and facilitates the use of depth cameras in applications that require high-speed depth estimation. We also showcase the effectiveness of the framework in upsampling different sparse datasets e.g. video object segmentation. As a demonstration of our method, we integrated our framework into existing body tracking systems and demonstrated the robustness of the proposed method in such applications. | ['Robert Xiao', 'Peyman Gholami'] | 2023-05-24 | null | null | null | null | ['video-object-segmentation', 'video-semantic-segmentation'] | ['computer-vision', 'computer-vision'] | [ 3.60879302e-01 -1.38014972e-01 -8.16555396e-02 -2.30032623e-01
-4.35407043e-01 -3.82854730e-01 -8.07089210e-02 -2.39477605e-01
-8.21283102e-01 4.64212388e-01 -1.67070419e-01 -5.30893430e-02
6.34910345e-01 -1.02680337e+00 -6.26770318e-01 -3.82963210e-01
1.62211478e-01 2.59805262e-01 7.55089045e-01 2.41678238e-01
7.20256269e-02 5.67479730e-01 -1.83590913e+00 1.77981883e-01
3.86243492e-01 1.14463937e+00 2.45641246e-01 1.06383848e+00
2.01719310e-02 7.06208527e-01 -5.26820242e-01 -1.95010677e-01
5.43878555e-01 -2.30984107e-01 -4.26361084e-01 2.34474942e-01
6.52574360e-01 -1.34587801e+00 -6.38708353e-01 7.61685491e-01
6.66223764e-01 -8.08753520e-02 -1.17903575e-01 -1.13271439e+00
3.64327908e-01 -1.05451591e-01 -8.04821670e-01 4.08213615e-01
7.10920751e-01 3.07546854e-01 2.82881320e-01 -5.43169796e-01
6.12199068e-01 1.06638372e+00 6.14406645e-01 8.66859555e-01
-8.01022172e-01 -7.06373692e-01 -1.64927319e-01 1.01763248e-01
-1.13911223e+00 -3.60265076e-01 5.91802359e-01 -1.53380737e-01
9.87431288e-01 2.59528533e-02 1.12810838e+00 7.82869339e-01
2.34272674e-01 5.77152789e-01 9.18617487e-01 -1.31206259e-01
4.12641108e-01 -2.64148921e-01 -2.64128894e-01 8.12007189e-01
3.32172722e-01 1.57754287e-01 -7.47995615e-01 8.21224377e-02
1.47439885e+00 2.90639400e-01 -3.85936260e-01 -5.47943003e-02
-1.25594664e+00 5.44398725e-01 2.66781986e-01 -1.42638966e-01
-2.57292897e-01 5.59662461e-01 4.30402637e-01 -3.59482765e-02
2.78745651e-01 -2.78089583e-01 -1.75909445e-01 -5.81542611e-01
-1.02201402e+00 2.57322416e-02 6.68095052e-01 8.81716967e-01
7.49929786e-01 -6.47090748e-03 4.35365081e-01 4.37080324e-01
3.44191194e-01 7.59910107e-01 4.85934287e-01 -1.64557505e+00
6.12465918e-01 3.67324352e-01 -8.99289101e-02 -9.29409981e-01
-3.68668258e-01 3.16289634e-01 -6.96348667e-01 6.53245270e-01
5.73447764e-01 -2.92470127e-01 -6.44100904e-01 1.10772789e+00
6.36671960e-01 5.07984579e-01 -8.80020186e-02 1.30514419e+00
7.56401718e-01 5.75664580e-01 -3.61493707e-01 -1.06652111e-01
1.23062527e+00 -7.25354970e-01 -5.02002418e-01 -3.26066852e-01
3.56062651e-01 -5.79816341e-01 8.11759949e-01 7.83285022e-01
-1.25473523e+00 -4.85470325e-01 -1.16551638e+00 -4.00682956e-01
2.78448880e-01 6.24902034e-03 7.58922935e-01 8.88814926e-01
-1.11565351e+00 6.35869920e-01 -1.53601468e+00 -3.37900460e-01
2.35750020e-01 8.40685189e-01 -4.36495841e-01 -3.01951587e-01
-6.18233323e-01 3.94335181e-01 2.87446588e-01 1.46545529e-01
-7.40121484e-01 -6.77799106e-01 -9.00355279e-01 -1.99942037e-01
1.45123929e-01 -9.31044281e-01 1.13443291e+00 -9.15525854e-01
-1.95561290e+00 8.21352839e-01 -2.16267332e-01 -4.41069126e-01
5.55197895e-01 -4.39891100e-01 8.99116024e-02 8.39357376e-01
-2.67990511e-02 8.62963438e-01 6.20429099e-01 -6.33545101e-01
-9.13232744e-01 -6.32448018e-01 2.92051345e-01 4.01466101e-01
-1.44893616e-01 -2.13829353e-01 -7.46756017e-01 -7.18318149e-02
4.33743507e-01 -8.89888704e-01 -1.32935449e-01 7.57041872e-01
4.76549715e-02 6.13825977e-01 8.70785534e-01 -3.94408554e-01
7.29057193e-01 -1.99921882e+00 -6.45692647e-02 1.08975451e-02
3.71965379e-01 3.03011835e-01 3.39566827e-01 1.02564867e-03
3.36645931e-01 -4.69034761e-01 -1.59309730e-01 -4.68459278e-01
-7.11524069e-01 4.48729515e-01 1.44478038e-01 6.98559105e-01
-3.53012681e-01 4.26894158e-01 -9.64071631e-01 -5.27624547e-01
8.53307128e-01 8.76873374e-01 -7.24812865e-01 2.87972003e-01
2.08027512e-01 5.58363199e-01 -8.33095312e-02 8.78826141e-01
7.39576042e-01 5.24404496e-02 2.24091664e-01 -2.63239712e-01
-1.58146441e-01 5.45141846e-02 -1.37903953e+00 2.12613010e+00
-4.53672200e-01 8.98346424e-01 2.63613045e-01 -4.46739733e-01
7.38827884e-01 1.43656522e-01 7.30804801e-01 -6.41506135e-01
2.47012258e-01 2.47616082e-01 -3.00334901e-01 -4.99404341e-01
6.21484578e-01 -1.53153285e-01 3.93661112e-01 3.80145490e-01
-1.18827097e-01 -2.33084232e-01 6.89777508e-02 3.49560864e-02
1.28479850e+00 4.13155526e-01 1.37336209e-01 2.37678126e-01
4.28826004e-01 -8.74219164e-02 5.63876688e-01 3.31000984e-01
-3.93293977e-01 9.27816391e-01 1.89517379e-01 -5.37967741e-01
-1.01090717e+00 -9.70645130e-01 4.46982756e-02 4.41451132e-01
5.69613814e-01 -5.54829478e-01 -6.45497918e-01 -1.78916276e-01
-9.50382128e-02 -1.83294535e-01 -1.71137333e-01 2.01666102e-01
-7.13410199e-01 -4.27855939e-01 5.53492785e-01 7.89308906e-01
8.49003673e-01 -6.24758244e-01 -1.41985965e+00 2.80204415e-01
-1.12156846e-01 -1.44523048e+00 -1.86135337e-01 -1.63549304e-01
-1.50082374e+00 -1.26084423e+00 -6.95800304e-01 -4.23860520e-01
4.90845293e-01 3.80739301e-01 9.18069303e-01 -6.54425994e-02
-4.24089074e-01 6.28790498e-01 -2.17548341e-01 1.49793163e-01
1.38938418e-02 -2.45595858e-01 2.63597239e-02 -2.99493909e-01
2.84856409e-01 -6.19465053e-01 -1.10246706e+00 2.14856312e-01
-1.00875914e+00 2.05018714e-01 1.22078165e-01 5.41763663e-01
7.02162206e-01 -1.97701588e-01 -1.69869125e-01 -6.72637284e-01
-5.26012965e-02 -7.24367872e-02 -9.78269756e-01 -5.16658366e-01
-3.38426918e-01 -2.82961041e-01 4.23952878e-01 -2.33169585e-01
-7.77038217e-01 5.63071728e-01 -4.15796906e-01 -6.58876300e-01
7.45840324e-03 -9.07965377e-02 5.37660718e-02 -2.81765103e-01
3.56912315e-01 8.81842300e-02 4.39349771e-01 -2.24158511e-01
1.09936908e-01 6.00256205e-01 9.35236037e-01 -3.48372430e-01
2.85017043e-01 1.04976141e+00 1.17738284e-01 -9.97977555e-01
-2.48954460e-01 -5.76070726e-01 -5.47931612e-01 -4.75331396e-01
8.64317536e-01 -1.39354491e+00 -1.21269846e+00 8.01366866e-01
-1.17941940e+00 -3.30440044e-01 -9.93764102e-02 7.65181005e-01
-5.66559076e-01 5.70041597e-01 -1.06013179e+00 -6.32899404e-01
-3.69530797e-01 -1.22258210e+00 1.35543001e+00 4.42846626e-01
-2.68685427e-02 -9.76709127e-01 -6.92304745e-02 5.11569679e-01
1.28850907e-01 4.28218126e-01 7.25276861e-03 4.73818928e-01
-8.34150493e-01 -1.12946574e-02 -1.82592735e-01 2.03870863e-01
1.29714921e-01 4.14058715e-02 -1.18701279e+00 -1.73378050e-01
2.37750337e-02 -2.05919147e-01 6.13091648e-01 6.26791418e-01
1.14778423e+00 1.39314190e-01 -2.24616036e-01 1.05467010e+00
1.86899328e+00 2.70753652e-01 1.01182055e+00 4.75401402e-01
9.90148604e-01 3.01913887e-01 7.05156505e-01 6.86402202e-01
4.20180470e-01 8.01915884e-01 7.14653254e-01 -1.60295665e-01
-1.45168900e-01 -8.80461652e-03 4.81671542e-01 6.64714336e-01
-4.22521174e-01 8.13450292e-02 -8.03726971e-01 2.73591518e-01
-1.64368606e+00 -9.84899580e-01 -2.80332834e-01 2.53197765e+00
7.76364207e-01 1.06238149e-01 2.29045004e-01 4.02696818e-01
5.69542527e-01 -5.02827987e-02 -5.04076660e-01 -3.71252537e-01
3.74576062e-01 5.16820967e-01 8.00974727e-01 6.28634810e-01
-8.62560332e-01 6.89590394e-01 5.82675457e+00 2.48009741e-01
-1.33214486e+00 6.32672757e-02 3.93345624e-01 -7.29330420e-01
8.86090547e-02 -2.44163677e-01 -8.24654520e-01 5.81366837e-01
9.39732432e-01 1.18109427e-01 2.72444785e-01 1.07747138e+00
3.42933208e-01 -7.32515395e-01 -1.12021077e+00 1.65575492e+00
1.12177350e-01 -1.26901329e+00 -3.72870982e-01 2.74310738e-01
4.46105450e-01 5.49152233e-02 -4.05963331e-01 -3.05864602e-01
-2.13424161e-01 -6.82399392e-01 5.11991441e-01 6.46070614e-02
1.07972419e+00 -8.58888209e-01 7.93026090e-01 2.44285896e-01
-1.22676384e+00 2.74694026e-01 -5.65427244e-01 -5.79581141e-01
3.78970772e-01 7.11123168e-01 -7.48748779e-01 1.18607700e-01
8.91194224e-01 8.05379093e-01 -1.22660801e-01 9.32708621e-01
-7.83075839e-02 1.58224732e-01 -7.10776389e-01 2.06814408e-01
-9.66669172e-02 -1.97457448e-01 2.05825999e-01 9.57650602e-01
3.59125912e-01 4.26786065e-01 -6.67220578e-02 4.69597042e-01
8.44099298e-02 -4.69167978e-01 -6.60729945e-01 4.46995467e-01
3.09299290e-01 1.13238692e+00 -9.52903748e-01 -4.60695535e-01
-6.02318347e-01 1.34296787e+00 -2.24716276e-01 -9.48065817e-02
-1.06690907e+00 -4.62289810e-01 9.51279104e-01 3.18135887e-01
1.42060265e-01 -4.40754056e-01 -4.51536655e-01 -1.38078868e+00
1.43492565e-01 -5.64231217e-01 7.34205320e-02 -8.59871566e-01
-6.54554427e-01 5.22462428e-01 -2.56115764e-01 -1.50214684e+00
-3.93119663e-01 -6.94360673e-01 -1.20527424e-01 5.47527194e-01
-1.51659536e+00 -6.19977653e-01 -9.79699194e-01 9.71480548e-01
5.56773961e-01 2.56076306e-01 5.92202187e-01 6.92176342e-01
-4.55588132e-01 2.02761218e-01 -1.66178733e-01 1.96413517e-01
4.96756673e-01 -1.05701315e+00 3.21633935e-01 9.25391436e-01
-1.32187605e-01 3.72870803e-01 3.55817884e-01 -6.39363289e-01
-1.75943291e+00 -8.27384710e-01 4.00208831e-01 -1.41474053e-01
1.22288644e-01 -2.67428666e-01 -4.48023707e-01 5.01504242e-01
-9.20248255e-02 4.21545058e-01 5.34805417e-01 -5.96298456e-01
9.36984941e-02 -4.13880676e-01 -1.35079265e+00 2.36511320e-01
9.76025820e-01 -4.30798471e-01 -1.77834928e-01 7.50871226e-02
4.15544063e-01 -1.20557368e+00 -7.53304303e-01 9.10082161e-02
9.21491146e-01 -1.65729880e+00 1.01953256e+00 5.49232125e-01
6.74289882e-01 -4.95578259e-01 -1.11041322e-01 -6.05464578e-01
2.37638682e-01 -3.87511134e-01 -2.17193127e-01 8.60982299e-01
-3.90195310e-01 -5.10356307e-01 1.36646259e+00 8.73765707e-01
2.13285685e-02 -6.18327677e-01 -1.15724814e+00 -3.85536641e-01
-5.79644084e-01 -7.96777129e-01 1.25314921e-01 3.37595314e-01
-1.10921552e-02 -4.63915914e-02 -2.30033308e-01 3.26429397e-01
8.32831323e-01 -7.50951990e-02 9.22011554e-01 -8.63749146e-01
-4.66009289e-01 9.98870954e-02 -9.64348853e-01 -1.38310599e+00
-2.63974488e-01 -2.70934880e-01 -1.91760480e-01 -1.35048366e+00
1.37637509e-02 -2.37026632e-01 4.31887329e-01 2.42409110e-01
1.41785949e-01 7.90619254e-01 2.54800051e-01 1.26467675e-01
-2.86617875e-01 1.70791477e-01 1.03875995e+00 1.58552721e-01
-3.05298269e-01 -5.03877699e-02 -9.71621089e-03 8.93346250e-01
4.22917843e-01 -3.71593595e-01 -4.60724622e-01 -6.47512436e-01
-8.35940521e-03 5.01856983e-01 4.70594943e-01 -1.55691862e+00
3.20466042e-01 2.11711451e-01 6.48412585e-01 -5.05895376e-01
8.11447680e-01 -1.02326834e+00 2.74870276e-01 8.52697432e-01
3.38972747e-01 2.24889517e-01 2.59943008e-01 3.44157577e-01
-2.85662115e-01 1.77212842e-02 7.69790053e-01 -4.24254239e-01
-9.89603817e-01 3.15885514e-01 -3.56524348e-01 -1.38896480e-01
1.18602037e+00 -7.63884902e-01 -2.02924266e-01 -3.84003311e-01
-4.83649939e-01 -1.15220226e-01 8.71118009e-01 6.09159432e-02
1.12533784e+00 -1.16769683e+00 -1.51106045e-01 5.04032552e-01
-2.98699766e-01 3.62095773e-01 1.87871084e-01 6.44004047e-01
-1.57496381e+00 2.17023298e-01 -4.84054238e-01 -1.13715410e+00
-1.40570915e+00 -6.84900284e-02 4.51595247e-01 1.29297137e-01
-9.71513569e-01 6.38366044e-01 -4.35135514e-02 2.33938783e-01
2.21569613e-01 -7.00416803e-01 1.86508149e-01 -3.10697198e-01
9.17568624e-01 7.29476392e-01 8.16832036e-02 -3.79675865e-01
-5.56217074e-01 9.45243061e-01 4.21786100e-01 -5.01442611e-01
1.12794030e+00 -3.26991886e-01 5.07708937e-02 3.66055548e-01
1.16206408e+00 -1.18971929e-01 -1.75064230e+00 3.34734589e-01
-5.44787228e-01 -9.68059897e-01 3.25717002e-01 -2.38107845e-01
-1.57141864e+00 9.10304129e-01 8.44160616e-01 -4.36005682e-01
1.44788301e+00 -3.22441995e-01 1.21509969e+00 -6.83354679e-03
8.92245173e-01 -8.76411378e-01 -3.20266038e-02 1.74090296e-01
3.09132412e-02 -1.34219205e+00 2.67193198e-01 -5.87016225e-01
-4.73848194e-01 1.47668433e+00 6.95889354e-01 -3.24582160e-01
2.02628881e-01 7.54534364e-01 1.43437251e-01 4.29585800e-02
-4.16257173e-01 -1.35601118e-01 -3.05198371e-01 7.76689112e-01
3.62756699e-01 -2.34914213e-01 -1.41978666e-01 -2.19888315e-01
9.24598053e-03 5.91149569e-01 1.07052088e+00 1.06211662e+00
-2.42356524e-01 -1.04661763e+00 -5.10173142e-01 1.45258009e-01
-5.86374938e-01 2.76122577e-02 2.62421280e-01 6.83703959e-01
2.20584929e-01 8.71899307e-01 4.28915113e-01 -3.67170930e-01
1.93537876e-01 -4.34010506e-01 7.97265232e-01 -6.06010556e-01
-5.90408444e-01 1.63884610e-01 -1.91351771e-01 -1.30411375e+00
-8.17683339e-01 -5.42893827e-01 -1.43258739e+00 -6.77070439e-01
9.81880724e-02 -4.11171824e-01 9.82978642e-01 5.98838270e-01
2.73667932e-01 2.19267234e-01 2.71083057e-01 -1.15224743e+00
2.59908259e-01 -4.52741593e-01 -7.22533464e-01 1.28812984e-01
4.52284306e-01 -4.06569779e-01 -2.73982555e-01 3.53749931e-01] | [8.868889808654785, -2.240626811981201] |
97ce08a8-b53b-450e-89ed-77991acce353 | unified-embedding-based-personalized | 2306.04833 | null | https://arxiv.org/abs/2306.04833v1 | https://arxiv.org/pdf/2306.04833v1.pdf | Unified Embedding Based Personalized Retrieval in Etsy Search | Embedding-based neural retrieval is a prevalent approach to address the semantic gap problem which often arises in product search on tail queries. In contrast, popular queries typically lack context and have a broad intent where additional context from users historical interaction can be helpful. In this paper, we share our novel approach to address both: the semantic gap problem followed by an end to end trained model for personalized semantic retrieval. We propose learning a unified embedding model incorporating graph, transformer and term-based embeddings end to end and share our design choices for optimal tradeoff between performance and efficiency. We share our learnings in feature engineering, hard negative sampling strategy, and application of transformer model, including a novel pre-training strategy and other tricks for improving search relevance and deploying such a model at industry scale. Our personalized retrieval model significantly improves the overall search experience, as measured by a 5.58% increase in search purchase rate and a 2.63% increase in site-wide conversion rate, aggregated across multiple A/B tests - on live traffic. | ['Thrivikrama Taula', 'Ethan Benjamin', 'Siddharth Subramaniyam', 'Rishikesh Jha'] | 2023-06-07 | null | null | null | null | ['feature-engineering', 'semantic-retrieval'] | ['methodology', 'natural-language-processing'] | [-6.30541891e-02 -3.22032779e-01 -6.41158044e-01 -4.42506850e-01
-1.28679574e+00 -7.21317410e-01 4.56741691e-01 1.74553707e-01
-2.98118174e-01 3.04850144e-03 3.10235620e-01 -4.79593068e-01
-6.08143270e-01 -6.15875661e-01 -3.83502603e-01 -5.87452129e-02
1.54989110e-02 4.14009750e-01 8.24171752e-02 -6.38038516e-01
2.80434847e-01 1.16956659e-01 -1.22695589e+00 1.61867768e-01
7.27445483e-01 1.51812005e+00 1.78807601e-01 6.68495893e-01
-9.44009274e-02 3.89174461e-01 -2.64545321e-01 -6.01974428e-01
5.99327147e-01 2.49944031e-01 -7.63967216e-01 -3.84507388e-01
6.35833621e-01 -5.34038603e-01 -8.08840513e-01 6.61323965e-01
9.23076630e-01 3.49808335e-01 5.51568866e-01 -1.27578914e+00
-1.29012120e+00 4.29879993e-01 -4.92460459e-01 3.18976879e-01
3.00129861e-01 1.39562249e-01 1.75524819e+00 -1.00484633e+00
3.36030066e-01 9.17920053e-01 5.14464796e-01 2.80124217e-01
-1.23575342e+00 -4.40386623e-01 2.24943906e-01 2.57272750e-01
-1.38431036e+00 -2.55743951e-01 6.99079216e-01 -2.10028276e-01
1.29914045e+00 3.73196393e-01 4.39716786e-01 1.13949215e+00
4.71540242e-02 9.85750496e-01 2.14915827e-01 -2.72594243e-01
3.85873653e-02 4.66239989e-01 5.06586969e-01 3.75166774e-01
2.79536992e-01 1.84319802e-02 -5.03497839e-01 -5.33793032e-01
4.93844539e-01 5.72590411e-01 2.82494202e-02 -4.28590089e-01
-4.55484837e-01 1.13816333e+00 5.33112466e-01 1.63910165e-02
-3.27915013e-01 5.25089860e-01 3.69712383e-01 4.91353840e-01
4.07383710e-01 7.57042825e-01 -8.37686658e-01 -3.18917930e-02
-8.54983270e-01 4.92832720e-01 1.00983632e+00 1.24724305e+00
6.85541093e-01 -2.75820941e-01 -5.04097581e-01 1.21023488e+00
4.06377196e-01 5.49660444e-01 6.21838331e-01 -8.60415816e-01
2.50191987e-01 3.74017715e-01 2.44946703e-01 -1.09489048e+00
6.31813034e-02 -8.20848107e-01 -3.09313256e-02 -6.33471787e-01
-1.99515268e-01 4.29089703e-02 -8.37462485e-01 1.54438257e+00
4.60212398e-03 -8.94754305e-02 -2.12896034e-01 7.12192535e-01
4.41700011e-01 6.55126333e-01 6.22224845e-02 3.23162049e-01
1.54205501e+00 -1.45703125e+00 -4.88735467e-01 -3.77285451e-01
8.03652227e-01 -9.83245850e-01 1.39614844e+00 2.04012007e-01
-1.03524423e+00 -3.86412084e-01 -9.62695956e-01 -4.67582256e-01
-6.78437173e-01 -1.74951404e-02 9.98719573e-01 6.21011317e-01
-1.14658093e+00 5.82585275e-01 -2.92400539e-01 -4.59827721e-01
2.09775075e-01 5.87256849e-01 1.18277542e-01 -2.41929665e-01
-1.32878625e+00 5.17815888e-01 -1.53011814e-01 -2.90442973e-01
-4.44834888e-01 -1.08205843e+00 -5.10836720e-01 6.05256140e-01
5.47996998e-01 -9.13736701e-01 1.71256506e+00 -2.57454842e-01
-1.22242177e+00 4.02638733e-01 -4.69124578e-02 -2.86355764e-01
-6.98563755e-02 -6.44283414e-01 -6.60041571e-01 -1.56581029e-01
8.06206390e-02 3.89465451e-01 8.46084595e-01 -7.76670814e-01
-7.34560072e-01 -3.43658090e-01 7.83910230e-02 3.04902852e-01
-8.05247962e-01 -1.71376333e-01 -1.07305229e+00 -6.60750329e-01
-3.11915696e-01 -8.60214531e-01 -3.34250391e-01 8.52670521e-02
-5.05298451e-02 -5.85417211e-01 8.11964035e-01 -6.31437600e-01
1.60358047e+00 -2.30456781e+00 -3.10675323e-01 5.13716221e-01
2.98452318e-01 1.77918613e-01 -4.90078628e-01 7.29348421e-01
3.16055238e-01 4.23645228e-01 6.37929916e-01 -1.40037909e-01
5.76937735e-01 -1.85140938e-01 -5.57471931e-01 -1.30685195e-01
1.04545406e-03 1.43340981e+00 -9.00598943e-01 -1.22068167e-01
-7.98492432e-02 4.48734492e-01 -8.79638672e-01 1.47867754e-01
-3.31288517e-01 -5.65692961e-01 -8.08006287e-01 9.95745242e-01
4.44988251e-01 -8.31814706e-01 1.04143456e-01 -2.75276095e-01
4.94964451e-01 5.00632942e-01 -6.57826543e-01 1.86757576e+00
-9.36124682e-01 4.10561293e-01 -9.28479433e-03 -5.25395155e-01
5.98832965e-01 1.32271433e-02 5.79534829e-01 -1.32857656e+00
-1.98033452e-02 2.45560437e-01 -5.32598436e-01 -3.74745965e-01
1.02901709e+00 5.97160280e-01 -1.63313285e-01 6.93852484e-01
-9.82197002e-03 1.33504882e-01 -2.14232709e-02 2.95795083e-01
1.34843123e+00 -2.91271955e-01 -1.54711068e-01 -1.56704053e-01
-5.42901941e-02 -1.54946640e-01 -2.05617268e-02 9.89795327e-01
4.44150763e-03 2.44645268e-01 9.15625095e-02 -3.30725968e-01
-1.02093625e+00 -9.19570327e-01 1.50362343e-01 1.78861165e+00
3.83463115e-01 -6.86298788e-01 -2.21523255e-01 -6.97621346e-01
4.98144478e-01 7.08089769e-01 -2.86172420e-01 -6.44826055e-01
-3.33240569e-01 -3.55237007e-01 6.31232783e-02 5.27529359e-01
1.63422748e-01 -5.72515845e-01 -1.23820819e-01 2.34729692e-01
1.20723113e-01 -8.36429656e-01 -1.38213527e+00 1.59766600e-01
-8.28822672e-01 -7.50972152e-01 -7.08248138e-01 -8.85576129e-01
1.20517381e-01 8.02484214e-01 1.19071877e+00 7.13847950e-02
-6.05879724e-01 8.07905257e-01 -3.76068801e-01 -1.01278692e-01
4.40624028e-01 4.68614370e-01 -1.25890598e-01 -4.56865162e-01
7.23606229e-01 -4.46268380e-01 -1.35669589e+00 5.27935445e-01
-8.65985751e-01 -8.22671890e-01 7.16250956e-01 1.00957441e+00
3.93290818e-01 -2.21195459e-01 5.72828293e-01 -5.90912580e-01
1.32288945e+00 -7.57307768e-01 -5.57969809e-01 5.14269888e-01
-1.58189011e+00 3.16523105e-01 2.04688400e-01 -4.45931286e-01
-6.61807120e-01 -4.57313001e-01 -3.63026187e-02 -4.12218422e-01
6.09864175e-01 5.24241030e-01 3.31241488e-01 -1.95066795e-01
8.42152596e-01 1.37355953e-01 -1.13874376e-02 -6.50305986e-01
7.80853629e-01 9.03756559e-01 -8.25735647e-03 -4.23040092e-01
6.19766176e-01 -4.46219966e-02 -5.50716817e-01 -4.55278456e-01
-5.47398984e-01 -1.05594766e+00 3.13201904e-01 4.79998887e-01
3.02743644e-01 -1.02574480e+00 -8.09673607e-01 -3.06843162e-01
-8.37978959e-01 -1.95632681e-01 -4.08873618e-01 2.49735758e-01
-3.20018411e-01 2.66016215e-01 -7.29102373e-01 -6.32727385e-01
-1.01665032e+00 -1.06909359e+00 1.40052736e+00 2.38440543e-01
-2.70111740e-01 -9.48390901e-01 4.15801965e-02 4.76254165e-01
1.06181931e+00 -5.87846518e-01 1.05699480e+00 -1.03626144e+00
-8.08681190e-01 -5.51579952e-01 -5.38720012e-01 2.87276477e-01
8.86904076e-02 -4.28963661e-01 -8.75454187e-01 -4.47639018e-01
-3.87962073e-01 -2.35790983e-01 7.48792648e-01 1.58885494e-01
1.22990239e+00 -3.20393920e-01 -5.97349226e-01 4.38449502e-01
1.55557895e+00 2.90448695e-01 2.82812953e-01 1.77690059e-01
3.18694800e-01 2.29370847e-01 5.85532904e-01 4.27235931e-01
1.83781877e-01 1.06371522e+00 1.85786352e-01 1.36224240e-01
-4.51713465e-02 -6.66234612e-01 2.64889002e-01 5.59040308e-01
6.22748315e-01 -5.87625146e-01 -3.20192844e-01 7.24201322e-01
-1.86524558e+00 -6.82492316e-01 6.45081758e-01 2.16927576e+00
6.97971344e-01 7.58286566e-02 9.43153054e-02 -4.22900140e-01
3.92313451e-01 1.70806631e-01 -8.47139359e-01 -4.30635780e-01
3.97231370e-01 4.47023183e-01 8.84314835e-01 5.34557939e-01
-7.14981496e-01 9.37515557e-01 7.05292559e+00 1.32859349e+00
-9.35877562e-01 1.76349819e-01 4.10137981e-01 -3.30330044e-01
-8.16173911e-01 -9.79109704e-02 -1.03227031e+00 3.44165355e-01
9.91993845e-01 -5.93274593e-01 7.35955596e-01 1.43805993e+00
-2.18068197e-01 4.19715703e-01 -1.31715453e+00 1.29624152e+00
1.40043244e-01 -1.54056752e+00 1.57688782e-01 3.70655417e-01
5.64323366e-01 6.85566217e-02 6.63132310e-01 7.87281036e-01
3.74981701e-01 -8.62300515e-01 1.88402355e-01 1.99504867e-01
7.48564363e-01 -4.82351303e-01 5.53024471e-01 -1.15576155e-01
-1.17265081e+00 -3.76223266e-01 -1.61663756e-01 3.75285298e-01
1.30149871e-01 4.76219386e-01 -8.35372627e-01 3.32412183e-01
6.45571172e-01 5.84734559e-01 -5.76631665e-01 1.00110984e+00
3.05881739e-01 4.08604622e-01 -5.10233879e-01 -4.45463806e-01
3.10466588e-01 8.87810346e-03 2.99680561e-01 1.11008036e+00
4.28248286e-01 -2.94644922e-01 2.17640176e-02 7.64837921e-01
-4.10631537e-01 2.52002060e-01 -7.37793267e-01 -5.05705118e-01
6.82522714e-01 1.29485488e+00 -9.25469548e-02 -2.03261197e-01
-4.67222422e-01 1.30045021e+00 8.29062611e-02 7.16574907e-01
-8.45283449e-01 -6.57404006e-01 9.64841843e-01 2.37383917e-01
6.67083323e-01 5.03151566e-02 9.21727344e-02 -1.04221654e+00
4.10464495e-01 -7.33111620e-01 4.17775005e-01 -6.41303301e-01
-1.67347133e+00 4.24866468e-01 -6.77885711e-02 -8.02097440e-01
-4.04150963e-01 -6.73145890e-01 -3.14691097e-01 1.00603926e+00
-1.61242688e+00 -1.00577962e+00 -3.25280167e-02 2.53782541e-01
7.93236434e-01 -3.26859117e-01 6.84300065e-01 8.12653482e-01
-1.74669504e-01 1.29661739e+00 3.88828784e-01 -3.58368516e-01
8.21741402e-01 -9.67993677e-01 5.70283949e-01 1.46649763e-01
2.15607479e-01 1.20047534e+00 3.38445723e-01 -3.76742095e-01
-1.95143366e+00 -8.71732593e-01 9.40471649e-01 -7.15017974e-01
9.82563257e-01 -4.53523517e-01 -4.34963077e-01 6.07450724e-01
2.78031319e-01 -1.65274650e-01 9.48488533e-01 6.75788581e-01
-6.53316379e-01 -4.57498282e-01 -1.06108141e+00 8.00601602e-01
1.14347088e+00 -8.60038042e-01 -1.69708893e-01 6.89032495e-01
1.41608310e+00 -2.74487305e-02 -9.85773504e-01 2.14426905e-01
8.69292855e-01 -4.38622832e-01 1.33264470e+00 -7.11457014e-01
1.16181284e-01 1.94474757e-01 -5.12534201e-01 -9.98498619e-01
-5.72872937e-01 -9.89972115e-01 -4.43715036e-01 8.09667349e-01
7.81662703e-01 -8.06393087e-01 1.04421782e+00 9.40213144e-01
1.09178767e-01 -1.22539496e+00 -3.63188624e-01 -8.89963746e-01
-1.77619621e-01 -3.59110028e-01 6.84696078e-01 4.41786587e-01
-8.63274783e-02 6.00668788e-01 -2.47023165e-01 -2.46057540e-01
3.63458902e-01 1.34219155e-01 5.77683210e-01 -9.73013222e-01
-6.55246317e-01 -5.76310813e-01 -1.63589656e-01 -1.93016410e+00
-3.07350338e-01 -9.49340701e-01 -4.53997344e-01 -1.32304788e+00
2.75740921e-01 -5.49667120e-01 -6.54975891e-01 2.31515676e-01
9.82013270e-02 2.60588601e-02 -2.74153426e-02 1.70848072e-01
-8.54386687e-01 5.76858878e-01 8.15972090e-01 -3.47299159e-01
-3.21779072e-01 -6.51738346e-02 -1.41237700e+00 -1.60308540e-01
5.17901301e-01 -2.20623121e-01 -9.82355118e-01 -7.30597258e-01
6.76926613e-01 -1.78463444e-01 1.70952529e-01 -3.37706059e-01
3.99958909e-01 7.82465264e-02 -1.76093020e-02 -5.30599713e-01
5.08473039e-01 -1.04764378e+00 -1.20520882e-01 8.98882225e-02
-6.88126028e-01 2.83925861e-01 1.25612557e-01 1.06616700e+00
-1.03034265e-01 -1.76527902e-01 -9.21464041e-02 1.15280703e-01
-6.73620641e-01 6.84121490e-01 1.18354611e-01 8.38673860e-02
6.19028449e-01 -1.00780137e-01 -2.59254873e-01 -8.23712289e-01
-5.35887599e-01 6.65641487e-01 1.13329567e-01 7.40377009e-01
4.95741189e-01 -1.54107261e+00 -1.20605722e-01 1.69350937e-01
5.16965151e-01 -5.73801458e-01 1.61510065e-01 4.53295916e-01
-1.70811877e-01 8.88793051e-01 4.11217749e-01 -2.02697679e-01
-9.17087555e-01 6.35734260e-01 1.50180962e-02 -4.87714559e-01
-2.13086829e-01 1.04263759e+00 5.99917315e-04 -2.13973552e-01
4.77187842e-01 -1.59408346e-01 1.95195496e-01 -1.16615348e-01
4.58507895e-01 4.28106159e-01 1.95694119e-01 2.00877517e-01
-7.87341446e-02 5.04117489e-01 -7.23597586e-01 -1.15015477e-01
1.12164569e+00 -2.82834679e-01 3.57192248e-01 3.06127947e-02
1.83482277e+00 1.21749438e-01 -8.23132157e-01 -6.09651029e-01
-7.75350034e-02 -5.97791791e-01 3.98984969e-01 -1.01961899e+00
-1.03807592e+00 5.03807485e-01 9.38997984e-01 4.61467803e-01
9.73120391e-01 2.72463620e-01 1.53954911e+00 7.81264544e-01
4.14965272e-01 -1.37740469e+00 4.09651399e-01 3.46675903e-01
7.37855911e-01 -1.33875358e+00 2.59611104e-03 -2.50890523e-01
-4.68995094e-01 7.63974249e-01 3.27771991e-01 -1.35929780e-02
9.33843017e-01 -2.53252655e-01 -1.89229488e-01 -5.79946637e-01
-9.78866816e-01 -1.56940401e-01 4.98202384e-01 3.29602242e-01
4.01465863e-01 -1.21867530e-01 -2.71914601e-01 6.76942706e-01
9.48376805e-02 -3.95477675e-02 -3.96511853e-01 1.05208087e+00
-3.68462831e-01 -1.39723408e+00 3.90977114e-01 8.06530893e-01
-3.82371485e-01 -5.79673886e-01 -2.70588368e-01 5.68042099e-01
-5.19443452e-01 9.58206058e-01 -7.54066706e-02 -7.98516631e-01
4.68492806e-01 2.96651959e-01 2.58230716e-01 -5.62787294e-01
-6.22746468e-01 1.18999995e-01 2.04558983e-01 -1.02258074e+00
3.37819755e-01 -3.01171839e-01 -6.09886050e-01 -3.14867169e-01
-7.78775811e-01 2.83792198e-01 8.26744974e-01 3.49425822e-01
1.25328588e+00 1.84297070e-01 8.19175303e-01 -3.01877558e-01
-1.28095269e+00 -8.21984947e-01 -6.41960382e-01 3.21716964e-01
1.63635969e-01 -5.55546284e-01 -5.58370173e-01 -3.48227620e-01] | [11.323907852172852, 7.402409553527832] |
194153e7-1e94-4e5c-abc4-4c02cc9081e0 | discourse-analysis-via-questions-and-answers | 2210.05905 | null | https://arxiv.org/abs/2210.05905v2 | https://arxiv.org/pdf/2210.05905v2.pdf | Discourse Analysis via Questions and Answers: Parsing Dependency Structures of Questions Under Discussion | Automatic discourse processing is bottlenecked by data: current discourse formalisms pose highly demanding annotation tasks involving large taxonomies of discourse relations, making them inaccessible to lay annotators. This work instead adopts the linguistic framework of Questions Under Discussion (QUD) for discourse analysis and seeks to derive QUD structures automatically. QUD views each sentence as an answer to a question triggered in prior context; thus, we characterize relationships between sentences as free-form questions, in contrast to exhaustive fine-grained taxonomies. We develop the first-of-its-kind QUD parser that derives a dependency structure of questions over full documents, trained using a large, crowdsourced question-answering dataset DCQA (Ko et al., 2022). Human evaluation results show that QUD dependency parsing is possible for language models trained with this crowdsourced, generalizable annotation scheme. We illustrate how our QUD structure is distinct from RST trees, and demonstrate the utility of QUD analysis in the context of document simplification. Our findings show that QUD parsing is an appealing alternative for automatic discourse processing. | ['Junyi Jessy Li', 'Greg Durrett', 'Dananjay Srinivas', 'Cutter Dalton', 'Yating Wu', 'Wei-Jen Ko'] | 2022-10-12 | null | null | null | null | ['dependency-parsing'] | ['natural-language-processing'] | [ 2.39160687e-01 1.20816422e+00 -7.51277879e-02 -4.25648093e-01
-1.33789337e+00 -1.03429627e+00 9.57122147e-01 4.94973689e-01
-2.94509828e-01 9.72243309e-01 1.05976331e+00 -6.94585323e-01
-4.68330691e-03 -6.88155115e-01 -4.47153389e-01 -1.02902181e-01
2.78213024e-01 5.73848486e-01 5.03801167e-01 -7.81101704e-01
1.79337621e-01 6.98507875e-02 -1.09848332e+00 7.69219995e-01
8.55088770e-01 3.18583071e-01 4.60069627e-02 8.31529260e-01
-3.66655707e-01 1.57865131e+00 -1.39556491e+00 -9.36150312e-01
-2.11620837e-01 -8.50364327e-01 -1.78808868e+00 7.39686862e-02
4.99806285e-01 -3.08173925e-01 -1.03791229e-01 7.94686079e-01
2.00849742e-01 1.13438554e-02 7.18487382e-01 -8.00059974e-01
-9.49982166e-01 8.22604418e-01 1.08526703e-02 4.05830532e-01
9.28021431e-01 -6.48926422e-02 1.48076534e+00 -5.24315357e-01
1.19175410e+00 1.57751000e+00 4.58614051e-01 8.67986739e-01
-1.16604424e+00 2.27855787e-01 1.17500342e-01 1.48377910e-01
-7.48148024e-01 -5.30682623e-01 5.48860013e-01 -6.47598386e-01
1.36014533e+00 7.08101511e-01 2.03975245e-01 9.04118538e-01
-1.98852867e-02 8.24556947e-01 9.17979956e-01 -7.97606289e-01
1.60310924e-01 -1.22743092e-01 5.68247437e-01 4.62971717e-01
1.52919531e-01 -6.15349710e-01 -5.34970284e-01 -2.55785048e-01
2.57139176e-01 -9.74134743e-01 -1.62356824e-01 4.24036711e-01
-9.52560902e-01 9.29179549e-01 -2.42531765e-02 5.49061537e-01
-1.90018281e-01 -1.55384436e-01 6.83134496e-01 5.56243002e-01
5.39928854e-01 1.00263715e+00 -1.78803146e-01 -2.59552449e-01
-4.21343237e-01 7.19457984e-01 1.22822475e+00 8.23547244e-01
3.67196858e-01 -3.09854537e-01 -5.05173802e-01 6.18083835e-01
8.17388445e-02 2.64876068e-01 2.76476771e-01 -1.76085424e+00
7.49356806e-01 7.28988528e-01 3.13583851e-01 -9.64509249e-01
-3.13688457e-01 4.50093597e-01 -2.00482458e-01 -3.74317765e-02
8.93254220e-01 -3.80474776e-01 -7.58904889e-02 1.65585029e+00
6.52976274e-01 -1.00930250e+00 3.12356114e-01 7.87520528e-01
1.19726837e+00 6.09578371e-01 2.69703478e-01 -4.46791321e-01
1.73598492e+00 -8.18792403e-01 -1.11898971e+00 -1.46608233e-01
1.14719605e+00 -7.78096437e-01 1.13943768e+00 1.62512675e-01
-1.36802685e+00 -1.42362833e-01 -8.12568188e-01 -9.26430225e-01
4.82984222e-02 -2.87822485e-01 5.12768090e-01 4.66602892e-01
-8.97223949e-01 3.62706214e-01 -6.84572756e-01 -2.17578113e-01
4.68901843e-01 -4.91301864e-02 -2.11803094e-01 3.05620253e-01
-1.17992389e+00 1.35813177e+00 3.82725954e-01 -1.19466513e-01
-4.52870578e-01 -4.90295678e-01 -1.11118174e+00 -2.25347430e-01
6.14141762e-01 -4.66797143e-01 1.96273363e+00 -1.04040623e+00
-1.58842754e+00 1.42587006e+00 -5.06911874e-01 -6.12639546e-01
2.99737573e-01 -4.05035973e-01 -9.07811373e-02 4.23480541e-01
4.34503168e-01 2.63705313e-01 4.90324378e-01 -9.20015752e-01
-4.72047597e-01 -3.37625027e-01 1.04939556e+00 2.99933583e-01
5.95583022e-03 7.04343259e-01 2.84620047e-01 -4.19018090e-01
-1.50321618e-01 -4.98278320e-01 9.44778845e-02 -1.45414472e-01
-5.13058245e-01 -8.89475465e-01 5.08827150e-01 -1.00632155e+00
1.57759595e+00 -1.66393554e+00 2.49138728e-01 -5.42875886e-01
5.15434921e-01 5.33473313e-01 7.97395036e-02 6.77987337e-01
1.05051957e-01 4.45252895e-01 -3.42197150e-01 -2.07595959e-01
3.73748004e-01 4.74109828e-01 -4.60222781e-01 2.43492976e-01
6.60833120e-01 1.20185590e+00 -8.64901006e-01 -8.47573221e-01
-9.61418077e-02 -6.68956414e-02 -5.76127112e-01 4.82328802e-01
-8.02060366e-01 2.93979168e-01 -6.15900338e-01 2.88722426e-01
1.79439694e-01 -2.86346585e-01 5.56501210e-01 1.14086531e-01
-1.95570007e-01 9.58558142e-01 -6.41059816e-01 1.53985453e+00
-2.57863909e-01 9.99077380e-01 3.86906624e-01 -8.10118496e-01
6.66313529e-01 4.98432398e-01 -2.46248245e-01 -5.19582808e-01
2.90837914e-01 2.41985321e-01 2.27051139e-01 -1.15245581e+00
7.28303075e-01 -2.93824434e-01 -5.38091183e-01 7.83330858e-01
9.59819928e-02 -5.04178286e-01 6.67623699e-01 4.29272532e-01
1.09976995e+00 3.10536444e-01 6.82408690e-01 -4.58764642e-01
6.96082890e-01 6.05508566e-01 3.28243941e-01 5.16705155e-01
-2.36780450e-01 4.60037500e-01 1.05030894e+00 -3.13505381e-01
-1.10765982e+00 -8.03795457e-01 -2.96897352e-01 1.23825502e+00
-1.19866036e-01 -6.11744046e-01 -1.25478709e+00 -7.18923628e-01
-3.08967382e-01 1.09052587e+00 -5.39762020e-01 7.00188518e-01
-1.44254804e+00 -3.62763137e-01 8.81045461e-01 2.37106323e-01
8.69788602e-02 -1.16657412e+00 -9.93896425e-01 3.74966174e-01
-4.97012615e-01 -1.18894374e+00 -5.76691516e-02 -2.71639198e-01
-3.40712249e-01 -1.63449204e+00 -3.66352022e-01 -8.33526909e-01
2.21845374e-01 -1.41613394e-01 1.36421108e+00 1.86996043e-01
1.71562910e-01 3.57443005e-01 -5.37824094e-01 -6.31625891e-01
-1.12675822e+00 6.83638230e-02 -5.45319319e-01 -5.83057106e-01
5.31433642e-01 -3.01132053e-01 -1.72266334e-01 -2.15192810e-01
-7.98796117e-01 3.48824938e-03 -1.63432047e-01 6.42334640e-01
1.99792713e-01 -8.49247634e-01 7.65050948e-01 -1.41515112e+00
1.20601034e+00 -4.04481322e-01 -3.93536329e-01 4.82534558e-01
3.79891098e-02 1.40687853e-01 6.30765438e-01 -1.60751883e-02
-1.60029745e+00 -6.37211740e-01 -5.01953185e-01 6.50645435e-01
-1.99466228e-01 4.57685679e-01 -2.65529782e-01 5.58378518e-01
1.23558033e+00 -4.91229832e-01 6.49356171e-02 -3.58834743e-01
6.87580466e-01 6.40584767e-01 5.92825890e-01 -9.21577752e-01
5.67181587e-01 3.38839650e-01 -1.29742056e-01 -1.10191309e+00
-1.44112539e+00 -1.93096772e-01 -7.13870525e-01 -2.92188860e-02
1.28289938e+00 -7.01410234e-01 -8.01928401e-01 -2.31925249e-01
-1.93551970e+00 -4.35612172e-01 -7.39013553e-01 1.68249160e-02
-5.56810856e-01 6.63300931e-01 -9.62107837e-01 -8.96014750e-01
-3.68291229e-01 -6.41697586e-01 1.09881914e+00 8.07518885e-02
-1.12969625e+00 -1.14260352e+00 2.42524967e-01 6.85346782e-01
8.04214627e-02 7.62390554e-01 1.20836651e+00 -1.00810969e+00
-2.52666384e-01 1.82793528e-01 -1.27796695e-01 3.16249609e-01
-7.75743276e-02 -2.13593170e-02 -9.67817187e-01 3.34144264e-01
3.78865093e-01 -8.24031532e-01 2.49256790e-01 -1.70826297e-02
6.33634865e-01 -8.29490304e-01 1.93338484e-01 -1.45554245e-01
9.00650680e-01 -1.58518061e-01 4.47433680e-01 3.83278489e-01
4.12863255e-01 1.17240512e+00 5.50191462e-01 2.79935226e-02
8.59930634e-01 3.34699988e-01 1.08858943e-01 3.14957976e-01
-4.53372151e-01 -1.37348324e-01 1.47980317e-01 1.12419164e+00
-1.54811786e-02 -3.44554096e-01 -1.06599844e+00 6.92835033e-01
-1.89863217e+00 -9.88280714e-01 -8.76560271e-01 1.42245388e+00
1.35485327e+00 3.11200023e-02 4.75745276e-03 -2.08088104e-02
6.19324923e-01 5.42004108e-01 1.14055954e-01 -8.29606473e-01
-3.80319357e-01 4.25069064e-01 -3.15080732e-01 1.13603401e+00
-9.17935252e-01 8.34335983e-01 6.09622717e+00 5.32260716e-01
-3.84713948e-01 4.75136995e-01 2.56476045e-01 2.73277968e-01
-5.56249440e-01 2.13869691e-01 -5.38132906e-01 -1.86816473e-02
1.06175268e+00 -4.00718957e-01 2.02436675e-03 4.94294256e-01
7.35186487e-02 -3.85920256e-01 -1.42350543e+00 3.46167922e-01
2.85980225e-01 -1.69796515e+00 4.47578989e-02 -2.84614265e-01
5.25320470e-01 -3.50287795e-01 -6.86474681e-01 1.37668356e-01
4.73787695e-01 -1.16121709e+00 9.54255104e-01 1.77537441e-01
5.67914546e-01 -1.57120869e-01 7.15553164e-01 7.02349246e-01
-5.44322133e-01 7.73081034e-02 -1.39204904e-01 -7.58899271e-01
5.29553592e-01 3.99320163e-02 -8.96267891e-01 5.59299111e-01
1.52012482e-01 8.20269138e-02 -5.76632380e-01 1.23426609e-01
-7.93479443e-01 7.75449634e-01 -1.17958218e-01 -5.42165458e-01
6.52024671e-02 4.25667986e-02 7.57735133e-01 1.43259752e+00
-4.00971949e-01 7.92856395e-01 8.49750936e-02 6.50887430e-01
-2.06890598e-01 1.44426763e-01 -4.20271069e-01 7.17336237e-02
6.29634917e-01 8.97452950e-01 -2.26321697e-01 -4.18059289e-01
-3.02623183e-01 4.54581261e-01 6.33492827e-01 1.22227222e-01
-3.12491834e-01 -3.47010195e-01 1.93132073e-01 1.44909665e-01
-5.04053570e-02 -1.57814637e-01 -3.61588538e-01 -1.14922345e+00
3.23563159e-01 -1.20422637e+00 4.05042946e-01 -6.61081731e-01
-1.29400778e+00 5.78544796e-01 2.64909297e-01 -6.92633927e-01
-6.80858195e-01 -6.96187794e-01 -5.95663428e-01 9.57942307e-01
-1.50982451e+00 -1.21775126e+00 1.99814904e-02 1.86629370e-01
7.86826432e-01 4.60138977e-01 1.22059357e+00 -1.47816449e-01
-3.38141322e-01 1.63342997e-01 -3.10098439e-01 5.32902300e-01
4.80888784e-01 -1.41548193e+00 4.36618149e-01 8.41113567e-01
-2.20528562e-02 5.91693461e-01 9.59168792e-01 -4.15034086e-01
-1.21255958e+00 -7.48361290e-01 1.61969388e+00 -1.20551181e+00
1.02713811e+00 -2.96324998e-01 -1.27429390e+00 8.18125546e-01
8.52043152e-01 -4.95774806e-01 8.46757531e-01 2.13764742e-01
-3.23612183e-01 4.28047776e-01 -9.20166850e-01 5.57198465e-01
1.02759409e+00 -8.58091950e-01 -1.68793857e+00 8.41922045e-01
1.03867519e+00 -7.59142637e-01 -9.61884379e-01 -4.08107378e-02
8.43419433e-02 -8.10140014e-01 4.02736098e-01 -1.03522098e+00
9.02101517e-01 -1.61035419e-01 -2.04107761e-01 -6.46320224e-01
2.24551931e-01 -1.03837967e+00 -2.70604730e-01 1.43290484e+00
6.11283720e-01 -2.57759303e-01 5.33410549e-01 1.10096121e+00
-3.68064791e-01 -4.21550721e-01 -1.11321545e+00 -4.80094761e-01
7.58812964e-01 -4.06875551e-01 3.69108260e-01 9.08322632e-01
6.69442117e-01 1.08323336e+00 4.14729834e-01 -2.52823800e-01
2.80274689e-01 1.51877478e-01 8.89486074e-01 -1.25163102e+00
-3.58112156e-01 -2.45248571e-01 9.42017063e-02 -1.08616257e+00
6.23296022e-01 -8.35661650e-01 1.57544866e-01 -1.80603015e+00
-2.56522179e-01 -4.85354923e-02 8.79902363e-01 9.50005129e-02
-2.15662032e-01 -1.25599682e-01 4.10858095e-01 3.47932398e-01
-9.07806933e-01 3.69864255e-01 1.39844334e+00 -2.08200398e-03
-1.16003759e-01 -3.60260755e-01 -8.31096411e-01 1.28050733e+00
5.44933677e-01 -4.29720908e-01 -2.88910419e-01 -9.14207995e-01
4.97208327e-01 5.00328183e-01 3.19236547e-01 -1.30231813e-01
3.82353783e-01 -2.72789747e-01 -2.82534719e-01 -2.50027955e-01
1.63492605e-01 -1.84530154e-01 -3.63812953e-01 2.00307131e-01
-7.71143973e-01 7.90321231e-02 3.51545848e-02 2.07864016e-01
-3.84659499e-01 -8.00891399e-01 2.50630677e-01 -3.88267875e-01
-3.74057710e-01 -6.51098311e-01 -6.32736206e-01 1.02710164e+00
8.73651206e-01 -1.17714249e-01 -9.53421831e-01 -3.75703007e-01
-6.93034351e-01 2.78379709e-01 2.66836107e-01 -6.04769439e-02
1.04390122e-01 -8.69728684e-01 -1.01525950e+00 -6.55039310e-01
-5.77280819e-02 5.57419538e-01 -7.22791776e-02 3.60906899e-01
-7.89232254e-01 4.85669017e-01 1.49539173e-01 -2.81150073e-01
-1.15827930e+00 3.16150367e-01 1.79892704e-01 -5.21043539e-01
-4.60526258e-01 8.70020926e-01 1.16465077e-01 -5.19918919e-01
-1.00225873e-01 -3.85633796e-01 -6.01127565e-01 3.34182471e-01
6.72125459e-01 2.86106467e-01 3.53304520e-02 -7.85866857e-01
-2.44452521e-01 1.74084865e-02 1.14168867e-01 -3.35815161e-01
1.19255340e+00 -4.92848635e-01 -5.39607644e-01 5.71810305e-01
9.01357353e-01 3.28225344e-01 -8.88372004e-01 -1.97981268e-01
6.57528758e-01 -2.21905839e-02 -7.68834949e-01 -6.67260706e-01
3.21007997e-01 8.12651873e-01 -5.81089437e-01 8.95579994e-01
6.50017679e-01 5.04619002e-01 7.40053296e-01 7.89867103e-01
6.77945912e-02 -1.18870103e+00 2.25899499e-02 8.75047803e-01
1.29533720e+00 -1.02102244e+00 1.26344189e-01 -7.54865587e-01
-6.44184887e-01 1.29817665e+00 5.86807907e-01 -2.37805083e-01
1.99042007e-01 1.90760761e-01 2.18860626e-01 -5.58700562e-01
-8.08422148e-01 -1.10835671e-01 3.06789074e-02 8.07590723e-01
7.71325707e-01 3.81298773e-02 -8.20122480e-01 6.58364952e-01
-6.88343167e-01 -3.39885294e-01 9.92768943e-01 1.11747730e+00
-5.12278855e-01 -1.13559437e+00 -4.09368217e-01 1.20654300e-01
-8.12981009e-01 -4.29500304e-02 -8.90310764e-01 1.08870077e+00
-1.81633070e-01 1.34906757e+00 1.34420007e-01 3.75063121e-01
5.95450819e-01 9.98880640e-02 6.22965455e-01 -1.12052464e+00
-1.01328433e+00 -4.25107718e-01 1.19295835e+00 -2.28828937e-01
-9.39711094e-01 -7.05784738e-01 -1.47598958e+00 -2.77663857e-01
-2.86019713e-01 4.01445627e-01 2.29807451e-01 1.47352076e+00
2.20817819e-01 4.17356014e-01 6.13316409e-02 -4.01990831e-01
-8.48735988e-01 -9.60069239e-01 5.41851409e-02 5.80833375e-01
5.61284304e-01 -7.00842664e-02 -5.60822561e-02 4.08429444e-01] | [10.80997371673584, 9.395743370056152] |
c25ce6f2-dbc0-4ab9-9ad7-e3b74e87a1b4 | beyond-cross-view-image-retrieval-highly | 2204.04752 | null | https://arxiv.org/abs/2204.04752v2 | https://arxiv.org/pdf/2204.04752v2.pdf | Beyond Cross-view Image Retrieval: Highly Accurate Vehicle Localization Using Satellite Image | This paper addresses the problem of vehicle-mounted camera localization by matching a ground-level image with an overhead-view satellite map. Existing methods often treat this problem as cross-view image retrieval, and use learned deep features to match the ground-level query image to a partition (eg, a small patch) of the satellite map. By these methods, the localization accuracy is limited by the partitioning density of the satellite map (often in the order of tens meters). Departing from the conventional wisdom of image retrieval, this paper presents a novel solution that can achieve highly-accurate localization. The key idea is to formulate the task as pose estimation and solve it by neural-net based optimization. Specifically, we design a two-branch {CNN} to extract robust features from the ground and satellite images, respectively. To bridge the vast cross-view domain gap, we resort to a Geometry Projection module that projects features from the satellite map to the ground-view, based on a relative camera pose. Aiming to minimize the differences between the projected features and the observed features, we employ a differentiable Levenberg-Marquardt ({LM}) module to search for the optimal camera pose iteratively. The entire pipeline is differentiable and runs end-to-end. Extensive experiments on standard autonomous vehicle localization datasets have confirmed the superiority of the proposed method. Notably, e.g., starting from a coarse estimate of camera location within a wide region of 40m x 40m, with an 80% likelihood our method quickly reduces the lateral location error to be within 5m on a new KITTI cross-view dataset. | ['Hongdong Li', 'Yujiao Shi'] | 2022-04-10 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Shi_Beyond_Cross-View_Image_Retrieval_Highly_Accurate_Vehicle_Localization_Using_Satellite_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Shi_Beyond_Cross-View_Image_Retrieval_Highly_Accurate_Vehicle_Localization_Using_Satellite_CVPR_2022_paper.pdf | cvpr-2022-1 | ['camera-localization'] | ['computer-vision'] | [-3.92291732e-02 -1.27780885e-01 5.36368378e-02 -5.13750255e-01
-1.26609278e+00 -9.07583654e-01 5.77448666e-01 -2.25149885e-01
-6.96065545e-01 2.55491942e-01 -3.69967699e-01 -1.10838838e-01
-8.18270147e-02 -7.74383128e-01 -1.06244624e+00 -6.38758719e-01
9.41803604e-02 5.21521747e-01 1.29288256e-01 -7.46615902e-02
4.45094079e-01 7.51396596e-01 -1.40833437e+00 -5.36561966e-01
7.37976551e-01 1.15606189e+00 4.42945004e-01 4.38764691e-01
3.08282137e-01 1.11084007e-01 -2.47603387e-01 -2.45164350e-01
5.35516918e-01 1.92284927e-01 -5.84418297e-01 4.91236746e-01
8.98596823e-01 -5.21854818e-01 -3.11634243e-01 1.15783668e+00
2.47974545e-01 1.16995148e-01 3.77024442e-01 -1.13164425e+00
-1.13277964e-01 -1.60743013e-01 -6.58468306e-01 1.04839429e-01
3.65591347e-01 5.84500842e-02 8.85182917e-01 -1.16219127e+00
5.34645438e-01 8.12745333e-01 9.67199981e-01 -8.87773111e-02
-1.18196404e+00 -5.08500338e-01 2.49707639e-01 -1.06928639e-01
-2.00910544e+00 -3.10690701e-01 6.58123791e-01 -5.99572301e-01
7.93002903e-01 1.53392525e-02 5.51033258e-01 4.06889558e-01
1.18779965e-01 2.03617454e-01 7.47101247e-01 -7.56486729e-02
1.28356010e-01 1.89380214e-01 -2.78861731e-01 7.70612240e-01
1.21207073e-01 1.64029244e-02 -3.59218746e-01 -1.30092844e-01
6.15708828e-01 1.96164906e-01 -3.92767042e-01 -7.97938347e-01
-1.20475769e+00 9.61208284e-01 6.92739785e-01 -4.69610747e-03
-2.96716928e-01 2.16894254e-01 -1.11479443e-02 1.32358139e-02
5.10289967e-01 2.91762143e-01 -2.27690056e-01 2.50131041e-01
-1.11222398e+00 2.99193382e-01 4.80257660e-01 9.97450531e-01
1.48462117e+00 -1.93744436e-01 5.46606302e-01 4.34004098e-01
4.57205623e-01 9.00131583e-01 6.75666258e-02 -8.27365279e-01
7.76581228e-01 5.32845080e-01 4.47639853e-01 -1.42158806e+00
-4.38158333e-01 -4.32597429e-01 -6.34901404e-01 5.66100366e-02
3.38660419e-01 -1.04203351e-01 -5.39282024e-01 1.38688254e+00
5.74843109e-01 4.22594696e-02 -1.88765898e-02 1.28668380e+00
3.00595969e-01 6.47315323e-01 -4.46643054e-01 2.40721151e-01
1.16520119e+00 -6.93362832e-01 -8.80003497e-02 -7.68029749e-01
6.32034719e-01 -7.16624379e-01 5.88574946e-01 -4.77500670e-02
-6.66622102e-01 -4.46684659e-01 -1.18052804e+00 4.36489806e-02
-3.76198590e-01 5.91610968e-01 1.16584919e-01 2.60057658e-01
-1.22629678e+00 4.39756006e-01 -8.54616702e-01 -5.50861716e-01
-5.15459366e-02 4.76262212e-01 -7.49661982e-01 -8.00000206e-02
-9.63002563e-01 9.73299861e-01 4.79665816e-01 5.48038721e-01
-8.75389993e-01 -5.70968091e-01 -1.09578645e+00 3.91885377e-02
4.01582420e-01 -6.64287508e-01 8.44414830e-01 -8.78149509e-01
-1.32908750e+00 1.00499988e+00 -1.41833380e-01 -2.71046340e-01
5.34927607e-01 -2.14723766e-01 -1.20266911e-03 2.01103956e-01
4.89553124e-01 7.38201976e-01 8.02170336e-01 -1.29380715e+00
-9.10471380e-01 -6.72010183e-01 4.57824133e-02 3.60297859e-01
-3.20808813e-02 -2.75867134e-01 -9.16009426e-01 -1.55294240e-01
7.13264823e-01 -1.35012221e+00 -2.62818128e-01 2.36193538e-02
-1.26641139e-01 1.71514377e-01 6.15661740e-01 -5.08906722e-01
7.00847685e-01 -2.24954939e+00 1.08188182e-01 3.64220738e-01
1.29237156e-02 -2.51045614e-01 1.94205400e-02 5.09075105e-01
5.69076017e-02 -1.96407214e-01 -2.00741678e-01 -3.84584874e-01
-6.38950989e-02 -3.40044051e-02 -3.09020489e-01 1.19375741e+00
1.43668264e-01 8.20849061e-01 -7.88019717e-01 -1.97043225e-01
3.81006032e-01 4.93078470e-01 -3.59193116e-01 2.70218730e-01
7.27287307e-03 4.34577078e-01 -5.25790811e-01 5.74200988e-01
1.11311865e+00 -1.75203159e-01 -6.19047545e-02 -2.83967555e-01
-5.04761159e-01 4.10812814e-03 -1.19475818e+00 1.87755823e+00
-5.74002504e-01 7.49729156e-01 2.97707051e-01 -8.22969496e-01
1.14181161e+00 -1.51080400e-01 3.85536402e-01 -5.13955951e-01
-2.71065772e-04 3.79292339e-01 -6.45656466e-01 -3.38103205e-01
7.08683789e-01 -1.05724800e-02 -1.60247475e-01 -4.45139632e-02
-1.45885646e-01 -3.21374089e-01 -1.40457451e-01 -8.84756520e-02
6.69775248e-01 1.45665631e-01 1.42755315e-01 -1.63671568e-01
7.38281429e-01 3.78689826e-01 4.43027258e-01 5.07243812e-01
1.15684770e-01 7.98809350e-01 3.06734204e-01 -6.39125407e-01
-9.75133121e-01 -8.28483999e-01 -1.21702693e-01 6.35635972e-01
6.36238098e-01 -1.86547115e-01 -7.46797860e-01 -6.42042875e-01
1.55172735e-01 1.58859715e-01 -5.02833843e-01 1.38245046e-01
-6.54018641e-01 -4.88011479e-01 3.01939756e-01 4.68698829e-01
6.42574906e-01 -2.90282696e-01 -7.97626972e-01 7.20160129e-03
-2.57586211e-01 -1.37621307e+00 -6.15794003e-01 7.45215341e-02
-7.95654178e-01 -9.22319531e-01 -6.10179186e-01 -7.54320800e-01
9.13254201e-01 7.58241057e-01 8.74250114e-01 -9.78849903e-02
1.34175271e-01 3.01512271e-01 -1.04351304e-01 3.17518860e-01
1.79238096e-01 2.53517210e-01 1.84489086e-01 2.81134188e-01
2.63489097e-01 -4.45974588e-01 -6.62761509e-01 5.61417043e-01
-6.35434449e-01 -2.05967307e-01 7.16879427e-01 6.43823683e-01
8.14240694e-01 -3.54718082e-02 -1.93530455e-01 -1.19608179e-01
-1.52938113e-01 -5.87237716e-01 -1.45728409e+00 -5.05675077e-02
-4.59080338e-01 -1.19647659e-01 5.29350579e-01 -1.10794410e-01
-4.06344265e-01 7.82329559e-01 1.18235782e-01 -7.17147052e-01
-4.71442938e-02 8.08482945e-01 -1.46506131e-01 -5.15305042e-01
2.82633305e-01 6.08531415e-01 1.52706563e-01 -2.76283473e-01
3.45453858e-01 6.59678221e-01 6.96640790e-01 -1.33712530e-01
1.18399978e+00 8.21582198e-01 2.65862718e-02 -7.24615335e-01
-6.82304025e-01 -6.90964282e-01 -9.56311405e-01 -2.88254440e-01
9.41063344e-01 -1.36117578e+00 -9.20377195e-01 1.93192035e-01
-1.08181787e+00 -5.85099980e-02 4.00853515e-01 6.44183934e-01
-6.50930762e-01 3.85483116e-01 -8.91714618e-02 -5.77090025e-01
-1.47568002e-01 -1.26301610e+00 1.66294301e+00 2.27972835e-01
2.17856169e-01 -8.90275121e-01 1.27032429e-01 3.39651972e-01
2.23202616e-01 2.47228011e-01 2.51680225e-01 -2.39810646e-01
-1.06055057e+00 -6.55399084e-01 -3.30851078e-01 -1.62608735e-03
-1.51142478e-01 -2.72828221e-01 -8.57591927e-01 -6.98767543e-01
7.84283411e-03 -4.46036644e-02 7.70809591e-01 3.56369734e-01
5.27803719e-01 5.96964248e-02 -5.65700054e-01 1.07493472e+00
1.85559833e+00 -5.81426695e-02 3.40998828e-01 8.21756244e-01
5.88972628e-01 4.82593983e-01 8.85438144e-01 3.48221421e-01
6.84987903e-01 1.01353860e+00 8.71933222e-01 9.61483642e-03
4.87024784e-01 -3.77249390e-01 3.81642818e-01 4.15544897e-01
3.20091903e-01 -2.33312603e-03 -9.76130068e-01 6.66538477e-01
-1.78493094e+00 -6.00965619e-01 4.85895649e-02 2.38690400e+00
6.96861818e-02 8.83340370e-03 -1.74506009e-01 -5.55337489e-01
6.52157366e-01 2.99527794e-01 -4.60025072e-01 3.49426329e-01
1.09921440e-01 -7.73186862e-01 1.18559515e+00 7.59615958e-01
-1.43051696e+00 1.02981913e+00 5.30679846e+00 4.29235846e-01
-1.49783015e+00 -1.39073730e-01 1.93469495e-01 -3.14440508e-03
-1.49606364e-02 2.53312588e-01 -1.15182030e+00 3.83916408e-01
7.09582746e-01 1.70713812e-01 3.88943434e-01 1.04528451e+00
2.65896767e-01 -3.86623055e-01 -1.06054866e+00 1.20832837e+00
3.39877218e-01 -1.39554095e+00 -2.93715745e-01 3.84187162e-01
5.74787617e-01 6.67102337e-01 2.54002064e-02 -4.36138771e-02
-1.81127712e-01 -6.55108213e-01 1.04420340e+00 4.32283223e-01
8.25909913e-01 -7.56627977e-01 9.34331059e-01 5.85300744e-01
-1.45240915e+00 -6.56811669e-02 -4.68838841e-01 1.68318510e-01
3.04004222e-01 2.56811112e-01 -7.90502369e-01 6.61623776e-01
8.49037707e-01 7.67136276e-01 -5.71864903e-01 9.71316099e-01
-2.48369225e-03 -6.47636689e-03 -5.93984902e-01 3.17511231e-01
6.80654585e-01 -6.11916006e-01 5.58075786e-01 8.39510500e-01
5.58128178e-01 -3.30241323e-02 4.06414330e-01 9.45461214e-01
7.63715804e-02 -1.39958948e-01 -8.96235824e-01 3.55740666e-01
4.91236389e-01 1.45664406e+00 -6.97882950e-01 -1.04840897e-01
-4.95787531e-01 8.52786601e-01 3.66695851e-01 3.07809025e-01
-8.34209621e-01 -3.30627054e-01 6.13084316e-01 2.86175549e-01
7.47678757e-01 -4.53973025e-01 1.27805173e-01 -1.15906727e+00
4.81354088e-01 -5.58106363e-01 -1.90823883e-01 -8.18103135e-01
-7.79296279e-01 7.16044009e-01 -9.77411345e-02 -1.74736476e+00
-2.63939857e-01 -3.92687470e-01 -3.68948579e-01 1.10313284e+00
-1.61256313e+00 -1.24795437e+00 -6.56028390e-01 2.50396371e-01
3.56535703e-01 -1.76533423e-02 3.13341320e-01 3.94023240e-01
-4.28757548e-01 2.75531769e-01 5.11780262e-01 2.67707556e-01
5.35806537e-01 -9.91023779e-01 5.01296878e-01 8.13201785e-01
1.61803812e-02 5.56609571e-01 5.43427885e-01 -4.54414427e-01
-1.71909368e+00 -1.31505358e+00 9.25760984e-01 -6.24975264e-01
7.65784025e-01 -4.73799199e-01 -7.99367309e-01 8.84954751e-01
-9.02009457e-02 3.28491569e-01 -2.63209660e-02 -2.87597567e-01
-1.41375035e-01 -4.63660479e-01 -8.91367912e-01 2.35014766e-01
5.03118992e-01 -7.13103712e-01 -3.02650869e-01 3.09557736e-01
3.46929640e-01 -7.40379989e-01 -8.46539378e-01 2.01748922e-01
4.28578436e-01 -9.04056311e-01 8.87878418e-01 -1.80586688e-02
2.00535194e-03 -8.24806392e-01 -4.74354059e-01 -1.31956995e+00
-1.60603717e-01 -4.18350071e-01 5.35441160e-01 1.06884074e+00
2.60856241e-01 -6.86388195e-01 9.43472803e-01 3.76597494e-01
-1.43060580e-01 -5.66574514e-01 -1.13391578e+00 -7.99145877e-01
-1.66691303e-01 -2.29672924e-01 3.84422272e-01 5.86829603e-01
-4.25616592e-01 2.29829460e-01 -2.59154707e-01 1.06989658e+00
7.25822508e-01 3.09788913e-01 1.11198878e+00 -1.11272228e+00
4.76475507e-02 -2.24422559e-01 -6.54981613e-01 -1.42122781e+00
4.29252177e-01 -7.23814487e-01 4.47763205e-01 -1.26556635e+00
7.54989982e-02 -2.98874944e-01 2.07246542e-01 -1.27145881e-02
5.40151112e-02 3.29025209e-01 2.25964844e-01 4.25693184e-01
-5.65147400e-01 5.88135958e-01 6.16990030e-01 -1.78332984e-01
-3.29509913e-03 9.76929441e-03 -1.98140264e-01 7.78751194e-01
4.43111748e-01 -6.43196881e-01 -8.68682414e-02 -8.42095494e-01
4.82855976e-01 2.96401262e-01 7.08267093e-01 -1.06105864e+00
5.37746131e-01 2.52936259e-02 2.88265437e-01 -9.96330678e-01
5.68753123e-01 -1.05248427e+00 2.80877948e-01 2.37064987e-01
5.84620684e-02 2.96024114e-01 -4.83300909e-02 7.14994669e-01
-4.69013572e-01 -2.34343126e-01 7.10597456e-01 -2.99170483e-02
-8.70726585e-01 3.16552162e-01 -2.13207349e-01 -2.30577841e-01
9.72215295e-01 -4.56653208e-01 -1.42242927e-02 -3.42037529e-01
-4.75833654e-01 4.49543178e-01 9.20457482e-01 2.15979949e-01
4.85146403e-01 -1.24590027e+00 -5.58951616e-01 3.21139425e-01
4.67606008e-01 1.97662875e-01 -8.69066641e-03 1.24476731e+00
-8.65714192e-01 7.30065286e-01 1.62668005e-01 -1.12072790e+00
-9.88851905e-01 4.87895608e-01 7.39403248e-01 1.86626092e-01
-6.36465788e-01 6.04144216e-01 1.86331227e-01 -6.56619072e-01
1.65987387e-03 -2.68136412e-01 5.03403693e-02 1.19262293e-01
4.80860889e-01 2.54110307e-01 1.69508651e-01 -1.24733329e+00
-7.02419877e-01 1.31004012e+00 1.20696150e-01 -1.22973055e-01
1.33049011e+00 -5.95330179e-01 -9.29178894e-02 1.07118696e-01
1.64498127e+00 -5.23169897e-02 -1.67555106e+00 -2.02096939e-01
7.78884217e-02 -7.11720467e-01 2.79353261e-01 -3.16475146e-02
-1.14738512e+00 7.80587494e-01 8.48732889e-01 2.52853930e-02
7.77958214e-01 1.65492833e-01 6.03187323e-01 4.90538806e-01
4.66804475e-01 -8.74477804e-01 -2.26923704e-01 5.30065715e-01
6.98843181e-01 -1.60626578e+00 8.63016248e-02 -1.84263155e-01
-4.47601140e-01 1.19129562e+00 5.53642154e-01 -3.09739679e-01
3.67914081e-01 -1.20827675e-01 2.72649348e-01 -2.51160026e-01
-2.72764325e-01 -2.48331606e-01 2.01975524e-01 3.27624500e-01
-7.49329105e-02 -1.44445121e-01 2.99218595e-01 1.49191516e-02
-9.13131237e-02 -2.56855965e-01 3.32096994e-01 7.56592453e-01
-5.97647429e-01 -4.50742990e-01 -7.24468827e-01 -9.32934210e-02
-2.91077197e-02 6.58859536e-02 -2.01186314e-01 8.73694718e-01
1.10860713e-01 7.55922019e-01 4.11532253e-01 -4.75140184e-01
4.72152382e-01 -2.70731956e-01 6.27969205e-02 -2.56214350e-01
-1.49901435e-01 1.63193822e-01 -3.48233461e-01 -8.10103178e-01
-3.45799834e-01 -7.37119615e-01 -1.02899861e+00 -1.46916837e-01
-4.19939846e-01 1.20359167e-01 8.95568848e-01 1.11605322e+00
3.82443577e-01 -2.88847357e-01 9.59447324e-01 -1.40250909e+00
-6.92108214e-01 -6.43444896e-01 -5.48083127e-01 -5.05873151e-02
6.55572891e-01 -6.58798575e-01 -4.96065497e-01 -1.10353604e-01] | [7.709802627563477, -2.1087300777435303] |
e47f502a-dc79-4464-8473-ccf3b3c5ff25 | resmem-learn-what-you-can-and-memorize-the | 2302.01576 | null | https://arxiv.org/abs/2302.01576v1 | https://arxiv.org/pdf/2302.01576v1.pdf | ResMem: Learn what you can and memorize the rest | The impressive generalization performance of modern neural networks is attributed in part to their ability to implicitly memorize complex training patterns. Inspired by this, we explore a novel mechanism to improve model generalization via explicit memorization. Specifically, we propose the residual-memorization (ResMem) algorithm, a new method that augments an existing prediction model (e.g. a neural network) by fitting the model's residuals with a $k$-nearest neighbor based regressor. The final prediction is then the sum of the original model and the fitted residual regressor. By construction, ResMem can explicitly memorize the training labels. Empirically, we show that ResMem consistently improves the test set generalization of the original prediction model across various standard vision and natural language processing benchmarks. Theoretically, we formulate a stylized linear regression problem and rigorously show that ResMem results in a more favorable test risk over the base predictor. | ['Sanjiv Kumar', 'Aditya Krishna Menon', 'Manzil Zaheer', 'Ankit Singh Rawat', 'Zonglin Li', 'Vaishnavh Nagarajan', 'Michal Lukasik', 'Zitong Yang'] | 2023-02-03 | null | null | null | null | ['memorization'] | ['natural-language-processing'] | [ 3.77715111e-01 9.69486013e-02 -2.04448909e-01 -6.38875127e-01
-2.44610101e-01 -1.42152742e-01 5.79460919e-01 4.51330133e-02
-5.60899436e-01 1.03158092e+00 1.39991820e-01 -3.28659326e-01
1.96659993e-02 -9.30945575e-01 -1.02591801e+00 -6.00089431e-01
2.01398104e-01 1.78633511e-01 -3.17848437e-02 6.28819177e-03
4.57564473e-01 3.73952687e-01 -1.46016502e+00 3.05031747e-01
9.70038414e-01 1.17195904e+00 1.05547130e-01 3.72559965e-01
-1.01624550e-02 1.00242984e+00 -3.42013210e-01 -1.38850912e-01
3.27061564e-01 -4.12813544e-01 -9.62715924e-01 -1.21840574e-01
7.04979956e-01 -3.25652771e-02 -3.48777443e-01 8.16234827e-01
1.03842378e-01 7.14750171e-01 8.85342956e-01 -7.03743756e-01
-1.14368308e+00 7.88843870e-01 -5.77865720e-01 2.57740974e-01
-6.16927780e-02 -7.90644512e-02 8.37948501e-01 -1.33054972e+00
4.29848671e-01 1.05876231e+00 9.96316433e-01 5.74497759e-01
-1.77991641e+00 -8.66008878e-01 2.83729583e-01 1.26464546e-01
-1.38422883e+00 -3.50758672e-01 6.30279064e-01 -4.71919924e-01
1.11515486e+00 3.23753744e-01 4.99158025e-01 7.42514551e-01
5.23233056e-01 6.59351647e-01 1.50480449e+00 -5.83070159e-01
1.17510334e-01 3.34406942e-01 8.04556847e-01 8.96892965e-01
9.52792764e-02 1.50423512e-01 -5.50109327e-01 9.35592130e-02
5.37348568e-01 3.61222237e-01 -3.38161767e-01 -1.89852476e-01
-8.05594921e-01 6.40095949e-01 9.13572371e-01 1.86987743e-01
-3.05386394e-01 1.94255143e-01 3.24793463e-03 4.87928212e-01
6.14832163e-01 8.07749331e-01 -5.07820666e-01 5.87590575e-01
-1.20222509e+00 -8.14375952e-02 5.72995722e-01 5.03272712e-01
1.28021777e+00 4.13844615e-01 -1.03531986e-01 1.09888661e+00
3.11628617e-02 3.42366040e-01 9.83589232e-01 -8.32127988e-01
-6.93264008e-02 4.93439049e-01 -3.63347709e-01 -9.22054172e-01
-4.59417909e-01 -8.78716648e-01 -1.18970728e+00 2.79292762e-01
1.18806116e-01 2.67630041e-01 -1.21724200e+00 1.95602369e+00
-2.19887614e-01 3.74401957e-01 -2.08329186e-02 3.98381889e-01
7.98036695e-01 7.49088705e-01 4.35320467e-01 -3.25886697e-01
6.23916566e-01 -1.15343642e+00 -5.15142344e-02 -3.99785370e-01
6.29103482e-01 -2.49313667e-01 1.07651591e+00 4.16599363e-01
-1.08349156e+00 -1.10147333e+00 -1.13244700e+00 1.28814057e-01
-5.05679369e-01 -4.46919724e-03 5.00175536e-01 3.12018216e-01
-1.32601964e+00 9.82156575e-01 -4.86933976e-01 -2.18946412e-01
3.16077739e-01 5.73855639e-01 -5.43686926e-01 -1.08380154e-01
-8.50744724e-01 1.17896354e+00 5.18330216e-01 -1.01462692e-01
-5.95719039e-01 -8.94253910e-01 -8.80831420e-01 1.07675180e-01
-7.55329877e-02 -7.89508522e-01 1.09562087e+00 -1.33194959e+00
-1.45109248e+00 9.88477707e-01 -5.26206076e-01 -7.40441084e-01
1.38432637e-01 -3.41515005e-01 -1.70801356e-01 -4.90559965e-01
-2.57187158e-01 9.33306932e-01 1.15840518e+00 -1.36535811e+00
-3.55349183e-01 -1.96632117e-01 -3.51327986e-01 -1.54603705e-01
-4.42942709e-01 -5.89330494e-01 7.60026881e-03 -8.10159206e-01
3.48991930e-01 -9.90293264e-01 -3.68814647e-01 -4.55183923e-01
-4.22943234e-01 -1.76570356e-01 4.33121473e-01 -6.67280078e-01
1.28568769e+00 -2.25208783e+00 5.06983884e-02 6.33174777e-01
5.91538727e-01 3.91858846e-01 -4.16297287e-01 -2.09805563e-01
-5.71515560e-01 1.54014736e-01 -4.29790974e-01 -4.32798117e-01
-2.05460757e-01 1.98785856e-01 -9.04766083e-01 3.05672556e-01
1.38025507e-01 1.08902693e+00 -5.27168453e-01 -1.73435837e-01
-4.36872914e-02 3.26925844e-01 -6.04427695e-01 1.24315679e-01
-3.87237258e-02 3.94526571e-01 6.41769320e-02 2.73232102e-01
7.09917068e-01 -4.92783517e-01 6.11755624e-03 5.10318987e-02
-2.75542624e-02 3.69840205e-01 -7.12397099e-01 1.36603248e+00
-4.15304959e-01 7.69954920e-01 -6.11688256e-01 -1.15240765e+00
1.28939950e+00 -2.23842680e-01 2.17869863e-01 -8.10995579e-01
-1.59666047e-01 1.54814959e-01 -1.89925924e-01 1.68133830e-03
6.53889477e-01 -3.12390059e-01 3.41326118e-01 5.94688535e-01
2.28917196e-01 1.73360109e-01 -1.61793217e-01 -1.22859858e-01
9.82649744e-01 2.23031178e-01 5.11096716e-01 -3.24473828e-01
4.28337634e-01 -9.85708833e-02 6.15649760e-01 1.19048893e+00
5.78606501e-02 3.48204732e-01 -1.95129722e-01 -8.04464221e-01
-8.98618162e-01 -1.42617512e+00 -3.50124180e-01 1.46471119e+00
-1.94183126e-01 -2.79320061e-01 -6.17268145e-01 -4.97474134e-01
1.88316956e-01 8.20898235e-01 -9.38257098e-01 -5.33699095e-01
-7.18366623e-01 -7.37161875e-01 3.71953756e-01 7.81703174e-01
6.27156675e-01 -1.18979061e+00 -2.70977288e-01 2.03104779e-01
8.03170949e-02 -4.14392531e-01 -2.24724218e-01 6.13694966e-01
-1.21327257e+00 -6.86548769e-01 -6.06954634e-01 -9.06445801e-01
9.26410377e-01 1.92632452e-01 1.34502447e+00 3.88645113e-01
-6.39597252e-02 1.53388366e-01 -1.05099268e-01 -2.64542490e-01
-2.91380554e-01 6.61297068e-02 4.75249738e-01 -1.96448192e-01
4.79441166e-01 -7.18691289e-01 -4.33768094e-01 2.07685977e-01
-7.99367130e-01 2.87795722e-01 6.67682946e-01 1.13715458e+00
8.76787424e-01 -1.76282078e-01 7.68690526e-01 -1.15065849e+00
6.44864202e-01 -5.04316628e-01 -2.80758142e-01 3.68036062e-01
-1.20493388e+00 4.47634846e-01 7.30516791e-01 -6.89949512e-01
-9.40177619e-01 2.79572755e-02 1.05949871e-01 -3.06787521e-01
-1.10805489e-01 8.50355744e-01 4.80087459e-01 -3.62940729e-01
9.49181199e-01 5.25418043e-01 -2.22319886e-01 -4.65954125e-01
4.10589248e-01 2.91926503e-01 6.92042589e-01 -7.08216012e-01
7.33032644e-01 1.55791402e-01 1.65326130e-02 -5.75551033e-01
-1.00118947e+00 -7.44927227e-02 -6.82267666e-01 -6.17372282e-02
3.91477585e-01 -6.28101587e-01 -5.85108161e-01 2.38623157e-01
-1.03884137e+00 -7.34045625e-01 -5.49169302e-01 5.23194432e-01
-5.96922696e-01 6.38282001e-02 -8.36655915e-01 -6.02800429e-01
-4.27020937e-01 -7.37806201e-01 1.37955219e-01 2.86918730e-01
-4.86940801e-01 -8.47752154e-01 3.41444016e-01 -2.08826140e-01
6.92236722e-01 -6.31811619e-02 1.31791604e+00 -7.58566439e-01
-3.34712267e-01 -1.00541227e-01 -3.83618385e-01 6.83294475e-01
-8.98338407e-02 -1.93872586e-01 -1.30308521e+00 -3.44937027e-01
1.70185924e-01 -3.28986049e-01 1.70578253e+00 5.62914670e-01
1.48790085e+00 -2.91567683e-01 -3.60031992e-01 7.91657805e-01
1.21643412e+00 -3.10050640e-02 7.24318862e-01 2.55881548e-01
5.10550320e-01 5.35017610e-01 8.14449321e-03 1.45564929e-01
1.50195405e-01 1.14132352e-01 -1.98103581e-02 -8.63225535e-02
-7.50335008e-02 -4.27012831e-01 3.56149882e-01 1.14781725e+00
-3.16698551e-01 4.19219509e-02 -9.15060937e-01 1.47668883e-01
-1.79712915e+00 -9.97809887e-01 2.70587683e-01 2.38160110e+00
9.63147700e-01 1.82776049e-01 -1.62995279e-01 2.03899801e-01
5.22470176e-01 -7.24532828e-02 -7.20610559e-01 -4.16013449e-01
-3.63528430e-01 7.98804879e-01 4.72743541e-01 5.14204502e-01
-9.35791194e-01 1.07967269e+00 7.71040869e+00 8.56627941e-01
-1.34235752e+00 4.02101651e-02 1.05259502e+00 -3.83539610e-02
-2.02846259e-01 -2.19341367e-01 -8.29016328e-01 4.34878469e-02
1.03958261e+00 -1.02016248e-01 6.47978723e-01 9.14470494e-01
-2.54605621e-01 -6.24676496e-02 -1.42514658e+00 9.25149381e-01
1.88096598e-01 -1.63233519e+00 3.17118585e-01 -1.31815776e-01
9.93950665e-01 2.02178974e-02 7.04247832e-01 7.73205519e-01
3.57031763e-01 -1.56568682e+00 5.48282564e-01 1.09497440e+00
6.98732316e-01 -5.23949921e-01 4.01583731e-01 3.66255015e-01
-9.02644753e-01 -2.12980837e-01 -8.53463352e-01 -3.03148627e-01
-4.53278393e-01 5.51532924e-01 -8.76450360e-01 -1.37198150e-01
6.84079587e-01 8.72149527e-01 -7.83033431e-01 1.09746695e+00
-5.02501540e-02 9.92425382e-01 -7.21317008e-02 1.55777067e-01
2.07773852e-03 -1.29948929e-01 2.36282721e-01 1.42085743e+00
2.91157275e-01 1.37642041e-01 2.18096320e-02 1.06722367e+00
-6.24056518e-01 1.87030733e-01 -5.38928151e-01 3.37421834e-01
3.17990363e-01 9.00867760e-01 -4.65079397e-01 -4.87414092e-01
-2.58355647e-01 8.26467872e-01 1.03606510e+00 7.00142443e-01
-3.10535192e-01 -1.75520614e-01 4.15278882e-01 9.14786309e-02
1.84195727e-01 -1.42594963e-01 -8.53837550e-01 -1.01554465e+00
-5.06460741e-02 -6.71282768e-01 1.74058169e-01 -8.99508595e-01
-1.46357822e+00 8.11159551e-01 -3.74435663e-01 -9.51931119e-01
-2.62252510e-01 -7.83916056e-01 -6.62073493e-01 1.08042133e+00
-1.56211579e+00 -8.31632912e-01 -2.37399325e-01 5.41516662e-01
3.21885586e-01 -3.61531228e-01 1.18100011e+00 -5.28240949e-02
-4.65105325e-01 8.35608840e-01 3.05606693e-01 1.04589783e-01
8.66770685e-01 -1.13052511e+00 2.27702618e-01 5.86091220e-01
2.49008894e-01 9.97568369e-01 5.44043481e-01 -6.54774010e-01
-8.57798576e-01 -1.19449723e+00 8.43306899e-01 -5.55428386e-01
5.16546428e-01 9.58174840e-02 -1.42437410e+00 1.03414583e+00
-7.14378152e-03 9.60402191e-03 9.57206190e-01 3.86380374e-01
-9.37740743e-01 -3.57649475e-01 -1.05653977e+00 5.25635362e-01
7.74784505e-01 -6.65232241e-01 -8.73551667e-01 -1.59236982e-01
6.56828225e-01 -5.39659224e-02 -6.40703142e-01 5.93370259e-01
6.88131094e-01 -7.52639711e-01 1.02612591e+00 -1.04735720e+00
5.81387222e-01 1.56227667e-02 -2.86874503e-01 -1.47705400e+00
-8.11447561e-01 -1.74819484e-01 -2.65798509e-01 7.61183441e-01
6.92015827e-01 -7.03017294e-01 8.61268520e-01 9.40862536e-01
-2.14676306e-01 -9.33654130e-01 -6.88370168e-01 -7.29205668e-01
4.91554707e-01 -5.35128653e-01 4.12786543e-01 1.08591235e+00
3.12368840e-01 1.27382725e-01 -5.00359058e-01 -1.99871227e-01
4.56185251e-01 1.11956008e-01 5.62673032e-01 -1.50597548e+00
-4.58911657e-01 -6.62715554e-01 -3.96599859e-01 -1.40932941e+00
6.47695601e-01 -1.21367061e+00 2.31253996e-01 -1.03648055e+00
3.89342338e-01 -7.48249054e-01 -1.04863906e+00 5.84966838e-01
-2.86009312e-01 5.86722791e-01 1.37414917e-01 4.84950423e-01
-4.98016179e-01 3.12017888e-01 1.01449108e+00 -3.72611374e-01
-4.92795587e-01 5.06484397e-02 -7.51424491e-01 8.33742559e-01
9.76585686e-01 -5.97000599e-01 -1.05094016e-01 -4.56103295e-01
2.57425457e-01 -2.58033335e-01 3.16378415e-01 -1.16199303e+00
4.37480658e-01 -1.87268496e-01 9.03679073e-01 -3.15411806e-01
3.42912734e-01 -4.07794297e-01 8.92051756e-02 4.56125826e-01
-8.65900278e-01 7.96360895e-02 4.11820799e-01 6.51787698e-01
-1.38755009e-01 -3.40528727e-01 9.83020902e-01 -2.44088154e-02
-7.40690112e-01 2.51948655e-01 -5.36110938e-01 -2.13252038e-01
5.79417408e-01 -3.59621227e-01 -4.79462266e-01 -1.06540836e-01
-1.02041364e+00 -1.59366906e-01 2.07728699e-01 -3.33172046e-02
9.44201112e-01 -1.31101048e+00 -6.73508584e-01 4.63095337e-01
-5.82874492e-02 -3.64377677e-01 -5.23069315e-02 6.79904521e-01
-1.38834342e-01 4.00794536e-01 -4.12093580e-01 -2.54936039e-01
-9.24907088e-01 6.21720254e-01 5.38914144e-01 -3.00816953e-01
-3.56751412e-01 1.10387206e+00 4.36365932e-01 -4.90145355e-01
1.99100345e-01 -4.74050224e-01 -8.21052119e-02 -4.38605249e-01
7.22769499e-01 4.51534420e-01 -1.14210412e-01 -5.18796921e-01
-3.77836972e-02 6.18043482e-01 -3.89284968e-01 9.20667052e-02
1.47889674e+00 1.06531270e-01 -3.45644623e-01 6.70660198e-01
1.09147704e+00 -7.90402219e-02 -1.12080371e+00 -9.01594996e-01
1.02063105e-01 -2.10396796e-01 -7.98889920e-02 -7.16000676e-01
-8.76670241e-01 8.39887202e-01 4.51178879e-01 3.34520638e-02
1.07736790e+00 -1.69872746e-01 4.93227690e-01 1.10941470e+00
2.19314069e-01 -1.15393245e+00 1.53580979e-01 9.43061054e-01
8.65587592e-01 -1.28954720e+00 3.47712561e-02 9.55779688e-04
-3.05275321e-01 1.21521175e+00 8.24667335e-01 -5.04246235e-01
1.04786742e+00 -7.87832364e-02 -1.17230237e-01 1.41982362e-01
-6.20533526e-01 1.25260949e-01 6.21965826e-01 5.39285600e-01
6.10075355e-01 1.22522034e-01 -6.60090614e-03 8.97260547e-01
-3.75707209e-01 1.74616247e-01 1.26457170e-01 5.81666410e-01
-8.28875601e-01 -7.42758334e-01 -2.55120039e-01 9.24016237e-01
-8.89270753e-02 -4.99379724e-01 -2.33501077e-01 3.23907644e-01
1.47145927e-01 6.86960638e-01 2.20221937e-01 -8.53532791e-01
3.53709161e-02 4.76236194e-01 4.12145525e-01 -7.36996412e-01
-6.12124920e-01 -4.38259095e-01 -5.38392425e-01 -4.27306384e-01
-2.48935506e-01 -1.69569492e-01 -1.21358049e+00 -6.19038582e-01
-2.51142144e-01 1.29166916e-01 3.66941363e-01 1.22045231e+00
1.51793376e-01 4.22785848e-01 4.13147628e-01 -8.37256789e-01
-5.80997884e-01 -1.06804657e+00 -5.86256027e-01 4.04156297e-01
3.59407604e-01 -5.52766800e-01 -2.71876216e-01 2.68008083e-01] | [8.866656303405762, 3.2313153743743896] |
100de743-96f2-42e4-b4ff-876ae03065ca | path-specific-causal-fair-prediction-via | null | null | https://openreview.net/forum?id=sWqjiqlUDso | https://openreview.net/pdf?id=sWqjiqlUDso | Path-specific Causal Fair Prediction via Auxiliary Graph Structure Learning | Algorithm fairness has become a trending topic, and it has a great impact on social welfare. Among different fairness definitions, path-specific causal fairness is a widely adopted one with great potentials, as it distinguishes the fair and unfair effects that the sensitive attributes exert on algorithm predictions. Existing methods based on path-specific causal fairness either require graph structure as the prior knowledge or have high complexity in the calculation of path-specific effect. To tackle these challenges, we propose a novel casual graph based fair prediction framework, which integrates graph structure learning into fair prediction to ensure that unfair pathways are excluded in the causal graph. Furthermore, we generalize the proposed framework to the scenarios where sensitive attributes can be non-root nodes and affected by other variables, which is commonly observed in real-world applications but hardly addressed by existing works. We provide theoretical analysis on the generalization bound for the proposed fair prediction method, and conduct a series of experiments on real-world datasets to demonstrate that the proposed framework can provide better prediction performance and algorithm fairness trade-off. | ['Jing Gao', 'Mengdi Huai', 'Jinduo Liu', 'Jingren Zhou', 'Bolin Ding', 'Yaliang Li', 'Liuyi Yao'] | 2021-09-29 | null | null | null | null | ['graph-structure-learning'] | ['graphs'] | [ 2.21291825e-01 -4.72298451e-02 -6.89585388e-01 -5.17244816e-01
-2.07056496e-02 -2.18946010e-01 2.83825159e-01 3.12094480e-01
-8.96703452e-02 9.35454607e-01 1.56991452e-01 -5.61263740e-01
-6.98854923e-01 -1.15421319e+00 -3.50365967e-01 -4.70726192e-01
-2.09845185e-01 1.97188750e-01 1.71036512e-01 -2.38577902e-01
4.39837068e-01 6.23036809e-02 -1.16327739e+00 -2.03843579e-01
1.54733336e+00 8.94958377e-01 -4.79657769e-01 1.21664666e-01
-5.93048632e-02 1.08390474e+00 -3.83183300e-01 -9.91964579e-01
4.00908053e-01 -6.83022141e-01 -7.25888371e-01 -3.60294133e-01
1.46855628e-02 -2.08897144e-01 -3.78547221e-01 1.27870393e+00
4.87847000e-01 2.31564753e-02 7.59462714e-01 -1.75956452e+00
-5.83247244e-01 9.86793876e-01 -9.90846336e-01 1.49498254e-01
1.44034788e-01 1.67868227e-01 1.35576916e+00 -4.82656509e-02
1.83558598e-01 1.57450855e+00 6.35309577e-01 3.85081142e-01
-1.21523845e+00 -1.22708452e+00 4.20542389e-01 6.02743447e-01
-1.14068115e+00 -1.27031416e-01 8.33775699e-01 -4.36812490e-01
2.90313601e-01 6.12530053e-01 5.05077481e-01 7.42374718e-01
3.66630495e-01 4.45013940e-01 1.26618648e+00 -8.79264995e-02
1.87458947e-01 -9.46175009e-02 3.07832599e-01 4.50361341e-01
6.86832190e-01 4.18793410e-01 -2.64971942e-01 -4.02374566e-01
4.51228917e-01 1.42473280e-01 -3.73417377e-01 -4.70608592e-01
-8.74957740e-01 9.75659013e-01 7.73743927e-01 -3.94325256e-01
-2.12585375e-01 6.94542304e-02 5.32728672e-01 3.77268404e-01
4.25778657e-01 1.86742976e-01 -1.94199756e-01 1.50035873e-01
-7.07178473e-01 2.36036807e-01 6.52391195e-01 9.36713755e-01
5.28257251e-01 -1.80610880e-01 -5.20677030e-01 6.29088342e-01
2.82779068e-01 2.79275179e-01 4.43113409e-02 -8.41255844e-01
5.23449361e-01 9.18548584e-01 7.79668465e-02 -1.55635929e+00
-3.96690071e-01 -6.32269025e-01 -1.19832623e+00 6.74206614e-02
4.73882467e-01 -1.86474621e-01 -4.18303668e-01 1.95154190e+00
5.73280573e-01 3.59885126e-01 -3.22413445e-01 1.06918967e+00
6.33886814e-01 4.84833956e-01 7.16080308e-01 -7.17093050e-01
9.58767831e-01 -6.89568937e-01 -6.68377519e-01 2.18249261e-01
5.36738276e-01 -6.21488035e-01 1.02868640e+00 9.93475989e-02
-6.39285862e-01 -1.94248021e-01 -8.35930407e-01 1.55433893e-01
3.52537297e-02 -3.99612367e-01 9.52642977e-01 9.53819394e-01
-4.51169759e-01 7.70468056e-01 -9.72295031e-02 -2.61113346e-01
7.11338162e-01 2.97555000e-01 1.86401065e-02 -8.98215771e-02
-1.68918920e+00 6.23944104e-01 3.19378167e-01 -3.77556086e-02
-6.53826296e-01 -9.27317798e-01 -3.51682037e-01 3.94596726e-01
9.71558273e-01 -8.74409556e-01 8.44955564e-01 -1.07396996e+00
-1.06031191e+00 5.52223742e-01 3.12384754e-01 -3.20377231e-01
1.06376171e+00 1.31284863e-01 -5.44879615e-01 -3.68028104e-01
-1.78545201e-03 -1.90022483e-01 6.63813293e-01 -1.02466333e+00
-6.59578145e-01 -5.17652869e-01 4.68867779e-01 3.53296667e-01
-4.92224991e-01 1.25707733e-02 8.85989442e-02 -6.07829928e-01
-2.02705607e-01 -7.27604508e-01 -3.76681298e-01 -5.90275601e-03
-7.52993643e-01 -3.51589322e-01 4.64294314e-01 -2.61166722e-01
1.68033814e+00 -1.82318377e+00 -1.65217668e-01 4.06358778e-01
5.96243560e-01 2.25184262e-01 6.40143314e-03 3.80701929e-01
-7.84030184e-02 5.27517498e-01 -2.69365042e-01 3.85962903e-01
1.59618244e-01 -3.88870984e-02 -1.04905412e-01 5.06788194e-01
-2.69729912e-01 4.73210573e-01 -1.00068617e+00 -6.61863029e-01
-2.34701131e-02 -1.90726817e-02 -4.81807679e-01 2.48260334e-01
4.83430326e-02 1.74089447e-01 -7.19509304e-01 3.19934577e-01
1.09358990e+00 -1.65772229e-01 4.80501533e-01 -1.81294400e-02
1.56253248e-01 1.06303878e-01 -1.34581900e+00 8.52011085e-01
-2.34530903e-02 -1.60653040e-01 1.03346780e-02 -1.11190307e+00
9.12676513e-01 1.10297501e-02 4.33476031e-01 -6.11879647e-01
2.38638267e-01 -2.67458148e-02 3.53110224e-01 -1.68159261e-01
3.79655480e-01 -4.71755534e-01 -1.63538218e-01 5.84524870e-01
-4.80246812e-01 3.08792591e-01 7.58567229e-02 3.95919472e-01
1.00195599e+00 -2.69142509e-01 8.24036300e-01 -5.95271826e-01
7.84559965e-01 -2.13670880e-01 1.21710455e+00 7.33662426e-01
-7.31664181e-01 -4.28796224e-02 1.02868581e+00 -3.32492441e-01
-6.54423296e-01 -8.41534734e-01 -9.68345180e-02 1.20024991e+00
6.95170760e-01 -2.85161763e-01 -5.93025267e-01 -9.78928804e-01
2.89705217e-01 6.76123381e-01 -6.09560370e-01 -5.34264147e-01
-7.15116858e-02 -1.08736789e+00 3.05413753e-01 8.37722048e-02
5.05539894e-01 -7.12162077e-01 -1.19820602e-01 5.79349184e-03
-2.06610069e-01 -7.26216495e-01 -5.41955650e-01 -2.32111827e-01
-7.57284820e-01 -1.43179345e+00 -3.46615732e-01 -2.06005648e-01
4.42920744e-01 3.64131868e-01 1.28456187e+00 4.71190542e-01
5.13003441e-03 -5.62711097e-02 -1.90536648e-01 -4.32403415e-01
-2.48935625e-01 5.00251055e-02 -6.10018037e-02 2.04194888e-01
2.78013170e-01 -6.65358186e-01 -8.68097186e-01 5.34259379e-01
-5.70488572e-01 4.81547080e-02 4.91647482e-01 7.82708883e-01
3.74038994e-01 3.00028652e-01 1.02430570e+00 -1.63424563e+00
1.05208707e+00 -8.48378778e-01 -5.91687858e-01 3.18506092e-01
-1.36571419e+00 -1.40331343e-01 9.90267515e-01 -5.04390076e-02
-1.17693079e+00 -2.75985569e-01 1.92229465e-01 -1.01873383e-01
1.83281928e-01 5.06065249e-01 -6.01587474e-01 -1.26253068e-01
6.02075279e-01 -2.47394159e-01 -2.35942438e-01 -1.82322040e-01
4.71624315e-01 4.41421151e-01 1.49088457e-01 -6.64983571e-01
7.80343115e-01 7.36600114e-03 6.38938904e-01 -2.35364102e-02
-7.67845571e-01 -2.31443077e-01 -3.28615785e-01 -2.25608468e-01
4.19847220e-01 -6.82529330e-01 -1.27599752e+00 1.75878182e-01
-8.36545527e-01 8.22953358e-02 7.53274113e-02 3.12561691e-01
-3.21934879e-01 4.80602741e-01 -4.82423991e-01 -1.01468766e+00
-4.55925673e-01 -8.63271296e-01 1.84294418e-01 2.97446787e-01
-2.35535055e-02 -9.13843989e-01 -1.34709433e-01 2.10786834e-01
1.75824896e-01 3.70912462e-01 1.34820199e+00 -5.19533813e-01
-5.56877077e-01 -1.07361600e-01 -6.69602394e-01 -8.53029191e-02
2.13276502e-02 4.50612195e-02 -5.82158506e-01 -1.34342834e-01
-3.78147751e-01 -1.81920066e-01 8.41528296e-01 2.82322228e-01
1.45078385e+00 -6.60971224e-01 -4.08860862e-01 4.41984415e-01
1.52718186e+00 -3.99622601e-03 5.49818218e-01 -2.82481741e-02
9.16574836e-01 7.42460430e-01 8.66733015e-01 7.03874588e-01
5.08653760e-01 5.93902290e-01 7.50762045e-01 -1.25932336e-01
1.68301567e-01 -4.26462442e-01 -8.12950581e-02 5.56884646e-01
-4.26618606e-01 -3.43751878e-01 -6.69239879e-01 1.63647115e-01
-2.16858244e+00 -1.12911534e+00 -6.27401829e-01 2.49928975e+00
8.01266670e-01 1.24826282e-01 4.97092307e-01 3.13758850e-01
1.17418528e+00 1.91641420e-01 -7.18366802e-01 -5.47530532e-01
1.57607704e-01 -1.24489486e-01 6.52143717e-01 4.22976851e-01
-1.02235615e+00 7.12592542e-01 6.08742332e+00 1.18493259e+00
-8.41902137e-01 4.36323276e-03 9.46556330e-01 -1.33006554e-02
-6.56331003e-01 2.07370743e-01 -2.28920564e-01 6.71714008e-01
5.83699226e-01 -9.60219622e-01 4.34631288e-01 8.04856539e-01
4.95885968e-01 2.02820152e-01 -9.22577024e-01 8.28173995e-01
-5.29984176e-01 -7.65467763e-01 1.72546640e-01 6.50350079e-02
7.26461709e-01 -6.42264903e-01 -1.02321878e-02 3.31455916e-01
6.88230217e-01 -1.10633671e+00 2.70150512e-01 5.15454173e-01
6.87667847e-01 -1.19648552e+00 7.13987827e-01 4.89173025e-01
-1.05407822e+00 -2.86489636e-01 -7.53722131e-01 -5.23275912e-01
-2.07999498e-01 1.00862682e+00 -4.11433816e-01 1.03253996e+00
3.84793043e-01 9.14816737e-01 -4.28100646e-01 1.13286138e+00
-3.82493585e-01 8.98701906e-01 2.07757041e-01 -2.35694692e-01
-2.36700565e-01 -3.40638429e-01 4.51523632e-01 9.21937346e-01
1.74442902e-01 3.08505327e-01 2.74504125e-01 8.95872414e-01
-4.01909888e-01 7.14563787e-01 -4.04937595e-01 2.90774852e-01
6.48663759e-01 1.37996387e+00 -6.11888289e-01 -1.12264924e-01
-3.00944060e-01 5.95056832e-01 3.91676992e-01 2.25037456e-01
-1.05624855e+00 -3.73914450e-01 7.14859903e-01 2.61753708e-01
-3.55922878e-01 5.43610871e-01 -6.09525144e-01 -1.10317934e+00
-2.43267089e-01 -8.44125509e-01 9.34744418e-01 -1.77801132e-01
-1.86981857e+00 1.44290894e-01 -3.06432754e-01 -1.40734780e+00
2.51408905e-01 -1.23187900e-01 -8.68051708e-01 9.52320635e-01
-1.58860886e+00 -1.02910364e+00 -4.53832984e-01 7.00794280e-01
-1.84532493e-01 6.87798113e-02 5.91470003e-01 4.32641298e-01
-7.65657842e-01 9.97612774e-01 -2.85276193e-02 -1.65178388e-01
8.43864381e-01 -1.06577826e+00 -1.75072029e-01 8.41522992e-01
-4.47636992e-01 5.79993606e-01 7.01364934e-01 -7.02617884e-01
-9.68594670e-01 -1.31672788e+00 8.00265074e-01 -1.13618940e-01
5.37346423e-01 -2.88055800e-02 -7.93660045e-01 4.49788004e-01
-6.65458441e-02 1.19570956e-01 7.78869390e-01 5.92258751e-01
-4.47103113e-01 -5.29231608e-01 -1.31648529e+00 5.77911019e-01
1.52958524e+00 -1.15798876e-01 -1.42065883e-01 1.29627824e-01
6.15653336e-01 -1.23816334e-01 -7.52391100e-01 5.85766554e-01
5.72746873e-01 -1.26460862e+00 7.85402715e-01 -7.89032578e-01
7.01536655e-01 -1.82444334e-01 2.26962790e-01 -1.47915649e+00
-7.54044950e-01 -6.33661032e-01 1.31971374e-01 1.41603243e+00
3.31014901e-01 -8.04686487e-01 6.77247286e-01 7.69642055e-01
2.95228899e-01 -6.78754389e-01 -7.14087188e-01 -7.70825624e-01
3.09610516e-01 -1.79634705e-01 1.14951384e+00 1.37763095e+00
2.10571393e-01 4.93636370e-01 -9.31396484e-01 -1.41354829e-01
9.04949903e-01 4.66728270e-01 7.69369304e-01 -1.68049514e+00
-1.95667326e-01 -6.97865963e-01 -5.93028724e-01 -5.87798655e-01
2.92228878e-01 -1.02730715e+00 -2.19618946e-01 -1.39305854e+00
7.34011531e-01 -7.82314360e-01 -6.43113852e-01 3.39630276e-01
-9.20867264e-01 -1.29098535e-01 1.67298943e-01 2.91914672e-01
-7.11362660e-01 7.22819388e-01 1.51262581e+00 -1.30414486e-01
3.21644880e-02 2.59170651e-01 -1.19264030e+00 5.92834711e-01
6.68027461e-01 -6.83495522e-01 -8.62644911e-01 1.84596583e-01
2.59564400e-01 3.57493699e-01 2.10767657e-01 -5.30857861e-01
-1.12891626e-02 -9.08862650e-01 6.85234740e-02 3.80997173e-02
-3.56570989e-01 -1.01054525e+00 2.08988443e-01 5.64929008e-01
-6.70980811e-01 -2.77343214e-01 -2.66211003e-01 9.36988771e-01
6.91314936e-02 1.78910360e-01 8.52039874e-01 9.09496769e-02
-4.50268984e-01 8.42909098e-01 1.92372605e-01 2.52121508e-01
1.29244852e+00 5.87034486e-02 -6.00009143e-01 -6.79030657e-01
-3.67014050e-01 5.07329285e-01 3.12226504e-01 1.71046287e-01
2.96314061e-01 -1.52092898e+00 -8.32549930e-01 -2.70913392e-01
2.16548085e-01 -6.21076882e-01 4.35740530e-01 8.54927063e-01
-8.24582726e-02 1.34749889e-01 -3.48679870e-01 -1.87053993e-01
-1.22700751e+00 8.83083522e-01 2.42728308e-01 -4.34570968e-01
-1.56567022e-01 5.06615460e-01 4.67638075e-01 -4.29095834e-01
-4.03054319e-02 3.27448159e-01 -3.42369944e-01 -1.50340676e-01
2.22661570e-01 7.97618032e-01 -2.86007315e-01 -6.00573063e-01
-3.63723040e-01 2.79689670e-01 1.82821140e-01 4.83246863e-01
9.50973928e-01 -3.08720976e-01 -3.39297712e-01 1.93699285e-01
7.53097177e-01 2.04461917e-01 -9.54776704e-01 -2.57740915e-01
-8.14657062e-02 -9.06179249e-01 2.20048260e-02 -8.95472944e-01
-1.18894112e+00 9.44075346e-01 3.68667811e-01 6.03941619e-01
1.33698857e+00 -5.38002729e-01 6.19917870e-01 -2.00488538e-01
4.24634874e-01 -9.59737241e-01 -4.37784106e-01 2.03345954e-01
5.40511310e-01 -1.26278019e+00 3.83851975e-01 -1.10060608e+00
-6.38572037e-01 8.50704134e-01 9.26870525e-01 -1.18700393e-01
7.94518590e-01 -3.50235760e-01 -2.09950641e-01 5.22209518e-02
-6.89162970e-01 -6.42470866e-02 3.71511549e-01 4.97306615e-01
5.61866701e-01 6.36336684e-01 -1.21284938e+00 1.14488173e+00
-1.34810582e-01 8.77558142e-02 5.64798832e-01 2.82374501e-01
-1.89407587e-01 -1.11629009e+00 -4.79645934e-03 9.53117907e-01
-6.10198319e-01 -1.17543504e-01 -4.67904955e-01 5.84535360e-01
1.52220055e-01 1.04751956e+00 -4.13018912e-01 -5.94331861e-01
4.33550417e-01 -3.48884076e-01 1.75291881e-01 -3.14988554e-01
-5.21601975e-01 -2.96791732e-01 1.08327612e-01 -7.14670956e-01
-3.25845629e-01 -3.45868051e-01 -1.12434101e+00 -1.30643868e+00
-4.11186010e-01 2.86176234e-01 -1.71088025e-01 7.62253821e-01
3.50285262e-01 5.54939330e-01 1.05410016e+00 8.03057700e-02
-6.98671460e-01 -6.06382906e-01 -9.70679879e-01 6.80477917e-01
1.13009000e-02 -8.26715648e-01 -4.53089774e-01 -4.24993366e-01] | [9.096705436706543, 5.401921272277832] |
8eac8e8c-332b-4df7-a55d-66f44e1f503d | a-data-augmentation-perspective-on-diffusion | 2304.10253 | null | https://arxiv.org/abs/2304.10253v1 | https://arxiv.org/pdf/2304.10253v1.pdf | A data augmentation perspective on diffusion models and retrieval | Diffusion models excel at generating photorealistic images from text-queries. Naturally, many approaches have been proposed to use these generative abilities to augment training datasets for downstream tasks, such as classification. However, diffusion models are themselves trained on large noisily supervised, but nonetheless, annotated datasets. It is an open question whether the generalization capabilities of diffusion models beyond using the additional data of the pre-training process for augmentation lead to improved downstream performance. We perform a systematic evaluation of existing methods to generate images from diffusion models and study new extensions to assess their benefit for data augmentation. While we find that personalizing diffusion models towards the target data outperforms simpler prompting strategies, we also show that using the training data of the diffusion model alone, via a simple nearest neighbor retrieval procedure, leads to even stronger downstream performance. Overall, our study probes the limitations of diffusion models for data augmentation but also highlights its potential in generating new training data to improve performance on simple downstream vision tasks. | ['Chris Russell', 'Francesco Locatello', 'Osama Makansi', 'Max Horn', 'Dominik Zietlow', 'Florian Wenzel', 'Max F. Burg'] | 2023-04-20 | null | null | null | null | ['open-question'] | ['natural-language-processing'] | [ 4.86400843e-01 6.69401824e-01 -3.05457935e-02 -4.36162204e-01
-7.66836643e-01 -8.29704583e-01 1.26540530e+00 1.91664755e-01
-6.80537701e-01 4.44901377e-01 5.47711909e-01 -5.91994464e-01
-7.12018739e-03 -8.50835085e-01 -4.54293966e-01 -5.99694550e-01
3.83861303e-01 5.93180835e-01 2.65429974e-01 -3.34574163e-01
3.47328275e-01 7.11002052e-01 -1.53778923e+00 4.83527392e-01
5.72107852e-01 6.15304232e-01 7.54670799e-02 8.34255278e-01
-1.46180227e-01 7.84553051e-01 -8.70969415e-01 -6.01738811e-01
6.29506350e-01 -4.98890787e-01 -7.82815516e-01 1.55188799e-01
7.56800771e-01 -8.54183495e-01 -5.15334368e-01 5.91302693e-01
4.96464938e-01 1.94649711e-01 8.32366347e-01 -1.24388468e+00
-1.59876764e+00 3.49518061e-01 -3.76848727e-01 4.60564017e-01
1.06599338e-01 6.73086584e-01 8.67099583e-01 -8.63502800e-01
8.52863848e-01 1.03184700e+00 7.31918931e-01 9.02441084e-01
-1.47101581e+00 -3.72307450e-01 2.41074413e-02 -1.50108695e-01
-9.96653020e-01 -6.52755976e-01 5.67298651e-01 -4.67101932e-01
1.14076710e+00 1.87738210e-01 6.27720475e-01 1.16700017e+00
-4.42518741e-01 7.03281343e-01 1.31322265e+00 -5.93088746e-01
-1.93388145e-02 4.58646417e-01 -1.53034613e-01 7.17464685e-01
1.85525119e-01 4.09077793e-01 -4.25432414e-01 -1.32367536e-01
8.92044246e-01 -4.31022733e-01 -2.17210650e-01 -2.97470152e-01
-1.22003901e+00 9.75564897e-01 6.92964375e-01 2.17344493e-01
-3.89741600e-01 2.06347525e-01 1.27341315e-01 3.54524255e-01
7.81132162e-01 1.02258801e+00 -2.40812823e-01 -4.54347394e-02
-8.98757517e-01 1.66007817e-01 5.95265269e-01 9.61152673e-01
8.47380638e-01 -1.35168537e-01 -6.53698921e-01 6.74446702e-01
1.13504596e-01 2.47291118e-01 5.83664536e-01 -1.13824320e+00
3.32902253e-01 6.65157259e-01 -5.45719899e-02 -7.01245546e-01
-7.47415498e-02 -2.43541777e-01 -4.08083767e-01 3.47294569e-01
7.23933876e-01 -3.05132180e-01 -1.20819414e+00 1.78895044e+00
3.08609724e-01 -2.74091005e-01 3.06708634e-01 7.06529200e-01
7.72517323e-01 5.85338533e-01 3.33206832e-01 1.88310742e-01
9.39795494e-01 -1.30448091e+00 -2.54258960e-01 -2.37129956e-01
1.23290920e+00 -8.84171963e-01 1.25814211e+00 5.54305725e-02
-1.05058539e+00 -4.82949078e-01 -8.59014213e-01 -5.09608746e-01
-4.58944440e-01 1.50678635e-01 8.96439075e-01 7.11841822e-01
-1.44309807e+00 5.38685083e-01 -4.21002299e-01 -7.00454473e-01
5.48251212e-01 3.20371211e-01 -4.42279667e-01 -2.37097934e-01
-8.32492769e-01 8.88318121e-01 6.64731562e-02 -3.00082266e-01
-7.94043183e-01 -7.24526763e-01 -6.87357247e-01 -1.55956417e-01
1.37493461e-01 -8.86477470e-01 1.31631923e+00 -1.07140911e+00
-1.15817761e+00 9.65542674e-01 1.63695022e-01 -5.01306117e-01
5.82002819e-01 -4.32861373e-02 1.71499014e-01 1.83653653e-01
8.78529623e-02 1.70424211e+00 7.81726539e-01 -1.24610603e+00
-3.81988823e-01 -2.98617363e-01 5.23100257e-01 5.21819592e-01
-8.54921281e-01 -2.53767818e-01 -3.92063230e-01 -7.81480670e-01
-3.11544359e-01 -1.07550001e+00 -5.50777733e-01 4.45997238e-01
-2.70949751e-01 -2.34115168e-01 8.38867366e-01 -3.80297214e-01
7.76411712e-01 -2.15640759e+00 -1.46996707e-01 1.21584117e-01
3.64356786e-01 4.51851040e-01 -5.97090185e-01 6.03947401e-01
1.32088140e-01 3.40496093e-01 -1.27798006e-01 -4.93733555e-01
-2.64593899e-01 1.53206021e-01 -4.88267928e-01 3.86242233e-02
3.72774631e-01 1.37539160e+00 -7.30932713e-01 -3.95471364e-01
9.38405022e-02 5.31472087e-01 -6.62137091e-01 1.48367390e-01
-3.61933887e-01 5.69105029e-01 -1.92177311e-01 1.56901106e-01
1.89777642e-01 -3.44775558e-01 -3.88470709e-01 -1.92830320e-02
1.29869863e-01 -9.86354076e-04 -6.94007158e-01 1.57357693e+00
-2.39589944e-01 1.01245391e+00 -1.78188369e-01 -6.17662907e-01
7.95591712e-01 1.31695554e-01 1.77777484e-01 -6.91832423e-01
-1.30977526e-01 1.50070041e-01 2.12019041e-01 -5.67531645e-01
7.59136379e-01 -2.65183985e-01 3.24547172e-01 8.07470679e-01
6.36484027e-02 -5.41013122e-01 2.39506930e-01 5.04284859e-01
1.19026816e+00 1.65520474e-01 -2.84131348e-01 3.12503381e-03
1.96006969e-01 6.37917697e-01 -1.63301989e-01 1.04037333e+00
-8.17448273e-02 7.64085591e-01 1.13536656e-01 -2.22466365e-01
-1.47703946e+00 -7.88139105e-01 1.49416566e-01 1.00783336e+00
-1.64173722e-01 -4.03209269e-01 -7.96763420e-01 -8.86001825e-01
-3.24434775e-04 7.72952795e-01 -7.68339336e-01 -4.07112926e-01
-1.68524742e-01 -8.11782837e-01 1.06413794e+00 8.90374482e-01
5.53315580e-01 -1.14215660e+00 -2.77900726e-01 -3.77467051e-02
2.03379303e-01 -9.93821621e-01 -4.26300704e-01 4.69684973e-02
-1.18465936e+00 -6.81508899e-01 -1.11440766e+00 -6.77925706e-01
9.19417858e-01 4.93049473e-01 9.08072829e-01 3.92856061e-01
-3.34183663e-01 8.87384593e-01 -2.95283347e-01 -4.25314069e-01
-8.02335620e-01 1.45499930e-01 -3.04935426e-01 -2.93088675e-01
5.86167693e-01 -3.14091653e-01 -7.76437044e-01 2.26015329e-01
-1.09227896e+00 1.56562790e-01 9.44756329e-01 6.77277684e-01
3.41132820e-01 -4.17395741e-01 5.87118983e-01 -1.09432447e+00
1.18189049e+00 -2.17467383e-01 -4.52328593e-01 4.35073636e-02
-9.48675215e-01 2.86986798e-01 1.20931476e-01 -7.72801459e-01
-1.21172345e+00 1.33234143e-01 -1.50282815e-01 -4.29567695e-01
-3.28417033e-01 3.10474336e-01 3.96573275e-01 -2.44360402e-01
1.38158262e+00 1.51968807e-01 2.95071274e-01 -2.63448626e-01
9.53328669e-01 5.86254776e-01 3.69294912e-01 -5.36705554e-01
9.18151021e-01 7.20481694e-01 -6.26767054e-02 -8.80421281e-01
-8.40077758e-01 -3.92421097e-01 -6.68583512e-01 -5.29737398e-03
9.51013923e-01 -7.39182830e-01 -1.71981864e-02 3.04519147e-01
-1.06556058e+00 -7.87565649e-01 -8.96660805e-01 2.95442164e-01
-5.76540291e-01 3.20612252e-01 -7.60020435e-01 -6.40520036e-01
-1.93798110e-01 -9.45953369e-01 1.21683896e+00 7.40227252e-02
-4.16220605e-01 -1.10791624e+00 1.53703794e-01 7.84042060e-01
6.74859047e-01 -3.46574128e-01 1.06035113e+00 -8.28831732e-01
-7.88010120e-01 -3.00816625e-01 -3.36957008e-01 4.87015158e-01
1.17920041e-01 1.21144252e-02 -1.19709980e+00 -6.18640222e-02
-2.35604927e-01 -7.00197339e-01 9.44627583e-01 2.97337547e-02
8.97282839e-01 -2.79633313e-01 -3.93687040e-01 3.69607717e-01
1.12748349e+00 6.55866414e-03 6.93867266e-01 3.52345228e-01
6.07911050e-01 1.03304935e+00 5.27726471e-01 -1.04000129e-01
2.93907523e-01 3.42004746e-01 1.10243060e-01 -4.46290612e-01
-6.76864088e-01 -3.91383201e-01 9.02755633e-02 1.16291322e-01
-4.20552716e-02 -3.20436120e-01 -9.16854620e-01 6.58576429e-01
-1.43610573e+00 -8.94745469e-01 -2.98776269e-01 1.97357929e+00
8.43912065e-01 6.84357062e-02 1.45398498e-01 -1.90364942e-01
5.32176256e-01 -6.99613243e-02 -6.42497718e-01 -2.38422260e-01
-1.91764846e-01 1.47124780e-02 4.25131619e-01 3.93010855e-01
-7.38179564e-01 1.22914839e+00 7.44181538e+00 5.53013563e-01
-1.04641283e+00 7.32531324e-02 9.71965551e-01 -2.21839249e-01
-5.77425718e-01 1.41201168e-01 -6.77325189e-01 -1.59294665e-01
6.48921788e-01 3.15711685e-02 3.35948318e-01 6.80441856e-01
8.58453196e-03 -1.64938122e-01 -1.30366349e+00 5.25838375e-01
2.53027081e-01 -1.59315431e+00 4.77821708e-01 5.12675047e-01
1.05303776e+00 8.52766931e-02 4.15186286e-01 2.85162568e-01
5.86222172e-01 -1.07445824e+00 3.95906031e-01 3.00912768e-01
6.86821580e-01 -2.47094527e-01 4.36888456e-01 5.25139987e-01
-3.31464916e-01 -7.52518550e-02 -2.92535841e-01 -1.82381570e-01
5.77729382e-02 2.84722358e-01 -1.18263912e+00 -8.32947344e-02
4.37385857e-01 2.82363802e-01 -1.15130806e+00 9.41127002e-01
-2.65855759e-01 4.54279810e-01 -3.28804553e-01 2.84501091e-02
4.46753144e-01 -7.16835409e-02 3.18589777e-01 9.64309275e-01
3.45053822e-01 8.55802745e-02 -2.45569348e-01 8.59688401e-01
-8.32002386e-02 2.16641992e-01 -1.27525902e+00 -5.25914967e-01
2.53157198e-01 1.28534079e+00 -6.44191504e-01 -6.02389574e-01
-4.43923563e-01 1.18197966e+00 5.02329588e-01 5.10589242e-01
-4.53864515e-01 -3.11289374e-02 3.87988269e-01 1.95373297e-01
1.92241117e-01 -3.89604360e-01 -4.44025844e-01 -9.17562187e-01
-5.82001656e-02 -8.18277299e-01 3.25636268e-01 -1.34235740e+00
-1.42980385e+00 6.08718038e-01 2.15035528e-02 -6.92950130e-01
-2.88887292e-01 -4.41340476e-01 -5.21208584e-01 9.77165401e-01
-1.25145411e+00 -1.44203091e+00 -4.84644175e-01 4.96139914e-01
6.39362931e-01 -2.22210586e-02 6.97696745e-01 -1.26011297e-01
-4.35554832e-02 4.12843227e-01 -1.29062951e-01 4.85449418e-05
8.79455328e-01 -1.39919293e+00 5.38075745e-01 6.91236377e-01
5.86756825e-01 6.74255610e-01 6.67060375e-01 -8.47573638e-01
-8.46565425e-01 -1.04427564e+00 7.54552245e-01 -9.73026752e-01
5.32566488e-01 -2.32859015e-01 -9.60012138e-01 8.51474226e-01
5.03436446e-01 -2.70976812e-01 7.31116652e-01 1.24308020e-02
-3.54380935e-01 1.86959386e-01 -1.09506416e+00 9.66015875e-01
1.32988441e+00 -5.54914296e-01 -4.99038965e-01 5.10817587e-01
7.78476655e-01 -1.21149704e-01 -7.51197040e-01 2.19249725e-01
1.10957295e-01 -9.40185726e-01 9.59156275e-01 -8.37737083e-01
7.77060032e-01 8.58165324e-02 7.52821863e-02 -1.27297270e+00
-2.84072757e-01 -4.57635880e-01 1.39167532e-01 1.57509875e+00
8.45985711e-01 -4.71944690e-01 1.24112284e+00 9.94337559e-01
4.74066734e-02 -5.11149406e-01 -4.35773730e-01 -6.98543787e-01
4.19260621e-01 -2.96496987e-01 3.33545923e-01 9.37338352e-01
-3.05684954e-01 8.03656280e-01 1.23092905e-02 -8.56982172e-02
3.25722992e-01 -1.90790862e-01 1.16546285e+00 -1.09228241e+00
-3.92489851e-01 -5.12992799e-01 -3.06400210e-01 -1.26882577e+00
-9.70285013e-02 -9.49975610e-01 -8.23305994e-02 -1.88154697e+00
2.97300946e-02 -8.58547688e-01 3.09906751e-01 5.74789524e-01
-2.34576628e-01 7.35420763e-01 1.10730290e-01 6.10505998e-01
-6.22372925e-02 5.12589812e-01 1.56872678e+00 -1.88057020e-01
-3.50440264e-01 -1.64209872e-01 -1.11500084e+00 4.61814165e-01
7.18945742e-01 -3.25142443e-01 -1.09168816e+00 -8.85145664e-01
2.00486183e-01 -4.49203640e-01 3.38368893e-01 -8.18663239e-01
3.17590594e-01 5.84138893e-02 4.75416303e-01 -1.19285565e-03
6.21272743e-01 -5.81032753e-01 -3.06777298e-01 2.47037500e-01
-6.55429244e-01 2.52312243e-01 4.13305074e-01 8.15251887e-01
7.22836405e-02 -5.26649773e-01 4.69549209e-01 -2.64748603e-01
-6.74990118e-01 1.46190181e-01 -6.66825593e-01 -7.58154839e-02
1.19348252e+00 -4.65096772e-01 -6.83258116e-01 -7.86938131e-01
-8.20291698e-01 -1.14549831e-01 8.48220110e-01 3.93615097e-01
3.82468939e-01 -1.13592494e+00 -5.10446966e-01 2.39715934e-01
3.71338606e-01 4.36672196e-02 -1.91294149e-01 7.31051862e-01
-4.45668131e-01 3.53122354e-01 -8.11891928e-02 -5.06319940e-01
-1.31211853e+00 9.53205645e-01 1.68668866e-01 -5.33307232e-02
-6.11992955e-01 1.12797570e+00 4.47366118e-01 -3.15070719e-01
8.46777707e-02 -1.43450230e-01 -7.64307454e-02 4.30061370e-02
1.63769156e-01 5.19596934e-02 -4.72152121e-02 -3.53508085e-01
1.86443567e-01 1.81695387e-01 -2.44621515e-01 -8.18334639e-01
1.27678239e+00 -1.07364379e-01 2.81993598e-01 -1.88026913e-02
8.27201009e-01 1.20049380e-01 -1.42964733e+00 -1.74263656e-01
-2.60439385e-02 -4.07725811e-01 -2.75278389e-02 -9.41344440e-01
-8.98120642e-01 1.07282495e+00 4.19606358e-01 2.93341905e-01
1.02395082e+00 3.08259308e-01 4.44852531e-01 6.89399540e-01
-6.84145540e-02 -8.96550655e-01 5.74773312e-01 9.95803103e-02
1.00110960e+00 -1.31794977e+00 -1.07565150e-01 -4.15260613e-01
-8.21455061e-01 8.82120848e-01 1.00422406e+00 8.26046914e-02
2.85608083e-01 2.24654302e-02 4.72769886e-01 -3.67079824e-01
-7.95153260e-01 -3.93951952e-01 1.30372629e-01 9.94157910e-01
4.46117789e-01 -6.42239034e-01 -4.94931340e-02 -3.18011850e-01
-2.38670543e-01 -1.04756765e-01 5.82218230e-01 9.73492086e-01
-2.81952381e-01 -1.29470301e+00 -1.94303945e-01 7.13190436e-01
-5.78193031e-02 -4.77178454e-01 -1.18261921e+00 9.35739815e-01
-1.43907532e-01 9.65845823e-01 2.95518011e-01 -1.62311450e-01
1.06352225e-01 2.36152187e-01 6.01214290e-01 -9.90634620e-01
-5.93856394e-01 -1.95622995e-01 4.10268128e-01 -1.82933450e-01
-4.34253305e-01 -5.73864043e-01 -8.28443706e-01 -9.23385620e-02
-4.85413611e-01 1.25362530e-01 7.67180443e-01 8.42105329e-01
6.50160789e-01 2.41022855e-01 -3.68418009e-03 -6.24077618e-01
-6.71011627e-01 -1.10791278e+00 -1.64355792e-03 5.97252846e-01
1.10281363e-01 -3.61644357e-01 -2.60230929e-01 3.30111593e-01] | [11.26136302947998, -0.10596626996994019] |
c2b88a50-17e5-42a6-b364-b631a8b138c0 | twistslam-fusing-multiple-modalities-for | 2209.07888 | null | https://arxiv.org/abs/2209.07888v2 | https://arxiv.org/pdf/2209.07888v2.pdf | TwistSLAM++: Fusing multiple modalities for accurate dynamic semantic SLAM | Most classical SLAM systems rely on the static scene assumption, which limits their applicability in real world scenarios. Recent SLAM frameworks have been proposed to simultaneously track the camera and moving objects. However they are often unable to estimate the canonical pose of the objects and exhibit a low object tracking accuracy. To solve this problem we propose TwistSLAM++, a semantic, dynamic, SLAM system that fuses stereo images and LiDAR information. Using semantic information, we track potentially moving objects and associate them to 3D object detections in LiDAR scans to obtain their pose and size. Then, we perform registration on consecutive object scans to refine object pose estimation. Finally, object scans are used to estimate the shape of the object and constrain map points to lie on the estimated surface within the BA. We show on classical benchmarks that this fusion approach based on multimodal information improves the accuracy of object tracking. | ['Jérôme Royan', 'Amine Kacete', 'Eric Marchand', 'Mathieu Gonzalez'] | 2022-09-16 | null | null | null | null | ['semantic-slam'] | ['computer-vision'] | [ 6.27928451e-02 -3.97247046e-01 -1.48734182e-01 -3.47515255e-01
-5.68561733e-01 -7.71326303e-01 7.26313531e-01 4.05774623e-01
-6.49586558e-01 3.82088095e-01 -4.65475261e-01 2.51077235e-01
-1.89545229e-01 -6.23987138e-01 -8.49643469e-01 -4.16145593e-01
2.54488409e-01 1.35495031e+00 8.08730245e-01 6.08243421e-02
5.25599718e-01 9.27376390e-01 -1.62230790e+00 -3.91978920e-01
6.04079783e-01 9.30753291e-01 4.79227901e-01 3.35855901e-01
-2.59658664e-01 -7.14028021e-03 -3.27834785e-01 3.54962982e-03
3.74910027e-01 2.06639722e-01 -2.98970282e-01 1.47757217e-01
7.15392709e-01 -1.80440173e-01 1.14086308e-02 1.05321276e+00
8.46697688e-02 -8.52725841e-03 4.40042168e-01 -1.50067616e+00
2.61949897e-01 1.83053881e-01 -7.29497313e-01 -2.03640506e-01
5.84604442e-01 -1.70552000e-01 7.44581401e-01 -1.08189654e+00
8.95206928e-01 1.18513572e+00 6.59741700e-01 1.91225797e-01
-1.34220016e+00 -7.30309129e-01 1.45815536e-01 2.60131121e-01
-1.66705596e+00 -3.73609602e-01 6.08587921e-01 -4.37518477e-01
4.49450046e-01 2.26000875e-01 6.81282997e-01 6.04241014e-01
7.81658590e-02 3.15916717e-01 8.85626137e-01 -2.21070156e-01
2.01207474e-01 1.33863121e-01 -1.20971901e-02 6.02219641e-01
6.26177907e-01 7.80617911e-03 -9.34525073e-01 -1.41299918e-01
5.39907753e-01 1.83348000e-01 7.49819353e-03 -1.35131395e+00
-1.40985703e+00 6.09125376e-01 6.30700767e-01 -4.06066030e-02
-3.54176641e-01 2.23371759e-01 -4.63591367e-02 -2.03493714e-01
1.22594513e-01 1.40911981e-01 -2.12447047e-01 2.53518838e-02
-8.23412418e-01 3.58298153e-01 5.26838720e-01 1.16261244e+00
1.15156269e+00 -5.01741588e-01 2.47797996e-01 1.30143568e-01
4.68380779e-01 1.02740717e+00 -8.90672952e-02 -1.00107741e+00
3.29999119e-01 9.78695989e-01 3.64472538e-01 -9.11866248e-01
-4.98263836e-01 -4.70155552e-02 -2.22899958e-01 2.33971670e-01
2.32134447e-01 4.59421277e-01 -8.27845395e-01 1.16816401e+00
8.34256351e-01 2.96184599e-01 -2.08548725e-01 9.81841683e-01
2.56083041e-01 1.04536869e-01 -8.19311813e-02 1.22488543e-01
1.26237810e+00 -4.53008592e-01 -2.87799805e-01 -3.77771527e-01
4.08195108e-01 -9.81441855e-01 4.29227591e-01 3.42299044e-01
-8.02471399e-01 -3.46183687e-01 -8.46542537e-01 1.23500317e-01
-2.65334189e-01 1.96109787e-01 3.57153118e-01 4.96041000e-01
-6.64758205e-01 2.00956240e-01 -1.00172043e+00 -4.12407398e-01
2.52175033e-01 6.22640312e-01 -5.20192623e-01 -1.67215168e-01
-4.48525667e-01 1.19129622e+00 6.36264443e-01 5.77389412e-02
-8.04098427e-01 -5.79139471e-01 -1.03733325e+00 -2.76694983e-01
6.21006370e-01 -6.75752044e-01 9.80362475e-01 -3.17269444e-01
-1.13173616e+00 8.78512502e-01 -3.51750582e-01 -4.89055932e-01
6.52541697e-01 -4.91807759e-01 -1.73387583e-02 -2.08450705e-02
3.04972410e-01 7.21121132e-01 8.53198886e-01 -1.34453511e+00
-9.26910639e-01 -7.74250150e-01 -2.10175797e-01 3.61102700e-01
3.79183322e-01 -3.34300011e-01 -7.45099187e-01 2.54144847e-01
9.49370384e-01 -1.19292545e+00 -1.49712667e-01 3.14409077e-01
-2.48376146e-01 -8.19093734e-02 1.35705018e+00 7.80666620e-02
4.58129764e-01 -2.01255322e+00 3.17810982e-01 3.98770154e-01
-1.01379298e-01 -2.75174100e-02 7.24200206e-03 1.44799098e-01
5.19116759e-01 -4.95335311e-01 4.00449038e-02 -5.75237989e-01
-3.58081679e-03 3.23780477e-01 -2.13660702e-01 8.26840699e-01
-1.42124653e-01 7.26590335e-01 -7.96504438e-01 -4.91987467e-01
6.63272917e-01 5.50854385e-01 -2.65199333e-01 7.32428720e-03
-3.98826033e-01 6.37298524e-01 -4.84577149e-01 5.81306398e-01
9.94458973e-01 2.16615662e-01 1.86008751e-01 -2.76764423e-01
-3.81199569e-01 1.04949236e-01 -1.50576460e+00 1.91098881e+00
-4.44014221e-01 3.89894038e-01 2.86764115e-01 -3.39691937e-01
9.63898718e-01 -9.34943929e-02 7.31898129e-01 -2.90992230e-01
1.70559883e-01 3.02765399e-01 -3.33231449e-01 -1.91997141e-02
6.36781216e-01 1.23343587e-01 9.26945060e-02 1.15352891e-01
-1.83275968e-01 -5.37192345e-01 8.97886530e-02 -2.20857313e-04
7.96261668e-01 3.83097917e-01 2.18655616e-01 -1.42643049e-01
6.89222753e-01 2.70102054e-01 3.67051244e-01 5.66297233e-01
2.03711823e-01 4.37651217e-01 -2.48422176e-01 -3.64775032e-01
-9.59285259e-01 -1.33746350e+00 -2.74562776e-01 6.86702430e-01
8.85815501e-01 -2.69934207e-01 -4.48187679e-01 -5.29031038e-01
5.22813022e-01 3.91012907e-01 -2.60847420e-01 -2.01049298e-02
-5.48093855e-01 -2.91625500e-01 3.50643415e-03 2.31419742e-01
2.05457836e-01 -5.50049245e-01 -1.30889547e+00 2.17189699e-01
-1.64585054e-01 -1.29630506e+00 -2.27017865e-01 3.82334590e-02
-6.69332504e-01 -1.35155976e+00 -1.65207043e-01 -3.39520156e-01
7.40737736e-01 2.94403046e-01 6.36032164e-01 5.79443425e-02
-3.16857427e-01 5.43651283e-01 -1.99825242e-01 -3.40489656e-01
-2.69183248e-01 5.59600107e-02 3.51754874e-01 1.58763766e-01
2.65169263e-01 -1.52005449e-01 -4.97272015e-01 7.52129138e-01
-4.55397010e-01 9.68356803e-02 3.63989532e-01 1.45480633e-01
8.87573004e-01 -1.50656387e-01 -1.22531481e-01 -4.00156051e-01
-2.48503521e-01 -7.96995461e-02 -1.25852299e+00 4.76373509e-02
-4.49226111e-01 1.19359305e-04 -1.58429578e-01 -4.80361760e-01
-5.34917831e-01 8.58414233e-01 2.50083536e-01 -5.90446115e-01
-1.66073337e-01 1.80626959e-01 -2.46543840e-01 -5.26470959e-01
2.90414244e-01 1.44231498e-01 -9.71688256e-02 -5.53753197e-01
2.35478565e-01 2.07022503e-01 6.82458162e-01 -5.07532179e-01
1.14024127e+00 1.02178264e+00 5.50261438e-01 -6.50707722e-01
-8.02232563e-01 -8.09450984e-01 -1.06428456e+00 -3.31398457e-01
8.05337548e-01 -8.58948469e-01 -9.74312782e-01 1.80860519e-01
-1.19729662e+00 6.73957318e-02 -9.37697813e-02 7.99259424e-01
-6.44047141e-01 3.46140385e-01 8.41519758e-02 -8.68056834e-01
-8.32764804e-02 -1.17871630e+00 1.52535594e+00 6.74762065e-03
-3.52397785e-02 -4.98562843e-01 -6.84493035e-02 1.98535740e-01
1.52700022e-01 1.87504902e-01 3.09031248e-01 -2.10567266e-01
-1.07271624e+00 -4.56673265e-01 -8.02580863e-02 -5.52309632e-01
-2.65995357e-02 3.32847871e-02 -5.63217819e-01 -3.96641493e-01
-2.51108676e-01 3.53819758e-01 5.72141469e-01 2.87639588e-01
5.80865562e-01 4.46392089e-01 -9.81270015e-01 5.37821889e-01
1.42573822e+00 -8.55935439e-02 1.58453345e-01 4.58781689e-01
6.68817401e-01 5.27320266e-01 1.23275292e+00 4.74642903e-01
3.81432086e-01 1.20087016e+00 9.54322398e-01 4.83317971e-01
1.33423343e-01 -2.22192869e-01 2.74181634e-01 3.99984539e-01
2.57255971e-01 -2.67394688e-02 -1.14610672e+00 3.60722452e-01
-1.88449693e+00 -2.94312984e-01 -4.44401950e-01 2.34807014e+00
2.12202519e-01 2.69232392e-01 -9.50002670e-03 -1.20778054e-01
8.30595791e-01 -2.51706809e-01 -5.09270310e-01 4.29448664e-01
5.26938178e-02 -1.23705618e-01 9.51879025e-01 6.34353101e-01
-9.11309898e-01 1.14536226e+00 5.51100779e+00 2.54733622e-01
-1.03177929e+00 1.21225014e-01 -5.56910336e-01 -1.27534717e-01
-2.82511897e-02 4.15824324e-01 -1.32926476e+00 2.85811961e-01
5.78416407e-01 -1.37902424e-01 1.27712339e-01 7.81493306e-01
-4.18444313e-02 -5.87931812e-01 -1.27608550e+00 1.05705619e+00
8.82314816e-02 -1.09777510e+00 -8.75875633e-03 2.04914689e-01
4.14164692e-01 2.83228487e-01 -7.37727433e-02 -8.77704397e-02
1.96855783e-01 -5.14338791e-01 1.13731158e+00 6.38460398e-01
4.67168748e-01 -5.89704037e-01 5.69689393e-01 7.00889528e-01
-1.57630265e+00 1.04457855e-01 -4.04948205e-01 4.91170883e-02
4.80028301e-01 3.22853655e-01 -1.23813796e+00 5.97138405e-01
5.56125104e-01 4.50660616e-01 -7.63467312e-01 1.39260399e+00
-1.71936393e-01 -8.23361129e-02 -7.52693772e-01 2.41290569e-01
4.49839085e-02 -3.55853796e-01 7.94656813e-01 4.39951360e-01
6.95404947e-01 -2.89005965e-01 5.47116160e-01 8.63896132e-01
1.77606255e-01 -3.57636586e-02 -4.68059450e-01 5.06708860e-01
7.28600740e-01 1.21704996e+00 -1.03538620e+00 -1.36848152e-01
-1.12332977e-01 7.76253521e-01 7.28431046e-02 -1.54297158e-01
-7.77322233e-01 3.43871385e-01 6.89244390e-01 4.65000957e-01
3.23607683e-01 -7.61356592e-01 -3.54542886e-03 -1.03704226e+00
1.75212994e-01 -2.93857217e-01 1.86826304e-01 -8.36114585e-01
-6.24274611e-01 2.15992451e-01 6.47724271e-02 -1.28189790e+00
-9.27907825e-02 -5.22050917e-01 -3.63203371e-03 5.07084310e-01
-1.33623016e+00 -1.38943064e+00 -6.94371402e-01 5.22687554e-01
3.91745865e-01 2.06783980e-01 4.24899966e-01 1.72734827e-01
-5.00915898e-03 -1.86369643e-01 -7.80307874e-02 -2.25398719e-01
7.11494446e-01 -1.06439555e+00 1.91174448e-01 5.75668097e-01
5.61940789e-01 4.51670349e-01 7.67639160e-01 -1.07224572e+00
-1.74008179e+00 -9.89418805e-01 5.63846290e-01 -9.60804641e-01
4.28051591e-01 -6.17319763e-01 -8.04112911e-01 8.33418787e-01
-5.15601397e-01 2.35404715e-01 -1.33762464e-01 -2.60151699e-02
-2.56114304e-01 -2.07348511e-01 -9.05492604e-01 1.77246585e-01
8.20423782e-01 -3.07210892e-01 -3.98246408e-01 2.47104600e-01
4.23830360e-01 -9.52245653e-01 -6.07739210e-01 6.19228303e-01
6.08765781e-01 -7.78002262e-01 1.11062944e+00 -1.10784486e-01
-5.74250102e-01 -9.02947068e-01 -2.65042484e-01 -7.99985051e-01
1.84422702e-01 -1.11374393e-01 8.85500759e-03 9.82008517e-01
-1.86453447e-01 -4.40654397e-01 9.81401742e-01 2.80260652e-01
7.96619952e-02 -4.51713474e-03 -1.48169374e+00 -9.27710891e-01
-5.40511429e-01 -4.29531872e-01 6.23619199e-01 4.94116575e-01
-5.76921523e-01 1.48001999e-01 -2.18945622e-01 7.57248759e-01
1.00042927e+00 5.71836770e-01 1.21001112e+00 -1.86220884e+00
3.50756019e-01 -2.76429951e-01 -8.09535265e-01 -9.13089097e-01
4.23011750e-01 -7.94400692e-01 2.77556181e-01 -1.19696629e+00
6.42119274e-02 -6.78135633e-01 6.86339960e-02 1.52477011e-01
2.31565580e-01 5.00782371e-01 1.38507992e-01 3.14179331e-01
-8.50451350e-01 2.55497366e-01 6.56107068e-01 6.27862364e-02
-3.50252688e-02 4.53737564e-02 3.16386223e-01 8.35762620e-01
4.48697597e-01 -7.76011288e-01 1.49103731e-01 -4.92152900e-01
2.46582508e-01 4.85150293e-02 8.85380089e-01 -1.30855978e+00
5.64108849e-01 -2.06927359e-01 2.68112510e-01 -1.32741845e+00
8.49275410e-01 -1.42825449e+00 6.14609182e-01 7.54531860e-01
1.06186211e-01 2.41265669e-02 2.28358097e-02 5.89899957e-01
6.18502684e-02 -3.40476096e-01 8.18392277e-01 5.24701551e-02
-7.90218294e-01 5.61125875e-01 -5.71351778e-03 -5.81700087e-01
1.29846752e+00 -1.48361355e-01 1.19074963e-01 -1.07204802e-01
-6.74996197e-01 3.42218071e-01 1.05467451e+00 5.46390057e-01
5.74848652e-01 -1.14677274e+00 -4.83570099e-01 3.48215580e-01
2.42437854e-01 4.02906358e-01 -2.94799060e-02 1.12710989e+00
-5.14342308e-01 4.15270716e-01 -1.14374593e-01 -1.34215128e+00
-1.70206773e+00 5.32823563e-01 2.60976940e-01 4.58834887e-01
-5.94712675e-01 4.11156118e-01 -4.85809557e-02 -4.44865972e-01
6.38515651e-02 -2.49589294e-01 1.63540885e-01 1.30316675e-01
2.25299090e-01 3.54024380e-01 9.15405154e-02 -1.04224670e+00
-8.04464161e-01 1.15979838e+00 1.87956154e-01 -2.47090474e-01
1.06135368e+00 -5.13821721e-01 -5.80293462e-02 4.72447246e-01
7.37421274e-01 1.87891260e-01 -1.16223252e+00 -5.89141250e-01
5.35398245e-01 -7.56901920e-01 -1.41982049e-01 -1.69079840e-01
-7.13137150e-01 7.96426177e-01 8.07917058e-01 -2.24406868e-01
5.83606541e-01 3.92908931e-01 4.13280696e-01 4.01333064e-01
9.24927294e-01 -7.84706056e-01 -1.31594777e-01 5.42918980e-01
5.34163296e-01 -1.22639906e+00 3.43660176e-01 -6.35992646e-01
-2.12283224e-01 1.07930160e+00 5.63129067e-01 -1.04432646e-02
2.17665598e-01 2.96645880e-01 -1.94329191e-02 -1.98233545e-01
-2.17602521e-01 -3.68037820e-01 3.58026177e-01 5.02037227e-01
-2.78732002e-01 -8.15435350e-02 1.02122620e-01 -1.83221713e-01
-2.37753108e-01 -1.97359666e-01 1.49415284e-01 9.90006685e-01
-7.22318769e-01 -1.08335888e+00 -7.58072257e-01 -2.20785644e-02
5.08432649e-02 4.47745174e-01 -3.15361202e-01 9.43676293e-01
2.99285531e-01 4.82979983e-01 3.41460824e-01 -5.58684170e-02
3.57030362e-01 -2.23063584e-02 8.00904453e-01 -6.75749779e-01
-1.03583992e-01 1.66254401e-01 -3.78800243e-01 -7.97718167e-01
-4.39668626e-01 -9.51721787e-01 -1.35254967e+00 3.51972990e-02
-6.38323009e-01 1.11623228e-01 1.24020231e+00 7.93934584e-01
2.29964390e-01 -1.20559007e-01 1.96729332e-01 -1.12591279e+00
-4.62480992e-01 -5.29013276e-01 -5.71557224e-01 2.96978027e-01
3.13064009e-01 -1.13128734e+00 5.04192617e-03 -1.97677955e-01] | [7.321204662322998, -2.312021493911743] |
696600c2-1b95-49a0-9dde-b9d6ca7c713a | disentangled-modeling-of-domain-and-relevance | 2208.05753 | null | https://arxiv.org/abs/2208.05753v1 | https://arxiv.org/pdf/2208.05753v1.pdf | Disentangled Modeling of Domain and Relevance for Adaptable Dense Retrieval | Recent advance in Dense Retrieval (DR) techniques has significantly improved the effectiveness of first-stage retrieval. Trained with large-scale supervised data, DR models can encode queries and documents into a low-dimensional dense space and conduct effective semantic matching. However, previous studies have shown that the effectiveness of DR models would drop by a large margin when the trained DR models are adopted in a target domain that is different from the domain of the labeled data. One of the possible reasons is that the DR model has never seen the target corpus and thus might be incapable of mitigating the difference between the training and target domains. In practice, unfortunately, training a DR model for each target domain to avoid domain shift is often a difficult task as it requires additional time, storage, and domain-specific data labeling, which are not always available. To address this problem, in this paper, we propose a novel DR framework named Disentangled Dense Retrieval (DDR) to support effective and flexible domain adaptation for DR models. DDR consists of a Relevance Estimation Module (REM) for modeling domain-invariant matching patterns and several Domain Adaption Modules (DAMs) for modeling domain-specific features of multiple target corpora. By making the REM and DAMs disentangled, DDR enables a flexible training paradigm in which REM is trained with supervision once and DAMs are trained with unsupervised data. Comprehensive experiments in different domains and languages show that DDR significantly improves ranking performance compared to strong DR baselines and substantially outperforms traditional retrieval methods in most scenarios. | ['Shaoping Ma', 'Min Zhang', 'Xiaohui Xie', 'Jiaxin Mao', 'Yiqun Liu', 'Qingyao Ai', 'Jingtao Zhan'] | 2022-08-11 | null | null | null | null | ['ad-hoc-information-retrieval'] | ['natural-language-processing'] | [ 1.64688855e-01 -2.47301042e-01 -4.82007205e-01 -3.19977403e-01
-1.16424894e+00 -6.79105699e-01 7.73301244e-01 2.75005698e-02
-4.94679064e-01 5.34110844e-01 3.45821470e-01 2.95728240e-02
-3.47361386e-01 -7.43412971e-01 -4.23589528e-01 -3.80601168e-01
3.24647814e-01 1.09380901e+00 2.75823742e-01 -4.16423082e-01
8.27588513e-02 4.12351966e-01 -1.55246186e+00 3.34619820e-01
9.69980061e-01 6.65214419e-01 4.99314308e-01 2.20848531e-01
-2.77128041e-01 2.65365690e-01 -6.48897350e-01 -3.41779768e-01
5.56377113e-01 -1.16674833e-01 -7.39583850e-01 -2.38181368e-01
3.74131352e-01 -5.02127409e-01 -7.07964659e-01 9.22963440e-01
6.30229056e-01 4.26771462e-01 7.30888546e-01 -7.84786165e-01
-1.02149105e+00 5.95647752e-01 -6.65489137e-01 2.28457227e-01
3.02028835e-01 -3.38890821e-01 1.11846972e+00 -1.15381336e+00
7.23753512e-01 1.34571195e+00 1.07814267e-01 6.87841177e-01
-1.24663568e+00 -8.98961365e-01 1.30467966e-01 1.52957579e-02
-1.52191246e+00 -4.30095643e-01 8.12685490e-01 -1.09283775e-01
8.75605345e-01 1.35304913e-01 7.22150579e-02 1.18032861e+00
-3.39633614e-01 8.71088564e-01 6.65750027e-01 -5.98476529e-01
2.00550050e-01 4.15786773e-01 3.76260847e-01 9.56127942e-02
4.46807206e-01 1.50251523e-01 -5.72402179e-01 -3.80701184e-01
4.68930304e-01 1.99515402e-01 -1.51011616e-01 -6.49860024e-01
-1.02953446e+00 9.05761302e-01 4.08710569e-01 4.49298471e-01
-2.06780881e-01 -3.99385840e-01 3.24531049e-01 5.59122562e-01
5.32536030e-01 8.08203399e-01 -4.30320382e-01 2.29829729e-01
-9.93259013e-01 3.69044870e-01 6.05657041e-01 1.08824825e+00
9.47079301e-01 -5.04500687e-01 -1.45304814e-01 1.34408295e+00
4.37443167e-01 7.40783393e-01 7.89958775e-01 -6.37070298e-01
6.64399564e-01 7.98160613e-01 7.12749884e-02 -1.01696265e+00
7.94025511e-02 -4.90755379e-01 -6.88220680e-01 -3.27203751e-01
1.46075785e-01 2.15103745e-01 -9.93243277e-01 1.74299729e+00
2.56144822e-01 -1.46228909e-01 4.01078612e-01 1.03833640e+00
7.40262628e-01 7.25703001e-01 1.98649570e-01 1.02675803e-01
1.20908356e+00 -9.23150539e-01 -5.49719334e-01 -7.67960787e-01
8.06216955e-01 -8.53778005e-01 1.13815665e+00 5.35059273e-02
-8.95512402e-01 -4.79284286e-01 -1.03964734e+00 -3.04427207e-01
-4.67951953e-01 7.86455646e-02 5.80526888e-01 2.14005858e-01
-7.76596785e-01 1.84834793e-01 -5.37964463e-01 -5.35805285e-01
1.19104750e-01 4.43606019e-01 -5.10492146e-01 -7.21266985e-01
-1.74270356e+00 8.64491701e-01 4.42379534e-01 -1.22559592e-01
-7.43942142e-01 -5.77508688e-01 -8.25299323e-01 2.75117069e-01
3.43839526e-01 -7.36741900e-01 1.28741205e+00 -7.62116551e-01
-1.13113666e+00 9.47277188e-01 -1.17299594e-01 -2.24214390e-01
2.03932881e-01 -4.71255422e-01 -5.69373608e-01 2.25855708e-01
1.63854659e-01 6.20095253e-01 9.35252786e-01 -1.03646660e+00
-3.93780887e-01 -6.02209330e-01 -6.19274005e-03 6.11501396e-01
-5.82505882e-01 4.46536765e-02 -8.40702534e-01 -6.41295314e-01
5.62081300e-02 -7.99658060e-01 -7.11511746e-02 -2.56988019e-01
8.79913419e-02 -2.73771346e-01 8.64095569e-01 -5.90258479e-01
1.37868285e+00 -2.40907598e+00 3.01873088e-01 3.22440803e-01
2.22541392e-01 6.88002229e-01 -4.78998810e-01 5.40829301e-01
1.46158338e-01 2.65603680e-02 4.63660322e-02 -2.82028556e-01
8.47125426e-02 2.63567597e-01 -6.10030353e-01 5.09337336e-02
2.62038466e-02 7.47352660e-01 -9.83565509e-01 -5.14470458e-01
-8.27329904e-02 2.95075655e-01 -6.44766808e-01 4.79603589e-01
-2.54079640e-01 1.32386878e-01 -7.89747834e-01 4.44530904e-01
7.84618676e-01 -4.05767024e-01 3.99328589e-01 -6.04318231e-02
4.51154858e-01 4.56688762e-01 -1.01814520e+00 1.93544436e+00
-5.92952311e-01 4.44442809e-01 -1.33707330e-01 -9.11397696e-01
1.09726918e+00 2.53621370e-01 3.42946112e-01 -1.24515712e+00
-2.92358071e-01 4.07830298e-01 -3.09661478e-01 -1.78789169e-01
7.68408179e-01 -2.45769173e-02 -2.87280500e-01 6.48276806e-01
-1.40078459e-02 1.06239744e-01 1.09997079e-01 5.79240263e-01
1.29395831e+00 -2.08869845e-01 6.10346310e-02 7.08798766e-02
3.63979727e-01 1.47218332e-01 6.30493581e-01 8.90763402e-01
1.80167705e-02 5.13641596e-01 2.87772156e-02 -5.63623458e-02
-9.53378141e-01 -1.19799769e+00 -1.56995550e-01 1.34436989e+00
5.37575006e-01 -2.86992759e-01 -2.77682573e-01 -8.27557683e-01
3.68557394e-01 5.35577834e-01 -3.06726336e-01 -6.50790155e-01
-6.33559823e-01 -4.01713699e-01 4.06427830e-01 4.37584877e-01
5.49442589e-01 -7.95760036e-01 1.79510526e-02 1.10146053e-01
-2.37039104e-01 -1.03738773e+00 -5.26404202e-01 1.21402815e-01
-8.39172661e-01 -7.65252590e-01 -7.82392323e-01 -7.54958332e-01
7.06975102e-01 9.66308594e-01 1.26499248e+00 -8.61218870e-02
-5.43718785e-03 3.14224839e-01 -5.78579545e-01 -4.03008722e-02
-2.49363571e-01 4.33090806e-01 7.96735585e-02 -3.64632428e-01
1.02050114e+00 -3.15225452e-01 -7.17552543e-01 5.69329798e-01
-1.35263801e+00 -2.63432503e-01 9.42790627e-01 1.00970697e+00
4.38136578e-01 -6.87063783e-02 8.37890804e-01 -1.02418709e+00
8.96740079e-01 -7.45358527e-01 -5.52137613e-01 5.75477958e-01
-8.96829367e-01 4.26702559e-01 2.54160732e-01 -5.85682392e-01
-1.28419709e+00 -3.35062951e-01 2.52007633e-01 -5.53021014e-01
5.90380132e-02 6.54187858e-01 -2.82041192e-01 2.46856496e-01
9.13379848e-01 5.92843667e-02 -2.12237220e-02 -8.16684365e-01
4.41534340e-01 1.13467431e+00 2.16766551e-01 -7.13903666e-01
9.82729435e-01 1.29156306e-01 -5.79586983e-01 -3.38241398e-01
-9.61324453e-01 -1.03484225e+00 -5.30252755e-01 3.41219813e-01
4.48438853e-01 -1.28133869e+00 3.32815230e-01 4.52740788e-02
-9.66765583e-01 -6.86701462e-02 -2.43485451e-01 5.17343462e-01
4.54630097e-03 3.87830138e-01 -3.11474532e-01 -1.88016981e-01
-3.55066180e-01 -1.01151764e+00 1.27666795e+00 1.60551220e-01
-3.40735346e-01 -1.01697516e+00 4.51605797e-01 6.07545674e-01
5.99067926e-01 -5.56409538e-01 1.12436247e+00 -1.15325832e+00
-5.96233428e-01 -5.25228977e-01 -3.52635801e-01 4.72536623e-01
2.36525267e-01 -4.76042718e-01 -8.66738498e-01 -4.96790856e-01
-2.18443349e-01 -5.50798476e-01 8.55808616e-01 -1.50560603e-01
8.21703732e-01 -1.22196026e-01 -6.35311544e-01 2.81659842e-01
1.23300993e+00 4.29526605e-02 5.09640932e-01 4.66266394e-01
3.46034467e-01 5.33673227e-01 1.00569201e+00 1.02675498e-01
2.33720869e-01 1.04828870e+00 -9.91565511e-02 -1.71881139e-01
-3.16181183e-01 -4.40027386e-01 3.50064099e-01 7.37041950e-01
6.10336900e-01 -2.72421390e-01 -8.72731805e-01 7.40030050e-01
-1.80143631e+00 -8.78156066e-01 5.14004052e-01 2.39291716e+00
1.08152330e+00 -1.59086108e-01 -1.42263144e-01 -2.79042721e-01
6.99890852e-01 6.06071688e-02 -6.98825300e-01 -1.35315731e-01
-8.29356164e-02 3.09481919e-01 2.53324360e-01 4.16369826e-01
-8.54517043e-01 1.07721603e+00 6.00763226e+00 8.94266546e-01
-9.61523890e-01 2.55082287e-02 8.62133410e-03 -2.38250777e-01
-6.30672455e-01 1.62957907e-01 -1.01818836e+00 3.14647973e-01
7.58972287e-01 -4.65208262e-01 2.75286198e-01 1.00920570e+00
-2.88229734e-01 1.65157124e-01 -1.24525309e+00 9.07397330e-01
3.45979631e-03 -8.91644955e-01 4.20021564e-01 1.56799763e-01
5.98015904e-01 1.56521291e-01 2.71519542e-01 8.45039189e-01
4.62638795e-01 -7.74874330e-01 4.38633226e-02 2.45200559e-01
7.41456747e-01 -5.28100491e-01 6.64909482e-01 4.45676506e-01
-7.18740046e-01 -2.56434139e-02 -7.35139310e-01 4.68457997e-01
-2.28361443e-01 6.09737754e-01 -9.58950281e-01 6.47717655e-01
4.98146147e-01 6.61293507e-01 -6.44252181e-01 7.21659124e-01
-7.58908316e-02 2.72293836e-01 -2.29538113e-01 2.89730549e-01
7.25285560e-02 -4.96953726e-02 5.78868508e-01 1.05061114e+00
1.85374081e-01 -3.85328196e-02 -6.50207177e-02 6.00614488e-01
-3.31377447e-01 3.19517367e-02 -8.47549736e-01 -2.79316247e-01
9.57968116e-01 1.15098917e+00 -7.35450396e-03 -3.46578211e-01
-4.99981761e-01 9.29930866e-01 5.42270958e-01 6.30496502e-01
-3.87863010e-01 -4.46551502e-01 8.63740206e-01 9.83174443e-02
2.12061062e-01 -5.08981235e-02 7.48086721e-02 -1.57601368e+00
1.77860111e-01 -1.20037758e+00 7.74197280e-01 -5.76434374e-01
-1.53349090e+00 4.74936634e-01 1.71527520e-01 -1.25863552e+00
-5.34490824e-01 -2.17945725e-01 8.47265646e-02 1.08974540e+00
-1.87031186e+00 -8.75401735e-01 -1.92533270e-01 6.98582888e-01
5.38767636e-01 -4.23476577e-01 9.96743798e-01 6.27645135e-01
-4.80755866e-01 9.10231769e-01 6.04605794e-01 7.20051974e-02
1.34704232e+00 -1.03016973e+00 1.52732134e-01 7.19709694e-01
2.69697905e-01 1.15555882e+00 3.99331897e-01 -5.90864480e-01
-1.34229386e+00 -9.35470462e-01 1.00514603e+00 -4.79779422e-01
4.40206200e-01 -4.17699963e-01 -1.33775723e+00 5.78581274e-01
2.04476342e-02 -1.12065747e-01 9.45648611e-01 4.36762780e-01
-8.25499773e-01 -3.57420236e-01 -1.12535167e+00 5.43859839e-01
9.96581137e-01 -8.52309883e-01 -1.02811611e+00 4.59649980e-01
8.01669776e-01 -2.75934249e-01 -8.00931633e-01 4.93032008e-01
3.80659610e-01 -5.28741360e-01 1.17248988e+00 -6.61502063e-01
1.20005824e-01 -2.02846572e-01 -3.62351358e-01 -1.19277620e+00
-4.57426190e-01 -1.73456609e-01 -3.66615832e-01 1.26042676e+00
3.41763437e-01 -5.70676327e-01 5.38132966e-01 9.45422232e-01
1.06945775e-01 -3.18263918e-01 -6.84692085e-01 -8.52606356e-01
1.76410049e-01 -8.40727314e-02 7.99076080e-01 9.59504366e-01
-5.88618368e-02 6.51976824e-01 -5.48078194e-02 2.64336765e-01
4.22091603e-01 3.64701271e-01 8.16148818e-01 -1.32208669e+00
-3.34880918e-01 -1.91469505e-01 -1.75559476e-01 -1.49119079e+00
4.05306190e-01 -9.58328366e-01 -1.57491490e-01 -1.46082318e+00
3.25235695e-01 -7.69002497e-01 -4.88428116e-01 5.03603578e-01
-3.02005887e-01 -1.48662373e-01 -6.98569268e-02 8.19658041e-01
-7.81856120e-01 7.40553439e-01 1.09436011e+00 -2.91862935e-01
-2.66553670e-01 -8.06170180e-02 -9.79573846e-01 1.55361906e-01
5.38252115e-01 -6.71996891e-01 -7.77878106e-01 -8.84431243e-01
2.20850125e-01 -1.06768891e-01 9.57066417e-02 -5.58806002e-01
4.14241880e-01 8.08280781e-02 3.23003709e-01 -4.73513335e-01
2.55737662e-01 -1.00716579e+00 -1.63647402e-02 -5.39337099e-02
-6.67641878e-01 -6.18390851e-02 1.52798295e-01 7.83372641e-01
-6.66885197e-01 -2.31990337e-01 5.85728228e-01 -1.20842475e-02
-9.27878857e-01 2.57084668e-01 1.03823654e-03 1.50875419e-01
8.03355694e-01 -4.05433699e-02 -5.20831466e-01 -2.47460812e-01
-4.08099294e-01 5.01794279e-01 6.21430635e-01 7.55540490e-01
6.51070476e-01 -1.43069065e+00 -5.79626918e-01 2.72928357e-01
5.82038403e-01 2.27747381e-01 2.74123758e-01 2.11892471e-01
-6.27629310e-02 7.81121373e-01 1.21586621e-01 -4.04667675e-01
-1.12910235e+00 6.64483249e-01 9.79445428e-02 -7.71524191e-01
-4.07579869e-01 6.35546863e-01 3.55259538e-01 -6.64270818e-01
2.23050490e-01 3.14206630e-01 -2.11208194e-01 6.83925226e-02
5.45011520e-01 9.48002096e-03 2.91592807e-01 -3.67595434e-01
-3.73924345e-01 4.06634778e-01 -9.08914030e-01 -2.12195605e-01
1.25239599e+00 -2.96351373e-01 -1.28440438e-02 -9.06832591e-02
1.36616969e+00 1.06801344e-02 -8.63388777e-01 -1.06769276e+00
1.86636642e-01 -6.91566050e-01 2.15732157e-01 -9.44123924e-01
-7.94275224e-01 8.28829110e-01 5.33272088e-01 -6.83102608e-02
1.34493816e+00 1.43600568e-01 9.24187541e-01 6.24798656e-01
4.56406534e-01 -1.25382769e+00 3.04774672e-01 5.26095748e-01
7.23178327e-01 -1.33858705e+00 -9.59599912e-02 -1.11630596e-01
-6.79841876e-01 8.02110314e-01 7.44097769e-01 5.22005409e-02
4.09766138e-01 -2.64119804e-01 1.23247758e-01 -3.16600233e-01
-8.83621514e-01 -2.27236882e-01 6.93423986e-01 5.37918568e-01
2.74384856e-01 -2.59790093e-01 -2.07232460e-01 4.15579349e-01
1.76413462e-01 -2.43301585e-01 -1.09192640e-01 9.69110548e-01
-3.95525217e-01 -1.62009990e+00 -1.76251560e-01 2.83987939e-01
-2.56585777e-01 -1.59456342e-01 -6.48470521e-01 7.62669444e-01
-4.07041520e-01 8.21847379e-01 -2.10782643e-02 -4.27202523e-01
3.79798681e-01 2.50977665e-01 2.00459167e-01 -1.09548151e+00
-3.56087983e-01 -8.63236338e-02 -6.77298903e-02 -4.76369262e-01
-2.44808614e-01 -4.69962716e-01 -1.02800059e+00 -1.41260132e-01
-5.24596512e-01 5.93051791e-01 4.27880675e-01 8.82576525e-01
8.60333204e-01 5.67888133e-02 7.23484337e-01 -2.05598027e-01
-1.08355582e+00 -8.80901754e-01 -6.19228661e-01 6.19769156e-01
2.46276751e-01 -7.33780742e-01 -4.04676497e-01 -3.25419277e-01] | [11.3319673538208, 7.722308158874512] |
7e7fd3af-67a0-4448-9c23-048c50bd611b | channel-adversarial-training-for-cross | 1902.09074 | null | http://arxiv.org/abs/1902.09074v1 | http://arxiv.org/pdf/1902.09074v1.pdf | Channel adversarial training for cross-channel text-independent speaker recognition | The conventional speaker recognition frameworks (e.g., the i-vector and
CNN-based approach) have been successfully applied to various tasks when the
channel of the enrolment dataset is similar to that of the test dataset.
However, in real-world applications, mismatch always exists between these two
datasets, which may severely deteriorate the recognition performance.
Previously, a few channel compensation algorithms have been proposed, such as
Linear Discriminant Analysis (LDA) and Probabilistic LDA. However, these
methods always require the collections of different channels from a specific
speaker, which is unrealistic to be satisfied in real scenarios. Inspired by
domain adaptation, we propose a novel deep-learning based speaker recognition
framework to learn the channel-invariant and speaker-discriminative speech
representations via channel adversarial training. Specifically, we first employ
a gradient reversal layer to remove variations across different channels. Then,
the compressed information is projected into the same subspace by adversarial
training. Experiments on test datasets with 54,133 speakers demonstrate that
the proposed method is not only effective at alleviating the channel mismatch
problem, but also outperforms state-of-the-art speaker recognition methods.
Compared with the i-vector-based method and the CNN-based method, our proposed
method achieves significant relative improvement of 44.7% and 22.6%
respectively in terms of the Top1 recall. | [] | 2019-02-25 | null | null | null | null | ['text-independent-speaker-recognition'] | ['speech'] | [ 3.34498405e-01 -4.40669596e-01 1.61092151e-02 -4.80568588e-01
-1.09176219e+00 -3.78861994e-01 4.66092497e-01 -2.83183336e-01
-2.43207291e-01 5.64732313e-01 4.06790674e-01 -2.85146028e-01
2.94795036e-01 -4.68752503e-01 -8.07242155e-01 -1.01792383e+00
2.75837690e-01 -7.77806789e-02 -1.27872914e-01 -4.57354710e-02
3.27040367e-02 2.36360714e-01 -1.26257873e+00 1.22155383e-01
8.21235359e-01 1.11220920e+00 -8.37504789e-02 2.37229839e-01
-3.76663506e-01 1.11058980e-01 -7.41068661e-01 -3.64971638e-01
1.82859182e-01 -5.05840540e-01 -1.69931620e-01 1.47438049e-01
2.53122151e-01 -2.96276718e-01 -8.58638465e-01 1.17611206e+00
9.33030605e-01 -5.38615622e-02 4.95925099e-01 -1.16514206e+00
-5.44218242e-01 5.55830479e-01 -6.72410250e-01 1.07027031e-01
3.79582167e-01 -1.29836917e-01 5.45756757e-01 -9.23448443e-01
7.36577529e-03 1.36451757e+00 5.77331901e-01 7.51694143e-01
-9.70570862e-01 -1.19197977e+00 2.55739748e-01 3.32976878e-01
-1.63419998e+00 -8.45842540e-01 1.11611247e+00 -1.97922453e-01
3.76027316e-01 1.53578922e-01 2.46616215e-01 1.33485794e+00
-1.97928652e-01 9.64820862e-01 1.17261279e+00 -4.70054269e-01
1.19348153e-01 4.03761953e-01 -1.88686714e-01 1.79610685e-01
-1.50775179e-01 8.92508924e-02 -7.96478033e-01 -2.07027853e-01
4.63968873e-01 7.72362053e-02 -5.54104030e-01 -2.13456422e-01
-1.19509387e+00 7.24902511e-01 2.89776027e-01 2.64360517e-01
-2.82334417e-01 -3.52153271e-01 5.25169790e-01 2.74857044e-01
2.13476524e-01 -4.10411626e-01 -3.30213428e-01 5.99421374e-02
-8.75961602e-01 1.33674070e-01 7.80924261e-01 8.58636141e-01
4.30095673e-01 5.19542277e-01 -8.24393928e-02 9.94889379e-01
6.00819111e-01 8.42052162e-01 8.32923770e-01 -6.09848052e-02
8.58921349e-01 1.34111166e-01 -4.77394760e-02 -1.08666086e+00
-1.30706787e-01 -8.94071341e-01 -1.07643735e+00 -4.35849279e-01
3.26736391e-01 -2.41017759e-01 -8.54322016e-01 1.85110509e+00
2.26596490e-01 6.26676679e-01 4.88709003e-01 9.51490223e-01
8.69966030e-01 8.17848265e-01 2.17819232e-02 -2.86331505e-01
1.05860901e+00 -7.00133979e-01 -8.25880170e-01 -2.46055126e-01
1.66794971e-01 -1.02445781e+00 9.20156837e-01 3.33837658e-01
-5.75787365e-01 -5.64458430e-01 -1.28203821e+00 6.60490990e-01
-1.24064393e-01 3.32201943e-02 3.91694009e-01 1.21768594e+00
-7.18141794e-01 -7.12348744e-02 -4.92758811e-01 -8.34503993e-02
4.75382864e-01 3.26310098e-01 -3.97556812e-01 -3.81493568e-01
-1.38837218e+00 2.54013360e-01 1.14481814e-01 4.01563674e-01
-1.15331399e+00 -2.60377407e-01 -6.46862388e-01 3.13553922e-02
8.95464644e-02 -5.84035218e-02 1.07379150e+00 -9.75417554e-01
-1.67355883e+00 2.56861418e-01 -3.69340271e-01 -3.58150929e-01
3.96627873e-01 -8.59888643e-02 -9.60201919e-01 -1.98995695e-01
-2.34058842e-01 1.74833342e-01 1.04612529e+00 -1.16720068e+00
-4.18307930e-01 -5.86742759e-01 -3.13995302e-01 3.64403188e-01
-7.85218716e-01 -9.40129347e-03 -5.81592798e-01 -6.58385158e-01
5.82298100e-01 -8.96078467e-01 1.49976075e-01 -2.82593906e-01
-6.85113251e-01 2.17374131e-01 1.03644621e+00 -9.83083725e-01
1.02001381e+00 -2.47636938e+00 -5.86506799e-02 4.30195719e-01
-3.05874914e-01 4.00451005e-01 -7.60674104e-02 3.48250359e-01
-5.24622910e-02 -1.38992354e-01 -3.43040377e-01 -2.77519107e-01
-2.03609392e-02 -3.74622047e-02 -3.70037854e-01 6.61425650e-01
-2.15398058e-01 3.77354175e-01 -3.83137852e-01 -2.91305870e-01
1.96863592e-01 9.59992707e-01 -4.28077340e-01 2.97787875e-01
2.95256317e-01 8.30676615e-01 -4.14027184e-01 7.74760604e-01
1.16557980e+00 3.51384789e-01 1.13232017e-01 -6.30453229e-02
2.52771527e-01 2.51218349e-01 -1.34876549e+00 1.72808731e+00
-6.50051415e-01 5.07795334e-01 2.12758616e-01 -1.20366299e+00
1.09290135e+00 6.54271483e-01 1.38121665e-01 -7.39086628e-01
-7.50070214e-02 3.55911881e-01 1.04458764e-01 -3.70371014e-01
2.41124444e-02 -3.75207394e-01 -1.07859910e-01 1.36305943e-01
-1.38497129e-01 3.41538459e-01 -6.82921112e-01 -3.08248252e-02
7.09883213e-01 -3.34717661e-01 -2.26140302e-02 9.55316871e-02
1.06683886e+00 -7.43725955e-01 8.00157905e-01 5.24600983e-01
-3.40639025e-01 5.28996766e-01 1.74482033e-01 -1.13062914e-02
-7.25709498e-01 -9.88684773e-01 -3.27642441e-01 7.72523046e-01
1.77608788e-01 1.15861982e-01 -9.02046859e-01 -5.34611523e-01
2.91178860e-02 4.29522216e-01 -2.03689858e-01 -2.59176314e-01
-4.17076916e-01 -8.26803923e-01 9.32683229e-01 4.14492399e-01
1.06806028e+00 -6.08658433e-01 2.20701560e-01 3.21714729e-01
-2.33340472e-01 -1.12882149e+00 -6.42302334e-01 -1.75005063e-01
-7.25519657e-01 -5.83155751e-01 -9.44004238e-01 -9.79377627e-01
5.67985356e-01 4.38497156e-01 3.80495697e-01 -2.03672186e-01
1.16318025e-01 1.30417913e-01 -4.25323308e-01 -2.24546283e-01
-2.68066108e-01 -2.27573514e-02 2.20749944e-01 7.26502061e-01
6.40776813e-01 -6.22741520e-01 -6.55847967e-01 4.82112288e-01
-8.26339364e-01 -4.88287270e-01 8.25417340e-01 1.17552590e+00
3.30536515e-01 2.50024140e-01 8.81439805e-01 -5.94707906e-01
5.51979661e-01 -3.70994478e-01 -2.31873631e-01 6.10679090e-02
-4.54201460e-01 -1.37917653e-01 6.73526227e-01 -5.43814719e-01
-1.22095108e+00 2.83620477e-01 -3.74510348e-01 -3.48584384e-01
-3.31774324e-01 5.56675673e-01 -9.23862278e-01 -2.21563637e-01
3.48790288e-01 9.11967874e-01 -4.07078341e-02 -5.68589568e-01
6.47458062e-02 1.42714214e+00 5.93290329e-01 -3.37522775e-01
9.51465607e-01 2.33532414e-01 -4.92550641e-01 -9.02765632e-01
-2.81125367e-01 -4.95702535e-01 -3.01923782e-01 2.47877445e-02
4.93645877e-01 -1.34892142e+00 -5.06721675e-01 8.74463260e-01
-1.04806840e+00 2.41504461e-01 4.53247935e-01 7.70720482e-01
-1.03898473e-01 5.20570993e-01 -3.31744164e-01 -8.77548516e-01
-2.85148323e-01 -1.43896353e+00 8.40850234e-01 3.73163670e-01
3.88627142e-01 -7.16710925e-01 -2.68773824e-01 4.98247445e-01
7.13020563e-01 -1.15558624e-01 7.40002811e-01 -9.91578639e-01
-4.06525552e-01 -5.14148474e-01 -9.27132089e-03 5.44470608e-01
1.10260725e-01 -5.67617297e-01 -1.39198506e+00 -5.30229449e-01
9.56229046e-02 2.40956116e-02 6.91412628e-01 2.25345999e-01
1.11127937e+00 -3.54011387e-01 -3.19943070e-01 6.10465646e-01
1.19431913e+00 5.42349935e-01 8.05465639e-01 1.22894965e-01
6.32306814e-01 4.19290125e-01 3.03742945e-01 5.04066586e-01
2.39172190e-01 9.19918716e-01 1.95485845e-01 -3.04829516e-02
3.92917097e-02 -4.64189947e-01 4.98794615e-01 1.01422429e+00
4.21009272e-01 -1.75263569e-01 -7.40049541e-01 4.59348202e-01
-1.44334292e+00 -8.16712916e-01 3.17776918e-01 2.52850223e+00
7.32221186e-01 1.57592148e-01 1.57900918e-02 5.05937278e-01
1.11504483e+00 2.73746401e-01 -6.80485845e-01 -1.65960133e-01
-2.42968991e-01 4.01673913e-02 4.95901167e-01 2.88246632e-01
-1.14464962e+00 6.98413849e-01 5.48573112e+00 8.95517051e-01
-1.59531057e+00 3.64818871e-01 5.92195928e-01 1.08180769e-01
-1.91953093e-01 -1.62000328e-01 -8.07201385e-01 6.44053876e-01
9.36590850e-01 -6.39204681e-02 5.05745649e-01 8.16796124e-01
7.32257366e-02 5.54035664e-01 -7.52897024e-01 1.42140162e+00
4.59329814e-01 -7.36730516e-01 7.69102620e-03 1.23845011e-01
5.28379917e-01 -3.30919176e-01 4.32303011e-01 4.65092659e-01
-4.69247699e-01 -1.01293123e+00 6.72391355e-01 1.78271800e-01
7.84778714e-01 -9.65172291e-01 1.03510749e+00 2.47051284e-01
-1.09701073e+00 -1.07145250e-01 -3.24452043e-01 2.02589586e-01
-5.77536896e-02 6.50290191e-01 -7.43313849e-01 4.87005860e-01
6.23404682e-01 4.90744084e-01 -1.82472080e-01 1.02449691e+00
-3.87716629e-02 8.78394842e-01 -1.82493553e-01 2.96803825e-02
-4.51415172e-03 1.77671582e-01 6.56014502e-01 1.13853693e+00
3.83027434e-01 -1.86973542e-01 7.93853961e-03 3.25768501e-01
-3.75427008e-01 4.47107255e-01 -5.96022606e-01 6.94555864e-02
8.63269687e-01 8.81681979e-01 -9.59070697e-02 -1.83596894e-01
-6.60548329e-01 1.01444316e+00 -2.38600433e-01 5.86143374e-01
-8.22307348e-01 -5.66708446e-01 5.72059989e-01 -5.01274057e-02
4.93811965e-01 -8.01263675e-02 -1.35534659e-01 -1.21908104e+00
3.02274674e-01 -1.25365841e+00 -1.27966783e-03 -2.14722753e-01
-1.21072352e+00 6.64782465e-01 -4.71354693e-01 -1.56193388e+00
-1.24131158e-01 -2.24133983e-01 -6.56970501e-01 1.12112355e+00
-1.62018335e+00 -1.14663208e+00 -3.04874748e-01 9.42181408e-01
5.47979474e-01 -7.83368051e-01 8.85915935e-01 7.60604680e-01
-7.74369717e-01 1.29322815e+00 6.04437292e-01 2.87215918e-01
8.44014585e-01 -7.03349888e-01 1.41488910e-01 8.82374525e-01
4.21319902e-02 7.00109541e-01 6.02140009e-01 -2.54947782e-01
-1.66884089e+00 -1.05575669e+00 6.12927616e-01 2.64071822e-01
1.69072941e-01 -4.48182374e-01 -9.67043757e-01 5.08936167e-01
1.40068561e-01 -8.24730918e-02 9.42604899e-01 5.45244012e-03
-7.15935946e-01 -5.87401330e-01 -1.17001724e+00 2.10880175e-01
6.30880713e-01 -8.23300958e-01 -2.44731858e-01 2.47883961e-01
4.79566336e-01 -5.02282023e-01 -6.39856160e-01 2.12920099e-01
6.83557749e-01 -7.89712250e-01 9.78628874e-01 -3.95160615e-01
-2.81699095e-02 -3.18297356e-01 -6.54134989e-01 -1.33570707e+00
1.81457385e-01 -4.34823185e-01 1.98975429e-01 1.58154047e+00
5.07815123e-01 -9.13179874e-01 7.38999128e-01 4.76104617e-01
-6.03884943e-02 -3.66817594e-01 -1.33577454e+00 -8.21768224e-01
1.15673892e-01 -3.62250984e-01 9.07785296e-01 8.73884320e-01
-1.48264751e-01 1.09186687e-01 -6.14229918e-01 4.65969741e-01
6.63545609e-01 -1.60704870e-02 9.09238219e-01 -9.62247014e-01
-3.90035808e-01 -1.67446494e-01 -6.75740778e-01 -1.31140149e+00
3.38905066e-01 -6.94627047e-01 2.74348825e-01 -1.04550207e+00
1.15775034e-01 -5.40759683e-01 -5.82563519e-01 1.77623406e-01
2.78662803e-04 6.32525980e-02 -3.78293321e-02 3.76032501e-01
-1.03934817e-01 7.49913752e-01 8.65550578e-01 -5.21947026e-01
-3.52029085e-01 3.16864014e-01 -8.85798097e-01 3.78819674e-01
8.07298839e-01 -4.17337507e-01 -1.61432058e-01 -3.81243974e-01
-5.58768332e-01 2.70690233e-01 1.91647589e-01 -1.14373457e+00
3.36681455e-01 2.74620801e-01 4.78026897e-01 -3.67063940e-01
4.40842628e-01 -8.43191564e-01 -1.45182312e-02 3.42023760e-01
-3.66669595e-01 -3.66911322e-01 3.09765279e-01 9.49598610e-01
-6.95066869e-01 8.15670565e-02 6.72975242e-01 2.45457187e-01
-5.46761215e-01 2.69900113e-01 -1.53327852e-01 -2.82008976e-01
7.58204579e-01 2.65549738e-02 -1.40255034e-01 -7.14004874e-01
-5.77553451e-01 -1.86645359e-01 6.91573843e-02 4.12238806e-01
7.24548101e-01 -1.38503635e+00 -8.96402597e-01 5.45867980e-01
1.29168883e-01 -2.47352168e-01 6.42451227e-01 7.32971191e-01
-9.70072448e-02 6.29858673e-01 -1.65197868e-02 -6.61591589e-01
-1.22743428e+00 3.20741385e-01 2.38487616e-01 1.71290055e-01
-2.85656840e-01 7.69976437e-01 2.40022451e-01 -5.09775937e-01
6.88848734e-01 1.50335014e-01 -7.28128031e-02 -2.47130781e-01
6.81591392e-01 4.18611802e-02 2.64493376e-01 -9.95712817e-01
-6.32090807e-01 5.71414590e-01 -3.07042986e-01 -3.09695393e-01
1.00484335e+00 -2.47523233e-01 2.98483461e-01 7.82018155e-02
1.48579717e+00 2.67714947e-01 -1.03925419e+00 -6.72349811e-01
-4.79282528e-01 -5.75625479e-01 2.56322652e-01 -7.07777262e-01
-1.37574923e+00 1.27483070e+00 1.04495120e+00 -2.68224347e-02
1.19864440e+00 -2.99527138e-01 1.00603139e+00 1.70322597e-01
2.50080884e-01 -8.38470042e-01 -2.43638396e-01 8.74561593e-02
8.26341629e-01 -1.34411728e+00 -2.88751364e-01 -3.22773576e-01
-6.27704620e-01 8.92365158e-01 3.98351401e-01 3.03176492e-01
8.71099412e-01 -1.10137239e-01 2.93249547e-01 4.40977693e-01
-2.96096474e-01 1.26169026e-01 8.22679326e-02 7.61076391e-01
4.22636420e-01 2.85142750e-01 -1.29146501e-01 7.30652869e-01
-8.96150321e-02 -3.12793434e-01 2.20024273e-01 7.47137785e-01
-2.57299662e-01 -1.12189257e+00 -7.77927995e-01 2.83702105e-01
-4.98963088e-01 -1.27428696e-01 -1.39503658e-01 4.27370995e-01
-1.24996826e-01 1.30711126e+00 -2.05289304e-01 -7.78450787e-01
4.36041534e-01 3.12445283e-01 7.97514021e-02 -1.89318970e-01
-2.41106287e-01 4.39121902e-01 -1.72784880e-01 -1.66412205e-01
-3.70929688e-01 -8.92583549e-01 -9.70802188e-01 -2.99340725e-01
-5.18442571e-01 2.37931609e-01 1.05670500e+00 7.96480656e-01
3.78783762e-01 3.83523554e-01 1.03491187e+00 -6.25928998e-01
-8.57051730e-01 -1.14191651e+00 -7.95791447e-01 3.36970687e-01
3.37045997e-01 -6.73916698e-01 -5.84075868e-01 5.27946558e-03] | [14.435523986816406, 6.015368461608887] |
97f26817-ef39-462e-9bd1-51ab6d073a8d | boosting-contrastive-self-supervised-learning | 2011.11765 | null | https://arxiv.org/abs/2011.11765v2 | https://arxiv.org/pdf/2011.11765v2.pdf | Boosting Contrastive Self-Supervised Learning with False Negative Cancellation | Self-supervised representation learning has made significant leaps fueled by progress in contrastive learning, which seeks to learn transformations that embed positive input pairs nearby, while pushing negative pairs far apart. While positive pairs can be generated reliably (e.g., as different views of the same image), it is difficult to accurately establish negative pairs, defined as samples from different images regardless of their semantic content or visual features. A fundamental problem in contrastive learning is mitigating the effects of false negatives. Contrasting false negatives induces two critical issues in representation learning: discarding semantic information and slow convergence. In this paper, we propose novel approaches to identify false negatives, as well as two strategies to mitigate their effect, i.e. false negative elimination and attraction, while systematically performing rigorous evaluations to study this problem in detail. Our method exhibits consistent improvements over existing contrastive learning-based methods. Without labels, we identify false negatives with 40% accuracy among 1000 semantic classes on ImageNet, and achieve 5.8% absolute improvement in top-1 accuracy over the previous state-of-the-art when finetuning with 1% labels. Our code is available at https://github.com/google-research/fnc. | ['Maryam Khademi', 'Michael Maire', 'Matthew R. Walter', 'Simon Kornblith', 'Tri Huynh'] | 2020-11-23 | null | null | null | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 4.56097454e-01 9.79988575e-02 -3.38298261e-01 -4.09208417e-01
-1.15753794e+00 -7.82142878e-01 8.19498360e-01 1.20969906e-01
-5.02107739e-01 7.98860431e-01 1.41765952e-01 -4.93578687e-02
1.08454637e-01 -6.53875411e-01 -9.26720202e-01 -5.38420200e-01
-1.31744236e-01 3.66384625e-01 1.62497938e-01 -1.23108037e-01
2.85785735e-01 2.79759884e-01 -1.76013958e+00 6.08007371e-01
7.82307029e-01 9.03202057e-01 2.07196511e-02 3.80783081e-01
-1.85833294e-02 9.35516119e-01 -7.76095033e-01 -1.50319368e-01
5.29652059e-01 -5.45330763e-01 -8.25105965e-01 -5.27528897e-02
1.04245472e+00 -2.89883435e-01 -6.20965660e-02 1.31416738e+00
3.02368522e-01 2.59272121e-02 6.85967028e-01 -1.56592810e+00
-9.55044925e-01 3.29786539e-01 -8.41594577e-01 3.13900143e-01
1.59707293e-01 3.37498009e-01 1.17560828e+00 -1.29896498e+00
5.80249190e-01 1.27049935e+00 5.98135352e-01 7.96440780e-01
-1.46616364e+00 -1.10734665e+00 1.44761056e-01 1.81985483e-01
-1.42201853e+00 -6.00104868e-01 6.66396320e-01 -4.75654334e-01
8.62883508e-01 3.59128118e-01 6.90206110e-01 1.10339141e+00
-1.14620678e-01 8.10236812e-01 1.26455235e+00 -4.63591933e-01
1.64783254e-01 3.19529504e-01 1.17882848e-01 6.00446641e-01
3.99003595e-01 2.64603436e-01 -5.56389511e-01 -1.36590466e-01
7.76779890e-01 5.17556220e-02 -2.20960036e-01 -5.74786007e-01
-1.12518382e+00 8.00281763e-01 7.36701548e-01 8.64922479e-02
-4.90097404e-02 2.57470965e-01 2.92860180e-01 3.85138661e-01
6.57378912e-01 7.70011842e-01 -3.38329583e-01 8.75024050e-02
-8.77352238e-01 2.52305329e-01 4.37414765e-01 8.84247780e-01
9.87652183e-01 4.07612026e-02 3.74572128e-02 9.02022302e-01
6.24007508e-02 4.09477681e-01 6.08820856e-01 -8.12599421e-01
4.53339696e-01 6.23160779e-01 1.46335559e-02 -1.07773542e+00
-1.62091643e-01 -4.97533262e-01 -7.20640481e-01 4.91685450e-01
4.83953297e-01 4.13119495e-02 -1.11149931e+00 1.97033286e+00
2.28556823e-02 3.80230755e-01 -1.58908609e-02 8.43518674e-01
8.35908234e-01 4.75648910e-01 1.76772401e-01 2.57441700e-02
1.05579603e+00 -1.02331829e+00 -3.90843660e-01 -4.54240084e-01
7.69453228e-01 -7.91990757e-01 1.37250352e+00 3.09207946e-01
-9.72929895e-01 -4.49410498e-01 -1.12193501e+00 -1.79308858e-02
-4.29075927e-01 3.13046388e-02 5.15007675e-01 4.02499139e-01
-1.13768137e+00 5.39711058e-01 -6.28276229e-01 -1.53611079e-01
7.86916018e-01 2.38129362e-01 -4.98220742e-01 -1.25591502e-01
-1.01336479e+00 9.02612031e-01 2.85997868e-01 -2.18276531e-01
-8.55009556e-01 -9.88853633e-01 -7.92847157e-01 -7.33582824e-02
2.84259468e-01 -3.38077575e-01 1.18801701e+00 -1.63066661e+00
-7.76988745e-01 1.24604249e+00 -1.67698085e-01 -4.17184442e-01
7.44365335e-01 -4.71628189e-01 -4.06807005e-01 -2.27350984e-02
3.29277158e-01 1.22581899e+00 8.49862695e-01 -1.60323095e+00
-5.98280132e-01 -3.09224337e-01 -2.27296408e-02 4.85660166e-01
-4.74488497e-01 -1.70991212e-01 -1.98544040e-01 -7.91276276e-01
2.02921137e-01 -9.64532793e-01 -2.01939404e-01 4.45259303e-01
-1.66870907e-01 -1.94256142e-01 7.68599153e-01 -3.21696520e-01
6.50413036e-01 -2.23881960e+00 -3.40581894e-01 1.77016944e-01
4.71267581e-01 3.90373856e-01 -4.02867198e-01 -5.70595041e-02
-4.25359607e-01 3.23076695e-01 -2.41227120e-01 -2.00127497e-01
-3.30008298e-01 8.65780190e-03 -4.60182041e-01 4.50702339e-01
6.29005313e-01 8.95912468e-01 -1.32673955e+00 -3.81025881e-01
2.74040788e-01 4.22576547e-01 -5.53356886e-01 -8.51103738e-02
9.58732218e-02 2.19372109e-01 -1.87111013e-02 4.98684496e-01
7.32049763e-01 -4.94636625e-01 1.64799228e-01 -3.99546146e-01
1.51225016e-01 3.46965641e-01 -1.12841964e+00 1.32762516e+00
-3.28136206e-01 7.93061078e-01 -3.07554394e-01 -1.16274345e+00
9.18887913e-01 -5.75909428e-02 3.07066411e-01 -9.71245706e-01
-9.39644501e-02 3.03240597e-01 -1.14924967e-01 -1.46274015e-01
6.06469214e-01 -2.44207039e-01 2.63733298e-01 3.67911756e-01
1.36585817e-01 -1.66082814e-01 6.40643612e-02 2.37650529e-01
9.78059232e-01 2.44175568e-02 3.41848880e-01 -4.96091157e-01
1.40740916e-01 1.68637574e-01 6.07126594e-01 8.31309736e-01
-4.41882581e-01 8.79524648e-01 3.61477703e-01 -4.74443763e-01
-8.14099967e-01 -1.44589972e+00 -1.00884162e-01 1.05948317e+00
4.33037907e-01 -4.33720708e-01 -4.85433280e-01 -9.91673529e-01
1.12564184e-01 6.60207868e-01 -8.52715731e-01 -4.96574789e-01
-4.92506742e-01 -7.08212018e-01 5.55970013e-01 6.30763948e-01
5.31841159e-01 -9.01829541e-01 -4.08064365e-01 -1.97947741e-01
-1.41795620e-01 -7.74790108e-01 -2.33244330e-01 2.56419152e-01
-8.50197315e-01 -1.29451704e+00 -5.86461961e-01 -9.20360506e-01
9.23662722e-01 7.22914636e-01 1.54132187e+00 6.68473318e-02
-3.40458393e-01 2.24950016e-01 -1.86679408e-01 -4.56831634e-01
-3.96591485e-01 -2.14383602e-01 -4.68404330e-02 -2.21047521e-01
4.75603551e-01 -3.79110545e-01 -7.08685577e-01 3.51434171e-01
-9.38873947e-01 2.14027986e-01 5.70333719e-01 1.02425766e+00
7.28893280e-01 -2.61345088e-01 7.52949119e-01 -1.17592394e+00
4.91199553e-01 -6.23802722e-01 -3.35983574e-01 3.12896609e-01
-7.04234123e-01 -4.51026037e-02 6.43057466e-01 -4.82526630e-01
-7.97613978e-01 1.55870663e-02 2.61305183e-01 -5.27653813e-01
-5.96305244e-02 8.59585628e-02 2.27072239e-01 1.71690586e-03
1.22559845e+00 7.40744844e-02 1.47299618e-01 -8.45609531e-02
4.17194098e-01 4.38753575e-01 5.21350741e-01 -5.51510394e-01
7.71385252e-01 6.18274570e-01 -4.55355465e-01 -6.92257047e-01
-1.04417813e+00 -5.51398218e-01 -3.68844002e-01 -2.62000710e-01
3.68955225e-01 -1.20186937e+00 -4.57724705e-02 2.44950011e-01
-7.73175180e-01 -3.69370252e-01 -5.66560268e-01 3.37309867e-01
-4.30033267e-01 7.16169327e-02 -3.75268072e-01 -4.37176853e-01
-2.40570694e-01 -9.68759656e-01 9.10654545e-01 1.67254806e-01
-5.96632063e-01 -8.58210802e-01 -4.90241349e-02 2.50424027e-01
3.46881151e-01 5.26517667e-02 5.85015774e-01 -5.92572868e-01
-3.83666039e-01 1.85361281e-02 -4.87298429e-01 6.98029220e-01
4.08091486e-01 -7.87290633e-02 -1.19073248e+00 -5.37117422e-01
-2.25147471e-01 -7.12011397e-01 1.08838820e+00 2.11497724e-01
1.25303698e+00 -4.06396329e-01 -2.85715014e-01 5.06385386e-01
1.41964281e+00 -5.49332313e-02 6.87458932e-01 2.76273221e-01
7.15477526e-01 5.85212171e-01 6.90296412e-01 2.81543106e-01
1.10858284e-01 4.40539569e-01 3.68117929e-01 -4.26244974e-01
-4.52856332e-01 -3.73860180e-01 3.29620279e-02 3.27093989e-01
1.46118358e-01 -2.43734661e-02 -1.03537703e+00 6.96278214e-01
-1.76675153e+00 -9.18737173e-01 1.11523807e-01 2.37540245e+00
1.00979173e+00 3.42067897e-01 -4.17853147e-02 8.59945491e-02
8.90435040e-01 2.01415017e-01 -7.00234056e-01 1.23218752e-01
-2.81054944e-01 3.89484525e-01 5.29700577e-01 4.36620533e-01
-1.18462336e+00 1.01151919e+00 6.27676105e+00 7.09727943e-01
-1.29968429e+00 -1.56991475e-03 9.91982639e-01 -3.82796824e-01
-5.13669610e-01 -6.13734722e-02 -5.72055042e-01 3.58951688e-01
3.98813009e-01 -1.15849413e-01 2.43316367e-01 9.98074114e-01
-7.12235644e-03 -2.05179229e-02 -1.17700005e+00 1.22068667e+00
3.26965183e-01 -1.59291506e+00 2.96243668e-01 -3.14522773e-01
1.07795584e+00 2.49662369e-01 4.21895683e-01 2.74168104e-01
5.11084080e-01 -1.10442734e+00 8.63439679e-01 1.31715864e-01
8.98976445e-01 -6.29354417e-01 4.52844411e-01 -9.60004795e-03
-9.25405622e-01 8.19983631e-02 -2.61002004e-01 -2.37281695e-01
-4.24783707e-01 7.02222347e-01 -7.68806398e-01 -5.07003926e-02
9.55449045e-01 1.00998020e+00 -7.71120667e-01 8.28640997e-01
-3.10518712e-01 6.51875496e-01 -1.58301413e-01 1.48592750e-02
1.67156681e-01 1.13187321e-01 2.74515599e-01 1.10806870e+00
7.40233585e-02 -2.35281572e-01 9.02412310e-02 1.00040579e+00
-3.82403463e-01 -3.06336815e-03 -8.14646602e-01 2.09494740e-01
7.65814006e-01 1.15447032e+00 -8.40780497e-01 -5.81041753e-01
-3.78570497e-01 8.64465892e-01 6.39127851e-01 4.14239347e-01
-7.89484084e-01 -1.48801431e-01 7.39396274e-01 2.20897242e-01
-2.08661202e-02 6.03212491e-02 -5.47361195e-01 -1.19305062e+00
2.33646974e-01 -9.92549360e-01 4.47389275e-01 -6.19972348e-01
-1.44650424e+00 3.79730672e-01 -1.24110267e-01 -1.50074756e+00
-8.88633579e-02 -4.63937998e-01 -5.25440276e-01 6.26267433e-01
-1.40840471e+00 -9.27163064e-01 -4.70220149e-01 4.58948463e-01
5.28144538e-01 -1.89184416e-02 7.80991375e-01 2.35112846e-01
-2.38378555e-01 9.04174685e-01 1.48105770e-01 1.07729726e-01
1.08464205e+00 -1.25966406e+00 4.87918317e-01 7.23666131e-01
3.40675384e-01 5.82946777e-01 4.51500505e-01 -4.82175291e-01
-9.09950495e-01 -1.19818699e+00 6.88909709e-01 -4.46200699e-01
5.34928799e-01 -3.92086744e-01 -1.01776803e+00 7.89264202e-01
-1.60121098e-01 4.65214014e-01 5.76979876e-01 1.28573418e-01
-9.23918188e-01 -9.36491862e-02 -1.25513566e+00 7.95135796e-01
1.21910775e+00 -5.57306468e-01 -4.50343311e-01 3.74644697e-01
4.47626650e-01 -2.86761582e-01 -3.42572898e-01 4.73681897e-01
5.43559611e-01 -1.05658710e+00 1.10990894e+00 -5.24007738e-01
6.74205899e-01 -2.62971103e-01 -1.15566313e-01 -1.43769538e+00
-3.52401584e-01 -2.40212992e-01 1.60281181e-01 1.00044167e+00
5.44103205e-01 -7.93032646e-01 1.01620233e+00 4.42988098e-01
2.09184084e-02 -7.28436887e-01 -7.03755677e-01 -1.06839991e+00
2.56056696e-01 -2.44948626e-01 2.97774523e-01 1.35252512e+00
-7.05761313e-02 5.23982167e-01 -2.17785403e-01 -7.54863322e-02
6.74414754e-01 -1.40484601e-01 7.19471216e-01 -9.99023676e-01
-5.98344989e-02 -5.87076545e-01 -5.83334804e-01 -8.93508971e-01
1.32710621e-01 -9.80687201e-01 1.54347628e-01 -1.27454865e+00
4.43843544e-01 -6.99438870e-01 -5.87632596e-01 6.47706926e-01
-4.30782139e-01 7.64231563e-01 1.66040137e-01 4.42914069e-01
-6.75480843e-01 5.07553101e-01 1.05040467e+00 -3.24105054e-01
-9.43148881e-02 -4.12140518e-01 -9.27363813e-01 8.13191593e-01
1.13805437e+00 -5.06948054e-01 -6.98807776e-01 -5.16226411e-01
1.37947619e-01 -4.94664967e-01 6.45537138e-01 -1.09214282e+00
-2.73877922e-02 -1.37919590e-01 7.18006730e-01 -4.24641967e-01
2.90928096e-01 -4.56907809e-01 -1.75930694e-01 5.62148988e-01
-8.16635787e-01 2.28531957e-01 2.72306472e-01 5.14811873e-01
-3.14635217e-01 -7.15552494e-02 1.03626895e+00 -2.80179799e-01
-9.85597312e-01 2.43938249e-02 -1.02854865e-02 5.44278562e-01
9.75921392e-01 -2.63900369e-01 -7.00844944e-01 -3.15543085e-01
-4.26611006e-01 8.51471499e-02 5.58688223e-01 6.62384391e-01
7.82963216e-01 -1.45451677e+00 -8.04638207e-01 3.48118842e-01
5.28794229e-01 -5.33407144e-02 2.00225800e-01 4.43121165e-01
-3.05854470e-01 -7.97343180e-02 -3.40948671e-01 -7.06053495e-01
-1.45469213e+00 1.97062954e-01 3.57132167e-01 -5.14428280e-02
-4.44622248e-01 1.17201054e+00 6.26573741e-01 -4.64483798e-01
2.29892865e-01 -1.01775929e-01 6.71176612e-03 6.45950611e-04
6.57190740e-01 2.87332952e-01 3.10699660e-02 -3.73873532e-01
-4.38385010e-01 3.33818108e-01 -5.34738064e-01 -9.98706892e-02
1.23222136e+00 2.28120774e-01 1.06579013e-01 4.30512816e-01
1.49562144e+00 -2.42804766e-01 -1.43546021e+00 -2.93683499e-01
-8.45090523e-02 -7.19247580e-01 -4.63013686e-02 -8.70059848e-01
-1.06262136e+00 6.64047658e-01 9.04648602e-01 1.41999617e-01
9.76043105e-01 1.55752286e-01 4.62076426e-01 2.85994083e-01
1.63410306e-01 -1.15956712e+00 5.83004832e-01 3.49126458e-01
9.53084052e-01 -1.71021819e+00 3.15620825e-02 -5.63699841e-01
-5.06795585e-01 7.26745427e-01 9.96475875e-01 -4.80788857e-01
5.28474331e-01 1.94433063e-01 2.74568975e-01 -2.08752364e-01
-7.82733679e-01 -3.92878912e-02 2.10218504e-01 7.44388103e-01
6.50930405e-01 3.12949806e-01 -1.51946753e-01 4.21306938e-02
-2.57909268e-01 -1.19715832e-01 4.14490581e-01 1.07280290e+00
-2.64059484e-01 -7.72494495e-01 -2.15308174e-01 7.93857872e-01
-2.74521828e-01 -3.31833959e-01 -5.77105880e-01 9.41148937e-01
-7.10747093e-02 7.46132731e-01 4.79277968e-01 -5.12863815e-01
2.63579696e-01 -1.83673933e-01 6.02309227e-01 -7.52114475e-01
-4.23220545e-01 -4.06705409e-01 1.53875183e-02 -5.83501995e-01
-4.41531271e-01 -6.07957304e-01 -1.20362973e+00 -1.98280901e-01
-2.78726280e-01 -6.03316315e-02 3.24590206e-01 6.76284313e-01
3.97262692e-01 3.28632951e-01 6.73240542e-01 -6.72798574e-01
-7.47992992e-01 -7.62754440e-01 -3.47570956e-01 9.04388785e-01
3.80953908e-01 -7.00376153e-01 -6.09293938e-01 -1.44522416e-03] | [9.576974868774414, 2.729806423187256] |
3d747f3f-0838-4f8f-ba02-af0867b768fb | semantic-validation-in-structure-from-motion | 2304.02420 | null | https://arxiv.org/abs/2304.02420v1 | https://arxiv.org/pdf/2304.02420v1.pdf | Semantic Validation in Structure from Motion | The Structure from Motion (SfM) challenge in computer vision is the process of recovering the 3D structure of a scene from a series of projective measurements that are calculated from a collection of 2D images, taken from different perspectives. SfM consists of three main steps; feature detection and matching, camera motion estimation, and recovery of 3D structure from estimated intrinsic and extrinsic parameters and features. A problem encountered in SfM is that scenes lacking texture or with repetitive features can cause erroneous feature matching between frames. Semantic segmentation offers a route to validate and correct SfM models by labelling pixels in the input images with the use of a deep convolutional neural network. The semantic and geometric properties associated with classes in the scene can be taken advantage of to apply prior constraints to each class of object. The SfM pipeline COLMAP and semantic segmentation pipeline DeepLab were used. This, along with planar reconstruction of the dense model, were used to determine erroneous points that may be occluded from the calculated camera position, given the semantic label, and thus prior constraint of the reconstructed plane. Herein, semantic segmentation is integrated into SfM to apply priors on the 3D point cloud, given the object detection in the 2D input images. Additionally, the semantic labels of matched keypoints are compared and inconsistent semantically labelled points discarded. Furthermore, semantic labels on input images are used for the removal of objects associated with motion in the output SfM models. The proposed approach is evaluated on a data-set of 1102 images of a repetitive architecture scene. This project offers a novel method for improved validation of 3D SfM models. | ['Joseph Rowell'] | 2023-04-05 | null | null | null | null | ['motion-estimation'] | ['computer-vision'] | [ 5.83228946e-01 1.61639169e-01 2.80168235e-01 -4.96957332e-01
-7.44599402e-01 -5.52037716e-01 4.56867486e-01 8.29408988e-02
-5.11757970e-01 8.98480490e-02 -5.72904229e-01 1.59926727e-01
-1.66306347e-01 -6.47610784e-01 -1.00582051e+00 -5.35569668e-01
3.28528643e-01 1.11820686e+00 6.59166753e-01 1.74795702e-01
6.46812677e-01 1.09765971e+00 -1.69106674e+00 1.79404750e-01
3.22669089e-01 9.73933816e-01 6.49276555e-01 5.29501021e-01
-2.84060270e-01 2.11287975e-01 -3.75494182e-01 -1.74002331e-02
5.28430462e-01 3.48296538e-02 -8.12926650e-01 7.45020628e-01
8.31538618e-01 -5.50475001e-01 1.10080242e-01 1.21724498e+00
-1.15847476e-01 1.26032084e-01 6.73946798e-01 -1.21410096e+00
3.96754831e-01 -2.63642073e-01 -4.39913839e-01 -1.84140369e-01
6.06424630e-01 2.10963443e-01 5.31192958e-01 -1.17499638e+00
1.00262749e+00 1.36994445e+00 7.46431053e-01 6.69543371e-02
-1.31035614e+00 -1.25941873e-01 -1.60918415e-01 2.05850109e-01
-1.47491527e+00 -4.29139316e-01 1.04039693e+00 -7.06215441e-01
1.03251433e+00 6.29761517e-02 6.22385621e-01 5.66516757e-01
-1.53130457e-01 3.36962163e-01 7.09298849e-01 -6.18432820e-01
3.45118374e-01 2.07343966e-01 1.17925726e-01 6.45280063e-01
4.78520319e-02 1.46387920e-01 -3.07902366e-01 -2.51906794e-02
8.36502433e-01 -1.23091146e-01 -1.10527396e-01 -9.05226529e-01
-1.08572447e+00 4.24330443e-01 3.99941236e-01 -2.50141192e-02
-3.99280488e-01 -2.40865014e-02 -3.24598402e-02 -2.52124220e-01
2.13824213e-01 1.94951981e-01 -6.04877055e-01 2.94071376e-01
-1.11552536e+00 7.86224604e-02 4.93774593e-01 1.07366681e+00
1.36600184e+00 -1.07056066e-01 5.18690109e-01 3.22458744e-01
5.64395666e-01 7.79183030e-01 3.49070504e-02 -1.54266357e+00
2.52149343e-01 8.89041662e-01 1.52027160e-01 -1.29549491e+00
-7.15732872e-01 -1.62628237e-02 -4.12157565e-01 6.17492020e-01
5.79676509e-01 6.16127372e-01 -1.19948328e+00 1.01329613e+00
8.34295988e-01 1.21440805e-01 2.37222314e-02 9.98907447e-01
7.68786311e-01 3.93391788e-01 -5.42620540e-01 2.04084907e-02
1.05916417e+00 -3.93535256e-01 -2.27661282e-01 -3.47078741e-01
4.84897166e-01 -1.14611423e+00 4.95438010e-01 4.08583313e-01
-1.12231791e+00 -7.16927588e-01 -8.62034023e-01 -2.44517714e-01
-2.80599624e-01 1.82558000e-01 1.13806777e-01 2.57774740e-01
-8.77202213e-01 7.35780060e-01 -7.61644542e-01 -4.60431784e-01
3.94725829e-01 6.12749636e-01 -6.38158977e-01 -2.61689514e-01
-5.40100992e-01 1.00325823e+00 7.57783413e-01 5.32883644e-01
-9.81336296e-01 -6.16691828e-01 -9.88937080e-01 -1.82498708e-01
4.25510854e-01 -8.09184313e-01 8.34705412e-01 -1.24308109e+00
-1.06126618e+00 1.36083877e+00 -1.53251737e-01 -2.87204564e-01
7.56498992e-01 -4.51076329e-02 1.87741011e-01 5.08719206e-01
3.58488470e-01 8.01557660e-01 1.04515600e+00 -1.40725541e+00
-7.18974292e-01 -4.41242009e-01 -2.42534071e-01 2.53090471e-01
7.01494992e-01 -1.87370211e-01 -8.01039398e-01 7.66195729e-02
9.03000474e-01 -8.71647835e-01 -3.60911816e-01 1.97677258e-02
-6.42788410e-01 3.08377445e-01 9.50477958e-01 -8.35813701e-01
1.46375611e-01 -2.09946823e+00 9.43557695e-02 6.12146854e-01
-1.58386961e-01 8.88073444e-02 -1.40313774e-01 -1.20569877e-01
4.83579747e-03 -2.00972140e-01 -5.33391178e-01 -5.70304811e-01
-3.54446590e-01 2.72402883e-01 1.08212419e-01 8.73381793e-01
2.35396966e-01 4.54369813e-01 -4.64700878e-01 -4.68050838e-01
1.06845081e+00 4.82952476e-01 -3.94020170e-01 2.09909618e-01
-6.15403473e-01 7.64196098e-01 -3.17574531e-01 6.20073497e-01
1.26164782e+00 1.72036290e-01 -1.86168626e-01 -3.16870838e-01
-2.64425486e-01 1.46748483e-01 -1.74905300e+00 1.93239081e+00
-1.23075582e-01 4.89728838e-01 1.94372237e-01 -9.17975545e-01
9.83446896e-01 1.48530109e-02 6.96850121e-01 -4.46587324e-01
3.33055884e-01 3.33417386e-01 -3.90432060e-01 -5.99065959e-01
6.19289637e-01 4.53721434e-01 3.60797435e-01 -7.61842206e-02
1.25763103e-01 -7.65316188e-01 -3.19978334e-02 -7.02502504e-02
4.70166117e-01 6.62022769e-01 -3.70363407e-02 -1.13009058e-01
8.52908850e-01 6.11896873e-01 7.11872339e-01 4.60934609e-01
1.93893000e-01 1.13714981e+00 2.70972520e-01 -6.15402579e-01
-1.38359725e+00 -7.82746613e-01 -3.18792015e-01 -5.74125908e-02
5.39272904e-01 2.72842310e-02 -8.57566714e-01 -3.73839468e-01
-8.91965777e-02 6.74903929e-01 -3.77899766e-01 5.36054820e-02
-6.19761050e-01 -3.05444330e-01 -1.33352697e-01 6.40304908e-02
4.63846058e-01 -8.10631394e-01 -1.11865997e+00 4.08899076e-02
-2.17840776e-01 -1.47544658e+00 8.78035054e-02 7.98349231e-02
-1.07385921e+00 -1.59275520e+00 -1.87223330e-01 -5.83604217e-01
8.90817940e-01 4.77775633e-01 1.08106923e+00 1.77877396e-01
-4.20008183e-01 5.35676301e-01 -7.42921680e-02 3.66290361e-02
-6.61850929e-01 -2.57346481e-01 -1.07476801e-01 2.04228610e-01
1.73058853e-01 -2.38312840e-01 -5.05881250e-01 6.51616096e-01
-8.32635522e-01 3.42857540e-01 2.37598866e-01 3.89472157e-01
1.03413761e+00 1.21263660e-01 -2.44280696e-01 -6.50086880e-01
-6.76846921e-01 -1.04016200e-01 -1.20276964e+00 4.90313768e-02
-3.19654942e-01 -1.10246882e-01 8.66699591e-02 -5.89829497e-02
-1.01317680e+00 8.77923429e-01 -1.55040056e-01 -7.93064535e-01
-6.77840710e-01 2.93341093e-02 -5.76087415e-01 -1.05114207e-01
5.21910608e-01 9.36856568e-02 1.98758051e-01 -5.30856967e-01
2.92895406e-01 3.52522373e-01 7.66289711e-01 -2.50839323e-01
7.20904231e-01 1.01169586e+00 5.43831468e-01 -1.02820170e+00
-6.43743992e-01 -9.04173970e-01 -1.44657779e+00 -4.36307341e-01
1.15002644e+00 -7.77801692e-01 -1.79697156e-01 5.50556660e-01
-1.56752002e+00 -1.07604273e-01 -2.14027449e-01 5.63498855e-01
-8.43156397e-01 4.94541228e-01 -5.71639091e-02 -7.38373101e-01
-6.04011565e-02 -1.59742808e+00 1.30920434e+00 1.59100160e-01
-9.43547934e-02 -7.87823498e-01 -5.02367079e-01 6.71391010e-01
-2.74831176e-01 3.59357744e-01 7.05545962e-01 -2.93312043e-01
-1.20950818e+00 -3.38049978e-01 2.13527866e-02 4.62302208e-01
-2.36917675e-01 4.31796908e-01 -1.20402229e+00 7.08421767e-02
2.06577584e-01 2.21463367e-01 5.47208190e-01 8.34974885e-01
5.95883071e-01 4.23676580e-01 -3.59653652e-01 9.76132512e-01
1.59405518e+00 7.83454105e-02 6.76615775e-01 6.72563374e-01
9.51992333e-01 9.20392692e-01 9.55113947e-01 1.42123505e-01
-2.61284318e-02 9.18245852e-01 1.08227742e+00 -1.21920491e-02
-1.55465573e-01 -1.20938458e-01 1.58968091e-01 8.71032178e-02
1.11172572e-02 3.39021802e-01 -1.03457034e+00 3.96318018e-01
-1.67785764e+00 -5.86614370e-01 -9.96800005e-01 2.57399774e+00
2.78921239e-02 2.39736542e-01 -2.09323958e-01 1.93460777e-01
8.09464931e-01 -4.35259193e-01 -5.76308012e-01 -1.01699591e-01
-3.65144163e-01 -1.39738977e-01 7.32271969e-01 8.29690099e-01
-1.05003393e+00 1.01909864e+00 4.90097523e+00 5.30231416e-01
-8.23118985e-01 -1.36574030e-01 3.93379152e-01 2.86994636e-01
1.99934449e-02 4.71117496e-01 -1.02582431e+00 2.04292104e-01
4.70938832e-01 4.53707784e-01 2.10372671e-01 8.72330546e-01
2.99844265e-01 -6.05423331e-01 -1.17934537e+00 1.26373231e+00
2.41212592e-01 -1.33076835e+00 -1.10560872e-01 -2.34271903e-02
7.61068523e-01 3.40203226e-01 -2.74884820e-01 -4.86523658e-01
-2.76901096e-01 -6.61895931e-01 1.16751719e+00 7.56045580e-01
4.97522682e-01 -8.84849191e-01 7.69513547e-01 6.46824419e-01
-9.19794500e-01 1.33881167e-01 -6.15704298e-01 2.52253771e-01
3.44275504e-01 6.48918629e-01 -1.10052741e+00 8.41697633e-01
8.06099832e-01 7.75337100e-01 -8.51090908e-01 1.05073881e+00
-2.41659060e-01 2.05445606e-02 -6.64929867e-01 7.43018091e-01
5.69380186e-02 -6.16644800e-01 8.25480938e-01 8.24647129e-01
3.01764190e-01 -3.05803180e-01 8.73325691e-02 1.17372632e+00
4.00258541e-01 -2.76242405e-01 -5.73141456e-01 6.68971300e-01
1.49435297e-01 1.39465451e+00 -1.19179857e+00 -2.42691964e-01
-1.02765493e-01 1.01059341e+00 -8.04799721e-02 3.09120476e-01
-4.26781416e-01 3.13699543e-01 3.86395782e-01 3.76356810e-01
2.60402948e-01 -4.16522115e-01 -2.74440020e-01 -9.15139318e-01
3.34317982e-02 -4.21359181e-01 1.89147711e-01 -1.24570692e+00
-7.34683752e-01 3.65312755e-01 1.79237545e-01 -1.38770151e+00
-3.72831702e-01 -6.21954858e-01 -2.57376134e-01 1.06445706e+00
-1.24404514e+00 -1.08984768e+00 -5.00125825e-01 5.74032247e-01
7.18860745e-01 1.08896762e-01 5.73031604e-01 3.57268155e-02
-1.94458738e-01 -3.49603653e-01 6.38677180e-02 -1.75770149e-01
2.76112020e-01 -9.80398238e-01 2.22181827e-01 1.00858068e+00
8.34412873e-02 6.12457953e-02 6.21424496e-01 -7.66829789e-01
-1.15033877e+00 -1.01439357e+00 5.84406674e-01 -7.44060934e-01
-5.16882241e-02 -2.84823805e-01 -1.00622165e+00 4.70083117e-01
-3.90513390e-01 2.01854762e-02 -3.91747542e-02 -6.20627105e-01
1.10255010e-01 2.19533712e-01 -1.29435062e+00 7.85958096e-02
7.25304902e-01 -3.32895279e-01 -4.78978783e-01 3.38442177e-01
5.17327070e-01 -6.87676489e-01 -5.43360710e-01 4.69723046e-01
2.30301663e-01 -1.16336644e+00 1.32376039e+00 -1.35173768e-01
1.46445990e-01 -8.36448312e-01 -3.56151521e-01 -8.78457487e-01
2.20811427e-01 -2.39110231e-01 3.79845262e-01 1.07945132e+00
-2.73689218e-02 -1.44637704e-01 1.01975024e+00 7.98471749e-01
-2.63598531e-01 -1.55676985e-02 -1.27103603e+00 -5.77308297e-01
-2.55263567e-01 -8.37361038e-01 3.79911065e-01 8.45531940e-01
-1.07178843e+00 1.54735640e-01 4.39613964e-03 7.60819376e-01
1.01762533e+00 1.07654549e-01 1.22579765e+00 -1.42366779e+00
-6.90039992e-02 -1.94685802e-01 -9.19334650e-01 -8.09227049e-01
1.85960338e-01 -8.77795100e-01 2.85474379e-02 -1.53514111e+00
-1.82801977e-01 -4.83289599e-01 4.12915021e-01 -1.38377532e-01
3.80369872e-01 1.64868250e-01 2.36845285e-01 5.49196184e-01
-3.06833744e-01 1.98451281e-01 8.75927866e-01 -4.01424132e-02
-2.39104658e-01 1.33503139e-01 4.55501974e-01 1.36199796e+00
5.86095750e-01 -5.66043496e-01 -1.18305251e-01 -5.30756712e-01
2.21982136e-01 9.02971327e-02 8.96266222e-01 -1.10486770e+00
2.15521634e-01 3.71551700e-02 6.78545952e-01 -1.40902269e+00
7.13469326e-01 -1.50400174e+00 7.36906946e-01 3.53342384e-01
5.98478392e-02 -1.41608834e-01 6.90245479e-02 4.17236626e-01
-4.02916856e-02 -7.21084297e-01 9.55169857e-01 -3.22737366e-01
-1.05336523e+00 1.26482382e-01 -1.98032185e-01 -3.59922409e-01
1.00263119e+00 -8.77944112e-01 3.48313123e-01 6.25891536e-02
-8.90461445e-01 9.16068032e-02 9.20654118e-01 1.36867777e-01
1.05029798e+00 -9.58802402e-01 -4.55248356e-01 7.04570413e-01
8.51349533e-02 1.02768528e+00 4.97207135e-01 8.85474145e-01
-1.10778165e+00 2.21393943e-01 -1.24494515e-01 -1.42445183e+00
-1.44640493e+00 5.58496892e-01 8.13353181e-01 4.61549371e-01
-9.58909094e-01 4.34150189e-01 1.46175355e-01 -5.71728885e-01
1.62168086e-01 -6.68011010e-01 -4.96452302e-02 -1.34615004e-01
3.05113047e-01 3.61533374e-01 3.81916165e-01 -1.25569439e+00
-4.05437499e-01 1.20569229e+00 2.60402054e-01 -2.08281666e-01
1.12841547e+00 -3.73852462e-01 -2.09509715e-01 2.31897265e-01
1.27690661e+00 -2.75089860e-01 -1.54241490e+00 -2.18942791e-01
7.41691142e-02 -7.97532618e-01 1.87881663e-01 -4.00919139e-01
-9.41791594e-01 8.41500401e-01 7.92531013e-01 -3.15990627e-01
8.35356534e-01 1.47060260e-01 2.07814023e-01 2.11820751e-01
3.81843626e-01 -1.18894482e+00 -3.67695361e-01 4.52408582e-01
6.69017196e-01 -1.41660154e+00 7.70813748e-02 -6.59337938e-01
-2.89034903e-01 1.55736864e+00 3.36595505e-01 -9.02277902e-02
4.68811423e-01 -1.72880158e-01 9.76555720e-02 -4.10487115e-01
9.43889990e-02 -1.99800059e-01 3.15529913e-01 8.31654549e-01
-3.23183060e-01 -2.17763871e-01 2.57625163e-01 -7.76835606e-02
-4.72087972e-02 -2.38133952e-01 6.92435861e-01 6.93538129e-01
-4.72838402e-01 -8.38637352e-01 -9.79165196e-01 -1.06060050e-01
-8.16343073e-03 2.78737217e-01 -3.17302078e-01 8.61242652e-01
5.26442289e-01 7.86862314e-01 5.01333177e-01 -1.07471064e-01
4.27033365e-01 -4.17480469e-02 5.68133593e-01 -5.37339747e-01
-1.99991450e-01 3.20116729e-01 -2.68828291e-02 -8.10795963e-01
-7.77420402e-01 -8.61337960e-01 -1.51043594e+00 1.68109775e-01
-3.46358746e-01 -2.97009259e-01 1.30775666e+00 9.23751831e-01
1.05550230e-01 1.63754284e-01 5.25021195e-01 -1.37048733e+00
8.65253969e-04 -6.03824735e-01 -5.25216818e-01 6.67164505e-01
2.19979346e-01 -7.71136999e-01 -4.77973461e-01 3.65247041e-01] | [7.644425392150879, -2.6629722118377686] |
a3fc3825-deb8-45d2-98bf-9e43214755ef | embedded-deep-bilinear-interactive | 2007.06143 | null | https://arxiv.org/abs/2007.06143v1 | https://arxiv.org/pdf/2007.06143v1.pdf | Embedded Deep Bilinear Interactive Information and Selective Fusion for Multi-view Learning | As a concrete application of multi-view learning, multi-view classification improves the traditional classification methods significantly by integrating various views optimally. Although most of the previous efforts have been demonstrated the superiority of multi-view learning, it can be further improved by comprehensively embedding more powerful cross-view interactive information and a more reliable multi-view fusion strategy in intensive studies. To fulfill this goal, we propose a novel multi-view learning framework to make the multi-view classification better aimed at the above-mentioned two aspects. That is, we seamlessly embed various intra-view information, cross-view multi-dimension bilinear interactive information, and a new view ensemble mechanism into a unified framework to make a decision via the optimization. In particular, we train different deep neural networks to learn various intra-view representations, and then dynamically learn multi-dimension bilinear interactive information from different bilinear similarities via the bilinear function between views. After that, we adaptively fuse the representations of multiple views by flexibly tuning the parameters of the view-weight, which not only avoids the trivial solution of weight but also provides a new way to select a few discriminative views that are beneficial to make a decision for the multi-view classification. Extensive experiments on six publicly available datasets demonstrate the effectiveness of the proposed method. | ['Junwei Han', 'Xiwen Yao', 'Xiangsen Zhang', 'Peicheng Zhou', 'Xinwang Liu', 'Jiantao Shen', 'Wenbin Li', 'Jinglin Xu'] | 2020-07-13 | null | null | null | null | ['multi-view-learning'] | ['computer-vision'] | [-4.02244478e-01 -6.61303282e-01 -2.18259618e-01 -5.84192634e-01
-8.93931448e-01 -6.17823839e-01 5.00697017e-01 -1.91209927e-01
-3.29735838e-02 1.63880542e-01 4.25766498e-01 1.22614361e-01
-2.82014310e-01 -8.61755908e-01 -3.86233449e-01 -1.08754289e+00
3.51761460e-01 4.83188443e-02 1.35405064e-01 -2.23800838e-01
2.42571443e-01 3.00110638e-01 -1.46986425e+00 5.67007780e-01
9.20999825e-01 1.02328646e+00 1.62913039e-01 1.07288107e-01
8.17047954e-02 3.40131879e-01 -2.16330916e-01 -3.54200304e-01
1.59737051e-01 -3.09151679e-01 -2.39842936e-01 3.68548274e-01
4.20015782e-01 -3.56623918e-01 3.93503197e-02 9.93186772e-01
7.16318607e-01 4.76204604e-02 4.17299002e-01 -1.00845981e+00
-7.07993090e-01 3.50155950e-01 -8.40969920e-01 2.17447311e-01
1.64397106e-01 5.13571277e-02 1.14211214e+00 -1.21915758e+00
2.21865520e-01 1.38526869e+00 4.79286522e-01 1.12135150e-01
-1.05017555e+00 -8.05084109e-01 6.55849040e-01 5.09060144e-01
-1.20053971e+00 -3.16138059e-01 1.20899606e+00 -4.15360898e-01
2.84303457e-01 2.84223437e-01 8.31654191e-01 1.11489463e+00
2.43760303e-01 8.33719373e-01 1.38749039e+00 2.83630993e-02
-2.93950349e-01 3.56518865e-01 4.49092872e-02 8.61067235e-01
7.94576034e-02 -1.42038450e-01 -2.77138323e-01 -1.94134280e-01
5.45132995e-01 5.97038448e-01 -5.42514503e-01 -8.18332493e-01
-1.43887103e+00 9.06164527e-01 4.42797065e-01 4.16225165e-01
-2.11767539e-01 -4.55197811e-01 6.87617064e-01 1.34681165e-01
5.59987783e-01 -6.86591864e-03 -3.71065527e-01 3.60907704e-01
-4.61807996e-01 -9.44340453e-02 3.24806899e-01 4.65957791e-01
7.93392181e-01 7.47485459e-02 -2.34475601e-02 1.10129428e+00
3.12151134e-01 3.74341756e-01 6.34207845e-01 -6.69590294e-01
8.34152400e-01 8.91115129e-01 -1.58326611e-01 -1.29969943e+00
-3.19601357e-01 -6.38418257e-01 -1.31772888e+00 1.40977651e-01
4.78491709e-02 -1.85375772e-02 -4.23368156e-01 1.57603073e+00
4.92294908e-01 1.59137458e-01 1.56870648e-01 9.22696114e-01
9.11158741e-01 5.66107571e-01 -2.65441179e-01 -3.23616713e-01
1.46871603e+00 -1.09152997e+00 -5.61513603e-01 6.88342601e-02
2.71585792e-01 -7.56657362e-01 7.51571000e-01 5.11839271e-01
-6.89984024e-01 -8.94944727e-01 -1.46766150e+00 1.38430402e-01
-3.57313991e-01 4.04949814e-01 7.07788765e-01 3.11117262e-01
-5.73763013e-01 1.07258797e-01 -5.37589014e-01 1.40231088e-01
2.56769240e-01 2.30135903e-01 -7.81593084e-01 -2.18858421e-01
-1.06945336e+00 5.54686785e-01 4.74532068e-01 4.03954297e-01
-6.09547794e-01 -5.11170387e-01 -9.33159590e-01 4.09184061e-02
5.15921414e-01 -8.82134438e-01 5.67203641e-01 -8.86643887e-01
-1.25238299e+00 6.34370863e-01 -5.97036704e-02 2.40003183e-01
3.78799200e-01 -1.16143338e-01 -6.88970566e-01 1.78261489e-01
1.42106175e-01 5.42606935e-02 9.28237438e-01 -1.64302146e+00
-5.76141179e-01 -8.09189618e-01 3.14901531e-01 5.10599911e-01
-5.42657375e-01 -2.18148887e-01 -5.88141084e-01 -7.55183816e-01
5.39717376e-01 -8.23769629e-01 -1.33334383e-01 -2.49446318e-01
-1.03298895e-01 -2.17817187e-01 9.83141363e-01 -6.48093164e-01
1.14267027e+00 -2.16362786e+00 5.12977362e-01 -4.06589136e-02
3.58110785e-01 1.52978584e-01 5.86508438e-02 3.62097055e-01
-2.07161650e-01 -9.00882930e-02 3.96370962e-02 -3.08385223e-01
-3.83608699e-01 1.16400339e-01 -1.36604607e-01 4.72534120e-01
-2.06847548e-01 5.04389644e-01 -7.35413730e-01 -4.61226016e-01
4.71802831e-01 5.31831980e-01 -6.31872952e-01 4.18615788e-01
3.12647283e-01 7.06276894e-01 -7.34001160e-01 6.03015661e-01
9.72852767e-01 -4.51937973e-01 2.19534293e-01 -8.20607364e-01
-7.44312182e-02 -2.74324924e-01 -1.41722322e+00 1.74031842e+00
-7.29299009e-01 -1.06993765e-01 1.23389319e-01 -1.22225749e+00
8.00433278e-01 2.84125805e-01 7.27176666e-01 -3.49941373e-01
4.26534824e-02 1.07391499e-01 -2.32138544e-01 -5.49088538e-01
1.78533375e-01 -3.45572770e-01 -1.53886989e-01 3.74681413e-01
2.10818723e-01 3.68278056e-01 -2.48931661e-01 -1.08434215e-01
1.73115835e-01 2.69225687e-02 5.17530859e-01 -2.97804214e-02
1.23230672e+00 -7.34915197e-01 7.85774887e-01 1.46826833e-01
-2.30265036e-03 6.99973166e-01 2.06028789e-01 -6.28310263e-01
-5.86918712e-01 -8.73367488e-01 -1.08129971e-01 1.13391471e+00
4.66411352e-01 -3.49967331e-01 -3.17430943e-01 -1.17699552e+00
1.07277809e-02 1.46495536e-01 -6.89968824e-01 -1.36405766e-01
-6.46701992e-01 -8.61884713e-01 -1.04116812e-01 7.13267207e-01
8.20594966e-01 -4.39785212e-01 1.02270238e-01 3.40564288e-02
-3.50412935e-01 -1.06389976e+00 -6.59632564e-01 7.01304898e-02
-8.99210513e-01 -1.03811324e+00 -8.09126258e-01 -6.65823996e-01
6.52748644e-01 9.14585233e-01 8.43659520e-01 2.04539727e-02
2.17672497e-01 4.48032886e-01 -3.59843999e-01 6.77905530e-02
2.44166385e-02 -5.85446618e-02 1.15913674e-01 6.06028318e-01
6.42328784e-02 -8.37677360e-01 -7.67607450e-01 5.81703126e-01
-8.03334117e-01 1.96606025e-01 5.48747241e-01 1.22926509e+00
8.31252158e-01 1.28417853e-02 5.48925459e-01 -8.47235084e-01
2.82360882e-01 -5.26182890e-01 -4.40611988e-01 6.12559915e-01
-6.06069803e-01 -7.04831257e-02 9.76899028e-01 -1.44774646e-01
-1.17320383e+00 -1.25532806e-01 -1.88974291e-01 -7.15341687e-01
-9.59254429e-03 4.47055340e-01 -7.32789278e-01 -2.51183778e-01
2.52252426e-02 5.94661832e-01 -1.30525649e-01 -5.23682714e-01
5.57746947e-01 5.48797250e-01 6.73624277e-02 -5.31793058e-01
8.05749416e-01 6.29509449e-01 -1.47621378e-01 -2.79497266e-01
-1.06833732e+00 -5.03375590e-01 -8.86868834e-01 -2.15781927e-01
9.67798829e-01 -1.35315061e+00 -7.91881382e-01 4.82255846e-01
-7.48365462e-01 5.53281307e-01 3.38538557e-01 5.64768314e-01
-2.88704902e-01 7.21810520e-01 -3.22873652e-01 -3.74335736e-01
-3.23555917e-01 -1.40349150e+00 1.11473048e+00 3.13438267e-01
5.67167103e-01 -1.15802395e+00 -6.13095947e-02 7.55277395e-01
3.79587077e-02 1.91953555e-01 8.25500846e-01 -5.70832968e-01
-5.16645849e-01 -1.19435720e-01 -2.84019083e-01 7.33596385e-01
4.27196473e-01 -1.18419386e-01 -9.21886802e-01 -6.04865372e-01
3.90703738e-01 -2.93934375e-01 9.60753024e-01 4.39959615e-02
1.45901608e+00 -1.80602267e-01 -2.67758131e-01 9.34736550e-01
1.73097789e+00 2.76891738e-01 5.67057729e-02 2.21424967e-01
1.14348269e+00 3.02556545e-01 6.04268849e-01 5.31840026e-01
8.60690475e-01 7.77331650e-01 4.70884025e-01 -1.33258547e-03
3.12622070e-01 -1.60370365e-01 2.64904737e-01 1.48403418e+00
-3.31140965e-01 8.46390352e-02 -4.34841990e-01 2.15113983e-01
-1.77492654e+00 -1.17235112e+00 2.54357308e-01 2.09914804e+00
3.70607942e-01 2.59793866e-02 1.43968582e-01 -9.51229632e-02
6.69220209e-01 7.29500592e-01 -4.47485030e-01 2.06515249e-02
-6.31562546e-02 -4.85241413e-01 9.75893289e-02 1.63042173e-01
-1.46166635e+00 3.40755433e-01 5.21438789e+00 8.09514642e-01
-1.40095234e+00 1.83941782e-01 6.34984195e-01 -1.91031545e-02
-5.97778797e-01 -1.05859794e-01 -9.24880087e-01 5.23998260e-01
1.45165175e-01 5.33254212e-03 3.81129831e-01 9.05480742e-01
3.31812873e-02 1.06817745e-01 -8.85036528e-01 1.38482487e+00
5.34396529e-01 -1.26491427e+00 4.55692530e-01 1.09609596e-01
7.31023192e-01 -2.74185598e-01 7.02779144e-02 4.10442591e-01
-7.84029216e-02 -4.94132042e-01 5.29546320e-01 5.87115884e-01
4.35306460e-01 -9.50287163e-01 7.25496292e-01 3.98754269e-01
-1.70029771e+00 -3.92314970e-01 -3.19578648e-01 3.08252513e-01
-5.98202981e-02 5.42183757e-01 4.44104970e-02 1.39033508e+00
6.35975897e-01 1.21600246e+00 -7.62691081e-01 7.53520727e-01
9.73873660e-02 3.67897525e-02 9.03696269e-02 4.33951557e-01
2.48949572e-01 -5.60015619e-01 4.40912187e-01 7.31567621e-01
3.52275044e-01 1.94178417e-01 6.13075316e-01 2.91277409e-01
1.49749875e-01 2.91343510e-01 -7.24692106e-01 4.53435630e-01
2.62464613e-01 1.57044816e+00 -4.25387293e-01 -3.74816120e-01
-8.58234763e-01 9.17062163e-01 4.45672482e-01 3.06932211e-01
-8.31063509e-01 1.00685759e-02 5.61461091e-01 -2.53987342e-01
6.64212465e-01 -2.28367552e-01 -1.77535415e-01 -1.86485672e+00
3.43427777e-01 -1.16345811e+00 7.19313562e-01 -5.49550235e-01
-1.47066379e+00 5.74032784e-01 -2.62173335e-03 -1.87970531e+00
-7.77215958e-02 -5.66143692e-01 -6.65143073e-01 8.11363041e-01
-1.31790316e+00 -1.56362331e+00 -4.30548698e-01 8.07269871e-01
5.96258163e-01 -3.77573311e-01 8.03117335e-01 4.81431544e-01
-8.21924388e-01 6.50571108e-01 1.93070561e-01 8.16094279e-02
8.71638775e-01 -1.13758743e+00 -2.48044327e-01 6.83762848e-01
1.13974921e-01 7.13799775e-01 1.50526881e-01 -7.77338669e-02
-1.58802593e+00 -8.89027297e-01 1.71327457e-01 -2.70759851e-01
4.71482158e-01 -2.51271069e-01 -9.24235284e-01 6.77990496e-01
2.99633533e-01 1.78665280e-01 1.12363839e+00 3.39684218e-01
-6.15832925e-01 -7.57555306e-01 -7.55087018e-01 3.58589292e-01
6.43233180e-01 -7.50617743e-01 -5.20122409e-01 8.11116099e-02
4.81912613e-01 -3.04653168e-01 -1.21426380e+00 7.68533409e-01
7.46945262e-01 -1.43968797e+00 1.21552944e+00 -4.25815076e-01
3.20878118e-01 -4.46478814e-01 -3.79497945e-01 -1.56105888e+00
-5.92887223e-01 8.14951584e-02 -5.02944365e-02 1.47947145e+00
-8.43012631e-02 -8.52830172e-01 3.55069518e-01 -2.00670608e-03
-1.00491166e-01 -1.35357380e+00 -8.16630423e-01 -4.07810122e-01
-3.66625935e-02 -6.26543164e-02 6.41677856e-01 1.14506698e+00
-3.83584946e-01 3.85188460e-01 -6.66628242e-01 4.42349434e-01
5.97182572e-01 8.46075654e-01 7.81530023e-01 -1.10285246e+00
-5.01584530e-01 -4.73437011e-01 -3.94034445e-01 -1.00165188e+00
9.65468585e-02 -8.81206691e-01 -3.40162486e-01 -1.44746435e+00
6.44700587e-01 -3.77241910e-01 -7.90319502e-01 2.34773785e-01
-6.36822402e-01 6.37742355e-02 2.27248549e-01 3.29805672e-01
-7.08306909e-01 7.84766674e-01 1.66857207e+00 -9.40089524e-02
1.71572238e-01 1.73039719e-01 -1.11466491e+00 8.62941384e-01
3.05720538e-01 -7.20614046e-02 -4.88902658e-01 -4.67654556e-01
-7.09995767e-03 3.99891049e-01 4.41909194e-01 -9.73328471e-01
3.82720493e-02 -1.53609335e-01 7.48481393e-01 -7.83375800e-01
4.51596379e-01 -1.01317358e+00 -5.11123203e-02 6.54661432e-02
8.44242498e-02 1.85337394e-01 -7.72877336e-02 9.67138529e-01
-6.83799446e-01 1.44322589e-01 7.43108034e-01 -3.37170541e-01
-8.54007244e-01 5.03551960e-01 1.16641775e-01 -1.41776830e-01
1.04931951e+00 -1.32746235e-01 -1.45385653e-01 -2.19640389e-01
-8.86590958e-01 3.76177073e-01 3.90859514e-01 6.27273977e-01
5.22021234e-01 -1.85902452e+00 -4.63941514e-01 5.22857070e-01
3.78672630e-01 -2.59738624e-01 8.58159244e-01 9.64251697e-01
8.11539590e-02 1.53027639e-01 -3.69892120e-01 -7.93327332e-01
-1.39812720e+00 7.07122326e-01 3.64540458e-01 -5.33549726e-01
-5.29035687e-01 6.36891544e-01 6.25180840e-01 -4.95211303e-01
-1.04365528e-01 -9.68481600e-02 -7.32427061e-01 6.54039145e-01
5.22049189e-01 2.36721218e-01 2.74558049e-02 -8.87525618e-01
-3.79588395e-01 1.05343366e+00 -2.00145975e-01 3.72946352e-01
1.43640971e+00 -4.08798158e-01 -2.30528951e-01 8.00769150e-01
1.63055277e+00 2.91460991e-01 -1.21061873e+00 -2.82974333e-01
-7.15254605e-01 -5.55595815e-01 -7.59067088e-02 -5.35009861e-01
-1.49997938e+00 1.04265666e+00 7.38728225e-01 3.19814295e-01
1.39282131e+00 -2.34405816e-01 6.90210342e-01 2.13950709e-01
4.47383553e-01 -7.43893504e-01 2.83664048e-01 2.75350124e-01
9.77660954e-01 -1.51855659e+00 2.74552196e-01 -4.38180059e-01
-7.54119217e-01 1.25164664e+00 8.57251227e-01 -1.14880219e-01
9.66167331e-01 -2.22894594e-01 1.87978625e-01 -1.96802750e-01
-5.60575664e-01 7.60886595e-02 6.18866444e-01 1.62372485e-01
2.91206628e-01 6.75669238e-02 -1.54697150e-01 8.20679724e-01
2.80312032e-01 -3.18320125e-01 1.24603393e-03 4.60784942e-01
-1.61874875e-01 -1.15491533e+00 -3.67700487e-01 3.29867840e-01
-3.43802899e-01 1.67719409e-01 1.27568617e-01 6.06476307e-01
4.66233641e-01 6.80279315e-01 -3.39079082e-01 -8.29394281e-01
2.64201969e-01 -7.89297968e-02 3.59776586e-01 -4.64675426e-01
-6.91729128e-01 1.70195222e-01 -1.43581972e-01 -5.34798443e-01
-6.74600899e-01 -6.51451170e-01 -6.00926757e-01 -1.33383712e-02
-4.09714341e-01 -1.88129693e-02 2.39073411e-01 1.14597785e+00
4.21506166e-01 4.96129334e-01 1.30848050e+00 -9.67003465e-01
-6.64271772e-01 -6.62853599e-01 -5.41419804e-01 4.81206954e-01
2.69111425e-01 -9.34042454e-01 -4.01819676e-01 -3.15398335e-01] | [8.496973037719727, 4.513453006744385] |
81b3a8ff-94a5-45de-97f0-28662521f480 | source-side-prediction-for-neural-headline | 1712.08302 | null | http://arxiv.org/abs/1712.08302v1 | http://arxiv.org/pdf/1712.08302v1.pdf | Source-side Prediction for Neural Headline Generation | The encoder-decoder model is widely used in natural language generation
tasks. However, the model sometimes suffers from repeated redundant generation,
misses important phrases, and includes irrelevant entities. Toward solving
these problems we propose a novel source-side token prediction module. Our
method jointly estimates the probability distributions over source and target
vocabularies to capture a correspondence between source and target tokens. The
experiments show that the proposed model outperforms the current
state-of-the-art method in the headline generation task. Additionally, we show
that our method has an ability to learn a reasonable token-wise correspondence
without knowing any true alignments. | ['Masaaki Nagata', 'Kentaro Inui', 'Sho Takase', 'Shun Kiyono', 'Jun Suzuki', 'Naoaki Okazaki'] | 2017-12-22 | null | null | null | null | ['headline-generation'] | ['natural-language-processing'] | [-1.33897692e-01 3.71535420e-01 -4.65278745e-01 -2.74887860e-01
-1.27315748e+00 -4.31797773e-01 9.10743833e-01 1.59821138e-01
-2.74299234e-01 1.24965942e+00 7.21011519e-01 9.55524854e-03
3.86485189e-01 -8.98095369e-01 -7.96380103e-01 -2.45967254e-01
2.36032039e-01 7.43977368e-01 1.30778342e-01 -5.98570287e-01
1.64361000e-01 -3.07563722e-01 -1.15626943e+00 2.99218923e-01
1.26613617e+00 5.02857983e-01 4.19491798e-01 3.18581045e-01
-5.87364793e-01 7.83913434e-01 -7.77040124e-01 -9.03346896e-01
5.26243560e-02 -5.52972555e-01 -6.71349585e-01 -2.94684857e-01
2.37564027e-01 -2.26385251e-01 -8.17658231e-02 1.09755826e+00
5.20894825e-01 -1.45960435e-01 8.29237401e-01 -1.15549731e+00
-1.07784891e+00 1.46334183e+00 -3.87710720e-01 -1.51048982e-04
6.14513874e-01 -4.33470495e-02 1.59152758e+00 -1.34158552e+00
7.69543707e-01 1.20692146e+00 3.60564232e-01 5.72303355e-01
-9.35937822e-01 -1.06870103e+00 2.40356073e-01 -2.63362434e-02
-1.70936143e+00 -3.96519333e-01 5.49548388e-01 -2.62275279e-01
1.20976603e+00 -1.88135177e-01 5.43822944e-01 1.32082570e+00
2.85263240e-01 8.96755934e-01 5.42901456e-01 -5.36067128e-01
-1.51064545e-01 3.28768581e-01 -1.52260125e-01 5.96412838e-01
5.53628027e-01 -7.39084855e-02 -7.71027625e-01 -2.55179852e-01
5.79359353e-01 -4.01490092e-01 -2.76095778e-01 1.40581101e-01
-1.57163131e+00 8.32082987e-01 5.83894812e-02 3.77762288e-01
-3.28641921e-01 3.59382242e-01 2.74728149e-01 8.36495087e-02
5.53136110e-01 6.79950714e-01 -4.81106669e-01 -2.20016569e-01
-9.34462488e-01 6.71248019e-01 8.66300285e-01 1.70319903e+00
6.81315005e-01 3.50334458e-02 -6.19812250e-01 7.42639720e-01
5.77424645e-01 6.99210107e-01 6.28913164e-01 -2.72996217e-01
1.01958382e+00 4.63653952e-01 4.10801917e-01 -7.93379486e-01
3.04785408e-02 -3.20815176e-01 -7.87396908e-01 -7.38638997e-01
6.52964190e-02 -3.82970244e-01 -7.10684478e-01 1.90539086e+00
1.07859336e-01 1.94100320e-01 2.83851683e-01 5.27676225e-01
9.26416159e-01 9.59098279e-01 2.05562234e-01 -2.61166066e-01
1.28856218e+00 -1.08326876e+00 -9.87242460e-01 -4.26185727e-01
5.30193686e-01 -1.00994170e+00 8.08587551e-01 -3.75184305e-02
-1.27849603e+00 -5.22960305e-01 -6.82264090e-01 -1.86379626e-01
-3.02724838e-01 4.82463837e-01 4.70463604e-01 2.61483908e-01
-7.76008010e-01 3.32798868e-01 -4.41658765e-01 -1.04289256e-01
-5.31793870e-02 1.20582759e-01 -2.51495808e-01 -1.84565336e-02
-1.85706484e+00 9.73156393e-01 7.14811027e-01 -2.62181144e-02
-7.15657651e-01 -7.73651361e-01 -1.07359409e+00 2.64434934e-01
1.13292441e-01 -1.13930416e+00 1.53841913e+00 -6.46395743e-01
-1.39952493e+00 3.90941501e-01 -6.34479761e-01 -2.79109120e-01
3.73439848e-01 -3.35468888e-01 -4.24312949e-01 -4.40341026e-01
5.26998699e-01 7.83677757e-01 6.00383639e-01 -9.52996790e-01
-7.54809618e-01 2.90238827e-01 -1.27476916e-01 3.86064589e-01
-2.15305358e-01 3.44989523e-02 -4.98106003e-01 -1.07689202e+00
-2.67992467e-01 -6.15725696e-01 -2.77774036e-01 -4.74618375e-01
-8.01548481e-01 -5.89751422e-01 -9.58147575e-04 -4.87230808e-01
1.59794331e+00 -1.75184929e+00 4.58999686e-02 8.99026915e-03
-5.03305420e-02 7.99145475e-02 -2.54356444e-01 1.00776708e+00
1.43922180e-01 3.23235184e-01 -2.26844288e-02 -3.91795486e-01
3.24517697e-01 -9.37957764e-02 -6.56560004e-01 -2.00946823e-01
2.33637929e-01 9.06453609e-01 -1.34584963e+00 -7.22043395e-01
-3.14209253e-01 2.73987919e-01 -4.96860087e-01 4.83830035e-01
-5.57713807e-01 8.29453245e-02 -5.69520652e-01 4.46751088e-01
4.44533676e-01 -3.20537478e-01 1.93118781e-01 9.23642516e-03
-4.23439406e-02 9.38440561e-01 -1.08229852e+00 1.64451718e+00
-5.20984054e-01 2.77513832e-01 -4.87317920e-01 -3.56401116e-01
1.07080197e+00 7.19516397e-01 5.60066253e-02 -2.83384144e-01
-2.17425320e-02 5.06526053e-01 -1.37262866e-01 -4.08302635e-01
1.05137730e+00 -3.07548910e-01 -4.03970391e-01 5.25499701e-01
2.50260383e-01 -1.35259911e-01 5.67440510e-01 4.40020233e-01
8.14781427e-01 5.56692183e-02 6.91353619e-01 -5.19741103e-02
3.29817981e-01 7.26331258e-03 6.27338588e-01 7.46807933e-01
2.46198863e-01 5.44995487e-01 4.46995348e-01 -1.82558689e-02
-1.08183682e+00 -1.06509519e+00 2.39153430e-01 8.42813194e-01
1.83525667e-01 -9.37308490e-01 -4.74971324e-01 -7.24774063e-01
-3.02545447e-02 1.12583137e+00 -4.05185908e-01 -7.00464398e-02
-5.75565517e-01 -5.15442669e-01 8.34342480e-01 6.27702177e-01
2.22483382e-01 -1.07850194e+00 2.61185676e-01 5.26395440e-01
-7.53761649e-01 -1.14917803e+00 -7.96245277e-01 -3.63561094e-01
-5.59089184e-01 -7.71091700e-01 -8.44564617e-01 -1.02687550e+00
7.35697269e-01 1.51477590e-01 1.43152893e+00 -1.63145706e-01
1.42122895e-01 -2.53421843e-01 -4.02249843e-01 -6.23272061e-01
-6.43084228e-01 5.08341551e-01 -6.13281876e-02 -3.72997522e-01
5.75750768e-01 -2.68482268e-01 -3.78845662e-01 -1.56666599e-02
-6.57824814e-01 3.72319311e-01 8.09157073e-01 9.61762965e-01
5.24558365e-01 -1.78427115e-01 9.17508900e-01 -1.05842471e+00
1.12305343e+00 -7.37977505e-01 -4.90165532e-01 5.91625035e-01
-6.61703885e-01 4.50016230e-01 8.61312211e-01 -2.87469298e-01
-1.27272010e+00 -1.13363996e-01 -2.60969847e-01 -9.54987556e-02
-8.22757545e-04 6.83698535e-01 -1.48304373e-01 6.74577773e-01
3.61638874e-01 4.25235093e-01 -5.97280681e-01 -4.24180746e-01
6.24404669e-01 6.35597706e-01 1.43386796e-01 -5.75703263e-01
9.25440252e-01 -8.59438255e-02 -5.83168626e-01 -4.03301448e-01
-1.01182652e+00 -5.38793743e-01 -4.93213534e-01 8.55245739e-02
5.40328503e-01 -1.37164235e+00 -1.25629827e-01 2.21171319e-01
-1.77390826e+00 2.18709316e-02 -2.37046167e-01 5.13675511e-01
-2.97488838e-01 2.19845772e-01 -5.59385717e-01 -5.15487552e-01
-7.11108088e-01 -8.41191947e-01 1.30258834e+00 2.09798649e-01
-4.32336926e-01 -1.08002758e+00 4.58901286e-01 6.89051375e-02
3.92174751e-01 -2.84740657e-01 9.59149718e-01 -8.57635140e-01
-7.83054829e-01 -1.77393824e-01 -1.93882152e-01 -8.81220922e-02
3.94456744e-01 1.11374311e-01 -5.32182992e-01 6.72668368e-02
-6.86092317e-01 -2.42882445e-01 8.04764569e-01 2.62916461e-02
7.16473758e-01 -7.97427237e-01 -4.73009378e-01 1.60648420e-01
1.32388067e+00 -2.59749264e-01 4.36165631e-01 -6.64678663e-02
6.33501470e-01 5.38551450e-01 8.36992383e-01 5.30435979e-01
9.32261109e-01 5.80401361e-01 -4.00803564e-03 7.74838701e-02
-6.60631582e-02 -9.99857366e-01 6.21033907e-01 1.29329014e+00
2.69527078e-01 -5.68358362e-01 -7.04386055e-01 1.08943534e+00
-1.97466111e+00 -1.17723334e+00 -1.47935599e-01 1.89085591e+00
1.38974619e+00 1.10033594e-01 -1.31168455e-01 -5.05494475e-01
6.93672240e-01 1.82003960e-01 -4.52786833e-02 -2.64950424e-01
6.86783344e-02 2.56006509e-01 5.73110998e-01 5.57382166e-01
-7.37752020e-01 1.43956709e+00 7.20635796e+00 9.96947050e-01
-6.80822670e-01 1.45916520e-02 9.40866545e-02 7.95194879e-02
-8.75224590e-01 7.01803267e-02 -1.47221768e+00 5.23853302e-01
6.77346289e-01 -9.07301247e-01 -9.77646280e-03 9.61644709e-01
2.60316193e-01 2.50386894e-01 -1.14952385e+00 6.42528832e-01
2.99989194e-01 -1.35758209e+00 6.31903291e-01 -2.17161521e-01
8.91639829e-01 -5.59956320e-02 -2.77076900e-01 5.84383190e-01
7.42753685e-01 -1.01811838e+00 8.43235016e-01 4.11355287e-01
7.52774894e-01 -6.75685048e-01 9.03057337e-01 6.40925229e-01
-1.37144744e+00 2.83237100e-01 -6.53330147e-01 2.19822098e-02
6.33454442e-01 7.66174197e-01 -1.38911331e+00 5.93360662e-01
-4.01901528e-02 7.93128848e-01 -2.73134202e-01 9.22302961e-01
-8.30542803e-01 5.64628720e-01 5.61201796e-02 -5.29521465e-01
2.96887010e-01 6.21285737e-02 4.48007077e-01 1.44287455e+00
7.18562484e-01 -3.94923333e-03 3.07224035e-01 1.10904229e+00
-5.93419254e-01 5.06854653e-01 -8.37581277e-01 -1.49994344e-01
8.90468836e-01 1.03068590e+00 -1.24115720e-01 -5.32299519e-01
-3.39762300e-01 8.70148838e-01 5.69685221e-01 2.52242804e-01
-8.07850599e-01 -5.85745752e-01 6.29676521e-01 -3.99175882e-02
4.64158028e-01 -5.75316511e-02 -8.59563276e-02 -1.43167853e+00
7.65458047e-02 -8.48415196e-01 1.04202084e-01 -7.26848841e-01
-1.53130960e+00 5.76589286e-01 1.77827284e-01 -1.29010093e+00
-6.66879952e-01 -1.87755853e-01 -7.86665618e-01 1.04055965e+00
-1.86849868e+00 -1.25184727e+00 8.75108391e-02 4.42535341e-01
8.29652250e-01 -3.47520709e-01 9.09281194e-01 2.60266453e-01
-3.40549827e-01 8.48453820e-01 -5.19426502e-02 4.08908486e-01
1.00649810e+00 -1.27482855e+00 8.83265972e-01 8.95262301e-01
3.07610542e-01 9.21427131e-01 7.37985432e-01 -8.73652756e-01
-9.85188127e-01 -1.40156686e+00 1.97843695e+00 -2.16181383e-01
6.83406830e-01 -3.34096342e-01 -5.01177609e-01 8.23097408e-01
6.41033173e-01 -4.88865405e-01 9.20597374e-01 1.83604419e-01
-3.75224650e-01 7.24207237e-02 -6.75506830e-01 7.54073560e-01
7.78503954e-01 -2.63541102e-01 -8.54361475e-01 5.70004404e-01
9.39757466e-01 -5.33105671e-01 -6.86718702e-01 4.68508266e-02
4.94378805e-01 -4.33684468e-01 6.52376592e-01 -5.92668355e-01
8.57698619e-01 -2.19406649e-01 1.87183842e-01 -1.72154260e+00
-3.26893777e-01 -6.31726801e-01 -2.05794290e-01 1.56483948e+00
1.13165832e+00 -3.63566101e-01 4.75854695e-01 5.20081997e-01
-1.05407819e-01 -5.60633898e-01 -7.40266323e-01 -8.10107052e-01
6.31358176e-02 -2.14768037e-01 8.38724911e-01 7.90365636e-01
3.90547961e-01 8.32872033e-01 -6.51023984e-01 2.32081339e-02
3.76627207e-01 2.45727792e-01 8.11523259e-01 -8.25865626e-01
-3.06049585e-01 -2.69723594e-01 6.85282499e-02 -1.44538045e+00
4.59932983e-01 -1.02253985e+00 3.70629758e-01 -1.87910378e+00
3.32071841e-01 -5.99190593e-01 -5.69021180e-02 3.46325427e-01
-6.92346752e-01 -1.44781977e-01 1.09247983e-01 6.25061095e-02
-6.36368573e-01 8.70273232e-01 1.26104879e+00 -1.41789138e-01
1.19866186e-03 -4.28102948e-02 -9.47333276e-01 4.61201906e-01
8.02955151e-01 -8.16613793e-01 -4.54451412e-01 -7.29415357e-01
6.47944927e-01 -3.59924580e-03 -9.60166678e-02 -5.45524120e-01
2.79189557e-01 -3.99729073e-01 2.26677228e-02 -6.35480583e-01
1.37935445e-01 -3.24950427e-01 -2.72365008e-02 1.88645735e-01
-7.41298795e-01 3.27981830e-01 -1.14652723e-01 5.31655848e-01
-5.13741076e-01 -4.30839419e-01 2.00687498e-01 -3.43375117e-01
-3.73854041e-01 3.10784787e-01 -3.79655600e-01 4.69084144e-01
8.17667544e-01 2.17491984e-01 -3.07017386e-01 -6.02607489e-01
-1.15073830e-01 4.63133961e-01 2.29845420e-01 5.80626309e-01
7.17688382e-01 -1.54908478e+00 -1.22978389e+00 2.74070036e-02
4.42174286e-01 1.44315898e-01 -1.64367825e-01 4.15178031e-01
-3.16579819e-01 7.32933879e-01 2.14950189e-01 -3.49327214e-02
-1.07595456e+00 2.06139654e-01 9.66643766e-02 -9.27091420e-01
-3.07767808e-01 1.03197062e+00 1.32234693e-01 -4.33241814e-01
3.60617153e-02 -2.90932864e-01 -2.70239562e-01 -3.64271849e-02
6.06646001e-01 1.69667266e-02 -1.61868513e-01 -6.07511938e-01
-1.40378058e-01 1.42338783e-01 -4.21414524e-01 -2.62536645e-01
1.04667413e+00 -1.09493189e-01 -1.39379159e-01 4.22294289e-01
9.44687307e-01 3.42130482e-01 -5.76279759e-01 -3.48394573e-01
2.07735430e-02 -3.17830056e-01 -4.32879090e-01 -6.49266899e-01
-6.79364860e-01 8.47619355e-01 -3.58403206e-01 -5.05658351e-02
6.74792528e-01 1.41529534e-02 1.29698968e+00 4.24724191e-01
4.59203422e-01 -1.10918427e+00 -1.36062130e-01 9.31790948e-01
8.34248960e-01 -1.14413929e+00 -1.03063904e-01 -5.85559607e-01
-9.46431220e-01 9.58843231e-01 8.74302924e-01 -3.09918895e-02
4.95563209e-01 2.10275680e-01 -6.35398775e-02 6.95139244e-02
-1.47967780e+00 -1.94946423e-01 3.41424584e-01 5.59537470e-01
9.28353965e-01 1.27476631e-02 -6.42215371e-01 7.24447072e-01
-6.10717535e-01 -1.14774041e-01 6.16583526e-01 5.78501284e-01
-4.37719613e-01 -1.53416467e+00 -4.90905605e-02 3.19487959e-01
-5.57485044e-01 -7.52255619e-01 -4.90035236e-01 7.45653987e-01
6.49307817e-02 6.98679030e-01 -1.17519610e-01 -2.11741358e-01
2.29870930e-01 2.31023937e-01 2.27401450e-01 -1.03401780e+00
-7.77478874e-01 7.14573264e-02 2.55729586e-01 -1.83975965e-01
-2.71403253e-01 -4.43641245e-01 -1.10935211e+00 -2.00287715e-01
-6.47825778e-01 5.25317490e-01 3.85902345e-01 8.96696150e-01
4.89791632e-01 3.73545945e-01 5.21703839e-01 -3.40709001e-01
-6.88540339e-01 -1.29595065e+00 -3.63409907e-01 3.12464863e-01
2.06397697e-01 -3.88978988e-01 -5.26236836e-03 1.76228613e-01] | [11.905048370361328, 9.024657249450684] |
ed2b05ff-fbf1-4aa4-81d3-ac965009da98 | learning-from-a-tiny-dataset-of-manual | 1812.00033 | null | https://arxiv.org/abs/1812.00033v3 | https://arxiv.org/pdf/1812.00033v3.pdf | Learning from a tiny dataset of manual annotations: a teacher/student approach for surgical phase recognition | Vision algorithms capable of interpreting scenes from a real-time video stream are necessary for computer-assisted surgery systems to achieve context-aware behavior. In laparoscopic procedures one particular algorithm needed for such systems is the identification of surgical phases, for which the current state of the art is a model based on a CNN-LSTM. A number of previous works using models of this kind have trained them in a fully supervised manner, requiring a fully annotated dataset. Instead, our work confronts the problem of learning surgical phase recognition in scenarios presenting scarce amounts of annotated data (under 25% of all available video recordings). We propose a teacher/student type of approach, where a strong predictor called the teacher, trained beforehand on a small dataset of ground truth-annotated videos, generates synthetic annotations for a larger dataset, which another model - the student - learns from. In our case, the teacher features a novel CNN-biLSTM-CRF architecture, designed for offline inference only. The student, on the other hand, is a CNN-LSTM capable of making real-time predictions. Results for various amounts of manually annotated videos demonstrate the superiority of the new CNN-biLSTM-CRF predictor as well as improved performance from the CNN-LSTM trained using synthetic labels generated for unannotated videos. For both offline and online surgical phase recognition with very few annotated recordings available, this new teacher/student strategy provides a valuable performance improvement by efficiently leveraging the unannotated data. | ['Tong Yu', 'Didier Mutter', 'Nicolas Padoy', 'Jacques Marescaux'] | 2018-11-30 | null | null | null | null | ['online-surgical-phase-recognition', 'surgical-phase-recognition'] | ['computer-vision', 'computer-vision'] | [ 6.13388479e-01 6.90431535e-01 -3.49514723e-01 -5.07328272e-01
-7.67938912e-01 -3.28670949e-01 4.61613744e-01 1.73344940e-01
-7.54599035e-01 6.68237209e-01 -8.26615691e-02 -3.90869200e-01
1.91018581e-01 -4.66240376e-01 -9.05469418e-01 -7.64251769e-01
6.29824102e-02 7.49095798e-01 8.65170434e-02 -1.13830045e-01
-1.28900945e-01 3.38681698e-01 -1.53591156e+00 5.00267029e-01
5.85472405e-01 1.19121766e+00 5.40717006e-01 8.13935220e-01
1.19328171e-01 8.95433307e-01 -5.56457102e-01 -9.97441560e-02
3.63086790e-01 -4.15391803e-01 -6.87116027e-01 1.00644849e-01
3.86336476e-01 -4.16300207e-01 -2.35701412e-01 6.59902394e-01
2.22325251e-01 -5.32483794e-02 2.43021473e-01 -8.32042754e-01
9.26962420e-02 7.29203522e-01 8.92615989e-02 -5.68544376e-04
2.96065271e-01 2.01544508e-01 4.40510929e-01 -5.44523418e-01
8.44570339e-01 5.72550893e-01 5.51840901e-01 7.69937873e-01
-1.22159111e+00 -3.76425683e-01 -1.10956758e-01 9.20487717e-02
-1.06662548e+00 -1.44985154e-01 6.52498901e-01 -6.36442423e-01
7.25446403e-01 1.25840783e-01 8.86400282e-01 1.64271450e+00
2.67521948e-01 7.26051271e-01 1.18177330e+00 -5.88755548e-01
1.93642989e-01 3.70453089e-01 -3.13352346e-01 6.94957316e-01
3.94770205e-02 4.77935433e-01 -3.74305844e-01 2.68994898e-01
8.63159359e-01 1.86986759e-01 -3.45302522e-01 -5.00057757e-01
-1.68333364e+00 6.38582587e-01 4.44450259e-01 5.95467091e-01
-3.15819770e-01 1.83366328e-01 7.74935842e-01 1.91416353e-01
2.92002738e-01 5.89745522e-01 -5.57945967e-01 -2.46054500e-01
-1.33401895e+00 -2.08113164e-01 1.06669021e+00 1.16567874e+00
7.01384008e-01 1.52966827e-01 -4.56141233e-01 3.70619208e-01
2.96415109e-02 1.71751082e-01 9.14931715e-01 -6.43201590e-01
9.32683349e-02 4.45968270e-01 -1.15753457e-01 -6.51201010e-01
-5.93767583e-01 -6.52462780e-01 -7.76039839e-01 6.42233491e-02
5.69421351e-01 -1.67574629e-01 -1.34238899e+00 1.52349627e+00
6.94582537e-02 4.30820435e-01 2.01584592e-01 7.77961731e-01
6.67724669e-01 2.70747781e-01 1.77346766e-01 -3.47588360e-01
1.26163185e+00 -1.20816445e+00 -6.39216661e-01 -1.10406615e-01
7.71435499e-01 -6.24891460e-01 7.60693550e-01 6.24187469e-01
-6.38962865e-01 -6.86141729e-01 -8.57103109e-01 9.86583158e-02
-4.12625968e-01 7.03296065e-01 4.34952438e-01 4.79039878e-01
-1.17199445e+00 5.65401435e-01 -1.09368384e+00 -3.65968525e-01
2.49918044e-01 7.54272640e-01 -5.52789569e-01 -8.77932757e-02
-8.41962874e-01 9.80304837e-01 6.56763911e-01 5.58766782e-01
-1.56553149e+00 -4.45589155e-01 -1.01471150e+00 8.22670534e-02
6.01475716e-01 -6.10675514e-01 1.52249467e+00 -1.65472496e+00
-1.70044172e+00 1.03831124e+00 1.67808712e-01 -9.32688236e-01
8.26394916e-01 -1.65270660e-02 -1.37568757e-01 2.94696182e-01
-3.90598774e-01 7.69644320e-01 1.01444566e+00 -1.40207350e+00
-3.62644404e-01 -8.27974454e-02 1.59272969e-01 -1.82188123e-01
-2.27796018e-01 -3.96215886e-01 -2.17846572e-01 -5.97509503e-01
-3.36485118e-01 -1.23922551e+00 -6.02299333e-01 -8.29755738e-02
-4.82702702e-01 1.81848764e-01 6.38906538e-01 -5.95605612e-01
8.00984442e-01 -1.89789653e+00 -9.45623741e-02 3.04808784e-02
9.82317999e-02 6.20226860e-01 -6.10232018e-02 4.04700339e-01
-2.04638347e-01 -3.08708638e-01 -1.88117936e-01 -6.72265708e-01
-4.98458594e-01 6.84523582e-01 -1.58457503e-01 3.80619109e-01
1.82157010e-01 6.37366891e-01 -1.24699354e+00 -7.14121640e-01
5.63222826e-01 4.76691365e-01 -5.50631106e-01 7.60734737e-01
-3.67593259e-01 1.07651055e+00 -2.61703759e-01 4.50084418e-01
1.46069974e-01 -1.10613897e-01 3.91662031e-01 -2.33479351e-01
-1.49038613e-01 -2.29115233e-01 -5.85142195e-01 2.27620101e+00
-8.47682476e-01 7.17267573e-01 -1.23382159e-01 -1.39991689e+00
8.96525025e-01 7.90455878e-01 6.38081312e-01 -3.75570178e-01
4.18044597e-01 5.58272362e-01 2.47055404e-02 -8.34456265e-01
4.29594457e-01 -1.73722535e-01 6.40972704e-02 -1.96423858e-01
6.37864232e-01 -4.48729424e-03 1.86044544e-01 -9.24327374e-02
1.28788888e+00 5.70407093e-01 4.05588031e-01 -1.43189095e-02
5.46256304e-01 5.70821702e-01 5.66141069e-01 8.13689649e-01
-1.15722500e-01 7.81347275e-01 2.79434562e-01 -7.70705998e-01
-1.11135483e+00 -7.07670271e-01 -2.14032902e-04 6.54017448e-01
-1.38316169e-01 -1.24822021e-01 -8.56172621e-01 -8.53932559e-01
-4.13097739e-01 5.65904200e-01 -8.90722573e-01 -1.05737217e-01
-7.69372165e-01 -1.94067091e-01 2.85913140e-01 4.60225433e-01
-5.55446073e-02 -1.38415706e+00 -1.16110599e+00 4.95934933e-01
-5.78534827e-02 -1.49678493e+00 6.74718898e-03 5.44047892e-01
-1.06158018e+00 -1.05418515e+00 -6.92300439e-01 -8.82237077e-01
9.76515770e-01 -2.53659546e-01 1.09360361e+00 -4.48777201e-03
-5.61084569e-01 4.24548954e-01 -5.94485581e-01 -5.09444952e-01
-6.53927982e-01 4.47608046e-02 -1.25848323e-01 2.52756983e-01
3.78263905e-03 -5.17348230e-01 -5.83754599e-01 -4.67651114e-02
-8.57085645e-01 3.89386088e-01 9.84636366e-01 1.34190869e+00
6.89587593e-01 -7.43956625e-01 3.27250570e-01 -1.41376591e+00
-1.01064913e-01 -4.02943522e-01 -6.00259960e-01 1.57457530e-01
-4.99571025e-01 8.92697498e-02 1.18722522e+00 -8.19822669e-01
-9.31541502e-01 6.55024767e-01 -6.21649623e-02 -1.08234954e+00
-5.22201180e-01 6.05124831e-01 3.62713575e-01 6.09118342e-02
6.32198870e-01 3.50595087e-01 1.13495737e-01 -3.31512272e-01
8.08456689e-02 4.80856091e-01 9.21462238e-01 -4.83005881e-01
5.28615892e-01 4.46186036e-01 2.84978170e-02 -6.81840539e-01
-1.02071989e+00 -6.75078511e-01 -7.40641296e-01 -4.92658138e-01
9.25790668e-01 -1.03278649e+00 -7.80910552e-01 2.31033936e-01
-1.13552928e+00 -6.47160232e-01 -6.57986403e-01 9.06380653e-01
-1.04589939e+00 -4.59338492e-03 -5.50307274e-01 -6.71625495e-01
-3.07397455e-01 -1.43401599e+00 1.15198803e+00 3.64419341e-01
-1.45037996e-03 -1.13004208e+00 2.90376544e-01 3.55265319e-01
5.19485116e-01 6.73845530e-01 4.83668000e-01 -1.15767479e+00
-5.19292295e-01 -5.75697601e-01 1.52269885e-01 4.82756555e-01
-6.06804788e-02 -1.19398542e-01 -1.20860243e+00 -2.98042625e-01
7.26461112e-02 -4.60660547e-01 7.03589618e-01 3.85284692e-01
1.51778066e+00 -3.23344111e-01 -4.22517300e-01 6.22084439e-01
1.60623944e+00 5.61606996e-02 4.22679245e-01 2.31093794e-01
5.71077704e-01 4.70302612e-01 8.74086320e-01 2.94684023e-01
2.40231976e-01 6.58446729e-01 8.29330027e-01 -4.42003220e-01
-3.22973765e-02 -2.95464993e-01 2.78899372e-01 9.14846778e-01
-2.94972986e-01 -1.26794815e-01 -9.43307638e-01 6.61160648e-01
-1.99366152e+00 -7.06914067e-01 1.22260891e-01 2.38257766e+00
8.95750642e-01 1.35374114e-01 -1.87571168e-01 -4.87941317e-02
3.90225768e-01 -1.76515132e-01 -2.99339265e-01 -3.04950178e-01
4.96671081e-01 4.62959528e-01 6.29728198e-01 1.11581147e-01
-1.12695312e+00 7.71133065e-01 5.01085711e+00 6.38529658e-01
-1.74283516e+00 2.40173981e-01 4.42220718e-01 1.09079741e-02
2.53716409e-01 1.05590045e-01 -5.25073588e-01 5.03933072e-01
1.43822634e+00 2.68466741e-01 1.72638804e-01 1.25177240e+00
5.00759184e-01 -3.03429693e-01 -1.47022998e+00 8.68587732e-01
3.65123123e-01 -1.41866994e+00 -1.95645511e-01 8.66442546e-02
4.96968359e-01 1.33957952e-01 -2.70335734e-01 4.23669100e-01
1.28291650e-02 -1.10474610e+00 7.02653229e-01 8.55156481e-01
1.00033486e+00 -1.29983515e-01 1.20188117e+00 6.06580555e-01
-7.48896658e-01 -1.31716490e-01 -2.03282554e-02 3.26158762e-01
1.72977731e-01 4.51819271e-01 -1.46948302e+00 8.11587453e-01
3.79668057e-01 6.92177474e-01 -4.19936836e-01 1.23359466e+00
-1.43832326e-01 6.77597404e-01 -2.62533635e-01 2.19620287e-01
3.50354850e-01 -1.47333916e-03 3.35133225e-01 1.51506364e+00
3.60955417e-01 -2.08297938e-01 6.41426384e-01 5.53842783e-01
1.33168280e-01 -5.15330844e-02 -7.42940545e-01 -9.08655003e-02
-2.98383739e-02 1.69249678e+00 -8.22018087e-01 -5.21280527e-01
-3.85490686e-01 6.32347763e-01 2.96913415e-01 8.78800824e-02
-6.14001036e-01 -3.80180962e-02 -1.12589352e-01 2.87329890e-02
2.09531993e-01 -2.89641768e-01 1.63440518e-02 -1.12310040e+00
-1.55899450e-01 -7.35288203e-01 3.14903975e-01 -7.74703622e-01
-8.27470124e-01 1.05924189e+00 -3.74425054e-02 -1.84531403e+00
-8.20810080e-01 -8.99827540e-01 -4.66129929e-01 4.57633525e-01
-1.57469893e+00 -1.58506119e+00 -7.98687339e-01 5.95605910e-01
8.18837762e-01 -2.92328540e-02 1.19803739e+00 1.97946504e-01
-4.11590040e-01 3.25132787e-01 -9.32157636e-02 2.39729002e-01
8.08966398e-01 -1.36789048e+00 -3.83356303e-01 5.36751032e-01
3.32547188e-01 3.44879061e-01 8.92325401e-01 -2.77801216e-01
-1.38141179e+00 -1.09385848e+00 6.00969017e-01 -2.31441125e-01
5.04224002e-01 -2.73511499e-01 -5.33001125e-01 9.40931439e-01
2.19280332e-01 4.35450226e-01 7.27916002e-01 -2.01236308e-01
8.14053267e-02 -1.58039942e-01 -9.46711302e-01 2.56887466e-01
5.71562290e-01 -3.67338240e-01 -5.58960676e-01 5.89333653e-01
4.76260185e-01 -9.59550023e-01 -1.12762272e+00 5.78792632e-01
5.38339138e-01 -8.09075296e-01 6.39407933e-01 -4.65191901e-01
7.36169994e-01 -1.33787066e-01 2.14084283e-01 -1.27596962e+00
6.49683237e-01 -3.99270803e-01 -4.45420779e-02 5.58332682e-01
3.30834478e-01 -3.36968809e-01 1.00499976e+00 3.04280877e-01
-6.34793043e-01 -9.93579984e-01 -8.74769151e-01 -7.07590759e-01
-4.52154011e-01 -3.89577240e-01 -1.45664558e-01 8.85334611e-01
-5.25475740e-02 2.93139238e-02 -4.89545435e-01 -1.36757210e-01
3.38756353e-01 1.08654037e-01 8.70074451e-01 -1.04530644e+00
-3.67722005e-01 1.61621660e-01 -6.00092888e-01 -6.62867129e-01
2.24112675e-01 -8.30449402e-01 5.45087099e-01 -1.38529062e+00
5.54920621e-02 -4.64203507e-01 -1.99270964e-01 7.58069217e-01
3.21181536e-01 1.96731463e-01 2.20136717e-01 2.49942154e-01
-4.29536551e-01 2.56537765e-01 1.29439747e+00 -4.88890037e-02
-9.97167528e-02 2.58379519e-01 1.61947966e-01 7.43229151e-01
3.38645995e-01 -6.85838521e-01 -2.24656731e-01 -2.64029920e-01
-1.97978884e-01 4.36286330e-01 5.52813590e-01 -1.25428963e+00
5.53346038e-01 9.11357701e-02 3.41294199e-01 -2.79916763e-01
5.90929389e-01 -1.13792145e+00 2.30876446e-01 8.14594746e-01
-4.16349888e-01 -1.69486225e-01 6.19253963e-02 7.00482845e-01
-6.34678245e-01 -4.74407375e-01 5.78287721e-01 -4.37793016e-01
-5.90720475e-01 2.64101535e-01 -4.24212098e-01 -2.19480991e-01
1.08373094e+00 -3.66366535e-01 -1.23289488e-01 -2.11421460e-01
-1.32644427e+00 -4.74878438e-02 2.94963956e-01 2.96468168e-01
4.86412495e-01 -8.44748616e-01 -4.54152703e-01 1.38808280e-01
1.87064692e-01 3.83944869e-01 4.02702421e-01 1.21058619e+00
-7.23286629e-01 6.49063170e-01 -4.73109484e-01 -1.01017666e+00
-1.02662885e+00 6.73686147e-01 2.95638889e-01 -7.75454402e-01
-5.96040368e-01 5.24423838e-01 1.90126821e-02 -4.05478984e-01
2.08791718e-01 -6.13362789e-01 -3.37944269e-01 2.87195127e-02
1.49130136e-01 -3.95449102e-01 3.44978273e-01 -4.07813936e-01
-3.82218957e-02 8.97382125e-02 -3.27060595e-02 1.53399214e-01
1.38386393e+00 4.73120689e-01 3.12601238e-01 5.58256090e-01
1.04025769e+00 -2.54070461e-01 -1.37076044e+00 -7.23456740e-02
6.64358884e-02 -2.33671412e-01 -3.47515866e-02 -7.43113756e-01
-1.05029905e+00 1.02525461e+00 7.07790792e-01 -8.73973779e-03
1.19806433e+00 -1.93809822e-01 6.07276440e-01 3.81404817e-01
7.61973619e-01 -8.34425986e-01 8.71207491e-02 3.45675766e-01
6.36283517e-01 -1.44124234e+00 -3.34616512e-01 -1.74751461e-01
-7.23402321e-01 1.50786114e+00 6.95846498e-01 -2.63608634e-01
4.98994619e-01 2.89223194e-01 2.26875842e-01 -2.92951688e-02
-7.32895374e-01 -3.19199592e-01 1.53586656e-01 5.00976622e-01
2.24075347e-01 1.83535829e-01 -1.36822596e-01 5.98197699e-01
-2.28549182e-01 4.20531839e-01 8.46116126e-01 9.96991396e-01
2.30322983e-02 -1.13252199e+00 -3.08651868e-02 5.38095772e-01
-4.21297610e-01 1.51483864e-02 1.89498857e-01 9.12747979e-01
1.97006807e-01 5.65918028e-01 1.31410763e-01 -3.67466539e-01
2.89814174e-01 -7.78234974e-02 4.59960133e-01 -1.06021440e+00
-1.19028902e+00 1.06276222e-01 1.00730166e-01 -8.10288370e-01
-7.65480757e-01 -4.24234241e-01 -1.09591806e+00 4.89960134e-01
-3.33918542e-01 2.03026414e-01 1.03120017e+00 1.12398970e+00
1.28272802e-01 8.44178379e-01 5.32495975e-01 -1.33743393e+00
-5.86542428e-01 -1.01270926e+00 -3.09787154e-01 2.61487633e-01
5.38147807e-01 -5.93240976e-01 -1.63386539e-01 4.83260483e-01] | [14.148472785949707, -3.2271206378936768] |
7cc79c46-8a60-46ac-925c-e2f179835587 | exploring-the-limits-of-chatgpt-for-query-or | 2302.08081 | null | https://arxiv.org/abs/2302.08081v1 | https://arxiv.org/pdf/2302.08081v1.pdf | Exploring the Limits of ChatGPT for Query or Aspect-based Text Summarization | Text summarization has been a crucial problem in natural language processing (NLP) for several decades. It aims to condense lengthy documents into shorter versions while retaining the most critical information. Various methods have been proposed for text summarization, including extractive and abstractive summarization. The emergence of large language models (LLMs) like GPT3 and ChatGPT has recently created significant interest in using these models for text summarization tasks. Recent studies \cite{goyal2022news, zhang2023benchmarking} have shown that LLMs-generated news summaries are already on par with humans. However, the performance of LLMs for more practical applications like aspect or query-based summaries is underexplored. To fill this gap, we conducted an evaluation of ChatGPT's performance on four widely used benchmark datasets, encompassing diverse summaries from Reddit posts, news articles, dialogue meetings, and stories. Our experiments reveal that ChatGPT's performance is comparable to traditional fine-tuning methods in terms of Rouge scores. Moreover, we highlight some unique differences between ChatGPT-generated summaries and human references, providing valuable insights into the superpower of ChatGPT for diverse text summarization tasks. Our findings call for new directions in this area, and we plan to conduct further research to systematically examine the characteristics of ChatGPT-generated summaries through extensive human evaluation. | ['Wei Cheng', 'Haifeng Chen', 'Xinlu Zhang', 'Yan Li', 'Xianjun Yang'] | 2023-02-16 | null | null | null | null | ['abstractive-text-summarization'] | ['natural-language-processing'] | [ 1.68308750e-01 4.67788041e-01 -3.77444565e-01 -1.84139058e-01
-1.20551610e+00 -5.27173102e-01 9.27740157e-01 7.61471391e-01
-2.67374903e-01 1.07312191e+00 1.24141347e+00 -9.74443853e-02
5.12332357e-02 -4.19154853e-01 -2.66721666e-01 -3.13873947e-01
-5.67421643e-03 5.38033903e-01 1.97095454e-01 -4.73238379e-01
9.71787274e-01 -2.34886669e-02 -1.16357899e+00 6.74507022e-01
1.44109142e+00 2.21532926e-01 1.64348468e-01 1.05437922e+00
-4.21293616e-01 7.63104379e-01 -1.22907889e+00 -4.79330003e-01
-3.45839351e-01 -7.39124060e-01 -1.08422041e+00 -1.00700416e-01
5.45879960e-01 -1.81753457e-01 -3.25622499e-01 8.28709126e-01
9.30046618e-01 3.80556762e-01 6.99065626e-01 -8.60271335e-01
-6.41798615e-01 1.15069735e+00 -6.99718595e-01 4.44010735e-01
8.13893735e-01 3.71872000e-02 1.39842629e+00 -4.61670578e-01
8.03200841e-01 1.48702562e+00 5.52349389e-01 5.85462153e-01
-9.40034807e-01 -2.13490263e-01 1.32887423e-01 -9.01030973e-02
-7.51077831e-01 -6.28142893e-01 4.11729485e-01 -1.96415074e-02
1.48056579e+00 8.30048621e-01 4.38640267e-01 9.89342630e-01
6.61812544e-01 1.32675016e+00 6.31062925e-01 -3.24855775e-01
1.23018689e-01 -3.46126705e-02 6.16264760e-01 2.10502416e-01
6.81040764e-01 -7.37232625e-01 -8.67283702e-01 -5.02732038e-01
4.47369106e-02 -3.87914300e-01 -3.52028936e-01 5.31924307e-01
-1.45036638e+00 8.44004929e-01 -9.17185694e-02 4.15734529e-01
-4.61827099e-01 -1.49291977e-02 1.00711989e+00 3.58270794e-01
9.15866256e-01 9.97453570e-01 -2.71017939e-01 -5.41260719e-01
-1.19982994e+00 6.28105581e-01 1.35468876e+00 1.19729006e+00
3.41130674e-01 -4.60916609e-02 -8.58966470e-01 1.01175892e+00
-2.16941565e-01 5.29875576e-01 8.50315154e-01 -9.72438574e-01
9.81341362e-01 5.99478304e-01 6.77229092e-02 -1.09421861e+00
-3.22793096e-01 -1.88256174e-01 -1.04137182e+00 -7.99377024e-01
-2.37421304e-01 -3.57709765e-01 -3.60337764e-01 1.16298330e+00
-2.10680738e-01 -5.21634281e-01 3.05291146e-01 3.85937184e-01
1.51899970e+00 1.14855361e+00 -1.36674628e-01 -6.77590430e-01
1.20833445e+00 -1.12709534e+00 -8.91838968e-01 -2.58815408e-01
7.08105922e-01 -9.32249069e-01 9.74116147e-01 3.18307817e-01
-1.43344820e+00 -3.28428894e-01 -8.19510818e-01 -2.77220488e-01
1.83159802e-02 2.28473008e-01 4.74462718e-01 3.93566996e-01
-1.15335262e+00 7.71082103e-01 -7.45469987e-01 -8.80761266e-01
2.80775309e-01 -2.05321629e-02 -5.66560104e-02 2.11943582e-01
-1.00436950e+00 8.68125737e-01 5.27213216e-01 -2.54797518e-01
-3.06354284e-01 -4.79912996e-01 -6.28667116e-01 2.13407010e-01
3.71326894e-01 -9.40237463e-01 1.72399473e+00 -1.47821039e-01
-1.52490902e+00 5.90556502e-01 -4.45178270e-01 -6.79944396e-01
4.30261165e-01 -5.95744669e-01 -1.91586196e-01 2.33195797e-01
3.43569964e-01 4.41222548e-01 3.01970214e-01 -8.88328791e-01
-7.59785593e-01 -7.04210624e-02 -2.57854104e-01 4.42447752e-01
-3.45568687e-01 3.04857165e-01 -3.11637431e-01 -8.06528986e-01
-2.57875741e-01 -6.50442541e-01 -3.14684033e-01 -1.03087461e+00
-1.12344944e+00 -6.72894537e-01 5.61134875e-01 -7.86169171e-01
1.74165273e+00 -1.59590995e+00 1.61987528e-01 -4.65159476e-01
3.11375380e-01 4.55929250e-01 -2.98257679e-01 1.38016891e+00
4.92921948e-01 4.68436480e-01 -1.63937464e-01 -5.64571202e-01
9.71666053e-02 -4.56131622e-02 -6.68568492e-01 1.51184518e-02
-7.70996064e-02 1.24614799e+00 -1.00869417e+00 -6.75720215e-01
-2.24787235e-01 -1.51705876e-01 -4.44162011e-01 8.52235854e-02
-5.00715435e-01 2.26675421e-01 -8.15393627e-01 3.68845791e-01
1.24108538e-01 -1.38600752e-01 -6.00531846e-02 2.09993646e-01
-4.00396645e-01 7.27399349e-01 -4.97781217e-01 1.61197853e+00
-8.41133147e-02 1.00340879e+00 -2.93438882e-01 -7.40407348e-01
7.90613413e-01 3.49426955e-01 2.09404171e-01 -5.50297260e-01
1.84962273e-01 1.22858204e-01 -3.48382980e-01 -6.05390608e-01
1.50389707e+00 1.72196135e-01 -5.18328249e-01 8.78193736e-01
-1.03473686e-01 -6.61868632e-01 8.96816552e-01 8.63673806e-01
1.19589221e+00 -3.84315461e-01 6.95452869e-01 -3.03889632e-01
3.15226763e-01 4.04032290e-01 2.93746263e-01 1.22945213e+00
1.04293495e-01 6.86124325e-01 8.68867040e-01 -5.20604476e-02
-7.95280933e-01 -5.30083299e-01 2.99567491e-01 1.18346381e+00
-8.27351213e-02 -1.02880383e+00 -8.53676736e-01 -6.06919765e-01
-1.44979835e-01 1.23526931e+00 -3.23092550e-01 -1.73884362e-01
-7.87710309e-01 -8.56299460e-01 7.93821394e-01 2.31649011e-01
6.06036067e-01 -1.44085991e+00 -5.68235397e-01 2.69781381e-01
-7.68573523e-01 -1.03482723e+00 -8.29722404e-01 -3.11623603e-01
-1.25898826e+00 -7.41768420e-01 -8.11129272e-01 -6.48029387e-01
3.69958073e-01 4.89187986e-01 1.28501022e+00 -1.82190672e-01
1.42965168e-01 4.86420959e-01 -7.10004508e-01 -7.29078710e-01
-9.05741811e-01 8.24271977e-01 4.39569587e-03 -5.10209978e-01
9.82926041e-02 -4.37585980e-01 -3.99194598e-01 -9.98219997e-02
-9.43824112e-01 2.25334570e-01 7.56797016e-01 5.43519080e-01
1.77422777e-01 -7.08733574e-02 1.07888782e+00 -1.14023077e+00
1.75219858e+00 -3.43268871e-01 1.37180001e-01 4.16335136e-01
-3.97475332e-01 1.12629458e-01 5.86139560e-01 -3.06352824e-01
-1.30392933e+00 -9.45392251e-01 -1.31781071e-01 5.67977011e-01
2.12973356e-01 9.40489650e-01 2.41782576e-01 6.18502855e-01
8.18639576e-01 5.29160798e-01 -1.69064462e-01 -5.44216573e-01
3.48798454e-01 8.95548701e-01 6.04105711e-01 -4.71917301e-01
3.86525363e-01 1.48078531e-01 -5.60782135e-01 -1.31989884e+00
-1.02089012e+00 -6.47279561e-01 -1.79097280e-01 4.90376614e-02
4.57544148e-01 -7.07096159e-01 -4.49205041e-01 2.87048012e-01
-1.38162458e+00 -1.45024776e-01 -5.37145257e-01 2.25846455e-01
-4.32963014e-01 9.38475490e-01 -8.49743783e-01 -6.53999269e-01
-1.34189665e+00 -5.63203514e-01 1.17843080e+00 5.23005426e-01
-1.02093089e+00 -9.67373192e-01 4.66799974e-01 4.13322210e-01
3.66240591e-01 2.23586187e-01 7.65218377e-01 -1.13841963e+00
-1.09114960e-01 -4.39047068e-01 2.23857015e-02 2.93997109e-01
2.02250630e-01 1.13740362e-01 -4.65002239e-01 -3.72114450e-01
3.29369400e-03 -2.89003104e-01 1.29770339e+00 5.67278922e-01
6.78776026e-01 -8.30188513e-01 -3.80329937e-01 -7.99280778e-02
7.60384142e-01 -8.15206841e-02 5.49656510e-01 3.35947901e-01
4.72356647e-01 5.89178324e-01 8.10480297e-01 6.20128453e-01
5.43545604e-01 1.26461223e-01 -1.86363921e-01 4.12393808e-01
-1.66208044e-01 -3.88682783e-01 5.55777967e-01 1.53494632e+00
-1.17316157e-01 -8.85896504e-01 -6.30866289e-01 5.44009030e-01
-2.09223104e+00 -1.21284282e+00 -2.72230536e-01 1.80348778e+00
1.04437554e+00 2.21366853e-01 5.07510528e-02 -1.01202883e-01
6.32052779e-01 6.10733271e-01 -4.23488349e-01 -8.13855171e-01
-3.39978516e-01 -1.29096910e-01 6.73117191e-02 2.28490680e-01
-7.13666737e-01 9.82940137e-01 6.28483868e+00 9.54140544e-01
-7.83264399e-01 -2.43910164e-01 4.77756232e-01 -1.73764884e-01
-3.44910741e-01 -1.61026344e-02 -9.06297088e-01 3.85546714e-01
1.00764394e+00 -1.02545011e+00 -1.23829477e-01 5.61901808e-01
5.69409728e-01 -4.02564347e-01 -9.91739452e-01 6.60779178e-01
4.15114373e-01 -1.56427932e+00 5.38563430e-01 -2.19096035e-01
1.04537404e+00 2.25628927e-01 -3.08681458e-01 7.07684517e-01
3.37565362e-01 -7.03208387e-01 4.44921285e-01 4.41777706e-01
4.51122195e-01 -6.23509347e-01 8.39543283e-01 7.63663828e-01
-7.44782090e-01 1.86598599e-01 -6.21803522e-01 -1.01395674e-01
5.46206415e-01 6.91688001e-01 -1.10096395e+00 9.57005441e-01
3.71533155e-01 9.90048289e-01 -7.03139246e-01 1.02786362e+00
-2.06860736e-01 8.36855769e-01 7.30549097e-02 -5.39903820e-01
2.34186485e-01 -1.06982961e-01 1.13459170e+00 1.64749503e+00
3.32507998e-01 2.12906331e-01 1.55021161e-01 4.47778642e-01
-3.10590684e-01 3.73422265e-01 -3.53378028e-01 -5.61152816e-01
3.63002419e-01 1.15517533e+00 -7.43191063e-01 -7.98506021e-01
-5.46251573e-02 8.36128950e-01 4.22304869e-02 2.37338766e-01
-4.50983673e-01 -6.63772821e-01 8.53888392e-02 -7.97485188e-02
3.81801389e-02 -2.38800511e-01 -4.04297799e-01 -1.37794209e+00
3.58388424e-02 -1.05467010e+00 3.92749548e-01 -6.88296616e-01
-1.13658106e+00 7.69744515e-01 2.90365189e-01 -1.02337992e+00
-3.33297312e-01 2.91277587e-01 -1.15958858e+00 4.31604028e-01
-1.18594289e+00 -6.79180026e-01 -1.35099411e-01 -7.77940080e-02
1.27621818e+00 -1.16262026e-01 5.83593428e-01 -3.01664203e-01
-7.63649046e-01 4.01656687e-01 2.95512706e-01 -1.81705415e-01
9.43180561e-01 -1.25890672e+00 1.00652909e+00 7.43687510e-01
-6.62954971e-02 7.09552526e-01 1.22644901e+00 -1.03167355e+00
-1.27278733e+00 -1.05528152e+00 1.41016126e+00 -6.01614416e-01
4.97923136e-01 4.78707030e-02 -9.57577050e-01 5.55895329e-01
8.79517794e-01 -1.13881338e+00 5.75886309e-01 4.89569530e-02
1.91320136e-01 2.17493325e-01 -7.61689603e-01 8.44885051e-01
7.92441368e-01 -3.00454423e-02 -1.22444212e+00 4.64908212e-01
1.13119328e+00 -3.85486573e-01 -6.65057719e-01 2.09115788e-01
3.26521695e-01 -8.79747629e-01 5.31309307e-01 -4.85446066e-01
7.42774427e-01 1.56863973e-01 3.28202337e-01 -1.77870393e+00
-2.27635466e-02 -1.18174839e+00 -2.74814487e-01 1.62901914e+00
3.19006890e-01 -7.56701291e-01 5.88489950e-01 4.17678833e-01
-5.81729174e-01 -6.24979675e-01 -4.51380670e-01 -6.37464941e-01
1.59438282e-01 5.33357225e-02 3.08818281e-01 5.74022114e-01
5.97232521e-01 9.60066080e-01 -3.00630540e-01 -4.86004144e-01
2.66663402e-01 2.56348968e-01 1.05682981e+00 -1.20988405e+00
6.79374114e-02 -7.85979152e-01 2.30127990e-01 -1.34307432e+00
5.21765463e-02 -8.59785259e-01 9.60092619e-02 -2.24286413e+00
6.81837440e-01 3.33662003e-01 4.31285113e-01 1.60303533e-01
-5.02692759e-01 -2.88670152e-01 1.17377274e-01 4.03964520e-01
-1.19145536e+00 8.37944925e-01 1.46599567e+00 -2.66420394e-01
-6.57699287e-01 3.21379960e-01 -1.28691256e+00 5.00149667e-01
1.13649452e+00 -4.17671293e-01 -3.27273309e-01 -3.13929468e-01
3.63650322e-01 3.26724917e-01 -3.19555223e-01 -6.69185162e-01
5.77559292e-01 -1.74503654e-01 -2.58540004e-01 -1.08133197e+00
-1.43006518e-01 2.99760818e-01 -4.15130764e-01 3.68494630e-01
-7.44096160e-01 1.99452460e-01 2.24091828e-01 5.32391012e-01
-4.01716888e-01 -2.33255357e-01 2.11500168e-01 -2.68464357e-01
-2.93134093e-01 9.56174266e-03 -8.02360535e-01 5.41348636e-01
4.41735476e-01 -2.41339356e-01 -6.92393541e-01 -7.32654870e-01
-1.49463609e-01 7.02984273e-01 2.78322786e-01 4.56261575e-01
6.09095395e-01 -7.29262948e-01 -1.33055413e+00 -4.85210210e-01
5.58669418e-02 1.12129211e-01 3.19048524e-01 9.22427118e-01
-5.16666472e-01 1.01097667e+00 1.50936767e-01 -3.39987487e-01
-1.38362181e+00 1.59639701e-01 -3.62146318e-01 -7.34456301e-01
-8.94140780e-01 4.95716184e-01 1.31463185e-02 -2.87238181e-01
1.13121822e-01 -6.31381571e-01 -4.57501441e-01 3.62856627e-01
8.49207640e-01 8.96772504e-01 5.24003878e-02 -2.76025981e-01
4.55907993e-02 1.00586213e-01 -5.42482734e-01 -6.19848967e-02
1.44975340e+00 -4.05371100e-01 -4.95169431e-01 4.51142758e-01
8.94838750e-01 3.28705877e-01 -5.56912363e-01 -2.73133963e-01
4.69386369e-01 7.44533213e-03 -3.68780524e-01 -6.96194589e-01
-4.28589523e-01 4.76064324e-01 -6.78661704e-01 6.28612518e-01
9.09580350e-01 1.63221791e-01 1.14231718e+00 8.47950161e-01
1.42479688e-01 -1.23281944e+00 3.53306919e-01 9.36663270e-01
1.35678279e+00 -1.10106015e+00 4.93211299e-01 -8.62154290e-02
-1.03170609e+00 1.04083145e+00 3.33403528e-01 -1.38559612e-02
-1.24882959e-01 -2.80958474e-01 -3.08330178e-01 -3.88044983e-01
-1.20619595e+00 2.04589918e-01 4.19553012e-01 1.93616256e-01
6.29168332e-01 1.52101013e-04 -7.95626462e-01 6.73336029e-01
-7.89405286e-01 -2.64370888e-01 1.06601322e+00 1.04082084e+00
-8.74427795e-01 -8.62623274e-01 -2.43479624e-01 9.03880119e-01
-5.87871373e-01 -1.92421541e-01 -8.81247520e-01 6.40290499e-01
-9.14934576e-01 1.29737771e+00 -2.79028147e-01 -6.98015466e-02
5.38047791e-01 -2.14637443e-02 2.03022644e-01 -1.16113615e+00
-9.38172460e-01 1.65304951e-02 6.04931295e-01 -1.84887797e-01
-3.22963655e-01 -8.48339558e-01 -1.16828406e+00 -6.64451122e-01
-3.48173141e-01 7.98465967e-01 2.89149463e-01 7.18227565e-01
5.31233609e-01 6.09183550e-01 3.47538561e-01 -8.16871166e-01
-7.30085075e-01 -1.47979975e+00 -3.54390740e-01 -2.91600153e-02
2.51782328e-01 1.96400151e-01 -3.87168169e-01 -6.23111539e-02] | [12.483285903930664, 9.438592910766602] |
6b06ad29-303a-4c4f-9485-c83bdcb34a35 | fusionseg-learning-to-combine-motion-and | 1701.05384 | null | http://arxiv.org/abs/1701.05384v2 | http://arxiv.org/pdf/1701.05384v2.pdf | FusionSeg: Learning to combine motion and appearance for fully automatic segmention of generic objects in videos | We propose an end-to-end learning framework for segmenting generic objects in
videos. Our method learns to combine appearance and motion information to
produce pixel level segmentation masks for all prominent objects in videos. We
formulate this task as a structured prediction problem and design a two-stream
fully convolutional neural network which fuses together motion and appearance
in a unified framework. Since large-scale video datasets with pixel level
segmentations are problematic, we show how to bootstrap weakly annotated videos
together with existing image recognition datasets for training. Through
experiments on three challenging video segmentation benchmarks, our method
substantially improves the state-of-the-art for segmenting generic (unseen)
objects. Code and pre-trained models are available on the project website. | ['Kristen Grauman', 'Suyog Dutt Jain', 'Bo Xiong'] | 2017-01-19 | null | null | null | cvpr-2017 | ['unsupervised-video-object-segmentation'] | ['computer-vision'] | [ 5.28527558e-01 7.55364373e-02 -5.20619273e-01 -5.13766646e-01
-9.58586633e-01 -7.73811638e-01 2.67119706e-01 -5.77978015e-01
-4.48338479e-01 2.97746301e-01 1.01370044e-01 -1.33998215e-01
5.10809839e-01 -2.43104607e-01 -1.34453869e+00 -4.63070393e-01
-1.13053918e-01 2.39234224e-01 8.49306226e-01 2.75382787e-01
-5.92814088e-02 2.78820038e-01 -1.29486620e+00 7.83758998e-01
4.98445719e-01 1.13668096e+00 1.43479630e-01 1.04209328e+00
1.03130162e-01 9.73244250e-01 -2.51319975e-01 -2.52472430e-01
6.38848245e-01 -3.18507433e-01 -1.35980058e+00 6.54209435e-01
1.13261783e+00 -8.56798649e-01 -4.50792789e-01 8.34096432e-01
2.45973879e-05 1.89182252e-01 4.94050324e-01 -1.27445567e+00
-3.86415064e-01 5.42601109e-01 -5.09320557e-01 3.69860113e-01
1.56374022e-01 4.51957077e-01 7.99198031e-01 -7.57146716e-01
9.51050639e-01 1.07542336e+00 6.71831012e-01 6.82607293e-01
-1.22511935e+00 -1.95081398e-01 6.01381898e-01 -3.82761396e-02
-9.96087551e-01 -5.52892029e-01 5.64577043e-01 -7.31065929e-01
8.76412868e-01 9.12905782e-02 6.01759553e-01 1.19029868e+00
-2.26168379e-01 1.46757758e+00 4.02053535e-01 4.39035036e-02
4.23048772e-02 -2.33688101e-01 1.69783339e-01 8.07532191e-01
8.76195207e-02 -1.57754943e-01 -3.27640742e-01 1.19743526e-01
1.01902544e+00 -2.70657986e-02 -2.44428009e-01 -7.58935809e-01
-1.14979863e+00 4.70490068e-01 3.43623936e-01 8.85645002e-02
-3.85196596e-01 6.76892042e-01 3.06988835e-01 -9.91233438e-02
4.23932612e-01 -3.65403742e-02 -9.73309994e-01 -2.61293352e-01
-1.21230471e+00 2.65979409e-01 6.20033622e-01 1.04277682e+00
7.80519426e-01 -5.18018603e-02 -4.07200366e-01 6.71405733e-01
2.09815174e-01 2.27867171e-01 3.60756256e-02 -1.83915913e+00
3.20350796e-01 4.25093204e-01 2.13561192e-01 -4.50490117e-01
-2.44750872e-01 -2.30676219e-01 -2.07741871e-01 -4.94746724e-03
6.30957842e-01 -4.80544329e-01 -1.44905519e+00 1.68042123e+00
4.20762718e-01 6.79729223e-01 -7.16308653e-02 1.02780855e+00
9.38077152e-01 8.49758744e-01 3.43833566e-01 6.48989454e-02
1.05009496e+00 -1.63307369e+00 -3.88359010e-01 -5.39633691e-01
5.68162262e-01 -5.59117615e-01 6.97823584e-01 3.35250199e-01
-1.36476612e+00 -8.16227376e-01 -6.36071861e-01 -2.36378595e-01
-2.70702578e-02 1.17041074e-01 7.21074700e-01 3.85068238e-01
-1.23757911e+00 6.82564795e-01 -1.27812231e+00 -3.49678189e-01
8.91602814e-01 3.69381577e-01 -3.69474083e-01 -2.16304120e-02
-6.24492586e-01 2.25239754e-01 6.08972728e-01 1.76461995e-01
-1.40783286e+00 -7.23982334e-01 -9.95796084e-01 -3.16687971e-01
5.75511694e-01 -9.10458326e-01 1.58907092e+00 -1.72255087e+00
-1.12921178e+00 8.60725522e-01 -1.74134403e-01 -4.83423173e-01
7.07420766e-01 -5.31803310e-01 6.56662509e-02 6.28674507e-01
1.83248773e-01 1.58103573e+00 9.22621548e-01 -1.44001317e+00
-9.59951460e-01 2.47546472e-02 2.37116680e-01 5.82246147e-02
-7.30336159e-02 2.23692760e-01 -1.29075766e+00 -6.84550405e-01
-2.33581796e-01 -9.52359974e-01 -5.39449871e-01 1.52487233e-01
-2.43244275e-01 -1.84618995e-01 1.30985522e+00 -9.54661608e-01
8.86495411e-01 -1.92424381e+00 3.37272853e-01 -2.79062063e-01
-4.48345616e-02 4.03466165e-01 -5.11687338e-01 -1.98548347e-01
-2.88848940e-04 -1.14868507e-02 -5.01654327e-01 -4.91104841e-01
-1.94289804e-01 2.78184086e-01 -3.83184887e-02 3.15963656e-01
4.93745536e-01 1.23883390e+00 -9.72939074e-01 -6.19291425e-01
2.07617164e-01 4.21676219e-01 -8.58998895e-01 4.13506478e-01
-6.79901361e-01 5.46234310e-01 -3.65254283e-01 9.92883444e-01
4.37228233e-01 -3.41047585e-01 -8.63625258e-02 -2.22075284e-01
2.91258991e-01 -1.99449867e-01 -1.01429260e+00 2.19846511e+00
-4.42644209e-03 7.04086125e-01 3.12253654e-01 -1.10599518e+00
1.60915524e-01 2.45412678e-01 7.68706441e-01 -8.90881792e-02
1.19296335e-01 -3.26655544e-02 -3.70229542e-01 -9.22172248e-01
3.46396625e-01 2.77173787e-01 -4.95417193e-02 1.27994940e-01
5.92085063e-01 9.53162555e-03 5.30162990e-01 3.01656634e-01
1.04672229e+00 9.61441219e-01 -3.31116289e-01 -2.68617608e-02
3.50214869e-01 1.43893570e-01 8.14718187e-01 6.03154659e-01
-5.07041812e-01 9.95644748e-01 5.09261191e-01 -6.28308296e-01
-1.03108037e+00 -1.10922551e+00 -3.61981764e-02 1.41640985e+00
1.92660570e-01 -1.70640320e-01 -1.17890930e+00 -1.13291192e+00
-1.27432108e-01 2.51447082e-01 -6.27753079e-01 3.18385720e-01
-8.06744576e-01 -3.57553005e-01 2.96213061e-01 1.09280813e+00
3.09710234e-01 -1.15527236e+00 -5.19412875e-01 1.63138956e-01
-3.97688210e-01 -1.69022417e+00 -7.36474752e-01 3.70385535e-02
-8.96355093e-01 -1.23415232e+00 -8.75390410e-01 -1.24805892e+00
6.85495079e-01 4.11883056e-01 1.27152467e+00 1.82931423e-01
-4.24948156e-01 7.01112509e-01 -2.52528101e-01 1.22139156e-01
-2.80813128e-01 1.83183495e-02 -4.02769566e-01 1.11856535e-01
1.32956535e-01 -2.17864737e-01 -8.51127386e-01 2.86515862e-01
-1.14273715e+00 2.43436769e-01 4.71366584e-01 3.77159476e-01
7.44663954e-01 -4.32452023e-01 2.33132690e-01 -8.66127968e-01
-2.63601780e-01 -4.46776062e-01 -6.43084228e-01 1.84620813e-01
2.26906136e-01 -1.47789165e-01 3.10341507e-01 -2.59733111e-01
-8.77791405e-01 9.00185645e-01 -2.15484858e-01 -7.37464905e-01
-6.99154019e-01 1.45759448e-01 -1.81467682e-01 -1.68788418e-01
1.81195021e-01 1.00046530e-01 -2.60454088e-01 -4.56480861e-01
7.50607669e-01 4.89897847e-01 1.11773956e+00 -6.53334141e-01
6.01645410e-01 5.49291670e-01 -3.71484399e-01 -6.17518067e-01
-1.02927732e+00 -7.26684153e-01 -1.01906419e+00 -3.56890261e-01
1.39929032e+00 -1.32452965e+00 -1.81610629e-01 4.64522600e-01
-1.20837831e+00 -9.83522892e-01 -1.96285248e-01 1.69023409e-01
-8.84477258e-01 4.47681010e-01 -9.17073548e-01 -3.53372097e-01
-2.09347397e-01 -1.36968505e+00 1.44366169e+00 3.37335497e-01
-1.05531536e-01 -9.48990703e-01 -4.67600897e-02 7.69412756e-01
-3.43198292e-02 4.15403664e-01 2.73330629e-01 -5.04497170e-01
-1.09500659e+00 -6.01869784e-02 -2.89915740e-01 6.77201152e-01
-2.16248840e-01 2.94663817e-01 -8.74249220e-01 -2.00405672e-01
-3.70099097e-01 -5.71934044e-01 1.38495541e+00 7.18427658e-01
1.53795016e+00 -2.21617952e-01 -4.76875246e-01 9.78337467e-01
1.26135147e+00 1.85245685e-02 4.07305300e-01 1.13285810e-01
1.17131209e+00 5.96669674e-01 6.03779912e-01 2.33767971e-01
3.60814482e-01 5.61580122e-01 3.40396732e-01 -1.90697402e-01
-3.60018998e-01 -1.49991244e-01 5.17282963e-01 1.65954128e-01
-3.87538932e-02 -2.37777680e-01 -7.49921679e-01 8.47193122e-01
-2.10179377e+00 -1.09413183e+00 -1.72490552e-01 1.69619215e+00
8.02410126e-01 1.80250272e-01 5.51626980e-01 -4.67578948e-01
6.88880920e-01 2.01479286e-01 -6.03447556e-01 -1.31471783e-01
2.70429011e-02 -1.00898417e-02 6.09255910e-01 4.73684251e-01
-1.85674167e+00 1.31281316e+00 6.93796682e+00 5.51051974e-01
-9.24618065e-01 7.31070340e-02 1.08538043e+00 -2.06324816e-01
-1.14377148e-01 6.34643510e-02 -6.89438760e-01 2.98547655e-01
8.56919050e-01 4.69099551e-01 2.44387135e-01 1.01176488e+00
2.47662053e-01 1.26149440e-02 -1.22850800e+00 7.57327318e-01
3.25614065e-02 -1.42079890e+00 9.63221863e-02 -1.84129953e-01
1.33159208e+00 4.68307078e-01 -1.03375547e-01 1.59100637e-01
3.39234442e-01 -8.77331793e-01 1.00195539e+00 2.40323663e-01
4.74704444e-01 -4.55284387e-01 5.17945588e-01 2.29655746e-02
-1.21707821e+00 -1.04670905e-01 -6.92605823e-02 -4.79607959e-04
4.72731113e-01 1.89890474e-01 -4.69812870e-01 2.77564049e-01
7.75285482e-01 1.12038970e+00 -6.83056176e-01 1.32119882e+00
-2.28890747e-01 7.80857682e-01 -2.94928104e-01 5.69608867e-01
6.48708045e-01 -6.90893158e-02 2.19121054e-01 1.46112514e+00
-5.57595268e-02 1.20929793e-01 5.92787266e-01 8.88447225e-01
-2.89648145e-01 -3.55135173e-01 -2.65802592e-01 -1.79104120e-01
3.97148728e-02 1.54463029e+00 -1.16701782e+00 -6.62122786e-01
-6.59409583e-01 1.26401150e+00 9.65527445e-02 6.91097081e-01
-1.01134503e+00 1.89100914e-02 6.82625115e-01 3.77797429e-03
9.12446320e-01 -2.42850259e-01 -9.44524407e-02 -1.24240160e+00
-1.34929091e-01 -7.84398854e-01 4.54096317e-01 -7.29737639e-01
-9.76532996e-01 3.43135208e-01 1.21931151e-01 -1.02295256e+00
-2.74013430e-01 -6.80983782e-01 -6.33200288e-01 2.29730994e-01
-1.23118877e+00 -1.34354603e+00 -4.60290253e-01 6.02904260e-01
1.03669262e+00 3.14707518e-01 2.68585682e-01 2.45836183e-01
-7.55498111e-01 2.79782593e-01 -1.69235617e-01 6.62620962e-01
4.15289015e-01 -1.23547208e+00 6.99339092e-01 1.27659345e+00
3.34804028e-01 2.01603711e-01 5.04060924e-01 -6.83935642e-01
-1.29944134e+00 -1.43798959e+00 2.25127146e-01 -7.17242777e-01
4.66684341e-01 -4.41030234e-01 -7.46229231e-01 1.01675022e+00
5.41734219e-01 5.98300517e-01 4.34424013e-01 -2.26418465e-01
-2.02427655e-01 2.20188007e-01 -7.43120015e-01 4.18370217e-01
1.31970954e+00 -3.01012665e-01 -5.30793488e-01 5.45464158e-01
9.46868062e-01 -5.40774763e-01 -6.92785919e-01 5.48322737e-01
4.21740234e-01 -7.66152561e-01 1.13874340e+00 -8.40057969e-01
6.67874813e-01 -4.78385091e-01 -1.87015720e-02 -8.26235950e-01
-1.39334932e-01 -7.49627829e-01 -3.44400316e-01 1.01134527e+00
3.64474684e-01 1.62417442e-01 1.24955392e+00 8.53187025e-01
-3.17428082e-01 -7.45101333e-01 -5.09037137e-01 -5.24047852e-01
1.93586701e-03 -5.05122602e-01 -1.36068106e-01 6.22854173e-01
-4.28387403e-01 1.26806021e-01 -2.60920376e-01 1.66448265e-01
6.35384619e-01 2.05898494e-01 9.12076771e-01 -7.61250079e-01
-4.53540295e-01 -4.01465356e-01 -4.14010793e-01 -1.47739005e+00
4.05969530e-01 -7.93475688e-01 4.13384616e-01 -1.82647991e+00
5.19084632e-01 2.81120185e-02 -3.09847236e-01 8.00773740e-01
-4.45466489e-01 7.32772708e-01 2.98664570e-01 -1.26537085e-02
-1.33757317e+00 1.30808949e-01 1.24736357e+00 -2.54809618e-01
-3.15590173e-01 -5.68285100e-02 -4.62915570e-01 8.26176584e-01
4.76605237e-01 -3.66350234e-01 -4.91298467e-01 -6.54527307e-01
-4.21186388e-01 1.52183965e-01 5.65389156e-01 -1.03552043e+00
6.37356490e-02 -2.45584488e-01 7.65260339e-01 -7.34513998e-01
3.82902801e-01 -5.75246036e-01 -5.51453531e-02 3.35715979e-01
-3.88381600e-01 -2.90297866e-01 3.48857790e-01 5.95454633e-01
-1.66771352e-01 -9.37541276e-02 7.20985591e-01 -2.77697474e-01
-1.26491535e+00 6.36051655e-01 -2.94227213e-01 2.90417641e-01
1.20092297e+00 -3.11669648e-01 -1.88844666e-01 -2.63228953e-01
-8.94176900e-01 4.37830836e-01 6.50763869e-01 7.26929665e-01
4.80592161e-01 -9.79979217e-01 -6.10005260e-01 -1.81816630e-02
-1.60477102e-01 2.66223758e-01 3.17811757e-01 8.35362315e-01
-8.64814341e-01 3.39575708e-01 -2.30344772e-01 -8.64759326e-01
-1.13295519e+00 6.54164076e-01 4.69922155e-01 3.19757283e-01
-4.64147002e-01 1.29955816e+00 4.90463763e-01 -1.04347818e-01
4.81439650e-01 -6.03896499e-01 1.79039061e-01 -1.56840429e-01
5.21354973e-01 2.51208376e-02 -3.18927109e-01 -7.92969346e-01
-3.22908789e-01 5.77932835e-01 -1.07014082e-01 1.22485887e-02
1.44599342e+00 -1.82437792e-01 4.67025228e-02 1.56270176e-01
1.64457047e+00 -4.10466939e-01 -2.23416543e+00 5.77434991e-03
-4.03298959e-02 -4.23872560e-01 3.68416198e-02 -6.16585732e-01
-1.42802727e+00 7.17189074e-01 3.60639393e-01 -1.25251934e-01
1.12453520e+00 2.13772535e-01 1.09470284e+00 1.90899327e-01
7.05424473e-02 -1.20417917e+00 2.91664273e-01 3.22183043e-01
4.11902905e-01 -1.35521054e+00 -2.04929039e-01 -6.14107847e-01
-7.81668603e-01 1.19743776e+00 9.34022725e-01 -2.75949150e-01
4.20683593e-01 2.91055948e-01 1.97078377e-01 2.04333723e-01
-7.10026741e-01 -3.25520337e-01 5.19673824e-01 5.29279947e-01
4.20593709e-01 -3.18057686e-01 -2.44296659e-02 6.28031731e-01
4.42578018e-01 2.97240585e-01 4.55150187e-01 1.02513993e+00
-4.67508405e-01 -1.16297901e+00 -1.96245238e-01 2.87593424e-01
-7.41605163e-01 1.18133053e-01 -4.08532083e-01 4.50556159e-01
3.50252688e-01 8.48473907e-01 1.46276712e-01 -2.42698282e-01
-4.64442745e-02 2.66528362e-03 5.72046101e-01 -5.42207062e-01
-3.79673004e-01 4.56253141e-01 1.76875547e-01 -1.12936795e+00
-7.94917107e-01 -8.06824028e-01 -1.35565877e+00 3.20436776e-01
1.54650345e-01 -2.66552508e-01 3.90136212e-01 9.99283433e-01
3.42359543e-01 5.30077219e-01 1.82014078e-01 -1.38561928e+00
-6.67532682e-02 -5.81850231e-01 -1.09913103e-01 7.64824688e-01
4.57051873e-01 -3.04885119e-01 6.62151215e-05 7.90392339e-01] | [9.175517082214355, -0.08169513195753098] |
baa21f1a-5e89-40b7-b438-a20912c77b1d | evolving-dictionary-representation-for-few | 2305.01885 | null | https://arxiv.org/abs/2305.01885v1 | https://arxiv.org/pdf/2305.01885v1.pdf | Evolving Dictionary Representation for Few-shot Class-incremental Learning | New objects are continuously emerging in the dynamically changing world and a real-world artificial intelligence system should be capable of continual and effectual adaptation to new emerging classes without forgetting old ones. In view of this, in this paper we tackle a challenging and practical continual learning scenario named few-shot class-incremental learning (FSCIL), in which labeled data are given for classes in a base session but very limited labeled instances are available for new incremental classes. To address this problem, we propose a novel and succinct approach by introducing deep dictionary learning which is a hybrid learning architecture that combines dictionary learning and visual representation learning to provide a better space for characterizing different classes. We simultaneously optimize the dictionary and the feature extraction backbone in the base session, while only finetune the dictionary in the incremental session for adaptation to novel classes, which can alleviate the forgetting on base classes compared to finetuning the entire model. To further facilitate future adaptation, we also incorporate multiple pseudo classes into the base session training so that certain space projected by dictionary can be reserved for future new concepts. The extensive experimental results on CIFAR100, miniImageNet and CUB200 validate the effectiveness of our approach compared to other SOTA methods. | ['Yuhong Guo', 'Xuejun Han'] | 2023-05-03 | null | null | null | null | ['class-incremental-learning', 'few-shot-class-incremental-learning', 'incremental-learning'] | ['computer-vision', 'methodology', 'methodology'] | [ 1.20379306e-01 -2.27831841e-01 -1.23365782e-01 -3.53584975e-01
6.81114867e-02 -4.47071582e-01 4.91257578e-01 2.65081227e-01
-5.93805373e-01 7.71733880e-01 -1.22976772e-01 1.51471913e-01
-4.97768335e-02 -1.02212179e+00 -4.44459736e-01 -6.99032187e-01
1.76383451e-01 5.87962270e-01 5.81135392e-01 -2.85992682e-01
4.85136770e-02 4.50666338e-01 -1.73660731e+00 2.27848396e-01
9.36998546e-01 8.13010037e-01 6.56442344e-01 1.55762956e-01
-3.52593809e-01 6.02431357e-01 -4.31856275e-01 -7.42038935e-02
3.34739864e-01 -2.54274994e-01 -4.50129956e-01 4.78489935e-01
3.99651170e-01 -2.12704718e-01 -4.61568803e-01 1.01617455e+00
6.18751526e-01 6.94057822e-01 3.96620810e-01 -9.97417927e-01
-7.72518396e-01 5.78090072e-01 -5.59031606e-01 6.47516608e-01
3.84787619e-02 1.89597413e-01 5.81505895e-01 -1.27243221e+00
7.52989352e-01 1.01339233e+00 3.79008561e-01 6.58058286e-01
-9.08151925e-01 -8.25037181e-01 8.77733946e-01 7.02900290e-01
-1.57423866e+00 -4.53853756e-01 9.29174721e-01 -2.02868208e-01
6.80434108e-01 -1.31464526e-01 8.55568171e-01 8.89509380e-01
-1.40446573e-01 8.94609630e-01 8.32779527e-01 -4.06673431e-01
3.85617048e-01 3.96155089e-01 2.49387145e-01 6.18235409e-01
1.13070734e-01 3.65191400e-02 -3.71859521e-01 1.34776458e-01
5.77086270e-01 6.38812542e-01 -2.76809275e-01 -7.16483951e-01
-1.20803070e+00 6.54111326e-01 4.60908771e-01 3.95556420e-01
-3.17112982e-01 -4.63304907e-01 5.72345138e-01 4.67181295e-01
2.97036588e-01 8.65457132e-02 -5.53618908e-01 1.63403794e-01
-7.39655614e-01 -1.80796310e-02 3.48779738e-01 1.16456211e+00
9.96091425e-01 4.69593674e-01 -1.66433722e-01 1.11463428e+00
-7.59851858e-02 5.77963769e-01 1.04373014e+00 -1.72249109e-01
3.74706894e-01 8.36654544e-01 -1.71056569e-01 -7.71158934e-01
-4.24535245e-01 -9.82585073e-01 -7.96503961e-01 -3.10915429e-02
-4.07011360e-02 -6.65984675e-02 -1.28098643e+00 1.55238199e+00
5.93986452e-01 6.11787498e-01 8.96679536e-02 6.14318013e-01
7.65980780e-01 9.87142384e-01 1.39617085e-01 -7.14979589e-01
9.87963259e-01 -1.07870650e+00 -6.94419503e-01 -3.79595667e-01
3.70584369e-01 -4.76284564e-01 1.24346793e+00 6.11483514e-01
-5.53615808e-01 -1.10305929e+00 -1.21459246e+00 2.64081359e-01
-6.17731988e-01 5.01845255e-02 5.78445852e-01 3.32881421e-01
-4.19654608e-01 3.65218133e-01 -6.49044871e-01 -3.84514391e-01
6.24739468e-01 3.34772736e-01 -2.21313104e-01 -3.74033093e-01
-1.11313641e+00 5.90242505e-01 9.50411022e-01 2.60936432e-02
-1.16736102e+00 -6.24095440e-01 -5.95967591e-01 5.30444682e-02
6.51378155e-01 -2.37974852e-01 9.22686040e-01 -1.07576239e+00
-1.19459760e+00 4.98220950e-01 7.13727400e-02 -4.31589842e-01
1.48006424e-01 3.68122160e-02 -8.24948490e-01 -1.07366756e-01
-6.90452829e-02 5.32036066e-01 1.03817141e+00 -1.23815644e+00
-1.03484571e+00 -4.39084172e-01 8.58603790e-02 6.10801935e-01
-8.95618320e-01 -6.91048741e-01 -5.41117728e-01 -8.79847109e-01
1.49771288e-01 -9.42322493e-01 -1.63257331e-01 -2.13263944e-01
2.09402457e-01 -2.25081116e-01 1.24406409e+00 -3.19002658e-01
1.44106686e+00 -2.19639564e+00 3.67657810e-01 -4.38239425e-02
6.71399385e-02 7.34896719e-01 -2.90107906e-01 1.04085431e-01
-1.17502876e-01 -4.75922793e-01 -2.40245327e-01 -2.27574840e-01
-3.71317625e-01 5.58315992e-01 -3.96635622e-01 1.24842077e-01
-2.41145656e-01 7.08056688e-01 -1.00728488e+00 -2.30270684e-01
3.22739452e-01 2.45309487e-01 -3.86546493e-01 1.90046221e-01
-1.68569580e-01 5.06214499e-01 -3.08684677e-01 8.24129045e-01
7.41595328e-01 -6.34068325e-02 7.76591245e-03 -2.26978362e-01
2.89902594e-02 -4.66346502e-01 -1.40088797e+00 1.95957208e+00
-5.69909513e-01 3.20603669e-01 -4.66809481e-01 -1.31430495e+00
1.05396867e+00 6.52078688e-02 2.90778011e-01 -8.50661993e-01
7.56738037e-02 -6.24996461e-02 1.12174638e-01 -4.95011985e-01
2.43597671e-01 -3.46439660e-01 1.65521890e-01 2.13191792e-01
4.15714890e-01 2.14951202e-01 3.53578061e-01 6.30684868e-02
5.50051451e-01 -4.32203598e-02 3.96449387e-01 -2.36962293e-03
7.89554060e-01 -1.94617793e-01 9.02968109e-01 7.34500647e-01
-3.44749749e-01 3.57444823e-01 -4.04461533e-01 -9.64629650e-01
-8.15020382e-01 -1.13344944e+00 -1.64674446e-01 1.19103372e+00
4.30567831e-01 -1.23678558e-01 -2.38182455e-01 -1.08231676e+00
-8.77266899e-02 7.03312278e-01 -6.34696245e-01 -5.90769589e-01
-6.17059708e-01 -9.38609362e-01 -2.81384379e-01 4.81198579e-01
7.06602871e-01 -1.26143980e+00 -4.36446458e-01 5.63651562e-01
6.64159134e-02 -9.26387429e-01 -5.17646194e-01 2.06985474e-01
-1.04440928e+00 -9.80652094e-01 -7.51004517e-01 -1.10073626e+00
8.46537173e-01 5.45692682e-01 6.62340105e-01 -3.79315913e-02
-3.88375223e-01 5.18857539e-01 -7.09519804e-01 -5.66110075e-01
-9.80141982e-02 2.22891346e-01 3.60098988e-01 4.34075981e-01
4.17590290e-01 -8.48258138e-01 -5.58081031e-01 1.73150152e-01
-1.00155461e+00 -1.06280088e-01 6.04615748e-01 1.03812265e+00
9.23595071e-01 4.18243706e-01 8.92160535e-01 -1.11661899e+00
3.04833055e-01 -7.78882504e-01 -5.31049550e-01 4.56366569e-01
-7.99247622e-01 -1.64611146e-01 9.72026944e-01 -1.06670499e+00
-1.21219146e+00 7.11000711e-02 8.88922140e-02 -5.49844742e-01
1.09384492e-01 4.84901518e-01 -3.39005917e-01 -1.46267682e-01
6.62133455e-01 6.80852652e-01 -3.68549109e-01 -6.05545163e-01
4.38148111e-01 5.23413360e-01 5.05887151e-01 -5.09896219e-01
9.74849761e-01 4.47012395e-01 -4.08738166e-01 -6.88537002e-01
-1.02387619e+00 -4.58517581e-01 -9.43320215e-01 -3.13008219e-01
3.18284363e-01 -9.26895678e-01 -1.67885646e-01 4.87470269e-01
-6.10336840e-01 -1.91299886e-01 -7.90112138e-01 4.31193173e-01
-1.81757033e-01 4.17503595e-01 2.55464334e-02 -4.78441834e-01
-2.18248919e-01 -8.46811891e-01 5.54684401e-01 6.79225981e-01
4.74959314e-01 -1.05384016e+00 1.25056654e-01 7.03850463e-02
3.03895533e-01 8.04995280e-03 7.58485138e-01 -7.83011138e-01
-4.71086085e-01 -1.98688552e-01 -3.03672906e-02 4.67494607e-01
4.53973979e-01 -4.98588741e-01 -7.80875921e-01 -9.27332222e-01
1.64281085e-01 -3.21787000e-01 1.05274498e+00 -1.05169795e-01
1.28636289e+00 -2.73298711e-01 -4.11566198e-01 7.31264412e-01
1.51600993e+00 8.00185561e-01 2.70846575e-01 3.73142391e-01
8.22445273e-01 1.90069959e-01 8.44653249e-01 7.23588645e-01
3.82396877e-01 6.07074142e-01 2.37243846e-01 5.09607084e-02
-3.78298163e-01 -2.54808396e-01 1.83358826e-02 1.02575517e+00
1.16764463e-01 -1.23924084e-01 -8.55806112e-01 7.73535550e-01
-1.74908245e+00 -8.70823920e-01 6.07704163e-01 2.31143260e+00
8.09528410e-01 3.66442412e-01 -2.19981104e-01 1.40307412e-01
6.60156906e-01 1.14434332e-01 -1.01032412e+00 6.39082491e-02
-1.71264648e-01 1.86266288e-01 1.92240134e-01 3.45241487e-01
-1.21325433e+00 1.08345497e+00 5.48800945e+00 1.08568680e+00
-1.35312235e+00 3.55816126e-01 4.03900713e-01 -1.44494578e-01
-4.27013636e-03 8.23063701e-02 -1.01526356e+00 4.94308054e-01
6.13587916e-01 -3.51622730e-01 5.20548165e-01 9.50426340e-01
-1.79455310e-01 3.92220020e-02 -8.22864473e-01 1.30139744e+00
3.08199137e-01 -1.29465532e+00 4.95334953e-01 -3.43948483e-01
9.10894692e-01 -8.68517235e-02 2.20595449e-01 8.14632535e-01
-3.43708061e-02 -3.58916521e-01 4.13527995e-01 6.88183069e-01
6.36304617e-01 -8.95793974e-01 4.88245100e-01 5.69514096e-01
-1.32282305e+00 -6.33494496e-01 -7.28024900e-01 1.20214693e-01
-3.13329473e-02 4.16633368e-01 -7.18644202e-01 4.73783880e-01
6.68315589e-01 1.09793389e+00 -8.53549123e-01 1.20041144e+00
-3.59265991e-02 4.72354770e-01 -7.28305280e-02 2.60091960e-01
1.80189013e-01 -6.07970022e-02 5.47963321e-01 8.82667959e-01
2.95783788e-01 4.45398450e-01 6.97843850e-01 2.41435707e-01
1.51416259e-02 2.96825856e-01 -4.58644748e-01 2.35927746e-01
6.53453827e-01 1.08400869e+00 -8.62068772e-01 -7.39798129e-01
-4.56687987e-01 1.11847782e+00 4.54433948e-01 4.40093458e-01
-6.98514044e-01 -3.89862567e-01 1.83873087e-01 1.42559316e-02
5.63937604e-01 -1.88373819e-01 1.34666964e-01 -1.61096680e+00
-2.35785991e-02 -8.05984020e-01 7.76012063e-01 -5.04794061e-01
-1.18950009e+00 8.00699890e-01 7.74716260e-03 -1.55050659e+00
1.62175655e-01 -2.92364866e-01 -4.35104311e-01 2.86314756e-01
-1.71914029e+00 -1.26800656e+00 -4.74309415e-01 1.03292179e+00
1.13351214e+00 -7.82985449e-01 5.95772624e-01 6.34628475e-01
-5.57365000e-01 7.11692154e-01 3.81082028e-01 -2.02727988e-01
6.51943207e-01 -8.67755413e-01 2.80347522e-02 8.63871217e-01
4.90635365e-01 4.51921970e-01 5.22416711e-01 -6.92182004e-01
-1.08044338e+00 -1.33166802e+00 4.32159662e-01 -3.21391784e-02
3.97122651e-01 -3.41847032e-01 -1.18928778e+00 6.56650841e-01
-1.57315865e-01 2.94806987e-01 6.92551017e-01 -2.30689701e-02
-2.88668424e-01 -7.22845137e-01 -9.52019453e-01 4.00085866e-01
1.04175663e+00 -2.91335493e-01 -8.06830823e-01 4.57079053e-01
8.04141104e-01 -2.32239366e-01 -4.84321326e-01 6.15950525e-01
3.23391020e-01 -5.90353012e-01 9.28420484e-01 -5.95067322e-01
-3.83372098e-01 -5.54954767e-01 -1.29572645e-01 -1.42961144e+00
-6.03544891e-01 -1.73752978e-01 -4.33447242e-01 1.12110043e+00
5.67199886e-02 -4.78729814e-01 7.31501818e-01 7.74974972e-02
-1.95577264e-01 -7.38695800e-01 -8.98303449e-01 -1.02613580e+00
-3.02829482e-02 -8.14646780e-02 6.94907010e-01 1.24129593e+00
-2.96413422e-01 3.62902045e-01 -6.15950048e-01 1.00679062e-01
6.11097693e-01 2.91968435e-01 6.11904621e-01 -1.33472395e+00
-3.17855477e-01 4.84273992e-02 -4.53392982e-01 -1.07348096e+00
-1.17470078e-01 -1.00413752e+00 -2.95459449e-01 -1.31444037e+00
3.80252659e-01 -6.56735301e-01 -8.64458740e-01 6.05555654e-01
-3.87601852e-01 9.39819887e-02 2.87335485e-01 4.43114817e-01
-9.22851384e-01 9.20538247e-01 1.33392382e+00 -5.15586317e-01
-5.36944866e-01 1.28956094e-01 -6.54648125e-01 6.03716612e-01
5.47824860e-01 -5.23078382e-01 -1.09833586e+00 -5.18075347e-01
-1.81286231e-01 -1.72497377e-01 -5.91116734e-02 -1.42098308e+00
5.23755431e-01 -2.66898096e-01 6.18824840e-01 -6.43419325e-01
1.48402497e-01 -9.11098003e-01 1.99853927e-02 4.26322997e-01
-1.98986396e-01 -2.71538377e-01 3.94345969e-01 9.21945155e-01
-1.91983923e-01 -2.60228068e-01 1.06988358e+00 -3.29891950e-01
-1.43152201e+00 9.12106693e-01 -6.44409284e-02 6.15945160e-02
1.33856034e+00 -3.77065271e-01 3.72057371e-02 6.23295307e-02
-1.33005762e+00 3.75212580e-01 2.47575760e-01 7.71692634e-01
9.45782185e-01 -1.45778382e+00 -5.60889065e-01 3.70723307e-01
4.59582239e-01 -1.81115512e-02 6.90194190e-01 2.68405199e-01
-1.28060848e-01 2.86104251e-02 -4.97121125e-01 -3.85197759e-01
-1.03049195e+00 1.22762716e+00 1.55235559e-01 -1.43876091e-01
-7.25102723e-01 9.81227279e-01 4.60041583e-01 -4.71825778e-01
3.57174039e-01 1.36676222e-01 -6.89059377e-01 3.45013767e-01
7.59224057e-01 1.73712358e-01 3.52903306e-02 -4.86172646e-01
-1.73240766e-01 5.03565252e-01 -6.64352834e-01 2.93954968e-01
1.42877591e+00 -5.40463507e-01 1.51413113e-01 8.68397892e-01
9.72261190e-01 -2.39838347e-01 -1.38307333e+00 -8.96772683e-01
-2.35227704e-01 -4.93403018e-01 1.58215426e-02 -9.23618436e-01
-1.33583236e+00 7.93951333e-01 1.25797021e+00 -2.06970364e-01
1.42742801e+00 -3.14048529e-01 9.98953044e-01 7.04730511e-01
5.25555849e-01 -1.36915815e+00 2.60399044e-01 4.10731196e-01
6.94397211e-01 -1.21495652e+00 1.57120824e-01 2.09941119e-02
-7.01354742e-01 1.19058597e+00 9.93983626e-01 -8.13265145e-02
9.20002222e-01 -3.23761702e-01 -2.33627036e-01 -5.82801625e-02
-6.62570298e-01 -2.29857564e-01 3.02420825e-01 6.37035847e-01
-1.64650217e-01 -1.50670201e-01 -2.22898602e-01 5.60195208e-01
2.19098836e-01 -1.17257312e-02 4.75626975e-01 1.03786421e+00
-8.30029249e-01 -1.25618494e+00 -5.27068451e-02 4.57527846e-01
1.85422823e-01 -1.05747916e-01 6.37270957e-02 6.09636247e-01
7.80172288e-01 5.77557445e-01 -2.55871918e-02 -4.20470089e-01
5.52591503e-01 1.55280694e-01 3.98902535e-01 -1.02334535e+00
-1.61686227e-01 3.11712176e-02 -5.31635761e-01 -7.30383024e-02
-3.40503991e-01 -6.48063183e-01 -1.14854991e+00 2.21910924e-01
-4.06463236e-01 2.22424477e-01 2.64615655e-01 9.89021242e-01
2.60749161e-01 6.25098586e-01 9.02218699e-01 -4.88834441e-01
-3.90641004e-01 -7.62017012e-01 -6.72895908e-01 3.42652410e-01
2.98636049e-01 -1.01664829e+00 -1.54129356e-01 2.38663524e-01] | [9.84841251373291, 3.371135711669922] |
ffe05410-f3ba-46df-b4e0-049e07e6db31 | one-shot-face-video-re-enactment-using-hybrid | 2302.07848 | null | https://arxiv.org/abs/2302.07848v1 | https://arxiv.org/pdf/2302.07848v1.pdf | One-Shot Face Video Re-enactment using Hybrid Latent Spaces of StyleGAN2 | While recent research has progressively overcome the low-resolution constraint of one-shot face video re-enactment with the help of StyleGAN's high-fidelity portrait generation, these approaches rely on at least one of the following: explicit 2D/3D priors, optical flow based warping as motion descriptors, off-the-shelf encoders, etc., which constrain their performance (e.g., inconsistent predictions, inability to capture fine facial details and accessories, poor generalization, artifacts). We propose an end-to-end framework for simultaneously supporting face attribute edits, facial motions and deformations, and facial identity control for video generation. It employs a hybrid latent-space that encodes a given frame into a pair of latents: Identity latent, $\mathcal{W}_{ID}$, and Facial deformation latent, $\mathcal{S}_F$, that respectively reside in the $W+$ and $SS$ spaces of StyleGAN2. Thereby, incorporating the impressive editability-distortion trade-off of $W+$ and the high disentanglement properties of $SS$. These hybrid latents employ the StyleGAN2 generator to achieve high-fidelity face video re-enactment at $1024^2$. Furthermore, the model supports the generation of realistic re-enactment videos with other latent-based semantic edits (e.g., beard, age, make-up, etc.). Qualitative and quantitative analyses performed against state-of-the-art methods demonstrate the superiority of the proposed approach. | ['Yaser Yacoob', 'Trevine Oorloff'] | 2023-02-15 | null | null | null | null | ['video-generation'] | ['computer-vision'] | [ 3.68247956e-01 2.13356808e-01 1.61480550e-02 -4.96268272e-01
-4.66617405e-01 -4.83847797e-01 6.35400236e-01 -7.48120010e-01
-4.96848375e-02 7.41713583e-01 1.53877035e-01 3.24281245e-01
-1.66489497e-01 -8.41084719e-01 -6.78192914e-01 -5.91375828e-01
1.21697165e-01 -1.61171673e-04 -4.48018521e-01 -2.26085976e-01
2.38038208e-02 6.72744155e-01 -1.83901358e+00 5.16953096e-02
7.71923304e-01 1.28065026e+00 -6.07100502e-02 6.01041496e-01
3.52186896e-02 4.68948215e-01 -3.36050153e-01 -1.11701059e+00
5.63753664e-01 -5.50014734e-01 -3.54002208e-01 2.11734951e-01
8.49894464e-01 -6.70883834e-01 -5.82559526e-01 1.07964134e+00
4.55120623e-01 1.36685699e-01 5.75797021e-01 -1.45084655e+00
-9.04806077e-01 -9.47798043e-02 -6.25146687e-01 -1.87729716e-01
5.48687160e-01 5.40544868e-01 7.83606231e-01 -1.00546587e+00
1.13314533e+00 1.40005314e+00 7.53999114e-01 7.82489717e-01
-1.24382806e+00 -1.11200058e+00 2.16421150e-02 2.77295969e-02
-1.56992936e+00 -8.94185901e-01 9.90144014e-01 -4.52883452e-01
5.12902319e-01 4.46274579e-01 6.48073673e-01 1.39625716e+00
-5.50866919e-03 2.01883361e-01 9.44609046e-01 -1.47290543e-01
-3.58082913e-02 -3.78804356e-02 -7.18257964e-01 1.09287786e+00
1.05614113e-02 3.53697568e-01 -6.60171807e-01 -1.01187736e-01
1.18481433e+00 3.25837061e-02 -3.69964242e-01 -1.73415944e-01
-9.31875527e-01 5.50076962e-01 2.83224639e-02 -1.61361948e-01
-1.67501450e-01 3.21435809e-01 1.81621373e-01 1.70055926e-01
4.77458626e-01 1.14946693e-01 -1.73209071e-01 -1.73918843e-01
-1.21968269e+00 3.84637505e-01 2.91617185e-01 1.46803379e+00
9.03572559e-01 3.86415213e-01 -2.96889722e-01 7.76911497e-01
4.41881120e-01 6.98101521e-01 1.00813642e-01 -1.44827306e+00
3.86489391e-01 3.63142997e-01 1.07789099e-01 -1.34453726e+00
1.59770817e-01 -1.14451818e-01 -1.08912539e+00 2.82882422e-01
-1.22520858e-02 -7.99391642e-02 -9.90274370e-01 2.21278858e+00
3.23162675e-01 5.20229161e-01 -2.61394083e-01 5.63166440e-01
6.69145226e-01 4.27382410e-01 7.23762959e-02 -5.12954593e-01
1.26786876e+00 -8.14542770e-01 -8.12606037e-01 -4.34246063e-02
1.39163256e-01 -9.86617804e-01 1.02118552e+00 8.35894421e-03
-1.48979020e+00 -8.76094639e-01 -8.66528034e-01 -2.21096680e-01
3.00940797e-02 2.04611510e-01 3.53696942e-01 7.40847290e-01
-1.11827028e+00 7.69117057e-01 -6.86900675e-01 -1.45048201e-01
5.44747770e-01 4.45030868e-01 -7.10017741e-01 -1.16976053e-01
-1.06320131e+00 4.62117463e-01 -7.25576282e-02 2.69580305e-01
-9.97588813e-01 -9.61355507e-01 -9.45022643e-01 -7.59316087e-02
3.99471998e-01 -9.66514945e-01 6.03520930e-01 -1.12287116e+00
-1.85419130e+00 1.05193388e+00 -7.40629807e-02 6.25834912e-02
8.32553625e-01 -1.84149072e-01 -4.75179076e-01 2.56930023e-01
-1.00304887e-01 8.64139736e-01 1.30334342e+00 -1.14201272e+00
-2.96187878e-01 -3.93496037e-01 1.09964631e-01 1.23792022e-01
-4.74852234e-01 1.46425754e-01 -5.92181981e-01 -1.10723937e+00
-7.07300752e-02 -1.04880834e+00 1.84144899e-01 6.28429890e-01
-3.42014372e-01 2.41824329e-01 9.19129610e-01 -9.11694288e-01
1.25705707e+00 -2.29375291e+00 3.63315523e-01 1.02760404e-01
3.22156489e-01 2.66973019e-01 -3.95231038e-01 1.70470893e-01
-2.42386088e-01 4.32685912e-01 -2.57530659e-01 -7.16771364e-01
1.25658184e-01 3.57336029e-02 -5.89786991e-02 5.62198460e-01
3.14707428e-01 8.62510681e-01 -7.11369455e-01 -6.15841568e-01
1.00457124e-01 8.21339130e-01 -9.39960003e-01 3.42436403e-01
-1.13318972e-01 3.66522253e-01 -2.05165848e-01 7.66778946e-01
9.29997683e-01 1.34198695e-01 1.69153407e-01 -5.69345593e-01
-2.01130342e-02 -3.21178138e-01 -1.28157902e+00 1.95913482e+00
-1.81275874e-01 3.15895289e-01 3.98797244e-01 -4.83420283e-01
1.01434565e+00 3.90565962e-01 6.05101347e-01 -3.88471067e-01
6.74365833e-02 1.34414643e-01 -3.56426984e-01 -4.33566183e-01
4.60564435e-01 -3.03582013e-01 2.35357329e-01 3.24560851e-01
2.64211148e-01 -2.21015126e-01 4.77797948e-02 -1.81491319e-02
7.82519758e-01 6.13138616e-01 -6.54651225e-02 -2.51020551e-01
5.15817344e-01 -7.60285556e-01 8.22995663e-01 1.65300295e-01
-2.26133138e-01 9.32472825e-01 3.88682574e-01 -2.30429515e-01
-1.43927324e+00 -1.15957844e+00 1.21246837e-01 6.39776051e-01
-3.07619721e-02 -4.78375673e-01 -1.10788512e+00 -4.69772846e-01
-1.75285965e-01 4.79358763e-01 -7.19339907e-01 -4.24133480e-01
-7.14968324e-01 -3.50724429e-01 7.43450403e-01 3.40571880e-01
7.17946470e-01 -8.82338941e-01 -3.57058555e-01 -7.18401000e-03
-3.08327705e-01 -1.11166275e+00 -9.44239497e-01 -8.21189106e-01
-6.56003952e-01 -9.73062873e-01 -7.65514672e-01 -6.06548369e-01
9.49943066e-01 -8.04052576e-02 8.34434688e-01 2.24288568e-01
-4.82327014e-01 3.55917156e-01 -2.09975690e-01 1.38978437e-01
-2.70492733e-01 -3.10366213e-01 3.16473126e-01 4.99784291e-01
-1.64489165e-01 -8.67039084e-01 -1.00731146e+00 4.82348889e-01
-1.02222621e+00 3.13331336e-01 1.93897113e-01 9.10247922e-01
5.27255237e-01 -1.45304739e-01 2.45905787e-01 -6.17930710e-01
2.21945718e-01 -1.81708232e-01 -3.62774163e-01 3.29806685e-01
-4.75017548e-01 -6.84763640e-02 6.65527701e-01 -4.77736175e-01
-1.27355647e+00 -1.50610372e-01 -2.76460618e-01 -8.87345493e-01
-1.83400083e-02 1.78586394e-02 -5.75139761e-01 -1.43107757e-01
3.42575520e-01 3.50892723e-01 3.67366850e-01 -3.33456278e-01
5.20760179e-01 3.54127318e-01 5.54337204e-01 -6.10705256e-01
1.18116748e+00 5.63628137e-01 -1.43057201e-02 -6.76884055e-01
-3.86317492e-01 2.68430144e-01 -6.10026181e-01 -2.84369767e-01
7.94236839e-01 -1.03911269e+00 -7.37682879e-01 6.67853892e-01
-1.01842058e+00 4.01733480e-02 -4.40715969e-01 1.90333173e-01
-9.43297207e-01 4.60047454e-01 -5.01546621e-01 -6.51645839e-01
-3.71467143e-01 -1.31444550e+00 9.44712460e-01 2.31072009e-01
-1.63701251e-01 -5.80681086e-01 -3.44223678e-01 5.23864627e-01
4.65779334e-01 6.61559403e-01 9.16863024e-01 1.72841445e-01
-6.86039329e-01 2.10791864e-02 -2.40901455e-01 5.03591478e-01
3.68789315e-01 4.42833066e-01 -8.82686198e-01 -5.83359480e-01
-7.74177015e-02 -9.12315175e-02 3.90134603e-01 5.52048795e-02
1.11646736e+00 -6.69207752e-01 -8.25785622e-02 1.16986752e+00
1.27054048e+00 2.28218362e-01 9.56518054e-01 -2.71494538e-01
7.58929670e-01 6.96827710e-01 4.26468819e-01 7.81762242e-01
2.76753187e-01 8.67164433e-01 5.11058867e-01 3.89001779e-02
-3.42021644e-01 -5.24149418e-01 5.94612300e-01 7.63592422e-01
-6.08827114e-01 -9.92738679e-02 -4.14341480e-01 2.07220435e-01
-1.46339941e+00 -1.18457413e+00 4.43852544e-01 2.15235639e+00
9.05580103e-01 -1.98469460e-01 -1.51686937e-01 -9.63102728e-02
7.38376379e-01 3.29358637e-01 -5.21734774e-01 -1.74152926e-01
-1.29342318e-01 4.26552206e-01 1.65505141e-01 4.57398027e-01
-6.89296186e-01 9.67487812e-01 5.49353504e+00 9.48436260e-01
-1.05266321e+00 -5.11067174e-03 8.65848243e-01 -3.99722099e-01
-7.15568364e-01 -1.63529888e-01 -8.14768374e-01 6.88655674e-01
6.87541246e-01 -1.89947739e-01 7.29284585e-01 5.75015783e-01
2.83195823e-01 3.90248835e-01 -1.01469064e+00 1.26809239e+00
3.79099816e-01 -1.54908371e+00 3.14411879e-01 1.48846999e-01
8.12791705e-01 -8.60303223e-01 4.33966041e-01 7.98731372e-02
-2.92442534e-02 -1.13638461e+00 1.01155436e+00 7.49429286e-01
1.83482158e+00 -7.14822292e-01 2.22311363e-01 -3.88623625e-02
-1.38789713e+00 -1.27151027e-01 -6.60573915e-02 3.94989610e-01
3.09951007e-01 1.85822457e-01 -1.97584614e-01 6.50741339e-01
5.84650218e-01 6.00117385e-01 -2.91188270e-01 2.31245756e-01
-5.34004308e-02 8.27479139e-02 -2.74561316e-01 6.38792872e-01
-1.47721544e-01 -4.38584298e-01 4.77032989e-01 8.40557158e-01
9.30495799e-01 4.77269053e-01 -1.37540251e-01 1.05062151e+00
-4.07464206e-01 -4.93823066e-02 -5.76420248e-01 -1.82682112e-01
5.70356131e-01 9.74064827e-01 -1.76806003e-01 -1.73290223e-01
-2.29772002e-01 1.26592314e+00 7.13179633e-02 3.37057292e-01
-1.13537455e+00 -2.45910048e-01 1.27044463e+00 5.34944832e-01
2.50897020e-01 -2.97220737e-01 -3.10277510e-02 -1.21928215e+00
4.20521237e-02 -9.75208223e-01 -1.29133552e-01 -7.97425508e-01
-9.89690244e-01 6.95322037e-01 6.11538664e-02 -1.16125596e+00
-3.30577999e-01 -3.97228539e-01 -3.46581548e-01 8.65784049e-01
-1.26769972e+00 -1.55332589e+00 -4.35778707e-01 8.71159971e-01
4.65813488e-01 -3.16938162e-01 8.37936163e-01 5.84580183e-01
-6.44591630e-01 1.17515731e+00 -6.22000806e-02 1.29083902e-01
7.69793630e-01 -4.76648003e-01 3.67342532e-01 8.75794351e-01
-1.54118910e-01 6.79576337e-01 4.93913531e-01 -6.33873940e-01
-1.52586889e+00 -1.13194060e+00 7.46984005e-01 -1.84760571e-01
2.55604625e-01 -4.58412647e-01 -4.81308937e-01 7.52874911e-01
-2.41173722e-04 2.19662353e-01 6.71546459e-01 -4.03900802e-01
-5.21383047e-01 -3.97830844e-01 -1.46182024e+00 8.87272835e-01
1.70240974e+00 -7.63666809e-01 5.69097232e-04 -7.92451277e-02
6.57199442e-01 -3.26177239e-01 -1.08075011e+00 5.24810135e-01
8.13229918e-01 -1.20648217e+00 1.18668485e+00 -3.98757994e-01
7.17410862e-01 -1.74240693e-01 -2.88624406e-01 -9.88428593e-01
-3.93648535e-01 -1.19505072e+00 -1.06125884e-01 1.63875854e+00
-3.87759954e-02 -2.32972488e-01 9.08302128e-01 8.36977780e-01
-7.63477311e-02 -7.01118231e-01 -1.10720515e+00 -4.31465030e-01
-2.16911659e-01 -2.18619004e-01 9.57267463e-01 1.02809107e+00
-5.09369910e-01 -2.16056153e-01 -9.09955263e-01 -1.10960476e-01
7.68378198e-01 -1.59313515e-01 7.92731702e-01 -9.81787086e-01
-1.72108367e-01 -3.17316324e-01 -4.30582941e-01 -8.76619279e-01
3.12798947e-01 -6.00609243e-01 -2.41195679e-01 -7.59078741e-01
4.25473899e-02 -5.01545250e-01 -4.32617441e-02 4.30868268e-01
-6.27444685e-02 5.07585526e-01 4.99097347e-01 1.08027063e-01
-2.71028802e-02 8.57132494e-01 1.57157862e+00 9.60152298e-02
8.78029466e-02 -4.56206828e-01 -6.58228755e-01 8.11998785e-01
4.31209981e-01 -2.13239685e-01 -5.36704361e-01 -5.72096944e-01
1.11775041e-01 3.05744231e-01 4.56843823e-01 -8.99715066e-01
-8.94391239e-02 -4.74625051e-01 2.70551711e-01 8.48955214e-02
7.73947060e-01 -8.18569958e-01 8.59545410e-01 2.34974608e-01
-1.69323266e-01 2.22468346e-01 5.65383881e-02 4.71826375e-01
-1.56128734e-01 7.78331161e-02 1.07301259e+00 -1.72559753e-01
-3.52266192e-01 9.61867034e-01 2.55959213e-01 -1.12780901e-02
1.09750903e+00 -6.98483944e-01 -1.62805721e-01 -4.04794872e-01
-6.00655377e-01 -3.02963793e-01 7.96892822e-01 5.13430238e-01
7.48233914e-01 -1.65876865e+00 -8.74183476e-01 7.69425035e-01
-3.07186097e-02 -5.53230681e-02 6.55249059e-01 5.58689117e-01
-7.03181922e-01 -1.45945609e-01 -6.77100956e-01 -2.19730020e-01
-1.38398302e+00 2.24454805e-01 1.35009184e-01 -1.43399611e-01
-4.16769892e-01 1.03430009e+00 2.56484240e-01 -1.12750307e-01
-5.81045225e-02 1.37157217e-01 2.17871174e-01 7.41335228e-02
4.16654617e-01 5.07813632e-01 -3.83486062e-01 -1.01381791e+00
-1.55288830e-01 9.23631728e-01 3.62663344e-02 -1.45345911e-01
1.27267647e+00 -2.88928568e-01 -1.10052481e-01 -1.69131592e-01
1.35859275e+00 5.16091362e-02 -1.82233417e+00 5.67428507e-02
-5.71998715e-01 -9.27031934e-01 -3.33352506e-01 -5.55388272e-01
-1.32569599e+00 7.96298563e-01 6.94773674e-01 -3.63094926e-01
1.27705991e+00 -4.25256282e-01 9.22970891e-01 -3.36167216e-01
4.83030856e-01 -1.16067934e+00 1.91551864e-01 1.63914159e-01
1.09034514e+00 -8.71080279e-01 1.01143502e-04 -5.65497160e-01
-4.41496640e-01 9.48613167e-01 5.92030287e-01 3.20980549e-02
6.83745444e-01 1.09836079e-01 -2.11589500e-01 1.46537200e-01
-5.94145536e-01 3.66666675e-01 2.08101019e-01 5.78757405e-01
3.04243743e-01 7.05289282e-03 -1.16145588e-01 4.60277766e-01
-4.23661798e-01 1.06665760e-01 3.17918688e-01 6.80225968e-01
9.94335040e-02 -1.14590037e+00 -2.72896022e-01 1.99704453e-01
-4.21378613e-01 -4.55461591e-02 9.16642025e-02 7.16626346e-01
6.88929141e-01 6.57859206e-01 5.46652675e-02 -5.00721157e-01
2.83009559e-01 8.20449218e-02 6.73574150e-01 -3.49467099e-01
-5.07269859e-01 4.12792377e-02 -1.00278839e-01 -7.50048876e-01
-4.69407290e-01 -4.90651339e-01 -8.17676723e-01 -8.98943603e-01
-3.23320813e-02 -2.32403800e-01 4.35575098e-01 6.24079525e-01
7.64172494e-01 1.89977139e-01 7.65262723e-01 -8.98063898e-01
-4.11326051e-01 -7.82539487e-01 -6.75127983e-01 8.53215218e-01
1.89620957e-01 -7.43550658e-01 -3.18849534e-01 6.25485003e-01] | [12.607306480407715, -0.24407434463500977] |
7e88e7f7-8d2f-4a43-b9e2-a291f53ca8de | mc-beit-multi-choice-discretization-for-image | 2203.15371 | null | https://arxiv.org/abs/2203.15371v4 | https://arxiv.org/pdf/2203.15371v4.pdf | mc-BEiT: Multi-choice Discretization for Image BERT Pre-training | Image BERT pre-training with masked image modeling (MIM) becomes a popular practice to cope with self-supervised representation learning. A seminal work, BEiT, casts MIM as a classification task with a visual vocabulary, tokenizing the continuous visual signals into discrete vision tokens using a pre-learned dVAE. Despite a feasible solution, the improper discretization hinders further improvements of image pre-training. Since image discretization has no ground-truth answers, we believe that the masked patch should not be assigned with a unique token id even if a better tokenizer can be obtained. In this work, we introduce an improved BERT-style image pre-training method, namely mc-BEiT, which performs MIM proxy tasks towards eased and refined multi-choice training objectives. Specifically, the multi-choice supervision for the masked image patches is formed by the soft probability vectors of the discrete token ids, which are predicted by the off-the-shelf image tokenizer and further refined by high-level inter-patch perceptions resorting to the observation that similar patches should share their choices. Extensive experiments on classification, segmentation, and detection tasks demonstrate the superiority of our method, e.g., the pre-trained ViT-B achieves 84.1% top-1 fine-tuning accuracy on ImageNet-1K classification, 49.2% AP^b and 44.0% AP^m of object detection and instance segmentation on COCO, 50.8% mIOU on ADE20K semantic segmentation, outperforming the competitive counterparts. The code will be available at https://github.com/lixiaotong97/mc-BEiT. | ['Ling-Yu Duan', 'Ying Shan', 'Zixuan Hu', 'Kun Yi', 'Yixiao Ge', 'Xiaotong Li'] | 2022-03-29 | null | null | null | null | ['self-supervised-image-classification'] | ['computer-vision'] | [ 3.33881855e-01 4.10304606e-01 -5.17574549e-01 -4.36866611e-01
-9.54386771e-01 -5.46940923e-01 2.71094054e-01 -1.64172500e-01
-5.61324596e-01 5.05173087e-01 -4.77640420e-01 -2.35905960e-01
3.29424471e-01 -5.24245322e-01 -1.19454467e+00 -8.48302960e-01
2.90076107e-01 4.63862956e-01 1.81287542e-01 2.07858145e-01
-1.47497535e-01 -1.21576190e-01 -1.48375964e+00 5.86410284e-01
8.49371016e-01 1.52819121e+00 5.65835059e-01 5.06661355e-01
-1.60476342e-01 7.49006867e-01 -6.44714832e-01 -5.79137564e-01
3.11230481e-01 -1.24147341e-01 -9.80444908e-01 4.31810766e-01
6.74634635e-01 -1.83049738e-01 1.86531413e-02 1.22053480e+00
1.08997196e-01 -1.43715695e-01 7.35681176e-01 -1.23122370e+00
-8.09219778e-01 9.58796978e-01 -7.47763872e-01 -8.84392932e-02
-3.64810497e-01 2.85333723e-01 1.10325003e+00 -1.02891171e+00
5.04508257e-01 9.88701105e-01 5.21763206e-01 6.45279884e-01
-1.41226172e+00 -7.03752398e-01 5.14922976e-01 3.26486200e-01
-1.64339876e+00 -1.28319532e-01 4.66268986e-01 -4.62414503e-01
7.54504383e-01 1.69156238e-01 5.67558706e-01 1.03272295e+00
-1.58659294e-01 1.22875476e+00 1.16943634e+00 -3.07032794e-01
1.67669177e-01 5.01896620e-01 1.37621403e-01 7.15920508e-01
3.81163098e-02 -1.43858805e-01 -3.74574363e-01 2.77393341e-01
8.03874254e-01 -3.83608155e-02 -8.70427266e-02 -1.84360653e-01
-1.10210323e+00 7.01092660e-01 8.10137987e-01 1.93263799e-01
-3.39134425e-01 3.17489386e-01 2.97191352e-01 3.11284531e-02
3.84933800e-01 1.74849227e-01 -6.97110236e-01 3.05546820e-01
-1.11880791e+00 -3.26950960e-02 2.91370898e-01 1.11769104e+00
1.14548111e+00 1.13213666e-01 -3.58847946e-01 8.32981706e-01
3.63971442e-01 4.54166859e-01 4.01976228e-01 -9.48900521e-01
2.80134052e-01 4.22955990e-01 -1.36441693e-01 -4.90142226e-01
-1.68732344e-03 -6.49507582e-01 -9.00367379e-01 1.22837260e-01
4.43689585e-01 2.27799709e-03 -1.49887693e+00 1.67114902e+00
3.11800838e-01 4.24562752e-01 4.00459617e-02 8.62800002e-01
8.00152063e-01 7.00046778e-01 2.39264384e-01 4.64138351e-02
1.54827464e+00 -1.17769659e+00 -4.58076328e-01 -3.96549106e-01
3.59800935e-01 -7.11536705e-01 1.06864822e+00 5.85169196e-01
-9.65316832e-01 -1.04204369e+00 -9.11314189e-01 -4.01046723e-02
-2.65411675e-01 4.91331607e-01 5.08576989e-01 4.69395816e-01
-9.74098384e-01 3.42491895e-01 -7.28137791e-01 4.41659689e-02
1.06336761e+00 3.67943257e-01 -1.89794630e-01 -1.14859834e-01
-9.55155432e-01 4.68614876e-01 6.24355376e-01 2.56369501e-01
-1.26081574e+00 -7.91512728e-01 -8.00025403e-01 -7.06901327e-02
6.09025240e-01 -4.41862941e-01 1.36136782e+00 -1.37021554e+00
-1.37234151e+00 1.22290039e+00 -1.33305624e-01 -7.88336754e-01
6.04969561e-01 -5.15101142e-02 -9.80721116e-02 2.21908078e-01
4.42358315e-01 1.29995918e+00 1.31241000e+00 -1.46992660e+00
-8.95950735e-01 -2.61646044e-02 2.58017574e-02 2.12459303e-02
-1.90819234e-01 -4.54731077e-01 -9.30806279e-01 -8.01860869e-01
-9.69804078e-02 -7.83939838e-01 -3.53764325e-01 1.15444668e-01
-6.62517786e-01 -3.67011160e-01 5.75539589e-01 -4.91068423e-01
7.23576069e-01 -2.34560204e+00 -3.86680327e-02 1.48000911e-01
2.96702981e-01 5.05823731e-01 -1.73892647e-01 -2.06150472e-01
6.58940896e-02 1.64163053e-01 -4.16397095e-01 -7.62750208e-01
1.57696053e-01 4.78718251e-01 -3.91461045e-01 4.09979343e-01
5.78817904e-01 1.08489025e+00 -7.23727524e-01 -7.24667966e-01
4.25502717e-01 3.23137820e-01 -5.19949377e-01 2.31821835e-01
-5.69892764e-01 3.32935780e-01 -1.89655766e-01 8.43205988e-01
8.30541134e-01 -6.05609834e-01 3.89048532e-02 -4.38262582e-01
-1.09228073e-02 2.21892577e-02 -1.20119584e+00 1.74723637e+00
-2.71527678e-01 5.10353446e-01 1.47027642e-01 -1.26974535e+00
6.81160629e-01 3.08419205e-02 3.49970281e-01 -7.58635283e-01
2.26291224e-01 4.62174490e-02 -3.40887815e-01 -2.83453196e-01
4.67871755e-01 1.44927382e-01 -1.09579824e-01 -2.87905894e-02
3.86711895e-01 5.92870116e-02 1.23388328e-01 2.18332365e-01
5.21514654e-01 1.44454017e-01 1.08542949e-01 -1.44289985e-01
3.15152764e-01 2.32977927e-01 8.00292134e-01 1.04220307e+00
-1.25279725e-01 7.82116294e-01 4.37895834e-01 -1.11656010e-01
-8.44296992e-01 -1.15417230e+00 -5.79697967e-01 1.06128061e+00
3.78315896e-01 -2.81790555e-01 -9.79644299e-01 -7.81077683e-01
1.66900158e-01 4.23199624e-01 -8.22393835e-01 -6.38105022e-03
-2.91740716e-01 -6.19955063e-01 6.30978763e-01 6.37674153e-01
7.96404541e-01 -1.12285626e+00 -3.19338143e-01 2.20255241e-01
-1.49336666e-01 -1.26964211e+00 -4.50695157e-01 5.67511976e-01
-4.21821475e-01 -8.24712157e-01 -8.12902927e-01 -1.16777039e+00
8.55624914e-01 2.47972950e-01 9.82019484e-01 6.37364909e-02
-4.63531494e-01 1.63618043e-01 -3.81160855e-01 -1.99367195e-01
-2.70090252e-01 8.01032111e-02 -1.67000279e-01 2.91025966e-01
1.35740727e-01 -1.01669848e-01 -7.05552518e-01 4.55031127e-01
-8.90227020e-01 2.75568545e-01 7.63735592e-01 1.11279392e+00
1.27096748e+00 -7.22224712e-02 4.41109121e-01 -9.86086071e-01
-2.76554704e-01 -3.67502570e-01 -5.99840403e-01 3.30841988e-01
-5.44394314e-01 -1.34567440e-01 5.27602792e-01 -7.98297465e-01
-9.15478826e-01 3.36240292e-01 -1.40002534e-01 -7.46760070e-01
-3.23610067e-01 3.41092288e-01 -3.69201452e-01 -1.42473042e-01
5.01897693e-01 3.11372548e-01 -2.87186414e-01 -4.72406030e-01
7.31793880e-01 7.59304702e-01 8.64727199e-01 -6.32766008e-01
7.18196750e-01 5.18497407e-01 -6.95810676e-01 -6.39988005e-01
-1.04956055e+00 -5.50740361e-01 -4.47205365e-01 -6.18448071e-02
1.11642230e+00 -1.31738055e+00 -6.57492578e-01 7.01399267e-01
-9.33158576e-01 -7.22995877e-01 -4.61913615e-01 1.27083123e-01
-4.60237205e-01 3.02258283e-01 -6.35218978e-01 -4.81328249e-01
-1.76702738e-01 -1.28893137e+00 1.05014801e+00 3.82394224e-01
2.84130443e-02 -6.36617362e-01 -6.01983786e-01 7.58493245e-01
1.61375657e-01 -1.84961762e-02 6.42095447e-01 -3.97489101e-01
-8.23413312e-01 1.44077376e-01 -5.69296598e-01 8.70916426e-01
1.08635277e-01 -1.33920759e-01 -1.35934067e+00 -2.47896418e-01
-2.02012047e-01 -5.77749670e-01 1.17431617e+00 6.17069840e-01
1.59496009e+00 -2.97529191e-01 -3.26723367e-01 9.06757355e-01
1.48971438e+00 -9.29453447e-02 4.94948715e-01 2.48690978e-01
1.00590253e+00 2.26116434e-01 7.22166002e-01 4.30330902e-01
4.95861351e-01 5.23521781e-01 6.27722621e-01 -4.48434114e-01
-3.45645905e-01 -2.62860596e-01 2.15737760e-01 4.44766074e-01
2.84530044e-01 -1.88047647e-01 -7.09803104e-01 6.85978055e-01
-1.87136054e+00 -6.16283119e-01 7.54278079e-02 1.85698855e+00
1.23510778e+00 4.19649482e-01 3.17120925e-02 -6.19473122e-02
7.08809435e-01 1.17483912e-02 -8.96439970e-01 6.36138692e-02
-2.44588941e-01 3.24603260e-01 9.61521566e-01 4.28645641e-01
-1.39178264e+00 1.47420621e+00 4.90485287e+00 1.36921835e+00
-1.26372385e+00 1.74896687e-01 1.05001485e+00 2.32891113e-01
8.92230961e-03 -4.58787428e-03 -1.06430078e+00 6.33515000e-01
4.50968474e-01 1.98004186e-01 2.66408682e-01 9.59651887e-01
-2.08101779e-01 -2.07347408e-01 -9.41641390e-01 1.04114306e+00
-1.33818746e-01 -1.56087208e+00 2.23167762e-01 -1.09900169e-01
8.59385967e-01 1.91723332e-01 3.94842684e-01 4.36124653e-01
1.79882035e-01 -1.01720440e+00 1.19084132e+00 2.72332400e-01
1.06460106e+00 -4.88742441e-01 6.15590394e-01 2.86900967e-01
-1.27110744e+00 1.14885323e-01 -5.33367932e-01 1.84287503e-01
-1.80685461e-01 5.86592436e-01 -9.50055242e-01 5.51671565e-01
1.02134001e+00 9.10417140e-01 -5.34018576e-01 8.99341881e-01
-4.30555224e-01 9.95944738e-01 -2.95058578e-01 3.30724925e-01
3.66795301e-01 8.78523514e-02 8.69813040e-02 1.31724310e+00
-2.77240843e-01 -1.35255605e-01 5.24351299e-01 1.05923808e+00
-3.25642377e-01 -2.13869780e-01 6.90115020e-02 1.28215402e-01
3.88855845e-01 1.35836732e+00 -9.68679428e-01 -7.19927430e-01
-2.90125966e-01 1.19753253e+00 4.02726799e-01 4.81796682e-01
-1.10426843e+00 -1.58466950e-01 7.72509813e-01 -2.26950228e-01
9.27983999e-01 1.18527986e-01 -4.30791974e-01 -1.08387113e+00
7.55118951e-02 -9.08184588e-01 3.23050231e-01 -5.34094274e-01
-1.38486874e+00 7.11141109e-01 -2.42940802e-02 -1.14285946e+00
8.00830424e-02 -7.79179096e-01 -3.13646019e-01 6.88661039e-01
-1.75609851e+00 -1.36337626e+00 -2.63720065e-01 6.39066637e-01
6.17401779e-01 1.98064387e-01 5.29202819e-01 4.17467147e-01
-7.15092063e-01 9.74176705e-01 1.88376848e-02 3.47995520e-01
7.07004488e-01 -1.45309520e+00 4.20005292e-01 7.90982068e-01
4.08496231e-01 1.98774010e-01 3.67379040e-01 -4.69526589e-01
-9.84097719e-01 -1.55951393e+00 4.37726319e-01 -2.47833923e-01
5.42974234e-01 -6.28370583e-01 -1.06299043e+00 6.72747374e-01
1.36952817e-01 3.78968447e-01 4.49252248e-01 -3.44708949e-01
-4.09377009e-01 -2.24138349e-01 -9.92484033e-01 3.67481321e-01
8.27606678e-01 -5.15910208e-01 -3.40563774e-01 4.25861329e-01
8.96884978e-01 -5.37758052e-01 -8.27859342e-01 3.41824681e-01
1.64583534e-01 -5.42077363e-01 1.09122181e+00 -2.28775963e-01
2.16270849e-01 -5.90137064e-01 -1.95677087e-01 -8.64577055e-01
-1.42647341e-01 -4.19491291e-01 1.49786219e-01 1.50263727e+00
4.30019498e-01 -4.31419045e-01 9.41071630e-01 3.87314379e-01
-1.73189431e-01 -8.56383801e-01 -1.00529182e+00 -8.22801232e-01
-9.41054896e-03 -6.89047873e-01 4.13527042e-01 8.62026334e-01
-5.76855242e-01 1.17272213e-01 -4.34169590e-01 4.25890684e-01
8.26770008e-01 1.23717517e-01 5.90240777e-01 -8.81500900e-01
-5.01172423e-01 -3.76870483e-01 -3.30033660e-01 -1.49853277e+00
2.24855930e-01 -1.06042266e+00 3.32098067e-01 -1.32635665e+00
6.65501878e-02 -7.44021237e-01 -5.12798309e-01 9.43924010e-01
-2.69850791e-01 6.19693935e-01 2.11972624e-01 2.14599386e-01
-9.26424205e-01 5.56173027e-01 1.25338876e+00 -6.12329066e-01
-1.96722578e-02 -4.75571007e-02 -6.43852472e-01 6.48520529e-01
8.57389569e-01 -6.85537338e-01 -2.77122557e-01 -5.58338046e-01
-1.31114244e-01 -2.31187537e-01 7.16463089e-01 -9.22547817e-01
1.84573010e-01 6.86215376e-03 4.89702016e-01 -6.23020232e-01
3.51037771e-01 -6.11294448e-01 -2.14684997e-02 3.10094327e-01
-3.44723016e-01 -5.35307825e-01 2.40791053e-01 5.55361629e-01
-3.22398067e-01 -1.55184165e-01 8.95728886e-01 -1.65456071e-01
-1.14925206e+00 4.55639958e-01 -2.75557220e-01 1.42463654e-01
1.00047493e+00 -3.62986118e-01 -1.61479875e-01 1.70904174e-01
-8.10399413e-01 5.12720764e-01 4.08388585e-01 2.39366397e-01
5.68544388e-01 -1.03423977e+00 -5.33402085e-01 2.19304398e-01
2.77593523e-01 5.87917089e-01 4.31550682e-01 6.49611890e-01
-1.67712003e-01 -2.56374804e-03 -1.01696052e-01 -1.06463420e+00
-1.11504352e+00 5.01762033e-01 3.24293345e-01 -7.04048350e-02
-6.25570357e-01 1.30067813e+00 5.04711211e-01 -1.98659644e-01
5.08485794e-01 -7.22225964e-01 -1.46339843e-02 9.67141241e-02
3.95886272e-01 -1.07334174e-01 4.05387133e-02 -5.56405783e-01
-3.96986872e-01 5.85628390e-01 -4.19425577e-01 1.38736263e-01
1.05450225e+00 -1.78071648e-01 6.03067828e-03 3.13335508e-01
1.17400420e+00 -5.23754120e-01 -1.81716061e+00 -6.09609902e-01
-1.37532458e-01 -2.36506194e-01 1.08499981e-01 -7.94720411e-01
-1.32959223e+00 7.46747196e-01 6.66966438e-01 -5.33719808e-02
1.08338547e+00 3.81973118e-01 6.75701082e-01 2.85492718e-01
3.34379315e-01 -1.06826460e+00 7.03793541e-02 2.81296045e-01
5.57566226e-01 -1.54067242e+00 -3.40434939e-01 -5.67922235e-01
-8.51009786e-01 6.96950614e-01 8.47301900e-01 -2.34319314e-01
5.73832631e-01 1.45058110e-01 3.95075172e-01 9.58195999e-02
-5.56479573e-01 -4.88852620e-01 3.57937425e-01 5.60574472e-01
-4.34509143e-02 3.26785475e-01 2.62825400e-01 8.50472689e-01
-8.37108791e-02 -1.35897323e-01 1.81043550e-01 5.95224142e-01
-2.68727094e-01 -9.17309403e-01 -2.81031013e-01 3.55424762e-01
-3.80219042e-01 -2.58298129e-01 -4.24066819e-02 6.63931668e-01
6.39932990e-01 8.29580665e-01 9.03728679e-02 -4.41874415e-01
7.03688860e-02 -2.47570232e-01 3.15982997e-01 -7.00484931e-01
-6.49415255e-01 2.22104937e-01 -3.05214614e-01 -4.71249282e-01
-4.61062849e-01 -4.65450674e-01 -1.44847178e+00 1.80423617e-01
-3.84103298e-01 1.29108518e-01 4.61687118e-01 9.03692722e-01
2.74816364e-01 6.72724247e-01 5.28270423e-01 -9.32904720e-01
-4.61614162e-01 -7.41839647e-01 -4.96887207e-01 3.61757159e-01
4.70525861e-01 -5.53752005e-01 -2.12706506e-01 4.51925933e-01] | [9.59519100189209, 0.7802178263664246] |
ef4471ce-88ce-4b3d-994d-6b41267a0501 | active-object-localization-with-deep | 1511.06015 | null | http://arxiv.org/abs/1511.06015v1 | http://arxiv.org/pdf/1511.06015v1.pdf | Active Object Localization with Deep Reinforcement Learning | We present an active detection model for localizing objects in scenes. The
model is class-specific and allows an agent to focus attention on candidate
regions for identifying the correct location of a target object. This agent
learns to deform a bounding box using simple transformation actions, with the
goal of determining the most specific location of target objects following
top-down reasoning. The proposed localization agent is trained using deep
reinforcement learning, and evaluated on the Pascal VOC 2007 dataset. We show
that agents guided by the proposed model are able to localize a single instance
of an object after analyzing only between 11 and 25 regions in an image, and
obtain the best detection results among systems that do not use object
proposals for object localization. | ['Svetlana Lazebnik', 'Juan C. Caicedo'] | 2015-11-18 | active-object-localization-with-deep-1 | http://openaccess.thecvf.com/content_iccv_2015/html/Caicedo_Active_Object_Localization_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Caicedo_Active_Object_Localization_ICCV_2015_paper.pdf | iccv-2015-12 | ['active-object-localization'] | ['computer-vision'] | [ 9.15378053e-03 4.66301769e-01 -9.53623578e-02 -2.96123087e-01
-8.42497468e-01 -7.36261308e-01 6.65059686e-01 5.93927383e-01
-1.08073151e+00 3.68179888e-01 -2.21008554e-01 2.80437589e-01
9.41471830e-02 -8.85741770e-01 -1.18082976e+00 -7.88147748e-01
-3.31526488e-01 7.25347161e-01 1.10414398e+00 2.27464810e-01
2.57430285e-01 1.02146530e+00 -1.20551860e+00 3.37900877e-01
4.22963262e-01 7.84676015e-01 7.76069343e-01 7.51746178e-01
9.39189494e-02 1.15226328e+00 -9.47205901e-01 5.02789244e-02
4.14937794e-01 -7.13154906e-03 -7.79180050e-01 2.31751233e-01
8.01073551e-01 -5.94559491e-01 -9.63098481e-02 1.06046677e+00
1.49381598e-02 2.11304188e-01 5.77863097e-01 -1.21493697e+00
-4.66095716e-01 5.77615917e-01 -4.74636495e-01 5.27467847e-01
3.05016816e-01 3.98560762e-01 1.03005922e+00 -9.06592786e-01
7.19571412e-01 1.17650807e+00 1.42099053e-01 6.15234077e-01
-1.24587154e+00 -2.70614296e-01 6.35860384e-01 4.43364948e-01
-1.31718016e+00 -3.09324026e-01 4.61054951e-01 -5.14423728e-01
1.19383299e+00 -1.05958087e-02 6.18667424e-01 8.10066819e-01
2.75329798e-01 9.73536313e-01 7.93663144e-01 -5.01595616e-01
7.30286181e-01 1.47453099e-01 1.62545562e-01 9.69278455e-01
3.46812069e-01 2.93178767e-01 -3.56967539e-01 -2.14602336e-01
6.16186738e-01 2.54765302e-02 1.86074540e-01 -1.03172970e+00
-1.22746134e+00 7.83399642e-01 1.04617214e+00 1.11891717e-01
-6.74149930e-01 4.29050356e-01 1.18812911e-01 -3.83821219e-01
1.94430500e-01 5.50068080e-01 -1.71295092e-01 5.43371081e-01
-6.58431530e-01 3.35168809e-01 3.49970013e-01 8.20404291e-01
9.17644560e-01 -2.31189787e-01 -3.30813318e-01 2.75232881e-01
2.34797627e-01 5.06069005e-01 -3.02040242e-02 -1.04340529e+00
2.31009260e-01 8.43488216e-01 4.53843504e-01 -7.33007610e-01
-4.47190017e-01 -3.43875229e-01 4.01401520e-02 9.64577913e-01
3.17323327e-01 -1.10091612e-01 -9.54141080e-01 1.54674327e+00
5.57005346e-01 -1.93938762e-02 1.92645833e-01 8.89016032e-01
6.36297524e-01 6.09625459e-01 5.10883629e-01 3.46729346e-02
1.10632670e+00 -1.22110283e+00 -1.69231787e-01 -6.83854222e-01
4.63273704e-01 -2.31882438e-01 5.14524519e-01 1.09365001e-01
-1.15267134e+00 -8.11068892e-01 -8.74912620e-01 3.26423571e-02
-5.54687619e-01 6.29601419e-01 5.49566984e-01 1.28799364e-01
-1.07855165e+00 1.10193171e-01 -1.03997099e+00 -5.30400991e-01
6.69024169e-01 3.74823034e-01 -5.05982220e-01 1.07509091e-01
-5.29827535e-01 1.12314427e+00 9.30896938e-01 6.03931732e-02
-1.81127667e+00 -4.82054472e-01 -8.75654519e-01 1.80648282e-01
6.48935556e-01 -3.79708469e-01 1.06332052e+00 -1.20324135e+00
-1.07830429e+00 8.13893080e-01 -1.01798676e-01 -1.05152893e+00
4.39663082e-01 -4.69566345e-01 4.97642206e-03 5.22660196e-01
2.96750456e-01 1.17207932e+00 9.17662621e-01 -1.42708576e+00
-1.30031359e+00 -4.09570128e-01 6.27939999e-01 2.82869697e-01
1.09368913e-01 2.18138963e-01 -4.23302174e-01 -6.12051114e-02
-1.05503881e-02 -8.98539066e-01 -3.78899097e-01 2.67604351e-01
-2.91221768e-01 -5.09905756e-01 1.28497696e+00 -3.28112602e-01
3.11354458e-01 -2.04660368e+00 -1.35849072e-02 -7.07346946e-02
7.54065886e-02 2.48609975e-01 -3.80012095e-01 2.85373870e-02
1.38534784e-01 -2.97159106e-01 -7.90427849e-02 -3.58478606e-01
-2.70231336e-01 6.33421913e-02 -6.32440984e-01 6.40035689e-01
4.47300196e-01 9.07775462e-01 -9.99309361e-01 -4.12461251e-01
4.72988814e-01 2.16777593e-01 -4.05537456e-01 4.93971318e-01
-5.62264323e-01 2.58663923e-01 -3.90034527e-01 4.66120869e-01
5.69991291e-01 -3.08925267e-02 -3.69404368e-02 -9.87885706e-03
-2.34099030e-01 6.89218044e-02 -1.09386492e+00 1.39634264e+00
-2.19850942e-01 6.27734900e-01 3.41769271e-02 -9.42191958e-01
8.77493024e-01 -1.18573070e-01 3.41802746e-01 -5.72230160e-01
-2.23194599e-01 -1.27636284e-01 1.49179727e-01 -4.17742103e-01
3.40477973e-01 6.76240385e-01 1.79531462e-02 3.36946666e-01
3.38646591e-01 -3.86814773e-02 4.79764432e-01 3.73840749e-01
1.34558558e+00 2.79702246e-01 3.64817768e-01 -5.85229509e-02
5.30559123e-01 2.96704262e-01 3.67370933e-01 1.24240541e+00
-4.45701718e-01 1.94609165e-01 3.05792898e-01 -5.55353701e-01
-7.59447098e-01 -1.30646002e+00 2.52845913e-01 1.46042752e+00
4.89904702e-01 -9.51090455e-03 -7.67374277e-01 -1.15826857e+00
-1.30486116e-01 8.29421103e-01 -9.81912613e-01 -9.59797576e-02
-7.16538310e-01 -2.63648629e-01 1.58185482e-01 7.06726432e-01
6.16889119e-01 -1.46129942e+00 -1.52569938e+00 1.26743186e-02
1.98733896e-01 -1.16876090e+00 -5.36234640e-02 4.42186028e-01
-5.68703890e-01 -1.28814125e+00 -5.13087928e-01 -8.34602475e-01
1.29157782e+00 2.67581910e-01 9.51506078e-01 -2.62306761e-02
-5.32736361e-01 7.37890542e-01 -3.31559986e-01 -5.82813621e-01
-4.18334574e-01 -5.72864786e-02 -2.02759102e-01 -1.66826323e-02
2.79998064e-01 3.21794093e-01 -5.62675416e-01 2.30316430e-01
-5.49519300e-01 -1.40668675e-01 7.34133780e-01 4.15591925e-01
6.48183048e-01 -1.24049157e-01 1.42158449e-01 -6.71947420e-01
1.03543915e-01 -3.85580398e-02 -1.18238735e+00 5.09140313e-01
1.72661021e-01 -9.60250944e-02 4.39689547e-01 -3.32640886e-01
-1.01095796e+00 8.85971487e-01 3.96428168e-01 -1.86279804e-01
-6.03242636e-01 -2.64605194e-01 3.73299830e-02 -2.18289912e-01
9.51857328e-01 3.49331617e-01 -5.85341573e-01 1.00585306e-02
3.69164467e-01 -7.53140897e-02 6.01024032e-01 -3.26796949e-01
7.35455573e-01 8.28485429e-01 -1.69543520e-01 -4.85141754e-01
-9.32907939e-01 -5.70770025e-01 -1.06303632e+00 -4.34609205e-01
1.06082869e+00 -8.76720130e-01 -6.90950096e-01 1.78635754e-02
-1.37412298e+00 -5.87971807e-01 -4.57330406e-01 4.92222518e-01
-7.37322330e-01 -1.72618881e-01 -1.26820683e-01 -7.39174366e-01
-1.54808074e-01 -1.17379284e+00 1.23797071e+00 2.81288087e-01
6.00806577e-03 -6.97027862e-01 9.91155431e-02 1.13839668e-03
2.53113180e-01 8.32194909e-02 6.20581388e-01 -7.92882264e-01
-1.10240126e+00 -2.76656300e-01 -1.38538569e-01 1.16280705e-01
-9.71907284e-03 2.34014988e-01 -9.13605392e-01 -4.41567332e-01
-3.35611522e-01 -3.04848939e-01 1.07475305e+00 2.69003689e-01
1.11473370e+00 -2.21601292e-01 -7.09185421e-01 1.01141654e-01
1.42584991e+00 6.02689147e-01 3.49780023e-01 2.85401493e-01
2.63253927e-01 4.31340426e-01 9.12443459e-01 1.79498121e-01
-2.10112371e-02 7.13338733e-01 1.03063285e+00 -3.43845308e-01
-1.21232249e-01 7.27670044e-02 3.33704561e-01 -7.16432750e-01
-7.10790828e-02 -1.36879772e-01 -1.01061416e+00 7.31110930e-01
-1.85371554e+00 -1.04581499e+00 4.61837769e-01 2.10027504e+00
2.38101751e-01 4.75522548e-01 2.05729499e-01 -3.85046601e-01
8.46357644e-01 4.77289036e-02 -6.84097826e-01 -1.22974716e-01
4.06622216e-02 -1.79179851e-02 6.09291792e-01 7.03565657e-01
-1.41358066e+00 1.37123907e+00 6.76868725e+00 2.52596796e-01
-1.11597991e+00 7.81246498e-02 3.50737959e-01 -8.26675817e-02
5.40303111e-01 1.35373607e-01 -1.06626022e+00 5.63778169e-03
2.43921041e-01 -3.91680095e-03 9.79095548e-02 1.30663872e+00
1.73124492e-01 -6.43180728e-01 -1.38988924e+00 3.01826537e-01
2.15321243e-01 -1.27159286e+00 1.24825677e-02 -3.61666232e-01
6.06649816e-01 1.39533743e-01 2.90392041e-02 3.36935282e-01
5.55824280e-01 -7.11779535e-01 1.07920885e+00 6.21209979e-01
-1.40839726e-01 -6.69354737e-01 3.82053345e-01 5.43992579e-01
-9.55965936e-01 -4.70284253e-01 -6.04909539e-01 1.05862707e-01
-1.47029296e-01 -1.03411980e-01 -1.58259189e+00 -2.10993305e-01
7.26256490e-01 3.39783877e-01 -8.86220276e-01 1.44074798e+00
-5.85932136e-01 3.94157052e-01 -1.80546463e-01 -2.32466772e-01
3.80613655e-01 2.63745278e-01 6.25778675e-01 1.24814749e+00
-8.29682872e-02 4.18297648e-02 5.28055847e-01 1.19730389e+00
1.68538988e-01 -2.14695379e-01 -3.77440155e-01 4.02616471e-01
3.93090963e-01 1.28496957e+00 -1.15923917e+00 -5.06319940e-01
-4.44399342e-02 1.12051368e+00 8.51753354e-01 4.68319088e-01
-1.06099105e+00 -2.32639074e-01 7.46751204e-02 8.05299655e-02
7.66200721e-01 -4.62130487e-01 2.12535679e-01 -5.76829433e-01
-2.20515504e-01 -3.91282856e-01 3.97699237e-01 -1.13049626e+00
-6.35667980e-01 5.27629018e-01 7.66031668e-02 -7.66848147e-01
-1.79689258e-01 -8.66679251e-01 -6.66074932e-01 5.56771934e-01
-1.19840527e+00 -1.14876246e+00 -4.90690529e-01 5.12933135e-01
5.90518475e-01 -2.71961600e-01 8.68414521e-01 -3.17607343e-01
-2.56402284e-01 1.10454522e-01 -2.36620456e-01 5.44737935e-01
3.41455609e-01 -1.26833165e+00 2.85094619e-01 1.05873919e+00
6.04144156e-01 6.66320801e-01 5.95956564e-01 -4.91842985e-01
-1.03987527e+00 -1.29830241e+00 3.07852507e-01 -6.26734078e-01
3.05017173e-01 -6.82040453e-01 -6.00966156e-01 9.54913735e-01
2.42571577e-01 5.70968866e-01 -4.32557911e-02 -2.65810937e-01
-3.45003694e-01 -2.43827581e-01 -1.31844079e+00 5.02828121e-01
7.19141185e-01 -3.37500572e-01 -7.11138844e-01 6.61695004e-01
4.73461896e-01 -5.62540412e-01 -7.99332187e-02 4.65513915e-01
3.31915595e-04 -9.84231234e-01 1.02447379e+00 -6.24488950e-01
-1.06914146e-02 -6.53895855e-01 1.35139897e-01 -1.04663908e+00
-6.24115944e-01 6.14093877e-02 -6.18057512e-02 5.68881691e-01
5.56208014e-01 -2.60800511e-01 9.03403282e-01 2.93352962e-01
1.83165018e-02 -4.28324014e-01 -9.00243640e-01 -5.17791867e-01
-4.26249236e-01 -1.27774447e-01 9.30282921e-02 3.92655969e-01
-3.74053955e-01 7.82428458e-02 2.43094787e-01 7.55143285e-01
6.24714792e-01 3.22310388e-01 9.66182232e-01 -8.02964985e-01
-3.94207120e-01 -2.37889916e-01 -7.08431304e-01 -8.25121701e-01
4.08795059e-01 -4.73814785e-01 3.94741923e-01 -1.81083786e+00
4.76079583e-01 -3.22976023e-01 -3.83599818e-01 9.77535188e-01
5.62016740e-02 4.22417611e-01 4.12523568e-01 -6.40863925e-02
-1.21621370e+00 1.30515382e-01 8.06977808e-01 -5.35562515e-01
-9.44304690e-02 1.70490831e-01 -3.39219198e-02 9.03936684e-01
5.77190518e-01 -5.48787177e-01 -1.70228794e-01 -3.84529978e-01
-6.92927465e-02 -1.61553636e-01 9.58488762e-01 -1.32498229e+00
4.58193183e-01 -3.94152462e-01 8.59634757e-01 -7.49351740e-01
5.31917095e-01 -1.17703354e+00 -3.70150864e-01 8.37096095e-01
-8.32021892e-01 -1.91672951e-01 4.58029211e-01 4.79702801e-01
1.84953474e-02 -4.53836292e-01 1.06292570e+00 -3.05179745e-01
-1.46618998e+00 1.29436553e-01 -4.42929596e-01 -4.75169450e-01
1.72444105e+00 7.15778545e-02 -4.30596471e-01 -2.53174221e-03
-1.12721765e+00 -4.99790162e-03 2.37365454e-01 4.99365687e-01
7.59110451e-01 -9.02857721e-01 -5.47123432e-01 1.02278978e-01
3.12217534e-01 4.97241393e-02 1.12331063e-02 2.51638710e-01
-7.72161722e-01 4.61538941e-01 -2.72959858e-01 -7.99567103e-01
-1.52453852e+00 9.38521564e-01 8.06444466e-01 3.45819034e-02
-4.87535298e-01 1.03680313e+00 6.64750218e-01 -7.17679113e-02
4.09839213e-01 -4.84613448e-01 -2.78850377e-01 -2.77289063e-01
6.41571760e-01 1.84211805e-01 4.10726927e-02 -7.08209157e-01
-6.05679095e-01 3.44597816e-01 -2.98353046e-01 -2.23343000e-02
1.05123878e+00 2.44949982e-01 -4.10334915e-02 1.70585424e-01
7.77851045e-01 7.59459063e-02 -1.61385715e+00 -2.55197078e-01
-6.32067397e-02 -3.84868592e-01 -2.70360615e-02 -1.05780005e+00
-8.20379555e-01 6.18283987e-01 8.92878771e-01 6.85885027e-02
8.68049145e-01 3.24684471e-01 -2.38092497e-01 7.73344636e-01
6.56592906e-01 -9.59643960e-01 6.18720591e-01 5.18164814e-01
1.02133739e+00 -1.19452143e+00 1.96287140e-01 -2.92801857e-01
-5.86560547e-01 1.08661151e+00 1.11455917e+00 -6.04248881e-01
1.67560503e-01 3.92541051e-01 -6.08223714e-02 -6.47270754e-02
-6.04873598e-01 -4.31394428e-01 2.79161304e-01 6.36795342e-01
-1.52940273e-01 -2.46083774e-02 1.81035981e-01 -2.01934040e-01
3.66646141e-01 -5.12745738e-01 4.11722720e-01 9.63628769e-01
-1.04026341e+00 -4.64947432e-01 -5.92519343e-01 -1.71568155e-01
-1.58546343e-01 3.22803944e-01 -7.13466704e-01 8.81253183e-01
4.22635734e-01 7.12640584e-01 5.66158116e-01 2.52739012e-01
1.91353381e-01 -2.81979620e-01 6.66419566e-01 -9.99987423e-01
-3.48468244e-01 -1.45462751e-01 -4.58438665e-01 -6.84657454e-01
-4.91522819e-01 -6.02128386e-01 -1.61693764e+00 4.89863008e-01
-2.61396736e-01 -5.18080741e-02 6.67702317e-01 8.53397548e-01
2.24064112e-01 4.85073954e-01 5.00308514e-01 -8.43000412e-01
-1.46570370e-01 -7.53996491e-01 -1.94879532e-01 2.79441774e-01
3.84170502e-01 -5.19952357e-01 -1.26985684e-01 5.97028583e-02] | [9.325366020202637, 0.570793092250824] |
f36dc77c-98cb-4ee3-90f4-34f6a4a52806 | are-neural-open-domain-dialog-systems-robust | 2008.07683 | null | https://arxiv.org/abs/2008.07683v1 | https://arxiv.org/pdf/2008.07683v1.pdf | Are Neural Open-Domain Dialog Systems Robust to Speech Recognition Errors in the Dialog History? An Empirical Study | Large end-to-end neural open-domain chatbots are becoming increasingly popular. However, research on building such chatbots has typically assumed that the user input is written in nature and it is not clear whether these chatbots would seamlessly integrate with automatic speech recognition (ASR) models to serve the speech modality. We aim to bring attention to this important question by empirically studying the effects of various types of synthetic and actual ASR hypotheses in the dialog history on TransferTransfo, a state-of-the-art Generative Pre-trained Transformer (GPT) based neural open-domain dialog system from the NeurIPS ConvAI2 challenge. We observe that TransferTransfo trained on written data is very sensitive to such hypotheses introduced to the dialog history during inference time. As a baseline mitigation strategy, we introduce synthetic ASR hypotheses to the dialog history during training and observe marginal improvements, demonstrating the need for further research into techniques to make end-to-end open-domain chatbots fully speech-robust. To the best of our knowledge, this is the first study to evaluate the effects of synthetic and actual ASR hypotheses on a state-of-the-art neural open-domain dialog system and we hope it promotes speech-robustness as an evaluation criterion in open-domain dialog. | ['Dilek Hakkani-Tur', 'Longshaokan Wang', 'Yang Liu', 'Karthik Gopalakrishnan', 'Behnam Hedayatnia'] | 2020-08-18 | null | null | null | null | ['open-domain-dialog'] | ['natural-language-processing'] | [-3.8992271e-02 6.8154103e-01 3.0777106e-01 -5.9708714e-01
-1.0613191e+00 -8.3146399e-01 8.0621856e-01 -5.4994577e-01
-3.9728507e-01 5.4375482e-01 4.4372007e-01 -6.1578882e-01
4.2326131e-01 -4.6267071e-01 -5.5946916e-01 -2.3358256e-01
2.9255483e-01 1.2485821e+00 3.8229334e-01 -7.8351188e-01
-3.6219403e-01 -2.8455187e-02 -8.6135578e-01 5.4235828e-01
7.4676955e-01 4.6723095e-01 5.0902581e-01 9.0044391e-01
-6.1742134e-02 1.1488175e+00 -1.1774700e+00 -6.0516900e-01
1.3904569e-01 -7.5561392e-01 -1.3273894e+00 -5.9135590e-02
2.0446846e-01 -6.8954158e-01 -5.1637012e-01 7.3220950e-01
6.9550943e-01 4.1159225e-01 5.2815700e-01 -1.0014186e+00
-1.0604224e+00 1.2567590e+00 2.7995905e-01 1.7179592e-01
2.7079561e-01 8.3491141e-01 1.1155787e+00 -6.9898564e-01
8.0723822e-01 1.5416096e+00 5.0941092e-01 1.0154803e+00
-1.3557304e+00 -5.7324862e-01 -4.3354239e-02 -2.6587903e-02
-9.1743189e-01 -9.7281772e-01 5.3489023e-01 -2.5451154e-01
1.4995238e+00 9.4496973e-02 8.3790077e-03 1.7992263e+00
-1.4656931e-01 7.7853841e-01 9.3301135e-01 -3.4488410e-01
1.8163295e-01 3.6750829e-01 1.7952460e-01 4.9142456e-01
-7.5139481e-01 3.3956090e-01 -5.0299513e-01 -4.5109909e-02
4.9649146e-01 -5.8144724e-01 -1.7149618e-01 3.3561090e-01
-7.5386810e-01 1.2110072e+00 4.0227476e-01 4.4473085e-01
-5.1849656e-02 -5.9358642e-02 5.6111372e-01 8.3783221e-01
5.8024186e-01 7.5626302e-01 -6.6022009e-01 -7.5139391e-01
-6.3419539e-01 7.4346349e-02 1.3697250e+00 1.0567321e+00
3.8175562e-01 2.0250097e-01 -3.9159507e-01 1.5488272e+00
1.8801534e-01 8.7123916e-02 6.0810280e-01 -1.1402014e+00
6.2851882e-01 2.8474814e-01 -2.5461346e-01 6.6108957e-02
-4.1304361e-02 1.6552510e-02 -1.3724872e-01 1.3001274e-01
8.1030673e-01 -6.3876873e-01 -7.8854561e-01 1.9387963e+00
-1.6388364e-01 -3.2616806e-01 4.8613539e-01 7.5097221e-01
1.1488631e+00 8.1620693e-01 -5.3497776e-02 1.3386934e-01
1.3900161e+00 -1.2233481e+00 -7.9650235e-01 -4.6209195e-01
6.9014245e-01 -8.1433392e-01 1.4579743e+00 9.2734687e-02
-1.1551201e+00 -3.4392670e-01 -6.5937984e-01 -3.4902906e-01
-1.6742407e-01 -1.6224043e-02 2.4577820e-01 7.6816922e-01
-1.1674906e+00 4.2889139e-01 -8.7789679e-01 -6.2528515e-01
1.0310836e-01 1.2505902e-01 -1.9921674e-01 1.6344136e-01
-1.4773045e+00 1.4467680e+00 6.5345027e-02 -3.5767743e-01
-1.1490922e+00 -6.6720015e-01 -6.9690841e-01 3.0673271e-01
4.6124420e-01 -4.2432868e-01 2.2725899e+00 -7.9484588e-01
-2.0112896e+00 7.9812974e-01 -5.8739889e-02 -7.0416570e-01
4.2615315e-01 -8.8443533e-02 -1.1158280e-01 -1.3267474e-02
-1.7104445e-01 8.7814415e-01 4.4475976e-01 -1.1794376e+00
-1.4651904e-01 -2.5639257e-01 2.5738817e-01 2.1318263e-01
-1.9996393e-01 3.0108950e-01 1.4353053e-01 -4.5260811e-01
-4.3979946e-01 -1.0659081e+00 6.9127930e-03 -2.6000959e-01
-2.9091239e-01 -5.8015186e-01 8.3849764e-01 -8.7639749e-01
7.7301675e-01 -2.0039320e+00 -2.2229500e-02 -5.0179237e-01
-7.5746581e-02 4.9261370e-01 -2.6050887e-01 8.1401414e-01
2.4326609e-01 6.7567460e-02 -1.3567263e-01 -8.1689590e-01
2.2068879e-01 5.4442990e-01 -5.9248471e-01 -2.3121335e-02
4.3058985e-01 1.0285593e+00 -6.8516618e-01 -2.1608342e-01
1.6175428e-01 3.9792454e-01 -5.4901755e-01 7.2573751e-01
-7.6872987e-01 5.8637983e-01 -6.8521425e-02 1.6959317e-01
3.9157923e-02 1.2891051e-01 2.0532755e-02 3.2991579e-01
-1.1061819e-02 1.2087041e+00 -1.5843523e-01 1.7701263e+00
-7.7845675e-01 8.0580831e-01 3.9342922e-01 -6.7926604e-01
1.0308930e+00 8.5897493e-01 -5.4809190e-02 -5.7202458e-01
5.8960789e-01 1.6464716e-01 5.3593224e-01 -2.5198951e-01
7.5292569e-01 -5.6423706e-01 -1.7202941e-01 5.3723961e-01
8.0888718e-01 -4.0341228e-01 -1.6113983e-01 3.9312449e-01
1.3474903e+00 -1.2767096e-01 -1.2627879e-01 -2.8811682e-02
-3.4751792e-02 1.9435735e-01 -2.4581354e-02 9.4058794e-01
-3.0474770e-01 7.7310139e-01 4.3168503e-01 1.7461771e-01
-9.7341794e-01 -1.0350006e+00 -4.5959265e-03 1.4908741e+00
-3.9501405e-01 -1.9657229e-01 -1.1930588e+00 -8.0609637e-01
-4.0619195e-01 1.2471551e+00 -2.4369414e-01 -2.4163358e-01
-4.2119038e-01 -1.6660424e-01 1.3975763e+00 4.7904566e-01
2.9142427e-01 -1.2762276e+00 -1.7266506e-01 4.7782752e-01
-5.3484809e-01 -1.2805731e+00 -6.8784672e-01 5.3390747e-01
-7.8328061e-01 -5.4685855e-01 -8.3407241e-01 -8.4719330e-01
1.9611295e-02 5.4201070e-02 1.2340252e+00 -2.1017157e-01
3.1531212e-01 2.8359482e-01 -6.7486674e-01 -2.0938459e-01
-1.5855395e+00 4.0186206e-01 -2.7869287e-01 -4.6293339e-01
3.8954988e-01 -6.6002965e-01 -8.7217437e-03 6.7481226e-01
-5.1327324e-01 -1.0915081e-01 3.2727504e-01 1.1779230e+00
-3.0492136e-01 -4.9152347e-01 6.8138415e-01 -8.4937805e-01
1.1090499e+00 -5.2461255e-01 -3.3502966e-01 1.5462600e-01
-7.1548820e-02 3.3723533e-01 7.3206383e-01 -7.6514971e-01
-1.5037845e+00 -5.9528958e-02 -7.7532452e-01 -4.6451858e-01
-3.4986761e-01 3.1669393e-01 -2.2144751e-01 3.0446932e-01
1.0726180e+00 1.6735747e-01 2.3118730e-01 -5.4484749e-01
8.2487494e-01 1.1633928e+00 7.2748286e-01 -6.3377845e-01
4.1250893e-01 -2.5599429e-02 -9.3854564e-01 -1.2743819e+00
-5.2203995e-01 -4.1841635e-01 -2.0076443e-01 -1.7996240e-02
9.9015790e-01 -9.1428500e-01 -5.9942812e-01 5.0081271e-01
-1.6552287e+00 -1.1151004e+00 -2.4145490e-01 3.2736611e-02
-7.8049046e-01 2.1415354e-01 -1.0641143e+00 -9.0218246e-01
-3.4986144e-01 -1.3890781e+00 8.3492982e-01 -3.5169326e-02
-7.7817589e-01 -1.0022957e+00 2.0247160e-01 8.3822036e-01
5.7453132e-01 -6.6601682e-01 6.1566532e-01 -1.4380859e+00
-4.6965292e-01 1.2254938e-01 1.2201397e-01 6.0074538e-01
-4.8734620e-02 -3.3532888e-01 -1.5281445e+00 -1.0082341e-01
3.5738865e-01 -9.1291034e-01 6.4820248e-01 4.8427042e-02
3.1083542e-01 -6.9308352e-01 1.3199372e-02 1.1083980e-01
6.1980486e-01 3.6398315e-01 7.0823652e-01 -2.4159886e-01
3.8919118e-01 7.4192053e-01 3.3802637e-01 6.8488091e-02
3.3282116e-01 8.1182998e-01 1.8743229e-01 3.6996145e-02
-4.3722114e-01 -4.0748280e-01 7.3890322e-01 8.0190843e-01
4.2570058e-01 -6.5180594e-01 -7.7647859e-01 6.2230480e-01
-1.5596005e+00 -1.0277749e+00 -2.8375823e-02 1.7371939e+00
1.2971601e+00 3.3296001e-01 2.3611295e-01 -3.4046572e-01
6.3753295e-01 1.8239446e-01 -3.2662600e-01 -7.5010282e-01
7.2618417e-02 2.7822751e-01 2.3920085e-01 8.8504815e-01
-6.6501689e-01 1.3020022e+00 5.5824265e+00 8.8782698e-01
-9.6541059e-01 7.0250738e-01 5.3127968e-01 4.0370218e-02
-1.1577030e-01 2.5445500e-01 -8.4726626e-01 4.0820405e-01
1.4409790e+00 2.2514197e-01 7.5678259e-01 9.8681045e-01
1.1009797e-01 1.9257151e-02 -1.3982016e+00 3.6948663e-01
-1.4428651e-01 -1.0031943e+00 -4.3174946e-01 5.7336327e-02
3.2191101e-01 5.4658401e-01 -1.2057176e-01 9.0400094e-01
1.2008548e+00 -1.0583680e+00 7.4894863e-01 -1.9362101e-01
5.6433338e-01 -2.2192451e-01 4.6350056e-01 6.3387018e-01
-6.0274369e-01 1.0899458e-01 -1.2160964e-01 -2.3692349e-01
6.0768509e-01 -7.0835792e-02 -1.7663107e+00 4.7813047e-02
2.5560459e-01 9.3088172e-02 -2.1868125e-01 3.7513167e-01
-2.8945589e-01 1.1374111e+00 -5.5538261e-01 -3.7741262e-01
4.0308177e-01 1.9039616e-01 7.6864058e-01 1.1203104e+00
-6.9347873e-02 1.3008489e-01 3.6476180e-02 1.3121103e+00
-1.9486913e-01 -4.1157898e-01 -7.7013457e-01 -4.1293740e-01
6.5418154e-01 6.8970752e-01 -4.6108934e-01 -1.6630010e-01
-4.5813161e-01 1.1676379e+00 5.6738305e-01 1.3568537e-01
-6.8841588e-01 -1.2301866e-01 8.1551987e-01 1.8854875e-02
2.7439725e-01 -2.6457551e-01 -3.1672612e-01 -9.8258173e-01
-1.9945060e-01 -1.1455761e+00 3.0626521e-01 -7.9289192e-01
-1.6369866e+00 9.8763448e-01 -2.0229982e-01 -5.6103909e-01
-6.9227618e-01 -4.1984800e-01 -8.7921333e-01 1.0282756e+00
-8.9679438e-01 -1.0464376e+00 1.2719236e-01 4.3462977e-01
1.2476192e+00 -2.6200005e-01 1.0154332e+00 -9.2751263e-03
-1.8554929e-01 8.5351878e-01 1.0173926e-01 5.6929874e-01
8.3166778e-01 -1.1792017e+00 1.0577807e+00 6.2539601e-01
2.3794399e-01 6.1618370e-01 9.5944619e-01 -6.1009717e-01
-1.0192553e+00 -7.3826510e-01 7.6047975e-01 -9.6218902e-01
7.8996086e-01 -6.9638103e-01 -1.1102201e+00 9.4062525e-01
8.1931490e-01 -5.4322910e-01 4.0075752e-01 3.8196969e-01
-2.7440888e-01 3.8685313e-01 -1.0435528e+00 7.5228280e-01
9.5248294e-01 -8.2184458e-01 -1.1981634e+00 1.9867308e-01
1.2245278e+00 -5.9490633e-01 -8.0294728e-01 3.6196426e-02
2.5243554e-01 -7.9276317e-01 6.9341743e-01 -6.0475802e-01
4.7031647e-01 4.7745681e-01 -8.9283757e-02 -1.7461325e+00
1.9210246e-01 -1.0664309e+00 2.2819158e-01 1.6646050e+00
7.7769482e-01 -5.5644786e-01 6.5789115e-01 7.1987259e-01
-7.2151089e-01 -3.3544451e-01 -1.2578703e+00 -9.6187395e-01
5.2248251e-01 -2.3191854e-01 5.2589491e-02 8.6910331e-01
5.1538199e-01 1.1860752e+00 -2.2559795e-01 -2.1592061e-01
-4.8540872e-02 -5.1386768e-01 9.5844066e-01 -9.3378288e-01
-6.7174190e-01 -4.5960855e-01 6.4213350e-02 -1.4202673e+00
3.5865852e-01 -7.9922819e-01 6.2088072e-01 -1.4298019e+00
-3.3127642e-01 -5.5823630e-01 5.3975499e-01 5.5724162e-01
5.7971783e-02 -2.4971330e-01 3.9560667e-01 2.8910690e-03
-2.9822972e-01 9.2778814e-01 1.0354471e+00 -4.3739032e-02
-4.2106918e-01 2.1129780e-01 -7.1192145e-01 3.8201204e-01
7.1331489e-01 -6.1021280e-01 -5.8369994e-01 -5.8480692e-01
-5.5620646e-01 5.6278741e-01 2.0361154e-01 -7.4488539e-01
2.7626184e-01 8.9132763e-02 -5.8338225e-01 -2.0969486e-01
8.9198941e-01 -4.4814771e-01 -1.9930963e-01 2.1645983e-01
-7.1396697e-01 -2.9206231e-01 2.7256298e-01 1.3649128e-01
-9.3359649e-02 -4.4650361e-01 9.5421535e-01 -4.8424664e-01
-3.3923602e-01 -3.2333520e-01 -9.6492112e-01 6.7412442e-01
5.4207152e-01 -1.8637368e-02 -7.3681122e-01 -1.0023915e+00
-6.7274171e-01 2.1250947e-01 3.1965587e-01 7.1426946e-01
3.4896237e-01 -7.7565098e-01 -7.7081275e-01 3.5930723e-03
-4.4867951e-02 -1.1946801e-01 4.2548593e-02 5.9469312e-01
-1.7626689e-01 5.2385128e-01 4.8855670e-02 -4.4617063e-01
-1.3683168e+00 2.4801368e-01 5.8211088e-01 -2.5624922e-01
-5.3746146e-01 1.2316682e+00 2.6777259e-01 -9.2848724e-01
6.0631919e-01 -2.2830562e-01 2.7022567e-01 -1.1898619e-01
2.1312819e-01 1.0773373e-01 2.5397110e-01 -5.1099789e-01
-5.2156959e-02 -5.4358399e-01 -3.1407377e-01 -8.6626405e-01
1.1919274e+00 -6.1458819e-02 5.1491058e-01 4.5715240e-01
1.0876472e+00 -2.2681072e-01 -1.4499352e+00 -2.3717418e-01
-2.7370659e-01 -2.9804127e-02 -1.9007455e-01 -1.2385510e+00
-6.1878276e-01 1.1115565e+00 1.2492778e-01 4.4573611e-01
5.2871346e-01 4.7921702e-01 1.1516517e+00 6.9554090e-01
2.6803124e-01 -9.0890557e-01 3.7901756e-01 1.3348083e+00
1.0611080e+00 -1.1823841e+00 -7.9832107e-01 -3.3451146e-01
-1.2501365e+00 6.9678485e-01 7.6163656e-01 -1.1471501e-01
3.0192462e-01 4.1049904e-01 5.2618802e-01 -1.2196071e-01
-1.1373291e+00 -1.9984704e-01 -2.9574814e-01 8.1723571e-01
5.1358229e-01 -1.4510629e-01 1.4953944e-01 7.6982427e-01
-7.7736717e-01 -2.5508785e-01 6.6642225e-01 7.8472954e-01
-3.7104496e-01 -1.2063361e+00 -2.7600828e-01 1.0677580e-01
-3.6689270e-01 -4.7548109e-01 -1.0741197e+00 5.4363984e-01
-4.7244734e-01 1.4704399e+00 -8.4510848e-02 -3.8276196e-01
3.6578313e-01 6.8744701e-01 3.1200382e-01 -1.1029398e+00
-1.1722894e+00 -1.4105956e-01 8.4931016e-01 -1.3821475e-01
1.8779112e-01 -4.7695529e-01 -1.1448160e+00 -1.9144288e-01
-8.6017644e-01 2.7002051e-01 5.6758136e-01 1.0543201e+00
2.5429294e-01 5.5847973e-01 2.8111279e-01 -7.6194632e-01
-1.1813382e+00 -1.5774714e+00 -1.1560590e-01 2.4024393e-01
2.3281389e-01 -4.9325779e-01 -3.2585028e-01 8.3556727e-02] | [12.826552391052246, 8.055208206176758] |
df1e604e-4ec0-4503-beca-695b46b9042f | strengthening-structural-baselines-for-graph | 2305.00724 | null | https://arxiv.org/abs/2305.00724v1 | https://arxiv.org/pdf/2305.00724v1.pdf | Strengthening structural baselines for graph classification using Local Topological Profile | We present the analysis of the topological graph descriptor Local Degree Profile (LDP), which forms a widely used structural baseline for graph classification. Our study focuses on model evaluation in the context of the recently developed fair evaluation framework, which defines rigorous routines for model selection and evaluation for graph classification, ensuring reproducibility and comparability of the results. Based on the obtained insights, we propose a new baseline algorithm called Local Topological Profile (LTP), which extends LDP by using additional centrality measures and local vertex descriptors. The new approach provides the results outperforming or very close to the latest GNNs for all datasets used. Specifically, state-of-the-art results were obtained for 4 out of 9 benchmark datasets. We also consider computational aspects of LDP-based feature extraction and model construction to propose practical improvements affecting execution speed and scalability. This allows for handling modern, large datasets and extends the portfolio of benchmarks used in graph representation learning. As the outcome of our work, we obtained LTP as a simple to understand, fast and scalable, still robust baseline, capable of outcompeting modern graph classification models such as Graph Isomorphism Network (GIN). We provide open-source implementation at \href{https://github.com/j-adamczyk/LTP}{GitHub}. | ['Wojciech Czech', 'Jakub Adamczyk'] | 2023-05-01 | null | null | null | null | ['graph-classification'] | ['graphs'] | [-2.22121067e-02 6.84373155e-02 -4.43919063e-01 -2.90897745e-03
-2.77857333e-01 -6.49554968e-01 9.04139102e-01 9.44876730e-01
-2.20440969e-01 5.19775748e-01 2.44971607e-02 -4.18805689e-01
-7.74161756e-01 -1.12376368e+00 -3.04719627e-01 -5.37042260e-01
-7.04252899e-01 6.89197421e-01 3.54261160e-01 -3.88663858e-01
4.45834100e-01 8.31160903e-01 -1.49948156e+00 1.06652595e-01
6.45291328e-01 1.00496519e+00 -1.08980842e-01 6.78328097e-01
6.91479519e-02 5.58202267e-01 -2.07502156e-01 -4.15591598e-01
1.80563420e-01 -9.12771374e-02 -1.03487790e+00 -4.45727974e-01
5.04844487e-01 3.64189535e-01 -5.37438393e-01 7.51033664e-01
4.42954898e-01 1.27464384e-01 7.29189992e-01 -1.37720740e+00
-3.91313642e-01 6.58097029e-01 -2.52454489e-01 3.56662780e-01
6.69234276e-01 -6.71142787e-02 1.43436766e+00 -6.67409718e-01
1.01167512e+00 9.82852697e-01 7.88188457e-01 4.03622016e-02
-1.42786288e+00 -2.65071660e-01 -4.96340357e-02 4.31130201e-01
-1.56425214e+00 -2.59200066e-01 7.79471636e-01 -4.14715022e-01
8.79616857e-01 6.63283885e-01 7.19736874e-01 9.44439471e-01
1.93121269e-01 3.31427902e-01 1.13704455e+00 -4.66597527e-01
-7.89518096e-03 -1.52148560e-01 4.36059982e-01 1.02409911e+00
5.05897939e-01 1.30438998e-01 -2.45277867e-01 -4.59864050e-01
5.79166293e-01 -3.14239472e-01 -2.07889825e-01 -7.28569090e-01
-1.30775583e+00 6.78733885e-01 6.72312200e-01 5.43982685e-01
-1.98529616e-01 8.68026242e-02 6.30025983e-01 6.09136820e-01
5.16999424e-01 3.84876162e-01 -1.95841357e-01 -1.04262702e-01
-7.36209631e-01 1.88805625e-01 9.76384997e-01 7.23101079e-01
7.67080247e-01 -2.72357225e-01 -1.76521853e-01 7.12172687e-01
5.58412820e-02 -1.06182778e-02 1.48774281e-01 -5.06354511e-01
3.06409925e-01 1.02189147e+00 -6.24746323e-01 -1.62785769e+00
-7.22910881e-01 -5.82166076e-01 -1.01372170e+00 -4.27800305e-02
3.25734764e-01 6.67013407e-01 -4.90496993e-01 1.43505669e+00
3.00066262e-01 1.98705032e-01 -3.55261624e-01 4.34249282e-01
1.08227873e+00 2.05158427e-01 7.63254836e-02 -2.66913865e-02
1.20859265e+00 -8.12895715e-01 -8.84997174e-02 3.10589224e-01
1.02293658e+00 -6.47304237e-01 9.48481798e-01 3.27098161e-01
-7.33309746e-01 -4.65785623e-01 -9.43405926e-01 2.11857259e-02
-7.80585468e-01 5.09485379e-02 9.47288573e-01 5.75017154e-01
-1.51436365e+00 1.06541467e+00 -7.91564345e-01 -8.79937530e-01
2.09191054e-01 4.51194406e-01 -8.38001013e-01 -6.82798699e-02
-9.80544448e-01 7.85159469e-01 5.70297360e-01 -1.61069855e-01
-4.93605703e-01 -6.07436538e-01 -7.27318585e-01 8.82480200e-03
4.03778464e-01 -6.73862040e-01 4.26777303e-01 -4.66861308e-01
-1.15339839e+00 1.13640690e+00 2.26513833e-01 -5.29167473e-01
4.22455370e-01 4.19104308e-01 -5.51835835e-01 2.07310900e-01
-1.12183169e-01 3.03270459e-01 3.87954175e-01 -1.04285705e+00
2.12662313e-02 -2.60791004e-01 1.49015680e-01 -1.05330206e-01
-3.82626772e-01 -7.72123486e-02 -5.01727581e-01 -6.31824553e-01
1.04807191e-01 -9.21211183e-01 -1.42344579e-01 -1.97940886e-01
-5.31763613e-01 -3.20251912e-01 5.14630735e-01 -5.03442287e-01
1.54467833e+00 -1.79738319e+00 3.92435580e-01 8.76450539e-01
7.49028683e-01 2.37138748e-01 -2.89445758e-01 1.13725781e+00
-4.65548515e-01 3.60858709e-01 -1.28155991e-01 -1.60792336e-01
4.93217967e-02 3.37059200e-02 7.61040375e-02 6.12430096e-01
-7.92732686e-02 9.69313860e-01 -8.48559618e-01 -5.16753376e-01
4.18935150e-01 3.66071939e-01 -3.22700500e-01 -1.26521915e-01
5.96919432e-02 3.46561581e-01 -3.40586662e-01 6.10728204e-01
5.73194385e-01 -4.24497336e-01 4.96338457e-01 -2.60203689e-01
1.05131522e-01 2.00475708e-01 -1.18732166e+00 1.61389029e+00
-1.89541906e-01 3.56525749e-01 -1.84053034e-01 -1.33822787e+00
1.21600318e+00 -8.35825726e-02 6.02767348e-01 -6.08695388e-01
2.30633229e-01 2.78159738e-01 5.16279461e-03 -1.10909961e-01
2.43603125e-01 5.85708737e-01 1.79112982e-02 3.21243972e-01
3.74695033e-01 9.11986455e-02 6.88088715e-01 5.87982059e-01
1.57624280e+00 -9.23318043e-02 6.43230557e-01 -8.66385341e-01
8.51460397e-01 -2.12804303e-01 -5.99547401e-02 6.18053615e-01
-1.36725277e-01 3.85625511e-01 1.02395856e+00 -6.78508759e-01
-8.41022789e-01 -1.04920948e+00 -4.12539728e-02 1.11701441e+00
-5.18360175e-03 -1.21854651e+00 -5.36523223e-01 -7.69511640e-01
1.29101261e-01 1.42719567e-01 -7.43255436e-01 -1.54907078e-01
-5.03509164e-01 -6.90615177e-01 6.73116386e-01 8.58456641e-02
6.67022020e-02 -8.50100100e-01 2.23818779e-01 -1.07214615e-01
5.89231178e-02 -1.06713319e+00 -9.32423249e-02 -1.26776099e-02
-9.42231834e-01 -1.50090647e+00 -2.52126753e-01 -6.92264497e-01
6.59108222e-01 1.07949495e-01 1.51904523e+00 7.73580909e-01
-4.32602674e-01 4.32822615e-01 -4.75199312e-01 3.37813288e-01
-4.78321612e-01 6.43539608e-01 -3.42714973e-02 -1.01182707e-01
-1.00060180e-01 -8.61118793e-01 -5.60825169e-01 3.12308937e-01
-7.41775870e-01 -9.00171697e-03 4.38786417e-01 5.81793070e-01
5.64971805e-01 -6.64888471e-02 2.92773724e-01 -1.17726111e+00
7.49062061e-01 -5.34132302e-01 -5.95472872e-01 2.81305194e-01
-9.56385672e-01 2.02601090e-01 7.74827719e-01 8.24603438e-02
-8.25656354e-02 -3.42465937e-01 -1.43938422e-01 -9.09735039e-02
-3.72043625e-02 7.75548458e-01 7.23933131e-02 -7.40154266e-01
6.47380888e-01 6.86047301e-02 7.34457150e-02 -4.99399394e-01
3.03531796e-01 2.12160751e-01 2.31195390e-01 -7.64391124e-01
8.04560483e-01 3.83175611e-01 6.17134750e-01 -9.85703409e-01
-1.70530200e-01 -6.02871001e-01 -8.86574388e-01 -2.34149650e-01
2.79489905e-01 -4.36089307e-01 -8.71899366e-01 2.76781917e-01
-7.08789587e-01 -2.18103886e-01 -2.90770456e-03 1.69276774e-01
-4.95970070e-01 7.56644726e-01 -4.56626713e-01 -4.44800466e-01
-3.34820330e-01 -8.86004031e-01 8.35234702e-01 -3.52170944e-01
-2.17228547e-01 -1.51660013e+00 3.47965360e-01 2.15011656e-01
5.78244627e-01 8.42315257e-01 1.28344524e+00 -8.91907573e-01
-5.16546845e-01 -3.71488899e-01 -2.93744832e-01 1.74577892e-01
-9.48471427e-02 3.48165572e-01 -5.78973472e-01 -6.14732325e-01
-7.95827508e-01 3.74942720e-02 1.04403019e+00 -3.15856002e-02
1.24875069e+00 -1.32320806e-01 -5.85068882e-01 7.59577811e-01
1.64657187e+00 -5.48727751e-01 6.34577751e-01 3.40819985e-01
7.80307055e-01 5.98348379e-01 3.03489476e-01 1.78362563e-01
4.60801035e-01 9.12084818e-01 5.30015886e-01 -1.07339434e-01
-3.64243895e-01 -2.65202314e-01 1.50727080e-02 1.14944851e+00
-6.40807390e-01 -2.72762269e-01 -1.30014670e+00 4.26524729e-01
-1.92024314e+00 -7.71200120e-01 -5.20334840e-01 2.44865608e+00
2.40497991e-01 2.44750977e-01 4.47087437e-01 2.80131161e-01
6.73198044e-01 3.63508135e-01 1.28527954e-01 -4.09583420e-01
-1.06765509e-01 6.45439148e-01 6.61032200e-01 6.38601601e-01
-1.04806399e+00 9.02750790e-01 5.99988270e+00 9.76181388e-01
-1.00288057e+00 1.07111176e-02 4.22941655e-01 3.78483206e-01
-1.55828819e-01 2.15314209e-01 -3.65430802e-01 2.09242821e-01
1.26482999e+00 -4.21581566e-01 6.40344024e-01 6.43983841e-01
1.20833144e-01 1.90812930e-01 -1.08530521e+00 8.66325319e-01
1.67242587e-02 -1.43749726e+00 1.48712933e-01 2.64209598e-01
4.08458441e-01 3.57491672e-01 -1.74975887e-01 2.40681753e-01
5.29875644e-02 -1.06866658e+00 2.96332181e-01 5.94637990e-01
8.22564542e-01 -6.31711602e-01 6.50716662e-01 -6.24469519e-02
-1.73092556e+00 1.16903381e-02 -2.47436702e-01 -6.52236193e-02
-3.08701307e-01 6.45972252e-01 -7.57938385e-01 1.33970451e+00
3.74965549e-01 9.18401062e-01 -1.14274514e+00 1.06287813e+00
-1.45033337e-02 5.11465430e-01 -3.86795402e-01 4.83736470e-02
1.17132634e-01 -2.88636267e-01 5.55885732e-01 1.37908840e+00
1.82106897e-01 -4.87361819e-01 3.06837380e-01 5.21231294e-01
-2.02911720e-01 6.47122979e-01 -9.40222144e-01 -1.86484575e-01
4.39372867e-01 1.57797694e+00 -1.15676200e+00 -8.24413896e-02
-1.53758869e-01 4.68623728e-01 7.82369256e-01 5.48810810e-02
-7.57576644e-01 -5.28680623e-01 3.61530781e-01 4.33572650e-01
1.02093711e-01 -3.73796731e-01 4.93813977e-02 -1.15310287e+00
1.22235447e-01 -7.99123049e-01 7.54060924e-01 -1.71991691e-01
-1.30963159e+00 8.62046003e-01 2.68826813e-01 -1.15870547e+00
-3.11868563e-02 -9.05575633e-01 -7.83756137e-01 4.89355683e-01
-1.36124516e+00 -1.38111544e+00 -6.48756325e-01 5.86551130e-01
-2.80020088e-01 -1.83839828e-01 1.05034912e+00 4.24043477e-01
-4.46026236e-01 6.89003289e-01 1.13902047e-01 -3.59848700e-02
4.89960700e-01 -1.30891728e+00 6.38852775e-01 6.10652030e-01
3.40433508e-01 6.25054181e-01 4.69466716e-01 -4.18212444e-01
-1.40219092e+00 -9.17962790e-01 8.35553229e-01 -5.10354161e-01
1.06086469e+00 -6.10641479e-01 -8.12810421e-01 4.82513040e-01
-8.98065194e-02 1.75875083e-01 4.82643306e-01 3.37035000e-01
-4.96057600e-01 -2.44231552e-01 -9.97974932e-01 3.67693216e-01
1.40823209e+00 -5.67432404e-01 -7.87936300e-02 5.57926178e-01
3.50986272e-01 -8.03197846e-02 -1.38437033e+00 5.37037075e-01
4.66064245e-01 -1.16111362e+00 1.07114863e+00 -4.82748210e-01
-4.39886004e-02 -8.45049769e-02 -3.94200608e-02 -1.13803244e+00
-5.65987051e-01 -7.23265707e-01 -1.98273197e-01 1.10750067e+00
4.52614397e-01 -1.03667605e+00 7.64430940e-01 -9.45789963e-02
-4.37211394e-02 -1.01100588e+00 -9.88437057e-01 -8.95895302e-01
2.45438404e-02 -2.61809200e-01 5.88325858e-01 1.04775739e+00
2.41474494e-01 8.38271156e-02 3.36710699e-02 -9.68663022e-02
6.58322096e-01 6.13746159e-02 9.05813098e-01 -1.53736341e+00
-1.60887614e-01 -8.12235653e-01 -1.24995708e+00 -3.43558908e-01
2.76123047e-01 -1.72113478e+00 -8.06533873e-01 -1.66206944e+00
1.22699454e-01 -5.55032492e-01 -5.98648667e-01 4.37857896e-01
1.14653684e-01 5.11087656e-01 1.97203293e-01 2.88471371e-01
-6.99556649e-01 2.97595561e-01 1.00862396e+00 -9.87560581e-03
1.12831190e-01 7.43889017e-03 -4.90186423e-01 3.92029375e-01
8.20446193e-01 -3.26289475e-01 -3.13252628e-01 4.11739834e-02
2.72840202e-01 -2.84086645e-01 4.82260168e-01 -1.17693722e+00
1.70247838e-01 1.52569443e-01 -4.01573442e-02 -2.24567667e-01
1.34433508e-01 -5.73437095e-01 3.20396394e-01 6.96966767e-01
-1.29475698e-01 3.49585116e-01 -5.17836474e-02 4.88671541e-01
-1.95505634e-01 5.22434935e-02 6.33846045e-01 -5.71340360e-02
-7.55413413e-01 6.34313345e-01 2.05301479e-01 3.07877120e-02
1.06857634e+00 -1.51702106e-01 -7.47450233e-01 -1.69488773e-01
-8.33738983e-01 8.81619677e-02 6.98614955e-01 4.54464287e-01
3.92286956e-01 -1.35990036e+00 -8.45280230e-01 1.70261562e-01
4.68041956e-01 -7.26524055e-01 8.92044455e-02 1.24984503e+00
-9.44297671e-01 5.03756762e-01 -3.36777657e-01 -5.24653852e-01
-1.39157581e+00 6.15627110e-01 1.87843710e-01 -7.08528817e-01
-5.80782771e-01 5.30107558e-01 -2.19615147e-01 -8.05595160e-01
2.74251029e-02 -3.43950838e-01 -2.93676287e-01 3.20852287e-02
5.08898124e-02 6.43900037e-01 5.73272705e-01 -7.85841227e-01
-6.09862506e-01 6.01119101e-01 1.46790713e-01 4.44805235e-01
1.34523165e+00 7.79606625e-02 -7.04180777e-01 3.65169942e-01
1.30354774e+00 -1.85385477e-02 -3.20843935e-01 -1.31476214e-02
3.34208429e-01 -5.05389869e-01 -1.34847686e-01 -5.31476438e-01
-9.15586293e-01 5.05824268e-01 4.27440166e-01 7.39652395e-01
9.69064116e-01 1.08199559e-01 3.11303914e-01 3.74965608e-01
6.43664718e-01 -5.73792517e-01 -1.40712515e-01 4.86097515e-01
9.02478576e-01 -9.30753648e-01 3.72216254e-01 -7.48377621e-01
-1.75931349e-01 1.23829448e+00 3.04522723e-01 -3.74828190e-01
7.85759032e-01 -1.91100538e-01 -4.05675799e-01 -5.61143100e-01
-8.75747681e-01 -2.41683900e-01 6.09835386e-01 6.20468378e-01
5.70779204e-01 2.60855496e-01 -7.17253745e-01 -1.45346755e-02
-3.89763415e-01 -3.88179213e-01 3.54191095e-01 6.68848932e-01
-1.04670152e-01 -1.60691321e+00 1.79291114e-01 6.94698691e-01
-2.26410434e-01 -1.11218542e-01 -6.98869824e-01 1.19737077e+00
-1.75417766e-01 6.09632909e-01 -3.41611892e-01 -6.67812169e-01
2.54381418e-01 -1.40108839e-01 5.42701781e-01 -5.05781114e-01
-6.76741362e-01 -5.20141602e-01 4.22934562e-01 -8.12873304e-01
-3.48844916e-01 -3.52099538e-01 -8.15913260e-01 -7.95753539e-01
-2.73030818e-01 3.08061540e-01 5.69120944e-01 5.31502306e-01
5.85283637e-01 4.03042704e-01 5.35542250e-01 -8.39815199e-01
-2.12667942e-01 -9.37348545e-01 -6.37244821e-01 5.10037661e-01
-6.28357828e-02 -7.46341348e-01 -4.55368429e-01 -4.02461112e-01] | [7.0267438888549805, 5.938589096069336] |
225bb5ff-f4f8-4be0-b60f-066a9f6fc7e4 | the-decomposition-of-the-higher-order | 2107.10970 | null | https://arxiv.org/abs/2107.10970v3 | https://arxiv.org/pdf/2107.10970v3.pdf | The decomposition of the higher-order homology embedding constructed from the $k$-Laplacian | The null space of the $k$-th order Laplacian $\mathbf{\mathcal L}_k$, known as the {\em $k$-th homology vector space}, encodes the non-trivial topology of a manifold or a network. Understanding the structure of the homology embedding can thus disclose geometric or topological information from the data. The study of the null space embedding of the graph Laplacian $\mathbf{\mathcal L}_0$ has spurred new research and applications, such as spectral clustering algorithms with theoretical guarantees and estimators of the Stochastic Block Model. In this work, we investigate the geometry of the $k$-th homology embedding and focus on cases reminiscent of spectral clustering. Namely, we analyze the {\em connected sum} of manifolds as a perturbation to the direct sum of their homology embeddings. We propose an algorithm to factorize the homology embedding into subspaces corresponding to a manifold's simplest topological components. The proposed framework is applied to the {\em shortest homologous loop detection} problem, a problem known to be NP-hard in general. Our spectral loop detection algorithm scales better than existing methods and is effective on diverse data such as point clouds and images. | ['Marina Meilă', 'Yu-Chia Chen'] | 2021-07-23 | null | http://proceedings.neurips.cc/paper/2021/hash/842424a1d0595b76ec4fa03c46e8d755-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/842424a1d0595b76ec4fa03c46e8d755-Paper.pdf | neurips-2021-12 | ['stochastic-block-model'] | ['graphs'] | [ 4.14460376e-02 4.85215694e-01 7.10216239e-02 6.12968095e-02
-3.52939874e-01 -7.85266936e-01 1.50558412e-01 4.56975214e-02
-1.03090882e-01 8.10860991e-02 -2.81876355e-01 -5.13395727e-01
-6.36495471e-01 -8.74072373e-01 -8.55748951e-01 -1.01821208e+00
-5.73660851e-01 5.07435322e-01 1.29947573e-01 -1.26934707e-01
3.37290764e-01 4.40028369e-01 -1.52429974e+00 -4.67445970e-01
8.51297557e-01 8.83967102e-01 9.49459001e-02 5.64803660e-01
-1.28324062e-01 1.20909605e-02 -1.50601054e-02 -2.36894831e-01
4.80504662e-01 -5.30283451e-01 -9.63916361e-01 1.40645728e-01
6.68729722e-01 4.18995053e-01 -4.77113783e-01 1.62044382e+00
8.02837759e-02 8.67324024e-02 7.31812477e-01 -1.52416849e+00
-5.82131088e-01 3.11507910e-01 -6.90745592e-01 2.48653948e-01
1.74404666e-01 -1.87586129e-01 1.24864423e+00 -9.12420809e-01
8.73000801e-01 1.11048841e+00 5.76178074e-01 6.24192618e-02
-1.75326884e+00 -4.88278329e-01 -3.02746177e-01 1.81168646e-01
-1.90979314e+00 -9.73170325e-02 1.13229084e+00 -9.73499358e-01
3.38540018e-01 4.66961384e-01 5.07442832e-01 1.71814397e-01
7.51397535e-02 1.83822632e-01 1.07987702e+00 -4.54182059e-01
1.82763696e-01 5.15038632e-02 2.94091403e-01 1.27363575e+00
4.71233010e-01 -3.17711025e-01 -1.93927601e-01 -2.55651921e-01
8.33660841e-01 -2.10213453e-01 -4.92435217e-01 -9.38502312e-01
-1.15864599e+00 9.49327409e-01 2.98062861e-01 2.82880157e-01
3.02808341e-02 7.25096017e-02 -1.22646622e-01 3.22263330e-01
3.18752676e-01 1.50437862e-01 1.45020366e-01 3.24320376e-01
-7.73364723e-01 -1.25318393e-01 9.32845294e-01 1.12857747e+00
1.46339846e+00 -3.30661803e-01 7.30818808e-01 4.05037224e-01
4.18540835e-01 5.29800713e-01 -2.28178546e-01 -1.19942677e+00
4.37183887e-01 9.77318764e-01 -1.38763040e-01 -1.67956817e+00
-5.91395199e-01 -2.14735344e-01 -1.26614594e+00 -1.90727338e-02
4.96712774e-01 3.38142991e-01 -2.16138229e-01 1.90967655e+00
4.28135246e-01 1.50534540e-01 -3.47935647e-01 7.06233859e-01
2.71804750e-01 4.79159892e-01 -4.39771265e-01 -2.89346814e-01
1.09170938e+00 -2.37094611e-01 -1.69717491e-01 2.66764522e-01
8.33577275e-01 -6.48465812e-01 1.01157236e+00 -3.97311077e-02
-1.06505418e+00 -2.40069777e-01 -1.14990878e+00 2.21755058e-01
-2.54348248e-01 -1.32664777e-02 -2.95576770e-02 5.10704517e-01
-1.32101917e+00 7.00855255e-01 -5.59756100e-01 -5.16694427e-01
8.19866918e-03 4.26415324e-01 -5.20638049e-01 -1.45577127e-02
-8.37505996e-01 3.73841465e-01 2.66614437e-01 1.26551660e-02
-2.15782106e-01 -4.96431261e-01 -8.06459129e-01 -2.94641614e-01
3.56732905e-01 -3.33390057e-01 2.63130844e-01 -5.42676985e-01
-7.27209032e-01 1.15587592e+00 -1.40345901e-01 -7.50432089e-02
3.21235597e-01 5.27025819e-01 -2.52473146e-01 4.60539669e-01
2.16292471e-01 2.27097049e-01 9.74546731e-01 -1.04710591e+00
-5.25257327e-02 -7.23172426e-01 -5.59868990e-03 -2.18652055e-01
-2.96555340e-01 -4.41627324e-01 4.81673516e-02 -5.04598260e-01
8.10198724e-01 -1.38510633e+00 -8.02709758e-02 -1.16470538e-01
-5.23747325e-01 -3.45371991e-01 1.01324439e+00 -5.22114992e-01
1.46773493e+00 -2.30279374e+00 7.19356894e-01 8.13939929e-01
5.05274713e-01 -1.56994045e-01 1.13366097e-01 7.64179587e-01
-3.08690429e-01 4.70082223e-01 -4.43505108e-01 2.12654173e-01
4.48893197e-02 -5.49669452e-02 -1.49780020e-01 1.36043358e+00
-1.04620129e-01 3.80709708e-01 -7.02445865e-01 -5.13508499e-01
8.05418491e-02 2.57046789e-01 -4.55489486e-01 -1.13050386e-01
-1.47798747e-01 3.87169451e-01 -3.23806942e-01 1.95444256e-01
9.66883302e-01 -5.56795418e-01 5.64277172e-02 -3.52000028e-01
-2.34408379e-01 -1.17637925e-01 -1.53618574e+00 1.58261752e+00
1.19571947e-01 6.59014821e-01 5.80601692e-01 -1.45077264e+00
7.19202697e-01 -1.29447162e-01 7.76239455e-01 -2.62202710e-01
5.76639026e-02 7.73805752e-02 -1.45673439e-01 -3.02738369e-01
1.22189932e-01 -3.84206064e-02 -6.94870576e-02 5.82951486e-01
-1.46362215e-01 1.60524398e-01 1.60733640e-01 4.72494036e-01
1.31259191e+00 -5.01539826e-01 -8.37328210e-02 -1.12851655e+00
8.01905513e-01 -1.35047883e-01 3.27795923e-01 3.18278730e-01
-7.59735480e-02 3.04650813e-01 7.53838658e-01 -1.15003578e-01
-1.20189786e+00 -1.31041467e+00 -3.07935268e-01 6.62333786e-01
5.35418451e-01 -6.05047166e-01 -1.12012589e+00 -2.54842132e-01
1.01612478e-01 4.45196211e-01 -5.25057077e-01 -4.43965882e-01
-5.34015119e-01 -5.48247576e-01 3.46275896e-01 -1.25077322e-01
4.36630607e-01 -4.95383322e-01 -3.71966600e-01 -1.03910811e-01
-3.03487152e-01 -1.03662360e+00 -7.79162586e-01 -3.36940318e-01
-9.53916728e-01 -1.47759640e+00 -2.54787117e-01 -8.11426342e-01
8.87343168e-01 3.85580063e-01 8.38324070e-01 1.30879581e-01
-6.16007209e-01 8.70338082e-01 -8.19287598e-02 2.54230857e-01
-2.72791207e-01 -7.59614259e-02 3.08281511e-01 3.41833949e-01
1.29110411e-01 -7.55271137e-01 -8.73474658e-01 7.80761838e-01
-1.09614813e+00 -1.31394088e-01 2.23044828e-01 3.40962201e-01
8.06404471e-01 5.48832178e-01 8.95174742e-02 -3.71266007e-01
2.83785999e-01 -4.88960415e-01 -8.09288800e-01 1.14558287e-01
-5.92662215e-01 4.54447389e-01 6.26165509e-01 -5.61778955e-02
-1.43756300e-01 4.50910255e-02 4.23935294e-01 -5.92525244e-01
2.66767561e-01 3.66780192e-01 -2.51624584e-01 -5.08811474e-01
3.31162453e-01 4.54883844e-01 1.34670302e-01 -6.00871384e-01
4.60087955e-01 2.71121800e-01 4.63998973e-01 -4.33849931e-01
1.08957124e+00 9.17253852e-01 6.63241982e-01 -1.53987443e+00
-4.98466164e-01 -7.28453338e-01 -8.57121289e-01 -1.76263571e-01
8.59390676e-01 -5.14447987e-01 -9.88961875e-01 -3.26935481e-03
-9.70354557e-01 2.97149387e-03 -1.03397638e-01 2.91739821e-01
-1.01641691e+00 8.12232792e-01 -4.12452430e-01 -7.15607345e-01
2.08744574e-02 -9.95365381e-01 1.03637540e+00 -3.58211011e-01
-7.57517368e-02 -1.01580882e+00 4.56789941e-01 2.50356317e-01
-8.44242573e-02 5.02894759e-01 1.40588450e+00 -1.47193760e-01
-9.18510258e-01 -2.24596500e-01 -2.14781180e-01 1.88013807e-01
-2.21239254e-01 -3.94898839e-02 -3.25368077e-01 -5.94475985e-01
1.70202628e-01 3.45493138e-01 6.58297777e-01 1.28879011e-01
1.05285668e+00 -6.01821125e-01 -3.13614130e-01 6.33558273e-01
1.46005702e+00 -3.73981982e-01 3.17594498e-01 -4.70093675e-02
8.47482145e-01 7.55988002e-01 1.57991722e-01 2.43578568e-01
2.60464638e-01 6.17756963e-01 4.70053494e-01 3.56168181e-01
2.14319170e-01 -2.39190951e-01 4.62979823e-01 1.32243216e+00
4.51022536e-02 4.28818673e-01 -9.75293756e-01 4.44833636e-01
-1.69890761e+00 -1.02504420e+00 -3.54103684e-01 2.47635436e+00
3.73323679e-01 -8.04526731e-02 2.41833180e-01 2.06801161e-01
1.09991920e+00 7.96374679e-02 -3.92031729e-01 -5.73457330e-02
-3.33753645e-01 3.32621813e-01 6.74013853e-01 7.42909014e-01
-9.17479455e-01 5.83621621e-01 4.94485903e+00 7.35080481e-01
-8.48893762e-01 1.77872479e-01 3.83629277e-02 3.29129249e-01
-4.24261063e-01 3.22502315e-01 -4.41462249e-01 5.51882625e-01
8.35325658e-01 -3.73978645e-01 5.04278839e-01 7.29756892e-01
1.67951047e-01 -1.11874558e-01 -1.16043866e+00 1.15035319e+00
6.40060678e-02 -1.46851027e+00 2.09702682e-02 7.91094661e-01
5.62113285e-01 -1.16395973e-01 3.98328274e-01 -4.87856477e-01
6.87849820e-02 -9.50573444e-01 5.89689672e-01 4.04646009e-01
1.08368027e+00 -7.84253836e-01 1.04798041e-01 5.17048538e-01
-1.76440847e+00 1.41316071e-01 -3.67222607e-01 2.23185956e-01
-8.27155486e-02 5.62750041e-01 -6.22518003e-01 4.43036109e-01
7.37731218e-01 7.21502066e-01 -5.51113486e-01 7.42236674e-01
4.79449749e-01 4.62563336e-01 -4.48357880e-01 2.68262327e-01
9.87169370e-02 -1.14639914e+00 9.83685374e-01 8.46629798e-01
5.18406928e-01 4.97321904e-01 2.96725512e-01 1.01186991e+00
-1.28739879e-01 3.63442361e-01 -1.02753341e+00 -2.45936245e-01
3.50395292e-01 1.09411359e+00 -1.10330594e+00 1.46573126e-01
-2.33876064e-01 9.23592925e-01 3.26493353e-01 2.20486611e-01
-4.29325134e-01 -5.02568841e-01 9.07183051e-01 6.50178492e-01
3.42363924e-01 -7.38333166e-01 -1.33305369e-02 -1.13591874e+00
2.08233610e-01 -4.16293710e-01 2.68311530e-01 -3.98761511e-01
-1.08247244e+00 2.17237964e-01 -2.78394148e-02 -1.01548266e+00
1.73465967e-01 -6.69877648e-01 -4.01814967e-01 6.26913965e-01
-6.99708521e-01 -5.77477634e-01 -1.95783138e-01 9.42090213e-01
-2.42447212e-01 2.24375471e-01 7.29268312e-01 1.22424953e-01
-4.54505384e-01 3.09869021e-01 5.82643211e-01 2.10417822e-01
1.54638454e-01 -1.25262594e+00 -5.42177558e-02 5.92142403e-01
1.69746473e-01 6.89519584e-01 8.04235816e-01 -5.62394977e-01
-2.05444431e+00 -1.19084287e+00 3.98263425e-01 -5.18402219e-01
1.09422481e+00 -5.81705213e-01 -8.34110737e-01 7.29707301e-01
-2.89300263e-01 -7.11388066e-02 6.46137297e-01 -4.29720968e-01
-4.52580154e-01 -1.97102249e-01 -1.26604629e+00 4.61522311e-01
1.38280237e+00 -8.00295055e-01 -1.63036063e-01 3.80978078e-01
7.32806861e-01 3.04794014e-01 -1.08424187e+00 3.07810068e-01
8.62931758e-02 -1.08876765e+00 9.68505144e-01 -3.52028996e-01
-1.34015158e-01 -5.03490090e-01 -4.83089298e-01 -8.47861946e-01
-3.02683532e-01 -9.62109149e-01 6.97848271e-04 7.07249582e-01
1.01935305e-01 -4.60063338e-01 9.69479620e-01 6.00651279e-02
1.50914758e-01 -4.55605268e-01 -1.45558763e+00 -7.61606395e-01
2.27611437e-01 -3.18551689e-01 -2.42909230e-02 1.03826272e+00
2.04737172e-01 2.66952634e-01 1.28205523e-01 5.18008888e-01
1.41990197e+00 1.66159749e-01 6.06890321e-01 -1.66620338e+00
-1.76405776e-02 -5.81398070e-01 -9.21315551e-01 -7.35079765e-01
3.61735612e-01 -1.30483222e+00 -2.83185959e-01 -9.07083690e-01
1.70482904e-01 -5.67755163e-01 4.16080244e-02 -2.72683889e-01
5.42097807e-01 1.00298114e-01 2.44518667e-02 4.48086411e-01
-6.92645371e-01 3.57554197e-01 8.77978146e-01 -1.22847356e-01
-8.43179971e-02 -2.50418603e-01 -2.40619659e-01 8.83542299e-01
5.26685536e-01 -1.53143078e-01 -2.64339924e-01 -5.13042957e-02
6.98141217e-01 1.34317070e-01 6.20136976e-01 -9.49251890e-01
3.75698537e-01 4.85580740e-03 -3.37019205e-01 -6.96323156e-01
2.49179617e-01 -8.84843767e-01 5.06914258e-01 3.66104990e-01
-6.32768944e-02 3.23660970e-01 -1.85790062e-01 1.04086161e+00
2.15242300e-02 -2.03540511e-02 9.64753807e-01 -1.29092885e-02
-2.14717075e-01 5.46995699e-01 -9.32021514e-02 3.58505934e-01
1.38405764e+00 -2.85684288e-01 -1.68671444e-01 -3.29197317e-01
-8.25911522e-01 6.86826035e-02 9.83476937e-01 -2.22069528e-02
5.86243927e-01 -1.41356802e+00 -3.93435448e-01 2.51867324e-01
2.33597234e-01 5.29746991e-03 1.92553595e-01 1.26230037e+00
-6.29397392e-01 4.42039728e-01 7.26233572e-02 -9.66043711e-01
-1.18232954e+00 7.66895831e-01 3.59049141e-01 8.56000185e-02
-5.38095891e-01 6.02735400e-01 3.37893218e-01 -2.47374609e-01
5.27541228e-02 1.61982309e-02 3.54196429e-01 -4.19130437e-02
2.00077355e-01 8.90425444e-01 -1.95320964e-01 -1.32887030e+00
-2.85394758e-01 1.08861077e+00 3.48849267e-01 -1.15158558e-01
9.76035535e-01 -4.22243863e-01 -9.80062962e-01 5.77061951e-01
1.89364409e+00 -3.26869152e-02 -7.19488978e-01 -2.23621070e-01
2.82642752e-01 -3.04401487e-01 -3.63331169e-01 2.54886121e-01
-8.84060502e-01 8.99609268e-01 5.21832526e-01 7.27557600e-01
7.33768106e-01 7.04610527e-01 5.54324508e-01 2.32487097e-01
6.29875064e-01 -1.10313404e+00 9.15311128e-02 2.88807064e-01
8.06077242e-01 -8.44773710e-01 -4.26648647e-01 -7.43941724e-01
2.23464206e-01 1.06258011e+00 9.74540710e-02 -4.85954106e-01
1.28573644e+00 -4.02631015e-01 -4.79436040e-01 -6.06148541e-01
-1.57210529e-01 -9.05687585e-02 4.35816526e-01 2.02137232e-01
7.10796565e-02 2.76031256e-01 -3.02029192e-01 3.77393551e-02
-6.25483632e-01 -5.51300764e-01 4.98094857e-01 5.40395319e-01
-7.38319218e-01 -9.08896983e-01 -5.07880151e-01 4.44714129e-01
-1.81798741e-01 1.93132177e-01 -6.25449359e-01 6.38773859e-01
2.59097397e-01 8.52394521e-01 7.46841356e-02 -5.38214684e-01
1.94111541e-01 1.10921837e-01 5.13117433e-01 -2.74451315e-01
4.54191506e-01 -1.14163466e-01 -5.65944672e-01 -7.01512396e-01
-3.25054079e-01 -8.67752731e-01 -1.27425241e+00 -6.62950218e-01
-5.44883013e-02 4.99410957e-01 5.97977459e-01 6.47060513e-01
4.38477069e-01 -3.41483802e-01 9.35911477e-01 -6.85974479e-01
-5.85290372e-01 -4.97118741e-01 -1.06706786e+00 6.02366865e-01
4.21365708e-01 -7.84465671e-01 -8.66462827e-01 -7.34668449e-02] | [7.324032783508301, 4.456172466278076] |
0c89e09c-224d-423f-a74b-3e134b8b59e5 | segvitv2-exploring-efficient-and-continual | 2306.06289 | null | https://arxiv.org/abs/2306.06289v1 | https://arxiv.org/pdf/2306.06289v1.pdf | SegViTv2: Exploring Efficient and Continual Semantic Segmentation with Plain Vision Transformers | We explore the capability of plain Vision Transformers (ViTs) for semantic segmentation using the encoder-decoder framework and introduce SegViTv2. In our work, we implement the decoder with the global attention mechanism inherent in ViT backbones and propose the lightweight Attention-to-Mask module that effectively converts the global attention map into semantic masks for high-quality segmentation results. Our decoder can outperform the most commonly-used decoder UpperNet in various ViT backbones while consuming only about 5\% of the computational cost. For the encoder, we address the concern of the relatively high computational cost in the ViT-based encoders and propose a Shrunk++ structure that incorporates edge-aware query-based down-sampling (EQD) and query-based up-sampling (QU) modules. The Shrunk++ structure reduces the computational cost of the encoder by up to $50\%$ while maintaining competitive performance. Furthermore, due to the flexibility of our ViT-based architecture, SegVit can be easily extended to semantic segmentation under the setting of continual learning, achieving nearly zero forgetting. Experiments show that our proposed SegViT outperforms recent segmentation methods on three popular benchmarks including ADE20k, COCO-Stuff-10k and PASCAL-Context datasets. The code is available through the following link: \url{https://github.com/zbwxp/SegVit}. | ['Yifan Liu', 'Chunhua Shen', 'Zhi Tian', 'Minh Hieu Phan', 'Liyang Liu', 'BoWen Zhang'] | 2023-06-09 | null | null | null | null | ['continual-semantic-segmentation'] | ['computer-vision'] | [ 1.62301928e-01 4.10660535e-01 -2.33816728e-02 -3.82324249e-01
-9.47622716e-01 -2.92570710e-01 1.45218909e-01 -1.99394777e-01
-7.25046396e-01 4.01665509e-01 -7.60306641e-02 -4.25871313e-01
5.25614023e-01 -8.93197477e-01 -1.05624437e+00 -3.69718134e-01
4.90548968e-01 2.93649167e-01 8.66805911e-01 -1.96452543e-01
-4.25861403e-03 -7.76579827e-02 -1.31004179e+00 8.15678611e-02
1.06133258e+00 1.15049922e+00 7.55479097e-01 6.61463499e-01
-2.12869748e-01 6.49017572e-01 -5.23679912e-01 -8.74661982e-01
3.08944821e-01 -3.50671113e-01 -9.94387507e-01 -1.04700230e-01
6.39678001e-01 -5.44402361e-01 -4.85472947e-01 1.02342808e+00
5.59543312e-01 1.75065566e-02 2.27873489e-01 -1.03817356e+00
-7.84809113e-01 6.37155235e-01 -7.55967438e-01 3.20212603e-01
-1.61301434e-01 3.06398988e-01 1.13702476e+00 -8.92764091e-01
5.95123947e-01 1.12681711e+00 7.74718702e-01 5.23144066e-01
-1.06478572e+00 -5.09561121e-01 4.18305427e-01 5.18396020e-01
-1.50547922e+00 -3.52805227e-01 4.33658123e-01 -1.00605153e-01
1.10005999e+00 5.99115305e-02 6.61048651e-01 8.01595151e-01
4.94681783e-02 1.22011161e+00 6.93234205e-01 -1.58493012e-01
1.97640136e-01 -1.95203006e-01 3.09309989e-01 9.77729142e-01
4.50096950e-02 -3.49736333e-01 -3.85611117e-01 4.36404169e-01
8.56146097e-01 -2.03150913e-01 -3.17259192e-01 -3.03162308e-03
-8.28101575e-01 7.50945091e-01 8.65973711e-01 -4.30868827e-02
-2.48505384e-01 6.99291945e-01 4.86180037e-01 9.12070125e-02
4.48510379e-01 1.43220976e-01 -6.17168665e-01 -2.04612672e-01
-1.04059815e+00 5.15026040e-03 5.34671783e-01 1.28532302e+00
9.82533574e-01 1.15311399e-01 -3.46782327e-01 1.03496325e+00
1.79349110e-01 3.33806038e-01 3.39030951e-01 -1.16086113e+00
4.02584821e-01 5.43761730e-01 -3.14271241e-01 -4.20446128e-01
-4.61437963e-02 -5.92918038e-01 -5.86674511e-01 -8.59418809e-02
2.23225251e-01 -1.08546853e-01 -1.41581702e+00 1.61252165e+00
3.42061996e-01 4.19729978e-01 -2.41076022e-01 8.74248981e-01
9.83015060e-01 6.23441339e-01 1.27316281e-01 2.29089946e-01
1.53303421e+00 -1.53238928e+00 -5.32332599e-01 -4.83799398e-01
5.26305020e-01 -6.11835897e-01 1.30003810e+00 -2.04292331e-02
-1.29814279e+00 -5.63820004e-01 -1.09176612e+00 -6.91772044e-01
-2.56687373e-01 1.09073864e-02 5.35576642e-01 4.79293704e-01
-1.43482637e+00 4.26868975e-01 -1.05351985e+00 -3.10391963e-01
7.45987773e-01 4.04390752e-01 6.58006892e-02 -1.98939562e-01
-9.53136027e-01 5.42443514e-01 3.65794241e-01 1.36839710e-02
-9.86505747e-01 -8.36759329e-01 -1.00677025e+00 2.01315328e-01
6.44260645e-01 -7.70843923e-01 1.63987601e+00 -8.42109978e-01
-1.41351724e+00 7.19339788e-01 -4.30718154e-01 -6.40393794e-01
5.45416832e-01 -2.42967024e-01 1.05157651e-01 3.52604300e-01
2.83039242e-01 1.28218377e+00 7.01509833e-01 -9.64840770e-01
-6.20950460e-01 -1.14178538e-01 2.39342049e-01 2.82915831e-01
-1.80661127e-01 -2.29597196e-01 -1.35956025e+00 -7.01496840e-01
8.60756487e-02 -7.81484723e-01 -2.47747630e-01 5.52843884e-02
-4.78531361e-01 -1.20503664e-01 1.04349101e+00 -7.17224777e-01
1.16634226e+00 -2.18943477e+00 -8.29264820e-02 -3.79211217e-01
1.79730341e-01 5.06912291e-01 -2.34179974e-01 8.37056190e-02
3.73032302e-01 7.59098008e-02 -6.95232630e-01 -8.22030306e-01
-7.71290511e-02 4.24579710e-01 2.99262330e-02 7.01101720e-02
2.34969988e-01 1.50179160e+00 -8.27868164e-01 -6.01401746e-01
3.33410382e-01 4.85892504e-01 -7.80774593e-01 4.85073738e-02
-4.71422732e-01 -1.06131015e-02 -2.83038944e-01 8.11331332e-01
7.25303352e-01 -4.35890257e-01 -6.65291622e-02 -1.53043538e-01
-4.53848392e-02 4.40261632e-01 -8.24028134e-01 2.00343227e+00
-4.48496282e-01 6.70147479e-01 2.13154107e-01 -8.60068083e-01
5.20738363e-01 9.57294405e-02 4.51861508e-02 -1.07117951e+00
2.40448371e-01 2.22251102e-01 -2.81623393e-01 -1.58423305e-01
7.73822367e-01 1.87935501e-01 6.23058267e-02 9.86459479e-02
3.73873293e-01 -1.07932657e-01 3.23164046e-01 3.60420227e-01
1.03227770e+00 1.91610739e-01 -1.04657382e-01 -2.65082717e-01
2.98902959e-01 3.07645928e-02 7.52701998e-01 7.04178393e-01
-2.68550187e-01 7.86123455e-01 5.20469427e-01 -1.51797816e-01
-9.63392913e-01 -1.00212622e+00 -6.45073652e-02 1.19773936e+00
4.71257955e-01 -4.78410512e-01 -1.15587425e+00 -7.61429667e-01
-1.50370196e-01 7.76294291e-01 -5.06348848e-01 -1.19227571e-02
-5.76315403e-01 -7.18472898e-01 7.02644646e-01 8.57589900e-01
9.84461665e-01 -1.06474733e+00 -6.38366044e-01 3.55956703e-01
-3.10489148e-01 -1.28140974e+00 -8.74548197e-01 2.07692519e-01
-8.27619314e-01 -8.12705636e-01 -1.00431585e+00 -9.58314061e-01
5.12691915e-01 2.88950562e-01 1.12935972e+00 1.30739138e-01
-3.05270463e-01 1.30851209e-01 -2.99667716e-01 -3.74155074e-01
-4.27055284e-02 4.12947178e-01 -7.33787894e-01 -2.19526902e-01
1.56997740e-01 -4.38928127e-01 -9.60464895e-01 3.02156627e-01
-8.89051020e-01 4.70931113e-01 5.40584743e-01 7.27028131e-01
9.44179535e-01 -3.26670378e-01 6.45208597e-01 -9.79122043e-01
1.06655188e-01 -2.87209928e-01 -6.41603410e-01 1.13746487e-01
-6.95304036e-01 -4.02107574e-02 3.34759831e-01 -4.13315222e-02
-1.01225030e+00 2.81091463e-02 -6.19985700e-01 -4.60457802e-01
1.52575314e-01 1.79256961e-01 -2.20175400e-01 -3.25825475e-02
1.91342413e-01 4.49601442e-01 -1.94225967e-01 -7.05469370e-01
5.29664218e-01 5.04026175e-01 7.20520735e-01 -4.21194673e-01
3.62701267e-01 5.08598208e-01 -4.83934730e-01 -6.35786355e-01
-8.66450429e-01 -3.94892871e-01 -3.51840854e-01 -2.83251181e-02
1.14566505e+00 -1.18550396e+00 -6.00444913e-01 7.19456434e-01
-1.05685294e+00 -8.10356379e-01 -4.64190632e-01 2.12381617e-03
-4.62809443e-01 3.74456316e-01 -1.04188716e+00 -3.38429034e-01
-7.50154138e-01 -1.56440556e+00 1.26831865e+00 3.29407454e-01
9.21035632e-02 -8.49701583e-01 -3.91527981e-01 6.53910756e-01
4.53116089e-01 2.31698677e-02 7.46401489e-01 -1.20118961e-01
-9.67805862e-01 2.19027832e-01 -6.79383814e-01 5.15644610e-01
-1.90171063e-01 -3.39648932e-01 -1.10346282e+00 -2.94941783e-01
-3.55897784e-01 -3.07826400e-01 1.33869231e+00 5.65527380e-01
1.37737083e+00 -7.73114488e-02 -2.13151768e-01 1.11393332e+00
1.64922583e+00 2.19866678e-01 8.06567490e-01 2.84555674e-01
1.06181502e+00 -2.53049992e-02 2.03671217e-01 8.23650062e-02
8.60081553e-01 5.49458206e-01 6.22465849e-01 -4.03325647e-01
-7.03432798e-01 -2.82109380e-01 2.88963735e-01 9.41418350e-01
1.67919502e-01 -3.85285944e-01 -8.20027590e-01 9.25652802e-01
-1.80141830e+00 -4.81374949e-01 -1.94452763e-01 1.83313811e+00
9.19070780e-01 3.19563717e-01 -1.89626321e-01 -1.68897077e-01
6.26299262e-01 2.60969430e-01 -7.80549049e-01 -6.85057938e-01
1.31600685e-02 5.83555758e-01 8.30100834e-01 6.47380233e-01
-1.01478469e+00 1.40508187e+00 5.68089676e+00 1.03246295e+00
-1.02608907e+00 4.30927187e-01 8.58059347e-01 -1.15889087e-01
-2.83638835e-01 -1.63769666e-02 -9.08537865e-01 4.88984793e-01
8.14044237e-01 -2.65244022e-02 2.11377382e-01 8.33489001e-01
5.59000671e-03 -2.83189118e-01 -7.96129882e-01 7.70809352e-01
-1.20883361e-01 -1.46700668e+00 1.43611021e-02 5.09511773e-03
7.04059064e-01 7.03398585e-01 3.84305865e-02 4.09786910e-01
2.15736642e-01 -7.61704504e-01 1.03565478e+00 -7.23067578e-03
1.14710069e+00 -5.66579401e-01 6.37438357e-01 8.44764337e-02
-1.37174749e+00 1.45268723e-01 -3.59590203e-01 1.58201873e-01
3.73198032e-01 5.38928688e-01 -7.92059898e-01 5.44727862e-01
9.08672631e-01 8.35294127e-01 -5.67022622e-01 1.08871233e+00
-3.68545949e-01 8.43190432e-01 -3.34621847e-01 3.40202928e-01
6.01029515e-01 -5.35103343e-02 3.51787925e-01 1.31569314e+00
1.64368689e-01 -1.51885021e-02 2.22177966e-03 8.82083654e-01
-4.81021285e-01 -1.54899761e-01 7.92097375e-02 2.50258207e-01
4.61770177e-01 1.05724549e+00 -1.02504337e+00 -5.50098658e-01
-5.17471552e-01 1.37997293e+00 3.80820423e-01 3.88506472e-01
-1.02924359e+00 -4.66967791e-01 6.78606629e-01 1.28539145e-01
9.16313171e-01 -6.89725578e-02 -4.90562648e-01 -9.94919121e-01
8.42225030e-02 -6.08866930e-01 3.25747609e-01 -8.45441997e-01
-7.36265898e-01 6.82351470e-01 -1.82484686e-01 -5.56307375e-01
2.19088256e-01 -2.13846266e-01 -5.23730516e-01 8.92227590e-01
-1.93052399e+00 -1.13877571e+00 -3.49664539e-01 5.40222406e-01
1.02605748e+00 3.09119582e-01 4.22506601e-01 5.32036543e-01
-8.16530824e-01 8.26272428e-01 -1.51358414e-02 2.77025342e-01
4.42194730e-01 -1.43239248e+00 1.06875563e+00 8.93357515e-01
6.32395968e-02 2.72416443e-01 2.90348947e-01 -4.98525023e-01
-1.10959291e+00 -1.41364431e+00 6.72138929e-01 -1.42465219e-01
3.02003860e-01 -5.05325973e-01 -9.54659343e-01 7.90813327e-01
4.12343621e-01 1.01882264e-01 1.84216052e-01 -2.01235712e-01
-3.37809920e-01 -1.12648807e-01 -1.00877488e+00 5.93522906e-01
1.17771149e+00 -3.75943542e-01 -4.00210023e-01 1.23481460e-01
1.26020646e+00 -6.89709425e-01 -6.24605298e-01 2.28929445e-01
3.78756732e-01 -9.29093778e-01 9.24845576e-01 1.97310477e-01
4.46802557e-01 -3.45420003e-01 -2.29876563e-02 -1.05853105e+00
-2.73698658e-01 -5.22694111e-01 -6.86587915e-02 1.17105341e+00
4.89427894e-01 -6.30212069e-01 9.19879079e-01 2.68607020e-01
-6.53336227e-01 -1.10194218e+00 -1.04564905e+00 -5.99238276e-01
1.92645937e-02 -6.69587851e-01 5.85892081e-01 5.29814661e-01
-5.58635294e-01 2.38201931e-01 -3.03616002e-02 1.33098289e-01
5.97739935e-01 -1.15520634e-01 4.13149834e-01 -6.59039497e-01
-5.23371756e-01 -4.07499015e-01 -2.83936799e-01 -1.48137498e+00
-1.76124796e-01 -9.71269369e-01 9.47783887e-02 -1.93685067e+00
1.65590838e-01 -3.55275989e-01 -2.62837678e-01 7.99511671e-01
-3.60574841e-01 7.25745499e-01 4.03849393e-01 3.98047157e-02
-8.07149410e-01 6.86390221e-01 1.51878667e+00 -1.47087827e-01
-7.53999054e-02 -3.30409735e-01 -8.37893724e-01 6.69053078e-01
6.99981987e-01 -3.66093457e-01 -5.93082666e-01 -1.06624353e+00
-7.88005218e-02 -1.95025921e-01 4.19968456e-01 -1.05228424e+00
1.64141521e-01 3.10812891e-01 1.07377283e-02 -5.79117656e-01
4.65904117e-01 -4.13341075e-01 -1.25966132e-01 4.59638923e-01
-7.59885013e-02 8.72413963e-02 3.90908360e-01 5.04063010e-01
-1.92802206e-01 -6.58725649e-02 8.94396782e-01 -3.04784656e-01
-1.11361742e+00 4.96282995e-01 -1.46196470e-01 3.93571466e-01
9.53033566e-01 -3.72417927e-01 -4.26185220e-01 -1.24028921e-01
-5.46192527e-01 6.18906975e-01 5.22085786e-01 2.82291472e-01
5.94915628e-01 -8.89670789e-01 -4.49486941e-01 4.65278514e-02
8.79677609e-02 4.96941149e-01 4.51130807e-01 9.01842356e-01
-8.40936899e-01 3.05390418e-01 6.52935132e-02 -6.28635108e-01
-1.02258801e+00 1.82468757e-01 4.46896553e-01 -7.68892169e-02
-9.56989050e-01 1.42779791e+00 5.54016411e-01 -9.70549658e-02
2.46772870e-01 -5.49348116e-01 2.15442479e-01 -1.25299484e-01
3.76170039e-01 3.06314260e-01 1.87903881e-01 -4.95692879e-01
-3.43288958e-01 3.50586236e-01 -4.11729932e-01 -1.28289843e-02
1.27965903e+00 -1.94193974e-01 6.02060556e-02 2.51153056e-02
1.19066691e+00 -4.80767578e-01 -1.73633480e+00 -3.02594602e-01
-1.52020201e-01 -1.59565076e-01 3.18530738e-01 -8.63369763e-01
-1.49724579e+00 9.17305410e-01 4.48048055e-01 -2.03547895e-01
1.30551231e+00 1.12252556e-01 1.43584657e+00 1.70185566e-02
2.75975376e-01 -1.11300707e+00 -1.93931416e-01 5.95017016e-01
5.12860954e-01 -1.04059017e+00 -2.57931262e-01 -6.58880115e-01
-6.91179395e-01 4.76594120e-01 7.60622680e-01 -1.53740317e-01
4.98954982e-01 4.50008065e-01 5.77525683e-02 -1.29676878e-01
-7.29576886e-01 -5.00032187e-01 1.22061238e-01 3.98025721e-01
3.82481724e-01 4.89518456e-02 -3.88387978e-01 4.64632928e-01
-1.97758630e-01 8.09540525e-02 5.45509160e-01 8.00852537e-01
-4.04801726e-01 -1.05577993e+00 5.38332909e-02 4.89139020e-01
-5.83646357e-01 -5.09006798e-01 -2.46258155e-02 5.96573472e-01
3.77055556e-01 7.59343445e-01 2.26533622e-01 -1.58885270e-01
4.04824793e-01 6.37083352e-02 2.32471496e-01 -6.23956919e-01
-6.38278544e-01 1.62856936e-01 2.31059100e-02 -8.89561236e-01
-1.88645437e-01 -4.46817517e-01 -1.41648340e+00 -2.37559527e-01
-2.38373443e-01 -1.58826172e-01 5.48737228e-01 8.22002769e-01
6.20155692e-01 9.42122579e-01 -1.25584211e-02 -6.47731245e-01
-1.50129065e-01 -7.88152814e-01 -3.74247432e-01 9.71933454e-02
2.85483778e-01 -4.01279539e-01 1.28225192e-01 1.46357194e-01] | [9.522019386291504, 0.15129072964191437] |
2c7bd6cd-3900-48d2-a947-66373b0eaefa | co-learning-meets-stitch-up-for-noisy-multi | 2307.00880 | null | https://arxiv.org/abs/2307.00880v1 | https://arxiv.org/pdf/2307.00880v1.pdf | Co-Learning Meets Stitch-Up for Noisy Multi-label Visual Recognition | In real-world scenarios, collected and annotated data often exhibit the characteristics of multiple classes and long-tailed distribution. Additionally, label noise is inevitable in large-scale annotations and hinders the applications of learning-based models. Although many deep learning based methods have been proposed for handling long-tailed multi-label recognition or label noise respectively, learning with noisy labels in long-tailed multi-label visual data has not been well-studied because of the complexity of long-tailed distribution entangled with multi-label correlation. To tackle such a critical yet thorny problem, this paper focuses on reducing noise based on some inherent properties of multi-label classification and long-tailed learning under noisy cases. In detail, we propose a Stitch-Up augmentation to synthesize a cleaner sample, which directly reduces multi-label noise by stitching up multiple noisy training samples. Equipped with Stitch-Up, a Heterogeneous Co-Learning framework is further designed to leverage the inconsistency between long-tailed and balanced distributions, yielding cleaner labels for more robust representation learning with noisy long-tailed data. To validate our method, we build two challenging benchmarks, named VOC-MLT-Noise and COCO-MLT-Noise, respectively. Extensive experiments are conducted to demonstrate the effectiveness of our proposed method. Compared to a variety of baselines, our method achieves superior results. | ['Yi Yang', 'Linchao Zhu', 'Zongxin Yang', 'Chao Liang'] | 2023-07-03 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [ 4.18017685e-01 -5.64127922e-01 -1.23179041e-01 -6.41255200e-01
-1.33338547e+00 -6.14257812e-01 3.75367194e-01 1.10253491e-01
-2.70568848e-01 6.08092189e-01 9.45787430e-02 1.42198965e-01
-2.14362238e-02 -2.21526951e-01 -6.09502435e-01 -1.11669517e+00
6.45016909e-01 3.03973645e-01 -2.34900311e-01 2.48071238e-01
-2.23106280e-01 5.73448837e-02 -1.58285594e+00 3.88915449e-01
6.93817973e-01 1.18171167e+00 8.83312672e-02 5.09416223e-01
-1.34565011e-01 7.91261673e-01 -7.42726803e-01 -4.70870584e-01
2.39667609e-01 -4.60214704e-01 -3.36142004e-01 3.71865690e-01
8.94355297e-01 7.35968724e-02 3.08917020e-03 1.53491318e+00
7.68759549e-01 9.52936709e-02 8.66371989e-01 -1.36173797e+00
-7.27898121e-01 4.07790065e-01 -1.01190627e+00 -1.24430656e-01
-8.67236927e-02 1.35103241e-01 1.07183015e+00 -9.82795775e-01
4.24968421e-01 1.52533638e+00 1.02561355e+00 5.48973024e-01
-1.31040335e+00 -1.11486983e+00 2.49982119e-01 1.02965079e-01
-1.35654151e+00 -1.57668740e-01 7.42928088e-01 -3.87216926e-01
1.58257499e-01 1.55435428e-01 1.82887718e-01 1.65623128e+00
2.74220016e-02 9.66926038e-01 1.66416097e+00 -2.63536304e-01
-3.58983167e-02 -1.40717756e-02 2.46441916e-01 5.83633482e-01
3.77125829e-01 4.84434702e-02 -3.95306736e-01 -1.83148995e-01
1.92914665e-01 2.71516293e-01 -1.02314882e-01 -1.47698000e-01
-1.18655407e+00 6.19121850e-01 2.96683073e-01 6.61826804e-02
2.19799392e-03 2.07653120e-01 8.28593612e-01 1.69871673e-01
7.06722438e-01 -2.87483074e-02 -5.24318635e-01 9.00981128e-02
-9.85399008e-01 2.16497466e-01 3.77801836e-01 1.09166586e+00
7.67220318e-01 4.42314148e-02 -6.21246696e-01 1.26440501e+00
3.72390628e-01 8.26047421e-01 4.66815501e-01 -6.20793760e-01
4.87297148e-01 3.15336734e-01 -1.04098000e-01 -9.75725234e-01
-5.60145140e-01 -8.63172054e-01 -1.40773976e+00 -1.51886251e-02
4.29780126e-01 -1.56659976e-01 -1.09141958e+00 1.87247443e+00
4.33030188e-01 3.90722752e-01 1.80183891e-02 8.10825825e-01
8.81895602e-01 5.13639033e-01 2.98213810e-01 -3.44993025e-01
1.43801570e+00 -1.36312425e+00 -1.00270557e+00 -1.45016029e-01
7.19775498e-01 -9.56472635e-01 1.19952321e+00 4.05341387e-01
-3.95797104e-01 -6.54605389e-01 -8.17056417e-01 -1.25133902e-01
-2.28953019e-01 1.95507511e-01 1.88968599e-01 6.16018116e-01
-3.62711370e-01 2.31818646e-01 -4.18242663e-01 -2.26259269e-02
8.12586725e-01 -1.26418754e-01 -4.20182019e-01 -6.41983390e-01
-9.01988328e-01 5.78954816e-01 3.46965671e-01 2.36671790e-01
-1.00236666e+00 -6.18676722e-01 -8.84796917e-01 -1.15226381e-01
7.41240323e-01 -5.42136908e-01 1.29382193e+00 -7.54358828e-01
-9.90330338e-01 8.61896873e-01 7.55018508e-03 5.62084727e-02
5.69041908e-01 -1.48979694e-01 -5.69859147e-01 -3.46299350e-01
4.12851423e-01 4.29328203e-01 9.33799684e-01 -1.62354660e+00
-7.52653837e-01 -3.95226955e-01 -4.52721000e-01 1.03140302e-01
-3.81680787e-01 -1.64460137e-01 -2.03741163e-01 -1.08766472e+00
1.02398396e-01 -1.12797344e+00 -2.58068107e-02 -1.91132396e-01
-7.12209940e-01 -2.68507302e-01 9.38555777e-01 -2.11125478e-01
1.12986100e+00 -2.24570966e+00 -1.52754202e-01 -1.36854336e-01
2.79660046e-01 3.42588991e-01 -5.73343933e-01 2.24168450e-01
-1.30188525e-01 2.48985395e-01 -2.11352721e-01 -9.88044500e-01
1.71104461e-01 5.19753218e-01 -2.73912936e-01 6.64393246e-01
1.02008350e-01 8.74712169e-01 -1.17845249e+00 -6.14810467e-01
1.95825808e-02 3.80276352e-01 -6.64950460e-02 3.21761161e-01
-2.04547361e-01 7.22568989e-01 -2.80005217e-01 1.05290067e+00
9.75553274e-01 -3.63404036e-01 -3.02235615e-02 -5.31417787e-01
2.64997631e-01 -4.87203270e-01 -1.29476666e+00 1.62480474e+00
-4.81000990e-01 2.19446048e-01 2.00802773e-01 -7.19039142e-01
8.42395723e-01 2.55767405e-01 2.91650862e-01 -6.45401001e-01
3.66199613e-01 4.50038135e-01 -3.64471853e-01 -6.37419760e-01
3.71727973e-01 -6.65072799e-01 -2.22197711e-01 3.80095333e-01
1.60780132e-01 -1.94874376e-01 2.21197426e-01 -2.91718468e-02
8.95813525e-01 3.11441794e-02 1.97792858e-01 3.97537835e-03
3.40713203e-01 -5.63554168e-01 9.93718386e-01 7.51165688e-01
-5.16127527e-01 9.59744513e-01 3.49915147e-01 -2.07976878e-01
-8.89269114e-01 -6.06278718e-01 -1.28347054e-01 1.28493571e+00
2.38058761e-01 -3.31734836e-01 -5.53958595e-01 -1.17608511e+00
2.76522413e-02 3.46850812e-01 -7.14348197e-01 -7.01491907e-02
-2.02160642e-01 -1.08017623e+00 7.86108911e-01 3.42888236e-01
4.59414393e-01 -8.22542667e-01 3.13387327e-02 5.10173757e-03
-3.18162978e-01 -1.35465384e+00 -8.26865971e-01 5.61854959e-01
-4.40955371e-01 -1.19638383e+00 -8.13539684e-01 -8.28139007e-01
5.65244853e-01 5.70007026e-01 1.13686728e+00 2.68155206e-02
-2.83323854e-01 5.99138774e-02 -4.60580170e-01 -3.91326070e-01
-3.89367670e-01 -2.11396348e-02 -1.33053422e-01 4.59523261e-01
2.05914766e-01 -4.26609397e-01 -4.03520346e-01 4.90417153e-01
-1.38075054e+00 -1.69612080e-01 6.81312561e-01 1.38726676e+00
1.02910125e+00 3.48229766e-01 7.18304753e-01 -1.20304489e+00
5.33366203e-01 -7.54484057e-01 -3.95805150e-01 3.84835511e-01
-6.55462623e-01 -1.49460873e-02 7.30170310e-01 -7.20421851e-01
-9.97261226e-01 1.29630208e-01 -1.69205204e-01 -6.74957037e-01
-3.95145446e-01 3.27872515e-01 -3.91868114e-01 9.68428478e-02
4.10957903e-01 1.70931090e-02 -2.65441805e-01 -6.96866691e-01
5.40484965e-01 8.65242541e-01 6.15558803e-01 -6.84180200e-01
8.18559110e-01 4.72377300e-01 1.77907199e-01 -3.06589961e-01
-1.72575903e+00 -7.82714903e-01 -3.61732990e-01 -1.18703209e-01
7.67510414e-01 -1.11750770e+00 -5.18219769e-01 1.02681851e+00
-1.02296507e+00 -9.39767435e-02 -2.21619204e-01 9.48886648e-02
-2.71872103e-01 5.31568766e-01 -6.10137343e-01 -7.99682379e-01
-3.59644204e-01 -1.38517559e+00 1.50670469e+00 3.29934061e-01
2.18733490e-01 -9.30133700e-01 2.73928523e-01 6.55213773e-01
2.95181662e-01 4.11826849e-01 8.60233128e-01 -8.52472663e-01
-1.55733883e-01 -2.43120909e-01 -5.31521440e-01 7.30852962e-01
1.61754414e-01 -2.98645884e-01 -1.25537694e+00 -5.23106277e-01
6.76944256e-02 -1.02503562e+00 9.42139089e-01 3.47598977e-02
1.37177086e+00 -3.56396362e-02 -1.37735367e-01 6.73689127e-01
1.50769985e+00 -3.43159497e-01 1.27325103e-01 1.75355375e-01
1.17707205e+00 4.57053244e-01 7.88111985e-01 5.35179615e-01
5.52203000e-01 5.62414467e-01 7.54741073e-01 -4.32375185e-02
-5.03103375e-01 -1.84021115e-01 2.34650299e-02 1.04480839e+00
5.83846331e-01 -4.59305465e-01 -6.43575370e-01 4.52055603e-01
-1.87304556e+00 -6.64791584e-01 -3.28051805e-01 1.78239083e+00
1.00201201e+00 -1.41630411e-01 -9.29019973e-02 1.91990390e-01
9.73578632e-01 3.45860094e-01 -7.55435526e-01 2.41695046e-01
-5.06735802e-01 -1.59261495e-01 5.03210366e-01 1.23580694e-01
-1.47592843e+00 6.45331085e-01 5.19740772e+00 1.50526655e+00
-1.03085935e+00 4.85581785e-01 7.99650192e-01 -1.82753541e-02
-2.36327007e-01 -3.94677103e-01 -9.89418983e-01 7.34675288e-01
5.59070230e-01 5.16327739e-01 8.77422616e-02 8.19059014e-01
-8.54938701e-02 8.85710195e-02 -9.42099273e-01 1.38152111e+00
2.20058590e-01 -7.67316341e-01 -1.82203427e-01 -4.14059050e-02
1.28218853e+00 -2.51526497e-02 3.28273356e-01 5.69726229e-01
3.60012680e-01 -9.37808275e-01 9.72765565e-01 2.84517825e-01
1.00449502e+00 -6.87622130e-01 1.10625732e+00 4.40656275e-01
-1.20296872e+00 -2.97451347e-01 -2.91212529e-01 2.43985623e-01
1.22740306e-01 1.33902264e+00 -3.75720054e-01 6.84357405e-01
4.86735821e-01 9.16637301e-01 -7.59271085e-01 1.01997066e+00
-2.96678782e-01 6.84935153e-01 -7.28250742e-02 2.81980574e-01
2.92742580e-01 -1.13241322e-01 2.55787045e-01 1.46702361e+00
2.49484688e-01 -2.60877550e-01 6.18091583e-01 5.16195238e-01
-4.59213406e-01 1.31168580e-02 -3.65077227e-01 7.39743337e-02
5.18365502e-01 1.72433209e+00 -7.23499596e-01 -3.72580230e-01
-4.99061137e-01 8.32286596e-01 3.75554174e-01 3.25169921e-01
-9.64217544e-01 -9.66827646e-02 4.97593671e-01 -3.90742987e-01
2.31954560e-01 1.67186949e-02 -3.52149695e-01 -1.32223046e+00
8.29117820e-02 -1.24858749e+00 4.57392335e-01 -5.59258640e-01
-2.04165554e+00 6.00620747e-01 -3.37531477e-01 -1.19358325e+00
2.58721679e-01 -4.14417267e-01 -1.56295702e-01 6.18976474e-01
-1.87860942e+00 -1.65444338e+00 -6.02592170e-01 5.48831940e-01
7.11642504e-01 -5.68059944e-02 7.44958520e-01 7.89716184e-01
-8.54143202e-01 8.74594152e-01 4.61082697e-01 -8.17645863e-02
1.23207366e+00 -1.35775185e+00 -3.36883403e-02 6.78100348e-01
2.90541947e-01 1.37847170e-01 4.34174240e-01 -5.87724984e-01
-1.04569137e+00 -1.62610614e+00 3.49080652e-01 -2.84397662e-01
6.27171755e-01 -4.53989893e-01 -1.01038527e+00 4.42462653e-01
2.54717702e-03 6.43040955e-01 8.02298188e-01 -5.31055667e-02
-9.35835123e-01 -3.42775047e-01 -1.01230121e+00 2.32151613e-01
8.85041356e-01 -6.83612764e-01 -1.31738737e-01 5.87782025e-01
6.55691564e-01 -4.21307355e-01 -6.61435425e-01 6.90974474e-01
4.14090276e-01 -8.04353595e-01 7.89547980e-01 -3.20722193e-01
3.09106886e-01 -3.94020855e-01 -4.74018514e-01 -1.57683277e+00
-2.86163300e-01 -3.69255245e-01 7.61098862e-02 1.65712893e+00
9.49033350e-02 -2.66340077e-01 6.30622387e-01 7.90311322e-02
-2.93925673e-01 -8.82454515e-01 -8.21030378e-01 -7.95299351e-01
5.96776009e-02 -3.30913544e-01 4.44295555e-01 1.10024631e+00
-8.14235866e-01 4.22802478e-01 -8.79287660e-01 1.52639911e-01
8.71540606e-01 8.92535150e-02 7.42817998e-01 -1.14210653e+00
-1.21458381e-01 -1.97800085e-01 -8.22207928e-02 -8.59304011e-01
4.15562361e-01 -9.47365880e-01 5.48023105e-01 -1.21586704e+00
6.26216352e-01 -7.87100017e-01 -5.51840246e-01 5.60667038e-01
-6.02553427e-01 7.51164436e-01 2.44083926e-01 2.50441849e-01
-1.11305547e+00 7.49793470e-01 1.38877273e+00 -3.50905627e-01
4.34631079e-01 -1.36507109e-01 -6.87090814e-01 7.90970802e-01
5.12881994e-01 -9.50937986e-01 -3.62284720e-01 -4.62889969e-01
2.26878628e-01 -3.10416996e-01 2.11388201e-01 -9.85157013e-01
1.17404066e-01 -7.01200888e-02 1.18492611e-01 -6.29035950e-01
2.24197492e-01 -1.06189227e+00 8.13773945e-02 1.13691069e-01
-4.56222951e-01 -6.21851720e-02 -1.26039714e-01 1.04176283e+00
-4.16376531e-01 -2.85784900e-01 1.06813145e+00 -6.06772155e-02
-4.70464766e-01 4.10167664e-01 4.76736352e-02 4.83260155e-01
9.57607150e-01 2.51022607e-01 -4.91205037e-01 -1.38091609e-01
-4.82773304e-01 3.05469126e-01 3.33320558e-01 3.48428041e-01
3.70104909e-01 -1.60307336e+00 -7.00695753e-01 1.19506463e-01
4.10005927e-01 2.99452275e-01 6.58644855e-01 7.36690283e-01
-7.89050013e-03 5.21247089e-03 1.06969737e-01 -6.91450000e-01
-1.40266705e+00 6.35842323e-01 1.78703204e-01 -6.34929538e-01
-2.43004307e-01 1.02454937e+00 2.23801911e-01 -7.87895679e-01
4.99956369e-01 -2.92100817e-01 1.96118941e-04 5.00067413e-01
4.78187114e-01 3.59721065e-01 1.30161166e-01 -8.01212788e-01
-1.44190043e-01 8.91903400e-01 -8.22373182e-02 3.76339346e-01
1.22506750e+00 -4.37127620e-01 -1.30026072e-01 7.07082570e-01
1.54566848e+00 -7.25540593e-02 -1.32938790e+00 -6.40564799e-01
-1.80536341e-02 -5.24743199e-01 -2.53963564e-02 -9.36104000e-01
-1.29599428e+00 9.21959817e-01 7.28284776e-01 1.01736031e-01
1.17161548e+00 -2.25230545e-01 1.12968290e+00 1.53734088e-01
1.69904888e-01 -1.09813762e+00 6.00475729e-01 4.51065540e-01
5.17507374e-01 -1.79035950e+00 -1.81825832e-01 -5.53717375e-01
-6.51100159e-01 7.47783899e-01 5.46350956e-01 2.24966273e-01
5.82457781e-01 3.36031467e-01 4.78095025e-01 -1.48470879e-01
-6.10979259e-01 -2.61852682e-01 2.16015086e-01 5.67215502e-01
4.30987179e-02 8.63721520e-02 1.79738291e-02 7.21289814e-01
2.59606600e-01 -1.20902024e-01 3.86604518e-01 7.57767498e-01
-2.84624487e-01 -1.16529429e+00 -5.96789896e-01 3.73853236e-01
-7.11680949e-01 -6.72202036e-02 1.16641663e-01 3.65771711e-01
7.64857769e-01 1.13595533e+00 -4.11369175e-01 -4.70543295e-01
3.82065892e-01 2.03414753e-01 1.61656842e-01 -7.18192518e-01
-5.75092614e-01 3.94223034e-01 -7.66802877e-02 -3.53499562e-01
-5.88422954e-01 -5.36692739e-01 -8.48734558e-01 -7.59146959e-02
-6.26619935e-01 -1.69108272e-01 7.04720497e-01 9.57557142e-01
2.15680346e-01 6.96391642e-01 6.91290617e-01 -7.28748858e-01
-9.50589597e-01 -1.17525458e+00 -9.48096156e-01 9.45408463e-01
5.43281913e-01 -9.02098775e-01 -5.43266535e-01 3.19141448e-02] | [9.43031120300293, 3.8495843410491943] |
affdb1b4-41e4-4e7c-a7fd-b320b76a4c23 | toward-fairness-through-fair-multi-exit | 2306.14518 | null | https://arxiv.org/abs/2306.14518v2 | https://arxiv.org/pdf/2306.14518v2.pdf | Toward Fairness Through Fair Multi-Exit Framework for Dermatological Disease Diagnosis | Fairness has become increasingly pivotal in medical image recognition. However, without mitigating bias, deploying unfair medical AI systems could harm the interests of underprivileged populations. In this paper, we observe that while features extracted from the deeper layers of neural networks generally offer higher accuracy, fairness conditions deteriorate as we extract features from deeper layers. This phenomenon motivates us to extend the concept of multi-exit frameworks. Unlike existing works mainly focusing on accuracy, our multi-exit framework is fairness-oriented; the internal classifiers are trained to be more accurate and fairer, with high extensibility to apply to most existing fairness-aware frameworks. During inference, any instance with high confidence from an internal classifier is allowed to exit early. Experimental results show that the proposed framework can improve the fairness condition over the state-of-the-art in two dermatological disease datasets. | ['Tsung-Yi Ho', 'Yiyu Shi', 'Yu-Jen Chen', 'Hao-Wei Chung', 'Ching-Hao Chiu'] | 2023-06-26 | null | null | null | null | ['fairness', 'fairness'] | ['computer-vision', 'miscellaneous'] | [ 2.08586916e-01 5.47734499e-01 -6.85628116e-01 -8.26948702e-01
-1.10659763e-01 2.24453956e-02 3.40229720e-01 2.58485764e-01
-8.87710989e-01 1.15833151e+00 -1.71973124e-01 -2.51105130e-01
-3.49482536e-01 -9.80238080e-01 -2.70325840e-01 -7.43522644e-01
2.19415985e-02 1.69186428e-01 -4.61071908e-01 -6.18042052e-03
3.12306613e-01 5.22706985e-01 -1.30599475e+00 2.54827738e-01
1.43119967e+00 1.19796252e+00 -7.88906813e-01 3.05977523e-01
1.35351673e-01 1.17255199e+00 -6.78237736e-01 -1.22490966e+00
3.56155604e-01 -1.53232217e-01 -9.02279615e-01 -5.63203633e-01
4.91291314e-01 -6.84142053e-01 -1.52462244e-01 1.28102386e+00
5.16227961e-01 -1.01406947e-01 5.25045156e-01 -1.47368169e+00
-9.12791550e-01 7.73947299e-01 -4.88256574e-01 2.19188601e-01
-2.87004679e-01 2.02530891e-01 1.04750621e+00 -1.61447227e-01
3.53590876e-01 1.25135648e+00 4.97527868e-01 1.02757382e+00
-8.61609995e-01 -9.23809290e-01 3.28359485e-01 2.24838257e-01
-1.00652719e+00 -5.64959347e-01 5.72768450e-01 -6.28403872e-02
3.64445776e-01 7.55415142e-01 6.17511809e-01 1.12723923e+00
3.62311125e-01 1.15750122e+00 1.38848460e+00 -1.24349765e-01
2.90663213e-01 1.84600651e-01 2.62955606e-01 5.81665397e-01
7.80080378e-01 5.34228563e-01 -3.52212489e-01 -4.90500242e-01
5.02618492e-01 2.66406059e-01 -6.28239587e-02 -3.93318338e-03
-6.62945092e-01 9.49852228e-01 8.41526508e-01 9.25290361e-02
-6.87239230e-01 1.81677058e-01 3.51821333e-01 1.34373710e-01
6.60160422e-01 5.85337162e-01 -2.18529001e-01 1.63986553e-02
-1.25175560e+00 3.58759791e-01 7.18200803e-01 3.26882482e-01
1.09673649e-01 -1.97216317e-01 -7.63162911e-01 5.10892570e-01
1.08886063e-01 2.91106790e-01 2.42037207e-01 -1.27793634e+00
-2.11931347e-05 8.10632229e-01 -1.80673301e-01 -1.13213909e+00
-2.76449531e-01 -9.05220032e-01 -1.24232137e+00 4.25042480e-01
5.40500581e-01 -8.51941407e-02 -7.15373635e-01 1.76269448e+00
3.47617149e-01 -1.65475830e-01 -1.10386327e-01 9.69549656e-01
7.03967869e-01 -3.79162058e-02 5.54608524e-01 -2.66520888e-01
1.29350483e+00 -9.03103590e-01 -8.91152859e-01 -1.19137922e-02
3.17363143e-01 -3.98135751e-01 7.11697757e-01 6.12287343e-01
-1.07571304e+00 -1.05085008e-01 -6.55054510e-01 -2.61063874e-02
-2.31371462e-01 2.20635775e-02 9.62900162e-01 1.12550068e+00
-6.65709615e-01 8.51056814e-01 -4.74459112e-01 -1.86548941e-03
1.51263189e+00 2.85488307e-01 -2.45221987e-01 -2.82170046e-02
-1.47531331e+00 9.87528265e-01 1.32165521e-01 1.51989341e-01
-6.87422693e-01 -9.79354680e-01 -5.51437080e-01 1.85403392e-01
3.31859708e-01 -1.00035715e+00 9.73712206e-01 -1.27818644e+00
-1.26629186e+00 1.12623250e+00 6.17768206e-02 -8.30788672e-01
1.16114831e+00 -4.97537434e-01 -2.00307101e-01 7.46287033e-02
-9.66150314e-02 6.61717772e-01 7.31252789e-01 -1.15688324e+00
-7.03866601e-01 -5.27091265e-01 2.72277862e-01 -5.63242137e-02
-4.97674823e-01 -5.23936115e-02 2.94502109e-01 -6.23611689e-01
-5.59440851e-01 -6.22342944e-01 -6.20571315e-01 5.54425299e-01
-7.15071082e-01 -1.07453056e-01 3.79298180e-01 -4.95273978e-01
1.47697735e+00 -1.89071655e+00 -3.58011782e-01 2.23539203e-01
8.09530020e-01 5.93114436e-01 1.46668628e-01 -3.73193234e-01
1.31818295e-01 4.16134685e-01 -2.25715771e-01 -6.30484521e-02
1.13792889e-01 2.39843607e-01 1.03885241e-01 5.70138752e-01
2.91664213e-01 8.94892573e-01 -7.84688950e-01 -7.90338397e-01
1.13258712e-01 3.32361788e-01 -8.82803142e-01 1.23915665e-01
2.94279277e-01 2.20946684e-01 -6.47027612e-01 8.57123017e-01
9.92111146e-01 -2.45121881e-01 1.35961343e-02 2.25753300e-02
3.43828350e-01 3.50681543e-02 -5.87970257e-01 1.06129420e+00
-3.11871558e-01 1.58259585e-01 -2.37197652e-02 -1.16822827e+00
7.56659865e-01 1.54698312e-01 3.30382317e-01 -7.14729488e-01
4.35933053e-01 1.87830552e-01 3.37995142e-01 -4.21580672e-01
5.08836389e-01 -3.30789000e-01 -9.05377865e-02 1.87026441e-01
-1.78005278e-01 4.61506307e-01 -1.96817115e-01 2.10898027e-01
6.48082256e-01 -3.67503226e-01 7.20800340e-01 -3.54018927e-01
6.13994598e-01 -2.35751763e-01 8.19985807e-01 1.07812810e+00
-1.09764946e+00 9.87732261e-02 7.53538847e-01 -7.02500343e-01
-6.82789385e-01 -9.40820336e-01 -5.25231242e-01 9.00210798e-01
5.79554867e-03 1.04628876e-01 -9.10725534e-01 -1.20865881e+00
2.74739087e-01 8.59326124e-01 -1.03158033e+00 -5.52290440e-01
-8.06419924e-02 -1.06322694e+00 1.02377927e+00 3.09502840e-01
6.10288620e-01 -8.01069736e-01 -1.11255252e+00 -7.88885504e-02
-1.52753592e-01 -8.55539560e-01 -1.22737937e-01 -1.87750041e-01
-9.50562716e-01 -1.16256487e+00 -9.25987482e-01 1.80436268e-01
7.74666786e-01 -4.78555053e-01 1.31701827e+00 4.32049692e-01
-5.45009196e-01 -2.29410753e-02 -6.70166686e-02 -8.89261723e-01
-4.28637505e-01 4.08017114e-02 -1.72051191e-01 1.04216032e-01
6.25567377e-01 -2.18253657e-01 -1.09322727e+00 -2.73062065e-02
-8.63106906e-01 -1.05998673e-01 5.88997304e-01 1.05999053e+00
2.42000878e-01 -2.04790995e-01 8.60192895e-01 -1.40600276e+00
7.79666960e-01 -4.22688097e-01 -1.62850916e-01 3.81896466e-01
-1.07487261e+00 -2.23874440e-03 8.20546865e-01 -1.95612743e-01
-1.22002256e+00 -4.34674948e-01 1.01306356e-01 -1.86542645e-01
-3.98441434e-01 7.18173664e-03 6.79165078e-03 -8.28580633e-02
7.06941426e-01 -3.28803450e-01 1.35381833e-01 -5.48723824e-02
3.34798574e-01 7.13581383e-01 2.13166699e-01 -6.01443529e-01
3.28171223e-01 4.31408793e-01 1.38811663e-01 -2.79811203e-01
-1.03193140e+00 7.08384588e-02 -5.98186478e-02 -4.39586729e-01
5.84546566e-01 -5.74601233e-01 -1.23063862e+00 4.19404119e-01
-8.37740779e-01 1.85891524e-01 -4.58183080e-01 3.91414434e-01
-1.32298753e-01 3.14992249e-01 -6.64603949e-01 -1.39071620e+00
-7.37801015e-01 -9.96120989e-01 6.76979065e-01 6.11382842e-01
-4.12780404e-01 -8.25699091e-01 -2.12219492e-01 6.18204534e-01
6.18739724e-01 3.10014993e-01 7.72748053e-01 -7.35136509e-01
-1.89392909e-01 -3.66726190e-01 -3.91665608e-01 4.21048194e-01
-3.62802558e-02 1.87866986e-01 -1.50357091e+00 7.49705806e-02
-7.80566111e-02 -4.69018817e-01 1.06421804e+00 6.89197540e-01
1.67145443e+00 -7.49602437e-01 -1.60867065e-01 5.32052159e-01
1.23697865e+00 -6.24187232e-04 7.77734816e-01 2.84634948e-01
3.62867713e-01 9.92519915e-01 6.49183631e-01 7.06675947e-01
4.40548748e-01 2.62643278e-01 7.58344233e-01 -5.24628103e-01
3.72197449e-01 8.65588933e-02 4.55680788e-02 -6.22553937e-02
-5.55971026e-01 -8.14585760e-02 -6.70354843e-01 3.80843848e-01
-1.90584695e+00 -9.57036197e-01 5.71373589e-02 2.35294390e+00
1.09105730e+00 3.10171619e-02 1.31984845e-01 1.32078558e-01
6.90186501e-01 5.24664894e-02 -7.47546613e-01 -7.86306500e-01
-7.50467852e-02 3.38913441e-01 6.33883119e-01 2.86860734e-01
-1.15112782e+00 7.59800971e-01 6.12286329e+00 1.06535053e+00
-9.54476178e-01 4.83694933e-02 1.64452434e+00 -4.64352727e-01
-5.52783906e-01 -3.13535392e-01 -3.01385760e-01 4.40410048e-01
7.08828807e-01 -3.09590727e-01 1.57084808e-01 8.62505138e-01
1.13039143e-01 -1.57645330e-01 -1.08129799e+00 6.84312940e-01
-1.81277350e-01 -1.10676229e+00 1.50304213e-01 2.31574833e-01
5.73267639e-01 -4.91520733e-01 4.21511352e-01 2.06582129e-01
5.03486216e-01 -1.58226299e+00 4.23452049e-01 8.86381686e-01
6.19111121e-01 -1.08733320e+00 1.12645423e+00 4.14109051e-01
-2.31668144e-01 -6.85851723e-02 -4.60691601e-01 -1.66890830e-01
-1.68349966e-01 1.32384408e+00 -4.10657167e-01 7.90269136e-01
6.90752625e-01 3.68453562e-01 -4.60407972e-01 9.52977717e-01
-1.59165889e-01 3.97551715e-01 2.25980263e-02 -1.79179534e-01
1.95151642e-01 1.96997121e-01 2.15470582e-01 9.66379941e-01
-8.48838761e-02 1.38356924e-01 -2.45594278e-01 1.06719780e+00
-2.81256258e-01 2.26636797e-01 -5.48339009e-01 -3.69259231e-02
2.10406899e-01 1.32369614e+00 -2.27047548e-01 -5.24576783e-01
1.28626660e-01 7.61223555e-01 3.53003204e-01 4.14355434e-02
-8.80929172e-01 -2.06788227e-01 7.82248855e-01 -1.96305647e-01
-4.97240692e-01 8.12723458e-01 -8.32796097e-01 -1.15092778e+00
-1.67612270e-01 -9.77471173e-01 7.12959230e-01 -1.90582067e-01
-1.69831038e+00 5.45860827e-01 -3.62661332e-01 -8.84263456e-01
3.93273756e-02 -5.86679041e-01 -3.89431268e-01 8.40344012e-01
-1.84904122e+00 -1.02559078e+00 -8.40488002e-02 5.18837214e-01
-1.31298214e-01 -1.65225595e-01 1.06180382e+00 3.41618538e-01
-5.89822471e-01 1.18616807e+00 -2.67303199e-01 2.09902734e-01
7.70245135e-01 -9.79505479e-01 -1.41250461e-01 8.07961822e-01
-2.76827127e-01 7.95788825e-01 4.08722579e-01 -5.10236740e-01
-5.47544837e-01 -9.29536462e-01 1.01209152e+00 -3.49945545e-01
6.71880916e-02 9.20354649e-02 -6.95078731e-01 1.72066212e-01
9.10779089e-02 2.34400675e-01 1.14815199e+00 2.82660633e-01
-4.42076594e-01 -2.31457114e-01 -1.85033941e+00 5.99267364e-01
9.23493624e-01 -2.85613835e-01 -3.39018911e-01 -1.49278969e-01
2.41130963e-01 -1.67256236e-01 -8.66060078e-01 6.13004506e-01
9.06419635e-01 -1.23354232e+00 1.02940297e+00 -1.13853371e+00
1.07200551e+00 3.06899816e-01 9.70804468e-02 -1.03058505e+00
-3.29198509e-01 -3.52948219e-01 -2.12436870e-01 9.15219665e-01
3.73814583e-01 -8.55490863e-01 9.66797054e-01 1.10693312e+00
5.41677654e-01 -1.00374770e+00 -1.15901887e+00 -5.13768375e-01
4.86467332e-01 -2.20234692e-01 8.31822097e-01 1.33209383e+00
1.53727857e-02 -2.68166661e-01 -6.60214245e-01 7.03078462e-03
1.08362651e+00 2.69247275e-02 4.25061345e-01 -1.43193960e+00
-6.08244464e-02 -8.29061627e-01 -3.54037136e-01 -3.02183688e-01
1.57373875e-01 -8.32600772e-01 -2.85506666e-01 -1.10082614e+00
6.55087829e-01 -5.75507998e-01 -7.34529138e-01 7.57803082e-01
-6.99199557e-01 2.69450724e-01 3.04106414e-01 -5.51848784e-02
-3.54671299e-01 5.10172844e-01 1.57432747e+00 -1.82440355e-01
3.40841740e-01 2.33755469e-01 -1.33876801e+00 6.99904978e-01
9.27138746e-01 -6.86968029e-01 -1.76127478e-01 -4.62865792e-02
-1.20876636e-02 -1.46325856e-01 5.87943017e-01 -5.76837301e-01
6.37540296e-02 -5.02112627e-01 6.57188237e-01 8.25195760e-02
-1.10777006e-01 -8.95979702e-01 -2.52152234e-01 1.01658452e+00
-7.90428579e-01 -5.95303893e-01 -2.12671030e-02 3.41634005e-01
-1.78220496e-02 -1.39449880e-01 1.06841218e+00 -2.70697474e-01
-4.65675890e-02 5.08201540e-01 -9.87216234e-02 5.70967272e-02
1.09048223e+00 -1.57938108e-01 -6.20193064e-01 -1.96029559e-01
-5.23622811e-01 2.84830362e-01 2.18655914e-01 1.36281982e-01
4.97483850e-01 -1.19812977e+00 -7.84743428e-01 5.95543943e-02
2.49161109e-01 -2.45380789e-01 5.37115693e-01 9.52548742e-01
-3.19166303e-01 3.33797336e-01 -4.56246972e-01 -3.88221651e-01
-1.10607123e+00 3.66856515e-01 5.38879573e-01 -6.36523128e-01
-1.86739549e-01 9.13269997e-01 4.56306547e-01 -6.06933892e-01
3.90579134e-01 -8.66338834e-02 -3.55288476e-01 -6.82061687e-02
5.50712943e-01 5.82845390e-01 -2.10909367e-01 -2.29718089e-01
-4.98484671e-01 -3.15647572e-02 -3.01642179e-01 2.62128621e-01
1.20573962e+00 1.76637396e-01 -2.51844555e-01 2.01336686e-02
7.20998466e-01 -1.21331796e-01 -9.24605787e-01 5.88054024e-02
-1.85755715e-01 -9.18151081e-01 3.15904468e-01 -1.22518778e+00
-1.53824055e+00 1.01724648e+00 7.87346303e-01 1.15590706e-01
1.28925931e+00 -5.29280066e-01 5.86985230e-01 1.08343005e-01
2.96226501e-01 -1.12131774e+00 -4.43814993e-01 -1.42521113e-01
3.75334024e-01 -1.40865684e+00 5.74007295e-02 -2.77858973e-01
-8.88664305e-01 9.74770367e-01 8.30985606e-01 -1.18226446e-01
5.55454612e-01 2.22407520e-01 2.03267083e-01 -1.91082865e-01
-7.10567355e-01 1.65220305e-01 4.07597244e-01 4.86620218e-01
5.63233793e-01 5.25146246e-01 -7.20264196e-01 9.71306086e-01
-1.09057866e-01 5.18211126e-01 4.67676789e-01 6.39157951e-01
-1.47785127e-01 -1.12832332e+00 -2.81660259e-01 8.79992366e-01
-1.12015891e+00 -2.44520426e-01 -5.37667692e-01 4.45894390e-01
3.31010938e-01 7.80178368e-01 2.88985185e-02 -1.24159820e-01
1.25428274e-01 -1.91943079e-01 3.26709241e-01 -1.98602006e-01
-9.96114492e-01 -4.48897362e-01 2.45342344e-01 -9.10866499e-01
-7.00021684e-01 -3.17414016e-01 -8.65593016e-01 -8.05622399e-01
-3.84607762e-01 8.68444294e-02 2.40490437e-01 8.64788115e-01
2.68200129e-01 6.72895193e-01 5.26975691e-01 -7.54458457e-03
-6.85989141e-01 -4.98625457e-01 -5.41450381e-01 2.95340151e-01
4.27163154e-01 -4.27751124e-01 -2.89436936e-01 -4.83193129e-01] | [8.969541549682617, 5.156896591186523] |
4a890052-cea3-465c-88d3-eaeb209fe939 | speaker-profiling-in-multi-party | null | null | https://openreview.net/forum?id=iozkB44VlOl | https://openreview.net/pdf?id=iozkB44VlOl | Speaker Profiling in Multi-party Conversations | In a conversation, individual speakers respond uniquely. Consequently, a `one size fits all' technique is not the best way for a dialog agent to generate responses. While many studies design personalized dialog agents with the help of persona information of speakers, all of them assume that speaker persona is supplied beforehand. However, this assumption does not hold for all conversational agents. This paper attempts to address this gap by exploring the task -- Speaker Profiling in Conversations (SPC). The aim of SPC is to produce a summary of the persona characteristics for individual speakers present in a dialog. We divide this task into two subtasks -- persona discovery and persona-type identification. Given a dialog, the former subtask aims to find all its utterances that carry persona information. The latter subtask focuses on evaluating these utterances to identify the type of persona information present in them. We present SPICE, a novel dataset that is specifically curated with labels to tackle the task of SPC. We show the performance of various baselines on SPICE and benchmark it using SPOT, a neural model based on GRU and Transformer. SPOT shows an improvement of ~7% for persona discovery and ~3% for persona-type identification over the best baseline. We further present a thorough analysis of SPOT to diagnose individual modules and their limitations, both quantitatively and qualitatively. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['speaker-profiling'] | ['speech'] | [ 4.79574911e-02 3.57782930e-01 1.79511830e-01 -7.96940207e-01
-8.23316753e-01 -6.90214932e-01 9.47350323e-01 -3.95893976e-02
-2.30608687e-01 4.46462512e-01 6.26684785e-01 -1.20219275e-01
-3.83194350e-02 -5.07885277e-01 -7.28849992e-02 -6.98721111e-01
1.96461365e-01 1.14801562e+00 8.19532722e-02 -4.92772430e-01
-1.95640605e-02 4.45501566e-01 -1.37724936e+00 6.28516495e-01
4.86395359e-01 5.98078489e-01 1.66129440e-01 7.53152788e-01
-5.27227521e-01 7.30087340e-01 -1.09775448e+00 -5.47404051e-01
-1.58784077e-01 -6.24247372e-01 -1.30486560e+00 3.91766787e-01
1.27910838e-01 -3.77148926e-01 -1.39482440e-02 7.11877108e-01
6.54706299e-01 2.14008451e-01 7.89590716e-01 -1.41406429e+00
1.92255780e-01 1.17488217e+00 4.05748375e-03 6.32265508e-02
7.81720579e-01 -3.88995931e-02 1.08933938e+00 -5.90138614e-01
3.26812893e-01 1.88126945e+00 5.43627560e-01 1.16197932e+00
-1.02296293e+00 -3.65110844e-01 3.89887393e-02 -2.97488004e-01
-1.12556231e+00 -7.74966300e-01 7.00530231e-01 -5.50511897e-01
8.58013630e-01 5.19337296e-01 3.09621394e-02 1.48906374e+00
-4.65696126e-01 8.76535475e-01 1.17067814e+00 -3.80095571e-01
2.21614525e-01 6.47664487e-01 7.72117615e-01 3.40885162e-01
-5.60040593e-01 -2.45406017e-01 -7.77313828e-01 -4.21448886e-01
2.84211069e-01 -3.05522978e-01 -2.75718182e-01 2.69429445e-01
-9.83291984e-01 9.03458893e-01 -6.30429834e-02 4.65053678e-01
-4.47557271e-01 -4.07775432e-01 4.13567364e-01 3.66802633e-01
3.11025262e-01 5.67558050e-01 -2.35212952e-01 -4.08406019e-01
-4.99096096e-01 6.69959247e-01 1.50018394e+00 1.03003478e+00
5.57743251e-01 -2.63856649e-01 -6.13943636e-01 1.23545063e+00
3.69425058e-01 3.39045405e-01 5.55684686e-01 -9.18066919e-01
4.91841167e-01 9.76527393e-01 2.97127306e-01 -6.87034011e-01
-5.76378465e-01 -1.09671876e-01 -5.26719809e-01 5.53439185e-02
7.77556837e-01 -4.68608707e-01 -3.84999514e-01 1.79291570e+00
3.04969341e-01 -4.28830445e-01 4.11358446e-01 7.80586541e-01
1.28091013e+00 7.22890854e-01 7.38069564e-02 -1.88021362e-01
1.86721206e+00 -8.47518086e-01 -8.67441654e-01 -1.34499073e-01
3.35124284e-01 -6.74617231e-01 1.02341211e+00 6.18696511e-02
-1.01005232e+00 -4.36025918e-01 -6.75589323e-01 2.30687857e-01
-4.10243481e-01 -2.26393975e-02 3.61513615e-01 1.29172838e+00
-8.94881904e-01 2.37422094e-01 -3.75572503e-01 -6.53161943e-01
-1.77392468e-01 4.34348911e-01 -2.94333339e-01 3.70144784e-01
-1.23497617e+00 8.08070898e-01 -4.28619049e-02 -1.58685073e-01
-8.69435132e-01 -4.35272694e-01 -6.42000794e-01 1.70228347e-01
4.06739414e-01 -4.71535891e-01 1.84504890e+00 -6.73736811e-01
-1.93069792e+00 1.03094542e+00 -6.09400213e-01 -6.55107379e-01
7.10084558e-01 9.80277658e-02 -4.33078527e-01 -1.07681915e-01
-1.27994409e-02 5.45842946e-01 5.02413392e-01 -1.35476565e+00
-9.79421139e-01 -3.08515012e-01 3.17260087e-01 3.28461260e-01
-2.08698645e-01 6.77494049e-01 -4.96411741e-01 -8.68061408e-02
-1.30719095e-01 -8.43475699e-01 1.43184289e-01 -7.60210574e-01
-6.67832494e-01 -9.98128951e-01 7.06719577e-01 -6.49708688e-01
1.02044666e+00 -1.95297074e+00 3.79227959e-02 8.77826065e-02
3.02804291e-01 6.17079698e-02 1.26480445e-01 8.02895129e-01
2.02290460e-01 -3.43805961e-02 5.20068556e-02 -1.11922336e+00
5.05763829e-01 2.76061241e-02 -4.90223229e-01 3.74401808e-02
-9.37518850e-02 5.09306669e-01 -5.96576095e-01 -3.57910722e-01
5.97471446e-02 3.36885929e-01 -3.04065466e-01 6.61273956e-01
-2.92098552e-01 5.65133333e-01 -3.24655026e-01 3.28137040e-01
4.12711650e-01 7.44477659e-02 3.14538330e-01 1.14354819e-01
-1.83344454e-01 7.33995140e-01 -1.06963801e+00 1.12579691e+00
-4.63618040e-01 5.43006599e-01 4.54561651e-01 -5.31788886e-01
1.14864755e+00 7.89656997e-01 4.75143820e-01 -1.44128934e-01
2.38036022e-01 2.35946253e-01 2.77805328e-01 -4.29672301e-01
6.28721237e-01 -2.66671162e-02 -5.09847164e-01 7.30739832e-01
3.22023891e-02 1.84810013e-01 2.25938350e-01 2.81595647e-01
1.08229411e+00 -4.55535740e-01 2.00462297e-01 -3.33832383e-01
9.65437591e-01 -1.79350913e-01 4.05792028e-01 9.82107222e-01
-4.32178557e-01 4.44143981e-01 7.23225534e-01 -2.38565579e-01
-6.28905058e-01 -8.48751724e-01 1.05580479e-01 1.52999175e+00
-1.38116553e-01 -3.79803002e-01 -1.20293689e+00 -8.25743735e-01
-1.98447809e-01 8.51162076e-01 -4.43888098e-01 2.87142396e-01
-6.44936204e-01 -6.61076963e-01 7.29845762e-01 9.69617292e-02
7.25646675e-01 -1.33551359e+00 -4.64774013e-01 2.38566980e-01
-4.87006634e-01 -1.33424127e+00 -5.88608384e-01 -3.10436971e-02
-6.20895885e-02 -7.86595166e-01 -6.45320714e-01 -8.46176088e-01
2.32244641e-01 7.74081051e-02 1.15391469e+00 -1.49424195e-01
2.93378681e-01 4.20745194e-01 -5.09114563e-01 -5.16442716e-01
-1.22097874e+00 3.23801547e-01 1.54086336e-01 3.39098811e-01
7.13249326e-01 -3.58790696e-01 -2.83957452e-01 7.66669869e-01
-2.28515819e-01 7.32018426e-02 2.40108714e-01 4.38612401e-01
-3.24819863e-01 1.90642610e-01 6.82852805e-01 -1.12885058e+00
9.76588547e-01 -3.32770050e-01 -1.61542997e-01 2.38242701e-01
-5.17859310e-02 -1.66081175e-01 4.68429178e-01 -2.11138055e-01
-1.33683431e+00 2.40784157e-02 -4.22478437e-01 2.23446965e-01
-6.66025460e-01 1.19361281e-01 -7.28239417e-01 5.04252493e-01
6.02208138e-01 2.16903493e-01 1.56640723e-01 -6.51270807e-01
-7.61377960e-02 1.36069238e+00 4.93563831e-01 -7.68652678e-01
3.93239290e-01 9.45997387e-02 -7.06379712e-01 -1.02956128e+00
-8.23776066e-01 -1.04258382e+00 -3.77514571e-01 -6.24891818e-01
9.43162382e-01 -8.55432928e-01 -1.26697206e+00 7.84311652e-01
-1.31359041e+00 -4.24211621e-01 1.06020622e-01 1.83958616e-02
-4.59979683e-01 8.78896713e-02 -5.57333171e-01 -1.30655360e+00
-4.83715504e-01 -1.32294858e+00 1.02956319e+00 2.00032175e-01
-7.77842104e-01 -9.37172651e-01 -1.48166120e-01 7.42071450e-01
3.42621565e-01 -1.87485904e-01 6.65951312e-01 -1.49508035e+00
-7.63740838e-02 -1.34041861e-01 8.90714005e-02 9.96588990e-02
4.31515396e-01 -4.79506791e-01 -1.47977090e+00 -1.76218376e-01
1.69166952e-01 -1.28474265e-01 3.67460668e-01 -2.53655333e-02
6.08156323e-01 -5.54862082e-01 -3.27310354e-01 -2.01398227e-02
5.67498326e-01 4.58827883e-01 2.94984967e-01 2.37223934e-02
4.58954781e-01 1.45640087e+00 2.25579411e-01 3.08579594e-01
6.90754831e-01 1.16325033e+00 2.95991488e-02 2.05399305e-01
-1.16253130e-01 -1.05707049e-01 6.28090084e-01 5.01543283e-01
-6.08047144e-03 -5.59591353e-01 -1.05166721e+00 4.87194955e-01
-1.71993589e+00 -1.02511930e+00 -1.91333786e-01 1.88800895e+00
1.00974798e+00 2.67152619e-02 8.74119520e-01 1.29712835e-01
9.67394650e-01 7.46435719e-03 -9.88771245e-02 -3.82470995e-01
-5.22254854e-02 -3.71975303e-01 -3.80975790e-02 8.16410244e-01
-1.10666966e+00 8.36092234e-01 5.63764238e+00 4.64052349e-01
-6.95412576e-01 7.21432343e-02 5.13948500e-01 2.75561690e-01
-8.47338513e-02 -2.29343042e-01 -1.46683562e+00 6.39647543e-01
1.33273017e+00 -1.66060477e-01 4.19117272e-01 7.19607294e-01
4.41667110e-01 1.82703048e-01 -1.44696283e+00 7.48371601e-01
1.80363178e-01 -9.72139537e-01 -3.19119170e-02 5.33798002e-02
2.83310682e-01 -4.18302208e-01 -2.02311859e-01 6.43106043e-01
6.56910837e-01 -8.95581663e-01 6.63227141e-01 2.44592443e-01
2.10611224e-01 -5.42722225e-01 8.97842705e-01 4.78017122e-01
-7.60371864e-01 -6.53399825e-02 -1.24853961e-02 1.17935374e-01
3.15401167e-01 1.63973451e-01 -1.43917704e+00 3.26670527e-01
6.62812829e-01 1.20342337e-01 -3.27458709e-01 6.89989448e-01
1.67135075e-02 6.42365396e-01 -2.25806400e-01 -3.53249758e-01
1.56905368e-01 -1.89511466e-03 8.72386813e-01 1.46742344e+00
9.07870904e-02 6.98366016e-02 3.22123289e-01 8.72077346e-01
5.50736114e-02 1.43540323e-01 -1.60474420e-01 1.41729504e-01
8.58272076e-01 1.24257064e+00 -4.37385172e-01 -3.72402191e-01
-2.95553684e-01 1.02731550e+00 6.53892905e-02 1.84516326e-01
-4.32816714e-01 -1.34146124e-01 6.42631233e-01 1.78938766e-03
-1.54154658e-01 1.80400267e-01 -1.48407882e-02 -5.03411651e-01
-2.65482634e-01 -1.29773676e+00 5.58186948e-01 -4.74021405e-01
-1.53678727e+00 8.85710359e-01 1.26015589e-01 -7.83653259e-01
-9.36473250e-01 -6.42086387e-01 -8.76360357e-01 1.29625762e+00
-8.63569617e-01 -1.26490784e+00 -4.99019325e-01 4.83724952e-01
1.04689085e+00 -6.47352159e-01 1.06097651e+00 4.86084595e-02
-7.94987559e-01 7.36713111e-01 -3.88206869e-01 4.55085784e-01
7.81477749e-01 -1.60054052e+00 5.08524358e-01 3.36346835e-01
-2.06637517e-01 8.18624556e-01 1.23277986e+00 -4.32567060e-01
-9.15373564e-01 -8.58866334e-01 1.34614503e+00 -7.24660456e-01
5.26410639e-01 -7.08198428e-01 -8.95229816e-01 5.93023181e-01
3.99933130e-01 -9.87860203e-01 5.46280861e-01 4.81961727e-01
-2.33137086e-01 2.24536061e-02 -1.12422764e+00 5.78637600e-01
7.00822592e-01 -5.80506563e-01 -6.74365878e-01 2.93308973e-01
6.13604844e-01 -3.69860679e-01 -7.35674977e-01 -6.12625033e-02
3.22906822e-01 -1.15677822e+00 8.69600236e-01 -3.73140603e-01
9.40465555e-02 1.87921580e-02 -1.23849697e-01 -1.20485806e+00
1.23295523e-01 -1.05743957e+00 2.02026904e-01 1.95356262e+00
5.09859562e-01 -7.20840216e-01 7.83251107e-01 8.55765283e-01
-1.81336641e-01 -1.29315719e-01 -8.32298160e-01 -6.48816228e-01
1.38142407e-01 -2.55834341e-01 9.51627791e-01 8.19102108e-01
2.43784353e-01 9.36831117e-01 -3.77419204e-01 2.81568587e-01
4.18273658e-01 -3.92561071e-02 1.09925818e+00 -1.27198493e+00
-3.31543744e-01 -5.96874535e-01 8.15253332e-02 -1.00773990e+00
4.03154761e-01 -5.80156684e-01 4.61970091e-01 -1.40934670e+00
3.05146188e-01 -4.47752982e-01 3.13683391e-01 3.15961957e-01
-2.47960642e-01 -3.69508505e-01 2.81191599e-02 3.86305779e-01
-3.46348017e-01 2.52941251e-01 6.20079994e-01 -2.05973744e-01
-5.53051710e-01 7.64861941e-01 -5.13285220e-01 6.44824684e-01
9.88458157e-01 -1.24995276e-01 -3.65071535e-01 -2.91302428e-02
-2.60673761e-01 3.13618243e-01 3.74017417e-01 -8.79299045e-01
3.43792200e-01 5.97065836e-02 -3.34534138e-01 -6.91576004e-01
6.72807455e-01 -6.35832489e-01 -1.33129194e-01 8.54357928e-02
-7.28794992e-01 -1.07995756e-01 -1.07536815e-01 2.86694795e-01
-1.35113806e-01 -5.96652210e-01 6.04279935e-01 -1.96411267e-01
-5.49126208e-01 -3.63950245e-02 -7.65713036e-01 -1.45074546e-01
6.61745429e-01 9.27567407e-02 -4.90316272e-01 -8.76624286e-01
-6.16059840e-01 3.28819424e-01 1.70433700e-01 4.42861468e-01
6.31673113e-02 -8.02609086e-01 -8.91993761e-01 -7.82053396e-02
1.89970613e-01 -2.56817132e-01 3.80531222e-01 5.39091766e-01
-1.41428802e-02 5.08533597e-01 4.35993820e-02 -5.10407388e-01
-1.75275874e+00 1.85104713e-01 6.91739500e-01 -3.49165887e-01
-4.78685141e-01 9.32205677e-01 3.98907065e-01 -7.85642266e-01
6.50697112e-01 1.53714731e-01 -8.50994408e-01 2.51940310e-01
8.38843703e-01 1.41594261e-01 -8.73721540e-02 -9.41357374e-01
-3.41508627e-01 -1.44447446e-01 -1.94639206e-01 -5.47617555e-01
1.00545061e+00 -3.17378372e-01 -1.70565352e-01 6.81359470e-01
9.05315578e-01 1.94348827e-01 -1.04216182e+00 -4.41596359e-01
1.75392091e-01 -2.19983384e-02 -4.01508808e-01 -1.02621770e+00
-4.39768016e-01 7.42787123e-01 3.53486061e-01 8.47485542e-01
6.02662027e-01 3.40457082e-01 5.53252697e-01 3.40482175e-01
2.26517275e-01 -9.29646909e-01 9.28058401e-02 8.47205460e-01
1.18950462e+00 -1.31051540e+00 -4.63616431e-01 -6.87836766e-01
-9.62251306e-01 9.02352095e-01 6.77552283e-01 5.46397448e-01
3.27642918e-01 -3.23347673e-02 4.29619521e-01 -2.82441497e-01
-6.97727382e-01 -2.43637621e-01 3.25349569e-01 5.87455571e-01
3.98561329e-01 2.26069048e-01 2.79916916e-02 9.54588652e-01
-8.81071687e-01 -7.84437597e-01 4.29211885e-01 4.94696140e-01
-3.54649544e-01 -1.23956037e+00 -5.08083820e-01 1.50261790e-01
-3.98391366e-01 1.78005368e-01 -8.52751255e-01 6.20957971e-01
-2.51028895e-01 1.71037781e+00 1.66466951e-01 -4.76593345e-01
5.59275389e-01 4.38896984e-01 -5.36666997e-02 -8.21895957e-01
-1.14200103e+00 3.04203406e-02 8.51117849e-01 -4.97206673e-02
-5.65296710e-01 -1.05427015e+00 -9.99706089e-01 -6.15097523e-01
-3.19897532e-02 5.22402048e-01 6.33516788e-01 1.03345644e+00
-1.11466177e-01 5.17652214e-01 7.92610765e-01 -4.97906774e-01
-6.37539744e-01 -1.35815871e+00 -3.73402864e-01 6.19748831e-01
1.88223794e-01 -4.46079075e-01 -5.43118596e-01 1.35724777e-02] | [12.781282424926758, 7.884990692138672] |
3d8b2e69-5f16-49c1-8521-5db3cf9e81d2 | memonav-selecting-informative-memories-for | 2208.09610 | null | https://arxiv.org/abs/2208.09610v1 | https://arxiv.org/pdf/2208.09610v1.pdf | MemoNav: Selecting Informative Memories for Visual Navigation | Image-goal navigation is a challenging task, as it requires the agent to navigate to a target indicated by an image in a previously unseen scene. Current methods introduce diverse memory mechanisms which save navigation history to solve this task. However, these methods use all observations in the memory for generating navigation actions without considering which fraction of this memory is informative. To address this limitation, we present the MemoNav, a novel memory mechanism for image-goal navigation, which retains the agent's informative short-term memory and long-term memory to improve the navigation performance on a multi-goal task. The node features on the agent's topological map are stored in the short-term memory, as these features are dynamically updated. To aid the short-term memory, we also generate long-term memory by continuously aggregating the short-term memory via a graph attention module. The MemoNav retains the informative fraction of the short-term memory via a forgetting module based on a Transformer decoder and then incorporates this retained short-term memory and the long-term memory into working memory. Lastly, the agent uses the working memory for action generation. We evaluate our model on a new multi-goal navigation dataset. The experimental results show that the MemoNav outperforms the SoTA methods by a large margin with a smaller fraction of navigation history. The results also empirically show that our model is less likely to be trapped in a deadlock, which further validates that the MemoNav improves the agent's navigation efficiency by reducing redundant steps. | ['Zhaoxiang Zhang', 'Shuqi Mei', 'Yuran Yang', 'Xu Yang', 'Hongxin Li'] | 2022-08-20 | null | null | null | null | ['action-generation'] | ['computer-vision'] | [ 2.25771498e-02 3.43742758e-01 1.27517805e-01 -4.41519618e-02
-4.75611836e-01 -3.41263175e-01 7.02186465e-01 -2.38619387e-01
-7.33628988e-01 8.29680383e-01 3.98997933e-01 -1.37177825e-01
-1.40186235e-01 -1.34977913e+00 -9.95963812e-01 -7.97927380e-01
-1.65759698e-01 5.38679361e-01 7.81514645e-01 -5.53020775e-01
5.57486773e-01 -3.96063849e-02 -1.57957494e+00 1.79744288e-02
9.33684111e-01 7.50812232e-01 1.05094481e+00 6.21628106e-01
-3.39795575e-02 9.75679874e-01 -5.28189659e-01 -9.76157188e-02
2.86576331e-01 -7.53524899e-01 -8.51224840e-01 -2.27947205e-01
1.16842858e-01 -5.47138512e-01 -7.45716393e-01 9.40718710e-01
5.25326431e-01 7.38189220e-01 3.22002321e-01 -1.06960821e+00
-7.50550568e-01 5.65350235e-01 -5.07345721e-02 3.46052468e-01
4.28667575e-01 3.98956209e-01 7.18477726e-01 -8.19265425e-01
9.45034683e-01 1.28165686e+00 2.82225102e-01 6.06721640e-01
-7.23275542e-01 -4.56998557e-01 6.66361094e-01 3.58416408e-01
-1.15572655e+00 -1.60120443e-01 5.68358183e-01 -1.36659190e-01
1.21662104e+00 -4.12110426e-02 9.36957598e-01 1.01091564e+00
9.23375309e-01 6.10960364e-01 8.36473823e-01 -1.57346919e-01
4.44882780e-01 -3.68514448e-01 1.23629622e-01 1.44026101e+00
4.54191677e-03 4.74650830e-01 -8.82966101e-01 1.84051424e-01
9.34490204e-01 2.61024758e-02 -3.11365277e-01 -4.36088920e-01
-1.03114700e+00 1.04209983e+00 9.26047266e-01 2.02385962e-01
-5.28066695e-01 4.16312784e-01 -1.85579378e-02 3.93035442e-01
-9.97220911e-03 5.33807456e-01 -7.76484311e-02 -2.25619510e-01
-3.83614391e-01 -3.31332162e-02 5.15757084e-01 8.80164385e-01
8.82649183e-01 1.13338731e-01 -2.63421983e-01 5.94205737e-01
3.65974486e-01 5.00721812e-01 9.18631077e-01 -9.87896323e-01
5.76952159e-01 8.07412982e-01 -6.83356598e-02 -1.14642632e+00
-5.39425671e-01 -5.18769026e-01 -6.12344444e-01 5.05064070e-01
5.06483503e-02 4.45561446e-02 -1.38205779e+00 2.24448466e+00
6.65066168e-02 6.44515827e-02 2.39046693e-01 1.07050192e+00
7.93010175e-01 8.43972087e-01 1.04956767e-02 -3.96714732e-02
9.64246988e-01 -1.74166930e+00 -4.73844677e-01 -9.51961696e-01
7.17885435e-01 -1.29692987e-01 1.20995200e+00 -1.59895763e-01
-9.99766290e-01 -7.29709625e-01 -1.13813460e+00 -8.28478634e-02
-6.80789948e-01 -2.89125293e-01 6.74696684e-01 2.24868700e-01
-1.48748672e+00 5.80438852e-01 -8.32379937e-01 -5.56693852e-01
1.72557086e-01 3.59535635e-01 -4.36050653e-01 -1.69166684e-01
-1.33085775e+00 1.12623048e+00 6.01967633e-01 4.65808362e-02
-1.32721126e+00 -6.33660480e-02 -1.27147484e+00 -3.34857889e-02
5.61465979e-01 -1.25743639e+00 8.44721258e-01 -5.45051873e-01
-1.51904202e+00 4.25908536e-01 -3.93348783e-01 -6.12597764e-01
2.92057574e-01 1.69387385e-02 -4.79191169e-02 3.23171169e-02
3.50893289e-01 1.05703139e+00 7.45756984e-01 -1.05757272e+00
-7.70985305e-01 -4.58405912e-01 4.20064718e-01 8.02118599e-01
-1.54613882e-01 -8.47452879e-01 -8.25846016e-01 -6.50495648e-01
4.51907843e-01 -1.19971180e+00 -5.56800604e-01 -2.73253828e-01
-3.04285049e-01 1.50673375e-01 6.66868150e-01 -5.46957850e-01
1.27226460e+00 -1.84068167e+00 5.24863482e-01 9.27896425e-02
1.15296319e-01 -5.53445444e-02 -6.06922984e-01 4.14656013e-01
3.07399184e-01 2.00457536e-02 -1.99401751e-01 -3.54243606e-01
-1.90680251e-01 4.08255994e-01 -3.70125264e-01 2.25457456e-02
-3.88042301e-01 1.33371735e+00 -1.00656700e+00 -2.15129286e-01
2.23086789e-01 2.48284549e-01 -6.13194823e-01 1.58507168e-01
-2.21748352e-01 4.20307040e-01 -5.00902534e-01 1.88227594e-01
2.72433281e-01 -3.67159337e-01 2.07638927e-02 1.53508827e-01
-1.02104573e-02 3.78278285e-01 -8.12849700e-01 2.12560511e+00
-4.65003520e-01 2.77273238e-01 -3.82640809e-01 -4.65811849e-01
8.03923666e-01 -4.11226988e-01 -9.19841304e-02 -1.51351810e+00
3.87552232e-02 8.25122595e-02 3.35195735e-02 -3.14757556e-01
7.64187217e-01 2.54581064e-01 -1.61604807e-01 6.84423029e-01
6.02641813e-02 1.12789102e-01 4.50978398e-01 5.12615383e-01
1.45020199e+00 2.69587129e-01 5.18177263e-02 -8.40425789e-02
5.06468534e-01 1.46713927e-01 3.42677444e-01 1.25613296e+00
-1.46530077e-01 5.17955005e-01 -1.51989171e-02 -4.14755732e-01
-6.37469888e-01 -1.05926991e+00 6.34908617e-01 1.44718897e+00
6.78781748e-01 -6.48524642e-01 -7.18881011e-01 -7.83615887e-01
-2.84831136e-01 8.61963272e-01 -9.78432059e-01 -8.77518594e-01
-9.02638912e-01 -5.37420630e-01 1.19017079e-01 5.79097629e-01
9.57360268e-01 -1.49040234e+00 -9.62145984e-01 2.92904109e-01
-3.66467059e-01 -4.95409131e-01 -6.18636310e-01 1.49908245e-01
-8.42965007e-01 -9.70404625e-01 -6.19279206e-01 -9.44963098e-01
8.86874318e-01 5.30878007e-01 1.02811456e+00 4.06045288e-01
8.70496780e-02 3.87865514e-01 -2.86888003e-01 3.12256068e-01
-2.73534536e-01 3.63458306e-01 -2.14134604e-01 -4.96170104e-01
2.38057282e-02 -5.19275606e-01 -6.69708669e-01 5.10681033e-01
-6.05259836e-01 4.34604228e-01 5.50302088e-01 9.94898081e-01
5.92893720e-01 7.12012500e-02 4.86617267e-01 -5.54582953e-01
6.18559480e-01 -4.90609556e-01 -6.48545265e-01 5.02373204e-02
-5.65815687e-01 4.70364809e-01 4.22454327e-01 -3.05754364e-01
-1.01717138e+00 -8.13861489e-02 -1.49893209e-01 5.83851151e-02
3.59302044e-01 7.55894005e-01 -1.74074352e-01 -1.24868676e-01
4.13443595e-01 7.73024261e-01 1.38973549e-01 -1.41638845e-01
3.26182574e-01 -1.19448535e-01 4.66264188e-01 -1.88943386e-01
5.34663320e-01 3.66069078e-01 5.39510064e-02 -2.95002341e-01
-6.14690483e-01 1.17608197e-01 -3.78176123e-01 -3.00315201e-01
8.12287629e-01 -6.50038540e-01 -7.43618906e-01 5.53757131e-01
-9.05206919e-01 -8.49106431e-01 -1.82439253e-01 3.66888314e-01
-7.79138803e-01 9.37153995e-02 -7.73497641e-01 -4.88089293e-01
-2.19873786e-01 -1.40204573e+00 8.36137354e-01 4.52791184e-01
5.55888005e-02 -1.10035145e+00 1.74702078e-01 1.05294876e-01
5.45138299e-01 -1.75684810e-01 8.56729329e-01 -3.14426690e-01
-1.19816291e+00 1.49125293e-01 -6.13255333e-03 -3.90169799e-01
-1.60229340e-01 -8.22893739e-01 -4.60358262e-01 -5.26561260e-01
-1.42517686e-01 -1.75254822e-01 1.52478623e+00 4.13007170e-01
7.21847177e-01 -2.69332081e-01 -6.06194794e-01 6.05705678e-01
1.26177645e+00 6.51837289e-01 8.60229909e-01 9.08768296e-01
5.10727406e-01 4.58322674e-01 8.23761284e-01 4.64140736e-02
9.57516372e-01 4.23904210e-01 9.34168756e-01 2.67655194e-01
-4.23674941e-01 -7.11391866e-01 6.04273975e-01 6.84506893e-01
7.45917931e-02 -5.17216921e-01 -8.16644907e-01 5.79480648e-01
-2.20149493e+00 -9.99239802e-01 3.14029038e-01 2.14297581e+00
4.43614751e-01 3.86186808e-01 -2.01502100e-01 -3.97842973e-01
6.37455404e-01 4.92477804e-01 -8.20697248e-01 -1.97343528e-03
-1.50693864e-01 -2.21610680e-01 4.19270903e-01 9.49587941e-01
-8.37748051e-01 1.46776223e+00 6.31998014e+00 6.11007035e-01
-8.79135013e-01 5.81756458e-02 3.49966794e-01 -3.59010771e-02
-2.21051469e-01 -5.48783364e-03 -9.99797404e-01 5.23075402e-01
8.77756834e-01 -1.63334846e-01 6.26292646e-01 9.10147786e-01
-1.04708582e-01 -7.12454021e-01 -8.94209266e-01 8.66630495e-01
5.01046538e-01 -1.46138430e+00 3.86410326e-01 2.58106798e-01
4.83160406e-01 8.04287419e-02 3.87733161e-01 6.63279593e-01
5.85643291e-01 -1.03662133e+00 8.01276147e-01 7.09139109e-01
2.43958876e-01 -7.39474952e-01 6.29411757e-01 5.16661525e-01
-1.30724466e+00 -5.00566542e-01 -4.93589103e-01 -2.48720944e-01
5.17347634e-01 6.87912107e-02 -5.14227331e-01 2.96130925e-01
5.01928329e-01 5.78157187e-01 -9.90383446e-01 1.00275362e+00
-3.55350465e-01 -1.30210603e-02 -1.49543926e-01 -3.57266255e-02
5.57006180e-01 -3.74990046e-01 5.98025978e-01 4.92060065e-01
6.22438192e-01 6.52789846e-02 3.55579793e-01 7.22050071e-01
1.52695790e-01 -3.20179731e-01 -8.23528349e-01 9.61424336e-02
4.76370543e-01 8.45466971e-01 -8.95308554e-01 -3.92278910e-01
-2.59792693e-02 1.34204590e+00 5.69261193e-01 4.65842158e-01
-1.00255501e+00 -2.33680233e-01 2.51709580e-01 -6.76510856e-02
2.28448436e-01 -3.92381608e-01 -1.54181778e-01 -8.17498863e-01
-6.91943094e-02 -3.69781971e-01 6.29086435e-01 -1.13448501e+00
-6.13715410e-01 1.13059068e+00 -3.80361676e-01 -8.39091241e-01
-5.89835763e-01 -2.77574211e-01 -4.18760151e-01 6.88130498e-01
-1.33643568e+00 -1.20937479e+00 -6.12426698e-01 6.77602053e-01
8.28450859e-01 -4.32830900e-01 9.44411516e-01 8.52116570e-03
-5.14944673e-01 3.09952348e-01 -4.49040234e-01 -7.52971917e-02
3.88203442e-01 -1.06546748e+00 6.06788993e-01 9.72053409e-01
-6.02856800e-02 7.76077569e-01 5.18238783e-01 -1.16545272e+00
-1.36924577e+00 -1.19821095e+00 6.99598730e-01 -4.53931361e-01
4.09988761e-01 -2.93173224e-01 -7.24423826e-01 1.15645611e+00
1.19586356e-01 -2.68985242e-01 2.83371806e-01 -2.34260336e-01
-1.82165168e-02 2.49445617e-01 -8.99657428e-01 1.03468168e+00
1.73558664e+00 -2.69223332e-01 -8.67179811e-01 2.57988032e-02
8.60976696e-01 -5.53390920e-01 -7.14601353e-02 8.08544680e-02
3.85283560e-01 -1.09908664e+00 8.08768511e-01 -3.76574516e-01
3.06838989e-01 -5.00515819e-01 -1.79684326e-01 -1.66109180e+00
-6.92139864e-01 -3.14405411e-01 -3.05852175e-01 7.02275038e-01
6.43357337e-01 -8.75414371e-01 9.37226832e-01 5.26267171e-01
-3.97949040e-01 -5.49892128e-01 -9.50363874e-01 -7.39506602e-01
-1.74765125e-01 -5.27380779e-02 4.64504212e-01 4.11429554e-01
-2.21971378e-01 6.60902917e-01 -4.72858638e-01 5.78076132e-02
2.67031133e-01 3.91121805e-01 7.32275784e-01 -8.26149642e-01
-6.76108971e-02 -2.78162450e-01 -2.08643213e-01 -1.54869103e+00
2.38114640e-01 -9.43791449e-01 1.71878368e-01 -2.07420158e+00
4.16481763e-01 -2.57886797e-01 -2.15215251e-01 6.03588760e-01
-2.00742394e-01 1.89476401e-01 3.78874391e-01 1.02682568e-01
-9.70109284e-01 9.46510375e-01 1.54185927e+00 -2.07478032e-01
-4.75460023e-01 -3.04612517e-01 -7.28307486e-01 7.81296432e-01
7.00822711e-01 -4.33718473e-01 -8.30122411e-01 -6.80892527e-01
2.48828024e-01 7.99140930e-02 3.81853104e-01 -1.22233021e+00
6.97135985e-01 -1.07530974e-01 4.85119641e-01 -6.41505837e-01
7.09294617e-01 -6.46029651e-01 1.52425870e-01 8.19888234e-01
-3.16526800e-01 4.90705341e-01 9.15090740e-02 7.71036983e-01
3.71649367e-04 -1.42991900e-01 6.55768335e-01 -5.09869993e-01
-1.42031324e+00 3.32986742e-01 -6.50394380e-01 -1.33579209e-01
1.09195495e+00 -4.59973633e-01 -6.04515731e-01 -5.22507131e-01
-1.06099546e+00 5.81107974e-01 5.66685021e-01 7.23719239e-01
9.90512788e-01 -1.47944927e+00 -8.01533908e-02 4.84838486e-01
-8.08001384e-02 -1.97002381e-01 5.07141650e-01 5.85779786e-01
-2.85589367e-01 4.97073650e-01 -5.17696142e-01 -3.79347265e-01
-8.97963703e-01 8.23874295e-01 5.03282785e-01 -3.73370618e-01
-7.97813952e-01 9.61004436e-01 4.18735474e-01 -4.81329799e-01
1.35046586e-01 -2.98588853e-02 -3.11374366e-01 -4.94200550e-02
5.82419097e-01 3.42708170e-01 -2.33487621e-01 -8.19827974e-01
-3.92062843e-01 6.27483189e-01 -1.21599160e-01 -4.74874884e-01
1.19144893e+00 -6.05510652e-01 -5.57719953e-02 1.94015697e-01
8.40847790e-01 -2.39557058e-01 -1.41962373e+00 -8.71103164e-03
-2.56482780e-01 -5.51917970e-01 7.85612781e-03 -9.42967057e-01
-1.20353317e+00 5.71321070e-01 4.70789760e-01 -1.78178206e-01
9.52625513e-01 -1.50756165e-01 9.74510610e-01 6.00376368e-01
9.56794620e-01 -1.15374613e+00 7.34120488e-01 1.05314863e+00
1.15643048e+00 -1.01358378e+00 -4.34251845e-01 -1.99428096e-01
-6.71958804e-01 8.27922523e-01 1.20074582e+00 -1.36405304e-01
4.34726983e-01 -9.29369852e-02 -7.73281977e-02 -2.17377454e-01
-8.46178114e-01 -5.19552886e-01 -5.12558445e-02 7.68415987e-01
-2.70826101e-01 -3.50913227e-01 -1.80870205e-01 5.98244071e-01
-5.88466525e-01 -2.04269052e-01 5.07315397e-01 1.23402238e+00
-8.49810600e-01 -8.46062243e-01 -2.24106647e-02 3.20459276e-01
2.78275371e-01 -2.58370459e-01 -3.46055299e-01 5.71903765e-01
-1.41383171e-01 9.66407895e-01 6.52076975e-02 -5.43773949e-01
2.27318227e-01 1.76663958e-02 3.24560195e-01 -5.87674081e-01
-3.24645400e-01 -7.52620474e-02 1.73135195e-02 -1.09623718e+00
-3.49760358e-03 -7.13201687e-02 -1.64450777e+00 -4.07198548e-01
-4.77433838e-02 1.01128176e-01 3.36640447e-01 6.78038538e-01
7.51488268e-01 8.27002645e-01 2.11591735e-01 -9.02480602e-01
-1.45407453e-01 -6.39948547e-01 -2.17930749e-01 2.99368143e-01
2.77389079e-01 -1.07503915e+00 -2.30154857e-01 -3.06122631e-01] | [4.509902000427246, 0.45255789160728455] |
94bbd35e-df11-4cc5-9191-9060d3199d2e | hybrid-facial-expression-recognition-fer2013 | 2206.09509 | null | https://arxiv.org/abs/2206.09509v2 | https://arxiv.org/pdf/2206.09509v2.pdf | Hybrid Facial Expression Recognition (FER2013) Model for Real-Time Emotion Classification and Prediction | Facial Expression Recognition is a vital research topic in most fields ranging from artificial intelligence and gaming to Human-Computer Interaction (HCI) and Psychology. This paper proposes a hybrid model for Facial Expression recognition, which comprises a Deep Convolutional Neural Network (DCNN) and Haar Cascade deep learning architectures. The objective is to classify real-time and digital facial images into one of the seven facial emotion categories considered. The DCNN employed in this research has more convolutional layers, ReLU Activation functions, and multiple kernels to enhance filtering depth and facial feature extraction. In addition, a haar cascade model was also mutually used to detect facial features in real-time images and video frames. Grayscale images from the Kaggle repository (FER-2013) and then exploited Graphics Processing Unit (GPU) computation to expedite the training and validation process. Pre-processing and data augmentation techniques are applied to improve training efficiency and classification performance. The experimental results show a significantly improved classification performance compared to state-of-the-art (SoTA) experiments and research. Also, compared to other conventional models, this paper validates that the proposed architecture is superior in classification performance with an improvement of up to 6%, totaling up to 70% accuracy, and with less execution time of 2098.8s. | ['Kanyifeechukwu Jane Oguine', 'Daniel Ofuani', 'Hashim Ibrahim Bisallah', 'Ozioma Collins Oguine'] | 2022-06-19 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [-7.04893470e-02 -2.84328222e-01 1.20529413e-01 -5.86222172e-01
1.07528426e-01 1.88563809e-01 2.91568249e-01 -4.63839084e-01
-6.11522436e-01 4.49955940e-01 -3.04026932e-01 -2.30068788e-02
2.89557308e-01 -8.07477772e-01 -2.04264164e-01 -7.47972310e-01
-2.19324678e-01 -3.16486031e-01 -3.35398525e-01 -3.67594838e-01
2.07067453e-04 1.19505513e+00 -2.13689852e+00 5.20019531e-01
2.85972357e-01 1.65209377e+00 -5.23266017e-01 6.17028534e-01
-1.06442101e-01 9.62275267e-01 -4.69728470e-01 -6.28969610e-01
1.19680665e-01 -2.30049357e-01 -4.49527442e-01 2.65208185e-01
3.69220585e-01 -4.36477989e-01 -2.01225445e-01 8.23631108e-01
6.98742986e-01 3.14656608e-02 3.93662989e-01 -1.56641769e+00
-4.24955875e-01 -3.35503638e-01 -8.89766335e-01 1.15376413e-01
1.78649470e-01 -1.67616963e-01 1.93855926e-01 -1.25768495e+00
2.68642753e-01 1.25618851e+00 8.05057645e-01 6.12346351e-01
-5.69240630e-01 -1.27869391e+00 -3.58884096e-01 4.68228132e-01
-1.72429359e+00 -4.81197089e-01 8.42757106e-01 -2.26262599e-01
1.06088960e+00 7.40113407e-02 1.00022125e+00 9.85073149e-01
4.12961602e-01 7.23404646e-01 1.33243787e+00 -5.04352212e-01
1.84178412e-01 1.05380386e-01 1.29492050e-02 1.21403384e+00
-1.23605072e-01 9.13957320e-03 -5.80502331e-01 -9.67138037e-02
9.69016194e-01 3.95882782e-03 1.53156713e-01 3.66496205e-01
-1.77636564e-01 9.94858861e-01 3.63695264e-01 3.56289148e-01
-8.48459601e-01 -1.10430680e-02 7.81731665e-01 2.90665299e-01
4.95086551e-01 -1.98229864e-01 -1.90942720e-01 -1.25557899e-01
-7.81883478e-01 1.17938034e-01 5.53855121e-01 6.30991101e-01
7.45048940e-01 5.60053885e-01 2.50192024e-02 8.45820904e-01
2.82621443e-01 3.59851658e-01 5.65143585e-01 -7.73001611e-01
-1.86299115e-01 9.02356088e-01 -2.80740798e-01 -1.35555637e+00
-5.42988062e-01 -8.12983438e-02 -1.22885394e+00 7.52653837e-01
1.70140900e-02 -3.54336172e-01 -9.60353136e-01 1.28174186e+00
2.92997122e-01 2.38441125e-01 1.07071675e-01 1.20825040e+00
1.05057251e+00 7.86683321e-01 3.80765378e-01 -2.65119553e-01
1.64201951e+00 -8.01698864e-01 -9.73358274e-01 1.57335952e-01
6.07189178e-01 -7.79462039e-01 7.77881742e-01 6.98228657e-01
-9.18971539e-01 -8.85748506e-01 -9.73098218e-01 -1.17499478e-01
-4.79180485e-01 6.80149078e-01 1.23987186e+00 9.38704729e-01
-1.16629648e+00 1.20237030e-01 -5.58213055e-01 -4.55252051e-01
7.50148118e-01 6.58565521e-01 -6.79439366e-01 1.29083037e-01
-1.09968340e+00 6.18080616e-01 9.46991518e-03 4.65583622e-01
-4.37593460e-01 -5.18811047e-01 -8.75758350e-01 9.38025787e-02
-2.90029466e-01 -2.67398745e-01 1.09366143e+00 -1.89292121e+00
-1.88106763e+00 1.01573586e+00 -1.79294869e-01 -3.07167411e-01
1.12702899e-01 -1.67772204e-01 -8.10986638e-01 3.61156374e-01
-4.18634057e-01 7.18760967e-01 8.94897282e-01 -6.92490816e-01
-6.06408000e-01 -7.80323386e-01 -1.40267670e-01 1.22701921e-01
-5.71470976e-01 5.25431395e-01 -3.71017903e-01 -4.98181015e-01
-7.32183084e-02 -8.83136451e-01 5.22058690e-03 2.69905210e-01
3.81458163e-01 -3.43957305e-01 1.29423118e+00 -5.98472536e-01
1.22445166e+00 -2.10523105e+00 -3.65300298e-01 4.05169934e-01
-1.09457569e-02 6.78813457e-01 -2.81804278e-02 -1.32484511e-01
-3.28614444e-01 -3.40505928e-01 1.25610858e-01 -3.34072769e-01
-3.59793395e-01 2.24429712e-01 1.02814212e-01 6.09498024e-01
4.84769523e-01 8.30580175e-01 -2.43427143e-01 -4.00422394e-01
3.56787533e-01 1.00026381e+00 -3.58497620e-01 2.55478084e-01
4.97536927e-01 2.09269255e-01 -2.95190543e-01 1.10054815e+00
9.18858767e-01 2.82492965e-01 -1.84966847e-01 -3.32982361e-01
-1.77940294e-01 -6.84945643e-01 -1.10551631e+00 1.27707076e+00
-5.23708344e-01 7.54738986e-01 3.87726009e-01 -9.14485335e-01
1.27605498e+00 4.15413499e-01 5.85571289e-01 -9.75022197e-01
8.01917315e-01 2.71704700e-02 -1.82009891e-01 -7.78674901e-01
3.36126208e-01 -1.32235602e-01 2.52485991e-01 4.66247126e-02
2.22450137e-01 3.39436799e-01 -1.44576848e-01 -3.97790134e-01
6.62024200e-01 -3.69085670e-02 1.82493672e-01 -1.53541774e-01
8.13321173e-01 -2.37650871e-01 5.46563506e-01 -3.24396007e-02
-2.70120919e-01 1.07091628e-01 4.17850077e-01 -9.81847763e-01
-7.68536091e-01 -3.62410367e-01 -7.39643127e-02 1.22278607e+00
-3.13890010e-01 -1.84378386e-01 -1.01136458e+00 -1.89256594e-01
-1.44826144e-01 5.48818745e-02 -9.34466302e-01 -1.07394516e-01
-3.81547302e-01 -7.89860070e-01 8.66329730e-01 7.13953793e-01
1.06392968e+00 -1.38227952e+00 -1.06253302e+00 -6.60364181e-02
4.07756180e-01 -1.08911693e+00 2.83055872e-01 -2.78786749e-01
-6.77155793e-01 -9.49767590e-01 -6.92525327e-01 -8.92439544e-01
6.56366229e-01 -1.55767044e-02 7.30860591e-01 1.46749884e-01
-7.56921411e-01 2.22139090e-01 -4.00060654e-01 -7.97322035e-01
5.66342995e-02 -3.90638560e-01 7.50479922e-02 5.82599163e-01
8.80776763e-01 -3.28311175e-01 -6.40670180e-01 5.39480075e-02
-9.68357801e-01 -8.94967392e-02 8.14115942e-01 8.59617054e-01
3.74810278e-01 -4.44248728e-02 3.27945977e-01 -6.38710320e-01
6.44437015e-01 -3.32751215e-01 -6.28532708e-01 -9.50443447e-02
-5.05909979e-01 -6.36260986e-01 5.36636651e-01 -3.28502327e-01
-1.16568089e+00 1.98842049e-01 -5.50566435e-01 -8.33704233e-01
-4.29403484e-01 4.27592039e-01 3.54489288e-03 -5.83960772e-01
4.87470061e-01 4.31972332e-02 5.96414804e-01 -3.59785780e-02
-1.26715630e-01 9.15207982e-01 2.94272274e-01 -8.72329623e-02
-8.89878199e-02 5.20512938e-01 2.84106433e-01 -1.19334674e+00
-3.84840816e-01 -3.79713237e-01 -5.56970835e-01 -6.69878781e-01
9.19856250e-01 -1.05776000e+00 -1.18797994e+00 1.05026031e+00
-1.01564753e+00 -3.82394195e-02 3.51397276e-01 5.16144753e-01
-3.30034643e-01 -2.18663029e-02 -9.80600595e-01 -1.14857602e+00
-7.76925445e-01 -9.44627404e-01 1.02073371e+00 6.66679442e-01
-1.55347288e-02 -6.77435756e-01 -4.19993728e-01 2.15269446e-01
5.85052133e-01 5.28871357e-01 5.31349957e-01 -2.38898601e-02
1.64471090e-01 -5.59827507e-01 -4.09402639e-01 5.34234166e-01
-1.80179343e-01 4.29179370e-01 -1.26875007e+00 -1.75671875e-01
-1.01918638e-01 -5.27024388e-01 4.31914359e-01 3.78009737e-01
1.48639035e+00 -1.98820859e-01 1.00044936e-01 6.94368303e-01
1.40340936e+00 5.00235200e-01 1.14315915e+00 3.20952326e-01
4.84657288e-01 7.08878160e-01 6.15489841e-01 8.05605054e-01
8.45824927e-02 6.04929149e-01 5.05321443e-01 -8.16951156e-01
1.05859146e-01 3.49990517e-01 2.96739310e-01 4.28788126e-01
-4.90874767e-01 2.67150939e-01 -7.35592723e-01 1.66806400e-01
-1.62567616e+00 -1.05852056e+00 -1.94558352e-01 1.69515967e+00
3.56647283e-01 -4.07486051e-01 1.62659556e-01 3.44323903e-01
4.38914180e-01 -2.08162352e-01 -2.07627594e-01 -1.10577333e+00
-2.27758735e-01 9.99889135e-01 1.21993467e-01 9.80219916e-02
-1.24390769e+00 1.09898376e+00 5.88521862e+00 7.95535505e-01
-1.90665805e+00 -4.01034532e-03 1.05999434e+00 1.13780499e-01
7.20226169e-01 -8.01234782e-01 -5.17710030e-01 2.48658508e-01
1.10373998e+00 3.08987439e-01 1.85025573e-01 1.14948082e+00
2.78833508e-01 -1.60170868e-01 -2.74685532e-01 1.39995265e+00
3.42342347e-01 -1.03030384e+00 1.04351506e-01 6.74297661e-02
3.67045730e-01 -4.12254959e-01 1.74434453e-01 5.38194716e-01
-4.03248936e-01 -1.42741072e+00 3.43255609e-01 5.92572510e-01
8.33466411e-01 -1.39307868e+00 1.31298244e+00 -1.73292994e-01
-1.07875001e+00 -2.88295090e-01 -4.24593389e-01 -5.37170291e-01
-3.14645320e-01 1.80641025e-01 -4.61452156e-01 2.93634951e-01
1.11344838e+00 4.69994336e-01 -3.76365155e-01 5.08802950e-01
1.20912977e-01 6.04844272e-01 -1.30100489e-01 -3.12524915e-01
4.36154872e-01 -3.45968425e-01 -2.70304859e-01 1.46613574e+00
3.41344148e-01 7.76784062e-01 -1.74745545e-01 3.25607717e-01
5.93522638e-02 4.72574294e-01 -4.74020511e-01 9.33097079e-02
-6.13958053e-02 1.86333656e+00 -4.65422541e-01 -3.32658678e-01
-5.77125967e-01 1.10842645e+00 1.64498225e-01 2.67324477e-01
-1.03548634e+00 -5.98486960e-01 9.50262189e-01 8.31121132e-02
1.55289516e-01 -1.54623628e-01 -1.42151028e-01 -6.76382422e-01
-1.54340416e-01 -6.07108176e-01 3.48343998e-01 -9.47429836e-01
-9.41945136e-01 8.92172873e-01 -3.32743019e-01 -9.60889161e-01
-1.98507935e-01 -1.13640857e+00 -5.45892954e-01 8.86993706e-01
-1.57676315e+00 -1.53542125e+00 -8.29848528e-01 9.66266453e-01
2.99434930e-01 -4.94358838e-01 1.16911888e+00 5.71508169e-01
-8.21863234e-01 8.79245579e-01 -3.93829674e-01 4.82237726e-01
3.23759049e-01 -5.59729517e-01 -3.20806295e-01 5.76618373e-01
-2.49662966e-01 4.38208252e-01 1.91096395e-01 -8.11079890e-02
-1.53825772e+00 -1.10367751e+00 6.98930740e-01 3.73645067e-01
2.86196887e-01 -2.35147163e-01 -7.66320646e-01 3.24928910e-01
2.87820607e-01 4.00279462e-01 1.04838073e+00 -2.08816677e-01
-2.40782842e-01 -3.36902171e-01 -1.38231385e+00 5.14434516e-01
5.75666130e-01 -2.84667283e-01 1.33176968e-01 -2.77675837e-01
-2.51201540e-01 -4.21546519e-01 -8.65670264e-01 6.34581685e-01
9.99735713e-01 -1.23215532e+00 5.14655769e-01 -7.32185781e-01
5.39644063e-01 -1.64088994e-01 -4.41308366e-03 -7.32691288e-01
-3.40953887e-01 -3.31208289e-01 -1.06107377e-01 9.65454340e-01
-6.18641376e-02 -3.65210772e-01 1.02754009e+00 6.59304738e-01
8.70324671e-02 -1.26975238e+00 -9.22886193e-01 -2.06557557e-01
-3.47466260e-01 -4.22777295e-01 4.13689137e-01 9.92674172e-01
-2.89840639e-01 3.35270047e-01 -3.29916000e-01 3.27942781e-02
1.89313501e-01 -6.54851571e-02 8.50587547e-01 -1.06243336e+00
4.31568414e-01 -5.04335821e-01 -1.15440142e+00 -2.33687043e-01
4.23923492e-01 -3.47957194e-01 -5.13248801e-01 -9.71544087e-01
-7.59993419e-02 -1.39793485e-01 -2.92164505e-01 8.23337138e-01
2.14564726e-01 9.34644282e-01 -3.82345952e-02 -2.66058922e-01
-1.63519323e-01 6.92398012e-01 9.04515803e-01 1.97893038e-01
-1.19423486e-01 -3.06816027e-02 -4.01976317e-01 7.82725096e-01
8.16272318e-01 1.69787750e-01 -2.07656428e-01 -1.61512360e-01
-2.72644550e-01 -1.13877490e-01 4.82663035e-01 -1.07223213e+00
1.58968613e-01 1.19478442e-01 1.04070413e+00 -1.92887068e-01
6.56779826e-01 -9.68672752e-01 1.53759092e-01 4.28102911e-01
5.24172634e-02 1.86720341e-01 6.40562832e-01 -9.65665132e-02
-5.12862086e-01 1.39332980e-01 1.07016504e+00 1.20195746e-01
-1.25165308e+00 4.77402329e-01 -6.96081936e-01 -1.00067139e+00
1.45330203e+00 -5.78854620e-01 1.40385553e-01 -3.56805682e-01
-7.08233833e-01 -2.61889160e-01 6.11777492e-02 4.62970436e-01
9.35767889e-01 -1.53565192e+00 -7.06546426e-01 6.56374753e-01
1.31643683e-01 -5.21592140e-01 4.35676306e-01 7.88140059e-01
-9.54716921e-01 1.71142787e-01 -9.86991107e-01 -4.38275278e-01
-1.79552245e+00 7.99861178e-03 3.98027033e-01 2.66109139e-01
-3.17360550e-01 9.56046879e-01 -1.57632664e-01 -7.94647168e-03
1.98785394e-01 1.32153258e-01 -6.16700053e-01 1.45044461e-01
8.51733506e-01 3.68223071e-01 3.65125179e-01 -1.11406684e+00
-3.09049368e-01 6.00322902e-01 7.09567890e-02 8.19460303e-02
1.20801497e+00 2.64771849e-01 -3.89410406e-01 -4.20039110e-02
1.47767711e+00 -4.17667657e-01 -8.00782204e-01 1.21208802e-01
-2.23098323e-01 -4.31918710e-01 4.19875562e-01 -6.45444930e-01
-1.42545187e+00 1.04673290e+00 1.20316184e+00 -2.50778854e-01
1.87895739e+00 -5.04584849e-01 6.00611031e-01 2.19554469e-01
2.73282349e-01 -1.17714083e+00 4.04395908e-02 4.81668532e-01
9.02543604e-01 -1.05341017e+00 -2.04277486e-01 -4.10580575e-01
-8.64983618e-01 1.57976115e+00 1.00375330e+00 -3.06057602e-01
8.11871827e-01 5.97208679e-01 3.62415165e-01 -3.41471434e-01
-4.98875052e-01 -2.15124920e-01 1.65351152e-01 3.29999804e-01
7.26760507e-01 -1.09109379e-01 -3.05783898e-01 7.30715394e-01
-1.17850058e-01 4.32258457e-01 4.50095162e-02 1.04396820e+00
-2.79633135e-01 -5.20878494e-01 -3.34440440e-01 4.05788809e-01
-7.83257961e-01 2.20122814e-01 -3.24583709e-01 9.35236812e-01
2.75886059e-01 7.59165406e-01 4.86498356e-01 -5.61611056e-01
3.38611960e-01 9.87089276e-02 3.97945195e-01 -1.91591419e-02
-7.72958875e-01 2.99490802e-02 -1.27315328e-01 -7.94930398e-01
-7.19216347e-01 -2.50355363e-01 -1.30772614e+00 -5.68217397e-01
-6.99517503e-02 -1.51383150e-02 9.82728422e-01 6.51677787e-01
5.88618994e-01 4.16411072e-01 8.27126145e-01 -9.60477114e-01
2.37916633e-01 -1.07800865e+00 -7.30876923e-01 2.52496064e-01
-5.91601580e-02 -7.47764945e-01 6.80965707e-02 2.27374449e-01] | [13.534934043884277, 1.775122880935669] |
f371429e-9ab6-43e1-8901-7cdc3c052dc8 | simplifying-graph-convolutional-networks | 1902.07153 | null | https://arxiv.org/abs/1902.07153v2 | https://arxiv.org/pdf/1902.07153v2.pdf | Simplifying Graph Convolutional Networks | Graph Convolutional Networks (GCNs) and their variants have experienced significant attention and have become the de facto methods for learning graph representations. GCNs derive inspiration primarily from recent deep learning approaches, and as a result, may inherit unnecessary complexity and redundant computation. In this paper, we reduce this excess complexity through successively removing nonlinearities and collapsing weight matrices between consecutive layers. We theoretically analyze the resulting linear model and show that it corresponds to a fixed low-pass filter followed by a linear classifier. Notably, our experimental evaluation demonstrates that these simplifications do not negatively impact accuracy in many downstream applications. Moreover, the resulting model scales to larger datasets, is naturally interpretable, and yields up to two orders of magnitude speedup over FastGCN. | ['Amauri Holanda de Souza Jr.', 'Kilian Q. Weinberger', 'Tianyi Zhang', 'Tao Yu', 'Felix Wu', 'Christopher Fifty'] | 2019-02-19 | null | null | null | null | ['node-classification-on-non-homophilic', 'graph-regression'] | ['graphs', 'graphs'] | [ 1.53645054e-01 4.86318827e-01 -1.24726519e-01 -1.78051978e-01
-3.00570309e-01 -7.10514605e-01 7.96674132e-01 6.15292430e-01
-3.07876796e-01 5.86518466e-01 8.36758912e-02 -7.15410650e-01
-2.84317791e-01 -9.34247732e-01 -7.40786195e-01 -6.35819197e-01
-3.72305781e-01 1.13596842e-02 1.94705427e-01 -2.88262367e-02
-1.56101296e-02 7.04867661e-01 -1.04551601e+00 2.08431453e-01
8.14852715e-01 8.53635550e-01 -2.08217159e-01 6.68037593e-01
-2.21410263e-02 8.39483202e-01 -4.49112147e-01 -7.90186584e-01
3.13965052e-01 -2.62843370e-01 -8.59376848e-01 -1.05454087e-01
4.64896083e-01 -1.51377201e-01 -7.67720222e-01 9.08076584e-01
2.59955466e-01 1.20141380e-01 4.99634206e-01 -1.22552359e+00
-6.35338604e-01 8.23870301e-01 -4.36802387e-01 1.02338187e-01
-1.80651456e-01 7.18899220e-02 1.39660478e+00 -5.00697434e-01
5.62312245e-01 1.16127908e+00 1.07800531e+00 2.78549910e-01
-1.55294895e+00 -2.84517676e-01 3.40224087e-01 -8.00862722e-03
-1.31332958e+00 -2.12031379e-01 4.30184871e-01 -3.16340834e-01
1.14604032e+00 2.36204535e-01 6.80896401e-01 8.21877003e-01
2.34099135e-01 7.12244987e-01 6.14500999e-01 -3.12811017e-01
9.35093611e-02 -2.85036504e-01 2.55377024e-01 9.93486524e-01
8.48749399e-01 -1.51851326e-01 -1.33828044e-01 -1.09218635e-01
8.06934357e-01 2.08076924e-01 -4.26282018e-01 -4.46413040e-01
-8.90268505e-01 9.04397309e-01 8.26046050e-01 1.53052986e-01
-7.02815950e-02 6.54315591e-01 6.37029707e-01 2.81657666e-01
4.30838823e-01 4.12241966e-01 -1.84861943e-01 1.15592852e-01
-5.67579687e-01 1.19103514e-01 8.16796422e-01 1.02752686e+00
7.48881102e-01 7.93880671e-02 7.74741033e-03 6.57082736e-01
-2.05471143e-01 6.11639544e-02 2.20816284e-01 -5.73914051e-01
3.69025677e-01 9.61738408e-01 -3.89626205e-01 -1.29631197e+00
-7.83074319e-01 -8.57064843e-01 -1.17266726e+00 6.19610548e-02
5.16622305e-01 -1.25481606e-01 -9.15713489e-01 1.67250955e+00
-7.98504651e-02 9.45985690e-02 -9.00228024e-02 6.02593839e-01
6.55910373e-01 4.34147358e-01 8.26074108e-02 1.84114918e-01
9.44168568e-01 -9.24189091e-01 -4.62073535e-01 -3.21087629e-01
1.01820278e+00 -2.76944906e-01 9.62593555e-01 3.40545416e-01
-8.67522001e-01 -2.40775257e-01 -1.12615311e+00 -3.67840648e-01
-5.03449321e-01 1.91853166e-01 1.14648819e+00 5.21574199e-01
-1.27338183e+00 1.14890814e+00 -1.09139574e+00 -3.33073199e-01
6.87638164e-01 5.17173111e-01 -4.22503740e-01 -9.33009386e-02
-9.21870470e-01 5.44959486e-01 5.01622915e-01 3.51392746e-01
-5.75624168e-01 -6.85023129e-01 -9.05334890e-01 5.92832863e-01
3.62332135e-01 -6.52578413e-01 1.07411683e+00 -7.94530749e-01
-1.16313899e+00 7.01206625e-01 -2.61309408e-02 -7.61245072e-01
4.07749861e-01 -1.79297000e-01 -2.26006970e-01 1.28045693e-01
-2.41798371e-01 4.02787358e-01 5.49277723e-01 -8.26132178e-01
-4.25716311e-01 -1.61682084e-01 4.93144512e-01 -2.32057855e-01
-7.46899843e-01 -4.39287364e-01 -5.95667958e-01 -7.30859339e-01
1.87737271e-01 -1.06795156e+00 -4.43214774e-01 2.24253669e-01
-5.10181665e-01 -5.58272563e-02 4.73974168e-01 -5.03479600e-01
1.47326624e+00 -2.16503119e+00 1.02675669e-01 3.15843791e-01
8.30146909e-01 3.42063338e-01 -7.84808919e-02 6.09339535e-01
-1.61533996e-01 3.46762389e-01 -3.81839842e-01 -1.45967796e-01
-1.54596167e-02 3.58013064e-01 -2.14581490e-01 5.46823442e-01
5.17558038e-01 1.29123318e+00 -9.80585575e-01 -1.57375962e-01
1.08393587e-01 4.51918334e-01 -6.15598977e-01 -1.12099834e-01
-1.30092382e-01 -1.29382282e-01 -2.95470983e-01 2.29987338e-01
5.37928998e-01 -8.22129250e-01 7.44417012e-01 -2.33642578e-01
1.71043888e-01 4.38959956e-01 -8.69086862e-01 1.35935259e+00
-4.66207176e-01 1.00871372e+00 2.43840907e-02 -1.47656095e+00
7.93274820e-01 5.11080201e-04 2.44549751e-01 -5.01862526e-01
2.39882037e-01 9.71369967e-02 2.16255546e-01 -9.49020833e-02
2.16239393e-01 4.65249568e-02 5.43109961e-02 2.48356849e-01
4.54873033e-02 2.14873463e-01 4.30032045e-01 3.52723658e-01
1.41206837e+00 -2.94017922e-02 4.34832245e-01 -4.76420820e-01
3.89044970e-01 -1.08420417e-01 5.75426221e-01 6.76007986e-01
-7.78627098e-02 4.74535197e-01 1.21368003e+00 -7.23842204e-01
-8.67989242e-01 -8.55054736e-01 1.40582129e-01 7.96243727e-01
2.68497020e-02 -7.29430318e-01 -7.73217082e-01 -7.14605689e-01
1.46182284e-01 2.49202982e-01 -6.81260467e-01 -4.93653059e-01
-8.65052104e-01 -8.89738798e-01 7.08188772e-01 8.47636521e-01
3.13360721e-01 -6.78330421e-01 -2.46776298e-01 3.12028289e-01
2.49928758e-01 -1.13951004e+00 -1.92133203e-01 3.75775784e-01
-1.07351661e+00 -1.27607155e+00 -4.50682849e-01 -8.13090920e-01
7.42504060e-01 3.51070374e-01 1.28394115e+00 4.81092095e-01
-2.99890220e-01 -1.55222386e-01 -2.12362379e-01 -1.23196244e-01
-3.05463970e-01 3.18940014e-01 -2.77981788e-01 -3.67952958e-02
1.78454652e-01 -8.04581523e-01 -5.48540831e-01 -2.70696878e-01
-8.57436299e-01 1.33134350e-01 6.90025687e-01 9.29606736e-01
3.43127012e-01 9.72625092e-02 4.99763072e-01 -1.38661277e+00
6.99298024e-01 -1.92603305e-01 -9.13613498e-01 1.77515283e-01
-7.66643107e-01 2.49913454e-01 1.15130723e+00 -6.06926344e-02
-6.00427151e-01 6.87706620e-02 -6.44304417e-03 -4.27403539e-01
1.05279863e-01 7.10766494e-01 8.88193846e-02 -1.89641207e-01
6.26596749e-01 -4.84567741e-03 1.66189913e-02 -3.80416989e-01
5.63641846e-01 1.23860188e-01 3.46806884e-01 -3.35511565e-01
8.52894247e-01 5.56382537e-01 3.82490426e-01 -7.47507870e-01
-7.35565007e-01 -8.04574564e-02 -5.42796075e-01 -2.32020440e-03
6.14369512e-01 -8.42375398e-01 -1.03681958e+00 2.06766665e-01
-1.08298624e+00 -4.71294880e-01 6.22808840e-03 3.05453360e-01
-1.50586516e-01 4.06379074e-01 -1.02674413e+00 -6.01387441e-01
-3.86442542e-01 -9.60345030e-01 7.63266742e-01 9.30695683e-02
-1.11335754e-01 -1.24789429e+00 -3.84844691e-01 -2.59238183e-01
3.17362666e-01 4.52707559e-01 1.33236444e+00 -5.17194450e-01
-6.74914956e-01 -2.52381325e-01 -6.12115085e-01 2.65686125e-01
7.88112208e-02 2.73670275e-02 -9.04535532e-01 -4.95721072e-01
-6.76965296e-01 1.97873861e-02 1.23263407e+00 3.08204561e-01
1.41647470e+00 -3.42927605e-01 -5.29318631e-01 8.72422397e-01
1.56723249e+00 -1.48697898e-01 4.60861117e-01 2.21632328e-03
1.01334488e+00 4.43968564e-01 -1.57799825e-01 6.23256415e-02
-2.63309409e-03 3.94935340e-01 5.10918200e-01 -3.94672245e-01
-2.19355255e-01 -2.79841334e-01 1.21538021e-01 9.27226186e-01
-1.67042851e-01 -5.46974778e-01 -9.63607907e-01 5.02019286e-01
-1.97541916e+00 -6.76506400e-01 -4.41736996e-01 2.08978724e+00
3.89019251e-01 3.42950493e-01 -5.83654232e-02 2.49712989e-01
5.20202875e-01 1.01822995e-01 -4.97322381e-01 -4.81895506e-01
-1.96822912e-01 5.80013990e-01 8.62297237e-01 4.26094264e-01
-1.07723570e+00 1.09007835e+00 6.83928347e+00 6.50617301e-01
-1.14413548e+00 -2.27530316e-01 6.79630935e-01 1.55028114e-02
-3.33411425e-01 9.63920429e-02 -3.95487010e-01 6.47528917e-02
9.16272700e-01 -3.51333618e-01 4.07703131e-01 8.27545345e-01
-3.76125192e-03 3.80287975e-01 -1.20921218e+00 7.33343720e-01
-4.09509808e-01 -1.64245439e+00 2.83354104e-01 1.94172025e-01
5.65122664e-01 7.12936744e-02 -1.81374684e-01 2.66423047e-01
6.00092173e-01 -1.23967135e+00 4.77126658e-01 1.57158852e-01
6.59296513e-01 -8.08703482e-01 6.81243718e-01 -7.09425542e-04
-1.35357988e+00 -2.34613881e-01 -6.43097997e-01 -3.48888487e-01
4.67032269e-02 8.32647920e-01 -6.63958728e-01 7.28337884e-01
5.57767987e-01 8.95299673e-01 -6.23057663e-01 9.22588944e-01
-2.89695084e-01 8.35503995e-01 -1.80210128e-01 -2.87080109e-01
4.83615994e-01 -3.17251772e-01 2.31471136e-01 1.40769184e+00
1.41638368e-01 -7.92978704e-02 -7.85661209e-03 8.50108862e-01
-4.06423897e-01 5.50146122e-03 -7.35672414e-01 -5.27218759e-01
2.67515928e-01 1.28407979e+00 -9.81401145e-01 -3.58338177e-01
-7.24060416e-01 8.00159693e-01 9.30699050e-01 4.71648157e-01
-7.43749022e-01 -7.06061780e-01 9.57934141e-01 2.91745998e-02
4.13211554e-01 -3.13493639e-01 -4.73295003e-01 -1.07919025e+00
4.78382409e-02 -5.61026454e-01 4.71278310e-01 -2.28115290e-01
-1.19469309e+00 6.25993729e-01 -4.09546345e-01 -1.00821221e+00
-3.67181711e-02 -1.05500340e+00 -4.93926048e-01 7.05341160e-01
-1.53635669e+00 -9.20541346e-01 -3.28949720e-01 2.10280508e-01
5.22545539e-02 2.18361661e-01 8.37282181e-01 4.67274338e-01
-8.29540849e-01 8.41484249e-01 1.52917832e-01 4.68888342e-01
1.14034176e-01 -1.35224545e+00 8.07147682e-01 1.03138888e+00
1.73315585e-01 9.65694010e-01 4.97372240e-01 -4.27305251e-01
-1.49255335e+00 -1.33757985e+00 7.25554287e-01 2.36347807e-03
8.64988506e-01 -7.59640276e-01 -1.04689121e+00 8.66941929e-01
1.38730658e-02 2.05394641e-01 4.27859783e-01 4.54494059e-01
-5.82768738e-01 -2.26671636e-01 -6.31529689e-01 8.21766973e-01
1.47968984e+00 -4.66754735e-01 7.68998265e-02 4.75579262e-01
7.86103249e-01 -3.17302793e-01 -8.24871480e-01 3.60828876e-01
5.36401868e-01 -9.85710680e-01 8.26790392e-01 -9.31221485e-01
5.01461685e-01 -6.22279495e-02 1.98500678e-01 -1.19573951e+00
-6.07334435e-01 -8.50484252e-01 -1.81643859e-01 8.43176186e-01
6.63772404e-01 -7.79499829e-01 9.30038750e-01 4.76834327e-01
-3.99599582e-01 -1.06213701e+00 -4.85770732e-01 -1.03188944e+00
1.99515745e-01 -2.65221506e-01 5.91647685e-01 9.87441599e-01
2.75223076e-01 3.63489628e-01 -1.54006198e-01 1.55569464e-01
5.07036686e-01 3.07112336e-01 7.74037063e-01 -1.27539384e+00
-4.78172958e-01 -8.94998908e-01 -6.13569677e-01 -1.28682566e+00
2.23802030e-01 -1.33403587e+00 -2.41530865e-01 -1.65361989e+00
3.10433619e-02 -4.42442775e-01 -4.45445269e-01 7.95680046e-01
-2.98367471e-01 3.40632677e-01 -5.67793809e-02 -9.90620777e-02
-2.08288774e-01 2.56063938e-01 9.49079931e-01 -1.10145912e-01
1.26247294e-02 -1.83361769e-01 -9.91311789e-01 6.11275852e-01
9.89603639e-01 -2.68385708e-01 -4.85345811e-01 -5.63038886e-01
5.14153004e-01 -3.58450770e-01 4.16888684e-01 -8.46588254e-01
7.12390319e-02 4.84260358e-02 2.12127388e-01 2.06728354e-02
-8.39233678e-03 -5.96598268e-01 1.12823546e-01 6.41378462e-01
-3.76313180e-01 9.89313945e-02 4.43613917e-01 6.85727775e-01
-1.37253240e-01 5.96435256e-02 7.36029088e-01 -4.07747878e-03
-4.00470167e-01 4.57225233e-01 -2.98823684e-01 -1.38722837e-01
6.79705739e-01 -1.09232038e-01 -4.45587873e-01 -4.65364814e-01
-5.47877312e-01 1.12938397e-01 2.62558401e-01 2.45856732e-01
2.63922364e-01 -1.24387181e+00 -5.30196011e-01 1.18882991e-01
-1.08137421e-01 1.10340714e-01 8.16951785e-03 8.59892309e-01
-8.41754556e-01 6.66470885e-01 8.32466707e-02 -4.50048596e-01
-9.79838192e-01 6.31604314e-01 3.31297457e-01 -4.27602500e-01
-1.04525387e+00 8.86499465e-01 4.27334219e-01 -3.03471893e-01
1.43252000e-01 -6.55709147e-01 5.08824922e-02 -2.30595365e-01
2.72172630e-01 3.96146804e-01 3.30753893e-01 -1.87827483e-01
-2.51497626e-01 2.53048420e-01 -1.26191229e-01 4.69893336e-01
1.57090068e+00 1.30125716e-01 -2.31339112e-01 1.69637188e-01
1.41473162e+00 -5.86776212e-02 -1.18337679e+00 -1.02638766e-01
2.62002528e-01 -1.07211754e-01 -5.29902764e-02 -2.79825330e-01
-1.38271284e+00 1.08960748e+00 -3.44309285e-02 6.10393882e-01
1.10113275e+00 -8.28988254e-02 8.01136851e-01 5.47679961e-01
9.89814624e-02 -6.40881896e-01 -2.54958034e-01 5.30589640e-01
5.87836325e-01 -8.18840742e-01 1.80448070e-01 -8.41342747e-01
5.30420579e-02 1.23764801e+00 4.16470021e-01 -2.92057604e-01
5.44065893e-01 3.14324141e-01 -2.08375350e-01 -2.47646049e-01
-8.15376103e-01 -2.04385519e-01 1.18525729e-01 5.65701425e-01
6.80253565e-01 1.27105430e-01 -5.53053617e-01 4.60365295e-01
-2.54894018e-01 -1.87684685e-01 5.10759056e-01 6.97023809e-01
-2.51923472e-01 -1.07303441e+00 3.57525706e-01 6.70833945e-01
-3.28267723e-01 -3.68824571e-01 -4.60441351e-01 9.72352922e-01
-3.70296746e-01 7.53810108e-01 8.78372416e-02 -3.71019572e-01
2.85451025e-01 -8.88012424e-02 3.89495432e-01 -4.72213537e-01
-5.75889170e-01 -1.16628535e-01 1.55529410e-01 -7.68327713e-01
3.19259726e-02 -2.54437715e-01 -1.42184341e+00 -5.83292961e-01
-3.76230061e-01 -2.05096193e-02 1.73845574e-01 7.32399046e-01
8.23587418e-01 7.88445115e-01 2.80306429e-01 -5.69119394e-01
-4.73076224e-01 -8.05935562e-01 -4.33195621e-01 3.21450800e-01
2.66964138e-01 -4.45311785e-01 -3.50908697e-01 -1.50515944e-01] | [6.901280879974365, 6.05600118637085] |
21c1d2d9-e326-408a-b41b-429d54b78039 | netsentry-a-deep-learning-approach-to | 2202.09873 | null | https://arxiv.org/abs/2202.09873v2 | https://arxiv.org/pdf/2202.09873v2.pdf | NetSentry: A Deep Learning Approach to Detecting Incipient Large-scale Network Attacks | Machine Learning (ML) techniques are increasingly adopted to tackle ever-evolving high-profile network attacks, including DDoS, botnet, and ransomware, due to their unique ability to extract complex patterns hidden in data streams. These approaches are however routinely validated with data collected in the same environment, and their performance degrades when deployed in different network topologies and/or applied on previously unseen traffic, as we uncover. This suggests malicious/benign behaviors are largely learned superficially and ML-based Network Intrusion Detection System (NIDS) need revisiting, to be effective in practice. In this paper we dive into the mechanics of large-scale network attacks, with a view to understanding how to use ML for Network Intrusion Detection (NID) in a principled way. We reveal that, although cyberattacks vary significantly in terms of payloads, vectors and targets, their early stages, which are critical to successful attack outcomes, share many similarities and exhibit important temporal correlations. Therefore, we treat NID as a time-sensitive task and propose NetSentry, perhaps the first of its kind NIDS that builds on Bidirectional Asymmetric LSTM (Bi-ALSTM), an original ensemble of sequential neural models, to detect network threats before they spread. We cross-evaluate NetSentry using two practical datasets, training on one and testing on the other, and demonstrate F1 score gains above 33% over the state-of-the-art, as well as up to 3 times higher rates of detecting attacks such as XSS and web bruteforce. Further, we put forward a novel data augmentation technique that boosts the generalization abilities of a broad range of supervised deep learning algorithms, leading to average F1 score gains above 35%. | ['Paul Patras', 'Haoyu Liu'] | 2022-02-20 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [ 1.62433863e-01 -7.19751418e-01 -3.44169587e-01 -1.15835175e-01
-9.61895287e-02 -9.44578826e-01 8.63865554e-01 2.60449667e-02
-3.80317450e-01 5.28862834e-01 -2.05391124e-01 -9.96163428e-01
-3.49219084e-01 -8.21475029e-01 -4.90773320e-01 -5.20973146e-01
-9.21066940e-01 6.02331460e-01 5.94885468e-01 -2.93504983e-01
1.97243407e-01 9.85483944e-01 -9.48735297e-01 4.24335808e-01
5.69112115e-02 1.04872596e+00 -7.98147798e-01 8.01462054e-01
-7.36182705e-02 9.96898949e-01 -1.01687193e+00 -5.70676804e-01
3.78195763e-01 -4.64919768e-02 -7.95813680e-01 -5.05501926e-01
1.58721417e-01 -5.48393369e-01 -7.29619145e-01 7.92413116e-01
3.26409191e-01 -2.25287542e-01 1.77117705e-01 -1.61075890e+00
-1.78156182e-01 6.80851281e-01 -4.87626821e-01 8.59907627e-01
1.28487155e-01 6.38041437e-01 8.31261575e-01 -1.82096407e-01
6.03894114e-01 1.18705738e+00 8.72960150e-01 5.23015976e-01
-1.31872654e+00 -9.44692075e-01 1.50084227e-01 2.91042596e-01
-7.01408565e-01 -3.77628893e-01 6.48423433e-01 -2.58231401e-01
1.10402274e+00 1.84804469e-01 2.94152498e-01 2.17160439e+00
1.75970674e-01 4.90634292e-01 8.01409543e-01 -5.07468209e-02
2.24843502e-01 -4.71297503e-02 1.41894609e-01 3.61909270e-01
4.42246318e-01 6.01494670e-01 -2.22133607e-01 -5.15335262e-01
5.18990040e-01 4.98918951e-01 1.47229172e-02 9.96334851e-02
-9.72641051e-01 1.09974182e+00 4.56475914e-01 5.81802249e-01
-4.68465894e-01 5.88809513e-02 9.82548654e-01 6.22309566e-01
2.58669376e-01 6.12056792e-01 -8.85280311e-01 -3.85505974e-01
-6.07628405e-01 -1.77570898e-02 1.21027350e+00 1.15916148e-01
3.56232077e-01 5.66976309e-01 3.00668001e-01 3.93096566e-01
2.39683818e-02 3.95661622e-01 3.06240499e-01 -4.06738192e-01
3.18114132e-01 5.02226233e-01 -4.88213837e-01 -1.05419552e+00
-6.29276931e-01 -6.39933944e-01 -7.76094198e-01 2.41697952e-01
5.30825436e-01 -5.16046107e-01 -9.25623477e-01 1.83822656e+00
1.93663940e-01 6.05027974e-01 -1.93836644e-01 2.87271053e-01
8.99858922e-02 4.77315396e-01 3.55039209e-01 -2.28793845e-02
1.05749357e+00 -2.69550890e-01 -2.82456219e-01 -2.77701437e-01
6.51178122e-01 -3.08574021e-01 5.66025436e-01 4.81995255e-01
-4.58313346e-01 -4.22030408e-03 -9.30746973e-01 9.44705307e-01
-9.12976801e-01 -8.57024968e-01 9.27441955e-01 1.03587055e+00
-7.05021799e-01 9.09656048e-01 -7.81054974e-01 -5.34407079e-01
6.88112378e-01 3.94763857e-01 -1.04937643e-01 2.30375573e-01
-1.51230323e+00 7.21666157e-01 4.47131753e-01 -1.84170574e-01
-1.26430690e+00 -9.03567433e-01 -3.30726743e-01 1.03133179e-01
5.76499820e-01 -1.26943380e-01 9.73162174e-01 -4.14916635e-01
-1.34276998e+00 4.23331916e-01 3.45034242e-01 -8.96326602e-01
2.75040060e-01 -4.89146374e-02 -1.08231366e+00 1.42154247e-01
-2.03897312e-01 -5.35175577e-02 8.99456739e-01 -1.17144895e+00
-6.33608639e-01 -3.59476775e-01 1.45606205e-01 -8.01025748e-01
-6.68698311e-01 6.09067380e-01 3.22232157e-01 -5.10641336e-01
-4.88446265e-01 -5.93940914e-01 -2.15182632e-01 -5.89726508e-01
-5.58414936e-01 -1.18185118e-01 1.62828362e+00 -5.53603053e-01
1.34427428e+00 -1.71217799e+00 -2.96596348e-01 6.49038732e-01
4.93174434e-01 1.01299143e+00 -3.09080005e-01 7.96152651e-01
-2.66436130e-01 5.65130711e-01 -4.42812452e-03 1.33888628e-02
1.13941662e-01 1.93955839e-01 -9.90194857e-01 2.55366325e-01
3.36045802e-01 9.71764147e-01 -7.75455832e-01 2.35488668e-01
1.56468108e-01 3.37005645e-01 -3.68373752e-01 2.71689087e-01
-3.04542035e-01 3.72355253e-01 -4.09634799e-01 8.92358303e-01
3.48410189e-01 -4.10486430e-01 2.07187757e-01 7.97537565e-02
1.15821488e-01 3.99620742e-01 -6.52890861e-01 7.60443330e-01
-3.49061728e-01 6.58743322e-01 4.83873151e-02 -1.21425259e+00
7.65224159e-01 4.24869657e-01 7.12694407e-01 -1.02606535e+00
3.20329428e-01 3.22104931e-01 5.39336443e-01 -4.04133797e-01
-4.08790618e-01 5.61315231e-02 4.79312278e-02 1.01635599e+00
1.98226511e-01 6.55278921e-01 3.93827856e-01 3.20038766e-01
1.87840867e+00 -6.64239883e-01 9.87820253e-02 2.08073661e-01
4.10372943e-01 -9.86478180e-02 4.80548799e-01 1.11881411e+00
-4.40301508e-01 -4.00457352e-01 8.44592929e-01 -9.64016199e-01
-1.03499341e+00 -1.14774001e+00 -1.01518780e-01 1.24047184e+00
-3.61494958e-01 -2.56640464e-01 -4.58534837e-01 -1.21003234e+00
6.11021705e-02 5.40881813e-01 -5.23030519e-01 -3.85959566e-01
-1.05374265e+00 -1.13652766e+00 1.28706884e+00 3.68639410e-01
4.52308685e-01 -1.29561663e+00 -4.19833302e-01 5.76669276e-01
4.20934737e-01 -1.59313393e+00 1.40745208e-01 3.89966816e-01
-6.20480776e-01 -1.40380847e+00 1.78248972e-01 -1.09925203e-01
7.48905391e-02 2.31083427e-02 9.93971884e-01 1.46827966e-01
-6.16870940e-01 3.46192151e-01 -3.27171057e-01 -5.08941948e-01
-5.22088945e-01 2.17544690e-01 5.34365177e-01 1.99102402e-01
6.94723427e-01 -1.21989954e+00 -2.71903098e-01 3.68353337e-01
-1.06346798e+00 -1.01071155e+00 7.89144874e-01 6.46634281e-01
-4.45526600e-01 3.29443365e-01 9.85257804e-01 -9.93747830e-01
8.49378169e-01 -9.58537042e-01 -5.11745334e-01 1.32274637e-02
-5.96154094e-01 -1.93764865e-01 1.13832104e+00 -9.96095121e-01
-6.03971362e-01 -7.98907340e-01 -2.04762176e-01 -5.35280704e-01
-4.75742012e-01 2.62409031e-01 7.15509728e-02 -4.14735883e-01
9.14316535e-01 2.21783653e-01 -1.08367480e-01 -3.36848766e-01
5.51221073e-02 3.87527496e-01 2.68908381e-01 -6.75226092e-01
1.42653978e+00 5.52952111e-01 1.76356256e-01 -9.36582923e-01
-6.84509039e-01 -2.01353908e-01 -3.93063337e-01 -1.71102375e-01
3.30258816e-01 -1.54176056e-01 -1.25557077e+00 6.79558516e-01
-1.16930497e+00 -3.48907501e-01 -6.74802586e-02 1.95770353e-01
1.76453218e-01 3.95766199e-01 -1.10932267e+00 -7.21589684e-01
-4.67477620e-01 -8.69162679e-01 3.38067859e-01 3.56976241e-02
-2.02753887e-01 -1.39296746e+00 4.02873993e-01 6.20932356e-02
1.10531628e+00 4.81741488e-01 1.14288700e+00 -1.80179036e+00
-4.50003028e-01 -5.10760307e-01 -4.67290401e-01 3.46705347e-01
-1.59742720e-02 1.51304260e-01 -8.99465561e-01 -5.33470511e-01
1.38119847e-01 -2.52080590e-01 6.26283824e-01 -8.97758305e-02
9.75193501e-01 -7.75712430e-01 -4.45137888e-01 6.35967374e-01
1.26441956e+00 6.90150917e-01 5.36854148e-01 5.55836439e-01
6.85548604e-01 5.43541670e-01 -3.39858741e-01 4.50420380e-01
-1.87356114e-01 4.63863790e-01 7.45659351e-01 -3.62576800e-03
2.65829504e-01 -2.41068214e-01 6.08211100e-01 3.88581902e-01
1.92738160e-01 -4.33901995e-01 -1.08808577e+00 1.43955156e-01
-1.24234748e+00 -1.36792600e+00 7.24921450e-02 2.02750301e+00
4.94671136e-01 7.53048062e-01 5.92538416e-01 3.12242329e-01
5.44672072e-01 5.21517456e-01 -7.24187434e-01 -6.48956060e-01
2.56095850e-03 7.38765001e-01 4.56416517e-01 1.92105964e-01
-1.20256376e+00 9.20059144e-01 6.49670506e+00 8.20590377e-01
-1.62546682e+00 4.65131029e-02 5.34938276e-01 -9.52302217e-02
2.35857248e-01 4.50040288e-02 -7.17576921e-01 6.06330156e-01
1.72115779e+00 4.59561385e-02 7.62627661e-01 7.12223113e-01
3.76988947e-02 6.19464397e-01 -9.73752499e-01 5.28226495e-01
-1.46914631e-01 -1.44833291e+00 1.56774133e-01 4.67846274e-01
3.38105589e-01 3.76382262e-01 2.51900494e-01 5.85760593e-01
7.04919875e-01 -1.17202294e+00 -8.22363794e-03 -1.04961656e-01
3.93782467e-01 -6.78975821e-01 6.63040996e-01 2.19137743e-01
-8.84451509e-01 -6.19358182e-01 5.30027300e-02 6.08295240e-02
2.39116371e-01 6.38023376e-01 -1.00585520e+00 3.15289646e-01
6.00399792e-01 7.03249037e-01 -5.46253741e-01 7.22689629e-01
-6.61912933e-02 1.12674427e+00 -4.49907243e-01 -5.95791116e-02
6.41159654e-01 2.23350018e-01 8.59312296e-01 1.24182475e+00
-1.14800945e-01 -3.33669662e-01 3.07673514e-01 7.35489368e-01
-1.28027365e-01 -3.97467852e-01 -7.74070919e-01 -5.14724493e-01
6.27148509e-01 1.29592288e+00 -5.32334745e-01 -1.24682717e-01
-1.75962597e-01 4.24887598e-01 5.11398315e-02 6.28231645e-01
-9.84089255e-01 -4.91794974e-01 1.10733199e+00 2.02268362e-01
1.12548389e-01 -2.04202473e-01 -1.99796960e-01 -1.14642274e+00
-1.84051558e-01 -1.21348739e+00 6.78961039e-01 1.42703384e-01
-1.66555643e+00 9.18286681e-01 -2.28512257e-01 -1.00741148e+00
-4.75642473e-01 -9.47202206e-01 -1.25527930e+00 2.69541144e-01
-1.34791100e+00 -1.01189148e+00 1.68097854e-01 6.43129289e-01
2.42424160e-01 -5.81474185e-01 6.68592036e-01 6.76766396e-01
-1.17554700e+00 5.84862709e-01 -7.96317235e-02 6.85256898e-01
4.15202141e-01 -8.09073389e-01 7.68879771e-01 1.18723142e+00
3.04792047e-01 7.68331945e-01 4.59755272e-01 -6.84024036e-01
-1.27091324e+00 -1.08826005e+00 3.44396770e-01 -5.98630071e-01
1.47336662e+00 -5.53225517e-01 -1.03732836e+00 9.64787245e-01
-3.96170616e-02 -3.02875624e-03 6.40577972e-01 1.69502780e-01
-1.04189932e+00 -1.27217948e-01 -1.33898866e+00 7.72544026e-01
1.15723252e+00 -5.95163882e-01 -1.90921903e-01 3.61432552e-01
8.91185403e-01 3.28908414e-01 -8.11659515e-01 6.82759047e-01
3.29698116e-01 -1.13627303e+00 1.28114212e+00 -1.44944990e+00
-1.07183330e-01 1.40028924e-01 -1.02493614e-01 -1.00443459e+00
-1.97072044e-01 -9.53195632e-01 -7.00850546e-01 1.23478878e+00
3.14015329e-01 -1.35760343e+00 7.99185395e-01 -1.11019367e-03
3.40484381e-01 -8.34997833e-01 -9.73835409e-01 -9.94632125e-01
-6.53964356e-02 -6.36679769e-01 5.99832475e-01 1.20780647e+00
-1.01427786e-01 2.12400898e-01 -4.05287743e-01 2.77717441e-01
8.48741353e-01 -3.78500998e-01 6.80578291e-01 -1.53204620e+00
-1.83952272e-01 -9.07023191e-01 -6.18272126e-01 -4.12862748e-01
2.48076692e-01 -5.69013238e-01 -7.97133148e-01 -5.19418180e-01
-2.72795856e-01 -4.14692700e-01 -4.97736573e-01 6.94381714e-01
3.97445291e-01 2.73127079e-01 3.66958231e-02 1.73123613e-01
-4.73005205e-01 6.89857751e-02 3.73645723e-01 -1.91133559e-01
-1.94367990e-01 7.89722651e-02 -4.84629333e-01 7.46075571e-01
1.01205599e+00 -5.74352086e-01 -1.48799181e-01 -4.61978763e-02
1.32842762e-02 -2.22226650e-01 5.57731986e-01 -1.11338186e+00
1.95164934e-01 -1.81179971e-01 4.26789016e-01 -2.18561739e-01
1.77780792e-01 -6.45072579e-01 -1.29905269e-01 8.46136391e-01
-1.15712687e-01 3.50164890e-01 3.55439365e-01 5.53312182e-01
8.79245996e-02 2.06764072e-01 7.36925185e-01 -4.44874051e-04
-5.52897573e-01 8.67162406e-01 -5.41418970e-01 3.07640791e-01
1.12971151e+00 1.23676173e-02 -8.55717540e-01 -3.55409145e-01
-5.03253698e-01 -7.27253482e-02 -3.44680920e-02 6.13106132e-01
5.45173526e-01 -9.97937560e-01 -6.17018998e-01 4.63372171e-01
-2.55093277e-01 -8.41493964e-01 1.72569707e-01 8.18602979e-01
-2.87472188e-01 6.02121234e-01 -4.39217329e-01 -4.70568627e-01
-8.90514493e-01 7.65830636e-01 4.22111481e-01 -8.05951476e-01
-5.30117214e-01 4.29548025e-01 -3.54760885e-01 -5.81664681e-01
3.30897540e-01 2.17165425e-01 -5.93945868e-02 5.49754966e-03
7.92610049e-01 5.06298900e-01 3.38928327e-02 -2.40613967e-01
-5.67017555e-01 1.03733249e-01 -5.06832540e-01 1.82013482e-01
1.51603913e+00 6.22990131e-01 -3.47514927e-01 5.86363534e-03
1.39559948e+00 -1.64073363e-01 -7.99697757e-01 -4.90424573e-01
5.70142329e-01 -4.39541787e-01 -3.45979273e-01 -9.16594565e-01
-1.23396230e+00 8.60584140e-01 3.64035398e-01 8.70870411e-01
9.54556286e-01 -1.70709908e-01 1.29280376e+00 5.73096752e-01
3.35028261e-01 -3.79524559e-01 4.78421807e-01 8.52369845e-01
8.84681940e-02 -9.28893685e-01 -3.68587017e-01 1.45555049e-01
-1.76122144e-01 1.25948930e+00 6.13876104e-01 -2.50027746e-01
7.16943383e-01 5.57537675e-01 -7.20426440e-02 -4.77912188e-01
-1.13028097e+00 3.09726018e-02 -3.21476370e-01 6.77933812e-01
-2.73938447e-01 -1.75649852e-01 3.97964001e-01 1.08571596e-01
1.50903508e-01 -5.22089601e-01 5.07253766e-01 8.23343456e-01
-3.22934359e-01 -1.21710205e+00 -2.42515907e-01 6.57753766e-01
-7.61615038e-01 5.73250726e-02 -5.35160124e-01 8.83380890e-01
-1.93031728e-01 9.51971829e-01 -1.27047926e-01 -1.06879485e+00
2.06602141e-01 2.17375591e-01 -1.39651895e-01 -2.52856970e-01
-9.16120708e-01 -5.64590216e-01 8.48343298e-02 -8.71465027e-01
2.49496147e-01 -4.38217461e-01 -7.88957596e-01 -9.25074339e-01
3.44239995e-02 2.60487366e-02 6.20210052e-01 1.14792478e+00
4.60548550e-01 5.96565664e-01 1.04088938e+00 -7.09016800e-01
-8.89520943e-01 -8.65461290e-01 -2.54935920e-01 5.22239029e-01
4.52647984e-01 -6.54506683e-01 -9.23846304e-01 -7.16768622e-01] | [5.349460124969482, 7.3437604904174805] |
6cb0b512-b41f-4104-8711-6c9097df41b3 | look-into-person-joint-body-parsing-pose | 1804.01984 | null | http://arxiv.org/abs/1804.01984v1 | http://arxiv.org/pdf/1804.01984v1.pdf | Look into Person: Joint Body Parsing & Pose Estimation Network and A New Benchmark | Human parsing and pose estimation have recently received considerable
interest due to their substantial application potentials. However, the existing
datasets have limited numbers of images and annotations and lack a variety of
human appearances and coverage of challenging cases in unconstrained
environments. In this paper, we introduce a new benchmark named "Look into
Person (LIP)" that provides a significant advancement in terms of scalability,
diversity, and difficulty, which are crucial for future developments in
human-centric analysis. This comprehensive dataset contains over 50,000
elaborately annotated images with 19 semantic part labels and 16 body joints,
which are captured from a broad range of viewpoints, occlusions, and background
complexities. Using these rich annotations, we perform detailed analyses of the
leading human parsing and pose estimation approaches, thereby obtaining
insights into the successes and failures of these methods. To further explore
and take advantage of the semantic correlation of these two tasks, we propose a
novel joint human parsing and pose estimation network to explore efficient
context modeling, which can simultaneously predict parsing and pose with
extremely high quality. Furthermore, we simplify the network to solve human
parsing by exploring a novel self-supervised structure-sensitive learning
approach, which imposes human pose structures into the parsing results without
resorting to extra supervision. The dataset, code and models are available at
http://www.sysu-hcp.net/lip/. | ['Liang Lin', 'Xiaodan Liang', 'Ke Gong', 'Xiaohui Shen'] | 2018-04-05 | null | null | null | null | ['human-parsing'] | ['computer-vision'] | [ 1.50124565e-01 3.44122052e-02 -2.14489102e-01 -5.30515194e-01
-8.37385893e-01 -4.73742396e-01 2.73730546e-01 -3.61679196e-01
-2.71136254e-01 4.10624593e-01 1.73684835e-01 2.70095944e-01
2.66980737e-01 -3.61294150e-01 -5.82947075e-01 -4.66120839e-01
4.30019759e-02 5.71549773e-01 2.80682683e-01 -1.50811523e-01
-3.00445706e-01 2.12731361e-01 -1.52155900e+00 -3.34067233e-02
5.69470167e-01 1.06703484e+00 2.06931427e-01 6.23705268e-01
2.97994792e-01 3.80407393e-01 -5.02214670e-01 -8.62841666e-01
2.23781288e-01 -2.75662601e-01 -7.37755299e-01 3.32526296e-01
6.91907108e-01 -3.99971515e-01 -2.42362618e-01 9.74507034e-01
7.71223128e-01 2.41966024e-02 1.42883599e-01 -1.27649665e+00
-2.68839747e-01 1.21124849e-01 -7.04821348e-01 -1.83645502e-01
6.93401694e-01 1.57236695e-01 9.85417485e-01 -7.43656337e-01
7.47157454e-01 1.35355532e+00 7.76763201e-01 7.60326505e-01
-8.23794186e-01 -5.98681092e-01 2.89012492e-01 1.21169433e-01
-1.17583358e+00 -4.47569966e-01 1.00314021e+00 -3.16841960e-01
4.67399657e-01 1.49882466e-01 7.93632805e-01 1.47276437e+00
-1.85376212e-01 1.08685756e+00 9.04007077e-01 -3.63415122e-01
1.58451758e-02 -9.81700122e-02 2.75144707e-02 1.10659719e+00
3.20596874e-01 -4.36204597e-02 -6.43025279e-01 7.06041530e-02
6.87271178e-01 -9.89464745e-02 2.42414931e-03 -7.85802603e-01
-1.10376740e+00 5.65688610e-01 4.05376971e-01 -1.02665812e-01
-5.90326488e-02 2.16447085e-01 4.58391607e-01 -3.60344142e-01
1.85620070e-01 6.68885410e-02 -6.52603507e-01 -1.88248437e-02
-5.01144111e-01 3.31696630e-01 5.77002525e-01 1.14812124e+00
4.32085723e-01 -2.31058076e-01 4.73023169e-02 9.57738042e-01
3.23823959e-01 6.69379771e-01 1.61410928e-01 -1.31013894e+00
6.83731854e-01 6.11268997e-01 2.73731370e-02 -1.16504371e+00
-9.14640903e-01 -3.24757934e-01 -7.59066105e-01 -6.27089143e-02
6.53431714e-01 -1.59984723e-01 -8.27002168e-01 1.95275569e+00
8.22453022e-01 -3.48893881e-01 -2.64228553e-01 1.13291967e+00
9.69417691e-01 1.68561295e-01 3.43758285e-01 1.76780313e-01
1.88263702e+00 -1.29040802e+00 -6.89972818e-01 -5.64006031e-01
3.83851588e-01 -6.70660734e-01 1.14720058e+00 4.64412451e-01
-1.06085229e+00 -7.88276076e-01 -9.03599262e-01 -4.41707820e-01
-2.59611100e-01 5.34199953e-01 9.17098522e-01 7.21908987e-01
-7.38506019e-01 3.18530202e-01 -1.02176046e+00 -5.95040977e-01
3.36693078e-01 2.68539935e-01 -5.33252776e-01 -1.61361784e-01
-1.01349056e+00 6.23190224e-01 1.97905079e-01 4.15039688e-01
-4.16919678e-01 -9.29829106e-02 -1.14248586e+00 -2.48842984e-01
8.18642795e-01 -9.14438069e-01 1.29522705e+00 -6.11492634e-01
-1.50617838e+00 9.67981040e-01 -3.32223289e-02 -4.94017778e-03
8.30104172e-01 -6.06103837e-01 -1.96250409e-01 2.81842679e-01
2.55314499e-01 1.00791168e+00 4.51807171e-01 -1.09016335e+00
-5.94324470e-01 -7.56287515e-01 4.01325300e-02 2.10823521e-01
-1.73443347e-01 1.46886840e-01 -1.28172064e+00 -6.48694575e-01
1.93103418e-01 -1.23708522e+00 -3.17188472e-01 1.87029794e-01
-6.52730465e-01 -1.04435444e-01 4.48943824e-01 -1.03610659e+00
9.85077143e-01 -1.92222536e+00 4.62237835e-01 -1.22367360e-01
1.82776764e-01 1.97139561e-01 -2.08096858e-02 2.05776662e-01
3.18591982e-01 -9.48890895e-02 -2.23809019e-01 -7.67997324e-01
2.09447205e-01 2.24259481e-01 2.93254107e-01 4.11793113e-01
3.29633318e-02 9.85687375e-01 -6.91561520e-01 -9.48256791e-01
2.74051577e-01 5.10860860e-01 -6.09666109e-01 3.48852038e-01
-8.88115019e-02 7.77671039e-01 -6.22653067e-01 1.19479299e+00
5.02020061e-01 -2.63359517e-01 4.16163474e-01 -4.71264124e-01
2.33884767e-01 -1.82444826e-01 -1.16179478e+00 1.96872568e+00
-1.86293945e-01 2.50553817e-01 2.24597797e-01 -7.21646965e-01
5.89293778e-01 1.72734603e-01 6.23681903e-01 -6.56622767e-01
3.18953246e-01 -7.38204047e-02 -3.52514088e-01 -5.95224619e-01
2.52273768e-01 2.13513374e-01 -3.77862930e-01 -8.46759155e-02
6.91239461e-02 2.15013400e-01 4.11216378e-01 -2.11053900e-02
8.10167074e-01 5.56818247e-01 3.97331417e-01 1.45788684e-01
4.85656798e-01 -1.28267571e-01 9.62785959e-01 3.13639283e-01
-6.06185257e-01 8.20017457e-01 4.12055552e-01 -5.95370293e-01
-8.57552648e-01 -1.01588118e+00 -2.86811050e-02 1.24903059e+00
3.52959603e-01 -5.99299610e-01 -1.05512083e+00 -8.16506147e-01
-9.53787193e-02 -4.06064466e-02 -5.89757144e-01 1.20198756e-01
-8.79450798e-01 -7.84220576e-01 5.74128032e-01 9.81049120e-01
6.58390760e-01 -1.11331093e+00 -7.54723847e-01 -9.57347378e-02
-6.56731367e-01 -1.63829672e+00 -4.93113101e-01 -1.46102041e-01
-6.39482260e-01 -1.41858315e+00 -8.59076679e-01 -8.02472532e-01
5.87142169e-01 -7.67702237e-02 1.12272191e+00 -1.20848984e-01
-6.67410493e-01 5.31697273e-01 -2.24620819e-01 -2.82902628e-01
-1.14299394e-02 2.65934408e-01 8.60604346e-02 -1.66863471e-01
4.07038145e-02 -2.78842360e-01 -7.81614065e-01 5.97737730e-01
-3.60186726e-01 1.87275127e-01 7.29320824e-01 6.08049393e-01
6.44403517e-01 -3.02229017e-01 2.79901534e-01 -8.57470989e-01
2.59343442e-02 -9.55892727e-02 -5.48980296e-01 4.03761059e-01
-3.07311445e-01 -2.37872854e-01 3.71244550e-01 -2.17685983e-01
-1.20701170e+00 5.26990354e-01 -3.90627772e-01 -2.00864390e-01
-3.84513170e-01 4.81815916e-03 -7.33329058e-01 6.96766656e-03
2.44204164e-01 -1.18750051e-01 -1.05074510e-01 -7.53492117e-01
4.03188705e-01 3.07279199e-01 9.25902128e-01 -8.16672862e-01
6.16500139e-01 4.76600021e-01 7.91812912e-02 -5.88250399e-01
-1.09387159e+00 -5.92258930e-01 -9.95942295e-01 -3.15151125e-01
1.12300348e+00 -1.00380111e+00 -8.25563252e-01 6.24001384e-01
-1.15065384e+00 -1.53887674e-01 1.09349340e-01 3.43575686e-01
-5.66608310e-01 7.74205744e-01 -7.90606141e-01 -6.77248955e-01
-3.57002258e-01 -1.20322120e+00 1.40918005e+00 2.51246333e-01
-2.82375365e-01 -6.50005698e-01 -1.17307775e-01 1.03088522e+00
-8.84573758e-02 5.27014136e-01 4.66387391e-01 -3.85689110e-01
-4.74722743e-01 -3.24732542e-01 -3.05308998e-01 2.43503988e-01
-7.82478303e-02 -1.84421852e-01 -7.52610564e-01 -2.92484045e-01
-4.02599424e-01 -5.09451747e-01 5.37986517e-01 3.97370100e-01
1.08601129e+00 -3.92876156e-02 -3.94830525e-01 7.96344936e-01
1.01755404e+00 -1.21092640e-01 2.31923372e-01 3.15989643e-01
1.07695305e+00 1.05944288e+00 7.56964207e-01 4.56641197e-01
7.00223804e-01 9.95971680e-01 4.65127498e-01 -2.42193848e-01
-3.26608986e-01 -4.08634961e-01 1.58995017e-01 7.87052810e-01
-4.04970735e-01 -1.56686470e-01 -9.31531489e-01 2.42951497e-01
-1.76968801e+00 -5.61463177e-01 -5.55993873e-04 1.89750171e+00
6.10801339e-01 1.11166716e-01 4.97059405e-01 -2.34615803e-01
8.70018899e-01 1.61613524e-01 -6.31719232e-01 2.25384459e-01
6.01522848e-02 -8.45760033e-02 3.63417953e-01 8.87990370e-02
-1.49037743e+00 1.00031722e+00 6.19023800e+00 5.75862706e-01
-6.51010871e-01 5.88714145e-02 5.57786703e-01 2.92976201e-02
2.76076555e-01 -2.90222794e-01 -1.05063212e+00 3.90521020e-01
4.38907474e-01 5.58076620e-01 1.57282472e-01 1.16494501e+00
-2.66913604e-02 -4.51269411e-02 -1.03764260e+00 1.09737957e+00
2.28935838e-01 -7.19074428e-01 -2.04264760e-01 -2.57435311e-02
3.99983525e-01 -2.87858874e-01 -2.76636537e-02 2.83332050e-01
4.89024408e-02 -7.29282677e-01 8.16367686e-01 3.28718811e-01
6.88838899e-01 -5.99902809e-01 7.08895266e-01 3.74362081e-01
-1.49990714e+00 -6.50000572e-02 -2.51086444e-01 8.78313631e-02
4.64294791e-01 3.19680870e-01 -2.20162600e-01 5.59388518e-01
1.03577328e+00 5.09739697e-01 -7.78581619e-01 6.82013631e-01
-4.26877528e-01 2.37160563e-01 -3.27793181e-01 2.15678424e-01
-2.04242542e-01 -6.25058040e-02 2.29611188e-01 1.23168933e+00
4.06058431e-02 8.92621279e-02 5.89715242e-01 5.03293514e-01
-7.34977201e-02 1.71907917e-01 -1.42946079e-01 2.00015917e-01
2.60128647e-01 1.43053401e+00 -9.23636138e-01 -2.82130510e-01
-4.29565102e-01 9.99131024e-01 4.76704836e-01 1.46189690e-01
-1.16875863e+00 -4.81457859e-02 5.37482500e-01 1.09726802e-01
1.83278412e-01 -3.38005334e-01 -1.51941761e-01 -1.34014821e+00
3.57816666e-01 -9.12337720e-01 5.42243183e-01 -6.27297759e-01
-1.09007061e+00 4.80050981e-01 1.74834698e-01 -9.60831463e-01
-3.03124636e-01 -7.82191038e-01 -1.14127703e-01 3.72230321e-01
-1.20074534e+00 -1.64370847e+00 -6.20507717e-01 4.51120257e-01
7.21837997e-01 6.79281503e-02 5.98030031e-01 5.03507376e-01
-1.00461352e+00 8.50723803e-01 -4.78629678e-01 4.69068885e-01
8.45161915e-01 -1.14522362e+00 4.93431151e-01 6.86848462e-01
5.61939329e-02 5.00938058e-01 5.76012135e-01 -6.88494802e-01
-1.36729681e+00 -9.90375042e-01 7.99922109e-01 -6.57025993e-01
2.15061083e-01 -5.47854900e-01 -5.25118291e-01 7.43834138e-01
-1.65949270e-01 1.88323021e-01 5.46820223e-01 3.11964154e-01
-3.11047792e-01 -8.01782832e-02 -9.79567707e-01 6.23656869e-01
1.47815144e+00 -1.13731645e-01 -3.82421136e-01 3.19769472e-01
6.00584507e-01 -7.21870065e-01 -7.82083333e-01 7.71986425e-01
9.09937263e-01 -1.12667596e+00 1.32594991e+00 -4.28614467e-01
3.06053489e-01 -1.04021631e-01 -2.34830752e-01 -6.28878295e-01
-2.20840842e-01 -2.99062312e-01 -2.01565653e-01 1.36541307e+00
6.04544021e-02 -3.35890651e-01 1.13190806e+00 7.60569751e-01
3.61695588e-02 -9.21328008e-01 -7.99282551e-01 -6.36003733e-01
-2.55798608e-01 -3.56155902e-01 2.91531682e-01 5.51551223e-01
-2.33635843e-01 3.14770311e-01 -7.74894118e-01 2.34495685e-01
8.84283721e-01 4.32394855e-02 1.03945303e+00 -1.06596136e+00
-3.97110879e-01 -1.39688164e-01 -3.96479189e-01 -1.06711602e+00
9.51660722e-02 -4.12147999e-01 1.36618882e-01 -1.36437690e+00
3.73120844e-01 -1.72324777e-01 -4.22386359e-03 6.30365312e-01
-3.90672952e-01 5.16050577e-01 4.06663537e-01 2.54920483e-01
-9.96099532e-01 5.00090361e-01 1.25724602e+00 9.32215080e-02
2.08284393e-01 2.12406859e-01 -3.70693475e-01 1.28646207e+00
6.31631136e-01 -2.16775030e-01 -2.34315798e-01 -4.49347615e-01
1.03658989e-01 2.18247324e-01 5.79496801e-01 -1.19589031e+00
1.84182405e-01 3.03277280e-02 6.79973781e-01 -6.54630184e-01
6.50145710e-01 -6.48186326e-01 1.04431257e-01 3.52999896e-01
-1.97850913e-01 2.29071289e-01 -9.01368856e-02 6.70860112e-01
-2.06551075e-01 9.54984650e-02 7.14239180e-01 -4.15512532e-01
-1.04403651e+00 4.79093224e-01 2.51027793e-01 2.84021974e-01
1.00457072e+00 -2.30963632e-01 1.59092814e-01 -1.86615691e-01
-9.50963736e-01 3.64543527e-01 4.55095440e-01 5.53213537e-01
3.10578406e-01 -1.21631706e+00 -4.11155015e-01 9.27026570e-02
2.00313151e-01 1.79836124e-01 5.07542074e-01 7.23664761e-01
-5.01966238e-01 3.49455237e-01 -3.01712334e-01 -6.91485524e-01
-1.44209623e+00 5.24975181e-01 1.33859292e-01 -3.66445482e-01
-6.80640459e-01 7.17146933e-01 3.89454097e-01 -7.60002553e-01
4.29048926e-01 -1.07842773e-01 -8.73751491e-02 -2.80168448e-02
2.67390013e-01 5.46323359e-01 -1.50029704e-01 -8.61926556e-01
-5.70333421e-01 9.97107983e-01 2.05538169e-01 1.50194824e-01
9.74801719e-01 -3.67802352e-01 2.63969928e-01 6.80011958e-02
1.14756691e+00 5.02069741e-02 -1.53068233e+00 -1.49156585e-01
1.10407290e-03 -4.01688457e-01 -5.33808291e-01 -7.82524049e-01
-1.27560604e+00 9.58972096e-01 5.58396161e-01 -3.92717361e-01
1.06920528e+00 3.28265995e-01 1.08848667e+00 2.09045470e-01
5.94109774e-01 -1.29823267e+00 2.76430428e-01 2.00659752e-01
7.76726186e-01 -1.52912664e+00 1.10367477e-01 -9.64662850e-01
-7.27087677e-01 9.13497031e-01 9.10512328e-01 1.49248868e-01
2.00204581e-01 2.22676709e-01 2.58125067e-01 -1.35443389e-01
-2.52994001e-01 -3.51384282e-01 4.28982973e-01 7.03305542e-01
4.41249818e-01 1.21317260e-01 -2.95161158e-01 9.22597468e-01
-3.50552589e-01 -1.49494931e-01 -1.20372452e-01 8.92498195e-01
-1.86272234e-01 -1.11277318e+00 -4.92217124e-01 -1.17447332e-01
-4.60668176e-01 3.94453049e-01 -4.02389586e-01 1.04683471e+00
3.22320193e-01 6.52307272e-01 -4.67292905e-01 -2.74389833e-01
5.73329747e-01 3.68297026e-02 5.70010424e-01 -4.10392702e-01
-1.41951650e-01 1.89797550e-01 2.87997574e-01 -8.85175169e-01
-4.37298059e-01 -7.95499027e-01 -1.17471743e+00 -9.03625786e-02
-1.87639818e-01 -2.90003479e-01 6.28219426e-01 8.48914921e-01
2.93773949e-01 4.82283652e-01 1.72231764e-01 -1.16377318e+00
-6.08343065e-01 -7.12936938e-01 -4.11055505e-01 6.59038842e-01
-7.00725913e-02 -9.41076577e-01 -2.03498062e-02 1.56144068e-01] | [7.9774932861328125, -0.3683710992336273] |
b555ca2e-896b-4277-964e-a6bbe5f487ae | the-many-ai-challenges-of-hearthstone | 1907.06562 | null | https://arxiv.org/abs/1907.06562v1 | https://arxiv.org/pdf/1907.06562v1.pdf | The Many AI Challenges of Hearthstone | Games have benchmarked AI methods since the inception of the field, with classic board games such as Chess and Go recently leaving room for video games with related yet different sets of challenges. The set of AI problems associated with video games has in recent decades expanded from simply playing games to win, to playing games in particular styles, generating game content, modeling players etc. Different games pose very different challenges for AI systems, and several different AI challenges can typically be posed by the same game. In this article we analyze the popular collectible card game Hearthstone (Blizzard 2014) and describe a varied set of interesting AI challenges posed by this game. Collectible card games are relatively understudied in the AI community, despite their popularity and the interesting challenges they pose. Analyzing a single game in-depth in the manner we do here allows us to see the entire field of AI and Games through the lens of a single game, discovering a few new variations on existing research topics. | ['Fernando De Mesentier Silva', 'Julian Togelius', 'Amy K. Hoover', 'Scott Lee'] | 2019-07-15 | null | null | null | null | ['board-games', 'card-games'] | ['playing-games', 'playing-games'] | [ 2.14167431e-01 1.68857388e-02 1.39324144e-01 2.49877304e-01
-4.32200521e-01 -1.00019777e+00 5.90657473e-01 -1.61890671e-01
-4.28556621e-01 6.41513348e-01 6.96103573e-02 -3.03948279e-02
-5.36265135e-01 -8.69749188e-01 -3.02367181e-01 -3.70865285e-01
-2.95445204e-01 7.29663789e-01 5.56865931e-01 -1.17983115e+00
4.36877251e-01 -2.27065742e-01 -1.92584229e+00 4.31074381e-01
3.62388611e-01 7.93203831e-01 2.67900139e-01 1.08688760e+00
-1.80225428e-02 1.38578856e+00 -8.39553773e-01 -9.30089891e-01
7.89862931e-01 -6.08826339e-01 -8.75988662e-01 -5.21025956e-02
2.00319871e-01 -1.48229823e-01 -4.90123481e-01 1.09394491e+00
3.67832154e-01 1.85027882e-01 2.93433160e-01 -1.72567332e+00
-1.37873292e-01 8.90158176e-01 -4.10749137e-01 3.98174405e-01
6.72355175e-01 2.35658452e-01 1.54537725e+00 -1.89552158e-01
5.90760708e-01 8.63662660e-01 6.18000984e-01 6.77337110e-01
-8.64411294e-01 -6.87060058e-01 1.07147008e-01 3.82086873e-01
-1.11051941e+00 -4.76568043e-02 5.74322641e-01 -4.88661885e-01
6.34517729e-01 5.39408445e-01 1.23916566e+00 8.06628227e-01
-1.23455785e-02 1.07298231e+00 9.49229181e-01 -1.39250636e-01
3.56082916e-01 -5.76425254e-01 -4.29356359e-02 2.93679208e-01
3.16161692e-01 2.25195616e-01 -4.57603157e-01 -1.65890858e-01
7.49395370e-01 -2.14395657e-01 2.81834424e-01 -2.60118425e-01
-1.00208604e+00 1.08095646e+00 -1.60637736e-01 4.59120274e-01
-1.17602073e-01 3.37917209e-01 5.50873697e-01 3.95380974e-01
6.45615831e-02 9.32825506e-01 -1.64049953e-01 -1.00208902e+00
-8.94205213e-01 1.33185029e+00 1.10141802e+00 8.17112327e-01
3.79527897e-01 1.40367463e-01 -1.44176157e-02 5.85589170e-01
-7.99545720e-02 -1.45585775e-01 3.08885932e-01 -1.19535971e+00
2.68976033e-01 4.36510384e-01 3.99667658e-02 -1.08768821e+00
-3.62859070e-01 -2.72126436e-01 -5.43975055e-01 4.04163331e-01
9.94691312e-01 -3.80135059e-01 -4.84727949e-01 1.70501196e+00
4.66092722e-03 1.94516093e-01 -2.82986432e-01 9.25077319e-01
8.28077435e-01 7.43800700e-01 -5.65926097e-02 3.10801864e-01
1.43078041e+00 -8.87675226e-01 -2.45138243e-01 -6.84167862e-01
4.05595243e-01 -4.86640632e-01 8.64187777e-01 9.88587320e-01
-1.56237924e+00 -2.23503724e-01 -1.04606509e+00 2.89624464e-02
-2.07350150e-01 -5.87460935e-01 1.18836677e+00 6.99698627e-01
-1.05535078e+00 6.06852829e-01 -4.55032438e-01 -3.41113210e-01
2.95612216e-01 5.57485104e-01 -9.00753513e-02 -1.16307609e-01
-1.29214478e+00 1.00268209e+00 4.61225837e-01 -3.47801685e-01
-8.43088031e-01 -6.88371778e-01 -6.45230711e-01 -2.74824221e-02
1.16521502e+00 -6.19278193e-01 1.56436908e+00 -1.14217222e+00
-1.66077864e+00 1.16413820e+00 4.79521900e-01 -5.34359753e-01
7.15056539e-01 -7.13476241e-02 8.82232487e-02 -3.87443244e-01
1.19063050e-01 3.32938910e-01 3.82827431e-01 -7.67279148e-01
-1.08518910e+00 -1.77421421e-01 1.13193715e+00 4.85975742e-01
1.95376486e-01 6.64348841e-01 -6.15588963e-01 -7.68412650e-01
-3.16011518e-01 -8.11173260e-01 -6.82043612e-01 -4.30994570e-01
-1.73357725e-01 -1.79091766e-01 1.47189647e-02 -7.88621008e-02
1.62887025e+00 -1.83007801e+00 6.98064208e-01 1.09132603e-02
8.17141116e-01 -1.55963497e-02 1.31528480e-02 7.22328842e-01
-6.65260479e-02 6.89691007e-02 -1.32293478e-01 2.50787824e-01
3.17472428e-01 2.37445116e-01 1.07447557e-01 4.02427241e-02
-1.21771462e-01 1.17891836e+00 -1.21740746e+00 -1.44658461e-01
1.59103945e-02 -2.95315474e-01 -8.79119098e-01 -1.90785788e-02
-3.88932109e-01 -2.68002022e-02 -5.18679082e-01 3.40608656e-01
2.19748020e-01 -6.55218512e-02 5.24195880e-02 5.35920143e-01
-1.46265551e-01 1.59973025e-01 -1.47969162e+00 1.76837480e+00
2.00471893e-01 7.08160698e-01 2.73175091e-01 -1.06976998e+00
5.48275828e-01 1.05401859e-01 7.53614843e-01 -6.06840789e-01
4.74221826e-01 -2.51347013e-02 6.98565066e-01 -3.16994727e-01
1.04698777e+00 -3.31385553e-01 -7.15279758e-01 7.45211542e-01
-1.13717571e-01 -6.83015108e-01 8.96000028e-01 3.51018280e-01
1.48739123e+00 1.04973726e-01 5.78617394e-01 -2.75087953e-01
1.68671589e-02 5.73645115e-01 3.50016952e-01 1.19178271e+00
-2.61636764e-01 6.67882383e-01 6.49800122e-01 -7.49490440e-01
-1.23024487e+00 -8.82389307e-01 2.96640098e-01 1.45536184e+00
1.86880127e-01 -1.06345391e+00 -8.87329578e-01 4.27274294e-02
-2.51464367e-01 1.17873974e-01 -6.55081093e-01 -4.23822962e-02
-6.44492567e-01 -8.94279420e-01 7.92076349e-01 2.47456849e-01
3.28690797e-01 -1.33934247e+00 -9.74639416e-01 4.88151461e-01
-4.83646952e-02 -9.99631763e-01 -8.84653851e-02 1.44205570e-01
-4.02146131e-01 -1.26145136e+00 -6.83895171e-01 -6.36383951e-01
-1.26049280e-01 2.73397267e-01 1.68051207e+00 2.16684967e-01
-6.98859394e-01 3.78643125e-01 -5.27140617e-01 -9.75408435e-01
-4.03160393e-01 1.81368947e-01 -4.08816375e-02 -2.36467481e-01
5.52380979e-01 -7.16705561e-01 -1.89376831e-01 1.19178481e-01
-9.79890943e-01 4.47853088e-01 1.16546229e-01 5.88733017e-01
3.59037854e-02 2.85514116e-01 8.91737044e-02 -9.73474205e-01
9.17309880e-01 -8.98590088e-01 -5.94341457e-01 -2.53010958e-01
1.37980834e-01 -3.63617957e-01 7.06721246e-01 -5.50129652e-01
-3.37416172e-01 -3.87411356e-01 -6.44196644e-02 4.50163968e-02
-1.50067270e-01 6.25507712e-01 2.33450979e-02 4.15942185e-02
8.70785654e-01 1.59640208e-01 -9.73088294e-02 -3.27643231e-02
4.29935575e-01 2.25913063e-01 5.97131014e-01 -9.58318949e-01
1.05734551e+00 2.07593709e-01 -5.30658849e-02 -6.66376770e-01
-6.89739406e-01 -2.09005907e-01 -2.49463111e-01 -5.76073885e-01
7.18843639e-01 -5.98833621e-01 -1.17542410e+00 8.59755993e-01
-8.38830173e-01 -4.53769952e-01 -6.60181105e-01 2.90055517e-02
-8.23498726e-01 3.23355682e-02 -6.80455804e-01 -7.65743673e-01
1.01407744e-01 -1.13419867e+00 6.12962604e-01 4.19068754e-01
-5.48678875e-01 -7.78464139e-01 4.26934958e-01 5.22365332e-01
4.76705045e-01 3.14899981e-01 6.91773355e-01 -5.70600748e-01
-5.66355765e-01 -3.68665576e-01 2.03935832e-01 -9.55672786e-02
1.55076720e-02 -8.62517953e-02 -5.67233920e-01 4.78715077e-02
-1.74611315e-01 -2.72252738e-01 3.18710774e-01 5.50647676e-01
8.03974807e-01 5.65609969e-02 8.44568312e-02 4.97818261e-01
1.36239946e+00 5.21673977e-01 8.72016847e-01 7.46060193e-01
5.38969338e-01 5.54687262e-01 3.97421539e-01 6.74655259e-01
5.21696210e-01 7.82351196e-01 5.38397074e-01 2.77940154e-01
4.66795564e-01 -5.13263457e-02 1.20397240e-01 5.44417858e-01
-8.06008339e-01 -4.01877463e-01 -1.01459801e+00 3.88755560e-01
-2.03212404e+00 -1.51942849e+00 1.22687221e-01 1.85963571e+00
6.11232758e-01 5.00422537e-01 9.10683513e-01 3.81731629e-01
5.76156378e-01 2.66213864e-01 -3.28737020e-01 -4.14313674e-01
-2.86882102e-01 6.84439719e-01 8.90623257e-02 4.44391847e-01
-9.90618289e-01 1.32727242e+00 7.25359344e+00 1.23109794e+00
-6.94785476e-01 -2.61576593e-01 4.92173135e-01 -4.81318474e-01
-1.02708079e-01 2.00271547e-01 -4.13592696e-01 4.17712063e-01
3.88090014e-01 -8.32261205e-01 9.64107573e-01 7.61117995e-01
-4.71165106e-02 -3.74622911e-01 -1.14553308e+00 1.20998466e+00
8.72485489e-02 -1.38700700e+00 -1.10855289e-01 2.25373164e-01
4.95474786e-01 -8.93187374e-02 1.01521775e-01 5.09303629e-01
1.13366020e+00 -1.52099383e+00 9.05543566e-01 1.03056408e-01
2.77738661e-01 -6.74832642e-01 1.93547696e-01 4.75259840e-01
-1.13768637e+00 -3.72493178e-01 -4.35525656e-01 -1.18031490e+00
3.45132723e-02 -1.72634557e-01 -3.05504296e-02 4.00013298e-01
6.61911964e-01 6.63935900e-01 -3.21974874e-01 1.32657981e+00
1.71070471e-01 3.91417623e-01 -4.05237138e-01 -3.71461511e-01
6.47625089e-01 -5.20428896e-01 8.26547384e-01 7.02453911e-01
2.39438117e-01 6.47461236e-01 2.11684197e-01 8.32943082e-01
1.49658442e-01 1.37626976e-01 -6.60676003e-01 -2.62775749e-01
1.16612196e-01 1.16344666e+00 -8.68631184e-01 -6.79480210e-02
-4.62541342e-01 7.68534243e-01 1.89704329e-01 -4.02649045e-02
-8.56154323e-01 -1.82064578e-01 1.09667230e+00 2.55289197e-01
5.11710793e-02 -1.15834907e-01 -1.68817088e-01 -1.31053030e+00
-1.31295055e-01 -1.48428679e+00 4.47726876e-01 -8.74361575e-01
-1.21240461e+00 5.08172214e-01 3.02226067e-01 -1.18968070e+00
-5.36401808e-01 -6.63389802e-01 -9.72451985e-01 3.42434257e-01
-6.58128500e-01 -6.20044529e-01 -2.65384138e-01 5.80248415e-01
7.55398989e-01 -4.02203083e-01 6.61327839e-01 2.07876340e-01
-3.10022891e-01 2.76255757e-01 7.58963600e-02 1.61425248e-01
1.00918770e-01 -1.15106153e+00 8.37268054e-01 6.12750530e-01
4.03245300e-01 8.90534446e-02 9.55327690e-01 -1.68946609e-01
-1.59845853e+00 -2.32374668e-01 3.78134102e-01 -7.71184742e-01
9.31842923e-01 -4.86310422e-01 -2.00215295e-01 6.23745441e-01
2.50428952e-02 -5.83899677e-01 4.10422325e-01 3.25188041e-02
-6.74078166e-02 2.87647724e-01 -5.78353286e-01 1.11598039e+00
1.26935136e+00 -2.44292378e-01 -5.56668520e-01 -7.58027099e-03
2.34213501e-01 -7.33376503e-01 -3.08836937e-01 -4.69542593e-02
8.38243067e-01 -1.18368185e+00 9.68397439e-01 -1.11597943e+00
9.67061460e-01 2.65477914e-02 -5.40870354e-02 -1.23908424e+00
-6.57823682e-01 -1.06002474e+00 3.33725095e-01 8.07404160e-01
1.38421044e-01 -5.54208197e-02 1.34559226e+00 7.52793014e-01
-4.28258888e-02 -8.54568303e-01 -6.76214159e-01 -6.20632946e-01
4.86043483e-01 -9.88148808e-01 6.18622184e-01 7.88500428e-01
4.73055989e-01 1.54934838e-01 -6.39063776e-01 -5.54555893e-01
5.43682873e-01 -1.45313039e-01 1.14397097e+00 -1.53776538e+00
-9.64453518e-01 -1.04928195e+00 -1.01288676e+00 -1.03318119e+00
-2.46403113e-01 -9.18143988e-01 -6.33724928e-02 -1.24685621e+00
3.80583733e-01 -2.87782431e-01 6.21963553e-02 1.49163306e-01
-2.20978141e-01 6.71597540e-01 8.14943969e-01 1.16143040e-02
-8.57027054e-01 -1.45859241e-01 1.12379873e+00 -2.30582014e-01
-6.70091808e-02 1.36355326e-01 -1.15764415e+00 8.74444544e-01
6.15979016e-01 -3.49112719e-01 -5.13622344e-01 -4.27758157e-01
1.06316304e+00 5.49919903e-02 1.18304104e-01 -1.22528481e+00
3.23376924e-01 -5.79522491e-01 -3.46219897e-01 3.42898607e-01
4.68122661e-01 -4.39910740e-01 4.15088236e-01 3.89918327e-01
-1.56532943e-01 2.23851502e-01 4.30386692e-01 8.19020718e-02
-2.64344662e-01 -5.18334806e-01 4.46589619e-01 -6.99768960e-01
-1.04368722e+00 4.27950084e-01 -1.01998413e+00 9.32628810e-01
1.38459587e+00 -8.57672691e-01 -3.35694402e-02 -8.36682558e-01
-8.91437352e-01 2.61562675e-01 3.64450634e-01 4.15617675e-01
2.24105313e-01 -9.53726232e-01 -9.58470702e-01 -2.43683085e-01
6.41724318e-02 -5.83558939e-02 3.54603946e-01 2.34049559e-01
-7.99509883e-01 -4.58361991e-02 -8.42477858e-01 -6.86077923e-02
-1.34354377e+00 2.10013360e-01 3.53394985e-01 -6.75174773e-01
-5.12180507e-01 1.07092464e+00 4.14880931e-01 -1.65223449e-01
1.72799200e-01 1.37023628e-02 -4.02638584e-01 5.86286001e-02
5.91118515e-01 4.02079970e-01 -2.56669819e-01 -4.53712076e-01
-1.54913753e-01 5.17293334e-01 -2.43981425e-02 -2.40507707e-01
1.62080193e+00 2.78163046e-01 -2.48315949e-02 5.70064127e-01
2.30107337e-01 -2.03051582e-01 -9.35195923e-01 -2.62841936e-02
-1.49236411e-01 -2.78654158e-01 -3.61223340e-01 -6.40645146e-01
-1.01130295e+00 6.96517169e-01 -1.44011542e-01 5.51408947e-01
1.06851518e+00 -1.25324335e-02 6.99959338e-01 1.47475019e-01
8.75519454e-01 -9.62708473e-01 6.92478986e-03 1.07634151e+00
4.85149562e-01 -8.27271879e-01 -1.14545755e-01 -2.50961393e-01
-8.19363832e-01 1.03204882e+00 6.43974483e-01 -6.48462355e-01
5.09866893e-01 7.20596015e-01 -2.91012973e-01 -4.34399694e-01
-8.12325358e-01 -3.89032185e-01 -2.01579612e-02 5.92502534e-01
4.15295959e-01 -5.52251451e-02 -3.09595972e-01 1.15781760e+00
-9.84059989e-01 1.97424278e-01 9.99451518e-01 1.10439169e+00
-4.56203461e-01 -1.35443616e+00 -3.82198662e-01 5.49347401e-01
-6.53448701e-01 -3.24559271e-01 -7.51829743e-01 9.55472767e-01
1.92899466e-01 8.51953924e-01 1.38525054e-01 -7.41121173e-01
3.22773248e-01 -1.76999614e-01 8.73026311e-01 -8.10455799e-01
-1.19860959e+00 -4.07679051e-01 1.55921131e-01 -5.55520236e-01
-1.69286966e-01 -5.77496290e-01 -6.86655164e-01 -1.05671430e+00
1.72586758e-02 3.32204521e-01 1.80749327e-01 1.14415097e+00
-1.55498818e-01 3.39875549e-01 5.16764633e-02 -1.07862949e+00
-1.90564588e-01 -5.78568459e-01 -1.05054343e+00 3.93455267e-01
-2.69043833e-01 -5.10494888e-01 -5.49745150e-02 -1.21788912e-01] | [3.496347427368164, 1.4493240118026733] |
ba25b523-cf73-4277-8083-8143c591c01e | personalized-keyword-spotting-through-multi | 2206.13708 | null | https://arxiv.org/abs/2206.13708v1 | https://arxiv.org/pdf/2206.13708v1.pdf | Personalized Keyword Spotting through Multi-task Learning | Keyword spotting (KWS) plays an essential role in enabling speech-based user interaction on smart devices, and conventional KWS (C-KWS) approaches have concentrated on detecting user-agnostic pre-defined keywords. However, in practice, most user interactions come from target users enrolled in the device which motivates to construct personalized keyword spotting. We design two personalized KWS tasks; (1) Target user Biased KWS (TB-KWS) and (2) Target user Only KWS (TO-KWS). To solve the tasks, we propose personalized keyword spotting through multi-task learning (PK-MTL) that consists of multi-task learning and task-adaptation. First, we introduce applying multi-task learning on keyword spotting and speaker verification to leverage user information to the keyword spotting system. Next, we design task-specific scoring functions to adapt to the personalized KWS tasks thoroughly. We evaluate our framework on conventional and personalized scenarios, and the results show that PK-MTL can dramatically reduce the false alarm rate, especially in various practical scenarios. | ['Simyung Chang', 'Inseop Chung', 'Byeonggeun Kim', 'Seunghan Yang'] | 2022-06-28 | null | null | null | null | ['keyword-spotting'] | ['speech'] | [ 3.01254869e-01 -2.73034513e-01 -4.20348078e-01 -5.07092237e-01
-1.62885928e+00 -4.08421636e-01 1.85136139e-01 -4.01028305e-01
-1.60639212e-01 3.34086657e-01 2.30522946e-01 -5.70353866e-01
-1.79442883e-01 5.25971390e-02 -3.85073245e-01 -5.48800409e-01
2.62223095e-01 1.91975057e-01 2.12236628e-01 2.38138810e-02
4.93720509e-02 -2.88813021e-02 -1.04094303e+00 3.94085228e-01
7.85886407e-01 1.08257961e+00 5.06232381e-01 9.85602558e-01
-1.38553027e-02 1.18900828e-01 -8.16685200e-01 -2.49031216e-01
3.20860324e-03 -2.23129645e-01 -5.44244230e-01 -3.10161531e-01
9.86405015e-02 -3.79929751e-01 -1.94167867e-01 6.83282197e-01
1.16773796e+00 2.02520207e-01 2.76926935e-01 -1.55930734e+00
-5.40910125e-01 8.27549100e-01 -3.72225761e-01 2.01945186e-01
5.22184253e-01 8.57356638e-02 9.91619289e-01 -1.01222193e+00
-4.13458586e-01 1.03544104e+00 8.59587252e-01 6.55612648e-01
-1.10711336e+00 -1.02316558e+00 2.99815714e-01 2.73237228e-01
-1.72292721e+00 -9.24198151e-01 7.62100577e-01 -2.40443483e-01
8.77207816e-01 5.37841737e-01 7.17762113e-02 1.27017438e+00
-3.28882366e-01 1.26992762e+00 7.80265749e-01 -2.62536258e-01
5.02885170e-02 4.13607389e-01 1.61342204e-01 1.08474120e-01
-2.98051864e-01 -3.15108538e-01 -1.00180924e+00 -5.44188082e-01
1.22161828e-01 -7.49536185e-03 -4.09873068e-01 5.48575595e-02
-1.23626411e+00 4.28720623e-01 -6.45447731e-01 1.72261283e-01
-2.65204936e-01 -7.71069080e-02 4.72291201e-01 3.84418875e-01
3.59268725e-01 2.12303236e-01 -1.16091597e+00 -4.71087813e-01
-1.04377496e+00 1.21520758e-01 6.71337903e-01 1.08250546e+00
6.62530541e-01 6.47471994e-02 -8.27430308e-01 1.15893829e+00
3.06124061e-01 9.32259381e-01 9.49046075e-01 -4.47766662e-01
5.55385470e-01 1.89733312e-01 2.89769471e-01 -3.98368031e-01
-3.69277924e-01 -2.44918093e-01 -4.29434091e-01 -7.91074872e-01
-1.24462306e-01 -4.60130215e-01 -7.17806458e-01 1.93554366e+00
2.68259317e-01 5.55744708e-01 -4.18208092e-02 5.50546288e-01
7.03601122e-01 6.83995485e-01 4.25178111e-02 -4.29040939e-01
1.40571237e+00 -8.56686056e-01 -9.56718743e-01 -1.06368206e-01
8.73561740e-01 -8.75047863e-01 1.67064345e+00 3.06914479e-01
-4.95411038e-01 -3.30411792e-01 -6.97028279e-01 4.00627673e-01
-3.12376887e-01 5.74343264e-01 3.52400452e-01 1.24161136e+00
-9.80375051e-01 -1.74098499e-02 -2.79002577e-01 -1.05311647e-01
1.09023683e-01 6.60144985e-01 5.80579555e-03 3.84668618e-01
-1.54336667e+00 4.41423804e-01 2.08600491e-01 -1.26485497e-01
-8.96968067e-01 -9.03041661e-01 -7.23721921e-01 1.49868131e-01
7.12016761e-01 -2.90742964e-01 1.90047610e+00 -3.58488083e-01
-1.91043806e+00 5.20053089e-01 -6.69582784e-01 -1.19467065e-01
6.58300444e-02 -3.20343047e-01 -9.61477339e-01 -2.69118458e-01
9.55867916e-02 -3.69206108e-02 1.31575739e+00 -1.15046144e+00
-8.70932758e-01 -1.50019512e-01 -1.79328457e-01 2.49635994e-01
-8.19488406e-01 2.27539942e-01 -7.38432884e-01 -7.23848820e-01
-3.00033301e-01 -6.11127436e-01 2.33531803e-01 -6.78526342e-01
-8.11399996e-01 -6.18631005e-01 1.24108136e+00 -9.97989476e-01
1.83174276e+00 -2.38976693e+00 -3.49980205e-01 2.12818831e-01
7.26124197e-02 7.62018979e-01 -8.58390629e-02 2.25127190e-01
8.55337679e-02 -1.28600076e-01 1.88956112e-01 -7.67447352e-01
2.00076699e-01 -7.44117498e-02 -3.10493439e-01 2.56997421e-02
-3.26411426e-01 1.11186981e+00 -8.46396506e-01 -4.64551479e-01
2.27499053e-01 2.62008369e-01 -2.80854315e-01 4.80500638e-01
-5.88827021e-02 1.64462909e-01 -6.24088705e-01 7.49105930e-01
4.71396208e-01 -2.57811606e-01 1.25695288e-01 -3.90862823e-01
7.27324933e-02 4.95074153e-01 -1.20975590e+00 1.46451759e+00
-7.93761253e-01 1.47327811e-01 1.54432878e-01 -1.00999355e+00
4.88737196e-01 7.18794882e-01 4.64247525e-01 -6.26396358e-01
-1.70360863e-01 2.03028902e-01 -3.35740030e-01 -6.33716524e-01
4.61606950e-01 1.95164323e-01 -3.58326405e-01 7.96429932e-01
1.75314099e-01 2.54392058e-01 -5.51805377e-01 2.47437164e-01
8.29260409e-01 -2.32572749e-01 3.35863143e-01 8.62337127e-02
6.22425556e-01 -8.03266764e-01 4.97353584e-01 1.06093860e+00
-4.29332942e-01 2.79672503e-01 7.43098333e-02 1.62206054e-01
-4.59638745e-01 -8.20132911e-01 1.14003718e-01 1.77624702e+00
4.25896235e-02 -5.65492034e-01 -8.03797722e-01 -7.38710821e-01
-7.50405621e-03 8.39372694e-01 -2.30477840e-01 -4.66379911e-01
-2.24591330e-01 -6.68425977e-01 9.13058519e-01 3.08821827e-01
1.85032099e-01 -8.15773845e-01 -1.85316384e-01 1.24788776e-01
-3.55580837e-01 -1.33648741e+00 -1.43400979e+00 2.53725559e-01
-9.79034603e-02 -6.80502415e-01 -1.02207386e+00 -6.25600636e-01
2.30323032e-01 6.57240689e-01 4.34042156e-01 -5.22112548e-01
-3.04712374e-02 8.71007502e-01 -4.90591913e-01 -3.34151566e-01
-1.96007982e-01 5.19726753e-01 6.29270852e-01 5.93644500e-01
4.20019120e-01 -5.20310521e-01 -4.81656909e-01 6.47034764e-01
-3.77403408e-01 -1.25799209e-01 8.28122258e-01 7.11097062e-01
6.22449405e-02 -1.64118484e-01 9.36990142e-01 -6.42090917e-01
1.04383731e+00 -2.61191189e-01 -2.03933209e-01 7.64749885e-01
-8.54514420e-01 -1.53509788e-02 5.42777836e-01 -1.11366856e+00
-8.81288230e-01 2.56149545e-02 -4.17423308e-01 -4.32835817e-01
1.15897506e-01 1.94481552e-01 -7.03952253e-01 -2.86993623e-01
2.74726808e-01 8.82216871e-01 -5.71148694e-01 -6.29440129e-01
2.68606097e-01 1.53344941e+00 4.65370953e-01 -4.23981726e-01
7.94298112e-01 -1.65270358e-01 -9.47705328e-01 -9.84243274e-01
-6.91431940e-01 -1.06981075e+00 -6.95397779e-02 -6.61028773e-02
4.68401521e-01 -8.57940018e-01 -1.10230207e+00 6.55238211e-01
-1.06240284e+00 -4.52061027e-01 3.78919020e-02 3.97963196e-01
-3.95405978e-01 4.78656679e-01 -2.65818268e-01 -1.42814291e+00
-7.56323993e-01 -1.19080091e+00 1.70423365e+00 1.23150773e-01
-3.91255230e-01 -7.82158077e-01 -1.21754631e-01 3.37562412e-01
6.56172812e-01 -7.69875467e-01 6.39756262e-01 -1.21965933e+00
-3.51713374e-02 -3.50991577e-01 -1.20496981e-01 3.30895483e-01
5.80847442e-01 -7.90713787e-01 -1.33418548e+00 -2.18155965e-01
-2.10035630e-02 -2.94944733e-01 3.30947608e-01 3.42414409e-01
1.52903163e+00 -5.65990984e-01 -6.20923936e-01 5.07983744e-01
6.28533840e-01 3.74819994e-01 1.10742435e-01 -1.36717051e-01
7.92020261e-01 1.71456635e-02 4.83352900e-01 6.12033546e-01
5.95752001e-01 1.28403592e+00 -4.09032375e-01 4.41098437e-02
1.76717564e-01 -5.55830836e-01 5.85392773e-01 8.08094203e-01
6.30853057e-01 -2.32126012e-01 -5.82853675e-01 5.24731696e-01
-1.92060518e+00 -6.04000270e-01 3.56851369e-01 2.42211199e+00
1.05208778e+00 3.01422291e-02 3.71703625e-01 3.10718566e-01
9.12842751e-01 8.33866969e-02 -8.91656339e-01 -1.21144511e-01
5.62238507e-02 8.54764283e-02 4.91778016e-01 6.37420654e-01
-1.11737776e+00 1.03583956e+00 6.14916658e+00 1.34497738e+00
-1.17584825e+00 6.14653707e-01 5.61397851e-01 -7.59698153e-02
-2.57036507e-01 -3.40218633e-01 -1.21520734e+00 8.30813408e-01
1.04247546e+00 -4.51355010e-01 4.93824452e-01 1.02879167e+00
5.09708643e-01 1.47059321e-01 -9.95391846e-01 1.54562438e+00
3.12124431e-01 -8.23472083e-01 -7.42460936e-02 -1.23660542e-01
2.81293124e-01 -2.06014320e-01 2.75235504e-01 6.88679218e-01
-1.71730891e-02 -5.46047449e-01 5.22389710e-01 2.24113435e-01
1.08536148e+00 -3.78302336e-01 4.61274475e-01 4.34853166e-01
-1.33690071e+00 -1.05014868e-01 2.22622439e-01 5.65169334e-01
4.84794855e-01 7.12641776e-01 -1.16221404e+00 5.35319924e-01
6.80377305e-01 5.25024021e-03 -1.13934308e-01 8.19514811e-01
5.25914840e-02 8.74720633e-01 -4.08383787e-01 -1.78239480e-01
-2.13337407e-01 3.85697901e-01 4.31675762e-01 1.29601669e+00
4.82016265e-01 -1.08332463e-01 3.09561908e-01 2.97710508e-01
-1.23964384e-01 1.97241753e-01 -1.89455613e-01 -8.46201926e-02
7.94840991e-01 1.34561157e+00 -2.80480862e-01 -3.18041176e-01
-3.13534737e-01 1.50044191e+00 -1.85797140e-01 5.80831528e-01
-7.98658133e-01 -5.22057652e-01 8.77572179e-01 -1.61808401e-01
3.57164443e-01 -4.21074405e-02 3.24321352e-02 -1.02336395e+00
1.02502681e-01 -8.05237889e-01 1.66564167e-01 -5.28104126e-01
-9.99266744e-01 1.73468471e-01 4.23483290e-02 -8.93382847e-01
-2.22585723e-01 -3.59827697e-01 -6.33643031e-01 1.07540703e+00
-1.37492776e+00 -1.23551500e+00 7.83919021e-02 7.67606735e-01
8.24426889e-01 -1.82173193e-01 9.20945525e-01 8.54281068e-01
-5.19395232e-01 1.54849577e+00 1.57721639e-02 -4.71802941e-03
1.01272762e+00 -9.72724497e-01 6.64912105e-01 6.36942387e-01
1.25534326e-01 8.21008086e-01 5.82641900e-01 -5.83051920e-01
-1.52732503e+00 -1.04430974e+00 1.13452268e+00 -4.90703046e-01
5.12036562e-01 -9.42658424e-01 -6.33692265e-01 5.73298752e-01
-2.97605067e-01 -2.42903978e-01 1.08709645e+00 4.03853685e-01
-3.46257269e-01 -2.69096434e-01 -9.74663496e-01 5.50993264e-01
9.54920232e-01 -1.10857892e+00 -1.88740849e-01 5.46588302e-01
1.17979324e+00 -3.33007127e-01 -4.19770300e-01 4.50033426e-01
8.36680055e-01 -4.59678829e-01 8.43670368e-01 -4.27339464e-01
-8.75167787e-01 -2.89527386e-01 -1.66355029e-01 -1.31369352e+00
3.48693528e-03 -1.36851621e+00 -5.87541938e-01 1.50095403e+00
8.51170599e-01 -6.43768609e-01 6.62364006e-01 6.53144538e-01
-2.54982591e-01 -6.40767574e-01 -1.13875937e+00 -8.87695134e-01
-7.69001603e-01 -9.68922317e-01 8.25351357e-01 8.55917215e-01
1.62844032e-01 4.67425823e-01 -1.06227934e+00 5.15323877e-01
3.04841578e-01 -1.38912350e-01 8.07009578e-01 -7.60197639e-01
-6.24503553e-01 -2.51960725e-01 9.14280713e-02 -1.52886081e+00
-3.60143982e-04 -4.75198299e-01 3.38059157e-01 -1.02972615e+00
1.23152085e-01 -4.10987198e-01 -3.69316131e-01 7.22526371e-01
-5.75436532e-01 -2.46189144e-02 -1.59599438e-01 1.12016633e-01
-1.10353684e+00 5.26761115e-01 5.18724144e-01 -9.99052078e-02
-5.87498486e-01 6.52637780e-01 -9.57473814e-01 2.12602839e-01
7.29926109e-01 -1.80303171e-01 -5.29936910e-01 -1.18326014e-02
-1.26107428e-02 -1.88710883e-01 -1.77236870e-01 -6.60618067e-01
3.21980000e-01 -3.37671526e-02 -1.30110294e-01 -4.86240357e-01
4.21910703e-01 -5.65455854e-01 -3.41562837e-01 1.05450779e-01
-3.09365094e-01 -2.59030521e-01 5.36816120e-02 4.09676611e-01
2.94706106e-01 8.95679221e-02 3.65505040e-01 3.36721063e-01
-7.21374929e-01 5.19070327e-01 -4.64298189e-01 -1.18523188e-01
9.21327055e-01 -1.08039789e-01 1.10519648e-01 -7.30533361e-01
-6.52340114e-01 6.05370402e-01 -3.02237153e-01 6.53595328e-01
5.94514608e-01 -1.33243132e+00 -1.63956881e-01 3.06850344e-01
5.00665903e-01 -2.73779750e-01 3.73969376e-01 8.89397919e-01
5.87358952e-01 7.29014874e-01 8.02934289e-01 -5.28638959e-01
-1.52828455e+00 3.71741861e-01 2.43637681e-01 -5.79821356e-02
-2.66928256e-01 1.08836675e+00 2.10059255e-01 -7.29625165e-01
7.82046974e-01 -3.68169487e-01 -1.79046970e-02 -1.62187189e-01
7.57209718e-01 1.39987305e-01 2.97296822e-01 -4.33755636e-01
-5.04794419e-01 3.06723744e-01 -2.49722540e-01 -1.97203636e-01
7.14891553e-01 -4.27022159e-01 5.67083597e-01 7.07024992e-01
1.06245482e+00 1.19378731e-01 -9.34996188e-01 -6.05466902e-01
2.70997792e-01 -1.27933234e-01 -4.02128976e-03 -9.26732838e-01
-5.36617756e-01 5.23258269e-01 7.61551261e-01 1.52503088e-01
1.10861540e+00 2.23588735e-01 1.38483357e+00 4.49176371e-01
4.32726502e-01 -1.24854898e+00 -2.99145449e-02 3.44786644e-01
3.48055482e-01 -1.22200000e+00 -5.13997614e-01 -2.42247820e-01
-7.95578182e-01 4.03766960e-01 4.81534302e-01 8.24915409e-01
7.79012561e-01 2.59262770e-01 9.68636647e-02 1.48357093e-01
-4.66169715e-01 -2.69695967e-01 4.75231349e-01 6.83062434e-01
1.58263683e-01 2.68290669e-01 2.70190705e-02 1.29597187e+00
-6.62651211e-02 -6.17882125e-02 -2.93235689e-01 6.42206192e-01
-2.79381216e-01 -1.35792971e+00 -2.70934016e-01 7.14181066e-01
-4.20794010e-01 -4.47646201e-01 -4.85308558e-01 -4.53158319e-02
-2.64805462e-02 1.07249975e+00 -5.77505112e-01 -9.98656929e-01
4.10454810e-01 6.49113119e-01 8.18934962e-02 -6.74630642e-01
-3.99526268e-01 1.17296897e-01 -4.77400497e-02 -4.53662246e-01
-7.08199963e-02 -5.52518129e-01 -7.82599032e-01 8.06226805e-02
-6.47155881e-01 4.86224741e-01 7.29340971e-01 1.30096471e+00
5.84189653e-01 4.93279874e-01 9.58473861e-01 -7.69964099e-01
-7.14724004e-01 -1.15836024e+00 -6.96042657e-01 1.10503085e-01
5.53027630e-01 -4.94736224e-01 -4.02917415e-01 -1.23683795e-01] | [14.218816757202148, 6.341888904571533] |
5aea2eea-9de2-418c-8800-4b24e586cbea | important-object-identification-with-semi | 2203.02634 | null | https://arxiv.org/abs/2203.02634v1 | https://arxiv.org/pdf/2203.02634v1.pdf | Important Object Identification with Semi-Supervised Learning for Autonomous Driving | Accurate identification of important objects in the scene is a prerequisite for safe and high-quality decision making and motion planning of intelligent agents (e.g., autonomous vehicles) that navigate in complex and dynamic environments. Most existing approaches attempt to employ attention mechanisms to learn importance weights associated with each object indirectly via various tasks (e.g., trajectory prediction), which do not enforce direct supervision on the importance estimation. In contrast, we tackle this task in an explicit way and formulate it as a binary classification ("important" or "unimportant") problem. We propose a novel approach for important object identification in egocentric driving scenarios with relational reasoning on the objects in the scene. Besides, since human annotations are limited and expensive to obtain, we present a semi-supervised learning pipeline to enable the model to learn from unlimited unlabeled data. Moreover, we propose to leverage the auxiliary tasks of ego vehicle behavior prediction to further improve the accuracy of importance estimation. The proposed approach is evaluated on a public egocentric driving dataset (H3D) collected in complex traffic scenarios. A detailed ablative study is conducted to demonstrate the effectiveness of each model component and the training strategy. Our approach also outperforms rule-based baselines by a large margin. | ['Chiho Choi', 'Masayoshi Tomizuka', 'Hengbo Ma', 'Haiming Gang', 'Jiachen Li'] | 2022-03-05 | null | null | null | null | ['relational-reasoning'] | ['natural-language-processing'] | [-5.82927326e-03 3.08677673e-01 -5.18355489e-01 -7.36788988e-01
-6.38896883e-01 -2.90908337e-01 7.96312928e-01 1.36230499e-01
-6.64557397e-01 7.05345869e-01 1.88690856e-01 -2.59382457e-01
-1.05122730e-01 -6.00751460e-01 -9.12272811e-01 -5.74758708e-01
2.47996300e-01 5.91999829e-01 3.18178415e-01 -2.80027211e-01
2.56060541e-01 5.87010741e-01 -1.85011673e+00 -2.30413824e-01
9.01812613e-01 9.71854270e-01 3.10062498e-01 4.67702955e-01
3.17444146e-01 1.13730931e+00 -2.13390812e-01 -4.96113956e-01
1.86895758e-01 2.05579802e-01 -7.46343493e-01 1.96064472e-01
2.81955332e-01 -6.40630901e-01 -3.53465647e-01 8.93117905e-01
8.24540667e-03 4.20142382e-01 7.52760828e-01 -1.58482862e+00
-1.04593813e-01 3.25974107e-01 -3.31820011e-01 3.52861702e-01
-1.78907678e-01 4.96291965e-01 1.09740841e+00 -8.01913798e-01
5.39331675e-01 1.07252610e+00 2.44586214e-01 3.04774165e-01
-9.13150489e-01 -4.70124811e-01 6.86397254e-01 1.02750885e+00
-1.28167987e+00 -7.78604567e-01 8.23893666e-01 -5.50356805e-01
8.67768168e-01 -3.47538367e-02 5.03088474e-01 9.52016652e-01
-3.34440991e-02 1.17327881e+00 4.74342793e-01 1.85892999e-01
4.06067699e-01 1.76607952e-01 3.82801712e-01 6.16538286e-01
3.57733399e-01 1.53725743e-01 -3.14255893e-01 1.88627124e-01
1.82827175e-01 -5.22697307e-02 1.62995324e-01 -6.63310349e-01
-1.18346131e+00 6.77794039e-01 3.72106642e-01 -2.72061110e-01
-8.15051973e-01 3.29079777e-01 3.36244702e-01 -4.26467881e-02
4.68211234e-01 2.01485023e-01 -3.75270605e-01 -3.07060122e-01
-3.33336353e-01 5.05332470e-01 3.84645015e-01 1.26221824e+00
8.93372238e-01 1.62565857e-01 -4.14832532e-01 4.15478110e-01
3.72381896e-01 5.55386007e-01 -5.59229851e-02 -1.14423084e+00
6.20062232e-01 5.88360608e-01 3.72768700e-01 -7.89993644e-01
-4.96485591e-01 -4.17702705e-01 -3.92517209e-01 3.34056765e-01
3.57488394e-01 -6.83210418e-02 -8.13033819e-01 1.83012748e+00
6.59864724e-01 4.76386964e-01 8.12687650e-02 1.21696413e+00
6.07941628e-01 2.84600645e-01 3.83665174e-01 1.05648980e-01
1.39088798e+00 -1.23220098e+00 -7.28584707e-01 -2.75890172e-01
6.53914332e-01 -3.93562585e-01 9.55552876e-01 1.90620080e-01
-9.49858546e-01 -6.34240627e-01 -9.89306509e-01 -2.50009924e-01
-1.92738622e-01 1.48456305e-01 7.86924183e-01 3.15638125e-01
-7.33870268e-01 2.61423647e-01 -8.83521974e-01 -1.81456834e-01
5.95201492e-01 3.39792490e-01 -4.50718015e-01 -8.53126273e-02
-1.06995821e+00 1.25939739e+00 4.32998657e-01 -1.89639106e-02
-1.42984033e+00 -4.71247017e-01 -1.15499318e+00 4.20903787e-02
6.74841523e-01 -6.03565454e-01 1.33230591e+00 -8.17688227e-01
-1.30197501e+00 5.52725971e-01 -3.50902587e-01 -8.43251765e-01
5.39357305e-01 -4.09517109e-01 -8.58615413e-02 1.72299773e-01
2.57635713e-01 8.18169653e-01 8.08515131e-01 -1.12252879e+00
-1.11380911e+00 -2.43276983e-01 7.81322002e-01 4.70136911e-01
-1.74801037e-01 -2.36430347e-01 -6.32795155e-01 -3.97874862e-01
-2.16876119e-01 -1.03188050e+00 -4.14029986e-01 -8.71392116e-02
-2.12936893e-01 -5.66790164e-01 8.65597785e-01 -4.24352109e-01
8.36471915e-01 -1.97626221e+00 1.42262667e-01 -8.41886327e-02
3.39586735e-01 1.17451444e-01 -1.01558670e-01 -1.58102646e-01
3.10534865e-01 -3.24642032e-01 -1.84718102e-01 -5.09314001e-01
1.17284790e-01 3.65982473e-01 -1.42593816e-01 4.82777029e-01
5.52770197e-01 1.04316998e+00 -1.05601490e+00 -4.77923691e-01
5.73409736e-01 3.48062694e-01 -7.34007955e-01 3.18908542e-01
-3.22112918e-01 6.78077042e-01 -6.91821396e-01 5.53084731e-01
4.13759202e-01 -9.85834077e-02 -6.16450757e-02 -2.03107223e-01
-8.32245573e-02 3.62343937e-01 -9.47074771e-01 1.41058314e+00
-5.26907682e-01 5.92217624e-01 -9.24947783e-02 -1.12064600e+00
5.27259588e-01 1.76945716e-01 4.39551353e-01 -6.97140634e-01
2.49482423e-01 -1.27056345e-01 1.63833484e-01 -7.30071723e-01
7.84017086e-01 2.21167326e-01 -1.42350093e-01 3.31901610e-01
-9.40746218e-02 7.12532923e-02 3.55906308e-01 1.94467515e-01
1.01454830e+00 4.04507697e-01 3.62373680e-01 -2.36170143e-01
8.43119860e-01 3.31590295e-01 7.77090073e-01 7.07040191e-01
-6.83076382e-01 2.33352050e-01 4.64797288e-01 -5.01796961e-01
-1.08719623e+00 -6.13507330e-01 -5.88856731e-03 1.11476362e+00
7.80218005e-01 -2.19903111e-01 -7.57149041e-01 -9.93668497e-01
-3.26430351e-02 1.23417819e+00 -6.46906734e-01 -2.52387226e-01
-5.59255779e-01 -4.85297203e-01 1.42934293e-01 7.60900736e-01
3.95474255e-01 -8.89772356e-01 -7.80724347e-01 1.11781210e-01
-3.74561220e-01 -1.58227253e+00 -2.29249209e-01 -5.25426120e-02
-4.25641418e-01 -1.16967773e+00 -1.47334531e-01 -2.84241647e-01
7.65575767e-01 5.62754214e-01 1.04499912e+00 -1.04633838e-01
1.00777097e-01 4.46603209e-01 -2.84572810e-01 -5.93617201e-01
-6.76069930e-02 1.58180356e-01 2.53457963e-01 2.80410320e-01
7.22697079e-01 -2.74841398e-01 -6.94679379e-01 4.88511831e-01
-4.30177778e-01 3.10990751e-01 7.38289773e-01 5.92195570e-01
6.69583023e-01 2.94036940e-02 6.41545534e-01 -6.34906113e-01
2.43224233e-01 -8.64684165e-01 -7.48972952e-01 -1.10286362e-01
-5.36963999e-01 2.98003286e-01 4.94017214e-01 -2.32393831e-01
-1.21501386e+00 1.85646936e-01 -2.16050133e-01 -4.96723294e-01
-4.66407418e-01 3.52953613e-01 -4.74802762e-01 2.45638430e-01
3.48079473e-01 4.20120768e-02 -2.19285920e-01 -2.35547572e-01
4.43618655e-01 4.90534544e-01 4.58176255e-01 -5.18088758e-01
9.40027595e-01 7.07567513e-01 8.01529139e-02 -6.15587771e-01
-9.86096501e-01 -6.41722322e-01 -7.18037486e-01 -4.31193262e-01
1.02904928e+00 -1.24907720e+00 -1.02695882e+00 1.36594608e-01
-1.07325089e+00 -4.29927170e-01 -2.35232681e-01 8.04607451e-01
-7.70595431e-01 1.66677639e-01 -2.02592373e-01 -9.18928146e-01
2.44683083e-02 -1.37194264e+00 1.25707698e+00 4.66944724e-02
-1.54675364e-01 -7.24285781e-01 -4.47856396e-01 7.66306579e-01
1.98964298e-01 -5.96132800e-02 7.21590579e-01 -5.86444676e-01
-1.07112598e+00 -1.39978185e-01 -3.91517460e-01 7.62108564e-02
-2.40811724e-02 -3.34101558e-01 -1.08150876e+00 6.83935732e-02
-1.40509591e-01 -2.39369988e-01 9.26489472e-01 2.98297375e-01
1.18731570e+00 -2.45789587e-01 -2.77122438e-01 3.74212325e-01
8.16630960e-01 4.21746001e-02 4.99017090e-01 3.53209347e-01
8.45151424e-01 1.04768085e+00 1.42659259e+00 4.22712475e-01
1.15974092e+00 8.89910638e-01 9.12663758e-01 1.29850715e-01
-6.38745353e-02 -5.58317266e-02 3.03084463e-01 2.69277215e-01
-2.93257743e-01 -2.56096005e-01 -7.88170397e-01 7.58676529e-01
-2.22325587e+00 -1.10139263e+00 -2.27867529e-01 2.03281355e+00
4.35361952e-01 4.75705922e-01 8.94125998e-02 -8.83820206e-02
5.48459113e-01 5.66896759e-02 -9.71788585e-01 1.62465926e-02
2.16461346e-01 -5.69296002e-01 5.80775499e-01 6.58785641e-01
-1.37772346e+00 1.19463921e+00 5.32046413e+00 6.10336900e-01
-9.37947750e-01 2.73044378e-01 6.81566417e-01 -1.68286666e-01
-2.69325674e-01 -6.15567788e-02 -1.04669344e+00 2.93041229e-01
8.65317822e-01 -6.73373714e-02 1.71045259e-01 1.17847526e+00
5.46469510e-01 -1.20240264e-01 -1.12233806e+00 8.33198488e-01
-3.14322039e-02 -1.18975830e+00 -1.37258485e-01 -3.11499368e-02
6.44536793e-01 2.33363450e-01 -4.38640006e-02 5.72147548e-01
4.61592197e-01 -7.78024912e-01 1.07846701e+00 5.51813841e-01
3.59781384e-01 -8.42048109e-01 7.81046629e-01 5.86889088e-01
-1.21654201e+00 -2.16912940e-01 -2.84882694e-01 -1.57204732e-01
3.93536061e-01 3.26599121e-01 -9.09666657e-01 4.04902160e-01
5.77966690e-01 9.18700218e-01 -4.75100309e-01 8.14676404e-01
-6.40777826e-01 4.62948412e-01 -1.01705059e-01 4.16127630e-02
3.99012983e-01 -7.69312680e-02 8.69716704e-01 8.24613690e-01
3.19594182e-02 1.69905186e-01 1.32388219e-01 7.95626044e-01
-1.09797806e-01 -1.60711437e-01 -5.91515124e-01 1.59347415e-01
3.11557561e-01 1.38546610e+00 -5.45621336e-01 -6.59686029e-01
-5.18400371e-01 5.72345912e-01 5.03849924e-01 3.67487878e-01
-1.25923800e+00 -7.60654211e-02 1.14987338e+00 2.01533079e-01
3.95452827e-01 -3.09158802e-01 -2.29401127e-01 -1.11113214e+00
1.55856848e-01 -5.12415648e-01 1.68458402e-01 -7.04939187e-01
-8.34835768e-01 5.39733768e-01 1.97081268e-01 -1.37263525e+00
-4.43642706e-01 -4.99399483e-01 -4.38780934e-01 7.01493025e-01
-2.19970751e+00 -1.30981565e+00 -5.94702005e-01 4.77532506e-01
9.73629475e-01 -2.22375169e-01 2.30953828e-01 2.68783718e-01
-6.85499907e-01 3.56163561e-01 -3.06409597e-01 -1.85467109e-01
5.26818991e-01 -1.12750947e+00 5.20823777e-01 1.01849341e+00
-2.79478490e-01 3.49822640e-01 7.54305840e-01 -6.83530271e-01
-1.42969120e+00 -1.53043592e+00 8.24735463e-01 -7.69451261e-01
4.90247190e-01 -3.62428457e-01 -6.24437749e-01 6.95788682e-01
-1.03534497e-01 4.44312431e-02 4.31031883e-01 1.28456011e-01
-1.10605516e-01 -1.33803055e-01 -9.00257587e-01 7.70916164e-01
1.27131224e+00 -3.08117062e-01 -5.73818505e-01 2.11639777e-01
7.93734789e-01 -2.19053432e-01 -5.36764503e-01 6.82897031e-01
4.32612062e-01 -7.17314661e-01 9.88181174e-01 -7.59960413e-01
4.52760428e-01 -5.93103528e-01 -2.02054441e-01 -9.66957927e-01
-3.16034555e-01 -2.08069816e-01 -5.48226833e-01 8.67626250e-01
3.55262935e-01 -5.07141352e-01 9.24326062e-01 8.82835686e-01
-5.27487874e-01 -5.98807633e-01 -7.14489937e-01 -5.29153585e-01
-4.86000568e-01 -9.07308221e-01 8.10391843e-01 6.63937151e-01
-2.80670553e-01 5.08313656e-01 -4.17130649e-01 5.65539539e-01
8.43774498e-01 -3.33036855e-02 1.12297618e+00 -1.15751040e+00
-9.58471280e-03 -3.77198398e-01 -5.66223443e-01 -1.18381846e+00
5.79159379e-01 -6.73500121e-01 3.53214294e-01 -1.43726492e+00
1.84343085e-01 -7.02464938e-01 -4.50873911e-01 3.77688169e-01
-5.30149519e-01 1.92070127e-01 -2.06392258e-02 3.30206156e-01
-1.11907375e+00 8.49716544e-01 1.07175708e+00 -4.37201291e-01
-7.51721188e-02 2.47152433e-01 -7.24432766e-01 9.74560797e-01
7.59732902e-01 -3.57492357e-01 -8.49706292e-01 -5.01374125e-01
1.10964291e-01 3.38707939e-02 5.54494977e-01 -9.21459615e-01
3.63809317e-01 -5.66828728e-01 6.52383938e-02 -8.69715154e-01
6.68579817e-01 -9.76773441e-01 -2.91613638e-01 1.76334441e-01
-3.50541264e-01 -2.35464662e-01 1.13799378e-01 7.28758574e-01
-1.45356610e-01 -2.30266020e-01 5.30296922e-01 2.10158780e-01
-1.34468389e+00 6.58301890e-01 -3.40732217e-01 -1.11622989e-01
1.47515082e+00 -8.56469050e-02 -1.63387582e-01 -4.37628657e-01
-4.89583313e-01 6.71737611e-01 3.36466670e-01 6.96879923e-01
5.71366549e-01 -1.30249679e+00 -7.98659682e-01 1.09989554e-01
6.14347696e-01 1.61859125e-01 3.84014785e-01 1.10106075e+00
-1.32520020e-01 6.20512128e-01 -2.05511346e-01 -6.71872675e-01
-1.03994513e+00 6.87938869e-01 1.93107605e-01 -6.44714236e-02
-4.67014462e-01 4.87318844e-01 7.25755990e-01 -3.78523260e-01
1.69196382e-01 -9.86504108e-02 -7.63130128e-01 6.90175071e-02
5.84706306e-01 3.43427479e-01 3.92946713e-02 -1.13081682e+00
-3.68601143e-01 1.91958532e-01 -2.00099275e-01 1.22713290e-01
1.25768769e+00 -5.03224909e-01 4.21872199e-01 1.08717531e-01
7.87189901e-01 -3.24677676e-01 -1.85368121e+00 -2.74503261e-01
5.08481264e-02 -4.15738583e-01 2.43329778e-01 -4.12375301e-01
-9.10416245e-01 1.00703192e+00 3.38453472e-01 -2.94374555e-01
7.84248948e-01 2.51487512e-02 5.66360831e-01 7.16608524e-01
4.39928472e-01 -1.19994307e+00 2.93677021e-02 5.41852534e-01
7.65209973e-01 -1.74730647e+00 -1.40371487e-01 -3.00867707e-01
-1.06173217e+00 6.73585773e-01 8.89548242e-01 3.12838137e-01
4.34188545e-01 -3.03344075e-02 -2.75641698e-02 -1.54577121e-01
-1.04070878e+00 -7.42839873e-01 4.95083392e-01 5.83304584e-01
6.73484828e-05 -5.31923361e-02 -6.70578033e-02 5.23834705e-01
6.41992614e-02 -1.58783481e-01 3.69549364e-01 7.30626464e-01
-5.62052846e-01 -7.71876454e-01 -6.81352988e-02 4.25425291e-01
-2.17291817e-01 1.37821540e-01 9.77163389e-02 6.52291417e-01
1.23742439e-01 1.12706029e+00 1.06391706e-01 -3.60683322e-01
3.95869017e-01 -2.08714753e-01 3.25804614e-02 -5.17530143e-01
-7.93674886e-02 -2.91119576e-01 3.54682237e-01 -7.58902729e-01
-4.69858587e-01 -8.22218120e-01 -1.26073313e+00 -3.43373656e-01
-1.71891347e-01 -3.93556058e-02 6.72623098e-01 1.13187814e+00
4.65473145e-01 7.80345142e-01 5.67780614e-01 -9.96413350e-01
-4.02512282e-01 -6.95376992e-01 -9.27141383e-02 6.52703404e-01
3.50937784e-01 -1.05723417e+00 -7.18052685e-02 1.37619317e-01] | [6.249282360076904, 0.6862313747406006] |
242ef5b8-79dc-4fff-b2fe-3cffd6e8e67f | spec-summary-preference-decomposition-for-low | 2303.14011 | null | https://arxiv.org/abs/2303.14011v1 | https://arxiv.org/pdf/2303.14011v1.pdf | SPEC: Summary Preference Decomposition for Low-Resource Abstractive Summarization | Neural abstractive summarization has been widely studied and achieved great success with large-scale corpora. However, the considerable cost of annotating data motivates the need for learning strategies under low-resource settings. In this paper, we investigate the problems of learning summarizers with only few examples and propose corresponding methods for improvements. First, typical transfer learning methods are prone to be affected by data properties and learning objectives in the pretext tasks. Therefore, based on pretrained language models, we further present a meta learning framework to transfer few-shot learning processes from source corpora to the target corpus. Second, previous methods learn from training examples without decomposing the content and preference. The generated summaries could therefore be constrained by the preference bias in the training set, especially under low-resource settings. As such, we propose decomposing the contents and preferences during learning through the parameter modulation, which enables control over preferences during inference. Third, given a target application, specifying required preferences could be non-trivial because the preferences may be difficult to derive through observations. Therefore, we propose a novel decoding method to automatically estimate suitable preferences and generate corresponding summary candidates from the few training examples. Extensive experiments demonstrate that our methods achieve state-of-the-art performance on six diverse corpora with 30.11%/33.95%/27.51% and 26.74%/31.14%/24.48% average improvements on ROUGE-1/2/L under 10- and 100-example settings. | ['Hong-Han Shuai', 'Yun-Zhu Song', 'Yi-Syuan Chen'] | 2023-03-24 | null | null | null | null | ['abstractive-text-summarization'] | ['natural-language-processing'] | [ 4.27010387e-01 -3.84147577e-02 -5.28347194e-01 -4.74884421e-01
-1.26402545e+00 -4.53256607e-01 5.72608948e-01 2.91718185e-01
-6.47590756e-01 1.07454014e+00 4.04073626e-01 3.54470394e-04
-1.17522292e-01 -6.13790572e-01 -7.67548621e-01 -6.33799911e-01
1.07339084e-01 5.26669502e-01 5.41645996e-02 -1.87537506e-01
4.25406903e-01 -1.80780485e-01 -1.60350037e+00 5.09732723e-01
1.54388952e+00 7.98973382e-01 6.37865245e-01 6.38422191e-01
-2.73393005e-01 4.64973658e-01 -8.11026096e-01 -4.17601883e-01
-1.20948002e-01 -5.74962795e-01 -6.70511365e-01 7.30587915e-02
3.23693633e-01 -2.50650585e-01 1.61922406e-02 1.06273735e+00
7.36336887e-01 4.06685472e-01 9.47884500e-01 -7.74242580e-01
-5.82810581e-01 1.13137341e+00 -3.73340398e-01 1.10843383e-01
2.50311404e-01 -1.40140932e-02 1.22748518e+00 -9.41129208e-01
3.28038305e-01 9.72721815e-01 2.03742266e-01 7.32721925e-01
-1.11052668e+00 -4.91716892e-01 3.76358390e-01 3.66567552e-01
-1.08151317e+00 -6.31715596e-01 8.06994975e-01 -8.02381188e-02
7.82994330e-01 3.75189036e-01 2.97107399e-01 1.27442837e+00
-1.24715604e-01 1.23233354e+00 8.06031883e-01 -5.14247417e-01
3.79142970e-01 2.89062440e-01 2.20188335e-01 3.56619626e-01
4.94513810e-01 -5.37790775e-01 -6.93181813e-01 -5.96405976e-02
1.94625258e-01 -3.83201987e-01 -4.31956947e-01 -1.21841229e-01
-1.10632050e+00 7.94045687e-01 9.44859982e-02 1.55171111e-01
-1.93116337e-01 -1.96313500e-01 5.28241217e-01 2.18474135e-01
6.80189908e-01 7.34411716e-01 -5.11599064e-01 -3.47839415e-01
-1.07982457e+00 1.26063973e-01 7.23889351e-01 1.19887662e+00
7.14812517e-01 4.02603336e-02 -4.30723697e-01 1.31892717e+00
7.57406130e-02 3.54258448e-01 8.65690708e-01 -6.98798776e-01
9.56658602e-01 3.26537997e-01 1.82900354e-01 -7.15002060e-01
-1.46609858e-01 -6.37302220e-01 -9.43178296e-01 -6.21234000e-01
2.19986409e-01 -3.26986879e-01 -6.67330623e-01 1.75160539e+00
-3.76302339e-02 7.38734305e-02 3.77912909e-01 6.65236890e-01
8.87828231e-01 9.98580933e-01 1.91645380e-02 -8.04911673e-01
1.05270004e+00 -1.17685926e+00 -7.77608693e-01 -6.11048222e-01
7.95336366e-01 -5.81853151e-01 1.58228910e+00 3.53930593e-01
-1.25734091e+00 -4.08963740e-01 -1.06652105e+00 7.83858970e-02
3.67600173e-02 4.79629904e-01 3.50977272e-01 3.71375948e-01
-6.73833311e-01 6.55566990e-01 -6.82211220e-01 -2.45253563e-01
2.95230776e-01 3.31277817e-01 2.12706462e-01 -2.08127528e-01
-1.29048383e+00 7.50747204e-01 6.64172053e-01 -1.50095463e-01
-6.59479380e-01 -6.10606432e-01 -6.97849214e-01 3.77138674e-01
5.36917984e-01 -6.54816329e-01 1.47574639e+00 -8.77880275e-01
-1.72134721e+00 4.13904041e-01 -1.64568186e-01 -3.38635385e-01
4.90818828e-01 -4.38954443e-01 -2.25891873e-01 -5.47245294e-02
3.38992961e-02 4.30903286e-01 5.22569060e-01 -1.17893755e+00
-7.06600904e-01 -4.26412523e-02 -1.17348671e-01 5.71109295e-01
-8.90866935e-01 -1.82715729e-01 -5.60516179e-01 -7.69108236e-01
-1.94215000e-01 -7.60909677e-01 -1.65097758e-01 -6.88648999e-01
-4.27654922e-01 -4.57579046e-01 2.50557691e-01 -5.05865216e-01
1.49367237e+00 -1.95897520e+00 2.40079120e-01 -1.79428816e-01
-3.94266784e-01 4.66091365e-01 -3.19353014e-01 4.02031988e-01
4.51039374e-01 1.52167901e-01 -3.42352778e-01 -3.57843995e-01
8.99811275e-03 4.27063741e-02 -4.44565207e-01 -2.11703274e-02
2.88877673e-02 5.88696122e-01 -1.10107195e+00 -5.97248971e-01
-1.43672824e-01 1.62866086e-01 -6.82856619e-01 5.65762699e-01
-3.39612216e-01 1.47588462e-01 -7.34325707e-01 3.96896273e-01
2.92993784e-01 -1.97549149e-01 2.00831294e-01 -1.65999532e-01
2.12563518e-02 5.98215103e-01 -1.02095103e+00 1.91671109e+00
-7.51772881e-01 3.39436650e-01 -2.39363045e-01 -1.25561178e+00
9.41219211e-01 2.53332853e-01 9.75006223e-02 -4.72728074e-01
2.14655474e-01 4.19461340e-01 1.54719669e-02 -6.45179987e-01
7.35955000e-01 -7.23030791e-02 -3.23533982e-01 4.46772218e-01
1.02061577e-01 -2.06703380e-01 5.55158734e-01 2.12217361e-01
8.07839394e-01 -4.34300005e-02 3.87449056e-01 -2.73392767e-01
4.37587678e-01 -1.53731748e-01 7.39144862e-01 8.60024631e-01
1.42619580e-01 6.50028288e-01 3.63381475e-01 7.81496614e-03
-6.50940180e-01 -7.46526241e-01 1.09909043e-01 1.39416325e+00
6.28003329e-02 -5.76009929e-01 -7.33654737e-01 -6.02979839e-01
-4.58283871e-01 1.17760158e+00 -2.58028299e-01 -3.21048528e-01
-6.68232977e-01 -9.31605697e-01 3.19587439e-01 4.85414207e-01
3.54791552e-01 -9.92232800e-01 -4.64149386e-01 3.94888043e-01
-4.48691219e-01 -9.90164578e-01 -6.52997553e-01 1.80620849e-01
-1.11853671e+00 -6.39876306e-01 -8.60750914e-01 -7.81776130e-01
7.22525775e-01 2.52686590e-01 1.12202358e+00 -1.17880367e-01
3.56274515e-01 -8.92053023e-02 -4.83005822e-01 -3.85696322e-01
-3.93670589e-01 6.25451207e-01 2.32727855e-01 -1.17719226e-01
1.38793588e-01 -6.12376451e-01 -3.47014427e-01 4.81327064e-02
-7.95721650e-01 4.27046776e-01 9.33570087e-01 1.00234890e+00
5.04896581e-01 -1.90551400e-01 1.01769519e+00 -1.01628530e+00
9.88389850e-01 -4.79770064e-01 -2.79864103e-01 6.32538080e-01
-6.63916826e-01 3.34957719e-01 1.03225136e+00 -7.00788736e-01
-1.28448117e+00 -1.47845268e-01 1.90844610e-01 -1.08067073e-01
-3.81163470e-02 8.26644897e-01 -5.88524640e-01 6.61358118e-01
6.98694706e-01 3.78869534e-01 -3.31553161e-01 -4.40148711e-01
2.91469485e-01 1.05244255e+00 3.31193835e-01 -9.33670223e-01
4.64054316e-01 -1.45135805e-01 -5.98103881e-01 -8.04880023e-01
-1.22457278e+00 -2.91949630e-01 -3.61047179e-01 4.34425995e-02
3.74009311e-01 -9.07577395e-01 -7.19556361e-02 1.18889757e-01
-1.12849534e+00 -3.77734333e-01 -4.26066890e-02 6.45338953e-01
-6.14550531e-01 4.13125426e-01 -4.64444518e-01 -6.42701507e-01
-7.72722721e-01 -1.24049520e+00 7.46930122e-01 3.52411777e-01
-4.35687006e-01 -8.32814217e-01 -1.76975876e-02 3.09376478e-01
4.61692452e-01 -2.74300605e-01 1.02670777e+00 -9.69345510e-01
-2.83916324e-01 4.24561799e-02 4.89812754e-02 3.41493398e-01
3.35783809e-01 -6.49208426e-02 -8.06826770e-01 -3.32675368e-01
-5.28649800e-02 -5.82322717e-01 8.57118428e-01 3.02850068e-01
1.44891274e+00 -7.43435740e-01 -1.33734480e-01 3.46511602e-01
1.12107611e+00 7.42185265e-02 2.43542820e-01 2.82095015e-01
6.12236917e-01 7.33745873e-01 9.02220905e-01 5.91568768e-01
3.11618805e-01 5.61770141e-01 -2.45323032e-02 5.91159344e-01
7.76629522e-02 -2.54192710e-01 5.89079499e-01 1.41257823e+00
-3.54430676e-02 -4.68837798e-01 -7.16447353e-01 6.11906707e-01
-1.98728907e+00 -7.15689600e-01 3.22211981e-01 2.24704909e+00
1.51716805e+00 3.18747014e-01 5.48695028e-03 -8.46707597e-02
8.25660408e-01 2.71058470e-01 -6.84180975e-01 -3.02315325e-01
5.28846532e-02 -8.90027806e-02 1.91327900e-01 3.86841327e-01
-8.42836857e-01 9.24308300e-01 5.01758909e+00 1.12878919e+00
-1.02928782e+00 -1.34900197e-01 7.65942395e-01 -4.55932707e-01
-6.02269232e-01 -2.34070122e-01 -9.89802718e-01 6.86096966e-01
1.13740993e+00 -6.95437908e-01 3.01059395e-01 7.21651793e-01
1.93216994e-01 6.32886738e-02 -1.33569717e+00 8.64104509e-01
2.47201040e-01 -1.19815218e+00 3.61134648e-01 -3.00106049e-01
8.90052974e-01 -1.16881862e-01 -3.65375727e-02 6.30895197e-01
1.59573451e-01 -7.81627417e-01 5.99325061e-01 2.00973690e-01
8.24832618e-01 -9.33549762e-01 6.55252099e-01 8.84672582e-01
-6.44525707e-01 3.90834641e-03 -6.86033547e-01 1.01079661e-02
7.01913312e-02 6.44401848e-01 -9.25267875e-01 4.88590896e-01
2.69169509e-01 7.11612225e-01 -3.97437841e-01 1.01248693e+00
-3.97176832e-01 8.68941844e-01 -1.72313660e-01 -6.10142887e-01
1.64465189e-01 -5.33387922e-02 4.73562896e-01 1.37802672e+00
6.35613501e-01 1.00818455e-01 1.05925485e-01 5.36801755e-01
-4.30606931e-01 3.82834047e-01 -2.49715507e-01 -6.69895904e-03
7.53431976e-01 1.07681835e+00 -4.07321990e-01 -4.52072650e-01
-2.16210410e-01 6.34100437e-01 5.54715395e-01 4.23021764e-01
-6.97999477e-01 -4.42749321e-01 3.05221409e-01 -1.72864109e-01
2.33358309e-01 1.31948680e-01 -3.26425016e-01 -1.48779881e+00
1.16500765e-01 -1.11554515e+00 4.16552186e-01 -3.76777798e-01
-1.16123962e+00 6.90525949e-01 3.11573982e-01 -1.33653879e+00
-4.08715516e-01 -1.77334011e-01 -8.39603901e-01 4.68290836e-01
-1.66506886e+00 -6.65987194e-01 -4.02374566e-02 5.30842841e-02
1.18937218e+00 -3.54781002e-01 8.60990226e-01 7.41889467e-03
-7.69458473e-01 8.51971805e-01 3.52392226e-01 -2.31470138e-01
1.03850317e+00 -1.25913155e+00 8.91477317e-02 7.81501472e-01
1.70099750e-01 5.86953759e-01 9.89450872e-01 -4.01813567e-01
-1.34899378e+00 -1.06831348e+00 9.94568884e-01 -3.63792889e-02
5.27812243e-01 -3.22458178e-01 -1.08430791e+00 3.44954938e-01
3.35774779e-01 -2.90796638e-01 9.20490861e-01 2.58178413e-01
-9.34625119e-02 -2.48620942e-01 -8.13722491e-01 8.92551243e-01
8.75614464e-01 -1.87978938e-01 -8.56145561e-01 3.03997755e-01
6.50416195e-01 -3.84931475e-01 -6.39703572e-01 4.59688574e-01
3.30504060e-01 -5.04359066e-01 6.51992619e-01 -5.20027220e-01
6.51492000e-01 5.38085438e-02 -1.76888689e-01 -1.78886449e+00
-1.50654793e-01 -6.25299156e-01 -3.71254772e-01 1.56438720e+00
7.09048510e-01 -2.44707003e-01 7.28711367e-01 6.94064498e-01
-3.53517383e-01 -9.19239044e-01 -6.60457432e-01 -6.78691745e-01
2.43437394e-01 -1.81429669e-01 5.72983563e-01 8.51849318e-01
2.71328360e-01 7.53086329e-01 -4.75710779e-01 -1.06526151e-01
5.36105335e-01 2.93581992e-01 6.95479929e-01 -9.85517979e-01
-4.24301237e-01 -4.66188133e-01 4.22086000e-01 -1.37366390e+00
4.42304909e-01 -8.20212841e-01 3.78326654e-01 -1.57598197e+00
3.47928286e-01 -4.53668833e-01 -4.72128153e-01 2.46892810e-01
-7.24982977e-01 -3.05323571e-01 7.97671154e-02 3.07355225e-01
-8.27446461e-01 8.70123029e-01 1.14803600e+00 -2.96327353e-01
-2.97120035e-01 2.61547327e-01 -9.77848053e-01 5.99077225e-01
1.15013027e+00 -4.79788780e-01 -7.51676798e-01 -7.52454042e-01
2.33533457e-01 9.59847718e-02 -3.54989648e-01 -7.06524134e-01
3.53887320e-01 -2.96045542e-01 -2.28393506e-02 -5.34670591e-01
2.37321153e-01 -3.73238087e-01 -4.51422125e-01 1.28214464e-01
-8.67255986e-01 -1.60798550e-01 -3.42872366e-02 6.31571531e-01
-2.71663427e-01 -6.29970551e-01 5.07801712e-01 -1.63642555e-01
-6.21054173e-01 4.85925861e-02 -2.02981889e-01 4.33699846e-01
6.79794610e-01 -1.63253695e-02 -3.33072364e-01 -4.76588994e-01
-3.09680223e-01 4.79611039e-01 2.45911449e-01 5.20350039e-01
5.63459694e-01 -1.21291602e+00 -9.92066562e-01 -5.32438681e-02
2.04718947e-01 3.63303035e-01 2.02646837e-01 4.65911478e-01
7.98860788e-02 3.44703794e-01 4.20453697e-02 -3.45522612e-01
-1.26278532e+00 2.49495804e-01 -1.40183404e-01 -4.01261598e-01
-2.26906776e-01 8.20908904e-01 6.82812706e-02 -2.47078314e-01
4.77965593e-01 -3.79405588e-01 -4.28560704e-01 3.48135084e-01
6.42429948e-01 4.00685281e-01 1.08218871e-01 -2.22492844e-01
-9.62640718e-02 2.77410299e-01 -4.26635236e-01 1.10999038e-02
1.53210902e+00 -9.08877328e-02 1.95874974e-01 6.55184984e-01
1.10279047e+00 -5.93890660e-02 -1.25703824e+00 -5.24516582e-01
2.66181231e-01 -2.31961101e-01 -1.04757011e-01 -6.98214352e-01
-6.90690577e-01 9.82523382e-01 -1.23621173e-01 2.13992745e-02
1.08112276e+00 -1.39643148e-01 8.56937408e-01 7.59896934e-01
2.03367755e-01 -1.39834392e+00 1.99232757e-01 6.94991469e-01
7.95625865e-01 -1.23534358e+00 8.83924589e-02 -2.11971864e-01
-8.77102375e-01 1.03530622e+00 7.35124767e-01 1.53385326e-01
1.15152538e-01 -5.96202724e-02 -6.58010915e-02 2.57570654e-01
-1.13535690e+00 2.08026037e-01 5.03215194e-01 1.82709008e-01
6.77240193e-01 1.00194529e-01 -5.51738679e-01 1.05616474e+00
-2.90012419e-01 -2.34657735e-01 6.52607739e-01 7.66329169e-01
-8.26917946e-01 -1.21106875e+00 2.75810491e-02 6.00080669e-01
-5.58819711e-01 -1.81148529e-01 -2.71547019e-01 4.22291994e-01
-3.99826676e-01 1.00364482e+00 -1.63507968e-01 -1.94002464e-01
3.62391531e-01 2.00724006e-01 3.07491362e-01 -1.01525354e+00
-4.00095433e-01 2.47497186e-01 3.55669707e-01 2.65613035e-03
-4.78466123e-01 -6.27833724e-01 -1.14569533e+00 6.36220053e-02
-6.40674651e-01 6.84041858e-01 3.73824745e-01 1.02061999e+00
4.35408384e-01 5.54725051e-01 6.75697148e-01 -7.33575225e-01
-1.17861903e+00 -1.31752610e+00 -2.99465686e-01 1.88524872e-01
6.49216473e-02 -3.92417610e-01 -4.68179107e-01 -1.74428932e-02] | [11.892318725585938, 8.942936897277832] |
445c5168-b440-4495-a033-a403fff387da | diffsound-discrete-diffusion-model-for-text | 2207.09983 | null | https://arxiv.org/abs/2207.09983v2 | https://arxiv.org/pdf/2207.09983v2.pdf | Diffsound: Discrete Diffusion Model for Text-to-sound Generation | Generating sound effects that humans want is an important topic. However, there are few studies in this area for sound generation. In this study, we investigate generating sound conditioned on a text prompt and propose a novel text-to-sound generation framework that consists of a text encoder, a Vector Quantized Variational Autoencoder (VQ-VAE), a decoder, and a vocoder. The framework first uses the decoder to transfer the text features extracted from the text encoder to a mel-spectrogram with the help of VQ-VAE, and then the vocoder is used to transform the generated mel-spectrogram into a waveform. We found that the decoder significantly influences the generation performance. Thus, we focus on designing a good decoder in this study. We begin with the traditional autoregressive decoder, which has been proved as a state-of-the-art method in previous sound generation works. However, the AR decoder always predicts the mel-spectrogram tokens one by one in order, which introduces the unidirectional bias and accumulation of errors problems. Moreover, with the AR decoder, the sound generation time increases linearly with the sound duration. To overcome the shortcomings introduced by AR decoders, we propose a non-autoregressive decoder based on the discrete diffusion model, named Diffsound. Specifically, the Diffsound predicts all of the mel-spectrogram tokens in one step and then refines the predicted tokens in the next step, so the best-predicted results can be obtained after several steps. Our experiments show that our proposed Diffsound not only produces better text-to-sound generation results when compared with the AR decoder but also has a faster generation speed, e.g., MOS: 3.56 \textit{v.s} 2.786, and the generation speed is five times faster than the AR decoder. | ['Dong Yu', 'Yuexian Zou', 'Chao Weng', 'Wen Wang', 'Helin Wang', 'Jianwei Yu', 'Dongchao Yang'] | 2022-07-20 | null | null | null | null | ['audio-generation'] | ['audio'] | [ 4.81292512e-03 -7.72976205e-02 3.47781986e-01 4.71207462e-02
-7.84886301e-01 -1.42092392e-01 2.89507508e-01 -2.17325240e-01
-1.20664202e-01 5.77584505e-01 4.13204253e-01 -1.82704881e-01
2.54685879e-01 -8.97965014e-01 -4.95553464e-01 -8.60200047e-01
5.92599034e-01 -2.22619902e-02 3.87209624e-01 -3.24731797e-01
1.51094839e-01 -2.56976962e-01 -1.51683056e+00 3.96929085e-02
1.00206411e+00 9.20319796e-01 7.43777692e-01 9.13421631e-01
-2.64448881e-01 8.13377917e-01 -7.52722025e-01 -4.19790417e-01
-1.24576397e-01 -1.11783957e+00 -2.67613322e-01 -2.84774095e-01
-3.03587824e-01 -4.11077440e-01 -4.47619706e-01 1.03338480e+00
8.22160244e-01 2.94645518e-01 8.94593894e-01 -8.60864282e-01
-9.02226865e-01 9.63957787e-01 -2.45274737e-01 3.73922102e-02
2.61944383e-01 -6.44291111e-04 1.00277853e+00 -8.15930247e-01
1.72189534e-01 1.14722705e+00 4.49607015e-01 6.60341620e-01
-8.34515452e-01 -8.24522793e-01 -1.63632751e-01 -1.43166939e-02
-1.50745165e+00 -5.43996096e-01 1.03433502e+00 -4.51672524e-01
7.77474344e-01 3.11101347e-01 5.89030266e-01 1.09697378e+00
5.03150046e-01 5.86961985e-01 6.59328163e-01 -6.06718123e-01
2.93193489e-01 1.45886123e-01 -1.63471535e-01 4.66327637e-01
-1.53120250e-01 2.47713923e-01 -4.35676515e-01 1.55258253e-01
9.27541971e-01 -2.09826022e-01 -3.20281953e-01 5.70468843e-01
-8.91079128e-01 8.83657396e-01 7.35349804e-02 5.64040065e-01
-5.71965277e-01 2.78616339e-01 1.19847007e-01 1.42736450e-01
4.99850839e-01 2.06580043e-01 -1.66316047e-01 -5.04599214e-01
-1.23124385e+00 2.96944469e-01 6.96339011e-01 6.91160262e-01
3.54809582e-01 6.68715596e-01 -4.27672207e-01 1.04710412e+00
6.35378361e-01 7.36081541e-01 7.96742380e-01 -6.60489023e-01
4.76690114e-01 -4.25036773e-02 -2.34043784e-02 -1.00631368e+00
-8.24263766e-02 -3.79972726e-01 -9.98297870e-01 -1.10552318e-01
6.89104646e-02 -5.82863092e-01 -6.51733279e-01 1.71891046e+00
1.25170127e-01 2.42658392e-01 2.07726598e-01 8.19361627e-01
8.61868978e-01 1.27425170e+00 -1.55345798e-01 -6.64041758e-01
1.17429578e+00 -9.90381360e-01 -1.27917731e+00 1.72034428e-01
3.26676607e-01 -1.04790711e+00 1.17303240e+00 4.40399945e-01
-1.14027166e+00 -1.03067005e+00 -1.04000378e+00 4.01869118e-02
2.20306795e-02 2.89594352e-01 -8.72568414e-03 7.30280936e-01
-8.92359197e-01 5.44152200e-01 -6.50780737e-01 1.38099760e-01
-1.79710686e-01 -9.23693553e-02 3.38607818e-01 4.71359938e-01
-1.51083851e+00 5.26539743e-01 4.42811877e-01 7.35042766e-02
-8.18358302e-01 -4.19565111e-01 -7.17180490e-01 2.31798381e-01
1.76318437e-01 -5.63718736e-01 1.46290231e+00 -6.69615924e-01
-2.15143895e+00 -1.59316763e-01 -3.81226063e-01 -2.46276483e-01
3.90086293e-01 -1.80722252e-01 -7.95458913e-01 -7.40359277e-02
-9.70048904e-02 4.20961499e-01 1.12271154e+00 -1.01051950e+00
-7.05711126e-01 2.04917148e-01 -2.33826801e-01 2.21888959e-01
-5.30025959e-01 -1.60942599e-01 -4.20771569e-01 -1.03038836e+00
-9.49643180e-03 -7.94079900e-01 -1.14382999e-02 -7.07646966e-01
-5.18176079e-01 -3.41176629e-01 5.59922516e-01 -8.27373683e-01
1.94097626e+00 -2.37455201e+00 -1.82371971e-03 -1.57375142e-01
-5.90010779e-03 2.78495520e-01 -5.22856340e-02 5.45997679e-01
1.12342924e-01 1.94726497e-01 -1.46944106e-01 -4.07580763e-01
1.16507418e-01 -9.15501490e-02 -5.72045863e-01 -3.54968235e-02
3.43363434e-02 7.06721246e-01 -8.79355609e-01 -5.75521231e-01
1.31065592e-01 6.79090500e-01 -7.63127685e-01 5.10223508e-01
-2.19597772e-01 4.13105965e-01 -4.70856518e-01 5.64762466e-02
4.62556720e-01 7.39000142e-02 -2.13170618e-01 -7.72909373e-02
-4.90383595e-01 3.72225106e-01 -1.29060936e+00 1.38047564e+00
-6.39995992e-01 5.37660778e-01 -2.74983823e-01 -6.10640943e-01
1.30936360e+00 7.12190807e-01 2.88691849e-01 -5.96396089e-01
3.03511441e-01 3.05326730e-01 9.94978622e-02 -6.67701840e-01
7.39887714e-01 -4.61203605e-01 1.48066238e-01 3.39146048e-01
4.74973917e-02 -4.76192534e-01 1.52893960e-01 -9.87661779e-02
6.33692086e-01 6.14274479e-02 1.53662786e-01 2.10530728e-01
5.54726660e-01 -4.70245451e-01 4.86739427e-01 5.82948029e-01
2.04544961e-01 7.52993822e-01 3.75759691e-01 2.03438714e-01
-1.02167130e+00 -8.99904728e-01 -4.49574478e-02 8.98332357e-01
8.74690339e-02 -6.18898153e-01 -1.07289374e+00 -2.54050314e-01
-4.45024043e-01 1.17400455e+00 -1.86456740e-01 -2.88729876e-01
-3.68056238e-01 -4.36364591e-01 7.48319507e-01 3.77871364e-01
4.79416341e-01 -1.32536352e+00 -3.04245800e-01 6.33678734e-01
-3.56311619e-01 -6.70892715e-01 -8.16330314e-01 -1.82472005e-01
-5.43723583e-01 -3.52299958e-01 -1.08825922e+00 -7.75742650e-01
2.19662532e-01 -5.96052893e-02 6.07669711e-01 -1.25296414e-01
3.08361113e-01 -1.24266014e-01 -7.96163142e-01 -6.17173970e-01
-7.42856801e-01 4.86194603e-02 1.53610274e-01 1.86386451e-01
1.39781088e-02 -5.71818829e-01 -5.57829320e-01 5.91666140e-02
-1.01331067e+00 2.68099248e-01 5.59636295e-01 8.11292827e-01
5.97107530e-01 4.90417957e-01 7.97685981e-01 -5.73148549e-01
1.14567006e+00 -4.67734218e-01 -4.71663117e-01 -6.00973815e-02
-6.00922287e-01 -1.24351615e-02 1.07965016e+00 -6.30242825e-01
-1.22558403e+00 -2.22269371e-01 -7.89062500e-01 -4.31532502e-01
-8.74540303e-03 6.35139883e-01 -8.22637454e-02 4.92946506e-01
5.58544695e-01 7.04440117e-01 -3.01110893e-01 -5.28155506e-01
3.27142358e-01 1.22779775e+00 3.68951172e-01 -2.30316281e-01
7.24842787e-01 -2.23920614e-01 -4.46928829e-01 -1.08510566e+00
-4.67255443e-01 4.41974290e-02 -2.80257255e-01 -1.80281043e-01
1.08025181e+00 -6.96644783e-01 -5.61363220e-01 5.16776264e-01
-1.48788214e+00 -2.28691325e-01 -3.13593596e-01 9.10316229e-01
-5.29803038e-01 3.21910292e-01 -8.55030119e-01 -1.23101711e+00
-5.19361436e-01 -1.18382370e+00 8.27977479e-01 3.93576890e-01
-2.12363571e-01 -7.73159802e-01 3.02560508e-01 5.00362851e-02
5.22040844e-01 -2.00908899e-01 8.03780913e-01 -4.21066940e-01
-2.78416216e-01 5.18298633e-02 1.72365636e-01 5.93166649e-01
2.74982959e-01 6.38971031e-02 -9.93178248e-01 3.25423889e-02
2.85372674e-01 6.65211901e-02 6.63264513e-01 5.59791327e-01
1.11782598e+00 -4.76255685e-01 1.27927497e-01 4.87164348e-01
1.11913252e+00 8.44481468e-01 7.64409006e-01 -2.18219042e-01
6.48868501e-01 2.25287735e-01 5.07063210e-01 6.12685204e-01
3.89909476e-01 5.26251972e-01 6.72029555e-02 -4.03229743e-02
-2.52257138e-01 -7.78862774e-01 5.80926180e-01 1.82426465e+00
-1.60842374e-01 -7.12234855e-01 -4.96151030e-01 4.28450644e-01
-1.56801641e+00 -9.11454678e-01 -3.18741947e-01 2.14244199e+00
1.09351194e+00 7.20635951e-02 -8.79242122e-02 4.94520903e-01
6.54376388e-01 3.58204752e-01 -2.95737863e-01 -5.08035004e-01
2.64093071e-01 3.78775835e-01 -9.83696803e-03 5.22719264e-01
-6.15881383e-01 9.41704571e-01 6.02540970e+00 1.19998026e+00
-1.24802685e+00 1.53475374e-01 2.97994018e-01 1.35064662e-01
-4.64910984e-01 -5.11266515e-02 -8.08469296e-01 9.10094023e-01
1.16334546e+00 -3.66546720e-01 6.27664864e-01 5.85959613e-01
5.85534096e-01 1.97499350e-01 -8.41571510e-01 1.04756677e+00
1.06215656e-01 -9.52833831e-01 1.57400712e-01 -1.19644247e-01
6.27395511e-01 -5.95665514e-01 1.18066505e-01 4.86163646e-01
-5.25867715e-02 -9.38039720e-01 8.86055529e-01 6.89955413e-01
9.75261211e-01 -7.68590033e-01 7.60119617e-01 6.11274898e-01
-1.38298094e+00 1.31550759e-01 -4.83683109e-01 -1.83468014e-01
4.70029384e-01 8.47943664e-01 -7.27227569e-01 6.91825509e-01
2.94615775e-01 4.27641302e-01 6.27999231e-02 9.05688882e-01
-4.01103348e-01 1.14810348e+00 -4.01074626e-02 -4.46090490e-01
7.90520199e-03 -3.28519315e-01 5.88473320e-01 1.18226779e+00
9.26577806e-01 2.76177704e-01 -7.76264910e-03 1.06346238e+00
-4.89506265e-03 5.06188452e-01 -4.00837779e-01 -3.25070232e-01
5.58571994e-01 8.55545104e-01 -3.15143019e-01 -3.79597813e-01
-1.64075032e-01 1.03517532e+00 -2.08830521e-01 5.34860790e-01
-1.02689207e+00 -8.68317306e-01 1.86413690e-01 -4.42662872e-02
3.66130650e-01 -1.44908890e-01 -8.11336637e-02 -9.32090223e-01
-1.45167693e-01 -6.93889916e-01 -7.22377524e-02 -9.92736101e-01
-1.04235423e+00 8.64644349e-01 -1.98835105e-01 -1.30925417e+00
-6.41032398e-01 -3.76463011e-02 -7.03542829e-01 1.15784860e+00
-1.34149265e+00 -6.30792558e-01 -1.24182291e-01 3.79381686e-01
9.24719214e-01 -1.68453246e-01 7.82285333e-01 3.09932590e-01
-6.39068902e-01 6.67977631e-01 1.15139261e-01 -4.96899225e-02
5.42489231e-01 -1.03378677e+00 3.14944714e-01 7.82080710e-01
3.25554073e-01 4.59892899e-01 7.37543941e-01 -6.84854269e-01
-1.05027962e+00 -1.04762530e+00 1.13580155e+00 -8.13603692e-04
4.76381361e-01 -3.09906244e-01 -1.00131524e+00 2.52128124e-01
3.58249724e-01 -4.36012089e-01 5.51850021e-01 -2.79301226e-01
2.64075935e-01 -1.43514186e-01 -6.78857684e-01 7.28417993e-01
6.52852356e-01 -5.45496345e-01 -7.65567064e-01 -2.14587920e-03
1.18344474e+00 -5.63056231e-01 -7.26174414e-01 6.99374229e-02
4.45430249e-01 -9.30612683e-01 4.72825140e-01 1.15191624e-01
7.54302323e-01 -4.63539273e-01 -1.59848675e-01 -1.65687144e+00
-4.76145595e-01 -8.20360839e-01 -1.59905165e-01 1.68682063e+00
5.26752770e-01 -5.88214993e-01 1.64591417e-01 6.84931427e-02
-2.87861198e-01 -5.43947041e-01 -6.49964452e-01 -6.53393507e-01
1.05293328e-02 -6.79790795e-01 6.78131461e-01 5.09986818e-01
-4.70659286e-02 5.37688017e-01 -6.81890488e-01 1.04175009e-01
2.20554322e-01 -7.88624808e-02 6.28239453e-01 -8.86415839e-01
-5.61039746e-01 -4.04735446e-01 2.09318683e-01 -1.56556284e+00
-1.71469629e-01 -6.38345242e-01 5.74025571e-01 -1.46546888e+00
-1.60870969e-01 -2.77431875e-01 -3.43551993e-01 1.37411892e-01
-4.64152575e-01 -1.50276065e-01 4.02703613e-01 8.18943977e-02
1.72388845e-03 9.84034657e-01 1.48735714e+00 7.66102076e-02
-6.49457276e-01 2.33389765e-01 -6.00669086e-01 5.62758446e-01
7.70592391e-01 -5.95432222e-01 -6.07550740e-01 -2.98383534e-01
-9.34581161e-02 5.78192770e-01 -7.18869418e-02 -1.09391248e+00
2.85264254e-01 -1.70605004e-01 1.35405362e-01 -7.09697008e-01
4.71130729e-01 -4.62096423e-01 2.24551141e-01 3.57193440e-01
-3.35926533e-01 -8.30683187e-02 -3.83733734e-02 4.73109752e-01
-4.11543459e-01 -5.23352265e-01 5.13021827e-01 1.09262250e-01
-2.03030676e-01 1.57402381e-01 -6.02572560e-01 3.92460357e-03
6.49700880e-01 -4.43754382e-02 -4.16999385e-02 -6.73863113e-01
-3.37377638e-01 -1.57413810e-01 -7.09280074e-02 4.62729812e-01
7.27650881e-01 -1.44729686e+00 -8.83048117e-01 3.63025635e-01
-3.62125039e-01 -9.04220566e-02 5.05539477e-01 5.90064049e-01
-3.03012401e-01 4.27242339e-01 2.77023226e-01 -3.11044216e-01
-9.98149931e-01 3.42617035e-01 1.32277593e-01 -1.28295302e-01
-5.46625435e-01 7.14048803e-01 3.36625129e-01 5.90663143e-02
1.59926698e-01 -4.52118784e-01 -5.31226575e-01 7.70036830e-03
5.60054302e-01 3.81080687e-01 -2.57886678e-01 -5.82314432e-01
8.16245824e-02 5.92062891e-01 2.46619970e-01 -6.41628861e-01
1.05249655e+00 -1.12268433e-01 1.18963525e-01 8.29291344e-01
1.07610559e+00 4.42700028e-01 -9.86472487e-01 1.27716184e-01
-5.71296155e-01 -2.53549874e-01 3.25091243e-01 -5.25275707e-01
-1.04304731e+00 1.02425754e+00 3.93308580e-01 5.60486972e-01
1.10965741e+00 -3.23759824e-01 1.29542267e+00 -1.71965897e-01
1.25634074e-01 -1.15102804e+00 1.94085896e-01 7.50472844e-01
9.22023952e-01 -6.72934413e-01 -4.93399650e-01 -1.97737753e-01
-8.80454481e-01 1.04377389e+00 4.55112010e-01 2.81070154e-02
6.37938976e-01 2.53922403e-01 1.38856366e-01 2.85157770e-01
-7.53805339e-01 -1.82700321e-01 2.89725125e-01 3.63821626e-01
6.55593693e-01 1.73951179e-01 -6.49532497e-01 1.11640465e+00
-7.91437328e-01 1.06155135e-01 5.00275850e-01 3.95454466e-01
-5.86161196e-01 -1.09080553e+00 -4.47359711e-01 3.40003788e-01
-5.20783424e-01 -3.91336828e-01 -9.64281429e-03 1.74958408e-01
2.56344169e-01 1.41079545e+00 5.20024002e-02 -8.67793143e-01
4.63236481e-01 1.42399535e-01 1.87373068e-02 -4.71071571e-01
-4.63336438e-01 6.01146877e-01 -1.53280228e-01 -1.08420132e-02
-5.76511547e-02 -3.58952641e-01 -1.45014095e+00 -1.51642308e-01
-7.27643073e-01 5.17428160e-01 6.37583077e-01 8.72583687e-01
1.26395836e-01 1.18555105e+00 1.00140405e+00 -5.86931884e-01
-5.06700218e-01 -1.26641762e+00 -7.91578293e-01 1.41290247e-01
2.99233347e-01 -3.25291842e-01 -4.98984575e-01 2.45097131e-01] | [15.275729179382324, 6.245105266571045] |
5460949b-78e5-4375-b017-cb1ec65f624b | effective-cloud-detection-and-segmentation | 1809.10801 | null | http://arxiv.org/abs/1809.10801v1 | http://arxiv.org/pdf/1809.10801v1.pdf | Effective Cloud Detection and Segmentation using a Gradient-Based Algorithm for Satellite Imagery; Application to improve PERSIANN-CCS | Being able to effectively identify clouds and monitor their evolution is one
important step toward more accurate quantitative precipitation estimation and
forecast. In this study, a new gradient-based cloud-image segmentation
technique is developed using tools from image processing techniques. This
method integrates morphological image gradient magnitudes to separable cloud
systems and patches boundaries. A varying scale-kernel is implemented to reduce
the sensitivity of image segmentation to noise and capture objects with various
finenesses of the edges in remote-sensing images. The proposed method is
flexible and extendable from single- to multi-spectral imagery. Case studies
were carried out to validate the algorithm by applying the proposed
segmentation algorithm to synthetic radiances for channels of the Geostationary
Operational Environmental Satellites (GOES-R) simulated by a high-resolution
weather prediction model. The proposed method compares favorably with the
existing cloud-patch-based segmentation technique implemented in the
PERSIANN-CCS (Precipitation Estimation from Remotely Sensed Information using
Artificial Neural Network - Cloud Classification System) rainfall retrieval
algorithm. Evaluation of event-based images indicates that the proposed
algorithm has potential to improve rain detection and estimation skills with an
average of more than 45% gain comparing to the segmentation technique used in
PERSIANN-CCS and identifying cloud regions as objects with accuracy rates up to
98%. | ['Kuo-lin Hsu', 'Soroosh Sorooshian', 'Negin Hayatbini', 'Yunji Zhang', 'Fuqing Zhang'] | 2018-09-27 | null | null | null | null | ['cloud-detection'] | ['computer-vision'] | [ 2.41924852e-01 -6.20752394e-01 2.33441040e-01 -4.11844999e-01
-3.90719414e-01 -6.37632489e-01 5.14143229e-01 1.76227018e-01
-5.59310794e-01 8.67517829e-01 -4.74386215e-01 -7.77680635e-01
-1.53535247e-01 -1.29552674e+00 1.25156343e-02 -9.24830437e-01
-3.56200457e-01 6.76210403e-01 3.25352214e-02 -5.38451672e-01
3.10395658e-01 1.08386207e+00 -1.69001484e+00 -1.26058728e-01
1.32178140e+00 5.99463224e-01 7.09512949e-01 8.57385635e-01
-3.97507906e-01 2.59291410e-01 -3.92897725e-01 6.86348677e-01
5.67225814e-01 -3.38998348e-01 -3.81116748e-01 3.55596781e-01
5.34333348e-01 -9.39823017e-02 6.23104334e-01 1.09856045e+00
2.10727483e-01 -3.55770513e-02 7.58297682e-01 -5.31515718e-01
-9.01524536e-03 -3.09210774e-02 -9.27269280e-01 6.21777594e-01
-3.36478472e-01 3.83106582e-02 4.89149630e-01 -1.04924679e+00
4.16984081e-01 9.41162467e-01 6.54964685e-01 -2.99057186e-01
-1.04536617e+00 -7.05362439e-01 -1.01787955e-01 -8.47709924e-02
-1.45403552e+00 8.57949480e-02 2.33668789e-01 -7.48535514e-01
8.55779529e-01 7.77517200e-01 1.11624646e+00 -6.31265581e-01
9.00631920e-02 2.13113502e-01 1.91297841e+00 -7.08174050e-01
3.28362316e-01 9.40730870e-02 4.18646574e-01 4.60911095e-01
8.23221684e-01 2.86889434e-01 1.93663076e-01 -2.38127485e-02
7.73087382e-01 2.29624242e-01 -2.50339508e-01 1.61550105e-01
-6.42213106e-01 1.20021129e+00 6.35743856e-01 5.76529622e-01
-7.93079019e-01 -4.09852177e-01 -1.72727495e-01 4.11118299e-01
1.01178551e+00 4.92842197e-01 -2.51104772e-01 3.22937846e-01
-1.99744093e+00 6.93909228e-01 3.89710873e-01 5.17470717e-01
8.68884563e-01 7.90162504e-01 1.11588679e-01 4.98919487e-01
4.94591415e-01 1.34540009e+00 1.23136654e-01 -6.71829224e-01
2.26587191e-01 4.76804912e-01 5.38756728e-01 -9.15755510e-01
-4.15658683e-01 -5.97487986e-01 -6.83000684e-01 8.87945831e-01
2.52764702e-01 -2.08702356e-01 -1.59871101e+00 5.88374019e-01
3.03594649e-01 -5.77771887e-02 3.23523805e-02 1.06461656e+00
5.43912888e-01 1.07036626e+00 3.25107902e-01 -3.49242598e-01
1.34283018e+00 -5.19351780e-01 -5.81264913e-01 -2.24874958e-01
2.72315979e-01 -8.51236641e-01 5.24795711e-01 1.74772829e-01
-5.78597903e-01 -5.80168307e-01 -9.17713165e-01 8.94152522e-01
-8.75298858e-01 3.34730685e-01 7.42137730e-01 9.20590997e-01
-1.04353404e+00 4.83625233e-01 -6.75873756e-01 -5.70702016e-01
1.20608568e-01 2.18519628e-01 8.83790106e-02 1.59318671e-01
-9.99476075e-01 9.72099543e-01 6.70080483e-01 5.98164916e-01
-3.54312062e-01 -4.97512758e-01 -6.75904274e-01 9.22057331e-02
-4.37734008e-01 -2.47308820e-01 6.25158787e-01 -1.48334312e+00
-1.03517699e+00 9.33259249e-01 -2.85326898e-01 -6.73863411e-01
1.88011795e-01 -5.55604324e-03 -4.74820137e-01 4.97515321e-01
2.87670463e-01 5.79413533e-01 9.04076755e-01 -1.52506828e+00
-1.10820997e+00 -6.01986587e-01 -4.14492965e-01 3.65503728e-01
4.12676960e-01 1.45418227e-01 2.37599224e-01 -7.38083124e-01
6.25746429e-01 -1.01523483e+00 -4.82362866e-01 -3.00358921e-01
3.22215974e-01 5.29379368e-01 1.14757872e+00 -9.52008367e-01
9.55232501e-01 -1.69890308e+00 -4.65783656e-01 8.70111883e-01
-8.38208050e-02 7.00572252e-01 9.30850804e-02 1.86007619e-01
-2.13340700e-01 6.79240599e-02 -9.04878378e-01 1.56035945e-01
-6.23884201e-01 5.20232737e-01 -3.66477489e-01 5.29581785e-01
2.68069386e-01 2.90982902e-01 -5.55112839e-01 -3.03228498e-01
6.16599441e-01 2.63927937e-01 2.31852785e-01 9.03297663e-02
-1.65424883e-01 4.96027261e-01 -3.16851854e-01 8.32827389e-01
1.70019209e+00 3.29410851e-01 -2.63946317e-02 6.02791905e-01
-8.64321768e-01 -4.44206148e-01 -1.33390880e+00 4.80586559e-01
-2.05248281e-01 6.84111357e-01 4.68279630e-01 -9.44086850e-01
1.35713530e+00 3.83345723e-01 1.36974037e-01 -6.39300227e-01
-2.88800031e-01 5.28012753e-01 -7.16573596e-02 -7.27643013e-01
9.55423355e-01 -4.70799237e-01 6.59642577e-01 -4.22243476e-02
-3.62948954e-01 -6.98853493e-01 2.58727729e-01 -3.22902054e-02
7.18215154e-03 1.47857890e-02 1.70303091e-01 -8.44435692e-01
5.01475930e-01 7.44728982e-01 2.66656756e-01 8.31883609e-01
-1.01109184e-01 5.81573844e-01 -2.52029181e-01 -5.65424919e-01
-1.03533840e+00 -8.27978373e-01 -5.55075943e-01 5.97127259e-01
-2.34868050e-01 5.17155707e-01 -5.80106020e-01 1.43910483e-01
1.03037447e-01 6.40217543e-01 -5.19472122e-01 9.56487536e-01
-1.76809520e-01 -1.42000949e+00 1.53340518e-01 1.46313785e-02
9.20476913e-01 -1.14200115e+00 -7.22705722e-01 4.31653827e-01
4.59120274e-02 -7.72286177e-01 6.46963596e-01 2.09805027e-01
-1.57605720e+00 -7.81763852e-01 -8.87665749e-01 -6.26013875e-01
6.87901020e-01 5.71023643e-01 1.27848661e+00 2.52622843e-01
-3.49109352e-01 1.67659789e-01 -5.36211848e-01 -8.67949486e-01
-1.44014746e-01 -1.71875089e-01 -7.09886551e-01 -1.81719780e-01
5.52713454e-01 -3.19346994e-01 -5.65159798e-01 -1.44122362e-01
-9.84342098e-01 -2.45765284e-01 4.27354991e-01 4.93698627e-01
5.84896564e-01 3.04128587e-01 2.14584768e-01 -9.74577129e-01
3.53467226e-01 -5.24611473e-01 -1.38637710e+00 5.15983291e-02
-9.19455409e-01 -4.90188241e-01 5.93568273e-02 4.29515541e-01
-1.23990536e+00 2.62629777e-01 2.69160748e-01 6.31844997e-03
-6.16987705e-01 8.90439391e-01 4.91275609e-01 -5.38899779e-01
5.96854031e-01 4.46283400e-01 1.94305144e-02 -2.53159404e-01
8.88611376e-02 7.28197873e-01 3.97134006e-01 -3.78186762e-01
9.87141073e-01 8.00907791e-01 1.60007805e-01 -1.44056845e+00
-2.20127299e-01 -1.18476701e+00 -7.07054138e-01 -5.34775615e-01
9.52693462e-01 -1.23763156e+00 2.78143704e-01 7.85575986e-01
-7.48044252e-01 -1.55155927e-01 1.07587703e-01 6.80827856e-01
1.67439878e-01 1.50198758e-01 -2.82256782e-01 -1.45098674e+00
-7.73224950e-01 -8.24260533e-01 7.60104477e-01 4.74527001e-01
3.39089096e-01 -9.98077869e-01 2.29362413e-01 1.60001338e-01
9.75146770e-01 5.45056581e-01 4.69725728e-01 8.66613388e-02
-5.58570981e-01 -1.55491650e-01 -3.71272385e-01 3.24045986e-01
-2.35549174e-03 5.53990126e-01 -9.83378410e-01 -1.69499502e-01
1.16915971e-01 2.02611059e-01 1.28768063e+00 8.17878067e-01
4.42556858e-01 8.47637728e-02 -6.10158294e-02 4.82674897e-01
2.31589222e+00 3.38346958e-01 9.12632883e-01 9.02047157e-01
2.67195106e-01 5.55698991e-01 8.68411481e-01 5.13090670e-01
-1.38989270e-01 1.93443596e-02 5.36843240e-01 -6.69688046e-01
3.78337234e-01 7.81774640e-01 -8.77563059e-02 3.74725223e-01
-7.87015796e-01 1.85230374e-01 -1.15568495e+00 7.97817588e-01
-1.45395875e+00 -1.24247622e+00 -9.08657134e-01 2.12129045e+00
4.08914387e-01 2.49140821e-02 -2.02878952e-01 1.90779179e-01
6.83149576e-01 3.07757258e-01 2.55220681e-01 -8.92644405e-01
-2.68128008e-01 1.12411165e+00 1.10820329e+00 9.25994217e-01
-1.41371083e+00 1.02284503e+00 5.59214497e+00 4.59276557e-01
-1.58310115e+00 7.09103718e-02 3.98879051e-01 4.26983058e-01
-1.30808756e-01 2.09919989e-01 -6.47468030e-01 2.93736041e-01
8.58117938e-01 4.05039549e-01 1.99770138e-01 4.06385094e-01
1.04083550e+00 -8.27938795e-01 4.34962869e-01 3.83864701e-01
-4.53811318e-01 -1.25613892e+00 1.08267561e-01 -1.46576121e-01
9.28096235e-01 3.62613320e-01 -1.85384601e-01 -1.16894292e-02
2.43377894e-01 -1.07552481e+00 5.57830632e-01 8.14480662e-01
6.15777850e-01 -6.97814047e-01 9.06072974e-01 6.48014396e-02
-1.48915577e+00 1.66695386e-01 -5.23298144e-01 -6.24576092e-01
-3.36594462e-01 1.04025459e+00 -8.71319175e-01 7.19488502e-01
7.83201456e-01 3.22733879e-01 -5.35582364e-01 1.33939672e+00
-3.18205059e-02 1.13220596e+00 -6.56127989e-01 3.04241896e-01
8.08662891e-01 -1.01013136e+00 4.96427536e-01 1.63132465e+00
5.75595021e-01 4.73677993e-01 -4.24719453e-02 6.62975550e-01
6.26762748e-01 3.49292696e-01 -5.00146747e-01 2.62632459e-01
4.55162913e-01 1.30412483e+00 -1.14397323e+00 -6.53507113e-01
-3.90936852e-01 4.94344175e-01 -5.29362917e-01 5.89073420e-01
-4.11528289e-01 -3.11790079e-01 4.31847215e-01 3.55543554e-01
5.47537923e-01 -4.46326792e-01 -6.48457706e-01 -6.10756636e-01
-3.12857062e-01 -6.37779474e-01 8.60708654e-02 -9.87640619e-01
-6.13428891e-01 4.62085664e-01 1.38745993e-01 -1.35371840e+00
3.90594713e-02 -5.21728158e-01 -1.03133714e+00 1.52952862e+00
-1.99520469e+00 -1.14988101e+00 -6.38395309e-01 4.03435588e-01
3.50415587e-01 -7.54494667e-02 1.08927894e+00 -6.33235574e-02
-4.56109121e-02 -4.96158630e-01 9.29610014e-01 -1.41464742e-02
1.38751477e-01 -1.54048586e+00 3.63967493e-02 1.26045084e+00
-3.46158147e-01 2.08768547e-01 8.93623769e-01 -7.35082328e-01
-6.16433859e-01 -1.44702423e+00 8.53285909e-01 5.41486502e-01
2.98404336e-01 2.26449460e-01 -9.64850426e-01 4.88981396e-01
3.57161105e-01 -5.22530749e-02 2.45417997e-01 -2.43254513e-01
1.24998733e-01 -1.93913221e-01 -1.35917866e+00 1.13620713e-01
-2.55492985e-01 -1.61015213e-01 -5.91160178e-01 4.08937603e-01
-2.20652327e-01 -2.78076351e-01 -7.03583121e-01 5.47825396e-01
4.14350718e-01 -1.17624998e+00 5.73014975e-01 -3.16123962e-02
1.91956371e-01 -6.18882954e-01 5.29494323e-02 -1.18425214e+00
-3.98931980e-01 5.25606051e-03 8.00022423e-01 8.60179901e-01
5.94154894e-01 -6.89179003e-01 6.97875500e-01 1.98270291e-01
1.55717805e-01 -2.17801463e-02 -7.82289624e-01 -4.90859240e-01
2.03126103e-01 -8.32477510e-02 2.35902280e-01 1.17286873e+00
-7.86642790e-01 -3.11568499e-01 1.79474980e-01 8.34537625e-01
6.74193501e-01 7.15540528e-01 5.30140281e-01 -1.70555997e+00
1.96322456e-01 -2.96533585e-01 -3.51563454e-01 -1.70609549e-01
-2.76251018e-01 -4.24203932e-01 -8.49793255e-02 -1.57553196e+00
-1.30373845e-02 -8.01168084e-01 -2.04005279e-02 3.06371510e-01
-3.51266980e-01 2.84143209e-01 3.08332890e-01 5.58555305e-01
3.30258369e-01 -2.81695127e-02 1.07899094e+00 -1.46413565e-01
-3.14652681e-01 1.68667287e-01 1.61459316e-02 6.45671844e-01
1.16232347e+00 -6.91666603e-01 1.09868888e-02 -1.94671646e-01
-2.47258246e-02 1.58887595e-01 3.01537991e-01 -1.34392667e+00
-3.51732373e-01 -3.89758497e-01 5.55742621e-01 -7.71987915e-01
-1.46820378e-02 -1.02906251e+00 4.76200789e-01 6.88018322e-01
3.59485388e-01 8.72816890e-04 5.29666007e-01 1.75554007e-01
-6.39398336e-01 -5.44718206e-01 1.07483256e+00 -5.94117999e-01
-1.20705092e+00 6.82745948e-02 -9.77297843e-01 -6.32809103e-01
9.95571017e-01 -5.09900630e-01 -1.27076566e-01 -3.76292884e-01
-8.96292686e-01 1.21000543e-01 4.64645624e-01 -2.55478680e-01
5.38479328e-01 -5.49918771e-01 -1.10262680e+00 3.35528910e-01
1.82695203e-02 -6.70516789e-02 2.71993261e-02 6.36663079e-01
-1.59230983e+00 4.07156825e-01 -4.60472137e-01 -7.31256962e-01
-1.59081352e+00 -5.32010719e-02 7.52257109e-01 -4.05321687e-01
-3.53411436e-01 5.92606664e-01 -4.37561691e-01 -3.77397746e-01
-4.31355894e-01 -5.24582446e-01 -7.37905383e-01 1.95245013e-01
2.89120793e-01 1.20773390e-01 3.64876688e-01 -7.74450421e-01
-1.10991165e-01 6.99300170e-01 3.41273338e-01 -2.08320498e-01
1.43991971e+00 -5.26847318e-02 -5.19168735e-01 3.86650085e-01
4.53479409e-01 1.03907384e-01 -9.09679055e-01 2.66248472e-02
2.09034793e-02 -5.79456568e-01 5.76822877e-01 -7.19908834e-01
-1.34661174e+00 7.99026132e-01 1.36838937e+00 2.24550143e-01
1.30061126e+00 -8.95829439e-01 1.87714517e-01 2.28417635e-01
-1.12761296e-02 -9.73441839e-01 -1.02433717e+00 5.80239594e-01
6.59832299e-01 -1.43101025e+00 5.67671001e-01 -3.81667256e-01
-4.69897866e-01 1.57971084e+00 2.19388828e-02 -3.63147110e-01
1.04208565e+00 1.99539453e-01 5.22684336e-01 -5.22175431e-01
1.52947560e-01 -8.08803022e-01 -2.07416147e-01 5.18535614e-01
4.69856501e-01 8.21690679e-01 -7.33098269e-01 -4.15893197e-01
8.90966207e-02 2.64161646e-01 6.51659548e-01 1.23974907e+00
-1.26307023e+00 -6.96965337e-01 -1.42341018e+00 8.67424190e-01
-4.33293104e-01 -6.69488668e-01 1.05916902e-01 9.40456390e-01
3.31233591e-01 9.55442786e-01 3.84516954e-01 3.55445623e-01
-2.27756023e-01 7.75399208e-02 1.27473146e-01 -4.59793836e-01
-6.89125657e-01 2.73608208e-01 -2.34414451e-03 1.32798254e-01
-1.26484966e+00 -8.15975726e-01 -1.20438457e+00 -3.31999689e-01
-4.14271474e-01 5.60507715e-01 1.00054121e+00 9.44043577e-01
-8.33188146e-02 9.88737121e-02 6.81326449e-01 -1.20165539e+00
1.26131624e-03 -1.22644448e+00 -1.49434614e+00 -1.75573528e-01
2.94968843e-01 -4.83165264e-01 -3.27296138e-01 2.05019221e-01] | [9.70119571685791, -1.7621225118637085] |
12ff4e87-3ef1-40df-a96f-c901129c0253 | daml-chinese-named-entity-recognition-with-a | null | null | https://openreview.net/forum?id=N-DZvl4bQsO | https://openreview.net/pdf?id=N-DZvl4bQsO | DAML: Chinese Named Entity Recognition with a fusion method of data-augmentation and meta-learning | Overfitting is still a common problem in NER with insufficient data. Latest methods such as Transfer Learning, which focuses on storing knowledge gained while solving one task and applying it to a different but related task, or Model-Agnostic Meta-Learning (MAML), which learns a model parameter initialization that generalizes better to similar tasks. However, these methods still need rich resources to pre-train. In this work, we present new perspectives on how to make the most of in-domain and out-domain information. By introducing a fusion method of data augmentation and MAML, we first use data augmentation to mine more information. With the augmented resources, we directly utilize out-domain and in-domain data with MAML, while avoiding performance degradation after domain transfer. To further improve the model’s generalization ability, we proposed a new data augmentation method based on a generative approach. We conduct experiments on six open Chinese NER datasets (MSRANER, PeopleDairyNER, CLUENER, WeiboNER, Resume NER, and BOSONNER). The results show that our method significantly reduces the impact of insufficient data and outperforms the state-of-the-art. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['chinese-named-entity-recognition'] | ['natural-language-processing'] | [-9.25458968e-02 3.29778567e-02 9.65342820e-02 -4.62338656e-01
-6.49388850e-01 -4.50269222e-01 6.45119786e-01 -3.05465423e-02
-9.39140916e-01 1.06480277e+00 3.38520080e-01 -1.06335677e-01
8.81064776e-03 -8.90056133e-01 -5.37536383e-01 -4.76495266e-01
5.94253719e-01 5.55216849e-01 7.50282481e-02 -5.89725256e-01
6.91455379e-02 1.69278905e-01 -1.14544535e+00 1.96969613e-01
1.47918057e+00 4.83658642e-01 6.46827340e-01 1.49836078e-01
-5.84452331e-01 6.56468093e-01 -7.88169086e-01 -4.81212676e-01
-1.03892889e-02 -4.66366053e-01 -1.16332173e+00 -2.79023468e-01
-1.61570981e-01 1.68093070e-02 -2.80539036e-01 1.01529956e+00
6.77427053e-01 4.95731413e-01 7.36310184e-01 -8.70843589e-01
-1.01471758e+00 8.61841381e-01 -3.01734716e-01 1.41530126e-01
-1.44381315e-01 -1.81756318e-01 3.02310407e-01 -8.17991495e-01
5.10279417e-01 1.13149774e+00 7.92465389e-01 1.09361827e+00
-7.91294992e-01 -8.24653983e-01 2.28390202e-01 1.12657599e-01
-1.04557562e+00 -2.62259692e-01 7.52590418e-01 -1.15912735e-01
6.89715028e-01 -4.49786410e-02 9.36252847e-02 1.26474321e+00
-2.93647230e-01 6.82495534e-01 1.38853574e+00 -6.06531203e-01
3.26872803e-02 2.61612415e-01 3.25484276e-01 3.43910247e-01
2.47620538e-01 -4.87556048e-02 -2.46883854e-01 -4.47623804e-02
6.33730948e-01 2.07075298e-01 -2.30420262e-01 3.32595706e-01
-1.17632914e+00 6.69335723e-01 5.40244997e-01 7.39757657e-01
-3.55628222e-01 -4.85768288e-01 2.59805411e-01 3.34058434e-01
7.36503541e-01 5.61334312e-01 -9.59384024e-01 -3.64005603e-02
-6.77588820e-01 -5.39362989e-02 7.71077275e-01 1.08302879e+00
9.98385370e-01 1.03816338e-01 -2.26590753e-01 1.14075243e+00
-6.32100552e-02 3.77473712e-01 1.19088173e+00 -3.69824082e-01
8.50563705e-01 8.57737660e-01 1.08062029e-01 -4.64010805e-01
-6.35179460e-01 -3.86979640e-01 -1.31854892e+00 -3.50885481e-01
6.71318695e-02 -7.26061583e-01 -1.25252402e+00 1.91170764e+00
8.16792101e-02 2.00214937e-01 5.99011242e-01 6.06487513e-01
1.13717353e+00 8.12407851e-01 4.58028764e-01 -9.96585488e-02
1.33124161e+00 -1.20068443e+00 -9.19969559e-01 -5.30643880e-01
1.00624132e+00 -5.73745728e-01 1.23085713e+00 2.26868331e-01
-7.08273649e-01 -8.86176586e-01 -8.24206293e-01 -2.35745654e-01
-8.01775277e-01 4.38043207e-01 5.17807245e-01 5.18863678e-01
-7.55735219e-01 7.33605802e-01 -5.46183109e-01 -3.62947762e-01
1.36330694e-01 2.78356194e-01 -5.91550589e-01 -2.31216371e-01
-1.55299842e+00 1.17758870e+00 1.25721598e+00 -1.66233182e-02
-5.99655330e-01 -7.75487900e-01 -9.04100180e-01 1.08738907e-01
3.20281416e-01 -5.58393300e-01 1.20047367e+00 -6.37509584e-01
-1.48729551e+00 6.11229062e-01 -5.02294861e-03 -2.83529758e-01
1.40175000e-01 -5.75855970e-01 -8.92665327e-01 -4.87696946e-01
1.07786907e-02 6.86651945e-01 2.47648746e-01 -1.29986441e+00
-5.72227538e-01 -3.74585718e-01 -2.97745429e-02 2.42480502e-01
-9.15482402e-01 -1.61734998e-01 -2.70080388e-01 -7.91999102e-01
-1.23001404e-01 -6.91511750e-01 -3.55650574e-01 -9.68194306e-01
-2.55152285e-01 -4.90764439e-01 7.79697061e-01 -9.37555432e-01
1.41188169e+00 -2.00430274e+00 9.37559605e-02 -1.71445012e-01
-1.95397679e-02 8.74880135e-01 -4.22191262e-01 5.88759184e-01
-3.80427763e-02 3.57210249e-01 -4.39911008e-01 -4.27110106e-01
-2.45675281e-01 4.73775178e-01 -1.30595729e-01 -2.68145889e-01
2.44926512e-01 1.00577819e+00 -9.67097223e-01 -4.49561924e-01
-4.47193608e-02 4.07912493e-01 -3.00528675e-01 5.32366037e-01
-1.09891377e-01 8.23041439e-01 -4.67515290e-01 3.24568838e-01
7.67227590e-01 -1.76403672e-01 3.21901105e-02 -3.47881794e-01
-1.37563437e-01 3.00619692e-01 -1.04279971e+00 1.95674288e+00
-8.03738475e-01 1.52475968e-01 -2.31325120e-01 -1.11664402e+00
1.47634029e+00 3.89416754e-01 4.98698233e-03 -8.28643560e-01
1.99492872e-01 2.32429415e-01 -5.49119189e-02 -5.22049904e-01
5.85484326e-01 -1.50456771e-01 -1.39561281e-01 3.14849526e-01
3.72817487e-01 2.93583304e-01 2.51360357e-01 -1.03073388e-01
8.02215517e-01 2.05857158e-01 3.67419332e-01 -2.97130913e-01
6.19388402e-01 8.66202936e-02 8.83404732e-01 4.68824685e-01
2.17270762e-01 2.53224164e-01 6.75126985e-02 -2.79641449e-01
-1.00320995e+00 -6.17595673e-01 4.87659452e-03 1.30720425e+00
3.76769304e-02 -3.35096568e-01 -1.03000009e+00 -1.00156212e+00
-4.51638371e-01 9.87084448e-01 -6.23465359e-01 -4.10789847e-01
-7.18746305e-01 -1.08906209e+00 6.86423957e-01 8.06898892e-01
1.17056215e+00 -1.25195694e+00 -2.14099772e-02 2.44640559e-01
-5.09661555e-01 -1.03637540e+00 -4.60034102e-01 2.66293734e-01
-1.24103785e+00 -6.91682577e-01 -1.00424457e+00 -1.03462553e+00
8.26257765e-01 5.35343289e-02 1.15897202e+00 -1.51546644e-02
8.69776532e-02 5.37842549e-02 -7.42407918e-01 -5.09973109e-01
-5.26687562e-01 6.16620183e-01 6.36851266e-02 -2.81863034e-01
4.76647019e-01 -4.82304454e-01 -2.44640648e-01 3.99830401e-01
-9.82126534e-01 1.05940744e-01 9.52262759e-01 1.15588272e+00
4.23490375e-01 2.89239492e-02 9.84330416e-01 -1.25606036e+00
8.43916178e-01 -6.39463425e-01 -2.19265580e-01 4.87626761e-01
-6.12653136e-01 4.28026527e-01 7.99752831e-01 -7.26480067e-01
-1.72196007e+00 -7.46862590e-02 -2.44884863e-01 -3.00600439e-01
-3.28737915e-01 7.26092458e-01 -7.02598512e-01 2.55379379e-01
8.05702984e-01 2.94894934e-01 -3.01313579e-01 -1.23918939e+00
3.94219667e-01 9.63919163e-01 6.78143799e-01 -8.14665496e-01
6.69085026e-01 6.63530454e-02 -4.25781250e-01 -6.60077572e-01
-1.12224519e+00 -3.03451091e-01 -8.17232251e-01 3.41181487e-01
8.30296934e-01 -9.02971089e-01 -2.21330360e-01 5.06966829e-01
-1.38439870e+00 -3.29591572e-01 -1.23420522e-01 5.73325694e-01
-8.27063434e-03 2.30206743e-01 -5.33110380e-01 -5.95637679e-01
-5.70416689e-01 -7.35826313e-01 5.49594402e-01 6.23490393e-01
2.75281250e-01 -1.22165108e+00 3.35064739e-01 1.98682070e-01
4.98973161e-01 -6.71084523e-02 8.86242747e-01 -1.37871945e+00
1.15805574e-01 -4.17953171e-03 -1.60910308e-01 7.55453408e-01
3.29095483e-01 -6.59828007e-01 -9.78340328e-01 -2.08050087e-01
7.85026774e-02 -3.57493848e-01 9.37227905e-01 -1.37592211e-01
1.23753679e+00 -3.36544454e-01 -4.03369546e-01 5.65582812e-01
1.25319505e+00 3.07203680e-01 7.78081059e-01 5.60397208e-01
7.43163049e-01 5.62916696e-01 7.44910717e-01 8.56201127e-02
4.87462521e-01 3.92944247e-01 2.23394614e-02 -4.08641875e-01
-1.22441344e-01 -4.32627559e-01 1.78747222e-01 1.20099652e+00
-5.15943944e-01 -3.44207287e-01 -8.91812265e-01 5.26043475e-01
-1.80173409e+00 -6.02120757e-01 1.47289738e-01 2.10291147e+00
1.11154664e+00 -9.23406705e-02 -1.34258077e-01 -1.40114263e-01
9.20473695e-01 -2.62392789e-01 -5.09159744e-01 -1.53055727e-01
-2.37926573e-01 3.10798287e-01 4.27309304e-01 2.38873273e-01
-1.22872710e+00 1.28198528e+00 5.44630527e+00 9.67400253e-01
-7.58313894e-01 4.71628070e-01 4.94586647e-01 6.88773811e-01
-2.15285867e-01 -4.53920700e-02 -1.22596788e+00 4.79693413e-01
1.09814012e+00 -1.31763265e-01 2.43063286e-01 8.48515809e-01
-2.16869548e-01 3.78259689e-01 -9.01182950e-01 7.82916069e-01
1.04037911e-01 -1.11239946e+00 1.92547947e-01 -6.28616661e-02
8.49974692e-01 4.34031747e-02 -4.17963684e-01 9.87364054e-01
4.40286994e-01 -9.67890918e-01 -1.70740136e-03 7.81532586e-01
5.52610815e-01 -8.58941853e-01 1.11232436e+00 8.40999186e-01
-1.00639880e+00 6.29464537e-02 -7.24793553e-01 2.08472982e-02
-3.59311253e-02 5.14644325e-01 -5.78621566e-01 1.12628329e+00
6.04263246e-01 5.46630442e-01 -6.16885424e-01 8.37551594e-01
-4.72487062e-01 6.17032528e-01 -6.91337734e-02 -3.58211920e-02
3.04860100e-02 -1.95911646e-01 6.31592199e-02 1.41001630e+00
4.49166089e-01 3.42276663e-01 2.33311743e-01 7.97878861e-01
-4.14026052e-01 1.94393933e-01 -5.58490813e-01 -6.03347458e-02
8.07632089e-01 1.30108595e+00 -2.70219803e-01 -4.28571105e-01
-3.93540710e-01 1.14107680e+00 5.29464841e-01 3.47302854e-01
-5.93176723e-01 -7.04464078e-01 2.14435533e-01 -2.16553822e-01
-4.68735397e-03 -1.72216445e-01 -9.42257121e-02 -1.37390745e+00
-1.17132708e-01 -7.47885227e-01 6.15432501e-01 -6.35993540e-01
-1.63673723e+00 8.94731343e-01 8.40913132e-02 -1.24726188e+00
-2.99173802e-01 -6.09105051e-01 -7.42891908e-01 1.19159472e+00
-1.67695630e+00 -1.31672466e+00 -4.91198182e-01 7.05827236e-01
6.41206086e-01 -4.66795474e-01 1.07958567e+00 4.13025975e-01
-5.99577963e-01 6.93055928e-01 2.10933030e-01 5.54542601e-01
9.19205546e-01 -1.13952971e+00 4.69746739e-01 6.66658282e-01
-1.02319382e-03 6.91952109e-01 1.51447669e-01 -7.70815611e-01
-8.96833122e-01 -1.25383413e+00 1.04067373e+00 -3.36883277e-01
2.29061902e-01 -1.68101549e-01 -1.31236231e+00 8.23633552e-01
2.93883711e-01 -1.82317883e-01 7.76455522e-01 3.34282517e-01
-1.70843050e-01 -1.27467036e-01 -1.17298567e+00 4.63346988e-01
9.75679815e-01 -1.22175641e-01 -1.27368712e+00 1.88387468e-01
1.01379061e+00 -3.82896006e-01 -1.12311125e+00 6.13212168e-01
-1.23491392e-01 -5.60625911e-01 8.32543433e-01 -1.06873381e+00
2.76243031e-01 -1.58298999e-01 2.08457652e-02 -1.90174758e+00
-3.78203571e-01 -3.24469715e-01 -3.12184840e-02 1.85380769e+00
4.71822053e-01 -6.26343489e-01 4.06450957e-01 4.83073145e-01
-4.71921653e-01 -3.15387994e-01 -6.62393212e-01 -1.08027494e+00
5.18729568e-01 -2.16696989e-02 8.51204991e-01 1.45819712e+00
-4.66344133e-02 6.33955717e-01 -5.29279113e-01 4.50135209e-02
2.72111416e-01 -3.07577044e-01 7.18518436e-01 -1.58830571e+00
3.14167477e-02 -1.71619970e-02 9.58484411e-02 -9.37922537e-01
2.84477025e-01 -1.06472528e+00 -1.95236113e-02 -1.64984870e+00
1.92614809e-01 -4.66264933e-01 -5.83296120e-01 9.48532403e-01
-4.88093823e-01 -1.56582981e-01 1.76041365e-01 2.05311254e-01
-4.55096632e-01 7.50027955e-01 1.37189543e+00 1.44448817e-01
-3.83760154e-01 -8.21318552e-02 -8.76902461e-01 8.46250892e-01
9.81723487e-01 -5.55673540e-01 -3.46660018e-01 -7.47538090e-01
-5.08252792e-02 -2.54640013e-01 1.79165870e-01 -1.09096849e+00
3.16666037e-01 -9.73232149e-04 6.19224012e-01 -5.76111436e-01
1.66758448e-01 -6.35423243e-01 -1.19063362e-01 2.90647626e-01
-2.48035952e-01 2.17194110e-01 5.19787788e-01 3.85660410e-01
-3.22972506e-01 -6.24137223e-01 7.36440361e-01 -3.47840756e-01
-1.04868412e+00 2.75094271e-01 9.29954275e-02 4.73983586e-01
6.93577826e-01 2.07101867e-01 -5.18410861e-01 2.42167469e-02
-7.95979083e-01 3.81915629e-01 3.01340930e-02 5.78938365e-01
4.47167486e-01 -1.40467560e+00 -9.81475592e-01 1.89842224e-01
1.00523435e-01 1.50921002e-01 5.58742523e-01 4.43752199e-01
-1.50041450e-02 2.72318721e-01 -2.77880222e-01 -1.96611509e-01
-7.73173511e-01 6.26298785e-01 2.17095062e-01 -6.90275371e-01
-4.86460179e-01 5.39844990e-01 9.78455320e-02 -1.21303952e+00
-2.28621941e-02 -7.25418478e-02 -6.55932188e-01 2.09338330e-02
5.13589501e-01 5.18975019e-01 2.39796326e-01 -3.10110509e-01
-8.15491229e-02 3.79416019e-01 -2.93014526e-01 -6.55083507e-02
1.33912516e+00 -2.19035689e-02 -6.62463307e-02 2.99944967e-01
9.14905012e-01 -1.21609420e-01 -9.77070153e-01 -5.84971905e-01
1.94536358e-01 -2.82873772e-02 3.34495306e-02 -1.16911054e+00
-9.19542432e-01 9.81929123e-01 5.49891472e-01 -7.78864324e-02
1.24462020e+00 -3.98258166e-03 8.18668067e-01 8.08468878e-01
2.99305946e-01 -1.34874082e+00 -1.34456277e-01 9.38522756e-01
7.57376075e-01 -1.31504369e+00 -3.63697350e-01 -1.88102782e-01
-8.50293159e-01 1.05601060e+00 9.97609258e-01 1.35735661e-01
6.82676077e-01 1.42928824e-01 5.36364689e-02 1.98098257e-01
-4.58657086e-01 -3.53473097e-01 4.13692296e-01 8.29389572e-01
6.10173583e-01 -8.34302679e-02 -5.31724513e-01 1.43185556e+00
-1.47603467e-01 2.49518007e-01 2.06281483e-01 8.36430311e-01
-5.64225316e-01 -1.54796076e+00 -2.39202872e-01 2.64002293e-01
-2.34304160e-01 -4.05296862e-01 -2.84852684e-01 1.06272876e+00
3.19484860e-01 9.36076522e-01 -6.61635995e-02 -5.30268431e-01
7.02990174e-01 4.09582406e-01 2.11819425e-01 -7.15814888e-01
-6.10330760e-01 -1.41292870e-01 1.30846471e-01 1.74875613e-02
-5.62059462e-01 -1.41940176e-01 -1.37350094e+00 -4.72014211e-02
-4.08847630e-01 6.10283911e-01 6.94013417e-01 1.26849055e+00
4.71555561e-01 7.20865250e-01 4.33606625e-01 -5.95740199e-01
-6.03512704e-01 -1.55882716e+00 -5.02693117e-01 4.51780736e-01
-2.15882793e-01 -6.27174497e-01 -2.02094197e-01 1.40777081e-01] | [9.840950965881348, 9.53619384765625] |
187c9e45-f26c-4da8-ac7f-cd15dba855e2 | liquid-structural-state-space-models | 2209.12951 | null | https://arxiv.org/abs/2209.12951v1 | https://arxiv.org/pdf/2209.12951v1.pdf | Liquid Structural State-Space Models | A proper parametrization of state transition matrices of linear state-space models (SSMs) followed by standard nonlinearities enables them to efficiently learn representations from sequential data, establishing the state-of-the-art on a large series of long-range sequence modeling benchmarks. In this paper, we show that we can improve further when the structural SSM such as S4 is given by a linear liquid time-constant (LTC) state-space model. LTC neural networks are causal continuous-time neural networks with an input-dependent state transition module, which makes them learn to adapt to incoming inputs at inference. We show that by using a diagonal plus low-rank decomposition of the state transition matrix introduced in S4, and a few simplifications, the LTC-based structural state-space model, dubbed Liquid-S4, achieves the new state-of-the-art generalization across sequence modeling tasks with long-term dependencies such as image, text, audio, and medical time-series, with an average performance of 87.32% on the Long-Range Arena benchmark. On the full raw Speech Command recognition, dataset Liquid-S4 achieves 96.78% accuracy with a 30% reduction in parameter counts compared to S4. The additional gain in performance is the direct result of the Liquid-S4's kernel structure that takes into account the similarities of the input sequence samples during training and inference. | ['Daniela Rus', 'Alexander Amini', 'Makram Chahine', 'Tsun-Hsuan Wang', 'Mathias Lechner', 'Ramin Hasani'] | 2022-09-26 | null | null | null | null | ['spo2-estimation', 'heart-rate-estimation', 'long-range-modeling'] | ['medical', 'medical', 'natural-language-processing'] | [ 4.71408814e-01 -1.76526770e-01 -4.72325921e-01 -2.82445759e-01
-3.98336351e-01 -4.38787431e-01 8.36924136e-01 -1.05273277e-01
-4.17737544e-01 3.62597078e-01 2.52509236e-01 -7.40893781e-01
-2.97073513e-01 -2.06202880e-01 -1.16459978e+00 -6.96458459e-01
-5.07491052e-01 4.28754449e-01 2.86905289e-01 -1.13406949e-01
-8.56314600e-02 5.54377973e-01 -1.04856527e+00 3.45083475e-01
3.89119089e-01 9.73884523e-01 5.33655174e-02 9.88870144e-01
-4.23987024e-02 1.14247620e+00 -3.36751044e-01 1.83323771e-01
2.08201781e-01 -3.10991049e-01 -8.34775984e-01 -2.54899263e-01
1.68673068e-01 -1.97834492e-01 -9.56432879e-01 5.04327595e-01
1.48122624e-01 2.93485135e-01 7.94104934e-01 -1.03877580e+00
-4.02792811e-01 7.60443628e-01 -1.89898476e-01 3.67592424e-01
-5.25293164e-02 3.17891300e-01 9.29242134e-01 -5.78177631e-01
3.64926547e-01 1.34719801e+00 7.84959316e-01 8.44245672e-01
-1.63397300e+00 -6.14679158e-01 6.49814665e-01 5.39278686e-01
-1.21035266e+00 -2.15780631e-01 3.58060896e-01 -4.29524630e-01
1.65906882e+00 2.99925745e-01 4.63418096e-01 1.51780570e+00
6.08727932e-01 1.02729690e+00 9.38052654e-01 -2.52638578e-01
3.06463897e-01 -4.04396951e-01 6.59261346e-01 5.33782065e-01
-8.35170895e-02 3.78402442e-01 -6.78281724e-01 -2.95052141e-01
7.89139271e-01 1.31740123e-01 -2.36612231e-01 -3.24219584e-01
-1.31951904e+00 4.59635764e-01 2.91562051e-01 3.22546333e-01
-3.88471603e-01 6.98732674e-01 6.56385541e-01 4.75366801e-01
2.80880809e-01 1.73574552e-01 -7.39437461e-01 -3.51575851e-01
-8.21709871e-01 -8.70269984e-02 8.84837389e-01 8.65928054e-01
2.73213029e-01 5.06237924e-01 -1.79708451e-01 5.96271098e-01
2.40766957e-01 7.07071424e-01 6.42796457e-01 -7.50616431e-01
3.98684561e-01 2.83413023e-01 -2.53958553e-01 -2.88754016e-01
-6.67671442e-01 -9.33502316e-01 -1.25449967e+00 -2.93561310e-01
1.80550456e-01 5.74049391e-02 -1.36504614e+00 2.05602264e+00
-2.04820469e-01 7.83305228e-01 1.66005522e-01 2.22701713e-01
1.45330191e-01 1.20369577e+00 -7.47766048e-02 -4.52561259e-01
1.01372659e+00 -7.30051160e-01 -7.03734577e-01 -3.99692446e-01
7.20815659e-01 -2.39454463e-01 7.35013545e-01 4.33174074e-01
-8.64252388e-01 -4.50053155e-01 -1.07115662e+00 2.38014489e-01
-9.13892388e-02 -9.24523324e-02 4.24705535e-01 2.67728716e-01
-1.09318304e+00 8.76292288e-01 -1.70281541e+00 -3.02537918e-01
-7.58816972e-02 4.45535809e-01 -2.87644088e-01 7.98876137e-02
-1.34566379e+00 9.47546721e-01 3.31856847e-01 1.90465838e-01
-1.23416853e+00 -9.92353559e-01 -5.76349318e-01 9.09400433e-02
3.67592424e-01 -7.64559746e-01 1.35231566e+00 -4.53590393e-01
-1.71556985e+00 1.57810792e-01 -2.78494358e-01 -1.06526589e+00
3.14167947e-01 -3.01054239e-01 -5.05901277e-01 -2.46818606e-02
-3.46193880e-01 3.54731917e-01 9.28930223e-01 -9.27093208e-01
-3.33220780e-01 -1.23031907e-01 -5.00989079e-01 -1.82748154e-01
-1.50364161e-01 -1.41953439e-01 -3.21804762e-01 -5.48582315e-01
6.02670796e-02 -1.39486063e+00 -4.60141718e-01 -4.43049788e-01
-3.44115734e-01 -2.59371817e-01 6.05372190e-01 -5.85427523e-01
1.52508140e+00 -2.04640198e+00 4.43205953e-01 2.63612926e-01
-1.76364601e-01 4.35266674e-01 -5.17081320e-01 8.70357454e-01
-4.86809403e-01 -1.49021707e-02 -2.52566427e-01 -5.46981037e-01
-2.88226102e-02 6.33739948e-01 -7.43875325e-01 4.50005054e-01
1.10615224e-01 9.76949751e-01 -6.50530100e-01 8.29713792e-02
1.32307515e-01 4.51354653e-01 -4.21104282e-01 1.21522941e-01
-2.59141803e-01 3.72015238e-01 -2.69430190e-01 -2.29858086e-02
1.82534009e-02 -5.12942076e-01 2.30351284e-01 -7.49685317e-02
-1.88458373e-03 4.80877489e-01 -8.04149210e-01 1.84799504e+00
-6.49331689e-01 7.13074744e-01 -2.98809230e-01 -1.02433598e+00
6.27118647e-01 5.00963748e-01 6.34187996e-01 -4.46565360e-01
-7.13815540e-02 2.47993812e-01 2.54592597e-01 -1.66258991e-01
3.46384831e-02 -1.22539535e-01 -3.89873125e-02 4.97331709e-01
3.07361633e-01 6.61455914e-02 9.06342641e-02 4.56109196e-01
1.55331230e+00 2.28719264e-02 9.18034986e-02 -2.59754658e-01
3.95259172e-01 -5.18535912e-01 5.34659863e-01 1.07601917e+00
1.31397724e-01 2.82108009e-01 4.88324493e-01 -4.83215064e-01
-1.11143458e+00 -1.13277793e+00 -1.54609233e-01 9.15245533e-01
-5.87622464e-01 -5.50095975e-01 -4.98463869e-01 -2.69981742e-01
-8.57674852e-02 1.00504923e+00 -6.94810331e-01 -7.35592365e-01
-7.44612932e-01 -6.59813821e-01 7.87592053e-01 7.58672655e-01
8.27070549e-02 -7.26653397e-01 -3.52643937e-01 4.85057443e-01
-3.13367285e-02 -1.07687926e+00 -5.12443364e-01 6.43611491e-01
-1.09539914e+00 -6.43521845e-01 -5.72274566e-01 -4.42830801e-01
2.93002427e-01 -2.78701991e-01 7.54579961e-01 -3.93613130e-01
-3.24553251e-02 5.48229992e-01 -1.06747754e-01 -6.39650077e-02
-7.17209697e-01 2.53442317e-01 4.02700424e-01 1.80174679e-01
-1.99858144e-01 -7.15981066e-01 -2.91510075e-01 2.70088464e-01
-1.12303436e+00 2.62761563e-01 7.06116259e-01 1.15314567e+00
3.92610341e-01 -2.18944371e-01 4.53362823e-01 -6.56361878e-01
2.76861370e-01 -3.61421734e-01 -5.13758481e-01 3.54411960e-01
-8.77828956e-01 6.80851877e-01 7.60017097e-01 -7.78807759e-01
-8.62717748e-01 1.21480137e-01 -2.18592212e-01 -8.21955919e-01
1.82923391e-01 7.01616347e-01 4.02695835e-01 3.26106071e-01
6.12563431e-01 8.42742443e-01 3.92934352e-01 -5.88725567e-01
3.22170168e-01 3.89183879e-01 6.15231872e-01 -6.31979346e-01
7.83791959e-01 2.70118088e-01 4.53910977e-01 -8.72926056e-01
-9.40035701e-01 -4.14938986e-01 -5.38949728e-01 8.89966637e-02
6.67666852e-01 -8.92496765e-01 -9.69923139e-01 8.89316320e-01
-1.09530461e+00 -9.12796557e-01 -2.61883110e-01 6.67966843e-01
-8.09630096e-01 2.43483439e-01 -1.30071712e+00 -8.50759625e-01
-2.73523331e-01 -8.85888875e-01 7.88099945e-01 -3.68391365e-01
-6.17029369e-02 -1.13746059e+00 1.64356202e-01 -1.13191798e-01
5.90564489e-01 -1.77845180e-01 1.35490644e+00 -7.98276782e-01
-6.62633896e-01 -2.11629376e-01 2.50680745e-01 8.12514126e-01
-2.83430293e-02 -2.42827013e-02 -6.53128207e-01 -4.92122561e-01
1.69291973e-01 1.12833060e-01 1.03731036e+00 5.89224100e-01
9.52652097e-01 -4.03523892e-01 -2.87194610e-01 6.04222178e-01
1.03177750e+00 3.79340112e-01 4.09221143e-01 4.40088138e-02
7.38906324e-01 2.10921630e-01 2.38857195e-01 3.95884544e-01
1.14014991e-01 6.03280425e-01 3.23728323e-01 2.21320480e-01
9.70788598e-02 -3.49978834e-01 7.90763021e-01 1.32256091e+00
9.92219225e-02 -1.82469487e-01 -1.07534480e+00 2.50064462e-01
-2.18012452e+00 -9.07835007e-01 -1.03343651e-01 2.21617723e+00
9.05513167e-01 2.91475534e-01 -4.63520326e-02 8.27004686e-02
3.37683022e-01 1.97830543e-01 -8.91563892e-01 -1.54559240e-01
-5.86733595e-02 2.58355975e-01 7.90458322e-01 5.96699774e-01
-9.03503418e-01 7.05447555e-01 6.94128942e+00 9.37919259e-01
-1.25568473e+00 4.33472618e-02 4.54732001e-01 -3.82054597e-01
2.71277390e-02 -3.14883851e-02 -9.42007542e-01 3.01739961e-01
1.91568851e+00 -5.95404729e-02 5.86787760e-01 3.95577490e-01
3.09692979e-01 4.42054272e-01 -1.47314811e+00 8.30882668e-01
-1.00255832e-01 -1.49157858e+00 1.28611520e-01 -1.37765417e-02
6.11060321e-01 4.21077907e-01 3.72779757e-01 5.68225384e-01
2.41755277e-01 -9.40389514e-01 7.99196124e-01 5.05194485e-01
8.55263412e-01 -5.94696701e-01 3.65249306e-01 6.75456583e-01
-1.14764953e+00 -3.27154458e-01 -9.72823203e-02 2.40228549e-02
3.90424669e-01 2.65089631e-01 -7.58934677e-01 4.77183074e-01
3.65028888e-01 1.15503180e+00 -3.30983162e-01 5.16046405e-01
-8.48661885e-02 1.39938569e+00 -4.45203781e-01 -1.02612283e-02
4.57293600e-01 4.52345870e-02 7.33614266e-01 1.32455397e+00
1.50422063e-02 -3.99339609e-02 2.75749732e-02 5.66278100e-01
3.19501370e-01 -4.31141406e-01 -3.49047542e-01 -2.51950055e-01
2.42824793e-01 6.88443005e-01 -2.66681343e-01 -4.18144286e-01
-2.53468990e-01 8.42565656e-01 1.69702232e-01 6.50546372e-01
-8.74963284e-01 -4.40891720e-02 7.36651778e-01 5.27418293e-02
2.70764530e-01 -6.40149653e-01 1.02111846e-01 -1.18339050e+00
-1.73079073e-01 -1.01089740e+00 2.03034505e-01 -5.93961835e-01
-1.15278220e+00 7.44710326e-01 2.48705640e-01 -1.06886578e+00
-7.57835329e-01 -7.15636969e-01 -3.28495592e-01 8.03362429e-01
-1.12763679e+00 -1.16897440e+00 5.01390398e-01 6.05201721e-01
4.83299732e-01 -1.36537611e-01 7.57364213e-01 1.80424944e-01
-6.84310138e-01 4.72411484e-01 4.19603854e-01 -2.62584649e-02
3.41956347e-01 -1.13323665e+00 9.53300536e-01 1.02217507e+00
2.27392897e-01 1.14237928e+00 7.81294048e-01 -6.10738635e-01
-1.74560678e+00 -1.26989138e+00 6.26347601e-01 -4.73106027e-01
1.13470232e+00 -5.28330386e-01 -1.21283996e+00 1.19908571e+00
-4.45522442e-02 -4.84039225e-02 4.67754126e-01 2.00149253e-01
-7.57747412e-01 -1.91099256e-01 -3.29764158e-01 6.41253352e-01
1.21228576e+00 -8.97526205e-01 -5.78231275e-01 2.97305137e-01
1.02781916e+00 -2.94312328e-01 -8.70804906e-01 4.38176066e-01
5.80967188e-01 -4.48365211e-01 1.14747357e+00 -9.86978650e-01
1.47714630e-01 -2.00362176e-01 -1.18669592e-01 -1.34453142e+00
-6.76406801e-01 -9.22993541e-01 -4.52674240e-01 7.35904872e-01
5.57858050e-01 -9.29515541e-01 5.24378717e-01 4.63685542e-01
-5.00705898e-01 -9.42352593e-01 -1.12605786e+00 -1.35379076e+00
2.06198677e-01 -6.39533401e-01 1.59307361e-01 3.70344937e-01
-3.69187593e-02 6.03293657e-01 -6.42617702e-01 1.29362494e-01
4.10149157e-01 -1.69891611e-01 4.63544607e-01 -1.01415968e+00
-6.41307831e-01 -3.09702963e-01 -2.44811401e-01 -1.55534577e+00
4.32728916e-01 -8.71529102e-01 1.89815417e-01 -1.20773733e+00
1.93492949e-01 -3.23522568e-01 -6.37208521e-01 5.86727500e-01
-1.24479756e-02 -4.27978396e-01 3.38922232e-01 3.22414041e-01
-3.71056169e-01 6.53842628e-01 8.80989254e-01 -1.74623579e-01
-1.54403463e-01 4.12355572e-01 -2.88403872e-02 3.36522490e-01
5.27900875e-01 -4.52389479e-01 -5.36608875e-01 -1.88575685e-01
2.37960204e-01 4.71586019e-01 2.60362148e-01 -1.14705098e+00
4.34803993e-01 -1.93694338e-01 2.82791965e-02 -7.16959417e-01
7.94845164e-01 -6.71764135e-01 5.46113312e-01 1.05431950e+00
-7.60729373e-01 1.45207882e-01 3.94985139e-01 1.05338001e+00
-5.03930002e-02 1.85252771e-01 7.46247590e-01 2.32249662e-01
-5.79754233e-01 3.62232864e-01 -8.57521176e-01 -2.64116704e-01
5.48672616e-01 2.11749196e-01 -3.16760302e-01 -5.83305240e-01
-8.54715228e-01 6.96730018e-02 -4.93680593e-03 6.19174063e-01
2.73923784e-01 -1.14294541e+00 -5.96558988e-01 2.87831694e-01
-7.15499446e-02 -3.06660026e-01 5.54093301e-01 1.00964224e+00
4.68003489e-02 9.24683452e-01 1.13799721e-02 -1.03263175e+00
-1.11510909e+00 5.23307204e-01 4.45759833e-01 -6.13609076e-01
-6.56770170e-01 7.93905437e-01 2.88788587e-01 -4.18714017e-01
2.35120103e-01 -9.06118274e-01 2.47894734e-01 -3.68222773e-01
2.99290866e-01 2.62779921e-01 3.86935547e-02 -4.38620269e-01
-4.41307127e-01 3.80700260e-01 -1.11068025e-01 -3.12905103e-01
1.31439793e+00 3.56565267e-02 1.94327846e-01 1.13334775e+00
1.33162808e+00 -7.12326884e-01 -1.51252878e+00 -5.85782230e-01
-2.47382186e-02 2.31142879e-01 -5.95726259e-02 -7.79977679e-01
-6.93482399e-01 9.92421091e-01 5.39013386e-01 -1.19951153e-02
8.19491565e-01 -1.37519553e-01 6.70709670e-01 7.38933563e-01
4.23231393e-01 -7.45960951e-01 1.43318892e-01 1.26286232e+00
8.81504059e-01 -5.89747429e-01 -2.92547226e-01 7.16262832e-02
-5.32787561e-01 9.20942843e-01 1.26116186e-01 -1.77574009e-01
7.62134969e-01 3.92455488e-01 -2.31008992e-01 2.66511232e-01
-1.59068894e+00 2.61011273e-01 3.69339138e-01 2.78149098e-01
1.53821588e-01 2.30939120e-01 2.52479762e-01 4.06647503e-01
6.48232326e-02 -2.02996150e-01 3.42465132e-01 8.91142547e-01
-2.52263129e-01 -1.16363394e+00 -4.91633303e-02 5.27974725e-01
-2.45862052e-01 -2.23276287e-01 1.05712183e-01 6.19061172e-01
-7.63811409e-01 7.58086920e-01 2.06359759e-01 -5.32510877e-01
2.77916610e-01 4.40476298e-01 5.37836730e-01 -6.44324481e-01
-6.24841630e-01 2.06869077e-02 -1.54433340e-01 -8.42060804e-01
-4.48899902e-02 -7.47370005e-01 -1.37405729e+00 -2.33410329e-01
-1.38366625e-01 -7.14571625e-02 8.16563129e-01 9.82348323e-01
6.37462199e-01 8.18225920e-01 3.74351531e-01 -8.25785220e-01
-1.27510619e+00 -1.23357701e+00 -3.78674775e-01 2.38094613e-01
7.11613297e-01 -4.31893080e-01 -4.60666537e-01 4.27332334e-02] | [7.5192413330078125, 3.4071807861328125] |
e229bb24-1cad-4886-bb15-27deb663ed9f | meeting-summarization-a-survey-of-the-state | 2212.08206 | null | https://arxiv.org/abs/2212.08206v1 | https://arxiv.org/pdf/2212.08206v1.pdf | Meeting Summarization: A Survey of the State of the Art | Information overloading requires the need for summarizers to extract salient information from the text. Currently, there is an overload of dialogue data due to the rise of virtual communication platforms. The rise of Covid-19 has led people to rely on online communication platforms like Zoom, Slack, Microsoft Teams, Discord, etc. to conduct their company meetings. Instead of going through the entire meeting transcripts, people can use meeting summarizers to select useful data. Nevertheless, there is a lack of comprehensive surveys in the field of meeting summarizers. In this survey, we aim to cover recent meeting summarization techniques. Our survey offers a general overview of text summarization along with datasets and evaluation metrics for meeting summarization. We also provide the performance of each summarizer on a leaderboard. We conclude our survey with different challenges in this domain and potential research opportunities for future researchers. | ['Arman Kabiri', 'Lakshmi Prasanna Kumar'] | 2022-12-16 | null | null | null | null | ['meeting-summarization'] | ['natural-language-processing'] | [ 2.09415093e-01 1.98226795e-01 -2.21418500e-01 -3.53412300e-01
-1.13735104e+00 -6.87925637e-01 4.60822761e-01 9.42629039e-01
-9.61429253e-02 9.29006636e-01 1.03025150e+00 -8.64540711e-02
-8.17011520e-02 -2.17344210e-01 2.90186793e-01 1.64359715e-02
1.45436645e-01 3.50151092e-01 -5.01105152e-02 -5.49640119e-01
1.00940478e+00 -8.31958875e-02 -1.42366707e+00 6.26832545e-01
1.21404147e+00 3.11840415e-01 4.81688708e-01 1.01540339e+00
-7.18252838e-01 8.14738989e-01 -1.60935271e+00 -1.58650830e-01
-3.07535857e-01 -6.00864708e-01 -1.06519163e+00 3.14742237e-01
5.47683775e-01 -9.55250338e-02 -5.70469238e-02 8.09346318e-01
8.08708429e-01 3.44079494e-01 2.91976333e-01 -1.09062457e+00
-1.43370911e-01 1.08041847e+00 -6.33061349e-01 5.14458835e-01
1.02929068e+00 -4.28730905e-01 1.33803272e+00 -7.85186410e-01
8.75711322e-01 1.12042475e+00 3.78173470e-01 4.01309818e-01
-6.35878086e-01 -4.25128996e-01 4.24695969e-01 -4.77523804e-02
-8.78869891e-01 -5.73588312e-01 7.29171157e-01 -3.45142424e-01
1.19016182e+00 9.30876374e-01 6.09618008e-01 8.65939379e-01
1.89454243e-01 8.04539740e-01 4.62994248e-01 -5.32485545e-01
1.00812376e-01 2.03641474e-01 9.67745662e-01 4.38867331e-01
5.01508534e-01 -1.15271926e+00 -1.09685981e+00 -5.20947456e-01
1.53114900e-01 -1.92316607e-01 -3.62415195e-01 1.75318331e-01
-1.33410943e+00 6.87099397e-01 -2.42857486e-01 5.14766216e-01
-3.44845593e-01 -2.48576120e-01 9.01433885e-01 5.36092579e-01
7.02198207e-01 1.08275294e+00 3.65884416e-02 -4.95430559e-01
-1.28837132e+00 5.43726027e-01 1.41790712e+00 9.21306551e-01
4.78753686e-01 -9.32757482e-02 -5.20949423e-01 9.39066648e-01
-7.40208179e-02 -1.66082364e-02 5.87828398e-01 -1.02274597e+00
8.93010259e-01 8.22444439e-01 3.53151470e-01 -1.20959044e+00
-6.12816036e-01 -3.32515568e-01 -7.21213579e-01 -5.09817541e-01
-7.18505979e-02 -5.09363472e-01 -1.64791867e-01 9.39373791e-01
-4.66334298e-02 -5.75841844e-01 -5.40012978e-02 4.77944314e-01
1.54572237e+00 7.20054269e-01 -3.81522477e-01 -8.97128284e-01
1.23556948e+00 -1.13447452e+00 -1.41432297e+00 -2.72780061e-01
7.66494811e-01 -1.12423766e+00 9.70997870e-01 4.61319745e-01
-1.37428546e+00 -3.22610319e-01 -1.10934365e+00 -1.30438119e-01
-3.49638835e-02 3.24236974e-02 6.14496946e-01 4.72255588e-01
-1.11179328e+00 4.95384783e-01 -5.81229627e-01 -8.61141920e-01
-1.28423199e-01 2.04729587e-01 7.41272345e-02 2.74756908e-01
-8.71951580e-01 6.96492493e-01 3.55409086e-02 -2.52733052e-01
-2.71878809e-01 -6.62748277e-01 -7.65737057e-01 -4.64155227e-02
4.10248756e-01 -9.24308300e-01 1.74144995e+00 -4.31271523e-01
-1.46354723e+00 6.16139352e-01 -4.73909616e-01 -1.70949876e-01
3.95720571e-01 -6.22136176e-01 -1.46426946e-01 -3.30381356e-02
3.56246829e-01 2.23824605e-02 5.81019595e-02 -1.06176639e+00
-8.20879221e-01 -1.66367933e-01 4.60326150e-02 7.13010907e-01
-4.65240270e-01 6.41539216e-01 -3.02424371e-01 -3.50130409e-01
-6.80314153e-02 -3.34518790e-01 -4.57326591e-01 -1.06992745e+00
-9.62337852e-01 -5.34939766e-01 7.09732831e-01 -7.83615708e-01
2.27247882e+00 -1.84109735e+00 1.37828305e-01 -1.22980699e-01
5.33582151e-01 8.51523057e-02 1.32232606e-01 1.42022943e+00
2.02726677e-01 3.34823191e-01 6.17207959e-02 -7.16064692e-01
-9.28583220e-02 -3.60334545e-01 -4.71758693e-01 1.29624173e-01
-4.68735009e-01 5.89867473e-01 -1.05609667e+00 -6.52647734e-01
1.25622734e-01 -9.78435278e-02 -2.32154906e-01 6.04366027e-02
-1.11630887e-01 2.46466100e-01 -6.81243360e-01 4.17275131e-01
3.18638593e-01 -4.46992010e-01 3.36174667e-01 4.32596236e-01
-7.44434059e-01 9.01506484e-01 -1.13407290e+00 1.82165670e+00
-2.93918848e-01 9.76103246e-01 2.60073632e-01 -7.21056700e-01
1.14753544e+00 4.57313418e-01 4.96638745e-01 -4.43895981e-02
2.31146459e-02 1.78575382e-01 -3.02027673e-01 -3.80254239e-01
1.42599034e+00 4.53519613e-01 -5.61308205e-01 8.36661100e-01
-3.04894298e-01 -3.58526230e-01 7.28110850e-01 9.35002923e-01
1.26041305e+00 -7.42806792e-01 7.30247200e-01 -2.94667125e-01
4.18722898e-01 3.32823485e-01 2.76507884e-01 9.62759435e-01
-1.40401363e-01 5.18168449e-01 4.80451822e-01 -3.90846610e-01
-6.26714528e-01 -5.61163306e-01 2.76255280e-01 1.22117364e+00
1.74607560e-01 -1.41746294e+00 -7.06987798e-01 -4.31735396e-01
-3.07241708e-01 8.23939621e-01 -1.56925067e-01 2.87617713e-01
-5.83770394e-01 -3.49085093e-01 1.56463221e-01 1.42605469e-01
4.22986478e-01 -9.97417927e-01 -6.37553990e-01 3.98741484e-01
-8.30427289e-01 -7.18380332e-01 -7.29851305e-01 3.69482413e-02
-9.41981316e-01 -6.16082430e-01 -4.79610473e-01 -8.42762232e-01
3.28872919e-01 7.96568811e-01 1.30921447e+00 2.24286050e-01
-1.47528097e-01 6.42832577e-01 -4.86295551e-01 -8.47697377e-01
-4.65126455e-01 7.87039042e-01 1.83614232e-02 -5.33849359e-01
3.38875800e-01 -4.34853166e-01 -4.95417714e-01 4.54283021e-02
-5.59433997e-01 2.18455747e-01 2.68879291e-02 3.03315401e-01
-5.04438430e-02 -2.72196352e-01 9.95207727e-01 -1.00139129e+00
1.64327765e+00 -2.55775034e-01 -4.09598313e-02 3.20200890e-01
-3.06491494e-01 -3.40623707e-01 2.15014055e-01 7.27857351e-02
-1.19349182e+00 -3.70054871e-01 2.90970892e-01 6.54363513e-01
2.39017680e-01 8.66525054e-01 2.64911890e-01 5.49679279e-01
8.99320722e-01 -8.03961456e-02 -1.66410238e-01 -4.73928064e-01
3.49438377e-02 1.14616311e+00 1.47769019e-01 -2.09091127e-01
2.99310058e-01 1.64753303e-01 -7.40195930e-01 -1.37773836e+00
-1.12425339e+00 -1.10629392e+00 -2.52216935e-01 -4.96492475e-01
5.88005483e-01 -7.46514082e-01 -5.86260974e-01 -2.59543192e-02
-1.51609981e+00 -4.46196720e-02 -3.05734932e-01 2.35468760e-01
-2.36377090e-01 7.76008368e-01 -7.01826632e-01 -8.89419734e-01
-8.30134153e-01 -8.38422716e-01 8.31480384e-01 5.95144868e-01
-1.35495353e+00 -9.68872130e-01 3.48562121e-01 7.06933677e-01
3.69207442e-01 1.62272036e-01 3.95259589e-01 -1.07535982e+00
-2.36755803e-01 -3.55611503e-01 2.00779364e-01 7.52445385e-02
5.49127698e-01 2.58182406e-01 -5.86566150e-01 -2.29828879e-01
1.58133474e-03 -1.74465239e-01 8.17763925e-01 6.60225630e-01
9.52049971e-01 -5.10439813e-01 -7.07587779e-01 -2.22765297e-01
7.82929480e-01 -7.61901662e-02 2.21640512e-01 3.29070240e-01
3.63406241e-01 8.96741807e-01 7.93392897e-01 1.04963017e+00
5.69131017e-01 3.50146651e-01 -6.43532141e-04 3.98792922e-01
5.32360226e-02 -3.86845879e-02 3.92511666e-01 1.78194237e+00
-3.93175557e-02 -5.40994167e-01 -8.58811915e-01 5.17148972e-01
-2.03977489e+00 -9.31754768e-01 -3.97113830e-01 1.96401620e+00
9.41859901e-01 -4.60130796e-02 1.78370327e-01 1.28426120e-01
9.20197070e-01 5.52601278e-01 -1.10508978e-01 -9.10106063e-01
5.00605702e-02 -2.62611508e-01 -3.26527171e-02 6.72467113e-01
-8.44262421e-01 6.69996858e-01 7.02611971e+00 4.07466233e-01
-6.91321492e-01 -3.29992115e-01 3.32132041e-01 -3.00708443e-01
-4.07371700e-01 -8.64634290e-02 -1.09075940e+00 2.25086272e-01
7.84982622e-01 -1.06152081e+00 -3.13945822e-02 7.49481142e-01
5.15876830e-01 -3.79028201e-01 -1.12356305e+00 1.05078828e+00
4.15757418e-01 -1.61177146e+00 5.88587224e-02 -8.25062841e-02
1.07379436e+00 -2.98241556e-01 -2.27877453e-01 3.64632994e-01
4.44641888e-01 -8.54738832e-01 8.96802694e-02 2.55841732e-01
1.69709280e-01 -6.79149628e-01 6.25159144e-01 5.85610688e-01
-9.67024624e-01 1.29418686e-01 -3.45583171e-01 -5.62090397e-01
4.30057585e-01 7.79052556e-01 -1.17320347e+00 8.75936210e-01
4.09472436e-01 7.37923145e-01 -3.85016054e-01 1.12290871e+00
1.10756587e-02 3.21436793e-01 1.55407004e-02 -4.19892520e-01
-3.14697698e-02 -1.77128106e-01 1.00692177e+00 1.51921678e+00
2.23330349e-01 -6.72451481e-02 7.10181177e-01 -1.15614450e-02
-1.93609253e-01 4.47019935e-01 -5.53938985e-01 -1.30665317e-01
8.44376028e-01 1.38414145e+00 -5.92945337e-01 -7.02955425e-01
-2.28772014e-01 8.80624354e-01 1.05990237e-02 2.25799426e-01
-3.06583494e-01 -7.87030518e-01 4.82757598e-01 1.09099373e-01
-3.28867584e-01 -4.65881646e-01 -6.18656456e-01 -9.41330314e-01
-5.44285774e-03 -1.28893697e+00 5.41965961e-01 -4.63648230e-01
-9.04187918e-01 4.24545109e-01 6.14564642e-02 -8.59311879e-01
-2.47243986e-01 2.06271946e-01 -1.04613125e+00 5.72421670e-01
-1.04518497e+00 -3.20768982e-01 -6.61813915e-01 -6.78708479e-02
1.37616265e+00 -4.00988683e-02 7.94953227e-01 -2.89184432e-02
-5.53936005e-01 2.16910750e-01 2.36750245e-02 -2.84992546e-01
1.28494322e+00 -1.35238695e+00 5.04548073e-01 2.78190553e-01
-3.83276008e-02 9.03836012e-01 1.25683868e+00 -8.49275231e-01
-1.17134595e+00 -6.50485814e-01 1.57713616e+00 -7.87370682e-01
6.69427156e-01 -1.80768132e-01 -7.68509328e-01 7.37152934e-01
9.90428150e-01 -1.11695755e+00 1.05863476e+00 4.45220977e-01
2.35026285e-01 3.15996222e-02 -5.92064381e-01 9.04203951e-01
9.11605299e-01 -1.63453862e-01 -9.13334370e-01 5.79110622e-01
8.35555077e-01 -5.18057585e-01 -5.05206347e-01 -1.21525079e-01
1.53465390e-01 -7.31431186e-01 3.56928468e-01 -4.06864494e-01
4.73288596e-01 -5.60991764e-02 2.45179474e-01 -1.60364497e+00
-1.43737311e-03 -1.35111260e+00 7.71177560e-02 1.52510548e+00
5.17381132e-01 -3.56196076e-01 6.52449846e-01 6.67582572e-01
-5.32843947e-01 -3.28875601e-01 -4.56298441e-01 -3.20084363e-01
-3.40426773e-01 -3.65765952e-02 1.89611956e-01 8.46153140e-01
1.17590845e+00 1.00685632e+00 -2.96724617e-01 -3.66363347e-01
3.24514568e-01 2.17411608e-01 1.12788248e+00 -1.60617387e+00
1.54368907e-01 -5.54861605e-01 8.87518823e-02 -1.29619479e+00
-1.82978988e-01 -6.80501163e-01 -6.24384917e-02 -2.50173926e+00
4.97310996e-01 3.34160149e-01 1.21935271e-01 -1.08698525e-01
-2.32263491e-01 -2.47130349e-01 1.85288340e-01 1.56569675e-01
-1.38925838e+00 2.39452556e-01 1.34419286e+00 -1.56808645e-01
-8.92588615e-01 2.07980916e-01 -1.22698402e+00 6.65416360e-01
1.07460153e+00 -2.24387914e-01 -4.38714832e-01 -2.22791255e-01
7.02449143e-01 3.17889631e-01 -4.16038007e-01 -8.63827348e-01
6.76950157e-01 -1.54495344e-01 -9.10869539e-02 -1.07899630e+00
2.25596905e-01 -5.44012301e-02 -2.60454893e-01 2.06715107e-01
-7.28885949e-01 4.01794553e-01 9.45599526e-02 3.47716421e-01
-4.84838068e-01 -3.30429584e-01 1.17484219e-01 -2.75033772e-01
-2.79706836e-01 -1.87938198e-01 -9.27722633e-01 2.78600395e-01
8.23787391e-01 -3.80223244e-01 -5.86976111e-01 -9.61294889e-01
-4.50661063e-01 7.25110412e-01 3.11758280e-01 4.96630341e-01
4.87418234e-01 -8.08458507e-01 -9.83467579e-01 -5.24425566e-01
1.48872077e-01 9.48655307e-02 3.60478520e-01 6.82084203e-01
-3.59657705e-01 5.89779615e-01 7.22534880e-02 -4.20561284e-01
-1.83080518e+00 -4.22627293e-02 -2.58987606e-01 -6.23434067e-01
-7.02471614e-01 6.42787397e-01 1.67499483e-01 -1.49219930e-01
4.85200971e-01 -2.31334865e-01 -6.16794765e-01 6.38969362e-01
1.14938772e+00 8.31989467e-01 2.38312975e-01 8.91846120e-02
-4.28782016e-01 -3.14348713e-02 -4.38702255e-01 -1.81358233e-01
1.13177764e+00 -6.72066033e-01 -3.18964541e-01 9.44436610e-01
7.21735954e-01 3.83086026e-01 -4.26197022e-01 -1.11209452e-01
5.35566151e-01 -1.71922967e-01 -3.14564914e-01 -6.59141243e-01
-1.17213503e-01 5.05140126e-01 -4.34867114e-01 9.07431543e-01
5.52794218e-01 5.69771789e-02 5.79804838e-01 6.49052143e-01
4.15027179e-02 -1.52650094e+00 6.26306057e-01 9.27649558e-01
1.33130538e+00 -1.09427738e+00 4.13211673e-01 -4.66973960e-01
-9.99751568e-01 1.11198258e+00 3.98251593e-01 -1.39274569e-02
3.00014853e-01 2.73599058e-01 1.71540186e-01 -4.44504201e-01
-1.09552002e+00 3.23315650e-01 5.33333309e-02 3.91813278e-01
1.09599447e+00 4.48586941e-02 -6.85909629e-01 6.69733644e-01
-7.62117565e-01 -2.27294669e-01 1.30693531e+00 9.70037401e-01
-1.04908466e+00 -1.04752624e+00 -4.01923627e-01 7.35595882e-01
-6.12821698e-01 3.38116437e-02 -1.03076172e+00 6.79420173e-01
-9.14950967e-01 1.62446237e+00 1.22196414e-01 -4.14421447e-02
5.33337593e-01 1.13644049e-01 8.40948001e-02 -1.33720911e+00
-1.21383619e+00 -1.17495239e-01 7.77299583e-01 -2.29729652e-01
-3.17907184e-01 -7.82303274e-01 -1.04329026e+00 -7.39138067e-01
-3.74892920e-01 8.13194215e-01 7.12717712e-01 5.28403580e-01
4.83024895e-01 8.24472725e-01 4.93741989e-01 -6.22319221e-01
-7.00834930e-01 -1.35341549e+00 -4.29899842e-01 -2.01515406e-01
3.88191611e-01 -1.02233663e-01 -2.79410779e-01 -8.95238444e-02] | [12.608647346496582, 9.399872779846191] |
83d9ccf1-b945-4708-a0a6-1433aeb77656 | a-statistical-method-for-object-counting | 1807.08335 | null | http://arxiv.org/abs/1807.08335v1 | http://arxiv.org/pdf/1807.08335v1.pdf | A Statistical Method for Object Counting | In this paper we present a new object counting method that is intended for
counting similarly sized and mostly round objects. Unlike many other algorithms
of the same purpose, the proposed method does not rely on identifying every
object, it uses statistical data obtained from the image instead. The method is
evaluated on images with human bone cells, oranges and pills achieving good
accuracy. Its strengths are ability to deal with touching and partly
overlapping objects, ability to work with different kinds of objects without
prior configuration and good performance. | ['Jans Glagolevs', 'Karlis Freivalds'] | 2018-07-22 | null | null | null | null | ['object-counting'] | ['computer-vision'] | [ 1.83134720e-01 -3.64712328e-01 1.86760053e-01 -4.13785614e-02
-1.67730048e-01 -3.95821869e-01 4.76887852e-01 4.87182826e-01
-9.63238478e-01 1.01480818e+00 -3.89250070e-01 4.24118293e-03
5.55935428e-02 -7.81121850e-01 -7.68776089e-02 -5.49520731e-01
1.06387779e-01 1.10190880e+00 1.01479816e+00 3.41369331e-01
6.72904730e-01 9.18510020e-01 -1.49298847e+00 5.63800782e-02
1.84671760e-01 5.91724932e-01 5.39117634e-01 1.04873109e+00
-2.18431488e-01 8.79040301e-01 -8.15235198e-01 -1.41684413e-01
2.64668148e-02 -7.43657202e-02 -7.44633853e-01 2.42514938e-01
2.82223761e-01 -3.95291388e-01 2.95495242e-01 7.35796571e-01
4.92728829e-01 -1.37709215e-01 1.20242822e+00 -9.34978545e-01
-3.33719581e-01 2.18872771e-01 -1.04243350e+00 6.88179374e-01
5.24837554e-01 -2.99735069e-01 3.47287446e-01 -9.62158084e-01
3.69414687e-01 1.22356999e+00 8.78307283e-01 4.32135582e-01
-9.28426266e-01 -6.13259971e-01 -4.44189429e-01 -2.32533649e-01
-1.62239420e+00 1.02503307e-01 6.03192784e-02 -5.27246416e-01
7.97235131e-01 3.00223589e-01 4.92362320e-01 2.87371844e-01
1.22878507e-01 6.29480541e-01 1.38918686e+00 -6.68046951e-01
1.70176417e-01 2.30105370e-01 2.79761821e-01 7.49958158e-01
9.01027739e-01 -2.43995190e-01 2.46978970e-03 -2.44489923e-01
9.53159750e-01 2.68145233e-01 2.82451868e-01 -5.30970767e-02
-1.36130369e+00 6.37905777e-01 -9.92332399e-02 7.98426628e-01
-2.29623854e-01 2.94777215e-01 2.99254209e-01 -4.64953214e-01
-5.25287502e-02 1.47282928e-01 -3.61553371e-01 -1.57065675e-01
-9.70110774e-01 1.23064399e-01 8.65747035e-01 9.69889641e-01
2.89797783e-01 -1.32077992e-01 -1.84931234e-01 5.82127392e-01
1.85210437e-01 5.95427275e-01 3.51675153e-01 -6.68202221e-01
2.41078734e-01 7.93534815e-01 4.28124875e-01 -7.90331364e-01
-8.75626922e-01 1.32006198e-01 -7.03926027e-01 5.47751844e-01
7.62942016e-01 1.63667306e-01 -1.29016352e+00 9.78319168e-01
3.38572592e-01 3.76282670e-02 -1.79397881e-01 4.26715970e-01
1.11794698e+00 4.30974483e-01 1.14560015e-01 -2.52715528e-01
1.75817990e+00 -6.47124410e-01 -6.20884597e-01 2.42935002e-01
1.64509937e-01 -1.15818751e+00 6.50290251e-01 6.65258586e-01
-1.22942507e+00 -4.52798784e-01 -8.95634651e-01 -7.54022300e-02
-7.27061868e-01 3.17263782e-01 5.98387003e-01 9.09003079e-01
-7.57509828e-01 4.25790876e-01 -6.15485489e-01 -6.56116843e-01
5.27067482e-01 8.08690608e-01 -2.71356463e-01 4.29697514e-01
-2.49937966e-01 1.07550657e+00 6.98270977e-01 -2.90059000e-01
-2.90174216e-01 -7.10525662e-02 -3.63962859e-01 -7.55591467e-02
1.73681572e-01 -5.41418135e-01 1.05717432e+00 -5.67520082e-01
-1.46437633e+00 1.13476026e+00 -3.56656343e-01 -3.06933135e-01
6.84175789e-01 -1.32663235e-01 -9.25988108e-02 1.35646701e-01
1.26093149e-01 5.85260987e-01 3.91661644e-01 -1.29994571e+00
-8.54242563e-01 -3.67587030e-01 -2.83074319e-01 -3.13245170e-02
3.11870407e-02 2.84896702e-01 -4.06155258e-01 -6.67866170e-01
2.30244577e-01 -7.42273450e-01 -4.66695763e-02 -1.69279203e-01
-3.34145844e-01 -3.23825568e-01 1.04532027e+00 -1.77770630e-01
8.73677373e-01 -1.66591930e+00 -4.15056169e-01 2.96545923e-01
1.39940426e-01 5.27146637e-01 2.85765707e-01 4.42095608e-01
4.54710215e-01 -1.03342928e-01 -8.55196416e-02 -3.87561917e-01
-4.28792924e-01 3.63860428e-01 2.13099644e-01 5.13126135e-01
2.45834212e-03 5.01295090e-01 -9.69655037e-01 -1.24015021e+00
4.34106559e-01 4.17982459e-01 5.68837002e-02 -2.83225060e-01
4.28201199e-01 2.60273576e-01 -4.60518837e-01 9.91579592e-01
8.77925754e-01 -3.37061912e-01 -1.12439424e-01 8.15499015e-03
-1.63111195e-01 -4.51448351e-01 -1.71606922e+00 8.60406816e-01
-1.66262507e-01 3.53761017e-01 -4.92602028e-02 -5.36357403e-01
9.46888149e-01 2.08192512e-01 4.82466877e-01 -3.83514196e-01
4.93967772e-01 3.24078411e-01 1.78802591e-02 -3.95698190e-01
6.30602956e-01 -3.25332910e-01 2.05601856e-01 5.56983173e-01
5.33242971e-02 -3.76464576e-01 7.36784995e-01 1.05801439e-02
9.79507387e-01 -3.05051625e-01 1.13970780e+00 -4.74253207e-01
8.08701873e-01 -1.24517143e-01 1.25979483e-01 8.47090721e-01
-1.89302832e-01 7.85527110e-01 -2.02420028e-03 -4.81543750e-01
-1.01502514e+00 -1.20407641e+00 -7.65067816e-01 1.03615928e+00
5.43149412e-01 1.74007967e-01 -6.94752395e-01 -4.76048440e-01
5.59636652e-02 3.42151165e-01 -8.43769133e-01 8.14978242e-01
-6.89124703e-01 -9.22368050e-01 3.53764623e-01 8.80939960e-01
4.68839616e-01 -1.25867295e+00 -1.19353640e+00 2.53547817e-01
8.43729302e-02 -1.15036309e+00 -1.90900937e-01 1.62856519e-01
-1.10426342e+00 -1.36393309e+00 -9.01340544e-01 -8.24509859e-01
8.36454690e-01 4.26496923e-01 1.35540712e+00 6.51970565e-01
-6.62527978e-01 3.78958255e-01 -2.63329595e-01 -1.07713425e+00
-2.52702475e-01 -1.41516477e-01 -9.91228372e-02 -4.77443755e-01
7.00592339e-01 -1.70810804e-01 -3.59734952e-01 6.19894743e-01
-7.11450160e-01 -2.49752566e-01 5.97002983e-01 8.20714116e-01
5.06683111e-01 1.78485334e-01 3.79977971e-01 -1.05425179e+00
4.34467465e-01 -1.18089415e-01 -7.31666982e-01 1.82891071e-01
-2.64882535e-01 -5.14313243e-02 3.50560576e-01 -5.53192377e-01
-7.58731008e-01 2.50830203e-01 2.51216620e-01 3.26082110e-01
-2.58796692e-01 -4.63594675e-01 3.44085902e-01 -1.94606259e-01
5.24852335e-01 1.49381533e-01 -1.46161720e-01 -4.19802129e-01
-1.48909882e-01 5.88294148e-01 6.11488044e-01 -2.38630012e-01
5.79433978e-01 1.07104135e+00 4.89278615e-01 -8.04981053e-01
-4.73290056e-01 -9.49279547e-01 -9.48525906e-01 -1.19983323e-01
8.18279684e-01 -2.87068933e-01 -1.13389838e+00 4.97825652e-01
-1.21906543e+00 8.29468146e-02 -1.99769303e-01 7.39638984e-01
-4.63510543e-01 4.65290695e-01 -6.44741178e-01 -1.22736132e+00
-5.61141610e-01 -7.44640529e-01 1.13707876e+00 6.12102747e-01
-1.86482012e-01 -9.40182269e-01 1.04099937e-01 6.85068918e-03
2.08991006e-01 2.57236183e-01 2.94612557e-01 -6.95000350e-01
-3.90300870e-01 -6.10268235e-01 -4.92534846e-01 -2.63138767e-02
4.30914730e-01 3.70888352e-01 -7.65799344e-01 -1.15666434e-01
-6.61998540e-02 -1.73433810e-01 8.02059114e-01 6.77586973e-01
1.11847317e+00 1.63966805e-01 -7.97123134e-01 8.80654529e-02
1.77202356e+00 5.09510398e-01 8.26167405e-01 2.65525818e-01
3.66164804e-01 4.62843299e-01 6.63484573e-01 5.01282692e-01
7.71935061e-02 4.27917033e-01 5.40509522e-01 -1.80681735e-01
-5.71731403e-02 3.01258236e-01 -4.17450607e-01 4.83956486e-01
-6.16688132e-01 -4.54254031e-01 -9.04694200e-01 5.46787500e-01
-1.44510090e+00 -1.23315144e+00 -5.72490573e-01 2.25864172e+00
4.16609615e-01 2.14668900e-01 4.32560384e-01 5.35311997e-01
1.01613843e+00 -5.16479075e-01 -9.32648331e-02 -3.67834806e-01
7.19175860e-02 5.13775826e-01 8.69871318e-01 3.97370726e-01
-1.10596979e+00 5.66981554e-01 7.99593544e+00 8.35967898e-01
-6.93100572e-01 1.50768697e-01 4.62370396e-01 4.97768074e-02
5.87432563e-01 -3.98444712e-01 -1.17185819e+00 5.52143157e-01
1.66212469e-01 2.63583571e-01 -1.12981960e-01 4.74097848e-01
-1.94262981e-01 -1.11026502e+00 -7.89041042e-01 1.03339529e+00
1.11638561e-01 -1.01074874e+00 -1.60066709e-01 -2.34328546e-02
5.63998640e-01 -2.79653937e-01 -3.09854090e-01 -8.92632976e-02
3.68358940e-01 -1.26798689e+00 5.70418417e-01 7.24216700e-01
4.51533973e-01 -6.09434366e-01 1.28415084e+00 3.57691228e-01
-1.32810533e+00 1.11956090e-01 -5.99078000e-01 -3.24967980e-01
8.67584497e-02 4.60457146e-01 -1.22223806e+00 8.20318311e-02
6.71840310e-01 2.84285769e-02 -6.71919584e-01 1.51047421e+00
2.68418133e-01 1.47683918e-01 -8.79983425e-01 -6.03404045e-01
-3.66479680e-02 -1.35063455e-01 1.40044406e-01 1.65433228e+00
5.11483371e-01 3.86699051e-01 2.61346120e-02 3.55307639e-01
4.49858993e-01 4.58351195e-01 -5.98736584e-01 5.39298117e-01
6.04868352e-01 1.34236717e+00 -1.77723837e+00 -7.47250140e-01
-4.45321172e-01 6.95496023e-01 7.62010440e-02 -1.59308672e-01
-7.58596301e-01 -2.08457872e-01 -3.78440857e-01 3.13252836e-01
5.31452656e-01 -1.42892137e-01 -5.23496449e-01 -4.71090347e-01
-2.29542553e-01 -1.89710513e-01 5.76058686e-01 -7.04631865e-01
-1.13127768e+00 4.39985305e-01 3.08128536e-01 -1.33452928e+00
1.52322561e-01 -1.01971459e+00 -6.89990938e-01 6.19779468e-01
-9.38925862e-01 -1.12391102e+00 -5.11046529e-01 4.98248845e-01
3.79812896e-01 -1.80238187e-01 7.52113581e-01 8.42780918e-02
-1.06357867e-02 2.50701606e-01 2.61120468e-01 1.40518680e-01
4.96308923e-01 -1.52066123e+00 -1.59970045e-01 4.06075716e-01
1.51514336e-01 4.64904934e-01 6.33315444e-01 -5.35714805e-01
-7.55754173e-01 -5.92640996e-01 7.24126220e-01 -7.12052107e-01
2.46562749e-01 -1.15176365e-01 -6.30560935e-01 2.26107582e-01
6.29472882e-02 3.46309543e-01 5.63653469e-01 -6.50740489e-02
1.03011511e-01 1.48635417e-01 -1.60460305e+00 -2.46302132e-02
6.10394597e-01 9.30514261e-02 -6.42455757e-01 4.95853990e-01
-3.11837137e-01 -5.86732686e-01 -5.88019371e-01 4.28831607e-01
7.26760328e-01 -1.16062641e+00 1.12226725e+00 -1.11317309e-02
-1.01894505e-01 -6.21800542e-01 1.05105042e-01 -3.70214254e-01
-3.17503065e-01 -1.13592111e-01 -2.08217241e-02 1.04060280e+00
5.60986474e-02 -5.76277673e-01 9.81669843e-01 -8.48761052e-02
3.92385751e-01 -5.23306131e-01 -1.10153735e+00 -1.09116554e+00
-1.57587737e-01 1.19635120e-01 4.79461670e-01 5.89158058e-01
-2.58659661e-01 2.79606968e-01 -2.40323514e-01 1.05380632e-01
7.10112095e-01 -7.38454564e-03 9.28218067e-01 -1.62671256e+00
-3.23153019e-01 -5.19936800e-01 -8.14415991e-01 -4.35808837e-01
-5.87970972e-01 -3.86099368e-01 -3.30632962e-02 -1.52309501e+00
7.32222319e-01 -5.83230555e-01 -1.27184361e-01 3.67020607e-01
-2.24243671e-01 9.60528255e-01 4.94450927e-02 1.84478655e-01
-6.61762953e-01 -3.91183823e-01 1.23563361e+00 1.67905524e-01
-3.28903296e-03 1.93122491e-01 -9.58389428e-04 1.15500510e+00
8.33867490e-01 -8.06386769e-01 -6.30077720e-02 -9.44047421e-02
-1.63243581e-02 -1.66534394e-01 4.43599343e-01 -1.44956625e+00
6.36675358e-02 -1.12664364e-01 6.75145924e-01 -1.12786388e+00
3.08495015e-01 -1.08132148e+00 3.91543061e-01 8.72187734e-01
4.71710503e-01 1.69879541e-01 1.32191226e-01 4.31645513e-01
-4.49943803e-02 -1.07905018e+00 1.29449725e+00 -4.98631895e-01
-4.53950167e-01 -1.66671008e-01 -3.61933321e-01 -2.81164587e-01
1.56405294e+00 -1.07370806e+00 -1.68836877e-01 7.20023960e-02
-8.25343847e-01 -1.22691713e-01 7.03536749e-01 -2.35850617e-01
2.16188371e-01 -1.14760149e+00 -5.67767262e-01 -2.05606475e-01
6.50637150e-02 5.20811714e-02 -2.27017373e-01 7.42162168e-01
-1.18055153e+00 4.96067554e-01 -3.50820601e-01 -7.38779664e-01
-1.76516044e+00 7.81265974e-01 5.38232401e-02 -5.61073959e-01
-4.47789907e-01 7.45501995e-01 2.01341093e-01 -1.05339065e-01
2.92042159e-02 -5.46944797e-01 -6.04771912e-01 1.01081654e-01
7.45278835e-01 9.38343406e-01 7.48786852e-02 -6.81429684e-01
-5.52811921e-01 1.06228328e+00 -2.53571719e-02 -5.23898266e-02
9.26196873e-01 9.41458791e-02 -2.66348626e-02 8.30542803e-01
6.93741739e-01 4.14420038e-01 -8.21987689e-01 -9.29941162e-02
1.22146243e-02 -7.60045767e-01 -3.96511704e-01 -6.80759192e-01
-7.61487067e-01 6.56052172e-01 8.83644164e-01 4.53256339e-01
8.53755713e-01 6.85912445e-02 3.22372377e-01 3.26191306e-01
7.73641348e-01 -1.20893204e+00 3.03810686e-01 4.58373666e-01
5.68062544e-01 -1.33371842e+00 7.05756485e-01 -5.36427617e-01
-4.01375264e-01 1.49109030e+00 4.89419669e-01 -3.71055245e-01
6.04252040e-01 8.13337922e-01 -1.38668954e-01 -3.65030617e-01
-2.31362864e-01 -4.65774417e-01 8.26427862e-02 9.32285488e-01
5.53096056e-01 7.01638535e-02 -8.62214267e-01 -1.45271823e-01
7.95113444e-02 3.78981203e-01 7.37390757e-01 1.34307373e+00
-7.57770956e-01 -8.86967361e-01 -9.23934281e-01 8.19654465e-01
-9.58049595e-01 9.76377875e-02 -1.42619044e-01 1.30890560e+00
4.48466629e-01 7.44809449e-01 3.35201442e-01 2.11636737e-01
6.57022372e-02 -2.91345954e-01 8.86625588e-01 -6.14219725e-01
-5.12966335e-01 -2.97089554e-02 -2.22113952e-01 -7.58206695e-02
-8.37795794e-01 -8.57826471e-01 -1.46265554e+00 -3.92692357e-01
-8.36870968e-01 -4.08699363e-02 4.94943261e-01 7.53938258e-01
-3.76544476e-01 1.32626727e-01 1.54263407e-01 -1.00220835e+00
2.51519103e-02 -1.25630414e+00 -1.02117884e+00 3.76474470e-01
5.01688570e-03 -1.06571674e+00 -1.16989374e-01 2.42699146e-01] | [14.673741340637207, -3.182264566421509] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.