Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files
xet
 Community
1
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
e6c2122d-4aad-486e-97ed-5c88553661b2
3d-human-pose-estimation-using-convolutional
1608.03075
null
http://arxiv.org/abs/1608.03075v2
http://arxiv.org/pdf/1608.03075v2.pdf
3D Human Pose Estimation Using Convolutional Neural Networks with 2D Pose Information
While there has been a success in 2D human pose estimation with convolutional neural networks (CNNs), 3D human pose estimation has not been thoroughly studied. In this paper, we tackle the 3D human pose estimation task with end-to-end learning using CNNs. Relative 3D positions between one joint and the other joints are learned via CNNs. The proposed method improves the performance of CNN with two novel ideas. First, we added 2D pose information to estimate a 3D pose from an image by concatenating 2D pose estimation result with the features from an image. Second, we have found that more accurate 3D poses are obtained by combining information on relative positions with respect to multiple joints, instead of just one root joint. Experimental results show that the proposed method achieves comparable performance to the state-of-the-art methods on Human 3.6m dataset.
['Sungheon Park', 'Jihye Hwang', 'Nojun Kwak']
2016-08-10
null
null
null
null
['2d-human-pose-estimation']
['computer-vision']
[-3.47203463e-01 1.83646798e-01 2.61323247e-02 -3.74214262e-01 -3.43907237e-01 -7.22541139e-02 3.56840253e-01 -1.61469266e-01 -9.59913015e-01 5.06062269e-01 3.82728577e-01 5.15494287e-01 2.80435383e-01 -3.60644966e-01 -8.07033777e-01 -2.07712725e-01 -1.87683463e-01 8.31013739e-01 2.95772016e-01 -3.81934583e-01 -1.34352118e-01 6.38591349e-01 -1.18808568e+00 -1.89130723e-01 1.96169153e-01 9.80787039e-01 -1.96262598e-01 5.16447723e-01 4.82629985e-01 3.36251169e-01 -7.64701009e-01 -2.22713679e-01 4.00608867e-01 -1.78686306e-01 -7.18768120e-01 2.23555878e-01 5.88966191e-01 -6.03416800e-01 -4.63488519e-01 5.89009166e-01 9.41035151e-01 1.55154973e-01 5.28874695e-01 -1.18651879e+00 -3.22562903e-02 -8.65403563e-02 -7.53050208e-01 -1.85975820e-01 6.84653699e-01 -8.42809752e-02 6.76935554e-01 -8.62174630e-01 6.41393244e-01 1.46419442e+00 9.83453870e-01 4.89606738e-01 -7.66891897e-01 -4.75115299e-01 -5.64895943e-02 6.23557083e-02 -1.57442307e+00 1.00424282e-01 9.16782439e-01 -5.04670620e-01 1.15104711e+00 -2.21438974e-01 1.13245940e+00 9.35221255e-01 4.76907343e-01 9.93756354e-01 7.99871206e-01 -4.50036973e-01 -2.53290594e-01 -4.54417884e-01 -3.30071032e-01 9.94033575e-01 2.51381546e-01 6.24588430e-02 -5.25622666e-01 1.31712765e-01 1.07033229e+00 6.17575981e-02 2.75419634e-02 -7.61873305e-01 -1.41243303e+00 5.95013618e-01 9.38331842e-01 1.14321858e-02 -6.04342222e-01 5.42321086e-01 4.44430023e-01 -2.18529552e-01 5.00752568e-01 2.52833188e-01 -7.00044692e-01 -2.06606165e-01 -7.22179174e-01 8.59510481e-01 5.77824414e-01 9.32041943e-01 4.12823647e-01 -2.98224717e-01 -1.11577228e-01 6.00000024e-01 3.17870617e-01 3.92421126e-01 2.16921464e-01 -9.31681991e-01 6.48565650e-01 6.31003261e-01 2.75333792e-01 -1.26273739e+00 -1.05498815e+00 -5.26116371e-01 -7.62231588e-01 1.60015479e-01 7.56487191e-01 -2.63730079e-01 -8.80667627e-01 1.64232326e+00 4.85365033e-01 -3.59502763e-01 -4.10983264e-01 1.45417380e+00 7.23793209e-01 2.67436713e-01 -4.78403941e-02 4.95430768e-01 1.48221505e+00 -1.10460675e+00 -5.58048010e-01 -4.72972274e-01 3.86872947e-01 -5.81882238e-01 3.38339627e-01 4.43365395e-01 -1.06917465e+00 -1.02034509e+00 -1.13965237e+00 -1.93073362e-01 -2.39142612e-01 5.35882056e-01 4.41870421e-01 4.24091965e-01 -7.05364883e-01 5.11275172e-01 -1.07234895e+00 -5.77013552e-01 1.39459893e-01 5.65320909e-01 -8.24671865e-01 2.59916723e-01 -1.31453705e+00 1.34312069e+00 5.08689046e-01 5.33102930e-01 -5.55598557e-01 -7.98979327e-02 -1.02338827e+00 -3.49086881e-01 5.47185123e-01 -1.11692667e+00 1.40275061e+00 -4.50345397e-01 -1.62881947e+00 9.91365373e-01 -5.24140671e-02 -4.72007573e-01 9.48244929e-01 -1.11151981e+00 1.79410979e-01 2.25530282e-01 5.94407469e-02 1.05431938e+00 6.05591297e-01 -1.05292141e+00 -3.91777962e-01 -7.93744147e-01 -6.74426183e-02 3.83610278e-01 -5.66894636e-02 -3.29322308e-01 -8.08775187e-01 -6.03289068e-01 4.59306598e-01 -1.25379443e+00 -4.16023582e-01 3.99545252e-01 -6.53447032e-01 -2.69086123e-01 5.49240768e-01 -7.83684611e-01 7.64806271e-01 -1.55379915e+00 5.52009642e-01 1.43014595e-01 2.34820351e-01 3.98912728e-01 1.42969951e-01 3.09125781e-01 3.27206627e-02 -3.57704997e-01 1.53454870e-01 -5.89175999e-01 8.36660638e-02 5.94437905e-02 4.15882230e-01 6.80354118e-01 3.03558290e-01 1.15485454e+00 -7.75082707e-01 -5.68514824e-01 4.80060756e-01 8.43558073e-01 -5.22506297e-01 3.89671236e-01 4.86344881e-02 6.88381255e-01 -3.47789645e-01 5.87802052e-01 4.45427835e-01 -9.04431418e-02 1.55874953e-01 -4.75593895e-01 2.32665166e-01 4.47683558e-02 -1.28402829e+00 2.20889688e+00 -2.18618095e-01 2.93799520e-01 -3.39747697e-01 -7.77114272e-01 1.05766261e+00 5.15451074e-01 5.96802354e-01 -3.28419477e-01 5.35030365e-01 1.36951223e-01 -2.16780603e-01 -3.59949440e-01 3.98600131e-01 2.50058118e-02 -3.92085105e-01 3.41941155e-02 3.89319062e-01 -3.85536477e-02 -9.18828323e-02 -4.01791751e-01 7.20841110e-01 7.16323495e-01 6.99359357e-01 1.32775187e-01 4.60949421e-01 -9.19291824e-02 3.83290172e-01 4.33352053e-01 -5.05302966e-01 8.62038732e-01 3.79761755e-01 -7.88562834e-01 -1.26758993e+00 -1.17038298e+00 3.09026986e-01 6.16647184e-01 -4.21862863e-02 -5.55133164e-01 -8.75950038e-01 -6.04478121e-01 2.10479826e-01 -2.22764820e-01 -6.81575477e-01 -6.21629283e-02 -1.02653432e+00 -3.00722867e-01 7.57947445e-01 1.01238573e+00 9.22448575e-01 -7.57269442e-01 -1.23813307e+00 2.15818554e-01 -4.89059955e-01 -1.30130899e+00 -5.04678488e-01 2.94045694e-02 -9.83425558e-01 -1.09931350e+00 -1.22893775e+00 -6.94233358e-01 4.74444330e-01 -1.58945486e-01 9.90264833e-01 -2.10071474e-01 -4.78620857e-01 2.00898230e-01 -2.79151857e-01 -3.31932247e-01 3.09669673e-01 3.31746936e-01 3.14075768e-01 -3.33731800e-01 3.40963840e-01 -2.48320118e-01 -6.72675848e-01 4.21039790e-01 -2.37311721e-01 4.81526740e-02 7.84550488e-01 6.38030529e-01 4.75464880e-01 -1.80725232e-01 1.89616859e-01 -3.16210330e-01 2.24639326e-01 2.87096918e-01 -2.52598822e-01 -1.48799837e-01 2.16594953e-02 1.92548241e-02 1.76228404e-01 -3.24023873e-01 -9.14134383e-01 8.34314108e-01 -4.01650071e-01 -3.51451576e-01 -5.53052664e-01 3.18081528e-01 -7.42354318e-02 1.34132177e-01 5.75743973e-01 -2.15409771e-01 4.32762623e-01 -6.50355756e-01 3.36821616e-01 4.03942585e-01 8.02870512e-01 -4.99123394e-01 7.66855597e-01 4.87019420e-01 2.94095397e-01 -5.61540067e-01 -9.25160885e-01 -6.14894509e-01 -1.50104690e+00 -5.27209163e-01 1.12970901e+00 -1.21641409e+00 -1.04149842e+00 7.46921480e-01 -1.49217784e+00 1.21899538e-01 1.91995457e-01 8.17361116e-01 -8.42910945e-01 3.27310354e-01 -7.73918808e-01 -7.36958146e-01 -4.37636405e-01 -1.13267362e+00 1.57925332e+00 -3.59603800e-02 -8.50435913e-01 -6.70966923e-01 1.09922916e-01 4.94548798e-01 -4.20686491e-02 8.71039569e-01 2.25761458e-01 -3.74330223e-01 -1.50900304e-01 -8.91089022e-01 1.10492751e-01 1.78953961e-01 -7.51030520e-02 -4.38935935e-01 -7.25690126e-01 -4.27167714e-01 -2.83482999e-01 -4.26884234e-01 7.07084298e-01 5.26351750e-01 7.26907372e-01 5.84344827e-02 -3.87992829e-01 4.48271573e-01 9.63538766e-01 -2.67230719e-01 5.60453296e-01 5.73398232e-01 9.19229448e-01 6.65641606e-01 7.09633946e-01 5.57063222e-01 3.07588965e-01 1.03776348e+00 4.70794708e-01 -1.70196950e-01 -1.25700280e-01 -3.77434880e-01 1.56936407e-01 5.79924524e-01 -6.12038672e-01 1.50285199e-01 -9.01351154e-01 2.81433642e-01 -2.01850009e+00 -5.18462479e-01 1.59115810e-02 2.01657844e+00 5.99599302e-01 4.61593240e-01 5.35005212e-01 1.72471836e-01 6.60994411e-01 1.70235872e-01 -4.19677347e-01 2.11370468e-01 3.08088124e-01 1.94828287e-01 5.03382802e-01 2.98971295e-01 -1.36169589e+00 9.25198615e-01 6.19266033e+00 3.03748071e-01 -8.50334167e-01 -1.65525496e-01 7.50157982e-02 -2.40128607e-01 6.45987034e-01 -3.39866281e-01 -9.68045890e-01 -5.41454274e-03 1.83632642e-01 4.89308923e-01 -2.80354083e-01 9.56018209e-01 9.46258828e-02 -9.42523777e-02 -1.21530271e+00 1.04848433e+00 2.56125510e-01 -6.78742409e-01 -8.12220350e-02 6.41032010e-02 5.18595397e-01 -3.40466321e-01 -1.58840120e-01 1.37499139e-01 -2.63938248e-01 -9.82521057e-01 1.07583857e+00 7.06173420e-01 4.70156938e-01 -1.09525633e+00 1.14954281e+00 6.08069122e-01 -1.39340520e+00 2.23034844e-01 -1.32384211e-01 -5.55047154e-01 2.89296925e-01 4.13537085e-01 -1.13458622e+00 5.87518036e-01 7.71022379e-01 6.94054604e-01 -7.11092472e-01 1.09523666e+00 -5.15279055e-01 -1.12917878e-01 -2.43453011e-01 -1.47438198e-01 1.32283717e-01 2.70649791e-01 5.11273503e-01 9.91704702e-01 1.28305927e-01 -2.03665599e-01 4.11053896e-01 6.13160074e-01 1.41456157e-01 -1.17203541e-01 -4.98467922e-01 2.76466370e-01 -2.62542837e-03 1.19421363e+00 -7.12899804e-01 -1.49928600e-01 4.56306676e-04 1.16347754e+00 3.06677848e-01 1.07474022e-01 -8.48032951e-01 -4.72359926e-01 5.85562110e-01 2.28066757e-01 2.42649078e-01 -7.82163382e-01 -4.53669354e-02 -9.11632597e-01 2.61585921e-01 -5.88600278e-01 2.43673623e-01 -8.88534367e-01 -1.13873541e+00 4.00954098e-01 2.62837440e-01 -1.27070820e+00 -5.66547334e-01 -8.80118310e-01 -1.13176301e-01 8.36866140e-01 -9.23448324e-01 -1.29762101e+00 -3.79704535e-01 3.52652878e-01 4.10628974e-01 1.44336507e-01 7.78718889e-01 1.04692593e-01 -1.54889241e-01 5.55599451e-01 -7.15891540e-01 8.22601318e-01 8.79219592e-01 -1.18360329e+00 7.42363513e-01 4.84809846e-01 -1.36490613e-01 6.35918498e-01 7.00640976e-01 -7.47848511e-01 -1.19646001e+00 -7.44477689e-01 1.24620736e+00 -6.86508834e-01 -1.93572901e-02 -4.70107019e-01 -2.54769385e-01 6.94722772e-01 -5.59000745e-02 -6.01321235e-02 2.69940108e-01 1.85077921e-01 -3.52228999e-01 3.71221341e-02 -1.05791843e+00 5.43437302e-01 1.12194014e+00 -2.78645992e-01 -7.99300075e-01 1.29010439e-01 6.32725775e-01 -8.70069742e-01 -9.78702784e-01 7.01885641e-01 1.00241745e+00 -7.28797555e-01 1.32087135e+00 -4.88714010e-01 3.03486288e-01 -4.77423519e-01 -1.18382208e-01 -1.24737704e+00 -2.65787035e-01 -1.82282984e-01 -1.91565275e-01 3.46279263e-01 4.62785037e-03 -1.89698711e-01 1.15669537e+00 3.20544302e-01 6.52542561e-02 -8.89451683e-01 -1.01323342e+00 -8.09594572e-01 -4.29526297e-03 -1.92029595e-01 3.49576205e-01 3.02881777e-01 -3.71013097e-02 4.41326648e-01 -8.02728415e-01 1.03380986e-01 6.62187278e-01 -1.06289543e-01 1.26014841e+00 -1.38046718e+00 -2.38391280e-01 -1.75040841e-01 -8.95894349e-01 -1.45203078e+00 1.23915160e-02 -3.94126981e-01 3.21039110e-01 -1.50744939e+00 1.49722338e-01 3.71390402e-01 9.94366780e-02 4.75135297e-01 -1.49235517e-01 4.82309520e-01 4.06796008e-01 -1.05771065e-01 -6.38009131e-01 4.92256641e-01 1.52071452e+00 3.83467264e-02 7.34715909e-02 6.03578053e-02 -1.31670916e-02 9.39280391e-01 6.78978562e-01 -2.95570672e-01 -7.32327020e-03 -2.87729889e-01 7.11107329e-02 2.17941493e-01 8.08552563e-01 -1.44826221e+00 2.61752635e-01 4.36766297e-01 1.27361882e+00 -1.24494684e+00 7.81034231e-01 -6.93616986e-01 8.48283470e-02 9.19700861e-01 -2.49858603e-01 2.92077810e-01 -1.54278567e-02 4.85936433e-01 -1.22279577e-01 2.05096304e-01 6.54740334e-01 -5.95747948e-01 -7.21016526e-01 4.08261031e-01 -2.09475607e-01 -9.82269272e-02 8.76049697e-01 -3.37685317e-01 4.37765241e-01 -4.30562437e-01 -1.09697437e+00 1.02326967e-01 2.07335368e-01 6.39655113e-01 7.96033919e-01 -1.58767831e+00 -5.54463267e-01 1.01089649e-01 6.54910430e-02 3.49356383e-01 1.75138637e-02 7.77414858e-01 -6.81943715e-01 7.48893082e-01 -5.90063334e-01 -9.30886924e-01 -1.23016477e+00 2.44038075e-01 5.19257665e-01 -2.35334188e-01 -5.08427739e-01 8.84241819e-01 -2.91685641e-01 -9.38194811e-01 6.11879528e-01 -2.72456050e-01 -1.23920269e-01 2.57591885e-02 2.50182986e-01 3.96277964e-01 1.40020117e-01 -1.05980551e+00 -6.44574463e-01 1.02354705e+00 5.47090247e-02 -1.86168820e-01 1.30992734e+00 1.09470770e-01 1.76001415e-01 3.88859183e-01 1.60775042e+00 -5.73717475e-01 -1.40547931e+00 -2.98942983e-01 -2.07659975e-01 -5.82356274e-01 -3.96676540e-01 -7.21524119e-01 -1.09781897e+00 9.13095117e-01 6.52158797e-01 -5.21240473e-01 7.97650814e-01 -4.17075418e-02 1.13422406e+00 6.19282544e-01 6.24107778e-01 -1.32749867e+00 7.19993412e-01 6.59522593e-01 9.43955719e-01 -1.20001698e+00 2.98938364e-01 -3.29615861e-01 -4.32275385e-01 1.20411086e+00 9.46512759e-01 -4.92201567e-01 6.88960671e-01 -6.76267743e-02 1.72026366e-01 -1.95381105e-01 -2.58530885e-01 -3.85855734e-01 7.10785687e-01 6.06138587e-01 7.89690375e-01 7.78149590e-02 -2.94139296e-01 3.58186692e-01 -2.55665153e-01 9.13051516e-02 -2.24903319e-02 1.18971276e+00 -5.63111544e-01 -9.95482564e-01 -6.73170865e-01 -9.71305072e-02 -3.76873791e-01 4.46227044e-01 -5.89009881e-01 1.16354167e+00 2.97269195e-01 6.70607686e-01 4.27178107e-02 -6.00209296e-01 8.50351989e-01 1.54226184e-01 9.99221444e-01 -5.28722823e-01 -4.22070473e-01 2.89827615e-01 -1.10587804e-03 -7.97735155e-01 -5.09748161e-01 -5.87006509e-01 -1.26499498e+00 -6.64869249e-02 -1.86174840e-01 -2.50911623e-01 6.79974377e-01 9.68253493e-01 9.76671204e-02 5.12824059e-01 -1.77172795e-02 -1.61007845e+00 -7.07492769e-01 -1.00883210e+00 -5.25928199e-01 4.56402928e-01 2.29765579e-01 -1.22886336e+00 1.13369919e-01 -1.58425063e-01]
[6.962892055511475, -0.840390682220459]
0ee6144d-1376-4b26-b79b-aea92ee504a7
domain-independent-svm-for-transfer-learning
1903.11020
null
http://arxiv.org/abs/1903.11020v1
http://arxiv.org/pdf/1903.11020v1.pdf
Domain Independent SVM for Transfer Learning in Brain Decoding
Brain imaging data are important in brain sciences yet expensive to obtain, with big volume (i.e., large p) but small sample size (i.e., small n). To tackle this problem, transfer learning is a promising direction that leverages source data to improve performance on related, target data. Most transfer learning methods focus on minimizing data distribution mismatch. However, a big challenge in brain imaging is the large domain discrepancies in cognitive experiment designs and subject-specific structures and functions. A recent transfer learning approach minimizes domain dependence to learn common features across domains, via the Hilbert-Schmidt Independence Criterion (HSIC). Inspired by this method, we propose a new Domain Independent Support Vector Machine (DI-SVM) for transfer learning in brain condition decoding. Specifically, DI-SVM simultaneously minimizes the SVM empirical risk and the dependence on domain information via a simplified HSIC. We use public data to construct 13 transfer learning tasks in brain decoding, including three interesting multi-source transfer tasks. Experiments show that DI-SVM's superior performance over eight competing methods on these tasks, particularly an improvement of more than 24% on multi-source transfer tasks.
['Christopher R. Cox', 'Wenwen Li', 'Shuo Zhou', 'Haiping Lu']
2019-03-26
null
null
null
null
['brain-decoding', 'brain-decoding']
['medical', 'miscellaneous']
[ 1.57630205e-01 -2.65299201e-01 -1.54091418e-01 -6.75713718e-01 -9.82543945e-01 -4.24300194e-01 4.17369574e-01 -3.01888227e-01 -5.67281306e-01 1.09355962e+00 1.70934811e-01 -7.74724931e-02 -1.97985709e-01 -2.33119577e-01 -7.88969994e-01 -6.15321517e-01 4.47516255e-02 4.81182545e-01 2.13069871e-01 -9.94444191e-02 3.55973095e-01 7.97137469e-02 -9.33863163e-01 3.14054489e-01 1.23534477e+00 1.02144790e+00 4.67272907e-01 -1.15503818e-01 2.35365853e-01 3.57907355e-01 -4.31665957e-01 -2.69925088e-01 2.10181043e-01 -6.00411952e-01 -7.40091741e-01 -2.73422569e-01 2.83157408e-01 -3.49952638e-01 -2.41589814e-01 1.30076611e+00 8.03031743e-01 1.16243400e-01 1.36164010e+00 -1.61272883e+00 -9.23271298e-01 3.04845333e-01 -6.71181679e-01 6.56598091e-01 -1.16587564e-01 1.02167033e-01 7.34284878e-01 -1.09120941e+00 3.63046914e-01 9.82982635e-01 5.80534875e-01 4.16641504e-01 -1.41736722e+00 -1.21851492e+00 -3.20511162e-01 5.17376304e-01 -1.25796878e+00 -4.89951164e-01 7.01967597e-01 -9.24875140e-01 5.91493666e-01 -1.88232228e-01 3.36873531e-01 1.49233437e+00 3.81991506e-01 7.34930992e-01 1.73001182e+00 -1.05008647e-01 4.77967054e-01 3.82587969e-01 2.13432521e-01 3.63868415e-01 2.05776811e-01 1.06416456e-01 -7.00707734e-01 -9.03577954e-02 9.53499973e-01 -2.08715945e-01 -2.14715809e-01 -4.79071498e-01 -1.45286918e+00 1.01905811e+00 4.96005327e-01 1.49067953e-01 -1.12264968e-01 -4.74102855e-01 5.22011340e-01 5.02988517e-01 6.58178091e-01 2.94067025e-01 -6.81632161e-01 2.02595040e-01 -8.77349079e-01 4.03056480e-02 5.45153916e-01 1.06358075e+00 6.25691533e-01 -6.41169678e-03 -3.63493502e-01 1.13682437e+00 1.16160952e-01 7.34144688e-01 8.95375788e-01 -4.93309528e-01 8.16649258e-01 1.29607737e-01 -3.22091728e-01 -8.13470840e-01 -5.14203548e-01 -3.73618215e-01 -1.10934722e+00 -1.79579090e-02 4.16621029e-01 -3.01067591e-01 -6.25683248e-01 1.95492637e+00 1.48961060e-02 2.03591913e-01 -1.63265735e-01 9.39261794e-01 8.00247371e-01 3.31297010e-01 7.86622539e-02 -1.24083921e-01 1.33031416e+00 -9.41942930e-01 -3.64778489e-01 -5.40783882e-01 6.12479508e-01 -3.28645736e-01 1.14096344e+00 2.63254493e-01 -7.24005163e-01 -4.13329214e-01 -8.33356738e-01 2.79312171e-02 -7.61944205e-02 2.20893130e-01 5.84854722e-01 6.84462667e-01 -8.33990991e-01 4.14834023e-01 -5.62839150e-01 -1.39691651e-01 1.02169836e+00 4.54079449e-01 -6.43783808e-01 -6.66044578e-02 -1.15018392e+00 1.10015297e+00 5.95388174e-01 -4.09652561e-01 -9.90302265e-01 -1.18306208e+00 -7.06067622e-01 1.06605515e-01 1.19770080e-01 -4.75576907e-01 9.00094390e-01 -1.08201075e+00 -1.52702606e+00 1.04932356e+00 1.31729007e-01 -3.14553797e-01 4.01466131e-01 -8.81777331e-02 -3.13536674e-01 1.28680661e-01 4.80315953e-01 6.79735243e-01 1.08097744e+00 -8.13896298e-01 -2.47461364e-01 -6.61773205e-01 -6.67082191e-01 2.08751634e-01 -6.78445995e-01 2.55435139e-01 1.89524993e-01 -6.92617059e-01 -6.36967719e-02 -9.22811449e-01 2.35038787e-01 -3.49657983e-02 -1.34273931e-01 -1.85112193e-01 2.54501492e-01 -1.10170209e+00 4.93919700e-01 -2.25490928e+00 4.24281031e-01 2.25685816e-02 3.83699626e-01 1.21832296e-01 -3.49868387e-01 -2.36653507e-01 -3.80823612e-01 -3.31170708e-01 -4.90655571e-01 -4.87167835e-02 -3.33626643e-02 -1.38223708e-01 -1.70221090e-01 7.41717577e-01 3.64008725e-01 1.10703337e+00 -7.00582564e-01 -4.61469054e-01 -2.12763280e-01 1.31245285e-01 -7.55488038e-01 2.56996423e-01 3.13156396e-01 1.00970566e+00 -3.04948688e-01 4.21807736e-01 8.35446894e-01 -3.87357146e-01 -2.24700183e-01 -2.19868809e-01 2.63785779e-01 1.67602837e-01 -6.40613198e-01 1.95014000e+00 -2.85904706e-01 6.38227880e-01 -7.53281415e-02 -1.74741137e+00 9.18029368e-01 1.72582909e-01 3.95772249e-01 -9.09823298e-01 3.03649545e-01 5.41260660e-01 4.64756519e-01 -3.59378427e-01 -4.13459569e-01 -4.16269243e-01 8.39223061e-03 3.06063533e-01 6.41147256e-01 -3.02193873e-02 -4.38230604e-01 -1.65501043e-01 9.29455101e-01 -1.47410274e-01 4.54926193e-01 -8.36614192e-01 4.00948912e-01 -3.68703514e-01 6.76057994e-01 3.40926200e-01 -6.70842648e-01 5.87010801e-01 5.69484711e-01 2.78769508e-02 -8.72988284e-01 -1.11508679e+00 -4.12633240e-01 1.22726452e+00 -8.44609812e-02 2.48754531e-01 -8.19076478e-01 -7.32943892e-01 6.93132058e-02 5.65913558e-01 -7.27568030e-01 -5.71089208e-01 -3.63138258e-01 -8.86141658e-01 5.62986076e-01 6.45704985e-01 7.16506839e-01 -1.01605678e+00 -2.22981259e-01 -1.92511082e-01 -2.00007007e-01 -1.03825104e+00 -7.30758607e-01 4.11031246e-01 -1.02603865e+00 -6.52412832e-01 -1.30286813e+00 -9.74993110e-01 5.70652783e-01 3.14664871e-01 8.65781367e-01 -6.85640752e-01 -2.00509652e-01 2.03096494e-01 -2.36938938e-01 -4.98951554e-01 5.79629242e-02 2.83764720e-01 3.43653381e-01 7.64296129e-02 4.63680059e-01 -6.34352922e-01 -6.42960548e-01 5.74388981e-01 -5.21792471e-01 -5.90681098e-02 7.70738602e-01 1.20148718e+00 4.34151083e-01 -3.18299115e-01 1.14524686e+00 -5.89623988e-01 7.49912560e-01 -1.09205425e+00 -3.82207930e-01 3.05034935e-01 -5.97274065e-01 1.73654333e-02 6.08494461e-01 -7.88592935e-01 -1.03775907e+00 -3.44749421e-01 2.70086825e-01 -4.11693662e-01 -1.57600522e-01 6.53621197e-01 -3.87886792e-01 -2.34758809e-01 8.30067337e-01 4.63706255e-01 2.62121677e-01 -4.28099334e-01 2.08256505e-02 6.65979922e-01 3.31103712e-01 -5.77409744e-01 5.78589737e-01 1.20793141e-01 -1.21074669e-01 -6.89053416e-01 -8.65462542e-01 -3.19781035e-01 -8.86092484e-01 9.63607281e-02 8.95820200e-01 -1.05285895e+00 -1.79384887e-01 5.81140757e-01 -9.55487788e-01 -4.73193288e-01 8.80435854e-02 9.69492853e-01 -7.18026340e-01 8.93329829e-02 -3.36842865e-01 -1.90254182e-01 -1.87781408e-01 -1.33519936e+00 7.81445682e-01 6.04323158e-03 -2.98746023e-02 -1.04279208e+00 2.58506965e-02 5.05585372e-01 5.67681491e-01 -1.41784370e-01 9.69855726e-01 -1.15508127e+00 -1.26883373e-01 2.08124489e-01 -7.31095731e-01 5.91657102e-01 2.37977296e-01 -9.30258334e-01 -9.00527656e-01 -5.59593022e-01 4.84899700e-01 -6.15878880e-01 8.09203923e-01 6.71756983e-01 1.43723583e+00 -5.89202344e-02 -2.77728587e-01 7.30377614e-01 9.96564269e-01 2.09612355e-01 5.43530107e-01 2.25493804e-01 6.01766765e-01 7.99790144e-01 4.05220449e-01 3.11447442e-01 3.23031485e-01 6.84686065e-01 -1.78424209e-01 2.51938868e-02 -1.28954053e-01 -4.59684059e-02 5.73153436e-01 1.05375612e+00 4.66155224e-02 4.67731267e-01 -1.05701792e+00 3.43373865e-01 -1.65463519e+00 -6.89051569e-01 -3.40497121e-02 2.22938156e+00 1.16302431e+00 -6.61531985e-02 3.77330303e-01 -2.10111931e-01 7.18699872e-01 -3.04789573e-01 -9.86092806e-01 2.50752717e-01 -1.68514431e-01 4.14396524e-01 5.37649095e-01 -2.72918530e-02 -1.12637675e+00 8.15882623e-01 6.09232569e+00 1.14700437e+00 -1.08755910e+00 7.62990236e-01 7.18736053e-01 -8.95724222e-02 4.61723842e-02 -3.93893331e-01 -7.03126669e-01 7.73833811e-01 9.80971694e-01 -4.72163141e-01 5.18741250e-01 9.18626487e-01 -1.41663060e-01 8.05323496e-02 -1.30038559e+00 1.53390908e+00 3.34298790e-01 -1.03953826e+00 -3.40068310e-01 1.50840789e-01 7.08241642e-01 3.43458295e-01 3.29462200e-01 5.57191432e-01 -1.76111430e-01 -1.29749978e+00 6.43195331e-01 1.74874395e-01 1.12476766e+00 -5.76380551e-01 4.82195199e-01 4.22166139e-01 -6.98472738e-01 2.97965985e-02 -7.11464345e-01 1.25851771e-02 -1.90795407e-01 5.30975163e-01 -4.69835699e-01 1.81886524e-01 7.26683259e-01 9.51471567e-01 -4.25570488e-01 1.13269281e+00 -6.62034657e-03 7.79032946e-01 -3.01195192e-04 6.46458715e-02 -2.00081453e-01 -4.02612209e-01 3.32000941e-01 1.09991169e+00 3.26882988e-01 1.23567969e-01 1.13324352e-01 1.16435063e+00 -1.49824649e-01 3.51743281e-01 -6.41055405e-01 2.85565585e-01 2.87599385e-01 1.13547528e+00 -6.67065442e-01 -4.43489365e-02 -5.99383652e-01 1.13622463e+00 5.86805165e-01 2.91315675e-01 -1.03086638e+00 -2.26646483e-01 4.98325437e-01 -7.33069405e-02 1.42152265e-01 -3.05137843e-01 -6.60882771e-01 -1.47350180e+00 9.47812274e-02 -9.49827731e-01 1.67180747e-01 -5.20690322e-01 -1.94375086e+00 4.55160141e-01 1.67589203e-01 -1.28818464e+00 2.01624081e-01 -8.95777225e-01 -3.82709593e-01 1.13458800e+00 -1.65705049e+00 -1.04450798e+00 -4.22355272e-02 1.15407741e+00 4.58091408e-01 -8.70167315e-01 6.58543885e-01 5.71459413e-01 -5.74433088e-01 8.31284940e-01 4.07825410e-01 1.90080881e-01 1.34606218e+00 -9.73188579e-01 -9.81764793e-02 2.58606941e-01 -5.76878749e-02 5.17371714e-01 1.25152081e-01 -6.26299262e-01 -1.22116053e+00 -9.40276444e-01 4.27720457e-01 -3.92140418e-01 8.38813663e-01 -5.21984100e-01 -1.14245939e+00 7.35079110e-01 -7.63937039e-03 6.49192035e-02 1.21359992e+00 7.08248653e-03 -6.32021666e-01 -4.68105488e-02 -1.30523872e+00 3.05548161e-01 1.11250281e+00 -6.32068515e-01 -8.90407622e-01 6.23548985e-01 4.82153714e-01 -1.65028378e-01 -9.82223988e-01 1.79539666e-01 3.73552442e-01 -7.15297222e-01 1.07128167e+00 -8.97794545e-01 7.20205188e-01 1.91771328e-01 -2.54638106e-01 -1.84723341e+00 -6.02571785e-01 -9.49520096e-02 1.72937408e-01 1.02137542e+00 3.42700332e-01 -8.72805774e-01 5.41799128e-01 7.99654782e-01 -9.99262631e-02 -3.62784684e-01 -1.24590397e+00 -1.30969429e+00 8.08799207e-01 -8.97724703e-02 2.79733211e-01 1.23967338e+00 1.24568701e-01 5.74760556e-01 -3.33129823e-01 -1.02859333e-01 8.70271623e-01 -1.47379413e-01 4.70349193e-01 -1.56087804e+00 -2.54360288e-01 -3.79107803e-01 -3.30100536e-01 -8.77587199e-01 7.38598883e-01 -1.60534966e+00 2.62477640e-02 -9.43033755e-01 6.22590542e-01 -4.54290628e-01 -5.43389320e-01 4.48303670e-01 -2.28929356e-01 -3.05062439e-02 -7.61094503e-03 5.09365857e-01 -3.27209592e-01 8.92207801e-01 1.34590197e+00 -2.21641958e-01 -1.24704599e-01 -1.41286924e-01 -7.70342767e-01 6.17989242e-01 9.35695708e-01 -6.77412748e-01 -5.62288284e-01 -5.07294536e-01 -2.68008351e-01 3.66948522e-03 4.94580626e-01 -9.58452702e-01 1.28856957e-01 -2.26091683e-01 6.19856656e-01 4.02636789e-02 5.43451607e-02 -6.64014459e-01 -3.36016953e-01 4.12487417e-01 -3.61947268e-01 -2.79690683e-01 2.03562468e-01 4.84963089e-01 -1.31480932e-01 -2.09562242e-01 1.01591086e+00 1.10741034e-01 -5.90230763e-01 5.28475881e-01 -9.90587026e-02 6.55848980e-01 9.99418736e-01 -4.49723490e-02 -2.64071941e-01 6.03758655e-02 -5.87079406e-01 4.12823074e-02 -1.02882192e-01 4.76990521e-01 6.11757576e-01 -1.59642136e+00 -1.05036378e+00 3.40682149e-01 2.33774900e-01 -3.79990041e-01 1.85159042e-01 1.23449171e+00 1.24004722e-01 4.40514654e-01 -1.00572407e+00 -6.69482112e-01 -9.63874340e-01 3.01641583e-01 1.95340708e-01 -2.06716247e-02 -5.32719254e-01 1.02497816e+00 7.44934261e-01 -5.78665733e-01 -1.31307274e-01 -1.52746171e-01 -1.97331250e-01 3.56609076e-01 5.92402995e-01 3.75515550e-01 2.15599909e-01 -4.29742038e-01 -5.02645433e-01 2.99632519e-01 -2.48512506e-01 -2.71961153e-01 1.62700093e+00 1.16943151e-01 -1.02533318e-01 4.29327101e-01 1.60238373e+00 -4.41640645e-01 -1.36547649e+00 -6.00758791e-01 1.69193387e-01 -4.54066753e-01 1.24089457e-01 -7.93020546e-01 -1.11270249e+00 1.02194953e+00 6.98620498e-01 -3.99715990e-01 9.77323115e-01 2.42806971e-02 6.45531833e-01 2.79192328e-01 6.59001410e-01 -1.20245254e+00 3.64032537e-01 6.45378590e-01 1.26734459e+00 -1.72353601e+00 -1.87969640e-01 -2.87205070e-01 -1.00504637e+00 7.94491827e-01 7.35674143e-01 -3.02366138e-01 1.01886892e+00 -3.60716041e-03 -4.48774457e-01 -6.64725676e-02 -2.44872034e-01 1.75456539e-01 6.21960342e-01 9.72851753e-01 2.43655786e-01 2.29362324e-01 -3.18678439e-01 1.28819156e+00 -1.37839884e-01 1.53620481e-01 5.40835112e-02 5.89209616e-01 -2.57359803e-01 -9.10094082e-01 -3.30799371e-01 9.30841088e-01 -1.81007981e-01 -3.34473640e-01 -1.24264151e-01 4.87201452e-01 -9.97657254e-02 5.77355027e-01 -9.84406322e-02 -2.71664739e-01 1.67522952e-01 1.22664563e-01 7.70869076e-01 -7.90450573e-01 -2.86269277e-01 -6.76610395e-02 -4.75066215e-01 -4.33816582e-01 -2.75180757e-01 -8.73868048e-01 -9.18167949e-01 2.65108049e-02 -2.26956874e-01 -2.00700924e-01 5.03980815e-01 1.08965087e+00 3.95810306e-01 3.14016819e-01 6.42036736e-01 -8.62247169e-01 -9.84881938e-01 -1.13096797e+00 -1.01666248e+00 5.22774458e-01 1.38824284e-01 -1.22241962e+00 -2.66508400e-01 7.62935430e-02]
[12.626970291137695, 3.3440327644348145]
e18237a4-c0e6-4c7e-b31a-c75cc4a657e4
bayesian-bilinear-neural-network-for
2203.03613
null
https://arxiv.org/abs/2203.03613v2
https://arxiv.org/pdf/2203.03613v2.pdf
Bayesian Bilinear Neural Network for Predicting the Mid-price Dynamics in Limit-Order Book Markets
The prediction of financial markets is a challenging yet important task. In modern electronically-driven markets, traditional time-series econometric methods often appear incapable of capturing the true complexity of the multi-level interactions driving the price dynamics. While recent research has established the effectiveness of traditional machine learning (ML) models in financial applications, their intrinsic inability to deal with uncertainties, which is a great concern in econometrics research and real business applications, constitutes a major drawback. Bayesian methods naturally appear as a suitable remedy conveying the predictive ability of ML methods with the probabilistically-oriented practice of econometric research. By adopting a state-of-the-art second-order optimization algorithm, we train a Bayesian bilinear neural network with temporal attention, suitable for the challenging time-series task of predicting mid-price movements in ultra-high-frequency limit-order book markets. We thoroughly compare our Bayesian model with traditional ML alternatives by addressing the use of predictive distributions to analyze errors and uncertainties associated with the estimated parameters and model forecasts. Our results underline the feasibility of the Bayesian deep-learning approach and its predictive and decisional advantages in complex econometric tasks, prompting future research in this direction.
['Alexandros Iosifidis', 'Mostafa Shabani', 'Martin Magris']
2022-03-07
null
null
null
null
['econometrics']
['miscellaneous']
[-5.00262678e-01 -3.08599651e-01 -1.14146687e-01 -4.18428123e-01 -8.40305030e-01 -4.42868143e-01 9.94430304e-01 -1.46305770e-01 -2.85807759e-01 6.86093569e-01 8.80517811e-02 -7.14597583e-01 -7.41222680e-01 -6.88887894e-01 -6.55476928e-01 -6.87717915e-01 -3.66259009e-01 9.37064648e-01 -2.18857870e-01 -1.70672536e-01 5.16150296e-01 3.72889340e-01 -1.33345687e+00 -5.58320656e-02 6.58639371e-01 1.71795344e+00 -1.02416508e-01 2.93384284e-01 -2.57399559e-01 9.41736400e-01 -3.23675334e-01 -6.49786949e-01 2.79768050e-01 -6.92184865e-02 -2.77476043e-01 -1.79429382e-01 -2.07801163e-01 -3.88403147e-01 -2.16716260e-01 8.05275440e-01 2.26370454e-01 2.07458273e-01 9.00689363e-01 -1.08650720e+00 -6.81139290e-01 6.57816768e-01 -5.31569183e-01 6.41143262e-01 -2.01804638e-01 2.14953646e-01 1.30178308e+00 -8.36619139e-01 1.43690601e-01 1.11163700e+00 7.87583530e-01 -9.02849734e-02 -1.43330860e+00 -5.00460505e-01 2.85152942e-01 2.25743338e-01 -8.95081103e-01 -2.91803837e-01 1.02798617e+00 -9.88013387e-01 8.52573097e-01 4.30999920e-02 6.15842819e-01 1.16316342e+00 7.72250354e-01 6.67528927e-01 1.59155142e+00 -9.52819660e-02 2.86729902e-01 9.40572917e-02 1.15064397e-01 -1.51236698e-01 -2.55359313e-03 8.79368067e-01 -6.54347956e-01 -2.94739038e-01 6.43670976e-01 2.29772720e-02 1.58982143e-01 -2.49131948e-01 -9.80404019e-01 1.02058041e+00 7.84134939e-02 3.72735023e-01 -7.92606771e-01 3.66181195e-01 3.02091449e-01 3.70234042e-01 1.14691126e+00 2.93944925e-01 -8.78608465e-01 -2.82003701e-01 -1.27003253e+00 8.15652132e-01 9.67713237e-01 3.58937889e-01 1.37694463e-01 3.90594900e-01 5.80096208e-02 4.17925775e-01 6.14913702e-01 4.26622599e-01 3.61992210e-01 -6.52825415e-01 2.10342288e-01 1.10479183e-02 4.77360159e-01 -1.05041528e+00 -4.26709265e-01 -7.22375333e-01 -8.28476191e-01 3.47862065e-01 7.69769549e-01 -1.67663351e-01 -3.24206620e-01 1.36434519e+00 -5.15322620e-03 1.42453387e-01 -1.64353311e-01 7.85974860e-01 -1.21471755e-01 8.17709208e-01 4.57727201e-02 -3.56381476e-01 1.19391537e+00 -4.39374149e-01 -8.24041426e-01 -1.45216912e-01 5.16040958e-02 -7.68973589e-01 5.84158063e-01 5.89709997e-01 -9.85221148e-01 -3.34633380e-01 -6.63356125e-01 3.79579872e-01 -4.28537846e-01 -2.60051310e-01 7.47651875e-01 3.49870384e-01 -6.50738120e-01 1.00184739e+00 -9.98478353e-01 3.80068272e-01 1.16967462e-01 2.14416131e-01 2.87285119e-01 5.50804615e-01 -1.32046020e+00 1.30256510e+00 2.26346120e-01 6.72316909e-01 -5.03505468e-01 -1.02241039e+00 -4.92521912e-01 1.77599549e-01 2.95005977e-01 -3.91590476e-01 1.58165252e+00 -8.29942703e-01 -1.47513795e+00 2.83612996e-01 1.85019776e-01 -9.61309850e-01 9.62696970e-01 -3.67827475e-01 -4.95164931e-01 -2.65309602e-01 -1.56595349e-01 1.65698230e-01 1.11933684e+00 -7.93201387e-01 -5.15788972e-01 -2.41358235e-01 -3.67578149e-01 -1.76389426e-01 2.17269063e-01 -5.12054302e-02 5.22099316e-01 -1.08107388e+00 9.13286209e-02 -5.66859245e-01 -2.88806170e-01 -4.30304319e-01 -9.01938677e-02 -4.07833904e-01 3.65428507e-01 -1.01379335e+00 1.21715951e+00 -2.06226063e+00 -9.04535726e-02 3.18885028e-01 -1.10669397e-01 -2.90028155e-01 3.35650593e-01 7.34755456e-01 -1.93673462e-01 9.30492803e-02 -3.45975250e-01 -3.20749372e-01 7.11331308e-01 -1.41182318e-02 -9.29838181e-01 5.78620970e-01 5.90465724e-01 1.07572591e+00 -6.05070531e-01 6.11552708e-02 3.17877829e-01 4.02779132e-01 -3.83334428e-01 1.05727725e-01 -6.43896580e-01 5.48076510e-01 -4.36197668e-01 4.56991345e-01 5.22326946e-01 -1.64912894e-01 6.48589851e-03 9.21947975e-03 -3.77228767e-01 5.59497654e-01 -1.30723786e+00 9.11126792e-01 -3.14010680e-01 6.92475200e-01 -1.49622574e-01 -1.44600308e+00 8.24709594e-01 3.62328678e-01 5.47690749e-01 -9.91546273e-01 4.36495170e-02 6.45773113e-01 1.75483480e-01 -3.26607704e-01 3.63979489e-01 -6.92042530e-01 -1.06484592e-01 5.30287981e-01 -1.26042113e-01 -2.63290852e-01 -4.83964235e-02 -5.06320238e-01 4.38139468e-01 5.57059228e-01 4.18953300e-02 -5.53788722e-01 -2.74163242e-02 -7.11610466e-02 4.92840052e-01 7.85897076e-01 6.61209505e-03 2.74268389e-01 7.44017780e-01 -8.84903729e-01 -1.12691522e+00 -9.07062590e-01 -6.80804849e-01 7.55635858e-01 -5.68688452e-01 2.81694293e-01 -1.60547659e-01 -2.01478615e-01 5.73796749e-01 1.05261993e+00 -5.94733357e-01 2.24554256e-01 -3.50419194e-01 -1.21043670e+00 -1.75963014e-01 4.27132636e-01 6.46701679e-02 -1.04931474e+00 -7.50623703e-01 7.62156546e-01 1.33969918e-01 -7.52708912e-01 5.38078835e-03 4.70624387e-01 -8.52129936e-01 -9.15333211e-01 -7.82767594e-01 -6.95219776e-03 -7.59325400e-02 -5.28605998e-01 1.29082990e+00 -6.31676674e-01 1.12770185e-01 4.19024616e-01 1.15352131e-01 -7.63301313e-01 -3.72941703e-01 -1.81901410e-01 1.00934915e-01 3.10308665e-01 5.08244693e-01 -7.70376265e-01 -4.95466024e-01 7.35766441e-02 -8.42538774e-01 -3.89690876e-01 3.75935435e-01 1.09575331e+00 3.49179089e-01 2.86182076e-01 8.28681946e-01 -4.81842130e-01 8.31599712e-01 -7.39825487e-01 -1.51038742e+00 -2.32228283e-02 -9.39848840e-01 2.39409715e-01 2.59156913e-01 -5.28792441e-01 -1.08099163e+00 -4.60815489e-01 -5.13219042e-03 -2.07229391e-01 6.79661483e-02 1.27274370e+00 4.83123004e-01 3.67961049e-01 8.93171784e-03 2.65612751e-01 7.39460364e-02 -5.58130324e-01 6.88476115e-02 2.08540782e-01 3.45617652e-01 -7.67425358e-01 7.22934306e-01 2.27056235e-01 1.42761007e-01 -5.49180686e-01 -7.09828496e-01 -8.79662856e-02 -5.33897758e-01 -2.39464015e-01 6.17994249e-01 -8.48962843e-01 -8.32371473e-01 7.32259035e-01 -1.15774441e+00 -2.36521319e-01 -3.56296986e-01 7.03062296e-01 -8.23985755e-01 7.36876875e-02 -6.91276431e-01 -1.67132223e+00 1.55946806e-01 -1.07013822e+00 8.16392541e-01 -1.99981228e-01 -2.56062895e-01 -1.38473821e+00 1.99162781e-01 2.98529297e-01 6.95358276e-01 4.68458444e-01 1.07174420e+00 -9.40956295e-01 -6.91641688e-01 -3.81655663e-01 -9.11644101e-03 4.38935429e-01 -3.37277681e-01 6.59832582e-02 -9.27452385e-01 1.55678287e-01 5.19727111e-01 -8.84807706e-02 9.00395513e-01 9.27846849e-01 7.66439319e-01 -4.00829732e-01 2.10387394e-01 2.97445267e-01 1.31798029e+00 4.28834945e-01 1.81845933e-01 5.13335586e-01 -5.48058115e-02 1.04827774e+00 4.57943052e-01 7.39762306e-01 4.83382583e-01 5.17210245e-01 4.37748730e-01 1.88734174e-01 8.20358872e-01 -2.03025013e-01 3.09687883e-01 8.25110197e-01 9.29268636e-03 3.12418975e-02 -1.16417646e+00 3.89392108e-01 -2.05327320e+00 -1.06799686e+00 -1.16886742e-01 2.09584975e+00 6.78856194e-01 4.53263432e-01 2.16467842e-01 6.37064278e-02 3.16754967e-01 2.74919063e-01 -6.09740913e-01 -3.14577639e-01 -3.27564090e-01 1.15282074e-01 4.08061445e-01 3.53009105e-01 -1.39991295e+00 3.43107432e-01 6.41215277e+00 6.51080072e-01 -9.15097296e-01 -1.02403566e-01 1.11552882e+00 1.12741008e-01 -3.85294050e-01 -3.92823294e-02 -6.72375321e-01 8.39730620e-01 1.37639761e+00 -1.49688393e-01 5.74648201e-01 6.60592198e-01 5.12783110e-01 -3.54552835e-01 -1.27646899e+00 7.22218871e-01 -4.87564325e-01 -1.54851329e+00 -3.31198603e-01 5.22881031e-01 6.16153896e-01 6.91475049e-02 6.22017026e-01 3.20389301e-01 2.19088614e-01 -1.13416064e+00 1.26826465e+00 1.12023473e+00 -1.09903030e-02 -7.45960116e-01 1.12347126e+00 5.19225061e-01 -6.85589552e-01 -3.38601261e-01 -1.66792989e-01 -5.31996369e-01 4.70460027e-01 9.94343996e-01 -1.45098791e-01 3.03398520e-01 7.85018265e-01 6.82883620e-01 -5.72098419e-02 8.72944653e-01 2.43297786e-01 7.35345781e-01 -5.60930133e-01 -3.48464176e-02 5.19144118e-01 -6.71642482e-01 4.94789958e-01 8.24721932e-01 5.02728164e-01 -1.05311327e-01 -2.44271442e-01 1.29272056e+00 3.93923461e-01 -1.75031662e-01 -6.35374188e-01 -4.37891543e-01 1.25853941e-01 5.94800949e-01 -4.71927047e-01 -3.82224098e-02 -5.46315849e-01 1.91941589e-01 2.52315570e-02 4.35238540e-01 -7.14404762e-01 1.39587775e-01 5.04442394e-01 1.78124502e-01 5.34698367e-01 -6.19693875e-01 -3.91496420e-01 -1.16604102e+00 4.55810428e-01 -8.48940372e-01 2.64508247e-01 -5.05427241e-01 -1.84763861e+00 2.78067768e-01 2.14915082e-01 -9.89385605e-01 -7.35568047e-01 -9.34709549e-01 -5.04680216e-01 1.03507721e+00 -1.84908319e+00 -8.50978076e-01 7.57506430e-01 1.37512460e-01 4.52116638e-01 -2.23854154e-01 5.80313146e-01 2.30661128e-02 -4.42381829e-01 -2.70324379e-01 8.16704929e-01 -5.61559610e-02 2.74368018e-01 -1.31499910e+00 4.97802675e-01 6.24143720e-01 1.55308992e-01 4.20838535e-01 9.34336066e-01 -5.49266636e-01 -1.34796524e+00 -5.34606040e-01 1.08711922e+00 -5.41569889e-01 1.56811666e+00 -3.79839599e-01 -9.88274455e-01 5.56987226e-01 1.48812100e-01 -1.60433754e-01 4.26107973e-01 1.51882619e-01 -2.58493125e-01 -2.73843080e-01 -7.17004836e-01 3.67564589e-01 5.91280609e-02 -6.44590974e-01 -9.09325480e-01 2.64501393e-01 3.54065120e-01 -1.52035560e-02 -1.00627422e+00 3.77836108e-01 7.53789008e-01 -1.14852595e+00 8.43741655e-01 -5.73995292e-01 4.10094380e-01 1.44832775e-01 -1.47489071e-01 -1.26557112e+00 -1.59702659e-01 -8.77167761e-01 -3.86615992e-01 1.21037614e+00 4.47003007e-01 -1.03970158e+00 4.19448197e-01 9.12881672e-01 3.42023045e-01 -5.99945843e-01 -1.52832031e+00 -7.85559297e-01 5.40123940e-01 -9.59211230e-01 5.34866154e-01 7.79269397e-01 -1.62334859e-01 -1.44382805e-01 -5.30471206e-01 4.68618236e-02 8.35232913e-01 2.21626639e-01 2.62650669e-01 -1.66841340e+00 -5.02316952e-01 -9.88236547e-01 -6.39497936e-02 -7.99079239e-01 4.14169550e-01 -3.52603883e-01 -8.53938237e-02 -7.73753703e-01 -2.86430687e-01 -3.40305358e-01 -5.12930930e-01 -3.80362719e-01 2.31932893e-01 -2.17993081e-01 1.30272791e-01 2.45324507e-01 4.40016612e-02 7.52775371e-01 6.95180953e-01 -1.30509049e-01 5.51214330e-02 4.60146546e-01 -3.06015164e-01 6.85633481e-01 7.46079803e-01 -4.22188491e-01 -9.05449316e-02 -1.60438985e-01 7.65664756e-01 3.80751282e-01 6.83581054e-01 -5.78231990e-01 1.81040168e-01 -3.28277707e-01 4.26002115e-01 -9.23713505e-01 3.70937228e-01 -8.79248381e-01 3.01654458e-01 4.50433761e-01 -2.99966723e-01 5.33994555e-01 3.27634007e-01 7.52690613e-01 -5.77015698e-01 -8.68937075e-02 4.45505798e-01 -1.90415025e-01 -4.49887365e-01 2.12021291e-01 -6.91069961e-01 -2.63889786e-02 6.77601933e-01 3.17920744e-02 2.40535498e-01 -4.79026973e-01 -7.91113973e-01 2.75834858e-01 -1.60008803e-01 3.68495256e-01 3.17808956e-01 -1.03700638e+00 -8.85696352e-01 1.52095214e-01 -2.68999815e-01 -3.19658995e-01 2.10355297e-01 1.08666885e+00 -2.65062481e-01 8.03627610e-01 1.27499476e-01 -5.36949337e-01 -5.80875464e-02 6.26122892e-01 5.88366210e-01 -5.32044590e-01 -4.84725296e-01 5.74480712e-01 1.08439974e-01 -2.04259202e-01 2.35575438e-01 -6.17325604e-01 -5.90515099e-02 6.16031289e-01 1.49819195e-01 4.20619488e-01 8.52237716e-02 -3.72800499e-01 -8.63617212e-02 3.74467760e-01 2.55728841e-01 -4.95212793e-01 1.74510479e+00 -4.98010330e-02 -8.33536088e-02 1.06708384e+00 6.94439530e-01 -4.73182946e-01 -1.60773563e+00 -3.08892906e-01 9.79258597e-01 -3.22690159e-02 3.65929693e-01 -8.19119871e-01 -7.23673463e-01 1.17998004e+00 3.81089300e-01 5.48593998e-01 7.21456587e-01 -2.47392595e-01 5.41971028e-01 2.01013282e-01 2.90568441e-01 -1.32601416e+00 -3.80244225e-01 4.84984040e-01 1.08529282e+00 -1.42520690e+00 -1.82414964e-01 4.27430838e-01 -5.09009302e-01 1.38156998e+00 -1.79034963e-01 -4.16358888e-01 1.26993871e+00 3.41259688e-01 4.51706015e-02 -1.50400281e-01 -1.07727146e+00 9.04685408e-02 5.57888448e-01 3.91678751e-01 3.75982434e-01 1.67456105e-01 -1.56232491e-01 8.77153397e-01 -2.85967261e-01 -1.58094928e-01 2.04756573e-01 7.31000900e-01 3.05468589e-03 -8.49329948e-01 -4.01193053e-01 4.10466492e-01 -8.27637315e-01 -2.70949811e-01 9.70417857e-02 9.97338295e-01 -2.87413746e-01 7.28863776e-01 3.78599912e-01 2.06167504e-01 1.78316578e-01 5.11322677e-01 1.84249766e-02 -6.31056875e-02 -7.11903155e-01 6.32439137e-01 -3.73490117e-02 -3.71161640e-01 -3.61992598e-01 -1.21821439e+00 -3.47494841e-01 -3.26777816e-01 -1.87109232e-01 1.11728638e-01 7.72821009e-01 1.16454637e+00 6.05220310e-02 3.84966105e-01 6.23036385e-01 -9.99054193e-01 -1.48484910e+00 -9.35795009e-01 -9.86295402e-01 2.68829335e-02 7.64120042e-01 -7.41384387e-01 -5.87308943e-01 -5.07472418e-02]
[4.811517715454102, 4.05569314956665]
a6bf7e0b-c1c1-4c5e-8794-62c813565fd6
less-is-more-micro-expression-recognition
1606.01721
null
http://arxiv.org/abs/1606.01721v3
http://arxiv.org/pdf/1606.01721v3.pdf
Less is More: Micro-expression Recognition from Video using Apex Frame
Despite recent interest and advances in facial micro-expression research, there is still plenty room for improvement in terms of micro-expression recognition. Conventional feature extraction approaches for micro-expression video consider either the whole video sequence or a part of it, for representation. However, with the high-speed video capture of micro-expressions (100-200 fps), are all frames necessary to provide a sufficiently meaningful representation? Is the luxury of data a bane to accurate recognition? A novel proposition is presented in this paper, whereby we utilize only two images per video: the apex frame and the onset frame. The apex frame of a video contains the highest intensity of expression changes among all frames, while the onset is the perfect choice of a reference frame with neutral expression. A new feature extractor, Bi-Weighted Oriented Optical Flow (Bi-WOOF) is proposed to encode essential expressiveness of the apex frame. We evaluated the proposed method on five micro-expression databases: CAS(ME)$^2$, CASME II, SMIC-HS, SMIC-NIR and SMIC-VIS. Our experiments lend credence to our hypothesis, with our proposed technique achieving a state-of-the-art F1-score recognition performance of 61% and 62% in the high frame rate CASME II and SMIC-HS databases respectively.
['Raphael C. -W. Phan', 'Sze-Teng Liong', 'John See', 'KokSheik Wong']
2016-06-06
null
null
null
null
['micro-expression-recognition']
['computer-vision']
[ 3.64538312e-01 -2.61323571e-01 -2.27996796e-01 -4.64041084e-01 -3.67931873e-01 -6.15492538e-02 4.18727398e-01 -2.76964605e-01 -5.14225185e-01 7.21173584e-01 2.23657023e-03 3.55247498e-01 2.53004599e-02 -3.43911111e-01 -2.24911898e-01 -9.92410958e-01 -2.06684560e-01 -4.17798489e-01 -1.06540270e-01 -5.50248563e-01 1.23861678e-01 6.77000105e-01 -1.87937164e+00 4.28825140e-01 3.31123888e-01 1.59048784e+00 -9.89805385e-02 7.76731312e-01 -1.19306706e-01 1.27373719e+00 -6.18417144e-01 -8.41463745e-01 1.84237376e-01 -5.74251890e-01 -6.96347475e-01 3.57763082e-01 4.77779299e-01 -3.77820820e-01 -3.32271159e-01 1.03596759e+00 5.18486202e-01 8.59215930e-02 4.05618072e-01 -1.51599610e+00 -1.74884811e-01 -2.59672344e-01 -8.92886579e-01 4.32341456e-01 8.74778748e-01 -1.53430432e-01 7.49373913e-01 -7.90099204e-01 1.08766842e+00 1.00373483e+00 3.49842191e-01 4.68758911e-01 -7.35211194e-01 -7.87497044e-01 -2.70829856e-01 4.39794213e-01 -1.62060106e+00 -8.98232818e-01 9.33081448e-01 -3.50950480e-01 8.11788976e-01 3.67841274e-01 9.28017318e-01 9.57532108e-01 2.61160821e-01 7.05715716e-01 1.31771457e+00 -5.99321723e-01 -5.72499298e-02 1.01788893e-01 -1.92886069e-01 8.98553371e-01 -3.81747901e-01 5.85185662e-02 -8.15899014e-01 -1.18543720e-02 7.58305848e-01 -1.79297045e-01 -3.95852536e-01 1.15286037e-01 -7.04250634e-01 4.93424892e-01 -1.28665775e-01 5.01053393e-01 -5.27274966e-01 -6.03412241e-02 6.72557771e-01 6.73279822e-01 3.82576406e-01 -1.68308824e-01 -8.50927159e-02 -8.36070836e-01 -1.09873331e+00 1.91937596e-01 6.33406997e-01 8.24260116e-01 7.13528693e-01 4.69319493e-01 1.88095927e-01 9.31708515e-01 -7.11792931e-02 4.29621071e-01 5.39150655e-01 -1.12544215e+00 1.12101860e-01 3.88693690e-01 1.46009669e-01 -1.48463631e+00 -1.60672873e-01 -4.98995036e-02 -6.92939043e-01 4.01415318e-01 2.93392241e-01 -2.19941258e-01 -4.56496477e-01 1.78585696e+00 3.82620096e-01 2.99547613e-01 1.33074269e-01 9.69414115e-01 6.17407978e-01 7.45753944e-01 1.62193879e-01 -8.70896220e-01 1.30522966e+00 -5.81977606e-01 -1.06028271e+00 3.46663147e-01 5.10707021e-01 -1.02288747e+00 6.66428626e-01 4.96364266e-01 -1.17477548e+00 -6.20229542e-01 -1.05259359e+00 2.13036761e-01 -1.46699008e-02 2.35790104e-01 6.36422515e-01 5.94476223e-01 -9.16565537e-01 3.14205766e-01 -5.73140383e-01 -2.71012425e-01 3.50433558e-01 6.04911804e-01 -1.04380536e+00 9.11033601e-02 -9.63603199e-01 7.77826071e-01 -3.01006027e-02 2.62683332e-01 -4.61377501e-01 -2.57116348e-01 -7.88072884e-01 -1.70610026e-01 1.00231379e-01 -7.44606853e-02 1.01352322e+00 -2.06958342e+00 -1.78087866e+00 1.21664882e+00 -5.47083557e-01 -1.57338113e-01 3.47204387e-01 -1.16658539e-01 -9.73803997e-01 7.78093874e-01 -2.89422333e-01 5.26307344e-01 1.05539894e+00 -9.70563233e-01 -6.76926613e-01 -4.61258203e-01 2.47955590e-01 1.09166354e-01 -1.24563262e-01 6.46203995e-01 -3.15109968e-01 -5.07668495e-01 -4.49588224e-02 -8.80259454e-01 3.60307574e-01 2.17424765e-01 2.48345528e-02 -6.80583864e-02 1.28366351e+00 -5.75438559e-01 1.27091432e+00 -2.41362047e+00 -1.80374011e-02 8.54925662e-02 1.87718533e-02 5.79384446e-01 -1.19975716e-01 1.85104012e-01 -3.45082104e-01 -2.10978240e-01 1.76018015e-01 -9.71049666e-02 -3.39190722e-01 3.18482995e-01 -4.01669666e-02 7.68826544e-01 1.94046497e-01 5.48773885e-01 -7.48115778e-01 -7.53900886e-01 3.03516626e-01 7.40365386e-01 -3.30197066e-01 3.40298861e-01 3.64778429e-01 2.62474805e-01 -4.77731436e-01 1.05151546e+00 7.74040639e-01 2.14400277e-01 -4.76826690e-02 -3.73052448e-01 -1.87562510e-01 -5.37325323e-01 -1.23478973e+00 1.40851116e+00 -2.38514036e-01 9.34283972e-01 3.64039779e-01 -1.11841035e+00 1.03637588e+00 6.77932978e-01 1.06001890e+00 -8.61370444e-01 3.83344024e-01 2.69337386e-01 -3.51701193e-02 -9.77059364e-01 5.05236924e-01 -4.32057083e-01 2.69826353e-01 4.67935635e-04 2.77382791e-01 2.95940965e-01 3.68537366e-01 -5.87616786e-02 7.79561937e-01 1.03797585e-01 6.10169649e-01 -1.18104815e-01 9.09402013e-01 -5.31100631e-01 6.80785894e-01 -2.12974660e-02 -7.55613029e-01 4.19117808e-01 7.87541986e-01 -4.66746032e-01 -8.24848711e-01 -8.59228492e-01 -1.08306818e-01 9.62619722e-01 3.47725837e-03 -5.98062873e-01 -8.24767649e-01 -2.82878160e-01 -3.63290459e-01 6.18767589e-02 -5.59848189e-01 1.62826091e-01 -5.28081656e-01 -5.59185147e-01 7.40312099e-01 4.23606485e-02 7.36249864e-01 -1.11634016e+00 -9.86584485e-01 1.52514383e-01 -2.52612978e-01 -1.38963068e+00 -2.35861927e-01 -2.20417067e-01 -5.59426785e-01 -1.13697934e+00 -7.23999918e-01 -5.98601222e-01 6.20331228e-01 6.07714467e-02 8.89339864e-01 1.32923231e-01 -4.51641351e-01 3.04316282e-01 -5.06479919e-01 1.49314676e-03 -2.09339157e-01 -7.35360503e-01 -4.35919315e-03 6.80766940e-01 4.45284009e-01 -6.63980722e-01 -5.39713740e-01 3.65707219e-01 -7.99319685e-01 -3.52873057e-02 2.98699766e-01 8.11867118e-01 5.73358715e-01 -1.99625611e-01 2.70526081e-01 -5.73464751e-01 3.16073030e-01 -2.95111418e-01 -2.94542164e-01 -9.50136688e-03 -1.18497290e-01 -5.18646836e-01 6.88414693e-01 -3.42078149e-01 -1.11916065e+00 1.19217522e-01 -4.85319555e-01 -6.65471911e-01 -1.40351668e-01 2.91595846e-01 -1.08231202e-01 -3.01883638e-01 3.78495008e-01 2.80052185e-01 3.62224996e-01 6.01534918e-02 3.24689746e-02 8.36952806e-01 7.06753910e-01 -4.87063020e-01 2.42316023e-01 6.22781277e-01 1.76645324e-01 -1.65707338e+00 -2.95198083e-01 -6.32531583e-01 -5.87741911e-01 -7.14582741e-01 8.63256574e-01 -8.67777944e-01 -1.00492585e+00 7.09480882e-01 -9.88424897e-01 2.41050750e-01 -7.93687552e-02 3.29306662e-01 -7.36443460e-01 5.91034591e-01 -5.12440920e-01 -1.12626243e+00 -2.79244542e-01 -1.32432544e+00 1.03561914e+00 3.46408993e-01 -3.09910804e-01 -7.89562643e-01 -1.35317326e-01 3.15903097e-01 3.05754364e-01 7.93089926e-01 4.83892441e-01 5.01089878e-02 -9.32127312e-02 -4.12408739e-01 -1.25763968e-01 4.91806209e-01 1.80016473e-01 5.37031770e-01 -1.09946632e+00 -8.44846480e-03 1.59967523e-02 -5.48373699e-01 3.20903838e-01 3.63420606e-01 1.05372190e+00 -1.99734762e-01 1.15206435e-01 8.44515204e-01 1.45345426e+00 6.33626461e-01 1.04636705e+00 1.26104966e-01 2.15450540e-01 5.52776754e-01 7.20434010e-01 8.26071382e-01 9.54785571e-02 1.12213206e+00 3.09038967e-01 -1.58589035e-01 -2.79061068e-02 1.20118417e-01 5.76186419e-01 7.87804306e-01 -7.20015228e-01 -8.13825577e-02 -3.33951533e-01 2.68827081e-01 -1.52077782e+00 -1.47157311e+00 3.07678916e-02 1.98928869e+00 5.46788275e-01 -3.02884102e-01 2.29810327e-01 5.62578142e-01 4.42244738e-01 4.63639587e-01 -6.48392737e-02 -8.28837276e-01 -4.14447397e-01 4.04748350e-01 -3.76631543e-02 4.10128564e-01 -1.00260019e+00 6.91080153e-01 6.18050814e+00 8.83356750e-01 -1.67333388e+00 4.90162848e-03 7.59494245e-01 -2.25861609e-01 2.86471158e-01 -2.47638062e-01 -4.90169704e-01 4.82928157e-01 1.09388745e+00 -6.83072582e-02 1.21700205e-01 9.82022166e-01 5.11755764e-01 -4.01030272e-01 -7.17326939e-01 1.52638566e+00 3.41873556e-01 -1.19191730e+00 -1.06264338e-01 -1.19026132e-01 3.99617761e-01 -5.91901302e-01 -3.76856960e-02 2.45093610e-02 -5.13556063e-01 -1.02893138e+00 5.24698436e-01 4.04207677e-01 1.15036237e+00 -9.38581765e-01 7.82600045e-01 -3.65229473e-02 -1.25069392e+00 6.58661202e-02 -2.57226288e-01 -7.93141797e-02 2.03060597e-01 1.97621644e-01 -2.55665839e-01 5.48523724e-01 6.10046506e-01 8.01608860e-01 -1.32692963e-01 4.42303419e-01 1.40115127e-01 4.73235160e-01 -1.94604725e-01 7.12504983e-02 2.73402631e-01 -3.28569651e-01 4.61004168e-01 1.51164126e+00 4.26343530e-01 5.27466118e-01 -2.05608517e-01 1.73820898e-01 2.10128620e-01 4.28885579e-01 -5.45120835e-01 -1.45621568e-01 -1.07873894e-01 1.36063576e+00 -3.62654656e-01 -3.16904753e-01 -6.01633251e-01 1.22612286e+00 4.95103858e-02 2.75711894e-01 -7.60419607e-01 -3.03153306e-01 9.98033762e-01 1.17931776e-01 5.22621386e-02 -1.44292593e-01 3.51656646e-01 -1.15979087e+00 1.02429828e-02 -1.08323634e+00 3.31277430e-01 -9.36596394e-01 -7.19659150e-01 8.39792609e-01 -6.81126267e-02 -1.28373528e+00 -6.06473029e-01 -6.91800416e-01 -3.85090172e-01 5.67609966e-01 -1.54044235e+00 -1.04192185e+00 -5.61226726e-01 1.05200744e+00 4.28106904e-01 -3.79961431e-01 1.04825878e+00 6.83098614e-01 -4.78974998e-01 8.00319493e-01 -1.65402859e-01 1.93730637e-01 4.67532337e-01 -6.82359517e-01 -6.63741529e-01 6.39556527e-01 2.21804187e-01 3.70175481e-01 8.19395542e-01 -3.65257934e-02 -1.64309895e+00 -4.69450027e-01 8.91235709e-01 1.46227598e-01 4.46269095e-01 9.10153333e-03 -5.00415683e-01 2.97944337e-01 1.70373961e-01 4.22477454e-01 8.95196259e-01 -4.25879151e-01 -1.36057615e-01 -4.72649187e-01 -1.20906639e+00 4.54482764e-01 8.29793632e-01 -6.23227417e-01 -2.49136090e-01 -1.16568200e-01 -1.34331688e-01 -3.76745254e-01 -1.04293203e+00 3.37293386e-01 1.12309229e+00 -1.57280767e+00 8.72216403e-01 -3.60334486e-01 6.60265207e-01 -1.75613776e-01 -5.49902499e-01 -8.33703578e-01 2.18169391e-01 -8.92858267e-01 7.61116296e-03 1.00837064e+00 -2.62048453e-01 -2.06012979e-01 9.53086793e-01 4.38624710e-01 2.34761596e-01 -1.00608277e+00 -1.35939407e+00 -4.89782453e-01 -5.56685746e-01 -5.32260656e-01 1.90442547e-01 8.82102191e-01 2.48074144e-01 -1.57786570e-02 -6.71550512e-01 -3.48495662e-01 3.63185436e-01 -1.33119081e-03 9.38260615e-01 -7.39664376e-01 -1.13083415e-01 -3.36506933e-01 -1.15339983e+00 -7.39142001e-01 3.70234936e-01 -3.94690067e-01 -3.92447621e-01 -7.56855011e-01 7.97582194e-02 4.83291373e-02 -2.45485932e-01 1.50771722e-01 2.47870162e-01 5.87337554e-01 3.74905884e-01 1.43212691e-01 -3.61212999e-01 4.45174545e-01 1.26237381e+00 1.05674341e-01 5.53230271e-02 -1.61109984e-01 -2.73187220e-01 8.24093044e-01 2.95368046e-01 1.04669984e-02 -4.64126378e-01 1.02106147e-01 -1.75730214e-01 6.16323650e-01 5.30138575e-02 -8.89034092e-01 -8.36661756e-02 -2.04570442e-01 4.43727434e-01 -1.94163322e-01 9.28224206e-01 -8.98001015e-01 3.70910436e-01 2.69865781e-01 -8.78598243e-02 3.57763648e-01 5.54097034e-02 1.70790464e-01 -8.52561712e-01 -1.41936049e-01 1.13781309e+00 -5.00506237e-02 -1.26660085e+00 3.03727627e-01 -4.59924698e-01 -1.28274933e-01 1.19002604e+00 -7.79646873e-01 -7.48585016e-02 -6.63132668e-01 -6.71192467e-01 -5.23575664e-01 3.04043084e-01 2.90785015e-01 8.17518115e-01 -1.28055060e+00 -5.48025250e-01 4.00294900e-01 7.98963010e-02 -6.91336751e-01 3.96698684e-01 1.08520555e+00 -7.46007621e-01 1.00874722e-01 -8.03580344e-01 -5.18350303e-01 -1.76784325e+00 -1.24266259e-02 4.07003105e-01 -4.41386923e-02 -4.68668699e-01 6.97927535e-01 7.99908787e-02 5.29468536e-01 -5.19040190e-02 5.16877398e-02 -3.67873251e-01 1.58067480e-01 7.72767067e-01 3.62127364e-01 -1.65027171e-01 -1.61312211e+00 -2.78750807e-01 7.00819254e-01 2.29434177e-01 -6.88554049e-02 1.18861485e+00 -1.94805652e-01 -2.53468603e-01 2.91496098e-01 1.62023795e+00 1.13444701e-01 -1.04710793e+00 2.39499256e-01 -1.72488451e-01 -9.27793324e-01 2.75430921e-02 -4.78354335e-01 -1.19469726e+00 8.23740482e-01 8.16826403e-01 -1.51520655e-01 1.65717065e+00 -5.10204136e-01 6.74204469e-01 2.54638563e-03 3.70882869e-01 -1.11005962e+00 1.93594247e-01 2.06229016e-01 8.81282568e-01 -1.13827395e+00 -5.97681999e-02 -4.51023966e-01 -8.82302523e-01 1.38851237e+00 4.89969403e-01 -1.11495629e-01 6.91697299e-01 5.15113711e-01 2.60662854e-01 -1.84545666e-01 -6.27343595e-01 -8.00872035e-03 5.39262556e-02 4.99660999e-01 7.67246723e-01 -1.41681254e-01 -3.06597680e-01 1.74231097e-01 -2.07017034e-01 4.25749272e-01 4.88617063e-01 9.59888220e-01 -3.98898363e-01 -8.01524580e-01 -3.25234443e-01 2.34333083e-01 -1.03587723e+00 3.81291062e-01 -1.13839850e-01 1.09537661e+00 1.44140825e-01 9.65159416e-01 3.99017595e-02 -4.70707655e-01 3.07806104e-01 1.98099867e-01 6.43272161e-01 6.64099306e-02 -3.46481383e-01 2.01771930e-01 3.16227317e-01 -8.76390636e-01 -1.06929147e+00 -6.79992020e-01 -1.18425584e+00 -6.38548017e-01 -5.74214831e-02 1.53217956e-01 6.31443202e-01 7.76150823e-01 1.16861857e-01 -1.88237093e-02 6.61358654e-01 -7.47777522e-01 -1.77026652e-02 -7.52781093e-01 -9.35448170e-01 8.41481328e-01 4.44603235e-01 -8.29246104e-01 -2.74562031e-01 5.02798378e-01]
[13.653576850891113, 1.8200126886367798]
425fa612-2b2b-46f4-9f01-d8d06f5ffd3b
deep-learning-with-convolutional-neural
1703.05051
null
http://arxiv.org/abs/1703.05051v5
http://arxiv.org/pdf/1703.05051v5.pdf
Deep learning with convolutional neural networks for EEG decoding and visualization
PLEASE READ AND CITE THE REVISED VERSION at Human Brain Mapping: http://onlinelibrary.wiley.com/doi/10.1002/hbm.23730/full Code available here: https://github.com/robintibor/braindecode
['Jost Tobias Springenberg', 'Michael Tangermann', 'Robin Tibor Schirrmeister', 'Lukas Dominique Josef Fiederer', 'Frank Hutter', 'Wolfram Burgard', 'Tonio Ball', 'Martin Glasstetter', 'Katharina Eggensperger']
2017-03-15
null
null
null
null
['eeg-decoding', 'eeg-decoding']
['medical', 'time-series']
[-7.51829803e-01 1.99708879e-01 -4.72662300e-01 -2.20674157e-01 -8.00899863e-01 -2.94661760e-01 3.96612525e-01 4.92117912e-01 -4.56484944e-01 1.02347851e+00 5.13282776e-01 -2.36375257e-01 2.54575282e-01 -5.28913379e-01 -6.63588881e-01 -4.78447497e-01 -5.30666746e-02 6.02029562e-01 2.14636430e-01 1.17580965e-01 1.76874295e-01 2.59418309e-01 -9.47914720e-01 -8.26628283e-02 9.22493458e-01 8.87209594e-01 8.48678112e-01 1.21217363e-01 3.58495831e-01 7.08393991e-01 4.81667668e-02 -5.50467789e-01 2.90353417e-01 -4.54069018e-01 -7.72384524e-01 -6.43144488e-01 -3.87342721e-01 -5.98050475e-01 -9.28382218e-01 1.21900797e+00 7.47854948e-01 -8.33006799e-02 4.50841039e-01 -1.10903180e+00 -7.21562445e-01 7.50784338e-01 -4.02207464e-01 1.17362797e+00 6.14040494e-01 3.27971101e-01 4.76804376e-01 -1.60614169e+00 5.22761166e-01 6.14524901e-01 -1.89776383e-02 7.55939841e-01 -6.85199618e-01 -1.18454146e+00 -3.17234874e-01 6.42065108e-01 -1.59922564e+00 -1.38714266e+00 4.44923252e-01 -3.71927410e-01 7.86416233e-01 -2.17710182e-01 7.69709527e-01 1.11043704e+00 5.49989462e-01 8.17840874e-01 1.15560544e+00 -2.23325893e-01 -8.96280184e-02 -2.74877816e-01 9.41260532e-02 7.86843836e-01 5.64465106e-01 3.17321748e-01 -2.33840019e-01 3.22970361e-01 9.38917816e-01 1.54918209e-01 -4.10761952e-01 -2.84275785e-02 -1.07137835e+00 7.66099751e-01 6.27646208e-01 5.77493072e-01 -7.18902409e-01 5.13371117e-02 3.56196374e-01 3.44670832e-01 3.30045193e-01 -1.13941126e-01 3.14548641e-01 -2.52757907e-01 -4.44347829e-01 1.28939837e-01 6.29024565e-01 7.36833870e-01 6.24276459e-01 -3.75126123e-01 5.81997156e-01 1.32260752e+00 1.05645239e-01 7.56139457e-02 4.27843064e-01 -7.24366009e-01 1.26145080e-01 -4.38251384e-02 -1.93307400e-01 -5.34223020e-01 -5.77032804e-01 -2.93140173e-01 -6.07936323e-01 -4.05971229e-01 4.23915535e-01 -1.48428395e-01 -2.96352386e-01 1.28205764e+00 2.71531224e-01 -4.67064857e-01 -4.46292311e-01 1.01358974e+00 1.30229115e+00 3.06947142e-01 -2.02766806e-01 -9.90259275e-02 1.20438063e+00 -9.65928376e-01 -7.22548425e-01 -1.28075257e-01 2.74294049e-01 -9.38297749e-01 5.71098745e-01 1.31141871e-01 -1.46898866e+00 -8.93277824e-02 -9.10890341e-01 -1.51815891e-01 -7.24869251e-01 -2.34901249e-01 7.25660324e-01 4.69213203e-02 -1.09913075e+00 1.46142110e-01 -1.18946195e+00 -4.05592322e-01 1.00388360e+00 6.72439039e-02 -2.18631074e-01 -4.22946632e-01 -1.36213017e+00 1.26855779e+00 3.10488969e-01 4.29699838e-01 -8.60818446e-01 -5.08858204e-01 -5.73905706e-01 -9.07418907e-01 6.42010272e-01 -1.14111088e-01 1.40672576e+00 -3.74114126e-01 -9.40439761e-01 1.01100969e+00 -3.35791767e-01 2.11134374e-01 3.06240529e-01 -2.90701419e-01 -2.48131931e-01 5.36281109e-01 3.22261542e-01 4.70976949e-01 -1.71921104e-02 -5.85914373e-01 -5.00942588e-01 -4.60134566e-01 -6.10915981e-02 -1.60826609e-01 -9.88401249e-02 3.66023481e-01 -2.42620438e-01 -5.65604448e-01 -1.07697949e-01 -5.90611696e-01 2.92893529e-01 3.12465310e-01 -3.57364953e-01 1.89678222e-01 -9.49350223e-02 -1.10639358e+00 9.92779613e-01 -1.98878777e+00 -3.36352944e-01 -7.32315183e-02 6.14685081e-02 -1.73729762e-01 9.71983895e-02 5.95803976e-01 -3.32825452e-01 2.21289620e-01 -6.17485881e-01 2.20963433e-02 -8.56320634e-02 -5.32186270e-01 8.29032302e-01 1.26962602e+00 -2.02208623e-01 1.14168763e+00 -1.16094851e+00 -2.68183142e-01 2.21216023e-01 6.50097489e-01 3.18844944e-01 -9.24661756e-02 8.30931723e-01 7.33857214e-01 -4.64210182e-01 8.62488627e-01 6.11179471e-01 -4.81134169e-02 -1.01170018e-01 1.63686112e-01 -4.02676970e-01 4.52549309e-01 -1.14842916e+00 1.51851153e+00 9.57166106e-02 5.20295560e-01 3.67336005e-01 -4.37958390e-01 2.04948217e-01 7.25059986e-01 5.17696023e-01 -1.21002269e+00 4.52679902e-01 4.41852301e-01 7.54465938e-01 3.88865955e-02 -1.57873183e-01 -2.73159146e-01 3.09807807e-01 2.56393343e-01 2.66413569e-01 -1.56851292e-01 3.78368884e-01 2.61482239e-01 4.83078301e-01 1.28491566e-01 8.35998535e-01 -1.70144513e-01 -7.82157201e-03 -4.57213260e-03 1.54216558e-01 2.87757576e-01 -2.93275177e-01 2.54069448e-01 2.01627985e-01 2.67619759e-01 -7.65469193e-01 -1.06305015e+00 -1.08151770e+00 2.30431065e-01 -1.54324463e-02 -1.21454634e-01 -6.19491339e-01 2.53793836e-01 -3.82275544e-02 1.15924668e+00 -6.37643576e-01 9.21578556e-02 4.01823521e-02 -2.30243087e-01 5.33477724e-01 3.91584277e-01 7.97146037e-02 -1.31829107e+00 -8.95392179e-01 -1.91192284e-01 -1.53915614e-01 -3.94514948e-01 -6.63250566e-01 5.03564179e-01 -6.81668699e-01 -7.69851983e-01 -1.05529118e+00 -3.43596756e-01 9.91043389e-01 -1.42653314e-02 4.98144507e-01 5.03579438e-01 -9.14995730e-01 4.13411736e-01 -5.97427487e-01 -8.15994665e-02 -2.73757428e-01 -1.84554711e-01 6.61264285e-02 -7.43804872e-01 5.10848641e-01 -5.51694095e-01 -1.08446038e+00 -3.63736972e-02 -6.37688994e-01 3.37431192e-01 7.87705481e-01 8.28116596e-01 6.36745214e-01 -3.42656881e-01 3.33741039e-01 -4.38404292e-01 3.74701530e-01 -5.81225038e-01 -9.71759737e-01 -5.94986975e-01 -1.16268110e+00 -6.54929340e-01 3.24357629e-01 -6.26205057e-02 -9.09970343e-01 -7.16308951e-01 -2.39518672e-01 1.62833892e-02 -8.38556111e-01 5.66512823e-01 8.13297629e-02 3.70843142e-01 2.95737624e-01 2.00674593e-01 2.41862208e-01 -5.44210017e-01 -6.86641112e-02 3.05813879e-01 2.66537577e-01 -2.83079743e-01 2.28387445e-01 2.85547704e-01 -5.80020785e-01 -3.86132956e-01 -1.47563040e-01 -6.92529008e-02 -1.12411702e+00 -5.88988662e-01 5.83503783e-01 -8.93089712e-01 -8.51353034e-02 3.25906307e-01 -3.38559330e-01 -8.37830186e-01 -2.61618108e-01 7.63285160e-01 -4.11013216e-01 1.21653877e-01 -3.58856708e-01 -1.40430361e-01 -3.43717933e-01 -1.00381458e+00 1.79873351e-02 2.72056580e-01 -6.87001467e-01 -9.02595699e-01 -2.73587734e-01 3.45868707e-01 4.85404044e-01 4.27921295e-01 1.84176788e-01 -6.72200501e-01 -3.22412550e-01 -6.14473045e-01 2.18485184e-02 1.45076320e-01 2.22036153e-01 -2.45469045e-02 -5.32433987e-01 -2.32004419e-01 2.01893432e-04 -9.21935886e-02 5.59586585e-01 7.00103998e-01 6.79362714e-01 -3.78721237e-01 -1.61372989e-01 3.30754787e-01 1.51754010e+00 6.05809152e-01 3.26718211e-01 4.41619545e-01 2.22438395e-01 4.21198189e-01 2.45686069e-01 5.44736803e-01 1.12467863e-01 1.02439725e+00 2.85923541e-01 4.95787114e-01 -2.21349299e-01 -5.52880645e-01 -2.24146903e-01 7.90484786e-01 -2.43828312e-01 3.97723228e-01 -1.94757593e+00 9.19563830e-01 -1.66100931e+00 -9.23413455e-01 -1.03908353e-01 2.14700365e+00 8.93708110e-01 7.30131716e-02 6.41722798e-01 -3.40135425e-01 5.07934988e-01 1.44514903e-01 -5.52264392e-01 -2.74046183e-01 2.67783821e-01 3.09623331e-01 7.46550381e-01 1.29330504e+00 -6.64756060e-01 8.07172120e-01 3.07191610e+00 6.73458815e-01 -1.34053731e+00 7.89104581e-01 2.58950233e-01 -9.64542806e-01 -1.34435907e-01 3.91013920e-02 -2.90787488e-01 3.10054809e-01 1.34649515e+00 -1.28347576e+00 1.15678430e+00 4.87675011e-01 5.65725327e-01 -3.07280153e-01 -2.65196949e-01 7.27170467e-01 -3.23444992e-01 -8.90730381e-01 -7.56972253e-01 6.30999207e-02 -7.43758157e-02 6.27964795e-01 -1.87684968e-01 -3.38130891e-01 -5.09803593e-01 -7.71262348e-01 9.29800391e-01 4.34168160e-01 9.91537035e-01 -4.33199048e-01 4.87279028e-01 -8.96552503e-02 -8.55165601e-01 -1.21494405e-01 -2.93839127e-01 -1.89966276e-01 3.40904415e-01 4.20382768e-01 -5.27283847e-01 2.43775651e-01 5.90829432e-01 8.47748637e-01 -4.97155607e-01 1.98578644e+00 -5.65748036e-01 4.29922849e-01 -1.79255188e-01 -8.32717791e-02 -1.73087150e-01 2.29584742e-02 4.73015338e-01 8.14828038e-01 3.18754405e-01 6.86284542e-01 -1.76975131e-01 9.55842078e-01 7.26226047e-02 4.97923940e-01 -6.35446727e-01 -2.14639410e-01 1.05509579e+00 1.32741034e+00 -1.22483432e+00 -1.04355589e-01 -8.32877040e-01 7.14259326e-01 2.56957173e-01 2.64951646e-01 -8.98267090e-01 -2.95817345e-01 -1.37921125e-01 5.75682461e-01 2.07705870e-01 2.08585411e-01 -4.77783650e-01 -1.13154852e+00 1.66819289e-01 -5.91838896e-01 5.21588862e-01 -8.60007167e-01 -1.78514361e-01 6.22704744e-01 7.82795787e-01 -1.50858140e+00 -8.79259557e-02 -6.33344173e-01 -1.76568225e-01 1.15762353e+00 -7.72585690e-01 -6.90616786e-01 4.62333821e-02 2.78232187e-01 5.56662679e-01 2.15428993e-01 7.53453910e-01 3.07577431e-01 -4.00098562e-01 3.54205579e-01 3.46919894e-01 -1.29767224e-01 3.25376064e-01 -9.11672294e-01 2.55199045e-01 6.52326286e-01 -3.43120009e-01 8.45420659e-01 4.55771834e-01 -6.49183750e-01 -1.13972247e+00 -6.78722262e-01 7.40921378e-01 -2.20944420e-01 8.32592249e-01 -4.65440243e-01 -9.10789371e-01 1.03475606e+00 6.43916547e-01 -4.95659888e-01 8.01052690e-01 -2.31523603e-01 2.25732714e-01 1.25378743e-01 -1.15093827e+00 7.02229798e-01 6.74560547e-01 -4.01969135e-01 -4.85599130e-01 7.48016417e-01 -2.56786406e-01 -8.07797670e-01 -1.51657021e+00 -1.87106296e-01 9.47306573e-01 -7.34887004e-01 3.63867849e-01 3.56699973e-01 1.04412150e+00 3.03187147e-02 -1.66350305e-01 -8.99790883e-01 -7.36343384e-01 -9.30791572e-02 3.20213027e-02 6.52665496e-01 7.43065655e-01 -1.85938895e+00 1.28912449e-01 3.76879871e-01 -2.85997063e-01 -1.38627601e+00 -1.14601552e+00 -6.61174595e-01 3.23489577e-01 -3.13881725e-01 1.39386266e-01 8.95531535e-01 8.75983775e-01 -3.51278812e-01 1.39529884e-01 -1.83250159e-01 5.87347686e-01 -2.62465417e-01 -4.29573089e-01 -4.86543894e-01 -1.88970700e-01 -9.77079213e-01 -1.29721850e-01 -1.58409886e-02 -4.07180846e-01 -1.26151419e+00 -4.09997664e-02 -1.91467416e+00 3.75041217e-01 3.85771573e-01 -1.67937368e-01 9.04343963e-01 2.29974911e-01 1.84998915e-01 6.02786362e-01 5.28830290e-01 -4.65149820e-01 5.43143928e-01 1.11466467e+00 4.87356126e-01 2.36338019e-01 -2.19202071e-01 -1.04931152e+00 2.26973951e-01 2.00755000e+00 -5.75131297e-01 -1.59300432e-01 -5.32125294e-01 -4.60285485e-01 4.94087428e-01 3.87117028e-01 -8.37396681e-01 1.87841967e-01 -3.17860931e-01 4.84892786e-01 -4.97935452e-02 6.43704176e-01 -2.91584790e-01 5.15303493e-01 9.72721994e-01 -2.70122081e-01 7.65540481e-01 8.32612634e-01 -4.61175054e-01 1.78364471e-01 -3.75486732e-01 1.25251138e+00 -4.90610808e-01 -9.48930740e-01 2.76224941e-01 -4.98631865e-01 1.07207194e-01 5.45853734e-01 -5.04404843e-01 -6.47950053e-01 -4.70583409e-01 -6.42571747e-01 3.63602668e-01 5.41835189e-01 4.03233975e-01 9.31873202e-01 -9.67674971e-01 -6.25688851e-01 -3.97385657e-01 1.06748320e-01 -4.25480694e-01 6.44455254e-01 2.00332236e+00 -4.05222058e-01 1.06403494e+00 -6.52434409e-01 3.49566728e-01 -7.70514071e-01 3.90546292e-01 5.02703905e-01 9.88175988e-01 -7.41687179e-01 7.11139679e-01 4.46777977e-02 9.69724432e-02 1.03222661e-01 2.21645012e-01 2.26271123e-01 1.99002966e-01 4.77998495e-01 3.96452367e-01 -3.43154848e-01 -1.23306763e+00 -8.00159693e-01 -3.39088947e-01 -3.31270128e-01 -1.15176231e-01 1.62822604e+00 -2.80549079e-01 -6.13334537e-01 8.16631317e-01 1.00802088e+00 -1.93432674e-01 -4.12622809e-01 -8.57993662e-02 -9.00105163e-02 -5.44901729e-01 -4.47544549e-03 -1.04737401e+00 -1.00940442e+00 5.82484841e-01 9.24470842e-01 -4.89829749e-01 1.11921108e+00 3.82005394e-01 4.27620739e-01 -4.57843870e-01 1.05993569e-01 -9.06038702e-01 -4.98056442e-01 2.38163456e-01 1.38210917e+00 -9.69569087e-01 5.71510315e-01 -8.46832618e-02 -8.89011145e-01 9.91373539e-01 3.03432822e-01 -6.72600344e-02 1.23139393e+00 1.76247627e-01 5.47866821e-02 -2.63509721e-01 -6.44790053e-01 -2.54486889e-01 -3.22283655e-01 3.50523859e-01 8.94366503e-01 5.16925395e-01 -9.58935618e-01 7.74601042e-01 -3.04505616e-01 4.21399534e-01 5.16071200e-01 1.10065150e+00 2.15761602e-01 -7.32389629e-01 -1.86241120e-01 9.19384718e-01 -4.32076305e-01 -5.07369757e-01 -2.44945735e-01 7.80700862e-01 -3.98886472e-01 7.19580352e-01 -1.19644091e-01 6.26476556e-02 1.62085280e-01 5.21619856e-01 4.01920706e-01 -5.93052208e-01 -1.26588056e-02 1.77023858e-01 2.89318442e-01 -6.62094235e-01 -1.52994946e-01 -1.21444964e+00 -1.19726729e+00 -7.33553350e-01 1.89402610e-01 2.43001282e-01 6.80562735e-01 2.91705459e-01 2.07144365e-01 3.52827072e-01 -8.29866603e-02 -9.94416952e-01 8.48314073e-03 -9.60493982e-01 -7.44088769e-01 2.23284378e-03 3.67892921e-01 -1.12910879e+00 -2.71885961e-01 -1.50797710e-01]
[14.180715560913086, -2.182055711746216]
5909854d-d8e9-4c76-9801-e2ddaed44f90
parameter-efficient-transfer-learning-of-pre
2210.16032
null
https://arxiv.org/abs/2210.16032v1
https://arxiv.org/pdf/2210.16032v1.pdf
Parameter-efficient transfer learning of pre-trained Transformer models for speaker verification using adapters
Recently, the pre-trained Transformer models have received a rising interest in the field of speech processing thanks to their great success in various downstream tasks. However, most fine-tuning approaches update all the parameters of the pre-trained model, which becomes prohibitive as the model size grows and sometimes results in overfitting on small datasets. In this paper, we conduct a comprehensive analysis of applying parameter-efficient transfer learning (PETL) methods to reduce the required learnable parameters for adapting to speaker verification tasks. Specifically, during the fine-tuning process, the pre-trained models are frozen, and only lightweight modules inserted in each Transformer block are trainable (a method known as adapters). Moreover, to boost the performance in a cross-language low-resource scenario, the Transformer model is further tuned on a large intermediate dataset before directly fine-tuning it on a small dataset. With updating fewer than 4% of parameters, (our proposed) PETL-based methods achieve comparable performances with full fine-tuning methods (Vox1-O: 0.55%, Vox1-E: 0.82%, Vox1-H:1.73%).
['Jan Černocký', 'Lukáš Burget', 'Ladislav Mošner', 'Oldřich Plchot', 'Rongzhi Gu', 'Themos Stafylakis', 'Junyi Peng']
2022-10-28
null
null
null
null
['speaker-verification']
['speech']
[ 1.36633426e-01 1.54685929e-01 -4.14274558e-02 -5.38570344e-01 -1.15159833e+00 -7.67376840e-01 3.57342780e-01 -1.34156853e-01 -5.25896609e-01 6.31707788e-01 8.83677378e-02 -5.76500893e-01 1.33013561e-01 -4.18617398e-01 -7.65941978e-01 -5.53083956e-01 3.01698774e-01 4.87383991e-01 2.74232239e-01 -1.22486748e-01 -1.64450899e-01 1.97101533e-01 -1.18690360e+00 2.34293476e-01 9.81463253e-01 8.70485425e-01 2.23076478e-01 4.27118003e-01 -2.76615527e-02 3.09134930e-01 -5.25057256e-01 -8.30482185e-01 1.39073521e-01 -2.50019461e-01 -8.75034094e-01 -1.28896341e-01 2.88735658e-01 -2.72649676e-01 -1.61102876e-01 9.17890131e-01 6.68169081e-01 2.15942100e-01 3.13162655e-01 -1.08098423e+00 -4.21180576e-01 1.10553336e+00 -2.06888124e-01 1.03957072e-01 -1.23222321e-01 2.50652403e-01 8.45195293e-01 -9.95666981e-01 2.12819397e-01 1.28094244e+00 7.56185174e-01 7.09035516e-01 -1.13702214e+00 -8.88636112e-01 3.47182363e-01 1.60391837e-01 -1.53087771e+00 -1.14630973e+00 5.39344251e-01 -2.44596332e-01 1.25666702e+00 1.22467682e-01 2.24045396e-01 1.09332454e+00 -2.49526769e-01 5.59692621e-01 8.12685311e-01 -3.21048081e-01 1.72489300e-01 4.22501266e-01 -2.37304404e-01 5.51233113e-01 -2.03827545e-01 6.70483708e-02 -6.99824870e-01 -4.08611633e-02 5.05971432e-01 -3.21729630e-01 -2.12263912e-01 -9.34126079e-02 -1.20678198e+00 7.35873997e-01 3.19369614e-01 2.81609803e-01 -5.68151362e-02 -1.79185092e-01 6.03559315e-01 5.54271519e-01 5.26961505e-01 1.45165145e-01 -9.69593763e-01 -4.42730844e-01 -1.02653098e+00 -2.37305075e-01 6.54668212e-01 1.12128162e+00 8.62886608e-01 1.18353866e-01 -3.23385745e-01 1.36345959e+00 2.17771992e-01 3.08270693e-01 7.66816616e-01 -5.05019546e-01 1.01323211e+00 5.45176148e-01 -2.69542813e-01 -2.43377730e-01 -2.58132577e-01 -7.06168652e-01 -1.00863910e+00 -4.93787169e-01 3.11278641e-01 -2.07787886e-01 -8.97202611e-01 1.95074010e+00 4.71463442e-01 2.48455033e-01 -4.82138470e-02 4.85386133e-01 7.33954847e-01 7.42182374e-01 2.59013683e-01 -2.37383261e-01 1.27437246e+00 -1.37305570e+00 -4.47388053e-01 -5.10847926e-01 8.09976995e-01 -9.36288357e-01 1.44422543e+00 1.45790190e-01 -1.08474362e+00 -7.37736821e-01 -8.68007243e-01 -4.35653236e-03 -2.19809070e-01 3.75512511e-01 2.54750311e-01 8.53027821e-01 -1.15764534e+00 4.22933787e-01 -7.32053638e-01 -1.93489075e-01 4.93299276e-01 5.67170918e-01 -4.16932762e-01 -3.40670273e-02 -1.17102385e+00 7.74784684e-01 4.24230963e-01 2.28973106e-01 -1.10544622e+00 -1.12665772e+00 -7.01176107e-01 2.73755491e-01 2.70662040e-01 -6.96505249e-01 1.55343890e+00 -7.02195227e-01 -2.15408707e+00 7.96687186e-01 -2.14114338e-01 -3.11541170e-01 5.47285199e-01 -1.05730355e-01 -3.45132291e-01 -3.16216856e-01 -2.32737049e-01 6.28975511e-01 1.23639429e+00 -7.56984949e-01 -4.95638609e-01 -1.20937280e-01 2.57360220e-01 1.78013761e-02 -6.12559021e-01 5.74220568e-02 -6.60560310e-01 -7.33685553e-01 -9.37905014e-02 -1.00727856e+00 -2.96583232e-02 -4.07178223e-01 -3.41388345e-01 -3.35603744e-01 4.91154462e-01 -7.45809615e-01 1.43393886e+00 -2.40872049e+00 2.21842170e-01 -4.51367870e-02 -4.66684662e-02 8.32034767e-01 -3.78673732e-01 3.75405639e-01 5.68720326e-02 1.28304839e-01 -5.46081543e-01 -6.91762388e-01 2.23682702e-01 9.99881849e-02 -2.40280345e-01 1.91225439e-01 2.38162324e-01 8.27963948e-01 -6.02833033e-01 -4.39007550e-01 2.15667024e-01 5.37652254e-01 -9.13364053e-01 4.93762642e-01 4.93449494e-02 6.22107863e-01 -3.06984395e-01 3.81779522e-01 8.22409034e-01 9.71233123e-04 1.95806801e-01 -3.18627715e-01 3.19709145e-02 8.20945203e-01 -1.05363476e+00 1.74144077e+00 -1.03681469e+00 2.55219072e-01 2.18868420e-01 -9.73089933e-01 8.64767969e-01 6.22975767e-01 6.55510873e-02 -6.72994792e-01 5.30338325e-02 2.28821024e-01 5.49938791e-02 -2.23751515e-01 1.27969384e-01 -3.00346822e-01 -1.05967514e-01 5.58271408e-01 3.34820986e-01 -1.13120928e-01 2.13522501e-02 -8.68362933e-02 9.01885152e-01 1.35228818e-03 2.99153388e-01 -3.63240391e-02 1.01454461e+00 -5.18866003e-01 6.91120803e-01 4.56640750e-01 -3.17366123e-01 3.37101460e-01 1.00400843e-01 -9.56454799e-02 -9.47521925e-01 -9.07873571e-01 -1.62819818e-01 1.49365258e+00 -4.37244296e-01 -5.72978735e-01 -1.31748044e+00 -6.12503827e-01 -1.51811078e-01 5.05332708e-01 -2.86405414e-01 -5.02116799e-01 -7.94097304e-01 -6.85732782e-01 8.98400962e-01 4.65995431e-01 6.76642179e-01 -1.00331259e+00 -6.31708950e-02 3.32402527e-01 -3.49020898e-01 -1.16935873e+00 -8.39199185e-01 1.47074729e-01 -1.00552368e+00 -5.38912177e-01 -5.55472732e-01 -8.25582922e-01 5.93016863e-01 5.55875674e-02 1.02570629e+00 5.04868366e-02 2.07223713e-01 -1.13028325e-01 -2.88757205e-01 -7.75326565e-02 -6.33057356e-01 7.24262297e-01 1.94735870e-01 2.65064895e-01 2.81521291e-01 -4.93516713e-01 -2.77695090e-01 5.18352568e-01 -5.27550459e-01 1.34935658e-02 6.00354493e-01 9.19429719e-01 4.26638812e-01 -2.49055594e-01 7.69354403e-01 -8.69330466e-01 3.97895426e-01 -2.09039882e-01 -7.42518187e-01 4.04639363e-01 -5.61949134e-01 2.01570392e-01 8.06602418e-01 -6.92163527e-01 -1.26153827e+00 -7.47545213e-02 -4.28101152e-01 -4.72655594e-01 5.95743721e-03 3.67326766e-01 -5.30372560e-01 -1.76101422e-03 3.96434635e-01 1.28177181e-01 -3.34065437e-01 -8.93391609e-01 3.28260481e-01 9.19300854e-01 2.43757471e-01 -4.68610227e-01 1.11722362e+00 -7.19335526e-02 -6.69536531e-01 -5.82397640e-01 -1.01160264e+00 -2.57817507e-01 -7.66682327e-01 2.39059314e-01 3.81804675e-01 -1.17802846e+00 -6.30348265e-01 5.28234959e-01 -9.83552694e-01 -7.54364133e-01 -2.15546936e-01 5.20527422e-01 -3.45262796e-01 1.70787901e-01 -6.56128287e-01 -5.62257886e-01 -6.55572832e-01 -1.36168921e+00 1.03411388e+00 1.69745069e-02 -1.37349069e-01 -9.14005816e-01 8.84826928e-02 6.43265784e-01 7.59278119e-01 -9.10142124e-01 1.03353274e+00 -5.88080585e-01 -4.22196746e-01 1.46819293e-01 -1.51223063e-01 5.19639850e-01 6.72027692e-02 -3.35373253e-01 -1.50497997e+00 -6.41929865e-01 -5.55174835e-02 -2.35747904e-01 6.04853809e-01 2.38200784e-01 1.20320630e+00 -4.28281963e-01 -2.52155036e-01 9.12093878e-01 8.53642583e-01 2.04287618e-02 2.99045920e-01 2.06045270e-01 6.47691786e-01 4.17442590e-01 4.25135463e-01 2.60837376e-01 4.21358824e-01 9.71030056e-01 -5.71537092e-02 6.36098385e-02 -3.67150187e-01 -5.51344752e-01 7.68689334e-01 1.52158260e+00 1.24850504e-01 -3.60221155e-02 -6.55134201e-01 4.19682592e-01 -1.50355697e+00 -6.33466542e-01 4.54804480e-01 2.43786669e+00 1.29967642e+00 8.12541023e-02 1.39777884e-01 2.68522471e-01 8.35983574e-01 2.55087297e-02 -7.08658934e-01 -4.99854922e-01 5.77190071e-02 3.23828489e-01 1.37419313e-01 7.82110512e-01 -1.02190924e+00 1.37798667e+00 6.09166431e+00 1.01498544e+00 -1.23842204e+00 5.58279991e-01 4.34446573e-01 -1.51368186e-01 -2.01108113e-01 3.00488509e-02 -1.14179921e+00 4.65577930e-01 1.41451240e+00 -7.24423155e-02 7.49614775e-01 8.05149794e-01 2.56426990e-01 4.54685777e-01 -1.27630687e+00 1.11878419e+00 -1.04537763e-01 -9.56984043e-01 9.10805315e-02 -1.10129952e-01 4.84722853e-01 2.41854846e-01 2.26243678e-02 8.02555263e-01 1.11118734e-01 -8.57131839e-01 7.87331820e-01 -4.19688895e-02 1.12361085e+00 -7.80805290e-01 6.58230841e-01 4.45605695e-01 -1.17949724e+00 -1.11459032e-01 -5.67835867e-01 1.81295380e-01 3.25732946e-01 7.63279140e-01 -1.04527354e+00 3.11607718e-01 7.82231688e-01 4.45501834e-01 -5.39618492e-01 8.14272344e-01 -3.83533627e-01 1.01972413e+00 -3.40031922e-01 2.65658766e-01 9.24068764e-02 -9.59781334e-02 2.73561299e-01 1.25135660e+00 3.59964132e-01 -2.35206544e-01 -1.21181995e-01 5.52619219e-01 -4.64511067e-01 2.00999841e-01 -6.78615868e-02 1.24907186e-02 7.32209682e-01 1.27583373e+00 -6.73173517e-02 -3.60653281e-01 -3.70690316e-01 7.90188730e-01 8.06287646e-01 2.85953492e-01 -9.51237500e-01 -1.52491301e-01 8.37289512e-01 -6.93313554e-02 4.85595047e-01 -4.09380533e-02 -3.53720225e-02 -1.29296780e+00 1.29462644e-01 -1.14618874e+00 3.06858242e-01 -3.81805986e-01 -1.16099274e+00 9.93847251e-01 -1.86375767e-01 -1.11219013e+00 -3.94711435e-01 -3.69429499e-01 -4.03934509e-01 1.00976574e+00 -1.65066588e+00 -1.26207650e+00 -1.23315632e-01 7.68966079e-01 6.50498271e-01 -3.22348982e-01 9.56060767e-01 7.73953974e-01 -9.48388636e-01 1.46609807e+00 3.38969752e-03 4.70371731e-02 9.86655116e-01 -7.95684576e-01 6.11489654e-01 8.28223169e-01 5.30543700e-02 6.99121058e-01 3.16371441e-01 -1.50095448e-01 -1.26351392e+00 -1.21057570e+00 1.23593092e+00 -2.60238826e-01 6.64419591e-01 -6.59417152e-01 -1.06828189e+00 8.21704328e-01 2.41314452e-02 3.12216561e-02 7.55195677e-01 3.31062257e-01 -6.43870234e-01 -5.06966293e-01 -1.03126526e+00 4.63338464e-01 1.18458724e+00 -9.46295500e-01 -4.14381742e-01 1.27960563e-01 8.47074568e-01 -3.79814237e-01 -1.05451453e+00 2.34632149e-01 3.89300048e-01 -6.29796922e-01 9.10567224e-01 -5.38975835e-01 -2.36696586e-01 -2.77076036e-01 -8.73825252e-02 -1.60445607e+00 -3.03737551e-01 -9.83497322e-01 -6.62205368e-02 1.69458246e+00 7.79260576e-01 -8.35234821e-01 4.85172987e-01 3.73488128e-01 -4.03195411e-01 -3.97821665e-01 -1.30452573e+00 -7.64000714e-01 1.96035549e-01 -4.91328895e-01 1.06148040e+00 8.60957801e-01 -1.69736997e-03 6.10961914e-01 -2.65446723e-01 1.40767708e-01 3.46637040e-01 -2.31871698e-02 8.73346806e-01 -9.74689364e-01 -4.54067826e-01 -3.46440673e-01 3.78969647e-02 -1.22155941e+00 2.83857405e-01 -1.15352440e+00 9.60542858e-02 -9.25939918e-01 8.47131610e-02 -9.26133275e-01 -4.64023203e-01 7.90284216e-01 -3.87324870e-01 1.22311741e-01 1.38655081e-01 1.02366067e-01 -2.62682527e-01 7.09989727e-01 1.03934455e+00 -2.35362202e-01 -2.97057569e-01 2.99676865e-01 -5.58333635e-01 4.68576968e-01 9.05842483e-01 -6.10075533e-01 -4.66568768e-01 -7.58944750e-01 2.81953048e-02 -2.18059197e-01 4.29927297e-02 -7.52208889e-01 1.63701773e-01 1.85343638e-01 -3.42720188e-02 -2.60494828e-01 4.01750714e-01 -6.11702800e-01 1.11456260e-01 2.74570316e-01 -1.84539795e-01 -1.27252549e-01 4.40220028e-01 5.49247488e-02 -3.45879138e-01 -2.81991839e-01 9.87402022e-01 2.33362347e-01 -3.98467064e-01 4.89959717e-01 -1.79036513e-01 1.50952622e-01 6.14279628e-01 6.19208068e-02 -1.03999346e-01 -1.68718323e-01 -6.29403532e-01 -7.35895708e-02 1.51602238e-01 6.12116218e-01 1.64365038e-01 -1.17672086e+00 -7.88109899e-01 5.61337769e-01 1.90874279e-01 1.41537130e-01 4.83149946e-01 9.99502957e-01 3.71095836e-02 6.15846038e-01 1.21897049e-01 -6.59341812e-01 -1.16382647e+00 5.04160285e-01 4.76347715e-01 -3.45567405e-01 -4.87116486e-01 1.10302222e+00 4.17119771e-01 -1.00588441e+00 3.01262051e-01 -3.96678478e-01 8.96751955e-02 3.66960503e-02 5.20729184e-01 3.08075219e-01 5.20712912e-01 -6.65595293e-01 -5.36382496e-01 5.47382832e-01 -2.30716124e-01 -9.87804532e-02 1.41964459e+00 -1.81438938e-01 4.57377210e-02 2.24331975e-01 1.19131517e+00 1.14952646e-01 -1.37970114e+00 -6.04599655e-01 -1.62550941e-01 -6.38886243e-02 1.71733558e-01 -7.35873580e-01 -1.20886600e+00 1.17667878e+00 3.93279076e-01 -1.51884109e-01 1.22836196e+00 5.79079688e-02 8.12526941e-01 3.55770350e-01 3.29910338e-01 -1.11255908e+00 -3.35798055e-01 7.45314717e-01 7.88507104e-01 -1.07728767e+00 -3.10959250e-01 -4.98460919e-01 -5.59362233e-01 7.18620420e-01 4.98787045e-01 4.64630514e-01 7.18401611e-01 2.93824911e-01 1.53761789e-01 4.87435162e-01 -8.66526365e-01 5.89790978e-02 2.03820810e-01 5.29169798e-01 5.70289135e-01 1.76896676e-01 1.97230399e-01 7.03553975e-01 -5.41437328e-01 -7.35681131e-02 -3.62444259e-02 4.44288880e-01 -1.49914175e-01 -1.36420727e+00 -1.67496815e-01 2.13690594e-01 -2.00607017e-01 -4.27429765e-01 -4.55158018e-02 4.92298275e-01 7.50195188e-03 1.06574702e+00 -9.93279070e-02 -3.13703209e-01 6.09681666e-01 3.08544487e-01 4.42101061e-01 -6.31156027e-01 -9.71957564e-01 4.79712673e-02 1.01714812e-01 -5.09474695e-01 -2.32547402e-01 -6.55670822e-01 -8.98500741e-01 -3.78272504e-01 -4.87616211e-01 3.19428205e-01 6.00715160e-01 1.10548258e+00 5.94662726e-01 5.88901877e-01 6.99108362e-01 -7.68997431e-01 -9.84009802e-01 -1.41118181e+00 -1.59752548e-01 3.67580093e-02 2.36326605e-01 -6.33183181e-01 -4.42470044e-01 2.14786902e-01]
[14.131278991699219, 6.620329856872559]
5412226f-ed55-4552-a7c7-e302fc3c2177
real-time-aerial-detection-and-reasoning-on
2305.12414
null
https://arxiv.org/abs/2305.12414v1
https://arxiv.org/pdf/2305.12414v1.pdf
Real-time Aerial Detection and Reasoning on Embedded-UAVs
We present a unified pipeline architecture for a real-time detection system on an embedded system for UAVs. Neural architectures have been the industry standard for computer vision. However, most existing works focus solely on concatenating deeper layers to achieve higher accuracy with run-time performance as the trade-off. This pipeline of networks can exploit the domain-specific knowledge on aerial pedestrian detection and activity recognition for the emerging UAV applications of autonomous surveying and activity reporting. In particular, our pipeline architectures operate in a time-sensitive manner, have high accuracy in detecting pedestrians from various aerial orientations, use a novel attention map for multi-activities recognition, and jointly refine its detection with temporal information. Numerically, we demonstrate our model's accuracy and fast inference speed on embedded systems. We empirically deployed our prototype hardware with full live feeds in a real-world open-field environment.
['Tin Lai']
2023-05-21
null
null
null
null
['pedestrian-detection']
['computer-vision']
[ 4.52202186e-02 -5.11016250e-01 -7.58173689e-02 -3.69092703e-01 -2.77397811e-01 -7.49439061e-01 2.39974365e-01 -1.39783874e-01 -4.72877532e-01 2.33874246e-01 -2.88922846e-01 -3.11532050e-01 1.93038985e-01 -7.58229971e-01 -6.13241553e-01 -4.95328724e-01 -6.04766488e-01 -2.45680854e-01 8.55844378e-01 3.67805362e-02 -2.56186962e-01 7.24598765e-01 -1.70817482e+00 1.00540280e-01 3.41291338e-01 1.18829072e+00 2.10801363e-01 1.20084250e+00 8.52432489e-01 8.28959167e-01 -7.72381604e-01 1.54924989e-02 6.58144772e-01 3.59971970e-01 -3.68217349e-01 1.74497604e-01 6.00653172e-01 -1.09397388e+00 -7.92589784e-01 9.82713640e-01 2.03239173e-01 -1.73754886e-01 2.26255596e-01 -1.29431939e+00 -2.58231491e-01 1.92209467e-01 -7.61732817e-01 5.78039348e-01 8.51137191e-02 9.61058021e-01 5.48115849e-01 -4.80864346e-01 -8.62168521e-02 1.11401963e+00 8.08108985e-01 2.08043516e-01 -8.38873208e-01 -7.45033860e-01 5.37665725e-01 1.58624575e-01 -1.26634324e+00 -3.97059143e-01 2.38642231e-01 -5.27516842e-01 1.24850535e+00 2.14568358e-02 1.15002584e+00 1.20967865e+00 2.74303526e-01 8.20232213e-01 5.41866004e-01 2.49252841e-01 1.41757503e-01 -2.99967617e-01 2.11174518e-01 1.09508491e+00 6.68040931e-01 4.58817303e-01 -4.48228806e-01 9.53660607e-02 1.02508771e+00 2.39682272e-01 -1.65727168e-01 3.08691543e-02 -1.45131767e+00 5.24849534e-01 7.00958192e-01 -2.61483282e-01 -4.32960331e-01 6.42976463e-01 2.46185914e-01 8.76091123e-02 2.53038675e-01 3.58045757e-01 -7.02489436e-01 2.78736968e-02 -9.79223907e-01 1.31275237e-01 7.00865388e-01 1.18571508e+00 5.31600952e-01 3.58644664e-01 -2.17459738e-01 1.42946139e-01 6.39118850e-01 9.11220014e-01 1.39390916e-01 -1.10223186e+00 1.20343596e-01 6.98864520e-01 3.42348695e-01 -7.32053399e-01 -5.14338017e-01 -6.08667076e-01 -6.67933702e-01 5.32456338e-01 2.62454271e-01 -7.03709841e-01 -8.60534251e-01 1.60421312e+00 3.76459450e-01 3.58088881e-01 1.18628323e-01 1.10345507e+00 4.02530313e-01 6.78395033e-01 2.56534606e-01 2.46334553e-01 1.60499167e+00 -1.02751100e+00 -1.66303709e-01 -8.28641891e-01 3.35251004e-01 -3.96977544e-01 5.33829808e-01 2.40283638e-01 -6.64004087e-01 -8.83829594e-01 -1.53975129e+00 1.33450404e-01 -3.78336430e-01 8.97420108e-01 9.74836707e-01 6.27380013e-01 -1.12566459e+00 4.01248217e-01 -1.43836045e+00 -7.19423890e-01 4.83332962e-01 4.89556968e-01 -1.65394247e-01 2.73920864e-01 -7.32809544e-01 6.13535225e-01 2.73625672e-01 2.72002578e-01 -1.56034768e+00 -4.85496044e-01 -8.17486465e-01 9.55820903e-02 4.54042017e-01 -8.69246900e-01 1.47074413e+00 -8.57606828e-01 -1.52475274e+00 5.24592221e-01 2.18501151e-01 -1.01140785e+00 2.07195461e-01 -4.60098833e-01 -3.56305748e-01 1.36191800e-01 9.44345742e-02 1.05956995e+00 9.89423335e-01 -5.00036120e-01 -1.17662489e+00 -3.25390786e-01 5.32068670e-01 5.87136634e-02 -4.19161201e-01 5.88527024e-02 -1.61606446e-01 -4.77760911e-01 -2.27955759e-01 -8.75734866e-01 -3.98842633e-01 6.26054227e-01 7.93253630e-02 5.95586095e-03 1.26036131e+00 -4.48675483e-01 7.29149044e-01 -2.11499786e+00 -2.55838484e-01 -3.11097890e-01 2.30042458e-01 5.07796347e-01 -2.02536602e-02 3.04852650e-02 3.74585837e-01 -3.75213802e-01 -1.81254834e-01 -1.36240661e-01 -2.76986480e-01 7.57748932e-02 -3.08049738e-01 6.27873957e-01 5.83575845e-01 6.16664708e-01 -1.08736193e+00 -3.31825227e-01 4.63296473e-01 3.52584273e-01 -6.07228935e-01 4.67879385e-01 -2.31433973e-01 1.89074054e-01 -3.84185344e-01 1.05004954e+00 4.00715321e-01 -2.80059248e-01 1.23433597e-01 -3.69465888e-01 -6.20911360e-01 -1.76309809e-01 -1.10405493e+00 1.50622523e+00 -2.58221030e-01 1.01404786e+00 4.79755908e-01 -5.20354629e-01 6.05543196e-01 -1.21197313e-01 2.42218122e-01 2.07455456e-02 3.33611250e-01 -2.66333193e-01 -2.28451788e-02 -3.50321651e-01 8.20422173e-01 8.57489645e-01 -3.12726349e-02 -1.02106147e-01 2.76936501e-01 9.64626446e-02 2.67399251e-01 -2.94498801e-02 1.58830953e+00 3.02600890e-01 3.69892150e-01 -2.54381984e-01 2.36460090e-01 3.44389230e-01 4.41467404e-01 7.60880053e-01 -6.57664597e-01 1.29119352e-01 1.52284250e-01 -6.60298228e-01 -6.26095951e-01 -1.02562714e+00 1.01379806e-03 1.18235767e+00 3.39165419e-01 -5.06838560e-01 -8.26777875e-01 -6.02115512e-01 -1.96390703e-01 1.37716755e-01 -3.63376707e-01 -1.12510830e-01 -3.20993304e-01 -9.95212853e-01 8.57212245e-01 9.03573513e-01 1.10701621e+00 -5.90815604e-01 -1.69410050e+00 3.01791012e-01 2.24637762e-01 -1.41093314e+00 -2.03260168e-01 4.66652185e-01 -5.91591597e-01 -1.34600592e+00 -3.81683946e-01 -5.98915219e-01 4.51354533e-01 8.53072882e-01 8.04244995e-01 -7.93632269e-02 -6.64685428e-01 6.53803587e-01 -6.87966719e-02 -5.18206596e-01 1.93162739e-01 -2.54983772e-02 2.50897616e-01 -4.11588885e-02 3.10004890e-01 -4.70832944e-01 -8.87768745e-01 3.30921203e-01 -5.19538283e-01 -7.18385652e-02 8.11903596e-01 6.54504299e-01 2.55464941e-01 1.03614248e-01 -1.28575161e-01 1.26796111e-01 1.95700601e-01 -4.97702122e-01 -1.55348492e+00 1.08335331e-01 -1.53942168e-01 -1.67588189e-01 3.75553668e-01 -5.34921288e-01 -7.17764616e-01 7.63824701e-01 8.67234021e-02 -5.30398726e-01 -4.80791986e-01 1.71589896e-01 -9.05676857e-02 -4.57575500e-01 5.38085163e-01 5.15469210e-03 -1.99722767e-01 -2.92930473e-03 9.21467394e-02 5.02041340e-01 8.14198494e-01 -3.15229803e-01 6.12549245e-01 5.90512633e-01 1.31449029e-01 -1.19002438e+00 -7.82018363e-01 -4.64941829e-01 -7.41052091e-01 -4.29825604e-01 1.00215542e+00 -1.67202008e+00 -1.08290696e+00 8.49852324e-01 -1.35947287e+00 -5.56289911e-01 9.87839699e-02 5.21943033e-01 -5.86809963e-02 2.58128136e-01 -5.18307567e-01 -8.90923381e-01 -3.80344599e-01 -1.20328629e+00 1.37269998e+00 6.62383974e-01 1.19123839e-01 -5.03808260e-01 -5.61608300e-02 -2.32880011e-01 3.34887177e-01 2.88622409e-01 -3.71325202e-02 -1.31119236e-01 -1.16037714e+00 -3.50964606e-01 -5.87653458e-01 2.07699150e-01 -2.30141804e-01 4.85479355e-01 -1.11620295e+00 -3.47747564e-01 -4.25091118e-01 -6.09159730e-02 9.09671783e-01 5.97364783e-01 1.14873672e+00 -2.55258709e-01 -6.26831114e-01 9.69314158e-01 1.21109378e+00 8.66122469e-02 1.37397185e-01 8.67438912e-02 5.45615792e-01 2.61020601e-01 8.27208936e-01 7.32797503e-01 5.26002109e-01 5.55205882e-01 1.03004825e+00 -6.46145120e-02 -9.06827822e-02 -2.31402554e-02 8.73623908e-01 1.35561734e-01 -1.53745592e-01 -3.62034351e-01 -8.01932693e-01 5.66469967e-01 -2.09715748e+00 -9.96749461e-01 -7.76551515e-02 2.06836510e+00 1.41815975e-01 1.85382172e-01 2.35086873e-01 -3.11566323e-01 4.89410311e-01 2.65362024e-01 -6.09978616e-01 3.80115181e-01 6.75171092e-02 -1.40218019e-01 1.20701289e+00 1.22039467e-01 -1.91848564e+00 1.06280744e+00 6.78086853e+00 1.36892736e-01 -1.03641605e+00 -6.72836555e-03 1.51866868e-01 -3.09856653e-01 8.81997347e-01 -1.10325120e-01 -1.23384309e+00 3.81662786e-01 1.17620289e+00 2.90177166e-01 2.62424707e-01 1.34222293e+00 9.37337987e-03 -2.71265328e-01 -9.63435650e-01 9.25270915e-01 -1.80885866e-01 -1.09708965e+00 -4.29551482e-01 1.18842259e-01 2.90601283e-01 5.60499370e-01 -5.70678040e-02 3.17660928e-01 8.37248564e-01 -6.70761466e-01 8.53365362e-01 2.35074982e-01 6.67152107e-01 -5.01580834e-01 7.73744702e-01 2.77114749e-01 -1.70594966e+00 -3.89094025e-01 -4.58625674e-01 -4.96113539e-01 1.16553910e-01 1.78358614e-01 -8.65608454e-01 9.37432945e-02 1.26681566e+00 1.01249850e+00 -9.71318543e-01 1.14320660e+00 -3.56897444e-01 6.14662111e-01 -6.85364783e-01 -3.15120786e-01 5.33400416e-01 1.88973874e-01 6.89590812e-01 1.33020198e+00 4.58145916e-01 1.75212398e-01 6.21722460e-01 6.34713650e-01 1.48834437e-01 -7.90908277e-01 -8.34465265e-01 -7.99541250e-02 3.98858458e-01 1.63645172e+00 -7.22034931e-01 -3.42634022e-01 -5.05850852e-01 8.57491434e-01 1.48423612e-01 9.76964682e-02 -1.32029474e+00 -4.45322484e-01 1.34983957e+00 6.52141869e-02 4.47255522e-01 -7.80312598e-01 6.32495619e-03 -1.14243793e+00 -1.44743085e-01 -4.34787899e-01 1.80826873e-01 -8.42072904e-01 -8.52421463e-01 6.76181734e-01 5.03106974e-02 -1.25746810e+00 1.17458366e-01 -1.21758950e+00 -5.30778050e-01 4.90917146e-01 -1.28113925e+00 -1.47932065e+00 -9.58372176e-01 4.70389158e-01 7.55418956e-01 -2.86790073e-01 7.90138483e-01 2.10272089e-01 -9.39893544e-01 9.10804197e-02 -5.21862149e-01 4.52082098e-01 5.17517149e-01 -1.17608309e+00 9.40516114e-01 1.76964879e+00 3.22915092e-02 4.08397228e-01 5.04368782e-01 -6.78092241e-01 -1.64800537e+00 -1.53400064e+00 -2.31054038e-01 -5.53563833e-01 8.09041917e-01 -2.14462459e-01 -4.65571582e-01 9.56800640e-01 3.99787605e-01 2.84815162e-01 4.30652738e-01 -1.21386565e-01 -2.04875037e-01 -3.25190693e-01 -9.03223693e-01 6.48805022e-01 1.19921112e+00 -1.80052832e-01 -2.83956021e-01 3.52779984e-01 8.47656190e-01 -5.96216619e-01 -7.23760009e-01 4.47393417e-01 6.41176999e-01 -8.18454325e-01 1.08843040e+00 -4.85949010e-01 -7.59901805e-03 -8.82899284e-01 -2.80787021e-01 -9.59076464e-01 -7.36856282e-01 -5.32122254e-01 -4.14400280e-01 7.63199449e-01 1.75154284e-02 -4.73452151e-01 6.33156657e-01 1.66452661e-01 -4.13422167e-01 -2.65929222e-01 -6.52419984e-01 -6.49345100e-01 -9.24918056e-01 -4.08778250e-01 5.67513227e-01 2.31105089e-01 -4.24533427e-01 2.27011725e-01 -4.79053259e-01 1.28518462e+00 9.04203475e-01 -1.28105953e-01 9.45954740e-01 -1.22222877e+00 -2.31181115e-01 -2.10468918e-01 -6.34675920e-01 -1.42603147e+00 -8.25463980e-02 -2.67594516e-01 3.13288510e-01 -1.22342122e+00 -2.19021782e-01 5.60564734e-02 -9.39012170e-02 7.77005553e-01 1.20777220e-01 2.04278409e-01 -3.65930945e-02 -9.34403241e-02 -9.37146962e-01 2.20341623e-01 4.21327770e-01 -3.97173285e-01 7.36886635e-02 7.47758299e-02 -4.25420702e-01 9.02230024e-01 7.01714933e-01 -2.77645588e-01 -1.74450099e-01 -8.28929603e-01 -1.78961292e-01 -1.52030319e-01 1.07326877e+00 -1.90363979e+00 6.07211649e-01 -1.84188738e-01 8.36518049e-01 -6.89389586e-01 5.38267970e-01 -1.02163172e+00 -6.71212301e-02 8.12844157e-01 1.97482258e-01 2.01524198e-01 5.69502473e-01 8.34229469e-01 9.22671184e-02 2.02749938e-01 9.31859493e-01 -2.62800436e-02 -1.39056742e+00 6.02602541e-01 -9.15075183e-01 -3.41895372e-01 1.32818103e+00 -1.52652860e-02 -5.61107039e-01 1.06755979e-01 -1.42639250e-01 3.21242988e-01 4.04018193e-01 5.61242700e-01 6.22600853e-01 -8.64776015e-01 -4.24331784e-01 5.35486817e-01 2.18936861e-01 5.29801175e-02 1.16267242e-01 6.77728891e-01 -8.62440348e-01 4.83720690e-01 -5.09747863e-01 -9.98461366e-01 -1.30590141e+00 4.37972575e-01 5.14362812e-01 -6.12144098e-02 -3.81652921e-01 8.91473711e-01 1.20136477e-01 9.67209861e-02 3.41469824e-01 -7.32725680e-01 1.80821475e-02 -2.67014563e-01 8.30551922e-01 3.31949860e-01 -1.52885810e-01 -2.06693158e-01 -5.75694203e-01 4.34168339e-01 2.26915091e-01 1.21993452e-01 9.84274924e-01 -4.11352664e-02 3.31156999e-01 -7.45567307e-02 4.30731475e-01 -5.28565586e-01 -2.07430863e+00 1.22615181e-01 -2.34535411e-01 -3.03687274e-01 5.40133059e-01 -6.60215437e-01 -1.03183842e+00 8.12811077e-01 8.29309285e-01 6.66661039e-02 1.05722010e+00 -2.73129493e-01 5.99309981e-01 8.40359688e-01 6.64718390e-01 -8.90493870e-01 1.59034971e-02 6.58424497e-01 5.65029442e-01 -1.43257058e+00 4.76779379e-02 -2.46512577e-01 -3.60531569e-01 1.22358227e+00 1.10017204e+00 -2.57454515e-01 5.44974744e-01 7.67865717e-01 -1.81529939e-01 -5.67644872e-02 -8.87965262e-01 -6.32520258e-01 -6.51216432e-02 9.78486598e-01 -9.85417143e-02 2.03994483e-01 6.89545393e-01 6.34187996e-01 1.29200116e-01 -1.82272345e-01 4.11379516e-01 9.86336827e-01 -7.53758550e-01 -3.99706870e-01 -3.44307393e-01 1.82136104e-01 -2.56180018e-01 1.63916312e-02 -2.43039861e-01 7.28611290e-01 1.38542458e-01 1.06286466e+00 8.83608386e-02 -6.01683676e-01 3.06085020e-01 -4.58368450e-01 4.72984612e-01 -5.23050427e-01 -6.86279774e-01 -1.97872356e-01 7.56768063e-02 -9.35508668e-01 -3.42041671e-01 -6.03040814e-01 -6.69392407e-01 -3.07993352e-01 -1.78379387e-01 -4.36214685e-01 7.46673763e-01 6.93994284e-01 6.63548648e-01 8.36387217e-01 4.04328227e-01 -1.14791143e+00 -4.37026411e-01 -8.07417750e-01 -1.70876101e-01 -5.07975221e-01 4.32686210e-01 -7.65490472e-01 -1.42808333e-01 2.11985469e-01]
[6.6969757080078125, -1.9073266983032227]
42df154a-63da-4468-b5df-94887436ac65
ego-body-pose-estimation-via-ego-head-pose
2212.04636
null
https://arxiv.org/abs/2212.04636v2
https://arxiv.org/pdf/2212.04636v2.pdf
Ego-Body Pose Estimation via Ego-Head Pose Estimation
Estimating 3D human motion from an egocentric video sequence plays a critical role in human behavior understanding and has various applications in VR/AR. However, naively learning a mapping between egocentric videos and human motions is challenging, because the user's body is often unobserved by the front-facing camera placed on the head of the user. In addition, collecting large-scale, high-quality datasets with paired egocentric videos and 3D human motions requires accurate motion capture devices, which often limit the variety of scenes in the videos to lab-like environments. To eliminate the need for paired egocentric video and human motions, we propose a new method, Ego-Body Pose Estimation via Ego-Head Pose Estimation (EgoEgo), which decomposes the problem into two stages, connected by the head motion as an intermediate representation. EgoEgo first integrates SLAM and a learning approach to estimate accurate head motion. Subsequently, leveraging the estimated head pose as input, EgoEgo utilizes conditional diffusion to generate multiple plausible full-body motions. This disentanglement of head and body pose eliminates the need for training datasets with paired egocentric videos and 3D human motion, enabling us to leverage large-scale egocentric video datasets and motion capture datasets separately. Moreover, for systematic benchmarking, we develop a synthetic dataset, AMASS-Replica-Ego-Syn (ARES), with paired egocentric videos and human motion. On both ARES and real data, our EgoEgo model performs significantly better than the current state-of-the-art methods.
['Jiajun Wu', 'C. Karen Liu', 'Jiaman Li']
2022-12-09
null
http://openaccess.thecvf.com//content/CVPR2023/html/Li_Ego-Body_Pose_Estimation_via_Ego-Head_Pose_Estimation_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Li_Ego-Body_Pose_Estimation_via_Ego-Head_Pose_Estimation_CVPR_2023_paper.pdf
cvpr-2023-1
['head-pose-estimation']
['computer-vision']
[-4.48454887e-01 -1.21350512e-01 -1.60784706e-01 -2.38181353e-01 -5.86952209e-01 -4.03129488e-01 3.77318293e-01 -7.79032052e-01 -2.21805602e-01 4.86716300e-01 7.73737848e-01 3.10055375e-01 2.90836602e-01 -2.96221763e-01 -7.78642356e-01 -5.29332221e-01 -1.91159435e-02 4.20241505e-01 1.82130158e-01 -1.08647458e-01 -2.70581186e-01 1.77754626e-01 -1.53953493e+00 -1.09961145e-01 3.98023009e-01 5.54870725e-01 1.29514664e-01 1.06331527e+00 5.95491469e-01 7.35600531e-01 -1.98816091e-01 -1.88185826e-01 3.00603807e-01 -4.55099970e-01 -4.96044427e-01 3.25609893e-01 7.54956782e-01 -9.14875329e-01 -9.07283366e-01 7.31258452e-01 9.26026940e-01 5.18547595e-01 3.87842923e-01 -1.45102894e+00 -1.32317171e-01 1.73442680e-02 -7.30401278e-01 1.07409276e-01 1.17856252e+00 4.49383855e-01 5.61142147e-01 -7.78966010e-01 9.10393655e-01 1.46182179e+00 5.19101262e-01 8.37748587e-01 -9.79846776e-01 -7.22584784e-01 2.22974628e-01 3.45046610e-01 -1.58648014e+00 -7.02674925e-01 8.00627828e-01 -6.44529998e-01 5.89907646e-01 2.08495632e-02 1.09490311e+00 1.57388377e+00 2.18005195e-01 1.15505862e+00 3.05958629e-01 -5.14464453e-03 2.77105331e-01 -3.91162425e-01 -3.65413278e-01 6.94229782e-01 1.97295919e-01 -8.52160752e-02 -8.22515726e-01 4.19201218e-02 1.07658756e+00 1.56319007e-01 -6.10215306e-01 -1.00994396e+00 -1.42535090e+00 5.97047985e-01 2.23952934e-01 -1.68305457e-01 -5.21532476e-01 3.39033246e-01 3.16100270e-01 -2.64458627e-01 2.84428239e-01 -1.56968951e-01 -2.17592582e-01 -3.85287404e-01 -9.21015501e-01 6.92718863e-01 6.38827443e-01 1.32691252e+00 4.59556043e-01 1.21307880e-01 -1.36613756e-01 2.59647131e-01 4.13632184e-01 7.32534170e-01 5.30472696e-01 -1.24588156e+00 6.21771395e-01 3.54858041e-01 3.42333466e-01 -1.17983198e+00 -5.45370221e-01 -1.45967931e-01 -4.72775072e-01 -5.01071692e-01 4.49137479e-01 -2.98196077e-01 -6.98571622e-01 2.15296435e+00 1.01141477e+00 6.18519723e-01 -1.39941409e-01 1.35173893e+00 7.56569505e-01 3.06329668e-01 1.89950764e-02 -1.68580949e-01 1.44133806e+00 -1.15923822e+00 -8.15468729e-01 -4.29851711e-01 7.59481490e-01 -3.59717011e-01 8.95295322e-01 1.03539266e-01 -1.06396365e+00 -5.20114720e-01 -7.31174886e-01 -1.55315191e-01 2.30755553e-01 -2.32603252e-02 5.54307878e-01 4.24359173e-01 -8.55007648e-01 1.86202765e-01 -1.24130666e+00 -5.63664734e-01 9.24688280e-02 2.35544488e-01 -9.13462818e-01 -3.05603713e-01 -1.08769357e+00 5.47630370e-01 2.68492550e-01 2.04275027e-01 -1.17353678e+00 -5.71736455e-01 -1.47127354e+00 -1.74874961e-01 5.28278589e-01 -1.29418874e+00 1.34460664e+00 -4.94056702e-01 -1.53268600e+00 7.11381078e-01 -3.92402947e-01 -1.65696532e-01 8.84476185e-01 -8.18196595e-01 -2.50761986e-01 3.39344591e-01 9.58720744e-02 6.91762567e-01 9.62268591e-01 -9.28676009e-01 -3.05892229e-01 -7.33536780e-01 -2.75152586e-02 5.97848475e-01 2.58828774e-02 -2.72241175e-01 -1.01446903e+00 -6.10884786e-01 3.22687745e-01 -1.24348807e+00 6.40276447e-02 -4.79514040e-02 -3.59229714e-01 9.57420021e-02 8.21544707e-01 -7.58854330e-01 1.13327587e+00 -2.02440596e+00 5.72376430e-01 -2.34364301e-01 3.36386085e-01 -1.06144123e-01 -2.12992150e-02 9.59196091e-02 -1.31806448e-01 -5.02269149e-01 2.51372814e-01 -7.08393812e-01 1.51821701e-02 1.90694898e-01 -9.71900299e-03 8.87915671e-01 -3.90815884e-01 1.00968122e+00 -1.24895072e+00 -6.13904655e-01 3.63756299e-01 6.74754441e-01 -1.06526160e+00 5.12990534e-01 2.00919852e-01 9.96213615e-01 -4.53926861e-01 6.38911784e-01 6.04452133e-01 -8.31848308e-02 1.98306262e-01 -2.88552284e-01 2.59238988e-01 4.16936353e-02 -1.30037105e+00 2.32087421e+00 -1.99293941e-01 3.89770925e-01 1.08560093e-01 -4.96556491e-01 2.92354316e-01 5.69427848e-01 7.19119668e-01 -2.97505081e-01 3.26990545e-01 -1.54081464e-01 -3.43098611e-01 -9.80915010e-01 4.95645285e-01 -4.93230298e-03 -1.29698724e-01 3.20830584e-01 2.79080331e-01 -2.20307112e-02 8.15657433e-03 3.05277258e-01 1.08489931e+00 7.42495894e-01 4.16443050e-01 1.69905588e-01 4.78384197e-01 -2.44782835e-01 9.42802310e-01 3.28147084e-01 -5.74073076e-01 8.92415762e-01 2.37923026e-01 -3.80195290e-01 -7.79794395e-01 -1.23464751e+00 4.23951298e-01 1.00104201e+00 3.06637794e-01 -6.35210574e-01 -1.09003818e+00 -5.37184417e-01 -1.18512496e-01 4.63551551e-01 -5.63081443e-01 -1.72862411e-01 -8.21662009e-01 -3.93013895e-01 3.77321959e-01 6.95438445e-01 4.48325157e-01 -8.19346309e-01 -7.81193018e-01 1.96599588e-03 -8.00826907e-01 -1.50761521e+00 -8.95580173e-01 -7.10282803e-01 -7.52627373e-01 -1.14553905e+00 -1.03322601e+00 -2.37808704e-01 5.60362160e-01 6.09912872e-01 1.00485671e+00 -4.23907936e-01 -2.78652757e-01 8.34366620e-01 -2.98750490e-01 1.10858127e-01 2.46776298e-01 -1.24289781e-01 6.59055352e-01 1.92089275e-01 4.30502683e-01 -6.88118041e-01 -1.06938052e+00 3.81965131e-01 -4.22629893e-01 2.28761345e-01 1.92907438e-01 6.61815226e-01 3.09334189e-01 -4.03600007e-01 5.50593920e-02 -4.88432676e-01 -8.98438096e-02 -7.51987934e-01 -2.12077826e-01 -2.15299070e-01 1.83823377e-01 -3.29668194e-01 2.06319362e-01 -5.37191689e-01 -1.00206172e+00 2.97879785e-01 -2.36585569e-02 -9.34274733e-01 -1.89476594e-01 3.75473052e-02 -6.59549832e-01 3.62119555e-01 4.22148496e-01 1.73629612e-01 -9.53504071e-02 -2.60175586e-01 5.00201941e-01 2.31902212e-01 1.01601911e+00 -4.33208257e-01 8.26745749e-01 7.24104464e-01 -5.59434183e-02 -9.93724525e-01 -6.32013142e-01 -6.95522368e-01 -9.25173879e-01 -3.64975154e-01 1.26322007e+00 -1.49974525e+00 -9.52339053e-01 5.35592318e-01 -1.05042839e+00 -2.95091748e-01 2.82966569e-02 1.01431692e+00 -1.05220866e+00 5.63765407e-01 -4.87342298e-01 -6.92917764e-01 -1.14871256e-01 -1.17559958e+00 1.46621704e+00 2.23204747e-01 -6.69737399e-01 -8.87572646e-01 1.82040334e-01 6.05352044e-01 -2.25175872e-01 4.58511680e-01 2.29166329e-01 -7.48338550e-02 -6.63781345e-01 -4.83777016e-01 2.66000628e-01 -2.45896831e-01 4.17779535e-02 -3.39652926e-01 -8.50232780e-01 -5.23857951e-01 3.72582972e-02 -3.02449584e-01 2.42614269e-01 5.80241621e-01 6.52746141e-01 -2.29787096e-01 -3.54659975e-01 1.04361427e+00 7.59840667e-01 -2.48027682e-01 5.68632782e-01 9.38023478e-02 1.23203075e+00 6.51457310e-01 7.77558744e-01 6.41893804e-01 9.33759034e-01 1.02965391e+00 3.75815660e-01 3.24214756e-01 -7.99783468e-02 -8.25543284e-01 6.01270020e-01 9.09599662e-01 -2.64567345e-01 -2.16158435e-01 -6.07362092e-01 5.66309810e-01 -2.07120347e+00 -1.12620842e+00 1.25401974e-01 2.35792017e+00 2.91794002e-01 -2.98581839e-01 4.82550234e-01 -1.42751351e-01 7.16095865e-01 2.45530948e-01 -7.83384085e-01 3.99788737e-01 2.78535575e-01 -4.96023595e-01 1.30683497e-01 3.93976122e-01 -1.04181850e+00 1.09541929e+00 5.52060843e+00 1.99916124e-01 -1.04470742e+00 1.54303327e-01 -1.60682965e-02 -5.61007202e-01 -6.19310699e-02 -1.26706794e-01 -8.24858665e-01 4.18900549e-01 6.24267876e-01 4.62275222e-02 2.57683486e-01 1.03020847e+00 6.05860114e-01 -1.63344011e-01 -1.31178117e+00 1.43980825e+00 3.58067751e-01 -6.88696444e-01 -1.14346676e-01 1.26485199e-01 6.24131083e-01 -3.62758599e-02 -1.29649252e-01 3.41938406e-01 1.53859615e-01 -6.21763825e-01 9.71428335e-01 4.82502282e-01 6.92273438e-01 -4.82249290e-01 4.51419652e-01 6.50152266e-01 -1.53018951e+00 3.38023645e-03 -1.88269049e-01 -5.87522164e-02 7.03106165e-01 6.56442195e-02 -4.43074971e-01 5.08205891e-01 9.22799647e-01 8.13631475e-01 -2.87785947e-01 7.28274822e-01 -5.38322806e-01 1.71801820e-01 -3.86689335e-01 5.86061895e-01 -1.15745224e-01 -1.25358850e-01 8.48939359e-01 8.84099305e-01 4.50993747e-01 4.37957376e-01 3.35888714e-01 5.44974387e-01 6.52176961e-02 -4.83965650e-02 -9.11569476e-01 2.04857334e-01 3.76732439e-01 1.13975120e+00 -4.10401076e-01 -3.70176613e-01 -3.64788562e-01 1.30691898e+00 1.93793356e-01 5.05277038e-01 -1.02832198e+00 1.48554042e-01 1.03212523e+00 2.77567983e-01 9.24632847e-02 -5.64394116e-01 4.52188373e-01 -1.86845219e+00 4.90653962e-02 -8.48127246e-01 3.75288963e-01 -1.06866288e+00 -5.19415319e-01 3.73518616e-01 2.63118953e-01 -1.42680287e+00 -9.86557186e-01 -1.34767428e-01 -4.76817578e-01 6.39572918e-01 -8.07436168e-01 -1.20294750e+00 -9.21991110e-01 8.90516758e-01 8.17762852e-01 1.56901732e-01 4.20676261e-01 2.97390819e-01 -7.94853389e-01 7.28202164e-01 -4.50539321e-01 8.11488181e-02 1.00869250e+00 -9.06143665e-01 6.08436763e-01 8.33506644e-01 -2.76895594e-02 7.92856395e-01 7.91574776e-01 -7.07059205e-01 -1.86001647e+00 -9.67826247e-01 6.00358725e-01 -9.92317259e-01 3.13293964e-01 -6.21548831e-01 -6.18765414e-01 1.11645114e+00 -2.76735812e-01 4.30375248e-01 5.12044430e-01 -2.65259564e-01 -6.83421493e-02 2.61966646e-01 -9.10741270e-01 9.47481751e-01 1.64571011e+00 -5.85323334e-01 -5.88358641e-01 7.26109818e-02 5.93753099e-01 -8.72845352e-01 -6.58807576e-01 3.48748237e-01 9.78616953e-01 -9.98401165e-01 1.12378144e+00 -6.02355480e-01 3.60154927e-01 -5.08168340e-01 -3.15460384e-01 -1.20606327e+00 -2.20050409e-01 -8.67428780e-01 -7.24827766e-01 9.34393525e-01 -3.76074374e-01 -1.98848873e-01 1.16929889e+00 8.28058183e-01 1.28196999e-01 -4.10225064e-01 -6.95550323e-01 -6.48422778e-01 -4.82544124e-01 -4.77853209e-01 5.54695368e-01 8.74961793e-01 -2.43578013e-03 2.84317344e-01 -8.57356131e-01 1.91651285e-01 7.33644426e-01 -1.19329773e-01 1.71615481e+00 -8.59429717e-01 -2.34777078e-01 1.73509613e-01 -7.18938053e-01 -1.59505951e+00 3.66092265e-01 -4.91431892e-01 1.65638372e-01 -1.12746596e+00 2.42536873e-01 3.02177012e-01 3.58372837e-01 5.67377247e-02 -3.27397704e-01 2.42349014e-01 2.07163736e-01 2.43749514e-01 -6.10233784e-01 9.46543396e-01 1.39053845e+00 1.95698172e-01 -3.97242457e-01 -9.87850800e-02 -2.54580587e-01 1.24369955e+00 1.41579747e-01 -4.06237394e-01 -7.39224732e-01 -4.59324002e-01 -2.77987942e-02 5.55765986e-01 4.89695996e-01 -1.15247262e+00 4.32322234e-01 -1.91587031e-01 5.58841944e-01 -6.86475098e-01 7.48169243e-01 -5.71784973e-01 4.23716217e-01 3.95162582e-01 1.44110695e-01 3.27439576e-01 -1.07315689e-01 8.54341209e-01 -6.02752380e-02 3.26946974e-01 4.98306364e-01 -2.86294669e-01 -9.26729620e-01 6.63345456e-01 -2.00350910e-01 3.13597441e-01 1.01799130e+00 -3.86570007e-01 -6.82975678e-03 -9.59401190e-01 -8.56840312e-01 4.07969445e-01 8.01006973e-01 7.16664672e-01 5.99699795e-01 -1.39567614e+00 -4.23647910e-01 3.82640243e-01 2.03563303e-01 2.74640530e-01 9.19152498e-01 1.12985933e+00 -6.73805058e-01 4.31928337e-01 -1.08889721e-01 -1.00018048e+00 -1.30345321e+00 6.15627587e-01 2.17590094e-01 1.39930695e-01 -9.57309842e-01 5.97460687e-01 8.60727012e-01 -4.20975059e-01 8.00374746e-02 -3.34642455e-02 -1.21834055e-01 -1.67488962e-01 6.42936468e-01 5.84300995e-01 -4.59804982e-01 -1.35928857e+00 -3.31418782e-01 8.26049626e-01 9.16590691e-02 -2.49841839e-01 9.78900373e-01 -6.47578001e-01 4.39978123e-01 3.19853812e-01 1.22301531e+00 3.54863964e-02 -1.66098273e+00 -1.17972955e-01 -3.83243889e-01 -8.24977636e-01 -1.78198919e-01 1.04306974e-01 -1.04812264e+00 8.92631888e-01 4.19523984e-01 -8.11218739e-01 8.42315197e-01 -8.32811370e-02 1.09661400e+00 3.94632965e-01 8.20254028e-01 -9.68897998e-01 3.27840686e-01 3.74071151e-01 7.85143971e-01 -1.25641000e+00 1.36134237e-01 -5.80217779e-01 -8.83489430e-01 6.58070505e-01 9.56759930e-01 -3.85699770e-03 5.59478581e-01 -2.10969597e-01 1.38853900e-02 -1.24926202e-01 -5.81104398e-01 -1.15177564e-01 2.77917743e-01 6.42688334e-01 2.38632008e-01 7.62662739e-02 1.02786347e-01 5.96315980e-01 -4.52510417e-01 1.59186646e-01 4.15665001e-01 9.09747064e-01 3.11825648e-02 -5.88402927e-01 -4.68895525e-01 -3.29499811e-01 -3.41828316e-01 4.02075320e-01 -2.41557974e-02 9.57380474e-01 -6.39474690e-02 7.04794884e-01 1.01314962e-01 -5.45003951e-01 4.05301064e-01 1.30456500e-02 6.43498957e-01 -5.53933859e-01 1.15067102e-01 2.94390887e-01 -1.84122860e-01 -1.15455747e+00 -3.02410454e-01 -7.68431485e-01 -1.13926530e+00 -5.82952559e-01 -1.04965664e-01 -6.60192072e-02 4.37376767e-01 9.80600297e-01 4.33534861e-01 3.63226563e-01 4.07678485e-01 -1.60936940e+00 -3.50414187e-01 -8.29077482e-01 -6.45953774e-01 6.88840032e-01 3.45200300e-01 -1.08398926e+00 -2.72920370e-01 1.76180378e-01]
[7.055445671081543, -0.820791482925415]
f7fd1b9a-690a-48f0-b911-660fda1d4e53
two-way-fixed-effects-and-differences-in
2112.04565
null
https://arxiv.org/abs/2112.04565v6
https://arxiv.org/pdf/2112.04565v6.pdf
Two-Way Fixed Effects and Differences-in-Differences with Heterogeneous Treatment Effects: A Survey
Linear regressions with period and group fixed effects are widely used to estimate policies' effects: 26 of the 100 most cited papers published by the American Economic Review from 2015 to 2019 estimate such regressions. It has recently been shown that those regressions may produce misleading estimates, if the policy's effect is heterogeneous between groups or over time, as is often the case. This survey reviews a fast-growing literature that documents this issue, and that proposes alternative estimators robust to heterogeneous effects. We use those alternative estimators to revisit Wolfers (2006).
["Xavier D'Haultfœuille", 'Clément de Chaisemartin']
2021-12-08
null
null
null
null
['econometrics']
['miscellaneous']
[-3.72124404e-01 -3.70229296e-02 -1.54396737e+00 -9.16159078e-02 -6.12289965e-01 -6.48490131e-01 8.86924565e-01 2.97620475e-01 -8.06343973e-01 1.05293179e+00 7.63284862e-01 -1.11882615e+00 -4.02006626e-01 -4.21794176e-01 -5.64119399e-01 -3.62058073e-01 1.99728622e-03 -3.68655436e-02 -5.10515571e-02 2.48280883e-01 8.31124783e-01 2.54591674e-01 -1.17420924e+00 -3.20230126e-01 1.18015730e+00 2.76526093e-01 -2.53385603e-02 3.30408454e-01 7.45067075e-02 8.16619039e-01 -8.05873692e-01 -4.86215979e-01 2.48946458e-01 -1.95606709e-01 -2.51803666e-01 -1.41335323e-01 -1.41418737e-03 -3.47073346e-01 -1.83018923e-01 6.63077414e-01 2.71340847e-01 5.83516583e-02 9.67444479e-01 -1.00520289e+00 -8.90457571e-01 8.80519450e-01 -7.71861613e-01 4.10490364e-01 1.69852972e-01 2.53201514e-01 6.21659040e-01 -3.63781631e-01 5.64429522e-01 1.59437132e+00 6.07164264e-01 7.96486810e-02 -1.31658018e+00 -1.14936233e+00 6.45168722e-01 1.55685052e-01 -9.99213994e-01 -5.16576231e-01 5.78729391e-01 -7.63823926e-01 6.87477648e-01 7.29624778e-02 6.50682449e-01 1.15543699e+00 5.99152982e-01 2.00029522e-01 1.79608679e+00 -6.56645536e-01 -8.31470639e-02 2.67369807e-01 4.51467097e-01 -1.59526601e-01 1.00141644e+00 4.68624085e-01 1.50777772e-01 -3.47491622e-01 8.43816400e-01 -2.34068409e-02 -3.33488211e-02 1.74628347e-01 -9.85625625e-01 1.15613687e+00 2.82040257e-02 6.01020753e-01 -7.46216238e-01 1.49612343e-02 5.47537148e-01 3.72143656e-01 7.93186069e-01 3.68703783e-01 -4.75012928e-01 -1.72882974e-01 -8.03300619e-01 5.15616179e-01 7.84191310e-01 4.69786018e-01 1.76607490e-01 1.95103437e-02 -2.35845521e-01 7.99268723e-01 1.44558191e-01 7.17120767e-01 4.43947732e-01 -1.08260429e+00 4.75831717e-01 2.12854758e-01 5.70407927e-01 -8.92373204e-01 -5.07014155e-01 -2.17393890e-01 -7.40557790e-01 7.42879882e-02 6.66852534e-01 -6.18505299e-01 -4.20974344e-01 1.72601926e+00 -1.22996353e-01 -2.87687600e-01 -1.61498010e-01 4.46157366e-01 4.69284594e-01 4.80083287e-01 5.07134020e-01 -9.18218195e-01 8.83603275e-01 -8.10857713e-01 -1.04146361e+00 -4.86749783e-02 4.51784432e-01 -5.84218085e-01 7.07014978e-01 2.44699672e-01 -1.23686457e+00 -3.70536596e-01 -2.86704868e-01 3.38816643e-01 -2.99059123e-01 -2.31404066e-01 8.51224303e-01 8.77312899e-01 -8.71252000e-01 3.32406342e-01 -5.71321845e-01 -4.23317581e-01 2.75023460e-01 2.03655154e-01 2.88223743e-01 2.62544930e-01 -1.06061673e+00 1.26317477e+00 -3.15404326e-01 -4.68852907e-01 -4.02361751e-01 -8.62596929e-01 -5.01303196e-01 -8.95849522e-03 5.55488765e-01 -7.70761847e-01 1.18967664e+00 -1.17822587e+00 -1.39211738e+00 4.90240276e-01 -1.70161143e-01 -3.78939956e-01 4.74879801e-01 1.37840256e-01 -6.88842237e-01 -3.45558643e-01 4.81070310e-01 -2.71114767e-01 3.95350039e-01 -8.58016670e-01 -7.34922647e-01 -5.53762913e-01 3.85496877e-02 5.28356917e-02 -6.07621893e-02 1.02936828e+00 2.60615677e-01 -9.39019561e-01 -4.46798652e-01 -8.46465111e-01 -5.32869101e-01 -9.18401539e-01 -1.30108818e-01 -6.30314887e-01 -1.39778420e-01 -6.91792130e-01 1.71329010e+00 -1.77444565e+00 -3.24949652e-01 8.57736394e-02 -1.69304162e-01 -1.53262317e-01 2.97926813e-01 6.95103407e-01 -3.33692819e-01 6.96242988e-01 -9.14741382e-02 9.67752468e-03 1.67917356e-01 -8.98580551e-02 -2.07218274e-01 9.13620889e-01 -2.85861373e-01 6.71782553e-01 -4.75071549e-01 -2.42537677e-01 2.45069504e-01 4.55818363e-02 -3.60916317e-01 -2.05663279e-01 5.74335575e-01 3.49991322e-01 -4.33857888e-01 3.06703776e-01 6.55313313e-01 5.58358096e-02 3.43290299e-01 4.16810662e-01 -1.20064735e+00 8.32246244e-01 -7.51637757e-01 8.43267500e-01 -4.45606917e-01 7.46925414e-01 4.46324870e-02 -1.53705573e+00 6.31823957e-01 6.12597644e-01 6.31042063e-01 -6.37697220e-01 5.71869798e-02 2.68031806e-01 3.53649527e-01 -3.65285993e-01 1.92597777e-01 -4.51815754e-01 -2.34481290e-01 2.80470818e-01 -3.55029464e-01 -8.90193284e-02 1.90110698e-01 -2.32319877e-01 8.13067436e-01 -1.68437809e-01 7.83301711e-01 -8.89569044e-01 3.90660435e-01 -4.28111758e-03 6.91024244e-01 1.03698754e+00 -1.24653473e-01 -2.28449494e-01 6.56562865e-01 -3.94387618e-02 -8.98479402e-01 -7.14062870e-01 -4.60559577e-01 1.04742193e+00 -5.20020247e-01 1.02149226e-01 -3.53881359e-01 -4.31119353e-01 4.22856718e-01 8.94206583e-01 -8.73644590e-01 2.13684738e-01 -3.62461805e-01 -7.73205757e-01 1.43424213e-01 2.85663366e-01 5.41368276e-02 -7.49408484e-01 -4.41247553e-01 2.09035233e-01 1.85758546e-01 -7.28438735e-01 -5.65318055e-02 -2.11480424e-01 -1.06468892e+00 -1.19646251e+00 -1.00911963e+00 -3.30501735e-01 3.40752393e-01 2.64502108e-01 9.73289967e-01 -1.94870517e-01 5.17320216e-01 4.73598987e-01 -4.15476352e-01 -1.10833120e+00 -4.86469030e-01 6.98018745e-02 1.96963355e-01 -5.15496254e-01 5.68922102e-01 -3.04081142e-01 -3.89601678e-01 6.16514459e-02 -5.71895897e-01 -3.94118965e-01 3.13004911e-01 5.23242831e-01 -9.24196914e-02 -1.09316736e-01 1.43090498e+00 -9.72857118e-01 6.81237936e-01 -7.69570231e-01 -9.72374439e-01 4.28883992e-02 -1.10529947e+00 -5.04913270e-01 5.22106946e-01 -5.70833385e-01 -1.18692756e+00 -1.12875104e+00 1.48201466e-01 2.45858386e-01 -1.61167443e-01 7.80136466e-01 4.29072410e-01 5.39087169e-02 4.67934072e-01 -6.04915977e-01 8.83293599e-02 -5.46181202e-01 -1.39326736e-01 7.62292922e-01 1.07148085e-02 -5.17480612e-01 6.76970065e-01 3.55725020e-01 -1.66572720e-01 -4.48077440e-01 -6.00976467e-01 -3.38629246e-01 -2.05441177e-01 -2.27901667e-01 6.77191913e-01 -1.10318244e+00 -4.53551799e-01 1.57860041e-01 -7.97672033e-01 -7.13802040e-01 -1.01176813e-01 1.35017514e+00 -5.63538611e-01 -1.79524168e-01 -4.58125025e-01 -1.28429878e+00 1.32481441e-01 -1.05588448e+00 1.03003234e-01 3.29832733e-01 -1.94693670e-01 -1.34851754e+00 9.60097164e-02 1.81535184e-01 3.65036726e-01 5.53033888e-01 7.19454050e-01 -3.88915956e-01 2.18398850e-02 -4.59644720e-02 -1.63019029e-03 -7.30423722e-03 4.12126362e-01 6.09700084e-01 -4.25522089e-01 -2.98693568e-01 1.52484477e-01 3.57786089e-01 8.07591617e-01 1.49987519e+00 1.03743243e+00 -8.06786597e-01 -4.97773826e-01 2.53581971e-01 1.46779680e+00 7.90321529e-01 3.81033003e-01 6.90540135e-01 1.80682212e-01 1.05008125e+00 8.68331909e-01 5.68857849e-01 6.04836345e-01 6.25521600e-01 -1.41350195e-01 -1.23356827e-01 4.25175995e-01 -1.42871872e-01 4.40400690e-01 7.59945333e-01 -5.98048866e-01 -9.26920027e-02 -8.03842306e-01 7.99596131e-01 -1.67271936e+00 -1.31304431e+00 -4.47669715e-01 2.47387314e+00 6.12676024e-01 1.98859379e-01 9.17574346e-01 -1.00029416e-01 7.63014674e-01 2.91928053e-01 -3.54051702e-02 -9.77237999e-01 -1.12288073e-01 1.39854223e-01 1.27741170e+00 5.27272582e-01 -5.89929283e-01 6.33834124e-01 8.37801552e+00 3.63247752e-01 -7.87671745e-01 1.94256961e-01 8.65660787e-01 -1.63069759e-02 -5.32338262e-01 4.39735651e-01 -6.22613013e-01 6.00576699e-01 1.40907609e+00 -8.81523073e-01 -1.27120586e-02 3.07003945e-01 1.06040764e+00 -5.73729992e-01 -4.05256629e-01 1.86339244e-01 -1.74237892e-01 -9.55658495e-01 -7.30591297e-01 5.45574129e-01 1.19519377e+00 -1.82062000e-01 3.42879593e-01 6.42819643e-01 7.84641325e-01 -8.04550529e-01 6.23984516e-01 5.47999859e-01 6.20162964e-01 -8.63254607e-01 5.15753210e-01 4.79206532e-01 -5.52305639e-01 -5.86862564e-01 -4.57814068e-01 -8.80987108e-01 1.60826027e-01 8.09259713e-01 -3.91552478e-01 5.83647788e-01 7.27294803e-01 8.02685857e-01 -3.07035327e-01 9.38061416e-01 -2.43336543e-01 1.33329093e+00 1.03844665e-01 1.19927406e-01 2.71971554e-01 -6.21357560e-01 4.86029208e-01 8.87536526e-01 3.46992075e-01 3.15181345e-01 -3.00923914e-01 5.07017732e-01 2.27577075e-01 3.58788639e-01 -1.03893805e+00 -1.49386346e-01 6.20237589e-01 6.76520228e-01 -5.71980536e-01 -4.52719063e-01 -1.09691954e+00 -2.67588533e-02 -3.57061215e-02 6.66650951e-01 -7.55271852e-01 6.30548075e-02 5.34658611e-01 4.22067970e-01 -1.57611609e-01 -2.55724549e-01 -4.35960919e-01 -9.03567731e-01 -4.21888828e-01 -8.62815857e-01 5.56396782e-01 -2.87365824e-01 -1.07840049e+00 -5.06758571e-01 6.50785863e-01 -7.41530418e-01 -2.68143237e-01 -3.18639040e-01 -4.24184144e-01 1.05277455e+00 -1.29355848e+00 -6.76550508e-01 4.73301739e-01 2.50461511e-02 5.59529841e-01 -5.39253000e-03 4.41204578e-01 1.58454403e-01 -8.68817687e-01 3.16230088e-01 4.86161590e-01 -3.97887915e-01 1.09905493e+00 -1.08936346e+00 6.25435561e-02 7.31127024e-01 -7.40681648e-01 8.59517694e-01 7.51922607e-01 -8.14224243e-01 -6.88465476e-01 -6.17935300e-01 1.15682840e+00 -3.49240214e-01 8.99546742e-01 -9.55187678e-02 -5.43685079e-01 1.13338220e+00 5.27456343e-01 -6.82816267e-01 7.15315819e-01 3.83166552e-01 4.15105820e-02 -1.10613897e-01 -1.00301600e+00 7.38645136e-01 7.60634959e-01 -1.60705969e-01 -5.22295892e-01 2.46506706e-01 6.67579472e-01 -1.40355127e-02 -1.40718699e+00 5.47874153e-01 8.03765774e-01 -1.20490551e+00 8.09652627e-01 -6.52150750e-01 2.17726752e-01 4.94212866e-01 3.02499592e-01 -1.34711325e+00 -7.57110596e-01 -6.22673512e-01 4.56267953e-01 1.31558692e+00 4.51837957e-01 -1.37606561e+00 5.59917055e-02 4.75432009e-01 -1.31711558e-01 -6.17099464e-01 -7.57107556e-01 -1.00423217e+00 7.34516323e-01 -1.50778377e-03 8.79643321e-01 1.36537981e+00 1.13929406e-01 -2.12658420e-01 -8.03769678e-02 -3.41151059e-01 3.36187243e-01 5.15890121e-03 8.93677056e-01 -1.40128410e+00 3.46629284e-02 -9.88245070e-01 9.88498926e-02 -4.18042302e-01 5.99350691e-01 -1.52096957e-01 -5.47703326e-01 -1.73211908e+00 4.95390922e-01 -5.49736321e-01 -3.58100504e-01 -5.66977113e-02 -3.79176855e-01 -4.66212988e-01 1.19599469e-01 1.45257726e-01 1.24454178e-01 1.08368196e-01 1.10891128e+00 2.71871746e-01 -3.40155661e-01 6.02082551e-01 -1.16285443e+00 6.42693162e-01 1.05392110e+00 -6.20200574e-01 -1.59645289e-01 2.57658493e-02 1.04175553e-01 5.29507875e-01 2.61182368e-01 -4.42953080e-01 -3.22308421e-01 -1.17623544e+00 2.34729722e-02 -4.31573808e-01 -2.96136200e-01 -7.05922723e-01 4.38402444e-01 5.70919037e-01 -3.99068534e-01 6.38042629e-01 3.39094758e-01 2.65897602e-01 -3.82930338e-02 -2.44008675e-01 4.92400974e-01 -6.10712580e-02 2.59011444e-02 1.15188546e-01 -8.96744132e-01 2.99347974e-02 1.06468463e+00 -6.30202070e-02 -3.74612421e-01 -6.35469317e-01 -1.12093419e-01 -2.58848257e-03 7.84994304e-01 2.72570580e-01 -2.30109483e-01 -1.22854424e+00 -7.93910623e-01 -5.67546964e-01 -2.07892463e-01 -7.23935902e-01 1.59708589e-01 1.60389268e+00 4.59778272e-02 9.44542944e-01 -2.84469873e-03 1.41938537e-01 -9.78733122e-01 9.79321837e-01 8.37283954e-02 -3.65810126e-01 -3.51938367e-01 3.04353833e-01 4.06184345e-01 1.24892667e-01 -2.04658404e-01 -3.92340899e-01 -5.86868882e-01 3.08819205e-01 3.35065812e-01 9.62902308e-01 -5.67820430e-01 -6.47768915e-01 -4.58114624e-01 3.49217951e-01 3.96853060e-01 -5.44517815e-01 1.29873443e+00 -4.77433234e-01 -2.39969924e-01 1.13560736e+00 8.28624845e-01 3.80239457e-01 -1.02742481e+00 2.21756124e-03 7.09857792e-02 -5.12020707e-01 8.12032223e-02 -4.58919793e-01 -7.17546225e-01 2.34612137e-01 2.45506182e-01 6.50745988e-01 1.19664848e+00 -4.15963024e-01 -2.75420725e-01 -4.78744447e-01 2.82846928e-01 -1.48882663e+00 -5.95464349e-01 6.81846738e-02 9.57447410e-01 -1.17722273e+00 4.69588757e-01 -2.72305965e-01 -4.09359783e-01 6.23611152e-01 5.86946383e-02 -3.61225426e-01 8.29412758e-01 -1.71474978e-01 -3.21502417e-01 1.48843512e-01 -8.05549800e-01 -2.53592599e-02 1.21874236e-01 7.61038423e-01 8.02984238e-01 5.23095369e-01 -1.80488944e+00 5.40022135e-01 -2.38849297e-01 -1.21655554e-01 8.63522530e-01 5.78354001e-01 -1.21873274e-01 -1.20279253e+00 -8.58573318e-01 9.46312129e-01 -1.14479184e+00 3.69906053e-02 -2.49641836e-01 1.52435088e+00 -1.22607745e-01 1.32518244e+00 4.27554220e-01 3.95988256e-01 2.76960492e-01 -4.76862937e-02 2.11350694e-01 -4.05821562e-01 -8.79434943e-01 3.30522180e-01 4.44886535e-01 -2.16052741e-01 -1.04315174e+00 -1.30980730e+00 -5.08624673e-01 -6.69305205e-01 -3.92274261e-01 3.30094039e-01 5.99485099e-01 8.82142186e-01 -2.30545864e-01 4.87735361e-01 1.09271967e+00 -4.09328312e-01 -6.35032356e-01 -1.32870483e+00 -7.32991278e-01 7.83980265e-02 5.11823595e-01 -1.06417096e+00 -7.61770368e-01 -4.14995462e-01]
[7.955697059631348, 5.194067478179932]
fbc61c07-c19b-4d4d-b379-34f262c27cb3
a-distance-geometric-method-for-recovering
2301.02051
null
https://arxiv.org/abs/2301.02051v2
https://arxiv.org/pdf/2301.02051v2.pdf
A Distance-Geometric Method for Recovering Robot Joint Angles From an RGB Image
Autonomous manipulation systems operating in domains where human intervention is difficult or impossible (e.g., underwater, extraterrestrial or hazardous environments) require a high degree of robustness to sensing and communication failures. Crucially, motion planning and control algorithms require a stream of accurate joint angle data provided by joint encoders, the failure of which may result in an unrecoverable loss of functionality. In this paper, we present a novel method for retrieving the joint angles of a robot manipulator using only a single RGB image of its current configuration, opening up an avenue for recovering system functionality when conventional proprioceptive sensing is unavailable. Our approach, based on a distance-geometric representation of the configuration space, exploits the knowledge of a robot's kinematic model with the goal of training a shallow neural network that performs a 2D-to-3D regression of distances associated with detected structural keypoints. It is shown that the resulting Euclidean distance matrix uniquely corresponds to the observed configuration, where joint angles can be recovered via multidimensional scaling and a simple inverse kinematics procedure. We evaluate the performance of our approach on real RGB images of a Franka Emika Panda manipulator, showing that the proposed method is efficient and exhibits solid generalization ability. Furthermore, we show that our method can be easily combined with a dense refinement technique to obtain superior results.
['Ivan Petrović', 'Ivan Marković', 'Filip Marić', 'Ivan Bilić']
2023-01-05
null
null
null
null
['motion-planning']
['robots']
[ 4.12827849e-01 1.13874517e-01 2.23550275e-01 7.97688738e-02 -4.46634650e-01 -6.68349802e-01 3.23228270e-01 1.79083839e-01 -7.07763791e-01 6.02106333e-01 -4.97775882e-01 -2.48623028e-01 -6.21254623e-01 -5.90172350e-01 -1.02012765e+00 -7.67435014e-01 -4.06825632e-01 6.17957532e-01 2.16090679e-01 -5.57001948e-01 5.47685623e-01 9.23319817e-01 -1.46600044e+00 -5.17138660e-01 4.47523206e-01 9.75340784e-01 6.11731589e-01 8.91689777e-01 4.45202947e-01 3.30685616e-01 -4.77215856e-01 3.61478180e-01 6.75097704e-01 -8.76766443e-02 -7.63380468e-01 1.98793933e-01 6.14624470e-02 -5.07066369e-01 -4.10682708e-01 8.71187747e-01 2.67700434e-01 2.73444891e-01 4.80220139e-01 -1.12001038e+00 -8.31255168e-02 -2.97724940e-02 -3.47586185e-01 -2.03682512e-01 6.19192839e-01 2.56706998e-02 6.83891416e-01 -7.01841772e-01 8.46127093e-01 8.57380629e-01 7.89869010e-01 1.46838754e-01 -1.13200617e+00 -2.57454574e-01 -3.14634323e-01 1.97307587e-01 -1.49212229e+00 -3.25319529e-01 6.56534970e-01 -2.82460362e-01 1.01848865e+00 -1.06821753e-01 6.90046132e-01 7.35836923e-01 5.19734800e-01 5.89484647e-02 6.91902757e-01 -5.42132974e-01 3.64975423e-01 -2.91432112e-01 -3.95017207e-01 9.00302052e-01 4.34553593e-01 7.92131200e-02 -4.11372095e-01 2.07419693e-02 1.11312699e+00 1.85233548e-01 -4.67589259e-01 -9.70215082e-01 -1.48576283e+00 5.16978383e-01 7.68914938e-01 3.07186637e-02 -5.84244132e-01 2.73014784e-01 -1.70626327e-01 6.49006069e-01 -2.20565125e-01 8.10403943e-01 -3.54831457e-01 -3.07471305e-01 -3.67528111e-01 1.64549634e-01 9.81405199e-01 1.20291829e+00 8.38488162e-01 -9.61078554e-02 1.01649523e+00 3.62160146e-01 2.29089394e-01 6.40856087e-01 2.73199022e-01 -1.29338205e+00 5.58441877e-01 3.82754147e-01 4.54677790e-01 -1.10607004e+00 -8.49362612e-01 5.19553982e-02 -6.17929637e-01 6.92797482e-01 3.68914157e-01 -1.36814460e-01 -7.64297843e-01 1.37311637e+00 3.44105244e-01 -3.39846939e-01 3.14691007e-01 1.00712097e+00 -1.59595191e-01 3.25165838e-01 -6.86568916e-01 -5.60275987e-02 8.91367555e-01 -3.49071026e-01 -2.92356342e-01 -2.26904795e-01 4.73419547e-01 -5.69644094e-01 5.80291629e-01 7.55418956e-01 -8.66673529e-01 -1.94543332e-01 -1.47495723e+00 -2.58013420e-02 -1.23145454e-01 -1.59009218e-01 4.14473116e-01 1.44269422e-01 -8.73001277e-01 9.98797357e-01 -1.34533942e+00 -6.24891162e-01 -1.96294725e-01 8.45178068e-01 -8.74297857e-01 -9.15257633e-02 -6.29818499e-01 1.38661957e+00 4.87641424e-01 4.71584678e-01 -7.17367589e-01 -5.31790704e-02 -8.50004017e-01 -2.64284492e-01 6.04200721e-01 -5.74176371e-01 1.14666915e+00 -3.03442955e-01 -1.63468540e+00 1.52653813e-01 1.60750344e-01 -2.63569236e-01 4.51544106e-01 -3.11084539e-01 2.37462312e-01 5.85868835e-01 -8.47279131e-02 3.74901265e-01 8.70333076e-01 -1.32212090e+00 -4.69564587e-01 -6.66669667e-01 2.37117246e-01 4.21059370e-01 -9.51207504e-02 -7.87042379e-01 -1.37582153e-01 -7.50474706e-02 9.28040862e-01 -1.37316442e+00 -4.57897544e-01 4.14701939e-01 -2.65978187e-01 2.41140813e-01 5.46287954e-01 -5.75426757e-01 3.51433724e-01 -1.91422105e+00 8.32906306e-01 5.64003348e-01 -6.00114614e-02 -2.44782984e-01 -3.08425054e-02 8.19258571e-01 2.30744675e-01 -3.04890096e-01 -2.62983233e-01 -1.11451127e-01 -2.54891604e-01 6.59971297e-01 -6.62152544e-02 1.02074885e+00 8.34370703e-02 2.46625528e-01 -8.46083879e-01 -6.56196401e-02 2.01493025e-01 4.69061166e-01 -3.68683934e-01 1.78490445e-01 5.69684282e-02 6.99709117e-01 -6.16049349e-01 6.53076291e-01 2.13024616e-01 2.00860977e-01 1.69700354e-01 -6.81703240e-02 -3.56607765e-01 2.02380374e-01 -1.35603619e+00 2.19465423e+00 -5.83182514e-01 4.79813367e-01 4.78382796e-01 -1.01542222e+00 9.65484381e-01 2.97659397e-01 6.28904402e-01 -5.07879257e-01 2.60754466e-01 5.45863330e-01 -2.06888076e-02 -5.78047216e-01 7.48973310e-01 -1.70047715e-01 -1.84068739e-01 2.75581002e-01 -1.77066848e-02 -6.14986539e-01 -2.10180670e-01 -1.39935404e-01 1.37274420e+00 4.58161831e-01 4.56162274e-01 5.19069806e-02 2.00936019e-01 3.97904694e-01 1.87257692e-01 5.25337040e-01 1.98383749e-01 4.79579121e-01 2.96495944e-01 -2.47329146e-01 -1.46928859e+00 -1.05876172e+00 8.39598197e-03 4.02024359e-01 5.33950210e-01 1.08343683e-01 -3.66715819e-01 2.14569476e-02 1.81008011e-01 2.31521428e-01 -4.78360265e-01 -3.14938962e-01 -9.20886636e-01 -1.77171633e-01 2.43424833e-01 4.38822567e-01 2.39494428e-01 -7.20945537e-01 -1.20647824e+00 2.13567555e-01 8.68220404e-02 -1.12818182e+00 2.75304943e-01 5.23879945e-01 -1.21369934e+00 -1.26544583e+00 -4.08237696e-01 -7.52589881e-01 8.47435474e-01 6.06848776e-01 3.64168644e-01 1.38438523e-01 -3.51008266e-01 5.73450208e-01 -5.35831630e-01 -1.64160561e-02 -3.52827251e-01 4.95460769e-03 4.67976451e-01 -5.18537223e-01 -3.14869583e-01 -8.44822943e-01 -4.73192751e-01 3.06354135e-01 -6.50697410e-01 -2.84052342e-01 8.16735864e-01 6.95300996e-01 5.19525111e-01 7.75072500e-02 2.79939055e-01 -1.08748138e-01 3.47640634e-01 -3.85363638e-01 -7.61453092e-01 -2.05716938e-01 -4.73323822e-01 1.91606164e-01 5.52841723e-01 -2.76181012e-01 -6.15792274e-01 6.25613987e-01 1.10930894e-02 -4.10521209e-01 -1.40475467e-01 6.77830637e-01 4.85720411e-02 -5.51046133e-01 6.92541599e-01 2.10934356e-01 4.58489716e-01 -5.32515824e-01 3.05074334e-01 7.10038066e-01 8.69790852e-01 -4.73067224e-01 8.41352046e-01 6.12576842e-01 6.25953376e-01 -1.11267841e+00 -9.66947302e-02 -5.87028503e-01 -1.19395864e+00 -1.73760295e-01 5.36038399e-01 -6.94076836e-01 -9.99950647e-01 2.72402078e-01 -1.24051547e+00 -2.14346871e-01 -5.16089238e-02 7.09965408e-01 -1.02797186e+00 6.79714799e-01 -3.86923164e-01 -8.37224066e-01 9.12085176e-02 -1.17675626e+00 1.14028215e+00 -1.61146790e-01 -1.56808943e-01 -4.88742113e-01 5.97203448e-02 -5.41258371e-04 1.14047276e-02 5.76219559e-01 5.50929606e-01 -2.96194643e-01 -8.09620619e-01 -5.67613006e-01 2.90371925e-01 -5.24443835e-02 1.98760465e-01 -2.94393510e-01 -4.13457394e-01 -6.20976925e-01 2.66144931e-01 -2.96726376e-01 4.20038164e-01 -1.24096014e-01 5.00068367e-01 -9.23520774e-02 -3.78054559e-01 5.40853560e-01 1.52270234e+00 1.40568942e-01 4.25570875e-01 8.53661358e-01 6.54803813e-01 6.74861014e-01 8.98454666e-01 4.72362518e-01 3.47604871e-01 7.39963353e-01 1.14238036e+00 4.38899517e-01 3.27342987e-01 -3.42541770e-03 2.36943588e-01 7.44484484e-01 -5.26747704e-01 9.44870263e-02 -8.48755658e-01 4.37846839e-01 -1.72988307e+00 -4.18648601e-01 -3.93484719e-02 2.55336809e+00 5.10881603e-01 3.12195607e-02 -1.84962720e-01 4.44705814e-01 3.38316679e-01 -4.39352483e-01 -7.82232225e-01 -2.79687196e-01 4.11508650e-01 6.16645738e-02 9.23314214e-01 5.78360140e-01 -7.14537859e-01 5.49495041e-01 5.74500942e+00 -1.02121145e-01 -9.36111510e-01 -2.77129024e-01 -6.63853228e-01 -3.28452848e-02 2.54839510e-01 -1.14452960e-02 -3.25807512e-01 4.12721112e-02 9.68269944e-01 1.76834792e-01 6.15431905e-01 8.42628658e-01 -7.32631609e-02 -5.74280798e-01 -1.21401012e+00 6.28146768e-01 2.08959803e-02 -9.44147229e-01 -4.35128391e-01 3.46078664e-01 1.93341285e-01 1.17139384e-01 -2.60961920e-01 -1.85809985e-01 5.23347035e-02 -6.71147943e-01 8.38506341e-01 5.05728841e-01 6.36836290e-01 -7.63032019e-01 5.97744346e-01 8.51007938e-01 -9.68588591e-01 -3.38060915e-01 -5.53153872e-01 -4.66913193e-01 2.74865597e-01 -2.29989737e-02 -1.24840653e+00 6.83813751e-01 5.51740468e-01 4.87106144e-01 -9.74598601e-02 1.15242696e+00 -2.46718138e-01 -2.37352595e-01 -6.92803979e-01 -2.09475994e-01 1.02190882e-01 -1.92139149e-01 7.84198642e-01 5.03299296e-01 7.83689141e-01 2.65375882e-01 9.42499340e-02 3.48538011e-01 3.22746634e-01 -3.41823280e-01 -8.79095972e-01 2.64560461e-01 4.55329269e-01 9.88827825e-01 -8.32130909e-01 1.18161477e-01 -1.64242432e-01 1.01340377e+00 3.98465097e-01 1.79716632e-01 -2.88243890e-01 -5.56124806e-01 4.93099838e-01 -1.29170299e-01 4.80143070e-01 -1.19389844e+00 -1.47329122e-01 -9.18526888e-01 2.69240260e-01 -6.10941350e-01 -9.29630473e-02 -9.15040731e-01 -6.50439203e-01 5.11020303e-01 1.18404096e-02 -1.64926052e+00 -6.80925429e-01 -8.55650365e-01 -4.35975976e-02 7.64641464e-01 -1.30562556e+00 -7.05520988e-01 -4.29602653e-01 6.06025219e-01 2.79215783e-01 1.60133600e-01 1.08979785e+00 -1.53939262e-01 -1.64670553e-02 -5.74440882e-02 1.34017393e-01 -1.51173428e-01 4.31444496e-01 -1.24086547e+00 1.95086166e-01 6.13709092e-01 -5.15467487e-02 7.01813519e-01 1.04623973e+00 -6.75419033e-01 -2.28165913e+00 -5.81302822e-01 3.32807750e-01 -3.38395029e-01 8.54144037e-01 -2.17522591e-01 -7.12234020e-01 8.57547283e-01 -2.87499368e-01 -1.33804783e-01 1.08168557e-01 -2.41819292e-01 -5.52960224e-02 4.65785749e-02 -1.15387654e+00 3.56444567e-01 9.73582268e-01 -3.54829490e-01 -8.84793222e-01 4.32175875e-01 5.84656537e-01 -8.17161679e-01 -1.01445806e+00 3.34522814e-01 8.32489789e-01 -6.90688312e-01 1.02468586e+00 -2.98104674e-01 1.66352436e-01 -5.09204805e-01 -2.69765347e-01 -1.29955471e+00 -3.76226231e-02 -6.32200301e-01 2.75908187e-02 4.42756116e-01 1.37366816e-01 -6.59670472e-01 7.01485515e-01 3.96735221e-01 -2.75561333e-01 -4.93246138e-01 -1.39414167e+00 -9.56285477e-01 -2.08645806e-01 -3.85110676e-01 1.92867920e-01 5.16928256e-01 3.23654801e-01 8.07739422e-02 -3.07931691e-01 8.51597011e-01 5.74471533e-01 1.74451098e-01 9.86359000e-01 -1.28874385e+00 -3.10903519e-01 -1.52502544e-02 -9.38319385e-01 -1.18762088e+00 2.13529244e-02 -4.21575189e-01 6.53478980e-01 -1.33588016e+00 -4.49173927e-01 -5.38673222e-01 1.94246247e-01 4.04157251e-01 5.87019324e-01 2.39023075e-01 1.85152635e-01 4.86857265e-01 -2.56167412e-01 4.45909113e-01 1.13980842e+00 2.47056574e-01 -2.56405681e-01 9.17056352e-02 -3.41120735e-02 8.54801059e-01 6.95028305e-01 -4.10736203e-01 -1.18404977e-01 -5.86313426e-01 4.40510333e-01 5.22043288e-01 4.76538092e-01 -1.23448360e+00 6.27399445e-01 8.00695177e-03 2.25530937e-01 -3.14362645e-01 8.15420747e-01 -1.33830297e+00 1.00668177e-01 8.15302789e-01 -6.46235123e-02 3.23138297e-01 -2.17095558e-02 9.86104906e-01 4.25755829e-02 -5.19191921e-01 2.38984525e-01 -2.22089991e-01 -8.01571250e-01 -6.17458597e-02 -4.77723807e-01 -7.06905246e-01 9.01005924e-01 -4.48193043e-01 -7.77806342e-02 -3.46313983e-01 -8.46596301e-01 1.31380230e-01 8.66956413e-01 2.63516188e-01 7.95692623e-01 -9.95499492e-01 -2.22574651e-01 2.83774137e-01 5.52928858e-02 4.38282251e-01 -1.02026239e-01 1.01251972e+00 -1.03563082e+00 2.83315480e-01 -4.68296289e-01 -8.64328682e-01 -1.06445563e+00 3.98066729e-01 1.90148383e-01 3.50194037e-01 -9.54484463e-01 5.23306429e-01 -4.98249978e-01 -4.71563548e-01 1.22120574e-01 -5.32139361e-01 9.53208208e-02 -3.24581861e-01 2.33104214e-01 3.55265170e-01 2.45925158e-01 -8.15380454e-01 -3.45590532e-01 7.54508555e-01 3.86275381e-01 -2.65769064e-01 1.59048045e+00 -3.69839489e-01 -1.89629033e-01 4.16878432e-01 1.11803889e+00 -2.04660848e-01 -1.49523449e+00 -6.24149404e-02 2.31151711e-02 -5.13917565e-01 -2.06114024e-01 -2.47815281e-01 -6.03418946e-01 8.15264702e-01 3.27783495e-01 1.99779436e-01 1.02351773e+00 -7.19312280e-02 5.29726505e-01 1.31609595e+00 1.15898085e+00 -8.71103883e-01 2.52841529e-03 5.78508794e-01 1.09983861e+00 -1.17191339e+00 3.03768575e-01 -2.98721015e-01 -1.39517054e-01 1.59022653e+00 5.11668250e-02 -4.31719482e-01 4.06594723e-01 2.19719797e-01 4.97548236e-03 -4.18441556e-02 -3.64181131e-01 -7.53130466e-02 -1.51618198e-02 6.60148501e-01 -1.51318669e-01 -6.94506839e-02 1.39160141e-01 -7.08805695e-02 -5.24204791e-01 -2.68950194e-01 9.82037127e-01 1.61707795e+00 -9.20184970e-01 -8.84068370e-01 -5.83473444e-01 -5.88683970e-02 -3.31694447e-02 3.62353504e-01 -1.96343616e-01 1.17097473e+00 -1.71154648e-01 8.45933676e-01 -2.35226541e-03 -4.81299311e-01 5.64878404e-01 -8.40392187e-02 8.84102583e-01 -6.26834810e-01 2.33635958e-02 2.99056619e-02 -6.20677788e-03 -9.18201685e-01 -4.68006909e-01 -8.27807367e-01 -1.68640471e+00 4.66393754e-02 -3.61186415e-01 -5.44360653e-02 1.28323925e+00 1.09095609e+00 1.61476389e-01 3.79417092e-01 4.71347779e-01 -1.49781191e+00 -8.42180848e-01 -8.20772707e-01 -7.00930119e-01 1.76160075e-02 8.31606269e-01 -9.73899901e-01 -2.67873764e-01 -2.96621043e-02]
[6.996329307556152, -1.8138326406478882]
1a7238f8-25aa-4f25-9975-30c41939ab37
identification-and-classification-of-1
2003.08209
null
https://arxiv.org/abs/2003.08209v1
https://arxiv.org/pdf/2003.08209v1.pdf
Identification and Classification of Phenomena in Multispectral Satellite Imagery Using a New Image Smoother Method and its Applications in Environmental Remote Sensing
In this paper a new method of image smoothing for satellite imagery and its applications in environmental remote sensing are presented. This method is based on the global gradient minimization over the whole image. With respect to the image discrete identity, the continuous minimization problem is discretized. Using the finite difference numerical method of differentiation, a simple yet efficient 5*5-pixel template is derived. Convolution of the derived template with the image in different bands results in the discrimination of various image elements. This method is extremely fast, besides being highly precise. A case study is presented for the northern Iran, covering parts of the Caspian Sea. Comparison of the method with the usual Laplacian template reveals that it is more capable of distinguishing phenomena in the image.
['M. Kiani']
2020-03-17
null
null
null
null
['image-smoothing']
['computer-vision']
[ 6.16917193e-01 -3.14007849e-01 5.77077866e-01 -3.66408616e-01 -6.59246981e-01 -3.23497772e-01 4.31459188e-01 -3.23178172e-01 -7.15585709e-01 8.41808796e-01 -2.06143349e-01 -2.70733118e-01 -3.09248865e-01 -9.29081976e-01 -5.15287369e-02 -1.28815329e+00 -1.44342244e-01 -1.45850897e-01 1.20817460e-02 -2.65744209e-01 2.73169309e-01 9.67385471e-01 -1.38756859e+00 -1.57791346e-01 1.27613306e+00 6.07609212e-01 4.92937177e-01 5.99094629e-01 -8.55197385e-02 8.51190537e-02 -5.66873610e-01 1.75308853e-01 4.86513734e-01 -5.00406742e-01 -6.32173836e-01 8.18530619e-01 6.65448725e-01 -4.72421236e-02 1.71217337e-01 1.52597797e+00 6.44433975e-01 5.35257518e-01 9.13488328e-01 -3.38128239e-01 -3.47996503e-01 1.18159642e-02 -9.31572914e-01 1.63969517e-01 -9.22096670e-02 -1.10304937e-01 4.14131433e-01 -1.03781843e+00 5.29025674e-01 1.14869475e+00 1.02149713e+00 -1.57528192e-01 -1.53983903e+00 -1.27716631e-01 -2.05979526e-01 -1.52681679e-01 -1.83456838e+00 -1.35448813e-01 5.04170537e-01 -3.38134080e-01 5.99070668e-01 7.18545794e-01 6.83001637e-01 -3.28916550e-01 2.61087239e-01 1.14740185e-01 1.72138369e+00 -6.52703464e-01 -2.54120566e-02 9.65335518e-02 1.45296440e-01 4.34873372e-01 4.53776836e-01 2.78624147e-02 3.16146523e-01 -2.24635974e-01 6.39365911e-01 -5.71960695e-02 -4.97938812e-01 2.84776598e-01 -6.10912800e-01 8.49025130e-01 4.66156423e-01 7.08824873e-01 -6.46382332e-01 -4.09750164e-01 -4.56661507e-02 1.17431119e-01 1.00160885e+00 5.43683656e-02 -8.00162554e-02 6.54479146e-01 -1.50242388e+00 1.95868060e-01 6.14254653e-01 1.00494139e-01 1.04319489e+00 5.65296769e-01 6.94265068e-02 6.24306023e-01 3.72410268e-01 1.06677711e+00 1.19793721e-01 -8.49212468e-01 -6.05330281e-02 4.04250413e-01 2.24689767e-01 -1.06539023e+00 -3.34945172e-01 -3.43585193e-01 -8.76135886e-01 1.08419764e+00 3.87709200e-01 -3.80500048e-01 -9.75576937e-01 9.83927906e-01 4.88859296e-01 1.32006004e-01 2.14239419e-01 7.16703236e-01 5.57338238e-01 1.05401778e+00 8.61392096e-02 -3.98904502e-01 1.27782869e+00 -4.50942427e-01 -7.82957613e-01 2.64553949e-02 2.18951061e-01 -8.33514750e-01 6.31803632e-01 3.41340154e-01 -7.46615350e-01 -5.81224084e-01 -7.86194384e-01 2.27803037e-01 -6.21489644e-01 5.42890370e-01 3.01026970e-01 7.14211762e-01 -1.13693678e+00 7.78924823e-01 -4.07084793e-01 -4.70105678e-01 4.43988061e-03 2.20288888e-01 -2.34365597e-01 4.29716855e-01 -8.90492499e-01 1.00944519e+00 3.41923714e-01 5.52497745e-01 -1.89974412e-01 -3.22598338e-01 -8.18937600e-01 2.56840307e-02 -3.12703967e-01 -6.69666287e-03 6.65157497e-01 -1.62801647e+00 -1.31826806e+00 9.91856277e-01 -3.63138318e-01 -3.25157583e-01 6.43070459e-01 8.26571882e-02 -6.69563174e-01 1.71833515e-01 2.05455005e-01 8.79094843e-03 1.17307830e+00 -1.20172882e+00 -7.02364683e-01 -4.47919250e-01 -6.86507285e-01 2.55183935e-01 1.50069699e-01 -3.68850194e-02 8.12534802e-03 -6.23785138e-01 3.04724962e-01 -6.81486368e-01 -5.15583098e-01 -5.88607788e-02 -8.64561647e-02 2.41930291e-01 9.70860183e-01 -1.21597433e+00 1.11050439e+00 -2.50883555e+00 -4.43927288e-01 8.34346592e-01 -3.36801052e-01 4.19227660e-01 2.13488820e-03 3.84021401e-01 -1.59851834e-01 -1.75918941e-03 -8.49234521e-01 2.47044235e-01 -3.80051851e-01 8.91215205e-02 -1.06167465e-01 1.05742478e+00 5.01777371e-03 2.66449094e-01 -5.95286429e-01 -5.54704964e-01 3.57887626e-01 7.98097253e-01 1.74279466e-01 -3.59321803e-01 3.73069882e-01 4.51598763e-01 -3.59061539e-01 4.08268750e-01 1.49414718e+00 1.62427127e-01 1.79196820e-01 -2.36017302e-01 -1.01046371e+00 -4.18754011e-01 -1.40238988e+00 9.60754633e-01 -7.80478865e-02 7.92994380e-01 6.59537733e-01 -1.03313291e+00 1.21263075e+00 3.37737918e-01 4.47118223e-01 -4.28001881e-01 -5.21483459e-02 2.85425156e-01 -2.20279425e-01 -6.13592863e-01 5.17548263e-01 -3.22680622e-01 6.36698186e-01 3.19034494e-02 -5.78204751e-01 -4.57281768e-01 3.08022439e-01 -1.72970235e-01 -4.13170923e-03 1.68861747e-01 6.29377306e-01 -9.50531304e-01 9.67819333e-01 5.09023666e-01 2.41065294e-01 5.56672812e-01 -9.68593210e-02 2.22108766e-01 -1.67827144e-01 -3.82034361e-01 -7.75157869e-01 -7.52589703e-01 -6.68588698e-01 5.28725326e-01 4.28208783e-02 4.02242482e-01 -1.03685105e+00 1.05096428e-02 2.07463428e-01 6.12400472e-01 -5.29224992e-01 5.85337400e-01 -2.70070732e-01 -1.23001707e+00 4.27271813e-01 -2.56909251e-01 1.20507789e+00 -8.03575516e-01 -6.04680598e-01 2.82596588e-01 -8.17371756e-02 -5.86715162e-01 -2.36346021e-01 -2.56475329e-01 -1.52484941e+00 -7.15560377e-01 -1.07423437e+00 -7.81845212e-01 9.56295252e-01 7.63851523e-01 7.75867820e-01 4.27630357e-02 -3.66851687e-01 1.27836570e-01 3.72554921e-02 -3.38586718e-01 -3.89903605e-01 -5.25624096e-01 -4.89652753e-01 3.27793658e-01 4.02550340e-01 -3.21800143e-01 -4.27970678e-01 3.46863829e-02 -9.00395632e-01 -4.61606175e-01 5.91987848e-01 7.74544239e-01 7.38874137e-01 7.39782631e-01 3.96236666e-02 -9.60232437e-01 6.40190959e-01 -8.26351270e-02 -1.10358214e+00 3.40275317e-01 -6.36614144e-01 -2.85940260e-01 9.91972685e-02 -1.33247795e-02 -1.72423410e+00 5.08778691e-01 1.12041555e-01 4.40821618e-01 -3.22822452e-01 6.80244029e-01 1.10839635e-01 -7.37574100e-01 9.78527963e-01 4.47448730e-01 4.16502833e-01 -8.45303476e-01 1.60884365e-01 7.55150497e-01 6.99918032e-01 -3.04368939e-02 7.84645081e-01 1.03336883e+00 2.67512292e-01 -1.82147884e+00 -3.23901564e-01 -7.55555689e-01 -6.68119013e-01 -4.84520584e-01 1.09164369e+00 -7.71728218e-01 -2.50582874e-01 6.95814550e-01 -7.74105430e-01 -1.28088728e-01 -2.37563401e-01 6.42832816e-01 -1.51761383e-01 7.58247256e-01 -2.85937726e-01 -1.16239572e+00 -4.76488441e-01 -7.63620734e-01 7.27702856e-01 4.61750984e-01 2.63037205e-01 -1.32445192e+00 2.42282078e-01 -9.93650779e-02 5.27429760e-01 4.20915514e-01 3.37912768e-01 -1.27836719e-01 -1.21550649e-01 -2.59813190e-01 -3.89119864e-01 5.64510942e-01 4.60387915e-01 2.80540764e-01 -9.91595626e-01 -2.59178996e-01 4.20765251e-01 3.77760738e-01 9.92956102e-01 7.62935400e-01 3.94206941e-01 -1.23081952e-01 -1.89011872e-01 7.12683797e-01 2.14849520e+00 1.32122070e-01 1.05312216e+00 5.44320941e-01 1.55091405e-01 6.98880732e-01 7.19177067e-01 2.72582799e-01 -2.67697930e-01 2.81194150e-01 2.25773439e-01 -8.21768701e-01 3.88049148e-02 5.14663458e-01 3.58597755e-01 1.53792948e-01 -7.01634228e-01 3.12481195e-01 -6.65036440e-01 4.40248460e-01 -1.43169439e+00 -1.44399881e+00 -8.92612278e-01 2.30879402e+00 5.90554833e-01 -5.30035138e-01 -1.56121969e-01 2.81465203e-01 9.64426339e-01 1.69477463e-01 1.64943852e-03 -5.51252127e-01 -4.83290166e-01 2.82736123e-01 1.15357816e+00 1.06551933e+00 -1.40562856e+00 6.64641380e-01 6.83759594e+00 9.17995512e-01 -1.53744888e+00 2.40248460e-02 3.05640399e-01 7.77630925e-01 -2.66065751e-03 1.18991490e-02 -5.90406299e-01 2.63956785e-01 5.14303982e-01 -1.12805851e-01 5.93383461e-02 3.37062627e-01 1.04526365e+00 -1.00384080e+00 2.78529644e-01 5.58128119e-01 -1.40920863e-01 -9.19403076e-01 1.59531310e-01 2.31153354e-01 1.03239799e+00 -1.14108115e-01 -1.40600473e-01 -5.13406813e-01 -1.11127511e-01 -8.51862967e-01 4.17549700e-01 9.32188809e-01 7.40927577e-01 -6.68920755e-01 5.80974519e-01 2.19872802e-01 -1.28015375e+00 2.51231581e-01 -4.82894123e-01 -2.84066081e-01 1.21234603e-01 8.56640637e-01 -4.82415825e-01 7.06630230e-01 6.53843582e-01 4.26938683e-01 -3.78903687e-01 1.34194803e+00 -1.68687060e-01 6.02209628e-01 -5.21710336e-01 3.04068565e-01 5.47659099e-01 -1.33855057e+00 7.54760742e-01 1.68453217e+00 6.43136978e-01 4.33647782e-01 -6.10459223e-02 6.73292637e-01 7.11371362e-01 5.19666672e-01 -7.28422821e-01 1.02879606e-01 5.87840937e-02 1.24147332e+00 -8.44957769e-01 -5.73023200e-01 -5.10188997e-01 9.88303721e-01 -4.62821960e-01 7.80569017e-01 -1.74184635e-01 -6.69919908e-01 5.00451624e-01 -8.29539597e-02 3.35204005e-01 -2.30765149e-01 -3.96202803e-01 -7.67468691e-01 -3.71034861e-01 -7.52150357e-01 3.72687280e-01 -6.42225266e-01 -8.97338033e-01 5.58136702e-01 1.74762253e-02 -1.29418647e+00 1.91860855e-01 -5.83142936e-01 -9.63697672e-01 1.76020598e+00 -1.60540926e+00 -8.39094579e-01 -2.65636683e-01 7.01042414e-01 3.64869028e-01 1.57733425e-01 7.79184759e-01 1.71230093e-01 -2.52396286e-01 -3.30982625e-01 7.85925746e-01 -1.14146583e-01 2.52572626e-01 -1.18542099e+00 8.39776099e-02 1.21620727e+00 -3.08445960e-01 2.60067821e-01 1.00826836e+00 -8.36386681e-01 -6.19638920e-01 -9.31073964e-01 1.15455151e+00 5.43884575e-01 4.47575808e-01 4.37922031e-01 -1.18391883e+00 2.53779709e-01 3.63706350e-01 -3.42657924e-01 4.20085073e-01 -7.51700759e-01 2.83986598e-01 -2.13737756e-01 -1.58157873e+00 4.04522598e-01 9.57883671e-02 -3.03447813e-01 -5.75186431e-01 3.78789306e-01 -2.99860835e-01 2.95832977e-02 -7.19727874e-01 2.34053701e-01 4.24028873e-01 -9.33798671e-01 1.02371204e+00 -2.10885089e-02 -3.75704765e-01 -7.14592814e-01 9.01479945e-02 -1.14746428e+00 -4.97362643e-01 -5.46855569e-01 1.14370227e+00 9.71300960e-01 4.10129249e-01 -1.02165914e+00 4.38664675e-01 1.85334340e-01 2.48413995e-01 4.56267335e-02 -8.84054959e-01 -7.95186996e-01 -1.96373150e-01 1.31538540e-01 1.66294932e-01 8.24001849e-01 -4.42460716e-01 -1.53659448e-01 -3.38242620e-01 7.11508632e-01 1.18222249e+00 3.12173754e-01 3.84387076e-01 -1.58993804e+00 2.28985459e-01 -2.89854586e-01 -3.32429975e-01 -5.40941834e-01 -4.73708548e-02 -5.09376228e-01 1.08281828e-01 -1.70632637e+00 -2.45445415e-01 -8.44131485e-02 1.94136426e-01 1.36806190e-01 -2.28826627e-01 5.85800707e-01 -7.48139853e-03 3.75559926e-01 5.53898633e-01 1.30779697e-02 1.30706644e+00 -2.73627847e-01 -3.58337283e-01 2.75831670e-01 -1.96611822e-01 8.70200098e-01 1.07661307e+00 -3.41246814e-01 -2.13561282e-02 -1.24103926e-01 -2.65080899e-01 -2.31171072e-01 3.51657689e-01 -8.89264643e-01 -3.50712091e-02 -1.62688136e-01 2.95582980e-01 -5.60220778e-01 1.77669317e-01 -1.01020265e+00 7.61099637e-01 7.65967071e-01 3.30416232e-01 -2.06490695e-01 2.38467962e-01 3.00588697e-01 -5.23065448e-01 -7.78924227e-01 1.19602239e+00 -3.94984365e-01 -1.21691513e+00 -1.20248362e-01 -8.03837895e-01 -4.90838557e-01 8.85257065e-01 -6.94176018e-01 1.03669129e-01 -3.95934790e-01 -1.02566481e+00 -1.80237815e-01 3.45310569e-01 -7.01455235e-01 3.85005981e-01 -8.91003311e-01 -1.13967061e+00 1.37612015e-01 -5.17025113e-01 -5.06319880e-01 3.55855703e-01 1.15697777e+00 -1.19165528e+00 -5.04429312e-03 -2.26362929e-01 -4.83493119e-01 -1.62137115e+00 2.62442231e-02 7.55961776e-01 5.38762808e-02 -6.63513601e-01 3.28872740e-01 2.20580455e-02 -1.77204572e-02 -2.86314249e-01 -5.20415120e-02 -4.91362154e-01 2.98341423e-01 6.71012044e-01 8.00010920e-01 -1.14945516e-01 -1.16879570e+00 -5.91660216e-02 1.07558203e+00 7.82218933e-01 -4.02738959e-01 1.21062541e+00 -5.00139832e-01 -7.91629136e-01 2.29787380e-01 1.00126612e+00 4.49514776e-01 -1.26543248e+00 -1.04277141e-01 3.18191610e-02 -6.78771377e-01 4.69380677e-01 -5.28195202e-01 -9.74286139e-01 6.85907781e-01 9.26094770e-01 4.41388696e-01 1.54430902e+00 -7.89870977e-01 4.20941263e-02 4.22108561e-01 -1.66870981e-01 -1.07934129e+00 -1.01220477e+00 4.94778007e-01 8.97837937e-01 -1.07208002e+00 5.78141332e-01 -6.28905058e-01 -4.39225972e-01 1.45271516e+00 -3.78745556e-01 -2.29069561e-01 8.84882450e-01 1.50602475e-01 6.15575373e-01 -2.65957624e-01 3.86618197e-01 -7.02808797e-01 3.40032935e-01 6.54730678e-01 5.19804060e-01 1.52872041e-01 -1.12849236e+00 -4.39440489e-01 4.27727669e-01 2.96198502e-02 4.52571601e-01 8.24611008e-01 -1.05626690e+00 -7.69010127e-01 -1.17676091e+00 2.30918095e-01 -6.62116289e-01 -2.66357958e-01 -1.24670789e-01 9.56049800e-01 1.23984270e-01 8.61466527e-01 -4.82543707e-02 5.05255640e-01 2.99975246e-01 6.48648441e-02 1.21869385e-01 -9.05765295e-02 -5.66985369e-01 5.92807591e-01 1.25858933e-01 -1.48081303e-01 -1.04143643e+00 -8.94111395e-01 -1.33176410e+00 -9.34706405e-02 -4.46017951e-01 7.33226597e-01 9.56254721e-01 6.46067739e-01 -2.21668512e-01 -9.29881558e-02 7.09999382e-01 -1.29836047e+00 -6.77901089e-01 -9.42205966e-01 -1.50872636e+00 2.04206690e-01 3.51931721e-01 -2.61299074e-01 -7.73014486e-01 4.19180572e-01]
[10.126562118530273, -2.042008876800537]
efd9500c-0a62-4700-976d-6d594cc9bdfb
nlnde-at-cantemist-neural-sequence-labeling
2010.12322
null
https://arxiv.org/abs/2010.12322v1
https://arxiv.org/pdf/2010.12322v1.pdf
NLNDE at CANTEMIST: Neural Sequence Labeling and Parsing Approaches for Clinical Concept Extraction
The recognition and normalization of clinical information, such as tumor morphology mentions, is an important, but complex process consisting of multiple subtasks. In this paper, we describe our system for the CANTEMIST shared task, which is able to extract, normalize and rank ICD codes from Spanish electronic health records using neural sequence labeling and parsing approaches with context-aware embeddings. Our best system achieves 85.3 F1, 76.7 F1, and 77.0 MAP for the three tasks, respectively.
['Jannik Strötgen', 'Heike Adel', 'Xiang Dai', 'Lukas Lange']
2020-10-23
null
null
null
null
['clinical-concept-extraction']
['medical']
[ 3.17384988e-01 1.01820409e-01 -4.58253145e-01 -6.99412942e-01 -1.30556071e+00 -5.61634600e-01 3.18874955e-01 9.20404613e-01 -8.73294473e-01 8.63198817e-01 4.68142390e-01 -4.91579235e-01 -1.51186615e-01 -4.74307895e-01 -5.89227557e-01 -6.32560968e-01 7.22957999e-02 8.42987359e-01 -1.16667353e-01 7.57605061e-02 -1.20504841e-01 6.16186619e-01 -9.68156993e-01 6.66245341e-01 5.32709360e-01 9.78812218e-01 -4.09589380e-01 8.90215218e-01 -3.09626400e-01 9.64955628e-01 -4.69699889e-01 -7.28800952e-01 -3.58102262e-01 7.77703943e-03 -8.58264804e-01 -6.67829275e-01 3.02514464e-01 -1.87259048e-01 -2.33719110e-01 1.42340648e+00 5.63284814e-01 -2.62391269e-01 9.36609507e-01 -9.75809753e-01 -9.69925821e-01 1.01006186e+00 -3.03692013e-01 3.06736946e-01 9.57204178e-02 -7.41324946e-02 1.03025830e+00 -6.83559656e-01 9.29096997e-01 9.08368886e-01 9.63010788e-01 8.78265381e-01 -1.11277485e+00 -7.06718504e-01 -2.57548779e-01 -6.72373474e-02 -1.60613966e+00 -5.87898076e-01 4.29295562e-03 -5.30826151e-01 1.37409627e+00 3.61861736e-01 1.26215190e-01 1.13162911e+00 5.15739024e-01 4.54936564e-01 6.68760896e-01 -2.88889408e-01 5.71714044e-02 -2.18211666e-01 6.47914708e-01 8.09395969e-01 5.83321750e-01 -1.75293446e-01 6.09824620e-02 -4.75188404e-01 2.43371680e-01 2.20811680e-01 -8.08580071e-02 8.39292929e-02 -1.14880979e+00 7.40551412e-01 4.05484200e-01 5.36420524e-01 -5.45627415e-01 5.74180372e-02 6.94965482e-01 -7.37667363e-03 4.29778546e-01 5.60358524e-01 -7.45865822e-01 -7.93901086e-02 -6.52338922e-01 -1.38393477e-01 6.89243674e-01 9.99530077e-01 -2.67161578e-02 -5.77317834e-01 -4.90536243e-01 9.28011417e-01 1.02733985e-01 3.39997292e-01 6.99456573e-01 -4.88129556e-01 4.30925757e-01 6.57047987e-01 -4.10378516e-01 -6.96122229e-01 -1.06981289e+00 -2.36128688e-01 -1.33374429e+00 -4.79115546e-01 2.03125328e-01 -2.46534601e-01 -1.38946462e+00 1.92999864e+00 7.46899459e-04 -8.02016556e-02 4.78631884e-01 2.76433468e-01 1.38831198e+00 1.40019223e-01 8.03198934e-01 -3.91477719e-02 1.69802594e+00 -8.57772112e-01 -1.09522951e+00 2.07193866e-02 1.14072526e+00 -6.75065994e-01 2.99989522e-01 -1.59898818e-01 -9.62553918e-01 -1.31024733e-01 -7.83148706e-01 -5.63597500e-01 -6.76523745e-01 2.10822508e-01 6.72606170e-01 4.28320289e-01 -9.97703552e-01 4.35346335e-01 -1.28071940e+00 -6.29228473e-01 7.38552332e-01 5.89716434e-01 -7.48853743e-01 -3.07573318e-01 -1.03095818e+00 8.89804184e-01 5.78708410e-01 -3.54314774e-01 -5.34277856e-01 -9.40547168e-01 -9.88963425e-01 2.16915563e-01 -4.28273939e-02 -7.42497981e-01 1.37047970e+00 -2.65273213e-01 -1.03469896e+00 1.26815414e+00 -1.17865033e-01 -6.15760148e-01 1.41405165e-01 1.00220263e-01 -7.14682102e-01 -1.07170552e-01 1.55333072e-01 7.46136844e-01 -4.04373020e-01 -2.79537916e-01 -5.00941515e-01 -6.41425669e-01 -5.98153234e-01 -1.11057803e-01 -3.73139143e-01 2.36988753e-01 -4.38057721e-01 -6.32665813e-01 7.32185785e-04 -8.80710721e-01 -6.15172088e-01 -9.42260250e-02 -6.32312238e-01 -2.98881948e-01 -2.18781959e-02 -1.05133224e+00 1.19851708e+00 -2.21464729e+00 -8.86621401e-02 1.05936207e-01 3.93182874e-01 1.39907166e-01 -1.51283085e-01 6.27404153e-02 -4.05568123e-01 5.99241555e-01 -5.00120640e-01 -2.37248212e-01 3.90170254e-02 3.28312397e-01 2.78100669e-01 3.19616735e-01 2.97195464e-01 1.27288198e+00 -6.88649297e-01 -5.88524222e-01 -1.92955986e-01 4.25107718e-01 -6.26331568e-01 1.16866618e-01 1.59114584e-01 -7.61835724e-02 -3.31673175e-01 8.95945072e-01 6.11122012e-01 -4.19699371e-01 4.58820969e-01 -2.81118214e-01 1.84395686e-01 1.97188571e-01 -5.60233235e-01 1.93568730e+00 -3.15773934e-02 3.04900974e-01 1.16388053e-01 -7.37475932e-01 6.24229670e-01 3.19674700e-01 9.06562448e-01 -5.32654464e-01 4.84311432e-01 2.36652628e-01 7.31936768e-02 -6.53939545e-01 3.26290607e-01 -3.72131839e-02 -5.04237413e-01 1.15783229e-01 2.38231823e-01 2.82196999e-01 1.54435918e-01 1.66430742e-01 1.74214149e+00 -5.79155147e-01 1.03139162e+00 -2.86220074e-01 4.68848884e-01 1.43368348e-01 7.83185363e-01 5.22688806e-01 -4.68922138e-01 6.33883655e-01 6.58715785e-01 -6.96229935e-01 -8.57896686e-01 -9.92077410e-01 -5.10764420e-01 8.36570203e-01 -4.05121893e-01 -3.71615797e-01 -5.94980478e-01 -1.05588925e+00 1.62458390e-01 4.68993038e-01 -8.87612522e-01 -2.82987475e-01 -4.53356683e-01 -1.38417566e+00 1.17395318e+00 9.09150958e-01 -8.51493329e-02 -8.49372625e-01 -9.58348438e-02 2.79689997e-01 -9.88387764e-02 -1.26364648e+00 -7.09586740e-01 7.28740275e-01 -8.06156397e-01 -1.46642053e+00 -6.31079018e-01 -1.14106011e+00 7.69226968e-01 -6.32553101e-01 1.30717158e+00 2.55321860e-02 -4.75743234e-01 -1.51416078e-01 -9.11108404e-02 -4.67564046e-01 -5.13452828e-01 5.92779279e-01 -2.08768755e-01 -3.69660318e-01 8.36739004e-01 -1.36986643e-01 -1.40976638e-01 -3.38028930e-02 -9.45319593e-01 1.88395064e-02 8.92306745e-01 9.76759851e-01 1.15757644e+00 -5.44632971e-01 3.27120662e-01 -1.20059896e+00 6.85456634e-01 -5.39908230e-01 -3.26510787e-01 5.23615956e-01 -5.16635656e-01 3.49445969e-01 5.47239840e-01 -1.43455684e-01 -5.49861670e-01 6.65539503e-01 -8.60453486e-01 6.61631823e-02 -5.84446669e-01 3.59529525e-01 -3.30390871e-01 6.00480847e-02 7.39197791e-01 -6.96890131e-02 -1.17166534e-01 -4.33086365e-01 3.20913076e-01 1.03107953e+00 9.13805008e-01 -1.66410059e-01 9.25457627e-02 1.65142655e-01 -1.90171357e-02 -3.52886617e-01 -8.50606024e-01 -5.32315135e-01 -6.47762179e-01 6.75142109e-01 1.48721027e+00 -7.82850862e-01 -7.22816527e-01 1.95168212e-01 -1.24517047e+00 -7.80594395e-03 -2.68335253e-01 5.12848854e-01 -2.44293988e-01 -1.80120319e-01 -1.07790232e+00 -6.42268807e-02 -8.68087173e-01 -1.21585190e+00 9.36357319e-01 1.12563312e-01 -4.63037729e-01 -7.83367217e-01 3.31464022e-01 5.40081821e-02 4.71813768e-01 3.84753883e-01 1.15379143e+00 -1.56311011e+00 1.75260007e-01 -4.86207545e-01 -5.52361071e-01 1.30105957e-01 3.45907122e-01 -1.22387655e-01 -8.85183215e-01 -3.64906341e-03 -5.45853913e-01 -2.87538588e-01 9.27278221e-01 4.47942793e-01 1.64904916e+00 -1.43761724e-01 -7.17581570e-01 1.20346129e+00 1.31053472e+00 4.09813255e-01 4.98082131e-01 1.14812039e-01 5.69017053e-01 1.62358925e-01 -7.90518969e-02 8.65147784e-02 4.71935689e-01 4.04963046e-01 2.89278716e-01 -2.90752888e-01 -2.17472062e-01 3.89286801e-02 -2.83052415e-01 9.62122083e-01 2.61784345e-01 -2.64137864e-01 -1.40470517e+00 5.97416103e-01 -1.47977769e+00 -5.88742077e-01 -1.68371484e-01 1.50346804e+00 1.06305468e+00 -1.04458854e-01 -4.74895328e-01 -2.14857772e-01 8.38792086e-01 -2.68304318e-01 -5.92382193e-01 -5.99183917e-01 -8.25333819e-02 4.28932965e-01 8.61771941e-01 2.30398878e-01 -1.42720664e+00 7.09506452e-01 6.94729900e+00 5.37033796e-01 -8.66174817e-01 3.62200528e-01 9.89356339e-01 -9.18511599e-02 7.95523822e-02 -8.25559795e-01 -8.48247588e-01 2.88630873e-01 1.45987201e+00 -1.17506690e-01 2.10644417e-02 6.84019864e-01 -5.91186464e-01 4.41438645e-01 -1.27347338e+00 9.54919398e-01 1.95215687e-01 -1.50452757e+00 5.80720343e-02 5.86120375e-02 6.95319533e-01 4.11610126e-01 -4.35797684e-02 4.26958829e-01 7.01972902e-01 -1.33782983e+00 1.54361520e-02 5.11482298e-01 1.12951767e+00 -8.61589432e-01 1.64029276e+00 -6.28540479e-03 -1.00663745e+00 2.96487808e-02 -1.28252087e-02 6.42403901e-01 9.91485938e-02 7.70839214e-01 -8.68633807e-01 5.70687294e-01 7.32141197e-01 7.56959498e-01 -5.95927238e-01 1.13643479e+00 -2.53151357e-01 3.97340417e-01 -2.20973894e-01 -9.23461914e-02 -2.17695329e-02 6.03729963e-01 -6.78710267e-02 1.65951276e+00 3.38104039e-01 4.26601380e-01 2.76360124e-01 3.70604396e-01 -8.86236310e-01 4.08856481e-01 -3.88109773e-01 -1.50925964e-01 2.32574359e-01 1.36534464e+00 -7.69054890e-01 -5.07655323e-01 -9.15346220e-02 7.04937518e-01 4.70456094e-01 -9.32224989e-02 -1.07211471e+00 -6.79388821e-01 9.03020740e-01 -5.21529198e-01 4.02630419e-02 3.28727663e-01 -5.67049205e-01 -1.04211760e+00 -2.49583736e-01 -8.67982447e-01 1.01568186e+00 -3.57853860e-01 -1.55293870e+00 8.81078482e-01 -4.68412220e-01 -7.38617241e-01 -1.61139145e-01 -9.97204244e-01 -4.84961346e-02 6.41180038e-01 -1.45995724e+00 -9.49983597e-01 -3.18693668e-01 3.86665046e-01 9.37453136e-02 -2.12573156e-01 1.40671527e+00 7.84329176e-01 -9.13685620e-01 1.05421102e+00 3.56188506e-01 7.13540554e-01 9.28066075e-01 -1.23869538e+00 5.37029743e-01 4.58600014e-01 -9.64548439e-03 3.99104446e-01 9.59113985e-02 -5.31534076e-01 -1.01591837e+00 -1.62633324e+00 1.60818183e+00 -6.20833993e-01 5.58534384e-01 -3.47718775e-01 -9.84510839e-01 7.98091054e-01 7.63443783e-02 4.32760596e-01 1.24813604e+00 1.28819555e-01 -3.96922499e-01 2.94396486e-02 -1.48634279e+00 1.89452946e-01 1.09869003e+00 -4.17484164e-01 -6.16090000e-01 4.82063085e-01 8.11585903e-01 -7.29104042e-01 -1.40677869e+00 6.02109015e-01 4.82899994e-01 -2.36402694e-02 9.87301588e-01 -1.14724684e+00 4.34227794e-01 2.61746466e-01 -4.53036159e-01 -1.13078904e+00 -6.33113205e-01 -6.48056343e-02 1.87297359e-01 1.20091033e+00 9.50635731e-01 -6.09694481e-01 7.34596789e-01 5.88868618e-01 -3.75534952e-01 -6.34201884e-01 -1.10818541e+00 -4.83272523e-01 4.28138137e-01 -4.06461656e-01 9.92941082e-01 1.15569949e+00 -1.57662943e-01 1.40505463e-01 1.31991401e-01 3.58926356e-01 3.09600979e-01 -2.01445758e-01 9.03845299e-03 -1.35749471e+00 1.86051458e-01 -6.28760874e-01 -4.83819246e-01 -1.99016809e-01 2.88587213e-01 -1.58307910e+00 2.56237268e-01 -1.60205436e+00 6.14740849e-01 -3.33347142e-01 -1.03748131e+00 9.66861486e-01 1.12118535e-02 2.90253371e-01 -1.99163586e-01 3.78736332e-02 -7.93931603e-01 7.65487030e-02 5.34885168e-01 -4.51147944e-01 1.57962125e-02 -3.37688625e-01 -9.45637405e-01 7.58705974e-01 9.71582651e-01 -1.18853426e+00 3.64292026e-01 -7.00583160e-01 -1.23429328e-01 -5.67171425e-02 -1.84041820e-02 -9.07142699e-01 4.99764204e-01 1.65770820e-03 5.85417271e-01 -5.04365325e-01 -2.44738013e-02 -7.00883448e-01 1.15224041e-01 8.61663222e-01 -8.11693192e-01 5.27064681e-01 6.18910074e-01 3.78949583e-01 -2.02468649e-01 3.30703333e-02 6.55123234e-01 1.31150961e-01 -3.31011027e-01 4.41239983e-01 -3.26241016e-01 1.93561777e-01 1.01175416e+00 4.81624633e-01 -4.82906640e-01 4.37392086e-01 -9.99297500e-01 3.98376167e-01 1.66941099e-02 4.07086670e-01 4.59112555e-01 -1.41547370e+00 -9.14901018e-01 1.59739763e-01 4.81385827e-01 -1.14581272e-01 2.44860232e-01 8.97878766e-01 -6.62157059e-01 8.02667439e-01 -3.47388417e-01 -4.84031528e-01 -1.56551778e+00 5.81433594e-01 3.94151628e-01 -8.40542853e-01 -3.17829877e-01 8.14481914e-01 -2.99127083e-02 -9.43785667e-01 2.13286534e-01 -7.34712005e-01 -5.65899551e-01 1.63267195e-01 8.21041346e-01 1.28947169e-01 6.89671278e-01 -4.94045883e-01 -9.26702082e-01 5.13417423e-01 -3.39609355e-01 3.32207501e-01 1.54401970e+00 5.96652567e-01 -3.56365979e-01 -2.05085605e-01 1.58802485e+00 -4.12083149e-01 -2.98863113e-01 -9.03211683e-02 4.91499007e-01 2.45665148e-01 1.19871788e-01 -1.20356333e+00 -1.18584275e+00 7.57323921e-01 8.26269209e-01 -1.84655428e-01 1.14188063e+00 3.81242096e-01 7.29080796e-01 5.24076521e-01 -1.06072210e-01 -6.01384699e-01 -7.37826228e-01 7.13868797e-01 5.46046674e-01 -1.21047568e+00 -1.76654667e-01 -2.65981436e-01 -2.61865497e-01 1.05659211e+00 4.05601203e-01 3.33323866e-01 5.86168647e-01 7.81377673e-01 1.61627829e-01 -2.79545277e-01 -1.01908052e+00 -8.63336176e-02 4.00346011e-01 5.21572948e-01 6.80447638e-01 2.98970371e-01 -3.60367328e-01 1.24135721e+00 -5.69705218e-02 1.58487752e-01 2.39001617e-01 8.33603024e-01 -1.73670202e-01 -1.06285107e+00 -1.30329639e-01 8.23033452e-01 -1.13922966e+00 -3.93070400e-01 -4.56062615e-01 7.69722819e-01 2.33811855e-01 5.54231763e-01 1.60609901e-01 -3.61821800e-01 7.37902284e-01 3.03839266e-01 -1.00099020e-01 -7.67986238e-01 -8.73843014e-01 -4.29579347e-01 2.35041976e-01 -6.48768425e-01 4.08992432e-02 -6.87237501e-01 -1.53098047e+00 -2.87057310e-02 1.34664103e-01 5.09240590e-02 6.79317415e-01 5.50314903e-01 6.59952462e-01 1.06061471e+00 -6.37733713e-02 7.16795251e-02 -4.56338376e-01 -8.68927300e-01 -1.09942749e-01 5.22006452e-01 3.31499428e-01 -1.30214781e-01 -1.35921165e-01 -1.00233525e-01]
[8.512078285217285, 8.726417541503906]
2229a0e1-d5bb-4aca-aff0-c1adceeaa51e
detecting-human-object-interaction-via
2103.08214
null
https://arxiv.org/abs/2103.08214v2
https://arxiv.org/pdf/2103.08214v2.pdf
Detecting Human-Object Interaction via Fabricated Compositional Learning
Human-Object Interaction (HOI) detection, inferring the relationships between human and objects from images/videos, is a fundamental task for high-level scene understanding. However, HOI detection usually suffers from the open long-tailed nature of interactions with objects, while human has extremely powerful compositional perception ability to cognize rare or unseen HOI samples. Inspired by this, we devise a novel HOI compositional learning framework, termed as Fabricated Compositional Learning (FCL), to address the problem of open long-tailed HOI detection. Specifically, we introduce an object fabricator to generate effective object representations, and then combine verbs and fabricated objects to compose new HOI samples. With the proposed object fabricator, we are able to generate large-scale HOI samples for rare and unseen categories to alleviate the open long-tailed issues in HOI detection. Extensive experiments on the most popular HOI detection dataset, HICO-DET, demonstrate the effectiveness of the proposed method for imbalanced HOI detection and significantly improve the state-of-the-art performance on rare and unseen HOI categories. Code is available at https://github.com/zhihou7/HOI-CL.
['DaCheng Tao', 'Xiaojiang Peng', 'Yu Qiao', 'Baosheng Yu', 'Zhi Hou']
2021-03-15
null
http://openaccess.thecvf.com//content/CVPR2021/html/Hou_Detecting_Human-Object_Interaction_via_Fabricated_Compositional_Learning_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Hou_Detecting_Human-Object_Interaction_via_Fabricated_Compositional_Learning_CVPR_2021_paper.pdf
cvpr-2021-1
['affordance-recognition']
['computer-vision']
[ 2.44271576e-01 -1.95998296e-01 -4.84981090e-02 -2.21301258e-01 -5.18142402e-01 -2.51394928e-01 4.12727535e-01 -9.38400999e-02 1.42494217e-01 4.30727988e-01 4.14855421e-01 2.96554267e-01 -1.15728475e-01 -6.77304089e-01 -8.58775675e-01 -6.90148771e-01 1.83578417e-01 4.82293636e-01 2.27655888e-01 1.24181136e-01 1.60488904e-01 2.02597871e-01 -2.21500492e+00 7.17515469e-01 7.99116611e-01 1.20417106e+00 8.82772505e-02 6.58118427e-01 2.89284945e-01 8.25581551e-01 -6.53417528e-01 -2.07664222e-01 7.34686255e-02 -4.59022105e-01 -4.88812745e-01 3.12461197e-01 6.66595936e-01 -3.60200316e-01 -3.65650952e-01 8.86224329e-01 7.49615431e-01 -9.00177434e-02 9.71490681e-01 -1.85979819e+00 -8.88808548e-01 3.32087487e-01 -6.07436717e-01 1.80502459e-01 7.27164030e-01 4.74527776e-01 1.01678514e+00 -1.37216318e+00 6.62625909e-01 1.70600474e+00 3.87052983e-01 5.30044556e-01 -9.11686361e-01 -1.07126689e+00 -3.81319635e-02 5.83406508e-01 -1.61991942e+00 -2.04204559e-01 8.70206237e-01 -5.95797956e-01 6.90321565e-01 4.11293387e-01 9.10819530e-01 1.36883879e+00 -1.25918895e-01 1.72443724e+00 8.53439629e-01 -2.59869486e-01 -1.35252684e-01 1.82457209e-01 4.73595768e-01 6.20513797e-01 6.68824315e-01 -4.46866229e-02 -6.44012392e-01 1.37956599e-02 4.85695809e-01 1.65507793e-01 -3.61497611e-01 -2.98942804e-01 -1.51375926e+00 6.94980383e-01 5.11726618e-01 2.81083258e-03 -1.86482638e-01 -9.94571149e-02 5.53393364e-01 -1.27817243e-01 7.31058568e-02 2.03675464e-01 1.11581631e-01 3.12463880e-01 -3.03829789e-01 7.36726403e-01 5.98200262e-01 1.28614247e+00 5.90452909e-01 -3.57329190e-01 -7.19487250e-01 8.67608726e-01 1.16576105e-01 5.66796303e-01 4.56354499e-01 -6.44735396e-01 5.69784343e-01 8.94817948e-01 1.30678892e-01 -1.05096364e+00 -3.21792096e-01 -4.05791283e-01 -9.86540914e-01 -3.57966274e-02 1.67940125e-01 3.17839116e-01 -7.62395442e-01 1.51602376e+00 7.52703607e-01 -1.52061693e-02 -1.48563936e-01 1.04299188e+00 1.41382420e+00 7.82084048e-01 8.35699439e-02 -2.30320636e-03 1.58905077e+00 -1.22230256e+00 -6.52862310e-01 -1.33835793e-01 5.10775626e-01 -5.76842129e-01 1.41307795e+00 3.98367882e-01 -7.99482644e-01 -7.98267603e-01 -1.04501522e+00 -3.89700353e-01 -3.90345097e-01 3.63064021e-01 5.38208246e-01 2.82719404e-01 -2.76520729e-01 -1.89861909e-01 -1.41692355e-01 -3.23475957e-01 9.04119730e-01 7.32185021e-02 -3.22892129e-01 -3.36688906e-01 -1.07151949e+00 4.54344213e-01 8.48683238e-01 2.71191746e-01 -1.31562364e+00 -7.45768309e-01 -8.97622764e-01 -1.02882601e-01 7.60719538e-01 -7.98664808e-01 9.35186028e-01 -6.83556974e-01 -8.46323967e-01 1.14115441e+00 -5.42105809e-02 -1.59434780e-01 5.85269988e-01 -3.73602420e-01 -2.66052932e-01 6.17475174e-02 2.04689756e-01 7.30927825e-01 1.07844925e+00 -1.46832216e+00 -7.25663245e-01 -4.73756760e-01 -1.43800035e-01 3.42730552e-01 -2.78454036e-01 1.31594405e-01 -2.53741950e-01 -6.48328066e-01 5.43245487e-03 -1.03970027e+00 6.34963632e-01 2.02032030e-01 -7.45891452e-01 -8.03481042e-01 1.15399146e+00 -5.88133931e-01 1.25263178e+00 -2.36657858e+00 7.65946135e-02 -2.08278194e-01 5.28128982e-01 4.31547195e-01 -1.69973627e-01 1.77488342e-01 -1.03243373e-01 -1.31205127e-01 -1.17019862e-01 -3.88743132e-02 1.87645823e-01 -9.68100782e-03 -1.36620626e-01 5.13711989e-01 3.04507554e-01 1.26353216e+00 -9.28334355e-01 -6.86594903e-01 2.87080199e-01 2.31803387e-01 -5.80593824e-01 7.48708606e-01 -1.12956576e-01 4.50037748e-01 -1.77689835e-01 1.10814381e+00 4.89965409e-01 -2.20878914e-01 -2.98541397e-01 -4.64758158e-01 1.80468589e-01 -8.38962442e-04 -1.24137688e+00 1.06712317e+00 -1.22506157e-01 4.28653955e-01 -1.33484274e-01 -1.05568027e+00 7.55167603e-01 1.84665129e-01 1.45710185e-01 -6.39142156e-01 2.15609387e-01 2.62161195e-01 8.39270353e-02 -8.97116065e-01 1.13365486e-01 1.65577039e-01 -2.78871119e-01 -1.31161194e-02 3.61777991e-02 -2.64385313e-01 2.46595547e-01 7.60866329e-02 7.13642955e-01 -1.06118470e-02 6.23277664e-01 -1.06357113e-01 5.06934583e-01 -2.10626841e-01 6.87109828e-01 9.13055480e-01 -6.11350656e-01 6.62421286e-01 4.03121382e-01 -5.77368081e-01 -1.12766695e+00 -1.16176438e+00 -1.86553717e-01 1.33654511e+00 5.29197991e-01 -6.89651296e-02 -6.89501643e-01 -6.06911063e-01 2.06427738e-01 6.19733214e-01 -7.12490499e-01 -2.43058458e-01 -5.16210735e-01 -7.52752602e-01 3.09878170e-01 6.10045850e-01 6.66390896e-01 -1.77491319e+00 -5.56590557e-01 -1.28260879e-02 -6.76836967e-01 -1.05693376e+00 -4.93848354e-01 -1.59354836e-01 -1.80878162e-01 -1.06072295e+00 -5.56398869e-01 -1.08824754e+00 4.35313761e-01 6.26707137e-01 1.19874537e+00 2.33366974e-02 -1.10013282e+00 1.67074025e-01 -4.12669122e-01 -7.74357378e-01 -1.34567931e-01 -1.53290734e-01 -1.13043387e-03 2.37022102e-01 4.52225357e-01 -2.29107127e-01 -8.36255610e-01 7.01258600e-01 -8.43476176e-01 3.73120159e-01 4.72380906e-01 9.89127636e-01 4.73845929e-01 2.87616551e-01 4.89562035e-01 -7.85345495e-01 2.18031928e-01 -5.91608524e-01 -1.57827497e-01 2.26145044e-01 9.57367867e-02 -4.04573113e-01 3.32325041e-01 -8.38981330e-01 -1.19829607e+00 -1.61348656e-01 2.30322734e-01 -5.15192330e-01 -6.12095296e-01 4.78010811e-02 -6.95903778e-01 1.11095458e-01 5.27053654e-01 1.41797841e-01 -4.36064661e-01 -2.30196178e-01 2.93251783e-01 9.83619034e-01 7.67007232e-01 -5.00065804e-01 7.05665648e-01 5.15213668e-01 -4.16209221e-01 -1.02641618e+00 -1.31085610e+00 -8.09668005e-01 -4.72685605e-01 -4.06442076e-01 1.01395631e+00 -1.14652216e+00 -8.84384274e-01 8.05594206e-01 -1.08028758e+00 -3.22284460e-01 -1.58570752e-01 1.81050286e-01 -6.02339268e-01 1.68255106e-01 -5.96188545e-01 -9.34715688e-01 -3.99815381e-01 -9.88977551e-01 1.48139632e+00 2.38731652e-01 -2.96986789e-01 -1.95082381e-01 -2.81930625e-01 9.98682737e-01 -1.72955871e-01 3.59101832e-01 1.08115542e+00 -4.72086012e-01 -7.48687267e-01 -2.75714666e-01 -6.64322078e-01 3.92153591e-01 -2.20764458e-01 -2.10688785e-01 -1.00734258e+00 -3.05426359e-01 2.49831565e-03 -8.58585954e-01 7.13694930e-01 1.18890824e-02 1.53198659e+00 -3.37514102e-01 -1.87216878e-01 4.74975884e-01 1.03529763e+00 7.98744559e-02 7.29688466e-01 5.44203371e-02 1.24155951e+00 8.28364789e-01 1.06158400e+00 7.12681770e-01 4.55335468e-01 6.89799130e-01 4.32756156e-01 2.16811299e-02 -5.96477449e-01 -4.25389946e-01 2.06611738e-01 7.75754333e-01 6.37105256e-02 -4.32974935e-01 -8.04652631e-01 7.68280268e-01 -1.96233869e+00 -9.68778670e-01 -3.14543784e-01 1.93086970e+00 5.58302104e-01 1.11233965e-01 3.17244053e-01 4.22372520e-01 9.59980905e-01 -6.53440878e-03 -6.86387956e-01 2.38871261e-01 -2.44103074e-01 -3.53465676e-01 1.68031380e-02 -9.52891633e-03 -1.33853114e+00 7.54952431e-01 4.99435568e+00 1.06418121e+00 -7.00262845e-01 2.47513488e-01 4.34301287e-01 -7.57451653e-02 7.52750412e-02 -5.35269082e-01 -1.18168056e+00 6.83680475e-01 2.65119463e-01 -3.33594643e-02 3.87149781e-01 1.09179926e+00 -2.70740222e-03 -8.23709667e-02 -1.20955062e+00 1.55959916e+00 5.35772145e-01 -9.48220551e-01 2.74244577e-01 -4.30846550e-02 7.27659225e-01 -3.50922704e-01 5.66106848e-02 5.63012600e-01 -1.77468523e-01 -9.32124674e-01 9.97803032e-01 2.65859067e-01 7.36805320e-01 -4.16801661e-01 8.15659583e-01 4.83258545e-01 -1.49349284e+00 -4.71854210e-01 -3.66268098e-01 -9.29093957e-02 -2.10658669e-01 4.47998583e-01 -6.90525472e-01 1.03920527e-01 9.37031329e-01 7.76944637e-01 -8.12502146e-01 1.15870810e+00 -2.82393813e-01 4.87765074e-01 -2.52513349e-01 -1.12437159e-01 -1.66081667e-01 1.12304643e-01 5.98528504e-01 1.19680321e+00 1.23719417e-01 2.46409118e-01 4.45551217e-01 8.46463323e-01 -1.11416593e-01 7.47111142e-02 -7.34519720e-01 -6.19515888e-02 3.75038534e-01 1.03821993e+00 -5.01949489e-01 -5.73167801e-01 -4.58732873e-01 1.16167450e+00 5.20850122e-01 2.67758071e-01 -1.21394122e+00 -4.49838609e-01 5.06145418e-01 1.91098556e-01 3.09999913e-01 1.65113091e-01 -3.11510582e-02 -1.33108532e+00 2.75315344e-01 -1.10814071e+00 7.22605944e-01 -9.69858646e-01 -1.53510892e+00 1.05594859e-01 3.07361990e-01 -1.32619321e+00 1.80798337e-01 -6.89656496e-01 -4.22513962e-01 2.73028225e-01 -1.19950843e+00 -1.45299542e+00 -1.11469793e+00 6.65060222e-01 9.65004563e-01 2.36922696e-01 3.88532728e-01 5.01402080e-01 -8.37336123e-01 5.77220201e-01 -3.32278997e-01 1.25961632e-01 6.14369512e-01 -9.51998532e-01 1.69153661e-02 4.76890534e-01 -1.09622506e-02 3.94691199e-01 6.37199938e-01 -6.79842353e-01 -1.35349083e+00 -1.31548464e+00 5.33598721e-01 -6.68052435e-01 4.45224315e-01 -8.93118858e-01 -9.95230496e-01 6.85651839e-01 -2.56271571e-01 4.09978509e-01 5.61484814e-01 -3.07215184e-01 -5.32567203e-01 -4.01219279e-02 -9.71471965e-01 5.28816104e-01 1.65651321e+00 -5.63370824e-01 -8.17019761e-01 6.56713307e-01 7.80105412e-01 -2.46471956e-01 -6.78449452e-01 7.32373893e-01 7.68138468e-01 -1.04192197e+00 1.23120475e+00 -3.67780447e-01 3.71897012e-01 -4.83233005e-01 -1.63289323e-01 -8.51487100e-01 -5.36700785e-01 -2.84762651e-01 -4.84511316e-01 1.15247834e+00 -8.90018791e-02 -4.02134359e-01 6.21297240e-01 2.12281883e-01 -6.48493692e-02 -7.19194472e-01 -4.63959575e-01 -8.54013920e-01 -4.24721450e-01 -3.21442664e-01 3.95885944e-01 6.33272171e-01 -2.75309652e-01 5.31387150e-01 -5.87983131e-01 2.80062795e-01 1.09906268e+00 4.12106633e-01 1.18320882e+00 -1.08079338e+00 -2.49951273e-01 -2.69892514e-01 -6.56024218e-01 -9.31581676e-01 1.37230620e-01 -7.17720151e-01 3.91168803e-01 -1.21538699e+00 9.51055706e-01 -2.18744099e-01 -1.98415637e-01 3.62014264e-01 -5.17811060e-01 8.16027999e-01 2.30597600e-01 4.45917308e-01 -9.28290188e-01 7.78812110e-01 1.35219026e+00 -3.86054844e-01 -1.20147601e-01 -2.56995469e-01 -4.48090762e-01 9.08180177e-01 5.53920448e-01 -2.39910737e-01 -4.09549832e-01 -1.12773895e-01 -3.46531942e-02 -2.87654132e-01 8.42648387e-01 -1.26414096e+00 -1.47930071e-01 -8.97233561e-02 4.32055175e-01 -1.04637158e+00 1.90249115e-01 -5.45585930e-01 1.15233138e-01 5.40123582e-01 -3.97177517e-01 -4.91366059e-01 -1.67323321e-01 5.52961707e-01 -2.08018854e-01 -9.06415209e-02 8.86564374e-01 -1.15131274e-01 -9.77475703e-01 3.73225093e-01 -1.19074918e-01 4.11791593e-01 1.32360840e+00 -2.00745255e-01 -4.50053215e-01 -5.22178486e-02 -5.44900179e-01 2.69425273e-01 2.37465382e-01 7.01294005e-01 7.90334940e-01 -1.32893324e+00 -7.58387744e-01 3.91986012e-01 8.68875682e-01 1.16787970e-01 4.67012048e-01 7.31468260e-01 -3.80705535e-01 2.75877893e-01 -2.20880032e-01 -7.35412359e-01 -1.43549192e+00 1.00091207e+00 -3.63837369e-02 1.45140335e-01 -8.75437379e-01 8.93215120e-01 1.04690230e+00 -4.59728986e-01 5.78844786e-01 -2.56924003e-01 -7.30796298e-03 1.43035680e-01 7.35738993e-01 4.93427455e-01 -3.25760990e-01 -7.52115905e-01 -3.58283758e-01 3.94221783e-01 5.02132513e-02 5.14726162e-01 8.35701048e-01 -2.20527947e-01 -1.01770103e-01 6.36954486e-01 1.38445246e+00 -5.81555843e-01 -1.02589107e+00 -1.63370237e-01 -3.91540527e-01 -6.64210975e-01 -4.17845249e-01 -5.40053785e-01 -5.41684210e-01 9.82601583e-01 6.96311116e-01 7.51044080e-02 8.28179359e-01 4.24002469e-01 1.04217255e+00 3.47233295e-01 4.51747119e-01 -8.70617509e-01 6.23267472e-01 1.47515610e-01 1.23363805e+00 -1.54660225e+00 -1.43030614e-01 -8.95914078e-01 -7.27509379e-01 5.45446455e-01 1.01223934e+00 -6.50046542e-02 2.40292937e-01 -1.27497017e-01 -3.00791830e-01 -4.25886422e-01 -4.83809382e-01 -3.69945049e-01 2.97004104e-01 6.55149817e-01 1.13610700e-01 4.24623042e-01 -1.75015897e-01 3.87417406e-01 -9.69455391e-02 -1.21969454e-01 1.97509989e-01 7.14267313e-01 -4.14912820e-01 -2.62900263e-01 -4.81903791e-01 6.13110185e-01 1.19263239e-01 1.75567299e-01 -3.06528926e-01 7.55433619e-01 5.71897507e-01 7.50396550e-01 -1.88920032e-02 -2.54912794e-01 4.36733872e-01 -2.23848969e-02 5.21537066e-01 -6.84173405e-01 -2.04311162e-01 -9.94142517e-02 2.91865394e-02 -8.54951680e-01 -2.70596087e-01 -5.31353116e-01 -8.35895777e-01 1.06009305e-01 -3.39812696e-01 -3.77290338e-01 2.89766818e-01 7.85661817e-01 2.57267177e-01 5.58661163e-01 3.97561729e-01 -1.16094613e+00 -3.15934807e-01 -9.34492826e-01 -6.97387278e-01 1.27663434e+00 2.58991361e-01 -1.20034170e+00 -5.49641967e-01 3.45794000e-02]
[9.527586936950684, 1.4299191236495972]
8b7a17bf-0957-47d9-8237-746facca3e7f
sumsum-fns-2020-shared-task
null
null
https://aclanthology.org/2020.fnp-1.25
https://aclanthology.org/2020.fnp-1.25.pdf
SUMSUM@FNS-2020 Shared Task
This paper describes the SUMSUM systems submitted to the Financial Narrative Summarization Shared Task (FNS-2020). We explore a section-based extractive summarization method tailored to the structure of financial reports: our best system parses the report Table of Contents (ToC), splits the report into narrative sections based on the ToC, and applies a BERT-based classifier to each section to determine whether it should be included in the summary. Our best system ranks 4<sup>th</sup>, 1<sup>st</sup>, 2<sup>nd</sup> and 17<sup>th</sup> on the Rouge-1, Rouge-2, Rouge-SU4, and Rouge-L official metrics, respectively. We also report results on the validation set using an alternative set of Rouge-based metrics that measure performance with respect to the best-matching of the available gold summaries.
['Claire Cardie', 'Anneliese Lu', 'Siyan Zheng']
null
null
null
null
fnp-coling-2020-12
['extractive-summarization']
['natural-language-processing']
[-1.45736799e-01 5.68689525e-01 -5.90007961e-01 -5.56505211e-02 -1.43340147e+00 -9.03407037e-01 9.28650677e-01 9.21408117e-01 -2.77951926e-01 1.12863958e+00 1.11142051e+00 -4.06054407e-01 -4.90281910e-01 -6.39906108e-01 -3.26067716e-01 -6.13421015e-02 -1.03711203e-01 4.66300666e-01 3.49842072e-01 -1.72251716e-01 9.41855609e-01 3.60847950e-01 -1.09633827e+00 8.52310002e-01 7.28805721e-01 1.04768920e+00 -1.93897322e-01 8.21307719e-01 -1.46914765e-01 8.79383624e-01 -8.72175694e-01 -6.10233605e-01 -2.11759061e-01 -8.45954359e-01 -1.07513165e+00 -1.86192337e-02 2.00900286e-01 -7.63568282e-02 -1.12545528e-01 7.54593313e-01 6.98102191e-02 6.65747598e-02 1.13289893e+00 -6.70606017e-01 -2.13411793e-01 1.02960563e+00 -6.48316741e-01 8.11419427e-01 9.03434575e-01 -2.30563298e-01 1.44488168e+00 -6.71409011e-01 1.15854895e+00 8.64942491e-01 3.94957721e-01 3.27769890e-02 -7.84632027e-01 -3.50967437e-01 5.75357154e-02 5.19584604e-02 -7.06403255e-01 -3.03643942e-01 4.39930439e-01 -5.85854292e-01 1.38494432e+00 5.57063401e-01 5.03504753e-01 8.46069634e-01 6.05406284e-01 7.42252529e-01 7.72799432e-01 -4.46672410e-01 2.32292548e-01 -1.30522266e-01 5.95989823e-01 2.08415478e-01 5.98936141e-01 -5.26066244e-01 -5.32614589e-01 -4.08487737e-01 3.80786210e-02 -5.76807678e-01 -2.57462531e-01 6.33834541e-01 -1.43572402e+00 8.37329686e-01 -8.45586136e-02 5.69067836e-01 -6.23086452e-01 -1.92008138e-01 7.19590485e-01 1.52395129e-01 9.07920837e-01 8.80069792e-01 -3.65574121e-01 -4.56332922e-01 -1.60499632e+00 7.74117827e-01 1.02912223e+00 8.56276751e-01 2.72747934e-01 -2.19533026e-01 -7.62033105e-01 7.06280291e-01 -1.06640175e-01 6.44745082e-02 2.81732261e-01 -1.11588871e+00 1.08703804e+00 6.64589882e-01 3.50572944e-01 -8.76604259e-01 -6.23776734e-01 -5.63560784e-01 -6.24563038e-01 -4.08022732e-01 -1.88775778e-01 -1.99695781e-01 -3.69381845e-01 1.02995563e+00 -2.18691722e-01 -5.87589025e-01 2.01107383e-01 4.36121792e-01 1.37229371e+00 8.62510085e-01 -2.56129354e-01 -1.16221416e+00 1.25877237e+00 -7.45917082e-01 -8.38875771e-01 1.99057087e-01 7.82529771e-01 -9.02175963e-01 4.82104838e-01 4.51979488e-01 -1.46080983e+00 -2.01028511e-01 -1.27231765e+00 1.50375724e-01 -2.33886197e-01 1.87810779e-01 2.34829769e-01 2.84746468e-01 -7.97508955e-01 7.13735640e-01 -5.77572405e-01 -2.67469049e-01 1.04901314e-01 -1.68338463e-01 -1.97217062e-01 3.02690893e-01 -1.04522359e+00 6.64853752e-01 8.19842875e-01 -5.16888440e-01 -2.22740248e-01 -7.34389186e-01 -5.14713526e-01 1.23533539e-01 6.96756721e-01 -3.64838451e-01 1.33770657e+00 1.84303984e-01 -8.60096574e-01 7.41402924e-01 3.22457729e-03 -6.98584914e-01 3.98898333e-01 -2.03973621e-01 -6.73619390e-01 5.88407815e-01 4.94367599e-01 9.56608504e-02 1.36750229e-02 -9.79580462e-01 -9.64110732e-01 -7.15123042e-02 -5.60680451e-03 -3.54419127e-02 5.02862222e-03 5.32700658e-01 -4.25777435e-02 -8.19246590e-01 1.18843742e-01 -4.46384668e-01 1.20396890e-01 -1.15313828e+00 -8.79033744e-01 -5.65922976e-01 4.82760519e-01 -9.06725883e-01 2.05767894e+00 -1.68744600e+00 3.07642855e-02 -2.42882017e-02 2.41804168e-01 -1.14005305e-01 2.28270233e-01 9.35233116e-01 4.37417999e-02 6.49090946e-01 -1.88812315e-01 -1.16109885e-01 -2.04789266e-02 -3.23645622e-01 -5.42852998e-01 1.34813100e-01 4.65102866e-02 6.63427472e-01 -8.60432684e-01 -7.72756994e-01 -4.68589246e-01 -5.99576235e-01 -3.90279889e-01 -4.71285023e-02 -2.93146551e-01 -1.63836345e-01 -5.34721076e-01 4.76704359e-01 2.70256460e-01 -7.90432002e-03 -4.99205627e-02 -1.69605568e-01 -6.35895491e-01 8.34207058e-01 -9.00258660e-01 1.12902129e+00 2.09998310e-01 7.10097790e-01 -2.71512240e-01 -6.58326447e-01 1.12970996e+00 3.97105366e-01 6.19181275e-01 -3.56482834e-01 -3.72420851e-04 3.50562662e-01 -2.77625442e-01 -3.98617446e-01 1.34717703e+00 3.30742002e-02 -6.83378577e-01 8.91827643e-01 1.62783325e-01 -4.61226165e-01 1.06489778e+00 7.96628237e-01 1.37551820e+00 -1.76587850e-01 6.83169127e-01 -2.99926698e-01 5.06526828e-01 2.72022784e-01 4.16538000e-01 6.22946620e-01 2.00151443e-01 1.00573492e+00 1.30555308e+00 -2.20399782e-01 -1.11317992e+00 -7.77978420e-01 -1.30070478e-01 4.97924805e-01 -4.27161366e-01 -1.03727400e+00 -7.80326366e-01 -9.60578620e-01 -6.94487244e-02 1.50813758e+00 -6.59223437e-01 -5.44558242e-02 -4.45298612e-01 -9.47717786e-01 7.43979931e-01 2.71791428e-01 2.60873288e-01 -1.07548237e+00 -6.83546782e-01 3.15497041e-01 -5.81776142e-01 -1.06931019e+00 -4.93463725e-01 2.34045669e-01 -7.55380452e-01 -1.09409428e+00 -6.11571670e-01 -5.10013849e-02 1.62671536e-01 -3.45218778e-01 9.91338372e-01 -3.15756470e-01 3.75663131e-01 1.88441515e-01 -8.90900493e-01 -4.00741965e-01 -7.80194819e-01 6.29409850e-01 -1.45875439e-01 -4.41741318e-01 1.17364954e-02 -1.09952599e-01 -5.46539277e-02 -2.49324217e-01 -7.81996369e-01 -2.11176932e-01 2.22865507e-01 5.34000456e-01 4.97740090e-01 -4.08686213e-02 9.92568731e-01 -8.28093529e-01 1.21571743e+00 -5.00975728e-01 -1.29407778e-01 3.64087850e-01 -8.05270672e-01 5.89156821e-02 5.78427851e-01 8.09979439e-02 -8.47389042e-01 -6.51725590e-01 2.30770156e-01 1.08739641e-02 1.65644139e-01 1.14165473e+00 6.47594109e-02 1.05852425e+00 6.76312327e-01 2.67335922e-02 -4.97224063e-01 -6.67826414e-01 1.88933700e-01 7.68523872e-01 9.84609008e-01 -2.78049737e-01 3.20746362e-01 -9.35029313e-02 -5.26172780e-02 -6.34048522e-01 -1.23500609e+00 -5.89753211e-01 -6.42572999e-01 -3.21666747e-01 7.69071937e-01 -5.32572687e-01 -3.30700278e-01 -9.18918401e-02 -1.35194457e+00 3.34109843e-01 -3.85432839e-01 7.01349676e-01 -6.76875532e-01 4.63701457e-01 -5.64472497e-01 -6.95674360e-01 -6.35673106e-01 -7.50628412e-01 6.34336889e-01 1.10131480e-01 -1.11108303e+00 -6.76607192e-01 2.46304899e-01 3.73916149e-01 -1.36781439e-01 6.49110734e-01 1.12230718e+00 -1.18579698e+00 9.32177901e-02 -5.18210828e-01 -2.46716321e-01 2.72044539e-01 -3.27405445e-02 4.68061924e-01 -1.78991064e-01 6.17792606e-02 -1.77572295e-01 -1.53911605e-01 1.28960168e+00 4.80129391e-01 6.25896513e-01 -9.75678205e-01 -4.43136722e-01 -2.88790697e-03 1.05169582e+00 2.97702491e-01 6.12291515e-01 6.74946606e-01 2.30176479e-01 5.86103141e-01 7.89602637e-01 7.24034607e-01 3.58863443e-01 3.73693228e-01 9.56664309e-02 7.96347797e-01 -9.57172960e-02 -1.70583040e-01 4.85031545e-01 1.01144826e+00 -2.51114190e-01 -5.16894400e-01 -8.68378937e-01 7.43923187e-01 -1.65600371e+00 -1.24008787e+00 -4.94806796e-01 1.94233894e+00 7.67235518e-01 8.43243480e-01 4.35526043e-01 4.08379495e-01 5.08569419e-01 6.06772125e-01 4.50707215e-04 -9.10590529e-01 -1.61877111e-01 2.40758080e-02 3.57216269e-01 2.38579229e-01 -9.13962364e-01 5.03433824e-01 6.19075346e+00 7.94976354e-01 -5.05309463e-01 -1.13956286e-02 5.75468421e-01 -3.52371216e-01 -4.64878082e-01 1.58540994e-01 -9.40150380e-01 6.09319746e-01 1.49364412e+00 -8.29335451e-01 -4.54430550e-01 6.04101658e-01 3.78779769e-01 -6.62876964e-01 -1.05946314e+00 2.93596655e-01 2.40760922e-01 -1.98194170e+00 2.43174031e-01 2.29126483e-01 9.82468367e-01 -2.04619929e-01 -3.31708550e-01 1.16779208e-01 8.21025968e-02 -5.23953557e-01 1.08430648e+00 7.50861824e-01 8.95186067e-01 -1.07579756e+00 1.01361525e+00 3.62851977e-01 -8.79636943e-01 5.29162735e-02 -1.99327618e-02 1.60048813e-01 3.78725797e-01 9.31317568e-01 -9.85575140e-01 1.30636656e+00 5.46432316e-01 8.65356266e-01 -5.25105238e-01 8.61914992e-01 -7.10249692e-02 7.90296197e-01 5.16561680e-02 -3.46026510e-01 4.16741610e-01 -2.32530698e-01 1.02528214e+00 1.66504145e+00 5.80131233e-01 4.36597079e-01 5.75147420e-02 5.51425457e-01 -4.44187880e-01 1.91297829e-01 -2.53512204e-01 -4.51090097e-01 4.08717394e-01 7.92364001e-01 -1.02879858e+00 -6.70398951e-01 6.49825037e-02 3.22027296e-01 -6.24827668e-02 -1.25922695e-01 -3.40133131e-01 -8.21237981e-01 -1.34517148e-01 3.30060095e-01 4.58605021e-01 -6.86902925e-02 -6.36265039e-01 -9.90933955e-01 2.08148807e-01 -4.91237223e-01 6.81919336e-01 -7.49773443e-01 -7.55574524e-01 9.18317199e-01 5.72816491e-01 -1.44986820e+00 -7.54380345e-01 6.38637021e-02 -5.91411650e-01 5.48734248e-01 -6.59465849e-01 -5.28836429e-01 3.04410338e-01 -4.64342147e-01 4.99198705e-01 -6.31530583e-01 6.54596508e-01 -1.48030475e-01 -3.84725690e-01 4.92128611e-01 3.08149546e-01 7.98987374e-02 4.88222390e-01 -1.17737925e+00 2.53234506e-01 7.32658565e-01 -1.83155686e-01 1.41618431e-01 1.20417225e+00 -1.08191133e+00 -6.84068084e-01 -1.03474164e+00 1.47285175e+00 -4.49616879e-01 8.97429109e-01 3.43909711e-01 -8.22108269e-01 5.70783257e-01 3.19024444e-01 -8.63652289e-01 4.05452996e-01 2.08375379e-01 -3.19190085e-01 9.71345231e-03 -1.10649216e+00 2.49200046e-01 5.73870540e-01 -1.79383725e-01 -1.19918871e+00 4.70551908e-01 9.77302134e-01 -4.01686311e-01 -1.30006325e+00 2.71581054e-01 4.22796905e-01 -8.46000373e-01 4.14784640e-01 -4.92337763e-01 1.08948982e+00 2.25723814e-02 -2.09578887e-01 -1.21293926e+00 -3.16413492e-01 -7.43528724e-01 -1.34747624e-01 1.49813294e+00 9.03823376e-01 -3.07298303e-01 4.88846749e-01 5.14566787e-02 -6.29886150e-01 -7.91715205e-01 -1.13973475e+00 -9.03927267e-01 2.03150570e-01 -4.91326660e-01 5.02993405e-01 6.76209927e-01 6.31209433e-01 3.92079771e-01 -7.99383670e-02 -4.58516568e-01 2.89482713e-01 1.35998011e-01 3.72573853e-01 -1.16591763e+00 9.53591317e-02 -8.71766210e-01 -7.81842619e-02 -3.34212631e-01 -8.70532990e-02 -9.62301075e-01 -2.18115479e-01 -2.20956063e+00 4.36249405e-01 3.02013278e-01 -2.50645876e-01 2.23300084e-01 9.82163772e-02 -2.61069387e-01 3.49947482e-01 5.26350498e-01 -9.14559484e-01 2.75675863e-01 9.53602552e-01 -2.17144340e-01 -3.14378351e-01 7.03761205e-02 -9.34147179e-01 4.47125584e-01 5.72998881e-01 -5.85238457e-01 1.16983801e-01 3.29160661e-01 5.52758157e-01 5.22874951e-01 -1.82607219e-01 -1.02476025e+00 1.33542642e-01 -4.29561466e-01 6.47201538e-02 -1.39500868e+00 -2.93293417e-01 1.78345308e-01 1.26574069e-01 3.65724564e-01 -8.05565476e-01 4.19610769e-01 1.43420771e-01 2.01520681e-01 -2.75875300e-01 -6.16266966e-01 3.91584903e-01 -1.58984259e-01 -2.38884110e-02 -3.12063694e-01 -6.72296822e-01 1.52867913e-01 6.97198749e-01 -9.27760452e-02 -6.31685197e-01 -3.28722924e-01 -4.48058933e-01 3.05374891e-01 1.34354547e-01 4.25560862e-01 4.62104768e-01 -1.04996169e+00 -1.38621008e+00 -6.97656572e-01 2.01147124e-01 -1.87546104e-01 1.05630927e-01 8.75931263e-01 -5.39150953e-01 1.00714684e+00 2.13271845e-02 6.81416411e-03 -9.86864686e-01 3.61181468e-01 -2.04308689e-01 -1.04252923e+00 -6.08725786e-01 3.60130548e-01 -3.55309397e-01 4.02379632e-01 -1.13248855e-01 -7.22440779e-01 -6.25894487e-01 7.70144701e-01 8.65618646e-01 8.76590252e-01 4.23132241e-01 -7.02422142e-01 -3.78884405e-01 3.58659439e-02 -4.79857504e-01 -5.19709706e-01 1.43932283e+00 -6.03354163e-02 -3.10998142e-01 7.45441675e-01 1.15975928e+00 1.55949473e-01 -6.28034294e-01 1.20502375e-01 6.64493144e-01 1.95836261e-01 1.23669254e-02 -1.04017150e+00 -3.99060428e-01 1.38846606e-01 -4.85429257e-01 7.13398039e-01 7.83586144e-01 4.09838021e-01 7.14273751e-01 2.66179800e-01 -6.43071756e-02 -1.49844921e+00 8.58598724e-02 7.44224131e-01 1.47134435e+00 -5.90500593e-01 9.19588029e-01 -1.11474872e-01 -8.42996955e-01 1.20822656e+00 -1.38713896e-01 -1.45727992e-01 3.67122620e-01 -9.90983937e-03 -4.85219955e-01 -4.20941889e-01 -1.18229735e+00 2.37824738e-01 5.95081627e-01 -1.28306404e-01 4.85219419e-01 1.30016744e-01 -1.18805683e+00 1.17458129e+00 -7.78526902e-01 -9.19771194e-02 1.18927133e+00 8.07412922e-01 -6.78311586e-01 -5.49048781e-01 -4.88771677e-01 1.03597569e+00 -8.93684685e-01 2.19504654e-01 -7.38065243e-01 9.42679346e-01 -1.52816787e-01 1.15565383e+00 4.70572598e-02 -5.26874542e-01 6.07652366e-01 1.60021856e-01 2.71455288e-01 -7.14220405e-01 -9.13096964e-01 3.11984479e-01 8.61190379e-01 -1.01160474e-01 -4.08041894e-01 -1.31970727e+00 -1.43564618e+00 -2.77788907e-01 1.81966163e-02 6.77204370e-01 4.91190225e-01 1.13683069e+00 2.07279325e-01 7.40363598e-01 6.48356020e-01 -4.66153502e-01 -5.28542817e-01 -1.43807364e+00 -5.73604882e-01 1.64364092e-02 2.86919177e-01 -4.42659140e-01 -3.97127748e-01 1.31740376e-01]
[12.43258285522461, 9.521369934082031]
a7b5fd4d-c0d8-4a8b-a3ba-aea02586cb0f
standardized-cyclegan-training-for
2301.13128
null
https://arxiv.org/abs/2301.13128v1
https://arxiv.org/pdf/2301.13128v1.pdf
Standardized CycleGAN training for unsupervised stain adaptation in invasive carcinoma classification for breast histopathology
Generalization is one of the main challenges of computational pathology. Slide preparation heterogeneity and the diversity of scanners lead to poor model performance when used on data from medical centers not seen during training. In order to achieve stain invariance in breast invasive carcinoma patch classification, we implement a stain translation strategy using cycleGANs for unsupervised image-to-image translation. We compare three cycleGAN-based approaches to a baseline classification model obtained without any stain invariance strategy. Two of the proposed approaches use cycleGAN's translations at inference or training in order to build stain-specific classification models. The last method uses them for stain data augmentation during training. This constrains the classification model to learn stain-invariant features. Baseline metrics are set by training and testing the baseline classification model on a reference stain. We assessed performances using three medical centers with H&E and H&E&S staining. Every approach tested in this study improves baseline metrics without needing labels on target stains. The stain augmentation-based approach produced the best results on every stain. Each method's pros and cons are studied and discussed in this paper. However, training highly performing cycleGANs models in itself represents a challenge. In this work, we introduce a systematical method for optimizing cycleGAN training by setting a novel stopping criterion. This method has the benefit of not requiring any visual inspection of cycleGAN results and proves superiority to methods using a predefined number of training epochs. In addition, we also study the minimal amount of data required for cycleGAN training.
['Stéphane Sockeel', 'Marie Sockeel', 'Rémy Peyret', 'Nicolas Nerrienet']
2023-01-30
null
null
null
null
['unsupervised-image-to-image-translation']
['computer-vision']
[ 7.48346150e-01 3.35664541e-01 -1.28690854e-01 -2.48559490e-01 -8.03691447e-01 -6.93406045e-01 4.41358089e-01 5.78010269e-02 -7.44051516e-01 7.11528003e-01 -4.58073795e-01 -6.15902781e-01 1.33059710e-01 -6.49639189e-01 -5.38899899e-01 -1.19846487e+00 1.85839981e-01 5.33692896e-01 2.00026125e-01 -5.15421759e-03 1.13688067e-01 8.60335410e-01 -1.08179677e+00 5.26219904e-01 4.60757673e-01 6.20691240e-01 2.98215777e-01 1.17731714e+00 -1.01388954e-01 6.08937442e-01 -6.89813077e-01 -2.57189125e-01 4.32036519e-01 -9.92059052e-01 -8.19080949e-01 3.06336999e-01 4.94026929e-01 1.19461529e-02 2.73084193e-01 8.71843338e-01 6.19063735e-01 -4.18635249e-01 8.93596590e-01 -9.91061330e-01 -3.33121091e-01 2.58608311e-01 -5.06489933e-01 4.22027141e-01 -1.04811914e-01 5.30200541e-01 6.26818478e-01 -5.53041160e-01 9.48520422e-01 7.33971536e-01 7.85358191e-01 9.38191712e-01 -1.43418694e+00 -5.30912340e-01 -4.77777630e-01 4.02003415e-02 -1.20488560e+00 -1.45384818e-01 5.95125258e-01 -6.00613654e-01 7.22210646e-01 6.44501925e-01 8.00373554e-01 6.27041280e-01 3.52109611e-01 4.91738409e-01 1.79403329e+00 -7.92845964e-01 1.77292675e-01 6.15185976e-01 -1.56013779e-02 9.52568293e-01 5.31346798e-01 -1.75017983e-01 -1.24656975e-01 6.64622337e-02 6.55099273e-01 -1.55841947e-01 -2.68925875e-01 -9.15697664e-02 -1.05828118e+00 7.46181488e-01 4.04749900e-01 7.99893022e-01 -1.47808362e-02 -4.00600918e-02 4.28525090e-01 4.29344267e-01 4.01478201e-01 7.01192081e-01 -1.12642311e-01 4.19622689e-01 -1.09230888e+00 -2.79660732e-01 6.12337470e-01 4.48949337e-01 7.59605765e-01 -9.95107889e-02 -1.86934292e-01 5.51242828e-01 1.49201483e-01 5.86455166e-01 6.57867253e-01 -5.21653891e-01 1.56887397e-02 8.54574978e-01 -4.18013543e-01 -7.97860265e-01 -7.38862753e-01 -7.30024278e-01 -9.78686690e-01 4.37311471e-01 5.95061183e-01 1.16238154e-01 -1.39786839e+00 1.28657877e+00 3.31325710e-01 -1.10804826e-01 1.90538973e-01 6.25765622e-01 6.77848279e-01 2.79630303e-01 -1.58697516e-02 -2.49708235e-01 1.47525585e+00 -8.79664421e-01 -6.88215196e-01 1.15979634e-01 1.31591022e+00 -8.52554023e-01 1.15778923e+00 3.32021087e-01 -8.21913183e-01 -3.10497552e-01 -1.35877383e+00 1.51365757e-01 -6.40318632e-01 2.49158844e-01 5.84884524e-01 1.17628253e+00 -1.32362461e+00 3.98538172e-01 -9.58171248e-01 -5.99062026e-01 5.93888402e-01 7.42682040e-01 -6.50220335e-01 -2.21273080e-02 -6.29543602e-01 1.00581849e+00 3.23512644e-01 1.88277543e-01 -5.46825469e-01 -6.08929217e-01 -5.99399865e-01 -2.95552969e-01 -1.53839335e-01 -5.24344325e-01 9.31647718e-01 -1.16805470e+00 -1.57614899e+00 1.48595369e+00 1.31197162e-02 -4.33602989e-01 8.15626085e-01 7.68313527e-01 -1.64108321e-01 3.82374674e-01 -2.11401597e-01 7.13148057e-01 6.22887552e-01 -1.27607000e+00 -5.40679395e-01 -7.95951262e-02 -2.07622305e-01 -4.40207906e-02 -3.96845758e-01 -2.84555584e-01 -3.34469110e-01 -3.20978969e-01 -1.63434491e-01 -1.15250027e+00 -2.19804749e-01 -5.20619787e-02 -3.99617702e-01 1.69197574e-01 7.37363458e-01 -5.88633299e-01 7.65914142e-01 -2.13366675e+00 -2.28094488e-01 5.97215712e-01 1.66034132e-01 4.37650800e-01 -2.21856385e-01 6.76072165e-02 -1.36253059e-01 4.58758920e-01 -3.11908633e-01 -4.35037255e-01 -2.26143077e-01 3.05837333e-01 4.64940995e-01 9.05514896e-01 3.82108152e-01 1.03738928e+00 -7.16076434e-01 -1.05015349e+00 3.19247365e-01 6.98823690e-01 -5.78019202e-01 1.63078159e-01 1.73216522e-01 5.33193648e-01 1.16755880e-01 6.63097501e-01 7.38392651e-01 -4.68078315e-01 4.47856814e-01 -2.93637395e-01 2.60884315e-01 -3.26400757e-01 -8.41423571e-01 1.14118874e+00 -4.42725837e-01 7.88761199e-01 -2.17605963e-01 -7.98902273e-01 8.19431067e-01 2.27891713e-01 2.36192897e-01 -5.64727187e-01 2.93262094e-01 5.93232989e-01 4.39736813e-01 -5.74474812e-01 1.16390977e-02 -5.56087017e-01 4.38946784e-01 4.00527418e-01 2.11961389e-01 -4.56455648e-01 4.76694852e-01 -2.25596964e-01 1.05976105e+00 -3.13669503e-01 4.37853843e-01 -3.96650821e-01 8.72198939e-01 1.75011724e-01 2.79593021e-01 5.04034817e-01 -1.42411232e-01 5.75037181e-01 7.46937990e-01 -4.43013519e-01 -1.14222884e+00 -4.80417579e-01 -2.27077007e-01 5.27952135e-01 -3.23526651e-01 2.38196373e-01 -9.36919272e-01 -9.16424096e-01 -4.48274761e-01 2.86971241e-01 -1.39989007e+00 5.08912280e-02 -6.87279820e-01 -1.22358775e+00 6.81724906e-01 1.70995206e-01 3.69205862e-01 -9.23564434e-01 -8.62696052e-01 -1.80582002e-01 1.87723890e-01 -7.72540808e-01 -1.97663397e-01 5.63003004e-01 -9.49503183e-01 -1.38508940e+00 -9.16377962e-01 -9.81244147e-01 1.41549182e+00 -7.89059624e-02 8.44915926e-01 5.46019614e-01 -5.29011667e-01 1.21231616e-01 -3.83526087e-01 -6.45003676e-01 -9.05176103e-01 5.62049806e-01 -4.50833499e-01 8.32887068e-02 3.76241505e-01 -1.08033463e-01 -7.13312566e-01 2.88291365e-01 -1.16638577e+00 1.27851143e-01 1.11488748e+00 1.20391870e+00 8.81120861e-01 -2.38508880e-01 6.62295297e-02 -1.43018103e+00 2.38602102e-01 4.11583716e-03 -3.53519708e-01 2.49945596e-01 -6.77904606e-01 -7.95808583e-02 5.77492237e-01 -4.19005960e-01 -6.70112073e-01 1.18873745e-01 -4.41153795e-02 -1.08924314e-01 -1.57147609e-02 2.81166315e-01 8.28101262e-02 -6.51571691e-01 8.66584122e-01 6.95253015e-02 5.06063700e-01 -1.11762970e-03 -2.47263297e-01 4.72143024e-01 2.70884782e-01 8.64626169e-02 9.07415152e-01 7.65010178e-01 4.26889002e-01 -7.48807311e-01 -4.46228027e-01 -5.70164919e-01 -7.01956630e-01 -2.17818618e-01 9.25123036e-01 -2.47054860e-01 -4.22496319e-01 5.29112577e-01 -7.81014264e-01 -8.09178531e-01 -4.47054654e-01 3.81318241e-01 -4.23112273e-01 1.25134766e-01 -6.15761399e-01 -5.04194260e-01 -5.05651951e-01 -1.09169519e+00 8.25687885e-01 3.72833401e-01 -1.80329144e-01 -1.38438678e+00 2.61092454e-01 1.52698129e-01 5.12381375e-01 6.84175730e-01 6.93647921e-01 -7.18968809e-01 -2.85160750e-01 -4.29070175e-01 -1.82321280e-01 5.26384294e-01 4.37862784e-01 2.55634457e-01 -1.06937420e+00 -4.93760139e-01 1.27554327e-01 -3.97017263e-02 7.56477177e-01 3.44510347e-01 1.00685918e+00 -1.99042469e-01 -4.57744330e-01 1.04896367e+00 1.74875200e+00 2.51701832e-01 8.22764158e-01 7.07159698e-01 5.12675047e-01 6.52963400e-01 4.69461292e-01 -4.05625939e-01 -1.60429314e-01 2.39429206e-01 3.26007873e-01 -8.93439114e-01 -5.02916932e-01 8.68440270e-02 -2.36422406e-03 4.85567629e-01 -3.06328803e-01 -3.67267877e-01 -1.00991547e+00 5.37288070e-01 -9.79182422e-01 -6.86619103e-01 -2.04663306e-01 1.62027383e+00 8.15945327e-01 1.37264326e-01 -6.85171336e-02 4.98946697e-01 4.83186305e-01 -2.34197527e-01 -2.86908835e-01 -4.84870851e-01 -3.85143787e-01 4.29722816e-01 8.43125284e-01 5.26036680e-01 -9.42425072e-01 5.99113405e-01 6.53982353e+00 7.35328913e-01 -1.65895772e+00 3.05909455e-01 1.12143660e+00 1.28174096e-01 -1.68314353e-01 -3.98010701e-01 -6.61562324e-01 1.86896041e-01 8.60623002e-01 3.18555683e-01 -1.29765034e-01 4.52930480e-01 7.03844661e-03 -3.06315571e-01 -1.09772944e+00 7.72197783e-01 3.06904763e-01 -1.44419050e+00 9.28208753e-02 4.55741733e-01 7.97337472e-01 -2.96758324e-01 2.86029637e-01 -1.47290349e-01 -1.37571454e-01 -1.22297704e+00 8.51582512e-02 3.92628342e-01 1.34801555e+00 -5.15127063e-01 1.47477627e+00 -2.29034826e-01 -6.41012251e-01 2.98066050e-01 -1.27712592e-01 4.21041936e-01 -2.64542580e-01 3.41138780e-01 -1.59534895e+00 3.55261058e-01 3.78843814e-01 3.13870102e-01 -1.07950771e+00 8.24543893e-01 -7.41971731e-02 8.65819871e-01 -2.07982883e-01 -3.52796018e-01 3.25678855e-01 1.96834147e-01 1.47013843e-01 1.86063838e+00 2.40258455e-01 -2.56669402e-01 -3.71175587e-01 3.87656718e-01 2.41303936e-01 4.69599009e-01 -3.83744180e-01 -4.00566757e-02 -8.27978477e-02 1.65872121e+00 -1.53849757e+00 -3.58802408e-01 -8.92449766e-02 8.10544431e-01 -6.44584224e-02 1.01681352e-01 -6.00444734e-01 -3.71162504e-01 -7.38207027e-02 4.78168368e-01 1.41559646e-01 2.31213540e-01 -4.36887562e-01 -5.76905727e-01 -3.21882159e-01 -9.55569625e-01 5.34669697e-01 -3.28813225e-01 -9.48105872e-01 7.92778552e-01 -1.62713289e-01 -1.32240140e+00 -2.67804950e-01 -1.05534577e+00 -4.96398181e-01 7.50225902e-01 -1.77176356e+00 -1.53396153e+00 -6.11542523e-01 3.31651449e-01 2.14998856e-01 -2.16059715e-01 1.00035906e+00 1.65228993e-01 -2.92212576e-01 1.00682771e+00 -8.64330083e-02 2.43495315e-01 8.54086459e-01 -1.47823119e+00 -1.60237461e-01 8.49417090e-01 -3.19771945e-01 3.67774010e-01 7.69647241e-01 -3.49917352e-01 -9.77938771e-01 -1.02086687e+00 7.38036394e-01 -3.25796574e-01 4.71768945e-01 -2.65856624e-01 -8.04579794e-01 7.38247931e-01 2.70475209e-01 1.27381399e-01 1.23294842e+00 -4.76505041e-01 -1.47912093e-02 -2.02812150e-01 -1.50163400e+00 5.15673995e-01 5.11049569e-01 -2.35940620e-01 -9.26737860e-03 3.83283913e-01 2.36862332e-01 -4.95566130e-01 -7.95537293e-01 2.07085758e-01 5.19960761e-01 -8.46929729e-01 5.24427712e-01 -1.45374954e-01 3.84675264e-01 -4.50765282e-01 1.15372986e-01 -1.16034651e+00 -1.89990819e-01 -3.85722697e-01 4.72745895e-01 1.04603362e+00 7.23186195e-01 -1.01088440e+00 1.09128690e+00 5.52144647e-02 -1.73691828e-02 -8.95859838e-01 -7.74169564e-01 -6.13358915e-01 8.15252811e-02 1.20698094e-01 4.07142907e-01 9.38956678e-01 -2.45442092e-01 -7.64506310e-02 8.30833316e-02 -1.76937029e-01 6.18364334e-01 -2.65327990e-01 9.59682882e-01 -7.85088360e-01 -3.67688507e-01 -5.72793424e-01 -9.11548793e-01 -7.13461563e-02 -1.73051611e-01 -1.14947748e+00 -1.61778301e-01 -1.34687877e+00 4.17960227e-01 -5.44539511e-01 -3.66952449e-01 6.48875892e-01 -1.68680310e-01 1.10601294e+00 1.84546322e-01 1.89914912e-01 -1.80956692e-01 -5.07169664e-01 1.63994467e+00 -4.68213439e-01 -1.59384042e-01 -2.31121063e-01 -7.10444391e-01 4.23482358e-01 1.14382970e+00 -6.16974533e-01 -2.45536879e-01 2.67095100e-02 2.39369512e-01 -5.83761513e-01 3.57444584e-01 -1.07545078e+00 1.06485307e-01 -7.66372075e-03 6.44484282e-01 -3.65434349e-01 5.80976009e-02 -7.82939911e-01 3.52379441e-01 1.18258870e+00 -1.18354172e-01 -7.98090547e-02 2.19283983e-01 1.13973282e-01 -9.34589952e-02 -4.31040347e-01 1.14768338e+00 -1.56220898e-01 -2.95230865e-01 -4.10516337e-02 -4.59341317e-01 -5.28213024e-01 1.22916377e+00 -7.76889086e-01 -2.71755219e-01 5.97898886e-02 -6.42319918e-01 -2.63578165e-02 7.67961860e-01 -1.91905186e-01 3.09474826e-01 -8.49832952e-01 -8.89672995e-01 2.81916499e-01 9.61051881e-02 5.07169999e-02 2.97243506e-01 1.28855443e+00 -1.27407265e+00 3.71497929e-01 -5.54353118e-01 -1.00568020e+00 -1.72145367e+00 5.18452883e-01 8.36106420e-01 -9.44775641e-01 -3.06060195e-01 9.82638776e-01 1.78734824e-01 -3.85494471e-01 -1.78360432e-01 -4.98672187e-01 -3.18682015e-01 8.79431590e-02 5.40320218e-01 6.75510094e-02 3.40490997e-01 -4.30646896e-01 -2.40953684e-01 6.69442356e-01 -2.49131888e-01 2.70673889e-03 1.34079075e+00 2.58783400e-01 -2.11708024e-01 6.35570168e-01 1.34739256e+00 2.22403947e-02 -8.36080730e-01 2.50500113e-01 -1.40377939e-01 -2.69108802e-01 -1.10949449e-01 -8.07025373e-01 -1.18249416e+00 7.33160317e-01 1.06487763e+00 7.54941180e-02 1.30073440e+00 -1.08487874e-01 2.18325019e-01 3.08457552e-03 2.14516800e-02 -8.63704681e-01 9.75034758e-02 -7.32999668e-02 5.74480832e-01 -1.28139400e+00 1.06900111e-01 -5.24337888e-01 -2.73124248e-01 1.22633088e+00 4.52577889e-01 -1.12106234e-01 2.83434808e-01 5.22415638e-01 5.62143803e-01 -3.09232086e-01 -5.08735478e-01 -7.35012144e-02 2.60258555e-01 8.18390548e-01 5.15837252e-01 8.17663372e-02 -5.95104277e-01 -9.36986804e-02 -3.95052522e-01 5.66824228e-02 5.92357039e-01 9.17205632e-01 -6.95126280e-02 -1.12955594e+00 -5.11393428e-01 6.21211350e-01 -7.32632220e-01 2.83452153e-01 -5.22382379e-01 1.32919860e+00 2.99209058e-01 5.30080378e-01 1.37542427e-01 -3.91818345e-01 2.03476384e-01 5.00650704e-02 6.91695988e-01 -6.37668073e-01 -9.26145375e-01 7.96969682e-02 -2.02636898e-01 -1.52488858e-01 -8.32277834e-01 -7.52080560e-01 -1.07839286e+00 -3.15347970e-01 -4.87557292e-01 1.33116305e-01 7.82929420e-01 7.02619493e-01 -1.97767362e-01 9.16425109e-01 4.50726718e-01 -6.49800420e-01 -9.64927003e-02 -1.11324966e+00 -5.93603015e-01 3.48452240e-01 4.46790308e-01 -2.55633801e-01 -5.69330692e-01 5.31104147e-01]
[15.032538414001465, -3.01418137550354]
2c5dbde1-d665-4809-af30-42650dc31a9b
drug-synergistic-combinations-predictions-via
2301.05931
null
https://arxiv.org/abs/2301.05931v1
https://arxiv.org/pdf/2301.05931v1.pdf
Drug Synergistic Combinations Predictions via Large-Scale Pre-Training and Graph Structure Learning
Drug combination therapy is a well-established strategy for disease treatment with better effectiveness and less safety degradation. However, identifying novel drug combinations through wet-lab experiments is resource intensive due to the vast combinatorial search space. Recently, computational approaches, specifically deep learning models have emerged as an efficient way to discover synergistic combinations. While previous methods reported fair performance, their models usually do not take advantage of multi-modal data and they are unable to handle new drugs or cell lines. In this study, we collected data from various datasets covering various drug-related aspects. Then, we take advantage of large-scale pre-training models to generate informative representations and features for drugs, proteins, and diseases. Based on that, a message-passing graph is built on top to propagate information together with graph structure learning flexibility. This is first introduced in the biological networks and enables us to generate pseudo-relations in the graph. Our framework achieves state-of-the-art results in comparison with other deep learning-based methods on synergistic prediction benchmark datasets. We are also capable of inferencing new drug combination data in a test on an independent set released by AstraZeneca, where 10% of improvement over previous methods is observed. In addition, we're robust against unseen drugs and surpass almost 15% AU ROC compared to the second-best model. We believe our framework contributes to both the future wet-lab discovery of novel drugs and the building of promising guidance for precise combination medicine.
['Yu Li', 'Le Song', 'Xin Gao', 'Irwin King', 'Taifeng Wang', 'Yucheng Guo', 'Qinze Yu', 'Zhihang Hu']
2023-01-14
null
null
null
null
['graph-structure-learning']
['graphs']
[ 3.50655466e-01 -1.73459694e-01 -5.83593249e-01 5.53641878e-02 -6.99570954e-01 -4.93139118e-01 4.92248744e-01 6.91741407e-01 1.27615303e-01 1.32271874e+00 3.59818600e-02 -6.07458055e-01 -5.63372135e-01 -9.71300602e-01 -8.14607680e-01 -8.69622171e-01 -2.53142893e-01 8.16693902e-01 -5.62228151e-02 -3.80068243e-01 -5.85439540e-02 7.37367570e-01 -9.52414215e-01 5.44451952e-01 1.03647590e+00 5.16603410e-01 -4.58511338e-03 3.15621018e-01 1.04400955e-01 4.23665583e-01 -2.03720495e-01 -3.46329838e-01 4.16429453e-02 -4.46161568e-01 -5.78366578e-01 -1.25960633e-01 -5.16865440e-02 -3.03249713e-02 -4.74566668e-01 6.01459146e-01 9.59693134e-01 -3.36646974e-01 5.71090698e-01 -9.39888477e-01 -5.26355922e-01 5.68915963e-01 -4.22192544e-01 -8.11365098e-02 4.69572335e-01 5.69309592e-01 1.07326186e+00 -7.05351114e-01 9.33896363e-01 8.74770403e-01 6.36200011e-01 5.92486620e-01 -1.37026286e+00 -6.85721576e-01 -2.37439513e-01 2.97913998e-01 -1.11773968e+00 -1.96144432e-01 5.49919665e-01 -3.65794450e-01 1.21892321e+00 3.90189976e-01 8.68095219e-01 1.27582276e+00 5.02160192e-01 6.50335729e-01 9.41566229e-01 -1.19062528e-01 2.90378690e-01 -6.41028136e-02 -1.69971362e-01 6.77965462e-01 4.18843716e-01 3.31197381e-01 -5.16466916e-01 -4.63324726e-01 4.10210460e-01 3.52577418e-01 -4.44488376e-01 -3.34986120e-01 -1.02553642e+00 8.55950058e-01 6.04289412e-01 5.47541738e-01 -5.88408113e-01 -2.65261829e-02 1.09007180e-01 -1.74317136e-02 3.69022608e-01 9.82285857e-01 -7.93659687e-01 2.67107069e-01 -5.88568926e-01 2.23230883e-01 8.09599936e-01 3.51634711e-01 3.24539840e-01 -2.71806270e-01 -1.69554785e-01 6.46639526e-01 2.52321899e-01 5.17752655e-02 2.95429856e-01 -2.18695417e-01 2.16285199e-01 9.26125944e-01 -1.20468803e-01 -8.57249975e-01 -1.00086534e+00 -1.00783610e+00 -9.69776750e-01 -7.96525776e-02 2.49346793e-01 -2.13220373e-01 -8.44219685e-01 1.62074125e+00 3.80148053e-01 4.28618580e-01 1.44929513e-01 4.86911237e-01 9.11705494e-01 3.84193987e-01 2.98502713e-01 -5.24915159e-01 1.18782055e+00 -6.28109634e-01 -6.26894951e-01 2.30371103e-01 9.87925410e-01 -5.76778710e-01 4.04535383e-01 5.68844557e-01 -7.66133845e-01 -3.98666747e-02 -1.15766084e+00 3.63876849e-01 -7.61099815e-01 -2.04579994e-01 1.24468327e+00 5.08499265e-01 -8.05456161e-01 1.11260927e+00 -6.70514047e-01 -1.97570652e-01 1.07338619e+00 9.24221396e-01 -5.89212596e-01 -3.55409265e-01 -1.47106302e+00 1.06836987e+00 2.62598306e-01 -3.52509622e-03 -9.19459522e-01 -1.15194011e+00 -6.44883633e-01 -6.16179220e-02 2.37830013e-01 -1.21791744e+00 5.09256124e-01 -1.90509468e-01 -1.57653213e+00 3.74330848e-01 1.37139797e-01 -5.82646728e-01 3.42820764e-01 2.15853959e-01 -4.75844413e-01 1.32256113e-02 -3.07710946e-01 5.39283872e-01 9.11528170e-02 -8.88697684e-01 -2.63336092e-01 -5.38143575e-01 -7.20563829e-02 7.36728907e-02 -2.61621296e-01 -4.24613804e-01 -1.55606583e-01 -4.20016408e-01 -3.53832990e-01 -1.01423442e+00 -6.23951674e-01 -1.54869109e-01 -6.93099380e-01 -1.31620094e-01 6.99447334e-01 -3.76393229e-01 1.25338995e+00 -1.65026903e+00 4.42059666e-01 2.47429356e-01 7.25362122e-01 6.36913717e-01 -3.75709087e-01 8.74348700e-01 -4.45477337e-01 3.14756662e-01 -1.04032785e-01 1.10890307e-01 -4.15505767e-01 -1.37766182e-01 5.20789549e-02 5.74123383e-01 3.75340402e-01 1.14907670e+00 -1.06034875e+00 -2.36491449e-02 3.46719772e-01 7.69329011e-01 -4.80966657e-01 -9.21502560e-02 -7.19574273e-01 7.70918548e-01 -6.45212471e-01 8.26103330e-01 6.99487209e-01 -7.81700075e-01 7.30161726e-01 -1.75440133e-01 3.09204310e-01 6.12394921e-02 -6.27677023e-01 1.56467581e+00 -3.13810349e-01 2.65251175e-02 -7.88860977e-01 -1.03229880e+00 7.66092777e-01 4.11265194e-01 8.47404242e-01 -6.12260222e-01 1.55419365e-01 3.93041342e-01 5.05784333e-01 -2.90985435e-01 -4.04228270e-01 -1.08037986e-01 4.76229429e-01 1.25577720e-02 -8.79500285e-02 -1.24627732e-01 1.88055694e-01 -2.32340917e-02 1.35376716e+00 -4.66080680e-02 3.66353035e-01 -6.21674545e-02 5.28339982e-01 -3.05565223e-02 2.96494335e-01 3.13942671e-01 2.12026179e-01 1.19345285e-01 5.92676640e-01 -7.43503511e-01 -7.05726683e-01 -5.65732181e-01 -3.86254162e-01 4.54124510e-01 -8.92190486e-02 -6.48093283e-01 -3.70360196e-01 -7.41118073e-01 1.40000924e-01 5.61423481e-01 -7.55822182e-01 -2.48252332e-01 -1.99729562e-01 -1.47597146e+00 4.68969971e-01 1.37394369e-01 1.42869562e-01 -5.02891183e-01 2.75598079e-01 6.80004716e-01 3.16658765e-01 -9.30290282e-01 7.40191787e-02 3.82530808e-01 -8.86333883e-01 -1.56868386e+00 -5.28199494e-01 -4.68295217e-01 4.02242750e-01 -3.68448310e-02 8.38029623e-01 2.76516348e-01 -6.27254009e-01 -3.51849914e-01 -4.66322266e-02 -4.57296491e-01 -5.57799399e-01 1.48070231e-02 -1.98941734e-02 -2.13118836e-01 3.93603683e-01 -6.84894443e-01 -9.14378941e-01 3.51440132e-01 -7.67419398e-01 1.04131117e-01 7.67766654e-01 1.14559543e+00 7.73580611e-01 1.66999072e-01 8.48678946e-01 -1.15438783e+00 5.13086438e-01 -7.23900914e-01 -5.54102063e-01 4.13053930e-01 -8.31735492e-01 1.91380903e-01 7.44290888e-01 -2.65714854e-01 -5.00558555e-01 3.75257462e-01 -2.58069038e-01 9.50229540e-03 1.90427247e-02 7.75892198e-01 -1.80645987e-01 -3.56208295e-01 7.72937953e-01 -3.71887088e-02 2.55598396e-01 -3.30915153e-01 2.68552214e-01 4.02055979e-01 -3.02930951e-01 -1.50723711e-01 3.52093548e-01 2.55699217e-01 8.04618478e-01 -5.44307709e-01 -5.44236839e-01 -1.98896483e-01 -2.20595777e-01 2.13870510e-01 5.96959114e-01 -8.26942265e-01 -1.11716330e+00 4.57207918e-01 -9.62853134e-01 -2.34081194e-01 2.46182531e-01 5.37036657e-01 -3.06708157e-01 3.21557939e-01 -7.22216547e-01 -2.83021837e-01 -4.88203317e-01 -1.58805776e+00 8.97961497e-01 4.25654799e-02 -3.00659835e-01 -1.45180702e+00 3.19831729e-01 4.92053092e-01 4.00128156e-01 7.75331140e-01 1.19679022e+00 -1.04457879e+00 -7.07394123e-01 -3.33889455e-01 -7.99001195e-03 -1.33242518e-01 4.77273375e-01 3.28192860e-02 -7.32347846e-01 -1.71636626e-01 -4.88806814e-01 -2.88006932e-01 8.31876695e-01 4.88509566e-01 1.30151916e+00 -2.20055029e-01 -1.03373218e+00 4.55426216e-01 1.50877476e+00 4.29881871e-01 8.18181694e-01 2.15459038e-02 6.11104488e-01 1.67034492e-01 2.89210707e-01 5.16611457e-01 1.14740893e-01 9.01794851e-01 5.17795503e-01 -4.56827313e-01 -1.70319110e-01 -1.97530642e-01 -1.24823779e-01 1.98559836e-01 -2.66196787e-01 -6.39248669e-01 -9.95121479e-01 7.70668983e-02 -1.79388392e+00 -8.02627683e-01 -4.34539706e-01 2.09577203e+00 1.26275635e+00 7.71064237e-02 5.70521578e-02 -5.67473210e-02 4.29182529e-01 -3.21842521e-01 -7.90075064e-01 -8.87183920e-02 -4.22318518e-01 6.79838777e-01 4.55190837e-01 4.46800947e-01 -9.86579418e-01 1.02394760e+00 6.54666519e+00 1.14693677e+00 -1.28310871e+00 -3.55153739e-01 1.04790771e+00 2.16744423e-01 -4.63476807e-01 -1.47832721e-01 -7.83095300e-01 3.74834955e-01 9.66826141e-01 -2.23526335e-03 4.03245091e-01 4.04952407e-01 3.64539325e-01 4.44036312e-02 -1.43027854e+00 8.08946192e-01 -1.92458242e-01 -2.16173220e+00 2.56465375e-01 4.17826086e-01 7.89045215e-01 7.17945546e-02 1.22970760e-01 -1.11968867e-01 5.99278152e-01 -1.46155930e+00 -4.69772637e-01 5.14295518e-01 6.86591864e-01 -6.53193593e-01 9.17883575e-01 5.96800409e-02 -6.54702425e-01 1.86131503e-02 -4.69108112e-02 -6.68920111e-05 -1.07711472e-01 8.83035123e-01 -1.47729778e+00 9.70624387e-01 4.34816591e-02 1.15322685e+00 -6.02657020e-01 1.24461544e+00 -8.10014829e-02 3.24794799e-01 -2.21443951e-01 -2.83069730e-01 6.20278865e-02 -3.84704173e-02 1.90991223e-01 8.90936136e-01 2.33605251e-01 4.47574817e-02 1.02872737e-01 5.53338289e-01 -1.71172783e-01 1.83656320e-01 -8.31905007e-01 -4.44582880e-01 2.85946608e-01 1.10802889e+00 -4.73188251e-01 -2.70541549e-01 -1.25989288e-01 6.10450745e-01 1.29402444e-01 1.93391114e-01 -8.69636714e-01 1.38716390e-02 5.99451065e-01 6.03767894e-02 -6.43506320e-03 1.33438423e-01 -3.68615001e-01 -7.32827365e-01 -6.65748060e-01 -1.16653335e+00 5.41700900e-01 -3.88453513e-01 -1.37771523e+00 6.14367902e-01 -3.88176024e-01 -1.21168816e+00 -9.00981575e-02 -7.66287863e-01 -2.36314327e-01 7.05036283e-01 -1.48463774e+00 -1.18099201e+00 8.18653256e-02 3.99675369e-01 2.37057507e-01 -3.03056061e-01 1.16029775e+00 4.29078341e-01 -7.37754166e-01 5.17638445e-01 3.53689611e-01 -4.85629708e-01 6.20996237e-01 -9.24053431e-01 1.98044106e-02 1.61974832e-01 3.55939791e-02 5.81621706e-01 7.89797485e-01 -8.02105129e-01 -1.62111998e+00 -9.13928151e-01 6.92724943e-01 -6.32830143e-01 9.83908653e-01 -1.36965260e-01 -7.08376586e-01 1.83031768e-01 1.57621726e-01 -9.33120549e-02 1.37542892e+00 1.00130334e-01 -2.38414437e-01 -4.73526195e-02 -1.01151669e+00 6.70581877e-01 1.02812111e+00 -1.22646615e-02 -6.63020536e-02 1.17291808e+00 7.88525999e-01 -4.79145020e-01 -1.14020920e+00 7.42364585e-01 4.22426879e-01 -7.26136625e-01 1.04132390e+00 -1.05889523e+00 5.80846071e-01 -4.33419138e-01 1.64651930e-01 -1.47224951e+00 -4.00554568e-01 -6.48923993e-01 -2.70052720e-02 5.09266138e-01 1.01959264e+00 -7.33403325e-01 7.50899792e-01 5.29728949e-01 -1.05289981e-01 -1.52899170e+00 -6.99941158e-01 -5.62402189e-01 -6.35120422e-02 -8.95666480e-02 7.94865668e-01 1.05575454e+00 3.28720927e-01 6.11646891e-01 -4.82029229e-01 -1.02807745e-01 3.42605919e-01 6.24205582e-02 6.50199234e-01 -1.26067173e+00 -6.75492048e-01 -6.51964724e-01 -4.14489239e-01 -3.84723186e-01 1.53611396e-02 -1.04644907e+00 -6.87095463e-01 -1.74286056e+00 4.16411549e-01 -4.79012311e-01 -4.99326766e-01 7.88569808e-01 -1.63039386e-01 1.18662737e-01 -2.66460359e-01 -5.77205941e-02 -4.48741734e-01 3.82981509e-01 1.60645974e+00 -5.50819576e-01 -3.49673927e-01 7.76501223e-02 -8.88185561e-01 2.22335607e-01 8.56608033e-01 -3.60289037e-01 -4.79130030e-01 8.49462599e-02 4.11931038e-01 2.37185940e-01 1.63652226e-01 -6.53520823e-01 7.67803714e-02 -3.56728017e-01 5.85507572e-01 -3.18333447e-01 2.10435644e-01 -7.47534335e-01 8.97450387e-01 1.06262231e+00 -1.57568529e-01 -4.27858144e-01 3.57722342e-01 7.27156699e-01 -5.34121767e-02 2.98123121e-01 6.43716276e-01 -2.33045071e-02 -1.97648272e-01 7.06694007e-01 -8.46006498e-02 -5.63921750e-01 1.09688425e+00 -4.07384522e-02 -6.57267034e-01 -1.32226080e-01 -7.32002079e-01 1.81840882e-01 2.26607099e-01 2.25005075e-01 5.81967115e-01 -1.10342515e+00 -6.92697704e-01 -1.14446804e-01 2.10299626e-01 -2.31422737e-01 4.35989171e-01 1.06264579e+00 -6.55995488e-01 6.61704183e-01 -9.84213501e-02 -5.04208505e-01 -1.13884223e+00 9.13221955e-01 4.00427014e-01 -7.17029929e-01 -2.91806161e-01 6.91284359e-01 1.66757897e-01 -9.46031734e-02 -1.11704759e-01 3.37771028e-02 -3.40896726e-01 -7.47531578e-02 4.10303086e-01 5.11914939e-02 3.65317583e-01 -1.30826980e-01 -6.08716130e-01 4.69034493e-01 -3.22322249e-01 7.03617990e-01 1.77765012e+00 5.36746204e-01 -2.50613332e-01 -1.63143024e-01 1.17326856e+00 -7.57549182e-02 -7.65192211e-01 2.36685853e-02 -7.50150681e-02 -2.13647544e-01 -1.05766699e-01 -1.37369895e+00 -1.03059435e+00 5.29700041e-01 6.24381840e-01 -1.36013925e-01 9.89917815e-01 4.45977338e-02 7.66001344e-01 4.42485839e-01 2.79808700e-01 -7.12241471e-01 1.53528497e-01 1.38155788e-01 9.78207469e-01 -1.28765869e+00 4.55412090e-01 -7.47312963e-01 -3.95509541e-01 1.22849000e+00 2.83912182e-01 2.63784379e-01 7.02771008e-01 1.18715212e-01 -2.94217885e-01 -5.36073327e-01 -1.06684029e+00 -6.94689527e-02 2.86719471e-01 6.87365353e-01 6.71260774e-01 2.57474333e-01 -6.18254304e-01 5.39970696e-01 3.00127566e-01 3.12255502e-01 3.04449260e-01 5.22323310e-01 -1.18663713e-01 -1.81753457e+00 2.41915300e-01 5.09627879e-01 -4.09416825e-01 -3.38094622e-01 -5.44431806e-01 8.12673986e-01 7.67689496e-02 8.74040842e-01 -6.27189875e-01 -6.07325792e-01 1.89191520e-01 -1.25350937e-01 5.84888518e-01 -4.82690901e-01 -5.07964909e-01 1.56062350e-01 3.28053921e-01 -4.81954902e-01 -3.58635008e-01 -2.51404077e-01 -1.14984858e+00 -2.85981297e-01 -5.33243716e-01 1.20883144e-01 5.84072530e-01 8.86550307e-01 9.76583481e-01 7.09783375e-01 7.58195996e-01 -3.75045896e-01 -1.74574032e-01 -5.85956872e-01 -4.23518717e-01 1.97534263e-01 1.88542679e-02 -7.27624416e-01 -1.04821175e-01 -3.10158134e-01]
[5.334403991699219, 5.772904396057129]
316ff7df-ffe3-462d-933e-f87997d6a6a1
using-a-waffle-iron-for-automotive-point
2301.10100
null
https://arxiv.org/abs/2301.10100v1
https://arxiv.org/pdf/2301.10100v1.pdf
Using a Waffle Iron for Automotive Point Cloud Semantic Segmentation
Semantic segmentation of point clouds in autonomous driving datasets requires techniques that can process large numbers of points over large field of views. Today, most deep networks designed for this task exploit 3D sparse convolutions to reduce memory and computational loads. The best methods then further exploit specificities of rotating lidar sampling patterns to further improve the performance, e.g., cylindrical voxels, or range images (for feature fusion from multiple point cloud representations). In contrast, we show that one can build a well-performing point-based backbone free of these specialized tools. This backbone, WaffleIron, relies heavily on generic MLPs and dense 2D convolutions, making it easy to implement, and contains just a few parameters easy to tune. Despite its simplicity, our experiments on SemanticKITTI and nuScenes show that WaffleIron competes with the best methods designed specifically for these autonomous driving datasets. Hence, WaffleIron is a strong, easy-to-implement, baseline for semantic segmentation of sparse outdoor point clouds.
['Renaud Marlet', 'Alexandre Boulch', 'Gilles Puy']
2023-01-24
null
null
null
null
['robust-3d-semantic-segmentation']
['computer-vision']
[-9.48955268e-02 -5.87263033e-02 -6.35818206e-03 -5.97950697e-01 -4.58066761e-01 -6.58330798e-01 6.73383296e-01 -7.02971965e-02 -6.08204722e-01 3.26049447e-01 -5.56070626e-01 -3.62117320e-01 -3.57710458e-02 -1.18892574e+00 -1.24081218e+00 -4.09504086e-01 -7.67630860e-02 1.00062680e+00 7.16834903e-01 -2.99035668e-01 1.56566113e-01 1.00373161e+00 -2.17425752e+00 4.10506828e-03 7.64361382e-01 1.07897747e+00 3.55820954e-01 5.93588829e-01 -7.05980182e-01 2.15434805e-01 -3.23187232e-01 -3.40123683e-01 5.53746164e-01 5.28171480e-01 -6.37243867e-01 -2.36698404e-01 9.75776017e-01 -3.34373772e-01 -2.04920158e-01 8.74202847e-01 3.37976664e-01 1.28725797e-01 4.19532895e-01 -1.28556633e+00 -1.38071448e-01 3.81674439e-01 -4.36509639e-01 4.40186448e-02 -2.46994838e-01 3.40743512e-01 7.80242085e-01 -1.06221271e+00 5.29651880e-01 1.36412811e+00 1.23152161e+00 3.29241961e-01 -1.09683323e+00 -6.99231088e-01 2.07220927e-01 -3.58576048e-03 -1.19541419e+00 -3.04340303e-01 6.33871675e-01 -3.62069637e-01 1.22180736e+00 2.27202669e-01 8.29067647e-01 9.81786132e-01 -6.39092550e-02 6.90700412e-01 8.50936592e-01 5.33737577e-02 3.58494133e-01 6.78221183e-03 2.02870533e-01 6.41992152e-01 4.28278983e-01 1.87255159e-01 -3.95328999e-01 -9.25986245e-02 8.12124372e-01 2.32289851e-01 1.97296739e-02 -7.90224373e-01 -1.12124801e+00 9.65120912e-01 7.30613291e-01 -3.46486792e-02 -1.02942936e-01 5.88402987e-01 3.11963677e-01 6.75325468e-02 4.43555027e-01 3.16604435e-01 -5.85223198e-01 -1.29633188e-01 -1.18830800e+00 6.69574082e-01 8.18614066e-01 1.16796291e+00 1.28856039e+00 -1.21946439e-01 2.99393654e-01 6.96376383e-01 2.41349146e-01 8.66112053e-01 8.97938162e-02 -1.45910251e+00 4.86307949e-01 4.51870382e-01 4.22533415e-02 -7.33007133e-01 -5.27709424e-01 -3.59267175e-01 -4.89484310e-01 6.78440034e-01 3.94427359e-01 2.97107771e-02 -1.24495876e+00 1.09973264e+00 3.71579945e-01 2.97716320e-01 -9.82844979e-02 9.25285459e-01 9.97345030e-01 4.69865739e-01 -1.15742132e-01 8.18899453e-01 1.21965027e+00 -9.11583960e-01 6.01235665e-02 -6.58267677e-01 5.98770678e-01 -3.98904592e-01 1.09676147e+00 2.76753306e-01 -9.87500429e-01 -6.17385864e-01 -1.18000269e+00 -3.74664247e-01 -6.68540955e-01 -1.52200699e-01 1.10290694e+00 7.42106974e-01 -1.30828273e+00 9.29635882e-01 -1.15505433e+00 -4.03893203e-01 8.99433374e-01 6.03363812e-01 -2.22170219e-01 -1.68661281e-01 -7.16722190e-01 8.34026575e-01 2.26468652e-01 7.79429525e-02 -6.69450819e-01 -9.84962344e-01 -7.78122783e-01 2.25310288e-02 2.43524626e-01 -1.05735517e+00 1.16100287e+00 -6.93316817e-01 -1.41809726e+00 1.09506917e+00 -1.69588342e-01 -8.79080653e-01 5.97303152e-01 -3.87372524e-01 1.46800414e-01 2.23134100e-01 7.72052258e-02 1.36297250e+00 7.71170378e-01 -1.24200058e+00 -6.19628966e-01 -5.47198236e-01 2.31251866e-01 1.19282298e-01 7.72191808e-02 -3.66780877e-01 -4.42232639e-01 -3.55707332e-02 3.33126128e-01 -1.05846024e+00 -5.82156897e-01 2.56119579e-01 -1.65239498e-01 -2.53758073e-01 9.96365786e-01 -3.56417522e-02 8.32052305e-02 -2.18466330e+00 -1.63004622e-01 2.61237323e-01 2.85354137e-01 1.90435514e-01 -1.16012245e-01 4.78386506e-02 1.47217765e-01 2.53595095e-02 -4.30712044e-01 -6.20029449e-01 1.58991814e-01 6.66248918e-01 -3.50945175e-01 5.07583618e-01 4.03954566e-01 1.03584826e+00 -6.50467038e-01 -4.17844981e-01 5.70228457e-01 6.15493894e-01 -7.66715467e-01 -2.00070798e-01 -5.00582635e-01 1.53800160e-01 -5.33505678e-01 5.48063636e-01 1.19042802e+00 -1.71801701e-01 -5.23033679e-01 5.76679967e-02 -3.61935645e-01 3.50262433e-01 -1.07551491e+00 2.12746644e+00 -5.04475296e-01 7.09215105e-01 4.23645079e-01 -9.35497224e-01 9.52029765e-01 -1.90466553e-01 6.05798364e-01 -4.05811250e-01 1.31152779e-01 4.54072833e-01 -4.25754249e-01 -1.12904266e-01 6.83478475e-01 6.97369082e-03 6.49150759e-02 -5.28641837e-03 2.11658478e-01 -8.70656967e-01 6.50161505e-02 2.38771904e-02 1.08580327e+00 4.78989422e-01 -4.15401250e-01 -2.86017448e-01 3.23052108e-01 4.83769894e-01 2.63240337e-01 9.64498639e-01 -9.38101262e-02 8.55710268e-01 1.61597282e-01 -7.60629952e-01 -1.08119249e+00 -1.10834718e+00 -3.30080926e-01 9.21989322e-01 4.31624353e-01 -3.22079062e-01 -7.87243009e-01 -5.39940417e-01 3.97400349e-01 5.76641679e-01 -1.68707579e-01 2.78212190e-01 -7.71915734e-01 -3.90085578e-01 5.88088870e-01 6.90244257e-01 5.33334851e-01 -8.39359224e-01 -8.90166640e-01 1.38682857e-01 2.48437628e-01 -1.33006525e+00 3.04061413e-01 6.85993791e-01 -1.26561236e+00 -9.04685974e-01 -4.98168111e-01 -5.48495173e-01 2.25356579e-01 6.20304644e-01 1.48834693e+00 -5.51360287e-02 -1.02026939e-01 3.71402144e-01 -4.32450399e-02 -7.81913459e-01 9.81105492e-02 4.27474469e-01 -2.51966834e-01 -6.19265437e-01 7.43603289e-01 -9.37754273e-01 -5.08910120e-01 2.26880759e-01 -5.93248487e-01 -1.20884091e-01 5.68955779e-01 3.80681008e-01 7.61921763e-01 -1.78080633e-01 -8.08765888e-02 -7.78171837e-01 4.72618995e-05 -2.96750396e-01 -1.04086244e+00 -4.97402221e-01 -2.78999507e-01 1.60560906e-02 5.03298223e-01 -8.29923898e-02 -4.70930547e-01 4.43156898e-01 -5.83959997e-01 -8.38026464e-01 -6.84758425e-01 -1.29977055e-02 1.50967851e-01 -5.55135787e-01 8.44241977e-01 3.97449322e-02 3.36396769e-02 -5.51605344e-01 7.58615375e-01 3.43118250e-01 6.76829755e-01 -6.63088858e-01 9.79054511e-01 1.06533515e+00 1.44432127e-01 -9.39494729e-01 -6.12604499e-01 -6.09723568e-01 -7.76482940e-01 9.79486555e-02 9.66246188e-01 -1.16846132e+00 -6.93538606e-01 3.16978306e-01 -1.29974556e+00 -3.94543350e-01 -6.61240220e-01 2.59834856e-01 -8.57978702e-01 1.21670954e-01 -3.36037487e-01 -5.25286257e-01 -1.75786644e-01 -1.27757668e+00 1.63654268e+00 6.42369315e-02 7.22669661e-02 -6.54733300e-01 -2.03839600e-01 3.58432710e-01 4.69832480e-01 2.58897841e-01 5.22134542e-01 -4.64739054e-01 -1.19609559e+00 -7.08708614e-02 -4.24318761e-01 4.08116072e-01 -5.80220580e-01 3.25908773e-02 -1.35066426e+00 1.28140170e-02 6.79301322e-02 -2.60357469e-01 1.15098751e+00 5.14738917e-01 1.49631715e+00 2.02007741e-01 -6.21155620e-01 1.13293278e+00 1.33237386e+00 -2.05253229e-01 5.92860103e-01 4.39750224e-01 1.01448500e+00 5.43603241e-01 3.08298856e-01 -8.93067382e-03 7.18974471e-01 5.46378791e-01 8.28648269e-01 -3.25963467e-01 -7.12443218e-02 1.67958066e-02 2.11733282e-02 3.86927605e-01 -1.74313530e-01 1.58282369e-01 -1.05961680e+00 5.08843124e-01 -1.65930533e+00 -7.87740827e-01 -5.16435981e-01 1.86782467e+00 2.54409820e-01 4.65964019e-01 -3.57958488e-02 8.67499560e-02 2.27759913e-01 3.25695962e-01 -5.89392126e-01 -4.58418220e-01 -5.38327135e-02 7.47465134e-01 1.15755522e+00 3.06456804e-01 -1.31041193e+00 1.16584134e+00 6.39314079e+00 9.00090635e-01 -1.21657956e+00 2.26423711e-01 2.36871094e-01 -2.76313305e-01 -3.94100875e-01 -3.60872224e-02 -1.04700840e+00 3.63486946e-01 1.12330174e+00 4.35352296e-01 3.69809747e-01 1.25411797e+00 2.70438697e-02 -2.37480439e-02 -1.05512094e+00 1.13912332e+00 -3.05969983e-01 -1.67405820e+00 -2.22792238e-01 9.79029760e-02 4.81418967e-01 1.07042718e+00 -9.71085756e-05 3.25350493e-01 5.98477602e-01 -1.17048538e+00 8.86153281e-01 2.65453577e-01 5.32366753e-01 -7.57038832e-01 5.45305967e-01 3.80173832e-01 -1.02212751e+00 1.73553620e-02 -6.96891367e-01 -9.01070163e-02 8.44633505e-02 7.76139379e-01 -5.33939123e-01 3.69069248e-01 1.12759805e+00 7.21283376e-01 -5.21133006e-01 1.02829123e+00 -7.09130690e-02 3.43692362e-01 -9.99828458e-01 7.25352615e-02 6.40933990e-01 -2.67672658e-01 5.57693243e-01 1.13775969e+00 3.74869466e-01 -2.45431885e-01 2.92047560e-01 1.12061310e+00 3.60560976e-02 -2.42726818e-01 -9.87309515e-01 2.62267262e-01 3.97770673e-01 1.30794036e+00 -8.11288178e-01 -2.97552645e-01 -4.91256416e-01 5.80493271e-01 3.26362818e-01 1.53649747e-01 -8.53422821e-01 -4.10137773e-01 1.13317204e+00 4.30061966e-01 7.92589188e-01 -6.90109134e-01 -7.45182335e-01 -9.11569715e-01 5.59101477e-02 -4.93470967e-01 -1.29774988e-01 -9.55192566e-01 -1.06439424e+00 5.69393396e-01 3.19601893e-02 -1.03583312e+00 -1.17715128e-01 -8.79393518e-01 -5.56941748e-01 9.34989691e-01 -1.99969625e+00 -1.13436294e+00 -6.08734846e-01 6.90871179e-01 4.78346467e-01 1.82113349e-01 6.02222502e-01 2.85623252e-01 -1.06399268e-01 9.68165398e-02 2.96054664e-03 -1.50259271e-01 1.94115564e-01 -1.26827347e+00 1.03861165e+00 6.24573171e-01 2.63724446e-01 4.77173209e-01 6.14637494e-01 -3.61321658e-01 -1.52841306e+00 -1.18119526e+00 4.16656643e-01 -7.49516249e-01 5.32113194e-01 -7.17440546e-01 -9.07049894e-01 6.53400302e-01 -1.22464262e-01 2.44015634e-01 1.54953644e-01 1.81105688e-01 -3.64770681e-01 -1.30021900e-01 -1.15249741e+00 3.83328944e-01 1.47312796e+00 -3.35744262e-01 -4.75080550e-01 6.96042597e-01 9.38861728e-01 -7.75712311e-01 -6.59466684e-01 4.91617173e-01 2.04473436e-01 -1.35034156e+00 1.43171310e+00 -1.31211057e-01 3.50376785e-01 -4.24766243e-01 -1.66228056e-01 -1.07158101e+00 -1.13933831e-01 -3.61134350e-01 9.52058956e-02 6.68942392e-01 2.19976455e-01 -8.65847230e-01 1.32095313e+00 3.27542901e-01 -7.03094065e-01 -5.35684407e-01 -1.00579393e+00 -8.60230148e-01 2.34207407e-01 -8.45897257e-01 9.25523341e-01 5.75464189e-01 -8.55601311e-01 9.62172449e-02 3.31528157e-01 2.97856927e-01 6.56358004e-01 3.03924054e-01 1.22083652e+00 -1.73272502e+00 -3.86154018e-02 -5.68999171e-01 -6.71621740e-01 -1.30366707e+00 1.98867694e-01 -9.88555253e-01 1.45301044e-01 -1.56466019e+00 -5.68063974e-01 -1.02444398e+00 2.86592454e-01 4.46543694e-01 2.43444413e-01 4.47456390e-01 1.67610705e-01 2.64384240e-01 -4.12150651e-01 4.94187176e-01 1.03484678e+00 -2.04145059e-01 -2.25367919e-01 4.24294099e-02 -4.40367043e-01 9.68900621e-01 7.31141448e-01 -5.73930621e-01 -2.41637245e-01 -8.51997077e-01 1.20302364e-01 -4.47011203e-01 7.66422868e-01 -1.55945003e+00 1.74122348e-01 2.45705936e-02 3.76559258e-01 -1.12025321e+00 8.65377426e-01 -8.80265951e-01 1.24288402e-01 1.09046154e-01 3.67448211e-01 -2.35117655e-02 3.62476796e-01 4.04887199e-01 -1.47019789e-01 -2.60664880e-01 6.89427435e-01 -5.30271113e-01 -8.10450017e-01 5.18448293e-01 -1.02703519e-01 -1.38478458e-01 8.04287791e-01 -7.66598165e-01 -3.07163298e-01 4.20228206e-02 -5.02926886e-01 3.22836459e-01 7.99467087e-01 3.73951286e-01 4.37033981e-01 -1.03128254e+00 -4.12434876e-01 4.02680606e-01 -1.55682303e-02 9.48945165e-01 1.23449363e-01 7.14168608e-01 -1.00427973e+00 4.37705308e-01 -1.40323058e-01 -1.26857674e+00 -8.26236427e-01 4.26848829e-01 4.67337251e-01 9.17139575e-02 -1.18991709e+00 1.19271553e+00 2.09169462e-01 -7.84013033e-01 1.05795935e-01 -8.14233661e-01 8.06160420e-02 -9.70393494e-02 1.01213284e-01 2.01440036e-01 5.72932243e-01 -5.13886094e-01 -4.05862600e-01 7.96504855e-01 2.46509567e-01 5.05850390e-02 1.53862846e+00 1.32848039e-01 -6.21158257e-02 2.88709819e-01 9.95980322e-01 -2.40935579e-01 -1.63944817e+00 3.54187638e-02 -7.40625486e-02 -5.00708997e-01 2.42674381e-01 -2.65629232e-01 -1.26290977e+00 1.23530579e+00 5.32331884e-01 1.58422455e-01 6.21194780e-01 -2.17907391e-02 9.00188863e-01 6.67920351e-01 6.92384362e-01 -7.48956084e-01 -4.20034856e-01 7.56075263e-01 6.05058551e-01 -1.15388131e+00 -7.50336200e-02 -6.42260313e-01 -1.48832366e-01 1.13304377e+00 4.57043707e-01 -6.25873208e-01 7.18081892e-01 5.54688692e-01 1.14029914e-01 -5.80627024e-01 -3.30886841e-01 -4.33008075e-01 -1.70070620e-03 8.94877493e-01 -1.27427593e-01 4.68452834e-02 3.27862293e-01 1.52995154e-01 -8.08909953e-01 8.15271661e-02 2.41063818e-01 8.86076868e-01 -6.39815390e-01 -8.72179270e-01 -5.32630980e-01 6.32671654e-01 3.11588421e-02 -1.32516131e-01 -6.69521168e-02 8.14173043e-01 3.58149678e-01 6.39476120e-01 5.01891732e-01 -2.05289438e-01 5.20523787e-01 8.55445117e-02 4.96541619e-01 -5.10509253e-01 -7.37776995e-01 -2.52557546e-01 -6.17650561e-02 -1.14572501e+00 -4.34680909e-01 -5.99010587e-01 -1.23311985e+00 -5.04844487e-01 -8.64025503e-02 -1.67575747e-01 1.23005462e+00 8.15312326e-01 5.86804092e-01 3.71236056e-01 1.07961759e-01 -1.58319211e+00 -3.25392574e-01 -5.84941804e-01 -3.79249781e-01 5.70509396e-02 2.80701607e-01 -8.82298112e-01 -2.80575961e-01 -2.98146367e-01]
[8.070289611816406, -3.0376627445220947]
46e68fb8-afa7-4d6b-8091-3f484bfb9871
computer-aided-diagnosis-of-lung-carcinoma
1803.05471
null
http://arxiv.org/abs/1803.05471v1
http://arxiv.org/pdf/1803.05471v1.pdf
Computer-aided diagnosis of lung carcinoma using deep learning - a pilot study
Aim: Early detection and correct diagnosis of lung cancer are the most important steps in improving patient outcome. This study aims to assess which deep learning models perform best in lung cancer diagnosis. Methods: Non-small cell lung carcinoma and small cell lung carcinoma biopsy specimens were consecutively obtained and stained. The specimen slides were diagnosed by two experienced pathologists (over 20 years). Several deep learning models were trained to discriminate cancer and non-cancer biopsies. Result: Deep learning models give reasonable AUC from 0.8810 to 0.9119. Conclusion: The deep learning analysis could help to speed up the detection process for the whole-slide image (WSI) and keep the comparable detection rate with human observer.
['Qiang Li', 'Tao Tan', 'Guoping Cai', 'Geert Litjens', 'Yuling Tang', 'Quchang Ouyang', 'Jun Tang', 'Jiaolong Xu', 'Zhi Duan', 'Ping Liu', 'Hui Chen', 'Zheyu Hu', 'Zhang Li']
2018-03-14
null
null
null
null
['lung-cancer-diagnosis']
['medical']
[-1.96325183e-01 -1.33808509e-01 -6.45904422e-01 2.13088155e-01 -9.68719363e-01 -4.53333467e-01 1.56298965e-01 2.55885780e-01 -4.89221126e-01 8.58154058e-01 -1.68590039e-01 -7.58945942e-01 2.41296012e-02 -8.59725475e-01 6.47465363e-02 -1.11760342e+00 9.84228402e-02 1.00179374e+00 4.26866204e-01 4.13834363e-01 -2.33825788e-01 9.96461332e-01 -7.29526281e-01 4.81203288e-01 5.08164585e-01 6.16629541e-01 2.36517385e-01 1.25740135e+00 -1.32292181e-01 9.06266809e-01 -2.93471754e-01 -7.18338415e-02 9.68457162e-02 -3.60581219e-01 -1.01137185e+00 -7.23936930e-02 1.96012020e-01 -6.30452096e-01 -2.51841038e-01 7.70995140e-01 7.39183187e-01 -5.17089307e-01 1.07132864e+00 -5.49824655e-01 -2.25604191e-01 2.45473087e-01 -6.96127117e-01 7.10879803e-01 -2.86008835e-01 3.16462934e-01 7.78898120e-01 -9.18266952e-01 3.06525767e-01 5.33997416e-01 9.11972344e-01 8.25558424e-01 -6.58738017e-01 -7.29128480e-01 -8.88942540e-01 4.13627893e-01 -1.44493508e+00 -2.70546544e-02 -6.18988574e-02 -5.37421703e-01 6.71104252e-01 5.01914144e-01 9.72284436e-01 7.64772713e-01 7.46446371e-01 7.20884204e-01 7.88444459e-01 -2.75117487e-01 -1.67261690e-01 3.33054841e-01 1.42317519e-01 1.09405529e+00 8.30324769e-01 1.34296864e-01 2.24109262e-01 -1.47008881e-01 8.69987547e-01 5.90567291e-01 -1.80223927e-01 2.29320973e-01 -1.30077362e+00 7.38804758e-01 5.75482965e-01 9.07121539e-01 -2.12481380e-01 6.01681545e-02 5.98937333e-01 1.64990708e-01 2.37014908e-02 6.38198704e-02 -7.71567449e-02 3.24659705e-01 -8.12790334e-01 -3.52743804e-01 5.59287965e-01 3.17572057e-01 1.56563282e-01 -1.35091364e-01 -5.09992599e-01 7.45877147e-01 3.31336170e-01 5.54783463e-01 1.03354597e+00 -3.98964554e-01 -1.41472772e-01 5.83233953e-01 -6.60195528e-03 -3.59835953e-01 -6.20527208e-01 -6.95594072e-01 -1.38433230e+00 4.24691178e-02 5.07555306e-01 5.93266189e-02 -1.10216296e+00 6.42786682e-01 3.27273495e-02 2.68263251e-01 -5.57339191e-02 6.35192990e-01 1.01272786e+00 3.61193210e-01 4.47855204e-01 -3.89514305e-02 1.78830016e+00 -9.52091813e-01 -4.68685925e-01 -9.01354328e-02 1.12452388e+00 -4.46389347e-01 7.15395808e-01 1.18755132e-01 -8.10656607e-01 -3.12465012e-01 -7.75333524e-01 -1.19034342e-01 -2.58265078e-01 7.01273620e-01 5.71171284e-01 6.91285849e-01 -1.05709267e+00 2.38196611e-01 -1.07385838e+00 -7.23361075e-01 7.46221125e-01 6.78695261e-01 -4.97629941e-01 -9.82845500e-02 -9.64734316e-01 8.08280349e-01 3.82226348e-01 5.52798696e-02 -1.07663345e+00 -5.15460432e-01 1.05787687e-01 5.97863086e-02 -1.67365879e-01 -1.24190426e+00 1.62174881e+00 -6.80987954e-01 -9.32096839e-01 1.70880020e+00 -2.20159397e-01 -4.96829122e-01 4.05701429e-01 3.10097307e-01 -2.61326253e-01 3.10104668e-01 7.24561363e-02 2.77773768e-01 2.34567791e-01 -8.33148181e-01 -1.19319463e+00 -3.81401479e-01 -5.25496304e-01 2.36390345e-02 -4.18452919e-01 -1.11666709e-01 -5.22162616e-01 -2.80136347e-01 -4.93895590e-01 -9.83998895e-01 -3.38491529e-01 2.93246299e-01 -3.94939870e-01 -5.91081619e-01 7.19462335e-01 -7.62491047e-01 1.15699565e+00 -1.88892901e+00 -4.34306473e-01 1.58031747e-01 7.75432527e-01 7.81952143e-01 1.24040566e-01 3.21246535e-02 4.32350412e-02 5.75891554e-01 2.29984969e-01 1.05802231e-01 -4.07599717e-01 3.71797010e-02 4.99759346e-01 6.44158840e-01 -7.59506524e-02 1.36738658e+00 -7.54381776e-01 -1.08002019e+00 1.24099836e-01 3.26620102e-01 8.80476162e-02 4.61146086e-01 3.32381427e-01 1.45378768e-01 -4.43357110e-01 8.70192647e-01 3.05434406e-01 -7.12229967e-01 1.47583887e-01 -5.50770052e-02 4.00209188e-01 -1.70043275e-01 -3.79085362e-01 7.78502345e-01 -2.84242779e-01 8.18185091e-01 -3.44452113e-02 -6.31580949e-01 6.13073885e-01 7.36105263e-01 4.52585429e-01 -3.84254545e-01 3.93074721e-01 5.27482867e-01 3.89798015e-01 -9.07088280e-01 -5.34566164e-01 -7.29100168e-01 5.34623265e-01 2.79968381e-01 -3.47507596e-01 -4.07045521e-02 1.20605968e-01 -8.34404230e-02 1.22902942e+00 -9.84248519e-01 1.10404432e+00 -2.84617454e-01 8.23774755e-01 1.93434328e-01 4.01282907e-01 6.36133730e-01 -5.87260485e-01 3.78619820e-01 4.14733618e-01 -6.87470555e-01 -7.08322167e-01 -1.12292194e+00 -2.72200882e-01 6.91198647e-01 -4.62980509e-01 4.32484955e-01 -4.06296104e-01 -9.32079256e-01 5.83243258e-02 3.50713491e-01 -8.92728090e-01 -3.81728522e-02 -5.04642010e-01 -6.79788828e-01 6.56113088e-01 8.47004473e-01 3.63276899e-01 -8.55668128e-01 -3.91108274e-01 1.81877479e-01 -1.08308345e-01 -3.97292972e-01 -2.88669080e-01 4.18081820e-01 -1.09046686e+00 -1.64304650e+00 -1.23024201e+00 -1.37424409e+00 7.47400641e-01 5.29022932e-01 1.04431999e+00 6.97469890e-01 -9.54690337e-01 -2.76807453e-02 1.50498167e-01 -7.12372780e-01 -7.67174065e-01 1.69519961e-01 -1.43547967e-01 -4.71364200e-01 9.19121683e-01 -2.95155030e-02 -9.42563891e-01 9.66481566e-02 -8.50280583e-01 -8.17062557e-02 1.24815416e+00 8.08381557e-01 7.83680260e-01 1.87410370e-01 3.68112415e-01 -1.08339894e+00 5.18088639e-01 -5.66269577e-01 -5.53160049e-02 2.99915314e-01 -4.94884193e-01 -4.08810228e-01 8.52623224e-01 -1.65213168e-01 -6.39711320e-01 -9.96899605e-02 -3.24753582e-01 -1.98966905e-01 -4.16534841e-01 2.73813605e-01 3.79141778e-01 5.33589674e-03 7.15442359e-01 8.54673237e-02 1.49755515e-02 2.67168768e-02 -4.86568004e-01 8.17655683e-01 4.13243830e-01 1.77082792e-01 8.21715236e-01 6.04161978e-01 3.41430336e-01 -6.89660013e-01 -8.97875547e-01 -1.08416581e+00 -6.90983713e-01 -2.26790071e-01 1.01061821e+00 -8.75116706e-01 -5.06654918e-01 4.64679271e-01 -7.77339935e-01 -4.26572382e-01 -2.02946320e-01 8.23545396e-01 6.41053123e-03 2.95672804e-01 -1.11166108e+00 -2.79794991e-01 -1.06831181e+00 -6.75890386e-01 9.21990633e-01 4.57188219e-01 -2.53544629e-01 -1.48182690e+00 5.66834629e-01 1.76655442e-01 5.48740506e-01 2.09726319e-01 9.76269960e-01 -1.09068918e+00 -3.10306102e-01 -9.91533875e-01 -5.65223932e-01 3.06603722e-02 5.18123984e-01 4.80834544e-01 -9.88660395e-01 -4.40001100e-01 -6.90047815e-03 -2.55737305e-01 9.78717148e-01 6.07482255e-01 1.21161914e+00 -1.47597104e-01 -1.17033255e+00 5.74879706e-01 1.68109679e+00 6.58420026e-01 5.54130077e-01 3.75943840e-01 6.30841255e-01 1.07563294e-01 2.78893918e-01 -7.75599806e-03 -9.52419862e-02 -1.89440206e-01 2.83094168e-01 -4.97480035e-01 -3.28321904e-01 1.66183338e-02 -1.51963815e-01 6.26550198e-01 -2.50251949e-01 -6.07691526e-01 -1.50625217e+00 8.07834387e-01 -1.02201533e+00 -9.67869639e-01 -5.83967805e-01 1.73158658e+00 6.06747627e-01 2.08986904e-02 1.47959134e-02 2.94982493e-01 6.04369402e-01 -2.92450815e-01 -4.84212250e-01 -4.71736304e-03 2.50559241e-01 3.29451621e-01 5.54075778e-01 2.58064330e-01 -9.92937624e-01 2.87071973e-01 7.02496529e+00 9.81847286e-01 -1.43324757e+00 3.11638806e-02 1.06543195e+00 -2.94161234e-02 -1.53189888e-02 -5.45054495e-01 -8.70960414e-01 9.22456831e-02 9.00679588e-01 -2.97145426e-01 -3.70606720e-01 7.87944615e-01 2.38041997e-01 -2.05691218e-01 -1.20019829e+00 7.41378784e-01 -2.34053046e-01 -1.50745463e+00 9.94427800e-02 3.72698456e-01 5.96953452e-01 1.89524338e-01 -1.38242589e-02 5.00941515e-01 1.75732180e-01 -1.22478700e+00 -4.53319460e-01 7.55931258e-01 1.34238565e+00 -6.32399738e-01 1.62053871e+00 5.60981810e-01 -1.22263348e+00 -9.34962779e-02 -3.13873172e-01 4.04075265e-01 -4.24182653e-01 5.86421609e-01 -1.98919177e+00 5.25763966e-02 5.61148465e-01 4.54230309e-01 -7.73408532e-01 1.04974747e+00 4.84464839e-02 1.03517556e+00 -8.11783746e-02 -5.94355643e-01 2.59166807e-01 3.34356308e-01 -7.63374120e-02 1.56034231e+00 4.39297855e-01 2.28674054e-01 -1.16885006e-01 3.84137183e-01 3.39922719e-02 1.73913449e-01 -4.29722011e-01 -1.48478955e-01 3.40990096e-01 1.53923774e+00 -1.09860694e+00 -4.99977857e-01 -5.64088225e-01 7.08172381e-01 1.72539592e-01 7.82047808e-02 -7.71975279e-01 -2.47370169e-01 4.67352383e-02 6.31943583e-01 3.55879031e-02 5.16940355e-01 -4.48304683e-01 -7.49195039e-01 -7.88900018e-01 -5.89849174e-01 8.77825737e-01 -3.90184850e-01 -1.41721988e+00 3.22365105e-01 -5.07322550e-01 -1.18335044e+00 -5.66667877e-02 -9.82399583e-01 -1.31143844e+00 5.55559754e-01 -1.62938261e+00 -1.03781617e+00 -5.86417913e-01 3.55438560e-01 4.28531647e-01 -3.20624560e-01 1.07521141e+00 4.30650301e-02 -7.10948110e-01 5.58835685e-01 5.09729028e-01 5.43911874e-01 7.00737894e-01 -1.51599419e+00 -1.46777838e-01 3.93393725e-01 -5.00755131e-01 4.29086685e-01 3.33663672e-02 -3.77471417e-01 -9.60216999e-01 -1.31750357e+00 9.99095082e-01 -1.61504626e-01 5.47617435e-01 4.07968074e-01 -7.31597364e-01 5.18473208e-01 2.08434671e-01 1.96569741e-01 1.39525962e+00 -5.12070775e-01 2.11895570e-01 -1.49759650e-01 -1.16337478e+00 4.94829059e-01 1.98803261e-01 -4.56640214e-01 -1.54936194e-01 4.59841818e-01 1.09909186e-02 -2.40748465e-01 -9.80231106e-01 3.94990146e-01 5.79082310e-01 -9.08256769e-01 1.07965910e+00 -4.77669030e-01 2.59384871e-01 -1.92354083e-01 4.47380424e-01 -7.61809349e-01 -1.09516668e+00 1.51042372e-01 5.87224104e-02 6.62608743e-01 4.71359998e-01 -6.72569811e-01 1.24672687e+00 3.29298228e-01 2.88324565e-01 -1.19884980e+00 -7.73659348e-01 -4.89183307e-01 3.85265261e-01 2.40862221e-01 4.22012918e-02 7.30927110e-01 -1.76992372e-01 1.42496273e-01 3.22699726e-01 -8.12137201e-02 3.19150954e-01 -1.20427206e-01 4.37195122e-01 -1.41178036e+00 -7.13016018e-02 -8.83005083e-01 -3.99468780e-01 -4.86542761e-01 -2.40703583e-01 -1.03683007e+00 -2.81856298e-01 -2.12812185e+00 9.87588584e-01 -1.31812230e-01 -6.89622760e-01 5.53058147e-01 -6.23628139e-01 3.19259316e-01 -4.39969808e-01 5.60073078e-01 -2.15647697e-01 -2.00760901e-01 1.42096686e+00 -3.62259448e-01 3.52758110e-01 5.44674993e-01 -7.01905310e-01 7.04471231e-01 1.22189498e+00 -5.97724974e-01 -9.39687788e-02 -8.10005218e-02 -1.94457114e-01 3.55524421e-01 3.65298420e-01 -1.18511832e+00 1.64145559e-01 -4.02824521e-01 8.26791465e-01 -8.55614901e-01 -2.38620624e-01 -9.57912683e-01 1.75771549e-01 1.33870792e+00 -2.58536339e-01 -5.46062112e-01 2.05444723e-01 5.71920872e-01 -1.77227587e-01 -5.67731500e-01 1.19087672e+00 -4.52168375e-01 -3.99430603e-01 7.05906868e-01 -9.79960024e-01 -4.28454936e-01 1.45634627e+00 -5.09704649e-01 -1.13826334e-01 -7.23265633e-02 -5.63763440e-01 1.02646351e-01 1.59607187e-01 -3.27754527e-01 5.70514023e-01 -1.24598825e+00 -8.86883736e-01 -1.09650776e-01 7.02610910e-02 2.59168327e-01 1.78472355e-01 1.21439993e+00 -1.27240360e+00 8.91669631e-01 -2.29203030e-02 -5.78760922e-01 -1.62237275e+00 4.78339821e-01 1.04115188e+00 -1.05715096e+00 -3.23264450e-01 1.43028796e+00 3.45002979e-01 -1.02613293e-01 3.19130480e-01 -4.77264613e-01 -3.49300265e-01 -8.70528445e-02 4.92096066e-01 6.01457655e-01 -7.21245706e-02 -8.33424032e-02 -4.07271117e-01 5.56307495e-01 -4.70325768e-01 4.67100590e-01 9.33284700e-01 2.37350047e-01 -1.92826241e-01 5.28059542e-01 1.55005670e+00 4.48082909e-02 -4.03863519e-01 -9.23653468e-02 -9.29132253e-02 -1.93062752e-01 1.86768785e-01 -7.13730574e-01 -1.08100247e+00 1.01869738e+00 1.03679061e+00 1.59392908e-01 1.05567551e+00 1.70091614e-02 9.52965021e-01 5.32736182e-01 -2.09490389e-01 -6.86757684e-01 1.60676807e-01 6.41132221e-02 3.47004712e-01 -1.51750255e+00 2.22694561e-01 -1.77478313e-01 -3.96949291e-01 1.81030500e+00 6.97204471e-01 -1.16086185e-01 7.34753728e-01 5.07000625e-01 3.67594123e-01 -3.68432522e-01 -9.99766171e-01 -2.56009907e-01 2.23414034e-01 5.11694431e-01 9.90301430e-01 3.24891269e-01 -2.00754493e-01 5.55039108e-01 2.38932148e-01 4.55336511e-01 1.59550443e-01 6.09141469e-01 -1.04131067e+00 -6.12253010e-01 -4.28241402e-01 1.03264189e+00 -7.68880188e-01 1.71531051e-01 -8.26313138e-01 1.11242008e+00 2.94663734e-03 6.16931975e-01 4.93960157e-02 -7.43608177e-02 -7.08846152e-02 1.14037748e-03 -2.35859305e-02 -5.03728688e-01 -5.97664833e-01 2.00260997e-01 -1.15289278e-01 1.35687843e-01 -2.40567178e-01 -3.53440791e-01 -1.52905667e+00 -3.20378661e-01 -4.96622980e-01 2.21874684e-01 -2.36160844e-03 6.51964188e-01 -2.25511059e-01 6.91480398e-01 6.81910753e-01 -1.57668874e-01 -3.77748251e-01 -1.10488486e+00 -9.02071953e-01 1.44421663e-02 5.53627908e-01 6.05395958e-02 -6.57576799e-01 2.04672456e-01]
[15.337623596191406, -2.5019073486328125]
2bb8dd60-57c2-457c-bb46-20a1f4a4f04b
visual-commonsense-aware-representation
2211.09469
null
https://arxiv.org/abs/2211.09469v1
https://arxiv.org/pdf/2211.09469v1.pdf
Visual Commonsense-aware Representation Network for Video Captioning
Generating consecutive descriptions for videos, i.e., Video Captioning, requires taking full advantage of visual representation along with the generation process. Existing video captioning methods focus on making an exploration of spatial-temporal representations and their relationships to produce inferences. However, such methods only exploit the superficial association contained in the video itself without considering the intrinsic visual commonsense knowledge that existed in a video dataset, which may hinder their capabilities of knowledge cognitive to reason accurate descriptions. To address this problem, we propose a simple yet effective method, called Visual Commonsense-aware Representation Network (VCRN), for video captioning. Specifically, we construct a Video Dictionary, a plug-and-play component, obtained by clustering all video features from the total dataset into multiple clustered centers without additional annotation. Each center implicitly represents a visual commonsense concept in the video domain, which is utilized in our proposed Visual Concept Selection (VCS) to obtain a video-related concept feature. Next, a Conceptual Integration Generation (CIG) is proposed to enhance the caption generation. Extensive experiments on three publicly video captioning benchmarks: MSVD, MSR-VTT, and VATEX, demonstrate that our method reaches state-of-the-art performance, indicating the effectiveness of our method. In addition, our approach is integrated into the existing method of video question answering and improves this performance, further showing the generalization of our method. Source code has been released at https://github.com/zchoi/VCRN.
['Heng Tao Shen', 'Jin Qian', 'Xiangpeng Li', 'Lianli Gao', 'Haonan Zhang', 'Pengpeng Zeng']
2022-11-17
null
null
null
null
['video-question-answering']
['computer-vision']
[ 2.80860513e-01 -1.06063783e-01 -3.03148746e-01 -2.54307419e-01 -6.49140179e-01 -6.38304532e-01 6.53093636e-01 4.33141887e-02 -2.77973595e-03 6.41263962e-01 6.20042443e-01 -1.92484275e-01 -8.12599994e-03 -6.50957823e-01 -9.18866754e-01 -4.13555443e-01 2.89972693e-01 1.04589999e-01 4.67118397e-02 -1.64468601e-01 2.58573771e-01 -2.34743077e-02 -1.69717097e+00 6.12319052e-01 9.78440821e-01 1.08377755e+00 4.64544058e-01 4.01097804e-01 -4.80927199e-01 1.09451950e+00 -5.10799408e-01 -4.47853684e-01 -1.23584867e-01 -7.84621418e-01 -8.65300417e-01 3.64532024e-01 3.50241035e-01 -3.48167628e-01 -5.18322527e-01 1.13352108e+00 1.24764524e-01 4.97983873e-01 5.55707395e-01 -1.65551329e+00 -1.16984808e+00 7.82411754e-01 -4.17455435e-01 2.86863685e-01 7.57870972e-01 2.13595003e-01 1.08076727e+00 -9.72025871e-01 7.91612506e-01 1.16293800e+00 1.22286238e-01 7.19450951e-01 -9.39719975e-01 -8.03395510e-01 2.78115988e-01 6.51541173e-01 -1.47402680e+00 -3.62782568e-01 1.06634331e+00 -5.01547813e-01 5.94455302e-01 3.88255388e-01 8.63160551e-01 1.26558769e+00 -5.86498380e-01 1.04756308e+00 8.14791143e-01 -1.41185924e-01 3.13589394e-01 7.29958415e-02 -4.36235890e-02 4.66384590e-01 1.77798957e-01 -3.85953844e-01 -5.76722980e-01 1.47014239e-03 8.30024302e-01 1.25334695e-01 -6.69597745e-01 -3.74523401e-01 -1.40591621e+00 7.67229021e-01 5.08262753e-01 4.07013148e-01 -4.86782670e-01 2.85679400e-01 4.74820554e-01 -6.47396892e-02 2.59372503e-01 4.10392553e-01 5.50856963e-02 -3.14963847e-01 -7.94432640e-01 1.75843611e-01 4.51774985e-01 1.31814468e+00 7.32186198e-01 -1.49133518e-01 -6.56404495e-01 7.43798494e-01 2.19068035e-01 4.29831773e-01 5.58907330e-01 -1.11342287e+00 6.70742035e-01 7.64723957e-01 -2.59978957e-02 -1.21481204e+00 8.12701359e-02 -5.51150218e-02 -6.61689937e-01 -4.73790288e-01 -1.03947753e-03 1.30849987e-01 -8.39273214e-01 1.87028253e+00 1.81877851e-01 6.63475275e-01 2.85733730e-01 1.21957314e+00 1.18713295e+00 8.73955965e-01 2.89378911e-01 -2.71224260e-01 1.48262310e+00 -9.71420765e-01 -8.98160994e-01 -2.20749199e-01 3.65668595e-01 -4.43858624e-01 1.22784483e+00 4.94550765e-02 -8.84755433e-01 -4.65749502e-01 -9.15844381e-01 -1.31746799e-01 -4.03231680e-01 -4.53273542e-02 6.69275403e-01 1.49058416e-01 -8.41437817e-01 2.51814514e-01 -4.13324982e-01 -3.70673984e-01 6.38756931e-01 -1.13051705e-01 -4.83181745e-01 -3.91337186e-01 -1.42993641e+00 4.34644014e-01 7.11213648e-01 4.86659929e-02 -9.80054438e-01 -5.87854326e-01 -1.04373193e+00 7.12403730e-02 6.62822723e-01 -8.22605252e-01 1.04625440e+00 -1.25318670e+00 -9.30859506e-01 5.78940332e-01 -2.66663849e-01 -3.54647279e-01 1.72157735e-01 -1.42625362e-01 -4.54927295e-01 7.27664471e-01 1.90034494e-01 1.11991477e+00 7.95106828e-01 -1.61376703e+00 -4.16745901e-01 -1.88104212e-02 5.64835370e-01 4.08760160e-01 -5.62530994e-01 -8.70687217e-02 -9.35789227e-01 -8.00470412e-01 -8.16385448e-02 -7.69070923e-01 1.21900879e-01 4.79213186e-02 -4.37169433e-01 -1.64845690e-01 7.14043677e-01 -8.69543791e-01 1.46429992e+00 -2.22966385e+00 3.74299854e-01 7.11337253e-02 2.61538237e-01 1.68748453e-01 -2.60072827e-01 4.29592133e-01 -2.16059908e-01 3.15097809e-01 -4.25315857e-01 -2.23896265e-01 -8.86998177e-02 2.41383284e-01 -3.01688313e-01 2.24557836e-02 2.83657283e-01 1.06023073e+00 -1.36466253e+00 -8.61946285e-01 2.26550326e-01 5.51434815e-01 -5.63246906e-01 2.45585829e-01 -5.18816411e-01 3.13296080e-01 -6.94089830e-01 6.59194648e-01 5.88444173e-01 -4.95743960e-01 1.90998822e-01 -4.96687680e-01 1.96290836e-01 -5.94734289e-02 -8.83254409e-01 2.00707841e+00 -2.53629684e-01 7.86235988e-01 -4.86606389e-01 -8.78765762e-01 8.20954621e-01 3.90555948e-01 5.08401990e-01 -6.84889257e-01 1.07767448e-01 -5.11130542e-02 -4.23857868e-01 -9.27142620e-01 6.02156341e-01 -5.47026843e-02 4.40975763e-02 1.86199695e-01 5.55030890e-02 2.11432755e-01 3.59570563e-01 7.60801613e-01 8.69072139e-01 3.56429011e-01 2.71980226e-01 2.03526869e-01 5.37927508e-01 2.57181644e-01 3.65668625e-01 4.25983429e-01 -2.47549549e-01 7.54376352e-01 6.48058653e-01 2.77935853e-03 -8.88822973e-01 -9.00690734e-01 2.85298705e-01 7.71145284e-01 5.24593771e-01 -7.23699152e-01 -8.20047796e-01 -6.36204422e-01 -1.10313855e-01 9.50665772e-01 -8.10493052e-01 -3.03942919e-01 -4.96129990e-01 -2.97510296e-01 4.00714666e-01 6.79361939e-01 4.98233885e-01 -1.30739760e+00 -4.58240509e-01 5.82577195e-03 -8.50574195e-01 -1.41643369e+00 -5.67750394e-01 -5.23669600e-01 -6.08254850e-01 -1.09079707e+00 -8.07960808e-01 -7.65212059e-01 7.14308679e-01 7.00468719e-01 1.04736960e+00 2.00566784e-01 3.33595686e-02 6.83169901e-01 -8.97057772e-01 6.66844985e-03 -2.94453770e-01 -3.55055541e-01 -1.66393444e-01 2.06496552e-01 4.25551534e-01 -5.04821122e-01 -6.81048512e-01 9.00152251e-02 -1.17150998e+00 4.87870336e-01 5.63627303e-01 6.14913285e-01 6.69320703e-01 -2.01704517e-01 6.60649538e-01 -5.82814395e-01 5.45355380e-01 -9.94398236e-01 -4.75681089e-02 3.48677069e-01 -1.17906474e-01 -7.00140819e-02 5.30885398e-01 -5.70317209e-01 -9.97662544e-01 -4.13683467e-02 2.44355962e-01 -1.16392446e+00 -1.28384933e-01 6.81178272e-01 -3.19183677e-01 2.88403481e-01 3.21473867e-01 5.76304674e-01 3.48707102e-02 -2.85067379e-01 7.50345528e-01 5.90813100e-01 6.78392172e-01 -5.23819149e-01 8.09885502e-01 5.01586139e-01 -3.07126611e-01 -5.82851589e-01 -7.69951761e-01 -5.51909924e-01 -3.53943169e-01 -5.18120766e-01 1.14901578e+00 -1.02851844e+00 -6.12552524e-01 -7.45549947e-02 -1.28301692e+00 8.46604556e-02 -1.45695552e-01 3.84723574e-01 -6.44715309e-01 5.54817319e-01 -2.87191987e-01 -6.58377349e-01 -2.38184050e-01 -9.60550964e-01 9.34191883e-01 2.76423633e-01 -9.45712999e-02 -8.25407743e-01 -3.15469593e-01 6.82253003e-01 2.01210558e-01 4.31367129e-01 8.21269691e-01 -6.83931768e-01 -7.75166988e-01 1.40863046e-01 -5.73852241e-01 2.89280474e-01 1.27296373e-01 -1.12482436e-01 -7.97877371e-01 -1.15285076e-01 -3.34298491e-01 -3.86295527e-01 8.93845797e-01 1.25249624e-01 1.56076133e+00 -5.10020733e-01 -2.49288857e-01 3.87002736e-01 1.53096199e+00 2.58248150e-01 8.21074963e-01 3.55709940e-01 1.04427767e+00 5.37082732e-01 8.44782114e-01 5.80374300e-01 6.12974644e-01 6.91470325e-01 6.58458173e-01 1.03969209e-01 -3.42826545e-01 -5.88335454e-01 4.18906182e-01 7.86572576e-01 -1.80537939e-01 -4.31284338e-01 -7.89039731e-01 7.33285785e-01 -2.09008837e+00 -1.20303094e+00 -1.26578575e-02 1.81547999e+00 7.46890306e-01 -2.09364086e-01 -1.21793505e-02 1.49999140e-03 9.40613091e-01 1.66607410e-01 -5.06645143e-01 -2.98878569e-02 -5.22229858e-02 -2.69320667e-01 6.29289076e-02 5.00659347e-02 -9.45175767e-01 9.21096921e-01 4.95591640e+00 9.72541809e-01 -7.73422062e-01 1.18485302e-01 3.89958590e-01 -6.26624450e-02 -6.87067509e-01 -3.54919978e-03 -3.35368544e-01 6.32628620e-01 6.38076723e-01 -3.37679058e-01 5.54781914e-01 7.57185519e-01 2.37777948e-01 9.71711949e-02 -1.07941222e+00 1.24812126e+00 5.65935373e-01 -1.51857436e+00 7.15285778e-01 -2.43375540e-01 5.78754187e-01 -5.38781464e-01 -7.99121335e-02 3.37478548e-01 -2.27622151e-01 -7.74295628e-01 9.11644578e-01 5.07578194e-01 8.97664011e-01 -6.70824289e-01 6.23542070e-01 -5.19438535e-02 -1.39444363e+00 -5.44042960e-02 -2.48356029e-01 2.05508694e-01 3.27342004e-01 3.28383476e-01 -6.94404900e-01 7.42705226e-01 6.24925494e-01 1.05005312e+00 -7.20117331e-01 1.00926232e+00 -2.07858965e-01 5.47206044e-01 2.17838585e-01 -8.86893421e-02 3.27567756e-01 -7.92500749e-02 7.24296868e-01 1.21589684e+00 4.05648649e-01 3.94048572e-01 7.29060024e-02 1.13074648e+00 -3.04715395e-01 2.04648241e-01 -4.37575310e-01 -3.39490265e-01 7.33938158e-01 1.15767634e+00 -7.47169137e-01 -6.41162097e-01 -4.75312650e-01 1.02374828e+00 2.45662048e-01 4.59198326e-01 -1.32774186e+00 -3.21156442e-01 6.02408946e-01 -6.39770851e-02 4.60737228e-01 1.13501567e-02 1.26432091e-01 -1.41900492e+00 2.78531700e-01 -9.55519080e-01 5.19790828e-01 -1.33325827e+00 -1.19441259e+00 6.53627753e-01 3.96983176e-01 -1.60284090e+00 -1.67154178e-01 -3.27678472e-01 -3.86168122e-01 4.09702927e-01 -1.45504904e+00 -1.28152227e+00 -8.50066364e-01 8.03740263e-01 8.32183838e-01 6.07383549e-02 4.19961184e-01 4.25589681e-01 -5.11184037e-01 3.19068432e-01 -3.57386708e-01 1.68576524e-01 5.86640358e-01 -9.20582712e-01 8.87547061e-02 8.81619573e-01 1.52952269e-01 5.83442807e-01 7.00835586e-01 -7.40938723e-01 -1.44128096e+00 -1.29501665e+00 7.00098872e-01 -3.16452652e-01 7.76318789e-01 -2.23256066e-01 -1.10555029e+00 7.24005640e-01 2.57648140e-01 -1.16862640e-01 7.82671213e-01 -3.46749961e-01 -5.97478449e-01 1.44778356e-01 -9.05540466e-01 7.13301659e-01 1.39691567e+00 -6.36053503e-01 -8.63203108e-01 3.38548452e-01 1.23774636e+00 -2.38854438e-01 -6.90250099e-01 1.01468712e-01 4.13985908e-01 -6.17193103e-01 1.10328829e+00 -6.85428023e-01 9.60766792e-01 -5.59664130e-01 -3.71457994e-01 -1.11113524e+00 -2.46943310e-01 -3.93242329e-01 -3.99668753e-01 1.65175259e+00 3.28235477e-01 -1.39189810e-01 4.25676435e-01 5.70272326e-01 -3.39016140e-01 -5.89328110e-01 -6.12714887e-01 -7.64061391e-01 -3.77423525e-01 -5.58185220e-01 6.85135722e-01 1.27442467e+00 1.81316376e-01 2.35005081e-01 -4.80322421e-01 4.97049354e-02 5.01585782e-01 1.67379886e-01 4.80793208e-01 -8.20031166e-01 -2.47381359e-01 -3.97829980e-01 -5.30382156e-01 -1.02580166e+00 2.18103573e-01 -9.95347619e-01 3.48039493e-02 -1.82488430e+00 6.97565138e-01 -1.63447484e-01 -3.79107267e-01 6.13501787e-01 -3.98367316e-01 3.58905852e-01 6.03188694e-01 3.52302879e-01 -1.02915001e+00 6.18719518e-01 1.48661113e+00 -2.48176575e-01 -1.18105054e-01 -6.35146439e-01 -9.42775249e-01 4.73284751e-01 7.52652168e-01 -2.77578801e-01 -6.69913530e-01 -4.96164262e-01 1.72207057e-01 1.72178596e-01 6.70413375e-01 -9.50645864e-01 9.42476615e-02 -3.07696849e-01 2.74312407e-01 -4.64614183e-01 4.92418766e-01 -7.26573467e-01 2.53224701e-01 1.03086069e-01 -5.03693342e-01 5.31888120e-02 1.65328354e-01 7.04610109e-01 -5.33430099e-01 -2.01373711e-01 3.12550724e-01 -1.06607556e-01 -1.27077699e+00 3.43120188e-01 -1.16377272e-01 2.53832549e-01 1.13519800e+00 -2.93359220e-01 -5.64078867e-01 -6.13626242e-01 -5.89541733e-01 3.75390798e-01 4.96556997e-01 8.07337224e-01 1.03967965e+00 -1.69394720e+00 -6.68983340e-01 -2.65284032e-01 6.54172242e-01 -1.81001887e-01 5.65141797e-01 7.08353996e-01 -3.92592013e-01 3.05180103e-01 -2.97500819e-01 -4.29019839e-01 -1.12313104e+00 9.91311312e-01 -5.09187430e-02 2.59852350e-01 -7.00413167e-01 6.93412781e-01 4.08693582e-01 3.44017982e-01 1.83014311e-02 -3.75186533e-01 -6.83511317e-01 1.98325381e-01 6.77712739e-01 1.53400898e-01 -4.96937245e-01 -9.05030489e-01 -4.28585261e-01 4.86045927e-01 2.42008626e-01 3.48484218e-02 1.00402641e+00 -3.31449419e-01 -9.28753987e-02 3.57027739e-01 1.29851711e+00 -3.05380613e-01 -1.04926205e+00 -1.32950768e-01 -1.90139964e-01 -5.58275938e-01 -1.18437305e-01 -5.65078020e-01 -1.11631715e+00 7.55867422e-01 1.40477672e-01 2.24077757e-02 1.33202732e+00 3.29071581e-01 9.53457057e-01 2.29120985e-01 1.02823921e-01 -7.35196829e-01 4.50892508e-01 1.14643857e-01 1.17206621e+00 -1.29800963e+00 -8.58836696e-02 -6.49102211e-01 -1.10183322e+00 1.02153897e+00 8.03608656e-01 1.66223601e-01 2.26935953e-01 -4.03465748e-01 -2.77020305e-01 -1.96997270e-01 -7.80509531e-01 -3.83104235e-01 4.84079838e-01 5.84704280e-01 1.89956814e-01 6.39307648e-02 -2.72299707e-01 8.36511254e-01 1.10232659e-01 1.31171331e-01 6.10128641e-01 8.46557021e-01 -3.79016429e-01 -6.55522346e-01 -2.12825820e-01 2.24541828e-01 -9.24147964e-02 -2.55810887e-01 -3.80847603e-01 6.10819101e-01 1.94272742e-01 9.92084682e-01 2.35187635e-01 -5.69090664e-01 1.85391232e-01 -3.32341120e-02 2.00610772e-01 -4.83350664e-01 -1.71481237e-01 -1.76905468e-01 7.81832933e-02 -6.67812943e-01 -7.93139815e-01 -5.62305510e-01 -1.46160376e+00 -7.26691782e-02 -9.88456309e-02 2.82066941e-01 5.42024493e-01 8.21955144e-01 4.58868921e-01 5.91012836e-01 4.33533400e-01 -6.98053300e-01 1.01386355e-02 -7.80046225e-01 -2.40073442e-01 9.54218924e-01 1.97276309e-01 -8.88327777e-01 -4.74107891e-01 4.81264263e-01]
[10.523992538452148, 0.9040728211402893]
7161098c-98ac-4415-a522-1bbec4714843
analogical-inference-enhanced-knowledge-graph
2301.00982
null
https://arxiv.org/abs/2301.00982v2
https://arxiv.org/pdf/2301.00982v2.pdf
Analogical Inference Enhanced Knowledge Graph Embedding
Knowledge graph embedding (KGE), which maps entities and relations in a knowledge graph into continuous vector spaces, has achieved great success in predicting missing links in knowledge graphs. However, knowledge graphs often contain incomplete triples that are difficult to inductively infer by KGEs. To address this challenge, we resort to analogical inference and propose a novel and general self-supervised framework AnKGE to enhance KGE models with analogical inference capability. We propose an analogical object retriever that retrieves appropriate analogical objects from entity-level, relation-level, and triple-level. And in AnKGE, we train an analogy function for each level of analogical inference with the original element embedding from a well-trained KGE model as input, which outputs the analogical object embedding. In order to combine inductive inference capability from the original KGE model and analogical inference capability enhanced by AnKGE, we interpolate the analogy score with the base model score and introduce the adaptive weights in the score function for prediction. Through extensive experiments on FB15k-237 and WN18RR datasets, we show that AnKGE achieves competitive results on link prediction task and well performs analogical inference.
['Huajun Chen', 'Yi Yang', 'Yufeng Huang', 'Mingyang Chen', 'Wen Zhang', 'Zhen Yao']
2023-01-03
null
null
null
null
['knowledge-graph-embedding']
['graphs']
[-1.53769672e-01 5.98119378e-01 -6.30066693e-01 -1.89158216e-01 -2.41393875e-02 -4.05298620e-01 4.83484983e-01 4.54886883e-01 -1.62317678e-01 9.20675159e-01 1.37745544e-01 -4.29441631e-01 -6.71687841e-01 -1.51976800e+00 -1.19678009e+00 -5.03246449e-02 -5.54190576e-02 8.34862888e-01 2.38630280e-01 -4.74057049e-01 -2.51063377e-01 2.34236106e-01 -1.14519417e+00 3.09495628e-01 1.29685187e+00 1.04811442e+00 -8.73065591e-02 2.31889337e-01 -4.74073023e-01 1.18119240e+00 -1.65483147e-01 -9.98108268e-01 -7.62087405e-02 -2.91384518e-01 -9.65345025e-01 -6.86934471e-01 4.04471427e-01 -1.04322232e-01 -9.25746441e-01 9.40617263e-01 1.71738118e-01 7.58433640e-02 9.11081851e-01 -1.60166657e+00 -1.56696129e+00 1.04265940e+00 -6.79211989e-02 6.84862882e-02 5.64767122e-01 -3.15576822e-01 1.49834144e+00 -1.19012618e+00 7.66149700e-01 1.30406713e+00 8.27435136e-01 1.97910935e-01 -1.32998478e+00 -7.55009353e-01 6.39972910e-02 9.79898393e-01 -1.58164108e+00 1.42373219e-01 9.59009111e-01 -2.64556438e-01 1.09490323e+00 -7.13758245e-02 8.48620653e-01 8.85614276e-01 -1.14354752e-01 8.62503171e-01 7.28524804e-01 -4.00750935e-01 -6.96280524e-02 3.44204605e-01 3.46462995e-01 9.98498499e-01 4.79695171e-01 1.24497535e-02 -5.91400504e-01 -1.25454679e-01 6.68625474e-01 -2.27165744e-02 -4.15015280e-01 -3.50952625e-01 -1.17819083e+00 8.43198895e-01 1.42717147e+00 2.85473634e-02 -3.86976361e-01 1.38653636e-01 1.64300069e-01 4.39165711e-01 5.07335514e-02 7.63238549e-01 -5.51572502e-01 7.00491846e-01 -2.95158714e-01 1.58344328e-01 9.43390369e-01 1.27975047e+00 9.60200369e-01 -3.22617620e-01 -3.83534700e-01 8.32888126e-01 3.54287654e-01 4.49525386e-01 3.34452897e-01 -6.35483265e-01 4.47198302e-01 1.12083495e+00 -2.52389759e-01 -1.27471817e+00 -2.88694590e-01 -6.62386596e-01 -7.81608284e-01 -2.55454421e-01 -7.15222256e-03 2.04268709e-01 -6.17206275e-01 1.74031413e+00 3.74630064e-01 6.08777344e-01 3.20450306e-01 5.90088129e-01 1.28778052e+00 6.77630007e-01 2.06489593e-01 2.30841897e-03 1.28602958e+00 -8.68003070e-01 -7.68678069e-01 -8.76616463e-02 7.57143795e-01 2.73995157e-02 1.00408947e+00 7.82427266e-02 -7.82222807e-01 -4.76003170e-01 -1.30516529e+00 -3.89787704e-01 -9.92704809e-01 6.34707510e-02 9.13656414e-01 -4.75768335e-02 -7.59050131e-01 6.99420214e-01 -3.38758588e-01 -1.37033939e-01 3.69707257e-01 3.26231986e-01 -5.76568961e-01 -2.41135433e-01 -2.12222290e+00 1.32238579e+00 1.24310815e+00 1.28245488e-01 -2.86618143e-01 -9.53106284e-01 -1.12972140e+00 3.92201453e-01 6.74930990e-01 -1.11848760e+00 3.34220111e-01 -1.79525927e-01 -1.16243529e+00 4.22005504e-01 2.14776501e-01 -5.15203774e-01 -1.15964361e-01 2.77368668e-02 -7.91996121e-01 -2.18837690e-02 6.67183474e-02 6.92993641e-01 3.46100032e-01 -1.27880883e+00 -3.86215746e-01 -4.10939515e-01 4.58750069e-01 2.40287587e-01 -6.83696806e-01 -7.82242715e-01 -4.26245481e-01 -4.84306216e-01 1.63569137e-01 -6.87798917e-01 3.30822915e-01 -2.09204927e-02 -5.20563543e-01 -7.75603414e-01 6.03903115e-01 -7.24361002e-01 1.37098038e+00 -1.87409866e+00 4.84070987e-01 5.15292525e-01 5.28378725e-01 2.92488515e-01 -1.44659817e-01 2.61106968e-01 -1.81415692e-01 -4.09603342e-02 3.99461528e-03 2.97856420e-01 2.11213410e-01 6.59730911e-01 -3.73472184e-01 -3.17899495e-01 2.90880591e-01 1.54183018e+00 -1.42949641e+00 -6.64059162e-01 2.13490035e-02 2.83781439e-01 -6.80129290e-01 4.21313122e-02 -3.97970259e-01 -3.53692412e-01 -4.21279579e-01 7.52910435e-01 4.29474324e-01 -6.23847783e-01 4.48750466e-01 -8.73922229e-01 6.60327613e-01 2.66622514e-01 -9.52459395e-01 1.53758931e+00 -5.90758622e-01 3.18945974e-01 -7.34507203e-01 -1.00485933e+00 9.28772449e-01 3.23504210e-02 3.21694873e-02 -6.44579828e-01 -2.58828551e-01 2.63451993e-01 -4.74719070e-02 -3.45977515e-01 4.31270570e-01 -2.14030668e-01 2.97044944e-02 -1.18957013e-01 6.97427273e-01 2.12059058e-02 6.34156615e-02 8.60076666e-01 1.10006189e+00 2.94997394e-01 3.76732051e-01 5.42245470e-02 6.51873291e-01 -4.81211953e-02 5.22812009e-01 4.89229977e-01 3.68427366e-01 -2.22677007e-01 6.63640380e-01 -2.09504962e-01 -5.73640704e-01 -1.58792782e+00 -1.12901755e-01 8.36317241e-01 2.89646238e-01 -6.74172819e-01 -1.54488280e-01 -1.04043460e+00 6.66995883e-01 9.54579353e-01 -7.89990604e-01 -7.79951096e-01 -3.01284730e-01 -4.01637703e-01 6.45020843e-01 8.76131237e-01 5.45320630e-01 -1.14069140e+00 6.24948442e-01 2.91368216e-01 -3.05564433e-01 -1.28242159e+00 -1.36020169e-01 1.81065723e-01 -6.45051420e-01 -1.08822024e+00 -1.03955790e-01 -9.02434707e-01 7.13352978e-01 -2.80861557e-01 1.28588665e+00 1.26223326e-01 -7.15008378e-02 1.94773078e-01 -3.23523581e-01 -1.74674898e-01 -1.97437942e-01 8.49072412e-02 1.96245104e-01 -1.62410945e-01 5.55339158e-01 -7.82464206e-01 -3.16435367e-01 1.05264872e-01 -7.40798652e-01 -1.84399895e-02 7.40800261e-01 1.09733593e+00 6.62082791e-01 1.13747463e-01 1.12463725e+00 -8.74283195e-01 5.96994340e-01 -7.13290453e-01 -4.88349855e-01 7.34366655e-01 -1.03762364e+00 3.67198199e-01 7.03253627e-01 -2.60116667e-01 -9.01272714e-01 -3.04471821e-01 2.87576690e-02 -6.41251624e-01 6.02334261e-01 1.20389307e+00 -3.43392879e-01 -1.22293398e-01 7.27157831e-01 9.80658978e-02 -1.86401665e-01 -2.42956936e-01 1.01104581e+00 4.08880740e-01 9.71897781e-01 -8.30642819e-01 9.31154072e-01 -2.85603367e-02 2.12248072e-01 -2.25851327e-01 -1.31090534e+00 -7.33803213e-02 -4.62702364e-01 1.32351592e-01 6.17169678e-01 -9.21845853e-01 -9.92455959e-01 -2.19303742e-01 -1.13818896e+00 -2.82938443e-02 -3.96895260e-01 5.26104748e-01 -2.32185721e-01 2.33559877e-01 -5.59778512e-01 -2.85543591e-01 -3.42812747e-01 -4.77156818e-01 5.28788328e-01 1.48699373e-01 -1.59938401e-03 -1.21403086e+00 -3.75864133e-02 2.54053086e-01 3.45576517e-02 1.17087074e-01 1.52687991e+00 -7.18161762e-01 -7.04199612e-01 -1.94259390e-01 -5.13916910e-01 3.16554338e-01 1.66984692e-01 -3.57796431e-01 -5.20293295e-01 1.44536197e-01 -8.20346713e-01 -4.85030562e-01 1.06176805e+00 -1.76984996e-01 1.20273805e+00 -4.33285058e-01 -6.77956998e-01 5.14294863e-01 1.58698297e+00 -1.89200327e-01 6.43505037e-01 2.74241537e-01 1.05797434e+00 1.50391936e-01 3.90651017e-01 -1.01129770e-01 9.13299918e-01 6.20487809e-01 2.39207104e-01 2.96025306e-01 -2.54935324e-01 -9.97675657e-01 9.08210501e-02 9.08431053e-01 -1.83470860e-01 -1.00092098e-01 -6.71414137e-01 6.94008708e-01 -1.99532759e+00 -9.22137856e-01 -8.05497915e-02 1.90840030e+00 1.43800676e+00 2.81986088e-01 -2.92840511e-01 8.13460425e-02 4.18683320e-01 -4.14631188e-01 -5.21613002e-01 -1.87929945e-05 -3.07923526e-01 2.15335876e-01 3.35068226e-01 7.28290439e-01 -6.48433566e-01 1.15413773e+00 5.16849232e+00 9.91470456e-01 -4.64056939e-01 1.09518945e-01 -1.95940688e-01 3.43994498e-01 -7.05779016e-01 2.63020426e-01 -8.54554892e-01 3.82734269e-01 6.38296902e-01 -4.55522090e-01 6.61242783e-01 6.48052335e-01 -7.63657928e-01 4.85604107e-01 -1.62706685e+00 8.73964071e-01 1.11013036e-02 -1.61709034e+00 6.63961649e-01 -1.12882406e-01 6.61968768e-01 -2.89502829e-01 -2.08815053e-01 1.06376076e+00 6.21036708e-01 -9.96153235e-01 5.56716584e-02 8.69010746e-01 5.96492290e-01 -5.82882404e-01 8.25611770e-01 9.52753723e-02 -1.28880847e+00 -1.72785923e-01 -5.09929597e-01 1.02649830e-01 1.23084977e-01 6.14169955e-01 -9.68264759e-01 1.23780406e+00 4.64945227e-01 8.92463386e-01 -6.82456970e-01 6.11769259e-01 -8.86118412e-01 3.17009777e-01 -2.31896728e-01 1.55721419e-02 -4.25869320e-03 -2.31091887e-01 2.17606053e-01 9.04716313e-01 9.89620313e-02 3.95745873e-01 1.83861237e-02 1.25923097e+00 -6.81868851e-01 -6.41210005e-02 -7.17559695e-01 -1.52106270e-01 8.31943095e-01 1.22553837e+00 -7.08311349e-02 -6.15278244e-01 -4.78552550e-01 8.88025761e-01 9.58417773e-01 5.02759576e-01 -8.43709767e-01 -6.86536908e-01 1.80657834e-01 -7.67291114e-02 2.38878712e-01 2.17471093e-01 -2.14201771e-02 -1.16440511e+00 1.09909840e-01 -3.96402866e-01 8.82607877e-01 -1.19064271e+00 -1.71134639e+00 2.55201578e-01 8.63355398e-02 -7.82306492e-01 -9.82456207e-02 -7.61504114e-01 -3.79608154e-01 7.42757082e-01 -1.62625480e+00 -1.24111533e+00 -3.68717879e-01 7.54282176e-01 -2.88565427e-01 -1.54551357e-01 8.95114541e-01 4.28267151e-01 -3.61650676e-01 9.49795723e-01 -2.06829190e-01 4.47619408e-01 5.12315333e-01 -1.50553894e+00 1.32750437e-01 9.77615640e-02 2.91058451e-01 8.61373246e-01 3.47822160e-01 -8.84234130e-01 -1.57088685e+00 -1.29249716e+00 9.90420103e-01 -7.29278266e-01 1.09404981e+00 -7.83517808e-02 -1.25829768e+00 1.23824608e+00 -9.83822122e-02 4.54559803e-01 6.77670002e-01 5.87669849e-01 -7.89993584e-01 -3.56433094e-01 -8.67866695e-01 5.53751051e-01 1.44586527e+00 -7.68102288e-01 -1.27790523e+00 4.61763859e-01 1.10678422e+00 -2.56057471e-01 -1.56969082e+00 7.39808500e-01 5.45446694e-01 -1.66128382e-01 1.35114515e+00 -1.02521503e+00 6.01978958e-01 -4.92352396e-01 -1.58539638e-01 -1.63466430e+00 -7.01847494e-01 1.14929438e-01 -1.00975704e+00 1.14101839e+00 8.29561949e-01 -7.49648690e-01 6.27222836e-01 2.59709924e-01 -4.02652100e-02 -1.04901409e+00 -6.14642620e-01 -1.13398468e+00 -8.59669223e-02 1.32421464e-01 6.59096718e-01 1.36042488e+00 5.37155569e-01 9.71497774e-01 -1.73590541e-01 3.67537290e-01 7.61008739e-01 3.05230647e-01 4.91832137e-01 -1.58954251e+00 -4.02954370e-01 -2.53262043e-01 -7.50053227e-01 -8.74164641e-01 6.41257405e-01 -1.76691639e+00 -4.56439972e-01 -2.04472518e+00 2.47732192e-01 -4.24897403e-01 -7.23540366e-01 8.31857383e-01 -4.46292698e-01 3.59628983e-02 -4.59591486e-02 -1.62689462e-01 -6.34249628e-01 9.89054263e-01 1.26749277e+00 -4.97359574e-01 -2.03280952e-02 -5.86952507e-01 -8.13451707e-01 5.31285584e-01 2.69538909e-01 -2.58664191e-01 -7.72336423e-01 -2.83646733e-01 8.56363654e-01 -8.69028922e-03 6.85359359e-01 -6.37687743e-01 6.49694860e-01 2.19721109e-01 5.81966877e-01 -5.12005925e-01 3.11404765e-01 -8.72275829e-01 4.69095372e-02 1.93697482e-01 -2.70513505e-01 -4.55908298e-01 1.37330294e-01 9.64690685e-01 -2.95430154e-01 -5.76627553e-02 1.53405800e-01 2.12781012e-01 -9.51022983e-01 4.06119645e-01 6.30615294e-01 7.59537965e-02 6.97814167e-01 -2.81024054e-02 -5.14891148e-01 -1.55960675e-02 -1.20749068e+00 7.58711994e-01 -1.31488293e-01 4.57451969e-01 8.96663189e-01 -2.00514603e+00 -6.57928705e-01 -6.44366965e-02 5.90238929e-01 -5.15213534e-02 9.10111293e-02 7.46387899e-01 -1.68336555e-02 2.71500349e-01 -1.50343776e-01 -2.20656037e-01 -5.86134970e-01 1.00133920e+00 2.51749903e-01 -3.98311108e-01 -6.07824445e-01 7.77166963e-01 -1.04510911e-01 -8.26643169e-01 2.73071113e-03 -2.53590196e-01 -3.60544801e-01 1.92814786e-02 9.34387669e-02 4.37711000e-01 5.73932007e-02 -1.01068147e-01 -4.67162043e-01 3.90403450e-01 -1.80999681e-01 2.61964411e-01 1.04946899e+00 1.60784200e-01 -3.63179058e-01 3.11030984e-01 1.12226117e+00 -2.38186121e-01 -5.64330697e-01 -7.74167061e-01 -7.58388545e-03 -1.42898589e-01 8.37781802e-02 -1.01018262e+00 -8.09748709e-01 4.26311344e-01 -7.77442055e-03 7.44918808e-02 8.44897509e-01 3.92772764e-01 7.13851929e-01 1.01228654e+00 3.98905158e-01 -8.30049157e-01 2.04051770e-02 5.13890684e-01 1.08855820e+00 -1.11699820e+00 2.19316185e-01 -8.19535911e-01 -2.46118024e-01 8.76763523e-01 9.10565794e-01 -4.24441062e-02 7.29873598e-01 -2.54267365e-01 -6.25458300e-01 -5.60196459e-01 -6.95621550e-01 -2.29759887e-01 7.03691483e-01 5.13301671e-01 1.29755542e-01 1.75824746e-01 -3.39470863e-01 9.85662401e-01 -2.51663983e-01 3.03891689e-01 4.22178172e-02 4.65077490e-01 -1.71529621e-01 -9.13903892e-01 1.11696713e-01 7.10578978e-01 2.61626095e-01 -4.91103113e-01 -4.76160407e-01 7.88902879e-01 9.46941078e-02 4.78113323e-01 -2.20530733e-01 -6.79643035e-01 5.33185244e-01 3.03593427e-01 7.40693808e-01 -4.60344851e-01 -4.76364978e-02 -8.80547404e-01 2.38041744e-01 -2.51731128e-01 -3.54401171e-02 6.37035072e-02 -1.65308440e+00 -2.30070159e-01 -5.14503837e-01 4.68309790e-01 1.38050243e-01 1.09332407e+00 4.87990350e-01 7.63479173e-01 2.88276851e-01 2.50737071e-02 -5.25278986e-01 -9.52762902e-01 -6.93707466e-01 5.16390741e-01 -8.35821033e-02 -1.09561896e+00 -2.09025279e-01 -2.00493947e-01]
[8.798065185546875, 7.869781970977783]
c96f82c7-7dff-4ae5-a3ab-0cddd938af54
bootstrapping-a-user-centered-task-oriented
2207.05223
null
https://arxiv.org/abs/2207.05223v2
https://arxiv.org/pdf/2207.05223v2.pdf
Bootstrapping a User-Centered Task-Oriented Dialogue System
We present TacoBot, a task-oriented dialogue system built for the inaugural Alexa Prize TaskBot Challenge, which assists users in completing multi-step cooking and home improvement tasks. TacoBot is designed with a user-centered principle and aspires to deliver a collaborative and accessible dialogue experience. Towards that end, it is equipped with accurate language understanding, flexible dialogue management, and engaging response generation. Furthermore, TacoBot is backed by a strong search engine and an automated end-to-end test suite. In bootstrapping the development of TacoBot, we explore a series of data augmentation strategies to train advanced neural language processing models and continuously improve the dialogue experience with collected real conversations. At the end of the semifinals, TacoBot achieved an average rating of 3.55/5.0.
['Huan Sun', 'Yu Su', 'Tianshu Zhang', 'Xiang Yue', 'Zhen Wang', 'Samuel Stevens', 'Lingbo Mo', 'Ashley Lewis', 'Xiang Deng', 'Ziru Chen', 'Shijie Chen']
2022-07-11
null
null
null
null
['dialogue-management']
['natural-language-processing']
[-3.37030947e-01 4.41020459e-01 2.38471419e-01 -5.08253098e-01 -7.34980702e-01 -7.43470609e-01 7.72864997e-01 3.98657061e-02 -4.55545008e-01 9.53439534e-01 5.52370071e-01 -4.08103853e-01 -1.16355876e-02 -4.51087087e-01 1.08154334e-01 6.69185519e-02 8.91732201e-02 9.45422888e-01 -2.19511464e-01 -1.09814465e+00 1.15898244e-01 -2.77359158e-01 -1.02600920e+00 5.58789253e-01 9.89554703e-01 9.14228261e-01 3.18224221e-01 1.31711173e+00 -2.34520495e-01 1.41536653e+00 -5.38310707e-01 -6.30759299e-01 -1.57871544e-01 -3.86306524e-01 -1.69036591e+00 -2.17357323e-01 -1.39295412e-02 -5.66250145e-01 -6.46348819e-02 2.51801819e-01 8.02682698e-01 4.01030332e-01 9.27458480e-02 -1.23162079e+00 -4.89836603e-01 8.57414722e-01 5.46772301e-01 -1.21456742e-01 8.64661515e-01 6.19469941e-01 1.28139758e+00 -9.29073215e-01 5.80838442e-01 1.03902316e+00 8.05610538e-01 1.01698196e+00 -1.08751976e+00 -1.34462282e-01 -3.27890903e-01 2.37534698e-02 -4.59512860e-01 -8.09836507e-01 6.80114090e-01 -3.22737724e-01 1.05552208e+00 5.55352449e-01 7.18030334e-01 1.47057950e+00 -5.64046681e-01 1.10293090e+00 9.62546468e-01 -4.67767924e-01 8.89747143e-02 4.49525446e-01 3.19149464e-01 7.32702434e-01 -8.12588930e-01 -2.76745945e-01 -8.57756495e-01 -2.02671453e-01 1.46520525e-01 -5.17375767e-01 -2.46866450e-01 -6.88819662e-02 -1.18268931e+00 1.10067844e+00 1.55781403e-01 1.94507197e-01 -4.92844522e-01 -2.27085501e-01 8.87770891e-01 5.96365452e-01 4.26401287e-01 1.08107162e+00 -8.55341196e-01 -9.47765291e-01 -2.54221618e-01 3.93628478e-01 1.47842336e+00 6.47018790e-01 1.41044572e-01 -3.59026343e-01 -6.73017979e-01 1.37581086e+00 2.46642455e-01 1.87516689e-01 3.93633574e-01 -1.24816751e+00 5.85134625e-01 6.26463473e-01 5.47938466e-01 -3.40317011e-01 -7.27256238e-01 2.34384462e-01 -6.36459231e-01 -5.45142293e-02 8.78843606e-01 -7.21524000e-01 -8.50878060e-02 1.65781069e+00 2.55247712e-01 -5.51849008e-01 2.19681308e-01 8.03479552e-01 1.15287852e+00 5.27563751e-01 3.43516618e-01 5.82948588e-02 1.40090561e+00 -1.39463103e+00 -7.90072918e-01 -1.67040378e-01 7.80110538e-01 -7.92169750e-01 1.69433308e+00 5.14842510e-01 -1.23209834e+00 -5.31697094e-01 -5.79187572e-01 -2.05486476e-01 -2.68281966e-01 -3.36511247e-02 7.96961248e-01 5.79246163e-01 -1.06219280e+00 4.09820467e-01 -3.86061817e-01 -3.97271901e-01 -1.37428626e-01 -2.24315934e-02 -3.61537337e-01 1.86919272e-01 -1.51996374e+00 1.15420699e+00 3.03939044e-01 3.38679329e-02 -5.56388378e-01 -6.32489204e-01 -7.03457713e-01 -8.47405791e-02 4.61079180e-01 -7.69426763e-01 2.05490375e+00 -4.64116633e-01 -2.39459538e+00 8.80686879e-01 4.04293239e-01 -4.94349450e-01 6.64528489e-01 -6.28810525e-01 -8.22083727e-02 -2.20590346e-02 -3.63412499e-02 5.32561183e-01 1.30368754e-01 -5.27332246e-01 -6.67710781e-01 4.36618812e-02 3.15176636e-01 5.69982946e-01 -3.68934751e-01 1.86975047e-01 -5.68728149e-02 -2.34523416e-02 -6.60008311e-01 -7.62620032e-01 -2.92898566e-01 -4.19133484e-01 -3.78519177e-01 -6.69188201e-01 5.53663433e-01 -9.36707437e-01 9.69962060e-01 -1.81873226e+00 1.10408187e-01 -3.45165402e-01 3.66961002e-01 5.50039649e-01 -1.60181358e-01 9.22204971e-01 5.04534006e-01 4.92127463e-02 2.80479282e-01 -6.00241065e-01 4.27608162e-01 -1.34793341e-01 -1.12282336e-01 -3.15265507e-01 1.21467970e-01 1.05315638e+00 -1.27190995e+00 -2.23505333e-01 3.99766713e-01 2.98649557e-02 -7.18322277e-01 9.67155337e-01 -5.78706384e-01 6.08081222e-01 -2.54504174e-01 1.90027937e-01 4.78293933e-02 -1.37040868e-01 1.12809509e-01 3.55962873e-01 -2.36341685e-01 9.11483645e-01 -5.10126412e-01 1.98013985e+00 -9.26854730e-01 6.06834292e-01 5.63437760e-01 -5.23953617e-01 1.02672517e+00 6.47839963e-01 3.51850927e-01 -7.24246860e-01 1.15439452e-01 1.43188685e-01 -1.80125833e-01 -8.21466327e-01 9.92413402e-01 8.38393494e-02 -3.48227203e-01 7.70531178e-01 3.34685683e-01 -6.01492465e-01 8.86510983e-02 4.50985193e-01 1.11118114e+00 9.88797992e-02 2.11533442e-01 -2.64865816e-01 5.23747921e-01 2.88395546e-02 -4.81289290e-02 8.21100771e-01 -4.75289524e-01 1.48699164e-01 2.59503126e-01 -5.60485363e-01 -1.05690086e+00 -7.99390078e-01 4.24521178e-01 1.79549003e+00 -4.32830095e-01 -7.07831204e-01 -8.18580806e-01 -7.43109047e-01 -4.36550379e-01 9.29831088e-01 -2.66810954e-01 -4.40225378e-03 -2.02621803e-01 -2.71754086e-01 8.98872077e-01 4.86625209e-02 9.19423103e-01 -1.52347910e+00 -6.23292923e-01 5.73083699e-01 -1.00494444e+00 -1.11337292e+00 -4.76738930e-01 4.05550301e-02 -1.40096039e-01 -1.05262542e+00 -3.62261683e-01 -7.35433102e-01 -2.29513466e-01 -4.74663563e-02 1.56637263e+00 1.96357027e-01 -7.51182139e-02 4.65205014e-01 -7.42996335e-01 -2.74913728e-01 -8.70727897e-01 5.20691454e-01 -8.64606500e-02 -4.27271307e-01 2.34812543e-01 -3.75013113e-01 -4.65649426e-01 2.88044453e-01 -1.34125188e-01 6.36382878e-01 4.79573198e-02 1.33661067e+00 -7.27291346e-01 -6.39497936e-01 8.66307855e-01 -6.61033034e-01 1.56763232e+00 -3.93642098e-01 -1.02144554e-01 2.40651369e-01 -5.03210008e-01 -1.63707480e-01 6.44524038e-01 -3.13838333e-01 -1.16668570e+00 -9.61611941e-02 -5.73331416e-01 5.82563400e-01 -1.23968013e-01 4.94639486e-01 1.95211172e-02 4.41043615e-01 9.30466592e-01 2.56798640e-02 3.19095343e-01 -3.68299603e-01 7.40329146e-01 1.23284888e+00 6.12568140e-01 -8.93313229e-01 2.16311038e-01 -4.39461112e-01 -8.91438663e-01 -9.50644076e-01 -8.68014276e-01 -3.40122819e-01 -4.63648796e-01 -5.13733268e-01 7.06582487e-01 -6.67751789e-01 -1.49097502e+00 7.48164535e-01 -1.27543676e+00 -1.26703179e+00 -2.25076243e-01 2.12441146e-01 -7.39766717e-01 5.28625101e-02 -9.38839376e-01 -1.16395438e+00 -9.46643174e-01 -7.16303289e-01 6.85002387e-01 2.89878696e-01 -9.91435885e-01 -1.00632250e+00 1.95092186e-01 1.09508753e+00 7.16107786e-01 -2.07974643e-01 4.93752599e-01 -1.14372969e+00 9.64864939e-02 -8.03661197e-02 -8.00673515e-02 3.57467830e-01 -8.90320465e-02 -1.73281595e-01 -9.70908463e-01 -7.32425451e-02 -1.18925624e-01 -1.24066186e+00 1.01927936e-01 -2.69843012e-01 4.21847761e-01 -6.76888525e-01 3.10007393e-01 1.12772582e-03 3.50704044e-01 6.14323886e-03 3.03787917e-01 2.59615123e-01 2.12256849e-01 8.74770761e-01 6.95147097e-01 5.24498165e-01 8.33116591e-01 9.13544059e-01 1.54726163e-01 2.78042704e-02 2.33970106e-01 -4.53790456e-01 3.93682182e-01 7.17393935e-01 2.18443885e-01 -1.84121817e-01 -9.42091465e-01 5.99160254e-01 -1.94737482e+00 -1.02190483e+00 -1.61053091e-01 1.80106735e+00 1.45768762e+00 1.21889196e-01 6.02567732e-01 -1.76294684e-01 3.41779858e-01 7.73595944e-02 -1.61865592e-01 -8.98937821e-01 3.30626190e-01 1.17740691e-01 -3.63494456e-01 8.76567185e-01 -9.17236686e-01 1.22266769e+00 6.32382679e+00 4.65879381e-01 -7.70513296e-01 7.77722970e-02 6.05408013e-01 9.45371389e-02 1.59015611e-01 -3.20034742e-01 -3.93219799e-01 3.24713290e-01 1.11925542e+00 -1.29402950e-01 8.53578746e-01 8.80535722e-01 3.02875251e-01 -1.63521282e-02 -9.32469070e-01 4.90486920e-01 -1.80015907e-01 -1.25887859e+00 -6.16803706e-01 -3.30273718e-01 1.76498711e-01 1.76796898e-01 -2.02015936e-01 9.83976543e-01 8.95270944e-01 -1.24160838e+00 3.94331992e-01 3.43123674e-01 5.71873963e-01 -5.72211564e-01 8.06739450e-01 9.36755180e-01 -5.59312522e-01 -1.84362069e-01 2.29570776e-01 -7.19649017e-01 3.52816820e-01 5.10065854e-02 -1.61112773e+00 2.21061751e-01 4.72128123e-01 -5.68594085e-03 -2.56348133e-01 5.14426351e-01 -3.39413226e-01 5.98874331e-01 -2.31436715e-01 -7.24149168e-01 2.07466125e-01 -3.94317172e-02 4.04997796e-01 1.25844085e+00 -2.76455015e-01 2.72058070e-01 5.29695213e-01 6.20498359e-01 -1.87028006e-01 2.79344469e-01 -5.34061491e-01 -9.82448608e-02 7.39014745e-01 1.75380576e+00 -6.07936420e-02 -1.60174981e-01 -2.23892163e-02 1.02183425e+00 6.69543028e-01 -8.29385519e-02 -5.32917142e-01 -4.29126203e-01 4.30792987e-01 -1.80210605e-01 -2.67795116e-01 -2.31475443e-01 -2.90310681e-01 -8.96527290e-01 -3.29177171e-01 -1.27317679e+00 1.52305111e-01 -8.65057170e-01 -1.38068724e+00 8.45890343e-01 -4.58330572e-01 -4.25770164e-01 -7.47628987e-01 -4.87487853e-01 -7.44024634e-01 8.82249177e-01 -7.69994378e-01 -1.37445438e+00 -3.38398993e-01 4.23850477e-01 7.75164366e-01 -2.90855169e-01 1.46231318e+00 1.92585826e-01 -3.13787043e-01 5.20904064e-01 -3.71542305e-01 3.22183549e-01 7.02079713e-01 -1.46439052e+00 6.14441216e-01 1.41532183e-01 -6.09527715e-02 6.15954459e-01 9.44697559e-01 -2.50396639e-01 -1.09607244e+00 -6.17217183e-01 1.18059087e+00 -7.88824081e-01 8.27036619e-01 -5.80049157e-01 -4.99604970e-01 4.66786116e-01 6.17134213e-01 -6.62381887e-01 7.82314181e-01 7.47424066e-01 -8.65215287e-02 2.73934066e-01 -1.17735040e+00 8.08208168e-01 8.25623453e-01 -8.02040994e-01 -8.39409769e-01 4.16763216e-01 7.88863659e-01 -6.67666912e-01 -8.56982768e-01 1.99688271e-01 6.29446328e-01 -9.76070642e-01 8.35340142e-01 -8.90720308e-01 5.61357141e-01 4.99025941e-01 1.46990491e-03 -1.65704143e+00 1.04451932e-01 -1.52019632e+00 -4.48215008e-03 1.46706116e+00 4.62695301e-01 -1.96690664e-01 7.10937977e-01 1.21691644e+00 -1.88359469e-01 -6.98006094e-01 -6.86716080e-01 -8.76561925e-02 -1.48799494e-01 -4.93108630e-01 3.75353366e-01 9.19389784e-01 8.65938067e-01 1.26484966e+00 -7.44412839e-01 -6.44036889e-01 3.43099609e-02 -3.32215667e-01 1.32601786e+00 -1.22266662e+00 -2.32458726e-01 -4.79821235e-01 4.90827978e-01 -1.28727484e+00 4.42216918e-02 -5.86686969e-01 3.21885884e-01 -1.42979527e+00 -2.35488772e-01 -3.88787925e-01 2.96521187e-01 6.22694731e-01 -2.94624686e-01 1.56619504e-01 2.74046719e-01 -3.44978422e-01 -9.29482520e-01 6.76979065e-01 1.05255306e+00 7.32770190e-02 -6.37130618e-01 3.40828985e-01 -9.40422416e-01 6.64047360e-01 9.86073017e-01 3.10447961e-01 -3.92996311e-01 -2.06066042e-01 1.95175663e-01 3.28113228e-01 3.14509094e-01 -5.65214157e-01 2.37389326e-01 -6.44344240e-02 -2.18871042e-01 -2.61045247e-01 5.67614913e-01 -3.87944490e-01 -4.24943626e-01 2.58650422e-01 -8.15127194e-01 -6.84000999e-02 4.34438884e-02 -1.99245930e-01 4.67755161e-02 -1.20077759e-01 4.82284009e-01 -3.49962294e-01 -5.79031885e-01 -1.86040759e-01 -6.46171987e-01 2.91661769e-01 6.55948162e-01 1.70661777e-01 -5.58420360e-01 -8.77537072e-01 -6.90759301e-01 8.86109471e-01 -1.39708981e-01 6.31707370e-01 3.70491415e-01 -1.10153985e+00 -9.17311013e-01 1.05202250e-01 3.95240970e-02 -2.26877611e-02 2.14070797e-01 6.13138497e-01 -4.57298905e-01 5.14586031e-01 -2.74003059e-01 -1.82762116e-01 -1.43876266e+00 -1.72799733e-02 3.84755403e-01 -8.29662442e-01 -5.13511002e-01 9.93234873e-01 -7.07569301e-01 -1.38713348e+00 3.57294559e-01 1.65939450e-01 -3.80102932e-01 1.64529562e-01 6.26258969e-01 3.36191654e-01 -7.39305466e-02 -4.96217348e-02 1.32436514e-01 -5.60250819e-01 3.37423608e-02 -7.31966734e-01 1.44850647e+00 -2.92708576e-01 -1.21671475e-01 4.90526259e-01 6.17871702e-01 5.75798936e-02 -1.06376255e+00 -1.15468889e-01 3.51837009e-01 -6.00507632e-02 -3.58049601e-01 -1.63633931e+00 -1.18309267e-01 6.19062006e-01 6.06352463e-02 7.50954509e-01 5.72684765e-01 -3.25833410e-01 1.14255822e+00 1.03289771e+00 3.73435020e-01 -1.32396948e+00 4.53547388e-01 1.28134799e+00 1.06898212e+00 -1.43516648e+00 -5.11287928e-01 1.37602445e-02 -1.14193666e+00 9.81889009e-01 9.31011498e-01 4.25385416e-01 6.50290027e-02 4.46441546e-02 4.24554616e-01 -1.39530838e-01 -1.24022675e+00 7.21647665e-02 4.01004180e-02 6.38517439e-01 7.23433971e-01 1.73944741e-01 -1.70421436e-01 9.93161023e-01 -7.39199102e-01 2.73620337e-01 2.41190255e-01 5.19550204e-01 -3.84183586e-01 -1.07037675e+00 7.15321675e-02 -4.75776158e-02 -2.38773391e-01 -2.54262775e-01 -8.28633547e-01 5.81717789e-01 -4.85473186e-01 1.49661744e+00 -3.03195864e-01 -6.41068697e-01 5.59362411e-01 5.28128386e-01 7.29448125e-02 -5.93621790e-01 -1.19570982e+00 -5.00963092e-01 1.07063556e+00 -3.91121984e-01 3.24403457e-02 -2.84601420e-01 -1.00272346e+00 -6.65611565e-01 -2.88955271e-01 9.18142617e-01 5.98860919e-01 9.88535643e-01 3.04930061e-01 3.37968796e-01 8.91707361e-01 -7.61662900e-01 -9.49518621e-01 -1.72865355e+00 -3.86463851e-02 5.39527059e-01 3.16721916e-01 -4.01170254e-02 9.27293971e-02 -3.28073442e-01]
[12.780008316040039, 8.018976211547852]
07d5084d-72de-42ac-9741-b5a74ea5b687
boosting-cross-task-transferability-of
2304.05402
null
https://arxiv.org/abs/2304.05402v1
https://arxiv.org/pdf/2304.05402v1.pdf
Boosting Cross-task Transferability of Adversarial Patches with Visual Relations
The transferability of adversarial examples is a crucial aspect of evaluating the robustness of deep learning systems, particularly in black-box scenarios. Although several methods have been proposed to enhance cross-model transferability, little attention has been paid to the transferability of adversarial examples across different tasks. This issue has become increasingly relevant with the emergence of foundational multi-task AI systems such as Visual ChatGPT, rendering the utility of adversarial samples generated by a single task relatively limited. Furthermore, these systems often entail inferential functions beyond mere recognition-like tasks. To address this gap, we propose a novel Visual Relation-based cross-task Adversarial Patch generation method called VRAP, which aims to evaluate the robustness of various visual tasks, especially those involving visual reasoning, such as Visual Question Answering and Image Captioning. VRAP employs scene graphs to combine object recognition-based deception with predicate-based relations elimination, thereby disrupting the visual reasoning information shared among inferential tasks. Our extensive experiments demonstrate that VRAP significantly surpasses previous methods in terms of black-box transferability across diverse visual reasoning tasks.
['Shunchang Liu', 'Yisong Xiao', 'Songze Li', 'Tony Ma']
2023-04-11
null
null
null
null
['object-recognition', 'visual-reasoning', 'visual-reasoning']
['computer-vision', 'computer-vision', 'reasoning']
[ 4.09344435e-01 2.50052005e-01 4.10235077e-01 -1.17893621e-01 -8.29594553e-01 -9.65060234e-01 9.19602215e-01 -1.11860305e-01 -3.53245795e-01 6.42217934e-01 -1.58058435e-01 -5.16077936e-01 4.41572629e-02 -7.08551645e-01 -1.11073387e+00 -4.60878372e-01 3.16680193e-01 1.61750734e-01 1.33697942e-01 -3.67541790e-01 9.13712755e-02 4.85493153e-01 -1.19881439e+00 5.75290620e-01 1.14379132e+00 8.86355877e-01 -3.60707372e-01 4.94192213e-01 9.53720436e-02 1.14587080e+00 -1.09488809e+00 -1.16491437e+00 2.53131390e-01 -4.56759542e-01 -7.86858797e-01 -3.07877034e-01 7.95723379e-01 -2.82865882e-01 -5.39140463e-01 1.20143044e+00 4.19609129e-01 2.46842131e-01 7.18063414e-01 -1.89834809e+00 -1.38071465e+00 3.80181491e-01 -3.18201929e-01 1.37875140e-01 3.54175180e-01 6.45926118e-01 8.43569696e-01 -6.48271263e-01 3.61779362e-01 1.57133400e+00 4.08896118e-01 9.26854610e-01 -1.57733190e+00 -8.74051869e-01 1.68323979e-01 4.96266574e-01 -1.24788189e+00 -3.19356561e-01 8.55220020e-01 -4.69491512e-01 8.22763681e-01 5.02512932e-01 2.10395709e-01 1.66691351e+00 2.01456770e-01 8.48128796e-01 1.32964957e+00 -1.51187867e-01 2.78586447e-01 3.67223740e-01 -3.12588483e-01 4.34600174e-01 1.30471036e-01 2.38697022e-01 -4.64643389e-01 -5.50900660e-02 5.50749302e-01 -1.35320842e-01 -5.15544474e-01 -4.03207570e-01 -1.14152181e+00 7.40446866e-01 8.29011858e-01 -1.00314952e-01 -2.11821757e-02 2.00355560e-01 5.71502447e-01 5.44159949e-01 3.53073180e-01 9.11458135e-01 -5.29383495e-02 2.37723604e-01 -3.21220875e-01 4.08106327e-01 5.27173042e-01 9.57003951e-01 4.41198856e-01 3.98079991e-01 -6.87218130e-01 6.09358966e-01 1.35079008e-02 5.57486057e-01 2.09683329e-01 -8.78699780e-01 8.02807868e-01 6.47517741e-01 -1.42130136e-01 -1.09244668e+00 1.99190497e-01 -1.33297190e-01 -9.98322845e-01 6.50344670e-01 4.75188076e-01 6.31665858e-03 -8.35375190e-01 2.01315093e+00 3.18965137e-01 -1.06050953e-01 2.98336685e-01 1.01624560e+00 9.44069684e-01 4.04391140e-01 4.50647086e-01 3.66822422e-01 1.29197848e+00 -7.96237171e-01 -4.22147036e-01 -4.23910052e-01 1.75904244e-01 -4.48914915e-01 1.51270473e+00 2.24764407e-01 -8.91970575e-01 -5.26535332e-01 -9.56928611e-01 -2.29102150e-01 -5.65434575e-01 -5.75723350e-01 5.07726669e-01 6.34026647e-01 -7.39146709e-01 3.23867559e-01 -3.15454572e-01 2.19907001e-01 1.13008082e+00 7.84284100e-02 -6.49091661e-01 -3.11697066e-01 -1.42289841e+00 1.17295825e+00 3.68491054e-01 1.83805148e-03 -1.38398528e+00 -9.78110790e-01 -1.05544603e+00 1.33186325e-01 3.98532391e-01 -9.10539985e-01 9.96291995e-01 -1.33560944e+00 -1.18137550e+00 9.64255512e-01 2.61849999e-01 -3.72135490e-01 9.76894200e-01 -2.28565514e-01 -2.33873188e-01 1.46715000e-01 -6.13601543e-02 7.87122905e-01 1.30400121e+00 -1.53039706e+00 8.29759836e-02 -3.07636440e-01 6.32618308e-01 3.25563192e-01 -1.69604853e-01 -2.10424624e-02 -6.98983520e-02 -9.41979706e-01 -5.84572375e-01 -8.42119873e-01 1.70457169e-01 3.61530781e-01 -6.17622375e-01 -1.50278181e-01 7.68071473e-01 -6.96000636e-01 4.97880697e-01 -2.34473038e+00 4.71372068e-01 -2.35492110e-01 4.64164287e-01 4.89166588e-01 -3.53238344e-01 1.92080319e-01 -3.50730091e-01 4.04926479e-01 -3.66057485e-01 -7.75971636e-02 1.87590405e-01 1.02874555e-01 -7.75342047e-01 2.99563020e-01 7.42661953e-01 1.43329477e+00 -8.96045744e-01 -4.77757573e-01 1.34803742e-01 4.40844774e-01 -3.57750058e-01 4.62241083e-01 -3.77494484e-01 6.85600340e-01 -2.91268855e-01 5.15520155e-01 7.42456496e-01 -9.34001878e-02 -1.69382736e-01 -3.12830620e-02 6.47038758e-01 -1.77593693e-01 -4.20539498e-01 1.30668545e+00 -3.80063564e-01 7.96856105e-01 -2.70589501e-01 -8.80833685e-01 5.89780688e-01 2.27577403e-01 -2.27159962e-01 -1.05027080e+00 -5.75759895e-02 -3.03879917e-01 1.71102270e-01 -4.70760018e-01 3.31199020e-01 -4.37462717e-01 -1.61315322e-01 2.23217309e-01 -2.10523997e-02 -5.33202231e-01 -2.11972654e-01 5.34782112e-01 1.15495479e+00 3.26587290e-01 1.54364035e-01 9.15006995e-02 4.13212270e-01 4.82510030e-02 3.68870825e-01 6.66890442e-01 -4.98115361e-01 6.66435122e-01 8.23794186e-01 -1.78390816e-01 -9.75924850e-01 -1.41300726e+00 1.18717745e-01 9.45044160e-01 3.49856049e-01 9.22156963e-03 -7.45314777e-01 -1.22427678e+00 3.08649510e-01 1.02808559e+00 -1.08171177e+00 -8.04770470e-01 -2.52432257e-01 -3.43745202e-01 9.50604498e-01 3.97840619e-01 6.22107565e-01 -1.48663378e+00 -4.50285435e-01 -4.08222854e-01 -1.43924475e-01 -1.22311389e+00 -2.45292813e-01 -1.86022639e-01 -4.33169276e-01 -1.21721041e+00 -7.05012739e-01 -4.65967566e-01 7.68824100e-01 1.96436346e-01 1.34201241e+00 3.83139923e-02 -2.40622535e-01 4.55685854e-01 -3.88811499e-01 -4.29244190e-01 -7.33497083e-01 -2.95102745e-01 -3.05711210e-01 5.60305528e-02 9.81788635e-02 -5.21800280e-01 -5.20680070e-01 3.26594710e-01 -1.27060354e+00 1.02439001e-01 6.24412596e-01 8.89638066e-01 3.04691881e-01 -2.51324713e-01 6.41134381e-01 -8.98580492e-01 8.06524634e-01 -5.62601030e-01 -4.61369425e-01 5.61887264e-01 -1.23541802e-01 7.58171305e-02 1.00998545e+00 -8.37172329e-01 -1.10654366e+00 -5.40859401e-01 8.41949731e-02 -9.09676790e-01 -2.90124565e-01 1.31998152e-01 -5.68815410e-01 -2.67486304e-01 1.01003528e+00 1.77108094e-01 -4.81966995e-02 6.10629730e-02 6.89940453e-01 1.63236588e-01 5.90724468e-01 -6.85939968e-01 1.27771080e+00 3.43122423e-01 -3.46692353e-02 -5.29124975e-01 -7.20085382e-01 2.59688556e-01 -1.36541143e-01 -2.18329176e-01 1.04747343e+00 -6.98577225e-01 -9.62343693e-01 5.36524832e-01 -1.39466345e+00 -4.96170580e-01 -3.72971743e-01 -6.37098029e-02 -5.61654449e-01 3.81943792e-01 -4.04825419e-01 -6.61553085e-01 -2.77184635e-01 -1.23606205e+00 8.71117949e-01 2.34602988e-02 -2.50081539e-01 -8.20849478e-01 -1.37667388e-01 7.47989237e-01 2.45026544e-01 6.23311639e-01 1.33727598e+00 -6.61450565e-01 -8.15507054e-01 -4.95608076e-02 -5.53045809e-01 5.88931024e-01 -8.10816661e-02 -3.36090088e-01 -1.25771105e+00 -3.16129088e-01 -1.32716358e-01 -9.60690022e-01 7.06225276e-01 -2.39600375e-01 1.35778379e+00 -4.49406147e-01 -4.96160835e-02 6.28379107e-01 1.19328356e+00 1.28344074e-01 1.09275353e+00 2.36183971e-01 1.08134925e+00 7.15893686e-01 4.41551924e-01 -1.61112607e-01 2.02748179e-01 6.51688099e-01 9.39819753e-01 -1.83890745e-01 -3.13258648e-01 -3.71199489e-01 3.95695478e-01 2.53085941e-02 1.96886927e-01 -4.67186838e-01 -9.05399203e-01 4.46866602e-01 -1.61075270e+00 -1.05146801e+00 1.31146967e-01 1.94835007e+00 9.45700824e-01 1.49652502e-02 -2.64706820e-01 -1.56941693e-02 5.72405636e-01 1.51003852e-01 -9.38428700e-01 -4.66388971e-01 -1.85918689e-01 1.39357030e-01 -1.19122025e-02 1.70845300e-01 -1.07070637e+00 1.04282272e+00 5.48239517e+00 8.36279869e-01 -7.82943130e-01 1.82179719e-01 7.09777355e-01 1.02756068e-01 -6.41846299e-01 -2.00777560e-01 -6.13526031e-02 3.81626785e-01 4.50549334e-01 -1.78340003e-01 4.95657593e-01 8.85612607e-01 -3.44034106e-01 9.53884721e-02 -1.43402338e+00 8.85626733e-01 2.71746606e-01 -9.54842210e-01 5.05864620e-01 -4.75458167e-02 5.95882297e-01 -4.04873371e-01 6.01866364e-01 6.43765450e-01 4.66443807e-01 -1.46749103e+00 8.09426308e-01 3.55737388e-01 1.01169479e+00 -7.96713352e-01 4.82976735e-01 1.98452815e-01 -5.74483216e-01 1.47247612e-01 -1.60100535e-01 6.17919564e-02 -1.52458519e-01 1.24960393e-01 -7.68768966e-01 5.28804123e-01 8.43668461e-01 1.44414499e-01 -7.22674668e-01 6.19749904e-01 -7.25009441e-01 3.40938300e-01 2.51425683e-01 1.50110766e-01 2.82393452e-02 8.93251002e-02 6.47531748e-01 7.03659832e-01 -1.06368572e-01 -2.38906473e-01 -3.05695444e-01 1.41189945e+00 -4.44899082e-01 -3.00012529e-01 -1.06851292e+00 -5.88027462e-02 3.14013690e-01 1.03486121e+00 -2.06072420e-01 -1.53600261e-01 -3.56333941e-01 1.27924311e+00 6.72905207e-01 6.07698202e-01 -1.32874370e+00 -4.73725230e-01 9.53183770e-01 -1.77222341e-01 8.08090493e-02 7.19808340e-02 -3.41299981e-01 -1.18537068e+00 2.63109386e-01 -1.51118243e+00 2.47379839e-01 -1.02548873e+00 -1.57967877e+00 6.24805570e-01 1.96890999e-02 -1.04673636e+00 -6.68444782e-02 -6.97507501e-01 -5.85863888e-01 1.04009795e+00 -1.33691061e+00 -1.34095466e+00 -4.43486184e-01 8.02164316e-01 4.72746193e-01 -1.87907949e-01 8.08922768e-01 2.13063546e-02 -5.46019375e-01 1.03893924e+00 -4.11156237e-01 2.72164017e-01 8.40539098e-01 -1.18856657e+00 6.67073905e-01 1.04989064e+00 7.27406070e-02 5.72855771e-01 6.26524270e-01 -4.99036223e-01 -1.26268482e+00 -1.31487644e+00 2.34446168e-01 -1.01748967e+00 6.30026340e-01 -6.02198124e-01 -1.28742504e+00 8.19315076e-01 3.32302302e-01 1.56416088e-01 5.77388227e-01 -4.59738187e-02 -1.01158881e+00 -5.82044423e-02 -1.28011250e+00 1.15628242e+00 1.06931770e+00 -9.53913033e-01 -8.48629892e-01 2.14537099e-01 1.04078245e+00 -3.49066079e-01 -6.09866679e-01 3.55181873e-01 3.20849478e-01 -8.69029462e-01 1.22600198e+00 -1.04500544e+00 8.89431179e-01 -3.16889942e-01 -1.35839924e-01 -1.58490527e+00 -2.53999621e-01 -4.62041080e-01 -8.36346745e-02 1.31622446e+00 3.93065065e-01 -8.33769977e-01 4.77975547e-01 9.14157450e-01 -2.88294889e-02 -2.10264981e-01 -9.29586232e-01 -8.48369122e-01 4.46422666e-01 -2.64968961e-01 5.28813183e-01 1.20612907e+00 -2.52634645e-01 4.29866821e-01 -3.74986678e-01 2.05507472e-01 6.17977679e-01 6.06115200e-02 9.98986363e-01 -8.19761992e-01 -5.01678288e-01 -4.65778828e-01 -5.31018376e-01 -3.87658775e-01 6.76976681e-01 -1.04350090e+00 7.10214972e-02 -1.12393880e+00 2.74106383e-01 -2.06927478e-01 -2.05620900e-01 7.15990543e-01 -6.20455801e-01 3.68833303e-01 5.08034706e-01 1.05060354e-01 -5.94861865e-01 8.74594033e-01 1.61663663e+00 -5.54581225e-01 3.86375904e-01 -3.03220123e-01 -9.57308292e-01 5.05186379e-01 7.44783282e-01 -5.18880010e-01 -7.67448545e-01 -6.57647789e-01 3.76387954e-01 -1.60226926e-01 1.01743507e+00 -8.21010768e-01 -1.57257795e-01 -3.11920106e-01 4.57163781e-01 1.65200204e-01 4.10576195e-01 -7.36556351e-01 5.86793534e-02 3.17345679e-01 -4.04223531e-01 5.29098585e-02 6.45810068e-01 7.38813281e-01 -2.25653291e-01 1.44261345e-01 7.25141585e-01 -7.76204765e-02 -7.73458600e-01 2.45739430e-01 -5.26544750e-02 5.70314467e-01 1.38412821e+00 -8.16282555e-02 -9.06674981e-01 -3.14344466e-01 -4.62582499e-01 1.26245230e-01 5.46253920e-01 6.46255672e-01 8.47571135e-01 -1.34979093e+00 -8.70791495e-01 4.76915203e-02 5.25802195e-01 2.51079984e-02 5.85180700e-01 4.59309489e-01 -4.04150367e-01 7.01825023e-02 -5.41823745e-01 -4.52412307e-01 -1.23953605e+00 1.18855476e+00 4.58398461e-01 -1.49579704e-01 -5.09809077e-01 1.05903709e+00 1.02034569e+00 -4.27752703e-01 3.34319808e-02 1.55582592e-01 7.32833147e-02 -2.94258207e-01 4.48421210e-01 1.24785259e-01 -1.17473483e-01 -3.92520964e-01 -3.83911073e-01 8.63756910e-02 -8.60154703e-02 -6.70990050e-02 7.69932151e-01 3.32155406e-01 2.14244165e-02 1.65995479e-01 1.03634977e+00 -2.35820219e-01 -1.57314527e+00 -1.74214598e-03 -4.32315618e-01 -5.37470102e-01 -3.56774151e-01 -1.14146280e+00 -8.98829103e-01 1.15939021e+00 2.67066389e-01 2.19163403e-01 1.08994913e+00 -4.69790660e-02 3.88567746e-01 2.35409603e-01 2.97402889e-01 -5.50304949e-01 5.44593632e-01 2.66735435e-01 1.48908305e+00 -1.51206458e+00 -1.87254220e-01 -4.09546375e-01 -1.05787575e+00 3.39136362e-01 1.05120564e+00 -2.59650834e-02 -1.32978693e-01 -3.63078266e-02 8.78034234e-02 4.42470349e-02 -7.00934052e-01 -6.49331659e-02 5.14243662e-01 1.11883152e+00 1.11999027e-02 2.23899838e-02 3.41787219e-01 2.92402059e-01 -1.01066478e-01 -4.12030071e-01 1.19963825e-01 6.92865849e-01 4.00924504e-01 -8.65291953e-01 -4.44204807e-01 4.06601541e-02 -5.01319945e-01 -3.64502162e-01 -7.31778443e-01 1.03952777e+00 -4.05111276e-02 8.49139154e-01 -1.13402694e-01 -3.91274512e-01 3.77469718e-01 6.98521659e-02 7.02535868e-01 -4.87747073e-01 -7.64861524e-01 -9.01370108e-01 1.39219627e-01 -6.06983781e-01 2.01815758e-02 -2.54599720e-01 -7.60081649e-01 -5.15316129e-01 1.69853810e-02 -1.10664621e-01 3.71247619e-01 8.81380320e-01 4.98750448e-01 8.44866812e-01 3.46611351e-01 -6.79140270e-01 -6.77034736e-01 -7.18490779e-01 -2.09384650e-01 1.01838148e+00 3.64623576e-01 -7.09969878e-01 -3.62485975e-01 -2.05372125e-02]
[10.867973327636719, 1.8247935771942139]
38f54ac0-d588-4fa6-b762-8ee1b768ce2c
cab-comprehensive-attention-benchmarking-on
2210.07661
null
https://arxiv.org/abs/2210.07661v3
https://arxiv.org/pdf/2210.07661v3.pdf
CAB: Comprehensive Attention Benchmarking on Long Sequence Modeling
Transformer has achieved remarkable success in language, image, and speech processing. Recently, various efficient attention architectures have been proposed to improve transformer's efficiency while largely preserving its efficacy, especially in modeling long sequences. A widely-used benchmark to test these efficient methods' capability on long-range modeling is Long Range Arena (LRA). However, LRA only focuses on the standard bidirectional (or noncausal) self attention, and completely ignores cross attentions and unidirectional (or causal) attentions, which are equally important to downstream applications. In this paper, we propose Comprehensive Attention Benchmark (CAB) under a fine-grained attention taxonomy with four distinguishable attention patterns, namely, noncausal self, causal self, noncausal cross, and causal cross attentions. CAB collects seven real-world tasks from different research areas to evaluate efficient attentions under the four attention patterns. Among these tasks, CAB validates efficient attentions in eight backbone networks to show their generalization across neural architectures. We conduct exhaustive experiments to benchmark the performances of nine widely-used efficient attention architectures designed with different philosophies on CAB. Extensive experimental results also shed light on the fundamental problems of efficient attentions, such as efficiency length against vanilla attention, performance consistency across attention patterns, the benefit of attention mechanisms, and interpolation/extrapolation on long-context language modeling.
['Lingpeng Kong', 'Lin Zheng', 'Jiangtao Feng', 'Shuyang Jiang', 'Jun Zhang']
2022-10-14
null
null
null
null
['long-range-modeling']
['natural-language-processing']
[-9.55649912e-02 -5.03140092e-01 -3.62510473e-01 -1.81109041e-01 -4.93403345e-01 -1.60245553e-01 6.35571241e-01 -2.66482770e-01 -4.59025264e-01 6.84260249e-01 5.47587156e-01 -5.19697070e-01 -2.99801379e-01 -6.38184428e-01 -8.31889689e-01 -6.47878170e-01 -1.78841308e-01 3.66317689e-01 2.20849231e-01 -4.80902791e-01 3.64800781e-01 4.14759278e-01 -1.46066701e+00 2.53333211e-01 8.73921931e-01 7.71802485e-01 5.41185915e-01 7.63947070e-01 -2.97862440e-01 1.02645373e+00 -7.57619679e-01 -3.48841876e-01 -2.58782655e-01 -2.87992716e-01 -8.68463874e-01 -6.80462897e-01 -1.44849852e-01 -6.39428645e-02 -5.09151578e-01 7.50489235e-01 7.02659190e-01 1.97206125e-01 5.89207649e-01 -1.47334564e+00 -1.61804116e+00 9.89586055e-01 -6.83741331e-01 8.27066004e-01 1.55315593e-01 4.73081589e-01 1.26326716e+00 -9.25779998e-01 8.80808979e-02 1.73058534e+00 8.25168371e-01 5.20189524e-01 -6.55415714e-01 -9.72275913e-01 6.05462849e-01 6.24621332e-01 -1.47577631e+00 -2.55638301e-01 4.27881628e-01 -3.58709008e-01 1.43920124e+00 4.19086605e-01 4.87442464e-01 1.35947430e+00 6.89935446e-01 9.45399046e-01 8.44573438e-01 -2.19687253e-01 -1.87076241e-01 -3.74420702e-01 6.43193185e-01 4.17734236e-01 -6.85103759e-02 3.14962685e-01 -5.26206315e-01 -4.93652858e-02 7.12214649e-01 -3.86695527e-02 -3.24324906e-01 5.00586212e-01 -1.44088995e+00 6.80336773e-01 6.42959595e-01 5.97749114e-01 -3.90732437e-01 4.54082131e-01 3.57752651e-01 6.70168996e-02 3.91226619e-01 2.61444002e-01 -6.83462083e-01 -6.11962862e-02 -3.58352035e-01 9.56007987e-02 2.47845724e-01 1.26515293e+00 4.80790854e-01 3.67769897e-01 -9.89632249e-01 1.04795074e+00 3.60874325e-01 4.54782724e-01 6.23966634e-01 -5.75168252e-01 3.09211552e-01 2.45013714e-01 -2.89672047e-01 -9.55112100e-01 -5.54363012e-01 -6.83449090e-01 -1.43479061e+00 -3.63306075e-01 6.39679357e-02 1.07694902e-01 -7.52366424e-01 2.06727147e+00 -3.30682397e-01 4.28176045e-01 -1.72845528e-01 9.03701007e-01 9.60933626e-01 9.52369094e-01 5.22235155e-01 -3.76714855e-01 1.46998906e+00 -1.28902328e+00 -8.97919774e-01 -2.91342646e-01 3.33804965e-01 -8.28918099e-01 1.53969443e+00 1.30166352e-01 -1.06486082e+00 -7.93120146e-01 -8.25348854e-01 -4.91123050e-01 -3.45402151e-01 -2.32321203e-01 6.79973602e-01 2.91547745e-01 -1.12083030e+00 4.02295411e-01 -4.04632568e-01 -1.21954992e-01 3.29678178e-01 2.16508552e-01 1.38648242e-01 4.58024442e-02 -1.85684967e+00 8.94955575e-01 7.55990520e-02 1.44188598e-01 -1.03308523e+00 -1.03736699e+00 -5.44153631e-01 3.67715150e-01 1.04957290e-01 -9.82262909e-01 1.13842630e+00 -8.62575114e-01 -1.33270371e+00 4.77082580e-01 -3.18587363e-01 -5.85669816e-01 2.23967284e-01 -4.61000919e-01 -4.31670636e-01 -5.03790379e-01 2.06093386e-01 8.96807373e-01 4.98665571e-01 -7.25833833e-01 -4.11045045e-01 -5.28403893e-02 2.31512979e-01 -1.09214000e-01 -2.97354162e-01 4.46427464e-01 -6.37968957e-01 -1.11330056e+00 -5.60239136e-01 -8.25942695e-01 -1.55128539e-01 -3.68506461e-01 -4.49663222e-01 -7.55422890e-01 7.79811442e-01 -4.11785483e-01 1.69435930e+00 -2.02280307e+00 1.71921864e-01 -4.63184476e-01 5.35131283e-02 1.48217201e-01 -4.86106038e-01 2.08127096e-01 -4.20616657e-01 4.85649765e-01 -2.81546235e-01 -1.60679042e-01 -1.23821259e-01 3.72097701e-01 -5.60843050e-01 1.27376989e-01 2.77835816e-01 1.38171422e+00 -9.26654935e-01 -5.32044590e-01 -3.05490531e-02 6.45745814e-01 -6.73661888e-01 1.96006224e-01 -2.39504233e-01 4.82610703e-01 -3.39858890e-01 5.91599584e-01 4.07245010e-01 -4.71370786e-01 -2.81486660e-01 -1.91610262e-01 -2.22302571e-01 2.56726623e-01 -4.17331278e-01 1.43112504e+00 -8.25808764e-01 8.07055175e-01 -4.18031007e-01 -7.20179677e-01 7.22023189e-01 1.87143043e-01 2.23806873e-01 -1.01591253e+00 1.10578515e-01 -8.25280622e-02 4.49152976e-01 -3.72833639e-01 5.77477515e-01 8.60793516e-02 1.71411708e-02 4.01440918e-01 2.95926500e-02 5.50461948e-01 9.76472944e-02 1.53580368e-01 9.67564464e-01 -1.24767579e-01 3.33671838e-01 -7.59424984e-01 6.76053464e-01 -4.90965396e-01 6.16524577e-01 1.01158202e+00 -2.78149813e-01 6.42661810e-01 4.49872375e-01 -4.78908002e-01 -1.05655730e+00 -7.21245289e-01 -5.86415380e-02 1.75180399e+00 2.04617441e-01 -4.10911292e-01 -4.15185094e-01 -5.39448619e-01 -3.57002079e-01 9.25641775e-01 -7.34090447e-01 -4.10571545e-01 -7.04391897e-01 -1.11549270e+00 1.01537418e+00 9.19520199e-01 6.53498650e-01 -1.56331897e+00 -4.46621358e-01 1.51194885e-01 -4.79938149e-01 -8.49498153e-01 -8.17055166e-01 -2.64337733e-02 -3.42725605e-01 -7.96842873e-01 -1.11352324e+00 -5.15491664e-01 8.67136940e-02 4.05914515e-01 1.33901644e+00 1.72688276e-01 1.95652898e-02 2.40697712e-01 -9.44119692e-02 -5.69861889e-01 -6.77838326e-02 2.63312340e-01 4.38061766e-02 -1.66095674e-01 1.76827654e-01 -5.15534580e-01 -4.51263905e-01 6.79796219e-01 -6.07559383e-01 -2.29766443e-02 9.59554970e-01 1.07481921e+00 5.35856485e-01 -2.77148604e-01 7.82457769e-01 -4.29637760e-01 9.60690677e-01 -9.04435992e-01 -3.56335342e-01 4.65691179e-01 -5.15376747e-01 3.35182875e-01 6.19420290e-01 -6.01594448e-01 -1.05468023e+00 -8.67214561e-01 -5.67165136e-01 -6.69693649e-01 1.35087058e-01 5.16943753e-01 -1.85321495e-01 2.90132910e-01 3.31231713e-01 4.83233273e-01 -4.24477130e-01 -4.10441786e-01 3.52485031e-01 4.37822759e-01 2.72912055e-01 -7.28599846e-01 1.07724011e-01 1.06084101e-01 -1.59861326e-01 -9.85368371e-01 -8.45593631e-01 -2.71546960e-01 -2.46119916e-01 3.25255133e-02 1.04639840e+00 -7.37413943e-01 -9.00458813e-01 6.20302796e-01 -1.62016964e+00 -3.98720205e-01 -5.82074523e-02 3.21508795e-01 -4.70777184e-01 6.89186454e-02 -8.07948351e-01 -7.30295777e-01 -6.18771732e-01 -1.45580053e+00 1.04414761e+00 8.89950320e-02 -1.55378446e-01 -9.43332434e-01 -2.28582650e-01 -3.90212722e-02 7.32483566e-01 -4.93394285e-01 1.20064378e+00 -4.70688969e-01 -6.68435335e-01 6.14797413e-01 -5.87070942e-01 2.10879371e-01 -2.64039874e-01 6.62484318e-02 -1.06263232e+00 -6.61091460e-03 -2.35016987e-01 -8.95370767e-02 1.02852786e+00 7.03217566e-01 1.62666440e+00 -2.87807941e-01 -2.93122679e-01 6.31013453e-01 8.02241981e-01 5.34980714e-01 8.15246463e-01 2.82184809e-01 6.55109286e-01 2.95119017e-01 4.09972519e-01 -1.83250784e-04 3.28169048e-01 5.85879564e-01 4.30431962e-01 7.05592334e-02 -2.94956297e-01 -1.13956295e-01 4.49087620e-01 1.32751477e+00 -3.42582494e-01 -7.04752088e-01 -9.85456467e-01 5.74681818e-01 -1.77848673e+00 -1.04656792e+00 -3.10180485e-01 1.74412358e+00 7.09790885e-01 2.55513817e-01 -2.58956343e-01 -8.60450640e-02 8.53972495e-01 3.33854139e-01 -5.05041540e-01 -7.05646396e-01 -4.28161561e-01 6.27156571e-02 3.02945852e-01 4.79254574e-01 -8.48383069e-01 9.06746328e-01 6.99609804e+00 1.25202537e+00 -1.01467335e+00 5.21114290e-01 9.74142909e-01 5.71342856e-02 -4.71024007e-01 -1.90080464e-01 -1.04008269e+00 4.83971030e-01 1.09690142e+00 -4.88800049e-01 3.12133372e-01 8.91985238e-01 1.26749158e-01 3.15793097e-01 -9.54551518e-01 9.35698032e-01 1.55821191e-02 -1.54433060e+00 4.92747962e-01 -2.12566797e-02 6.68568790e-01 3.23475569e-01 2.29388714e-01 7.78632402e-01 3.11386257e-01 -1.19229507e+00 7.23692536e-01 6.59104943e-01 7.22147465e-01 -8.37531805e-01 7.89376259e-01 3.74842733e-01 -1.37791574e+00 -2.58858204e-01 -4.13943887e-01 -3.44922543e-01 1.32537201e-01 3.35174352e-01 -1.46264762e-01 6.18084311e-01 9.66729701e-01 9.58217204e-01 -4.32621628e-01 7.94912755e-01 -2.84882575e-01 6.10282481e-01 -7.99649954e-02 -2.34628186e-01 5.66449702e-01 8.23621154e-02 7.29428053e-01 1.55213881e+00 4.17428970e-01 1.98810712e-01 -2.61939377e-01 1.01131666e+00 -1.86014354e-01 -3.05969771e-02 -5.93398571e-01 1.80393308e-01 4.98597413e-01 8.32434356e-01 -4.41152811e-01 -3.42680812e-01 -4.10909891e-01 5.36875427e-01 2.90588111e-01 4.28619325e-01 -1.44277406e+00 -1.89969182e-01 9.43366945e-01 5.56649156e-02 -3.95212360e-02 -1.38924301e-01 -2.14448452e-01 -9.41618741e-01 -4.70813066e-01 -7.53871143e-01 5.71750283e-01 -1.21802378e+00 -1.39321876e+00 7.69176066e-01 1.69714525e-01 -7.93936014e-01 1.43829927e-01 -3.32425117e-01 -8.90048504e-01 1.15996933e+00 -1.54783762e+00 -1.11396313e+00 -2.12877974e-01 7.37029910e-01 1.19782710e+00 -1.84287235e-01 4.76657003e-01 6.55170321e-01 -8.65119100e-01 8.18433642e-01 -2.83486813e-01 -3.08172125e-02 4.92524654e-01 -7.83778489e-01 5.14503598e-01 8.48081291e-01 1.58344507e-02 1.03866291e+00 3.95931870e-01 -3.90921593e-01 -1.20305622e+00 -1.41014826e+00 1.04261410e+00 -1.55168250e-01 8.42887461e-01 -1.76355287e-01 -1.23250902e+00 1.01501739e+00 6.54110074e-01 1.68839730e-02 2.19296530e-01 4.12906498e-01 -4.04342979e-01 -5.71259670e-02 -3.35879475e-01 7.78620124e-01 1.51097965e+00 -3.46736699e-01 -5.84889174e-01 2.07176760e-01 1.34565043e+00 -7.93161690e-02 -5.54611981e-01 5.91762900e-01 4.98975337e-01 -1.02895212e+00 1.23571754e+00 -7.68377244e-01 6.59412980e-01 -2.98705064e-02 -2.11392567e-01 -1.14022315e+00 -8.91423821e-01 -5.28946340e-01 -5.04042916e-02 1.35061705e+00 1.67521060e-01 -7.85231650e-01 1.65367033e-02 -8.16851035e-02 -6.52728200e-01 -1.15760899e+00 -9.76541102e-01 -9.50844646e-01 4.47141230e-01 -8.39667916e-01 8.34035814e-01 1.01153648e+00 -6.11970603e-01 7.60451078e-01 -5.98659098e-01 1.49411798e-01 2.65819758e-01 4.42558676e-02 4.48046148e-01 -8.77229035e-01 -1.71420246e-01 -1.00644934e+00 -1.04359746e-01 -1.37270677e+00 3.29281837e-01 -8.37413728e-01 -1.71174258e-01 -1.28636074e+00 3.51796508e-01 -5.88530958e-01 -6.16540909e-01 4.98742759e-01 -3.91409189e-01 1.33181382e-02 1.52419209e-01 3.61223549e-01 -6.69228017e-01 8.37306559e-01 1.29478908e+00 -3.56366783e-01 1.04018122e-01 -2.15748832e-01 -8.48586202e-01 9.62691009e-01 6.41921878e-01 -3.34199160e-01 -5.72443128e-01 -8.75985384e-01 2.46745706e-01 -1.41772777e-01 3.78516644e-01 -9.68069792e-01 3.58037740e-01 -4.21484530e-01 2.60803308e-02 -7.93136179e-01 3.71293910e-02 -3.62198949e-01 9.40734521e-02 4.83714640e-01 -5.26892364e-01 5.19557953e-01 4.46527869e-01 4.65939492e-01 -2.65760750e-01 1.32435992e-01 8.94116342e-01 -1.15663320e-01 -1.03051674e+00 5.57892501e-01 -2.66330093e-01 4.79191810e-01 9.17634845e-01 2.38971651e-01 -4.83996481e-01 -2.50082731e-01 -7.16068149e-01 3.66557807e-01 -2.86478072e-01 1.02135289e+00 6.00353837e-01 -1.48903763e+00 -8.57565224e-01 2.09771439e-01 6.71429262e-02 -2.03917935e-01 6.47844434e-01 9.78398919e-01 -1.49921924e-01 9.42473948e-01 -1.44217774e-01 -8.68035793e-01 -1.05621469e+00 1.02042186e+00 3.48229945e-01 -6.07141256e-01 -3.52922440e-01 1.06435657e+00 1.11252224e+00 -2.21199200e-01 1.04001746e-01 -5.61316669e-01 -2.90008426e-01 -1.89849213e-01 6.25523627e-01 4.38456625e-01 -1.57594889e-01 -6.16415083e-01 -4.84649092e-01 6.50332272e-01 1.06526259e-02 3.58314574e-01 8.93869102e-01 -2.40790427e-01 -6.69604391e-02 7.42288888e-01 9.52134848e-01 -2.27241695e-01 -8.88767898e-01 -1.23855539e-01 -1.74218416e-01 -1.61905661e-01 2.41256002e-02 -8.22167039e-01 -1.09460855e+00 1.33490920e+00 1.98195770e-01 2.38009661e-01 1.27128673e+00 1.39210019e-02 6.77658737e-01 3.15323398e-02 2.25064158e-01 -4.90543336e-01 1.66990906e-01 1.09936011e+00 1.61362123e+00 -9.68012094e-01 -2.82607317e-01 -9.03508067e-02 -6.03515744e-01 7.48182058e-01 8.59597445e-01 9.74689871e-02 7.25455225e-01 3.45500767e-01 -3.41220379e-01 7.92060941e-02 -1.28935122e+00 -1.99396342e-01 2.84061581e-01 4.06541973e-01 5.48402965e-01 -1.38378650e-01 -3.57565433e-01 8.21089625e-01 4.69205454e-02 -1.05371952e-01 1.07499808e-01 3.81034046e-01 -9.33759809e-02 -7.19262838e-01 -3.08458805e-01 2.22591698e-01 -6.13102913e-01 -6.83416009e-01 6.81350706e-03 1.05008423e+00 2.95291662e-01 7.18164265e-01 3.11811447e-01 -2.24111289e-01 3.69221866e-01 -5.38569391e-02 4.48569134e-02 -1.22548290e-01 -7.04638183e-01 8.60121772e-02 -1.28515512e-01 -5.56472301e-01 -3.63619894e-01 -2.72902906e-01 -1.09744239e+00 -5.94328403e-01 -2.60507017e-01 3.84703167e-02 1.87938631e-01 1.00963593e+00 7.37824261e-01 1.28868449e+00 2.81526923e-01 -6.56234324e-01 -2.17752248e-01 -1.28139508e+00 -5.11245150e-03 1.99742511e-01 2.52672762e-01 -8.83876801e-01 -9.38562378e-02 -2.39841044e-01]
[10.907670021057129, 6.636829853057861]
dfd5bbb1-1753-4791-b36e-bafc909ff8bc
fuzzy-controller-of-reward-of-reinforcement
1812.07028
null
http://arxiv.org/abs/1812.07028v1
http://arxiv.org/pdf/1812.07028v1.pdf
Fuzzy Controller of Reward of Reinforcement Learning For Handwritten Digit Recognition
Recognition of human environment with computer systems always was a big deal in artificial intelligence. In this area handwriting recognition and conceptualization of it to computer is an important area in it. In the past years with growth of machine learning in artificial intelligence, efforts to using this technique increased. In this paper is tried to using fuzzy controller, to optimizing amount of reward of reinforcement learning for recognition of handwritten digits. For this aim first a sample of every digit with 10 standard computer fonts, given to actor and then actor is trained. In the next level is tried to test the actor with dataset and then results show improvement of recognition when using fuzzy controller of reinforcement learning.
['Saber Malekzadeh']
2018-12-17
null
null
null
null
['handwritten-digit-recognition']
['computer-vision']
[ 3.84629220e-01 1.37160450e-01 5.02111092e-02 -4.79644388e-01 5.49881101e-01 -7.84300268e-01 5.70481479e-01 -4.18589078e-02 -5.38606226e-01 8.86970460e-01 -8.07282999e-02 -3.53361368e-01 -2.59492368e-01 -8.28109980e-01 -3.29404712e-01 -3.46376121e-01 2.95600265e-01 7.22997844e-01 2.65265405e-01 -2.20080063e-01 8.78451228e-01 1.05301237e+00 -1.17642963e+00 2.97482163e-01 3.78229797e-01 4.87120211e-01 1.32546604e-01 1.20619416e+00 -1.64605409e-01 1.29967225e+00 -8.52938175e-01 -1.38905987e-01 5.95191538e-01 -6.51314855e-01 -9.66887534e-01 6.09595120e-01 -1.04259208e-01 -2.25938007e-01 -2.32227996e-01 9.61359859e-01 3.10094088e-01 3.04554373e-01 1.04276812e+00 -9.71869588e-01 -7.18506455e-01 5.70203066e-01 -3.88208836e-01 4.20981109e-01 1.77328870e-01 3.31654519e-01 1.12095281e-01 -2.83638507e-01 5.28494596e-01 1.07796633e+00 9.47271138e-02 6.15653038e-01 -1.01287568e+00 -2.23016158e-01 -3.92576754e-01 3.49589497e-01 -8.00329328e-01 1.78726539e-01 6.31681740e-01 -4.49815631e-01 1.32176614e+00 -5.11411019e-02 6.96829617e-01 4.18018818e-01 4.79952209e-02 6.82444155e-01 1.68483138e+00 -7.72764623e-01 1.02815710e-01 6.41592979e-01 6.68629289e-01 4.45460320e-01 -4.70297970e-03 4.22349960e-01 -3.51106450e-02 6.25039518e-01 8.71052861e-01 -1.10872164e-02 3.36822748e-01 5.84333837e-01 -8.51767957e-01 5.16645730e-01 1.13500044e-01 9.25361931e-01 -6.93915188e-01 1.11873373e-01 3.43497396e-01 4.60197210e-01 -5.71492374e-01 6.65906847e-01 -2.11379468e-01 -4.75831389e-01 -5.82477033e-01 -9.71749984e-03 8.33820045e-01 6.19687438e-01 2.43307605e-01 4.58658457e-01 1.60653263e-01 4.93024141e-01 4.49084379e-02 6.02548420e-01 8.33279729e-01 -6.53715014e-01 2.46640041e-01 1.04164398e+00 -1.05289202e-02 -8.59775245e-01 -3.15860331e-01 -1.41779054e-02 -6.42258823e-01 1.12140918e+00 6.69441938e-01 -3.44740331e-01 -1.09431159e+00 7.29586542e-01 2.64194980e-02 -5.14386185e-02 7.08275139e-01 7.11498559e-01 -1.89453196e-02 8.52294385e-01 -1.74089625e-01 -3.02558810e-01 1.16259408e+00 -6.87685788e-01 -7.18829393e-01 3.31406593e-01 -9.66251735e-03 -1.00259984e+00 8.72530878e-01 1.18192148e+00 -7.29174614e-01 -7.87209272e-01 -1.20118034e+00 6.00199401e-01 -6.33648276e-01 5.13961017e-01 3.97892475e-01 8.40014040e-01 -5.96067131e-01 1.08248901e+00 -6.90266967e-01 -5.53143740e-01 1.43700093e-01 7.36442268e-01 -2.68201590e-01 2.09558740e-01 -8.32784593e-01 1.27643752e+00 1.00640070e+00 -1.54825047e-01 -3.64147604e-01 1.25554949e-01 -1.38287634e-01 -1.93580136e-01 1.06896035e-01 3.19976240e-01 1.09566128e+00 -1.79825985e+00 -1.80579889e+00 5.24004579e-01 3.98611605e-01 -8.78191173e-01 5.25155425e-01 6.55886307e-02 -5.53139567e-01 1.10658169e-01 -6.55081153e-01 4.30971026e-01 8.90306950e-01 -7.91597724e-01 -8.00475061e-01 -4.12941366e-01 -2.06610933e-01 2.54389077e-01 -1.67294502e-01 1.42148901e-02 2.12172091e-01 -3.91762286e-01 -2.14853585e-01 -6.15285873e-01 6.56332672e-02 -5.60896397e-01 1.51337430e-01 -3.27463776e-01 8.51885676e-01 -7.39913464e-01 9.23358917e-01 -1.84579968e+00 -7.94035569e-02 5.45326591e-01 -6.28646970e-01 7.63408661e-01 1.35238543e-01 3.21223497e-01 1.00366019e-01 -9.21312645e-02 2.05621123e-02 8.72856677e-01 -3.54415715e-01 3.69957715e-01 -5.28622270e-02 6.67535961e-02 4.86199260e-01 2.07145005e-01 -6.29289269e-01 -5.70714176e-01 3.51520240e-01 8.17934945e-02 -6.27770498e-02 1.66214779e-01 -3.48068297e-01 1.25403598e-01 -5.99262714e-01 5.71266413e-01 5.75587153e-01 1.59280419e-01 1.83088228e-01 -1.38989538e-01 -2.14489385e-01 -5.34767449e-01 -1.39361382e+00 6.23617768e-01 -3.92788788e-03 7.65488505e-01 -5.80272555e-01 -1.43360353e+00 1.49015260e+00 4.28934336e-01 1.33108199e-01 -7.99544513e-01 3.04024935e-01 2.79125422e-01 5.82223892e-01 -9.50736940e-01 9.72334668e-02 1.38964120e-03 6.82675242e-01 4.80162203e-01 2.03493923e-01 -2.78436273e-01 4.99902904e-01 -3.32831711e-01 8.29739690e-01 4.86800462e-01 4.69385058e-01 -2.32800026e-03 8.53602409e-01 5.10119975e-01 1.22746034e-02 4.78556305e-01 -8.28310996e-02 -8.53555426e-02 3.61845016e-01 -5.49999893e-01 -1.51151597e+00 -3.49220008e-01 -1.11707821e-01 3.72609168e-01 -4.87624317e-01 7.19293237e-01 -8.64613354e-01 -5.33659220e-01 1.11158088e-01 5.49517632e-01 -3.99110466e-01 1.08283237e-01 -8.88374090e-01 -2.77608842e-01 5.95941603e-01 2.71519572e-01 1.05559599e+00 -1.57625794e+00 -1.16660142e+00 6.05430007e-01 1.04795516e+00 -5.89648485e-01 1.67457476e-01 4.21988428e-01 -1.26694608e+00 -1.18978012e+00 -8.53321433e-01 -7.75437891e-01 6.46066189e-01 -5.78543305e-01 4.84142154e-01 2.31142268e-01 -6.43053651e-01 2.06254125e-01 -6.59241080e-01 -8.75699162e-01 -7.08408713e-01 -1.19492933e-01 -1.04172491e-01 -2.00110987e-01 5.76317906e-01 -2.94349223e-01 -2.19819412e-01 3.76678631e-02 -8.94970953e-01 -2.68513590e-01 1.08493757e+00 9.90289986e-01 -1.14797078e-01 2.75638342e-01 5.20559251e-01 -7.42502868e-01 9.31956351e-01 1.02582254e-01 -1.02119231e+00 5.21752477e-01 -9.03624475e-01 3.96107525e-01 9.24132586e-01 -5.97142339e-01 -7.73586154e-01 3.55587929e-01 2.23597899e-01 1.24959135e-02 -4.61328566e-01 3.31803530e-01 2.69773066e-01 -5.81657253e-02 9.58438456e-01 4.42154765e-01 5.09724379e-01 -1.26446798e-01 -1.09082937e-01 1.16557515e+00 3.89897317e-01 -3.25194657e-01 3.98453385e-01 -3.86889189e-01 3.61828893e-01 -9.65415001e-01 2.04065874e-01 -6.55495971e-02 -5.47395825e-01 -6.74398065e-01 8.18523705e-01 2.55904019e-01 -1.09630418e+00 7.26432264e-01 -1.24178100e+00 -2.18854889e-01 1.44602954e-02 6.92239642e-01 -5.18015206e-01 -1.25905909e-02 -2.10660264e-01 -1.58870399e+00 -2.22307563e-01 -1.05652392e+00 9.16062947e-03 7.67186761e-01 2.41641253e-02 -7.90527463e-01 -5.32963164e-02 5.65497279e-02 1.75408155e-01 3.56662571e-01 7.20217586e-01 -7.57670462e-01 -2.64134616e-01 -4.38561469e-01 -1.86443813e-02 9.53649700e-01 4.25088108e-01 4.07835215e-01 -5.54018736e-01 2.85507470e-01 1.25282094e-01 -4.89934951e-01 4.32278812e-01 2.28780508e-02 7.50548661e-01 -4.44015890e-01 -1.13762021e-02 -1.02634303e-01 1.74581575e+00 1.57552123e+00 7.89564312e-01 5.12325108e-01 1.14057204e-02 2.96786398e-01 9.49189723e-01 4.72836763e-01 -5.99714875e-01 2.66200751e-01 -1.23967223e-01 2.37399265e-01 4.83369455e-02 1.19283788e-01 4.10639733e-01 2.75072545e-01 -6.55486703e-01 -1.30517364e-01 -1.19855344e+00 3.43644097e-02 -1.47604871e+00 -1.21723175e+00 8.36821124e-02 1.83627081e+00 9.15676117e-01 4.37132150e-01 2.98238039e-01 9.66879368e-01 7.91362941e-01 -6.27531111e-01 -5.11220574e-01 -8.37972701e-01 -1.88175172e-01 4.56065565e-01 6.83440864e-01 8.66173804e-01 -6.03560150e-01 1.07450318e+00 5.16256952e+00 3.04897994e-01 -1.71892357e+00 -5.70467710e-01 8.20788443e-01 4.26976025e-01 5.19732177e-01 -4.76530828e-02 -7.62977779e-01 5.59070468e-01 7.37163961e-01 -1.84619337e-01 1.08579576e+00 6.60559654e-01 4.01734531e-01 -5.44711292e-01 -8.60549569e-01 7.92211890e-01 8.66275355e-02 -9.74310100e-01 3.52114022e-01 -1.85856625e-01 6.65767074e-01 -8.41182590e-01 -8.20316449e-02 -8.29161033e-02 1.86704442e-01 -1.45383620e+00 2.97655731e-01 1.04247773e+00 1.46159902e-01 -1.03265584e+00 8.70964885e-01 4.84235913e-01 -5.27157128e-01 -3.88654798e-01 -7.01394439e-01 -1.82210848e-01 -1.65712938e-01 -1.46907806e-01 -1.34120500e+00 4.87369746e-02 2.04212278e-01 5.28369009e-01 -4.97987300e-01 1.05388165e+00 1.37089387e-01 8.35643828e-01 -3.43626469e-01 -9.78949428e-01 5.24885058e-01 -4.29201126e-01 1.24434970e-01 1.27469862e+00 1.97617739e-01 6.11336708e-01 -2.73217350e-01 6.06807888e-01 2.29536623e-01 2.23883554e-01 -8.98528814e-01 -5.96993327e-01 3.67059976e-01 7.68568695e-01 -9.58299339e-01 -6.27187908e-01 1.62276447e-01 9.27488983e-01 2.00725213e-01 1.82416484e-01 -3.77738833e-01 -8.18117797e-01 -2.79119581e-01 8.31422880e-02 -1.15257334e-02 -2.96923012e-01 -4.31046844e-01 -2.73750216e-01 -1.73705563e-01 -1.08119595e+00 1.66807562e-01 -6.93673611e-01 -6.62899375e-01 6.17939949e-01 1.88576733e-03 -1.09581709e+00 -3.20541143e-01 -1.37498212e+00 -6.30898416e-01 1.03818333e+00 -6.95049584e-01 -5.58242381e-01 -2.10950047e-01 6.59810185e-01 9.33255672e-01 -1.21589375e+00 6.04714930e-01 -1.22318789e-01 -1.49912462e-01 2.23302305e-01 3.06774557e-01 4.45168614e-01 5.98373652e-01 -1.42454410e+00 -3.87803078e-01 7.85525382e-01 2.17790619e-01 3.08044523e-01 7.74933279e-01 -7.99860835e-01 -1.29732144e+00 -4.68875408e-01 4.15317565e-01 1.98899820e-01 4.09755230e-01 3.82876664e-01 -4.90363002e-01 4.28902239e-01 7.92879343e-01 -3.76420319e-01 1.51012257e-01 -7.05496192e-01 4.93955821e-01 -3.45594287e-01 -1.42526126e+00 3.54149401e-01 1.12705819e-01 8.54995847e-03 -6.56485081e-01 3.12214315e-01 -2.29000703e-01 -2.81334162e-01 -8.27895761e-01 -1.58554420e-01 4.38762307e-01 -8.50603700e-01 4.04946893e-01 -1.01504207e+00 4.44487780e-01 -4.00086462e-01 5.07622818e-03 -1.26014352e+00 -5.83184287e-02 -6.55368507e-01 3.69079411e-01 1.06081772e+00 5.24993062e-01 -3.46367151e-01 1.23314393e+00 5.54089367e-01 3.33216459e-01 -4.88426328e-01 -4.56870764e-01 -6.66630685e-01 1.14441380e-01 3.00289869e-01 2.67705888e-01 8.01022887e-01 2.62308866e-01 1.70725554e-01 -2.75050610e-01 -9.57279354e-02 4.32540685e-01 -4.91847128e-01 4.77585912e-01 -1.03274405e+00 -6.64996207e-01 -5.20924747e-01 -1.01517344e+00 -5.19414961e-01 -4.61738080e-01 -6.83893561e-01 -3.95190388e-01 -1.35790086e+00 -3.87274802e-01 -1.46837592e-01 -6.32225201e-02 5.54062486e-01 1.14906598e-02 -1.57095164e-01 3.06533545e-01 -6.58919066e-02 1.81199506e-01 -2.79857725e-01 1.21662939e+00 -3.39572459e-01 -2.73583621e-01 3.99398923e-01 2.65121400e-01 3.94443691e-01 1.15645039e+00 -2.53860921e-01 -5.02308369e-01 -5.62353432e-02 -1.12261496e-01 5.70469737e-01 -8.22229907e-02 -1.35912383e+00 2.84823447e-01 -5.06720960e-01 9.34083045e-01 -3.22144061e-01 -1.92207202e-01 -1.31371737e+00 5.43127619e-02 9.37991381e-01 -5.49555004e-01 3.64720374e-01 1.27019256e-01 1.54936016e-01 -2.77779549e-01 -1.20071054e+00 1.06205225e+00 -4.54793632e-01 -6.95324481e-01 -4.67044562e-01 -7.73642778e-01 -3.52241158e-01 1.48325431e+00 -7.80796289e-01 1.88391954e-01 -1.03153132e-01 -8.72158349e-01 -6.83947280e-02 -2.40497701e-02 6.39256742e-03 5.40688872e-01 -9.91307676e-01 -4.79833543e-01 -1.69230506e-01 -5.33502996e-01 -6.41689718e-01 -2.76117831e-01 4.84850407e-01 -1.31876850e+00 4.80035782e-01 -1.35297644e+00 1.90144982e-02 -1.83552051e+00 6.23529136e-01 7.45721936e-01 -2.98026860e-01 -1.97207466e-01 2.55242854e-01 -1.06852639e+00 1.55969009e-01 4.43703502e-01 -3.71623009e-01 -9.16745067e-01 -3.71721983e-01 3.89570236e-01 6.73837602e-01 -1.80315778e-01 -1.81778148e-01 1.81368187e-01 6.79329932e-01 -6.74693063e-02 -6.98925674e-01 1.37529790e+00 7.26411283e-01 4.09447998e-02 3.42152059e-01 6.88064337e-01 -3.17474514e-01 -1.17237675e+00 3.32986683e-01 6.08720124e-01 -4.33474332e-01 -3.74050289e-01 -1.47060120e+00 -7.59116054e-01 9.28529799e-01 1.20483553e+00 2.03048885e-01 1.07409060e+00 -9.66092348e-01 -1.48279428e-01 1.36977386e+00 3.37280393e-01 -1.66174865e+00 3.29414874e-01 6.80184722e-01 1.09217811e+00 -1.28890073e+00 -6.75488785e-02 2.47850150e-01 -1.27106977e+00 2.13417053e+00 7.41170585e-01 -7.74733603e-01 6.47957623e-01 5.98416984e-01 7.96527043e-02 2.66831014e-02 -2.85976201e-01 8.17062929e-02 -1.54178828e-01 8.36357296e-01 6.36480689e-01 -2.23495603e-01 -8.48518670e-01 3.51576269e-01 1.45472912e-02 7.58823216e-01 7.82262444e-01 1.23637855e+00 -8.08053493e-01 -1.48349667e+00 -7.91468859e-01 6.37928605e-01 -3.95536214e-01 8.68162140e-02 -6.47042871e-01 9.32455897e-01 3.00933599e-01 7.39762723e-01 -7.38481507e-02 -3.21910560e-01 4.88781452e-01 2.11076185e-01 7.43284762e-01 -1.22561371e-02 -1.19299042e+00 -4.43718880e-01 -9.85352769e-02 2.46506527e-01 -2.88387567e-01 -6.46722853e-01 -1.63259614e+00 -5.92006892e-02 -7.85013586e-02 4.19052273e-01 1.09588480e+00 5.76220572e-01 -1.28889620e-01 2.99010605e-01 5.95883787e-01 -3.14620882e-01 -9.34084058e-01 -1.40207493e+00 -5.80981791e-01 4.82163608e-01 -1.80339441e-01 -3.47350329e-01 1.33140102e-01 4.78341341e-01]
[11.834882736206055, 2.7021188735961914]
58dcbb3f-5b0e-4f54-a24a-3f4f9135df34
centroid-distance-keypoint-detector-for
2210.01298
null
https://arxiv.org/abs/2210.01298v2
https://arxiv.org/pdf/2210.01298v2.pdf
Centroid Distance Keypoint Detector for Colored Point Clouds
Keypoint detection serves as the basis for many computer vision and robotics applications. Despite the fact that colored point clouds can be readily obtained, most existing keypoint detectors extract only geometry-salient keypoints, which can impede the overall performance of systems that intend to (or have the potential to) leverage color information. To promote advances in such systems, we propose an efficient multi-modal keypoint detector that can extract both geometry-salient and color-salient keypoints in colored point clouds. The proposed CEntroid Distance (CED) keypoint detector comprises an intuitive and effective saliency measure, the centroid distance, that can be used in both 3D space and color space, and a multi-modal non-maximum suppression algorithm that can select keypoints with high saliency in two or more modalities. The proposed saliency measure leverages directly the distribution of points in a local neighborhood and does not require normal estimation or eigenvalue decomposition. We evaluate the proposed method in terms of repeatability and computational efficiency (i.e. running time) against state-of-the-art keypoint detectors on both synthetic and real-world datasets. Results demonstrate that our proposed CED keypoint detector requires minimal computational time while attaining high repeatability. To showcase one of the potential applications of the proposed method, we further investigate the task of colored point cloud registration. Results suggest that our proposed CED detector outperforms state-of-the-art handcrafted and learning-based keypoint detectors in the evaluated scenes. The C++ implementation of the proposed method is made publicly available at https://github.com/UCR-Robotics/CED_Detector.
['Konstantinos Karydis', 'Amit K. Roy-Chowdhury', 'Xinyue Kan', 'Dimitrios Chatziparaschis', 'Hanzhe Teng']
2022-10-04
null
null
null
null
['point-cloud-registration']
['computer-vision']
[-9.12467837e-02 -3.36471051e-01 -9.82323885e-02 1.76463172e-01 -8.29005957e-01 -7.05040514e-01 8.03814232e-01 4.10036772e-01 -5.00860810e-01 1.49391919e-01 -2.07200870e-01 -6.58296198e-02 -1.63542241e-01 -5.03077328e-01 -6.81595504e-01 -5.97149253e-01 -1.35965601e-01 1.18116990e-01 6.48875117e-01 -2.12367341e-01 7.44201839e-01 1.00416374e+00 -1.86530209e+00 -3.79246950e-01 9.25364375e-01 1.09054470e+00 5.54777563e-01 5.13221383e-01 5.68339825e-02 6.27916306e-02 -2.89008498e-01 6.93088621e-02 6.92640781e-01 6.27686828e-02 -3.42656612e-01 -3.29924047e-01 8.73153687e-01 7.03017507e-03 -8.45597386e-02 1.21497464e+00 4.20682639e-01 1.54955685e-01 5.07157147e-01 -1.62882042e+00 -6.14983380e-01 -1.46202162e-01 -9.37793314e-01 3.11818033e-01 2.63244301e-01 3.09631806e-02 1.02136564e+00 -1.31198621e+00 6.26433313e-01 9.30194557e-01 4.86950576e-01 1.42330796e-01 -9.47573185e-01 -7.37245202e-01 1.61603112e-02 8.88370350e-03 -1.49221134e+00 -2.56466478e-01 1.13898218e+00 -3.35509092e-01 6.52522326e-01 4.38732684e-01 7.23944545e-01 3.69997710e-01 1.20603807e-01 8.46421599e-01 1.06961870e+00 -3.36586893e-01 3.55238765e-01 4.85070422e-03 -5.64904474e-02 8.19974005e-01 4.52160746e-01 1.87397167e-01 -6.77036345e-01 -3.23301047e-01 9.86352086e-01 3.03660542e-01 -1.52880251e-01 -9.92484927e-01 -1.71005511e+00 5.58872283e-01 8.78689647e-01 1.12075344e-01 -4.10562277e-01 3.27011168e-01 -9.80518907e-02 -1.19280703e-01 2.52060980e-01 6.38787448e-01 -1.46877587e-01 -1.25825197e-01 -1.02559912e+00 3.15741509e-01 2.64543533e-01 1.15320981e+00 1.05270278e+00 -1.54089138e-01 3.00982874e-02 5.43953240e-01 3.93385440e-01 1.05611753e+00 1.71918467e-01 -8.23789835e-01 1.64090782e-01 8.82469237e-01 3.52516443e-01 -1.35414517e+00 -4.86536384e-01 -2.92955011e-01 -5.18178284e-01 5.94213367e-01 1.66075066e-01 2.47273117e-01 -9.36179161e-01 1.30276895e+00 7.26076424e-01 4.80364352e-01 -1.83711380e-01 1.14048946e+00 6.46848321e-01 3.45402539e-01 -1.82189703e-01 1.04362674e-01 1.25445986e+00 -6.84719861e-01 -8.54678676e-02 -3.28966022e-01 1.96551561e-01 -1.06297529e+00 9.83517528e-01 1.01678833e-01 -9.28215981e-01 -4.15331185e-01 -1.18362641e+00 9.51910298e-03 -3.88197690e-01 3.42787474e-01 6.98041677e-01 2.90952533e-01 -1.05909836e+00 2.51088858e-01 -9.10043359e-01 -5.85286081e-01 2.10057527e-01 2.40695834e-01 -3.57944548e-01 1.06041789e-01 -6.51293814e-01 8.97100806e-01 2.02674687e-01 -4.55824053e-03 -5.77083170e-01 -1.01750779e+00 -6.48143709e-01 -1.14351340e-01 2.72071987e-01 -5.91331780e-01 9.36831117e-01 -7.08941281e-01 -1.09108067e+00 9.13054407e-01 -1.62231609e-01 -2.27723509e-01 4.58364308e-01 -5.17730176e-01 -1.77871376e-01 5.69678545e-01 3.55319113e-01 9.30093765e-01 1.15587330e+00 -1.46301782e+00 -1.13365042e+00 -2.98740685e-01 -3.44178975e-02 3.52299303e-01 -1.21090978e-01 -8.08199644e-02 -5.86993873e-01 -5.90239763e-01 5.06802082e-01 -1.21495402e+00 -3.24174128e-02 2.95094937e-01 -4.26522285e-01 -2.06019714e-01 1.03281856e+00 -2.91189663e-02 6.82900012e-01 -2.17711711e+00 -9.81208906e-02 2.68262178e-01 2.26351187e-01 1.19500123e-01 -6.74415901e-02 4.77535397e-01 1.01831099e-02 -1.73317894e-01 -9.88424569e-02 -8.17649886e-02 3.63849732e-03 -3.36020201e-01 -3.60180676e-01 9.48823154e-01 2.70826489e-01 7.68261790e-01 -1.13188744e+00 -4.58320111e-01 7.94249773e-01 4.45808768e-01 -2.90351033e-01 -1.73661187e-01 -4.24997434e-02 1.74869224e-01 -6.48297846e-01 1.01830757e+00 8.74680519e-01 -1.29938275e-01 -4.53453302e-01 -3.76771927e-01 -4.72809702e-01 -1.92942515e-01 -1.50103283e+00 1.82873499e+00 -1.25959292e-01 6.13091707e-01 -3.19269821e-02 -4.80135888e-01 1.06214774e+00 -1.72239617e-01 6.93114161e-01 -5.75083852e-01 -5.43044694e-02 3.42636973e-01 -3.50346148e-01 1.19675077e-01 1.07546484e+00 3.11391652e-01 1.10078733e-02 2.56646395e-01 -1.10100195e-01 -3.44939679e-01 1.58357583e-02 3.57383251e-01 9.47081864e-01 3.69545594e-02 3.35467011e-01 -4.54316020e-01 3.20739090e-01 3.38951617e-01 3.26835930e-01 6.08196020e-01 -3.37542355e-01 7.28348017e-01 -8.08279067e-02 -1.06979705e-01 -1.05424869e+00 -1.16361940e+00 -1.87017377e-02 9.59279954e-01 9.29397702e-01 -2.60870546e-01 -3.47990274e-01 -1.55759260e-01 3.37638348e-01 3.21647048e-01 -4.78276581e-01 4.47607413e-03 -3.87198538e-01 -1.95921600e-01 2.52222061e-01 3.22818816e-01 3.64840835e-01 -7.19440162e-01 -1.25483644e+00 -1.93290174e-01 1.45665020e-01 -1.02877712e+00 -4.29032475e-01 -6.37381673e-02 -7.66647637e-01 -1.24429715e+00 -8.24559450e-01 -7.54524112e-01 7.79978454e-01 1.11328864e+00 8.24894965e-01 1.63007870e-01 -3.13115478e-01 9.29342031e-01 -5.26288509e-01 -5.23959100e-01 2.53208607e-01 4.82802652e-02 2.22204104e-01 -2.94122994e-02 4.71827686e-01 -4.52235043e-01 -9.90720809e-01 3.32454860e-01 -7.94872701e-01 1.72448784e-01 8.13768804e-01 4.97882187e-01 7.82375336e-01 -3.41980129e-01 1.71930239e-01 -1.59895569e-01 3.70484173e-01 -2.11156666e-01 -1.00049436e+00 1.30777538e-01 -4.76208955e-01 1.44528851e-01 1.46342382e-01 -2.51751930e-01 -4.57345814e-01 4.71304834e-01 4.19379175e-01 -6.96600318e-01 -1.10598594e-01 3.35981876e-01 2.71324635e-01 -6.38911963e-01 5.84101617e-01 3.56177151e-01 -2.30995387e-01 -2.76240915e-01 7.27379322e-01 4.52901065e-01 6.96266234e-01 -5.86337030e-01 1.35893559e+00 8.76770973e-01 3.19760442e-01 -1.03782845e+00 -2.95397431e-01 -1.09368443e+00 -7.65572786e-01 -3.35703254e-01 5.82900167e-01 -9.89927888e-01 -7.46695161e-01 3.85130346e-01 -9.90788519e-01 1.88604176e-01 -5.43867648e-02 4.08471853e-01 -5.20047724e-01 3.16894978e-01 -3.20184305e-02 -8.47689927e-01 -4.94001359e-01 -9.63931382e-01 1.51506281e+00 5.96793056e-01 2.15115119e-03 -6.23916507e-01 3.11761051e-02 -1.17262006e-01 2.27096319e-01 4.86518532e-01 5.06954372e-01 -3.95075768e-01 -1.12610102e+00 -3.54472846e-01 -4.27135617e-01 -2.48977944e-01 8.24600011e-02 1.78602517e-01 -8.75871718e-01 -4.35524464e-01 -4.67543066e-01 -2.85711773e-02 7.53031313e-01 3.46441597e-01 4.54271525e-01 3.12611490e-01 -5.07871509e-01 5.52404940e-01 1.72027457e+00 -8.23170170e-02 3.02833796e-01 7.11932957e-01 8.37552130e-01 2.55792409e-01 1.05523002e+00 4.86151308e-01 4.66765493e-01 6.69110358e-01 6.84853911e-01 -2.73497611e-01 2.69402713e-02 -3.78983587e-01 1.00104496e-01 4.98470873e-01 -1.36711493e-01 5.34541309e-01 -1.19060087e+00 8.90160561e-01 -1.91196728e+00 -7.34162509e-01 -1.06064953e-01 2.27768922e+00 4.05815542e-01 -1.54504515e-02 1.25566348e-01 9.67617147e-03 7.57225335e-01 5.15158325e-02 -6.95926011e-01 1.11291669e-01 -6.44851997e-02 9.75583494e-02 9.77008641e-01 2.62599647e-01 -1.27110600e+00 1.02770293e+00 5.31004000e+00 5.66067100e-01 -1.43801606e+00 -1.55958861e-01 -4.07390520e-02 -8.28056112e-02 -1.54191822e-01 1.11133695e-01 -6.41856074e-01 1.81745380e-01 3.48128498e-01 -3.68811786e-01 1.92101687e-01 1.27306175e+00 9.69584063e-02 -3.84761751e-01 -9.87925410e-01 1.27787471e+00 3.13132927e-02 -1.46767330e+00 -1.86054066e-01 -5.84982038e-02 6.50007129e-01 3.93848956e-01 3.32842201e-01 -2.57563055e-01 1.16886020e-01 -5.47720730e-01 1.07273519e+00 3.64028662e-01 5.91792464e-01 -7.83850729e-01 2.92547971e-01 1.32876679e-01 -1.58506346e+00 7.63419457e-03 -4.13804680e-01 1.15922429e-01 -1.82860777e-01 4.73126322e-01 -8.63914907e-01 5.61308980e-01 8.20560694e-01 7.53566325e-01 -9.06375051e-01 1.53382516e+00 -1.92194164e-01 9.30339620e-02 -6.37161255e-01 -1.71725512e-01 3.31100941e-01 -1.48311272e-01 8.36696208e-01 1.07745910e+00 5.57528019e-01 -4.35580350e-02 3.60763162e-01 7.78047681e-01 1.87905893e-01 7.06446916e-02 -5.01941085e-01 2.01549649e-01 1.02654183e+00 1.56211233e+00 -1.17411184e+00 -1.08909860e-01 -4.44982827e-01 8.66158843e-01 1.56152189e-01 3.53919238e-01 -6.17493033e-01 -5.77905118e-01 9.46344554e-01 1.06432103e-01 3.94256026e-01 -6.30091310e-01 -4.62654978e-01 -1.02155387e+00 1.37124360e-01 -4.74775463e-01 8.88793319e-02 -1.10791099e+00 -9.90736783e-01 2.62936324e-01 7.76817724e-02 -1.66436255e+00 1.64404903e-02 -6.02835655e-01 -5.24523258e-01 7.88618147e-01 -1.75048637e+00 -1.56245244e+00 -6.05495453e-01 7.65443206e-01 2.33288646e-01 4.56795096e-02 4.77988362e-01 3.21490341e-03 -8.78800973e-02 3.21257472e-01 2.02001899e-01 -1.39110032e-02 7.20722735e-01 -1.33581352e+00 5.39283156e-01 1.26007915e+00 3.61350775e-01 8.82099748e-01 7.82562077e-01 -5.66805005e-01 -1.98372877e+00 -9.26200151e-01 2.49159306e-01 -5.05318463e-01 7.00548828e-01 -3.85389745e-01 -6.50537252e-01 2.63637185e-01 -2.08709344e-01 2.29917079e-01 1.89500049e-01 -1.07075617e-01 -2.63675034e-01 -1.59624472e-01 -9.98799086e-01 7.30902970e-01 7.65336633e-01 -5.73153615e-01 -6.37560666e-01 1.04049772e-01 6.86186016e-01 -5.52998424e-01 -5.97390354e-01 2.83242762e-01 5.32977402e-01 -8.71209741e-01 1.30156636e+00 1.27447322e-01 3.32611538e-02 -8.09120476e-01 -1.46761477e-01 -1.20026219e+00 -4.38554764e-01 -5.40777206e-01 -6.52505681e-02 9.73006725e-01 9.45121795e-02 -6.10165894e-01 8.39280367e-01 5.76170564e-01 -2.28918508e-01 -4.89283055e-01 -1.01291084e+00 -8.01342845e-01 -3.38572800e-01 -4.64502692e-01 5.66876769e-01 8.67575288e-01 -5.99339604e-03 -4.21337448e-02 -1.33892447e-01 6.48543417e-01 7.63604462e-01 5.63669920e-01 1.03109038e+00 -1.36830735e+00 3.44144642e-01 -4.51173455e-01 -9.11355376e-01 -9.25176442e-01 -2.05866516e-01 -6.51647925e-01 4.39991765e-02 -1.48023653e+00 -6.63467422e-02 -7.02844799e-01 -5.56800008e-01 3.73008579e-01 -2.69920945e-01 3.22856247e-01 5.31804979e-01 4.50665474e-01 -5.74757636e-01 4.74577278e-01 9.84725475e-01 -1.12382844e-01 -3.44184130e-01 -2.82982767e-01 -6.04816496e-01 6.98940754e-01 6.95448160e-01 -2.76779711e-01 -3.03147435e-01 -2.06191599e-01 2.90559560e-01 -3.48777950e-01 7.59185493e-01 -1.29066980e+00 6.42397821e-01 -4.34593976e-01 4.37727779e-01 -8.34912777e-01 4.11444932e-01 -9.55080688e-01 -4.88188304e-02 3.41067553e-01 1.37325719e-01 2.74492830e-01 3.11031729e-01 6.04291379e-01 -1.22929059e-01 1.31915867e-01 7.74091780e-01 5.36429994e-02 -1.22658122e+00 3.62427175e-01 2.08784297e-01 -2.01740474e-01 1.29805160e+00 -5.23259699e-01 -4.63040352e-01 -8.44778270e-02 1.12913847e-01 1.03509352e-01 1.05292821e+00 6.85106099e-01 1.01100326e+00 -1.35458636e+00 -4.74861205e-01 1.53142527e-01 5.74327767e-01 -5.88132329e-02 -4.49115857e-02 9.32120025e-01 -7.66164124e-01 4.39998478e-01 -4.17954504e-01 -9.50537562e-01 -1.32109177e+00 6.50589168e-01 6.36850968e-02 4.66359258e-01 -7.47672021e-01 7.40952611e-01 3.70262749e-03 -2.22596228e-01 -4.03443240e-02 -6.21436000e-01 1.05003819e-01 -8.47458541e-02 4.36950862e-01 4.52318221e-01 6.22471906e-02 -8.23175907e-01 -7.49300897e-01 1.00380063e+00 1.07448436e-01 -9.61356238e-02 1.28276300e+00 -1.28862560e-01 -6.99543133e-02 3.85953903e-01 9.30501878e-01 1.48953989e-01 -1.23398435e+00 -1.85964674e-01 1.29854143e-01 -8.17988753e-01 2.41336331e-01 -4.04792845e-01 -7.86294103e-01 5.92114270e-01 1.01243138e+00 4.76752929e-02 1.14217663e+00 1.00093922e-02 7.00567186e-01 3.40395808e-01 7.55062103e-01 -9.91791725e-01 -2.80146394e-02 2.56231099e-01 7.50737727e-01 -1.45285380e+00 2.51089185e-01 -4.14692879e-01 -4.94604945e-01 8.76340210e-01 4.35393363e-01 -5.08123636e-01 5.70138931e-01 -8.14698357e-03 2.34193474e-01 -4.13059890e-01 -2.97233522e-01 -4.71717417e-01 5.99016607e-01 7.48421013e-01 8.36901590e-02 7.59087577e-02 -3.64508145e-02 -6.08460233e-02 -2.35558316e-01 -1.69414863e-01 3.98773193e-01 1.23081088e+00 -7.57377505e-01 -6.35939956e-01 -7.00184405e-01 2.91282088e-01 -1.43987769e-02 -1.16483517e-01 -3.59791756e-01 7.74269819e-01 -1.22614391e-01 8.22867036e-01 5.14144897e-02 -3.74811888e-01 3.44323367e-01 -3.60471129e-01 2.96821952e-01 -3.93589079e-01 -2.27497101e-01 -4.00534384e-02 -5.30012667e-01 -8.62397909e-01 -6.99166000e-01 -5.84002316e-01 -1.34120798e+00 -1.91735953e-01 -3.85358483e-01 1.45375775e-02 9.36755955e-01 6.37880981e-01 6.10774994e-01 2.65557840e-02 4.69381034e-01 -1.31143963e+00 -1.36519894e-01 -6.72367096e-01 -5.71232080e-01 4.90549594e-01 6.16751850e-01 -1.00711715e+00 -1.08352676e-01 -5.23796305e-02]
[7.8264031410217285, -2.1289138793945312]
225b9973-1b80-4e45-a98b-894db95ec709
balanced-mse-for-imbalanced-visual-regression
2203.16427
null
https://arxiv.org/abs/2203.16427v1
https://arxiv.org/pdf/2203.16427v1.pdf
Balanced MSE for Imbalanced Visual Regression
Data imbalance exists ubiquitously in real-world visual regressions, e.g., age estimation and pose estimation, hurting the model's generalizability and fairness. Thus, imbalanced regression gains increasing research attention recently. Compared to imbalanced classification, imbalanced regression focuses on continuous labels, which can be boundless and high-dimensional and hence more challenging. In this work, we identify that the widely used Mean Square Error (MSE) loss function can be ineffective in imbalanced regression. We revisit MSE from a statistical view and propose a novel loss function, Balanced MSE, to accommodate the imbalanced training label distribution. We further design multiple implementations of Balanced MSE to tackle different real-world scenarios, particularly including the one that requires no prior knowledge about the training label distribution. Moreover, to the best of our knowledge, Balanced MSE is the first general solution to high-dimensional imbalanced regression. Extensive experiments on both synthetic and three real-world benchmarks demonstrate the effectiveness of Balanced MSE.
['Ziwei Liu', 'Cunjun Yu', 'Mingyuan Zhang', 'Jiawei Ren']
2022-03-30
null
http://openaccess.thecvf.com//content/CVPR2022/html/Ren_Balanced_MSE_for_Imbalanced_Visual_Regression_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Ren_Balanced_MSE_for_Imbalanced_Visual_Regression_CVPR_2022_paper.pdf
cvpr-2022-1
['age-estimation', 'imbalanced-classification', 'age-estimation']
['computer-vision', 'miscellaneous', 'miscellaneous']
[ 1.52396530e-01 -1.20372050e-01 -5.86261213e-01 -5.29568672e-01 -6.04761362e-01 -1.67330459e-01 5.88526092e-02 2.99440861e-01 -3.05987269e-01 1.02099741e+00 -2.53395379e-01 -3.05460483e-01 -6.12828229e-03 -5.07811010e-01 -5.65332651e-01 -6.74265981e-01 2.20403969e-01 3.43224287e-01 -3.94562602e-01 -9.87787545e-02 2.04141468e-01 1.85259163e-01 -1.60721922e+00 -1.37433812e-01 1.17288995e+00 1.09025049e+00 -4.92744625e-01 1.85396418e-01 2.01599121e-01 8.00177872e-01 -1.10511076e+00 -7.25121975e-01 4.42388594e-01 -2.63842911e-01 -3.79831672e-01 1.66644156e-02 7.75235772e-01 -2.77430087e-01 -4.38574493e-01 9.69952345e-01 7.88120806e-01 -7.32833222e-02 7.65174091e-01 -1.99077427e+00 -5.27677000e-01 2.25842074e-01 -1.30275011e+00 2.35245287e-01 3.62108531e-03 -1.47287667e-01 8.81459236e-01 -8.02278042e-01 2.95306981e-01 1.25010610e+00 6.86217189e-01 5.83258271e-01 -1.09921050e+00 -1.12624085e+00 5.28194547e-01 1.72545210e-01 -1.14891958e+00 -2.56802410e-01 7.70759344e-01 -4.14405316e-01 3.22077423e-01 5.56912243e-01 4.35380816e-01 1.15240586e+00 6.63254559e-02 9.35237586e-01 1.46015310e+00 -2.54302114e-01 1.41953811e-01 7.23758759e-03 2.63204843e-01 3.36290091e-01 8.30402255e-01 -2.15083063e-01 -5.25006890e-01 -1.60709545e-01 3.73100251e-01 1.80586249e-01 -2.01670930e-01 -6.34421170e-01 -1.08917403e+00 6.92005515e-01 2.78231949e-01 -2.57095516e-01 3.50847095e-02 1.03926815e-01 6.86522245e-01 4.54900324e-01 9.73196924e-01 2.94949353e-01 -3.96554530e-01 -5.97758144e-02 -7.82433867e-01 5.48067033e-01 7.07064629e-01 8.34473133e-01 4.52394485e-01 -1.11818947e-01 -2.88793236e-01 1.03886890e+00 8.10850337e-02 7.29701161e-01 2.57182062e-01 -7.52821803e-01 9.48884428e-01 6.35401845e-01 5.28078787e-02 -1.55549216e+00 -5.30558705e-01 -8.38532269e-01 -1.31203723e+00 8.91382024e-02 6.63206518e-01 -3.47470231e-02 -6.89429820e-01 1.78022397e+00 4.88020003e-01 2.55048070e-02 -2.71772444e-01 1.02628231e+00 5.76024771e-01 2.16851249e-01 7.26799443e-02 -5.19921660e-01 9.38182294e-01 -9.09527957e-01 -9.37302291e-01 -2.26332396e-01 5.70517004e-01 -6.05128407e-01 9.99310315e-01 5.24535894e-01 -9.57437515e-01 -1.89994588e-01 -1.12326241e+00 -4.67927940e-02 -3.09440523e-01 1.40457992e-02 6.28462732e-01 8.98065150e-01 -4.95715439e-01 1.89399302e-01 -4.61731255e-01 1.77185357e-01 7.69534409e-01 4.55482364e-01 -4.18584883e-01 -4.08665866e-01 -1.15523779e+00 8.27973187e-01 -1.36145070e-01 3.06900233e-01 -2.85250187e-01 -8.33030343e-01 -7.04160035e-01 -2.63219565e-01 5.02921641e-01 -6.95333123e-01 1.12658572e+00 -9.53615129e-01 -8.80428076e-01 1.06016672e+00 -1.13031395e-01 -1.09218746e-01 9.71034586e-01 -1.88113660e-01 -1.62743181e-01 -4.56590116e-01 1.35127604e-02 2.04153880e-01 9.40074623e-01 -1.28407633e+00 -4.40027982e-01 -9.75703657e-01 6.70460537e-02 3.10601711e-01 -4.83495146e-01 -4.44692597e-02 -2.35870242e-01 -8.11867237e-01 5.95692024e-02 -8.73534262e-01 -1.47756428e-01 4.93759327e-02 -5.98127842e-01 -8.01002905e-02 6.38661027e-01 -6.53460324e-01 1.67548382e+00 -2.01323056e+00 1.19437315e-01 1.49367884e-01 7.38235414e-01 1.99135467e-01 1.94541633e-01 -7.81325325e-02 -2.58795559e-01 3.36582929e-01 -2.46753410e-01 -4.34393734e-01 1.20652564e-01 3.75235975e-02 2.06594443e-04 8.61501455e-01 1.14291862e-01 6.76261783e-01 -8.13671649e-01 -5.16928852e-01 -1.91044435e-01 3.23674738e-01 -6.16442502e-01 1.92786291e-01 2.02904180e-01 3.93225372e-01 -3.04378897e-01 9.66646552e-01 1.15287018e+00 -2.62669981e-01 1.12344697e-01 -1.22764684e-01 2.31654510e-01 -1.13250412e-01 -1.15084183e+00 1.22403586e+00 -5.00858009e-01 6.85934484e-01 -8.37896913e-02 -1.25950742e+00 1.08835053e+00 -9.43071544e-02 5.69420457e-01 -1.13781071e+00 5.50046237e-03 4.53201026e-01 -3.78900059e-02 -3.15385997e-01 6.13248825e-01 -3.21402907e-01 -3.36485177e-01 2.18462661e-01 -5.73005199e-01 2.62469739e-01 1.41832054e-01 -3.26261893e-02 8.60804975e-01 -1.53354719e-01 3.78967732e-01 -2.43905000e-02 1.28977910e-01 -3.70951802e-01 9.23464358e-01 5.76974332e-01 -4.94395554e-01 7.42233276e-01 1.00182152e+00 -5.19043446e-01 -9.10201728e-01 -7.27449179e-01 -2.51528919e-01 1.00236821e+00 5.26398122e-01 -3.12709183e-01 -5.42644739e-01 -1.07994545e+00 4.20987129e-01 1.66303411e-01 -7.93925583e-01 -2.47336969e-01 -6.37283862e-01 -1.41890335e+00 4.81027246e-01 2.21430987e-01 2.33057663e-01 -4.19999570e-01 -4.82797146e-01 -9.80666503e-02 -3.95608842e-01 -9.89000559e-01 -3.43827546e-01 1.61208346e-01 -6.96161687e-01 -1.32311916e+00 -7.73276269e-01 -4.77313131e-01 8.31287086e-01 2.66327322e-01 1.53275585e+00 1.49565011e-01 -4.62605655e-01 -6.80663669e-03 -2.31355652e-01 -7.44117618e-01 -5.25712259e-02 3.37545037e-01 -1.40393563e-02 -1.93928846e-03 4.74317402e-01 -2.64516145e-01 -8.43172014e-01 7.43448138e-01 -7.44580150e-01 2.07512975e-02 3.67333114e-01 8.96595061e-01 5.72846889e-01 7.41944611e-02 9.89484012e-01 -1.29716408e+00 5.21890461e-01 -7.70715535e-01 -3.36417586e-01 1.82638913e-01 -7.30071247e-01 -3.34357798e-01 4.83292371e-01 -5.50978720e-01 -7.65161037e-01 -5.62303364e-01 1.07254468e-01 -3.31252366e-01 1.70362577e-01 2.14457333e-01 -3.18830490e-01 1.87704399e-01 5.12745380e-01 -3.77248466e-01 5.87835815e-03 -3.84307355e-01 -8.60658139e-02 9.48606491e-01 9.74694788e-02 -4.89998668e-01 5.52525222e-01 3.60949695e-01 2.89159775e-01 -3.26801419e-01 -1.16672492e+00 -3.92328054e-01 -2.88340360e-01 -2.08717361e-01 1.59929946e-01 -1.02629554e+00 -9.22637045e-01 8.46872866e-01 -8.58717442e-01 -2.20999375e-01 -2.01587453e-01 3.16890895e-01 -5.33334851e-01 1.29841954e-01 -2.27477774e-01 -8.31599057e-01 -1.14749953e-01 -1.02942896e+00 1.09615874e+00 1.91201389e-01 -2.50432082e-02 -8.34206343e-01 -2.75218021e-03 7.27500439e-01 3.19553494e-01 6.80454969e-01 8.91550243e-01 -4.13257062e-01 -1.93455771e-01 -3.09180617e-01 -5.31141162e-01 4.29158866e-01 1.28859282e-01 -6.67825853e-03 -1.00696480e+00 -5.05912125e-01 -6.12715259e-02 -5.00251114e-01 9.06225383e-01 2.52596468e-01 1.64376116e+00 -1.62510201e-01 -1.96213335e-01 6.44400835e-01 1.39403117e+00 -6.07117899e-02 6.66937709e-01 3.41533363e-01 1.01321781e+00 7.83012450e-01 1.17506361e+00 6.67393386e-01 6.65566444e-01 9.50593889e-01 7.27669477e-01 -5.33562779e-01 3.74739394e-02 -4.53326553e-02 5.15954830e-02 9.94675338e-01 -6.64091185e-02 -3.84149075e-01 -8.36294353e-01 2.59516031e-01 -1.97778618e+00 -6.95210755e-01 -1.25466317e-01 2.50862598e+00 1.01592934e+00 2.19539013e-02 2.17058584e-01 5.60646534e-01 8.78228486e-01 3.05051327e-01 -9.19723153e-01 -4.22020614e-01 -2.29414150e-01 -2.25848645e-01 6.95515513e-01 5.83068021e-02 -1.25041902e+00 4.86111313e-01 5.93879223e+00 1.00933123e+00 -1.08109891e+00 1.38396546e-01 1.26033711e+00 -5.03455758e-01 -2.95417130e-01 -4.41176683e-01 -8.21202040e-01 8.24523926e-01 4.54207003e-01 -2.09958971e-01 1.21922255e-01 7.06062555e-01 -2.71280315e-02 2.26859320e-02 -1.10240591e+00 1.42059648e+00 3.22417289e-01 -6.29115224e-01 -3.21427315e-01 2.21793726e-01 8.58511627e-01 -3.39081138e-01 3.74705523e-01 5.11981726e-01 -1.63717002e-01 -1.35904944e+00 4.65371430e-01 3.11416626e-01 1.10365975e+00 -1.02168608e+00 8.41497898e-01 3.35375875e-01 -6.92186713e-01 -1.34140164e-01 -4.71763283e-01 -1.81864873e-01 -3.17907810e-01 1.08942878e+00 -3.58886421e-01 4.27123487e-01 6.75366700e-01 1.01277590e+00 -6.54600680e-01 1.03085899e+00 1.91270441e-01 4.45587426e-01 -6.50389642e-02 1.02953464e-01 -2.58138031e-01 -2.47470066e-02 2.47799620e-01 1.06210697e+00 1.63815320e-01 -4.07155365e-01 1.89386070e-01 1.89709485e-01 -4.16529268e-01 4.71806705e-01 -5.93608737e-01 4.56713408e-01 4.04740512e-01 1.10229802e+00 -3.89697164e-01 5.08178733e-02 -4.40365165e-01 6.65560603e-01 4.92456079e-01 3.27097625e-01 -1.02252984e+00 -2.05402493e-01 8.93130541e-01 1.69969946e-01 -3.52615677e-02 2.35574782e-01 -6.29649520e-01 -1.35891318e+00 3.95632148e-01 -1.37463379e+00 4.37436730e-01 -1.83090419e-01 -1.60848117e+00 3.05988222e-01 -1.30889565e-01 -1.39931762e+00 -4.30290215e-03 -4.80945349e-01 1.10153101e-01 6.52861238e-01 -2.00637674e+00 -9.13243830e-01 -6.62781835e-01 3.13755900e-01 2.92813182e-01 4.12008399e-03 4.88127381e-01 8.62819314e-01 -1.10313845e+00 1.02704644e+00 1.22847229e-01 -1.08138658e-01 1.21478057e+00 -1.35148144e+00 1.42629072e-01 3.43764961e-01 -3.19329023e-01 2.59014696e-01 8.02812636e-01 -3.64517719e-01 -1.17536139e+00 -1.10724330e+00 7.25210845e-01 -4.10837352e-01 3.02208662e-01 -4.30837721e-01 -8.65192473e-01 3.93829614e-01 -3.86939585e-01 3.84108543e-01 9.12822843e-01 2.02284887e-01 -7.24307716e-01 -5.57688534e-01 -1.28656673e+00 4.38962579e-01 1.20623457e+00 -9.28973854e-02 6.19232841e-02 3.19436610e-01 4.19522047e-01 -7.33790994e-01 -9.47523355e-01 7.79931366e-01 8.01901460e-01 -1.04864514e+00 1.07032979e+00 -7.08222032e-01 6.00081384e-01 -1.30453467e-01 -2.83055782e-01 -1.35962474e+00 4.39728610e-02 -2.57040709e-01 -4.54378664e-01 1.18692446e+00 1.80135086e-01 -7.30974019e-01 9.06863928e-01 5.55358231e-01 2.82141238e-01 -1.33376122e+00 -9.35905576e-01 -8.03660154e-01 3.07661057e-01 -2.17995450e-01 7.11366475e-01 1.13755119e+00 -2.41854653e-01 5.80417179e-02 -9.35680032e-01 -1.91520482e-01 8.58990252e-01 1.10278241e-01 9.91711497e-01 -1.50841343e+00 -1.29112035e-01 -3.39177608e-01 -7.05855668e-01 -9.14822876e-01 4.03528005e-01 -6.39065206e-01 -1.86304882e-01 -1.09621382e+00 4.78860110e-01 -9.88913417e-01 -4.65850949e-01 3.17981094e-01 -6.54724181e-01 7.09306777e-01 2.43077710e-01 1.85532674e-01 -7.86038280e-01 5.58893263e-01 1.39352739e+00 -2.82992512e-01 1.60463318e-01 2.62282878e-01 -1.13815033e+00 7.23927975e-01 1.08455324e+00 -8.08893442e-01 -4.55933720e-01 -2.66773134e-01 7.67609179e-01 -1.92003012e-01 1.32607609e-01 -5.30057132e-01 -3.01472824e-02 -4.02285039e-01 3.17524791e-01 -3.54124010e-01 1.19052373e-01 -8.83654714e-01 -2.43280932e-01 2.95662016e-01 -2.69916266e-01 1.36472598e-01 -1.26063854e-01 5.98728478e-01 -4.20925260e-01 9.41589922e-02 8.17217827e-01 2.47718230e-01 -2.00162664e-01 5.08442760e-01 3.62839401e-01 6.79584563e-01 1.22736943e+00 -2.65337139e-01 -7.04461336e-01 -3.46369207e-01 -2.41715744e-01 5.07932663e-01 6.15669668e-01 5.40753663e-01 4.85542983e-01 -1.34821939e+00 -9.54230785e-01 1.19362168e-01 3.52737457e-01 1.20276250e-01 2.64361799e-01 1.12622118e+00 -3.46540421e-01 -1.95871964e-02 -1.18279412e-01 -5.31594694e-01 -1.48862100e+00 6.54617071e-01 2.75424510e-01 -6.02723181e-01 -2.16067433e-01 5.29094458e-01 3.33620369e-01 -5.25539696e-01 5.62946320e-01 -4.13137078e-02 -1.34335726e-01 4.70599741e-01 4.67415571e-01 8.71486962e-01 2.19891489e-01 -6.01082146e-01 -4.89348382e-01 6.11902118e-01 -1.56696960e-01 4.07789618e-01 1.07166278e+00 -3.27436656e-01 -1.85546339e-01 7.21966922e-01 1.24274194e+00 7.84323290e-02 -1.12408662e+00 -1.54324144e-01 -7.20114484e-02 -9.75635350e-01 -1.78052217e-01 -6.90251291e-01 -1.41270328e+00 8.45404625e-01 4.24402267e-01 2.57008553e-01 1.33059192e+00 -3.78202796e-01 6.67939007e-01 1.11786813e-01 3.67699385e-01 -1.19032001e+00 1.25568435e-01 9.79274064e-02 7.34159410e-01 -1.68828857e+00 3.82459790e-01 -6.22470140e-01 -6.72360718e-01 6.44243658e-01 1.01788211e+00 5.46317659e-02 6.15650833e-01 2.39400342e-01 3.19462299e-01 1.09141193e-01 -7.92331994e-01 2.10330293e-01 1.77018240e-01 4.96969163e-01 4.41595644e-01 2.57884771e-01 -6.08553350e-01 4.17486757e-01 -6.77540526e-02 -8.39418024e-02 4.63812441e-01 6.94129169e-01 1.24550432e-01 -1.10860944e+00 -3.61487836e-01 7.96532094e-01 -6.39964819e-01 2.03102335e-01 -1.60162508e-01 7.38687873e-01 3.67316812e-01 8.47842693e-01 2.89525360e-01 -1.55010223e-01 5.71959615e-01 -2.13122576e-01 5.35667539e-01 -3.70698720e-01 -5.55476725e-01 -3.70760590e-01 -2.09312085e-02 -5.28263748e-01 -5.10561228e-01 -4.64873016e-01 -6.97695673e-01 -6.52897179e-01 -4.53899741e-01 -2.22106725e-01 9.02611196e-01 6.11623526e-01 2.58587271e-01 6.00667894e-01 1.08624876e+00 -3.53101581e-01 -8.28495085e-01 -7.41135240e-01 -7.66811371e-01 7.12220848e-01 6.12375915e-01 -8.73511195e-01 -5.30541480e-01 -3.56132925e-01]
[9.052103996276855, 3.9695677757263184]
9a45f69c-0bd8-4ef4-9f23-1e5c244fa501
correlation-clustering-algorithm-for-dynamic
2301.00384
null
https://arxiv.org/abs/2301.00384v1
https://arxiv.org/pdf/2301.00384v1.pdf
Correlation Clustering Algorithm for Dynamic Complete Signed Graphs: An Index-based Approach
In this paper, we reduce the complexity of approximating the correlation clustering problem from $O(m\times\left( 2+ \alpha (G) \right)+n)$ to $O(m+n)$ for any given value of $\varepsilon$ for a complete signed graph with $n$ vertices and $m$ positive edges where $\alpha(G)$ is the arboricity of the graph. Our approach gives the same output as the original algorithm and makes it possible to implement the algorithm in a full dynamic setting where edge sign flipping and vertex addition/removal are allowed. Constructing this index costs $O(m)$ memory and $O(m\times\alpha(G))$ time. We also studied the structural properties of the non-agreement measure used in the approximation algorithm. The theoretical results are accompanied by a full set of experiments concerning seven real-world graphs. These results shows superiority of our index-based algorithm to the non-index one by a decrease of %34 in time on average.
['Ali Shakiba']
2023-01-01
null
null
null
null
['graph-clustering']
['graphs']
[ 2.37501442e-01 3.10895920e-01 -3.52000110e-02 -1.18774809e-01 -6.17951572e-01 -8.05659175e-01 2.78151780e-02 4.81733501e-01 -7.22909808e-01 6.72571301e-01 -8.08393717e-01 -5.27591109e-01 -4.24590915e-01 -1.25309169e+00 -6.68787837e-01 -6.84321582e-01 -1.03686285e+00 9.49775457e-01 5.33192277e-01 -1.51424855e-01 4.81118560e-01 4.83339995e-01 -1.50148356e+00 -4.26968157e-01 5.44160903e-01 1.04087961e+00 -1.20946459e-01 1.06289804e+00 -7.47903809e-02 1.87319860e-01 -3.08760613e-01 -7.52750218e-01 7.71481991e-01 -4.76242036e-01 -1.04547739e+00 2.10451454e-01 1.92943871e-01 4.73577827e-01 -2.66702563e-01 1.42700243e+00 2.93562382e-01 9.83583704e-02 3.87605786e-01 -1.53547704e+00 6.88258931e-02 9.93098140e-01 -1.24608600e+00 1.55646518e-01 3.19941908e-01 -2.37330675e-01 1.12310374e+00 -3.25303644e-01 7.96302557e-01 9.53420103e-01 5.80469787e-01 -9.69514549e-02 -1.24222875e+00 -7.15780973e-01 -1.21791802e-01 2.99377851e-02 -1.88544822e+00 -6.65684640e-02 6.09179080e-01 -4.92289178e-02 8.48314881e-01 6.37282908e-01 5.91351390e-01 -3.98337901e-01 -2.25201815e-01 1.31548913e-02 1.22412455e+00 -7.21235633e-01 2.15715557e-01 -1.78077698e-01 3.24088365e-01 1.10875130e+00 1.01984656e+00 1.97662767e-02 1.48977693e-02 -3.66851956e-01 4.08216119e-01 -2.04270318e-01 -1.02346927e-01 -4.06753093e-01 -7.94053257e-01 7.93457925e-01 4.07125324e-01 4.17601019e-01 2.69731954e-02 8.08795512e-01 2.09203795e-01 6.57275081e-01 1.05092093e-01 1.65009946e-01 -4.26037014e-01 -1.07280210e-01 -1.04747486e+00 -1.19814761e-01 1.03086090e+00 1.27343988e+00 1.10597599e+00 -2.28474304e-01 6.36620581e-01 5.51730692e-01 -5.90813830e-02 8.02899063e-01 -4.46565956e-01 -1.29199147e+00 6.48199201e-01 8.36095333e-01 6.06861375e-02 -1.35300529e+00 -6.27019346e-01 -3.96843642e-01 -1.10282147e+00 2.70898640e-01 6.92025661e-01 -1.67497098e-01 -4.73870903e-01 1.94082963e+00 1.35618374e-01 -3.33967179e-01 -5.35804152e-01 2.48987034e-01 -6.29970804e-02 4.51542228e-01 -6.51639640e-01 -8.60379875e-01 1.38013196e+00 -7.57443786e-01 -2.58011699e-01 5.29097579e-02 1.05112410e+00 -8.77453864e-01 6.16948843e-01 5.46395123e-01 -1.79608834e+00 -6.23270571e-02 -1.06881821e+00 4.17421550e-01 -4.73504126e-01 -2.99757749e-01 7.49490738e-01 1.09990025e+00 -1.58344674e+00 5.61029136e-01 -4.87030774e-01 -1.46968156e-01 3.48413587e-02 9.04338658e-01 -3.40781152e-01 -3.00034881e-01 -5.45760274e-01 4.14560467e-01 2.23200783e-01 -1.18072078e-01 3.50522920e-02 -6.09790683e-02 -6.75043106e-01 1.52391410e-02 6.72340155e-01 -4.33115214e-01 5.09124875e-01 -6.92110479e-01 -6.21557176e-01 9.78570521e-01 -2.11175993e-01 -3.19723845e-01 2.40927950e-01 7.99499154e-01 -1.02051295e-01 4.72682446e-01 4.37684469e-02 3.04557115e-01 1.45709589e-01 -1.24530137e+00 -5.74027777e-01 -6.84690833e-01 1.51486576e-01 -1.50179684e-01 5.00774197e-03 1.69190601e-01 -7.31132329e-01 -2.21123487e-01 6.02670193e-01 -1.41314757e+00 -5.32029331e-01 -1.45595148e-01 -1.64546460e-01 8.37203860e-02 2.65482038e-01 -1.71213925e-01 1.57959294e+00 -1.92851460e+00 7.64554888e-02 1.31845570e+00 5.17812133e-01 3.91579326e-03 -4.11476158e-02 6.99873567e-01 -1.93070754e-01 3.46724898e-01 -5.28084278e-01 -3.01682986e-02 -8.14700723e-02 1.89560115e-01 4.89057153e-01 7.97681212e-01 -9.75845039e-01 2.72538990e-01 -6.66953087e-01 -5.68361938e-01 -2.86254823e-01 -1.71780456e-02 -5.69182634e-01 -4.54677075e-01 1.68434799e-01 -4.97129619e-01 6.98326379e-02 5.23333490e-01 1.29778361e+00 -5.01949787e-01 8.48954856e-01 1.81167841e-01 -4.83171269e-02 -2.74979532e-01 -1.81902266e+00 1.21263814e+00 -3.93240601e-01 1.73852727e-01 6.21591628e-01 -8.78990054e-01 7.13763297e-01 -1.91431660e-02 7.49239266e-01 -6.29424512e-01 2.84685135e-01 5.95927000e-01 1.85831428e-01 1.32758722e-01 3.53656650e-01 -8.13206378e-03 -2.43379042e-01 1.11596835e+00 -4.78132337e-01 7.44005293e-02 7.39105105e-01 6.75968647e-01 1.68074059e+00 -8.13167751e-01 1.13985300e-01 -5.13105631e-01 6.06562912e-01 -1.30491555e-01 3.61614406e-01 5.86547554e-01 -2.63403058e-01 -1.95820574e-02 9.25031245e-01 -1.73028171e-01 -9.59826410e-01 -6.29233956e-01 2.83148557e-01 7.66193390e-01 4.09448981e-01 -6.21185839e-01 -1.07771969e+00 -5.03773987e-01 -1.88681945e-01 1.81603879e-01 -7.75451660e-01 1.72064483e-01 -7.70989180e-01 -9.92718220e-01 3.68357003e-01 5.72481692e-01 4.19725865e-01 -6.35542095e-01 -4.96633530e-01 -1.95198879e-02 -1.38752870e-02 -9.25790191e-01 -7.09034681e-01 2.34786808e-01 -1.07448125e+00 -1.60225844e+00 -1.01267889e-01 -9.85559404e-01 1.43770874e+00 5.65917969e-01 9.64900255e-01 5.77492356e-01 -3.15832734e-01 1.88378587e-01 -2.21575946e-01 2.33362451e-01 -5.01270369e-02 -3.47446017e-02 -1.33441612e-01 -6.01885438e-01 2.37043098e-01 -7.13228643e-01 -7.14528561e-01 3.33214462e-01 -9.48903143e-01 -4.98374522e-01 4.26715255e-01 7.45476484e-01 6.62322640e-01 5.02861679e-01 -1.02322184e-01 -1.21259093e+00 3.87554854e-01 -2.39557251e-01 -1.18967056e+00 6.02732189e-02 -1.18922317e+00 2.10698143e-01 7.47748554e-01 6.59851432e-02 -2.92356253e-01 1.94455802e-01 1.50920361e-01 1.21813335e-01 7.16380894e-01 3.54937911e-01 1.13120250e-01 -5.10215819e-01 2.28435323e-01 -8.47449005e-02 1.18370585e-01 -1.56061977e-01 3.61050993e-01 7.12894127e-02 3.44937533e-01 -2.99677849e-01 8.54857147e-01 6.92552984e-01 9.35373366e-01 -3.90328646e-01 1.20863020e-01 -3.68920654e-01 -4.10051137e-01 5.22708446e-02 1.27992168e-01 -3.45460713e-01 -1.50041950e+00 1.35174423e-01 -7.87722647e-01 -8.98699537e-02 -7.30876401e-02 1.64430633e-01 -5.30873358e-01 8.15198004e-01 -5.50896347e-01 -1.05799139e+00 -3.43001664e-01 -9.22775209e-01 4.45857316e-01 -3.17527801e-01 -2.67987177e-02 -8.18481505e-01 -1.67949498e-02 3.55908662e-01 2.92356968e-01 2.73443967e-01 1.06135154e+00 -3.82472336e-01 -8.00571322e-01 -5.60387611e-01 -5.52801192e-01 -7.27519095e-02 -2.55330771e-01 -1.62536912e-02 -6.06987737e-02 -6.01336420e-01 -2.99287349e-01 1.40941188e-01 6.31466925e-01 1.70352697e-01 1.01008129e+00 -4.46349323e-01 -5.56140542e-01 4.48940128e-01 1.94005430e+00 3.36967081e-01 6.03418648e-01 2.34568827e-02 2.57784128e-01 2.45803967e-01 6.60316229e-01 5.26084304e-01 2.59907782e-01 4.80655074e-01 5.73606849e-01 -1.03365444e-01 6.13826513e-02 2.74026722e-01 6.35559391e-03 1.10355580e+00 -3.38180393e-01 -3.69822800e-01 -7.96035111e-01 6.93840206e-01 -1.54261291e+00 -9.19171333e-01 -4.30801541e-01 2.59165239e+00 5.33291578e-01 3.57527018e-01 3.31857622e-01 6.81728303e-01 9.94407356e-01 -2.70510204e-02 1.20575473e-01 -8.81686926e-01 -3.32048498e-02 8.30236554e-01 1.16419661e+00 9.01225865e-01 -3.75977933e-01 4.81190711e-01 6.26372766e+00 9.26072240e-01 -4.47635710e-01 6.49522170e-02 6.51136994e-01 -1.83720872e-01 -2.87148952e-01 4.11969543e-01 -2.52849519e-01 5.85492969e-01 1.10753596e+00 -1.75769612e-01 6.63117766e-01 7.36213326e-01 -1.76227540e-01 -6.34904265e-01 -7.90375292e-01 1.16653287e+00 -2.15907861e-02 -1.00378001e+00 -3.86421055e-01 5.27180493e-01 7.78402448e-01 -2.93782681e-01 -8.64347816e-02 -2.82259941e-01 5.44280946e-01 -7.31254220e-01 -1.50784664e-02 -2.03058288e-01 1.02256227e+00 -1.22294939e+00 8.75954151e-01 2.21188799e-01 -1.75740421e+00 -6.31382018e-02 -1.87950104e-01 -4.61216755e-02 7.30815008e-02 6.01361692e-01 -5.29953361e-01 5.08441806e-01 7.21705556e-01 -3.52962255e-01 -2.48590067e-01 7.81835556e-01 4.17919815e-01 4.10244837e-02 -8.82929742e-01 -1.71249598e-01 2.72651523e-01 -6.24600887e-01 3.34208459e-01 1.03213429e+00 5.02344310e-01 3.92927259e-01 -3.65911350e-02 4.26084064e-02 -4.03614551e-01 3.15895200e-01 -5.71384788e-01 1.84881359e-01 7.74967253e-01 1.07394552e+00 -1.47062624e+00 -3.56273800e-01 -2.16912672e-01 8.02911818e-01 2.27304861e-01 -1.08925000e-01 -7.61241734e-01 -9.25655723e-01 2.87045181e-01 5.76975524e-01 6.69701815e-01 -4.73155886e-01 -2.40884006e-01 -5.00602245e-01 1.62377343e-01 -6.53854370e-01 7.10956991e-01 -2.60907024e-01 -6.38593972e-01 6.17895305e-01 3.48845012e-02 -8.31879139e-01 -6.66889101e-02 -5.26145697e-01 -1.39030531e-01 5.29034853e-01 -1.01773882e+00 -4.06347692e-01 -1.40270054e-01 6.00097239e-01 -4.37347829e-01 3.71135473e-01 7.89968431e-01 4.21976596e-01 -1.47680894e-01 9.20263588e-01 2.58852214e-01 -7.48202875e-02 3.37752849e-01 -1.14366329e+00 8.49610567e-02 1.04105067e+00 -1.83372155e-01 6.69725657e-01 7.75380611e-01 -3.30473542e-01 -1.54508460e+00 -5.53510249e-01 1.22863102e+00 1.72148854e-01 4.92729336e-01 -2.53569067e-01 -2.59550959e-01 5.29778242e-01 2.29886472e-01 1.39018491e-01 8.33793104e-01 -6.81696758e-02 -4.26240951e-01 -5.45665979e-01 -1.68628788e+00 6.61677957e-01 1.51642430e+00 -1.50520489e-01 4.20211107e-01 2.76266873e-01 2.76294231e-01 -1.23385295e-01 -1.13373888e+00 2.54211068e-01 2.75683403e-01 -1.10113049e+00 6.82747066e-01 1.75060689e-01 -3.05361420e-01 -5.25837839e-01 -2.39156783e-01 -7.01550782e-01 -3.14229131e-01 -1.09632003e+00 -1.48009369e-02 6.24070764e-01 7.32298970e-01 -8.54847670e-01 9.06124234e-01 4.20044303e-01 2.89239645e-01 -8.52814615e-01 -1.17366982e+00 -6.53413177e-01 2.90253479e-02 -3.64831448e-01 5.60367763e-01 7.69771218e-01 4.46776479e-01 1.06516413e-01 -1.39460620e-02 6.43738806e-02 9.26740825e-01 4.25740838e-01 7.67644763e-01 -1.21058524e+00 -6.52371764e-01 -5.76979458e-01 -5.55957556e-01 -7.70513892e-01 -3.95339906e-01 -9.20089960e-01 -4.88707095e-01 -1.20226800e+00 5.28093219e-01 -8.71365249e-01 -1.28845200e-01 4.38874125e-01 2.47957617e-01 5.77620327e-01 7.98621699e-02 -1.13991499e-01 -7.59187698e-01 -2.53920943e-01 8.92165780e-01 1.38751745e-01 6.81624189e-02 -4.92084073e-03 -8.19840729e-01 5.81536055e-01 7.02163458e-01 -6.67542636e-01 -4.70086008e-01 -1.57493591e-01 8.38136137e-01 6.60247505e-01 -2.38955423e-01 -8.03222537e-01 3.39225441e-01 1.69966578e-01 -2.92643279e-01 -6.98543191e-01 3.63158703e-01 -9.60560501e-01 5.41926682e-01 9.33655679e-01 3.05174571e-02 7.99443543e-01 9.07117408e-03 5.56591153e-01 -9.30850208e-02 -3.59751433e-01 9.05585468e-01 -1.44237623e-01 -3.24402712e-02 4.28175986e-01 1.77507728e-01 -3.74910869e-02 1.35217702e+00 -6.35751426e-01 -5.87806880e-01 -4.36149597e-01 -7.76466250e-01 1.88101217e-01 9.36710417e-01 -5.34168959e-01 3.18608820e-01 -9.15050507e-01 -2.94205904e-01 3.43149342e-02 -7.53862411e-02 -1.60656154e-01 3.07330340e-01 1.14534974e+00 -1.08738613e+00 3.01372290e-01 3.37741487e-02 -3.19573134e-01 -1.80577135e+00 9.37594056e-01 3.18451524e-02 -6.25466704e-01 -7.69119412e-02 9.37183380e-01 -5.42506933e-01 -3.96789312e-02 2.92937439e-02 2.34136195e-03 6.35007024e-01 -1.81614622e-01 8.39219168e-02 9.64850307e-01 3.44734602e-02 -5.91591358e-01 -4.45217848e-01 7.86798596e-01 -1.26226291e-01 -5.42260528e-01 1.15910745e+00 -2.32786149e-01 -8.42630029e-01 -1.52099073e-01 1.41003692e+00 4.54177171e-01 -5.22778094e-01 -4.04056311e-02 -1.72965955e-02 -4.40414965e-01 -3.53788048e-01 -3.15659195e-01 -1.50968575e+00 4.98437196e-01 4.37984109e-01 5.84783196e-01 1.46033287e+00 -1.19285993e-02 7.48962879e-01 3.04397702e-01 1.06758475e+00 -1.38359284e+00 -2.37911314e-01 1.14350803e-01 1.49733737e-01 -6.94013774e-01 3.33957762e-01 -1.03492999e+00 -1.95543736e-01 7.89254785e-01 2.30531856e-01 -4.12298530e-01 1.06720853e+00 4.86954272e-01 -5.10144532e-01 -3.24877888e-01 -6.77339017e-01 -1.96294010e-01 -5.20771801e-01 2.76729763e-01 2.44849294e-01 2.50375777e-01 -1.14881492e+00 2.22749859e-01 -2.65705287e-01 -3.23254853e-01 8.18106234e-01 1.18260348e+00 -6.74389839e-01 -1.68526471e+00 -2.95082957e-01 4.38701808e-01 -5.33007979e-01 -9.09808949e-02 -4.32275742e-01 1.01688790e+00 1.36569679e-01 1.13892853e+00 3.25963609e-02 -3.98563832e-01 2.04176381e-02 -4.59740832e-02 8.37575495e-01 -1.40675426e-01 -6.54174328e-01 4.62300293e-02 2.93961823e-01 -5.32728612e-01 -4.33864236e-01 -5.69265068e-01 -1.64153755e+00 -1.15218139e+00 -4.84856874e-01 5.95596790e-01 6.68150961e-01 2.43959084e-01 4.89580482e-01 9.18602198e-02 9.40306962e-01 -1.00701720e-01 -2.29381219e-01 -6.30149722e-01 -9.88768160e-01 2.08214208e-01 -4.26191986e-01 -4.64902073e-01 -6.50626659e-01 -3.46740723e-01]
[6.786737442016602, 5.0807271003723145]
1a10a86b-613c-4dce-a274-93f7e3216078
training-a-deep-q-learning-agent-inside-a
2301.01913
null
https://arxiv.org/abs/2301.01913v1
https://arxiv.org/pdf/2301.01913v1.pdf
Training a Deep Q-Learning Agent Inside a Generic Constraint Programming Solver
Constraint programming is known for being an efficient approach for solving combinatorial problems. Important design choices in a solver are the branching heuristics, which are designed to lead the search to the best solutions in a minimum amount of time. However, developing these heuristics is a time-consuming process that requires problem-specific expertise. This observation has motivated many efforts to use machine learning to automatically learn efficient heuristics without expert intervention. To the best of our knowledge, it is still an open research question. Although several generic variable-selection heuristics are available in the literature, the options for a generic value-selection heuristic are more scarce. In this paper, we propose to tackle this issue by introducing a generic learning procedure that can be used to obtain a value-selection heuristic inside a constraint programming solver. This has been achieved thanks to the combination of a deep Q-learning algorithm, a tailored reward signal, and a heterogeneous graph neural network architecture. Experiments on graph coloring, maximum independent set, and maximum cut problems show that our framework is able to find better solutions close to optimality without requiring a large amounts of backtracks while being generic.
['Louis-Martin Rousseau', 'Quentin Cappart', 'Louis Gauthier', 'Pierre Tessier', 'Tristan François', 'Tom Marty']
2023-01-05
null
null
null
null
['variable-selection']
['methodology']
[ 0.19256894 0.25010923 -0.38713413 -0.13636525 -0.6208722 -0.64627755 0.11122739 0.45482364 -0.18934305 0.87879485 -0.4961141 -0.38190413 -0.45669284 -0.9181313 -0.62264544 -0.59778607 -0.1708463 0.9305864 0.22654107 -0.1953625 0.49476883 0.6688637 -1.4370329 0.01184747 1.091217 1.0171622 0.4165934 0.39341512 -0.3938735 0.36953896 -0.4172183 -0.40613145 0.50934 -0.68015796 -1.0507978 0.26195127 -0.05671796 0.23248257 0.52595514 0.97017586 0.45500824 0.08249777 0.17754507 -1.1586903 -0.10172407 0.9143197 -0.40041977 0.08492462 0.3938311 0.23322627 1.2738582 -0.1352076 0.8839622 0.96766645 0.2114027 0.43039757 -1.3525156 -0.2804768 0.33770424 0.39146012 -1.2328359 0.04583787 1.0070711 -0.23236772 0.90851307 0.30584422 1.0878634 0.9128113 0.0758688 0.8252104 1.315711 -0.655872 0.77268535 0.4274516 -0.10618997 0.6229385 0.33225945 0.18881965 -0.26034194 0.10133567 0.53491676 -0.36512107 -0.14519797 -1.0073583 -0.6128256 1.1949416 0.521545 0.52076715 -0.34214124 0.18855457 0.33273473 0.5652644 0.22001842 1.1538874 -0.55911404 -0.12290727 -1.0310436 0.34945711 1.1665437 0.7288032 0.86417806 -0.03986403 -0.03160735 0.3847955 -0.05612459 0.15607715 -0.060739 -0.9706159 0.34104156 0.8098969 0.12945618 -1.1282549 -0.76679283 -0.84222996 -0.35859305 0.44054964 0.6263143 -0.40315887 -0.52471703 1.4936467 0.4853343 -0.01983836 -0.11000943 1.2783215 0.07556485 0.55266684 -0.21265833 -0.4173557 1.000438 -1.0662749 -0.4462748 -0.1806548 0.6043294 -0.4869933 0.754448 0.69381934 -1.1937654 -0.06447603 -0.8698839 0.40780625 -0.3203654 -0.2916886 0.9433374 0.9375 -0.92851067 0.7552767 -0.8649136 -0.42554733 0.37892655 0.65348214 -0.0335166 -0.2180732 -0.9277397 1.1213682 0.63104546 0.21472752 -0.5525379 -0.31972542 -0.481649 0.30334386 1.2657928 -0.8506649 1.032938 -1.4623007 -1.931954 0.4289655 0.13948388 -0.63080245 0.61262447 0.329398 0.07940435 0.20445108 -0.26747578 0.3918931 0.78662527 -1.220879 -0.658294 -0.19023283 0.541058 0.10197079 -0.3198553 0.04531721 -0.49105147 -0.23375344 -0.10137676 -1.1601347 -0.85204554 -0.3017593 -0.2368115 -0.18376637 0.12292402 -0.29846698 1.1094252 -1.6550275 0.77176553 0.5361947 -0.01995178 0.20831898 -0.18333867 0.48990473 0.12393706 0.06890762 -0.23269415 0.18573512 0.0329931 0.1735348 0.19939402 0.41773066 0.30431023 0.7689582 -0.9660677 -0.34132716 0.24440244 0.14511263 -1.0528156 0.15138523 -0.884417 0.48438707 -0.69642407 0.57658356 0.48723263 -0.18844774 0.7279039 0.27195817 -0.25861102 0.11903202 -1.6488609 1.7166519 -0.70759004 0.22444248 0.25868604 -1.4835368 1.021151 0.01635654 0.57797974 -0.851792 0.47694352 0.514028 0.28100595 -0.46297163 0.24177898 -0.02381969 -0.0357632 0.3925262 -0.08228624 -0.24172263 0.7060162 -0.3457521 1.2382815 0.15806167 0.20367464 -0.18346092 0.65466225 0.29489923 0.6998152 0.6950377 0.19315413 0.356562 1.0117058 -0.59361464 -0.8287021 -0.34617972 0.1824558 0.94202155 0.13457902 -0.1894308 -0.82876414 -0.67001885 -0.23909834 0.64288646 -0.47224477 -0.02249094 -0.68955463 -0.49537638 -0.29247564 0.21447325 0.03388539 -1.3352222 -1.1699799 0.56752586 0.16564666 -1.0102386 0.02697386 0.6778322 -1.045224 -1.1204017 -0.6104855 -0.6744228 0.70955384 -0.04799198 1.1783068 0.32598433 -0.38108486 0.12760876 -0.6525706 -0.10875493 -0.20840617 0.7377153 -0.51235497 -0.14483562 0.03067538 -0.487802 -0.23567267 0.24093579 -0.74603385 -0.0737784 0.8108139 0.8832606 0.55439603 0.37344837 0.29574293 -1.08551 0.6451861 -0.47353768 -1.3929516 0.23162709 -0.7873869 0.4772592 1.0207978 -0.31398186 -0.5325611 0.4693 0.02703868 -0.36153957 -0.18215571 0.7831796 -0.17984502 -0.31319094 0.44758493 -0.03116921 -0.2782089 -0.32104534 0.3542538 0.11954983 -0.12564494 -0.8530117 0.5850151 0.09349101 0.33045742 -0.43763304 -0.77149266 -0.3626713 -0.44519636 -0.13294329 0.6075088 -0.28403905 -0.7671814 -0.04156414 -0.95086515 -0.45156807 -0.21999843 0.22322749 -0.67721176 0.13552175 -0.02599158 -0.78825366 -0.0751856 -1.4834391 0.56415784 0.44005314 -0.15755813 -0.93106735 0.14294209 0.40108928 0.5505095 0.42425308 0.76522017 -0.53943294 -0.97800696 -0.13599181 0.0383979 -0.09227407 -0.2532405 0.05418703 -0.43997073 -0.31518865 -0.26257572 -0.24269798 0.6784572 0.3938528 1.2830753 -0.3911697 -0.40072647 0.65793365 1.7224554 0.2529762 0.38909802 0.64087784 0.21586224 0.72401464 0.70856696 0.58320826 0.21237333 1.0384591 0.8850746 0.15573263 0.14768742 0.18771057 -0.02222267 0.33784327 -0.24908939 -0.38927436 -1.0655487 0.68721116 -1.8809984 -0.87556905 -0.07809623 2.1079423 0.6493187 0.3930283 0.3697879 0.27849856 0.53596294 -0.08747853 -0.49905175 -0.99860096 0.24192087 0.5099955 0.42471474 0.50137645 -0.9653567 1.0582442 5.259455 0.4747407 -1.4218729 -0.24195156 0.32410568 -0.27358443 -0.46689802 0.3586724 -0.5882993 0.4857788 0.9927297 -0.08478101 1.0031688 1.0087388 0.0855569 -0.20646694 -0.9340849 0.6770897 -0.15803327 -1.3306509 -0.38902658 -0.0320584 0.879571 -0.32452634 -0.15756877 0.18381988 0.2736519 -0.9943713 0.60469544 0.41853094 0.03781197 -1.1820894 0.80592823 0.35199928 -0.8758019 -0.4385876 -0.31648642 -0.1556513 0.10609562 0.73572403 -0.7750401 0.73443127 0.59605587 0.29296207 -0.35721913 1.6362084 -0.3641146 0.46491578 -0.2830016 -0.5818397 0.45508486 -0.2825672 0.560609 0.9584255 0.0868001 -0.17970131 0.5584735 1.036726 0.23546584 0.4074386 -0.30776635 -0.26164904 0.16311754 1.5819212 -1.2776889 0.2874495 -0.15603252 0.6836876 0.6221328 0.02075758 -0.75302655 -0.28202355 0.16618809 0.07032008 0.6761642 -0.295314 -0.47737876 -0.85847324 0.02295159 -0.9826997 0.4217021 -0.23547724 -0.91279614 0.58032894 -0.18466231 -0.9344827 -0.4172866 -0.4920176 -0.4072985 0.54368 -1.7840672 -0.63576794 -0.19891432 0.388912 0.33514827 -0.10499755 0.73347485 0.06344771 -0.71646553 0.30558306 -0.0995144 -0.3859035 0.2384732 -1.4591223 -0.1189372 0.7132121 0.03370005 0.3418035 1.0995811 -0.23975751 -1.8456064 -0.47191498 0.5503664 0.10745557 0.7489446 -0.26661986 -0.73391694 0.20831464 0.26658276 0.0375397 0.43875667 0.4339924 0.07371638 -0.24269138 -0.95873153 0.534522 0.67880154 0.22944835 -0.19274996 0.38731208 0.27478427 -0.6394306 -0.6004415 0.04854531 0.15518543 -1.0250088 0.5840674 -0.30601302 0.4804625 -0.318183 0.2751553 -1.4608008 -0.28012067 -0.7955938 -0.05846158 1.0500567 0.5539977 -0.53403515 0.96380514 0.6943012 -0.03872728 -1.0835955 -0.9078135 -0.88010556 0.05576289 -0.29157004 0.5432251 0.91138226 -0.01330591 0.19920222 -0.29477474 0.1876875 0.39665565 0.8522374 0.7569932 -1.3862066 -0.5978309 -0.78590095 -0.14508231 -0.5620523 0.38714725 -1.0018442 -0.05373879 -1.7496169 -0.0834577 -0.8098847 -0.16314887 0.5903974 0.1754281 -0.27860937 0.33003017 -0.35612637 -0.7654867 0.3276853 1.3548868 -0.15622377 -0.48662373 0.19268696 -0.7674776 0.44387764 0.97053283 -0.77405876 -0.25375926 -0.33059493 0.779409 0.46896714 -0.01406758 -1.0435108 0.23463309 -0.668019 -0.09897167 -0.08123142 0.08408924 -1.2961934 0.18816629 0.59963936 -0.33444762 -0.20805264 0.14137603 0.3598345 -0.07430363 -0.7850378 0.75582176 -0.34116295 -0.7460646 0.01859394 -0.36384127 0.01212192 1.392801 -0.15726686 0.14674592 -0.17920232 -0.6103663 0.48412767 0.3990655 0.35122102 0.15314934 -0.6761571 -0.48518616 -0.0984243 -0.18427531 -0.02193802 -0.05879158 0.71649235 -0.5610691 0.4966709 -0.46026245 -0.4245579 -0.9312705 0.92820424 0.5067645 -0.71493995 -0.5396822 0.5336114 -0.8125344 -0.39839745 0.1266406 -0.25075912 -0.2509844 0.2854991 0.12698114 0.40061194 0.20171168 0.03205245 -0.3987643 0.52664566 0.01326505 0.13238008 1.6425807 0.08558322 -0.12574403 0.03165395 0.79978186 0.00794432 -1.1099126 0.1407073 0.37209442 -0.4423011 0.17298515 -1.022523 -1.620236 0.67200816 0.33294922 0.609736 1.2980862 -0.25487998 0.53980225 0.58212084 0.86013234 -1.3051518 -0.01458068 0.42977872 0.72794217 -1.1082338 0.08283371 -0.4591939 -0.4964896 1.4341661 0.6029147 -0.3871418 0.22550453 0.3432948 -0.07034795 -0.1125839 -0.8315999 -0.56830025 0.02304563 0.3588917 0.2066181 -0.01919113 -0.84171 0.30897006 -0.17356917 0.12981586 0.5613452 0.842517 -0.3650779 -1.6235198 -0.30066925 0.21094754 -0.35358816 0.26825067 -0.36529523 0.745156 0.02158403 0.980564 -0.35247698 0.04648206 0.23175338 -0.1762812 0.69658726 -0.6225545 -0.97164816 -0.09100666 0.15675409 -0.83262646 -0.47930196 -0.5524972 -1.2038985 -0.0138065 -0.5053484 0.20037188 0.74863577 1.051839 0.14928746 0.7382908 0.72284776 -0.7247463 -0.36385027 -0.2526599 -0.44933954 0.02270789 -0.1285025 -0.77780133 -0.14731619 -0.48449707]
[5.277121543884277, 3.0971405506134033]
71368df0-3b95-40cf-a5e3-4a09d21fdbe9
a-novel-multi-stage-approach-for-hierarchical
null
null
https://ieeexplore.ieee.org/document/10077796
https://ieeexplore.ieee.org/document/10077796
A Novel Multi-Stage Approach for Hierarchical Intrusion Detection
An intrusion detection system (IDS), traditionally an example of an effective security monitoring system, is facing significant challenges due to the ongoing digitization of our modern society. The growing number and variety of connected devices are not only causing a continuous emergence of new threats that are not recognized by existing systems, but the amount of data to be monitored is also exceeding the capabilities of a single system. This raises the need for a scalable IDS capable of detecting unknown, zero-day, attacks. In this paper, a novel multi-stage approach for hierarchical intrusion detection is proposed. The proposed approach is validated on the public benchmark datasets, CIC-IDS-2017 and CSE-CIC-IDS-2018. Results demonstrate that our proposed approach besides effective and robust zero-day detection, outperforms both the baseline and existing approaches, achieving high classification performance, up to 96% balanced accuracy. Additionally, the proposed approach is easily adaptable without any retraining and takes advantage of n-tier deployments to reduce bandwidth and computational requirements while preserving privacy constraints. The best-performing models with a balanced set of thresholds correctly classified 87% or 41 out of 47 zero-day attacks, while reducing the bandwidth requirements up to 69%.
['Filip De Turck', 'Bruno Volckaert', 'Tim Wauters', 'Ying-Dar Lin', 'Didik Sudyana', 'Laurens D’hooge', 'Miel Verkerken']
2023-03-21
null
null
null
ieee-transactions-on-network-and-service-1
['network-intrusion-detection']
['miscellaneous']
[ 7.05393329e-02 -4.65690792e-01 -9.61681604e-02 -1.86656371e-01 -4.15926635e-01 -7.68610477e-01 6.82644725e-01 6.62687659e-01 -7.12718964e-01 6.20948672e-01 -5.25590181e-01 -5.86454153e-01 -3.19346935e-01 -8.13187420e-01 -6.01668209e-02 -4.41311449e-01 -2.11934194e-01 4.74296808e-01 6.34019315e-01 -5.28036654e-02 4.21463907e-01 9.98474896e-01 -1.34125960e+00 4.04476002e-03 5.03163517e-01 1.50592196e+00 -7.13067651e-01 4.84761268e-01 1.05120026e-01 2.18016766e-02 -8.56103241e-01 -5.53584754e-01 5.55476010e-01 5.48058003e-02 -4.77300495e-01 -3.02848518e-01 2.64619023e-01 -4.95672137e-01 -2.29332924e-01 9.90974069e-01 4.17885482e-01 1.16543714e-02 1.05298355e-01 -1.53779066e+00 -3.62436287e-02 1.91908091e-01 -8.48698974e-01 5.85267246e-01 2.00699165e-01 -1.35194749e-01 6.25891149e-01 -2.98205137e-01 3.82472366e-01 8.99467528e-01 3.88302892e-01 5.17309248e-01 -1.22340608e+00 -1.12168431e+00 6.43105879e-02 2.36942559e-01 -1.43038845e+00 -4.81231213e-01 4.67252821e-01 -1.40045851e-01 7.87298083e-01 5.57451069e-01 2.65932661e-02 1.06526947e+00 -5.12136631e-02 4.69109267e-02 1.01605928e+00 -2.22986475e-01 6.13035977e-01 5.74809849e-01 4.82140362e-01 1.89655840e-01 9.77535188e-01 1.27885371e-01 -2.15911090e-01 -7.41724491e-01 2.26300970e-01 5.00394762e-01 -3.22999209e-02 -2.45949537e-01 -6.89038038e-01 6.62314057e-01 5.66217974e-02 4.84532624e-01 -4.32591230e-01 -6.99607253e-01 8.23097229e-01 2.39463776e-01 2.42410123e-01 3.21674258e-01 -5.83013356e-01 -2.75471658e-01 -8.54542136e-01 -1.57622918e-01 9.89500344e-01 5.91873288e-01 3.96939963e-01 1.10948861e-01 2.53877848e-01 3.67827356e-01 -9.41033959e-02 4.25861984e-01 2.64275134e-01 -1.98530212e-01 3.14707071e-01 8.86161804e-01 -8.59733149e-02 -1.21186256e+00 -6.31464899e-01 -7.05438912e-01 -1.23070490e+00 -8.83518159e-02 1.67597055e-01 -1.45995721e-01 -6.06559753e-01 1.58156490e+00 6.16096616e-01 4.12709475e-01 -7.91903883e-02 2.71555364e-01 -3.77071323e-03 3.86192709e-01 1.72925666e-01 -4.28176790e-01 1.39708054e+00 -1.82033673e-01 -5.84501147e-01 -4.32559848e-02 2.93787003e-01 -5.88673711e-01 4.88816053e-01 6.67105019e-01 -6.09454274e-01 -1.46072477e-01 -1.23799014e+00 7.20802128e-01 -7.67675281e-01 -3.66765440e-01 5.74211717e-01 1.31826210e+00 -7.62354136e-01 3.13514888e-01 -7.46071160e-01 -7.43976355e-01 6.70927286e-01 5.12739062e-01 -5.53308845e-01 -5.21856919e-02 -1.14675784e+00 6.22564375e-01 5.20526290e-01 -3.66600156e-01 -6.56485021e-01 -5.93759954e-01 -2.96797603e-01 1.31477982e-01 5.21641076e-01 -5.43574356e-02 7.61556387e-01 -2.43185282e-01 -1.07667768e+00 5.01216710e-01 2.84336478e-01 -7.74690092e-01 3.82045001e-01 -1.70063078e-02 -1.02796054e+00 3.17755222e-01 -2.13367254e-01 -9.26494598e-02 7.40663350e-01 -9.70746040e-01 -9.58842397e-01 -7.94730067e-01 6.37445599e-03 -3.07832867e-01 -1.10668623e+00 4.06542331e-01 -8.91411901e-02 -2.74592787e-01 -1.69631526e-01 -8.32876146e-01 -3.62338811e-01 -2.61306912e-01 -4.97977078e-01 1.07187085e-01 1.65817690e+00 -4.91748571e-01 1.58058643e+00 -2.08882976e+00 -5.16519725e-01 6.41500890e-01 1.59658313e-01 1.04703891e+00 2.04586774e-01 5.18711984e-01 -7.63306394e-02 2.88043916e-01 -3.47649246e-01 -4.16344345e-01 -1.00145206e-01 1.06486134e-01 -3.61796409e-01 4.73906338e-01 -4.50601242e-02 3.56079787e-02 -4.05008167e-01 8.48981589e-02 4.07352477e-01 6.29076362e-01 -3.72861385e-01 2.86825776e-01 2.04284862e-01 4.25404668e-01 -3.94751638e-01 7.86267757e-01 1.01199841e+00 -1.60839021e-01 2.93199599e-01 2.16330409e-01 3.45443450e-02 -3.27118039e-02 -1.39835191e+00 9.42825675e-01 -1.80887431e-01 7.38075450e-02 2.29327023e-01 -1.14796221e+00 1.16171634e+00 5.10683656e-01 6.58139229e-01 -6.98800266e-01 2.67256767e-01 2.16222346e-01 4.67422456e-02 8.46668426e-03 1.99756469e-03 2.24830866e-01 -3.33591342e-01 6.91737592e-01 -2.64244646e-01 7.30323613e-01 2.21489891e-01 3.50777864e-01 1.51458514e+00 -9.30702507e-01 5.99067271e-01 -4.85526659e-02 9.35589671e-01 -4.85286027e-01 6.46883667e-01 7.00141430e-01 -6.47506118e-01 -1.64872304e-01 4.28030491e-01 -5.12547016e-01 -6.15711272e-01 -8.83884966e-01 -3.46560746e-01 7.29335845e-01 -4.99290489e-02 -4.81003255e-01 -7.75523424e-01 -8.92837048e-01 6.94023222e-02 5.50334036e-01 -3.66430134e-01 -2.52938598e-01 -4.42026407e-01 -9.34467018e-01 8.95656466e-01 1.21814877e-01 9.52831209e-01 -7.20682502e-01 -7.76372194e-01 2.33600691e-01 2.45884627e-01 -1.50477350e+00 5.72030097e-02 -4.12268117e-02 -6.95192277e-01 -1.31672966e+00 8.47682431e-02 -2.36062497e-01 5.30540586e-01 3.13923746e-01 6.68284953e-01 2.03475952e-01 -7.49106348e-01 3.07570010e-01 -2.34308928e-01 -5.86259782e-01 -1.97941348e-01 2.35987872e-01 4.88950759e-01 4.02088255e-01 7.60170758e-01 -7.20011413e-01 -5.91033399e-01 3.48105580e-01 -1.06935275e+00 -7.93411195e-01 5.32982409e-01 4.23390955e-01 1.49474323e-01 4.14001286e-01 1.12892962e+00 -1.06640184e+00 5.71085095e-01 -8.05676103e-01 -7.39287138e-01 6.36063740e-02 -9.73542035e-01 -5.65283895e-01 9.91845787e-01 -4.24481183e-01 -8.12875450e-01 -1.10296533e-01 -7.30481595e-02 1.19777629e-03 -6.77598417e-01 -2.25710869e-02 -3.57039005e-01 -2.77867436e-01 6.08813047e-01 5.22232354e-02 -7.31283426e-02 -4.75627869e-01 -3.37350257e-02 1.01346004e+00 3.65247130e-01 -2.81220734e-01 1.10698009e+00 5.81166863e-01 3.06744486e-01 -1.07550120e+00 -4.19562459e-01 -6.70809627e-01 -5.53184450e-01 1.20298535e-01 3.38405013e-01 -4.00215894e-01 -1.01995790e+00 5.39003849e-01 -8.71127009e-01 5.59564650e-01 2.18064502e-01 1.72681302e-01 3.86025488e-01 6.60721600e-01 -5.24150908e-01 -1.13080931e+00 -8.92224312e-01 -8.64067078e-01 4.33082312e-01 2.52227247e-01 -9.58143175e-02 -7.48622477e-01 -2.42523905e-02 2.60856360e-01 9.75546360e-01 5.54409504e-01 6.60836756e-01 -1.52815521e+00 -3.71328473e-01 -1.12469661e+00 -4.05897677e-01 3.68735701e-01 4.07303840e-01 -7.03588203e-02 -8.80004823e-01 -8.71825993e-01 1.76948786e-01 3.09848543e-02 2.57365406e-01 -2.17463732e-01 1.31043553e+00 -4.50210959e-01 -3.46150786e-01 4.28900540e-01 1.48477244e+00 7.34412789e-01 3.85919929e-01 5.67494392e-01 2.52551377e-01 4.72675204e-01 5.31063020e-01 9.39369619e-01 4.09508124e-03 5.88138819e-01 7.46995509e-01 1.44930203e-02 6.79075301e-01 1.19454615e-01 -3.01733650e-02 3.98999184e-01 3.81890655e-01 -1.81969851e-01 -1.06425142e+00 3.03103030e-01 -1.52366638e+00 -9.36106861e-01 3.93658392e-02 2.71563935e+00 2.72823244e-01 5.61943769e-01 3.76374930e-01 7.14753687e-01 7.28578925e-01 4.20294553e-02 -7.64567852e-01 -6.53355241e-01 1.22688442e-01 6.35294437e-01 5.95873058e-01 6.72597587e-02 -1.26508343e+00 4.36614186e-01 5.56972313e+00 5.04480541e-01 -1.17359829e+00 1.39285073e-01 7.09063172e-01 -4.61170822e-03 6.12549245e-01 -1.49600789e-01 -8.63395333e-01 5.66063762e-01 1.53142238e+00 -3.42602402e-01 2.59249598e-01 9.84352112e-01 -2.41851956e-02 1.62569612e-01 -7.35778213e-01 8.48736167e-01 1.49928197e-01 -1.12529993e+00 -2.13762205e-02 4.62398410e-01 2.70521164e-01 -2.43566275e-01 1.41529441e-01 1.46696836e-01 2.66847685e-02 -7.30270267e-01 -1.98349759e-01 3.11462209e-02 6.56340480e-01 -1.25829875e+00 8.52953553e-01 5.94225168e-01 -1.17321861e+00 -4.10314947e-01 -1.52702764e-01 -5.26883937e-02 -5.34695685e-02 6.14747584e-01 -7.69835591e-01 6.82863593e-01 8.00022542e-01 2.84493059e-01 -7.47680902e-01 1.02172542e+00 2.10653335e-01 7.08909512e-01 -5.96747875e-01 -3.24624777e-02 5.52412234e-02 1.28846198e-01 6.29578412e-01 1.06074929e+00 1.92705885e-01 5.98690398e-02 4.19660360e-01 -5.08179031e-02 -7.94272572e-02 2.21393138e-01 -5.93376279e-01 2.66723502e-02 9.09041762e-01 1.47652769e+00 -6.73709393e-01 -2.29541212e-01 -2.54863232e-01 5.68936110e-01 3.64926718e-02 -9.81508195e-02 -7.39600778e-01 -5.51405013e-01 8.76090288e-01 3.94589037e-01 -6.02386370e-02 -1.23164997e-01 -2.74575979e-01 -9.49356139e-01 -4.27484736e-02 -1.09811771e+00 1.00476336e+00 3.37251037e-01 -1.25957417e+00 9.65199828e-01 -1.17467739e-01 -1.12908590e+00 1.92805787e-03 -5.39342046e-01 -6.32895291e-01 4.66476053e-01 -1.33242297e+00 -9.09373403e-01 -2.30265528e-01 7.12075114e-01 1.24075256e-01 -4.18463856e-01 1.27224135e+00 8.26063216e-01 -1.22551763e+00 1.06022239e+00 1.46893263e-01 6.00996725e-02 6.56007230e-01 -7.87133873e-01 3.42620969e-01 1.32792950e+00 -1.16502516e-01 9.49806809e-01 3.58022004e-01 -5.25057733e-01 -1.11855376e+00 -1.13123512e+00 3.54871124e-01 -2.01805100e-01 6.87508583e-01 -6.24115407e-01 -1.12891102e+00 5.13028502e-01 5.47668599e-02 1.17387131e-01 1.09531188e+00 -6.05617091e-02 -6.43941045e-01 -6.15970850e-01 -1.86619496e+00 3.30584705e-01 6.02110505e-01 -2.70210177e-01 -1.13131953e-02 1.72656298e-01 4.25359786e-01 8.54726136e-02 -8.21396351e-01 4.74083871e-01 2.83993870e-01 -1.08935010e+00 8.39828253e-01 -6.56328499e-01 -6.57634437e-01 -1.63309500e-01 -7.89004713e-02 -7.23213196e-01 -7.30080083e-02 -9.47083414e-01 -4.69902933e-01 1.53917587e+00 2.08540469e-01 -1.28002763e+00 1.01690280e+00 7.51022816e-01 6.52574062e-01 -6.54613912e-01 -1.18592358e+00 -7.66699970e-01 -4.18033093e-01 -2.28112116e-01 7.85689294e-01 1.19779420e+00 -6.47731423e-02 2.85474844e-02 -3.97638738e-01 4.52639103e-01 1.06850922e+00 -3.78643930e-01 8.60751152e-01 -1.63014042e+00 -1.21250138e-01 -2.23058924e-01 -8.72093081e-01 -1.89733222e-01 -2.86131173e-01 -5.05852342e-01 -8.33302498e-01 -7.71041274e-01 -3.56261618e-02 -5.30196369e-01 -8.64739239e-01 5.93869090e-01 2.71777391e-01 3.82957697e-01 -8.00045729e-02 1.42429059e-03 -5.23438573e-01 -3.93711068e-02 1.43010855e-01 2.64138013e-01 -7.44864047e-02 2.50799924e-01 -7.90499270e-01 6.34644032e-01 1.21159077e+00 -4.94797111e-01 -3.95362854e-01 3.13392401e-01 -3.08451474e-01 -1.75317526e-01 4.91498969e-02 -1.28677619e+00 4.71772552e-01 -5.69340922e-02 4.22685832e-01 -4.78603572e-01 3.01782757e-01 -1.22189856e+00 2.07995594e-01 8.80429864e-01 9.37739685e-02 3.48982394e-01 4.09737557e-01 6.29143834e-01 -7.41219446e-02 2.47308999e-01 1.13914812e+00 2.30928734e-01 -4.04468477e-01 7.06973612e-01 -1.40964895e-01 -3.43537390e-01 1.67879820e+00 -1.89092293e-01 -6.17916226e-01 1.25828357e-02 -3.20811003e-01 9.82195735e-02 3.04221928e-01 5.84054530e-01 5.28862834e-01 -8.88058782e-01 -3.60576570e-01 5.69494128e-01 1.28085345e-01 -4.70082760e-01 3.63221884e-01 5.17103612e-01 -3.63966972e-01 5.18744409e-01 -4.71112639e-01 -2.34519467e-01 -1.59856999e+00 7.31319845e-01 -6.38927221e-02 -5.68801582e-01 -5.24949312e-01 4.45337653e-01 -3.23682606e-01 -1.91923514e-01 6.34787440e-01 4.41268802e-01 -1.22763172e-01 3.74729484e-02 1.06230068e+00 7.07885802e-01 3.21272969e-01 -5.92194974e-01 -7.25669742e-01 3.97098102e-02 -5.91620147e-01 1.96402967e-01 1.27310252e+00 8.56676046e-03 -2.92633623e-01 6.14615995e-03 1.08344948e+00 -2.90588364e-02 -4.61442560e-01 -2.16076806e-01 4.47102845e-01 -7.73992956e-01 -1.11974642e-01 -9.63094234e-01 -1.13687015e+00 8.22968483e-01 8.66413653e-01 7.22723186e-01 1.38901985e+00 -6.78357363e-01 1.01973474e+00 3.28918248e-01 6.72328651e-01 -6.78992867e-01 -6.27635270e-02 1.64119467e-01 1.22343808e-01 -1.12321663e+00 -3.17808725e-02 -3.15909535e-01 -3.19751948e-01 9.40835655e-01 8.46865892e-01 1.66975204e-02 8.47066224e-01 3.55000913e-01 -3.48069482e-02 -2.41706129e-02 -7.44410515e-01 3.23518604e-01 -2.44287834e-01 7.65890181e-01 -1.53497845e-01 1.53021906e-02 -3.29527527e-01 5.25377393e-01 2.62096792e-01 -5.48967421e-01 4.46117014e-01 1.13819635e+00 -3.89185101e-01 -1.35767055e+00 -4.28360671e-01 6.69776678e-01 -8.68864298e-01 4.12265927e-01 -5.28670907e-01 6.55828834e-01 1.50549179e-02 1.26268375e+00 -5.17788082e-02 -6.09781623e-01 4.52960610e-01 2.12790743e-02 -4.15720254e-01 -4.54142511e-01 -7.65795112e-01 -3.99219990e-01 -8.09997320e-02 -5.80240905e-01 1.10958092e-01 -5.68936825e-01 -9.30216849e-01 -7.31006145e-01 -3.66209298e-01 3.98615152e-01 9.90508795e-01 5.65641284e-01 7.77206302e-01 2.94746637e-01 1.23033834e+00 -1.69783980e-01 -8.46083283e-01 -8.01085472e-01 -6.09484732e-01 3.18091720e-01 1.52585581e-01 -7.31851518e-01 -6.59627140e-01 -7.39226162e-01]
[5.270878791809082, 7.198757648468018]
bc7cd91c-1537-456c-82e0-d325ed68df36
progressive-unsupervised-person-re
1910.11560
null
https://arxiv.org/abs/1910.11560v1
https://arxiv.org/pdf/1910.11560v1.pdf
Progressive Unsupervised Person Re-identification by Tracklet Association with Spatio-Temporal Regularization
Existing methods for person re-identification (Re-ID) are mostly based on supervised learning which requires numerous manually labeled samples across all camera views for training. Such a paradigm suffers the scalability issue since in real-world Re-ID application, it is difficult to exhaustively label abundant identities over multiple disjoint camera views. To this end, we propose a progressive deep learning method for unsupervised person Re-ID in the wild by Tracklet Association with Spatio-Temporal Regularization (TASTR). In our approach, we first collect tracklet data within each camera by automatic person detection and tracking. Then, an initial Re-ID model is trained based on within-camera triplet construction for person representation learning. After that, based on the person visual feature and spatio-temporal constraint, we associate cross-camera tracklets to generate cross-camera triplets and update the Re-ID model. Lastly, with the refined Re-ID model, better visual feature of person can be extracted, which further promote the association of cross-camera tracklets. The last two steps are iterated multiple times to progressively upgrade the Re-ID model.
['Guo-Jun Qi', 'Qiaokang Xie', 'Wengang Zhou', 'Qi Tian', 'Houqiang Li']
2019-10-25
null
null
null
null
['unsupervised-person-re-identification']
['computer-vision']
[-4.49089184e-02 -5.56527972e-01 4.42529377e-03 -3.49199086e-01 -5.87625682e-01 -7.19173551e-01 5.82454503e-01 2.78987288e-01 -5.31981051e-01 6.11174047e-01 2.86576867e-01 4.30511028e-01 -1.13164149e-01 -4.94897872e-01 -5.42923212e-01 -5.89836061e-01 2.20203519e-01 5.23622572e-01 2.44639833e-02 2.44904369e-01 -1.52737163e-02 5.53534091e-01 -1.51458681e+00 3.07660196e-02 6.55267537e-01 6.15174770e-01 -3.72041017e-02 4.23731834e-01 1.61713827e-02 5.15219450e-01 -3.49728882e-01 -6.86187208e-01 3.62660885e-01 -4.14094985e-01 -5.01019657e-01 4.41481858e-01 8.69748414e-01 -3.34457308e-01 -4.88462329e-01 1.13677180e+00 4.48552877e-01 4.77695137e-01 4.56618935e-01 -1.33383346e+00 -6.61338627e-01 3.36913377e-01 -9.69388664e-01 1.71330646e-01 5.16143322e-01 -1.13458194e-01 8.84187281e-01 -9.02275443e-01 4.56247360e-01 1.17716396e+00 1.02831197e+00 6.22983694e-01 -1.44301116e+00 -8.94108176e-01 5.56824923e-01 2.36494079e-01 -1.75546050e+00 -4.15301830e-01 8.11067641e-01 -5.41530430e-01 3.15717250e-01 5.05190976e-02 8.86004090e-01 1.00587451e+00 -5.50493300e-01 6.46932304e-01 6.81625783e-01 -3.37250352e-01 -1.89502820e-01 2.19262928e-01 2.83528298e-01 5.78840077e-01 3.53089124e-01 1.36448905e-01 -5.12736022e-01 -3.82083692e-02 7.90042281e-01 4.62428331e-01 -1.45101100e-02 -5.14935136e-01 -1.18999338e+00 5.29859304e-01 5.16672552e-01 3.57301772e-01 -3.79060596e-01 -1.19989544e-01 5.42170107e-01 -3.72446030e-02 2.15235546e-01 4.48826654e-03 1.77794229e-02 2.46916652e-01 -1.17660058e+00 3.99109304e-01 2.32778817e-01 9.46792603e-01 9.14567411e-01 -2.58906335e-01 -2.80732304e-01 1.06793332e+00 2.02105299e-01 5.61565995e-01 4.16048020e-01 -6.23177409e-01 4.78053331e-01 7.15516150e-01 4.66956288e-01 -1.19920886e+00 -3.47970545e-01 -6.33200049e-01 -1.17571509e+00 -7.48035908e-02 5.24156749e-01 -1.08693972e-01 -7.55589306e-01 1.74700236e+00 4.67309296e-01 7.17562258e-01 -4.83962335e-02 9.66937959e-01 6.70564950e-01 3.99749547e-01 2.81449169e-01 -1.26070738e-01 1.52059853e+00 -9.57500637e-01 -5.55577159e-01 -3.30243222e-02 4.24872696e-01 -5.58644414e-01 5.41641057e-01 1.40148118e-01 -6.95513010e-01 -1.00958157e+00 -8.75925839e-01 1.69985041e-01 -2.16125757e-01 6.74007118e-01 2.63209850e-01 6.26381099e-01 -8.20272267e-01 1.32694647e-01 -5.34218669e-01 -6.15498662e-01 4.03650433e-01 4.44009453e-01 -7.02710390e-01 -2.51584858e-01 -8.91363621e-01 4.68091905e-01 5.56787431e-01 2.08201200e-01 -8.20871115e-01 -6.69732451e-01 -8.41978788e-01 -5.33435717e-02 1.60247475e-01 -7.38790214e-01 7.56668389e-01 -1.03795648e+00 -1.10138428e+00 1.12290883e+00 -3.68863583e-01 -3.40747535e-01 5.20086884e-01 -2.28355274e-01 -7.40591347e-01 -7.55216330e-02 4.58111584e-01 4.67993617e-01 9.21067476e-01 -1.64037597e+00 -1.06635630e+00 -4.98631209e-01 6.65422231e-02 3.07400525e-01 -3.75608116e-01 5.47517687e-02 -8.87866318e-01 -7.32371926e-01 1.25235379e-01 -1.16551816e+00 -5.66427857e-02 -1.03296198e-01 -3.96984458e-01 -3.24952155e-01 5.98099589e-01 -7.35652506e-01 1.20645118e+00 -2.27202392e+00 1.35122612e-01 3.45235109e-01 3.14891040e-01 4.29367870e-01 -1.68137133e-01 1.77810058e-01 -3.01423818e-01 -2.20459774e-01 1.11550651e-01 -7.73735583e-01 -1.21553183e-01 -1.96050659e-01 -9.92353633e-03 5.57237685e-01 -1.01022720e-01 6.63660467e-01 -9.24412966e-01 -6.78410709e-01 5.01671970e-01 3.55952948e-01 -3.87028098e-01 2.06344828e-01 3.13249916e-01 7.11934507e-01 -2.77293354e-01 4.59555775e-01 9.28908765e-01 -2.30114207e-01 8.79770145e-02 -5.65189779e-01 -2.94203430e-01 -5.71510673e-01 -1.41666996e+00 1.61769998e+00 -1.58503726e-01 2.80229509e-01 -1.48749426e-01 -9.23838913e-01 9.57653761e-01 1.75844371e-01 8.56033206e-01 -4.18238610e-01 -1.44300386e-01 -3.46256383e-02 -4.02588189e-01 -4.00529832e-01 5.10340691e-01 6.14423379e-02 1.18835373e-02 6.81879342e-01 -2.13867798e-02 9.59309518e-01 2.15742573e-01 1.26766473e-01 4.62761939e-01 1.74609706e-01 1.25410363e-01 1.29736722e-01 1.11965668e+00 -1.51757345e-01 7.42215037e-01 7.54579008e-01 -4.45822597e-01 6.07200444e-01 -2.65902609e-01 -8.40444982e-01 -1.26593816e+00 -1.06454647e+00 1.05491001e-02 1.07796097e+00 4.64187205e-01 -3.94020230e-01 -7.37741113e-01 -8.46828759e-01 -4.81800325e-02 2.00483948e-01 -7.37863839e-01 4.72996198e-02 -6.95890665e-01 -7.37997651e-01 6.34427786e-01 5.40502906e-01 7.80204713e-01 -7.70789504e-01 2.08888780e-02 1.94257990e-01 -4.53596443e-01 -1.12987423e+00 -9.63092089e-01 -4.46138769e-01 -3.50921363e-01 -1.12776935e+00 -1.24917424e+00 -1.03738332e+00 1.00401771e+00 7.31253743e-01 5.52674949e-01 2.18114898e-01 -1.48606420e-01 4.88464534e-01 -3.31807435e-01 9.09746140e-02 5.55565581e-02 -1.32694423e-01 4.46957082e-01 6.96641326e-01 6.73165202e-01 -4.55593884e-01 -6.91570878e-01 3.27895790e-01 -3.75455618e-01 -9.63668823e-02 3.22888970e-01 8.10406983e-01 7.15983152e-01 2.53201008e-01 4.46870416e-01 -5.81576943e-01 2.67187536e-01 -2.90771633e-01 -5.57844818e-01 6.30499482e-01 -1.79711863e-01 -2.54330218e-01 4.50397044e-01 -7.14085281e-01 -1.15016806e+00 3.97595495e-01 8.74149874e-02 -6.96377873e-01 -2.63128400e-01 1.53638542e-01 -2.71052241e-01 -6.72238469e-02 4.10758913e-01 4.79250044e-01 -3.69889826e-01 -4.88767385e-01 2.92063296e-01 5.45917928e-01 8.08990061e-01 -3.88365149e-01 1.11562502e+00 6.44909739e-01 -3.88265073e-01 -5.27738631e-01 -1.04512835e+00 -7.63223946e-01 -1.14983106e+00 -4.67902035e-01 1.26057065e+00 -1.26682556e+00 -8.49207997e-01 6.73175514e-01 -1.08616030e+00 8.94741043e-02 -5.03253117e-02 7.81500340e-01 -1.03442870e-01 7.29262352e-01 -3.23806494e-01 -9.69109237e-01 -2.98978359e-01 -7.65381396e-01 1.13359463e+00 5.66356957e-01 -7.56240487e-02 -9.50934768e-01 2.62463659e-01 3.60724539e-01 -9.21360478e-02 4.15801667e-02 2.97125220e-01 -5.76670945e-01 -4.70264673e-01 -5.24261475e-01 -5.28846800e-01 1.50084244e-02 3.99707347e-01 -3.82796466e-01 -8.57993245e-01 -5.97565114e-01 -4.56506908e-01 -5.91790415e-02 6.55568600e-01 3.01734179e-01 1.01414526e+00 -7.67893195e-02 -7.33952463e-01 8.67359877e-01 1.30253458e+00 9.38440710e-02 2.02422664e-01 4.93097335e-01 1.22454298e+00 6.35645330e-01 5.84879458e-01 5.64485788e-01 7.62845337e-01 9.16408896e-01 -1.26938969e-01 -2.03205004e-01 -4.91810367e-02 -5.38839757e-01 1.44485131e-01 5.09520471e-01 -4.14684951e-01 -5.36809154e-02 -6.69690132e-01 5.52913904e-01 -1.98809874e+00 -1.31326401e+00 5.81712555e-03 2.42612576e+00 2.44716480e-01 -2.42021412e-01 6.66036427e-01 -2.35095876e-03 1.54752612e+00 -6.15462512e-02 -5.48071921e-01 4.84554440e-01 -5.32738082e-02 -5.24869502e-01 4.83604372e-01 2.98773050e-01 -1.39620447e+00 9.44298863e-01 5.38498402e+00 5.81587136e-01 -8.33604813e-01 6.76725432e-02 2.64318258e-01 2.41754465e-02 4.83181365e-02 -8.59491751e-02 -1.19161928e+00 6.07644618e-01 2.27082014e-01 -1.67373285e-01 5.29185116e-01 5.11627078e-01 7.55418986e-02 1.66775599e-01 -1.08580053e+00 1.76357925e+00 4.21218812e-01 -1.10468876e+00 4.00841143e-03 1.28664628e-01 7.72548020e-01 -4.89892274e-01 -1.58475801e-01 1.90510198e-01 3.93314630e-01 -6.72135830e-01 7.17526972e-01 6.87035561e-01 7.92366147e-01 -9.59505737e-01 6.86520457e-01 2.88551837e-01 -1.82878399e+00 -3.41895908e-01 -4.71990794e-01 4.56269741e-01 3.76488924e-01 1.35369584e-01 -2.82219499e-01 8.14870536e-01 9.62182105e-01 1.09288657e+00 -7.25788832e-01 1.16384602e+00 2.57420152e-01 5.90716749e-02 -2.22174019e-01 6.06796443e-01 -7.11818831e-03 -3.39151204e-01 4.53105897e-01 1.13034916e+00 3.78106624e-01 1.65005118e-01 6.08700752e-01 5.77967405e-01 5.01801213e-03 -9.31364894e-02 -3.83611023e-01 4.42588866e-01 7.02505171e-01 1.25490916e+00 -6.50182784e-01 -3.52177083e-01 -6.24084055e-01 1.24721050e+00 5.35963595e-01 5.50484359e-01 -9.46371853e-01 5.57700656e-02 5.44351220e-01 8.74759406e-02 3.71314317e-01 -1.60974115e-01 2.96022445e-01 -1.41860342e+00 6.51666000e-02 -6.62890911e-01 8.08142304e-01 -5.48774540e-01 -1.75483942e+00 6.70008004e-01 -2.73275040e-02 -1.61094952e+00 3.80667262e-02 -2.35073134e-01 -4.83379185e-01 8.01578641e-01 -1.19773448e+00 -1.83438087e+00 -6.26220345e-01 1.11166728e+00 2.98299670e-01 -4.99969214e-01 6.88597739e-01 7.36654103e-01 -9.70013559e-01 1.10985398e+00 2.36832141e-03 5.90916276e-01 8.94077003e-01 -9.87567842e-01 3.44708353e-01 1.06040263e+00 1.43485546e-01 8.81467819e-01 2.13552460e-01 -9.07785594e-01 -9.26399946e-01 -1.45726585e+00 8.43401730e-01 -4.08924371e-01 3.88645142e-01 -4.30606931e-01 -8.30041051e-01 9.41396594e-01 -2.03512043e-01 1.56044975e-01 8.57906044e-01 3.53283912e-01 -4.29835945e-01 -4.02706891e-01 -9.58750367e-01 4.84733552e-01 1.19961119e+00 -6.66980028e-01 -3.50351036e-01 2.81275660e-01 2.39573285e-01 -1.27696663e-01 -8.44145119e-01 1.60059988e-01 6.42552853e-01 -6.87475145e-01 1.35662758e+00 -5.06176114e-01 -3.30767930e-01 -8.00500512e-01 2.95758918e-02 -9.49421525e-01 -7.52420843e-01 -3.93787354e-01 1.96537450e-01 1.81940019e+00 -1.51695400e-01 -4.20732290e-01 8.74399483e-01 5.42337775e-01 2.20660090e-01 5.07959053e-02 -8.01490784e-01 -7.83016264e-01 -3.71457338e-01 -2.42142845e-02 7.89037764e-01 1.02586079e+00 -2.51280069e-01 2.15476394e-01 -9.11136925e-01 5.80910444e-01 1.14350581e+00 1.62842706e-01 1.07135499e+00 -1.43918049e+00 -6.99261725e-02 -1.03401400e-01 -4.97270763e-01 -1.06763327e+00 2.12792724e-01 -9.07048464e-01 -5.52185886e-02 -1.31287038e+00 7.44033277e-01 -7.89981127e-01 -4.02223200e-01 3.05485845e-01 -4.26736712e-01 4.72097963e-01 4.42006260e-01 6.92246020e-01 -9.71087694e-01 4.63533968e-01 8.99715662e-01 -1.74886420e-01 -3.71516228e-01 5.50841205e-02 -5.77180803e-01 6.59865379e-01 4.50486928e-01 -3.77637267e-01 -2.51394540e-01 -4.40148860e-01 -2.42867559e-01 -1.08656235e-01 7.81033218e-01 -1.31007934e+00 6.09730661e-01 6.61828592e-02 9.48749125e-01 -8.09131980e-01 3.53271931e-01 -9.06058252e-01 4.19163316e-01 2.23100901e-01 -3.43127012e-01 1.59081578e-01 1.41754083e-03 7.76857376e-01 -1.79823697e-01 -1.98357955e-01 8.14056933e-01 -2.38313541e-01 -7.92002738e-01 8.17090333e-01 -1.99530199e-01 -1.32243484e-01 1.20720959e+00 -4.98287618e-01 4.90154810e-02 -1.50128126e-01 -1.04613006e+00 3.77498448e-01 6.16862059e-01 3.95785302e-01 4.84221995e-01 -1.74153018e+00 -9.04474378e-01 3.06198984e-01 4.15095985e-01 -2.81335950e-01 7.54202604e-01 6.35667741e-01 -1.61824763e-01 1.85478047e-01 -2.90296108e-01 -8.22124779e-01 -1.53711104e+00 9.25157011e-01 4.73682225e-01 -2.74210840e-01 -9.20726478e-01 7.00997829e-01 4.83942688e-01 -3.83496940e-01 3.25737059e-01 4.73069578e-01 -5.48040867e-01 2.05925465e-01 7.98861504e-01 2.80527830e-01 -4.12551880e-01 -1.22003913e+00 -5.55681109e-01 1.07794452e+00 -3.55313450e-01 -1.75798759e-01 1.10783482e+00 -6.13897562e-01 3.66863944e-02 1.75736785e-01 1.05368555e+00 1.48635387e-01 -1.33849287e+00 -5.09368598e-01 -9.92557555e-02 -6.31763279e-01 -4.64356840e-01 -3.23806047e-01 -1.05347669e+00 5.95718205e-01 9.53362226e-01 -1.52860656e-01 1.06392074e+00 -1.63856238e-01 8.73995245e-01 1.98699698e-01 3.54614049e-01 -1.01586175e+00 1.76006034e-01 2.25888237e-01 4.81826127e-01 -1.22421992e+00 7.78629035e-02 -3.30811769e-01 -7.27871716e-01 8.31270754e-01 7.25588620e-01 -1.11220367e-01 5.63319504e-01 -1.58267185e-01 6.23964369e-02 9.10351351e-02 4.32959273e-02 -4.36561733e-01 2.84963697e-01 9.43485677e-01 8.89430419e-02 7.09416298e-03 2.23588794e-01 6.52519047e-01 -3.27984318e-02 1.05017051e-03 3.27500887e-02 4.72586930e-01 -4.69691083e-02 -1.20088029e+00 -6.65599167e-01 1.71439394e-01 -1.72916993e-01 1.72284439e-01 -2.30191574e-01 5.58063030e-01 6.44175112e-01 9.08394873e-01 1.61997840e-01 -4.90003884e-01 2.67629266e-01 -1.97251350e-01 5.50008357e-01 -4.76400584e-01 -5.84853172e-01 6.46046922e-02 -9.78661403e-02 -2.02578321e-01 -7.10819185e-01 -9.34157789e-01 -9.01486695e-01 -3.95854443e-01 -2.85788149e-01 1.28643185e-01 1.17597312e-01 8.92200708e-01 2.20541120e-01 2.25017250e-01 5.87125361e-01 -9.23823833e-01 4.28207219e-02 -6.41740561e-01 -5.91209352e-01 8.57150316e-01 4.22700822e-01 -8.65638793e-01 -8.83655772e-02 5.10412931e-01]
[14.770283699035645, 1.0224322080612183]
7f3fd312-3269-4e00-b429-c189779fbcd9
iaunet-global-context-aware-feature-learning
2009.01035
null
https://arxiv.org/abs/2009.01035v1
https://arxiv.org/pdf/2009.01035v1.pdf
IAUnet: Global Context-Aware Feature Learning for Person Re-Identification
Person re-identification (reID) by CNNs based networks has achieved favorable performance in recent years. However, most of existing CNNs based methods do not take full advantage of spatial-temporal context modeling. In fact, the global spatial-temporal context can greatly clarify local distractions to enhance the target feature representation. To comprehensively leverage the spatial-temporal context information, in this work, we present a novel block, Interaction-Aggregation-Update (IAU), for high-performance person reID. Firstly, Spatial-Temporal IAU (STIAU) module is introduced. STIAU jointly incorporates two types of contextual interactions into a CNN framework for target feature learning. Here the spatial interactions learn to compute the contextual dependencies between different body parts of a single frame. While the temporal interactions are used to capture the contextual dependencies between the same body parts across all frames. Furthermore, a Channel IAU (CIAU) module is designed to model the semantic contextual interactions between channel features to enhance the feature representation, especially for small-scale visual cues and body parts. Therefore, the IAU block enables the feature to incorporate the globally spatial, temporal, and channel context. It is lightweight, end-to-end trainable, and can be easily plugged into existing CNNs to form IAUnet. The experiments show that IAUnet performs favorably against state-of-the-art on both image and video reID tasks and achieves compelling results on a general object categorization task. The source code is available at https://github.com/blue-blue272/ImgReID-IAnet.
['Xilin Chen', 'Ruibing Hou', 'Xinqian Gu', 'Bingpeng Ma', 'Shiguang Shan', 'Hong Chang']
2020-09-02
null
null
null
null
['object-categorization']
['computer-vision']
[-3.78392965e-01 -6.13535702e-01 -1.63785443e-01 -6.66615725e-01 -3.69698524e-01 -1.83238909e-01 6.29750431e-01 -1.92668848e-02 -6.44139469e-01 4.90915805e-01 6.42451882e-01 1.09829336e-01 6.82070339e-03 -5.83735943e-01 -7.05482185e-01 -5.70306063e-01 -3.76141258e-02 -4.31650691e-02 -6.90617934e-02 -1.63138598e-01 -2.40817472e-01 3.43377680e-01 -1.48449254e+00 9.26350877e-02 6.07077777e-01 1.03442919e+00 -8.71190988e-03 3.10619354e-01 -2.22149193e-02 5.26256859e-01 -4.43029821e-01 -5.39237857e-01 3.21497023e-01 -1.90769061e-01 -6.21590793e-01 -2.99375206e-01 7.77603984e-01 -6.70656502e-01 -7.47084260e-01 8.87883246e-01 8.77434790e-01 5.16367733e-01 2.84613252e-01 -1.34666634e+00 -8.46061885e-01 4.46548939e-01 -8.55708361e-01 4.99070615e-01 4.27319676e-01 3.58796895e-01 6.28817916e-01 -8.96174371e-01 4.32549536e-01 1.48386955e+00 8.17016780e-01 7.91778386e-01 -9.08709526e-01 -1.32053995e+00 7.04186082e-01 4.47141051e-01 -1.52832556e+00 -4.15946126e-01 6.41625285e-01 -3.08921725e-01 9.46770847e-01 9.17922035e-02 7.48939395e-01 1.32430100e+00 -5.92046604e-02 1.07543552e+00 7.14116633e-01 5.28648756e-02 -2.74215847e-01 -4.16290760e-01 3.31099719e-01 6.23861432e-01 1.37054384e-01 3.82830322e-01 -7.75028408e-01 8.42082575e-02 8.36350441e-01 5.35104036e-01 -1.65180057e-01 -3.36914808e-02 -1.10180891e+00 5.31733334e-01 9.92456198e-01 2.22471654e-01 -1.78473964e-01 4.20797408e-01 6.36597574e-01 -3.72507386e-02 4.85205203e-01 -1.35455534e-01 -3.95417511e-01 -6.89591169e-02 -6.34466469e-01 5.35810590e-01 2.29714602e-01 1.17135739e+00 7.47247934e-01 -7.99664780e-02 -7.44020283e-01 9.26780045e-01 1.94949925e-01 4.02033001e-01 3.09516668e-01 -7.19260633e-01 4.92089987e-01 6.66715682e-01 4.14856859e-02 -8.87944639e-01 -6.12503648e-01 -5.37087679e-01 -9.54533577e-01 -2.72680879e-01 3.61632973e-01 -1.88661501e-01 -1.00267363e+00 2.09741974e+00 6.12448633e-01 5.47604322e-01 -3.98694009e-01 1.19745290e+00 1.32876515e+00 2.60002285e-01 6.36192501e-01 2.13647708e-01 1.58239186e+00 -1.14814365e+00 -6.42745972e-01 -2.10518718e-01 5.99656999e-01 -5.44674873e-01 7.62912750e-01 -2.82718390e-01 -8.99859250e-01 -9.83881414e-01 -7.71415055e-01 -4.37846959e-01 -4.88234997e-01 4.08987403e-01 6.55608416e-01 3.91564816e-01 -1.02499235e+00 3.09416711e-01 -6.95011318e-01 -4.65599477e-01 5.38539767e-01 5.12965620e-01 -6.62052453e-01 -3.97569984e-01 -1.25848091e+00 6.08095706e-01 2.65095830e-01 4.26696688e-01 -6.86251044e-01 -7.01724172e-01 -1.14575255e+00 1.20576613e-01 2.75322765e-01 -1.07735181e+00 1.23684406e+00 -8.48513842e-01 -1.14424050e+00 7.02121854e-01 -5.92503250e-01 -1.99264899e-01 6.55862391e-01 -3.77963573e-01 -4.11799580e-01 -4.04357053e-02 2.91392565e-01 1.15096641e+00 7.36563444e-01 -9.56070960e-01 -7.86184907e-01 -4.71521974e-01 1.08903393e-01 1.71721578e-01 -4.14540827e-01 2.74784952e-01 -1.07535923e+00 -1.04691875e+00 -3.03782612e-01 -8.98549139e-01 5.06518297e-02 8.83195847e-02 -4.47223574e-01 -5.37985086e-01 7.79095888e-01 -7.03982949e-01 1.09300506e+00 -2.26748538e+00 -5.24383364e-03 9.75501817e-03 3.56493324e-01 3.13785940e-01 -2.86164075e-01 1.26917958e-01 -3.29778075e-01 -3.14287283e-02 1.75461516e-01 -9.49425697e-01 5.53913414e-02 -7.34867109e-03 2.43651852e-01 5.46364129e-01 -5.67858666e-02 1.32487202e+00 -9.29730833e-01 -4.48241264e-01 3.72124881e-01 8.93401146e-01 -4.84725684e-01 1.55413136e-01 3.44242603e-01 7.22106934e-01 -4.38979357e-01 7.81272352e-01 9.07056689e-01 -7.04711080e-02 -1.65617853e-01 -4.44493264e-01 -1.26761153e-01 1.88438687e-02 -9.81345594e-01 1.90789258e+00 -1.67542577e-01 4.76857543e-01 -4.21648845e-02 -6.18961513e-01 6.36671782e-01 6.43571168e-02 4.41433549e-01 -1.09091711e+00 4.87487882e-01 -1.54539585e-01 -1.61017865e-01 -3.15585583e-01 6.21455193e-01 2.96597332e-01 -9.26737338e-02 1.13935798e-01 2.29545593e-01 9.63370442e-01 1.55451193e-01 3.29413921e-01 7.98570037e-01 1.91595420e-01 6.01072386e-02 -1.77155241e-01 5.22620857e-01 -3.72773975e-01 9.34837520e-01 7.94909239e-01 -5.78420222e-01 6.53724551e-01 -1.58911720e-02 -7.51896799e-01 -7.94299901e-01 -8.36514175e-01 -2.69017909e-02 1.35408258e+00 5.57210624e-01 -6.64445460e-01 -7.47968137e-01 -6.46499634e-01 2.46257693e-01 8.53933021e-02 -1.09728670e+00 -1.37216434e-01 -7.42745399e-01 -5.79125404e-01 5.33876777e-01 1.20745814e+00 1.07096732e+00 -8.78738940e-01 -1.69830784e-01 1.30806968e-01 -3.47374976e-01 -1.10299349e+00 -1.16948640e+00 -3.96015644e-01 -4.74117339e-01 -1.06078708e+00 -9.71032679e-01 -6.56094790e-01 5.66924691e-01 7.76121318e-01 7.48819232e-01 2.97822624e-01 -4.63503689e-01 5.85076869e-01 -4.53837782e-01 -4.05108750e-01 6.28798306e-01 3.56336869e-02 1.82290226e-01 3.09227705e-02 6.68149352e-01 -4.33233410e-01 -1.11824071e+00 5.91584980e-01 -5.51181614e-01 1.47231549e-01 4.40129906e-01 9.71510053e-01 5.05694151e-01 -2.85052866e-01 5.02704740e-01 -4.16347355e-01 2.89670050e-01 -4.81864452e-01 -1.79468170e-01 4.94503900e-02 -2.13722974e-01 -4.27330375e-01 2.94711679e-01 -6.07011795e-01 -1.29231870e+00 -7.91031644e-02 -1.87834829e-01 -5.97657859e-01 -3.16550523e-01 3.84092689e-01 -3.59989107e-01 -4.38301824e-02 2.94516057e-01 1.60655826e-01 -1.57699451e-01 -5.23314059e-01 1.91645399e-01 3.46153200e-01 9.14665043e-01 -6.84704483e-01 6.61874235e-01 5.23794889e-01 -3.14897925e-01 -4.59029406e-01 -6.87380731e-01 -8.29261422e-01 -6.68288350e-01 -2.98853517e-01 1.01128185e+00 -1.37840438e+00 -9.13305163e-01 8.83805037e-01 -1.11270499e+00 -3.67020339e-01 1.42954871e-01 4.18606907e-01 -4.42845114e-02 3.44628215e-01 -7.93196619e-01 -5.25606453e-01 -6.02138519e-01 -1.06293082e+00 1.15513790e+00 7.70421922e-01 -9.61537510e-02 -7.17797995e-01 -3.62812698e-01 4.63891834e-01 4.30237472e-01 1.38281643e-01 3.88882667e-01 -3.29238564e-01 -5.30692101e-01 -4.03950840e-01 -7.35967159e-01 -1.52561748e-02 7.21012875e-02 -3.64733338e-01 -1.04195833e+00 -5.48978388e-01 -6.61017835e-01 -1.23907901e-01 1.04012406e+00 5.54533660e-01 1.36608160e+00 -1.31061375e-01 -5.75934231e-01 9.08959389e-01 9.63808954e-01 3.02478243e-02 5.58791161e-01 5.31378686e-01 1.17830920e+00 3.78571451e-01 5.13448417e-01 3.66409153e-01 9.11954343e-01 9.48242486e-01 1.12585790e-01 -2.49599829e-01 -4.16152596e-01 -3.47884685e-01 1.67900264e-01 2.41760761e-01 -2.90358603e-01 -1.00987017e-01 -7.58357882e-01 5.76836228e-01 -2.18107486e+00 -1.00444627e+00 3.34804803e-02 2.00259519e+00 5.25606453e-01 -2.15049788e-01 3.68908197e-01 -4.00496691e-01 9.33442652e-01 1.43593386e-01 -6.93983853e-01 1.68899864e-01 -6.27787262e-02 -2.49308929e-01 4.83633548e-01 2.29011863e-01 -1.39147651e+00 9.60215569e-01 5.32504940e+00 7.75205731e-01 -8.98669183e-01 3.90078634e-01 5.54265141e-01 -3.00214350e-01 1.90813571e-01 -5.00946045e-01 -1.06157351e+00 6.44705355e-01 4.90005761e-01 7.05614090e-02 1.31137833e-01 8.90245557e-01 2.75621265e-01 -3.31930667e-02 -1.18198633e+00 1.30893898e+00 1.03932709e-01 -9.97523963e-01 1.30662657e-02 -6.16435185e-02 5.45569539e-01 -6.55654743e-02 2.22335264e-01 4.88601625e-01 5.78031987e-02 -1.03648674e+00 8.29759717e-01 4.51473504e-01 8.76405835e-01 -9.18192208e-01 9.60620582e-01 -1.55497584e-02 -1.79763889e+00 -3.34813058e-01 -3.03163648e-01 -2.02955604e-02 1.29787624e-01 1.65094301e-01 -1.70164451e-01 6.89525723e-01 1.22039247e+00 1.00680578e+00 -7.30856717e-01 1.32535613e+00 -7.38910139e-02 1.85303658e-01 -2.13706225e-01 2.89925039e-01 2.53598690e-01 2.74470389e-01 3.24128091e-01 1.53994179e+00 1.14864521e-01 3.61161649e-01 3.57357204e-01 6.22982204e-01 -2.09853008e-01 -2.65977345e-02 -2.27743655e-01 5.83785832e-01 5.24165034e-01 1.18969536e+00 -4.14298326e-01 -4.02050376e-01 -6.66688800e-01 1.02292836e+00 3.81819993e-01 5.68533421e-01 -9.81829941e-01 -1.88402638e-01 1.06774104e+00 -3.57690677e-02 2.32712165e-01 -1.33182332e-01 -1.45680830e-01 -1.36703253e+00 1.82832852e-01 -6.33455276e-01 6.39284730e-01 -6.44414663e-01 -1.59700942e+00 4.08300966e-01 2.51338273e-01 -1.23282659e+00 1.76337540e-01 -4.89673287e-01 -8.13039839e-01 1.13572228e+00 -1.60968852e+00 -1.70660400e+00 -8.85947824e-01 9.88208652e-01 6.16465926e-01 -4.40576151e-02 6.23285353e-01 6.54730320e-01 -1.10414481e+00 1.23624229e+00 -3.77766341e-01 7.64575720e-01 1.03659987e+00 -9.65201199e-01 5.54241061e-01 1.03931880e+00 -4.68518525e-01 1.08154190e+00 3.95143926e-01 -7.89713800e-01 -1.18230236e+00 -1.42377996e+00 7.52236307e-01 -3.77935469e-01 2.13198483e-01 -3.82159680e-01 -7.97711730e-01 8.43270302e-01 4.50158864e-02 4.88022596e-01 7.59637892e-01 3.49533141e-01 -8.32747459e-01 -2.20907345e-01 -9.37838197e-01 5.14510274e-01 1.50447929e+00 -6.16969466e-01 -3.89785081e-01 4.91958261e-02 5.74547470e-01 -6.84733510e-01 -6.37007654e-01 3.39458525e-01 9.50835824e-01 -8.15735400e-01 1.28694916e+00 -4.95303094e-01 3.19807529e-01 -4.58282381e-01 1.10227562e-01 -1.16331744e+00 -6.73135102e-01 -2.06144512e-01 -1.95518956e-01 1.42139649e+00 3.20116207e-02 -5.83010435e-01 6.15940392e-01 1.02106977e+00 -2.04154089e-01 -6.46665633e-01 -9.27902102e-01 -8.47773194e-01 9.80553310e-03 -3.40685785e-01 9.11606371e-01 9.89009202e-01 -6.53789863e-02 5.62879741e-02 -6.05396628e-01 7.79562667e-02 6.41530812e-01 -6.05638474e-02 1.02376497e+00 -1.01044297e+00 -1.42526804e-02 -5.79225600e-01 -5.43784082e-01 -1.26783943e+00 2.29162171e-01 -7.52708316e-01 -1.59091204e-01 -1.42294228e+00 6.25611663e-01 -4.63907152e-01 -5.17507434e-01 8.14287961e-01 -7.07274675e-01 4.10352707e-01 4.44596738e-01 2.02921197e-01 -7.65577555e-01 6.47112608e-01 1.13323450e+00 -3.99208337e-01 -1.38899505e-01 -2.24521548e-01 -7.91248798e-01 5.14700234e-01 7.15690076e-01 -1.83773547e-01 -1.86898068e-01 -7.12799609e-01 -3.36037576e-01 -4.13774639e-01 7.74728656e-01 -1.05436206e+00 5.50071180e-01 1.42317802e-01 1.14146471e+00 -6.77122891e-01 4.23411727e-01 -5.24824202e-01 8.64146426e-02 6.05954528e-02 -1.79365754e-01 1.74387977e-01 3.21436256e-01 6.64711654e-01 -2.59388953e-01 4.06960547e-01 5.74232280e-01 -8.79588649e-02 -1.06289232e+00 9.37762022e-01 9.33715180e-02 -3.28677028e-01 8.54859054e-01 -2.39343092e-01 -4.85483259e-01 -3.63627642e-01 -6.06490970e-01 6.71998084e-01 3.14284384e-01 8.22384536e-01 6.22492015e-01 -1.56640124e+00 -5.91654360e-01 7.34497532e-02 3.44741255e-01 -3.77087109e-02 1.01817131e+00 9.21984434e-01 -7.54367113e-02 3.98964435e-01 -3.20678115e-01 -5.77795744e-01 -1.38012588e+00 6.37408376e-01 4.73539859e-01 -9.35788825e-03 -7.75412202e-01 1.16816509e+00 7.42275536e-01 -3.04075211e-01 5.68906188e-01 -1.58475250e-01 -3.17183226e-01 -2.14198045e-02 1.08030951e+00 3.31252664e-01 -2.33656719e-01 -1.08332849e+00 -4.80803519e-01 6.19413555e-01 -3.31881642e-01 3.24811995e-01 1.05692518e+00 -3.65868419e-01 7.35954791e-02 -2.23557100e-01 1.36507058e+00 -3.41327369e-01 -1.47840786e+00 -4.98090386e-01 -4.55647439e-01 -6.99020386e-01 -6.35260344e-02 -8.28547060e-01 -1.40335059e+00 8.01769614e-01 7.73493528e-01 -4.67616290e-01 1.14218175e+00 -9.52020437e-02 1.10138810e+00 1.20359831e-01 4.38781887e-01 -9.22039866e-01 1.89414062e-02 5.03047943e-01 7.72481382e-01 -1.26384509e+00 -1.27488643e-01 -5.52972198e-01 -5.75072110e-01 9.01537478e-01 1.15514493e+00 5.89027740e-02 6.29432142e-01 -3.57166789e-02 1.90597940e-02 -2.54056491e-02 -2.32627660e-01 -5.41493833e-01 5.00048935e-01 6.66006625e-01 4.03080165e-01 1.98715419e-01 -1.97108805e-01 9.07124579e-01 -9.13809240e-02 5.99056249e-03 -1.51369065e-01 8.77029717e-01 6.69680312e-02 -1.00977695e+00 -2.70870209e-01 3.43036503e-01 -4.52154845e-01 -3.86484638e-02 -1.77200094e-01 8.53695929e-01 6.71300054e-01 8.43475819e-01 5.04250079e-02 -6.39369607e-01 4.59701627e-01 -1.49713427e-01 3.17834765e-01 -3.28615278e-01 -7.28190005e-01 1.36134326e-01 -4.43894118e-02 -8.38129938e-01 -5.37745655e-01 -7.09113836e-01 -1.07606411e+00 -6.70268774e-01 -1.39787555e-01 -1.88502774e-01 3.74240041e-01 1.00978446e+00 5.48654437e-01 4.79493082e-01 1.95105255e-01 -1.31734133e+00 2.02284485e-01 -1.06194842e+00 -3.71508121e-01 5.14808536e-01 4.63981479e-01 -1.13863897e+00 1.22110926e-01 -1.08420596e-01]
[14.734582901000977, 0.939357578754425]
2952de44-e827-464d-93d2-23f0daf14f6d
towards-human-cognition-level-based
2211.00103
null
https://arxiv.org/abs/2211.00103v1
https://arxiv.org/pdf/2211.00103v1.pdf
Towards Human Cognition Level-based Experiment Design for Counterfactual Explanations (XAI)
Explainable Artificial Intelligence (XAI) has recently gained a swell of interest, as many Artificial Intelligence (AI) practitioners and developers are compelled to rationalize how such AI-based systems work. Decades back, most XAI systems were developed as knowledge-based or expert systems. These systems assumed reasoning for the technical description of an explanation, with little regard for the user's cognitive capabilities. The emphasis of XAI research appears to have turned to a more pragmatic explanation approach for better understanding. An extensive area where cognitive science research may substantially influence XAI advancements is evaluating user knowledge and feedback, which are essential for XAI system evaluation. To this end, we propose a framework to experiment with generating and evaluating the explanations on the grounds of different cognitive levels of understanding. In this regard, we adopt Bloom's taxonomy, a widely accepted model for assessing the user's cognitive capability. We utilize the counterfactual explanations as an explanation-providing medium encompassed with user feedback to validate the levels of understanding about the explanation at each cognitive level and improvise the explanation generation methods accordingly.
['Alessandro Bogliolo', 'Muhammad Yaseen Khan', 'Muhammad Suffian']
2022-10-31
null
null
null
null
['explanation-generation']
['natural-language-processing']
[ 1.72183946e-01 8.38863254e-01 -2.25957766e-01 -5.24964750e-01 1.47157028e-01 -6.25861287e-01 8.24088693e-01 2.75441766e-01 9.60276946e-02 5.67738116e-01 5.02379358e-01 -1.08799064e+00 -7.25602210e-01 -6.45778716e-01 -3.71573687e-01 4.25656401e-02 3.37541431e-01 4.47962403e-01 -2.57922322e-01 -3.22454453e-01 9.34710026e-01 2.87283540e-01 -2.06668353e+00 5.49554050e-01 1.37383008e+00 4.14157450e-01 7.88466632e-02 5.76897264e-01 -4.52701271e-01 1.02469063e+00 -8.61052811e-01 -6.51457787e-01 -1.58199906e-01 -6.35875583e-01 -1.23803520e+00 -1.27743661e-01 -2.01869868e-02 -2.13961214e-01 1.16257705e-01 8.57576013e-01 -1.25625834e-01 -3.22226365e-03 5.42863965e-01 -1.75634456e+00 -1.21321476e+00 9.45435762e-01 2.18819395e-01 7.79952630e-02 8.16968858e-01 2.90791571e-01 9.05393660e-01 -5.83806157e-01 4.53509808e-01 1.26173687e+00 2.90332705e-01 5.73352277e-01 -8.53240132e-01 -4.96876657e-01 3.50096285e-01 7.50941157e-01 -1.02337277e+00 -1.05976656e-01 8.16145062e-01 -5.12044191e-01 8.47289503e-01 7.45687962e-01 9.21738267e-01 7.71201193e-01 2.95323551e-01 7.35348463e-01 1.27611876e+00 -9.75156248e-01 5.29348731e-01 6.90219223e-01 7.04461277e-01 3.83172870e-01 8.58596325e-01 3.78841192e-01 -6.55935585e-01 9.43073712e-04 4.34160352e-01 3.43294479e-02 -2.38742203e-01 -1.33232430e-01 -9.59465742e-01 8.52962911e-01 2.93703049e-01 6.22084022e-01 -7.38398433e-01 -2.42041439e-01 6.47905841e-02 3.85017097e-01 9.64799449e-02 1.17681205e+00 -3.21771145e-01 -3.02105963e-01 -3.73173296e-01 5.89594781e-01 8.80605221e-01 6.78696811e-01 5.30491531e-01 1.34638980e-01 -2.45261431e-01 1.92780927e-01 7.94741750e-01 2.15463191e-01 7.63647437e-01 -1.15594459e+00 -1.43704087e-01 1.33193088e+00 3.26872021e-01 -8.69259715e-01 -2.38391489e-01 -5.33784509e-01 -1.66324258e-01 8.00892353e-01 3.27158689e-01 4.54237275e-02 -7.01428354e-01 1.57553983e+00 2.18615960e-02 -2.86306292e-01 4.02759135e-01 9.80371773e-01 6.77258551e-01 2.84129202e-01 1.99775204e-01 -1.13023311e-01 1.36078238e+00 -7.16337979e-01 -9.87641335e-01 -2.80697316e-01 6.11935318e-01 -5.59212923e-01 1.45674574e+00 4.52725083e-01 -1.06330132e+00 -5.85068643e-01 -1.25092924e+00 1.90602705e-01 -5.43753982e-01 -2.01523975e-01 1.13554943e+00 1.10281825e+00 -9.69402671e-01 4.15700853e-01 -3.19374412e-01 -4.32684660e-01 -1.11699123e-02 6.98774904e-02 2.66718388e-01 4.34379373e-03 -1.22274685e+00 1.18716884e+00 5.95128238e-01 1.33588901e-02 -4.10765588e-01 -6.33874595e-01 -6.46371126e-01 3.95295769e-01 3.03103328e-01 -1.02377713e+00 1.47382534e+00 -1.30387723e+00 -1.37359393e+00 3.38540107e-01 -3.32880728e-02 -4.58290279e-01 2.77839214e-01 1.15765510e-02 -6.06609046e-01 -1.85740963e-01 2.81687230e-01 4.25554186e-01 2.06911370e-01 -1.59178507e+00 -8.05718064e-01 -4.72151339e-01 8.44439447e-01 3.29457462e-01 -6.35080412e-02 -2.08800703e-01 2.49908939e-01 -1.98386058e-01 1.49591863e-01 -8.59178066e-01 -5.17459288e-02 -3.27953935e-01 -1.84262350e-01 -4.55585808e-01 4.44766551e-01 -2.55015701e-01 1.51028466e+00 -1.75208211e+00 -4.36622232e-01 3.73624951e-01 4.28076297e-01 1.72739357e-01 2.20745772e-01 5.74067712e-01 -1.77710846e-01 5.58655977e-01 3.15709591e-01 3.38866383e-01 3.69264960e-01 1.52226806e-01 -2.70027071e-01 -4.77700561e-01 -2.27924868e-01 6.82411432e-01 -9.20962870e-01 -1.22338995e-01 3.17793250e-01 2.18371779e-01 -7.03500926e-01 4.19662058e-01 -2.86783963e-01 2.32384071e-01 -5.15718222e-01 5.35906315e-01 2.12052464e-01 -4.84348923e-01 1.24158598e-01 6.77276999e-02 -5.82899988e-01 4.90207583e-01 -7.79760420e-01 1.40076005e+00 -2.51076043e-01 6.94342434e-01 -6.36773527e-01 -5.10290384e-01 7.58340240e-01 5.58575988e-01 -2.70051435e-02 -5.42004764e-01 2.11116090e-01 9.14903283e-02 6.42681360e-01 -7.23433673e-01 3.08752000e-01 -2.28683934e-01 2.40614444e-01 7.68645585e-01 -4.10595775e-01 -1.05674848e-01 7.57736042e-02 3.63137960e-01 9.24053192e-01 -1.96858421e-02 7.31109500e-01 -3.43970656e-01 4.73491281e-01 5.71133077e-01 1.08989902e-01 1.11200726e+00 -5.64916134e-02 1.05520412e-01 1.45659283e-01 -7.93603122e-01 -6.90387785e-01 -7.58299887e-01 1.17989264e-01 8.25632393e-01 3.05385143e-01 -4.17002320e-01 -1.01893497e+00 -4.67798859e-01 -3.61666471e-01 1.94790065e+00 -7.44114995e-01 -3.85230124e-01 1.42634347e-01 -1.33221641e-01 1.56328470e-01 2.92219162e-01 5.73805869e-01 -1.43109846e+00 -1.40086687e+00 9.21834260e-02 -3.84080559e-01 -2.35797629e-01 3.04511964e-01 -9.30402130e-02 -8.39217663e-01 -1.14909875e+00 8.18319097e-02 -8.94192830e-02 6.06170774e-01 5.83637059e-01 1.12213063e+00 7.66173899e-01 1.84183493e-02 8.46408308e-01 -7.33087301e-01 -1.14043915e+00 -6.03781462e-01 -2.67547190e-01 -9.22904983e-02 -6.09463692e-01 7.70029962e-01 -4.84651953e-01 -7.17528999e-01 3.80893886e-01 -1.07914960e+00 4.50052530e-01 7.36809611e-01 3.48263979e-01 -7.30948523e-02 2.44956240e-01 4.65881556e-01 -1.03008139e+00 1.36151040e+00 -5.13550222e-01 -3.52779143e-02 6.33157194e-01 -1.51895809e+00 3.17536563e-01 3.33205611e-01 -1.89627320e-01 -1.61509144e+00 -7.18110085e-01 1.80233970e-01 1.58977166e-01 -5.11056125e-01 9.41225529e-01 -2.18531236e-01 5.87308742e-02 1.10024965e+00 2.28260122e-02 -2.00568009e-02 -1.80466194e-02 6.35053217e-01 5.31016409e-01 5.69663763e-01 -7.46490896e-01 6.66283011e-01 4.95443046e-02 -5.24055302e-01 -1.59608692e-01 -8.65531623e-01 6.32963926e-02 -1.87771857e-01 -7.62115061e-01 7.05419540e-01 -1.62611201e-01 -1.14897954e+00 -3.56230319e-01 -1.22769749e+00 -1.55667141e-01 -2.24545971e-01 6.38695836e-01 -5.84749758e-01 7.33195411e-05 1.80194974e-01 -1.17013645e+00 -2.29682922e-01 -1.09295797e+00 2.75354266e-01 5.10172188e-01 -9.49962854e-01 -8.69854629e-01 2.30573788e-02 6.17604554e-01 5.88219404e-01 3.22460779e-03 1.51336300e+00 -9.22415435e-01 -5.16515851e-01 -2.37852812e-01 -3.58420089e-02 -2.17574194e-01 5.48700280e-02 -4.92687002e-02 -8.49325240e-01 3.05985153e-01 2.76435465e-01 2.77475994e-02 -1.69281885e-01 2.84608096e-01 9.10480976e-01 -6.47019029e-01 -1.49336904e-01 -8.29703659e-02 1.27822351e+00 9.72333431e-01 7.37548351e-01 7.73101747e-01 7.96640813e-02 8.90387356e-01 8.65708828e-01 4.35224950e-01 4.60948676e-01 4.20349121e-01 3.92617792e-01 1.84684411e-01 -4.14827354e-02 -2.68270671e-01 -4.45708185e-02 2.15781152e-01 -5.21818221e-01 -2.43912861e-01 -1.15791261e+00 2.69464076e-01 -1.95884371e+00 -1.16040254e+00 -1.13910556e-01 2.11189342e+00 2.49690816e-01 3.63705724e-01 -2.34141037e-01 4.09852982e-01 3.95558804e-01 -5.45395255e-01 -5.59698880e-01 -6.87718928e-01 2.46519655e-01 -2.74757445e-01 -2.87641555e-01 7.13147461e-01 -4.77069728e-02 6.37473762e-01 6.65340805e+00 -8.20943862e-02 -7.88420558e-01 -1.83845073e-01 4.70052958e-01 2.90725887e-01 -1.03136158e+00 3.42462629e-01 -2.20868051e-01 3.84946704e-01 9.75625277e-01 -9.26928580e-01 4.56853062e-01 1.04609656e+00 4.00854528e-01 -1.98471397e-01 -1.26061308e+00 6.15855455e-01 -5.86826392e-02 -1.39914739e+00 4.69993174e-01 2.93558035e-02 3.83285731e-01 -9.22004938e-01 1.86939806e-01 5.46765327e-01 3.36381763e-01 -1.00016415e+00 9.78192866e-01 6.08747005e-01 1.50142819e-01 -6.83960974e-01 9.49262559e-01 4.22218978e-01 -5.87101698e-01 -2.00655073e-01 -2.26862609e-01 -9.60092723e-01 -7.41564259e-02 -7.39163980e-02 -1.24312842e+00 4.12246555e-01 5.28895855e-01 -2.40465641e-01 -5.88605940e-01 7.81337798e-01 -5.99658310e-01 6.86446548e-01 2.00294361e-01 -3.43057036e-01 3.11587155e-02 -2.24364445e-01 3.03933650e-01 5.91328144e-01 4.35511082e-01 8.06508243e-01 -1.06811054e-01 1.12065518e+00 4.95733172e-01 6.12877831e-02 -7.67282903e-01 -5.74475303e-02 9.55370128e-01 6.57008946e-01 -7.00147033e-01 -7.47256219e-01 -4.59796876e-01 6.62837565e-01 -2.09189858e-02 3.98827434e-01 -7.22579837e-01 -5.14327735e-02 6.60368562e-01 2.51312882e-01 -4.94210303e-01 2.44259715e-01 -1.12659669e+00 -8.41661572e-01 -2.74570018e-01 -1.09385896e+00 2.59302497e-01 -1.51577985e+00 -7.90713310e-01 7.25436747e-01 4.56321061e-01 -9.92934704e-01 -6.10998154e-01 -4.24989611e-01 -8.54372859e-01 8.69720399e-01 -1.04368579e+00 -9.44823682e-01 -7.30199099e-01 1.38929039e-01 6.34067357e-01 -7.55237043e-02 9.98639643e-01 -2.57570505e-01 -2.58766502e-01 2.02161893e-01 -5.74496806e-01 -6.18538678e-01 3.67910385e-01 -1.22263265e+00 1.58965334e-01 8.29021335e-01 1.49049744e-01 1.14060247e+00 1.32819378e+00 -7.77005613e-01 -1.19810903e+00 -3.63183200e-01 1.03728163e+00 -6.08424485e-01 2.96876073e-01 3.28110278e-01 -1.09417796e+00 8.90637338e-01 5.45085371e-01 -1.03410423e+00 1.16802227e+00 4.24743474e-01 1.04233846e-02 3.00007790e-01 -1.06239879e+00 9.70882297e-01 1.06651294e+00 -8.44922587e-02 -1.20487463e+00 1.26775252e-02 6.30553544e-01 -1.18932240e-02 -5.72187901e-01 2.85388649e-01 6.67528749e-01 -1.39226568e+00 7.82823443e-01 -7.17394233e-01 5.27776480e-01 -5.46190739e-01 9.07354653e-02 -1.30322170e+00 -7.27353096e-01 -4.38912481e-01 2.01004803e-01 9.97897387e-01 7.22281218e-01 -7.02724338e-01 7.36859679e-01 1.61683047e+00 -4.83676434e-01 -4.83148932e-01 -3.03589433e-01 -3.92250389e-01 -4.52223182e-01 -7.31116951e-01 1.13228369e+00 8.93278956e-01 7.19200373e-01 4.71385628e-01 -2.85869278e-03 3.39465022e-01 3.66885304e-01 2.37791657e-01 9.58833277e-01 -1.67661941e+00 -1.46293253e-01 -6.51607931e-01 -2.71294266e-01 -5.85980237e-01 -4.78857309e-02 -6.60193264e-01 -1.79102078e-01 -1.79774952e+00 2.54551262e-01 9.78327319e-02 -1.47270426e-01 5.25845051e-01 -4.61046070e-01 -3.57059002e-01 2.91545063e-01 4.84719038e-01 -2.59435266e-01 1.92775264e-01 1.00321424e+00 7.76005760e-02 -4.30844516e-01 -6.15036599e-02 -1.72944725e+00 9.51868415e-01 8.82641494e-01 -2.48611420e-01 -1.15603030e+00 -3.33449274e-01 4.99225527e-01 1.31778330e-01 2.73715228e-01 -1.12187231e+00 3.07297647e-01 -4.91898596e-01 2.79077947e-01 -3.32037091e-01 -8.22949335e-02 -1.00282753e+00 6.94004476e-01 5.34046829e-01 -6.76461101e-01 2.67100275e-01 2.13392779e-01 1.99698046e-01 -2.14983150e-01 -5.82702279e-01 2.28864983e-01 -2.39118505e-02 -8.25690925e-01 -2.60914654e-01 -7.64848709e-01 -5.14837384e-01 9.72882807e-01 -8.21273446e-01 -2.59257406e-01 -6.46986604e-01 -5.89058697e-01 -5.91897853e-02 5.15531123e-01 4.44346726e-01 8.04336905e-01 -1.15120590e+00 -3.66165459e-01 5.49710579e-02 4.45919335e-01 -5.89526355e-01 2.76997924e-01 3.06468427e-01 -4.46831763e-01 7.76796341e-01 -5.86528659e-01 -6.09652586e-02 -1.08276284e+00 7.96674967e-01 1.72492694e-02 7.25695863e-02 -4.44552422e-01 4.81420457e-01 4.04022157e-01 -5.71902506e-02 1.11642204e-01 -2.54917324e-01 -6.15356624e-01 -4.82882291e-01 7.77733505e-01 4.26503509e-01 -3.16726506e-01 8.11895207e-02 -2.43098855e-01 -1.31374195e-01 -1.09778970e-01 -4.09186065e-01 8.79267454e-01 -2.96843797e-01 9.08021182e-02 6.23893797e-01 1.52521431e-01 -1.92568392e-01 -6.80495679e-01 2.15887919e-01 3.11459064e-01 -7.42567122e-01 -1.51462391e-01 -1.56057084e+00 -6.37257919e-02 6.18461311e-01 4.79359061e-01 6.64784074e-01 9.76951659e-01 4.78857243e-03 -2.05646843e-01 4.97009695e-01 3.79139990e-01 -7.92946756e-01 -4.81795333e-02 4.02375124e-02 1.19637847e+00 -1.06532359e+00 -5.48641868e-02 -4.91725504e-01 -8.34811091e-01 1.31630969e+00 8.90825629e-01 5.14596641e-01 2.79967934e-01 -1.83441952e-01 4.58592921e-01 -6.17929339e-01 -9.59515095e-01 1.04220688e-01 3.77739549e-01 6.64730489e-01 9.58631754e-01 2.59910405e-01 -7.92952538e-01 1.00433314e+00 -7.09800601e-01 1.82264715e-01 7.14054883e-01 6.64074361e-01 -7.03875542e-01 -9.85026181e-01 -6.36029780e-01 3.47088277e-01 5.86951524e-02 1.04031585e-01 -8.86813700e-01 1.03771949e+00 4.74615134e-02 1.44041455e+00 -3.13443094e-01 -3.24827194e-01 4.08306032e-01 3.06896389e-01 1.76545575e-01 -5.73180199e-01 -6.36285543e-01 -5.19905210e-01 1.52970046e-01 -3.61290008e-01 -3.19008917e-01 -4.62114871e-01 -1.32835114e+00 -5.64378500e-01 -2.17414409e-01 7.36614048e-01 8.26674461e-01 1.24036169e+00 3.55082273e-01 7.54962444e-01 1.49952129e-01 -2.93972373e-01 -2.93478638e-01 -9.63371098e-01 -1.10394634e-01 5.86378574e-01 -3.41520995e-01 -7.08983004e-01 -3.32556218e-01 -1.24095622e-02]
[8.9550142288208, 6.122028350830078]
bfe2e560-1c36-4022-b1c0-bfc2845360fc
online-action-detection-in-streaming-videos
2010.03016
null
https://arxiv.org/abs/2010.03016v1
https://arxiv.org/pdf/2010.03016v1.pdf
Online Action Detection in Streaming Videos with Time Buffers
We formulate the problem of online temporal action detection in live streaming videos, acknowledging one important property of live streaming videos that there is normally a broadcast delay between the latest captured frame and the actual frame viewed by the audience. The standard setting of the online action detection task requires immediate prediction after a new frame is captured. We illustrate that its lack of consideration of the delay is imposing unnecessary constraints on the models and thus not suitable for this problem. We propose to adopt the problem setting that allows models to make use of the small `buffer time' incurred by the delay in live streaming videos. We design an action start and end detection framework for this online with buffer setting with two major components: flattened I3D and window-based suppression. Experiments on three standard temporal action detection benchmarks under the proposed setting demonstrate the effectiveness of the proposed framework. We show that by having a suitable problem setting for this problem with wide-applications, we can achieve much better detection accuracy than off-the-shelf online action detection models.
['Yuanjun Xiong', 'Meng Wang', 'Hao Chen', 'BoWen Zhang']
2020-10-06
null
null
null
null
['online-action-detection']
['computer-vision']
[ 7.50979841e-01 6.70704991e-02 -4.24079567e-01 -6.97904602e-02 -6.31448567e-01 -4.36296165e-01 3.83858651e-01 3.95507887e-02 -5.64776301e-01 1.63953424e-01 2.91622430e-01 -1.80788100e-01 1.50603084e-02 -3.27183872e-01 -8.51213574e-01 -6.57314837e-01 -5.89168251e-01 -9.98615697e-02 1.02751613e+00 5.05111702e-02 1.82600111e-01 3.21807206e-01 -1.59849560e+00 5.19559085e-01 2.00822219e-01 1.32705104e+00 2.41832346e-01 1.09193742e+00 1.75134122e-01 1.28018379e+00 -6.05863392e-01 7.73808220e-03 7.71824002e-01 -5.03324866e-01 -7.13899434e-01 7.40664423e-01 3.12438399e-01 -1.10365343e+00 -6.04059756e-01 7.02491701e-01 4.70123857e-01 2.81571686e-01 6.48749545e-02 -1.41421747e+00 1.14913456e-01 5.88984549e-01 -6.70199633e-01 9.06049013e-01 7.71845579e-01 1.13813929e-01 9.84471083e-01 -4.79518771e-01 6.82642341e-01 1.01905262e+00 4.48507756e-01 4.98133600e-01 -7.94199407e-01 -2.92704165e-01 4.71213728e-01 4.16113734e-01 -1.06138182e+00 -7.51743793e-01 4.56370920e-01 -2.47504413e-01 9.82145131e-01 4.23291624e-01 6.68248296e-01 7.67234623e-01 3.70917544e-02 1.19032526e+00 4.84658211e-01 -4.03956950e-01 3.21576446e-01 -4.28733259e-01 -5.42489067e-02 3.20092559e-01 -1.61200210e-01 -1.10702805e-01 -7.79286146e-01 -2.52328426e-01 8.00411642e-01 2.64519364e-01 -5.24480641e-01 -2.70707101e-01 -1.36431324e+00 4.90852684e-01 -2.15242445e-01 1.03913523e-01 -3.02988082e-01 3.29839081e-01 6.96630180e-01 5.39554656e-01 5.11622787e-01 -1.42646343e-01 -3.37036908e-01 -7.38271058e-01 -1.06800401e+00 1.81712300e-01 7.05214560e-01 1.18551159e+00 2.69045532e-01 -1.84173211e-01 -2.03570068e-01 4.05954540e-01 -1.33700863e-01 2.10327893e-01 2.93608397e-01 -1.59033704e+00 6.36406302e-01 3.38690609e-01 4.51006114e-01 -1.08046365e+00 -1.48097441e-01 1.67462230e-01 -3.01223278e-01 8.32309052e-02 6.09629869e-01 -1.29219070e-01 -3.56453359e-01 1.56477630e+00 6.45114183e-01 7.31272042e-01 -2.38178417e-01 8.20392489e-01 1.53669536e-01 7.53037632e-01 -1.74863562e-01 -8.03660631e-01 1.04391801e+00 -9.40229118e-01 -7.65128076e-01 -9.15583745e-02 8.29257846e-01 -5.66993237e-01 6.51887417e-01 5.32486141e-01 -1.23151875e+00 -3.17589670e-01 -9.13184762e-01 1.22615904e-01 3.06819350e-01 -1.63282186e-01 4.49771643e-01 4.83031154e-01 -8.88969004e-01 5.69003165e-01 -1.10069835e+00 -5.11584759e-01 1.61244020e-01 2.93778509e-01 -2.29986817e-01 7.38207921e-02 -7.43185937e-01 2.73252636e-01 2.03837082e-01 4.07721242e-03 -1.02129579e+00 -6.16400838e-01 -5.09084642e-01 -1.53796244e-02 1.06882000e+00 -2.36168519e-01 1.76524091e+00 -1.52153265e+00 -1.42141342e+00 6.44079149e-01 -2.29175061e-01 -6.98818922e-01 9.85853255e-01 -5.78710318e-01 -1.66240871e-01 8.08540225e-01 2.16404162e-02 2.95097411e-01 1.03898442e+00 -4.66796666e-01 -1.08985043e+00 -8.46181810e-02 7.43751884e-01 3.12020957e-01 -4.74947602e-01 5.96816063e-01 -5.66590846e-01 -5.99150658e-01 -1.16989993e-01 -8.65560830e-01 -1.67473763e-01 4.08634663e-01 -2.88425256e-02 -8.62992927e-02 9.72886980e-01 -4.14650232e-01 1.44466496e+00 -2.37063408e+00 -2.16838181e-01 -1.25572488e-01 3.64028029e-02 3.34138572e-01 7.03508093e-04 5.95637500e-01 -4.65391912e-02 -1.02565363e-01 1.08609542e-01 -2.27543831e-01 -1.42862201e-01 1.91188842e-01 -3.81788373e-01 7.78482080e-01 -1.15501173e-01 3.21090400e-01 -1.07811046e+00 -6.26160979e-01 2.16574475e-01 9.84037071e-02 -8.12531412e-01 4.57809091e-01 -2.62129068e-01 2.01544523e-01 -4.97937083e-01 4.30299342e-01 3.86994570e-01 -3.89301687e-01 1.68423131e-01 2.30192348e-01 -2.58220613e-01 3.08941633e-01 -1.66366374e+00 1.60907006e+00 -1.51977435e-01 6.41744912e-01 3.64703625e-01 -9.24993992e-01 1.97347164e-01 5.17479658e-01 1.02713490e+00 -4.75143075e-01 1.83061883e-02 -2.38202214e-01 -9.18484479e-02 -7.14605272e-01 3.51300299e-01 2.25719467e-01 3.54297161e-01 6.74466789e-01 -2.43462190e-01 6.30243063e-01 4.31519270e-01 3.32999557e-01 1.70437872e+00 2.21405029e-01 4.36795205e-01 1.52490526e-01 4.15942013e-01 -3.33808035e-01 7.68920183e-01 1.01715255e+00 -6.25741601e-01 5.77006578e-01 7.59830952e-01 -3.64009798e-01 -7.61516511e-01 -6.05883181e-01 3.19066703e-01 1.59679735e+00 1.09630421e-01 -5.52493513e-01 -7.78671920e-01 -6.67337060e-01 -2.26979360e-01 6.14121705e-02 -5.40254235e-01 1.12680890e-01 -8.12045515e-01 -2.96410084e-01 2.37122640e-01 6.32603168e-01 3.33650768e-01 -9.21971262e-01 -1.31901562e+00 4.30324197e-01 -3.07573527e-01 -1.47599983e+00 -8.51503670e-01 -1.80677980e-01 -8.79527688e-01 -1.21107805e+00 -7.48499572e-01 -4.87793148e-01 3.41848910e-01 9.05865908e-01 9.11117911e-01 3.36939901e-01 -1.11624479e-01 8.41806710e-01 -6.95411325e-01 -7.00932294e-02 -3.67219716e-01 -3.50506186e-01 -5.16487360e-02 5.01021802e-01 -4.41977382e-02 -4.10984308e-01 -9.07222092e-01 5.15735984e-01 -1.25910938e+00 1.01931468e-01 2.42720798e-01 2.72148639e-01 4.17025536e-01 1.22092590e-01 3.68875414e-01 -5.63441157e-01 -6.74758926e-02 -4.59391534e-01 -4.69090044e-01 8.29017758e-02 -9.10516679e-02 -4.47700083e-01 4.66087699e-01 -6.90840125e-01 -8.59151542e-01 1.81843534e-01 4.99289148e-02 -4.24923599e-01 -3.84371765e-02 5.98390289e-02 -1.28189743e-01 -5.10696098e-02 2.03879490e-01 2.36908108e-01 -6.08164519e-02 -4.28635448e-01 -1.23623805e-02 5.92699766e-01 4.17834163e-01 -2.06449479e-01 2.35585764e-01 9.42269742e-01 -7.78310522e-02 -1.01728809e+00 -7.09231853e-01 -9.50421095e-01 -4.79525357e-01 -4.25944567e-01 6.88719511e-01 -9.77642536e-01 -9.35021758e-01 6.16664469e-01 -1.13019693e+00 -4.04321283e-01 -3.35552603e-01 3.78440231e-01 -8.65355551e-01 9.81367290e-01 -7.42588699e-01 -1.09302783e+00 -3.34549844e-02 -8.90507817e-01 1.09731364e+00 -9.90134254e-02 -3.79916169e-02 -7.54295409e-01 -1.00985775e-02 1.29260808e-01 1.26765698e-01 2.99124807e-01 2.18552887e-01 -6.21491730e-01 -6.74409866e-01 -2.74265617e-01 -6.20017350e-02 4.81067032e-01 7.57568255e-02 1.62252411e-01 -7.07727969e-01 -2.27103427e-01 2.09425211e-01 -8.02672505e-02 7.91778922e-01 5.37642300e-01 1.18750799e+00 -3.71418417e-01 -1.76035956e-01 4.30512697e-01 1.24741554e+00 2.37901673e-01 5.33753753e-01 3.71947944e-01 3.97770256e-01 4.30560380e-01 1.13095999e+00 1.08673620e+00 5.73224686e-02 8.61003578e-01 6.52038276e-01 1.69706464e-01 2.59981453e-01 -7.84741994e-03 9.11257267e-01 3.91434908e-01 -2.20344216e-01 -5.24228275e-01 -5.83740115e-01 6.60148680e-01 -2.21246982e+00 -1.29521573e+00 7.43081644e-02 2.49198771e+00 5.63498378e-01 3.70751441e-01 6.42017961e-01 4.73722517e-01 6.48322701e-01 4.60444242e-01 -3.60159427e-01 -2.13090256e-01 1.23160332e-01 -2.16967955e-01 6.94059193e-01 3.65116835e-01 -1.42771065e+00 4.21616256e-01 6.35664701e+00 6.62279904e-01 -9.25782979e-01 8.34406242e-02 4.71619695e-01 -5.11205256e-01 3.86427879e-01 -2.96784658e-02 -8.23910832e-01 6.14598513e-01 1.09837365e+00 -1.71933919e-01 2.67393172e-01 8.64471555e-01 9.17314708e-01 -3.86336505e-01 -1.43237436e+00 9.64464605e-01 8.91508684e-02 -1.13208365e+00 -1.46657676e-01 -6.28563985e-02 3.33372772e-01 -2.61824906e-01 -3.19937944e-01 -6.60743043e-02 -1.88905254e-01 -4.71875787e-01 8.80127370e-01 2.55068481e-01 5.64740598e-01 -4.35118377e-01 2.55963981e-01 5.37738979e-01 -1.23609018e+00 -5.19198596e-01 -2.72377670e-01 -4.72008526e-01 5.97488999e-01 4.92444009e-01 -4.85822886e-01 1.91019535e-01 6.45370841e-01 9.87102807e-01 -2.70628422e-01 1.21900594e+00 1.60329998e-01 8.20243001e-01 -5.69624007e-01 3.46682340e-01 1.86035529e-01 1.77830935e-01 6.86006963e-01 1.30287421e+00 4.37868953e-01 2.21808493e-01 4.15092319e-01 -4.67224494e-02 1.87012047e-01 1.44988492e-01 -4.94384319e-01 4.95419353e-02 1.64885506e-01 7.67556906e-01 -7.79538810e-01 -4.23240542e-01 -9.69511807e-01 1.16698480e+00 -1.39918998e-01 1.61396042e-01 -1.22934568e+00 1.06829740e-01 6.60187066e-01 4.92360622e-01 7.67513216e-01 -2.98910558e-01 3.31155479e-01 -1.16959083e+00 3.55509311e-01 -9.74490285e-01 6.34557664e-01 -4.75486189e-01 -7.16297448e-01 -2.63354927e-02 1.46679223e-01 -1.52984178e+00 -3.43730301e-01 -4.72539306e-01 -6.40126228e-01 -5.32113388e-02 -1.35945511e+00 -7.14869142e-01 -8.42666775e-02 8.33007395e-01 8.57169330e-01 3.79567146e-01 3.64758253e-01 5.95605671e-01 -4.70541239e-01 2.42513612e-01 -1.91911943e-02 9.83120799e-02 6.72158301e-01 -1.06990075e+00 1.82317644e-01 1.12697506e+00 6.57149628e-02 -2.49865502e-02 9.52914476e-01 -4.28347319e-01 -1.69862282e+00 -8.97071362e-01 7.10109055e-01 -2.73434401e-01 7.96707988e-01 -3.68652701e-01 -7.51979709e-01 8.11925769e-01 3.76889191e-04 3.57504040e-01 5.51695943e-01 -4.54903364e-01 1.94072761e-02 -2.14956313e-01 -7.17242777e-01 3.22817832e-01 1.37134492e+00 -2.95885533e-01 -3.50119144e-01 5.61445832e-01 6.90679312e-01 -3.61769676e-01 -6.60270214e-01 1.35247424e-01 7.11578071e-01 -1.25768757e+00 8.96962702e-01 -6.39298320e-01 2.90330350e-01 -1.80867821e-01 -1.02474302e-01 -5.23415923e-01 -1.10882223e-02 -1.20168781e+00 -5.07509649e-01 1.09579515e+00 -7.71104693e-02 -1.94214687e-01 8.42494309e-01 8.31184447e-01 7.51558542e-02 -5.75532675e-01 -1.30506945e+00 -9.00769711e-01 -6.10241354e-01 -7.07375109e-01 1.13356650e-01 4.42959219e-01 2.39127800e-01 -1.56892389e-01 -7.32727051e-01 2.23088592e-01 3.04941148e-01 -5.13734221e-02 9.55534995e-01 -6.36631072e-01 -7.87513554e-01 -2.50502694e-02 -5.72931409e-01 -1.90012002e+00 -1.77630544e-01 -2.19484493e-02 1.11471616e-01 -1.00113165e+00 3.57198954e-01 8.36320668e-02 -4.01995927e-01 1.78786352e-01 -8.78529996e-02 3.72299328e-02 3.77631396e-01 1.79858059e-01 -1.33319509e+00 2.48579219e-01 9.93266761e-01 3.05270374e-01 -1.97633058e-01 3.21054369e-01 -1.77685991e-01 9.32299733e-01 3.71614814e-01 -4.87468451e-01 -5.66614687e-01 -3.96931380e-01 3.35468560e-01 5.59232593e-01 3.59447449e-01 -1.07017982e+00 2.21721143e-01 -4.25806552e-01 -2.85870433e-01 -2.91246414e-01 3.36557835e-01 -1.05154908e+00 -1.86971836e-02 4.13363427e-01 -4.97627884e-01 8.37219507e-03 -1.21280663e-01 9.22300339e-01 -5.84389530e-02 -8.30807388e-02 7.18486905e-01 -7.17974380e-02 -9.50498164e-01 3.23684901e-01 -6.82563722e-01 1.95401087e-01 1.29017007e+00 -5.15527010e-01 -7.42477775e-02 -7.84436107e-01 -8.15068483e-01 1.21214636e-01 5.18630922e-01 3.55561823e-01 4.79253680e-01 -1.02312112e+00 -5.60758889e-01 -1.98394470e-02 -1.32768983e-02 -4.55853283e-01 2.84732163e-01 1.35078597e+00 -5.59340298e-01 1.30980074e-01 7.74049908e-02 -6.47541761e-01 -1.66647077e+00 8.60776305e-01 1.41566768e-01 -2.28950679e-01 -8.89059544e-01 6.22614026e-01 3.88395876e-01 6.28530681e-01 5.69425344e-01 -6.50082350e-01 -1.83999985e-02 -6.26343638e-02 1.12547672e+00 7.80784369e-01 -1.54479891e-01 -7.08141446e-01 -2.86026061e-01 6.86131045e-02 3.98875438e-02 -1.09974071e-01 1.17300069e+00 -6.25339866e-01 2.71385133e-01 5.33001721e-01 1.01858830e+00 2.11030953e-02 -1.83341241e+00 -2.12832004e-01 2.05045845e-02 -8.93768728e-01 -1.07592776e-01 -1.08185038e-01 -9.91945744e-01 6.01690948e-01 5.62671959e-01 5.39453506e-01 1.33588922e+00 -1.92647651e-01 1.10860550e+00 2.69050628e-01 4.88838613e-01 -1.31106246e+00 1.26940876e-01 3.89782190e-01 6.32613599e-01 -1.15350592e+00 -6.61711842e-02 -6.33506835e-01 -3.02428842e-01 1.08380044e+00 2.99992681e-01 -2.06840590e-01 6.54345989e-01 3.08493227e-01 -1.06927082e-01 2.03008667e-01 -1.06848478e+00 -1.10659830e-01 -1.83838412e-01 3.09755355e-01 2.51727700e-01 -2.65584111e-01 -3.88050139e-01 2.81901956e-01 6.09516025e-01 2.47295678e-01 8.75342309e-01 1.35462976e+00 -6.61241829e-01 -8.66342485e-01 -3.37532103e-01 2.07338214e-01 -1.09791493e+00 1.12360716e-01 -2.54918277e-01 6.40441179e-01 7.67629920e-03 1.16829503e+00 1.47762135e-01 -5.21699116e-02 2.35271543e-01 -1.15861304e-01 4.32065904e-01 -6.49514854e-01 -3.39856625e-01 3.38135272e-01 2.04949647e-01 -1.24760985e+00 -8.54459941e-01 -1.00001001e+00 -1.10539651e+00 -2.39928201e-01 -2.81071335e-01 -1.40432075e-01 9.46474969e-02 9.56048191e-01 4.70769376e-01 1.49002373e-01 7.67261863e-01 -8.98468912e-01 -6.55652285e-01 -7.16104627e-01 -7.60672092e-01 5.64013660e-01 6.70011878e-01 -5.13835907e-01 -6.05692089e-01 4.71321821e-01]
[8.33199405670166, 0.4387865662574768]
e1003fad-7ded-4482-ba1d-7bdef78ff3cf
iitk-detox-at-semeval-2021-task-5-semi
2104.01566
null
https://arxiv.org/abs/2104.01566v1
https://arxiv.org/pdf/2104.01566v1.pdf
IITK@Detox at SemEval-2021 Task 5: Semi-Supervised Learning and Dice Loss for Toxic Spans Detection
In this work, we present our approach and findings for SemEval-2021 Task 5 - Toxic Spans Detection. The task's main aim was to identify spans to which a given text's toxicity could be attributed. The task is challenging mainly due to two constraints: the small training dataset and imbalanced class distribution. Our paper investigates two techniques, semi-supervised learning and learning with Self-Adjusting Dice Loss, for tackling these challenges. Our submitted system (ranked ninth on the leader board) consisted of an ensemble of various pre-trained Transformer Language Models trained using either of the above-proposed techniques.
['Ashutosh Modi', 'Abhay Kaushik', 'Archit Bansal']
2021-04-04
null
https://aclanthology.org/2021.semeval-1.24
https://aclanthology.org/2021.semeval-1.24.pdf
semeval-2021
['toxic-spans-detection']
['natural-language-processing']
[ 2.40998656e-01 -1.58610255e-01 -2.30885088e-01 -8.06929469e-02 -1.44629312e+00 -6.91278398e-01 6.31921053e-01 7.65833735e-01 -5.09518147e-01 1.43119979e+00 3.51005256e-01 -3.54508907e-01 -1.42875880e-01 -3.62669945e-01 -7.06503570e-01 -5.00384450e-01 -1.67112663e-01 4.76549804e-01 3.26094151e-01 6.98737502e-02 4.33476508e-01 5.44074774e-01 -9.86901164e-01 5.79604030e-01 7.43116558e-01 1.03528845e+00 -8.11710656e-02 5.47335744e-01 3.97167820e-03 1.32283950e+00 -5.90250492e-01 -5.68183064e-01 3.19078863e-02 -2.93547750e-01 -1.02583492e+00 -5.83825350e-01 2.63769418e-01 -4.16247807e-02 -3.20614100e-01 8.82798076e-01 8.08581829e-01 -2.46965989e-01 9.98311281e-01 -1.29015887e+00 5.92139997e-02 8.69819820e-01 -5.53957224e-01 6.30737066e-01 4.82163131e-01 1.38080735e-02 8.84656847e-01 -8.96908522e-01 4.63075519e-01 1.00791979e+00 9.47214961e-01 5.67982912e-01 -1.11197591e+00 -9.25637960e-01 -2.37732217e-01 2.70919383e-01 -1.52493119e+00 -4.58666384e-01 5.53637922e-01 -8.85042131e-01 1.18474305e+00 1.33311048e-01 1.22450784e-01 1.32394838e+00 6.33572817e-01 5.02538025e-01 1.43552208e+00 -2.71546274e-01 4.24399942e-01 2.75528550e-01 2.49720126e-01 2.79787898e-01 9.12698060e-02 8.29019845e-02 -8.88974726e-01 -5.89291334e-01 -1.02202132e-01 -4.67216998e-01 -4.83310074e-02 2.72615612e-01 -9.36284184e-01 7.17580914e-01 6.82490319e-02 2.23256290e-01 -2.41660178e-01 9.56090167e-02 9.55840707e-01 2.07075670e-01 9.27800655e-01 3.96487057e-01 -7.86199212e-01 4.60405508e-03 -1.41816926e+00 2.07636535e-01 5.98778963e-01 7.54969001e-01 2.05967546e-01 -1.91301718e-01 -5.45121908e-01 6.23408198e-01 3.55084896e-01 8.89622271e-02 3.06484818e-01 -3.84900928e-01 8.89458179e-01 3.55625331e-01 -7.44815469e-02 -3.66175115e-01 -5.59919655e-01 -3.32343102e-01 -6.73828542e-01 7.99744427e-02 5.14333844e-01 -4.18487608e-01 -8.70544434e-01 1.74305272e+00 -1.11602098e-01 -2.18569160e-01 -3.03685348e-02 2.97226489e-01 8.40001941e-01 5.70502520e-01 8.29336762e-01 -3.26157093e-01 1.28831077e+00 -6.38498187e-01 -6.37890816e-01 4.44332510e-02 7.28601456e-01 -8.70405436e-01 5.79306960e-01 5.86304784e-01 -9.67631757e-01 -1.02406338e-01 -1.04212749e+00 1.72614262e-01 -8.06158483e-01 2.21926004e-01 2.40864322e-01 5.88243842e-01 -8.33817720e-01 7.51160026e-01 -2.84916461e-01 -2.65604645e-01 6.53062284e-01 4.85594571e-01 -2.94256926e-01 2.02332050e-01 -1.65088558e+00 1.25834763e+00 4.30850148e-01 -2.99341500e-01 -1.19756436e+00 -7.68740535e-01 -6.22943044e-01 -1.81622773e-01 4.33368683e-02 -4.09116060e-01 1.12819123e+00 -3.55974734e-01 -9.18401361e-01 9.92354929e-01 1.02063231e-01 -7.21205294e-01 9.29903746e-01 -3.05511057e-01 -3.18693221e-01 -1.03458412e-01 1.80027172e-01 4.07397538e-01 2.32680708e-01 -6.86939478e-01 -3.36783320e-01 -6.52709544e-01 -2.54200578e-01 -7.61220008e-02 -5.89888215e-01 5.24933338e-01 -1.44495741e-02 -6.14201665e-01 -6.58644915e-01 -6.88241661e-01 -3.10015142e-01 -2.65392572e-01 -7.15623975e-01 -5.69515526e-01 7.16557384e-01 -8.82919252e-01 1.53704667e+00 -1.49292302e+00 -1.74633428e-01 -3.93954590e-02 1.23800308e-01 1.68561637e-01 -4.15050127e-02 7.95001209e-01 -3.25736403e-01 1.34568498e-01 -2.66902328e-01 -3.44470531e-01 -1.35331511e-01 -5.60109973e-01 -4.95909065e-01 5.76031804e-01 2.10412398e-01 5.45838535e-01 -8.85749340e-01 -7.99084425e-01 -9.10001993e-02 2.47865900e-01 8.01106766e-02 2.91365325e-01 -2.90030658e-01 2.49009892e-01 -2.84903407e-01 7.35758781e-01 6.07234597e-01 2.98276663e-01 -1.08253360e-01 -1.87480181e-01 -1.61247641e-01 4.51937646e-01 -6.71594381e-01 1.52933085e+00 -2.41115659e-01 4.32286501e-01 -4.81494009e-01 -8.04525733e-01 8.83000851e-01 4.40878153e-01 5.43327987e-01 -5.57504594e-01 -9.16612614e-03 2.15634197e-01 -1.29540384e-01 -4.93157029e-01 1.86430998e-02 -3.65119547e-01 -3.26791465e-01 3.19834530e-01 1.86467022e-01 -1.17654912e-04 2.90825307e-01 2.15911090e-01 1.42903709e+00 2.49141157e-01 2.92474896e-01 -5.01098394e-01 7.59650171e-01 -2.04539031e-01 6.11875057e-01 4.22060341e-01 -3.29919666e-01 4.63438958e-01 8.01244497e-01 -3.28811705e-01 -1.02288091e+00 -1.00343728e+00 -2.78783619e-01 8.81758451e-01 -5.07062435e-01 -5.28090596e-01 -8.55971873e-01 -1.01517272e+00 -2.25405753e-01 9.49516535e-01 -6.99163914e-01 -7.46283531e-02 -2.08725125e-01 -1.02567899e+00 1.20410156e+00 7.41591692e-01 1.79129630e-01 -8.46335590e-01 -4.31924045e-01 2.33231723e-01 -1.30666092e-01 -9.70846057e-01 -6.37869239e-01 7.58687377e-01 -6.98241472e-01 -1.10481930e+00 -8.60056520e-01 -5.45925736e-01 2.75106430e-01 -4.03161407e-01 1.08497083e+00 -3.59955698e-01 -3.81793737e-01 -1.99245811e-01 -8.60334858e-02 -7.86172509e-01 -3.63485068e-01 9.15085301e-02 1.38893813e-01 -4.77222890e-01 5.22701621e-01 -3.84628564e-01 -6.11380160e-01 1.81649208e-01 -7.22633421e-01 -3.96807432e-01 5.12780845e-01 5.82148969e-01 5.17820716e-01 1.56699196e-01 1.03825426e+00 -1.01110995e+00 8.05784345e-01 -7.63261318e-01 -3.82325649e-01 4.86436725e-01 -7.70477533e-01 5.81522249e-02 7.43162870e-01 -3.84052336e-01 -7.21122324e-01 4.00888294e-01 -2.82846987e-01 -1.00419201e-01 1.80674233e-02 5.68990946e-01 -5.43500662e-01 7.56531209e-02 7.25623906e-01 2.57052809e-01 -3.64619523e-01 -4.59536791e-01 7.78487325e-02 1.00258279e+00 2.80169666e-01 -6.26365423e-01 5.27658343e-01 -1.89171255e-01 9.82672572e-02 -5.83877444e-01 -9.55673337e-01 -4.04289395e-01 -6.63298130e-01 -3.75469923e-01 8.29325020e-01 -1.18175292e+00 -2.72705674e-01 8.09254348e-01 -1.44062901e+00 -3.76700222e-01 -1.41138762e-01 2.67533839e-01 -4.37912822e-01 4.15597379e-01 -7.08862424e-01 -9.42541599e-01 -9.49872255e-01 -7.71716595e-01 8.59922171e-01 -9.34391469e-02 -4.78615820e-01 -1.01502204e+00 5.76986074e-01 3.96311283e-01 3.46006989e-01 5.38182616e-01 1.00447965e+00 -1.17151427e+00 2.45247632e-01 -3.41944695e-01 -2.08961353e-01 3.42799604e-01 -3.50587219e-02 -2.14737192e-01 -1.17287242e+00 -2.69441009e-01 -5.98811656e-02 -9.12640810e-01 9.13481951e-01 2.80342996e-01 1.39069963e+00 -2.25129142e-01 -2.88040191e-01 -1.09465182e-01 1.41434872e+00 7.52416030e-02 7.44763970e-01 3.00975442e-01 2.63907641e-01 6.99459910e-01 4.75209892e-01 5.80221772e-01 2.00984940e-01 6.28757775e-01 2.45315701e-01 -2.98343524e-02 -2.98505038e-01 -5.00945807e-01 5.62321901e-01 4.55149025e-01 2.07412839e-01 -4.18770820e-01 -1.02055705e+00 5.88730931e-01 -1.53015137e+00 -1.00610971e+00 -3.62481236e-01 2.24852562e+00 1.16347194e+00 5.76357007e-01 4.28874165e-01 3.89048696e-01 5.99215209e-01 -5.92717081e-02 -5.86818635e-01 -6.22034550e-01 -1.80372596e-01 1.37863919e-01 6.76870704e-01 1.37714416e-01 -1.14273763e+00 6.67415142e-01 7.31217003e+00 1.38533223e+00 -8.96534801e-01 2.18048424e-01 1.04070520e+00 1.29925072e-01 -1.03635237e-01 -1.89771563e-01 -9.21361923e-01 7.12642848e-01 1.52079976e+00 -2.28362843e-01 5.48017211e-02 2.14777187e-01 4.28698897e-01 -1.74314976e-01 -1.30152833e+00 4.59514707e-01 3.76126796e-01 -1.03306365e+00 -8.89560580e-03 -2.45175883e-03 4.02047336e-01 1.12591177e-01 7.75160044e-02 2.30730280e-01 1.81334913e-01 -1.23019016e+00 8.62180471e-01 6.51931465e-01 1.06951165e+00 -7.90866017e-01 7.24395454e-01 4.96861756e-01 -1.00014210e+00 8.53378791e-03 -2.14221478e-01 -2.34493539e-02 -1.30042017e-01 1.04061913e+00 -7.69834757e-01 4.25048411e-01 5.09248853e-01 5.89840055e-01 -6.62504971e-01 1.44213772e+00 -1.70050994e-01 9.45846915e-01 -2.20804185e-01 -1.23212263e-01 9.88262296e-02 3.27351332e-01 3.83925438e-01 1.60853994e+00 2.77429700e-01 -2.08717152e-01 3.58251110e-02 8.30743790e-01 -3.46646249e-01 3.17701161e-01 -7.14440227e-01 4.51957854e-03 4.08261359e-01 1.18970108e+00 -6.67607188e-01 -2.22539648e-01 -2.81604975e-01 6.75556421e-01 3.19746196e-01 -2.27388740e-01 -9.42886233e-01 -5.60860157e-01 -6.88882247e-02 3.79429698e-01 -7.83358365e-02 2.07009420e-01 -5.91093481e-01 -7.47078061e-01 -3.56058061e-01 -6.80521429e-01 8.23196411e-01 -7.02894092e-01 -1.47264433e+00 8.20215583e-01 7.69746378e-02 -1.21573007e+00 1.25074238e-01 -6.92917466e-01 -8.59478474e-01 9.57072794e-01 -1.38166642e+00 -1.16743612e+00 3.44070971e-01 3.53782326e-01 8.64506841e-01 -4.46730018e-01 8.33462834e-01 5.57785153e-01 -5.19381881e-01 8.85027766e-01 7.26824105e-02 -8.60384703e-02 9.40972447e-01 -1.35505223e+00 3.48358899e-02 6.27592921e-01 -1.87265992e-01 2.38060132e-01 9.62607384e-01 -7.53071129e-01 -8.48038614e-01 -1.34091759e+00 1.55971813e+00 -6.86463773e-01 8.87265325e-01 -4.81338799e-01 -6.31161690e-01 2.54437923e-01 2.74516761e-01 -2.65894651e-01 1.00579405e+00 -1.62833154e-01 -4.18668568e-01 -1.15846120e-01 -1.35005069e+00 2.99569853e-02 5.96109271e-01 -6.28485084e-01 -5.73935449e-01 7.96847105e-01 3.24703068e-01 -3.01694751e-01 -1.01775336e+00 3.82574975e-01 2.77584046e-01 -4.47145879e-01 8.61023664e-01 -9.41132903e-01 8.59274328e-01 -2.01852679e-01 -1.62866428e-01 -1.21147811e+00 -9.44093764e-02 -4.58204120e-01 -9.22364667e-02 1.57706904e+00 7.89942741e-01 -1.47711948e-01 7.28663743e-01 5.26203156e-01 1.75962240e-01 -1.02952838e+00 -1.13770890e+00 -8.04242730e-01 7.59096861e-01 -4.05174732e-01 2.61771142e-01 6.07730865e-01 1.97962821e-01 8.03580046e-01 -6.05480731e-01 -2.02871397e-01 8.21832538e-01 -4.07601446e-01 1.65583313e-01 -1.12657619e+00 5.14692962e-02 -3.02407742e-01 -3.01903069e-01 -3.43112528e-01 1.94113255e-01 -8.86897922e-01 1.12741873e-01 -1.20890212e+00 9.20826435e-01 -1.39060050e-01 -7.13892579e-01 6.07581735e-01 2.21876949e-02 3.09278607e-01 -4.51543964e-02 2.89325863e-02 -7.79696226e-01 4.05639917e-01 5.56963980e-01 -5.19650757e-01 1.84918255e-01 3.60327363e-01 -8.51550758e-01 5.31754017e-01 8.89069796e-01 -7.95546889e-01 -3.18144143e-01 -1.07812341e-02 3.49543691e-01 1.22977514e-02 2.43089840e-01 -1.03802609e+00 3.86183679e-01 -2.52830684e-02 5.54088116e-01 -1.03002632e+00 -9.14261043e-02 -4.94372427e-01 2.18297839e-01 7.21838593e-01 -8.93729389e-01 2.15451181e-01 5.19208431e-01 4.48161304e-01 3.07985153e-02 -4.93130863e-01 7.38671184e-01 -1.09853342e-01 -1.17369927e-01 1.60147563e-01 -6.22843146e-01 2.49310195e-01 1.06635189e+00 2.48823613e-01 -3.55901122e-01 -1.19602688e-01 -6.52320743e-01 3.40689898e-01 -7.22557753e-02 1.25240237e-01 5.00042379e-01 -1.21316028e+00 -1.13395739e+00 -6.20250106e-01 2.51737654e-01 -5.32123983e-01 1.58717915e-01 9.40003753e-01 -2.32827574e-01 6.48976922e-01 -1.75844431e-01 -4.83510569e-02 -1.43087852e+00 6.95978820e-01 2.83578992e-01 -9.71698821e-01 -1.00955255e-01 7.01118708e-01 -2.03457206e-01 -4.00536686e-01 3.99259657e-01 9.44463611e-02 -5.13845026e-01 3.82418960e-01 6.04662716e-01 6.89119518e-01 4.48635757e-01 -6.76703691e-01 -6.32165074e-01 3.34899187e-01 -3.26820612e-01 1.39145013e-02 1.38453889e+00 1.11578807e-01 -1.79924250e-01 4.81055528e-01 1.23673368e+00 -1.38375342e-01 -7.21367121e-01 -6.31252527e-02 3.81498396e-01 1.04566671e-01 2.17815667e-01 -1.51392102e+00 -6.30822837e-01 1.12600875e+00 8.47181678e-01 -1.85189784e-01 1.03376424e+00 -2.77552903e-02 8.04251552e-01 1.71518877e-01 2.91921407e-01 -1.35155678e+00 3.55307274e-02 4.32650208e-01 8.39220881e-01 -8.64412665e-01 2.98086733e-01 -1.27749681e-01 -5.07800639e-01 1.23672414e+00 6.19251132e-01 2.52636075e-01 5.59589565e-01 4.69440341e-01 -1.72773272e-01 -1.42106652e-01 -1.13782561e+00 1.47639066e-01 3.73296559e-01 4.12360728e-01 6.85486972e-01 -1.90339208e-01 -9.27057505e-01 9.15970087e-01 4.43545356e-03 1.58239126e-01 5.24988055e-01 6.26614213e-01 -3.25345606e-01 -9.81122494e-01 -2.63487816e-01 7.01326847e-01 -1.00788069e+00 -2.50990987e-01 -6.64510131e-01 2.82254577e-01 3.17618340e-01 9.62140322e-01 -5.83531320e-01 -5.39191723e-01 3.11737031e-01 7.72560388e-02 4.32110667e-01 -3.85645390e-01 -7.34311342e-01 -5.75467795e-02 2.48750418e-01 -2.58700192e-01 -2.90678263e-01 -8.01551461e-01 -9.64466155e-01 2.06518844e-02 -4.10827637e-01 3.38206500e-01 5.32836497e-01 9.83740747e-01 -7.09577948e-02 3.08450997e-01 7.06790328e-01 -2.68469900e-01 -9.70719337e-01 -1.11369038e+00 -5.52689910e-01 2.17431098e-01 4.80750948e-02 -4.62179899e-01 -3.27114940e-01 3.25470343e-02]
[8.959778785705566, 10.62716007232666]
ba47adfc-d709-4a8e-b3d1-5f3923615826
egocentric-image-captioning-for-privacy
2107.00372
null
https://arxiv.org/abs/2107.00372v2
https://arxiv.org/pdf/2107.00372v2.pdf
Egocentric Image Captioning for Privacy-Preserved Passive Dietary Intake Monitoring
Camera-based passive dietary intake monitoring is able to continuously capture the eating episodes of a subject, recording rich visual information, such as the type and volume of food being consumed, as well as the eating behaviours of the subject. However, there currently is no method that is able to incorporate these visual clues and provide a comprehensive context of dietary intake from passive recording (e.g., is the subject sharing food with others, what food the subject is eating, and how much food is left in the bowl). On the other hand, privacy is a major concern while egocentric wearable cameras are used for capturing. In this paper, we propose a privacy-preserved secure solution (i.e., egocentric image captioning) for dietary assessment with passive monitoring, which unifies food recognition, volume estimation, and scene understanding. By converting images into rich text descriptions, nutritionists can assess individual dietary intake based on the captions instead of the original images, reducing the risk of privacy leakage from images. To this end, an egocentric dietary image captioning dataset has been built, which consists of in-the-wild images captured by head-worn and chest-worn cameras in field studies in Ghana. A novel transformer-based architecture is designed to caption egocentric dietary images. Comprehensive experiments have been conducted to evaluate the effectiveness and to justify the design of the proposed architecture for egocentric dietary image captioning. To the best of our knowledge, this is the first work that applies image captioning for dietary intake assessment in real life settings.
['Benny Lo', 'Gary Frost', 'Mingui Sun', 'Edward Sazonov', 'Megan A McCrory', 'Alex K. Anderson', 'Matilda Steiner-Asiedu', 'Tom Baranowski', 'Wenyan Jia', 'Modou L. Jobarteh', 'Xiao Gu', 'Frank P. -W. Lo', 'Jianing Qiu']
2021-07-01
null
null
null
null
['food-recognition']
['computer-vision']
[ 3.45082581e-01 -5.58480583e-02 -3.75353247e-01 -7.32964277e-01 -3.69533300e-01 -7.41572380e-01 -1.47225946e-01 5.84751904e-01 -7.78950974e-02 9.02832225e-02 7.61675298e-01 2.35069185e-01 2.41809472e-01 -7.98427880e-01 -9.46620882e-01 -6.26864314e-01 1.68016609e-02 -1.12940729e-01 -5.87725282e-01 2.76777178e-01 -2.06320718e-01 -8.30616727e-02 -9.84877706e-01 3.77505183e-01 3.02590162e-01 9.82124805e-01 2.07562223e-01 6.53683484e-01 3.26621026e-01 7.63658822e-01 -1.87104642e-01 -5.04554331e-01 3.64426613e-01 -4.63110298e-01 -1.31304517e-01 5.44755042e-01 6.31476581e-01 -1.21390915e+00 -3.19859624e-01 1.06656218e+00 4.47906554e-01 -2.17526689e-01 2.29376838e-01 -1.04092765e+00 -1.15943992e+00 5.63504219e-01 -6.15764022e-01 -1.07273338e-02 7.92301238e-01 2.08206013e-01 4.17922884e-01 -1.37107685e-01 2.69200683e-01 9.14271653e-01 6.32146955e-01 5.32210290e-01 -1.24083340e+00 -8.93959045e-01 5.01240119e-02 1.60648406e-01 -1.17021680e+00 -5.26759446e-01 1.00498462e+00 -5.01614094e-01 3.87845755e-01 4.72026020e-01 1.06381011e+00 1.27893364e+00 -7.35186040e-03 4.91055548e-01 1.19397616e+00 -3.71028066e-01 3.94152015e-01 7.08094537e-01 -1.37471575e-02 4.99405891e-01 5.44432342e-01 -1.04123518e-01 -3.71525764e-01 -2.29088902e-01 5.23249805e-01 3.64790738e-01 -2.54120022e-01 -9.40038562e-01 -1.11339140e+00 9.95381773e-01 5.89226663e-01 -1.61551848e-01 -5.62910974e-01 4.98863757e-02 5.33653378e-01 -2.42214650e-01 6.54394984e-01 -1.58496246e-01 -9.44751278e-02 4.62318778e-01 -7.97223747e-01 -4.02027182e-02 6.66408122e-01 1.07866585e+00 3.01753819e-01 -1.18516393e-01 -2.89954603e-01 1.80705115e-01 6.56670451e-01 9.93515849e-01 3.97878617e-01 -6.76105917e-01 4.72631186e-01 7.47470260e-01 2.65599996e-01 -1.41737783e+00 -1.74469948e-01 9.73742902e-02 -6.54615939e-01 -1.84515700e-01 2.82024860e-01 -2.60882229e-01 -7.31381893e-01 1.86807394e+00 6.34397686e-01 -7.63977244e-02 1.69082955e-01 1.18614852e+00 1.00617743e+00 5.18138826e-01 6.11013412e-01 -1.22468323e-01 2.15218687e+00 -5.97986042e-01 -1.11483073e+00 -2.70665616e-01 4.05456394e-01 -7.29201019e-01 7.07876921e-01 -6.97332397e-02 -8.12782645e-01 -2.44547427e-01 -1.11454272e+00 4.25188243e-02 -4.33950543e-01 3.04971814e-01 6.22254252e-01 1.16767180e+00 -4.26932931e-01 -1.54901862e-01 -9.04768407e-01 -8.69284868e-01 2.75049090e-01 1.07858151e-01 -6.01012111e-01 -2.88453758e-01 -9.07974303e-01 5.20695686e-01 2.53141314e-01 1.30873337e-01 -9.19984221e-01 -8.40021312e-01 -1.29732704e+00 -6.38461709e-02 1.11000963e-01 -7.50006795e-01 1.16476977e+00 -1.19314146e+00 -1.14352679e+00 1.00549662e+00 1.12647517e-02 -5.94673872e-01 4.52297598e-01 -1.42181188e-01 -4.76099223e-01 5.22182405e-01 2.06606731e-01 6.27953529e-01 5.26206017e-01 -1.04542577e+00 -9.08377320e-02 -1.05323195e+00 -1.36583643e-02 6.06282055e-01 -3.60549510e-01 1.72861576e-01 -1.56846698e-02 -4.13863420e-01 -1.94090888e-01 -8.37271690e-01 1.14939749e-01 7.28933513e-01 -3.12955827e-01 5.15248716e-01 8.11452687e-01 -1.06069458e+00 9.46677923e-01 -2.41395402e+00 -6.53432846e-01 -1.00657143e-01 1.71336136e-03 3.26971948e-01 2.03495920e-01 1.35636449e-01 -3.83039936e-02 -1.47281393e-01 -1.15698628e-01 -6.53577074e-02 1.23797506e-01 4.66779806e-02 1.30045235e-01 1.09015405e+00 -3.39351296e-01 9.28143740e-01 -8.30190420e-01 -5.60123444e-01 7.25861311e-01 6.45679235e-01 -4.50336933e-01 4.05674011e-01 -1.51542634e-01 2.13674664e-01 -3.75337601e-01 6.52633786e-01 1.03078508e+00 8.56219009e-02 4.07555044e-01 -7.52321303e-01 2.37934813e-02 -2.59759724e-01 -1.01887906e+00 1.73547518e+00 -6.45565763e-02 2.16002181e-01 5.19592941e-01 -6.80941164e-01 6.55929029e-01 7.78190553e-01 6.66813314e-01 -7.50887215e-01 2.48451397e-01 -3.80156487e-01 -5.74703038e-01 -1.12444186e+00 2.24945307e-01 7.45647103e-02 -8.86404216e-02 3.60486418e-01 -3.35225701e-01 3.93922836e-01 -1.51881874e-01 -1.38429299e-01 5.27589142e-01 3.68113041e-01 5.98228753e-01 -1.39076665e-01 2.75491059e-01 3.63896787e-01 2.36172140e-01 2.42735624e-01 -6.63821220e-01 5.04907906e-01 -3.09655815e-01 -6.16725087e-01 -1.29215395e+00 -1.05542946e+00 -5.45928590e-02 7.62302816e-01 2.08816931e-01 -1.77748784e-01 -1.04194593e+00 -5.21462083e-01 -3.55519354e-02 7.68786907e-01 -7.38835871e-01 -2.04287186e-01 -3.29418033e-01 -7.06530154e-01 2.66685486e-01 4.06199992e-01 9.36989188e-01 -7.78654456e-01 -1.05314541e+00 1.93781093e-01 -6.52616501e-01 -1.04282439e+00 -1.10347915e+00 -1.40019625e-01 -6.38827324e-01 -1.34727001e+00 -8.57507348e-01 -7.31183171e-01 7.19364643e-01 6.93509758e-01 8.31281126e-01 -6.45483017e-01 -5.21774054e-01 7.70922303e-01 -2.03418791e-01 -7.58387744e-01 -1.46596104e-01 -5.50136268e-01 -2.59575665e-01 2.12751657e-01 1.18284857e+00 -2.38265797e-01 -1.36154866e+00 2.43896365e-01 -7.20358968e-01 -4.30898033e-02 2.14628741e-01 1.87647462e-01 7.13694572e-01 -1.76935807e-01 2.51483005e-02 -8.70215178e-01 2.80006796e-01 -8.70464206e-01 -3.22510540e-01 2.01638147e-01 -3.46456587e-01 -4.83985960e-01 4.94776845e-01 -6.20003581e-01 -1.01646829e+00 6.12664700e-01 3.51368546e-01 -2.89526016e-01 -7.09259629e-01 5.01286574e-02 -1.50910765e-01 1.49553716e-01 5.06251514e-01 2.77684569e-01 2.73190498e-01 -4.56354916e-01 4.64440942e-01 1.02273369e+00 5.60514450e-01 4.05437499e-02 3.82748336e-01 7.86900699e-01 -3.00939173e-01 -8.03736985e-01 -6.93058074e-01 -6.66676641e-01 -3.28203470e-01 -1.35269076e-01 1.28712225e+00 -1.39451742e+00 -8.61785769e-01 2.99564153e-01 -6.96489155e-01 2.32982025e-01 -2.32352942e-01 9.74157810e-01 -5.15269637e-01 3.73388320e-01 -2.54106939e-01 -9.47696507e-01 -8.72504592e-01 -8.96744490e-01 9.16283071e-01 1.67057872e-01 -2.97583431e-01 -7.41175473e-01 4.46308404e-02 8.09914827e-01 3.78416151e-01 7.99418867e-01 5.72298765e-01 -2.10410759e-01 -3.89520079e-01 -2.88785666e-01 -3.28352302e-01 7.42472932e-02 4.06755239e-01 -6.59015715e-01 -1.00521338e+00 -4.16240364e-01 4.64255691e-01 -6.18566237e-02 1.29798219e-01 7.35180378e-01 6.26665890e-01 -9.53313053e-01 -2.27793753e-01 6.52177334e-01 1.69747138e+00 3.79452467e-01 6.07244968e-01 1.69974238e-01 6.39997423e-01 5.76816082e-01 3.51887703e-01 4.28622901e-01 7.27461219e-01 7.95990348e-01 6.05925441e-01 -3.62802297e-01 -2.25329578e-01 -5.66220701e-01 6.78749979e-01 2.75783449e-01 6.91834569e-01 -3.00377011e-01 -4.97047268e-02 6.77043319e-01 -1.41476464e+00 -9.10336196e-01 -1.28432289e-01 2.15843940e+00 4.79296446e-01 -7.73460090e-01 3.25649410e-01 -2.30201125e-01 8.86841595e-01 1.18747831e-03 -8.16131294e-01 -6.72241330e-01 3.99567693e-01 -4.65759128e-01 9.42917407e-01 -3.00705135e-02 -1.29082656e+00 1.40500203e-01 5.71145630e+00 -9.86779109e-02 -8.63071442e-01 2.20103130e-01 6.09918594e-01 -1.70207709e-01 -1.00759931e-01 -4.14386690e-01 -2.36634731e-01 6.07616484e-01 9.59398985e-01 4.79122400e-02 5.16768336e-01 9.53159213e-01 4.86783028e-01 -2.56806612e-01 -1.04170835e+00 1.03832507e+00 6.59268558e-01 -4.66446340e-01 -1.77174345e-01 2.04024427e-02 1.95148021e-01 -3.31985891e-01 9.71021950e-02 -4.19459552e-01 8.42527226e-02 -7.69473374e-01 8.26346040e-01 2.01250494e-01 7.86995113e-01 -4.65667635e-01 5.37383854e-01 9.36282352e-02 -1.40693927e+00 -1.03065513e-01 -3.63445550e-01 1.65306777e-01 4.54702973e-02 3.16105694e-01 -8.31175447e-01 4.34303939e-01 8.65537465e-01 4.97393161e-01 -4.22308981e-01 1.06760490e+00 1.69276521e-01 5.25195062e-01 -4.60282594e-01 7.01399446e-02 4.93678935e-02 -3.63186896e-01 3.57518941e-01 1.11428010e+00 5.91415465e-01 4.98768747e-01 3.36630791e-02 1.02132487e+00 5.81112131e-02 5.54265320e-01 -9.68560159e-01 2.06781179e-02 -9.52911228e-02 1.17259312e+00 -3.34216505e-01 -2.16683932e-03 -7.30849862e-01 9.53474104e-01 -2.33886912e-01 1.00971736e-01 -7.82574177e-01 2.20701203e-01 5.88182807e-01 4.25565153e-01 3.16779017e-01 1.92424312e-01 9.79393423e-02 -1.11108887e+00 1.68191478e-01 -7.60411799e-01 7.52168715e-01 -1.03887904e+00 -1.04881310e+00 1.00938387e-01 1.88359961e-01 -9.45454836e-01 7.68215954e-02 -1.61153361e-01 -2.47440323e-01 6.00140989e-01 -1.33980215e+00 -1.51523364e+00 -7.70961583e-01 7.64542282e-01 4.48404372e-01 2.66857296e-01 1.06977701e+00 4.45831776e-01 -5.51459551e-01 5.66849887e-01 1.70583561e-01 2.32059255e-01 5.99747300e-01 -1.05181766e+00 -2.26176888e-01 5.45235217e-01 -1.06402859e-02 2.82555342e-01 7.30995834e-01 -8.92541647e-01 -1.85724616e+00 -1.40185273e+00 6.72441423e-01 -2.98415303e-01 3.88904400e-02 -5.39262354e-01 -3.58114094e-01 8.37984562e-01 3.44221503e-01 -1.65857092e-01 1.04136026e+00 -6.07130826e-01 -3.76163721e-01 -4.46473986e-01 -1.90948176e+00 3.98726314e-01 5.75096726e-01 -4.29394871e-01 -3.36456418e-01 5.10313034e-01 4.24298018e-01 -2.97740132e-01 -9.00094569e-01 -1.26656651e-01 8.53533506e-01 -6.56491876e-01 1.22948396e+00 -2.59835254e-02 2.42383227e-01 -3.58971000e-01 -4.69123095e-01 -1.01212716e+00 -2.35882580e-01 -1.34354368e-01 1.54446550e-02 1.56281149e+00 -1.24239318e-01 -3.87993127e-01 1.00280154e+00 1.04849780e+00 4.18653250e-01 1.84286237e-02 -4.74915653e-01 -1.93271458e-01 -4.67200339e-01 1.86098501e-01 9.17761564e-01 9.21036959e-01 7.34579936e-02 -1.74268067e-01 -6.61419332e-01 5.23063064e-01 1.06345510e+00 1.34131297e-01 5.03464341e-01 -4.85745639e-01 2.23991990e-01 5.51642179e-01 -4.70266223e-01 -3.90047163e-01 -4.74457681e-01 -9.16083872e-01 -1.77460209e-01 -1.42728412e+00 7.96202302e-01 2.11406246e-01 8.83941948e-02 4.00857747e-01 1.47946849e-01 4.73072648e-01 2.42101461e-01 -1.25792641e-02 -2.68189043e-01 7.77305290e-02 1.09255028e+00 -4.67554867e-01 -4.31758240e-02 -3.59479934e-01 -9.48431373e-01 4.21905577e-01 9.84927356e-01 -6.44915164e-01 -6.12302303e-01 -4.41795468e-01 -1.79811597e-01 8.77223164e-02 6.75397635e-01 -8.73849928e-01 1.88116595e-01 -1.64561719e-01 7.55613029e-01 -4.66058761e-01 2.66631544e-01 -1.59316897e+00 9.33566630e-01 6.23118818e-01 -4.69593287e-01 7.07720444e-02 1.16444670e-01 7.31654465e-01 1.65132344e-01 -2.35402420e-01 5.04075050e-01 -8.17248464e-01 -2.25656807e-01 1.72555700e-01 -2.70785481e-01 -4.22033876e-01 1.45309734e+00 -4.27872986e-01 -4.09386367e-01 -2.71785647e-01 -5.58509111e-01 1.79496676e-01 5.61816335e-01 2.00367525e-01 3.88245165e-01 -1.43678927e+00 -7.65010655e-01 4.57073659e-01 3.97692353e-01 -3.76384854e-01 5.85472941e-01 5.40993690e-01 -7.10283577e-01 5.98265767e-01 -3.77727866e-01 -4.94599193e-01 -1.45947623e+00 1.15642107e+00 1.95287973e-01 1.12156995e-01 -9.28661525e-01 -4.07646224e-02 5.79793096e-01 -8.91486779e-02 3.00549179e-01 -2.67256886e-01 -3.01581770e-01 -1.17301866e-02 7.61558771e-01 3.01170081e-01 -1.62447453e-01 -8.34513545e-01 -2.79272199e-01 6.04969561e-01 2.16537356e-01 4.41174835e-01 1.14935338e+00 -9.52501595e-01 2.22938851e-01 3.43715288e-02 1.24985135e+00 -2.32884273e-01 -1.16198027e+00 -8.38969722e-02 -7.01365352e-01 -7.04183877e-01 2.68545836e-01 -1.05656159e+00 -1.32650638e+00 6.10016465e-01 1.40892053e+00 5.36354929e-02 1.36442590e+00 -3.25525433e-01 9.03141022e-01 -2.76591092e-01 1.63657069e-01 -7.78594553e-01 -1.74325094e-01 -7.79507220e-01 6.90347493e-01 -1.23530936e+00 1.58685699e-01 -4.22351867e-01 -6.85886621e-01 7.73780048e-01 4.09464478e-01 1.75364837e-01 4.56320792e-01 5.85919134e-02 2.59491146e-01 -4.01620746e-01 -1.66093241e-02 2.20965326e-01 2.22375765e-02 9.99450684e-01 1.62305459e-01 5.73617041e-01 -1.85041837e-02 4.23881501e-01 -1.32682472e-01 4.04914558e-01 4.44934726e-01 7.75640965e-01 -2.14069467e-02 -4.97740418e-01 -8.08602691e-01 2.93501884e-01 -6.05488777e-01 9.30383131e-02 -1.73679560e-01 4.30957794e-01 2.12103471e-01 1.29997671e+00 2.14961674e-02 1.93918809e-01 4.47709262e-01 3.38194109e-02 4.21365649e-01 -3.18931669e-01 -7.85288811e-01 9.90371034e-02 9.80192423e-02 -4.93917346e-01 -6.03981078e-01 -6.87522769e-01 -4.75707293e-01 -5.26196480e-01 -1.84591353e-01 -1.48744255e-01 1.00468409e+00 3.67047995e-01 1.77600488e-01 3.09807733e-02 5.59232712e-01 -3.91301274e-01 -5.05532682e-01 -6.95021987e-01 -7.01202095e-01 9.78362679e-01 3.01876307e-01 -1.08721711e-01 -1.61475316e-01 4.11737949e-01]
[11.566555976867676, 4.403940200805664]
2e52054e-c3c4-4aa3-b298-7ab88b24646a
knn-classification-with-one-step-computation
2012.06047
null
https://arxiv.org/abs/2012.06047v2
https://arxiv.org/pdf/2012.06047v2.pdf
KNN Classification with One-step Computation
KNN classification is an improvisational learning mode, in which they are carried out only when a test data is predicted that set a suitable K value and search the K nearest neighbors from the whole training sample space, referred them to the lazy part of KNN classification. This lazy part has been the bottleneck problem of applying KNN classification due to the complete search of K nearest neighbors. In this paper, a one-step computation is proposed to replace the lazy part of KNN classification. The one-step computation actually transforms the lazy part to a matrix computation as follows. Given a test data, training samples are first applied to fit the test data with the least squares loss function. And then, a relationship matrix is generated by weighting all training samples according to their influence on the test data. Finally, a group lasso is employed to perform sparse learning of the relationship matrix. In this way, setting K value and searching K nearest neighbors are both integrated to a unified computation. In addition, a new classification rule is proposed for improving the performance of one-step KNN classification. The proposed approach is experimentally evaluated, and demonstrated that the one-step KNN classification is efficient and promising
['Jiaye Li', 'Shichao Zhang']
2020-12-09
null
null
null
null
['sparse-learning']
['methodology']
[ 2.95988500e-01 -2.23655269e-01 -4.66049433e-01 -4.17767763e-01 -7.06658542e-01 -2.12617651e-01 2.40796641e-01 2.80880611e-02 -3.70956421e-01 6.36567175e-01 -1.43262848e-01 -1.32636353e-01 -6.48490071e-01 -7.95647681e-01 -5.00391126e-01 -9.51711297e-01 3.00500870e-01 4.73234683e-01 9.56970174e-03 2.33512625e-01 2.93002069e-01 4.58044112e-01 -1.31387627e+00 3.02786529e-01 1.05563974e+00 1.28994727e+00 6.51911348e-02 2.65014082e-01 -1.76124811e-01 8.79527211e-01 -1.81549221e-01 -4.79380507e-03 4.43070650e-01 -6.67815268e-01 -5.34994304e-01 1.96546927e-01 3.43179554e-01 -1.31998941e-01 4.11691032e-02 1.10063148e+00 3.40950459e-01 5.50706148e-01 7.13751554e-01 -1.23641968e+00 -2.43676811e-01 7.18397021e-01 -5.13941050e-01 -1.52512178e-01 1.63671583e-01 -1.58444911e-01 9.28553641e-01 -1.26291764e+00 3.99311066e-01 8.59291673e-01 6.77832544e-01 2.26072401e-01 -1.30979717e+00 -7.70499468e-01 -9.96631086e-02 4.25704807e-01 -1.58987820e+00 -2.05392584e-01 1.03765714e+00 -4.61452425e-01 4.80259389e-01 4.22742069e-01 9.27633464e-01 4.51087505e-01 -1.25203699e-01 7.57864177e-01 5.46237946e-01 -5.23953319e-01 4.72120434e-01 2.58074939e-01 4.11566943e-01 6.08251691e-01 1.09793179e-01 -1.77835450e-01 -3.99736017e-01 -3.00785601e-01 4.66981888e-01 5.06699800e-01 -2.02985048e-01 -5.96021056e-01 -9.74537015e-01 9.24618363e-01 6.25313580e-01 2.84588665e-01 -5.36193073e-01 -2.18302652e-01 4.31306660e-01 2.52498806e-01 4.30869490e-01 1.71133190e-01 -2.99445033e-01 1.04173504e-01 -1.24439073e+00 7.08708316e-02 8.01617861e-01 4.69709098e-01 1.15695214e+00 -8.35669562e-02 2.09009815e-02 1.04798293e+00 7.94174448e-02 3.12777221e-01 7.70798624e-01 -7.05303013e-01 5.25786579e-01 1.07270479e+00 -1.23915493e-01 -1.46367824e+00 -4.03235942e-01 -6.29060388e-01 -1.11210680e+00 -1.05933353e-01 3.92367929e-01 3.80746052e-02 -5.85330009e-01 1.20211089e+00 5.97993076e-01 5.36495984e-01 5.33593595e-02 1.00664341e+00 4.97447968e-01 6.60224795e-01 -2.19279557e-01 -5.29704630e-01 8.74780834e-01 -1.13174319e+00 -4.10176337e-01 3.27147506e-02 8.92909765e-01 -7.45999217e-01 8.83513987e-01 4.77895647e-01 -5.84778070e-01 -5.26174128e-01 -8.05815041e-01 1.16004400e-01 -1.78041786e-01 4.94434744e-01 5.16982198e-01 2.38717794e-01 -4.60046053e-01 5.90812266e-01 -5.80792308e-01 -3.79870459e-03 2.51094550e-01 2.40510866e-01 -4.41334456e-01 -4.54366058e-01 -9.81025994e-01 4.67207372e-01 6.26654208e-01 6.40784919e-01 -4.95408475e-01 -7.89736271e-01 -8.38342905e-01 2.11102933e-01 6.75727725e-01 -2.48423278e-01 7.23552644e-01 -1.00076401e+00 -1.15159571e+00 5.07350326e-01 -2.43482918e-01 -3.74182135e-01 4.55282986e-01 -8.69684666e-02 -2.77496189e-01 -4.73963656e-02 1.67349920e-01 -4.94047292e-02 1.07633376e+00 -8.46014202e-01 -5.21781564e-01 -3.18026394e-01 -3.65888387e-01 1.75071478e-01 -4.36648011e-01 -4.14809555e-01 -3.54020596e-01 -6.40924931e-01 8.56462121e-01 -7.83687353e-01 -2.75154978e-01 -1.63711891e-01 -3.28126818e-01 -3.80616486e-01 8.89726460e-01 -5.53365111e-01 1.55282068e+00 -2.42233443e+00 2.80825883e-01 8.72722566e-01 2.54696101e-01 3.89571637e-01 -7.76712373e-02 2.69501328e-01 -3.04751724e-01 -2.92616040e-01 -3.13399583e-01 -1.13849625e-01 -2.58581728e-01 2.13937506e-01 -2.72369534e-01 4.03815061e-01 -1.66049659e-01 6.17995739e-01 -8.27864647e-01 -5.75855434e-01 1.83496222e-01 2.56311923e-01 -7.00473249e-01 1.41493142e-01 -5.96864969e-02 2.09428564e-01 -5.47413468e-01 5.93451262e-01 7.11288154e-01 -2.97122091e-01 1.62466243e-01 -5.48825920e-01 -2.79619507e-02 -9.75196809e-02 -1.62530410e+00 1.21050000e+00 -2.04243049e-01 3.51293422e-02 -1.26515090e-01 -1.53612304e+00 1.23119867e+00 4.63170707e-02 5.45381367e-01 -2.20145047e-01 -3.23024392e-02 5.08504331e-01 -1.23737670e-01 -5.45108438e-01 -1.13338485e-01 -2.08493341e-02 6.55995160e-02 2.52241939e-01 -2.20422775e-01 5.95381893e-02 1.09945923e-01 5.80182299e-02 7.27586031e-01 -8.30031782e-02 5.95498264e-01 -9.72752422e-02 9.92422402e-01 1.74420461e-01 9.45035338e-01 6.45294547e-01 1.24193117e-01 3.55400205e-01 5.47143996e-01 -9.99372900e-01 -7.54052997e-01 -6.07450545e-01 -1.45990532e-02 8.14010143e-01 -3.51780131e-02 -4.59669650e-01 -5.80408871e-01 -9.17068243e-01 9.58163068e-02 3.62225801e-01 -6.82889998e-01 -3.90119255e-01 -5.47490954e-01 -6.09526575e-01 1.71544235e-02 2.06613898e-01 5.04715800e-01 -1.03965008e+00 -1.71835035e-01 1.75725013e-01 2.71215588e-02 -6.24253452e-01 -5.82512617e-01 3.08869421e-01 -9.06931043e-01 -1.31623185e+00 -5.89974582e-01 -7.67286599e-01 9.32870030e-01 2.86558002e-01 3.81099641e-01 1.45537511e-01 -2.19008341e-01 -1.53687239e-01 -6.05121076e-01 1.35954976e-01 -9.88714471e-02 8.56622159e-02 2.14554220e-01 6.98595285e-01 4.26710099e-01 -5.73063791e-01 -5.81869841e-01 3.36150020e-01 -6.95930183e-01 -2.34308057e-02 8.52731228e-01 1.13888288e+00 9.53483105e-01 2.96836019e-01 4.60254312e-01 -1.00876093e+00 3.83099884e-01 -5.34791052e-01 -7.32927263e-01 3.27189088e-01 -6.92582905e-01 -1.36380836e-01 1.17306733e+00 -6.61898017e-01 -5.82988381e-01 4.38827634e-01 1.82379827e-01 -7.79373109e-01 2.67492920e-01 8.41767967e-01 -1.55023798e-01 -1.27300024e-01 4.93716389e-01 6.49366438e-01 2.46366099e-01 -5.47580779e-01 6.69046864e-02 6.12487197e-01 2.36995518e-01 -3.21770012e-01 8.60440016e-01 1.32679939e-01 2.23792523e-01 -6.69440746e-01 -9.66533959e-01 -5.85899830e-01 -6.42901063e-01 -1.21012546e-01 5.61409354e-01 -6.06461346e-01 -6.47198141e-01 3.93867999e-01 -5.92254281e-01 -1.26801550e-01 -4.66616511e-01 8.65266502e-01 -2.60726810e-01 3.28109771e-01 -3.88738096e-01 -6.67816222e-01 -3.72161716e-01 -1.08813393e+00 5.90184331e-01 1.86494738e-01 -9.01820064e-02 -7.62093842e-01 -5.74481301e-02 4.21176791e-01 2.25411713e-01 -2.67605036e-01 9.14356470e-01 -1.14356220e+00 -3.83838415e-01 -5.61522543e-01 -2.33330995e-01 6.32942617e-01 2.46552333e-01 -2.13950276e-01 -5.09023607e-01 -4.02842760e-01 2.46493891e-01 -2.90398806e-01 7.32237875e-01 2.74260104e-01 1.35648608e+00 -4.28615689e-01 -2.69051045e-01 8.97189975e-01 1.50832582e+00 2.87926465e-01 2.48474061e-01 1.51830867e-01 8.36365163e-01 2.22156987e-01 8.35406005e-01 5.15649974e-01 2.81888735e-03 5.56958437e-01 1.34590924e-01 4.20053601e-02 2.48759657e-01 -2.73570001e-01 8.96158740e-02 1.11255836e+00 2.15283424e-01 3.20215702e-01 -9.13702786e-01 5.32729402e-02 -2.09629965e+00 -9.37627435e-01 -3.85659337e-02 2.27917981e+00 5.75913846e-01 -7.10656196e-02 -1.60988420e-01 6.14834905e-01 7.42248178e-01 1.18324988e-01 -7.72544980e-01 6.04323223e-02 1.19507127e-01 -1.84889272e-01 1.34875551e-01 6.39127195e-01 -9.34059143e-01 7.82393277e-01 5.34279442e+00 1.26076293e+00 -1.31634772e+00 -2.97046095e-01 5.74810207e-01 -8.85328501e-02 -7.84536600e-02 1.82225406e-01 -9.52053249e-01 6.30099773e-01 3.08099866e-01 -6.54860027e-03 6.70579433e-01 1.12560570e+00 3.90909851e-01 -1.83516234e-01 -1.07818556e+00 1.22298694e+00 2.03011826e-01 -1.20587337e+00 2.46933758e-01 -1.66566491e-01 5.21010816e-01 -4.79220778e-01 -2.69520044e-01 3.79693091e-01 -2.30842471e-01 -6.98075056e-01 3.71232927e-01 6.28941834e-01 4.51860636e-01 -7.41735220e-01 7.32489765e-01 7.72777975e-01 -1.18832397e+00 -3.77793163e-01 -6.06170058e-01 -3.46039571e-02 -2.10453436e-01 1.18704247e+00 -8.99493575e-01 6.11327112e-01 6.21143043e-01 8.55598748e-01 -4.32995498e-01 1.10560250e+00 -2.49173641e-02 5.94792604e-01 -3.77214283e-01 -1.04937658e-01 3.39241207e-01 -8.02463472e-01 4.24584687e-01 7.23523855e-01 4.43823218e-01 1.97131529e-01 6.09868765e-01 5.78761578e-01 2.26084478e-02 5.14368594e-01 -3.30097020e-01 3.03970605e-01 5.48436522e-01 1.29140449e+00 -7.06345081e-01 -3.29594553e-01 -2.70033658e-01 9.25229013e-01 5.97463369e-01 3.76540244e-01 -6.07187390e-01 -4.27917719e-01 8.68353322e-02 -1.09409234e-02 3.71513367e-01 6.89306483e-02 -1.58780426e-01 -1.18760777e+00 2.95000345e-01 -8.67187858e-01 5.58472633e-01 -5.27515054e-01 -1.35051394e+00 3.95988762e-01 -3.00160170e-01 -1.59360921e+00 -1.80745572e-01 -3.76024961e-01 -4.27132666e-01 8.35988402e-01 -1.28292620e+00 -9.62261438e-01 -3.81680638e-01 7.23562419e-01 3.48186791e-01 -2.67404765e-01 7.34141111e-01 4.31532413e-01 -8.87237072e-01 6.19415462e-01 3.52856874e-01 2.34742224e-01 7.36049175e-01 -8.46633196e-01 -6.19874179e-01 7.67419457e-01 -3.12566794e-02 6.27149522e-01 2.11074367e-01 -6.25970185e-01 -1.21921706e+00 -1.18243408e+00 9.88418221e-01 1.49755985e-01 5.36450148e-01 -2.26221055e-01 -9.41586912e-01 5.92877209e-01 -7.06084371e-01 5.35153866e-01 8.86246026e-01 -2.52374243e-02 -2.75722951e-01 -7.68636107e-01 -1.00476444e+00 3.69492173e-01 5.61078727e-01 -4.17591751e-01 -4.52854007e-01 4.13734376e-01 5.26033819e-01 -2.53282815e-01 -8.28430414e-01 4.03861910e-01 6.00353122e-01 -6.92263186e-01 8.28492284e-01 -5.81176281e-01 3.15135390e-01 -5.82565069e-01 -2.69493818e-01 -1.04742968e+00 -3.89883578e-01 -3.64608198e-01 -6.28745109e-02 1.17911649e+00 4.52229470e-01 -6.17434740e-01 1.10309529e+00 5.73527098e-01 6.62509501e-02 -1.21285415e+00 -9.45546031e-01 -5.68926275e-01 -4.49035525e-01 -3.25641274e-01 4.83538270e-01 1.13499153e+00 -8.18906873e-02 2.30964065e-01 -5.95456839e-01 7.45253563e-02 7.39218771e-01 5.73080182e-01 6.96733832e-01 -1.10879540e+00 -2.77867556e-01 -1.22388057e-01 -4.06799555e-01 -8.69971395e-01 1.31981179e-01 -1.12843645e+00 -2.24053338e-01 -1.10038030e+00 3.56397599e-01 -8.07961762e-01 -4.82396334e-01 6.05503201e-01 -2.88217813e-01 1.97151840e-01 2.42139921e-01 5.26643872e-01 -5.70192933e-01 3.90457779e-01 1.05528808e+00 -3.71558405e-02 -5.26974320e-01 3.33747178e-01 -4.36043352e-01 6.59253240e-01 6.08906031e-01 -5.93125403e-01 -6.18354619e-01 -1.39007375e-01 1.24178424e-01 2.61313766e-01 2.49530897e-01 -1.00686049e+00 7.62167871e-01 -2.04313532e-01 5.88737190e-01 -6.74128413e-01 1.26835912e-01 -1.16531479e+00 1.16182737e-01 5.68465948e-01 -3.77247781e-01 -3.66480798e-01 -4.62980539e-01 4.57722396e-01 -4.34043497e-01 -4.27448213e-01 8.28980863e-01 1.25433832e-01 -6.10683203e-01 4.42389369e-01 4.93227281e-02 -1.81063011e-01 1.17635286e+00 -4.25047338e-01 4.16589111e-01 -1.53012320e-01 -1.00015485e+00 3.34731758e-01 -9.90972016e-03 -1.33348405e-01 8.47408056e-01 -1.46959424e+00 -5.57918191e-01 6.65407121e-01 -7.56469667e-02 2.78430313e-01 2.84217656e-01 9.38899636e-01 -3.32249820e-01 2.51048833e-01 1.04378507e-01 -6.77743673e-01 -1.10837722e+00 6.28874242e-01 2.18257055e-01 -4.80848014e-01 -5.24909496e-01 6.99567556e-01 -1.76060125e-01 -5.69991529e-01 3.60601008e-01 -1.00368746e-01 -3.88416857e-01 3.98887813e-01 5.67384422e-01 4.98307168e-01 6.10310175e-02 -2.67168611e-01 -3.23744833e-01 8.63327801e-01 -3.62978578e-02 4.51777905e-01 1.46130300e+00 1.52323367e-02 -6.88867807e-01 5.16020954e-01 1.44949889e+00 -4.16746251e-02 -8.75685394e-01 -4.46633369e-01 -9.77673531e-02 -5.45502722e-01 -1.40276045e-01 -5.08508503e-01 -1.20874143e+00 4.49234068e-01 3.23402017e-01 4.94570732e-02 1.36037159e+00 -1.95821688e-01 6.73449516e-01 7.31058300e-01 1.39022574e-01 -9.38474834e-01 -9.08864737e-02 5.71537375e-01 6.66278780e-01 -1.24923480e+00 1.37946337e-01 -6.63776577e-01 -2.79196978e-01 1.15299273e+00 7.02789903e-01 -2.32272282e-01 9.33536351e-01 -2.06450336e-02 -2.19028711e-01 7.82672837e-02 -3.65182281e-01 4.03425753e-01 3.06123048e-01 1.09659657e-01 5.35125621e-02 -1.16261609e-01 -6.43572986e-01 9.38846171e-01 -1.56337932e-01 3.15010160e-01 -6.18726341e-03 4.81705815e-01 -4.88766640e-01 -9.47499573e-01 -3.46220285e-01 7.34575748e-01 -1.34884909e-01 -4.06694226e-02 -5.42568155e-02 5.19712567e-01 1.84069633e-01 6.72674954e-01 -9.85898972e-02 -5.52352428e-01 3.10395628e-01 2.25242600e-01 -7.17066079e-02 -4.90545005e-01 -4.02608901e-01 2.67055947e-02 -3.71850371e-01 -6.97484612e-01 -1.33995101e-01 -4.18097198e-01 -1.18513227e+00 -9.53626037e-02 -3.55751276e-01 6.34339631e-01 4.21336681e-01 8.46245050e-01 1.56855509e-01 8.78833681e-02 1.07825887e+00 -7.07959294e-01 -9.21266854e-01 -5.34037173e-01 -7.18540549e-01 5.32974660e-01 2.47249231e-01 -3.52905720e-01 -7.05316007e-01 -1.51185691e-01]
[8.878568649291992, 3.7284891605377197]
5de5b855-5d66-4193-8f71-774c8cbbb53f
sentiment-analysis-of-twitter-data-for
1610.09225
null
http://arxiv.org/abs/1610.09225v1
http://arxiv.org/pdf/1610.09225v1.pdf
Sentiment Analysis of Twitter Data for Predicting Stock Market Movements
Predicting stock market movements is a well-known problem of interest. Now-a-days social media is perfectly representing the public sentiment and opinion about current events. Especially, twitter has attracted a lot of attention from researchers for studying the public sentiments. Stock market prediction on the basis of public sentiments expressed on twitter has been an intriguing field of research. Previous studies have concluded that the aggregate public mood collected from twitter may well be correlated with Dow Jones Industrial Average Index (DJIA). The thesis of this work is to observe how well the changes in stock prices of a company, the rises and falls, are correlated with the public opinions being expressed in tweets about that company. Understanding author's opinion from a piece of text is the objective of sentiment analysis. The present paper have employed two different textual representations, Word2vec and N-gram, for analyzing the public sentiments in tweets. In this paper, we have applied sentiment analysis and supervised machine learning principles to the tweets extracted from twitter and analyze the correlation between stock market movements of a company and sentiments in tweets. In an elaborate way, positive news and tweets in social media about a company would definitely encourage people to invest in the stocks of that company and as a result the stock price of that company would increase. At the end of the paper, it is shown that a strong correlation exists between the rise and falls in stock prices with the public sentiments in tweets.
['Kamal Nayan Reddy Challa', 'Babita Majhi', 'Ganapati Panda', 'Venkata Sasank Pagolu']
2016-10-28
null
null
null
null
['stock-market-prediction']
['time-series']
[-5.87009907e-01 -3.37210238e-01 -4.90469873e-01 -2.35862359e-01 4.61693220e-02 -6.01938605e-01 8.15326989e-01 6.21864557e-01 -3.69059443e-01 6.40487909e-01 6.29202604e-01 -2.70469040e-01 3.02771717e-01 -1.25802696e+00 -1.55138552e-01 -5.24848700e-01 2.15927199e-01 -5.54757416e-02 -1.78898182e-02 -9.71443534e-01 7.83313453e-01 4.29596931e-01 -1.62827063e+00 1.06663726e-01 -4.83361557e-02 1.14140332e+00 -1.71458144e-02 3.21160495e-01 -7.53612041e-01 1.43683052e+00 -6.98748291e-01 -4.02079463e-01 1.40384600e-01 -3.25437307e-01 -5.42297423e-01 1.32453263e-01 -3.45757902e-01 2.90718824e-01 1.00703418e-01 1.01442695e+00 1.56617388e-01 -2.36418515e-01 4.24600840e-01 -1.13697898e+00 -4.01630729e-01 8.57379019e-01 -7.78366208e-01 6.72144473e-01 1.03266910e-01 -3.45232695e-01 1.42109632e+00 -7.25265265e-01 5.96657038e-01 6.69152975e-01 3.93375069e-01 -3.28276485e-01 -4.77602005e-01 -8.82527590e-01 1.58476621e-01 -6.19536154e-02 -8.89377356e-01 2.77457505e-01 9.11938071e-01 -8.87555957e-01 6.82374418e-01 3.40077370e-01 1.12662363e+00 4.30069804e-01 7.09557354e-01 5.08508921e-01 1.31255960e+00 -3.34305614e-01 1.57347709e-01 8.34555447e-01 4.17529434e-01 5.87145574e-02 4.85517621e-01 -3.12454998e-01 -5.46308756e-01 -6.27075285e-02 -7.56687531e-03 3.88336748e-01 7.29141533e-02 4.37742203e-01 -1.03975332e+00 1.40362644e+00 2.65721172e-01 1.04454410e+00 -7.96846151e-01 -2.07795799e-01 6.85081542e-01 5.35803974e-01 1.19704401e+00 3.85574013e-01 -6.45947158e-01 -3.95104468e-01 -8.34571898e-01 3.17584187e-01 9.67198253e-01 2.56759226e-01 7.87698269e-01 1.77939266e-01 4.98130739e-01 3.28885257e-01 3.38648140e-01 7.07353890e-01 1.15545249e+00 7.36424774e-02 2.31771857e-01 1.08199298e+00 1.07399777e-01 -1.89230287e+00 -4.61438298e-01 -6.64750218e-01 -3.32782865e-01 1.51430249e-01 4.58068624e-02 -5.65671504e-01 -1.11359037e-01 1.17622197e+00 1.20025374e-01 -8.98915157e-02 2.45621413e-01 4.75864589e-01 7.58543193e-01 1.07181835e+00 -1.07095465e-01 -5.42738914e-01 1.31104100e+00 -3.45425576e-01 -1.16572165e+00 -2.13602901e-01 5.43787897e-01 -1.10880649e+00 3.90241355e-01 9.96418223e-02 -7.07984746e-01 -4.18462455e-01 -1.02201605e+00 6.84950352e-01 -1.08047819e+00 -3.32701296e-01 6.44832730e-01 5.62850893e-01 -6.00327611e-01 6.31494582e-01 -5.28156042e-01 -1.53453037e-01 -2.04382036e-02 2.76621372e-01 5.24526797e-02 7.14511514e-01 -1.36130917e+00 9.68664885e-01 4.88764532e-02 -2.52145916e-01 3.58254701e-01 -3.18723619e-01 -6.23334467e-01 -2.77765155e-01 2.59888601e-02 4.13189121e-02 1.04737008e+00 -1.33974421e+00 -1.01601410e+00 8.94423485e-01 -3.95162590e-02 -4.98794675e-01 1.21502727e-01 2.27652080e-02 -8.82779956e-01 -3.41143131e-01 3.51406634e-01 -3.05591673e-01 5.16285360e-01 -7.11042285e-01 -1.07456946e+00 -5.98705530e-01 -1.12027586e-01 -1.40610531e-01 -5.87885916e-01 4.22437638e-01 2.56870747e-01 -7.66356289e-01 7.77011588e-02 -9.04313684e-01 -1.47100702e-01 -1.19430542e+00 -2.28474483e-01 -3.93778056e-01 8.25332046e-01 -3.22091371e-01 1.35536397e+00 -1.98041189e+00 -5.41199386e-01 5.30115426e-01 -5.29075740e-03 -9.58548039e-02 6.36427045e-01 8.36017370e-01 -2.91632593e-01 3.56065959e-01 2.73603976e-01 3.66763681e-01 -1.44918477e-02 2.54539885e-02 -8.61740351e-01 6.53500080e-01 9.84079167e-02 5.81836700e-01 -6.29995286e-01 1.86298992e-02 5.68348579e-02 3.46938133e-01 -6.27224222e-02 -1.21177264e-01 -2.20946632e-02 1.19822338e-01 -6.40793562e-01 3.19570720e-01 4.31232810e-01 -3.13373655e-01 -1.03028268e-01 -7.50827715e-02 -9.72120106e-01 3.81052464e-01 -9.24835086e-01 3.48922163e-01 -2.13770658e-01 1.08216190e+00 -5.77836215e-01 -7.04489112e-01 1.38689268e+00 3.89049977e-01 7.73822308e-01 -7.04484820e-01 7.11283207e-01 3.44364703e-01 2.92495135e-02 -2.72434622e-01 8.79641235e-01 -6.04290128e-01 -2.71624237e-01 7.66782761e-01 -7.11530328e-01 -2.26414382e-01 5.99988937e-01 -1.19750887e-01 3.45383018e-01 -4.72788960e-01 6.68453872e-01 -2.58117706e-01 6.66992724e-01 1.80467039e-01 7.74563253e-02 -1.28402993e-01 1.37326136e-01 5.64595722e-02 8.03670347e-01 -6.61931574e-01 -6.99918747e-01 -1.19366772e-01 -1.53541222e-01 1.01319122e+00 -1.21124871e-01 -3.22381645e-01 -3.31619680e-01 -4.95224968e-02 1.24569356e-01 6.92340851e-01 -7.57645190e-01 3.99251401e-01 -2.59806454e-01 -1.20981383e+00 -2.21558139e-01 2.18569294e-01 2.93712080e-01 -1.07325220e+00 -8.51670861e-01 3.05624276e-01 1.56697080e-01 -9.31407094e-01 7.16761723e-02 3.20035785e-01 -6.62450671e-01 -1.05152035e+00 -7.05001831e-01 -5.99446893e-01 4.65025693e-01 1.94047987e-01 9.73564804e-01 -2.62441397e-01 2.40580007e-01 3.23614515e-02 -6.63078487e-01 -1.32080209e+00 -2.75677234e-01 1.66856393e-01 2.90324334e-02 4.03518796e-01 7.09762692e-01 -2.81677276e-01 -2.45386660e-01 9.26876068e-02 -9.71837521e-01 -3.60976338e-01 8.55831429e-02 5.78115657e-02 1.74668983e-01 5.73724210e-01 7.37864494e-01 -1.09616852e+00 9.04182792e-01 -9.91026878e-01 -6.61571622e-01 -5.12098312e-01 -7.95795560e-01 -1.98918596e-01 2.50376582e-01 -2.56465320e-02 -7.28794158e-01 -4.50776786e-01 -7.42370859e-02 3.80282551e-01 2.28283033e-01 1.15021825e+00 5.64264297e-01 3.76915991e-01 2.20212191e-01 2.54905522e-01 1.91568643e-01 -1.69205237e-02 -1.27513573e-01 7.66699553e-01 -8.74368101e-02 3.05740416e-01 8.35344613e-01 7.08896399e-01 -2.69113749e-01 -9.38181162e-01 -1.11850560e+00 -8.33859146e-01 -2.70780355e-01 -5.08261144e-01 7.93246031e-01 -1.07768190e+00 -7.43452311e-01 6.05386436e-01 -7.28463531e-01 4.58085477e-01 -2.76314527e-01 6.55962527e-01 -5.85372001e-02 -2.29737252e-01 -3.45823258e-01 -1.04753721e+00 -3.75079125e-01 -9.39613104e-01 2.64734209e-01 3.24116588e-01 -7.55107403e-01 -1.39342666e+00 4.38576460e-01 1.96132109e-01 6.84973836e-01 5.89982629e-01 5.20980179e-01 -1.18958187e+00 -1.46709189e-01 -8.79336417e-01 3.68983448e-02 4.42894042e-01 6.28359318e-01 2.81885237e-01 -7.82195508e-01 3.67717333e-02 4.66332048e-01 5.91035187e-02 4.34229493e-01 3.65432501e-01 2.72482187e-01 -3.54155958e-01 7.83477724e-02 -9.60948542e-02 1.50336051e+00 4.01246160e-01 4.11441177e-01 1.15455854e+00 3.15519065e-01 7.27711320e-01 6.33724451e-01 1.06864762e+00 2.80648530e-01 6.49969280e-02 4.54152912e-01 -1.93885081e-02 8.03311348e-01 1.56962022e-01 7.92580307e-01 1.38327515e+00 -1.68573976e-01 2.03382550e-03 -7.60486186e-01 4.46548790e-01 -1.41816151e+00 -1.14912808e+00 -6.84672475e-01 1.54351294e+00 5.97098410e-01 4.16458994e-01 3.78120542e-01 6.59936309e-01 6.57340527e-01 7.58055270e-01 -9.63572562e-02 -5.03972352e-01 -4.45703506e-01 1.68296427e-01 9.65468705e-01 1.67386442e-01 -9.47210670e-01 5.52694798e-01 5.48077965e+00 2.24264309e-01 -1.70248902e+00 -2.73860425e-01 7.43965328e-01 8.34049061e-02 -4.86540496e-01 -2.33396649e-01 -9.90607083e-01 5.90929449e-01 1.13215363e+00 -7.66710401e-01 -4.24123406e-01 8.91360879e-01 5.09685993e-01 -2.96348184e-01 -3.83373588e-01 6.45951688e-01 2.58545816e-01 -1.53766179e+00 -1.52600408e-01 4.01051551e-01 8.76597762e-01 1.23712711e-01 4.46984082e-01 -1.38227269e-02 6.67267218e-02 -4.67739433e-01 8.03871393e-01 4.37608451e-01 -1.52497545e-01 -1.15617442e+00 1.42348111e+00 2.11273089e-01 -1.09441376e+00 -5.36728418e-03 -1.67576194e-01 -7.05646515e-01 2.76663601e-01 9.63741124e-01 -8.65740061e-01 1.77554324e-01 5.82355261e-01 1.06439734e+00 -3.41438055e-01 2.58798748e-01 5.28862700e-02 5.34729302e-01 2.54847948e-02 -8.25177670e-01 6.76467657e-01 -5.12603223e-01 2.07970530e-01 1.09326756e+00 2.97191530e-01 -5.66979572e-02 -2.33998165e-01 3.93447697e-01 7.01530725e-02 8.26440454e-01 -9.25683200e-01 -6.47786200e-01 -1.79936081e-01 1.04305017e+00 -9.87638295e-01 -3.88435364e-01 -8.48750889e-01 2.76230633e-01 -3.80644858e-01 -1.67072237e-01 -4.69975740e-01 -3.14061493e-01 7.66916394e-01 5.10599673e-01 2.77857661e-01 8.01900998e-02 -4.48536932e-01 -7.91312516e-01 -1.46370068e-01 -6.24667466e-01 9.72568616e-03 -4.81039017e-01 -1.34733796e+00 7.73276508e-01 -3.91479135e-01 -1.35494578e+00 -1.76108137e-01 -5.81526160e-01 -8.67387176e-01 7.91491926e-01 -1.78785574e+00 -4.23511922e-01 1.61404133e-01 4.25886899e-01 4.25837398e-01 -5.98137140e-01 6.04032755e-01 -8.71321261e-02 -3.28727514e-01 -3.48912776e-01 1.66079521e-01 4.57479537e-01 2.87879795e-01 -1.14983606e+00 3.03897738e-01 3.79836380e-01 6.51170835e-02 4.59930003e-01 1.04359972e+00 -8.27196181e-01 -1.01508510e+00 -8.25423241e-01 1.55016816e+00 -2.80168980e-01 1.69270825e+00 5.69640696e-01 -4.45294023e-01 5.79344451e-01 6.85628355e-01 -5.60057104e-01 1.23758328e+00 -1.47482306e-01 -1.18488772e-03 -1.55248418e-01 -7.07615793e-01 5.63354075e-01 -3.40246379e-01 -4.20416117e-01 -8.41567695e-01 5.27500451e-01 3.43007922e-01 1.21860586e-01 -8.17194164e-01 -1.12485148e-01 4.70259666e-01 -1.11184299e+00 5.70006788e-01 -3.87763858e-01 5.62423170e-01 -1.90824598e-01 -1.65869087e-01 -1.46896374e+00 4.68703620e-02 -2.32509926e-01 8.11101019e-01 1.13523114e+00 6.87879145e-01 -1.00503123e+00 8.08384955e-01 5.59263766e-01 4.37255919e-01 -5.87817669e-01 -2.75123656e-01 -2.90412575e-01 1.26303703e-01 -5.68076193e-01 6.46755993e-01 1.10219562e+00 4.62101072e-01 4.17672217e-01 -2.22110376e-01 -5.18056490e-02 -7.12501705e-02 3.88746500e-01 9.04568791e-01 -1.43969965e+00 2.85501063e-01 -5.98391354e-01 -6.26837254e-01 -2.77550757e-01 2.91229375e-02 -7.29508162e-01 -6.85762525e-01 -1.29262781e+00 -2.74966925e-01 8.69054347e-03 -2.83907354e-01 -2.04344094e-01 2.69656837e-01 2.00841397e-01 3.40433568e-01 3.12753230e-01 9.23585966e-02 1.33295298e-01 1.21841681e+00 -1.44218609e-01 -3.25053096e-01 3.89117390e-01 -9.36074078e-01 1.10347033e+00 1.04662883e+00 -4.21650261e-01 -3.06264218e-02 4.53039646e-01 1.39026690e+00 -1.19789727e-01 -3.12656730e-01 -7.36369431e-01 1.71337754e-01 -2.01405704e-01 1.77748874e-01 -1.11090887e+00 1.25696480e-01 -9.64016616e-01 1.09966502e-01 5.84361196e-01 -3.46204787e-01 8.14553857e-01 9.03778896e-03 2.22747490e-01 -9.36618268e-01 -3.40452462e-01 4.09664482e-01 -3.23218524e-01 -7.08772302e-01 3.41368467e-02 -9.40054953e-01 -1.96149051e-01 1.16320217e+00 -2.76701212e-01 -8.43414813e-02 -7.16583550e-01 -4.52002108e-01 -5.11131361e-02 1.11879624e-01 5.47632754e-01 3.34401488e-01 -9.98172402e-01 -7.70504475e-01 -2.84943487e-02 -4.43305410e-02 -6.62552118e-01 -1.76112503e-01 7.99073398e-01 -6.34256661e-01 3.59875321e-01 -1.29744977e-01 3.87272835e-02 -1.16942608e+00 3.02232087e-01 -1.06081687e-01 -2.27734342e-01 -1.20883927e-01 6.45584524e-01 -2.97043413e-01 1.32277995e-01 -2.07907081e-01 -5.39442837e-01 -1.05600691e+00 1.42906928e+00 7.81241775e-01 3.38106781e-01 6.41513094e-02 -1.28967977e+00 -1.85779914e-01 7.74821341e-01 -1.19703725e-01 -7.11798519e-02 1.83435249e+00 -4.73362692e-02 -4.95977253e-01 1.37443149e+00 1.41398227e+00 5.36653936e-01 -2.12162286e-01 4.96859774e-02 5.26495159e-01 -3.26704174e-01 2.38367587e-01 -2.24314407e-01 -1.38648856e+00 3.41556609e-01 2.53251642e-01 1.14794409e+00 7.32330620e-01 7.63573945e-02 9.08774972e-01 2.23032773e-01 6.95032179e-02 -1.38596976e+00 -1.01743713e-01 7.37836003e-01 6.80933595e-01 -1.40680981e+00 2.29093358e-01 -1.51373997e-01 -1.07881284e+00 1.31609559e+00 -2.11878583e-01 -3.55207145e-01 1.38766527e+00 3.38625878e-01 5.14664829e-01 -5.56887269e-01 -4.82217312e-01 -3.58393431e-01 -5.20380735e-02 -1.12840794e-01 9.11220491e-01 2.87296623e-01 -5.51210403e-01 6.40680432e-01 -9.32621419e-01 -6.93365857e-02 9.68161345e-01 1.02676880e+00 -8.22151959e-01 -9.44201887e-01 -5.55780888e-01 5.36716163e-01 -1.25499856e+00 1.29439123e-02 -5.20083010e-01 9.57199514e-01 -4.24715644e-03 1.07763445e+00 5.01985669e-01 -4.42583978e-01 3.29837710e-01 1.43332481e-02 -5.28403878e-01 -5.68269670e-01 -9.93851662e-01 2.19887737e-02 2.94581708e-02 -1.69618484e-02 -1.03945208e+00 -9.85607266e-01 -1.31625354e+00 -6.26903117e-01 -3.02511513e-01 4.43668604e-01 1.13811922e+00 1.05211031e+00 -2.03914359e-01 4.93667394e-01 1.19444501e+00 -4.59104449e-01 -2.56643683e-01 -8.90691638e-01 -1.01533532e+00 3.49696726e-01 4.52722639e-01 -2.84213930e-01 -7.17462540e-01 7.46722594e-02]
[4.52412223815918, 4.401846408843994]
6b42a885-5ea4-4a54-81e9-677dd95e75fc
occupancy-networks-learning-3d-reconstruction
1812.03828
null
http://arxiv.org/abs/1812.03828v2
http://arxiv.org/pdf/1812.03828v2.pdf
Occupancy Networks: Learning 3D Reconstruction in Function Space
With the advent of deep neural networks, learning-based approaches for 3D reconstruction have gained popularity. However, unlike for images, in 3D there is no canonical representation which is both computationally and memory efficient yet allows for representing high-resolution geometry of arbitrary topology. Many of the state-of-the-art learning-based 3D reconstruction approaches can hence only represent very coarse 3D geometry or are limited to a restricted domain. In this paper, we propose Occupancy Networks, a new representation for learning-based 3D reconstruction methods. Occupancy networks implicitly represent the 3D surface as the continuous decision boundary of a deep neural network classifier. In contrast to existing approaches, our representation encodes a description of the 3D output at infinite resolution without excessive memory footprint. We validate that our representation can efficiently encode 3D structure and can be inferred from various kinds of input. Our experiments demonstrate competitive results, both qualitatively and quantitatively, for the challenging tasks of 3D reconstruction from single images, noisy point clouds and coarse discrete voxel grids. We believe that occupancy networks will become a useful tool in a wide variety of learning-based 3D tasks.
['Sebastian Nowozin', 'Michael Niemeyer', 'Lars Mescheder', 'Michael Oechsle', 'Andreas Geiger']
2018-12-10
occupancy-networks-learning-3d-reconstruction-1
http://openaccess.thecvf.com/content_CVPR_2019/html/Mescheder_Occupancy_Networks_Learning_3D_Reconstruction_in_Function_Space_CVPR_2019_paper.html
http://openaccess.thecvf.com/content_CVPR_2019/papers/Mescheder_Occupancy_Networks_Learning_3D_Reconstruction_in_Function_Space_CVPR_2019_paper.pdf
cvpr-2019-6
['3d-shape-representation']
['computer-vision']
[-3.52107920e-02 9.98338908e-02 -2.13734619e-02 -1.94531620e-01 -9.10866737e-01 -3.05608094e-01 7.28127182e-01 2.48576865e-01 -1.87988982e-01 5.00687480e-01 4.93807159e-02 -4.33982581e-01 -5.88852987e-02 -1.10597813e+00 -1.14877570e+00 -4.02876645e-01 -1.93364501e-01 1.16936994e+00 2.49939859e-01 1.77371785e-01 2.92480230e-01 1.22232866e+00 -1.87191164e+00 2.61405766e-01 2.95262575e-01 1.30281126e+00 1.13380358e-01 3.34723294e-01 -5.86511612e-01 3.96745831e-01 -2.28642389e-01 6.17551029e-01 3.69181663e-01 1.61372349e-01 -8.09427202e-01 4.48211543e-02 6.64528370e-01 -4.31348056e-01 -4.81857955e-01 6.63060606e-01 4.03281003e-01 1.83666766e-01 1.06677341e+00 -6.68947697e-01 -3.83252949e-01 -6.01095930e-02 -3.37683737e-01 -1.01818256e-01 4.05619204e-01 -5.38360439e-02 6.66046858e-01 -1.29727554e+00 6.88974857e-01 1.30410457e+00 9.04490530e-01 3.53426665e-01 -1.42555904e+00 -2.72250026e-01 -8.79772287e-03 -1.41130388e-01 -1.51585412e+00 -2.58252531e-01 9.25205588e-01 -5.78631580e-01 1.52117205e+00 2.02353925e-01 1.09950781e+00 8.71887803e-01 2.34078854e-01 6.80725217e-01 1.25925744e+00 -4.25721973e-01 5.33882082e-01 -3.39799732e-01 -1.77921623e-01 8.86873662e-01 1.08399309e-01 1.74943775e-01 -4.93154556e-01 -2.81259596e-01 1.61619329e+00 1.88458443e-01 3.04049160e-02 -9.28671300e-01 -1.13514066e+00 8.79676163e-01 6.77428782e-01 2.51677725e-02 -3.10721576e-01 6.40852988e-01 2.28563800e-01 -5.48024476e-02 8.57151091e-01 4.69759136e-01 -2.78271288e-01 3.18197757e-02 -9.10704374e-01 4.06986773e-01 6.84360623e-01 8.64269912e-01 9.44073737e-01 1.80010870e-01 3.11855435e-01 5.84903896e-01 3.48292589e-01 3.21470112e-01 -1.42403354e-03 -1.25746393e+00 -5.35552166e-02 5.23756206e-01 2.40312845e-01 -7.45750308e-01 -6.17763996e-01 -3.36752236e-01 -1.19840038e+00 6.96694672e-01 1.46493152e-01 5.57131708e-01 -1.17686892e+00 1.30134189e+00 4.85649049e-01 2.48207033e-01 -3.06300640e-01 8.85756135e-01 9.32101607e-01 6.95594609e-01 -4.33917463e-01 8.30600932e-02 8.62096012e-01 -3.05654377e-01 -1.83872506e-01 6.17888384e-03 4.17993516e-01 -4.51417744e-01 8.05827439e-01 3.59467298e-01 -1.33014369e+00 -5.23433149e-01 -1.05527318e+00 -4.18979496e-01 -3.88835430e-01 -5.82925677e-01 7.31586456e-01 1.98012605e-01 -1.26176524e+00 8.37399423e-01 -1.03425586e+00 -2.09028274e-01 8.42639923e-01 4.10347551e-01 -3.32535326e-01 -2.12760702e-01 -7.88427234e-01 1.13290620e+00 1.93767428e-01 -7.51624331e-02 -1.13687706e+00 -6.98457837e-01 -1.00050938e+00 -1.58908889e-01 -1.57884359e-02 -9.50314224e-01 1.12390709e+00 -2.14075148e-01 -1.15211546e+00 1.17562902e+00 -6.32609874e-02 -3.91234517e-01 4.86912966e-01 -2.35087257e-02 2.81503260e-01 -1.10398822e-01 -3.16085368e-02 8.72528434e-01 6.35805249e-01 -1.50226760e+00 -1.80788830e-01 -5.97292840e-01 7.57938391e-03 2.26432502e-01 3.88113976e-01 -7.80891240e-01 -2.13147625e-01 -1.47429794e-01 8.73116016e-01 -5.79473615e-01 -6.09294236e-01 5.67826450e-01 -1.71160907e-01 -3.16738814e-01 8.02315414e-01 -4.70510684e-02 3.88207346e-01 -1.95791268e+00 2.32506335e-01 3.57541293e-01 4.72855210e-01 -1.77679464e-01 1.39149666e-01 2.15355054e-01 2.13893309e-01 2.88018972e-01 -4.10648376e-01 -4.81332034e-01 2.77440578e-01 5.56834698e-01 -3.19791317e-01 8.62163365e-01 1.83481604e-01 9.05444682e-01 -7.07773447e-01 -3.87284935e-01 8.91484141e-01 9.86765385e-01 -5.90975881e-01 1.79684699e-01 -6.21878684e-01 4.55952883e-01 -4.35987115e-01 5.67723155e-01 6.44266844e-01 -4.14051414e-01 -7.76256025e-02 -1.09455369e-01 -2.29425222e-01 5.04082143e-01 -1.06323493e+00 2.23374176e+00 -6.92249835e-01 5.96837580e-01 8.90829042e-02 -1.09308767e+00 1.19413173e+00 2.37839222e-01 7.01084197e-01 -6.73991263e-01 1.00883424e-01 3.53082806e-01 -5.24440229e-01 2.16862913e-02 4.71158862e-01 -3.72664362e-01 -3.88742030e-01 6.29150271e-01 -1.26399964e-01 -9.04658675e-01 -5.69127500e-01 -1.61058933e-01 1.00276518e+00 2.95624942e-01 4.25522864e-01 -3.02158147e-01 -5.54992035e-02 1.02844447e-01 1.66548133e-01 9.65748429e-01 4.87320870e-02 8.16796303e-01 1.64015219e-01 -1.09553874e+00 -1.59954560e+00 -1.37431729e+00 -6.14209592e-01 3.71840626e-01 2.85775125e-01 -2.10581988e-01 -4.03371274e-01 -7.19513074e-02 4.11362827e-01 3.68579239e-01 -6.35337591e-01 2.57257700e-01 -7.32818067e-01 -2.20015734e-01 3.06264997e-01 5.49519718e-01 2.75010824e-01 -9.69041824e-01 -8.76155257e-01 3.37329507e-01 2.76449502e-01 -1.01554632e+00 2.25646347e-01 5.65042138e-01 -1.33980989e+00 -9.23954844e-01 -4.04101104e-01 -8.47437859e-01 6.56500280e-01 2.31581047e-01 1.48199511e+00 2.19059810e-01 -4.87267703e-01 3.99000823e-01 9.72500741e-02 -2.15438738e-01 -2.90852517e-01 4.63335179e-02 2.53980607e-01 -7.84518182e-01 3.49428862e-01 -1.16297758e+00 -5.13834119e-01 -3.21932323e-02 -6.70778275e-01 3.83064091e-01 3.44692767e-01 6.84791744e-01 1.37904382e+00 -6.07243739e-02 1.35179669e-01 -7.56580889e-01 2.46331155e-01 -4.40743923e-01 -6.71486855e-01 -3.39568347e-01 -8.49900022e-02 2.85909683e-01 6.02731645e-01 -1.19515054e-01 -4.46641147e-01 3.81257474e-01 -5.12988091e-01 -8.85087669e-01 -5.89583039e-01 3.17231625e-01 1.56703636e-01 -1.41493991e-01 7.55248249e-01 3.68310332e-01 -3.45910490e-02 -7.47813642e-01 4.33447063e-01 3.78280193e-01 4.31422889e-01 -8.64776552e-01 5.66100836e-01 8.21057379e-01 4.52832639e-01 -9.10512328e-01 -7.78499126e-01 -1.60056397e-01 -1.12333500e+00 -1.77738905e-01 7.86395192e-01 -8.46376777e-01 -5.99888682e-01 2.02609897e-01 -1.40935707e+00 -4.30965871e-01 -6.25919640e-01 1.77886456e-01 -1.03933167e+00 3.41865309e-02 -5.78300238e-01 -9.46321487e-01 -1.50504664e-01 -1.19270825e+00 1.42006516e+00 -3.46154749e-01 -3.37973684e-01 -9.44393337e-01 1.31958917e-01 -3.73573527e-02 3.62085789e-01 8.43560576e-01 1.24947035e+00 -2.75050700e-02 -1.12748873e+00 -1.37609810e-01 -2.46075079e-01 -1.45110086e-01 -8.32775831e-02 -3.42835546e-01 -9.98032212e-01 -9.47789848e-02 -1.01537980e-01 -5.79312146e-01 9.20277715e-01 6.07175946e-01 1.67557490e+00 -6.48431182e-02 -3.80196720e-01 8.14588130e-01 1.58588815e+00 -2.90718377e-01 4.51453269e-01 4.59788628e-02 6.88783467e-01 2.26248562e-01 1.10720880e-01 5.10115802e-01 1.94278091e-01 5.21550000e-01 9.50491309e-01 -2.40940899e-01 -1.25994608e-01 -4.40336436e-01 -3.70245963e-01 9.18207347e-01 -6.84829503e-02 7.46321678e-02 -1.23043382e+00 4.68117058e-01 -1.71794856e+00 -6.49574041e-01 1.19875759e-01 2.16827083e+00 6.53143048e-01 3.50422561e-01 -1.62067637e-01 2.26126969e-01 3.12150747e-01 3.71744066e-01 -9.46601570e-01 -4.34697568e-01 3.65154967e-02 6.32375777e-01 4.31807637e-01 6.33305430e-01 -1.07235837e+00 7.63318658e-01 7.27210474e+00 5.95323026e-01 -9.70474720e-01 1.28078461e-01 5.82128108e-01 -3.33807558e-01 -6.14902794e-01 -3.74614269e-01 -5.79331100e-01 -4.82905470e-02 6.87918007e-01 2.80038267e-01 3.89835149e-01 9.69721496e-01 -3.84477973e-02 4.20326181e-02 -1.50053799e+00 1.33126688e+00 -1.10329665e-01 -2.15021563e+00 3.13409358e-01 3.30500573e-01 6.93520784e-01 4.65761214e-01 1.93236191e-02 6.56085238e-02 4.29682374e-01 -1.73511398e+00 8.62812817e-01 5.69752395e-01 1.24805164e+00 -8.29291403e-01 3.74545157e-01 6.90689266e-01 -1.23434687e+00 4.23577994e-01 -6.81562543e-01 -3.71205091e-01 2.08566114e-01 7.03610659e-01 -7.12794423e-01 1.19483918e-01 7.88489223e-01 7.41785765e-01 1.23661183e-01 8.45378518e-01 2.34659016e-01 2.91249342e-02 -7.40418434e-01 -3.60659398e-02 4.69880223e-01 5.61335497e-02 4.11641896e-01 8.93457830e-01 3.20439279e-01 2.13991150e-01 4.71957356e-01 1.22772002e+00 -2.33784810e-01 -2.49857679e-01 -1.29052639e+00 2.50636816e-01 7.09822714e-01 6.51475251e-01 -8.29192519e-01 -1.29766583e-01 -2.41130516e-01 6.22709930e-01 7.43525743e-01 3.18209454e-02 -4.32181060e-01 6.36027008e-02 8.23166966e-01 5.46091497e-01 4.76653129e-01 -8.28417003e-01 -6.59149528e-01 -7.19323039e-01 -1.66807428e-01 -2.80280679e-01 -2.09900588e-01 -7.92788446e-01 -1.26881349e+00 4.91658509e-01 1.55940913e-02 -1.05686653e+00 -2.03299433e-01 -8.42368603e-01 -1.46167457e-01 8.06331038e-01 -1.45799196e+00 -8.32320690e-01 -2.70674169e-01 3.56340557e-01 4.31243300e-01 1.73035458e-01 1.25323796e+00 -3.06647643e-02 7.54606649e-02 -1.25323430e-01 2.18001127e-01 -4.05686945e-02 -1.33569136e-01 -1.21735859e+00 8.01151514e-01 4.24423218e-02 3.77935320e-01 3.00617874e-01 4.13970619e-01 -4.44149166e-01 -1.75677240e+00 -1.09327030e+00 3.86145175e-01 -6.65971994e-01 7.81386867e-02 -7.21234500e-01 -1.01125515e+00 5.47378659e-01 -4.05439764e-01 3.92694056e-01 5.02372146e-01 1.50120035e-01 -5.01311779e-01 3.71999204e-01 -1.24803185e+00 3.04371864e-01 1.47077560e+00 -8.59636903e-01 -5.79103649e-01 3.60352069e-01 6.77539825e-01 -8.97723496e-01 -1.14607263e+00 4.58625704e-01 5.90241790e-01 -1.06299877e+00 1.38506651e+00 -4.64669496e-01 3.97943825e-01 -2.84075946e-01 -6.29378021e-01 -1.07219112e+00 -5.10309756e-01 -1.00010917e-01 -5.08389235e-01 1.44057661e-01 3.65938358e-02 -2.52187043e-01 1.17324245e+00 3.20956975e-01 -4.74690408e-01 -1.10579717e+00 -1.47581303e+00 -5.78256845e-01 5.28801084e-01 -8.70079279e-01 7.29330063e-01 7.55486190e-01 -3.11460465e-01 9.95350555e-02 -1.52757213e-01 1.05701350e-01 9.12307143e-01 4.44388241e-01 6.77902818e-01 -1.58479095e+00 7.49539509e-02 -5.37492216e-01 -6.24830008e-01 -1.51434672e+00 2.99224377e-01 -1.08686054e+00 2.27327600e-01 -1.92647743e+00 -4.12956998e-02 -8.52271080e-01 -8.61971527e-02 2.50558943e-01 7.74232686e-01 5.64651966e-01 -2.57088631e-01 4.01140511e-01 -5.86202323e-01 6.73573613e-01 1.36082757e+00 -1.36672124e-01 -2.62783542e-02 -4.89388019e-01 -1.73803777e-01 9.01004314e-01 6.72237635e-01 -3.29784840e-01 -1.86085761e-01 -7.30956912e-01 1.73992813e-01 1.65219635e-01 5.69831967e-01 -9.61029053e-01 1.49404019e-01 -8.17502290e-02 9.02579904e-01 -1.26953769e+00 9.05987978e-01 -8.49434912e-01 3.14475983e-01 2.24551469e-01 -9.70975533e-02 -2.84479797e-01 3.46355736e-01 5.20995617e-01 1.10238753e-01 1.75082743e-01 8.56539667e-01 -8.18975866e-01 -5.84954798e-01 7.70912886e-01 -3.93197507e-01 -4.48957793e-02 6.97261333e-01 -6.04207277e-01 9.90301892e-02 -1.00972570e-01 -7.48420954e-01 -1.75334871e-01 9.41636860e-01 3.83506082e-02 1.15465057e+00 -1.59388220e+00 -6.02990568e-01 5.24379194e-01 -1.20217867e-01 9.12046611e-01 1.55525103e-01 1.16089918e-01 -9.18607116e-01 6.44633651e-01 -4.10900533e-01 -1.24313712e+00 -5.80275834e-01 3.24849188e-01 6.31726861e-01 -3.28693874e-02 -1.20428360e+00 6.87262774e-01 1.67274341e-01 -7.74010539e-01 4.39525813e-01 -6.12380207e-01 3.04796845e-01 -4.36307758e-01 2.78353900e-01 -1.94609945e-03 9.86264348e-02 -6.35389686e-01 -3.51774395e-01 6.85827315e-01 1.95304140e-01 -1.39358258e-02 1.63184500e+00 2.64369667e-01 -5.58153503e-02 8.92477751e-01 1.24667776e+00 -6.87871993e-01 -1.42185879e+00 -3.01372200e-01 -3.96627575e-01 -5.72693825e-01 4.08003747e-01 -2.68125147e-01 -7.71250069e-01 1.23563051e+00 4.83426958e-01 2.13942260e-01 4.82502729e-01 4.05025959e-01 6.42273068e-01 5.56271255e-01 9.02765691e-01 -6.59520328e-01 6.36166483e-02 8.13006401e-01 1.04292989e+00 -1.19695580e+00 2.26185992e-01 -1.90658197e-01 4.00744408e-01 1.11608028e+00 4.73022997e-01 -5.77107370e-01 9.27266240e-01 4.77091968e-01 -3.14775467e-01 -5.90301692e-01 -7.31715977e-01 1.22357696e-01 1.76365629e-01 9.01503921e-01 4.03383136e-01 1.20103046e-01 5.96353531e-01 6.86462894e-02 -3.32703859e-01 -2.11906403e-01 2.97181517e-01 9.86189604e-01 -8.78469646e-01 -7.76235938e-01 -4.76981252e-01 6.82605326e-01 1.49137676e-01 1.50835151e-02 -1.61658511e-01 8.88397753e-01 2.93247756e-02 7.32082874e-02 7.74673402e-01 -2.57580757e-01 1.99188098e-01 4.50089984e-02 1.05993903e+00 -8.60368192e-01 -3.99252921e-02 -1.02932543e-01 -1.35404915e-01 -8.33546519e-01 -3.41543645e-01 -5.76207995e-01 -1.40340507e+00 -5.08955598e-01 1.12978004e-01 -6.82306439e-02 8.30025315e-01 8.21262956e-01 3.37796599e-01 3.85884553e-01 3.08820188e-01 -1.66535568e+00 -2.18598306e-01 -6.08371735e-01 -7.45604813e-01 1.67081848e-01 5.16418695e-01 -1.05772424e+00 -1.53744757e-01 -4.93511796e-01]
[8.419380187988281, -3.617124557495117]
22fe418d-b6e2-4ad1-9638-2c0b29d1c8c6
conditional-local-filters-with-explainers-for
2101.01000
null
https://arxiv.org/abs/2101.01000v3
https://arxiv.org/pdf/2101.01000v3.pdf
Conditional Local Convolution for Spatio-temporal Meteorological Forecasting
Spatio-temporal forecasting is challenging attributing to the high nonlinearity in temporal dynamics as well as complex location-characterized patterns in spatial domains, especially in fields like weather forecasting. Graph convolutions are usually used for modeling the spatial dependency in meteorology to handle the irregular distribution of sensors' spatial location. In this work, a novel graph-based convolution for imitating the meteorological flows is proposed to capture the local spatial patterns. Based on the assumption of smoothness of location-characterized patterns, we propose conditional local convolution whose shared kernel on nodes' local space is approximated by feedforward networks, with local representations of coordinate obtained by horizon maps into cylindrical-tangent space as its input. The established united standard of local coordinate system preserves the orientation on geography. We further propose the distance and orientation scaling terms to reduce the impacts of irregular spatial distribution. The convolution is embedded in a Recurrent Neural Network architecture to model the temporal dynamics, leading to the Conditional Local Convolution Recurrent Network (CLCRN). Our model is evaluated on real-world weather benchmark datasets, achieving state-of-the-art performance with obvious improvements. We conduct further analysis on local pattern visualization, model's framework choice, advantages of horizon maps and etc.
['Ling Li', 'Yongjie Xu', 'Stan. Z. Li', 'Lirong Wu', 'Zhangyang Gao', 'Haitao Lin']
2021-01-04
null
null
null
null
['spatio-temporal-forecasting']
['time-series']
[-4.59215850e-01 -5.62853277e-01 3.08308184e-01 -4.44179475e-01 5.51308930e-01 -4.84240234e-01 7.24265695e-01 -3.72156613e-02 -1.48337319e-01 2.20987469e-01 5.05738080e-01 -7.76231945e-01 -4.33582425e-01 -1.08943570e+00 -4.89006817e-01 -6.63373470e-01 -7.70363927e-01 -3.65311354e-01 1.56971291e-01 -4.48185265e-01 8.23825896e-02 7.90128410e-01 -1.16707158e+00 1.99978575e-02 1.00300312e+00 8.33322883e-01 2.27526903e-01 8.01000655e-01 -2.36132681e-01 1.01648557e+00 -4.26508158e-01 1.47776157e-01 1.79241583e-01 -2.57842034e-01 -3.70030820e-01 -2.08978429e-01 -6.76263198e-02 -1.83035776e-01 -6.37503088e-01 8.76218736e-01 2.60192841e-01 5.89360118e-01 6.91671431e-01 -1.02134776e+00 -1.18296134e+00 3.26995254e-01 -4.01990861e-01 7.97626615e-01 -2.32798204e-01 1.23986766e-01 6.87664092e-01 -7.33444273e-01 1.85996696e-01 1.12416530e+00 8.18844795e-01 -2.23808289e-01 -8.86908293e-01 -3.37103188e-01 8.24726820e-01 1.31932214e-01 -1.68254685e+00 1.90166354e-01 9.36158121e-01 -6.15817070e-01 1.02409697e+00 3.22265118e-01 5.55229127e-01 6.64331138e-01 5.11674523e-01 1.27147049e-01 8.44704926e-01 4.44551371e-02 -6.89765364e-02 -9.15613919e-02 1.19264834e-01 4.36649591e-01 -3.08597624e-01 3.66169989e-01 -1.55172691e-01 -3.57532725e-02 1.06682038e+00 6.67783141e-01 -5.03304064e-01 1.59527823e-01 -1.16271424e+00 6.47441268e-01 1.32327032e+00 6.81452513e-01 -5.91575980e-01 9.57256928e-02 1.38585985e-01 2.65348345e-01 1.08912313e+00 8.71036872e-02 -3.72023851e-01 4.39709686e-02 -8.10332656e-01 3.08279634e-01 5.58228791e-01 8.11428487e-01 6.98702216e-01 4.92018104e-01 -2.26824030e-01 5.67201793e-01 2.42868111e-01 7.02633739e-01 6.08683705e-01 -2.29731426e-01 5.64359963e-01 9.16150689e-01 1.67061791e-01 -1.86968374e+00 -9.41301465e-01 -9.26621437e-01 -1.70255208e+00 -1.32222742e-01 1.82122439e-01 -3.25287879e-01 -8.58274877e-01 1.52852798e+00 3.33051026e-01 8.39330852e-01 -5.44994883e-02 1.08436441e+00 5.89229524e-01 1.43454349e+00 1.38073742e-01 2.97375601e-02 1.10428917e+00 -6.99817836e-01 -8.39742720e-01 3.56243670e-01 6.14742160e-01 -4.45284903e-01 1.12445664e+00 -3.16287935e-01 -4.52872217e-01 -7.21928060e-01 -6.61016524e-01 1.45200416e-01 -1.11039007e+00 3.85003954e-01 3.36677253e-01 1.16918065e-01 -1.28265131e+00 7.78887987e-01 -7.01658607e-01 -2.33034328e-01 -2.90330350e-01 -2.54041344e-01 7.01210424e-02 4.42566186e-01 -1.63989437e+00 6.65390015e-01 1.88753843e-01 1.03141820e+00 -3.51304710e-01 -1.09587932e+00 -8.09962034e-01 3.65470648e-01 -2.34676301e-01 -4.67128694e-01 6.38072968e-01 -4.90633905e-01 -1.36519098e+00 2.68641394e-02 -1.55453265e-01 -3.80002558e-01 2.85296768e-01 2.25562211e-02 -1.04654419e+00 -3.25145185e-01 -1.46309033e-01 -4.58713435e-02 7.10070729e-01 -6.37227774e-01 -4.94078457e-01 -1.15425810e-01 -5.00633009e-02 2.74069279e-01 -3.84125412e-01 -3.47237252e-02 -1.08747467e-01 -1.13913214e+00 1.46938726e-01 -5.99434733e-01 -5.90038002e-01 -2.93967187e-01 -3.03620994e-01 -3.05839479e-01 9.04622734e-01 -9.48551595e-01 1.94800508e+00 -2.25294352e+00 -6.13919720e-02 4.59663451e-01 2.59796113e-01 1.14460744e-01 -5.52067049e-02 5.87727785e-01 -8.56809169e-02 2.23634660e-01 -4.49181676e-01 -1.40399829e-01 -4.68317010e-02 1.96359500e-01 -9.48624253e-01 6.08698010e-01 3.47631663e-01 9.62969482e-01 -7.69923449e-01 2.06521779e-01 3.65897566e-01 7.83150852e-01 -3.26449424e-02 4.06782717e-01 -1.03068709e-01 5.60859084e-01 -5.45207441e-01 1.89274713e-01 1.06044960e+00 -4.04121816e-01 -2.17062265e-01 6.53478724e-04 -8.57786894e-01 3.43390524e-01 -1.19411063e+00 1.22717142e+00 -4.95669723e-01 3.36907864e-01 -1.95327386e-01 -7.86094427e-01 1.34768188e+00 2.59283453e-01 6.60738871e-02 -7.75463939e-01 -2.94260561e-01 3.73089202e-02 -1.27024502e-01 -4.04816747e-01 5.65750599e-01 2.65653461e-01 2.24592566e-01 3.14611316e-01 -4.91314709e-01 3.27071577e-01 -3.52362335e-01 7.81839564e-02 7.38081872e-01 1.49788484e-01 -1.16971835e-01 -6.58651114e-01 3.91346484e-01 -6.90591037e-02 5.05677521e-01 3.35298061e-01 1.28147632e-01 5.82013011e-01 4.66489762e-01 -1.33462536e+00 -7.74525404e-01 -6.58100545e-01 -9.00230855e-02 1.01660705e+00 5.04899509e-02 -3.64594012e-01 -2.61314124e-01 -3.10545623e-01 5.70365265e-02 4.87331003e-01 -9.67630625e-01 1.09045416e-01 -6.65482402e-01 -1.12159932e+00 5.41207492e-01 5.76550245e-01 6.53225660e-01 -1.10911894e+00 -5.67773640e-01 3.04709017e-01 2.35263646e-01 -7.96630561e-01 -6.07569814e-01 -5.09358011e-02 -7.37866342e-01 -7.25578606e-01 -1.03327191e+00 -4.26574588e-01 6.99160933e-01 4.87244964e-01 8.56893957e-01 -1.09553836e-01 1.32031128e-01 -2.28447989e-02 -3.67158294e-01 5.42618986e-03 4.13969874e-01 2.87523150e-01 -3.81717794e-02 5.59636116e-01 4.26942632e-02 -9.95448649e-01 -8.88015807e-01 5.11349022e-01 -9.27111089e-01 5.18672206e-02 2.55419374e-01 6.44629598e-01 2.77658671e-01 1.84822530e-01 3.52293283e-01 -6.45207703e-01 8.94629478e-01 -1.07512379e+00 -8.09055150e-01 3.48853379e-01 -4.91884619e-01 -6.62123635e-02 1.07432115e+00 -4.32429522e-01 -1.02935994e+00 -3.45051527e-01 3.84390563e-01 -6.87360048e-01 -4.06503752e-02 1.14339280e+00 3.83248121e-01 -1.82400066e-02 6.24026418e-01 5.13684690e-01 -3.72013837e-01 -6.59232199e-01 4.33539361e-01 2.32271224e-01 5.43019116e-01 -2.52743512e-01 9.98456419e-01 2.68474817e-01 4.17200066e-02 -9.54964817e-01 -3.61842245e-01 -4.23184037e-01 -5.85722983e-01 -1.52586862e-01 7.07720518e-01 -8.62780750e-01 -7.03363299e-01 6.51270092e-01 -1.38270330e+00 -5.62234938e-01 4.43201885e-02 5.19687951e-01 2.22202331e-01 -6.92181364e-02 -5.49797237e-01 -8.99000287e-01 -2.58453131e-01 -6.17197394e-01 6.88281178e-01 3.55091691e-01 2.50626117e-01 -1.64507329e+00 4.80915517e-01 -9.88694727e-01 1.22020328e+00 4.97146368e-01 7.69453347e-01 -6.87842593e-02 -5.96022427e-01 1.23525446e-03 -5.11294425e-01 4.31558192e-02 2.89850593e-01 2.34311804e-01 -8.22380006e-01 -1.04977190e-01 -1.87535603e-02 5.50982893e-01 8.31001997e-01 6.05333388e-01 1.10610712e+00 -6.54121876e-01 -1.23341918e-01 1.17674804e+00 1.32119095e+00 1.78401008e-01 4.21347827e-01 2.49328241e-02 1.11156714e+00 8.98494303e-01 5.98231182e-02 5.88090479e-01 6.50940657e-01 4.15518045e-01 2.35546887e-01 -4.69428152e-01 2.71428019e-01 -4.14144307e-01 7.73665980e-02 1.11846018e+00 -4.69328195e-01 -2.90679008e-01 -1.09493864e+00 7.20978260e-01 -2.16775012e+00 -8.63642693e-01 -3.23275089e-01 1.91358614e+00 1.09810576e-01 -3.39159042e-01 4.25434485e-02 -3.37960422e-01 8.40534151e-01 7.77512491e-01 -4.24988091e-01 -4.64999110e-01 -3.47008377e-01 -1.75231159e-01 8.26743722e-01 6.30532205e-01 -1.25448942e+00 8.38922143e-01 5.55402088e+00 6.34108603e-01 -1.69677985e+00 -6.84966333e-03 8.27715337e-01 2.53448635e-01 -4.34823602e-01 -2.21778944e-01 -5.52440226e-01 6.55991077e-01 1.10121810e+00 -5.64300902e-02 5.99302053e-01 5.75848520e-01 6.96212709e-01 5.85624814e-01 -1.59878522e-01 7.29803383e-01 -3.86278600e-01 -1.45214701e+00 1.99559152e-01 -5.80826700e-02 6.79202855e-01 2.81125844e-01 3.65557969e-01 1.44808412e-01 4.07580823e-01 -1.27198386e+00 4.73364890e-01 9.55047548e-01 7.26319253e-01 -6.81898355e-01 4.95563716e-01 3.63541394e-01 -2.01735997e+00 3.86183932e-02 -6.14856422e-01 -4.64412779e-01 1.30000249e-01 6.54229164e-01 -5.28304636e-01 9.08917427e-01 9.71182525e-01 1.40265596e+00 -3.69738728e-01 8.46995950e-01 -1.18871562e-01 8.13654721e-01 -5.87473989e-01 -8.20796713e-02 7.68832803e-01 -7.18374908e-01 4.00438219e-01 1.40632164e+00 5.17383218e-01 2.61175334e-01 1.31600484e-01 1.16324437e+00 3.86750758e-01 1.31361410e-01 -9.90202963e-01 1.68659806e-01 4.15889859e-01 1.19016051e+00 -5.20355523e-01 -1.43616349e-01 -2.34410584e-01 5.83417773e-01 2.69634128e-01 1.14159358e+00 -7.21337497e-01 -6.41646862e-01 6.70746863e-01 6.52513131e-02 2.10257947e-01 -6.56591952e-01 -2.01850131e-01 -1.13196421e+00 2.73202538e-01 -1.63633034e-01 3.56194884e-01 -6.59789205e-01 -1.33260775e+00 1.07210863e+00 -8.03672522e-02 -1.43089473e+00 -1.77637219e-01 -5.12596250e-01 -1.37662566e+00 1.53277791e+00 -1.94928563e+00 -1.40777314e+00 -3.94935012e-01 8.97305846e-01 -2.61928216e-02 8.17927867e-02 7.32127905e-01 3.29458088e-01 -6.67957246e-01 2.72343487e-01 1.96302161e-01 1.68587998e-01 2.35243365e-01 -1.25039351e+00 9.98140872e-01 1.13777864e+00 -1.25501752e-01 7.99887836e-01 2.67560929e-01 -6.27668202e-01 -1.09389663e+00 -1.66578019e+00 1.23448002e+00 -4.17999066e-02 9.37912762e-01 -3.45850140e-01 -1.32990432e+00 7.49159753e-01 6.47299513e-02 5.46670735e-01 1.59275770e-01 7.08326623e-02 -4.65297520e-01 -2.96823919e-01 -7.20255494e-01 5.89061499e-01 9.63947892e-01 -6.58268750e-01 -2.07421541e-01 4.01634336e-01 1.08657658e+00 -4.89450991e-01 -7.23900795e-01 3.85806024e-01 1.80955291e-01 -7.03382730e-01 7.26379156e-01 -5.38656175e-01 1.76766112e-01 -7.32850611e-01 1.88724138e-02 -1.57489359e+00 -8.64321232e-01 -8.15421581e-01 1.32978559e-01 9.39870477e-01 4.81113493e-01 -1.12787271e+00 1.85473457e-01 3.73397350e-01 -2.79633224e-01 -8.16374123e-01 -9.39324260e-01 -5.79108000e-01 -5.36672249e-02 -3.18004966e-01 1.13928115e+00 1.28470838e+00 -1.60223749e-02 -7.81065300e-02 -6.88888013e-01 7.57133961e-01 7.62577541e-03 2.94226378e-01 4.65377510e-01 -9.82094288e-01 -4.58769873e-02 -5.82361341e-01 -3.96465749e-01 -1.21375358e+00 -1.05389029e-01 -7.36045241e-01 -4.13380325e-01 -1.20755756e+00 -6.37529492e-01 -6.63233042e-01 -7.66832829e-01 3.64683598e-01 -1.44373596e-01 -1.71160206e-01 -1.82317253e-02 3.48753750e-01 -1.69582143e-01 9.32749510e-01 1.20618320e+00 -2.21911557e-02 -6.30662382e-01 7.04641044e-02 -1.56758517e-01 3.48780394e-01 8.43411326e-01 -4.70161401e-02 -5.67916274e-01 -6.64164066e-01 4.61305618e-01 -1.16411179e-01 4.87229824e-01 -8.88010919e-01 5.40931463e-01 -4.11742598e-01 1.77179024e-01 -6.80458069e-01 -4.18516546e-02 -7.82818198e-01 3.39835525e-01 2.15445727e-01 -2.73161292e-01 8.96703005e-01 3.88345242e-01 5.99699318e-01 -5.38493276e-01 6.35809124e-01 1.74473032e-01 1.55268133e-01 -6.44244432e-01 7.04435706e-01 -3.44039083e-01 -2.62576133e-01 6.22503042e-01 1.77617401e-01 -4.69790429e-01 -4.25406784e-01 -5.78319252e-01 4.61334467e-01 4.02567498e-02 3.74047518e-01 6.63395762e-01 -1.48216343e+00 -7.80384541e-01 5.19627035e-01 -9.90591645e-02 1.32462487e-01 6.46174073e-01 9.01251614e-01 -8.35564137e-01 5.28123379e-01 4.71256860e-02 -5.06845832e-01 -4.50485528e-01 5.33798158e-01 6.83099866e-01 -3.87647271e-01 -1.05897880e+00 5.29222608e-01 4.95630205e-01 -7.83576608e-01 3.15910280e-02 -1.15986812e+00 -5.28970957e-01 -1.53638214e-01 5.23432374e-01 4.15380001e-01 -8.31711814e-02 -5.73601484e-01 -2.74215132e-01 8.24531734e-01 4.15725261e-01 7.45590106e-02 1.43231571e+00 -2.22209558e-01 -2.55027860e-01 7.42303252e-01 1.09024763e+00 -2.41037041e-01 -1.29119313e+00 -4.14254606e-01 -1.81889579e-01 -2.75752276e-01 5.10064512e-03 -5.34872711e-01 -1.32795191e+00 1.14787126e+00 4.73070294e-01 6.96882606e-01 1.04488516e+00 -6.50970817e-01 6.37794018e-01 1.09882377e-01 -7.89532885e-02 -5.54581940e-01 -5.99055469e-01 1.01779389e+00 1.16356897e+00 -7.64067471e-01 -3.99423748e-01 -1.40589595e-01 -6.16329014e-01 1.19754255e+00 2.27757320e-01 -6.08619273e-01 1.45327139e+00 -1.61777735e-02 2.59045184e-01 -3.58793259e-01 -5.60601175e-01 2.51063388e-02 6.12266243e-01 2.48125955e-01 3.74392211e-01 5.47025383e-01 1.34605557e-01 4.37265366e-01 -1.85135916e-01 -2.81303287e-01 2.78466307e-02 5.66156983e-01 -7.05189481e-02 -3.52813661e-01 -2.46540040e-01 1.85656011e-01 -3.96762490e-02 -3.53726357e-01 1.57137319e-01 5.37883222e-01 9.34795439e-02 7.62201667e-01 6.54842913e-01 -5.74324191e-01 3.81672591e-01 -1.60816252e-01 -5.50928593e-01 -1.28945231e-01 -6.52305126e-01 -2.03346778e-02 -5.10759532e-01 -5.70204914e-01 -2.93118626e-01 -2.55689859e-01 -1.01638591e+00 -5.38753629e-01 2.19802395e-01 2.75386572e-01 5.56324720e-01 6.89366281e-01 8.60680759e-01 7.48385370e-01 1.10345209e+00 -9.34977472e-01 -1.77268863e-01 -1.17104399e+00 -9.72921550e-01 2.63425291e-01 7.65247464e-01 -4.62398410e-01 -5.37550867e-01 -1.68353885e-01]
[6.651744842529297, 2.76419997215271]
5c7b38ef-4594-4d59-9728-f81fe1e27aa0
elsed-enhanced-line-segment-drawing
2108.03144
null
https://arxiv.org/abs/2108.03144v1
https://arxiv.org/pdf/2108.03144v1.pdf
ELSED: Enhanced Line SEgment Drawing
Detecting local features, such as corners, segments or blobs, is the first step in the pipeline of many Computer Vision applications. Its speed is crucial for real time applications. In this paper we present ELSED, the fastest line segment detector in the literature. The key for its efficiency is a local segment growing algorithm that connects gradient aligned pixels in presence of small discontinuities. The proposed algorithm not only runs in devices with very low end hardware, but may also be parametrized to foster the detection of short or longer segments, depending on the task at hand. We also introduce new metrics to evaluate the accuracy and repeatability of segment detectors. In our experiments with different public benchmarks we prove that our method is the most efficient in the literature and quantify the accuracy traded for such gain.
['Luis Baumela', 'José M. Buenaposada', 'Iago Suárez']
2021-08-06
null
null
null
null
['line-segment-detection', 'line-detection']
['computer-vision', 'computer-vision']
[ 3.20666373e-01 -2.69288510e-01 -3.64495307e-01 -1.52251154e-01 -4.02212143e-01 -7.34950483e-01 5.67988515e-01 6.89792216e-01 -4.42515284e-01 1.76807836e-01 -4.26962107e-01 -4.45762396e-01 2.76724756e-01 -6.60492361e-01 -6.50324643e-01 -4.16199714e-01 -1.21461980e-01 1.33906171e-01 1.22242820e+00 -9.55238193e-03 6.86329246e-01 1.04258549e+00 -1.29611731e+00 8.91315285e-03 5.24192691e-01 1.14025784e+00 -5.38686141e-02 8.88984025e-01 2.94959312e-03 3.84277165e-01 -3.74939770e-01 -2.68552303e-01 5.48854053e-01 -2.58547634e-01 -5.64192772e-01 7.30631799e-02 5.59589028e-01 -3.55338365e-01 1.86623052e-01 7.21336305e-01 4.85326976e-01 -5.48214734e-01 6.49588585e-01 -1.06883562e+00 2.99026549e-01 5.91810167e-01 -1.05306232e+00 4.64087188e-01 4.39058572e-01 1.27745390e-01 7.25270152e-01 -6.72721505e-01 9.22276318e-01 6.65309548e-01 1.00345004e+00 1.17338032e-01 -1.25442135e+00 -2.36490682e-01 -1.02351584e-01 -5.21988124e-02 -1.34236538e+00 -3.52705657e-01 6.74403191e-01 -5.21309912e-01 8.15954030e-01 1.81704789e-01 6.41564310e-01 5.91449738e-01 3.13192636e-01 7.04963326e-01 8.58940542e-01 -6.46298528e-01 3.53581607e-01 -4.45754193e-02 4.80160475e-01 9.11751986e-01 6.04725718e-01 -1.06455736e-01 -2.76593655e-01 -2.71102544e-02 1.01313961e+00 -2.37429559e-01 -1.38690293e-01 -7.02206910e-01 -1.13892555e+00 6.82407916e-01 5.09064674e-01 3.13594431e-01 -2.15020880e-01 1.28378138e-01 5.56424081e-01 1.36616677e-01 7.92480484e-02 1.58886313e-01 -1.40312314e-01 -1.50408313e-01 -1.31380641e+00 6.00342522e-04 7.47210681e-01 8.74886453e-01 6.21892333e-01 -4.30788606e-01 -1.48622701e-02 4.57923263e-01 1.20531663e-01 1.60264149e-01 1.69188276e-01 -7.81012058e-01 2.69089133e-01 9.11586881e-01 1.31988034e-01 -7.78692722e-01 -8.41965377e-01 -3.87211889e-01 -5.16677260e-01 6.17120504e-01 7.56365061e-01 -1.88590482e-01 -8.37630570e-01 1.06522596e+00 5.85937500e-01 2.53955483e-01 -3.47753793e-01 6.28827870e-01 5.37132502e-01 4.63010162e-01 -3.07509899e-01 -7.99051672e-02 1.30289495e+00 -9.66400146e-01 -1.06379744e-02 -2.99180061e-01 6.63971364e-01 -1.13449824e+00 5.33000946e-01 3.93141448e-01 -8.95670772e-01 -4.84589845e-01 -1.35634768e+00 -2.87313819e-01 -8.39144439e-02 1.85781211e-01 2.52974123e-01 7.17913330e-01 -1.02023625e+00 5.58617771e-01 -1.00463307e+00 -7.18799770e-01 1.83415741e-01 4.45764005e-01 -1.20235518e-01 2.32293680e-01 -1.62991390e-01 6.44918382e-01 2.21648872e-01 1.21042892e-01 -1.52568638e-01 -3.27154636e-01 -5.45649707e-01 -9.27303433e-02 2.05852509e-01 -4.94512618e-01 1.01839697e+00 -8.42453897e-01 -1.31078243e+00 1.21552217e+00 -1.54858053e-01 -1.06535637e+00 1.02318418e+00 -1.70353889e-01 9.48139727e-02 2.71329343e-01 -3.25962633e-01 6.68905556e-01 8.34950566e-01 -9.28032637e-01 -1.07616508e+00 -2.31925383e-01 1.06223710e-01 -7.22568259e-02 2.22066551e-01 1.91860810e-01 -5.45357704e-01 -2.00530589e-01 1.61663353e-01 -9.52032089e-01 -2.31080338e-01 3.76372457e-01 -5.43534219e-01 -3.31404239e-01 1.05639863e+00 -2.58676022e-01 9.64963675e-01 -2.18409181e+00 -3.85684639e-01 4.85342801e-01 1.26383051e-01 4.65508372e-01 2.16104805e-01 4.24534947e-01 4.38290358e-01 -9.63447541e-02 -1.36484236e-01 -3.03944379e-01 -3.78905863e-01 -1.73370942e-01 -3.42028588e-02 8.32253635e-01 2.49659643e-01 5.88639736e-01 -5.02617300e-01 -7.49304712e-01 5.63548803e-01 4.96149629e-01 -1.27167493e-01 -2.95581311e-01 1.54560462e-01 6.43113405e-02 -3.52705389e-01 5.04739165e-01 8.06703448e-01 -1.30707055e-01 -2.91336507e-01 -2.29680926e-01 -8.16736042e-01 -3.56395133e-02 -1.44274807e+00 1.52724457e+00 -3.01149487e-02 1.05732179e+00 2.36853212e-03 -7.05396712e-01 1.16889226e+00 -1.23640023e-01 4.48696196e-01 -4.39143181e-01 3.68050247e-01 4.05569822e-01 -1.36714041e-01 -1.98827803e-01 5.44220746e-01 5.86277962e-01 1.60873517e-01 2.31937051e-01 -3.77976865e-01 -2.56715459e-03 4.07347381e-01 -3.71002243e-03 1.20177042e+00 3.52100991e-02 6.61151648e-01 -4.26135033e-01 4.70969319e-01 3.55732977e-01 2.87678212e-01 8.24650049e-01 -4.88921344e-01 6.58468366e-01 6.49983585e-01 -4.01978523e-01 -1.10099399e+00 -1.03485131e+00 -4.27425414e-01 6.56240344e-01 5.38416564e-01 -2.73209155e-01 -8.70673895e-01 -3.24099362e-01 -2.78428286e-01 1.66877851e-01 -3.62448663e-01 2.00373515e-01 -7.35359311e-01 -2.41019115e-01 5.90085030e-01 5.31423271e-01 4.60031450e-01 -8.70316982e-01 -1.69172347e+00 1.83332339e-01 4.35171127e-01 -1.22243452e+00 -4.81090754e-01 2.06086025e-01 -1.10653400e+00 -1.14216554e+00 -7.90496051e-01 -1.09588420e+00 6.57410026e-01 4.82473433e-01 1.01284146e+00 1.26048252e-01 -5.37677586e-01 2.87649985e-02 -4.38214242e-01 -2.33567506e-01 -2.76427358e-01 2.77300119e-01 -5.30711174e-01 -1.05732441e-01 1.63704932e-01 -1.01778723e-01 -1.02517748e+00 4.98976201e-01 -6.59712315e-01 1.68599755e-01 5.39598465e-01 3.95537943e-01 6.56317770e-01 -3.40281785e-01 -1.04558632e-01 -8.33625376e-01 3.24549526e-01 -8.31238925e-02 -1.09722137e+00 1.75097302e-01 -3.10823143e-01 7.96612054e-02 3.49000424e-01 -2.06327006e-01 -5.80384731e-01 7.32879341e-01 -9.64849591e-02 1.10299625e-02 -2.37389609e-01 2.45663837e-01 3.70378226e-01 -5.53713262e-01 8.86929750e-01 -6.12959713e-02 -2.84260899e-01 -3.26097339e-01 3.17842871e-01 5.44667721e-01 5.97960889e-01 -9.84381437e-02 5.16659200e-01 8.86171162e-01 3.91276091e-01 -1.21541965e+00 -6.56909943e-02 -9.72268462e-01 -9.31586504e-01 -2.68793821e-01 6.55566752e-01 -5.88596940e-01 -6.03731036e-01 6.62190557e-01 -1.19308019e+00 -3.97770703e-01 -2.60401428e-01 9.18902755e-02 -4.60973501e-01 4.46512997e-01 -6.00532770e-01 -7.91388333e-01 -5.27448714e-01 -1.08306646e+00 1.08346272e+00 7.33068228e-01 -5.77593632e-02 -8.07806730e-01 1.49718106e-01 -2.04895288e-01 2.77474344e-01 8.12816560e-01 3.65845621e-01 -3.49228293e-01 -6.17887855e-01 -4.83332634e-01 -4.30949807e-01 5.19911908e-02 -1.19382702e-01 7.42404938e-01 -7.02760577e-01 -9.93867218e-02 -3.49755973e-01 1.01046115e-01 8.89185429e-01 6.97369397e-01 5.01976728e-01 1.99374124e-01 -5.72835147e-01 5.58476925e-01 1.93066490e+00 5.51934466e-02 5.56714475e-01 4.75689262e-01 4.21726257e-01 5.13899028e-01 6.04410887e-01 4.02467698e-01 5.20759523e-02 8.01704943e-01 2.58671492e-01 -3.72719496e-01 -2.28104606e-01 2.90444847e-02 3.02275300e-01 3.62599283e-01 -2.60770768e-02 4.42601927e-03 -1.02020919e+00 5.32295704e-01 -1.75310326e+00 -7.54328072e-01 -7.41385520e-01 2.39548087e+00 4.29233730e-01 6.67970598e-01 5.91294646e-01 3.84467542e-01 8.23652744e-01 -2.63018072e-01 -5.08743346e-01 -8.00255835e-01 -6.86184093e-02 3.63465279e-01 8.81799996e-01 4.36052650e-01 -1.31478822e+00 7.38260627e-01 6.46121597e+00 6.34881735e-01 -1.47609437e+00 -2.74542183e-01 4.87114489e-01 2.74236262e-01 5.17171860e-01 1.71832666e-02 -1.09796822e+00 1.96699381e-01 6.46570563e-01 1.17908362e-02 -1.33105919e-01 9.45083618e-01 1.95471615e-01 -6.17922127e-01 -1.16732287e+00 8.47692430e-01 -1.35495350e-01 -1.22261763e+00 -5.44587433e-01 -1.11363187e-01 6.92632020e-01 3.95599246e-01 -1.32558987e-01 -5.20465732e-01 2.60081259e-04 -5.33455789e-01 8.59569967e-01 9.25833285e-02 5.61568439e-01 -6.70268714e-01 6.76828504e-01 3.52208465e-01 -1.35478079e+00 2.43718430e-01 -3.47959042e-01 8.63498971e-02 1.93766460e-01 7.36047268e-01 -1.23007786e+00 1.23219110e-01 4.07307237e-01 4.31743979e-01 -9.98080909e-01 1.84997070e+00 -3.08880564e-02 7.18750536e-01 -9.94883955e-01 -1.50710046e-01 2.73525238e-01 -2.04521477e-01 3.94833058e-01 1.71286321e+00 2.25841761e-01 -5.49416006e-01 -1.93018056e-02 3.77087981e-01 2.33246535e-01 3.93311977e-01 -6.27596498e-01 5.67847311e-01 4.96395528e-01 1.23997629e+00 -1.59201193e+00 -1.59714535e-01 -4.64254111e-01 1.13706791e+00 1.28201023e-01 -1.95692237e-02 -7.77262807e-01 -7.28672326e-01 2.34715596e-01 2.69316643e-01 5.64119399e-01 -5.77903509e-01 -4.84381706e-01 -7.49147356e-01 1.04596518e-01 -3.38893116e-01 3.07589442e-01 -3.04689974e-01 -7.42735445e-01 3.74741554e-01 -3.55993569e-01 -1.25170887e+00 -1.96029812e-01 -7.07312882e-01 -6.77814960e-01 4.04459268e-01 -1.40946698e+00 -1.06700587e+00 -5.27559936e-01 5.29523790e-01 5.88835359e-01 6.27417684e-01 2.90873736e-01 7.96833411e-02 -5.64464808e-01 6.31503463e-01 4.02018249e-01 3.54661316e-01 5.77471495e-01 -1.06680369e+00 8.86229277e-01 1.23503184e+00 3.86526763e-01 5.10856509e-01 9.16778743e-01 -3.11861217e-01 -1.04516613e+00 -6.35726511e-01 6.49957955e-01 2.02864292e-03 4.20742005e-01 -3.99387091e-01 -6.91386044e-01 3.82458985e-01 5.96485734e-02 1.37981519e-01 2.67754197e-01 -2.52355188e-01 -3.88333738e-01 -1.20196685e-01 -1.14278805e+00 4.67807651e-01 9.17113841e-01 -1.97694376e-01 -2.30878457e-01 5.16831875e-02 3.34992379e-01 -6.25239789e-01 -5.16014457e-01 1.63590953e-01 7.57931471e-01 -1.63811612e+00 1.03412366e+00 7.84240440e-02 1.91098794e-01 -5.33873677e-01 2.86010951e-01 -6.96107626e-01 -1.33630380e-01 -5.40876627e-01 1.09605536e-01 1.37938499e+00 4.93127674e-01 -4.57001805e-01 9.74370122e-01 2.25857005e-01 -9.73449741e-03 -7.51964211e-01 -1.09344232e+00 -8.35640609e-01 -3.06928068e-01 -2.49656975e-01 2.23524928e-01 3.82766843e-01 -9.99287665e-02 1.99883580e-01 3.09990142e-02 1.79991588e-01 7.27895677e-01 3.91565055e-01 8.59318256e-01 -1.26011920e+00 -2.81752013e-02 -8.67374897e-01 -1.03970468e+00 -1.21993196e+00 -7.11887360e-01 -5.37090898e-01 9.81626287e-02 -1.50171149e+00 -8.37931931e-02 -5.61886549e-01 -1.72691792e-02 1.60018995e-01 1.09958909e-02 5.82215846e-01 1.88892394e-01 2.25744113e-01 -5.59075415e-01 -4.34302360e-01 6.87164545e-01 2.99352467e-01 -6.05750144e-01 2.72014022e-01 -5.26944548e-02 8.42500925e-01 9.35160339e-01 -3.00593823e-01 4.93129715e-02 -2.20822036e-01 1.71425506e-01 -3.52498710e-01 2.58655071e-01 -1.48795283e+00 4.70266074e-01 1.81010246e-01 2.74754852e-01 -7.12773800e-01 1.26421005e-01 -8.32086444e-01 -8.40276033e-02 8.14058661e-01 -7.25134090e-02 1.36216447e-01 1.40817121e-01 3.39608699e-01 1.12137496e-01 -6.34554505e-01 1.08803606e+00 3.67644280e-01 -8.78373623e-01 3.24255787e-02 -2.50370979e-01 -1.03194781e-01 1.45516634e+00 -6.18710101e-01 -4.29744065e-01 5.89119978e-02 -1.17561631e-01 1.48169130e-01 9.26471114e-01 7.90837035e-02 5.48178852e-01 -8.40702891e-01 -5.83963454e-01 1.02351196e-01 2.05062091e-01 -5.09779081e-02 -2.17391744e-01 8.79890025e-01 -1.57273436e+00 3.48815054e-01 -3.78531039e-01 -9.54454660e-01 -1.66380298e+00 5.15107334e-01 2.41938621e-01 -1.16157547e-01 -7.08009660e-01 7.39769340e-01 -3.17352295e-01 5.22342205e-01 3.02792966e-01 -7.74368167e-01 4.87962998e-02 7.32465312e-02 4.29012030e-01 6.31881595e-01 1.56559870e-01 -6.45075977e-01 -4.22349095e-01 1.10807359e+00 -1.06794104e-01 1.18378829e-02 9.30905282e-01 -1.74989760e-01 1.37973428e-01 3.96497071e-01 1.03372252e+00 1.45224288e-01 -1.40482593e+00 -3.69123779e-02 3.27282757e-01 -3.13862056e-01 6.97352812e-02 -2.31865704e-01 -8.18945646e-01 8.07568610e-01 9.18084204e-01 3.98167998e-01 1.11910403e+00 6.05207169e-03 8.03308904e-01 1.08151324e-01 5.75645030e-01 -1.02666521e+00 -3.89802992e-01 9.81379375e-02 3.21153432e-01 -1.02495766e+00 2.44245827e-01 -7.12277174e-01 -2.75023252e-01 1.45402145e+00 1.16372809e-01 -5.38986385e-01 5.85858583e-01 5.66338599e-01 2.36718565e-01 -1.93775985e-02 -1.97510704e-01 -3.70857090e-01 1.10227302e-01 5.57730436e-01 3.02596241e-01 -3.43981870e-02 -7.79369771e-01 -3.08373749e-01 4.03753556e-02 1.53791830e-02 4.83192384e-01 9.85510468e-01 -8.29686165e-01 -1.09033072e+00 -4.64423120e-01 3.72442007e-01 -3.58442187e-01 3.10923785e-01 -5.57459474e-01 7.67454267e-01 2.50056744e-01 1.02727616e+00 8.75248909e-02 -1.75933257e-01 3.32108527e-01 -3.22287709e-01 5.02564073e-01 -1.61685571e-01 -7.76698709e-01 1.01412222e-01 -1.39446765e-01 -6.43473744e-01 -4.87597436e-01 -8.61051619e-01 -1.44761586e+00 -2.46866390e-01 -5.31248510e-01 -3.45181882e-01 9.71364796e-01 5.51309347e-01 3.48444611e-01 -6.13364503e-02 5.65002620e-01 -8.31578910e-01 -1.83049925e-02 -4.59200710e-01 -3.67043465e-01 2.78270781e-01 4.02959287e-01 -3.54187548e-01 -1.84016258e-01 2.43170679e-01]
[8.287923812866211, -1.5581036806106567]
a90d94d3-80ee-4a6c-9367-23364ecde054
argument-pair-extraction-with-mutual-guidance
null
null
https://aclanthology.org/2021.emnlp-main.319
https://aclanthology.org/2021.emnlp-main.319.pdf
Argument Pair Extraction with Mutual Guidance and Inter-sentence Relation Graph
Argument pair extraction (APE) aims to extract interactive argument pairs from two passages of a discussion. Previous work studied this task in the context of peer review and rebuttal, and decomposed it into a sequence labeling task and a sentence relation classification task. However, despite the promising performance, such an approach obtains the argument pairs implicitly by the two decomposed tasks, lacking explicitly modeling of the argument-level interactions between argument pairs. In this paper, we tackle the APE task by a mutual guidance framework, which could utilize the information of an argument in one passage to guide the identification of arguments that can form pairs with it in another passage. In this manner, two passages can mutually guide each other in the process of APE. Furthermore, we propose an inter-sentence relation graph to effectively model the inter-relations between two sentences and thus facilitates the extraction of argument pairs. Our proposed method can better represent the holistic argument-level semantics and thus explicitly capture the complex correlations between argument pairs. Experimental results show that our approach significantly outperforms the current state-of-the-art model.
['Ruifeng Xu', 'Min Yang', 'Yice Zhang', 'Jingyi Sun', 'Bin Liang', 'Jianzhu Bao']
null
null
null
null
emnlp-2021-11
['argument-pair-extraction-ape', 'relation-classification']
['natural-language-processing', 'natural-language-processing']
[ 3.76968920e-01 5.31325102e-01 -3.05168301e-01 -3.53635579e-01 -6.35215282e-01 -5.77560425e-01 1.00259233e+00 9.14413154e-01 -2.93178290e-01 7.02553451e-01 4.96534109e-01 -6.59392953e-01 -3.39335322e-01 -8.67540359e-01 -4.79387224e-01 -2.96055913e-01 2.80734777e-01 3.43566090e-01 3.42607677e-01 -5.72207868e-01 5.81836522e-01 -6.49084151e-02 -1.42258811e+00 6.55025542e-01 1.22952855e+00 5.72925746e-01 2.11744774e-02 4.31080997e-01 -7.76950717e-01 1.09836912e+00 -6.50589228e-01 -7.45425642e-01 -2.55931169e-01 -8.19402277e-01 -1.23427606e+00 -2.07824156e-01 4.28238772e-02 2.16654107e-01 -1.16931051e-02 9.36732590e-01 1.89731926e-01 -1.83738936e-02 6.54073656e-01 -1.09940720e+00 -3.30792427e-01 1.15766191e+00 -4.20745492e-01 3.30575258e-01 5.64117193e-01 -3.13286275e-01 1.56421995e+00 -7.15843022e-01 6.88510299e-01 1.39167893e+00 3.02385777e-01 2.18812391e-01 -8.09212923e-01 -6.11976683e-01 5.16953886e-01 4.03561860e-01 -5.63759148e-01 -2.52486378e-01 1.18462491e+00 -3.44503820e-01 7.83567369e-01 2.89833635e-01 8.02275240e-01 8.15080225e-01 -2.04730779e-01 1.03037786e+00 7.94035614e-01 -4.89448518e-01 -1.78981572e-01 -9.12066177e-02 9.68430638e-01 4.15813208e-01 1.27893046e-01 -5.88837981e-01 -6.19858563e-01 -4.17549103e-01 2.19003960e-01 -3.11438024e-01 -3.43585491e-01 1.03754014e-01 -1.06863070e+00 7.01786757e-01 3.67316782e-01 5.25626242e-01 -3.34935725e-01 -3.05161983e-01 6.15056455e-01 3.44812214e-01 4.27831501e-01 5.12844443e-01 -3.62116992e-01 -2.11343676e-01 -4.81993109e-01 3.83027762e-01 1.16035819e+00 7.18692720e-01 6.49417877e-01 -9.24041688e-01 -3.19042325e-01 7.40808368e-01 5.00558913e-01 -2.18531154e-02 1.76638961e-01 -6.15474224e-01 1.07603157e+00 1.21891940e+00 3.54030132e-02 -1.22017920e+00 -3.17706108e-01 -3.18634927e-01 -5.10685861e-01 -4.50339884e-01 4.06945080e-01 -2.04969049e-01 -6.33791089e-02 1.51267898e+00 5.54241240e-01 1.27610505e-01 3.90284300e-01 6.28293812e-01 1.36628759e+00 4.78367299e-01 6.27209023e-02 -5.15881836e-01 1.68773603e+00 -1.16305101e+00 -1.11175680e+00 -3.41206610e-01 9.46797073e-01 -1.08692539e+00 7.43778050e-01 -7.41658062e-02 -1.12284625e+00 -3.38584274e-01 -1.14984846e+00 -3.50788414e-01 -4.72626537e-02 2.07431599e-01 6.51791811e-01 -1.03496537e-01 -3.52075189e-01 5.89224219e-01 -3.37690979e-01 -1.69679131e-02 3.18335801e-01 1.17692798e-01 -1.76519632e-01 2.20735788e-01 -1.74030232e+00 8.78297269e-01 3.35934639e-01 3.13570797e-01 1.30292535e-01 -4.76586550e-01 -9.40265417e-01 1.91516533e-01 7.94923007e-01 -7.33332396e-01 1.18055093e+00 -7.41118431e-01 -1.30078042e+00 8.31471384e-01 -3.97979379e-01 -3.75853896e-01 3.51632863e-01 -5.37442088e-01 -2.86783338e-01 1.71429798e-01 2.59214371e-01 1.61283031e-01 4.05372381e-01 -1.09967434e+00 -4.64939207e-01 -3.33729386e-01 4.49526876e-01 6.42150223e-01 -2.78444052e-01 2.98838645e-01 -4.97302741e-01 -3.55674773e-01 2.48457953e-01 -6.82222188e-01 -4.82624471e-02 -2.15105549e-01 -6.32302821e-01 -1.00226033e+00 6.83602810e-01 -5.16838372e-01 1.51304901e+00 -2.01306486e+00 2.62993544e-01 3.22559066e-02 4.97457623e-01 4.93458748e-01 -9.31838453e-02 8.47990036e-01 -5.42371273e-02 4.19268101e-01 -2.59645760e-01 -2.10886776e-01 -1.17323354e-01 2.18543917e-01 -3.25970381e-01 1.40438825e-01 2.35851377e-01 9.48136449e-01 -1.21201324e+00 -7.98724890e-01 -2.08134130e-01 1.64135173e-01 -1.80030331e-01 3.92448604e-01 -3.03761303e-01 3.62982541e-01 -8.05136323e-01 1.76571961e-02 6.09787405e-01 -3.95296425e-01 6.66319847e-01 -2.91015804e-01 1.97177343e-02 1.03964877e+00 -7.66053200e-01 1.52012086e+00 -4.15279865e-01 6.27967477e-01 5.52653931e-02 -1.23454249e+00 1.12730753e+00 3.27694088e-01 1.58923984e-01 -5.63105881e-01 1.33008674e-01 3.44140112e-01 5.35142064e-01 -7.12349415e-01 2.64525056e-01 2.59989202e-01 -1.54622074e-03 7.55163908e-01 -3.71540576e-01 -2.65241712e-02 6.34862065e-01 6.46504223e-01 9.23874974e-01 1.00070655e-01 5.70953310e-01 -6.87713921e-02 1.15703690e+00 -2.27324307e-01 6.35191321e-01 6.70639813e-01 6.43074960e-02 2.28805155e-01 1.07298517e+00 -2.57447064e-01 -7.09309638e-01 -5.82476437e-01 -5.99520188e-03 5.56943059e-01 5.85594773e-01 -9.13957894e-01 -6.14362478e-01 -1.19167578e+00 -1.01450361e-01 5.25071740e-01 -7.11638093e-01 1.21330380e-01 -1.07364559e+00 -4.42559361e-01 3.71249765e-01 2.62839854e-01 5.90715706e-01 -1.09943998e+00 -2.42945582e-01 2.81394154e-01 -7.55060017e-01 -1.21586382e+00 -2.77830511e-01 -1.06551886e-01 -5.92829525e-01 -1.78481376e+00 -1.48444906e-01 -7.88459957e-01 5.85206270e-01 2.00592205e-01 1.17745256e+00 7.03119159e-01 1.87864915e-01 -1.02972910e-01 -5.32297015e-01 -2.51037538e-01 -5.43446779e-01 5.09409606e-02 -4.57799137e-01 -5.64614916e-03 5.05293190e-01 -5.59813976e-01 -5.21888793e-01 2.25935042e-01 -6.45277977e-01 4.92125630e-01 3.83052528e-01 9.76420164e-01 2.10225612e-01 -3.31479132e-01 8.19601655e-01 -1.29443848e+00 1.18725646e+00 -7.21090496e-01 -7.83658624e-02 6.41819656e-01 -5.35526991e-01 1.78254604e-01 6.16168439e-01 -1.59414113e-01 -1.32641351e+00 -7.66801536e-01 -1.72250628e-01 3.70867342e-01 1.46865934e-01 7.38535166e-01 -3.04652303e-01 5.53172052e-01 2.10554555e-01 -6.80413321e-02 1.58165708e-01 -3.10959220e-01 4.77507234e-01 6.85147107e-01 2.39985675e-01 -7.71680117e-01 5.50361097e-01 1.88277274e-01 1.05431862e-01 -5.20202219e-01 -1.24070561e+00 -5.66522241e-01 -7.09024072e-01 -9.36822444e-02 6.94617629e-01 -6.00330234e-01 -9.04428899e-01 1.34524420e-01 -1.72840202e+00 1.98419914e-01 -5.38066737e-02 2.91385829e-01 -1.05753474e-01 8.05137038e-01 -7.11058378e-01 -7.26231873e-01 -4.81291085e-01 -9.33969557e-01 8.15029621e-01 4.20814872e-01 -5.58516741e-01 -1.16634262e+00 8.44180305e-03 6.95312083e-01 -1.99236616e-01 -8.98430869e-02 1.11392653e+00 -1.18592978e+00 -2.47804403e-01 7.55359046e-03 -4.22193736e-01 6.97186515e-02 2.41730779e-01 8.30911994e-02 -4.17598367e-01 2.86751717e-01 2.09052756e-01 -5.94739318e-02 7.39093244e-01 -1.16530143e-01 7.61118412e-01 -4.44492936e-01 -4.85014379e-01 -2.56951042e-02 6.27399623e-01 5.73842674e-02 4.24819350e-01 4.82608527e-01 7.13966846e-01 1.26663291e+00 1.03738427e+00 1.45757616e-01 6.49542451e-01 6.96421683e-01 3.11425894e-01 -9.44309607e-02 -9.69865695e-02 -2.17175350e-01 4.32521887e-02 1.13527811e+00 8.96641463e-02 -4.94042337e-01 -7.62848675e-01 4.15236354e-01 -2.54392529e+00 -8.84976029e-01 -9.88563895e-01 1.62715948e+00 9.91306782e-01 2.61358380e-01 -1.30744040e-01 2.66426295e-01 8.23993146e-01 3.03939193e-01 -1.58781081e-01 -5.18289804e-01 -2.35933274e-01 -1.33279994e-01 -3.39626670e-01 4.76563603e-01 -9.04242694e-01 7.26042747e-01 5.17474079e+00 7.48311281e-01 -3.12333316e-01 -2.71718144e-01 5.69389880e-01 3.85221779e-01 -6.68616295e-01 4.56819206e-01 -8.19359958e-01 4.37380522e-01 5.13247430e-01 -4.29781109e-01 -1.85238168e-01 3.02713186e-01 2.39735603e-01 -3.87057327e-02 -1.23351133e+00 5.36094189e-01 6.44361824e-02 -1.36722839e+00 1.32834435e-01 -1.22169122e-01 4.21451807e-01 -6.55905724e-01 -3.86792243e-01 9.98743922e-02 8.12761337e-02 -6.79957628e-01 2.40063250e-01 3.73229414e-01 -7.88661167e-02 -6.09545588e-01 1.07801020e+00 5.45637429e-01 -1.21747243e+00 1.87458843e-02 2.85798069e-02 -5.28999031e-01 3.63448650e-01 7.70468652e-01 -6.70550704e-01 1.08527827e+00 2.09185421e-01 1.15274096e+00 -4.72586155e-01 6.88915372e-01 -1.04283857e+00 6.50294960e-01 -9.31150615e-02 -3.56825560e-01 3.08247536e-01 -5.65787911e-01 7.34909415e-01 1.00409341e+00 -7.16755912e-02 4.04907286e-01 2.19151482e-01 6.85416877e-01 -2.48652354e-01 5.42942226e-01 -4.31730151e-01 -5.84830120e-02 6.74054265e-01 1.35691345e+00 -6.01057231e-01 -6.14715874e-01 -4.34924603e-01 5.77821195e-01 6.77868366e-01 1.15855142e-01 -6.86820388e-01 -3.49699974e-01 5.48824728e-01 -1.33170143e-01 5.06936759e-02 1.88691959e-01 -2.89837867e-01 -1.13935328e+00 5.48903942e-01 -8.03270876e-01 4.71552044e-01 -5.51488876e-01 -1.44245148e+00 5.67764163e-01 -1.00869253e-01 -1.23317969e+00 -1.85410187e-01 -1.87898457e-01 -1.07758188e+00 9.48647141e-01 -1.49252915e+00 -1.21390498e+00 -1.04381396e-02 1.00493617e-01 6.93131804e-01 2.75238335e-01 5.84502399e-01 9.60990340e-02 -6.34437084e-01 3.25036734e-01 -3.45619947e-01 3.75891238e-01 6.21065021e-01 -1.21214926e+00 4.82909888e-01 9.01834130e-01 2.08409607e-01 1.07710361e+00 4.96119469e-01 -8.35751653e-01 -9.57860231e-01 -4.67350453e-01 1.51012409e+00 -2.70287722e-01 9.48580861e-01 -1.40722632e-01 -1.26037335e+00 2.54363507e-01 5.33708036e-01 -2.73852170e-01 9.86722291e-01 5.43884814e-01 -3.40022475e-01 2.83483118e-01 -5.03575206e-01 7.45334864e-01 1.25188196e+00 -6.62315190e-01 -1.34994519e+00 3.18936706e-01 8.70660186e-01 -3.29663426e-01 -6.03468299e-01 4.23850775e-01 3.30173969e-01 -9.07127500e-01 8.96788716e-01 -7.06331551e-01 9.93065536e-01 -3.99412125e-01 5.52561820e-01 -1.09796441e+00 6.19224869e-02 -6.22839272e-01 -2.96432555e-01 1.76463723e+00 6.35000587e-01 -5.88842690e-01 3.44053358e-01 5.13468146e-01 -9.25670117e-02 -9.94378567e-01 -4.46466744e-01 -3.39790076e-01 3.64370225e-03 -1.51404902e-01 6.57926321e-01 8.62122834e-01 6.71208024e-01 9.84065652e-01 -2.44132038e-02 -9.19821337e-02 5.11075318e-01 6.79876626e-01 7.40867794e-01 -1.45656526e+00 -2.70334274e-01 -6.43559694e-01 5.83105758e-02 -1.33703876e+00 6.79078281e-01 -9.73697722e-01 -9.40173045e-02 -1.94265699e+00 3.46194834e-01 -4.42803591e-01 -4.14530337e-02 1.46937028e-01 -1.07259643e+00 -4.01468158e-01 1.66727036e-01 4.43489820e-01 -7.25314856e-01 6.74404621e-01 1.47711730e+00 -2.02593088e-01 -2.19178170e-01 3.81489843e-02 -8.96820128e-01 9.14257109e-01 5.39834261e-01 -4.62289065e-01 -6.13761723e-01 -1.98348299e-01 5.40353000e-01 2.13903666e-01 7.28657767e-02 -2.14143977e-01 5.39681017e-01 -9.12646130e-02 -2.38974169e-01 -6.73518002e-01 -6.48255646e-02 -4.71492618e-01 -3.43964487e-01 1.26886025e-01 -7.61986852e-01 3.06802476e-03 -1.06583685e-01 7.17002094e-01 -6.64052308e-01 -5.00588119e-01 -4.42507491e-03 9.66030955e-02 -2.88650662e-01 -1.49875969e-01 -1.50735021e-01 2.79801041e-01 9.01651323e-01 1.14464320e-01 -6.94144964e-01 -2.65082240e-01 -6.27649665e-01 7.22950637e-01 1.66958541e-01 5.64339161e-01 5.99793017e-01 -1.03290057e+00 -1.03070283e+00 -2.20899045e-01 2.62689851e-02 3.12191308e-01 1.78470835e-01 9.49160576e-01 -1.08449824e-01 1.66042238e-01 3.59540522e-01 -4.07830924e-01 -1.89029825e+00 3.31924796e-01 -7.11456314e-02 -9.61907923e-01 -5.82754493e-01 7.70285547e-01 3.05475801e-01 -3.80369306e-01 -1.33041501e-01 5.37283383e-02 -1.13299036e+00 4.20945674e-01 5.20840943e-01 1.94154829e-02 -1.59076974e-01 -5.73996961e-01 -4.15235758e-01 5.84064424e-01 -4.04389620e-01 1.19258547e-02 1.16765642e+00 -2.45642155e-01 -7.83581734e-01 4.38529581e-01 1.08987784e+00 2.11194515e-01 -7.81804144e-01 -4.04571921e-01 3.91852081e-01 -2.23152757e-01 -4.18494195e-01 -4.94841337e-01 -4.71049756e-01 8.55857670e-01 -5.66406250e-01 4.96187299e-01 7.27307737e-01 1.83348700e-01 8.45724106e-01 3.28517169e-01 -2.22860873e-01 -8.60154808e-01 1.29947931e-01 8.06428969e-01 8.05152595e-01 -1.12115538e+00 2.77435720e-01 -1.25596023e+00 -5.70109844e-01 1.40919447e+00 7.26712704e-01 2.31032461e-01 4.46577996e-01 7.20829293e-02 1.49806670e-03 -6.00801289e-01 -9.61937726e-01 -1.80569649e-01 5.74871719e-01 8.04107338e-02 8.80455256e-01 -2.19543561e-01 -1.25994766e+00 7.51557112e-01 -2.83535966e-03 -4.31592405e-01 2.86797941e-01 8.58122408e-01 -1.23656847e-01 -1.72564626e+00 2.05705017e-02 3.59081596e-01 -3.65231782e-01 -3.73055458e-01 -8.66862416e-01 5.21715462e-01 -6.11481182e-02 1.25395000e+00 7.30048642e-02 7.24891527e-03 2.03545645e-01 1.15757130e-01 2.27653950e-01 -6.54309154e-01 -1.19446027e+00 -1.37434915e-01 7.78482795e-01 -1.73408538e-01 -8.14156175e-01 -6.38379097e-01 -1.56034815e+00 -1.70920957e-02 -6.25364482e-01 8.26981127e-01 4.63365763e-01 1.52199483e+00 3.41653347e-01 7.59332955e-01 6.10798776e-01 -1.84943140e-01 -3.17854255e-01 -9.39941347e-01 -7.88174868e-02 8.15349281e-01 3.75194848e-02 -5.55619359e-01 -4.39897954e-01 -2.27423251e-01]
[9.757883071899414, 9.205171585083008]
71c45b6c-e393-4bf8-8930-f2dd3145574e
attention-enhanced-cross-modal-localization
2212.02757
null
https://arxiv.org/abs/2212.02757v1
https://arxiv.org/pdf/2212.02757v1.pdf
Attention-Enhanced Cross-modal Localization Between 360 Images and Point Clouds
Visual localization plays an important role for intelligent robots and autonomous driving, especially when the accuracy of GNSS is unreliable. Recently, camera localization in LiDAR maps has attracted more and more attention for its low cost and potential robustness to illumination and weather changes. However, the commonly used pinhole camera has a narrow Field-of-View, thus leading to limited information compared with the omni-directional LiDAR data. To overcome this limitation, we focus on correlating the information of 360 equirectangular images to point clouds, proposing an end-to-end learnable network to conduct cross-modal visual localization by establishing similarity in high-dimensional feature space. Inspired by the attention mechanism, we optimize the network to capture the salient feature for comparing images and point clouds. We construct several sequences containing 360 equirectangular images and corresponding point clouds based on the KITTI-360 dataset and conduct extensive experiments. The results demonstrate the effectiveness of our approach.
['Sebastian Scherer', 'Wen Yang', 'Chenwei Lyv', 'Huai Yu', 'Zhipeng Zhao']
2022-12-06
null
null
null
null
['camera-localization', 'visual-localization']
['computer-vision', 'computer-vision']
[-4.31053370e-01 -6.33511603e-01 -3.38815272e-01 -6.45702302e-01 -3.58893543e-01 -4.77438509e-01 4.01286274e-01 -1.35079324e-01 -7.53327549e-01 5.42485714e-01 -2.48142079e-01 -2.21474499e-01 -3.53189170e-01 -7.43225873e-01 -8.15236330e-01 -5.19892335e-01 7.91281834e-02 3.27692002e-01 1.04487561e-01 -1.17244750e-01 3.33945304e-01 7.62541294e-01 -1.55716467e+00 -5.58445275e-01 9.76088285e-01 8.37109208e-01 7.83479691e-01 1.34996876e-01 -1.70927748e-01 3.28636169e-01 -1.54743731e-01 9.43232924e-02 3.37713182e-01 1.22847602e-01 -1.92231447e-01 2.85978615e-02 6.55411661e-01 -1.55879736e-01 -4.14504200e-01 1.38223660e+00 4.17643875e-01 1.73378587e-01 3.08877319e-01 -1.58485973e+00 -9.56963181e-01 7.30423629e-02 -9.31856453e-01 1.92149863e-01 1.52603045e-01 7.44452402e-02 8.01684737e-01 -1.04823375e+00 3.87442172e-01 1.24226069e+00 6.40859008e-01 8.17502290e-02 -8.11208844e-01 -8.43538582e-01 2.89148986e-01 4.48448330e-01 -1.80081666e+00 -1.77528188e-01 8.08927715e-01 -3.43073726e-01 5.60072720e-01 -3.71315211e-01 6.17247343e-01 7.09051549e-01 3.48344475e-01 4.81976420e-01 7.63970494e-01 -7.83800185e-02 5.72094545e-02 9.62045267e-02 -2.27518976e-01 8.26458573e-01 4.69541073e-01 2.73925871e-01 -4.96893853e-01 2.16642693e-01 7.83206046e-01 4.77128685e-01 -4.09098327e-01 -9.34436679e-01 -1.34734035e+00 8.46429706e-01 1.04787755e+00 1.49824142e-01 -2.61516035e-01 3.20164561e-01 1.39851525e-01 1.40537769e-01 9.12950709e-02 2.57308722e-01 -1.86969265e-01 1.08217880e-01 -4.73927230e-01 -5.81177808e-02 1.70446649e-01 1.26260340e+00 1.16019630e+00 1.56715795e-01 5.84392011e-01 7.97388911e-01 6.14581227e-01 9.55678821e-01 4.18918163e-01 -8.79900932e-01 5.82723975e-01 6.64337695e-01 2.21202433e-01 -1.29886508e+00 -5.60180008e-01 -4.82097179e-01 -9.43651438e-01 2.76315749e-01 -3.99790658e-03 5.34278899e-02 -7.05100954e-01 1.60164189e+00 1.82634667e-01 1.93449587e-01 2.66492553e-02 1.34141815e+00 6.42193377e-01 5.40531158e-01 -9.90889445e-02 1.75063103e-01 1.14441109e+00 -8.29478920e-01 -4.66721684e-01 -6.87663019e-01 3.04755419e-01 -5.39316893e-01 9.99511480e-01 -1.28410950e-01 -3.36900234e-01 -7.82650888e-01 -1.29825950e+00 -9.78242755e-02 -3.66933316e-01 1.97130471e-01 5.57422936e-01 2.47025371e-01 -9.79948223e-01 6.79723034e-03 -7.50216603e-01 -5.47032833e-01 1.80801347e-01 2.84184545e-01 -6.56737030e-01 -3.85063261e-01 -1.11520159e+00 1.04910553e+00 4.09883410e-01 2.80697078e-01 -4.95777994e-01 -4.12492871e-01 -1.23955286e+00 -7.43184686e-02 1.75175026e-01 -5.21378577e-01 7.99072266e-01 -3.75084192e-01 -1.13742757e+00 6.35214925e-01 -2.35511437e-01 -3.73325557e-01 3.92330438e-01 -2.33091488e-01 -3.77352715e-01 3.17829698e-02 6.18410170e-01 9.58480060e-01 6.07958555e-01 -1.25982511e+00 -9.98502672e-01 -5.24036348e-01 1.04650982e-01 4.04041439e-01 -6.87593073e-02 -4.24289495e-01 -5.25855601e-01 -8.65905359e-02 7.04119921e-01 -1.03755665e+00 -2.24513784e-01 2.43911698e-01 -1.65358782e-01 -2.05360308e-01 1.14455199e+00 4.60094139e-02 4.29810196e-01 -2.17242837e+00 -1.82480812e-01 -3.80727574e-02 3.21511589e-02 -8.46943706e-02 -1.06406733e-01 2.18508214e-01 1.20663084e-01 -1.24364153e-01 -5.45460582e-02 -2.80677646e-01 -1.85834005e-01 3.82576674e-01 -3.76866251e-01 7.90779173e-01 1.42401904e-01 9.64498580e-01 -1.05312788e+00 -5.74452758e-01 7.28388965e-01 4.48009282e-01 -1.26412645e-01 6.66368455e-02 1.97912186e-01 5.52093267e-01 -7.04123557e-01 8.43418241e-01 8.91042590e-01 -1.53768823e-01 -4.52735394e-01 -2.49510810e-01 -5.15768051e-01 -1.21835597e-01 -9.22322869e-01 1.92530644e+00 -6.92994177e-01 9.96720374e-01 -1.51191399e-01 -7.82087624e-01 1.32843304e+00 -2.78045505e-01 2.70696759e-01 -1.01136708e+00 -1.38888853e-02 3.47895294e-01 -1.99933350e-01 -3.95994037e-01 6.36083245e-01 -1.04610130e-01 -2.83609301e-01 -7.74630159e-02 -1.33982718e-01 -5.08443356e-01 -1.07292540e-01 -1.97652243e-02 5.27946651e-01 1.08967595e-01 3.57895702e-01 -8.55486020e-02 6.51985109e-01 1.03535689e-01 6.05107009e-01 5.88804126e-01 -2.49954224e-01 6.51236594e-01 -1.66562602e-01 -5.94864130e-01 -9.50425386e-01 -1.01902020e+00 -2.67147928e-01 3.94310236e-01 9.53746498e-01 -2.16294117e-02 -1.06529094e-01 -4.03593153e-01 2.41268843e-01 3.57909560e-01 -3.29756975e-01 -1.05496049e-01 -5.01512885e-01 -2.01304004e-01 3.42266291e-01 6.07964098e-01 8.42983723e-01 -8.03332925e-01 -8.87356043e-01 1.81326326e-02 -2.01866627e-01 -1.31831586e+00 -3.44253153e-01 1.80142000e-01 -6.52501464e-01 -9.44925904e-01 -5.38377583e-01 -9.32378590e-01 6.10126257e-01 1.13535929e+00 8.27876925e-01 -1.28571451e-01 -1.50024462e-02 3.78215820e-01 -6.46158606e-02 -4.11659271e-01 3.06945056e-01 1.42372638e-01 3.29907864e-01 -1.18633665e-01 6.27575636e-01 -5.10733604e-01 -5.51008165e-01 5.09243846e-01 -5.42167664e-01 -7.86947682e-02 8.27393711e-01 7.10661232e-01 6.92184448e-01 -6.92748502e-02 4.39454466e-01 -1.24827847e-01 2.31435195e-01 -5.37262440e-01 -1.09502852e+00 -1.26217425e-01 -4.79185909e-01 -9.19491351e-02 6.54369235e-01 -7.79126883e-02 -5.48022866e-01 4.23199862e-01 1.34331256e-01 -7.99323678e-01 -1.92059934e-01 6.93717837e-01 -1.35530010e-01 -4.85049069e-01 3.00314158e-01 3.11272889e-01 -6.86251000e-02 -1.16546057e-01 3.91696900e-01 6.74811721e-01 8.48540962e-01 -2.11879909e-01 1.16756022e+00 7.46683300e-01 3.86201218e-02 -1.01993680e+00 -6.95947230e-01 -6.33329332e-01 -7.61144936e-01 -3.03093046e-01 7.57408977e-01 -1.22054148e+00 -8.89511049e-01 2.16177806e-01 -1.19342387e+00 2.90393829e-01 1.19437501e-01 8.69648099e-01 -5.02731621e-01 5.00950754e-01 -7.14701042e-02 -7.10457981e-01 -6.64600283e-02 -1.25094950e+00 1.19580090e+00 5.92257500e-01 2.25974858e-01 -7.16056168e-01 9.31550190e-02 4.92380559e-02 1.64661780e-01 3.25639509e-02 5.65598786e-01 -7.36466870e-02 -1.09379768e+00 -3.57777268e-01 -6.46254718e-01 5.93842044e-02 1.32511944e-01 -3.15369219e-01 -6.50650740e-01 -3.95352155e-01 -5.13852760e-02 -3.89740944e-01 7.72032559e-01 3.60000044e-01 9.14389789e-01 3.68106425e-01 -5.20342827e-01 9.08438742e-01 1.65592849e+00 2.01754004e-01 2.31652826e-01 6.86789632e-01 8.92839193e-01 4.40199465e-01 1.00628710e+00 1.57143369e-01 7.09972262e-01 6.18373275e-01 9.66296554e-01 -6.37034252e-02 4.20881242e-01 -5.08861780e-01 1.15787089e-01 5.80873787e-01 2.00852454e-01 -9.66172665e-02 -8.91389966e-01 8.28207076e-01 -1.89589739e+00 -6.55737221e-01 -7.79728293e-02 2.17155337e+00 7.69580305e-02 3.78658362e-02 -4.23321217e-01 -3.58074844e-01 8.58236969e-01 4.33986694e-01 -7.36170650e-01 8.93342048e-02 -2.29946434e-01 -5.08755207e-01 8.95107448e-01 3.18516940e-01 -1.07024097e+00 1.04519427e+00 5.06481028e+00 5.99125862e-01 -1.48437786e+00 -1.09944239e-01 2.09358633e-02 1.93146288e-01 -1.64875492e-01 4.02766913e-02 -9.12359059e-01 5.83734989e-01 4.69314247e-01 -1.39869645e-01 2.96307355e-01 1.13650942e+00 2.36793458e-01 -1.39922902e-01 -8.64561558e-01 1.30256033e+00 7.07132742e-02 -1.26075935e+00 -9.69631523e-02 1.16347231e-01 6.03802800e-01 6.50565803e-01 1.61004201e-01 7.35853612e-02 2.11603358e-01 -9.07753527e-01 7.68534303e-01 2.89780617e-01 8.89926910e-01 -9.45679903e-01 7.49278486e-01 5.96773326e-01 -1.58791590e+00 -1.52184039e-01 -8.46077144e-01 -1.09089687e-01 3.82282227e-01 3.24392557e-01 -8.72035205e-01 5.46984494e-01 9.37107205e-01 1.09974194e+00 -4.66808289e-01 1.19884300e+00 -3.38808835e-01 -1.15567364e-01 -4.97069836e-01 -6.54241070e-02 6.08034015e-01 -5.37978828e-01 4.27395195e-01 7.12243378e-01 6.45310462e-01 -1.18825659e-01 4.33424562e-01 9.37679291e-01 -8.28259334e-04 -1.12187438e-01 -1.07554603e+00 2.98831612e-01 8.77807319e-01 1.23771298e+00 -6.26991034e-01 -8.66833031e-02 -6.19767845e-01 6.29021287e-01 3.95692885e-01 3.32875073e-01 -7.87863135e-01 -6.46134377e-01 8.21411908e-01 -2.03929365e-01 3.05784762e-01 -6.88109100e-01 -1.63141534e-01 -1.15485454e+00 1.06396638e-01 -2.94161528e-01 -3.38673703e-02 -1.17057276e+00 -1.17922986e+00 6.71258152e-01 -9.67654511e-02 -1.62750304e+00 -3.41914669e-02 -4.63341355e-01 -4.52189118e-01 1.01821041e+00 -1.95496154e+00 -1.21375060e+00 -9.40662742e-01 5.51254511e-01 5.15979946e-01 -7.37770498e-02 3.47378910e-01 2.58359551e-01 -2.70242155e-01 1.81827247e-01 2.28112847e-01 4.24152426e-02 7.18290985e-01 -9.77459669e-01 5.60876012e-01 8.75506938e-01 1.45668894e-01 7.60714173e-01 5.70949554e-01 -4.85664338e-01 -1.56979918e+00 -1.26997709e+00 7.78777301e-01 -1.84259132e-01 5.73621929e-01 -2.74186611e-01 -9.34308946e-01 7.22973466e-01 1.87812969e-02 2.83876061e-01 -7.66522139e-02 -2.90586412e-01 -1.49132371e-01 -3.75839502e-01 -8.82634163e-01 6.36428893e-01 1.13518941e+00 -5.68936527e-01 -4.73371506e-01 1.40122771e-01 8.14180851e-01 -6.06168211e-01 -4.25841421e-01 5.27056515e-01 4.92273331e-01 -9.42152679e-01 1.15424752e+00 1.76550467e-02 1.15802467e-01 -8.16493988e-01 -2.34227568e-01 -1.34766674e+00 -3.93622130e-01 -5.04153548e-03 4.92982209e-01 9.63843942e-01 8.55791867e-02 -9.21096861e-01 8.22888613e-01 -2.87238299e-03 -2.29437441e-01 -4.32302684e-01 -1.03872108e+00 -7.43742287e-01 -2.83525318e-01 -4.01869923e-01 5.26025355e-01 8.47997487e-01 -3.93263429e-01 5.25109828e-01 -4.81841326e-01 6.68687940e-01 7.43634641e-01 4.51414943e-01 9.37644541e-01 -1.34928370e+00 4.18154597e-01 -1.46804124e-01 -8.06532145e-01 -1.45772815e+00 5.35951853e-01 -7.86327362e-01 3.69844824e-01 -1.44300222e+00 -1.36269301e-01 -6.88192487e-01 -1.63553759e-01 2.92725623e-01 9.53393281e-02 5.41675448e-01 1.55017659e-01 4.46598828e-01 -6.23314321e-01 8.84439588e-01 1.00647259e+00 -2.47262701e-01 -4.05157208e-02 -1.10093899e-01 -4.47540373e-01 7.79860437e-01 6.81011915e-01 -4.72734988e-01 -5.41676879e-01 -8.41044128e-01 2.04269692e-01 1.22153290e-01 4.65262562e-01 -1.20345557e+00 5.69712162e-01 -2.56867737e-01 4.30630535e-01 -1.12433243e+00 4.81336296e-01 -1.12166286e+00 -3.14128660e-02 4.19060856e-01 1.06332272e-01 4.67774868e-01 8.98896754e-02 8.84600282e-01 -5.47211587e-01 -1.84327319e-01 7.49322593e-01 -1.14247419e-01 -1.37546492e+00 6.22025728e-01 -2.29486734e-01 -2.31063485e-01 9.61841226e-01 -4.46535707e-01 -3.90258223e-01 -4.44009334e-01 1.24936417e-01 7.00336218e-01 9.46157992e-01 8.54917526e-01 9.93136048e-01 -1.39637911e+00 -4.38158214e-01 5.02443016e-01 6.40776038e-01 3.54521990e-01 2.23853290e-01 7.98549354e-01 -7.17557311e-01 7.20504522e-01 -4.30408299e-01 -1.23010659e+00 -9.41286147e-01 6.07992768e-01 3.15589577e-01 4.90108192e-01 -6.78728461e-01 5.30915260e-01 2.60277510e-01 -8.04402590e-01 -4.50950637e-02 -3.65785122e-01 -2.92569011e-01 -2.57328182e-01 2.34631866e-01 7.45706409e-02 -1.19511068e-01 -9.98837471e-01 -7.88708389e-01 1.25892889e+00 1.58301473e-01 5.78582697e-02 1.13659108e+00 -6.81845486e-01 -3.27135511e-02 4.95443434e-01 1.38502705e+00 -9.74359885e-02 -1.28922045e+00 -3.77635658e-01 -2.44177785e-02 -8.51439416e-01 1.21074036e-01 -2.34101154e-02 -1.02276385e+00 1.06069541e+00 7.55039692e-01 -1.14124715e-01 7.00668275e-01 -8.71099085e-02 6.61655426e-01 7.12677479e-01 9.83611822e-01 -6.66657627e-01 -1.01028390e-01 7.73386419e-01 7.01012731e-01 -1.69981003e+00 8.64029229e-02 -1.71869263e-01 -5.46896219e-01 1.05425680e+00 7.86607027e-01 -2.89798170e-01 3.89826089e-01 -1.08250022e-01 4.31894958e-01 -1.23065613e-01 -2.83188790e-01 -2.87171960e-01 7.55522326e-02 8.12684476e-01 -7.27166459e-02 -1.34674981e-01 7.49469995e-02 -1.10885039e-01 -2.45485291e-01 -6.18393868e-02 3.93194020e-01 9.41192627e-01 -6.98463976e-01 -6.01870298e-01 -3.55755985e-01 3.07262260e-02 5.74498959e-02 6.40585572e-02 8.18931386e-02 9.16786075e-01 1.63164601e-01 8.77763033e-01 3.05847675e-01 -4.39685285e-01 2.99676955e-01 -4.25708503e-01 1.61701068e-01 -3.40031862e-01 2.56603748e-01 -1.98271632e-01 -4.20174390e-01 -4.66096461e-01 -4.45086628e-01 -4.31983620e-01 -1.32223356e+00 -2.13582054e-01 -4.47851419e-01 3.05320770e-01 1.05957639e+00 8.86246979e-01 5.26096463e-01 1.41983643e-01 7.88026154e-01 -1.24557495e+00 -3.38317007e-01 -5.98425806e-01 -5.41627586e-01 -4.75619687e-03 6.13474786e-01 -1.02176976e+00 -2.65987903e-01 -3.30388933e-01]
[7.6315155029296875, -2.075411319732666]
9a02317b-c59c-46d6-954a-39783976076e
understanding-convolution-for-semantic
1702.08502
null
http://arxiv.org/abs/1702.08502v3
http://arxiv.org/pdf/1702.08502v3.pdf
Understanding Convolution for Semantic Segmentation
Recent advances in deep learning, especially deep convolutional neural networks (CNNs), have led to significant improvement over previous semantic segmentation systems. Here we show how to improve pixel-wise semantic segmentation by manipulating convolution-related operations that are of both theoretical and practical value. First, we design dense upsampling convolution (DUC) to generate pixel-level prediction, which is able to capture and decode more detailed information that is generally missing in bilinear upsampling. Second, we propose a hybrid dilated convolution (HDC) framework in the encoding phase. This framework 1) effectively enlarges the receptive fields (RF) of the network to aggregate global information; 2) alleviates what we call the "gridding issue" caused by the standard dilated convolution operation. We evaluate our approaches thoroughly on the Cityscapes dataset, and achieve a state-of-art result of 80.1% mIOU in the test set at the time of submission. We also have achieved state-of-the-art overall on the KITTI road estimation benchmark and the PASCAL VOC2012 segmentation task. Our source code can be found at https://github.com/TuSimple/TuSimple-DUC .
['Garrison Cottrell', 'Ye Yuan', 'Xiaodi Hou', 'Pengfei Chen', 'Zehua Huang', 'Panqu Wang', 'Ding Liu']
2017-02-27
null
null
null
null
['thermal-image-segmentation']
['computer-vision']
[ 3.80771250e-01 1.82933241e-01 1.31722074e-02 -5.27576387e-01 -6.89106286e-01 -3.54763538e-01 4.53637362e-01 -2.59012640e-01 -6.36886179e-01 6.15840435e-01 4.86335903e-03 -4.73552793e-01 3.97555947e-01 -9.99867857e-01 -9.21349704e-01 -6.66099548e-01 3.65783051e-02 8.94586146e-02 5.74021518e-01 -1.63661584e-01 1.01874061e-01 3.54522198e-01 -1.57498717e+00 4.87152725e-01 1.22821558e+00 1.28411436e+00 3.92603159e-01 6.19558513e-01 -1.15826711e-01 6.33956611e-01 -4.13992912e-01 -4.10175443e-01 3.77108037e-01 -2.08706781e-01 -1.05761206e+00 -8.11161771e-02 5.32194316e-01 -4.04592097e-01 -3.37767452e-01 1.18487692e+00 3.18796515e-01 9.14758220e-02 4.19071466e-01 -8.84282708e-01 -4.95740801e-01 6.70914531e-01 -6.42171323e-01 8.48966166e-02 -4.20916826e-01 7.62809068e-02 8.25385451e-01 -6.74490988e-01 4.25895125e-01 1.15896833e+00 7.93249190e-01 5.44846058e-01 -9.60885704e-01 -5.96741736e-01 1.69688761e-01 2.72427469e-01 -1.37545776e+00 -1.59001797e-01 5.48618913e-01 -2.58884579e-01 7.53723741e-01 2.71699995e-01 6.46680474e-01 8.42285335e-01 -1.10311739e-01 1.09825480e+00 1.08019352e+00 -2.58995593e-01 1.57795489e-01 -1.33030370e-01 6.45485893e-02 6.82217836e-01 1.09665945e-01 6.27321154e-02 4.76079881e-02 5.01765132e-01 8.93648982e-01 -1.44893900e-01 -2.91771352e-01 5.89535534e-02 -9.93358314e-01 8.53935838e-01 9.25879240e-01 1.88727394e-01 -3.79860044e-01 4.77010995e-01 3.07365239e-01 -9.00937542e-02 6.67033851e-01 2.27934569e-01 -5.68323493e-01 5.70198335e-03 -1.01201081e+00 2.49046326e-01 6.11153126e-01 8.30987632e-01 9.48225796e-01 -7.64315650e-02 -2.52446115e-01 1.04750574e+00 1.22176912e-02 2.65626162e-01 3.09568018e-01 -1.09136128e+00 4.84243125e-01 4.51902449e-01 -1.97910845e-01 -5.57769418e-01 -3.60027254e-01 -6.74131691e-01 -9.66108084e-01 2.08163798e-01 4.97207761e-01 -3.03398192e-01 -1.34066272e+00 1.43295085e+00 1.96462855e-01 2.98162848e-01 3.34260240e-02 8.92661691e-01 9.94260550e-01 6.46741629e-01 1.60136893e-01 4.38399643e-01 1.40673256e+00 -1.45579636e+00 -4.34009939e-01 -4.10285562e-01 7.24235058e-01 -6.73899651e-01 9.71708059e-01 2.19372839e-01 -9.37543988e-01 -7.56514251e-01 -1.10293317e+00 -3.87852520e-01 -6.15275800e-01 3.76262665e-01 7.84972012e-01 6.89368725e-01 -1.11999941e+00 8.31577122e-01 -9.51872826e-01 -1.17907584e-01 9.58313048e-01 2.51604170e-01 -3.07305083e-02 -8.62860829e-02 -1.22534680e+00 5.33018351e-01 5.86486280e-01 3.19508880e-01 -6.19624794e-01 -8.61884892e-01 -7.68226802e-01 7.60401934e-02 3.96587938e-01 -4.62567806e-01 1.36392820e+00 -1.13001740e+00 -1.36702716e+00 7.00983465e-01 -2.46730968e-02 -7.34303594e-01 6.36366665e-01 -3.27895582e-01 -1.42051607e-01 1.88282549e-01 1.91299096e-02 1.48164940e+00 4.58362341e-01 -1.17211008e+00 -8.13024104e-01 -1.45622283e-01 -1.78009563e-03 1.20580636e-01 -1.12258226e-01 -2.89441228e-01 -7.44407058e-01 -8.97524595e-01 1.60420179e-01 -9.36851740e-01 -4.82874334e-01 9.26556182e-04 -5.00127733e-01 -1.44315232e-02 9.47503984e-01 -8.23741615e-01 1.04382420e+00 -2.00066519e+00 -2.56059289e-01 4.04081345e-02 8.61191750e-03 6.89938128e-01 -1.56346262e-01 -3.46342251e-02 -5.03197983e-02 2.02629685e-01 -7.07047105e-01 -5.99904180e-01 -1.27128229e-01 2.34966084e-01 -2.03200892e-01 1.18827052e-01 3.78007114e-01 1.26869404e+00 -6.47128224e-01 -3.68442297e-01 3.60711902e-01 6.06512487e-01 -6.40259922e-01 -1.00576423e-01 -2.33780861e-01 4.43478405e-01 -3.35943639e-01 6.48602605e-01 1.04427421e+00 -2.03712493e-01 -2.03617498e-01 -2.69590408e-01 -3.00223291e-01 3.52267116e-01 -9.57253993e-01 1.67943811e+00 -3.84409487e-01 7.64471948e-01 -8.42247233e-02 -1.08950222e+00 8.47916484e-01 -7.81280473e-02 1.76137239e-01 -8.59175384e-01 2.93845266e-01 2.66027421e-01 -1.39171302e-01 -1.57789737e-01 5.38885534e-01 3.86040211e-01 6.68310598e-02 -1.46199867e-01 3.19565497e-02 -1.14953354e-01 1.46603629e-01 -7.92161897e-02 8.38924766e-01 3.13335925e-01 -6.51703253e-02 -5.38216531e-01 6.04899049e-01 1.56357035e-01 5.10224879e-01 6.36942327e-01 -1.83899835e-01 9.00000930e-01 6.46846890e-01 -4.51502264e-01 -1.01043487e+00 -9.02441680e-01 -2.50560969e-01 9.65210855e-01 1.84084997e-01 -2.36959234e-02 -1.28280544e+00 -6.50710583e-01 -2.61715561e-01 6.38826191e-01 -6.63644135e-01 1.26966730e-01 -6.58348441e-01 -8.97464991e-01 8.28055143e-01 9.52337027e-01 1.31197643e+00 -9.93239641e-01 -6.13174677e-01 2.29660690e-01 -3.23756993e-01 -1.54271090e+00 -3.81021053e-01 2.25879580e-01 -9.59808230e-01 -7.79572725e-01 -9.45063233e-01 -9.11793590e-01 6.88824236e-01 1.54462487e-01 9.49367940e-01 -4.29244787e-02 -3.24796587e-01 -3.54736954e-01 -4.24202651e-01 -3.25515062e-01 -1.38525322e-01 3.66636634e-01 -7.15607762e-01 -1.46870315e-01 2.68172801e-01 -2.79411048e-01 -8.53470862e-01 3.18064898e-01 -8.07210088e-01 5.04126310e-01 7.40532279e-01 6.89233601e-01 7.02749372e-01 7.85596482e-03 3.52959603e-01 -9.65392172e-01 1.30697384e-01 -1.52511403e-01 -7.47520685e-01 -3.35950479e-02 -3.33712012e-01 2.49262433e-02 4.84060317e-01 -1.39228895e-01 -1.20943761e+00 2.18061462e-01 -7.13974774e-01 -8.52507427e-02 -4.97568756e-01 1.84605092e-01 -9.42112356e-02 -1.59983695e-01 4.26153660e-01 3.18806805e-02 -1.00357644e-01 -6.33791506e-01 4.40864861e-01 8.21527839e-01 8.01906765e-01 -5.19448340e-01 3.61432493e-01 6.25213146e-01 -1.22754142e-01 -8.39827776e-01 -9.15762663e-01 -5.27421296e-01 -6.20149076e-01 -8.70275870e-03 1.20108330e+00 -1.10433638e+00 -4.69690382e-01 9.58985150e-01 -1.09038877e+00 -8.96199703e-01 -1.89779252e-01 2.99041271e-01 -3.67059648e-01 1.13526829e-01 -7.46097147e-01 -3.61182868e-01 -3.94844621e-01 -1.41711736e+00 1.09731400e+00 5.42954743e-01 1.93027720e-01 -8.96437883e-01 -3.97512466e-01 4.42373961e-01 5.67402422e-01 3.47374082e-01 5.55472732e-01 -3.18802625e-01 -7.54873812e-01 1.39666140e-01 -8.16733181e-01 7.48628378e-01 -2.14059427e-01 -1.43544182e-01 -1.28741515e+00 5.42716235e-02 -4.59025115e-01 -5.66464551e-02 1.44893861e+00 5.68235576e-01 1.75794911e+00 -4.03909050e-02 -2.53365904e-01 9.64771211e-01 1.50924933e+00 1.23477198e-01 1.13624239e+00 3.93802077e-01 9.09682333e-01 6.09113872e-01 4.56916362e-01 1.51926115e-01 5.26364148e-01 5.96863508e-01 5.27048945e-01 -5.60268462e-01 -6.15489542e-01 -2.34842405e-01 -7.78876469e-02 3.63902748e-01 -2.65807807e-01 -2.10637182e-01 -8.85150373e-01 6.83711708e-01 -1.82632744e+00 -5.80684364e-01 -4.47545946e-01 1.84050453e+00 7.21221328e-01 1.87706143e-01 -8.25033188e-02 4.59613279e-02 6.38297796e-01 1.37340188e-01 -3.62489730e-01 -4.93188024e-01 -7.38476589e-02 6.90731764e-01 1.17924571e+00 5.82867086e-01 -1.56317639e+00 1.41532302e+00 5.39408684e+00 1.16453826e+00 -1.25921154e+00 1.58460930e-01 1.13486660e+00 3.40878725e-01 -5.22859879e-02 -2.00320512e-01 -8.95617366e-01 4.33573127e-01 8.07805955e-01 5.19157350e-01 2.59679228e-01 8.23357761e-01 1.45409375e-01 -3.15097779e-01 -5.67003131e-01 6.81328058e-01 -3.44766706e-01 -1.52816868e+00 -7.96266645e-02 1.23228431e-01 1.09617090e+00 3.86805177e-01 1.73208177e-01 3.86499137e-01 1.77599102e-01 -1.08655357e+00 7.39657938e-01 2.11385876e-01 9.06094432e-01 -7.78949738e-01 9.76008773e-01 9.98767316e-02 -1.28940189e+00 -4.75396626e-02 -4.00202215e-01 4.29039933e-02 1.34762824e-01 7.27167070e-01 -6.77729607e-01 4.52086091e-01 9.84838128e-01 6.08302832e-01 -6.42131686e-01 1.10611296e+00 -4.87986743e-01 6.88056946e-01 -3.35544229e-01 1.45537958e-01 7.56895363e-01 -1.66283965e-01 9.02511775e-02 1.36551654e+00 1.95177987e-01 7.29551911e-02 -8.00050646e-02 9.98613834e-01 -2.81549543e-01 -7.61260763e-02 -9.83171165e-02 2.63525516e-01 2.21831560e-01 1.19469309e+00 -1.07948220e+00 -4.98138487e-01 -2.65105397e-01 1.13227522e+00 1.98354468e-01 4.58982825e-01 -1.02892447e+00 -6.16848528e-01 8.07049036e-01 -1.00225816e-02 7.13558376e-01 -1.72110200e-01 -7.08392024e-01 -8.21358621e-01 2.95862905e-03 -4.88941073e-01 3.07525620e-02 -5.28694153e-01 -6.33967042e-01 5.86681902e-01 -3.55459563e-02 -8.21645796e-01 1.69491857e-01 -7.41445124e-01 -6.49898291e-01 8.91370714e-01 -1.94153905e+00 -1.16409779e+00 -5.66713929e-01 3.01938891e-01 7.03431487e-01 2.76684999e-01 4.49970603e-01 6.36681437e-01 -5.72068214e-01 6.06118619e-01 3.21082138e-02 4.09815043e-01 3.67338151e-01 -1.26068401e+00 9.17777598e-01 9.08915937e-01 -2.04058573e-01 1.37754157e-01 3.71324301e-01 -4.13279116e-01 -6.13012254e-01 -1.55482936e+00 7.43974566e-01 3.15870009e-02 4.28489894e-01 -3.85166466e-01 -9.55179572e-01 5.22577822e-01 1.06678875e-02 1.39650971e-01 1.60475224e-01 -2.62685329e-01 -2.82310605e-01 -7.23231509e-02 -1.21444023e+00 5.64991176e-01 1.02012885e+00 -1.59511626e-01 -2.24850059e-01 7.62995631e-02 1.00815928e+00 -6.21537507e-01 -6.42792523e-01 6.94213808e-01 5.14881313e-01 -1.05636275e+00 1.05821848e+00 -4.58065905e-02 6.87823117e-01 -3.16051990e-01 -9.33555216e-02 -1.14650536e+00 -1.50096238e-01 -2.75514871e-01 2.37091631e-01 9.34466779e-01 5.32430947e-01 -8.17059636e-01 8.80277038e-01 1.80164620e-01 -6.04507446e-01 -1.01176953e+00 -6.89122796e-01 -7.15472162e-01 2.80771106e-01 -5.33118367e-01 7.18588173e-01 6.06357396e-01 -6.43127084e-01 -1.68088481e-01 3.04695740e-02 2.16830760e-01 5.96504152e-01 -1.73093557e-01 4.14777517e-01 -1.00185335e+00 -1.08506456e-01 -7.24915087e-01 -2.32753322e-01 -1.46303618e+00 -4.74995468e-03 -6.96268439e-01 2.23399714e-01 -1.73372233e+00 -1.36084691e-01 -5.61387837e-01 -1.26302436e-01 6.94306552e-01 -1.14758648e-01 7.46618152e-01 1.46850199e-01 -4.09587324e-02 -4.14487779e-01 4.00786042e-01 1.54430485e+00 -6.50007054e-02 -2.04104394e-01 -1.44109770e-03 -7.06638753e-01 8.18409741e-01 1.13345182e+00 -1.54061958e-01 -1.96227744e-01 -5.75280130e-01 -3.07421118e-01 -3.98718685e-01 7.08603799e-01 -1.21884584e+00 -2.27230787e-03 1.70479909e-01 3.40758592e-01 -7.56827831e-01 1.88910052e-01 -4.89559650e-01 -1.49358720e-01 5.80830038e-01 -2.45504305e-01 -3.92676502e-01 4.86589223e-01 1.76546380e-01 -3.16520065e-01 -2.26358607e-01 1.04973626e+00 -9.00577828e-02 -1.18179965e+00 3.15500975e-01 -7.37159923e-02 4.93983924e-02 8.11876357e-01 -2.19441369e-01 -5.00997007e-01 -2.03406624e-02 -5.18928230e-01 3.11629325e-01 2.92164862e-01 3.75598043e-01 3.49150538e-01 -1.03559816e+00 -5.29370129e-01 2.16846153e-01 -7.19076768e-02 4.68173414e-01 4.04024750e-01 8.97528827e-01 -1.07023299e+00 4.68191713e-01 -1.83887601e-01 -6.02360070e-01 -9.77328181e-01 5.28633259e-02 4.16219682e-01 -1.45814613e-01 -7.23278463e-01 1.15843642e+00 4.12678212e-01 -3.63463134e-01 1.44923836e-01 -6.81382000e-01 -1.99693531e-01 -3.61438803e-02 4.92343605e-01 4.02910084e-01 2.28039414e-01 -5.48437953e-01 -2.26945192e-01 5.65236807e-01 -1.20666161e-01 7.74099156e-02 1.21879864e+00 -1.78292617e-02 2.27692863e-03 -1.32102698e-01 1.43246925e+00 -5.73788881e-01 -1.67983413e+00 -9.56073999e-02 -2.77499050e-01 -3.62989902e-01 4.34920490e-01 -9.12599385e-01 -1.38105655e+00 1.12335527e+00 6.79482818e-01 2.90778317e-02 1.22588456e+00 -1.46702528e-02 1.24183834e+00 1.73590988e-01 5.92261590e-02 -1.16318583e+00 -3.08981210e-01 6.61464989e-01 6.00288689e-01 -1.21464431e+00 -2.46734515e-01 -7.91817725e-01 -6.66373312e-01 1.07140386e+00 5.78191996e-01 -2.55944282e-01 6.46499634e-01 3.13097715e-01 1.11662946e-03 -1.30742835e-02 -2.24674761e-01 -7.30326176e-01 2.50521600e-01 4.72514093e-01 3.38941157e-01 4.10901397e-01 -3.12086374e-01 4.32500213e-01 -2.62064278e-01 1.63382869e-02 3.57921213e-01 6.72746956e-01 -6.37762845e-01 -9.37558591e-01 -1.81226641e-01 2.37318158e-01 -4.33315933e-01 -3.08960110e-01 4.22576591e-02 7.68871605e-01 4.83341038e-01 7.04111397e-01 4.07608598e-01 -1.86487049e-01 2.19777539e-01 -2.21596599e-01 3.57616067e-01 -3.19586545e-01 -3.27773243e-01 1.45411506e-01 1.01191685e-01 -7.56412089e-01 -3.36446792e-01 -4.59578544e-01 -1.49284637e+00 -2.93037027e-01 -9.09395069e-02 -1.98576003e-01 8.81906629e-01 9.01573479e-01 2.47186631e-01 8.96059215e-01 2.50934362e-01 -1.00568759e+00 -2.06248239e-01 -9.23286080e-01 -3.28229874e-01 1.99644968e-01 2.35280961e-01 -4.61686730e-01 -1.59316987e-01 1.51647598e-01]
[9.489218711853027, 0.010646681301295757]
2e56439e-3810-4bc0-b020-0052367e1441
efficient-domain-generalization-via-common
2003.12815
null
https://arxiv.org/abs/2003.12815v2
https://arxiv.org/pdf/2003.12815v2.pdf
Efficient Domain Generalization via Common-Specific Low-Rank Decomposition
Domain generalization refers to the task of training a model which generalizes to new domains that are not seen during training. We present CSD (Common Specific Decomposition), for this setting,which jointly learns a common component (which generalizes to new domains) and a domain specific component (which overfits on training domains). The domain specific components are discarded after training and only the common component is retained. The algorithm is extremely simple and involves only modifying the final linear classification layer of any given neural network architecture. We present a principled analysis to understand existing approaches, provide identifiability results of CSD,and study effect of low-rank on domain generalization. We show that CSD either matches or beats state of the art approaches for domain generalization based on domain erasure, domain perturbed data augmentation, and meta-learning. Further diagnostics on rotated MNIST, where domains are interpretable, confirm the hypothesis that CSD successfully disentangles common and domain specific components and hence leads to better domain generalization.
['Sunita Sarawagi', 'Praneeth Netrapalli', 'Vihari Piratla']
2020-03-28
null
https://proceedings.icml.cc/static/paper_files/icml/2020/4649-Paper.pdf
https://proceedings.icml.cc/static/paper_files/icml/2020/4649-Paper.pdf
icml-2020-1
['rotated-mnist']
['computer-vision']
[ 7.01192677e-01 2.34920606e-01 -4.07535434e-01 -4.22766060e-01 -4.03282911e-01 -1.00322473e+00 8.09556007e-01 1.97357405e-02 -3.59784245e-01 1.03716028e+00 3.20837379e-01 -4.69880790e-01 -5.48386097e-01 -5.46349227e-01 -8.63923132e-01 -6.17016852e-01 -3.09729874e-01 7.01933205e-01 -1.06032610e-01 -3.44367266e-01 -1.02309883e-01 5.62392592e-01 -1.33178115e+00 4.50435817e-01 9.48465526e-01 7.50511169e-01 6.59223720e-02 2.41917580e-01 1.73446402e-01 2.87843943e-01 -5.37663281e-01 1.99977309e-02 5.28545082e-01 -2.23947272e-01 -1.11537790e+00 1.37274295e-01 5.54471970e-01 -3.29028875e-01 -5.29555857e-01 9.37376142e-01 1.90137386e-01 5.35706699e-01 9.56496894e-01 -1.24896610e+00 -1.24100208e+00 5.51348567e-01 -1.76443636e-01 3.33092541e-01 2.75976121e-01 -1.52074650e-01 9.45377588e-01 -6.86547458e-01 9.79838967e-01 1.25004530e+00 5.86912036e-01 9.55956042e-01 -1.84150922e+00 -9.15663183e-01 3.88773710e-01 -1.62249520e-01 -9.99148309e-01 -3.57665509e-01 8.36491466e-01 -6.97241604e-01 8.98065090e-01 -1.19907655e-01 9.79872495e-02 1.81425679e+00 -8.39771405e-02 6.93739116e-01 1.32974911e+00 -3.40313435e-01 6.04488194e-01 3.25427681e-01 3.84996295e-01 7.10888952e-02 6.41623318e-01 6.64717138e-01 -5.51102698e-01 -1.60996854e-01 7.21906364e-01 -1.06319271e-01 -2.77006894e-01 -8.76210213e-01 -1.20135212e+00 7.12654948e-01 4.13837433e-01 1.21162415e-01 -4.12569791e-01 -3.97909522e-01 4.72188115e-01 8.09055448e-01 5.95307827e-01 8.28882873e-01 -9.37645912e-01 2.70865291e-01 -6.76511884e-01 5.29025555e-01 6.39596283e-01 1.14931488e+00 7.61919200e-01 2.58391380e-01 7.69710615e-02 9.15529251e-01 -2.17075795e-01 4.18210864e-01 7.40935981e-01 -1.04429626e+00 4.90274012e-01 6.97807252e-01 -9.20563862e-02 -4.81122434e-01 -5.22680402e-01 -8.95964146e-01 -1.00414789e+00 2.45702922e-01 2.89195061e-01 -3.54118496e-01 -1.09328544e+00 2.27762413e+00 -7.01624304e-02 3.53344381e-01 6.14978075e-01 5.39902329e-01 6.59576714e-01 2.07169011e-01 1.67521462e-01 9.33632702e-02 9.22780037e-01 -3.70686740e-01 -2.61695743e-01 -7.04218745e-01 8.05060506e-01 -4.66916785e-02 1.04869115e+00 4.75876808e-01 -8.15097272e-01 -6.88445091e-01 -1.34341145e+00 1.37353376e-01 -7.43684947e-01 -3.17897908e-02 6.44293904e-01 7.58246064e-01 -9.48294163e-01 8.89277995e-01 -4.96175736e-01 -4.84421104e-01 7.15727508e-01 5.92153668e-01 -8.63030314e-01 -3.48271579e-01 -1.50223279e+00 8.52588236e-01 9.45305526e-01 -6.93404794e-01 -9.52729881e-01 -1.08385110e+00 -1.02099204e+00 -7.97190070e-02 2.55742788e-01 -1.03299737e+00 1.13948798e+00 -9.01178479e-01 -1.17309713e+00 8.91198754e-01 -1.07496448e-01 -7.62255371e-01 1.76835641e-01 -2.59160489e-01 -6.05492651e-01 -7.00805336e-02 2.38956898e-01 6.71941936e-01 8.82216394e-01 -1.39500797e+00 -5.07479787e-01 -4.80107158e-01 2.53231041e-02 1.46432593e-01 -4.17062640e-01 -7.41906822e-01 2.79876202e-01 -8.21260631e-01 3.75515252e-01 -1.07387173e+00 -7.93902352e-02 -5.20301223e-01 -2.19729602e-01 -2.98401266e-02 7.54923403e-01 -7.02028692e-01 9.72617984e-01 -2.39022613e+00 5.96521080e-01 3.28039885e-01 2.97084242e-01 2.44407102e-01 -3.51579487e-01 2.66703993e-01 -8.13558340e-01 1.84786826e-01 -5.04809320e-01 -2.28571534e-01 1.51735157e-01 4.97265786e-01 -7.61719882e-01 3.51947635e-01 4.22449350e-01 5.20437062e-01 -8.57772768e-01 3.42968196e-01 -4.11109962e-02 1.05435684e-01 -8.82172525e-01 -1.85735106e-01 -3.13636631e-01 6.52763963e-01 -1.88182309e-01 2.61974752e-01 1.08045208e+00 -3.01836401e-01 4.22318429e-01 -2.24482566e-02 3.08873266e-01 5.09460866e-01 -1.33839619e+00 1.83395243e+00 -3.66601288e-01 5.12833476e-01 -1.55549720e-01 -1.49148309e+00 9.83233869e-01 1.36922538e-01 2.21561760e-01 -6.67560041e-01 -2.30823651e-01 3.58177245e-01 -1.14808790e-02 -2.08260641e-01 2.82402933e-01 -4.75995183e-01 -1.88416719e-01 4.40942258e-01 6.46748006e-01 2.18023658e-01 -4.78551276e-02 2.93820202e-01 1.11721408e+00 2.26949677e-01 4.05118763e-01 -3.97882104e-01 4.68358934e-01 5.93885966e-02 5.48170984e-01 9.40416813e-01 -2.88518593e-02 3.47656041e-01 5.28397322e-01 -1.31347105e-01 -1.15749264e+00 -1.35126793e+00 -2.34139085e-01 1.32243896e+00 4.39953208e-02 3.18106338e-02 -3.89448106e-01 -8.25569510e-01 4.49353069e-01 1.03825927e+00 -8.24527204e-01 -8.23738754e-01 -3.71259511e-01 -6.77376509e-01 6.39016330e-01 7.94221878e-01 5.95867217e-01 -6.95987165e-01 8.63634497e-02 3.30029726e-02 7.21109435e-02 -1.13536096e+00 6.46927059e-02 6.23690724e-01 -1.36223352e+00 -8.51387858e-01 -7.68736601e-01 -8.28633964e-01 5.55052459e-01 1.97772235e-01 1.01397693e+00 -5.32887459e-01 2.49684080e-01 4.76181775e-01 -2.84618199e-01 -1.61477104e-01 -6.01947665e-01 3.19240540e-01 6.56648278e-01 -1.82801157e-01 5.60601175e-01 -1.08107853e+00 -4.89641912e-02 1.27169937e-01 -1.05036116e+00 -3.96066159e-01 7.76805639e-01 1.17983055e+00 3.37270707e-01 1.93023130e-01 1.05117071e+00 -1.20492697e+00 8.93831551e-01 -7.87869036e-01 -3.45553190e-01 4.60968278e-02 -6.81762278e-01 5.02200365e-01 6.62019610e-01 -6.28555179e-01 -1.19368041e+00 -3.34534869e-02 2.40940645e-01 -4.00272042e-01 -6.25449598e-01 5.36099911e-01 -1.95807427e-01 6.49625733e-02 1.30500543e+00 4.51831311e-01 1.99619401e-02 -8.42878461e-01 5.22760808e-01 4.22988743e-01 6.61710143e-01 -8.12049329e-01 1.00439537e+00 5.45785785e-01 3.12167462e-02 -8.90862644e-01 -7.84116566e-01 -3.86806965e-01 -9.99871552e-01 5.51121891e-01 4.23666507e-01 -1.19121838e+00 -2.57438064e-01 1.97091162e-01 -9.27938223e-01 -3.50981027e-01 -5.53556144e-01 5.84409058e-01 -7.11716712e-01 2.80785859e-01 -1.44863650e-01 -3.71634990e-01 1.82001933e-01 -5.86241186e-01 8.13749611e-01 -1.02663741e-01 -3.73439372e-01 -1.29350054e+00 1.44759834e-01 9.77025554e-02 1.76879406e-01 1.68859556e-01 1.31898999e+00 -1.42470503e+00 -1.83684677e-01 2.80376207e-02 -1.59955442e-01 8.26140046e-01 1.41127273e-01 -8.93721700e-01 -1.09220481e+00 -5.74466467e-01 -7.66438618e-02 -3.44765931e-01 1.22000527e+00 3.57418031e-01 9.95170653e-01 -3.40932310e-01 -5.23288369e-01 8.48394871e-01 1.16214192e+00 2.15773910e-01 4.11318630e-01 6.43941104e-01 2.58758396e-01 5.50265133e-01 3.18091393e-01 2.44253665e-01 9.93734039e-03 2.57418156e-01 1.41976327e-01 -6.08440414e-02 -2.68039256e-01 -2.12832555e-01 3.38516444e-01 5.01964614e-02 2.06015464e-02 -2.84650251e-02 -9.14491177e-01 6.51407182e-01 -1.48371041e+00 -9.16944206e-01 3.41920048e-01 2.14014554e+00 7.94533908e-01 2.66739100e-01 2.05726072e-01 1.93704069e-01 5.04868269e-01 -8.55989382e-02 -1.02911127e+00 -2.51556605e-01 -4.91634339e-01 6.35120571e-01 4.98653740e-01 4.37646538e-01 -1.16813612e+00 9.38639820e-01 6.95052671e+00 4.63201612e-01 -7.44567335e-01 -1.00398928e-01 1.23519853e-01 1.81585327e-02 -3.83897036e-01 -7.12600425e-02 -6.89411283e-01 1.01553299e-01 6.63530767e-01 -3.16511184e-01 6.44853175e-01 1.11851859e+00 -4.69600260e-01 3.34897280e-01 -1.77469480e+00 6.87172353e-01 -1.66869611e-01 -1.40933001e+00 3.89746130e-01 1.40167654e-01 1.05892956e+00 4.71957400e-02 6.51575625e-01 7.52843022e-01 6.53530538e-01 -9.62651253e-01 2.11150333e-01 2.89576262e-01 8.96595955e-01 -5.18937290e-01 5.29775679e-01 4.67873126e-01 -4.93767977e-01 -5.46805680e-01 -5.47199845e-01 -1.55312672e-01 -4.20663118e-01 2.97077119e-01 -1.21350443e+00 7.20318496e-01 5.44238806e-01 1.06299949e+00 -6.07817173e-01 6.43363535e-01 -7.70592391e-02 4.99164641e-01 -1.31381288e-01 6.28217518e-01 8.29304531e-02 2.55320352e-02 9.12005842e-01 1.14676750e+00 3.63529325e-01 6.60347492e-02 5.84613197e-02 1.05357933e+00 -2.14293748e-01 -4.56424415e-01 -9.41726387e-01 -1.34059370e-01 6.59205019e-01 4.85366553e-01 -1.33665964e-01 -4.12615478e-01 -3.01244766e-01 1.10453796e+00 1.55049235e-01 9.66040134e-01 -2.61653721e-01 -3.70938778e-01 1.17325139e+00 -1.52508514e-02 3.87076378e-01 -2.90882617e-01 -5.73928535e-01 -1.26319623e+00 -1.17428318e-01 -1.04298532e+00 6.73009455e-01 -4.50160235e-01 -1.59183037e+00 4.12592888e-01 2.23502368e-01 -1.38014174e+00 -4.28547174e-01 -9.16203022e-01 -1.84634611e-01 1.13456106e+00 -1.58121574e+00 -9.66705382e-01 -9.17753503e-02 7.53568649e-01 2.68407762e-01 -7.44492888e-01 1.05137277e+00 2.22163931e-01 -3.49493593e-01 8.57393265e-01 5.87383926e-01 4.09129709e-02 8.85284007e-01 -1.26436222e+00 4.82425630e-01 7.09051609e-01 -2.39089817e-01 1.15904260e+00 7.15140998e-01 -8.16640198e-01 -8.51815104e-01 -1.17543721e+00 6.96063578e-01 -6.60060167e-01 6.13078356e-01 -4.11697358e-01 -1.21096420e+00 1.08562326e+00 -2.65165359e-01 -3.79932404e-01 8.53578448e-01 5.81965208e-01 -8.08458030e-01 -5.58676757e-02 -1.23904836e+00 4.89754051e-01 1.39426351e+00 -6.96189582e-01 -1.08076620e+00 1.54813513e-01 1.00646758e+00 -1.51161745e-01 -7.96822786e-01 3.93625200e-01 3.72656912e-01 -7.65708208e-01 1.39200091e+00 -1.53137863e+00 3.94782990e-01 -2.90441792e-02 -5.71005344e-01 -1.76311052e+00 -6.58563733e-01 -4.19082016e-01 -2.40814224e-01 9.88511026e-01 4.70018774e-01 -9.51997399e-01 7.59953856e-01 4.40567464e-01 -1.83884814e-01 -1.70910284e-01 -9.16113257e-01 -1.53553760e+00 7.82108784e-01 -2.24559933e-01 7.69125462e-01 1.37164128e+00 4.17727716e-02 5.12613952e-01 -2.55557030e-01 4.04434144e-01 6.51511729e-01 -3.68234590e-02 5.33414185e-01 -1.76786995e+00 -3.19540203e-01 -2.75484741e-01 -3.41989249e-01 -1.24442267e+00 5.21666884e-01 -1.33023071e+00 -6.12287998e-01 -1.21979940e+00 -5.27051091e-02 -4.29178029e-01 -3.85553390e-01 5.28487325e-01 1.31575838e-01 -2.57243484e-01 -9.43586081e-02 1.90873966e-01 -2.24172726e-01 5.21241367e-01 1.16825008e+00 -2.21250325e-01 -5.76815128e-01 1.58014432e-01 -1.35566592e+00 6.81918919e-01 8.58793557e-01 -4.47664887e-01 -7.50068009e-01 -2.90842712e-01 1.35774359e-01 -1.87135905e-01 6.50620639e-01 -1.08438909e+00 5.30441627e-02 -1.55295506e-01 7.73627102e-01 -1.96456552e-01 4.21700567e-01 -8.45617115e-01 -1.93290412e-01 4.06568885e-01 -7.06718147e-01 -4.13477033e-01 5.57375073e-01 8.93990993e-01 -4.88475943e-03 -1.15372203e-01 7.31751800e-01 -6.41777962e-02 -1.07232463e+00 1.92025810e-01 -1.85559615e-01 2.23265335e-01 7.92746782e-01 -3.63218516e-01 -4.81397152e-01 -2.83236265e-01 -1.31520832e+00 9.10754353e-02 2.95719802e-01 5.13171732e-01 4.94368702e-01 -1.48591506e+00 -7.93013096e-01 6.26624107e-01 1.73434585e-01 -3.47042605e-02 3.17461848e-01 2.98761696e-01 3.32573950e-01 5.72632492e-01 -3.84042948e-01 -5.35459816e-01 -7.52072513e-01 7.26816595e-01 3.45055699e-01 -6.28087744e-02 -5.39585173e-01 1.06672478e+00 5.49860656e-01 -9.45584953e-01 1.29683703e-01 -3.56011003e-01 -2.14791209e-01 1.48150682e-01 2.62733161e-01 4.35228378e-01 6.01922832e-02 -3.01928818e-01 -4.44261909e-01 3.47597808e-01 -3.71839583e-01 -1.72349781e-01 1.48794985e+00 -3.29713598e-02 2.90540189e-01 3.70542914e-01 1.30474246e+00 -4.95894104e-01 -1.26068914e+00 -7.85566747e-01 1.25287265e-01 -3.39323014e-01 -2.36180052e-01 -1.27560365e+00 -4.06115085e-01 9.26367104e-01 5.85263968e-01 7.55728856e-02 1.28605843e+00 2.02119812e-01 4.20209050e-01 6.86905324e-01 1.53813228e-01 -1.13007009e+00 3.00180316e-01 8.51269424e-01 1.02860582e+00 -1.13708854e+00 2.04055607e-02 -1.42346397e-01 -7.82374263e-01 8.58860373e-01 5.53910434e-01 -3.92118961e-01 6.69327676e-01 -2.52065152e-01 -4.08717096e-01 -6.74769841e-03 -6.62891209e-01 -8.88816118e-02 5.15514553e-01 1.31042719e+00 -5.58676124e-02 -4.19342108e-02 2.24000841e-01 9.58364964e-01 -3.76124620e-01 -3.93304676e-02 4.53301370e-01 7.14324951e-01 -4.08097833e-01 -1.21574712e+00 -2.46993378e-01 4.23287064e-01 -2.29363050e-02 -1.04953565e-01 -6.01559401e-01 1.06014979e+00 3.03272665e-01 6.25114202e-01 -1.18757434e-01 -6.43860996e-01 5.67904234e-01 6.72562003e-01 4.74581152e-01 -9.32396472e-01 -3.54653643e-03 -3.95960122e-01 2.05958560e-02 -3.33412468e-01 -2.21207976e-01 -8.26295733e-01 -1.00743020e+00 -2.03097746e-01 1.14756301e-02 -1.78025186e-01 3.47576559e-01 8.88838470e-01 6.84688807e-01 7.16169059e-01 4.21984434e-01 -5.64597547e-01 -1.03852367e+00 -8.21269333e-01 -8.11593235e-01 6.95294380e-01 9.03036892e-01 -1.00472844e+00 -5.14455676e-01 2.52182275e-01]
[10.283004760742188, 3.016594886779785]
b767a98c-bb52-45b8-9352-084d67e31eb5
selfd-self-learning-large-scale-driving
2204.10320
null
https://arxiv.org/abs/2204.10320v1
https://arxiv.org/pdf/2204.10320v1.pdf
SelfD: Self-Learning Large-Scale Driving Policies From the Web
Effectively utilizing the vast amounts of ego-centric navigation data that is freely available on the internet can advance generalized intelligent systems, i.e., to robustly scale across perspectives, platforms, environmental conditions, scenarios, and geographical locations. However, it is difficult to directly leverage such large amounts of unlabeled and highly diverse data for complex 3D reasoning and planning tasks. Consequently, researchers have primarily focused on its use for various auxiliary pixel- and image-level computer vision tasks that do not consider an ultimate navigational objective. In this work, we introduce SelfD, a framework for learning scalable driving by utilizing large amounts of online monocular images. Our key idea is to leverage iterative semi-supervised training when learning imitative agents from unlabeled data. To handle unconstrained viewpoints, scenes, and camera parameters, we train an image-based model that directly learns to plan in the Bird's Eye View (BEV) space. Next, we use unlabeled data to augment the decision-making knowledge and robustness of an initially trained model via self-training. In particular, we propose a pseudo-labeling step which enables making full use of highly diverse demonstration data through "hypothetical" planning-based data augmentation. We employ a large dataset of publicly available YouTube videos to train SelfD and comprehensively analyze its generalization benefits across challenging navigation scenarios. Without requiring any additional data collection or annotation efforts, SelfD demonstrates consistent improvements (by up to 24%) in driving performance evaluation on nuScenes, Argoverse, Waymo, and CARLA.
['Eshed Ohn-Bar', 'Ruizhao Zhu', 'Jimuyang Zhang']
2022-04-21
null
http://openaccess.thecvf.com//content/CVPR2022/html/Zhang_SelfD_Self-Learning_Large-Scale_Driving_Policies_From_the_Web_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Zhang_SelfD_Self-Learning_Large-Scale_Driving_Policies_From_the_Web_CVPR_2022_paper.pdf
cvpr-2022-1
['self-learning']
['natural-language-processing']
[-1.07769646e-01 -2.68424465e-03 -2.20820993e-01 -5.04845202e-01 -6.02213144e-01 -8.56564045e-01 5.78077018e-01 -3.92672420e-01 -6.03587151e-01 6.62648916e-01 1.31603286e-01 -5.56394994e-01 2.81196386e-01 -6.34070098e-01 -9.68993187e-01 -3.23866695e-01 1.72456399e-01 2.76580989e-01 3.08966756e-01 -4.10312951e-01 1.35302335e-01 4.24422324e-01 -1.76075959e+00 -2.82159746e-01 1.02134669e+00 7.68049061e-01 3.08788985e-01 4.95207846e-01 4.73730922e-01 7.27433264e-01 -5.36262058e-02 -3.28320302e-02 7.25870013e-01 1.22137971e-01 -4.66684103e-01 2.53637016e-01 6.98147058e-01 -8.72793257e-01 -8.31385672e-01 9.47428286e-01 2.62440205e-01 4.48765129e-01 3.46646249e-01 -1.49570096e+00 -4.11433011e-01 7.26529360e-02 -2.75004894e-01 2.23454356e-01 3.46798092e-01 9.82303023e-01 6.56159222e-01 -6.76488936e-01 9.49495196e-01 1.01795864e+00 4.29136217e-01 4.92395550e-01 -8.54627550e-01 -7.06164002e-01 4.74298149e-01 3.68671954e-01 -1.12052679e+00 -5.40898860e-01 6.14542961e-01 -5.50102353e-01 7.94884682e-01 -2.71630257e-01 5.24224699e-01 1.44204986e+00 -6.58714622e-02 8.42753470e-01 1.13791597e+00 1.82952389e-01 3.82458895e-01 -5.81580251e-02 -2.31334437e-02 9.17448759e-01 2.15353504e-01 5.58158398e-01 -5.62621951e-01 4.44584340e-01 5.18233597e-01 -5.10751596e-03 -2.22875521e-01 -8.57995152e-01 -1.40058613e+00 6.36709452e-01 7.17713356e-01 -4.94875401e-01 -2.38139883e-01 1.34112045e-01 1.59611195e-01 3.04476351e-01 9.04264450e-02 5.24155736e-01 -4.65842038e-01 -3.37840050e-01 -4.21248406e-01 3.48777086e-01 5.04186749e-01 1.32266831e+00 9.51251686e-01 2.59439766e-01 2.31010884e-01 3.61256570e-01 3.19690794e-01 9.55388963e-01 2.78416604e-01 -1.48301423e+00 7.18389452e-01 6.55441940e-01 3.87999564e-01 -7.43051112e-01 -5.58939517e-01 -3.35155547e-01 -2.28796050e-01 6.87389135e-01 4.20786917e-01 -4.39524919e-01 -1.06929052e+00 1.91653776e+00 5.54795504e-01 4.05563079e-02 2.56460816e-01 1.13610053e+00 7.19521284e-01 2.70429581e-01 -1.75719410e-01 3.98877174e-01 8.42335939e-01 -1.48696005e+00 -2.47449532e-01 -5.90594351e-01 6.26870573e-01 -3.00114155e-01 1.39201939e+00 1.25701666e-01 -6.16267264e-01 -6.23430073e-01 -1.28839135e+00 -2.27080762e-01 -5.45396686e-01 2.94779073e-02 6.12638056e-01 3.23060960e-01 -9.57352281e-01 9.61399972e-02 -9.44037020e-01 -4.80550915e-01 5.54052472e-01 1.99402422e-01 -7.81086683e-01 -4.68514174e-01 -8.24806392e-01 9.82088804e-01 1.21671833e-01 -1.29662439e-01 -1.60141051e+00 -4.85865384e-01 -1.21003711e+00 -3.08856040e-01 8.35053682e-01 -8.19864035e-01 1.26719511e+00 -4.74769831e-01 -1.50823617e+00 5.49352288e-01 -1.03489317e-01 -6.37997687e-01 7.31527746e-01 -2.98438787e-01 -1.91667542e-01 4.45858389e-02 4.17558610e-01 1.21673501e+00 5.26841581e-01 -1.31016874e+00 -7.64460564e-01 -5.62034488e-01 6.50026619e-01 6.37403369e-01 1.91142689e-02 -6.51055038e-01 -6.87787771e-01 -1.58841357e-01 2.40911901e-01 -1.39195716e+00 -3.55573773e-01 2.38663509e-01 -3.11417311e-01 1.65975437e-01 8.45446885e-01 -3.89010251e-01 3.29820722e-01 -2.14297986e+00 8.78366828e-02 -1.13677308e-01 2.12343216e-01 3.45522426e-02 -2.56216049e-01 1.68978959e-01 4.72798288e-01 -1.29979745e-01 -8.03725347e-02 -4.28426206e-01 6.27841800e-02 3.54633003e-01 -2.80486614e-01 3.79578918e-01 -1.21852690e-02 1.13409126e+00 -1.04246068e+00 -1.57059774e-01 6.07111931e-01 1.83649614e-01 -8.28760564e-01 1.57242104e-01 -3.03537399e-01 9.30218637e-01 -4.98846263e-01 8.41400325e-01 3.93316060e-01 -1.28491849e-01 -2.22531900e-01 -1.21132016e-01 -2.81600863e-01 1.47527859e-01 -8.62430096e-01 2.13320065e+00 -5.15582263e-01 8.92820477e-01 -5.65678328e-02 -5.26172519e-01 7.04820931e-01 -2.54363030e-01 2.60739654e-01 -8.91290784e-01 1.95914909e-01 1.08609967e-01 -9.05687734e-02 -5.86152136e-01 6.25162065e-01 2.64650255e-01 -1.61362678e-01 2.56744146e-01 5.87894246e-02 -3.97279292e-01 1.92953214e-01 2.70749062e-01 1.29467392e+00 5.32381117e-01 -8.96939337e-02 9.20416042e-02 2.81688362e-01 5.45558453e-01 4.98023689e-01 7.25624382e-01 -7.31749952e-01 5.00869274e-01 1.24618210e-01 -4.30972576e-01 -1.19727874e+00 -1.08540392e+00 -1.32636717e-02 8.81943345e-01 6.22468650e-01 -6.51473925e-02 -5.84907353e-01 -7.85968900e-01 1.72730163e-01 7.60507047e-01 -5.66867411e-01 -1.74940825e-01 -2.33391300e-01 -2.69807070e-01 4.62736666e-01 5.87303698e-01 9.22752202e-01 -7.90071130e-01 -8.09091985e-01 -1.39070645e-01 -2.07369536e-01 -1.63149238e+00 -3.18245530e-01 2.05137469e-02 -5.91582060e-01 -1.21206498e+00 -1.42896488e-01 -3.76983017e-01 6.89121723e-01 8.30059171e-01 7.04884052e-01 -4.34488446e-01 1.97466072e-02 5.81282616e-01 -2.43203476e-01 -2.19923586e-01 -1.43084705e-01 1.12003379e-01 3.93038869e-01 -2.98162222e-01 1.95568278e-01 -5.34231007e-01 -5.86239398e-01 6.92318857e-01 -4.56057966e-01 2.53347903e-01 6.09123707e-01 6.65297985e-01 5.60234427e-01 -3.26697201e-01 4.47661966e-01 -5.58578491e-01 1.21639580e-01 -5.24969041e-01 -8.97825181e-01 -1.62042782e-01 -6.89996779e-01 2.00671926e-02 5.55719137e-01 -2.09948540e-01 -1.10478508e+00 1.17123626e-01 7.73296580e-02 -6.72344089e-01 -3.77521724e-01 4.29931551e-01 -3.45575064e-01 -3.94364625e-01 8.12384069e-01 1.47081062e-01 1.14839330e-01 2.93934289e-02 6.73803449e-01 5.52408755e-01 8.83789659e-01 -4.40117866e-01 1.00510442e+00 8.70824158e-01 -2.01056972e-01 -4.67018127e-01 -8.39426696e-01 -3.73254836e-01 -6.25083447e-01 -3.32198650e-01 9.00868535e-01 -1.40091586e+00 -7.01146066e-01 3.28307778e-01 -5.08262813e-01 -7.96191037e-01 -2.15727046e-01 6.73004746e-01 -7.39514470e-01 1.90224305e-01 -1.35188743e-01 -4.39310372e-01 1.86196014e-01 -1.40158045e+00 8.99841785e-01 3.42967212e-01 1.74297899e-01 -6.48426354e-01 2.43990552e-02 9.60913062e-01 2.93478310e-01 1.84094623e-01 2.75272340e-01 -3.90162677e-01 -1.20003998e+00 -2.86175936e-01 -2.83517629e-01 2.37106636e-01 5.51800523e-03 -2.88546085e-01 -9.29521978e-01 -2.77249247e-01 -2.71202713e-01 -8.41979325e-01 7.36278176e-01 2.66209207e-02 8.67309153e-01 9.47296768e-02 -1.93929568e-01 1.01051533e+00 1.07922733e+00 1.58303291e-01 2.43627891e-01 6.83123589e-01 7.94456780e-01 5.56418955e-01 9.13153648e-01 1.08866312e-01 1.03828573e+00 4.87386465e-01 8.43193531e-01 1.75889060e-01 -1.56488299e-01 -7.45850027e-01 4.10165995e-01 3.28221738e-01 7.33328015e-02 -2.68791497e-01 -9.65120673e-01 6.22502208e-01 -1.92014611e+00 -8.68783534e-01 4.17520076e-01 1.97034085e+00 3.11991096e-01 3.81099582e-01 -1.76050708e-01 -4.57864016e-01 3.38626683e-01 2.08526909e-01 -1.38640225e+00 3.09360951e-01 -2.57208347e-01 -5.66885352e-01 8.29071820e-01 4.29490715e-01 -1.16207612e+00 1.15014899e+00 5.92944384e+00 2.80237406e-01 -1.15204442e+00 1.06828310e-01 4.06772733e-01 -3.15009654e-01 -3.12753797e-01 7.80682042e-02 -7.14585304e-01 3.06886047e-01 7.12262988e-01 -1.64487287e-02 5.78421593e-01 1.16076636e+00 4.86007303e-01 -2.92743206e-01 -1.11119080e+00 9.96368825e-01 8.25694799e-02 -1.37358499e+00 -3.13416272e-01 2.53167599e-01 1.06735098e+00 8.52191269e-01 2.54835814e-01 6.63762093e-01 7.39090085e-01 -8.13071549e-01 7.86823750e-01 1.81662887e-01 6.20304406e-01 -3.54526311e-01 5.62843442e-01 5.43169379e-01 -8.46338153e-01 -3.12943429e-01 -2.46660918e-01 -2.78959751e-01 3.15646261e-01 -9.61388201e-02 -7.51182914e-01 4.03213978e-01 7.47175276e-01 9.37298715e-01 -5.60739279e-01 9.17498052e-01 -4.89595860e-01 2.83999562e-01 -5.39434493e-01 1.96458772e-01 5.48222482e-01 -1.53766260e-01 5.67128420e-01 3.82820070e-01 1.76690027e-01 1.89907148e-01 4.54589367e-01 7.75982618e-01 -1.82316810e-01 -2.90784091e-01 -9.61753845e-01 2.18090504e-01 5.95642447e-01 1.16162467e+00 -4.42094266e-01 -2.72448272e-01 -5.93901515e-01 6.52724802e-01 4.10410315e-01 5.66581726e-01 -9.24155891e-01 1.17689818e-01 9.50934470e-01 -3.15792933e-02 2.56612182e-01 -8.06939363e-01 -3.18986744e-01 -1.39402819e+00 2.12623373e-01 -7.86689341e-01 -7.32993260e-02 -1.13722312e+00 -8.17815542e-01 6.35042846e-01 -4.19932380e-02 -1.43750942e+00 -2.53784388e-01 -5.90574622e-01 -3.75405371e-01 5.38232863e-01 -1.90867186e+00 -1.28375161e+00 -9.64073598e-01 7.09219754e-01 7.52020955e-01 -4.32409227e-01 3.56548369e-01 1.16916366e-01 -6.71125531e-01 3.60708445e-01 4.02243733e-02 -2.82337493e-03 7.90104389e-01 -9.90855515e-01 6.43411517e-01 1.07085061e+00 1.03654571e-01 3.72335196e-01 6.04782999e-01 -5.91433227e-01 -1.63266969e+00 -1.15656459e+00 1.11876108e-01 -7.71743774e-01 7.66175389e-01 -1.89601019e-01 -3.95236492e-01 9.35363948e-01 -1.81932487e-02 2.25696221e-01 3.13253731e-01 -6.68361560e-02 -4.55427796e-01 -7.16421753e-02 -9.93924737e-01 9.67480183e-01 1.54087698e+00 -5.75634539e-01 -4.51534122e-01 3.19635928e-01 9.26959693e-01 -7.45432019e-01 -4.76193845e-01 5.69766700e-01 4.95477051e-01 -1.02619219e+00 9.29486990e-01 -6.00227416e-01 3.84557664e-01 -6.98188365e-01 -4.49223697e-01 -1.25177610e+00 8.85991827e-02 -4.06735837e-01 4.66395579e-02 6.07454419e-01 5.63723385e-01 -7.01057553e-01 1.05371177e+00 8.36977661e-01 -6.02805674e-01 -5.68167269e-01 -7.85413504e-01 -6.61653280e-01 -2.49274790e-01 -7.02823222e-01 4.56063002e-01 6.47556782e-01 -7.02014789e-02 2.36400262e-01 -3.12164277e-01 4.45329726e-01 5.61186671e-01 1.42206341e-01 1.56450033e+00 -7.92515218e-01 5.76698370e-02 -4.26216125e-02 -6.78091645e-01 -1.53800356e+00 3.47036421e-01 -6.82843328e-01 2.44986847e-01 -1.48729038e+00 -1.42661124e-01 -6.51623666e-01 -3.13236229e-02 5.61628759e-01 -2.92788018e-02 2.34065622e-01 1.96423367e-01 3.70409340e-01 -9.08667147e-01 8.91662002e-01 1.41134620e+00 -2.52947450e-01 -1.96156472e-01 -1.04991324e-01 -6.66004777e-01 8.16056907e-01 7.62136102e-01 4.42831293e-02 -8.57127726e-01 -9.68329191e-01 4.80505489e-02 -2.78398748e-02 6.14594281e-01 -1.32139361e+00 4.52714920e-01 -3.47225517e-01 9.41243321e-02 -6.00915492e-01 5.56348920e-01 -7.49139011e-01 -1.52451441e-01 2.10867926e-01 -1.37830108e-01 1.01294599e-01 2.71504194e-01 8.49361837e-01 -1.48652777e-01 1.76795006e-01 4.21482295e-01 -1.79472461e-01 -1.44648588e+00 5.44170737e-01 -2.14322701e-01 2.51017094e-01 1.28645825e+00 -2.94963688e-01 -6.26498342e-01 -5.76112092e-01 -5.77357590e-01 8.54958177e-01 1.01836431e+00 7.22772241e-01 5.82500637e-01 -1.15048468e+00 -4.05694872e-01 4.51287210e-01 5.98967552e-01 4.02901143e-01 4.83474046e-01 7.03207731e-01 -5.58632612e-01 4.51854408e-01 -4.79781330e-01 -8.49832594e-01 -7.11715102e-01 5.67421675e-01 2.55888969e-01 3.33669186e-01 -7.30617046e-01 6.73584938e-01 3.13272953e-01 -8.81867826e-01 6.95725083e-02 -3.49403441e-01 7.97330588e-03 -3.87409002e-01 2.41372079e-01 1.21165961e-01 -1.04111828e-01 -7.72156715e-01 -1.88205600e-01 4.12357479e-01 3.86706330e-02 -3.12291801e-01 1.05802763e+00 -5.89809179e-01 6.03722274e-01 2.55926371e-01 9.44020212e-01 -5.27567975e-02 -2.29352665e+00 -2.32176319e-01 -3.85282367e-01 -5.92768669e-01 5.82573600e-02 -7.48734474e-01 -9.81305838e-01 8.53544891e-01 5.53204477e-01 -4.97325599e-01 6.20907605e-01 -3.12000178e-02 6.89780533e-01 9.25403178e-01 8.41522694e-01 -1.12500250e+00 2.51347512e-01 6.63093448e-01 5.98038435e-01 -1.88549447e+00 -2.86113441e-01 -1.84804916e-01 -8.67784381e-01 7.34513581e-01 1.14387715e+00 4.07813713e-02 4.53926235e-01 1.49313444e-02 3.66703242e-01 -4.34525087e-02 -7.61904120e-01 -4.45974797e-01 3.79186086e-02 8.48111987e-01 -4.87482756e-01 2.41058762e-03 3.82556289e-01 2.63594836e-01 -4.86243576e-01 -2.47514024e-02 6.82536006e-01 9.62721825e-01 -5.41330099e-01 -5.80784738e-01 -1.05891200e-02 1.87474936e-01 2.98365116e-01 1.26089901e-01 -4.80486546e-03 8.88115704e-01 1.70322657e-01 1.12772214e+00 -1.27155289e-01 -6.75921977e-01 3.59756678e-01 -1.94769695e-01 2.46212721e-01 -5.65485418e-01 2.22387165e-01 -3.70095521e-01 1.37151212e-01 -8.68941307e-01 -3.31469506e-01 -8.20154786e-01 -1.20027041e+00 -3.25595528e-01 -7.03114718e-02 -3.10374916e-01 9.17204022e-01 1.06433582e+00 6.69933498e-01 4.74278182e-01 5.36982358e-01 -1.12213922e+00 -5.43438196e-01 -5.91598928e-01 -1.66315529e-02 3.74817252e-01 3.75793874e-01 -1.11917055e+00 -3.48404616e-01 -8.17457959e-02]
[4.504973888397217, 0.6570364832878113]
11084c83-19fc-4220-8029-2c9a9d84eed3
long-term-spatio-temporal-forecasting-via
2204.11008
null
https://arxiv.org/abs/2204.11008v4
https://arxiv.org/pdf/2204.11008v4.pdf
Long-term Spatio-temporal Forecasting via Dynamic Multiple-Graph Attention
Many real-world ubiquitous applications, such as parking recommendations and air pollution monitoring, benefit significantly from accurate long-term spatio-temporal forecasting (LSTF). LSTF makes use of long-term dependency between spatial and temporal domains, contextual information, and inherent pattern in the data. Recent studies have revealed the potential of multi-graph neural networks (MGNNs) to improve prediction performance. However, existing MGNN methods cannot be directly applied to LSTF due to several issues: the low level of generality, insufficient use of contextual information, and the imbalanced graph fusion approach. To address these issues, we construct new graph models to represent the contextual information of each node and the long-term spatio-temporal data dependency structure. To fuse the information across multiple graphs, we propose a new dynamic multi-graph fusion module to characterize the correlations of nodes within a graph and the nodes across graphs via the spatial attention and graph attention mechanisms. Furthermore, we introduce a trainable weight tensor to indicate the importance of each node in different graphs. Extensive experiments on two large-scale datasets demonstrate that our proposed approaches significantly improve the performance of existing graph neural network models in LSTF prediction tasks.
['Flora Salim', 'Junshan Zhang', 'Zhaofeng Zhang', 'Hamid Menouar', 'Xiao Xiao', 'Yufan Kang', 'Shuo Wang', 'Zhiling Jin', 'Wei Shao']
2022-04-23
null
null
null
null
['spatio-temporal-forecasting']
['time-series']
[-1.63199618e-01 -3.07588518e-01 -3.21155608e-01 -3.36360544e-01 -1.61100760e-01 -9.78109613e-02 3.25700015e-01 4.13762182e-01 1.12419620e-01 5.34409106e-01 3.93351883e-01 -3.73643935e-01 -5.74842036e-01 -1.04596353e+00 -5.98470628e-01 -6.62452459e-01 -3.13029528e-01 1.73797905e-01 3.33496243e-01 -2.83347964e-01 -1.59115940e-01 3.85132790e-01 -1.26396120e+00 2.01086581e-01 1.12051201e+00 1.15285778e+00 2.92442977e-01 4.02252942e-01 -3.57057631e-01 9.44631815e-01 -2.55715400e-01 -2.53459185e-01 -1.10883787e-01 -6.44902959e-02 -3.66456687e-01 8.68929625e-02 3.98032725e-01 1.45871401e-01 -6.93568408e-01 9.27206278e-01 3.94951731e-01 3.68471831e-01 1.69222385e-01 -1.57923830e+00 -1.07535565e+00 6.48112774e-01 -6.80344343e-01 5.15216947e-01 8.06972012e-03 1.17584532e-02 1.27876163e+00 -7.40396261e-01 2.83324867e-01 1.14960384e+00 8.40316713e-01 -7.52187893e-02 -8.57722819e-01 -6.41744435e-01 9.48163688e-01 5.33110857e-01 -1.32161438e+00 -9.55552310e-02 1.12882781e+00 -3.49694848e-01 1.10692108e+00 1.67809442e-01 6.89082503e-01 8.97581518e-01 3.64343494e-01 4.78861392e-01 6.76688969e-01 -4.75248806e-02 -1.68419629e-01 -3.73999387e-01 3.97979945e-01 8.49003494e-01 2.45901048e-01 -6.86990991e-02 -6.23011172e-01 -2.12108642e-01 4.78082865e-01 4.19064701e-01 -3.78578342e-02 -2.23728463e-01 -1.08150315e+00 5.25604546e-01 9.77111757e-01 5.46584308e-01 -6.36295199e-01 3.55194449e-01 2.88808286e-01 1.81744676e-02 9.81785119e-01 -4.32680435e-02 -2.82358468e-01 3.60598117e-01 -5.74146748e-01 3.64188873e-03 3.33855927e-01 7.83716857e-01 8.71917546e-01 2.29449064e-01 -3.47336829e-01 8.08532953e-01 4.50040430e-01 5.01318336e-01 3.57057333e-01 -4.34093982e-01 8.73685420e-01 9.63656187e-01 -3.52538794e-01 -1.79765368e+00 -5.48792899e-01 -8.08089495e-01 -1.12657070e+00 -5.09680629e-01 -5.41834235e-02 -2.07426231e-02 -9.92622256e-01 1.79347324e+00 3.94125491e-01 8.39363933e-01 -4.11704570e-01 6.64643466e-01 1.05436599e+00 5.34297109e-01 2.41657704e-01 -1.25910938e-01 9.30868149e-01 -1.01082957e+00 -9.19227064e-01 -4.56633508e-01 7.00499833e-01 -1.46343097e-01 9.31272447e-01 -2.76609004e-01 -6.49279714e-01 -6.12183213e-01 -8.60590041e-01 1.32520750e-01 -7.20921874e-01 -1.41981512e-01 7.89270818e-01 2.16259256e-01 -1.10327542e+00 4.98615593e-01 -5.72123170e-01 -3.03907365e-01 3.72636050e-01 3.09256583e-01 -2.43930042e-01 -2.90721536e-01 -1.48509800e+00 4.98604208e-01 2.01877236e-01 5.92587829e-01 -2.01652452e-01 -5.65745950e-01 -9.59499240e-01 2.49914065e-01 5.04358947e-01 -6.89902008e-01 5.57048142e-01 -6.69322431e-01 -5.22947967e-01 9.89682823e-02 -3.29559505e-01 -2.81869084e-01 -1.98503420e-01 6.27001747e-02 -9.15208459e-01 -2.62423635e-01 2.42045313e-01 7.56836385e-02 6.19287670e-01 -9.20482814e-01 -6.68845713e-01 -7.63471663e-01 -2.72590984e-02 3.00842553e-01 -6.58183217e-01 -3.63167912e-01 -5.48604429e-01 -8.06056976e-01 8.97819176e-02 -8.46631706e-01 -3.64487261e-01 -4.34566468e-01 -3.68092954e-01 -5.40314078e-01 1.00603986e+00 -6.88200295e-01 1.74957740e+00 -2.22781277e+00 1.22005865e-01 2.70444542e-01 6.92985237e-01 2.11796254e-01 -2.95428604e-01 4.10867095e-01 3.96255264e-03 1.57218426e-01 1.43709421e-01 -2.82336265e-01 -1.18804403e-01 4.67741579e-01 -9.30881724e-02 2.00328395e-01 -3.97138707e-02 1.33566964e+00 -9.93510604e-01 -4.23750550e-01 2.84776509e-01 6.28175080e-01 -1.30447850e-01 3.20472643e-02 -2.25681067e-01 1.57824054e-01 -7.31916487e-01 5.73511064e-01 6.08846426e-01 -8.33698750e-01 3.16655934e-01 -4.88263935e-01 2.84457296e-01 1.94683716e-01 -8.38720202e-01 1.55415297e+00 -3.34287286e-01 2.37277821e-01 -1.71800077e-01 -1.13698125e+00 7.45740235e-01 1.12936094e-01 8.20454121e-01 -1.14145601e+00 -1.41079351e-01 -6.60682842e-02 -1.47716448e-01 -4.94857371e-01 5.25460601e-01 1.86067104e-01 -4.15730029e-02 3.25960070e-01 -5.41374683e-02 6.99160159e-01 1.58058852e-01 3.39589834e-01 1.09722531e+00 -2.52953529e-01 -6.08781502e-02 -7.80536756e-02 4.13796395e-01 -3.21438342e-01 6.99928820e-01 4.63589221e-01 -3.35684806e-01 2.58439362e-01 2.38820150e-01 -8.35276365e-01 -5.23262858e-01 -6.30879700e-01 5.36063552e-01 1.31438053e+00 1.79376081e-01 -6.14265859e-01 -1.83887005e-01 -9.07232881e-01 4.09933001e-01 4.67805386e-01 -7.56514192e-01 -4.57933903e-01 -6.46290481e-01 -9.33078170e-01 2.16581687e-01 7.75225222e-01 3.27275157e-01 -8.24077845e-01 2.77171850e-01 2.71101445e-01 -4.49975610e-01 -1.48635483e+00 -6.31181121e-01 -1.02346286e-01 -8.15210462e-01 -1.01182318e+00 -1.35052025e-01 -5.95201969e-01 4.73321974e-01 9.46612477e-01 1.22487032e+00 2.09106803e-01 9.85475034e-02 3.16956967e-01 -1.84870332e-01 -6.18654154e-02 1.87810317e-01 2.42426485e-01 -5.94600290e-02 2.91709989e-01 2.96962827e-01 -9.11513269e-01 -5.28380275e-01 3.27352524e-01 -6.29785001e-01 5.18919937e-02 4.12203729e-01 6.15031838e-01 6.30375206e-01 3.30190450e-01 8.39601159e-01 -8.54138315e-01 8.73838484e-01 -9.45299387e-01 -2.45573744e-01 5.63454628e-01 -9.18302357e-01 1.15386315e-01 5.44376373e-01 -2.98842221e-01 -7.57913351e-01 -2.83564359e-01 1.18548103e-01 -7.37753630e-01 1.22545928e-01 1.13209867e+00 -1.62530094e-01 -7.71506131e-02 2.30604008e-01 8.20729434e-02 -2.83638626e-01 -3.32119614e-01 4.19317216e-01 2.01721773e-01 3.11357021e-01 -3.67823333e-01 6.24605060e-01 2.65139580e-01 2.23658577e-01 -4.66859192e-01 -8.97241652e-01 -5.05120516e-01 -6.02193654e-01 -3.99252385e-01 6.95222497e-01 -1.04205132e+00 -4.86693382e-01 3.86752367e-01 -8.93201649e-01 -2.42184892e-01 7.04903826e-02 2.35309869e-01 8.10935795e-02 4.90123630e-01 -3.43732566e-01 -7.79488862e-01 -3.91217858e-01 -7.67201722e-01 1.08081758e+00 9.06765088e-03 3.14774841e-01 -1.45748103e+00 -1.13463804e-01 4.11703557e-01 5.08965790e-01 3.61772537e-01 1.04668355e+00 -3.83389384e-01 -6.36506379e-01 -2.86213398e-01 -5.98935425e-01 2.77576689e-02 4.55883890e-01 -1.92494035e-01 -6.08297765e-01 -2.40709543e-01 -4.21591222e-01 1.60671473e-01 1.04768419e+00 5.04710495e-01 1.30287766e+00 -4.53514755e-01 -6.46555781e-01 5.42987823e-01 1.26557338e+00 -1.81942210e-02 1.68420315e-01 -9.95300338e-02 1.60926759e+00 4.99800682e-01 3.70003402e-01 2.78274328e-01 1.09337080e+00 6.19302571e-01 7.44945407e-01 -3.91574204e-02 -2.57043093e-01 -3.27751756e-01 1.28957883e-01 1.15524733e+00 -1.71437174e-01 -6.03854418e-01 -9.59451616e-01 7.68942237e-01 -2.38694286e+00 -8.59433591e-01 -3.44566315e-01 1.97420716e+00 -3.57214063e-02 3.18944044e-02 1.98209256e-01 -7.70915598e-02 9.91231263e-01 6.93503082e-01 -5.85041523e-01 2.25873967e-03 -2.22411841e-01 -3.48805875e-01 5.83995461e-01 3.57308418e-01 -1.00767982e+00 7.62861669e-01 5.83431244e+00 7.87707210e-01 -1.19381344e+00 1.78224921e-01 6.25104785e-01 -4.01126593e-02 -5.75345874e-01 -2.38846660e-01 -5.72905481e-01 5.98896205e-01 1.13968027e+00 -1.29597649e-01 6.06889546e-01 5.30477881e-01 2.02169031e-01 4.43299174e-01 -5.86697221e-01 8.65961075e-01 1.78520575e-01 -1.36104238e+00 1.73405871e-01 2.69812971e-01 6.01284266e-01 5.28452098e-01 2.59833243e-02 3.10235560e-01 4.35141146e-01 -9.57297325e-01 2.50766695e-01 6.20573640e-01 3.92989129e-01 -6.90682769e-01 5.76772928e-01 3.03515136e-01 -1.95454037e+00 -2.77679443e-01 -1.37665510e-01 -6.52213842e-02 1.03754655e-01 9.81561840e-01 -7.47870922e-01 1.10765517e+00 6.55593336e-01 1.28692305e+00 -7.69681633e-01 7.20113993e-01 4.35029194e-02 5.29208720e-01 -2.27101445e-01 6.76184446e-02 3.46443832e-01 -1.72686845e-01 4.07936096e-01 8.51433277e-01 4.66964602e-01 -1.35658070e-01 4.36878622e-01 5.52611589e-01 -6.35311827e-02 8.29312801e-02 -8.47671866e-01 -2.94998497e-01 4.06188577e-01 1.20037234e+00 -4.08484310e-01 -1.31445587e-01 -7.59373546e-01 5.53316295e-01 7.82362759e-01 6.61734402e-01 -9.54198718e-01 2.86637787e-02 6.63009763e-01 2.05885410e-01 4.22756016e-01 -5.07709563e-01 -2.81642139e-01 -1.14614511e+00 3.67932975e-01 -3.39911163e-01 8.66748095e-01 -8.28159153e-01 -1.54687345e+00 7.66167521e-01 -2.01294705e-01 -9.19151127e-01 -6.87813088e-02 -2.85420060e-01 -6.21721089e-01 6.89421535e-01 -1.65650928e+00 -1.78522396e+00 -5.46115279e-01 8.40699494e-01 1.98280796e-01 -9.84051898e-02 5.13666749e-01 5.67179620e-01 -8.06032717e-01 4.21426952e-01 -8.91865045e-02 8.45762119e-02 3.79232645e-01 -1.06754410e+00 7.91681468e-01 9.81992483e-01 2.60848135e-01 3.80861223e-01 2.38782182e-01 -1.03053260e+00 -1.49880838e+00 -1.66481256e+00 1.12089860e+00 -2.35845044e-01 9.22426760e-01 -3.55210841e-01 -1.10282433e+00 8.13892305e-01 -8.85894895e-02 6.74210191e-01 4.67868119e-01 6.69171095e-01 -5.12523592e-01 -5.10466516e-01 -6.51552439e-01 3.42584431e-01 1.32241273e+00 -8.05198073e-01 9.69018638e-02 4.43400055e-01 1.05890214e+00 -6.87151700e-02 -9.56493676e-01 7.52352536e-01 4.71479863e-01 -7.49571502e-01 1.05394351e+00 -6.26645803e-01 7.35673234e-02 -3.13527644e-01 -2.27261916e-01 -1.44509232e+00 -8.75264943e-01 -1.79431856e-01 -5.61055779e-01 1.08499360e+00 3.72618794e-01 -7.77317047e-01 6.52575731e-01 5.23932815e-01 -2.36911044e-01 -8.98550391e-01 -9.63761330e-01 -5.49996018e-01 -3.57656032e-01 -6.81649923e-01 1.05110395e+00 1.15739691e+00 -8.75095204e-02 6.06364846e-01 -7.47655213e-01 4.72496003e-01 4.46363956e-01 4.39802885e-01 4.65759903e-01 -1.51893520e+00 -5.24173491e-02 -3.19620758e-01 -4.57300931e-01 -7.27674842e-01 2.99594104e-01 -1.00035310e+00 -3.33145499e-01 -2.01553559e+00 7.09391534e-02 -5.76365709e-01 -9.71854091e-01 6.44431531e-01 -5.88590205e-01 3.01405992e-02 4.03950885e-02 9.26081911e-02 -1.01512301e+00 6.44437671e-01 1.23759842e+00 -2.77737707e-01 -1.22415304e-01 -5.29563464e-02 -7.89428830e-01 3.37272674e-01 5.79298735e-01 -2.46727884e-01 -7.71416724e-01 -7.44639754e-01 5.69428682e-01 1.03065893e-01 3.42524827e-01 -9.73493874e-01 4.53175902e-01 -4.12225574e-01 1.88917503e-01 -7.40015507e-01 2.51319587e-01 -8.94038141e-01 2.98200279e-01 4.65189219e-02 7.06054941e-02 4.12384808e-01 1.15208708e-01 1.17765260e+00 -2.14115381e-01 8.11014295e-01 1.66356236e-01 -3.57951745e-02 -8.93733203e-01 9.71323550e-01 5.08602746e-02 -1.88857883e-01 8.12379181e-01 -1.11047499e-01 -6.00656033e-01 -4.26919281e-01 -7.83888042e-01 5.68948209e-01 9.41664055e-02 7.63059080e-01 6.37510538e-01 -1.76449144e+00 -4.46784556e-01 3.02797973e-01 3.09810430e-01 -8.52590129e-02 6.39966190e-01 9.68859553e-01 2.76586890e-01 3.72280240e-01 -2.78666262e-02 -5.31082451e-01 -1.08840287e+00 8.39999914e-01 2.69554645e-01 -6.22634053e-01 -5.46671271e-01 5.27637601e-01 3.07402462e-01 -4.18883741e-01 1.44643942e-02 -5.51696658e-01 -3.50931585e-01 3.46653834e-02 4.19067591e-02 2.80719608e-01 3.07051092e-01 -9.83580709e-01 -4.86385107e-01 5.56630254e-01 2.47279093e-01 4.89378899e-01 1.42479527e+00 -4.72167134e-01 -1.76351473e-01 5.49541533e-01 1.00726998e+00 -2.61719018e-01 -8.57852936e-01 -6.90131664e-01 -5.08972593e-02 -4.12678003e-01 3.18791538e-01 -5.83523929e-01 -1.56060839e+00 7.57137656e-01 2.32505068e-01 5.82773387e-01 1.16495454e+00 -9.19398069e-02 1.19223785e+00 1.26734868e-01 2.77724057e-01 -9.61890340e-01 1.21595293e-01 5.44327378e-01 7.58210838e-01 -1.17125392e+00 -1.03499368e-02 -5.44135332e-01 -5.70113003e-01 7.66090155e-01 7.67810464e-01 -5.12582660e-02 1.08788180e+00 -1.43254949e-02 -2.79216856e-01 -5.14212132e-01 -1.02753532e+00 -4.56958830e-01 7.56109416e-01 5.30188441e-01 2.13731289e-01 3.03108543e-01 1.94437638e-01 5.60224950e-01 2.91547894e-01 -1.31476715e-01 -3.35605405e-02 6.02729440e-01 -3.11397463e-01 -9.27349269e-01 1.12266175e-01 7.28304923e-01 -2.33514950e-01 -9.72160026e-02 -3.04192454e-01 4.25916612e-01 2.20135778e-01 1.06422853e+00 9.86981089e-04 -9.72493589e-01 2.58054644e-01 -1.94816235e-02 2.81981044e-02 -3.27272058e-01 -4.23979819e-01 -1.33070890e-02 2.16573074e-01 -7.12044179e-01 -6.08821988e-01 -3.82879883e-01 -9.85226393e-01 -4.71707851e-01 -3.26476306e-01 1.64914519e-01 5.20947576e-01 1.11903965e+00 1.00513649e+00 9.39756453e-01 5.72766960e-01 -5.16272128e-01 2.21428007e-01 -9.06979084e-01 -6.77752972e-01 5.08967936e-01 3.88166785e-01 -8.14446092e-01 -2.01772571e-01 -5.08004367e-01]
[6.680901050567627, 2.5162508487701416]
3f5d9c57-497c-45bb-b838-130c56ee124b
bayesian-pseudo-labels-expectation
2208.04435
null
https://arxiv.org/abs/2208.04435v3
https://arxiv.org/pdf/2208.04435v3.pdf
Bayesian Pseudo Labels: Expectation Maximization for Robust and Efficient Semi-Supervised Segmentation
This paper concerns pseudo labelling in segmentation. Our contribution is fourfold. Firstly, we present a new formulation of pseudo-labelling as an Expectation-Maximization (EM) algorithm for clear statistical interpretation. Secondly, we propose a semi-supervised medical image segmentation method purely based on the original pseudo labelling, namely SegPL. We demonstrate SegPL is a competitive approach against state-of-the-art consistency regularisation based methods on semi-supervised segmentation on a 2D multi-class MRI brain tumour segmentation task and a 3D binary CT lung vessel segmentation task. The simplicity of SegPL allows less computational cost comparing to prior methods. Thirdly, we demonstrate that the effectiveness of SegPL may originate from its robustness against out-of-distribution noises and adversarial attacks. Lastly, under the EM framework, we introduce a probabilistic generalisation of SegPL via variational inference, which learns a dynamic threshold for pseudo labelling during the training. We show that SegPL with variational inference can perform uncertainty estimation on par with the gold-standard method Deep Ensemble.
['Joseph Jacob', 'Yipeng Hu', 'Neil P. Oxtoby', 'Daniel C. Alexander', 'Marius de Groot', 'Chen Jin', 'Yukun Zhou', 'Mou-Cheng Xu']
2022-08-08
null
null
null
null
['semi-supervised-medical-image-segmentation']
['computer-vision']
[ 4.84173983e-01 8.40558290e-01 2.22817019e-01 -5.09783149e-01 -1.41201377e+00 -4.37224984e-01 6.48756742e-01 6.32417202e-02 -6.56651735e-01 6.61379576e-01 -4.26632762e-02 -3.33801389e-01 -1.81126580e-01 -4.85263944e-01 -8.71733367e-01 -1.12116361e+00 2.16573969e-01 8.80789518e-01 4.44456488e-01 3.99505913e-01 -7.51239657e-02 3.20757866e-01 -8.87814939e-01 7.84914345e-02 7.68229663e-01 7.88163006e-01 -4.30935919e-02 7.16176212e-01 -5.51319160e-02 5.24106026e-01 -4.85744685e-01 -4.51136500e-01 2.44483650e-01 -5.27928293e-01 -1.20046258e+00 5.39724052e-01 2.39431292e-01 -1.47374541e-01 3.98072377e-02 1.33935797e+00 5.96932828e-01 -7.28855208e-02 1.06112909e+00 -1.20487952e+00 1.90708078e-02 8.60870242e-01 -5.42742312e-01 1.05934730e-02 -1.45904511e-01 -3.30631025e-02 6.05398059e-01 -4.74678040e-01 6.79409623e-01 1.00809038e+00 8.19061875e-01 6.90027535e-01 -1.65642560e+00 2.66973767e-02 -1.74907118e-01 -2.77372509e-01 -1.25937903e+00 -1.90854892e-01 3.99314016e-01 -6.05416238e-01 4.18251693e-01 3.25229019e-01 2.96462893e-01 1.02375090e+00 2.34474331e-01 1.06629527e+00 1.64755845e+00 -5.36039710e-01 6.24377787e-01 1.31212950e-01 1.25140041e-01 7.40575612e-01 5.10978699e-02 1.03426054e-01 -5.45470156e-02 -4.92539823e-01 7.27566838e-01 -5.65096915e-01 -2.16098398e-01 -5.77780604e-01 -1.08580458e+00 7.79547274e-01 4.94986437e-02 3.61321151e-01 -2.74378300e-01 4.40436602e-01 4.64591950e-01 -1.57069758e-01 5.91280341e-01 1.99651290e-02 -3.70961905e-01 1.08940251e-01 -1.59344602e+00 -4.77887280e-02 8.48861158e-01 8.17186058e-01 4.00902241e-01 -1.97629735e-01 -3.10384125e-01 5.72918415e-01 8.57650518e-01 4.80765790e-01 3.16220820e-01 -1.32329905e+00 -2.09665865e-01 -1.75188363e-01 -2.31047004e-01 -3.27688992e-01 -4.68415529e-01 -5.34985185e-01 -9.35736358e-01 3.95600975e-01 6.27405047e-01 -9.56816450e-02 -1.12977660e+00 1.70086467e+00 4.29157227e-01 4.79654282e-01 -8.98814499e-02 5.52334487e-01 6.55539930e-01 6.22607907e-03 5.43750301e-02 -4.81542438e-01 1.32970214e+00 -8.47348452e-01 -6.46975100e-01 4.66305297e-03 5.68120599e-01 -4.75541025e-01 2.13530108e-01 5.06186843e-01 -1.18698657e+00 -2.19329327e-01 -8.65963459e-01 3.24255228e-01 9.75808650e-02 -1.82581112e-01 2.51126170e-01 1.22448909e+00 -1.28333592e+00 8.78260672e-01 -1.57301414e+00 3.35413665e-02 7.14656055e-01 3.21129292e-01 -2.27896407e-01 3.08172315e-01 -9.33887959e-01 1.01442063e+00 5.48359513e-01 1.04588442e-01 -9.76621091e-01 -6.61996186e-01 -9.89568532e-01 -6.18061185e-01 2.57792264e-01 -7.78652251e-01 1.35213435e+00 -5.95419884e-01 -1.73870873e+00 1.36663234e+00 -2.05469489e-01 -8.73605907e-01 1.24255896e+00 8.90267082e-03 3.42982635e-02 4.55097258e-01 2.94799149e-01 7.94499218e-01 1.18433893e+00 -1.52306116e+00 -3.21584731e-01 -2.09843561e-01 -6.09969079e-01 -2.64041692e-01 6.88311934e-01 -1.76758066e-01 -4.20490474e-01 -5.08176148e-01 3.61935198e-01 -1.06600642e+00 -8.09987247e-01 -2.99914479e-01 -8.88700485e-01 5.87807177e-03 3.07723045e-01 -5.67471147e-01 8.20802033e-01 -1.94431365e+00 2.16334566e-01 6.02797747e-01 3.11927289e-01 -2.52520032e-02 3.65203798e-01 -1.81743279e-01 -1.45829886e-01 2.02285692e-01 -1.13573945e+00 -8.86562645e-01 2.47063130e-01 6.01314425e-01 -1.45168379e-01 9.74676847e-01 1.23815630e-02 1.07659340e+00 -8.63639593e-01 -9.73587275e-01 4.76173580e-01 5.72796047e-01 -2.21330538e-01 -1.72013491e-01 -1.41578555e-01 1.17654979e+00 -3.86326641e-01 3.22619557e-01 9.96249378e-01 -2.36870155e-01 1.55954836e-02 -9.84344035e-02 1.88027307e-01 -1.33338884e-01 -1.13274491e+00 2.05220270e+00 -2.07297266e-01 2.69006193e-01 2.88259774e-01 -1.18997204e+00 4.17436570e-01 4.89705831e-01 7.07522392e-01 -7.93650001e-03 4.79651183e-01 5.26047528e-01 -3.98370713e-01 -2.25730643e-01 -4.03987914e-02 -4.67558473e-01 -2.53849458e-02 5.50181925e-01 3.08530360e-01 -7.38392711e-01 7.01326737e-03 2.65315652e-01 9.68454719e-01 4.97447908e-01 3.21435690e-01 -6.19489670e-01 5.86276174e-01 -2.20779106e-01 4.33993667e-01 1.06379461e+00 -5.50890923e-01 1.04240358e+00 6.24257982e-01 1.52441010e-01 -7.37190127e-01 -1.35450935e+00 -1.06683099e+00 3.26252639e-01 -1.97907254e-01 -3.78486812e-02 -1.23364830e+00 -9.22892392e-01 -2.11443469e-01 8.99185717e-01 -7.65188932e-01 3.32075506e-01 -4.00781065e-01 -1.34788692e+00 8.51652861e-01 2.77342647e-01 5.56794763e-01 -7.13078678e-01 -8.11437130e-01 3.21476370e-01 -1.40690997e-01 -1.30961943e+00 -1.53240889e-01 3.59785885e-01 -1.11746645e+00 -9.67567503e-01 -1.01982093e+00 -4.80250835e-01 8.59156966e-01 -7.26604342e-01 1.27760243e+00 -2.11336643e-01 -2.66859651e-01 8.39027464e-01 -3.37561667e-02 -3.58208805e-01 -1.06577432e+00 -2.50656337e-01 -2.35459030e-01 -4.67454456e-02 1.30546778e-01 -6.19448543e-01 -4.98401821e-01 2.10804224e-01 -1.01322150e+00 -1.46898583e-01 4.36926156e-01 9.59073007e-01 1.14028144e+00 1.98936358e-01 1.02898590e-01 -1.41624904e+00 1.95458189e-01 -3.49158198e-01 -5.35349786e-01 1.33349568e-01 -6.76278949e-01 2.94268250e-01 -1.06622420e-01 3.14915478e-02 -1.06991565e+00 3.17386240e-01 -5.91270328e-01 -2.68202394e-01 -5.46828210e-01 2.00389415e-01 1.23509832e-01 -1.49712443e-01 6.44643009e-01 1.71351999e-01 3.03527832e-01 -3.15934062e-01 6.35551214e-01 4.39400166e-01 8.43284190e-01 -7.41338074e-01 5.51664889e-01 1.11576831e+00 4.70374018e-01 -5.94585180e-01 -8.60695302e-01 -4.79072422e-01 -1.06723642e+00 -5.63361496e-02 1.31036091e+00 -5.80065250e-01 -3.43273640e-01 7.40662873e-01 -1.10593271e+00 -4.86602455e-01 -5.89186847e-01 4.26665962e-01 -1.06847823e+00 8.79239023e-01 -7.32360125e-01 -8.10894072e-01 -2.39664569e-01 -1.58285201e+00 1.49331260e+00 -1.31518066e-01 -1.09690636e-01 -1.44972074e+00 5.89335449e-02 4.34649229e-01 1.73339039e-01 7.71316409e-01 4.24280941e-01 -8.61789584e-01 -3.80535185e-01 -3.07480320e-02 -3.42886895e-02 6.98003471e-01 -1.64837644e-01 -1.13959268e-01 -1.24945128e+00 -1.95195332e-01 6.11700714e-01 -1.61938116e-01 1.10099816e+00 1.07457530e+00 1.17336965e+00 2.22043395e-01 -3.89009595e-01 5.74109375e-01 1.37562752e+00 -3.06776017e-01 5.63354373e-01 2.82133639e-01 5.50925612e-01 5.76528907e-01 3.54560822e-01 2.22701505e-01 4.23294194e-02 2.27485657e-01 5.50456583e-01 -1.30393520e-01 1.36393495e-02 1.74202219e-01 -2.01121382e-02 6.29169583e-01 1.10745050e-01 1.54224753e-01 -9.16264415e-01 5.51763356e-01 -1.89744508e+00 -5.27352154e-01 -5.16386867e-01 2.09043384e+00 9.65128005e-01 3.32586080e-01 -3.37307737e-03 1.68150589e-01 6.82255745e-01 -1.54020013e-02 -2.67028719e-01 -2.73147915e-02 -7.62357712e-02 5.48697650e-01 9.81893837e-01 6.45532012e-01 -1.41819811e+00 5.80779552e-01 6.38754702e+00 1.14107299e+00 -5.50128818e-01 6.98695421e-01 9.31685507e-01 5.30384004e-01 -2.90867865e-01 -4.77160588e-02 -4.01028961e-01 3.85115862e-01 9.40938413e-01 4.06102985e-01 -1.45834774e-01 6.66200340e-01 -1.82104588e-01 -4.66308147e-01 -1.14152277e+00 8.16819251e-01 8.20317864e-02 -1.12674618e+00 -4.23366666e-01 2.24034116e-01 8.83044958e-01 3.22242737e-01 4.10711914e-02 -1.67510256e-01 4.91712958e-01 -1.14116323e+00 8.38428617e-01 4.27205116e-01 6.25345707e-01 -5.35733938e-01 9.43972766e-01 4.35214669e-01 -6.64693058e-01 7.54735887e-01 -1.10167392e-01 7.46099949e-01 7.25909948e-01 1.01471817e+00 -6.72730207e-01 8.80286217e-01 4.62020665e-01 4.53077614e-01 -2.80611753e-01 1.06910980e+00 -6.02817118e-01 8.61565888e-01 -5.81134081e-01 5.90744436e-01 2.92969644e-01 -1.34390652e-01 1.02100587e+00 1.50221968e+00 -1.59227714e-01 -1.99500114e-01 1.86924174e-01 1.13095081e+00 2.44315490e-01 -2.11411759e-01 -2.44711220e-01 3.54167312e-01 -4.74682413e-02 1.18544483e+00 -1.32236731e+00 -3.56249481e-01 9.44213867e-02 1.08282483e+00 -2.10514307e-01 1.01598173e-01 -8.33526552e-01 2.09234431e-01 -1.08620316e-01 -3.21830958e-01 2.23289624e-01 2.48928275e-02 -6.50593758e-01 -9.06515062e-01 -2.93018937e-01 -4.84439552e-01 4.87856120e-01 -4.70517039e-01 -1.28518069e+00 6.38360023e-01 4.89822388e-01 -8.78199935e-01 -3.05981398e-01 -7.47961640e-01 -5.35848737e-01 7.36559927e-01 -1.37986147e+00 -1.02253032e+00 1.59583658e-01 2.71371603e-01 2.65722185e-01 2.79484451e-01 9.55688179e-01 -9.11083594e-02 -3.01432461e-01 3.18416625e-01 1.73003092e-01 2.03627661e-01 4.61095512e-01 -1.86194372e+00 2.78720796e-01 1.09899974e+00 2.84012735e-01 3.08423877e-01 9.34141159e-01 -4.93767142e-01 -5.28944612e-01 -1.07195580e+00 4.25277352e-01 -8.55991662e-01 6.59376621e-01 -3.95451486e-02 -7.60491073e-01 8.26587915e-01 3.59199017e-01 3.11406940e-01 7.23303020e-01 -4.63610411e-01 3.00329123e-02 6.18143559e-01 -1.66727936e+00 3.00684333e-01 7.26998150e-01 -3.90988261e-01 -9.11033511e-01 6.11449957e-01 4.41699803e-01 -8.51076245e-01 -1.09736145e+00 5.61274946e-01 1.18123889e-01 -1.01373100e+00 9.30370629e-01 -5.87763311e-03 1.71241254e-01 -3.51773083e-01 1.29847750e-02 -1.24234664e+00 1.29848525e-01 -8.85135412e-01 1.54788002e-01 9.63789999e-01 3.32954586e-01 -7.88235903e-01 7.75364995e-01 6.89089656e-01 -1.59097776e-01 -5.49122572e-01 -1.39456964e+00 -6.47760510e-01 4.80366558e-01 -1.03491485e+00 6.63678050e-02 7.35066712e-01 -2.16161713e-01 -1.67306483e-01 -6.08940087e-02 2.61773258e-01 1.46314061e+00 -2.91756868e-01 3.11842859e-01 -1.12180734e+00 -6.89988077e-01 -3.93646181e-01 -4.83111292e-01 -8.01141798e-01 4.54305679e-01 -1.15391791e+00 3.84188920e-01 -1.44643855e+00 1.78097680e-01 -2.17017680e-01 -3.47476184e-01 1.66106433e-01 1.16891712e-02 3.31923217e-01 -1.51712865e-01 2.12573677e-01 -6.40672803e-01 1.82224050e-01 1.26334834e+00 -1.00198165e-01 9.56552476e-02 4.20792580e-01 -2.73600847e-01 9.92473185e-01 5.45808792e-01 -8.92115176e-01 -2.41891921e-01 -5.95490672e-02 -8.74498338e-02 7.48732388e-02 6.57789707e-01 -7.45065570e-01 2.33266890e-01 3.85775208e-01 -2.31596306e-02 -6.57167733e-01 -3.79480198e-02 -8.35641503e-01 2.04752997e-01 5.54740131e-01 -2.09840462e-01 -7.09415376e-01 2.13922843e-01 7.52712846e-01 -1.01838917e-01 -8.38656962e-01 1.11971521e+00 -4.18159544e-01 -2.47194782e-01 3.18404026e-02 -4.07573879e-01 3.41113716e-01 8.41837287e-01 -1.07806303e-01 1.16600901e-01 -5.25689796e-02 -1.42487884e+00 2.75668204e-01 2.19380096e-01 -3.27792495e-01 2.68766463e-01 -8.48997772e-01 -9.49349999e-01 2.61470433e-02 -3.31903696e-01 6.58426702e-01 3.65914196e-01 1.59601080e+00 -5.86217344e-01 1.99477851e-01 2.31824309e-01 -1.26447380e+00 -1.12434626e+00 1.69994652e-01 7.08198488e-01 -5.28681993e-01 -7.76005745e-01 1.19063246e+00 2.19322607e-01 -6.79782450e-01 1.19981237e-01 -4.07631397e-01 1.28999785e-01 -2.41144046e-01 2.66239345e-01 3.16254467e-01 1.36321738e-01 -7.98726380e-01 -4.75260258e-01 4.34127122e-01 1.20915383e-01 -7.58158028e-01 1.08865976e+00 -2.74194866e-01 -3.62465113e-01 5.01243949e-01 1.07516146e+00 -4.28488776e-02 -1.37532818e+00 -2.46873140e-01 2.79684037e-01 3.46567370e-02 3.78088593e-01 -7.54532516e-01 -1.02389109e+00 8.40372205e-01 6.38137579e-01 1.33172020e-01 8.42706680e-01 2.16865793e-01 5.67136467e-01 -1.32881775e-01 4.97433245e-01 -1.17943060e+00 -5.84198236e-01 1.13439545e-01 5.29959679e-01 -1.32789409e+00 1.33871004e-01 -5.67518592e-01 -6.81255400e-01 1.01470006e+00 -2.88883477e-01 -2.03483149e-01 1.13808286e+00 5.04082024e-01 1.64409339e-01 -2.09842354e-01 -1.19903952e-01 -3.11313629e-01 4.80676949e-01 6.08752787e-01 1.42650336e-01 1.89945713e-01 -2.51846373e-01 3.26300621e-01 -1.38754621e-01 -1.03636254e-02 4.13631797e-01 9.37345684e-01 -1.52737603e-01 -1.24993968e+00 -3.77254099e-01 1.83146685e-01 -1.14186108e+00 -1.23585172e-01 -6.21063635e-03 7.07407176e-01 1.13997526e-01 8.24493289e-01 -5.02018593e-02 2.19476894e-01 -1.11603640e-01 3.88616651e-01 7.43616760e-01 -5.80830991e-01 -3.72354090e-01 5.36582410e-01 -1.07667521e-01 -5.16455948e-01 -1.00119531e+00 -1.02358198e+00 -1.63137043e+00 3.19677502e-01 -4.51802850e-01 2.10907876e-01 7.59201646e-01 1.29579818e+00 -1.56585529e-01 7.45709240e-01 1.03416733e-01 -7.53795922e-01 -7.80444920e-01 -8.88785005e-01 -6.65663958e-01 4.06478554e-01 2.36856982e-01 -5.90002239e-01 -7.48335183e-01 1.48167431e-01]
[14.468591690063477, -2.095590829849243]
e8995d9e-d7d5-467b-ae3c-ef42eb65e7ae
a-pretraining-numerical-reasoning-model-for
null
null
https://aclanthology.org/2021.findings-emnlp.159
https://aclanthology.org/2021.findings-emnlp.159.pdf
A Pretraining Numerical Reasoning Model for Ordinal Constrained Question Answering on Knowledge Base
Knowledge Base Question Answering (KBQA) is to answer natural language questions posed over knowledge bases (KBs). This paper targets at empowering the IR-based KBQA models with the ability of numerical reasoning for answering ordinal constrained questions. A major challenge is the lack of explicit annotations about numerical properties. To address this challenge, we propose a pretraining numerical reasoning model consisting of NumGNN and NumTransformer, guided by explicit self-supervision signals. The two modules are pretrained to encode the magnitude and ordinal properties of numbers respectively and can serve as model-agnostic plugins for any IR-based KBQA model to enhance its numerical reasoning ability. Extensive experiments on two KBQA benchmarks verify the effectiveness of our method to enhance the numerical reasoning ability for IR-based KBQA models.
['Hong Chen', 'Cuiping Li', 'Quan Liu', 'Lemao Liu', 'Wayne Xin Zhao', 'Gaole He', 'Jing Zhang', 'Yu Feng']
null
null
null
null
findings-emnlp-2021-11
['knowledge-base-question-answering']
['natural-language-processing']
[-3.59326690e-01 6.16020739e-01 -3.38373452e-01 -6.18511915e-01 -1.03396285e+00 -6.80953801e-01 2.23712713e-01 9.96975601e-02 -3.15665960e-01 8.83301973e-01 2.47044295e-01 -6.31073773e-01 -1.91017315e-01 -1.38343489e+00 -9.93353248e-01 7.57441223e-02 1.24601431e-01 7.97855198e-01 1.40912786e-01 -9.39621806e-01 -1.09289512e-01 2.73551524e-01 -1.33887613e+00 8.34903657e-01 1.25266552e+00 1.62758839e+00 -4.96072888e-01 7.75298417e-01 -4.05025542e-01 1.68498206e+00 -5.25454164e-01 -9.48137343e-01 2.13456482e-01 -1.10624246e-01 -1.41295576e+00 -8.60571086e-01 5.89235842e-01 -7.36116588e-01 -3.33758950e-01 8.22562575e-01 1.13688506e-01 2.97232270e-01 7.45779634e-01 -1.44195068e+00 -1.25146830e+00 9.76638675e-01 6.83737695e-02 1.16371356e-01 8.08450341e-01 2.62568176e-01 1.77320421e+00 -3.94925624e-01 1.61079273e-01 1.51231992e+00 6.52733028e-01 7.88580418e-01 -7.53234982e-01 -3.00462514e-01 -2.15786435e-02 7.06128657e-01 -1.14839160e+00 -1.14407569e-01 8.35778058e-01 -5.19285612e-02 1.02433670e+00 4.87080306e-01 4.43609506e-01 5.59596896e-01 -4.90022868e-01 7.70662367e-01 7.86769271e-01 -4.30547982e-01 4.25250560e-01 1.60617799e-01 4.73637909e-01 9.99923408e-01 1.53164510e-02 -2.83192813e-01 -2.59597033e-01 -3.91961962e-01 8.83203983e-01 -6.12012208e-01 6.64433539e-02 -2.76133120e-01 -1.01772046e+00 1.14445829e+00 8.30523908e-01 -1.63103029e-01 -2.39641592e-01 5.40529311e-01 5.59940636e-01 5.78994215e-01 -1.21963650e-01 1.02362096e+00 -9.11870480e-01 -9.28832889e-02 -1.67869866e-01 7.36316323e-01 9.76631761e-01 8.15578341e-01 8.00475538e-01 -2.46483549e-01 -5.71182072e-01 7.58550704e-01 2.24748597e-01 6.17715120e-01 2.71696031e-01 -1.22057784e+00 6.77138150e-01 1.13855922e+00 2.54732281e-01 -7.62123406e-01 -4.93829072e-01 2.32533559e-01 -4.12504613e-01 -2.16965288e-01 5.29063404e-01 -7.08004236e-02 -6.72251582e-01 1.84522080e+00 5.19464493e-01 2.18508877e-02 5.32753646e-01 1.07381129e+00 1.36195493e+00 7.55918443e-01 2.95186460e-01 1.74971908e-01 1.81195498e+00 -8.08834314e-01 -6.46068573e-01 7.23978281e-02 1.01940918e+00 1.03970371e-01 1.77942491e+00 1.62156224e-01 -1.15493202e+00 -2.74303705e-01 -8.44615936e-01 -8.29103053e-01 -5.92917383e-01 6.91482574e-02 9.96445715e-01 5.87379873e-01 -9.68186080e-01 -1.51155919e-01 -4.12009597e-01 2.53400683e-01 3.27087015e-01 3.78819466e-01 -2.56098621e-02 -6.92609772e-02 -2.04906321e+00 1.09265077e+00 7.27370381e-01 1.32514209e-01 -4.06536400e-01 -9.39043880e-01 -1.18379426e+00 3.87530237e-01 5.71101546e-01 -1.11076438e+00 1.55395412e+00 -6.10714972e-01 -1.68077934e+00 4.89538044e-01 2.36434802e-01 -6.80313945e-01 2.58764148e-01 -1.07845180e-01 -3.59462589e-01 3.60175461e-01 -2.51014203e-01 9.38431978e-01 1.94819719e-01 -1.03830874e+00 -3.87908459e-01 -2.61866063e-01 1.08813930e+00 3.66948277e-01 -5.44392407e-01 -2.09225193e-01 -3.26154500e-01 -3.80000025e-01 -2.54786849e-01 -5.16869545e-01 -1.72992945e-01 -7.98019245e-02 -1.00454040e-01 -7.81752467e-01 3.84852797e-01 -7.37735152e-01 1.07691121e+00 -1.58253372e+00 -3.10326610e-02 2.84847319e-01 -1.28892049e-01 -1.41275674e-02 -4.39625651e-01 -1.10021746e-02 2.43822575e-01 5.55426627e-03 3.15841138e-02 5.40098131e-01 4.34254080e-01 5.57505369e-01 -7.58078694e-01 -2.47822180e-01 7.66930580e-01 1.46047652e+00 -1.20061111e+00 -7.77282059e-01 -1.81398079e-01 -9.71417427e-02 -1.04179251e+00 4.86202866e-01 -1.31297767e+00 -1.52799219e-01 -5.97153008e-01 8.65289330e-01 7.41959810e-01 -3.58786821e-01 1.42339557e-01 -3.97329599e-01 6.99954927e-01 6.69913292e-01 -1.02950203e+00 1.60472226e+00 -7.20839500e-01 -1.29016623e-01 -3.48446906e-01 -8.32120299e-01 7.53228545e-01 2.79811740e-01 3.83858606e-02 -9.55263972e-01 5.20125963e-02 6.46052882e-02 -1.00724399e-01 -6.09359324e-01 8.27406406e-01 -3.86438429e-01 -6.28677011e-01 3.18951249e-01 5.22405766e-02 -6.57416344e-01 4.16102737e-01 4.06160504e-01 9.79553342e-01 -9.41629112e-02 1.80835962e-01 1.80307757e-02 9.83441591e-01 1.74938515e-01 3.94657254e-01 6.84672296e-01 -5.65604493e-03 -8.80805589e-03 7.35155165e-01 -5.30671775e-01 -7.63478935e-01 -1.29099262e+00 -7.70694576e-03 1.64078450e+00 -3.04176733e-02 -3.25555772e-01 -4.43637639e-01 -9.55427885e-01 1.44532964e-01 6.88543797e-01 -6.92628384e-01 -4.11611974e-01 -6.98974431e-01 -3.85331333e-01 1.07176030e+00 9.46070552e-01 4.90939558e-01 -1.27427757e+00 -3.98802727e-01 -2.46650185e-02 -8.00210953e-01 -1.34906042e+00 1.14065148e-01 4.12345119e-02 -9.00766373e-01 -8.66409183e-01 -2.22927868e-01 -6.71672523e-01 5.05838096e-01 -5.51185846e-01 1.50635743e+00 1.60842314e-01 1.22856289e-01 5.31083286e-01 -4.67765719e-01 -4.73973423e-01 -5.24918079e-01 2.27599025e-01 -5.96683025e-02 -5.18530607e-01 4.22788948e-01 -2.80337602e-01 -4.27570820e-01 1.08352229e-01 -9.96150672e-01 -1.90872937e-01 4.07166928e-01 7.88654089e-01 1.85930729e-01 -2.20586553e-01 9.29257274e-01 -7.11598516e-01 8.90301824e-01 -4.09683973e-01 -7.36194253e-01 6.55557752e-01 -1.86468750e-01 5.24449348e-01 1.01044953e+00 -3.31738800e-01 -1.10337758e+00 -1.71221375e-01 -1.30133510e-01 -7.65438452e-02 1.84602886e-01 9.10097659e-01 -4.18801397e-01 1.42278805e-01 8.95769358e-01 -1.36047989e-01 -2.69546270e-01 -5.19471839e-02 1.05579770e+00 5.21879137e-01 9.06107247e-01 -1.52261400e+00 8.31723809e-01 3.45053732e-01 -5.38028777e-02 -1.01516746e-01 -1.39260101e+00 -4.94398624e-01 -2.17370912e-01 2.11294889e-01 6.55827522e-01 -1.06022692e+00 -1.35889804e+00 -4.08603996e-02 -1.06956542e+00 -4.81020510e-01 -3.98250937e-01 9.89812836e-02 -9.98981237e-01 2.37361535e-01 -8.35506201e-01 -8.23525071e-01 -6.66557848e-01 -6.31032944e-01 8.51284087e-01 1.48623839e-01 -1.29399702e-01 -9.86647546e-01 1.98344395e-01 9.90644217e-01 3.12143862e-01 -8.35038498e-02 1.57146358e+00 -5.68914592e-01 -3.93671602e-01 -5.05340211e-02 -6.20226264e-01 3.99690211e-01 -4.22833487e-02 -2.89196044e-01 -9.91258323e-01 3.69005084e-01 -3.85315716e-01 -1.34386516e+00 7.27256298e-01 -3.97579223e-02 1.43799102e+00 -9.00235415e-01 5.29462516e-01 3.30634117e-01 1.18733275e+00 -3.36273581e-01 7.86204159e-01 4.19252366e-01 5.59023678e-01 2.70076752e-01 7.22327292e-01 4.90722567e-01 1.15927279e+00 5.11723161e-01 6.37753904e-01 2.19201088e-01 3.27298343e-01 -2.69911587e-01 1.57333046e-01 4.13905293e-01 -1.59542248e-01 1.31602556e-01 -1.10606444e+00 5.87488472e-01 -1.89852881e+00 -8.52759600e-01 1.66740388e-01 1.63442338e+00 1.86536431e+00 -6.72331750e-02 2.12370217e-01 8.86820480e-02 -7.78579041e-02 -1.67491809e-01 -6.57053828e-01 -8.16911936e-01 1.73569396e-02 2.59413660e-01 7.47340396e-02 3.78699332e-01 -1.00462019e+00 1.08850777e+00 6.05892563e+00 6.67978346e-01 -5.87148070e-01 -4.76296514e-01 5.37742257e-01 1.85061663e-01 -6.03466928e-01 -2.00894594e-01 -6.29495263e-01 2.69961972e-02 1.16873860e+00 4.99554388e-02 6.49605691e-01 9.14064169e-01 -3.56500328e-01 1.54380992e-01 -1.34830177e+00 7.25291312e-01 -1.84261531e-01 -1.38719094e+00 8.04321110e-01 -6.79889321e-01 6.28275216e-01 -4.43083137e-01 1.78283930e-01 1.17175746e+00 8.72103274e-01 -1.23385417e+00 6.44812346e-01 6.44290328e-01 8.69172156e-01 -8.38346720e-01 9.57283437e-01 2.18211442e-01 -1.06115031e+00 -4.63922888e-01 -6.60265803e-01 -4.07016128e-01 -9.34255347e-02 3.23400825e-01 -1.12774026e+00 8.28530908e-01 6.40985727e-01 2.57495362e-02 -6.84759319e-01 4.18134898e-01 -4.34864849e-01 4.83849943e-01 -3.04856718e-01 -2.09123850e-01 3.46946388e-01 -7.70087615e-02 -2.80508131e-01 8.51826370e-01 -2.08761111e-01 6.66329145e-01 -1.29237577e-01 1.03931093e+00 -5.31272829e-01 1.30910531e-01 2.50487067e-02 -2.13953152e-01 3.08751941e-01 1.28249753e+00 -3.92626449e-02 -7.09232569e-01 -4.75940913e-01 5.35461128e-01 8.89071763e-01 2.10311264e-01 -1.05376208e+00 -2.22785041e-01 5.04435837e-01 -1.26218870e-01 1.72182187e-01 1.70896262e-01 -5.58324456e-01 -1.28697681e+00 1.98444068e-01 -1.31275439e+00 1.19507253e+00 -1.12186348e+00 -1.57115579e+00 2.24336594e-01 5.34189381e-02 -5.67619324e-01 -4.89516556e-01 -9.16241586e-01 -3.17004621e-01 7.94246018e-01 -1.89826918e+00 -1.44731057e+00 -4.70047921e-01 7.89429069e-01 -2.45986938e-01 2.03506067e-01 1.01348352e+00 -1.69241782e-02 2.30691470e-02 8.57376277e-01 -4.55799818e-01 6.43963516e-01 4.63499278e-01 -1.60354519e+00 1.52220622e-01 2.73299590e-02 -1.02170601e-01 8.01420271e-01 5.49079597e-01 -3.20096850e-01 -1.71808314e+00 -1.00230098e+00 5.77925682e-01 -9.42359388e-01 1.01653850e+00 -2.31946707e-01 -1.00983298e+00 7.07150400e-01 -2.84513831e-01 3.27105999e-01 7.95655966e-01 4.58156854e-01 -1.10618043e+00 -2.19058305e-01 -1.20357800e+00 5.99857390e-01 7.18617141e-01 -7.88534045e-01 -1.26939666e+00 2.14904621e-01 1.13228261e+00 -6.34635508e-01 -1.18374610e+00 7.25498796e-01 4.74498361e-01 -3.56449008e-01 1.34191060e+00 -1.27242160e+00 8.42566073e-01 -2.76754498e-01 -1.99052677e-01 -9.60213304e-01 1.13783993e-01 3.83985043e-02 -6.73789620e-01 8.97043705e-01 4.20529991e-01 -2.60428727e-01 1.03427947e+00 1.07007647e+00 2.13391542e-01 -7.71023273e-01 -9.56425965e-01 -7.04608321e-01 5.90112209e-01 -4.48630303e-01 1.05880380e+00 1.11627519e+00 5.67053080e-01 5.05763412e-01 1.26314327e-01 3.30682218e-01 -5.73919751e-02 4.05276954e-01 8.59254599e-01 -1.15851295e+00 -3.30655068e-01 -4.05926377e-01 -3.00173342e-01 -1.02690780e+00 3.39582294e-01 -9.06915843e-01 -1.75760746e-01 -1.77591860e+00 1.33014068e-01 -5.68708658e-01 -3.12397093e-01 1.05958748e+00 -4.15458858e-01 8.56065378e-02 -7.67646416e-04 -4.24700588e-01 -1.05156493e+00 7.34954953e-01 1.35314643e+00 -2.55869508e-01 -8.46752301e-02 -3.11086982e-01 -9.75064337e-01 6.24578595e-01 6.56135023e-01 2.32899673e-02 -6.67817414e-01 -3.81710678e-01 1.19893515e+00 3.93034667e-01 6.09911382e-01 -6.53788924e-01 2.74422079e-01 -5.54034412e-01 1.13700166e-01 -5.30860424e-01 3.14144105e-01 -8.55715692e-01 -7.85072207e-01 5.49718253e-02 -6.54169261e-01 1.86139010e-02 4.78120774e-01 2.71634728e-01 -3.79453868e-01 -2.03902915e-01 4.62150723e-01 -1.34026960e-01 -6.34294987e-01 8.21127519e-02 1.21656924e-01 5.98337173e-01 5.29523671e-01 3.67645532e-01 -6.22525275e-01 -5.52082956e-01 -3.88156295e-01 6.67564094e-01 3.51971686e-02 1.28335655e-01 5.17168939e-01 -1.44779634e+00 -6.45813525e-01 -2.23181769e-01 5.67487538e-01 5.35965800e-01 2.27702856e-01 3.17729175e-01 -7.16288090e-01 7.32481778e-01 -2.48205453e-01 8.75416994e-02 -6.80707753e-01 8.52898717e-01 5.40769160e-01 -8.06306064e-01 6.66431570e-03 8.30341935e-01 -4.28149402e-02 -1.25473785e+00 2.61225939e-01 -1.05953622e+00 -4.60704774e-01 -1.83779970e-01 5.32609880e-01 2.78541982e-01 1.72092140e-01 -1.38605097e-02 -3.17237139e-01 4.49662581e-02 -6.12460002e-02 -5.47941364e-02 1.09318233e+00 3.57212901e-01 -4.33932334e-01 2.35900745e-01 8.23564708e-01 -1.35910302e-01 -7.80811608e-01 -4.51457381e-01 2.03070983e-01 7.72108957e-02 -3.30562502e-01 -1.39572692e+00 -6.40287995e-01 6.73480093e-01 -3.62888612e-02 -1.15639269e-02 1.05377519e+00 1.80445567e-01 8.90984476e-01 1.31144083e+00 2.54798472e-01 -9.56112564e-01 3.66361916e-01 9.17293072e-01 1.02841806e+00 -1.36556828e+00 -2.19976127e-01 -3.77562553e-01 -6.46226406e-01 1.18940854e+00 1.17350507e+00 1.03248805e-01 1.21601313e-01 7.09209889e-02 1.96318969e-01 -2.53531277e-01 -1.03574944e+00 -1.68688998e-01 7.02732503e-01 4.01873440e-01 3.58299017e-01 -3.01885083e-02 -4.47767042e-02 1.24375212e+00 -7.90246248e-01 2.21239924e-01 4.48486418e-01 7.23246753e-01 -2.99106658e-01 -6.95539594e-01 -5.20165265e-01 4.90109116e-01 -1.72391132e-01 -4.58312273e-01 -4.34113383e-01 5.75638771e-01 -3.52899432e-01 7.76873529e-01 -2.34773420e-02 -4.71856259e-02 4.00729507e-01 3.05122524e-01 6.16298199e-01 -4.54598635e-01 -6.27058446e-01 -1.00846362e+00 3.88236135e-01 -3.57996374e-01 -1.02802418e-01 -1.83396551e-04 -1.87049806e+00 -3.12362611e-01 -2.81649739e-01 5.38912654e-01 1.43338367e-01 8.47127140e-01 3.31663080e-02 4.44617957e-01 2.02466652e-01 3.59089911e-01 -1.23811555e+00 -7.85436690e-01 -2.93227404e-01 6.03305101e-01 2.97399193e-01 -3.50787431e-01 -2.39695325e-01 -1.00985229e-01]
[10.413310050964355, 7.90489387512207]
663a01bb-f12a-49e2-a30f-fa9fa5a9dee6
cleaning-noisy-and-heterogeneous-metadata-for
1906.08470
null
https://arxiv.org/abs/1906.08470v1
https://arxiv.org/pdf/1906.08470v1.pdf
Cleaning Noisy and Heterogeneous Metadata for Record Linking Across Scholarly Big Datasets
Automatically extracted metadata from scholarly documents in PDF formats is usually noisy and heterogeneous, often containing incomplete fields and erroneous values. One common way of cleaning metadata is to use a bibliographic reference dataset. The challenge is to match records between corpora with high precision. The existing solution which is based on information retrieval and string similarity on titles works well only if the titles are cleaned. We introduce a system designed to match scholarly document entities with noisy metadata against a reference dataset. The blocking function uses the classic BM25 algorithm to find the matching candidates from the reference data that has been indexed by ElasticSearch. The core components use supervised methods which combine features extracted from all available metadata fields. The system also leverages available citation information to match entities. The combination of metadata and citation achieves high accuracy that significantly outperforms the baseline method on the same test dataset. We apply this system to match the database of CiteSeerX against Web of Science, PubMed, and DBLP. This method will be deployed in the CiteSeerX system to clean metadata and link records to other scholarly big datasets.
['Jian Wu', 'Cornelia Caragea', 'Athar Sefid', 'Prasenjit Mitra', 'Lu Liu', 'Jing Zhao', 'C. Lee Giles', 'Allen C. Ge']
2019-06-20
null
null
null
null
['record-linking']
['natural-language-processing']
[-4.23436791e-01 -3.91234048e-02 -5.45891285e-01 -4.43221107e-02 -1.82661486e+00 -8.73035669e-01 7.09043443e-01 8.21398377e-01 -6.13549471e-01 1.12037969e+00 5.96156001e-01 4.39746454e-02 -6.09196603e-01 -8.09251428e-01 -9.02251422e-01 -2.80431449e-01 4.13712770e-01 6.40868306e-01 2.30797917e-01 2.20513776e-01 8.92650843e-01 3.97970974e-01 -1.56696939e+00 4.43753004e-01 1.01150084e+00 6.39405847e-01 3.70161891e-01 2.79858589e-01 -7.76033938e-01 5.73754013e-01 -8.52698207e-01 -5.46685278e-01 2.25667521e-01 4.25332859e-02 -9.19566870e-01 -1.01367211e+00 8.73480678e-01 5.29932618e-01 -4.50654298e-01 1.21003306e+00 5.99708319e-01 -2.02920899e-01 4.23069984e-01 -8.86951625e-01 -6.70018196e-01 1.12648749e+00 -3.63572687e-01 3.94181013e-01 7.71545053e-01 -3.90221030e-01 9.89830673e-01 -7.78917551e-01 1.63287926e+00 1.10995436e+00 8.52263391e-01 -4.67536338e-02 -1.16490877e+00 -6.11793458e-01 -6.36471212e-01 3.32687169e-01 -1.42722476e+00 -6.95467770e-01 8.22715014e-02 -5.27518809e-01 8.73987556e-01 5.66001415e-01 6.62514791e-02 9.24071312e-01 2.73001671e-01 2.07952440e-01 9.83704627e-01 -5.65174222e-01 3.26781869e-01 9.03228894e-02 6.99846804e-01 2.18412489e-01 8.69179130e-01 -3.63180012e-01 -7.72371650e-01 -6.45336390e-01 -2.92429715e-01 1.41192749e-01 -2.75416791e-01 2.66695350e-01 -1.43318224e+00 1.57706112e-01 -2.91190762e-02 5.96518457e-01 -6.54606640e-01 -3.54950726e-01 5.08976519e-01 3.82373333e-01 1.73394799e-01 1.27606058e+00 -3.38221669e-01 -5.12482934e-02 -1.67343473e+00 5.82232773e-01 1.20223868e+00 1.29826999e+00 4.80608791e-01 -1.07686603e+00 -6.05719745e-01 6.51778340e-01 4.07799691e-01 4.45100904e-01 4.65955824e-01 -1.23856592e+00 3.99508029e-01 7.80537486e-01 1.41025245e-01 -8.98620725e-01 -3.28338474e-01 -3.11741829e-01 -2.65348315e-01 -1.38679534e-01 4.43987697e-01 3.74090314e-01 -9.92455900e-01 1.22154975e+00 4.01869386e-01 -8.70092884e-02 -9.18673631e-03 4.98595327e-01 1.49921751e+00 3.80803615e-01 3.97788703e-01 -2.35211790e-01 1.33779705e+00 -2.78606713e-01 -1.35186565e+00 7.09825516e-01 7.12144732e-01 -1.31118405e+00 4.68202263e-01 2.44196132e-01 -1.14296520e+00 -5.65827973e-02 -9.38425779e-01 -4.93703723e-01 -1.18422210e+00 -8.18039700e-02 2.00641885e-01 1.09562270e-01 -8.93894315e-01 8.94966722e-01 -5.99653721e-01 -5.55623233e-01 4.98280883e-01 1.15057975e-01 -6.37918472e-01 -3.66363168e-01 -1.09235847e+00 1.06075668e+00 3.68473232e-01 -7.13614643e-01 -1.54791325e-01 -1.55357814e+00 -4.75272983e-01 2.97216505e-01 3.09964389e-01 -7.16989696e-01 9.27856684e-01 -8.54398608e-02 -5.73181391e-01 9.68247175e-01 -1.35105461e-01 -4.79513586e-01 4.15509582e-01 -1.54106840e-01 -6.63714170e-01 1.14202961e-01 8.21770549e-01 6.59732744e-02 -3.77737135e-02 -1.13956082e+00 -7.51258254e-01 -4.47354496e-01 -7.37788558e-01 -3.81902069e-01 -8.77987407e-03 2.27372214e-01 -9.53209579e-01 -7.92142689e-01 5.83530143e-02 -5.99005699e-01 1.44035622e-01 -4.39100683e-01 -7.65407085e-01 -4.51073766e-01 4.78768259e-01 -1.00509226e+00 1.67529953e+00 -1.56176674e+00 2.11174339e-02 5.72375953e-01 3.53330523e-01 -4.85092066e-02 1.83398202e-02 6.95570350e-01 1.30837023e-01 4.92645055e-01 2.31673628e-01 -6.55637607e-02 6.17529713e-02 -1.32199422e-01 -3.60635072e-01 4.92773533e-01 -3.47167462e-01 7.78168023e-01 -9.94634032e-01 -9.24241126e-01 -3.81179303e-01 2.37842813e-01 -3.01080734e-01 1.99507177e-01 -2.56070465e-01 -1.51906088e-01 -4.50511456e-01 9.61304307e-01 7.20889866e-01 -3.02231401e-01 8.74462873e-02 -4.98913348e-01 -4.25238878e-01 6.98778212e-01 -1.34999764e+00 1.96545410e+00 2.53617883e-01 5.32961965e-01 3.33143882e-02 -4.49844539e-01 7.06486762e-01 1.13004759e-01 8.29981744e-01 -6.31567419e-01 -1.24130033e-01 6.15585327e-01 -4.69077408e-01 -6.75788879e-01 6.30177319e-01 6.53853655e-01 -2.43764035e-02 2.93580890e-01 3.45340937e-01 8.25527608e-02 8.88059556e-01 8.06457579e-01 1.79362583e+00 9.86467153e-02 9.16589946e-02 -5.29003859e-01 3.35945338e-01 4.74131286e-01 5.72964370e-01 1.17330480e+00 4.03929383e-01 3.95337820e-01 2.72583723e-01 -1.09567801e-02 -1.08196568e+00 -7.81141639e-01 -6.63514316e-01 6.38715804e-01 -6.34499192e-02 -7.96153605e-01 -6.00338399e-01 -3.80623430e-01 6.86984599e-01 7.19182193e-01 -5.88554621e-01 1.14710905e-01 -2.31632590e-01 -5.19226015e-01 5.92738688e-01 -1.18833363e-01 -1.59046888e-01 -8.14833760e-01 -1.50215775e-01 2.18044430e-01 -1.45719633e-01 -6.95732772e-01 -2.12333232e-01 9.81480777e-02 -2.97862947e-01 -1.32036281e+00 -5.44099092e-01 -3.89627963e-01 3.90512973e-01 -2.67950352e-02 1.31448436e+00 2.62433052e-01 -3.93901259e-01 3.01732928e-01 -2.22625300e-01 -4.73830253e-01 -4.76063728e-01 5.60152292e-01 6.45650923e-02 -7.66035974e-01 9.40693796e-01 -2.04839960e-01 -2.97546029e-01 -2.20365241e-01 -1.05983508e+00 -4.73329842e-01 5.32292247e-01 7.82855570e-01 1.10656881e+00 -3.73529881e-01 7.24116862e-01 -1.02169883e+00 4.64508802e-01 -1.01942682e+00 -7.19543755e-01 7.64662206e-01 -1.23507106e+00 4.44612235e-01 3.73845279e-01 8.74083787e-02 -9.24121201e-01 -1.24293320e-01 -3.29165868e-02 -2.45696694e-01 -8.38390067e-02 9.30855632e-01 -1.83899283e-01 1.18658029e-01 4.35244977e-01 -4.61852312e-01 -3.39694649e-01 -1.35328233e+00 3.40437770e-01 1.16130066e+00 1.13552070e+00 -6.23531401e-01 7.12858379e-01 2.33365655e-01 5.78206219e-02 -2.28823408e-01 -9.11737025e-01 -8.53415489e-01 -5.26463747e-01 1.79627627e-01 5.51780403e-01 -9.73633528e-01 -6.50724173e-01 -2.36191452e-01 -1.03237903e+00 5.78431666e-01 -2.70571947e-01 6.17792726e-01 1.09177351e-01 1.35062307e-01 -5.72356761e-01 -9.57610831e-02 -6.02021098e-01 -7.80804873e-01 1.09780490e+00 4.31134194e-01 -3.82274866e-01 -2.94239193e-01 6.31155372e-01 4.10147518e-01 2.88096815e-01 1.76348478e-01 8.82537782e-01 -1.36532021e+00 -7.12143064e-01 -5.07904530e-01 -1.49738565e-01 -5.63262701e-01 2.78626718e-02 5.33877373e-01 -7.33234704e-01 2.83615496e-02 -4.54858035e-01 7.27636218e-02 9.93181229e-01 3.24887633e-01 1.21259797e+00 -3.37228000e-01 -8.22142899e-01 9.08955812e-01 1.62238300e+00 1.19050615e-01 7.24362731e-01 1.26185393e+00 6.23589516e-01 4.08266515e-01 4.61476892e-01 2.17173591e-01 2.64093280e-01 4.48528945e-01 -8.57243687e-02 1.40301704e-01 -4.12715487e-02 -1.95924670e-01 -5.17829895e-01 8.66288245e-01 1.78085536e-01 -2.54095614e-01 -1.21562290e+00 6.25215054e-01 -1.61825562e+00 -1.44302058e+00 -3.69679093e-01 2.11185408e+00 1.50609899e+00 -4.77528535e-02 -3.89894217e-01 -3.48532706e-01 6.68817520e-01 -5.00642359e-01 -1.81159481e-01 -8.62667058e-03 -5.69592357e-01 4.01458174e-01 1.05349171e+00 2.17210308e-01 -8.72828126e-01 5.65992177e-01 6.32362556e+00 7.93890476e-01 -6.84079587e-01 5.35570756e-02 1.68801636e-01 -5.15824854e-01 -5.14690876e-01 1.96104705e-01 -1.35156512e+00 1.06714916e+00 1.42674267e+00 -8.36911261e-01 7.94062316e-02 7.95956552e-01 4.93416116e-02 -2.58931458e-01 -1.08489752e+00 7.37471521e-01 -1.54535487e-01 -2.28796482e+00 -7.25661665e-02 2.00740248e-02 6.02634251e-01 2.98797965e-01 -3.24644715e-01 3.15919459e-01 5.37672639e-01 -9.72102225e-01 5.23701847e-01 1.34925306e+00 7.20927715e-01 -5.21023929e-01 9.06459868e-01 -1.49812520e-01 -4.32554871e-01 4.32263017e-01 -4.18181807e-01 7.83925414e-01 -1.85358137e-01 9.59552646e-01 -4.80075985e-01 6.85920656e-01 1.41079509e+00 8.07815254e-01 -9.54107046e-01 1.69292295e+00 4.08497602e-01 3.17054868e-01 -4.45555538e-01 1.87865824e-01 -2.86413133e-01 -8.30124766e-02 6.73794210e-01 1.62401235e+00 5.56104600e-01 -5.58532737e-02 -1.44465044e-01 8.67368460e-01 -8.99799824e-01 4.19473320e-01 -4.73671108e-01 -8.69414806e-02 1.32968020e+00 1.34548450e+00 -3.47749621e-01 -6.16930366e-01 -3.37521851e-01 1.09957635e-01 3.47419083e-01 9.18114930e-02 -2.65996337e-01 -9.99019146e-01 5.44927537e-01 1.32148620e-02 1.78862691e-01 4.58343118e-01 -3.50196272e-01 -1.08988273e+00 5.48496507e-02 -1.05558181e+00 8.13772440e-01 -6.99557602e-01 -1.66447759e+00 2.85552233e-01 -1.69709966e-01 -1.02708125e+00 -6.67098984e-02 -1.34591043e-01 -2.81538755e-01 1.04280150e+00 -1.34674811e+00 -6.90165162e-01 -3.07216585e-01 1.10804252e-01 5.56957126e-02 -1.68601096e-01 7.39473283e-01 7.66906381e-01 -6.26126170e-01 5.88982761e-01 1.10340834e+00 1.97675005e-01 1.62825251e+00 -1.49877727e+00 6.39001057e-02 6.85623407e-01 -1.53837845e-01 1.38214827e+00 8.05940390e-01 -1.39376569e+00 -1.89166725e+00 -1.15702641e+00 1.31267774e+00 -9.86417472e-01 8.36817026e-01 1.99536122e-02 -1.63892698e+00 3.59809011e-01 5.55265188e-01 -1.84266776e-01 9.78098452e-01 -3.79052907e-02 -4.09030885e-01 2.85727326e-02 -1.22164333e+00 1.23945720e-01 7.84154594e-01 -3.41613621e-01 -1.15914631e+00 5.91926277e-01 4.83203113e-01 -4.82544124e-01 -1.74366772e+00 2.41275147e-01 5.12691855e-01 -9.01203137e-03 1.02950931e+00 -9.66571689e-01 4.26373363e-01 -5.49960613e-01 -1.11123353e-01 -1.08437431e+00 -3.75811219e-01 -7.87676930e-01 -2.72247136e-01 1.92603481e+00 4.26394850e-01 -2.00609222e-01 2.31951281e-01 8.27015519e-01 -1.51238680e-01 -1.78705454e-01 -1.00689054e+00 -7.45858371e-01 1.31700501e-01 2.43613735e-01 8.84963870e-01 1.23797894e+00 3.35253716e-01 1.92565098e-01 2.93018758e-01 -9.22182575e-03 7.83832014e-01 3.96628439e-01 6.75382257e-01 -1.78627181e+00 3.15352827e-01 -5.09179175e-01 -1.86576456e-01 7.73772821e-02 1.90036044e-01 -1.35963237e+00 -1.11846119e-01 -1.77756298e+00 4.69631732e-01 -6.89198017e-01 -5.36071062e-01 3.80559862e-01 -4.44970250e-01 -1.14035293e-01 -3.27682823e-01 9.92536604e-01 -8.40764403e-01 -1.29229844e-01 3.97950858e-01 -6.45430982e-02 -1.22372709e-01 -8.33027899e-01 -7.69680679e-01 3.91234070e-01 2.37198204e-01 -1.20176089e+00 6.63023710e-01 -1.34437099e-01 4.16853875e-01 -2.99007654e-01 -7.21562281e-02 -1.01013446e+00 9.50969756e-01 -1.86779767e-01 6.58982038e-01 -1.11123466e+00 -5.71770728e-01 -6.07839584e-01 7.16906905e-01 -7.62664713e-03 -9.20032978e-01 2.33022884e-01 9.42122191e-02 6.58680797e-01 -1.05506860e-01 -2.93777764e-01 6.61067545e-01 -1.89997375e-01 -3.94232661e-01 -5.77472933e-02 -2.02887207e-01 2.90343404e-01 5.27965665e-01 4.44130659e-01 -1.07891941e+00 4.44391340e-01 -3.44716221e-01 4.27260637e-01 9.00457799e-01 4.38500226e-01 -8.19621887e-03 -1.18018365e+00 -8.01912069e-01 -3.89086932e-01 2.40850091e-01 -4.38411593e-01 -4.04450327e-01 9.33756530e-01 -3.93644452e-01 7.47764111e-01 -3.02855402e-01 -2.44442701e-01 -1.29398918e+00 8.50526035e-01 -1.78817198e-01 -2.14424804e-01 -5.55716753e-01 3.73303801e-01 -8.10404181e-01 -1.75031587e-01 5.38320124e-01 5.38836606e-02 -4.03113782e-01 5.10188878e-01 8.45014513e-01 5.53574026e-01 7.74191439e-01 -4.53838140e-01 -5.88579178e-01 2.08379984e-01 -2.40275875e-01 1.17124930e-01 1.79846394e+00 -8.71553496e-02 -5.74803054e-01 2.93164074e-01 1.33344364e+00 6.19558573e-01 1.02040665e-02 -5.20067036e-01 1.02068532e+00 -4.78284329e-01 3.03867698e-01 -9.83891428e-01 -7.86498725e-01 3.17901336e-02 6.48501873e-01 1.13420472e-01 5.00504911e-01 2.31492147e-01 4.56714392e-01 6.37709558e-01 -8.10708967e-04 -1.38186741e+00 -9.23797250e-01 2.96213448e-01 6.24656558e-01 -1.26958227e+00 4.76702690e-01 4.95431833e-02 1.65071845e-01 1.16872382e+00 2.71485567e-01 2.23200545e-01 6.07257724e-01 5.06712198e-01 -1.31979091e-02 -7.12622046e-01 -7.71019757e-01 -1.00964487e-01 6.62576854e-01 1.55741572e-01 5.13122499e-01 -4.18740541e-01 -9.05374527e-01 7.72443533e-01 -8.37492868e-02 2.89247662e-01 3.90422165e-01 1.14412725e+00 -3.42852086e-01 -1.16477168e+00 -7.22183526e-01 1.27996337e+00 -1.41028404e+00 -2.09754601e-01 -5.53207636e-01 5.62952399e-01 -3.13773640e-02 6.81161523e-01 1.08247995e-02 1.77678972e-01 3.49483818e-01 3.30189139e-01 -8.80103782e-02 -5.37787557e-01 -9.95832145e-01 7.55976886e-02 1.54689535e-01 -6.73388481e-01 -4.61978525e-01 -9.35349822e-01 -1.30598545e+00 -4.95304704e-01 -2.90903211e-01 7.33463883e-01 9.73848164e-01 6.86217129e-01 9.83193994e-01 3.02201658e-01 5.54350205e-02 -2.79114932e-01 -3.63142878e-01 -8.56262088e-01 -2.47331172e-01 7.41657555e-01 2.08385304e-01 -6.87908590e-01 -4.27652955e-01 4.60760325e-01]
[9.452275276184082, 8.38356876373291]
d54cc66b-58a8-46de-9bb7-aafaf16ac898
motion-capture-from-pan-tilt-cameras-with
1908.11676
null
https://arxiv.org/abs/1908.11676v1
https://arxiv.org/pdf/1908.11676v1.pdf
Motion Capture from Pan-Tilt Cameras with Unknown Orientation
In sports, such as alpine skiing, coaches would like to know the speed and various biomechanical variables of their athletes and competitors. Existing methods use either body-worn sensors, which are cumbersome to setup, or manual image annotation, which is time consuming. We propose a method for estimating an athlete's global 3D position and articulated pose using multiple cameras. By contrast to classical markerless motion capture solutions, we allow cameras to rotate freely so that large capture volumes can be covered. In a first step, tight crops around the skier are predicted and fed to a 2D pose estimator network. The 3D pose is then reconstructed using a bundle adjustment method. Key to our solution is the rotation estimation of Pan-Tilt cameras in a joint optimization with the athlete pose and conditioning on relative background motion computed with feature tracking. Furthermore, we created a new alpine skiing dataset and annotated it with 2D pose labels, to overcome shortcomings of existing ones. Our method estimates accurate global 3D poses from images only and provides coaches with an automatic and fast tool for measuring and improving an athlete's performance.
['Jörg Spörri', 'Pascal Fua', 'Roman Bachmann', 'Helge Rhodin']
2019-08-30
null
null
null
null
['markerless-motion-capture']
['computer-vision']
[-1.08709857e-01 -9.41333398e-02 -2.59788960e-01 -5.08740768e-02 -6.39627695e-01 -7.58533716e-01 3.47222351e-02 -6.00520521e-02 -7.84424245e-01 3.75832230e-01 1.40156748e-03 4.07467902e-01 7.51483664e-02 -4.37431753e-01 -7.65378475e-01 -4.40264046e-01 6.99934289e-02 8.69117439e-01 5.82782567e-01 -3.09621811e-01 2.54964501e-01 5.79948723e-01 -1.56566060e+00 -4.16675508e-01 3.61854583e-01 5.90809226e-01 2.54622281e-01 1.01598442e+00 4.73035306e-01 5.36986768e-01 -5.16075075e-01 -4.70994145e-01 5.32575667e-01 -3.96083266e-01 -4.14108276e-01 4.15499091e-01 8.35467339e-01 -1.70852825e-01 -3.26188445e-01 6.91833079e-01 5.82604468e-01 3.41009915e-01 6.49013445e-02 -9.16888416e-01 5.81716955e-01 4.42198403e-02 -6.64509296e-01 2.47076303e-02 6.67969882e-01 1.80877686e-01 6.65306509e-01 -6.22303843e-01 9.87999976e-01 8.48951161e-01 1.09038997e+00 5.16560793e-01 -1.17494988e+00 -5.15387237e-01 -1.22159496e-01 2.62224317e-01 -1.45249438e+00 -3.61026317e-01 1.06202662e+00 -6.09921575e-01 2.44656563e-01 3.38989615e-01 1.58130610e+00 7.43224204e-01 1.94319472e-01 6.19899273e-01 9.16077554e-01 -3.14057499e-01 3.39793116e-02 -2.29832128e-01 -1.86924860e-01 5.81035852e-01 2.59311974e-01 -8.93097892e-02 -9.93383169e-01 1.40360996e-01 1.16331458e+00 1.43475711e-01 -1.60189435e-01 -1.04835844e+00 -1.30686033e+00 6.02172196e-01 1.12088926e-01 -3.91555458e-01 -3.12283039e-01 5.78771710e-01 3.64812315e-01 -1.96621060e-01 3.18248779e-01 4.61603999e-01 -1.48535907e-01 -8.25258195e-01 -1.01059008e+00 6.29067659e-01 7.25140929e-01 5.85978806e-01 6.06049001e-01 -2.29802385e-01 4.43992168e-01 5.68922758e-01 2.12822065e-01 4.97184724e-01 3.02843601e-01 -1.25659132e+00 5.18850327e-01 6.02639198e-01 1.22706763e-01 -1.16383135e+00 -6.00330770e-01 -5.16619563e-01 -2.07913309e-01 2.72808522e-01 7.67793417e-01 -7.88032040e-02 -5.30969083e-01 1.29447532e+00 6.80270851e-01 2.47446373e-01 -2.98806399e-01 1.60442746e+00 2.49163672e-01 2.78160367e-02 -4.19783503e-01 8.02872851e-02 1.41260314e+00 -9.50686991e-01 -4.49935645e-01 -6.87045813e-01 6.02325737e-01 -8.65482211e-01 7.98544884e-01 6.50406420e-01 -1.15276110e+00 -4.86483365e-01 -9.45161760e-01 -8.37939978e-03 2.65308470e-01 1.41324013e-01 4.33675170e-01 7.55471587e-01 -3.80890340e-01 5.99724829e-01 -1.21628416e+00 -1.54909343e-01 -2.00355247e-01 4.96112645e-01 -6.89061582e-01 1.97445095e-01 -8.85399401e-01 1.25920117e+00 1.34801954e-01 6.23046935e-01 -7.82168984e-01 -4.30509865e-01 -1.02955246e+00 -4.31399018e-01 9.32933390e-01 -6.49197578e-01 1.18664694e+00 -6.59235656e-01 -1.88573039e+00 9.64419127e-01 2.32587725e-01 -3.03582489e-01 1.00504923e+00 -9.26050723e-01 1.90007165e-01 3.60474735e-01 -3.43725877e-03 1.35632217e-01 5.53498626e-01 -1.13682544e+00 -2.91134357e-01 -5.77338696e-01 -1.17029198e-01 8.16273987e-01 1.91010073e-01 -2.43082479e-01 -9.91725445e-01 -3.52068841e-01 7.36704946e-01 -1.38150477e+00 -5.62623739e-01 1.86779946e-01 -2.26149589e-01 4.73582625e-01 6.49810553e-01 -7.79761910e-01 9.09304082e-01 -1.85003018e+00 4.80957448e-01 3.72053832e-01 3.70834291e-01 2.55452573e-01 4.51354325e-01 2.12569833e-01 2.61099607e-01 -5.13724864e-01 9.62684825e-02 -3.20163667e-01 -3.51930588e-01 3.78744930e-01 4.87355083e-01 9.75527763e-01 -5.11717796e-01 4.74748075e-01 -8.65529537e-01 -7.99927950e-01 4.77278054e-01 4.67495978e-01 -5.39329588e-01 1.35791942e-01 1.89668238e-01 7.21714377e-01 -3.76245260e-01 4.99459177e-01 4.04726774e-01 2.79856831e-01 2.84959316e-01 -1.97564498e-01 -3.33515376e-01 3.38625684e-02 -1.83179486e+00 2.27333069e+00 -4.12540942e-01 4.79396850e-01 4.50680792e-01 -8.99451196e-01 8.74358118e-01 3.93394470e-01 7.06968904e-01 -1.24888867e-01 1.95639774e-01 2.35972226e-01 -2.24523857e-01 -5.43530345e-01 6.91319346e-01 -3.01043332e-01 -2.79033870e-01 1.07836716e-01 -1.47504792e-01 -4.09689069e-01 4.44825552e-02 -1.07931577e-01 8.61457646e-01 7.96321929e-01 3.01948905e-01 1.22343138e-01 2.92766631e-01 2.43002996e-01 6.76224053e-01 3.54521841e-01 -1.00298502e-01 9.29143310e-01 2.45844319e-01 -6.75368965e-01 -1.09791672e+00 -9.12028015e-01 2.53348112e-01 4.79506493e-01 4.35265154e-01 -5.48959315e-01 -8.34222794e-01 -1.84443399e-01 2.24747509e-02 -2.37950251e-01 -3.31960082e-01 -1.80495054e-01 -1.12012446e+00 -3.39471310e-01 3.05430174e-01 5.66793501e-01 2.97020584e-01 -2.61777759e-01 -1.24860179e+00 3.52950454e-01 -4.19902116e-01 -1.09881449e+00 -6.58836722e-01 -5.88562489e-02 -9.05909717e-01 -1.30514002e+00 -7.60906696e-01 -5.06552100e-01 3.32267702e-01 1.76131129e-01 9.04212713e-01 -1.29231691e-01 -3.92427713e-01 3.27506512e-01 -2.63078690e-01 -1.85129791e-01 2.36804150e-02 -1.14746749e-01 4.06181097e-01 4.31287736e-02 -1.37407139e-01 -4.88359034e-01 -8.95281136e-01 6.63174272e-01 -2.90242761e-01 1.96857229e-01 4.76342112e-01 5.29798448e-01 9.48577106e-01 -4.82461065e-01 -4.03555244e-01 -6.85562432e-01 4.69794013e-02 1.81068212e-01 -7.07457066e-01 -1.91334054e-01 -1.66332096e-01 -3.23728859e-01 1.10771112e-01 -5.99838018e-01 -7.18228877e-01 1.00047576e+00 -2.26424590e-01 -6.27188623e-01 3.47046950e-03 2.00569317e-01 -9.56015736e-02 -3.05016190e-01 7.04077542e-01 -2.69475998e-03 5.20856738e-01 -6.92608356e-01 2.89892077e-01 2.71948040e-01 9.94204640e-01 -4.70814556e-01 7.98771143e-01 5.04893482e-01 3.27638865e-01 -7.22616255e-01 -8.62447619e-01 -9.16723013e-01 -1.33388710e+00 -9.61472154e-01 8.53944540e-01 -1.04067242e+00 -1.17707443e+00 3.15832257e-01 -9.09125149e-01 -4.01274227e-02 -4.27021891e-01 1.34196520e+00 -8.57366145e-01 6.88225806e-01 -4.97557819e-01 -7.61145949e-01 -1.43008560e-01 -1.13647091e+00 1.28531575e+00 1.29676446e-01 -3.85399103e-01 -7.33663619e-01 5.76871037e-01 9.07559931e-01 -1.63037151e-01 6.83237791e-01 -3.25769484e-01 -3.78653109e-02 -3.05784553e-01 -6.29493237e-01 5.84047496e-01 1.37531444e-01 -4.07844573e-01 -2.16346592e-01 -5.62803090e-01 -2.32935533e-01 -1.19222114e-02 -1.18685149e-01 2.46010706e-01 6.24768615e-01 7.87168562e-01 5.90453111e-02 -2.71316826e-01 9.14681733e-01 1.25316679e+00 -4.58118975e-01 4.29629445e-01 6.97141588e-01 9.69353735e-01 7.73809493e-01 1.05567181e+00 4.55565959e-01 4.36194748e-01 1.38329506e+00 6.33583367e-01 -6.08585365e-02 -1.21007927e-01 -4.05793786e-01 3.81486863e-01 9.84234691e-01 -8.09331834e-01 3.61843884e-01 -9.71456766e-01 3.59487474e-01 -1.89465356e+00 -7.47488618e-01 -6.08269751e-01 2.45697260e+00 6.20912492e-01 6.24042861e-02 5.29601157e-01 2.30688870e-01 6.96559787e-01 -1.45755425e-01 -4.56797242e-01 -9.44953263e-02 2.13162571e-01 -1.14973737e-02 1.04656768e+00 4.08727765e-01 -7.97658443e-01 8.94916892e-01 6.06040668e+00 4.03527677e-01 -9.27671134e-01 2.82140244e-02 -2.36611292e-01 -5.88592350e-01 2.61399955e-01 4.06875998e-01 -8.28433871e-01 2.50458747e-01 7.60200143e-01 2.43782565e-01 -1.72078293e-02 9.08090472e-01 2.64763564e-01 -4.68706161e-01 -7.27977335e-01 1.04870558e+00 3.84036005e-01 -1.15680742e+00 -7.12481320e-01 3.00575018e-01 3.58305067e-01 -1.90277487e-01 -4.38786566e-01 -2.07984462e-01 7.34509230e-02 -4.70105827e-01 1.00686896e+00 8.84367287e-01 3.89078796e-01 -7.44819582e-01 6.44990742e-01 5.97069085e-01 -1.23261821e+00 2.38987997e-01 -1.38725787e-01 -4.64987695e-01 5.07789671e-01 4.07457471e-01 -6.43484712e-01 3.55559140e-01 5.78909218e-01 5.99553883e-01 -4.90298450e-01 1.32108939e+00 -1.90263763e-01 5.12274086e-01 -5.98663568e-01 2.15522766e-01 -1.35611758e-01 -4.35574472e-01 8.35417032e-01 7.25041568e-01 3.29222023e-01 -1.41421914e-01 4.87452120e-01 1.81222767e-01 3.62510383e-01 2.40062177e-01 -3.39953721e-01 5.22738695e-01 3.61983180e-02 1.43510771e+00 -8.52240622e-01 -1.35833666e-01 -1.36593893e-01 8.70917201e-01 1.36880409e-02 -1.35750651e-01 -7.95463264e-01 -8.94815326e-02 5.46651125e-01 7.67491758e-01 -3.95051204e-02 -7.04366267e-01 -3.00414190e-02 -1.56330299e+00 2.72425532e-01 -7.65752792e-01 2.78448880e-01 -7.86170483e-01 -4.87285674e-01 9.29583609e-02 1.20538749e-01 -1.64611125e+00 -4.61083531e-01 -4.33139950e-01 -2.07485884e-01 5.53416193e-01 -8.31154883e-01 -1.11472297e+00 -3.74379456e-01 5.40978849e-01 5.34744084e-01 3.89377177e-01 3.67012501e-01 3.07173640e-01 -4.25408155e-01 9.02123228e-02 -3.28336179e-01 1.70367911e-01 8.91503155e-01 -1.23642027e+00 4.25729305e-02 8.30309510e-01 3.21660846e-01 2.79893398e-01 1.09831500e+00 -6.44978285e-01 -1.80136096e+00 -4.01812226e-01 5.72657704e-01 -7.27960229e-01 5.59853792e-01 -5.30350685e-01 -4.63342875e-01 8.00467312e-01 -5.17157018e-01 2.29509071e-01 6.27126515e-01 9.03258622e-02 2.84060597e-01 -3.42354506e-01 -5.36908269e-01 3.14550698e-01 9.96386468e-01 -1.65516824e-01 -5.16937673e-01 3.62207085e-01 -6.11823797e-02 -1.27516961e+00 -1.06931341e+00 -1.97849441e-02 1.02184772e+00 -7.16795683e-01 1.12230623e+00 -2.17944533e-01 -4.81532477e-02 -6.06078267e-01 2.80488968e-01 -1.04232121e+00 1.82748511e-01 -6.73543513e-01 6.56509846e-02 8.22442770e-01 -1.70045018e-01 -1.05874412e-01 1.45415235e+00 5.24713695e-01 -2.26561829e-01 -2.98186809e-01 -1.18120003e+00 -6.77927852e-01 -3.98471564e-01 -6.05236888e-01 1.45960003e-01 5.92522323e-01 -6.36232458e-03 9.01814401e-02 -9.96638596e-01 2.89090633e-01 9.59426045e-01 2.19689801e-01 1.55459404e+00 -1.44998074e+00 -4.64814335e-01 1.17697813e-01 -1.08378053e+00 -1.18681610e+00 -2.73642361e-01 -3.53551835e-01 2.57021338e-01 -1.07445288e+00 4.66674492e-02 -3.48420860e-03 3.08521479e-01 1.94039017e-01 1.18755378e-01 7.73490727e-01 1.74775437e-01 4.08356428e-01 -6.14564002e-01 4.58097905e-02 1.46484709e+00 5.80989838e-01 -1.56468883e-01 3.57289493e-01 -2.93403845e-02 1.05081940e+00 3.62458497e-01 -6.91099763e-01 -4.04684581e-02 -3.95264655e-01 5.62840819e-01 6.18374705e-01 7.88385034e-01 -1.31737900e+00 3.16578001e-01 -1.97596833e-01 3.84320676e-01 -5.84755540e-01 7.25942850e-01 -7.34121263e-01 6.80798650e-01 7.07574964e-01 -9.35554802e-02 1.42055631e-01 -2.66975790e-01 5.43061256e-01 -3.81276548e-01 -2.49748826e-01 5.81716537e-01 -4.76238698e-01 -6.39838517e-01 2.70202279e-01 -4.13630247e-01 -1.49038240e-01 1.14484918e+00 -7.98304796e-01 3.33048344e-01 -4.08624589e-01 -1.21993542e+00 1.87135994e-01 8.07477355e-01 1.99043438e-01 5.17856658e-01 -1.11265993e+00 -4.97537613e-01 1.53489903e-01 3.87808457e-02 3.85650694e-01 3.85039985e-01 1.32137918e+00 -1.24620211e+00 -7.29240254e-02 -3.71354610e-01 -1.05514300e+00 -1.58651745e+00 -2.64027677e-02 4.15245682e-01 -1.69549242e-01 -9.40791965e-01 5.12725651e-01 -3.49692553e-01 -4.01483536e-01 -1.77684858e-01 -4.31744903e-02 -2.43088320e-01 2.97367070e-02 1.07039481e-01 5.62711656e-01 8.95164311e-02 -1.15165520e+00 -2.68205792e-01 1.24776399e+00 2.84429848e-01 -3.19207817e-01 1.30283666e+00 -4.62057561e-01 2.13915169e-01 4.11794245e-01 8.80822003e-01 4.31483418e-01 -1.65667975e+00 3.25809531e-02 -2.70733893e-01 -8.83786976e-01 -2.99818683e-02 -9.53994319e-02 -1.06635141e+00 9.10619736e-01 6.69023037e-01 -3.56198251e-01 8.51748109e-01 -1.36061572e-02 7.94607401e-01 1.06089540e-01 2.75604725e-01 -1.49696040e+00 -1.48671418e-02 1.63534254e-01 5.85361362e-01 -9.60544169e-01 4.75318670e-01 -4.63416070e-01 -6.86692655e-01 1.20860362e+00 5.37654161e-01 -3.75195950e-01 2.28689641e-01 2.51592427e-01 4.21435297e-01 -3.11102778e-01 -2.65508085e-01 -1.43923014e-01 4.91185576e-01 4.83326882e-01 2.80000865e-01 3.44971754e-02 -5.16564727e-01 2.36399081e-02 -5.80653906e-01 -6.97240904e-02 6.82079434e-01 1.19502425e+00 -3.60979944e-01 -1.22963190e+00 -7.03913391e-01 -1.98587142e-02 -5.46040714e-01 5.44391215e-01 -2.56302565e-01 1.00604391e+00 2.99028188e-01 5.03333926e-01 1.77178681e-02 -3.93807679e-01 6.78800523e-01 -9.37329903e-02 6.27395511e-01 -6.87077403e-01 -6.85415387e-01 5.01912236e-01 2.05245808e-01 -8.82478714e-01 -6.43543720e-01 -9.84389722e-01 -9.44031954e-01 -5.29982224e-02 -6.02463484e-01 2.25366533e-01 1.03280962e+00 7.71314681e-01 -1.57255590e-01 1.27220571e-01 3.13238740e-01 -1.34184563e+00 -2.32938692e-01 -6.42536581e-01 -7.81178057e-01 6.38800263e-01 1.68096483e-01 -9.79171932e-01 -8.40339437e-02 2.68994242e-01]
[7.207302093505859, -0.8509618639945984]
f0fc307e-1ed5-441c-b04d-ff6dd9d5f4d4
scan-a-spatial-context-attentive-network-for
2102.00109
null
https://arxiv.org/abs/2102.00109v2
https://arxiv.org/pdf/2102.00109v2.pdf
SCAN: A Spatial Context Attentive Network for Joint Multi-Agent Intent Prediction
Safe navigation of autonomous agents in human centric environments requires the ability to understand and predict motion of neighboring pedestrians. However, predicting pedestrian intent is a complex problem. Pedestrian motion is governed by complex social navigation norms, is dependent on neighbors' trajectories, and is multimodal in nature. In this work, we propose SCAN, a Spatial Context Attentive Network that can jointly predict socially-acceptable multiple future trajectories for all pedestrians in a scene. SCAN encodes the influence of spatially close neighbors using a novel spatial attention mechanism in a manner that relies on fewer assumptions, is parameter efficient, and is more interpretable compared to state-of-the-art spatial attention approaches. Through experiments on several datasets we demonstrate that our approach can also quantitatively outperform state of the art trajectory prediction methods in terms of accuracy of predicted intent.
['Cody Fleming', 'Jasmine Sekhon']
2021-01-29
null
null
null
null
['social-navigation']
['robots']
[-3.63514632e-01 -4.72788885e-02 -3.26986849e-01 -4.35606688e-01 -1.34767100e-01 -4.52389330e-01 7.97566473e-01 1.05567843e-01 -7.37641335e-01 8.49255800e-01 7.77404606e-01 -5.96440494e-01 -8.95338282e-02 -9.66826916e-01 -7.45544612e-01 -3.69261146e-01 -3.73269767e-01 3.65925461e-01 5.84461689e-01 -4.76044953e-01 3.52214307e-01 3.61410737e-01 -1.30662632e+00 -9.16043371e-02 8.42810750e-01 5.03111839e-01 1.32612184e-01 7.51213610e-01 3.84386033e-01 1.26286113e+00 6.51548579e-02 -3.78364265e-01 1.67139769e-01 -2.40629435e-01 -6.18825614e-01 -4.82484668e-01 3.21117967e-01 -3.85932654e-01 -9.19616878e-01 6.51960254e-01 1.38749942e-01 8.63329232e-01 8.47105920e-01 -1.44814527e+00 -9.88089979e-01 3.93050134e-01 -9.65504795e-02 4.36943799e-01 4.21580791e-01 5.68883181e-01 1.02257872e+00 -7.37885237e-01 6.82194710e-01 1.36358964e+00 7.42257714e-01 4.85223949e-01 -1.05666924e+00 -4.34932977e-01 7.98238337e-01 6.21934474e-01 -1.28193200e+00 -2.95037329e-01 4.94700372e-01 -5.27030587e-01 1.01583695e+00 1.60902843e-01 9.61873710e-01 1.29975402e+00 3.20012152e-01 6.85339868e-01 4.16773260e-01 2.46505618e-01 5.37694514e-01 -3.54874849e-01 1.24979891e-01 8.03185463e-01 1.44160360e-01 3.41718346e-01 -4.10709292e-01 -5.25131933e-02 4.29250211e-01 1.49371237e-01 -1.41853511e-01 -3.32571715e-01 -1.31525517e+00 9.76068258e-01 9.92976964e-01 -1.04999006e-01 -4.58826482e-01 6.68320537e-01 7.33087137e-02 -3.15181911e-01 5.02060890e-01 2.60667980e-01 -8.41087103e-02 -4.44767594e-01 -4.92095768e-01 6.53705776e-01 6.68052256e-01 1.02511406e+00 5.86122870e-01 -7.48292878e-02 -3.15813959e-01 2.76110798e-01 5.44807971e-01 7.67484307e-01 8.24498087e-02 -1.30104280e+00 4.46561724e-01 5.20024061e-01 5.51471412e-01 -1.31327164e+00 -6.89249873e-01 -2.25787535e-01 -7.43740261e-01 2.23937944e-01 4.98546124e-01 -4.43982571e-01 -7.57430077e-01 2.02891994e+00 2.43684515e-01 5.15642107e-01 -6.63102791e-02 1.16694975e+00 7.46549547e-01 7.27223575e-01 5.39182365e-01 4.19255465e-01 8.81384313e-01 -1.25213861e+00 -3.65015596e-01 -4.85271364e-01 6.84531510e-01 -1.34312794e-01 9.14874554e-01 -3.52139562e-01 -8.03319633e-01 -4.90821183e-01 -8.06003392e-01 -3.37678224e-01 -6.17394686e-01 -1.12627923e-01 6.55928016e-01 6.80989683e-01 -1.36999321e+00 2.80344695e-01 -8.97952914e-01 -6.99561119e-01 6.59512579e-01 3.01007539e-01 -2.39830598e-01 -8.43017474e-02 -9.40391719e-01 1.04786515e+00 1.53741658e-01 -2.22380146e-01 -7.26572156e-01 -7.72924304e-01 -1.09427178e+00 9.20684263e-03 2.09982246e-01 -1.06502438e+00 9.78899539e-01 -5.11781812e-01 -1.28438056e+00 1.17659077e-01 -6.21846259e-01 -6.52914286e-01 7.05558360e-01 -3.64095092e-01 -2.53303200e-01 -1.22550674e-01 4.99694347e-01 1.31992149e+00 1.67329416e-01 -1.06944704e+00 -9.89526331e-01 -1.53298661e-01 2.05968350e-01 4.34131891e-01 6.55909181e-02 -3.31212938e-01 -3.39263439e-01 -4.03748870e-01 -2.91254193e-01 -1.32432115e+00 -9.28814769e-01 1.42914474e-01 -3.85595977e-01 -2.18951523e-01 9.86995578e-01 -3.92171532e-01 9.42949057e-01 -1.98413444e+00 8.06107745e-02 2.56367683e-01 3.08111757e-01 1.09858043e-03 -2.68545896e-01 4.78267014e-01 4.93619651e-01 3.31565917e-01 -1.12470172e-01 -6.11471415e-01 2.31069118e-01 1.99124917e-01 -4.10263717e-01 5.45768142e-01 -6.64529949e-02 1.27313793e+00 -1.23614597e+00 -3.01767617e-01 5.55172265e-01 5.90004504e-01 -8.48438978e-01 -9.28917974e-02 -3.04012597e-01 9.23243284e-01 -6.27498627e-01 4.20886427e-01 3.60999584e-01 -2.48535395e-01 -1.32224381e-01 3.63399595e-01 -4.44707185e-01 1.52707502e-01 -5.49282789e-01 1.40489650e+00 -3.03765148e-01 1.11638379e+00 -5.62652707e-01 -4.43681300e-01 5.16445816e-01 -8.04453567e-02 3.43667299e-01 -7.75921404e-01 3.36678959e-02 -7.87852928e-02 -1.20680004e-01 -4.98194665e-01 8.73239279e-01 4.61561561e-01 -2.39064842e-01 6.01538837e-01 -5.35254478e-01 5.41328728e-01 2.13924646e-01 2.55333692e-01 1.08602381e+00 5.78482971e-02 3.07010949e-01 -3.38803887e-01 2.47956395e-01 2.38800496e-01 6.21714354e-01 1.19247806e+00 -6.41811728e-01 2.82909751e-01 2.18531951e-01 -8.77127469e-01 -1.19469774e+00 -1.19689393e+00 4.22687382e-01 1.30245101e+00 7.21455634e-01 -3.68329197e-01 -6.02448165e-01 -5.44265926e-01 -4.85528372e-02 1.12388885e+00 -8.38040233e-01 -1.08120209e-02 -9.17037487e-01 -3.81812483e-01 3.52708757e-01 9.01878655e-01 5.53872287e-01 -8.99471581e-01 -8.22987258e-01 7.18597323e-02 -4.17942524e-01 -1.26368773e+00 -6.95832491e-01 -6.25530243e-01 -2.73638844e-01 -1.12699533e+00 -5.40542781e-01 -4.74346250e-01 6.29173338e-01 6.50439382e-01 9.98404503e-01 -1.42816696e-02 1.59005046e-01 6.27698302e-01 -3.62202674e-01 -3.89804482e-01 8.40278715e-02 2.82072723e-01 1.66424140e-01 6.77907392e-02 6.00055575e-01 -6.13579214e-01 -1.13939834e+00 4.16190594e-01 -1.86203137e-01 2.63238847e-01 2.78112650e-01 4.85392839e-01 2.96939909e-01 -1.24112979e-01 4.26111042e-01 -4.40874904e-01 5.22730291e-01 -1.04601848e+00 -4.90812242e-01 4.39693267e-03 -2.23695934e-01 -1.83149084e-01 6.69527769e-01 -1.97832108e-01 -1.09200871e+00 1.44309059e-01 6.44147247e-02 1.03874348e-01 -4.73571181e-01 2.02163666e-01 5.47893047e-02 -2.61420190e-01 5.98626912e-01 1.60627261e-01 -2.29928553e-01 1.41681641e-01 8.02416384e-01 1.76352173e-01 4.09145921e-01 -2.87500650e-01 5.80148637e-01 8.77149880e-01 3.11246425e-01 -1.02128816e+00 -3.55141282e-01 -3.81683975e-01 -7.84327745e-01 -7.09773064e-01 1.10946763e+00 -1.07548273e+00 -1.21323252e+00 1.40522078e-01 -1.26319587e+00 -6.45982325e-01 -4.34762659e-03 6.37026429e-01 -5.93067467e-01 2.19891131e-01 -4.38576728e-01 -9.43881810e-01 2.33456567e-01 -1.20954907e+00 8.31575274e-01 2.68251598e-01 -5.20251036e-01 -1.26159465e+00 3.47314142e-02 3.09114307e-01 5.10778487e-01 1.87060952e-01 6.69240177e-01 -3.86938900e-01 -1.42113435e+00 3.03569883e-02 -3.89412254e-01 -5.72760999e-01 -4.56785262e-02 -1.96794689e-01 -4.95057374e-01 -1.58063732e-02 -7.65669346e-01 1.72917828e-01 1.07957864e+00 7.45535791e-01 7.71276832e-01 -6.09822631e-01 -8.53113592e-01 4.50208992e-01 1.05571151e+00 2.89505094e-01 5.81438184e-01 4.07943457e-01 9.19380546e-01 7.90041506e-01 6.82485223e-01 4.40866560e-01 1.25484204e+00 6.34506583e-01 7.61422038e-01 2.10949451e-01 1.42435431e-01 -4.60855663e-01 2.66679168e-01 8.11381191e-02 -3.61474484e-01 -7.18491077e-01 -1.04122460e+00 9.25187469e-01 -2.65502095e+00 -1.31376588e+00 -3.86055261e-01 1.73020601e+00 -1.46239996e-01 -2.45725006e-01 4.95284379e-01 -5.16137779e-01 4.51704711e-01 3.58261526e-01 -7.22230256e-01 -1.17707185e-01 -1.13252655e-01 -6.54220045e-01 8.77903998e-01 8.75155210e-01 -1.39389908e+00 1.30876505e+00 6.82241583e+00 3.18127334e-01 -5.63733757e-01 1.20943628e-01 7.67983198e-01 -2.88401008e-01 -4.89261866e-01 -1.42235085e-01 -6.45192862e-01 5.94489932e-01 8.30661058e-01 -5.68041578e-02 4.36730683e-01 9.02892411e-01 7.20700145e-01 -2.47284397e-01 -1.14555061e+00 5.94787598e-01 -6.72933459e-02 -1.69415808e+00 -1.73501484e-03 2.75561810e-01 8.13010216e-01 3.63578945e-01 4.27294850e-01 2.03221157e-01 7.82545507e-01 -1.30957472e+00 9.88302290e-01 8.97357285e-01 -1.27900496e-01 -9.50560808e-01 5.83643496e-01 5.54156840e-01 -1.40649664e+00 -4.76880044e-01 -2.20091775e-01 -5.48756182e-01 6.40328467e-01 4.47869562e-02 -6.41340554e-01 1.32379949e-01 8.12858701e-01 1.13305688e+00 -5.06611288e-01 1.18934679e+00 -1.09167933e-01 5.07624984e-01 -2.13405818e-01 -6.38921797e-01 8.19927037e-01 -3.72496456e-01 8.21314812e-01 1.19283140e+00 4.94784415e-01 3.34900528e-01 3.34678471e-01 9.05072570e-01 3.00453514e-01 1.03827707e-01 -1.21423960e+00 4.64835614e-01 4.88506675e-01 5.51027060e-01 -5.93402505e-01 -1.09071299e-01 -3.94529015e-01 8.13111961e-01 5.22664487e-01 7.33123422e-01 -1.02747738e+00 2.23113969e-01 1.19085550e+00 1.50931925e-01 5.35183728e-01 -5.55967271e-01 -4.79160368e-01 -8.30500901e-01 -5.57622546e-03 4.31569703e-02 8.38519782e-02 -7.51165271e-01 -1.21949935e+00 5.76001823e-01 -2.44627297e-01 -1.18222368e+00 -3.37429315e-01 -4.22884881e-01 -8.78697634e-01 6.22222364e-01 -1.44992530e+00 -1.42376375e+00 -4.11937356e-01 5.93505859e-01 4.53776240e-01 -2.67570853e-01 5.79277694e-01 1.78033173e-01 -4.48433399e-01 2.60478497e-01 4.69087549e-02 1.11248367e-01 2.25367844e-01 -1.11032629e+00 9.23795104e-01 9.80059445e-01 -2.23566726e-01 4.63621199e-01 6.98132098e-01 -9.36632097e-01 -9.48015153e-01 -1.71579766e+00 1.11928737e+00 -6.87677085e-01 5.77686965e-01 -1.40768975e-01 -3.99785906e-01 1.18576753e+00 2.39651799e-01 -6.17556907e-02 9.45940733e-01 1.29472584e-01 -1.63490936e-01 3.43036413e-01 -8.85101140e-01 1.51437521e+00 1.60441327e+00 -3.07855725e-01 -1.67231977e-01 2.93785781e-01 8.21741402e-01 -1.50271505e-01 -2.87299842e-01 1.02353610e-01 5.07453382e-01 -1.23425686e+00 1.34659374e+00 -8.15582633e-01 2.04737410e-01 -3.92925441e-01 -3.93833220e-01 -1.33788145e+00 -8.28004599e-01 -3.79800558e-01 4.96822745e-02 6.74712420e-01 5.98562539e-01 -7.30046928e-01 1.04031801e+00 9.94134784e-01 -2.41009995e-01 -6.08727813e-01 -1.15220320e+00 -6.16000772e-01 9.53651667e-02 -7.41897047e-01 7.99062014e-01 5.51403761e-01 2.57613137e-02 1.06520571e-01 -8.21747780e-01 4.34297889e-01 7.24526644e-01 -2.32474297e-01 8.30881238e-01 -1.12972641e+00 3.14235091e-01 -7.07220852e-01 -6.65024459e-01 -1.43531644e+00 5.96079826e-01 -6.44754946e-01 6.33742884e-02 -1.73435175e+00 -1.73643082e-02 -4.71038252e-01 9.53755900e-02 1.70130923e-01 -1.19612508e-01 1.51010588e-01 2.77019083e-01 1.82041943e-01 -1.16013408e+00 9.36084032e-01 1.05550528e+00 -3.07392448e-01 -2.54999757e-01 2.93673545e-01 -6.56079710e-01 8.51857901e-01 8.56608212e-01 -2.10448682e-01 -6.07644916e-01 -3.38060945e-01 2.19328806e-01 -1.17201537e-01 8.69140089e-01 -1.22656739e+00 7.84766018e-01 -5.05460203e-01 3.07031363e-01 -8.29987526e-01 7.87330747e-01 -7.24882543e-01 -1.26040559e-02 4.37498599e-01 -4.73216653e-01 2.63756424e-01 7.96739943e-03 1.16995180e+00 2.18829244e-01 3.99650633e-01 3.77129674e-01 -2.13631749e-01 -1.18928492e+00 7.41641223e-01 -1.04021716e+00 5.87695763e-02 1.30730414e+00 -3.36906999e-01 -4.60748315e-01 -9.49106157e-01 -5.08049846e-01 6.28399909e-01 5.70945024e-01 5.26279986e-01 8.36612046e-01 -1.62201560e+00 -6.20827317e-01 -1.53088614e-01 2.24929988e-01 -2.94585288e-01 4.96820033e-01 7.07480669e-01 -5.88616312e-01 9.26472187e-01 -2.27189109e-01 -5.23371220e-01 -8.86295855e-01 5.03307462e-01 3.14140499e-01 9.43238884e-02 -6.92139983e-01 6.78956151e-01 4.54549104e-01 -6.32350445e-01 9.55272093e-02 -2.82963067e-01 -4.64590669e-01 -4.64707643e-01 5.66815794e-01 7.43111670e-01 -8.44390154e-01 -1.30571783e+00 -3.76418978e-01 5.21210790e-01 3.11430871e-01 -2.63463169e-01 1.16561699e+00 -5.78620732e-01 1.80830508e-01 2.51778513e-01 8.66798580e-01 -2.89027065e-01 -1.68010545e+00 2.83762603e-03 -1.07328109e-01 -7.04726219e-01 -2.11855799e-01 -6.61256373e-01 -7.40431905e-01 5.75038433e-01 3.31781536e-01 1.33907616e-01 4.95192826e-01 -9.17234421e-02 1.04455483e+00 6.27337217e-01 5.09895027e-01 -9.21503782e-01 5.07990457e-03 8.53188932e-01 6.56732023e-01 -1.64058113e+00 -4.61435795e-01 -4.57946837e-01 -8.31885278e-01 6.70987189e-01 9.33467388e-01 -7.10935220e-02 1.04739153e+00 -3.26216221e-01 -7.42519423e-02 4.99095470e-02 -7.39729643e-01 -4.26814139e-01 3.87918890e-01 1.13969398e+00 -7.92116895e-02 2.47617632e-01 2.44974673e-01 3.85594219e-01 -3.76293540e-01 -2.04084769e-01 5.05437613e-01 5.60778797e-01 -6.32138431e-01 -6.75686955e-01 -9.01388600e-02 2.17016384e-01 1.08648181e-01 -1.26949608e-01 -3.18303972e-01 7.25755394e-01 1.85483724e-01 1.31092048e+00 2.89866090e-01 -4.56999183e-01 1.78946465e-01 -4.92029518e-01 -3.10496055e-02 -1.88407525e-01 -2.10674569e-01 -7.21649349e-01 2.37234607e-01 -9.00111198e-01 -3.87741566e-01 -9.64485347e-01 -1.25397980e+00 -8.76790047e-01 5.08373201e-01 -2.41198137e-01 3.39772373e-01 1.14922249e+00 9.27847862e-01 3.13228458e-01 2.26747185e-01 -1.23432457e+00 3.40173811e-01 -4.39586550e-01 -8.46441910e-02 3.08517486e-01 5.18020928e-01 -8.15564871e-01 -9.72004458e-02 -2.48808995e-01]
[6.018428802490234, 0.7580181360244751]
635d1130-b1e4-43ea-8a3f-dda90cf8784b
drum-end-to-end-differentiable-rule-mining-on
1911.00055
null
https://arxiv.org/abs/1911.00055v1
https://arxiv.org/pdf/1911.00055v1.pdf
DRUM: End-To-End Differentiable Rule Mining On Knowledge Graphs
In this paper, we study the problem of learning probabilistic logical rules for inductive and interpretable link prediction. Despite the importance of inductive link prediction, most previous works focused on transductive link prediction and cannot manage previously unseen entities. Moreover, they are black-box models that are not easily explainable for humans. We propose DRUM, a scalable and differentiable approach for mining first-order logical rules from knowledge graphs which resolves these problems. We motivate our method by making a connection between learning confidence scores for each rule and low-rank tensor approximation. DRUM uses bidirectional RNNs to share useful information across the tasks of learning rules for different relations. We also empirically demonstrate the efficiency of DRUM over existing rule mining methods for inductive link prediction on a variety of benchmark datasets.
['Patrick Ding', 'Ali Sadeghian', 'Mohammadreza Armandpour', 'Daisy Zhe Wang']
2019-10-31
drum-end-to-end-differentiable-rule-mining-on-1
http://papers.nips.cc/paper/9669-drum-end-to-end-differentiable-rule-mining-on-knowledge-graphs
http://papers.nips.cc/paper/9669-drum-end-to-end-differentiable-rule-mining-on-knowledge-graphs.pdf
neurips-2019-12
['inductive-link-prediction', 'inductive-knowledge-graph-completion']
['graphs', 'knowledge-base']
[-6.09908924e-02 7.99934745e-01 -9.01262760e-01 -5.73846221e-01 -1.54630765e-01 -2.44173065e-01 2.92637080e-01 3.87071759e-01 2.98146486e-01 1.07120109e+00 4.70521688e-01 -9.82036293e-01 -9.93707955e-01 -1.22578967e+00 -1.03869498e+00 1.02684058e-01 -9.38401759e-01 1.05986333e+00 2.70085186e-01 -4.17848736e-01 -2.28946373e-01 2.86753058e-01 -1.35267365e+00 8.77771199e-01 7.49915361e-01 8.54125798e-01 -6.40255511e-01 5.55914581e-01 -1.59121707e-01 1.86789477e+00 1.27113119e-01 -9.53791738e-01 -1.46872267e-01 7.14234263e-02 -1.56607997e+00 -7.07688093e-01 2.85451233e-01 -2.89668143e-01 -5.72398126e-01 5.96996844e-01 -1.87996969e-01 3.02854069e-02 6.97043359e-01 -1.65320098e+00 -1.01298094e+00 1.77855897e+00 -1.86867833e-01 3.07770133e-01 4.57732648e-01 -7.10540354e-01 1.94462335e+00 -9.91697550e-01 8.93598974e-01 1.39220440e+00 7.91883588e-01 5.06912231e-01 -1.66974783e+00 -4.37998652e-01 2.66589284e-01 8.85638535e-01 -1.21771729e+00 -3.10686618e-01 6.25427902e-01 -4.52177763e-01 1.17253780e+00 4.44363415e-01 4.72230077e-01 1.17179763e+00 -1.59905568e-01 9.06726301e-01 7.28961229e-01 -3.80258083e-01 -1.46567389e-01 -3.58848609e-02 6.89536333e-01 9.83739078e-01 6.97759509e-01 2.60253489e-01 -1.05911791e+00 -3.86408955e-01 3.40427458e-01 -3.04671764e-01 -1.08813651e-01 -5.86145639e-01 -1.36885262e+00 8.41092825e-01 6.34136558e-01 1.17048979e-01 -1.10205986e-01 5.28178930e-01 3.88427347e-01 4.49352622e-01 4.27007765e-01 4.64876086e-01 -9.59331989e-01 2.35149011e-01 -3.04799050e-01 -2.33695637e-02 1.33696353e+00 9.76454914e-01 5.72297454e-01 -1.99403286e-01 -1.09333754e-01 4.54135567e-01 6.35982215e-01 1.55337453e-02 -2.43126944e-01 -7.97051907e-01 5.22152007e-01 6.93480134e-01 8.22980050e-03 -1.32485330e+00 -5.77019870e-01 -3.99103016e-01 -6.51756942e-01 -3.17695290e-01 4.02178079e-01 -4.73391376e-02 -4.33755785e-01 1.58231544e+00 4.19769078e-01 4.27197129e-01 3.00468832e-01 5.10429800e-01 1.09011424e+00 4.73794937e-01 1.47467732e-01 -2.79688537e-01 9.93974626e-01 -7.99423575e-01 -5.96829891e-01 1.69709563e-01 9.11534786e-01 -2.77535677e-01 6.48375094e-01 2.93257594e-01 -8.91164064e-01 -8.20175111e-02 -8.87027264e-01 -2.88922518e-01 -4.20414388e-01 -1.40321717e-01 1.56159067e+00 1.33336410e-02 -8.91374350e-01 7.17964351e-01 -6.93926632e-01 -3.02569866e-01 3.36410165e-01 6.36968970e-01 -4.30214614e-01 8.40616748e-02 -1.62561643e+00 9.14274752e-01 9.42476749e-01 2.96373695e-01 -5.12611508e-01 -8.75113189e-01 -7.06741512e-01 1.67851374e-01 8.42917740e-01 -9.72482681e-01 8.88002932e-01 -3.77658904e-01 -1.05401456e+00 5.10609150e-01 -1.61407411e-01 -7.58830190e-01 9.63995755e-02 -4.47430909e-01 -7.30556071e-01 -2.27811992e-01 6.99491799e-02 5.61983883e-01 3.67144197e-01 -1.40172517e+00 -7.60908544e-01 -2.50142533e-02 3.09628814e-01 -2.86946326e-01 -3.82056266e-01 -2.66130388e-01 -1.28161356e-01 -3.89276654e-01 8.34148824e-02 -7.22776949e-01 1.70626063e-02 -3.60781759e-01 -1.08521557e+00 -7.55490899e-01 6.27020895e-01 -2.74898320e-01 1.27839231e+00 -1.58489072e+00 2.20962033e-01 6.83708012e-01 6.09590471e-01 -1.55758098e-01 -5.52185923e-02 4.14149582e-01 -4.98401284e-01 4.68466938e-01 3.12294543e-01 2.78311044e-01 5.37287071e-02 6.54316068e-01 -9.14875448e-01 -1.63612857e-01 3.62362146e-01 1.32036734e+00 -1.08931231e+00 -7.03172982e-01 -4.30966198e-01 2.66246468e-01 -5.17085135e-01 -2.74702832e-02 -6.08127892e-01 -1.23776026e-01 -5.27848005e-01 5.95180869e-01 6.28975183e-02 -7.17457831e-01 9.29767072e-01 -5.55268168e-01 4.02864665e-01 8.78875852e-01 -1.04242337e+00 1.03660715e+00 -2.67651286e-02 4.99213427e-01 -5.09230494e-01 -1.16833353e+00 7.19527066e-01 3.38645905e-01 4.72187459e-01 -1.55627146e-01 -2.50936478e-01 1.03608780e-01 2.12229177e-01 -6.23231769e-01 3.01789314e-01 -1.76973969e-01 3.42984974e-01 5.10482013e-01 1.22129284e-01 4.89701599e-01 4.69413131e-01 6.00723982e-01 1.24297738e+00 2.85684139e-01 1.54240444e-01 -2.29550782e-03 2.29049951e-01 2.28981808e-01 6.74205363e-01 7.24796534e-01 3.76695842e-01 -2.88364470e-01 7.55415618e-01 -9.90250051e-01 -5.31894326e-01 -1.30430293e+00 -6.20457493e-02 1.42545712e+00 -5.14060166e-03 -8.60856891e-01 1.96085796e-01 -1.10658073e+00 4.76039857e-01 6.96381271e-01 -7.39664495e-01 -1.96740478e-02 -4.18218762e-01 -7.39368916e-01 6.75442398e-01 7.75524497e-01 -1.33483008e-01 -7.99992681e-01 4.37674433e-01 1.76735520e-01 -4.54650640e-01 -1.27642572e+00 3.01115423e-01 4.53824997e-01 -1.02048421e+00 -1.55455673e+00 7.20745146e-01 -7.06471026e-01 6.36553288e-01 7.31218830e-02 1.69261158e+00 4.45215225e-01 2.34617908e-02 2.82953769e-01 -3.33959311e-01 -2.80571520e-01 -3.45370084e-01 1.81992412e-01 5.11934161e-01 -1.77243441e-01 6.84167862e-01 -7.38256216e-01 -1.61742456e-02 2.18158200e-01 -5.75509012e-01 2.23993540e-01 3.50515038e-01 9.81852293e-01 5.87523699e-01 3.03833663e-01 6.76237643e-01 -1.17993438e+00 8.37737441e-01 -7.27449477e-01 -3.45493436e-01 6.59176052e-01 -8.92793238e-01 4.59077120e-01 3.22391003e-01 -2.89551437e-01 -8.17205429e-01 -1.62391484e-01 3.02498013e-01 -4.16159537e-03 2.45619297e-01 1.14070618e+00 2.36314803e-01 8.21784213e-02 7.74996936e-01 -4.26981866e-01 -3.72227013e-01 -1.41095191e-01 8.12628925e-01 2.72082090e-02 3.21153998e-01 -1.08037162e+00 1.17194247e+00 4.41361994e-01 4.42689210e-01 -1.64358228e-01 -1.48682845e+00 -4.95802090e-02 -7.49630094e-01 1.17429636e-01 6.07105374e-01 -6.81025565e-01 -1.45414352e+00 -6.55286729e-01 -1.29513752e+00 -1.74928412e-01 -2.54528433e-01 5.89327931e-01 -3.46411288e-01 2.76999563e-01 -9.17719662e-01 -7.65605628e-01 -2.78355002e-01 -2.20171288e-01 2.27632537e-01 -3.60571623e-01 -6.82236791e-01 -1.24878621e+00 1.66382298e-01 6.04772091e-01 -1.10236727e-01 9.62370262e-02 1.51917589e+00 -8.09864640e-01 -8.76182914e-01 1.06171638e-01 -3.81152302e-01 -1.11499280e-02 4.20538820e-02 2.07007572e-01 -6.24258101e-01 3.91057611e-01 -9.22534585e-01 -6.47306979e-01 1.02673995e+00 -8.58923569e-02 1.09540057e+00 -7.86204159e-01 -7.09383786e-01 2.48045146e-01 9.69571829e-01 -2.30914161e-01 4.98168021e-01 2.49711752e-01 1.09490538e+00 7.35740781e-01 5.08302152e-01 -1.40643362e-02 9.76640940e-01 3.79802078e-01 4.11811322e-01 7.11420504e-03 1.31115273e-01 -7.65913665e-01 1.26217574e-01 8.44741523e-01 -7.30319679e-01 -1.48220835e-02 -1.10015106e+00 5.10779798e-01 -2.35301709e+00 -1.37901688e+00 -5.76630592e-01 1.55944705e+00 1.27818930e+00 3.76813978e-01 1.31403357e-01 2.03070760e-01 4.96173352e-01 -5.00531673e-01 -2.51299322e-01 -3.56226414e-01 -2.24941030e-01 -1.37768779e-02 2.72028118e-01 6.84904337e-01 -1.16182816e+00 1.08007026e+00 7.17128372e+00 5.81783831e-01 -5.37388504e-01 -5.90806231e-02 5.27283967e-01 -7.19327405e-02 -8.21826875e-01 3.24846566e-01 -7.89711535e-01 -1.15833849e-01 7.57427871e-01 4.65133302e-02 5.01090169e-01 8.54439795e-01 -1.38781175e-01 3.66218328e-01 -1.78975511e+00 6.01904154e-01 -3.85222942e-01 -1.89994764e+00 3.53280395e-01 -6.79924041e-02 7.93196499e-01 1.29561394e-01 -3.33048016e-01 3.65805954e-01 1.20513010e+00 -1.29695213e+00 2.96692729e-01 9.19849634e-01 3.18790853e-01 -6.01473331e-01 6.14844859e-01 4.12016525e-04 -9.22118664e-01 -2.23081559e-01 -3.98670912e-01 -3.27155560e-01 -1.46491388e-02 1.14188409e+00 -1.22533357e+00 8.14365327e-01 6.45242870e-01 8.97851944e-01 -3.98099840e-01 5.23640454e-01 -7.37498581e-01 9.87636268e-01 -3.80782753e-01 -2.14508921e-01 -2.18012109e-01 1.73514783e-01 3.99578035e-01 1.15111792e+00 -8.14188868e-02 7.15384334e-02 3.00301090e-02 1.01010847e+00 -4.76727605e-01 -1.53441593e-01 -9.26963449e-01 -1.64313883e-01 4.73403543e-01 1.13665295e+00 -3.72290581e-01 -2.49028906e-01 -3.15944463e-01 1.69947267e-01 8.86111081e-01 4.51442093e-01 -6.25612795e-01 -8.23396817e-03 5.49226820e-01 8.58135968e-02 2.99142212e-01 -3.00403774e-01 -3.52774292e-01 -1.25737584e+00 6.09846413e-02 -6.27658963e-01 9.15350318e-01 -6.58766925e-01 -1.73593223e+00 3.32066894e-01 -7.43326964e-03 -7.07690120e-01 -3.48398656e-01 -8.79166901e-01 -2.82533556e-01 2.46924862e-01 -1.50260544e+00 -1.49404883e+00 2.09489793e-01 5.82003891e-01 -2.24052086e-01 -2.33078785e-02 9.77074027e-01 1.62877232e-01 -5.54629505e-01 4.03333038e-01 -2.73706436e-01 4.37907070e-01 4.44430202e-01 -1.36612284e+00 2.35899597e-01 4.67475295e-01 6.50799096e-01 1.04499221e+00 6.36530399e-01 -8.95603597e-01 -1.64491701e+00 -1.16961813e+00 1.27951276e+00 -9.06730711e-01 1.45392215e+00 -2.45091394e-01 -8.82074058e-01 1.23938572e+00 -1.16120666e-01 3.77212882e-01 1.12997043e+00 1.37942529e+00 -1.00495744e+00 -4.12227213e-01 -5.49100339e-01 5.65460503e-01 1.55425596e+00 -5.97077429e-01 -7.82186806e-01 4.94354635e-01 8.73219669e-01 1.03599817e-01 -1.24285460e+00 8.32420409e-01 5.94634652e-01 -5.93885541e-01 1.18859208e+00 -1.61776590e+00 8.33100140e-01 -4.07918423e-01 -4.37438861e-03 -1.04408240e+00 -7.03648090e-01 -5.53546607e-01 -1.16364360e+00 1.09497190e+00 1.20932710e+00 -2.73647338e-01 8.73004138e-01 6.12090886e-01 3.02717477e-01 -1.07319534e+00 -6.65711164e-01 -7.27629662e-01 -2.54124254e-01 -9.89656687e-01 4.87207234e-01 1.52927208e+00 7.77962148e-01 1.08084762e+00 -5.73543251e-01 2.98633248e-01 8.17839563e-01 4.43832785e-01 5.82607150e-01 -1.75450420e+00 -5.03399074e-01 -1.97684407e-01 -4.65460688e-01 -7.94836402e-01 5.35726130e-01 -1.45577860e+00 -3.51560920e-01 -1.79893756e+00 3.36545825e-01 -6.91688716e-01 -4.13055718e-01 1.01320350e+00 -3.37749161e-02 8.26518983e-02 -2.16858968e-01 1.72982469e-01 -9.16520357e-01 3.18088174e-01 9.18715537e-01 -5.65157413e-01 6.20598868e-02 -2.07564056e-01 -1.00806117e+00 7.14396238e-01 6.42330945e-01 -4.87039685e-01 -7.73213446e-01 -5.79987586e-01 1.47210109e+00 -1.09757438e-01 6.15370870e-01 -3.50700021e-01 3.82442892e-01 -5.27725339e-01 3.12989712e-01 -6.73571348e-01 6.76852167e-02 -7.01949537e-01 1.02338918e-01 3.09404641e-01 -5.63566327e-01 -3.13373566e-01 -8.84412676e-02 9.02614057e-01 2.77493596e-02 1.26437172e-01 -1.61991611e-01 2.92243719e-01 -7.69281983e-01 2.85260350e-01 -2.81000659e-02 2.34057512e-02 7.72183001e-01 1.77405879e-01 -6.61058486e-01 -3.58774453e-01 -1.11031508e+00 4.64554608e-01 -4.61723387e-01 5.16566932e-01 8.37432861e-01 -1.64400542e+00 -8.70117366e-01 -5.27654946e-01 1.80075765e-01 -1.29888296e-01 -4.48614299e-01 9.59429324e-01 -1.34531245e-01 6.37743950e-01 2.41645068e-01 -1.36600122e-01 -1.15809107e+00 9.44063842e-01 2.53390789e-01 -5.11671841e-01 -4.28687155e-01 9.23607528e-01 -1.79109395e-01 -5.65115392e-01 2.66244173e-01 -5.17092526e-01 -4.35470074e-01 3.74535583e-02 3.30857396e-01 4.72432256e-01 -2.21958384e-01 -4.87069227e-02 -5.29747069e-01 6.71237633e-02 -1.60289586e-01 4.00897503e-01 1.66788650e+00 1.57345161e-01 -8.25245738e-01 7.00708032e-01 6.55212224e-01 1.00137644e-01 -3.60386252e-01 -3.60882550e-01 7.00825572e-01 4.89952788e-03 -2.63650119e-01 -9.06504571e-01 -7.06478953e-01 3.57633799e-01 -2.07816735e-01 6.57274723e-01 4.38079536e-01 5.32467425e-01 5.17774045e-01 1.26683760e+00 3.24074835e-01 -9.78414834e-01 -6.18417049e-03 9.13950443e-01 8.24492037e-01 -1.29499865e+00 1.73307300e-01 -1.07153726e+00 -2.86638916e-01 1.31612289e+00 5.55282414e-01 2.47806937e-01 7.69542396e-01 2.58585751e-01 -3.43314558e-01 -5.63420713e-01 -1.37755203e+00 -2.84888089e-01 7.00960755e-01 6.21491432e-01 8.10652018e-01 3.00673455e-01 -1.74710646e-01 5.88438928e-01 -2.43653521e-01 1.91932112e-01 2.49589756e-01 5.30409217e-01 -2.27118239e-01 -1.21431243e+00 -1.49312630e-01 7.27836430e-01 -3.28072608e-01 -3.87456179e-01 -8.27236235e-01 6.39156520e-01 1.48722574e-01 1.09014976e+00 -2.49410734e-01 -9.54422772e-01 4.88446876e-02 1.18956141e-01 4.62104440e-01 -5.24786592e-01 4.12550867e-02 -7.79802084e-01 9.96626258e-01 -4.09491718e-01 -6.32801056e-01 -2.92574674e-01 -1.40156353e+00 -6.98578119e-01 -3.42709601e-01 4.37017441e-01 4.66446653e-02 1.14530861e+00 5.47739804e-01 5.06423533e-01 2.97135830e-01 1.72116876e-01 -1.90847233e-01 -7.50920057e-01 -3.62195402e-01 2.21761987e-01 1.26416013e-01 -8.95909071e-01 -1.32993609e-02 1.55274451e-01]
[8.881119728088379, 7.766942977905273]
7df7abcc-f2c5-40db-9449-3bed1894ddb0
comstreamclust-a-communicative-text
2010.05349
null
https://arxiv.org/abs/2010.05349v2
https://arxiv.org/pdf/2010.05349v2.pdf
ComStreamClust: a communicative multi-agent approach to text clustering in streaming data
Topic detection is the task of determining and tracking hot topics in social media. Twitter is arguably the most popular platform for people to share their ideas with others about different issues. One such prevalent issue is the COVID-19 pandemic. Detecting and tracking topics on these kinds of issues would help governments and healthcare companies deal with this phenomenon. In this paper, we propose a novel, multi-agent, communicative clustering approach, so-called ComStreamClust for clustering sub-topics inside a broader topic, e.g., COVID-19. The proposed approach is parallelizable, and can simultaneously handle several data-point. The LaBSE sentence embedding is used to measure the semantic similarity between two tweets. ComStreamClust has been evaluated on two datasets: the COVID-19 and the FA CUP. The results obtained from ComStreamClust approve the effectiveness of the proposed approach when compared to existing methods.
['Meysam Asgari-Chenaghlu', 'Ali Mohammadpur-Fard', 'Rahim Dehkharghani', 'Araz Gholipour-Shilabin', 'Ali Najafi']
2020-10-11
null
null
null
null
['text-clustering']
['natural-language-processing']
[-3.26221168e-01 -2.54750121e-02 -1.37569591e-01 -8.83481652e-02 -5.24355292e-01 -3.68058056e-01 9.50931966e-01 1.06790841e+00 -4.96759832e-01 6.33993566e-01 5.40234327e-01 2.53521591e-01 -1.74599618e-01 -9.41558242e-01 6.57715797e-02 -7.34695852e-01 -2.27047116e-01 8.26845527e-01 3.03064734e-01 -1.95386261e-01 4.24483180e-01 -7.57865086e-02 -1.39274848e+00 1.10733338e-01 7.72755206e-01 6.50206029e-01 4.19788122e-01 3.53589147e-01 -4.91060317e-01 4.24836874e-01 -6.63484216e-01 -2.77936935e-01 -2.03412056e-01 -1.78236857e-01 -8.74047935e-01 -1.56885862e-01 -2.47082025e-01 1.17604263e-01 3.96343768e-01 9.81660903e-01 3.84726077e-01 1.91086709e-01 7.41236567e-01 -1.43608236e+00 1.66644409e-01 6.43213272e-01 -1.00986743e+00 5.17390668e-01 3.60380977e-01 -4.90579456e-01 1.09611166e+00 -6.89696670e-01 8.45469892e-01 1.32985914e+00 5.72348475e-01 -2.30674949e-02 -8.27434540e-01 -7.74115980e-01 2.36179903e-01 1.68441653e-01 -1.53637886e+00 1.10458069e-01 8.54217112e-01 -7.56842613e-01 5.26587248e-01 4.80248898e-01 5.69645345e-01 8.09514284e-01 3.74436229e-01 6.21372998e-01 1.28429914e+00 -2.77971942e-02 4.86286432e-01 6.26913905e-01 7.21429110e-01 1.49207428e-01 5.59863925e-01 -6.91226423e-01 -3.79877627e-01 -7.85169244e-01 -1.50227413e-01 3.28765303e-01 -5.34929559e-02 3.18560489e-02 -1.24075902e+00 1.38364279e+00 1.05349310e-01 7.76288211e-01 -6.34611785e-01 -3.31558257e-01 6.12590313e-01 -3.53612229e-02 1.18468404e+00 4.16638941e-01 -1.11046493e-01 -1.83794811e-01 -7.55933166e-01 3.33359003e-01 7.98847377e-01 4.95430887e-01 8.37888420e-01 -6.66029394e-01 1.15713745e-01 6.38981640e-01 6.00106299e-01 3.84994179e-01 6.69075966e-01 -1.95431858e-01 4.98210490e-01 8.94415081e-01 1.66472957e-01 -1.62444580e+00 -8.16549957e-01 -2.79893160e-01 -7.48724043e-01 -3.93420994e-01 -1.09476015e-01 -5.90171456e-01 -1.85882166e-01 1.38780236e+00 9.61787343e-01 5.11755288e-01 9.17232335e-02 7.41183281e-01 1.00913501e+00 1.10709119e+00 3.61453712e-01 -3.86544049e-01 1.73430681e+00 -5.62571168e-01 -9.65552688e-01 1.00300193e-01 4.50937778e-01 -9.22160268e-01 4.65175867e-01 1.08856864e-01 -5.67861378e-01 -2.16370851e-01 -5.73557854e-01 4.05815750e-01 -7.65042901e-01 -2.76788026e-01 4.82985526e-01 5.16652822e-01 -8.77847254e-01 -1.75593048e-02 -6.75594985e-01 -9.02089417e-01 7.20260516e-02 2.33493865e-01 -1.94326639e-02 3.42301190e-01 -1.36689270e+00 5.83075941e-01 7.63846636e-01 -5.28536618e-01 -3.98062408e-01 -6.50804281e-01 -3.96609008e-01 -1.11158393e-01 2.73999661e-01 -4.04744685e-01 8.49804342e-01 -4.86751378e-01 -1.02598274e+00 7.84881175e-01 -2.30363861e-01 -4.73041385e-01 1.86715424e-01 -3.20405841e-01 -6.36927009e-01 1.92948729e-01 5.50668716e-01 4.51072901e-01 7.06286252e-01 -1.13467848e+00 -9.07429159e-01 -4.50025439e-01 -1.66420579e-01 3.31780106e-01 -7.87363768e-01 5.52381456e-01 -1.06978320e-01 -5.71166515e-01 -3.24914485e-01 -1.05645156e+00 -1.40208632e-01 -6.92698359e-01 -8.24751318e-01 -8.42432857e-01 1.48002553e+00 -3.98220748e-01 1.31697726e+00 -1.98498344e+00 -7.09647387e-02 1.53184935e-01 5.58837414e-01 1.45048022e-01 4.13781315e-01 9.48015809e-01 2.02046990e-01 1.17459029e-01 1.56769417e-02 -3.40419471e-01 -7.44079277e-02 -1.97858185e-01 -3.33306611e-01 5.02357960e-01 2.17247624e-02 1.80780664e-01 -8.78318727e-01 -6.60637081e-01 3.68186384e-02 4.66886848e-01 -3.30938369e-01 1.87948436e-01 -4.10261512e-01 3.61151665e-01 -8.88723195e-01 1.64588820e-02 7.15195775e-01 -5.19206345e-01 2.16270328e-01 2.39457618e-02 -4.34887439e-01 1.41731367e-01 -1.12819874e+00 1.20682013e+00 -1.47981599e-01 7.59999573e-01 2.07009800e-02 -7.95229197e-01 9.39851344e-01 4.84243631e-01 8.65921199e-01 -1.15545265e-01 5.13287485e-01 -1.47610083e-01 -1.39176682e-01 -4.63464916e-01 9.11783576e-01 5.07427529e-02 -2.41870254e-01 1.00386739e+00 -4.16539311e-01 3.71581346e-01 3.92984390e-01 3.94873708e-01 8.62678230e-01 -6.66835070e-01 6.13293827e-01 -6.20901287e-01 4.51360106e-01 1.89694986e-01 3.32975328e-01 3.07908028e-01 -1.82732075e-01 2.30610564e-01 3.24995697e-01 -2.05723643e-01 -6.83695555e-01 -6.85379028e-01 -1.78855330e-01 1.09632921e+00 4.10792470e-01 -6.88659966e-01 -7.41123974e-01 -6.39193177e-01 2.68632304e-02 5.66058695e-01 -6.04341328e-01 3.53251219e-01 -3.36689860e-01 -1.18324029e+00 1.45949930e-01 -2.67860830e-01 7.13634133e-01 -8.14032733e-01 -7.16602206e-01 3.67202878e-01 -5.68251491e-01 -8.83686066e-01 -3.52088779e-01 -3.33588719e-01 -4.90502596e-01 -1.00269604e+00 -5.83531857e-01 -7.91208506e-01 4.37089145e-01 5.09825945e-01 7.11647987e-01 -1.67314768e-01 -3.94518971e-02 3.25864255e-01 -5.01266420e-01 -6.35789931e-01 -2.50592470e-01 3.92415285e-01 2.11885720e-01 4.96359974e-01 6.38468385e-01 -2.96031415e-01 -7.24336684e-01 3.88927221e-01 -8.59693706e-01 -2.20582243e-02 7.10012987e-02 2.53167063e-01 1.92923114e-01 5.69314420e-01 7.34523237e-01 -1.21391726e+00 9.98569727e-01 -1.34012365e+00 -2.34878376e-01 -1.05941586e-01 -5.68922341e-01 -4.04024392e-01 3.10289353e-01 -3.08539987e-01 -1.02695751e+00 -3.87372851e-01 5.27825579e-02 3.24028164e-01 -3.61841053e-01 7.23148942e-01 1.93892583e-01 4.98031765e-01 2.32356563e-01 -9.60646272e-02 -2.03197792e-01 -4.90328878e-01 1.76536098e-01 1.11632073e+00 -1.78577080e-01 -6.46619648e-02 7.34066784e-01 8.83248210e-01 -5.54729879e-01 -1.32853138e+00 -6.23347044e-01 -1.21640313e+00 -7.03453347e-02 -3.04587692e-01 1.29977584e+00 -9.57388878e-01 -8.52648795e-01 3.10729921e-01 -1.32941675e+00 3.63543510e-01 1.74835578e-01 5.86508870e-01 4.58948575e-02 2.11728618e-01 -5.10884881e-01 -8.30957234e-01 -6.71741784e-01 -7.76031971e-01 9.37888920e-01 3.67896825e-01 -6.58823907e-01 -1.47403824e+00 6.82860851e-01 2.91514158e-01 5.72620392e-01 4.94917572e-01 6.98243976e-01 -1.37687778e+00 -1.71091095e-01 -4.09978181e-02 -2.45196208e-01 -4.13581282e-01 4.94658083e-01 -1.26060173e-01 -8.96188259e-01 -4.10425276e-01 1.22047432e-01 5.02477493e-03 5.05351901e-01 1.67930961e-01 5.09077430e-01 -6.16700292e-01 -8.99735212e-01 -3.46632361e-01 1.29921055e+00 4.05085206e-01 1.36215225e-01 5.79174340e-01 4.36575294e-01 6.90459549e-01 7.34163404e-01 7.84174621e-01 6.72423840e-01 7.24648297e-01 2.59813726e-01 -1.97659820e-01 3.77243191e-01 1.20465882e-01 2.86340863e-01 1.41924369e+00 3.55701745e-01 -4.76981997e-01 -1.08970273e+00 6.69665992e-01 -1.73579288e+00 -9.74972427e-01 -3.01405460e-01 1.61564600e+00 7.01547861e-01 2.21412834e-02 3.57861608e-01 1.28147349e-01 1.12218142e+00 3.54954928e-01 -9.72440466e-02 -3.20368797e-01 4.71966267e-02 -2.62707710e-01 3.84063981e-02 5.05091548e-01 -1.34194517e+00 5.91597676e-01 4.87750626e+00 8.71495366e-01 -1.12651134e+00 4.90941405e-01 7.05705225e-01 3.72355103e-01 -2.03707114e-01 -2.06046447e-01 -1.01095080e+00 7.60936201e-01 9.94994164e-01 -5.50030351e-01 -3.52863818e-01 7.50841916e-01 5.16108453e-01 -3.29842955e-01 -4.51934338e-01 6.98753297e-01 4.09874529e-01 -1.26701689e+00 -2.26981223e-01 5.43930903e-02 8.29197168e-01 1.44827351e-01 9.88871232e-02 -1.02521844e-01 3.76492560e-01 -3.63173306e-01 2.06162646e-01 1.03435852e-01 1.57683328e-01 -7.94278800e-01 9.69267309e-01 3.31537038e-01 -1.34093451e+00 2.27409899e-01 -1.41680524e-01 6.59835041e-02 3.20813924e-01 8.33385587e-01 -1.34425724e+00 4.92290795e-01 6.60690963e-01 7.79351115e-01 -3.96980971e-01 1.20411706e+00 2.69294024e-01 8.45829427e-01 -2.84591645e-01 -6.58291101e-01 3.70511144e-01 -3.15992028e-01 8.82808506e-01 1.40508020e+00 3.21576327e-01 -5.91945089e-02 3.52232754e-01 3.48679334e-01 2.67449301e-03 5.65076888e-01 -6.94007933e-01 -1.44886136e-01 4.75725055e-01 1.40004992e+00 -1.26402950e+00 -3.59203696e-01 -2.06327423e-01 4.95544523e-01 1.92454718e-02 8.08706228e-03 -8.61609936e-01 -4.77982908e-01 6.00774646e-01 2.44801581e-01 1.87501594e-01 2.30380171e-03 3.75561751e-02 -7.27232337e-01 -3.87764007e-01 -5.98557770e-01 6.27461076e-01 -3.74877036e-01 -1.24740696e+00 8.08433652e-01 3.27949464e-01 -1.14041877e+00 -7.97527060e-02 1.30133319e-03 -9.26560700e-01 2.70458460e-01 -1.37745059e+00 -8.12013268e-01 -3.32401305e-01 6.15963280e-01 6.42303169e-01 -1.05235629e-01 7.63141513e-01 3.23920190e-01 -4.86351162e-01 2.58805417e-03 4.08807039e-01 -3.68347093e-02 5.00723958e-01 -1.08410394e+00 1.12384796e-01 4.36383009e-01 -2.44388785e-02 6.72964752e-01 9.28461671e-01 -1.00244355e+00 -9.60294843e-01 -1.35734594e+00 1.18608344e+00 -1.64419964e-01 8.70398700e-01 -4.93641734e-01 -6.32919133e-01 3.91037554e-01 9.17278767e-01 -7.90360451e-01 1.08255911e+00 1.67899147e-01 8.68343711e-02 -7.70403892e-02 -1.17545390e+00 5.12972176e-01 1.96955964e-01 -1.76553518e-01 -7.54840374e-01 7.55294085e-01 9.82528389e-01 6.19982183e-02 -7.58855581e-01 -3.12024564e-01 3.96193154e-02 -8.13813746e-01 6.43675089e-01 -3.69847983e-01 4.78867218e-02 -3.10261726e-01 8.25060457e-02 -1.42573166e+00 -5.91351017e-02 -8.82687092e-01 2.72357285e-01 1.50977349e+00 2.73577899e-01 -8.99821162e-01 7.43040204e-01 2.44919673e-01 2.37759560e-01 -2.23630086e-01 -9.13909614e-01 -4.38842624e-01 -2.62626499e-01 -2.35564247e-01 5.48904896e-01 1.38254666e+00 2.41271853e-01 4.65201885e-01 -2.74088353e-01 2.18076169e-01 5.82653522e-01 2.18058661e-01 8.48220110e-01 -1.57906437e+00 -4.52365959e-03 -1.52284756e-01 -4.76386130e-01 -4.73206431e-01 -1.77589521e-01 -7.29176879e-01 -1.87864423e-01 -1.46021044e+00 4.69229460e-01 -5.78493237e-01 -1.00094892e-01 1.08205797e-02 -1.38719141e-01 4.70232107e-02 1.78185657e-01 3.41396183e-01 -1.00178313e+00 4.39590126e-01 6.93828225e-01 -2.09865972e-01 -4.08099264e-01 3.38977352e-02 -6.46991670e-01 7.97307491e-01 1.13023341e+00 -9.17208433e-01 -3.69991779e-01 8.30291212e-02 3.98263693e-01 -1.81976318e-01 -9.05172974e-02 -9.27763939e-01 3.25534880e-01 -1.36629209e-01 -3.06505620e-01 -9.43426013e-01 3.36177826e-01 -8.00964355e-01 2.54915327e-01 6.42117202e-01 -3.28845054e-01 5.18876612e-01 1.86367050e-01 7.32118726e-01 -3.19919080e-01 1.24496613e-02 4.81810093e-01 2.15430543e-01 -3.93678188e-01 1.07138768e-01 -7.74201930e-01 3.97923440e-01 1.47523689e+00 3.34065348e-01 -6.95432544e-01 -5.13959408e-01 -5.13046384e-01 3.43704671e-01 4.38614674e-02 5.88603377e-01 5.69072604e-01 -1.23135197e+00 -9.22733247e-01 -2.90451050e-01 1.05575293e-01 -2.27405727e-01 1.70412436e-01 1.08866775e+00 -3.73901904e-01 6.03597283e-01 1.00619301e-01 -5.87288857e-01 -1.56583047e+00 4.93042737e-01 -4.46692854e-01 -5.38852930e-01 -6.02027178e-01 5.30469358e-01 2.62710273e-01 -2.16198742e-01 4.34189029e-02 -2.18859389e-02 -8.68448853e-01 6.67073667e-01 7.60589898e-01 5.89417398e-01 -1.67330042e-01 -8.48698556e-01 -5.57790399e-01 5.23927510e-01 -3.85619134e-01 -1.50128916e-01 1.31633294e+00 -2.92686045e-01 -4.67101961e-01 7.39329517e-01 1.37988067e+00 6.08367920e-02 -3.39980304e-01 -2.60127932e-01 4.89179999e-01 -1.43318340e-01 -5.96384518e-02 -4.20862198e-01 -7.16568410e-01 5.91703475e-01 7.11779594e-01 8.82526755e-01 7.06304193e-01 1.60987228e-01 9.94998276e-01 2.39022732e-01 1.07133150e-01 -1.18010616e+00 1.17092468e-01 2.52590179e-01 6.60064459e-01 -1.19491518e+00 7.87112117e-02 -4.07201558e-01 -6.92504883e-01 7.37669766e-01 2.97724932e-01 3.76283862e-02 1.08724606e+00 -1.09365135e-01 3.44392136e-02 -5.79336703e-01 -7.17080116e-01 -8.86655673e-02 7.69888535e-02 2.76930958e-01 4.34383303e-01 2.31988594e-01 -6.89344823e-01 2.34223634e-01 -9.86684710e-02 -5.02992809e-01 5.12570143e-01 8.74720454e-01 -6.59205079e-01 -9.52502728e-01 -6.08620405e-01 5.03822386e-01 -6.58045053e-01 1.33721381e-01 -4.63646382e-01 9.18904006e-01 1.88395828e-01 1.31718695e+00 2.85937577e-01 -4.30496097e-01 -3.93189698e-01 -1.79901361e-01 -6.20976925e-01 -6.13514543e-01 -8.13005567e-01 1.70715183e-01 2.21937224e-01 -6.23545833e-02 -7.79743552e-01 -9.24527824e-01 -1.27521181e+00 -2.69080669e-01 -3.91696841e-01 9.92803216e-01 7.84669697e-01 8.01267743e-01 5.52968502e-01 1.80395812e-01 1.01681781e+00 -3.02739769e-01 1.30962789e-01 -8.93457770e-01 -3.99522960e-01 5.00836730e-01 1.50172099e-01 -7.11701572e-01 -2.76400983e-01 -2.45854795e-01]
[10.335811614990234, 7.315977573394775]
1dd50640-b359-4e8e-920a-3b75bd03b4f7
videonavqa-bridging-the-gap-between-visual
1908.04950
null
https://arxiv.org/abs/1908.04950v1
https://arxiv.org/pdf/1908.04950v1.pdf
VideoNavQA: Bridging the Gap between Visual and Embodied Question Answering
Embodied Question Answering (EQA) is a recently proposed task, where an agent is placed in a rich 3D environment and must act based solely on its egocentric input to answer a given question. The desired outcome is that the agent learns to combine capabilities such as scene understanding, navigation and language understanding in order to perform complex reasoning in the visual world. However, initial advancements combining standard vision and language methods with imitation and reinforcement learning algorithms have shown EQA might be too complex and challenging for these techniques. In order to investigate the feasibility of EQA-type tasks, we build the VideoNavQA dataset that contains pairs of questions and videos generated in the House3D environment. The goal of this dataset is to assess question-answering performance from nearly-ideal navigation paths, while considering a much more complete variety of questions than current instantiations of the EQA task. We investigate several models, adapted from popular VQA methods, on this new benchmark. This establishes an initial understanding of how well VQA-style methods can perform within this novel EQA paradigm.
['Pietro Liò', 'Cătălina Cangea', 'Eugene Belilovsky', 'Aaron Courville']
2019-08-14
null
null
null
null
['embodied-question-answering']
['computer-vision']
[-7.16202557e-02 1.86125696e-01 5.48892856e-01 -3.05613875e-01 -6.84470713e-01 -8.86790693e-01 1.00207233e+00 -1.78057671e-01 -5.38571775e-01 5.15705943e-01 2.28993624e-01 -5.98865032e-01 -2.21889228e-01 -7.26240396e-01 -7.42702365e-01 -2.70502895e-01 -3.87072526e-02 7.67980456e-01 2.62470454e-01 -7.65719354e-01 3.79567474e-01 3.66992265e-01 -1.68078125e+00 2.21524835e-01 6.42697930e-01 6.52208686e-01 5.24786115e-01 1.19775701e+00 -1.63241312e-01 1.32702076e+00 -3.34851146e-01 -2.44059771e-01 3.30883473e-01 -6.42601728e-01 -1.40357673e+00 -4.57880553e-03 6.67545736e-01 -6.61552906e-01 -5.81463754e-01 6.14598691e-01 2.89262950e-01 4.63538855e-01 6.03816211e-01 -1.43143749e+00 -5.77501237e-01 -7.03111887e-02 2.17222288e-01 2.65705377e-01 1.28320086e+00 6.09288692e-01 1.13408566e+00 -7.40079522e-01 1.05103993e+00 1.41431475e+00 3.78324360e-01 7.48337805e-01 -9.94802833e-01 2.98911966e-02 2.23551393e-01 5.35999537e-01 -8.02794516e-01 -5.39285660e-01 4.55265999e-01 -2.23129615e-01 1.33767164e+00 1.14759095e-01 7.13670075e-01 1.21067870e+00 2.81141847e-01 1.06130791e+00 1.01594555e+00 -2.82029867e-01 5.83691180e-01 -1.34973645e-01 2.41754316e-02 9.57172394e-01 -1.78828627e-01 7.80729949e-02 -7.74880350e-01 8.90585035e-02 7.87809134e-01 -4.01020378e-01 -3.58355850e-01 -1.22292018e+00 -1.54291737e+00 8.03174555e-01 6.25218987e-01 1.73889130e-01 -5.31629264e-01 3.44891667e-01 1.66607022e-01 7.36355960e-01 -2.59243757e-01 9.72666681e-01 -3.31805348e-01 -4.05423403e-01 -3.70667875e-01 9.44247901e-01 1.12275219e+00 8.85043263e-01 6.65844083e-01 -2.81604361e-02 -1.70106217e-01 1.81481004e-01 4.99611109e-01 5.08692741e-01 1.58823326e-01 -1.80502498e+00 4.61645275e-01 5.92888653e-01 4.09042209e-01 -8.59932959e-01 -6.73465312e-01 1.52814582e-01 6.58302102e-03 8.13994229e-01 1.03212321e+00 4.08635065e-02 -9.11799252e-01 1.87043822e+00 5.02614439e-01 -1.48725510e-01 5.17284274e-01 1.05668318e+00 1.02186441e+00 6.07714117e-01 -9.98307541e-02 3.88653189e-01 1.33994079e+00 -1.17826509e+00 -4.77125585e-01 -3.13672245e-01 7.69928098e-01 -2.36119404e-01 1.38475013e+00 2.95663089e-01 -1.33345544e+00 -5.23804486e-01 -9.32819486e-01 -5.96412957e-01 -4.97767478e-01 -5.01581907e-01 6.57343745e-01 5.36775649e-01 -1.48013783e+00 2.81530708e-01 -6.98441386e-01 -9.54487741e-01 2.97856152e-01 7.70172477e-02 -6.89684808e-01 -3.38807940e-01 -9.36366618e-01 1.27092803e+00 -2.37418450e-02 4.83774617e-02 -1.57520115e+00 -3.94326329e-01 -1.19050312e+00 -1.03562653e-01 5.93182623e-01 -1.32070529e+00 1.70302224e+00 -5.01578689e-01 -1.53090954e+00 9.74732578e-01 -8.75866115e-02 -4.71099824e-01 5.65705597e-01 -2.64008284e-01 2.56443441e-01 5.83428323e-01 3.12242836e-01 9.47172403e-01 5.43832481e-01 -1.29777658e+00 -4.10948634e-01 -7.39742577e-01 9.84941483e-01 6.64082944e-01 6.01088762e-01 -4.43427324e-01 -3.04183751e-01 2.00447235e-02 5.09246923e-02 -9.02506351e-01 -2.73473978e-01 3.35466057e-01 8.84998292e-02 -2.14074239e-01 6.88346982e-01 -5.98557234e-01 1.92346290e-01 -1.58311856e+00 4.79376048e-01 -1.38773859e-01 3.11336875e-01 -1.15267432e-03 -4.70813543e-01 7.69686639e-01 2.25853086e-01 -2.18407273e-01 -2.07308844e-01 -3.14777762e-01 2.71228015e-01 5.58501482e-01 -3.48699003e-01 2.92952627e-01 1.17148384e-01 1.39769292e+00 -1.23296368e+00 -3.10578316e-01 2.85860002e-01 1.50517866e-01 -8.85115325e-01 5.09463191e-01 -6.64117575e-01 7.50273883e-01 -5.80823660e-01 5.29873431e-01 1.44444138e-01 -1.59030214e-01 -6.57630339e-02 2.73199707e-01 4.49235775e-02 1.51128560e-01 -8.13901246e-01 2.54674053e+00 -4.72228587e-01 8.03726494e-01 3.44226718e-01 -9.69871461e-01 4.85083222e-01 2.63274074e-01 2.77574688e-01 -1.09328437e+00 1.58982262e-01 -8.07847977e-02 2.97036320e-02 -1.16653037e+00 4.96868879e-01 -7.86044300e-02 -5.37261963e-02 4.69350874e-01 3.44881564e-01 -8.46137583e-01 2.49109551e-01 6.57077670e-01 1.51233757e+00 7.70940781e-01 1.48530319e-01 -2.90369749e-01 4.78298247e-01 6.29653513e-01 -2.01177597e-01 1.24651647e+00 -7.03347147e-01 5.80227911e-01 5.09438515e-01 -3.99483919e-01 -1.02982020e+00 -1.44575489e+00 4.67669457e-01 1.06744838e+00 3.66946667e-01 -1.75848991e-01 -7.78678834e-01 -7.77432978e-01 -2.85450190e-01 1.08584571e+00 -8.19069624e-01 8.34932271e-03 -3.59308392e-01 1.65717155e-01 5.44857502e-01 3.91143113e-01 6.66922390e-01 -1.38155591e+00 -1.44169867e+00 4.14825715e-02 -3.25148314e-01 -1.22072065e+00 6.66844919e-02 -1.77071959e-01 -5.26709497e-01 -1.41926956e+00 -9.10074115e-01 -8.09532583e-01 3.23868304e-01 5.24401248e-01 1.58112359e+00 1.13789104e-02 -1.28461629e-01 1.58142793e+00 -6.62258804e-01 -2.98455983e-01 -3.34221601e-01 -1.88965484e-01 -2.01472566e-01 -4.30777401e-01 2.67466158e-01 -4.74361151e-01 -7.53552258e-01 1.70185149e-01 -7.82101691e-01 2.04575583e-02 3.54925275e-01 6.81217432e-01 -4.06423248e-02 -5.03676593e-01 4.48902875e-01 -3.74135047e-01 8.55474234e-01 -5.11018336e-01 -3.43337059e-01 2.27379978e-01 -3.33972350e-02 1.29050091e-01 3.93779427e-01 -1.27534926e-01 -1.19740069e+00 -1.55417830e-01 -4.40089822e-01 -1.13259256e-01 -3.96422118e-01 3.69762242e-01 -1.67845264e-01 -7.95071796e-02 8.84127438e-01 4.10048723e-01 2.38497242e-01 2.92007662e-02 8.76818061e-01 5.78349270e-02 7.90252268e-01 -7.52525806e-01 6.78680897e-01 7.29656696e-01 2.15651728e-02 -9.15654778e-01 -3.89954448e-01 -4.79464769e-01 -6.54113889e-01 -3.68641913e-01 1.22087431e+00 -7.99393952e-01 -1.09697270e+00 2.43072644e-01 -1.22339189e+00 -8.17180991e-01 -3.84882540e-01 2.90328473e-01 -1.26270545e+00 4.04713571e-01 -1.56213731e-01 -7.72117853e-01 9.88341048e-02 -1.28632593e+00 1.12302077e+00 3.32629174e-01 -3.68585736e-01 -1.03540266e+00 3.98053497e-01 7.43682444e-01 4.09348816e-01 1.49230376e-01 8.59788060e-01 -6.01096988e-01 -8.70571971e-01 2.12873995e-01 -1.86357915e-01 -1.72056571e-01 -2.68620610e-01 -5.31846225e-01 -7.99804389e-01 -2.42013007e-01 2.56173521e-01 -9.19468224e-01 6.49279833e-01 1.07772062e-02 5.81683874e-01 3.01193804e-01 3.47009748e-02 2.32560411e-01 1.21214676e+00 3.71065021e-01 7.07643986e-01 5.24172843e-01 3.14202189e-01 8.20913970e-01 5.34502327e-01 9.65111181e-02 1.15671885e+00 6.16017222e-01 9.15531039e-01 2.21652344e-01 -2.75519248e-02 -2.99121618e-01 3.42146516e-01 2.54691958e-01 3.55814621e-02 -2.90743589e-01 -1.20250070e+00 7.39452362e-01 -1.89523935e+00 -1.09556615e+00 -3.66716795e-02 1.68787360e+00 3.01217049e-01 -1.53675675e-01 1.88234653e-02 -1.44704327e-01 -1.81342736e-01 2.70555854e-01 -7.56760716e-01 -5.67798078e-01 1.08382113e-01 1.09711416e-01 -2.30908945e-01 4.50803548e-01 -7.00755835e-01 9.15561795e-01 7.03621101e+00 6.80595711e-02 -4.74501789e-01 -4.88022342e-02 1.12282611e-01 1.57410204e-02 -3.54839742e-01 5.92017174e-02 -1.41719997e-01 -2.45132312e-01 7.20339715e-01 1.70947328e-01 6.60354316e-01 6.97131336e-01 8.45343098e-02 -6.46024644e-01 -1.59480500e+00 1.03024065e+00 4.18731958e-01 -1.21446431e+00 6.69967532e-02 -2.59815425e-01 4.18260217e-01 1.19252503e-01 6.85529560e-02 9.56575036e-01 6.31334782e-01 -1.17109632e+00 6.91635430e-01 6.54669166e-01 6.07923828e-02 -1.82029977e-01 4.27067935e-01 6.08721972e-01 -8.49562705e-01 -1.93287984e-01 -1.17632300e-01 -5.09978235e-01 4.26852047e-01 -6.07162774e-01 -8.45859110e-01 6.04694366e-01 8.09741855e-01 3.06356311e-01 -5.97148597e-01 9.27518070e-01 -2.46203914e-01 3.33989561e-02 -1.59391224e-01 -2.02064693e-01 7.16556668e-01 -2.84566730e-01 6.92712247e-01 5.84519804e-01 1.32999867e-01 4.39407468e-01 -5.15831448e-02 8.80699277e-01 2.51282871e-01 -6.24992102e-02 -1.04088378e+00 1.13177866e-01 -2.27667555e-01 9.51881826e-01 -5.74784756e-01 -1.46209881e-01 -5.23860276e-01 1.14570940e+00 5.09789467e-01 6.38034165e-01 -8.02463114e-01 -1.37578502e-01 8.05253208e-01 -1.46239057e-01 2.28118598e-01 -6.95319116e-01 1.87198639e-01 -1.30488431e+00 -4.42925021e-02 -1.26556540e+00 2.31389895e-01 -1.67713022e+00 -6.92311406e-01 3.72961402e-01 -2.84118801e-02 -7.15665817e-01 -6.40880048e-01 -9.95792329e-01 -5.81711888e-01 5.17435253e-01 -1.59215713e+00 -1.16861129e+00 -7.55726337e-01 9.20221090e-01 9.64049876e-01 -1.39269426e-01 8.59634936e-01 -2.37641677e-01 -3.09971664e-02 -1.26470879e-01 -1.47870541e-01 -8.61093551e-02 4.30749148e-01 -1.50468159e+00 6.80153549e-01 6.67157769e-01 5.43294787e-01 4.15230393e-01 8.20492625e-01 -2.51441002e-01 -1.95225036e+00 -3.33484709e-01 3.78490090e-01 -1.34818649e+00 5.61500490e-01 -4.55506921e-01 -7.84339786e-01 8.32777321e-01 5.43803453e-01 -2.77035683e-01 2.67179281e-01 -1.42476827e-01 -4.27454770e-01 4.43029553e-01 -1.08260477e+00 1.02385402e+00 1.36462986e+00 -7.83597827e-01 -1.27829182e+00 2.67324567e-01 5.75924158e-01 -4.30962890e-01 -4.83953327e-01 2.31925547e-01 6.17394328e-01 -1.37151563e+00 1.15832496e+00 -9.85930860e-01 5.41747153e-01 -4.66952950e-01 -5.23773313e-01 -1.34070790e+00 -3.76934558e-02 -5.61787605e-01 -4.94353436e-02 6.59983754e-01 1.89153880e-01 -4.51808393e-01 8.23263407e-01 7.29581594e-01 2.99846698e-02 -4.61724252e-01 -1.03965807e+00 -5.02140880e-01 2.67453343e-01 -6.40945673e-01 3.19370538e-01 5.99520862e-01 4.95160595e-02 5.21378875e-01 5.57902306e-02 1.17490821e-01 4.84745950e-01 7.01497309e-03 1.32482994e+00 -9.10843432e-01 -2.34579906e-01 -3.72629970e-01 -3.72307450e-01 -1.47420192e+00 2.50319719e-01 -8.34290445e-01 2.71960586e-01 -2.12347484e+00 -1.79741517e-01 -6.12700416e-04 1.41810432e-01 1.06732734e-01 6.59293160e-02 -7.63490647e-02 4.97507423e-01 -3.87363620e-02 -1.01608503e+00 7.82269299e-01 1.58031440e+00 -1.37510195e-01 -7.01067969e-02 -3.71496767e-01 -4.76791829e-01 7.31841505e-01 5.88832021e-01 1.72729462e-01 -8.73471141e-01 -8.97896886e-01 4.83226568e-01 4.52368319e-01 8.30894172e-01 -9.76133466e-01 4.85444903e-01 1.01101637e-01 3.41184586e-01 -5.72935045e-01 7.49981463e-01 -8.08222353e-01 -2.78109819e-01 4.02173370e-01 -3.23676288e-01 2.80794293e-01 1.96296513e-01 6.79431379e-01 -1.07941113e-01 -1.62083283e-01 2.38307968e-01 -6.46659911e-01 -1.44422150e+00 -2.35306006e-03 -7.41056919e-01 3.98683518e-01 1.11252654e+00 -4.50447261e-01 -2.95908898e-01 -1.03402150e+00 -7.39542782e-01 7.89857626e-01 6.15447998e-01 5.22541106e-01 8.04767966e-01 -9.86299753e-01 -6.48751557e-01 -1.88326351e-02 4.31558818e-01 4.26528305e-02 4.05817777e-01 5.93555391e-01 -9.66523468e-01 6.14390910e-01 -5.22445798e-01 -6.49140537e-01 -8.85863483e-01 6.73708320e-01 6.04979217e-01 -6.86887801e-02 -5.46625376e-01 9.17065918e-01 5.51349461e-01 -9.75048065e-01 7.53935203e-02 -2.72439271e-01 -4.53356057e-01 -2.32843891e-01 4.21451509e-01 2.19264314e-01 -3.18060756e-01 -6.12192154e-01 -3.54569048e-01 5.24882674e-01 2.26081789e-01 -5.73712826e-01 9.69626665e-01 -4.08448815e-01 2.91861653e-01 2.43446454e-01 9.02714908e-01 -3.63936961e-01 -1.31472349e+00 1.50851132e-02 -1.01270445e-01 -3.83985788e-01 -2.70658404e-01 -7.94462562e-01 -2.84714967e-01 1.10785580e+00 3.57910216e-01 2.42417559e-01 7.06250489e-01 1.75509900e-01 4.23017830e-01 1.03974175e+00 7.03314483e-01 -8.44297051e-01 7.09986091e-01 7.22042203e-01 1.16196370e+00 -1.45479226e+00 -2.77378231e-01 2.06393704e-01 -7.49234498e-01 9.20651615e-01 7.83628702e-01 -1.55882478e-01 1.75597504e-01 -4.40873533e-01 3.98807257e-01 -5.39637446e-01 -8.04123342e-01 -6.01890564e-01 -7.83180147e-02 9.99560177e-01 -1.77847922e-01 -4.92870331e-01 3.85072827e-01 -1.36710210e-02 -4.09116775e-01 -1.68132380e-01 7.27181017e-01 1.10530388e+00 -3.39503407e-01 -6.06080651e-01 -2.86797881e-01 -5.28053902e-02 -1.07873715e-02 3.45634699e-01 -5.42612553e-01 1.13691890e+00 -2.77090698e-01 1.31791580e+00 -8.19802377e-03 -2.11118795e-02 5.85496247e-01 1.67750806e-01 8.77144396e-01 -5.15001237e-01 -2.66311616e-01 -6.54571652e-01 1.83547854e-01 -1.13822758e+00 -6.63593650e-01 -7.27361977e-01 -1.31037366e+00 -1.15129049e-03 1.49406642e-01 2.34037898e-02 7.68727601e-01 1.14663100e+00 2.18067303e-01 4.51222450e-01 -2.82963342e-03 -1.05745304e+00 -4.96370047e-01 -5.71153879e-01 -1.50632963e-01 5.55299819e-01 5.28836191e-01 -5.94001055e-01 -2.75195390e-01 -2.21309677e-01]
[4.390598773956299, 0.566823422908783]
93ccd812-807d-44c2-8e0d-a0428ed80c51
controlling-epidemic-spread-using
2202.08296
null
https://arxiv.org/abs/2202.08296v1
https://arxiv.org/pdf/2202.08296v1.pdf
Controlling Epidemic Spread using Probabilistic Diffusion Models on Networks
The spread of an epidemic is often modeled by an SIR random process on a social network graph. The MinINF problem for optimal social distancing involves minimizing the expected number of infections, when we are allowed to break at most $B$ edges; similarly the MinINFNode problem involves removing at most $B$ vertices. These are fundamental problems in epidemiology and network science. While a number of heuristics have been considered, the complexity of these problems remains generally open. In this paper, we present two bicriteria approximation algorithms for MinINF, which give the first non-trivial approximations for this problem. The first is based on the cut sparsification result of Karger \cite{karger:mathor99}, and works when the transmission probabilities are not too small. The second is a Sample Average Approximation (SAA) based algorithm, which we analyze for the Chung-Lu random graph model. We also extend some of our results to tackle the MinINFNode problem.
['Anil Vullikanti', 'Leonidas Tsepenekas', 'Aravind Srinivasan', 'Michael Dinitz', 'Amy Babay']
2022-02-16
null
null
null
null
['epidemiology']
['medical']
[ 1.70436069e-01 6.84225261e-01 -1.59169883e-01 2.87201196e-01 8.09126720e-03 -5.05746901e-01 1.55731484e-01 3.42183322e-01 -4.59301680e-01 9.50722992e-01 -5.19216001e-01 -5.18800080e-01 -9.01540220e-01 -1.24281907e+00 -6.78467155e-01 -7.42641568e-01 -8.59795511e-01 1.13956904e+00 2.84098893e-01 -4.47575897e-01 9.65334177e-02 4.25517321e-01 -9.04457927e-01 -5.22042274e-01 7.93477654e-01 6.30877912e-01 -2.03628421e-01 1.06689823e+00 -1.45487279e-01 3.88484448e-01 -5.69283485e-01 -5.81992209e-01 4.48722661e-01 -5.30251682e-01 -9.65320230e-01 1.69378221e-02 -5.55939913e-01 -4.00871672e-02 -2.08662257e-01 1.10583532e+00 3.38142097e-01 -2.09502712e-01 7.43370771e-01 -1.81288958e+00 -1.38013631e-01 1.21712363e+00 -1.26139557e+00 4.13053274e-01 3.95154446e-01 -5.54279327e-01 8.22563767e-01 6.74192756e-02 8.97285819e-01 1.12754965e+00 6.74263716e-01 2.04016581e-01 -1.16178155e+00 -6.30430460e-01 6.17344230e-02 -1.72314018e-01 -1.34444726e+00 3.46390605e-02 4.53956723e-01 -2.66062200e-01 7.14957833e-01 5.97558975e-01 1.03617716e+00 3.10883552e-01 1.87639281e-01 3.72854114e-01 8.79307628e-01 -4.54392552e-01 2.45110646e-01 -3.16218520e-03 1.83083192e-01 6.18629158e-01 1.27520955e+00 -1.49167985e-01 5.18528819e-02 -6.76681757e-01 8.46164286e-01 2.30400398e-01 -3.73226851e-01 -1.52523935e-01 -6.33240998e-01 1.06445253e+00 3.28081906e-01 4.66882288e-01 -3.67518395e-01 4.44929183e-01 3.85488868e-02 5.88023365e-01 6.70853019e-01 1.59730390e-01 -3.41796905e-01 1.92097232e-01 -8.90988052e-01 3.47196847e-01 1.40676510e+00 9.68895972e-01 6.32110178e-01 -4.36567694e-01 4.83609915e-01 3.55300218e-01 2.86565423e-01 7.86846995e-01 -7.29854703e-01 -9.74532187e-01 3.58858466e-01 2.49178588e-01 1.80557251e-01 -1.21082854e+00 -6.01137519e-01 -2.46285141e-01 -1.14759481e+00 -1.65954992e-01 4.82919097e-01 -4.81650531e-01 -5.37446141e-01 1.73380637e+00 3.47744286e-01 1.39794558e-01 -3.87250483e-01 2.25667477e-01 1.29150763e-01 8.12590420e-01 -3.28431964e-01 -1.06624472e+00 9.68098402e-01 -6.23160303e-01 -6.80973053e-01 1.85074080e-02 6.75901890e-01 -6.70753896e-01 -1.20784529e-01 5.44860303e-01 -1.41691935e+00 5.62482655e-01 -5.63404620e-01 5.60959280e-01 -1.39200628e-01 -8.40642095e-01 6.53301835e-01 8.48117709e-01 -1.63708186e+00 4.97432679e-01 -5.88547230e-01 -8.63702238e-01 2.54749805e-01 4.89606708e-01 4.81992550e-02 -4.56470788e-01 -1.05937684e+00 4.89518732e-01 6.59477487e-02 -8.32056903e-05 -4.37874913e-01 -3.56650800e-01 -3.90948206e-01 -3.45498770e-02 6.78268611e-01 -6.97248518e-01 7.64117539e-01 -6.19627237e-01 -7.49949336e-01 7.69931674e-01 -3.31123434e-02 -5.58330774e-01 5.99518716e-01 3.11809540e-01 -4.92597669e-02 5.21147966e-01 -2.82881223e-02 1.31897017e-01 2.67760932e-01 -1.26884437e+00 -4.20140207e-01 -4.93720472e-01 3.72359604e-01 -1.55380264e-01 -9.55132917e-02 3.75781655e-01 -3.18738967e-01 -4.82829213e-01 -5.06950542e-02 -1.14189100e+00 -9.41361070e-01 -1.48989037e-01 -7.09385872e-01 -3.42671871e-01 3.88860554e-01 -2.06322834e-01 1.38041437e+00 -1.46846688e+00 2.79897869e-01 8.66705060e-01 5.50604284e-01 1.67155378e-02 -1.38177946e-01 1.02678406e+00 8.17030743e-02 6.89078033e-01 -4.06131238e-01 1.17050268e-01 -3.16438526e-01 2.39501566e-01 1.47677943e-01 7.75678754e-01 -4.59759742e-01 4.84705240e-01 -8.13661754e-01 -4.83396709e-01 -4.57789123e-01 2.07171142e-01 -7.98715115e-01 -1.47000447e-01 -7.15161264e-02 -2.31401473e-01 -6.63151264e-01 2.86580175e-01 1.07750332e+00 -5.08294404e-01 5.17695248e-01 5.42850137e-01 6.58541694e-02 -2.09374532e-01 -1.34216774e+00 9.69092131e-01 -5.30204102e-02 1.67571768e-01 7.15595007e-01 -1.17445314e+00 4.16920185e-01 3.16008657e-01 9.64815557e-01 1.19780399e-01 4.58541960e-01 2.01489851e-01 -3.35223041e-02 -3.73227596e-01 7.72064254e-02 -3.60228330e-01 -1.13935746e-01 1.09914708e+00 -6.31056070e-01 1.10886499e-01 5.90837717e-01 8.25537920e-01 1.76214266e+00 -7.12796628e-01 3.41615260e-01 -5.35231471e-01 -9.76522267e-02 -2.09822017e-03 4.37404841e-01 1.03954589e+00 -1.52066693e-01 7.67759904e-02 1.08904767e+00 -1.08899772e-01 -7.61105359e-01 -1.10346210e+00 8.34023729e-02 4.90320474e-01 3.46028149e-01 -5.33652246e-01 -1.08812225e+00 -4.81758624e-01 -4.86212969e-02 4.18050021e-01 -7.56889641e-01 7.88423195e-02 -3.36922139e-01 -1.15415347e+00 2.50451088e-01 -5.34794405e-02 9.41323861e-02 -6.86424077e-01 -3.57693017e-01 3.41730803e-01 -2.54532039e-01 -7.90115595e-01 -2.86597133e-01 -1.04295202e-02 -8.09223711e-01 -1.52416277e+00 -7.11009264e-01 -5.74683189e-01 8.50543141e-01 3.47921103e-01 9.11915600e-01 6.62858665e-01 -4.86756027e-01 5.31281471e-01 -2.76631504e-01 -4.92851675e-01 -2.65232921e-01 1.03500672e-01 -8.47155042e-03 -3.36054772e-01 1.69292644e-01 -6.18998826e-01 -7.58284986e-01 1.98110223e-01 -1.03965020e+00 -3.80501360e-01 3.34688395e-01 1.64305687e-01 4.27488446e-01 4.95246768e-01 6.82130694e-01 -1.29832518e+00 7.64572382e-01 -1.00463295e+00 -4.63984936e-01 2.25243241e-01 -4.87689525e-01 -1.15418516e-01 3.21281761e-01 -1.32233188e-01 -4.78398919e-01 -3.06877404e-01 -1.41327441e-01 1.13423727e-01 3.20159614e-01 5.94453156e-01 1.69706270e-01 -1.47744343e-01 2.40692630e-01 -3.42779368e-01 8.00519958e-02 -2.03315154e-01 1.86762482e-01 5.25239825e-01 -4.05526638e-01 -1.73495784e-01 8.47726464e-01 7.12549746e-01 3.88934046e-01 -1.18286479e+00 -4.64556754e-01 -3.94377023e-01 8.89258534e-02 -2.41223514e-01 6.28024042e-01 -2.26522669e-01 -1.08112168e+00 3.52869928e-01 -1.12731624e+00 -5.27711511e-01 -3.26162547e-01 1.50427610e-01 -5.93449950e-01 4.81462806e-01 -9.27252769e-01 -1.30146432e+00 -2.19107106e-01 -6.25771582e-01 3.20469946e-01 -4.30728793e-02 -9.36599597e-02 -1.07515299e+00 3.83315682e-01 1.24607921e-01 4.80892956e-01 6.22400343e-01 8.41414690e-01 -6.94300890e-01 -5.67253828e-01 -2.41517052e-01 -2.43162170e-01 -3.80061090e-01 1.65040456e-02 2.27228209e-01 9.08831693e-03 -5.20684540e-01 -1.61751047e-01 3.44624482e-02 8.43424559e-01 7.77023673e-01 9.08808827e-01 -7.10086942e-01 -9.69763219e-01 2.49768436e-01 1.67372358e+00 1.23857722e-01 6.75344586e-01 2.31877435e-02 2.65138239e-01 7.11593926e-01 3.07325423e-01 5.74075103e-01 5.11408985e-01 1.96661949e-01 5.99070191e-01 -1.40927628e-01 4.88003135e-01 1.32450894e-01 -3.48318592e-02 7.47519076e-01 -5.34594715e-01 -9.76212859e-01 -8.39761019e-01 5.89490533e-01 -1.72873747e+00 -1.18997848e+00 -5.71006179e-01 2.09841204e+00 6.67212486e-01 1.92277193e-01 5.80095887e-01 4.32498574e-01 1.00218666e+00 3.66769619e-02 -1.70284599e-01 -6.51215732e-01 1.23696037e-01 3.40985119e-01 8.78863335e-01 8.92084956e-01 -4.79548156e-01 5.72378278e-01 6.83317232e+00 8.23632002e-01 -3.14233303e-01 1.43993899e-01 8.06581736e-01 -2.41596669e-01 -6.07413054e-01 6.49121553e-02 -6.16382062e-01 2.39570037e-01 1.15003932e+00 -6.13079727e-01 7.53911674e-01 5.13047397e-01 1.80152074e-01 -4.37537491e-01 -5.50428748e-01 5.23537934e-01 -8.20482150e-02 -1.03784466e+00 -2.32587442e-01 6.67616963e-01 9.75670695e-01 -1.09952748e-01 -2.51947016e-01 -1.84929013e-01 7.77825356e-01 -1.02969539e+00 6.26287088e-02 2.71332353e-01 6.92323625e-01 -1.27602708e+00 5.07854879e-01 5.69101334e-01 -1.17765820e+00 -4.44463305e-02 -3.59136701e-01 -4.15039919e-02 7.01439857e-01 1.05760849e+00 -2.79169440e-01 4.94618565e-01 6.47831738e-01 2.58366078e-01 7.23318383e-02 1.27120233e+00 1.46762893e-01 5.33701897e-01 -9.44448173e-01 -1.77934662e-01 8.36562142e-02 -4.53773528e-01 7.76253879e-01 8.72792542e-01 4.33670282e-01 5.62345803e-01 1.09042890e-01 4.55915481e-01 -1.59429207e-01 1.79347172e-01 -8.16952348e-01 -1.61104292e-01 4.51254219e-01 1.14892852e+00 -1.32482660e+00 6.48166984e-02 2.20287573e-02 7.28056073e-01 2.00162768e-01 4.97281551e-01 -6.49289370e-01 -7.54359901e-01 5.90923965e-01 6.25532091e-01 4.46226835e-01 -9.59851146e-02 2.42429134e-02 -8.00543070e-01 -4.03439045e-01 -4.85168546e-01 6.14269435e-01 -1.83803350e-01 -1.09282911e+00 6.44567192e-01 2.52427399e-01 -2.79311895e-01 -1.76145479e-01 -1.64524704e-01 -6.60058498e-01 3.88176382e-01 -1.16996491e+00 -5.83749354e-01 3.13297480e-01 5.20151317e-01 -3.16856205e-02 6.55931950e-01 3.26374531e-01 2.11564571e-01 -5.70994675e-01 2.37701967e-01 2.02814609e-01 -2.20525384e-01 7.18243141e-03 -9.98656154e-01 1.39472960e-02 9.43238378e-01 -4.03918266e-01 5.47240853e-01 1.19475853e+00 -9.42475140e-01 -1.41227341e+00 -9.31533575e-01 1.15573215e+00 3.86589654e-02 6.45572364e-01 -3.45210075e-01 -4.39679950e-01 7.89129674e-01 2.65398830e-01 -9.24715623e-02 6.07522964e-01 -2.70324200e-02 2.18197480e-01 -6.40941337e-02 -1.55719316e+00 5.28515816e-01 1.45972264e+00 2.76438296e-01 2.02297002e-01 6.19268775e-01 6.16704226e-01 1.40803069e-01 -8.07648480e-01 1.58253357e-01 2.62340933e-01 -7.39523709e-01 8.58023942e-01 -5.45756519e-01 1.54402748e-01 -4.70511243e-02 2.23243043e-01 -1.34616685e+00 -2.68489361e-01 -1.07659972e+00 1.05961777e-01 9.50083613e-01 5.62447548e-01 -9.49859262e-01 8.06195915e-01 1.30637929e-01 7.62276173e-01 -1.01088095e+00 -1.10786986e+00 -8.50487709e-01 1.72605351e-01 -1.75600961e-01 4.39334184e-01 8.95820796e-01 2.87905127e-01 1.68927163e-01 -3.92061770e-01 3.17334756e-02 1.13647199e+00 -1.06143169e-01 5.69922984e-01 -1.50117218e+00 -4.21691209e-01 -4.94290769e-01 -7.27830082e-02 -8.84544253e-01 -2.10485697e-01 -6.69113398e-01 -7.59358704e-02 -1.98512435e+00 3.36169064e-01 -9.38395917e-01 1.52909085e-01 4.16189767e-02 8.40203464e-02 1.70172632e-01 -1.93911716e-02 -3.92898545e-02 -7.42858350e-01 -5.71399257e-02 9.10172105e-01 2.18528941e-01 -2.53610492e-01 4.39485252e-01 -8.16099226e-01 7.62033105e-01 9.70561743e-01 -9.28078890e-01 -6.52158856e-01 -3.20478193e-02 9.43853617e-01 5.91236234e-01 1.24811670e-02 -3.75752747e-01 1.90484121e-01 -4.21991199e-01 -3.88076246e-01 -5.44148922e-01 5.01866452e-02 -8.54911685e-01 6.22433841e-01 1.09887576e+00 -1.38492092e-01 2.58510262e-01 -3.18601042e-01 1.10464573e+00 5.72341084e-01 -3.52888912e-01 7.95912683e-01 -1.24128409e-01 5.04006565e-01 7.06262767e-01 -6.63014948e-01 4.77801442e-01 1.50481272e+00 -7.81920552e-03 -5.36961436e-01 -8.22141469e-01 -8.24603677e-01 4.70308095e-01 4.40006465e-01 -3.33645493e-01 3.99529487e-01 -7.05447555e-01 -7.70010889e-01 -3.81033331e-01 -4.67011154e-01 -5.71711063e-02 3.21298808e-01 1.26654625e+00 -8.54211986e-01 1.57415092e-01 2.02129498e-01 -1.39675885e-01 -1.24780905e+00 9.07858670e-01 1.56544045e-01 -7.08909750e-01 -1.83912382e-01 9.60110486e-01 2.70126238e-02 1.52544633e-01 5.92961833e-02 1.17318563e-01 -8.70880410e-02 -8.92442069e-04 5.61741531e-01 8.98694992e-01 -4.30065840e-01 -3.30018133e-01 -3.96652997e-01 4.10745442e-01 -2.99574375e-01 -2.42470667e-01 1.79515779e+00 -4.06387717e-01 -7.37630725e-01 9.58761647e-02 8.76054764e-01 2.51910686e-01 -5.03597558e-01 1.82181243e-02 -6.25507906e-02 -1.99526802e-01 -2.89942592e-01 -2.83788532e-01 -1.38100326e+00 3.33975405e-01 -1.23864956e-01 1.28073204e+00 1.26639688e+00 3.63806158e-01 7.16267943e-01 1.55531287e-01 7.37427652e-01 -7.80711710e-01 -2.82943577e-01 3.18890899e-01 5.95013738e-01 -3.99615884e-01 1.36061013e-01 -1.02911067e+00 -3.41903940e-02 8.14509869e-01 2.58792132e-01 -3.19318563e-01 1.30541039e+00 5.85653245e-01 -6.84918582e-01 -4.74066257e-01 -6.40231967e-01 -3.09613049e-01 -6.52451038e-01 4.95353132e-01 1.10089630e-01 2.87315361e-02 -1.06697118e+00 4.81676072e-01 -6.10855110e-02 6.00745948e-03 9.36844707e-01 1.04340029e+00 -7.62254238e-01 -1.25080705e+00 -2.65580118e-01 8.00528049e-01 -7.38823354e-01 -1.44079134e-01 -5.97178638e-01 8.96632850e-01 -2.49627344e-02 1.14431667e+00 -6.91825002e-02 -1.09423675e-01 -4.01934199e-02 -5.64541757e-01 6.09779954e-01 -4.53910798e-01 -4.50120330e-01 6.50970116e-02 2.32628793e-01 -2.92796254e-01 -5.22567391e-01 -6.32564008e-01 -1.06148219e+00 -1.15082479e+00 -6.75285935e-01 5.55661142e-01 3.54453087e-01 8.68746340e-01 8.79862383e-02 1.93919316e-01 7.80630469e-01 -2.21300140e-01 -1.60103962e-01 -7.05301642e-01 -1.24152827e+00 -2.55893856e-01 1.11551292e-01 -3.67219508e-01 -8.35333228e-01 -4.21290338e-01]
[6.709329605102539, 5.157647132873535]
0301a3c7-41a1-45d3-b39f-41069347986d
task-specific-alignment-and-multiple-level
2307.01985
null
https://arxiv.org/abs/2307.01985v1
https://arxiv.org/pdf/2307.01985v1.pdf
Task-Specific Alignment and Multiple Level Transformer for Few-Shot Action Recognition
In the research field of few-shot learning, the main difference between image-based and video-based is the additional temporal dimension for videos. In recent years, many approaches for few-shot action recognition have followed the metric-based methods, especially, since some works use the Transformer to get the cross-attention feature of the videos or the enhanced prototype, and the results are competitive. However, they do not mine enough information from the Transformer because they only focus on the feature of a single level. In our paper, we have addressed this problem. We propose an end-to-end method named "Task-Specific Alignment and Multiple Level Transformer Network (TSA-MLT)". In our model, the Multiple Level Transformer focuses on the multiple-level feature of the support video and query video. Especially before Multiple Level Transformer, we use task-specific TSA to filter unimportant or misleading frames as a pre-processing. Furthermore, we adopt a fusion loss using two kinds of distance, the first is L2 sequence distance, which focuses on temporal order alignment. The second one is Optimal transport distance, which focuses on measuring the gap between the appearance and semantics of the videos. Using a simple fusion network, we fuse the two distances element-wise, then use the cross-entropy loss as our fusion loss. Extensive experiments show our method achieves state-of-the-art results on the HMDB51 and UCF101 datasets and a competitive result on the benchmark of Kinetics and something-2-something V2 datasets. Our code will be available at the URL: https://github.com/cofly2014/tsa-mlt.git
['YiWang Wang', 'Li Zhu', 'Fei Guo']
2023-07-05
null
null
null
null
['few-shot-action-recognition', 'action-recognition-in-videos', 'few-shot-learning']
['computer-vision', 'computer-vision', 'methodology']
[-4.61013056e-02 -6.06072545e-01 -2.00124428e-01 -3.87405336e-01 -6.85456276e-01 1.96597576e-02 3.30541402e-01 -1.38403729e-01 -7.22914994e-01 3.45090598e-01 3.67319316e-01 3.34895551e-01 -1.14853464e-01 -5.17568350e-01 -6.50486350e-01 -7.89948940e-01 1.16407447e-01 -5.51714674e-02 8.07594895e-01 -2.36654505e-01 1.80779621e-01 -1.01627178e-01 -1.47927654e+00 5.60376108e-01 5.99030554e-01 1.39443302e+00 3.21986496e-01 1.74451113e-01 -2.20224664e-01 9.85174000e-01 -2.49386251e-01 -3.55146378e-01 3.52376461e-01 -6.67042673e-01 -5.66934645e-01 -3.71078923e-02 2.20518097e-01 -4.48503554e-01 -6.27053738e-01 1.23941410e+00 7.06699312e-01 2.86100864e-01 2.47344926e-01 -1.46818328e+00 -2.72691905e-01 4.70913172e-01 -7.67474592e-01 5.48531294e-01 2.98182130e-01 4.75984782e-01 8.55925500e-01 -9.74012733e-01 5.64713120e-01 1.31359875e+00 4.73884523e-01 3.91314745e-01 -6.23255312e-01 -6.09053135e-01 3.04448813e-01 9.84332323e-01 -1.51117218e+00 -3.73746306e-01 8.48881185e-01 -5.36789834e-01 6.67676926e-01 5.11571839e-02 6.77920878e-01 1.04641783e+00 1.53808624e-01 1.08564913e+00 9.94108021e-01 -2.03685299e-01 1.55635485e-02 -5.97234592e-02 9.35723856e-02 6.57156944e-01 -2.45394647e-01 -9.50154290e-02 -5.06058037e-01 3.25624883e-01 4.54942197e-01 4.25373703e-01 -5.41799664e-01 -3.08575273e-01 -1.25332272e+00 6.47530317e-01 2.95456171e-01 6.22782826e-01 -2.52881348e-01 4.09611687e-02 7.09564328e-01 3.46320599e-01 4.57399487e-01 -1.32379442e-01 -3.67797166e-01 -3.48596662e-01 -8.55708957e-01 3.24898958e-02 3.30596924e-01 6.77320659e-01 6.59041941e-01 -3.00014526e-01 -6.04428887e-01 9.22359705e-01 2.37746209e-01 2.68293530e-01 8.59035790e-01 -7.93639243e-01 6.57846332e-01 5.07747591e-01 -1.41759947e-01 -1.15285873e+00 -2.19769120e-01 -1.99175432e-01 -7.20895708e-01 2.65131313e-02 2.78620511e-01 -5.83425490e-03 -7.81772316e-01 1.78952003e+00 3.22373927e-01 5.89901268e-01 -1.59649491e-01 1.21425152e+00 8.42381179e-01 6.57919288e-01 -1.87465817e-01 -4.89881307e-01 1.28582275e+00 -1.30740416e+00 -7.68885195e-01 -1.26376534e-02 6.41296029e-01 -7.15995789e-01 1.22314382e+00 2.92892933e-01 -1.02195990e+00 -7.19873786e-01 -1.07078433e+00 -6.32668808e-02 -3.62555921e-01 3.46137397e-02 1.17883965e-01 9.48368832e-02 -6.52613819e-01 6.95409954e-01 -7.55488157e-01 -3.29126984e-01 3.91737044e-01 -1.97940677e-01 -3.01318645e-01 -2.92417467e-01 -1.43810451e+00 7.56878793e-01 5.24315417e-01 4.43923436e-02 -7.72093892e-01 -5.55910885e-01 -7.27220297e-01 5.08135110e-02 8.06802809e-01 -3.65989566e-01 1.14320397e+00 -1.04530656e+00 -1.29332566e+00 5.74530840e-01 -9.26963519e-03 -4.68731523e-01 8.15929711e-01 -2.46005863e-01 -5.44717491e-01 2.71486521e-01 1.65123299e-01 5.11910021e-01 6.76580787e-01 -6.71330631e-01 -8.28944027e-01 -1.82600334e-01 1.11357190e-01 3.03903222e-01 -5.04127204e-01 1.23918235e-01 -8.58815134e-01 -7.26066768e-01 -1.28002211e-01 -7.14077234e-01 1.14062548e-01 1.51320964e-01 -8.87690187e-02 -3.62127364e-01 1.04037642e+00 -7.16741383e-01 1.41855347e+00 -2.52990842e+00 2.58189738e-01 -2.14099228e-01 6.45609349e-02 4.26368773e-01 -2.79725879e-01 2.41688162e-01 -2.93016527e-02 -2.72415131e-02 -2.29449987e-01 -2.31229812e-01 -7.41778389e-02 -3.37308794e-02 8.20399672e-02 4.25461978e-01 2.52508130e-02 6.94758415e-01 -9.10841405e-01 -8.57316911e-01 3.49077672e-01 3.85618329e-01 -3.67739826e-01 2.08658680e-01 -1.51073739e-01 2.77289391e-01 -3.96768063e-01 4.91538674e-01 5.81030250e-01 -8.36418793e-02 -2.58937746e-01 -5.67427099e-01 -1.91178307e-01 -8.41763616e-02 -1.15997326e+00 2.11116982e+00 -2.05690846e-01 4.01854157e-01 -2.92444825e-01 -1.05066419e+00 6.83399022e-01 2.41734281e-01 8.19906712e-01 -9.34183538e-01 2.85938889e-01 1.84257478e-01 1.16685601e-02 -7.94123232e-01 -4.98222336e-02 -2.15256959e-02 1.28316790e-01 5.72571643e-02 8.53377879e-02 4.18947905e-01 5.30243099e-01 1.25398785e-01 1.05442429e+00 2.95223147e-01 1.95371345e-01 1.02261700e-01 7.88612545e-01 -2.86716998e-01 9.57814038e-01 1.76227927e-01 -4.61132288e-01 8.20642292e-01 6.00051284e-01 -3.81412625e-01 -8.50805402e-01 -7.42613554e-01 1.84526265e-01 8.23222339e-01 5.76275289e-01 -6.30001068e-01 -6.35766923e-01 -8.01803648e-01 -1.89207062e-01 5.73191285e-01 -5.74197829e-01 -4.47767764e-01 -5.89403510e-01 -4.34405833e-01 3.23589772e-01 4.76585090e-01 1.03143549e+00 -1.07258618e+00 -6.84069335e-01 2.52420366e-01 -5.46120882e-01 -1.25930727e+00 -8.99434209e-01 -3.19648206e-01 -5.89010119e-01 -1.12239087e+00 -9.42795217e-01 -6.13390982e-01 1.83809102e-01 5.38277566e-01 6.07014954e-01 9.87714082e-02 -3.08388352e-01 1.67064384e-01 -7.04105020e-01 -5.20708412e-02 2.44392082e-01 -2.90164590e-01 -4.86352928e-02 4.72661525e-01 4.50115114e-01 -5.51274478e-01 -7.45866358e-01 5.95383406e-01 -8.97644043e-01 8.21849704e-02 6.57353759e-01 7.05851078e-01 7.66050100e-01 1.82042271e-01 2.10070610e-01 -2.71046996e-01 3.89028877e-01 -4.68318284e-01 -2.21130282e-01 4.33736503e-01 -4.35506701e-01 -7.49058798e-02 6.76096201e-01 -5.39360523e-01 -9.72139716e-01 -1.00264572e-01 -1.59121588e-01 -1.13402522e+00 9.76564735e-02 5.25955081e-01 -4.63158697e-01 1.24253079e-01 1.72683284e-01 4.73227143e-01 3.68015058e-02 -5.39937139e-01 1.04314171e-01 5.89679241e-01 3.28304112e-01 -2.16554120e-01 4.60766733e-01 4.21900868e-01 -2.01505244e-01 -6.12839341e-01 -8.60727549e-01 -6.18729651e-01 -3.76211047e-01 -3.70699137e-01 1.18612504e+00 -7.62043417e-01 -5.43101668e-01 8.24366748e-01 -1.03519261e+00 -1.55778587e-01 -3.15534949e-01 8.74560773e-01 -6.61464036e-01 6.43236458e-01 -5.27203619e-01 -5.91556072e-01 -2.44975671e-01 -1.35830522e+00 7.89789379e-01 3.06636006e-01 3.51863146e-01 -4.95865911e-01 2.01871499e-01 2.95645028e-01 2.48431727e-01 2.37657249e-01 5.16494453e-01 -5.85258543e-01 -4.34678167e-01 3.18540707e-02 -3.16749245e-01 6.34189844e-01 4.40921970e-02 -9.16273966e-02 -6.77010357e-01 -3.57459456e-01 2.21037254e-01 -3.19064915e-01 1.21167910e+00 3.41439247e-01 1.31093001e+00 -2.07449391e-01 -1.62464976e-01 6.68584943e-01 1.45782030e+00 4.31574494e-01 8.85116994e-01 2.65740484e-01 7.61606634e-01 3.48805189e-01 1.06212473e+00 4.63049084e-01 2.87091911e-01 1.00330174e+00 3.28942567e-01 1.72478557e-01 -2.04266950e-01 -2.09918275e-01 6.30124032e-01 9.36868966e-01 -2.46550307e-01 -1.52937651e-01 -6.53032780e-01 3.56417984e-01 -2.29296684e+00 -1.39461398e+00 1.21253565e-01 2.23659658e+00 6.67319119e-01 3.33411664e-01 3.22386265e-01 -3.24764885e-02 9.69314158e-01 5.24416208e-01 -6.66318178e-01 9.44114402e-02 -1.96074679e-01 -2.58072257e-01 3.19205791e-01 2.17360258e-01 -1.21994889e+00 6.77498400e-01 3.96913671e+00 1.38696611e+00 -1.04885566e+00 4.45036858e-01 4.80825484e-01 -4.28346664e-01 1.59765065e-01 -6.46635816e-02 -6.41469359e-01 1.02598107e+00 6.06006145e-01 -2.15916067e-01 2.80856937e-01 6.76915884e-01 3.48917454e-01 2.26917304e-03 -9.59180892e-01 1.25779474e+00 2.44500399e-01 -1.07959914e+00 -7.74779962e-03 -1.24303043e-01 2.83706397e-01 -1.11080660e-02 -9.81917605e-02 4.62186962e-01 -4.15580690e-01 -5.88512599e-01 8.67082596e-01 7.85991788e-01 6.65349543e-01 -6.50455654e-01 9.09193099e-01 2.76549667e-01 -1.55536497e+00 -2.21891850e-01 -4.14008707e-01 1.76929325e-01 3.31104726e-01 4.70647454e-01 7.69289061e-02 7.40600169e-01 1.01869297e+00 1.16896951e+00 -5.89586735e-01 1.37470055e+00 -2.23349463e-02 3.26566219e-01 -1.80567488e-01 4.61816899e-02 4.05035824e-01 -2.21337110e-01 6.39251173e-01 1.07457221e+00 4.34020728e-01 1.97856918e-01 4.42614824e-01 3.68244410e-01 2.09815856e-02 2.45948046e-01 -3.60664427e-01 1.22873880e-01 2.18513742e-01 1.15336823e+00 -5.68466842e-01 -4.95603234e-01 -7.03959286e-01 1.06542826e+00 1.39771104e-01 1.56845655e-02 -1.19810355e+00 -6.30446017e-01 4.54196692e-01 6.04104474e-02 5.52597344e-01 9.02226269e-02 3.97297174e-01 -1.44470847e+00 4.02892798e-01 -9.13879752e-01 6.67499721e-01 -7.33419359e-01 -1.24602211e+00 4.05995071e-01 2.25778043e-01 -1.64321852e+00 1.55462667e-01 -4.16504711e-01 -7.07075298e-01 4.03130352e-01 -1.39068675e+00 -9.81751382e-01 -4.61941004e-01 9.10012066e-01 9.68967915e-01 -1.77100420e-01 1.93188146e-01 7.86938071e-01 -6.94061339e-01 5.84166765e-01 6.43334016e-02 2.76748568e-01 9.59283710e-01 -7.84259498e-01 1.27465499e-03 8.42551410e-01 5.24844602e-02 1.14946783e-01 5.12839913e-01 -4.92869794e-01 -1.10140324e+00 -9.57464397e-01 5.34366310e-01 4.85775508e-02 5.93182504e-01 -5.61617948e-02 -9.94551063e-01 4.03142273e-01 1.92435637e-01 2.90884197e-01 3.23839307e-01 -4.01726335e-01 -3.50496262e-01 -4.29577768e-01 -9.95053649e-01 3.99126709e-01 1.27712643e+00 -3.01447690e-01 -5.87835550e-01 3.10322493e-01 9.49193597e-01 -1.42459735e-01 -7.14850903e-01 6.14865720e-01 6.40774846e-01 -1.23819387e+00 8.60950649e-01 -4.52704996e-01 4.86873865e-01 -6.08479559e-01 -3.07397217e-01 -1.12694097e+00 -4.05164152e-01 -2.66112775e-01 -7.24573582e-02 1.24743521e+00 5.58055528e-02 -4.25529718e-01 4.20659751e-01 4.20353338e-02 -1.94729269e-01 -9.70566392e-01 -1.02375758e+00 -1.09795952e+00 -2.16424420e-01 -2.62204438e-01 4.65334803e-01 8.57623577e-01 -7.62359947e-02 2.98327804e-01 -6.96682036e-01 -2.68663138e-01 5.18833756e-01 8.07166547e-02 4.75855768e-01 -9.19207871e-01 -4.66943920e-01 -6.25389278e-01 -7.04277754e-01 -9.86328304e-01 -6.94367811e-02 -6.65418327e-01 4.90692183e-02 -1.32885945e+00 5.21352172e-01 8.24871212e-02 -7.85901010e-01 4.81584877e-01 -1.80270538e-01 2.22061593e-02 4.97945487e-01 3.42940748e-01 -1.04997635e+00 8.75328362e-01 1.23909760e+00 -3.29245925e-01 -3.54024470e-02 -2.19153255e-01 -1.72024250e-01 7.80648053e-01 9.79599595e-01 -4.41889793e-01 -5.05134761e-01 -4.32024211e-01 -2.30865121e-01 1.52224605e-03 4.33452874e-01 -1.37175405e+00 3.84598553e-01 -2.64028311e-01 2.13642284e-01 -6.95825815e-01 5.18553793e-01 -8.19780529e-01 -3.75111550e-02 6.69125080e-01 -2.79157698e-01 2.33928189e-02 -1.75379634e-01 7.17400789e-01 -4.28920627e-01 -1.51775211e-01 9.47742999e-01 -2.05715463e-01 -1.16338599e+00 6.82361066e-01 1.66662216e-01 3.17215562e-01 1.32426167e+00 -1.18688978e-01 -4.47437108e-01 -2.42590517e-01 -5.81493855e-01 4.41267669e-01 3.95546436e-01 5.74072123e-01 8.27157140e-01 -1.57952392e+00 -7.48795211e-01 -1.03867732e-01 2.88616061e-01 -4.08300638e-01 5.70745647e-01 1.36705148e+00 -2.00273424e-01 1.49159700e-01 -5.18989742e-01 -5.33318877e-01 -1.39086807e+00 8.48161519e-01 3.34941655e-01 -2.39998698e-01 -6.78930104e-01 7.14319885e-01 3.25782657e-01 1.48430005e-01 3.16837728e-01 -6.29167631e-02 -3.19521934e-01 2.26294950e-01 7.76175499e-01 4.42965835e-01 -2.00513646e-01 -7.36040831e-01 -6.36918843e-01 8.07978213e-01 -1.23048007e-01 -1.41611114e-01 1.16933894e+00 -8.09352249e-02 7.86169916e-02 6.68158352e-01 1.50878239e+00 -3.87355149e-01 -1.29629028e+00 -4.62644786e-01 -2.17202336e-01 -7.33149290e-01 1.41065186e-02 -5.18757284e-01 -1.42543030e+00 1.04261816e+00 8.87769699e-01 -4.08049002e-02 1.26335430e+00 -8.81913826e-02 1.09113896e+00 2.16234162e-01 2.86315382e-01 -1.26533854e+00 3.05852085e-01 3.57502222e-01 8.39291096e-01 -1.24781907e+00 -1.14919849e-01 -1.23333640e-01 -7.30565965e-01 9.06195164e-01 8.77960861e-01 2.32833438e-03 6.73954487e-01 -1.16999470e-01 -2.52778143e-01 -9.22978222e-02 -6.87913477e-01 -4.49406475e-01 2.53872693e-01 1.74989223e-01 1.69430971e-01 -2.25433737e-01 -8.53101611e-01 5.89040160e-01 3.24659973e-01 3.27102423e-01 1.96130529e-01 9.70648646e-01 -5.90308428e-01 -9.80959475e-01 3.60824876e-02 3.62504929e-01 -4.07997340e-01 7.75668956e-03 -6.39903992e-02 6.40343606e-01 4.16423798e-01 9.01990592e-01 -4.67966609e-02 -7.98224151e-01 5.01658320e-01 2.17958037e-02 4.42516685e-01 -2.56454527e-01 -3.05195421e-01 1.14688233e-01 -9.31299180e-02 -9.08525109e-01 -6.09235823e-01 -6.16526067e-01 -1.18558073e+00 -3.03750932e-01 -2.92737454e-01 2.35219911e-01 2.52782196e-01 8.75703990e-01 3.46934974e-01 3.58549535e-01 8.10699344e-01 -5.79466760e-01 -5.19870758e-01 -9.37304199e-01 -5.64564824e-01 6.14835322e-01 9.04632434e-02 -8.79804671e-01 -4.22991544e-01 -1.08257867e-01]
[8.585698127746582, 0.7590094208717346]
730ae128-a07d-4e03-8b61-fedb329f4f08
fast-and-large-scale-unsupervised-relation
null
null
https://aclanthology.org/Y15-1012
https://aclanthology.org/Y15-1012.pdf
Fast and Large-scale Unsupervised Relation Extraction
null
['Kentaro Inui', 'Sho Takase', 'Naoaki Okazaki']
2015-10-01
fast-and-large-scale-unsupervised-relation-1
https://aclanthology.org/Y15-1012
https://aclanthology.org/Y15-1012.pdf
paclic-2015-10
['open-information-extraction']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.3260817527771, 3.858095407485962]
270ab9b1-6ed5-4a5d-8619-6d04a481e469
impara-impact-based-metric-for-gec-using
null
null
https://aclanthology.org/2022.coling-1.316
https://aclanthology.org/2022.coling-1.316.pdf
IMPARA: Impact-Based Metric for GEC Using Parallel Data
Automatic evaluation of grammatical error correction (GEC) is essential in developing useful GEC systems. Existing methods for automatic evaluation require multiple reference sentences or manual scores. However, such resources are expensive, thereby hindering automatic evaluation for various domains and correction styles. This paper proposes an Impact-based Metric for GEC using PARAllel data, IMPARA, which utilizes correction impacts computed by parallel data comprising pairs of grammatical/ungrammatical sentences. As parallel data is cheaper than manually assessing evaluation scores, IMPARA can reduce the cost of data creation for automatic evaluation. Correlations between IMPARA and human scores indicate that IMPARA is comparable or better than existing evaluation methods. Furthermore, we find that IMPARA can perform evaluations that fit different domains and correction styles trained on various parallel data.
['Naoaki Okazaki', 'Masahiro Kaneko', 'Koki Maeda']
null
null
null
null
coling-2022-10
['grammatical-error-correction']
['natural-language-processing']
[-1.45016998e-01 -8.35871026e-02 4.21961963e-01 -7.16518760e-01 -1.07954311e+00 -7.38080442e-01 1.78345799e-01 8.81907165e-01 -7.81060576e-01 7.56904721e-01 2.95957088e-01 -2.46749640e-01 1.62639152e-02 -7.57595062e-01 -3.73663813e-01 5.48118092e-02 4.71228331e-01 5.36669970e-01 3.25914919e-02 -6.57922804e-01 6.77361429e-01 -1.15272559e-01 -1.40641344e+00 2.67973691e-01 1.56966019e+00 2.20772281e-01 3.87075692e-01 8.88267338e-01 -1.08098105e-01 5.25041342e-01 -1.23354530e+00 -1.18716764e+00 -2.01636225e-01 -5.58572710e-01 -8.87244403e-01 -8.20543408e-01 4.33116525e-01 -9.91978124e-03 6.92133188e-01 1.13103354e+00 7.64546931e-01 5.58022000e-02 7.05065906e-01 -8.55297089e-01 -9.06207383e-01 8.96899879e-01 8.46821666e-02 1.58319995e-01 6.71239972e-01 1.30899316e-02 9.46162879e-01 -1.02419317e+00 6.96150839e-01 1.20233011e+00 1.02912927e+00 7.38700032e-01 -1.00770807e+00 -3.65392536e-01 -3.22046578e-01 8.35311264e-02 -1.05862665e+00 -1.05101191e-01 5.03113568e-01 -3.56437385e-01 1.43626404e+00 2.90149093e-01 4.76421267e-01 7.49101698e-01 9.26359892e-02 3.90083224e-01 1.20082116e+00 -1.04639280e+00 -1.17645659e-01 1.92171007e-01 4.09346223e-01 5.03183722e-01 4.90809560e-01 -7.97798112e-02 -3.42552513e-01 6.54503480e-02 1.35832176e-01 -7.14676201e-01 -1.12446807e-01 5.89594007e-01 -9.50387597e-01 7.55388379e-01 -2.91944612e-02 2.93777406e-01 -1.61568135e-01 -4.86336090e-03 7.72239804e-01 6.78372979e-01 5.26154935e-01 9.62655902e-01 -6.28018260e-01 -6.36427283e-01 -7.54306614e-01 5.38219690e-01 6.96305096e-01 1.02478456e+00 4.66376096e-01 2.89971400e-02 -3.75452727e-01 1.16165674e+00 4.19656426e-01 8.59852493e-01 5.29746413e-01 -8.64251971e-01 8.85490537e-01 1.01874220e+00 1.68302417e-01 -1.13942850e+00 -3.95789355e-01 -3.58378142e-01 -2.85357386e-01 8.85330215e-02 4.18522567e-01 7.16835037e-02 -4.08218145e-01 1.80506778e+00 3.80913951e-02 -9.00771916e-01 -2.10087020e-02 6.77534699e-01 9.15663719e-01 3.90560210e-01 4.99383450e-01 -7.61905909e-02 8.77461553e-01 -6.38480663e-01 -7.78326392e-01 6.28666282e-02 1.37271821e+00 -1.06191778e+00 1.80236030e+00 2.34514385e-01 -1.25444674e+00 -5.55006087e-01 -1.01731241e+00 -4.40871209e-01 -3.72791618e-01 2.29120746e-01 3.62174094e-01 9.06334519e-01 -1.06277013e+00 8.52279544e-01 -5.55356681e-01 -2.58724749e-01 -2.96540558e-01 8.36744308e-02 -2.97240198e-01 -8.92331898e-02 -1.16244006e+00 1.50490367e+00 3.92181963e-01 -1.83658525e-01 -4.01425689e-01 -4.15301591e-01 -9.98203874e-01 -1.18951246e-01 -2.18472257e-01 -2.70747006e-01 1.39396334e+00 -7.57518589e-01 -1.01420045e+00 9.04025257e-01 -1.17759019e-01 -1.12422720e-01 4.69843984e-01 -2.61160105e-01 -6.76711380e-01 -4.07288879e-01 3.68747383e-01 4.27807242e-01 9.64017734e-02 -7.64710844e-01 -5.76128483e-01 -3.42166305e-01 5.30581400e-02 4.15124267e-01 -3.03275496e-01 7.71943510e-01 -2.12256938e-01 -7.78997540e-01 -2.38286629e-01 -5.92903018e-01 2.11648226e-01 -6.49296403e-01 7.93726966e-02 -6.20060623e-01 2.06483945e-01 -1.20740867e+00 1.90872896e+00 -1.74705100e+00 1.27272636e-01 2.23130152e-01 4.61383946e-02 6.66653156e-01 -4.29281861e-01 2.98476160e-01 6.08732328e-02 6.28041506e-01 -1.95946157e-01 -4.17573929e-01 1.51575655e-01 1.58815220e-01 4.87920016e-01 -1.99204266e-01 1.72314346e-01 8.84942949e-01 -1.12057900e+00 -6.90238357e-01 -1.29362896e-01 1.42685577e-01 -7.34189630e-01 7.51565397e-02 -4.50387262e-02 1.70337960e-01 1.79545850e-01 7.46381044e-01 6.38634324e-01 2.80528665e-01 2.16646686e-01 2.68055022e-01 -1.33184850e-01 8.36236000e-01 -8.49703074e-01 1.27710474e+00 -6.38780951e-01 5.46037018e-01 -5.55957615e-01 -4.79960501e-01 1.25274301e+00 1.46117598e-01 -4.90060717e-01 -9.11431432e-01 2.89638460e-01 6.98115826e-01 5.66553809e-02 -3.31992388e-01 1.01045120e+00 1.31587416e-01 -4.74968731e-01 5.82179904e-01 2.89062172e-01 -4.84155655e-01 9.34519947e-01 1.93946198e-01 1.07487428e+00 1.88842528e-02 3.55112225e-01 -2.82615900e-01 3.85017484e-01 1.61909476e-01 9.69991803e-01 5.68094492e-01 -3.18336099e-01 4.86945808e-01 2.49809757e-01 -2.19854593e-01 -1.29939008e+00 -8.86415243e-01 3.85190472e-02 1.01525414e+00 -2.35070884e-01 -1.07839143e+00 -1.24078763e+00 -9.37393188e-01 -3.22709560e-01 1.20372987e+00 -2.54332423e-01 -1.38010249e-01 -6.80820167e-01 -6.44159317e-01 8.98220658e-01 7.32509971e-01 3.23695809e-01 -1.15908325e+00 -3.39635521e-01 3.72316658e-01 -7.27018893e-01 -5.30896723e-01 -5.70579350e-01 -3.49158406e-01 -9.29642022e-01 -1.15762734e+00 -1.43431351e-01 -7.75176585e-01 7.29701638e-01 -6.07733056e-02 1.76326835e+00 7.56122172e-01 2.67896891e-01 1.86919525e-01 -7.86574960e-01 -7.71692693e-01 -1.07790256e+00 -9.60666046e-04 5.06572276e-02 -1.07147551e+00 7.83385277e-01 -1.07288666e-01 -4.77924049e-02 3.51909339e-01 -5.55604517e-01 1.23831630e-01 2.07539603e-01 8.45868945e-01 4.09788996e-01 -3.39250654e-01 6.44635797e-01 -1.15871274e+00 1.16516423e+00 -1.32991925e-01 -4.61252272e-01 7.16447294e-01 -1.17303383e+00 5.21111377e-02 4.07077938e-01 4.76452857e-02 -1.16301537e+00 -6.75019503e-01 -3.36629570e-01 4.18463379e-01 2.01337919e-01 8.69672418e-01 -6.15915507e-02 -7.06115365e-02 1.07574272e+00 -3.82420599e-01 -2.97963709e-01 -4.76577997e-01 -1.48464620e-01 7.78703272e-01 4.94698644e-01 -8.90774846e-01 3.45202386e-01 -8.15509260e-01 -4.29447502e-01 -2.15349644e-01 -6.68706179e-01 -5.53315729e-02 -6.90713227e-01 -4.92425382e-01 4.38109875e-01 -7.39808500e-01 -1.99611366e-01 5.00184476e-01 -1.51173735e+00 -1.18127331e-01 -1.26299914e-02 5.83582342e-01 -2.10702583e-01 3.41578007e-01 -7.03563988e-01 -5.35712242e-01 -7.46807635e-01 -1.03000093e+00 8.34576130e-01 1.58865303e-01 -8.39741409e-01 -1.15891039e+00 4.37137455e-01 5.46695113e-01 4.92360771e-01 2.46925279e-01 1.11946905e+00 -4.22112674e-01 -1.06799705e-02 -4.99282062e-01 -8.54910612e-02 8.63727868e-01 -1.94894537e-01 3.33951205e-01 -4.37652111e-01 -4.09262255e-02 -3.86852771e-01 -3.41239035e-01 2.92675465e-01 -1.92980487e-02 8.13081026e-01 -3.25256556e-01 3.10447305e-01 1.97622448e-01 1.34018195e+00 5.50595298e-02 8.03301513e-01 3.87890637e-01 7.09254980e-01 5.03333092e-01 1.04783988e+00 1.10747732e-01 8.67687702e-01 5.61385989e-01 1.13331955e-02 4.31532651e-01 -3.10438007e-01 -4.75359797e-01 6.63299799e-01 1.65177405e+00 -5.23580074e-01 -3.80839974e-01 -1.25143588e+00 7.84929991e-01 -1.49106967e+00 -7.10641563e-01 -7.39884019e-01 2.27184749e+00 1.17382002e+00 1.68834068e-02 -1.91293240e-01 4.07651573e-01 8.73796403e-01 -5.41813970e-01 2.92136706e-02 -1.20782661e+00 -4.95050222e-01 5.31333804e-01 1.83005571e-01 5.48304319e-01 -6.50848091e-01 9.46374774e-01 6.97476578e+00 6.16691470e-01 -6.16103172e-01 4.88087237e-01 2.73625493e-01 1.82401687e-02 -7.98120618e-01 8.09958726e-02 -9.70033705e-01 6.53030634e-01 1.40393066e+00 -4.20409173e-01 2.63658524e-01 7.20385611e-01 2.12775603e-01 -1.44765273e-01 -9.61336553e-01 7.87347436e-01 2.82293677e-01 -9.23191726e-01 -3.87310376e-03 -4.12102640e-01 8.85236621e-01 -3.40422504e-02 -3.28488320e-01 6.80454910e-01 6.81403995e-01 -6.13591015e-01 9.62637484e-01 4.09539163e-01 9.11570609e-01 -7.26638317e-01 1.17190850e+00 2.21526995e-01 -6.60265326e-01 2.97084063e-01 -5.87447703e-01 -4.20585632e-01 5.95714487e-02 5.16492665e-01 -7.69764245e-01 3.52278739e-01 7.47294426e-01 4.19389963e-01 -1.24827516e+00 8.02110434e-01 -8.32269430e-01 7.69125402e-01 2.56996769e-02 -3.59910309e-01 -3.10918987e-01 -1.07178785e-01 4.05040413e-01 1.44882047e+00 7.24478424e-01 -2.26871729e-01 -3.10570240e-01 6.61782801e-01 -1.61004230e-01 6.08074307e-01 -5.56402087e-01 -7.56068975e-02 8.31460416e-01 9.17034268e-01 -2.07070395e-01 -3.31222415e-01 -2.42192924e-01 8.79934251e-01 7.59556293e-01 -1.87617153e-01 -7.90102661e-01 -6.59651875e-01 2.13854179e-01 -1.59624964e-02 -5.42045891e-01 -1.81872666e-01 -8.49005103e-01 -1.11507928e+00 2.54291564e-01 -1.06248665e+00 4.48176563e-01 -8.58607292e-01 -1.03186584e+00 5.82919002e-01 -7.49560520e-02 -1.22284293e+00 -3.70478958e-01 -6.38095260e-01 -6.71618938e-01 1.14541984e+00 -1.06842995e+00 -1.01461220e+00 -4.19633657e-01 1.73897400e-01 3.78725737e-01 -3.61191481e-02 1.09231389e+00 5.94090462e-01 -5.88668823e-01 1.17410493e+00 2.52297223e-02 8.57285783e-02 1.07656789e+00 -1.59173405e+00 5.89718938e-01 1.31912398e+00 -1.50167316e-01 6.94953859e-01 4.90676731e-01 -9.30768132e-01 -4.12507117e-01 -1.10994911e+00 1.94208348e+00 -8.44429970e-01 1.53369606e-01 1.37267234e-02 -9.60852921e-01 4.02133167e-01 1.21328816e-01 -6.30658031e-01 8.59511137e-01 4.04665440e-01 -2.80061334e-01 7.73581415e-02 -1.32968712e+00 4.12052989e-01 1.09281445e+00 -6.00229800e-01 -7.49495506e-01 8.80395249e-02 5.26769519e-01 -5.07872283e-01 -1.01268029e+00 3.27847511e-01 1.90239444e-01 -6.32185340e-01 3.41489434e-01 -5.21686018e-01 8.37293684e-01 -4.09898013e-01 -2.11898953e-01 -1.83897209e+00 -3.90044034e-01 -1.12544209e-01 3.98306787e-01 1.73706973e+00 9.97490525e-01 -4.57074285e-01 1.78312638e-03 1.04543483e+00 -6.41066253e-01 -7.64670223e-02 -3.83075476e-01 -8.54639292e-01 2.92136133e-01 -7.60586381e-01 8.56337607e-01 1.08203351e+00 2.55687505e-01 2.11295813e-01 1.00481361e-02 -3.46784890e-02 4.17428643e-01 -5.10576725e-01 5.77119887e-01 -1.29895377e+00 -4.24418077e-02 -3.08195174e-01 -3.17602217e-01 1.43760413e-01 4.22307700e-02 -1.13125145e+00 5.04612550e-02 -1.43002403e+00 1.59156904e-01 -6.19417489e-01 -6.92524314e-02 6.04918599e-01 -6.28887773e-01 2.88936168e-01 1.79956704e-01 1.30752847e-01 -4.59326774e-01 3.61827374e-01 9.62228060e-01 3.26809986e-03 -7.85788745e-02 -5.88212013e-01 -6.76629782e-01 7.41258502e-01 1.04703403e+00 -7.03809321e-01 -6.81643188e-02 -1.10941470e+00 8.29201937e-01 -3.47368866e-01 9.36678424e-03 -1.04153502e+00 -3.06103025e-02 -8.89790580e-02 1.83845013e-01 -2.96847522e-01 -4.61750329e-01 5.00584356e-02 1.68414548e-01 4.01533656e-02 -2.45296821e-01 9.44919288e-01 1.72697186e-01 -1.59934387e-01 -5.12385190e-01 -9.09862220e-01 7.02623546e-01 -2.66848475e-01 -6.61570013e-01 -3.91929537e-01 -1.91022500e-01 2.76182920e-01 5.29366612e-01 -1.83394730e-01 -5.52142203e-01 3.23164836e-02 -4.16249901e-01 2.05650717e-01 7.45634198e-01 3.36327255e-01 5.12188733e-01 -1.44418645e+00 -9.91075397e-01 -5.17304204e-02 5.87594748e-01 -1.86194807e-01 -1.33445142e-02 5.77839196e-01 -1.07551265e+00 1.33591309e-01 -3.81384879e-01 -2.47376680e-01 -1.73156309e+00 -1.05370924e-01 6.57615811e-02 -4.76114810e-01 4.23592925e-02 7.76876867e-01 -5.51545680e-01 -1.01857078e+00 -1.67178944e-01 -2.09573850e-01 -4.50391620e-01 -1.09364733e-01 5.67170918e-01 8.08474302e-01 5.73554397e-01 -6.16224885e-01 -7.82038271e-02 2.84545213e-01 -1.72985047e-01 -2.05050200e-01 1.13182986e+00 -1.75168619e-01 -4.63862896e-01 2.84088463e-01 7.45337486e-01 1.69718027e-01 -3.67766738e-01 -4.08199318e-02 3.47118706e-01 -6.11057520e-01 3.00539434e-02 -1.40775669e+00 -4.46678221e-01 8.10423434e-01 4.20009911e-01 -3.76591623e-01 1.17701936e+00 -5.01332819e-01 4.65484679e-01 4.67304140e-01 6.24014616e-01 -1.77489591e+00 -2.57516176e-01 9.27905738e-01 1.09792078e+00 -1.41610754e+00 -4.78092968e-01 -3.31068367e-01 -7.55338728e-01 9.23252285e-01 9.69757736e-01 2.58429319e-01 1.41426116e-01 -1.14900991e-01 2.22646371e-01 -1.82386771e-01 -5.45591295e-01 1.76428288e-01 2.99431354e-01 7.21812248e-01 1.11887681e+00 5.25573611e-01 -1.49526250e+00 8.85036588e-01 -6.45502448e-01 -1.48443535e-01 8.57449651e-01 7.55227685e-01 -1.61699161e-01 -1.63679743e+00 -2.46994928e-01 4.63282943e-01 -4.31507796e-01 -2.73109138e-01 -6.00798965e-01 6.63293421e-01 2.30897799e-01 1.15497279e+00 -1.65706426e-01 -4.28610593e-01 5.58507562e-01 1.64190829e-01 6.68161273e-01 -8.29435408e-01 -1.04617441e+00 -5.27451575e-01 6.71641409e-01 -3.86586130e-01 -3.23099613e-01 -8.47515762e-01 -1.16975594e+00 -7.00371802e-01 -6.49291158e-01 3.94058377e-01 8.17284524e-01 8.22774768e-01 3.50118786e-01 2.76388258e-01 2.89289832e-01 -2.46305928e-01 -4.82149720e-01 -1.60441995e+00 -2.81810075e-01 8.09737980e-01 -4.54559416e-01 -5.65462053e-01 -1.38562083e-01 2.75532585e-02]
[11.07904052734375, 10.680402755737305]
a043866f-f34d-4f63-8433-a251fe257fab
exploiting-pseudo-future-contexts-for-emotion
2306.15376
null
https://arxiv.org/abs/2306.15376v1
https://arxiv.org/pdf/2306.15376v1.pdf
Exploiting Pseudo Future Contexts for Emotion Recognition in Conversations
With the extensive accumulation of conversational data on the Internet, emotion recognition in conversations (ERC) has received increasing attention. Previous efforts of this task mainly focus on leveraging contextual and speaker-specific features, or integrating heterogeneous external commonsense knowledge. Among them, some heavily rely on future contexts, which, however, are not always available in real-life scenarios. This fact inspires us to generate pseudo future contexts to improve ERC. Specifically, for an utterance, we generate its future context with pre-trained language models, potentially containing extra beneficial knowledge in a conversational form homogeneous with the historical ones. These characteristics make pseudo future contexts easily fused with historical contexts and historical speaker-specific contexts, yielding a conceptually simple framework systematically integrating multi-contexts. Experimental results on four ERC datasets demonstrate our method's superiority. Further in-depth analyses reveal that pseudo future contexts can rival real ones to some extent, especially in relatively context-independent conversations.
['Guanglu Wan', 'Tong Mo', 'Wei Ye', 'Hailei Yan', 'Shuaipeng Liu', 'Yinyi Wei']
2023-06-27
null
null
null
null
['emotion-recognition']
['computer-vision']
[ 2.57098436e-01 -1.05171613e-01 -1.02217562e-01 -6.29530907e-01 -5.80410957e-01 -5.20850718e-01 8.72847736e-01 7.42293596e-02 -2.30404615e-01 8.64721179e-01 8.40950012e-01 -1.26083910e-01 1.95333421e-01 -5.68672061e-01 -1.74265712e-01 -5.39920926e-01 1.27349928e-01 -1.51301801e-01 -2.45232120e-01 -7.24152684e-01 1.57320291e-01 8.79635587e-02 -1.61212003e+00 3.43986034e-01 1.11551440e+00 8.48677516e-01 3.84505719e-01 2.99458951e-01 -5.71982324e-01 6.41110659e-01 -6.67603016e-01 -8.39914560e-01 -2.50445783e-01 -6.43770576e-01 -8.53591442e-01 2.17627972e-01 -4.41332579e-01 7.34859183e-02 -1.52949229e-01 7.56711185e-01 4.10717249e-01 3.17038864e-01 2.25062892e-01 -1.09531784e+00 -7.17266619e-01 9.83206987e-01 -1.02166913e-01 1.34276256e-01 8.21477413e-01 2.49531612e-01 1.07651043e+00 -6.93137050e-01 5.78708947e-01 1.25707424e+00 6.28170669e-01 8.19689274e-01 -9.61108327e-01 -5.79326868e-01 6.80006623e-01 3.32243681e-01 -1.14802420e+00 -5.78859091e-01 1.42131448e+00 -4.18702513e-02 6.67072415e-01 4.74395961e-01 7.55140066e-01 1.95276809e+00 -1.96759298e-01 9.33659732e-01 1.22915387e+00 -4.78703618e-01 2.31199563e-01 4.59500194e-01 -6.16652556e-02 -1.53349191e-01 -5.91357052e-01 -1.64264739e-01 -7.07140207e-01 -1.48007125e-01 2.01672420e-01 9.89445746e-02 -6.44321740e-01 3.35385025e-01 -1.24286389e+00 6.33328140e-01 4.32722792e-02 7.03713000e-01 -5.39389312e-01 -3.44013363e-01 5.91735065e-01 2.53846407e-01 6.64575338e-01 3.01224738e-01 -5.03547907e-01 -7.49372482e-01 -4.34817672e-01 7.45210722e-02 8.37030232e-01 1.00267768e+00 7.67137170e-01 -2.52977964e-02 1.05411150e-02 1.11901772e+00 -2.46533789e-02 3.25704306e-01 7.96355665e-01 -8.59624028e-01 5.26859164e-01 6.14056170e-01 1.78424180e-01 -1.23305631e+00 -2.38190949e-01 -5.16978145e-01 -6.85313880e-01 -8.98284614e-01 1.56838641e-01 -3.39690834e-01 -7.21303150e-02 2.24136925e+00 3.80432338e-01 5.41006088e-01 3.84956598e-01 7.18266606e-01 5.38195610e-01 5.57746172e-01 1.42819807e-01 -7.19701409e-01 1.21031845e+00 -6.34128511e-01 -1.00299227e+00 -3.94835234e-01 5.25767982e-01 -7.75175929e-01 1.33942473e+00 2.39696503e-01 -5.47101498e-01 -2.83848345e-01 -8.05487394e-01 2.12673694e-01 -3.88116062e-01 -6.97785467e-02 8.63865852e-01 5.74563205e-01 -7.04310834e-01 3.37921619e-01 -2.89796650e-01 -2.83499241e-01 -1.90613031e-01 -2.34503761e-01 -1.86393082e-01 1.78492069e-01 -1.79206109e+00 7.91196525e-01 2.97952890e-01 5.05455375e-01 -3.82176459e-01 -3.39759350e-01 -8.80940914e-01 -8.70214477e-02 7.60212362e-01 -3.25023264e-01 1.32465446e+00 -1.18996322e+00 -1.93655705e+00 5.55511236e-01 -5.56440353e-01 -3.92117083e-01 4.73845184e-01 -1.66633934e-01 -1.03783333e+00 1.78676806e-02 -7.03853890e-02 1.54021159e-01 7.21460581e-01 -1.45885885e+00 -5.51076055e-01 -2.80628055e-01 4.00935024e-01 3.37947011e-01 -7.81517208e-01 4.74582836e-02 -5.13478994e-01 -6.55016303e-01 -3.79521586e-02 -9.77196455e-01 -2.17280611e-01 -6.02200866e-01 -2.22011924e-01 -4.46566463e-01 8.87247920e-01 -4.63396788e-01 1.60647988e+00 -2.14649820e+00 5.30560873e-02 -4.98211719e-02 -1.38703749e-01 -2.19707321e-02 4.16482314e-02 6.56310081e-01 3.14576514e-02 9.73058715e-02 -2.19791532e-01 -4.94058579e-01 8.14491287e-02 3.59583229e-01 -6.21926248e-01 -9.59396642e-03 2.00868979e-01 8.28710079e-01 -1.14705694e+00 -5.91223657e-01 1.73076704e-01 4.63714391e-01 -4.27035272e-01 1.32114649e-01 -2.76700884e-01 6.65187418e-01 -7.05264390e-01 4.01320308e-01 3.72082382e-01 -1.81173265e-01 5.58429778e-01 1.19866729e-01 -2.09459551e-02 3.14678043e-01 -8.77290785e-01 1.82369161e+00 -1.10431981e+00 4.98553604e-01 1.21219501e-01 -1.00776386e+00 1.12349367e+00 5.98435998e-01 2.96590745e-01 -4.92896438e-01 1.29618660e-01 7.68300444e-02 -2.43325025e-01 -6.91244960e-01 8.07082236e-01 -6.64541662e-01 -4.61860031e-01 4.75663126e-01 -1.86877027e-01 2.72716098e-02 -2.15175860e-02 2.42435262e-01 6.61402583e-01 1.00964149e-02 3.03602904e-01 2.07950771e-01 8.94843102e-01 -4.62239414e-01 1.02283311e+00 4.01906788e-01 -6.88959956e-01 4.85030532e-01 4.72214133e-01 8.26947764e-03 -4.11177337e-01 -5.54407418e-01 -6.67167380e-02 9.72714365e-01 4.01723444e-01 -6.89059675e-01 -5.49587905e-01 -6.81824207e-01 -4.11419153e-01 1.01911926e+00 -6.00564301e-01 -1.70247063e-01 -6.03879035e-01 -5.75287104e-01 3.70671302e-01 3.66950631e-01 4.16546881e-01 -1.25240862e+00 -4.13735420e-01 5.67397416e-01 -8.64376485e-01 -1.40290308e+00 -4.81230974e-01 -1.40871704e-01 -6.42916501e-01 -6.58911884e-01 -6.00285709e-01 -3.02618712e-01 -3.05236485e-02 4.29746032e-01 1.07531106e+00 -4.30156291e-02 7.66012520e-02 5.87399662e-01 -7.58828521e-01 -3.42403144e-01 -5.75358748e-01 5.04545681e-03 1.64833874e-01 4.55980629e-01 6.57780588e-01 -1.05778360e+00 -4.69544977e-01 2.67746359e-01 -6.29494965e-01 1.80952132e-01 2.57165730e-01 9.36655998e-01 -4.63808998e-02 -5.26526533e-02 1.43574381e+00 -7.69097567e-01 8.77782464e-01 -8.27475011e-01 1.39698446e-01 3.98394942e-01 -2.84630179e-01 -1.62808985e-01 8.66809666e-01 -5.93162775e-01 -1.92354774e+00 -4.83425170e-01 -1.90895513e-01 -2.54785448e-01 -1.93145245e-01 6.77637875e-01 -5.30297101e-01 6.58044875e-01 4.49564666e-01 4.71867114e-01 -2.09359795e-01 -3.01906556e-01 4.39427376e-01 1.04615748e+00 5.12020051e-01 -1.05465770e+00 3.70534122e-01 3.83522928e-01 -6.69396937e-01 -8.14525723e-01 -9.64967966e-01 -4.51618850e-01 -3.54139894e-01 -5.73881686e-01 5.81278205e-01 -8.74127030e-01 -6.95881784e-01 3.39593709e-01 -1.29903531e+00 -2.25719493e-02 -7.34709995e-03 5.64574599e-01 -3.12685877e-01 5.20796835e-01 -5.31549454e-01 -1.27267194e+00 -8.73643458e-02 -1.03096950e+00 9.28528309e-01 4.13202822e-01 -4.21198964e-01 -1.19105661e+00 -1.30035460e-01 4.50553536e-01 4.74108279e-01 2.38848731e-01 6.37770534e-01 -6.86295569e-01 -2.57430524e-01 -1.14878170e-01 1.37642890e-01 3.50897640e-01 3.82225126e-01 -8.99957493e-02 -1.25195408e+00 1.83541059e-01 2.89457440e-01 -3.11143696e-01 4.80228305e-01 -2.77997345e-01 1.23478782e+00 -3.04928094e-01 -3.49124521e-01 5.16964123e-02 1.03915310e+00 3.41387630e-01 3.34725738e-01 1.13994792e-01 2.34163105e-01 1.05631626e+00 7.10102201e-01 8.68125975e-01 7.35247910e-01 6.06955707e-01 1.45071819e-01 4.68083560e-01 2.68912941e-01 -2.51322806e-01 3.98607075e-01 1.36131990e+00 -3.58801149e-02 -2.00823948e-01 -7.69072592e-01 8.01222026e-01 -1.86651599e+00 -1.24015021e+00 1.71451584e-01 1.78277016e+00 1.27762437e+00 9.54989791e-02 -3.24019551e-01 2.34200247e-02 9.73506868e-01 5.04697442e-01 -5.10164797e-01 -2.71556020e-01 -3.95498067e-01 -2.14245752e-01 -5.54468751e-01 2.25881889e-01 -8.11667442e-01 8.07333589e-01 5.38995314e+00 8.40254664e-01 -1.15780354e+00 7.83995911e-02 5.97309947e-01 -1.21410750e-02 -8.43438148e-01 2.18109176e-01 -6.17287338e-01 7.28221118e-01 9.93788421e-01 -6.45731270e-01 4.68296260e-01 9.42752302e-01 3.60575408e-01 1.08787559e-01 -1.04425883e+00 1.11528349e+00 7.63139874e-02 -1.09680724e+00 -1.35363117e-01 -1.29580393e-01 5.53002834e-01 -4.05569166e-01 -1.43664956e-01 8.86077106e-01 3.82228270e-02 -4.84279275e-01 5.68600357e-01 4.42617804e-01 4.48060423e-01 -8.06458175e-01 6.73461258e-01 5.53820431e-01 -1.03830481e+00 -1.60585359e-01 -6.45374740e-03 -2.88823605e-01 4.47958082e-01 7.14618385e-01 -6.78941250e-01 9.14599776e-01 5.92413902e-01 8.11316669e-01 -2.73494452e-01 2.04906702e-01 -7.67295808e-02 6.66387200e-01 -1.88398898e-01 -1.84375197e-01 1.81653649e-01 -3.42817634e-01 6.53021395e-01 1.40725124e+00 2.68049836e-01 4.12541568e-01 7.90238157e-02 7.31791258e-01 -2.31314123e-01 3.74737233e-01 -6.92203343e-01 -3.98339108e-02 8.62037718e-01 1.36605144e+00 -3.38258088e-01 -4.37418669e-01 -5.20668685e-01 9.08252180e-01 3.05217296e-01 4.64769423e-01 -9.22782123e-01 -2.66424000e-01 7.24998236e-01 -5.42489767e-01 -1.11478776e-01 -3.26992497e-02 -6.94168359e-02 -1.58086908e+00 3.79354954e-01 -9.40756261e-01 2.54673183e-01 -6.64473534e-01 -1.61526799e+00 7.70029187e-01 1.15422904e-02 -1.19808960e+00 -5.55514455e-01 -1.36234745e-01 -9.88032877e-01 6.38145208e-01 -1.58830452e+00 -1.03923619e+00 -3.21789235e-01 5.14316499e-01 7.89808273e-01 9.08414498e-02 8.63954604e-01 1.34597540e-01 -7.88894892e-01 5.72183967e-01 -1.12630785e-01 -1.54340729e-01 9.04850543e-01 -1.00564516e+00 -8.00458789e-02 7.29293525e-01 -6.35184050e-02 1.04096544e+00 8.69876683e-01 -4.60165590e-01 -1.30932212e+00 -7.01866210e-01 1.17948389e+00 -2.76603639e-01 9.28802907e-01 -4.25603330e-01 -1.20164931e+00 6.21923268e-01 3.51346999e-01 -4.04414475e-01 1.01286626e+00 6.61105156e-01 -2.34306410e-01 -1.20582521e-01 -8.15516353e-01 1.09995723e+00 1.36407328e+00 -9.21698034e-01 -1.01828897e+00 1.64156985e-02 9.96934950e-01 -1.20972849e-01 -7.43902802e-01 4.61260259e-01 4.18820649e-01 -1.05743396e+00 6.96714103e-01 -3.65866840e-01 4.79903221e-01 -2.82866135e-02 -1.76906273e-01 -1.40483665e+00 4.14249897e-01 -1.13285697e+00 -1.20963827e-02 1.86366820e+00 2.92332888e-01 -7.34516084e-01 4.02403593e-01 1.00640750e+00 -2.38846704e-01 -6.69628680e-01 -9.18853462e-01 -6.20405912e-01 -1.31579086e-01 -9.65538085e-01 9.98935163e-01 1.34947467e+00 5.61272562e-01 4.40792829e-01 -5.27866304e-01 -1.35691479e-01 -6.59822598e-02 4.86493230e-01 7.96918094e-01 -1.04491246e+00 -3.19592357e-01 -5.16234279e-01 -4.89123072e-03 -1.14099693e+00 7.40571797e-01 -6.69537127e-01 1.28707200e-01 -1.03052711e+00 -9.58544668e-04 -5.55517435e-01 -2.76315480e-01 3.52515936e-01 -6.35461688e-01 -3.09924364e-01 1.28709897e-01 3.87232043e-02 -5.83743632e-01 1.23669863e+00 1.19234812e+00 5.13305180e-02 -3.85791302e-01 -4.12727483e-02 -8.57510209e-01 8.30867171e-01 8.61846924e-01 2.36986000e-02 -5.70407271e-01 2.37813871e-03 -1.61111727e-03 3.55405211e-01 1.19071409e-01 -5.85574627e-01 6.90937415e-02 -6.32116497e-01 -1.90223187e-01 -3.55426818e-01 5.75616181e-01 -6.81970716e-01 7.89638311e-02 -2.84835577e-01 -4.28241134e-01 -1.89632088e-01 1.16256505e-01 1.00342751e+00 -6.32514358e-01 -2.51616806e-01 1.67938650e-01 -2.04617217e-01 -9.45527136e-01 -4.98216525e-02 -3.11313719e-01 8.96299332e-02 9.59586561e-01 -1.86969280e-01 -1.45319715e-01 -7.07821190e-01 -7.00115263e-01 2.95213729e-01 3.35469663e-01 7.76357412e-01 4.54683006e-01 -1.28655505e+00 -5.30750692e-01 -1.16364442e-01 3.85055512e-01 -2.36034125e-01 7.93189824e-01 7.95000136e-01 4.18929517e-01 2.89531916e-01 3.04499835e-01 -4.62652922e-01 -1.08926690e+00 6.07477427e-01 7.18476772e-02 -1.28100872e-01 -5.79184890e-01 5.25856972e-01 3.10667247e-01 -3.86288881e-01 9.61264968e-02 -1.35612801e-01 -2.73615301e-01 3.32426101e-01 5.14712393e-01 -7.18255714e-02 -2.99095273e-01 -6.91320837e-01 -3.70711267e-01 1.56245843e-01 5.39566614e-02 -2.36240312e-01 1.28551197e+00 -7.55219281e-01 -6.90388447e-03 8.78473997e-01 1.20459485e+00 2.73705661e-01 -1.16421938e+00 -6.61638618e-01 3.02012295e-01 -4.55123991e-01 -1.87768444e-01 -7.59344518e-01 -7.33174324e-01 8.94237578e-01 -1.49493560e-01 3.77595097e-01 1.30987203e+00 2.71574911e-02 1.03368807e+00 3.88411582e-01 5.89194000e-01 -1.35664856e+00 1.93468049e-01 4.98241454e-01 1.15515339e+00 -1.29671967e+00 -4.45409209e-01 -4.61920351e-01 -1.08698702e+00 1.05212414e+00 6.43971503e-01 5.67481339e-01 5.24914384e-01 -7.65053630e-02 1.94244072e-01 1.87723026e-01 -1.11200905e+00 -2.29796499e-01 -1.40464351e-01 3.30696702e-01 5.32239079e-01 1.59645557e-01 -4.61048096e-01 1.15153837e+00 -3.45920175e-01 -6.40181601e-02 5.52347779e-01 8.75892222e-01 1.99261885e-02 -1.18289125e+00 -1.08940654e-01 -2.89327819e-02 -3.74860853e-01 -1.09114893e-01 -3.78046572e-01 6.10866010e-01 -9.59950760e-02 1.40071881e+00 -2.53955215e-01 -4.06509757e-01 1.61832795e-01 5.40365219e-01 -8.16085115e-02 -3.49740326e-01 -6.38385415e-01 4.02060933e-02 5.12798190e-01 -3.24227154e-01 -7.46597528e-01 -7.60063231e-01 -1.15860021e+00 -2.64873981e-01 -5.93853354e-01 2.84996063e-01 6.21599853e-01 1.17310953e+00 5.79744577e-01 3.38516861e-01 1.04612386e+00 -6.61895335e-01 -5.13980806e-01 -1.08036566e+00 -3.48164439e-01 6.57913029e-01 1.94047391e-01 -6.76517606e-01 -6.89454854e-01 9.07604545e-02]
[13.028871536254883, 6.092994689941406]
bd843c68-eb47-4b24-a4f7-9ead75bac21d
benchmarking-joint-face-spoofing-and-forgery
2208.05401
null
https://arxiv.org/abs/2208.05401v1
https://arxiv.org/pdf/2208.05401v1.pdf
Benchmarking Joint Face Spoofing and Forgery Detection with Visual and Physiological Cues
Face anti-spoofing (FAS) and face forgery detection play vital roles in securing face biometric systems from presentation attacks (PAs) and vicious digital manipulation (e.g., deepfakes). Despite promising performance upon large-scale data and powerful deep models, the generalization problem of existing approaches is still an open issue. Most of recent approaches focus on 1) unimodal visual appearance or physiological (i.e., remote photoplethysmography (rPPG)) cues; and 2) separated feature representation for FAS or face forgery detection. On one side, unimodal appearance and rPPG features are respectively vulnerable to high-fidelity face 3D mask and video replay attacks, inspiring us to design reliable multi-modal fusion mechanisms for generalized face attack detection. On the other side, there are rich common features across FAS and face forgery detection tasks (e.g., periodic rPPG rhythms and vanilla appearance for bonafides), providing solid evidence to design a joint FAS and face forgery detection system in a multi-task learning fashion. In this paper, we establish the first joint face spoofing and forgery detection benchmark using both visual appearance and physiological rPPG cues. To enhance the rPPG periodicity discrimination, we design a two-branch physiological network using both facial spatio-temporal rPPG signal map and its continuous wavelet transformed counterpart as inputs. To mitigate the modality bias and improve the fusion efficacy, we conduct a weighted batch and layer normalization for both appearance and rPPG features before multi-modal fusion. We find that the generalization capacities of both unimodal (appearance or rPPG) and multi-modal (appearance+rPPG) models can be obviously improved via joint training on these two tasks. We hope this new benchmark will facilitate the future research of both FAS and deepfake detection communities.
['Alex C. Kot', 'Jingang Shi', 'Wenhan Yang', 'Zhi Li', 'Rizhao Cai', 'Zitong Yu']
2022-08-10
null
null
null
null
['face-anti-spoofing']
['computer-vision']
[ 1.58967614e-01 -3.87912333e-01 5.80675602e-02 -2.57928163e-01 -5.76579452e-01 -5.50667882e-01 4.54367578e-01 -6.73666894e-01 8.95655379e-02 4.61318880e-01 7.67611191e-02 -4.61115967e-03 8.61042961e-02 -5.58471680e-01 -5.85227907e-01 -1.32628882e+00 -6.46984801e-02 -5.30085266e-01 -2.01703146e-01 -2.66928166e-01 1.31657332e-01 7.75204182e-01 -1.76969051e+00 5.88751435e-01 5.11170030e-01 1.38808537e+00 -3.24374676e-01 5.00909626e-01 2.30308428e-01 9.32064578e-02 -6.10822320e-01 -6.11152947e-01 2.61859179e-01 -3.55473846e-01 -3.71124744e-01 -1.81659430e-01 4.72492933e-01 -3.70679379e-01 -4.46100026e-01 1.11153018e+00 1.11680210e+00 -2.42695719e-01 5.76200843e-01 -1.58186853e+00 -9.74452376e-01 -5.73432818e-02 -1.10526788e+00 3.23745966e-01 5.35015345e-01 4.88474935e-01 2.30246812e-01 -1.11445224e+00 1.56686276e-01 1.53327835e+00 8.87449801e-01 8.31525385e-01 -9.69653428e-01 -1.18296778e+00 -2.51595587e-01 2.96855271e-01 -1.44352067e+00 -9.54118848e-01 1.21295583e+00 -1.23327792e-01 5.88642120e-01 2.91872978e-01 3.35338503e-01 1.76341677e+00 3.85339409e-01 4.87458795e-01 1.38773143e+00 1.53450761e-02 -2.77266383e-01 1.25398502e-01 -2.90204555e-01 6.81109548e-01 1.09441899e-01 4.61429983e-01 -6.52157605e-01 -4.44879770e-01 1.01379728e+00 -1.08621061e-01 -5.83609223e-01 3.70194703e-01 -7.39922523e-01 4.92178857e-01 9.17813331e-02 3.67633313e-01 -2.80419141e-01 6.87528178e-02 5.08422792e-01 6.01648331e-01 1.98851898e-01 1.46674439e-02 -3.18540394e-01 1.69523612e-01 -7.88000822e-01 1.03668626e-02 5.72543025e-01 3.10385257e-01 7.08638012e-01 2.85395294e-01 -1.48954764e-01 1.00561845e+00 4.97209042e-01 7.43467927e-01 6.93291426e-01 -6.69840276e-01 2.37678304e-01 1.40984550e-01 -1.97417006e-01 -1.62276542e+00 -4.76759076e-01 -2.31863968e-02 -9.25354481e-01 1.34650096e-01 4.27826196e-01 -1.39873102e-01 -6.81011856e-01 1.98894048e+00 3.87752473e-01 6.56089842e-01 -5.56950644e-02 9.74938929e-01 1.14772689e+00 4.22356337e-01 6.07903376e-02 -6.39589727e-01 1.69891369e+00 -3.15556675e-01 -9.12866473e-01 6.14276249e-03 2.91800708e-01 -8.79476070e-01 9.05843079e-01 4.36561853e-01 -7.98731744e-01 -7.69014716e-01 -9.80425298e-01 1.17991552e-01 -2.74058104e-01 1.56834006e-01 5.34526825e-01 1.31967354e+00 -1.03495848e+00 5.62922359e-01 -4.75013494e-01 -9.14738402e-02 6.41486645e-01 4.33691144e-01 -7.92718232e-01 -1.69206876e-02 -1.48143888e+00 6.25350714e-01 -5.28426096e-02 5.92384875e-01 -9.59525764e-01 -5.84627211e-01 -7.29907811e-01 -1.94833994e-01 -4.26223278e-02 -2.95556366e-01 4.68706816e-01 -1.00463951e+00 -1.64564180e+00 1.13873053e+00 -5.40295877e-02 1.32059619e-01 2.23105416e-01 1.61897182e-01 -1.09463990e+00 4.13883358e-01 -3.07582319e-01 4.18514341e-01 1.75062096e+00 -1.18347883e+00 1.14426181e-01 -9.13895726e-01 -4.34325039e-01 -6.62312061e-02 -7.62870669e-01 4.94955242e-01 7.66310394e-02 -7.85851777e-01 4.46981899e-02 -5.64168215e-01 5.59562862e-01 2.67752200e-01 -9.44425166e-02 -3.15930724e-01 1.35803449e+00 -1.06492019e+00 9.79292393e-01 -2.35509181e+00 -2.09191397e-01 6.54744403e-03 1.75332367e-01 6.70363426e-01 -6.08043909e-01 1.79458171e-01 -4.84729052e-01 2.17155337e-01 1.57721918e-02 -3.44180912e-01 -2.82308668e-01 -1.11945882e-01 -3.50610048e-01 9.78329182e-01 4.77437735e-01 9.23974991e-01 -6.40601873e-01 -4.13252264e-01 6.09414279e-02 9.07464087e-01 -2.70508528e-01 1.61037892e-02 3.98972541e-01 5.45569479e-01 -3.29664826e-01 1.17939723e+00 1.38954365e+00 2.22523153e-01 -2.33674988e-01 -7.44240582e-01 3.46521944e-01 -1.64407969e-01 -1.04720986e+00 1.43286359e+00 -3.11118573e-01 4.72033948e-01 5.32147825e-01 -1.05927205e+00 1.12538290e+00 7.54008353e-01 6.55410945e-01 -6.16202772e-01 3.54343474e-01 2.63851762e-01 -7.13303685e-02 -8.88684452e-01 -9.02420729e-02 -4.62721616e-01 1.97480455e-01 4.02524382e-01 4.40507382e-01 2.84515113e-01 -6.20419145e-01 -3.79445642e-01 8.43430638e-01 2.37942729e-02 -1.65541649e-01 -2.11881116e-01 8.59383464e-01 -1.04453886e+00 7.32101381e-01 2.80661881e-01 -8.25704098e-01 6.57053769e-01 3.98411006e-01 -3.71621519e-01 -5.22189736e-01 -9.44711089e-01 -4.03237641e-01 9.65497851e-01 1.25566557e-01 -1.48033813e-01 -5.35148919e-01 -7.41071761e-01 3.36702466e-01 -9.05449688e-02 -6.41342700e-01 -5.79457521e-01 -5.72726429e-01 -1.09060299e+00 1.41141617e+00 3.54961812e-01 7.45672822e-01 -1.16967452e+00 -3.49566303e-02 -4.01563980e-02 -2.40279108e-01 -1.15476656e+00 -3.23788226e-01 -3.63197535e-01 -5.68560421e-01 -1.14414632e+00 -8.22825789e-01 -7.86696970e-01 3.32503289e-01 3.99861485e-01 4.37851876e-01 1.34161040e-01 -4.97016668e-01 3.75980228e-01 -1.11227795e-01 -2.24962622e-01 -2.61497140e-01 -8.23160648e-01 6.13756299e-01 6.31424904e-01 4.45026428e-01 -9.20848548e-01 -7.51363039e-01 6.54317856e-01 -8.74102294e-01 -4.20798659e-01 5.11364758e-01 8.17780197e-01 2.27888543e-02 7.86367208e-02 1.02583456e+00 -1.13595858e-01 8.36710930e-01 -3.40057373e-01 -1.22467421e-01 2.57933348e-01 -1.82123393e-01 -4.39670354e-01 5.49861908e-01 -7.65996039e-01 -1.26027286e+00 -2.85705477e-01 -3.51224631e-01 -9.39298511e-01 -3.56481403e-01 2.87303507e-01 -5.09164572e-01 -7.91553020e-01 8.10743570e-01 5.26705205e-01 3.70676190e-01 -3.48268569e-01 3.01092006e-02 8.70969713e-01 6.37069285e-01 -5.49729407e-01 8.95079911e-01 5.59444010e-01 7.05781952e-02 -1.12849259e+00 1.74704585e-02 -2.78571188e-01 -2.45870575e-01 -3.66071254e-01 5.51787198e-01 -9.85415816e-01 -1.29692316e+00 1.28192174e+00 -1.32357061e+00 3.17464739e-01 4.87573773e-01 3.47273260e-01 -3.27557445e-01 1.09036589e+00 -8.52615893e-01 -1.12299323e+00 -4.20615226e-01 -9.22131062e-01 1.22074783e+00 3.29210073e-01 2.37639248e-01 -7.14389145e-01 -6.48300052e-02 5.07564783e-01 4.72503185e-01 4.12019432e-01 5.43722153e-01 -4.21176314e-01 1.06167600e-01 -3.33603799e-01 -4.75767910e-01 6.46736681e-01 3.98965597e-01 1.18605345e-01 -1.79871058e+00 -5.11055529e-01 8.01936865e-01 -5.96154571e-01 9.15801108e-01 3.19260776e-01 1.23295820e+00 -3.16354990e-01 -2.45968997e-01 7.98433721e-01 1.06163132e+00 1.39108449e-01 9.01453078e-01 -4.09224242e-01 6.25852585e-01 8.45178783e-01 1.19981416e-01 5.17295778e-01 -1.05214618e-01 5.29479861e-01 4.31159526e-01 -4.48843986e-02 -1.86231375e-01 -1.32404342e-01 8.17645550e-01 3.54509383e-01 -3.78841817e-01 -8.95370394e-02 -5.43869495e-01 2.01413274e-01 -1.37119639e+00 -1.10144663e+00 2.05863118e-01 2.11197805e+00 6.64961040e-01 -4.83881950e-01 3.16399559e-02 3.02206010e-01 1.01414645e+00 3.46618563e-01 -5.52786231e-01 -2.12904468e-01 -5.79260349e-01 2.10393667e-01 7.06075579e-02 -5.76261021e-02 -1.17978275e+00 5.96785188e-01 5.24135303e+00 1.10936809e+00 -1.42016733e+00 2.57743686e-01 6.09977305e-01 2.25721039e-02 -5.79268932e-02 -4.99735922e-01 -8.74499500e-01 7.60838032e-01 8.31710577e-01 3.07173073e-01 6.14960670e-01 3.93779159e-01 1.87002033e-01 3.66231441e-01 -7.70862877e-01 1.63311625e+00 4.21881735e-01 -9.37278450e-01 -8.84778574e-02 2.15617061e-01 2.10130140e-01 -4.22374189e-01 3.51170480e-01 1.47999078e-01 -4.09520864e-01 -1.25127780e+00 5.34867793e-02 1.93657726e-01 1.03424489e+00 -5.26636302e-01 6.08783305e-01 5.11933118e-02 -1.29023635e+00 -2.91579664e-01 -4.00324345e-01 2.95503438e-01 -9.46138927e-04 3.54087383e-01 -1.46403030e-01 7.89139867e-01 7.22425044e-01 6.86166048e-01 -2.69803911e-01 6.42420530e-01 2.56939675e-03 3.16289157e-01 -4.44589436e-01 4.66962308e-01 -2.77982146e-01 4.05940339e-02 6.15650058e-01 1.05572820e+00 3.80932570e-01 1.62361771e-01 -2.37892210e-01 9.33392048e-01 -4.59894873e-02 -9.22806635e-02 -7.17715800e-01 -3.28425378e-01 3.71587515e-01 1.60970604e+00 -3.62885118e-01 1.66741550e-01 -4.39090937e-01 9.97076452e-01 -1.79693639e-01 4.47224617e-01 -9.38322306e-01 -3.46835613e-01 9.93295014e-01 -1.24879204e-01 -7.97861218e-02 6.08987957e-02 -1.48589253e-01 -1.40577626e+00 1.75768882e-01 -9.23436463e-01 3.89791042e-01 -4.95189637e-01 -1.56986618e+00 5.51801920e-01 -4.03342932e-01 -1.06999075e+00 1.66281909e-01 -8.69086504e-01 -8.71504188e-01 1.06042874e+00 -1.74796903e+00 -1.45851445e+00 -4.60075498e-01 1.10928202e+00 4.89529260e-02 -1.68315277e-01 8.07714701e-01 5.77062368e-01 -8.53602469e-01 1.06048298e+00 -4.15169448e-01 2.52514899e-01 9.82144952e-01 -3.87154311e-01 1.11101218e-01 6.88471854e-01 -3.05557817e-01 6.60078347e-01 3.08868498e-01 -5.52814722e-01 -1.78016925e+00 -7.45257497e-01 4.00524437e-01 -2.25946516e-01 4.68997359e-01 -2.45950088e-01 -1.11078429e+00 1.77797422e-01 -4.45589833e-02 3.17376703e-01 7.76319861e-01 -2.63427258e-01 -6.92530096e-01 -1.75915062e-01 -1.66604435e+00 2.89798707e-01 9.21462297e-01 -9.81893361e-01 -2.66302884e-01 3.03449154e-01 2.59405673e-01 1.35569423e-01 -9.94086802e-01 7.90284038e-01 9.95556474e-01 -1.10806930e+00 1.09954226e+00 -5.06659508e-01 1.05954021e-01 -1.28729284e-01 -1.38223737e-01 -9.57215250e-01 -2.03848332e-01 -9.48795199e-01 -3.12582791e-01 1.40997303e+00 -2.34492227e-01 -8.85940611e-01 5.85018814e-01 2.45447367e-01 1.54966516e-02 -5.47232032e-01 -1.17915559e+00 -8.53110850e-01 -7.93377757e-02 -3.18802059e-01 4.87588435e-01 1.15740871e+00 1.26137614e-01 -1.84255615e-01 -6.44490659e-01 3.65343928e-01 6.16532564e-01 -2.50075251e-01 3.96539032e-01 -1.22218800e+00 -3.44749540e-02 -4.44108218e-01 -5.17423451e-01 -6.13537490e-01 3.21407408e-01 -6.26172125e-01 -3.11915487e-01 -4.64153647e-01 -9.79051068e-02 -6.44867122e-02 -5.57490706e-01 6.21979058e-01 8.44639074e-03 6.67313814e-01 -1.50038674e-01 1.35702029e-01 3.13120812e-01 6.88283443e-01 1.34703505e+00 8.00012946e-02 -1.41272739e-01 -6.11424558e-02 -6.65408194e-01 5.25967777e-01 6.20255470e-01 -7.27573708e-02 -2.39986435e-01 -1.78100932e-02 -1.04059100e-01 3.67664307e-01 7.68917024e-01 -8.85144114e-01 1.31024957e-01 -1.34478267e-02 8.22375357e-01 -1.26922622e-01 6.80746555e-01 -5.58356166e-01 2.40898486e-02 4.68978971e-01 2.79892266e-01 -1.70655757e-01 4.54631597e-01 5.30563891e-01 -2.23609433e-01 3.19488049e-01 1.15184355e+00 -1.15478545e-01 -3.65397692e-01 5.44616461e-01 -1.71147600e-01 -4.22689497e-01 8.22499275e-01 -6.27234280e-01 -7.23246932e-01 -2.62041837e-01 -6.81284726e-01 -1.14018194e-01 1.70079350e-01 7.23894298e-01 1.05346906e+00 -1.44187224e+00 -7.58746445e-01 9.42548454e-01 -9.96858180e-02 -7.84909904e-01 6.65274918e-01 1.15734923e+00 2.14535533e-03 2.03212574e-01 -7.52316117e-01 -4.72070038e-01 -1.39009023e+00 6.61728978e-01 5.52138209e-01 3.24256837e-01 -1.36855274e-01 1.05343163e+00 3.23186755e-01 -2.21223786e-01 -3.63565376e-03 2.02517793e-01 -3.03091228e-01 2.82139808e-01 8.88245344e-01 4.60310072e-01 1.51463464e-01 -8.84654045e-01 -5.13728023e-01 7.89811730e-01 3.55795063e-02 2.48271272e-01 1.08002150e+00 -9.50767547e-02 -3.03325504e-01 -9.32846293e-02 1.48492467e+00 -1.35099173e-01 -1.03024209e+00 -9.09803957e-02 -5.49469054e-01 -7.09773481e-01 3.56551260e-02 -5.83912909e-01 -1.34595823e+00 1.12744808e+00 9.30495679e-01 2.20808104e-01 1.41652703e+00 -2.17204496e-01 7.57667661e-01 -1.86300986e-02 2.62148499e-01 -7.57519126e-01 3.71554703e-01 -1.34677365e-01 1.16153741e+00 -1.16442657e+00 -3.04634243e-01 -6.49226427e-01 -3.38543981e-01 1.35262311e+00 6.98423207e-01 2.96537317e-02 9.99570906e-01 2.45208040e-01 6.23910725e-02 -3.18703234e-01 -4.10092622e-01 2.63477355e-01 3.84236038e-01 9.25241470e-01 2.33875692e-01 -2.28503555e-01 -1.78223670e-01 7.76012599e-01 1.28831640e-01 -9.66332853e-03 3.50272469e-02 4.88849729e-01 -1.21898592e-01 -8.22536111e-01 -6.01028740e-01 2.60159075e-01 -6.36784494e-01 1.59752980e-01 -3.94694865e-01 3.69043022e-01 2.01538533e-01 1.07254934e+00 -2.13739440e-01 -7.55913496e-01 -6.54640868e-02 3.03687155e-01 7.15628326e-01 -1.10987134e-01 -6.33048713e-01 2.05915689e-01 -2.59188086e-01 -6.72640741e-01 -4.10221726e-01 -4.27729696e-01 -6.81029320e-01 -5.65484524e-01 -5.94168305e-01 -4.14615780e-01 5.61559021e-01 8.66851509e-01 4.35410202e-01 -1.07690699e-01 1.05636859e+00 -1.08868814e+00 -6.03752077e-01 -1.06783450e+00 -8.37317705e-01 5.41206121e-01 5.52214861e-01 -9.08164144e-01 -6.67572737e-01 -1.23014301e-01]
[13.0045804977417, 1.1496691703796387]
56b284ca-4438-44d1-8431-f76a7bfe8e82
sequential-edge-detection-using-joint
2302.14247
null
https://arxiv.org/abs/2302.14247v1
https://arxiv.org/pdf/2302.14247v1.pdf
Sequential edge detection using joint hierarchical Bayesian learning
This paper introduces a new sparse Bayesian learning (SBL) algorithm that jointly recovers a temporal sequence of edge maps from noisy and under-sampled Fourier data. The new method is cast in a Bayesian framework and uses a prior that simultaneously incorporates intra-image information to promote sparsity in each individual edge map with inter-image information to promote similarities in any unchanged regions. By treating both the edges as well as the similarity between adjacent images as random variables, there is no need to separately form regions of change. Thus we avoid both additional computational cost as well as any information loss resulting from pre-processing the image. Our numerical examples demonstrate that our new method compares favorably with more standard SBL approaches.
['Guohui Song', 'Anne Gelb', 'Yao Xiao']
2023-02-28
null
null
null
null
['edge-detection']
['computer-vision']
[ 3.98407727e-01 8.58263019e-03 4.47222143e-02 -3.52061808e-01 -9.05588627e-01 -2.53976583e-01 6.50739551e-01 -1.33321825e-02 -4.18463349e-01 7.25424349e-01 1.84180737e-01 3.52605432e-01 -3.94127369e-01 -5.73326766e-01 -6.30729258e-01 -9.23066258e-01 -3.43835384e-01 9.44812819e-02 5.65255165e-01 2.04513714e-01 2.50164419e-01 3.18364322e-01 -1.20814300e+00 1.57524318e-01 4.08984274e-01 8.96944344e-01 3.38343918e-01 7.31953263e-01 1.85613751e-01 7.45513439e-01 -9.53294560e-02 1.26829818e-01 4.40715939e-01 -3.70004922e-01 -6.32437885e-01 5.53514659e-01 5.36721110e-01 -2.50117093e-01 -7.15486884e-01 1.28003407e+00 3.80237907e-01 4.02943194e-01 6.84401512e-01 -7.51269400e-01 -1.03170849e-01 1.58628255e-01 -1.10596013e+00 4.60796893e-01 3.05712491e-01 -1.56774595e-01 7.08200514e-01 -8.08801770e-01 8.79509747e-01 1.06764543e+00 1.01089954e+00 -2.09028229e-01 -1.66426349e+00 -1.27171680e-01 4.77248356e-02 4.09832984e-01 -1.48175490e+00 -6.34063423e-01 1.19876933e+00 -4.92381573e-01 4.77364987e-01 3.90527174e-02 6.07491791e-01 6.03149295e-01 5.74950635e-01 5.94134271e-01 1.23986149e+00 -6.52222395e-01 2.23408163e-01 -1.78051144e-01 4.18322533e-02 8.77970815e-01 7.13412613e-02 4.10726845e-01 -8.72783959e-01 -2.89892614e-01 9.65500832e-01 -2.04322293e-01 -4.76655930e-01 -7.23417044e-01 -1.16192639e+00 7.51601636e-01 1.76147595e-01 2.86092460e-01 -6.63561225e-01 3.36974710e-01 1.20802380e-01 1.49020210e-01 7.76624739e-01 -1.49761871e-01 -1.70532316e-01 1.46959228e-02 -1.40491319e+00 8.68132710e-03 6.57553971e-01 7.11681366e-01 9.88751709e-01 1.33189157e-01 1.94596171e-01 6.03757620e-01 3.60402107e-01 4.00820941e-01 2.14793816e-01 -1.48998523e+00 -7.06142783e-02 -3.54654282e-01 2.79415607e-01 -1.39215481e+00 -7.01761842e-02 -3.03476453e-01 -8.89927208e-01 3.80755663e-01 3.48976940e-01 -9.17490199e-02 -8.65348756e-01 1.69390428e+00 2.36461714e-01 7.29650795e-01 -2.96604902e-01 5.39400637e-01 4.89848822e-01 9.36987221e-01 -4.41824913e-01 -7.78565347e-01 8.67717743e-01 -5.67371249e-01 -9.43621755e-01 -1.85827464e-01 -1.27326697e-01 -1.02933455e+00 3.04318219e-01 4.72513109e-01 -1.41406083e+00 -5.73164940e-01 -1.04376221e+00 1.61976114e-01 -6.61880185e-04 -3.79335344e-01 4.55820382e-01 2.05324262e-01 -1.35115373e+00 7.66817153e-01 -1.18923199e+00 -7.22135901e-02 9.71813351e-02 2.55065233e-01 -3.52268130e-01 -2.46532232e-01 -9.10275757e-01 7.74955213e-01 2.53435373e-01 -2.91500222e-02 -7.01360345e-01 -6.67307973e-01 -1.04302597e+00 -1.50564432e-01 2.65784293e-01 -6.13182724e-01 1.04248953e+00 -8.06896150e-01 -1.30554163e+00 6.41198635e-01 -6.91655099e-01 -2.50565857e-01 4.35894877e-01 7.34864175e-03 -2.66035229e-01 4.11282063e-01 1.87790290e-01 7.17386603e-01 1.35808074e+00 -1.54729748e+00 -5.83077550e-01 -1.92756474e-03 -3.96849662e-01 2.96176642e-01 2.21372664e-01 -2.84289032e-01 -6.68873250e-01 -1.01481521e+00 7.56227672e-01 -7.64562130e-01 -3.69634628e-01 2.91929960e-01 -1.44140432e-02 2.72006780e-01 1.01952755e+00 -8.18841636e-01 1.01937318e+00 -2.16399240e+00 2.27436230e-01 5.97941041e-01 2.47728914e-01 -3.85574818e-01 -1.16946906e-01 8.23568255e-02 -2.42603377e-01 -3.73244703e-01 -5.97180843e-01 -3.89382511e-01 -5.17269194e-01 3.29102784e-01 -9.46858376e-02 9.58440423e-01 -8.88763219e-02 3.61108452e-01 -1.17819774e+00 -6.59475923e-01 4.84311074e-01 7.24546492e-01 -4.81823891e-01 -3.46014276e-02 2.26171687e-01 6.52806222e-01 -1.40738204e-01 2.91401476e-01 6.66481316e-01 -1.36894301e-01 1.31667210e-02 -5.09543180e-01 -1.71979740e-01 6.37966916e-02 -1.81945670e+00 1.74244106e+00 -3.45017612e-01 8.34998250e-01 6.51036441e-01 -1.27439749e+00 7.70420194e-01 5.40259719e-01 1.00697005e+00 -3.28618407e-01 -1.46638945e-01 1.00156300e-01 -3.47807944e-01 -4.60768163e-01 2.32019991e-01 -4.67079043e-01 2.43822098e-01 5.11072695e-01 7.30504245e-02 -4.70400095e-01 1.40859127e-01 2.19374299e-01 1.05433297e+00 1.61463410e-01 4.19515848e-01 -5.41224837e-01 3.33302706e-01 -1.24713950e-01 8.89639199e-01 8.78487945e-01 -2.58971900e-01 6.12666905e-01 -7.94714168e-02 -4.10624921e-01 -9.68535066e-01 -1.39543450e+00 -3.73481065e-01 5.35688579e-01 4.49427813e-01 -1.82684645e-01 -3.44861984e-01 -2.87502885e-01 -1.59027517e-01 2.18672022e-01 -3.64005983e-01 1.50985122e-01 -6.55503273e-01 -9.30311382e-01 -1.42557383e-01 1.87930375e-01 6.09845519e-01 -6.27730727e-01 -4.28444594e-01 4.01003003e-01 -4.08580780e-01 -9.62768137e-01 -6.75186217e-01 2.38968462e-01 -1.04572582e+00 -7.99735487e-01 -6.43798828e-01 -1.07839537e+00 8.22050333e-01 4.51577961e-01 1.02820313e+00 -4.23444897e-01 -4.69681978e-01 6.77339375e-01 -9.94333774e-02 1.67154789e-01 -2.57109970e-01 -5.67332387e-01 -2.80112214e-02 2.12244168e-01 -6.12401962e-02 -8.82184386e-01 -3.99197251e-01 9.94207710e-02 -6.72828794e-01 -4.57061492e-02 3.34119171e-01 1.09830642e+00 9.66412604e-01 6.99539483e-01 2.27728993e-01 -6.34293437e-01 1.73742771e-01 -3.98106813e-01 -8.17181349e-01 -1.08837020e-02 -5.64792335e-01 8.39551613e-02 -7.81036466e-02 -2.56517559e-01 -1.51050949e+00 3.45009148e-01 2.12603793e-01 -3.41475010e-01 -2.71904040e-02 6.80889249e-01 2.04871103e-01 -3.31795692e-01 3.97953987e-01 2.56560773e-01 1.81331977e-01 -5.31240523e-01 3.95814717e-01 2.85614550e-01 1.07035422e+00 -5.29635489e-01 7.96451628e-01 9.32578087e-01 2.86621958e-01 -1.14018643e+00 -6.65992975e-01 -8.39057088e-01 -8.37835371e-01 -3.66090149e-01 6.05491161e-01 -9.65058327e-01 -2.25600824e-01 5.78961253e-01 -1.08061039e+00 -1.94335133e-01 -4.37433869e-01 8.24527502e-01 -7.77246535e-01 8.46480191e-01 -9.01118934e-01 -5.80087125e-01 1.03499576e-01 -9.10256624e-01 8.83164227e-01 6.28829515e-03 -3.30088913e-01 -1.32441556e+00 2.61970818e-01 -3.88823755e-05 3.15868109e-01 1.77419469e-01 5.31689942e-01 7.02465773e-02 -7.35977292e-01 -2.73119926e-01 -1.08473860e-01 2.40685210e-01 3.97505313e-01 -2.46134371e-01 -6.89912260e-01 -5.27465701e-01 7.03918755e-01 -6.05916791e-02 9.75702643e-01 9.61792290e-01 7.73972511e-01 -1.68167993e-01 -4.57469940e-01 6.74273551e-01 1.61234343e+00 1.85668975e-01 4.73451197e-01 9.32185799e-02 5.07041991e-01 6.50070608e-01 4.29950178e-01 4.54761177e-01 -6.04991652e-02 5.69770813e-01 -3.33588980e-02 -3.52669507e-01 -3.26252609e-01 1.22523479e-01 2.00029939e-01 9.89320159e-01 7.59850768e-03 2.55553797e-02 -6.55969977e-01 8.43687832e-01 -1.79299378e+00 -1.29988122e+00 -1.63667440e-01 2.16172671e+00 1.05426180e+00 -7.45308995e-02 -2.50833601e-01 2.04543129e-01 9.59521890e-01 4.70873028e-01 -4.14215952e-01 1.20718241e-01 -1.52441248e-01 3.34203064e-01 6.65346086e-01 1.22980487e+00 -1.30003142e+00 3.93673748e-01 7.93924665e+00 7.52074897e-01 -6.61899269e-01 1.83950379e-01 6.14771962e-01 -2.47266497e-02 -1.93166673e-01 9.65090767e-02 -4.32957113e-01 3.22734684e-01 3.93283427e-01 -2.66273230e-01 3.95925879e-01 3.65812540e-01 5.99998593e-01 -4.90367830e-01 -8.25797975e-01 1.10270321e+00 1.98702693e-01 -1.31711411e+00 -3.36638004e-01 -1.18473671e-01 1.13950014e+00 1.60877138e-01 -2.72201359e-01 -5.02785146e-01 4.46787953e-01 -6.30241930e-01 6.44709945e-01 7.62256265e-01 5.19717872e-01 -4.92542803e-01 7.52591014e-01 1.68158874e-01 -1.33007169e+00 3.09111476e-01 -1.68555617e-01 -1.58655047e-01 6.34868026e-01 1.14642632e+00 -3.54108840e-01 4.17290479e-01 8.86931837e-01 1.04340398e+00 1.95845757e-02 1.15428662e+00 -2.26802722e-01 5.22422075e-01 -7.18730628e-01 7.89808929e-01 1.48099646e-01 -4.72901255e-01 9.87129092e-01 1.12882090e+00 3.87874663e-01 1.77442655e-01 5.78014314e-01 6.58126712e-01 2.63279229e-01 -3.03941637e-01 -5.79277992e-01 5.31539500e-01 4.77795035e-01 9.35169160e-01 -8.85060191e-01 -5.21706641e-01 -5.94102740e-01 1.06686521e+00 -2.91950613e-01 6.59266531e-01 -3.16376686e-01 -2.69297898e-01 4.28320289e-01 2.13203907e-01 5.76339364e-01 -3.87172461e-01 -4.05162215e-01 -1.08288515e+00 -2.77421791e-02 -4.41027462e-01 5.22789836e-01 -8.26757967e-01 -1.37391150e+00 1.76286265e-01 3.47971022e-01 -1.18248272e+00 -3.17146182e-01 -9.64288190e-02 -5.93826115e-01 8.36032808e-01 -1.54704010e+00 -9.31463778e-01 -8.36444348e-02 8.22384357e-01 5.05437195e-01 1.95013314e-01 3.15512598e-01 2.24952355e-01 -1.26872852e-01 1.13418818e-01 5.03946483e-01 -1.08013041e-01 7.74164259e-01 -1.06775439e+00 -4.72947806e-02 1.20488942e+00 1.18048251e-01 3.50703329e-01 9.07819211e-01 -8.38866711e-01 -7.80811369e-01 -7.17929006e-01 9.27272737e-01 -1.86858419e-02 7.03313410e-01 2.34280169e-01 -9.77908254e-01 7.58659244e-01 2.78858662e-01 2.40477636e-01 3.00907612e-01 -6.00952618e-02 -7.24976510e-02 -2.32352301e-01 -1.17614901e+00 2.48003811e-01 7.16538608e-01 -6.50049448e-01 -7.19077766e-01 1.97242260e-01 1.78622231e-01 -2.24694803e-01 -9.85097408e-01 5.35916090e-01 2.87350357e-01 -9.24307406e-01 1.15500951e+00 1.95264935e-01 -4.79920246e-02 -7.06923783e-01 -2.23611578e-01 -1.27432489e+00 -6.24935746e-01 -9.01730597e-01 -1.34725168e-01 1.01844227e+00 2.13933121e-02 -5.01472116e-01 8.80557716e-01 3.22747558e-01 -1.71878487e-01 -2.16093540e-01 -1.16472435e+00 -8.78307879e-01 -3.75827134e-01 -4.62107480e-01 -1.45680949e-01 1.09258389e+00 1.62486091e-01 1.53779075e-01 -6.01297200e-01 3.62442017e-01 1.28564477e+00 7.12434649e-02 1.48955569e-01 -1.07972074e+00 -5.83048403e-01 -2.57332653e-01 -5.36280513e-01 -1.08424211e+00 -3.76918241e-02 -7.11097956e-01 6.40982389e-01 -1.23585999e+00 3.83645564e-01 -2.83188045e-01 -2.45141953e-01 2.42510453e-01 -1.47909805e-01 4.00532722e-01 -2.41389826e-01 3.32617134e-01 -2.24660411e-01 4.21482205e-01 1.03152299e+00 -6.45896494e-02 -1.92814320e-01 -1.54091850e-01 -1.16069943e-01 1.09806645e+00 3.47305983e-01 -5.65938175e-01 -4.06812221e-01 -3.28513056e-01 -2.56490499e-01 1.77793592e-01 4.38297302e-01 -8.34868193e-01 4.77825761e-01 -4.46938463e-02 4.03112143e-01 -5.84381878e-01 4.16599482e-01 -7.65860140e-01 3.65733922e-01 4.62489486e-01 -8.46921057e-02 -7.42382705e-02 -2.08682753e-02 9.24062788e-01 -5.89317560e-01 -3.71946841e-01 1.27115691e+00 -2.16222838e-01 -9.23984170e-01 1.57301247e-01 -6.50539815e-01 -3.21183324e-01 8.41302156e-01 -3.34944636e-01 3.04607242e-01 -7.96977639e-01 -1.14628041e+00 7.08044097e-02 4.50473577e-01 -1.42575085e-01 4.26134735e-01 -1.24241686e+00 -6.39787436e-01 3.90589327e-01 -3.45448494e-01 7.34521076e-02 3.66764694e-01 8.35339963e-01 -4.23989266e-01 1.23917265e-02 -3.17112878e-02 -8.07017148e-01 -1.29179549e+00 4.54650998e-01 2.62209892e-01 -3.55461270e-01 -9.82508481e-01 9.44237947e-01 9.56171677e-02 2.19904948e-02 3.33219439e-01 1.39771119e-01 1.37258321e-01 3.62165831e-02 4.20621127e-01 6.72722697e-01 -2.01167256e-01 -7.30928898e-01 -4.68313098e-02 9.38564897e-01 7.89802000e-02 -5.55848002e-01 1.32726097e+00 -5.98364890e-01 -2.84367532e-01 5.26398480e-01 1.20834243e+00 -3.17297131e-02 -1.67588425e+00 -5.65565705e-01 4.55829799e-02 -6.43469214e-01 7.52436697e-01 -3.03480744e-01 -1.12087858e+00 3.81817877e-01 7.20721304e-01 -9.14558992e-02 1.32543790e+00 8.83309022e-02 6.39551818e-01 2.33794421e-01 4.31049019e-01 -1.25899220e+00 1.85541198e-01 3.05463552e-01 5.88247180e-01 -1.15758407e+00 4.50360626e-01 -7.40746439e-01 -4.84679528e-02 8.61357331e-01 -9.08274576e-02 -4.02935356e-01 1.31143653e+00 4.71881986e-01 -1.68031961e-01 -2.36521065e-01 -4.87789154e-01 -9.69453380e-02 1.58315957e-01 6.27250314e-01 3.08115453e-01 -3.16495717e-01 -3.00487399e-01 -1.05414659e-01 3.16693425e-01 2.22471476e-01 4.56982404e-01 1.25536740e+00 -4.73042220e-01 -1.09277141e+00 -7.66448319e-01 3.42312574e-01 -2.97396153e-01 -4.80635613e-02 3.71274263e-01 6.48221076e-01 2.33071610e-01 8.28788757e-01 3.42216790e-01 7.38815814e-02 -8.38719457e-02 2.83885133e-02 6.40580833e-01 -4.78729308e-01 1.76791817e-01 8.20283234e-01 7.84906298e-02 -5.47891140e-01 -9.02979255e-01 -1.02693236e+00 -1.03043807e+00 -1.69125214e-01 -2.59689599e-01 1.46336630e-01 4.02909875e-01 1.03796768e+00 8.22979361e-02 3.66790473e-01 6.96644008e-01 -1.24148691e+00 -2.57072330e-01 -7.58592010e-01 -9.56111550e-01 3.70014280e-01 5.46303630e-01 -6.94324374e-01 -7.06341088e-01 7.15151727e-01]
[11.511202812194824, -2.377856492996216]
08f6cb65-7bb1-4229-a53f-2fc0de4947ef
picture-word-interference-in-language
2303.09201
null
https://arxiv.org/abs/2303.09201v1
https://arxiv.org/pdf/2303.09201v1.pdf
Picture-word interference in language production studies: Exploring the roles of attention and processing times
The picture-word interference paradigm (participants name target pictures while ignoring distractor words) is often used to model the planning processes involved in word production. The participants' naming times are delayed in the presence of a distractor (general interference). The size of this effect depends on the relationship between the target and distractor words. Distractors of the same semantic category create more interference (semantic interference), distractors overlapping in phonology create less interference (phonological facilitation). The present study examines the relationships between these experimental effects, processing times, and attention in order to better understand the cognitive processes underlying participants' behavior in this paradigm. Participants named pictures with a superimposed line of Xs, semantically related distractors, phonologically related distractors, or unrelated distractors. General interference, semantic interference, and phonological facilitation effects are replicated. Distributional analyses reveal that general and semantic interference effects increase with naming times, while phonological facilitation decreases. The phonological facilitation and semantic interference effects are found to depend on the synchronicity in processing times between the planning of the picture's name and the processing of the distractor word. Finally, EEG power in the alpha band before stimulus onset varied with the position of the trial in the experiment and with repetition but did not predict the size of interference or facilitation effects. Taken together, these results suggest that experimental effects in the picture-word interference paradigm depend on processing times to both the target word and distractor word and that distributional patterns could party reflect this dependency.
['Sylvain Madec', 'Audrey Bürki']
2023-03-16
null
null
null
null
['eeg', 'eeg']
['methodology', 'time-series']
[ 1.75444201e-01 -5.49713314e-01 2.56124824e-01 5.80976077e-04 2.67979354e-01 -5.63324869e-01 8.24932218e-01 4.23859864e-01 -9.06567693e-01 9.27141979e-02 4.95826155e-01 -1.49183795e-01 -8.06780607e-02 -6.04810953e-01 -2.75869697e-01 -7.01397419e-01 -1.61165781e-02 1.33589908e-01 1.98524907e-01 1.15493797e-01 8.59400392e-01 5.99847913e-01 -1.39907503e+00 -1.73724201e-02 5.24812639e-01 5.49645126e-01 7.63325036e-01 2.03307673e-01 -8.37638751e-02 3.15852493e-01 -9.15609181e-01 1.71721727e-01 2.29218200e-01 -9.70878065e-01 -5.25058925e-01 -1.38146877e-01 1.02651529e-01 7.34857768e-02 -6.73024058e-02 9.95360613e-01 4.39260393e-01 6.54675782e-01 5.48274875e-01 -9.66681659e-01 -8.14242959e-01 6.16009116e-01 -5.47948003e-01 7.30167687e-01 6.21424019e-01 2.17434466e-01 6.59128606e-01 -8.15810978e-01 2.76739836e-01 1.38124239e+00 -2.06078544e-01 2.68395752e-01 -1.32840371e+00 -9.13679302e-01 2.99445897e-01 5.42573333e-01 -1.50840175e+00 -5.39492011e-01 2.46236503e-01 -4.97766376e-01 1.15967345e+00 5.79384677e-02 1.08224058e+00 1.09311056e+00 6.98619246e-01 -2.95624197e-01 1.22076738e+00 -5.64529121e-01 2.85911262e-01 5.30391298e-02 3.91807765e-01 -1.83150321e-01 5.54548144e-01 6.42648458e-01 -6.30644500e-01 1.04717895e-01 6.35375381e-01 -3.42888087e-01 -7.22013950e-01 3.19186032e-01 -1.24517357e+00 5.69112659e-01 2.63000578e-01 1.18460536e+00 -6.50986612e-01 -1.40336320e-01 3.07075292e-01 6.80256188e-02 -4.31844518e-02 6.40092909e-01 -5.85098527e-02 -8.34029168e-02 -6.11032486e-01 8.07014033e-02 2.12031454e-01 6.48711920e-02 1.01184025e-01 3.56063187e-01 -5.97972393e-01 8.45185459e-01 1.73463583e-01 6.72924757e-01 8.52555692e-01 -3.47741246e-01 -3.63109969e-02 -6.92968741e-02 -2.90297214e-02 -1.04891717e+00 -8.45209002e-01 -2.77110748e-02 -5.08354092e-03 4.31445122e-01 5.73652565e-01 3.21601242e-01 -4.79638159e-01 2.23557615e+00 -2.51768231e-01 -2.49174133e-01 -3.09525132e-01 9.99657929e-01 1.50463358e-01 7.41800725e-01 7.49691427e-01 -1.08186448e+00 1.66089892e+00 -5.24051428e-01 -9.00403619e-01 -7.62442470e-01 3.70878369e-01 -7.27473974e-01 1.21811545e+00 -5.87988459e-02 -9.47186947e-01 -9.98926878e-01 -8.18355918e-01 1.85967699e-01 -1.34479135e-01 -7.24295378e-01 1.12366788e-01 3.59172732e-01 -9.24826086e-01 5.10872900e-01 -4.29340035e-01 -4.59613562e-01 -1.53233618e-01 4.83824462e-02 -4.88145113e-01 -1.35610595e-01 -9.99213219e-01 1.73436308e+00 7.81387568e-01 -8.17729235e-02 -3.25355619e-01 -3.69933426e-01 -4.95004833e-01 4.29082572e-01 3.18915620e-02 -7.49810815e-01 9.29095387e-01 -1.73947918e+00 -1.02295446e+00 8.23575735e-01 -4.32637632e-01 3.48193496e-01 -6.37493849e-01 3.19285057e-02 -8.72909665e-01 2.94961095e-01 4.87656564e-01 4.77501839e-01 8.92359912e-01 -4.19858545e-01 -6.10043369e-02 -6.87678754e-01 -6.29911363e-01 6.65859461e-01 5.05573392e-01 4.70522672e-01 1.85016915e-01 -6.73517704e-01 1.60009339e-01 -6.56416953e-01 3.22460800e-01 -3.61830175e-01 2.27538675e-01 -5.48457980e-01 7.96345770e-02 -2.29044735e-01 8.74207675e-01 -2.61895466e+00 -4.23847027e-02 2.15665132e-01 -4.18354906e-02 -1.05680460e-02 -7.31116354e-01 5.15061021e-01 -8.25216174e-01 6.53213933e-02 5.69107831e-02 7.62203872e-01 1.54558986e-01 -1.41645059e-01 1.49564445e-01 7.24616230e-01 -1.07506514e-01 8.30434799e-01 -6.77276433e-01 9.86130759e-02 -5.13596274e-02 3.31054240e-01 -2.09323928e-01 -8.24125633e-02 -8.33699666e-03 3.48263979e-01 4.22516584e-01 -1.41669914e-01 6.77801192e-01 8.35145339e-02 1.74499914e-01 -5.88592365e-02 -6.66987836e-01 9.76854444e-01 -5.94171405e-01 1.00009644e+00 -1.04508780e-01 9.72376049e-01 -2.67390966e-01 -6.56145036e-01 6.11381769e-01 4.45718050e-01 -3.31763178e-01 -1.72992170e+00 5.89454114e-01 2.12171152e-01 1.44619215e+00 -3.58301073e-01 -2.91375369e-02 -7.83936858e-01 1.01782188e-01 7.28624940e-01 -2.60875434e-01 9.59758088e-02 5.75113654e-01 -8.48863721e-02 6.61831141e-01 -2.50883192e-01 6.31541133e-01 -5.41742504e-01 1.78594634e-01 -5.39001405e-01 4.03900832e-01 5.61747432e-01 -4.34415042e-02 2.51125544e-01 4.46065009e-01 9.54118967e-02 -5.47392190e-01 -1.21449697e+00 -2.19760403e-01 1.12913275e+00 2.04551369e-01 1.48124322e-01 -4.21842754e-01 -2.89699379e-02 -3.38857204e-01 1.63793683e+00 -6.65684044e-01 -6.26962543e-01 -5.04359722e-01 -3.65527481e-01 -1.38645574e-01 5.64492524e-01 1.21949725e-01 -1.77192390e+00 -7.20062912e-01 2.11652607e-01 -5.52323237e-02 -1.06391120e+00 -8.33251059e-01 -9.82006490e-02 -7.13812709e-01 -1.03513849e+00 -8.34731013e-02 -4.76101756e-01 6.46124601e-01 4.20594454e-01 8.12214375e-01 8.34573582e-02 -2.76193798e-01 6.08005285e-01 -8.40128958e-02 -3.10675770e-01 -1.55303150e-01 -5.32395005e-01 -6.44017607e-02 -3.95242214e-01 3.21769863e-01 -6.80854738e-01 -4.45029497e-01 -6.35645017e-02 -9.67889369e-01 -4.57435958e-02 6.27479792e-01 1.47222966e-01 2.74636000e-01 -2.78710276e-01 4.48595583e-01 5.87009266e-02 8.75007808e-01 -5.95181942e-01 -3.72994184e-01 -2.88514137e-01 -4.13435131e-01 -2.92899553e-02 4.99383092e-01 -7.85088480e-01 -7.00142860e-01 -1.09901726e+00 9.99320745e-02 3.26568000e-02 -4.90551889e-01 4.87558037e-01 -2.45798513e-01 3.10483396e-01 6.46624148e-01 3.66191715e-01 -2.63817292e-02 7.60615319e-02 -2.08185032e-01 -9.49354917e-02 4.53049466e-02 -1.86677694e-01 3.00197810e-01 -1.46940544e-01 -6.61714608e-03 -7.70735562e-01 -5.21564066e-01 1.74562290e-01 -4.84090984e-01 -9.98042598e-02 1.06407988e+00 -8.29155862e-01 -5.75376689e-01 2.89424270e-01 -9.01709974e-01 -1.94151729e-01 -1.82144716e-01 9.94934916e-01 -2.57895708e-01 1.88424349e-01 -3.04854810e-01 -5.15842080e-01 -3.03488195e-01 -9.62485909e-01 1.00607663e-01 1.05319619e-01 -7.42819011e-01 -6.75191879e-01 -1.00750491e-01 -2.69210905e-01 3.05694669e-01 -1.29817367e-01 1.34780192e+00 -1.12854290e+00 -3.25387776e-01 1.27188733e-03 -2.59063721e-01 -7.83128291e-02 5.23178577e-01 -2.93058276e-01 -2.62724400e-01 2.21262172e-01 8.39727819e-01 3.53795946e-01 3.31798106e-01 5.50453067e-01 3.42429280e-01 -2.33248636e-01 -2.25146785e-01 -6.24389015e-02 1.39312100e+00 6.47829115e-01 6.33042037e-01 1.95254050e-02 4.83155578e-01 5.62003136e-01 4.72702831e-01 -1.41053379e-01 -1.66490719e-01 5.10034144e-01 -1.79182723e-01 6.40514493e-01 -1.52274426e-02 1.11633047e-01 5.00055194e-01 5.96280515e-01 4.87589329e-01 -2.93426931e-01 -9.55808103e-01 4.65224653e-01 -7.34915078e-01 -1.26507413e+00 -6.81217909e-01 2.68339252e+00 3.33110631e-01 7.21599907e-02 -2.64565706e-01 2.34258667e-01 6.75078630e-01 2.52807409e-01 -2.94938475e-01 -6.17553592e-01 -4.84671704e-02 4.52884346e-01 -3.43555026e-02 4.78855401e-01 1.09655611e-01 8.30310822e-01 6.10348749e+00 5.71807683e-01 -1.16190267e+00 1.33802027e-01 9.04882699e-02 -6.06693327e-01 -1.65970251e-01 -9.72914323e-02 -2.29011998e-01 5.13773203e-01 1.36149788e+00 -7.26557016e-01 4.80062962e-01 1.85071424e-01 8.51105690e-01 -1.23209369e+00 -1.20019102e+00 7.48445392e-01 1.94998875e-01 -3.44930381e-01 4.87827994e-02 6.15940019e-02 -2.44782805e-01 -2.88511187e-01 2.00145487e-02 1.28753915e-01 -6.65929437e-01 -9.05205131e-01 1.04366124e+00 3.79532993e-01 5.98361492e-01 -5.07698596e-01 2.75681078e-01 3.77443105e-01 -8.54234815e-01 1.79583639e-01 -3.23387951e-01 -8.62611294e-01 4.23553407e-01 2.03810453e-01 -3.89097154e-01 -3.07716668e-01 1.19639367e-01 -1.45608351e-01 -2.50613421e-01 8.06172550e-01 -5.88847816e-01 4.80421185e-01 -9.42154601e-02 -6.98525682e-02 -9.16300938e-02 -1.74066931e-01 4.92615968e-01 5.66127896e-01 4.27900732e-01 5.47331154e-01 -5.19498050e-01 1.17901254e+00 3.82758290e-01 5.83463252e-01 -4.01977390e-01 -2.86888510e-01 8.28245521e-01 7.73535728e-01 -1.14686549e+00 -1.53122559e-01 -5.67281365e-01 1.13748670e+00 1.56242326e-01 5.49386024e-01 -6.72548532e-01 5.68452738e-02 4.08416897e-01 3.65420401e-01 3.38865757e-01 -1.57273158e-01 -1.54242158e-01 -4.54843313e-01 1.04602240e-01 -6.42971754e-01 -2.44308263e-01 -1.02781177e+00 -9.95039999e-01 4.30921048e-01 4.42029089e-01 -3.55792493e-01 -3.55017781e-02 -2.86496758e-01 -4.69211668e-01 1.68225574e+00 -8.10389161e-01 -2.23861441e-01 1.50340334e-01 4.88411009e-01 6.32368147e-01 4.58155394e-01 1.06730831e+00 -3.00338417e-01 -3.10787171e-01 -5.23363985e-02 -4.48900372e-01 -3.70089680e-01 8.72119069e-01 -5.23952484e-01 -1.14922211e-01 4.67663050e-01 2.00212263e-02 8.03226054e-01 4.19766784e-01 -8.30798268e-01 -5.24971068e-01 -1.29777864e-01 1.39125752e+00 8.99236053e-02 4.10292268e-01 -2.78352797e-01 -9.85830426e-01 3.48190337e-01 6.36518776e-01 -2.61107892e-01 9.90608990e-01 -6.23058677e-02 -5.13051271e-01 2.13947341e-01 -7.03666091e-01 8.57763648e-01 1.06489515e+00 -6.25869215e-01 -9.87431467e-01 2.10667402e-02 4.72143233e-01 3.13468486e-01 -2.43108302e-01 -2.59088408e-02 4.53998953e-01 -1.08932972e+00 5.16548336e-01 -4.39128540e-02 -2.72332672e-02 5.32339141e-03 2.26701260e-01 -1.48898315e+00 -1.31989300e+00 3.33528101e-01 4.93543833e-01 9.01969671e-01 9.51872468e-02 -7.21372545e-01 -5.21525264e-01 4.29880381e-01 4.34041619e-02 2.73148596e-01 -6.82618260e-01 -7.35242486e-01 2.25745723e-01 -3.03991139e-01 1.23196900e-01 5.92338026e-01 3.38608861e-01 9.17791188e-01 5.15708804e-01 1.41170606e-01 -2.50641294e-02 -4.57311869e-02 -5.54337859e-01 -1.23004460e+00 -2.15523079e-01 -1.06680548e+00 8.88468847e-02 -2.79771745e-01 3.97680521e-01 -6.15689218e-01 -2.97495052e-02 -1.27331650e+00 2.77442783e-01 1.67753085e-01 -1.11037672e-01 2.62161613e-01 -6.80317760e-01 -3.23640466e-01 6.50269449e-01 2.38565326e-01 2.23419547e-01 5.25858402e-01 9.27037001e-01 4.14554685e-01 -6.15113556e-01 -2.53255308e-01 -8.59896660e-01 3.83491248e-01 9.58413124e-01 -5.60022354e-01 -5.38140535e-01 -3.69695157e-01 8.26306641e-02 2.77255237e-01 2.06481189e-01 -1.04696143e+00 -2.20312059e-01 1.12968162e-02 5.74322820e-01 2.77399231e-04 1.59675986e-01 -8.03357244e-01 3.91763568e-01 5.55094242e-01 -1.36044979e-01 5.92292011e-01 9.10806417e-01 5.21773519e-03 1.38676465e-01 -8.87879550e-01 1.11549163e+00 -8.27195570e-02 -3.48770440e-01 -4.56981540e-01 -1.06770563e+00 1.27092376e-01 1.02520263e+00 -3.52919579e-01 -1.16604283e-01 -1.51768044e-01 -9.75837648e-01 -5.93062460e-01 2.31887788e-01 7.45348811e-01 4.41853583e-01 -1.21057248e+00 -5.91952920e-01 6.25050813e-02 -4.37657714e-01 -9.13405955e-01 4.31423306e-01 1.26372719e+00 -1.21242344e-01 5.64927399e-01 -4.59120572e-01 8.49789008e-02 -1.20304227e+00 1.30189097e+00 4.48276326e-02 2.28664905e-01 -2.63711751e-01 8.85488331e-01 8.24575424e-01 7.32880354e-01 -1.31137252e-01 1.04025997e-01 -2.88888514e-01 5.94793618e-01 1.02540767e+00 3.63997936e-01 -2.43742466e-01 -7.52169490e-01 -3.71282786e-01 3.38605583e-01 1.04137231e-02 -5.20752490e-01 6.94567621e-01 -1.44376531e-01 -5.19640505e-01 1.35250807e+00 8.18546653e-01 2.64202476e-01 -4.61424351e-01 1.19173214e-01 -2.40819067e-01 -4.38670635e-01 -1.42884701e-01 -1.04835665e+00 -7.01224744e-01 6.34197235e-01 6.44659817e-01 -9.45673808e-02 1.14284658e+00 2.30808035e-01 -1.05196819e-01 -5.34163535e-01 3.29234675e-02 -1.02893651e+00 1.90892875e-01 4.44660276e-01 1.18333983e+00 -4.21745151e-01 -1.94667816e-01 -3.47817659e-01 -5.51368833e-01 6.99748218e-01 6.66127563e-01 -1.72949389e-01 3.12138289e-01 -1.35024279e-01 -4.46023524e-01 -2.67760456e-01 -8.01257849e-01 -4.86769080e-01 2.60236353e-01 5.70251763e-01 7.27458835e-01 -1.45764798e-01 -1.25684226e+00 5.97518623e-01 -3.21654290e-01 -4.08913225e-01 2.49281585e-01 7.49103069e-01 -5.42927802e-01 -8.87977362e-01 -8.66812170e-01 2.23019034e-01 -2.98970938e-01 -6.96455240e-01 -3.57544601e-01 1.04614496e+00 3.80821884e-01 7.04746544e-01 1.04703283e+00 1.21337749e-01 4.92645353e-01 5.94614744e-01 8.58518600e-01 -8.63277972e-01 -7.44293630e-01 4.16928411e-01 -1.06198467e-01 -2.86405653e-01 -2.22160801e-01 -1.13949716e+00 -1.49898815e+00 -1.23227052e-01 -3.77863705e-01 1.79021239e-01 4.03672069e-01 1.20966005e+00 1.72379985e-01 4.68866140e-01 -4.28851647e-03 -3.72108608e-01 2.18068272e-01 -1.37236726e+00 -7.38617718e-01 2.14241505e-01 -2.95178652e-01 -5.47736049e-01 -3.98602664e-01 -2.26564094e-01]
[10.290511131286621, 2.286548376083374]
e46ee0f2-65aa-4bee-a05b-d95bbcb43938
3d-cartoon-face-generation-with-controllable
2207.14425
null
https://arxiv.org/abs/2207.14425v1
https://arxiv.org/pdf/2207.14425v1.pdf
3D Cartoon Face Generation with Controllable Expressions from a Single GAN Image
In this paper, we investigate an open research task of generating 3D cartoon face shapes from single 2D GAN generated human faces and without 3D supervision, where we can also manipulate the facial expressions of the 3D shapes. To this end, we discover the semantic meanings of StyleGAN latent space, such that we are able to produce face images of various expressions, poses, and lighting by controlling the latent codes. Specifically, we first finetune the pretrained StyleGAN face model on the cartoon datasets. By feeding the same latent codes to face and cartoon generation models, we aim to realize the translation from 2D human face images to cartoon styled avatars. We then discover semantic directions of the GAN latent space, in an attempt to change the facial expressions while preserving the original identity. As we do not have any 3D annotations for cartoon faces, we manipulate the latent codes to generate images with different poses and lighting, such that we can reconstruct the 3D cartoon face shapes. We validate the efficacy of our method on three cartoon datasets qualitatively and quantitatively.
['Chunyan Miao', 'Steven C. H. Hoi', 'Guosheng Lin', 'Hao Wang']
2022-07-29
null
null
null
null
['face-model']
['computer-vision']
[ 2.18107626e-01 4.80863124e-01 2.63682783e-01 -3.69584531e-01 -9.70856845e-02 -1.04528451e+00 6.70813203e-01 -1.07635057e+00 2.44771898e-01 4.93805081e-01 2.70428240e-01 7.32524320e-02 5.43174446e-01 -8.42996478e-01 -8.01968694e-01 -6.71600878e-01 4.82703239e-01 4.96687412e-01 -5.37460864e-01 -1.70682460e-01 -1.23703234e-01 7.95562744e-01 -1.33073592e+00 2.20976412e-01 3.38212579e-01 7.25849032e-01 -3.46198618e-01 4.70266014e-01 2.20850576e-02 4.20878977e-01 -4.85678077e-01 -7.23813117e-01 6.49690151e-01 -8.20175946e-01 -4.39262986e-01 6.94084704e-01 5.34104466e-01 -3.06714982e-01 -2.70114511e-01 1.02898526e+00 1.50047883e-01 -3.49072099e-01 8.64230692e-01 -1.52125525e+00 -1.09473896e+00 3.29054952e-01 -7.23166823e-01 -7.88083613e-01 4.85794008e-01 2.15262488e-01 7.20506489e-01 -8.23070049e-01 9.68152702e-01 1.89499712e+00 3.37613910e-01 1.06395614e+00 -1.49744928e+00 -1.03103292e+00 2.08942220e-01 -3.34645510e-01 -1.16762280e+00 -6.00810707e-01 1.07701159e+00 -5.28129280e-01 -1.89863835e-02 1.63049981e-01 7.74147630e-01 1.47094595e+00 1.02649182e-02 3.97484452e-01 1.17368770e+00 -3.96468729e-01 9.85360593e-02 2.58147955e-01 -8.18601072e-01 9.18464601e-01 -8.44238922e-02 -7.88861886e-02 -4.44174320e-01 -6.19371124e-02 1.14154410e+00 1.65643364e-01 -1.72048494e-01 -6.73968077e-01 -1.18315482e+00 8.78082812e-01 1.88657776e-01 5.00001982e-02 -2.44359896e-01 1.04735732e-01 -1.02220580e-01 3.49554718e-01 4.32633638e-01 4.16005909e-01 -2.74819642e-01 2.81455606e-01 -4.31585789e-01 1.64603129e-01 6.32200003e-01 1.31050146e+00 7.98440516e-01 2.54400760e-01 -1.85920209e-01 7.19915390e-01 4.02603686e-01 9.01121020e-01 1.22762315e-01 -1.14717269e+00 3.35698053e-02 5.55103779e-01 6.64138943e-02 -1.04380763e+00 1.65518910e-01 1.46842986e-01 -8.53534222e-01 3.06711316e-01 3.00191522e-01 -3.14973861e-01 -9.97137904e-01 1.97494936e+00 2.78398603e-01 -9.30616260e-02 9.78098288e-02 8.93725276e-01 4.16604549e-01 5.61936319e-01 -1.99647680e-01 7.73783866e-03 1.45531654e+00 -7.04559267e-01 -7.54571497e-01 -1.70174792e-01 1.54916257e-01 -8.52334976e-01 1.27536941e+00 -3.58067043e-02 -1.16063082e+00 -5.06726444e-01 -8.61265779e-01 -1.36080729e-02 4.09559309e-02 2.83208817e-01 2.76979089e-01 7.35858917e-01 -1.04726291e+00 2.13414147e-01 -6.93218887e-01 -2.55350947e-01 5.26644289e-01 1.09223507e-01 -6.17545724e-01 -1.21313609e-01 -8.48646522e-01 5.60142338e-01 -8.54421332e-02 1.20750599e-01 -9.97341931e-01 -5.11347830e-01 -8.99085879e-01 -1.17968798e-01 1.28830612e-01 -7.08489597e-01 9.21612501e-01 -1.39118207e+00 -1.90451479e+00 1.22894871e+00 -1.35689661e-01 1.63677782e-01 7.47977376e-01 2.65516371e-01 -1.63256899e-01 -5.96711114e-02 1.42610982e-01 1.11656749e+00 1.32471514e+00 -1.43577754e+00 -1.66275442e-01 -5.26389539e-01 1.00468144e-01 1.06240682e-01 -2.83062994e-01 -1.51122198e-01 -6.39406741e-01 -8.76747012e-01 1.21562421e-01 -1.46468532e+00 8.04119930e-02 3.90232176e-01 -5.89238405e-01 3.88069421e-01 1.18912518e+00 -4.59026635e-01 2.22220838e-01 -2.33532596e+00 6.51290536e-01 2.97986209e-01 1.64873332e-01 -2.59382039e-01 -4.33983713e-01 1.27399325e-01 -5.91161139e-02 3.00222754e-01 -2.50645578e-01 -5.66430628e-01 1.42564818e-01 3.39972347e-01 -4.57730561e-01 3.04739356e-01 4.11455303e-01 1.02555180e+00 -7.14828134e-01 -2.40125209e-01 -9.49684605e-02 7.70472527e-01 -8.48775744e-01 3.89713019e-01 -4.26183313e-01 1.09014416e+00 -5.22832990e-01 4.65372950e-01 8.13635945e-01 -3.53340246e-02 5.60645044e-01 -1.98658243e-01 2.05599770e-01 -2.18756601e-01 -7.96683848e-01 1.59664273e+00 -5.76885939e-01 6.87518477e-01 1.61655441e-01 -4.63438749e-01 1.19393468e+00 3.20601732e-01 3.76298755e-01 -4.81481642e-01 2.22001389e-01 4.69634719e-02 -2.64965743e-01 -2.25529507e-01 5.70452176e-02 -6.26027226e-01 -2.04412788e-01 6.67787135e-01 1.67187806e-02 -4.76095438e-01 -2.15174332e-01 -4.23753820e-02 5.99478602e-01 2.54409373e-01 -2.66978055e-01 -2.36413796e-02 4.04363304e-01 -5.28302193e-01 3.21430624e-01 8.05790201e-02 4.24701989e-01 7.37436116e-01 1.04471183e+00 -5.18487155e-01 -1.32289004e+00 -1.06109953e+00 2.28836849e-01 7.59556353e-01 -1.01935223e-01 -5.70049286e-02 -9.44210291e-01 -7.90914059e-01 -9.87485126e-02 5.15422881e-01 -9.32079375e-01 -1.34331822e-01 -6.32171214e-01 -2.67102182e-01 5.77975810e-01 2.54651338e-01 4.62298632e-01 -1.07675064e+00 -4.22412425e-01 -2.18250230e-01 -1.16649747e-01 -1.40576947e+00 -8.66617501e-01 -4.46578234e-01 -5.53100765e-01 -1.12061822e+00 -8.10014784e-01 -9.17495906e-01 1.21794820e+00 -7.35915676e-02 8.68582904e-01 -2.00966615e-02 9.65978652e-02 3.70742172e-01 -3.35351050e-01 -2.09690005e-01 -7.05943823e-01 -2.57329404e-01 7.09608197e-02 5.71864665e-01 -1.10636979e-01 -8.76191735e-01 -4.08691257e-01 4.58899468e-01 -1.00287342e+00 5.83441079e-01 3.36572200e-01 6.58020556e-01 4.21947181e-01 -2.58933008e-01 2.02712774e-01 -1.21541369e+00 3.42737228e-01 -3.13704833e-02 -8.27941775e-01 1.42395884e-01 -1.65674478e-01 2.21893921e-01 6.78420067e-01 -5.63623726e-01 -9.91013885e-01 1.87748492e-01 -1.70936137e-01 -8.93483818e-01 -2.51403928e-01 -3.48457932e-01 -6.40504658e-01 -1.99933834e-02 4.35915411e-01 6.21716417e-02 1.78738102e-01 -4.25176620e-01 8.65021229e-01 4.42404568e-01 4.14414912e-01 -6.97811246e-01 1.21782041e+00 7.59832025e-01 8.57704598e-03 -5.52684486e-01 -5.59665203e-01 5.68032682e-01 -9.67863321e-01 -1.11330301e-01 1.25726128e+00 -7.67472327e-01 -6.94571733e-01 5.71916044e-01 -1.23251331e+00 -3.87650490e-01 -2.93172657e-01 -2.33664930e-01 -7.66079962e-01 -4.14303653e-02 -3.91807050e-01 -4.58740145e-01 -9.29504186e-02 -1.34860814e+00 1.52402592e+00 -2.05093268e-02 -1.95603102e-01 -6.85256362e-01 -1.33376420e-01 9.08342004e-02 9.23724994e-02 6.79220200e-01 1.24290538e+00 8.96864980e-02 -5.92407525e-01 7.31349811e-02 -1.71498850e-01 1.75191358e-01 5.47879696e-01 -2.40363553e-02 -9.23249125e-01 -2.88914055e-01 -2.62913257e-02 -2.08037630e-01 2.82176256e-01 -4.67428528e-02 1.02504098e+00 -7.02554345e-01 -1.47833362e-01 9.58633780e-01 8.62265408e-01 4.17330742e-01 5.56826234e-01 -2.17075303e-01 9.13230360e-01 8.02006066e-01 6.91958591e-02 2.56714493e-01 1.39234379e-01 7.13965058e-01 2.47235969e-01 -1.20544262e-01 -1.41410157e-01 -9.14874971e-01 3.63071948e-01 4.00463521e-01 -7.95360003e-03 -1.57365084e-01 -5.66562235e-01 1.93316057e-01 -1.34903216e+00 -6.09968305e-01 5.30154586e-01 1.75537241e+00 7.17708349e-01 -1.59826174e-01 -1.37888929e-02 -1.58619285e-01 8.26265574e-01 2.23716259e-01 -6.32608712e-01 -2.49681339e-01 2.90002208e-02 1.43522486e-01 1.23795711e-01 3.42578858e-01 -6.73146367e-01 1.10605240e+00 5.87019730e+00 1.54862344e-01 -1.47417080e+00 4.62503061e-02 7.57069826e-01 -2.44958788e-01 -7.89436758e-01 9.75115076e-02 -4.77225482e-01 4.67810363e-01 3.00465673e-01 9.41016078e-02 7.91528940e-01 5.77182651e-01 3.58719602e-02 7.27472007e-01 -1.35947156e+00 1.03265417e+00 7.11385459e-02 -1.25411689e+00 4.96107221e-01 2.96842694e-01 1.05713630e+00 -7.28433132e-01 2.42681727e-01 3.70464101e-02 4.48200256e-01 -1.02109504e+00 1.05889082e+00 5.05941272e-01 1.42951858e+00 -6.12424552e-01 1.82711184e-01 1.26984224e-01 -8.26784909e-01 9.74927470e-02 -1.17271699e-01 2.18239203e-01 2.24068254e-01 1.61824226e-01 -6.32649124e-01 9.35183093e-02 4.79451001e-01 6.59313262e-01 -3.15528721e-01 2.15558428e-02 -5.22060871e-01 1.58054978e-01 -8.82578641e-02 2.22863063e-01 -2.80291643e-02 -5.31983793e-01 4.83453661e-01 6.41892314e-01 6.43824279e-01 2.58225322e-01 -5.03780879e-02 1.29364967e+00 -7.34678984e-01 -2.16839164e-01 -1.01766491e+00 -1.88071206e-01 4.20022488e-01 1.10237706e+00 -6.10413432e-01 4.94954817e-04 -1.82369247e-01 1.35063064e+00 1.65883098e-02 5.29677689e-01 -8.54682028e-01 1.50728412e-02 8.68782103e-01 1.96193114e-01 1.90504745e-01 -3.08126688e-01 -2.85611935e-02 -1.36621404e+00 -9.59539190e-02 -1.10527980e+00 -2.82055259e-01 -1.04212391e+00 -1.18821681e+00 8.37481141e-01 -1.80019677e-01 -1.00070107e+00 -2.65189409e-01 -4.96401548e-01 -5.39088130e-01 6.84680641e-01 -9.97981489e-01 -1.44468868e+00 -4.26361531e-01 6.55823588e-01 2.90635377e-01 -2.76989877e-01 8.14289868e-01 7.53328800e-02 -2.77239501e-01 6.01894975e-01 -2.94521302e-01 4.11768198e-01 6.84352577e-01 -8.16896677e-01 7.65790641e-01 4.52537268e-01 2.79665202e-01 4.75587249e-01 5.05661666e-01 -5.32046318e-01 -1.62754655e+00 -1.14521706e+00 3.53697836e-01 -5.27560174e-01 3.02155554e-01 -9.09614861e-01 -5.46578884e-01 1.00915766e+00 2.07343712e-01 1.13340449e-02 4.47241038e-01 -2.02509210e-01 -5.76133370e-01 -3.87244560e-02 -1.27575409e+00 8.91158700e-01 1.37865412e+00 -6.05970979e-01 -1.13856964e-01 2.90299475e-01 7.78415978e-01 -3.82905751e-01 -5.99709332e-01 2.55325735e-01 8.32046211e-01 -7.16758072e-01 8.43763709e-01 -6.97846711e-01 5.58392346e-01 -3.03285748e-01 -2.47753099e-01 -1.36349118e+00 -6.71997592e-02 -5.58202624e-01 3.14369142e-01 1.21871173e+00 3.82981479e-01 -6.32845521e-01 8.48052919e-01 5.51460862e-01 2.74559557e-01 -3.79421890e-01 -7.23330498e-01 -3.20090860e-01 1.79387435e-01 9.38357338e-02 1.19446564e+00 9.77208972e-01 -5.43376088e-01 2.91078717e-01 -7.41883993e-01 3.18917483e-02 4.66274649e-01 4.65187937e-01 1.11191118e+00 -1.03473723e+00 -1.00957006e-01 -2.14950234e-01 -2.70113319e-01 -8.96333277e-01 6.42565906e-01 -8.64156723e-01 -1.25327498e-01 -8.28093886e-01 4.26814407e-02 -5.28857470e-01 2.39318743e-01 7.35852182e-01 2.65685707e-01 7.60042250e-01 5.08013010e-01 1.32646158e-01 2.10936517e-02 8.70788991e-01 1.82265019e+00 -2.46619284e-01 -4.13876176e-02 -2.08840102e-01 -9.66044903e-01 6.68785334e-01 5.74388325e-01 -5.10143936e-01 -5.37112534e-01 -7.35342503e-01 1.30119428e-01 3.12969118e-01 5.54855645e-01 -4.53134179e-01 -3.52161825e-01 -2.64377803e-01 6.71627939e-01 2.65343636e-02 4.58393961e-01 -9.71303761e-01 5.94223917e-01 2.59552389e-01 -4.04316694e-01 8.71612504e-02 4.65017855e-02 3.15068245e-01 1.26964469e-02 2.06933096e-01 1.05031228e+00 -2.28793874e-01 -6.88896030e-02 5.71453869e-01 -1.12807117e-01 -7.43281767e-02 1.00486147e+00 -6.26026792e-03 7.32400492e-02 -6.45420074e-01 -8.26330364e-01 -2.78085656e-03 1.10229290e+00 7.73004532e-01 5.78798294e-01 -1.65144217e+00 -6.72394335e-01 9.60510254e-01 1.55919269e-02 -6.90589175e-02 7.47708306e-02 2.48010471e-01 -5.65885186e-01 2.25062624e-01 -7.52994835e-01 -5.47476590e-01 -1.10547948e+00 6.75742507e-01 4.21895891e-01 3.26807439e-01 -5.46181262e-01 4.24290478e-01 6.59782767e-01 -8.49843144e-01 -6.24426901e-02 -1.33314878e-02 -3.20897661e-02 7.12257857e-03 1.46325558e-01 -2.34254196e-01 -3.00701976e-01 -7.05170870e-01 -7.88083076e-02 9.80199516e-01 2.33142704e-01 -2.83235282e-01 1.33599448e+00 3.05327680e-02 -1.88126802e-01 8.19550082e-02 1.28794003e+00 4.44751143e-01 -1.53192961e+00 1.02808505e-01 -5.66889048e-01 -7.89765179e-01 -5.24239600e-01 -4.29114163e-01 -1.68531752e+00 8.78217876e-01 5.22371173e-01 -1.49285212e-01 1.21485198e+00 5.12647405e-02 7.56234646e-01 7.71628544e-02 4.92497861e-01 -3.88873786e-01 2.49450713e-01 2.76352555e-01 1.10167277e+00 -8.55380297e-01 -4.77833658e-01 -2.80490696e-01 -7.71456897e-01 9.50716972e-01 5.45419276e-01 -1.02457501e-01 5.39752126e-01 2.25740343e-01 2.06775934e-01 -2.21042365e-01 -5.70472836e-01 3.58764619e-01 2.19867244e-01 5.65842986e-01 2.62338519e-01 1.91777691e-01 2.09361032e-01 2.96929866e-01 -6.41214967e-01 -1.19577281e-01 5.12004495e-01 4.62449104e-01 2.23338693e-01 -1.36849582e+00 -4.11495507e-01 -5.99503033e-02 -2.18494400e-01 2.33667508e-01 -7.51366973e-01 6.75114036e-01 2.31981322e-01 5.21788895e-01 1.35008529e-01 -4.65051740e-01 5.06932974e-01 2.53930211e-01 7.17129350e-01 -5.93179762e-01 5.78345060e-02 1.57076299e-01 -2.20358655e-01 -3.51261795e-01 -2.42496759e-01 -5.01397312e-01 -9.43310082e-01 -4.33014810e-01 2.05489352e-01 9.02162567e-02 4.92037237e-01 6.19512856e-01 4.34344023e-01 3.59076530e-01 9.34658051e-01 -1.12530065e+00 -9.20736231e-03 -6.70926630e-01 -5.24111509e-01 6.22540534e-01 3.60325873e-01 -5.81190288e-01 -2.35676035e-01 4.18546587e-01]
[12.568704605102539, -0.2928411066532135]
01524ae4-7456-4065-b20a-f3df08591c0a
computational-tradeoff-in-minimum-obstacle
2302.07114
null
https://arxiv.org/abs/2302.07114v1
https://arxiv.org/pdf/2302.07114v1.pdf
Computational Tradeoff in Minimum Obstacle Displacement Planning for Robot Navigation
In this paper, we look into the minimum obstacle displacement (MOD) planning problem from a mobile robot motion planning perspective. This problem finds an optimal path to goal by displacing movable obstacles when no path exists due to collision with obstacles. However this problem is computationally expensive and grows exponentially in the size of number of movable obstacles. This work looks into approximate solutions that are computationally less intensive and differ from the optimal solution by a factor of the optimal cost.
['Michela Robba', 'Fulvio Mastrogiovanni', 'Giulio Ferro', 'Antony Thomas']
2023-02-14
null
null
null
null
['robot-navigation', 'motion-planning']
['robots', 'robots']
[ 2.51867384e-01 4.86672699e-01 -4.39849764e-01 1.73642144e-01 -4.87319291e-01 -7.84537435e-01 2.87566453e-01 2.15983853e-01 -9.17687535e-01 1.03587198e+00 -3.38088810e-01 -8.42291832e-01 -3.18800807e-01 -1.15537727e+00 -4.37048972e-01 -4.33970094e-01 -6.83042824e-01 8.49543393e-01 9.55098152e-01 -8.71499419e-01 8.02289426e-01 6.40540004e-01 -8.83888066e-01 -2.55033284e-01 7.70621300e-01 1.32078931e-01 6.71926439e-01 1.05523694e+00 -1.22715950e-01 2.77838290e-01 -1.40048087e-01 3.37582439e-01 5.02212405e-01 -2.12583080e-01 -1.41889274e+00 -1.14696138e-01 -2.96921700e-01 -8.13388675e-02 -6.95226640e-02 9.57345665e-01 2.05552891e-01 5.65744102e-01 5.91428339e-01 -1.65487027e+00 3.16640735e-02 1.94330782e-01 -2.93062449e-01 4.19267446e-01 5.20335555e-01 -1.84077486e-01 4.45333332e-01 -7.31031060e-01 9.84644473e-01 1.15220833e+00 6.90237761e-01 3.56693119e-01 -5.99650204e-01 2.72871345e-01 1.34790719e-01 4.25075382e-01 -1.52888322e+00 -1.91110432e-01 1.88810676e-01 1.16203412e-01 1.61698914e+00 5.19239306e-01 3.48242790e-01 1.20289527e-01 8.36856127e-01 1.14511680e-02 2.68857956e-01 -4.34847802e-01 4.05695856e-01 -2.39863247e-01 -2.11095084e-02 6.84279680e-01 7.56654620e-01 -1.82954833e-01 4.42625642e-01 -3.57376151e-02 5.49866199e-01 -2.86980510e-01 -1.13499314e-01 -3.24789375e-01 -9.06599641e-01 9.04931605e-01 4.70189154e-01 3.88078153e-01 -4.83936757e-01 4.01370019e-01 -2.39236757e-01 2.31456101e-01 -4.59780902e-01 5.88303268e-01 -3.28469396e-01 -3.80513012e-01 -4.07677114e-01 6.26906395e-01 7.23964632e-01 1.13127720e+00 9.86819625e-01 7.93363005e-02 5.50573885e-01 2.27549329e-01 3.28599930e-01 5.26181161e-01 7.32685775e-02 -1.39613831e+00 9.18478608e-01 3.93865705e-01 8.39611888e-01 -1.21718717e+00 -9.45706606e-01 3.90981019e-01 -2.78984427e-01 8.67468297e-01 5.58262348e-01 -3.00889313e-01 -8.02362323e-01 9.96855557e-01 4.74027514e-01 -4.01819974e-01 3.39788914e-01 5.67444742e-01 2.83736080e-01 8.04313183e-01 -1.69753432e-01 -3.44972819e-01 8.40172112e-01 -1.33463097e+00 -6.74540341e-01 -6.34937227e-01 1.07127571e+00 -7.61344731e-01 5.83969414e-01 2.77278330e-02 -1.23796117e+00 -6.44033253e-02 -1.34038246e+00 2.11737566e-02 -3.36193830e-01 -6.48402512e-01 1.86530903e-01 5.37087977e-01 -1.41097701e+00 7.22675383e-01 -8.44791412e-01 -5.33571303e-01 -1.63375869e-01 6.14011586e-01 -3.99689168e-01 -1.95799038e-01 -9.15285170e-01 1.36258936e+00 6.44664824e-01 -9.13699940e-02 -3.76492441e-01 2.98884809e-01 -8.94639850e-01 -3.33896697e-01 4.87157702e-01 -3.01409751e-01 1.01257765e+00 -4.73935485e-01 -1.40793622e+00 3.57644171e-01 -6.42550111e-01 -3.88091296e-01 5.30678093e-01 2.59380251e-01 -2.23269612e-02 -1.06212735e-01 3.13679844e-01 5.30909657e-01 1.40213668e-01 -1.15737462e+00 -9.26371932e-01 -6.65217042e-02 5.05694509e-01 5.44070482e-01 3.81754816e-01 -2.39529267e-01 -4.40654606e-01 3.66164483e-02 4.90208566e-01 -1.46048570e+00 -1.40395057e+00 -2.93414623e-01 -2.39631087e-01 -1.77789807e-01 5.72069108e-01 -2.53014594e-01 1.28838515e+00 -1.68205488e+00 1.67338759e-01 2.80042529e-01 -1.90235406e-01 -2.44276319e-02 -2.39005297e-01 7.98839748e-01 4.43939954e-01 2.85342872e-01 -3.45873892e-01 9.35990084e-03 -2.93024153e-01 6.87333763e-01 3.50035378e-03 5.05654514e-01 -2.75077820e-01 7.01063871e-01 -1.04813147e+00 -5.42215824e-01 2.30241399e-02 -7.01361597e-02 -5.47309458e-01 -4.54258293e-01 1.35644197e-01 -3.07040010e-02 -5.76717615e-01 7.35541642e-01 9.89442050e-01 2.82674491e-01 3.21135581e-01 7.30021238e-01 -1.08946002e+00 1.15017951e-01 -1.59288347e+00 1.33638132e+00 -1.53077304e-01 6.53227389e-01 1.67502180e-01 -1.02713358e+00 4.78032023e-01 6.27731532e-02 5.28617859e-01 -5.56517899e-01 9.22979936e-02 8.03324282e-01 -1.54293343e-01 -5.05925715e-01 1.16193128e+00 -6.86349720e-02 -1.94469050e-01 1.96727678e-01 -9.13205385e-01 2.61537191e-02 2.68761545e-01 -2.86060423e-01 1.53360569e+00 -1.84727684e-01 8.50598693e-01 -4.25224960e-01 5.67456007e-01 1.18518341e+00 3.65426242e-01 9.19805110e-01 -4.85365778e-01 1.81536287e-01 8.05988908e-02 -6.32540762e-01 -9.15549040e-01 -8.79022658e-01 1.32413119e-01 6.09148204e-01 1.22505832e+00 -1.98559776e-01 -6.22853100e-01 -7.69106075e-02 -2.67753303e-01 4.83285248e-01 -2.82029569e-01 4.25153941e-01 -1.44745064e+00 -6.82310522e-01 2.34769940e-01 3.30224723e-01 2.44322434e-01 -9.15922046e-01 -1.20160306e+00 6.70112073e-01 -3.67099106e-01 -1.00288939e+00 -2.12534800e-01 4.66369558e-03 -8.84720325e-01 -1.12392771e+00 -5.14374197e-01 -9.86065090e-01 1.02803791e+00 1.01372623e+00 4.41257477e-01 4.36457843e-01 -2.43575528e-01 1.82272837e-01 -4.48800474e-01 -4.51854110e-01 -3.31141293e-01 1.19867399e-02 9.95395035e-02 -9.74977553e-01 -2.06590608e-01 -2.63979077e-01 -3.88260633e-01 4.65387672e-01 -4.60194856e-01 -3.26640517e-01 3.40865284e-01 1.34518117e-01 7.81832457e-01 7.69796789e-01 1.41245887e-01 -2.70189852e-01 6.95319235e-01 -9.84188676e-01 -6.91842914e-01 -1.65919676e-01 -6.93304002e-01 3.34466957e-02 5.80400288e-01 -1.36968881e-01 -4.01589543e-01 2.43262723e-01 -1.81376383e-01 3.70494187e-01 1.24627814e-01 4.14141923e-01 -3.61079862e-03 -5.59777856e-01 4.75249797e-01 -4.32462692e-02 -2.92296618e-01 -1.38965726e-01 3.22531670e-01 2.99323499e-01 3.84853512e-01 1.31186813e-01 8.00841451e-01 7.45375037e-01 8.46547365e-01 -8.40358436e-01 4.43652689e-01 -4.37957674e-01 -6.60090446e-01 -3.05322647e-01 8.01826417e-01 -2.19198465e-01 -5.14707208e-01 8.30502138e-02 -1.38189638e+00 -4.55352247e-01 -7.40106553e-02 2.86913484e-01 -8.09133589e-01 5.59122503e-01 -7.72614852e-02 -8.96288872e-01 1.12008877e-01 -1.16029882e+00 3.42551857e-01 8.47114921e-02 -5.02359092e-01 -1.06724131e+00 2.60907590e-01 -3.23414505e-01 5.85421920e-01 7.31944561e-01 5.81570268e-01 -6.10041618e-02 -8.54908705e-01 -4.49226946e-01 1.80323705e-01 -9.50471103e-01 4.45323363e-02 -3.62730056e-01 2.56100863e-01 -2.11684495e-01 -2.37368092e-01 6.56048417e-01 4.14122403e-01 5.05198538e-01 -8.86812508e-02 -7.64025748e-01 -9.68176603e-01 2.29229033e-01 1.97464132e+00 8.50556552e-01 8.59416962e-01 1.52595484e+00 3.66213530e-01 7.87739515e-01 1.26372886e+00 2.32649565e-01 6.54852331e-01 8.48187447e-01 9.40167785e-01 3.95411134e-01 2.39878803e-01 1.32289752e-01 3.15778941e-01 2.26855382e-01 -5.39161026e-01 -7.59955406e-01 -1.50961268e+00 9.14891481e-01 -2.03381467e+00 -1.05544651e+00 -8.92827451e-01 2.09590673e+00 2.09801346e-02 2.20994741e-01 1.05730928e-01 5.66651583e-01 5.45551896e-01 -6.78432733e-02 -2.25833848e-01 -1.05700874e+00 1.19271092e-01 -3.21184665e-01 1.24343920e+00 1.60812378e+00 -8.61426115e-01 9.54321027e-01 7.47738600e+00 4.18425053e-01 -8.97210658e-01 1.91141576e-01 -3.51189017e-01 1.74509600e-01 -2.97261715e-01 3.83207351e-01 -8.88757110e-01 2.24682733e-01 6.93898320e-01 -6.11871541e-01 3.90276879e-01 1.02557814e+00 2.78784692e-01 -6.56017005e-01 -2.79181421e-01 3.96303624e-01 -2.20746949e-01 -1.31933308e+00 -1.04071140e-01 3.12065274e-01 6.15251601e-01 2.81143524e-02 -2.30521411e-01 -7.74558112e-02 3.65154713e-01 -1.01664329e+00 7.88478255e-01 2.95137092e-02 1.82288021e-01 -9.26318049e-01 7.46289372e-01 8.93969178e-01 -1.60503078e+00 -2.12267816e-01 -6.18516207e-01 -7.34475374e-01 1.02825129e+00 -1.56482026e-01 -1.05153692e+00 5.81092894e-01 5.89622498e-01 -1.52330492e-02 1.23983584e-01 1.19525278e+00 9.90401506e-02 -4.00600016e-01 -5.29873550e-01 -3.56146455e-01 9.92236853e-01 -1.93710536e-01 1.05381382e+00 9.95365083e-01 6.74269557e-01 4.87588435e-01 4.77036983e-01 1.04849473e-01 7.07511723e-01 1.64367110e-01 -9.63991225e-01 4.95543391e-01 5.32196343e-01 7.19495595e-01 -1.20700192e+00 9.18758810e-02 -6.93950132e-02 8.94077599e-01 1.54047698e-01 1.76589444e-01 -7.38694549e-01 -8.49161446e-01 5.34994364e-01 3.97776008e-01 2.01556936e-01 -9.15171444e-01 -2.77505726e-01 -1.02838688e-01 -7.00579733e-02 -1.05034548e-03 1.61350351e-02 -3.51889789e-01 -3.56569998e-02 7.58725584e-01 1.20042570e-01 -1.60121834e+00 -4.62022692e-01 -6.57355428e-01 -8.19313228e-01 7.40788460e-01 -1.54226387e+00 -4.36899990e-01 -1.71022490e-01 3.04748893e-01 8.74100208e-01 2.53249258e-01 7.88990915e-01 2.83111874e-02 -2.00044349e-01 7.44338557e-02 2.00118572e-01 -5.12562513e-01 -1.20910384e-01 -8.05108249e-01 5.00050008e-01 1.03720212e+00 -8.52712750e-01 4.14361268e-01 1.20599663e+00 -7.39924788e-01 -1.52556062e+00 -8.80969405e-01 1.45179355e+00 -4.53858793e-01 3.43066126e-01 6.34236336e-01 -3.83647770e-01 7.33678281e-01 8.48704427e-02 -2.94844598e-01 2.35098094e-01 -6.09118223e-01 6.74700618e-01 7.38180101e-01 -1.37535870e+00 9.70320344e-01 1.30592227e+00 5.18599391e-01 -3.47067237e-01 9.44519415e-02 5.30569196e-01 -7.09031999e-01 -1.88057318e-01 4.84666914e-01 4.89965409e-01 -4.98602390e-01 1.10458076e+00 -4.25171137e-01 -1.70354411e-01 -6.31770372e-01 -3.87035161e-01 -9.26075220e-01 -5.28248012e-01 -8.62156630e-01 4.51643914e-01 5.06765485e-01 6.71329081e-01 -5.86131752e-01 8.36081982e-01 6.40625000e-01 -3.50117952e-01 -9.41506922e-01 -1.59931242e+00 -1.09978795e+00 1.10194065e-01 -5.73469877e-01 2.93781072e-01 8.18626404e-01 4.09022182e-01 -9.98209491e-02 -3.77513140e-01 5.90181649e-01 4.43426728e-01 -1.52928278e-01 7.32065082e-01 -8.69879365e-01 3.37567478e-01 -4.19120669e-01 -5.70621610e-01 -1.08063650e+00 -1.14075951e-01 -6.77078724e-01 4.30241942e-01 -2.22997761e+00 -4.96858239e-01 -8.89764488e-01 3.30286562e-01 3.67206335e-01 2.38516062e-01 -7.21178800e-02 3.06271911e-01 3.82682592e-01 -6.89451158e-01 -5.35992645e-02 1.04571998e+00 -8.61269087e-02 -6.42910957e-01 9.83510017e-02 -3.49318802e-01 1.15430450e+00 1.18929708e+00 -8.57667804e-01 -4.91903454e-01 -6.33588612e-01 4.83228147e-01 5.74200749e-01 -3.56355220e-01 -1.00160360e+00 5.05377650e-01 -1.10627937e+00 -5.97142160e-01 -8.02384734e-01 3.95563364e-01 -8.36356461e-01 3.22237670e-01 1.11239886e+00 3.33550483e-01 7.02778339e-01 3.12143326e-01 6.70347154e-01 1.66478336e-01 -9.97943163e-01 6.14076376e-01 -3.95862281e-01 -1.21793091e+00 -2.01641340e-02 -1.34419310e+00 -2.92133451e-01 1.64677465e+00 -8.12351704e-01 -3.36682022e-01 -2.08603024e-01 -7.40782619e-01 5.03569722e-01 6.41587794e-01 3.04175019e-01 7.52228498e-01 -1.22206080e+00 -2.90151507e-01 -4.42382395e-01 -3.50275815e-01 -3.05578895e-02 2.24629696e-02 7.81911671e-01 -1.44034564e+00 9.70962524e-01 -4.57125157e-01 -1.88691795e-01 -1.42630577e+00 7.90706635e-01 2.71253496e-01 -2.94829190e-01 -4.58208889e-01 6.63655043e-01 -5.41165352e-01 -2.63061821e-01 -1.99158847e-01 -1.14178091e-01 -5.85182011e-01 -3.46254885e-01 5.07324278e-01 1.22565413e+00 -5.60888231e-01 -9.86324906e-01 -9.10421193e-01 1.11629224e+00 6.68100238e-01 -5.71967661e-01 8.54825795e-01 -6.91581070e-01 -2.48562008e-01 1.15549952e-01 8.05004537e-01 3.66170287e-01 -6.74985468e-01 4.13312703e-01 4.32999134e-01 -5.87549388e-01 -5.31638324e-01 6.15186989e-02 -8.84292573e-02 2.34223768e-01 3.80505592e-01 3.43200117e-01 6.32455885e-01 -1.29802957e-01 7.49189019e-01 8.27505648e-01 1.05062473e+00 -1.08207655e+00 -1.08020388e-01 1.13310421e+00 8.31712425e-01 -9.51240838e-01 1.26788452e-01 -9.34018314e-01 -3.70388269e-01 1.12066019e+00 5.21890283e-01 -4.00867790e-01 6.53788388e-01 4.44403946e-01 2.53609680e-02 2.51504421e-01 -3.31535757e-01 -3.61152172e-01 -3.76276582e-01 1.02861857e+00 -3.76005024e-01 6.95087761e-02 -1.18150854e+00 3.20601970e-01 -2.61043400e-01 -3.29685092e-01 1.14315605e+00 1.59694672e+00 -1.52259386e+00 -1.27481973e+00 -7.10204840e-01 -3.05464894e-01 -4.91076224e-02 7.78681189e-02 -2.07266971e-01 1.20466459e+00 3.72303993e-01 1.38755071e+00 1.12537898e-01 -2.54470378e-01 5.52390635e-01 -3.17806154e-01 2.74164796e-01 -4.06441152e-01 2.38907814e-01 -4.30177480e-01 5.02625883e-01 -7.08577037e-01 -3.68952274e-01 -4.72889841e-01 -2.29781985e+00 -5.35200238e-01 -6.50626495e-02 2.46523485e-01 6.45225286e-01 9.46089864e-01 1.18396133e-01 1.35300606e-01 4.55335498e-01 -1.11461449e+00 -7.86065832e-02 -3.11010331e-01 -2.01509550e-01 -4.16803569e-01 3.27123225e-01 -5.35818160e-01 -3.04984212e-01 -3.71369302e-01]
[5.020934581756592, 1.637241244316101]
811a0604-fdf8-4b16-b57b-5596a38b50a8
pseco-pseudo-labeling-and-consistency
2203.16317
null
https://arxiv.org/abs/2203.16317v2
https://arxiv.org/pdf/2203.16317v2.pdf
PseCo: Pseudo Labeling and Consistency Training for Semi-Supervised Object Detection
In this paper, we delve into two key techniques in Semi-Supervised Object Detection (SSOD), namely pseudo labeling and consistency training. We observe that these two techniques currently neglect some important properties of object detection, hindering efficient learning on unlabeled data. Specifically, for pseudo labeling, existing works only focus on the classification score yet fail to guarantee the localization precision of pseudo boxes; For consistency training, the widely adopted random-resize training only considers the label-level consistency but misses the feature-level one, which also plays an important role in ensuring the scale invariance. To address the problems incurred by noisy pseudo boxes, we design Noisy Pseudo box Learning (NPL) that includes Prediction-guided Label Assignment (PLA) and Positive-proposal Consistency Voting (PCV). PLA relies on model predictions to assign labels and makes it robust to even coarse pseudo boxes; while PCV leverages the regression consistency of positive proposals to reflect the localization quality of pseudo boxes. Furthermore, in consistency training, we propose Multi-view Scale-invariant Learning (MSL) that includes mechanisms of both label- and feature-level consistency, where feature consistency is achieved by aligning shifted feature pyramids between two images with identical content but varied scales. On COCO benchmark, our method, termed PSEudo labeling and COnsistency training (PseCo), outperforms the SOTA (Soft Teacher) by 2.0, 1.8, 2.0 points under 1%, 5%, and 10% labelling ratios, respectively. It also significantly improves the learning efficiency for SSOD, e.g., PseCo halves the training time of the SOTA approach but achieves even better performance. Code is available at https://github.com/ligang-cs/PseCo.
['Shanshan Zhang', 'Ding Liang', 'Yichao Wu', 'Yujie Wang', 'Xiang Li', 'Gang Li']
2022-03-30
null
null
null
null
['semi-supervised-object-detection']
['computer-vision']
[-1.56191215e-02 4.66665486e-03 -4.27934378e-01 -4.17753279e-01 -1.09965491e+00 -7.07249820e-01 5.88373065e-01 1.12505727e-01 -2.98387200e-01 4.57673609e-01 -2.25813851e-01 -3.11401375e-02 5.28065823e-02 -4.05615240e-01 -7.25015104e-01 -8.65258634e-01 2.25113556e-01 5.20770967e-01 7.32651830e-01 1.89590991e-01 -1.54566802e-02 3.78547817e-01 -1.47014022e+00 3.23424488e-01 6.19994938e-01 1.17151964e+00 1.78784236e-01 4.56278414e-01 1.19398594e-01 7.28025019e-01 -2.49569535e-01 -3.64258319e-01 4.59107935e-01 -1.74810484e-01 -5.36589086e-01 2.79106945e-01 8.60428154e-01 -4.65036303e-01 -7.37996101e-02 1.26599514e+00 3.64554882e-01 -2.16238096e-01 7.41011441e-01 -1.51325154e+00 -5.71365714e-01 3.86855215e-01 -9.53635335e-01 1.91315673e-02 -4.08042185e-02 4.36139479e-02 1.29392970e+00 -1.24826801e+00 6.66660011e-01 1.09305453e+00 7.31329441e-01 4.34594721e-01 -1.32706058e+00 -7.84141958e-01 1.46161854e-01 9.12028179e-03 -1.41153514e+00 -3.79145443e-01 6.35972798e-01 -5.11929929e-01 4.25875723e-01 1.92352712e-01 3.82485092e-01 8.00190985e-01 1.56647384e-01 8.60245228e-01 1.57865238e+00 -3.71519834e-01 3.98817897e-01 2.91049480e-01 3.14268559e-01 8.16954315e-01 5.71044683e-01 1.78543136e-01 -3.85958314e-01 -1.87290996e-01 8.31195056e-01 2.20386580e-01 -1.45759642e-01 -8.28734100e-01 -1.08895409e+00 6.72222078e-01 5.96656024e-01 9.78655666e-02 4.18738686e-02 8.90435427e-02 3.65148246e-01 1.25706360e-01 3.96478325e-01 2.45732978e-01 -6.13410890e-01 3.49625021e-01 -9.97805893e-01 5.46694621e-02 3.96381617e-01 1.11386430e+00 9.11222041e-01 -2.47307256e-01 -2.89601624e-01 7.69533098e-01 2.83742905e-01 6.75337315e-01 2.97835678e-01 -9.81995046e-01 2.58694738e-01 7.10491002e-01 2.17618048e-01 -8.04213583e-01 -6.00702822e-01 -5.67910373e-01 -7.37544298e-01 2.81268716e-01 5.81518233e-01 1.37059227e-01 -1.16576982e+00 1.72991705e+00 6.50334895e-01 7.58551583e-02 -2.01984346e-01 9.90004420e-01 7.79282331e-01 2.50383109e-01 3.93109247e-02 -1.73368737e-01 1.61230600e+00 -1.26425469e+00 -4.65373158e-01 -3.83062333e-01 7.49857426e-01 -8.59074414e-01 1.03688109e+00 3.99148524e-01 -9.29124951e-01 -6.43797576e-01 -1.08958292e+00 9.01291817e-02 -1.49395213e-01 3.79363090e-01 5.37596345e-01 5.62080085e-01 -9.20404732e-01 3.61370564e-01 -1.00148356e+00 -2.81354189e-01 5.97269833e-01 2.34072670e-01 -5.97373247e-01 -1.84820518e-01 -6.88454866e-01 7.88124144e-01 3.32089067e-01 -2.09619313e-01 -8.06362510e-01 -5.78994811e-01 -7.07656145e-01 -1.64324492e-02 6.35371447e-01 -3.36421967e-01 1.20641434e+00 -9.32901382e-01 -1.07978773e+00 1.00788128e+00 -1.82275012e-01 -3.11453521e-01 6.12923384e-01 -1.41801000e-01 -1.47947937e-01 2.26868615e-01 3.87931675e-01 1.01048839e+00 1.03929496e+00 -1.54865050e+00 -7.78533697e-01 -3.26277852e-01 1.82702709e-02 8.44301060e-02 -2.05723509e-01 -1.17288709e-01 -7.19630182e-01 -5.31511486e-01 6.34798706e-01 -1.15112507e+00 -7.49140307e-02 3.71871710e-01 -2.93864101e-01 -2.88984001e-01 8.37416053e-01 -3.93213898e-01 9.56552148e-01 -2.24014115e+00 -2.33003750e-01 1.35869965e-01 2.87785709e-01 8.83180797e-02 -1.25758618e-01 9.94155183e-03 9.94217321e-02 -9.17141214e-02 -1.31885558e-01 -6.52165592e-01 -2.34713387e-02 2.77565956e-01 -2.26956934e-01 6.93201423e-01 2.38604650e-01 9.59515989e-01 -8.79589736e-01 -7.43417084e-01 1.40166819e-01 1.61683902e-01 -6.35453522e-01 -3.28529365e-02 -8.67652223e-02 4.12114322e-01 -2.72641897e-01 8.70314479e-01 9.12964046e-01 -7.82425582e-01 3.76049548e-01 -4.01681364e-01 2.09960137e-02 1.22896954e-02 -1.51594269e+00 1.25456190e+00 -3.05673480e-01 2.09212393e-01 7.36624748e-02 -5.96766531e-01 7.72299170e-01 1.35088235e-01 4.91928846e-01 -6.41675711e-01 -2.32772063e-02 4.77633864e-01 -2.14031443e-01 -1.85592882e-02 4.23778385e-01 1.22181825e-01 1.17782699e-02 4.49528903e-01 1.67854160e-01 -1.62324652e-01 1.46396875e-01 3.29076111e-01 8.93111467e-01 2.85786569e-01 5.57495356e-01 -1.84013426e-01 3.56720507e-01 -1.45004809e-01 7.67220020e-01 8.94092977e-01 -5.77470541e-01 8.73892665e-01 4.37202603e-01 -2.83277690e-01 -1.12066734e+00 -1.03072715e+00 -5.52388549e-01 1.16510606e+00 5.87050319e-01 -2.88870960e-01 -5.84507167e-01 -1.20986211e+00 1.84522197e-01 2.87579864e-01 -6.24633133e-01 9.58770514e-03 -4.52760756e-01 -6.16123259e-01 2.19216466e-01 6.33914590e-01 4.55922425e-01 -7.29698718e-01 -3.93369645e-01 -1.33185774e-01 -2.48724557e-02 -1.29498208e+00 -7.08641052e-01 5.30097604e-01 -7.00175166e-01 -1.10657847e+00 -5.13112187e-01 -7.03120828e-01 9.67874467e-01 7.08363831e-01 9.00193453e-01 1.93757862e-01 4.35607135e-02 2.74447173e-01 -4.80883718e-01 -1.29223228e-01 -2.38390237e-01 -7.12742209e-02 1.73740253e-01 -5.76454215e-02 2.14564666e-01 -3.72546375e-01 -6.35019898e-01 8.08602631e-01 -8.44196975e-01 9.77255180e-02 6.81791604e-01 1.22297585e+00 8.90779614e-01 -2.37742975e-01 5.03090322e-01 -1.02497160e+00 -2.53059477e-01 -2.46740371e-01 -8.28724086e-01 3.35417241e-01 -9.61280286e-01 -2.78016645e-02 4.23321456e-01 -5.09984672e-01 -7.26483405e-01 3.21659237e-01 1.04751959e-01 -6.49768293e-01 -1.71727464e-01 -1.11446403e-01 5.94430566e-02 -4.15590793e-01 7.26370871e-01 1.48916319e-01 -2.09242597e-01 -2.57468611e-01 3.73596013e-01 5.01204908e-01 3.70641470e-01 -4.05781776e-01 1.06801593e+00 9.07702446e-01 -1.39471054e-01 -1.97456971e-01 -1.36335027e+00 -8.13883007e-01 -6.92262650e-01 -5.97347617e-02 5.61208665e-01 -1.07402980e+00 -3.45645308e-01 2.07025036e-01 -9.00038779e-01 -2.25793958e-01 -3.26834351e-01 3.95446450e-01 -4.57061768e-01 4.34989125e-01 -6.32136285e-01 -6.64108038e-01 -3.19484860e-01 -1.16060519e+00 1.43259382e+00 1.61659643e-01 3.36201377e-02 -6.71292424e-01 -1.94211364e-01 4.05540019e-01 3.03599268e-01 -4.11171280e-02 5.89857101e-01 -7.95977175e-01 -6.89218163e-01 -2.45242372e-01 -6.31541073e-01 4.07271624e-01 -1.31105864e-02 -1.71728522e-01 -1.16811419e+00 -6.38431430e-01 -3.33544686e-02 -7.02925742e-01 8.18463326e-01 4.40467060e-01 8.93380702e-01 -5.47035672e-02 -3.73320192e-01 5.25909781e-01 1.39509797e+00 -1.40271187e-01 2.18215480e-01 3.06239814e-01 5.92419624e-01 5.48539042e-01 1.01138675e+00 4.60032493e-01 3.86997968e-01 8.69168699e-01 5.11790991e-01 -2.48857602e-01 -4.16793495e-01 -4.18198258e-01 3.42667252e-01 5.52262962e-01 1.92510784e-01 1.14966258e-01 -7.10207343e-01 3.11153382e-01 -1.83192456e+00 -7.32316792e-01 -1.07041955e-01 2.19071341e+00 8.02686691e-01 3.53822798e-01 8.64896402e-02 1.04899742e-01 7.94524252e-01 9.55999643e-02 -5.26271343e-01 2.83651114e-01 -5.43958433e-02 -9.72791687e-02 7.62729764e-01 3.97105932e-01 -1.43068612e+00 9.98023152e-01 5.16702652e+00 1.12519479e+00 -1.08362401e+00 5.50284445e-01 7.40590870e-01 1.06777444e-01 1.40885194e-03 1.80342272e-01 -1.24780357e+00 3.72249812e-01 2.92548478e-01 3.62843961e-01 1.47670776e-01 1.38099980e+00 1.81079526e-02 -2.86822706e-01 -1.08857727e+00 8.82792711e-01 3.20614725e-02 -1.08757913e+00 -1.67206436e-01 -2.82870382e-02 1.07487595e+00 1.73523560e-01 1.86075643e-01 3.93218756e-01 1.85973898e-01 -5.75348318e-01 1.12605464e+00 2.81681325e-02 8.42334509e-01 -3.82588774e-01 9.03458953e-01 3.45023662e-01 -1.37901902e+00 -1.29240453e-01 -4.22741652e-01 3.44278365e-01 1.47338435e-02 6.84504867e-01 -7.88429976e-01 5.01713157e-01 7.30659723e-01 5.71884036e-01 -9.52519000e-01 9.00705338e-01 -4.34268802e-01 6.03446424e-01 -3.76956224e-01 2.63608277e-01 9.29350480e-02 -1.50228217e-02 3.55531871e-01 1.05957544e+00 1.92261878e-02 -2.08717361e-01 4.92555052e-01 7.33166695e-01 -6.04271106e-02 -4.31513712e-02 -1.64345205e-01 5.12668371e-01 7.29192793e-01 1.60534132e+00 -1.08614242e+00 -3.25628340e-01 -4.27920789e-01 8.28660965e-01 5.00830293e-01 1.20026283e-01 -9.30207610e-01 6.01138249e-02 1.16476819e-01 1.22875765e-01 4.99101579e-01 -1.24091148e-01 -4.08321530e-01 -1.24631166e+00 1.12208299e-01 -7.60892630e-01 4.58152950e-01 -4.98147875e-01 -1.39650893e+00 4.66359556e-01 -1.32858828e-02 -1.40493727e+00 1.33792311e-01 -5.63623309e-01 -3.43577683e-01 4.72754240e-01 -1.75439489e+00 -1.54906106e+00 -4.55432594e-01 2.37487376e-01 5.25441945e-01 1.37206078e-01 4.18444067e-01 3.57819200e-01 -5.90682089e-01 7.65003383e-01 1.17957480e-01 6.36585727e-02 1.07410634e+00 -1.32139730e+00 2.79404968e-01 7.51714826e-01 4.11061674e-01 5.41120887e-01 4.73789990e-01 -5.52547336e-01 -1.04082978e+00 -1.35846686e+00 7.14645624e-01 -6.69215441e-01 5.52623510e-01 -4.49273437e-01 -9.55209553e-01 6.55428767e-01 -4.51777786e-01 7.30245829e-01 3.02699119e-01 -4.91072498e-02 -7.38261878e-01 -1.66405722e-01 -1.14214122e+00 4.26090449e-01 1.05386281e+00 -4.59357113e-01 -3.04484487e-01 4.43582952e-01 6.77996516e-01 -5.42731822e-01 -6.50445819e-01 6.96847498e-01 6.48980856e-01 -1.13369012e+00 8.60704005e-01 -1.38391361e-01 1.77020416e-01 -6.55810177e-01 -2.36366585e-01 -6.30367875e-01 -4.51660067e-01 -8.42565596e-02 -2.22783297e-01 1.26707685e+00 3.04868221e-01 -6.55889153e-01 9.20093775e-01 3.35758746e-01 -4.59716991e-02 -9.67726886e-01 -1.13516474e+00 -1.05250490e+00 -1.89038217e-01 -4.31108832e-01 2.57782310e-01 9.93324995e-01 -4.42911595e-01 1.51321948e-01 -3.62540513e-01 3.96973729e-01 6.16194963e-01 3.57122481e-01 8.59407604e-01 -1.09565651e+00 -5.43789029e-01 -2.91060179e-01 -2.73683906e-01 -1.16836405e+00 -1.53868208e-02 -9.29678619e-01 9.80589241e-02 -1.11800182e+00 5.74097335e-01 -8.62397492e-01 -3.39286208e-01 6.98478162e-01 -3.23963284e-01 7.89940059e-01 2.80044526e-01 6.12655759e-01 -1.04797399e+00 3.10216337e-01 1.21410871e+00 3.90005149e-02 -5.66982217e-02 -5.83101511e-02 -4.30741638e-01 9.04645503e-01 5.81442356e-01 -7.58564174e-01 -7.11402670e-02 -1.04501761e-01 8.70859483e-04 -1.88018858e-01 5.81806600e-01 -1.00958836e+00 2.66554624e-01 -9.32597518e-02 4.65404272e-01 -5.86372614e-01 2.08757833e-01 -8.56693149e-01 -1.09463111e-01 6.56718135e-01 -1.74434841e-01 -1.39609843e-01 -6.43699989e-02 7.74479866e-01 -4.58166711e-02 -2.78648645e-01 1.21091199e+00 1.19237132e-01 -6.49320483e-01 2.91052133e-01 2.32226685e-01 4.77964170e-02 1.06647658e+00 -1.21963896e-01 -4.45625693e-01 -1.35814026e-01 -5.03310919e-01 3.71001691e-01 7.24660218e-01 1.84984937e-01 4.04014260e-01 -1.33707738e+00 -3.96474898e-01 2.07509860e-01 4.26748693e-01 2.37562940e-01 1.99237317e-01 1.22149444e+00 -2.77228862e-01 3.58145863e-01 1.67342976e-01 -1.02119529e+00 -1.27960801e+00 4.85850096e-01 2.20910430e-01 -4.67065543e-01 -4.48045880e-01 8.97584915e-01 4.41290170e-01 -5.25586128e-01 3.46146971e-01 -2.62333184e-01 7.42963254e-02 -6.69068396e-02 3.11392039e-01 3.01928669e-01 1.03457786e-01 -5.42958617e-01 -4.11802232e-01 7.30061710e-01 -3.42781872e-01 6.16141632e-02 9.28125978e-01 -1.04521386e-01 1.67772874e-01 3.38172734e-01 9.52664733e-01 1.80019006e-01 -1.64481103e+00 -5.93978584e-01 2.28730232e-01 -4.43838716e-01 1.33452518e-02 -6.60760105e-01 -8.98539007e-01 5.34744203e-01 7.26978362e-01 2.69833207e-02 9.24052060e-01 3.36502433e-01 5.53802848e-01 1.69274077e-01 6.77970767e-01 -9.25219119e-01 4.41866159e-01 3.45660686e-01 6.24178886e-01 -1.72464752e+00 3.37863714e-01 -5.92710316e-01 -7.13131487e-01 9.31869864e-01 8.74789894e-01 -6.65632784e-02 3.88940692e-01 2.77187824e-01 -1.41943749e-02 -1.05061971e-01 -6.78622067e-01 -3.02508086e-01 4.77692664e-01 2.67397225e-01 1.81211501e-01 -3.42212319e-02 -1.32455647e-01 5.61983883e-01 2.07372919e-01 -3.11351150e-01 1.43348083e-01 8.82293880e-01 -6.00243330e-01 -1.05199254e+00 -5.26168168e-01 4.87768888e-01 -3.69109064e-01 2.38766074e-02 -1.52646184e-01 8.43865693e-01 3.97558302e-01 7.13188350e-01 -2.42290813e-02 -2.92134911e-01 1.54314265e-01 -1.29752114e-01 3.02680463e-01 -7.66625285e-01 -3.36977780e-01 3.27338845e-01 -3.05194229e-01 -5.19963443e-01 -5.44803679e-01 -6.32037461e-01 -1.29381990e+00 -8.78079794e-03 -9.87559199e-01 7.03750029e-02 4.32997197e-01 7.69886851e-01 2.50337929e-01 1.95329562e-01 7.84268498e-01 -9.93079543e-01 -9.00357842e-01 -8.82562160e-01 -6.31886423e-01 4.59696949e-01 2.86337137e-01 -9.05210257e-01 -6.29288912e-01 -3.47087681e-02]
[9.203326225280762, 1.2724902629852295]
b1e03163-6a80-4c82-a6cd-c835b8cdf860
deep-smart-contract-intent-detection
2211.10724
null
https://arxiv.org/abs/2211.10724v1
https://arxiv.org/pdf/2211.10724v1.pdf
Deep Smart Contract Intent Detection
Nowadays, security activities in smart contracts concentrate on vulnerability detection. Despite early success, we find that developers' intent to write smart contracts is a more noteworthy security concern because smart contracts with malicious intent have caused significant users' financial loss. Unfortunately, current approaches to identify the aforementioned malicious smart contracts rely on smart contract security audits, which entail huge manpower consumption and financial expenditure. To resolve this issue, we propose a novel deep learning-based approach, SmartIntentNN, to conduct automated smart contract intent detection. SmartIntentNN consists of three primary parts: a pre-trained sentence encoder to generate the contextual representations of smart contracts, a K-means clustering method to highlight intent-related representations, and a bidirectional LSTM-based (long-short term memory) multi-label classification network to predict the intents in smart contracts. To evaluate the performance of SmartIntentNN, we collect more than 40,000 real smart contracts and perform a series of comparison experiments with our selected baseline approaches. The experimental results demonstrate that SmartIntentNN outperforms all baselines by up to 0.8212 in terms of the f1-score metric.
['Youshuai Tan', 'Sen Fang', 'Tao Zhang', 'Youwei Huang']
2022-11-19
null
null
null
null
['vulnerability-detection', 'intent-detection']
['miscellaneous', 'natural-language-processing']
[ 1.57174021e-01 1.26879215e-01 -2.29798734e-01 -4.82312977e-01 -1.06567073e+00 -8.21023285e-01 5.29073179e-01 -4.33428675e-01 1.19756185e-01 1.78521663e-01 5.82454801e-01 -8.28356802e-01 2.20063999e-01 -6.41030133e-01 -3.35256666e-01 -6.39161348e-01 1.79901332e-01 4.00312282e-02 -1.69610139e-02 -7.87008703e-02 7.42295682e-01 -1.26741678e-01 -8.14091861e-01 8.86653543e-01 7.66158640e-01 8.48017871e-01 -1.80193633e-01 3.97593141e-01 8.64631310e-02 1.35511506e+00 -8.25259566e-01 -1.16938090e+00 3.06031644e-01 -7.29288533e-02 -7.94335663e-01 -5.87981105e-01 -1.53801352e-01 -6.78468227e-01 -2.47979194e-01 1.30023897e+00 5.17164707e-01 -5.54656029e-01 3.97842556e-01 -1.22975791e+00 -9.73159671e-01 1.20831466e+00 -5.59634745e-01 1.12050712e-01 1.78132728e-01 9.81683582e-02 1.48235714e+00 -6.75057590e-01 5.15774071e-01 7.97385275e-01 8.32538426e-01 4.90629077e-01 -7.63282835e-01 -7.97200859e-01 -1.42811999e-01 2.06953183e-01 -9.52923656e-01 -1.85988724e-01 1.09821856e+00 -7.01443672e-01 1.50027227e+00 5.78027889e-02 6.69041798e-02 1.49877954e+00 7.22729325e-01 6.74571157e-01 8.68558764e-01 -1.06439956e-01 9.45570692e-02 3.37241232e-01 -4.58371304e-02 3.96775812e-01 2.09113219e-04 2.40957811e-01 -4.61037531e-02 -7.19150305e-01 2.46715814e-01 3.19107980e-01 1.88887864e-02 1.29394084e-01 -7.21036911e-01 1.21286702e+00 -1.05137311e-01 7.07106411e-01 -2.87896723e-01 2.79474199e-01 9.26264524e-01 6.19533420e-01 6.27461672e-01 4.35918868e-01 -1.00501978e+00 -6.43080413e-01 -5.73026419e-01 1.13284932e-02 9.70330775e-01 7.58374870e-01 4.45068151e-01 -2.70791687e-02 -8.35604370e-02 6.17433846e-01 5.11456728e-01 3.28541875e-01 6.76652849e-01 -8.04590225e-01 1.07038546e+00 8.34807694e-01 -3.73600423e-02 -1.03545034e+00 2.05131602e-02 -5.53287804e-01 -5.55723071e-01 1.28031999e-01 -1.42629236e-01 -3.92121106e-01 -1.45797938e-01 1.24416935e+00 -1.93408087e-01 6.56278133e-02 1.84690341e-01 3.35593700e-01 5.14250398e-01 5.89986026e-01 7.18325330e-03 -2.25842863e-01 1.27417588e+00 -1.02947426e+00 -4.44878072e-01 5.47762476e-02 1.17219639e+00 -8.07926059e-01 1.06128514e+00 2.35252082e-01 -5.62924385e-01 -4.50316770e-03 -8.40609848e-01 5.14213026e-01 -4.03651178e-01 8.96410793e-02 7.10380495e-01 1.20448887e+00 -6.87088668e-01 6.35352612e-01 -5.39492369e-01 3.09856772e-01 5.35670400e-01 9.88147929e-02 6.05449872e-03 2.86558300e-01 -1.21675885e+00 6.78743780e-01 2.10602526e-02 -2.09506080e-01 -9.86537337e-01 -5.87633729e-01 -8.30038369e-01 3.00837278e-01 2.97201753e-01 -1.42086409e-02 1.52646923e+00 -7.39449799e-01 -1.23459435e+00 6.75200880e-01 1.89527109e-01 -3.24156463e-01 1.30492672e-01 -3.23775262e-01 -5.77365279e-01 -2.15435773e-01 1.08181104e-01 -2.02047035e-01 7.19693184e-01 -8.91379476e-01 -5.07043481e-01 -1.47670254e-01 2.65413702e-01 -1.43870771e-01 -6.83603525e-01 8.63519907e-01 1.59560010e-01 -8.82248580e-01 -3.61970246e-01 -8.96978140e-01 -2.39521846e-01 -1.11669290e+00 -3.83535534e-01 -5.96542060e-01 8.62628281e-01 -9.51839209e-01 1.47145092e+00 -2.05397201e+00 -4.40198958e-01 4.49847020e-02 4.77305725e-02 3.87928665e-01 -3.55492741e-01 8.46300185e-01 -2.25306153e-01 3.67401540e-01 -2.04640836e-01 -3.36037338e-01 4.12991911e-01 -1.22678511e-01 -8.87938857e-01 2.32122004e-01 -5.05599044e-02 1.20650613e+00 -6.96424961e-01 -3.49028647e-01 -7.08368197e-02 2.65897483e-01 -4.38246995e-01 3.07905376e-01 -1.86662883e-01 1.52282417e-01 -6.79538429e-01 8.98746669e-01 5.33189833e-01 -3.18900943e-01 2.27157325e-01 2.79924780e-01 -1.70859359e-02 7.72518516e-01 -3.51220161e-01 1.53757811e+00 -7.30289280e-01 4.17481869e-01 -2.87186444e-01 -9.26984906e-01 9.39097345e-01 7.89150000e-01 6.45827055e-01 -7.09451258e-01 1.12080127e-01 3.58684480e-01 -8.84574279e-02 -9.04608428e-01 1.28613979e-01 -4.13481519e-02 -4.18975145e-01 1.30806494e+00 -5.54654360e-01 2.26449803e-01 -3.06134641e-01 1.07185796e-01 1.78957963e+00 6.25456944e-02 8.54839087e-02 1.77817583e-01 5.80909789e-01 -4.31981862e-01 1.03655028e+00 3.96604598e-01 -2.00067967e-01 3.91496778e-01 8.59969318e-01 -1.05738711e+00 -8.27267945e-01 -4.53007758e-01 2.75504142e-01 1.18454206e+00 -2.55765796e-01 -4.76763546e-01 -8.50270331e-01 -1.67554724e+00 -1.83267951e-01 8.82173955e-01 -4.07101512e-01 -1.90055057e-01 -7.44200528e-01 -8.24509144e-01 7.65012801e-01 5.68438530e-01 3.71187091e-01 -1.81120539e+00 -7.85463274e-01 1.86035991e-01 -1.71216860e-01 -9.86166537e-01 -6.73038006e-01 1.17048901e-02 -6.20254099e-01 -1.30564320e+00 -3.92953753e-01 -7.30156183e-01 5.77823818e-01 -1.75842062e-01 1.00244021e+00 3.32765877e-01 -5.28603382e-02 -8.65667909e-02 -5.89994788e-01 -4.45422158e-02 -6.63652718e-01 2.84204185e-01 -2.04237774e-01 -5.54207526e-02 8.03336620e-01 -7.61557221e-01 -4.50368375e-01 1.92492455e-01 -7.88476408e-01 -3.09093326e-01 1.04974294e+00 6.41799808e-01 -2.21159190e-01 3.97942722e-01 9.27337050e-01 -1.31960881e+00 8.95914853e-01 -4.81924236e-01 -6.63840175e-01 4.80472654e-01 -1.19294751e+00 -7.71150514e-02 8.27978551e-01 -1.84155390e-01 -1.25641155e+00 -2.66872793e-02 -5.26632130e-01 -2.86197569e-02 -5.56874871e-02 6.16034627e-01 -1.79128140e-01 -7.45706335e-02 2.62876779e-01 3.57337683e-01 -5.72366416e-01 -5.66176474e-01 1.80960268e-01 1.26159728e+00 1.52894557e-01 -1.23414382e-01 1.15726721e+00 1.00107774e-01 -6.95968509e-01 -2.47418229e-02 -7.40620017e-01 -1.86203241e-01 -2.99855471e-01 9.86942872e-02 6.44921005e-01 -5.53144097e-01 -8.65323305e-01 5.81176281e-01 -1.61277068e+00 -1.27369866e-01 2.16462016e-01 2.82486618e-01 -4.04347986e-01 7.05548823e-01 -1.06189477e+00 -8.74701321e-01 -1.03843999e+00 -1.43627357e+00 8.55806112e-01 -7.89825618e-02 -2.37358540e-01 -9.54423189e-01 4.82657671e-01 9.59684432e-01 4.93083507e-01 4.31417897e-02 1.27596807e+00 -1.05581045e+00 -4.51443762e-01 -4.74195749e-01 -2.37789869e-01 5.32029748e-01 2.06333116e-01 -2.28593305e-01 -9.90175247e-01 -3.31264168e-01 6.54399395e-01 -2.92558968e-01 6.34480417e-01 8.21237341e-02 1.01902461e+00 -9.74212348e-01 -2.28425696e-01 4.62764978e-01 1.35086191e+00 8.60621393e-01 6.68174744e-01 4.32288200e-01 5.67169189e-01 4.59022582e-01 5.63959897e-01 6.16130531e-01 5.68236589e-01 6.09653652e-01 7.09420860e-01 2.12618813e-01 5.44462204e-01 -2.36267611e-01 8.00073206e-01 8.45892489e-01 -1.46535560e-02 1.65725738e-01 -9.80187833e-01 4.78576303e-01 -1.81067502e+00 -1.19716465e+00 1.46554798e-01 1.88264859e+00 8.89781892e-01 2.81837344e-01 -2.56313056e-01 7.79688731e-02 4.62405562e-01 3.04722279e-01 -4.78977680e-01 -6.59353912e-01 1.91815019e-01 2.41840794e-03 3.27919215e-01 1.79173887e-01 -1.31648242e+00 8.73361826e-01 5.58071756e+00 7.97608852e-01 -9.64163482e-01 5.38174093e-01 8.01110923e-01 6.16846755e-02 -8.56746256e-01 2.50244677e-01 -4.75435197e-01 8.93873215e-01 7.99882531e-01 -1.76039524e-02 3.13245296e-01 1.31433761e+00 4.36398014e-02 5.12648225e-01 -8.08504522e-01 7.81207979e-01 2.06050381e-01 -1.48599946e+00 -4.93337959e-01 3.33541334e-01 9.06219006e-01 2.18505338e-01 1.44477919e-01 4.71140236e-01 6.19190574e-01 -8.85869503e-01 4.14661497e-01 1.61828488e-01 5.31917870e-01 -7.49271929e-01 1.30636513e+00 3.96868736e-01 -1.06243193e+00 -5.90290666e-01 -2.95441389e-01 -1.27285764e-01 1.07779369e-01 5.65036595e-01 -9.82564867e-01 5.38246870e-01 6.92224145e-01 9.36413646e-01 -4.11999941e-01 5.24687171e-01 -4.75964725e-01 6.74872100e-01 4.79034811e-01 -1.67888775e-01 1.17848314e-01 -1.19492985e-01 8.66182446e-02 1.12452948e+00 3.57143342e-01 -6.08580820e-02 -1.38757586e-01 9.90534961e-01 -1.66497439e-01 -1.10764325e-01 -8.02181184e-01 -3.63126844e-01 4.86697018e-01 1.48018420e+00 -4.74956483e-01 -2.04162464e-01 -1.01319742e+00 8.59999001e-01 7.56832808e-02 3.80708814e-01 -1.00822508e+00 -7.05962300e-01 7.18402565e-01 -2.22108305e-01 4.65679020e-01 5.49929775e-02 -5.01878858e-01 -1.20516789e+00 2.60349661e-01 -1.12208784e+00 3.67509902e-01 -3.72830182e-01 -1.17005777e+00 9.63651299e-01 -6.62710488e-01 -1.41533375e+00 -4.65102673e-01 -2.19605044e-01 -1.12416065e+00 5.76669037e-01 -1.41533303e+00 -1.24620807e+00 3.16428244e-01 1.90028161e-01 6.50611699e-01 -8.50722075e-01 8.64946961e-01 3.56382936e-01 -6.15673244e-01 7.17187166e-01 -2.58020982e-02 6.06189311e-01 3.89012039e-01 -1.06240761e+00 9.22401726e-01 8.95756900e-01 4.18499321e-01 8.16630065e-01 2.93442369e-01 -9.13748145e-01 -1.19893444e+00 -1.00916231e+00 1.45941722e+00 -6.16882086e-01 7.45658398e-01 -3.55645865e-01 -5.00810325e-01 8.79160881e-01 5.34382045e-01 -3.92524838e-01 9.22841668e-01 -7.63813257e-02 -7.50272751e-01 -6.57826886e-02 -9.77469444e-01 2.17600107e-01 7.80726850e-01 -8.86557937e-01 -6.35531306e-01 4.87950653e-01 9.18069482e-01 1.23004191e-01 -9.91510451e-01 4.19637293e-01 5.65216601e-01 -1.30805409e+00 6.36946082e-01 -5.40409923e-01 7.20918179e-01 5.98351173e-02 -3.55647728e-02 -8.69605720e-01 -5.03936186e-02 -8.05567443e-01 -5.33272326e-02 1.39640057e+00 6.38993859e-01 -7.64799654e-01 1.07925975e+00 7.96904743e-01 -2.63495386e-01 -1.03031933e+00 -9.97777224e-01 -7.49082208e-01 2.50201263e-02 -5.45985401e-01 9.43208277e-01 1.28164852e+00 6.11199975e-01 2.25289792e-01 -5.56980729e-01 2.04401929e-02 5.18198669e-01 7.06600845e-01 3.62863451e-01 -8.62951815e-01 -5.56789100e-01 -4.60069746e-01 -1.16675615e-01 -7.71423757e-01 4.93765354e-01 -6.16680443e-01 -1.95906028e-01 -1.22298729e+00 5.60536683e-01 -3.37360620e-01 -3.45136523e-01 9.13974822e-01 1.12911172e-01 1.72929335e-02 -1.99900746e-01 3.43270421e-01 -6.10600531e-01 4.39249486e-01 6.32022977e-01 -4.54974174e-01 -3.24107148e-02 6.33561075e-01 -9.14778769e-01 6.84955060e-01 9.62262988e-01 -7.41121590e-01 -2.76239574e-01 -4.76531208e-01 6.90051913e-01 2.29721740e-01 -1.84908107e-01 -4.60169077e-01 1.40551701e-01 -2.32896760e-01 -3.21896553e-01 -5.30245662e-01 -1.84480682e-01 -7.18498051e-01 -5.86101459e-03 9.32639003e-01 -4.45388377e-01 1.87329322e-01 -5.29109418e-01 1.92070559e-01 -2.38985568e-01 -7.09115446e-01 1.75561070e-01 -1.56418607e-01 -3.97169858e-01 1.93419263e-01 -4.26388144e-01 -1.67480871e-01 9.47983682e-01 1.35043234e-01 -7.09792256e-01 -4.00765628e-01 -2.70211846e-02 -9.58679318e-02 3.22159022e-01 9.15437996e-01 7.09357142e-01 -1.22856081e+00 -6.96611583e-01 2.71812350e-01 6.86874762e-02 -5.64167440e-01 2.45588422e-01 7.01857150e-01 -3.73200953e-01 6.82300389e-01 1.90178931e-01 8.22543129e-02 -1.46196854e+00 5.77904999e-01 1.99363396e-01 -7.67549992e-01 -6.35113835e-01 8.48745346e-01 8.23951587e-02 -5.37339866e-01 2.27061063e-01 -1.56857535e-01 -4.69767570e-01 -2.63843864e-01 5.20224750e-01 1.43145591e-01 1.15415966e-02 -5.44384658e-01 -5.19921184e-01 5.16209602e-01 -3.64211708e-01 1.12402558e-01 1.70803440e+00 3.62031788e-01 -3.69754314e-01 -2.20247611e-01 1.21779895e+00 1.57514185e-01 -1.10459161e+00 -2.57976621e-01 5.90610564e-01 -5.22592068e-01 -2.78128505e-01 -9.98610139e-01 -1.55805016e+00 7.11232007e-01 1.35434702e-01 6.29255235e-01 1.02988100e+00 1.96858868e-01 1.26923168e+00 4.48429465e-01 5.24291277e-01 -1.07547557e+00 2.57649809e-01 4.36248660e-01 8.56596291e-01 -1.22614932e+00 -3.69215339e-01 -1.51194155e-01 -8.59996140e-01 9.44130421e-01 5.36028147e-01 1.81439608e-01 6.87951505e-01 4.53208596e-01 6.00725353e-01 -3.44284892e-01 -9.04570341e-01 5.58283806e-01 -2.21211061e-01 5.15235484e-01 5.18480420e-01 -3.28435078e-02 -2.91218281e-01 1.22552621e+00 -2.14323640e-01 -3.22304964e-01 6.23987198e-01 7.65648961e-01 -1.97362244e-01 -1.62724113e+00 6.80698827e-02 4.15420264e-01 -1.04259789e+00 -5.10848999e-01 -5.71644068e-01 4.40189987e-03 5.29615171e-02 1.22531343e+00 -4.56892788e-01 -1.06991601e+00 7.21113905e-02 2.69595739e-02 -3.44249427e-01 -7.89920628e-01 -1.25931752e+00 -9.52908769e-02 9.85204652e-02 -6.63395822e-01 -1.63901821e-01 -8.11581850e-01 -1.12478244e+00 -1.82938740e-01 -4.18377340e-01 2.96630949e-01 6.01676643e-01 1.12069952e+00 4.67751503e-01 1.74614027e-01 1.39297903e+00 -3.80373672e-02 -1.11462116e+00 -1.23859692e+00 -4.28943336e-01 4.84973252e-01 3.34976539e-02 -1.29685715e-01 -4.69394743e-01 -4.54006754e-02]
[6.83836030960083, 7.345325946807861]
588f1a80-5f74-41e7-be54-9e749e4ff8a9
confidence-estimation-for-knowledge-base
null
null
https://aclanthology.org/R13-1051
https://aclanthology.org/R13-1051.pdf
Confidence Estimation for Knowledge Base Population
null
['Xiang Li', 'Ralph Grishman']
2013-09-01
confidence-estimation-for-knowledge-base-1
https://aclanthology.org/R13-1051
https://aclanthology.org/R13-1051.pdf
ranlp-2013-9
['knowledge-base-population']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.507970333099365, 3.5807580947875977]
8e85a784-fd63-4712-8f57-cba955109769
deep-learning-framework-for-real-time-fetal
2205.01675
null
https://arxiv.org/abs/2205.01675v1
https://arxiv.org/pdf/2205.01675v1.pdf
Deep Learning Framework for Real-time Fetal Brain Segmentation in MRI
Fetal brain segmentation is an important first step for slice-level motion correction and slice-to-volume reconstruction in fetal MRI. Fast and accurate segmentation of the fetal brain on fetal MRI is required to achieve real-time fetal head pose estimation and motion tracking for slice re-acquisition and steering. To address this critical unmet need, in this work we analyzed the speed-accuracy performance of a variety of deep neural network models, and devised a symbolically small convolutional neural network that combines spatial details at high resolution with context features extracted at lower resolutions. We used multiple branches with skip connections to maintain high accuracy while devising a parallel combination of convolution and pooling operations as an input downsampling module to further reduce inference time. We trained our model as well as eight alternative, state-of-the-art networks with manually-labeled fetal brain MRI slices and tested on two sets of normal and challenging test cases. Experimental results show that our network achieved the highest accuracy and lowest inference time among all of the compared state-of-the-art real-time segmentation methods. We achieved average Dice scores of 97.99\% and 84.04\% on the normal and challenging test sets, respectively, with an inference time of 3.36 milliseconds per image on an NVIDIA GeForce RTX 2080 Ti. Code, data, and the trained models are available at https://github.com/bchimagine/real_time_fetal_brain_segmentation.
['Ali Gholipour', 'Deniz Erdogmus', 'Davood Karimi', 'Razieh Faghihpirayesh']
2022-05-02
null
null
null
null
['head-pose-estimation']
['computer-vision']
[ 7.90898949e-02 1.72539920e-01 2.39790052e-01 -8.59382629e-01 -6.83073938e-01 -4.50513870e-01 4.60289791e-02 -1.35233253e-01 -5.48822284e-01 4.90153104e-01 -2.27402281e-02 -4.11688685e-01 -1.94729269e-01 -4.98146325e-01 -6.55970991e-01 -6.59167588e-01 -6.06442511e-01 5.64649343e-01 3.12226593e-01 4.36501980e-01 2.15304881e-01 7.39397585e-01 -9.39062953e-01 3.04210335e-01 9.69984114e-01 9.55574453e-01 1.31101221e-01 8.60513926e-01 2.03251079e-01 6.82748854e-01 -2.84092218e-01 -4.21537459e-01 2.86553591e-01 -2.88578093e-01 -9.32444870e-01 -3.25458854e-01 6.64581597e-01 -6.27979398e-01 -3.70638669e-01 8.40588570e-01 1.04929924e+00 7.05522746e-02 4.05019492e-01 -7.40402699e-01 -2.84349591e-01 5.30322313e-01 -7.17894018e-01 9.05239165e-01 -1.13651520e-02 2.89177634e-02 -1.13650888e-01 -1.05119193e+00 5.96698821e-01 4.99325395e-01 9.46761906e-01 8.23817253e-01 -8.41591477e-01 -1.15769875e+00 -4.63279903e-01 2.21201017e-01 -1.44392622e+00 -6.12616122e-01 4.41172987e-01 -5.64436555e-01 8.96682560e-01 3.74142021e-01 8.22459638e-01 6.17775500e-01 5.34774303e-01 2.65198797e-01 8.49546850e-01 -5.14434762e-02 -1.59941893e-02 -4.50750768e-01 -1.11551166e-01 1.14595711e+00 2.47232318e-01 -1.49869919e-01 -3.51239443e-01 -8.42860565e-02 1.13802683e+00 -1.44506603e-01 -5.25578260e-01 4.00328264e-03 -1.09978616e+00 6.49950981e-01 5.52492023e-01 5.33041418e-01 -4.65927482e-01 2.60341167e-01 4.12483186e-01 -3.60801578e-01 5.33502102e-01 2.12120131e-01 -3.70884269e-01 -2.11904466e-01 -1.32965040e+00 -1.27897859e-02 6.84605420e-01 6.29091799e-01 1.36447087e-01 1.03437997e-01 -2.48547524e-01 6.85841620e-01 1.59580216e-01 2.42766961e-01 7.23054171e-01 -1.22186601e+00 4.32809711e-01 -2.73607690e-02 -3.55532616e-01 -9.58758593e-01 -1.09186614e+00 -5.13182104e-01 -1.02614033e+00 -2.11552624e-02 4.43895102e-01 -3.53089422e-01 -1.35067272e+00 1.50708473e+00 5.18211007e-01 5.40309072e-01 -4.42769468e-01 1.14188790e+00 1.20321620e+00 3.17668587e-01 8.06360133e-03 -4.27171111e-01 1.22706354e+00 -9.18990254e-01 -6.99697733e-01 -5.96727058e-02 6.76304638e-01 -6.47554100e-01 5.92281759e-01 2.92480409e-01 -1.56660914e+00 -1.68072686e-01 -9.61519837e-01 7.29314461e-02 1.54662430e-01 -3.45537290e-02 6.38628960e-01 8.00313830e-01 -1.43178570e+00 9.58680928e-01 -1.54940701e+00 1.75432518e-01 8.47811282e-01 7.04906404e-01 -5.18831551e-01 -1.03365496e-01 -9.48948145e-01 9.79425609e-01 3.75836641e-02 3.61136079e-01 -8.27796221e-01 -1.22425747e+00 -8.10163260e-01 1.81359779e-02 1.00446515e-01 -5.22023141e-01 1.43555832e+00 -6.32546127e-01 -1.32359064e+00 1.01742780e+00 -8.60427991e-02 -2.12550983e-01 7.27971792e-01 8.40119086e-03 -3.85110646e-01 5.57602644e-01 1.09217197e-01 6.81914389e-01 5.38609147e-01 -7.42900133e-01 -2.39860326e-01 -5.05571485e-01 -5.92451632e-01 -5.57038642e-04 8.14199075e-02 3.02731961e-01 -5.38473070e-01 -5.10800242e-01 5.65282464e-01 -8.91859472e-01 -1.17016181e-01 1.64674610e-01 -7.73646589e-03 2.77102053e-01 3.70967090e-01 -1.34762406e+00 1.12137556e+00 -1.58994150e+00 -4.55848813e-01 2.51635700e-01 5.61745346e-01 5.35084665e-01 1.08977869e-01 -4.18762714e-01 -4.30302531e-01 3.96837778e-02 -5.67575693e-01 -1.44728273e-01 -7.35045373e-01 -1.47196248e-01 5.79691231e-01 9.73227620e-01 -1.45871595e-01 9.27845597e-01 -8.66943777e-01 -7.51987994e-01 2.25298181e-01 8.27523172e-01 -7.13712811e-01 3.40331614e-01 7.15030968e-01 9.13811088e-01 -1.81915238e-01 5.82416713e-01 9.65240777e-01 -1.40177846e-01 2.38817617e-01 -2.32921734e-01 -1.26416802e-01 1.46565929e-01 -8.25795770e-01 2.09246802e+00 -3.38395000e-01 6.45941794e-01 2.16969460e-01 -8.23084235e-01 5.37163079e-01 6.92356408e-01 7.70437956e-01 -8.00690711e-01 3.76963198e-01 4.20989513e-01 4.48315859e-01 -9.62172329e-01 -1.07651725e-01 -1.81419790e-01 5.60020804e-01 3.52483660e-01 8.84022489e-02 -6.30170181e-02 7.46135563e-02 -1.40802875e-01 9.94053185e-01 5.44790775e-02 1.06880283e-02 -6.00713432e-01 2.71084189e-01 -3.54373127e-01 7.42281795e-01 5.63284695e-01 -3.85727793e-01 1.23865986e+00 1.96991339e-01 -7.93038011e-01 -8.20846736e-01 -9.11014080e-01 -5.68632603e-01 8.32157910e-01 -2.34692112e-01 1.40602544e-01 -1.26256895e+00 -5.54082930e-01 -3.65216196e-01 5.19587159e-01 -7.94262409e-01 2.16378689e-01 -1.17304981e+00 -1.08013499e+00 8.41737688e-01 8.44764054e-01 3.36792260e-01 -1.07399225e+00 -1.07175958e+00 4.00351137e-01 -2.50888675e-01 -9.39918518e-01 -6.09613597e-01 4.07710252e-03 -9.55964029e-01 -1.06045103e+00 -9.40605521e-01 -7.19304264e-01 9.13005173e-01 -1.74741358e-01 1.08379459e+00 2.94644058e-01 -5.74962676e-01 -3.17136973e-01 1.08528160e-01 -2.19836652e-01 -2.24868372e-01 -5.26293293e-02 8.10426176e-02 -4.25570041e-01 -2.72660464e-01 -8.17116022e-01 -1.17847359e+00 2.98923701e-02 -7.45097220e-01 4.04790789e-01 2.93388724e-01 7.26606607e-01 3.55475217e-01 -7.20553279e-01 2.91670054e-01 -7.52740264e-01 2.51830101e-01 -5.77381670e-01 -7.22679138e-01 1.30236700e-01 -6.19673729e-01 -2.68120795e-01 4.14031744e-01 -3.24755728e-01 -1.02058911e+00 -1.10873386e-01 -6.14060938e-01 -4.28897768e-01 -4.46697287e-02 4.37196732e-01 3.25000614e-01 -3.94745022e-01 5.87132990e-01 7.37901181e-02 6.61026761e-02 -3.17853957e-01 -2.14181975e-01 4.03667301e-01 8.02114427e-01 -3.62642169e-01 1.37973174e-01 5.47875404e-01 1.51770934e-01 -3.97332668e-01 -4.99001294e-01 -9.46014896e-02 -8.33681643e-01 -4.03391629e-01 1.09813511e+00 -5.66702843e-01 -5.98527670e-01 4.17959481e-01 -1.29836011e+00 -5.18605232e-01 2.19766170e-01 6.96370423e-01 -3.68592083e-01 2.27235854e-01 -9.77780759e-01 -2.82442153e-01 -1.00395334e+00 -1.47580242e+00 4.55642700e-01 3.75778645e-01 -1.42250106e-01 -8.06995571e-01 -2.50173863e-02 3.84434670e-01 7.56054997e-01 5.62411785e-01 5.36861002e-01 -7.22905695e-01 -1.95972398e-01 -1.12190634e-01 -4.10027921e-01 3.16094197e-02 4.37110849e-02 9.49109718e-03 -1.01062274e+00 -2.92358607e-01 2.11675510e-01 2.12533355e-01 4.13573354e-01 8.88845742e-01 1.71991646e+00 -2.39023849e-01 -1.84361815e-01 1.28115833e+00 1.16261947e+00 5.89205205e-01 4.87644315e-01 4.53639515e-02 9.01091695e-01 4.00008112e-01 2.79829234e-01 3.72413397e-01 4.78695750e-01 3.43288094e-01 3.96922857e-01 -2.49608561e-01 -1.95355058e-01 2.98151314e-01 -5.05020201e-01 1.06788516e+00 -4.78923947e-01 7.09468052e-02 -1.63276541e+00 6.17886603e-01 -1.39925575e+00 -6.15017354e-01 -1.76222384e-01 2.19574714e+00 9.80696261e-01 -1.43583417e-01 -2.23533005e-01 -1.49111882e-01 7.61149406e-01 1.48361018e-02 -3.33542615e-01 -4.50525910e-01 5.64970255e-01 5.88265777e-01 6.36070788e-01 4.23339933e-01 -1.02968049e+00 5.40767670e-01 5.61862373e+00 5.42218566e-01 -1.59459746e+00 7.40803838e-01 9.56453145e-01 -2.54981458e-01 1.88744187e-01 -5.04364192e-01 -2.19789684e-01 5.04505217e-01 1.13275480e+00 1.54537439e-01 3.67466122e-01 6.35398924e-01 3.20118666e-01 -1.40027227e-02 -9.65081871e-01 8.94864619e-01 -5.17584383e-03 -1.66030324e+00 -5.63278973e-01 -2.11723283e-01 7.55929887e-01 4.26570565e-01 -3.72998655e-01 -1.83757961e-01 -3.90243828e-01 -1.35659826e+00 5.79009593e-01 4.83835965e-01 1.19987965e+00 -7.95124650e-01 9.01986003e-01 2.27943093e-01 -8.97586405e-01 3.10562342e-01 -1.77845471e-02 1.18246101e-01 6.54032081e-02 3.55590820e-01 -1.02067447e+00 2.82054543e-01 1.09638977e+00 2.01549694e-01 -2.30918512e-01 1.22640920e+00 -2.79589873e-02 6.22075081e-01 -3.20216089e-01 4.16766614e-01 1.69194013e-01 1.68896809e-01 3.22956890e-01 1.61715889e+00 3.46519738e-01 5.01057267e-01 -3.67401630e-01 7.00987577e-01 -1.16177976e-01 3.93274687e-02 -9.30967778e-02 5.23088157e-01 4.36291695e-01 1.51838851e+00 -1.09817135e+00 -3.92983407e-01 -3.37611884e-01 6.35973155e-01 3.40300411e-01 1.19285583e-01 -1.21357667e+00 -3.12429458e-01 3.48039299e-01 4.27337050e-01 2.74594873e-02 -1.03148140e-01 -6.46800697e-01 -9.78663802e-01 -5.79380756e-03 -3.78746271e-01 2.43254364e-01 -6.34170890e-01 -7.04297066e-01 7.41080940e-01 7.25650564e-02 -8.83514106e-01 -2.82837123e-01 -3.07886422e-01 -9.53987062e-01 8.61697137e-01 -1.37326646e+00 -8.23945284e-01 -6.61268532e-01 4.49993461e-01 3.10346842e-01 2.51880109e-01 8.48813176e-01 7.67291367e-01 -5.46355128e-01 8.36287022e-01 -2.00363159e-01 5.00812054e-01 3.97210360e-01 -9.20099318e-01 4.78878707e-01 1.01602745e+00 -4.89158720e-01 7.10508108e-01 3.58632147e-01 -6.87416077e-01 -1.12665558e+00 -1.04804468e+00 6.10832989e-01 3.87905911e-02 2.59095132e-01 9.26136319e-03 -1.19905651e+00 8.02238584e-01 1.24551430e-01 7.76395202e-01 6.27551377e-01 -5.94753027e-01 2.38595292e-01 5.19186556e-02 -1.48161328e+00 3.28551441e-01 1.04515290e+00 9.62954238e-02 -4.04127061e-01 2.32407898e-01 4.33776438e-01 -1.39255381e+00 -1.12772346e+00 8.56907487e-01 8.30821574e-01 -1.09912753e+00 9.25520599e-01 -3.12026441e-01 6.33198023e-01 1.29584176e-02 3.97233188e-01 -9.87046421e-01 -3.58751714e-01 -4.77863789e-01 -1.71374545e-01 7.79954255e-01 3.51795465e-01 -7.32834816e-01 9.15228009e-01 9.93360579e-01 -5.03479838e-01 -1.25409019e+00 -1.39796448e+00 -2.03371957e-01 2.15386271e-01 -5.40310621e-01 5.70140898e-01 1.13041985e+00 -2.10721478e-01 -3.99374157e-01 8.77094194e-02 3.03125501e-01 6.42350197e-01 -8.70866999e-02 -1.96051970e-02 -7.22506046e-01 2.08889738e-01 -4.81758028e-01 -3.79043967e-01 -3.46937060e-01 -9.15073380e-02 -1.05826008e+00 3.17756087e-04 -1.58949292e+00 2.86696345e-01 -3.42669785e-01 -2.51231343e-01 6.27677798e-01 -6.57860562e-02 6.20151818e-01 4.14163806e-02 -9.12145525e-02 -1.30246028e-01 -1.68135781e-02 1.57417810e+00 3.37621361e-01 -6.25035092e-02 3.94534245e-02 -3.00543934e-01 9.35135722e-01 8.08850408e-01 -6.62545860e-01 -9.99026075e-02 -8.24117601e-01 -2.52946854e-01 6.83221996e-01 1.82231307e-01 -1.16223109e+00 3.07295501e-01 3.76011841e-02 9.20881689e-01 -5.67426503e-01 1.71610668e-01 -5.12148142e-01 2.33453125e-01 7.52932906e-01 -1.55508950e-01 2.13801771e-01 2.63421148e-01 -4.95379955e-01 6.95418641e-02 -1.73250020e-01 1.15894449e+00 -2.76755452e-01 -3.39982599e-01 7.52611995e-01 -2.28203550e-01 1.29778042e-01 8.48425925e-01 -2.99822658e-01 -6.28971383e-02 -1.53330609e-01 -9.27045763e-01 1.42428741e-01 7.54403323e-02 2.45718926e-01 9.35760617e-01 -1.03222179e+00 -9.57369387e-01 3.09281379e-01 -5.00986755e-01 3.58473033e-01 6.14999413e-01 1.50408077e+00 -1.40754688e+00 2.74919480e-01 -4.37972784e-01 -7.27633059e-01 -1.16841316e+00 4.38412502e-02 8.28524172e-01 -9.89732891e-02 -7.70667374e-01 1.18025303e+00 1.13305664e-02 -4.44180936e-01 3.12114190e-02 -4.38110948e-01 -9.13902745e-02 -2.24298328e-01 6.51516616e-01 6.03724003e-01 5.06898105e-01 -6.72512233e-01 -7.64900506e-01 6.24048889e-01 1.29240468e-01 5.99342510e-02 1.51144505e+00 -2.97472645e-02 -2.87307233e-01 -2.19021782e-01 1.32828057e+00 -2.04289526e-01 -1.19308567e+00 1.92835093e-01 -1.37659699e-01 -6.28890276e-01 1.37586534e-01 -8.34652662e-01 -1.77161992e+00 1.03807259e+00 1.07999802e+00 -1.93303391e-01 1.19134092e+00 -2.25015953e-01 8.07482123e-01 -3.26844215e-01 2.17180714e-01 -7.11873472e-01 -3.68466109e-01 3.05874705e-01 9.13879216e-01 -1.29379928e+00 1.60929918e-01 -3.50293607e-01 -4.61720794e-01 1.25868654e+00 9.08373356e-01 -1.62611738e-01 6.66770041e-01 6.55019343e-01 2.67786056e-01 -3.15877289e-01 -2.07850739e-01 5.13994157e-01 5.01911521e-01 3.63980651e-01 8.62217546e-01 1.66084290e-01 -5.18877447e-01 6.87672436e-01 -2.81515300e-01 2.62694091e-01 4.30151194e-01 1.02395356e+00 -1.06281586e-01 -3.06839406e-01 -2.75058061e-01 6.35108054e-01 -1.24340904e+00 -2.31189638e-01 6.07573628e-01 6.86132967e-01 2.71240979e-01 5.12018502e-01 1.25506863e-01 -7.66537115e-02 7.24859983e-02 -7.63745829e-02 6.58686697e-01 -4.74218398e-01 -9.13072944e-01 4.46162969e-02 -4.41576950e-02 -9.86552000e-01 -7.78848156e-02 -6.21997595e-01 -1.75930750e+00 -2.88439810e-01 -6.06744736e-02 -1.46624595e-01 1.14499307e+00 8.95515442e-01 3.27575654e-01 7.75594592e-01 3.85619581e-01 -1.38194132e+00 -6.99803978e-02 -1.11037374e+00 -2.86840916e-01 9.06430632e-02 4.13217336e-01 -3.75630111e-01 -4.15015630e-02 -8.60418826e-02]
[14.019425392150879, -2.402336359024048]
d7d5452c-cb3e-4ff7-a7cc-73d9d0c0f3d6
pointit-a-fast-tracking-framework-based-on-3d
1902.06379
null
http://arxiv.org/abs/1902.06379v1
http://arxiv.org/pdf/1902.06379v1.pdf
PointIT: A Fast Tracking Framework Based on 3D Instance Segmentation
Recently most popular tracking frameworks focus on 2D image sequences. They seldom track the 3D object in point clouds. In this paper, we propose PointIT, a fast, simple tracking method based on 3D on-road instance segmentation. Firstly, we transform 3D LiDAR data into the spherical image with the size of 64 x 512 x 4 and feed it into instance segment model to get the predicted instance mask for each class. Then we use MobileNet as our primary encoder instead of the original ResNet to reduce the computational complexity. Finally, we extend the Sort algorithm with this instance framework to realize tracking in the 3D LiDAR point cloud data. The model is trained on the spherical images dataset with the corresponding instance label masks which are provided by KITTI 3D Object Track dataset. According to the experiment results, our network can achieve on Average Precision (AP) of 0.617 and the performance of multi-tracking task has also been improved.
['Yu-An Wang', 'Yang Yu', 'Ming Liu']
2019-02-18
null
null
null
null
['3d-instance-segmentation-1']
['computer-vision']
[-1.89425558e-01 -3.58056575e-01 -3.93184721e-01 -2.67527908e-01 -5.36538363e-01 -4.67031300e-01 3.71442378e-01 -5.50572515e-01 -4.99109656e-01 4.01964337e-01 -4.03388232e-01 -3.52627307e-01 9.58524495e-02 -8.40713263e-01 -1.04850340e+00 -3.34604979e-01 2.94710308e-01 8.59563410e-01 9.07720625e-01 3.00945491e-01 2.18413115e-01 7.54343629e-01 -1.60366488e+00 -2.04236791e-01 8.58581245e-01 1.22352934e+00 6.29927695e-01 5.18779457e-01 -7.24191666e-01 3.15802932e-01 -4.00727421e-01 -3.00403923e-01 5.85437655e-01 4.84287173e-01 -4.53031719e-01 2.09969096e-02 7.85532117e-01 -4.90691394e-01 -4.02498871e-01 1.25504589e+00 5.58268070e-01 -1.44852102e-01 4.04084861e-01 -1.39963603e+00 -5.38651764e-01 4.78383601e-01 -9.83512282e-01 3.01162571e-01 -1.92302242e-01 1.25600308e-01 4.66682822e-01 -8.98184359e-01 5.87264717e-01 1.44792986e+00 8.84914994e-01 6.18108749e-01 -5.93283713e-01 -1.29682207e+00 2.99718767e-01 2.31418625e-01 -1.57482708e+00 -8.53492469e-02 5.28519034e-01 -4.61973131e-01 7.87871122e-01 8.41784999e-02 8.81301641e-01 4.48338956e-01 8.54409933e-02 1.12334740e+00 7.20522404e-01 2.59883821e-01 -4.15944993e-01 -1.27048656e-01 1.67702317e-01 7.14838684e-01 3.19049239e-01 3.56599599e-01 2.76196338e-02 3.44856769e-01 9.06994462e-01 4.03365105e-01 -6.20889813e-02 -4.18847442e-01 -1.35532916e+00 3.53049666e-01 7.58860350e-01 1.74970552e-01 -2.79673606e-01 5.26400447e-01 3.66457015e-01 -4.80896756e-02 6.48877382e-01 -4.61174726e-01 -6.58876538e-01 1.47299260e-01 -1.04129839e+00 1.63742796e-01 2.22224578e-01 1.67161691e+00 8.19336832e-01 1.36468792e-02 -1.42819196e-01 5.34874320e-01 6.97713137e-01 1.09081638e+00 1.76895827e-01 -9.33889508e-01 5.66591263e-01 9.16451812e-01 2.86448181e-01 -6.27238154e-01 -3.13334674e-01 -5.85494041e-01 -4.86287475e-01 2.97270924e-01 1.19389854e-01 -2.50831693e-02 -1.30897331e+00 1.29909754e+00 7.93284416e-01 9.66470182e-01 -2.43016735e-01 9.35050547e-01 9.90193367e-01 7.00141251e-01 5.03722280e-02 -6.61617219e-02 1.19606185e+00 -1.18470109e+00 -4.64014232e-01 -5.49583696e-02 5.72244525e-01 -6.62406862e-01 7.41711318e-01 -6.39500022e-02 -1.01985514e+00 -1.10483003e+00 -8.98359716e-01 -5.15158065e-02 -3.22974294e-01 4.76317108e-01 1.91943079e-01 5.06955981e-01 -8.43109131e-01 4.22910988e-01 -1.01993513e+00 -1.56142488e-01 8.10165882e-01 6.39223814e-01 -5.55242971e-02 -9.58662704e-02 -7.57787228e-01 8.27561200e-01 6.17678761e-01 2.20123291e-01 -7.95327365e-01 -8.87122333e-01 -5.79762220e-01 -9.49828029e-02 3.37742716e-01 -7.03382969e-01 1.22662163e+00 -2.28586525e-01 -1.20899880e+00 1.00522673e+00 -2.68805385e-01 -5.03166020e-01 7.41301835e-01 -5.70275664e-01 -3.46357286e-01 -2.88534999e-01 4.92820531e-01 1.03788149e+00 6.69830143e-01 -1.18993819e+00 -1.38357937e+00 -5.20224750e-01 -3.56395662e-01 1.84550315e-01 3.00715029e-01 -1.55666918e-01 -1.05534184e+00 2.38875002e-02 2.67105967e-01 -1.05282903e+00 -1.76598385e-01 4.76212464e-02 -4.82830793e-01 -5.79922855e-01 1.37352729e+00 -1.70268327e-01 9.53490436e-01 -2.19439244e+00 -3.42426270e-01 2.70921774e-02 2.23498687e-01 5.06211817e-01 -3.09671219e-02 -3.32555383e-01 1.04276076e-01 3.57549312e-03 3.21736299e-02 -4.20005113e-01 4.48314399e-02 1.79646581e-01 -3.43405575e-01 4.86929983e-01 -2.70461589e-02 1.30798733e+00 -7.44169712e-01 -9.10199106e-01 6.40371382e-01 4.86681700e-01 -3.36464375e-01 -3.26612145e-02 -3.81633699e-01 4.66829211e-01 -7.79727101e-01 5.74827373e-01 1.37512290e+00 -3.48494560e-01 -6.85378492e-01 -2.20507607e-01 -5.86632371e-01 2.20351443e-02 -1.34109211e+00 1.77625084e+00 -1.28637910e-01 5.41407466e-01 -2.61021614e-01 -3.83371115e-01 1.10347152e+00 5.89776374e-02 7.64047742e-01 -4.40451741e-01 2.24989980e-01 3.06210011e-01 -2.10765004e-01 -3.44206661e-01 5.96235871e-01 1.33941755e-01 8.95492062e-02 1.54638603e-01 -2.31620565e-01 -2.02789471e-01 -1.46752968e-02 -1.74413890e-01 6.87452734e-01 5.17265201e-01 -4.64236468e-01 8.79775360e-02 7.68539667e-01 5.00878870e-01 6.26750529e-01 6.72974885e-01 -3.72388810e-01 6.05725706e-01 -2.70293862e-01 -4.12911654e-01 -1.13107646e+00 -9.72098887e-01 -4.13946748e-01 5.28455496e-01 8.39344084e-01 -3.50077711e-02 -5.72055936e-01 -7.67379999e-01 3.11559796e-01 4.50503409e-01 -2.26930022e-01 1.44359440e-01 -8.13396096e-01 -2.18865588e-01 4.90593463e-01 7.31805801e-01 9.96004403e-01 -9.69295204e-01 -4.91474569e-01 2.19604611e-01 1.08551025e-01 -1.30716276e+00 -6.33582711e-01 -1.85052738e-01 -1.00086832e+00 -1.08956563e+00 -6.65632486e-01 -9.36117709e-01 3.55302602e-01 6.30862653e-01 7.39385605e-01 2.18085989e-01 1.25598490e-01 2.17435062e-01 -3.26138809e-02 -6.45434499e-01 2.11250857e-01 3.08733732e-01 -3.42420954e-03 -3.78983408e-01 7.88273454e-01 -2.41458908e-01 -6.79748714e-01 4.57359165e-01 -4.77217317e-01 7.91898444e-02 4.36411798e-01 2.90764481e-01 9.82273340e-01 7.12880269e-02 1.04201883e-01 -6.83402419e-01 9.57265049e-02 -1.84767962e-01 -1.06557727e+00 3.90764587e-02 -4.06995624e-01 -3.17995310e-01 2.35198177e-02 -4.82630372e-01 -8.37533891e-01 5.20615697e-01 -2.41621032e-01 -1.22085500e+00 -2.04390883e-01 -2.46818420e-02 -1.26277115e-02 -1.64543927e-01 1.37685075e-01 4.30799335e-01 -1.13125294e-01 -4.81592536e-01 4.70202982e-01 8.15298617e-01 6.53581142e-01 -3.21225375e-01 1.18163621e+00 5.42694330e-01 -1.37757108e-01 -3.63039851e-01 -1.12306094e+00 -5.21129429e-01 -8.61476123e-01 -2.70416737e-01 1.23735523e+00 -1.31708813e+00 -1.13342381e+00 4.36959565e-01 -1.33817303e+00 -3.42369676e-01 -2.13066161e-01 6.72918737e-01 -5.70541561e-01 1.71803132e-01 -5.56312740e-01 -7.64993906e-01 -5.73443949e-01 -1.21097565e+00 1.61884117e+00 5.46565592e-01 5.58629155e-01 -5.34924269e-01 3.38511206e-02 1.81060970e-01 1.61886290e-01 -9.32194293e-02 2.42123812e-01 -4.06179786e-01 -1.34236526e+00 -2.45748594e-01 -6.94063425e-01 4.47468497e-02 -1.19052276e-01 1.06348105e-01 -7.35859752e-01 -9.26984772e-02 -9.72638801e-02 2.54395783e-01 7.97095716e-01 6.31103218e-01 1.38574815e+00 3.26316923e-01 -9.84209418e-01 8.28032672e-01 1.43683004e+00 3.86280626e-01 4.44658041e-01 4.00979042e-01 1.05886292e+00 8.10277984e-02 1.27675164e+00 -8.63359869e-02 6.02634072e-01 5.99381745e-01 6.96610153e-01 9.53359827e-02 -2.27792650e-01 -4.53473091e-01 3.95434536e-02 7.60463119e-01 -2.39672046e-02 -9.19606537e-03 -8.93587589e-01 4.91739571e-01 -1.88347697e+00 -1.03310108e+00 -7.40583658e-01 1.91901004e+00 3.72872621e-01 4.67985958e-01 1.53451962e-02 -2.76603103e-01 1.18413615e+00 8.28093588e-02 -6.56347692e-01 3.87139440e-01 2.46569872e-01 4.51922528e-02 9.90780234e-01 5.24095356e-01 -1.14722908e+00 1.51846838e+00 5.33871841e+00 9.84437466e-01 -1.25483549e+00 1.39559269e-01 4.71871570e-02 -1.01130269e-02 1.96541101e-02 -9.75574851e-02 -1.62992549e+00 9.18366432e-01 6.67843044e-01 -7.64264539e-02 -3.72902751e-02 1.10079789e+00 7.11067244e-02 2.54528522e-01 -7.24348307e-01 1.10727012e+00 -4.17871833e-01 -1.35329926e+00 -3.67667191e-02 1.40062988e-01 5.92130423e-01 6.75806940e-01 2.44783852e-02 4.99478519e-01 2.86817074e-01 -7.52804637e-01 8.80937397e-01 4.39077795e-01 8.89652491e-01 -7.96169639e-01 5.46212494e-01 8.08640420e-01 -1.72448289e+00 2.25921422e-02 -8.90299916e-01 3.78562808e-01 5.24076641e-01 3.95148009e-01 -8.47696722e-01 4.85479802e-01 9.47049081e-01 1.04152119e+00 -2.83302456e-01 1.72903526e+00 -1.08098872e-01 3.24786127e-01 -7.56859004e-01 3.45256403e-02 3.78287196e-01 -3.09534997e-01 6.09281301e-01 8.77910316e-01 6.36488795e-01 1.65270269e-01 3.52965504e-01 8.20396304e-01 -2.57954728e-02 -2.16337606e-01 -7.24917233e-01 4.35056269e-01 9.29487646e-01 1.28007138e+00 -6.49556935e-01 -5.94485343e-01 -3.67883682e-01 6.16101503e-01 5.93294390e-03 -2.61872094e-02 -1.25039113e+00 -2.96807289e-01 4.59087640e-01 1.04826376e-01 8.66593421e-01 -3.42205822e-01 -2.05655679e-01 -9.26374614e-01 -1.15672037e-01 -1.90050781e-01 2.52036545e-02 -1.22629941e+00 -1.03174019e+00 5.75668156e-01 -2.11140022e-01 -1.61140025e+00 2.55251765e-01 -7.16478825e-01 -4.95741338e-01 9.54503655e-01 -1.84390604e+00 -1.43233228e+00 -6.15145206e-01 7.05695033e-01 6.41936779e-01 8.86776596e-02 -9.79308970e-03 8.83861959e-01 -6.28204525e-01 2.64932901e-01 -8.07301551e-02 2.01115116e-01 4.91973609e-01 -1.06076729e+00 8.74933243e-01 5.58475018e-01 -2.66303867e-02 4.34641600e-01 2.78842330e-01 -1.08348560e+00 -1.12014043e+00 -1.70392859e+00 7.58182049e-01 -9.17488217e-01 4.29557055e-01 -1.61171183e-01 -9.13229346e-01 1.06996775e+00 -1.52541503e-01 2.90754974e-01 -1.38548966e-02 -3.42892438e-01 1.26069218e-01 -1.69918478e-01 -1.09045470e+00 3.83278102e-01 1.67258334e+00 -2.37093911e-01 -5.26917338e-01 3.47373605e-01 1.23151231e+00 -1.18464768e+00 -9.37988818e-01 6.22375965e-01 2.74199247e-01 -4.72585976e-01 1.24488521e+00 -3.17853004e-01 -3.14945430e-02 -1.10480082e+00 4.70854901e-03 -7.18701601e-01 -3.27869505e-01 -2.99781002e-02 -1.91618949e-01 1.13320577e+00 2.04327360e-01 -4.96398449e-01 1.37117505e+00 1.46912143e-01 -5.99408746e-01 -5.81933677e-01 -9.80310619e-01 -7.20038772e-01 -1.34748995e-01 -7.20271349e-01 1.27406490e+00 6.33780658e-01 -1.03734922e+00 3.29302728e-01 -1.82436198e-01 4.90642637e-01 8.59357357e-01 4.94006783e-01 1.17526519e+00 -1.55299973e+00 4.64152426e-01 -2.58574814e-01 -4.41870809e-01 -1.92531002e+00 2.50378191e-01 -1.07874680e+00 3.37625258e-02 -1.48982954e+00 -2.56287865e-03 -9.16991055e-01 -8.70554373e-02 2.40547583e-01 -1.53208777e-01 2.57727653e-01 3.49709392e-01 5.19003332e-01 -8.77547801e-01 5.28419197e-01 1.64409614e+00 -2.94281512e-01 -1.86851919e-01 5.26018500e-01 -2.57884353e-01 5.35864651e-01 5.86984754e-01 -5.62470376e-01 -2.84456372e-01 -8.36001456e-01 -1.82275191e-01 1.03132904e-01 3.88996214e-01 -1.29024255e+00 5.87727845e-01 -3.43761742e-02 7.30258465e-01 -1.95975065e+00 3.96758914e-01 -1.34546387e+00 3.93729240e-01 6.26720667e-01 1.36581585e-01 1.23474307e-01 3.01292628e-01 6.36163652e-01 1.92285180e-02 -1.22098140e-01 6.76555336e-01 -3.98379341e-02 -8.82965088e-01 1.19440520e+00 4.31261122e-01 -2.40839779e-01 1.23854673e+00 -6.63738489e-01 -3.68423611e-01 3.74199867e-01 -6.84988201e-01 8.45285654e-01 5.76068163e-01 6.09268248e-01 5.79318821e-01 -1.79506433e+00 -5.07300317e-01 1.44548997e-01 -1.84496000e-01 6.35267258e-01 2.86114693e-01 8.59891593e-01 -6.15114570e-01 5.67047596e-01 -7.88443014e-02 -1.42206860e+00 -1.21369195e+00 6.97915196e-01 5.86767614e-01 1.05753727e-01 -8.92455876e-01 7.63007462e-01 2.19465479e-01 -7.88302660e-01 1.18930563e-01 -3.35352570e-01 -3.27213556e-01 -2.45402187e-01 2.86931932e-01 2.97096163e-01 -2.68433124e-01 -7.86847532e-01 -5.41803420e-01 1.30252767e+00 -1.21748656e-01 1.69607639e-01 1.34340465e+00 -2.35439196e-01 9.97208282e-02 2.03954071e-01 1.12153137e+00 -1.83120161e-01 -1.58319378e+00 -2.35090449e-01 -8.29267967e-03 -7.09379494e-01 -9.78364125e-02 -2.17655987e-01 -1.42185521e+00 9.77748394e-01 1.03568029e+00 1.49479136e-01 5.14401913e-01 1.12814613e-01 1.19284582e+00 1.35863066e-01 5.38121998e-01 -7.51627922e-01 -3.85148585e-01 8.30610156e-01 2.02287093e-01 -1.12245798e+00 -1.99673623e-01 -4.66464460e-01 -3.06457013e-01 7.23032832e-01 9.83496308e-01 -5.38092673e-01 6.92765951e-01 2.33342499e-01 1.36145890e-01 -3.79303336e-01 -4.08282906e-01 -5.65958679e-01 1.77465066e-01 7.31364369e-01 -1.18729367e-04 -7.59007111e-02 -3.55053134e-02 9.48601365e-02 -3.28027129e-01 3.45344931e-01 3.08130179e-02 5.87627888e-01 -8.38363886e-01 -9.15861189e-01 -4.48723912e-01 5.47031105e-01 -2.56520629e-01 1.93399578e-01 1.66344434e-01 1.01993060e+00 4.21342075e-01 4.46729183e-01 4.79995817e-01 -5.14263093e-01 6.03354394e-01 -1.98759615e-01 4.41103041e-01 -5.63168764e-01 -3.92059416e-01 -1.06988857e-02 -3.76965225e-01 -4.63300705e-01 -4.21698689e-01 -7.05358326e-01 -1.65295458e+00 -4.13643807e-01 -8.15600395e-01 1.80000022e-01 7.54665852e-01 7.99986064e-01 5.01822293e-01 5.21628022e-01 5.55188954e-01 -1.10287917e+00 -1.97234184e-01 -7.84288108e-01 -3.63634974e-01 6.85140863e-03 2.93068290e-01 -9.32239771e-01 1.76464930e-01 -2.31448159e-01]
[6.643890857696533, -2.3702685832977295]
7f1c506f-fe82-4eec-99c1-58ae65e565c9
bbs-net-rgb-d-salient-object-detection-with-a
2007.02713
null
https://arxiv.org/abs/2007.02713v3
https://arxiv.org/pdf/2007.02713v3.pdf
Bifurcated backbone strategy for RGB-D salient object detection
Multi-level feature fusion is a fundamental topic in computer vision. It has been exploited to detect, segment and classify objects at various scales. When multi-level features meet multi-modal cues, the optimal feature aggregation and multi-modal learning strategy become a hot potato. In this paper, we leverage the inherent multi-modal and multi-level nature of RGB-D salient object detection to devise a novel cascaded refinement network. In particular, first, we propose to regroup the multi-level features into teacher and student features using a bifurcated backbone strategy (BBS). Second, we introduce a depth-enhanced module (DEM) to excavate informative depth cues from the channel and spatial views. Then, RGB and depth modalities are fused in a complementary way. Our architecture, named Bifurcated Backbone Strategy Network (BBS-Net), is simple, efficient, and backbone-independent. Extensive experiments show that BBS-Net significantly outperforms eighteen SOTA models on eight challenging datasets under five evaluation measures, demonstrating the superiority of our approach ($\sim 4 \%$ improvement in S-measure $vs.$ the top-ranked model: DMRA-iccv2019). In addition, we provide a comprehensive analysis on the generalization ability of different RGB-D datasets and provide a powerful training set for future research.
['Ling Shao', 'Deng-Ping Fan', 'Liang Wang', 'Junwei Han', 'Ali Borji', 'Jufeng Yang', 'Yingjie Zhai']
2020-07-06
null
null
null
null
['rgb-d-salient-object-detection']
['computer-vision']
[ 7.21291676e-02 -3.07859719e-01 1.24837393e-02 -4.26604062e-01 -9.67988372e-01 -2.49038920e-01 3.85618001e-01 -1.93208363e-02 -3.35770667e-01 3.09017032e-01 3.87165025e-02 1.56993028e-02 -2.53171504e-01 -7.91107595e-01 -4.77711350e-01 -8.84056151e-01 8.56262520e-02 -2.03234926e-01 7.80181944e-01 -2.03410760e-01 2.06664592e-01 7.05775261e-01 -1.78691983e+00 2.40368679e-01 8.03784132e-01 1.48507392e+00 4.14296806e-01 2.57371336e-01 1.90941785e-02 5.47513723e-01 -1.64380640e-01 -3.11814070e-01 4.90278929e-01 -1.85251325e-01 -5.89646637e-01 3.69102567e-01 4.25222099e-01 -2.37082079e-01 -2.16188550e-01 8.92182112e-01 6.65883660e-01 6.23351559e-02 5.40247381e-01 -1.25369203e+00 -3.92057985e-01 2.58538932e-01 -1.11514628e+00 3.88747960e-01 1.39488250e-01 1.16914757e-01 8.74163449e-01 -1.17716873e+00 4.61834699e-01 1.12266612e+00 5.55647314e-01 2.93305039e-01 -9.20177579e-01 -8.21230710e-01 4.47523504e-01 3.02892774e-01 -1.37103581e+00 -6.00971021e-02 1.21127319e+00 -3.27893466e-01 5.56042790e-01 5.67385443e-02 7.27518260e-01 6.82759345e-01 6.35115206e-02 1.10354114e+00 1.36861789e+00 -2.84529120e-01 1.01042660e-02 -1.53259384e-02 -5.11661507e-02 1.07643759e+00 1.55610844e-01 6.27361536e-02 -8.33572567e-01 1.97857216e-01 7.87351131e-01 2.70559400e-01 -1.61263093e-01 -6.01263642e-01 -1.24629748e+00 7.38435566e-01 9.50677574e-01 1.44304454e-01 -4.91602898e-01 -4.72714640e-02 7.44353160e-02 5.70186675e-02 4.65447992e-01 1.56220391e-01 -4.07241166e-01 2.99097806e-01 -7.75584042e-01 6.66383207e-02 7.98597783e-02 7.44981885e-01 1.01317954e+00 1.04690827e-02 -2.09500678e-02 8.53653550e-01 5.21933794e-01 4.90683496e-01 3.61330569e-01 -9.50808764e-01 3.24516416e-01 9.97062564e-01 -1.89434513e-01 -1.07726479e+00 -7.05158412e-01 -7.24395692e-01 -1.03439248e+00 1.01441264e-01 1.38568386e-01 5.73235825e-02 -9.80450988e-01 1.47181368e+00 6.52806520e-01 2.16577545e-01 -1.18424699e-01 1.06112456e+00 1.17784381e+00 4.41888362e-01 2.79859211e-02 -1.13780320e-01 1.18498886e+00 -9.43577766e-01 -1.13235526e-01 -2.07406148e-01 3.48316997e-01 -5.89128792e-01 9.12515879e-01 3.37389261e-01 -1.01332521e+00 -7.62684345e-01 -1.02278519e+00 -2.48327665e-02 -3.34690154e-01 1.02128521e-01 9.43028688e-01 2.55799234e-01 -9.26723659e-01 2.58026928e-01 -7.19620585e-01 -3.62928770e-02 6.12135351e-01 2.85114974e-01 -3.79822612e-01 -4.02787268e-01 -9.97943461e-01 5.25804222e-01 3.96504134e-01 1.15198500e-01 -8.61586809e-01 -7.15681911e-01 -8.89025331e-01 -3.99416685e-01 4.16078329e-01 -6.25734091e-01 9.13751006e-01 -5.03751099e-01 -1.17689848e+00 8.44369650e-01 -9.25389230e-02 -9.36804414e-02 1.49821386e-01 6.82493895e-02 -1.77651286e-01 5.06251216e-01 1.26233205e-01 9.45395827e-01 8.99711251e-01 -1.55044425e+00 -1.23286867e+00 -6.77740395e-01 1.12646833e-01 3.74915212e-01 -5.03963709e-01 -3.22175473e-01 -6.82182193e-01 -5.80621243e-01 7.36602128e-01 -6.83076739e-01 -4.01840001e-01 2.44107340e-02 -3.28633577e-01 -3.98249060e-01 7.66887844e-01 -3.13609064e-01 1.17805648e+00 -2.26686931e+00 3.41954082e-01 1.35233343e-01 4.50389475e-01 9.69058946e-02 -1.93956997e-02 -3.17178736e-03 8.39018002e-02 -1.32975087e-01 -2.27300614e-01 -3.36687177e-01 -1.33450255e-01 -5.65295992e-03 2.90324613e-02 4.99392092e-01 2.22094342e-01 9.56849337e-01 -8.19888890e-01 -8.04269791e-01 4.58032519e-01 4.76050973e-01 -3.81122679e-01 -1.30176228e-02 8.61996859e-02 3.71160835e-01 -7.22585142e-01 1.26139402e+00 6.91949248e-01 -4.45732862e-01 -3.90322983e-01 -5.78601658e-01 -2.77318448e-01 -2.56435990e-01 -1.26963961e+00 1.86127234e+00 -3.24134976e-01 3.18939030e-01 5.91503829e-03 -9.11239862e-01 9.17654097e-01 -1.27913505e-01 6.28177166e-01 -8.87527525e-01 1.99506000e-01 2.88811713e-01 -2.46617824e-01 -3.53984267e-01 3.84650648e-01 -4.07619774e-02 -2.64797419e-01 1.20119199e-01 8.91165957e-02 -3.49768192e-01 2.00873818e-02 1.62143886e-01 8.65662038e-01 1.15105562e-01 1.77420944e-01 -1.79203376e-02 6.64410889e-01 -1.46007434e-01 7.39626646e-01 4.90681827e-01 -4.27970409e-01 6.39157951e-01 1.17381617e-01 -3.22735518e-01 -4.86035943e-01 -1.06776631e+00 -1.60824612e-01 1.15735269e+00 5.83573580e-01 -2.23008424e-01 -3.78341764e-01 -8.12094927e-01 1.17094338e-01 1.48998767e-01 -6.60129011e-01 -8.39672238e-02 -3.99955839e-01 -7.81367302e-01 1.98458165e-01 6.80969894e-01 9.28469121e-01 -8.25635970e-01 -8.08586657e-01 7.02821370e-03 -2.27284268e-01 -1.12579429e+00 -1.29692554e-01 3.43068242e-01 -9.55563962e-01 -1.04045773e+00 -7.68279374e-01 -8.14905822e-01 4.23584700e-01 8.51812720e-01 8.59036148e-01 -7.05113681e-03 -2.54124135e-01 4.53741521e-01 -5.53907573e-01 -4.45973128e-01 4.05011117e-01 1.77196518e-01 -1.87832057e-01 1.15480736e-01 2.79829532e-01 -6.22854471e-01 -1.02167034e+00 3.97912413e-01 -8.47143829e-01 1.52241111e-01 1.07890213e+00 6.35655642e-01 9.49097693e-01 6.27382994e-02 4.56948310e-01 -3.14949721e-01 2.25627944e-01 -3.31614256e-01 -4.54952508e-01 1.62731573e-01 -5.05951703e-01 -2.58572429e-01 2.94883251e-01 -1.83058642e-02 -8.78940403e-01 2.65844524e-01 -1.78674743e-01 -5.64504027e-01 -1.42028108e-01 4.50178534e-01 -3.92672777e-01 -3.18915784e-01 3.14803421e-01 4.43713337e-01 -2.64778882e-01 -4.70228553e-01 3.32434207e-01 6.20299757e-01 5.64452648e-01 -4.34310883e-01 9.45175886e-01 7.46320724e-01 1.98613957e-01 -8.61069024e-01 -1.16717041e+00 -6.58804059e-01 -6.97822392e-01 -4.80548531e-01 7.90762067e-01 -1.17913592e+00 -6.63842022e-01 8.54720592e-01 -8.29678118e-01 -9.23166424e-02 -2.08660856e-01 4.55148488e-01 -3.35576564e-01 1.66613460e-01 -3.93368572e-01 -6.03556037e-01 -2.55038887e-01 -1.17302597e+00 1.28813326e+00 6.25533402e-01 5.11665404e-01 -6.48629010e-01 -2.56590366e-01 5.50492346e-01 2.28493318e-01 4.64935601e-01 6.26160026e-01 -2.28679672e-01 -7.87641883e-01 -8.30683932e-02 -6.66931987e-01 2.19643623e-01 6.04009181e-02 -2.17824697e-01 -9.94149029e-01 -2.32400686e-01 -2.21486077e-01 -5.09765804e-01 1.12787521e+00 4.67565238e-01 1.36145103e+00 4.20644879e-01 -3.04001004e-01 8.96049917e-01 1.51684105e+00 -3.26690339e-02 2.90056735e-01 5.36849141e-01 8.72598231e-01 5.77473462e-01 8.38872790e-01 6.23973668e-01 7.79334068e-01 5.38481414e-01 7.92860091e-01 -2.62460560e-01 -2.05677152e-01 2.67690439e-02 1.49918631e-01 8.40193748e-01 -1.55148879e-01 3.57157066e-02 -9.06552732e-01 5.65970480e-01 -1.61842668e+00 -7.04323292e-01 -8.56133029e-02 1.72160614e+00 6.54773176e-01 2.00191930e-01 2.28675038e-01 2.00687215e-01 5.24263978e-01 2.51708984e-01 -7.36208141e-01 2.10622668e-01 -4.05104607e-01 1.91822723e-01 3.83596539e-01 1.16818525e-01 -1.34109461e+00 7.38979280e-01 5.12481022e+00 9.46677446e-01 -1.20135403e+00 1.21738888e-01 5.80829680e-01 -6.86796010e-02 -2.22742975e-01 -2.98747987e-01 -9.59776640e-01 3.60491276e-01 4.06878650e-01 1.57060131e-01 -1.46360278e-01 7.56149709e-01 6.75273361e-04 -3.59430522e-01 -7.13325202e-01 1.16630578e+00 1.99428126e-01 -1.24306273e+00 -8.04825053e-02 3.28590139e-03 7.66598761e-01 2.74632573e-01 1.56151786e-01 3.23073328e-01 2.37731770e-01 -6.85672820e-01 7.19570518e-01 3.87087852e-01 5.74665010e-01 -9.43328738e-01 6.86648130e-01 2.60800838e-01 -1.63259697e+00 -3.05878818e-01 -1.50825083e-01 2.79068172e-01 4.44368310e-02 5.93941092e-01 -8.12608600e-02 9.99429584e-01 1.10707831e+00 1.06635547e+00 -9.46451962e-01 1.13383520e+00 -1.88012883e-01 2.41498858e-01 -3.33715171e-01 1.38668433e-01 4.40821320e-01 8.42619985e-02 3.79218906e-01 1.01270866e+00 3.33921462e-01 3.76821697e-01 3.49463135e-01 3.89560491e-01 3.87118123e-02 -1.45520404e-01 -2.43162960e-01 4.15021837e-01 3.97917807e-01 1.56450820e+00 -9.19763267e-01 -6.27231970e-02 -5.51144898e-01 7.77730703e-01 2.31469616e-01 1.31900102e-01 -7.33465672e-01 -3.31887454e-01 4.77433413e-01 -4.23335545e-02 5.85256040e-01 -2.30836242e-01 -1.95621535e-01 -1.10605562e+00 -3.85375624e-03 -5.78496933e-01 5.03207922e-01 -7.98899353e-01 -1.20371866e+00 6.50427997e-01 6.42014071e-02 -1.52677023e+00 1.01552494e-01 -6.03640795e-01 -2.82610834e-01 6.80588067e-01 -2.09557366e+00 -1.46198952e+00 -7.53372610e-01 9.60151136e-01 5.74439049e-01 -3.29785161e-02 3.77642393e-01 3.30135763e-01 -6.49889112e-01 4.95242923e-01 -2.30048373e-01 9.11848620e-02 4.78491098e-01 -1.13484824e+00 -1.53993607e-01 9.42114711e-01 -3.38121597e-03 2.35115021e-01 2.97323942e-01 -3.17243010e-01 -1.47507787e+00 -1.21446729e+00 3.26870114e-01 -7.91659132e-02 4.72993672e-01 -1.10985368e-01 -7.13903546e-01 2.44698822e-01 -2.94505097e-02 4.66560602e-01 6.63202107e-01 -1.58150017e-01 -2.79535204e-01 -4.31273699e-01 -1.01435363e+00 2.64005721e-01 1.08350253e+00 -4.03473943e-01 -4.62667733e-01 -1.53505970e-02 8.45317245e-01 -4.68503475e-01 -9.75141585e-01 8.11463952e-01 5.59272289e-01 -1.44458950e+00 1.20150900e+00 -7.11646453e-02 5.25118470e-01 -4.74182457e-01 -5.49706936e-01 -1.14701366e+00 -3.06452125e-01 -2.00534165e-01 -2.84672976e-01 1.18670774e+00 1.66326240e-01 -4.03948456e-01 8.40963006e-01 1.70558169e-01 -4.46201295e-01 -1.27709579e+00 -9.03456092e-01 -5.51865578e-01 -6.55607656e-02 -6.26987040e-01 5.65291643e-01 7.37876773e-01 -5.42998433e-01 2.15743840e-01 -9.39938352e-02 3.15685689e-01 8.65734279e-01 5.68145990e-01 6.56565487e-01 -1.38384521e+00 -1.09075278e-01 -5.80559611e-01 -4.83850539e-01 -1.29556417e+00 -1.36934906e-01 -7.82716334e-01 -7.71887181e-03 -1.61019683e+00 2.82825708e-01 -6.70642555e-01 -7.09546804e-01 5.28718889e-01 -2.33600244e-01 7.49149919e-01 3.31830591e-01 1.83476314e-01 -8.09813321e-01 7.71827936e-01 1.52277565e+00 -7.93769434e-02 -2.39171609e-01 2.16853409e-03 -9.21051145e-01 9.31857467e-01 6.29953742e-01 -1.59247249e-01 -3.37938249e-01 -3.51161182e-01 1.52864177e-02 4.12920304e-02 6.27824068e-01 -1.12104619e+00 3.46138358e-01 -1.94883645e-01 6.35222971e-01 -1.19263256e+00 4.09245729e-01 -8.97456288e-01 -4.22474474e-01 3.50289404e-01 6.51490316e-02 -5.56167401e-02 1.51184112e-01 7.07195401e-01 -3.74057770e-01 2.28596494e-01 9.11648095e-01 -7.20792562e-02 -1.24260128e+00 6.10862255e-01 1.10804550e-01 -3.10056787e-02 1.27804780e+00 -5.02895117e-01 -2.37340927e-01 8.85388851e-02 -5.21776259e-01 4.03610826e-01 2.90166289e-01 4.22181547e-01 1.03524709e+00 -1.29968214e+00 -5.38589776e-01 2.87316829e-01 3.25327367e-01 4.74467337e-01 5.18814683e-01 1.08709741e+00 -1.33054957e-01 2.79728264e-01 -2.14437842e-01 -1.00091553e+00 -1.08886588e+00 2.14517638e-01 1.79774672e-01 -1.59699827e-01 -5.22211134e-01 1.26530766e+00 1.99063212e-01 -2.25721464e-01 2.92605340e-01 -2.38780603e-01 -2.74759203e-01 4.57356632e-01 3.87349069e-01 3.02008182e-01 1.37898147e-01 -8.72647345e-01 -6.18809462e-01 9.71594036e-01 -8.50703567e-02 1.65030345e-01 1.66174054e+00 -4.39473420e-01 -3.12347766e-02 4.14289713e-01 1.17581558e+00 -2.65560389e-01 -1.45689154e+00 -6.09910667e-01 -2.24616110e-01 -5.44944286e-01 3.59886855e-01 -5.45318127e-01 -1.49965441e+00 8.26821387e-01 7.76114285e-01 9.49997902e-02 1.61226344e+00 2.62502819e-01 8.73489857e-01 8.07126462e-02 4.69763517e-01 -8.78616810e-01 4.90660131e-01 2.77531713e-01 7.04741240e-01 -1.52528405e+00 2.42429137e-01 -5.33567905e-01 -5.71494997e-01 8.75074625e-01 9.92579341e-01 -6.65572733e-02 9.02504206e-01 5.51090129e-02 -3.13200094e-02 -4.69322741e-01 -4.95531887e-01 -6.23187482e-01 4.90775585e-01 5.38550019e-01 8.63856226e-02 -1.57080352e-01 3.68325599e-02 6.95148885e-01 4.06058365e-03 -1.61699578e-01 3.51056643e-02 1.14161611e+00 -6.85311735e-01 -8.09187889e-01 -3.25471014e-01 4.94391769e-01 -1.03076182e-01 7.42859021e-02 -2.00816602e-01 1.01245952e+00 5.09628117e-01 9.67692196e-01 -1.25802502e-01 -8.40889275e-01 3.83169770e-01 -2.79817790e-01 5.38533568e-01 -4.13579166e-01 -4.88389462e-01 1.61325738e-01 -5.45875192e-01 -7.56473780e-01 -8.91314089e-01 -7.63761520e-01 -1.18906713e+00 -7.50392005e-02 -3.87146145e-01 -1.51661351e-01 5.65693378e-01 9.08245862e-01 3.85046095e-01 6.42244935e-01 9.44851398e-01 -1.16233611e+00 -1.08372703e-01 -7.04310179e-01 -5.76104820e-01 8.69717002e-02 3.63941640e-01 -1.04870605e+00 -3.74988675e-01 -1.55337751e-01]
[9.653342247009277, -0.8716728091239929]
9434b4dd-f47e-4871-9520-5aa9edf9d809
unsupervised-machine-translation-using
1711.00043
null
http://arxiv.org/abs/1711.00043v2
http://arxiv.org/pdf/1711.00043v2.pdf
Unsupervised Machine Translation Using Monolingual Corpora Only
Machine translation has recently achieved impressive performance thanks to recent advances in deep learning and the availability of large-scale parallel corpora. There have been numerous attempts to extend these successes to low-resource language pairs, yet requiring tens of thousands of parallel sentences. In this work, we take this research direction to the extreme and investigate whether it is possible to learn to translate even without any parallel data. We propose a model that takes sentences from monolingual corpora in two different languages and maps them into the same latent space. By learning to reconstruct in both languages from this shared feature space, the model effectively learns to translate without using any labeled data. We demonstrate our model on two widely used datasets and two language pairs, reporting BLEU scores of 32.8 and 15.1 on the Multi30k and WMT English-French datasets, without using even a single parallel sentence at training time.
["Marc'Aurelio Ranzato", 'Ludovic Denoyer', 'Guillaume Lample', 'Alexis Conneau']
2017-10-31
unsupervised-machine-translation-using-1
https://openreview.net/forum?id=rkYTTf-AZ
https://openreview.net/pdf?id=rkYTTf-AZ
iclr-2018-1
['unsupervised-machine-translation']
['natural-language-processing']
[-4.27492447e-02 -3.21608931e-01 -2.99138814e-01 -4.28526253e-01 -1.46244109e+00 -8.73060107e-01 9.56133544e-01 -1.41275272e-01 -5.46635747e-01 1.14360964e+00 2.38805115e-01 -5.75823843e-01 3.73307556e-01 -5.20261467e-01 -9.73862112e-01 -4.02416885e-01 1.83596030e-01 9.27873850e-01 -2.35040531e-01 -2.89711207e-01 -6.88416585e-02 1.38588265e-01 -7.45265007e-01 5.08743644e-01 9.12008524e-01 3.15990001e-01 3.15390319e-01 6.06323838e-01 -1.20779440e-01 3.33135456e-01 -3.48831624e-01 -7.81754136e-01 5.45094609e-01 -4.22929227e-01 -9.77743804e-01 -3.55379619e-02 5.04218340e-01 -2.89676189e-01 -1.89666763e-01 7.83592105e-01 5.22669017e-01 -3.43518555e-01 4.76947933e-01 -8.23844850e-01 -8.08139920e-01 7.46684492e-01 -3.78169835e-01 1.39890730e-01 3.92961562e-01 7.13359490e-02 1.20035684e+00 -1.01360476e+00 8.12325060e-01 1.11203277e+00 6.32905304e-01 3.46548378e-01 -1.54917860e+00 -7.00182378e-01 -4.09769058e-01 4.18824032e-02 -1.32528043e+00 -6.70375764e-01 4.48753059e-01 -4.68558550e-01 1.39874995e+00 -1.21249877e-01 3.99769276e-01 1.52772534e+00 5.87117195e-01 6.78670108e-01 1.50084615e+00 -7.27978587e-01 -1.78179443e-01 1.71880037e-01 -2.82892823e-01 6.29901648e-01 -1.46912469e-04 1.02674082e-01 -6.76984847e-01 -2.58901156e-02 5.24739027e-01 -2.28603497e-01 -9.16836262e-02 -2.99805552e-01 -1.65034258e+00 8.97943735e-01 1.90670624e-01 5.60440481e-01 -1.65249616e-01 -1.53376937e-01 4.58832771e-01 9.22269523e-01 7.37822831e-01 5.37280262e-01 -7.16417432e-01 -2.37849653e-01 -9.72526491e-01 -1.19258836e-03 9.22543168e-01 1.03755033e+00 8.94437730e-01 -2.12886199e-01 1.73664376e-01 7.62424588e-01 1.58188958e-02 6.49663746e-01 7.89263785e-01 -5.15973806e-01 1.09107268e+00 2.68914938e-01 6.82259202e-02 -6.83589637e-01 -2.04991981e-01 -3.67485166e-01 -8.61389518e-01 -3.78192633e-01 4.89748567e-01 -2.84074366e-01 -4.99999613e-01 1.83062136e+00 -2.53778789e-02 -5.12828305e-02 5.56085169e-01 6.43295825e-01 2.55938202e-01 7.94230342e-01 -3.31395417e-01 -1.39039993e-01 1.05791163e+00 -1.25931072e+00 -3.90188962e-01 -3.90208185e-01 8.67402136e-01 -1.29886568e+00 1.38215137e+00 2.99290627e-01 -1.02999759e+00 -6.67935312e-01 -1.09877300e+00 -2.88474828e-01 -1.84119746e-01 2.37071291e-01 4.86052573e-01 3.10886294e-01 -1.25161731e+00 7.91690528e-01 -8.86463284e-01 -7.15079248e-01 2.54217982e-02 4.24875259e-01 -7.67555296e-01 -2.82667339e-01 -1.22076821e+00 1.17811167e+00 2.08194152e-01 8.94871482e-04 -7.93649077e-01 -2.83335507e-01 -5.75585961e-01 -1.67279616e-01 6.17883094e-02 -7.87191570e-01 1.31353164e+00 -1.13256037e+00 -1.74350166e+00 1.00673652e+00 -2.32291296e-01 -4.65074927e-01 7.39238739e-01 -4.24368650e-01 -4.24447417e-01 -1.46699503e-01 4.01079327e-01 5.52426994e-01 5.27759731e-01 -7.17612505e-01 -3.62691879e-01 -2.88902164e-01 1.71578936e-02 2.00075075e-01 -5.03804684e-01 2.78360248e-01 -2.01187894e-01 -4.44946200e-01 3.21041569e-02 -1.12372744e+00 -4.44259420e-02 -4.75160390e-01 -4.15945053e-01 -1.56379864e-01 1.70088977e-01 -9.48122501e-01 6.58047497e-01 -1.77692997e+00 6.47096813e-01 -3.37119967e-01 3.50081362e-02 1.51379704e-01 -6.18992150e-01 8.25692475e-01 1.11796618e-01 1.63709484e-02 -1.63585171e-01 -6.65817797e-01 -4.11505103e-02 2.92927027e-01 -4.50203449e-01 5.44297457e-01 1.97432533e-01 1.13625300e+00 -9.29530323e-01 -2.44393483e-01 -2.02912703e-01 3.10509712e-01 -3.47592533e-01 2.28243351e-01 -1.01467498e-01 6.76952839e-01 -7.95641989e-02 2.52277255e-01 4.24231857e-01 -2.15784341e-01 6.22792304e-01 2.03513160e-01 -1.35814235e-01 8.29124331e-01 -4.88024324e-01 2.13623381e+00 -9.34146821e-01 7.27028966e-01 -2.57371992e-01 -1.02170074e+00 9.52514827e-01 6.02437556e-01 2.36757249e-01 -8.49929035e-01 -6.12294301e-02 7.36704588e-01 1.46434277e-01 -4.25809950e-01 2.99597532e-01 -5.67276359e-01 -3.01797926e-01 8.43192160e-01 3.62285078e-01 -7.91701749e-02 1.91664010e-01 -1.85491100e-01 9.28433895e-01 1.18459329e-01 1.93506017e-01 -3.43528122e-01 4.31749433e-01 2.87464559e-02 4.22518909e-01 4.85778630e-01 9.88846496e-02 4.14572954e-01 2.78852701e-01 -7.59469390e-01 -1.49537432e+00 -1.36967552e+00 5.89142926e-02 8.86072159e-01 -3.09188634e-01 -4.39015388e-01 -5.47809601e-01 -7.06916869e-01 -1.55578777e-01 5.19927204e-01 -2.10918710e-01 -4.88847271e-02 -9.44548726e-01 -6.74674690e-01 7.59200633e-01 2.63938576e-01 4.46609169e-01 -8.36610138e-01 8.51454288e-02 2.26212740e-01 -5.45302451e-01 -1.11601567e+00 -5.15081584e-01 7.50596225e-02 -1.00109386e+00 -6.99176013e-01 -7.60427952e-01 -1.06449711e+00 4.94593650e-01 7.51248151e-02 1.53848815e+00 -3.22439909e-01 1.89571232e-01 -2.14014769e-01 -1.63385123e-01 1.04046635e-01 -9.19099092e-01 7.35945523e-01 5.11909842e-01 -8.82130191e-02 5.52239418e-01 -7.58041918e-01 -8.14584345e-02 2.08093867e-01 -5.40782928e-01 4.52728957e-01 8.34268212e-01 1.01247692e+00 3.83267015e-01 -5.23367882e-01 6.46952987e-01 -6.95201695e-01 5.73326170e-01 -5.72729766e-01 -6.68005943e-01 4.37001646e-01 -4.38489616e-01 2.13831633e-01 1.09800529e+00 -3.51187140e-01 -5.95349431e-01 -1.23216107e-01 -2.15577018e-02 -7.81588852e-02 -6.07538372e-02 5.30413806e-01 -7.50366747e-02 3.57309222e-01 5.79187453e-01 5.17241180e-01 -1.72920659e-01 -6.65667772e-01 4.16821301e-01 9.40907657e-01 2.03707352e-01 -6.31718457e-01 9.15227115e-01 1.33169949e-01 -1.78030074e-01 -4.61114973e-01 -8.54626536e-01 -1.96287081e-01 -1.06155980e+00 3.96967769e-01 7.20506191e-01 -1.12448895e+00 -1.14088967e-01 3.27091843e-01 -1.35540855e+00 -4.14669454e-01 3.88389453e-03 7.65176535e-01 -7.81208336e-01 2.26827785e-01 -8.83603394e-01 -8.38790908e-02 -4.52981621e-01 -1.25676107e+00 1.00858545e+00 -3.10148090e-01 -3.10026854e-01 -1.07806718e+00 5.54780185e-01 4.46562439e-01 4.57110345e-01 -2.71829724e-01 8.92891288e-01 -7.33801126e-01 -6.06593907e-01 -1.45933211e-01 1.42084369e-02 5.07775068e-01 5.55909753e-01 -4.09170747e-01 -6.35489583e-01 -7.52790630e-01 7.59047717e-02 -6.49685740e-01 6.04523122e-01 -1.72966257e-01 3.13602686e-01 -3.24310511e-01 -1.42637536e-01 7.22836375e-01 1.43867433e+00 -2.51836479e-01 2.59412706e-01 2.87354648e-01 4.94373739e-01 5.46062648e-01 2.92865280e-02 -2.52836615e-01 4.51840281e-01 8.72266889e-01 -2.84374893e-01 -2.90985163e-02 -1.02893017e-01 -5.17366290e-01 7.16125906e-01 1.77084005e+00 7.60033727e-02 -1.63749114e-01 -1.14778686e+00 5.85372388e-01 -1.53217423e+00 -7.18587339e-01 1.03200518e-01 2.22710443e+00 1.24060607e+00 8.78706053e-02 -1.53185800e-01 -3.34026068e-01 4.61894721e-01 1.13697480e-02 -2.51752287e-01 -5.64339638e-01 -3.04916650e-01 3.08428377e-01 2.65538722e-01 7.68226147e-01 -9.73441184e-01 1.23269236e+00 6.45842028e+00 5.18348575e-01 -1.49480140e+00 4.31669533e-01 7.50420749e-01 -1.72777519e-01 -3.00570369e-01 2.14499697e-01 -7.80349076e-01 3.32133532e-01 1.39408505e+00 -3.08332324e-01 7.69625366e-01 3.40605825e-01 6.16634935e-02 2.40251467e-01 -1.45910609e+00 8.12118113e-01 1.02233224e-01 -1.32549345e+00 3.21462937e-02 -3.95552590e-02 1.06228793e+00 7.53363907e-01 6.71508163e-02 4.79372501e-01 3.57292771e-01 -1.17215085e+00 5.43594480e-01 3.85633379e-01 1.00846207e+00 -5.06683290e-01 7.38510311e-01 7.32035398e-01 -7.68105209e-01 2.79859781e-01 -5.87263227e-01 -3.93325031e-01 1.28500894e-01 3.77044767e-01 -1.23123825e+00 7.58366406e-01 3.58895153e-01 8.83354425e-01 -5.65261900e-01 5.35418093e-01 -3.59835953e-01 7.60288239e-01 -3.00522178e-01 5.82064241e-02 2.93077707e-01 -4.52776939e-01 2.92656332e-01 9.86449659e-01 5.51203787e-01 -6.65863633e-01 3.69279742e-01 7.00159788e-01 -4.83480901e-01 4.32582676e-01 -7.77166605e-01 -2.04944462e-01 1.04119346e-01 1.03803170e+00 -4.54338759e-01 -3.42197090e-01 -7.42611527e-01 1.50027454e+00 8.88458073e-01 2.54884720e-01 -5.61086059e-01 -2.61035208e-02 4.84619349e-01 -4.51838672e-02 -2.05311328e-02 -6.25456095e-01 -1.92941338e-01 -1.66670406e+00 3.00169826e-01 -1.05691850e+00 -2.00642139e-01 -5.19118845e-01 -1.63727248e+00 9.75105524e-01 -3.48814458e-01 -1.14635980e+00 -6.89797699e-01 -4.39379513e-01 -3.71906847e-01 1.38711262e+00 -1.47644877e+00 -1.37268293e+00 5.08267224e-01 4.17531788e-01 6.65094376e-01 -6.58578217e-01 1.16354036e+00 4.48341399e-01 -4.03418869e-01 7.75363088e-01 7.73351669e-01 2.65701562e-01 1.15081298e+00 -1.09848309e+00 9.80508327e-01 7.47665823e-01 7.79262602e-01 8.37395668e-01 4.78440702e-01 -4.58354175e-01 -1.44314373e+00 -8.93411577e-01 1.81614971e+00 -7.54678965e-01 9.51507926e-01 -9.59313869e-01 -6.47289693e-01 9.04474020e-01 6.41507864e-01 -9.45890397e-02 7.64241159e-01 3.41994166e-01 -5.98815143e-01 -1.40304849e-01 -7.55636096e-01 6.93796933e-01 9.50818479e-01 -8.75129819e-01 -6.99398041e-01 8.09381127e-01 7.56553411e-01 -2.02296868e-01 -8.27993214e-01 1.52392492e-01 5.19077003e-01 -7.21800923e-01 7.03617275e-01 -8.85828555e-01 7.09727943e-01 4.80462722e-02 -4.45880443e-01 -1.61584818e+00 -1.33297652e-01 -6.19292557e-01 3.15265179e-01 1.13931358e+00 7.63740420e-01 -6.95132792e-01 5.37163854e-01 6.00520987e-03 -7.05701336e-02 -7.96587884e-01 -1.20104849e+00 -9.91006851e-01 8.41615558e-01 -8.39187577e-02 5.84266007e-01 1.02784562e+00 -7.01862760e-03 9.96832132e-01 -6.79895759e-01 -1.01664752e-01 2.14776784e-01 5.26999354e-01 9.53552186e-01 -1.00149870e+00 -6.51598155e-01 -2.05533966e-01 -4.57610749e-03 -1.13900399e+00 5.55774868e-01 -1.57693875e+00 -8.05084556e-02 -1.39359605e+00 4.01869446e-01 -4.16702837e-01 -9.17526111e-02 4.30216998e-01 -5.85980117e-02 5.46227336e-01 1.50173575e-01 5.07335842e-01 -4.25741702e-01 5.71943343e-01 1.05030775e+00 -1.99055851e-01 9.30940658e-02 -9.29277390e-02 -5.20077229e-01 4.72619861e-01 1.05510628e+00 -5.50477445e-01 -2.14040041e-01 -1.12931502e+00 3.70129198e-01 -1.97661668e-02 2.53249072e-02 -8.38143110e-01 -2.70377807e-02 -3.58885564e-02 3.33622009e-01 -1.16421863e-01 2.26454705e-01 -5.98127007e-01 1.25445366e-01 4.34669971e-01 -4.22461271e-01 6.15832269e-01 1.60557926e-01 3.62980306e-01 -3.20895940e-01 -6.51370734e-03 5.75422466e-01 -1.51154563e-01 -2.36702174e-01 9.84100029e-02 -2.33448133e-01 8.44490081e-02 7.56984115e-01 3.95963907e-01 -1.80053324e-01 -2.49066204e-01 -5.56061029e-01 -1.26793468e-02 7.67984927e-01 7.57508159e-01 9.06629264e-02 -1.49180198e+00 -1.12271583e+00 5.11606097e-01 7.32593462e-02 -4.75550175e-01 -1.67990610e-01 7.30430245e-01 -6.48665845e-01 8.07089269e-01 -3.66138518e-01 -6.36152327e-01 -9.19255495e-01 5.04270077e-01 2.42780209e-01 -7.20024586e-01 -4.98414546e-01 5.65133989e-01 -1.94283292e-01 -1.10360670e+00 -3.47725242e-01 -9.89980847e-02 3.85331601e-01 -1.56624094e-01 2.39386752e-01 -1.16475895e-02 2.44270623e-01 -8.69615972e-01 -1.61729604e-01 5.83351493e-01 -3.08006555e-01 -4.48262483e-01 1.37401283e+00 -1.75915688e-01 -3.00719053e-01 7.44674087e-01 1.57320762e+00 3.64576846e-01 -9.97354329e-01 -5.06586015e-01 2.89530367e-01 -3.21103543e-01 -5.10006368e-01 -7.88824141e-01 -7.17563391e-01 1.17540300e+00 2.80205518e-01 -1.43199325e-01 6.97844565e-01 5.29253967e-02 1.12445736e+00 6.28357172e-01 7.30317891e-01 -8.50389004e-01 6.62083328e-02 9.04126585e-01 7.24453986e-01 -1.43979681e+00 -2.82222271e-01 7.71566927e-02 -2.65741706e-01 1.02523243e+00 2.66563833e-01 -2.22645417e-01 2.78312892e-01 2.05811143e-01 3.26592952e-01 3.91722053e-01 -9.03882504e-01 3.57870936e-01 1.99051037e-01 2.86631227e-01 8.16353500e-01 2.57458895e-01 -4.30627584e-01 3.67196709e-01 -5.62686741e-01 -7.17417896e-02 3.64982635e-01 6.44342303e-01 -6.15358949e-02 -1.87269366e+00 -1.39143154e-01 2.93896884e-01 -4.88708317e-01 -4.03377652e-01 -2.53159910e-01 5.42887568e-01 -1.26992434e-01 8.58544469e-01 3.88406999e-02 -3.54149759e-01 -1.59303956e-02 4.50203747e-01 6.19794369e-01 -7.56127656e-01 -5.01842976e-01 -1.62781194e-01 1.46873733e-02 -3.77191782e-01 -2.50475675e-01 -7.68901229e-01 -7.88680971e-01 -4.74914968e-01 -4.61456999e-02 2.77407050e-01 6.43124461e-01 1.01920438e+00 4.38868433e-01 3.93701904e-02 6.76538706e-01 -6.93985522e-01 -1.02568233e+00 -1.18225741e+00 -2.10352302e-01 1.70644775e-01 3.22356641e-01 -1.95710212e-01 -9.11373720e-02 -4.99881990e-03]
[11.597250938415527, 10.271446228027344]
5f184a27-26b2-4d98-8f2d-7ae84d5fc4a0
tsanet-temporal-and-scale-alignment-for
2303.04376
null
https://arxiv.org/abs/2303.04376v1
https://arxiv.org/pdf/2303.04376v1.pdf
TSANET: Temporal and Scale Alignment for Unsupervised Video Object Segmentation
Unsupervised Video Object Segmentation (UVOS) refers to the challenging task of segmenting the prominent object in videos without manual guidance. In other words, the network detects the accurate region of the target object in a sequence of RGB frames without prior knowledge. In recent works, two approaches for UVOS have been discussed that can be divided into: appearance and appearance-motion based methods. Appearance based methods utilize the correlation information of inter-frames to capture target object that commonly appears in a sequence. However, these methods does not consider the motion of target object due to exploit the correlation information between randomly paired frames. Appearance-motion based methods, on the other hand, fuse the appearance features from RGB frames with the motion features from optical flow. Motion cue provides useful information since salient objects typically show distinctive motion in a sequence. However, these approaches have the limitation that the dependency on optical flow is dominant. In this paper, we propose a novel framework for UVOS that can address aforementioned limitations of two approaches in terms of both time and scale. Temporal Alignment Fusion aligns the saliency information of adjacent frames with the target frame to leverage the information of adjacent frames. Scale Alignment Decoder predicts the target object mask precisely by aggregating differently scaled feature maps via continuous mapping with implicit neural representation. We present experimental results on public benchmark datasets, DAVIS 2016 and FBMS, which demonstrate the effectiveness of our method. Furthermore, we outperform the state-of-the-art methods on DAVIS 2016.
['Sangyoun Lee', 'Minhyeok Lee', 'Dogyoon Lee', 'Suhwan Cho', 'Seunghoon Lee']
2023-03-08
null
null
null
null
['video-object-segmentation', 'video-semantic-segmentation', 'unsupervised-video-object-segmentation']
['computer-vision', 'computer-vision', 'computer-vision']
[ 3.44170094e-01 -3.36661965e-01 -4.07615840e-01 -1.64502397e-01 -3.61981869e-01 -4.63758349e-01 2.31099427e-01 -2.61399359e-01 -4.56377566e-01 4.93270665e-01 9.78143588e-02 3.68264675e-01 1.34359136e-01 -4.11446512e-01 -7.37972260e-01 -9.01528656e-01 2.73427039e-01 -3.58417273e-01 8.31434429e-01 -9.71215963e-02 3.16969156e-01 2.24209934e-01 -1.46371222e+00 1.39877588e-01 8.58630359e-01 1.43156219e+00 6.13124311e-01 3.73031050e-01 -1.32056773e-01 8.43643844e-01 -3.76522064e-01 7.61077330e-02 5.09449065e-01 -4.44713533e-01 -6.28900170e-01 3.02717149e-01 5.87894559e-01 -6.63080513e-01 -6.36342585e-01 1.26674807e+00 2.05863670e-01 2.56895602e-01 3.54168147e-01 -1.38789308e+00 -5.81722736e-01 2.55426556e-01 -9.72466946e-01 7.52006710e-01 1.99663162e-01 3.94788504e-01 7.52166450e-01 -9.71258640e-01 6.23452723e-01 1.18941200e+00 2.94219464e-01 4.54218358e-01 -8.60620260e-01 -6.01994038e-01 5.49115837e-01 5.36992610e-01 -1.19389200e+00 -4.30749625e-01 1.08060980e+00 -5.56978583e-01 5.52710235e-01 5.05907461e-02 7.46860504e-01 8.52501392e-01 -2.72025634e-02 1.13215244e+00 1.01154816e+00 -1.59636036e-01 3.91783491e-02 -6.40128031e-02 1.46368876e-01 7.45378137e-01 2.30839312e-01 1.74267858e-01 -6.14909649e-01 2.82972336e-01 9.25894439e-01 3.27892691e-01 -5.95497668e-01 -4.30929095e-01 -1.41631806e+00 3.86487752e-01 5.00838578e-01 1.56534865e-01 -3.86105448e-01 1.71531603e-01 3.09001148e-01 -1.46062389e-01 2.55650729e-01 -6.56431317e-02 -4.40658003e-01 -2.81182881e-02 -9.94875550e-01 -1.05930008e-01 2.05929354e-01 1.10480475e+00 8.77798498e-01 2.37289190e-01 -3.78383458e-01 3.91904682e-01 5.08838415e-01 5.52467346e-01 5.99750817e-01 -1.04376578e+00 5.20241678e-01 6.60089135e-01 2.36236364e-01 -1.17103314e+00 -2.01915011e-01 -1.04489580e-01 -5.84461510e-01 1.61011770e-01 4.63683188e-01 -5.84764704e-02 -1.09327352e+00 1.72258341e+00 5.88200450e-01 6.65058434e-01 1.02982759e-01 1.46240306e+00 1.04394031e+00 6.41418338e-01 1.63279306e-02 -4.49025810e-01 1.11930203e+00 -1.33169341e+00 -8.97753716e-01 -3.02592069e-01 9.83815640e-03 -7.34869480e-01 8.15827847e-01 1.03345990e-01 -1.09242487e+00 -8.82951736e-01 -1.08614576e+00 8.25223401e-02 -1.87730089e-01 2.60748595e-01 5.44140816e-01 3.48460972e-01 -8.84326816e-01 4.96865004e-01 -9.15359139e-01 -1.70695812e-01 4.97261167e-01 2.55357653e-01 -2.01767877e-01 1.54678881e-01 -1.14484799e+00 5.38244247e-01 5.43506384e-01 3.67669046e-01 -1.19502866e+00 -4.08589393e-01 -8.05615127e-01 -1.42625943e-01 6.64702654e-01 -4.94659275e-01 1.02512765e+00 -1.66223037e+00 -1.42732799e+00 4.66677487e-01 -3.87817472e-01 -3.88052702e-01 6.20782137e-01 -3.76134783e-01 -3.02697182e-01 5.91960609e-01 1.40390754e-01 7.34788716e-01 1.14100051e+00 -1.12629938e+00 -1.06230974e+00 -1.08618632e-01 1.75480857e-01 3.20909411e-01 -2.29875699e-01 4.34738733e-02 -7.48323679e-01 -7.41590559e-01 1.60692111e-01 -8.77373517e-01 -1.14442714e-01 2.48379409e-01 -2.85511881e-01 -1.66336223e-01 1.25701523e+00 -6.69935048e-01 1.27341247e+00 -2.23847914e+00 2.68416524e-01 -2.95844376e-01 1.80412322e-01 5.01494229e-01 -1.36600330e-01 -1.18729062e-01 1.15488395e-01 -1.39166772e-01 -2.43954465e-01 2.60472056e-02 -3.50517035e-01 2.44957358e-02 -4.03307289e-01 6.27175748e-01 3.71702850e-01 1.03008723e+00 -1.08123267e+00 -9.49939787e-01 4.62017149e-01 3.27600032e-01 -3.62464160e-01 3.54677647e-01 -1.88488573e-01 6.97031736e-01 -6.47938073e-01 7.71198332e-01 6.77722037e-01 -2.09018856e-01 -2.52705246e-01 -5.43542147e-01 -1.38306454e-01 -3.72116603e-02 -1.19235528e+00 1.94715798e+00 1.24572545e-01 5.99936664e-01 -6.67946711e-02 -9.19242084e-01 5.78067005e-01 1.94127977e-01 7.88488030e-01 -4.44555044e-01 1.74614236e-01 1.57217875e-01 -2.15217136e-02 -7.68472314e-01 3.68215740e-01 4.49664205e-01 3.01147878e-01 1.44884095e-01 2.48767082e-02 2.85483003e-01 2.98144639e-01 2.86678765e-02 7.25763083e-01 6.13593996e-01 2.13892549e-01 6.58108220e-02 7.49903262e-01 -1.29626587e-01 1.04539609e+00 6.40513480e-01 -7.66952991e-01 8.46944511e-01 3.07451058e-02 -4.33281690e-01 -7.96389282e-01 -1.05603421e+00 1.21543370e-01 8.55945230e-01 1.02780461e+00 -1.97588861e-01 -7.46511817e-01 -9.31567788e-01 -1.08134121e-01 2.44208649e-01 -7.67126381e-01 -1.10220261e-01 -8.08952212e-01 -4.27985400e-01 -1.18546030e-02 6.44013286e-01 8.40248406e-01 -1.11802065e+00 -1.02922297e+00 1.09525569e-01 -4.81510788e-01 -1.46730983e+00 -9.42380965e-01 -2.57097185e-01 -8.81510913e-01 -1.04836452e+00 -8.59046638e-01 -8.61501694e-01 6.78862274e-01 8.79500449e-01 6.66336954e-01 -8.31907906e-04 -3.26867521e-01 3.83021921e-01 -4.16680962e-01 -1.30265936e-01 4.26687039e-02 -2.68480808e-01 8.50467980e-02 4.75348413e-01 2.31384024e-01 -4.17445689e-01 -1.22406316e+00 5.57999611e-01 -9.53925312e-01 3.58617425e-01 7.62770236e-01 5.57660043e-01 7.09501386e-01 -3.46490800e-01 4.27632630e-01 -2.15540871e-01 -1.96913928e-01 -4.24688250e-01 -4.37531292e-01 3.64993334e-01 -2.73063093e-01 -7.36758560e-02 3.65077287e-01 -5.70588946e-01 -1.04236281e+00 4.06161040e-01 5.07660329e-01 -9.35277104e-01 -8.05993751e-02 8.80651325e-02 -1.61980897e-01 -1.09230459e-01 1.26705408e-01 5.53580821e-01 -1.24991098e-02 -2.44772181e-01 2.98177332e-01 6.24770284e-01 7.07370162e-01 -3.11533928e-01 7.05146730e-01 7.75665045e-01 -8.11662674e-02 -5.10415077e-01 -9.77929294e-01 -7.41478682e-01 -7.33619750e-01 -5.65363228e-01 1.15029907e+00 -9.90084350e-01 -5.61897695e-01 4.47322071e-01 -1.20288444e+00 -6.78182840e-02 -1.25598535e-01 5.55451393e-01 -5.93267083e-01 6.57545507e-01 -5.28223574e-01 -7.34078109e-01 -4.11297709e-01 -1.34437811e+00 1.20896673e+00 7.09269166e-01 1.61234513e-01 -5.66508055e-01 -3.72483224e-01 2.56684989e-01 1.79704323e-01 4.03343499e-01 3.38126898e-01 -2.88435400e-01 -1.00585115e+00 1.85527578e-01 -4.68065470e-01 4.03497636e-01 4.95625645e-01 1.03604190e-01 -9.06317472e-01 -2.65429348e-01 6.48067445e-02 -4.02779318e-02 9.71683621e-01 6.37926102e-01 1.08857024e+00 -1.66710317e-01 -3.68094206e-01 6.71779513e-01 1.40375006e+00 3.54650706e-01 3.24641675e-01 4.17170852e-01 1.03866899e+00 5.69180369e-01 1.13682473e+00 2.89660931e-01 1.35659561e-01 8.82933438e-01 5.39063334e-01 -1.21471949e-01 -3.67908627e-01 -1.21776156e-01 6.61403835e-01 6.28656924e-01 -2.33449385e-01 1.09966844e-01 -4.82673556e-01 5.58687389e-01 -2.15926576e+00 -1.01306880e+00 2.11815778e-02 2.11500096e+00 7.53767014e-01 1.37015700e-01 1.53962895e-01 -1.85772419e-01 1.03086901e+00 3.12216431e-01 -8.37208569e-01 2.42533356e-01 -2.90733725e-01 -3.76145512e-01 6.02154255e-01 1.01993687e-01 -1.35217714e+00 9.46953714e-01 5.35569239e+00 7.61280715e-01 -1.26208222e+00 2.36608580e-01 6.90901101e-01 -1.91289604e-01 1.13017172e-01 6.02414533e-02 -7.52594411e-01 7.97098875e-01 3.44598681e-01 -1.84901893e-01 3.59443039e-01 7.51236200e-01 4.24264729e-01 -3.22193325e-01 -1.03121364e+00 1.13466322e+00 1.97694078e-01 -1.14307380e+00 -9.89332274e-02 -2.99078286e-01 1.05425739e+00 -2.94501968e-02 2.08536103e-01 -3.06585319e-02 -1.81698039e-01 -7.36989558e-01 1.10092616e+00 6.42373323e-01 5.07480979e-01 -4.38497454e-01 5.53328335e-01 4.36820388e-02 -1.57087815e+00 -1.79521650e-01 -2.73043275e-01 1.02078564e-01 3.15148413e-01 2.81599104e-01 -4.23814416e-01 7.35538363e-01 9.53996241e-01 1.21005869e+00 -6.90157592e-01 1.27828395e+00 -1.44070432e-01 5.96012771e-01 -1.47327900e-01 1.98068231e-01 3.72548580e-01 -4.07787532e-01 7.14094877e-01 1.06338894e+00 2.17369035e-01 8.79150257e-02 4.59392577e-01 8.20203960e-01 4.04454559e-01 5.04450984e-02 -2.67760426e-01 1.15003943e-01 2.65269309e-01 1.29441750e+00 -9.44642365e-01 -4.50415820e-01 -6.43473029e-01 1.17383158e+00 -3.03059947e-02 5.69054544e-01 -1.05396998e+00 -8.07254240e-02 5.06450355e-01 -1.40680866e-02 7.46658504e-01 -2.42342308e-01 7.08884448e-02 -1.23862255e+00 1.47377089e-01 -6.75829887e-01 3.04453760e-01 -8.44175458e-01 -9.36502874e-01 4.72221404e-01 -2.38293540e-02 -1.65667808e+00 -2.62580998e-02 -4.14765835e-01 -4.83780831e-01 6.45803273e-01 -1.61290216e+00 -1.05302465e+00 -6.96792126e-01 5.15960336e-01 1.01192629e+00 2.56369933e-02 1.90234073e-02 1.59593821e-01 -8.08896720e-01 3.16903085e-01 7.41756633e-02 3.72612536e-01 6.88313186e-01 -1.09072280e+00 1.71214655e-01 1.14396393e+00 1.11838922e-01 4.16347295e-01 5.70956945e-01 -7.53823519e-01 -1.33004630e+00 -1.23943198e+00 2.35148072e-01 -2.45863780e-01 5.31200886e-01 -3.68210338e-02 -9.11269188e-01 4.43627685e-01 1.71110690e-01 5.74783504e-01 1.47187620e-01 -7.39989579e-01 -3.68290357e-02 -2.08111405e-01 -7.37860322e-01 4.89540786e-01 1.20640719e+00 -3.34498852e-01 -5.83939910e-01 2.35792883e-02 9.39407408e-01 -6.02452815e-01 -5.26519597e-01 6.77109659e-01 5.33507049e-01 -1.02140117e+00 1.04924548e+00 -4.26084757e-01 4.99719232e-01 -9.55481589e-01 -4.48951237e-02 -7.87990034e-01 -1.68194219e-01 -6.63186193e-01 -3.16987365e-01 1.29143071e+00 -1.84223190e-01 -4.37906861e-01 6.95814371e-01 3.80815268e-01 -9.37909260e-02 -8.76403809e-01 -7.40887761e-01 -6.94386363e-01 -5.13712823e-01 -1.41661867e-01 3.98154616e-01 6.71465278e-01 -3.66902500e-01 5.21402471e-02 -4.03715283e-01 2.03097075e-01 6.43782198e-01 4.38204497e-01 6.63984895e-01 -8.61387014e-01 -2.17155993e-01 -4.52262551e-01 -7.01030850e-01 -1.29392874e+00 1.03417851e-01 -6.31859720e-01 2.38413826e-01 -1.36069608e+00 3.81828785e-01 -1.08059108e-01 -7.04858959e-01 2.45119393e-01 -5.90955675e-01 3.43567640e-01 4.03126448e-01 4.92321104e-01 -9.97844756e-01 6.57858431e-01 1.47423875e+00 -3.05099636e-01 -2.53162801e-01 -1.91581264e-01 -3.04008096e-01 8.58271062e-01 5.09761691e-01 -2.37930715e-01 -4.85073179e-01 -3.74332398e-01 -3.19495708e-01 -1.62733588e-02 5.48924983e-01 -1.09721291e+00 2.67488539e-01 -3.38349193e-01 6.11892879e-01 -7.17257440e-01 1.95552558e-01 -8.29218745e-01 1.89835280e-02 4.55641210e-01 -2.57623613e-01 8.46262276e-02 1.01071954e-01 1.01067245e+00 -3.43667924e-01 -1.18204534e-01 8.01265419e-01 4.67848331e-02 -1.38152778e+00 5.35991967e-01 -2.45469492e-02 1.02524601e-01 1.27791345e+00 -4.61159706e-01 -2.05794260e-01 -2.02013299e-01 -4.67616290e-01 2.08133370e-01 4.45474297e-01 7.31949270e-01 8.71824026e-01 -1.30138433e+00 -5.36228895e-01 2.13643853e-02 2.58441437e-02 1.56253591e-01 3.61561656e-01 1.29693723e+00 -4.57021326e-01 3.31068635e-01 -3.58010024e-01 -1.01265931e+00 -1.22958863e+00 9.05026615e-01 3.89217377e-01 3.18656057e-01 -6.43639386e-01 7.12574244e-01 8.60691369e-01 4.14772749e-01 2.46926084e-01 -4.92324561e-01 -4.88225609e-01 -7.65699744e-02 6.62720263e-01 2.52048045e-01 -3.88874948e-01 -1.12849414e+00 -4.35003608e-01 9.22413528e-01 -1.28788009e-01 9.35939774e-02 9.66120541e-01 -5.48557401e-01 4.39103544e-02 6.04788899e-01 1.25730240e+00 -2.32021004e-01 -2.04505372e+00 -5.07267475e-01 -1.49698734e-01 -8.41479301e-01 3.97172123e-02 -5.20749032e-01 -1.46063817e+00 6.71438098e-01 8.84982884e-01 -1.84898585e-01 1.37782681e+00 -1.25779599e-01 8.94983232e-01 -1.08264066e-01 3.27812344e-01 -1.22361147e+00 4.66754556e-01 7.26546273e-02 6.58485353e-01 -1.51822233e+00 -1.54413125e-02 -6.60776615e-01 -7.53887415e-01 1.15574801e+00 9.61509585e-01 -9.15265158e-02 3.78510773e-01 -1.74061432e-01 8.05829465e-02 1.66709140e-01 -3.73792648e-01 -5.40631831e-01 5.56680679e-01 4.81687337e-01 -2.43957229e-02 -3.41810852e-01 -2.76657820e-01 5.31973481e-01 3.31025004e-01 -3.23919319e-02 3.81123543e-01 8.98810029e-01 -5.28375745e-01 -5.42172611e-01 -4.09370154e-01 1.13662772e-01 -4.71644521e-01 -1.27752349e-01 -1.99746653e-01 6.00558162e-01 2.71965533e-01 1.01331747e+00 -2.89174495e-03 -3.02267253e-01 7.39728138e-02 -2.67354727e-01 4.85531092e-01 -2.57364035e-01 -1.25475526e-01 3.17025870e-01 -4.08978522e-01 -8.15225601e-01 -9.50834036e-01 -7.67406821e-01 -1.38714623e+00 2.71275014e-01 -3.98851991e-01 -1.31755739e-01 3.00315022e-01 9.88132298e-01 2.65703768e-01 5.24924994e-01 8.07822466e-01 -1.21371722e+00 -2.79184639e-01 -7.54476666e-01 -3.90276670e-01 6.12027347e-01 6.06360435e-01 -9.01807964e-01 -3.36268693e-01 4.09588158e-01]
[9.282495498657227, -0.2823661267757416]
de9108b7-7521-4266-ad1b-317aa4161265
dr-vic-decomposition-and-reasoning-for-video
2203.12335
null
https://arxiv.org/abs/2203.12335v2
https://arxiv.org/pdf/2203.12335v2.pdf
DR.VIC: Decomposition and Reasoning for Video Individual Counting
Pedestrian counting is a fundamental tool for understanding pedestrian patterns and crowd flow analysis. Existing works (e.g., image-level pedestrian counting, crossline crowd counting et al.) either only focus on the image-level counting or are constrained to the manual annotation of lines. In this work, we propose to conduct the pedestrian counting from a new perspective - Video Individual Counting (VIC), which counts the total number of individual pedestrians in the given video (a person is only counted once). Instead of relying on the Multiple Object Tracking (MOT) techniques, we propose to solve the problem by decomposing all pedestrians into the initial pedestrians who existed in the first frame and the new pedestrians with separate identities in each following frame. Then, an end-to-end Decomposition and Reasoning Network (DRNet) is designed to predict the initial pedestrian count with the density estimation method and reason the new pedestrian's count of each frame with the differentiable optimal transport. Extensive experiments are conducted on two datasets with congested pedestrians and diverse scenes, demonstrating the effectiveness of our method over baselines with great superiority in counting the individual pedestrians. Code: https://github.com/taohan10200/DRNet.
['Wanli Ouyang', 'Qi Wang', 'Junyu Gao', 'Lei Bai', 'Tao Han']
2022-03-23
null
http://openaccess.thecvf.com//content/CVPR2022/html/Han_DR.VIC_Decomposition_and_Reasoning_for_Video_Individual_Counting_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Han_DR.VIC_Decomposition_and_Reasoning_for_Video_Individual_Counting_CVPR_2022_paper.pdf
cvpr-2022-1
['video-individual-counting']
['computer-vision']
[-3.39859843e-01 -5.96963644e-01 1.64921302e-02 -1.75026566e-01 -7.37846419e-02 -2.21353248e-01 4.70575333e-01 3.49959657e-02 -8.43101203e-01 9.24693584e-01 1.15484498e-01 -2.56686479e-01 6.95055962e-01 -1.02492821e+00 -5.39830923e-01 -6.66150630e-01 3.57181090e-03 3.59712899e-01 6.50363028e-01 1.93543971e-01 6.93067443e-03 2.05968916e-01 -1.49443638e+00 -6.24219216e-02 7.50037491e-01 7.23471105e-01 6.34634495e-02 1.16424799e+00 -1.91495940e-01 1.42060387e+00 -6.76862895e-01 -6.02996826e-01 1.41584679e-01 -2.20234096e-01 -5.84417760e-01 2.23625094e-01 6.83344781e-01 -1.08419430e+00 -6.38418913e-01 1.11271572e+00 4.63021994e-01 3.96455675e-01 4.57014441e-01 -1.57815492e+00 -5.40888965e-01 2.01200008e-01 -1.28336370e+00 7.97004044e-01 4.44611996e-01 4.71068323e-01 3.93181682e-01 -8.74661326e-01 1.01561807e-01 1.47900796e+00 6.37536883e-01 6.23473883e-01 -6.78713441e-01 -7.70734549e-01 4.35293853e-01 3.39641243e-01 -1.53023386e+00 -5.20371497e-01 3.67574513e-01 -6.95600271e-01 4.90097314e-01 1.73438117e-01 1.02213013e+00 7.08249569e-01 -4.17553872e-01 1.05364251e+00 5.75542331e-01 -2.71669596e-01 7.30822384e-02 -1.25599369e-01 4.69040573e-01 9.60398972e-01 6.74866498e-01 -1.34067267e-01 -2.76930362e-01 6.99273050e-02 8.07236254e-01 2.47375220e-01 -9.57300887e-03 1.80245012e-01 -1.28381181e+00 6.95221543e-01 3.37264985e-01 -9.38799605e-02 -3.37038130e-01 4.77796853e-01 6.15750909e-01 -4.63517874e-01 5.10727286e-01 -6.75483644e-01 1.33105135e-02 -1.14543267e-01 -1.09804320e+00 3.50781143e-01 7.47459829e-01 1.30079210e+00 7.15249896e-01 -7.81884044e-02 -6.16521597e-01 5.58541775e-01 1.71758056e-01 7.32546687e-01 -1.99085802e-01 -1.21223176e+00 8.27649772e-01 5.78682721e-01 5.37641644e-01 -1.24278772e+00 -2.06460997e-01 -2.37272475e-02 -1.11216104e+00 -1.05027109e-01 7.67156422e-01 -6.25502765e-01 -5.19116044e-01 1.44611943e+00 6.03918135e-01 7.19981909e-01 -2.72778183e-01 1.09685779e+00 8.96057844e-01 8.35885286e-01 5.80371141e-01 -1.28388807e-01 1.59930086e+00 -1.25908566e+00 -4.95342135e-01 -2.02106982e-01 4.44532365e-01 -5.31673253e-01 7.05232501e-01 -3.24303694e-02 -1.07579410e+00 -7.37949252e-01 -5.63734233e-01 -9.94010195e-02 -2.73387939e-01 4.52416688e-01 4.56371129e-01 8.02798808e-01 -8.32406998e-01 -3.39962845e-03 -7.64311075e-01 -4.41752970e-01 7.52006292e-01 1.30144998e-01 9.07641500e-02 -2.53231227e-01 -9.59538639e-01 6.30602181e-01 4.05709684e-01 2.59168774e-01 -9.33401346e-01 -6.29805446e-01 -7.41672873e-01 8.22320133e-02 4.25568402e-01 -1.11725092e+00 1.02853394e+00 -5.97161114e-01 -9.22176659e-01 7.27810144e-01 -6.43090606e-01 -4.13694650e-01 1.00008452e+00 -2.80443162e-01 -1.19246639e-01 1.64522767e-01 6.81826532e-01 8.90051067e-01 2.38452345e-01 -1.21095598e+00 -1.47126949e+00 -7.76257440e-02 4.18227524e-01 8.31151605e-02 -2.77162939e-01 2.68574417e-01 -7.55337298e-01 -1.74872890e-01 -5.52159131e-01 -6.68394923e-01 -2.43090272e-01 9.15443376e-02 -5.33820510e-01 -3.76243234e-01 7.39849091e-01 -8.81168664e-01 1.39244545e+00 -1.75788116e+00 -2.01251671e-01 -3.22504580e-01 7.06714630e-01 3.81953776e-01 1.62293777e-01 -8.23659152e-02 4.46196914e-01 -3.73242497e-02 2.52585132e-02 -6.97809935e-01 -2.77250946e-01 1.78757887e-02 1.39522403e-01 6.25240982e-01 1.31151555e-02 7.31276929e-01 -1.36926758e+00 -1.19987810e+00 7.27826595e-01 5.31428099e-01 -4.07746494e-01 1.35397375e-01 1.59477532e-01 6.48316026e-01 -3.15123796e-01 8.05700481e-01 1.12453222e+00 -2.26745456e-01 -2.68522292e-01 -1.68671072e-01 -5.34192979e-01 -5.03162444e-01 -1.41618359e+00 9.48540926e-01 -2.43466228e-01 6.03201389e-01 -1.58138976e-01 -5.07574141e-01 3.70264888e-01 9.39339027e-02 5.35148919e-01 -4.58855569e-01 2.77592361e-01 -5.89241423e-02 -3.03931087e-01 -5.62040091e-01 7.13139951e-01 3.22330445e-01 -4.30230312e-02 6.06241934e-02 -1.84951797e-01 7.18667030e-01 9.90642667e-01 3.73702228e-01 8.50769162e-01 -5.47068678e-02 3.58461648e-01 -6.20062500e-02 1.05055201e+00 8.72352645e-02 6.99608266e-01 8.10586452e-01 -9.24972355e-01 3.69352221e-01 3.06093484e-01 -8.28044832e-01 -1.21268034e+00 -1.00479853e+00 1.76030979e-01 1.03812623e+00 5.25420308e-01 -2.65849441e-01 -1.07863510e+00 -3.30076277e-01 -2.00277105e-01 4.61540312e-01 -4.70396310e-01 3.90890598e-01 -1.00239098e+00 -7.86432385e-01 5.91868579e-01 7.40788877e-01 1.25240791e+00 -7.32812107e-01 -7.62257457e-01 2.11827040e-01 -9.25304651e-01 -1.52497017e+00 -9.09940720e-01 -8.42240274e-01 -3.54355395e-01 -1.22503471e+00 -1.16616738e+00 -6.08964145e-01 7.71055281e-01 8.69498730e-01 1.05055034e+00 5.56919873e-01 -2.76929229e-01 4.19242412e-01 -1.98420603e-02 -2.90932149e-01 4.54582684e-02 6.59332238e-03 1.13576338e-01 1.65522277e-01 7.96980500e-01 -1.62568510e-01 -1.04119539e+00 3.21242690e-01 -2.58661121e-01 3.08505803e-01 1.05649985e-01 4.19283777e-01 3.26804489e-01 9.22253057e-02 4.36203927e-02 -4.73234504e-01 4.02626246e-01 -6.78226769e-01 -6.97745621e-01 2.60497868e-01 2.84495652e-01 -5.78119934e-01 6.37615442e-01 -4.64355350e-01 -1.05696416e+00 2.95838919e-02 1.20316289e-01 -3.73295426e-01 -4.63723928e-01 -2.94392645e-01 -1.68620944e-01 2.32494503e-01 1.96521923e-01 2.99211174e-01 -4.23643202e-01 2.04934715e-03 2.75471538e-01 4.33959365e-01 6.79364026e-01 -7.18923032e-01 6.19683683e-01 7.79885888e-01 -1.54666407e-02 -9.12764966e-01 -9.83964741e-01 -9.78072345e-01 -9.88800347e-01 -8.81517649e-01 1.39264357e+00 -1.21914303e+00 -1.55204010e+00 6.72149658e-01 -1.65851569e+00 -7.81713873e-02 -4.05705087e-02 4.17007983e-01 -2.30733156e-01 7.08153844e-01 -7.96324730e-01 -1.40684104e+00 -2.05804944e-01 -1.03644669e+00 1.14431775e+00 6.47619009e-01 1.45024911e-01 -9.97836292e-01 3.94189823e-03 3.34005952e-01 8.48632529e-02 4.28663641e-01 1.91612110e-01 8.33332241e-02 -9.14840281e-01 -1.59838378e-01 -8.24941993e-01 1.57479525e-01 -9.01304781e-02 2.01002166e-01 -7.76213884e-01 -3.88642251e-02 -5.50191641e-01 2.45116860e-01 9.37515378e-01 7.98273385e-01 1.01494241e+00 -3.04437727e-01 -4.63342130e-01 5.06154537e-01 1.46721125e+00 9.65608805e-02 6.01373076e-01 2.73166090e-01 1.24707091e+00 4.97092664e-01 3.96917641e-01 5.69416881e-01 1.03487515e+00 4.09313351e-01 2.63851821e-01 -1.17829040e-01 -3.16878378e-01 -3.80882263e-01 1.90434784e-01 7.42208183e-01 -7.77168632e-01 -4.13776577e-01 -8.99579763e-01 5.91196299e-01 -2.17912507e+00 -1.57820821e+00 -6.55111909e-01 2.06707263e+00 1.86025888e-01 -7.65907988e-02 8.25673699e-01 1.02633447e-03 1.44694805e+00 6.38609231e-02 -3.58022034e-01 2.14366049e-01 -6.32655621e-02 -5.56512594e-01 8.81549358e-01 7.17268050e-01 -1.34576154e+00 9.63051200e-01 5.41786051e+00 7.11062253e-01 -2.01460734e-01 2.29000375e-01 9.51678813e-01 -1.10999420e-01 4.98138011e-01 -1.30030811e-01 -1.34641957e+00 1.01632273e+00 5.65531731e-01 -5.70844673e-02 4.59196866e-01 6.61026061e-01 5.45941770e-01 -6.15999579e-01 -7.34447002e-01 1.22843468e+00 -5.16062714e-02 -1.31635749e+00 1.73466608e-01 -2.88395537e-03 5.60120583e-01 -3.63225877e-01 -3.67460221e-01 3.61896634e-01 4.85400110e-01 -6.40926480e-01 1.07688308e+00 8.48132730e-01 5.71337581e-01 -7.13722289e-01 8.85079861e-01 5.85949659e-01 -1.99180532e+00 -6.88996911e-02 -5.84166050e-01 -4.06057596e-01 7.83645213e-01 5.16054034e-01 -2.65274912e-01 2.91051626e-01 8.01669896e-01 7.76880562e-01 -5.54044962e-01 1.28674364e+00 2.06821263e-02 4.85471874e-01 -3.89745891e-01 -1.69838101e-01 1.74606666e-01 -4.36711609e-01 4.25527394e-01 1.61324942e+00 2.22972721e-01 3.10365170e-01 5.58914840e-01 7.38054693e-01 5.10553420e-02 -9.71810445e-02 -3.39349002e-01 6.48966968e-01 5.88527858e-01 1.29264081e+00 -9.98692572e-01 -7.80310214e-01 -7.67686605e-01 8.78953278e-01 3.53249222e-01 4.38841522e-01 -1.35082543e+00 -1.74816865e-02 6.34541035e-01 3.29425126e-01 2.30222434e-01 -3.00324440e-01 -1.11248784e-01 -1.36126232e+00 1.12693831e-01 -1.48196340e-01 4.30152863e-01 -5.09383619e-01 -1.31126058e+00 2.28417143e-01 2.62740105e-01 -1.17194617e+00 3.27015907e-01 -3.62216473e-01 -9.08294022e-01 8.10188174e-01 -1.57642126e+00 -1.16337228e+00 -9.01702046e-01 7.67079413e-01 6.16258085e-01 -1.00589646e-02 6.60042167e-02 9.86814320e-01 -9.86443222e-01 4.14309710e-01 -4.48907793e-01 7.91456163e-01 1.53545171e-01 -1.03715229e+00 5.14313221e-01 1.16840792e+00 -5.10397553e-01 2.95174867e-01 5.18325269e-01 -7.24325418e-01 -7.74467766e-01 -1.38012981e+00 7.98592806e-01 -6.54423535e-01 4.34591413e-01 -3.25066119e-01 -3.90871286e-01 6.05146945e-01 4.16213870e-02 4.09327269e-01 3.58552396e-01 -3.90382707e-01 1.04121044e-01 9.50877517e-02 -1.08813226e+00 6.55081987e-01 1.33371830e+00 2.26195399e-02 2.54883710e-02 3.64574075e-01 6.80119276e-01 -3.81910294e-01 -5.05712032e-01 -1.99462444e-01 5.29909253e-01 -1.05763435e+00 1.36883831e+00 -3.71546090e-01 5.01462102e-01 -7.98888147e-01 -1.57311082e-01 -7.06788182e-01 -4.52428639e-01 -7.83092827e-02 -3.70993555e-01 1.52825975e+00 -2.29287654e-01 -4.85114545e-01 8.82827938e-01 7.38928556e-01 1.08645260e-01 -3.74128133e-01 -9.40389514e-01 -6.21989071e-01 -1.21127941e-01 -1.75982058e-01 8.04585338e-01 5.62609553e-01 -4.85277534e-01 1.96865276e-01 -6.79898977e-01 3.22752655e-01 1.22648442e+00 -4.63246912e-01 1.23226833e+00 -1.04416680e+00 9.74263102e-02 -3.78128946e-01 -5.40139496e-01 -1.38446188e+00 9.06585250e-04 -4.55044329e-01 -5.80278151e-02 -1.57746518e+00 8.14064860e-01 -2.28951529e-01 2.41784871e-01 -3.89129259e-02 -5.18541276e-01 8.34487826e-02 5.37043214e-01 3.18636209e-01 -1.26858222e+00 2.86367625e-01 1.30192018e+00 -2.96312958e-01 5.86899631e-02 1.54048637e-01 -3.29238862e-01 9.94957268e-01 7.83873677e-01 -3.55441451e-01 -1.67019162e-02 -6.70067132e-01 -1.72374561e-01 1.75806329e-01 1.00415516e+00 -1.36958981e+00 6.50901556e-01 -2.17825353e-01 6.44246936e-01 -9.72073913e-01 1.93356216e-01 -6.57644212e-01 -1.63215667e-01 6.85407639e-01 1.62836224e-01 2.72372812e-01 3.71876769e-02 7.60831058e-01 -3.78620885e-02 -1.55959621e-01 7.76049614e-01 -4.66328263e-01 -8.94542396e-01 8.03128779e-01 -3.87133360e-01 3.05495679e-01 1.34780455e+00 -5.09851396e-01 -6.35361433e-01 -1.25520930e-01 -4.96627390e-01 6.41338348e-01 2.56575584e-01 2.77110133e-02 4.88946557e-01 -1.57505178e+00 -9.47877169e-01 -2.51423001e-01 -7.13650137e-02 1.65447071e-01 4.89977807e-01 8.84598792e-01 -8.31578135e-01 4.02185947e-01 -7.50238597e-02 -7.88919508e-01 -1.24304557e+00 6.16561234e-01 4.92100745e-01 -2.01433197e-01 -6.12574279e-01 7.28704095e-01 3.68012220e-01 -1.45503581e-01 1.49173170e-01 -1.86057985e-01 -3.76942933e-01 -6.57390282e-02 9.05745924e-01 1.25275373e+00 -7.03131735e-01 -1.22129059e+00 -4.68007475e-01 6.76767945e-01 9.73796695e-02 1.32230952e-01 8.92594278e-01 -5.02369285e-01 1.66227873e-02 3.30596775e-01 1.01459754e+00 -2.36232474e-01 -1.49085271e+00 -8.99060518e-02 -3.83848101e-01 -9.95917022e-01 -3.21451694e-01 -2.84587387e-02 -1.31932676e+00 9.67967451e-01 5.31080127e-01 -2.30589695e-02 7.30507493e-01 -1.84336275e-01 1.15429425e+00 2.53906287e-02 7.07144976e-01 -1.01610410e+00 -9.55360979e-02 4.27626848e-01 1.48956493e-01 -1.45684671e+00 7.26432949e-02 -5.16362309e-01 -4.70857352e-01 9.98567700e-01 1.02019763e+00 -2.20959768e-01 5.77174604e-01 3.30898374e-01 -4.82758760e-01 -1.66750327e-02 -2.91215777e-01 -5.74042559e-01 -2.20034838e-01 8.38768601e-01 2.15374187e-01 3.64285022e-01 -3.53395902e-02 2.41473675e-01 9.17164534e-02 3.35643232e-01 6.52852774e-01 6.61696255e-01 -6.64500296e-01 -4.15505767e-01 -6.83429003e-01 5.04232109e-01 -9.93829072e-02 -5.16059995e-02 1.73550785e-01 7.14117110e-01 6.78438246e-01 1.23470414e+00 3.55520010e-01 -5.74313775e-02 4.55440462e-01 -4.60148990e-01 4.87842619e-01 -2.73060322e-01 -2.56623149e-01 -3.98567528e-01 -1.02253733e-02 -2.57753253e-01 -7.27730989e-01 -1.02669537e+00 -1.03472722e+00 -1.26584589e+00 -3.04959089e-01 -3.27293903e-01 6.44410774e-02 8.21818292e-01 -2.43152127e-01 5.26891649e-01 3.59180063e-01 -1.15662205e+00 2.27990508e-01 -7.43278921e-01 -3.44432950e-01 4.33729529e-01 3.26524287e-01 -8.42177570e-01 -3.25779796e-01 3.97769630e-01]
[8.282828330993652, -0.42254510521888733]
7220972b-2e5d-4563-a51b-b9b3cc9278be
covidcare-transferring-knowledge-from
2007.08848
null
https://arxiv.org/abs/2007.08848v1
https://arxiv.org/pdf/2007.08848v1.pdf
CovidCare: Transferring Knowledge from Existing EMR to Emerging Epidemic for Interpretable Prognosis
Due to the characteristics of COVID-19, the epidemic develops rapidly and overwhelms health service systems worldwide. Many patients suffer from systemic life-threatening problems and need to be carefully monitored in ICUs. Thus the intelligent prognosis is in an urgent need to assist physicians to take an early intervention, prevent the adverse outcome, and optimize the medical resource allocation. However, in the early stage of the epidemic outbreak, the data available for analysis is limited due to the lack of effective diagnostic mechanisms, rarity of the cases, and privacy concerns. In this paper, we propose a deep-learning-based approach, CovidCare, which leverages the existing electronic medical records to enhance the prognosis for inpatients with emerging infectious diseases. It learns to embed the COVID-19-related medical features based on massive existing EMR data via transfer learning. The transferred parameters are further trained to imitate the teacher model's representation behavior based on knowledge distillation, which embeds the health status more comprehensively in the source dataset. We conduct the length of stay prediction experiments for patients on a real-world COVID-19 dataset. The experiment results indicate that our proposed model consistently outperforms the comparative baseline methods. CovidCare also reveals that, 1) hs-cTnI, hs-CRP and Platelet Counts are the most fatal biomarkers, whose abnormal values usually indicate emergency adverse outcome. 2) Normal values of gamma-GT, AP and eGFR indicate the overall improvement of health. The medical findings extracted by CovidCare are empirically confirmed by human experts and medical literatures.
['Yasha Wang', 'Xinyu Ma', 'Wenjie Ruan', 'Jiangtao Wang', 'Liantao Ma', 'Chaohe Zhang', 'Xianfeng Jiao', 'Wen Tang', 'Junyi Gao', 'Zhihao Yu']
2020-07-17
null
null
null
null
['length-of-stay-prediction']
['medical']
[-2.17565939e-01 -3.00772697e-01 -1.49214402e-01 -2.31714606e-01 -3.58435750e-01 -1.39826730e-01 -2.70028800e-01 4.94438916e-01 -5.44677734e-01 1.00890827e+00 2.68911391e-01 -4.74919289e-01 -5.64290822e-01 -7.67080843e-01 -2.83308923e-01 -9.37519908e-01 -3.78805518e-01 8.68830979e-01 -4.95076627e-01 9.04868320e-02 -2.34987840e-01 2.64446527e-01 -6.70330524e-01 2.76249558e-01 9.83536780e-01 1.17444289e+00 1.28787041e-01 4.44032580e-01 2.70945281e-01 8.56881857e-01 -4.89988476e-01 -7.71694034e-02 3.13641518e-01 -3.45123589e-01 -3.84534657e-01 -4.44536448e-01 -7.08399296e-01 -7.68913448e-01 -5.00520825e-01 7.20887065e-01 7.14399099e-01 -1.74766079e-01 7.19963014e-01 -1.17005098e+00 -4.45686281e-01 3.86824995e-01 -2.94086456e-01 4.03343648e-01 -1.00908414e-01 5.72062314e-01 3.91707599e-01 -4.01458204e-01 1.91413045e-01 7.93369353e-01 8.59680355e-01 6.17225409e-01 -6.83368027e-01 -8.14321399e-01 -1.10385105e-01 1.50166005e-01 -1.33655620e+00 1.06923379e-01 2.91471332e-01 -6.02226317e-01 6.68078184e-01 4.91124205e-03 6.97175622e-01 1.27734888e+00 5.65110624e-01 3.22516322e-01 8.16077471e-01 3.00176412e-01 2.30699610e-02 1.24072477e-01 -1.70159526e-02 6.37250662e-01 3.58378947e-01 1.58829495e-01 1.75811537e-02 -4.84408528e-01 8.01581860e-01 1.23146939e+00 -5.71577132e-01 4.46191709e-03 -1.46178567e+00 7.48046219e-01 3.60861331e-01 1.81772023e-01 -9.40940917e-01 -2.97686696e-01 6.96430087e-01 1.91877559e-01 3.35327029e-01 2.60926008e-01 -1.11969340e+00 -1.99974373e-01 -4.45380718e-01 -2.09790275e-01 5.53037941e-01 5.80109298e-01 3.67647141e-01 -1.55009493e-01 -3.06031227e-01 4.48887646e-01 2.41254225e-01 7.94313669e-01 6.99372947e-01 -6.46899879e-01 4.95217592e-01 9.09025967e-01 1.09372683e-01 -9.36086059e-01 -6.40568316e-01 -4.99525398e-01 -1.47225690e+00 -6.84117734e-01 3.46717867e-03 -7.23261952e-01 -5.62508523e-01 1.64532566e+00 2.07861245e-01 5.87793112e-01 3.56225848e-01 7.98735917e-01 4.79486674e-01 6.35810316e-01 3.97098064e-01 -5.47371387e-01 1.48448622e+00 -4.82567519e-01 -7.07893968e-01 1.79982349e-01 7.73130000e-01 -1.74221963e-01 5.87058425e-01 3.44597161e-01 -6.17515385e-01 -1.30941197e-01 -5.49910188e-01 6.29819691e-01 -3.12940121e-01 1.52319789e-01 3.92696977e-01 7.03404844e-02 -6.91028535e-01 4.99694198e-01 -8.09820175e-01 -3.44833165e-01 7.18777359e-01 4.47251767e-01 -3.12305003e-01 -2.52389520e-01 -1.53240407e+00 6.69497132e-01 5.49733639e-01 2.39036530e-01 -8.68555367e-01 -9.81412590e-01 -5.70912063e-01 9.83442143e-02 1.82244405e-01 -1.27978826e+00 8.28972995e-01 -2.93262482e-01 -9.03785348e-01 5.62141657e-01 2.14686334e-01 -3.25970501e-01 4.50753629e-01 -2.54450172e-01 -4.24787760e-01 3.63326788e-01 -2.01541170e-01 1.74348503e-01 4.83019501e-01 -8.23714972e-01 -5.15477598e-01 -5.46325266e-01 -2.79326767e-01 8.61779898e-02 -6.00809157e-01 -7.69165754e-02 8.50788206e-02 -6.60977423e-01 -5.36007404e-01 -7.42050588e-01 -4.72174764e-01 -3.00670147e-01 -3.01172376e-01 1.69361513e-02 8.64355683e-01 -7.16891229e-01 1.26242650e+00 -2.22918177e+00 -5.65818250e-02 1.24812111e-01 6.23373330e-01 5.87681115e-01 1.68095171e-01 4.46260363e-01 -9.18832347e-02 1.43820435e-01 -3.29614133e-01 1.49294049e-01 -3.63671154e-01 3.20103228e-01 -2.60576248e-01 4.50696975e-01 1.35623768e-01 8.83805215e-01 -1.03792751e+00 -5.98003745e-01 2.06073970e-01 6.74962938e-01 -4.73900020e-01 9.15151775e-01 1.08187065e-01 9.34828222e-01 -9.76797044e-01 6.18179202e-01 4.65962768e-01 -8.28784347e-01 -5.79067913e-04 -2.54993346e-02 3.45711738e-01 -1.60921887e-01 -4.52606589e-01 1.22193277e+00 -3.01610053e-01 3.57563235e-02 4.02833745e-02 -1.26321101e+00 7.43625998e-01 7.48687983e-01 1.09622943e+00 -3.42433572e-01 6.42495632e-01 -3.92326973e-02 1.42412543e-01 -1.07000852e+00 -3.74481350e-01 -3.22440863e-01 1.66313380e-01 5.84804237e-01 -5.63604593e-01 6.00124896e-01 -4.89896536e-01 1.41593382e-01 1.25302958e+00 -5.08276284e-01 4.81762439e-01 -1.09011561e-01 5.01962483e-01 -6.90944120e-02 8.93709064e-01 3.31346959e-01 -3.93668294e-01 2.84548253e-01 1.28129438e-01 -5.94449699e-01 -5.21065772e-01 -1.13880563e+00 -2.86653191e-01 4.59529579e-01 -1.16829112e-01 -2.16836836e-02 -5.81386387e-01 -7.51003981e-01 -2.54277815e-03 3.37524563e-01 -7.22123921e-01 -6.32569015e-01 -5.00982761e-01 -1.20393109e+00 4.86462057e-01 7.31808007e-01 4.10148680e-01 -1.21401894e+00 -1.04710567e+00 4.41193908e-01 -4.10161227e-01 -8.78903568e-01 -3.13726544e-01 5.49669191e-02 -1.05405080e+00 -1.29791045e+00 -8.75487149e-01 -5.42418122e-01 7.36838341e-01 -4.67394758e-03 8.24961901e-01 4.70602393e-01 -5.80747604e-01 3.70341301e-01 -3.39009911e-01 -6.67956293e-01 -2.96962857e-01 3.95749919e-02 4.49588537e-01 1.52115701e-02 7.06739366e-01 -3.87214184e-01 -1.25829613e+00 1.21548645e-01 -9.29296434e-01 -1.28307253e-01 6.84016585e-01 9.97720540e-01 4.64932680e-01 -3.71854939e-02 1.10236776e+00 -8.85092020e-01 7.87914813e-01 -1.17232978e+00 -1.04893267e-01 1.79595947e-01 -1.04035556e+00 -3.02917272e-01 8.82294536e-01 -4.04664338e-01 -6.91178083e-01 -4.00989771e-01 9.22216251e-02 -7.82884181e-01 -2.91341215e-01 6.69289351e-01 1.10095188e-01 8.65429997e-01 6.71104118e-02 2.78374076e-01 1.81838199e-01 -3.59661877e-01 -4.37330961e-01 1.08127093e+00 3.19313943e-01 -3.17276418e-01 4.93922889e-01 2.58076072e-01 -1.65535927e-01 -3.19607139e-01 -7.98355937e-01 -4.98493731e-01 -3.08288395e-01 1.09030135e-01 1.17884338e+00 -1.14329517e+00 -1.22304296e+00 4.87013996e-01 -1.07440734e+00 -1.51999772e-01 -1.51487172e-01 8.94698024e-01 -2.76956171e-01 1.05859034e-01 -9.88936186e-01 -6.13808811e-01 -9.04797494e-01 -1.12515366e+00 7.51966298e-01 1.44580498e-01 -2.91146487e-02 -1.21079588e+00 2.76938826e-01 1.37204885e-01 5.32736361e-01 4.01160061e-01 1.42772973e+00 -1.00275552e+00 -3.24994057e-01 -2.21900389e-01 -4.09275174e-01 4.53143984e-01 6.78056359e-01 -2.08082721e-01 -5.94049096e-01 -4.05656010e-01 1.24127999e-01 -1.72681376e-01 4.96579051e-01 5.15436471e-01 1.43277586e+00 -5.63144267e-01 -5.79286277e-01 8.78571212e-01 1.18440592e+00 6.43057108e-01 3.25198203e-01 6.19751550e-02 7.02110171e-01 4.16966826e-01 4.32286561e-01 1.01702821e+00 6.50788128e-01 -1.27815455e-01 5.37017524e-01 -3.17428678e-01 6.65139616e-01 -1.14680454e-01 1.44375414e-01 1.33697093e+00 -2.05218941e-01 -2.98606068e-01 -1.13588417e+00 4.07457113e-01 -1.76415360e+00 -7.16246128e-01 -6.19775169e-02 1.94947815e+00 1.00018275e+00 -4.18533862e-01 -9.83573794e-02 -3.13088596e-01 6.23665631e-01 -4.40270692e-01 -9.55871105e-01 -1.48918584e-01 1.90644056e-01 5.97225651e-02 2.32787654e-01 -2.10392207e-01 -8.79051447e-01 4.49006557e-01 5.93216419e+00 4.43196297e-02 -1.28241622e+00 1.16689056e-01 9.66003537e-01 -2.33901888e-02 1.79797679e-01 -5.29236972e-01 -3.75914216e-01 7.64913857e-01 1.10655570e+00 -1.47070423e-01 3.93345177e-01 7.11321592e-01 3.13806802e-01 4.69145209e-01 -1.08578372e+00 1.04159379e+00 -2.07594290e-01 -1.11137235e+00 -1.75314605e-01 3.58664095e-02 5.11973202e-01 3.03230673e-01 1.93590641e-01 3.87005091e-01 4.13149565e-01 -1.12446225e+00 -2.21155912e-01 8.96467626e-01 1.14384067e+00 -6.30933344e-01 1.27250862e+00 5.44588506e-01 -8.70287180e-01 -3.02610934e-01 -3.10765773e-01 1.57096654e-01 1.68183088e-01 5.29056191e-01 -1.18056250e+00 4.51017857e-01 8.12141657e-01 8.70566368e-01 -1.32991135e-01 7.60151982e-01 1.07966587e-01 8.46407592e-01 -4.99928445e-02 2.19632804e-01 3.06075048e-02 -3.83904934e-01 2.23300278e-01 1.08767056e+00 5.69253206e-01 6.91773832e-01 3.49953771e-01 6.60013556e-01 -2.07588315e-01 2.56350011e-01 -7.90796340e-01 -1.76909804e-01 3.00734997e-01 1.12930667e+00 -2.52375722e-01 -6.21534467e-01 -3.02480429e-01 6.34269238e-01 2.16309056e-02 5.41849673e-01 -9.09194529e-01 -1.28940731e-01 7.58253932e-01 8.94677415e-02 3.92807983e-02 2.32738674e-01 -1.50420025e-01 -1.29513693e+00 -3.58472943e-01 -1.06475794e+00 7.68315911e-01 -5.27618647e-01 -1.63614190e+00 8.33991826e-01 -6.84204474e-02 -1.24917686e+00 -4.18012768e-01 -4.63292003e-01 -7.59167433e-01 8.55268538e-01 -1.62261558e+00 -5.26493967e-01 -4.25880462e-01 9.80629921e-01 9.59703773e-02 -3.07958722e-01 1.23761106e+00 4.18434650e-01 -1.02449512e+00 4.76381391e-01 2.07099125e-01 4.05972362e-01 8.39341283e-01 -6.66907370e-01 -3.05207700e-01 2.31045857e-01 -9.06888068e-01 9.85162735e-01 2.30995327e-01 -7.51517057e-01 -1.20993602e+00 -1.45160687e+00 5.79386652e-01 -2.80980229e-01 6.43093288e-01 1.08539566e-01 -1.20031369e+00 7.19458044e-01 1.32921919e-01 1.88173622e-01 1.10583615e+00 -3.22608352e-01 -6.37039244e-02 -1.33681491e-01 -1.26051843e+00 2.52399236e-01 7.84747183e-01 -2.82004625e-01 -7.27596939e-01 4.56078500e-01 7.79199660e-01 7.30505884e-02 -1.26917648e+00 7.58333206e-01 2.46168107e-01 -4.06296253e-01 8.58255327e-01 -1.33369279e+00 4.39296126e-01 2.84191594e-02 -5.75115718e-02 -1.51068044e+00 -3.12398642e-01 -3.57530981e-01 -9.60870013e-02 7.94329822e-01 1.12466246e-01 -1.09451413e+00 3.73995692e-01 6.48559093e-01 1.70165479e-01 -1.23400080e+00 -6.83612883e-01 -2.79653430e-01 2.24180873e-02 6.00394718e-02 9.35926795e-01 1.30510485e+00 3.99696045e-02 2.08785817e-01 -4.74155545e-01 1.88763514e-01 3.06249499e-01 1.49250090e-01 4.41746831e-01 -1.42821681e+00 -3.02536637e-01 -6.66333288e-02 -1.00596152e-01 -4.28121477e-01 -4.35921103e-02 -8.54746878e-01 -1.47245005e-01 -1.69384670e+00 5.60868859e-01 -7.02361405e-01 -1.15955901e+00 6.97768331e-01 -5.20614505e-01 -2.07772955e-01 -1.90784827e-01 2.59331226e-01 -4.25017595e-01 6.45863414e-01 1.30401897e+00 -1.70700550e-01 -2.63261944e-01 -1.35463607e-02 -6.50861979e-01 7.82936871e-01 1.11206126e+00 -6.85289741e-01 -5.15784502e-01 -4.08882171e-01 1.48422241e-01 6.17250264e-01 2.28802770e-01 -6.60247266e-01 7.55047128e-02 -5.21163762e-01 5.44257581e-01 -4.60657299e-01 1.36736825e-01 -1.17285490e+00 1.06971107e-01 1.00459325e+00 -3.77915353e-02 5.75168252e-01 7.76024759e-02 6.25206530e-01 -8.74592811e-02 3.20842087e-01 4.90241945e-01 -8.73723850e-02 -8.52645859e-02 9.16973054e-01 -5.12880325e-01 4.23215777e-01 1.21419096e+00 1.94506228e-01 -4.32306468e-01 -2.23268971e-01 -4.45003450e-01 4.62259799e-01 2.00932324e-01 2.24918216e-01 8.64045799e-01 -1.02364826e+00 -9.71579552e-01 3.17814499e-01 1.97919697e-01 2.24234417e-01 5.82862854e-01 1.41966534e+00 -5.45542300e-01 1.80181086e-01 -4.61635888e-02 -5.50652802e-01 -8.98378372e-01 1.07561946e+00 2.21207768e-01 -4.67844278e-01 -8.52657855e-01 4.77168649e-01 6.31915271e-01 -1.69928983e-01 2.23221377e-01 -2.84954965e-01 -9.99335200e-02 -8.39725807e-02 6.59712374e-01 2.53335327e-01 -1.69120938e-01 -4.70518768e-01 -5.70379972e-01 3.09136808e-01 -2.78299898e-01 6.20804846e-01 1.65266812e+00 2.19774507e-02 -2.51512229e-01 2.56860971e-01 1.26997745e+00 -4.08024937e-01 -1.04879677e+00 -2.40369141e-01 -4.44638997e-01 -1.73690543e-01 -1.34705529e-01 -9.24880862e-01 -1.40734017e+00 1.09377038e+00 8.12977016e-01 -1.65210336e-01 1.34910667e+00 -1.20849513e-01 1.26873696e+00 4.70837682e-01 2.51659572e-01 -5.65546870e-01 -4.52713408e-02 2.98207134e-01 3.39051872e-01 -1.22159898e+00 -3.68854165e-01 2.04856202e-01 -8.32345426e-01 8.36732030e-01 4.94732320e-01 1.83682606e-01 1.02103555e+00 3.44666213e-01 5.67797482e-01 -2.74489403e-01 -9.15734768e-01 4.82749045e-01 -2.33880207e-01 5.15556335e-01 -6.32678941e-02 2.89667100e-01 6.74920902e-02 1.12110853e+00 2.94004679e-01 1.62019238e-01 1.14288092e-01 7.70252883e-01 -1.12043865e-01 -8.10535312e-01 -3.01652849e-01 8.57613206e-01 -6.85354531e-01 -3.26311409e-01 -1.06036961e-01 4.56873119e-01 2.11406007e-01 7.94845760e-01 -9.75485593e-02 -5.15682101e-01 1.91627190e-01 2.76678979e-01 -4.30659913e-02 -4.12398666e-01 -5.51657200e-01 -2.17180565e-01 -5.90584397e-01 -2.87440091e-01 -2.14994490e-01 -4.46743101e-01 -1.66417444e+00 -1.47451580e-01 5.20666949e-02 3.08416873e-01 3.51486385e-01 8.27924073e-01 8.77010345e-01 7.85393596e-01 7.56410480e-01 8.82172666e-04 -8.30695152e-01 -9.94455457e-01 -5.63523531e-01 4.27087277e-01 5.81011593e-01 -6.08582735e-01 -3.07518899e-01 7.04094246e-02]
[7.924874305725098, 6.213657855987549]
0abdd4cd-814b-431f-9b25-eebde80ac787
a-cnn-bilstm-model-with-attention-mechanism
2112.13444
null
https://arxiv.org/abs/2112.13444v1
https://arxiv.org/pdf/2112.13444v1.pdf
A CNN-BiLSTM Model with Attention Mechanism for Earthquake Prediction
Earthquakes, as natural phenomena, have continuously caused damage and loss of human life historically. Earthquake prediction is an essential aspect of any society's plans and can increase public preparedness and reduce damage to a great extent. Nevertheless, due to the stochastic character of earthquakes and the challenge of achieving an efficient and dependable model for earthquake prediction, efforts have been insufficient thus far, and new methods are required to solve this problem. Aware of these issues, this paper proposes a novel prediction method based on attention mechanism (AM), convolution neural network (CNN), and bi-directional long short-term memory (BiLSTM) models, which can predict the number and maximum magnitude of earthquakes in each area of mainland China-based on the earthquake catalog of the region. This model takes advantage of LSTM and CNN with an attention mechanism to better focus on effective earthquake characteristics and produce more accurate predictions. Firstly, the zero-order hold technique is applied as pre-processing on earthquake data, making the model's input data more proper. Secondly, to effectively use spatial information and reduce dimensions of input data, the CNN is used to capture the spatial dependencies between earthquake data. Thirdly, the Bi-LSTM layer is employed to capture the temporal dependencies. Fourthly, the AM layer is introduced to highlight its important features to achieve better prediction performance. The results show that the proposed method has better performance and generalize ability than other prediction methods.
['Amin Ramezani', 'Ehsan Jahani', 'Mohammadreza Kavianpour', 'Parisa Kavianpour']
2021-12-26
null
null
null
null
['earthquake-prediction']
['computer-vision']
[-2.25244656e-01 -4.93861258e-01 2.33379647e-01 -1.93847105e-01 -2.80222327e-01 3.63799155e-01 1.51789472e-01 8.15165490e-02 -5.78145802e-01 6.59664571e-01 5.32029688e-01 -2.99345911e-01 -1.70525700e-01 -1.27364063e+00 -3.59640628e-01 -8.31872523e-01 -3.11123461e-01 -1.30264282e-01 3.49671185e-01 -4.35024649e-01 4.85022545e-01 5.59277594e-01 -1.20772183e+00 3.29074502e-01 8.74546885e-01 9.44952786e-01 7.69284785e-01 3.60548317e-01 9.49152857e-02 1.01475501e+00 -5.31164587e-01 -7.66954012e-03 -3.23073357e-01 -2.83548981e-01 -8.12847733e-01 -3.75062406e-01 -7.33592331e-01 -5.63405573e-01 -6.70858622e-01 7.25363433e-01 7.23971844e-01 3.48124713e-01 5.24899125e-01 -5.63827991e-01 -8.15587163e-01 5.84780753e-01 -3.85034919e-01 6.90697491e-01 -5.11076339e-02 1.68114826e-01 7.33536839e-01 -1.00911522e+00 -1.27559736e-01 9.19823647e-01 8.00329208e-01 9.00439322e-02 -4.68784958e-01 -6.48369730e-01 7.31931701e-02 6.24726713e-01 -1.54524338e+00 -3.37702520e-02 8.47470105e-01 -4.28455919e-01 1.28680182e+00 6.74122348e-02 7.35404611e-01 7.37242222e-01 5.99750340e-01 6.42085254e-01 3.23503584e-01 -1.51657760e-01 -1.20647825e-01 -3.91278714e-01 -1.44173056e-01 2.05029160e-01 7.58776907e-04 -3.78127303e-03 -1.87827915e-01 1.77710891e-01 9.18677449e-01 5.66964447e-01 -2.40551174e-01 7.95404017e-01 -1.16885030e+00 6.71631932e-01 9.60588455e-01 8.68492544e-01 -8.69234741e-01 1.29711211e-01 4.77776349e-01 -1.98452383e-01 5.45678735e-01 3.21428418e-01 -4.38979149e-01 -1.77718937e-01 -8.94931555e-01 6.01981319e-02 1.49562970e-01 3.52959305e-01 4.62294906e-01 4.79134351e-01 -2.75495928e-03 8.48302007e-01 7.61632770e-02 5.65206051e-01 8.40789437e-01 -2.69485056e-01 7.84824610e-01 6.47726893e-01 -9.90975425e-02 -1.78196990e+00 -6.76344216e-01 -5.99428475e-01 -1.40923345e+00 -4.55429912e-01 -8.95823687e-02 -3.02190989e-01 -6.08595729e-01 1.55739725e+00 -2.69687027e-01 2.42070839e-01 4.89437170e-02 8.18323433e-01 6.09429479e-01 1.20272791e+00 3.28923672e-01 -8.32510069e-02 1.14093995e+00 -5.50338328e-01 -7.71300197e-01 -4.02361840e-01 8.27867389e-01 -5.02825439e-01 9.74276841e-01 -8.26189592e-02 -8.05230618e-01 -7.73138583e-01 -8.03131402e-01 9.23098400e-02 -5.03559291e-01 2.87362814e-01 3.51410240e-01 -5.41816317e-02 -6.73139095e-01 6.42933547e-01 -1.03161693e+00 -1.73409000e-01 4.34578866e-01 2.90743828e-01 -5.38019883e-03 8.02282989e-02 -1.96067035e+00 1.10221422e+00 7.60251641e-01 7.75777221e-01 -2.85750985e-01 -4.24860865e-01 -7.66142666e-01 4.49238092e-01 -1.96660101e-01 -2.89746583e-01 9.91266131e-01 -5.58604956e-01 -9.10284579e-01 -1.04711773e-02 -7.21394792e-02 -4.23952758e-01 6.85011968e-02 -2.42432266e-01 -6.35787189e-01 -1.87408887e-02 4.19228859e-02 2.76479691e-01 8.37570280e-02 -6.11833990e-01 -7.73486793e-01 -2.22726762e-01 -3.52968544e-01 3.34382564e-01 -8.83507490e-01 4.34503034e-02 -9.54325721e-02 -9.33885992e-01 1.50531530e-01 -4.28092688e-01 -5.57090461e-01 -8.76526952e-01 -3.67864937e-01 -1.65448308e-01 8.07184994e-01 -1.36605132e+00 1.87152183e+00 -2.30335212e+00 -1.77092820e-01 -5.34600439e-03 -1.67166129e-01 4.71735597e-01 1.40114248e-01 7.82790065e-01 -9.74020511e-02 1.27700135e-01 -4.55785573e-01 1.21006824e-01 -4.06443149e-01 3.22576106e-01 -5.84092617e-01 1.34382740e-01 3.41627210e-01 8.11497211e-01 -7.29605138e-01 -4.43677187e-01 3.76757115e-01 6.22245908e-01 -2.27132514e-01 2.55334467e-01 1.81402534e-01 5.69965601e-01 -8.30944180e-01 3.27593535e-01 5.23019791e-01 -1.38495699e-01 -2.32987627e-01 -5.22955460e-03 -4.82545406e-01 2.20620632e-01 -9.15848732e-01 9.92141068e-01 -4.85743791e-01 3.61899734e-01 -6.87343359e-01 -1.19061780e+00 1.21443594e+00 6.43558323e-01 4.53445166e-01 -9.28780735e-01 1.02787837e-01 4.43869799e-01 8.82965922e-02 -9.30423975e-01 4.43677187e-01 -1.31072879e-01 -1.23724520e-01 1.55122370e-01 -5.34176111e-01 2.38093749e-01 6.95790723e-02 -2.28067949e-01 7.68708766e-01 -1.68266565e-01 1.45128593e-01 -1.16885640e-01 5.55196822e-01 -5.01758270e-02 7.50334620e-01 3.66329551e-01 1.82684436e-01 5.76322973e-01 1.36195317e-01 -9.46448326e-01 -1.01744294e+00 -3.86067003e-01 2.25315336e-02 6.79755509e-01 2.42496468e-02 2.22865790e-01 -5.71762383e-01 2.00375076e-02 -3.18016320e-01 8.31986964e-01 -5.87032914e-01 -4.57844734e-01 -9.44927156e-01 -1.13385475e+00 5.41898429e-01 1.01824391e+00 1.06162226e+00 -1.63328218e+00 -8.04674745e-01 5.92389226e-01 -5.09987175e-01 -6.12874985e-01 -2.60638654e-01 -3.88149768e-02 -9.22758937e-01 -6.93707168e-01 -8.98249388e-01 -1.04601789e+00 4.47762579e-01 1.78415507e-01 6.86618030e-01 3.26556206e-01 1.00011393e-01 -3.97626281e-01 -3.15669209e-01 -2.55389184e-01 1.71375498e-01 4.15612340e-01 -9.01204795e-02 -6.71687424e-02 4.01280224e-01 -8.03735316e-01 -7.57371902e-01 9.28569809e-02 -9.64237392e-01 2.25122795e-01 8.66124451e-01 8.25124025e-01 3.26986074e-01 5.57139516e-01 9.70238805e-01 -3.65078598e-01 6.63096488e-01 -8.79108369e-01 -1.76094711e-01 7.86760375e-02 -2.90700942e-01 -2.68223137e-01 1.09639621e+00 -2.86467642e-01 -1.26515388e+00 -1.93560645e-01 -5.30170202e-01 -4.91293035e-02 -6.74414188e-02 1.24654436e+00 -1.20701931e-01 3.60349625e-01 1.49924293e-01 6.81705892e-01 -4.54585969e-01 -5.56885600e-01 -3.31441343e-01 8.15874934e-01 5.28186262e-01 -1.41244084e-01 3.28083813e-01 1.14334241e-01 -1.26363486e-01 -8.69248033e-01 -7.63498902e-01 -2.27832034e-01 -6.82659924e-01 -2.02959433e-01 9.69794571e-01 -8.56889427e-01 -6.25299156e-01 8.76980901e-01 -1.42142475e+00 -2.36954153e-01 1.86597124e-01 8.48848879e-01 -4.71142493e-02 1.41853943e-01 -1.01849282e+00 -8.81483912e-01 -4.15268838e-01 -9.16906714e-01 4.30411309e-01 4.69274133e-01 4.51061688e-02 -1.17351019e+00 -9.41029117e-02 -1.72111303e-01 8.34150732e-01 2.36928210e-01 9.20275629e-01 -5.93480170e-01 -3.65324557e-01 -4.51778471e-01 -3.33160192e-01 3.47278386e-01 1.87851965e-01 -9.99384895e-02 -6.47480667e-01 4.60670888e-02 2.40707025e-01 8.41564015e-02 8.72832119e-01 6.52423084e-01 1.48695350e+00 -5.78587592e-01 -2.45779693e-01 4.19590622e-01 1.45546436e+00 6.51203334e-01 9.32546139e-01 6.27238512e-01 7.35314488e-01 5.13432980e-01 4.98444229e-01 6.55069649e-01 7.33263254e-01 3.07352364e-01 4.35560793e-01 -4.08792078e-01 2.94021457e-01 -2.47114256e-01 1.86857823e-02 1.36467695e+00 -3.99965823e-01 -1.64160505e-01 -1.25733161e+00 9.51925218e-01 -1.72337997e+00 -1.47057831e+00 -4.65432107e-01 2.00569558e+00 6.49193287e-01 9.70528945e-02 -2.62436330e-01 4.33185160e-01 7.52777517e-01 2.71529377e-01 -3.09733391e-01 -2.17469111e-01 -2.44198427e-01 -5.91191985e-02 4.65091079e-01 1.32754460e-01 -1.07330549e+00 6.69718266e-01 5.62158728e+00 8.84476781e-01 -1.54665840e+00 -6.48548752e-02 9.74065721e-01 3.41264486e-01 -5.18604480e-02 -1.94916844e-01 -6.37295723e-01 9.04473007e-01 9.60249603e-01 1.72317903e-02 1.03893057e-01 6.98732078e-01 5.70475101e-01 -3.19664665e-02 -3.13671827e-01 6.48864090e-01 -2.61338234e-01 -1.35332847e+00 -9.33094323e-02 -4.41900752e-02 6.77179694e-01 -7.99531676e-03 1.34414464e-01 3.61508161e-01 -6.39494732e-02 -1.07466817e+00 4.22274709e-01 1.04724431e+00 4.31142360e-01 -1.09275532e+00 1.41951299e+00 7.17750549e-01 -1.49660504e+00 -4.05018896e-01 -5.23550451e-01 -6.72175407e-01 4.07826960e-01 7.83663511e-01 -4.42853421e-01 7.22182751e-01 9.09610808e-01 8.64668310e-01 -3.59484881e-01 1.10630083e+00 -1.60373732e-01 8.89966249e-01 -1.60628155e-01 -1.57012463e-01 4.21091765e-01 -8.74179155e-02 1.64221935e-02 1.22059846e+00 7.13972569e-01 4.18206424e-01 1.39143988e-01 3.95640135e-01 2.00943068e-01 1.58807129e-01 -4.89454091e-01 1.39376163e-01 5.85297465e-01 7.56543696e-01 -5.11839688e-01 -3.32920462e-01 -4.06189650e-01 6.13816381e-01 2.04901218e-01 4.45570737e-01 -9.02433336e-01 -8.06228638e-01 7.19506368e-02 1.16108201e-01 2.89247185e-01 -2.98355252e-01 -6.97829902e-01 -9.12853062e-01 1.34061854e-02 -3.09545577e-01 4.15904135e-01 -8.15083802e-01 -1.05657685e+00 8.40784073e-01 -1.63972795e-01 -1.22491622e+00 -1.82001114e-01 -2.07700312e-01 -1.22355509e+00 1.29982293e+00 -1.55483687e+00 -1.10656214e+00 -1.46125883e-01 3.71891886e-01 5.82116306e-01 1.52945947e-02 7.27309465e-01 5.92636347e-01 -8.33336055e-01 1.38365269e-01 2.85180449e-01 4.82621431e-01 1.58782646e-01 -6.91954076e-01 3.79805744e-01 1.05525362e+00 -5.00110805e-01 4.67223853e-01 4.19441521e-01 -6.95714235e-01 -6.64045811e-01 -1.36807001e+00 1.55517185e+00 2.39510834e-01 6.91995919e-01 2.46780559e-01 -1.33351493e+00 6.55065238e-01 5.92361344e-03 -1.38947770e-01 4.63020205e-01 -1.86393753e-01 4.09942150e-01 -2.32667789e-01 -6.83808506e-01 6.09170258e-01 4.58114415e-01 -3.65563720e-01 -6.60135329e-01 2.74404824e-01 7.39790797e-01 -1.38565764e-01 -9.27284002e-01 5.98536134e-01 1.96799606e-01 -8.89069676e-01 9.63036537e-01 -2.02211618e-01 7.77963698e-01 -1.83374509e-01 6.64158165e-03 -1.24508131e+00 -6.91647291e-01 9.99191850e-02 1.45596281e-01 1.12058818e+00 3.58580023e-01 -8.52947593e-01 3.95286947e-01 5.08178294e-01 -6.06895745e-01 -1.22447348e+00 -8.28481555e-01 -5.63489735e-01 9.64448452e-02 -4.14812833e-01 8.55888247e-01 8.72863352e-01 2.21201070e-02 1.45929411e-01 -6.79130852e-01 2.72918969e-01 6.87471181e-02 3.55079025e-02 1.74940258e-01 -9.34193552e-01 1.62490845e-01 -6.17021441e-01 -2.57713646e-01 -9.80272591e-01 -2.34841853e-01 -3.69862348e-01 -9.73936543e-03 -1.82159317e+00 2.57998630e-02 -4.12384808e-01 -5.79798877e-01 7.29302645e-01 -4.45550263e-01 1.15890950e-01 -1.54153973e-01 5.16819239e-01 -9.30940211e-02 9.03890312e-01 1.26611805e+00 2.98814084e-02 -1.40867576e-01 3.11345935e-01 -4.11656052e-01 7.26365805e-01 1.12706733e+00 -3.55497748e-01 -1.18488260e-01 -8.20419431e-01 2.29983285e-01 4.18366343e-01 2.86880016e-01 -1.32757306e+00 4.22759503e-01 -4.73807871e-01 6.91853046e-01 -9.18418169e-01 1.28901735e-01 -5.26851892e-01 1.17765456e-01 9.44846570e-01 -2.17454672e-01 4.32794005e-01 2.01398358e-01 1.38404146e-01 -7.07711458e-01 -9.14063975e-02 4.36535150e-01 -3.93788852e-02 -7.56773829e-01 5.42318165e-01 -6.06710494e-01 -2.55931109e-01 8.15164208e-01 -1.23730488e-01 2.24322565e-02 -2.18117982e-01 -4.61752385e-01 3.69257212e-01 -9.49554890e-02 1.74566373e-01 8.44602346e-01 -1.43936205e+00 -8.78837168e-01 1.87967017e-01 -2.78975695e-01 1.64449930e-01 1.03325689e+00 9.05680716e-01 -7.52653301e-01 5.43213725e-01 -2.14506775e-01 -9.68952775e-02 -6.13750756e-01 4.88721758e-01 1.94105074e-01 -5.24745584e-01 -5.98161578e-01 7.28883386e-01 1.27695575e-01 -2.50065416e-01 -1.63843423e-01 -3.57863277e-01 -7.99044669e-01 -1.96042433e-01 6.83731616e-01 3.12689573e-01 -4.12699506e-02 -8.04703295e-01 -3.44632655e-01 4.36812878e-01 2.89027661e-01 1.48305967e-01 1.68201041e+00 -1.51765510e-01 -2.38678187e-01 5.83668530e-01 9.68607366e-01 -3.16628247e-01 -1.16315663e+00 -1.42924353e-01 8.68113562e-02 -2.79180914e-01 6.25395998e-02 -5.59656203e-01 -1.18743503e+00 1.37000513e+00 1.50236338e-01 3.40893537e-01 1.36896133e+00 -4.14847672e-01 1.37204897e+00 2.10435107e-01 7.79950172e-02 -1.00051701e+00 -6.53026626e-02 8.20381880e-01 8.94596696e-01 -8.70070398e-01 -3.02473307e-01 1.33454934e-01 -5.76546311e-01 1.25463319e+00 6.11554325e-01 -2.08272800e-01 6.82335436e-01 1.18829697e-01 -2.15646669e-01 1.17120504e-01 -4.78222579e-01 -4.52284068e-02 1.27434939e-01 2.96647638e-01 4.38835770e-01 -2.78217755e-02 -2.82244176e-01 1.08722115e+00 -5.49644306e-02 -2.95477863e-02 1.69288382e-01 7.76997626e-01 -7.99174488e-01 -6.29533827e-01 -4.74421352e-01 5.27899921e-01 -4.64271665e-01 -4.41728354e-01 4.23218101e-01 5.92788279e-01 2.08316684e-01 8.12195659e-01 1.73269540e-01 -7.31120586e-01 2.62361020e-01 -2.82882512e-01 -3.43502402e-01 -3.31726402e-01 -5.61477423e-01 -2.01466724e-01 -3.87737632e-01 -2.69835293e-02 -1.56338602e-01 -4.81592715e-01 -1.52708220e+00 -3.70004982e-01 -3.87344658e-01 3.68001670e-01 4.40464795e-01 1.35046220e+00 3.84304374e-01 8.30242395e-01 9.24734056e-01 -9.08438265e-01 -4.21168506e-01 -1.37717319e+00 -5.58222175e-01 2.89286047e-01 -2.20860876e-02 -4.74859685e-01 -1.93435937e-01 -3.70473077e-04]
[6.843992710113525, 2.69747257232666]
9c5fc76f-d85c-40c2-9b5e-7587b5bbabe4
gesturediffuclip-gesture-diffusion-model-with
2303.14613
null
https://arxiv.org/abs/2303.14613v3
https://arxiv.org/pdf/2303.14613v3.pdf
GestureDiffuCLIP: Gesture Diffusion Model with CLIP Latents
The automatic generation of stylized co-speech gestures has recently received increasing attention. Previous systems typically allow style control via predefined text labels or example motion clips, which are often not flexible enough to convey user intent accurately. In this work, we present GestureDiffuCLIP, a neural network framework for synthesizing realistic, stylized co-speech gestures with flexible style control. We leverage the power of the large-scale Contrastive-Language-Image-Pre-training (CLIP) model and present a novel CLIP-guided mechanism that extracts efficient style representations from multiple input modalities, such as a piece of text, an example motion clip, or a video. Our system learns a latent diffusion model to generate high-quality gestures and infuses the CLIP representations of style into the generator via an adaptive instance normalization (AdaIN) layer. We further devise a gesture-transcript alignment mechanism that ensures a semantically correct gesture generation based on contrastive learning. Our system can also be extended to allow fine-grained style control of individual body parts. We demonstrate an extensive set of examples showing the flexibility and generalizability of our model to a variety of style descriptions. In a user study, we show that our system outperforms the state-of-the-art approaches regarding human likeness, appropriateness, and style correctness.
['Libin Liu', 'Zeyi Zhang', 'Tenglong Ao']
2023-03-26
null
null
null
null
['gesture-generation']
['robots']
[ 5.59004188e-01 -6.31191209e-02 -1.83613777e-01 -6.89341128e-01 -8.11091423e-01 -8.98109555e-01 9.34578061e-01 -6.28806949e-01 -2.80628979e-01 2.28009343e-01 6.65987194e-01 3.55718732e-02 2.83341974e-01 -3.80598634e-01 -5.48736274e-01 -4.64555055e-01 2.77599633e-01 4.99294907e-01 6.42266646e-02 -2.89572895e-01 1.43630192e-01 6.46310866e-01 -1.34017777e+00 6.33770525e-01 4.51387405e-01 6.82023227e-01 1.16233304e-01 9.79555786e-01 -2.85669267e-01 4.54906315e-01 -8.33286285e-01 -2.69106209e-01 3.21885437e-01 -7.85510421e-01 -7.40709603e-01 2.52101094e-01 6.74008369e-01 -6.61749303e-01 -1.78503677e-01 5.82332075e-01 7.02247858e-01 2.15144530e-01 8.95160198e-01 -1.12195086e+00 -7.13626862e-01 7.04349875e-01 -3.83409619e-01 -4.65680093e-01 6.17506623e-01 4.50488120e-01 9.46099818e-01 -6.82833076e-01 1.01970816e+00 1.62460995e+00 3.62098098e-01 1.27085948e+00 -1.19789386e+00 -7.43184566e-01 3.26502144e-01 -4.28315699e-01 -1.13733721e+00 -5.31035483e-01 7.77085960e-01 -3.68764758e-01 7.42764175e-01 4.65262443e-01 8.24148715e-01 1.71528625e+00 -7.77083635e-02 1.20114410e+00 8.22158277e-01 -4.98539776e-01 2.09048152e-01 -4.58461881e-01 -3.20993364e-01 5.58698118e-01 -3.86464179e-01 2.04937384e-01 -7.24815369e-01 -5.54017685e-02 1.28716290e+00 -1.23120965e-02 -9.43165272e-02 -1.45037398e-01 -1.34484470e+00 5.19625306e-01 2.93595761e-01 3.42700005e-01 -9.41445380e-02 4.31073189e-01 2.88094938e-01 3.08089167e-01 2.67743170e-01 5.80860972e-01 -1.69265628e-01 -3.70724469e-01 -1.23128009e+00 5.43731570e-01 8.22209537e-01 1.27762747e+00 2.53861137e-02 2.34384105e-01 -7.35155761e-01 8.57729256e-01 1.19960278e-01 7.12131143e-01 5.98213434e-01 -1.17651176e+00 3.46725643e-01 4.48927581e-01 1.50702655e-01 -5.66898346e-01 -3.14792156e-01 1.14314780e-01 -6.91458046e-01 5.11782706e-01 6.10963523e-01 -1.93487510e-01 -1.20707130e+00 1.90641332e+00 2.90278774e-02 -1.06946640e-01 -1.71797171e-01 1.12701285e+00 6.46491170e-01 5.19633889e-01 4.22995299e-01 1.77520245e-01 1.14599228e+00 -9.93536174e-01 -7.24686623e-01 -1.06381305e-01 3.69368792e-01 -8.55052829e-01 1.76223814e+00 3.83958519e-01 -1.19259608e+00 -4.76257682e-01 -7.88912594e-01 -2.76581228e-01 -1.45693952e-02 3.33359122e-01 5.08510172e-01 4.56544608e-01 -9.11404252e-01 7.04460025e-01 -9.48334694e-01 -5.29615819e-01 3.44386756e-01 4.20255393e-01 -2.67728180e-01 2.04759344e-01 -6.65620983e-01 5.63598156e-01 5.10605462e-02 -2.41208434e-01 -5.94026506e-01 -3.53876621e-01 -9.13710952e-01 -1.49759457e-01 2.77366936e-01 -8.44609797e-01 1.56536865e+00 -1.51324916e+00 -2.25427628e+00 7.78535068e-01 -1.03687704e-01 9.05696228e-02 7.94286788e-01 -5.34984231e-01 -1.05097435e-01 1.38067886e-01 -2.05923602e-01 1.26419163e+00 1.04959762e+00 -1.20520306e+00 -3.73586088e-01 -1.20826721e-01 1.32202730e-01 5.06585002e-01 -2.32297555e-01 2.72431523e-01 -7.82846212e-01 -1.38374841e+00 -5.37064597e-02 -1.35981524e+00 -2.08438158e-01 3.53886634e-01 -6.22570753e-01 -1.10307736e-02 7.37065196e-01 -4.98495162e-01 1.26846254e+00 -2.06388783e+00 6.68700457e-01 3.01120847e-01 -1.19168505e-01 4.42465506e-02 -5.65951049e-01 1.75622076e-01 1.69307634e-01 1.95275486e-01 -2.98952281e-01 -7.24606156e-01 2.30746672e-01 3.27907681e-01 -2.35138461e-01 -1.45441638e-02 2.37062499e-01 1.16719222e+00 -8.23393047e-01 -4.95240986e-01 5.59059978e-02 5.07120371e-01 -7.28640676e-01 5.59771359e-01 -6.59267128e-01 7.70488501e-01 -4.13167477e-01 6.74271464e-01 6.07083067e-02 -9.42776576e-02 2.69303709e-01 -1.75837591e-01 8.73378739e-02 1.18638165e-01 -1.10883391e+00 2.30310464e+00 -5.39982796e-01 4.83126014e-01 -3.19474339e-02 2.48664003e-02 6.93533897e-01 3.26280177e-01 3.70221645e-01 -3.31080735e-01 1.78549021e-01 8.39358121e-02 -1.24777578e-01 -4.84120995e-01 5.60968220e-01 -1.81010127e-01 -4.84045714e-01 8.88645232e-01 -9.65064019e-03 -5.81178010e-01 -3.88349369e-02 6.98946416e-02 8.67549896e-01 8.36876750e-01 2.53301114e-01 5.29532954e-02 9.15636420e-02 -2.67045110e-01 1.11923896e-01 7.26327479e-01 8.66160020e-02 1.21135545e+00 2.87503392e-01 -2.47198969e-01 -1.18570828e+00 -1.01068306e+00 5.21136284e-01 1.58492422e+00 -5.94427139e-02 -3.85581076e-01 -1.04188859e+00 -8.27388167e-01 -2.00469568e-01 6.20590448e-01 -5.31445265e-01 2.05997571e-01 -9.15082991e-01 -1.93185136e-01 8.96103740e-01 8.25228631e-01 2.24325255e-01 -1.46497476e+00 -6.65848732e-01 8.50186497e-02 -3.47503908e-02 -9.83922303e-01 -1.18138683e+00 -2.80687392e-01 -7.04770982e-01 -6.37303531e-01 -1.11427629e+00 -7.07872450e-01 7.02019930e-01 -2.89805949e-01 1.06903887e+00 1.33982733e-01 -6.61332086e-02 4.87870425e-01 -4.24096495e-01 -1.97386935e-01 -7.54956007e-01 2.49702796e-01 6.56002611e-02 2.14292109e-02 9.36757103e-02 -6.13180757e-01 -5.28095007e-01 2.95038640e-01 -1.24988413e+00 4.62602168e-01 5.68367302e-01 8.40927362e-01 5.10943353e-01 -9.43212748e-01 1.39851376e-01 -7.55539119e-01 8.99166822e-01 6.67907223e-02 -2.92550802e-01 1.92513928e-01 -1.75404280e-01 4.78167027e-01 3.41267109e-01 -1.01756406e+00 -1.13759017e+00 4.95031774e-01 -2.19137862e-01 -5.24892151e-01 -3.99805129e-01 -3.94865014e-02 -3.90290469e-01 1.64650723e-01 7.64601529e-01 7.35142604e-02 -2.30117189e-03 -5.62156022e-01 9.90906775e-01 6.89753056e-01 7.01413810e-01 -8.51940274e-01 8.59005332e-01 3.28747362e-01 -2.55043060e-01 -6.53090179e-01 -4.88772303e-01 -1.07453257e-01 -8.41229498e-01 -1.81939363e-01 9.16838467e-01 -6.07717514e-01 -5.63026547e-01 6.00875616e-01 -1.30515242e+00 -8.94074440e-01 -4.18685347e-01 2.18218729e-01 -9.42229986e-01 8.60644802e-02 -8.70982409e-01 -6.13049150e-01 -4.19707566e-01 -1.29414117e+00 1.70675850e+00 6.33705556e-02 -1.05782807e+00 -6.38809562e-01 -6.92288345e-03 -8.37849528e-02 3.75457346e-01 4.01156992e-01 6.08178437e-01 -4.71161306e-01 -4.65751648e-01 7.81261921e-02 8.53617340e-02 -2.70145983e-02 3.92622113e-01 1.31371751e-01 -8.06711614e-01 -3.09281610e-02 -5.49448550e-01 -4.64103222e-01 5.73768258e-01 2.20551968e-01 1.09730685e+00 -6.18167698e-01 -1.76796868e-01 9.09289718e-01 8.18724632e-01 1.97710007e-01 6.18983805e-01 4.81379554e-02 1.03218102e+00 5.60943484e-01 3.43468428e-01 3.41725677e-01 9.42235515e-02 1.10096753e+00 -8.00925046e-02 -1.83117270e-01 -5.16287804e-01 -5.17952979e-01 4.81398225e-01 5.24692535e-01 -3.38301897e-01 -3.46625775e-01 -4.78326321e-01 1.99363023e-01 -1.57981598e+00 -1.16986752e+00 5.04569113e-01 1.78274810e+00 1.22265470e+00 2.28609741e-02 4.99327064e-01 -6.41946048e-02 4.45910901e-01 1.95728615e-01 -6.25335038e-01 -5.94579756e-01 -2.53835525e-02 3.77596051e-01 1.32347941e-01 6.15872979e-01 -1.08383977e+00 1.44296718e+00 6.72599936e+00 7.35509813e-01 -1.43898332e+00 -5.87987937e-02 5.30113697e-01 -6.13468170e-01 -5.03066540e-01 -4.89566505e-01 -5.95652282e-01 3.11476052e-01 4.49858367e-01 2.80045599e-01 6.39240265e-01 8.22521925e-01 3.51412326e-01 3.82100701e-01 -1.45623028e+00 1.08714199e+00 1.49375409e-01 -1.29841781e+00 4.48782980e-01 -1.13337517e-01 7.02733994e-01 -3.81078690e-01 2.03729600e-01 7.33155981e-02 4.49042708e-01 -1.06499088e+00 1.12480485e+00 5.61300576e-01 1.53413951e+00 -3.25589389e-01 -2.92247105e-02 5.59671223e-02 -1.00854051e+00 4.91003059e-02 3.53046983e-01 -1.16080567e-02 3.94744933e-01 -1.53661355e-01 -6.52980924e-01 -3.22561548e-03 3.41049224e-01 4.32306468e-01 -4.00812715e-01 4.67419147e-01 -4.24061239e-01 5.58229864e-01 -4.81670082e-01 -2.09275767e-01 1.85084283e-01 1.34844318e-01 6.12025261e-01 1.64794362e+00 3.78841758e-01 3.39198411e-01 1.92092314e-01 9.42072332e-01 -5.28885238e-02 2.61824131e-01 -5.63059211e-01 -1.95458576e-01 3.90873432e-01 1.04341686e+00 -6.47188962e-01 -4.27070081e-01 -1.72613457e-01 1.67427385e+00 9.71864238e-02 4.96399403e-01 -4.79207993e-01 -1.02366522e-01 8.11016321e-01 7.84304962e-02 1.77864358e-01 -4.66881603e-01 -5.71730852e-01 -1.27505696e+00 -1.12709202e-01 -1.14056289e+00 7.65064806e-02 -8.39436710e-01 -1.22017145e+00 8.06921899e-01 1.21110998e-01 -1.38996148e+00 -8.87830496e-01 -5.83048403e-01 -6.69046819e-01 7.83858120e-01 -7.32187212e-01 -1.62754846e+00 -3.29212487e-01 5.15907764e-01 9.14483309e-01 -2.69430518e-01 9.64578331e-01 -2.78203916e-02 -1.96355477e-01 9.13675785e-01 -4.74001497e-01 3.21091384e-01 9.20771480e-01 -1.26755893e+00 8.19520414e-01 5.54243624e-01 3.33563298e-01 7.07562625e-01 7.10197747e-01 -7.99734652e-01 -1.18250239e+00 -9.27552044e-01 5.71390390e-01 -5.40111601e-01 3.22150022e-01 -6.34687662e-01 -4.98311520e-01 6.77705348e-01 2.09393039e-01 -2.96538383e-01 6.24157786e-01 -1.24057397e-01 -4.90404934e-01 2.76238352e-01 -9.90361512e-01 1.20017469e+00 1.53155220e+00 -4.53098625e-01 -6.37161791e-01 -3.76966558e-02 7.08987892e-01 -7.51917958e-01 -4.97290641e-01 1.31801635e-01 1.18371344e+00 -6.59365237e-01 7.81664848e-01 -7.78802276e-01 5.59700966e-01 -1.50812387e-01 -6.17978051e-02 -1.32050109e+00 -2.70969212e-01 -1.04262519e+00 2.41799690e-02 1.16084254e+00 3.55711520e-01 1.60834074e-01 8.05675924e-01 8.04518402e-01 -7.11047575e-02 -5.21621823e-01 -4.88414407e-01 -5.88859081e-01 3.74357551e-02 -5.94442070e-01 9.28659499e-01 7.03958035e-01 1.20378323e-01 3.06614220e-01 -5.51815331e-01 -2.05455646e-01 2.49051347e-01 1.09190464e-01 1.03592312e+00 -8.58290493e-01 -5.99239707e-01 -7.84323275e-01 -6.94031119e-02 -1.44643652e+00 2.72473246e-01 -7.79727340e-01 2.18612641e-01 -1.30516541e+00 -3.56714949e-02 -4.22565997e-01 1.98146254e-01 7.51119435e-01 -1.54205933e-01 4.34985995e-01 7.25603521e-01 3.58302444e-01 -4.12544161e-01 4.73270237e-01 1.63855326e+00 -2.19606340e-01 -7.60388613e-01 3.49780060e-02 -5.39690793e-01 8.38001609e-01 6.52735054e-01 -2.22142130e-01 -4.05430555e-01 -5.84601223e-01 -1.70542151e-01 1.11582816e-01 1.12861775e-01 -9.06426311e-01 -1.02463722e-01 -5.04614055e-01 4.95024294e-01 -7.25654513e-02 5.51146328e-01 -6.14217341e-01 1.12448484e-01 1.51519865e-01 -9.30444181e-01 1.94124207e-01 9.66423228e-02 2.29740977e-01 1.37165844e-01 1.63398057e-01 5.12593687e-01 -2.17612848e-01 -5.27457476e-01 3.86906594e-01 -3.74183148e-01 -1.53738558e-01 6.15158319e-01 -1.77543998e-01 1.07719786e-01 -7.48216987e-01 -9.30510461e-01 -1.22756422e-01 7.25192070e-01 8.06589901e-01 6.99075937e-01 -1.74878693e+00 -6.53030276e-01 3.26465160e-01 1.67672530e-01 -1.35394588e-01 -2.40870044e-01 1.45170480e-01 -4.53419566e-01 2.44338512e-02 -3.42805922e-01 -5.85437953e-01 -1.36299455e+00 2.45321557e-01 3.03350508e-01 -2.96781790e-02 -7.19480991e-01 6.77817941e-01 2.96126097e-01 -2.93400854e-01 3.10514271e-01 -6.63770378e-01 2.96034198e-02 -2.81284392e-01 6.98424041e-01 2.42315661e-02 -4.51440454e-01 -7.77749717e-01 -4.74715941e-02 7.56538987e-01 2.46328935e-01 -9.18829918e-01 1.11612642e+00 8.02197456e-02 3.38501722e-01 4.29409325e-01 8.54397655e-01 1.62237152e-01 -1.73152196e+00 1.08466502e-02 -9.18840319e-02 -2.72389621e-01 -5.52374542e-01 -1.12374389e+00 -9.22632754e-01 8.04375708e-01 4.97923195e-01 -4.02745456e-01 1.15264094e+00 2.94370367e-03 9.32831645e-01 3.37458819e-01 2.65688777e-01 -9.94811177e-01 5.06941378e-01 5.45232236e-01 1.35314620e+00 -8.45099986e-01 -4.31641191e-01 -1.60524741e-01 -1.04267323e+00 1.15786099e+00 5.33344626e-01 -8.65571871e-02 2.65932888e-01 6.69103026e-01 4.98051733e-01 2.59844176e-02 -4.00382876e-01 -5.05821407e-02 6.96897209e-01 6.68331385e-01 7.18652010e-01 2.77506083e-01 -2.97534198e-01 6.99757934e-01 -5.22409201e-01 2.08777368e-01 2.14827731e-01 9.26778316e-01 -1.08033121e-01 -1.48079240e+00 -3.71109158e-01 1.75904185e-01 -3.42444211e-01 -1.16110928e-01 -7.51673698e-01 6.18889749e-01 -3.43717150e-02 6.56161964e-01 6.30547926e-02 -5.64588130e-01 4.48145092e-01 4.07325685e-01 8.48023295e-01 -7.41972268e-01 -7.34543025e-01 4.24187750e-01 1.08115617e-02 -8.49381387e-01 -4.17045385e-01 -6.63824737e-01 -1.34010637e+00 -3.02848481e-02 1.77504629e-01 -5.03368378e-01 4.89068925e-01 1.01529109e+00 3.06563169e-01 3.31155837e-01 1.42949119e-01 -1.52081549e+00 -3.82294625e-01 -1.12829590e+00 -3.38091552e-01 9.58374381e-01 1.72723711e-01 -5.60687840e-01 -1.97481979e-02 5.37288666e-01]
[5.667873859405518, -0.13692258298397064]
66221ca0-3020-43a8-ab4a-7af2c136176e
achieving-diversity-in-objective-space-for
2306.13780
null
https://arxiv.org/abs/2306.13780v1
https://arxiv.org/pdf/2306.13780v1.pdf
Achieving Diversity in Objective Space for Sample-efficient Search of Multiobjective Optimization Problems
Efficiently solving multi-objective optimization problems for simulation optimization of important scientific and engineering applications such as materials design is becoming an increasingly important research topic. This is due largely to the expensive costs associated with said applications, and the resulting need for sample-efficient, multiobjective optimization methods that efficiently explore the Pareto frontier to expose a promising set of design solutions. We propose moving away from using explicit optimization to identify the Pareto frontier and instead suggest searching for a diverse set of outcomes that satisfy user-specified performance criteria. This method presents decision makers with a robust pool of promising design decisions and helps them better understand the space of good solutions. To achieve this outcome, we introduce the Likelihood of Metric Satisfaction (LMS) acquisition function, analyze its behavior and properties, and demonstrate its viability on various problems.
['Michael McCourt', 'Bolong Cheng', 'Eric Hans Lee']
2023-06-23
null
null
null
null
['multiobjective-optimization']
['methodology']
[ 1.30527526e-01 -5.71342647e-01 -2.56641805e-01 -2.09958062e-01 -9.96624529e-01 -5.50728798e-01 -4.53977957e-02 3.35429877e-01 -3.10845971e-01 9.70395148e-01 -2.41813455e-02 -4.45206165e-01 -9.21149075e-01 -6.22017086e-01 -3.86704415e-01 -7.26625741e-01 -5.04574813e-02 7.52346754e-01 -1.85300216e-01 -8.20828453e-02 7.24199533e-01 6.64894164e-01 -1.68431687e+00 -1.90700606e-01 1.34144664e+00 1.02356648e+00 1.59745604e-01 3.51453304e-01 -3.17454524e-02 -1.69447169e-01 -7.83292949e-01 -1.17658220e-01 1.36773825e-01 -4.96057391e-01 -6.11802638e-01 -5.05034663e-02 -3.82128179e-01 -8.64744037e-02 7.78958440e-01 1.00051701e+00 7.02402890e-01 5.03598273e-01 6.94545269e-01 -1.28795016e+00 -1.48601383e-01 2.04881743e-01 -7.83827484e-01 1.29766360e-01 2.75414377e-01 3.06977510e-01 1.18615246e+00 -8.14215481e-01 4.41924274e-01 1.09241569e+00 3.33700359e-01 1.44784078e-01 -1.43028414e+00 -5.12854636e-01 -5.38093336e-02 1.24799572e-01 -1.58833766e+00 -3.66996139e-01 7.50048101e-01 -2.91217655e-01 7.21971154e-01 6.55601442e-01 7.15317130e-01 3.72530013e-01 4.13637668e-01 5.00521421e-01 1.04202247e+00 -3.60608131e-01 7.40506411e-01 1.87624410e-01 3.25991958e-02 3.65519792e-01 7.43243277e-01 1.42445967e-01 -3.78272086e-01 -3.18951219e-01 1.94679871e-01 -2.92893440e-01 -3.06428403e-01 -6.02239311e-01 -7.84429550e-01 7.70778120e-01 1.42353147e-01 1.35667875e-01 -6.46232545e-01 8.26029256e-02 -2.60347147e-02 1.37385026e-01 1.59612283e-01 1.19004297e+00 -3.37088078e-01 -3.45915020e-01 -8.63526762e-01 5.63759625e-01 6.84831262e-01 4.47318643e-01 6.28443062e-01 1.17466927e-01 -7.17312992e-02 7.57481337e-01 4.75061834e-01 2.02322558e-01 -7.06796646e-02 -9.46588814e-01 3.78813326e-01 7.55501926e-01 5.90714335e-01 -8.93761337e-01 -4.19491738e-01 -9.71802711e-01 -1.97847411e-01 6.97182059e-01 4.18573171e-01 -3.95570070e-01 -5.70787489e-01 1.32725501e+00 4.26249444e-01 -6.63685381e-01 -1.31178126e-01 1.05848014e+00 -2.54272670e-03 7.78853893e-01 -2.73638143e-04 -4.42971528e-01 1.00827360e+00 -4.95828420e-01 -4.85797197e-01 -1.32714137e-01 4.14666444e-01 -8.33775938e-01 1.09850681e+00 4.44773048e-01 -1.18332613e+00 9.97024495e-03 -1.19753528e+00 5.56158662e-01 -6.56345934e-02 -8.99600051e-03 5.78860760e-01 8.60136807e-01 -8.01104426e-01 7.14940369e-01 -6.01218879e-01 -8.43720064e-02 5.15603542e-01 7.30157375e-01 3.15899253e-01 -1.26091748e-01 -6.20085418e-01 1.07152009e+00 3.49989772e-01 1.85296535e-01 -6.71135724e-01 -8.15648615e-01 -4.56291825e-01 3.56880754e-01 7.07157075e-01 -6.55643702e-01 7.74305582e-01 -9.14612591e-01 -1.51899171e+00 4.42448445e-02 -2.48469189e-01 1.57567292e-01 4.63604897e-01 8.38934910e-03 -2.45159239e-01 -1.49344116e-01 -1.01602189e-01 1.10587843e-01 5.39612293e-01 -1.29183781e+00 -5.56866348e-01 -1.50355116e-01 -1.45237669e-01 3.35057259e-01 -4.42935616e-01 3.64535064e-01 5.75222559e-02 -3.33307624e-01 -1.47892553e-02 -7.26008236e-01 -7.84441590e-01 -3.39514345e-01 -3.71158928e-01 -8.74217134e-03 4.17738289e-01 -4.78326738e-01 1.58521819e+00 -1.76074708e+00 2.57920265e-01 8.45940530e-01 -1.25870049e-01 -9.78121907e-02 -5.73910624e-02 4.88626897e-01 3.00625473e-01 3.88620585e-01 -6.07472733e-02 -1.32807449e-03 9.08410698e-02 -1.22223176e-01 2.38647744e-01 4.29000169e-01 3.28688264e-01 7.39058375e-01 -8.02118599e-01 -1.93816945e-01 1.86447397e-01 1.69229191e-02 -7.72945404e-01 9.10644159e-02 -2.87920177e-01 4.63066101e-01 -9.80562925e-01 1.08431673e+00 3.82858604e-01 -5.28357267e-01 1.37767956e-01 -1.76828563e-01 -4.55110103e-01 2.38110051e-02 -1.50436711e+00 1.04640353e+00 -4.64116365e-01 2.78024435e-01 1.90067485e-01 -1.12025249e+00 9.69057381e-01 2.80495454e-02 7.52194166e-01 -5.89539349e-01 4.45158601e-01 5.89903891e-01 2.70463586e-01 -1.86710119e-01 5.06855488e-01 -5.14943480e-01 -7.58678094e-02 4.28853035e-01 -3.42732728e-01 -1.98830128e-01 3.32295537e-01 -5.91111660e-01 9.54696059e-01 -5.64279370e-02 1.61973253e-01 -7.56775081e-01 3.90484393e-01 2.43874460e-01 7.85716474e-01 3.94500583e-01 -2.06729528e-02 3.90099525e-01 3.00203562e-01 -1.35112852e-01 -9.45580602e-01 -8.68435264e-01 -9.51927677e-02 6.25784636e-01 3.12963247e-01 -9.35969427e-02 -1.06042877e-01 -2.20544666e-01 2.09091976e-01 1.00675035e+00 -2.48052552e-01 -8.26091170e-02 -5.32923520e-01 -9.68573153e-01 -2.55795121e-01 3.44424486e-01 -7.65926838e-02 -6.17251515e-01 -1.10068011e+00 4.55377519e-01 2.23683435e-02 -3.40711027e-01 -2.46011645e-01 2.86490589e-01 -8.80200803e-01 -9.06129301e-01 -8.16001892e-01 -4.15003389e-01 8.34959805e-01 -5.77169731e-02 1.17753601e+00 1.06328987e-01 -5.93826771e-01 2.63135403e-01 -7.98624530e-02 -3.99937332e-01 -3.60278636e-01 -1.20851144e-01 -3.52048576e-02 -1.25737220e-01 -1.02348149e-01 -4.84529138e-01 -5.28638303e-01 7.86986649e-01 -5.68132997e-01 -2.50500649e-01 5.81890583e-01 7.43545413e-01 7.21128881e-01 5.01571834e-01 9.46549058e-01 -1.06494084e-01 8.65091324e-01 -5.16217887e-01 -1.16946948e+00 6.84985757e-01 -9.99399900e-01 4.14898366e-01 5.38187921e-01 -3.59914750e-01 -7.90742099e-01 -2.78132498e-01 8.91936719e-02 -1.72735840e-01 2.66876459e-01 8.45662951e-01 -3.40720654e-01 -2.99956590e-01 4.14331079e-01 -3.92822891e-01 3.05518624e-03 -4.25890654e-01 6.58125728e-02 2.64283389e-01 -4.19817073e-03 -9.56462204e-01 7.08971739e-01 1.03565648e-01 5.51388979e-01 -7.27811098e-01 -2.96644539e-01 -3.73061687e-01 9.49439928e-02 -6.69035494e-01 4.88824338e-01 -1.72888935e-01 -9.81441736e-01 -1.69321954e-01 -7.60065794e-01 9.88917053e-02 -3.93773317e-01 6.28830910e-01 -4.00175273e-01 -4.65693474e-02 3.10867637e-01 -1.16345739e+00 -1.97423503e-01 -1.53452122e+00 5.85341632e-01 5.43460429e-01 -8.25027823e-01 -8.84256661e-01 -2.09064167e-02 2.33003259e-01 6.14711702e-01 6.61520600e-01 1.20608485e+00 -3.03003877e-01 -8.49371195e-01 -2.05338687e-01 1.02199577e-01 -8.80061369e-03 2.32027471e-01 2.23681271e-01 -2.03586578e-01 -5.46516895e-01 -7.35085160e-02 -9.80508998e-02 2.63513744e-01 9.30692255e-01 1.01340938e+00 -3.59066486e-01 -4.82203335e-01 6.11773789e-01 1.79806173e+00 7.37716019e-01 2.82575518e-01 6.20929658e-01 1.18559197e-01 1.02167773e+00 8.84898722e-01 7.94068217e-01 3.99474055e-02 7.35903859e-01 3.34114134e-01 1.26190921e-02 3.87655079e-01 2.03600198e-01 6.67682290e-02 2.46325076e-01 -6.29241094e-02 -6.42909825e-01 -1.02140689e+00 6.52020931e-01 -1.77299345e+00 -6.21528625e-01 2.15300262e-01 2.38198709e+00 4.33149785e-01 1.76495224e-01 3.81339341e-01 1.92984194e-01 7.48869359e-01 -2.07628801e-01 -8.02551925e-01 -7.06884861e-01 -9.46309939e-02 3.81063551e-01 6.72019660e-01 3.37001085e-01 -4.92490798e-01 1.30805701e-01 6.74253130e+00 8.17572474e-01 -9.88652289e-01 -4.29183662e-01 8.96581113e-01 -4.63871479e-01 -8.42225075e-01 2.17719376e-01 -5.44085920e-01 3.67460698e-01 9.12952960e-01 -9.48425293e-01 5.53393483e-01 5.71134806e-01 6.93605542e-01 -3.37002814e-01 -8.58291924e-01 6.51119113e-01 -4.25785989e-01 -1.57225168e+00 -2.52920747e-01 4.02335048e-01 9.90164459e-01 -6.06091678e-01 1.32094830e-01 -2.25400656e-01 2.32961491e-01 -1.33491611e+00 8.87301028e-01 3.88636917e-01 4.68316048e-01 -1.17959273e+00 4.48110253e-01 -2.74916179e-02 -1.08674228e+00 -5.28954506e-01 3.61752547e-02 -2.49998700e-02 6.06491745e-01 6.77906692e-01 -7.36449659e-01 5.71794569e-01 5.16846061e-01 3.56050394e-02 -9.74715054e-02 1.69639719e+00 1.46112472e-01 5.28861344e-01 -6.32364869e-01 -7.88487136e-01 2.23223284e-01 -4.90977228e-01 7.90131986e-01 4.90344763e-01 7.85168052e-01 -1.12994470e-01 1.26383290e-01 1.17658043e+00 2.65681684e-01 3.13688368e-01 -1.92048609e-01 -4.07578737e-01 5.93842626e-01 9.70898509e-01 -1.05905056e+00 2.77813047e-01 -8.80307780e-05 1.51228189e-01 -2.03686386e-01 4.00721401e-01 -7.10611582e-01 -4.44392920e-01 7.65274644e-01 -7.24652130e-03 2.69505888e-01 -3.79817396e-01 -9.94984031e-01 -3.70882988e-01 1.52823672e-01 -8.50549042e-01 3.27295512e-01 -3.14946234e-01 -8.99378538e-01 9.07558352e-02 1.34216383e-01 -1.15426481e+00 -2.68947542e-01 -4.09152836e-01 -8.40469182e-01 1.02541435e+00 -1.34680271e+00 -4.88301575e-01 7.53505528e-03 -4.12991680e-02 4.42997932e-01 -1.44481119e-02 3.19888979e-01 3.26057345e-01 -6.19022310e-01 5.32633722e-01 5.27403057e-01 -7.95731664e-01 2.12639526e-01 -9.80218053e-01 -1.48762092e-01 7.56324112e-01 -4.81589884e-01 4.66893166e-01 1.21712983e+00 -6.31547928e-01 -1.65475750e+00 -4.77614015e-01 5.81577718e-01 -1.94697767e-01 5.58446646e-01 1.42349645e-01 -5.97613752e-01 -2.71018803e-01 -2.56118596e-01 -7.05687582e-01 6.23018026e-01 2.17856929e-01 6.99030161e-01 -2.32490629e-01 -1.23276508e+00 8.72843504e-01 7.36068070e-01 2.33402923e-01 -3.84612828e-02 -1.96625274e-02 2.88236439e-01 -7.00762272e-02 -9.58442807e-01 7.56184876e-01 6.82844341e-01 -6.16717875e-01 9.71575022e-01 -4.34869558e-01 4.56031114e-01 -4.09221739e-01 -1.36943668e-01 -1.39946282e+00 -2.15149462e-01 -9.03948367e-01 1.46113589e-01 1.30824161e+00 7.37222612e-01 -7.15605736e-01 9.38182592e-01 1.02038097e+00 -1.56127751e-01 -1.43124056e+00 -1.04568422e+00 -1.08615696e+00 -1.17433676e-03 -1.54503658e-01 8.88132036e-01 4.14019853e-01 -2.31134683e-01 7.20913336e-02 -7.73871392e-02 1.26323596e-01 8.63576353e-01 6.38009608e-01 5.11332750e-01 -1.05884826e+00 -3.88371944e-01 -9.06431377e-01 -4.03456986e-02 -6.71243668e-01 -2.88730323e-01 -4.55519855e-01 2.13928267e-01 -1.61923015e+00 -8.03707317e-02 -8.68385196e-01 -3.44206393e-01 -4.05800603e-02 -2.30959088e-01 -2.77377933e-01 3.17070261e-02 -2.94793341e-02 -3.91592205e-01 7.40084946e-01 1.07013369e+00 7.13577569e-02 -4.50959116e-01 2.68982261e-01 -1.10602498e+00 4.60557133e-01 7.35363901e-01 -4.52403784e-01 -3.69859606e-01 -1.09653652e-01 4.42831218e-01 2.55894154e-01 8.57603699e-02 -8.95775735e-01 -1.56928226e-01 -9.22108829e-01 3.25788915e-01 -5.97357512e-01 2.13771001e-01 -8.21290612e-01 4.98194873e-01 4.51052070e-01 -1.20411113e-01 3.05056512e-01 3.13347071e-01 3.30305010e-01 -1.49215441e-02 -5.59073269e-01 7.27811694e-01 3.10008556e-01 -6.30758941e-01 -4.70309556e-02 -3.05254668e-01 -3.75843905e-02 1.18692064e+00 -5.46961725e-01 -5.04573397e-02 -2.28154972e-01 -2.62573928e-01 7.33957052e-01 7.76054740e-01 2.04834178e-01 5.54370224e-01 -1.10391486e+00 -5.06330371e-01 -1.10617273e-01 -9.89265665e-02 -2.78340220e-01 3.86898853e-02 6.17787957e-01 -4.57078904e-01 2.69324988e-01 -1.64288998e-01 -4.20907229e-01 -1.10496247e+00 1.01574689e-01 2.81931788e-01 -3.18697989e-01 -1.41544566e-01 8.81560922e-01 -2.37229511e-01 1.07272714e-01 -1.59972794e-02 -1.71329424e-01 3.62721719e-02 -1.65768212e-03 1.20830104e-01 1.09138429e+00 6.75156293e-03 -1.14104800e-01 -7.68181384e-01 5.18599510e-01 5.16608715e-01 -3.53386492e-01 1.45202005e+00 7.79461581e-03 6.86038211e-02 9.07361135e-02 1.24757028e+00 4.28319573e-02 -1.20723784e+00 5.23324788e-01 4.45991188e-01 -8.73939693e-01 2.45699346e-01 -1.00786376e+00 -7.71959484e-01 3.19866210e-01 4.79882538e-01 2.83832550e-02 1.23917687e+00 -1.97401837e-01 6.28016353e-01 6.31554127e-02 3.78461957e-01 -1.35560668e+00 1.61611900e-01 2.78534293e-02 8.74455035e-01 -8.99935961e-01 4.18182433e-01 -1.49605438e-01 -3.03095430e-01 1.09260893e+00 4.72325593e-01 1.27338395e-01 4.50182766e-01 1.54537007e-01 -1.78499132e-01 -2.28425279e-01 -6.32530868e-01 -6.65734261e-02 4.07555461e-01 2.91572511e-01 1.19504757e-01 3.17673050e-02 -7.96224058e-01 3.85504216e-01 -9.95151582e-04 -2.23717749e-01 2.55730838e-01 1.30287600e+00 -6.28327787e-01 -1.41781795e+00 -7.78771579e-01 8.15164328e-01 -3.50222528e-01 2.13396579e-01 -4.52248678e-02 5.63068449e-01 -1.95757374e-01 1.25749636e+00 -1.89360395e-01 -1.02517866e-01 4.56470937e-01 -2.36327112e-01 3.43361229e-01 -4.53739733e-01 -5.52347779e-01 9.34641734e-02 4.65411305e-01 -3.36412609e-01 -9.53228846e-02 -8.87911260e-01 -1.08342731e+00 -2.64907867e-01 -8.03783357e-01 4.40363616e-01 9.73015130e-01 9.56194520e-01 4.02569473e-01 7.65276372e-01 9.72650468e-01 -6.61851704e-01 -7.54271090e-01 -1.84165302e-03 -3.36946279e-01 -5.58037460e-02 1.39025301e-01 -9.94665504e-01 -2.67104954e-01 -7.98988640e-01]
[5.876138687133789, 3.575511932373047]
91805cec-e934-40f4-ab42-0f266e5e88cb
a-tutorial-on-vaes-from-bayes-rule-to
2006.10273
null
https://arxiv.org/abs/2006.10273v2
https://arxiv.org/pdf/2006.10273v2.pdf
A Tutorial on VAEs: From Bayes' Rule to Lossless Compression
The Variational Auto-Encoder (VAE) is a simple, efficient, and popular deep maximum likelihood model. Though usage of VAEs is widespread, the derivation of the VAE is not as widely understood. In this tutorial, we will provide an overview of the VAE and a tour through various derivations and interpretations of the VAE objective. From a probabilistic standpoint, we will examine the VAE through the lens of Bayes' Rule, importance sampling, and the change-of-variables formula. From an information theoretic standpoint, we will examine the VAE through the lens of lossless compression and transmission through a noisy channel. We will then identify two common misconceptions over the VAE formulation and their practical consequences. Finally, we will visualize the capabilities and limitations of VAEs using a code example (with an accompanying Jupyter notebook) on toy 2D data.
['Ronald Yu']
2020-06-18
null
null
null
null
['misconceptions']
['miscellaneous']
[ 2.33878851e-01 1.21127404e-01 -3.08171269e-02 -2.88221836e-01 -8.71172428e-01 -1.63777113e-01 4.32007909e-01 -2.41310418e-01 -2.29641676e-01 9.27341104e-01 2.54253149e-01 -6.02174520e-01 -5.50594151e-01 -5.46457052e-01 -7.26939738e-01 -7.17930496e-01 -4.56184387e-01 5.33199161e-02 -2.38190562e-01 2.01908499e-01 1.97646752e-01 1.65047735e-01 -1.22520351e+00 -1.22460112e-01 5.60382247e-01 1.00212538e+00 3.62107933e-01 7.26631999e-01 2.13257387e-01 9.16880727e-01 -6.99796557e-01 -6.96047544e-01 3.26631874e-01 -5.36655545e-01 -5.34615397e-01 1.06137656e-01 -1.21765345e-01 -8.86693120e-01 -8.35387766e-01 1.07979214e+00 5.80635905e-01 2.61432230e-02 8.63773286e-01 -1.55563474e+00 -5.51932096e-01 4.04383421e-01 -4.40406591e-01 2.82118767e-01 1.96259633e-01 -1.41387239e-01 1.20659947e+00 -7.16730833e-01 5.63409448e-01 1.19175661e+00 7.16222227e-01 2.82722145e-01 -1.04840255e+00 -3.82641166e-01 -4.35963362e-01 2.55584687e-01 -1.84108484e+00 -5.11138201e-01 5.34432113e-01 -4.17367846e-01 8.61199439e-01 1.67384192e-01 5.97789884e-01 1.15770388e+00 5.90527952e-01 1.17085373e+00 7.95651734e-01 -4.84416455e-01 4.07200754e-01 2.26725340e-01 -8.17780793e-02 6.71687663e-01 3.86052698e-01 3.55111003e-01 -8.67240489e-01 -5.14601707e-01 8.81817460e-01 -3.77365947e-01 -4.00787085e-01 -7.30908692e-01 -7.14369893e-01 1.12418914e+00 -1.34814933e-01 -2.11251065e-01 -2.19247609e-01 4.54285502e-01 4.18010391e-02 1.91540375e-01 3.32880050e-01 -6.20553643e-02 -2.55344301e-01 -2.90031731e-01 -1.15543759e+00 2.60287821e-01 9.44370210e-01 1.19836020e+00 3.10771316e-01 6.01282120e-02 9.97624127e-04 6.94036186e-01 1.12189436e+00 5.97936273e-01 -1.80984199e-01 -1.31895435e+00 2.08088607e-01 -5.72077870e-01 8.77756998e-02 -8.24246287e-01 4.83497716e-02 -4.96048063e-01 -6.91165388e-01 4.02889729e-01 1.16084747e-01 -4.04867053e-01 -4.93350059e-01 1.71468747e+00 -1.76118940e-01 -6.41616657e-02 1.56382650e-01 7.68474936e-01 4.69318569e-01 7.95323431e-01 -2.01813638e-01 -6.72358036e-01 9.54720378e-01 -4.12656754e-01 -1.08075559e+00 2.51354091e-02 3.23952913e-01 -7.85541594e-01 4.99629289e-01 5.09607971e-01 -1.50431907e+00 5.57099283e-02 -1.26843309e+00 -3.23031992e-01 1.38929501e-01 -1.73620462e-01 6.70847774e-01 8.99368227e-01 -1.10213602e+00 7.51842260e-01 -1.14729738e+00 -3.44255179e-01 6.45590365e-01 1.63401868e-02 7.73955733e-02 -4.97431159e-02 -1.11047113e+00 9.64305699e-01 -7.86747560e-02 -2.04293057e-01 -1.13599229e+00 -4.96808499e-01 -9.27434266e-01 3.02226931e-01 2.05510646e-01 -9.01021123e-01 1.45955145e+00 -4.26497400e-01 -1.62934148e+00 3.03697586e-01 -3.77612054e-01 -5.03207386e-01 6.36899114e-01 -3.48005325e-01 -1.82074949e-01 2.97043920e-01 -2.22834140e-01 3.16611022e-01 5.93838930e-01 -1.23377669e+00 -3.38872194e-01 1.56341922e-02 -6.64405525e-02 2.50660539e-01 1.07979096e-01 -1.39512077e-01 -4.66780096e-01 -5.30611694e-01 6.29470795e-02 -7.00271189e-01 1.14176944e-01 5.71667671e-01 -3.80622029e-01 8.13267604e-02 5.42994320e-01 -5.65807581e-01 1.54000056e+00 -2.31835866e+00 -4.33060080e-02 2.44292408e-01 3.35175484e-01 -6.04722649e-02 2.50325680e-01 7.87927091e-01 -2.54121888e-02 3.67507607e-01 -4.94450808e-01 -6.59093857e-01 2.14648187e-01 2.72753417e-01 -1.58671573e-01 8.09368908e-01 -1.79626808e-01 5.20294964e-01 -9.43427861e-01 -5.32993734e-01 2.68956244e-01 8.90534401e-01 -7.30600297e-01 1.58797756e-01 1.62149370e-01 4.67636064e-03 -3.39864314e-01 5.95496774e-01 7.71628499e-01 -4.53075409e-01 1.72950760e-01 -7.30059221e-02 -6.05611540e-02 4.10952687e-01 -1.05584025e+00 1.60792303e+00 -3.56985599e-01 1.25637448e+00 3.28512371e-01 -8.06693077e-01 3.62810522e-01 4.40063477e-01 4.20774460e-01 -3.92471820e-01 1.28461093e-01 2.08253101e-01 -9.88766104e-02 -5.53674221e-01 5.60658753e-01 -3.24429363e-01 2.58842528e-01 7.45035112e-01 1.60624951e-01 -1.10286765e-01 -8.46720040e-02 5.95322192e-01 1.10899568e+00 1.45741880e-01 5.95351100e-01 -2.15992928e-01 -2.22016081e-01 -2.76415408e-01 3.30089957e-01 1.15725994e+00 -4.33676988e-01 6.92852676e-01 5.80859482e-01 6.36701286e-02 -1.31348825e+00 -1.37111163e+00 -7.69577563e-01 3.11868340e-01 1.57218456e-01 -7.96745837e-01 -5.36852837e-01 -1.26437336e-01 -1.16000555e-01 1.09189773e+00 -1.97902679e-01 3.20272408e-02 1.63104728e-01 -1.00213420e+00 6.35346472e-01 2.19881330e-02 5.01071990e-01 -3.01802367e-01 -6.14207864e-01 -1.69551581e-01 -4.82025981e-01 -1.00181389e+00 9.45180357e-02 4.05330919e-02 -6.47266448e-01 -7.54842222e-01 -7.49281466e-01 -1.81851372e-01 2.67665237e-01 2.26939186e-01 9.53513265e-01 -1.12406582e-01 -2.05467030e-01 7.36188054e-01 -2.14481339e-01 -3.79179716e-01 -5.59183419e-01 -5.10025203e-01 1.21716052e-01 -4.00083840e-01 3.99488747e-01 -7.39534438e-01 -5.66950858e-01 -1.05884127e-01 -8.02138686e-01 -6.69305325e-02 4.55005646e-01 7.08389938e-01 4.07319903e-01 2.96900332e-01 3.94285172e-02 -5.99687755e-01 7.53471911e-01 -9.01432872e-01 -6.21208251e-01 1.53354228e-01 -8.12532902e-01 1.91012979e-01 -7.66770691e-02 2.23060191e-01 -1.09065604e+00 -4.28200841e-01 -4.81435239e-01 -4.07120258e-01 2.53237367e-01 5.55225909e-01 4.90002101e-03 8.18631873e-02 4.59551662e-01 8.91319290e-02 4.09753919e-02 -5.01341403e-01 3.54579240e-01 1.01875627e+00 2.11387679e-01 -5.29452801e-01 4.50640351e-01 4.90128398e-01 9.54201370e-02 -1.14670110e+00 -5.32934487e-01 1.31505700e-02 -1.93888023e-01 -1.18565306e-01 6.86536312e-01 -9.70371068e-01 -6.89254165e-01 4.62915301e-01 -1.27183712e+00 -2.52747804e-01 -3.61634493e-01 8.28705370e-01 -1.01393056e+00 5.49049020e-01 -8.06308210e-01 -1.12626183e+00 1.36881500e-01 -1.39238989e+00 9.34238493e-01 2.12169308e-02 -1.08187720e-01 -1.12099123e+00 1.00135922e-01 1.71652898e-01 3.75465214e-01 -2.07241103e-02 9.88520026e-01 -2.33129963e-01 -9.02630866e-01 -2.37318445e-02 -2.37056687e-01 3.64509672e-01 -1.11075044e-01 9.78792384e-02 -1.05632472e+00 -3.99377525e-01 3.51922780e-01 -1.35327563e-01 6.28696442e-01 7.58299768e-01 1.14199901e+00 -2.76034683e-01 -2.23415837e-01 9.49522555e-01 1.80754292e+00 1.32583991e-01 7.45290995e-01 -3.02637834e-02 -2.05275957e-02 2.25630209e-01 1.43618077e-01 1.03282499e+00 2.24023774e-01 4.18042839e-01 4.45183992e-01 4.34592783e-01 -2.07213268e-01 -3.33189785e-01 4.10045207e-01 1.13611591e+00 1.83583163e-02 -8.43215585e-01 -4.83836442e-01 2.99865693e-01 -1.55411887e+00 -9.96483207e-01 -2.66171638e-02 2.27178526e+00 7.01607585e-01 -4.56749322e-03 -4.34264123e-01 -3.68540734e-02 6.00837171e-01 3.87422949e-01 -4.22534764e-01 -2.97898740e-01 -1.50360420e-01 -2.53960080e-02 6.86904728e-01 8.58360946e-01 -7.66462207e-01 4.34703052e-01 8.33849335e+00 8.75912070e-01 -5.75042307e-01 3.27707410e-01 4.40086871e-01 -2.48539135e-01 -4.86150116e-01 1.50146976e-01 -5.85784316e-01 4.50298905e-01 1.05422544e+00 -3.43084514e-01 5.74236274e-01 5.46825051e-01 3.05005252e-01 -4.27676857e-01 -1.13627410e+00 1.16830373e+00 4.96027730e-02 -1.22322261e+00 -1.28366858e-01 3.84264469e-01 3.76041293e-01 3.32576603e-01 1.16186626e-01 -9.01435465e-02 5.22527397e-01 -1.02498889e+00 7.21180797e-01 5.43279231e-01 9.33555365e-01 -5.11303127e-01 5.21976531e-01 2.99228817e-01 -5.65229833e-01 8.37153196e-02 -5.14816523e-01 -2.06082553e-01 7.18169510e-01 7.19981074e-01 -5.16019106e-01 4.27220762e-01 6.45989060e-01 5.21221101e-01 5.85448258e-02 1.08836687e+00 -2.79746085e-01 6.59442723e-01 -6.49777412e-01 -3.21894847e-02 -2.20500212e-02 -1.46950170e-01 9.29826200e-01 1.23328853e+00 6.67783499e-01 3.32930148e-01 -6.00599051e-01 1.10956812e+00 1.60963550e-01 -4.19591010e-01 -6.35969102e-01 -1.76493391e-01 6.73293412e-01 6.33595169e-01 -2.28692070e-01 -2.47755423e-02 -6.95999146e-01 9.08375740e-01 -1.31156415e-01 6.96295857e-01 -7.72349656e-01 -4.22765344e-01 9.89341319e-01 -1.05946794e-01 3.79678249e-01 -4.42243606e-01 -2.57215530e-01 -1.17566752e+00 -1.92534462e-01 -7.14364350e-01 -5.21986745e-03 -1.26481056e+00 -1.18418610e+00 1.03495039e-01 4.82736170e-01 -1.04567587e+00 -2.74984360e-01 -6.44711614e-01 -1.73882991e-01 8.05991590e-01 -1.34271932e+00 -1.55026436e-01 1.28000543e-01 3.96029532e-01 2.92041928e-01 -7.33515844e-02 8.48696113e-01 4.84967381e-01 -5.22654176e-01 5.34934580e-01 7.27202773e-01 -2.90942863e-02 2.65857100e-01 -9.11560059e-01 3.60935122e-01 8.47122014e-01 1.42442390e-01 7.54553080e-01 1.21388853e+00 -5.09128988e-01 -1.40816414e+00 -4.55124676e-01 7.88707495e-01 -3.49287957e-01 6.22432351e-01 -2.25784793e-01 -6.06790841e-01 9.34513509e-01 4.74122286e-01 -3.23332816e-01 9.49144065e-01 -1.30757466e-01 -3.12276417e-03 4.33865279e-01 -1.29925251e+00 6.16844773e-01 9.92188692e-01 -7.52173960e-01 -5.08310437e-01 5.05651951e-01 4.60357487e-01 -4.08195019e-01 -8.22091699e-01 1.51122361e-01 8.18333685e-01 -1.31719363e+00 9.20225859e-01 -2.29126781e-01 5.71156919e-01 9.39219370e-02 -6.78779662e-01 -1.10476089e+00 -2.09057793e-01 -1.00449646e+00 -3.08064967e-01 8.12202692e-01 1.05895169e-01 -3.35169196e-01 6.07133806e-01 6.64983511e-01 -1.38851842e-02 -7.06128001e-01 -1.16323614e+00 -6.72377586e-01 8.89737159e-02 -1.01395535e+00 2.16952130e-01 5.95868707e-01 2.51907974e-01 9.95496362e-02 -7.79777884e-01 2.51392633e-01 1.14735413e+00 -5.15876830e-01 3.69066030e-01 -8.12837958e-01 -6.05157375e-01 -1.64698541e-01 -4.00057137e-01 -1.78654408e+00 -3.38051230e-01 -8.07124555e-01 5.07949814e-02 -1.42584884e+00 3.81525934e-01 -2.12554380e-01 -3.86331938e-02 -1.94149420e-01 3.17854941e-01 -7.56950378e-02 2.60907710e-01 2.26490706e-01 -4.29251224e-01 7.54225135e-01 9.45988119e-01 1.36793479e-01 2.18366370e-01 1.09446101e-01 -6.01794541e-01 8.07164192e-01 6.04403615e-01 -7.11236238e-01 -6.66970372e-01 -7.35841870e-01 6.59636676e-01 4.36912626e-01 4.29467261e-01 -5.86604536e-01 3.32576841e-01 1.85433075e-01 1.69841424e-01 -6.46224380e-01 6.77743375e-01 -7.11463511e-01 2.56519198e-01 2.21660167e-01 -3.13779414e-01 -1.40773162e-01 -7.42088482e-02 8.34878981e-01 -3.00939716e-02 -7.16661930e-01 6.42755866e-01 -5.41928038e-02 -5.15177250e-01 1.99644044e-01 -7.79938757e-01 2.93828007e-02 6.82314754e-01 7.00387880e-02 -2.36944661e-01 -9.46811616e-01 -7.11006284e-01 1.63525552e-01 3.57550651e-01 -2.17845310e-02 7.91888118e-01 -1.20484972e+00 -6.00444138e-01 3.21421206e-01 -2.70240605e-01 -2.36599654e-01 2.22360656e-01 6.28854811e-01 -6.21327281e-01 3.72171313e-01 2.85990536e-01 -6.28786445e-01 -1.04971433e+00 3.62644464e-01 3.92019540e-01 9.59559008e-02 -5.05239844e-01 9.16515887e-01 1.54381767e-01 1.42772436e-01 5.34095407e-01 -1.00288235e-01 3.57061297e-01 -3.26297909e-01 5.16593754e-01 5.29150009e-01 -3.49432409e-01 -4.80593950e-01 -2.87626684e-01 2.63037711e-01 2.75629044e-01 -5.77478588e-01 1.06576896e+00 -8.70037496e-01 1.44985810e-01 5.44889390e-01 1.41282260e+00 5.83748184e-02 -1.52473688e+00 -2.82838047e-01 -4.77934361e-01 -5.64993918e-01 3.04506928e-01 -5.69782436e-01 -8.72505367e-01 1.14827764e+00 6.65899992e-01 3.57398570e-01 8.51788163e-01 2.15411442e-03 7.26401091e-01 3.34045112e-01 4.18423921e-01 -9.16341126e-01 -3.62633079e-01 3.58943343e-01 6.38654828e-01 -1.02573884e+00 5.13329089e-01 -2.56934404e-01 -4.67494279e-01 9.61610556e-01 -3.61055970e-01 1.96547821e-01 1.36635959e+00 4.95880455e-01 -1.97142959e-01 -2.04717934e-01 -8.39622855e-01 8.31384584e-02 -2.50451207e-01 7.38187850e-01 4.40223426e-01 -1.59677207e-01 -3.36004883e-01 4.88894731e-01 -3.37074786e-01 1.36627197e-01 8.66077363e-01 9.40479100e-01 -3.41107398e-01 -1.04555821e+00 -2.21921161e-01 3.25560689e-01 -7.83104479e-01 -4.04335350e-01 1.48113355e-01 7.78846562e-01 -1.45705953e-01 1.13118815e+00 1.57410294e-01 -2.25346103e-01 -2.10841775e-01 1.23702303e-01 5.44351935e-01 -2.78707534e-01 3.93734753e-01 1.85889706e-01 7.60168731e-02 -5.03904104e-01 -5.97153366e-01 -8.13598037e-01 -6.87794268e-01 -8.57223272e-01 -4.37393516e-01 1.04039386e-01 9.08773959e-01 1.04294109e+00 3.03069651e-01 4.65201229e-01 4.08236265e-01 -4.82298166e-01 -7.75766492e-01 -7.26810634e-01 -9.08602417e-01 -1.54530823e-01 6.46683097e-01 -5.40112078e-01 -8.47597480e-01 -6.91549480e-02]
[7.211519241333008, 3.9171230792999268]
6c3afd81-54d6-4d2b-94ee-4ffbf23d9aa0
roto-translation-equivariant-convolutional
2002.08725
null
https://arxiv.org/abs/2002.08725v1
https://arxiv.org/pdf/2002.08725v1.pdf
Roto-Translation Equivariant Convolutional Networks: Application to Histopathology Image Analysis
Rotation-invariance is a desired property of machine-learning models for medical image analysis and in particular for computational pathology applications. We propose a framework to encode the geometric structure of the special Euclidean motion group SE(2) in convolutional networks to yield translation and rotation equivariance via the introduction of SE(2)-group convolution layers. This structure enables models to learn feature representations with a discretized orientation dimension that guarantees that their outputs are invariant under a discrete set of rotations. Conventional approaches for rotation invariance rely mostly on data augmentation, but this does not guarantee the robustness of the output when the input is rotated. At that, trained conventional CNNs may require test-time rotation augmentation to reach their full capability. This study is focused on histopathology image analysis applications for which it is desirable that the arbitrary global orientation information of the imaged tissues is not captured by the machine learning models. The proposed framework is evaluated on three different histopathology image analysis tasks (mitosis detection, nuclei segmentation and tumor classification). We present a comparative analysis for each problem and show that consistent increase of performances can be achieved when using the proposed framework.
['Erik J. Bekkers', 'Maxime W. Lafarge', 'Mitko Veta', 'Josien P. W. Pluim', 'Remco Duits']
2020-02-20
null
null
null
null
['colorectal-gland-segmentation', 'breast-tumour-classification', 'multi-tissue-nucleus-segmentation', 'mitosis-detection']
['medical', 'medical', 'medical', 'medical']
[ 5.49839795e-01 2.92515427e-01 -2.76434928e-01 -2.97438949e-01 -2.54828066e-01 -4.63384449e-01 6.70080602e-01 1.09161168e-01 -6.87099695e-01 5.51763296e-01 -1.18167199e-01 -4.29039508e-01 -1.54766276e-01 -5.64651549e-01 -6.21888041e-01 -1.15145445e+00 -7.43352771e-02 1.64967120e-01 -2.94724270e-03 -2.81922907e-01 1.45955995e-01 1.24998498e+00 -1.08535898e+00 1.66902110e-01 4.47632253e-01 6.21795058e-01 -1.15438430e-02 9.77110326e-01 2.66508937e-01 4.58249122e-01 -1.87403604e-01 1.64582223e-01 2.99305111e-01 -2.13934571e-01 -1.06375253e+00 2.96695530e-01 1.45543262e-01 -7.33593106e-02 -3.09269309e-01 1.02595520e+00 2.78828949e-01 1.47000570e-02 8.79123926e-01 -8.42081785e-01 -3.80143106e-01 3.64354759e-01 -3.81959647e-01 4.79041398e-01 -1.57534301e-01 -6.86383322e-02 7.28412092e-01 -4.80459809e-01 9.94835079e-01 7.90835619e-01 5.52042902e-01 4.79850113e-01 -1.21394110e+00 -7.26402849e-02 -1.11099124e-01 9.43261851e-03 -1.34207237e+00 -3.36601704e-01 8.08001459e-01 -3.75412524e-01 8.06465030e-01 5.35308480e-01 3.54863465e-01 7.35874355e-01 7.47938275e-01 2.52795666e-01 9.85770822e-01 -3.13568830e-01 8.14075246e-02 1.10875443e-03 2.06037238e-01 8.51358891e-01 3.72511387e-01 -3.76267433e-01 4.04627388e-03 1.66769683e-01 1.11027956e+00 1.86287075e-01 -4.01104271e-01 -4.65566874e-01 -1.56351244e+00 6.21883094e-01 8.42654288e-01 6.59601569e-01 -2.04319417e-01 7.09747970e-02 4.94184613e-01 4.24724966e-01 3.07952613e-01 3.69617343e-01 -3.06196809e-01 4.15468991e-01 -7.86010861e-01 -7.04915375e-02 2.89068997e-01 7.76091695e-01 4.04203981e-01 1.78979397e-01 -1.40643939e-01 3.18152070e-01 1.50832161e-01 3.49075906e-02 8.81184399e-01 -4.47830260e-01 -1.45891942e-02 7.20433176e-01 -1.96127996e-01 -1.00401187e+00 -1.03828418e+00 -1.03052402e+00 -1.43990791e+00 1.16961643e-01 5.30856133e-01 2.03883201e-01 -1.12640274e+00 1.58518732e+00 4.07804728e-01 -1.04280468e-02 2.59195209e-01 7.32718825e-01 7.70290911e-01 1.27221480e-01 -1.56357899e-01 -2.68898457e-01 1.44418550e+00 -6.19415700e-01 -7.33860731e-01 1.82131127e-01 1.14211810e+00 -7.83301294e-01 7.31393576e-01 2.22615674e-01 -7.27756858e-01 -4.97737586e-01 -1.28986251e+00 -1.69900939e-01 -4.75679010e-01 3.95920068e-01 6.66940808e-01 6.30274475e-01 -1.13645911e+00 5.90578318e-01 -1.24648201e+00 -5.64684153e-01 3.49299759e-01 8.15803587e-01 -8.75477493e-01 -1.31545141e-02 -7.28292406e-01 7.34512448e-01 3.80203784e-01 5.30981660e-01 -5.64675689e-01 -3.99988085e-01 -8.55195582e-01 -1.13375328e-01 -3.82752568e-01 -6.05870545e-01 9.03809488e-01 -8.95033777e-01 -1.30548155e+00 9.71379876e-01 -3.23244110e-02 -4.33975965e-01 7.43803859e-01 3.43822658e-01 -1.39709890e-01 3.50311816e-01 -2.32896328e-01 6.48968458e-01 7.87314951e-01 -7.56968021e-01 -2.40275502e-01 -4.19598162e-01 1.25038167e-02 2.25973397e-01 -4.72346038e-01 -2.52782285e-01 -1.17097229e-01 -7.60819376e-01 6.99697375e-01 -1.20205474e+00 -4.00631070e-01 -3.76455188e-02 -4.58497494e-01 1.35928378e-01 9.44937110e-01 -5.59549987e-01 7.00973988e-01 -2.06123352e+00 2.03604236e-01 3.22200269e-01 6.01327829e-02 2.79042516e-02 -2.95926426e-02 1.23859398e-01 -5.84363043e-01 1.62885308e-01 3.47383171e-02 -4.95518409e-02 -5.34591973e-01 1.97515279e-01 9.03155804e-02 1.29829741e+00 4.69822496e-01 8.84880841e-01 -4.94882703e-01 -4.56592143e-01 2.82116205e-01 6.68211937e-01 -6.31381571e-01 -1.05247259e-01 1.75243080e-01 9.26440895e-01 -4.85954106e-01 5.21652341e-01 6.83789432e-01 -3.27344060e-01 2.67722398e-01 -5.23377597e-01 -5.41633694e-03 -1.47625774e-01 -9.77136612e-01 1.44871867e+00 -2.70420849e-01 5.28568208e-01 -8.56745094e-02 -1.37884581e+00 7.80908346e-01 5.45191765e-01 8.18527341e-01 -2.80720204e-01 3.97269607e-01 1.17888704e-01 4.70431834e-01 -4.04616058e-01 2.27309823e-01 6.18284612e-05 8.90449360e-02 1.98852912e-01 1.86524123e-01 1.90319791e-01 5.28892390e-02 -2.62156725e-01 1.06158555e+00 -1.00290135e-01 5.60336292e-01 -6.45891249e-01 9.08873498e-01 -2.52809692e-02 3.83909971e-01 3.94944847e-01 -7.95359761e-02 6.42120063e-01 5.23432255e-01 -8.38449359e-01 -1.30445015e+00 -6.64547503e-01 -5.32420933e-01 6.69316471e-01 -2.26728380e-01 3.28589618e-01 -7.88031101e-01 -6.64393961e-01 -4.53291684e-01 -1.29699707e-02 -1.04278481e+00 -2.08734259e-01 -8.71303976e-01 -1.15824103e+00 4.64804590e-01 4.82235283e-01 4.17790890e-01 -8.65482986e-01 -7.25739777e-01 1.05672337e-01 1.55678853e-01 -1.26085138e+00 -2.20437050e-01 5.48088729e-01 -1.31427109e+00 -1.11897492e+00 -7.73667693e-01 -1.16026688e+00 1.40832055e+00 2.58499414e-01 3.71357590e-01 1.66599005e-01 -3.99378002e-01 9.83739570e-02 -4.29797322e-01 -1.33658871e-01 -4.23824877e-01 5.13701141e-01 -7.81176658e-03 1.78022191e-01 8.21448416e-02 -5.42924523e-01 -6.45700097e-01 3.69073510e-01 -1.30452132e+00 1.02503747e-01 7.69697905e-01 1.00717258e+00 6.45448983e-01 9.11992639e-02 5.59143007e-01 -8.50880027e-01 2.19538942e-01 -1.27922967e-01 -2.43457407e-01 7.33816177e-02 -1.46489143e-01 2.39449337e-01 7.54315972e-01 -4.25767630e-01 -5.74977160e-01 4.24602002e-01 -1.78590029e-01 -2.29749128e-01 -2.96376616e-01 5.76393664e-01 2.78872028e-02 -5.45186341e-01 7.75933802e-01 3.67855400e-01 2.35350534e-01 -1.19414568e-01 1.30755097e-01 3.82365286e-01 6.22203052e-01 -9.48083028e-02 8.26331735e-01 9.47721720e-01 6.53318822e-01 -1.06862402e+00 -3.62276345e-01 -5.53941846e-01 -1.02690244e+00 -1.25785053e-01 9.70382571e-01 -6.33397460e-01 -5.48051178e-01 5.66603363e-01 -1.02057981e+00 1.37370586e-01 -6.74088076e-02 6.58437014e-01 -7.05719650e-01 3.34585100e-01 -6.35015726e-01 -3.37998539e-01 -4.72251803e-01 -1.35609818e+00 7.99741626e-01 1.80126578e-01 -6.94517791e-02 -1.11872733e+00 -1.54881924e-01 5.94702847e-02 4.11073297e-01 5.59827328e-01 1.18036604e+00 -9.52704549e-01 -4.30075943e-01 -6.05264425e-01 2.06708219e-02 2.18484223e-01 3.92855614e-01 -8.27807561e-03 -8.81477773e-01 -6.94483638e-01 3.18906493e-02 -6.26830012e-02 8.06184649e-01 5.04878342e-01 1.19734967e+00 -2.80951828e-01 -3.50647032e-01 8.85176659e-01 1.35024953e+00 5.61713018e-02 5.89775622e-01 4.19683576e-01 6.08248472e-01 4.43543643e-01 3.81816506e-01 1.07428886e-01 -9.14743394e-02 4.86396879e-01 6.74917698e-01 -5.99381506e-01 -4.24140282e-02 3.25557828e-01 -1.61526981e-03 8.47031653e-01 -4.26886737e-01 -1.20588519e-01 -8.85288537e-01 5.38679123e-01 -1.64615989e+00 -6.44547820e-01 -1.77199692e-01 2.14776087e+00 5.25976896e-01 1.69345766e-01 -2.68998146e-01 5.58934689e-01 7.01656759e-01 -4.60693613e-03 -2.87694573e-01 -4.72455442e-01 -1.64531469e-02 3.24132562e-01 6.83605969e-01 3.33969414e-01 -1.34963918e+00 6.37317181e-01 6.23264980e+00 5.71723580e-01 -1.60887754e+00 -1.28411248e-01 7.95557797e-01 4.23911870e-01 9.46959928e-02 -3.95989388e-01 -4.81030494e-01 -2.29524434e-01 5.58314323e-01 2.12915719e-01 -1.50176972e-01 6.28351450e-01 3.63763779e-01 2.46518493e-01 -1.18099773e+00 6.27807319e-01 -1.22579448e-01 -1.35396016e+00 1.51313663e-01 2.71584570e-01 7.70018041e-01 -2.65347183e-01 4.20060635e-01 -1.36764660e-01 -2.95989811e-01 -1.26338482e+00 9.24856067e-02 5.48866212e-01 7.99211442e-01 -8.92255187e-01 1.12061894e+00 2.02277631e-01 -1.00779760e+00 2.36046519e-02 -4.31060940e-01 2.03416660e-01 -2.61746645e-01 2.35379815e-01 -1.49869478e+00 8.79642725e-01 2.92788655e-01 5.99619567e-01 -6.02195323e-01 9.11514282e-01 2.38419861e-01 3.97464812e-01 -1.67506605e-01 1.37401894e-01 3.41450632e-01 -6.93598762e-02 4.94713962e-01 1.28471255e+00 4.22139585e-01 -2.36235917e-01 -7.43661150e-02 3.18626910e-01 -3.49903293e-02 3.38796973e-01 -7.08883941e-01 -1.19748376e-02 -1.17087550e-01 1.57971215e+00 -1.39447320e+00 2.27932017e-02 -1.90824017e-01 7.87544847e-01 1.40114382e-01 2.45400056e-01 -6.20764077e-01 -3.19218874e-01 5.03285408e-01 2.81563461e-01 2.94371605e-01 -4.08480078e-01 -1.89336374e-01 -9.46539283e-01 -2.39783928e-01 -7.79705048e-01 3.07545215e-01 -3.61684829e-01 -7.53850579e-01 7.08074749e-01 -2.39136249e-01 -1.38556933e+00 -3.25178057e-01 -9.57685411e-01 -6.12304688e-01 5.54767370e-01 -1.48444152e+00 -1.59521091e+00 -2.90003628e-01 6.85348928e-01 2.99772799e-01 -1.65612459e-01 9.61185873e-01 1.37446538e-01 -4.59725767e-01 7.51061618e-01 1.39928401e-01 3.34062308e-01 5.94209254e-01 -1.18832469e+00 -7.53867626e-02 8.98998797e-01 -3.24857421e-02 7.93404043e-01 8.85298789e-01 -3.65403086e-01 -1.29653621e+00 -1.31953073e+00 6.91592216e-01 -9.39197019e-02 4.20292079e-01 -1.61248073e-01 -7.25218177e-01 8.39008033e-01 5.80654256e-02 5.96702993e-01 7.16723680e-01 -3.61828744e-01 -7.36224875e-02 -5.20538315e-02 -1.28337073e+00 6.03482783e-01 6.50195539e-01 -2.85055518e-01 -1.92592263e-01 4.44056273e-01 5.32448530e-01 -4.31801826e-01 -1.01552260e+00 6.46915853e-01 4.66106474e-01 -5.42594194e-01 8.69782925e-01 -1.00907302e+00 2.97655314e-01 -4.91200268e-01 -7.70832449e-02 -1.04470515e+00 -4.08560514e-01 -2.46575639e-01 3.05835366e-01 6.40848696e-01 2.89898098e-01 -5.18541873e-01 8.34254086e-01 -5.93156256e-02 -2.61548668e-01 -7.56226838e-01 -1.12342870e+00 -4.76503730e-01 2.87044376e-01 5.90981953e-02 1.88938975e-01 1.14890277e+00 -1.02676906e-01 -1.57560065e-01 -4.93814871e-02 4.23518807e-01 2.91092455e-01 -1.97125778e-01 5.85665703e-01 -1.00762022e+00 -1.46630621e-02 -2.41195306e-01 -1.21087670e+00 -5.63718379e-01 5.41877560e-02 -1.18541551e+00 -2.98686683e-01 -1.17845690e+00 9.30478796e-02 -1.60510585e-01 -5.69687903e-01 4.80097920e-01 2.97397971e-01 5.86999476e-01 -2.42712498e-01 1.95068046e-01 -2.13330865e-01 1.55231580e-01 1.54446077e+00 -6.45225346e-02 -1.41337976e-01 2.74530053e-01 -5.22605836e-01 6.97872698e-01 1.02124178e+00 -3.03090692e-01 -4.05928612e-01 -1.31324917e-01 1.12923488e-01 -1.10963777e-01 4.37030375e-01 -1.17868221e+00 2.84162611e-01 -6.68785945e-02 6.98996425e-01 -3.67862165e-01 5.66171035e-02 -8.15070868e-01 2.17689618e-01 8.66368592e-01 -3.88517439e-01 1.83213592e-01 4.43002507e-02 3.44873518e-01 -2.04361320e-01 -1.76486924e-01 1.12945771e+00 -8.34660139e-03 -1.57660469e-01 3.04672748e-01 -5.50606668e-01 -6.49815977e-01 1.04791689e+00 -4.06530350e-01 -1.72142506e-01 1.14905415e-02 -1.00845301e+00 -4.33108270e-01 4.62660044e-01 1.56314760e-01 4.74322140e-01 -1.30244529e+00 -6.31590962e-01 2.88679004e-01 1.16775818e-01 2.14971572e-01 3.52331519e-01 1.20075834e+00 -1.08372343e+00 6.67099535e-01 -6.95480049e-01 -7.65566826e-01 -1.32038939e+00 5.55060446e-01 6.30844176e-01 -7.09272981e-01 -5.51881194e-01 4.52650756e-01 1.88317105e-01 -4.72394168e-01 -1.99522287e-01 -7.01620340e-01 -6.66045308e-01 -2.03747049e-01 2.39336774e-01 -6.93619996e-02 5.25312841e-01 -7.60650218e-01 -3.01607341e-01 4.64944035e-01 -3.27934444e-01 2.36346051e-01 1.47765040e+00 1.05987787e-01 -1.18285879e-01 3.87133919e-02 1.31032467e+00 -2.49364942e-01 -8.66189301e-01 -7.75210485e-02 -8.84282589e-02 3.49933729e-02 1.44053876e-01 -2.19924137e-01 -1.18730521e+00 9.00577843e-01 7.66290009e-01 1.07850969e-01 1.10196698e+00 -2.82301307e-01 8.94449651e-02 6.22410655e-01 3.01640667e-02 -7.60741353e-01 3.65646444e-02 3.51324916e-01 8.71679783e-01 -9.26102042e-01 -5.56855313e-02 -4.22808975e-01 -2.48588011e-01 1.53609991e+00 3.87270689e-01 -4.68112230e-01 6.48956954e-01 1.20595358e-01 1.41203462e-03 -8.07860494e-02 -4.32364881e-01 -2.52170283e-02 4.44348693e-01 5.67026913e-01 7.06828237e-01 -1.34789407e-01 -5.10079324e-01 2.69887835e-01 -2.21965432e-01 -2.28278354e-01 7.41759181e-01 1.02105963e+00 -3.75033051e-01 -9.63153839e-01 -3.38638008e-01 2.82652229e-01 -8.60421658e-01 3.16663176e-01 -1.20725505e-01 1.13872302e+00 -1.02818171e-02 3.68922532e-01 1.44804910e-01 -1.39419824e-01 -4.68240725e-03 -4.72659245e-02 5.33576846e-01 -5.00860572e-01 -3.08658808e-01 2.02438086e-01 -2.74985135e-01 -2.18363672e-01 -6.20478630e-01 -4.79603440e-01 -1.29819012e+00 1.67589821e-02 -1.44560501e-01 -1.46586433e-01 7.50357985e-01 8.68135333e-01 -4.41247746e-02 9.08343256e-01 7.36174285e-01 -6.75408900e-01 -5.57480216e-01 -9.55741048e-01 -6.47688270e-01 2.59536952e-01 6.37463570e-01 -5.06082654e-01 -1.63399458e-01 2.96259880e-01]
[15.073692321777344, -2.874262809753418]
f80c7ef9-0930-4a4d-9bc0-4617be34850e
object-level-depth-reconstruction-for
2204.01586
null
https://arxiv.org/abs/2204.01586v2
https://arxiv.org/pdf/2204.01586v2.pdf
Object Level Depth Reconstruction for Category Level 6D Object Pose Estimation From Monocular RGB Image
Recently, RGBD-based category-level 6D object pose estimation has achieved promising improvement in performance, however, the requirement of depth information prohibits broader applications. In order to relieve this problem, this paper proposes a novel approach named Object Level Depth reconstruction Network (OLD-Net) taking only RGB images as input for category-level 6D object pose estimation. We propose to directly predict object-level depth from a monocular RGB image by deforming the category-level shape prior into object-level depth and the canonical NOCS representation. Two novel modules named Normalized Global Position Hints (NGPH) and Shape-aware Decoupled Depth Reconstruction (SDDR) module are introduced to learn high fidelity object-level depth and delicate shape representations. At last, the 6D object pose is solved by aligning the predicted canonical representation with the back-projected object-level depth. Extensive experiments on the challenging CAMERA25 and REAL275 datasets indicate that our model, though simple, achieves state-of-the-art performance.
['Jun He', 'Hongyan Liu', 'Kejian Wu', 'Zhicheng Wang', 'Jian Xu', 'Zhenbo Song', 'Zhaoxin Fan']
2022-04-04
null
null
null
null
['6d-pose-estimation']
['computer-vision']
[ 1.63395405e-01 2.96357483e-01 -2.71958500e-01 -6.70931876e-01 -9.21600342e-01 -3.22896570e-01 4.15882677e-01 -3.56799364e-01 -3.22026938e-01 2.59052962e-01 1.01233259e-01 1.44115835e-01 6.53026924e-02 -7.53808916e-01 -9.19777811e-01 -6.32214844e-01 4.64132071e-01 6.98105276e-01 4.60187763e-01 3.94597590e-01 5.61623394e-01 9.36453581e-01 -1.60119092e+00 2.75478870e-01 5.65682411e-01 1.49082136e+00 4.65940773e-01 5.34290850e-01 -1.63537845e-01 5.45852542e-01 -2.14081600e-01 -1.54778287e-01 7.04104841e-01 -2.24141255e-02 -5.99740446e-01 5.92349887e-01 6.65165365e-01 -9.07135069e-01 -5.38507402e-01 8.32939744e-01 4.58385050e-01 -4.26266417e-02 5.33972144e-01 -1.11924076e+00 -3.14135909e-01 -2.01222509e-01 -7.74503708e-01 -1.72495097e-01 4.05638009e-01 2.27677703e-01 6.45754278e-01 -1.10970294e+00 6.82668149e-01 1.47522581e+00 2.80140817e-01 7.41290271e-01 -1.00338626e+00 -4.71983284e-01 4.50871944e-01 9.56922676e-03 -1.16283214e+00 -4.93217371e-02 1.17350233e+00 -3.31442356e-01 1.13731027e+00 -1.04553901e-01 7.68395305e-01 9.34979320e-01 -2.84817088e-02 9.66573954e-01 1.11788380e+00 -1.39588326e-01 2.51272529e-01 -1.04095131e-01 -1.61924660e-01 7.40927637e-01 1.80206701e-01 1.14615589e-01 -7.71898746e-01 1.02733932e-01 1.17957115e+00 2.27934062e-01 -2.22202063e-01 -9.76377666e-01 -1.25183713e+00 4.52804625e-01 8.06573868e-01 -1.82900637e-01 -3.06400388e-01 4.41444933e-01 -1.27992332e-01 -3.21819842e-01 5.42951167e-01 7.74202943e-02 -7.27754891e-01 -2.46372335e-02 -5.14541030e-01 3.73501509e-01 4.07141298e-01 1.15662932e+00 9.34764504e-01 -1.48125842e-01 1.06274765e-02 5.61754644e-01 7.60309756e-01 5.97300529e-01 1.94196060e-01 -1.36782503e+00 6.01413906e-01 1.05501521e+00 2.18079820e-01 -8.02145183e-01 -4.08058852e-01 -3.99876148e-01 -5.87899446e-01 3.22244704e-01 3.69584262e-01 4.39643830e-01 -1.13314486e+00 1.28105342e+00 7.78141141e-01 1.38193354e-01 -8.37394893e-02 1.37367284e+00 9.86231446e-01 6.94909573e-01 -4.16344017e-01 -2.38045901e-02 1.12172043e+00 -9.38501835e-01 -2.85338998e-01 -4.95736688e-01 2.43862212e-01 -5.14042079e-01 8.36542606e-01 4.96890724e-01 -1.34984243e+00 -4.95070249e-01 -1.01579785e+00 -5.52565515e-01 -1.15906529e-01 1.31650001e-01 7.71059036e-01 3.17303956e-01 -8.79456997e-01 3.25311095e-01 -1.03196824e+00 -9.01791751e-02 5.02848327e-01 5.41667640e-01 -4.93299246e-01 -4.03973699e-01 -4.10753876e-01 7.27249861e-01 4.71296757e-01 1.72902480e-01 -1.11700809e+00 -8.36825192e-01 -8.73353243e-01 -4.22808200e-01 4.77659464e-01 -9.27286804e-01 1.24227142e+00 -4.02208179e-01 -1.76014376e+00 1.17293835e+00 -3.20026092e-02 -3.83442603e-02 4.64075148e-01 -2.36468002e-01 3.45322967e-01 3.32994074e-01 -6.64758608e-02 1.13935053e+00 8.00156295e-01 -1.62524998e+00 -7.17057526e-01 -1.15478694e+00 1.24812536e-01 5.91161311e-01 -4.59274128e-02 -4.16622698e-01 -9.49041486e-01 -3.53168130e-01 1.12001836e+00 -7.22314298e-01 -2.54262894e-01 7.61321306e-01 -1.29943982e-01 -2.27394372e-01 7.70370841e-01 -4.30603504e-01 4.22226161e-01 -1.91205883e+00 5.12318611e-01 -1.11250624e-01 1.67897165e-01 -1.36762068e-01 -1.35265693e-01 -9.43250731e-02 2.51182795e-01 -3.47507894e-01 -2.15651229e-01 -8.50864530e-01 2.42286157e-02 4.25447315e-01 -9.86139849e-02 6.52289987e-01 1.81464761e-01 9.73987103e-01 -8.68535340e-01 -3.06356102e-01 4.69298065e-01 6.45203531e-01 -8.84829998e-01 5.23256242e-01 -5.05051136e-01 5.90590656e-01 -5.49109578e-01 1.11587191e+00 1.16845131e+00 -1.46595299e-01 -1.83601469e-01 -6.00360096e-01 -3.97923775e-02 2.11118802e-01 -1.07351887e+00 2.29207563e+00 -3.58881682e-01 1.50554240e-01 1.35074064e-01 -5.72254717e-01 9.27317321e-01 -1.83162346e-01 6.58600152e-01 -7.26527035e-01 1.22313075e-01 1.79745808e-01 -4.12036300e-01 -2.22215563e-01 4.14876401e-01 -1.29264563e-01 -7.32097104e-02 6.51967376e-02 1.73464000e-01 -7.86293447e-01 -5.72759390e-01 2.11787634e-02 7.31803060e-01 6.90155447e-01 1.42668188e-01 1.54673785e-01 3.78973275e-01 -1.19363748e-01 5.97630501e-01 2.28104830e-01 -1.65658027e-01 1.00971234e+00 3.57839853e-01 -3.41154248e-01 -9.63730872e-01 -1.20920682e+00 -2.00624853e-01 4.48343217e-01 6.70373917e-01 -1.27212375e-01 -5.28475046e-01 -7.19002485e-01 3.43819290e-01 3.34046870e-01 -5.19929707e-01 -1.48399044e-02 -6.83572114e-01 -3.72632116e-01 4.14131721e-03 7.43437290e-01 6.63274586e-01 -7.95481980e-01 -7.41570055e-01 7.49393702e-02 -1.00025907e-01 -1.39667940e+00 -2.26676434e-01 3.51717979e-01 -1.32063580e+00 -8.87926757e-01 -8.11503708e-01 -6.35924578e-01 6.91627860e-01 4.35530305e-01 8.36023092e-01 -3.55201989e-01 -2.62683839e-01 5.00587702e-01 -1.51640207e-01 -8.87035877e-02 2.09320694e-01 -2.01314971e-01 7.29575083e-02 2.62740012e-02 2.26690039e-01 -6.03517592e-01 -1.09298646e+00 4.20840174e-01 -7.31532753e-01 1.42636225e-01 7.64256299e-01 3.96527737e-01 1.03387797e+00 -1.81189686e-01 2.59650126e-02 -1.03104621e-01 -3.64596486e-01 -4.28416282e-02 -9.61166263e-01 -2.02802479e-01 -3.87008071e-01 -4.57744375e-02 5.64305931e-02 -2.25372285e-01 -1.08527863e+00 5.50543308e-01 -2.35461920e-01 -8.49980235e-01 -1.35650769e-01 -3.05815842e-02 -6.34634197e-01 -3.13217759e-01 2.81234920e-01 2.76312828e-01 -1.26825929e-01 -7.26233959e-01 2.63885677e-01 5.00737011e-01 6.18819237e-01 -5.10010540e-01 9.14457858e-01 7.85364926e-01 2.15604872e-01 -4.28269565e-01 -1.27953064e+00 -6.81482136e-01 -9.25667226e-01 -2.30142087e-01 1.07290411e+00 -1.34747052e+00 -1.01960802e+00 6.96940303e-01 -1.41022503e+00 -3.62628907e-01 -1.70359924e-01 5.17820597e-01 -8.71781409e-01 2.17056125e-01 -4.69908565e-01 -7.52445638e-01 -8.62933993e-02 -1.13651025e+00 1.86675119e+00 2.09268685e-02 4.00547534e-01 -5.10294557e-01 -2.21696302e-01 7.99876094e-01 -1.41970441e-01 4.79523897e-01 6.09917581e-01 2.34812990e-01 -1.51673031e+00 -8.55906159e-02 -5.35550892e-01 3.17043841e-01 -4.16730754e-02 -4.65142280e-01 -1.06620157e+00 -1.76166147e-01 3.41168791e-01 -4.08152997e-01 7.15734005e-01 4.00321335e-01 1.51495159e+00 -3.98078822e-02 -1.29028678e-01 1.18070889e+00 1.71053600e+00 -2.87951194e-02 5.75583875e-01 3.03608328e-01 1.21647215e+00 5.20938039e-01 1.05259955e+00 6.32243276e-01 6.68992937e-01 9.42680717e-01 1.24671769e+00 3.12636882e-01 -3.20367396e-01 -5.06772518e-01 2.61030316e-01 6.19520426e-01 -7.58154094e-02 -1.05689675e-01 -8.42617333e-01 2.72579551e-01 -1.58605802e+00 -4.07404840e-01 -1.89813182e-01 1.99131787e+00 5.63999534e-01 3.46416861e-01 1.08662099e-01 2.71863770e-02 1.53229222e-01 1.16931461e-01 -1.00709665e+00 1.55677482e-01 -1.03667796e-01 -1.71561062e-01 6.52997315e-01 5.02822101e-01 -7.86021829e-01 9.96026576e-01 5.08157921e+00 6.34262502e-01 -9.20519948e-01 3.98657843e-02 4.19658035e-01 -2.67724276e-01 -3.43507051e-01 -7.07772896e-02 -1.18078125e+00 1.71307102e-01 2.54873425e-01 3.27982932e-01 2.50089407e-01 1.06568444e+00 -1.28129935e-02 -3.85137171e-01 -1.33372843e+00 1.57915699e+00 2.30171859e-01 -1.04937041e+00 2.14006558e-01 3.89043719e-01 8.19340885e-01 6.59869164e-02 1.13264896e-01 9.76097286e-02 -2.45892227e-01 -7.43719935e-01 1.09491467e+00 5.54001987e-01 8.22034478e-01 -7.69279242e-01 5.02518296e-01 6.69338405e-01 -1.18113005e+00 -2.61642307e-01 -6.10870600e-01 -5.64905517e-02 1.88025728e-01 4.86292779e-01 -6.71953261e-01 5.59575915e-01 8.54221106e-01 9.05831814e-01 -4.58103418e-01 9.32968080e-01 -2.02447072e-01 -1.39495760e-01 -3.19740176e-01 2.91705787e-01 2.59442270e-01 -7.63224624e-03 4.29711252e-01 4.80843931e-01 3.12271982e-01 5.08951843e-01 4.25830670e-02 9.57089245e-01 -8.08296055e-02 -3.84886026e-01 -4.10048097e-01 2.32318327e-01 1.02611527e-01 1.19356930e+00 -8.09008121e-01 -2.59028580e-02 -2.80789793e-01 1.20830023e+00 2.64289677e-01 3.81683968e-02 -7.03945398e-01 2.50423223e-01 8.10516238e-01 3.56762797e-01 3.76871258e-01 -4.54816282e-01 -5.04660130e-01 -1.32366228e+00 2.97117829e-01 -4.76515621e-01 1.79887153e-02 -1.23220575e+00 -1.13352013e+00 2.59844840e-01 1.99410599e-02 -1.40987206e+00 1.20971680e-01 -1.05610001e+00 1.27366960e-01 7.48629689e-01 -1.68659246e+00 -1.17644548e+00 -7.64709473e-01 5.72990894e-01 8.20279360e-01 3.31477195e-01 3.52572948e-01 1.12547286e-01 -7.31606334e-02 3.92040551e-01 -3.42342615e-01 -2.12011427e-01 4.67676789e-01 -1.11190534e+00 2.01978222e-01 3.54622006e-01 -2.03918606e-01 9.65711623e-02 2.51732498e-01 -6.83664203e-01 -2.03154802e+00 -1.24935544e+00 4.61736292e-01 -9.66138601e-01 2.59691656e-01 -7.51456797e-01 -5.72891355e-01 4.76324230e-01 -4.76799667e-01 4.56386745e-01 -1.27810150e-01 -5.26859820e-01 -3.07730556e-01 -3.98215771e-01 -1.35450268e+00 3.55316609e-01 1.62705421e+00 -6.26779139e-01 -4.98452425e-01 3.41585964e-01 1.05582821e+00 -9.44702208e-01 -9.53965187e-01 6.93326950e-01 5.42502165e-01 -1.19424009e+00 1.34119558e+00 4.24298011e-02 5.05152166e-01 -5.23716033e-01 -6.84761167e-01 -7.24712253e-01 -1.60662793e-02 -7.69584402e-02 -5.87673306e-01 8.74476016e-01 -2.11992294e-01 -2.12442651e-01 1.37779474e+00 5.28793693e-01 -4.82574672e-01 -1.10377574e+00 -1.01713216e+00 -6.82145715e-01 -1.54639944e-01 -5.48838139e-01 4.97050703e-01 4.95648056e-01 -5.66443682e-01 2.56449580e-01 -1.27834216e-01 5.03369331e-01 1.04777634e+00 3.39970052e-01 9.73044634e-01 -1.18477499e+00 -3.08597505e-01 -2.34086528e-01 -8.17724347e-01 -1.89529312e+00 9.62708965e-02 -7.94651806e-01 1.54698461e-01 -1.64257729e+00 3.02049458e-01 -2.96397388e-01 3.03456634e-02 2.02994168e-01 2.17940778e-01 5.67307949e-01 1.05887711e-01 2.46032551e-02 -6.84565187e-01 9.49653149e-01 1.77296853e+00 -1.14662936e-02 -1.26902416e-01 1.82066653e-02 -3.01606327e-01 8.56014729e-01 3.72219115e-01 -4.47935492e-01 -4.69136298e-01 -5.95332146e-01 1.75537854e-01 3.72150183e-01 6.25772476e-01 -1.00470710e+00 1.91417024e-01 -1.40266195e-01 6.38969600e-01 -1.54497910e+00 1.03242898e+00 -1.13380194e+00 -1.34201303e-01 4.17291462e-01 3.90478633e-02 -2.14795455e-01 -3.50438207e-02 7.50934660e-01 -5.47450036e-02 8.35303515e-02 9.03423429e-01 -3.51400197e-01 -9.00865793e-01 8.25554788e-01 3.27337235e-01 -2.14726135e-01 1.07525659e+00 -7.67715991e-01 8.83618072e-02 2.32887268e-02 -6.75641954e-01 1.98348194e-01 8.35988164e-01 4.30287898e-01 1.12413883e+00 -1.40971911e+00 -4.01703864e-01 3.15010220e-01 3.82116497e-01 8.61267090e-01 3.03817689e-01 5.74618101e-01 -6.11314416e-01 5.11300325e-01 -1.03940591e-01 -1.30712020e+00 -8.86939406e-01 5.63770473e-01 4.72387969e-01 1.44606724e-01 -6.43700898e-01 1.18109441e+00 5.23297012e-01 -8.64130020e-01 6.17706895e-01 -6.43539429e-01 3.06007236e-01 -1.84130400e-01 2.78483808e-01 2.82420754e-01 1.08471595e-01 -6.53234720e-01 -5.14391541e-01 1.21410429e+00 1.70877919e-01 -2.79777013e-02 1.57499123e+00 -3.11850786e-01 -7.48222023e-02 4.77520436e-01 1.46337867e+00 -4.99265343e-01 -2.08465195e+00 -1.82187349e-01 -2.23051921e-01 -9.91164923e-01 1.66954070e-01 -6.98750377e-01 -1.06122196e+00 1.05942810e+00 7.34699428e-01 -5.84617555e-01 1.10339212e+00 3.24950784e-01 6.37388825e-01 4.48570937e-01 9.44913268e-01 -8.43160093e-01 7.24656105e-01 4.46260124e-01 1.06963480e+00 -1.46565342e+00 9.26247910e-02 -6.86495423e-01 -1.84473053e-01 9.95785654e-01 1.16825628e+00 -2.76430696e-01 5.74955165e-01 -1.38032092e-02 -1.40578225e-01 -1.89550951e-01 -5.14825940e-01 -3.47808003e-02 3.74796540e-01 6.49413347e-01 -1.36087731e-01 -2.78744459e-01 3.32448155e-01 3.95613492e-01 2.37384588e-02 -2.28145421e-01 2.29972422e-01 9.24978912e-01 -4.38372731e-01 -8.81308794e-01 -5.44434965e-01 1.09694891e-01 -8.63931850e-02 2.28110433e-01 -3.59821379e-01 6.75311029e-01 3.60484213e-01 4.36765254e-01 2.37725422e-01 -2.72893161e-01 4.60618675e-01 -3.27444345e-01 1.09518588e+00 -8.11281502e-01 -2.02940628e-01 2.24127486e-01 -3.44013184e-01 -1.08068466e+00 -5.46912670e-01 -5.64640701e-01 -1.38277042e+00 1.11051112e-01 -2.15969399e-01 -4.18621302e-01 9.57153320e-01 7.19054580e-01 1.65769264e-01 2.05810651e-01 6.25979900e-01 -1.57729673e+00 -3.21739018e-01 -6.88786685e-01 -6.31794751e-01 5.29244728e-02 4.57451254e-01 -8.32328677e-01 -4.95323449e-01 -2.03876883e-01]
[7.613234519958496, -2.665158271789551]
36577112-5b5a-40fe-bdfb-0e3cc4b43722
on-class-imbalance-and-background-filtering
1903.08456
null
http://arxiv.org/abs/1903.08456v2
http://arxiv.org/pdf/1903.08456v2.pdf
On Class Imbalance and Background Filtering in Visual Relationship Detection
In this paper we investigate the problems of class imbalance and irrelevant relationships in Visual Relationship Detection (VRD). State-of-the-art deep VRD models still struggle to predict uncommon classes, limiting their applicability. Moreover, many methods are incapable of properly filtering out background relationships while predicting relevant ones. Although these problems are very apparent, they have both been overlooked so far. We analyse why this is the case and propose modifications to both model and training to alleviate the aforementioned issues, as well as suggesting new measures to complement existing ones and give a more holistic picture of the efficacy of a model.
['Tingting Mu', 'Alessio Sarullo']
2019-03-20
null
null
null
null
['visual-relationship-detection']
['computer-vision']
[ 1.58670112e-01 1.63789257e-01 -2.83662111e-01 -3.72193724e-01 -5.91660403e-02 -4.16899413e-01 7.33377755e-01 6.53368294e-01 -2.93876737e-01 9.30043697e-01 1.21417986e-02 -4.44345832e-01 -5.81236959e-01 -7.55053341e-01 -3.79861534e-01 -5.43936789e-01 -1.56888649e-01 5.97997010e-01 5.00939965e-01 -3.86367947e-01 2.37372771e-01 8.97996604e-01 -2.05120111e+00 2.93495595e-01 7.13743508e-01 6.41660750e-01 -3.25043887e-01 4.33339149e-01 -1.18038192e-01 9.40044641e-01 -8.92962337e-01 -6.51170075e-01 -1.09254934e-01 -4.62759644e-01 -6.35682404e-01 8.26341137e-02 7.28440642e-01 -2.88759291e-01 -4.90665615e-01 7.99431562e-01 6.74404740e-01 -5.71186095e-02 8.18193138e-01 -1.55519521e+00 -5.86282074e-01 2.74735034e-01 -8.56843412e-01 7.65285671e-01 3.75436425e-01 -6.58318400e-02 1.03907228e+00 -7.88543224e-01 8.46414030e-01 1.25173497e+00 6.41598225e-01 4.45970356e-01 -1.24583602e+00 -4.73695606e-01 4.95658964e-01 4.49176878e-01 -1.24555230e+00 -4.43570048e-01 8.17357957e-01 -5.23605764e-01 1.29320955e+00 4.91080880e-01 9.11006093e-01 1.19959354e+00 1.09042022e-02 7.24416614e-01 1.03571165e+00 -4.49338853e-01 -7.00119436e-02 3.62399548e-01 3.47093701e-01 3.83827150e-01 7.37616777e-01 1.44309327e-02 -7.07573175e-01 -1.13644607e-01 6.76229656e-01 -4.13927853e-01 -2.54968792e-01 -7.42038131e-01 -9.27013040e-01 6.69521689e-01 3.38676572e-01 6.32101595e-01 -1.84506148e-01 -2.79988140e-01 3.85042608e-01 4.32402462e-01 6.24457777e-01 4.00674403e-01 -1.67120814e-01 2.24161983e-01 -7.46564448e-01 5.27805626e-01 5.78321457e-01 5.95212758e-01 4.48070794e-01 -1.07876919e-01 -3.33936721e-01 8.55958521e-01 -1.14995919e-01 -3.18656489e-02 1.03717789e-01 -5.30801594e-01 2.40789145e-01 9.15718317e-01 -1.62237808e-01 -1.52086532e+00 -6.34423673e-01 -8.51622164e-01 -7.45973110e-01 4.68870848e-01 5.54028451e-01 4.63217616e-01 -7.82652974e-01 1.52583385e+00 1.63058579e-01 -1.08657882e-01 -2.78010219e-01 8.87014925e-01 1.25438476e+00 1.16569310e-01 1.83774143e-01 -3.86163980e-01 1.16660213e+00 -6.21258020e-01 -8.90477240e-01 -3.78395498e-01 3.87729645e-01 -8.08806002e-01 8.24532926e-01 5.64509630e-01 -1.03799808e+00 -5.42250335e-01 -1.13513803e+00 -2.66907886e-02 -7.46723652e-01 -1.01086628e-02 1.13924074e+00 6.18083775e-01 -1.05672085e+00 7.54635334e-01 -3.59427363e-01 -4.69224930e-01 6.34653986e-01 5.80030620e-01 -4.67382759e-01 2.19501331e-02 -1.11679006e+00 1.26524496e+00 1.04713827e-01 2.02416852e-01 -3.20417583e-01 -3.22357923e-01 -5.15519559e-01 -2.95791812e-02 3.38879943e-01 -7.00848639e-01 7.92281866e-01 -9.08986747e-01 -6.56616092e-01 1.25845110e+00 -7.72179142e-02 -4.00085241e-01 8.09348226e-01 -3.31864543e-02 -5.86212397e-01 -5.75836524e-02 -2.38915101e-01 5.91664195e-01 5.59446454e-01 -1.50904584e+00 -5.46217501e-01 -4.77875829e-01 1.99645832e-01 2.80249149e-01 -4.34806019e-01 7.11548999e-02 -4.53790754e-01 -6.41008556e-01 1.05652899e-01 -4.61951792e-01 -4.46899468e-03 1.49695709e-01 -3.65918338e-01 -2.69589961e-01 8.58764052e-01 -4.01125640e-01 1.38780582e+00 -1.92796922e+00 1.37463868e-01 4.07260098e-02 6.60770893e-01 5.60554564e-01 -3.21034752e-02 2.82398790e-01 -3.41069639e-01 2.54583031e-01 -1.04204670e-01 -4.48575795e-01 -1.82451829e-01 3.24913085e-01 -1.77088261e-01 5.43474555e-01 5.46500921e-01 8.13215911e-01 -8.05415571e-01 -5.87082267e-01 4.31985915e-01 7.77702689e-01 -1.23321228e-01 -1.05166975e-02 -7.92259127e-02 3.01337093e-01 -2.09934086e-01 6.39910161e-01 8.07251692e-01 -1.44971296e-01 2.15498403e-01 -2.54184883e-02 -1.02983102e-01 3.90634120e-01 -1.30938268e+00 7.83763885e-01 1.43497035e-01 1.07891750e+00 -4.03372973e-01 -1.28063512e+00 1.07086730e+00 1.49060460e-02 3.46967489e-01 -1.12048364e+00 2.18655597e-02 2.60882974e-01 2.32168779e-01 -5.79883456e-01 6.35065615e-01 -1.25660390e-01 5.53493977e-01 -1.08347693e-03 -4.77989279e-02 2.77767986e-01 3.45517367e-01 7.75607163e-03 8.55499864e-01 2.65420228e-01 6.27607048e-01 9.05991495e-02 4.83711839e-01 -1.14165023e-02 3.62961113e-01 7.91011333e-01 -3.83854777e-01 7.10765362e-01 9.87581193e-01 -5.96161127e-01 -7.91214645e-01 -9.14387584e-01 -2.28023708e-01 7.78181016e-01 2.65214026e-01 -4.28782731e-01 -1.27478719e-01 -9.33975458e-01 1.24155015e-01 4.86510813e-01 -8.81704748e-01 -3.18920225e-01 -6.64398134e-01 -1.08826351e+00 2.85000086e-01 4.96788979e-01 2.85646375e-02 -9.65795338e-01 -7.01901615e-01 8.65457281e-02 -4.52954471e-02 -1.06201804e+00 5.41665614e-01 3.50968093e-01 -9.05500710e-01 -1.23702681e+00 -6.06221318e-01 -5.97480297e-01 6.60966456e-01 5.38280904e-01 1.70114493e+00 5.07971585e-01 -3.68816316e-01 1.00561954e-01 -3.34517270e-01 -4.93520916e-01 -4.43531275e-01 2.46144049e-02 -1.54495761e-01 -2.98920244e-01 6.73352897e-01 -5.17697513e-01 -6.28874063e-01 3.56903017e-01 -6.86343431e-01 -2.46836826e-01 5.00451505e-01 6.32353485e-01 4.46617723e-01 2.46336833e-01 5.84056973e-01 -9.90939498e-01 7.67105997e-01 -2.56247908e-01 -3.62264991e-01 1.26527935e-01 -7.48095214e-01 -2.89347112e-01 2.72733867e-01 -5.31613410e-01 -6.93540633e-01 -2.38615438e-01 -1.87553942e-01 -3.44873130e-01 -2.80903965e-01 2.95321912e-01 -1.41166821e-02 -8.51793140e-02 7.12080061e-01 -2.94991881e-01 -3.43499303e-01 -4.12181854e-01 -1.38556525e-01 2.16797277e-01 3.92175108e-01 1.49673954e-01 7.77797878e-01 4.02840316e-01 2.46810272e-01 -7.75168359e-01 -8.31424475e-01 -4.59420353e-01 -7.27122188e-01 -4.22446847e-01 4.37092811e-01 -6.74221635e-01 -5.43418646e-01 1.22741729e-01 -1.06808412e+00 1.22968391e-01 -4.03422087e-01 1.19256727e-01 -1.03507414e-01 3.09421092e-01 -1.96046859e-01 -9.49874580e-01 2.27173328e-01 -9.54492271e-01 7.68592477e-01 2.54777163e-01 -3.87353957e-01 -1.01878452e+00 1.83683317e-02 2.16707706e-01 3.43934447e-01 5.67832112e-01 1.15544593e+00 -7.74813354e-01 -3.89311016e-01 -2.85084844e-01 -5.31479716e-01 1.20808281e-01 7.50571415e-02 2.85863936e-01 -1.26294315e+00 -1.56033367e-01 -2.72940516e-01 -7.31570125e-02 1.11199999e+00 4.41034138e-01 1.06350291e+00 2.11365074e-01 -6.38895333e-01 1.54136792e-01 1.28894532e+00 2.05126151e-01 8.33465576e-01 6.40889645e-01 5.54425776e-01 9.47393060e-01 7.53002048e-01 2.11631283e-01 3.33124042e-01 1.14627910e+00 8.54583979e-01 -5.52512646e-01 -4.49515074e-01 -1.23389177e-01 -1.86833039e-01 3.64830196e-01 -3.24973732e-01 -5.44628978e-01 -9.94793653e-01 6.02597833e-01 -1.83440018e+00 -8.41808677e-01 -8.06311309e-01 2.14986515e+00 4.30398017e-01 6.60194993e-01 4.12483722e-01 7.69048154e-01 5.78162432e-01 2.47167751e-01 -2.71009266e-01 -4.27744955e-01 -5.59310555e-01 -2.70837378e-02 8.38338360e-02 3.01501006e-01 -1.16048276e+00 7.02854395e-01 6.93067074e+00 4.75437969e-01 -9.73106265e-01 -3.10035706e-01 9.06178117e-01 1.32361472e-01 -4.72798645e-01 -5.08355573e-02 -8.56284857e-01 3.83436829e-02 4.91947383e-01 2.21233994e-01 -1.34647004e-02 5.45013905e-01 -6.08778931e-02 -3.20453703e-01 -1.10791111e+00 9.19564128e-01 3.16892892e-01 -1.11021054e+00 2.12505817e-01 9.92607847e-02 2.62141675e-01 -4.68349457e-01 7.80010968e-02 2.32523054e-01 -3.23733985e-01 -1.35250974e+00 5.56289673e-01 4.90441918e-01 4.87591714e-01 -8.83601665e-01 9.62875485e-01 -6.40162453e-02 -7.86705911e-01 5.26880585e-02 -4.77693260e-01 -4.44170922e-01 -2.88878143e-01 8.03224981e-01 -7.67209530e-01 6.51598632e-01 7.59696305e-01 6.99382663e-01 -1.24232769e+00 1.42993534e+00 -1.24904558e-01 2.66710639e-01 -4.26109135e-02 1.84563175e-02 -3.54483090e-02 2.88056821e-01 6.51545644e-01 1.05298364e+00 4.81063016e-02 -3.25637996e-01 -3.36571664e-01 6.73458934e-01 2.27269053e-01 1.24051064e-01 -8.00324142e-01 3.56296152e-02 2.85017312e-01 1.02888238e+00 -1.13403487e+00 -1.32293925e-01 -5.37160397e-01 5.91187775e-01 4.64944899e-01 1.95789307e-01 -7.09723055e-01 -2.32163630e-02 5.72056413e-01 3.19694638e-01 1.63805217e-01 -9.17710597e-04 -4.15834308e-01 -1.00350606e+00 1.62368998e-01 -9.29161727e-01 6.63200796e-01 -5.67709029e-01 -1.30144954e+00 6.53072476e-01 1.13156691e-01 -1.44228709e+00 -7.10202381e-02 -6.60558045e-01 -2.50763029e-01 5.21684825e-01 -1.86023223e+00 -9.97222483e-01 -4.04295266e-01 4.02028531e-01 2.79492378e-01 -3.92889464e-03 7.00491130e-01 5.49527943e-01 -6.20407939e-01 5.77192366e-01 -3.74392837e-01 -4.28226173e-01 7.70910442e-01 -1.35609901e+00 3.65253031e-01 8.29768240e-01 3.48293185e-01 5.55688798e-01 1.14178896e+00 -5.49230576e-01 -7.77745306e-01 -5.50178349e-01 1.20050001e+00 -5.89896679e-01 4.69874382e-01 -3.90678585e-01 -1.23902130e+00 2.48708591e-01 2.23275833e-02 1.00454815e-01 5.81300795e-01 3.93737376e-01 -4.61585194e-01 -5.78183755e-02 -8.27271760e-01 6.68161571e-01 1.29841566e+00 -1.73158363e-01 -6.23447835e-01 2.19749361e-01 2.56391734e-01 -3.95918936e-01 -5.03638208e-01 6.93726003e-01 5.25189936e-01 -1.53513467e+00 1.34388089e+00 -5.09227395e-01 5.46529710e-01 -2.15853199e-01 1.97343171e-01 -9.48284030e-01 -2.54288584e-01 -3.04295003e-01 -3.15641463e-01 1.46920037e+00 3.70400459e-01 -5.01800597e-01 8.15836012e-01 1.65909290e-01 1.31153762e-01 -8.75280201e-01 -1.04396617e+00 -6.34161830e-01 -1.48436487e-01 -3.74091476e-01 4.70027477e-01 1.25335252e+00 -2.90102869e-01 5.40029943e-01 -6.06423914e-01 4.24167663e-02 4.59596336e-01 1.19708829e-01 6.59085333e-01 -1.73986542e+00 -4.17766422e-02 -8.51590455e-01 -6.85731471e-01 -3.20848733e-01 -8.04462880e-02 -5.05698860e-01 -5.28065324e-01 -1.91885209e+00 3.71090174e-01 -3.40136796e-01 -5.37833095e-01 3.78993362e-01 -4.42942739e-01 7.24098265e-01 -1.22406892e-02 1.04493067e-01 -5.18428206e-01 2.60287583e-01 1.23240697e+00 -4.41125706e-02 -1.37598649e-01 1.48456752e-01 -8.82379532e-01 6.59740567e-01 7.12672472e-01 -5.33417523e-01 -5.52343011e-01 -2.17539877e-01 7.42657900e-01 -2.21420765e-01 7.89435565e-01 -9.12778080e-01 -8.85982253e-03 1.41507313e-02 8.11318278e-01 -8.58348727e-01 2.83939183e-01 -8.36890578e-01 8.35802034e-02 3.80469233e-01 -2.88520098e-01 1.68913752e-01 4.71532404e-01 4.61814672e-01 -2.62807250e-01 -1.41964614e-01 6.06516659e-01 -6.34677261e-02 -7.75432944e-01 -1.59642380e-02 -2.94022948e-01 -1.53632194e-01 9.10864651e-01 -5.39637804e-01 -6.96883142e-01 -3.67851794e-01 -7.32626081e-01 1.45440012e-01 5.66581845e-01 7.04510450e-01 7.11627126e-01 -1.08027267e+00 -5.37351012e-01 8.99100155e-02 4.19544488e-01 -3.06696445e-01 2.96550930e-01 7.90680289e-01 -2.88890213e-01 2.96287417e-01 -3.96936208e-01 -4.80719477e-01 -1.66688824e+00 7.41057158e-01 4.11563903e-01 -4.22475874e-01 -7.52098680e-01 8.08197200e-01 1.46821644e-02 -4.91421036e-02 5.85780740e-01 6.24930225e-02 -9.00296092e-01 6.47600353e-01 4.25103694e-01 2.53196448e-01 4.39841926e-01 -5.62917292e-01 -7.48632133e-01 5.62442780e-01 -1.25243545e-01 5.18022597e-01 1.52123487e+00 -1.65680021e-01 -7.89202750e-02 5.70449829e-01 7.42799699e-01 3.88088971e-02 -8.84257436e-01 2.28311657e-03 7.67399594e-02 -7.13055015e-01 5.55239012e-03 -1.01985657e+00 -9.80677843e-01 1.06807649e+00 8.33431244e-01 6.52508795e-01 1.31557488e+00 6.19233437e-02 8.54834169e-02 -4.82028835e-02 -1.32897750e-01 -8.91403794e-01 6.51244894e-02 2.25034863e-01 8.95352542e-01 -1.36396348e+00 7.07894981e-01 -6.80756927e-01 -5.76762438e-01 1.21273601e+00 6.48181975e-01 4.52895798e-02 2.00650364e-01 1.84612453e-01 1.95782213e-03 -4.95645344e-01 -8.97083104e-01 -6.79364085e-01 5.13061404e-01 9.66841817e-01 6.56011045e-01 -3.44341785e-01 -5.61889172e-01 -3.67140956e-02 -1.06363045e-02 -1.04067281e-01 5.86024582e-01 8.65918815e-01 -4.56889421e-02 -1.07952201e+00 -2.91233748e-01 4.43367988e-01 -5.40418863e-01 1.20530330e-01 -8.29366744e-01 1.30720389e+00 3.37388873e-01 8.77572596e-01 8.65962803e-02 -4.37729269e-01 7.28995264e-01 -1.81339964e-01 5.78030467e-01 -2.04910770e-01 -6.64657533e-01 5.91450222e-02 5.57572544e-01 -3.12204599e-01 -7.57527232e-01 -5.45017362e-01 -4.77009267e-01 -1.98775232e-01 -2.97241300e-01 -3.01533461e-01 2.24451989e-01 7.33402133e-01 3.46863978e-02 9.07136738e-01 2.54336715e-01 -6.17028892e-01 -1.03660159e-01 -6.59332275e-01 -3.83826256e-01 4.80819941e-01 5.27240098e-01 -1.21115363e+00 -4.63215768e-01 -3.51484925e-01]
[10.258487701416016, 1.7238986492156982]
7eb3b3b1-705d-4745-845b-ab2d3bd526aa
understanding-label-bias-in-single-positive
2305.15584
null
https://arxiv.org/abs/2305.15584v1
https://arxiv.org/pdf/2305.15584v1.pdf
Understanding Label Bias in Single Positive Multi-Label Learning
Annotating data for multi-label classification is prohibitively expensive because every category of interest must be confirmed to be present or absent. Recent work on single positive multi-label (SPML) learning shows that it is possible to train effective multi-label classifiers using only one positive label per image. However, the standard benchmarks for SPML are derived from traditional multi-label classification datasets by retaining one positive label for each training example (chosen uniformly at random) and discarding all other labels. In realistic settings it is not likely that positive labels are chosen uniformly at random. This work introduces protocols for studying label bias in SPML and provides new empirical results.
['Elijah Cole', 'Pietro Perona', 'Julio Arroyo']
2023-05-24
null
null
null
null
['multi-label-learning']
['methodology']
[ 7.58361220e-01 2.76632365e-02 -7.39589155e-01 -8.36472809e-01 -1.17304647e+00 -9.14467156e-01 4.74787354e-01 7.28341639e-01 -8.00830364e-01 1.02634966e+00 -4.01179284e-01 -2.29486689e-01 7.42169991e-02 -5.42466760e-01 -5.75662017e-01 -9.76866066e-01 3.34964618e-02 8.63390028e-01 1.85514659e-01 5.39759696e-01 1.17506355e-01 3.00654352e-01 -1.70574403e+00 6.55384064e-01 4.04104926e-02 7.29922414e-01 -1.24808416e-01 7.99292564e-01 1.65453896e-01 9.75260019e-01 -5.97429335e-01 -4.00110096e-01 1.99595913e-01 -3.65752876e-01 -1.19024563e+00 1.21932733e-03 9.90001917e-01 -7.95208812e-02 6.19581103e-01 1.21188211e+00 4.96692389e-01 -1.07963689e-01 1.03163040e+00 -1.71715474e+00 -4.41534102e-01 5.72848678e-01 -9.90669429e-01 7.78798386e-02 7.84157664e-02 -2.26272807e-01 1.54430211e+00 -8.56019676e-01 7.92109609e-01 1.27972209e+00 9.02029872e-01 7.80915916e-01 -1.48784637e+00 -9.03545022e-01 1.28025725e-01 -9.60872397e-02 -1.32356775e+00 -1.93961233e-01 1.81767374e-01 -4.54046488e-01 4.95322436e-01 5.92217922e-01 1.89573497e-01 1.16392231e+00 2.88646281e-01 8.25838387e-01 1.72690511e+00 -8.68962526e-01 1.18999571e-01 4.73195195e-01 5.76236486e-01 5.80484629e-01 3.49979371e-01 1.04147747e-01 -3.54067624e-01 -6.38150394e-01 1.39184415e-01 -2.73021534e-02 2.49534085e-01 -5.20217299e-01 -1.37172616e+00 1.00880134e+00 3.85653414e-02 2.97806621e-01 -1.49414405e-01 2.86897719e-01 6.20267510e-01 3.51448566e-01 4.34009403e-01 3.68824244e-01 -7.53144741e-01 4.09280300e-01 -9.04756367e-01 2.40992412e-01 4.36646789e-01 9.44685400e-01 9.72979426e-01 -7.20922410e-01 -4.07382697e-01 1.16650963e+00 2.87980735e-01 4.40356046e-01 6.23131216e-01 -1.11985016e+00 -1.67173430e-01 3.94904286e-01 1.60866335e-01 -6.05075359e-01 -7.97803283e-01 -3.86035532e-01 -5.27284682e-01 3.14580947e-01 3.68847847e-01 -4.78560217e-02 -7.58232057e-01 1.75148821e+00 3.80084604e-01 5.43514192e-02 -1.12580165e-01 3.89560103e-01 7.58468032e-01 1.10427596e-01 7.72823334e-01 -3.35480928e-01 1.40152347e+00 -1.01134813e+00 -4.96082157e-01 -3.80410522e-01 1.42724729e+00 -7.29730964e-01 9.56692576e-01 2.76960224e-01 -5.42828083e-01 -2.06242561e-01 -7.07210243e-01 2.56627291e-01 -4.24940199e-01 1.98816106e-01 7.32038796e-01 8.50893259e-01 -1.00565290e+00 2.03931481e-01 -6.09044954e-02 -3.73949081e-01 5.31116784e-01 5.57832181e-01 -7.47257710e-01 -3.43900502e-01 -1.12840331e+00 9.33854282e-01 6.38846815e-01 -3.91072839e-01 -1.05281305e+00 -4.11772996e-01 -6.36525631e-01 -3.46796513e-01 1.54346064e-01 -3.28800619e-01 1.53757882e+00 -1.02773416e+00 -4.23620164e-01 1.71670508e+00 -3.04651797e-01 -1.75945505e-01 3.66718352e-01 3.65732938e-01 -3.44302326e-01 9.33563113e-02 5.99893987e-01 1.12179542e+00 6.15350127e-01 -1.70704341e+00 -1.22825778e+00 -1.43999616e-02 1.72858670e-01 2.12010637e-01 -2.69673854e-01 4.18794423e-01 3.07925671e-01 -3.90814066e-01 1.64484903e-01 -1.25476277e+00 -1.97535917e-01 -1.43228620e-01 -2.91481286e-01 -5.41197360e-01 6.42421067e-01 1.01783596e-01 8.09035361e-01 -1.91591251e+00 -5.06801307e-01 1.87367201e-01 4.54892516e-01 -1.92635074e-01 -2.73282021e-01 -9.53371450e-02 -5.07731080e-01 5.66588104e-01 1.46565855e-01 -4.05838668e-01 -1.17759094e-01 3.77100408e-01 4.60236240e-03 6.95372880e-01 3.54928374e-02 7.41490006e-01 -1.23802030e+00 -1.12147474e+00 1.23467341e-01 -9.94805247e-02 -4.01008278e-01 -2.71177679e-01 2.83634495e-02 3.53196114e-01 6.52855635e-02 8.79141510e-01 5.68187296e-01 -8.44749629e-01 3.68293613e-01 1.69521682e-02 2.02014089e-01 -1.47102207e-01 -1.26134729e+00 7.87639618e-01 -3.85530919e-01 3.02798808e-01 -2.07271770e-01 -8.97255719e-01 4.81608272e-01 6.93145633e-01 7.73658633e-01 -3.61960232e-01 1.18249021e-01 4.03115243e-01 -2.88338482e-01 -4.31355722e-02 2.16827676e-01 -8.08250248e-01 -2.34736115e-01 9.24056828e-01 1.62411809e-01 6.99031651e-02 3.15508634e-01 1.61190286e-01 1.10346174e+00 -4.09400374e-01 5.40713310e-01 -2.52483219e-01 3.91874969e-01 8.99438635e-02 7.74007261e-01 9.60517526e-01 -6.30995095e-01 5.47617078e-01 4.64801013e-01 -3.01404327e-01 -8.65326285e-01 -8.36501181e-01 -5.34428895e-01 1.75213230e+00 8.84800404e-02 -2.01663882e-01 -3.45307022e-01 -1.36589277e+00 -3.33822444e-02 5.31676888e-01 -7.44692385e-01 6.34762645e-02 -1.75909922e-01 -1.16727591e+00 5.30687034e-01 2.23941579e-01 -9.28666517e-02 -1.10182023e+00 -2.22376451e-01 8.88857618e-02 -4.03488696e-01 -8.68275225e-01 -2.99798727e-01 9.00293231e-01 -5.50908804e-01 -1.22160447e+00 -6.18798077e-01 -1.10544503e+00 1.04772353e+00 4.28680331e-01 1.39061487e+00 4.19788569e-01 -3.69965658e-02 2.66951531e-01 -2.85326093e-01 -4.77221429e-01 -7.70883024e-01 4.94652018e-02 5.41269407e-02 8.31734166e-02 6.51332080e-01 1.46137848e-01 -2.33518749e-01 6.12086117e-01 -6.88148975e-01 -8.55943710e-02 4.74089056e-01 1.06177616e+00 9.67564166e-01 3.83506387e-01 9.13968146e-01 -1.58774698e+00 2.94742852e-01 -6.94548190e-01 -9.03476477e-02 6.70984805e-01 -7.41385221e-01 -3.71875048e-01 1.92903578e-01 -7.58103251e-01 -5.69462359e-01 3.92346323e-01 6.54586405e-02 -2.38631163e-02 -5.19824505e-01 -5.81900496e-03 2.91534930e-01 -3.68028224e-01 7.99794257e-01 -2.65552282e-01 -3.86316419e-01 -1.15473494e-01 3.24003786e-01 7.37117171e-01 3.18711884e-02 -5.47894239e-01 2.96490610e-01 4.72557425e-01 1.86206847e-01 -4.73735362e-01 -1.42102444e+00 -9.27891254e-01 -9.26030695e-01 -5.00720680e-01 6.44948542e-01 -9.71726239e-01 -7.03345060e-01 3.39748144e-01 -8.12318742e-01 -2.90379524e-01 -1.70054644e-01 5.29540956e-01 -4.50420558e-01 2.58356929e-01 -5.71770787e-01 -6.26089394e-01 -6.60993829e-02 -1.10856986e+00 1.13411558e+00 -8.04399773e-02 -8.23403716e-01 -1.22736061e+00 2.54703797e-02 3.28992873e-01 -2.84257978e-02 9.43121985e-02 9.95810151e-01 -1.01966083e+00 5.70724644e-02 -5.70784390e-01 -2.20197722e-01 3.66206467e-01 -1.44195287e-02 -1.18233979e-01 -1.19073558e+00 -5.05671799e-01 -1.24032326e-01 -1.08331561e+00 7.12504804e-01 2.89342523e-01 1.00612640e+00 2.48682499e-02 -7.43235588e-01 -2.31907547e-01 1.55820513e+00 -3.70230600e-02 -1.16500139e-01 3.55084091e-01 5.58710456e-01 6.14600062e-01 9.91932571e-01 2.10221738e-01 4.04099882e-01 3.39735746e-01 5.14531136e-01 -3.27284992e-01 -2.48528924e-02 1.97614774e-01 -4.21032161e-02 2.35201329e-01 4.19761211e-01 -4.00022149e-01 -9.99786496e-01 6.73860490e-01 -1.48037958e+00 -8.99080098e-01 -6.05681121e-01 2.16768551e+00 1.17624116e+00 1.17850192e-01 -2.18293928e-02 3.06771487e-01 1.12468982e+00 -1.35466382e-01 -2.72772074e-01 -3.83313149e-02 -4.21038538e-01 1.63690299e-01 8.66210043e-01 5.01925290e-01 -1.67871404e+00 5.27793229e-01 8.03289795e+00 8.89537811e-01 -9.15238440e-01 6.76026046e-01 8.75576317e-01 -3.23584765e-01 -1.98018759e-01 5.01022525e-02 -1.32809865e+00 2.91477293e-01 1.05337870e+00 1.24485329e-01 -3.26889724e-01 9.54494834e-01 -2.79324383e-01 -4.90297943e-01 -1.23524320e+00 8.41530144e-01 -1.17619429e-02 -8.15822840e-01 -1.05537876e-01 4.45262529e-02 1.03127480e+00 5.29454462e-02 -6.07017282e-05 3.35262686e-01 8.62673640e-01 -8.79024088e-01 8.13863873e-01 5.52667715e-02 9.82934237e-01 -4.99798983e-01 8.51326942e-01 4.03501689e-01 -8.79519820e-01 -3.23200315e-01 -2.98354208e-01 1.03843458e-01 -1.65931523e-01 7.51496494e-01 -8.40752423e-01 3.17956172e-02 5.84965706e-01 6.17846012e-01 -9.85755146e-01 9.45099533e-01 -1.59906074e-01 7.24021912e-01 -2.54260123e-01 5.07034510e-02 2.72188962e-01 4.91214603e-01 -1.44267857e-01 1.26307464e+00 5.52320331e-02 -3.56947124e-01 6.06929243e-01 -1.64951812e-02 -1.88254327e-01 3.77158403e-01 -6.37462497e-01 3.53903741e-01 4.52662289e-01 1.53104675e+00 -1.26624179e+00 -4.96255398e-01 -7.02565789e-01 6.30442798e-01 3.61581743e-01 -6.83882311e-02 -7.60921955e-01 1.86306104e-01 7.74563849e-02 -1.88608661e-01 -2.48290956e-01 4.55171525e-01 -4.29566711e-01 -7.22936094e-01 -4.78593647e-01 -7.28586435e-01 9.22207117e-01 -6.04000151e-01 -1.64607048e+00 8.37733075e-02 -9.38020274e-02 -1.18951273e+00 -1.03911601e-01 -5.78034461e-01 1.23248406e-01 8.31305146e-01 -1.40283775e+00 -1.41867495e+00 -4.20728996e-02 2.17741281e-01 1.54365733e-01 1.33549303e-01 1.29719007e+00 5.13063908e-01 -3.66559029e-01 7.12711632e-01 1.09413862e-01 -9.74755287e-02 1.36621594e+00 -1.48228776e+00 -1.73946202e-01 2.44862944e-01 1.30234689e-01 4.11666632e-01 4.78728682e-01 -5.41544497e-01 -2.83243746e-01 -1.22794592e+00 1.37492406e+00 -7.35790670e-01 4.93008077e-01 -5.28643057e-02 -8.06994677e-01 9.83249724e-01 -1.70398757e-01 4.06244189e-01 1.30283976e+00 1.92677006e-01 -4.33380365e-01 2.01800495e-01 -1.42741549e+00 2.08201826e-01 7.06297636e-01 -5.18986821e-01 -8.62935036e-02 1.03638196e+00 2.45651767e-01 -9.01216194e-02 -1.00077641e+00 4.59614515e-01 5.93341708e-01 -6.10217929e-01 9.32570934e-01 -4.99520212e-01 2.63267964e-01 -2.97821164e-01 -2.93840379e-01 -1.15257204e+00 -4.96518642e-01 2.96849430e-01 2.82941967e-01 1.24557281e+00 6.63670242e-01 -6.28946543e-01 7.22933292e-01 6.83944941e-01 3.85066569e-01 -6.50413454e-01 -8.00807059e-01 -7.69151747e-01 3.43324900e-01 -2.15005800e-01 3.52639347e-01 1.57071853e+00 -2.40932092e-01 5.57515562e-01 -3.57591361e-01 1.76944077e-01 8.62372398e-01 -1.82212424e-02 3.23535562e-01 -1.67993510e+00 -3.45780738e-02 -3.45647037e-01 -2.02537835e-01 -4.60558027e-01 4.77622867e-01 -1.34836018e+00 1.79856479e-01 -1.37500763e+00 8.55977178e-01 -1.10227108e+00 -7.82846987e-01 1.00937760e+00 -3.27421874e-01 1.02503061e+00 -1.99948192e-01 4.06493396e-01 -1.02680242e+00 -4.34541523e-01 1.07868791e+00 -2.70582467e-01 4.31745708e-01 -1.37008645e-03 -7.51230657e-01 7.69847751e-01 8.34561348e-01 -1.15502548e+00 -2.96918362e-01 1.05861053e-01 3.98208618e-01 -2.98947632e-01 3.23705375e-01 -7.70471871e-01 4.65736836e-02 -4.45424408e-01 3.03293675e-01 -6.81737483e-01 6.47780895e-02 -9.59131360e-01 2.83197790e-01 3.72495174e-01 -8.32288742e-01 -2.68896222e-01 -8.85065868e-02 4.80646282e-01 6.86045736e-02 -9.17328656e-01 1.31572568e+00 -4.82714087e-01 -8.79430413e-01 3.80087532e-02 -2.61805773e-01 1.19194314e-01 1.51715970e+00 1.24660447e-01 -3.83868009e-01 2.39255354e-01 -9.16547835e-01 2.38311410e-01 5.95917881e-01 1.82996765e-01 5.98448068e-02 -1.52764583e+00 -6.84934080e-01 6.73454851e-02 6.55596375e-01 -4.74756181e-01 1.07073784e-01 6.70524299e-01 -1.14671834e-01 3.29626679e-01 -6.82365447e-02 -7.19882727e-01 -1.97235632e+00 6.53794408e-01 4.34580684e-01 -4.89959657e-01 -6.58451021e-02 1.11348963e+00 2.13064030e-01 -8.62927198e-01 1.54266670e-01 2.93644607e-01 -3.20045978e-01 4.88097280e-01 5.24911761e-01 2.67980933e-01 6.67685717e-02 -8.55449677e-01 -4.06432420e-01 6.46033585e-01 -3.11054498e-01 -1.20315123e-02 9.18951988e-01 -1.63787574e-01 -5.29147863e-01 1.06140590e+00 1.33618152e+00 -2.47379109e-01 -5.27339578e-01 -1.52881876e-01 2.83001065e-01 -3.06098998e-01 -5.98144755e-02 -9.19069707e-01 -7.27308273e-01 6.27852976e-01 6.96407676e-01 2.69078910e-01 7.43073285e-01 2.14413837e-01 1.96649849e-01 3.08557034e-01 7.50289321e-01 -1.17009580e+00 1.01392530e-01 1.81665465e-01 3.32633227e-01 -1.74874353e+00 5.16399965e-02 -5.37699819e-01 -5.20388484e-01 7.65493631e-01 5.70171773e-01 3.37682158e-01 7.44983017e-01 1.83144376e-01 3.76828939e-01 -2.51336336e-01 -7.79614627e-01 7.32430816e-02 -9.58153158e-02 4.61649090e-01 7.74853945e-01 3.97686303e-01 -4.83625084e-01 -5.25887609e-02 2.64094174e-01 -1.17267422e-01 5.62794983e-01 1.14135814e+00 -4.45017278e-01 -1.48089731e+00 -5.97663105e-01 1.05276930e+00 -8.54674459e-01 8.85013342e-02 -2.53460228e-01 4.77397829e-01 5.63576818e-01 1.16540217e+00 -1.13592036e-01 -1.90236315e-01 -2.99351476e-03 5.52304566e-01 4.52191412e-01 -1.15153956e+00 -7.34297454e-01 -2.03585923e-01 1.43794626e-01 -6.24947585e-02 -9.30399537e-01 -8.39036226e-01 -1.33692181e+00 -1.61975607e-01 -7.14929104e-01 -5.93643412e-02 3.73287261e-01 7.68394411e-01 -1.92011073e-01 3.17230284e-01 6.04795635e-01 -4.08278853e-01 -4.30678189e-01 -9.53223288e-01 -8.76302004e-01 5.80037534e-01 3.30621660e-01 -9.52847362e-01 -6.77040279e-01 2.48816758e-01]
[9.52323055267334, 4.256229400634766]
1b712f84-0eb4-4963-9fc0-ce24d46c3c21
reconfigurable-intelligent-surface-assisted-24
2307.04438
null
https://arxiv.org/abs/2307.04438v1
https://arxiv.org/pdf/2307.04438v1.pdf
Reconfigurable Intelligent Surface Assisted Railway Communications: A survey
The number of train passengers and the demand for high data rates to handle new technologies such as video streaming and IoT technologies are continuously increasing. Therefore the exploration of millimeter waves (mmWave) band is a key technology to meet this demand. However, the high penetration loss makes mmWave very sensitive to blocking, limiting its coverage area. One promising, efficient, and low-cost solution is the reconfigurable intelligent surface (RIS). This paper reviews the state of the art of RIS for railway communications in the mmWave context. First, we present the different types of RIS and review some optimization algorithms used in the literature to find the RIS phase shift. Then, we review recent works on RIS in the railway domain and provide future directions.
['Marion Berbineau', 'Charlotte Langlais', 'Ammar El Falou', 'Aline Habib']
2023-07-10
null
null
null
null
['blocking']
['natural-language-processing']
[ 9.56258923e-02 1.08314551e-01 -1.27276167e-01 -1.56005710e-01 -3.02252293e-01 -3.14969182e-01 -1.27908155e-01 -3.53081018e-01 -1.96521297e-01 9.09161866e-01 -5.91490418e-03 -3.74472439e-01 -5.84855080e-01 -1.39528382e+00 -2.27725387e-01 -1.12241435e+00 -7.67409950e-02 2.09530115e-01 1.46418065e-01 -5.90529978e-01 2.58159906e-01 7.65358210e-01 -1.26858187e+00 -5.03183827e-02 7.95301199e-01 1.13293850e+00 3.09648514e-01 1.61614522e-01 -7.05749020e-02 -9.25827920e-02 -8.26302230e-01 -3.90474409e-01 1.44540340e-01 -4.39633310e-01 -3.91238816e-02 -1.38202980e-01 -2.69592643e-01 2.13099584e-01 -7.27747738e-01 6.45789802e-01 1.00798166e+00 -8.21087975e-03 4.61297065e-01 -7.83774257e-01 2.98282802e-01 6.02892339e-01 -5.96615374e-01 1.87512040e-01 1.85175359e-01 -4.48705018e-01 3.20467144e-01 -3.64589989e-01 4.43124771e-01 4.00452852e-01 3.15238625e-01 2.12730259e-01 -3.27376902e-01 -6.11660957e-01 -5.93013689e-02 4.07683313e-01 -1.36007285e+00 -3.71605128e-01 7.38890707e-01 7.52997100e-02 6.38556004e-01 6.36997938e-01 9.03843701e-01 5.05330563e-01 5.05738556e-01 1.30101472e-01 7.84627080e-01 -6.47967577e-01 2.17406094e-01 -1.43260090e-02 -1.04940146e-01 5.55819392e-01 8.34214568e-01 1.26525909e-01 -6.14224136e-01 3.66559982e-01 7.68816650e-01 -1.35029912e-01 -4.37283516e-01 -8.18681940e-02 -7.93878734e-01 3.26072901e-01 5.31333148e-01 8.25479388e-01 -4.75136936e-01 3.03389788e-01 -2.88836658e-01 4.90290552e-01 3.14115226e-01 6.15669966e-01 -2.89470077e-01 -2.17698544e-01 -5.52122772e-01 -6.73149154e-02 5.08711100e-01 1.18270648e+00 2.83541650e-01 1.34461299e-01 -2.21406370e-02 6.03580832e-01 6.40252352e-01 8.01702678e-01 -2.86696672e-01 -4.00884598e-02 7.67190814e-01 2.25961521e-01 -1.49223253e-01 -6.48575187e-01 -1.02034271e+00 -1.48700094e+00 -8.27770293e-01 -1.99386388e-01 7.52479285e-02 -6.24573648e-01 -7.03257680e-01 9.48494196e-01 -1.57667436e-02 1.46602578e-02 2.60417640e-01 6.67540550e-01 9.60505962e-01 7.27924109e-01 -6.18649721e-01 -5.53791523e-01 1.29198670e+00 -4.41507876e-01 -7.07435071e-01 -4.23517585e-01 4.37473476e-01 -8.19665492e-01 2.02872187e-01 3.81933421e-01 -1.11764276e+00 1.22471653e-01 -1.35509360e+00 8.20707977e-01 5.80245927e-02 1.56340990e-02 4.55945730e-01 1.48068094e+00 -4.35174614e-01 -1.86931834e-01 -6.65061653e-01 -3.07263970e-01 3.15068901e-01 4.80638444e-01 4.01179373e-01 -3.08343589e-01 -1.16790771e+00 7.74584591e-01 -1.83289036e-01 2.87937045e-01 1.34453759e-01 -7.40990162e-01 -2.83582628e-01 -3.72154042e-02 -8.13565552e-02 -5.97321272e-01 7.20943272e-01 -1.51854269e-02 -1.79295456e+00 2.98523217e-01 6.08483776e-02 -2.31235638e-01 2.51359373e-01 1.22438774e-01 -8.51383567e-01 1.04231074e-01 -4.99791950e-01 -3.53056937e-01 2.68967450e-01 -9.91121233e-01 -9.77268696e-01 -3.77821088e-01 -3.12335640e-02 -1.52820870e-02 -2.45370746e-01 -1.91614628e-01 -1.96763828e-01 -4.66823816e-01 8.17213237e-01 -9.56858099e-01 -4.39475924e-01 -8.63339424e-01 -1.92666501e-01 2.87291974e-01 6.50656760e-01 -8.62508416e-02 1.20808792e+00 -1.94079936e+00 -5.40869720e-02 7.89246857e-01 -2.01670974e-01 -6.91319630e-02 1.42325327e-01 8.41392696e-01 3.64059597e-01 -4.88657504e-01 1.99813992e-02 3.99674773e-01 -3.43471080e-01 -7.53271878e-02 -3.95950526e-02 7.30147302e-01 -5.73197544e-01 5.87946296e-01 -3.88092250e-01 2.79556125e-01 7.61744156e-02 4.55312729e-01 -4.15386140e-01 -1.19987756e-01 3.05466205e-01 1.03891909e+00 -7.26660132e-01 7.64263868e-01 1.02524960e+00 3.49423647e-01 -1.07358843e-02 -2.21520722e-01 -6.73065722e-01 2.82854587e-01 -1.04236746e+00 1.24656999e+00 -9.02960837e-01 5.79589963e-01 -4.28505056e-02 -1.10055888e+00 1.30219734e+00 4.60530519e-01 8.99434924e-01 -1.15840852e+00 9.30759534e-02 6.41325533e-01 5.48419505e-02 -9.31423545e-01 1.33156344e-01 -4.50079083e-01 -3.87436122e-01 1.86585158e-01 -3.30400974e-01 -2.41030753e-01 3.31063420e-02 -3.54154587e-01 1.39271820e+00 -2.76194245e-01 -2.27887407e-01 -3.94955546e-01 4.18044388e-01 -1.84739411e-01 6.62539065e-01 4.38757628e-01 3.99893373e-01 1.42305121e-01 -9.67613608e-02 -3.31650943e-01 -3.24509114e-01 -1.12167180e+00 -3.82543832e-01 4.16345596e-01 7.99897373e-01 9.19281617e-02 -2.67234027e-01 -1.29042879e-01 -1.74923792e-01 6.73659563e-01 4.15165424e-01 -5.35150170e-02 -7.65399456e-01 -1.35260165e+00 1.88546732e-01 9.43118110e-02 9.22156811e-01 -4.33256835e-01 -8.78569245e-01 4.94139284e-01 -5.25521696e-01 -1.31520975e+00 3.37607324e-01 7.87248537e-02 -1.04528582e+00 -5.63332498e-01 -6.36434913e-01 -9.95925367e-01 7.31789887e-01 4.55781460e-01 8.73923302e-01 -2.29630798e-01 -2.91142702e-01 3.16866368e-01 -5.66126525e-01 -5.98048925e-01 1.08908199e-01 4.30184424e-01 3.96468937e-02 -6.55182302e-02 9.75828394e-02 -7.93657064e-01 -7.75973380e-01 6.46623552e-01 -3.77463669e-01 -5.34314699e-02 7.92916715e-01 1.65028945e-01 2.15502188e-01 7.96617627e-01 1.07097280e+00 -6.83620393e-01 -9.80732380e-04 -4.32297319e-01 -8.57938588e-01 4.38745245e-02 -2.91550487e-01 -1.27588332e-01 1.32330924e-01 4.06416029e-01 -1.20788252e+00 -1.56806409e-01 -7.58021951e-01 9.18965816e-01 1.16801701e-01 6.17146552e-01 -4.37108189e-01 -7.49070823e-01 4.11352336e-01 -1.70130402e-01 -7.45260894e-01 -2.50318438e-01 -1.98830083e-01 8.25944424e-01 1.09572224e-01 -9.79939401e-02 1.06440520e+00 5.69483995e-01 5.19772291e-01 -1.49259651e+00 -7.49035954e-01 -4.74439830e-01 -4.34763543e-02 -7.58528829e-01 4.75392997e-01 -7.02920854e-01 -6.54727221e-01 1.39340758e-01 -1.01679146e+00 -1.23899383e-02 -3.37848030e-02 1.17149186e+00 -4.61584091e-01 -2.51644820e-01 -2.67501563e-01 -9.63173687e-01 -2.41685912e-01 -7.66973376e-01 3.91839981e-01 4.80949461e-01 1.56130701e-01 -5.84817350e-01 -9.57915932e-02 5.59416831e-01 8.36247563e-01 3.56238693e-01 8.51962805e-01 3.23291242e-01 -1.09749472e+00 -5.70376337e-01 4.53938358e-02 -6.48107648e-01 9.14701521e-02 -7.77461946e-01 -7.24340141e-01 -2.67310888e-01 3.33216518e-01 7.69217074e-01 7.01379895e-01 8.18593264e-01 7.94812202e-01 -1.28161106e-02 -7.72609472e-01 8.96513760e-01 1.69813466e+00 4.99787092e-01 1.22020674e+00 3.82480472e-01 9.23852995e-02 6.31824493e-01 8.85282397e-01 4.57519382e-01 1.57750443e-01 7.20686316e-01 6.19300902e-01 -8.21505487e-02 1.93675280e-01 4.95122820e-01 4.62238789e-02 1.15552163e+00 -7.82314003e-01 -7.26979494e-01 -6.20776236e-01 5.03522418e-02 -1.56266582e+00 -8.54455292e-01 -4.93733495e-01 2.25929928e+00 -5.42134531e-02 1.88660771e-01 -1.07034363e-01 3.95142853e-01 5.92830718e-01 -2.40238115e-01 2.01437343e-02 -6.30341470e-02 -1.37933746e-01 4.66568708e-01 1.30829656e+00 2.88637578e-01 -6.51031852e-01 4.01705325e-01 5.87437582e+00 5.29144645e-01 -1.15649009e+00 1.05664104e-01 4.81813215e-02 -2.09459737e-01 -6.46327972e-01 -4.44041669e-01 -8.37333322e-01 1.87363133e-01 7.11005569e-01 -5.63810989e-02 -2.73734331e-02 -1.75710171e-02 3.76829773e-01 -3.99677217e-01 -1.68377995e-01 1.23748839e+00 -8.31929147e-02 -1.50878274e+00 -4.94995922e-01 3.07065248e-01 8.35435927e-01 -2.75950041e-02 8.91782418e-02 -3.78220528e-02 -5.13657928e-01 -4.77906197e-01 4.90607411e-01 5.36265731e-01 6.21048510e-01 -1.19598484e+00 1.00533056e+00 2.49243870e-01 -1.29391885e+00 -1.75796375e-01 -2.91835696e-01 -1.00501083e-01 5.01453757e-01 1.23457658e+00 -3.07148069e-01 1.05916309e+00 6.76871359e-01 3.36269766e-01 1.89692229e-01 1.69089806e+00 -2.56230205e-01 6.12216532e-01 -4.75593626e-01 -4.96289760e-01 -4.61609550e-02 -4.94011372e-01 7.97019064e-01 8.89762819e-01 1.00492871e+00 4.13491547e-01 -1.92986071e-01 -1.25655327e-02 7.82768875e-02 2.71245912e-02 -5.60596824e-01 5.12277424e-01 3.22358459e-01 9.17871058e-01 -9.57180560e-01 4.77551967e-01 -6.20804012e-01 3.88576776e-01 -5.98361135e-01 5.60483456e-01 -9.02864575e-01 -8.14878285e-01 6.50071561e-01 5.63688993e-01 3.09969243e-02 -8.50847363e-01 -7.13012755e-01 -5.18444896e-01 2.69572977e-02 -6.63403496e-02 2.64959306e-01 -4.07867879e-02 -7.06603110e-01 4.95512277e-01 -8.70389864e-02 -1.65481341e+00 3.59215766e-01 -2.23369926e-01 -3.40793490e-01 4.89993930e-01 -1.81648290e+00 -6.46385729e-01 -5.52226245e-01 2.56816000e-02 4.04058337e-01 -2.13498846e-01 6.43999577e-01 7.99305499e-01 -2.92612791e-01 4.26416427e-01 5.09100676e-01 -4.15377766e-01 4.10537153e-01 -2.91463196e-01 4.57683392e-02 6.17141128e-01 -1.87056392e-01 7.38484040e-02 9.86769915e-01 -4.62987393e-01 -1.97748721e+00 -8.08107078e-01 6.95942998e-01 3.48144948e-01 9.04997140e-02 -3.61955673e-01 -3.49677494e-03 1.74948677e-01 6.40659630e-02 -3.90185118e-01 9.32256103e-01 -3.06771393e-03 4.32985038e-01 -6.38111889e-01 -1.23442030e+00 7.37743020e-01 1.33143330e+00 4.93282229e-01 1.81373194e-01 3.52766305e-01 1.24374546e-01 -6.72673047e-01 -7.35287666e-01 7.76540041e-01 6.99762821e-01 -7.24575222e-01 8.85836124e-01 1.92458764e-01 -1.65560707e-01 -1.97271898e-01 -1.63421020e-01 -1.45463955e+00 -4.48125958e-01 -7.78883696e-01 1.98394150e-01 9.30365741e-01 6.21590793e-01 -1.13059103e+00 1.32959092e+00 1.98999885e-02 -4.71422851e-01 -9.05538857e-01 -1.37415433e+00 -1.00551331e+00 -4.73737329e-01 -4.31371093e-01 6.83237016e-01 2.94537067e-01 4.68867093e-01 3.58486146e-01 -2.26398438e-01 6.21257424e-01 7.57847369e-01 2.44071141e-01 4.72276926e-01 -1.37100983e+00 -2.75896341e-01 -3.10191274e-01 -9.44817543e-01 -1.39754093e+00 -5.26390970e-01 -7.61040330e-01 1.64075345e-02 -2.16421890e+00 -5.83259821e-01 -9.64912951e-01 -1.98645759e-02 -3.97665381e-01 6.44214272e-01 6.87527239e-01 -2.80740380e-01 -2.90040642e-01 -5.03569432e-02 4.25211132e-01 1.20646107e+00 -2.69143790e-01 -3.92129362e-01 1.15823388e+00 -4.75003392e-01 5.92448294e-01 1.27271235e+00 -4.06052887e-01 -9.26208422e-02 -4.66255963e-01 9.36955988e-01 2.84630328e-01 -4.62010831e-01 -1.44833064e+00 4.20811743e-01 -3.39772373e-01 2.55980015e-01 -7.50801265e-01 3.54311854e-01 -1.25199533e+00 1.72878087e-01 8.56569767e-01 4.30355519e-01 -3.58944029e-01 -8.87697339e-02 5.52259445e-01 -1.60201043e-02 -3.10666531e-01 9.22333479e-01 3.53633702e-01 -3.56549293e-01 2.81639606e-01 -1.05870473e+00 -3.25904846e-01 1.33865392e+00 -4.22582418e-01 -4.74130422e-01 -4.58523601e-01 -4.40052003e-01 2.88933873e-01 -2.51519620e-01 3.05200398e-01 6.37341917e-01 -9.77948666e-01 -7.94533432e-01 1.10325322e-01 1.80772543e-01 -3.24158847e-01 4.48383927e-01 9.36885953e-01 -7.87499964e-01 6.31157160e-01 -1.21218406e-01 -4.67866212e-01 -1.11527145e+00 -8.98374617e-02 4.32145655e-01 6.33796379e-02 -7.78268278e-01 8.87434661e-01 -2.38229200e-01 1.73869386e-01 1.21403866e-01 1.65259093e-01 -3.95821333e-01 -1.64403602e-01 3.34582269e-01 7.74105549e-01 7.07431853e-01 -2.68977016e-01 -6.06918454e-01 1.15681183e+00 7.14246929e-01 -3.15779984e-01 1.48720837e+00 -3.33937824e-01 -8.28901827e-02 2.06845418e-01 5.64504623e-01 6.00390434e-01 -4.57801253e-01 2.04383418e-01 -6.05917117e-03 -6.94978774e-01 8.93396437e-02 -4.16369438e-01 -1.36561036e+00 4.30029541e-01 5.87968230e-01 2.78879344e-01 1.36502278e+00 1.29340678e-01 9.42638218e-01 5.38926899e-01 1.33344984e+00 -1.20972800e+00 -1.08228110e-01 4.19579327e-01 6.04640543e-01 -2.93784261e-01 2.11320996e-01 -1.08184004e+00 1.44343972e-01 1.29532111e+00 1.91565156e-01 -1.92065295e-02 1.17648721e+00 7.58099318e-01 1.46766946e-01 -2.94911236e-01 -1.15595110e-01 -3.44449967e-01 -2.40930244e-01 6.87183321e-01 6.82092011e-01 -1.94346849e-02 -8.57124388e-01 3.40174228e-01 -4.97683018e-01 -3.71425092e-01 5.75432181e-01 1.18359423e+00 -8.81096363e-01 -1.43003833e+00 -5.97933292e-01 7.37931073e-01 -4.73246843e-01 5.22700906e-01 3.43450278e-01 2.48900965e-01 2.54757255e-02 1.50287080e+00 6.80711470e-04 -4.02671605e-01 8.48953247e-01 -5.07959962e-01 1.03898263e+00 -4.83226299e-01 -3.95363331e-01 -3.49386856e-02 3.63234699e-01 -5.87903103e-03 -4.37522620e-01 -3.96358788e-01 -1.37270844e+00 -1.43181771e-01 -6.14760637e-01 4.00702417e-01 1.19061220e+00 1.02115035e+00 2.04475209e-01 1.01182568e+00 1.00938570e+00 -2.88238555e-01 3.04860562e-01 -4.60266203e-01 -8.12119722e-01 -6.37011707e-01 -1.03783540e-01 -7.69638896e-01 3.67458235e-03 -5.66631556e-01]
[6.257157325744629, 1.2192310094833374]
c62afdee-2fd1-4cac-9df7-72a58e21fe13
refteacher-a-strong-baseline-for-semi
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Sun_RefTeacher_A_Strong_Baseline_for_Semi-Supervised_Referring_Expression_Comprehension_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Sun_RefTeacher_A_Strong_Baseline_for_Semi-Supervised_Referring_Expression_Comprehension_CVPR_2023_paper.pdf
RefTeacher: A Strong Baseline for Semi-Supervised Referring Expression Comprehension
Referring expression comprehension (REC) often requires a large number of instance-level annotations for fully supervised learning, which are laborious and expensive. In this paper, we present the first attempt of semi-supervised learning for REC and propose a strong baseline method called RefTeacher. Inspired by the recent progress in computer vision, RefTeacher adopts a teacher-student learning paradigm, where the teacher REC network predicts pseudo-labels for optimizing the student one. This paradigm allows REC models to exploit massive unlabeled data based on a small fraction of labeled. In particular, we also identify two key challenges in semi-supervised REC, namely, sparse supervision signals and worse pseudo-label noise. To address these issues, we equip RefTeacher with two novel designs called Attention-based Imitation Learning (AIL) and Adaptive Pseudo-label Weighting (APW). AIL can help the student network imitate the recognition behaviors of the teacher, thereby obtaining sufficient supervision signals. APW can help the model adaptively adjust the contributions of pseudo-labels with varying qualities, thus avoiding confirmation bias. To validate RefTeacher, we conduct extensive experiments on three REC benchmark datasets. Experimental results show that RefTeacher obtains obvious gains over the fully supervised methods. More importantly, using only 10% labeled data, our approach allows the model to achieve near 100% fully supervised performance, e.g., only -2.78% on RefCOCO.
['Rongrong Ji', 'Zhiyu Wang', 'Guannan Jiang', 'Xiaoshuai Sun', 'Yiyi Zhou', 'Gen Luo', 'Jiamu Sun']
2023-01-01
null
null
null
cvpr-2023-1
['referring-expression', 'pseudo-label']
['computer-vision', 'miscellaneous']
[ 2.40825608e-01 4.21503812e-01 -5.08956552e-01 -6.24591470e-01 -9.59487677e-01 -4.23995852e-01 3.60367775e-01 -1.91735268e-01 -4.80233431e-01 6.67740583e-01 -1.19217369e-03 -1.35803863e-01 1.91600934e-01 -4.30782050e-01 -8.48258018e-01 -6.74140394e-01 5.85629702e-01 4.33235735e-01 1.39027357e-01 -8.96780491e-02 1.11756045e-02 4.20681387e-02 -1.39768898e+00 1.76171824e-01 1.20593345e+00 9.06975448e-01 1.81748077e-01 2.54820347e-01 -1.24244161e-01 1.26270676e+00 -4.68002439e-01 -3.87037188e-01 -1.59139723e-01 -5.24420917e-01 -8.61256778e-01 2.82799006e-01 3.30524921e-01 -4.09708560e-01 -2.12681726e-01 1.13928032e+00 3.91028434e-01 2.44619265e-01 7.61671960e-01 -1.26946855e+00 -8.25497806e-01 1.09872735e+00 -6.99800372e-01 -1.03814609e-01 2.55785614e-01 2.15201110e-01 1.07750499e+00 -1.09409797e+00 2.61945784e-01 1.39300740e+00 4.28515941e-01 8.77789676e-01 -1.23137736e+00 -1.12081790e+00 4.24809068e-01 3.19693685e-01 -1.33176506e+00 -5.26386499e-01 1.07092345e+00 -2.91140229e-01 4.35987413e-01 -1.39133275e-01 1.84244946e-01 1.19997692e+00 -6.08173668e-01 1.28207350e+00 1.32698429e+00 -4.57291663e-01 1.39838174e-01 3.04204136e-01 6.22447133e-01 8.22795212e-01 -1.44363999e-01 -3.58280912e-02 -5.64454496e-01 3.34481336e-02 5.56840301e-01 7.33811110e-02 -5.12665033e-01 -3.60561818e-01 -1.09792185e+00 6.93509698e-01 5.54936111e-01 1.42343089e-01 -6.98441640e-02 1.69211566e-01 2.31364474e-01 4.11743373e-01 2.55986661e-01 3.21040064e-01 -5.69275856e-01 -1.09198056e-01 -6.03588343e-01 -2.80323863e-01 5.45109212e-01 1.30819309e+00 9.40276861e-01 7.92640448e-02 -3.17105085e-01 1.11231315e+00 4.46307451e-01 4.58968073e-01 7.13615417e-01 -1.15662324e+00 5.22928357e-01 7.11753070e-01 3.86090055e-02 -6.21793568e-01 -1.77869797e-01 -3.91605377e-01 -7.95871735e-01 -2.74084538e-01 4.41340297e-01 -1.89782694e-01 -7.88489640e-01 1.95385933e+00 3.04239124e-01 5.16842067e-01 1.01550594e-01 7.96049953e-01 9.79334652e-01 5.02820194e-01 2.33781204e-01 -3.18410307e-01 1.14320600e+00 -1.57941163e+00 -7.06414878e-01 -5.70069812e-02 8.75136316e-01 -4.00673151e-01 1.35697377e+00 3.82146269e-01 -9.56063688e-01 -5.86617768e-01 -8.97245944e-01 1.95926726e-02 -2.85231945e-04 3.71972948e-01 5.07760286e-01 3.43544275e-01 -7.27769971e-01 3.88452858e-01 -7.26210952e-01 6.78392351e-02 6.15611553e-01 3.64719987e-01 -1.26880348e-01 -2.46178806e-01 -1.08097064e+00 5.65356612e-01 2.37227440e-01 -2.60924231e-02 -1.24339020e+00 -6.84005380e-01 -8.90770197e-01 2.68260717e-01 6.63321078e-01 -1.56611696e-01 1.61247563e+00 -1.32881117e+00 -1.88881886e+00 9.71712232e-01 -2.25022897e-01 -2.12482944e-01 5.44137895e-01 -1.93615437e-01 -2.48502314e-01 1.90680951e-01 1.08450770e-01 7.74074435e-01 8.81515741e-01 -1.32047570e+00 -5.29828787e-01 -2.16759965e-01 1.84654575e-02 1.34557500e-01 -5.05240679e-01 -7.22483620e-02 -3.66440982e-01 -4.74698812e-01 1.46368533e-01 -9.73332942e-01 1.77758168e-02 -6.18712157e-02 -4.32867914e-01 -9.64771926e-01 6.12293899e-01 -5.60915656e-03 1.01364720e+00 -2.23232985e+00 9.84906852e-02 -1.80016816e-01 4.53651190e-01 4.60092723e-01 -4.32378232e-01 -1.28553897e-01 -2.77994275e-01 3.11571136e-02 -2.63163030e-01 -4.63416278e-01 -6.33147582e-02 2.61662126e-01 -3.89003128e-01 3.69589716e-01 3.26367080e-01 1.01575577e+00 -1.25285923e+00 -5.55507660e-01 -1.64716512e-01 1.65473670e-01 -5.06635010e-01 8.31779003e-01 -2.99106151e-01 6.57781184e-01 -7.92802870e-01 7.47368753e-01 4.76628393e-01 -6.28765821e-01 8.76813382e-02 -1.51214495e-01 2.14211613e-01 2.50419319e-01 -9.58955467e-01 1.56183708e+00 -5.84335148e-01 3.42598349e-01 1.55816212e-01 -1.32781839e+00 9.99507904e-01 3.25938404e-01 2.24324897e-01 -6.05991960e-01 3.02105635e-01 2.73401082e-01 -1.61538795e-01 -5.71122408e-01 7.48422742e-02 1.16927363e-01 7.51959309e-02 7.65448451e-01 3.23807359e-01 1.02293245e-01 -1.10958286e-01 3.53760034e-01 8.06270003e-01 3.49354774e-01 8.53717998e-02 -1.06943853e-01 8.38329792e-01 -1.48808569e-01 8.04396451e-01 6.05310440e-01 -5.55281639e-01 3.81551445e-01 4.51381177e-01 -8.68325382e-02 -4.74945128e-01 -7.42759109e-01 2.03478653e-02 1.64463043e+00 3.68507504e-01 -6.82263300e-02 -8.10553074e-01 -1.18586540e+00 -1.35782257e-01 7.89821506e-01 -5.74916840e-01 -3.55870068e-01 -6.30341470e-01 -4.93284225e-01 4.52817023e-01 6.71493471e-01 6.85658991e-01 -1.19892395e+00 -2.72433043e-01 5.73665649e-02 -2.10097998e-01 -1.14483786e+00 -7.40045130e-01 2.99686521e-01 -7.65429199e-01 -1.13484943e+00 -7.47754097e-01 -9.11843300e-01 1.03596246e+00 6.00263357e-01 1.12288833e+00 2.58093476e-01 3.24445486e-01 4.03180361e-01 -5.03030956e-01 -1.57779649e-01 -3.57182771e-01 1.54939622e-01 9.67944115e-02 1.52192056e-01 6.03689313e-01 -6.73456907e-01 -3.07449818e-01 6.09725773e-01 -5.81928194e-01 1.74967512e-01 6.19584739e-01 1.22507417e+00 5.91032505e-01 -5.45858860e-01 9.87219572e-01 -1.28007531e+00 4.36667800e-01 -6.51807547e-01 -5.78600407e-01 4.32956755e-01 -9.42745447e-01 1.07881896e-01 8.80582690e-01 -9.38826740e-01 -1.18976128e+00 1.01890586e-01 4.71834093e-02 -7.71321237e-01 -2.51340091e-01 3.07424068e-01 -2.32793599e-01 -6.41521066e-02 4.98918056e-01 2.54742056e-01 -1.01833440e-01 -5.27619541e-01 4.88606066e-01 9.84507859e-01 6.23796105e-01 -9.68722224e-01 7.16265082e-01 1.18941046e-01 -4.80028838e-01 -3.12515587e-01 -1.73322940e+00 -4.97016042e-01 -5.59465051e-01 -1.59413099e-01 5.04184783e-01 -1.20721674e+00 -8.34537923e-01 4.35729563e-01 -9.20668244e-01 -6.42361760e-01 -2.85421133e-01 3.23271126e-01 -5.16517699e-01 1.36619225e-01 -6.10890269e-01 -6.63804829e-01 -2.34852299e-01 -1.27067983e+00 8.09655190e-01 6.11753881e-01 5.51035032e-02 -9.08337116e-01 -1.81398705e-01 8.36847723e-01 3.21686357e-01 -2.60380298e-01 7.35802174e-01 -1.07598817e+00 -4.59132254e-01 1.21241584e-01 -5.17532408e-01 3.04809451e-01 7.76110813e-02 -2.50143945e-01 -1.25744033e+00 -2.49990225e-01 -4.90637608e-02 -1.28171086e+00 8.30512226e-01 -2.66997278e-01 1.53091550e+00 -5.56401201e-02 -1.53912500e-01 6.21419132e-01 8.51022780e-01 -1.68192387e-02 1.25530079e-01 1.53475538e-01 1.08645117e+00 6.63518071e-01 8.82572055e-01 1.72333375e-01 7.55191445e-01 5.03357470e-01 4.31104600e-01 1.31259263e-01 -2.66746879e-01 -6.95708275e-01 5.42412877e-01 1.18375206e+00 2.02717349e-01 -3.23864780e-02 -7.01511502e-01 5.40585816e-01 -1.94591951e+00 -4.86419737e-01 -2.15531383e-02 1.87360275e+00 1.33596134e+00 -4.13237363e-02 -9.68965441e-02 -1.39317773e-02 6.71807706e-01 7.65251219e-02 -9.17677701e-01 -8.11326951e-02 1.32577553e-01 2.77033057e-02 2.28255853e-01 3.80896866e-01 -9.47932780e-01 1.27049327e+00 5.63647556e+00 9.69423294e-01 -1.13809836e+00 3.29645216e-01 7.24548757e-01 1.85247943e-01 -2.77186751e-01 -1.06283195e-01 -1.02318764e+00 4.52068716e-01 9.38748062e-01 -1.24110898e-03 4.16572124e-01 1.06303561e+00 1.01489186e-01 1.06417462e-01 -1.30737531e+00 1.04807496e+00 2.40372017e-01 -7.84178317e-01 -3.66406292e-02 -4.05821204e-01 9.15573061e-01 3.22675556e-02 -5.15397415e-02 8.46007109e-01 6.99275196e-01 -9.35966492e-01 5.02878666e-01 3.09224695e-01 9.30164039e-01 -5.54355443e-01 7.02279627e-01 7.42268085e-01 -7.91637778e-01 -1.59378499e-01 -3.45488667e-01 8.10790882e-02 -1.56476766e-01 2.27195337e-01 -5.29869854e-01 1.78650960e-01 5.76132119e-01 1.00269592e+00 -5.01000762e-01 6.09566450e-01 -9.87682939e-01 1.28194523e+00 -2.29231969e-01 -1.64205655e-01 3.35744113e-01 -1.49951711e-01 1.97003156e-01 1.04069602e+00 -5.31207994e-02 4.28826153e-01 4.16243821e-01 9.47779477e-01 -6.42535865e-01 1.20421521e-01 -1.10415690e-01 6.08786680e-02 7.55575716e-01 1.26678813e+00 -1.23357199e-01 -4.21821296e-01 -4.31492060e-01 8.69408607e-01 9.85937655e-01 4.90519494e-01 -7.49360621e-01 -1.69533059e-01 1.29516959e-01 -2.05142453e-01 7.53633007e-02 9.46778432e-02 5.59691451e-02 -1.44707227e+00 -2.18996257e-01 -9.85981882e-01 4.10390407e-01 -9.51164782e-01 -1.22235358e+00 3.87271881e-01 -2.06362814e-01 -1.15626299e+00 -1.38241857e-01 -3.90354544e-01 -7.23018408e-01 6.23522997e-01 -2.03744340e+00 -1.17231214e+00 -4.39775139e-01 6.88551724e-01 7.66382456e-01 -2.51844138e-01 7.21281826e-01 2.06451148e-01 -1.00586164e+00 1.09763610e+00 -3.01386695e-02 2.92120904e-01 9.54796612e-01 -1.33083200e+00 -1.99797377e-01 4.35574353e-01 3.14347953e-01 4.71758366e-01 3.99606287e-01 -1.50355756e-01 -1.30425191e+00 -1.17089844e+00 7.79659867e-01 -2.92361557e-01 7.88638532e-01 -1.21137336e-01 -1.13173854e+00 8.83025527e-01 5.42296916e-02 3.17744732e-01 7.80384004e-01 3.84361781e-02 -6.42299354e-01 -1.51676133e-01 -1.08401442e+00 5.25618315e-01 1.00461233e+00 -5.00222981e-01 -8.31398964e-01 4.08114821e-01 9.75222588e-01 -3.12521458e-01 -7.61144400e-01 3.82257670e-01 1.96149021e-01 -6.92587852e-01 5.95502913e-01 -6.60553992e-01 4.98048455e-01 -1.55439377e-01 6.91079348e-02 -1.33982921e+00 -1.83123171e-01 -5.89821577e-01 -3.18215340e-01 1.63622117e+00 3.46595347e-01 -5.97652555e-01 8.53392839e-01 7.19673634e-01 -4.05755453e-02 -9.91979659e-01 -6.25710368e-01 -5.85276067e-01 1.34623706e-01 -2.92235523e-01 4.67166722e-01 1.31976628e+00 -5.33748344e-02 5.93840361e-01 -3.95444781e-01 -4.12977375e-02 6.83194458e-01 3.80196363e-01 7.81450510e-01 -1.11459994e+00 -2.32434884e-01 -2.45963678e-01 2.24142060e-01 -1.75243890e+00 7.58938015e-01 -1.04121864e+00 1.66034520e-01 -8.35065603e-01 4.03267622e-01 -7.69631147e-01 -5.82033038e-01 8.52430463e-01 -5.34935534e-01 1.26460403e-01 1.65026821e-02 4.72847432e-01 -1.17820787e+00 1.00931287e+00 1.43593085e+00 -2.68994063e-01 -1.23936214e-01 2.30590980e-02 -8.83351982e-01 1.02209067e+00 6.99411333e-01 -7.38802552e-01 -4.67375755e-01 -6.71540082e-01 -1.68876171e-01 -1.64956227e-02 1.11849956e-01 -4.38315541e-01 3.85640979e-01 -2.69630760e-01 -8.19438398e-02 -3.67481261e-01 4.92340885e-02 -8.86323392e-01 -6.70210958e-01 6.11201040e-02 -7.36209512e-01 -3.84348333e-01 -2.89786398e-01 6.23032928e-01 -2.87973940e-01 -4.42966461e-01 1.01775110e+00 -3.67610157e-02 -6.31849766e-01 2.79027879e-01 -4.77153882e-02 3.83032441e-01 9.21583772e-01 2.18986347e-01 -4.13499415e-01 -4.63822424e-01 -5.60990155e-01 1.02684522e+00 2.44488969e-01 2.94183701e-01 5.68014145e-01 -1.28610289e+00 -6.64794207e-01 4.30117808e-02 3.17343384e-01 3.51760030e-01 4.52221520e-02 7.88313627e-01 5.87036908e-02 3.22774351e-01 1.73510641e-01 -6.90311372e-01 -1.09765112e+00 6.37189806e-01 3.95213097e-01 -3.01679581e-01 -3.30232471e-01 1.17685485e+00 1.97712496e-01 -8.68274808e-01 5.12217045e-01 -2.12633535e-02 -5.13867557e-01 1.78958867e-02 5.43187022e-01 2.15976194e-01 -4.63255405e-01 -6.61745369e-01 -2.07143910e-02 6.41057849e-01 -3.83715898e-01 2.52395004e-01 1.23296821e+00 -1.38203025e-01 5.94692640e-02 6.23523414e-01 1.15729082e+00 -9.19767246e-02 -1.57040071e+00 -8.58473182e-01 9.68057737e-02 -1.46424323e-01 6.54620975e-02 -7.21299589e-01 -1.29113412e+00 1.09742200e+00 2.72569597e-01 -1.68119118e-01 1.07672453e+00 1.12383068e-01 7.46626973e-01 7.48435497e-01 4.59060609e-01 -9.33158875e-01 3.82333338e-01 4.93947655e-01 6.00511193e-01 -1.59853697e+00 -1.58537149e-01 -5.14760673e-01 -7.80353546e-01 7.93420136e-01 1.12830055e+00 -6.87459037e-02 5.27600288e-01 -5.92969693e-02 2.77205527e-01 1.72088221e-02 -9.72421408e-01 -4.78028059e-02 5.08506820e-02 2.42701545e-01 5.64443827e-01 5.01308590e-02 -1.34417206e-01 1.12616599e+00 1.24661036e-01 -4.14348058e-02 4.82652605e-01 6.17625415e-01 -4.38488007e-01 -1.12854075e+00 -2.06507325e-01 3.40717942e-01 -3.36093664e-01 -1.20074479e-02 -3.13630491e-01 4.02696669e-01 -9.74538848e-02 1.06875932e+00 -2.33910948e-01 -2.58818001e-01 3.77479166e-01 1.53040230e-01 1.47744820e-01 -8.02336931e-01 -6.44159317e-01 -1.16165141e-02 -7.29308575e-02 -6.27657771e-01 -6.88170910e-01 -4.70244050e-01 -1.52855599e+00 2.23206595e-01 -6.77405775e-01 4.55587417e-01 1.35654360e-01 1.08495438e+00 1.93260834e-01 5.42676449e-01 9.84996140e-01 -5.52276671e-01 -1.12758398e+00 -1.14067888e+00 -3.85297567e-01 5.45633256e-01 2.52274960e-01 -8.30297649e-01 -6.54570818e-01 1.39542714e-01]
[9.563163757324219, 3.3704612255096436]
9cdc4038-f846-41ac-8d9a-74fa7caaccaa
technical-report-temporal-aggregate
2106.03152
null
https://arxiv.org/abs/2106.03152v2
https://arxiv.org/pdf/2106.03152v2.pdf
Technical Report: Temporal Aggregate Representations
This technical report extends our work presented in [9] with more experiments. In [9], we tackle long-term video understanding, which requires reasoning from current and past or future observations and raises several fundamental questions. How should temporal or sequential relationships be modelled? What temporal extent of information and context needs to be processed? At what temporal scale should they be derived? [9] addresses these questions with a flexible multi-granular temporal aggregation framework. In this report, we conduct further experiments with this framework on different tasks and a new dataset, EPIC-KITCHENS-100.
['Angela Yao', 'Dibyadip Chatterjee', 'Fadime Sener']
2021-06-06
null
null
null
null
['action-anticipation']
['computer-vision']
[ 1.31402820e-01 -7.22194016e-02 -4.37391520e-01 -4.81902063e-01 -2.08054036e-01 -5.67000508e-01 6.49233043e-01 1.98914960e-01 -4.87571329e-01 7.75866568e-01 4.21878695e-01 -2.30538800e-01 -5.87246954e-01 -5.37815213e-01 -4.18721706e-01 -2.50343263e-01 -7.50202537e-01 8.88586044e-04 7.74966955e-01 -2.11588979e-01 2.19811454e-01 2.13878676e-01 -1.75429046e+00 8.40601146e-01 1.84616432e-01 1.16083384e+00 1.34499401e-01 1.08950341e+00 1.97665244e-01 1.40712392e+00 -4.09366876e-01 -1.01055898e-01 1.36522204e-01 -3.73786658e-01 -1.05245829e+00 5.47061145e-01 2.22575590e-01 -5.22697628e-01 -2.85970211e-01 5.13955891e-01 -7.75376558e-02 4.17361230e-01 3.70020300e-01 -1.31371808e+00 -3.28526944e-01 5.58520019e-01 -3.75076413e-01 9.43110108e-01 8.67790103e-01 1.78315848e-01 9.02919054e-01 -3.49923462e-01 7.77483761e-01 1.16351902e+00 6.70611978e-01 3.70832175e-01 -9.57794011e-01 -2.38776669e-01 9.25180912e-01 7.75257647e-01 -1.13543570e+00 -3.37335110e-01 7.19484210e-01 -4.65527862e-01 1.25263011e+00 3.00949752e-01 9.45541263e-01 1.30493164e+00 1.68502167e-01 1.01631343e+00 9.96131897e-01 -3.84038895e-01 2.62911737e-01 -2.64161974e-01 3.68425936e-01 2.85427779e-01 -2.55109191e-01 3.09333742e-01 -1.04001760e+00 -7.42242336e-02 7.73856401e-01 -1.10078551e-01 -4.58139777e-02 3.90079468e-02 -1.42695081e+00 4.46581423e-01 -2.68207520e-01 4.99259651e-01 -5.10963678e-01 2.92211384e-01 5.97963154e-01 6.37044072e-01 5.10953605e-01 -7.22507983e-02 -5.28326213e-01 -5.28809309e-01 -8.90942276e-01 3.56252283e-01 9.01248455e-01 1.02281404e+00 3.30948383e-01 -2.62347460e-01 -3.06466855e-02 2.94404000e-01 2.19015494e-01 -3.25139314e-02 1.42871246e-01 -1.45632637e+00 6.33132696e-01 -4.06159610e-02 4.02006119e-01 -8.86530221e-01 -4.20228839e-01 -3.66039500e-02 -3.54574323e-01 -8.96128342e-02 4.71993506e-01 -3.10447574e-01 -5.88948071e-01 1.75450790e+00 1.97094157e-01 8.15566123e-01 2.43515149e-01 7.88836539e-01 4.86833602e-01 6.37364388e-01 3.24148417e-01 -6.22282088e-01 1.26733243e+00 -1.01506948e+00 -1.16740811e+00 -4.01125342e-01 4.74826574e-01 -5.78553140e-01 6.04552567e-01 5.21134377e-01 -1.13195765e+00 -7.20750690e-01 -1.06718647e+00 8.78257453e-02 -4.41500098e-01 -9.23365057e-02 8.07164788e-01 3.02294284e-01 -1.42131114e+00 6.82637453e-01 -1.09988379e+00 -8.81348968e-01 -2.83074975e-01 1.69239879e-01 -8.80394354e-02 2.44900212e-01 -1.52755964e+00 6.97287083e-01 6.68198764e-01 1.52758747e-01 -6.17748082e-01 -4.45817918e-01 -7.39637494e-01 -1.71323448e-01 1.00140405e+00 -9.53983009e-01 1.43297827e+00 -9.46707606e-01 -1.37694967e+00 5.68435311e-01 -3.28115255e-01 -8.18005085e-01 3.65733385e-01 -5.04940152e-01 -6.95971310e-01 4.22179013e-01 -1.74158901e-01 3.16473186e-01 4.58293617e-01 -9.44529116e-01 -1.08137500e+00 -3.91971260e-01 7.37177253e-01 2.89580137e-01 -5.26374392e-02 3.50116283e-01 -6.30793095e-01 -9.71760929e-01 -1.46123823e-02 -1.12381792e+00 -2.85471976e-01 -1.64549246e-01 2.54281551e-01 -1.91226259e-01 8.88349354e-01 -5.73737025e-01 1.67640638e+00 -1.99507558e+00 3.15770864e-01 -9.94939432e-02 -4.03125919e-02 -1.29794143e-02 5.11657037e-02 6.61796093e-01 -1.02943674e-01 1.86291188e-01 7.20193684e-02 -1.86422795e-01 -1.51409864e-01 4.74860698e-01 -4.97070879e-01 2.57621519e-02 -9.64175165e-02 8.30044031e-01 -1.06087017e+00 -6.46391749e-01 4.44683969e-01 2.20055014e-01 -2.48736084e-01 1.80830732e-01 -5.01627326e-01 3.83264363e-01 -6.56535566e-01 5.80901086e-01 1.66639611e-01 -2.96899557e-01 3.54864657e-01 -9.33810696e-02 -9.49503332e-02 -1.24532603e-01 -1.17739582e+00 1.79855132e+00 -2.89984435e-01 7.07266271e-01 -9.92271826e-02 -9.68086421e-01 4.19753909e-01 6.80800498e-01 6.43682897e-01 -7.34385073e-01 -2.07012430e-01 -5.21821439e-01 -3.54536802e-01 -9.32048082e-01 5.42902887e-01 -4.57153618e-02 -2.06817925e-01 5.41196525e-01 -1.46586642e-01 2.18375936e-01 6.61305547e-01 3.65718544e-01 1.26796079e+00 5.54218709e-01 4.49865431e-01 -3.09005558e-01 6.97075248e-01 7.35096559e-02 6.73638940e-01 8.14436853e-01 -4.23604697e-01 4.16556925e-01 6.61012590e-01 -7.86795020e-01 -4.96612877e-01 -1.15086377e+00 2.79789507e-01 1.37996852e+00 3.36619943e-01 -8.87228727e-01 -3.77276689e-01 -6.29403710e-01 -4.40262944e-01 5.89508593e-01 -9.55161572e-01 3.33233476e-01 -6.77204490e-01 -5.49890697e-01 3.78717154e-01 9.11656201e-01 2.87155211e-01 -1.08565080e+00 -1.09054554e+00 4.97130454e-01 -7.93916464e-01 -1.53903091e+00 -3.42489928e-01 -8.77437741e-02 -1.12251234e+00 -1.12127614e+00 -2.74644822e-01 -2.13035166e-01 6.78547919e-02 2.23768532e-01 1.32208991e+00 3.24786804e-03 5.95917515e-02 9.86233950e-01 -7.32339799e-01 -4.82122958e-01 -1.19846269e-01 -1.51589274e-01 -9.16581079e-02 1.02299444e-01 3.99829388e-01 -7.52957642e-01 -6.04342639e-01 5.05302906e-01 -9.51567769e-01 3.05855453e-01 3.44233572e-01 5.00338972e-01 6.56770587e-01 2.59503603e-01 4.45565581e-01 -8.05687547e-01 4.17905271e-01 -5.99265754e-01 -4.76330310e-01 5.58030069e-01 -2.41465509e-01 9.26900059e-02 4.04666960e-01 -5.08927345e-01 -1.57768142e+00 -4.83620502e-02 -1.95869040e-02 -3.93960774e-01 -3.28063071e-01 7.19098032e-01 1.61252767e-01 7.29102969e-01 3.37473959e-01 9.32877958e-02 -4.11332965e-01 -3.25684756e-01 2.61458993e-01 6.97109327e-02 5.34337759e-01 -7.89919019e-01 3.43468428e-01 9.66387093e-01 -1.19499303e-01 -6.55107021e-01 -1.14296877e+00 -6.81943774e-01 -1.12272346e+00 -6.12882316e-01 1.04692888e+00 -8.92657757e-01 -6.39937758e-01 1.86286151e-01 -1.18129301e+00 -7.15303898e-01 -3.23553532e-01 4.09901917e-01 -9.93626833e-01 3.38532925e-01 -6.57291830e-01 -1.03053343e+00 1.33048877e-01 -7.03251898e-01 9.50846672e-01 1.55311776e-03 -5.95705152e-01 -1.41507077e+00 1.18684702e-01 3.26653779e-01 1.71121195e-01 5.51379442e-01 3.60285193e-01 -1.30442291e-01 -7.08972752e-01 1.84264973e-01 2.44264156e-02 -4.87474427e-02 2.46022809e-02 1.04932718e-01 -6.78045154e-01 -1.68470517e-01 1.99540332e-01 -2.05203772e-01 9.78868902e-01 3.59335750e-01 1.19353878e+00 -2.56761074e-01 -3.30306321e-01 3.39420080e-01 1.21766984e+00 6.47348464e-01 5.77037752e-01 4.49721038e-01 2.03753978e-01 8.64522278e-01 1.23453772e+00 5.55805802e-01 6.85531616e-01 7.64941990e-01 2.64043629e-01 5.40345907e-01 3.08985375e-02 -9.02669579e-02 5.13150036e-01 7.77417243e-01 -7.19861269e-01 -6.44127548e-01 -7.36051261e-01 8.73787522e-01 -2.61807680e+00 -1.49751770e+00 8.85326136e-03 1.63775766e+00 3.88184994e-01 2.76903093e-01 1.73771024e-01 1.71977699e-01 3.22776645e-01 7.62418628e-01 -3.48522723e-01 -3.18587512e-01 -2.52721142e-02 -3.59516948e-01 8.54687169e-02 5.16029894e-01 -1.25558603e+00 6.94961488e-01 7.38690042e+00 6.11259580e-01 -8.09763551e-01 1.13830209e-01 6.23496056e-01 -4.35397118e-01 6.49600849e-03 4.09112543e-01 -6.48484588e-01 4.06976253e-01 1.36517131e+00 -2.81837851e-01 2.63070017e-01 2.79176056e-01 4.59735572e-01 -6.24727547e-01 -1.55864179e+00 9.44732904e-01 1.05070338e-01 -1.35259664e+00 -1.97192639e-01 -9.73271430e-02 4.43466753e-01 -8.33860114e-02 -3.32212262e-02 1.67944893e-01 4.10255820e-01 -7.28454709e-01 9.45242643e-01 1.05703795e+00 2.29916632e-01 -4.04253274e-01 5.54312766e-01 5.60616314e-01 -1.74869454e+00 -3.11318338e-01 2.25432336e-01 -4.84702706e-01 7.66339719e-01 3.45389426e-01 -2.24662513e-01 9.52103198e-01 1.08940136e+00 1.10510349e+00 -4.81474042e-01 4.31484908e-01 -2.71124631e-01 6.06396139e-01 -4.64333177e-01 2.81395882e-01 3.17314267e-01 1.78251751e-02 5.83485961e-01 1.42421281e+00 3.45074683e-01 6.35083497e-01 3.35638225e-01 2.51690626e-01 5.56266069e-01 -4.04213458e-01 -4.82831359e-01 2.55225271e-01 1.92508027e-01 7.60613978e-01 -9.05655444e-01 -6.27430141e-01 -6.46834970e-01 7.95197725e-01 1.19634643e-01 6.51381314e-01 -1.09188461e+00 2.68653810e-01 4.51771021e-01 3.07995975e-01 3.25830072e-01 -3.19163352e-01 1.65678844e-01 -1.34557486e+00 2.74069279e-01 -4.46661294e-01 1.30103528e+00 -9.95106280e-01 -1.05113900e+00 5.90134025e-01 7.35937834e-01 -1.36162436e+00 -7.05935657e-01 -1.79059356e-01 -4.97255355e-01 2.77094752e-01 -1.38208616e+00 -1.15634334e+00 -1.77700713e-01 6.70329690e-01 1.03094721e+00 2.47041747e-01 4.36793953e-01 6.99013770e-02 -2.03823075e-01 -1.57731712e-01 -4.34341878e-01 -1.08285576e-01 6.04079545e-01 -1.24242866e+00 3.95888329e-01 9.20330644e-01 8.68332162e-02 4.37387735e-01 9.85194087e-01 -6.71551764e-01 -1.22663653e+00 -8.74162078e-01 8.72009158e-01 -7.04076588e-01 1.08563411e+00 -8.76912102e-02 -8.85447383e-01 1.19134402e+00 4.90774423e-01 1.91470817e-01 6.28335953e-01 1.87343433e-01 -2.98172474e-01 -2.03296505e-02 -7.81630456e-01 5.16981661e-01 1.59871423e+00 -5.16787291e-01 -8.35271835e-01 1.60313159e-01 9.60600376e-01 -3.42641056e-01 -1.01550162e+00 6.48111761e-01 8.14709187e-01 -1.39836073e+00 8.91886473e-01 -4.53743637e-01 1.93309218e-01 -2.98336565e-01 -1.92255616e-01 -9.57100987e-01 -9.38968733e-02 -6.83260858e-01 -7.47348547e-01 1.00681949e+00 1.49686664e-01 -2.55252033e-01 7.16424227e-01 7.07095981e-01 2.73975134e-01 -6.91559970e-01 -8.43687057e-01 -1.05509818e+00 -2.98721462e-01 -1.14475799e+00 4.40586567e-01 7.44253337e-01 1.73297942e-01 5.23420423e-03 -6.23803794e-01 2.05746263e-01 4.20976132e-01 2.18319938e-01 5.90594232e-01 -9.49731529e-01 -3.54952335e-01 -1.58224568e-01 -3.88419628e-01 -1.27551603e+00 1.93497185e-02 -1.10407390e-01 -3.16828310e-01 -1.46018171e+00 -9.97967795e-02 1.15869418e-01 -6.00453615e-01 3.42278302e-01 1.47454441e-01 -1.39394313e-01 3.53377461e-01 2.20214054e-02 -1.44204819e+00 2.97687203e-01 8.33001256e-01 3.60246859e-02 -1.37939095e-01 5.93686104e-02 -2.15320751e-01 9.48990226e-01 4.39333916e-01 -9.12706852e-02 -8.63790154e-01 -4.79365051e-01 4.70786452e-01 9.20622230e-01 3.62749428e-01 -1.08893657e+00 6.06819451e-01 -5.83543241e-01 2.78295372e-02 -1.06889951e+00 6.10796511e-01 -1.05022252e+00 5.64631343e-01 1.26382872e-01 -4.53963965e-01 4.12827075e-01 4.00302082e-01 1.03288579e+00 -4.83534873e-01 2.42721692e-01 3.62422466e-01 -2.52209514e-01 -1.47231221e+00 2.01261505e-01 -8.29446793e-01 -1.65853649e-01 1.50314605e+00 -2.97201693e-01 -1.12916782e-01 -6.42934620e-01 -1.59983349e+00 5.68540394e-01 1.66763559e-01 8.25784802e-01 6.33260548e-01 -1.31174099e+00 -5.32519519e-01 -3.57792407e-01 2.30584309e-01 -4.01082397e-01 7.57375598e-01 8.83644044e-01 -1.37098745e-01 5.32662332e-01 -1.46291768e-02 -7.09164381e-01 -1.48692977e+00 9.68030870e-01 1.04617789e-01 -6.34429574e-01 -6.87766552e-01 2.80605406e-01 1.81039125e-01 2.95738518e-01 1.95657238e-01 -5.58011472e-01 -5.51297426e-01 3.96287262e-01 7.52801716e-01 4.77620006e-01 -3.49830955e-01 -7.32536495e-01 -5.28562725e-01 5.75004399e-01 3.22580449e-02 -5.40149391e-01 1.23160946e+00 -7.89518476e-01 1.57150418e-01 1.05751431e+00 7.79453576e-01 -4.20529932e-01 -1.63559151e+00 -4.54293668e-01 5.07686734e-01 -4.39127952e-01 -6.81488141e-02 -7.53687263e-01 -9.51570392e-01 4.73547786e-01 4.30865437e-01 5.57397127e-01 1.47373414e+00 2.47495160e-01 6.74736679e-01 3.18796664e-01 6.61792696e-01 -1.34437060e+00 1.60951495e-01 6.89842820e-01 9.81672645e-01 -1.17349863e+00 1.32621184e-01 -4.70123231e-01 -7.26841033e-01 1.00461328e+00 6.57199919e-01 2.15176463e-01 9.66692567e-01 4.91471350e-01 -1.75478999e-02 -4.06606138e-01 -1.65577710e+00 -4.92784888e-01 2.71815717e-01 4.37229574e-01 1.53864026e-01 -1.56278372e-01 -1.68818906e-01 6.06372297e-01 1.55290812e-01 4.85121429e-01 7.08370864e-01 1.34638429e+00 -1.33758724e-01 -7.92237461e-01 -4.18438315e-01 1.04857944e-01 -3.89724791e-01 3.22828650e-01 -2.55251110e-01 7.90035486e-01 2.74014115e-01 1.39220929e+00 2.14282349e-01 -4.30909038e-01 3.88001323e-01 -5.41833751e-02 4.68866378e-01 -5.13655543e-01 -3.26568544e-01 2.35431120e-01 3.75571191e-01 -9.89680588e-01 -1.24270535e+00 -1.12299240e+00 -1.12660444e+00 -1.25314191e-01 1.11704938e-01 1.21233791e-01 1.34706169e-01 1.26887381e+00 2.17838228e-01 7.40903318e-01 1.24377847e-01 -5.22198021e-01 2.04357371e-01 -7.24852979e-01 -3.41249019e-01 3.41703385e-01 3.87431920e-01 -6.18684888e-01 -2.58936346e-01 6.59942746e-01]
[8.408903121948242, 0.5090969204902649]
8015cfdc-1926-4cb3-a9b2-0964eb55ccd5
doublemix-simple-interpolation-based-data
2209.05297
null
https://arxiv.org/abs/2209.05297v1
https://arxiv.org/pdf/2209.05297v1.pdf
DoubleMix: Simple Interpolation-Based Data Augmentation for Text Classification
This paper proposes a simple yet effective interpolation-based data augmentation approach termed DoubleMix, to improve the robustness of models in text classification. DoubleMix first leverages a couple of simple augmentation operations to generate several perturbed samples for each training data, and then uses the perturbed data and original data to carry out a two-step interpolation in the hidden space of neural models. Concretely, it first mixes up the perturbed data to a synthetic sample and then mixes up the original data and the synthetic perturbed data. DoubleMix enhances models' robustness by learning the "shifted" features in hidden space. On six text classification benchmark datasets, our approach outperforms several popular text augmentation methods including token-level, sentence-level, and hidden-level data augmentation techniques. Also, experiments in low-resource settings show our approach consistently improves models' performance when the training data is scarce. Extensive ablation studies and case studies confirm that each component of our approach contributes to the final performance and show that our approach exhibits superior performance on challenging counterexamples. Additionally, visual analysis shows that text features generated by our approach are highly interpretable. Our code for this paper can be found at https://github.com/declare-lab/DoubleMix.git.
['Soujanya Poria', 'Diyi Yang', 'Wei Han', 'Hui Chen']
2022-09-12
null
https://aclanthology.org/2022.coling-1.409
https://aclanthology.org/2022.coling-1.409.pdf
coling-2022-10
['text-augmentation']
['natural-language-processing']
[ 2.60092437e-01 5.89272380e-02 -3.09881121e-01 -3.54935855e-01 -1.03572738e+00 -3.07275534e-01 7.82672405e-01 3.68651420e-01 -4.59862500e-01 6.57978833e-01 3.89270514e-01 -4.59060848e-01 5.42659998e-01 -5.18522263e-01 -6.61062479e-01 -5.29253900e-01 2.32116878e-01 2.68115461e-01 -2.49876425e-01 -2.40238935e-01 1.44745097e-01 -1.61689473e-03 -1.25395036e+00 6.14953458e-01 9.82432425e-01 7.89520741e-01 -2.20240548e-01 7.29973733e-01 -2.51300097e-01 7.01573431e-01 -7.53289700e-01 -4.09439981e-01 2.95165658e-01 -3.88295829e-01 -8.76082182e-01 -7.01955240e-03 3.29885274e-01 -3.42305750e-01 -3.10361445e-01 6.92359865e-01 5.88478863e-01 2.06449449e-01 3.99172664e-01 -1.53113294e+00 -7.99677312e-01 9.10689235e-01 -6.06260657e-01 -1.57326125e-02 2.62958795e-01 2.73601800e-01 9.54398692e-01 -1.23552990e+00 2.64221221e-01 1.10272002e+00 9.64855790e-01 6.04523599e-01 -1.50684285e+00 -9.07694936e-01 2.95693398e-01 -2.29421556e-01 -1.09330630e+00 -5.18735707e-01 7.00350642e-01 -3.36768329e-01 8.13069999e-01 3.63536686e-01 3.82703513e-01 1.47080696e+00 -9.46759060e-02 1.21040475e+00 1.06647205e+00 -7.36502707e-01 1.56305954e-01 2.67374545e-01 5.07304251e-01 5.60252249e-01 2.22722933e-01 -5.41151799e-02 -5.49875081e-01 -3.36650938e-01 3.10449660e-01 7.00830743e-02 -2.31643274e-01 2.05413043e-01 -1.28071034e+00 8.56642008e-01 3.15360725e-01 3.03398222e-01 -2.33624861e-01 5.15333936e-02 5.24361312e-01 2.68708825e-01 8.43461514e-01 4.64197934e-01 -5.17456293e-01 -1.48032993e-01 -1.08435655e+00 3.44226927e-01 6.72852874e-01 9.46814179e-01 5.49130499e-01 2.66976744e-01 -5.31801939e-01 1.00724459e+00 1.31548420e-01 1.03024982e-01 8.54607821e-01 -5.40488541e-01 1.00630713e+00 8.09543490e-01 7.24826753e-03 -6.81222200e-01 -4.22006011e-01 -4.27737981e-01 -9.90847707e-01 3.96086983e-02 4.99810785e-01 -2.77175099e-01 -1.04854190e+00 1.68341982e+00 2.30928734e-01 2.87331603e-02 8.10626596e-02 3.48050863e-01 8.81704748e-01 7.19604254e-01 2.23043069e-01 8.65219161e-02 1.16747832e+00 -1.27453494e+00 -7.53390193e-01 -4.14647162e-01 1.08017635e+00 -6.74882531e-01 1.58713996e+00 1.66459784e-01 -1.20256460e+00 -4.44363117e-01 -9.69365597e-01 -2.52355278e-01 -5.28962016e-01 4.22186136e-01 2.67846763e-01 6.63079023e-01 -6.56778514e-01 5.78417897e-01 -7.17975676e-01 -4.95968089e-02 7.54699945e-01 2.04000801e-01 -3.95167083e-01 3.05362418e-02 -1.24065197e+00 6.71724558e-01 5.02871931e-01 6.22412376e-02 -4.42550838e-01 -9.03089583e-01 -9.05160964e-01 -5.14631458e-02 1.34684801e-01 -5.18688500e-01 1.38063467e+00 -7.83249557e-01 -1.24462271e+00 6.26067042e-01 -4.14868027e-01 -4.90777671e-01 8.92903507e-01 -3.82977813e-01 -2.10406661e-01 -2.53385454e-01 9.79280695e-02 7.65410721e-01 9.61922467e-01 -1.37945223e+00 -4.32979196e-01 -8.28665346e-02 -2.81496942e-01 1.25590786e-01 -8.27202737e-01 -1.91707283e-01 -5.03480375e-01 -1.17952514e+00 -1.06380992e-01 -8.55059266e-01 -2.79281974e-01 -1.71720132e-01 -9.25404191e-01 -1.82174817e-02 8.46601486e-01 -8.22104514e-01 1.52635896e+00 -2.14541793e+00 -1.57386601e-01 7.91034102e-02 3.87258381e-01 3.62470001e-01 -1.88990653e-01 4.04628009e-01 -4.32064861e-01 5.90314806e-01 -5.47679722e-01 -1.09422672e+00 1.12734854e-01 -1.19584240e-01 -4.12141263e-01 1.74877480e-01 3.77279997e-01 1.17376304e+00 -6.11981034e-01 -4.23874140e-01 5.20193344e-03 5.16417027e-01 -5.67184687e-01 9.12791118e-03 -2.22080886e-01 4.25822407e-01 -2.66357630e-01 4.25156951e-01 6.15189910e-01 -4.70261514e-01 -1.14824824e-01 1.11383185e-01 2.09382549e-01 4.17784989e-01 -1.00946820e+00 1.35146403e+00 -4.88839686e-01 7.45387852e-01 -2.92809665e-01 -7.25087225e-01 7.98807263e-01 4.14351702e-01 1.85854942e-01 -4.66608614e-01 2.15392530e-01 -8.78433660e-02 -3.06608140e-01 -3.19177181e-01 7.74891794e-01 5.92301823e-02 -5.29765934e-02 9.30006742e-01 -2.38005206e-01 1.27526343e-01 7.49364346e-02 4.34527367e-01 9.00933921e-01 8.92671645e-02 2.19044894e-01 1.89827457e-01 2.31653243e-01 -1.47869738e-04 2.66983032e-01 8.41242731e-01 -1.06599562e-01 6.73233449e-01 5.77938974e-01 -2.66860247e-01 -1.38622928e+00 -6.06962800e-01 -1.77642718e-01 1.22037029e+00 -5.67628503e-01 -6.21570349e-01 -9.07515526e-01 -9.65636134e-01 8.96125138e-02 9.93820310e-01 -1.16725349e+00 -3.20852041e-01 -3.90623748e-01 -1.03073025e+00 8.62095237e-01 7.81990886e-01 6.04288995e-01 -9.90636647e-01 -6.59357980e-02 1.75923780e-02 -4.83423829e-01 -9.61279809e-01 -6.10444546e-01 1.73574597e-01 -9.54401135e-01 -7.40273356e-01 -5.61267614e-01 -7.10624874e-01 9.48343337e-01 2.20632598e-01 8.38870168e-01 5.37463546e-01 -3.69433500e-02 -1.29476056e-01 -5.25175631e-01 -4.26234484e-01 -7.86114097e-01 3.39105070e-01 -7.39981160e-02 3.28214057e-02 3.13529789e-01 -1.78608820e-01 -3.14027190e-01 1.60456672e-01 -1.08830059e+00 4.23957258e-01 3.77411097e-01 1.22114015e+00 2.94093877e-01 -7.89470971e-02 5.67906916e-01 -1.08126104e+00 9.58387613e-01 -6.38682842e-01 -1.75577179e-01 -3.71935806e-04 -7.55787551e-01 8.81778672e-02 8.45905721e-01 -6.27805471e-01 -8.51399720e-01 -4.43704203e-02 -1.65883631e-01 -2.35895708e-01 -1.78199321e-01 5.73489785e-01 -3.30125988e-02 4.51979578e-01 9.87966955e-01 2.48869151e-01 1.49144337e-01 -5.82059383e-01 4.28728133e-01 8.61443579e-01 4.05568808e-01 -4.59175289e-01 1.07171226e+00 4.09601599e-01 -4.94640172e-01 -5.82818985e-01 -8.14558029e-01 -7.22116008e-02 -6.90528572e-01 2.12707072e-01 4.26568747e-01 -8.29496682e-01 -3.62990499e-01 6.04805231e-01 -9.04795945e-01 -7.35620439e-01 -4.18759555e-01 2.73171604e-01 -1.37659475e-01 3.42404515e-01 -7.03272402e-01 -7.29828477e-01 -6.02035999e-01 -9.75832582e-01 9.69811082e-01 -1.15458794e-01 -4.47960198e-01 -1.23831153e+00 -1.67315081e-01 4.72780675e-01 3.64478976e-01 1.47858560e-01 9.78279293e-01 -1.04503691e+00 -2.54577454e-02 -6.47095323e-01 -7.58684054e-03 4.27001804e-01 1.33200243e-01 1.31352007e-01 -1.24694943e+00 -3.80110741e-01 -2.96365708e-01 -4.33744520e-01 9.96699750e-01 2.91365027e-01 1.41698039e+00 -6.24929488e-01 -2.65020967e-01 5.25158823e-01 7.73172677e-01 -2.53138065e-01 5.21671474e-01 5.30839980e-01 8.80826354e-01 5.38560569e-01 5.76306283e-01 5.33922195e-01 3.65022749e-01 5.05776465e-01 1.77312806e-01 -5.25716722e-01 6.76624551e-02 -4.06024724e-01 3.17017198e-01 7.55343199e-01 2.33306184e-01 -2.79405355e-01 -1.04796124e+00 3.83563012e-01 -1.78748655e+00 -9.34875369e-01 -2.91462570e-01 2.16594982e+00 1.16580367e+00 3.02745521e-01 2.47322336e-01 6.60571218e-01 7.25044787e-01 -2.81190816e-02 -5.13617218e-01 -1.92364022e-01 -1.16610132e-01 1.36660561e-01 3.12391400e-01 5.84333301e-01 -1.17023408e+00 9.73664403e-01 6.47788143e+00 7.94677794e-01 -9.91255462e-01 1.99376538e-01 9.45126295e-01 -4.46352243e-01 -3.59543860e-01 -2.10439488e-01 -8.89705122e-01 6.65478468e-01 9.58469868e-01 -2.47400180e-01 1.97472274e-01 7.42054939e-01 3.69002879e-01 1.51407957e-01 -1.10857034e+00 5.86391032e-01 -5.85078038e-02 -1.51185930e+00 2.29317233e-01 3.75338248e-03 7.64388859e-01 -1.10432237e-01 4.10987318e-01 5.37648022e-01 4.45004582e-01 -1.17597032e+00 7.03584850e-01 2.59197563e-01 8.13220382e-01 -6.38785064e-01 7.30260789e-01 3.91626805e-01 -8.83523941e-01 -1.47805661e-01 8.32951292e-02 -5.12908213e-02 -1.09641835e-01 5.80639184e-01 -1.26120603e+00 3.41454268e-01 5.98252714e-01 6.19312823e-01 -1.06969595e+00 6.13704026e-01 -2.18952864e-01 1.02665961e+00 -1.88556522e-01 -3.43649415e-04 4.50313576e-02 2.51263440e-01 3.59136701e-01 1.46414065e+00 -7.40311071e-02 -2.00048253e-01 -2.06760392e-02 8.96878958e-01 -3.91495705e-01 2.24858090e-01 -5.07693410e-01 -6.48589209e-02 6.14611924e-01 1.22771740e+00 -4.16553169e-01 -6.46003783e-01 -2.71365792e-01 8.70880842e-01 2.93295443e-01 5.77756166e-01 -8.03220272e-01 -5.57838202e-01 3.52626413e-01 -8.94647315e-02 -4.62555774e-02 -2.57529609e-05 -8.11161876e-01 -1.21579897e+00 2.67522067e-01 -1.16637146e+00 4.58329201e-01 -7.03529656e-01 -1.10400462e+00 7.93795884e-01 -9.85518023e-02 -1.29902112e+00 -3.48645121e-01 -2.67604202e-01 -7.02180028e-01 1.10825264e+00 -1.20974278e+00 -1.23977387e+00 -5.01597345e-01 5.18416643e-01 6.69939160e-01 -2.96238482e-01 9.37143266e-01 9.70031545e-02 -1.09459913e+00 1.25211298e+00 3.43279064e-01 5.10040224e-01 6.61827385e-01 -1.24240613e+00 1.09625328e+00 8.70528817e-01 -6.66606352e-02 6.61771297e-01 4.62986559e-01 -6.79107189e-01 -9.16994214e-01 -1.32815945e+00 8.48407149e-01 -7.11895704e-01 5.96716464e-01 -6.99003935e-01 -1.39402843e+00 1.06006205e+00 1.33146718e-01 -4.30577947e-03 9.50809717e-01 7.59506226e-02 -5.14656007e-01 2.65220284e-01 -1.11480379e+00 9.19092715e-01 5.90544820e-01 -4.23848748e-01 -5.40463448e-01 2.99505591e-01 7.86233127e-01 -5.31979620e-01 -8.24632227e-01 1.70723379e-01 5.22049427e-01 -5.43449879e-01 7.07160294e-01 -1.02740073e+00 7.19403505e-01 2.32091248e-02 -4.99905571e-02 -1.48024166e+00 -2.03450352e-01 -7.80629098e-01 -2.62041181e-01 1.38000202e+00 8.21464598e-01 -9.21675265e-01 9.31911945e-01 7.31699646e-01 -8.23940113e-02 -9.08665478e-01 -6.53908193e-01 -5.63910067e-01 5.82417428e-01 -6.24967575e-01 6.53546214e-01 1.41972697e+00 1.40040264e-01 2.78837740e-01 -3.10561717e-01 -1.02816366e-01 3.80700707e-01 -2.87575036e-01 9.74538028e-01 -1.02580440e+00 -2.02137351e-01 -5.34254670e-01 1.34696320e-01 -8.21239710e-01 2.23023906e-01 -1.21642196e+00 -2.11974293e-01 -1.52294564e+00 8.21577981e-02 -7.23910391e-01 -5.39217070e-02 1.10008109e+00 -7.22713709e-01 4.41824913e-01 3.14610213e-01 3.37530553e-01 -1.00755140e-01 6.53082430e-01 1.11430788e+00 -1.64649144e-01 -5.30191958e-01 1.02151051e-01 -1.03182626e+00 5.06001174e-01 1.08472621e+00 -5.89564979e-01 -7.97377303e-02 -3.09351116e-01 1.25253409e-01 -4.83924270e-01 3.78587693e-01 -7.82716215e-01 2.13236809e-02 3.62010524e-02 7.34156251e-01 -4.56673652e-01 3.07507604e-01 -5.34780681e-01 -3.39158267e-01 4.98466581e-01 -8.17884386e-01 1.39158219e-01 6.29175723e-01 2.86974281e-01 1.59575380e-02 -2.50249684e-01 7.81811357e-01 1.98479682e-01 -6.37094816e-03 2.10352495e-01 -2.61223286e-01 2.09805861e-01 7.60759711e-01 -2.96662241e-01 -5.52511513e-01 -5.16408980e-01 -7.08403707e-01 3.70981783e-01 3.47444803e-01 5.55727482e-01 4.09848988e-01 -1.37297058e+00 -9.93652940e-01 4.10776407e-01 1.68069392e-01 1.07082382e-01 1.00111738e-02 9.14935589e-01 -1.65638506e-01 2.52182573e-01 1.23435244e-01 -4.95888591e-01 -1.27653313e+00 4.26365405e-01 2.26671889e-01 -3.31287354e-01 -6.91927314e-01 7.53172159e-01 -1.00972705e-01 -6.96991503e-01 4.41773057e-01 -5.04170299e-01 1.14944000e-02 9.06356201e-02 6.21451557e-01 3.58404130e-01 2.05540314e-01 -4.66921329e-01 -4.79522068e-03 1.02324829e-01 -5.46454251e-01 -2.53321677e-01 1.23596728e+00 9.60804746e-02 2.42887586e-01 6.70092106e-01 1.29488611e+00 1.34300739e-01 -1.28144133e+00 -4.49514180e-01 -1.80220470e-01 -4.08535391e-01 -4.94828112e-02 -9.09586728e-01 -8.35471392e-01 9.07277346e-01 2.48843759e-01 3.23192149e-01 1.09192801e+00 -3.05008918e-01 6.77891254e-01 3.72775912e-01 -4.28696156e-01 -1.02783799e+00 2.98179507e-01 5.34911394e-01 9.77584720e-01 -1.31007683e+00 1.63911488e-02 -2.33158946e-01 -9.57970917e-01 8.11581850e-01 6.17732644e-01 3.15781027e-01 4.36162502e-01 3.74200046e-01 2.15846539e-01 2.47489035e-01 -1.00003362e+00 2.19764084e-01 1.87882856e-01 3.28074366e-01 6.17795944e-01 -1.91172093e-01 9.77880806e-02 6.49353623e-01 -5.30789852e-01 -1.30686283e-01 4.49377090e-01 1.02532339e+00 -1.99640960e-01 -1.19685805e+00 -5.96862793e-01 7.33774841e-01 -4.73024786e-01 -4.65324134e-01 -4.55865383e-01 9.99862731e-01 -1.71136424e-01 8.62106800e-01 8.84668976e-02 -5.42114377e-01 3.40424895e-01 6.10657811e-01 -5.33694886e-02 -6.27599418e-01 -9.25602615e-01 -1.75500020e-01 3.74996774e-02 -1.74421862e-01 1.23845816e-01 -6.21304154e-01 -1.44630396e+00 -4.41242307e-01 -4.13810760e-01 1.56654000e-01 7.66000330e-01 8.75844240e-01 5.37272990e-01 6.43268228e-01 6.99700356e-01 -8.54872346e-01 -6.48749173e-01 -1.32301784e+00 -2.54042652e-02 6.68044090e-01 5.97210407e-01 -3.60308528e-01 -4.98082310e-01 2.67703503e-01]
[10.749519348144531, 8.146260261535645]
ec1dd753-fd80-4732-b757-f46fd192bcd1
masonnlp-at-semeval-2023-task-8-extracting
2304.13875
null
https://arxiv.org/abs/2304.13875v1
https://arxiv.org/pdf/2304.13875v1.pdf
MasonNLP+ at SemEval-2023 Task 8: Extracting Medical Questions, Experiences and Claims from Social Media using Knowledge-Augmented Pre-trained Language Models
In online forums like Reddit, users share their experiences with medical conditions and treatments, including making claims, asking questions, and discussing the effects of treatments on their health. Building systems to understand this information can effectively monitor the spread of misinformation and verify user claims. The Task-8 of the 2023 International Workshop on Semantic Evaluation focused on medical applications, specifically extracting patient experience- and medical condition-related entities from user posts on social media. The Reddit Health Online Talk (RedHot) corpus contains posts from medical condition-related subreddits with annotations characterizing the patient experience and medical conditions. In Subtask-1, patient experience is characterized by personal experience, questions, and claims. In Subtask-2, medical conditions are characterized by population, intervention, and outcome. For the automatic extraction of patient experiences and medical condition information, as a part of the challenge, we proposed language-model-based extraction systems that ranked $3^{rd}$ on both subtasks' leaderboards. In this work, we describe our approach and, in addition, explore the automatic extraction of this information using domain-specific language models and the inclusion of external knowledge.
['Ozlem Uzuner', 'Kevin Lybarger', 'Haritha Gangavarapu', 'Giridhar Kaushik Ramachandran']
2023-04-26
null
null
null
null
['misinformation']
['miscellaneous']
[ 1.11290209e-01 8.19570065e-01 -4.64434296e-01 -4.13935632e-01 -1.22930551e+00 -5.73201656e-01 3.41983199e-01 1.30749297e+00 -4.27802086e-01 8.58989477e-01 1.02250540e+00 -1.79630473e-01 -4.22718078e-01 -4.75157797e-01 -2.44533911e-01 -6.87575191e-02 -3.58371772e-02 7.16976225e-01 -8.38036835e-02 -2.43202031e-01 1.83473647e-01 -9.39998999e-02 -1.00238454e+00 8.95895481e-01 5.68392515e-01 1.03149545e+00 6.59646615e-02 5.18453479e-01 -3.00688863e-01 1.51402450e+00 -6.08966410e-01 -5.62693417e-01 -4.24671590e-01 -3.18360329e-01 -1.37071621e+00 8.65109712e-02 -9.43138376e-02 1.10980179e-02 1.15011856e-02 8.90904188e-01 6.23688102e-01 -2.44925901e-01 2.50266701e-01 -1.01939929e+00 -4.95893151e-01 9.38515365e-01 1.13570526e-01 -4.55070138e-02 1.00796711e+00 -2.87225574e-01 1.18216431e+00 -2.19622180e-01 1.45215559e+00 9.77316737e-01 8.69261682e-01 6.02887511e-01 -9.20496285e-01 -2.62925833e-01 -1.01940095e-01 -1.17022768e-01 -1.10699916e+00 -3.67678702e-01 8.15573633e-02 -7.88752794e-01 1.04386342e+00 8.17754447e-01 4.94263649e-01 1.08577204e+00 6.20756857e-02 7.39605188e-01 1.04952848e+00 -7.38426521e-02 3.35503787e-01 7.59024501e-01 4.88653868e-01 8.16100180e-01 6.67738616e-02 -3.38743001e-01 -4.26085979e-01 -9.63346481e-01 2.03722492e-01 9.38059948e-03 -4.40711230e-01 3.59543085e-01 -1.22773719e+00 6.11530125e-01 3.21365952e-01 4.43576157e-01 -8.30472231e-01 -3.64279807e-01 7.20090389e-01 1.18491858e-01 8.42116714e-01 8.21013689e-01 -1.02320123e+00 -2.31400639e-01 -8.26519191e-01 3.47125143e-01 1.42356157e+00 7.07168818e-01 3.86135668e-01 -1.21617496e+00 -6.07897699e-01 9.45662856e-01 -1.54442508e-02 1.73854947e-01 5.04465282e-01 -9.38670099e-01 5.06660759e-01 1.02091205e+00 5.26966631e-01 -1.07773280e+00 -1.01433134e+00 -3.79793972e-01 -6.08000219e-01 -8.06855798e-01 -1.50887862e-01 -7.20716298e-01 -6.70432687e-01 1.55244637e+00 4.90077615e-01 -1.09463511e-02 2.70168334e-01 5.39948404e-01 1.48660123e+00 1.56944498e-01 5.84031045e-01 -4.83435988e-01 1.93931520e+00 -5.92025459e-01 -1.24634719e+00 8.20616856e-02 1.29951012e+00 -9.66043890e-01 3.97990584e-01 -4.26702239e-02 -1.11624575e+00 6.21176548e-02 -1.57483503e-01 3.60242985e-02 -6.31072879e-01 1.38190418e-01 4.12685096e-01 9.85314772e-02 -7.21350372e-01 6.77026868e-01 -4.51311350e-01 -6.12611234e-01 4.79372948e-01 -1.15855187e-01 -3.69378209e-01 2.25056946e-01 -1.48982775e+00 9.61099863e-01 1.39121890e-01 -4.55678463e-01 -4.35533255e-01 -1.26516473e+00 -7.78560877e-01 -5.82886720e-03 4.15117353e-01 -1.06389499e+00 1.52274776e+00 -5.04817009e-01 -8.33584309e-01 1.40219128e+00 -6.46082237e-02 -5.53424358e-01 2.59971231e-01 -1.42316118e-01 -8.58174324e-01 4.73474972e-02 4.68750834e-01 4.72813606e-01 1.49750829e-01 -6.44109786e-01 -8.53394806e-01 -4.73145276e-01 1.05062202e-01 1.67657837e-01 -1.64256170e-01 3.59607786e-01 -2.23648921e-01 -4.11949694e-01 -2.38744751e-01 -8.75000358e-01 -5.45752048e-01 -4.43618208e-01 -8.35331619e-01 -3.15558881e-01 3.71480763e-01 -1.01636088e+00 1.56491280e+00 -2.03430223e+00 -1.10653050e-01 2.73934323e-02 4.90362078e-01 4.82605696e-02 2.37997323e-01 8.96933317e-01 1.39286950e-01 5.55276752e-01 -7.39710219e-03 -1.88367650e-01 -1.21951267e-01 3.73198502e-02 -4.16347086e-02 -1.23279944e-01 4.41585667e-02 7.76060760e-01 -1.14872277e+00 -6.12449944e-01 -2.27284297e-01 4.89782214e-01 -4.89662886e-01 1.60486385e-01 -4.22970474e-01 2.67946362e-01 -8.39792311e-01 4.13174391e-01 3.17262039e-02 -9.38379288e-01 3.34246606e-01 -4.04384911e-01 -9.95420963e-02 1.00317156e+00 -6.40604019e-01 1.63340068e+00 -6.97578490e-01 2.64861435e-01 4.77708757e-01 -5.54610729e-01 4.00624543e-01 9.28541303e-01 1.03367794e+00 -2.95105368e-01 2.19872311e-01 1.12625748e-01 -4.93417144e-01 -1.11773932e+00 3.61068338e-01 -7.02485889e-02 -4.36893135e-01 3.83423448e-01 -8.24568197e-02 7.00476021e-02 -1.20232955e-01 5.40618896e-01 1.72396839e+00 -4.06899005e-01 6.49208724e-01 -3.24980050e-01 3.79428834e-01 3.28776836e-01 4.28817362e-01 5.43207586e-01 -5.71989492e-02 2.56707937e-01 5.06372571e-01 -2.34364077e-01 -3.97947907e-01 -4.99735862e-01 -3.50265801e-01 8.23393524e-01 -9.66471657e-02 -8.29306364e-01 -6.33954525e-01 -8.85956943e-01 3.47902142e-02 8.14983010e-01 -7.71921933e-01 3.39190274e-01 8.86794627e-02 -7.76872158e-01 4.67614532e-01 1.93433419e-01 5.49581885e-01 -9.74884033e-01 -7.17447579e-01 5.38114130e-01 -9.04668093e-01 -1.26145446e+00 -3.98593813e-01 -1.03429876e-01 -4.47846353e-01 -1.35998225e+00 -6.65196359e-01 -6.86158001e-01 4.40120846e-01 -5.60561419e-01 1.29088521e+00 -1.02646854e-02 -7.24753618e-01 6.66131258e-01 -3.32581013e-01 -5.77868581e-01 -4.43680584e-01 1.88857932e-02 -5.50417304e-01 -1.21144541e-01 4.69891578e-01 -1.20825313e-01 -7.97287285e-01 3.28765996e-02 -7.71551490e-01 2.10833564e-01 2.77033150e-01 4.89739001e-01 2.97310412e-01 -2.28344187e-01 6.16964996e-01 -1.55447233e+00 9.92582440e-01 -9.87591445e-01 6.83054626e-02 3.00166368e-01 -5.35803556e-01 -1.10421628e-01 -3.24363336e-02 1.14328295e-01 -1.17966688e+00 -9.39863026e-02 -3.24409366e-01 4.21763539e-01 -3.33531201e-01 1.05354810e+00 2.81865954e-01 6.74777925e-01 1.00018644e+00 -3.79121065e-01 -1.63525701e-01 -6.56620383e-01 2.53494978e-01 1.22464168e+00 2.77159780e-01 -1.17023692e-01 -1.30322371e-02 4.96552527e-01 -4.04226065e-01 -6.42660141e-01 -1.62486684e+00 -1.19599974e+00 2.05781519e-01 -4.78552654e-02 1.26411390e+00 -1.09404588e+00 -8.65171313e-01 8.66045654e-02 -1.14257407e+00 -1.40405625e-01 -4.51555938e-01 4.04007643e-01 -2.08550423e-01 -2.09127754e-01 -9.50821400e-01 -6.72270656e-01 -5.05957365e-01 -6.32582366e-01 9.83499169e-01 -1.24455869e-01 -9.86463666e-01 -1.32859004e+00 3.11027229e-01 7.79796898e-01 4.66889739e-01 4.22562033e-01 8.16575110e-01 -1.06439722e+00 1.47772908e-01 -4.41095650e-01 -4.08616573e-01 3.06758061e-02 5.41819632e-01 -5.59604228e-01 -6.33032382e-01 1.87463865e-01 -1.41499072e-01 -4.21802163e-01 4.70576525e-01 4.78719920e-01 1.19083107e+00 -7.23766387e-01 -7.61576891e-01 -3.05476397e-01 1.10423875e+00 -1.24762230e-01 4.35255647e-01 7.16548041e-02 3.34526718e-01 9.75831747e-01 4.98632878e-01 7.84590304e-01 6.40786469e-01 5.12060583e-01 3.71209532e-02 -2.43631467e-01 -5.14184870e-02 -2.91873634e-01 2.03133635e-02 6.27321005e-01 -3.34511735e-02 -5.73410802e-02 -8.71466517e-01 8.43083382e-01 -1.71391046e+00 -8.18257332e-01 -2.41673067e-01 1.88554752e+00 1.26016033e+00 9.18975566e-04 -7.77252426e-04 -4.20250595e-01 5.92024982e-01 -1.86894044e-01 -2.79755324e-01 -2.31403515e-01 1.41254514e-01 2.05590174e-01 6.00148201e-01 5.04941344e-01 -8.87580395e-01 5.18488109e-01 5.50880384e+00 5.61491847e-01 -6.62457407e-01 3.93684834e-01 7.73141503e-01 5.73418550e-02 -4.44551945e-01 -9.88340005e-02 -9.28976119e-01 2.69924372e-01 1.11977315e+00 -1.35681003e-01 9.10783000e-03 5.12147427e-01 4.70934063e-01 -9.30458531e-02 -9.68154252e-01 5.94380260e-01 4.68450300e-02 -1.67318702e+00 -5.23771763e-01 -1.39812976e-01 7.89986789e-01 1.71296701e-01 -1.59118146e-01 2.20213473e-01 4.30785835e-01 -7.65038252e-01 3.13349336e-01 6.96402431e-01 7.00488627e-01 -9.95535925e-02 9.81664598e-01 3.22837472e-01 -8.72207403e-01 4.02977392e-02 4.11725730e-01 1.10459283e-01 5.03982484e-01 1.00584233e+00 -1.27523446e+00 6.70923054e-01 7.36667871e-01 9.47703719e-01 -1.66487664e-01 9.06397641e-01 -1.86304897e-02 6.67854607e-01 -4.26491834e-02 -1.68375358e-01 2.84354505e-03 3.26027483e-01 5.41991293e-01 1.48924792e+00 2.83094734e-01 5.04956663e-01 2.10175946e-01 7.27891743e-01 -5.70374012e-01 5.01287401e-01 -3.94475311e-01 -6.32426918e-01 3.21298271e-01 1.34015298e+00 -4.55797553e-01 -6.79005504e-01 -1.21677786e-01 7.02291310e-01 1.02958806e-01 1.19696878e-01 -4.49915141e-01 -2.68037349e-01 5.05732596e-01 5.99491000e-01 -1.57013252e-01 6.61590636e-01 -1.42969370e-01 -9.83696282e-01 -2.72392958e-01 -7.06788778e-01 7.82535672e-01 -6.64609551e-01 -1.43338501e+00 6.18216872e-01 -2.01505348e-01 -9.52655673e-01 -2.95387298e-01 -1.64860591e-01 3.86011712e-02 5.99696159e-01 -1.06455731e+00 -7.85341084e-01 -1.44878641e-01 5.38066626e-01 2.85042942e-01 2.18765989e-01 1.33152640e+00 5.69870114e-01 -3.24865788e-01 1.01744846e-01 -3.74410778e-01 1.24749944e-01 9.81588721e-01 -1.05731475e+00 -7.57631809e-02 -4.31372583e-01 -2.50035048e-01 5.28619826e-01 7.88972318e-01 -1.09844899e+00 -6.26514018e-01 -1.24190485e+00 1.91853666e+00 -7.10143685e-01 8.22229207e-01 1.14767492e-01 -7.08740115e-01 5.53848028e-01 2.79017746e-01 -4.54076260e-01 1.30365419e+00 5.40441573e-01 -1.33756995e-01 3.94320399e-01 -1.47565961e+00 2.91221619e-01 9.48085487e-01 -8.12269211e-01 -8.03579032e-01 1.05530727e+00 7.38156617e-01 -5.10243773e-01 -1.35240817e+00 3.05996835e-01 1.01572409e-01 -3.17305207e-01 6.87887013e-01 -1.12202930e+00 6.58223689e-01 2.33954087e-01 -6.55260608e-02 -1.33712602e+00 1.55945882e-01 -8.33871245e-01 1.79548874e-01 9.69173074e-01 1.01040900e+00 -4.75986838e-01 4.15621728e-01 1.04903090e+00 -1.61068887e-02 -6.98553801e-01 -5.25619566e-01 -1.57528430e-01 -4.79617715e-01 -2.70817637e-01 2.29982346e-01 1.38842058e+00 7.05198526e-01 5.08689702e-01 -2.17772275e-01 1.36875033e-01 2.24030718e-01 8.24290365e-02 -3.68597768e-02 -1.36624193e+00 -3.60102654e-01 -3.45909707e-02 -4.52184901e-02 -5.46611965e-01 -1.70157790e-01 -8.89800727e-01 -1.40479222e-01 -2.10721993e+00 5.82027316e-01 -5.03182292e-01 -2.47093439e-01 6.80879354e-01 -9.87552777e-02 -1.46579683e-01 -3.09988230e-01 7.93101937e-02 -8.31841171e-01 -1.66096970e-01 8.35060537e-01 -1.11225978e-01 -3.51228386e-01 2.66163528e-01 -8.60528171e-01 8.01583827e-01 6.72532797e-01 -6.16797268e-01 -1.18657604e-01 -1.36600155e-02 8.04595649e-01 4.97902036e-01 4.08053577e-01 -4.99782711e-01 1.61207616e-01 -1.25258416e-01 -3.96936476e-01 -2.86113858e-01 2.34543756e-01 -6.69906855e-01 1.92547128e-01 6.40862346e-01 -1.14972651e+00 -2.26380035e-01 2.17020571e-01 4.55652148e-01 -3.23983699e-01 -1.28141433e-01 1.97462365e-01 -5.11230350e-01 -1.66974083e-01 8.02142769e-02 -6.64422870e-01 4.44581717e-01 7.25131273e-01 7.94866621e-01 -5.12651443e-01 -5.94677269e-01 -1.42845893e+00 4.35214520e-01 -9.30729508e-02 3.61097515e-01 3.60233307e-01 -8.12229037e-01 -9.81007993e-01 -5.15430689e-01 2.54426599e-01 -3.70156258e-01 6.96093976e-01 1.22336042e+00 -2.55474299e-01 4.13141668e-01 2.81622022e-01 -2.96231471e-02 -1.59842205e+00 5.06504357e-01 7.31902495e-02 -8.01567376e-01 -2.95275867e-01 9.66759384e-01 -1.02436289e-01 -5.05608857e-01 3.62530649e-01 -3.33388954e-01 -3.00647616e-01 3.16313386e-01 7.05559731e-01 2.49183014e-01 2.57634670e-01 -3.31830859e-01 -3.26606512e-01 -1.65707514e-01 -1.44119024e-01 -2.33788326e-01 1.44735134e+00 -2.22021371e-01 -4.55056548e-01 3.22115362e-01 1.28994846e+00 -1.17057571e-02 -6.24199472e-02 -3.01311970e-01 3.13481510e-01 1.30960196e-01 2.77945071e-01 -1.53463137e+00 -6.51257932e-01 3.33002359e-01 3.30902040e-01 6.80298686e-01 1.00135279e+00 4.61096078e-01 9.56455171e-01 2.73588836e-01 -7.43366405e-02 -1.32349372e+00 -2.14970082e-01 1.74560294e-01 8.67539585e-01 -1.19121253e+00 -3.12115550e-01 -7.30251133e-01 -9.10570264e-01 4.99470711e-01 8.79253075e-02 6.82353377e-01 1.11760592e+00 6.80810064e-02 2.24261537e-01 -9.38337505e-01 -1.03076422e+00 -2.62997061e-01 3.92118216e-01 2.98497379e-02 7.68869638e-01 4.79480535e-01 -5.38116455e-01 8.29014301e-01 -4.00891677e-02 4.12886411e-01 2.70621508e-01 1.06109166e+00 -1.73858389e-01 -1.08531022e+00 1.34197310e-01 7.80515194e-01 -9.74950910e-01 -2.60883898e-01 -6.85656071e-01 2.22095266e-01 3.91946822e-01 1.14218593e+00 -2.53547102e-01 -2.91016698e-01 5.51583886e-01 1.41293377e-01 -1.33260891e-01 -1.06984258e+00 -1.20236683e+00 -1.73292890e-01 1.07323599e+00 -6.95618451e-01 -8.08153868e-01 -6.29256546e-01 -1.33855522e+00 1.03964403e-01 -1.69819862e-01 6.27500176e-01 8.25762391e-01 9.50046420e-01 9.45706069e-01 6.87016428e-01 2.99715072e-01 3.06836486e-01 -2.74786472e-01 -9.11337733e-01 -3.20785731e-01 9.14244115e-01 3.02847177e-01 -1.12737060e-01 -1.81883425e-01 1.25098243e-01]
[8.644426345825195, 8.87440013885498]
972ea3d8-83f6-40a5-b6bf-be861edb9ee7
co-optimal-transport
2002.03731
null
https://arxiv.org/abs/2002.03731v3
https://arxiv.org/pdf/2002.03731v3.pdf
CO-Optimal Transport
Optimal transport (OT) is a powerful geometric and probabilistic tool for finding correspondences and measuring similarity between two distributions. Yet, its original formulation relies on the existence of a cost function between the samples of the two distributions, which makes it impractical when they are supported on different spaces. To circumvent this limitation, we propose a novel OT problem, named COOT for CO-Optimal Transport, that simultaneously optimizes two transport maps between both samples and features, contrary to other approaches that either discard the individual features by focusing on pairwise distances between samples or need to model explicitly the relations between them. We provide a thorough theoretical analysis of our problem, establish its rich connections with other OT-based distances and demonstrate its versatility with two machine learning applications in heterogeneous domain adaptation and co-clustering/data summarization, where COOT leads to performance improvements over the state-of-the-art methods.
['Rémi Flamary', 'Titouan Vayer', 'Ievgen Redko', 'Nicolas Courty']
2020-02-10
null
http://proceedings.neurips.cc/paper/2020/hash/cc384c68ad503482fb24e6d1e3b512ae-Abstract.html
http://proceedings.neurips.cc/paper/2020/file/cc384c68ad503482fb24e6d1e3b512ae-Paper.pdf
neurips-2020-12
['data-summarization']
['miscellaneous']
[ 1.29509479e-01 -3.20294350e-01 -2.27824777e-01 -2.85381675e-01 -1.07393932e+00 -6.47394180e-01 6.13841891e-01 6.22026145e-01 -2.58781374e-01 5.78718543e-01 1.38587564e-01 -5.17655574e-02 -7.63339520e-01 -6.55330062e-01 -5.39059937e-01 -9.29071069e-01 -1.86751246e-01 8.84922147e-01 2.74833113e-01 -8.35109726e-02 2.95814842e-01 4.25668031e-01 -1.48739707e+00 1.03905395e-01 1.01041293e+00 9.73806381e-01 2.55138546e-01 4.93677527e-01 -2.82676339e-01 7.20265284e-02 -4.89122450e-01 -3.49450022e-01 2.08699062e-01 -5.58150411e-01 -8.97845626e-01 1.11939445e-01 4.94917065e-01 2.53024936e-01 -9.80220884e-02 1.16721225e+00 6.51940823e-01 1.77839339e-01 1.18798423e+00 -1.47685027e+00 -7.05304027e-01 2.99602211e-01 -5.36024988e-01 3.57183889e-02 2.98044205e-01 -3.26468259e-01 1.27997124e+00 -7.03101754e-01 5.03688633e-01 1.13508201e+00 1.06843317e+00 2.55957782e-01 -1.70244646e+00 -5.94243035e-02 -2.11229548e-03 3.38298827e-01 -1.50494552e+00 -3.61878246e-01 5.08459091e-01 -5.90000153e-01 4.57079172e-01 3.28854829e-01 5.89988768e-01 7.45798469e-01 -1.02579094e-01 9.33777392e-01 8.67072701e-01 -3.93416941e-01 4.20495898e-01 1.58772931e-01 4.76521924e-02 5.02013624e-01 1.74852937e-01 -3.03025723e-01 -5.00104010e-01 -4.27879542e-01 4.60454494e-01 5.48873655e-02 -3.67322952e-01 -8.68636668e-01 -1.41266620e+00 8.20300162e-01 3.04030240e-01 5.07329345e-01 -6.57737404e-02 -3.76173407e-02 3.48650455e-01 2.91280419e-01 5.72694957e-01 4.85889196e-01 -4.20283645e-01 -5.98039031e-02 -9.67561543e-01 3.06717902e-01 7.75573373e-01 1.14381933e+00 8.59448314e-01 -6.00042105e-01 -2.10101575e-01 9.47582841e-01 1.32397369e-01 4.34355587e-01 2.24743798e-01 -7.16145456e-01 5.74744582e-01 5.13778090e-01 1.02997892e-01 -1.17521608e+00 -2.63047576e-01 -2.31926918e-01 -8.03806603e-01 -3.57767075e-01 8.24440300e-01 1.79433465e-01 -3.24813038e-01 1.60657895e+00 5.10396242e-01 3.38944763e-01 -1.04142345e-01 7.27995396e-01 3.88340831e-01 3.47364962e-01 -3.82144451e-01 -4.96775247e-02 1.11558592e+00 -7.06359208e-01 -4.73992378e-01 3.24377805e-01 7.29841530e-01 -7.69081354e-01 1.13178968e+00 1.62462562e-01 -1.00754738e+00 -1.17143787e-01 -9.23640072e-01 -6.01149313e-02 -4.33563530e-01 5.32496199e-02 4.09676403e-01 5.19212842e-01 -9.98799086e-01 9.71786737e-01 -7.31739044e-01 -6.45120561e-01 5.92857420e-01 4.04050857e-01 -3.02682638e-01 -1.69470176e-01 -7.28275478e-01 6.92107320e-01 5.70245236e-02 -1.61497101e-01 -3.89702052e-01 -1.05969012e+00 -6.97211742e-01 -5.67241013e-02 2.18592703e-01 -8.33036721e-01 8.30728710e-01 -6.07542932e-01 -1.49104929e+00 6.17010951e-01 -3.80450189e-01 -1.82673275e-01 7.70838797e-01 -1.53197870e-01 -1.20706789e-01 -4.73487787e-02 2.85720944e-01 3.98575932e-01 7.20504820e-01 -1.01483166e+00 -7.19731510e-01 -3.26812923e-01 -1.88075706e-01 1.13347232e-01 -4.16805744e-01 -1.36414468e-01 -5.05613387e-01 -5.48928201e-01 1.20988898e-01 -8.12867939e-01 -1.91582382e-01 1.64159089e-01 -8.37075710e-01 -4.24483985e-01 8.53575647e-01 -2.36416996e-01 9.41143811e-01 -2.12551403e+00 5.60352743e-01 3.46737117e-01 2.41763234e-01 -4.82179709e-02 -1.64048716e-01 5.27849078e-01 2.30152264e-01 9.46773142e-02 -7.05738544e-01 -5.43387711e-01 2.69628078e-01 2.08480388e-01 -1.37067571e-01 9.89868820e-01 7.96165615e-02 6.33498788e-01 -1.25523901e+00 -4.94428247e-01 2.88299024e-01 5.48354506e-01 -5.24367213e-01 9.32938419e-03 -1.42357379e-01 6.44237041e-01 -3.36400926e-01 4.19844687e-01 9.49384153e-01 -2.45987594e-01 1.37416601e-01 -2.48532176e-01 -3.03992070e-03 2.02645719e-01 -1.28123307e+00 1.85352910e+00 -4.01780844e-01 5.93507886e-01 -1.14040747e-01 -1.26736677e+00 7.88064539e-01 1.06542399e-02 9.80400920e-01 -3.08938593e-01 -1.45687284e-02 4.31993634e-01 -3.71703774e-01 -4.36838150e-01 4.51007575e-01 -2.39145070e-01 -1.56239972e-01 3.97576094e-01 -1.79712288e-02 -1.56175226e-01 2.20095709e-01 -3.49399038e-02 1.04250431e+00 -2.62951478e-02 1.50902390e-01 -7.14671195e-01 6.76112413e-01 -1.49517491e-01 3.86808097e-01 6.49586797e-01 -9.35581252e-02 9.43457723e-01 6.11976922e-01 2.97994558e-02 -1.10734022e+00 -1.35042143e+00 -4.12603110e-01 7.36019909e-01 5.17204940e-01 -4.79056209e-01 -7.66660392e-01 -1.05822861e+00 3.39172482e-01 4.96481776e-01 -6.51350975e-01 1.63339585e-01 -4.01365042e-01 -7.86433697e-01 4.41794723e-01 3.74494970e-01 3.11559767e-01 -2.00544745e-01 -9.94615853e-02 2.06736326e-02 -3.64531308e-01 -1.06395698e+00 -8.97432983e-01 -1.76685248e-02 -8.97438943e-01 -1.29588497e+00 -9.19310868e-01 -5.53759038e-01 7.40637004e-01 3.09612721e-01 9.95104074e-01 -4.50437441e-02 -1.96904019e-01 4.24409747e-01 -2.43410841e-01 -1.18667029e-01 -1.27155915e-01 4.92032260e-01 1.02689005e-01 2.93160051e-01 2.75016367e-01 -7.09213912e-01 -5.20676374e-01 7.59458721e-01 -7.92159379e-01 -2.61250734e-01 2.60126650e-01 9.32935059e-01 8.08948338e-01 1.11560673e-01 5.90813518e-01 -7.86508918e-01 5.26455224e-01 -7.50517964e-01 -5.16832650e-01 5.13702571e-01 -3.68550658e-01 4.68561947e-01 6.72296464e-01 -3.29235554e-01 -6.63019419e-01 -5.71542457e-02 9.45068225e-02 -2.31381714e-01 -5.77481650e-02 2.53106564e-01 -3.71049583e-01 -8.23903531e-02 4.23907965e-01 1.87428266e-01 6.56970814e-02 -6.09906852e-01 5.76027572e-01 6.44190848e-01 4.36129957e-01 -7.47907758e-01 8.50713491e-01 9.33068395e-01 2.90396929e-01 -1.06318712e+00 -9.06119764e-01 -8.09619725e-01 -1.18426836e+00 1.88290715e-01 7.57685244e-01 -2.08118886e-01 -7.49962151e-01 7.05731213e-02 -1.01473641e+00 -8.42775926e-02 -5.41455209e-01 5.58478236e-01 -8.09473574e-01 4.93428648e-01 -4.52198833e-02 -5.97741067e-01 1.14768431e-01 -1.08841431e+00 1.29294503e+00 -2.92162567e-01 -2.37218380e-01 -1.55147278e+00 2.63846368e-01 -4.88038957e-02 4.08782780e-01 3.19115549e-01 1.18970311e+00 -6.35567367e-01 -5.07594526e-01 -1.86812326e-01 -3.26860726e-01 1.49066091e-01 3.20530921e-01 9.92763694e-03 -6.54798806e-01 -2.72818118e-01 -1.99929476e-01 1.81649253e-01 7.55966723e-01 5.56119800e-01 1.18353999e+00 -3.17524374e-01 -6.89815998e-01 7.52125859e-01 1.51316655e+00 -3.75654042e-01 4.24900085e-01 1.78574219e-01 7.79826045e-01 9.06962216e-01 4.24305499e-01 4.05624568e-01 5.09501517e-01 1.08923650e+00 1.67261571e-01 1.58455908e-01 -1.49575084e-01 -2.68023610e-01 2.12635592e-01 1.00002849e+00 7.39048645e-02 -1.44959226e-01 -4.46119606e-01 7.85354555e-01 -2.14482903e+00 -9.21231091e-01 -4.70649093e-01 2.63631582e+00 5.90812445e-01 -4.90214288e-01 4.12742317e-01 1.76817104e-01 9.54971373e-01 4.06116992e-02 -3.66838336e-01 -4.95309047e-02 -3.24693710e-01 1.20386615e-01 6.59423470e-01 4.81657773e-01 -1.14939153e+00 5.37009358e-01 6.89670992e+00 1.17854178e+00 -9.31443691e-01 2.24007562e-01 2.41776839e-01 -1.49260595e-01 -5.81141651e-01 -5.92795946e-02 -6.44910097e-01 6.09317601e-01 6.04084849e-01 -2.94586152e-01 4.14030492e-01 5.38230598e-01 6.46714717e-02 -8.01001042e-02 -1.51649892e+00 9.54755366e-01 -5.29096909e-02 -1.36936855e+00 -6.68480545e-02 2.25859508e-01 8.37207556e-01 -6.84796786e-03 2.93916076e-01 -2.25514069e-01 2.81892031e-01 -9.66368079e-01 7.47030437e-01 4.91937220e-01 6.23286724e-01 -8.01342607e-01 7.04022944e-01 2.11802632e-01 -1.27204800e+00 1.19618081e-01 -5.22972226e-01 3.75612527e-01 2.74515212e-01 9.26413953e-01 -8.90135884e-01 9.39354599e-01 5.28546393e-01 1.08302534e+00 -4.45677638e-01 1.34922409e+00 1.59118026e-01 2.78945506e-01 -5.18027544e-01 3.68470959e-02 1.72367301e-02 -4.02251124e-01 8.58152866e-01 1.27840471e+00 6.23132944e-01 -5.92824101e-01 2.14953348e-01 9.32022154e-01 -3.80152278e-02 3.81801158e-01 -6.79862261e-01 2.18058795e-01 6.11947656e-01 9.12803292e-01 -6.81196272e-01 -1.12546690e-01 -3.93138707e-01 1.09048295e+00 3.40983957e-01 2.84246087e-01 -7.55351067e-01 -4.42052096e-01 9.01468456e-01 2.61941344e-01 4.24517930e-01 -3.81544530e-01 -4.75971490e-01 -1.07240486e+00 2.34946221e-01 -3.52294505e-01 4.32014465e-01 -6.66461140e-02 -1.72058213e+00 3.19749445e-01 2.13256925e-01 -1.42838526e+00 6.79184869e-02 -4.11394864e-01 -4.49352145e-01 7.46393204e-01 -1.48119080e+00 -1.04004610e+00 -4.60367464e-02 7.59487927e-01 2.19582826e-01 -2.10639089e-02 7.56887019e-01 5.53812921e-01 -4.38876539e-01 6.77890301e-01 7.88614869e-01 -3.53443265e-01 6.40544891e-01 -1.56861615e+00 2.13041469e-01 5.58729112e-01 1.66604161e-01 4.43972766e-01 6.24951780e-01 -3.32504481e-01 -1.30707181e+00 -1.16318119e+00 1.07696402e+00 -6.34730160e-01 7.34795272e-01 -4.65024471e-01 -7.79036939e-01 3.95139307e-01 -1.01729492e-02 1.61639139e-01 7.95582533e-01 1.93723023e-01 -3.07793587e-01 -2.02430516e-01 -1.30816400e+00 5.08995533e-01 1.30636406e+00 -4.04406160e-01 -2.92094707e-01 5.24495363e-01 4.50813413e-01 2.75675226e-02 -1.06320155e+00 2.18828768e-01 4.32039618e-01 -1.16037822e+00 1.12074828e+00 -4.39581245e-01 2.46287435e-02 -6.18413091e-01 -6.06271863e-01 -1.50544310e+00 -1.34048030e-01 -7.46477187e-01 1.37370124e-01 1.30713320e+00 3.60358685e-01 -8.09054911e-01 6.94586158e-01 1.42726168e-01 -2.27583721e-02 -7.58551002e-01 -1.18560672e+00 -1.21199286e+00 2.28256479e-01 -5.78965783e-01 8.82773817e-01 1.10545123e+00 9.26607028e-02 1.58778146e-01 -2.10580498e-01 2.47844085e-01 8.06329906e-01 3.44255716e-02 7.77395070e-01 -1.52039373e+00 -1.78826571e-01 -8.28950822e-01 -6.76606894e-01 -1.33981681e+00 2.79467225e-01 -1.28054082e+00 7.94739798e-02 -1.56527996e+00 1.44639015e-01 -9.96578872e-01 1.75257921e-02 2.41595483e-03 8.06313008e-02 2.14095470e-02 -4.69837710e-02 5.01802564e-01 -8.08945119e-01 8.28126550e-01 1.08855402e+00 -1.60317495e-01 -3.96464586e-01 1.96688429e-01 -6.86115503e-01 5.70815444e-01 5.01286745e-01 -6.26344621e-01 -3.80871296e-01 -4.36870337e-01 -1.25086889e-01 -3.54443550e-01 2.98318833e-01 -9.30963933e-01 3.56931031e-01 -2.10216239e-01 -9.82712023e-03 -5.45285463e-01 1.43995404e-01 -8.59753072e-01 1.01161852e-01 7.41616413e-02 -3.80710661e-01 4.88028191e-02 -1.76855847e-01 8.60940278e-01 -2.34333798e-01 -2.97044605e-01 8.26951027e-01 2.31069729e-01 -2.45345339e-01 4.59282726e-01 -3.98268104e-02 1.67578503e-01 1.08780575e+00 -2.21311063e-01 -2.18409598e-01 -1.87637389e-01 -5.79639971e-01 1.04677893e-01 6.80113494e-01 1.08325563e-01 5.70157290e-01 -1.58493268e+00 -7.47187078e-01 4.18295227e-02 3.00556034e-01 2.24835575e-01 1.44894883e-01 1.46039259e+00 -3.47930431e-01 3.52315664e-01 2.07830071e-01 -1.04798591e+00 -1.19617641e+00 4.42540228e-01 3.83801132e-01 -2.39558235e-01 -7.22044945e-01 7.36916125e-01 2.05957338e-01 -7.10106850e-01 1.51187584e-01 -3.74435335e-01 3.61671038e-02 1.06109239e-01 2.67302692e-01 8.26718450e-01 1.02808587e-01 -7.79433191e-01 -5.08412600e-01 8.85138512e-01 1.31692946e-01 -2.72971004e-01 1.29126740e+00 -3.16590577e-01 -1.96895659e-01 5.97542882e-01 1.65231037e+00 1.28669426e-01 -1.23269367e+00 -4.92597789e-01 4.16908443e-01 -8.89693975e-01 -3.20965230e-01 -1.92917258e-01 -9.51811016e-01 9.21063483e-01 2.64918387e-01 3.87354851e-01 8.60721767e-01 3.98479939e-01 8.53767753e-01 3.34066153e-01 3.02732170e-01 -1.10042000e+00 1.61483541e-01 4.30909991e-01 7.53878474e-01 -9.88263071e-01 -3.69174406e-02 -6.02653861e-01 -5.08753061e-01 1.12386727e+00 -3.77021916e-02 -1.19114488e-01 9.19756174e-01 -1.23045482e-02 -4.17447090e-01 -1.27748773e-01 -3.18333954e-01 -4.72029358e-01 5.03114879e-01 9.66047466e-01 3.85675877e-01 6.79953620e-02 -1.03301413e-01 1.33203864e-01 -2.68240064e-01 -3.17699373e-01 1.40426040e-01 6.72391593e-01 -2.23753601e-01 -1.30638528e+00 -2.50794470e-01 4.51274395e-01 -2.66552031e-01 -2.16504689e-02 -3.49585801e-01 7.46391058e-01 1.30192548e-01 8.42063665e-01 2.41698116e-01 -3.37363392e-01 4.46096838e-01 -2.11074665e-01 6.76871121e-01 -4.96825963e-01 -1.57826379e-01 1.40528437e-02 -1.87578931e-01 -3.49050224e-01 -5.82199693e-01 -1.06140506e+00 -1.04464495e+00 -5.08562624e-01 -4.16791946e-01 2.66467035e-01 6.64361954e-01 1.12215197e+00 5.39315164e-01 2.41190672e-01 7.54591048e-01 -8.01727891e-01 -5.89888275e-01 -6.75833106e-01 -8.87683570e-01 6.15953624e-01 4.78097498e-01 -9.40543294e-01 -3.52261603e-01 -8.07909593e-02]
[7.779326438903809, 4.109753608703613]
623db609-5e1a-40b2-9e0b-0c3ae3b7dbef
parameterized-pseudo-differential-operators
null
null
https://openreview.net/forum?id=Y45i-hDynr
https://openreview.net/pdf?id=Y45i-hDynr
Parameterized Pseudo-Differential Operators for Graph Convolutional Neural Networks
We present a novel graph convolutional layer that is fast, conceptually simple, and provides high accuracy with reduced overfitting. Based on pseudo-differential operators, our layer operates on graphs with relative position information available for each pair of connected nodes. The new layer outperforms multiple recent architectures on superpixel image classification tasks using the MNIST and CIFAR100 superpixel datasets and performs comparably with recent results on the CIFAR10 superpixel dataset and FAUST node correspondence task. We measure test accuracy without bias to the test set by selecting the model with the best training accuracy. The new layer achieves a test error rate of 0.80% on the MNIST superpixel dataset, beating the closest reported rate of 0.95% by a factor of more than 15%. After dropping roughly 70% of the edge connections from the input by performing a Delaunay triangulation, our model still achieves a competitive error rate of 1.04%.
['John Tencer', 'Matthew David Smith', 'Steven Richard Sleder', 'Kevin M. Potter']
2021-01-01
null
null
null
null
['superpixel-image-classification']
['computer-vision']
[ 8.44125077e-02 4.95876342e-01 -3.27721417e-01 -3.54458869e-01 -7.76832640e-01 -6.19572639e-01 5.04643321e-01 1.88837320e-01 -6.94209158e-01 7.87824988e-01 -2.95787454e-01 -2.41544187e-01 3.87267247e-02 -8.87513578e-01 -9.35900867e-01 -7.24959791e-01 -2.55332291e-01 5.29056966e-01 4.14087474e-01 3.07575196e-01 8.04712158e-03 2.76637286e-01 -8.68833363e-01 4.92573231e-01 7.35367298e-01 1.33134961e+00 -1.82000756e-01 7.94539630e-01 2.83261418e-01 7.43751109e-01 -4.64294821e-01 -5.23713112e-01 4.53026205e-01 -2.04715967e-01 -9.92724776e-01 -1.60512030e-01 1.44467318e+00 -3.59953642e-01 -4.92537796e-01 1.02977979e+00 3.35347265e-01 1.14797603e-03 6.32825255e-01 -1.08541751e+00 -1.09634459e+00 9.57388282e-01 -7.06494689e-01 6.79254651e-01 -1.89321786e-01 1.41930627e-02 1.44980264e+00 -6.79928362e-01 6.92229331e-01 8.03917885e-01 1.03135276e+00 3.04128617e-01 -1.57378995e+00 -6.13668501e-01 2.33844310e-01 -4.28482220e-02 -1.69209802e+00 1.74346879e-01 3.79255265e-01 -2.61771828e-01 1.17741501e+00 -1.82255302e-02 6.98955178e-01 8.95195067e-01 1.67741954e-01 6.82041705e-01 9.95385408e-01 1.03807732e-01 8.44672769e-02 -2.00831249e-01 5.41079342e-01 9.07670677e-01 4.71583545e-01 -3.07494640e-01 -6.56825528e-02 4.31709029e-02 1.15450776e+00 -2.19684273e-01 -2.39458591e-01 -2.75231540e-01 -1.25672567e+00 6.31377101e-01 1.40735233e+00 3.20507526e-01 -8.84118304e-02 7.33569205e-01 1.18331008e-01 2.28979990e-01 5.58343887e-01 4.77019131e-01 -3.58095199e-01 3.52845490e-01 -1.11055648e+00 1.14321746e-02 6.20055377e-01 9.61533368e-01 5.96503973e-01 1.85726285e-01 -3.42830688e-01 6.30934775e-01 -2.55597997e-02 2.42447257e-01 1.70798451e-01 -1.04217958e+00 4.93991971e-01 6.99203551e-01 -2.46684298e-01 -7.79322982e-01 -6.77678168e-01 -1.32513022e+00 -8.74540150e-01 3.14866066e-01 8.04307640e-01 -9.66712981e-02 -1.34578419e+00 1.57000995e+00 -1.03727199e-01 3.84607613e-01 -3.55683774e-01 7.91694880e-01 1.09703624e+00 3.15877944e-01 8.53197724e-02 2.76673138e-01 9.61756825e-01 -1.05083621e+00 2.86029372e-02 -3.69955331e-01 8.13184261e-01 -3.37863892e-01 8.20283532e-01 3.72774392e-01 -9.65268016e-01 -5.92426836e-01 -1.28457558e+00 -3.56786638e-01 -5.11238515e-01 2.05440655e-01 8.60517561e-01 3.61256838e-01 -1.66902208e+00 9.50773537e-01 -7.65223563e-01 -1.20057568e-01 1.12959671e+00 7.06095576e-01 -5.06548643e-01 -4.62478325e-02 -7.43640125e-01 5.38368225e-01 5.10665119e-01 -1.20209761e-01 -7.54271328e-01 -1.08408761e+00 -7.67568350e-01 2.09410891e-01 -1.08794302e-01 -6.50238752e-01 1.01874304e+00 -1.18240356e+00 -8.17957640e-01 1.09294367e+00 2.43757263e-01 -9.65872645e-01 7.69662797e-01 8.30091164e-02 -8.67295116e-02 2.61797041e-01 7.27910325e-02 1.22549009e+00 4.74013478e-01 -7.65665770e-01 -5.51242888e-01 -3.74389738e-01 1.28226116e-01 -5.15316464e-02 -2.47376859e-01 -5.29921055e-01 -4.89927679e-01 -6.69470429e-01 3.17435384e-01 -1.05462563e+00 -1.67004883e-01 1.81691796e-01 -5.25223076e-01 -2.70312190e-01 6.39802337e-01 -4.34549719e-01 7.87658274e-01 -1.97618210e+00 3.18339677e-03 2.13139951e-01 7.86809027e-01 3.22679192e-01 -3.28844219e-01 -1.99336812e-01 -2.99649924e-01 3.78996551e-01 -3.19005042e-01 -2.51418352e-01 -3.38151544e-01 -2.05873892e-01 1.56703293e-01 4.64761525e-01 3.85978043e-01 1.32441890e+00 -1.03193164e+00 -3.37063551e-01 1.39691681e-01 5.37467420e-01 -5.70171475e-01 -5.03779948e-01 -7.23421648e-02 1.15359515e-01 -2.47037068e-01 5.40959656e-01 6.10149503e-01 -9.01347756e-01 3.73895355e-02 -3.25066328e-01 3.36605757e-01 3.48414361e-01 -9.32322919e-01 1.65567744e+00 -1.09594487e-01 1.04448676e+00 -9.03377831e-02 -8.31464529e-01 7.08777487e-01 2.59678345e-02 3.85959983e-01 -7.06972003e-01 1.38917565e-01 2.08047464e-01 2.97332048e-01 1.08937696e-01 3.94240618e-02 3.54964405e-01 2.61767149e-01 1.58779129e-01 4.67034847e-01 2.14866638e-01 2.49641091e-01 4.74913269e-01 1.30606365e+00 -1.32082812e-02 -7.56237283e-02 -6.83847547e-01 4.70189936e-02 1.64618180e-03 2.85873115e-01 8.02850366e-01 -2.09902555e-01 9.39482927e-01 8.00849497e-01 -6.87498152e-01 -9.81705427e-01 -1.32493508e+00 -2.87584156e-01 7.09212780e-01 1.68385338e-02 -2.28701293e-01 -8.71310115e-01 -8.72160554e-01 1.73366219e-01 4.88339663e-01 -1.01163518e+00 -1.53882978e-02 -4.72613096e-01 -9.78819251e-01 6.56432211e-01 1.01791608e+00 1.06620264e+00 -5.91116250e-01 -2.16938322e-04 1.54071614e-01 5.60927689e-02 -1.36886215e+00 -3.38091433e-01 2.52825111e-01 -1.12172151e+00 -1.21525049e+00 -9.97840464e-01 -1.20330179e+00 8.09425414e-01 1.54234260e-01 1.35659528e+00 1.55472204e-01 -3.08673650e-01 -8.26113224e-02 -1.19896822e-01 6.68150485e-02 6.17361926e-02 4.85307604e-01 -4.93674338e-01 -1.57316834e-01 2.94175148e-01 -5.49630105e-01 -8.37291420e-01 9.47118178e-02 -3.85461032e-01 1.78031042e-01 4.50308174e-01 9.17355478e-01 8.67850721e-01 -2.10944653e-01 4.05198395e-01 -1.13043392e+00 9.05553028e-02 -3.53677362e-01 -7.67961919e-01 9.59378704e-02 -7.34455943e-01 6.89223036e-02 5.55809677e-01 -3.68147194e-01 -4.24728185e-01 1.61192819e-01 8.06169286e-02 -4.40624684e-01 5.73285446e-02 8.33223015e-02 3.08542281e-01 -5.33713818e-01 8.30064178e-01 -2.29361981e-01 -2.79083610e-01 -2.39216372e-01 2.75100589e-01 1.30056858e-01 7.17703819e-01 2.38384325e-02 5.26280761e-01 5.40763140e-01 2.87906677e-01 -4.65213090e-01 -9.31013703e-01 -2.39485770e-01 -9.16734695e-01 4.79873754e-02 1.05020452e+00 -8.91857028e-01 -4.67385054e-01 6.44406796e-01 -8.50546241e-01 -8.95157039e-01 -4.88952771e-02 4.15391833e-01 -2.08597451e-01 5.35100289e-02 -9.11008954e-01 -2.08920643e-01 -3.74488682e-01 -1.00776887e+00 1.07037306e+00 9.87046510e-02 -5.41339256e-02 -1.25758827e+00 -4.26317066e-01 3.21262091e-01 3.04954380e-01 6.16343498e-01 8.45977485e-01 -7.24504590e-01 -8.15869927e-01 -4.44319785e-01 -7.65868485e-01 4.51460034e-01 -7.80832469e-02 8.43838677e-02 -1.06184161e+00 -3.13208640e-01 -7.65813172e-01 -4.34015453e-01 1.55676329e+00 6.99510932e-01 1.38871109e+00 -1.24423653e-01 -6.15745842e-01 9.48444426e-01 1.62454331e+00 -3.06113213e-01 5.03737748e-01 9.68882442e-02 1.03735542e+00 -4.00822498e-02 -8.15435946e-02 -3.49874705e-01 3.91559839e-01 3.11321110e-01 5.71544766e-01 -3.25966418e-01 -7.67009437e-01 -1.35166898e-01 -1.98245928e-01 1.25606939e-01 -1.58458486e-01 -4.66370285e-01 -1.07498932e+00 7.05291092e-01 -1.68290281e+00 -6.91688776e-01 -5.05201161e-01 2.15551782e+00 3.21320891e-01 7.25763261e-01 -6.65790141e-02 5.59430867e-02 5.58012605e-01 2.87555397e-01 -6.37654901e-01 -1.65952563e-01 -3.10583770e-01 3.64696890e-01 1.07847965e+00 6.89527392e-01 -1.53963137e+00 1.10245407e+00 6.74434376e+00 6.68881178e-01 -1.12718606e+00 -5.70216440e-02 1.32298541e+00 -1.94793463e-01 -1.23578303e-01 -3.87759626e-01 -6.68572068e-01 2.45669588e-01 6.72096491e-01 2.35779330e-01 3.64950716e-01 6.59462154e-01 -2.69405365e-01 -1.27635049e-02 -1.26829612e+00 1.04610598e+00 9.74575356e-02 -1.68624008e+00 6.28049066e-03 7.81205967e-02 1.15355051e+00 8.59958529e-01 3.05906951e-01 4.30433974e-02 7.55914867e-01 -1.55339432e+00 4.47608739e-01 1.20675094e-01 9.45161819e-01 -5.58461368e-01 7.50774205e-01 8.66475180e-02 -1.15116572e+00 -1.91553812e-02 -3.83035809e-01 -1.38429925e-01 -3.43864650e-01 6.49294138e-01 -8.63375783e-01 1.11317106e-01 8.96753907e-01 8.91833007e-01 -9.88537908e-01 1.29379833e+00 -3.41900557e-01 7.15037882e-01 -5.40132046e-01 4.12758905e-03 6.65452361e-01 -6.98845759e-02 1.86235204e-01 1.17078972e+00 -1.09963175e-02 -8.47391859e-02 2.27656867e-02 1.00377274e+00 -7.62606502e-01 -2.05935031e-01 -6.07214332e-01 8.93023610e-02 3.36103261e-01 1.47806716e+00 -1.00623345e+00 -3.58858705e-01 -3.68109614e-01 9.93739426e-01 8.46971452e-01 6.00506067e-01 -8.68332565e-01 -3.75700533e-01 2.81015575e-01 3.79449353e-02 4.13866758e-01 -2.53082156e-01 -7.07257986e-01 -7.95167804e-01 1.19880196e-02 -3.78885359e-01 5.68302095e-01 -7.65853882e-01 -1.00480914e+00 8.26044679e-01 -3.22324306e-01 -7.88505733e-01 -8.98595303e-02 -9.97056842e-01 -7.10984707e-01 7.96460867e-01 -1.52746129e+00 -1.13416529e+00 -4.65474010e-01 2.42268831e-01 8.42340142e-02 1.18671991e-01 6.11667991e-01 2.65675396e-01 -5.59179723e-01 9.72719848e-01 6.67046383e-02 8.18377912e-01 3.90133172e-01 -1.52951515e+00 9.45990324e-01 7.16663897e-01 5.23539662e-01 1.76253855e-01 2.04216197e-01 -4.25546110e-01 -8.02055418e-01 -1.43696499e+00 8.40049028e-01 -5.91815650e-01 7.46540248e-01 -3.89718473e-01 -8.47298205e-01 1.01391757e+00 -6.95224060e-03 6.89454973e-01 2.68817753e-01 4.03122641e-02 -7.71939993e-01 -2.78467059e-01 -1.17111731e+00 4.22958136e-01 1.42143512e+00 -2.39719331e-01 -2.20827401e-01 6.62019730e-01 6.89012408e-01 -4.94678140e-01 -9.88170445e-01 4.99636531e-01 4.40988392e-01 -8.03364158e-01 9.38595831e-01 -6.43091261e-01 2.90397376e-01 -1.14390023e-01 8.94176066e-02 -1.31000912e+00 -8.25315118e-01 -1.96208507e-01 7.66911656e-02 8.02889884e-01 9.76235807e-01 -6.50070846e-01 1.13981247e+00 2.34852821e-01 -9.96093675e-02 -8.98988247e-01 -9.16823268e-01 -8.12755704e-01 3.40050340e-01 -2.77616471e-01 9.75776389e-02 8.36183071e-01 -3.94382507e-01 4.05824512e-01 3.22203636e-01 1.69205099e-01 8.68888617e-01 9.88488346e-02 3.39371115e-01 -1.28417361e+00 -3.06213528e-01 -8.67390096e-01 -8.03872764e-01 -9.92405474e-01 1.83829218e-01 -1.46578944e+00 -3.71291190e-01 -1.89554858e+00 1.90565243e-01 -3.63758862e-01 -5.47448814e-01 7.30103195e-01 -1.33470729e-01 1.08953071e+00 -1.23378955e-01 -1.00545071e-01 -6.79317355e-01 -1.31065935e-01 1.31849921e+00 -4.61941600e-01 1.33145198e-01 -1.94992945e-01 -7.14759111e-01 7.85363913e-01 9.37207460e-01 -1.13383301e-01 -4.27002758e-01 -1.09033287e+00 1.64454654e-01 -5.28700173e-01 7.06163824e-01 -1.27545905e+00 1.09165244e-01 3.57604861e-01 8.83281887e-01 -2.23715216e-01 1.15139410e-01 -4.04123008e-01 -2.03646168e-01 5.93603194e-01 -5.20663202e-01 -1.29462972e-01 3.27237129e-01 5.05087137e-01 9.87807438e-02 2.49329761e-01 9.94824350e-01 -8.52434188e-02 -8.20606291e-01 6.94334745e-01 9.41354632e-02 2.80638576e-01 1.13522327e+00 -2.19536632e-01 -7.69793868e-01 -1.36333212e-01 -7.20490456e-01 4.51695800e-01 2.95126438e-01 2.80727625e-01 2.39001423e-01 -1.22830367e+00 -8.29854548e-01 1.37780577e-01 5.24894707e-02 8.77981186e-02 -2.01009512e-02 7.56596923e-01 -8.75316918e-01 4.65720743e-01 -1.56151086e-01 -8.11141014e-01 -1.07050681e+00 3.24707836e-01 8.86753142e-01 -1.18787400e-01 -7.17595160e-01 1.49554718e+00 1.78278163e-01 -7.37086758e-02 2.91244954e-01 -6.67578220e-01 8.45280662e-03 -2.27454677e-01 3.58664274e-01 4.60733116e-01 3.86526138e-01 -5.30248523e-01 -3.90022755e-01 5.56358397e-01 -4.33520004e-02 2.69356430e-01 1.25389898e+00 3.82745355e-01 4.10383381e-03 3.03847820e-01 1.51173413e+00 -3.56704980e-01 -1.46394980e+00 -2.46634871e-01 -8.39330032e-02 -6.78711310e-02 4.71823275e-01 -9.73714650e-01 -1.54122818e+00 9.21735823e-01 5.91030538e-01 2.52091646e-01 7.13948309e-01 2.76197076e-01 5.89230239e-01 4.53684568e-01 1.81477442e-01 -1.01823556e+00 -9.09536779e-02 3.90935242e-01 6.91334963e-01 -1.35411263e+00 4.32125330e-02 -5.21591425e-01 -2.31748238e-01 9.30265248e-01 7.43210614e-01 -8.10488760e-01 7.74914861e-01 4.25096542e-01 -4.68714200e-02 -2.40038499e-01 -5.93993425e-01 -1.66011766e-01 6.62826598e-01 6.54234707e-01 5.92705965e-01 2.62902349e-01 -7.57381460e-03 2.53944159e-01 -2.83673704e-01 -1.14669487e-01 2.04047695e-01 3.09072286e-01 -4.68207747e-01 -5.18388510e-01 2.44042471e-01 9.07447875e-01 -4.11150515e-01 -4.61345196e-01 -7.00301468e-01 1.01140499e+00 6.17975183e-03 5.78176260e-01 7.48262882e-01 -3.21959168e-01 9.80764534e-03 1.13587141e-01 6.14520490e-01 -5.43719828e-01 -5.08929133e-01 -1.63674206e-01 8.57223198e-02 -7.02334821e-01 -2.47942790e-01 -3.26941192e-01 -1.46525204e+00 -2.68651992e-01 -2.52574027e-01 -2.07248524e-01 2.22945184e-01 6.87869668e-01 4.09993827e-01 7.03561723e-01 2.63623297e-01 -6.88514411e-01 -1.91767395e-01 -8.78537536e-01 -5.28842747e-01 2.50223190e-01 3.23191702e-01 -2.89423734e-01 -2.02225313e-01 -3.13614219e-01]
[9.553826332092285, 1.1225996017456055]
50c32550-27d1-482d-aea3-577030681431
a-simple-attempt-for-3d-occupancy-estimation
2303.10076
null
https://arxiv.org/abs/2303.10076v2
https://arxiv.org/pdf/2303.10076v2.pdf
A Simple Attempt for 3D Occupancy Estimation in Autonomous Driving
The task of estimating 3D occupancy from surrounding view images is an exciting development in the field of autonomous driving, following the success of Birds Eye View (BEV) perception.This task provides crucial 3D attributes of the driving environment, enhancing the overall understanding and perception of the surrounding space. However, there is still a lack of a baseline to define the task, such as network design, optimization, and evaluation. In this work, we present a simple attempt for 3D occupancy estimation, which is a CNN-based framework designed to reveal several key factors for 3D occupancy estimation. In addition, we explore the relationship between 3D occupancy estimation and other related tasks, such as monocular depth estimation, stereo matching, and BEV perception (3D object detection and map segmentation), which could advance the study on 3D occupancy estimation. For evaluation, we propose a simple sampling strategy to define the metric for occupancy evaluation, which is flexible for current public datasets. Moreover, we establish a new benchmark in terms of the depth estimation metric, where we compare our proposed method with monocular depth estimation methods on the DDAD and Nuscenes datasets.The relevant code will be available in https://github.com/GANWANSHUI/SimpleOccupancy
['Naoto Yokoya', 'Hongbin Xu', 'Ningkai Mo', 'Wanshui Gan']
2023-03-17
null
null
null
null
['stereo-matching-1']
['computer-vision']
[-1.24029882e-01 -2.78645843e-01 -1.31766647e-01 -5.15280068e-01 -1.56399041e-01 -4.76586878e-01 5.98393261e-01 -1.60145342e-01 -6.18319988e-01 4.88891363e-01 2.06977576e-01 -2.13748515e-01 1.33688614e-01 -8.68218958e-01 -7.66684473e-01 -7.44163692e-01 1.52042732e-01 1.44785181e-01 3.20493042e-01 -3.05387348e-01 3.31650048e-01 5.62126815e-01 -2.32487035e+00 -2.59415060e-01 5.99859059e-01 1.20305133e+00 5.33693850e-01 5.34438729e-01 2.52064556e-01 5.10566354e-01 -3.01922679e-01 5.03602587e-02 3.42674494e-01 1.66099872e-02 -4.51307356e-01 -1.05234608e-01 6.23213947e-01 -6.37577236e-01 -6.20869935e-01 7.66666114e-01 7.84787893e-01 2.81323731e-01 6.61337554e-01 -1.48349690e+00 -6.71072081e-02 8.88290927e-02 -4.05454904e-01 4.70346451e-01 1.71059623e-01 3.42369765e-01 8.25604081e-01 -8.52334201e-01 3.15287977e-01 1.03991961e+00 3.65647197e-01 3.64333928e-01 -6.56333327e-01 -7.95794070e-01 3.30321848e-01 4.09509301e-01 -1.63415384e+00 -4.50608253e-01 5.78340054e-01 -6.69572115e-01 9.42554474e-01 4.46520537e-01 9.16290104e-01 9.77231085e-01 2.57398725e-01 9.73411441e-01 1.20527148e+00 -2.71273032e-03 3.41316640e-01 1.04048103e-01 1.22492090e-01 4.66682792e-01 1.46059915e-01 5.70468068e-01 -7.07589805e-01 3.33208472e-01 5.58096528e-01 3.41601064e-03 -2.52709657e-01 -6.63831294e-01 -1.12255478e+00 7.97145903e-01 5.82337320e-01 -1.12566568e-01 1.61096510e-02 2.74099827e-01 2.08271056e-01 -8.27848762e-02 6.31693661e-01 2.42241174e-01 -2.21764207e-01 -2.29227260e-01 -6.54150188e-01 5.86371541e-01 4.24343646e-01 9.59088445e-01 1.20185184e+00 -2.20169693e-01 -1.75316095e-01 5.50292790e-01 3.79147977e-01 7.16440856e-01 9.50901583e-02 -1.06655586e+00 4.20786083e-01 4.93290156e-01 2.39861727e-01 -7.39771068e-01 -5.78827858e-01 -4.15933281e-01 -7.36648500e-01 3.46642315e-01 2.07757041e-01 1.09424375e-01 -8.36625516e-01 1.58271980e+00 5.59543610e-01 8.23624358e-02 -1.31908879e-01 1.20422459e+00 1.17694831e+00 5.04001379e-01 -4.32103395e-01 3.26904565e-01 1.28087997e+00 -1.05904567e+00 -4.51047748e-01 -6.19688511e-01 5.23181796e-01 -2.66897053e-01 9.18781281e-01 1.31139089e-03 -6.75644279e-01 -7.68729568e-01 -1.21760225e+00 -4.03430015e-01 -6.20245457e-01 -5.46484813e-02 6.14191592e-01 6.17825389e-01 -1.03902352e+00 1.57284718e-02 -7.72501469e-01 -3.59787136e-01 3.32439154e-01 9.74879712e-02 -2.60022640e-01 -2.20632747e-01 -1.35962558e+00 1.07505524e+00 4.59006965e-01 3.14597487e-01 -1.15287471e+00 -4.36115772e-01 -1.21889293e+00 -2.51112401e-01 4.09530103e-01 -7.14236736e-01 1.11362267e+00 -1.10472143e-01 -1.08565104e+00 1.01993740e+00 -3.53867531e-01 -3.57380599e-01 4.45031285e-01 -8.93328190e-02 1.02023177e-01 -4.08998489e-01 2.43160233e-01 1.18616188e+00 4.25209641e-01 -1.17477345e+00 -9.58850682e-01 -6.48979723e-01 4.77613926e-01 6.62649810e-01 1.33786201e-01 -5.55925190e-01 -5.75210690e-01 8.16384181e-02 2.03031197e-01 -9.14663136e-01 -3.70542854e-01 6.84053898e-02 -2.67994851e-01 -2.09558934e-01 3.31057221e-01 -2.22070917e-01 9.91174161e-01 -2.43099093e+00 1.30667225e-01 -2.44531445e-02 5.77218890e-01 -2.42093518e-01 2.70174086e-01 3.53572726e-01 4.00611401e-01 -4.66770045e-02 -2.98696816e-01 -4.65953380e-01 3.51193160e-01 1.65713117e-01 -1.06055029e-01 6.96406901e-01 1.97882131e-02 9.02964413e-01 -8.69219899e-01 -3.02914709e-01 7.70531774e-01 4.82437849e-01 -4.82714862e-01 2.93842167e-01 2.88243890e-02 5.26700497e-01 -3.82577658e-01 6.99676871e-01 9.51535165e-01 1.70672834e-01 -4.01018590e-01 -5.72413718e-03 -6.90012157e-01 2.71011740e-01 -1.09606063e+00 1.59779000e+00 -5.00461459e-01 1.07087803e+00 -1.08671248e-01 -6.38592601e-01 1.07769704e+00 -2.00013176e-01 3.38923693e-01 -9.78328288e-01 3.47913414e-01 1.97131589e-01 -2.94169128e-01 -3.88906956e-01 8.99090469e-01 3.29861432e-01 -2.13481396e-01 1.18999526e-01 -3.22908223e-01 -4.53076750e-01 5.63093945e-02 -1.99418679e-01 7.84176409e-01 1.01795472e-01 4.06953543e-01 -2.73467869e-01 2.83290476e-01 -5.11690825e-02 3.15232217e-01 7.28570282e-01 -7.11622775e-01 6.52606428e-01 3.19579303e-01 -6.95014298e-01 -1.02328050e+00 -1.03048241e+00 -5.98659575e-01 8.49542081e-01 9.11859572e-01 -2.85389572e-01 -5.74909747e-01 -2.54442304e-01 2.16727436e-01 4.25338775e-01 -7.98848093e-01 -7.08968146e-03 -1.71701744e-01 -7.56315708e-01 3.31273347e-01 5.67759037e-01 8.14021409e-01 -8.35030019e-01 -1.08339632e+00 -1.96233913e-01 -4.51852560e-01 -1.31807148e+00 -2.64897197e-01 5.21737933e-01 -3.72174710e-01 -9.26845014e-01 -4.49568748e-01 -4.90645945e-01 2.31041119e-01 7.66602516e-01 1.16880679e+00 -9.30279028e-03 -1.58671692e-01 1.46797255e-01 -2.77741849e-01 -6.72758698e-01 2.00935811e-01 4.59109604e-01 -2.06334498e-02 -3.11540216e-01 7.78813720e-01 -5.19329190e-01 -9.77328181e-01 7.10533202e-01 -5.89371681e-01 4.49148089e-01 3.88117313e-01 2.72728503e-01 7.83920765e-01 -6.83055371e-02 5.55661460e-03 -3.15960377e-01 1.13520451e-01 -5.68381250e-01 -8.22817624e-01 -3.90355974e-01 -3.43349367e-01 -2.20978677e-01 7.99399689e-02 1.48898020e-01 -5.89752614e-01 2.53051817e-01 -4.73016262e-01 -2.98587054e-01 -5.15641570e-01 2.24960774e-01 -2.29777634e-01 -8.65983889e-02 6.56572104e-01 3.03575903e-01 -1.65167809e-01 -1.74674124e-01 2.77977943e-01 8.93078387e-01 4.34807360e-01 -9.38021317e-02 6.69504464e-01 8.51208448e-01 -1.27221376e-01 -8.44364703e-01 -9.73160446e-01 -9.30147231e-01 -7.72931874e-01 -4.63494331e-01 1.06459367e+00 -1.35609996e+00 -8.31967473e-01 5.59827566e-01 -1.00856960e+00 -4.77623403e-01 -2.35819761e-02 4.95478302e-01 -7.55887151e-01 4.93913554e-02 -9.85151529e-02 -9.67286110e-01 4.48371246e-02 -1.46161830e+00 1.48987150e+00 2.20077097e-01 3.51333320e-02 -7.20898449e-01 2.34173298e-01 6.19044065e-01 3.82458538e-01 3.15356314e-01 3.23522091e-01 1.68512180e-01 -1.01961863e+00 6.53866753e-02 -2.57640988e-01 1.89744651e-01 -2.12638050e-01 -2.34666303e-01 -1.48760366e+00 -1.29024446e-01 -8.97465572e-02 -3.24220061e-01 1.24029040e+00 6.52125418e-01 1.27888954e+00 3.95676434e-01 -4.35549170e-01 9.90051568e-01 1.28100145e+00 1.08172871e-01 8.17794681e-01 5.74472010e-01 7.65573382e-01 7.66808629e-01 8.38069499e-01 5.58854938e-01 9.32117105e-01 8.36659670e-01 1.04848182e+00 -3.67513090e-01 7.76553601e-02 -3.45664620e-01 8.47979188e-02 4.67463851e-01 -1.59389645e-01 -4.20869857e-01 -9.85286176e-01 5.08704722e-01 -1.86830449e+00 -8.04998517e-01 2.47122254e-03 2.02605152e+00 1.42477542e-01 1.49905011e-01 7.07856268e-02 -2.14353222e-02 5.27408600e-01 5.88947237e-01 -8.65669072e-01 -1.04562432e-01 -1.18962981e-01 -2.86511511e-01 8.30636084e-01 6.87262774e-01 -1.37390304e+00 9.45301235e-01 6.08151627e+00 7.76874959e-01 -1.00366330e+00 -9.63683054e-02 6.97877169e-01 -2.35765412e-01 -3.99052173e-01 -3.01261902e-01 -1.37918472e+00 3.50417793e-01 6.02878094e-01 1.97836205e-01 5.07612228e-01 9.98967111e-01 4.30294633e-01 -5.24715781e-01 -1.07659340e+00 1.14128077e+00 1.53676301e-01 -1.08449519e+00 -4.55890805e-01 4.71691191e-01 6.14211679e-01 5.15439153e-01 1.09400839e-01 3.20211858e-01 -5.23217656e-02 -1.09994698e+00 7.70791411e-01 3.31137180e-01 7.15765953e-01 -7.46020675e-01 9.11014736e-01 7.14180708e-01 -1.42829978e+00 2.54553780e-02 -6.84906006e-01 -3.86917531e-01 1.61322542e-02 4.06390905e-01 -5.85786879e-01 2.91195482e-01 1.10115302e+00 9.01060581e-01 -5.65346539e-01 1.06631184e+00 1.04752947e-02 -3.18813249e-02 -2.78894901e-01 -1.76224411e-01 3.31919044e-01 -1.20372534e-01 2.97650546e-01 8.65943193e-01 2.73372918e-01 -1.35156691e-01 1.41612619e-01 1.06592464e+00 1.45627037e-01 -4.42030430e-02 -1.10248959e+00 3.70939732e-01 4.99347657e-01 1.25808179e+00 -5.74907243e-01 2.59688664e-02 -3.71717721e-01 5.20716906e-01 3.40717226e-01 1.59618795e-01 -1.02354395e+00 -1.21810868e-01 1.08584285e+00 3.00287962e-01 3.03939849e-01 -4.79255855e-01 -1.98735237e-01 -9.60105896e-01 3.49756181e-02 -4.52472836e-01 -2.78112620e-01 -8.68804216e-01 -7.87787139e-01 5.85817456e-01 3.66954833e-01 -1.32516491e+00 -1.05830297e-01 -6.57304883e-01 -2.12662637e-01 8.51701677e-01 -1.99156010e+00 -6.97853744e-01 -1.09514439e+00 2.08904028e-01 5.75676262e-01 1.31866246e-01 3.62317711e-01 3.83839101e-01 -4.84940350e-01 2.26692408e-01 1.27912899e-02 -2.53316700e-01 3.87146473e-01 -1.13727570e+00 8.92902792e-01 5.93570292e-01 -9.47353616e-02 -7.55731463e-02 5.76467335e-01 -1.83394343e-01 -1.21750259e+00 -1.14206231e+00 6.00447476e-01 -9.18682933e-01 1.05274394e-01 -8.59842837e-01 -5.43877423e-01 3.55899066e-01 -5.53921014e-02 -8.70404765e-02 5.50223291e-01 -2.91452017e-02 6.48383200e-02 -7.67966956e-02 -7.92708457e-01 5.57595432e-01 1.56271374e+00 -5.03075540e-01 2.40297131e-02 1.82610154e-01 8.76978219e-01 -8.06015193e-01 -5.70378840e-01 6.55336678e-01 7.48906851e-01 -1.55799758e+00 9.03159916e-01 2.14816034e-01 3.88279945e-01 -5.96099317e-01 -6.08335674e-01 -1.13360488e+00 -4.02594805e-01 7.62642398e-02 4.90988307e-02 7.07938313e-01 3.10575128e-01 -6.57315731e-01 1.00424945e+00 2.35904038e-01 -5.42967558e-01 -9.29322124e-01 -1.03265369e+00 -5.16756296e-01 -2.28866041e-01 -7.14883506e-01 8.18798602e-01 4.60583180e-01 -4.25334305e-01 3.38577211e-01 -3.33611906e-01 3.85204196e-01 4.47774202e-01 8.95633996e-02 1.14948428e+00 -1.03569996e+00 1.50373861e-01 -3.28650147e-01 -7.51113296e-01 -1.62237203e+00 2.19136760e-01 -7.10024774e-01 3.72905582e-01 -1.69806230e+00 1.65206164e-01 -3.02969337e-01 -1.62596449e-01 1.74474679e-02 -8.14597011e-02 4.29409653e-01 -1.08954079e-01 1.78760037e-01 -6.05002999e-01 9.74154949e-01 1.46239316e+00 -1.34155214e-01 -6.42978847e-02 6.13754429e-02 -5.16292214e-01 4.73797888e-01 1.03266001e+00 -1.31481513e-01 -5.76315105e-01 -4.39113081e-01 2.60008365e-01 -4.96193059e-02 5.78005075e-01 -1.17564499e+00 1.27724439e-01 -2.43936464e-01 2.63175368e-01 -1.31667566e+00 7.45588005e-01 -7.10146725e-01 -1.42391650e-02 2.28727743e-01 7.47295469e-02 8.92284326e-03 1.47223338e-01 3.91046286e-01 -2.47514978e-01 -6.70771599e-02 5.30341685e-01 -2.49126881e-01 -1.32506859e+00 6.57963276e-01 -2.75638491e-01 2.31323130e-02 9.99814034e-01 -7.02124178e-01 -4.72952902e-01 -2.91070521e-01 -2.23917410e-01 6.52408123e-01 6.77452683e-01 5.18685222e-01 6.34569645e-01 -1.31075716e+00 -6.58728600e-01 5.17131865e-01 6.28681839e-01 4.75698531e-01 4.59940314e-01 7.31189311e-01 -6.53873086e-01 6.32474363e-01 -2.40926757e-01 -1.06850040e+00 -8.71285200e-01 3.09625208e-01 7.17121303e-01 2.72301376e-01 -4.17415887e-01 7.31092095e-01 8.63660693e-01 -7.54900634e-01 3.80082488e-01 -5.20306468e-01 -4.24234062e-01 -1.20980009e-01 4.60433424e-01 2.83558875e-01 1.93340965e-02 -7.22175539e-01 -3.72746140e-01 5.69972456e-01 4.16226268e-01 5.87989241e-02 1.04445827e+00 -6.53483093e-01 1.17053054e-01 6.81033373e-01 1.02044976e+00 -4.58839536e-01 -1.66357470e+00 -7.98643529e-02 -5.43364763e-01 -7.46114314e-01 4.10567492e-01 -1.94824472e-01 -8.27024221e-01 1.09906673e+00 1.01618028e+00 1.79648235e-01 8.04346859e-01 2.68747777e-01 5.70868373e-01 3.12008202e-01 5.61976850e-01 -9.94589865e-01 -1.18133470e-01 7.63283074e-01 8.04011941e-01 -1.72897315e+00 -2.01934040e-01 -4.29061651e-01 -4.54267234e-01 6.44765198e-01 1.04773724e+00 2.06085056e-01 7.65335798e-01 1.85125127e-01 3.39682102e-02 -4.56194311e-01 -6.22926950e-01 -7.52813756e-01 2.31398150e-01 6.98028505e-01 2.38909304e-01 2.66164392e-01 2.21791029e-01 9.18138772e-02 -6.67329073e-01 -3.32345605e-01 3.33206296e-01 6.08306468e-01 -8.03003132e-01 -4.63896096e-01 -1.84425220e-01 3.25812638e-01 1.70119092e-01 -8.75867084e-02 -2.09044442e-01 8.62703145e-01 5.73802590e-01 7.92584479e-01 3.95929873e-01 -6.92770958e-01 4.99737412e-01 -3.70004296e-01 4.25454408e-01 -6.33953929e-01 2.95018088e-02 -1.93161353e-01 6.00997843e-02 -6.10713959e-01 -5.14555812e-01 -4.96540457e-01 -6.68279767e-01 -6.42145634e-01 -3.64562511e-01 -2.68157601e-01 8.89651418e-01 9.03709829e-01 1.71402752e-01 4.76598501e-01 8.04132164e-01 -1.26913249e+00 -2.95713767e-02 -7.68387318e-01 -7.53026903e-01 -1.10484548e-01 4.94105488e-01 -1.09037161e+00 -6.06164873e-01 -5.37316620e-01]
[8.161608695983887, -2.4065141677856445]
ec22ec02-5314-4bf6-8575-1a29972e3787
automated-detection-of-oral-pre-cancerous
1909.08987
null
https://arxiv.org/abs/1909.08987v1
https://arxiv.org/pdf/1909.08987v1.pdf
Automated detection of oral pre-cancerous tongue lesions using deep learning for early diagnosis of oral cavity cancer
Discovering oral cavity cancer (OCC) at an early stage is an effective way to increase patient survival rate. However, current initial screening process is done manually and is expensive for the average individual, especially in developing countries worldwide. This problem is further compounded due to the lack of specialists in such areas. Automating the initial screening process using artificial intelligence (AI) to detect pre-cancerous lesions can prove to be an effective and inexpensive technique that would allow patients to be triaged accordingly to receive appropriate clinical management. In this study, we have applied and evaluated the efficacy of six deep convolutional neural network (DCNN) models using transfer learning, for identifying pre-cancerous tongue lesions directly using a small data set of clinically annotated photographic images to diagnose early signs of OCC. DCNN model based on Vgg19 architecture was able to differentiate between benign and pre-cancerous tongue lesions with a mean classification accuracy of 0.98, sensitivity 0.89 and specificity 0.97. Additionally, the ResNet50 DCNN model was able to distinguish between five types of tongue lesions i.e. hairy tongue, fissured tongue, geographic tongue, strawberry tongue and oral hairy leukoplakia with a mean classification accuracy of 0.97. Preliminary results using an (AI+Physician) ensemble model demonstrate that an automated initial screening process of tongue lesions using DCNNs can achieve near-human level classification performance for diagnosing early signs of OCC in patients.
['Mohammed Usman', 'Syed Ali', 'Mohammad Shiblee', 'Mohammed Zubair M. Shamim', 'Sadatullah Syed']
2019-09-18
null
null
null
null
['small-data']
['computer-vision']
[ 1.65045068e-01 2.85419524e-01 -4.33772683e-01 3.52566987e-02 -1.02321386e+00 -5.00350356e-01 1.14050023e-01 2.89580673e-01 -3.74709755e-01 6.09646678e-01 1.57160774e-01 -8.10178220e-01 -1.28505439e-01 -7.22553849e-01 -9.64220706e-03 -9.60797727e-01 1.84419289e-01 8.60925317e-01 9.45150945e-03 -1.70904607e-01 -3.08747441e-02 5.16218781e-01 -1.32467413e+00 3.43585044e-01 1.05847824e+00 7.83067584e-01 2.86156029e-01 9.41031218e-01 -3.62965353e-02 6.51511431e-01 -5.50139606e-01 -1.36426672e-01 -1.10085867e-01 -3.51220936e-01 -8.25975120e-01 8.94505829e-02 4.51721609e-01 -3.07776034e-01 1.81347132e-01 9.75204110e-01 6.72308922e-01 -5.37037373e-01 8.48634124e-01 -7.08148479e-01 -5.65408826e-01 1.29418269e-01 -3.19102079e-01 8.70830342e-02 1.91777200e-01 3.35832030e-01 6.73742115e-01 -4.23302621e-01 7.41660714e-01 5.66686213e-01 1.13536358e+00 8.47233713e-01 -6.64039612e-01 -3.01955700e-01 -4.03624922e-01 1.33805111e-01 -1.03613842e+00 -1.42869100e-01 7.98695982e-02 -2.11778536e-01 9.03634846e-01 3.93793643e-01 8.08003604e-01 7.28274941e-01 3.25224280e-01 6.72444999e-01 1.21083069e+00 -7.23366797e-01 3.50814968e-01 4.07369286e-01 3.45597953e-01 8.60174358e-01 5.10469675e-01 1.55759439e-01 9.83047485e-02 -1.88922938e-02 5.17101765e-01 -5.01894718e-03 -1.01329230e-01 6.26314163e-01 -4.05323267e-01 1.05217946e+00 6.90606654e-01 6.86485410e-01 -3.26084584e-01 -1.50870770e-01 4.09201115e-01 3.25497866e-01 3.32965881e-01 1.49857998e-01 -3.90684426e-01 4.60513532e-01 -5.69150984e-01 -4.45722789e-01 5.94046056e-01 2.80930966e-01 9.12842527e-02 -1.71622515e-01 2.85873022e-02 9.49197769e-01 6.33620143e-01 2.55303025e-01 1.32610035e+00 -4.81032699e-01 -3.75494182e-01 1.09318984e+00 -4.68117237e-01 -1.83346763e-01 -9.24289465e-01 -3.09592307e-01 -6.92160904e-01 5.07200956e-01 5.07838011e-01 -2.73987830e-01 -1.81893969e+00 1.04364777e+00 3.21152806e-01 -3.20709616e-01 2.85386115e-01 7.22521901e-01 9.40902352e-01 2.13830426e-01 4.67173427e-01 1.70497820e-01 1.88751304e+00 -7.00273275e-01 -4.64103609e-01 -8.41395557e-02 1.13747394e+00 -6.40428305e-01 8.38522375e-01 2.38436252e-01 -7.01385260e-01 -6.28632605e-02 -8.09932470e-01 -1.42140135e-01 -7.71951497e-01 5.37255585e-01 8.43701720e-01 1.22211146e+00 -1.53073668e+00 1.02551617e-01 -9.54510093e-01 -1.09980726e+00 7.37060189e-01 9.67864573e-01 -5.49108148e-01 -1.42778680e-01 -7.68635273e-01 1.03999424e+00 2.66676992e-01 3.50929680e-03 -5.36994934e-01 -6.07595026e-01 -6.78695560e-01 -1.95817947e-01 -1.09860264e-01 -8.58282328e-01 1.42516088e+00 -1.26597452e+00 -1.69343460e+00 1.10708010e+00 -1.11403510e-01 -3.99020940e-01 4.71159458e-01 4.11100179e-01 -5.32979667e-01 2.31223270e-01 -3.99769749e-04 7.17819273e-01 3.20311069e-01 -9.56075251e-01 -1.09750772e+00 -7.55389392e-01 -3.77224296e-01 2.46287078e-01 -3.01340729e-01 -1.44920841e-01 -3.39435458e-01 -1.28209487e-01 5.01909517e-02 -1.22955668e+00 -4.22299951e-01 3.65131646e-01 -4.75927532e-01 -4.50383961e-01 1.01465321e+00 -1.00653708e+00 5.66015065e-01 -1.97096205e+00 -5.23986757e-01 4.23481792e-01 2.02918679e-01 6.82379961e-01 3.97339687e-02 -1.22451140e-02 1.97346792e-01 4.36701447e-01 3.87247838e-02 -1.94598511e-01 -4.71464515e-01 2.84137070e-01 7.60979712e-01 5.51383674e-01 1.80810511e-01 9.50816154e-01 -6.59884930e-01 -6.67504549e-01 3.03522289e-01 8.07439864e-01 -1.71833619e-01 -4.49794158e-02 -2.87052274e-01 2.04699546e-01 -3.97383302e-01 1.30109203e+00 3.66499364e-01 -4.50877041e-01 2.12480739e-01 -3.70663516e-02 1.74239323e-01 -2.08657593e-01 -5.12157381e-01 9.39167857e-01 -5.86033702e-01 8.09013009e-01 2.25491077e-01 -3.24538112e-01 3.86067301e-01 7.81641960e-01 2.54646480e-01 -7.35087872e-01 5.58320403e-01 6.40884757e-01 3.13017517e-01 -7.64160156e-01 4.69374983e-03 -4.50249761e-01 4.25626755e-01 8.48154873e-02 -1.02994308e-01 -4.28719399e-03 1.62299424e-01 -2.28658855e-01 1.01180744e+00 -3.92398626e-01 5.80203712e-01 -1.96923658e-01 5.99187315e-01 4.33068544e-01 1.60053656e-01 2.44450599e-01 -5.43145478e-01 6.05658233e-01 1.11252308e-01 -5.51122069e-01 -4.86946553e-01 -8.59665036e-01 -4.14644748e-01 6.88602209e-01 -3.43390673e-01 4.78711784e-01 -6.78241074e-01 -7.57292390e-01 -1.08873941e-01 5.28702617e-01 -9.55090642e-01 -8.90684277e-02 -2.91823179e-01 -1.08925092e+00 5.92717767e-01 5.98995149e-01 4.50045794e-01 -1.02118969e+00 -3.23463053e-01 3.40438992e-01 1.05978742e-01 -7.82350004e-01 -2.36910224e-01 3.94776821e-01 -8.90596211e-01 -1.24635911e+00 -1.20242667e+00 -1.47110653e+00 1.12945533e+00 -1.03241265e-01 6.82249963e-01 6.07774913e-01 -7.46623516e-01 -3.76576595e-02 -2.27567911e-01 -7.86517441e-01 -9.64377224e-01 1.20951168e-01 -3.02549332e-01 -2.36632764e-01 7.49951780e-01 -2.98204646e-02 -7.87417412e-01 3.27971876e-01 -8.26942742e-01 -6.64914027e-02 1.09894371e+00 8.84006500e-01 6.91204131e-01 1.66043654e-01 5.94446778e-01 -8.87786567e-01 5.48238039e-01 -6.51147246e-01 -3.59346032e-01 1.97916090e-01 -5.88759542e-01 -4.56891149e-01 4.82331902e-01 -4.42410648e-01 -9.47357476e-01 5.26126564e-01 -4.50054497e-01 3.05826813e-01 -4.08509016e-01 5.18806756e-01 4.63826895e-01 -3.42280924e-01 7.35808790e-01 -5.42055964e-02 4.35655951e-01 -1.42176971e-01 -2.78715581e-01 1.17475867e+00 4.28555787e-01 1.72911316e-01 2.73422092e-01 7.47149646e-01 3.24240997e-02 -1.00020385e+00 -5.83439112e-01 -8.09512436e-01 -4.19535786e-01 -2.11506501e-01 1.17448592e+00 -8.82996023e-01 -5.57782531e-01 6.76522613e-01 -5.23042917e-01 -5.77719927e-01 -4.77012843e-02 3.17434341e-01 6.76394925e-02 -1.11494973e-01 -8.07218790e-01 -7.18587160e-01 -9.07127082e-01 -1.03587449e+00 1.34153426e+00 7.73677588e-01 -3.27090740e-01 -1.43233764e+00 1.24471284e-01 5.51539183e-01 6.30143702e-01 4.82053667e-01 9.54769969e-01 -7.04776525e-01 -1.36297405e-01 -7.40353346e-01 -1.44593716e-01 1.46786600e-01 4.87818867e-01 2.98937261e-01 -1.16754806e+00 -1.95420966e-01 -4.32509869e-01 -2.45997503e-01 7.82844126e-01 7.08262324e-01 5.53915501e-01 1.90929361e-02 -8.85162890e-01 2.57951796e-01 1.67373192e+00 6.59350097e-01 6.33710325e-01 3.36828738e-01 3.68014604e-01 4.18650389e-01 -4.41083573e-02 -3.53812367e-01 3.62814844e-01 2.26013795e-01 4.26745355e-01 -5.52461743e-01 -6.73475027e-01 1.17375232e-01 1.29620180e-01 4.12238181e-01 -3.17395866e-01 -3.76253068e-01 -1.04363608e+00 8.43642652e-01 -9.83303785e-01 -5.04797101e-01 -3.35839748e-01 1.72381186e+00 6.37317598e-01 1.44365486e-02 1.41251117e-01 4.75219637e-01 7.16084719e-01 -9.27517354e-01 -4.69855428e-01 -5.30622661e-01 -4.87427041e-02 3.96759540e-01 6.58450425e-01 3.62114727e-01 -8.41360629e-01 7.93160141e-01 6.11650753e+00 5.12335777e-01 -1.51677310e+00 6.13483787e-02 1.02050221e+00 3.39200884e-01 2.43219063e-01 -5.34177005e-01 -8.62174213e-01 2.09889859e-01 1.03873658e+00 1.67594895e-01 -1.86303079e-01 7.16347337e-01 6.02732636e-02 -3.03003281e-01 -6.56737030e-01 2.11430460e-01 -7.61196762e-02 -1.34497404e+00 -1.88758567e-01 4.68953550e-01 9.10955667e-01 3.85955095e-01 1.43098518e-01 2.90669203e-01 6.18330300e-01 -1.24892581e+00 -1.15321562e-01 8.47091600e-02 1.10086548e+00 -2.86281705e-01 1.38789070e+00 1.90040581e-02 -8.92853498e-01 -2.50663787e-01 9.35893878e-03 3.66061985e-01 -6.49154559e-02 1.99523672e-01 -1.68571734e+00 1.54296950e-01 7.35417426e-01 1.51771814e-01 -7.01910317e-01 1.33687270e+00 2.21636295e-02 8.14788699e-01 -7.44296610e-01 -4.57199872e-01 4.84627664e-01 5.35964251e-01 -2.22938918e-02 7.87071645e-01 4.61670756e-01 1.76643178e-01 -2.48239487e-01 1.41375616e-01 1.51766300e-01 2.74342537e-01 -1.76045761e-01 1.20159522e-01 8.27601627e-02 1.29985118e+00 -1.17041492e+00 -1.60211474e-01 -3.21463019e-01 7.99982429e-01 -2.24152029e-01 -3.72128412e-02 -3.00931782e-01 -1.28207639e-01 1.54505774e-01 6.26394451e-02 -1.62882134e-02 6.34336293e-01 -4.55561519e-01 -4.02731180e-01 -4.95701492e-01 -8.89746904e-01 8.14678490e-01 -6.24786377e-01 -1.15160477e+00 7.41239190e-01 -7.21577168e-01 -1.11685371e+00 -5.72505146e-02 -1.19867432e+00 -1.10771477e+00 8.21886182e-01 -1.64363623e+00 -1.56178343e+00 -6.71199858e-01 4.84774053e-01 6.00779831e-01 -1.13019966e-01 1.30985105e+00 3.24651375e-02 -5.77104747e-01 8.45082164e-01 7.74891824e-02 1.81230634e-01 3.79169613e-01 -1.44427216e+00 -3.07215810e-01 2.80268565e-02 -4.79363084e-01 3.08435678e-01 2.19301134e-01 -6.05392814e-01 -1.08521128e+00 -1.26711929e+00 7.59356141e-01 -2.30277479e-01 4.49435562e-01 4.11681384e-01 -7.64876664e-01 5.33722460e-01 1.62288204e-01 -9.13122967e-02 1.19680381e+00 -3.69687021e-01 1.35441482e-01 2.31635198e-01 -1.72958612e+00 6.63161576e-01 2.17034176e-01 -2.40602434e-01 -2.71454118e-02 9.01855469e-01 2.12862998e-01 -4.71529782e-01 -1.05468607e+00 4.07054394e-01 7.25614905e-01 -5.56583226e-01 7.26588845e-01 -2.57928848e-01 2.22768605e-01 -9.08473060e-02 1.69060990e-01 -1.16094637e+00 -1.17512316e-01 -2.53635556e-01 3.63665223e-01 7.27422118e-01 8.70495141e-01 -8.58505011e-01 1.34747899e+00 3.57940793e-01 -3.17409724e-01 -8.47726941e-01 -9.95247245e-01 -3.77709001e-01 3.14463466e-01 1.34094566e-01 3.24068427e-01 6.84654236e-01 -1.37018263e-01 -2.23337999e-03 5.34472823e-01 3.81865054e-01 2.33273908e-01 -4.61385608e-01 1.52150914e-01 -1.40321338e+00 4.90368493e-02 -8.20708156e-01 -7.76913226e-01 1.19514838e-01 -4.07947540e-01 -9.81647551e-01 -1.76826507e-01 -1.98510087e+00 4.37564738e-02 -4.38436747e-01 -3.70504200e-01 7.47054398e-01 -1.25415102e-01 6.10353589e-01 -3.32458228e-01 1.88321128e-01 3.98084164e-01 -4.15604472e-01 1.37352121e+00 -1.95369840e-01 -2.95282364e-01 4.03499633e-01 -8.29855025e-01 8.25662911e-01 1.10730410e+00 -2.61514574e-01 -2.22538695e-01 -1.82842135e-01 -2.67445594e-01 3.94284762e-02 2.66974151e-01 -1.05850184e+00 2.92769969e-01 1.01415724e-01 5.26372433e-01 -4.52082992e-01 1.62386551e-01 -9.92363393e-01 1.32420957e-01 1.08274198e+00 -4.95320298e-02 -4.62095201e-01 5.64515829e-01 4.20410216e-01 -1.70767203e-01 -2.76757985e-01 9.92117822e-01 -3.09936255e-01 -6.96701109e-01 9.84442383e-02 -8.49820912e-01 -6.87165976e-01 1.34400558e+00 -5.76659620e-01 -6.32144272e-01 -2.31448933e-01 -8.05141151e-01 1.49935052e-01 2.79417485e-01 1.20207250e-01 1.87664196e-01 -8.37203324e-01 -6.84290707e-01 4.37697731e-02 2.51980573e-01 3.28120566e-03 1.80815935e-01 1.07633913e+00 -1.20423770e+00 4.35148597e-01 -1.69102371e-01 -6.46032393e-01 -1.66345298e+00 -1.38147399e-01 7.98526227e-01 -5.10652661e-01 -5.04521132e-01 1.05135012e+00 -3.79630506e-01 -3.88525903e-01 2.38955617e-01 -5.10884166e-01 -6.20877922e-01 4.64292336e-03 4.87017453e-01 9.34527889e-02 4.56657141e-01 -7.11571038e-01 -1.93671122e-01 8.82356763e-01 -5.75007975e-01 2.86792010e-01 1.02075601e+00 3.39174211e-01 2.02504516e-01 -1.79983750e-01 1.43358552e+00 -1.84027955e-01 -4.78871495e-01 2.02004254e-01 1.13313325e-01 3.62336904e-01 5.59714437e-01 -1.61525428e+00 -1.26502311e+00 5.78600049e-01 1.49482834e+00 2.02790052e-01 1.34226549e+00 2.33430639e-02 8.00546169e-01 2.54015476e-01 1.29549280e-01 -9.68881726e-01 -2.54591793e-01 -8.43865350e-02 3.60459000e-01 -1.78288686e+00 -1.08305022e-01 -5.61806083e-01 -4.10419762e-01 1.28458154e+00 5.40358961e-01 -2.65059266e-02 6.68844640e-01 2.67968684e-01 9.00546908e-01 -4.96741563e-01 -5.15562713e-01 -4.99966592e-01 1.56567022e-01 7.68157125e-01 5.82236946e-01 4.34374213e-01 -3.60170037e-01 2.07396224e-01 -1.04773529e-01 3.20323229e-01 5.12471437e-01 1.04059231e+00 -4.30962920e-01 -1.01100600e+00 -3.77733290e-01 8.52988303e-01 -8.06138933e-01 1.11436583e-01 -7.59810865e-01 1.34301949e+00 2.36250341e-01 8.84908140e-01 9.04714316e-02 3.50774569e-03 1.12719692e-01 -4.35559452e-02 2.55340580e-02 -5.63106954e-01 -9.38971639e-01 1.52135476e-01 2.83302754e-01 1.35002315e-01 -3.07444572e-01 -4.72628206e-01 -1.46922159e+00 1.31284520e-01 -7.05716550e-01 -6.40205443e-02 1.11538792e+00 7.59778738e-01 1.53847307e-01 6.96390808e-01 1.55331135e-01 -3.67260486e-01 -2.52499461e-01 -1.13121450e+00 -3.81983966e-01 -2.39125453e-02 3.79093081e-01 -2.39015162e-01 -4.74451452e-01 2.22067147e-01]
[15.403705596923828, -3.0137808322906494]
cf6fdbc3-4d3d-4b11-a1cc-a74503d74ef2
the-devil-is-in-the-middle-exploiting-mid
1711.08106
null
http://arxiv.org/abs/1711.08106v2
http://arxiv.org/pdf/1711.08106v2.pdf
The Devil is in the Middle: Exploiting Mid-level Representations for Cross-Domain Instance Matching
Many vision problems require matching images of object instances across different domains. These include fine-grained sketch-based image retrieval (FG-SBIR) and Person Re-identification (person ReID). Existing approaches attempt to learn a joint embedding space where images from different domains can be directly compared. In most cases, this space is defined by the output of the final layer of a deep neural network (DNN), which primarily contains features of a high semantic level. In this paper, we argue that both high and mid-level features are relevant for cross-domain instance matching (CDIM). Importantly, mid-level features already exist in earlier layers of the DNN. They just need to be extracted, represented, and fused properly with the final layer. Based on this simple but powerful idea, we propose a unified framework for CDIM. Instantiating our framework for FG-SBIR and ReID, we show that our simple models can easily beat the state-of-the-art models, which are often equipped with much more elaborate architectures.
['Yi-Zhe Song', 'Qian Yu', 'Timothy M. Hospedales', 'Tao Xiang', 'Xiaobin Chang']
2017-11-22
null
null
null
null
['sketch-based-image-retrieval']
['computer-vision']
[-7.79135004e-02 -2.69832641e-01 -1.03255831e-01 -5.46844184e-01 -5.74044943e-01 -6.01228893e-01 1.01787722e+00 8.95542577e-02 -6.71723723e-01 4.71972108e-01 1.18196115e-01 3.40085238e-01 -2.64390826e-01 -9.18855071e-01 -6.78119540e-01 -3.45715404e-01 1.48829445e-01 5.05215704e-01 2.85962135e-01 -1.92108050e-01 4.36902232e-02 6.58317924e-01 -1.73399627e+00 3.34977180e-01 4.66392487e-01 1.05815208e+00 -7.75948539e-02 4.64261979e-01 -2.87292063e-01 5.22180080e-01 -4.62279379e-01 -9.39157665e-01 4.56494272e-01 -6.95669726e-02 -8.70777249e-01 -1.09030932e-01 1.01874197e+00 -7.21531928e-01 -8.75163853e-01 1.14349365e+00 4.90255922e-01 1.42674685e-01 6.75268948e-01 -1.57169759e+00 -1.04686725e+00 2.09509090e-01 -3.55269611e-01 -5.75974733e-02 3.35281879e-01 -7.01846229e-03 9.17276978e-01 -9.61616457e-01 8.38602245e-01 1.58869445e+00 7.65288889e-01 6.30066931e-01 -1.26622307e+00 -6.70712650e-01 2.87268847e-01 4.58872646e-01 -1.41959584e+00 -3.72165799e-01 9.59056914e-01 -4.19086248e-01 4.98564899e-01 -3.99485007e-02 5.60070992e-01 1.29123223e+00 -3.77173245e-01 9.84798729e-01 8.95076275e-01 -1.83050245e-01 4.70406897e-02 1.43255755e-01 3.07602584e-01 6.31530046e-01 4.78216082e-01 1.95447728e-01 -6.64516568e-01 -1.13267757e-01 1.12152886e+00 4.44331288e-01 -4.42731865e-02 -7.09032416e-01 -1.27673423e+00 7.31181622e-01 6.90175295e-01 5.06404996e-01 -1.99536726e-01 1.57445416e-01 3.92277479e-01 5.38411140e-01 1.61608383e-01 3.37955236e-01 -1.77024543e-01 3.12742859e-01 -8.64386559e-01 5.39660931e-01 6.77101254e-01 1.00996661e+00 9.28002119e-01 -5.10918438e-01 -3.58353227e-01 9.95863438e-01 2.30308473e-01 2.11641952e-01 1.58312559e-01 -8.03300083e-01 3.72353017e-01 6.83437586e-01 2.74891824e-01 -1.00293243e+00 -1.00181783e-02 -3.42622846e-01 -8.98887992e-01 2.52525032e-01 6.97893202e-01 3.14310163e-01 -7.12042272e-01 1.99738514e+00 1.50874093e-01 3.20396811e-01 5.30831628e-02 1.05323422e+00 9.25264657e-01 2.31304064e-01 1.40760392e-01 5.91458857e-01 1.40618813e+00 -9.49512959e-01 -3.48885506e-01 -2.24585563e-01 8.97820741e-02 -4.35397714e-01 8.60490739e-01 -8.17313045e-02 -1.14828813e+00 -1.00517619e+00 -1.10011196e+00 -5.83472788e-01 -9.80260432e-01 3.15575242e-01 4.48068887e-01 2.54105300e-01 -1.27719641e+00 8.34562361e-01 -2.29612529e-01 -6.52574837e-01 6.17385507e-01 2.71208316e-01 -8.79897058e-01 -3.32347631e-01 -1.27927351e+00 9.12567735e-01 1.26152068e-01 6.39427677e-02 -6.72817290e-01 -5.86073458e-01 -8.89396727e-01 9.60296318e-02 2.02488136e-02 -1.14928949e+00 1.02712715e+00 -9.90767062e-01 -1.14970970e+00 1.43908668e+00 -1.44207269e-01 -3.03660601e-01 8.54487181e-01 -2.68206328e-01 -2.90152103e-01 1.63427293e-01 2.12196976e-01 8.54770124e-01 1.00022542e+00 -1.33062577e+00 -5.33023238e-01 -7.79060781e-01 4.31510508e-01 -1.29755391e-02 -3.43302161e-01 1.11467376e-01 -7.47707903e-01 -6.28422678e-01 -1.21677183e-01 -7.21139371e-01 2.35101935e-02 8.00314784e-01 -1.39896557e-01 -6.32831335e-01 6.58162117e-01 -6.93686128e-01 7.71514297e-01 -2.11658025e+00 3.24419171e-01 7.47143663e-03 3.31416339e-01 4.26918954e-01 -5.45848787e-01 4.35643286e-01 -9.44575369e-02 -4.21186090e-02 -5.07602617e-02 -6.60958290e-01 4.08118635e-01 7.98388571e-02 -2.60625362e-01 4.92326409e-01 4.65457439e-01 1.32342184e+00 -9.26929653e-01 -3.59433353e-01 3.38059038e-01 5.98149538e-01 -1.50693089e-01 3.73693138e-01 1.55798584e-01 1.27775580e-01 -3.24853987e-01 5.41683972e-01 8.43551576e-01 -3.15618634e-01 4.11947258e-02 -4.90017831e-01 3.94993797e-02 1.05654486e-01 -1.28168225e+00 2.08241844e+00 -2.22504809e-01 6.81743443e-01 1.81970164e-01 -1.25533867e+00 8.96866679e-01 1.74824726e-02 2.38281995e-01 -8.82749796e-01 -5.88906556e-02 1.40655637e-01 -3.56858194e-01 -1.73784405e-01 4.43929732e-01 1.57538399e-01 -1.67493522e-01 4.85564768e-01 3.55030417e-01 2.95917839e-01 4.25407253e-02 1.81383342e-01 7.88246095e-01 8.92182216e-02 1.44397259e-01 -1.40393063e-01 7.98378825e-01 -2.43236199e-01 2.25703567e-01 1.07498765e+00 -4.51392531e-01 5.74922442e-01 2.39985853e-01 -8.09073627e-01 -1.35096240e+00 -1.37804639e+00 -1.25569955e-01 1.12028158e+00 5.48159361e-01 -2.61091381e-01 -6.74826443e-01 -6.97883546e-01 4.86723959e-01 -5.67751899e-02 -7.90753305e-01 -1.37753651e-01 -5.84782839e-01 -1.20588273e-01 6.41098619e-01 6.98474348e-01 7.95132935e-01 -8.99173677e-01 -1.85295656e-01 2.15379238e-01 2.11259723e-01 -1.28664720e+00 -3.21936786e-01 -3.04980278e-01 -5.43331981e-01 -1.17969739e+00 -1.16659188e+00 -8.90829861e-01 5.36906958e-01 4.81617481e-01 1.35579002e+00 2.90430516e-01 -4.33432400e-01 5.51950574e-01 -1.39188468e-01 -2.26717606e-01 -8.68907273e-02 3.67410714e-03 1.08552709e-01 2.30410814e-01 6.49046004e-01 -7.09061682e-01 -7.15898335e-01 2.90460169e-01 -8.96462679e-01 -5.25048599e-02 6.21422112e-01 9.39933300e-01 4.07563120e-01 -3.56326729e-01 4.90135729e-01 -5.63073397e-01 5.70421100e-01 -1.54622570e-02 -6.34919643e-01 6.79909348e-01 -1.22437038e-01 2.34499320e-01 5.38174272e-01 -5.44750631e-01 -7.44461417e-01 -7.81932175e-02 -1.05290316e-01 -6.15338206e-01 -4.15785402e-01 5.59591502e-02 -2.97167420e-01 -2.95820981e-01 4.30320501e-01 1.56051874e-01 1.12314718e-02 -7.34012365e-01 5.91279745e-01 6.75998926e-01 9.01493847e-01 -7.08593428e-01 1.11972010e+00 6.78722918e-01 -1.32108405e-01 -4.89207178e-01 -9.03785765e-01 -4.79165375e-01 -1.00010848e+00 -4.05028239e-02 7.98259139e-01 -1.08918929e+00 -7.98988223e-01 7.12128997e-01 -1.40105093e+00 -1.58808425e-01 -2.32962713e-01 1.74155340e-01 -4.64994669e-01 3.29260260e-01 -5.86840272e-01 -5.55901289e-01 -2.68685043e-01 -9.01087880e-01 1.47753787e+00 2.90858328e-01 1.20746396e-01 -9.10815597e-01 -7.47335851e-02 9.96041149e-02 3.90829653e-01 6.91020563e-02 8.10511529e-01 -5.98448634e-01 -5.69379091e-01 -2.20933884e-01 -9.26052928e-01 2.00350791e-01 4.67161238e-02 -2.49611974e-01 -1.25327456e+00 -3.95292521e-01 -4.34354395e-01 -2.79886127e-01 1.18701506e+00 8.39791521e-02 1.43480313e+00 -1.36608213e-01 -4.70013261e-01 7.17654049e-01 1.46866012e+00 -3.46658796e-01 6.75088942e-01 4.41081017e-01 7.35882938e-01 7.79556990e-01 2.83532619e-01 1.50557086e-01 6.21496379e-01 9.25505400e-01 1.78136587e-01 -2.46484175e-01 -3.92871141e-01 -3.95476431e-01 1.40766904e-01 2.06207767e-01 -2.32989108e-03 1.03275195e-01 -6.10593498e-01 7.21111715e-01 -2.09437537e+00 -1.20064282e+00 3.51172835e-01 2.13668680e+00 6.32272005e-01 -2.34664708e-01 2.42183745e-01 -2.50404868e-02 9.33061004e-01 2.37180188e-01 -7.27388203e-01 -6.12301975e-02 -2.53284156e-01 1.99810013e-01 3.09911281e-01 2.78134137e-01 -1.39379144e+00 9.19141948e-01 6.24808455e+00 6.24687433e-01 -7.76080251e-01 -1.86063312e-02 2.79223889e-01 2.44538024e-01 -1.85730875e-01 -2.80045152e-01 -7.25522339e-01 4.80832249e-01 3.86472255e-01 -1.78073630e-01 5.28994739e-01 8.22501063e-01 -5.41525304e-01 3.22309047e-01 -1.69191968e+00 1.36992753e+00 4.02176790e-02 -1.56198323e+00 2.45377064e-01 -3.86962034e-02 5.23661256e-01 -2.13800624e-01 1.39117748e-01 3.55433524e-01 2.41920128e-01 -1.05848682e+00 6.34187281e-01 6.75945342e-01 9.75908577e-01 -6.26460493e-01 5.90920091e-01 8.87688398e-02 -1.33616841e+00 -1.59267873e-01 -8.32849860e-01 1.47614509e-01 -1.05436295e-01 3.45458120e-01 -1.77232578e-01 7.42387295e-01 7.54862785e-01 9.95956481e-01 -6.25007272e-01 9.50968206e-01 -1.84900269e-01 -3.22347730e-01 -1.89270377e-01 3.49808514e-01 6.48393482e-02 -5.06703220e-02 2.77553171e-01 1.22712147e+00 2.07578361e-01 -7.56786540e-02 1.25981092e-01 1.21899724e+00 -3.29027146e-01 -3.77435714e-01 -8.12793911e-01 1.21961057e-01 5.80894589e-01 1.23591387e+00 -1.01031303e-01 -6.35719955e-01 -6.08787537e-01 1.26350760e+00 7.27241457e-01 3.77163082e-01 -6.83995187e-01 -3.44456255e-01 1.33038163e+00 -5.84805273e-02 2.45315924e-01 -1.64265200e-01 1.20691648e-02 -1.47468555e+00 2.54764616e-01 -6.56827390e-01 4.88704532e-01 -7.13728547e-01 -2.11413336e+00 4.23325002e-01 -1.70709640e-01 -1.10427427e+00 -2.35384569e-01 -1.01720667e+00 -4.98144925e-01 1.18262792e+00 -1.91423666e+00 -1.56318367e+00 -5.50943792e-01 1.13037372e+00 2.43774205e-01 -2.19975442e-01 8.91458035e-01 5.57939768e-01 -4.27953571e-01 1.03506768e+00 -3.46102566e-02 7.47515082e-01 8.06570888e-01 -1.10227740e+00 7.57948637e-01 7.28478670e-01 2.98549801e-01 8.60252023e-01 1.24670602e-01 -3.31708759e-01 -1.50550616e+00 -1.19710016e+00 9.99499857e-01 -4.67145979e-01 5.74697196e-01 -5.85488141e-01 -9.69430804e-01 4.75404263e-01 -9.14954320e-02 3.44718546e-01 3.56052101e-01 1.30911261e-01 -9.98361766e-01 -2.67760813e-01 -1.23525834e+00 5.28784752e-01 1.43914557e+00 -1.25098884e+00 -8.03732634e-01 1.40583456e-01 4.29968327e-01 -5.09867165e-03 -1.00842333e+00 2.02004328e-01 1.03503335e+00 -1.05874550e+00 1.59429443e+00 -1.00231111e+00 2.41232887e-01 -2.49059051e-01 -2.38502145e-01 -9.58842635e-01 -5.01196265e-01 -2.03019202e-01 -8.34136605e-02 1.11816931e+00 -1.44274876e-01 -5.30151665e-01 5.89941323e-01 7.42847741e-01 3.04240942e-01 -2.91205943e-01 -8.38244855e-01 -1.01835477e+00 1.74465671e-01 -1.67991072e-01 1.02906454e+00 1.01495647e+00 -3.01023543e-01 2.67263532e-01 -3.00402939e-01 3.48924667e-01 9.33643520e-01 3.86583120e-01 9.53177929e-01 -1.90531826e+00 -2.71575209e-02 -6.17319822e-01 -7.07885146e-01 -1.08751953e+00 4.22356635e-01 -8.95017982e-01 -3.38802725e-01 -1.68427050e+00 4.76235926e-01 -3.91835093e-01 -4.97246653e-01 3.99306566e-01 -2.31167391e-01 3.95957232e-01 5.03660262e-01 3.88405681e-01 -6.84351742e-01 4.22473371e-01 1.04277968e+00 -4.00458902e-01 2.05142856e-01 -3.85806143e-01 -6.01979315e-01 6.23804808e-01 5.04264474e-01 -1.09824963e-01 -4.20814082e-02 -8.76891255e-01 -1.41337723e-01 -2.08877042e-01 1.14037895e+00 -1.01621985e+00 5.03060818e-01 3.71688940e-02 8.52947712e-01 -4.20643330e-01 5.33584476e-01 -8.57923150e-01 1.02931693e-01 1.40873834e-01 -4.12426144e-01 -6.29357100e-02 -3.06607014e-03 4.66325998e-01 -3.89646232e-01 -4.20111371e-03 7.84405231e-01 -2.18813956e-01 -1.07729352e+00 8.08345139e-01 2.52055913e-01 -2.19964311e-01 7.58194983e-01 -3.14933062e-01 -4.56011772e-01 -2.63706028e-01 -6.54506922e-01 1.31456390e-01 7.38236606e-01 6.68643773e-01 5.92630327e-01 -1.69916034e+00 -6.49625063e-01 3.89032245e-01 5.59697747e-01 -1.91495195e-01 2.90786266e-01 3.20712060e-01 4.05497849e-03 3.49378377e-01 -6.16090596e-01 -5.87144434e-01 -9.28731620e-01 8.15775633e-01 4.46979314e-01 -2.04366431e-01 -5.39621115e-01 8.55265737e-01 4.28626657e-01 -6.16132200e-01 2.45755956e-01 1.23337097e-01 -7.17138797e-02 1.99546427e-01 7.61530340e-01 1.47906527e-01 -9.51501802e-02 -7.71691740e-01 -5.44753969e-01 8.19286764e-01 -7.62435347e-02 -1.10424504e-01 1.39627266e+00 1.76024958e-02 -1.66048259e-01 4.89910394e-02 1.49866068e+00 -4.84389693e-01 -1.31585252e+00 -7.13779509e-01 8.73086527e-02 -6.10863090e-01 -2.14430690e-01 -6.49872541e-01 -9.48221385e-01 1.10595429e+00 6.98208809e-01 -2.37534139e-02 9.54737782e-01 2.06832454e-01 6.74487233e-01 5.56879640e-01 6.22285366e-01 -1.05913913e+00 9.27520096e-02 3.30141515e-01 8.81816447e-01 -1.37699234e+00 -1.44615069e-01 -1.01377275e-02 -1.84217483e-01 1.14561987e+00 5.74892998e-01 -4.58196968e-01 5.27863264e-01 -1.70288105e-02 -2.29327947e-01 -7.80325904e-02 -4.00833488e-01 -5.14623702e-01 5.48405588e-01 8.42338204e-01 1.96654171e-01 -6.57807812e-02 -3.13492604e-02 7.01886952e-01 1.14895046e-01 -4.24142415e-03 -1.09196194e-01 6.63753748e-01 -2.56757438e-01 -1.42460489e+00 -1.96868017e-01 2.49034628e-01 -1.27375886e-01 1.38671249e-01 -6.51712596e-01 8.68915856e-01 1.97576031e-01 6.01970911e-01 2.41046384e-01 -2.58239716e-01 3.66882980e-01 4.83358419e-03 7.28794754e-01 -2.70538926e-01 -4.30492789e-01 -5.57808876e-01 -2.25845769e-01 -8.73605728e-01 -6.42609477e-01 -4.37650234e-01 -5.15749633e-01 -4.83477384e-01 1.79028347e-01 -1.95918143e-01 5.63819051e-01 8.25906932e-01 4.93641317e-01 3.77141461e-02 3.86689454e-01 -1.02600193e+00 -5.75536668e-01 -6.88604355e-01 -6.14232600e-01 7.68931150e-01 5.35228550e-01 -8.87004554e-01 1.09453395e-01 -8.37857276e-02]
[11.689029693603516, 0.6059616208076477]
b73b0c82-258d-455d-9fb8-54a9e04540dd
unsupervised-summarization-for-chat-logs-with
2012.07300
null
https://arxiv.org/abs/2012.07300v2
https://arxiv.org/pdf/2012.07300v2.pdf
Unsupervised Summarization for Chat Logs with Topic-Oriented Ranking and Context-Aware Auto-Encoders
Automatic chat summarization can help people quickly grasp important information from numerous chat messages. Unlike conventional documents, chat logs usually have fragmented and evolving topics. In addition, these logs contain a quantity of elliptical and interrogative sentences, which make the chat summarization highly context dependent. In this work, we propose a novel unsupervised framework called RankAE to perform chat summarization without employing manually labeled data. RankAE consists of a topic-oriented ranking strategy that selects topic utterances according to centrality and diversity simultaneously, as well as a denoising auto-encoder that is carefully designed to generate succinct but context-informative summaries based on the selected utterances. To evaluate the proposed method, we collect a large-scale dataset of chat logs from a customer service environment and build an annotated set only for model evaluation. Experimental results show that RankAE significantly outperforms other unsupervised methods and is able to generate high-quality summaries in terms of relevance and topic coverage.
['Xiaozhong Liu', 'Xuanjing Huang', 'Qi Zhang', 'Changlong Sun', 'Zhuoren Jiang', 'Yangyang Kang', 'Lujun Zhao', 'Jun Lin', 'Yicheng Zou']
2020-12-14
null
null
null
null
['topic-coverage']
['natural-language-processing']
[ 2.36874506e-01 2.11055696e-01 3.25174600e-01 -5.33973992e-01 -1.32291365e+00 -3.29253972e-01 7.74633408e-01 6.37668312e-01 -2.17648655e-01 7.45177746e-01 9.70914602e-01 4.41790164e-01 2.44974103e-02 -3.97819728e-01 -1.27446935e-01 -5.28593242e-01 2.15523466e-01 9.14270401e-01 3.08569938e-01 -2.11917654e-01 7.18831182e-01 -1.83813691e-01 -1.44452739e+00 6.15615368e-01 1.25988448e+00 3.95676225e-01 7.20350444e-01 8.69911492e-01 -4.11146849e-01 1.01991105e+00 -1.38334095e+00 -4.48468387e-01 -5.04345775e-01 -9.77983534e-01 -9.40530360e-01 4.32780057e-01 1.03781268e-01 -2.13155165e-01 3.30060348e-02 7.03042269e-01 6.76046252e-01 5.34186065e-01 7.77950525e-01 -7.04488039e-01 -6.82944283e-02 1.07518125e+00 9.21482500e-03 1.78902552e-01 6.58702195e-01 -8.02626312e-02 1.36424088e+00 -5.88204145e-01 8.25306714e-01 1.24223781e+00 4.43112046e-01 3.91299158e-01 -1.09288311e+00 -6.07894920e-02 8.77216756e-02 1.53045446e-01 -1.04480183e+00 -4.58647668e-01 1.02182639e+00 -3.38857114e-01 9.12335753e-01 5.78592718e-01 5.86711705e-01 1.23083544e+00 1.44283334e-03 1.01359642e+00 6.89468324e-01 -3.94082636e-01 5.14882922e-01 2.53919154e-01 4.22338605e-01 1.18324891e-01 -1.68108732e-01 -9.85690355e-01 -6.69288516e-01 -4.33481723e-01 1.68007642e-01 -6.89815208e-02 -1.88914165e-01 2.61090696e-01 -1.14056027e+00 8.39006782e-01 -2.17203021e-01 4.44715381e-01 -7.39293218e-01 -7.32855648e-02 6.92093134e-01 2.69949198e-01 8.09844673e-01 6.03614151e-01 -7.79806972e-02 -6.62894666e-01 -1.09009695e+00 4.32441682e-01 1.26565874e+00 9.38840806e-01 5.54701269e-01 -1.67790562e-01 -6.55380666e-01 1.21558404e+00 6.43783435e-02 1.53173417e-01 5.72503626e-01 -9.11955833e-01 6.95165455e-01 6.79686785e-01 -2.66607944e-02 -9.38004494e-01 -2.02430964e-01 -3.36573094e-01 -8.15658629e-01 -5.54618835e-01 4.34800200e-02 -2.11487427e-01 -2.90413886e-01 1.29596400e+00 8.00086111e-02 -7.38200322e-02 1.07775033e-01 6.37823343e-01 1.10171854e+00 9.09433484e-01 -6.91147298e-02 -7.26994097e-01 1.38335025e+00 -1.05445743e+00 -1.18002415e+00 8.44925791e-02 5.16258299e-01 -9.01408792e-01 1.26105392e+00 5.19122779e-01 -1.25308990e+00 -1.94064468e-01 -6.76955938e-01 -1.32252052e-01 2.07066521e-01 2.34596521e-01 1.62917241e-01 2.85241276e-01 -8.15926611e-01 4.68399376e-01 -6.84698403e-01 -4.66934621e-01 2.84406133e-02 -6.83297291e-02 -1.90376174e-02 1.76106378e-01 -9.76614833e-01 7.71901190e-01 4.03461844e-01 -2.40910038e-01 -6.74808860e-01 -5.00268877e-01 -6.99214339e-01 5.00089526e-01 4.89751160e-01 -5.15390813e-01 1.65920651e+00 -6.45233989e-01 -1.87734354e+00 2.05844656e-01 -3.83322865e-01 -5.94403267e-01 6.05596602e-01 -2.36307353e-01 1.37251943e-01 6.55094624e-01 1.74807869e-02 2.38902599e-01 6.33937657e-01 -1.15489554e+00 -6.87336564e-01 -5.72950020e-02 -1.46702304e-01 4.86981481e-01 -4.01821047e-01 2.57508159e-01 -4.69564438e-01 -6.72440529e-01 -8.80073830e-02 -4.41547930e-01 -2.06921548e-01 -1.06870937e+00 -6.25638068e-01 -5.70649266e-01 7.59559751e-01 -9.64100897e-01 1.51129341e+00 -1.87358618e+00 2.98704535e-01 -7.80853629e-02 3.08979779e-01 -8.14457983e-02 8.55797231e-02 1.02419305e+00 5.86807728e-01 -1.91145726e-02 -2.65330255e-01 -8.82598281e-01 9.29891616e-02 -4.24358100e-02 -1.26960412e-01 -1.54750412e-02 1.40879765e-01 4.89248097e-01 -9.98546481e-01 -9.32467937e-01 1.21912889e-01 3.76458555e-01 -6.07003987e-01 5.14208078e-01 -5.47023833e-01 4.96924818e-01 -6.33187294e-01 2.11986393e-01 2.26850495e-01 -2.19575629e-01 2.87338287e-01 1.87705323e-01 4.26844768e-02 8.06062937e-01 -8.69678676e-01 1.48691678e+00 -5.37409961e-01 8.94921780e-01 3.31283882e-02 -8.99343073e-01 1.17207038e+00 5.86489439e-01 5.18043280e-01 -2.11344719e-01 2.56179124e-01 -5.22964559e-02 -2.71945715e-01 -5.26252568e-01 1.10684979e+00 8.27872753e-02 -3.62091601e-01 8.53591323e-01 2.68719792e-01 -3.64797473e-01 6.80667639e-01 6.41890943e-01 1.25456321e+00 -4.92542297e-01 2.47907296e-01 2.44043861e-03 5.35908639e-01 5.19552326e-04 3.91350031e-01 7.16121912e-01 1.52728751e-01 9.60995257e-01 8.77924144e-01 3.66900042e-02 -9.41777825e-01 -6.82611227e-01 2.81291485e-01 1.19206917e+00 -1.16653360e-01 -9.63567495e-01 -1.00057411e+00 -6.04231656e-01 -6.32424414e-01 1.08893692e+00 -2.28115156e-01 2.39802282e-02 -6.27736211e-01 -4.64736462e-01 2.54822820e-01 1.04634196e-01 3.57200176e-01 -1.33781815e+00 -4.63639528e-01 4.86687779e-01 -8.97519827e-01 -9.62868094e-01 -7.40389884e-01 4.86596823e-02 -8.10255229e-01 -8.56788278e-01 -7.39627421e-01 -7.37076461e-01 5.16309917e-01 2.00977266e-01 1.22811794e+00 5.46333753e-02 1.10401422e-01 3.15179884e-01 -9.15359259e-01 -4.69428003e-01 -8.91464889e-01 6.23587966e-01 -3.24572653e-01 4.66360115e-02 2.11870104e-01 -7.93021619e-01 -2.47997135e-01 6.84522912e-02 -8.20389867e-01 3.44633050e-02 5.78370631e-01 9.15126920e-01 1.88857242e-01 8.74993727e-02 1.06942999e+00 -1.03546143e+00 1.42374015e+00 -3.70661616e-01 -2.01681018e-01 2.42344245e-01 -3.17789674e-01 -7.60440007e-02 7.25845575e-01 -3.77434969e-01 -1.48382008e+00 -1.24796927e-01 -2.08270237e-01 1.54527649e-01 -1.82250977e-01 6.98201060e-01 -4.35974002e-02 8.33237827e-01 5.50382435e-01 4.69381988e-01 -1.46582350e-01 -8.05802286e-01 1.72339901e-01 1.06258321e+00 4.29243028e-01 -4.08443630e-01 3.05700332e-01 2.95936819e-02 -7.50817835e-01 -1.31347513e+00 -6.25607669e-01 -8.24188888e-01 -6.21823072e-01 -4.60079432e-01 6.49809361e-01 -6.39394522e-01 -5.78450680e-01 1.80518866e-01 -1.43411469e+00 -1.19738318e-01 -3.45325381e-01 4.47263986e-01 -6.49259746e-01 6.49184763e-01 -8.66946459e-01 -1.01962876e+00 -7.48347580e-01 -8.37787807e-01 1.07087958e+00 3.02974761e-01 -7.14900553e-01 -8.93763483e-01 1.94664612e-01 5.77414691e-01 2.39649773e-01 3.08757462e-02 6.65134311e-01 -1.26604939e+00 -3.84971201e-01 -3.37369531e-01 2.34415844e-01 3.68770331e-01 9.88614112e-02 8.58200416e-02 -7.93641448e-01 -1.01126857e-01 3.62898856e-02 -3.27091902e-01 9.03339088e-01 3.41394424e-01 7.36229420e-01 -7.20445514e-01 -9.12534446e-02 -2.49449298e-01 7.98905134e-01 1.01241559e-01 4.89866704e-01 4.67401184e-02 3.46726805e-01 8.68012071e-01 6.61711156e-01 8.45624149e-01 5.14547646e-01 5.38257897e-01 6.07148334e-02 4.51010674e-01 1.61978006e-01 -2.01205373e-01 5.28804004e-01 1.62573910e+00 -1.37758004e-02 -5.25925279e-01 -5.94125569e-01 6.60207093e-01 -1.86936033e+00 -1.13977242e+00 -1.30462036e-01 1.83729208e+00 1.14131439e+00 2.03771353e-01 3.27324659e-01 1.06610790e-01 7.97761858e-01 2.37793565e-01 -1.92886069e-02 -5.05102813e-01 4.66241725e-02 -1.11859426e-01 -2.91887164e-01 5.35487771e-01 -7.81404793e-01 7.27644742e-01 5.99011803e+00 9.37031448e-01 -5.82171261e-01 2.10672393e-01 3.96202892e-01 -1.03007682e-01 -4.60461944e-01 -1.09114133e-01 -5.62172651e-01 5.91321051e-01 1.04275107e+00 -5.43243408e-01 -1.44131836e-02 9.46241438e-01 5.34674108e-01 -2.66718805e-01 -9.32688773e-01 6.10562980e-01 3.53136063e-01 -1.42819822e+00 -6.10843487e-02 -1.36556447e-01 7.36811697e-01 -2.71638662e-01 -4.02559400e-01 3.54009777e-01 3.26744527e-01 -6.19610369e-01 4.66662973e-01 5.63815534e-01 1.19595766e-01 -8.54372263e-01 1.22711718e+00 8.22214603e-01 -7.61764765e-01 2.51980305e-01 -3.37313592e-01 -5.96231557e-02 5.08775413e-01 6.07501209e-01 -1.35967243e+00 6.41809642e-01 3.84067476e-01 6.82720482e-01 -4.47844267e-01 1.02270138e+00 -1.77763417e-01 1.00737369e+00 -2.92066216e-01 -5.40262222e-01 1.86046705e-01 -2.58621246e-01 9.48982775e-01 1.58761942e+00 1.41677141e-01 1.58104703e-01 2.62932301e-01 6.92135096e-01 -1.02108195e-01 3.57998937e-01 -2.57074744e-01 -1.91429287e-01 6.82064474e-01 1.24823141e+00 -8.91216278e-01 -5.68246543e-01 8.66972134e-02 9.66591597e-01 3.65653522e-02 1.66063428e-01 -4.46576685e-01 -5.22375286e-01 7.09136426e-02 -1.04412504e-01 3.01016152e-01 -1.15067996e-01 -1.53008983e-01 -1.14009392e+00 1.59926534e-01 -8.69710982e-01 3.46248537e-01 -5.23412466e-01 -1.23884463e+00 8.64500880e-01 1.13419048e-01 -1.18893862e+00 -7.03909099e-01 2.94008702e-01 -1.08462977e+00 4.87695307e-01 -9.69130218e-01 -8.39708924e-01 -3.57381821e-01 1.54841021e-01 1.32150948e+00 -2.36342117e-01 8.02509904e-01 8.77130181e-02 -6.08703136e-01 3.64236683e-01 3.02094370e-01 -3.85620929e-02 7.27001667e-01 -1.44543719e+00 1.59437805e-01 5.40208876e-01 6.17628954e-02 4.80078965e-01 1.18074644e+00 -6.99130118e-01 -8.68109405e-01 -9.12087858e-01 1.30062568e+00 -3.90442133e-01 3.64407867e-01 -3.52128625e-01 -1.12748778e+00 4.90237385e-01 6.34928584e-01 -1.06549942e+00 7.02613294e-01 3.29316378e-01 1.96712002e-01 2.13042218e-02 -8.38198245e-01 5.71812689e-01 5.11014700e-01 -2.25585371e-01 -1.15719140e+00 6.25439405e-01 7.86944866e-01 -2.57880270e-01 -7.23159790e-01 -2.11843058e-01 1.55994907e-01 -9.55126941e-01 4.53230560e-01 -2.24515945e-01 6.32063746e-01 8.08379203e-02 1.47767201e-01 -1.44616735e+00 4.34492491e-02 -9.65097249e-01 -9.20573175e-02 1.84187841e+00 2.05554038e-01 -2.46737152e-01 7.83259034e-01 1.88671947e-01 -4.25818056e-01 -5.11580944e-01 -7.56870866e-01 -4.58211720e-01 -4.26774144e-01 -2.10507587e-01 3.90849322e-01 5.39019525e-01 5.49485326e-01 1.03063440e+00 -4.85142112e-01 -2.53434271e-01 4.64522570e-01 6.66357949e-02 9.67464745e-01 -1.35408723e+00 -4.82091643e-02 -5.33847094e-01 -2.18600914e-01 -1.05180156e+00 1.26248837e-01 -4.80120480e-01 5.14254868e-01 -1.92700732e+00 3.88720989e-01 -2.97644716e-02 2.82019377e-01 -7.14004934e-02 -3.59924465e-01 -2.09743664e-01 1.17286772e-01 3.66568834e-01 -1.24288607e+00 9.31126833e-01 9.36375260e-01 -1.76267609e-01 -6.41399026e-01 4.63183969e-01 -5.67298949e-01 7.56110847e-01 8.27442169e-01 -5.49760282e-01 -5.08976698e-01 4.63235117e-02 -3.25817764e-02 3.19028407e-01 -1.51502207e-01 -7.36617863e-01 3.69457781e-01 -3.43611836e-02 -2.14709803e-01 -1.12846160e+00 4.14489836e-01 -3.15126032e-01 -2.54113883e-01 8.94233957e-02 -8.39089870e-01 -1.21029422e-01 -3.23702842e-01 5.93873739e-01 -6.62464142e-01 -5.32416999e-01 3.89177114e-01 -1.29448339e-01 -1.92385063e-01 -2.95903265e-01 -9.01481450e-01 2.63796061e-01 5.52187979e-01 -2.59144558e-03 -1.66868389e-01 -1.15662277e+00 -5.62361836e-01 3.94114047e-01 1.77679762e-01 1.46802008e-01 6.56842411e-01 -1.05173361e+00 -1.15771782e+00 -3.96228164e-01 -3.37878265e-03 2.67174125e-01 4.57497895e-01 7.21995413e-01 -5.34078479e-01 4.03984427e-01 8.13442320e-02 -6.44272089e-01 -1.53561628e+00 2.62178909e-02 -3.52890462e-01 -6.65002823e-01 -1.02689672e+00 6.16939664e-01 -1.90996468e-01 -2.43181810e-01 4.21610385e-01 -3.68700117e-01 -6.59924448e-01 5.23531973e-01 7.96405315e-01 4.73027855e-01 1.31748542e-01 -4.55499500e-01 5.16556613e-02 -1.01923987e-01 -2.61320114e-01 -3.81383300e-01 1.56814253e+00 -4.42358047e-01 -1.85195029e-01 6.64133191e-01 1.04883516e+00 7.55057856e-02 -1.06260741e+00 -2.99795270e-01 4.82991427e-01 -2.40845993e-01 -3.24928969e-01 -5.57098329e-01 -3.90557855e-01 7.44981349e-01 -2.69570619e-01 7.96843410e-01 1.00422323e+00 2.16631398e-01 8.84084940e-01 7.71034777e-01 -1.43110901e-01 -1.45010209e+00 5.76450288e-01 7.71915555e-01 1.17847204e+00 -1.04779816e+00 3.23285200e-02 -3.95310074e-01 -1.18029606e+00 1.01959097e+00 3.68507743e-01 -5.79320118e-02 1.05935305e-01 -1.21233426e-02 4.84146527e-04 -2.33593047e-01 -1.20491266e+00 -6.75286651e-02 1.76455125e-01 4.58546996e-01 4.12539124e-01 -1.80007175e-01 -6.66931748e-01 8.94295514e-01 -6.17950201e-01 -3.77977878e-01 9.49592948e-01 8.10430706e-01 -6.71241164e-01 -9.79420900e-01 -3.44559640e-01 5.21692634e-01 -4.46708769e-01 1.20542966e-01 -6.71699703e-01 5.31351984e-01 -6.63681746e-01 1.29559433e+00 1.33177996e-01 -3.33101928e-01 3.76592129e-01 3.18142653e-01 4.15728725e-02 -1.03318608e+00 -9.24330950e-01 4.40809280e-01 5.98099709e-01 4.77323905e-02 -3.85186791e-01 -8.66528511e-01 -1.12102652e+00 -2.31038794e-01 -5.80861628e-01 8.48805130e-01 5.95235586e-01 1.09356630e+00 2.40693390e-01 8.50822091e-01 8.36018682e-01 -8.50781620e-01 -6.89684689e-01 -1.72383356e+00 -4.44799036e-01 4.59717453e-01 1.98591501e-01 -2.75265366e-01 -3.39831680e-01 2.79434204e-01]
[12.629449844360352, 9.178641319274902]
f7b08058-a775-4d14-bc94-b8ccf210e06d
learning-a-natural-language-interface-with
1611.08945
null
http://arxiv.org/abs/1611.08945v4
http://arxiv.org/pdf/1611.08945v4.pdf
Learning a Natural Language Interface with Neural Programmer
Learning a natural language interface for database tables is a challenging task that involves deep language understanding and multi-step reasoning. The task is often approached by mapping natural language queries to logical forms or programs that provide the desired response when executed on the database. To our knowledge, this paper presents the first weakly supervised, end-to-end neural network model to induce such programs on a real-world dataset. We enhance the objective function of Neural Programmer, a neural network with built-in discrete operations, and apply it on WikiTableQuestions, a natural language question-answering dataset. The model is trained end-to-end with weak supervision of question-answer pairs, and does not require domain-specific grammars, rules, or annotations that are key elements in previous approaches to program induction. The main experimental result in this paper is that a single Neural Programmer model achieves 34.2% accuracy using only 10,000 examples with weak supervision. An ensemble of 15 models, with a trivial combination technique, achieves 37.7% accuracy, which is competitive to the current state-of-the-art accuracy of 37.1% obtained by a traditional natural language semantic parser.
['Martin Abadi', 'Andrew McCallum', 'Quoc V. Le', 'Dario Amodei', 'Arvind Neelakantan']
2016-11-28
null
null
null
null
['program-induction']
['computer-code']
[ 2.09756494e-01 6.73001885e-01 -4.99511153e-01 -8.14111233e-01 -1.04630971e+00 -5.70536375e-01 3.59108627e-01 4.96807665e-01 -5.07686853e-01 4.16895241e-01 -8.97368267e-02 -8.63985240e-01 1.73997924e-01 -1.09175718e+00 -1.45813298e+00 2.58003920e-01 -4.45899367e-02 1.01708508e+00 4.61525768e-01 -4.60127354e-01 3.12745199e-02 -1.24140549e-02 -1.64302146e+00 8.54571044e-01 7.92667329e-01 9.13782239e-01 -7.66737908e-02 9.15043056e-01 -6.48678362e-01 1.54618335e+00 -5.27855277e-01 -6.36858463e-01 -7.22228512e-02 -1.48064330e-01 -1.32980061e+00 -6.64173484e-01 7.53002703e-01 -4.66509223e-01 8.12513009e-02 9.75753069e-01 1.19479209e-01 -2.53707975e-01 7.95295909e-02 -1.31283367e+00 -9.04410005e-01 9.97368038e-01 -7.92267621e-02 -3.84288207e-02 6.67719662e-01 9.51761305e-02 1.52018940e+00 -7.76335359e-01 7.95728385e-01 1.38413334e+00 5.79985619e-01 6.57133460e-01 -1.56252611e+00 -4.58590299e-01 5.80282733e-02 1.83203239e-02 -1.00926650e+00 -3.91291648e-01 4.79403198e-01 -3.69725823e-01 1.58557832e+00 4.45377976e-02 -1.08456969e-01 7.08260655e-01 5.85284224e-03 8.46959472e-01 7.33906984e-01 -7.11071074e-01 1.74811423e-01 5.19578516e-01 8.59712005e-01 1.19971299e+00 1.96225196e-01 -6.63831015e-04 -4.49783176e-01 -4.53727007e-01 2.82506168e-01 -2.14519978e-01 1.35320872e-01 -4.59723949e-01 -9.28411186e-01 9.76943135e-01 5.10849416e-01 1.06006555e-01 1.49092972e-02 2.38041982e-01 6.48694813e-01 6.94837868e-01 6.30236492e-02 8.09442043e-01 -9.20412064e-01 -1.10723667e-01 -2.23094061e-01 4.89491105e-01 1.58164954e+00 1.23582053e+00 9.78613496e-01 -2.53776789e-01 1.25320479e-01 6.70478940e-01 4.20359895e-02 3.67353171e-01 1.37462661e-01 -1.03980339e+00 6.87439978e-01 1.13014984e+00 -5.20269275e-02 -5.39384186e-01 -3.32733274e-01 6.26179739e-04 -7.64924288e-02 2.69349664e-01 6.68631852e-01 -1.96915790e-01 -4.39438760e-01 1.80892301e+00 1.34090826e-01 -2.82732576e-01 3.82148713e-01 6.29033625e-01 1.09361923e+00 7.09966362e-01 1.71096727e-01 3.99665087e-01 1.54222858e+00 -1.04567027e+00 -3.32460821e-01 -5.15556276e-01 9.92797256e-01 -2.04918563e-01 1.70737493e+00 3.64989161e-01 -1.06337547e+00 -5.27217686e-01 -1.03675020e+00 -4.75286305e-01 -5.84365368e-01 -7.86164552e-02 9.90998089e-01 3.36040765e-01 -1.16396999e+00 8.94293189e-02 -6.91309810e-01 -5.42915404e-01 1.42166778e-01 4.97148544e-01 -2.97914594e-01 -7.32790679e-02 -1.09451592e+00 8.26966643e-01 5.27610064e-01 -4.48887885e-01 -7.04964697e-01 -1.05846941e+00 -1.14573908e+00 1.69486940e-01 7.80107439e-01 -6.79556370e-01 1.91441798e+00 -1.04611707e+00 -1.30798185e+00 1.22504163e+00 -4.74435568e-01 -7.30745614e-01 -4.05472070e-02 -3.19047123e-01 -1.43469945e-01 -6.46927431e-02 2.35673800e-01 6.77337885e-01 2.79446300e-02 -1.28373420e+00 -5.34434855e-01 -4.81932938e-01 6.84920192e-01 -2.41388097e-01 -1.01389624e-01 5.12148440e-01 -4.70246822e-01 -6.00101911e-02 -2.22846359e-01 -6.73651755e-01 -1.15415245e-01 1.17789529e-01 -2.00891271e-01 -7.24533916e-01 3.32401633e-01 -6.90533519e-01 9.96434748e-01 -1.94011843e+00 -1.43760636e-01 9.59583670e-02 8.41950923e-02 -7.43947104e-02 -2.52612084e-01 2.88754493e-01 -1.97014526e-01 1.83607370e-01 -3.82063478e-01 9.65730026e-02 4.50731128e-01 2.48805583e-01 -5.59452355e-01 -1.88208312e-01 6.16186321e-01 1.06109107e+00 -8.39052439e-01 -4.01067138e-01 -3.57884288e-01 -2.10074022e-01 -1.07160640e+00 6.91009760e-01 -1.21627748e+00 -3.33854973e-01 -3.05550158e-01 7.29215205e-01 2.82168478e-01 -4.19985235e-01 2.87537485e-01 2.00027823e-01 1.80422008e-01 8.58442605e-01 -8.86940420e-01 2.00295234e+00 -7.91757762e-01 5.22355795e-01 2.48963043e-01 -1.12596035e+00 1.03193140e+00 1.25284314e-01 -7.95714855e-02 -9.46528673e-01 -3.11578363e-01 3.35245699e-01 -2.84291599e-02 -7.83818543e-01 1.90478757e-01 2.56450046e-02 -6.90254331e-01 6.17748320e-01 3.01308781e-01 -1.66068450e-01 3.07145834e-01 3.99150878e-01 1.49534035e+00 1.56251192e-01 1.99646041e-01 -3.23819965e-01 5.58766067e-01 5.64355493e-01 3.86106014e-01 9.90406275e-01 2.33033031e-01 2.62664676e-01 1.02225435e+00 -8.57555270e-01 -9.02847528e-01 -8.84429693e-01 1.27093747e-01 1.85543656e+00 -2.22405046e-01 -4.45766211e-01 -8.55152130e-01 -7.65447199e-01 1.25219196e-01 1.07781875e+00 -3.25429857e-01 -1.63755462e-01 -1.00185406e+00 -2.08779752e-01 7.79278338e-01 6.50894761e-01 2.90979773e-01 -1.51710558e+00 -4.67275470e-01 2.72348255e-01 -3.12875360e-02 -1.33675516e+00 -1.61263645e-01 5.05218625e-01 -6.94704056e-01 -1.30254734e+00 2.13457733e-01 -1.25284243e+00 6.93345606e-01 -4.84248817e-01 2.03958941e+00 4.41455752e-01 -1.49512798e-01 2.20470041e-01 1.34636881e-02 -5.65225363e-01 -7.18417883e-01 3.92996877e-01 -4.97403115e-01 -4.78463680e-01 9.99017894e-01 -3.94368231e-01 8.79652426e-02 -6.94819614e-02 -8.42541933e-01 3.58846858e-02 5.66937387e-01 9.46508169e-01 3.51634890e-01 -2.43447870e-01 3.97411674e-01 -1.65209270e+00 5.74765682e-01 -4.67630386e-01 -9.35574770e-01 4.11650181e-01 -6.14364922e-01 6.40452564e-01 1.05687165e+00 -1.25423148e-01 -1.13721013e+00 1.92451149e-01 -2.97166020e-01 1.27369761e-01 -6.00255549e-01 7.18393564e-01 -3.35169792e-01 2.13175952e-01 1.15661311e+00 1.06384590e-01 -1.83342043e-02 -3.85874480e-01 5.02413929e-01 5.35208762e-01 8.87611747e-01 -1.29590797e+00 6.81818724e-01 5.00050969e-02 -4.22065467e-01 -2.85740018e-01 -1.03023255e+00 -3.40230048e-01 -4.59295660e-01 5.82204163e-01 7.61612535e-01 -8.76998425e-01 -1.01836908e+00 8.06398764e-02 -1.50888121e+00 -8.38284016e-01 -1.90372512e-01 -1.82536289e-01 -6.34510398e-01 -1.69257104e-01 -7.62768328e-01 -4.89527643e-01 -4.88372475e-01 -1.00427341e+00 1.14245379e+00 -9.14832875e-02 -5.68061590e-01 -7.73834407e-01 -7.20361713e-03 5.03802121e-01 3.73258501e-01 4.52818535e-02 1.55920255e+00 -9.81184065e-01 -8.65637004e-01 -1.12040006e-01 -3.68763655e-01 3.33699226e-01 -2.92954057e-01 -2.53520489e-01 -7.80135155e-01 1.84117734e-01 -1.35434628e-01 -9.72349107e-01 5.06916046e-01 -1.08853646e-01 1.29386914e+00 -5.43757319e-01 -4.06320468e-02 4.66168910e-01 1.51420236e+00 2.08028257e-01 4.06751931e-01 4.24025565e-01 5.48405528e-01 6.27253294e-01 3.78468096e-01 -2.78068590e-03 8.50797117e-01 4.18177933e-01 3.79570097e-01 -5.57856001e-02 2.03636393e-01 -5.72911620e-01 3.52366626e-01 1.73461914e-01 5.72805643e-01 1.50155783e-01 -1.48904562e+00 5.95678449e-01 -1.83375239e+00 -6.03630304e-01 -3.25884670e-02 1.95573723e+00 1.23880064e+00 4.78683203e-01 2.84197815e-02 -2.61763722e-01 3.75672519e-01 -8.42845887e-02 -6.63923919e-01 -8.49397242e-01 2.45200098e-01 7.01565325e-01 4.10734713e-01 7.92829573e-01 -9.38112736e-01 1.10242879e+00 6.02718306e+00 1.08158000e-01 -8.44821990e-01 -1.55790532e-02 4.43487376e-01 2.01541990e-01 -4.85117137e-01 1.94365323e-01 -9.64258790e-01 3.37168314e-02 1.54021037e+00 -1.06679574e-01 6.77510440e-01 1.24409044e+00 -2.35825300e-01 -5.29057533e-02 -1.77324140e+00 4.82785970e-01 1.65630281e-01 -1.31840062e+00 1.74492989e-02 -5.06189346e-01 3.12492251e-01 2.33272444e-02 -3.83409470e-01 9.33866262e-01 8.94844711e-01 -1.15173411e+00 6.47574782e-01 4.18716431e-01 6.61268592e-01 -7.09975481e-01 7.00940132e-01 6.08145773e-01 -8.43289852e-01 -2.23384708e-01 -1.68221086e-01 -2.31548384e-01 -2.75279403e-01 3.22419405e-01 -9.15181458e-01 8.22273418e-02 7.57868767e-01 3.98647219e-01 -7.11269796e-01 3.18182975e-01 -3.86258036e-01 7.36735523e-01 -3.88947517e-01 -4.11295265e-01 1.05085537e-01 2.21259296e-01 -1.69222150e-02 1.26813924e+00 -2.15875491e-01 1.36871129e-01 2.95636654e-01 1.28819001e+00 -4.95968461e-01 -4.31446359e-02 -7.45057106e-01 -1.23704009e-01 3.06209981e-01 7.93752432e-01 -5.28356843e-02 -7.04306543e-01 -7.49030828e-01 5.66483021e-01 8.37016284e-01 3.97665948e-01 -6.58384562e-01 -6.92485332e-01 3.75294030e-01 1.84537321e-01 4.26132604e-02 1.99087918e-01 -6.70540988e-01 -1.07011151e+00 6.76850021e-01 -1.53109419e+00 7.08309114e-01 -7.76751935e-01 -1.26248538e+00 5.31713307e-01 -4.96912301e-02 -3.55546683e-01 -5.43318748e-01 -8.57012808e-01 -6.15424812e-01 9.97520268e-01 -1.41158664e+00 -1.00294352e+00 -2.69025356e-01 3.45026374e-01 4.37096506e-01 -1.91824287e-01 1.21566427e+00 3.35360497e-01 -3.31479967e-01 7.55812109e-01 -5.57723880e-01 4.86343056e-01 6.37742043e-01 -1.59668994e+00 7.85209835e-01 6.49872184e-01 -1.04816467e-01 1.04029357e+00 5.17470837e-01 -2.20135525e-01 -1.97671866e+00 -1.16045296e+00 1.35973823e+00 -8.52060437e-01 9.36831832e-01 -6.88332200e-01 -1.25394952e+00 1.06312120e+00 2.38139078e-01 1.99010402e-01 5.58095515e-01 5.83000600e-01 -8.33661556e-01 -3.94716799e-01 -9.31778133e-01 3.62082899e-01 9.71472561e-01 -9.67614174e-01 -8.31368983e-01 4.93198127e-01 1.24682820e+00 -6.34217739e-01 -8.04308653e-01 4.38191205e-01 3.04222465e-01 -7.40545869e-01 8.76708329e-01 -1.25444567e+00 8.39148283e-01 -1.25358671e-01 -1.85806721e-01 -7.97863305e-01 9.01720747e-02 -3.87096077e-01 -2.44803801e-01 1.12396204e+00 9.99774098e-01 -5.94569266e-01 1.03353775e+00 1.13877380e+00 -1.41030416e-01 -7.83748865e-01 -3.34449530e-01 -4.45854872e-01 2.55856007e-01 -5.55734396e-01 6.41168952e-01 7.44692922e-01 2.89887011e-01 7.84592569e-01 2.80723661e-01 2.07197323e-01 4.75898862e-01 4.95141804e-01 1.08509147e+00 -1.17891181e+00 -6.94207788e-01 -4.20139790e-01 6.06765337e-02 -1.20877445e+00 7.19334781e-01 -1.10386479e+00 2.62438238e-01 -1.40716040e+00 2.29320437e-01 -4.27796394e-01 2.30972115e-02 8.24036002e-01 -1.83844656e-01 -6.15678012e-01 -1.80651933e-01 -2.21438900e-01 -7.00624228e-01 1.59456022e-02 4.96723741e-01 -4.45340067e-01 -1.15589447e-01 -9.57198814e-02 -1.19668031e+00 5.61135232e-01 7.95061529e-01 -6.35301828e-01 -4.30883676e-01 -8.14438999e-01 4.88455504e-01 3.41222435e-01 3.96969169e-01 -6.73878074e-01 5.44223189e-01 -6.70113191e-02 -3.39121938e-01 -2.67395169e-01 -9.69056264e-02 -7.58396864e-01 -2.81605721e-01 4.24694180e-01 -9.37669098e-01 3.78033012e-01 3.98484707e-01 2.41355181e-01 -4.22925174e-01 -4.34688002e-01 5.24229705e-01 -4.19420511e-01 -9.36715961e-01 -4.14626449e-02 -1.32706985e-01 5.73070467e-01 6.59883082e-01 4.15968418e-01 -5.18805504e-01 -1.56943887e-01 -4.01500493e-01 4.35758352e-01 3.61625612e-01 5.24880052e-01 2.41291732e-01 -8.77525389e-01 -6.56374514e-01 1.54731780e-01 5.59931397e-01 3.47501099e-01 -5.45515478e-01 1.47043690e-01 -8.53683531e-01 6.05104327e-01 1.85313165e-01 -5.47208905e-01 -1.01066446e+00 6.53099239e-01 4.49171603e-01 -6.05978549e-01 -3.03608954e-01 9.08596992e-01 6.23648204e-02 -1.30133438e+00 5.32437921e-01 -6.93929076e-01 1.94009304e-01 -5.98521948e-01 4.73214269e-01 -3.33487719e-01 3.47202986e-01 1.48087785e-01 -3.87015402e-01 6.88279793e-02 -1.96206540e-01 8.79755691e-02 1.39947629e+00 7.67588675e-01 -6.10455096e-01 4.74820465e-01 1.32327056e+00 -4.10223424e-01 -6.23679161e-01 -6.34321630e-01 5.69214821e-01 1.37232244e-02 -2.93289989e-01 -9.75343764e-01 -7.05658317e-01 9.34151411e-01 1.48140237e-01 8.75436887e-02 7.84184992e-01 3.61435384e-01 7.80631542e-01 1.36586273e+00 4.70497668e-01 -7.09009111e-01 8.71473029e-02 8.77504528e-01 7.79522359e-01 -1.57064176e+00 -4.83052999e-01 -4.56611335e-01 -2.74843037e-01 9.61819410e-01 1.36949468e+00 -1.78007185e-01 2.81985283e-01 7.49355376e-01 8.83823037e-02 -3.72641146e-01 -1.31533957e+00 8.52156878e-02 -1.59417883e-01 4.30584431e-01 7.08094478e-01 -8.58393833e-02 3.69161591e-02 8.60304415e-01 -3.29350084e-01 2.06386447e-01 2.53825307e-01 1.28818297e+00 -3.01548243e-01 -1.18965292e+00 -1.24237090e-01 5.34329593e-01 -3.71072859e-01 -3.83185804e-01 -4.98453796e-01 7.24742830e-01 -2.68915683e-01 7.55474567e-01 1.64702654e-01 -1.23302780e-01 8.11925709e-01 4.38193321e-01 2.16956869e-01 -1.09510946e+00 -8.73697639e-01 -8.57491374e-01 5.64206898e-01 -6.82370663e-01 1.22716539e-01 -2.74456263e-01 -1.69658208e+00 -2.45904654e-01 -5.22479834e-03 3.28524828e-01 4.19066072e-01 6.46995425e-01 5.95108688e-01 3.60468715e-01 1.83283299e-01 -5.24042211e-02 -8.55263710e-01 -5.92423618e-01 8.11980367e-02 6.27160609e-01 3.65807950e-01 -4.31532320e-03 -1.56458750e-01 3.01708311e-01]
[9.653413772583008, 7.628544807434082]
fda1c7bd-d331-49b8-bcf1-20b635caf6cc
vision-deduction-and-alignment-an-empirical
2302.08774
null
https://arxiv.org/abs/2302.08774v2
https://arxiv.org/pdf/2302.08774v2.pdf
Vision, Deduction and Alignment: An Empirical Study on Multi-modal Knowledge Graph Alignment
Entity alignment (EA) for knowledge graphs (KGs) plays a critical role in knowledge engineering. Existing EA methods mostly focus on utilizing the graph structures and entity attributes (including literals), but ignore images that are common in modern multi-modal KGs. In this study we first constructed Multi-OpenEA -- eight large-scale, image-equipped EA benchmarks, and then evaluated some existing embedding-based methods for utilizing images. In view of the complementary nature of visual modal information and logical deduction, we further developed a new multi-modal EA method named LODEME using logical deduction and multi-modal KG embedding, with state-of-the-art performance achieved on Multi-OpenEA and other existing multi-modal EA benchmarks.
['Hai-Tao Zheng', 'Xi Chen', 'Yuejia Xiang', 'Yinghui Li', 'Jiaoyan Chen', 'Yangning Li']
2023-02-17
null
null
null
null
['entity-alignment', 'multi-modal-knowledge-graph', 'entity-alignment']
['knowledge-base', 'knowledge-base', 'natural-language-processing']
[-3.80634815e-01 7.39949718e-02 -3.47380012e-01 -5.99446567e-03 -4.27901089e-01 -5.73894262e-01 3.24368924e-01 3.64549845e-01 -1.12937979e-01 5.45444906e-01 3.30390394e-01 -1.51081353e-01 -3.92013550e-01 -1.39693582e+00 -7.63841391e-01 -3.18904519e-01 -2.83812344e-01 5.14723718e-01 2.93890357e-01 -3.81077409e-01 -1.79920569e-01 4.61469591e-01 -1.52097058e+00 4.40959483e-01 8.45993161e-01 1.00818193e+00 -1.98330134e-01 4.24094349e-01 -5.11345744e-01 1.12790596e+00 -2.40276888e-01 -1.05784273e+00 -7.58911250e-03 -4.18893732e-02 -1.02944505e+00 -1.62204072e-01 8.89213383e-01 7.69710774e-03 -6.06268466e-01 1.17601407e+00 5.19353867e-01 1.85203273e-02 7.06767082e-01 -1.71658564e+00 -1.37839997e+00 7.25142539e-01 -7.46441543e-01 2.12714244e-02 5.21816373e-01 3.23036201e-02 1.48742473e+00 -1.02963102e+00 1.22195423e+00 1.27228570e+00 7.27270722e-01 1.88464709e-02 -1.13437760e+00 -4.85349029e-01 3.05063110e-02 1.02887928e+00 -1.66891360e+00 -9.82468128e-02 1.00128484e+00 -3.13548595e-01 1.17341995e+00 1.02856271e-01 1.03584540e+00 7.11845934e-01 1.94919616e-01 8.89816701e-01 1.02810645e+00 -5.87298930e-01 -1.56680852e-01 1.01864628e-01 9.16830003e-02 1.23387527e+00 4.70779806e-01 -4.42638606e-01 -7.05038846e-01 1.30382508e-01 6.69106305e-01 -4.29306418e-01 -3.87706906e-01 -9.33769345e-01 -1.60911298e+00 8.26829791e-01 8.30368519e-01 8.97048116e-02 -3.66029948e-01 2.19801575e-01 5.95300853e-01 3.10101837e-01 4.00103591e-02 2.02794403e-01 -2.67407060e-01 2.27872774e-01 -5.44234633e-01 1.24732815e-02 5.97330034e-01 1.26590359e+00 1.03395474e+00 6.97803497e-02 -1.38505206e-01 6.24397695e-01 2.29118913e-01 5.45170844e-01 6.01506159e-02 -6.14145517e-01 5.87610722e-01 1.31196082e+00 -5.00034988e-01 -1.47789586e+00 -3.91024500e-01 -3.25114369e-01 -8.29641700e-01 8.80994275e-03 -4.95872162e-02 4.49104100e-01 -9.11778152e-01 1.54299092e+00 4.75083232e-01 1.81933150e-01 4.61361587e-01 5.65758407e-01 1.33271039e+00 3.49653244e-01 1.40382543e-01 3.70289646e-02 1.50178552e+00 -1.11189163e+00 -8.26669097e-01 3.76431048e-02 8.19090664e-01 -5.16346633e-01 1.35183895e+00 7.72511726e-03 -8.22987378e-01 -3.23572665e-01 -1.08785594e+00 -4.86743182e-01 -1.23778164e+00 3.86091247e-02 7.69304693e-01 3.41775686e-01 -1.05068886e+00 1.42045328e-02 -2.60247558e-01 -3.70068252e-01 5.82151353e-01 -1.36030183e-04 -9.39993739e-01 -4.11015004e-01 -1.45031750e+00 8.81473958e-01 8.23316514e-01 3.93886827e-02 -3.36478204e-01 -9.65183973e-01 -1.16436863e+00 8.08065236e-02 8.09334874e-01 -1.13434041e+00 4.44874644e-01 -3.70466799e-01 -8.52622867e-01 8.85526955e-01 2.40218341e-01 -1.13154007e-02 2.63711274e-01 -1.99401602e-01 -1.12458849e+00 4.68719274e-01 1.00223228e-01 7.00796902e-01 5.40899456e-01 -1.57427585e+00 -4.15987790e-01 -4.70434517e-01 5.00803351e-01 2.79972583e-01 -7.32805967e-01 -4.61018741e-01 -8.70414615e-01 -5.34437895e-01 -1.37448832e-01 -5.68757415e-01 2.47188672e-01 3.54766957e-02 -7.51214921e-01 -6.02200814e-02 1.02738774e+00 -8.51478100e-01 1.48997760e+00 -2.02172446e+00 5.56768715e-01 4.48577076e-01 6.02382898e-01 -1.93610266e-02 -9.00512636e-02 5.56418180e-01 -2.10159451e-01 1.42248934e-02 3.40278409e-02 1.09072551e-01 3.72834444e-01 1.82655841e-01 -1.13155313e-01 3.41090374e-02 -4.66389507e-02 1.53289711e+00 -1.05313504e+00 -1.30290687e+00 3.07049304e-01 5.13144732e-01 -3.47650856e-01 -2.43132606e-01 -3.50236923e-01 -3.73897284e-01 -3.00842255e-01 1.21380782e+00 6.11949742e-01 -5.86108506e-01 4.83746469e-01 -1.24665856e+00 3.15849036e-01 -4.73647773e-01 -1.32673609e+00 1.82044327e+00 -2.85939157e-01 4.81810987e-01 -3.67790073e-01 -6.03624940e-01 4.03254986e-01 -5.83351441e-02 5.19562542e-01 -8.17452490e-01 -1.24067664e-02 2.09974367e-02 -3.27801198e-01 -3.04131657e-01 9.53439295e-01 1.99686483e-01 -1.42300636e-01 -1.98225845e-02 2.91297108e-01 2.06485435e-01 4.21487927e-01 8.49362075e-01 1.00105333e+00 2.09405333e-01 4.78657216e-01 -8.85735452e-02 4.21708733e-01 4.39523757e-01 2.66013741e-01 1.68610394e-01 6.69798180e-02 1.92040741e-01 7.32029736e-01 -3.13622117e-01 -8.12906146e-01 -1.32744741e+00 2.04416458e-02 6.14269912e-01 5.35049140e-01 -1.19367206e+00 -3.94392043e-01 -1.03545976e+00 3.57388556e-01 6.50109649e-01 -6.51630938e-01 -2.69883096e-01 -2.15858877e-01 -4.83878702e-01 6.45762265e-01 7.99772501e-01 7.79497981e-01 -9.45586741e-01 -2.84126937e-01 5.35372198e-02 -2.36715138e-01 -1.38515937e+00 -2.38715231e-01 -1.91347867e-01 -3.39966714e-01 -1.49549890e+00 -4.74397540e-01 -8.71862650e-01 7.16636062e-01 1.20654762e-01 1.60340250e+00 -7.76848942e-02 -4.15150404e-01 9.39427733e-01 -3.76326621e-01 -1.88618079e-01 -8.13800283e-03 -2.76072621e-02 -1.83278024e-01 -1.68708473e-01 4.29570794e-01 -5.06195128e-01 -4.70341474e-01 1.09661043e-01 -1.01910043e+00 3.91669929e-01 8.29421878e-01 7.88688362e-01 1.06606150e+00 4.61332053e-01 4.69028831e-01 -1.04935622e+00 4.45811540e-01 -3.17392677e-01 -3.31703603e-01 1.11967313e+00 -6.75348401e-01 3.29143777e-02 4.47701782e-01 -1.50806054e-01 -8.35603297e-01 -3.11053246e-01 1.48183227e-01 -8.29862535e-01 2.71504104e-01 1.11011243e+00 -5.98196447e-01 -3.85692805e-01 3.33492279e-01 -1.73875578e-02 -3.38689923e-01 -2.09145650e-01 1.02473843e+00 2.32952416e-01 8.47885609e-01 -7.71157682e-01 9.56711113e-01 3.23796511e-01 4.37015086e-01 -6.36970222e-01 -5.56799650e-01 -2.55283713e-01 -7.13020205e-01 -3.59708458e-01 9.78087366e-01 -1.07104015e+00 -6.80319130e-01 3.83835614e-01 -8.28396261e-01 6.86024688e-03 -4.78559554e-01 1.72832668e-01 -4.02538389e-01 6.33044958e-01 -5.13924241e-01 -4.65460606e-02 -3.67573649e-01 -7.49433219e-01 9.16851580e-01 7.30712488e-02 2.53014803e-01 -1.16271257e+00 3.00616529e-02 5.15433311e-01 1.64319053e-01 5.38010478e-01 1.62140799e+00 -1.51381865e-01 -7.55804360e-01 -2.60740817e-02 -6.36001289e-01 1.06601641e-01 3.49631384e-02 1.40370548e-01 -5.67248940e-01 -1.47543728e-01 -1.01981091e+00 -4.78192806e-01 7.88890719e-01 -6.42705858e-02 1.06968367e+00 -8.05687383e-02 -4.93262261e-01 6.39804125e-01 2.02645326e+00 -2.42971390e-01 1.00239229e+00 6.61329985e-01 1.44788206e+00 2.94537485e-01 8.28889012e-01 1.54992044e-01 1.07574046e+00 7.04414606e-01 5.57823360e-01 -1.71070248e-01 -3.26430231e-01 -4.74074036e-01 2.30565533e-01 1.25529981e+00 -1.09093800e-01 -3.02302241e-01 -9.92310345e-01 9.15172160e-01 -2.04087853e+00 -8.62303913e-01 -3.67639542e-01 1.66273916e+00 8.25540006e-01 -5.97867891e-02 5.82932122e-02 -3.97698283e-02 6.57296717e-01 2.10627690e-01 -4.16305751e-01 -9.71311238e-03 -7.49601722e-01 1.07873768e-01 5.71107805e-01 -1.00247376e-02 -1.03982008e+00 9.57215965e-01 5.73682213e+00 1.07615805e+00 -6.55408084e-01 -1.12599200e-02 9.93254259e-02 1.08847111e-01 -7.97470272e-01 -6.94974586e-02 -5.91187835e-01 4.87216525e-02 5.29551625e-01 -3.34633678e-01 1.97387293e-01 8.66948307e-01 -8.33382070e-01 -4.79228143e-03 -9.77705121e-01 1.19618666e+00 1.32091492e-01 -1.80320358e+00 5.43307006e-01 4.57146205e-02 7.72283256e-01 -4.23536241e-01 -1.27077550e-01 3.51033092e-01 3.63981754e-01 -9.34837043e-01 6.13650441e-01 5.72312593e-01 1.24148107e+00 -9.29111838e-01 8.00166368e-01 -4.55780983e-01 -1.95343578e+00 3.13559234e-01 -2.29923472e-01 6.50734067e-01 3.02572101e-01 4.94710892e-01 -2.81309456e-01 1.60346413e+00 9.81640697e-01 1.05504680e+00 -1.07377946e+00 6.69286191e-01 -4.18297201e-01 5.62113151e-02 -1.23005383e-01 4.75432754e-01 2.28984393e-02 -1.08999461e-01 3.17655772e-01 1.22408092e+00 2.85359979e-01 -3.11632365e-01 -4.26636226e-02 6.71220183e-01 -4.39964205e-01 2.64522314e-01 -8.07736754e-01 -4.28455412e-01 4.66606200e-01 1.51797128e+00 -7.29746819e-01 -4.04766142e-01 -8.98939669e-01 8.67474616e-01 5.64817429e-01 3.55487019e-01 -1.12456524e+00 -5.95160544e-01 5.91234446e-01 -3.61908674e-02 5.58615506e-01 -1.05886981e-01 6.42425641e-02 -1.38213277e+00 2.37033889e-01 -7.78259158e-01 9.79841173e-01 -1.19378734e+00 -1.40290153e+00 4.86271739e-01 3.20353925e-01 -1.10958457e+00 2.07023278e-01 -7.87732661e-01 -2.01667294e-01 4.50014770e-01 -1.93396568e+00 -2.00996852e+00 -5.79981983e-01 1.02136683e+00 -9.75019187e-02 -2.35951364e-01 8.78201783e-01 6.26991689e-01 -6.27752244e-01 6.61294758e-01 4.37810272e-02 1.88884065e-01 7.62513578e-01 -1.67644429e+00 7.45215937e-02 8.03781092e-01 5.20406425e-01 6.18080258e-01 2.42433071e-01 -7.55106449e-01 -1.92516303e+00 -1.18324852e+00 4.14421290e-01 -3.79951328e-01 9.79833901e-01 -1.18706793e-01 -9.45072114e-01 1.10731709e+00 4.15275544e-01 3.25773567e-01 7.57791758e-01 1.28618315e-01 -6.50809705e-01 -1.41531602e-01 -1.01137781e+00 6.93382382e-01 1.38715982e+00 -7.91265786e-01 -6.14592612e-01 1.23666562e-01 8.61911714e-01 -5.23063064e-01 -1.57715487e+00 6.48983538e-01 5.81927836e-01 -6.05064273e-01 1.43436420e+00 -7.15422511e-01 5.60561359e-01 -6.51417077e-01 -3.80058348e-01 -1.45423675e+00 -5.30992508e-01 -1.12306718e-02 -7.60403097e-01 1.52707970e+00 1.52344808e-01 -4.33008254e-01 7.08457232e-01 1.95606217e-01 -1.99713439e-01 -8.20341647e-01 -6.15723491e-01 -8.79982710e-01 -2.85755515e-01 -1.51702717e-01 8.12090218e-01 1.36124563e+00 8.32833536e-03 3.99574250e-01 -2.78047860e-01 4.24006969e-01 7.29375541e-01 5.20696461e-01 8.31074953e-01 -1.00836694e+00 -2.43777912e-02 -1.84003860e-01 -1.12651336e+00 -3.15752983e-01 1.44242272e-01 -1.07999611e+00 -6.64529800e-01 -2.06084871e+00 3.10441971e-01 -6.54031783e-02 -6.37828290e-01 9.24700737e-01 -3.24773788e-02 4.18491662e-01 1.86207250e-01 -1.16425291e-01 -1.00673366e+00 7.02219903e-01 1.21240604e+00 -5.40163636e-01 2.01423377e-01 -1.23794484e+00 -6.34309471e-01 4.68699157e-01 2.75133342e-01 -2.10111737e-02 -7.75245845e-01 -3.26102674e-01 8.63260746e-01 -2.66753167e-01 6.56313598e-01 -9.14842308e-01 4.35314029e-01 -1.43883169e-01 3.48546714e-01 -7.74498582e-01 2.07173735e-01 -1.12077153e+00 6.98732018e-01 -9.02603567e-02 3.05482924e-01 3.03948313e-01 5.64112604e-01 7.06590235e-01 -6.77268386e-01 2.08731323e-01 2.44855255e-01 -3.42579624e-05 -1.84476960e+00 2.61487961e-01 3.78136575e-01 2.81163305e-01 1.39292467e+00 -4.12426949e-01 -1.09446549e+00 -2.91810688e-02 -6.47997797e-01 5.50528049e-01 6.98066711e-01 3.62033188e-01 8.63567293e-01 -1.91087937e+00 -3.45176518e-01 -9.54507142e-02 8.80721867e-01 9.30996165e-02 5.66598415e-01 9.43772733e-01 -7.76439846e-01 1.55152887e-01 -6.70951366e-01 -3.08304816e-01 -1.44802082e+00 7.66468287e-01 -2.79689208e-02 -4.32724506e-01 -8.19372594e-01 5.88456988e-01 2.63249099e-01 -3.06429595e-01 -6.75351247e-02 -6.24189116e-02 -2.24320874e-01 3.00241828e-01 1.32282525e-01 4.75665152e-01 1.04077771e-01 -8.39518070e-01 -5.85414886e-01 7.63788521e-01 -1.02032371e-01 3.92193675e-01 1.32890701e+00 1.41232898e-02 -3.23574126e-01 4.82129782e-01 1.06117094e+00 2.72123784e-01 -5.52348018e-01 -4.27886605e-01 -1.10178158e-01 -5.55317700e-01 1.69357553e-01 -8.34410012e-01 -1.36228919e+00 6.08866215e-01 3.43622774e-01 4.96211536e-02 1.36093557e+00 9.27300453e-02 8.79803956e-01 4.19939250e-01 6.08664751e-01 -1.15649736e+00 2.98714191e-01 1.33360490e-01 8.48926663e-01 -1.13011372e+00 3.98532510e-01 -9.46002901e-01 -9.07571971e-01 1.10513568e+00 9.57850456e-01 4.24986243e-01 5.63681662e-01 -1.73090259e-03 -9.60115790e-02 -9.84708726e-01 -4.44150597e-01 -3.77233952e-01 6.65115476e-01 6.88456535e-01 -6.75898343e-02 1.30878478e-01 9.80508178e-02 5.22843540e-01 1.53705394e-02 6.36903718e-02 2.06533998e-01 1.15372527e+00 5.05710654e-02 -1.03924048e+00 8.14112872e-02 5.64363182e-01 7.95692429e-02 -1.93568632e-01 -5.10897756e-01 1.32944167e+00 2.26578504e-01 3.24412942e-01 -2.97867507e-01 -6.71355307e-01 4.49616969e-01 5.53331859e-02 6.32162154e-01 -2.29087010e-01 -2.42896333e-01 -5.44174254e-01 3.88369709e-01 -5.95747828e-01 -6.26137435e-01 -3.02243173e-01 -1.30018437e+00 -6.90688372e-01 -3.44399482e-01 -2.67349899e-01 2.41413161e-01 5.51749587e-01 6.56706393e-01 7.97320068e-01 -1.12195149e-01 -2.04422548e-01 2.48703644e-01 -3.65495801e-01 -9.24485266e-01 6.39279485e-01 -2.70643741e-01 -9.67960715e-01 2.22331826e-02 1.23303466e-01]
[8.692697525024414, 7.732250213623047]
be6a0186-a83a-47d0-9aa9-e75277bebd46
learning-invariant-visual-representations-for
2206.00415
null
https://arxiv.org/abs/2206.00415v3
https://arxiv.org/pdf/2206.00415v3.pdf
Learning Invariant Visual Representations for Compositional Zero-Shot Learning
Compositional Zero-Shot Learning (CZSL) aims to recognize novel compositions using knowledge learned from seen attribute-object compositions in the training set. Previous works mainly project an image and a composition into a common embedding space to measure their compatibility score. However, both attributes and objects share the visual representations learned above, leading the model to exploit spurious correlations and bias towards seen pairs. Instead, we reconsider CZSL as an out-of-distribution generalization problem. If an object is treated as a domain, we can learn object-invariant features to recognize the attributes attached to any object reliably. Similarly, attribute-invariant features can also be learned when recognizing the objects with attributes as domains. Specifically, we propose an invariant feature learning framework to align different domains at the representation and gradient levels to capture the intrinsic characteristics associated with the tasks. Experiments on two CZSL benchmarks demonstrate that the proposed method significantly outperforms the previous state-of-the-art.
['Jun Guo', 'Zhanyu Ma', 'Xian Sun', 'Ruoyi Du', 'Kongming Liang', 'Tian Zhang']
2022-06-01
null
null
null
null
['compositional-zero-shot-learning']
['computer-vision']
[ 2.87868530e-01 -1.95484683e-01 -3.85776520e-01 -5.89948535e-01 -6.15404725e-01 -5.56495726e-01 8.24802935e-01 6.17807396e-02 -1.54966488e-01 3.11579198e-01 2.76037782e-01 5.37144959e-01 4.46294853e-03 -7.12104559e-01 -7.53172755e-01 -1.02908266e+00 1.30109787e-01 4.87164021e-01 1.78642452e-01 9.70852375e-02 1.84943065e-01 2.98396349e-01 -1.70598948e+00 6.28090203e-01 7.61936724e-01 1.26437569e+00 1.85898349e-01 9.37239826e-02 -5.10166228e-01 7.09340930e-01 -3.31243247e-01 -4.00115252e-01 5.00007391e-01 -5.09837508e-01 -5.97298324e-01 3.44449699e-01 9.65033114e-01 -1.24119762e-02 -2.55352914e-01 1.34349227e+00 2.61218399e-01 1.94795668e-01 1.08886707e+00 -1.58933425e+00 -1.29903293e+00 3.73817921e-01 -5.33729732e-01 -1.29725903e-01 3.96338046e-01 2.56058574e-01 1.53835905e+00 -1.34026778e+00 9.25184786e-01 1.11080873e+00 2.26499021e-01 5.85003018e-01 -1.63661087e+00 -7.98126400e-01 3.91177118e-01 6.91583812e-01 -1.37723708e+00 -3.57850611e-01 1.06819904e+00 -7.73293078e-01 5.35034120e-01 1.35146990e-01 6.33807838e-01 1.34384537e+00 -3.04842085e-01 9.55935359e-01 1.05995095e+00 -1.54221654e-01 4.85145211e-01 3.58477265e-01 6.81270733e-02 4.50479120e-01 2.59370178e-01 4.14233878e-02 -8.29545975e-01 -7.32280836e-02 3.60408306e-01 2.46544167e-01 -1.37358353e-01 -1.34861112e+00 -1.40547442e+00 6.82720482e-01 4.98821139e-01 1.67575240e-01 -3.03498387e-01 -1.85672894e-01 4.35585499e-01 5.34010887e-01 3.28069299e-01 5.10929465e-01 -2.82007158e-01 1.92411035e-01 -3.23025435e-01 3.27234864e-01 5.23761928e-01 1.42128551e+00 1.10256433e+00 3.52860317e-02 -3.99851471e-01 8.73999000e-01 8.65813419e-02 3.35914403e-01 4.06514943e-01 -7.34011769e-01 1.92769989e-01 9.16949213e-01 -1.89445540e-01 -7.85576284e-01 1.52171969e-01 -3.27137470e-01 -6.92416370e-01 3.57574373e-01 2.37031117e-01 5.55726647e-01 -7.31660843e-01 1.92500532e+00 3.87077332e-01 3.74764323e-01 1.39538914e-01 1.03349936e+00 6.95310712e-01 3.77399355e-01 1.29381016e-01 8.12885314e-02 1.25751495e+00 -9.65493619e-01 -5.29239118e-01 -2.43896008e-01 3.26129943e-01 -5.96513748e-01 1.31164932e+00 1.39323890e-01 -7.91118562e-01 -7.28491664e-01 -1.33319283e+00 -5.22863586e-03 -4.92380083e-01 -2.49726444e-01 2.87385613e-01 2.56814301e-01 -3.96343589e-01 5.63160360e-01 -3.55822682e-01 -2.89256901e-01 6.14744186e-01 7.65398890e-02 -5.67070007e-01 -3.53189856e-01 -8.32164347e-01 7.08497405e-01 3.64993721e-01 -6.40632808e-01 -1.11798966e+00 -9.36428010e-01 -1.01424587e+00 1.87560797e-01 5.19395173e-01 -4.60858166e-01 8.65790784e-01 -1.33629549e+00 -1.11879230e+00 1.05658960e+00 -1.84452519e-01 1.44292444e-01 2.82454669e-01 -2.57844571e-02 -4.75769430e-01 -1.52754495e-02 2.30907962e-01 4.04405683e-01 1.21814930e+00 -1.49907660e+00 -6.22394502e-01 -4.57875609e-01 -1.39307722e-01 1.26727208e-01 -5.78682601e-01 -1.43609613e-01 -4.38437201e-02 -7.52957940e-01 3.10074657e-01 -7.37006962e-01 3.24964613e-01 4.33400929e-01 -1.08235531e-01 -2.86705583e-01 8.63398671e-01 -2.26659209e-01 4.96052027e-01 -2.62836909e+00 5.42304039e-01 1.48562253e-01 2.40816101e-01 -9.24934223e-02 -4.57679510e-01 3.71566653e-01 -2.75141567e-01 -2.18196064e-01 -2.23564848e-01 -1.85501263e-01 1.66227460e-01 3.03010851e-01 -3.62070173e-01 6.61273897e-01 5.47438622e-01 8.02006185e-01 -1.22693026e+00 -5.00692368e-01 5.28263897e-02 2.29503527e-01 -4.25912261e-01 5.35052180e-01 -6.11784570e-02 5.26508629e-01 -2.81833321e-01 8.37390304e-01 7.13991046e-01 -1.65375054e-01 3.50052595e-01 -3.59821975e-01 3.18318993e-01 5.30452915e-02 -1.27204120e+00 1.79928398e+00 -3.55421215e-01 3.63579869e-01 -3.26065034e-01 -1.25630307e+00 1.03013921e+00 1.37461752e-01 4.64059532e-01 -7.09370852e-01 -1.31729424e-01 3.21787000e-01 -2.46545393e-03 -5.42144835e-01 1.14671730e-01 -3.32717091e-01 2.16787420e-02 3.31381500e-01 7.46069014e-01 1.21018849e-01 -1.54155388e-01 7.84342438e-02 8.55261683e-01 1.41395792e-01 4.85253781e-01 -3.19133312e-01 5.84406495e-01 -3.89995545e-01 8.48368049e-01 5.43022573e-01 -3.03115934e-01 6.26268625e-01 3.26534659e-01 -5.19653618e-01 -1.30579340e+00 -1.70455933e+00 -9.45554748e-02 1.43172836e+00 5.84028304e-01 -2.56435186e-01 -1.57884538e-01 -1.01117623e+00 3.54087830e-01 6.57494783e-01 -7.47465193e-01 -3.82310897e-01 -2.79495209e-01 -3.40715609e-02 -2.07047462e-02 7.04727352e-01 1.97186202e-01 -8.80899370e-01 -3.86135370e-01 1.03361875e-01 7.85350129e-02 -1.10971737e+00 -5.54156005e-01 2.22630411e-01 -5.16569018e-01 -1.12486625e+00 -6.89165294e-01 -1.08281875e+00 7.42967963e-01 3.50784153e-01 1.12773561e+00 -2.57883161e-01 -4.28352028e-01 5.13461471e-01 -5.06182909e-01 -2.28693336e-01 -2.47285664e-01 -3.15353483e-01 3.35800618e-01 7.01639295e-01 7.29451239e-01 -7.33587027e-01 -5.55803478e-01 3.55275899e-01 -7.74948597e-01 -6.39317259e-02 7.14897335e-01 1.25141096e+00 7.08040714e-01 -5.01943946e-01 6.07154667e-01 -8.23680758e-01 1.68300360e-01 -4.52169687e-01 -2.46074602e-01 5.70367277e-01 -7.05259681e-01 3.54343802e-01 6.84775829e-01 -8.44729960e-01 -9.63725448e-01 2.64970750e-01 5.16001582e-01 -8.53573263e-01 -3.09247583e-01 -3.25014591e-02 -6.67701960e-01 -8.60058796e-03 5.23366034e-01 4.75684226e-01 1.00364164e-01 -4.56054330e-01 6.46480143e-01 4.11766261e-01 4.78214055e-01 -8.95068824e-01 9.70418572e-01 6.91783845e-01 -7.75997937e-02 -6.59216464e-01 -8.34327936e-01 -7.86738455e-01 -8.60318124e-01 -1.00594699e-01 8.34943771e-01 -9.50465858e-01 -4.88455683e-01 2.63841957e-01 -9.50783193e-01 1.86004654e-01 -8.05852056e-01 4.03198242e-01 -7.87060201e-01 2.96778649e-01 -1.49725288e-01 -5.38725197e-01 4.06978354e-02 -1.08733010e+00 1.09541988e+00 -1.01638352e-02 -2.67987877e-01 -7.02110827e-01 1.78798318e-01 -1.24584138e-01 2.29336485e-01 6.34218156e-02 1.31002522e+00 -1.00213480e+00 -6.40257716e-01 1.02998028e-02 -3.28150392e-01 4.31168973e-01 3.33344191e-01 -2.23434716e-01 -1.18149543e+00 -4.06462550e-01 7.68399313e-02 -5.49197078e-01 9.30218160e-01 -2.21125349e-01 1.22862840e+00 -3.33269507e-01 -2.17592835e-01 7.49460697e-01 1.41328442e+00 1.37878666e-02 4.18633223e-01 1.47686228e-01 7.57113218e-01 8.40171576e-01 5.85242093e-01 4.68404114e-01 8.20235386e-02 7.32452869e-01 3.82628381e-01 1.59261882e-01 -1.99444547e-01 -5.39352596e-01 3.51333261e-01 7.09611058e-01 1.31011516e-01 2.55529195e-01 -6.06776118e-01 5.11761487e-01 -1.85647547e+00 -1.10059249e+00 2.94691682e-01 2.19477844e+00 7.99575627e-01 -2.10633025e-01 6.58079225e-04 -1.41694963e-01 7.81283736e-01 3.24392349e-01 -7.04357266e-01 -2.43015811e-02 -3.08933049e-01 2.02290639e-01 1.16293557e-01 -6.47444054e-02 -1.03378940e+00 7.65572250e-01 5.52749395e+00 7.20669687e-01 -9.90228772e-01 2.06813917e-01 2.29480579e-01 -1.30596921e-01 -5.01603067e-01 9.43887532e-02 -5.68697631e-01 3.83138567e-01 1.24109499e-01 -3.91547829e-01 4.66796547e-01 1.16949308e+00 -6.82830513e-01 3.41811001e-01 -1.79656625e+00 1.09240139e+00 4.54274803e-01 -1.07629478e+00 4.35982794e-01 2.69557871e-02 8.45467150e-01 -3.82676184e-01 5.73638141e-01 5.23256898e-01 1.68318421e-01 -7.88348734e-01 7.95286775e-01 5.48275530e-01 8.21998358e-01 -4.60752070e-01 4.67595994e-01 -5.79476589e-03 -1.31380630e+00 -2.01134607e-01 -7.11482167e-01 -6.55718073e-02 -2.88807034e-01 1.49527505e-01 -7.02939272e-01 5.83887994e-01 6.78649604e-01 1.09289563e+00 -5.59716761e-01 8.41531873e-01 -1.87069327e-01 9.03993770e-02 3.31814259e-01 1.24274246e-01 3.83966640e-02 -3.82882893e-01 5.99903405e-01 8.23273480e-01 3.40063602e-01 -1.82011500e-01 3.20543945e-01 1.19575393e+00 -2.31149912e-01 1.57598674e-01 -7.54338145e-01 -1.04871258e-01 3.59486818e-01 1.12009859e+00 -1.88011453e-01 -5.72784841e-01 -8.34991217e-01 1.27391708e+00 6.38481677e-01 4.63344336e-01 -4.96181965e-01 -1.17699482e-01 1.31113815e+00 -6.26007169e-02 5.94611406e-01 2.13032719e-02 -2.65896171e-01 -1.40293002e+00 1.32481098e-01 -9.52218950e-01 3.93081695e-01 -5.14867425e-01 -2.01366329e+00 2.84753561e-01 -1.33602947e-01 -1.72187078e+00 4.40874994e-02 -7.44650900e-01 -5.73943675e-01 7.45033979e-01 -1.46708179e+00 -1.26555228e+00 -5.41184545e-01 6.65441453e-01 7.23893642e-01 -5.72641850e-01 9.61130083e-01 1.22844033e-01 -7.90599957e-02 7.59885609e-01 4.07064259e-01 1.07387051e-01 1.11724114e+00 -1.33602130e+00 1.06983565e-01 5.40497482e-01 4.51261669e-01 6.22117758e-01 6.10942066e-01 -4.80331063e-01 -1.55699837e+00 -1.11841869e+00 4.91457641e-01 -3.36743921e-01 7.26722896e-01 -7.08835244e-01 -1.25144064e+00 5.72706103e-01 1.66280016e-01 5.39898336e-01 8.99155080e-01 3.34054202e-01 -1.27880085e+00 -3.41539323e-01 -9.12120461e-01 5.36693394e-01 1.33732736e+00 -9.27727818e-01 -7.83872128e-01 1.21354342e-01 5.37178576e-01 9.53525454e-02 -7.68650353e-01 3.94911259e-01 8.75126243e-01 -8.98643255e-01 1.07551563e+00 -9.36325729e-01 4.97752130e-01 -2.30448708e-01 -5.48704088e-01 -1.60296273e+00 -7.01885521e-01 4.81230021e-02 -1.41193807e-01 1.24350584e+00 1.97501332e-01 -5.16103029e-01 5.38774490e-01 3.64560932e-01 8.30037519e-03 -4.58240956e-01 -9.61070955e-01 -1.30597878e+00 -2.51535270e-02 1.07552379e-03 8.73987556e-01 1.24121618e+00 9.02484283e-02 4.24165010e-01 -2.67515868e-01 6.73872530e-02 8.70419204e-01 4.63773668e-01 7.76745260e-01 -1.44594145e+00 -3.10852647e-01 -4.99727994e-01 -9.30666208e-01 -7.24141240e-01 6.34990156e-01 -1.27327704e+00 2.61901822e-02 -9.99103904e-01 6.36976302e-01 -3.22108179e-01 -7.20562398e-01 3.61252636e-01 -2.65497237e-01 -3.30609120e-02 3.37351352e-01 2.35064343e-01 -6.60424352e-01 1.01520669e+00 1.15629995e+00 -7.56896317e-01 2.26176187e-01 -4.13152665e-01 -4.90784079e-01 4.82462823e-01 5.11140347e-01 -3.94301593e-01 -4.11210239e-01 -2.59779185e-01 2.19180863e-02 -5.08028746e-01 5.21774530e-01 -9.96802509e-01 9.86912176e-02 -3.80070746e-01 5.86092830e-01 -2.95744032e-01 4.44541842e-01 -1.16865683e+00 9.45955813e-02 2.26497382e-01 -5.86697817e-01 -2.83642501e-01 -3.37134510e-01 8.77120793e-01 -3.21837604e-01 -1.98708743e-01 9.52315390e-01 -3.60746719e-02 -1.10894489e+00 4.26541388e-01 1.84944510e-01 2.18456089e-01 1.38472712e+00 -2.59859502e-01 -3.36235881e-01 -1.15398325e-01 -5.80927908e-01 2.98194550e-02 8.68692636e-01 7.37082064e-01 1.00568748e+00 -1.82834375e+00 -5.72419286e-01 6.61183417e-01 1.16326547e+00 -3.29815060e-01 2.36202076e-01 5.58202982e-01 1.76901460e-01 -2.20364302e-01 -6.66238606e-01 -6.94174528e-01 -1.17027104e+00 1.11317360e+00 9.96917710e-02 2.57551223e-01 -6.86689675e-01 8.51035178e-01 8.32242608e-01 -5.24424911e-01 2.24465877e-01 1.10549666e-01 7.49815330e-02 2.60669023e-01 5.70272326e-01 1.68923691e-01 -2.37986878e-01 -6.78653121e-01 -4.14613754e-01 5.98208606e-01 -2.80670941e-01 4.53508981e-02 1.29467833e+00 9.84146893e-02 -6.42526289e-03 8.92889082e-01 1.50525975e+00 -2.59794146e-01 -1.33932996e+00 -9.28551495e-01 2.05418289e-01 -9.67786789e-01 -4.61829484e-01 -5.37974954e-01 -8.57654750e-01 1.07728732e+00 7.90247381e-01 -2.00385913e-01 8.36418211e-01 4.35198456e-01 3.31949621e-01 3.31033409e-01 3.12328398e-01 -1.16369414e+00 5.64678848e-01 3.25911760e-01 1.00817859e+00 -1.45760334e+00 -1.42625511e-01 -3.34882021e-01 -9.96123493e-01 1.02172184e+00 8.82139206e-01 -2.87486702e-01 4.41207498e-01 -1.28465220e-01 -2.13111341e-01 -8.47415105e-02 -6.91742420e-01 -3.49085182e-01 5.70764661e-01 8.63667071e-01 2.25251332e-01 2.03812331e-01 1.39520168e-02 5.14030635e-01 2.54909307e-01 -5.14696360e-01 1.48864180e-01 7.84257591e-01 -4.28136498e-01 -1.17299080e+00 -4.93544200e-03 4.66606140e-01 1.29908323e-01 6.87196851e-02 -6.12687707e-01 4.68810230e-01 1.56805679e-01 4.59686279e-01 2.53215641e-01 -5.39140105e-01 3.21353346e-01 4.62454230e-01 5.43186724e-01 -8.44918847e-01 -2.57481068e-01 -2.74070203e-01 -3.26553941e-01 -5.98415911e-01 -3.77567023e-01 -6.08572304e-01 -8.89676392e-01 8.07415545e-02 7.52511015e-03 -2.18484610e-01 3.50593179e-01 6.44892454e-01 3.30908626e-01 4.42592651e-01 8.71955097e-01 -6.92689300e-01 -9.51595962e-01 -8.07696998e-01 -9.21319008e-01 1.22934210e+00 3.68637651e-01 -1.20526576e+00 -4.07422185e-01 4.68759499e-02]
[10.1381196975708, 2.286712408065796]
497d12a9-8b8e-4c7a-a865-5ba68537a63a
reliable-and-interpretable-drift-detection-in
2305.17750
null
https://arxiv.org/abs/2305.17750v1
https://arxiv.org/pdf/2305.17750v1.pdf
Reliable and Interpretable Drift Detection in Streams of Short Texts
Data drift is the change in model input data that is one of the key factors leading to machine learning models performance degradation over time. Monitoring drift helps detecting these issues and preventing their harmful consequences. Meaningful drift interpretation is a fundamental step towards effective re-training of the model. In this study we propose an end-to-end framework for reliable model-agnostic change-point detection and interpretation in large task-oriented dialog systems, proven effective in multiple customer deployments. We evaluate our approach and demonstrate its benefits with a novel variant of intent classification training dataset, simulating customer requests to a dialog system. We make the data publicly available.
['Ateret Anaby-Tavor', 'Samuel Ackerman', 'Matan Vetzler', 'Ella Rabinovich']
2023-05-28
null
null
null
null
['intent-classification', 'change-point-detection']
['natural-language-processing', 'time-series']
[ 3.69379856e-02 -2.08425045e-01 -3.46118212e-01 -8.93837869e-01 -6.17763221e-01 -8.64779770e-01 6.36542261e-01 3.89074445e-01 -2.58124471e-01 6.75469160e-01 1.39253423e-01 -4.86619115e-01 -5.19170798e-02 4.42182831e-02 -3.31514567e-01 -2.43235663e-01 -1.34470224e-01 1.09460425e+00 4.23161060e-01 -4.03771818e-01 5.22425711e-01 4.36941653e-01 -1.20449817e+00 5.46197474e-01 7.03565598e-01 8.89525712e-01 -2.08059430e-01 9.05383945e-01 -1.93940073e-01 9.29567814e-01 -1.18440199e+00 -2.91023910e-01 -6.60220459e-02 -1.12086095e-01 -8.61233532e-01 -2.74942559e-03 3.42349976e-01 -4.58087415e-01 1.59565464e-01 7.46764421e-01 4.95931178e-01 -5.95679581e-02 3.65022361e-01 -1.68039656e+00 -2.25070596e-01 4.90084618e-01 -1.57955125e-01 8.01177382e-01 2.22870350e-01 4.25079644e-01 7.51261652e-01 -6.55058742e-01 4.62343365e-01 1.48143733e+00 7.22826242e-01 1.00387836e+00 -1.41474998e+00 -4.95015115e-01 5.09979129e-01 1.81314975e-01 -8.27706158e-01 -7.94107378e-01 7.70222127e-01 -7.08523750e-01 1.34452224e+00 3.53845596e-01 1.01486102e-01 1.35587025e+00 3.61790806e-01 8.21203947e-01 6.59410477e-01 -3.61820787e-01 6.32792532e-01 4.45392996e-01 8.55374873e-01 1.91250592e-01 1.89455345e-01 2.42552515e-02 -8.87983739e-01 -6.93830132e-01 3.04316133e-02 9.31859910e-02 1.12038251e-05 -1.73643321e-01 -4.11017179e-01 7.24797249e-01 9.37313735e-02 1.75384030e-01 -4.56092954e-01 1.64499179e-01 5.96744657e-01 6.56792223e-01 6.91362739e-01 1.71791911e-01 -9.07164693e-01 -6.91618741e-01 -6.62987888e-01 4.75717306e-01 8.66957664e-01 1.00137496e+00 1.16429560e-01 -1.79868657e-02 -3.00142229e-01 8.05408895e-01 3.86885613e-01 1.69883698e-01 9.71976221e-01 -5.96446633e-01 3.94062668e-01 9.65692341e-01 4.44765598e-01 -2.67151535e-01 -5.63999832e-01 -2.36058235e-01 -1.25502765e-01 1.71407357e-01 2.57342786e-01 -9.12467167e-02 -8.08858216e-01 1.70105922e+00 3.63987476e-01 -1.37366444e-01 1.53164521e-01 5.04840374e-01 3.17943215e-01 2.68438365e-02 2.49865681e-01 -2.54744083e-01 1.23759866e+00 -6.34292424e-01 -9.31695819e-01 -6.52782857e-01 7.44122505e-01 -5.75485706e-01 1.39464712e+00 4.94883209e-01 -8.62448633e-01 -3.13281000e-01 -1.08307123e+00 2.72522777e-01 -1.39505923e-01 -4.55261409e-01 3.79577726e-01 7.93757021e-01 -6.80143535e-01 4.93074626e-01 -9.01689291e-01 -4.37911451e-01 3.68292719e-01 4.69859153e-01 3.23174417e-01 2.18629166e-01 -9.30522978e-01 9.91499960e-01 5.73287569e-02 -6.31985217e-02 -8.67114544e-01 -1.00699151e+00 -2.15823248e-01 -2.21601918e-01 2.41875365e-01 -4.29765701e-01 2.37173152e+00 -6.43478930e-01 -1.10125673e+00 7.20967591e-01 -5.22729158e-01 -7.74761558e-01 1.08157992e+00 -6.40770853e-01 -6.40525341e-01 -7.50382781e-01 -9.46743712e-02 2.25668266e-01 1.03338611e+00 -1.35480845e+00 -7.69652009e-01 -7.19089270e-01 -3.48932445e-01 -1.69659317e-01 -3.77288043e-01 1.98720247e-01 2.28260104e-02 -3.40805352e-01 -9.59705934e-02 -8.48978341e-01 -1.53367212e-02 -6.62277400e-01 -2.49912232e-01 -3.21736127e-01 1.18699217e+00 -7.44565368e-01 1.51821709e+00 -1.75667906e+00 -2.01274216e-01 -2.20207438e-01 1.08024985e-01 3.16458821e-01 1.69420704e-01 5.26050985e-01 6.21089786e-02 1.44797876e-01 -3.18328023e-01 -6.77868009e-01 -8.29287246e-02 2.49659896e-01 -6.16308510e-01 8.23094994e-02 2.07473218e-01 7.29634225e-01 -7.59763360e-01 2.25212246e-01 -5.35034016e-03 1.61070898e-02 -3.75386596e-01 4.27318186e-01 -8.34963679e-01 4.49678004e-01 -1.48704290e-01 9.14823413e-01 5.83300233e-01 1.39030710e-01 9.09435675e-02 3.27291399e-01 1.67473227e-01 4.86771077e-01 -6.20246589e-01 1.31132495e+00 -4.99111682e-01 7.50344157e-01 -2.85962410e-02 -4.39101756e-01 9.37138855e-01 1.24508351e-01 2.47275993e-01 -6.41203284e-01 4.89574531e-03 -7.29418620e-02 1.69918418e-01 -5.61773419e-01 5.50023615e-01 -1.61281019e-01 -1.89013228e-01 4.67248261e-01 -3.25462341e-01 1.40257731e-01 -6.59742579e-02 2.90215552e-01 1.41599786e+00 -4.47004884e-01 6.50912151e-02 -1.41205102e-01 3.25125694e-01 3.32723230e-01 4.56741571e-01 8.60345662e-01 -5.98444760e-01 1.37529984e-01 4.83568311e-01 -6.28571391e-01 -6.96461618e-01 -8.26178730e-01 -1.87102348e-01 1.32840228e+00 -2.67928392e-01 -3.00251454e-01 -7.29615331e-01 -1.02039409e+00 6.30380809e-01 1.42804599e+00 -5.39153993e-01 -6.13655031e-01 -6.73186600e-01 -8.62687111e-01 4.85740542e-01 6.04720473e-01 4.00476336e-01 -1.02810025e+00 -7.33571470e-01 5.34089446e-01 -1.72136724e-02 -8.79937112e-01 -4.61332291e-01 3.29059631e-01 -1.29001319e+00 -1.31155241e+00 8.91800970e-02 -2.77287453e-01 3.77694488e-01 1.45350426e-01 1.36293733e+00 1.10932522e-01 -4.24547762e-01 4.76594687e-01 1.64321344e-02 -8.67090881e-01 -1.19001722e+00 1.81127563e-01 3.92918169e-01 -3.13922107e-01 9.06466484e-01 -1.82449758e-01 -5.02723992e-01 5.53347468e-01 -8.35666656e-01 -6.10441327e-01 3.05184778e-02 7.49038935e-01 -2.53735244e-01 -8.20576325e-02 9.72923517e-01 -1.28521025e+00 1.57127810e+00 -4.21847552e-01 -7.02151537e-01 3.08387160e-01 -1.59067929e+00 3.52377101e-04 2.05023214e-01 -3.32018107e-01 -1.24828899e+00 -9.89178196e-02 -3.21605168e-02 -1.36848062e-01 -3.06301147e-01 2.50638515e-01 2.15881635e-02 5.10881007e-01 9.42715704e-01 -1.95720956e-01 1.69279589e-03 -8.97019863e-01 1.94775999e-01 1.15770233e+00 5.97099960e-01 -2.42325127e-01 2.22858369e-01 2.79027253e-01 -5.95232368e-01 -8.50816429e-01 -3.58669609e-01 -8.28176320e-01 -8.39248419e-01 -2.94257939e-01 8.36354345e-02 -8.11397076e-01 -6.61509693e-01 6.95002913e-01 -1.22909629e+00 -4.65344161e-01 -1.27171770e-01 -3.13542217e-01 -2.56462395e-01 2.69705262e-02 -4.56716210e-01 -1.30762219e+00 -6.33594990e-01 -7.87458658e-01 8.93790245e-01 3.33441421e-02 -9.88182783e-01 -1.17405641e+00 4.13749576e-01 5.14254212e-01 6.42206252e-01 -1.29684493e-01 1.12323141e+00 -1.22135174e+00 -1.84342951e-01 -5.33387423e-01 3.84606838e-01 3.05821627e-01 3.42330962e-01 1.11069053e-03 -1.28334272e+00 -7.28960276e-01 4.90761131e-01 -9.85053182e-03 7.36372650e-01 2.34186485e-01 4.68743831e-01 -4.83920097e-01 -5.33156276e-01 -1.93357319e-01 1.06047893e+00 6.51726961e-01 1.35650292e-01 4.00255859e-01 4.14455712e-01 5.85066080e-01 8.00140262e-01 5.96899331e-01 2.86941230e-01 1.02804327e+00 3.73782218e-01 4.90626901e-01 4.68489937e-02 -1.55304283e-01 6.84834301e-01 3.79662156e-01 7.61865914e-01 -2.18508080e-01 -1.33233047e+00 5.81967413e-01 -2.09846759e+00 -5.43189228e-01 -2.59572327e-01 2.26278281e+00 7.79471815e-01 5.52327573e-01 5.24586439e-01 -1.25368347e-03 5.80279469e-01 -2.13930264e-01 -1.19954336e+00 -5.89742720e-01 3.12447548e-01 -4.20968026e-01 3.52260083e-01 7.09383249e-01 -7.87823498e-01 9.95111763e-01 7.06037855e+00 -1.67921692e-01 -9.62695837e-01 3.62471700e-01 4.38653350e-01 -5.05407393e-01 -5.68682663e-02 -2.92276777e-02 -1.03381228e+00 6.45569682e-01 1.43960428e+00 -3.58126909e-01 1.73524201e-01 1.16993630e+00 4.51288670e-01 -2.70541430e-01 -1.72490251e+00 6.71230137e-01 -7.80417696e-02 -1.01050627e+00 -3.89247239e-01 -2.02287436e-01 3.54751468e-01 1.06010571e-01 4.30512019e-02 4.73290324e-01 7.29538620e-01 -5.66581190e-01 6.77562952e-01 2.69774497e-01 1.71180069e-01 -4.95075643e-01 7.04183102e-01 5.49062490e-01 -4.33928788e-01 -5.13350248e-01 -5.55715524e-02 -2.17086002e-02 3.02308440e-01 4.73306656e-01 -1.79724634e+00 -4.86179963e-02 7.44015157e-01 2.82151192e-01 -8.19405496e-01 7.97904909e-01 9.33820978e-02 9.35245454e-01 -9.83567610e-02 -9.81873646e-03 -2.62957603e-01 3.18566263e-01 7.51336634e-01 1.21392989e+00 -2.44465992e-01 -3.21800649e-01 -1.60746321e-01 6.03719652e-01 -3.37829180e-02 -4.80375171e-01 -5.94380319e-01 1.20358184e-01 6.66780591e-01 6.99171662e-01 -4.11820680e-01 -3.51558894e-01 6.05258998e-03 1.03313982e+00 1.29674479e-01 2.40365639e-01 -6.98717952e-01 2.20242649e-01 1.07489967e+00 1.17520273e-01 -4.04754691e-02 -2.77079437e-02 -4.83254969e-01 -7.53268957e-01 4.01523858e-01 -9.57320094e-01 7.02094316e-01 -1.50566369e-01 -1.43735659e+00 5.47587752e-01 1.63798869e-01 -8.15058768e-01 -7.83597827e-01 -4.04203147e-01 -8.03720534e-01 8.41999054e-01 -1.28299308e+00 -7.29113936e-01 -5.54371297e-01 2.59650946e-01 1.18078661e+00 -3.00319403e-01 6.70135140e-01 6.31658733e-02 -9.55507100e-01 6.26694798e-01 1.31150618e-01 -4.78438288e-01 9.88843203e-01 -1.36887288e+00 9.56833005e-01 7.84603834e-01 -2.93281794e-01 6.92372680e-01 1.07777596e+00 -1.10055184e+00 -1.20015395e+00 -1.07506657e+00 6.94872558e-01 -1.19738412e+00 6.14076912e-01 -5.92570961e-01 -1.32201350e+00 9.01116014e-01 1.21943764e-01 -8.06139529e-01 6.00400448e-01 3.23980719e-01 -1.46575302e-01 -2.40116507e-01 -1.16553676e+00 4.16469425e-01 9.81838167e-01 -3.19350094e-01 -6.42624855e-01 4.62636918e-01 7.56284237e-01 -4.34607327e-01 -5.67704260e-01 3.40344340e-01 2.92471766e-01 -1.08650875e+00 6.00184321e-01 -1.21172571e+00 -2.49535665e-01 8.69946033e-02 -5.09474613e-02 -1.39090133e+00 9.21923593e-02 -9.22484636e-01 -4.90547389e-01 1.52914023e+00 4.93122935e-01 -7.39534676e-01 8.74054432e-01 1.41018391e+00 1.43930111e-02 -2.13072434e-01 -1.00765765e+00 -9.22820330e-01 4.29797135e-02 -5.92594326e-01 6.97901189e-01 6.81311309e-01 8.42141360e-02 5.98148108e-01 -3.03310066e-01 7.78471604e-02 4.03233975e-01 -1.90493673e-01 1.10127127e+00 -1.48877668e+00 5.35798408e-02 -4.85742241e-01 -4.44831029e-02 -8.33856046e-01 -7.26859225e-03 -4.54447001e-01 1.88893452e-01 -1.04954493e+00 -1.58412516e-01 -6.15667820e-01 -2.10033312e-01 3.92053246e-01 -2.46853635e-01 -6.90534532e-01 3.85257900e-02 7.76620924e-01 -3.61173481e-01 3.73669982e-01 1.79053292e-01 -2.50614345e-01 -7.37424731e-01 6.42094493e-01 -6.04065001e-01 4.84378338e-01 1.01108038e+00 -8.67394686e-01 -6.15775287e-01 -1.41519830e-01 -1.78555071e-01 -2.78885990e-01 3.01038742e-01 -8.30422699e-01 3.08429509e-01 -4.18558717e-02 1.17228575e-01 -6.83306336e-01 1.70404956e-01 -8.76994252e-01 -5.60587924e-03 9.33569014e-01 -6.84530079e-01 5.56538343e-01 6.58223093e-01 7.77709246e-01 6.81147724e-02 -4.21922386e-01 6.17911220e-01 1.92308933e-01 -7.22074270e-01 -1.75258860e-01 -3.19677949e-01 2.31267232e-02 1.17145944e+00 2.28321701e-01 -6.72952116e-01 -5.22716522e-01 -4.58279520e-01 5.46962321e-01 4.28853929e-01 1.19394743e+00 3.48772794e-01 -8.42323720e-01 -4.93365854e-01 3.16044658e-01 4.31592882e-01 -3.36916834e-01 -2.30450425e-02 5.06718457e-01 -1.10246964e-01 7.59938657e-01 1.25837430e-01 -1.04921591e+00 -1.77788639e+00 4.42672163e-01 4.26518410e-01 -1.50939256e-01 -3.53889763e-01 6.76467955e-01 -3.52002114e-01 -2.62782842e-01 7.66350687e-01 -6.16099596e-01 9.60882157e-02 1.45885408e-01 6.66439116e-01 5.18909454e-01 7.32555032e-01 -1.46897078e-01 -7.37694204e-01 -3.45007062e-01 -8.07940900e-01 -2.37081945e-01 9.76586938e-01 -2.11998731e-01 1.78144440e-01 9.24208641e-01 8.04689527e-01 -4.08081234e-01 -1.43842673e+00 -2.54047602e-01 8.55422020e-01 -5.01659811e-01 -1.62353843e-01 -1.32028866e+00 -3.61369461e-01 6.50185764e-01 1.13409984e+00 6.50778949e-01 8.62577260e-01 -4.23828512e-02 6.04443610e-01 5.22339344e-01 1.76680773e-01 -1.19137442e+00 2.01152027e-01 3.65313470e-01 1.09135664e+00 -1.63176775e+00 -2.46060565e-01 8.08333009e-02 -9.34834898e-01 7.70456016e-01 9.21778381e-01 5.51771879e-01 5.02234578e-01 2.72323132e-01 6.22437298e-01 -3.64316016e-01 -1.51192415e+00 3.42849016e-01 -1.31690979e-01 6.00177050e-01 3.16394687e-01 -3.97484452e-02 6.61153570e-02 6.72563910e-01 8.61850232e-02 7.49844536e-02 3.96122068e-01 1.32115853e+00 -4.15227443e-01 -1.36386514e+00 -2.98218727e-01 3.82836401e-01 -1.74858093e-01 1.47224128e-01 -1.09174871e+00 7.19054401e-01 -5.21839201e-01 1.23649418e+00 1.14045128e-01 -5.28689384e-01 8.59439909e-01 8.88296902e-01 2.64912024e-02 -4.82013524e-01 -9.83404577e-01 -2.47927353e-01 2.92905480e-01 -4.97092307e-01 2.07817256e-01 -9.78298306e-01 -1.32164526e+00 -4.08732593e-01 -2.38305226e-01 1.13470946e-02 9.55200434e-01 7.26146221e-01 8.74131918e-01 7.15761185e-01 8.02656054e-01 -1.79026693e-01 -1.21712375e+00 -1.40580153e+00 -4.75363769e-02 8.55124712e-01 5.76581955e-01 -6.15835428e-01 -4.47907925e-01 1.87903240e-01]
[12.58251667022705, 7.667848110198975]
cc5f0d5c-efc9-4c32-b091-c8af153e5a67
deep-ehr-spotlight-a-framework-and-mechanism
2103.14161
null
https://arxiv.org/abs/2103.14161v1
https://arxiv.org/pdf/2103.14161v1.pdf
Deep EHR Spotlight: a Framework and Mechanism to Highlight Events in Electronic Health Records for Explainable Predictions
The wide adoption of Electronic Health Records (EHR) has resulted in large amounts of clinical data becoming available, which promises to support service delivery and advance clinical and informatics research. Deep learning techniques have demonstrated performance in predictive analytic tasks using EHRs yet they typically lack model result transparency or explainability functionalities and require cumbersome pre-processing tasks. Moreover, EHRs contain heterogeneous and multi-modal data points such as text, numbers and time series which further hinder visualisation and interpretability. This paper proposes a deep learning framework to: 1) encode patient pathways from EHRs into images, 2) highlight important events within pathway images, and 3) enable more complex predictions with additional intelligibility. The proposed method relies on a deep attention mechanism for visualisation of the predictions and allows predicting multiple sequential outcomes.
['Joao H. Bettencourt-Silva', 'Gurdeep S. Mannu', 'Natasha Mulligan', 'Thanh Nguyen-Duc']
2021-03-25
null
null
null
null
['deep-attention', 'deep-attention']
['computer-vision', 'natural-language-processing']
[ 2.13408470e-01 2.60899395e-01 -9.89768282e-03 -4.95784342e-01 -4.47805315e-01 -3.60962331e-01 5.66791654e-01 9.75557685e-01 -5.91092110e-02 6.55410171e-01 6.13706231e-01 -6.68741882e-01 -3.00176471e-01 -7.69942880e-01 -4.28891182e-01 -4.68311489e-01 -6.33015871e-01 3.64953339e-01 -5.24569988e-01 3.36758643e-01 -1.80706859e-01 5.89869916e-01 -1.07212520e+00 9.49229062e-01 7.55404770e-01 1.16674542e+00 2.47864760e-02 6.52888238e-01 -2.88886726e-02 1.00029910e+00 -4.95078266e-01 -4.05579090e-01 -5.59767261e-02 -2.66332686e-01 -5.69407642e-01 -2.75762767e-01 -1.63639456e-01 -3.19091856e-01 -6.16884112e-01 7.93722987e-01 6.46003604e-01 -3.17712933e-01 3.74451697e-01 -9.06538844e-01 -1.09672797e+00 4.35783297e-01 -4.12218720e-01 1.79494649e-01 5.36370099e-01 6.10624135e-01 7.49073386e-01 -4.73872095e-01 7.64101446e-01 1.06712568e+00 7.62520134e-01 3.93984646e-01 -1.30089414e+00 -4.22223389e-01 -7.88791105e-02 4.22341585e-01 -1.13192868e+00 -2.05042839e-01 3.89036238e-01 -6.29098892e-01 1.20221829e+00 6.39574826e-01 1.07892370e+00 1.25920522e+00 5.39948642e-01 6.75289512e-01 8.90715003e-01 2.15187117e-01 5.27604902e-03 7.79975802e-02 6.42271638e-02 5.31859040e-01 2.19713211e-01 9.96707827e-02 -1.67853087e-01 -1.15543678e-01 7.95529723e-01 1.01124978e+00 -5.29348493e-01 1.61244869e-01 -1.78231657e+00 5.78204751e-01 6.63146377e-01 5.78668825e-02 -8.70329678e-01 -1.86876731e-03 7.88036942e-01 1.74701065e-01 1.83580130e-01 4.31633711e-01 -5.74656606e-01 -2.45274901e-01 -6.63697600e-01 2.42480591e-01 4.75719035e-01 9.59043384e-01 1.45608708e-01 -1.04505703e-01 -4.52960819e-01 4.00693178e-01 2.22982287e-01 2.37240121e-01 5.38498282e-01 -3.61012936e-01 5.88251472e-01 1.03882682e+00 1.79413080e-01 -1.21882987e+00 -9.44126904e-01 -1.86167091e-01 -1.48981762e+00 -2.37685487e-01 1.00653723e-01 9.60349739e-02 -9.71114516e-01 1.20712352e+00 2.77459621e-01 7.89096504e-02 1.28217891e-01 7.63064682e-01 1.06473565e+00 8.26697707e-01 6.99604273e-01 -1.54347986e-01 1.55100238e+00 -4.10385519e-01 -1.09576356e+00 4.77383614e-01 9.54959273e-01 -4.83554214e-01 1.21170902e+00 2.80088753e-01 -1.05301619e+00 -4.48216915e-01 -8.09179902e-01 -3.20626706e-01 -5.84371805e-01 1.29878119e-01 7.47432411e-01 1.41675457e-01 -7.68660188e-01 6.67850494e-01 -1.16133726e+00 -2.54852712e-01 1.10137022e+00 3.48498285e-01 -4.86071676e-01 -1.32432982e-01 -1.12938464e+00 7.79645085e-01 4.45175320e-01 1.96807519e-01 -6.09548926e-01 -1.10466218e+00 -9.53689754e-01 5.30375123e-01 -1.87828571e-01 -1.01146138e+00 9.32953060e-01 -4.87091690e-01 -1.04580903e+00 5.91391206e-01 9.52565074e-02 -4.61813152e-01 7.27622509e-01 -3.46894801e-01 -6.24952734e-01 -3.13586779e-02 -4.09150124e-01 4.54455376e-01 3.03241819e-01 -7.63644636e-01 -4.12568241e-01 -6.65355206e-01 -3.97949517e-01 5.14861271e-02 -2.92783052e-01 -4.74196486e-02 -2.63053119e-01 -7.69037545e-01 -1.72098190e-01 -4.47575152e-01 -5.10581970e-01 6.22741356e-02 -7.48341143e-01 -2.36387365e-02 8.98691356e-01 -1.23450971e+00 1.50696194e+00 -2.05923057e+00 -1.42715961e-01 -8.90014879e-03 9.08531845e-01 2.63772547e-01 2.11116552e-01 4.26360130e-01 -3.37476492e-01 2.33582884e-01 -2.61468049e-02 -1.57825321e-01 -1.19824484e-01 1.13310456e-01 -2.64274389e-01 4.38225091e-01 3.43994200e-01 1.38454115e+00 -8.18330407e-01 -5.04246533e-01 4.60397333e-01 8.58763337e-01 -5.20787060e-01 3.75304192e-01 -8.85147452e-02 8.18909943e-01 -3.69710594e-01 7.34012961e-01 4.67817783e-01 -8.91938388e-01 1.79441735e-01 -3.69583875e-01 -4.80257608e-02 4.95954514e-01 -7.13926077e-01 1.34462464e+00 -2.97668546e-01 6.21082783e-01 -3.51526082e-01 -9.07288194e-01 6.63045108e-01 5.31100869e-01 6.64746583e-01 -6.85842156e-01 -4.49470952e-02 -3.85160625e-01 8.66765305e-02 -8.82544875e-01 2.23157614e-01 -7.94532883e-04 1.26776742e-02 1.90806061e-01 -3.35186154e-01 4.89118427e-01 -3.32912147e-01 1.05018601e-01 1.23230052e+00 -1.74339011e-01 6.54318929e-01 2.91121513e-01 1.98186725e-01 -2.46927179e-02 6.27843916e-01 3.13681424e-01 -2.52578910e-02 5.12180686e-01 6.43279612e-01 -1.14397132e+00 -1.25054240e+00 -1.02285039e+00 -2.77042389e-01 3.97382110e-01 -1.09862663e-01 -4.78509367e-01 -3.45453799e-01 -5.19129395e-01 1.88506126e-01 5.04592061e-01 -7.40055442e-01 -1.35685235e-01 -4.04823214e-01 -7.90393829e-01 4.09802347e-01 9.36805367e-01 8.58256370e-02 -1.21530747e+00 -1.04415667e+00 4.04569715e-01 -7.96286687e-02 -8.63042235e-01 -4.95843068e-02 1.32726520e-01 -1.11363506e+00 -1.07299674e+00 -6.33837104e-01 -3.78889322e-01 6.88095570e-01 -3.88859183e-01 1.03565598e+00 -9.97404158e-02 -8.09868157e-01 -1.90599076e-02 -8.81171376e-02 -7.46048152e-01 -3.92959386e-01 -3.30182403e-01 -2.28936344e-01 -1.40343398e-01 4.47964728e-01 -3.69831264e-01 -9.91247356e-01 -2.29084581e-01 -9.43865597e-01 6.96731687e-01 6.45231485e-01 1.02223325e+00 8.63026440e-01 -3.33116621e-01 7.20218420e-01 -1.12025833e+00 8.55272174e-01 -7.68338323e-01 -4.35177267e-01 2.34001398e-01 -8.11521649e-01 -1.67723745e-01 9.93213832e-01 -1.82514161e-01 -9.82808113e-01 -8.10845867e-02 -5.85944299e-03 -4.03027713e-01 -5.20156503e-01 7.88866520e-01 -9.99200270e-02 5.47585845e-01 3.45964968e-01 1.31808951e-01 -5.93615249e-02 -4.78305906e-01 2.81607091e-01 7.08305478e-01 4.73553360e-01 1.63769037e-01 8.01158026e-02 4.24076945e-01 2.02022009e-02 -5.09260356e-01 -3.00740123e-01 -2.57136524e-01 -4.50450808e-01 1.07823655e-01 1.05249465e+00 -8.43225837e-01 -1.27854276e+00 1.72902104e-02 -1.27855849e+00 -9.92092341e-02 -3.51180702e-01 5.16356349e-01 -4.31036085e-01 3.40481818e-01 -1.01559579e+00 -5.42823017e-01 -7.36917794e-01 -9.47799802e-01 8.80903244e-01 5.22475466e-02 -6.67692304e-01 -9.77165401e-01 -1.42635942e-01 8.70758519e-02 3.95628244e-01 7.42756724e-01 1.48987746e+00 -8.13797832e-01 -6.79890871e-01 -5.01974106e-01 -7.00350523e-01 -3.16314518e-01 1.98693857e-01 -6.24627583e-02 -8.25797915e-01 -9.31799859e-02 -3.10928345e-01 2.65137013e-02 6.18621111e-01 6.17550254e-01 1.96137106e+00 -7.98568904e-01 -6.92953229e-01 1.01978981e+00 1.13365173e+00 6.19783282e-01 7.39325166e-01 1.70016944e-01 9.26748574e-01 6.62809014e-01 2.33036518e-01 9.44736660e-01 7.09420085e-01 2.25953385e-01 5.94095349e-01 -6.22375131e-01 2.46627331e-01 -2.59099156e-01 -3.31639737e-01 9.40869212e-01 -1.85276896e-01 -9.51697603e-02 -1.13745415e+00 4.52219784e-01 -2.07423067e+00 -8.91670823e-01 -2.45139629e-01 2.05390882e+00 6.94379985e-01 -3.63147587e-01 -1.23904847e-01 9.57456976e-03 4.03088957e-01 -1.69846103e-01 -8.10812831e-01 -3.58951092e-01 5.83921447e-02 9.51686054e-02 2.74814904e-01 -1.39424771e-01 -1.13383818e+00 3.83905143e-01 6.29695988e+00 3.41730230e-02 -1.18593395e+00 -4.12033424e-02 1.18951547e+00 -3.46158564e-01 -3.61570835e-01 -6.98456347e-01 -2.29835600e-01 6.23964131e-01 1.14775467e+00 -2.24557012e-01 -6.61102012e-02 6.72321200e-01 5.36710382e-01 4.78215396e-01 -1.47066009e+00 1.44400764e+00 -6.85484827e-01 -1.78470111e+00 3.10673177e-01 3.39911371e-01 6.11140728e-01 -1.54322326e-01 2.91452348e-01 5.08442447e-02 3.66156548e-01 -1.58912563e+00 1.21984996e-01 9.31626141e-01 1.03351939e+00 -6.12279534e-01 8.09494615e-01 9.27102193e-03 -1.01813412e+00 -4.39630896e-01 -2.07125381e-01 7.16314763e-02 1.96521625e-01 5.07300556e-01 -1.20743334e+00 5.61663866e-01 9.17921662e-01 1.08383000e+00 -3.82773548e-01 1.22910333e+00 1.79373100e-01 5.76410770e-01 3.95341069e-02 2.09551789e-02 4.57044020e-02 -8.43273774e-02 1.76177308e-01 1.49251938e+00 5.51922619e-01 2.02561423e-01 6.35481477e-02 8.96027267e-01 -8.61866102e-02 2.54988521e-01 -7.21053481e-01 -4.42105472e-01 2.33133078e-01 1.12622762e+00 -6.17126107e-01 -6.63465858e-01 -6.45712376e-01 7.65494049e-01 1.44808978e-01 4.91555989e-01 -7.30087638e-01 -2.22826153e-01 9.20157790e-01 3.37255776e-01 -3.29606086e-02 7.41022080e-02 -7.76026964e-01 -1.16350400e+00 -1.09758914e-01 -8.52874279e-01 8.26000810e-01 -7.23076820e-01 -1.22575200e+00 7.15860844e-01 -4.45364147e-01 -1.26142156e+00 -2.53788829e-01 -4.75526720e-01 -2.92951584e-01 1.06234467e+00 -1.42475450e+00 -1.13199651e+00 -5.54308116e-01 7.09056497e-01 2.53105581e-01 -7.49839991e-02 1.25887203e+00 5.27796149e-01 -6.33408308e-01 4.31570530e-01 2.05861613e-01 1.42668724e-01 5.43398023e-01 -1.34389639e+00 4.99767601e-01 2.80332208e-01 -1.13421060e-01 7.83141434e-01 4.63939130e-01 -6.10867202e-01 -1.38387203e+00 -1.49459708e+00 9.36558723e-01 -3.72465611e-01 3.86724532e-01 -6.03098869e-02 -1.19047534e+00 8.44164193e-01 7.01832920e-02 5.85049093e-02 1.20918345e+00 -3.71999480e-02 -2.10889071e-01 -1.39518976e-01 -1.09298694e+00 7.65104115e-01 6.74410522e-01 -4.08924669e-01 -3.64793897e-01 2.85278499e-01 6.45553768e-01 -3.43713105e-01 -1.26034522e+00 5.43414235e-01 5.56228161e-01 -5.99108160e-01 1.00023580e+00 -1.20327318e+00 7.79050112e-01 -1.37580872e-01 3.99642617e-01 -1.10741961e+00 -3.87943208e-01 -6.22560978e-01 -6.94403470e-01 6.45788968e-01 3.42834800e-01 -4.78897750e-01 5.71411014e-01 9.18377876e-01 -1.10105917e-01 -1.18302798e+00 -5.77091813e-01 5.60229830e-02 -5.11892915e-01 -3.23258460e-01 1.01411080e+00 1.39291167e+00 4.17970747e-01 -7.88777508e-03 -3.82592022e-01 8.07749406e-02 1.88378364e-01 2.89275706e-01 5.17569661e-01 -1.47652078e+00 -1.63989924e-02 -5.46692550e-01 -7.78697968e-01 -7.51973331e-01 -5.61338603e-01 -8.27629268e-01 -5.29970467e-01 -1.84905148e+00 2.93739766e-01 -4.66139972e-01 -6.97905660e-01 7.73879886e-01 -3.09794664e-01 -2.16124691e-02 -3.10781002e-02 4.09782022e-01 -6.13556504e-01 4.10029173e-01 1.23366225e+00 -2.69330442e-01 -4.05221552e-01 -1.59599334e-01 -6.06694579e-01 5.71827710e-01 6.13015592e-01 -2.10273668e-01 -2.07323790e-01 -2.67778635e-01 3.07696134e-01 5.84275484e-01 4.73317444e-01 -6.07083678e-01 -1.99196339e-02 2.40157731e-02 1.11894000e+00 -6.69778287e-01 4.08997275e-02 -1.00412345e+00 3.94800186e-01 6.76290631e-01 -6.34472013e-01 4.60879028e-01 4.10212994e-01 8.28527570e-01 -3.68034810e-01 5.96961617e-01 3.32158804e-01 -9.98557657e-02 -4.61284071e-01 5.93850970e-01 -4.24915791e-01 -5.18616438e-01 1.16720653e+00 -2.93063849e-01 -9.12467018e-02 -3.47690284e-01 -1.06541514e+00 3.18844944e-01 1.89446479e-01 3.82511854e-01 8.03828061e-01 -1.34662342e+00 -5.94015956e-01 1.59329534e-01 1.91114947e-01 2.83601224e-01 7.08252251e-01 9.32915688e-01 -8.73340011e-01 6.06263638e-01 -3.01338762e-01 -7.14172006e-01 -1.40527940e+00 8.86920750e-01 -1.17666528e-01 -5.77008903e-01 -1.22255027e+00 4.15849984e-01 5.75231493e-01 -2.81353474e-01 2.56413043e-01 -7.37482548e-01 -3.92733008e-01 -8.91669244e-02 8.82194042e-01 1.91018432e-01 4.79894057e-02 -2.53588974e-01 -2.24739268e-01 -1.05376795e-01 -2.52542078e-01 4.33499038e-01 1.65513384e+00 -4.17072996e-02 5.58567829e-02 2.95594752e-01 1.04905701e+00 -6.64184093e-01 -1.25288928e+00 -2.90758852e-02 2.72713363e-01 -3.29389513e-01 -2.43822291e-01 -1.09387696e+00 -8.43633652e-01 1.36830521e+00 6.37150407e-01 3.43810171e-01 1.28171527e+00 -1.44768000e-01 9.56451774e-01 2.69370854e-01 -2.21504807e-01 -6.30960643e-01 -3.25219870e-01 -7.09022284e-02 7.58637071e-01 -1.26482403e+00 -4.35162745e-02 -1.04167163e-01 -8.43245625e-01 1.27213490e+00 2.55656391e-01 4.70668137e-01 5.05310416e-01 4.23572540e-01 2.49028772e-01 -4.57504034e-01 -1.07448947e+00 3.60491872e-01 3.65361840e-01 7.79703438e-01 8.01216900e-01 3.73035014e-01 -7.90076628e-02 1.02468443e+00 -1.07863257e-02 1.99519902e-01 1.87913850e-01 6.57042325e-01 1.98227867e-01 -4.48448807e-01 -2.91713446e-01 8.91677380e-01 -6.83064818e-01 -2.62073457e-01 7.32906302e-03 6.32594585e-01 -2.52513587e-02 3.78793776e-01 1.56486630e-01 -2.96969533e-01 3.05885613e-01 1.63153753e-01 3.87177728e-02 -3.12007070e-01 -7.87810504e-01 1.57942042e-01 -1.27807647e-01 -6.44305944e-01 1.45146817e-01 -4.84196365e-01 -1.31277597e+00 -2.72643626e-01 2.10443348e-01 -1.46006748e-01 6.06162250e-01 5.42012811e-01 1.18472672e+00 9.87257659e-01 1.12449959e-01 -4.43722904e-01 -3.80797029e-01 -9.14092660e-01 -2.32915401e-01 7.33315051e-01 5.12598574e-01 -1.89041480e-01 4.39949840e-01 5.44176936e-01]
[7.899590969085693, 6.374433994293213]
c49523bb-7983-4a06-bf6d-08ca4830b753
about-the-cost-of-global-privacy-in-density
2306.14535
null
https://arxiv.org/abs/2306.14535v1
https://arxiv.org/pdf/2306.14535v1.pdf
About the Cost of Global Privacy in Density Estimation
We study non-parametric density estimation for densities in Lipschitz and Sobolev spaces, and under global privacy. In particular, we investigate regimes where the privacy budget is not supposed to be constant. We consider the classical definition of global differential privacy, but also the more recent notion of global concentrated differential privacy. We recover the result of Barber \& Duchi (2014) stating that histogram estimators are optimal against Lipschitz distributions for the L2 risk, and under regular differential privacy, and we extend it to other norms and notions of privacy. Then, we investigate higher degrees of smoothness, drawing two conclusions: First, and contrary to what happens with constant privacy budget (Wasserman \& Zhou, 2010), there are regimes where imposing privacy degrades the regular minimax risk of estimation on Sobolev densities. Second, so-called projection estimators are near-optimal against the same classes of densities in this new setup with pure differential privacy, but contrary to the constant privacy budget case, it comes at the cost of relaxation. With zero concentrated differential privacy, there is no need for relaxation, and we prove that the estimation is optimal.
['Rémi Gribonval', 'Aurélien Garivier', 'Clément Lalanne']
2023-06-26
null
null
null
null
['density-estimation']
['methodology']
[ 1.08003048e-02 5.11984885e-01 -2.87407245e-02 -1.43878907e-01 -8.44522893e-01 -6.91318154e-01 2.01025113e-01 2.20144346e-01 -5.40964723e-01 1.00658834e+00 3.25118154e-01 -1.50048777e-01 -3.16232920e-01 -9.17140603e-01 -8.80264759e-01 -1.10628235e+00 -1.96312353e-01 1.06338702e-01 -6.56673778e-03 9.39953327e-02 1.82495594e-01 6.96362734e-01 -1.23354876e+00 -3.54106158e-01 7.90306985e-01 1.11012566e+00 -4.77323085e-01 6.47040367e-01 2.71202654e-01 4.64185208e-01 -1.78453133e-01 -7.65647054e-01 6.32280767e-01 -4.06277210e-01 -6.60591722e-01 4.17684577e-03 7.34467447e-01 -5.02063751e-01 -2.08020717e-01 1.74142873e+00 6.06314123e-01 1.34503558e-01 8.38783026e-01 -1.14716280e+00 -8.43232334e-01 4.15462524e-01 -6.47860229e-01 -2.54005492e-02 9.92262438e-02 -1.53053403e-01 9.04555142e-01 -5.05804479e-01 4.69358057e-01 1.06176412e+00 1.19826961e+00 5.90671897e-01 -1.64572537e+00 -3.89000803e-01 -1.72795042e-01 -4.00671393e-01 -1.60755575e+00 -6.02717280e-01 1.25308767e-01 -6.28759682e-01 2.82917857e-01 3.85932177e-01 2.33561307e-01 1.07644415e+00 4.57016863e-02 4.52981591e-01 1.15909994e+00 -1.36557877e-01 4.92701977e-01 6.46701872e-01 1.36305392e-01 5.20800233e-01 7.37502575e-01 -1.13510350e-02 -9.69416182e-03 -4.86891806e-01 7.29030132e-01 -4.09913156e-03 -6.62491918e-01 -6.58395290e-01 -7.27323890e-01 1.03176463e+00 -1.81389511e-01 4.25270319e-01 -3.00347686e-01 -2.48754159e-01 3.56268853e-01 6.00453734e-01 7.95391262e-01 2.81713068e-01 -4.00836706e-01 -1.93101484e-02 -7.85024941e-01 4.90624338e-01 1.35373580e+00 1.36278999e+00 8.07829440e-01 -1.85998619e-01 -3.63687515e-01 4.07618791e-01 -1.51715532e-01 6.04832590e-01 4.99746129e-02 -1.07437408e+00 4.28048223e-01 -2.38497481e-01 4.50803578e-01 -9.98930454e-01 -6.77477196e-02 -1.87103003e-01 -9.97501791e-01 9.73774195e-02 8.68304551e-01 -4.88104373e-01 -1.68513022e-02 2.11930227e+00 1.93684310e-01 7.19203353e-02 1.03444509e-01 6.90604806e-01 -2.32378513e-01 4.56019223e-01 -1.43634349e-01 -6.47216320e-01 1.01642561e+00 -2.43958354e-01 -7.75149763e-01 3.84516984e-01 6.95836067e-01 -1.25150099e-01 1.01237571e+00 3.94790649e-01 -1.04168415e+00 -4.64846753e-03 -8.66123080e-01 -2.11748600e-01 -3.31163764e-01 -4.54616785e-01 2.13066906e-01 1.22453439e+00 -1.31617439e+00 9.15418506e-01 -4.53217536e-01 -5.88012218e-01 5.32772005e-01 8.89628381e-02 -4.14694160e-01 -8.14198554e-02 -1.13880312e+00 6.35797977e-01 -3.31050940e-02 -2.49561027e-01 -6.64334476e-01 -1.17529690e+00 -8.62885594e-01 3.21657211e-01 1.77777767e-01 -4.53791499e-01 8.80451500e-01 -8.39623690e-01 -1.49426579e+00 7.64158607e-01 -2.75524389e-02 -8.01885605e-01 1.31325865e+00 -9.51238200e-02 1.43700391e-01 4.20036651e-02 -4.78699096e-02 1.48654714e-01 6.30356431e-01 -8.87057126e-01 -3.90899807e-01 -8.07563484e-01 1.18191406e-01 2.93202084e-02 -4.99824017e-01 -1.58989623e-01 2.16590002e-01 -3.44184846e-01 -4.95599657e-01 -6.89186811e-01 -1.24697499e-01 5.28072596e-01 -2.14255884e-01 1.26759127e-01 7.87968993e-01 -7.32012630e-01 8.89722228e-01 -2.42744875e+00 3.71323503e-03 3.76101643e-01 1.43131942e-01 -7.32777044e-02 2.70318478e-01 8.27175751e-02 1.17930703e-01 4.20086235e-01 -8.54827344e-01 -8.43334675e-01 5.09059548e-01 1.39645249e-01 -2.27642283e-01 1.32971811e+00 -7.66360089e-02 6.09445810e-01 -6.34666383e-01 -3.52140516e-01 7.07481196e-03 6.82534635e-01 -9.14833665e-01 -5.16281389e-02 4.82216403e-02 7.25313842e-01 -3.01171392e-01 2.14300737e-01 1.27247822e+00 1.07202843e-01 7.67467311e-03 3.16422999e-01 -2.86118567e-01 -3.31418544e-01 -1.35501218e+00 1.25232100e+00 -3.85113865e-01 6.23708129e-01 7.79011607e-01 -1.24371052e+00 5.40008068e-01 3.91601592e-01 4.36518788e-01 -2.29414746e-01 2.79226303e-01 3.11230719e-01 -6.86288238e-01 -3.91240805e-01 3.04794759e-01 -6.89054310e-01 -8.49365294e-02 1.55044675e-01 -4.85722460e-02 4.86646481e-02 -2.65007317e-01 -2.14321822e-01 9.95952249e-01 -3.19730371e-01 2.38989651e-01 -1.04443836e+00 5.64227402e-01 -6.99911356e-01 5.07333815e-01 8.06007564e-01 -3.69946659e-01 6.45857871e-01 1.13391232e+00 2.01972961e-01 -1.24130237e+00 -1.09744620e+00 -8.75111341e-01 9.08367932e-01 -1.08710406e-02 -6.99336082e-02 -9.06020403e-01 -5.52095234e-01 4.99732465e-01 6.90208793e-01 -1.08170962e+00 2.26630364e-02 6.07888475e-02 -5.79558492e-01 7.03314006e-01 -3.38773280e-02 5.82625866e-01 -3.88610154e-01 -1.95310399e-01 -3.06222528e-01 3.00064057e-01 -9.93146837e-01 -8.39958370e-01 -7.33250156e-02 -9.64498758e-01 -7.98888683e-01 -1.26005101e+00 -3.79156440e-01 5.52631319e-01 -2.56943345e-01 5.68286598e-01 -5.15411019e-01 1.51428968e-01 7.68007636e-01 -1.15927212e-01 -2.84863770e-01 -1.61323979e-01 -8.09564888e-02 8.66642594e-02 4.03873056e-01 2.41214499e-01 -7.05904365e-01 -5.65509796e-01 2.33702332e-01 -1.06285179e+00 -7.70952284e-01 1.47203684e-01 4.99403954e-01 5.88369370e-01 2.35102460e-01 4.95071948e-01 -1.17701650e+00 5.14250755e-01 -8.70977163e-01 -9.86671627e-01 -2.28649080e-02 -6.28863215e-01 8.99766758e-02 8.50710750e-01 -2.33293846e-01 -9.63904440e-01 -2.77824670e-01 -4.07693796e-02 -4.97262269e-01 -8.84790793e-02 4.86374199e-02 -4.70634490e-01 -4.01530653e-01 5.83809078e-01 8.76235515e-02 1.89534172e-01 -6.47974908e-01 3.06535363e-01 6.47663116e-01 1.88602433e-01 -6.98880076e-01 5.33930302e-01 8.78538668e-01 1.91193730e-01 -1.21160865e+00 -6.54713392e-01 -2.54659444e-01 -4.66737002e-01 3.17496091e-01 9.60561156e-01 -5.25484324e-01 -9.89338458e-01 4.29035425e-01 -8.06873322e-01 -3.38968784e-01 -1.00765908e+00 3.56069863e-01 -1.11409676e+00 9.02842104e-01 -6.04305565e-01 -1.46606982e+00 1.10402420e-01 -9.53968227e-01 8.35304558e-01 -1.46549091e-01 1.63493171e-01 -1.45417583e+00 1.47788733e-01 -1.96769550e-01 8.08147490e-01 4.40083444e-01 4.89021540e-01 -6.20385945e-01 -5.20885922e-02 -2.75000006e-01 -2.67306328e-01 7.56504297e-01 -1.29235432e-01 -4.69883800e-01 -1.06551099e+00 -3.80030245e-01 7.39108920e-01 2.48041432e-02 6.48308635e-01 6.43078983e-01 1.41809905e+00 -1.12349236e+00 1.38644367e-01 9.67704594e-01 1.61197507e+00 -3.22643906e-01 8.33762705e-01 -1.36404440e-01 4.00639862e-01 1.04185092e+00 1.18053369e-01 6.89586639e-01 1.63587853e-01 4.64861006e-01 1.75121799e-01 2.30906993e-01 6.11507535e-01 -1.87528893e-01 4.54783291e-01 1.57014355e-01 1.38127327e-01 2.22844700e-03 -3.05592954e-01 6.12720251e-01 -1.73419571e+00 -1.14946055e+00 -8.22690800e-02 2.90104246e+00 9.75872040e-01 -3.49764645e-01 6.17953122e-01 -2.50052810e-01 9.52552676e-01 7.41312355e-02 -3.44182789e-01 -5.79227686e-01 -4.17100310e-01 1.76061556e-01 1.34385586e+00 9.45271015e-01 -1.08739567e+00 4.42515165e-01 6.63317728e+00 8.49714398e-01 -7.75406778e-01 5.35148442e-01 8.14726949e-01 -5.23143113e-02 -5.05604267e-01 -4.92856316e-02 -5.41665852e-01 6.37526989e-01 1.09953952e+00 -4.40482914e-01 4.27643269e-01 9.71823692e-01 -5.16176522e-02 -6.11973591e-02 -1.24138713e+00 8.06376457e-01 -1.19532563e-01 -8.09465587e-01 -6.02064848e-01 8.12830031e-01 7.34791577e-01 -3.38205427e-01 3.05940807e-01 1.72398195e-01 2.10908532e-01 -1.00015819e+00 4.01561111e-01 7.22697914e-01 8.30885947e-01 -9.84479547e-01 8.81984830e-01 3.47896069e-01 -9.15655911e-01 7.45418435e-03 -7.08610594e-01 3.13700624e-02 1.60155296e-01 7.77093649e-01 1.24266267e-01 3.67124915e-01 4.75953907e-01 6.71587169e-01 -2.42332593e-02 9.87759650e-01 3.25924397e-01 3.76851410e-01 -5.15569031e-01 3.14282179e-01 -1.94558240e-02 -6.67485237e-01 6.85714185e-01 1.52251732e+00 6.76213145e-01 3.24963406e-02 -4.07576144e-01 9.21361029e-01 -1.79951817e-01 2.53712207e-01 -1.00030315e+00 6.10144325e-02 2.63409495e-01 7.87976980e-01 -4.66153286e-02 -1.07597336e-01 -4.94629145e-01 1.10088384e+00 2.12191045e-01 3.82725835e-01 -6.46287680e-01 -4.28455919e-01 9.01729941e-01 4.09682900e-01 4.67039973e-01 -1.17563471e-01 -3.45711917e-01 -1.26414311e+00 1.36203766e-01 -3.25015068e-01 4.69451398e-01 6.51594549e-02 -1.72423744e+00 -3.29621851e-01 2.60336250e-01 -8.55099857e-01 2.71319538e-01 -7.12579608e-01 -5.12093484e-01 9.52036142e-01 -1.35367572e+00 -4.53622162e-01 3.57206047e-01 6.26955807e-01 -2.06243366e-01 3.28839809e-01 6.62379980e-01 4.19959843e-01 -3.59070539e-01 9.02561545e-01 6.55787289e-01 -9.15582478e-02 5.20472348e-01 -1.60411894e+00 4.55504172e-02 7.53147602e-01 -3.76069665e-01 6.75936520e-01 8.06703210e-01 -4.85800982e-01 -1.50988150e+00 -1.15174496e+00 8.06050599e-01 -4.33667451e-01 8.55385184e-01 -5.70904434e-01 -1.18534982e+00 8.12382340e-01 1.20678589e-01 1.79380774e-01 7.42484212e-01 -2.88762506e-02 -3.40905845e-01 -5.50569743e-02 -2.11030030e+00 4.36635226e-01 1.04561400e+00 -5.30891240e-01 -1.77224621e-01 4.99640614e-01 7.02407777e-01 8.74987543e-02 -1.21683586e+00 6.69843610e-03 5.42464614e-01 -1.08645725e+00 6.89654529e-01 -4.69107568e-01 -1.09128885e-01 9.49691162e-02 -5.99298298e-01 -8.15721571e-01 -4.84542549e-02 -9.68291581e-01 -4.23653275e-02 1.27426958e+00 3.28092396e-01 -1.21748257e+00 7.04908073e-01 8.34653437e-01 1.94470495e-01 -4.93637145e-01 -1.39596486e+00 -1.12506497e+00 7.12327719e-01 -3.57916236e-01 3.49080563e-01 1.20266736e+00 1.80287436e-01 -4.12466615e-01 -6.32361352e-01 1.58986181e-01 1.01416814e+00 -7.31636524e-01 6.42941892e-01 -1.04565144e+00 -6.00338221e-01 -6.39812946e-01 -6.96413815e-01 -1.07564557e+00 3.05777669e-01 -7.50980318e-01 1.40909240e-01 -9.07273471e-01 1.00498065e-01 -2.29598790e-01 -1.15918182e-01 -2.13407770e-01 2.18394369e-01 -1.69253983e-02 -1.11527719e-01 -2.89624948e-02 -3.98648947e-01 7.24016368e-01 9.46280181e-01 1.48905143e-01 7.73706660e-03 3.41183573e-01 -9.68722284e-01 4.61715341e-01 5.43226182e-01 -2.70625591e-01 -3.49090338e-01 1.01041868e-01 2.16833308e-01 5.73132001e-03 4.23990667e-01 -8.28665376e-01 2.22014815e-01 -1.39341190e-01 -3.56674418e-02 1.48827508e-01 2.02231944e-01 -9.18401122e-01 9.58627537e-02 2.55640686e-01 -3.77494901e-01 -2.90616751e-01 5.20151369e-02 9.31461155e-01 3.04013848e-01 -3.97341460e-01 1.28372002e+00 1.42697036e-01 2.08602637e-01 6.02697670e-01 -3.35810512e-01 3.50084245e-01 1.04469681e+00 -2.62574553e-01 -1.68679297e-01 -6.77463055e-01 -9.33652759e-01 1.05108708e-01 8.05782616e-01 -3.49512577e-01 2.68985897e-01 -1.24948728e+00 -7.56457329e-01 1.48145974e-01 -2.50652712e-03 -4.13004160e-02 4.85254705e-01 1.47914636e+00 -3.36748958e-01 8.64868239e-02 1.58288658e-01 -3.35771054e-01 -6.81592107e-01 8.08987617e-01 6.31888688e-01 1.46381900e-01 -6.91398740e-01 8.14217746e-01 6.24339700e-01 -1.50928766e-01 3.44188005e-01 -3.00820142e-01 2.26103693e-01 1.71135738e-02 4.76840466e-01 5.67364395e-01 -2.50362635e-01 -5.58961689e-01 -2.12340012e-01 4.68685806e-01 1.08938277e-01 -3.57003123e-01 1.09969866e+00 -3.12363625e-01 -6.52427748e-02 6.34746134e-01 1.77198207e+00 5.19515634e-01 -1.51792896e+00 -7.15858191e-02 4.43107486e-02 -6.71132028e-01 -2.93647826e-01 -3.89008969e-02 -1.07604909e+00 8.53075147e-01 4.78664786e-01 7.42829978e-01 8.51871908e-01 -3.96296307e-02 4.25812244e-01 1.16111584e-01 3.78093511e-01 -9.28931594e-01 -7.63555288e-01 4.63503361e-01 8.64842474e-01 -9.89855647e-01 -1.46699920e-01 -1.04486987e-01 -5.85006893e-01 7.29959369e-01 3.27425040e-02 -5.38025320e-01 1.04641294e+00 3.35314631e-01 -6.00660503e-01 1.36984020e-01 -1.30424708e-01 1.14234273e-04 -4.22294959e-02 8.34502101e-01 2.01427713e-01 3.41085233e-02 -5.37285686e-01 5.59000432e-01 -1.35090917e-01 -2.18336150e-01 5.01347423e-01 6.42691970e-01 -3.37904006e-01 -7.31614769e-01 -1.96254447e-01 4.53573883e-01 -7.74202526e-01 9.69420001e-02 -3.71097118e-01 7.99175501e-01 -5.33244275e-02 6.50194585e-01 2.39156216e-01 1.33948326e-01 2.93754935e-01 2.73907222e-02 3.90243471e-01 -3.65804195e-01 -2.77863145e-01 -2.56701201e-01 -1.79673299e-01 -4.76842701e-01 -1.69687316e-01 -8.81867945e-01 -4.69037712e-01 -9.32256162e-01 -1.05809852e-01 4.52594250e-01 4.19056773e-01 7.11827219e-01 2.05890834e-01 -1.44505963e-01 5.55948853e-01 -5.50332785e-01 -1.03441286e+00 -8.73697937e-01 -1.29799223e+00 3.04593593e-01 8.26157212e-01 -2.69213647e-01 -1.12315214e+00 -3.71810168e-01]
[7.094205379486084, 4.288792610168457]
6d96f3c0-1d97-483d-9604-f97213c44762
hybrid-window-attention-based-transformer
2209.07704
null
https://arxiv.org/abs/2209.07704v1
https://arxiv.org/pdf/2209.07704v1.pdf
Hybrid Window Attention Based Transformer Architecture for Brain Tumor Segmentation
As intensities of MRI volumes are inconsistent across institutes, it is essential to extract universal features of multi-modal MRIs to precisely segment brain tumors. In this concept, we propose a volumetric vision transformer that follows two windowing strategies in attention for extracting fine features and local distributional smoothness (LDS) during model training inspired by virtual adversarial training (VAT) to make the model robust. We trained and evaluated network architecture on the FeTS Challenge 2022 dataset. Our performance on the online validation dataset is as follows: Dice Similarity Score of 81.71%, 91.38% and 85.40%; Hausdorff Distance (95%) of 14.81 mm, 3.93 mm, 11.18 mm for the enhancing tumor, whole tumor, and tumor core, respectively. Overall, the experimental results verify our method's effectiveness by yielding better performance in segmentation accuracy for each tumor sub-region. Our code implementation is publicly available : https://github.com/himashi92/vizviva_fets_2022
['Mehrtash Harandi', 'Gary Egan', 'Zhaolin Chen', 'Munawar Hayat', 'Himashi Peiris']
2022-09-16
null
null
null
null
['brain-tumor-segmentation']
['medical']
[-1.94270629e-02 2.56224066e-01 -1.57263860e-01 -4.80189621e-01 -1.08270919e+00 -4.13714796e-01 4.27025378e-01 -7.56686479e-02 -6.26335621e-01 8.53371620e-01 1.29117459e-01 -3.28120470e-01 -5.82406558e-02 -4.53728557e-01 -5.98588765e-01 -8.11266124e-01 -2.04793841e-01 2.84818769e-01 3.46444786e-01 5.16650546e-03 1.66579306e-01 6.05403662e-01 -5.93940198e-01 2.35633671e-01 1.02967143e+00 1.28646243e+00 3.13355982e-01 5.40303648e-01 7.48398453e-02 6.80012524e-01 -3.38085026e-01 -2.26833746e-01 2.96222210e-01 -1.69177924e-03 -1.02683115e+00 -2.57673830e-01 2.74408251e-01 -3.32344681e-01 -5.10296226e-01 1.37452435e+00 3.92387539e-01 -6.14692755e-02 9.11635876e-01 -1.05973840e+00 -8.92499030e-01 4.72921133e-01 -7.82826006e-01 6.42846167e-01 -3.65769684e-01 3.79194915e-01 5.23064494e-01 -8.96788418e-01 7.07379580e-01 5.03699899e-01 8.17060173e-01 7.21550703e-01 -7.30069041e-01 -7.49708176e-01 -2.89644808e-01 2.01787084e-01 -1.50855720e+00 -2.59168476e-01 5.07411420e-01 -7.38269448e-01 5.79586029e-01 5.37976325e-01 5.74550986e-01 1.05840850e+00 8.16785634e-01 6.24498725e-01 1.28475261e+00 5.83945327e-02 2.87718568e-02 9.28568747e-03 2.24751368e-01 8.44507873e-01 1.36786819e-01 -3.24608199e-02 1.11968957e-01 6.41217753e-02 9.25141096e-01 1.09259598e-01 -5.16263306e-01 -2.17319012e-01 -1.24806356e+00 8.78401399e-01 9.53288019e-01 5.30632079e-01 -3.99353236e-01 4.34753522e-02 6.19924545e-01 -1.28353104e-01 5.09712875e-01 1.15730815e-01 -2.29271427e-01 2.49178067e-01 -8.71840358e-01 -3.92766669e-02 3.67474332e-02 8.14275920e-01 -1.03930943e-03 6.84458092e-02 -5.39242506e-01 8.54900301e-01 2.24714711e-01 4.38144177e-01 1.09985042e+00 -7.87028670e-01 2.69341975e-01 3.09031039e-01 -1.50130212e-01 -8.33331287e-01 -6.98996246e-01 -6.69578195e-01 -1.21369922e+00 -1.56725049e-02 5.19926608e-01 -6.93712234e-02 -1.18067026e+00 1.66693103e+00 2.37037390e-01 3.40067387e-01 -2.93952167e-01 9.25483346e-01 1.10712218e+00 5.30071221e-02 1.93605393e-01 -1.72186702e-01 1.15905941e+00 -1.05750275e+00 -5.85388124e-01 7.96134025e-02 6.33662701e-01 -4.92157400e-01 1.07661307e+00 -1.60059139e-01 -1.12101305e+00 -2.22737148e-01 -8.05647492e-01 1.57999679e-01 -3.35518330e-01 4.11701910e-02 6.83669031e-01 4.17938590e-01 -1.23515749e+00 5.07216632e-01 -1.18413126e+00 -1.93788409e-02 1.09008861e+00 3.11959803e-01 -3.52907240e-01 1.22121051e-02 -1.03553176e+00 8.59695315e-01 2.89606810e-01 1.36080431e-03 -1.06494200e+00 -1.22672439e+00 -5.91827154e-01 -1.77110553e-01 -9.48528945e-02 -5.85501969e-01 1.12892199e+00 -9.26292896e-01 -1.21937191e+00 1.15655923e+00 1.24715872e-01 -6.57514095e-01 7.42872298e-01 1.89480260e-01 -2.94998437e-01 4.48975176e-01 1.89573690e-01 7.27468610e-01 4.14394826e-01 -1.00275648e+00 -1.63483426e-01 -7.59062469e-01 -3.92663687e-01 5.55999056e-02 -1.79364651e-01 2.65465807e-02 -2.11805746e-01 -6.96808517e-01 4.25579548e-02 -7.86233008e-01 -3.77967387e-01 5.14685363e-02 -6.44244611e-01 9.05469134e-02 5.92774868e-01 -1.11300588e+00 8.36672008e-01 -1.85416842e+00 -1.27078354e-01 2.49985099e-01 4.99638796e-01 3.25326502e-01 4.01827991e-02 -4.78807181e-01 -2.58066952e-01 1.52808055e-01 -6.23573124e-01 -1.41758267e-02 -2.63909370e-01 -2.37287372e-01 1.73518121e-01 7.40182996e-01 5.47539406e-02 1.36589622e+00 -7.27018416e-01 -6.75391138e-01 2.29779631e-01 5.70212245e-01 -3.95203680e-01 4.41582575e-02 3.51793408e-01 7.63549864e-01 -6.24830961e-01 8.29965055e-01 6.62229180e-01 -2.15532675e-01 -2.79181421e-01 -2.14902535e-01 1.47574320e-01 -4.61487323e-01 -4.26257908e-01 1.92391753e+00 -2.16238782e-01 5.32127261e-01 1.51003227e-01 -1.14471698e+00 7.34471917e-01 1.89918444e-01 9.09986436e-01 -8.16658914e-01 4.93199795e-01 1.75361454e-01 1.57858998e-01 -6.54036701e-01 2.28855945e-02 -1.28262892e-01 2.17893541e-01 -6.64708689e-02 4.27131467e-02 -1.59873679e-01 -1.34295523e-01 2.07096294e-01 1.13298941e+00 -4.49783087e-01 1.31565794e-01 -5.35605729e-01 5.11322021e-01 -9.62316245e-02 4.02519673e-01 4.46360290e-01 -9.34440732e-01 8.29733968e-01 3.29596460e-01 -2.53092080e-01 -8.93765807e-01 -1.37117302e+00 -5.96933842e-01 4.75764900e-01 -4.90110926e-03 1.56244308e-01 -9.82901692e-01 -9.03343797e-01 -1.73855618e-01 5.69013238e-01 -1.06198204e+00 -2.43805930e-01 -3.53623211e-01 -8.27948093e-01 6.46295369e-01 7.57247627e-01 4.94559109e-01 -1.00506616e+00 -5.52642047e-01 -1.30143911e-01 -1.99087173e-01 -1.01564658e+00 -7.34009326e-01 -1.03350751e-01 -9.46469128e-01 -1.24659848e+00 -1.23341167e+00 -8.51492763e-01 8.76563847e-01 -8.88653621e-02 8.81416023e-01 -1.78251974e-02 -6.02495015e-01 1.68223962e-01 -1.23089133e-02 -3.02619010e-01 -5.82441352e-02 -7.17832986e-03 -2.34528720e-01 -1.46960795e-01 3.55446944e-03 -5.37863672e-01 -8.55655134e-01 1.72116950e-01 -8.23906004e-01 2.17163950e-01 5.98502517e-01 1.02569330e+00 9.65536177e-01 -4.85753357e-01 5.35924435e-01 -7.32203603e-01 5.52665949e-01 -7.10800171e-01 -3.71622533e-01 2.63776958e-01 -4.22222406e-01 -4.85383213e-01 5.17406404e-01 -3.75614911e-01 -6.57446861e-01 -1.02154471e-01 -1.80568233e-01 -6.70371950e-01 -2.13277385e-01 4.29580182e-01 3.02749783e-01 -2.48039767e-01 5.77183187e-01 3.73537064e-01 2.08566219e-01 1.06021300e-01 2.34290045e-02 5.25528014e-01 6.36281252e-01 -1.17158160e-01 5.81665397e-01 6.39489532e-01 -3.67349498e-02 -5.58316827e-01 -6.93647265e-01 -2.99159259e-01 -6.74442410e-01 -3.47435772e-01 1.08429039e+00 -6.55015171e-01 -5.11859238e-01 4.92588311e-01 -6.57461584e-01 -5.66564262e-01 -2.59342015e-01 5.10755301e-01 -5.31769931e-01 1.96236476e-01 -7.32162535e-01 -1.45580649e-01 -9.95374203e-01 -1.55179477e+00 6.35866284e-01 4.40709621e-01 -1.63354725e-02 -1.06583631e+00 -1.37647301e-01 4.72991109e-01 9.31736529e-01 7.27622449e-01 6.48950696e-01 -7.81662226e-01 -1.44799173e-01 -2.23599419e-01 -5.19852519e-01 2.72354037e-01 6.83242455e-02 -9.16095674e-02 -7.37562358e-01 -3.20701718e-01 -7.83434957e-02 -3.53177249e-01 7.86094606e-01 1.00672686e+00 1.83607042e+00 -4.03188802e-02 -4.10338432e-01 1.01419568e+00 1.40637541e+00 3.26302320e-01 5.78238308e-01 2.93199033e-01 6.78169370e-01 1.79723993e-01 3.35349470e-01 3.31191182e-01 3.03378552e-01 3.89535815e-01 6.73940182e-01 -1.84660822e-01 -3.44051659e-01 2.54244447e-01 -8.90740901e-02 9.23320830e-01 -4.57012057e-02 2.46472389e-01 -1.29583502e+00 7.49673605e-01 -1.24206471e+00 -7.54082739e-01 -1.46899417e-01 1.95100582e+00 8.47978175e-01 3.33163962e-02 -1.75182074e-01 -4.15785909e-01 7.62488008e-01 8.26844946e-02 -7.79454529e-01 -1.56794950e-01 5.93113191e-02 1.84175208e-01 7.94676185e-01 5.19505501e-01 -1.21458077e+00 6.54535472e-01 5.41254234e+00 8.97390187e-01 -1.29486156e+00 4.61498350e-01 1.24042714e+00 -2.13638484e-01 -1.56700060e-01 -5.94567418e-01 -2.84984052e-01 6.57069743e-01 8.63823593e-01 -2.54005551e-01 2.35153869e-01 5.64866722e-01 2.22030535e-01 3.68559845e-02 -4.81237531e-01 9.12248313e-01 -5.04923649e-02 -1.40392566e+00 -2.14617088e-01 -1.70423873e-02 8.89575601e-01 5.20875931e-01 4.20176417e-01 2.53014028e-01 7.99652189e-02 -1.48695171e+00 3.70409369e-01 7.24863946e-01 1.12921345e+00 -7.59413600e-01 9.53854620e-01 7.63317645e-02 -9.74899411e-01 2.92749375e-01 -2.99857169e-01 6.51166618e-01 -1.88863993e-01 4.53037322e-01 -9.23487246e-01 5.58294296e-01 7.87383020e-01 5.72919309e-01 -7.83469915e-01 1.12538886e+00 2.13790759e-01 5.19994259e-01 -9.99316052e-02 1.87492877e-01 3.94933254e-01 -1.28829196e-01 3.43251586e-01 1.09875691e+00 1.83017179e-01 1.47045344e-01 -7.69990608e-02 9.03954566e-01 -2.49503106e-01 2.64322549e-01 -3.73232871e-01 2.12873042e-01 3.78231972e-01 1.42092645e+00 -9.43246424e-01 -2.57061630e-01 -3.13277364e-01 6.89349353e-01 4.65270936e-01 3.27129900e-01 -1.26953518e+00 -3.57810736e-01 3.23206753e-01 -8.55696946e-03 1.46105453e-01 1.90569758e-01 -6.15926445e-01 -1.06657410e+00 -1.05976209e-01 -4.97801423e-01 3.35095644e-01 -7.45133936e-01 -1.20583868e+00 7.00004637e-01 -3.10140431e-01 -7.70590663e-01 1.63131461e-01 -5.40551126e-01 -9.22233343e-01 8.76612186e-01 -1.41221964e+00 -1.06653666e+00 -6.26087248e-01 8.37302983e-01 4.73605841e-01 -2.49085918e-01 6.68175042e-01 1.74316838e-01 -6.92997694e-01 9.52961981e-01 4.62380558e-01 4.58016157e-01 5.36593795e-01 -1.08733082e+00 -8.94974545e-02 5.88098645e-01 -4.38880324e-01 4.08312201e-01 2.99223840e-01 -4.37607855e-01 -1.03230035e+00 -1.41975844e+00 2.92717397e-01 -1.92506433e-01 8.99162173e-01 7.30726346e-02 -1.00559080e+00 8.39883685e-01 2.22234249e-01 6.41278863e-01 6.03564799e-01 -4.97012049e-01 -9.16022435e-02 1.79399848e-02 -1.63109648e+00 5.87833464e-01 7.71760345e-01 -1.83428437e-01 -3.46710950e-01 5.76154351e-01 5.81289768e-01 -7.63427794e-01 -1.46573400e+00 6.08399153e-01 4.93259341e-01 -7.28463471e-01 1.13426292e+00 -5.86595595e-01 6.34423018e-01 8.17966834e-02 -1.06827997e-01 -1.22576213e+00 -3.32905442e-01 4.35124934e-02 9.87302214e-02 7.87787855e-01 4.30554718e-01 -6.95262730e-01 6.76037371e-01 6.31609499e-01 -6.09628618e-01 -1.42796195e+00 -1.11712301e+00 -5.32732785e-01 6.45821929e-01 -2.42203534e-01 4.26377505e-01 1.22969151e+00 -2.86608964e-01 -3.29223633e-01 1.59846053e-01 1.63430020e-01 7.17960775e-01 -1.60750389e-01 1.66927218e-01 -8.87363255e-01 1.03132360e-01 -8.51398110e-01 -3.96661073e-01 -1.91627041e-01 2.27944583e-01 -1.35156155e+00 -3.54489237e-01 -1.50647986e+00 5.94278097e-01 -2.82733798e-01 -8.49912882e-01 4.87597495e-01 -2.11579818e-02 3.06376696e-01 -6.92408979e-02 2.06724912e-01 -4.09545690e-01 6.14158809e-01 1.79081953e+00 -3.81286830e-01 1.40881523e-01 -7.36399367e-02 -6.47187471e-01 7.97903836e-01 1.37924457e+00 -3.39549154e-01 -2.62834519e-01 -3.67791444e-01 -8.88534129e-01 1.39564767e-01 4.13850844e-01 -1.00218034e+00 1.55573338e-01 -2.17135563e-01 7.13935792e-01 -2.86601931e-01 8.52211118e-02 -5.54953814e-01 -8.08200762e-02 8.42247486e-01 -4.16005701e-01 -9.51160640e-02 3.33449692e-01 1.77769110e-01 7.73168206e-02 -5.82737736e-02 1.26506376e+00 -2.00987794e-02 -5.35306394e-01 8.83752525e-01 -3.27147506e-02 3.83396298e-01 1.43079686e+00 -8.86148363e-02 -2.15448827e-01 7.94257596e-02 -8.48957777e-01 1.35050863e-01 1.33108616e-01 2.79178381e-01 7.97701955e-01 -1.43683934e+00 -8.31905186e-01 -5.91409917e-04 1.50804478e-03 -6.34923503e-02 4.96766686e-01 1.43718326e+00 -6.77884936e-01 5.69698691e-01 -5.14469683e-01 -6.52141869e-01 -1.16696179e+00 3.09762686e-01 8.16233158e-01 -4.38593626e-01 -7.01918066e-01 1.09646678e+00 3.15922260e-01 -3.16543162e-01 1.73041001e-01 -4.85403597e-01 -2.21148834e-01 -4.23919559e-01 5.47199845e-01 7.16074482e-02 6.44873083e-02 -7.00580835e-01 -4.83217627e-01 4.29136068e-01 -3.55334967e-01 1.55910909e-01 1.34939682e+00 1.39562218e-02 -9.01174545e-03 1.14838541e-01 1.39378142e+00 -3.44755381e-01 -1.25538385e+00 -2.42492363e-01 -1.71367556e-01 -2.77313352e-01 3.82498056e-01 -1.04159963e+00 -1.68701541e+00 6.63642406e-01 1.15055525e+00 -1.80550124e-02 1.08164239e+00 1.57457024e-01 8.52548838e-01 -2.88314939e-01 8.11545029e-02 -8.44517887e-01 -2.12114811e-01 4.62185770e-01 1.16721380e+00 -1.45614541e+00 -2.69694775e-01 -1.22638978e-01 -7.30050921e-01 7.87584603e-01 7.20922351e-01 -2.76371866e-01 7.09528148e-01 2.99559295e-01 6.47275373e-02 -3.25928092e-01 -3.42706561e-01 1.93384871e-01 4.08391237e-01 6.33920848e-01 4.25871462e-01 3.81103873e-01 -4.22195375e-01 8.35605204e-01 -2.08437800e-01 -1.15867704e-01 3.10740530e-01 6.63561821e-01 -4.03236806e-01 -2.79551536e-01 -1.28860250e-01 8.55195343e-01 -8.04301322e-01 -1.60247415e-01 -9.27881077e-02 1.02360034e+00 -1.28931418e-01 4.19703692e-01 3.96307707e-02 -2.01198503e-01 7.50215426e-02 -1.11123053e-02 4.41864550e-01 -2.53943890e-01 -4.41801041e-01 -1.66307017e-01 -4.38606709e-01 -7.59481430e-01 -7.94809833e-02 -6.27877235e-01 -1.50799060e+00 -2.00688481e-01 -4.45470549e-02 7.80804409e-03 7.81048179e-01 7.69520700e-01 1.95485204e-01 8.87931883e-01 6.04448140e-01 -4.63227659e-01 -4.43792611e-01 -1.02605259e+00 -6.01785362e-01 4.23199207e-01 1.30830362e-01 -7.42658079e-01 -1.26239866e-01 -1.06786944e-01]
[14.389050483703613, -2.342435836791992]